`langchain_pinecone.Pinecone` is deprecated in favor of
`PineconeVectorStore`, and is currently a subclass of
`PineconeVectorStore`.
```python
@deprecated(since="0.0.3", removal="0.2.0", alternative="PineconeVectorStore")
class Pinecone(PineconeVectorStore):
"""Deprecated. Use PineconeVectorStore instead."""
pass
```
**Description:** AzureSearch vector store has no tests. This PR adds
initial tests to validate the code can be imported and used.
**Issue:** N/A
**Dependencies:** azure-search-documents and azure-identity are added as
optional dependencies for testing
---------
Co-authored-by: Matt Gotteiner <[email protected]>
Co-authored-by: Bagatur <baskaryan@gmail.com>
**Description**:
_PebbloSafeLoader_: Add support for pebblo server and client version
**Documentation:** NA
**Unit test:** NA
**Issue:** NA
**Dependencies:** None
---------
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
- [ ] **Kinetica Document Loader**: "community: a class to load
Documents from Kinetica"
- [ ] **Kinetica Document Loader**:
- **Description:** implemented KineticaLoader in `kinetica_loader.py`
- **Dependencies:** install the Kinetica API using `pip install
gpudb==7.2.0.1 `
**Description:** Fixes a bug in the HuggingGPT task execution logic
here:
except Exception as e:
self.status = "failed"
self.message = str(e)
self.status = "completed"
self.save_product()
where a caught exception effectively just sets `self.message` and can
then throw an exception if, e.g., `self.product` is not defined.
**Issue:** None that I'm aware of.
**Dependencies:** None
**Twitter handle:** https://twitter.com/michaeljschock
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
- **Description:** Changes
`lanchain_core.output_parsers.CommaSeparatedListOutputParser` to handle
`,` as a delimiter alongside the previous implementation which used `, `
as delimiter.
- **Issue:** Started noticing that some results returned by LLMs were
not getting parsed correctly when the output contained `,` instead of `,
`.
- **Dependencies:** No
- **Twitter handle:** not active on twitter.
<!---
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
-->
- **Description**:
- **add support for more data types**: by default `IpexLLM` will load
the model in int4 format. This PR adds more data types support such as
`sym_in5`, `sym_int8`, etc. Data formats like NF3, NF4, FP4 and FP8 are
only supported on GPU and will be added in future PR.
- Fix a small issue in saving/loading, update api docs
- **Dependencies**: `ipex-llm` library
- **Document**: In `docs/docs/integrations/llms/ipex_llm.ipynb`, added
instructions for saving/loading low-bit model.
- **Tests**: added new test cases to
`libs/community/tests/integration_tests/llms/test_ipex_llm.py`, added
config params.
- **Contribution maintainer**: @shane-huang
Description: Add support for Semantic topics and entities.
Classification done by pebblo-server is not used to enhance metadata of
Documents loaded by document loaders.
Dependencies: None
Documentation: Updated.
Signed-off-by: Rahul Tripathi <rauhl.psit.ec@gmail.com>
Co-authored-by: Rahul Tripathi <rauhl.psit.ec@gmail.com>
Thank you for contributing to LangChain!
- [x] **PR title**
- [x] **PR message**:
- **Description:** Deprecate persist method in Chroma no longer exists
in Chroma 0.4.x
- **Issue:** #20851
- **Dependencies:** None
- **Twitter handle:** AndresAlgaba1
- [x] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [x] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
---------
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
**Description:**
This PR removes an unnecessary code snippet from the documentation. The
snippet in question is not relevant to the content and does not
contribute to the overall understanding of the topic. It contained
redundant imports and unused code, potentially causing confusion for
readers.
**Issue:**
There is no specific issue number associated with this change.
**Dependencies:**
No additional dependencies are required for this change.
---------
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
**Description:**
The RecursiveUrlLoader loader offers a link_regex parameter that can
filter out URLs. However, this filtering capability is limited, and if
the internal links of the website change, unexpected resources may be
loaded. These resources, such as font files, can cause problems in
subsequent embedding processing.
>
https://blog.langchain.dev/assets/fonts/source-sans-pro-v21-latin-ext_latin-regular.woff2?v=0312715cbf
We can add the Content-Type in the HTTP response headers to the document
metadata so developers can choose which resources to use. This allows
developers to make their own choices.
For example, the following may be a good choice for text knowledge.
- text/plain - simple text file
- text/html - HTML web page
- text/xml - XML format file
- text/json - JSON format data
- application/pdf - PDF file
- application/msword - Word document
and ignore the following
- text/css - CSS stylesheet
- text/javascript - JavaScript script
- application/octet-stream - binary data
- image/jpeg - JPEG image
- image/png - PNG image
- image/gif - GIF image
- image/svg+xml - SVG image
- audio/mpeg - MPEG audio files
- video/mp4 - MP4 video file
- application/font-woff - WOFF font file
- application/font-ttf - TTF font file
- application/zip - ZIP compressed file
- application/octet-stream - binary data
**Twitter handle:** @coolbeevip
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
Thank you for contributing to LangChain!
- [x] **PR title**: "package: description"
- Where "package" is whichever of langchain, community, core,
experimental, etc. is being modified. Use "docs: ..." for purely docs
changes, "templates: ..." for template changes, "infra: ..." for CI
changes.
- Example: "community: add foobar LLM"
**Description:** In VoyageAI text-embedding examples use voyage-law-2
model
- [x] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [x] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
Thank you for contributing to LangChain!
- [x] **PR title**: Fix misplaced zep cloud example links
- [x] **PR message**:
- **Description:** Fixes misplaced links for vector store and memory zep
cloud examples
- [x] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [x] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
- **Description:** Adapt JinaEmbeddings to run with the new Jina AI
Rerank API
- **Twitter handle:** https://twitter.com/JinaAI_
- [ ] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [ ] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
---------
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
OpenAI API compatible server may not support `safe_len_embedding`，
use `disable_safe_len_embeddings=True` to disable it.
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
* Updating the provider docs page.
The RAG example was meant to be moved to cookbook, but was merged by
mistake.
* Fix bug in Groundedness Check
---------
Co-authored-by: JuHyung-Son <sonju0427@gmail.com>
Co-authored-by: Erick Friis <erick@langchain.dev>
Currently, when a new dev container is created, poetry does not work in
it with the error "No module named 'rapidfuzz'".
Install Poetry outside the project venv so that poetry and project
dependencies do not get mixed. Use pipx to install poetry securely in
its own isolated environment.
Issue: #12237
Twitter handle: https://twitter.com/ibratoev
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
- **Description:** Currently, the regex is static (`r"(?<=[.?!])\s+"`),
which is only useful for certain use cases. The current change only
moves this to be a parameter of split_text(). Which adds flexibility
without making it more complex (as the default regex is still the same).
- **Issue:** Not applicable (I searched, no one seems to have created
this issue yet).
- **Dependencies:** None.
_If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17._
---------
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
Description: MarkdownHeaderTextSplitter Fails to Parse Headers with
non-printable characters. more #20643
The following is the official test case. Just replacing `# Foo\n\n` with
`\ufeff# Foo\n\n` will cause the test case to fail.
chunk metadata is empty
```python
def test_md_header_text_splitter_1() -> None:
"""Test markdown splitter by header: Case 1."""
markdown_document = (
"\ufeff# Foo\n\n"
" ## Bar\n\n"
"Hi this is Jim\n\n"
"Hi this is Joe\n\n"
" ## Baz\n\n"
" Hi this is Molly"
)
headers_to_split_on = [
("#", "Header 1"),
("##", "Header 2"),
]
markdown_splitter = MarkdownHeaderTextSplitter(
headers_to_split_on=headers_to_split_on,
)
output = markdown_splitter.split_text(markdown_document)
expected_output = [
Document(
page_content="Hi this is Jim \nHi this is Joe",
metadata={"Header 1": "Foo", "Header 2": "Bar"},
),
Document(
page_content="Hi this is Molly",
metadata={"Header 1": "Foo", "Header 2": "Baz"},
),
]
assert output == expected_output
```
twitter: @coolbeevip
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Description :
- added functionalities - delete, index creation, using existing
connection object etc.
- updated usage
- Added LaceDB cloud OSS support
make lint_diff , make test checks done
- **Description:** fix a bug in the agent_token_buffer_memory
- **Issue:** agent_token_buffer_memory was not working with openai tools
- **Dependencies:** None
- **Twitter handle:** @pokidyshef
**Description:** Adds the command to install packages required before
using _Unstructured_ and _PDFMiner_ from `langchain.community`
**Documentation Page Being Updated:** [LangChain > Retrieval > Document
loaders > PDF > Using
Unstructured](https://python.langchain.com/docs/modules/data_connection/document_loaders/pdf/#using-unstructured)
**Issue:** #20719
**Dependencies:** no dependencies
**Twitter handle:** SalikaDave
<!--
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17. -->
---------
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
## Description
Add `aprep_output` method to `langchain/chains/base.py`. Some downstream
`ChatMessageHistory` objects that use async connections require an async
way to append to the context.
It turned out that `ainvoke()` was calling `prep_output` which is
synchronous.
---------
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
# Proxy Fix for Groq Class 🐛🚀
## Description
This PR fixes a bug related to proxy settings in the `Groq` class,
allowing users to connect to LangChain services via a proxy.
## Changes Made
- ✅ FIX support for specifying proxy settings in the `Groq` class.
- ✅ Resolved the bug causing issues with proxy settings.
- ❌ Did not include unit tests and documentation updates.
- ❌ Did not run make format, make lint, and make test to ensure code
quality and functionality because I couldn't get it to run, so I don't
program in Python and couldn't run `ruff`.
- ❔ Ensured that the changes are backwards compatible.
- ✅ No additional dependencies were added to `pyproject.toml`.
### Error Before Fix
```python
Traceback (most recent call last):
File "/home/bg/Documents/code/github.com/back2nix/test/groq/main.py", line 9, in <module>
chat = ChatGroq(
^^^^^^^^^
File "/home/bg/Documents/code/github.com/back2nix/test/groq/venv310/lib/python3.11/site-packages/langchain_core/load/serializable.py", line 120, in __init__
super().__init__(**kwargs)
File "/home/bg/Documents/code/github.com/back2nix/test/groq/venv310/lib/python3.11/site-packages/pydantic/v1/main.py", line 341, in __init__
raise validation_error
pydantic.v1.error_wrappers.ValidationError: 1 validation error for ChatGroq
__root__
Invalid `http_client` argument; Expected an instance of `httpx.AsyncClient` but got <class 'httpx.Client'> (type=type_error)
```
### Example usage after fix
```python3
import os
import httpx
from langchain_core.prompts import ChatPromptTemplate
from langchain_groq import ChatGroq
chat = ChatGroq(
temperature=0,
groq_api_key=os.environ.get("GROQ_API_KEY"),
model_name="mixtral-8x7b-32768",
http_client=httpx.Client(
proxies="socks5://127.0.0.1:1080",
transport=httpx.HTTPTransport(local_address="0.0.0.0"),
),
http_async_client=httpx.AsyncClient(
proxies="socks5://127.0.0.1:1080",
transport=httpx.HTTPTransport(local_address="0.0.0.0"),
),
)
system = "You are a helpful assistant."
human = "{text}"
prompt = ChatPromptTemplate.from_messages([("system", system), ("human", human)])
chain = prompt | chat
out = chain.invoke({"text": "Explain the importance of low latency LLMs"})
print(out)
```
---------
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
Implemented the ability to enable full-text search within the
SingleStore vector store, offering users a versatile range of search
strategies. This enhancement allows users to seamlessly combine
full-text search with vector search, enabling the following search
strategies:
* Search solely by vector similarity.
* Conduct searches exclusively based on text similarity, utilizing
Lucene internally.
* Filter search results by text similarity score, with the option to
specify a threshold, followed by a search based on vector similarity.
* Filter results by vector similarity score before conducting a search
based on text similarity.
* Perform searches using a weighted sum of vector and text similarity
scores.
Additionally, integration tests have been added to comprehensively cover
all scenarios.
Updated notebook with examples.
CC: @baskaryan, @hwchase17
---------
Co-authored-by: Volodymyr Tkachuk <vtkachuk-ua@singlestore.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
- added guard on the `pyTigerGraph` import
- added a missed example page in the `docs/integrations/graphs/`
- formatted the `docs/integrations/providers/` page to the consistent
format. Added links.
- **Description:**
This PR adds support for advanced filtering to the integration of HANA
Vector Engine.
The newly supported filtering operators are: $eq, $ne, $gt, $gte, $lt,
$lte, $between, $in, $nin, $like, $and, $or
- **Issue:** N/A
- **Dependencies:** no new dependencies added
Added integration tests to:
`libs/community/tests/integration_tests/vectorstores/test_hanavector.py`
Description of the new capabilities in notebook:
`docs/docs/integrations/vectorstores/hanavector.ipynb`
Thank you for contributing to LangChain!
community:perplexity[patch]: standardize init args
updated pplx_api_key and request_timeout so that aliased to api_key, and
timeout respectively. Added test that both continue to set the same
underlying attributes.
Related to
[20085](https://github.com/langchain-ai/langchain/issues/20085)
---------
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Thank you for contributing to LangChain!
- [x] **PR title**: docs: Update Zep Messaging, add links to Zep Cloud
Docs
- [x] **PR message**:
- **Description:** This PR updates Zep messaging in the docs + links to
Langchain Zep Cloud examples in our documentation
- **Twitter handle:** @paulpaliychuk51
- [x] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [x] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
This PR moves the interface and the logic to core.
The following changes to namespaces:
`indexes` -> `indexing`
`indexes._api` -> `indexing.api`
Testing code is intentionally duplicated for now since it's testing
different
implementations of the record manager (in-memory vs. SQL).
Common logic will need to be pulled out into the test client.
A follow up PR will move the SQL based implementation outside of
LangChain.
**Description:**
This PR fixes an issue in message formatting function for Anthropic
models on Amazon Bedrock.
Currently, LangChain BedrockChat model will crash if it uses Anthropic
models and the model return a message in the following type:
- `AIMessageChunk`
Moreover, when use BedrockChat with for building Agent, the following
message types will trigger the same issue too:
- `HumanMessageChunk`
- `FunctionMessage`
**Issue:**
https://github.com/langchain-ai/langchain/issues/18831
**Dependencies:**
No.
**Testing:**
Manually tested. The following code was failing before the patch and
works after.
```
@tool
def square_root(x: str):
"Useful when you need to calculate the square root of a number"
return math.sqrt(int(x))
llm = ChatBedrock(
model_id="anthropic.claude-3-sonnet-20240229-v1:0",
model_kwargs={ "temperature": 0.0 },
)
prompt = ChatPromptTemplate.from_messages(
[
("system", FUNCTION_CALL_PROMPT),
("human", "Question: {user_input}"),
MessagesPlaceholder(variable_name="agent_scratchpad"),
]
)
tools = [square_root]
tools_string = format_tool_to_anthropic_function(square_root)
agent = (
RunnablePassthrough.assign(
user_input=lambda x: x['user_input'],
agent_scratchpad=lambda x: format_to_openai_function_messages(
x["intermediate_steps"]
)
)
| prompt
| llm
| AnthropicFunctionsAgentOutputParser()
)
agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True, return_intermediate_steps=True)
output = agent_executor.invoke({
"user_input": "What is the square root of 2?",
"tools_string": tools_string,
})
```
List of messages returned from Bedrock:
```
<SystemMessage> content='You are a helpful assistant.'
<HumanMessage> content='Question: What is the square root of 2?'
<AIMessageChunk> content="Okay, let's calculate the square root of 2.<scratchpad>\nTo calculate the square root of a number, I can use the square_root tool:\n\n<function_calls>\n <invoke>\n <tool_name>square_root</tool_name>\n <parameters>\n <__arg1>2</__arg1>\n </parameters>\n </invoke>\n</function_calls>\n</scratchpad>\n\n<function_results>\n<search_result>\nThe square root of 2 is approximately 1.414213562373095\n</search_result>\n</function_results>\n\n<answer>\nThe square root of 2 is approximately 1.414213562373095\n</answer>" id='run-92363df7-eff6-4849-bbba-fa16a1b2988c'"
<FunctionMessage> content='1.4142135623730951' name='square_root'
```
Hi! My name is Alex, I'm an SDK engineer from
[Comet](https://www.comet.com/site/)
This PR updates the `CometTracer` class.
Fixed an issue when `CometTracer` failed while logging the data to Comet
because this data is not JSON-encodable.
The problem was in some of the `Run` attributes that could contain
non-default types inside, now these attributes are taken not from the
run instance, but from the `run.dict()` return value.
Causes an issue for this code
```python
from langchain.chat_models.openai import ChatOpenAI
from langchain.output_parsers.openai_tools import JsonOutputToolsParser
from langchain.schema import SystemMessage
prompt = SystemMessage(content="You are a nice assistant.") + "{question}"
llm = ChatOpenAI(
model_kwargs={
"tools": [
{
"type": "function",
"function": {
"name": "web_search",
"description": "Searches the web for the answer to the question.",
"parameters": {
"type": "object",
"properties": {
"query": {
"type": "string",
"description": "The question to search for.",
},
},
},
},
}
],
},
streaming=True,
)
parser = JsonOutputToolsParser(first_tool_only=True)
llm_chain = prompt | llm | parser | (lambda x: x)
for chunk in llm_chain.stream({"question": "tell me more about turtles"}):
print(chunk)
# message = llm_chain.invoke({"question": "tell me more about turtles"})
# print(message)
```
Instead by definition, we'll assume that RunnableLambdas consume the
entire stream and that if the stream isn't addable then it's the last
message of the stream that's in the usable format.
---
If users want to use addable dicts, they can wrap the dict in an
AddableDict class.
---
Likely, need to follow up with the same change for other places in the
code that do the upgrade
- **Description:** In January, Laiyer.ai became part of ProtectAI, which
means the model became owned by ProtectAI. In addition to that,
yesterday, we released a new version of the model addressing issues the
Langchain's community and others mentioned to us about false-positives.
The new model has a better accuracy compared to the previous version,
and we thought the Langchain community would benefit from using the
[latest version of the
model](https://huggingface.co/protectai/deberta-v3-base-prompt-injection-v2).
- **Issue:** N/A
- **Dependencies:** N/A
- **Twitter handle:** @alex_yaremchuk
This PR moves the implementations for chat history to core. So it's
easier to determine which dependencies need to be broken / add
deprecation warnings
Fixed an error in the sample code to ensure that the code can run
directly.
Thank you for contributing to LangChain!
- [ ] **PR title**: "package: description"
- Where "package" is whichever of langchain, community, core,
experimental, etc. is being modified. Use "docs: ..." for purely docs
changes, "templates: ..." for template changes, "infra: ..." for CI
changes.
- Example: "community: add foobar LLM"
- [ ] **PR message**: ***Delete this entire checklist*** and replace
with
- **Description:** a description of the change
- **Issue:** the issue # it fixes, if applicable
- **Dependencies:** any dependencies required for this change
- **Twitter handle:** if your PR gets announced, and you'd like a
mention, we'll gladly shout you out!
- [ ] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [ ] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
langchain_community.document_loaders depricated
new langchain_google_community
Thank you for contributing to LangChain!
- [ ] **PR title**: "package: description"
- Where "package" is whichever of langchain, community, core,
experimental, etc. is being modified. Use "docs: ..." for purely docs
changes, "templates: ..." for template changes, "infra: ..." for CI
changes.
- Example: "community: add foobar LLM"
- [ ] **PR message**: ***Delete this entire checklist*** and replace
with
- **Description:** a description of the change
- **Issue:** the issue # it fixes, if applicable
- **Dependencies:** any dependencies required for this change
- **Twitter handle:** if your PR gets announced, and you'd like a
mention, we'll gladly shout you out!
- [ ] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [ ] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
docs: Fix link for `partition_pdf` in Semi_Structured_RAG.ipynb cookbook
- **Description:** Fix incorrect link to unstructured-io `partition_pdf`
section
Vector indexes in ClickHouse are experimental at the moment and can
sometimes break/change behaviour. So this PR makes it possible to say
that you don't want to specify an index type.
Any queries against the embedding column will be brute force/linear
scan, but that gives reasonable performance for small-medium dataset
sizes.
---------
Co-authored-by: Erick Friis <erick@langchain.dev>
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Thank you for contributing to LangChain!
- [ ] **PR title**: "docs: added a description of differences
langchain_google_genai vs langchain_google_vertexai"
- [ ]
- **Description:** added a description of differences
langchain_google_genai vs langchain_google_vertexai
**Description:** implemented GraphStore class for Apache Age graph db
**Dependencies:** depends on psycopg2
Unit and integration tests included. Formatting and linting have been
run.
---------
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
Update Neo4j Cypher templates to use function callback to pass context
instead of passing it in user prompt.
Co-authored-by: Erick Friis <erick@langchain.dev>
**Description:** This pull request removes a duplicated `--quiet` flag
in the pip install command found in the LangSmith Walkthrough section of
the documentation.
**Issue:** N/A
**Dependencies:** None
Thank you for contributing to LangChain!
- [ ] **PR title**: "package: docs"
- [ ] **PR message**:
- **Description:** Updated Tutorials for Vertex Vector Search
- **Issue:** NA
- **Dependencies:** NA
- **Twitter handle:** if your PR gets announced, and you'd like a
mention, we'll gladly shout you out!
@lkuligin for review
---------
Co-authored-by: adityarane@google.com <adityarane@google.com>
Co-authored-by: Leonid Kuligin <lkuligin@yandex.ru>
Co-authored-by: Chester Curme <chester.curme@gmail.com>
This pull request corrects a mistake in the variable name within the
example code. The variable doc_schema has been changed to dog_schema to
fix the error.
Description: you don't need to pass a version for Replicate official
models. That was broken on LangChain until now!
You can now run:
```
llm = Replicate(
model="meta/meta-llama-3-8b-instruct",
model_kwargs={"temperature": 0.75, "max_length": 500, "top_p": 1},
)
prompt = """
User: Answer the following yes/no question by reasoning step by step. Can a dog drive a car?
Assistant:
"""
llm(prompt)
```
I've updated the replicate.ipynb to reflect that.
twitter: @charliebholtz
---------
Co-authored-by: Erick Friis <erick@langchain.dev>
ZhipuAI API only accepts `temperature` parameter between `(0, 1)` open
interval, and if `0` is passed, it responds with status code `400`.
However, 0 and 1 is often accepted by other APIs, for example, OpenAI
allows `[0, 2]` for temperature closed range.
This PR truncates temperature parameter passed to `[0.01, 0.99]` to
improve the compatibility between langchain's ecosystem's and ZhipuAI
(e.g., ragas `evaluate` often generates temperature 0, which results in
a lot of 400 invalid responses). The PR also truncates `top_p` parameter
since it has the same restriction.
Reference: [glm-4 doc](https://open.bigmodel.cn/dev/api#glm-4) (which
unfortunately is in Chinese though).
---------
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
faster-whisper is a reimplementation of OpenAI's Whisper model using
CTranslate2, which is up to 4 times faster than enai/whisper for the
same accuracy while using less memory. The efficiency can be further
improved with 8-bit quantization on both CPU and GPU.
It can automatically detect the following 14 languages and transcribe
the text into their respective languages: en, zh, fr, de, ja, ko, ru,
es, th, it, pt, vi, ar, tr.
The gitbub repository for faster-whisper is :
https://github.com/SYSTRAN/faster-whisper
---------
Co-authored-by: Eugene Yurtsev <eyurtsev@gmail.com>
VSDX data contains EMF files. Some of these apparently can contain
exploits with some Adobe tools.
This is likely a false positive from antivirus software, but we
can remove it nonetheless.
Hey @eyurtsev, I noticed that the notebook isn't displaying the outputs
properly. I've gone ahead and rerun the cells to ensure that readers can
easily understand the functionality without having to run the code
themselves.
Replaced `from langchain.prompts` with `from langchain_core.prompts`
where it is appropriate.
Most of the changes go to `langchain_experimental`
Similar to #20348
…gFaceTextGenInference)
- [x] **PR title**: community[patch]: Invoke callback prior to yielding
token fix for [HuggingFaceTextGenInference]
- [x] **PR message**:
- **Description:** Invoke callback prior to yielding token in stream
method in [HuggingFaceTextGenInference]
- **Issue:** https://github.com/langchain-ai/langchain/issues/16913
- **Dependencies:** None
- **Twitter handle:** @bolun_zhang
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
---------
Co-authored-by: Chester Curme <chester.curme@gmail.com>
fix timeout issue
fix zhipuai usecase notebookbook
Thank you for contributing to LangChain!
- [ ] **PR title**: "package: description"
- Where "package" is whichever of langchain, community, core,
experimental, etc. is being modified. Use "docs: ..." for purely docs
changes, "templates: ..." for template changes, "infra: ..." for CI
changes.
- Example: "community: add foobar LLM"
- [ ] **PR message**: ***Delete this entire checklist*** and replace
with
- **Description:** a description of the change
- **Issue:** the issue # it fixes, if applicable
- **Dependencies:** any dependencies required for this change
- **Twitter handle:** if your PR gets announced, and you'd like a
mention, we'll gladly shout you out!
- [ ] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [ ] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
fixed broken `LangGraph` hyperlink
Thank you for contributing to LangChain!
- [ ] **PR title**: "package: description"
- Where "package" is whichever of langchain, community, core,
experimental, etc. is being modified. Use "docs: ..." for purely docs
changes, "templates: ..." for template changes, "infra: ..." for CI
changes.
- Example: "community: add foobar LLM"
- [ ] **PR message**: ***Delete this entire checklist*** and replace
with
- **Description:** a description of the change
- **Issue:** the issue # it fixes, if applicable
- **Dependencies:** any dependencies required for this change
- **Twitter handle:** if your PR gets announced, and you'd like a
mention, we'll gladly shout you out!
- [ ] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [ ] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
Thank you for contributing to LangChain!
- [x] **PR title**: "package: description"
- Where "package" is whichever of langchain, community, core,
experimental, etc. is being modified. Use "docs: ..." for purely docs
changes, "templates: ..." for template changes, "infra: ..." for CI
changes.
- Example: "community: add foobar LLM"
- [ ] **PR message**: ***Delete this entire checklist*** and replace
with
- **Description:** a description of the change
- **Issue:** the issue # it fixes, if applicable
- **Dependencies:** any dependencies required for this change
- **Twitter handle:** if your PR gets announced, and you'd like a
mention, we'll gladly shout you out!
- [ ] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [ ] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
@rgupta2508 I believe this change is necessary following
https://github.com/langchain-ai/langchain/pull/20318 because of how
Milvus handles defaults:
59bf5e811a/pymilvus/client/prepare.py (L82-L85)
```python
num_shards = kwargs[next(iter(same_key))]
if not isinstance(num_shards, int):
msg = f"invalid num_shards type, got {type(num_shards)}, expected int"
raise ParamError(message=msg)
req.shards_num = num_shards
```
this way lets Milvus control the default value (instead of maintaining a
separate default in Langchain).
Let me know if I've got this wrong or you feel it's unnecessary. Thanks.
To support number of the shards for the collection to create in milvus
vvectorstores.
Thank you for contributing to LangChain!
- [ ] **PR title**: "package: description"
- Where "package" is whichever of langchain, community, core,
experimental, etc. is being modified. Use "docs: ..." for purely docs
changes, "templates: ..." for template changes, "infra: ..." for CI
changes.
- Example: "community: add foobar LLM"
- [ ] **PR message**: ***Delete this entire checklist*** and replace
with
- **Description:** a description of the change
- **Issue:** the issue # it fixes, if applicable
- **Dependencies:** any dependencies required for this change
- **Twitter handle:** if your PR gets announced, and you'd like a
mention, we'll gladly shout you out!
- [ ] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [ ] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
**Description:** Move `FileCallbackHandler` from community to core
**Issue:** #20493
**Dependencies:** None
(imo) `FileCallbackHandler` is a built-in LangChain callback handler
like `StdOutCallbackHandler` and should properly be in in core.
- **Description:** added the headless parameter as optional argument to
the langchain_community.document_loaders AsyncChromiumLoader class
- **Dependencies:** None
- **Twitter handle:** @perinim_98
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
---------
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
- would happen when user's code tries to access attritbute that doesnt
exist, we prefer to let this crash in the user's code, rather than here
- also catch more cases where a runnable is invoked/streamed inside a
lambda. before we weren't seeing these as deps
**Description:** currently, the `DirectoryLoader` progress-bar maximum value is based on an incorrect number of files to process
In langchain_community/document_loaders/directory.py:127:
```python
paths = p.rglob(self.glob) if self.recursive else p.glob(self.glob)
items = [
path
for path in paths
if not (self.exclude and any(path.match(glob) for glob in self.exclude))
]
```
`paths` returns both files and directories. `items` is later used to determine the maximum value of the progress-bar which gives an incorrect progress indication.
- Add functions (_stream, _astream)
- Connect to _generate and _agenerate
Thank you for contributing to LangChain!
- [x] **PR title**: "community: Add streaming logic in ChatHuggingFace"
- [x] **PR message**: ***Delete this entire checklist*** and replace
with
- **Description:** Addition functions (_stream, _astream) and connection
to _generate and _agenerate
- **Issue:** #18782
- **Dependencies:** none
- **Twitter handle:** @lunara_x
**Community: Unify Titan Takeoff Integrations and Adding Embedding
Support**
**Description:**
Titan Takeoff no longer reflects this either of the integrations in the
community folder. The two integrations (TitanTakeoffPro and
TitanTakeoff) where causing confusion with clients, so have moved code
into one place and created an alias for backwards compatibility. Added
Takeoff Client python package to do the bulk of the work with the
requests, this is because this package is actively updated with new
versions of Takeoff. So this integration will be far more robust and
will not degrade as badly over time.
**Issue:**
Fixes bugs in the old Titan integrations and unified the code with added
unit test converge to avoid future problems.
**Dependencies:**
Added optional dependency takeoff-client, all imports still work without
dependency including the Titan Takeoff classes but just will fail on
initialisation if not pip installed takeoff-client
**Twitter**
@MeryemArik9
Thanks all :)
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Description: Add support for authorized identities in PebbloSafeLoader.
Now with this change, PebbloSafeLoader will extract
authorized_identities from metadata and send it to pebblo server
Dependencies: None
Documentation: None
Signed-off-by: Rahul Tripathi <rauhl.psit.ec@gmail.com>
Co-authored-by: Rahul Tripathi <rauhl.psit.ec@gmail.com>
From `langchain_community 0.0.30`, there's a bug that cannot send a
file-like object via `file` parameter instead of `file path` due to
casting the `file_path` to str type even if `file_path` is None.
which means that when I call the `partition_via_api()`, exactly one of
`filename` and `file` must be specified by the following error message.
however, from `langchain_community 0.0.30`, `file_path` is casted into
`str` type even `file_path` is None in `get_elements_from_api()` and got
an error at `exactly_one(filename=filename, file=file)`.
here's an error message
```
---> 51 exactly_one(filename=filename, file=file)
53 if metadata_filename and file_filename:
54 raise ValueError(
55 "Only one of metadata_filename and file_filename is specified. "
56 "metadata_filename is preferred. file_filename is marked for deprecation.",
57 )
File /opt/homebrew/lib/python3.11/site-packages/unstructured/partition/common.py:441, in exactly_one(**kwargs)
439 else:
440 message = f"{names[0]} must be specified."
--> 441 raise ValueError(message)
ValueError: Exactly one of filename and file must be specified.
```
So, I simply made a change that casting to str type when `file_path` is
not None.
I use `UnstructuredAPIFileLoader` like below.
```
from langchain_community.document_loaders.unstructured import UnstructuredAPIFileLoader
documents: list = UnstructuredAPIFileLoader(
file_path=None,
file=file, # file-like object, io.BytesIO type
mode='elements',
url='http://127.0.0.1:8000/general/v0/general',
content_type='application/pdf',
metadata_filename='asdf.pdf',
).load_and_split()
```
- [x] **PR title**: "community: improve kuzu cypher generation prompt"
- [x] **PR message**: ***Delete this entire checklist*** and replace
with
- **Description:** Improves the Kùzu Cypher generation prompt to be more
robust to open source LLM outputs
- **Issue:** N/A
- **Dependencies:** N/A
- **Twitter handle:** @kuzudb
- [x] **Add tests and docs**: If you're adding a new integration, please
include
No new tests (non-breaking. change)
- [x] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
## Description:
The PR introduces 3 changes:
1. added `recursive` property to `O365BaseLoader`. (To keep the behavior
unchanged, by default is set to `False`). When `recursive=True`,
`_load_from_folder()` also recursively loads all nested folders.
2. added `folder_id` to SharePointLoader.(similar to (this
PR)[https://github.com/langchain-ai/langchain/pull/10780] ) This
provides an alternative to `folder_path` that doesn't seem to reliably
work.
3. when none of `document_ids`, `folder_id`, `folder_path` is provided,
the loader fetches documets from root folder. Combined with
`recursive=True` this provides an easy way of loading all compatible
documents from SharePoint.
The PR contains the same logic as [this stale
PR](https://github.com/langchain-ai/langchain/pull/10780) by
@WaleedAlfaris. I'd like to ask his blessing for moving forward with
this one.
## Issue:
- As described in https://github.com/langchain-ai/langchain/issues/19938
and https://github.com/langchain-ai/langchain/pull/10780 the sharepoint
loader often does not seem to work with folder_path.
- Recursive loading of subfolders is a missing functionality
## Dependecies: None
Twitter handle:
@martintriska1 @WRhetoric
This is my first PR here, please be gentle :-)
Please review @baskaryan
---------
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
This PR updates OctoAIEndpoint LLM to subclass BaseOpenAI as OctoAI is
an OpenAI-compatible service. The documentation and tests have also been
updated.
**Description:** Adds ThirdAI NeuralDB retriever integration. NeuralDB
is a CPU-friendly and fine-tunable text retrieval engine. We previously
added a vector store integration but we think that it will be easier for
our customers if they can also find us under under
langchain-community/retrievers.
---------
Co-authored-by: kartikTAI <129414343+kartikTAI@users.noreply.github.com>
Co-authored-by: Kartik Sarangmath <kartik@thirdai.com>
**Description:** Make ChatDatabricks model supports stream
**Issue:** N/A
**Dependencies:** MLflow nightly build version (we will release next
MLflow version soon)
**Twitter handle:** N/A
Manually test:
(Before testing, please install `pip install
git+https://github.com/mlflow/mlflow.git`)
```python
# Test Databricks Foundation LLM model
from langchain.chat_models import ChatDatabricks
chat_model = ChatDatabricks(
endpoint="databricks-llama-2-70b-chat",
max_tokens=500
)
from langchain_core.messages import AIMessageChunk
for chunk in chat_model.stream("What is mlflow?"):
print(chunk.content, end="|")
```
- [x] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [x] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
---------
Signed-off-by: Weichen Xu <weichen.xu@databricks.com>
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
- Add conditional: bool property to json representation of the graphs
- Add option to generate mermaid graph stripped of styles (useful as a
text representation of graph)
…s arg too
Thank you for contributing to LangChain!
- [ ] **PR title**: "package: description"
- Where "package" is whichever of langchain, community, core,
experimental, etc. is being modified. Use "docs: ..." for purely docs
changes, "templates: ..." for template changes, "infra: ..." for CI
changes.
- Example: "community: add foobar LLM"
- [ ] **PR message**: ***Delete this entire checklist*** and replace
with
- **Description:** a description of the change
- **Issue:** the issue # it fixes, if applicable
- **Dependencies:** any dependencies required for this change
- **Twitter handle:** if your PR gets announced, and you'd like a
mention, we'll gladly shout you out!
- [ ] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [ ] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
- **Description:**
This PR adds a callback handler for UpTrain. It performs evaluations in
the RAG pipeline to check the quality of retrieved documents, generated
queries and responses.
- **Dependencies:**
- The UpTrainCallbackHandler requires the uptrain package
---------
Co-authored-by: Eugene Yurtsev <eugene@langchain.dev>
enviroment variable ANTHROPIC_API_URL will not work if anthropic_api_url
has default value
---------
Co-authored-by: Eugene Yurtsev <eugene@langchain.dev>
**Description**: Support filter by OR and AND for deprecated PGVector
version
**Issue**: #20445
**Dependencies**: N/A
**Twitter** handle: @martinferenaz
- **Description:**Add Google Firestore Vector store docs
- **Issue:** NA
- **Dependencies:** NA
---------
Co-authored-by: Erick Friis <erick@langchain.dev>
- **Description:** a description of the change
- **Issue:** the issue # it fixes, if applicable
- **Dependencies:** any dependencies required for this change
- **Twitter handle:** if your PR gets announced, and you'd like a
mention, we'll gladly shout you out!
Description: fixes LangChainDeprecationWarning: The class
`langchain_community.embeddings.cohere.CohereEmbeddings` was deprecated
in langchain-community 0.0.30 and will be removed in 0.2.0. An updated
version of the class exists in the langchain-cohere package and should
be used instead. To use it run `pip install -U langchain-cohere` and
import as `from langchain_cohere import CohereEmbeddings`.
Dependencies : langchain_cohere
Twitter handle: @Mo_Noumaan
Description of features on mermaid graph renderer:
- Fixing CDN to use official Mermaid JS CDN:
https://www.jsdelivr.com/package/npm/mermaid?tab=files
- Add device_scale_factor to allow increasing quality of resulting PNG.
- [x] **PR title**: community[patch]: Invoke callback prior to yielding
token fix for [DeepInfra]
- [x] **PR message**:
- **Description:** Invoke callback prior to yielding token in stream
method in [DeepInfra]
- **Issue:** https://github.com/langchain-ai/langchain/issues/16913
- **Dependencies:** None
- **Twitter handle:** @bolun_zhang
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
Description: This update refines the documentation for
`RunnablePassthrough` by removing an unnecessary import and correcting a
minor syntactical error in the example provided. This change enhances
the clarity and correctness of the documentation, ensuring that users
have a more accurate guide to follow.
Issue: N/A
Dependencies: None
This PR focuses solely on documentation improvements, specifically
targeting the `RunnablePassthrough` class within the `langchain_core`
module. By clarifying the example provided in the docstring, users are
offered a more straightforward and error-free guide to utilizing the
`RunnablePassthrough` class effectively.
As this is a documentation update, it does not include changes that
require new integrations, tests, or modifications to dependencies. It
adheres to the guidelines of minimal package interference and backward
compatibility, ensuring that the overall integrity and functionality of
the LangChain package remain unaffected.
Thank you for considering this documentation refinement for inclusion in
the LangChain project.
Fix of YandexGPT embeddings.
The current version uses a single `model_name` for queries and
documents, essentially making the `embed_documents` and `embed_query`
methods the same. Yandex has a different endpoint (`model_uri`) for
encoding documents, see
[this](https://yandex.cloud/en/docs/yandexgpt/concepts/embeddings). The
bug may impact retrievers built with `YandexGPTEmbeddings` (for instance
FAISS database as retriever) since they use both `embed_documents` and
`embed_query`.
A simple snippet to test the behaviour:
```python
from langchain_community.embeddings.yandex import YandexGPTEmbeddings
embeddings = YandexGPTEmbeddings()
q_emb = embeddings.embed_query('hello world')
doc_emb = embeddings.embed_documents(['hello world', 'hello world'])
q_emb == doc_emb[0]
```
The response is `True` with the current version and `False` with the
changes I made.
Twitter: @egor_krash
---------
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
**Description:** Updates the documentation for Portkey and Langchain.
Also updates the notebook. The current documentation is fairly old and
is non-functional.
**Twitter handle:** @portkeyai
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
**Description:**
`_ListSQLDatabaseToolInput` raise error if model returns `{}`.
For example, gpt-4-turbo returns `{}` with SQL Agent initialized by
`create_sql_agent`.
So, I set default value `""` for `_ListSQLDatabaseToolInput` tool_input.
This is actually a gpt-4-turbo issue, not a LangChain issue, but I
thought it would be helpful to set a default value `""`.
This problem is discussed in detail in the following Issue.
**Issue:** https://github.com/langchain-ai/langchain/issues/20405
**Dependencies:** none
Sorry, I did not add or change the test code, as tests for this
components was not exist .
However, I have tested the following code based on the [SQL Agent
Document](https://python.langchain.com/docs/use_cases/sql/agents/), to
make sure it works.
```
from langchain_community.agent_toolkits.sql.base import create_sql_agent
from langchain_community.utilities.sql_database import SQLDatabase
from langchain_openai import ChatOpenAI
db = SQLDatabase.from_uri("sqlite:///Chinook.db")
llm = ChatOpenAI(model="gpt-4-turbo", temperature=0)
agent_executor = create_sql_agent(llm, db=db, agent_type="openai-tools", verbose=True)
result = agent_executor.invoke("List the total sales per country. Which country's customers spent the most?")
print(result["output"])
```
- **Description:** Complete the support for Lua code in
langchain.text_splitter module.
- **Dependencies:** No
- **Twitter handle:** @saberuster
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
```python
from langchain.agents import AgentExecutor, create_tool_calling_agent, tool
from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder
from langchain_groq import ChatGroq
prompt = ChatPromptTemplate.from_messages(
[
("system", "You are a helpful assistant"),
("human", "{input}"),
MessagesPlaceholder("agent_scratchpad"),
]
)
model = ChatGroq(model_name="mixtral-8x7b-32768", temperature=0)
@tool
def magic_function(input: int) -> int:
"""Applies a magic function to an input."""
return input + 2
tools = [magic_function]
agent = create_tool_calling_agent(model, tools, prompt)
agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True)
agent_executor.invoke({"input": "what is the value of magic_function(3)?"})
```
```
> Entering new AgentExecutor chain...
Invoking: `magic_function` with `{'input': 3}`
5The value of magic\_function(3) is 5.
> Finished chain.
{'input': 'what is the value of magic_function(3)?',
'output': 'The value of magic\\_function(3) is 5.'}
```
**Description:** Masking of the API key for AI21 models
**Issue:** Fixes#12165 for AI21
**Dependencies:** None
Note: This fix came in originally through #12418 but was possibly missed
in the refactor to the AI21 partner package
---------
Co-authored-by: Erick Friis <erick@langchain.dev>
Replaced all `from langchain.callbacks` into `from
langchain_core.callbacks` .
Changes in the `langchain` and `langchain_experimental`
---------
Co-authored-by: Erick Friis <erick@langchain.dev>
- **Description**: The pydantic schema fields are supposed to be
optional but the use of `...` makes them required. This causes a
`ValidationError` when running the example code. I replaced `...` with
`default=None` to make the fields optional as intended. I also
standardized the format for all fields.
- **Issue**: n/a
- **Dependencies**: none
- **Twitter handle**: https://twitter.com/m_atoms
---------
Co-authored-by: Chester Curme <chester.curme@gmail.com>
- [x] **PR title**: community[patch]: Invoke callback prior to yielding
token fix for Llamafile
- [x] **PR message**:
- **Description:** Invoke callback prior to yielding token in stream
method in community llamafile.py
- **Issue:** https://github.com/langchain-ai/langchain/issues/16913
- **Dependencies:** None
- **Twitter handle:** @bolun_zhang
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
spelling error fixed
Thank you for contributing to LangChain!
- [ ] **PR title**: "package: description"
- Where "package" is whichever of langchain, community, core,
experimental, etc. is being modified. Use "docs: ..." for purely docs
changes, "templates: ..." for template changes, "infra: ..." for CI
changes.
- Example: "community: add foobar LLM"
- [ ] **PR message**: ***Delete this entire checklist*** and replace
with
- **Description:** a description of the change
- **Issue:** the issue # it fixes, if applicable
- **Dependencies:** any dependencies required for this change
- **Twitter handle:** if your PR gets announced, and you'd like a
mention, we'll gladly shout you out!
- [ ] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [ ] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
- [x] **PR title**: community[patch]: Invoke callback prior to yielding
token fix for HuggingFaceEndpoint
- [x] **PR message**:
- **Description:** Invoke callback prior to yielding token in stream
method in community HuggingFaceEndpoint
- **Issue:** https://github.com/langchain-ai/langchain/issues/16913
- **Dependencies:** None
- **Twitter handle:** @bolun_zhang
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
---------
Co-authored-by: Chester Curme <chester.curme@gmail.com>
Added the [FireCrawl](https://firecrawl.dev) document loader. Firecrawl
crawls and convert any website into LLM-ready data. It crawls all
accessible subpages and give you clean markdown for each.
- **Description:** Adds FireCrawl data loader
- **Dependencies:** firecrawl-py
- **Twitter handle:** @mendableai
ccing contributors: (@ericciarla @nickscamara)
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
LLMs might sometimes return invalid response for LLM graph transformer.
Instead of failing due to pydantic validation, we skip it and manually
check and optionally fix error where we can, so that more information
gets extracted
- **Description:** Added cross-links for easy access of api
documentation of each output parser class from it's description page.
- **Issue:** related to issue #19969
Co-authored-by: Haris Ali <haris.ali@formulatrix.com>
avaliable -> available
- **Description:** fixed typo
- **Issue:** the issue # it fixes, if applicable
- **Dependencies:** any dependencies required for this change
- **Twitter handle:** if your PR gets announced, and you'd like a
mention, we'll gladly shout you out!
Mistral gives us one ID per response, no individual IDs for tool calls.
```python
from langchain.agents import AgentExecutor, create_tool_calling_agent, tool
from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder
from langchain_mistralai import ChatMistralAI
prompt = ChatPromptTemplate.from_messages(
[
("system", "You are a helpful assistant"),
("human", "{input}"),
MessagesPlaceholder("agent_scratchpad"),
]
)
model = ChatMistralAI(model="mistral-large-latest", temperature=0)
@tool
def magic_function(input: int) -> int:
"""Applies a magic function to an input."""
return input + 2
tools = [magic_function]
agent = create_tool_calling_agent(model, tools, prompt)
agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True)
agent_executor.invoke({"input": "what is the value of magic_function(3)?"})
```
---------
Co-authored-by: Eugene Yurtsev <eyurtsev@gmail.com>
**Description:** Adds chroma to the partners package. Tests & code
mirror those in the community package.
**Dependencies:** None
**Twitter handle:** @akiradev0x
---------
Co-authored-by: Erick Friis <erick@langchain.dev>
This PR should make it easier for linters to do type checking and for IDEs to jump to definition of code.
See #20050 as a template for this PR.
- As a byproduct: Added 3 missed `test_imports`.
- Added missed `SolarChat` in to __init___.py Added it into test_import
ut.
- Added `# type: ignore` to fix linting. It is not clear, why linting
errors appear after ^ changes.
---------
Co-authored-by: Eugene Yurtsev <eyurtsev@gmail.com>
```python
from langchain.agents import AgentExecutor, create_tool_calling_agent, tool
from langchain_anthropic import ChatAnthropic
from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder
prompt = ChatPromptTemplate.from_messages(
[
("system", "You are a helpful assistant"),
MessagesPlaceholder("chat_history", optional=True),
("human", "{input}"),
MessagesPlaceholder("agent_scratchpad"),
]
)
model = ChatAnthropic(model="claude-3-opus-20240229")
@tool
def magic_function(input: int) -> int:
"""Applies a magic function to an input."""
return input + 2
tools = [magic_function]
agent = create_tool_calling_agent(model, tools, prompt)
agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True)
agent_executor.invoke({"input": "what is the value of magic_function(3)?"})
```
```
> Entering new AgentExecutor chain...
Invoking: `magic_function` with `{'input': 3}`
responded: [{'text': '<thinking>\nThe user has asked for the value of magic_function applied to the input 3. Looking at the available tools, magic_function is the relevant one to use here, as it takes an integer input and returns an integer output.\n\nThe magic_function has one required parameter:\n- input (integer)\n\nThe user has directly provided the value 3 for the input parameter. Since the required parameter is present, we can proceed with calling the function.\n</thinking>', 'type': 'text'}, {'id': 'toolu_01HsTheJPA5mcipuFDBbJ1CW', 'input': {'input': 3}, 'name': 'magic_function', 'type': 'tool_use'}]
5
Therefore, the value of magic_function(3) is 5.
> Finished chain.
{'input': 'what is the value of magic_function(3)?',
'output': 'Therefore, the value of magic_function(3) is 5.'}
```
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
core[minor], langchain[patch], openai[minor], anthropic[minor], fireworks[minor], groq[minor], mistralai[minor]
```python
class ToolCall(TypedDict):
name: str
args: Dict[str, Any]
id: Optional[str]
class InvalidToolCall(TypedDict):
name: Optional[str]
args: Optional[str]
id: Optional[str]
error: Optional[str]
class ToolCallChunk(TypedDict):
name: Optional[str]
args: Optional[str]
id: Optional[str]
index: Optional[int]
class AIMessage(BaseMessage):
...
tool_calls: List[ToolCall] = []
invalid_tool_calls: List[InvalidToolCall] = []
...
class AIMessageChunk(AIMessage, BaseMessageChunk):
...
tool_call_chunks: Optional[List[ToolCallChunk]] = None
...
```
Important considerations:
- Parsing logic occurs within different providers;
- ~Changing output type is a breaking change for anyone doing explicit
type checking;~
- ~Langsmith rendering will need to be updated:
https://github.com/langchain-ai/langchainplus/pull/3561~
- ~Langserve will need to be updated~
- Adding chunks:
- ~AIMessage + ToolCallsMessage = ToolCallsMessage if either has
non-null .tool_calls.~
- Tool call chunks are appended, merging when having equal values of
`index`.
- additional_kwargs accumulate the normal way.
- During streaming:
- ~Messages can change types (e.g., from AIMessageChunk to
AIToolCallsMessageChunk)~
- Output parsers parse additional_kwargs (during .invoke they read off
tool calls).
Packages outside of `partners/`:
- https://github.com/langchain-ai/langchain-cohere/pull/7
- https://github.com/langchain-ai/langchain-google/pull/123/files
---------
Co-authored-by: Chester Curme <chester.curme@gmail.com>
Description: When multithreading is set to True and using the
DirectoryLoader, there was a bug that caused the return type to be a
double nested list. This resulted in other places upstream not being
able to utilize the from_documents method as it was no longer a
`List[Documents]` it was a `List[List[Documents]]`. The change made was
to just loop through the `future.result()` and yield every item.
Issue: #20093
Dependencies: N/A
Twitter handle: N/A
This unit test fails likely validation by the openai client.
Newer openai library seems to be doing more validation so the existing
test fails since http_client needs to be of httpx instance
- **Description**: fixes BooleanOutputParser detecting sub-words ("NOW
this is likely (YES)" -> `True`, not `AmbiguousError`)
- **Issue(s)**: fixes#11408 (follow-up to #17810)
- **Dependencies**: None
- **GitHub handle**: @casperdcl
<!-- if unreviewd after a few days, @-mention one of baskaryan, efriis,
eyurtsev, hwchase17 -->
- [x] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [ ] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
---------
Co-authored-by: Eugene Yurtsev <eyurtsev@gmail.com>
**Description:**
Use the `Stream` context managers in `ChatOpenAi` `stream` and `astream`
method.
Using the context manager returned by the OpenAI client makes it
possible to terminate the stream early since the response connection
will be closed when the context manager exists.
**Issue:** #5340
**Twitter handle:** @snopoke
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
- **Description:** Bug fix. Removed extra line in `GCSDirectoryLoader`
to allow catching Exceptions. Now also logs the file path if Exception
is raised for easier debugging.
- **Issue:** #20198 Bug since langchain-community==0.0.31
- **Dependencies:** No change
- **Twitter handle:** timothywong731
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
- make Tencent Cloud VectorDB support metadata filtering.
- implement delete function for Tencent Cloud VectorDB.
- support both Langchain Embedding model and Tencent Cloud VDB embedding
model.
- Tencent Cloud VectorDB support filter search keyword, compatible with
langchain filtering syntax.
- add Tencent Cloud VectorDB TranslationVisitor, now work with self
query retriever.
- more documentations.
---------
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
- **Description:** In this PR I fixed the links which points to the API
docs for classes in OpenAI functions and OpenAI tools section of output
parsers.
- **Issue:** It fixed the issue #19969
Co-authored-by: Haris Ali <haris.ali@formulatrix.com>
Issue `langchain_community.cross_encoders` didn't have flattening
namespace code in the __init__.py file.
Changes:
- added code to flattening namespaces (used #20050 as a template)
- added ut for a change
- added missed `test_imports` for `chat_loaders` and
`chat_message_histories` modules
This PR make `request_timeout` and `max_retries` configurable for
ChatAnthropic.
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
Co-authored-by: Erick Friis <erick@langchain.dev>
Thank you for contributing to LangChain!
- [ ] **PR title**: "community: Add semantic caching and memory using
MongoDB"
- [ ] **PR message**:
- **Description:** This PR introduces functionality for adding semantic
caching and chat message history using MongoDB in RAG applications. By
leveraging the MongoDBCache and MongoDBChatMessageHistory classes,
developers can now enhance their retrieval-augmented generation
applications with efficient semantic caching mechanisms and persistent
conversation histories, improving response times and consistency across
chat sessions.
- **Issue:** N/A
- **Dependencies:** Requires `datasets`, `langchain`,
`langchain-mongodb`, `langchain-openai`, `pymongo`, and `pandas` for
implementation. MongoDB Atlas is used for database services, and the
OpenAI API for model access.
- **Twitter handle:** @richmondalake
Co-authored-by: Erick Friis <erick@langchain.dev>
Issue:
When async_req is the default value True, pinecone client return the
multiprocessing AsyncResult object.
When async_req is set to False, pinecone client return the result
directly. `[{'upserted_count': 1}]` . Calling get() method will throw an
error in this case.
- [x] **PR title**: "package: description"
- Where "package" is whichever of langchain, community, core,
experimental, etc. is being modified. Use "docs: ..." for purely docs
changes, "templates: ..." for template changes, "infra: ..." for CI
changes.
- Example: "community: add foobar LLM"
- [x] **PR message**: ***Delete this entire checklist*** and replace
with
- **Description:** Langchain-Predibase integration was failing, because
it was not current with the Predibase SDK; in addition, Predibase
integration tests were instantiating the Langchain Community `Predibase`
class with one required argument (`model`) missing. This change updates
the Predibase SDK usage and fixes the integration tests.
- **Twitter handle:** `@alexsherstinsky`
- [x] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
---------
Co-authored-by: Erick Friis <erick@langchain.dev>
Last year Microsoft [changed the
name](https://learn.microsoft.com/en-us/azure/search/search-what-is-azure-search)
of Azure Cognitive Search to Azure AI Search. This PR updates the
Langchain Azure Retriever API and it's associated docs to reflect this
change. It may be confusing for users to see the name Cognitive here and
AI in the Microsoft documentation which is why this is needed. I've also
added a more detailed example to the Azure retriever doc page.
There are more places that need a similar update but I'm breaking it up
so the PRs are not too big 😄 Fixing my errors from the previous PR.
Twitter: @marlene_zw
Two new tests added to test backward compatibility in
`libs/community/tests/integration_tests/retrievers/test_azure_cognitive_search.py`
---------
Co-authored-by: Chester Curme <chester.curme@gmail.com>
- **Description:** update langchain anthropic templates to support
Claude 3 (iterative search, chain of note, summarization, and XML
response)
- **Issue:** issue # N/A. Stability issues and errors encountered when
trying to use older langchain and anthropic libraries.
- **Dependencies:**
- langchain_anthropic version 0.1.4\
- anthropic package version in the range ">=0.17.0,<1" to support
langchain_anthropic.
- **Twitter handle:** @d_w_b7
- [ x]**Add tests and docs**: If you're adding a new integration, please
include
1. used instructions in the README for testing
- [ x] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
---------
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Co-authored-by: Erick Friis <erick@langchain.dev>
After this PR it will be possible to pass a cache instance directly to a
language model. This is useful to allow different language models to use
different caches if needed.
- **Issue:** close#19276
---------
Co-authored-by: Eugene Yurtsev <eyurtsev@gmail.com>
- Added missed providers
- Added links, descriptions in related examples
- Formatted in a consistent format
Co-authored-by: Erick Friis <erick@langchain.dev>
Updated a page with existing document loaders with links to examples.
Fixed formatting of one example.
Co-authored-by: Erick Friis <erick@langchain.dev>
Issue: The `graph` code was moved into the `community` package a long
ago. But the related documentation is still in the
[use_cases](https://python.langchain.com/docs/use_cases/graph/integrations/diffbot_graphtransformer)
section and not in the `integrations`.
Changes:
- moved the `use_cases/graph/integrations` notebooks into the
`integrations/graphs`
- renamed files and changed titles to follow the consistent format
- redirected old page URLs to new URLs in `vercel.json` and in several
other pages
- added descriptions and links when necessary
- formatted into the consistent format
Should hopefully avoid weird broken link edge cases.
Relative links now trip up the Docusaurus broken link checker, so this
PR also removes them.
Also snuck in a small addition about asyncio
**Description:**
The `LocalFileStore` class can be used to create an on-disk
`CacheBackedEmbeddings` cache. However, the default `umask` settings
gives file/directory write permissions only to the original user. Once
the cache directory is created by the first user, other users cannot
write their own cache entries into the directory.
To make the cache usable by multiple users, this pull request updates
the `LocalFileStore` constructor to allow the permissions for newly
created directories and files to be specified. The specified permissions
override the default `umask` values.
For example, when configured as follows:
```python
file_store = LocalFileStore(temp_dir, chmod_dir=0o770, chmod_file=0o660)
```
then "user" and "group" (but not "other") have permissions to access the
store, which means:
* Anyone in our group could contribute embeddings to the cache.
* If we implement cache cleanup/eviction in the future, anyone in our
group could perform the cleanup.
The default values for the `chmod_dir` and `chmod_file` parameters is
`None`, which retains the original behavior of using the default `umask`
settings.
**Issue:**
Implements enhancement #18075.
**Testing:**
I updated the `LocalFileStore` unit tests to test the permissions.
---------
Signed-off-by: chrispy <chrispy@synopsys.com>
Co-authored-by: Eugene Yurtsev <eyurtsev@gmail.com>
- **Description:** Adds async variants of afrom_texts and
afrom_embeddings into `OpenSearchVectorSearch`, which allows for
`afrom_documents` to be called.
- **Issue:** I implemented this because my use case involves an async
scraper generating documents as and when they're ready to be ingested by
Embedding/OpenSearch
- **Dependencies:** None that I'm aware
Co-authored-by: Ben Mitchell <b.mitchell@reply.com>
This PR supports using Pydantic v2 objects to generate the schema for
the JSONOutputParser (#19441). This also adds a `json_schema` parameter
to allow users to pass any JSON schema to validate with, not just
pydantic.
core/langchain_core/_api[Patch]: mypy ignore fixes#17048
Related to #17048
Applied mypy fixes to below two files:
libs/core/langchain_core/_api/deprecation.py
libs/core/langchain_core/_api/beta_decorator.py
Summary of Fixes:
**Issue 1**
class _deprecated_property(type(obj)): # type: ignore
error: Unsupported dynamic base class "type" [misc]
Fix:
1. Added an __init__ method to _deprecated_property to initialize the
fget, fset, fdel, and __doc__ attributes.
2. In the __get__, __set__, and __delete__ methods, we now use the
self.fget, self.fset, and self.fdel attributes to call the original
methods after emitting the warning.
3. The finalize function now creates an instance of _deprecated_property
with the fget, fset, fdel, and doc attributes from the original obj
property.
**Issue 2**
def finalize( # type: ignore
wrapper: Callable[..., Any], new_doc: str
) -> T:
error: All conditional function variants must have identical
signatures
Fix: Ensured that both definitions of the finalize function have the
same signature
Twitter Handle -
https://x.com/gupteutkarsha?s=11&t=uwHe4C3PPpGRvoO5Qpm1aA
**Description:** Citations are the main addition in this PR. We now emit
them from the multihop agent! Additionally the agent is now more
flexible with observations (`Any` is now accepted), and the Cohere SDK
version is bumped to fix an issue with the most recent version of
pydantic v1 (1.10.15)
- **Description:** In order to use index and aindex in
libs/langchain/langchain/indexes/_api.py, I implemented delete method
and all async methods in opensearch_vector_search
- **Dependencies:** No changes
- **Description:** Improvement for #19599: fixing missing return of
graph.draw_mermaid_png and improve it to make the saving of the rendered
image optional
Co-authored-by: Angel Igareta <angel.igareta@klarna.com>
Thank you for contributing to LangChain!
- [ ] **PR title**: "package: description"
- Where "package" is whichever of langchain, community, core,
experimental, etc. is being modified. Use "docs: ..." for purely docs
changes, "templates: ..." for template changes, "infra: ..." for CI
changes.
- Example: "community: add foobar LLM"
- [ ] **PR message**: ***Delete this entire checklist*** and replace
with
- **Description:** a description of the change
- **Issue:** the issue # it fixes, if applicable
- **Dependencies:** any dependencies required for this change
- **Twitter handle:** if your PR gets announced, and you'd like a
mention, we'll gladly shout you out!
- [ ] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [ ] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
**Description:** Update of Cohere documentation (main provider page)
**Issue:** After addition of the Cohere partner package, the
documentation was out of date
**Dependencies:** None
---------
Co-authored-by: Chester Curme <chester.curme@gmail.com>
Thank you for contributing to LangChain!
- [ ] **PR title**: "community: deprecating integrations moved to
langchain_google_community"
- [ ] **PR message**: deprecating integrations moved to
langchain_google_community
---------
Co-authored-by: ccurme <chester.curme@gmail.com>
Removes required usage of `requests` from `langchain-core`, all of which
has been deprecated.
- removes Tracer V1 implementations
- removes old `try_load_from_hub` github-based hub implementations
Removal done in a way where imports will still succeed, and usage will
fail with a `RuntimeError`.
Thank you for contributing to LangChain!
- [x] **PR title**: "package: description"
- Where "package" is whichever of langchain, community, core,
experimental, etc. is being modified. Use "docs: ..." for purely docs
changes, "templates: ..." for template changes, "infra: ..." for CI
changes.
- Example: "community: add foobar LLM"
- [x] **PR message**:
- **Description:** mention not-caching methods in CacheBackedEmbeddings
- **Issue:** n/a I almost created one until I read the code
- **Dependencies:** n/a
- **Twitter handle:** `tarsylia`
- [ ] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [ ] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
**Description**: Improves the stability of all Cohere partner package
integration tests. Fixes a bug with document parsing (both dicts and
Documents are handled).
**Description**: This PR simplifies an integration test within the
Cohere partner package:
* It no longer relies on exact model answers
* It no longer relies on a third party tool
This PR completes work for PR #18798 to expose raw tool output in
on_tool_end.
Affected APIs:
* astream_log
* astream_events
* callbacks sent to langsmith via langsmith-sdk
* Any other code that relies on BaseTracer!
---------
Co-authored-by: Eugene Yurtsev <eyurtsev@gmail.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
- This ensures ids are stable across streamed chunks
- Multiple messages in batch call get separate ids
- Also fix ids being dropped when combining message chunks
Thank you for contributing to LangChain!
- [ ] **PR title**: "package: description"
- Where "package" is whichever of langchain, community, core,
experimental, etc. is being modified. Use "docs: ..." for purely docs
changes, "templates: ..." for template changes, "infra: ..." for CI
changes.
- Example: "community: add foobar LLM"
- [ ] **PR message**: ***Delete this entire checklist*** and replace
with
- **Description:** a description of the change
- **Issue:** the issue # it fixes, if applicable
- **Dependencies:** any dependencies required for this change
- **Twitter handle:** if your PR gets announced, and you'd like a
mention, we'll gladly shout you out!
- [ ] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [ ] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
- **Description:** add `remove_comments` option (default: True): do not
extract html _comments_,
- **Issue:** None,
- **Dependencies:** None,
- **Tag maintainer:** @nfcampos ,
- **Twitter handle:** peter_v
I ran `make format`, `make lint` and `make test`.
Discussion: I my use case, I prefer to not have the comments in the
extracted text:
* e.g. from a Google tag that is added in the html as comment
* e.g. content that the authors have temporarily hidden to make it non
visible to the regular reader
Removing the comments makes the extracted text more alike the intended
text to be seen by the reader.
**Choice to make:** do we prefer to make the default for this
`remove_comments` option to be True or False?
I have changed it to True in a second commit, since that is how I would
prefer to use it by default. Have the
cleaned text (without technical Google tags etc.) and also closer to the
actually visible and intended content.
I am not sure what is best aligned with the conventions of langchain in
general ...
INITIAL VERSION (new version above):
~**Choice to make:** do we prefer to make the default for this
`ignore_comments` option to be True or False?
I have set it to False now to be backwards compatible. On the other
hand, I would use it mostly with True.
I am not sure what is best aligned with the conventions of langchain in
general ...~
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
As in #19346, this PR exposes `request_timeout` in `BaseCohere`, while
`max_retires` is no longer a parameter of the beneath client
(`cohere.Client`) and it is already configured in
`langchain_cohere.llms.Cohere`.
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
- **Description:** the layout of html pages can be variant based on the
bootstrap framework or the styles of the pages. So we need to have a
splitter to transform the html tags to a proper layout and then split
the html content based on the provided list of tags to determine its
html sections. We are using BS4 library along with xslt structure to
split the html content using an section aware approach.
- **Dependencies:** No new dependencies
- **Twitter handle:** @m_setayesh
Please make sure your PR is passing linting and testing before
submitting. Run `make format`, `make lint` and `make test` from the root
of the package you've modified to check this locally.
See contribution guidelines for more information on how to write/run
tests, lint, etc: https://python.langchain.com/docs/contributing/
If you're adding a new integration, please include:
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
If no one reviews your PR within a few days, please @-mention one of
@baskaryan, @eyurtsev, @hwchase17.
-->
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
[Dria](https://dria.co/) is a hub of public RAG models for developers to
both contribute and utilize a shared embedding lake. This PR adds a
retriever that can retrieve documents from Dria.
Thank you for contributing to LangChain!
- [x] **PR title**: "package: description"
- Where "package" is whichever of langchain, community, core,
experimental, etc. is being modified. Use "docs: ..." for purely docs
changes, "templates: ..." for template changes, "infra: ..." for CI
changes.
- Example: "community: add foobar LLM"
- [x] **PR message**:
- **Description:** Fix argument translation from OpenAPI spec to OpenAI
function call (and similar)
- **Issue:** OpenGPTs failures with calling Action Server based actions.
- **Dependencies:** None
- **Twitter handle:** mikkorpela
- [x] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
~2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.~
- [x] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
Description: Update `ChatZhipuAI` to support the latest `glm-4` model.
Issue: N/A
Dependencies: httpx, httpx-sse, PyJWT
The previous `ChatZhipuAI` implementation requires the `zhipuai`
package, and cannot call the latest GLM model. This is because
- The old version `zhipuai==1.*` doesn't support the latest model.
- `zhipuai==2.*` requires `pydantic V2`, which is incompatible with
'langchain-community'.
This re-implementation invokes the GLM model by sending HTTP requests to
[open.bigmodel.cn](https://open.bigmodel.cn/dev/api) via the `httpx`
package, and uses the `httpx-sse` package to handle stream events.
---------
Co-authored-by: zR <2448370773@qq.com>
- **Description:** Add functionality to generate Mermaid syntax and
render flowcharts from graph data. This includes support for custom node
colors and edge curve styles, as well as the ability to export the
generated graphs to PNG images using either the Mermaid.INK API or
Pyppeteer for local rendering.
- **Dependencies:** Optional dependencies are `pyppeteer` if rendering
wants to be done using Pypeteer and Javascript code.
---------
Co-authored-by: Angel Igareta <angel.igareta@klarna.com>
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
**Description:** An additional `U` argument was added for the
instructions to install the pip packages for the MediaWiki Dump Document
loader which was leading to error in installing the package. Removing
the argument fixed the command to install.
**Issue:** #19820
**Dependencies:** No dependency change requierd
**Twitter handle:** [@vardhaman722](https://twitter.com/vardhaman722)
* Replace `source_documents` with `documents`
* Pass `documents` as a named arg vs keyword
* Make `parsed_docs` more robust
* Fix edge case of doc page_content being `None`
- **Updating Together.ai Endpoint**: "langchain_together: Updated
Deprecated endpoint for partner package"
- Description: The inference API of together is deprecates, do replaced
with completions and made corresponding changes.
- Twitter handle: @dev_yashmathur
---------
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
- **Description:** Add attribution_token within
GoogleVertexAISearchRetriever so user can provide this information to
Google support team or product team during debug session.
Reference:
https://cloud.google.com/generative-ai-app-builder/docs/view-analytics#user-events
Attribution tokens. Attribution tokens are unique IDs generated by
Vertex AI Search and returned with each search request. Make sure to
include that attribution token as UserEvent.attributionToken with any
user events resulting from a search. This is needed to identify if a
search is served by the API. Only user events with a Google-generated
attribution token are used to compute metrics.
- **Issue:** No
- **Dependencies:** No
- **Twitter handle:** abehsu1992626
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
- **Description:** Support reranking based on cross encoder models
available from HuggingFace.
- Added `CrossEncoder` schema
- Implemented `HuggingFaceCrossEncoder` and
`SagemakerEndpointCrossEncoder`
- Implemented `CrossEncoderReranker` that performs similar functionality
to `CohereRerank`
- Added `cross-encoder-reranker.ipynb` to demonstrate how to use it.
Please let me know if anything else needs to be done to make it visible
on the table-of-contents navigation bar on the left, or on the card list
on [retrievers documentation
page](https://python.langchain.com/docs/integrations/retrievers).
- **Issue:** N/A
- **Dependencies:** None other than the existing ones.
---------
Co-authored-by: Kenny Choe <kchoe@amazon.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
Description: Video imagery to text (Closed Captioning)
This pull request introduces the VideoCaptioningChain, a tool for
automated video captioning. It processes audio and video to generate
subtitles and closed captions, merging them into a single SRT output.
Issue: https://github.com/langchain-ai/langchain/issues/11770
Dependencies: opencv-python, ffmpeg-python, assemblyai, transformers,
pillow, torch, openai
Tag maintainer:
@baskaryan
@hwchase17
Hello! We are a group of students from the University of Toronto
(@LunarECL, @TomSadan, @nicoledroi1, @A2113S) that want to make a
contribution to the LangChain community! We have ran make format, make
lint and make test locally before submitting the PR. To our knowledge,
our changes do not introduce any new errors.
Thank you for taking the time to review our PR!
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
### Description
This implementation adds functionality from the AlphaVantage API,
renowned for its comprehensive financial data. The class encapsulates
various methods, each dedicated to fetching specific types of financial
information from the API.
### Implemented Functions
- **`search_symbols`**:
- Searches the AlphaVantage API for financial symbols using the provided
keywords.
- **`_get_market_news_sentiment`**:
- Retrieves market news sentiment for a specified stock symbol from the
AlphaVantage API.
- **`_get_time_series_daily`**:
- Fetches daily time series data for a specific symbol from the
AlphaVantage API.
- **`_get_quote_endpoint`**:
- Obtains the latest price and volume information for a given symbol
from the AlphaVantage API.
- **`_get_time_series_weekly`**:
- Gathers weekly time series data for a particular symbol from the
AlphaVantage API.
- **`_get_top_gainers_losers`**:
- Provides details on top gainers, losers, and most actively traded
tickers in the US market from the AlphaVantage API.
### Issue:
- #11994
### Dependencies:
- 'requests' library for HTTP requests. (import requests)
- 'pytest' library for testing. (import pytest)
---------
Co-authored-by: Adam Badar <94140103+adam-badar@users.noreply.github.com>
Co-authored-by: Harrison Chase <hw.chase.17@gmail.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
- [x] **PR title**: "package: description"
- Where "package" is whichever of langchain, community, core,
experimental, etc. is being modified. Use "docs: ..." for purely docs
changes, "templates: ..." for template changes, "infra: ..." for CI
changes.
- Example: "community: add foobar LLM"
- [x] **PR message**: ***Delete this entire checklist*** and replace
with
- **Description:** Langchain-Predibase integration was failing, because
it was not current with the Predibase SDK; in addition, Predibase
integration tests were instantiating the Langchain Community `Predibase`
class with one required argument (`model`) missing. This change updates
the Predibase SDK usage and fixes the integration tests.
- **Twitter handle:** `@alexsherstinsky`
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
- **Description:** Code written by following, the official documentation
of [Google Drive
Loader](https://python.langchain.com/docs/integrations/document_loaders/google_drive),
gives errors. I have opened an issue regarding this. See #14725. This is
a pull request for modifying the documentation to use an approach that
makes the code work. Basically, the change is that we need to always set
the GOOGLE_APPLICATION_CREDENTIALS env var to an emtpy string, rather
than only in case of RefreshError. Also, rewrote 2 paragraphs to make
the instructions more clear.
- **Issue:** See this related [issue #
14725](https://github.com/langchain-ai/langchain/issues/14725)
- **Dependencies:** NA
- **Tag maintainer:** @baskaryan
- **Twitter handle:** NA
Co-authored-by: Snehil <snehil@example.com>
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Thank you for contributing to LangChain!
- [x] **PR title**: "community: added support for llmsherpa library"
- [x] **Add tests and docs**:
1. Integration test:
'docs/docs/integrations/document_loaders/test_llmsherpa.py'.
2. an example notebook:
`docs/docs/integrations/document_loaders/llmsherpa.ipynb`.
- [x] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
---------
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
this pr also drops the community added action for checking broken links
in mdx. It does not work well for our use case, throwing errors for
local paths, plus the rest of the errors our in house solution had.
# Description
Implementing `_combine_llm_outputs` to `ChatMistralAI` to override the
default implementation in `BaseChatModel` returning `{}`. The
implementation is inspired by the one in `ChatOpenAI` from package
`langchain-openai`.
# Issue
None
# Dependencies
None
# Twitter handle
None
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
**Description:**
This template utilizes Chroma and TGI (Text Generation Inference) to
execute RAG on the Intel Xeon Scalable Processors. It serves as a
demonstration for users, illustrating the deployment of the RAG service
on the Intel Xeon Scalable Processors and showcasing the resulting
performance enhancements.
**Issue:**
None
**Dependencies:**
The template contains the poetry project requirements to run this
template.
CPU TGI batching is WIP.
**Twitter handle:**
None
---------
Signed-off-by: lvliang-intel <liang1.lv@intel.com>
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
**Description:** We'd like to support passing additional kwargs in
`with_structured_output`. I believe this is the accepted approach to
enable additional arguments on API calls.
- **Description:** Haskell language support added in text_splitter
module
- **Dependencies:** No
- **Twitter handle:** @nisargtr
If no one reviews your PR within a few days, please @-mention one of
@baskaryan, @eyurtsev, @hwchase17.
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
**Description:** PR adds support for limiting number of messages
preserved in a session history for DynamoDBChatMessageHistory
---------
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
### Subject: Fix Type Misdeclaration for index_schema in redis/base.py
I noticed a type misdeclaration for the index_schema column in the
redis/base.py file.
When following the instructions outlined in [Redis Custom Metadata
Indexing](https://python.langchain.com/docs/integrations/vectorstores/redis)
to create our own index_schema, it leads to a Pylance type error. <br/>
**The error message indicates that Dict[str, list[Dict[str, str]]] is
incompatible with the type Optional[Union[Dict[str, str], str,
os.PathLike]].**
```
index_schema = {
"tag": [{"name": "credit_score"}],
"text": [{"name": "user"}, {"name": "job"}],
"numeric": [{"name": "age"}],
}
rds, keys = Redis.from_texts_return_keys(
texts,
embeddings,
metadatas=metadata,
redis_url="redis://localhost:6379",
index_name="users_modified",
index_schema=index_schema,
)
```
Therefore, I have created this pull request to rectify the type
declaration problem.
---------
Co-authored-by: Eugene Yurtsev <eyurtsev@gmail.com>
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
## Feature
- Set additional headers in constructor
- Headers will be sent in post request
This feature is useful if deploying Ollama on a cloud service such as
hugging face, which requires authentication tokens to be passed in the
request header.
## Tests
- Test if header is passed
- Test if header is not passed
Similar to https://github.com/langchain-ai/langchain/pull/15881
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
If `prompt` is passed into `create_sql_agent()`, then
`toolkit.get_context()` shouldn't be executed against the database
unless relevant prompt variables (`table_info` or `table_names`) are
present .
Thank you for contributing to LangChain!
- [x] **PR title**: "community: Implement DirectoryLoader lazy_load
function"
- [x] **Description**: The `lazy_load` function of the `DirectoryLoader`
yields each document separately. If the given `loader_cls` of the
`DirectoryLoader` also implemented `lazy_load`, it will be used to yield
subdocuments of the file.
- [x] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access:
`libs/community/tests/unit_tests/document_loaders/test_directory_loader.py`
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory:
`docs/docs/integrations/document_loaders/directory.ipynb`
- [x] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
---------
Co-authored-by: Eugene Yurtsev <eyurtsev@gmail.com>
**Description:**
When using the SQLDatabaseChain with Llama2-70b LLM and, SQLite
database. I was getting `Warning: You can only execute one statement at
a time.`.
```
from langchain.sql_database import SQLDatabase
from langchain_experimental.sql import SQLDatabaseChain
sql_database_path = '/dccstor/mmdataretrieval/mm_dataset/swimming_record/rag_data/swimmingdataset.db'
sql_db = get_database(sql_database_path)
db_chain = SQLDatabaseChain.from_llm(mistral, sql_db, verbose=True, callbacks = [callback_obj])
db_chain.invoke({
"query": "What is the best time of Lance Larson in men's 100 meter butterfly competition?"
})
```
Error:
```
Warning Traceback (most recent call last)
Cell In[31], line 3
1 import langchain
2 langchain.debug=False
----> 3 db_chain.invoke({
4 "query": "What is the best time of Lance Larson in men's 100 meter butterfly competition?"
5 })
File ~/.conda/envs/guardrails1/lib/python3.9/site-packages/langchain/chains/base.py:162, in Chain.invoke(self, input, config, **kwargs)
160 except BaseException as e:
161 run_manager.on_chain_error(e)
--> 162 raise e
163 run_manager.on_chain_end(outputs)
164 final_outputs: Dict[str, Any] = self.prep_outputs(
165 inputs, outputs, return_only_outputs
166 )
File ~/.conda/envs/guardrails1/lib/python3.9/site-packages/langchain/chains/base.py:156, in Chain.invoke(self, input, config, **kwargs)
149 run_manager = callback_manager.on_chain_start(
150 dumpd(self),
151 inputs,
152 name=run_name,
153 )
154 try:
155 outputs = (
--> 156 self._call(inputs, run_manager=run_manager)
157 if new_arg_supported
158 else self._call(inputs)
159 )
160 except BaseException as e:
161 run_manager.on_chain_error(e)
File ~/.conda/envs/guardrails1/lib/python3.9/site-packages/langchain_experimental/sql/base.py:198, in SQLDatabaseChain._call(self, inputs, run_manager)
194 except Exception as exc:
195 # Append intermediate steps to exception, to aid in logging and later
196 # improvement of few shot prompt seeds
197 exc.intermediate_steps = intermediate_steps # type: ignore
--> 198 raise exc
File ~/.conda/envs/guardrails1/lib/python3.9/site-packages/langchain_experimental/sql/base.py:143, in SQLDatabaseChain._call(self, inputs, run_manager)
139 intermediate_steps.append(
140 sql_cmd
141 ) # output: sql generation (no checker)
142 intermediate_steps.append({"sql_cmd": sql_cmd}) # input: sql exec
--> 143 result = self.database.run(sql_cmd)
144 intermediate_steps.append(str(result)) # output: sql exec
145 else:
File ~/.conda/envs/guardrails1/lib/python3.9/site-packages/langchain_community/utilities/sql_database.py:436, in SQLDatabase.run(self, command, fetch, include_columns)
425 def run(
426 self,
427 command: str,
428 fetch: Literal["all", "one"] = "all",
429 include_columns: bool = False,
430 ) -> str:
431 """Execute a SQL command and return a string representing the results.
432
433 If the statement returns rows, a string of the results is returned.
434 If the statement returns no rows, an empty string is returned.
435 """
--> 436 result = self._execute(command, fetch)
438 res = [
439 {
440 column: truncate_word(value, length=self._max_string_length)
(...)
443 for r in result
444 ]
446 if not include_columns:
File ~/.conda/envs/guardrails1/lib/python3.9/site-packages/langchain_community/utilities/sql_database.py:413, in SQLDatabase._execute(self, command, fetch)
410 elif self.dialect == "postgresql": # postgresql
411 connection.exec_driver_sql("SET search_path TO %s", (self._schema,))
--> 413 cursor = connection.execute(text(command))
414 if cursor.returns_rows:
415 if fetch == "all":
File ~/.conda/envs/guardrails1/lib/python3.9/site-packages/sqlalchemy/engine/base.py:1416, in Connection.execute(self, statement, parameters, execution_options)
1414 raise exc.ObjectNotExecutableError(statement) from err
1415 else:
-> 1416 return meth(
1417 self,
1418 distilled_parameters,
1419 execution_options or NO_OPTIONS,
1420 )
File ~/.conda/envs/guardrails1/lib/python3.9/site-packages/sqlalchemy/sql/elements.py:516, in ClauseElement._execute_on_connection(self, connection, distilled_params, execution_options)
514 if TYPE_CHECKING:
515 assert isinstance(self, Executable)
--> 516 return connection._execute_clauseelement(
517 self, distilled_params, execution_options
518 )
519 else:
520 raise exc.ObjectNotExecutableError(self)
File ~/.conda/envs/guardrails1/lib/python3.9/site-packages/sqlalchemy/engine/base.py:1639, in Connection._execute_clauseelement(self, elem, distilled_parameters, execution_options)
1627 compiled_cache: Optional[CompiledCacheType] = execution_options.get(
1628 "compiled_cache", self.engine._compiled_cache
1629 )
1631 compiled_sql, extracted_params, cache_hit = elem._compile_w_cache(
1632 dialect=dialect,
1633 compiled_cache=compiled_cache,
(...)
1637 linting=self.dialect.compiler_linting | compiler.WARN_LINTING,
1638 )
-> 1639 ret = self._execute_context(
1640 dialect,
1641 dialect.execution_ctx_cls._init_compiled,
1642 compiled_sql,
1643 distilled_parameters,
1644 execution_options,
1645 compiled_sql,
1646 distilled_parameters,
1647 elem,
1648 extracted_params,
1649 cache_hit=cache_hit,
1650 )
1651 if has_events:
1652 self.dispatch.after_execute(
1653 self,
1654 elem,
(...)
1658 ret,
1659 )
File ~/.conda/envs/guardrails1/lib/python3.9/site-packages/sqlalchemy/engine/base.py:1848, in Connection._execute_context(self, dialect, constructor, statement, parameters, execution_options, *args, **kw)
1843 return self._exec_insertmany_context(
1844 dialect,
1845 context,
1846 )
1847 else:
-> 1848 return self._exec_single_context(
1849 dialect, context, statement, parameters
1850 )
File ~/.conda/envs/guardrails1/lib/python3.9/site-packages/sqlalchemy/engine/base.py:1988, in Connection._exec_single_context(self, dialect, context, statement, parameters)
1985 result = context._setup_result_proxy()
1987 except BaseException as e:
-> 1988 self._handle_dbapi_exception(
1989 e, str_statement, effective_parameters, cursor, context
1990 )
1992 return result
File ~/.conda/envs/guardrails1/lib/python3.9/site-packages/sqlalchemy/engine/base.py:2346, in Connection._handle_dbapi_exception(self, e, statement, parameters, cursor, context, is_sub_exec)
2344 else:
2345 assert exc_info[1] is not None
-> 2346 raise exc_info[1].with_traceback(exc_info[2])
2347 finally:
2348 del self._reentrant_error
File ~/.conda/envs/guardrails1/lib/python3.9/site-packages/sqlalchemy/engine/base.py:1969, in Connection._exec_single_context(self, dialect, context, statement, parameters)
1967 break
1968 if not evt_handled:
-> 1969 self.dialect.do_execute(
1970 cursor, str_statement, effective_parameters, context
1971 )
1973 if self._has_events or self.engine._has_events:
1974 self.dispatch.after_cursor_execute(
1975 self,
1976 cursor,
(...)
1980 context.executemany,
1981 )
File ~/.conda/envs/guardrails1/lib/python3.9/site-packages/sqlalchemy/engine/default.py:922, in DefaultDialect.do_execute(self, cursor, statement, parameters, context)
921 def do_execute(self, cursor, statement, parameters, context=None):
--> 922 cursor.execute(statement, parameters)
Warning: You can only execute one statement at a time.
```
**Issue:**
The Error occurs because when generating the SQLQuery, the llm_input
includes the stop character of "\nSQLResult:", so for this user query
the LLM generated response is **SELECT Time FROM men_butterfly_100m
WHERE Swimmer = 'Lance Larson';\nSQLResult:** it is required to remove
the SQLResult suffix on the llm response before executing it on the
database.
```
llm_inputs = {
"input": input_text,
"top_k": str(self.top_k),
"dialect": self.database.dialect,
"table_info": table_info,
"stop": ["\nSQLResult:"],
}
sql_cmd = self.llm_chain.predict(
callbacks=_run_manager.get_child(),
**llm_inputs,
).strip()
if SQL_RESULT in sql_cmd:
sql_cmd = sql_cmd.split(SQL_RESULT)[0].strip()
result = self.database.run(sql_cmd)
```
<!-- Thank you for contributing to LangChain!
Please title your PR "<package>: <description>", where <package> is
whichever of langchain, community, core, experimental, etc. is being
modified.
Replace this entire comment with:
- **Description:** a description of the change,
- **Issue:** the issue # it fixes if applicable,
- **Dependencies:** any dependencies required for this change,
- **Twitter handle:** we announce bigger features on Twitter. If your PR
gets announced, and you'd like a mention, we'll gladly shout you out!
Please make sure your PR is passing linting and testing before
submitting. Run `make format`, `make lint` and `make test` from the root
of the package you've modified to check this locally.
See contribution guidelines for more information on how to write/run
tests, lint, etc: https://python.langchain.com/docs/contributing/
If you're adding a new integration, please include:
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
If no one reviews your PR within a few days, please @-mention one of
@baskaryan, @eyurtsev, @hwchase17.
-->
---------
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Description: Fix xml parser to handle strings that only contain the root
tag
Issue: N/A
Dependencies: None
Twitter handle: N/A
A valid xml text can contain only the root level tag. Example: <body>
Some text here
</body>
The example above is a valid xml string. If parsed with the current
implementation the result is {"body": []}. This fix checks if the root
level text contains any non-whitespace character and if that's the case
it returns {root.tag: root.text}. The result is that the above text is
correctly parsed as {"body": "Some text here"}
@ale-delfino
Thank you for contributing to LangChain!
Checklist:
- [x] PR title: Please title your PR "package: description", where
"package" is whichever of langchain, community, core, experimental, etc.
is being modified. Use "docs: ..." for purely docs changes, "templates:
..." for template changes, "infra: ..." for CI changes.
- Example: "community: add foobar LLM"
- [x] PR message: **Delete this entire template message** and replace it
with the following bulleted list
- **Description:** a description of the change
- **Issue:** the issue # it fixes, if applicable
- **Dependencies:** any dependencies required for this change
- **Twitter handle:** if your PR gets announced, and you'd like a
mention, we'll gladly shout you out!
- [x] Pass lint and test: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified to check that you're
passing lint and testing. See contribution guidelines for more
information on how to write/run tests, lint, etc:
https://python.langchain.com/docs/contributing/
- [x] Add tests and docs: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
@baskaryan, @efriis, @eyurtsev, @hwchase17.
---------
Co-authored-by: Eugene Yurtsev <eyurtsev@gmail.com>
When testing Nomic embeddings --
```
from langchain_community.embeddings import LlamaCppEmbeddings
embd_model_path = "/Users/rlm/Desktop/Code/llama.cpp/models/nomic-embd/nomic-embed-text-v1.Q4_K_S.gguf"
embd_lc = LlamaCppEmbeddings(model_path=embd_model_path)
embedding_lc = embd_lc.embed_query(query)
```
We were seeing this error for strings > a certain size --
```
File ~/miniforge3/envs/llama2/lib/python3.9/site-packages/llama_cpp/llama.py:827, in Llama.embed(self, input, normalize, truncate, return_count)
824 s_sizes = []
826 # add to batch
--> 827 self._batch.add_sequence(tokens, len(s_sizes), False)
828 t_batch += n_tokens
829 s_sizes.append(n_tokens)
File ~/miniforge3/envs/llama2/lib/python3.9/site-packages/llama_cpp/_internals.py:542, in _LlamaBatch.add_sequence(self, batch, seq_id, logits_all)
540 self.batch.token[j] = batch[i]
541 self.batch.pos[j] = i
--> 542 self.batch.seq_id[j][0] = seq_id
543 self.batch.n_seq_id[j] = 1
544 self.batch.logits[j] = logits_all
ValueError: NULL pointer access
```
The default `n_batch` of llama-cpp-python's Llama is `512` but we were
explicitly setting it to `8`.
These need to be set to equal for embedding models.
* The embedding.cpp example has an assertion to make sure these are
always equal.
* Apparently this is not being done properly in llama-cpp-python.
With `n_batch` set to 8, if more than 8 tokens are passed the batch runs
out of space and it crashes.
This also explains why the CPU compute buffer size was small:
raw client with default `n_batch=512`
```
llama_new_context_with_model: CPU input buffer size = 3.51 MiB
llama_new_context_with_model: CPU compute buffer size = 21.00 MiB
```
langchain with `n_batch=8`
```
llama_new_context_with_model: CPU input buffer size = 0.04 MiB
llama_new_context_with_model: CPU compute buffer size = 0.33 MiB
```
We can work around this by passing `n_batch=512`, but this will not be
obvious to some users:
```
embedding = LlamaCppEmbeddings(model_path=embd_model_path,
n_batch=512)
```
From discussion w/ @cebtenzzre. Related:
https://github.com/abetlen/llama-cpp-python/issues/1189
Co-authored-by: Bagatur <baskaryan@gmail.com>
**Description:** The base URL for OpenAI is retrieved from the
environment variable "OPENAI_BASE_URL", whereas for langchain it is
obtained from "OPENAI_API_BASE". By adding `base_url =
os.environ.get("OPENAI_API_BASE")`, the OpenAI proxy can execute
correctly.
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
Thank you for contributing to LangChain!
- **Description:** added unit tests for NotebookLoader. Linked PR:
https://github.com/langchain-ai/langchain/pull/17614
- **Issue:**
[#17614](https://github.com/langchain-ai/langchain/pull/17614)
- **Twitter handle:** @paulodoestech
- [x] Pass lint and test: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified to check that you're
passing lint and testing. See contribution guidelines for more
information on how to write/run tests, lint, etc:
https://python.langchain.com/docs/contributing/
- [x] Add tests and docs: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
---------
Co-authored-by: lachiewalker <lachiewalker1@hotmail.com>
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
**Description:** Created a Langchain Tool for OpenAI DALLE Image
Generation.
**Issue:**
[#15901](https://github.com/langchain-ai/langchain/issues/15901)
**Dependencies:** n/a
**Twitter handle:** @paulodoestech
- [x] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [x] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
**Description:**: adding checking codes for calling AI model get error
in chat_models/base.py and llms/base.py
**Issue**: Sometimes the AI Model calling will get error, we should
raise it.
Otherwise, the next code 'choices.extend(response["choices"])' will
throw a "TypeError: 'NoneType' object is not iterable" error to mask the
true error.
Because 'response["choices"]' is None.
**Dependencies**: None
---------
Co-authored-by: yangkx <yangkx@asiainfo-int.com>
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
## PR message
**Description:** This PR adds a README file for the Together API in the
`libs/partners` folder of this repository. The README includes:
- A brief description of the package
- Installation instructions and class introductions
- Simple usage examples
**Issue:** #17545
This PR only contains document changes.
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
- **Description:**
1. Fix the BiliBiliLoader that can receive cookie parameters, it
requires 3 other parameters to run. The change is backward compatible.
2. Add test;
3. Add example in docs
- **Issue:** [#14213]
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
**Description:** A few grammatical changes to improve readability of the
LCEL .ipynb and tidy some null characters.
**Issue:** N/A
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
- [x] **PR title**: "community: Support streaming in Azure ML and few
naming changes"
- [x] **PR message**:
- **Description:** Added support for streaming for azureml_endpoint.
Also, renamed and AzureMLEndpointApiType.realtime to
AzureMLEndpointApiType.dedicated. Also, added new classes
CustomOpenAIChatContentFormatter and CustomOpenAIContentFormatter and
updated the classes LlamaChatContentFormatter and LlamaContentFormatter
to now show a deprecated warning message when instantiated.
---------
Co-authored-by: Sachin Paryani <saparan@microsoft.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
**Description:** At times, BaseChatMemory._get_input_output may acquire
some extra keys such as 'intermediate_steps' (agent_executor with
return_intermediate_steps set to True) and 'messages'
(agent_executor.iter with memory). In these instances, _get_input_output
can raise an error due to the presence of multiple keys. The 'output'
field should be used as the default field in these cases.
**Issue:** #16791
Previous markdown code was not working as intended, new code should add
green box around the tip so it is highlighted
Co-authored-by: Hershenson, Isaac (Extern) <isaac.hershenson.extern@bayer04.de>
Co-authored-by: Bagatur <baskaryan@gmail.com>
- Description: Added missing `from_documents` method to `KNNRetriever`,
providing the ability to supply metadata to LangChain `Document`s, and
to give it parity to the other retrievers, which do have
`from_documents`.
- Issue: None
- Dependencies: None
- Twitter handle: None
Co-authored-by: Victor Adan <vadan@netroadshow.com>
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Relates to #17048
Description : Applied fix to dynamodb and elasticsearch file.
Error was : `Cannot override writeable attribute with read-only
property`
Suggestion:
instead of adding
```
@messages.setter
def messages(self, messages: List[BaseMessage]) -> None:
raise NotImplementedError("Use add_messages instead")
```
we can change base class property
`messages: List[BaseMessage]`
to
```
@property
def messages(self) -> List[BaseMessage]:...
```
then we don't need to add `@messages.setter` in all child classes.
**Description:**
While not technically incorrect, the TypeVar used for the `@beta`
decorator prevented pyright (and thus most vscode users) from correctly
seeing the types of functions/classes decorated with `@beta`.
This is in part due to a small bug in pyright
(https://github.com/microsoft/pyright/issues/7448 ) - however, the
`Type` bound in the typevar `C = TypeVar("C", Type, Callable)` is not
doing anything - classes are `Callables` by default, so by my
understanding binding to `Type` does not actually provide any more
safety - the modified annotation still works correctly for both
functions, properties, and classes.
---------
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
In this small PR I added the `template_tool_response` arg to the
`create_json_chat` function, so that users can customize this prompt in
case of need.
Thanks for your reviews!
---------
Co-authored-by: taamedag <Davide.Menini@swisscom.com>
This patch updates multiple function "run" to "invoke" in
llm_symbolic_math.ipynb.
Without this patch, you see following message.
The function `run` was deprecated in LangChain 0.1.0
and will be removed in 0.2.0. Use invoke instead.
Signed-off-by: Masanari Iida <standby24x7@gmail.com>
**Description:** Adds support for `with_structured_output` to Cohere,
which supports single function calling.
---------
Co-authored-by: BeatrixCohere <128378696+BeatrixCohere@users.noreply.github.com>
- [x] **PR title**: "community: fix baidu qianfan missing stop
parameter"
- [x] **PR message**:
- **Description: Baidu Qianfan lost the stop parameter when requesting
service due to extracting it from kwargs. This bug can cause the agent
to receive incorrect results
---------
Co-authored-by: ligang33 <ligang33@baidu.com>
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
Bug fixes in this PR:
* allows for other params such as "message" not just the input param to
the prompt for the cohere tools agent
* fixes to documents kwarg from messages
* fixes to tool_calls API call
---------
Co-authored-by: Harry M <127103098+harry-cohere@users.noreply.github.com>
- **Issue:** When passing an empty list to MergerRetriever it fails with
error: ValueError: max() arg is an empty sequence
- **Description:** We have a use case where we dynamically select
retrievers and use MergerRetriever for merging the output of the
retrievers. We faced this issue when the retriever_docs list is empty.
Adding a default 0 for cases when retriever_docs is an empty list to
avoid "ValueError: max() arg is an empty sequence". Also, changed to use
map() which is more than twice as fast compared to the current
implementation.
```
import timeit
# Sample retriever_docs with varying lengths of sublists
retriever_docs = [[i for i in range(j)] for j in range(1, 1000)]
# First code snippet
code1 = '''
max_docs = max(len(docs) for docs in retriever_docs)
'''
# Second code snippet
code2 = '''
max_docs = max(map(len, retriever_docs), default=0)
'''
# Benchmarking
time1 = timeit.timeit(stmt=code1, globals=globals(), number=10000)
time2 = timeit.timeit(stmt=code2, globals=globals(), number=10000)
# Output
print(f"Execution time for code snippet 1: {time1} seconds")
print(f"Execution time for code snippet 2: {time2} seconds")
```
- **Dependencies:** none
The previous version didn't had Voyage rerank in the init file
- [ ] **PR title**: langchain_voyageai reranker is not working
- [ ] **PR message**:
- **Description:** This fix let you run reranker from voyage
- **Issue:** Was not able to run reranker from voyage
@efriis
Due to changes in the OpenAI SDK, the previous method of setting the
OpenAI proxy in ChatOpenAI no longer works. This PR fixes this issue,
making the previous way of setting the OpenAI proxy in ChatOpenAI
effective again.
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
This is a follow up to #18371. These are the changes:
- New **Azure AI Services** toolkit and tools to replace those of
**Azure Cognitive Services**.
- Updated documentation for Microsoft platform.
- The image analysis tool has been rewritten to use the new package
`azure-ai-vision-imageanalysis`, doing a proper replacement of
`azure-ai-vision`.
These changes:
- Update outdated naming from "Azure Cognitive Services" to "Azure AI
Services".
- Update documentation to use non-deprecated methods to create and use
agents.
- Removes need to depend on yanked python package (`azure-ai-vision`)
There is one new dependency that is needed as a replacement to
`azure-ai-vision`:
- `azure-ai-vision-imageanalysis`. This is optional and declared within
a function.
There is a new `azure_ai_services.ipynb` notebook showing usage; Changes
have been linted and formatted.
I am leaving the actions of adding deprecation notices and future
removal of Azure Cognitive Services up to the LangChain team, as I am
not sure what the current practice around this is.
---
If this PR makes it, my handle is @galo@mastodon.social
---------
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
Co-authored-by: ccurme <chester.curme@gmail.com>
- **Description**: `bigdl-llm` library has been renamed to
[`ipex-llm`](https://github.com/intel-analytics/ipex-llm). This PR
migrates the `bigdl-llm` integration to `ipex-llm` .
- **Issue**: N/A. The original PR of `bigdl-llm` is
https://github.com/langchain-ai/langchain/pull/17953
- **Dependencies**: `ipex-llm` library
- **Contribution maintainer**: @shane-huang
Updated doc: docs/docs/integrations/llms/ipex_llm.ipynb
Updated test:
libs/community/tests/integration_tests/llms/test_ipex_llm.py
- **Description:** Add support for Intel Lab's [Visual Data Management
System (VDMS)](https://github.com/IntelLabs/vdms) as a vector store
- **Dependencies:** `vdms` library which requires protobuf = "4.24.2".
There is a conflict with dashvector in `langchain` package but conflict
is resolved in `community`.
- **Contribution maintainer:** [@cwlacewe](https://github.com/cwlacewe)
- **Added tests:**
libs/community/tests/integration_tests/vectorstores/test_vdms.py
- **Added docs:** docs/docs/integrations/vectorstores/vdms.ipynb
- **Added cookbook:** cookbook/multi_modal_RAG_vdms.ipynb
---------
Co-authored-by: Eugene Yurtsev <eugene@langchain.dev>
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
If you use an embedding dist function in an eval loop, you get warned
every time. Would prefer to just check once and forget about it.
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
- .stream() and .astream() call on_llm_new_token, removing the need for
subclasses to do so. Backwards compatible because now we don't pass
run_manager into ._stream and ._astream
- .generate() and .agenerate() now handle `stream: bool` kwarg for
_generate and _agenerate. Subclasses handle this arg by delegating to
._stream(), now one less thing they need to do. Backwards compat because
this is an optional arg that we now never pass to the subclasses
- .generate() and .agenerate() now inspect callback handlers to decide
on a default value for stream:bool if not passed in. This auto enables
streaming when using astream_events and astream_log
- as a result of these three changes any usage of .astream_events and
.astream_log should now yield chat model stream events
- In future PRs we can update all subclasses to reflect these two things
now handled by base class, but in meantime all will continue to work
* **Description**: add `None` type for `file_path` along with `str` and
`List[str]` types.
* `file_path`/`filename` arguments in `get_elements_from_api()` and
`partition()` can be `None`, however, there's no `None` type hint for
`file_path` in `UnstructuredAPIFileLoader` and `UnstructuredFileLoader`
currently.
* calling the function with `file_path=None` is no problem, but my IDE
annoys me lol.
* **Issue**: N/A
* **Dependencies**: N/A
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
- **Description:** Updates Meilisearch vectorstore for compatibility
with v1.6 and above. Adds embedders settings and embedder_name which are
now required.
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
**Description:**
This PR adds a slightly more helpful message to a Tool Exception
```
# current state
langchain_core.tools.ToolException: Too many arguments to single-input tool
# proposed state
langchain_core.tools.ToolException: Too many arguments to single-input tool. Consider using a StructuredTool instead.
```
**Issue:** Somewhat discussed here 👉#6197
**Dependencies:** None
**Twitter handle:** N/A
---------
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Thank you for contributing to LangChain!
- [ ] **cookbook** - update example for SalesGPT - include Stripe
Payment Link Generation
- **Description:** We updated the Jupyter notebook example with the
ability of the AI Agent to negotiate with customers and then close the
deal by generating a custom Stripe payment link.
- **Issue:** N/A
- **Dependencies:** N/a
- **Twitter handle:** @FilipMichalsky @0xtotaylor
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
---------
Co-authored-by: Filip Michalsky <filip_michalsky@g.harvard.edu>
Co-authored-by: Bagatur <baskaryan@gmail.com>
As mentioned in #18322, the current PydanticOutputParser won't work for
anyone trying to parse to pydantic v2 models. This PR adds a separate
`PydanticV2OutputParser`, as well as a `langchain_core.pydantic_v2`
namespace that will fail on import to any projects using pydantic<2.
Happy to update the docs for output parsers if this is something we're
interesting in adding.
On a separate note, I also updated `check_pydantic.sh` to detect
pydantic imports with leading whitespace and excluded the internal
namespaces. That change can be separated into its own PR if needed.
---------
Co-authored-by: Jan Nissen <jan23@gmail.com>
- **Description:** I've made a fix to a ParseError call in the
XMLOutputParser documentation.
- **Issue:** None
- **Dependencies:** None
Co-authored-by: Eugene Yurtsev <eyurtsev@gmail.com>
**Description:**
PebbloSafeLoader: Add support for non-file-based Document Loaders
This pull request enhances PebbloSafeLoader by introducing support for
several non-file-based Document Loaders. With this update,
PebbloSafeLoader now seamlessly integrates with the following loaders:
- GoogleDriveLoader
- SlackDirectoryLoader
- Unstructured EmailLoader
**Issue:** NA
**Dependencies:** - None
**Twitter handle:** @Raj__725
---------
Co-authored-by: Rahul Tripathi <rauhl.psit.ec@gmail.com>
Patch potential XML vulnerability CVE-2024-1455
This patches a potential XML vulnerability in the XMLOutputParser in
langchain-core. The vulnerability in some situations could lead to a
denial of service attack.
At risk are users that:
1) Running older distributions of python that have older version of
libexpat
2) Are using XMLOutputParser with an agent
3) Accept inputs from untrusted sources with this agent (e.g., endpoint
on the web that allows an untrusted user to interact wiith the parser)
Introduction
[Intel® Extension for
Transformers](https://github.com/intel/intel-extension-for-transformers)
is an innovative toolkit designed to accelerate GenAI/LLM everywhere
with the optimal performance of Transformer-based models on various
Intel platforms
Description
adding ITREX runtime embeddings using intel-extension-for-transformers.
added mdx documentation and example notebooks
added embedding import testing.
---------
Signed-off-by: yuwenzho <yuwen.zhou@intel.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
- [x] **PR title**: "experimental: Enhance LLMGraphTransformer with
async processing and improved readability"
- [x] **PR message**:
- **Description:** This pull request refactors the `process_response`
and `convert_to_graph_documents` methods in the LLMGraphTransformer
class to improve code readability and adds async versions of these
methods for concurrent processing.
The main changes include:
- Simplifying list comprehensions and conditional logic in the
process_response method for better readability.
- Adding async versions aprocess_response and
aconvert_to_graph_documents to enable concurrent processing of
documents.
These enhancements aim to improve the overall efficiency and
maintainability of the `LLMGraphTransformer` class.
- **Issue:** N/A
- **Dependencies:** No additional dependencies required.
- **Twitter handle:** @jjovalle99
- [x] **Add tests and docs**: N/A (This PR does not introduce a new
integration)
- [x] **Lint and test**: Ran make format, make lint, and make test from
the root of the modified package(s). All tests pass successfully.
Additional notes:
- The changes made in this PR are backwards compatible and do not
introduce any breaking changes.
- The PR touches only the `LLMGraphTransformer` class within the
experimental package.
---------
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
- **Description:** Implemented try-except block for
`GCSDirectoryLoader`. Reason: Users processing large number of
unstructured files in a folder may experience many different errors. A
try-exception block is added to capture these errors. A new argument
`use_try_except=True` is added to enable *silent failure* so that error
caused by processing one file does not break the whole function.
- **Issue:** N/A
- **Dependencies:** no new dependencies
- **Twitter handle:** timothywong731
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
Thank you for contributing to LangChain!
- [ ] **PR title**: "package: description"
- Where "package" is whichever of langchain, community, core,
experimental, etc. is being modified. Use "docs: ..." for purely docs
changes, "templates: ..." for template changes, "infra: ..." for CI
changes.
- Example: "community: add foobar LLM"
- [ ] **PR message**: ***Delete this entire checklist*** and replace
with
- **Description:** Adding oracle autonomous database document loader
integration. This will allow users to connect to oracle autonomous
database through connection string or TNS configuration.
https://www.oracle.com/autonomous-database/
- **Issue:** None
- **Dependencies:** oracledb python package
https://pypi.org/project/oracledb/
- **Twitter handle:** if your PR gets announced, and you'd like a
mention, we'll gladly shout you out!
- [ ] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
Unit test and doc are added.
- [ ] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
---------
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
- **Description:** Currently the semantic_configurations are not used
when creating an AzureSearch instance, instead creating a new one with
default values. This PR changes the behavior to use the passed
semantic_configurations if it is present, and the existing default
configuration if not.
---------
Co-authored-by: Adam Law <adamlaw@microsoft.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
Thank you for contributing to LangChain!
- [ ] **PR title**: "package: description"
- Where "package" is whichever of langchain, community, core,
experimental, etc. is being modified. Use "docs: ..." for purely docs
changes, "templates: ..." for template changes, "infra: ..." for CI
changes.
- Example: "community: add foobar LLM"
- [ ] **PR message**: ***Delete this entire checklist*** and replace
with
- **Description:** a description of the change
- **Issue:** the issue # it fixes, if applicable
- **Dependencies:** any dependencies required for this change
- **Twitter handle:** if your PR gets announced, and you'd like a
mention, we'll gladly shout you out!
- [ ] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [ ] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
DefusedXML is causing parsing errors on previously functional code with
the 0.7.x versions. These do not seem to support newer version of python
well. 0.8.x has only been released as rc, so we're not going to to use
it in the core package
* Adds support for `additional_kwargs` in `get_cohere_chat_request`
* This functionality passes in Cohere SDK specific parameters from
`BaseMessage` based classes to the API
---------
Co-authored-by: Erick Friis <erick@langchain.dev>
Thank you for contributing to LangChain!
- [x] **Add len() implementation to Chroma**: "package: community"
- [x] **PR message**:
- **Description:** add an implementation of the __len__() method for the
Chroma vectostore, for convenience.
- **Issue:** no exposed method to know the size of a Chroma vectorstore
- **Dependencies:** None
- **Twitter handle:** lowrank_adrian
- [x] **Add tests and docs**
- [x] **Lint and test**
---------
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
- **Description:** Be more explicit with the `model_kwargs` and
`encode_kwargs` for `HuggingFaceEmbeddings`.
- **Issue:** -
- **Dependencies:** -
I received some reports by my users that they didn't realise that you
could change the default `batch_size` with `HuggingFaceEmbeddings`,
which may be attributed to how the `model_kwargs` and `encode_kwargs`
don't give much information about what you can specify.
I've added some parameter names & links to the Sentence Transformers
documentation to help clear it up. Let me know if you'd rather have
Markdown/Sphinx-style hyperlinks rather than a "bare URL".
- Tom Aarsen
So this arose from the
https://github.com/langchain-ai/langchain/pull/18397 problem of document
loaders not supporting `pathlib.Path`.
This pull request provides more uniform support for Path as an argument.
The core ideas for this upgrade:
- if there is a local file path used as an argument, it should be
supported as `pathlib.Path`
- if there are some external calls that may or may not support Pathlib,
the argument is immidiately converted to `str`
- if there `self.file_path` is used in a way that it allows for it to
stay pathlib without conversion, is is only converted for the metadata.
Twitter handle: https://twitter.com/mwmajewsk
### Issue
Recently, the new `allow_dangerous_deserialization` flag was introduced
for preventing unsafe model deserialization that relies on pickle
without user's notice (#18696). Since then some LLMs like Databricks
requires passing in this flag with true to instantiate the model.
However, this breaks existing functionality to loading such LLMs within
a chain using `load_chain` method, because the underlying loader
function
[load_llm_from_config](f96dd57501/libs/langchain/langchain/chains/loading.py (L40))
(and load_llm) ignores keyword arguments passed in.
### Solution
This PR fixes this issue by propagating the
`allow_dangerous_deserialization` argument to the class loader iff the
LLM class has that field.
---------
Co-authored-by: Eugene Yurtsev <eyurtsev@gmail.com>
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Create a Class which allows to use the "text2vec" open source embedding
model.
It should install the model by running 'pip install -U text2vec'.
Example to call the model through LangChain:
from langchain_community.embeddings.text2vec import Text2vecEmbeddings
embedding = Text2vecEmbeddings()
bookend.embed_documents([
"This is a CoSENT(Cosine Sentence) model.",
"It maps sentences to a 768 dimensional dense vector space.",
])
bookend.embed_query(
"It can be used for text matching or semantic search."
)
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
Co-authored-by: Eugene Yurtsev <eugene@langchain.dev>
Co-authored-by: Eugene Yurtsev <eyurtsev@gmail.com>
## Description
This PR proposes a modification to the `libs/langchain/dev.Dockerfile`
configuration to copy the `libs/langchain/poetry.lock` into the working
directory. The change aims to address the issue where the Poetry install
command, the last command in the `dev.Dockerfile`, takes excessively
long hours, and to ensure the reproducibility of the poetry environment
in the devcontainer.
## Problem
The `dev.Dockerfile`, prepared for development environments such as
`.devcontainer`, encounters an unending dependency resolution when
attempting the Poetry installation.
### Steps to Reproduce
Execute the following build command:
```bash
docker build -f libs/langchain/dev.Dockerfile .
```
### Current Behavior
The Docker build process gets stuck at the following step, which, in my
experience, did not conclude even after an entire night:
```
=> [langchain-dev-dependencies 4/6] COPY libs/community/ ../community/ 0.9s
=> [langchain-dev-dependencies 5/6] COPY libs/text-splitters/ ../text-splitters/ 0.0s
=> [langchain-dev-dependencies 6/6] RUN poetry install --no-interaction --no-ansi --with dev,test,docs 12.3s
=> => # Updating dependencies
=> => # Resolving dependencies...
```
### Expected Behavior
The Docker build completes in a realistic timeframe. By applying this
PR, the build finishes within a few minutes.
### Analysis
The complexity of LangChain's dependencies has reached a point where
Poetry is required to resolve dependencies akin to threading a needle.
Consequently, poetry install fails to complete in a practical timeframe.
## Solution
The solution for dependency resolution is already recorded in
`libs/langchain/poetry.lock`, so we can use it. When copying
`project.toml` and `poetry.toml`, the `poetry.lock` located in the same
directory should also be copied.
```diff
# Copy only the dependency files for installation
-COPY libs/langchain/pyproject.toml libs/langchain/poetry.toml ./
+COPY libs/langchain/pyproject.toml libs/langchain/poetry.toml libs/langchain/poetry.lock ./
```
## Note
I am not intimately familiar with the historical context of the
`dev.Dockerfile` and thus do not know why `poetry.lock` has not been
copied until now. It might have been an oversight, or perhaps dependency
resolution used to complete quickly even without the `poetry.lock` file
in the past. However, if there are deliberate reasons why copying
`poetry.lock` is not advisable, please just close this PR.
Description:
this change fixes the pydantic validation error when looking up from
GPTCache, the `ChatOpenAI` class returns `ChatGeneration` as response
which is not handled.
use the existing `_loads_generations` and `_dumps_generations` functions
to handle it
Trace
```
File "/home/theinhumaneme/Documents/NebuLogic/conversation-bot/development/scripts/chatbot-postgres-test.py", line 90, in <module>
print(llm.invoke("tell me a joke"))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/theinhumaneme/Documents/NebuLogic/conversation-bot/venv/lib/python3.11/site-packages/langchain_core/language_models/chat_models.py", line 166, in invoke
self.generate_prompt(
File "/home/theinhumaneme/Documents/NebuLogic/conversation-bot/venv/lib/python3.11/site-packages/langchain_core/language_models/chat_models.py", line 544, in generate_prompt
return self.generate(prompt_messages, stop=stop, callbacks=callbacks, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/theinhumaneme/Documents/NebuLogic/conversation-bot/venv/lib/python3.11/site-packages/langchain_core/language_models/chat_models.py", line 408, in generate
raise e
File "/home/theinhumaneme/Documents/NebuLogic/conversation-bot/venv/lib/python3.11/site-packages/langchain_core/language_models/chat_models.py", line 398, in generate
self._generate_with_cache(
File "/home/theinhumaneme/Documents/NebuLogic/conversation-bot/venv/lib/python3.11/site-packages/langchain_core/language_models/chat_models.py", line 585, in _generate_with_cache
cache_val = llm_cache.lookup(prompt, llm_string)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/theinhumaneme/Documents/NebuLogic/conversation-bot/venv/lib/python3.11/site-packages/langchain_community/cache.py", line 807, in lookup
return [
^
File "/home/theinhumaneme/Documents/NebuLogic/conversation-bot/venv/lib/python3.11/site-packages/langchain_community/cache.py", line 808, in <listcomp>
Generation(**generation_dict) for generation_dict in json.loads(res)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/theinhumaneme/Documents/NebuLogic/conversation-bot/venv/lib/python3.11/site-packages/langchain_core/load/serializable.py", line 120, in __init__
super().__init__(**kwargs)
File "/home/theinhumaneme/Documents/NebuLogic/conversation-bot/venv/lib/python3.11/site-packages/pydantic/v1/main.py", line 341, in __init__
raise validation_error
pydantic.v1.error_wrappers.ValidationError: 1 validation error for Generation
type
unexpected value; permitted: 'Generation' (type=value_error.const; given=ChatGeneration; permitted=('Generation',))
```
Although I don't seem to find any issues here, here's an
[issue](https://github.com/zilliztech/GPTCache/issues/585) raised in
GPTCache. Please let me know if I need to do anything else
Thank you
---------
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Few-Shot prompt template may use a `SemanticSimilarityExampleSelector`
that in turn uses a `VectorStore` that does I/O operations.
So to work correctly on the event loop, we need:
* async methods for the `VectorStore` (OK)
* async methods for the `SemanticSimilarityExampleSelector` (this PR)
* async methods for `BasePromptTemplate` and `BaseChatPromptTemplate`
(future work)
This is a small breaking change but I think it should be done as:
* No external dependency needs to be installed anymore for the default
to work
* It is vendor-neutral
This patch updates function "run" to "invoke" in fake_llm.ipynb. Without
this patch, you see following warning.
LangChainDeprecationWarning: The function `run` was deprecated in
LangChain 0.1.0 and will be removed in 0.2.0. Use invoke instead.
Signed-off-by: Masanari Iida <standby24x7@gmail.com>
Fixing some issues for AzureCosmosDBSemanticCache
- Added the entry for "AzureCosmosDBSemanticCache" which was missing in
langchain/cache.py
- Added application name when creating the MongoClient for the
AzureCosmosDBVectorSearch, for tracking purposes.
@baskaryan, can you please review this PR, we need this to go in asap.
These are just small fixes which we found today in our testing.
- **Description:** The `semantic_hybrid_search_with_score_and_rerank`
method of `AzureSearch` contains a hardcoded field name "metadata" for
the document metadata in the Azure AI Search Index. Adding such a field
is optional when creating an Azure AI Search Index, as other snippets
from `AzureSearch` test for the existence of this field before trying to
access it. Furthermore, the metadata field name shouldn't be hardcoded
as "metadata" and use the `FIELDS_METADATA` variable that defines this
field name instead. In the current implementation, any index without a
metadata field named "metadata" will yield an error if a semantic answer
is returned by the search in
`semantic_hybrid_search_with_score_and_rerank`.
- **Issue:** https://github.com/langchain-ai/langchain/issues/18731
- **Prior fix to this bug:** This bug was fixed in this PR
https://github.com/langchain-ai/langchain/pull/15642 by adding a check
for the existence of the metadata field named `FIELDS_METADATA` and
retrieving a value for the key called "key" in that metadata if it
exists. If the field named `FIELDS_METADATA` was not present, an empty
string was returned. This fix was removed in this PR
https://github.com/langchain-ai/langchain/pull/15659 (see
ed1ffca911#).
@lz-chen: could you confirm this wasn't intentional?
- **New fix to this bug:** I believe there was an oversight in the logic
of the fix from
[#1564](https://github.com/langchain-ai/langchain/pull/15642) which I
explain below.
The `semantic_hybrid_search_with_score_and_rerank` method creates a
dictionary `semantic_answers_dict` with semantic answers returned by the
search as follows.
5c2f7e6b2b/libs/community/langchain_community/vectorstores/azuresearch.py (L574-L581)
The keys in this dictionary are the unique document ids in the index, if
I understand the [documentation of semantic
answers](https://learn.microsoft.com/en-us/azure/search/semantic-answers)
in Azure AI Search correctly. When the method transforms a search result
into a `Document` object, an "answer" key is added to the document's
metadata. The value for this "answer" key should be the semantic answer
returned by the search from this document, if such an answer is
returned. The match between a `Document` object and the semantic answers
returned by the search should be done through the unique document id,
which is used as a key for the `semantic_answers_dict` dictionary. This
id is defined in the search result's field named `FIELDS_ID`. I added a
check to avoid any error in case no field named `FIELDS_ID` exists in a
search result (which shouldn't happen in theory).
A benefit of this approach is that this fix should work whether or not
the Azure AI Search Index contains a metadata field.
@levalencia could you confirm my analysis and test the fix?
@raunakshrivastava7 do you agree with the fix?
Thanks for the help!
### Prem SDK integration in LangChain
This PR adds the integration with [PremAI's](https://www.premai.io/)
prem-sdk with langchain. User can now access to deployed models
(llms/embeddings) and use it with langchain's ecosystem. This PR adds
the following:
### This PR adds the following:
- [x] Add chat support
- [X] Adding embedding support
- [X] writing integration tests
- [X] writing tests for chat
- [X] writing tests for embedding
- [X] writing unit tests
- [X] writing tests for chat
- [X] writing tests for embedding
- [X] Adding documentation
- [X] writing documentation for chat
- [X] writing documentation for embedding
- [X] run `make test`
- [X] run `make lint`, `make lint_diff`
- [X] Final checks (spell check, lint, format and overall testing)
---------
Co-authored-by: Anindyadeep Sannigrahi <anindyadeepsannigrahi@Anindyadeeps-MacBook-Pro.local>
Co-authored-by: Bagatur <baskaryan@gmail.com>
Co-authored-by: Erick Friis <erick@langchain.dev>
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
- **Description:** PgVector class always runs "create extension" on init
and this statement crashes on ReadOnly databases (read only replicas).
but wierdly the next create collection etc work even in readOnly
databases
- **Dependencies:** no new dependencies
- **Twitter handle:** @VenOmaX666
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Thank you for contributing to LangChain!
When run command langchain app new my-app, i get this error:
File
"/home/mauricio/.local/lib/python3.8/site-packages/langchain_cli/utils/pyproject.py",
line 15, in <module>
pyproject_toml: Path, local_editable_dependencies: Iterable[tuple[str,
Path]]
TypeError: 'type' object is not subscriptable
This PR fix the error.
The existing default list of separators for the `RecursiveTextSplitter`
assumes spaces are word boundaries. Some languages [don't use spaces
between
words](https://en.wikipedia.org/wiki/Category:Writing_systems_without_word_boundaries)
(Chinese, Japanese, Thai, Burmese).
This PR extends the documentation to explain how to cater for those
languages by adding additional punctuation to the separators and
zero-width spaces which are used by some typesetters and will assist the
splitter to not split in words.
Ideally, **these separators could be a constant in the module** but for
now, defining them in the documentation is a start.
**Description:**
- minor PR to speed up onboarding by not trying to add a dataset, if a
model is already present.
- replace batch publish API with streaming when single events are
published.
**Dependencies:** any dependencies required for this change
**Twitter handle:** behalder
Co-authored-by: Barun Halder <barun@fiddler.ai>
This PR aims to enhance the documentation for TiDB integration, driven
by feedback from our users. It provides detailed introductions to key
features, ensuring developers can fully leverage TiDB for AI application
development.
**Description:**
Expanding version in all the Confluence API calls so to get when the
page was last modified/created in all cases.
**Issue:** #12812
**Twitter handle:** zzste
This PR adds code to make sure that the correct base URL is being
created for the Azure Cognitive Search retriever. At the moment an
incorrect base URL is being generated. I think this is happening because
the original code was based on a depreciated API version. No
dependencies need to be added. I've also added more context to the test
doc strings.
I should also note that ACS is now Azure AI Search. I will open a
separate PR to make these changes as that would be a breaking change and
should potentially be discussed.
Twitter: @marlene_zw
- No new tests added, however the current ACS retriever tests are now
passing when I run them.
- Code was linted.
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
- **Description:** This commit introduces support for the newly
available GPU index types introduced in Milvus 2.4 within the LangChain
project's `milvus.py`. With the release of Milvus 2.4, a range of
GPU-accelerated index types have been added, offering enhanced search
capabilities and performance optimizations for vector search operations.
This update ensures LangChain users can fully utilize the new
performance benefits for vector search operations.
- Reference: https://milvus.io/docs/gpu_index.md
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Corrected a broken link within the semantic-chunker.ipynb notebook,
ensuring that users can access the referenced resource.
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
This patch fixes the #18022 issue, converting the SimSIMD internal
zero-copy outputs to NumPy.
I've also noticed, that oftentimes `dtype=np.float32` conversion is used
before passing to SimSIMD. Which numeric types do LangChain users
generally care about? We support `float64`, `float32`, `float16`, and
`int8` for cosine distances and `float16` seems reasonable for
practically any kind of embeddings and any modern piece of hardware, so
we can change that part as well 🤗
- **Description:** Added support for lower-case and mixed-case names
The names for tables and columns previouly had to be UPPER_CASE.
With this enhancement, also lower_case and MixedCase are supported,
- **Issue:** N/A
- **Dependencies:** no new dependecies added
- **Twitter handle:** @sapopensource
- **Description:** Since the implicit `__call__` has been deprecated in
favor of `invoke`, the local_llms article also needed to be updated.
This article was my introduction to Lanchain, and as it was helpful in
getting me setup with running LLMs locally, it is nice to not have any
warnings when running the example code. With this change, the warnings
go away when running the example code.
- **Issue:** N/A
- **Dependencies:** N/A
- **Twitter handle:** clarkerican
Previous PR passed _parser attribute which apparently is not meant to be
used by user code and causes non deterministic failures on CI when
testing the transform and a transform methods. Reverting this change
temporarily.
This mitigates a security concern for users still using older versions of libexpat that causes an attacker to compromise the availability of the system if an attacker manages to surface malicious payload to this XMLParser.
**Description:** This change passes through `batch_size` to
`add_documents()`/`aadd_documents()` on calls to `index()` and
`aindex()` such that the documents are processed in the expected batch
size.
**Issue:** #19415
**Dependencies:** N/A
**Twitter handle:** N/A
Updated `HuggingFacePipeline` docs to be in sync with list of supported
tasks, including translation.
- [x] **PR title**: "community: Update docs for `HuggingFacePipeline`"
- Where "package" is whichever of langchain, community, core,
experimental, etc. is being modified. Use "docs: ..." for purely docs
changes, "templates: ..." for template changes, "infra: ..." for CI
changes.
- Example: "community: add foobar LLM"
- [x] **PR message**:
- **Description:** Update docs for `HuggingFacePipeline`, was earlier
missing `translation` as a valid task
- **Issue:** N/A
- **Dependencies:** N/A
- **Twitter handle:** None
- [x] **Add tests and docs**:
- [x] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
**Description:**
This PR adds [Dappier](https://dappier.com/) for the chat model. It
supports generate, async generate, and batch functionalities. We added
unit and integration tests as well as a notebook with more details about
our chat model.
**Dependencies:**
No extra dependencies are needed.
- **Description:** [CVE
2024-21503](https://www.cve.org/CVERecord?id=CVE-2024-21503) was
recently identified. The python linter "black" suffers from a potential
Regex-related denial of service attack. Updated version from the
vulnerable 24.2.0 to the patched 24.3.0.
- **Issue:** N/A
- **Dependencies:** The 'black' package in both `langchain` (top-level)
and `templates/python-lint`.
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
DuckDB has a cosine similarity function along list and array data types,
which can be used as a vector store.
- **Description:** The latest version of DuckDB features a cosine
similarity function, which can be used with its support for list or
array column types. This PR surfaces this functionality to langchain.
- **Dependencies:** duckdb 0.10.0
- **Twitter handle:** @igocrite
---------
Co-authored-by: Eugene Yurtsev <eyurtsev@gmail.com>
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
**Description:** Update s3_file.py to use arguments **mode** and
**post_processors** from the base class **UnstructuredBaseLoader** to
include more metadata about the files from the S3 bucket such as
*'page_number', 'languages'* etc.
**Issue:** NA
**Dependencies:** None
**Twitter handle:** preak95
---------
Co-authored-by: ccurme <chester.curme@gmail.com>
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Looking at tokens / page of our docs, we see a few outliers:
<img width="761" alt="image"
src="https://github.com/langchain-ai/langchain/assets/122662504/677aa2d6-0a29-45e4-882a-db2bbf46d02b">
It is due to non-rendering images in one case, and output spamming.
Clean these, along with other cases of excessing output spamming in
docs.
All get sucked into chat-langchain for retrieval.
Thank you for contributing to LangChain!
bilibili-api-python use https://github.com/Nemo2011/bilibili-api repo.
Change to the correct address.
- [x] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
**Description:** Update module imports for Fireworks documentation
**Issue:** Module imports not present or in incorrect location
**Dependencies:** None
**Description:** Update import paths and move to lcel for llama.cpp
examples
**Issue:** Update import paths to reflect package refactoring and move
chains to LCEL in examples
**Dependencies:** None
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
**Description:** Invoke callback prior to yielding token for BaseOpenAI
& OpenAIChat
**Issue:** [Callback for on_llm_new_token should be invoked before the
token is yielded by the model
#16913](https://github.com/langchain-ai/langchain/issues/16913)
**Dependencies:** None
**Description:** Invoke callback prior to yielding token for Fireworks
**Issue:** [Callback for on_llm_new_token should be invoked before the
token is yielded by the model
#16913](https://github.com/langchain-ai/langchain/issues/16913)
**Dependencies:** None
**Description:** Moving FireworksEmbeddings documentation to the
location docs/integration/text_embedding/ from langchain_fireworks/docs/
**Issue:** FireworksEmbeddings documentation was not in the correct
location
**Dependencies:** None
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
I have a small dataset, and I tried to use docarray:
``DocArrayHnswSearch ``. But when I execute, it returns:
```bash
raise ImportError(
ImportError: Could not import docarray python package. Please install it with `pip install "langchain[docarray]"`.
```
Instead of docarray it needs to be
```bash
docarray[hnswlib]
```
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
RecursiveUrlLoader does not currently provide an option to set
`base_url` other than the `url`, though it uses a function with such an
option.
For example, this causes it unable to parse the
`https://python.langchain.com/docs`, as it returns the 404 page, and
`https://python.langchain.com/docs/get_started/introduction` has no
child routes to parse.
`base_url` allows setting the `https://python.langchain.com/docs` to
filter by, while the starting URL is anything inside, that contains
relevant links to continue crawling.
I understand that for this case, the docusaurus loader could be used,
but it's a common issue with many websites.
---------
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
- [ ] **PR message**: ***Delete this entire checklist*** and replace
with
- **Description:** a description of the change
- **Issue:** the issue # it fixes, if applicable
- **Dependencies:** any dependencies required for this change
- **Twitter handle:** if your PR gets announced, and you'd like a
mention, we'll gladly shout you out!
- [ ] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [ ] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
---------
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
**Description:** Delete MistralAIEmbeddings usage document from folder
partners/mistralai/docs
**Issue:** The document is present in the folder docs/docs
**Dependencies:** None
**Description:** Invoke callback prior to yielding token for llama.cpp
**Issue:** [Callback for on_llm_new_token should be invoked before the
token is yielded by the model
#16913](https://github.com/langchain-ai/langchain/issues/16913)
**Dependencies:** None
```python
from langchain.agents import tool
from langchain_mistralai import ChatMistralAI
llm = ChatMistralAI(model="mistral-large-latest", temperature=0)
@tool
def get_word_length(word: str) -> int:
"""Returns the length of a word."""
return len(word)
tools = [get_word_length]
llm_with_tools = llm.bind_tools(tools)
llm_with_tools.invoke("how long is the word chrysanthemum")
```
currently raises
```
AttributeError: 'dict' object has no attribute 'model_dump'
```
Same with `.with_structured_output`
```python
from langchain_mistralai import ChatMistralAI
from langchain_core.pydantic_v1 import BaseModel
class AnswerWithJustification(BaseModel):
"""An answer to the user question along with justification for the answer."""
answer: str
justification: str
llm = ChatMistralAI(model="mistral-large-latest", temperature=0)
structured_llm = llm.with_structured_output(AnswerWithJustification)
structured_llm.invoke("What weighs more a pound of bricks or a pound of feathers")
```
This appears to fix.
For prompt templates with only 1 variable (common in e.g.,
MessageGraph), it's convenient to wrap the incoming object in the
variable before formatting.
The downside of this, of course, would be that some number of
invocations will successfully format when the user may have intended to
format it properly before
This is a basic VectorStore implementation using an in-memory dict to
store the documents.
It doesn't need any extra/optional dependency as it uses numpy which is
already a dependency of langchain.
This is useful for quick testing, demos, examples.
Also it allows to write vendor-neutral tutorials, guides, etc...
Classes and functions defined in __init__.py are not parsed into the API
Reference.
For example:
- libs/core/langchain_core/messages/__init__.py : AnyMessage,
MessageLikeRepresentation, get_buffer_string(), messages_from_dict(),
...
Opinionated: __init__.py is not a typical place to define artifacts.
Moved artifacts from __init__ into utils.py.
Added `MessageLikeRepresentation` to __all__ since it is used outside of
`messages`, for example, in
`libs/core/langchain_core/language_models/base.py`
Added `_message_from_dict` to __all__ since it is used outside of
`messages`(???) I would add `message_from_dict` (without underscore) as
an alias. Please, advise.
Covered by tests in
`libs/core/tests/unit_tests/language_models/chat_models/test_base.py`,
`libs/core/tests/unit_tests/language_models/llms/test_base.py` and
`libs/core/tests/unit_tests/runnables/test_runnable_events.py`
**Description:**
Currently, `CacheBackedEmbeddings` computes vectors for *all* uncached
documents before updating the store. This pull request updates the
embedding computation loop to compute embeddings in batches, updating
the store after each batch.
I noticed this when I tried `CacheBackedEmbeddings` on our 30k document
set and the cache directory hadn't appeared on disk after 30 minutes.
The motivation is to minimize compute/data loss when problems occur:
* If there is a transient embedding failure (e.g. a network outage at
the embedding endpoint triggers an exception), at least the completed
vectors are written to the store instead of being discarded.
* If there is an issue with the store (e.g. no write permissions), the
condition is detected early without computing (and discarding!) all the
vectors.
**Issue:**
Implements enhancement #18026.
**Testing:**
I was unable to run unit tests; details in [this
post](https://github.com/langchain-ai/langchain/discussions/15019#discussioncomment-8576684).
---------
Signed-off-by: chrispy <chrispy@synopsys.com>
Co-authored-by: Eugene Yurtsev <eyurtsev@gmail.com>
## Description
Semantic Cache can retrieve noisy information if the score threshold for
the value is too low. Adding the ability to set a `score_threshold` on
cache construction can allow for less noisy scores to appear.
- [x] **Add tests and docs**
1. Added tests that confirm the `score_threshold` query is valid.
- [x] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
---------
Co-authored-by: Erick Friis <erick@langchain.dev>
The `retryFailed` option will retry all failed links, once at a time
with the goal of not triggering bot protection
`microsoft.com` is now hard coded into the whitelist
Classes and functions defined in __init__.py are not parsed into the API
Reference.
For example: libs/core/langchain_core/globals/__init__.py :
`set_verbose` `get_llm_cache`, `set_llm_cache`, ...
And the whole `langchain_core.globals` namespace is not visible in the
API Reference. The refactoring is just file renaming.
- **Description:** Enhanced the `BaseChatModel` to support an
`Optional[Union[bool, BaseCache]]` type for the `cache` attribute,
allowing for both boolean flags and custom cache implementations.
Implemented logic within chat model methods to utilize the provided
custom cache implementation effectively. This change aims to provide
more flexibility in caching strategies for chat models.
- **Issue:** Implements enhancement request #17242.
- **Dependencies:** No additional dependencies required for this change.
---------
Co-authored-by: Eugene Yurtsev <eyurtsev@gmail.com>
Thank you for contributing to LangChain!
- [x] **PR title**: "package: description"
- Where "package" is whichever of langchain, community, core,
experimental, etc. is being modified. Use "docs: ..." for purely docs
changes, "templates: ..." for template changes, "infra: ..." for CI
changes.
- Example: "community: add foobar LLM"
- **PR message**:
- **Description:** Update the slack toolkit doc to use an agent that
support multiple inputs. Using ReAct agent will cause a ValidationError
when invoking the slack tools. This is because the agent return a string
like `'{"channel": "C05LDF54S21", "message": "Hello, world!"}'` but the
ReAct agent does not support multiple inputs.
- **Issue:** This is related to this
[Discussion#18083](https://github.com/langchain-ai/langchain/discussions/18083)
- **Dependencies:** No dependencies required
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
---------
Co-authored-by: Chester Curme <chester.curme@gmail.com>
Changing OpenAIAssistantRunnable.create_assistant to send the `file_ids`
parameter to openai.beta.assistants.create
Co-authored-by: Frederico Wu <fred.diaswu@coxautoinc.com>
**Description**:
this PR enable VectorStore autoconfiguration for Infinispan: if
metadatas are only of basic types, protobuf
config will be automatically generated for the user.
When creating a new index, if we use a retrieval strategy that expects a
model to be deployed in Elasticsearch, check if a model with this name
is indeed deployed before creating an index. This lowers the probability
to get into a state in which an index was created with a faulty model
ID, which cannot be overwritten any more (the index has to manually be
deleted).
Add `keep_alive` parameter to control how long the model will stay
loaded into memory with Ollama。
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
**Description**
This PR adds some missing details from the "Split by tokens" page in the
documentation. Specifically:
- The `.from_tiktoken_encoder()` class methods for both the
`CharacterTextSplitter` and `RecursiveCharacterTextSplitter` default to
the old `gpt-2` encoding. I've added a comment to suggest specifying
`model_name` or `encoding`
- The docs didn't mention that the `from_tiktoken_encoder()` class
method passes additional kwargs down to the constructor of the splitter.
I only discovered this by reading the source code
- Added an example of using the `.from_tiktoken_encoder()` class method
with `RecursiveCharacterTextSplitter` which is the recommended approach
for most scenarios above `CharacterTextSplitter`
- Added a warning that `TokenTextSplitter` can split characters which
have multiple tokens (e.g. 猫 has 3 cl100k_base tokens) between multiple
chunks which creates malformed Unicode strings and should not be used in
these situations.
Side note: I think the default argument of `gpt2` for
`.from_tiktoken_encoder()` should be updated?
**Twitter handle** anthonypjshaw
---------
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
Issue : For functions which have an argument with the name 'title', the
convert_pydantic_to_openai_function generates an incorrect output and
omits the argument all together. This is because the _rm_titles function
removes all instances of the the key 'title' from the output.
Description : Updates the _rm_titles function to check the presence of
the 'type' key as well before removing the 'title' key. As the title key
that we wish to omit always has a type key along with it.
Potential gap if there is a function defined which has both title and
key as argument names, in which case this would fail. Maybe we could set
a filter on the function argument names and reject those with keyword
argument names.
No dependencies. Passed all tests.
- [x] **PR title**: "package: description"
- Where "package" is whichever of langchain, community, core,
experimental, etc. is being modified. Use "docs: ..." for purely docs
changes, "templates: ..." for template changes, "infra: ..." for CI
changes.
- Example: "community: add foobar LLM"
- [x] **PR message**: ***Delete this entire checklist*** and replace
with
- **Description:** a description of the change
- **Issue:** the issue # it fixes, if applicable
- **Dependencies:** any dependencies required for this change
- **Twitter handle:** if your PR gets announced, and you'd like a
mention, we'll gladly shout you out!
- [x] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [x] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
---------
Co-authored-by: Eugene Yurtsev <eyurtsev@gmail.com>
- **Description:** There was no formatter for mistral models for Azure
ML endpoints. Adding that, plus a configurable timeout (it was hard
coded before)
- **Dependencies:** none
- **Twitter handle:** @tjaffri @docugami
Thank you for contributing to LangChain!
- [ ] **PR title**: "package: description"
- Where "package" is whichever of langchain, community, core,
experimental, etc. is being modified. Use "docs: ..." for purely docs
changes, "templates: ..." for template changes, "infra: ..." for CI
changes.
- Example: "community: add foobar LLM"
- [ ] **PR message**: ***Delete this entire checklist*** and replace
with
- **Description:** a description of the change
- **Issue:** the issue # it fixes, if applicable
- **Dependencies:** any dependencies required for this change
- **Twitter handle:** if your PR gets announced, and you'd like a
mention, we'll gladly shout you out!
- [ ] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [x] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
Thank you for contributing to LangChain!
- [x] **PR title**: "Updating format of pip install in two files of
docs/cookbook"
- pip install is not reflecting properly in some of the files in
cookbook
- Example:
[docs/expression_language/cookbook/sql_db](https://python.langchain.com/docs/expression_language/cookbook/sql_db)
- [x] **PR message**: Updating format of pip install in two files of
docs/cookbook
- **Description:** a description of the change
- **Issue:** #19197
- Note - let's do squash merge for the PR
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
add **kwargs in add_documents for upsert, to make it use for other
argument also.
Lets use this, it was unused as of now.
- [ ] **PR title**: "package: description"
- Where "package" is whichever of langchain, community, core,
experimental, etc. is being modified. Use "docs: ..." for purely docs
changes, "templates: ..." for template changes, "infra: ..." for CI
changes.
- Example: "community: add foobar LLM"
- [ ] **PR message**: ***Delete this entire checklist*** and replace
with
- **Description:** a description of the change
- **Issue:** the issue # it fixes, if applicable
- **Dependencies:** any dependencies required for this change
- **Twitter handle:** if your PR gets announced, and you'd like a
mention, we'll gladly shout you out!
- [ ] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [ ] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
Co-authored-by: Rohit Gupta <rohit.gupta2@walmart.com>
## Description
This PR modifies the settings in `libs/langchain/dev.Dockerfile` to
ensure that the `text-splitters` directory is copied before the poetry
installation process begins.
Without this modification, the `docker build` command fails for
`dev.Dockerfile`, preventing the setup of some development environments,
including `.devcontainer`.
## Bug Details
### Repro
Run the following command:
```bash
docker build -f libs/langchain/dev.Dockerfile .
```
### Current Behavior
The docker build command fails, raising the following error:
```
...
=> [langchain-dev-dependencies 4/5] COPY libs/community/ ../community/ 0.4s
=> ERROR [langchain-dev-dependencies 5/5] RUN poetry install --no-interaction --no-ansi --with dev,test,docs 1.1s
------
> [langchain-dev-dependencies 5/5] RUN poetry install --no-interaction --no-ansi --with dev,test,docs:
#13 0.970
#13 0.970 Directory ../text-splitters does not exist
------
executor failed running [/bin/sh -c poetry install --no-interaction --no-ansi --with dev,test,docs]: exit code: 1
```
### Expected Behavior
The `docker build` command successfully completes without the poetry
error.
### Analysis
The error occurs because the `text-splitters` directory is not copied
into the build environment, unlike the other packages under the `libs`
directory. I suspect that the `COPY` setting was overlooked since
`text-splitters` was separated in a recent PR.
## Fix
Add the following lines to the `libs/langchain/dev.Dockerfile`:
```dockerfile
# Copy the text-splitters library for installation
COPY libs/text-splitters/ ../text-splitters/
```
- **Description:** Tests fail to do value lookup because it does not
specify the index name
- **Issue:** the issue # Failing integration test
- [x] **Add tests and docs**: Tests now pass
- [x] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
The root run id (~trace id's) is useful for assigning feedback, but the
current recommended approach is to use callbacks to retrieve it, which
has some drawbacks:
1. Doesn't work for streaming until after the first event
2. Doesn't let you call other endpoints with the same trace ID in
parallel (since you have to wait until the call is completed/started to
use
This PR lets you provide = "run_id" in the runnable config.
Couple considerations:
1. For batch calls, we split the trace up into separate trees (to permit
better rendering). We keep the provided run ID for the first one and
generate a unique one for other elements of the batch.
2. For nested calls, the provided ID is ONLY used on the top root/trace.
```
chain.invoke("foo", {"run_id": uuid.uuid4()})
```
Classes are missed in __all__ and in different places of __init__.py
- BaichuanLLM
- ChatDatabricks
- ChatMlflow
- Llamafile
- Mlflow
- Together
Added classes to __all__. I also sorted __all__ list.
---------
Co-authored-by: Erick Friis <erick@langchain.dev>
Thank you for contributing to LangChain!
- [x] **PR title**: "community: deprecate DocugamiLoader"
- [x] **PR message**: Deprecate the langchain_community and use the
docugami_langchain DocugamiLoader
---------
Co-authored-by: Kenzie Mihardja <kenzie28@cs.washington.edu>
## Description
* In memory cache easily gets out of sync with the server cache, so we
will remove it entirely to reduce the issues around invalidated caches.
## Dependencies
None
- [x] If you're adding a new integration, please include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [x] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Co-authored-by: Erick Friis <erick@langchain.dev>
## Description
Returning the embedding is not necessary in the vector search
functionality unless specified as a debugging step. This change defaults
the behavior such that the server _only_ returns the embedding key if
explicitly requested, such as in the case of
`max_marginal_relevance_search`.
- [x] **Add tests and docs**: If you're adding a new integration, please
include
* Added `test_from_documents_no_embedding_return`
- [x] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
---------
Co-authored-by: Erick Friis <erick@langchain.dev>
- Description:
- Updated the import path for `StreamingStdOutCallbackHandler` in the
streaming response example within `huggingface_endpoint.py`. This change
corrects the import statement to reflect the actual location of
`StreamingStdOutCallbackHandler` in
`langchain_core.callbacks.streaming_stdout`.
- Issue:
- None
- Dependencies:
- No additional dependencies are required for this change.
- Twitter handle:
- None
## Note:
I have tested this change locally and confirmed that the
`StreamingStdOutCallbackHandler` works as expected with the updated
import path. This PR does not require the addition of new tests since it
is a correction to documentation/examples rather than functional code.
- [x] **Support for translation**: "community: Add support for
translation in `HuggingFacePipeline`"
- [x] **Add support for translation in `HuggingFacePipeline`**:
- **Description:** Add support for translation in `HuggingFacePipeline`,
which earlier used to support only text summarization and generation.
- **Issue:** N/A
- **Dependencies:** N/A
- **Twitter handle:** None
- Description:
- This pull request is to fix a bug where page numbers were not set
correctly. In the current code, all chunks share the same metadata
object doc_metadata, so the page number is set with the same value for
all documents. To fix this, I changed to using separate metadata objects
for each chunk.
- Issue:
- None
- Dependencies:
- No additional dependencies are required for this change.
- Twitter handle:
- @eycjur
- Test
- Even if it's not a bug, there are cases where everything ends up with
the same number of pages, so it's very difficult for me to write
integration tests.
Thank you for contributing to LangChain!
- [ ] **PR title**: "package: description"
- Where "package" is whichever of langchain, community, core,
experimental, etc. is being modified. Use "docs: ..." for purely docs
changes, "templates: ..." for template changes, "infra: ..." for CI
changes.
- Example: "community: add foobar LLM"
- [ ] **PR message**: ***Delete this entire checklist*** and replace
with
- **Description:** a description of the change
- **Issue:** the issue # it fixes, if applicable
- **Dependencies:** any dependencies required for this change
- **Twitter handle:** if your PR gets announced, and you'd like a
mention, we'll gladly shout you out!
- [ ] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [ ] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
I think that cell type for pip command may be 'code'.
Please check, thank you :)
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
Line `from langchain_openai import ChatOpenAI` is put twice in Get
Started / Serving with LangServe section.
Imports on lines 559 and 566 are identical
Co-authored-by: Vitalii <vitalii@localhost>
**Description:**
#18040 forces `fastembed>2.0`, and this causes dependency conflicts with
the new `unstructured` package (different `onnxruntime`). There may be
other dependency conflicts.. The only way to use
`langchain-community>=0.0.28` is rollback to `unstructured 0.10.X`. But
new `unstructured` contains many fixes.
This PR allows to use both `fastembed` `v1` and `v2`.
How to reproduce:
`pyproject.toml`:
```toml
[tool.poetry]
name = "depstest"
version = "0.0.0"
description = "test"
authors = ["<dev@example.org>"]
[tool.poetry.dependencies]
python = ">=3.10,<3.12"
langchain-community = "^0.0.28"
fastembed = "^0.2.0"
unstructured = {extras = ["pdf"], version = "^0.12"}
```
```bash
$ poetry lock
```
Co-authored-by: Sergey Kozlov <sergey.kozlov@ludditelabs.io>
- **Description:** This modification addresses the issue of mutable
default parameters in functions. In the original code, the `chunks`
parameter is defaulted to a list containing an empty dictionary, which
is mutable. Since default parameters in Python are evaluated only once
at function definition time, modifications to the parameter would
persist across future calls. By changing the default to `None` and
checking/initializing within the function, a new list is created for
each call, thus avoiding potential issues.
---------
Co-authored-by: sixiang <sixiang@lixiang.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
**Description:** Update stales link in Together AI documentation
**Issue:** Some links pointed to legacy webpages on the Together AI
website
**Dependencies:** None
**Lint and test**: `make format`, `make lint` were run
**Description:** Update docstring of Together class to show example and
update API URL
**Issue:** Improves usability
**Dependencies:** None
**Lint and test**: `make format`, `make lint` and `make test` were run
- [ ] **PR title**: "docs: correction in
"https://github.com/langchain-ai/langchain/blob/master/docs/docs/get_started/quickstart.mdx",
line 289".
- Where "package" is whichever of langchain, community, core,
experimental, etc. is being modified. Use "docs: ..." for purely docs
changes, "templates: ..." for template changes, "infra: ..." for CI
changes.
- Example: "community: add foobar LLM"
- [ ] **PR message**:
- Corrected the spelling mistake
- #18981
Fixed Grammar in Considerations of Model I/O Concepts documentation page
- Update concepts.mdx
Page Link:
https://python.langchain.com/docs/modules/model_io/concepts#considerations
- **Description:** Fixed Grammar in Considerations of Model I/O
Documentation Page
- **Issue:** "to work well with the model are you using" # "to work well
with the model you are using"
- **Dependencies:** None
- **Twitter handle:** @Anubhav_Madhav
(https://twitter.com/Anubhav_Madhav)
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
## Description
This PR addresses a documentation issue in the
[Indexing](https://python.langchain.com/docs/modules/data_connection/indexing)
page. Specifically, it corrects the execution results of the Jupyter
notebook under the
[Source](https://python.langchain.com/docs/modules/data_connection/indexing#source)
section, which were broken as detailed below.
## Problem
The execution results following the statement, `This should delete the
old versions of documents associated with doggy.txt source and replace
them with the new versions.`, appear to be incorrect, as described
below.
### Current Behavior
- For some reason, the `index` function fails to add the new content of
`doggy.txt`. Although it deletes the document objects associated with
the `doggy.txt` source, it does not add the objects in
`changed_doggy_docs`. Consequently, the execution result displays
`num_added: 0`.
- This unexpected behavior also impacts the results of
`vectorstore.similarity_search("dog", k=30)`, showing only the contents
of `kitty.txt`. It appears as though the contents of `doggy.txt` have
been completely removed from the index:
```
Document(page_content='tty kitty', metadata={'source': 'kitty.txt'}),
Document(page_content='tty kitty ki', metadata={'source': 'kitty.txt'}),
Document(page_content='kitty kit', metadata={'source': 'kitty.txt'})]
```
### Expected Behavior
- The `index` function should successfully add the objects in
`changed_doggy_docs` after removing the old content of `doggy.txt`. The
anticipated execution result is `num_added: 2`.
- Subsequently, the modified content of `doggy.txt` should appear in the
results of `vectorstore.similarity_search("dog", k=30)` as follows:
```
[Document(page_content='woof woof', metadata={'source': 'doggy.txt'}),
Document(page_content='woof woof woof', metadata={'source': 'doggy.txt'}),
Document(page_content='tty kitty', metadata={'source': 'kitty.txt'}),
Document(page_content='tty kitty ki', metadata={'source': 'kitty.txt'}),
Document(page_content='kitty kit', metadata={'source': 'kitty.txt'})]
```
## Fix
I reran `docs/docs/modules/data_connection/indexing.ipynb` and have
included the diff in this PR.
- Description: [a description of the change] Add in code documentation
to core Runnable with_fallbacks method (docs only)
- Issue: the issue #18804
@eyurtsev PTAL
Docs fix: replace column name search with source.
The Xata integration expects metadata column named "source".
The docs suggest the name "search", which if used, yields the following
error:
```
File "/usr/local/lib/python3.11/site-packages/langchain_community/vectorstores/xata.py", line 95, in _add_vectors
raise Exception(f"Error adding vectors to Xata: {r.status_code} {r}")
Exception: Error adding vectors to Xata: 400 {'errors': [{'status': 400, 'message': 'invalid record: column [source]: column not found'}]}
```
**Description:** Many LLM steps complete in sub-second duration, which
can lead to non-collection of duration field for Fiddler. This PR
updates duration from seconds to milliseconds.
**Issue:** [INTERNAL] FDL-17568
**Dependencies:** NA
**Twitter handle:** behalder
Co-authored-by: Barun Halder <barun@fiddler.ai>
- **Description:** Handling fallbacks when calling async streaming for a
LLM that doesn't support it.
- **Issue:** #18920
- **Twitter handle:**@maximeperrin_
---------
Co-authored-by: Maxime Perrin <mperrin@doing.fr>
**Description:** This PR adds updates the fiddler events schema to also
pass user feedback, and llm status to fiddler
**Tickets:** [INTERNAL] FDL-17559
**Dependencies:** NA
**Twitter handle:** behalder
Co-authored-by: Barun Halder <barun@fiddler.ai>
- **Description:** This modification adds pydantic input definition for
sql_database tools. This helps for function calling capability in
LangGraph. Since actions nodes will usually check for the args_schema
attribute on tools, This update should make these tools compatible with
it (only implemented on the InfoSQLDatabaseTool)
- **Issue:** N/A
- **Dependencies:** N/A
- **Twitter handle:** juanfe8881
poetry can't reliably handle resolving the number of optional "extended
test" dependencies we have. If we instead just rely on pip to install
extended test deps in CI, this isn't an issue.
- **Description:** add async tests, add tokenize support
- **Dependencies:**
[ibm-watsonx-ai](https://pypi.org/project/ibm-watsonx-ai/),
- **Tag maintainer:**
Please make sure your PR is passing linting and testing before
submitting. Run `make format`, `make lint` and `make test` to check this
locally -> ✅
Please make sure integration_tests passing locally -> ✅
---------
Co-authored-by: Erick Friis <erick@langchain.dev>
This PR makes the following updates in the pgvector database:
1. Use JSONB field for metadata instead of JSON
2. Update operator syntax to include required `$` prefix before the
operators (otherwise there will be name collisions with fields)
3. The change is non-breaking, old functionality is still the default,
but it will emit a deprecation warning
4. Previous functionality has bugs associated with comparisons due to
casting to text (so lexical ordering is used incorrectly for numeric
fields)
5. Adds an a GIN index on the JSONB field for more efficient querying
**PR message**: ***Delete this entire checklist*** and replace with
- **Description:** [a description of the change](docs: Add in code
documentation to core Runnable assign method)
- **Issue:** the issue #18804
Fixed typo in line 661 - from 'mimimize' to 'minimize
- [ ] **PR message**:
- **Description:** Fixed typo in streaming document - change 'mimimize'
to 'minimize
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
- **Description:** When calling the `_stream_log_implementation` from
the `astream_log` method in the `Runnable` class, it is not handing over
the `kwargs` argument. Therefore, even if i want to customize APIHandler
and implement additional features with additional arguments, it is not
possible. Conversely, the `astream_events` method normally handing over
the `kwargs` argument.
- **Issue:** https://github.com/langchain-ai/langchain/issues/19054
- **Dependencies:**
- **Twitter handle:** if your PR gets announced, and you'd like a
mention, we'll gladly shout you out!
Co-authored-by: hyungwookyang <hyungwookyang@worksmobile.com>
Thank you for contributing to LangChain!
- [ ] **PR title**: "package: description"
- Where "package" is whichever of langchain, community, core,
experimental, etc. is being modified. Use "docs: ..." for purely docs
changes, "templates: ..." for template changes, "infra: ..." for CI
changes.
- Example: "community: add foobar LLM"
- [ ] **PR message**: ***Delete this entire checklist*** and replace
with
- **Description:** a description of the change
- **Issue:** the issue # it fixes, if applicable
- **Dependencies:** any dependencies required for this change
- **Twitter handle:** if your PR gets announced, and you'd like a
mention, we'll gladly shout you out!
- [ ] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [ ] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
- **Description:** This change fixes a bug where attempts to load data
from Notion using the NotionDBLoader resulted in a 400 Bad Request
error. The issue was traced to the unconditional addition of an empty
'filter' object in the request payload, which Notion's API does not
accept. The modification ensures that the 'filter' object is only
included in the payload when it is explicitly provided and not empty,
thus preventing the 400 error from occurring.
- **Issue:** Fixes
[#18009](https://github.com/langchain-ai/langchain/issues/18009)
- **Dependencies:** None
- **Twitter handle:** @gunnzolder
Co-authored-by: Anton Parkhomenko <anton@merge.rocks>
**Description:**
Updates to LangChain-MongoDB documentation: updates to the Atlas vector
search index definition
**Issue:**
NA
**Dependencies:**
NA
**Twitter handle:**
iprakul
This PR adds `batch as completed` method to the standard Runnable
interface. It takes in a list of inputs and yields the corresponding
outputs as the inputs are completed.
Add documentation notebook for `ElasticsearchRetriever`.
## Dependencies
- [ ] Release new `langchain-elasticsearch` version 0.2.0 that includes
`ElasticsearchRetriever`
**Description:** Circular dependencies when parsing references leading
to `RecursionError: maximum recursion depth exceeded` issue. This PR
address the issue by handling previously seen refs as in any typical DFS
to avoid infinite depths.
**Issue:** https://github.com/langchain-ai/langchain/issues/12163
**Twitter handle:** https://twitter.com/theBhulawat
- [x] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [x] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
Co-authored-by: Bagatur <22008038+baskaryan@users.noreply.github.com>
**Description:** Refactor code of FAISS vectorcstore and update the
related documentation.
Details:
- replace `.format()` with f-strings for strings formatting;
- refactor definition of a filtering function to make code more readable
and more flexible;
- slightly improve efficiency of
`max_marginal_relevance_search_with_score_by_vector` method by removing
unnecessary looping over the same elements;
- slightly improve efficiency of `delete` method by using set data
structure for checking if the element was already deleted;
**Issue:** fix small inconsistency in the documentation (the old example
was incorrect and unappliable to faiss vectorstore)
**Dependencies:** basic langchain-community dependencies and `faiss`
(for CPU or for GPU)
**Twitter handle:** antonenkodev
Issue : _call method of LLMRouterChain uses predict_and_parse, which is
slated for deprecation.
Description : Instead of using predict_and_parse, this replaces it with
individual predict and parse functions.
Added deps:
- `@supabase/supabase-js` - for sending inserts
- `supabase` - dev dep, for generating types via cli
- `dotenv` for loading env vars
Added script:
- `yarn gen` - will auto generate the database schema types using the
supabase CLI. Not necessary for development, but is useful. Requires
authing with the supabase CLI (will error out w/ instructions if you're
not authed).
Added functionality:
- pulls users IP address (using a free endpoint: `https://api.ipify.org`
so we can filter out abuse down the line)
TODO:
- [x] add env vars to vercel
**Description:** Update the docstring of OpenAI, OpenAIEmbeddings and
ChatOpenAI classes
**Issue:** Update import module paths to the current LangChain API
**Dependencies:** None
**Lint and test**: `make format` and `make lint` were run
This incorporates the review comments from langchain-ai/langchain#18637
which I closed due to an issue I had in updating that pr branch
---------
Co-authored-by: Erick Friis <erick@langchain.dev>
community: fix - change sparkllm spark_app_url to spark_api_url
- **Description:**
- Change the variable name from `sparkllm spark_app_url` to
`spark_api_url` in the community package.
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
**Description:**
Variable name was `openai_poem` but it didn't pass in the `"prompt":
"poem"` config, so the examples were showing a joke being returned from
a variable called `*_poem`.
We could have gone one of two ways:
1. Updating the config line and the output line, or
2. Updating the variable name
The latter seemed simpler, so that's what I went with. But I'd be glad
to re-do this PR if you prefer the former.
Thanks for everything, y'all. You rock 🤘
**Issue:** N/A
**Dependencies:** N/A
**Twitter handle:** `conroywhitney`
**Description:** Update AnthropicLLM deprecation message import path for
ChatAnthropic
**Issue:** Incorrect import path in deprecation message
**Dependencies:** None
**Lint and test**: `make format`, `make lint` and `make test` were run
This PR updates the on_tool_end handlers to return the raw output from the tool instead of casting it to a string.
This is technically a breaking change, though it's impact is expected to be somewhat minimal. It will fix behavior in `astream_events` as well.
Fixes the following issue #18760 raised by @eyurtsev
---------
Co-authored-by: Eugene Yurtsev <eyurtsev@gmail.com>
**Description:** Update callbacks documentation
**Issue:** Change some module imports and a method invocation to reflect
the current LangChainAPI
**Dependencies:** None
BasePDFLoader doesn't parse the suffix of the file correctly when
parsing S3 presigned urls. This fix enables the proper detection and
parsing of S3 presigned URLs to prevent errors such as `OSError: [Errno
36] File name too long`.
No additional dependencies required.
Created the `facebook` page from `facebook_faiss` and `facebook_chat`
pages. Added another Facebook integrations into this page.
Updated `discord` page.
Deduplicate documents using MD5 of the page_content. Also allows for
custom deduplication with graph ingestion method by providing metadata
id attribute
---------
Co-authored-by: Harrison Chase <hw.chase.17@gmail.com>
- **Description:** Adding an optional parameter `linearization_config`
to the `AmazonTextractPDFLoader` so the caller can define how the output
will be linearized, instead of forcing a predefined set of linearization
configs. It will still have a default configuration as this will be an
optional parameter.
- **Issue:** #17457
- **Dependencies:** The same ones that already exist for
`AmazonTextractPDFLoader`
- **Twitter handle:** [@lvieirajr19](https://twitter.com/lvieirajr19)
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
*Description**: My previous
[PR](https://github.com/langchain-ai/langchain/pull/18521) was
mistakenly closed, so I am reopening this one. Context: AWS released two
Mistral models on Bedrock last Friday (March 1, 2024). This PR includes
some code adjustments to ensure their compatibility with the Bedrock
class.
---------
Co-authored-by: Anis ZAKARI <anis.zakari@hymaia.com>
Co-authored-by: Erick Friis <erick@langchain.dev>
- **Description:** Update azuresearch vectorstore from_texts() method to
include fields argument, necessary for creating an Azure AI Search index
with custom fields.
- **Issue:** Currently index fields are fixed to default fields if Azure
Search index is created using from_texts() method
- **Dependencies:** None
- **Twitter handle:** None
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
Small improvement to the openapi prompt.
The agent was not finding the server base URL (looping through all
nodes). This small change narrows the search and enables finding the url
faster.
No dependency
Twitter : @al1pra
# Proper example for AzureOpenAI usage in error message
The original error message is wrong in part of a usage example it gives.
Corrected to the right one.
Co-authored-by: Dzmitry Kankalovich <dzmitry_kankalovich@epam.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
This PR is a successor to this PR -
https://github.com/langchain-ai/langchain/pull/17436
This PR updates the cookbook README with the notebook so that it is
available on langchain docs for discoverability.
cc: @baskaryan, @3coins
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
**Description:** Fix lists display issues in **Docs > Use Cases > Q&A
with RAG > Quickstart**.
In essence, this PR changes:
```markdown
Some paragraph.
- Item a.
- Item b.
```
to:
```markdown
Some paragraph.
- Item a.
- Item b.
```
There needs an extra empty line to make the list rendered properly.
FYI, the old version is displayed not properly as:
<img width="856" alt="image"
src="https://github.com/langchain-ai/langchain/assets/22856433/65202577-8ea2-47c6-b310-39bf42796fac">
- [x] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Co-authored-by: Bagatur <baskaryan@gmail.com>
- **Description:** `S3DirectoryLoader` is failing if prefix is a folder
(ex: `my_folder/`) because `S3FileLoader` will try to load that folder
and will fail. This PR skip nested directories so prefix can be set to
folder instead of `my_folder/files_prefix`.
- **Issue:**
- #11917
- #6535
- #4326
- **Dependencies:** none
- **Twitter handle:** @Falydoor
- [x] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [x] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
- [ ] Title: Mongodb: MongoDB connection performance improvement.
- [ ] Message:
- **Description:** I made collection index_creation as optional. Index
Creation is one time process.
- **Issue:** MongoDBChatMessageHistory class object is attempting to
create an index during connection, causing each request to take longer
than usual. This should be optional with a parameter.
- **Dependencies:** N/A
- **Branch to be checked:** origin/mongo_index_creation
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
- **Description:** Add embedding instruction to
HuggingFaceBgeEmbeddings, so that it can be compatible with nomic and
other models that need embedding instruction.
---------
Co-authored-by: Tao Wu <tao.wu@rwth-aachen.de>
Co-authored-by: Bagatur <baskaryan@gmail.com>
_generate() and _agenerate() both accept **kwargs, then pass them on to
_format_output; but _format_output doesn't accept **kwargs. Attempting
to pass, e.g.,
timeout=50
to _generate (or invoke()) results in a TypeError.
Thank you for contributing to LangChain!
- [ ] **PR title**: "package: description"
- Where "package" is whichever of langchain, community, core,
experimental, etc. is being modified. Use "docs: ..." for purely docs
changes, "templates: ..." for template changes, "infra: ..." for CI
changes.
- Example: "community: add foobar LLM"
- [ ] **PR message**: ***Delete this entire checklist*** and replace
with
- **Description:** a description of the change
- **Issue:** the issue # it fixes, if applicable
- **Dependencies:** any dependencies required for this change
- **Twitter handle:** if your PR gets announced, and you'd like a
mention, we'll gladly shout you out!
- [ ] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [ ] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
---------
Co-authored-by: Erick Friis <erick@langchain.dev>
## Add Passio Nutrition AI Food Search Tool to Community Package
### Description
We propose adding a new tool to the `community` package, enabling
integration with Passio Nutrition AI for food search functionality. This
tool will provide a simple interface for retrieving nutrition facts
through the Passio Nutrition AI API, simplifying user access to
nutrition data based on food search queries.
### Implementation Details
- **Class Structure:** Implement `NutritionAI`, extending `BaseTool`. It
includes an `_run` method that accepts a query string and, optionally, a
`CallbackManagerForToolRun`.
- **API Integration:** Use `NutritionAIAPI` for the API wrapper,
encapsulating all interactions with the Passio Nutrition AI and
providing a clean API interface.
- **Error Handling:** Implement comprehensive error handling for API
request failures.
### Expected Outcome
- **User Benefits:** Enable easy querying of nutrition facts from Passio
Nutrition AI, enhancing the utility of the `langchain_community` package
for nutrition-related projects.
- **Functionality:** Provide a straightforward method for integrating
nutrition information retrieval into users' applications.
### Dependencies
- `langchain_core` for base tooling support
- `pydantic` for data validation and settings management
- Consider `requests` or another HTTP client library if not covered by
`NutritionAIAPI`.
### Tests and Documentation
- **Unit Tests:** Include tests that mock network interactions to ensure
tool reliability without external API dependency.
- **Documentation:** Create an example notebook in
`docs/docs/integrations/tools/passio_nutrition_ai.ipynb` showing usage,
setup, and example queries.
### Contribution Guidelines Compliance
- Adhere to the project's linting and formatting standards (`make
format`, `make lint`, `make test`).
- Ensure compliance with LangChain's contribution guidelines,
particularly around dependency management and package modifications.
### Additional Notes
- Aim for the tool to be a lightweight, focused addition, not
introducing significant new dependencies or complexity.
- Potential future enhancements could include caching for common queries
to improve performance.
### Twitter Handle
- Here is our Passio AI [twitter handle](https://twitter.com/@passio_ai)
where we announce our products.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
"community: added a feature to filter documents in Mongoloader"
- **Description:** added a feature to filter documents in Mongoloader
- **Feature:** the feature #18251
- **Dependencies:** No
- **Twitter handle:** https://twitter.com/im_Kushagra
For some DBs with lots of tables, reflection of all the tables can take
very long. So this change will make the tables be reflected lazily when
get_table_info() is called and `lazy_table_reflection` is True.
Allows all chat models that implement _stream, but not _astream to still have async streaming to work.
Amongst other things this should resolve issues with streaming community model implementations through langserve since langserve is exclusively async.
**Description:** Replacing the deprecated predict() and apredict()
methods in the unit tests
**Issue:** Not applicable
**Dependencies:** None
**Lint and test**: `make format`, `make lint` and `make test` have been
run
**Description:** Minor update to Anthropic documentation
**Issue:** Not applicable
**Dependencies:** None
**Lint and test**: `make format` and `make lint` was done
This path updates function "run" to "invoke" in llm_bash.ipynb.
Without this path, you see following warning.
LangChainDeprecationWarning: The function `run` was deprecated in
LangChain 0.1.0
and will be removed in 0.2.0. Use invoke instead.
Signed-off-by: Masanari Iida <standby24x7@gmail.com>
Fixing a minor typo in the package name.
Thank you for contributing to LangChain!
- [ ] **PR title**: "package: description"
- Where "package" is whichever of langchain, community, core,
experimental, etc. is being modified. Use "docs: ..." for purely docs
changes, "templates: ..." for template changes, "infra: ..." for CI
changes.
- Example: "community: add foobar LLM"
- [ ] **PR message**: ***Delete this entire checklist*** and replace
with
- **Description:** a description of the change
- **Issue:** the issue # it fixes, if applicable
- **Dependencies:** any dependencies required for this change
- **Twitter handle:** if your PR gets announced, and you'd like a
mention, we'll gladly shout you out!
- [ ] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [ ] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
- [ ] **PR title:** docs: Fix link to HF TEI in
text_embeddings_inference.ipynb
- [ ] **PR message:**
- **Description:** Fix the link to [Hugging Face Text Embeddings
Inference
(TEI)](https://huggingface.co/docs/text-embeddings-inference/index) in
text_embeddings_inference.ipynb
- **Issue:** Fix#18576
Make `ElasticsearchRetriever` available as top-level import.
The `langchain` package depends on `langchain-community` so we do not
need to depend on it explicitly.
## Description
- Add [Friendli](https://friendli.ai/) integration for `Friendli` LLM
and `ChatFriendli` chat model.
- Unit tests and integration tests corresponding to this change are
added.
- Documentations corresponding to this change are added.
## Dependencies
- Optional dependency
[`friendli-client`](https://pypi.org/project/friendli-client/) package
is added only for those who use `Frienldi` or `ChatFriendli` model.
## Twitter handle
- https://twitter.com/friendliai
This pull request introduces initial support for the TiDB vector store.
The current version is basic, laying the foundation for the vector store
integration. While this implementation provides the essential features,
we plan to expand and improve the TiDB vector store support with
additional enhancements in future updates.
Upcoming Enhancements:
* Support for Vector Index Creation: To enhance the efficiency and
performance of the vector store.
* Support for max marginal relevance search.
* Customized Table Structure Support: Recognizing the need for
flexibility, we plan for more tailored and efficient data store
solutions.
Simple use case exmaple
```python
from typing import List, Tuple
from langchain.docstore.document import Document
from langchain_community.vectorstores import TiDBVectorStore
from langchain_openai import OpenAIEmbeddings
db = TiDBVectorStore.from_texts(
embedding=embeddings,
texts=['Andrew like eating oranges', 'Alexandra is from England', 'Ketanji Brown Jackson is a judge'],
table_name="tidb_vector_langchain",
connection_string=tidb_connection_url,
distance_strategy="cosine",
)
query = "Can you tell me about Alexandra?"
docs_with_score: List[Tuple[Document, float]] = db.similarity_search_with_score(query)
for doc, score in docs_with_score:
print("-" * 80)
print("Score: ", score)
print(doc.page_content)
print("-" * 80)
```
## **Description:**
MongoDB integration tests link to a provided Atlas Cluster. We have very
stringent permissions set against the cluster provided. In order to make
it easier to track and isolate the collections each test gets run
against, we've updated the collection names to map the test file name.
i.e. `langchain_{filename}` => `langchain_test_vectorstores`
Fixes integration test results
## **Dependencies:**
Provided MONGODB_ATLAS_URI
- [x] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
cc: @shaneharvey, @blink1073 , @NoahStapp , @caseyclements
- **Description:** Chroma use uuid4 instead of uuid1 as random ids. Use
uuid1 may leak mac address, changing to uuid4 will not cause other
effects.
- **Issue:** None
- **Dependencies:** None
- **Twitter handle:** None
Fixes#18513.
## Description
This PR attempts to fix the support for Anthropic Claude v3 models in
BedrockChat LLM. The changes here has updated the payload to use the
`messages` format instead of the formatted text prompt for all models;
`messages` API is backwards compatible with all models in Anthropic, so
this should not break the experience for any models.
## Notes
The PR in the current form does not support the v3 models for the
non-chat Bedrock LLM. This means, that with these changes, users won't
be able to able to use the v3 models with the Bedrock LLM. I can open a
separate PR to tackle this use-case, the intent here was to get this out
quickly, so users can start using and test the chat LLM. The Bedrock LLM
classes have also grown complex with a lot of conditions to support
various providers and models, and is ripe for a refactor to make future
changes more palatable. This refactor is likely to take longer, and
requires more thorough testing from the community. Credit to PRs
[18579](https://github.com/langchain-ai/langchain/pull/18579) and
[18548](https://github.com/langchain-ai/langchain/pull/18548) for some
of the code here.
---------
Co-authored-by: Erick Friis <erick@langchain.dev>
**Description:**
This integrates Infinispan as a vectorstore.
Infinispan is an open-source key-value data grid, it can work as single
node as well as distributed.
Vector search is supported since release 15.x
For more: [Infinispan Home](https://infinispan.org)
Integration tests are provided as well as a demo notebook
Thank you for contributing to LangChain!
- [x] **PR title**: "templates: rag-multi-modal typo, replace serch with
search "
- **Description:** Two little typos in multi modal templates (replace
serch string with search)
Signed-off-by: José Luis Di Biase <josx@interorganic.com.ar>
ValidationError: 2 validation errors for DocArrayDoc
text
Field required [type=missing, input_value={'embedding': [-0.0191128...9, 0.01005221541175212]}, input_type=dict]
For further information visit https://errors.pydantic.dev/2.5/v/missing
metadata
Field required [type=missing, input_value={'embedding': [-0.0191128...9, 0.01005221541175212]}, input_type=dict]
For further information visit https://errors.pydantic.dev/2.5/v/missing
```
In the `_get_doc_cls` method, the `DocArrayDoc` class is defined as
follows:
```python
class DocArrayDoc(BaseDoc):
text: Optional[str]
embedding: Optional[NdArray] = Field(**embeddings_params)
metadata: Optional[dict]
```
This is a PR that adds a dangerous load parameter to force users to opt in to use pickle.
This is a PR that's meant to raise user awareness that the pickling module is involved.
This is a patch for `CVE-2024-2057`:
https://www.cve.org/CVERecord?id=CVE-2024-2057
This affects users that:
* Use the `TFIDFRetriever`
* Attempt to de-serialize it from an untrusted source that contains a
malicious payload
**Description:** Update to the streaming tutorial notebook in the LCEL
documentation
**Issue:** Fixed an import and (minor) changes in documentation language
**Dependencies:** None
- **Description:** Databricks SerDe uses cloudpickle instead of pickle
when serializing a user-defined function transform_input_fn since pickle
does not support functions defined in `__main__`, and cloudpickle
supports this.
- **Dependencies:** cloudpickle>=2.0.0
Added a unit test.
- **Description:** Fixed some typos and copy errors in the Beta
Structured Output docs
- **Issue:** N/A
- **Dependencies:** Docs only
- **Twitter handle:** @psvann
Co-authored-by: P.S. Vann <psvann@yahoo.com>
Description:
This pull request addresses two key improvements to the langchain
repository:
**Fix for Crash in Flight Search Interface**:
Previously, the code would crash when encountering a failure scenario in
the flight ticket search interface. This PR resolves this issue by
implementing a fix to handle such scenarios gracefully. Now, the code
handles failures in the flight search interface without crashing,
ensuring smoother operation.
**Documentation Update for Amadeus Toolkit**:
Prior to this update, examples provided in the documentation for the
Amadeus Toolkit were unable to run correctly due to outdated
information. This PR includes an update to the documentation, ensuring
that all examples can now be executed successfully. With this update,
users can effectively utilize the Amadeus Toolkit with accurate and
functioning examples.
These changes aim to enhance the reliability and usability of the
langchain repository by addressing issues related to error handling and
ensuring that documentation remains up-to-date and actionable.
Issue: https://github.com/langchain-ai/langchain/issues/17375
Twitter Handle: SingletonYxx
### Description
Changed the value specified for `content_key` in JSONLoader from a
single key to a value based on jq schema.
I created [similar
PR](https://github.com/langchain-ai/langchain/pull/11255) before, but it
has several conflicts because of the architectural change associated
stable version release, so I re-create this PR to fit new architecture.
### Why
For json data like the following, specify `.data[].attributes.message`
for page_content and `.data[].attributes.id` or
`.data[].attributes.attributes. tags`, etc., the `content_key` must also
parse the json structure.
<details>
<summary>sample json data</summary>
```json
{
"data": [
{
"attributes": {
"message": "message1",
"tags": [
"tag1"
]
},
"id": "1"
},
{
"attributes": {
"message": "message2",
"tags": [
"tag2"
]
},
"id": "2"
}
]
}
```
</details>
<details>
<summary>sample code</summary>
```python
def metadata_func(record: dict, metadata: dict) -> dict:
metadata["source"] = None
metadata["id"] = record.get("id")
metadata["tags"] = record["attributes"].get("tags")
return metadata
sample_file = "sample1.json"
loader = JSONLoader(
file_path=sample_file,
jq_schema=".data[]",
content_key=".attributes.message", ## content_key is parsable into jq schema
is_content_key_jq_parsable=True, ## this is added parameter
metadata_func=metadata_func
)
data = loader.load()
data
```
</details>
### Dependencies
none
### Twitter handle
[kzk_maeda](https://twitter.com/kzk_maeda)
Neo4j tools use particular node labels and relationship types to store
metadata, but are irrelevant for text2cypher or graph generation, so we
want to ignore them in the schema representation.
This patch updates function "run" to "invoke" in smart_llm.ipynb.
Without this patch, you see following warning.
LangChainDeprecationWarning: The function `run` was deprecated in
LangChain 0.1.0 and will be removed in 0.2.0. Use invoke instead.
Signed-off-by: Masanari Iida <standby24x7@gmail.com>
Signed-off-by: Masanari Iida <standby24x7@gmail.com>
Deprecates the old langchain-hub repository. Does *not* deprecate the
new https://smith.langchain.com/hub
@PinkDraconian has correctly raised that in the event someone is loading
unsanitized user input into the `try_load_from_hub` function, they have
the ability to load files from other locations in github than the
hwchase17/langchain-hub repository.
This PR adds some more path checking to that function and deprecates the
functionality in favor of the hub built into LangSmith.
**Description:**
modified the user_name to username to conform with the expected inputs
to TelegramChatApiLoader
**Issue:**
Current code fails in langchain-community 0.0.24
<loader = TelegramChatApiLoader(
chat_entity="<CHAT_URL>", # recommended to use Entity here
api_hash="<API HASH >",
api_id="<API_ID>",
user_name="", # needed only for caching the session.
)>
## Description
Adding in Unit Test variation for `MongoDBChatMessageHistory` package
Follow-up to #18590
- [x] **Add tests and docs**: Unit test is what's being added
- [x] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
## **Description**
Migrate the `MongoDBChatMessageHistory` to the managed
`langchain-mongodb` partner-package
## **Dependencies**
None
## **Twitter handle**
@mongodb
## **tests and docs**
- [x] Migrate existing integration test
- [x ]~ Convert existing integration test to a unit test~ Creation is
out of scope for this ticket
- [x ] ~Considering delaying work until #17470 merges to leverage the
`MockCollection` object. ~
- [x] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
---------
Co-authored-by: Erick Friis <erick@langchain.dev>
# Description
- **Description:** Adding MongoDB LLM Caching Layer abstraction
- **Issue:** N/A
- **Dependencies:** None
- **Twitter handle:** @mongodb
Checklist:
- [x] PR title: Please title your PR "package: description", where
"package" is whichever of langchain, community, core, experimental, etc.
is being modified. Use "docs: ..." for purely docs changes, "templates:
..." for template changes, "infra: ..." for CI changes.
- Example: "community: add foobar LLM"
- [x] PR Message (above)
- [x] Pass lint and test: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified to check that you're
passing lint and testing. See contribution guidelines for more
information on how to write/run tests, lint, etc:
https://python.langchain.com/docs/contributing/
- [ ] Add tests and docs: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
@baskaryan, @efriis, @eyurtsev, @hwchase17.
---------
Co-authored-by: Jib <jib@byblack.us>
Thank you for contributing to LangChain!
- [ ] **PR title**: "community: deprecate vectorstores.MatchingEngine"
- [ ] **PR message**:
- **Description:** announced a deprecation since this integration has
been moved to langchain_google_vertexai
**Description:** Update docstrings of ChatAnthropic class
**Issue:** Change to ChatAnthropic from ChatAnthropicMessages
**Dependencies:** None
**Lint and test**: `make format`, `make lint` and `make test` passed
- **Description:**
This PR fixes some issues in the Jupyter notebook for the VectorStore
"SAP HANA Cloud Vector Engine":
* Slight textual adaptations
* Fix of wrong column name VEC_META (was: VEC_METADATA)
- **Issue:** N/A
- **Dependencies:** no new dependecies added
- **Twitter handle:** @sapopensource
path to notebook:
`docs/docs/integrations/vectorstores/hanavector.ipynb`
Currently llm_checker.ipynb uses a function "run".
Update to "invoke" to avoid following warning.
LangChainDeprecationWarning: The function `run` was deprecated in
LangChain 0.1.0
and will be removed in 0.2.0. Use invoke instead.
Signed-off-by: Masanari Iida <standby24x7@gmail.com>
This patch updates function "run" to "invoke".
Without this patch you see following warning.
LangChainDeprecationWarning: The function `run` was deprecated in
LangChain 0.1.0 and will be removed in 0.2.0. Use invoke instead.
Signed-off-by: Masanari Iida <standby24x7@gmail.com>
## PR title
Docs: Updated callbacks/index.mdx adding example on runnable methods
## PR message
- **Description:** Updated callbacks/index.mdx adding an example on how
to pass callbacks to the runnable methods (invoke, batch, ...)
- **Issue:** #16379
- **Dependencies:** None
- **Description:** finishes adding the you.com functionality including:
- add async functions to utility and retriever
- add the You.com Tool
- add async testing for utility, retriever, and tool
- add a tool integration notebook page
- **Dependencies:** any dependencies required for this change
- **Twitter handle:** @scottnath
- **Description:** add tools_renderer for various non-openai agents,
make tools can be render in different ways for your LLM.
- **Issue:** N/A
- **Dependencies:** N/A
---------
Co-authored-by: Harrison Chase <hw.chase.17@gmail.com>
Description:
This pull request introduces several enhancements for Azure Cosmos
Vector DB, primarily focused on improving caching and search
capabilities using Azure Cosmos MongoDB vCore Vector DB. Here's a
summary of the changes:
- **AzureCosmosDBSemanticCache**: Added a new cache implementation
called AzureCosmosDBSemanticCache, which utilizes Azure Cosmos MongoDB
vCore Vector DB for efficient caching of semantic data. Added
comprehensive test cases for AzureCosmosDBSemanticCache to ensure its
correctness and robustness. These tests cover various scenarios and edge
cases to validate the cache's behavior.
- **HNSW Vector Search**: Added HNSW vector search functionality in the
CosmosDB Vector Search module. This enhancement enables more efficient
and accurate vector searches by utilizing the HNSW (Hierarchical
Navigable Small World) algorithm. Added corresponding test cases to
validate the HNSW vector search functionality in both
AzureCosmosDBSemanticCache and AzureCosmosDBVectorSearch. These tests
ensure the correctness and performance of the HNSW search algorithm.
- **LLM Caching Notebook** - The notebook now includes a comprehensive
example showcasing the usage of the AzureCosmosDBSemanticCache. This
example highlights how the cache can be employed to efficiently store
and retrieve semantic data. Additionally, the example provides default
values for all parameters used within the AzureCosmosDBSemanticCache,
ensuring clarity and ease of understanding for users who are new to the
cache implementation.
@hwchase17,@baskaryan, @eyurtsev,
**Description:** Update to the pathspec for 'git grep' in lint check in
the Makefile
**Issue:** The pathspec {docs/docs,templates,cookbook} is not handled
correctly leading to the error during 'make lint' -
"fatal: ambiguous argument '{docs/docs,templates,cookbook}': unknown
revision or path not in the working tree."
See changes made in https://github.com/langchain-ai/langchain/pull/18058
Co-authored-by: Erick Friis <erick@langchain.dev>
### Description
Fixed a small bug in chroma.py add_images(), previously whenever we are
not passing metadata the documents is containing the base64 of the uris
passed, but when we are passing the metadata the documents is containing
normal string uris which should not be the case.
### Issue
In add_images() method when we are calling upsert() we have to use
"b64_texts" instead of normal string "uris".
### Twitter handle
https://twitter.com/whitepegasus01
- [X] Gemini Agent Executor imported `agent.py` has Gemini agent
executor which was not utilised in current template of gemini function
agent 🧑💻 instead openai_function_agent has been used
@sbusso @jarib please someone review it
---------
Co-authored-by: Erick Friis <erick@langchain.dev>
* **Description:** adds `LlamafileEmbeddings` class implementation for
generating embeddings using
[llamafile](https://github.com/Mozilla-Ocho/llamafile)-based models.
Includes related unit tests and notebook showing example usage.
* **Issue:** N/A
* **Dependencies:** N/A
- [x] **PR message**: ***Delete this entire checklist*** and replace
with
- **Description:** Remove the assert statement on the `count_documents`
in setup_class. It should just delete if there are documents present
- **Issue:** the issue # Crashes on class setup
- **Dependencies:** None
- **Twitter handle:** @mongodb
- [x] **Add tests and docs**: If you're adding a new integration, please
include
1. N/A
- [ ] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
Co-authored-by: Jib <jib@byblack.us>
Current implementation doesn't have an indexed property that would
optimize the import. I have added a `baseEntityLabel` parameter that
allows you to add a secondary node label, which has an indexed id
`property`. By default, the behaviour is identical to previous version.
Since multi-labeled nodes are terrible for text2cypher, I removed the
secondary label from schema representation object and string, which is
used in text2cypher.
**Description:**
(a) Update to the module import path to reflect the splitting up of
langchain into separate packages
(b) Update to the documentation to include the new calling method
(invoke)
This PR makes `cohere_api_key` in `llms/cohere` a SecretStr, so that the
API Key is not leaked when `Cohere.cohere_api_key` is represented as a
string.
---------
Signed-off-by: Arun <arun@arun.blog>
Co-authored-by: Eugene Yurtsev <eyurtsev@gmail.com>
**Description:**
The URL of the data to index, specified to `WebBaseLoader` to import is
incorrect, causing the `langsmith_search` retriever to return a `404:
NOT_FOUND`.
Incorrect URL: https://docs.smith.langchain.com/overview
Correct URL: https://docs.smith.langchain.com
**Issue:**
This commit corrects the URL and prevents the LangServe Playground from
returning an error from its inability to use the retriever when
inquiring, "how can langsmith help with testing?".
**Dependencies:**
None.
**Twitter Handle:**
@ryanmeinzer
**Description:** Fix `metadata_extractor` type for `RecursiveUrlLoader`,
the default `_metadata_extractor` returns `dict` instead of `str`.
**Issue:** N/A
**Dependencies:** N/A
**Twitter handle:** N/A
Signed-off-by: Hemslo Wang <hemslo.wang@gmail.com>
- **Description:** Removing this line
```python
response = index.query(query, response_mode="no_text", **self.query_kwargs)
```
to
```python
response = index.query(query, **self.query_kwargs)
```
Since llama index query does not support response_mode anymore : ``` |
TypeError: BaseQueryEngine.query() got an unexpected keyword argument
'response_mode'````
- **Twitter handle:** @maximeperrin_
---------
Co-authored-by: Maxime Perrin <mperrin@doing.fr>
- [ ] **PR title**: "cookbook: using Gemma on LangChain"
- [ ] **PR message**:
- **Description:** added a tutorial how to use Gemma with LangChain
(from VertexAI or locally from Kaggle or HF)
- **Dependencies:** langchain-google-vertexai==0.0.7
- **Twitter handle:** lkuligin
In this commit we update the documentation for Google El Carro for Oracle Workloads. We amend the documentation in the Google Providers page to use the correct name which is El Carro for Oracle Workloads. We also add changes to the document_loaders and memory pages to reflect changes we made in our repo.
If the document loader recieves Pathlib path instead of str, it reads
the file correctly, but the problem begins when the document is added to
Deeplake.
This problem arises from casting the path to str in the metadata.
```python
deeplake = True
fname = Path('./lorem_ipsum.txt')
loader = TextLoader(fname, encoding="utf-8")
docs = loader.load_and_split()
text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=100)
chunks= text_splitter.split_documents(docs)
if deeplake:
db = DeepLake(dataset_path=ds_path, embedding=embeddings, token=activeloop_token)
db.add_documents(chunks)
else:
db = Chroma.from_documents(docs, embeddings)
```
So using this snippet of code the error message for deeplake looks like
this:
```
[part of error message omitted]
Traceback (most recent call last):
File "/home/mwm/repositories/sources/fixing_langchain/main.py", line 53, in <module>
db.add_documents(chunks)
File "/home/mwm/repositories/sources/langchain/libs/core/langchain_core/vectorstores.py", line 139, in add_documents
return self.add_texts(texts, metadatas, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/mwm/repositories/sources/langchain/libs/community/langchain_community/vectorstores/deeplake.py", line 258, in add_texts
return self.vectorstore.add(
^^^^^^^^^^^^^^^^^^^^^
File "/home/mwm/anaconda3/envs/langchain/lib/python3.11/site-packages/deeplake/core/vectorstore/deeplake_vectorstore.py", line 226, in add
return self.dataset_handler.add(
^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/mwm/anaconda3/envs/langchain/lib/python3.11/site-packages/deeplake/core/vectorstore/dataset_handlers/client_side_dataset_handler.py", line 139, in add
dataset_utils.extend_or_ingest_dataset(
File "/home/mwm/anaconda3/envs/langchain/lib/python3.11/site-packages/deeplake/core/vectorstore/vector_search/dataset/dataset.py", line 544, in extend_or_ingest_dataset
extend(
File "/home/mwm/anaconda3/envs/langchain/lib/python3.11/site-packages/deeplake/core/vectorstore/vector_search/dataset/dataset.py", line 505, in extend
dataset.extend(batched_processed_tensors, progressbar=False)
File "/home/mwm/anaconda3/envs/langchain/lib/python3.11/site-packages/deeplake/core/dataset/dataset.py", line 3247, in extend
raise SampleExtendError(str(e)) from e.__cause__
deeplake.util.exceptions.SampleExtendError: Failed to append a sample to the tensor 'metadata'. See more details in the traceback. If you wish to skip the samples that cause errors, please specify `ignore_errors=True`.
```
Which is does not explain the error well enough.
The same error for chroma looks like this
```
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/home/mwm/repositories/sources/fixing_langchain/main.py", line 56, in <module>
db = Chroma.from_documents(docs, embeddings)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/mwm/repositories/sources/langchain/libs/community/langchain_community/vectorstores/chroma.py", line 778, in from_documents
return cls.from_texts(
^^^^^^^^^^^^^^^
File "/home/mwm/repositories/sources/langchain/libs/community/langchain_community/vectorstores/chroma.py", line 736, in from_texts
chroma_collection.add_texts(
File "/home/mwm/repositories/sources/langchain/libs/community/langchain_community/vectorstores/chroma.py", line 309, in add_texts
raise ValueError(e.args[0] + "\n\n" + msg)
ValueError: Expected metadata value to be a str, int, float or bool, got lorem_ipsum.txt which is a <class 'pathlib.PosixPath'>
Try filtering complex metadata from the document using langchain_community.vectorstores.utils.filter_complex_metadata.
```
Which is way more user friendly, so I just added information about
possible mismatch of the type in the error message, the same way it is
covered in chroma
https://github.com/langchain-ai/langchain/blob/master/libs/community/langchain_community/vectorstores/chroma.py#L224
Thank you for contributing to LangChain!
- [x] **PR title**: "package: description"
- Where "package" is whichever of langchain, community, core,
experimental, etc. is being modified. Use "docs: ..." for purely docs
changes, "templates: ..." for template changes, "infra: ..." for CI
changes.
- Example: "community: add foobar LLM"
- [x] **PR message**: ***Delete this entire checklist*** and replace
with
- **Description:** a description of the change
- **Issue:** the issue # it fixes, if applicable
- **Dependencies:** any dependencies required for this change
- **Twitter handle:** if your PR gets announced, and you'd like a
mention, we'll gladly shout you out!
- [x] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [x] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
---------
Co-authored-by: Erick Friis <erick@langchain.dev>
- **Description**:
[`bigdl-llm`](https://github.com/intel-analytics/BigDL) is a library for
running LLM on Intel XPU (from Laptop to GPU to Cloud) using
INT4/FP4/INT8/FP8 with very low latency (for any PyTorch model). This PR
adds bigdl-llm integrations to langchain.
- **Issue**: NA
- **Dependencies**: `bigdl-llm` library
- **Contribution maintainer**: @shane-huang
Examples added:
- docs/docs/integrations/llms/bigdl.ipynb
Nvidia provider page is missing a Triton Inference Server package
reference.
Changes:
- added the Triton Inference Server reference
- copied the example notebook from the package into the doc files.
- added the Triton Inference Server description and links, the link to
the above example notebook
- formatted page to the consistent format
NOTE:
It seems that the [example
notebook](https://github.com/langchain-ai/langchain/blob/master/libs/partners/nvidia-trt/docs/llms.ipynb)
was originally created in wrong place. It should be in the LangChain
docs
[here](https://github.com/langchain-ai/langchain/tree/master/docs/docs/integrations/llms).
So, I've created a copy of this example. The original example is still
in the nvidia-trt package.
Description-
- Changed the GitHub endpoint as existing was not working and giving 404
not found error
- Also the existing function was failing if file_filter is not passed as
the tree api return all paths including directory as well, and when
get_file_content was iterating over these path, the function was failing
for directory as the api was returning list of files inside the
directory, so added a condition to ignore the paths if it a directory
- Fixes this issue -
https://github.com/langchain-ai/langchain/issues/17453
Co-authored-by: Radhika Bansal <Radhika.Bansal@veritas.com>
## Description
Updates the `langchain_community.embeddings.fastembed` provider as per
the recent updates to [`FastEmbed`](https://github.com/qdrant/fastembed)
library.
Thank you for contributing to LangChain!
- [ ] **PR title**: "package: description"
- Where "package" is whichever of langchain, community, core,
experimental, etc. is being modified. Use "docs: ..." for purely docs
changes, "templates: ..." for template changes, "infra: ..." for CI
changes.
- Example: "community: add foobar LLM"
- [ ] **PR message**: ***Delete this entire checklist*** and replace
with
- **Description:** a description of the change
- **Issue:** the issue # it fixes, if applicable
- **Dependencies:** any dependencies required for this change
- **Twitter handle:** if your PR gets announced, and you'd like a
mention, we'll gladly shout you out!
- [ ] **Add tests and docs**: If you're adding a new integration, please
include
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory.
- [ ] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
This PR migrates the existing MongoDBAtlasVectorSearch abstraction from
the `langchain_community` section to the partners package section of the
codebase.
- [x] Run the partner package script as advised in the partner-packages
documentation.
- [x] Add Unit Tests
- [x] Migrate Integration Tests
- [x] Refactor `MongoDBAtlasVectorStore` (autogenerated) to
`MongoDBAtlasVectorSearch`
- [x] ~Remove~ deprecate the old `langchain_community` VectorStore
references.
## Additional Callouts
- Implemented the `delete` method
- Included any missing async function implementations
- `amax_marginal_relevance_search_by_vector`
- `adelete`
- Added new Unit Tests that test for functionality of
`MongoDBVectorSearch` methods
- Removed [`del
res[self._embedding_key]`](e0c81e1cb0/libs/community/langchain_community/vectorstores/mongodb_atlas.py (L218))
in `_similarity_search_with_score` function as it would make the
`maximal_marginal_relevance` function fail otherwise. The `Document`
needs to store the embedding key in metadata to work.
Checklist:
- [x] PR title: Please title your PR "package: description", where
"package" is whichever of langchain, community, core, experimental, etc.
is being modified. Use "docs: ..." for purely docs changes, "templates:
..." for template changes, "infra: ..." for CI changes.
- Example: "community: add foobar LLM"
- [x] PR message
- [x] Pass lint and test: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified to check that you're
passing lint and testing. See contribution guidelines for more
information on how to write/run tests, lint, etc:
https://python.langchain.com/docs/contributing/
- [x] Add tests and docs: If you're adding a new integration, please
include
1. Existing tests supplied in docs/docs do not change. Updated
docstrings for new functions like `delete`
2. an example notebook showing its use. It lives in
`docs/docs/integrations` directory. (This already exists)
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, hwchase17.
---------
Co-authored-by: Steven Silvester <steven.silvester@ieee.org>
Co-authored-by: Erick Friis <erick@langchain.dev>
## PR title
partners: changed the README file for the Fireworks integration in the
libs/partners/fireworks folder
## PR message
Description: Changed the README file of partners/fireworks following the
docs on https://python.langchain.com/docs/integrations/llms/Fireworks
The README includes:
- Brief description
- Installation
- Setting-up instructions (API key, model id, ...)
- Basic usage
Issue: https://github.com/langchain-ai/langchain/issues/17545
Dependencies: None
Twitter handle: None
- **Description:** The current embedchain implementation seems to handle
document metadata differently than done in the current implementation of
langchain and a KeyError is thrown. I would love for someone else to
test this...
---------
Co-authored-by: KKUGLER <kai.kugler@mercedes-benz.com>
Co-authored-by: Harrison Chase <hw.chase.17@gmail.com>
Co-authored-by: Deshraj Yadav <deshraj@gatech.edu>
Sometimes, you want to use various parameters in the retrieval query of
Neo4j Vector to personalize/customize results. Before, when there were
only predefined chains, it didn't really make sense. Now that it's all
about custom chains and LCEL, it is worth adding since users can inject
any params they wish at query time. Isn't prone to SQL injection-type
attacks since we use parameters and not concatenating strings.
**Description:** Add facility to pass the optional output parser to
customize the parsing logic
---------
Co-authored-by: hasan <hasan@m2sys.com>
Co-authored-by: Erick Friis <erick@langchain.dev>
This PR adds links to some more free resources for people to get
acquainted with Langhchain without having to configure their system.
<!-- If no one reviews your PR within a few days, please @-mention one
of baskaryan, efriis, eyurtsev, hwchase17. -->
Co-authored-by: Filip Schouwenaars <filipsch@users.noreply.github.com>
**Description:**
In this PR, I am adding a `PolygonFinancials` tool, which can be used to
get financials data for a given ticker. The financials data is the
fundamental data that is found in income statements, balance sheets, and
cash flow statements of public US companies.
**Twitter**:
[@virattt](https://twitter.com/virattt)
[](https://vscode.dev/redirect?url=vscode://ms-vscode-remote.remote-containers/cloneInVolume?url=https://github.com/langchain-ai/langchain)
[](https://codespaces.new/langchain-ai/langchain)
[](https://vscode.dev/redirect?url=vscode://ms-vscode-remote.remote-containers/cloneInVolume?url=https://github.com/langchain-ai/langchain)
[](https://codespaces.new/langchain-ai/langchain)
[](https://star-history.com/#langchain-ai/langchain)
For these applications, LangChain simplifies the entire application lifecycle:
- **Open-source libraries**: Build your applications using LangChain's [modular building blocks](https://python.langchain.com/v0.2/docs/concepts/#langchain-expression-language-lcel) and [components](https://python.langchain.com/v0.2/docs/concepts/#components). Integrate with hundreds of [third-party providers](https://python.langchain.com/v0.2/docs/integrations/platforms/).
- **Productionization**: Inspect, monitor, and evaluate your apps with [LangSmith](https://docs.smith.langchain.com/) so that you can constantly optimize and deploy with confidence.
- **Deployment**: Turn any chain into a REST API with [LangServe](https://python.langchain.com/v0.2/docs/langserve/).
- **Open-source libraries**: Build your applications using LangChain's [modular building blocks](https://python.langchain.com/docs/expression_language/) and [components](https://python.langchain.com/docs/modules/). Integrate with hundreds of [third-party providers](https://python.langchain.com/docs/integrations/platforms/).
- **Productionization**: Inspect, monitor, and evaluate your apps with [LangSmith](https://python.langchain.com/docs/langsmith/) so that you can constantly optimize and deploy with confidence.
- **Deployment**: Turn any chain into a REST API with [LangServe](https://python.langchain.com/docs/langserve).
### Open-source libraries
- **`langchain-core`**: Base abstractions and LangChain Expression Language.
- **`langchain-community`**: Third party integrations.
- Some integrations have been further split into **partner packages** that only rely on **`langchain-core`**. Examples include **`langchain_openai`** and **`langchain_anthropic`**.
- **`langchain`**: Chains, agents, and retrieval strategies that make up an application's cognitive architecture.
- **[`LangGraph`](https://langchain-ai.github.io/langgraph/)**: A library for building robust and stateful multi-actor applications with LLMs by modeling steps as edges and nodes in a graph.
- **[LangGraph](https://python.langchain.com/docs/langgraph)**: A library for building robust and stateful multi-actor applications with LLMs by modeling steps as edges and nodes in a graph.
### Productionization:
- **[LangSmith](https://docs.smith.langchain.com/)**: A developer platform that lets you debug, test, evaluate, and monitor chains built on any LLM framework and seamlessly integrates with LangChain.
- **[LangSmith](https://python.langchain.com/docs/langsmith)**: A developer platform that lets you debug, test, evaluate, and monitor chains built on any LLM framework and seamlessly integrates with LangChain.
### Deployment:
- **[LangServe](https://python.langchain.com/v0.2/docs/langserve/)**: A library for deploying LangChain chains as REST APIs.
- **[LangServe](https://python.langchain.com/docs/langserve)**: A library for deploying LangChain chains as REST APIs.
@@ -61,20 +61,20 @@ For these applications, LangChain simplifies the entire application lifecycle:
- End-to-end Example: [Web LangChain (web researcher chatbot)](https://weblangchain.vercel.app) and [repo](https://github.com/langchain-ai/weblangchain)
And much more! Head to the [Tutorials](https://python.langchain.com/v0.2/docs/tutorials/) section of the docs for more.
And much more! Head to the [Use cases](https://python.langchain.com/docs/use_cases/) section of the docs for more.
## 🚀 How does LangChain help?
The main value props of the LangChain libraries are:
@@ -87,50 +87,49 @@ Off-the-shelf chains make it easy to get started. Components make it easy to cus
LCEL is the foundation of many of LangChain's components, and is a declarative way to compose chains. LCEL was designed from day 1 to support putting prototypes in production, with no code changes, from the simplest “prompt + LLM” chain to the most complex chains.
- **[Overview](https://python.langchain.com/v0.2/docs/concepts/#langchain-expression-language-lcel)**: LCEL and its benefits
- **[Interface](https://python.langchain.com/v0.2/docs/concepts/#runnable-interface)**: The standard Runnable interface for LCEL objects
- **[Primitives](https://python.langchain.com/v0.2/docs/how_to/#langchain-expression-language-lcel)**: More on the primitives LCEL includes
- **[Cheatsheet](https://python.langchain.com/v0.2/docs/how_to/lcel_cheatsheet/)**: Quick overview of the most common usage patterns
- **[Overview](https://python.langchain.com/docs/expression_language/)**: LCEL and its benefits
- **[Interface](https://python.langchain.com/docs/expression_language/interface)**: The standard interface for LCEL objects
- **[Primitives](https://python.langchain.com/docs/expression_language/primitives)**: More on the primitives LCEL includes
## Components
Components fall into the following **modules**:
**📃 Model I/O**
**📃 Model I/O:**
This includes [prompt management](https://python.langchain.com/v0.2/docs/concepts/#prompt-templates), [prompt optimization](https://python.langchain.com/v0.2/docs/concepts/#example-selectors), a generic interface for [chat models](https://python.langchain.com/v0.2/docs/concepts/#chat-models) and [LLMs](https://python.langchain.com/v0.2/docs/concepts/#llms), and common utilities for working with [model outputs](https://python.langchain.com/v0.2/docs/concepts/#output-parsers).
This includes [prompt management](https://python.langchain.com/docs/modules/model_io/prompts/), [prompt optimization](https://python.langchain.com/docs/modules/model_io/prompts/example_selectors/), a generic interface for [chat models](https://python.langchain.com/docs/modules/model_io/chat/) and [LLMs](https://python.langchain.com/docs/modules/model_io/llms/), and common utilities for working with [model outputs](https://python.langchain.com/docs/modules/model_io/output_parsers/).
**📚 Retrieval**
**📚 Retrieval:**
Retrieval Augmented Generation involves [loading data](https://python.langchain.com/v0.2/docs/concepts/#document-loaders) from a variety of sources, [preparing it](https://python.langchain.com/v0.2/docs/concepts/#text-splitters), then [searching over (a.k.a. retrieving from)](https://python.langchain.com/v0.2/docs/concepts/#retrievers) it for use in the generation step.
Retrieval Augmented Generation involves [loading data](https://python.langchain.com/docs/modules/data_connection/document_loaders/) from a variety of sources, [preparing it](https://python.langchain.com/docs/modules/data_connection/document_loaders/), [then retrieving it](https://python.langchain.com/docs/modules/data_connection/retrievers/) for use in the generation step.
**🤖 Agents**
**🤖 Agents:**
Agents allow an LLM autonomy over how a task is accomplished. Agents make decisions about which Actions to take, then take that Action, observe the result, and repeat until the task is complete done. LangChain provides a [standard interface for agents](https://python.langchain.com/v0.2/docs/concepts/#agents) along with the [LangGraph](https://github.com/langchain-ai/langgraph) extension for building custom agents.
Agents allow an LLM autonomy over how a task is accomplished. Agents make decisions about which Actions to take, then take that Action, observe the result, and repeat until the task is complete done. LangChain provides a [standard interface for agents](https://python.langchain.com/docs/modules/agents/), a [selection of agents](https://python.langchain.com/docs/modules/agents/agent_types/) to choose from, and examples of end-to-end agents.
## 📖 Documentation
Please see [here](https://python.langchain.com) for full documentation, which includes:
- [Introduction](https://python.langchain.com/v0.2/docs/introduction/): Overview of the framework and the structure of the docs.
- [Tutorials](https://python.langchain.com/docs/use_cases/): If you're looking to build something specific or are more of a hands-on learner, check out our tutorials. This is the best place to get started.
-[How-to guides](https://python.langchain.com/v0.2/docs/how_to/): Answers to “How do I….?” type questions. These guides are goal-oriented and concrete; they're meant to help you complete a specific task.
- [Conceptual guide](https://python.langchain.com/v0.2/docs/concepts/): Conceptual explanations of the key parts of the framework.
- [API Reference](https://api.python.langchain.com): Thorough documentation of every class and method.
- [Getting started](https://python.langchain.com/docs/get_started/introduction): installation, setting up the environment, simple examples
- [Use case](https://python.langchain.com/docs/use_cases/) walkthroughs and best practice [guides](https://python.langchain.com/docs/guides/)
-Overviews of the [interfaces](https://python.langchain.com/docs/expression_language/), [components](https://python.langchain.com/docs/modules/), and [integrations](https://python.langchain.com/docs/integrations/providers)
You can also check out the full [API Reference docs](https://api.python.langchain.com).
## 🌐 Ecosystem
- [🦜🛠️ LangSmith](https://docs.smith.langchain.com/): Tracing and evaluating your language model applications and intelligent agents to help you move from prototype to production.
- [🦜🕸️ LangGraph](https://langchain-ai.github.io/langgraph/): Creating stateful, multi-actor applications with LLMs, built on top of (and intended to be used with) LangChain primitives.
- [🦜🛠️ LangSmith](https://python.langchain.com/docs/langsmith/): Tracing and evaluating your language model applications and intelligent agents to help you move from prototype to production.
- [🦜🕸️ LangGraph](https://python.langchain.com/docs/langgraph): Creating stateful, multi-actor applications with LLMs, built on top of (and intended to be used with) LangChain primitives.
- [🦜🏓 LangServe](https://python.langchain.com/docs/langserve): Deploying LangChain runnables and chains as REST APIs.
- [LangChain Templates](https://python.langchain.com/v0.2/docs/templates/): Example applications hosted with LangServe.
- [LangChain Templates](https://python.langchain.com/docs/templates/): Example applications hosted with LangServe.
## 💁 Contributing
As an open-source project in a rapidly developing field, we are extremely open to contributions, whether it be in the form of a new feature, improved infrastructure, or better documentation.
For detailed information on how to contribute, see [here](https://python.langchain.com/v0.2/docs/contributing/).
For detailed information on how to contribute, see [here](https://python.langchain.com/docs/contributing/).
[two_agent_debate_tools.ipynb](https://github.com/langchain-ai/langchain/tree/master/cookbook/two_agent_debate_tools.ipynb) | Simulate multi-agent dialogues where the agents can utilize various tools.
[two_player_dnd.ipynb](https://github.com/langchain-ai/langchain/tree/master/cookbook/two_player_dnd.ipynb) | Simulate a two-player dungeons & dragons game, where a dialogue simulator class is used to coordinate the dialogue between the protagonist and the dungeon master.
[wikibase_agent.ipynb](https://github.com/langchain-ai/langchain/tree/master/cookbook/wikibase_agent.ipynb) | Create a simple wikibase agent that utilizes sparql generation, with testing done on http://wikidata.org.
[oracleai_demo.ipynb](https://github.com/langchain-ai/langchain/tree/master/cookbook/oracleai_demo.ipynb) | This guide outlines how to utilize Oracle AI Vector Search alongside Langchain for an end-to-end RAG pipeline, providing step-by-step examples. The process includes loading documents from various sources using OracleDocLoader, summarizing them either within or outside the database with OracleSummary, and generating embeddings similarly through OracleEmbeddings. It also covers chunking documents according to specific requirements using Advanced Oracle Capabilities from OracleTextSplitter, and finally, storing and indexing these documents in a Vector Store for querying with OracleVS.
" raise ValueError(\"a KEYSPACE environment variable must be set\")\n",
"\n",
"session.set_keyspace(keyspace)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Setup Database"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"This needs to be done one time only!"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Download Data"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"The dataset used is from Kaggle, the [Environmental Sensor Telemetry Data](https://www.kaggle.com/datasets/garystafford/environmental-sensor-data-132k?select=iot_telemetry_data.csv). The next cell will download and unzip the data into a Pandas dataframe. The following cell is instructions to download manually. \n",
"\n",
"The net result of this section is you should have a Pandas dataframe variable `df`."
"with zip_file.open(csv_file_name) as csv_file:\n",
" df = pd.read_csv(csv_file)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"#### Download Manually"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"You can download the `.zip` file and unpack the `.csv` contained within. Comment in the next line, and adjust the path to this `.csv` file appropriately."
"WITH COMMENT = 'Data from environmental IoT room sensors. Columns include device identifier, timestamp (ts) of the data collection, carbon monoxide level (co), relative humidity, light presence, LPG concentration, motion detection, smoke concentration, and temperature (temp). Data is partitioned by day and device.';\n",
" description=\"A Python shell. Use this to execute python commands. Input should be a valid python command. If you want to see the output of a value, you should print it out with `print(...)`.\",\n",
"Here is your task: In the {keyspace} keyspace, find the total number of times the temperature of each device has exceeded 23 degrees on July 14, 2020.\n",
" Create a summary report including the name of the room. Use Pandas if helpful.\n",
"# Oracle AI Vector Search with Document Processing\n",
"Oracle AI Vector Search is designed for Artificial Intelligence (AI) workloads that allows you to query data based on semantics, rather than keywords.\n",
"One of the biggest benefits of Oracle AI Vector Search is that semantic search on unstructured data can be combined with relational search on business data in one single system.\n",
"This is not only powerful but also significantly more effective because you don't need to add a specialized vector database, eliminating the pain of data fragmentation between multiple systems.\n",
"\n",
"In addition, your vectors can benefit from all of Oracle Database’s most powerful features, like the following:\n",
"This guide demonstrates how Oracle AI Vector Search can be used with Langchain to serve an end-to-end RAG pipeline. This guide goes through examples of:\n",
"\n",
" * Loading the documents from various sources using OracleDocLoader\n",
" * Summarizing them within/outside the database using OracleSummary\n",
" * Generating embeddings for them within/outside the database using OracleEmbeddings\n",
" * Chunking them according to different requirements using Advanced Oracle Capabilities from OracleTextSplitter\n",
" * Storing and Indexing them in a Vector Store and querying them for queries in OracleVS"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"If you are just starting with Oracle Database, consider exploring the [free Oracle 23 AI](https://www.oracle.com/database/free/#resources) which provides a great introduction to setting up your database environment. While working with the database, it is often advisable to avoid using the system user by default; instead, you can create your own user for enhanced security and customization. For detailed steps on user creation, refer to our [end-to-end guide](https://github.com/langchain-ai/langchain/blob/master/cookbook/oracleai_demo.ipynb) which also shows how to set up a user in Oracle. Additionally, understanding user privileges is crucial for managing database security effectively. You can learn more about this topic in the official [Oracle guide](https://docs.oracle.com/en/database/oracle/oracle-database/19/admqs/administering-user-accounts-and-security.html#GUID-36B21D72-1BBB-46C9-A0C9-F0D2A8591B8D) on administering user accounts and security."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Prerequisites\n",
"\n",
"Please install Oracle Python Client driver to use Langchain with Oracle AI Vector Search. "
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# pip install oracledb"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Create Demo User\n",
"First, create a demo user with all the required privileges. "
]
},
{
"cell_type": "code",
"execution_count": 37,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Connection successful!\n",
"User setup done!\n"
]
}
],
"source": [
"import sys\n",
"\n",
"import oracledb\n",
"\n",
"# please update with your username, password, hostname and service_name\n",
"# please make sure this user has sufficient privileges to perform all below\n",
"Consider the following scenario: users possess documents stored either in an Oracle Database or a file system and intend to utilize this data with Oracle AI Vector Search powered by Langchain.\n",
"\n",
"To prepare the documents for analysis, a comprehensive preprocessing workflow is necessary. Initially, the documents must be retrieved, summarized (if required), and chunked as needed. Subsequent steps involve generating embeddings for these chunks and integrating them into the Oracle AI Vector Store. Users can then conduct semantic searches on this data.\n",
"\n",
"The Oracle AI Vector Search Langchain library encompasses a suite of document processing tools that facilitate document loading, chunking, summary generation, and embedding creation.\n",
"\n",
"In the sections that follow, we will detail the utilization of Oracle AI Langchain APIs to effectively implement each of these processes."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Connect to Demo User\n",
"The following sample code will show how to connect to Oracle Database. By default, python-oracledb runs in a ‘Thin’ mode which connects directly to Oracle Database. This mode does not need Oracle Client libraries. However, some additional functionality is available when python-oracledb uses them. Python-oracledb is said to be in ‘Thick’ mode when Oracle Client libraries are used. Both modes have comprehensive functionality supporting the Python Database API v2.0 Specification. See the following [guide](https://python-oracledb.readthedocs.io/en/latest/user_guide/appendix_a.html#featuresummary) that talks about features supported in each mode. You might want to switch to thick-mode if you are unable to use thin-mode."
]
},
{
"cell_type": "code",
"execution_count": 45,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Connection successful!\n"
]
}
],
"source": [
"import sys\n",
"\n",
"import oracledb\n",
"\n",
"# please update with your username, password, hostname and service_name\n",
" create_table_sql = \"\"\"create table demo_tab (id number, data clob)\"\"\"\n",
" cursor.execute(create_table_sql)\n",
"\n",
" insert_row_sql = \"\"\"insert into demo_tab values (:1, :2)\"\"\"\n",
" rows_to_insert = [\n",
" (\n",
" 1,\n",
" \"If the answer to any preceding questions is yes, then the database stops the search and allocates space from the specified tablespace; otherwise, space is allocated from the database default shared temporary tablespace.\",\n",
" ),\n",
" (\n",
" 2,\n",
" \"A tablespace can be online (accessible) or offline (not accessible) whenever the database is open.\\nA tablespace is usually online so that its data is available to users. The SYSTEM tablespace and temporary tablespaces cannot be taken offline.\",\n",
" ),\n",
" (\n",
" 3,\n",
" \"The database stores LOBs differently from other data types. Creating a LOB column implicitly creates a LOB segment and a LOB index. The tablespace containing the LOB segment and LOB index, which are always stored together, may be different from the tablespace containing the table.\\nSometimes the database can store small amounts of LOB data in the table itself rather than in a separate LOB segment.\",\n",
"With the inclusion of a demo user and a populated sample table, the remaining configuration involves setting up embedding and summary functionalities. Users are presented with multiple provider options, including local database solutions and third-party services such as Ocigenai, Hugging Face, and OpenAI. Should users opt for a third-party provider, they are required to establish credentials containing the necessary authentication details. Conversely, if selecting a database as the provider for embeddings, it is necessary to upload an ONNX model to the Oracle Database. No additional setup is required for summary functionalities when using the database option."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Load ONNX Model\n",
"\n",
"Oracle accommodates a variety of embedding providers, enabling users to choose between proprietary database solutions and third-party services such as OCIGENAI and HuggingFace. This selection dictates the methodology for generating and managing embeddings.\n",
"\n",
"***Important*** : Should users opt for the database option, they must upload an ONNX model into the Oracle Database. Conversely, if a third-party provider is selected for embedding generation, uploading an ONNX model to Oracle Database is not required.\n",
"\n",
"A significant advantage of utilizing an ONNX model directly within Oracle is the enhanced security and performance it offers by eliminating the need to transmit data to external parties. Additionally, this method avoids the latency typically associated with network or REST API calls.\n",
"\n",
"Below is the example code to upload an ONNX model into Oracle Database:"
"When selecting third-party providers for generating embeddings, users are required to establish credentials to securely access the provider's endpoints.\n",
"\n",
"***Important:*** No credentials are necessary when opting for the 'database' provider to generate embeddings. However, should users decide to utilize a third-party provider, they must create credentials specific to the chosen provider.\n",
"Users have the flexibility to load documents from either the Oracle Database, a file system, or both, by appropriately configuring the loader parameters. For comprehensive details on these parameters, please consult the [Oracle AI Vector Search Guide](https://docs.oracle.com/en/database/oracle/oracle-database/23/arpls/dbms_vector_chain1.html#GUID-73397E89-92FB-48ED-94BB-1AD960C4EA1F).\n",
"\n",
"A significant advantage of utilizing OracleDocLoader is its capability to process over 150 distinct file formats, eliminating the need for multiple loaders for different document types. For a complete list of the supported formats, please refer to the [Oracle Text Supported Document Formats](https://docs.oracle.com/en/database/oracle/oracle-database/23/ccref/oracle-text-supported-document-formats.html).\n",
"\n",
"Below is a sample code snippet that demonstrates how to use OracleDocLoader"
"Now that the user loaded the documents, they may want to generate a summary for each document. The Oracle AI Vector Search Langchain library offers a suite of APIs designed for document summarization. It supports multiple summarization providers such as Database, OCIGENAI, HuggingFace, among others, allowing users to select the provider that best meets their needs. To utilize these capabilities, users must configure the summary parameters as specified. For detailed information on these parameters, please consult the [Oracle AI Vector Search Guide book](https://docs.oracle.com/en/database/oracle/oracle-database/23/arpls/dbms_vector_chain1.html#GUID-EC9DDB58-6A15-4B36-BA66-ECBA20D2CE57)."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"***Note:*** The users may need to set proxy if they want to use some 3rd party summary generation providers other than Oracle's in-house and default provider: 'database'. If you don't have proxy, please remove the proxy parameter when you instantiate the OracleSummary."
]
},
{
"cell_type": "code",
"execution_count": 22,
"metadata": {},
"outputs": [],
"source": [
"# proxy to be used when we instantiate summary and embedder object\n",
"proxy = \"\""
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"The following sample code will show how to generate summary:"
"The documents may vary in size, ranging from small to very large. Users often prefer to chunk their documents into smaller sections to facilitate the generation of embeddings. A wide array of customization options is available for this splitting process. For comprehensive details regarding these parameters, please consult the [Oracle AI Vector Search Guide](https://docs.oracle.com/en/database/oracle/oracle-database/23/arpls/dbms_vector_chain1.html#GUID-4E145629-7098-4C7C-804F-FC85D1F24240).\n",
"\n",
"Below is a sample code illustrating how to implement this:"
"Now that the documents are chunked as per requirements, the users may want to generate embeddings for these chunks. Oracle AI Vector Search provides multiple methods for generating embeddings, utilizing either locally hosted ONNX models or third-party APIs. For comprehensive instructions on configuring these alternatives, please refer to the [Oracle AI Vector Search Guide](https://docs.oracle.com/en/database/oracle/oracle-database/23/arpls/dbms_vector_chain1.html#GUID-C6439E94-4E86-4ECD-954E-4B73D53579DE)."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"***Note:*** Users may need to configure a proxy to utilize third-party embedding generation providers, excluding the 'database' provider that utilizes an ONNX model."
]
},
{
"cell_type": "code",
"execution_count": 12,
"metadata": {},
"outputs": [],
"source": [
"# proxy to be used when we instantiate summary and embedder object\n",
"proxy = \"\""
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"The following sample code will show how to generate embeddings:"
"Now that you know how to use Oracle AI Langchain library APIs individually to process the documents, let us show how to integrate with Oracle AI Vector Store to facilitate the semantic searches."
"print(f\"Vector Store Table: {vectorstore.table_name}\")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"The example provided illustrates the creation of a vector store using the DOT_PRODUCT distance strategy. Users have the flexibility to employ various distance strategies with the Oracle AI Vector Store, as detailed in our [comprehensive guide](https://python.langchain.com/v0.1/docs/integrations/vectorstores/oracle/)."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"With embeddings now stored in vector stores, it is advisable to establish an index to enhance semantic search performance during query execution.\n",
"\n",
"***Note*** Should you encounter an \"insufficient memory\" error, it is recommended to increase the ***vector_memory_size*** in your database configuration\n",
"\n",
"Below is a sample code snippet for creating an index:"
"This example demonstrates the creation of a default HNSW index on embeddings within the 'oravs' table. Users may adjust various parameters according to their specific needs. For detailed information on these parameters, please consult the [Oracle AI Vector Search Guide book](https://docs.oracle.com/en/database/oracle/oracle-database/23/vecse/manage-different-categories-vector-indexes.html).\n",
"\n",
"Additionally, various types of vector indices can be created to meet diverse requirements. More details can be found in our [comprehensive guide](https://python.langchain.com/v0.1/docs/integrations/vectorstores/oracle/).\n"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Perform Semantic Search\n",
"All set!\n",
"\n",
"We have successfully processed the documents and stored them in the vector store, followed by the creation of an index to enhance query performance. We are now prepared to proceed with semantic searches.\n",
"\n",
"Below is the sample code for this process:"
]
},
{
"cell_type": "code",
"execution_count": 58,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"[Document(page_content='The database stores LOBs differently from other data types. Creating a LOB column implicitly creates a LOB segment and a LOB index. The tablespace containing the LOB segment and LOB index, which are always stored together, may be different from the tablespace containing the table. Sometimes the database can store small amounts of LOB data in the table itself rather than in a separate LOB segment.', metadata={'_oid': '662f2f257677f3c2311a8ff999fd34e5', '_rowid': 'AAAR/xAAEAAAAAnAAC', 'id': '662f2f257677f3c2311a8ff999fd34e5$3$1', 'document_id': '3', 'document_summary': 'Sometimes the database can store small amounts of LOB data in the table itself rather than in a separate LOB segment.\\n\\n'})]\n",
"[]\n",
"[(Document(page_content='The database stores LOBs differently from other data types. Creating a LOB column implicitly creates a LOB segment and a LOB index. The tablespace containing the LOB segment and LOB index, which are always stored together, may be different from the tablespace containing the table. Sometimes the database can store small amounts of LOB data in the table itself rather than in a separate LOB segment.', metadata={'_oid': '662f2f257677f3c2311a8ff999fd34e5', '_rowid': 'AAAR/xAAEAAAAAnAAC', 'id': '662f2f257677f3c2311a8ff999fd34e5$3$1', 'document_id': '3', 'document_summary': 'Sometimes the database can store small amounts of LOB data in the table itself rather than in a separate LOB segment.\\n\\n'}), 0.055675752460956573)]\n",
"[]\n",
"[Document(page_content='If the answer to any preceding questions is yes, then the database stops the search and allocates space from the specified tablespace; otherwise, space is allocated from the database default shared temporary tablespace.', metadata={'_oid': '662f2f253acf96b33b430b88699490a2', '_rowid': 'AAAR/xAAEAAAAAnAAA', 'id': '662f2f253acf96b33b430b88699490a2$1$1', 'document_id': '1', 'document_summary': 'If the answer to any preceding questions is yes, then the database stops the search and allocates space from the specified tablespace; otherwise, space is allocated from the database default shared temporary tablespace.\\n\\n'})]\n",
"[Document(page_content='If the answer to any preceding questions is yes, then the database stops the search and allocates space from the specified tablespace; otherwise, space is allocated from the database default shared temporary tablespace.', metadata={'_oid': '662f2f253acf96b33b430b88699490a2', '_rowid': 'AAAR/xAAEAAAAAnAAA', 'id': '662f2f253acf96b33b430b88699490a2$1$1', 'document_id': '1', 'document_summary': 'If the answer to any preceding questions is yes, then the database stops the search and allocates space from the specified tablespace; otherwise, space is allocated from the database default shared temporary tablespace.\\n\\n'})]\n"
" f\". Valid values are {sorted(latest_price['starrating'].value_counts().index.tolist())}\"\n",
")\n",
"attribute_info[3][\"description\"] += (\n",
" f\". Valid values are {sorted(latest_price['maxoccupancy'].value_counts().index.tolist())}\"\n",
")\n",
"attribute_info[-3][\"description\"] += (\n",
" f\". Valid values are {sorted(latest_price['country'].value_counts().index.tolist())}\"\n",
")"
"attribute_info[-2][\n",
" \"description\"\n",
"] += f\". Valid values are {sorted(latest_price['starrating'].value_counts().index.tolist())}\"\n",
"attribute_info[3][\n",
" \"description\"\n",
"] += f\". Valid values are {sorted(latest_price['maxoccupancy'].value_counts().index.tolist())}\"\n",
"attribute_info[-3][\n",
" \"description\"\n",
"] += f\". Valid values are {sorted(latest_price['country'].value_counts().index.tolist())}\""
]
},
{
@@ -688,9 +688,9 @@
"metadata": {},
"outputs": [],
"source": [
"attribute_info[-3][\"description\"] += (\n",
" \". NOTE: Only use the 'eq' operator if a specific country is mentioned. If a region is mentioned, include all relevant countries in filter.\"\n",
")\n",
"attribute_info[-3][\n",
" \"description\"\n",
"] += \". NOTE: Only use the 'eq' operator if a specific country is mentioned. If a region is mentioned, include all relevant countries in filter.\"\n",
| `2401.04088v1` [Mixtral of Experts](http://arxiv.org/abs/2401.04088v1) | Albert Q. Jiang, Alexandre Sablayrolles, Antoine Roux, et al. | 2024-01-08 | `Cookbook:` [together_ai](https://github.com/langchain-ai/langchain/blob/master/cookbook/together_ai.ipynb)
| `2312.06648v2` [Dense X Retrieval: What Retrieval Granularity Should We Use?](http://arxiv.org/abs/2312.06648v2) | Tong Chen, Hongwei Wang, Sihao Chen, et al. | 2023-12-11 | `Template:` [propositional-retrieval](https://python.langchain.com/docs/templates/propositional-retrieval)
| `2311.09210v1` [Chain-of-Note: Enhancing Robustness in Retrieval-Augmented Language Models](http://arxiv.org/abs/2311.09210v1) | Wenhao Yu, Hongming Zhang, Xiaoman Pan, et al. | 2023-11-15 | `Template:` [chain-of-note-wiki](https://python.langchain.com/docs/templates/chain-of-note-wiki)
| `2310.11511v1` [Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflection](http://arxiv.org/abs/2310.11511v1) | Akari Asai, Zeqiu Wu, Yizhong Wang, et al. | 2023-10-17 | `Cookbook:` [langgraph_self_rag](https://github.com/langchain-ai/langchain/blob/master/cookbook/langgraph_self_rag.ipynb)
| `2310.06117v2` [Take a Step Back: Evoking Reasoning via Abstraction in Large Language Models](http://arxiv.org/abs/2310.06117v2) | Huaixiu Steven Zheng, Swaroop Mishra, Xinyun Chen, et al. | 2023-10-09 | `Template:` [stepback-qa-prompting](https://python.langchain.com/docs/templates/stepback-qa-prompting), `Cookbook:` [stepback-qa](https://github.com/langchain-ai/langchain/blob/master/cookbook/stepback-qa.ipynb)
| `2307.09288v2` [Llama 2: Open Foundation and Fine-Tuned Chat Models](http://arxiv.org/abs/2307.09288v2) | Hugo Touvron, Louis Martin, Kevin Stone, et al. | 2023-07-18 | `Cookbook:` [Semi_Structured_RAG](https://github.com/langchain-ai/langchain/blob/master/cookbook/Semi_Structured_RAG.ipynb)
| `2305.14283v3` [Query Rewriting for Retrieval-Augmented Large Language Models](http://arxiv.org/abs/2305.14283v3) | Xinbei Ma, Yeyun Gong, Pengcheng He, et al. | 2023-05-23 | `Template:` [rewrite-retrieve-read](https://python.langchain.com/docs/templates/rewrite-retrieve-read), `Cookbook:` [rewrite](https://github.com/langchain-ai/langchain/blob/master/cookbook/rewrite.ipynb)
| `2305.08291v1` [Large Language Model Guided Tree-of-Thought](http://arxiv.org/abs/2305.08291v1) | Jieyi Long | 2023-05-15 | `API:` [langchain_experimental.tot](https://api.python.langchain.com/en/latest/experimental_api_reference.html#module-langchain_experimental.tot), `Cookbook:` [tree_of_thought](https://github.com/langchain-ai/langchain/blob/master/cookbook/tree_of_thought.ipynb)
| `2305.04091v3` [Plan-and-Solve Prompting: Improving Zero-Shot Chain-of-Thought Reasoning by Large Language Models](http://arxiv.org/abs/2305.04091v3) | Lei Wang, Wanyu Xu, Yihuai Lan, et al. | 2023-05-06 | `Cookbook:` [plan_and_execute_agent](https://github.com/langchain-ai/langchain/blob/master/cookbook/plan_and_execute_agent.ipynb)
| `2304.03442v2` [Generative Agents: Interactive Simulacra of Human Behavior](http://arxiv.org/abs/2304.03442v2) | Joon Sung Park, Joseph C. O'Brien, Carrie J. Cai, et al. | 2023-04-07 | `Cookbook:` [multiagent_bidding](https://github.com/langchain-ai/langchain/blob/master/cookbook/multiagent_bidding.ipynb), [generative_agents_interactive_simulacra_of_human_behavior](https://github.com/langchain-ai/langchain/blob/master/cookbook/generative_agents_interactive_simulacra_of_human_behavior.ipynb)
| `2303.17760v2` [CAMEL: Communicative Agents for "Mind" Exploration of Large Language Model Society](http://arxiv.org/abs/2303.17760v2) | Guohao Li, Hasan Abed Al Kader Hammoud, Hani Itani, et al. | 2023-03-31 | `Cookbook:` [camel_role_playing](https://github.com/langchain-ai/langchain/blob/master/cookbook/camel_role_playing.ipynb)
| `2303.17580v4` [HuggingGPT: Solving AI Tasks with ChatGPT and its Friends in Hugging Face](http://arxiv.org/abs/2303.17580v4) | Yongliang Shen, Kaitao Song, Xu Tan, et al. | 2023-03-30 | `API:` [langchain_experimental.autonomous_agents](https://api.python.langchain.com/en/latest/experimental_api_reference.html#module-langchain_experimental.autonomous_agents), `Cookbook:` [hugginggpt](https://github.com/langchain-ai/langchain/blob/master/cookbook/hugginggpt.ipynb)
| `2303.08774v6` [GPT-4 Technical Report](http://arxiv.org/abs/2303.08774v6) | OpenAI, Josh Achiam, Steven Adler, et al. | 2023-03-15 | `Docs:` [docs/integrations/vectorstores/mongodb_atlas](https://python.langchain.com/docs/integrations/vectorstores/mongodb_atlas)
| `2301.10226v4` [A Watermark for Large Language Models](http://arxiv.org/abs/2301.10226v4) | John Kirchenbauer, Jonas Geiping, Yuxin Wen, et al. | 2023-01-24 | `API:` [langchain_community.llms...HuggingFaceEndpoint](https://api.python.langchain.com/en/latest/llms/langchain_community.llms.huggingface_endpoint.HuggingFaceEndpoint.html#langchain_community.llms.huggingface_endpoint.HuggingFaceEndpoint), [langchain_community.llms...HuggingFaceTextGenInference](https://api.python.langchain.com/en/latest/llms/langchain_community.llms.huggingface_text_gen_inference.HuggingFaceTextGenInference.html#langchain_community.llms.huggingface_text_gen_inference.HuggingFaceTextGenInference), [langchain_huggingface.llms...HuggingFaceEndpoint](https://api.python.langchain.com/en/latest/llms/langchain_huggingface.llms.huggingface_endpoint.HuggingFaceEndpoint.html#langchain_huggingface.llms.huggingface_endpoint.HuggingFaceEndpoint), [langchain_community.llms...OCIModelDeploymentTGI](https://api.python.langchain.com/en/latest/llms/langchain_community.llms.oci_data_science_model_deployment_endpoint.OCIModelDeploymentTGI.html#langchain_community.llms.oci_data_science_model_deployment_endpoint.OCIModelDeploymentTGI)
| `2212.10496v1` [Precise Zero-Shot Dense Retrieval without Relevance Labels](http://arxiv.org/abs/2212.10496v1) | Luyu Gao, Xueguang Ma, Jimmy Lin, et al. | 2022-12-20 | `API:` [langchain.chains...HypotheticalDocumentEmbedder](https://api.python.langchain.com/en/latest/chains/langchain.chains.hyde.base.HypotheticalDocumentEmbedder.html#langchain.chains.hyde.base.HypotheticalDocumentEmbedder), `Template:` [hyde](https://python.langchain.com/docs/templates/hyde), `Cookbook:` [hypothetical_document_embeddings](https://github.com/langchain-ai/langchain/blob/master/cookbook/hypothetical_document_embeddings.ipynb)
| `2212.07425v3` [Robust and Explainable Identification of Logical Fallacies in Natural Language Arguments](http://arxiv.org/abs/2212.07425v3) | Zhivar Sourati, Vishnu Priya Prasanna Venkatesh, Darshan Deshpande, et al. | 2022-12-12 | `API:` [langchain_experimental.fallacy_removal](https://api.python.langchain.com/en/latest/experimental_api_reference.html#module-langchain_experimental.fallacy_removal)
| `2211.13892v2` [Complementary Explanations for Effective In-Context Learning](http://arxiv.org/abs/2211.13892v2) | Xi Ye, Srinivasan Iyer, Asli Celikyilmaz, et al. | 2022-11-25 | `API:` [langchain_core.example_selectors...MaxMarginalRelevanceExampleSelector](https://api.python.langchain.com/en/latest/example_selectors/langchain_core.example_selectors.semantic_similarity.MaxMarginalRelevanceExampleSelector.html#langchain_core.example_selectors.semantic_similarity.MaxMarginalRelevanceExampleSelector)
| `2211.10435v2` [PAL: Program-aided Language Models](http://arxiv.org/abs/2211.10435v2) | Luyu Gao, Aman Madaan, Shuyan Zhou, et al. | 2022-11-18 | `API:` [langchain_experimental.pal_chain](https://api.python.langchain.com/en/latest/experimental_api_reference.html#module-langchain_experimental.pal_chain), [langchain_experimental.pal_chain...PALChain](https://api.python.langchain.com/en/latest/pal_chain/langchain_experimental.pal_chain.base.PALChain.html#langchain_experimental.pal_chain.base.PALChain), `Cookbook:` [program_aided_language_model](https://github.com/langchain-ai/langchain/blob/master/cookbook/program_aided_language_model.ipynb)
| `2209.10785v2` [Deep Lake: a Lakehouse for Deep Learning](http://arxiv.org/abs/2209.10785v2) | Sasun Hambardzumyan, Abhinav Tuli, Levon Ghukasyan, et al. | 2022-09-22 | `Docs:` [docs/integrations/providers/activeloop_deeplake](https://python.langchain.com/docs/integrations/providers/activeloop_deeplake)
| `2205.12654v1` [Bitext Mining Using Distilled Sentence Representations for Low-Resource Languages](http://arxiv.org/abs/2205.12654v1) | Kevin Heffernan, Onur Çelebi, Holger Schwenk | 2022-05-25 | `API:` [langchain_community.embeddings...LaserEmbeddings](https://api.python.langchain.com/en/latest/embeddings/langchain_community.embeddings.laser.LaserEmbeddings.html#langchain_community.embeddings.laser.LaserEmbeddings)
| `2204.00498v1` [Evaluating the Text-to-SQL Capabilities of Large Language Models](http://arxiv.org/abs/2204.00498v1) | Nitarshan Rajkumar, Raymond Li, Dzmitry Bahdanau | 2022-03-15 | `API:` [langchain_community.utilities...SparkSQL](https://api.python.langchain.com/en/latest/utilities/langchain_community.utilities.spark_sql.SparkSQL.html#langchain_community.utilities.spark_sql.SparkSQL), [langchain_community.utilities...SQLDatabase](https://api.python.langchain.com/en/latest/utilities/langchain_community.utilities.sql_database.SQLDatabase.html#langchain_community.utilities.sql_database.SQLDatabase)
| `2202.00666v5` [Locally Typical Sampling](http://arxiv.org/abs/2202.00666v5) | Clara Meister, Tiago Pimentel, Gian Wiher, et al. | 2022-02-01 | `API:` [langchain_community.llms...HuggingFaceEndpoint](https://api.python.langchain.com/en/latest/llms/langchain_community.llms.huggingface_endpoint.HuggingFaceEndpoint.html#langchain_community.llms.huggingface_endpoint.HuggingFaceEndpoint), [langchain_community.llms...HuggingFaceTextGenInference](https://api.python.langchain.com/en/latest/llms/langchain_community.llms.huggingface_text_gen_inference.HuggingFaceTextGenInference.html#langchain_community.llms.huggingface_text_gen_inference.HuggingFaceTextGenInference), [langchain_huggingface.llms...HuggingFaceEndpoint](https://api.python.langchain.com/en/latest/llms/langchain_huggingface.llms.huggingface_endpoint.HuggingFaceEndpoint.html#langchain_huggingface.llms.huggingface_endpoint.HuggingFaceEndpoint)
| `2103.00020v1` [Learning Transferable Visual Models From Natural Language Supervision](http://arxiv.org/abs/2103.00020v1) | Alec Radford, Jong Wook Kim, Chris Hallacy, et al. | 2021-02-26 | `API:` [langchain_experimental.open_clip](https://api.python.langchain.com/en/latest/experimental_api_reference.html#module-langchain_experimental.open_clip)
| `1909.05858v2` [CTRL: A Conditional Transformer Language Model for Controllable Generation](http://arxiv.org/abs/1909.05858v2) | Nitish Shirish Keskar, Bryan McCann, Lav R. Varshney, et al. | 2019-09-11 | `API:` [langchain_community.llms...HuggingFaceEndpoint](https://api.python.langchain.com/en/latest/llms/langchain_community.llms.huggingface_endpoint.HuggingFaceEndpoint.html#langchain_community.llms.huggingface_endpoint.HuggingFaceEndpoint), [langchain_community.llms...HuggingFaceTextGenInference](https://api.python.langchain.com/en/latest/llms/langchain_community.llms.huggingface_text_gen_inference.HuggingFaceTextGenInference.html#langchain_community.llms.huggingface_text_gen_inference.HuggingFaceTextGenInference), [langchain_huggingface.llms...HuggingFaceEndpoint](https://api.python.langchain.com/en/latest/llms/langchain_huggingface.llms.huggingface_endpoint.HuggingFaceEndpoint.html#langchain_huggingface.llms.huggingface_endpoint.HuggingFaceEndpoint)
| `1908.10084v1` [Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks](http://arxiv.org/abs/1908.10084v1) | Nils Reimers, Iryna Gurevych | 2019-08-27 | `Docs:` [docs/integrations/text_embedding/sentence_transformers](https://python.langchain.com/docs/integrations/text_embedding/sentence_transformers)
## Self-Discover: Large Language Models Self-Compose Reasoning Structures
- **arXiv id:** 2402.03620v1
- **Title:** Self-Discover: Large Language Models Self-Compose Reasoning Structures
- **Authors:** Pei Zhou, Jay Pujara, Xiang Ren, et al.
### [Tutorials on YouTube](/docs/additional_resources/tutorials/#tutorials)
### Introduction to LangChain with Harrison Chase, creator of LangChain
- [Building the Future with LLMs, `LangChain`, & `Pinecone`](https://youtu.be/nMniwlGyX-c) by [Pinecone](https://www.youtube.com/@pinecone-io)
- [LangChain and Weaviate with Harrison Chase and Bob van Luijt - Weaviate Podcast #36](https://youtu.be/lhby7Ql7hbk) by [Weaviate • Vector Database](https://www.youtube.com/@Weaviate)
- [LangChain Demo + Q&A with Harrison Chase](https://youtu.be/zaYTXQFR0_s?t=788) by [Full Stack Deep Learning](https://www.youtube.com/@The_Full_Stack)
- [LangChain Agents: Build Personal Assistants For Your Data (Q&A with Harrison Chase and Mayo Oshin)](https://youtu.be/gVkF8cwfBLI) by [Chat with data](https://www.youtube.com/@chatwithdata)
## Videos (sorted by views)
Only videos with 40K+ views:
- [Using `ChatGPT` with YOUR OWN Data. This is magical. (LangChain OpenAI API)](https://youtu.be/9AXP7tCI9PI) by [TechLead](https://www.youtube.com/@TechLead)
- [First look - `ChatGPT` + `WolframAlpha` (`GPT-3.5` and Wolfram|Alpha via LangChain by James Weaver)](https://youtu.be/wYGbY811oMo) by [Dr Alan D. Thompson](https://www.youtube.com/@DrAlanDThompson)
- [LangChain explained - The hottest new Python framework](https://youtu.be/RoR4XJw8wIc) by [AssemblyAI](https://www.youtube.com/@AssemblyAI)
- [Chatbot with INFINITE MEMORY using `OpenAI` & `Pinecone` - `GPT-3`, `Embeddings`, `ADA`, `Vector DB`, `Semantic`](https://youtu.be/2xNzB7xq8nk) by [David Shapiro ~ AI](https://www.youtube.com/@DaveShap)
- [LangChain for LLMs is... basically just an Ansible playbook](https://youtu.be/X51N9C-OhlE) by [David Shapiro ~ AI](https://www.youtube.com/@DaveShap)
- [Build your own LLM Apps with LangChain & `GPT-Index`](https://youtu.be/-75p09zFUJY) by [1littlecoder](https://www.youtube.com/@1littlecoder)
- [`BabyAGI` - New System of Autonomous AI Agents with LangChain](https://youtu.be/lg3kJvf1kXo) by [1littlecoder](https://www.youtube.com/@1littlecoder)
- [Run `BabyAGI` with Langchain Agents (with Python Code)](https://youtu.be/WosPGHPObx8) by [1littlecoder](https://www.youtube.com/@1littlecoder)
- [How to Use Langchain With `Zapier` | Write and Send Email with GPT-3 | OpenAI API Tutorial](https://youtu.be/p9v2-xEa9A0) by [StarMorph AI](https://www.youtube.com/@starmorph)
- [Use Your Locally Stored Files To Get Response From GPT - `OpenAI` | Langchain | Python](https://youtu.be/NC1Ni9KS-rk) by [Shweta Lodha](https://www.youtube.com/@shweta-lodha)
- [`Langchain JS` | How to Use GPT-3, GPT-4 to Reference your own Data | `OpenAI Embeddings` Intro](https://youtu.be/veV2I-NEjaM) by [StarMorph AI](https://www.youtube.com/@starmorph)
- [The easiest way to work with large language models | Learn LangChain in 10min](https://youtu.be/kmbS6FDQh7c) by [Sophia Yang](https://www.youtube.com/@SophiaYangDS)
- [4 Autonomous AI Agents: “Westworld” simulation `BabyAGI`, `AutoGPT`, `Camel`, `LangChain`](https://youtu.be/yWbnH6inT_U) by [Sophia Yang](https://www.youtube.com/@SophiaYangDS)
- [AI CAN SEARCH THE INTERNET? Langchain Agents + OpenAI ChatGPT](https://youtu.be/J-GL0htqda8) by [tylerwhatsgood](https://www.youtube.com/@tylerwhatsgood)
- [Query Your Data with GPT-4 | Embeddings, Vector Databases | Langchain JS Knowledgebase](https://youtu.be/jRnUPUTkZmU) by [StarMorph AI](https://www.youtube.com/@starmorph)
- [`Weaviate` + LangChain for LLM apps presented by Erika Cardenas](https://youtu.be/7AGj4Td5Lgw) by [`Weaviate` • Vector Database](https://www.youtube.com/@Weaviate)
- [Langchain Overview — How to Use Langchain & `ChatGPT`](https://youtu.be/oYVYIq0lOtI) by [Python In Office](https://www.youtube.com/@pythoninoffice6568)
- [Langchain Overview - How to Use Langchain & `ChatGPT`](https://youtu.be/oYVYIq0lOtI) by [Python In Office](https://www.youtube.com/@pythoninoffice6568)
- [LangChain Tutorials](https://www.youtube.com/watch?v=FuqdVNB_8c0&list=PL9V0lbeJ69brU-ojMpU1Y7Ic58Tap0Cw6) by [Edrick](https://www.youtube.com/@edrickdch):
- [LangChain, Chroma DB, OpenAI Beginner Guide | ChatGPT with your PDF](https://youtu.be/FuqdVNB_8c0)
- [LangChain 101: The Complete Beginner's Guide](https://youtu.be/P3MAbZ2eMUI)
- [Custom langchain Agent & Tools with memory. Turn any `Python function` into langchain tool with Gpt 3](https://youtu.be/NIG8lXk0ULg) by [echohive](https://www.youtube.com/@echohive)
- [Building AI LLM Apps with LangChain (and more?) - LIVE STREAM](https://www.youtube.com/live/M-2Cj_2fzWI?feature=share) by [Nicholas Renotte](https://www.youtube.com/@NicholasRenotte)
- [`ChatGPT` with any `YouTube` video using langchain and `chromadb`](https://youtu.be/TQZfB2bzVwU) by [echohive](https://www.youtube.com/@echohive)
- [How to Talk to a `PDF` using LangChain and `ChatGPT`](https://youtu.be/v2i1YDtrIwk) by [Automata Learning Lab](https://www.youtube.com/@automatalearninglab)
- [Langchain Document Loaders Part 1: Unstructured Files](https://youtu.be/O5C0wfsen98) by [Merk](https://www.youtube.com/@heymichaeldaigler)
- [LangChain - Prompt Templates (what all the best prompt engineers use)](https://youtu.be/1aRu8b0XNOQ) by [Nick Daigler](https://www.youtube.com/@nickdaigler)
- [LangChain. Crear aplicaciones Python impulsadas por GPT](https://youtu.be/DkW_rDndts8) by [Jesús Conde](https://www.youtube.com/@0utKast)
- [Easiest Way to Use GPT In Your Products | LangChain Basics Tutorial](https://youtu.be/fLy0VenZyGc) by [Rachel Woods](https://www.youtube.com/@therachelwoods)
- [`BabyAGI` + `GPT-4` Langchain Agent with Internet Access](https://youtu.be/wx1z_hs5P6E) by [tylerwhatsgood](https://www.youtube.com/@tylerwhatsgood)
- [Learning LLM Agents. How does it actually work? LangChain, AutoGPT & OpenAI](https://youtu.be/mb_YAABSplk) by [Arnoldas Kemeklis](https://www.youtube.com/@processusAI)
- [Get Started with LangChain in `Node.js`](https://youtu.be/Wxx1KUWJFv4) by [Developers Digest](https://www.youtube.com/@DevelopersDigest)
- [LangChain + `OpenAI` tutorial: Building a Q&A system w/ own text data](https://youtu.be/DYOU_Z0hAwo) by [Samuel Chan](https://www.youtube.com/@SamuelChan)
- [Langchain + `Zapier` Agent](https://youtu.be/yribLAb-pxA) by [Merk](https://www.youtube.com/@heymichaeldaigler)
- [Connecting the Internet with `ChatGPT` (LLMs) using Langchain And Answers Your Questions](https://youtu.be/9Y0TBC63yZg) by [Kamalraj M M](https://www.youtube.com/@insightbuilder)
- [Build More Powerful LLM Applications for Business’s with LangChain (Beginners Guide)](https://youtu.be/sp3-WLKEcBg) by[ No Code Blackbox](https://www.youtube.com/@nocodeblackbox)
- [LangFlow LLM Agent Demo for 🦜🔗LangChain](https://youtu.be/zJxDHaWt-6o) by [Cobus Greyling](https://www.youtube.com/@CobusGreylingZA)
- [Chatbot Factory: Streamline Python Chatbot Creation with LLMs and Langchain](https://youtu.be/eYer3uzrcuM) by [Finxter](https://www.youtube.com/@CobusGreylingZA)
- [LangChain Tutorial - ChatGPT mit eigenen Daten](https://youtu.be/0XDLyY90E2c) by [Coding Crashkurse](https://www.youtube.com/@codingcrashkurse6429)
- [Chat with a `CSV` | LangChain Agents Tutorial (Beginners)](https://youtu.be/tjeti5vXWOU) by [GoDataProf](https://www.youtube.com/@godataprof)
- [Introdução ao Langchain - #Cortes - Live DataHackers](https://youtu.be/fw8y5VRei5Y) by [Prof. João Gabriel Lima](https://www.youtube.com/@profjoaogabriellima)
- [LangChain: Level up `ChatGPT` !? | LangChain Tutorial Part 1](https://youtu.be/vxUGx8aZpDE) by [Code Affinity](https://www.youtube.com/@codeaffinitydev)
- [Chat with Audio: Langchain, `Chroma DB`, OpenAI, and `Assembly AI`](https://youtu.be/Kjy7cx1r75g) by [AI Anytime](https://www.youtube.com/@AIAnytime)
- [QA over documents with Auto vector index selection with Langchain router chains](https://youtu.be/9G05qybShv8) by [echohive](https://www.youtube.com/@echohive)
- [Build your own custom LLM application with `Bubble.io` & Langchain (No Code & Beginner friendly)](https://youtu.be/O7NhQGu1m6c) by [No Code Blackbox](https://www.youtube.com/@nocodeblackbox)
- [Simple App to Question Your Docs: Leveraging `Streamlit`, `Hugging Face Spaces`, LangChain, and `Claude`!](https://youtu.be/X4YbNECRr7o) by [Chris Alexiuk](https://www.youtube.com/@chrisalexiuk)
- [LANGCHAIN AI- `ConstitutionalChainAI` + Databutton AI ASSISTANT Web App](https://youtu.be/5zIU6_rdJCU) by [Avra](https://www.youtube.com/@Avra_b)
- [LANGCHAIN AI AUTONOMOUS AGENT WEB APP - 👶 `BABY AGI` 🤖 with EMAIL AUTOMATION using `DATABUTTON`](https://youtu.be/cvAwOGfeHgw) by [Avra](https://www.youtube.com/@Avra_b)
- [The Future of Data Analysis: Using A.I. Models in Data Analysis (LangChain)](https://youtu.be/v_LIcVyg5dk) by [Absent Data](https://www.youtube.com/@absentdata)
- [Memory in LangChain | Deep dive (python)](https://youtu.be/70lqvTFh_Yg) by [Eden Marco](https://www.youtube.com/@EdenMarco)
- [Use Large Language Models in Jupyter Notebook | LangChain | Agents & Indexes](https://youtu.be/JSe11L1a_QQ) by [Abhinaw Tiwari](https://www.youtube.com/@AbhinawTiwariAT)
- [How to Talk to Your Langchain Agent | `11 Labs` + `Whisper`](https://youtu.be/N4k459Zw2PU) by [VRSEN](https://www.youtube.com/@vrsen)
- [LangChain Deep Dive: 5 FUN AI App Ideas To Build Quickly and Easily](https://youtu.be/mPYEPzLkeks) by [James NoCode](https://www.youtube.com/@jamesnocode)
- [LangChain 101: Models](https://youtu.be/T6c_XsyaNSQ) by [Mckay Wrigley](https://www.youtube.com/@realmckaywrigley)
- [LangChain with JavaScript Tutorial #1 | Setup & Using LLMs](https://youtu.be/W3AoeMrg27o) by [Leon van Zyl](https://www.youtube.com/@leonvanzyl)
- [LangChain Overview & Tutorial for Beginners: Build Powerful AI Apps Quickly & Easily (ZERO CODE)](https://youtu.be/iI84yym473Q) by [James NoCode](https://www.youtube.com/@jamesnocode)
- [LangChain In Action: Real-World Use Case With Step-by-Step Tutorial](https://youtu.be/UO699Szp82M) by [Rabbitmetrics](https://www.youtube.com/@rabbitmetrics)
- [Summarizing and Querying Multiple Papers with LangChain](https://youtu.be/p_MQRWH5Y6k) by [Automata Learning Lab](https://www.youtube.com/@automatalearninglab)
- [Using Langchain (and `Replit`) through `Tana`, ask `Google`/`Wikipedia`/`Wolfram Alpha` to fill out a table](https://youtu.be/Webau9lEzoI) by [Stian Håklev](https://www.youtube.com/@StianHaklev)
- [Langchain PDF App (GUI) | Create a ChatGPT For Your `PDF` in Python](https://youtu.be/wUAUdEw5oxM) by [Alejandro AO - Software & Ai](https://www.youtube.com/@alejandro_ao)
- [Auto-GPT with LangChain 🔥 | Create Your Own Personal AI Assistant](https://youtu.be/imDfPmMKEjM) by [Data Science Basics](https://www.youtube.com/@datasciencebasics)
- [Create Your OWN Slack AI Assistant with Python & LangChain](https://youtu.be/3jFXRNn2Bu8) by [Dave Ebbelaar](https://www.youtube.com/@daveebbelaar)
- [How to Create LOCAL Chatbots with GPT4All and LangChain [Full Guide]](https://youtu.be/4p1Fojur8Zw) by [Liam Ottley](https://www.youtube.com/@LiamOttley)
- [Build a `Multilingual PDF` Search App with LangChain, `Cohere` and `Bubble`](https://youtu.be/hOrtuumOrv8) by [Menlo Park Lab](https://www.youtube.com/@menloparklab)
- [Building a LangChain Agent (code-free!) Using `Bubble` and `Flowise`](https://youtu.be/jDJIIVWTZDE) by [Menlo Park Lab](https://www.youtube.com/@menloparklab)
- [Build a LangChain-based Semantic PDF Search App with No-Code Tools Bubble and Flowise](https://youtu.be/s33v5cIeqA4) by [Menlo Park Lab](https://www.youtube.com/@menloparklab)
- [LangChain Memory Tutorial | Building a ChatGPT Clone in Python](https://youtu.be/Cwq91cj2Pnc) by [Alejandro AO - Software & Ai](https://www.youtube.com/@alejandro_ao)
- [ChatGPT For Your DATA | Chat with Multiple Documents Using LangChain](https://youtu.be/TeDgIDqQmzs) by [Data Science Basics](https://www.youtube.com/@datasciencebasics)
- [`Llama Index`: Chat with Documentation using URL Loader](https://youtu.be/XJRoDEctAwA) by [Merk](https://www.youtube.com/@heymichaeldaigler)
- [Using OpenAI, LangChain, and `Gradio` to Build Custom GenAI Applications](https://youtu.be/1MsmqMg3yUc) by [David Hundley](https://www.youtube.com/@dkhundley)
- [LangChain, Chroma DB, OpenAI Beginner Guide | ChatGPT with your PDF](https://youtu.be/FuqdVNB_8c0)
- [Build AI chatbot with custom knowledge base using OpenAI API and GPT Index](https://youtu.be/vDZAZuaXf48) by [Irina Nik](https://www.youtube.com/@irina_nik)
- [Build Your Own Auto-GPT Apps with LangChain (Python Tutorial)](https://youtu.be/NYSWn1ipbgg) by [Dave Ebbelaar](https://www.youtube.com/@daveebbelaar)
- [Chat with Multiple `PDFs` | LangChain App Tutorial in Python (Free LLMs and Embeddings)](https://youtu.be/dXxQ0LR-3Hg) by [Alejandro AO - Software & Ai](https://www.youtube.com/@alejandro_ao)
- [Chat with a `CSV` | `LangChain Agents` Tutorial (Beginners)](https://youtu.be/tjeti5vXWOU) by [Alejandro AO - Software & Ai](https://www.youtube.com/@alejandro_ao)
- [Create Your Own ChatGPT with `PDF` Data in 5 Minutes (LangChain Tutorial)](https://youtu.be/au2WVVGUvc8) by [Liam Ottley](https://www.youtube.com/@LiamOttley)
- [Build a Custom Chatbot with OpenAI: `GPT-Index` & LangChain | Step-by-Step Tutorial](https://youtu.be/FIDv6nc4CgU) by [Fabrikod](https://www.youtube.com/@fabrikod)
- [`Flowise` is an open-source no-code UI visual tool to build 🦜🔗LangChain applications](https://youtu.be/CovAPtQPU0k) by [Cobus Greyling](https://www.youtube.com/@CobusGreylingZA)
- [LangChain & GPT 4 For Data Analysis: The `Pandas` Dataframe Agent](https://youtu.be/rFQ5Kmkd4jc) by [Rabbitmetrics](https://www.youtube.com/@rabbitmetrics)
- [`GirlfriendGPT` - AI girlfriend with LangChain](https://youtu.be/LiN3D1QZGQw) by [Girlfriend GPT](https://www.youtube.com/@girlfriendGPT)
- [How to build with Langchain 10x easier | ⛓️ LangFlow & `Flowise`](https://youtu.be/Ya1oGL7ZTvU) by [AI Jason](https://www.youtube.com/@AIJasonZ)
- [Getting Started With LangChain In 20 Minutes- Build Celebrity Search Application](https://youtu.be/_FpT1cwcSLg) by [Krish Naik](https://www.youtube.com/@krishnaik06)
- ⛓ [Vector Embeddings Tutorial – Code Your Own AI Assistant with `GPT-4 API` + LangChain + NLP](https://youtu.be/yfHHvmaMkcA?si=5uJhxoh2tvdnOXok) by [FreeCodeCamp.org](https://www.youtube.com/@freecodecamp)
- ⛓ [Fully LOCAL `Llama 2` Q&A with LangChain](https://youtu.be/wgYctKFnQ74?si=UX1F3W-B3MqF4-K-) by [1littlecoder](https://www.youtube.com/@1littlecoder)
- ⛓ [Fully LOCAL `Llama 2` Langchain on CPU](https://youtu.be/yhECvKMu8kM?si=IvjxwlA1c09VwHZ4) by [1littlecoder](https://www.youtube.com/@1littlecoder)
- ⛓ [Build LangChain Audio Apps with Python in 5 Minutes](https://youtu.be/7w7ysaDz2W4?si=BvdMiyHhormr2-vr) by [AssemblyAI](https://www.youtube.com/@AssemblyAI)
- ⛓ [`Voiceflow` & `Flowise`: Want to Beat Competition? New Tutorial with Real AI Chatbot](https://youtu.be/EZKkmeFwag0?si=-4dETYDHEstiK_bb) by [AI SIMP](https://www.youtube.com/@aisimp)
- ⛓ [THIS Is How You Build Production-Ready AI Apps (`LangSmith` Tutorial)](https://youtu.be/tFXm5ijih98?si=lfiqpyaivxHFyI94) by [Dave Ebbelaar](https://www.youtube.com/@daveebbelaar)
- ⛓ [Build POWERFUL LLM Bots EASILY with Your Own Data - `Embedchain` - Langchain 2.0? (Tutorial)](https://youtu.be/jE24Y_GasE8?si=0yEDZt3BK5Q-LIuF) by [WorldofAI](https://www.youtube.com/@intheworldofai)
- ⛓ [`Code Llama` powered Gradio App for Coding: Runs on CPU](https://youtu.be/AJOhV6Ryy5o?si=ouuQT6IghYlc1NEJ) by [AI Anytime](https://www.youtube.com/@AIAnytime)
- ⛓ [LangChain Complete Course in One Video | Develop LangChain (AI) Based Solutions for Your Business](https://youtu.be/j9mQd-MyIg8?si=_wlNT3nP2LpDKztZ) by [UBprogrammer](https://www.youtube.com/@UBprogrammer)
- ⛓ [How to Run `LLaMA` Locally on CPU or GPU | Python & Langchain & CTransformers Guide](https://youtu.be/SvjWDX2NqiM?si=DxFml8XeGhiLTzLV) by [Code With Prince](https://www.youtube.com/@CodeWithPrince)
- ⛓ [Prompt Engineering in Web Development | Using LangChain and Templates with OpenAI](https://youtu.be/pK6WzlTOlYw?si=fkcDQsBG2h-DM8uQ) by [Akamai Developer
](https://www.youtube.com/@AkamaiDeveloper)
- ⛓ [Retrieval-Augmented Generation (RAG) using LangChain and `Pinecone` - The RAG Special Episode](https://youtu.be/J_tCD_J6w3s?si=60Mnr5VD9UED9bGG) by [Generative AI and Data Science On AWS](https://www.youtube.com/@GenerativeAIOnAWS)
- ⛓ [`LLAMA2 70b-chat` Multiple Documents Chatbot with Langchain & Streamlit |All OPEN SOURCE|Replicate API](https://youtu.be/vhghB81vViM?si=dszzJnArMeac7lyc) by [DataInsightEdge](https://www.youtube.com/@DataInsightEdge01)
- ⛓ [Chatting with 44K Fashion Products: LangChain Opportunities and Pitfalls](https://youtu.be/Zudgske0F_s?si=8HSshHoEhh0PemJA) by [Rabbitmetrics](https://www.youtube.com/@rabbitmetrics)
- ⛓ [Structured Data Extraction from `ChatGPT` with LangChain](https://youtu.be/q1lYg8JISpQ?si=0HctzOHYZvq62sve) by [MG](https://www.youtube.com/@MG_cafe)
- ⛓ [Chat with Multiple PDFs using `Llama 2`, `Pinecone` and LangChain (Free LLMs and Embeddings)](https://youtu.be/TcJ_tVSGS4g?si=FZYnMDJyoFfL3Z2i) by [Muhammad Moin](https://www.youtube.com/@muhammadmoinfaisal)
- ⛓ [Integrate Audio into `LangChain.js` apps in 5 Minutes](https://youtu.be/hNpUSaYZIzs?si=Gb9h7W9A8lzfvFKi) by [AssemblyAI](https://www.youtube.com/@AssemblyAI)
- ⛓ [`ChatGPT` for your data with Local LLM](https://youtu.be/bWrjpwhHEMU?si=uM6ZZ18z9og4M90u) by [Jacob Jedryszek](https://www.youtube.com/@jj09)
- ⛓ [Training `Chatgpt` with your personal data using langchain step by step in detail](https://youtu.be/j3xOMde2v9Y?si=179HsiMU-hEPuSs4) by [NextGen Machines](https://www.youtube.com/@MayankGupta-kb5yc)
- ⛓ [Use ANY language in `LangSmith` with REST](https://youtu.be/7BL0GEdMmgY?si=iXfOEdBLqXF6hqRM) by [Nerding I/O](https://www.youtube.com/@nerding_io)
- ⛓ [How to Leverage the Full Potential of LLMs for Your Business with Langchain - Leon Ruddat](https://youtu.be/vZmoEa7oWMg?si=ZhMmydq7RtkZd56Q) by [PyData](https://www.youtube.com/@PyDataTV)
- ⛓ [`ChatCSV` App: Chat with CSV files using LangChain and `Llama 2`](https://youtu.be/PvsMg6jFs8E?si=Qzg5u5gijxj933Ya) by [Muhammad Moin](https://www.youtube.com/@muhammadmoinfaisal)
- ⛓ [Build Chat PDF app in Python with LangChain, OpenAI, Streamlit | Full project | Learn Coding](https://www.youtube.com/watch?v=WYzFzZg4YZI) by [Jutsupoint](https://www.youtube.com/@JutsuPoint)
- ⛓ [Build Eminem Bot App with LangChain, Streamlit, OpenAI | Full Python Project | Tutorial | AI ChatBot](https://www.youtube.com/watch?v=a2shHB4MRZ4) by [Jutsupoint](https://www.youtube.com/@JutsuPoint)
### [Prompt Engineering and LangChain](https://www.youtube.com/watch?v=muXbPpG_ys4&list=PLEJK-H61Xlwzm5FYLDdKt_6yibO33zoMW) by [Venelin Valkov](https://www.youtube.com/@venelin_valkov)
- [Getting Started with LangChain: Load Custom Data, Run OpenAI Models, Embeddings and `ChatGPT`](https://www.youtube.com/watch?v=muXbPpG_ys4)
- [Loaders, Indexes & Vectorstores in LangChain: Question Answering on `PDF` files with `ChatGPT`](https://www.youtube.com/watch?v=FQnvfR8Dmr0)
- [LangChain Chains: Use `ChatGPT` to Build Conversational Agents, Summaries and Q&A on Text With LLMs](https://www.youtube.com/watch?v=h1tJZQPcimM)
- [Analyze Custom CSV Data with `GPT-4` using Langchain](https://www.youtube.com/watch?v=Ew3sGdX8at4)
- [Build ChatGPT Chatbots with LangChain Memory: Understanding and Implementing Memory in Conversations](https://youtu.be/CyuUlf54wTs)
- [Using `ChatGPT` with YOUR OWN Data. This is magical. (LangChain `OpenAI API`)](https://youtu.be/9AXP7tCI9PI)
- [Chat with Multiple `PDFs` | LangChain App Tutorial in Python (Free LLMs and Embeddings)](https://youtu.be/dXxQ0LR-3Hg?si=pjXKhsHRzn10vOqX)
- [`Hugging Face` + Langchain in 5 mins | Access 200k+ FREE AI models for your AI apps](https://youtu.be/_j7JEDWuqLE?si=psimQscN3qo2dOa9)
- [LangChain Crash Course For Beginners | LangChain Tutorial](https://youtu.be/nAmC7SoVLd8?si=qJdvyG5-rnjqfdj1)
- [Vector Embeddings Tutorial – Code Your Own AI Assistant with GPT-4 API + LangChain + NLP](https://youtu.be/yfHHvmaMkcA?si=UBP3yw50cLm3a2nj)
- [Development with Large Language Models Tutorial – `OpenAI`, Langchain, Agents, `Chroma`](https://youtu.be/xZDB1naRUlk?si=v8J1q6oFHRyTkf7Y)
- [Langchain: `PDF` Chat App (GUI) | ChatGPT for Your PDF FILES | Step-by-Step Tutorial](https://youtu.be/RIWbalZ7sTo?si=LbKsCcuyv0BtnrTY)
- [Vector Search `RAG` Tutorial – Combine Your Data with LLMs with Advanced Search](https://youtu.be/JEBDfGqrAUA?si=pD7oxpfwWeJCxfBt)
- [LangChain Crash Course for Beginners](https://youtu.be/lG7Uxts9SXs?si=Yte4S5afN7KNCw0F)
- [Learn `RAG` From Scratch – Python AI Tutorial from a LangChain Engineer](https://youtu.be/sVcwVQRHIc8?si=_LN4g0vOgSdtlB3S)
- [`Llama 2` in LangChain — FIRST Open Source Conversational Agent!](https://youtu.be/6iHVJyX2e50?si=rtq1maPrzWKHbwVV)
- [LangChain Tutorial for Beginners | Generative AI Series](https://youtu.be/cQUUkZnyoD0?si=KYz-bvcocdqGh9f_)
- [Chatbots with `RAG`: LangChain Full Walkthrough](https://youtu.be/LhnCsygAvzY?si=yS7T98VLfcWdkDek)
- [LangChain Explained In 15 Minutes - A MUST Learn For Python Programmers](https://youtu.be/mrjq3lFz23s?si=wkQGcSKUJjuiiEPf)
- [LLM Project | End to End LLM Project Using Langchain, `OpenAI` in Finance Domain](https://youtu.be/MoqgmWV1fm8?si=oVl-5kJVgd3a07Y_)
- [What is LangChain?](https://youtu.be/1bUy-1hGZpI?si=NZ0D51VM5y-DhjGe)
- [`RAG` + Langchain Python Project: Easy AI/Chat For Your Doc](https://youtu.be/tcqEUSNCn8I?si=RLcWPBVLIErRqdmU)
- [Getting Started With LangChain In 20 Minutes- Build Celebrity Search Application](https://youtu.be/_FpT1cwcSLg?si=X9qVazlXYucN_JBP)
- [LangChain GEN AI Tutorial – 6 End-to-End Projects using OpenAI, Google `Gemini Pro`, `LLAMA2`](https://youtu.be/x0AnCE9SE4A?si=_92gJYm7kb-V2bi0)
- [Complete Langchain GEN AI Crash Course With 6 End To End LLM Projects With OPENAI, `LLAMA2`, `Gemini Pro`](https://youtu.be/aWKrL4z5H6w?si=NVLi7Yiq0ccE7xXE)
- [AI Leader Reveals The Future of AI AGENTS (LangChain CEO)](https://youtu.be/9ZhbA0FHZYc?si=1r4P6kRvKVvEhRgE)
- [Learn How To Query Pdf using Langchain Open AI in 5 min](https://youtu.be/5Ghv-F1wF_0?si=ZZRjrWfeiFOVrcvu)
- [Reliable, fully local RAG agents with `LLaMA3`](https://youtu.be/-ROS6gfYIts?si=75CXA8W_BbnkIxcV)
- [Learn `LangChain.js` - Build LLM apps with JavaScript and `OpenAI`](https://youtu.be/HSZ_uaif57o?si=Icj-RAhwMT-vHaYA)
- [LLM Project | End to End LLM Project Using LangChain, Google Palm In Ed-Tech Industry](https://youtu.be/AjQPRomyd-k?si=eC3NT6kn02Lhpz-_)
- [Chatbot Answering from Your Own Knowledge Base: Langchain, `ChatGPT`, `Pinecone`, and `Streamlit`: | Code](https://youtu.be/nAKhxQ3hcMA?si=9Zd_Nd_jiYhtml5w)
- [LangChain is AMAZING | Quick Python Tutorial](https://youtu.be/I4mFqyqFkxg?si=aJ66qh558OfNAczD)
- [`GirlfriendGPT` - AI girlfriend with LangChain](https://youtu.be/LiN3D1QZGQw?si=kZR-lnJwixeVrjmh)
- [Using NEW `MPT-7B` in `Hugging Face` and LangChain](https://youtu.be/DXpk9K7DgMo?si=99JDpV_ueimwJhMi)
- [LangChain - COMPLETE TUTORIAL - Basics to advanced concept!](https://youtu.be/a89vqgK-Qcs?si=0aVO2EOqsw7GE5e3)
- [Chat With Multiple `PDF` Documents With Langchain And Google `Gemini Pro`](https://youtu.be/uus5eLz6smA?si=YUwvHtaZsGeIl0WD)
- [LLM Project | End to end LLM project Using Langchain, `Google Palm` in Retail Industry](https://youtu.be/4wtrl4hnPT8?si=_eOKPpdLfWu5UXMQ)
- [Tutorial | Chat with any Website using Python and Langchain](https://youtu.be/bupx08ZgSFg?si=KRrjYZFnuLsstGwW)
- [Prompt Engineering And LLM's With LangChain In One Shot-Generative AI](https://youtu.be/t2bSApmPzU4?si=87vPQQtYEWTyu2Kx)
- [Build a Custom Chatbot with `OpenAI`: `GPT-Index` & LangChain | Step-by-Step Tutorial](https://youtu.be/FIDv6nc4CgU?si=gR1u3DUG9lvzBIKK)
- [Search Your `PDF` App using Langchain, `ChromaDB`, and Open Source LLM: No OpenAI API (Runs on CPU)](https://youtu.be/rIV1EseKwU4?si=UxZEoXSiPai8fXgl)
- [Building a `RAG` application from scratch using Python, LangChain, and the `OpenAI API`](https://youtu.be/BrsocJb-fAo?si=hvkh9iTGzJ-LnsX-)
- [Function Calling via `ChatGPT API` - First Look With LangChain](https://youtu.be/0-zlUy7VUjg?si=Vc6LFseckEc6qvuk)
- [Private GPT, free deployment! Langchain-Chachat helps you easily play with major mainstream AI models! | Zero Degree Commentary](https://youtu.be/3LLUyaHP-3I?si=AZumEeFXsvqaLl0f)
- [Create a ChatGPT clone using `Streamlit` and LangChain](https://youtu.be/IaTiyQ2oYUQ?si=WbgsYmqPDnMidSUK)
- [What's next for AI agents ft. LangChain's Harrison Chase](https://youtu.be/pBBe1pk8hf4?si=H4vdBF9nmkNZxiHt)
- [`LangFlow`: Build Chatbots without Writing Code - LangChain](https://youtu.be/KJ-ux3hre4s?si=TJuDu4bAlva1myNL)
- [Building a LangChain Custom Medical Agent with Memory](https://youtu.be/6UFtRwWnHws?si=wymYad26VgigRkHy)
As of release 0.2.0, `langchain` is required to be integration-agnostic. This means that code in `langchain` should not by default instantiate any specific chat models, llms, embedding models, vectorstores etc; instead, the user will be required to specify those explicitly.
The following functions and classes require an explicit LLM to be passed as an argument:
The following classes now require passing an explicit Embedding model as an argument:
- `langchain.indexes.VectostoreIndexCreator`
The following code has been removed:
- `langchain.natbot.NatBotChain.from_default` removed in favor of the `from_llm` class method.
### Deprecated
We have two main types of deprecations:
1. Code that was moved from `langchain` into another package (e.g, `langchain-community`)
If you try to import it from `langchain`, the import will keep on working, but will raise a deprecation warning. The warning will provide a replacement import statement.
LangChainDeprecationWarning: Importing UnstructuredMarkdownLoader from langchain.document_loaders is deprecated. Please replace deprecated imports:
>> from langchain.document_loaders import UnstructuredMarkdownLoader
with new imports of:
>> from langchain_community.document_loaders import UnstructuredMarkdownLoader
```
We will continue supporting the imports in `langchain` until release 0.4 as long as the relevant package where the code lives is installed. (e.g., as long as `langchain_community` is installed.)
However, we advise for users to not rely on these imports and instead migrate to the new imports. To help with this process, we’re releasing a migration script via the LangChain CLI. See further instructions in migration guide.
1. Code that has better alternatives available and will eventually be removed, so there’s only a single way to do things. (e.g., `predict_messages` method in ChatModels has been deprecated in favor of `invoke`).
Many of these were marked for removal in 0.2. We have bumped the removal to 0.3.
## 0.1.0 (Jan 5, 2024)
### Deleted
#### Deleted
No deletions.
### Deprecated
#### Deprecated
Deprecated classes and methods will be removed in 0.2.0
This section contains introductions to key parts of LangChain.
## Architecture
LangChain as a framework consists of a number of packages.
### `langchain-core`
This package contains base abstractions of different components and ways to compose them together.
The interfaces for core components like LLMs, vectorstores, retrievers and more are defined here.
No third party integrations are defined here.
The dependencies are kept purposefully very lightweight.
### Partner packages
While the long tail of integrations are in `langchain-community`, we split popular integrations into their own packages (e.g. `langchain-openai`, `langchain-anthropic`, etc).
This was done in order to improve support for these important integrations.
### `langchain`
The main `langchain` package contains chains, agents, and retrieval strategies that make up an application's cognitive architecture.
These are NOT third party integrations.
All chains, agents, and retrieval strategies here are NOT specific to any one integration, but rather generic across all integrations.
### `langchain-community`
This package contains third party integrations that are maintained by the LangChain community.
Key partner packages are separated out (see below).
This contains all integrations for various components (LLMs, vectorstores, retrievers).
All dependencies in this package are optional to keep the package as lightweight as possible.
LangChain Expression Language, or LCEL, is a declarative way to chain LangChain components.
LCEL was designed from day 1 to **support putting prototypes in production, with no code changes**, from the simplest “prompt + LLM” chain to the most complex chains (we’ve seen folks successfully run LCEL chains with 100s of steps in production). To highlight a few of the reasons you might want to use LCEL:
**First-class streaming support**
When you build your chains with LCEL you get the best possible time-to-first-token (time elapsed until the first chunk of output comes out). For some chains this means eg. we stream tokens straight from an LLM to a streaming output parser, and you get back parsed, incremental chunks of output at the same rate as the LLM provider outputs the raw tokens.
**Async support**
Any chain built with LCEL can be called both with the synchronous API (eg. in your Jupyter notebook while prototyping) as well as with the asynchronous API (eg. in a [LangServe](/docs/langserve/) server). This enables using the same code for prototypes and in production, with great performance, and the ability to handle many concurrent requests in the same server.
**Optimized parallel execution**
Whenever your LCEL chains have steps that can be executed in parallel (eg if you fetch documents from multiple retrievers) we automatically do it, both in the sync and the async interfaces, for the smallest possible latency.
**Retries and fallbacks**
Configure retries and fallbacks for any part of your LCEL chain. This is a great way to make your chains more reliable at scale. We’re currently working on adding streaming support for retries/fallbacks, so you can get the added reliability without any latency cost.
**Access intermediate results**
For more complex chains it’s often very useful to access the results of intermediate steps even before the final output is produced. This can be used to let end-users know something is happening, or even just to debug your chain. You can stream intermediate results, and it’s available on every [LangServe](/docs/langserve) server.
**Input and output schemas**
Input and output schemas give every LCEL chain Pydantic and JSONSchema schemas inferred from the structure of your chain. This can be used for validation of inputs and outputs, and is an integral part of LangServe.
Any chain created with LCEL can be easily deployed using [LangServe](/docs/langserve).
### Runnable interface
To make it as easy as possible to create custom chains, we've implemented a ["Runnable"](https://api.python.langchain.com/en/stable/runnables/langchain_core.runnables.base.Runnable.html#langchain_core.runnables.base.Runnable) protocol. Many LangChain components implement the `Runnable` protocol, including chat models, LLMs, output parsers, retrievers, prompt templates, and more. There are also several useful primitives for working with runnables, which you can read about below.
This is a standard interface, which makes it easy to define custom chains as well as invoke them in a standard way.
The standard interface includes:
- [`stream`](#stream): stream back chunks of the response
- [`invoke`](#invoke): call the chain on an input
- [`batch`](#batch): call the chain on a list of inputs
These also have corresponding async methods that should be used with [asyncio](https://docs.python.org/3/library/asyncio.html) `await` syntax for concurrency:
- `astream`: stream back chunks of the response async
- `ainvoke`: call the chain on an input async
- `abatch`: call the chain on a list of inputs async
- `astream_log`: stream back intermediate steps as they happen, in addition to the final response
- `astream_events`: **beta** stream events as they happen in the chain (introduced in `langchain-core` 0.1.14)
The **input type** and **output type** varies by component:
| Component | Input Type | Output Type |
| --- | --- | --- |
| Prompt | Dictionary | PromptValue |
| ChatModel | Single string, list of chat messages or a PromptValue | ChatMessage |
| LLM | Single string, list of chat messages or a PromptValue | String |
| OutputParser | The output of an LLM or ChatModel | Depends on the parser |
| Retriever | Single string | List of Documents |
| Tool | Single string or dictionary, depending on the tool | Depends on the tool |
All runnables expose input and output **schemas** to inspect the inputs and outputs:
- `input_schema`: an input Pydantic model auto-generated from the structure of the Runnable
- `output_schema`: an output Pydantic model auto-generated from the structure of the Runnable
## Components
LangChain provides standard, extendable interfaces and external integrations for various components useful for building with LLMs.
Some components LangChain implements, some components we rely on third-party integrations for, and others are a mix.
### Chat models
Language models that use a sequence of messages as inputs and return chat messages as outputs (as opposed to using plain text).
These are traditionally newer models (older models are generally `LLMs`, see above).
Chat models support the assignment of distinct roles to conversation messages, helping to distinguish messages from the AI, users, and instructions such as system messages.
Although the underlying models are messages in, message out, the LangChain wrappers also allow these models to take a string as input. This means you can easily use chat models in place of LLMs.
When a string is passed in as input, it is converted to a HumanMessage and then passed to the underlying model.
LangChain does not provide any ChatModels, rather we rely on third party integrations.
We have some standardized parameters when constructing ChatModels:
- `model`: the name of the model
ChatModels also accept other parameters that are specific to that integration.
:::important
**Tool Calling** Some chat models have been fine-tuned for tool calling and provide a dedicated API for tool calling.
Generally, such models are better at tool calling than non-fine-tuned models, and are recommended for use cases that require tool calling.
Please see the [tool calling section](/docs/concepts/#functiontool-calling) for more information.
:::
### LLMs
Language models that takes a string as input and returns a string.
These are traditionally older models (newer models generally are `ChatModels`, see below).
Although the underlying models are string in, string out, the LangChain wrappers also allow these models to take messages as input.
This makes them interchangeable with ChatModels.
When messages are passed in as input, they will be formatted into a string under the hood before being passed to the underlying model.
LangChain does not provide any LLMs, rather we rely on third party integrations.
### Messages
Some language models take a list of messages as input and return a message.
There are a few different types of messages.
All messages have a `role`, `content`, and `response_metadata` property.
The `role` describes WHO is saying the message.
LangChain has different message classes for different roles.
The `content` property describes the content of the message.
This can be a few different things:
- A string (most models deal this type of content)
- A List of dictionaries (this is used for multimodal input, where the dictionary contains information about that input type and that input location)
#### HumanMessage
This represents a message from the user.
#### AIMessage
This represents a message from the model. In addition to the `content` property, these messages also have:
**`response_metadata`**
The `response_metadata` property contains additional metadata about the response. The data here is often specific to each model provider.
This is where information like log-probs and token usage may be stored.
**`tool_calls`**
These represent a decision from an language model to call a tool. They are included as part of an `AIMessage` output.
They can be accessed from there with the `.tool_calls` property.
This property returns a list of dictionaries. Each dictionary has the following keys:
- `name`: The name of the tool that should be called.
- `args`: The arguments to that tool.
- `id`: The id of that tool call.
#### SystemMessage
This represents a system message, which tells the model how to behave. Not every model provider supports this.
#### FunctionMessage
This represents the result of a function call. In addition to `role` and `content`, this message has a `name` parameter which conveys the name of the function that was called to produce this result.
#### ToolMessage
This represents the result of a tool call. This is distinct from a FunctionMessage in order to match OpenAI's `function` and `tool` message types. In addition to `role` and `content`, this message has a `tool_call_id` parameter which conveys the id of the call to the tool that was called to produce this result.
### Prompt templates
Prompt templates help to translate user input and parameters into instructions for a language model.
This can be used to guide a model's response, helping it understand the context and generate relevant and coherent language-based output.
Prompt Templates take as input a dictionary, where each key represents a variable in the prompt template to fill in.
Prompt Templates output a PromptValue. This PromptValue can be passed to an LLM or a ChatModel, and can also be cast to a string or a list of messages.
The reason this PromptValue exists is to make it easy to switch between strings and messages.
There are a few different types of prompt templates
#### String PromptTemplates
These prompt templates are used to format a single string, and generally are used for simpler inputs.
For example, a common way to construct and use a PromptTemplate is as follows:
```python
from langchain_core.prompts import PromptTemplate
prompt_template = PromptTemplate.from_template("Tell me a joke about {topic}")
prompt_template.invoke({"topic": "cats"})
```
#### ChatPromptTemplates
These prompt templates are used to format a list of messages. These "templates" consist of a list of templates themselves.
For example, a common way to construct and use a ChatPromptTemplate is as follows:
```python
from langchain_core.prompts import ChatPromptTemplate
("placeholder", "{msgs}") # <-- This is the changed part
])
```
### Example selectors
One common prompting technique for achieving better performance is to include examples as part of the prompt.
This gives the language model concrete examples of how it should behave.
Sometimes these examples are hardcoded into the prompt, but for more advanced situations it may be nice to dynamically select them.
Example Selectors are classes responsible for selecting and then formatting examples into prompts.
### Output parsers
:::note
The information here refers to parsers that take a text output from a model try to parse it into a more structured representation.
More and more models are supporting function (or tool) calling, which handles this automatically.
It is recommended to use function/tool calling rather than output parsing.
See documentation for that [here](/docs/concepts/#function-tool-calling).
:::
Responsible for taking the output of a model and transforming it to a more suitable format for downstream tasks.
Useful when you are using LLMs to generate structured data, or to normalize output from chat models and LLMs.
LangChain has lots of different types of output parsers. This is a list of output parsers LangChain supports. The table below has various pieces of information:
**Name**: The name of the output parser
**Supports Streaming**: Whether the output parser supports streaming.
**Has Format Instructions**: Whether the output parser has format instructions. This is generally available except when (a) the desired schema is not specified in the prompt but rather in other parameters (like OpenAI function calling), or (b) when the OutputParser wraps another OutputParser.
**Calls LLM**: Whether this output parser itself calls an LLM. This is usually only done by output parsers that attempt to correct misformatted output.
**Input Type**: Expected input type. Most output parsers work on both strings and messages, but some (like OpenAI Functions) need a message with specific kwargs.
**Output Type**: The output type of the object returned by the parser.
**Description**: Our commentary on this output parser and when to use it.
| Name | Supports Streaming | Has Format Instructions | Calls LLM | Input Type | Output Type | Description |
| [JSON](https://api.python.langchain.com/en/latest/output_parsers/langchain_core.output_parsers.json.JsonOutputParser.html#langchain_core.output_parsers.json.JsonOutputParser) | ✅ | ✅ | | `str` \| `Message` | JSON object | Returns a JSON object as specified. You can specify a Pydantic model and it will return JSON for that model. Probably the most reliable output parser for getting structured data that does NOT use function calling. |
| [XML](https://api.python.langchain.com/en/latest/output_parsers/langchain_core.output_parsers.xml.XMLOutputParser.html#langchain_core.output_parsers.xml.XMLOutputParser) | ✅ | ✅ | | `str` \| `Message` | `dict` | Returns a dictionary of tags. Use when XML output is needed. Use with models that are good at writing XML (like Anthropic's). |
| [CSV](https://api.python.langchain.com/en/latest/output_parsers/langchain_core.output_parsers.list.CommaSeparatedListOutputParser.html#langchain_core.output_parsers.list.CommaSeparatedListOutputParser) | ✅ | ✅ | | `str` \| `Message` | `List[str]` | Returns a list of comma separated values. |
| [OutputFixing](https://api.python.langchain.com/en/latest/output_parsers/langchain.output_parsers.fix.OutputFixingParser.html#langchain.output_parsers.fix.OutputFixingParser) | | | ✅ | `str` \| `Message` | | Wraps another output parser. If that output parser errors, then this will pass the error message and the bad output to an LLM and ask it to fix the output. |
| [RetryWithError](https://api.python.langchain.com/en/latest/output_parsers/langchain.output_parsers.retry.RetryWithErrorOutputParser.html#langchain.output_parsers.retry.RetryWithErrorOutputParser) | | | ✅ | `str` \| `Message` | | Wraps another output parser. If that output parser errors, then this will pass the original inputs, the bad output, and the error message to an LLM and ask it to fix it. Compared to OutputFixingParser, this one also sends the original instructions. |
| [Pydantic](https://api.python.langchain.com/en/latest/output_parsers/langchain_core.output_parsers.pydantic.PydanticOutputParser.html#langchain_core.output_parsers.pydantic.PydanticOutputParser) | | ✅ | | `str` \| `Message` | `pydantic.BaseModel` | Takes a user defined Pydantic model and returns data in that format. |
| [YAML](https://api.python.langchain.com/en/latest/output_parsers/langchain.output_parsers.yaml.YamlOutputParser.html#langchain.output_parsers.yaml.YamlOutputParser) | | ✅ | | `str` \| `Message` | `pydantic.BaseModel` | Takes a user defined Pydantic model and returns data in that format. Uses YAML to encode it. |
| [Enum](https://api.python.langchain.com/en/latest/output_parsers/langchain.output_parsers.enum.EnumOutputParser.html#langchain.output_parsers.enum.EnumOutputParser) | | ✅ | | `str` \| `Message` | `Enum` | Parses response into one of the provided enum values. |
| [Datetime](https://api.python.langchain.com/en/latest/output_parsers/langchain.output_parsers.datetime.DatetimeOutputParser.html#langchain.output_parsers.datetime.DatetimeOutputParser) | | ✅ | | `str` \| `Message` | `datetime.datetime` | Parses response into a datetime string. |
| [Structured](https://api.python.langchain.com/en/latest/output_parsers/langchain.output_parsers.structured.StructuredOutputParser.html#langchain.output_parsers.structured.StructuredOutputParser) | | ✅ | | `str` \| `Message` | `Dict[str, str]` | An output parser that returns structured information. It is less powerful than other output parsers since it only allows for fields to be strings. This can be useful when you are working with smaller LLMs. |
### Chat history
Most LLM applications have a conversational interface.
An essential component of a conversation is being able to refer to information introduced earlier in the conversation.
At bare minimum, a conversational system should be able to access some window of past messages directly.
The concept of `ChatHistory` refers to a class in LangChain which can be used to wrap an arbitrary chain.
This `ChatHistory` will keep track of inputs and outputs of the underlying chain, and append them as messages to a message database
Future interactions will then load those messages and pass them into the chain as part of the input.
### Documents
A Document object in LangChain contains information about some data. It has two attributes:
- `page_content: str`: The content of this document. Currently is only a string.
- `metadata: dict`: Arbitrary metadata associated with this document. Can track the document id, file name, etc.
### Document loaders
These classes load Document objects. LangChain has hundreds of integrations with various data sources to load data from: Slack, Notion, Google Drive, etc.
Each DocumentLoader has its own specific parameters, but they can all be invoked in the same way with the `.load` method.
An example use case is as follows:
```python
from langchain_community.document_loaders.csv_loader import CSVLoader
loader = CSVLoader(
... # <-- Integration specific parameters here
)
data = loader.load()
```
### Text splitters
Once you've loaded documents, you'll often want to transform them to better suit your application. The simplest example is you may want to split a long document into smaller chunks that can fit into your model's context window. LangChain has a number of built-in document transformers that make it easy to split, combine, filter, and otherwise manipulate documents.
When you want to deal with long pieces of text, it is necessary to split up that text into chunks. As simple as this sounds, there is a lot of potential complexity here. Ideally, you want to keep the semantically related pieces of text together. What "semantically related" means could depend on the type of text. This notebook showcases several ways to do that.
At a high level, text splitters work as following:
1. Split the text up into small, semantically meaningful chunks (often sentences).
2. Start combining these small chunks into a larger chunk until you reach a certain size (as measured by some function).
3. Once you reach that size, make that chunk its own piece of text and then start creating a new chunk of text with some overlap (to keep context between chunks).
That means there are two different axes along which you can customize your text splitter:
1. How the text is split
2. How the chunk size is measured
### Embedding models
The Embeddings class is a class designed for interfacing with text embedding models. There are lots of embedding model providers (OpenAI, Cohere, Hugging Face, etc) - this class is designed to provide a standard interface for all of them.
Embeddings create a vector representation of a piece of text. This is useful because it means we can think about text in the vector space, and do things like semantic search where we look for pieces of text that are most similar in the vector space.
The base Embeddings class in LangChain provides two methods: one for embedding documents and one for embedding a query. The former takes as input multiple texts, while the latter takes a single text. The reason for having these as two separate methods is that some embedding providers have different embedding methods for documents (to be searched over) vs queries (the search query itself).
### Vector stores
One of the most common ways to store and search over unstructured data is to embed it and store the resulting embedding vectors,
and then at query time to embed the unstructured query and retrieve the embedding vectors that are 'most similar' to the embedded query.
A vector store takes care of storing embedded data and performing vector search for you.
Vector stores can be converted to the retriever interface by doing:
```python
vectorstore = MyVectorStore()
retriever = vectorstore.as_retriever()
```
### Retrievers
A retriever is an interface that returns documents given an unstructured query.
It is more general than a vector store.
A retriever does not need to be able to store documents, only to return (or retrieve) them.
Retrievers can be created from vectorstores, but are also broad enough to include [Wikipedia search](/docs/integrations/retrievers/wikipedia/) and [Amazon Kendra](/docs/integrations/retrievers/amazon_kendra_retriever/).
Retrievers accept a string query as input and return a list of Document's as output.
### Tools
Tools are interfaces that an agent, a chain, or a chat model / LLM can use to interact with the world.
A tool consists of the following components:
1. The name of the tool
2. A description of what the tool does
3. JSON schema of what the inputs to the tool are
4. The function to call
5. Whether the result of a tool should be returned directly to the user (only relevant for agents)
The name, description and JSON schema are provided as context
to the LLM, allowing the LLM to determine how to use the tool
appropriately.
Given a list of available tools and a prompt, an LLM can request
that one or more tools be invoked with appropriate arguments.
Generally, when designing tools to be used by a chat model or LLM, it is important to keep in mind the following:
- Chat models that have been fine-tuned for tool calling will be better at tool calling than non-fine-tuned models.
- Non fine-tuned models may not be able to use tools at all, especially if the tools are complex or require multiple tool calls.
- Models will perform better if the tools have well-chosen names, descriptions, and JSON schemas.
- Simpler tools are generally easier for models to use than more complex tools.
### Toolkits
Toolkits are collections of tools that are designed to be used together for specific tasks. They have convenient loading methods.
All Toolkits expose a `get_tools` method which returns a list of tools.
You can therefore do:
```python
# Initialize a toolkit
toolkit = ExampleTookit(...)
# Get list of tools
tools = toolkit.get_tools()
```
### Agents
By themselves, language models can't take actions - they just output text.
A big use case for LangChain is creating **agents**.
Agents are systems that use an LLM as a reasoning enginer to determine which actions to take and what the inputs to those actions should be.
The results of those actions can then be fed back into the agent and it determine whether more actions are needed, or whether it is okay to finish.
[LangGraph](https://github.com/langchain-ai/langgraph) is an extension of LangChain specifically aimed at creating highly controllable and customizable agents.
Please check out that documentation for a more in depth overview of agent concepts.
There is a legacy agent concept in LangChain that we are moving towards deprecating: `AgentExecutor`.
AgentExecutor was essentially a runtime for agents.
It was a great place to get started, however, it was not flexible enough as you started to have more customized agents.
In order to solve that we built LangGraph to be this flexible, highly-controllable runtime.
If you are still using AgentExecutor, do not fear: we still have a guide on [how to use AgentExecutor](/docs/how_to/agent_executor).
It is recommended, however, that you start to transition to LangGraph.
In order to assist in this we have put together a [transition guide on how to do so](/docs/how_to/migrate_agent)
### Multimodal
Some models are multimodal, accepting images, audio and even video as inputs. These are still less common, meaning model providers haven't standardized on the "best" way to define the API. Multimodal **outputs** are even less common. As such, we've kept our multimodal abstractions fairly light weight and plan to further solidify the multimodal APIs and interaction patterns as the field matures.
In LangChain, most chat models that support multimodal inputs also accept those values in OpenAI's content blocks format. So far this is restricted to image inputs. For models like Gemini which support video and other bytes input, the APIs also support the native, model-specific representations.
### Callbacks
LangChain provides a callbacks system that allows you to hook into the various stages of your LLM application. This is useful for logging, monitoring, streaming, and other tasks.
You can subscribe to these events by using the `callbacks` argument available throughout the API. This argument is list of handler objects, which are expected to implement one or more of the methods described below in more detail.
| Chat model start | When a chat model starts | `on_chat_model_start` |
| LLM start | When a llm starts | `on_llm_start` |
| LLM new token | When an llm OR chat model emits a new token | `on_llm_new_token` |
| LLM ends | When an llm OR chat model ends | `on_llm_end` |
| LLM errors | When an llm OR chat model errors | `on_llm_error` |
| Chain start | When a chain starts running | `on_chain_start` |
| Chain end | When a chain ends | `on_chain_end` |
| Chain error | When a chain errors | `on_chain_error` |
| Tool start | When a tool starts running | `on_tool_start` |
| Tool end | When a tool ends | `on_tool_end` |
| Tool error | When a tool errors | `on_tool_error` |
| Agent action | When an agent takes an action | `on_agent_action` |
| Agent finish | When an agent ends | `on_agent_finish` |
| Retriever start | When a retriever starts | `on_retriever_start` |
| Retriever end | When a retriever ends | `on_retriever_end` |
| Retriever error | When a retriever errors | `on_retriever_error` |
| Text | When arbitrary text is run | `on_text` |
| Retry | When a retry event is run | `on_retry` |
#### Callback handlers
Callback handlers can either be `sync` or `async`:
* Sync callback handlers implement the [BaseCallbackHandler](https://api.python.langchain.com/en/latest/callbacks/langchain_core.callbacks.base.BaseCallbackHandler.html) interface.
* Async callback handlers implement the [AsyncCallbackHandler](https://api.python.langchain.com/en/latest/callbacks/langchain_core.callbacks.base.AsyncCallbackHandler.html) interface.
During run-time LangChain configures an appropriate callback manager (e.g., [CallbackManager](https://api.python.langchain.com/en/latest/callbacks/langchain_core.callbacks.manager.CallbackManager.html) or [AsyncCallbackManager](https://api.python.langchain.com/en/latest/callbacks/langchain_core.callbacks.manager.AsyncCallbackManager.html) which will be responsible for calling the appropriate method on each "registered" callback handler when the event is triggered.
#### Passing callbacks
The `callbacks` property is available on most objects throughout the API (Models, Tools, Agents, etc.) in two different places:
The callbacks are available on most objects throughout the API (Models, Tools, Agents, etc.) in two different places:
- **Request time callbacks**: Passed at the time of the request in addition to the input data.
Available on all standard `Runnable` objects. These callbacks are INHERITED by all children
of the object they are defined on. For example, `chain.invoke({"number": 25}, {"callbacks": [handler]})`.
- **Constructor callbacks**: `chain = TheNameOfSomeChain(callbacks=[handler])`. These callbacks
are passed as arguments to the constructor of the object. The callbacks are scoped
only to the object they are defined on, and are **not** inherited by any children of the object.
:::warning
Constructor callbacks are scoped only to the object they are defined on. They are **not** inherited by children
of the object.
:::
If you're creating a custom chain or runnable, you need to remember to propagate request time
callbacks to any child objects.
:::important Async in Python<=3.10
Any `RunnableLambda`, a `RunnableGenerator`, or `Tool` that invokes other runnables
and is running async in python<=3.10, will have to propagate callbacks to child
objects manually. This is because LangChain cannot automatically propagate
callbacks to child objects in this case.
This is a common reason why you may fail to see events being emitted from custom
runnables or tools.
:::
## Techniques
### Function/tool calling
:::info
We use the term tool calling interchangeably with function calling. Although
function calling is sometimes meant to refer to invocations of a single function,
we treat all models as though they can return multiple tool or function calls in
each message.
:::
Tool calling allows a model to respond to a given prompt by generating output that
matches a user-defined schema. While the name implies that the model is performing
some action, this is actually not the case! The model is coming up with the
arguments to a tool, and actually running the tool (or not) is up to the user -
for example, if you want to [extract output matching some schema](/docs/tutorials/extraction)
from unstructured text, you could give the model an "extraction" tool that takes
parameters matching the desired schema, then treat the generated output as your final
result.
A tool call includes a name, arguments dict, and an optional identifier. The
arguments dict is structured `{argument_name: argument_value}`.
Many LLM providers, including [Anthropic](https://www.anthropic.com/),
| [Vectorstore](/docs/how_to/vectorstore_retriever/) | Vectorstore | No | If you are just getting started and looking for something quick and easy. | This is the simplest method and the one that is easiest to get started with. It involves creating embeddings for each piece of text. |
| [ParentDocument](/docs/how_to/parent_document_retriever/) | Vectorstore + Document Store | No | If your pages have lots of smaller pieces of distinct information that are best indexed by themselves, but best retrieved all together. | This involves indexing multiple chunks for each document. Then you find the chunks that are most similar in embedding space, but you retrieve the whole parent document and return that (rather than individual chunks). |
| [Multi Vector](/docs/how_to/multi_vector/) | Vectorstore + Document Store | Sometimes during indexing | If you are able to extract information from documents that you think is more relevant to index than the text itself. | This involves creating multiple vectors for each document. Each vector could be created in a myriad of ways - examples include summaries of the text and hypothetical questions. |
| [Self Query](/docs/how_to/self_query/) | Vectorstore | Yes | If users are asking questions that are better answered by fetching documents based on metadata rather than similarity with the text. | This uses an LLM to transform user input into two things: (1) a string to look up semantically, (2) a metadata filer to go along with it. This is useful because oftentimes questions are about the METADATA of documents (not the content itself). |
| [Contextual Compression](/docs/how_to/contextual_compression/) | Any | Sometimes | If you are finding that your retrieved documents contain too much irrelevant information and are distracting the LLM. | This puts a post-processing step on top of another retriever and extracts only the most relevant information from retrieved documents. This can be done with embeddings or an LLM. |
| [Time-Weighted Vectorstore](/docs/how_to/time_weighted_vectorstore/) | Vectorstore | No | If you have timestamps associated with your documents, and you want to retrieve the most recent ones | This fetches documents based on a combination of semantic similarity (as in normal vector retrieval) and recency (looking at timestamps of indexed documents) |
| [Multi-Query Retriever](/docs/how_to/MultiQueryRetriever/) | Any | Yes | If users are asking questions that are complex and require multiple pieces of distinct information to respond | This uses an LLM to generate multiple queries from the original one. This is useful when the original query needs pieces of information about multiple topics to be properly answered. By generating multiple queries, we can then fetch documents for each of them. |
| [Ensemble](/docs/how_to/ensemble_retriever/) | Any | No | If you have multiple retrieval methods and want to try combining them. | This fetches documents from multiple retrievers and then combines them. |
### Text splitting
LangChain offers many different types of `text splitters`.
These all live in the `langchain-text-splitters` package.
Table columns:
- **Name**: Name of the text splitter
- **Classes**: Classes that implement this text splitter
- **Splits On**: How this text splitter splits text
- **Adds Metadata**: Whether or not this text splitter adds metadata about where each chunk came from.
- **Description**: Description of the splitter, including recommendation on when to use it.
| Name | Classes | Splits On | Adds Metadata | Description |
| Recursive | [RecursiveCharacterTextSplitter](/docs/how_to/recursive_text_splitter/), [RecursiveJsonSplitter](/docs/how_to/recursive_json_splitter/) | A list of user defined characters | | Recursively splits text. This splitting is trying to keep related pieces of text next to each other. This is the `recommended way` to start splitting text. |
| HTML | [HTMLHeaderTextSplitter](/docs/how_to/HTML_header_metadata_splitter/), [HTMLSectionSplitter](/docs/how_to/HTML_section_aware_splitter/) | HTML specific characters | ✅ | Splits text based on HTML-specific characters. Notably, this adds in relevant information about where that chunk came from (based on the HTML) |
| Markdown | [MarkdownHeaderTextSplitter](/docs/how_to/markdown_header_metadata_splitter/), | Markdown specific characters | ✅ | Splits text based on Markdown-specific characters. Notably, this adds in relevant information about where that chunk came from (based on the Markdown) |
| Code | [many languages](/docs/how_to/code_splitter/) | Code (Python, JS) specific characters | | Splits text based on characters specific to coding languages. 15 different languages are available to choose from. |
| Token | [many classes](/docs/how_to/split_by_token/) | Tokens | | Splits text on tokens. There exist a few different ways to measure tokens. |
| Character | [CharacterTextSplitter](/docs/how_to/character_text_splitter/) | A user defined character | | Splits text based on a user defined character. One of the simpler methods. |
| Semantic Chunker (Experimental) | [SemanticChunker](/docs/how_to/semantic-chunker/) | Sentences | | First splits on sentences. Then combines ones next to each other if they are semantically similar enough. Taken from [Greg Kamradt](https://github.com/FullStackRetrieval-com/RetrievalTutorials/blob/main/tutorials/LevelsOfTextSplitting/5_Levels_Of_Text_Splitting.ipynb) |
| Integration: AI21 Semantic | [AI21SemanticTextSplitter](/docs/integrations/document_transformers/ai21_semantic_text_splitter/) | ✅ | Identifies distinct topics that form coherent pieces of text and splits along those. |
@@ -16,15 +16,15 @@ LangChain's documentation aspires to follow the [Diataxis framework](https://dia
Under this framework, all documentation falls under one of four categories:
- **Tutorials**: Lessons that take the reader by the hand through a series of conceptual steps to complete a project.
- An example of this is our [LCEL streaming guide](/docs/how_to/streaming).
- Our guides on [custom components](/docs/how_to/custom_chat_model) is another one.
- An example of this is our [LCEL streaming guide](/docs/expression_language/streaming).
- Our guides on [custom components](/docs/modules/model_io/chat/custom_chat_model) is another one.
- **How-to guides**: Guides that take the reader through the steps required to solve a real-world problem.
- The clearest examples of this are our [Use case](/docs/how_to#use-cases) quickstart pages.
- The clearest examples of this are our [Use case](/docs/use_cases/) quickstart pages.
- **Reference**: Technical descriptions of the machinery and how to operate it.
- Our [Runnable interface](/docs/concepts#interface) page is an example of this.
- Our [Runnable interface](/docs/expression_language/interface) page is an example of this.
- The [API reference pages](https://api.python.langchain.com/) are another.
- **Explanation**: Explanations that clarify and illuminate a particular topic.
- The [LCEL primitives pages](/docs/how_to/sequence) are an example of this.
- The [LCEL primitives pages](/docs/expression_language/primitives/sequence) are an example of this.
Each category serves a distinct purpose and requires a specific approach to writing and structuring the content.
@@ -35,14 +35,14 @@ when contributing new documentation:
### Getting started
The [getting started section](/docs/introduction) includes a high-level introduction to LangChain, a quickstart that
The [getting started section](/docs/get_started/introduction) includes a high-level introduction to LangChain, a quickstart that
tours LangChain's various features, and logistical instructions around installation and project setup.
It contains elements of **How-to guides** and **Explanations**.
### Use cases
[Use cases](/docs/how_to#use-cases) are guides that are meant to show how to use LangChain to accomplish a specific task (RAG, information extraction, etc.).
[Use cases](/docs/use_cases/) are guides that are meant to show how to use LangChain to accomplish a specific task (RAG, information extraction, etc.).
The quickstarts should be good entrypoints for first-time LangChain developers who prefer to learn by getting something practical prototyped,
then taking the pieces apart retrospectively. These should mirror what LangChain is good at.
@@ -55,7 +55,7 @@ The below sections are listed roughly in order of increasing level of abstractio
### Expression Language
[LangChain Expression Language (LCEL)](/docs/concepts#langchain-expression-language) is the fundamental way that most LangChain components fit together, and this section is designed to teach
[LangChain Expression Language (LCEL)](/docs/expression_language/) is the fundamental way that most LangChain components fit together, and this section is designed to teach
developers how to use it to build with LangChain's primitives effectively.
This section should contains **Tutorials** that teach how to stream and use LCEL primitives for more abstract tasks, **Explanations** of specific behaviors,
@@ -63,7 +63,7 @@ and some **References** for how to use different methods in the Runnable interfa
### Components
The [components section](/docs/concepts) covers concepts one level of abstraction higher than LCEL.
The [components section](/docs/modules) covers concepts one level of abstraction higher than LCEL.
Abstract base classes like `BaseChatModel` and `BaseRetriever` should be covered here, as well as core implementations of these base classes,
such as `ChatPromptTemplate` and `RecursiveCharacterTextSplitter`. Customization guides belong here too.
@@ -88,7 +88,7 @@ Concepts covered in `Integrations` should generally exist in `langchain_communit
### Guides and Ecosystem
The [Guides](/docs/tutorials) and [Ecosystem](https://docs.smith.langchain.com/) sections should contain guides that address higher-level problems than the sections above.
The [Guides](/docs/guides) and [Ecosystem](/docs/langsmith/) sections should contain guides that address higher-level problems than the sections above.
This includes, but is not limited to, considerations around productionization and development workflows.
These should contain mostly **How-to guides**, **Explanations**, and **Tutorials**.
@@ -102,7 +102,7 @@ LangChain's API references. Should act as **References** (as the name implies) w
We have set up our docs to assist a new developer to LangChain. Let's walk through the intended path:
- The developer lands on https://python.langchain.com, and reads through the introduction and the diagram.
- If they are just curious, they may be drawn to the [Quickstart](/docs/tutorials/llm_chain) to get a high-level tour of what LangChain contains.
- If they are just curious, they may be drawn to the [Quickstart](/docs/get_started/quickstart) to get a high-level tour of what LangChain contains.
- If they have a specific task in mind that they want to accomplish, they will be drawn to the Use-Case section. The use-case should provide a good, concrete hook that shows the value LangChain can provide them and be a good entrypoint to the framework.
- They can then move to learn more about the fundamentals of LangChain through the Expression Language sections.
- Next, they can learn about LangChain's various components and integrations.
"'El país donde se encuentra la ciudad de Honolulu, donde nació Barack Obama, el 44º Presidente de los Estados Unidos, es Estados Unidos. Honolulu se encuentra en la isla de Oahu, en el estado de Hawái.'"
"ChatPromptValue(messages=[HumanMessage(content='What is the color of strawberry and the flag of China?', additional_kwargs={}, example=False)])"
]
},
"execution_count": 9,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"question_generator.invoke(\"warm\")"
]
},
{
"cell_type": "code",
"execution_count": 10,
"id": "b4a9812b-bead-4fd9-ae27-0b8be57e5dc1",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"AIMessage(content='The color of an apple is typically red or green. The flag of China is predominantly red with a large yellow star in the upper left corner and four smaller yellow stars surrounding it.', additional_kwargs={}, example=False)"
]
},
"execution_count": 10,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"prompt = question_generator.invoke(\"warm\")\n",
"model.invoke(prompt)"
]
},
{
"cell_type": "markdown",
"id": "6d75a313-f1c8-4e94-9a17-24e0bf4a2bdc",
"metadata": {},
"source": [
"### Branching and Merging\n",
"\n",
"You may want the output of one component to be processed by 2 or more other components. [RunnableParallels](https://api.python.langchain.com/en/latest/runnables/langchain_core.runnables.base.RunnableParallel.html#langchain_core.runnables.base.RunnableParallel) let you split or fork the chain so multiple components can process the input in parallel. Later, other components can join or merge the results to synthesize a final response. This type of chain creates a computation graph that looks like the following:\n",
"\n",
"```text\n",
" Input\n",
" / \\\n",
" / \\\n",
" Branch1 Branch2\n",
" \\ /\n",
" \\ /\n",
" Combine\n",
"```"
]
},
{
"cell_type": "code",
"execution_count": 11,
"id": "247fa0bd-4596-4063-8cb3-1d7fc119d982",
"metadata": {},
"outputs": [],
"source": [
"planner = (\n",
" ChatPromptTemplate.from_template(\"Generate an argument about: {input}\")\n",
"'While Scrum has its potential cons and challenges, many organizations have successfully embraced and implemented this project management framework to great effect. The cons mentioned above can be mitigated or overcome with proper training, support, and a commitment to continuous improvement. It is also important to note that not all cons may be applicable to every organization or project.\\n\\nFor example, while Scrum may be complex initially, with proper training and guidance, teams can quickly grasp the concepts and practices. The lack of predictability can be mitigated by implementing techniques such as velocity tracking and release planning. The limited documentation can be addressed by maintaining a balance between lightweight documentation and clear communication among team members. The dependency on team collaboration can be improved through effective communication channels and regular team-building activities.\\n\\nScrum can be scaled and adapted to larger projects by using frameworks like Scrum of Scrums or LeSS (Large Scale Scrum). Concerns about speed versus quality can be addressed by incorporating quality assurance practices, such as continuous integration and automated testing, into the Scrum process. Scope creep can be managed by having a well-defined and prioritized product backlog, and a strong product owner can be developed through training and mentorship.\\n\\nResistance to change can be overcome by providing proper education and communication to stakeholders and involving them in the decision-making process. Ultimately, the cons of Scrum can be seen as opportunities for growth and improvement, and with the right mindset and support, they can be effectively managed.\\n\\nIn conclusion, while Scrum may have its challenges and potential cons, the benefits and advantages it offers in terms of collaboration, flexibility, adaptability, transparency, and customer satisfaction make it a widely adopted and successful project management framework. With proper implementation and continuous improvement, organizations can leverage Scrum to drive innovation, efficiency, and project success.'"
"Almost any other chains you build will use this building block."
]
},
{
"cell_type": "markdown",
"id": "93aa2c87",
"metadata": {},
"source": [
"## PromptTemplate + LLM\n",
"\n",
"The simplest composition is just combining a prompt and model to create a chain that takes user input, adds it to a prompt, passes it to a model, and returns the raw model output.\n",
"\n",
"Note, you can mix and match PromptTemplate/ChatPromptTemplates and LLMs/ChatModels as you like here."
"LCEL makes it easy to build complex chains from basic components, and supports out of the box functionality such as streaming, parallelism, and logging."
]
},
{
"cell_type": "markdown",
"id": "9a9acd2e",
"metadata": {},
"source": [
"## Basic example: prompt + model + output parser\n",
"\n",
"The most basic and common use case is chaining a prompt template and a model together. To see how this works, let's create a chain that takes a topic and generates a joke:"
"prompt = ChatPromptTemplate.from_template(\"tell me a short joke about {topic}\")\n",
"output_parser = StrOutputParser()\n",
"\n",
"chain = prompt | model | output_parser\n",
"\n",
"chain.invoke({\"topic\": \"ice cream\"})"
]
},
{
"cell_type": "markdown",
"id": "81c502c5-85ee-4f36-aaf4-d6e350b7792f",
"metadata": {},
"source": [
"Notice this line of the code, where we piece together these different components into a single chain using LCEL:\n",
"\n",
"```\n",
"chain = prompt | model | output_parser\n",
"```\n",
"\n",
"The `|` symbol is similar to a [unix pipe operator](https://en.wikipedia.org/wiki/Pipeline_(Unix)), which chains together the different components, feeding the output from one component as input into the next component. \n",
"\n",
"In this chain the user input is passed to the prompt template, then the prompt template output is passed to the model, then the model output is passed to the output parser. Let's take a look at each component individually to really understand what's going on."
]
},
{
"cell_type": "markdown",
"id": "aa1b77fa",
"metadata": {},
"source": [
"### 1. Prompt\n",
"\n",
"`prompt` is a `BasePromptTemplate`, which means it takes in a dictionary of template variables and produces a `PromptValue`. A `PromptValue` is a wrapper around a completed prompt that can be passed to either an `LLM` (which takes a string as input) or `ChatModel` (which takes a sequence of messages as input). It can work with either language model type because it defines logic both for producing `BaseMessage`s and for producing a string."
]
},
{
"cell_type": "code",
"execution_count": 2,
"id": "b8656990",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"ChatPromptValue(messages=[HumanMessage(content='tell me a short joke about ice cream')])"
"And lastly we pass our `model` output to the `output_parser`, which is a `BaseOutputParser` meaning it takes either a string or a \n",
"`BaseMessage` as input. The specific `StrOutputParser` simply converts any input into a string."
]
},
{
"cell_type": "code",
"execution_count": 13,
"id": "533e59a8",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"\"Why did the ice cream go to therapy? \\n\\nBecause it had too many toppings and couldn't find its cone-fidence!\""
]
},
"execution_count": 13,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"output_parser.invoke(message)"
]
},
{
"cell_type": "markdown",
"id": "9851e842",
"metadata": {},
"source": [
"### 4. Entire Pipeline\n",
"\n",
"To follow the steps along:\n",
"\n",
"1. We pass in user input on the desired topic as `{\"topic\": \"ice cream\"}`\n",
"2. The `prompt` component takes the user input, which is then used to construct a PromptValue after using the `topic` to construct the prompt. \n",
"3. The `model` component takes the generated prompt, and passes into the OpenAI LLM model for evaluation. The generated output from the model is a `ChatMessage` object. \n",
"4. Finally, the `output_parser` component takes in a `ChatMessage`, and transforms this into a Python string, which is returned from the invoke method. \n"
"Note that if you’re curious about the output of any components, you can always test out a smaller version of the chain such as `prompt` or `prompt | model` to see the intermediate results:\n",
"\n",
":::"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "11089b6f-23f8-474f-97ec-8cae8d0ca6d4",
"metadata": {},
"outputs": [],
"source": [
"input = {\"topic\": \"ice cream\"}\n",
"\n",
"prompt.invoke(input)\n",
"# > ChatPromptValue(messages=[HumanMessage(content='tell me a short joke about ice cream')])\n",
"\n",
"(prompt | model).invoke(input)\n",
"# > AIMessage(content=\"Why did the ice cream go to therapy?\\nBecause it had too many toppings and couldn't cone-trol itself!\")"
]
},
{
"cell_type": "markdown",
"id": "cc7d3b9d-e400-4c9b-9188-f29dac73e6bb",
"metadata": {},
"source": [
"## RAG Search Example\n",
"\n",
"For our next example, we want to run a retrieval-augmented generation chain to add some context when responding to questions."
"chain = setup_and_retrieval | prompt | model | output_parser\n",
"\n",
"chain.invoke(\"where did harrison work?\")"
]
},
{
"cell_type": "markdown",
"id": "f0999140-6001-423b-970b-adf1dfdb4dec",
"metadata": {},
"source": [
"In this case, the composed chain is: "
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "5b88e9bb-f04a-4a56-87ec-19a0e6350763",
"metadata": {},
"outputs": [],
"source": [
"chain = setup_and_retrieval | prompt | model | output_parser"
]
},
{
"cell_type": "markdown",
"id": "6e929e15-40a5-4569-8969-384f636cab87",
"metadata": {},
"source": [
"To explain this, we first can see that the prompt template above takes in `context` and `question` as values to be substituted in the prompt. Before building the prompt template, we want to retrieve relevant documents to the search and include them as part of the context. \n",
"\n",
"As a preliminary step, we’ve setup the retriever using an in memory store, which can retrieve documents based on a query. This is a runnable component as well that can be chained together with other components, but you can also try to run it separately:"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "a7319ef6-613b-4638-ad7d-4a2183702c1d",
"metadata": {},
"outputs": [],
"source": [
"retriever.invoke(\"where did harrison work?\")"
]
},
{
"cell_type": "markdown",
"id": "e6833844-f1c4-444c-a3d2-31b3c6b31d46",
"metadata": {},
"source": [
"We then use the `RunnableParallel` to prepare the expected inputs into the prompt by using the entries for the retrieved documents as well as the original user question, using the retriever for document search, and `RunnablePassthrough` to pass the user’s question:"
"chain = setup_and_retrieval | prompt | model | output_parser"
]
},
{
"cell_type": "markdown",
"id": "5c6f5f74-b387-48a0-bedd-1fae202cd10a",
"metadata": {},
"source": [
"With the flow being:\n",
"\n",
"1. The first steps create a `RunnableParallel` object with two entries. The first entry, `context` will include the document results fetched by the retriever. The second entry, `question` will contain the user’s original question. To pass on the question, we use `RunnablePassthrough` to copy this entry. \n",
"2. Feed the dictionary from the step above to the `prompt` component. It then takes the user input which is `question` as well as the retrieved document which is `context` to construct a prompt and output a PromptValue. \n",
"3. The `model` component takes the generated prompt, and passes into the OpenAI LLM model for evaluation. The generated output from the model is a `ChatMessage` object. \n",
"4. Finally, the `output_parser` component takes in a `ChatMessage`, and transforms this into a Python string, which is returned from the invoke method.\n",
"\n",
"```mermaid\n",
"graph LR\n",
" A(Question) --> B(RunnableParallel)\n",
" B -->|Question| C(Retriever)\n",
" B -->|Question| D(RunnablePassThrough)\n",
" C -->|context=retrieved docs| E(PromptTemplate)\n",
" D -->|question=Question| E\n",
" E -->|PromptValue| F(ChatModel) \n",
" F -->|ChatMessage| G(StrOutputParser)\n",
" G --> |String| H(Result)\n",
"```\n",
"\n"
]
},
{
"cell_type": "markdown",
"id": "8c2438df-164e-4bbe-b5f4-461695e45b0f",
"metadata": {},
"source": [
"## Next steps\n",
"\n",
"We recommend reading our [Advantages of LCEL](/docs/expression_language/why) section next to see a side-by-side comparison of the code needed to produce common functionality with and without LCEL."
"# Create a runnable with the @chain decorator\n",
"\n",
"You can also turn an arbitrary function into a chain by adding a `@chain` decorator. This is functionaly equivalent to wrapping in a [`RunnableLambda`](/docs/expression_language/primitives/functions).\n",
"\n",
"This will have the benefit of improved observability by tracing your chain correctly. Any calls to runnables inside this function will be traced as nested childen.\n",
"\n",
"It will also allow you to use this as any other runnable, compose it in chain, etc.\n",
"Once you create a runnable with LCEL, you may often want to inspect it to get a better sense for what is going on. This notebook covers some methods for doing so.\n",
"\n",
"This guide assumes familiarity with the following concepts:\n",
"- [LangChain Expression Language (LCEL)](/docs/concepts/#langchain-expression-language)\n",
"Once you create a runnable with [LangChain Expression Language](/docs/concepts/#langchain-expression-language), you may often want to inspect it to get a better sense for what is going on. This notebook covers some methods for doing so.\n",
"\n",
"This guide shows some ways you can programmatically introspect the internal steps of chains. If you are instead interested in debugging issues in your chain, see [this section](/docs/how_to/debugging) instead.\n",
"\n",
"First, let's create an example chain. We will create one that does retrieval:"
"First, let's create an example LCEL. We will create one that does retrieval"
"You can use the `get_graph()` method to get a graph representation of the runnable:"
"You can get a graph of the runnable"
]
},
{
@@ -94,7 +100,7 @@
"source": [
"## Print a graph\n",
"\n",
"While that is not super legible, you can use the `print_ascii()` method to show that graph in a way that's easier to understand:"
"While that is not super legible, you can print it to get a display that's easier to understand"
]
},
{
@@ -160,7 +166,7 @@
"source": [
"## Get the prompts\n",
"\n",
"You may want to see just the prompts that are used in a chain with the `get_prompts()` method:"
"An important part of every chain is the prompts that are used. You can get the prompts present in the chain:"
]
},
{
@@ -184,18 +190,6 @@
"chain.get_prompts()"
]
},
{
"cell_type": "markdown",
"id": "c5a74bd5",
"metadata": {},
"source": [
"## Next steps\n",
"\n",
"You've now learned how to introspect your composed LCEL chains.\n",
"\n",
"Next, check out the other how-to guides on runnables in this section, or the related how-to guide on [debugging your chains](/docs/how_to/debugging)."
"The `RunnableWithMessageHistory` lets us add message history to certain types of chains. It wraps another Runnable and manages the chat message history for it.\n",
"\n",
"Specifically, it can be used for any Runnable that takes as input one of\n",
"\n",
"* a sequence of `BaseMessage`\n",
"* a dict with a key that takes a sequence of `BaseMessage`\n",
"* a dict with a key that takes the latest message(s) as a string or sequence of `BaseMessage`, and a separate key that takes historical messages\n",
"\n",
"And returns as output one of\n",
"\n",
"* a string that can be treated as the contents of an `AIMessage`\n",
"* a sequence of `BaseMessage`\n",
"* a dict with a key that contains a sequence of `BaseMessage`\n",
"\n",
"Let's take a look at some examples to see how it works. First we construct a runnable (which here accepts a dict as input and returns a message as output):"
"2. A callable that returns an instance of `BaseChatMessageHistory`.\n",
"\n",
"Check out the [memory integrations](https://integrations.langchain.com/memory) page for implementations of chat message histories using Redis and other providers. Here we demonstrate using an in-memory `ChatMessageHistory` as well as more persistent storage using `RedisChatMessageHistory`."
]
},
{
"cell_type": "markdown",
"id": "3d83adad-9672-496d-9f25-5747e7b8c8bb",
"metadata": {},
"source": [
"## In-memory\n",
"\n",
"Below we show a simple example in which the chat history lives in memory, in this case via a global Python dict.\n",
"\n",
"We construct a callable `get_session_history` that references this dict to return an instance of `ChatMessageHistory`. The arguments to the callable can be specified by passing a configuration to the `RunnableWithMessageHistory` at runtime. By default, the configuration parameter is expected to be a single string `session_id`. This can be adjusted via the `history_factory_config` kwarg.\n",
"Note that we've specified `input_messages_key` (the key to be treated as the latest input message) and `history_messages_key` (the key to add historical messages to).\n",
"\n",
"When invoking this new runnable, we specify the corresponding chat history via a configuration parameter:"
]
},
{
"cell_type": "code",
"execution_count": 3,
"id": "01384412-f08e-4634-9edb-3f46f475b582",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"AIMessage(content='Cosine is a trigonometric function that calculates the ratio of the adjacent side to the hypotenuse of a right triangle.')"
]
},
"execution_count": 3,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"with_message_history.invoke(\n",
" {\"ability\": \"math\", \"input\": \"What does cosine mean?\"},\n",
"The configuration parameters by which we track message histories can be customized by passing in a list of ``ConfigurableFieldSpec`` objects to the ``history_factory_config`` parameter. Below, we use two parameters: a `user_id` and `conversation_id`."
"### Examples with runnables of different signatures\n",
"\n",
"The above runnable takes a dict as input and returns a BaseMessage. Below we show some alternatives."
]
},
{
"cell_type": "markdown",
"id": "48eae1bf-b59d-4a61-8e62-b6dbf667e866",
"metadata": {},
"source": [
"#### Messages input, dict output"
]
},
{
"cell_type": "code",
"execution_count": 7,
"id": "17733d4f-3a32-4055-9d44-5d58b9446a26",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"{'output_message': AIMessage(content=\"Simone de Beauvoir believed in the existence of free will. She argued that individuals have the ability to make choices and determine their own actions, even in the face of social and cultural constraints. She rejected the idea that individuals are purely products of their environment or predetermined by biology or destiny. Instead, she emphasized the importance of personal responsibility and the need for individuals to actively engage in creating their own lives and defining their own existence. De Beauvoir believed that freedom and agency come from recognizing one's own freedom and actively exercising it in the pursuit of personal and collective liberation.\")}"
"{'output_message': AIMessage(content='Simone de Beauvoir\\'s views on free will were closely aligned with those of her contemporary and partner Jean-Paul Sartre. Both de Beauvoir and Sartre were existentialist philosophers who emphasized the importance of individual freedom and the rejection of determinism. They believed that human beings have the capacity to transcend their circumstances and create their own meaning and values.\\n\\nSartre, in his famous work \"Being and Nothingness,\" argued that human beings are condemned to be free, meaning that we are burdened with the responsibility of making choices and defining ourselves in a world that lacks inherent meaning. Like de Beauvoir, Sartre believed that individuals have the ability to exercise their freedom and make choices in the face of external and internal constraints.\\n\\nWhile there may be some nuanced differences in their philosophical writings, overall, de Beauvoir and Sartre shared a similar belief in the existence of free will and the importance of individual agency in shaping one\\'s own life.')}"
]
},
"execution_count": 8,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"with_message_history.invoke(\n",
" [HumanMessage(content=\"How did this compare to Sartre\")],\n",
"In many cases it is preferable to persist conversation histories. `RunnableWithMessageHistory` is agnostic as to how the `get_session_history` callable retrieves its chat message histories. See [here](https://github.com/langchain-ai/langserve/blob/main/examples/chat_with_persistence_and_user/server.py) for an example using a local filesystem. Below we demonstrate how one could use Redis. Check out the [memory integrations](https://integrations.langchain.com/memory) page for implementations of chat message histories using other providers."
]
},
{
"cell_type": "markdown",
"id": "6bca45e5-35d9-4603-9ca9-6ac0ce0e35cd",
"metadata": {},
"source": [
"### Setup\n",
"\n",
"We'll need to install Redis if it's not installed already:"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "477d04b3-c2b6-4ba5-962f-492c0d625cd5",
"metadata": {},
"outputs": [],
"source": [
"%pip install --upgrade --quiet redis"
]
},
{
"cell_type": "markdown",
"id": "6a0ec9e0-7b1c-4c6f-b570-e61d520b47c6",
"metadata": {},
"source": [
"Start a local Redis Stack server if we don't have an existing Redis deployment to connect to:\n",
"```bash\n",
"docker run -d -p 6379:6379 -p 8001:8001 redis/redis-stack:latest\n",
"```"
]
},
{
"cell_type": "code",
"execution_count": 9,
"id": "cd6a250e-17fe-4368-a39d-1fe6b2cbde68",
"metadata": {},
"outputs": [],
"source": [
"REDIS_URL = \"redis://localhost:6379/0\""
]
},
{
"cell_type": "markdown",
"id": "36f43b87-655c-4f64-aa7b-bd8c1955d8e5",
"metadata": {},
"source": [
"### [LangSmith](/docs/langsmith)\n",
"\n",
"LangSmith is especially useful for something like message history injection, where it can be hard to otherwise understand what the inputs are to various parts of the chain.\n",
"\n",
"Note that LangSmith is not needed, but it is helpful.\n",
"If you do want to use LangSmith, after you sign up at the link above, make sure to uncoment the below and set your environment variables to start logging traces:"
"AIMessage(content='The inverse of cosine is the arccosine function, denoted as acos or cos^-1, which gives the angle corresponding to a given cosine value.')"
]
},
"execution_count": 12,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"with_message_history.invoke(\n",
" {\"ability\": \"math\", \"input\": \"What's its inverse\"},\n",
"Looking at the Langsmith trace for the second call, we can see that when constructing the prompt, a \"history\" variable has been injected which is a list of two messages (our first input and first output)."
"Routing allows you to create non-deterministic chains where the output of a previous step defines the next step. Routing can help provide structure and consistency around interactions with models by allowing you to define states and use information related to those states as context to model calls.\n",
"Routing allows you to create non-deterministic chains where the output of a previous step defines the next step. Routing helps provide structure and consistency around interactions with LLMs.\n",
"\n",
"There are two ways to perform routing:\n",
"\n",
"1. Conditionally return runnables from a [`RunnableLambda`](/docs/how_to/functions) (recommended)\n",
"2. Using a `RunnableBranch` (legacy)\n",
"1. Conditionally return runnables from a [`RunnableLambda`](/docs/expression_language/primitives/functions) (recommended)\n",
"2. Using a `RunnableBranch`.\n",
"\n",
"We'll illustrate both methods using a two step sequence where the first step classifies an input question as being about `LangChain`, `Anthropic`, or `Other`, then routes to a corresponding prompt chain."
LangChain Expression Language, or LCEL, is a declarative way to easily compose chains together.
LCEL was designed from day 1 to **support putting prototypes in production, with no code changes**, from the simplest “prompt + LLM” chain to the most complex chains (we’ve seen folks successfully run LCEL chains with 100s of steps in production). To highlight a few of the reasons you might want to use LCEL:
When you build your chains with LCEL you get the best possible time-to-first-token (time elapsed until the first chunk of output comes out). For some chains this means eg. we stream tokens straight from an LLM to a streaming output parser, and you get back parsed, incremental chunks of output at the same rate as the LLM provider outputs the raw tokens.
Any chain built with LCEL can be called both with the synchronous API (eg. in your Jupyter notebook while prototyping) as well as with the asynchronous API (eg. in a [LangServe](/docs/langsmith) server). This enables using the same code for prototypes and in production, with great performance, and the ability to handle many concurrent requests in the same server.
Whenever your LCEL chains have steps that can be executed in parallel (eg if you fetch documents from multiple retrievers) we automatically do it, both in the sync and the async interfaces, for the smallest possible latency.
[**Retries and fallbacks**](/docs/guides/productionization/fallbacks)
Configure retries and fallbacks for any part of your LCEL chain. This is a great way to make your chains more reliable at scale. We’re currently working on adding streaming support for retries/fallbacks, so you can get the added reliability without any latency cost.
For more complex chains it’s often very useful to access the results of intermediate steps even before the final output is produced. This can be used to let end-users know something is happening, or even just to debug your chain. You can stream intermediate results, and it’s available on every [LangServe](/docs/langserve) server.
[**Input and output schemas**](/docs/expression_language/interface#input-schema)
Input and output schemas give every LCEL chain Pydantic and JSONSchema schemas inferred from the structure of your chain. This can be used for validation of inputs and outputs, and is an integral part of LangServe.
[**Seamless LangSmith tracing**](/docs/langsmith)
As your chains get more and more complex, it becomes increasingly important to understand what exactly is happening at every step.
With LCEL, **all** steps are automatically logged to [LangSmith](/docs/langsmith/) for maximum observability and debuggability.
"- [Passing data through](/docs/how_to/passthrough)\n",
"\n",
":::\n",
"\n",
"An alternate way of [passing data through](/docs/how_to/passthrough) steps of a chain is to leave the current values of the chain state unchanged while assigning a new value under a given key. The [`RunnablePassthrough.assign()`](https://api.python.langchain.com/en/latest/runnables/langchain_core.runnables.passthrough.RunnablePassthrough.html#langchain_core.runnables.passthrough.RunnablePassthrough.assign) static method takes an input value and adds the extra arguments passed to the assign function.\n",
"\n",
"This is useful in the common [LangChain Expression Language](/docs/concepts/#langchain-expression-language) pattern of additively creating a dictionary to use as input to a later step.\n",
"This is useful when additively creating a dictionary to use as input to a later step, which is a common LCEL pattern.\n",
"\n",
"Here's an example:"
]
},
{
"cell_type": "code",
"execution_count": null,
"execution_count": 1,
"metadata": {},
"outputs": [],
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"\u001b[33mWARNING: You are using pip version 22.0.4; however, version 24.0 is available.\n",
"You should consider upgrading via the '/Users/jacoblee/.pyenv/versions/3.10.5/bin/python -m pip install --upgrade pip' command.\u001b[0m\u001b[33m\n",
"\u001b[0mNote: you may need to restart the kernel to use updated packages.\n"
"One convenient feature of this method is that it allows values to pass through as soon as they are available. To show this off, we'll use `RunnablePassthrough.assign()` to immediately return source docs in a retrieval chain:"
"One nice feature of this method is that it allows values to pass through as soon as they are available. To show this off, we'll use `RunnablePassthrough.assign()` to immediately return source docs in a retrieval chain:"
]
},
{
"cell_type": "code",
"execution_count": 3,
"execution_count": 4,
"metadata": {},
"outputs": [
{
@@ -152,13 +147,7 @@
"cell_type": "markdown",
"metadata": {},
"source": [
"We can see that the first chunk contains the original `\"question\"` since that is immediately available. The second chunk contains `\"context\"` since the retriever finishes second. Finally, the output from the `generation_chain` streams in chunks as soon as it is available.\n",
"\n",
"## Next steps\n",
"\n",
"Now you've learned how to pass data through your chains to help to help format the data flowing through your chains.\n",
"\n",
"To learn more, see the other how-to guides on runnables in this section."
"We can see that the first chunk contains the original `\"question\"` since that is immediately available. The second chunk contains `\"context\"` since the retriever finishes second. Finally, the output from the `generation_chain` streams in chunks as soon as it is available."
"# How to add default invocation args to a Runnable\n",
"# Binding: Attach runtime args\n",
"\n",
":::info Prerequisites\n",
"Sometimes we want to invoke a Runnable within a Runnable sequence with constant arguments that are not part of the output of the preceding Runnable in the sequence, and which are not part of the user input. We can use `Runnable.bind()` to pass these arguments in.\n",
"\n",
"This guide assumes familiarity with the following concepts:\n",
"- [LangChain Expression Language (LCEL)](/docs/concepts/#langchain-expression-language)\n",
"Sometimes we want to invoke a [`Runnable`](https://api.python.langchain.com/en/latest/runnables/langchain_core.runnables.base.Runnable.html) within a [RunnableSequence](https://api.python.langchain.com/en/latest/runnables/langchain_core.runnables.base.RunnableSequence.html) with constant arguments that are not part of the output of the preceding Runnable in the sequence, and which are not part of the user input. We can use the [`Runnable.bind()`](https://api.python.langchain.com/en/latest/runnables/langchain_core.runnables.base.Runnable.html#langchain_core.runnables.base.Runnable.bind) method to set these arguments ahead of time.\n",
"\n",
"## Binding stop sequences\n",
"\n",
"Suppose we have a simple prompt + model chain:"
"Suppose we have a simple prompt + model sequence:"
" {\"equation_statement\": RunnablePassthrough()} | prompt | model | StrOutputParser()\n",
")\n",
@@ -103,12 +94,12 @@
"id": "929c9aba-a4a0-462c-adac-2cfc2156e117",
"metadata": {},
"source": [
"and want to call the model with certain `stop` words so that we shorten the output as is useful in certain types of prompting techniques. While we can pass some arguments into the constructor, other runtime args use the `.bind()` method as follows:"
"and want to call the model with certain `stop` words:"
]
},
{
"cell_type": "code",
"execution_count": 3,
"execution_count": 12,
"id": "32e0484a-78c5-4570-a00b-20d597245a96",
"metadata": {},
"outputs": [
@@ -129,25 +120,92 @@
" | model.bind(stop=\"SOLUTION\")\n",
" | StrOutputParser()\n",
")\n",
"\n",
"print(runnable.invoke(\"x raised to the third plus seven equals 12\"))"
]
},
{
"cell_type": "markdown",
"id": "f4bd641f-6b58-4ca9-a544-f69095428f16",
"metadata": {},
"source": [
"## Attaching OpenAI functions\n",
"\n",
"One particularly useful application of binding is to attach OpenAI functions to a compatible OpenAI model:"
]
},
{
"cell_type": "code",
"execution_count": 3,
"id": "f66a0fe4-fde0-4706-8863-d60253f211c7",
"metadata": {},
"outputs": [],
"source": [
"function = {\n",
" \"name\": \"solver\",\n",
" \"description\": \"Formulates and solves an equation\",\n",
" \"parameters\": {\n",
" \"type\": \"object\",\n",
" \"properties\": {\n",
" \"equation\": {\n",
" \"type\": \"string\",\n",
" \"description\": \"The algebraic expression of the equation\",\n",
" },\n",
" \"solution\": {\n",
" \"type\": \"string\",\n",
" \"description\": \"The solution to the equation\",\n",
"runnable.invoke(\"x raised to the third plus seven equals 12\")"
]
},
{
"cell_type": "markdown",
"id": "f07d7528-9269-4d6f-b12e-3669592a9e03",
"metadata": {},
"source": [
"What you can bind to a Runnable will depend on the extra parameters you can pass when invoking it.\n",
"\n",
"## Attaching OpenAI tools\n",
"\n",
"Another common use-case is tool calling. While you should generally use the [`.bind_tools()`](/docs/how_to/tool_calling/) method for tool-calling models, you can also bind provider-specific args directly if you want lower level control:"
"model.invoke(\"What's the weather in SF, NYC and LA?\")"
]
},
{
"cell_type": "markdown",
"id": "095001f7",
"metadata": {},
"source": [
"## Next steps\n",
"\n",
"You now know how to bind runtime arguments to a Runnable.\n",
"\n",
"To learn more, see the other how-to guides on runnables in this section, including:\n",
"\n",
"- [Using configurable fields and alternatives](/docs/how_to/configure) to change parameters of a step in a chain, or even swap out entire steps, at runtime"
"Sometimes you may want to experiment with, or even expose to the end user, multiple different ways of doing things within your chains.\n",
"This can include tweaking parameters such as temperature or even swapping out one model for another.\n",
"Oftentimes you may want to experiment with, or even expose to the end user, multiple different ways of doing things.\n",
"In order to make this experience as easy as possible, we have defined two methods.\n",
"\n",
"- A `configurable_fields` method. This lets you configure particular fields of a runnable.\n",
" - This is related to the [`.bind`](/docs/how_to/binding) method on runnables, but allows you to specify parameters for a given step in a chain at runtime rather than specifying them beforehand.\n",
"- A `configurable_alternatives` method. With this method, you can list out alternatives for any particular runnable that can be set during runtime, and swap them for those specified alternatives."
"First, a `configurable_fields` method. \n",
"This lets you configure particular fields of a runnable.\n",
"\n",
"Second, a `configurable_alternatives` method.\n",
"With this method, you can list out alternatives for any particular runnable that can be set during runtime."
]
},
{
@@ -41,53 +34,34 @@
"id": "f2347a11",
"metadata": {},
"source": [
"## Configurable Fields\n",
"\n",
"Let's walk through an example that configures chat model fields like temperature at runtime:"
"## Configuration Fields"
]
},
{
"cell_type": "markdown",
"id": "a06f6e2d",
"metadata": {},
"source": [
"### With LLMs\n",
"With LLMs we can configure things like temperature"
]
},
{
"cell_type": "code",
"execution_count": 1,
"execution_count": null,
"id": "40ed76a2",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"\u001b[33mWARNING: You are using pip version 22.0.4; however, version 24.0 is available.\n",
"You should consider upgrading via the '/Users/jacoblee/.pyenv/versions/3.10.5/bin/python -m pip install --upgrade pip' command.\u001b[0m\u001b[33m\n",
"\u001b[0mNote: you may need to restart the kernel to use updated packages.\n"
"Above, we defined `temperature` as a [`ConfigurableField`](https://api.python.langchain.com/en/latest/runnables/langchain_core.runnables.utils.ConfigurableField.html#langchain_core.runnables.utils.ConfigurableField) that we can set at runtime. To do so, we use the [`with_config`](https://api.python.langchain.com/en/latest/runnables/langchain_core.runnables.base.Runnable.html#langchain_core.runnables.base.Runnable.with_config) method like this:"
" description=\"The Hub commit to pull from\",\n",
" )\n",
")"
]
},
{
"cell_type": "code",
"execution_count": 47,
"id": "c4a62cee",
"metadata": {},
"outputs": [
{
"data": {
@@ -210,28 +227,18 @@
"ChatPromptValue(messages=[HumanMessage(content=\"You are an assistant for question-answering tasks. Use the following pieces of retrieved context to answer the question. If you don't know the answer, just say that you don't know. Use three sentences maximum and keep the answer concise.\\nQuestion: foo \\nContext: bar \\nAnswer:\")])"
"ChatPromptValue(messages=[HumanMessage(content=\"[INST]<<SYS>> You are an assistant for question-answering tasks. Use the following pieces of retrieved context to answer the question. If you don't know the answer, just say that you don't know. Use three sentences maximum and keep the answer concise.<</SYS>> \\nQuestion: foo \\nContext: bar \\nAnswer: [/INST]\")])"
]
},
"execution_count": 7,
"execution_count": 49,
"metadata": {},
"output_type": "execute_result"
}
@@ -266,32 +273,22 @@
"id": "ac733d35",
"metadata": {},
"source": [
"The `configurable_alternatives()` method allows us to swap out steps in a chain with an alternative. Below, we swap out one chat model for another:"
"### With LLMs\n",
"\n",
"Let's take a look at doing this with LLMs"
]
},
{
"cell_type": "code",
"execution_count": 8,
"id": "3db59f45",
"execution_count": 4,
"id": "430ab8cc",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"\u001b[33mWARNING: You are using pip version 22.0.4; however, version 24.0 is available.\n",
"You should consider upgrading via the '/Users/jacoblee/.pyenv/versions/3.10.5/bin/python -m pip install --upgrade pip' command.\u001b[0m\u001b[33m\n",
"\u001b[0mNote: you may need to restart the kernel to use updated packages.\n"
"AIMessage(content=\"Here's a bear joke for you:\\n\\nWhy don't bears wear socks? \\nBecause they have bear feet!\\n\\nHow's that? I tried to come up with a simple, silly pun-based joke about bears. Puns and wordplay are a common way to create humorous bear jokes. Let me know if you'd like to hear another one!\", response_metadata={'id': 'msg_018edUHh5fUbWdiimhrC3dZD', 'model': 'claude-3-haiku-20240307', 'stop_reason': 'end_turn', 'stop_sequence': None, 'usage': {'input_tokens': 13, 'output_tokens': 80}}, id='run-775bc58c-28d7-4e6b-a268-48fa6661f02f-0')"
" # When configuring the end runnable, we can then use this id to configure this field\n",
" ConfigurableField(id=\"llm\"),\n",
@@ -333,25 +312,44 @@
" # You can add more configuration options here\n",
")\n",
"prompt = PromptTemplate.from_template(\"Tell me a joke about {topic}\")\n",
"chain = prompt | llm\n",
"\n",
"chain = prompt | llm"
]
},
{
"cell_type": "code",
"execution_count": 19,
"id": "e598b1f1",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"AIMessage(content=\" Here's a silly joke about bears:\\n\\nWhat do you call a bear with no teeth?\\nA gummy bear!\")"
]
},
"execution_count": 19,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# By default it will call Anthropic\n",
"chain.invoke({\"topic\": \"bears\"})"
]
},
{
"cell_type": "code",
"execution_count": 19,
"execution_count": 20,
"id": "48b45337",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"AIMessage(content=\"Why don't bears like fast food?\\n\\nBecause they can't catch it!\", response_metadata={'token_usage': {'completion_tokens': 15, 'prompt_tokens': 13, 'total_tokens': 28}, 'model_name': 'gpt-3.5-turbo', 'system_fingerprint': 'fp_c2295e73ad', 'finish_reason': 'stop', 'logprobs': None}, id='run-7bdaa992-19c9-4f0d-9a0c-1f326bc992d4-0')"
"AIMessage(content=\"Sure, here's a bear joke for you:\\n\\nWhy don't bears wear shoes?\\n\\nBecause they already have bear feet!\")"
]
},
"execution_count": 19,
"execution_count": 20,
"metadata": {},
"output_type": "execute_result"
}
@@ -363,17 +361,17 @@
},
{
"cell_type": "code",
"execution_count": 20,
"execution_count": 21,
"id": "42647fb7",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"AIMessage(content=\"Here's a bear joke for you:\\n\\nWhy don't bears wear socks? \\nBecause they have bear feet!\\n\\nHow's that? I tried to come up with a simple, silly pun-based joke about bears. Puns and wordplay are a common way to create humorous bear jokes. Let me know if you'd like to hear another one!\", response_metadata={'id': 'msg_01BZvbmnEPGBtcxRWETCHkct', 'model': 'claude-3-haiku-20240307', 'stop_reason': 'end_turn', 'stop_sequence': None, 'usage': {'input_tokens': 13, 'output_tokens': 80}}, id='run-59b6ee44-a1cd-41b8-a026-28ee67cdd718-0')"
"AIMessage(content=\"Here's a silly joke about bears:\\n\\nWhat do you call a bear with no teeth?\\nA gummy bear!\")"
]
},
"execution_count": 20,
"execution_count": 21,
"metadata": {},
"output_type": "execute_result"
}
@@ -395,23 +393,12 @@
},
{
"cell_type": "code",
"execution_count": 22,
"execution_count": 25,
"id": "9f6a7c6c",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"AIMessage(content=\"Here's a bear joke for you:\\n\\nWhy don't bears wear socks? \\nBecause they have bear feet!\", response_metadata={'id': 'msg_01DtM1cssjNFZYgeS3gMZ49H', 'model': 'claude-3-haiku-20240307', 'stop_reason': 'end_turn', 'stop_sequence': None, 'usage': {'input_tokens': 13, 'output_tokens': 28}}, id='run-8199af7d-ea31-443d-b064-483693f2e0a1-0')"
" poem=PromptTemplate.from_template(\"Write a short poem about {topic}\"),\n",
" # You can add more configuration options here\n",
")\n",
"chain = prompt | llm\n",
"\n",
"chain = prompt | llm"
]
},
{
"cell_type": "code",
"execution_count": 26,
"id": "97eda915",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"AIMessage(content=\" Here's a silly joke about bears:\\n\\nWhat do you call a bear with no teeth?\\nA gummy bear!\")"
]
},
"execution_count": 26,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# By default it will write a joke\n",
"chain.invoke({\"topic\": \"bears\"})"
]
},
{
"cell_type": "code",
"execution_count": 23,
"execution_count": 27,
"id": "927297a1",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"AIMessage(content=\"Here is a short poem about bears:\\n\\nMajestic bears, strong and true,\\nRoaming the forests, wild and free.\\nPowerful paws, fur soft and brown,\\nCommanding respect, nature's crown.\\n\\nForaging for berries, fishing streams,\\nProtecting their young, fierce and keen.\\nMighty bears, a sight to behold,\\nGuardians of the wilderness, untold.\\n\\nIn the wild they reign supreme,\\nEmbodying nature's grand theme.\\nBears, a symbol of strength and grace,\\nCaptivating all who see their face.\", response_metadata={'id': 'msg_01Wck3qPxrjURtutvtodaJFn', 'model': 'claude-3-haiku-20240307', 'stop_reason': 'end_turn', 'stop_sequence': None, 'usage': {'input_tokens': 13, 'output_tokens': 134}}, id='run-69414a1e-51d7-4bec-a307-b34b7d61025e-0')"
"AIMessage(content=' Here is a short poem about bears:\\n\\nThe bears awaken from their sleep\\nAnd lumber out into the deep\\nForests filled with trees so tall\\nForaging for food before nightfall \\nTheir furry coats and claws so sharp\\nSniffing for berries and fish to nab\\nLumbering about without a care\\nThe mighty grizzly and black bear\\nProud creatures, wild and free\\nRulingtheir domain majestically\\nWandering the woods they call their own\\nBefore returning to their dens alone')"
]
},
"execution_count": 23,
"execution_count": 27,
"metadata": {},
"output_type": "execute_result"
}
@@ -466,25 +472,12 @@
},
{
"cell_type": "code",
"execution_count": 25,
"execution_count": 28,
"id": "97538c23",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"AIMessage(content=\"In the forest deep and wide,\\nBears roam with grace and pride.\\nWith fur as dark as night,\\nThey rule the land with all their might.\\n\\nIn winter's chill, they hibernate,\\nIn spring they emerge, hungry and great.\\nWith claws sharp and eyes so keen,\\nThey hunt for food, fierce and lean.\\n\\nBut beneath their tough exterior,\\nLies a gentle heart, warm and superior.\\nThey love their cubs with all their might,\\nProtecting them through day and night.\\n\\nSo let us admire these majestic creatures,\\nIn awe of their strength and features.\\nFor in the wild, they reign supreme,\\nThe mighty bears, a timeless dream.\", response_metadata={'token_usage': {'completion_tokens': 133, 'prompt_tokens': 13, 'total_tokens': 146}, 'model_name': 'gpt-3.5-turbo', 'system_fingerprint': 'fp_c2295e73ad', 'finish_reason': 'stop', 'logprobs': None}, id='run-5eec0b96-d580-49fd-ac4e-e32a0803b49b-0')"
" # When configuring the end runnable, we can then use this id to configure this field\n",
" ConfigurableField(id=\"llm\"),\n",
@@ -510,8 +503,27 @@
" poem=PromptTemplate.from_template(\"Write a short poem about {topic}\"),\n",
" # You can add more configuration options here\n",
")\n",
"chain = prompt | llm\n",
"\n",
"chain = prompt | llm"
]
},
{
"cell_type": "code",
"execution_count": 29,
"id": "1dcc7ccc",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"AIMessage(content=\"In the forest, where tall trees sway,\\nA creature roams, both fierce and gray.\\nWith mighty paws and piercing eyes,\\nThe bear, a symbol of strength, defies.\\n\\nThrough snow-kissed mountains, it does roam,\\nA guardian of its woodland home.\\nWith fur so thick, a shield of might,\\nIt braves the coldest winter night.\\n\\nA gentle giant, yet wild and free,\\nThe bear commands respect, you see.\\nWith every step, it leaves a trace,\\nOf untamed power and ancient grace.\\n\\nFrom honeyed feast to salmon's leap,\\nIt takes its place, in nature's keep.\\nA symbol of untamed delight,\\nThe bear, a wonder, day and night.\\n\\nSo let us honor this noble beast,\\nIn forests where its soul finds peace.\\nFor in its presence, we come to know,\\nThe untamed spirit that in us also flows.\")"
]
},
"execution_count": 29,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# We can configure it write a poem with OpenAI\n",
"AIMessage(content=\"Why did the bear break up with his girlfriend? \\nBecause he couldn't bear the relationship anymore!\", response_metadata={'token_usage': {'completion_tokens': 20, 'prompt_tokens': 13, 'total_tokens': 33}, 'model_name': 'gpt-3.5-turbo', 'system_fingerprint': 'fp_c2295e73ad', 'finish_reason': 'stop', 'logprobs': None}, id='run-391ebd55-9137-458b-9a11-97acaff6a892-0')"
"AIMessage(content=\"Why don't bears wear shoes?\\n\\nBecause they have bear feet!\")"
"You can use arbitrary functions as [Runnables](https://api.python.langchain.com/en/latest/runnables/langchain_core.runnables.base.Runnable.html#langchain_core.runnables.base.Runnable). This is useful for formatting or when you need functionality not provided by other LangChain components, and custom functions used as Runnables are called [`RunnableLambdas`](https://api.python.langchain.com/en/latest/runnables/langchain_core.runnables.base.RunnableLambda.html).\n",
"\n",
"Note that all inputs to these functions need to be a SINGLE argument. If you have a function that accepts multiple arguments, you should write a wrapper that accepts a single dict input and unpacks it into multiple argument.\n",
"\n",
"This guide will cover:\n",
"\n",
"- How to explicitly create a runnable from a custom function using the `RunnableLambda` constructor and the convenience `@chain` decorator\n",
"- Coercion of custom functions into runnables when used in chains\n",
"- How to accept and use run metadata in your custom function\n",
"- How to stream with custom functions by having them return generators\n",
"\n",
"## Using the constructor\n",
"\n",
"Below, we explicitly wrap our custom logic using the `RunnableLambda` constructor:"
"Note that all inputs to these functions need to be a SINGLE argument. If you have a function that accepts multiple arguments, you should write a wrapper that accepts a single input and unpacks it into multiple argument."
"You can also turn an arbitrary function into a chain by adding a `@chain` decorator. This is functionaly equivalent to wrapping the function in a `RunnableLambda` constructor as shown above. Here's an example:"
")"
]
},
{
"cell_type": "code",
"execution_count": 3,
"id": "3142a516",
"execution_count": 2,
"id": "5488ec85",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"'The subject of the joke is the bear and his girlfriend.'"
"Above, the `@chain` decorator is used to convert `custom_chain` into a runnable, which we invoke with the `.invoke()` method.\n",
"## Accepting a Runnable Config\n",
"\n",
"If you are using a tracing with [LangSmith](https://docs.smith.langchain.com/), you should see a `custom_chain` trace in there, with the calls to OpenAI nested underneath.\n",
"\n",
"## Automatic coercion in chains\n",
"\n",
"When using custom functions in chains with the pipe operator (`|`), you can omit the `RunnableLambda` or `@chain` constructor and rely on coercion. Here's a simple example with a function that takes the output from the model and returns the first five letters of it:"
"Runnable lambdas can optionally accept a [RunnableConfig](https://api.python.langchain.com/en/latest/runnables/langchain_core.runnables.config.RunnableConfig.html#langchain_core.runnables.config.RunnableConfig), which they can use to pass callbacks, tags, and other configuration information to nested runs."
"Note that we didn't need to wrap the custom function `(lambda x: x.content[:5])` in a `RunnableLambda` constructor because the `model` on the left of the pipe operator is already a Runnable. The custom function is **coerced** into a runnable. See [this section](/docs/how_to/sequence/#coercion) for more information.\n",
"\n",
"## Passing run metadata\n",
"\n",
"Runnable lambdas can optionally accept a [RunnableConfig](https://api.python.langchain.com/en/latest/runnables/langchain_core.runnables.config.RunnableConfig.html#langchain_core.runnables.config.RunnableConfig) parameter, which they can use to pass callbacks, tags, and other configuration information to nested runs."
"You can use generator functions (ie. functions that use the `yield` keyword, and behave like iterators) in a chain.\n",
"You can use generator functions (ie. functions that use the `yield` keyword, and behave like iterators) in a LCEL pipeline.\n",
"\n",
"The signature of these generators should be `Iterator[Input] -> Iterator[Output]`. Or for async generators: `AsyncIterator[Input] -> AsyncIterator[Output]`.\n",
"\n",
@@ -310,13 +188,30 @@
"- implementing a custom output parser\n",
"- modifying the output of a previous step, while preserving streaming capabilities\n",
"\n",
"Here's an example of a custom output parser for comma-separated lists. First, we create a chain that generates such a list as text:"
"Here's an example of a custom output parser for comma-separated lists:"
]
},
{
"cell_type": "code",
"execution_count": 6,
"id": "29f55c38",
"metadata": {},
"outputs": [],
"source": [
"from typing import Iterator, List\n",
"\n",
"prompt = ChatPromptTemplate.from_template(\n",
" \"Write a comma-separated list of 5 animals similar to: {animal}. Do not include numbers\"\n",
")\n",
"model = ChatOpenAI(temperature=0.0)\n",
"\n",
"str_chain = prompt | model | StrOutputParser()"
]
},
{
"cell_type": "code",
"execution_count": 7,
"id": "29f55c38",
"id": "75aa946b",
"metadata": {},
"outputs": [
{
@@ -328,44 +223,37 @@
}
],
"source": [
"from typing import Iterator, List\n",
"\n",
"prompt = ChatPromptTemplate.from_template(\n",
" \"Write a comma-separated list of 5 animals similar to: {animal}. Do not include numbers\"\n",
")\n",
"\n",
"str_chain = prompt | model | StrOutputParser()\n",
"\n",
"for chunk in str_chain.stream({\"animal\": \"bear\"}):\n",
" print(chunk, end=\"\", flush=True)"
]
},
{
"cell_type": "markdown",
"id": "46345323",
"cell_type": "code",
"execution_count": 8,
"id": "d002a7fe",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"'lion, tiger, wolf, gorilla, panda'"
]
},
"execution_count": 8,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"Next, we define a custom function that will aggregate the currently streamed output and yield it when the model generates the next comma in the list:"
"str_chain.invoke({\"animal\": \"bear\"})"
]
},
{
"cell_type": "code",
"execution_count": 8,
"execution_count": 9,
"id": "f08b8a5b",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"['lion']\n",
"['tiger']\n",
"['wolf']\n",
"['gorilla']\n",
"['raccoon']\n"
]
}
],
"outputs": [],
"source": [
"# This is a custom parser that splits an iterator of llm tokens\n",
"# into a list of strings separated by commas\n",
@@ -384,58 +272,23 @@
" # save the rest for the next iteration\n",
" buffer = buffer[comma_index + 1 :]\n",
" # yield the last chunk\n",
" yield [buffer.strip()]\n",
"\n",
"\n",
"list_chain = str_chain | split_into_list\n",
"\n",
"for chunk in list_chain.stream({\"animal\": \"bear\"}):\n",
" print(chunk, flush=True)"
]
},
{
"cell_type": "markdown",
"id": "0a5adb69",
"metadata": {},
"source": [
"Invoking it gives a full array of values:"
]
},
{
"cell_type": "code",
"execution_count": 9,
"id": "9ea4ddc6",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"['lion', 'tiger', 'wolf', 'gorilla', 'raccoon']"
]
},
"execution_count": 9,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"list_chain.invoke({\"animal\": \"bear\"})"
]
},
{
"cell_type": "markdown",
"id": "96e320ed",
"metadata": {},
"source": [
"## Async version\n",
"\n",
"If you are working in an `async` environment, here is an `async` version of the above example:"
" yield [buffer.strip()]"
]
},
{
"cell_type": "code",
"execution_count": 10,
"id": "569dbbef",
"id": "02e414aa",
"metadata": {},
"outputs": [],
"source": [
"list_chain = str_chain | split_into_list"
]
},
{
"cell_type": "code",
"execution_count": 11,
"id": "7ed8799d",
"metadata": {},
"outputs": [
{
@@ -450,6 +303,46 @@
]
}
],
"source": [
"for chunk in list_chain.stream({\"animal\": \"bear\"}):\n",
"The `RunnableParallel` primitive is essentially a dict whose values are runnables (or things that can be coerced to runnables, like functions). It runs all of its values in parallel, and each value is called with the overall input of the `RunnableParallel`. The final return value is a dict with the results of each value under its appropriate key.\n",
"\n",
"This guide assumes familiarity with the following concepts:\n",
"- [LangChain Expression Language (LCEL)](/docs/concepts/#langchain-expression-language)\n",
"It is useful for parallelizing operations, but can also be useful for manipulating the output of one Runnable to match the input format of the next Runnable in a sequence.\n",
"\n",
":::\n",
"\n",
"The [`RunnableParallel`](https://api.python.langchain.com/en/latest/runnables/langchain_core.runnables.base.RunnableParallel.html) primitive is essentially a dict whose values are runnables (or things that can be coerced to runnables, like functions). It runs all of its values in parallel, and each value is called with the overall input of the `RunnableParallel`. The final return value is a dict with the results of each value under its appropriate key.\n",
"\n",
"## Formatting with `RunnableParallels`\n",
"\n",
"`RunnableParallels` are useful for parallelizing operations, but can also be useful for manipulating the output of one Runnable to match the input format of the next Runnable in a sequence. You can use them to split or fork the chain so that multiple components can process the input in parallel. Later, other components can join or merge the results to synthesize a final response. This type of chain creates a computation graph that looks like the following:\n",
"\n",
"```text\n",
" Input\n",
" / \\\n",
" / \\\n",
" Branch1 Branch2\n",
" \\ /\n",
" \\ /\n",
" Combine\n",
"```\n",
"\n",
"Below, the input to prompt is expected to be a map with keys `\"context\"` and `\"question\"`. The user input is just the question. So we need to get the context using our retriever and passthrough the user input under the `\"question\"` key.\n"
"Here the input to prompt is expected to be a map with keys \"context\" and \"question\". The user input is just the question. So we need to get the context using our retriever and passthrough the user input under the \"question\" key.\n"
"See the section on [coercion for more](/docs/how_to/sequence/#coercion)."
"\n"
]
},
{
@@ -150,7 +119,7 @@
},
{
"cell_type": "code",
"execution_count": 3,
"execution_count": 6,
"id": "84fc49e1-2daf-4700-ae33-a0a6ed47d5f6",
"metadata": {},
"outputs": [
@@ -160,7 +129,7 @@
"'Harrison ha lavorato a Kensho.'"
]
},
"execution_count": 3,
"execution_count": 6,
"metadata": {},
"output_type": "execute_result"
}
@@ -209,23 +178,23 @@
"source": [
"## Parallelize steps\n",
"\n",
"RunnableParallels make it easy to execute multiple Runnables in parallel, and to return the output of these Runnables as a map."
"RunnableParallel (aka. RunnableMap) makes it easy to execute multiple Runnables in parallel, and to return the output of these Runnables as a map."
]
},
{
"cell_type": "code",
"execution_count": 4,
"execution_count": 1,
"id": "31f18442-f837-463f-bef4-8729368f5f8b",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"{'joke': AIMessage(content=\"Why don't bears like fast food? Because they can't catch it!\", response_metadata={'token_usage': {'completion_tokens': 15, 'prompt_tokens': 13, 'total_tokens': 28}, 'model_name': 'gpt-3.5-turbo', 'system_fingerprint': 'fp_d9767fc5b9', 'finish_reason': 'stop', 'logprobs': None}, id='run-fe024170-c251-4b7a-bfd4-64a3737c67f2-0'),\n",
" 'poem': AIMessage(content='In the quiet of the forest, the bear roams free\\nMajestic and wild, a sight to see.', response_metadata={'token_usage': {'completion_tokens': 24, 'prompt_tokens': 15, 'total_tokens': 39}, 'model_name': 'gpt-3.5-turbo', 'system_fingerprint': 'fp_c2295e73ad', 'finish_reason': 'stop', 'logprobs': None}, id='run-2707913e-a743-4101-b6ec-840df4568a76-0')}"
"{'joke': AIMessage(content=\"Why don't bears wear shoes?\\n\\nBecause they have bear feet!\"),\n",
" 'poem': AIMessage(content=\"In the wild's embrace, bear roams free,\\nStrength and grace, a majestic decree.\")}"
]
},
"execution_count": 4,
"execution_count": 1,
"metadata": {},
"output_type": "execute_result"
}
@@ -258,7 +227,7 @@
},
{
"cell_type": "code",
"execution_count": 5,
"execution_count": 6,
"id": "38e47834-45af-4281-991f-86f150001510",
"metadata": {},
"outputs": [
@@ -266,7 +235,7 @@
"name": "stdout",
"output_type": "stream",
"text": [
"610 ms ± 64 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
"958 ms ± 402 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
]
}
],
@@ -278,7 +247,7 @@
},
{
"cell_type": "code",
"execution_count": 6,
"execution_count": 7,
"id": "d0cd40de-b37e-41fa-a2f6-8aaa49f368d6",
"metadata": {},
"outputs": [
@@ -286,7 +255,7 @@
"name": "stdout",
"output_type": "stream",
"text": [
"599 ms ± 73.3 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
"1.22 s ± 508 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
]
}
],
@@ -298,7 +267,7 @@
},
{
"cell_type": "code",
"execution_count": 7,
"execution_count": 8,
"id": "799894e1-8e18-4a73-b466-f6aea6af3920",
"metadata": {},
"outputs": [
@@ -306,7 +275,7 @@
"name": "stdout",
"output_type": "stream",
"text": [
"643 ms ± 77.8 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
"1.15 s ± 119 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
]
}
],
@@ -315,26 +284,6 @@
"\n",
"map_chain.invoke({\"topic\": \"bear\"})"
]
},
{
"cell_type": "markdown",
"id": "7d4492e1",
"metadata": {},
"source": [
"## Next steps\n",
"\n",
"You now know some ways to format and parallelize chain steps with `RunnableParallel`.\n",
"\n",
"To learn more, see the other how-to guides on runnables in this section."
"When composing chains with several steps, sometimes you will want to pass data from previous steps unchanged for use as input to a later step. The [`RunnablePassthrough`](https://api.python.langchain.com/en/latest/runnables/langchain_core.runnables.passthrough.RunnablePassthrough.html) class allows you to do just this, and is typically is used in conjuction with a [RunnableParallel](/docs/how_to/parallel/) to pass data through to a later step in your constructed chains.\n",
"RunnablePassthrough on its own allows you to pass inputs unchanged. This typically is used in conjuction with RunnableParallel to pass data through to a new key in the map. \n",
"In the example below, we see a more real-world use case where we use `RunnablePassthrough` along with `RunnableParallel` in a chain to properly format inputs to a prompt:"
"In the example below, we see a use case where we use `RunnablePassthrough` along with `RunnableParallel`. "
]
},
{
"cell_type": "code",
"execution_count": 3,
"execution_count": 17,
"id": "267d1460-53c1-4fdb-b2c3-b6a1eb7fccff",
"metadata": {},
"outputs": [
@@ -109,7 +94,7 @@
"'Harrison worked at Kensho.'"
]
},
"execution_count": 3,
"execution_count": 17,
"metadata": {},
"output_type": "execute_result"
}
@@ -148,13 +133,7 @@
"id": "392cd4c4-e7ed-4ab8-934d-f7a4eca55ee1",
"metadata": {},
"source": [
"Here the input to prompt is expected to be a map with keys \"context\" and \"question\". The user input is just the question. So we need to get the context using our retriever and passthrough the user input under the \"question\" key. The `RunnablePassthrough` allows us to pass on the user's question to the prompt and model. \n",
"\n",
"## Next steps\n",
"\n",
"Now you've learned how to pass data through your chains to help to help format the data flowing through your chains.\n",
"\n",
"To learn more, see the other how-to guides on runnables in this section."
"Here the input to prompt is expected to be a map with keys \"context\" and \"question\". The user input is just the question. So we need to get the context using our retriever and passthrough the user input under the \"question\" key. In this case, the RunnablePassthrough allows us to pass on the user's question to the prompt and model. \n"
"One key advantage of the `Runnable` interface is that any two runnables can be \"chained\" together into sequences. The output of the previous runnable's `.invoke()` call is passed as input to the next runnable. This can be done using the pipe operator (`|`), or the more explicit `.pipe()` method, which does the same thing. The resulting `RunnableSequence` isitself a runnable, which means it can be invoked, streamed, or piped just like any other runnable.\n",
"\n",
"This guide assumes familiarity with the following concepts:\n",
"- [LangChain Expression Language (LCEL)](/docs/concepts/#langchain-expression-language)\n",
"One point about [LangChain Expression Language](/docs/concepts/#langchain-expression-language) is that any two runnables can be \"chained\" together into sequences. The output of the previous runnable's `.invoke()` call is passed as input to the next runnable. This can be done using the pipe operator (`|`), or the more explicit `.pipe()` method, which does the same thing.\n",
"\n",
"The resulting [`RunnableSequence`](https://api.python.langchain.com/en/latest/runnables/langchain_core.runnables.base.RunnableSequence.html) is itself a runnable, which means it can be invoked, streamed, or further chained just like any other runnable. Advantages of chaining runnables in this way are efficient streaming (the sequence will stream output as soon as it is available), and debugging and tracing with tools like [LangSmith](/docs/how_to/debugging).\n",
"\n",
"## The pipe operator: `|`\n",
"\n",
"To show off how this works, let's go through an example. We'll walk through a common pattern in LangChain: using a [prompt template](/docs/how_to#prompt-templates) to format input into a [chat model](/docs/how_to#chat-models), and finally converting the chat message output into a string with an [output parser](/docs/how_to#output-parsers).\n",
"\n",
"```{=mdx}\n",
"import ChatModelTabs from \"@theme/ChatModelTabs\";\n",
"\n",
"<ChatModelTabs\n",
" customVarName=\"model\"\n",
"/>\n",
"```"
"To show off how this works, let's go through an example. We'll walk through a common pattern in LangChain: using a [prompt template](/docs/modules/model_io/prompts/) to format input into a [chat model](/docs/modules/model_io/chat/), and finally converting the chat message output into a string with an [output parser](/docs/modules/model_io/output_parsers/)."
"\"Here's a bear joke for you:\\n\\nWhy did the bear dissolve in water?\\nBecause it was a polar bear!\""
"\"Here's a bear joke for you:\\n\\nWhy don't bears wear socks? \\nBecause they have bear feet!\\n\\nHow's that? I tried to keep it light and silly. Bears can make for some fun puns and jokes. Let me know if you'd like to hear another one!\""
]
},
"execution_count": 3,
@@ -118,7 +86,7 @@
"\n",
"For example, let's say we wanted to compose the joke generating chain with another chain that evaluates whether or not the generated joke was funny.\n",
"\n",
"We would need to be careful with how we format the input into the next chain. In the below example, the dict in the chain is automatically parsed and converted into a [`RunnableParallel`](/docs/how_to/parallel), which runs all of its values in parallel and returns a dict with the results.\n",
"We would need to be careful with how we format the input into the next chain. In the below example, the dict in the chain is automatically parsed and converted into a [`RunnableParallel`](/docs/expression_language/primitives/parallel), which runs all of its values in parallel and returns a dict with the results.\n",
"\n",
"This happens to be the same format the next prompt template expects. Here it is in action:"
]
@@ -127,25 +95,32 @@
"cell_type": "code",
"execution_count": 4,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"'Haha, that\\'s a clever play on words! Using \"polar\" to imply the bear dissolved or became polar/polarized when put in water. Not the most hilarious joke ever, but it has a cute, groan-worthy pun that makes it mildly amusing. I appreciate a good pun or wordplay joke.'"
"analysis_prompt = ChatPromptTemplate.from_template(\"is this a funny joke? {joke}\")\n",
"\n",
"composed_chain = {\"joke\": chain} | analysis_prompt | model | StrOutputParser()\n",
"\n",
"composed_chain = {\"joke\": chain} | analysis_prompt | model | StrOutputParser()"
]
},
{
"cell_type": "code",
"execution_count": 5,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"\"That's a pretty classic and well-known bear pun joke. Whether it's considered funny is quite subjective, as humor is very personal. Some people may find that type of pun-based joke amusing, while others may not find it that humorous. Ultimately, the funniness of a joke is in the eye (or ear) of the beholder. If you enjoyed the joke and got a chuckle out of it, then that's what matters most.\""
]
},
"execution_count": 5,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"composed_chain.invoke({\"topic\": \"bears\"})"
]
},
@@ -158,20 +133,9 @@
},
{
"cell_type": "code",
"execution_count": 5,
"execution_count": 6,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"\"Haha, that's a cute and punny joke! I like how it plays on the idea of beets blushing or turning red like someone blushing. Food puns can be quite amusing. While not a total knee-slapper, it's a light-hearted, groan-worthy dad joke that would make me chuckle and shake my head. Simple vegetable humor!\""
]
},
"execution_count": 5,
"metadata": {},
"output_type": "execute_result"
}
],
"outputs": [],
"source": [
"composed_chain_with_lambda = (\n",
" chain\n",
@@ -179,8 +143,26 @@
" | analysis_prompt\n",
" | model\n",
" | StrOutputParser()\n",
")\n",
"\n",
")"
]
},
{
"cell_type": "code",
"execution_count": 7,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"'I appreciate the effort, but I have to be honest - I didn\\'t find that joke particularly funny. Beet-themed puns can be quite hit-or-miss, and this one falls more on the \"miss\" side for me. The premise is a bit too straightforward and predictable. While I can see the logic behind it, the punchline just doesn\\'t pack much of a comedic punch. \\n\\nThat said, I do admire your willingness to explore puns and wordplay around vegetables. Cultivating a good sense of humor takes practice, and not every joke is going to land. The important thing is to keep experimenting and finding what works. Maybe try for a more unexpected or creative twist on beet-related humor next time. But thanks for sharing - I always appreciate when humans test out jokes on me, even if they don\\'t always make me laugh out loud.'"
"However, keep in mind that using functions like this may interfere with operations like streaming. See [this section](/docs/how_to/functions) for more information."
"However, keep in mind that using functions like this may interfere with operations like streaming. See [this section](/docs/expression_language/primitives/functions) for more information."
]
},
{
@@ -202,20 +184,9 @@
},
{
"cell_type": "code",
"execution_count": 6,
"execution_count": 8,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"\"I cannot reproduce any copyrighted material verbatim, but I can try to analyze the humor in the joke you provided without quoting it directly.\\n\\nThe joke plays on the idea that the Cylon raiders, who are the antagonists in the Battlestar Galactica universe, failed to locate the human survivors after attacking their home planets (the Twelve Colonies) due to using an outdated and poorly performing operating system (Windows Vista) for their targeting systems.\\n\\nThe humor stems from the juxtaposition of a futuristic science fiction setting with a relatable real-world frustration – the use of buggy, slow, or unreliable software or technology. It pokes fun at the perceived inadequacies of Windows Vista, which was widely criticized for its performance issues and other problems when it was released.\\n\\nBy attributing the Cylons' failure to locate the humans to their use of Vista, the joke creates an amusing and unexpected connection between a fictional advanced race of robots and a familiar technological annoyance experienced by many people in the real world.\\n\\nOverall, the joke relies on incongruity and relatability to generate humor, but without reproducing any copyrighted material directly.\""
"'That\\'s a pretty good Battlestar Galactica-themed pun! I appreciated the clever play on words with \"Centurion\" and \"center on.\" It\\'s the kind of nerdy, science fiction-inspired humor that fans of the show would likely enjoy. The joke is clever and demonstrates a good understanding of the Battlestar Galactica universe. I\\'d be curious to hear any other Battlestar-related jokes you might have up your sleeve. As long as they don\\'t reproduce copyrighted material, I\\'m happy to provide my thoughts on the humor and appeal for fans of the show.'"
]
},
"execution_count": 9,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"## Related\n",
"\n",
"- [Streaming](/docs/how_to/streaming/): Check out the streaming guide to understand the streaming behavior of a chain\n"
The `langchain-community` package contains third-party integrations. Install with:
The `langchain-community` package contains third-party integrations. It is automatically installed by `langchain`, but can also be used separately. Install with:
**LangChain** is a framework for developing applications powered by large language models (LLMs).
LangChain simplifies every stage of the LLM application lifecycle:
- **Development**: Build your applications using LangChain's open-source [building blocks](/docs/expression_language/) and [components](/docs/modules/). Hit the ground running using [third-party integrations](/docs/integrations/platforms/) and [Templates](/docs/templates).
- **Productionization**: Use [LangSmith](/docs/langsmith/) to inspect, monitor and evaluate your chains, so that you can continuously optimize and deploy with confidence.
- **Deployment**: Turn any chain into an API with [LangServe](/docs/langserve).
import ThemedImage from '@theme/ThemedImage';
<ThemedImage
alt="Diagram outlining the hierarchical organization of the LangChain framework, displaying the interconnected parts across multiple layers."
sources={{
light: '/svg/langchain_stack.svg',
dark: '/svg/langchain_stack_dark.svg',
}}
title="LangChain Framework Overview"
/>
Concretely, the framework consists of the following open-source libraries:
- **`langchain-core`**: Base abstractions and LangChain Expression Language.
- **`langchain-community`**: Third party integrations.
- Partner packages (e.g. **`langchain-openai`**, **`langchain-anthropic`**, etc.): Some integrations have been further split into their own lightweight packages that only depend on **`langchain-core`**.
- **`langchain`**: Chains, agents, and retrieval strategies that make up an application's cognitive architecture.
- **[langgraph](/docs/langgraph)**: Build robust and stateful multi-actor applications with LLMs by modeling steps as edges and nodes in a graph.
- **[langserve](/docs/langserve)**: Deploy LangChain chains as REST APIs.
The broader ecosystem includes:
- **[LangSmith](/docs/langsmith)**: A developer platform that lets you debug, test, evaluate, and monitor LLM applications and seamlessly integrates with LangChain.
## Get started
We recommend following our [Quickstart](/docs/get_started/quickstart) guide to familiarize yourself with the framework by building your first LangChain application.
[See here](/docs/get_started/installation) for instructions on how to install LangChain, set up your environment, and start building.
:::note
These docs focus on the Python LangChain library. [Head here](https://js.langchain.com) for docs on the JavaScript LangChain library.
:::
## Use cases
If you're looking to build something specific or are more of a hands-on learner, check out our [use-cases](/docs/use_cases).
They're walkthroughs and techniques for common end-to-end tasks, such as:
- [Question answering with RAG](/docs/use_cases/question_answering/)
LangChain Expression Language (LCEL) is the foundation of many of LangChain's components, and is a declarative way to compose chains. LCEL was designed from day 1 to support putting prototypes in production, with no code changes, from the simplest “prompt + LLM” chain to the most complex chains.
- **[Get started](/docs/expression_language/)**: LCEL and its benefits
- **[Runnable interface](/docs/expression_language/interface)**: The standard interface for LCEL objects
- **[Primitives](/docs/expression_language/primitives)**: More on the primitives LCEL includes
- and more!
## Ecosystem
### [🦜🛠️ LangSmith](/docs/langsmith)
Trace and evaluate your language model applications and intelligent agents to help you move from prototype to production.
### [🦜🕸️ LangGraph](/docs/langgraph)
Build stateful, multi-actor applications with LLMs, built on top of (and intended to be used with) LangChain primitives.
### [🦜🏓 LangServe](/docs/langserve)
Deploy LangChain runnables and chains as REST APIs.
## [Security](/docs/security)
Read up on our [Security](/docs/security) best practices to make sure you're developing safely with LangChain.
## Additional resources
### [Components](/docs/modules/)
LangChain provides standard, extendable interfaces and integrations for many different components, including:
### [Integrations](/docs/integrations/providers/)
LangChain is part of a rich ecosystem of tools that integrate with our framework and build on top of it. Check out our growing list of [integrations](/docs/integrations/providers/).
- Get setup with LangChain, LangSmith and LangServe
- Use the most basic and common components of LangChain: prompt templates, models, and output parsers
- Use LangChain Expression Language, the protocol that LangChain is built on and which facilitates component chaining
- Build a simple application with LangChain
- Trace your application with LangSmith
- Serve your application with LangServe
That's a fair amount to cover! Let's dive in.
## Setup
### Jupyter Notebook
This guide (and most of the other guides in the documentation) uses [Jupyter notebooks](https://jupyter.org/) and assumes the reader is as well. Jupyter notebooks are perfect for learning how to work with LLM systems because oftentimes things can go wrong (unexpected output, API down, etc) and going through guides in an interactive environment is a great way to better understand them.
You do not NEED to go through the guide in a Jupyter Notebook, but it is recommended. See [here](https://jupyter.org/install) for instructions on how to install.
For more details, see our [Installation guide](/docs/get_started/installation).
### LangSmith
Many of the applications you build with LangChain will contain multiple steps with multiple invocations of LLM calls.
As these applications get more and more complex, it becomes crucial to be able to inspect what exactly is going on inside your chain or agent.
The best way to do this is with [LangSmith](https://smith.langchain.com).
Note that LangSmith is not needed, but it is helpful.
If you do want to use LangSmith, after you sign up at the link above, make sure to set your environment variables to start logging traces:
```shell
export LANGCHAIN_TRACING_V2="true"
export LANGCHAIN_API_KEY="..."
```
## Building with LangChain
LangChain enables building application that connect external sources of data and computation to LLMs.
In this quickstart, we will walk through a few different ways of doing that.
We will start with a simple LLM chain, which just relies on information in the prompt template to respond.
Next, we will build a retrieval chain, which fetches data from a separate database and passes that into the prompt template.
We will then add in chat history, to create a conversation retrieval chain. This allows you to interact in a chat manner with this LLM, so it remembers previous questions.
Finally, we will build an agent - which utilizes an LLM to determine whether or not it needs to fetch data to answer questions.
We will cover these at a high level, but there are lot of details to all of these!
We will link to relevant docs.
## LLM Chain
We'll show how to use models available via API, like OpenAI, and local open source models, using integrations like Ollama.
<Tabs>
<TabItem value="openai" label="OpenAI" default>
First we'll need to import the LangChain x OpenAI integration package.
```shell
pip install langchain-openai
```
Accessing the API requires an API key, which you can get by creating an account and heading [here](https://platform.openai.com/account/api-keys). Once we have a key we'll want to set it as an environment variable by running:
```shell
export OPENAI_API_KEY="..."
```
We can then initialize the model:
```python
from langchain_openai import ChatOpenAI
llm = ChatOpenAI()
```
If you'd prefer not to set an environment variable you can pass the key in directly via the `api_key` named parameter when initiating the OpenAI LLM class:
[Ollama](https://ollama.ai/) allows you to run open-source large language models, such as Llama 2, locally.
First, follow [these instructions](https://github.com/jmorganca/ollama) to set up and run a local Ollama instance:
* [Download](https://ollama.ai/download)
* Fetch a model via `ollama pull llama2`
Then, make sure the Ollama server is running. After that, you can do:
```python
from langchain_community.llms import Ollama
llm = Ollama(model="llama2")
```
</TabItem>
<TabItem value="anthropic" label="Anthropic">
First we'll need to import the LangChain x Anthropic package.
```shell
pip install langchain-anthropic
```
Accessing the API requires an API key, which you can get by creating an account [here](https://claude.ai/login). Once we have a key we'll want to set it as an environment variable by running:
If you'd prefer not to set an environment variable you can pass the key in directly via the `api_key` named parameter when initiating the Anthropic Chat Model class:
```python
llm = ChatAnthropic(api_key="...")
```
</TabItem>
<TabItem value="cohere" label="Cohere">
First we'll need to import the Cohere SDK package.
```shell
pip install langchain-cohere
```
Accessing the API requires an API key, which you can get by creating an account and heading [here](https://dashboard.cohere.com/api-keys). Once we have a key we'll want to set it as an environment variable by running:
```shell
export COHERE_API_KEY="..."
```
We can then initialize the model:
```python
from langchain_cohere import ChatCohere
llm = ChatCohere()
```
If you'd prefer not to set an environment variable you can pass the key in directly via the `cohere_api_key` named parameter when initiating the Cohere LLM class:
```python
from langchain_cohere import ChatCohere
llm = ChatCohere(cohere_api_key="...")
```
</TabItem>
</Tabs>
Once you've installed and initialized the LLM of your choice, we can try using it!
Let's ask it what LangSmith is - this is something that wasn't present in the training data so it shouldn't have a very good response.
```python
llm.invoke("how can langsmith help with testing?")
```
We can also guide its response with a prompt template.
Prompt templates convert raw user input to better input to the LLM.
```python
from langchain_core.prompts import ChatPromptTemplate
prompt = ChatPromptTemplate.from_messages([
("system", "You are a world class technical documentation writer."),
("user", "{input}")
])
```
We can now combine these into a simple LLM chain:
```python
chain = prompt | llm
```
We can now invoke it and ask the same question. It still won't know the answer, but it should respond in a more proper tone for a technical writer!
```python
chain.invoke({"input": "how can langsmith help with testing?"})
```
The output of a ChatModel (and therefore, of this chain) is a message. However, it's often much more convenient to work with strings. Let's add a simple output parser to convert the chat message to a string.
```python
from langchain_core.output_parsers import StrOutputParser
output_parser = StrOutputParser()
```
We can now add this to the previous chain:
```python
chain = prompt | llm | output_parser
```
We can now invoke it and ask the same question. The answer will now be a string (rather than a ChatMessage).
```python
chain.invoke({"input": "how can langsmith help with testing?"})
```
### Diving Deeper
We've now successfully set up a basic LLM chain. We only touched on the basics of prompts, models, and output parsers - for a deeper dive into everything mentioned here, see [this section of documentation](/docs/modules/model_io).
## Retrieval Chain
To properly answer the original question ("how can langsmith help with testing?"), we need to provide additional context to the LLM.
We can do this via *retrieval*.
Retrieval is useful when you have **too much data** to pass to the LLM directly.
You can then use a retriever to fetch only the most relevant pieces and pass those in.
In this process, we will look up relevant documents from a *Retriever* and then pass them into the prompt.
A Retriever can be backed by anything - a SQL table, the internet, etc - but in this instance we will populate a vector store and use that as a retriever. For more information on vectorstores, see [this documentation](/docs/modules/data_connection/vectorstores).
First, we need to load the data that we want to index. To do this, we will use the WebBaseLoader. This requires installing [BeautifulSoup](https://beautiful-soup-4.readthedocs.io/en/latest/):
```shell
pip install beautifulsoup4
```
After that, we can import and use WebBaseLoader.
```python
from langchain_community.document_loaders import WebBaseLoader
Next, we need to index it into a vectorstore. This requires a few components, namely an [embedding model](/docs/modules/data_connection/text_embedding) and a [vectorstore](/docs/modules/data_connection/vectorstores).
For embedding models, we once again provide examples for accessing via API or by running local models.
Now that we have this data indexed in a vectorstore, we will create a retrieval chain.
This chain will take an incoming question, look up relevant documents, then pass those documents along with the original question into an LLM and ask it to answer the original question.
First, let's set up the chain that takes a question and the retrieved documents and generates an answer.
```python
from langchain.chains.combine_documents import create_stuff_documents_chain
prompt = ChatPromptTemplate.from_template("""Answer the following question based only on the provided context:
We can now invoke this chain. This returns a dictionary - the response from the LLM is in the `answer` key
```python
response = retrieval_chain.invoke({"input": "how can langsmith help with testing?"})
print(response["answer"])
# LangSmith offers several features that can help with testing:...
```
This answer should be much more accurate!
### Diving Deeper
We've now successfully set up a basic retrieval chain. We only touched on the basics of retrieval - for a deeper dive into everything mentioned here, see [this section of documentation](/docs/modules/data_connection).
## Conversation Retrieval Chain
The chain we've created so far can only answer single questions. One of the main types of LLM applications that people are building are chat bots. So how do we turn this chain into one that can answer follow up questions?
We can still use the `create_retrieval_chain` function, but we need to change two things:
1. The retrieval method should now not just work on the most recent input, but rather should take the whole history into account.
2. The final LLM chain should likewise take the whole history into account
**Updating Retrieval**
In order to update retrieval, we will create a new chain. This chain will take in the most recent input (`input`) and the conversation history (`chat_history`) and use an LLM to generate a search query.
```python
from langchain.chains import create_history_aware_retriever
from langchain_core.prompts import MessagesPlaceholder
# First we need a prompt that we can pass into an LLM to generate this search query
We can test this out by passing in an instance where the user asks a follow-up question.
```python
from langchain_core.messages import HumanMessage, AIMessage
chat_history = [HumanMessage(content="Can LangSmith help test my LLM applications?"), AIMessage(content="Yes!")]
retriever_chain.invoke({
"chat_history": chat_history,
"input": "Tell me how"
})
```
You should see that this returns documents about testing in LangSmith. This is because the LLM generated a new query, combining the chat history with the follow-up question.
Now that we have this new retriever, we can create a new chain to continue the conversation with these retrieved documents in mind.
```python
prompt = ChatPromptTemplate.from_messages([
("system", "Answer the user's questions based on the below context:\n\n{context}"),
chat_history = [HumanMessage(content="Can LangSmith help test my LLM applications?"), AIMessage(content="Yes!")]
retrieval_chain.invoke({
"chat_history": chat_history,
"input": "Tell me how"
})
```
We can see that this gives a coherent answer - we've successfully turned our retrieval chain into a chatbot!
## Agent
We've so far created examples of chains - where each step is known ahead of time.
The final thing we will create is an agent - where the LLM decides what steps to take.
**NOTE: for this example we will only show how to create an agent using OpenAI models, as local models are not reliable enough yet.**
One of the first things to do when building an agent is to decide what tools it should have access to.
For this example, we will give the agent access to two tools:
1. The retriever we just created. This will let it easily answer questions about LangSmith
2. A search tool. This will let it easily answer questions that require up-to-date information.
First, let's set up a tool for the retriever we just created:
```python
from langchain.tools.retriever import create_retriever_tool
retriever_tool = create_retriever_tool(
retriever,
"langsmith_search",
"Search for information about LangSmith. For any questions about LangSmith, you must use this tool!",
)
```
The search tool that we will use is [Tavily](/docs/integrations/retrievers/tavily). This will require an API key (they have generous free tier). After creating it on their platform, you need to set it as an environment variable:
```shell
export TAVILY_API_KEY=...
```
If you do not want to set up an API key, you can skip creating this tool.
```python
from langchain_community.tools.tavily_search import TavilySearchResults
search = TavilySearchResults()
```
We can now create a list of the tools we want to work with:
```python
tools = [retriever_tool, search]
```
Now that we have the tools, we can create an agent to use them. We will go over this pretty quickly - for a deeper dive into what exactly is going on, check out the [Agent's Getting Started documentation](/docs/modules/agents)
Install langchain hub first
```bash
pip install langchainhub
```
Install the langchain-openai package
To interact with OpenAI we need to use langchain-openai which connects with OpenAI SDK[https://github.com/langchain-ai/langchain/tree/master/libs/partners/openai].
```bash
pip install langchain-openai
```
Now we can use it to get a predefined prompt
```python
from langchain_openai import ChatOpenAI
from langchain import hub
from langchain.agents import create_openai_functions_agent
We can now invoke the agent and see how it responds! We can ask it questions about LangSmith:
```python
agent_executor.invoke({"input": "how can langsmith help with testing?"})
```
We can ask it about the weather:
```python
agent_executor.invoke({"input": "what is the weather in SF?"})
```
We can have conversations with it:
```python
chat_history = [HumanMessage(content="Can LangSmith help test my LLM applications?"), AIMessage(content="Yes!")]
agent_executor.invoke({
"chat_history": chat_history,
"input": "Tell me how"
})
```
### Diving Deeper
We've now successfully set up a basic agent. We only touched on the basics of agents - for a deeper dive into everything mentioned here, see [this section of documentation](/docs/modules/agents).
## Serving with LangServe
Now that we've built an application, we need to serve it. That's where LangServe comes in.
LangServe helps developers deploy LangChain chains as a REST API. You do not need to use LangServe to use LangChain, but in this guide we'll show how you can deploy your app with LangServe.
While the first part of this guide was intended to be run in a Jupyter Notebook, we will now move out of that. We will be creating a Python file and then interacting with it from the command line.
Install with:
```bash
pip install "langserve[all]"
```
### Server
To create a server for our application we'll make a `serve.py` file. This will contain our logic for serving our application. It consists of three things:
1. The definition of our chain that we just built above
2. Our FastAPI app
3. A definition of a route from which to serve the chain, which is done with `langserve.add_routes`
```python
#!/usr/bin/env python
from typing import List
from fastapi import FastAPI
from langchain_core.prompts import ChatPromptTemplate
from langchain_openai import ChatOpenAI
from langchain_community.document_loaders import WebBaseLoader
from langchain_openai import OpenAIEmbeddings
from langchain_community.vectorstores import FAISS
from langchain_text_splitters import RecursiveCharacterTextSplitter
from langchain.tools.retriever import create_retriever_tool
from langchain_community.tools.tavily_search import TavilySearchResults
from langchain import hub
from langchain.agents import create_openai_functions_agent
from langchain.agents import AgentExecutor
from langchain.pydantic_v1 import BaseModel, Field
we should see our chain being served at localhost:8000.
### Playground
Every LangServe service comes with a simple built-in UI for configuring and invoking the application with streaming output and visibility into intermediate steps.
Head to http://localhost:8000/agent/playground/ to try it out! Pass in the same question as before - "how can langsmith help with testing?" - and it should respond same as before.
### Client
Now let's set up a client for programmatically interacting with our service. We can easily do this with the `[langserve.RemoteRunnable](/docs/langserve#client)`.
Using this, we can interact with the served chain as if it were running client-side.
"chat_history": [] # Providing an empty list as this is the first call
})
```
To learn more about the many other features of LangServe [head here](/docs/langserve).
## Next steps
We've touched on how to build an application with LangChain, how to trace it with LangSmith, and how to serve it with LangServe.
There are a lot more features in all three of these than we can cover here.
To continue on your journey, we recommend you read the following (in order):
- All of these features are backed by [LangChain Expression Language (LCEL)](/docs/expression_language) - a way to chain these components together. Check out that documentation to better understand how to create custom chains.
- [Model IO](/docs/modules/model_io) covers more details of prompts, LLMs, and output parsers.
- [Retrieval](/docs/modules/data_connection) covers more details of everything related to retrieval
- [Agents](/docs/modules/agents) covers details of everything related to agents
- Explore common [end-to-end use cases](/docs/use_cases/) and [template applications](/docs/templates)
- [Read up on LangSmith](/docs/langsmith/), the platform for debugging, testing, monitoring and more
- Learn more about serving your applications with [LangServe](/docs/langserve)
If you're building with LLMs, at some point something will break, and you'll need to debug. A model call will fail, or the model output will be misformatted, or there will be some nested model calls and it won't be clear where along the way an incorrect output was created.
Here are a few different tools and functionalities to aid in debugging.
## Tracing
Platforms with tracing capabilities like [LangSmith](/docs/langsmith/) are the most comprehensive solutions for debugging. These platforms make it easy to not only log and visualize LLM apps, but also to actively debug, test and refine them.
When building production-grade LLM applications, platforms like this are essential.
## `set_debug` and `set_verbose`
If you're prototyping in Jupyter Notebooks or running Python scripts, it can be helpful to print out the intermediate steps of a Chain run.
There are a number of ways to enable printing at varying degrees of verbosity.
Let's suppose we have a simple agent, and want to visualize the actions it takes and tool outputs it receives. Without any debugging, here's what we see:
agent.run("Who directed the 2023 film Oppenheimer and what is their age? What is their age in days (assume 365 days per year)?")
```
<CodeOutputBlock lang="python">
```
'The director of the 2023 film Oppenheimer is Christopher Nolan and he is approximately 19345 days old in 2023.'
```
</CodeOutputBlock>
### `set_debug(True)`
Setting the global `debug` flag will cause all LangChain components with callback support (chains, models, agents, tools, retrievers) to print the inputs they receive and outputs they generate. This is the most verbose setting and will fully log raw inputs and outputs.
```python
fromlangchain.globalsimportset_debug
set_debug(True)
agent.run("Who directed the 2023 film Oppenheimer and what is their age? What is their age in days (assume 365 days per year)?")
```
<details> <summary>Console output</summary>
<CodeOutputBlock lang="python">
```
[chain/start] [1:RunTypeEnum.chain:AgentExecutor] Entering Chain run with input:
{
"input": "Who directed the 2023 film Oppenheimer and what is their age? What is their age in days (assume 365 days per year)?"
}
[chain/start] [1:RunTypeEnum.chain:AgentExecutor > 2:RunTypeEnum.chain:LLMChain] Entering Chain run with input:
{
"input": "Who directed the 2023 film Oppenheimer and what is their age? What is their age in days (assume 365 days per year)?",
"agent_scratchpad": "",
"stop": [
"\nObservation:",
"\n\tObservation:"
]
}
[llm/start] [1:RunTypeEnum.chain:AgentExecutor > 2:RunTypeEnum.chain:LLMChain > 3:RunTypeEnum.llm:ChatOpenAI] Entering LLM run with input:
{
"prompts": [
"Human: Answer the following questions as best you can. You have access to the following tools:\n\nduckduckgo_search: A wrapper around DuckDuckGo Search. Useful for when you need to answer questions about current events. Input should be a search query.\nCalculator: Useful for when you need to answer questions about math.\n\nUse the following format:\n\nQuestion: the input question you must answer\nThought: you should always think about what to do\nAction: the action to take, should be one of [duckduckgo_search, Calculator]\nAction Input: the input to the action\nObservation: the result of the action\n... (this Thought/Action/Action Input/Observation can repeat N times)\nThought: I now know the final answer\nFinal Answer: the final answer to the original input question\n\nBegin!\n\nQuestion: Who directed the 2023 film Oppenheimer and what is their age? What is their age in days (assume 365 days per year)?\nThought:"
]
}
[llm/end] [1:RunTypeEnum.chain:AgentExecutor > 2:RunTypeEnum.chain:LLMChain > 3:RunTypeEnum.llm:ChatOpenAI] [5.53s] Exiting LLM run with output:
{
"generations": [
[
{
"text": "I need to find out who directed the 2023 film Oppenheimer and their age. Then, I need to calculate their age in days. I will use DuckDuckGo to find out the director and their age.\nAction: duckduckgo_search\nAction Input: \"Director of the 2023 film Oppenheimer and their age\"",
"generation_info": {
"finish_reason": "stop"
},
"message": {
"lc": 1,
"type": "constructor",
"id": [
"langchain",
"schema",
"messages",
"AIMessage"
],
"kwargs": {
"content": "I need to find out who directed the 2023 film Oppenheimer and their age. Then, I need to calculate their age in days. I will use DuckDuckGo to find out the director and their age.\nAction: duckduckgo_search\nAction Input: \"Director of the 2023 film Oppenheimer and their age\"",
"additional_kwargs": {}
}
}
}
]
],
"llm_output": {
"token_usage": {
"prompt_tokens": 206,
"completion_tokens": 71,
"total_tokens": 277
},
"model_name": "gpt-4"
},
"run": null
}
[chain/end] [1:RunTypeEnum.chain:AgentExecutor > 2:RunTypeEnum.chain:LLMChain] [5.53s] Exiting Chain run with output:
{
"text": "I need to find out who directed the 2023 film Oppenheimer and their age. Then, I need to calculate their age in days. I will use DuckDuckGo to find out the director and their age.\nAction: duckduckgo_search\nAction Input: \"Director of the 2023 film Oppenheimer and their age\""
}
[tool/start] [1:RunTypeEnum.chain:AgentExecutor > 4:RunTypeEnum.tool:duckduckgo_search] Entering Tool run with input:
"Director of the 2023 film Oppenheimer and their age"
[tool/end] [1:RunTypeEnum.chain:AgentExecutor > 4:RunTypeEnum.tool:duckduckgo_search] [1.51s] Exiting Tool run with output:
"Capturing the mad scramble to build the first atomic bomb required rapid-fire filming, strict set rules and the construction of an entire 1940s western town. By Jada Yuan. July 19, 2023 at 5:00 a ... In Christopher Nolan's new film, "Oppenheimer," Cillian Murphy stars as J. Robert Oppenheimer, the American physicist who oversaw the Manhattan Project in Los Alamos, N.M. Universal Pictures... Oppenheimer: Directed by Christopher Nolan. With Cillian Murphy, Emily Blunt, Robert Downey Jr., Alden Ehrenreich. The story of American scientist J. Robert Oppenheimer and his role in the development of the atomic bomb. Christopher Nolan goes deep on 'Oppenheimer,' his most 'extreme' film to date. By Kenneth Turan. July 11, 2023 5 AM PT. For Subscribers. Christopher Nolan is photographed in Los Angeles ... Oppenheimer is a 2023 epic biographical thriller film written and directed by Christopher Nolan.It is based on the 2005 biography American Prometheus by Kai Bird and Martin J. Sherwin about J. Robert Oppenheimer, a theoretical physicist who was pivotal in developing the first nuclear weapons as part of the Manhattan Project and thereby ushering in the Atomic Age."
[chain/start] [1:RunTypeEnum.chain:AgentExecutor > 5:RunTypeEnum.chain:LLMChain] Entering Chain run with input:
{
"input": "Who directed the 2023 film Oppenheimer and what is their age? What is their age in days (assume 365 days per year)?",
"agent_scratchpad": "I need to find out who directed the 2023 film Oppenheimer and their age. Then, I need to calculate their age in days. I will use DuckDuckGo to find out the director and their age.\nAction: duckduckgo_search\nAction Input: \"Director of the 2023 film Oppenheimer and their age\"\nObservation: Capturing the mad scramble to build the first atomic bomb required rapid-fire filming, strict set rules and the construction of an entire 1940s western town. By Jada Yuan. July 19, 2023 at 5:00 a ... In Christopher Nolan's new film, \"Oppenheimer,\" Cillian Murphy stars as J. Robert Oppenheimer, the American physicist who oversaw the Manhattan Project in Los Alamos, N.M. Universal Pictures... Oppenheimer: Directed by Christopher Nolan. With Cillian Murphy, Emily Blunt, Robert Downey Jr., Alden Ehrenreich. The story of American scientist J. Robert Oppenheimer and his role in the development of the atomic bomb. Christopher Nolan goes deep on 'Oppenheimer,' his most 'extreme' film to date. By Kenneth Turan. July 11, 2023 5 AM PT. For Subscribers. Christopher Nolan is photographed in Los Angeles ... Oppenheimer is a 2023 epic biographical thriller film written and directed by Christopher Nolan.It is based on the 2005 biography American Prometheus by Kai Bird and Martin J. Sherwin about J. Robert Oppenheimer, a theoretical physicist who was pivotal in developing the first nuclear weapons as part of the Manhattan Project and thereby ushering in the Atomic Age.\nThought:",
"stop": [
"\nObservation:",
"\n\tObservation:"
]
}
[llm/start] [1:RunTypeEnum.chain:AgentExecutor > 5:RunTypeEnum.chain:LLMChain > 6:RunTypeEnum.llm:ChatOpenAI] Entering LLM run with input:
{
"prompts": [
"Human: Answer the following questions as best you can. You have access to the following tools:\n\nduckduckgo_search: A wrapper around DuckDuckGo Search. Useful for when you need to answer questions about current events. Input should be a search query.\nCalculator: Useful for when you need to answer questions about math.\n\nUse the following format:\n\nQuestion: the input question you must answer\nThought: you should always think about what to do\nAction: the action to take, should be one of [duckduckgo_search, Calculator]\nAction Input: the input to the action\nObservation: the result of the action\n... (this Thought/Action/Action Input/Observation can repeat N times)\nThought: I now know the final answer\nFinal Answer: the final answer to the original input question\n\nBegin!\n\nQuestion: Who directed the 2023 film Oppenheimer and what is their age? What is their age in days (assume 365 days per year)?\nThought:I need to find out who directed the 2023 film Oppenheimer and their age. Then, I need to calculate their age in days. I will use DuckDuckGo to find out the director and their age.\nAction: duckduckgo_search\nAction Input: \"Director of the 2023 film Oppenheimer and their age\"\nObservation: Capturing the mad scramble to build the first atomic bomb required rapid-fire filming, strict set rules and the construction of an entire 1940s western town. By Jada Yuan. July 19, 2023 at 5:00 a ... In Christopher Nolan's new film, \"Oppenheimer,\" Cillian Murphy stars as J. Robert Oppenheimer, the American physicist who oversaw the Manhattan Project in Los Alamos, N.M. Universal Pictures... Oppenheimer: Directed by Christopher Nolan. With Cillian Murphy, Emily Blunt, Robert Downey Jr., Alden Ehrenreich. The story of American scientist J. Robert Oppenheimer and his role in the development of the atomic bomb. Christopher Nolan goes deep on 'Oppenheimer,' his most 'extreme' film to date. By Kenneth Turan. July 11, 2023 5 AM PT. For Subscribers. Christopher Nolan is photographed in Los Angeles ... Oppenheimer is a 2023 epic biographical thriller film written and directed by Christopher Nolan.It is based on the 2005 biography American Prometheus by Kai Bird and Martin J. Sherwin about J. Robert Oppenheimer, a theoretical physicist who was pivotal in developing the first nuclear weapons as part of the Manhattan Project and thereby ushering in the Atomic Age.\nThought:"
]
}
[llm/end] [1:RunTypeEnum.chain:AgentExecutor > 5:RunTypeEnum.chain:LLMChain > 6:RunTypeEnum.llm:ChatOpenAI] [4.46s] Exiting LLM run with output:
{
"generations": [
[
{
"text": "The director of the 2023 film Oppenheimer is Christopher Nolan. Now I need to find out his age.\nAction: duckduckgo_search\nAction Input: \"Christopher Nolan age\"",
"generation_info": {
"finish_reason": "stop"
},
"message": {
"lc": 1,
"type": "constructor",
"id": [
"langchain",
"schema",
"messages",
"AIMessage"
],
"kwargs": {
"content": "The director of the 2023 film Oppenheimer is Christopher Nolan. Now I need to find out his age.\nAction: duckduckgo_search\nAction Input: \"Christopher Nolan age\"",
"additional_kwargs": {}
}
}
}
]
],
"llm_output": {
"token_usage": {
"prompt_tokens": 550,
"completion_tokens": 39,
"total_tokens": 589
},
"model_name": "gpt-4"
},
"run": null
}
[chain/end] [1:RunTypeEnum.chain:AgentExecutor > 5:RunTypeEnum.chain:LLMChain] [4.46s] Exiting Chain run with output:
{
"text": "The director of the 2023 film Oppenheimer is Christopher Nolan. Now I need to find out his age.\nAction: duckduckgo_search\nAction Input: \"Christopher Nolan age\""
}
[tool/start] [1:RunTypeEnum.chain:AgentExecutor > 7:RunTypeEnum.tool:duckduckgo_search] Entering Tool run with input:
"Christopher Nolan age"
[tool/end] [1:RunTypeEnum.chain:AgentExecutor > 7:RunTypeEnum.tool:duckduckgo_search] [1.33s] Exiting Tool run with output:
"Christopher Edward Nolan CBE (born 30 July 1970) is a British and American filmmaker. Known for his Hollywood blockbusters with complex storytelling, Nolan is considered a leading filmmaker of the 21st century. His films have grossed $5 billion worldwide. The recipient of many accolades, he has been nominated for five Academy Awards, five BAFTA Awards and six Golden Globe Awards. July 30, 1970 (age 52) London England Notable Works: "Dunkirk" "Tenet" "The Prestige" See all related content → Recent News Jul. 13, 2023, 11:11 AM ET (AP) Cillian Murphy, playing Oppenheimer, finally gets to lead a Christopher Nolan film July 11, 2023 5 AM PT For Subscribers Christopher Nolan is photographed in Los Angeles. (Joe Pugliese / For The Times) This is not the story I was supposed to write. Oppenheimer director Christopher Nolan, Cillian Murphy, Emily Blunt and Matt Damon on the stakes of making a three-hour, CGI-free summer film. Christopher Nolan, the director behind such films as "Dunkirk," "Inception," "Interstellar," and the "Dark Knight" trilogy, has spent the last three years living in Oppenheimer's world, writing ..."
[chain/start] [1:RunTypeEnum.chain:AgentExecutor > 8:RunTypeEnum.chain:LLMChain] Entering Chain run with input:
{
"input": "Who directed the 2023 film Oppenheimer and what is their age? What is their age in days (assume 365 days per year)?",
"agent_scratchpad": "I need to find out who directed the 2023 film Oppenheimer and their age. Then, I need to calculate their age in days. I will use DuckDuckGo to find out the director and their age.\nAction: duckduckgo_search\nAction Input: \"Director of the 2023 film Oppenheimer and their age\"\nObservation: Capturing the mad scramble to build the first atomic bomb required rapid-fire filming, strict set rules and the construction of an entire 1940s western town. By Jada Yuan. July 19, 2023 at 5:00 a ... In Christopher Nolan's new film, \"Oppenheimer,\" Cillian Murphy stars as J. Robert Oppenheimer, the American physicist who oversaw the Manhattan Project in Los Alamos, N.M. Universal Pictures... Oppenheimer: Directed by Christopher Nolan. With Cillian Murphy, Emily Blunt, Robert Downey Jr., Alden Ehrenreich. The story of American scientist J. Robert Oppenheimer and his role in the development of the atomic bomb. Christopher Nolan goes deep on 'Oppenheimer,' his most 'extreme' film to date. By Kenneth Turan. July 11, 2023 5 AM PT. For Subscribers. Christopher Nolan is photographed in Los Angeles ... Oppenheimer is a 2023 epic biographical thriller film written and directed by Christopher Nolan.It is based on the 2005 biography American Prometheus by Kai Bird and Martin J. Sherwin about J. Robert Oppenheimer, a theoretical physicist who was pivotal in developing the first nuclear weapons as part of the Manhattan Project and thereby ushering in the Atomic Age.\nThought:The director of the 2023 film Oppenheimer is Christopher Nolan. Now I need to find out his age.\nAction: duckduckgo_search\nAction Input: \"Christopher Nolan age\"\nObservation: Christopher Edward Nolan CBE (born 30 July 1970) is a British and American filmmaker. Known for his Hollywood blockbusters with complex storytelling, Nolan is considered a leading filmmaker of the 21st century. His films have grossed $5 billion worldwide. The recipient of many accolades, he has been nominated for five Academy Awards, five BAFTA Awards and six Golden Globe Awards. July 30, 1970 (age 52) London England Notable Works: \"Dunkirk\" \"Tenet\" \"The Prestige\" See all related content → Recent News Jul. 13, 2023, 11:11 AM ET (AP) Cillian Murphy, playing Oppenheimer, finally gets to lead a Christopher Nolan film July 11, 2023 5 AM PT For Subscribers Christopher Nolan is photographed in Los Angeles. (Joe Pugliese / For The Times) This is not the story I was supposed to write. Oppenheimer director Christopher Nolan, Cillian Murphy, Emily Blunt and Matt Damon on the stakes of making a three-hour, CGI-free summer film. Christopher Nolan, the director behind such films as \"Dunkirk,\" \"Inception,\" \"Interstellar,\" and the \"Dark Knight\" trilogy, has spent the last three years living in Oppenheimer's world, writing ...\nThought:",
"stop": [
"\nObservation:",
"\n\tObservation:"
]
}
[llm/start] [1:RunTypeEnum.chain:AgentExecutor > 8:RunTypeEnum.chain:LLMChain > 9:RunTypeEnum.llm:ChatOpenAI] Entering LLM run with input:
{
"prompts": [
"Human: Answer the following questions as best you can. You have access to the following tools:\n\nduckduckgo_search: A wrapper around DuckDuckGo Search. Useful for when you need to answer questions about current events. Input should be a search query.\nCalculator: Useful for when you need to answer questions about math.\n\nUse the following format:\n\nQuestion: the input question you must answer\nThought: you should always think about what to do\nAction: the action to take, should be one of [duckduckgo_search, Calculator]\nAction Input: the input to the action\nObservation: the result of the action\n... (this Thought/Action/Action Input/Observation can repeat N times)\nThought: I now know the final answer\nFinal Answer: the final answer to the original input question\n\nBegin!\n\nQuestion: Who directed the 2023 film Oppenheimer and what is their age? What is their age in days (assume 365 days per year)?\nThought:I need to find out who directed the 2023 film Oppenheimer and their age. Then, I need to calculate their age in days. I will use DuckDuckGo to find out the director and their age.\nAction: duckduckgo_search\nAction Input: \"Director of the 2023 film Oppenheimer and their age\"\nObservation: Capturing the mad scramble to build the first atomic bomb required rapid-fire filming, strict set rules and the construction of an entire 1940s western town. By Jada Yuan. July 19, 2023 at 5:00 a ... In Christopher Nolan's new film, \"Oppenheimer,\" Cillian Murphy stars as J. Robert Oppenheimer, the American physicist who oversaw the Manhattan Project in Los Alamos, N.M. Universal Pictures... Oppenheimer: Directed by Christopher Nolan. With Cillian Murphy, Emily Blunt, Robert Downey Jr., Alden Ehrenreich. The story of American scientist J. Robert Oppenheimer and his role in the development of the atomic bomb. Christopher Nolan goes deep on 'Oppenheimer,' his most 'extreme' film to date. By Kenneth Turan. July 11, 2023 5 AM PT. For Subscribers. Christopher Nolan is photographed in Los Angeles ... Oppenheimer is a 2023 epic biographical thriller film written and directed by Christopher Nolan.It is based on the 2005 biography American Prometheus by Kai Bird and Martin J. Sherwin about J. Robert Oppenheimer, a theoretical physicist who was pivotal in developing the first nuclear weapons as part of the Manhattan Project and thereby ushering in the Atomic Age.\nThought:The director of the 2023 film Oppenheimer is Christopher Nolan. Now I need to find out his age.\nAction: duckduckgo_search\nAction Input: \"Christopher Nolan age\"\nObservation: Christopher Edward Nolan CBE (born 30 July 1970) is a British and American filmmaker. Known for his Hollywood blockbusters with complex storytelling, Nolan is considered a leading filmmaker of the 21st century. His films have grossed $5 billion worldwide. The recipient of many accolades, he has been nominated for five Academy Awards, five BAFTA Awards and six Golden Globe Awards. July 30, 1970 (age 52) London England Notable Works: \"Dunkirk\" \"Tenet\" \"The Prestige\" See all related content → Recent News Jul. 13, 2023, 11:11 AM ET (AP) Cillian Murphy, playing Oppenheimer, finally gets to lead a Christopher Nolan film July 11, 2023 5 AM PT For Subscribers Christopher Nolan is photographed in Los Angeles. (Joe Pugliese / For The Times) This is not the story I was supposed to write. Oppenheimer director Christopher Nolan, Cillian Murphy, Emily Blunt and Matt Damon on the stakes of making a three-hour, CGI-free summer film. Christopher Nolan, the director behind such films as \"Dunkirk,\" \"Inception,\" \"Interstellar,\" and the \"Dark Knight\" trilogy, has spent the last three years living in Oppenheimer's world, writing ...\nThought:"
]
}
[llm/end] [1:RunTypeEnum.chain:AgentExecutor > 8:RunTypeEnum.chain:LLMChain > 9:RunTypeEnum.llm:ChatOpenAI] [2.69s] Exiting LLM run with output:
{
"generations": [
[
{
"text": "Christopher Nolan was born on July 30, 1970, which makes him 52 years old in 2023. Now I need to calculate his age in days.\nAction: Calculator\nAction Input: 52*365",
"generation_info": {
"finish_reason": "stop"
},
"message": {
"lc": 1,
"type": "constructor",
"id": [
"langchain",
"schema",
"messages",
"AIMessage"
],
"kwargs": {
"content": "Christopher Nolan was born on July 30, 1970, which makes him 52 years old in 2023. Now I need to calculate his age in days.\nAction: Calculator\nAction Input: 52*365",
"additional_kwargs": {}
}
}
}
]
],
"llm_output": {
"token_usage": {
"prompt_tokens": 868,
"completion_tokens": 46,
"total_tokens": 914
},
"model_name": "gpt-4"
},
"run": null
}
[chain/end] [1:RunTypeEnum.chain:AgentExecutor > 8:RunTypeEnum.chain:LLMChain] [2.69s] Exiting Chain run with output:
{
"text": "Christopher Nolan was born on July 30, 1970, which makes him 52 years old in 2023. Now I need to calculate his age in days.\nAction: Calculator\nAction Input: 52*365"
}
[tool/start] [1:RunTypeEnum.chain:AgentExecutor > 10:RunTypeEnum.tool:Calculator] Entering Tool run with input:
"52*365"
[chain/start] [1:RunTypeEnum.chain:AgentExecutor > 10:RunTypeEnum.tool:Calculator > 11:RunTypeEnum.chain:LLMMathChain] Entering Chain run with input:
{
"question": "52*365"
}
[chain/start] [1:RunTypeEnum.chain:AgentExecutor > 10:RunTypeEnum.tool:Calculator > 11:RunTypeEnum.chain:LLMMathChain > 12:RunTypeEnum.chain:LLMChain] Entering Chain run with input:
{
"question": "52*365",
"stop": [
"```output"
]
}
[llm/start] [1:RunTypeEnum.chain:AgentExecutor > 10:RunTypeEnum.tool:Calculator > 11:RunTypeEnum.chain:LLMMathChain > 12:RunTypeEnum.chain:LLMChain > 13:RunTypeEnum.llm:ChatOpenAI] Entering LLM run with input:
{
"prompts": [
"Human: Translate a math problem into a expression that can be executed using Python's numexpr library. Use the output of running this code to answer the question.\n\nQuestion: ${Question with math problem.}\n```text\n${single line mathematical expression that solves the problem}\n```\n...numexpr.evaluate(text)...\n```output\n${Output of running the code}\n```\nAnswer: ${Answer}\n\nBegin.\n\nQuestion: What is 37593 * 67?\n```text\n37593 * 67\n```\n...numexpr.evaluate(\"37593 * 67\")...\n```output\n2518731\n```\nAnswer: 2518731\n\nQuestion: 37593^(1/5)\n```text\n37593**(1/5)\n```\n...numexpr.evaluate(\"37593**(1/5)\")...\n```output\n8.222831614237718\n```\nAnswer: 8.222831614237718\n\nQuestion: 52*365"
]
}
[llm/end] [1:RunTypeEnum.chain:AgentExecutor > 10:RunTypeEnum.tool:Calculator > 11:RunTypeEnum.chain:LLMMathChain > 12:RunTypeEnum.chain:LLMChain > 13:RunTypeEnum.llm:ChatOpenAI] [2.89s] Exiting LLM run with output:
[chain/end] [1:RunTypeEnum.chain:AgentExecutor > 10:RunTypeEnum.tool:Calculator > 11:RunTypeEnum.chain:LLMMathChain] [2.90s] Exiting Chain run with output:
{
"answer": "Answer: 18980"
}
[tool/end] [1:RunTypeEnum.chain:AgentExecutor > 10:RunTypeEnum.tool:Calculator] [2.90s] Exiting Tool run with output:
"Answer: 18980"
[chain/start] [1:RunTypeEnum.chain:AgentExecutor > 14:RunTypeEnum.chain:LLMChain] Entering Chain run with input:
{
"input": "Who directed the 2023 film Oppenheimer and what is their age? What is their age in days (assume 365 days per year)?",
"agent_scratchpad": "I need to find out who directed the 2023 film Oppenheimer and their age. Then, I need to calculate their age in days. I will use DuckDuckGo to find out the director and their age.\nAction: duckduckgo_search\nAction Input: \"Director of the 2023 film Oppenheimer and their age\"\nObservation: Capturing the mad scramble to build the first atomic bomb required rapid-fire filming, strict set rules and the construction of an entire 1940s western town. By Jada Yuan. July 19, 2023 at 5:00 a ... In Christopher Nolan's new film, \"Oppenheimer,\" Cillian Murphy stars as J. Robert Oppenheimer, the American physicist who oversaw the Manhattan Project in Los Alamos, N.M. Universal Pictures... Oppenheimer: Directed by Christopher Nolan. With Cillian Murphy, Emily Blunt, Robert Downey Jr., Alden Ehrenreich. The story of American scientist J. Robert Oppenheimer and his role in the development of the atomic bomb. Christopher Nolan goes deep on 'Oppenheimer,' his most 'extreme' film to date. By Kenneth Turan. July 11, 2023 5 AM PT. For Subscribers. Christopher Nolan is photographed in Los Angeles ... Oppenheimer is a 2023 epic biographical thriller film written and directed by Christopher Nolan.It is based on the 2005 biography American Prometheus by Kai Bird and Martin J. Sherwin about J. Robert Oppenheimer, a theoretical physicist who was pivotal in developing the first nuclear weapons as part of the Manhattan Project and thereby ushering in the Atomic Age.\nThought:The director of the 2023 film Oppenheimer is Christopher Nolan. Now I need to find out his age.\nAction: duckduckgo_search\nAction Input: \"Christopher Nolan age\"\nObservation: Christopher Edward Nolan CBE (born 30 July 1970) is a British and American filmmaker. Known for his Hollywood blockbusters with complex storytelling, Nolan is considered a leading filmmaker of the 21st century. His films have grossed $5 billion worldwide. The recipient of many accolades, he has been nominated for five Academy Awards, five BAFTA Awards and six Golden Globe Awards. July 30, 1970 (age 52) London England Notable Works: \"Dunkirk\" \"Tenet\" \"The Prestige\" See all related content → Recent News Jul. 13, 2023, 11:11 AM ET (AP) Cillian Murphy, playing Oppenheimer, finally gets to lead a Christopher Nolan film July 11, 2023 5 AM PT For Subscribers Christopher Nolan is photographed in Los Angeles. (Joe Pugliese / For The Times) This is not the story I was supposed to write. Oppenheimer director Christopher Nolan, Cillian Murphy, Emily Blunt and Matt Damon on the stakes of making a three-hour, CGI-free summer film. Christopher Nolan, the director behind such films as \"Dunkirk,\" \"Inception,\" \"Interstellar,\" and the \"Dark Knight\" trilogy, has spent the last three years living in Oppenheimer's world, writing ...\nThought:Christopher Nolan was born on July 30, 1970, which makes him 52 years old in 2023. Now I need to calculate his age in days.\nAction: Calculator\nAction Input: 52*365\nObservation: Answer: 18980\nThought:",
"stop": [
"\nObservation:",
"\n\tObservation:"
]
}
[llm/start] [1:RunTypeEnum.chain:AgentExecutor > 14:RunTypeEnum.chain:LLMChain > 15:RunTypeEnum.llm:ChatOpenAI] Entering LLM run with input:
{
"prompts": [
"Human: Answer the following questions as best you can. You have access to the following tools:\n\nduckduckgo_search: A wrapper around DuckDuckGo Search. Useful for when you need to answer questions about current events. Input should be a search query.\nCalculator: Useful for when you need to answer questions about math.\n\nUse the following format:\n\nQuestion: the input question you must answer\nThought: you should always think about what to do\nAction: the action to take, should be one of [duckduckgo_search, Calculator]\nAction Input: the input to the action\nObservation: the result of the action\n... (this Thought/Action/Action Input/Observation can repeat N times)\nThought: I now know the final answer\nFinal Answer: the final answer to the original input question\n\nBegin!\n\nQuestion: Who directed the 2023 film Oppenheimer and what is their age? What is their age in days (assume 365 days per year)?\nThought:I need to find out who directed the 2023 film Oppenheimer and their age. Then, I need to calculate their age in days. I will use DuckDuckGo to find out the director and their age.\nAction: duckduckgo_search\nAction Input: \"Director of the 2023 film Oppenheimer and their age\"\nObservation: Capturing the mad scramble to build the first atomic bomb required rapid-fire filming, strict set rules and the construction of an entire 1940s western town. By Jada Yuan. July 19, 2023 at 5:00 a ... In Christopher Nolan's new film, \"Oppenheimer,\" Cillian Murphy stars as J. Robert Oppenheimer, the American physicist who oversaw the Manhattan Project in Los Alamos, N.M. Universal Pictures... Oppenheimer: Directed by Christopher Nolan. With Cillian Murphy, Emily Blunt, Robert Downey Jr., Alden Ehrenreich. The story of American scientist J. Robert Oppenheimer and his role in the development of the atomic bomb. Christopher Nolan goes deep on 'Oppenheimer,' his most 'extreme' film to date. By Kenneth Turan. July 11, 2023 5 AM PT. For Subscribers. Christopher Nolan is photographed in Los Angeles ... Oppenheimer is a 2023 epic biographical thriller film written and directed by Christopher Nolan.It is based on the 2005 biography American Prometheus by Kai Bird and Martin J. Sherwin about J. Robert Oppenheimer, a theoretical physicist who was pivotal in developing the first nuclear weapons as part of the Manhattan Project and thereby ushering in the Atomic Age.\nThought:The director of the 2023 film Oppenheimer is Christopher Nolan. Now I need to find out his age.\nAction: duckduckgo_search\nAction Input: \"Christopher Nolan age\"\nObservation: Christopher Edward Nolan CBE (born 30 July 1970) is a British and American filmmaker. Known for his Hollywood blockbusters with complex storytelling, Nolan is considered a leading filmmaker of the 21st century. His films have grossed $5 billion worldwide. The recipient of many accolades, he has been nominated for five Academy Awards, five BAFTA Awards and six Golden Globe Awards. July 30, 1970 (age 52) London England Notable Works: \"Dunkirk\" \"Tenet\" \"The Prestige\" See all related content → Recent News Jul. 13, 2023, 11:11 AM ET (AP) Cillian Murphy, playing Oppenheimer, finally gets to lead a Christopher Nolan film July 11, 2023 5 AM PT For Subscribers Christopher Nolan is photographed in Los Angeles. (Joe Pugliese / For The Times) This is not the story I was supposed to write. Oppenheimer director Christopher Nolan, Cillian Murphy, Emily Blunt and Matt Damon on the stakes of making a three-hour, CGI-free summer film. Christopher Nolan, the director behind such films as \"Dunkirk,\" \"Inception,\" \"Interstellar,\" and the \"Dark Knight\" trilogy, has spent the last three years living in Oppenheimer's world, writing ...\nThought:Christopher Nolan was born on July 30, 1970, which makes him 52 years old in 2023. Now I need to calculate his age in days.\nAction: Calculator\nAction Input: 52*365\nObservation: Answer: 18980\nThought:"
]
}
[llm/end] [1:RunTypeEnum.chain:AgentExecutor > 14:RunTypeEnum.chain:LLMChain > 15:RunTypeEnum.llm:ChatOpenAI] [3.52s] Exiting LLM run with output:
{
"generations": [
[
{
"text": "I now know the final answer\nFinal Answer: The director of the 2023 film Oppenheimer is Christopher Nolan and he is 52 years old. His age in days is approximately 18980 days.",
"generation_info": {
"finish_reason": "stop"
},
"message": {
"lc": 1,
"type": "constructor",
"id": [
"langchain",
"schema",
"messages",
"AIMessage"
],
"kwargs": {
"content": "I now know the final answer\nFinal Answer: The director of the 2023 film Oppenheimer is Christopher Nolan and he is 52 years old. His age in days is approximately 18980 days.",
"additional_kwargs": {}
}
}
}
]
],
"llm_output": {
"token_usage": {
"prompt_tokens": 926,
"completion_tokens": 43,
"total_tokens": 969
},
"model_name": "gpt-4"
},
"run": null
}
[chain/end] [1:RunTypeEnum.chain:AgentExecutor > 14:RunTypeEnum.chain:LLMChain] [3.52s] Exiting Chain run with output:
{
"text": "I now know the final answer\nFinal Answer: The director of the 2023 film Oppenheimer is Christopher Nolan and he is 52 years old. His age in days is approximately 18980 days."
}
[chain/end] [1:RunTypeEnum.chain:AgentExecutor] [21.96s] Exiting Chain run with output:
{
"output": "The director of the 2023 film Oppenheimer is Christopher Nolan and he is 52 years old. His age in days is approximately 18980 days."
}
'The director of the 2023 film Oppenheimer is Christopher Nolan and he is 52 years old. His age in days is approximately 18980 days.'
```
</CodeOutputBlock>
</details>
### `set_verbose(True)`
Setting the `verbose` flag will print out inputs and outputs in a slightly more readable format and will skip logging certain raw outputs (like the token usage stats for an LLM call) so that you can focus on application logic.
```python
fromlangchain.globalsimportset_verbose
set_verbose(True)
agent.run("Who directed the 2023 film Oppenheimer and what is their age? What is their age in days (assume 365 days per year)?")
```
<details> <summary>Console output</summary>
<CodeOutputBlock lang="python">
```
> Entering new AgentExecutor chain...
> Entering new LLMChain chain...
Prompt after formatting:
Answer the following questions as best you can. You have access to the following tools:
duckduckgo_search: A wrapper around DuckDuckGo Search. Useful for when you need to answer questions about current events. Input should be a search query.
Calculator: Useful for when you need to answer questions about math.
Use the following format:
Question: the input question you must answer
Thought: you should always think about what to do
Action: the action to take, should be one of [duckduckgo_search, Calculator]
Action Input: the input to the action
Observation: the result of the action
... (this Thought/Action/Action Input/Observation can repeat N times)
Thought: I now know the final answer
Final Answer: the final answer to the original input question
Begin!
Question: Who directed the 2023 film Oppenheimer and what is their age? What is their age in days (assume 365 days per year)?
Thought:
> Finished chain.
First, I need to find out who directed the film Oppenheimer in 2023 and their birth date to calculate their age.
Action: duckduckgo_search
Action Input: "Director of the 2023 film Oppenheimer"
Observation: Oppenheimer: Directed by Christopher Nolan. With Cillian Murphy, Emily Blunt, Robert Downey Jr., Alden Ehrenreich. The story of American scientist J. Robert Oppenheimer and his role in the development of the atomic bomb. In Christopher Nolan's new film, "Oppenheimer," Cillian Murphy stars as J. Robert ... 2023, 12:16 p.m. ET. ... including his role as the director of the Manhattan Engineer District, better ... J Robert Oppenheimer was the director of the secret Los Alamos Laboratory. It was established under US president Franklin D Roosevelt as part of the Manhattan Project to build the first atomic bomb. He oversaw the first atomic bomb detonation in the New Mexico desert in July 1945, code-named "Trinity". In this opening salvo of 2023's Oscar battle, Nolan has enjoined a star-studded cast for a retelling of the brilliant and haunted life of J. Robert Oppenheimer, the American physicist whose... Oppenheimer is a 2023 epic biographical thriller film written and directed by Christopher Nolan.It is based on the 2005 biography American Prometheus by Kai Bird and Martin J. Sherwin about J. Robert Oppenheimer, a theoretical physicist who was pivotal in developing the first nuclear weapons as part of the Manhattan Project and thereby ushering in the Atomic Age.
Thought:
> Entering new LLMChain chain...
Prompt after formatting:
Answer the following questions as best you can. You have access to the following tools:
duckduckgo_search: A wrapper around DuckDuckGo Search. Useful for when you need to answer questions about current events. Input should be a search query.
Calculator: Useful for when you need to answer questions about math.
Use the following format:
Question: the input question you must answer
Thought: you should always think about what to do
Action: the action to take, should be one of [duckduckgo_search, Calculator]
Action Input: the input to the action
Observation: the result of the action
... (this Thought/Action/Action Input/Observation can repeat N times)
Thought: I now know the final answer
Final Answer: the final answer to the original input question
Begin!
Question: Who directed the 2023 film Oppenheimer and what is their age? What is their age in days (assume 365 days per year)?
Thought:First, I need to find out who directed the film Oppenheimer in 2023 and their birth date to calculate their age.
Action: duckduckgo_search
Action Input: "Director of the 2023 film Oppenheimer"
Observation: Oppenheimer: Directed by Christopher Nolan. With Cillian Murphy, Emily Blunt, Robert Downey Jr., Alden Ehrenreich. The story of American scientist J. Robert Oppenheimer and his role in the development of the atomic bomb. In Christopher Nolan's new film, "Oppenheimer," Cillian Murphy stars as J. Robert ... 2023, 12:16 p.m. ET. ... including his role as the director of the Manhattan Engineer District, better ... J Robert Oppenheimer was the director of the secret Los Alamos Laboratory. It was established under US president Franklin D Roosevelt as part of the Manhattan Project to build the first atomic bomb. He oversaw the first atomic bomb detonation in the New Mexico desert in July 1945, code-named "Trinity". In this opening salvo of 2023's Oscar battle, Nolan has enjoined a star-studded cast for a retelling of the brilliant and haunted life of J. Robert Oppenheimer, the American physicist whose... Oppenheimer is a 2023 epic biographical thriller film written and directed by Christopher Nolan.It is based on the 2005 biography American Prometheus by Kai Bird and Martin J. Sherwin about J. Robert Oppenheimer, a theoretical physicist who was pivotal in developing the first nuclear weapons as part of the Manhattan Project and thereby ushering in the Atomic Age.
Thought:
> Finished chain.
The director of the 2023 film Oppenheimer is Christopher Nolan. Now I need to find out his birth date to calculate his age.
Action: duckduckgo_search
Action Input: "Christopher Nolan birth date"
Observation: July 30, 1970 (age 52) London England Notable Works: "Dunkirk" "Tenet" "The Prestige" See all related content → Recent News Jul. 13, 2023, 11:11 AM ET (AP) Cillian Murphy, playing Oppenheimer, finally gets to lead a Christopher Nolan film Christopher Edward Nolan CBE (born 30 July 1970) is a British and American filmmaker. Known for his Hollywood blockbusters with complex storytelling, Nolan is considered a leading filmmaker of the 21st century. His films have grossed $5 billion worldwide. The recipient of many accolades, he has been nominated for five Academy Awards, five BAFTA Awards and six Golden Globe Awards. Christopher Nolan is currently 52 according to his birthdate July 30, 1970 Sun Sign Leo Born Place Westminster, London, England, United Kingdom Residence Los Angeles, California, United States Nationality Education Chris attended Haileybury and Imperial Service College, in Hertford Heath, Hertfordshire. Christopher Nolan's next movie will study the man who developed the atomic bomb, J. Robert Oppenheimer. Here's the release date, plot, trailers & more. July 2023 sees the release of Christopher Nolan's new film, Oppenheimer, his first movie since 2020's Tenet and his split from Warner Bros. Billed as an epic thriller about "the man who ...
Thought:
> Entering new LLMChain chain...
Prompt after formatting:
Answer the following questions as best you can. You have access to the following tools:
duckduckgo_search: A wrapper around DuckDuckGo Search. Useful for when you need to answer questions about current events. Input should be a search query.
Calculator: Useful for when you need to answer questions about math.
Use the following format:
Question: the input question you must answer
Thought: you should always think about what to do
Action: the action to take, should be one of [duckduckgo_search, Calculator]
Action Input: the input to the action
Observation: the result of the action
... (this Thought/Action/Action Input/Observation can repeat N times)
Thought: I now know the final answer
Final Answer: the final answer to the original input question
Begin!
Question: Who directed the 2023 film Oppenheimer and what is their age? What is their age in days (assume 365 days per year)?
Thought:First, I need to find out who directed the film Oppenheimer in 2023 and their birth date to calculate their age.
Action: duckduckgo_search
Action Input: "Director of the 2023 film Oppenheimer"
Observation: Oppenheimer: Directed by Christopher Nolan. With Cillian Murphy, Emily Blunt, Robert Downey Jr., Alden Ehrenreich. The story of American scientist J. Robert Oppenheimer and his role in the development of the atomic bomb. In Christopher Nolan's new film, "Oppenheimer," Cillian Murphy stars as J. Robert ... 2023, 12:16 p.m. ET. ... including his role as the director of the Manhattan Engineer District, better ... J Robert Oppenheimer was the director of the secret Los Alamos Laboratory. It was established under US president Franklin D Roosevelt as part of the Manhattan Project to build the first atomic bomb. He oversaw the first atomic bomb detonation in the New Mexico desert in July 1945, code-named "Trinity". In this opening salvo of 2023's Oscar battle, Nolan has enjoined a star-studded cast for a retelling of the brilliant and haunted life of J. Robert Oppenheimer, the American physicist whose... Oppenheimer is a 2023 epic biographical thriller film written and directed by Christopher Nolan.It is based on the 2005 biography American Prometheus by Kai Bird and Martin J. Sherwin about J. Robert Oppenheimer, a theoretical physicist who was pivotal in developing the first nuclear weapons as part of the Manhattan Project and thereby ushering in the Atomic Age.
Thought:The director of the 2023 film Oppenheimer is Christopher Nolan. Now I need to find out his birth date to calculate his age.
Action: duckduckgo_search
Action Input: "Christopher Nolan birth date"
Observation: July 30, 1970 (age 52) London England Notable Works: "Dunkirk" "Tenet" "The Prestige" See all related content → Recent News Jul. 13, 2023, 11:11 AM ET (AP) Cillian Murphy, playing Oppenheimer, finally gets to lead a Christopher Nolan film Christopher Edward Nolan CBE (born 30 July 1970) is a British and American filmmaker. Known for his Hollywood blockbusters with complex storytelling, Nolan is considered a leading filmmaker of the 21st century. His films have grossed $5 billion worldwide. The recipient of many accolades, he has been nominated for five Academy Awards, five BAFTA Awards and six Golden Globe Awards. Christopher Nolan is currently 52 according to his birthdate July 30, 1970 Sun Sign Leo Born Place Westminster, London, England, United Kingdom Residence Los Angeles, California, United States Nationality Education Chris attended Haileybury and Imperial Service College, in Hertford Heath, Hertfordshire. Christopher Nolan's next movie will study the man who developed the atomic bomb, J. Robert Oppenheimer. Here's the release date, plot, trailers & more. July 2023 sees the release of Christopher Nolan's new film, Oppenheimer, his first movie since 2020's Tenet and his split from Warner Bros. Billed as an epic thriller about "the man who ...
Thought:
> Finished chain.
Christopher Nolan was born on July 30, 1970. Now I need to calculate his age in 2023 and then convert it into days.
Action: Calculator
Action Input: (2023 - 1970) * 365
> Entering new LLMMathChain chain...
(2023 - 1970) * 365
> Entering new LLMChain chain...
Prompt after formatting:
Translate a math problem into a expression that can be executed using Python's numexpr library. Use the output of running this code to answer the question.
Question: ${Question with math problem.}
```text
${single line mathematical expression that solves the problem}
```
...numexpr.evaluate(text)...
```output
${Output of running the code}
```
Answer: ${Answer}
Begin.
Question: What is 37593 * 67?
```text
37593 * 67
```
...numexpr.evaluate("37593 * 67")...
```output
2518731
```
Answer: 2518731
Question: 37593^(1/5)
```text
37593**(1/5)
```
...numexpr.evaluate("37593**(1/5)")...
```output
8.222831614237718
```
Answer: 8.222831614237718
Question: (2023 - 1970) * 365
> Finished chain.
```text
(2023 - 1970) * 365
```
...numexpr.evaluate("(2023 - 1970) * 365")...
Answer: 19345
> Finished chain.
Observation: Answer: 19345
Thought:
> Entering new LLMChain chain...
Prompt after formatting:
Answer the following questions as best you can. You have access to the following tools:
duckduckgo_search: A wrapper around DuckDuckGo Search. Useful for when you need to answer questions about current events. Input should be a search query.
Calculator: Useful for when you need to answer questions about math.
Use the following format:
Question: the input question you must answer
Thought: you should always think about what to do
Action: the action to take, should be one of [duckduckgo_search, Calculator]
Action Input: the input to the action
Observation: the result of the action
... (this Thought/Action/Action Input/Observation can repeat N times)
Thought: I now know the final answer
Final Answer: the final answer to the original input question
Begin!
Question: Who directed the 2023 film Oppenheimer and what is their age? What is their age in days (assume 365 days per year)?
Thought:First, I need to find out who directed the film Oppenheimer in 2023 and their birth date to calculate their age.
Action: duckduckgo_search
Action Input: "Director of the 2023 film Oppenheimer"
Observation: Oppenheimer: Directed by Christopher Nolan. With Cillian Murphy, Emily Blunt, Robert Downey Jr., Alden Ehrenreich. The story of American scientist J. Robert Oppenheimer and his role in the development of the atomic bomb. In Christopher Nolan's new film, "Oppenheimer," Cillian Murphy stars as J. Robert ... 2023, 12:16 p.m. ET. ... including his role as the director of the Manhattan Engineer District, better ... J Robert Oppenheimer was the director of the secret Los Alamos Laboratory. It was established under US president Franklin D Roosevelt as part of the Manhattan Project to build the first atomic bomb. He oversaw the first atomic bomb detonation in the New Mexico desert in July 1945, code-named "Trinity". In this opening salvo of 2023's Oscar battle, Nolan has enjoined a star-studded cast for a retelling of the brilliant and haunted life of J. Robert Oppenheimer, the American physicist whose... Oppenheimer is a 2023 epic biographical thriller film written and directed by Christopher Nolan.It is based on the 2005 biography American Prometheus by Kai Bird and Martin J. Sherwin about J. Robert Oppenheimer, a theoretical physicist who was pivotal in developing the first nuclear weapons as part of the Manhattan Project and thereby ushering in the Atomic Age.
Thought:The director of the 2023 film Oppenheimer is Christopher Nolan. Now I need to find out his birth date to calculate his age.
Action: duckduckgo_search
Action Input: "Christopher Nolan birth date"
Observation: July 30, 1970 (age 52) London England Notable Works: "Dunkirk" "Tenet" "The Prestige" See all related content → Recent News Jul. 13, 2023, 11:11 AM ET (AP) Cillian Murphy, playing Oppenheimer, finally gets to lead a Christopher Nolan film Christopher Edward Nolan CBE (born 30 July 1970) is a British and American filmmaker. Known for his Hollywood blockbusters with complex storytelling, Nolan is considered a leading filmmaker of the 21st century. His films have grossed $5 billion worldwide. The recipient of many accolades, he has been nominated for five Academy Awards, five BAFTA Awards and six Golden Globe Awards. Christopher Nolan is currently 52 according to his birthdate July 30, 1970 Sun Sign Leo Born Place Westminster, London, England, United Kingdom Residence Los Angeles, California, United States Nationality Education Chris attended Haileybury and Imperial Service College, in Hertford Heath, Hertfordshire. Christopher Nolan's next movie will study the man who developed the atomic bomb, J. Robert Oppenheimer. Here's the release date, plot, trailers & more. July 2023 sees the release of Christopher Nolan's new film, Oppenheimer, his first movie since 2020's Tenet and his split from Warner Bros. Billed as an epic thriller about "the man who ...
Thought:Christopher Nolan was born on July 30, 1970. Now I need to calculate his age in 2023 and then convert it into days.
Action: Calculator
Action Input: (2023 - 1970) * 365
Observation: Answer: 19345
Thought:
> Finished chain.
I now know the final answer
Final Answer: The director of the 2023 film Oppenheimer is Christopher Nolan and he is 53 years old in 2023. His age in days is 19345 days.
> Finished chain.
'The director of the 2023 film Oppenheimer is Christopher Nolan and he is 53 years old in 2023. His age in days is 19345 days.'
```
</CodeOutputBlock>
</details>
### `Chain(..., verbose=True)`
You can also scope verbosity down to a single object, in which case only the inputs and outputs to that object are printed (along with any additional callbacks calls made specifically by that object).
```python
# Passing verbose=True to initialize_agent will pass that along to the AgentExecutor (which is a Chain).
agent=initialize_agent(
tools,
llm,
agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION,
verbose=True,
)
agent.run("Who directed the 2023 film Oppenheimer and what is their age? What is their age in days (assume 365 days per year)?")
```
<details> <summary>Console output</summary>
<CodeOutputBlock lang="python">
```
> Entering new AgentExecutor chain...
First, I need to find out who directed the film Oppenheimer in 2023 and their birth date. Then, I can calculate their age in years and days.
Action: duckduckgo_search
Action Input: "Director of 2023 film Oppenheimer"
Observation: Oppenheimer: Directed by Christopher Nolan. With Cillian Murphy, Emily Blunt, Robert Downey Jr., Alden Ehrenreich. The story of American scientist J. Robert Oppenheimer and his role in the development of the atomic bomb. In Christopher Nolan's new film, "Oppenheimer," Cillian Murphy stars as J. Robert Oppenheimer, the American physicist who oversaw the Manhattan Project in Los Alamos, N.M. Universal Pictures... J Robert Oppenheimer was the director of the secret Los Alamos Laboratory. It was established under US president Franklin D Roosevelt as part of the Manhattan Project to build the first atomic bomb. He oversaw the first atomic bomb detonation in the New Mexico desert in July 1945, code-named "Trinity". A Review of Christopher Nolan's new film 'Oppenheimer' , the story of the man who fathered the Atomic Bomb. Cillian Murphy leads an all star cast ... Release Date: July 21, 2023. Director ... For his new film, "Oppenheimer," starring Cillian Murphy and Emily Blunt, director Christopher Nolan set out to build an entire 1940s western town.
Thought:The director of the 2023 film Oppenheimer is Christopher Nolan. Now I need to find out his birth date to calculate his age.
Action: duckduckgo_search
Action Input: "Christopher Nolan birth date"
Observation: July 30, 1970 (age 52) London England Notable Works: "Dunkirk" "Tenet" "The Prestige" See all related content → Recent News Jul. 13, 2023, 11:11 AM ET (AP) Cillian Murphy, playing Oppenheimer, finally gets to lead a Christopher Nolan film Christopher Edward Nolan CBE (born 30 July 1970) is a British and American filmmaker. Known for his Hollywood blockbusters with complex storytelling, Nolan is considered a leading filmmaker of the 21st century. His films have grossed $5 billion worldwide. The recipient of many accolades, he has been nominated for five Academy Awards, five BAFTA Awards and six Golden Globe Awards. Christopher Nolan is currently 52 according to his birthdate July 30, 1970 Sun Sign Leo Born Place Westminster, London, England, United Kingdom Residence Los Angeles, California, United States Nationality Education Chris attended Haileybury and Imperial Service College, in Hertford Heath, Hertfordshire. Christopher Nolan's next movie will study the man who developed the atomic bomb, J. Robert Oppenheimer. Here's the release date, plot, trailers & more. Date of Birth: 30 July 1970 . ... Christopher Nolan is a British-American film director, producer, and screenwriter. His films have grossed more than US$5 billion worldwide, and have garnered 11 Academy Awards from 36 nominations. ...
Thought:Christopher Nolan was born on July 30, 1970. Now I can calculate his age in years and then in days.
Final Answer: The director of the 2023 film Oppenheimer is Christopher Nolan. He is 53 years old in 2023, which is approximately 19345 days.
> Finished chain.
'The director of the 2023 film Oppenheimer is Christopher Nolan. He is 53 years old in 2023, which is approximately 19345 days.'
```
</CodeOutputBlock>
</details>
## Other callbacks
`Callbacks` are what we use to execute any functionality within a component outside the primary component logic. All of the above solutions use `Callbacks` under the hood to log intermediate steps of components. There are a number of `Callbacks` relevant for debugging that come with LangChain out of the box, like the [FileCallbackHandler](/docs/modules/callbacks/filecallbackhandler). You can also implement your own callbacks to execute custom functionality.
See here for more info on [Callbacks](/docs/modules/callbacks/), how to use them, and customize them.
Extending LangChain's base abstractions, whether you're planning to contribute back to the open-source repo or build a bespoke internal integration, is encouraged.
Check out these guides for building your own custom classes for the following modules:
- [Chat models](/docs/modules/model_io/chat/custom_chat_model) for interfacing with chat-tuned language models.
- [LLMs](/docs/modules/model_io/llms/custom_llm) for interfacing with text language models.
- [Output parsers](/docs/modules/model_io/output_parsers/custom) for handling language model outputs.
"1. `Base model`: What is the base-model and how was it trained?\n",
"2. `Fine-tuning approach`: Was the base-model fine-tuned and, if so, what [set of instructions](https://cameronrwolfe.substack.com/p/beyond-llama-the-power-of-open-llms#%C2%A7alpaca-an-instruction-following-llama-model) was used?\n",
- Pydantic v2 was released in June, 2023 (https://docs.pydantic.dev/2.0/blog/pydantic-v2-final/)
- v2 contains has a number of breaking changes (https://docs.pydantic.dev/2.0/migration/)
- Pydantic v2 and v1 are under the same package name, so both versions cannot be installed at the same time
## LangChain Pydantic migration plan
As of `langchain>=0.0.267`, LangChain will allow users to install either Pydantic V1 or V2.
* Internally LangChain will continue to [use V1](https://docs.pydantic.dev/latest/migration/#continue-using-pydantic-v1-features).
* During this time, users can pin their pydantic version to v1 to avoid breaking changes, or start a partial
migration using pydantic v2 throughout their code, but avoiding mixing v1 and v2 code for LangChain (see below).
User can either pin to pydantic v1, and upgrade their code in one go once LangChain has migrated to v2 internally, or they can start a partial migration to v2, but must avoid mixing v1 and v2 code for LangChain.
Below are two examples of showing how to avoid mixing pydantic v1 and v2 code in
the case of inheritance and in the case of passing objects to LangChain.
**Example 1: Extending via inheritance**
**YES**
```python
frompydantic.v1importroot_validator,validator
classCustomTool(BaseTool):# BaseTool is v1 code
x:int=Field(default=1)
def_run(*args,**kwargs):
return"hello"
@validator('x')# v1 code
@classmethod
defvalidate_x(cls,x:int)->int:
return1
CustomTool(
name='custom_tool',
description="hello",
x=1,
)
```
Mixing Pydantic v2 primitives with Pydantic v1 primitives can raise cryptic errors
In today's fast-paced technological landscape, the use of Large Language Models (LLMs) is rapidly expanding. As a result, it is crucial for developers to understand how to effectively deploy these models in production environments. LLM interfaces typically fall into two categories:
In this scenario, most of the computational burden is handled by the LLM providers, while LangChain simplifies the implementation of business logic around these services. This approach includes features such as prompt templating, chat message generation, caching, vector embedding database creation, preprocessing, etc.
- **Case 2: Self-hosted Open-Source Models**
Alternatively, developers can opt to use smaller, yet comparably capable, self-hosted open-source LLM models. This approach can significantly decrease costs, latency, and privacy concerns associated with transferring data to external LLM providers.
Regardless of the framework that forms the backbone of your product, deploying LLM applications comes with its own set of challenges. It's vital to understand the trade-offs and key considerations when evaluating serving frameworks.
## Outline
This guide aims to provide a comprehensive overview of the requirements for deploying LLMs in a production setting, focusing on:
- **Designing a Robust LLM Application Service**
- **Maintaining Cost-Efficiency**
- **Ensuring Rapid Iteration**
Understanding these components is crucial when assessing serving systems. LangChain integrates with several open-source projects designed to tackle these issues, providing a robust framework for productionizing your LLM applications. Some notable frameworks include:
These links will provide further information on each ecosystem, assisting you in finding the best fit for your LLM deployment needs.
## Designing a Robust LLM Application Service
When deploying an LLM service in production, it's imperative to provide a seamless user experience free from outages. Achieving 24/7 service availability involves creating and maintaining several sub-systems surrounding your application.
### Monitoring
Monitoring forms an integral part of any system running in a production environment. In the context of LLMs, it is essential to monitor both performance and quality metrics.
**Performance Metrics:** These metrics provide insights into the efficiency and capacity of your model. Here are some key examples:
- Query per second (QPS): This measures the number of queries your model processes in a second, offering insights into its utilization.
- Latency: This metric quantifies the delay from when your client sends a request to when they receive a response.
- Tokens Per Second (TPS): This represents the number of tokens your model can generate in a second.
**Quality Metrics:** These metrics are typically customized according to the business use-case. For instance, how does the output of your system compare to a baseline, such as a previous version? Although these metrics can be calculated offline, you need to log the necessary data to use them later.
### Fault tolerance
Your application may encounter errors such as exceptions in your model inference or business logic code, causing failures and disrupting traffic. Other potential issues could arise from the machine running your application, such as unexpected hardware breakdowns or loss of spot-instances during high-demand periods. One way to mitigate these risks is by increasing redundancy through replica scaling and implementing recovery mechanisms for failed replicas. However, model replicas aren't the only potential points of failure. It's essential to build resilience against various failures that could occur at any point in your stack.
### Zero down time upgrade
System upgrades are often necessary but can result in service disruptions if not handled correctly. One way to prevent downtime during upgrades is by implementing a smooth transition process from the old version to the new one. Ideally, the new version of your LLM service is deployed, and traffic gradually shifts from the old to the new version, maintaining a constant QPS throughout the process.
### Load balancing
Load balancing, in simple terms, is a technique to distribute work evenly across multiple computers, servers, or other resources to optimize the utilization of the system, maximize throughput, minimize response time, and avoid overload of any single resource. Think of it as a traffic officer directing cars (requests) to different roads (servers) so that no single road becomes too congested.
There are several strategies for load balancing. For example, one common method is the *Round Robin* strategy, where each request is sent to the next server in line, cycling back to the first when all servers have received a request. This works well when all servers are equally capable. However, if some servers are more powerful than others, you might use a *Weighted Round Robin* or *Least Connections* strategy, where more requests are sent to the more powerful servers, or to those currently handling the fewest active requests. Let's imagine you're running a LLM chain. If your application becomes popular, you could have hundreds or even thousands of users asking questions at the same time. If one server gets too busy (high load), the load balancer would direct new requests to another server that is less busy. This way, all your users get a timely response and the system remains stable.
## Maintaining Cost-Efficiency and Scalability
Deploying LLM services can be costly, especially when you're handling a large volume of user interactions. Charges by LLM providers are usually based on tokens used, making a chat system inference on these models potentially expensive. However, several strategies can help manage these costs without compromising the quality of the service.
### Self-hosting models
Several smaller and open-source LLMs are emerging to tackle the issue of reliance on LLM providers. Self-hosting allows you to maintain similar quality to LLM provider models while managing costs. The challenge lies in building a reliable, high-performing LLM serving system on your own machines.
### Resource Management and Auto-Scaling
Computational logic within your application requires precise resource allocation. For instance, if part of your traffic is served by an OpenAI endpoint and another part by a self-hosted model, it's crucial to allocate suitable resources for each. Auto-scaling—adjusting resource allocation based on traffic—can significantly impact the cost of running your application. This strategy requires a balance between cost and responsiveness, ensuring neither resource over-provisioning nor compromised application responsiveness.
### Utilizing Spot Instances
On platforms like AWS, spot instances offer substantial cost savings, typically priced at about a third of on-demand instances. The trade-off is a higher crash rate, necessitating a robust fault-tolerance mechanism for effective use.
### Independent Scaling
When self-hosting your models, you should consider independent scaling. For example, if you have two translation models, one fine-tuned for French and another for Spanish, incoming requests might necessitate different scaling requirements for each.
### Batching requests
In the context of Large Language Models, batching requests can enhance efficiency by better utilizing your GPU resources. GPUs are inherently parallel processors, designed to handle multiple tasks simultaneously. If you send individual requests to the model, the GPU might not be fully utilized as it's only working on a single task at a time. On the other hand, by batching requests together, you're allowing the GPU to work on multiple tasks at once, maximizing its utilization and improving inference speed. This not only leads to cost savings but can also improve the overall latency of your LLM service.
In summary, managing costs while scaling your LLM services requires a strategic approach. Utilizing self-hosting models, managing resources effectively, employing auto-scaling, using spot instances, independently scaling models, and batching requests are key strategies to consider. Open-source libraries such as Ray Serve and BentoML are designed to deal with these complexities.
## Ensuring Rapid Iteration
The LLM landscape is evolving at an unprecedented pace, with new libraries and model architectures being introduced constantly. Consequently, it's crucial to avoid tying yourself to a solution specific to one particular framework. This is especially relevant in serving, where changes to your infrastructure can be time-consuming, expensive, and risky. Strive for infrastructure that is not locked into any specific machine learning library or framework, but instead offers a general-purpose, scalable serving layer. Here are some aspects where flexibility plays a key role:
### Model composition
Deploying systems like LangChain demands the ability to piece together different models and connect them via logic. Take the example of building a natural language input SQL query engine. Querying an LLM and obtaining the SQL command is only part of the system. You need to extract metadata from the connected database, construct a prompt for the LLM, run the SQL query on an engine, collect and feedback the response to the LLM as the query runs, and present the results to the user. This demonstrates the need to seamlessly integrate various complex components built in Python into a dynamic chain of logical blocks that can be served together.
## Cloud providers
Many hosted solutions are restricted to a single cloud provider, which can limit your options in today's multi-cloud world. Depending on where your other infrastructure components are built, you might prefer to stick with your chosen cloud provider.
## Infrastructure as Code (IaC)
Rapid iteration also involves the ability to recreate your infrastructure quickly and reliably. This is where Infrastructure as Code (IaC) tools like Terraform, CloudFormation, or Kubernetes YAML files come into play. They allow you to define your infrastructure in code files, which can be version controlled and quickly deployed, enabling faster and more reliable iterations.
## CI/CD
In a fast-paced environment, implementing CI/CD pipelines can significantly speed up the iteration process. They help automate the testing and deployment of your LLM applications, reducing the risk of errors and enabling faster feedback and iteration.
"[](https://colab.research.google.com/github/langchain-ai/langchain/blob/master/docs/docs/guides/evaluation/comparison/custom.ipynb)\n",
"\n",
"You can make your own pairwise string evaluators by inheriting from `PairwiseStringEvaluator` class and overwriting the `_evaluate_string_pairs` method (and the `_aevaluate_string_pairs` method if you want to use the evaluator asynchronously).\n",
"\n",
"In this example, you will make a simple custom evaluator that just returns whether the first prediction has more whitespace tokenized 'words' than the second.\n",
"\n",
"You can check out the reference docs for the [PairwiseStringEvaluator interface](https://api.python.langchain.com/en/latest/evaluation/langchain.evaluation.schema.PairwiseStringEvaluator.html#langchain.evaluation.schema.PairwiseStringEvaluator) for more info.\n"
" prediction=\"The quick brown fox jumped over the lazy dog.\",\n",
" prediction_b=\"The quick brown fox jumped over the dog.\",\n",
")"
]
},
{
"cell_type": "markdown",
"id": "d90f128f-6f49-42a1-b05a-3aea568ee03b",
"metadata": {},
"source": [
"## LLM-Based Example\n",
"\n",
"That example was simple to illustrate the API, but it wasn't very useful in practice. Below, use an LLM with some custom instructions to form a simple preference scorer similar to the built-in [PairwiseStringEvalChain](https://api.python.langchain.com/en/latest/evaluation/langchain.evaluation.comparison.eval_chain.PairwiseStringEvalChain.html#langchain.evaluation.comparison.eval_chain.PairwiseStringEvalChain). We will use `ChatAnthropic` for the evaluator chain."
" \"\"\"Which option is preferred? Do not take order into account. Evaluate based on accuracy and helpfulness. If neither is preferred, respond with C. Provide your reasoning, then finish with Preference: A/B/C\n",
"\n",
"Input: How do I get the path of the parent directory in python 3.8?\n",
"Reasoning: Both options return the same result. However, since option B is more concise and easily understand, it is preferred.\n",
"Preference: B\n",
"\n",
"Which option is preferred? Do not take order into account. Evaluate based on accuracy and helpfulness. If neither is preferred, respond with C. Provide your reasoning, then finish with Preference: A/B/C\n",
"{'reasoning': 'Option B is preferred over option A for importing from a relative directory, because it is more straightforward and concise.\\n\\nOption A uses the importlib module, which allows importing a module by specifying the full name as a string. While this works, it is less clear compared to option B.\\n\\nOption B directly imports from the relative path using dot notation, which clearly shows that it is a relative import. This is the recommended way to do relative imports in Python.\\n\\nIn summary, option B is more accurate and helpful as it uses the standard Python relative import syntax.',\n",
" 'value': 'B',\n",
" 'score': 0.0}"
]
},
"execution_count": 7,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"evaluator.evaluate_string_pairs(\n",
" input=\"How do I import from a relative directory?\",\n",
Comparison evaluators in LangChain help measure two different chains or LLM outputs. These evaluators are helpful for comparative analyses, such as A/B testing between two language models, or comparing different versions of the same model. They can also be useful for things like generating preference scores for ai-assisted reinforcement learning.
These evaluators inherit from the `PairwiseStringEvaluator` class, providing a comparison interface for two strings - typically, the outputs from two different prompts or models, or two versions of the same model. In essence, a comparison evaluator performs an evaluation on a pair of strings and returns a dictionary containing the evaluation score and other relevant details.
To create a custom comparison evaluator, inherit from the `PairwiseStringEvaluator` class and overwrite the `_evaluate_string_pairs` method. If you require asynchronous evaluation, also overwrite the `_aevaluate_string_pairs` method.
Here's a summary of the key methods and properties of a comparison evaluator:
- `evaluate_string_pairs`: Evaluate the output string pairs. This function should be overwritten when creating custom evaluators.
- `aevaluate_string_pairs`: Asynchronously evaluate the output string pairs. This function should be overwritten for asynchronous evaluation.
- `requires_input`: This property indicates whether this evaluator requires an input string.
- `requires_reference`: This property specifies whether this evaluator requires a reference label.
:::note LangSmith Support
The [run_on_dataset](https://api.python.langchain.com/en/latest/langchain_api_reference.html#module-langchain.smith) evaluation method is designed to evaluate only a single model at a time, and thus, doesn't support these evaluators.
:::
Detailed information about creating custom evaluators and the available built-in comparison evaluators is provided in the following sections.
"[](https://colab.research.google.com/github/langchain-ai/langchain/blob/master/docs/docs/guides/evaluation/comparison/pairwise_embedding_distance.ipynb)\n",
"\n",
"One way to measure the similarity (or dissimilarity) between two predictions on a shared or similar input is to embed the predictions and compute a vector distance between the two embeddings.<a name=\"cite_ref-1\"></a>[<sup>[1]</sup>](#cite_note-1)\n",
"\n",
"You can load the `pairwise_embedding_distance` evaluator to do this.\n",
"\n",
"**Note:** This returns a **distance** score, meaning that the lower the number, the **more** similar the outputs are, according to their embedded representation.\n",
"\n",
"Check out the reference docs for the [PairwiseEmbeddingDistanceEvalChain](https://api.python.langchain.com/en/latest/evaluation/langchain.evaluation.embedding_distance.base.PairwiseEmbeddingDistanceEvalChain.html#langchain.evaluation.embedding_distance.base.PairwiseEmbeddingDistanceEvalChain) for more info."
" prediction=\"Seattle is hot in June\", prediction_b=\"Seattle is cool in June.\"\n",
")"
]
},
{
"cell_type": "code",
"execution_count": 12,
"metadata": {
"tags": []
},
"outputs": [
{
"data": {
"text/plain": [
"{'score': 0.21018880025138598}"
]
},
"execution_count": 12,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"hf_evaluator.evaluate_string_pairs(\n",
" prediction=\"Seattle is warm in June\", prediction_b=\"Seattle is cool in June.\"\n",
")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<a name=\"cite_note-1\"></a><i>1. Note: When it comes to semantic similarity, this often gives better results than older string distance metrics (such as those in the `PairwiseStringDistanceEvalChain`), though it tends to be less reliable than evaluators that use the LLM directly (such as the `PairwiseStringEvalChain`) </i>"
"[](https://colab.research.google.com/github/langchain-ai/langchain/blob/master/docs/docs/guides/evaluation/comparison/pairwise_string.ipynb)\n",
"\n",
"Often you will want to compare predictions of an LLM, Chain, or Agent for a given input. The `StringComparison` evaluators facilitate this so you can answer questions like:\n",
"\n",
"- Which LLM or prompt produces a preferred output for a given question?\n",
"- Which examples should I include for few-shot example selection?\n",
"- Which output is better to include for fine-tuning?\n",
"\n",
"The simplest and often most reliable automated way to choose a preferred prediction for a given input is to use the `pairwise_string` evaluator.\n",
"\n",
"Check out the reference docs for the [PairwiseStringEvalChain](https://api.python.langchain.com/en/latest/evaluation/langchain.evaluation.comparison.eval_chain.PairwiseStringEvalChain.html#langchain.evaluation.comparison.eval_chain.PairwiseStringEvalChain) for more info."
"{'reasoning': 'Both responses are relevant to the question asked, as they both provide a numerical answer to the question about the number of dogs in the park. However, Response A is incorrect according to the reference answer, which states that there are four dogs. Response B, on the other hand, is correct as it matches the reference answer. Neither response demonstrates depth of thought, as they both simply provide a numerical answer without any additional information or context. \\n\\nBased on these criteria, Response B is the better response.\\n',\n",
" 'value': 'B',\n",
" 'score': 0}"
]
},
"execution_count": 2,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"evaluator.evaluate_string_pairs(\n",
" prediction=\"there are three dogs\",\n",
" prediction_b=\"4\",\n",
" input=\"how many dogs are in the park?\",\n",
" reference=\"four\",\n",
")"
]
},
{
"cell_type": "markdown",
"id": "7491d2e6-4e77-4b17-be6b-7da966785c1d",
"metadata": {},
"source": [
"## Methods\n",
"\n",
"\n",
"The pairwise string evaluator can be called using [evaluate_string_pairs](https://api.python.langchain.com/en/latest/evaluation/langchain.evaluation.comparison.eval_chain.PairwiseStringEvalChain.html#langchain.evaluation.comparison.eval_chain.PairwiseStringEvalChain.evaluate_string_pairs) (or async [aevaluate_string_pairs](https://api.python.langchain.com/en/latest/evaluation/langchain.evaluation.comparison.eval_chain.PairwiseStringEvalChain.html#langchain.evaluation.comparison.eval_chain.PairwiseStringEvalChain.aevaluate_string_pairs)) methods, which accept:\n",
"\n",
"- prediction (str) – The predicted response of the first model, chain, or prompt.\n",
"- prediction_b (str) – The predicted response of the second model, chain, or prompt.\n",
"- input (str) – The input question, prompt, or other text.\n",
"- reference (str) – (Only for the labeled_pairwise_string variant) The reference response.\n",
"\n",
"They return a dictionary with the following values:\n",
"\n",
"- value: 'A' or 'B', indicating whether `prediction` or `prediction_b` is preferred, respectively\n",
"- score: Integer 0 or 1 mapped from the 'value', where a score of 1 would mean that the first `prediction` is preferred, and a score of 0 would mean `prediction_b` is preferred.\n",
"- reasoning: String \"chain of thought reasoning\" from the LLM generated prior to creating the score"
]
},
{
"cell_type": "markdown",
"id": "ed353b93-be71-4479-b9c0-8c97814c2e58",
"metadata": {},
"source": [
"## Without References\n",
"\n",
"When references aren't available, you can still predict the preferred response.\n",
"The results will reflect the evaluation model's preference, which is less reliable and may result\n",
"{'reasoning': 'Both responses are correct and relevant to the question. However, Response B is more helpful and insightful as it provides a more detailed explanation of what addition is. Response A is correct but lacks depth as it does not explain what the operation of addition entails. \\n\\nFinal Decision: [[B]]',\n",
" 'value': 'B',\n",
" 'score': 0}"
]
},
"execution_count": 4,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"evaluator.evaluate_string_pairs(\n",
" prediction=\"Addition is a mathematical operation.\",\n",
" prediction_b=\"Addition is a mathematical operation that adds two numbers to create a third number, the 'sum'.\",\n",
" input=\"What is addition?\",\n",
")"
]
},
{
"cell_type": "markdown",
"id": "4a09b21d-9851-47e8-93d3-90044b2945b0",
"metadata": {
"tags": []
},
"source": [
"## Defining the Criteria\n",
"\n",
"By default, the LLM is instructed to select the 'preferred' response based on helpfulness, relevance, correctness, and depth of thought. You can customize the criteria by passing in a `criteria` argument, where the criteria could take any of the following forms:\n",
"\n",
"- [`Criteria`](https://api.python.langchain.com/en/latest/evaluation/langchain.evaluation.criteria.eval_chain.Criteria.html#langchain.evaluation.criteria.eval_chain.Criteria) enum or its string value - to use one of the default criteria and their descriptions\n",
"- [Constitutional principal](https://api.python.langchain.com/en/latest/chains/langchain.chains.constitutional_ai.models.ConstitutionalPrinciple.html#langchain.chains.constitutional_ai.models.ConstitutionalPrinciple) - use one any of the constitutional principles defined in langchain\n",
"- Dictionary: a list of custom criteria, where the key is the name of the criteria, and the value is the description.\n",
"- A list of criteria or constitutional principles - to combine multiple criteria in one.\n",
"\n",
"Below is an example for determining preferred writing responses based on a custom style."
]
},
{
"cell_type": "code",
"execution_count": 5,
"id": "8539e7d9-f7b0-4d32-9c45-593a7915c093",
"metadata": {
"tags": []
},
"outputs": [],
"source": [
"custom_criteria = {\n",
" \"simplicity\": \"Is the language straightforward and unpretentious?\",\n",
" \"clarity\": \"Are the sentences clear and easy to understand?\",\n",
" \"precision\": \"Is the writing precise, with no unnecessary words or details?\",\n",
" \"truthfulness\": \"Does the writing feel honest and sincere?\",\n",
" \"subtext\": \"Does the writing suggest deeper meanings or themes?\",\n",
"{'reasoning': 'Response A is simple, clear, and precise. It uses straightforward language to convey a deep and sincere message about families. The metaphor of joy and sorrow as music is effective and easy to understand.\\n\\nResponse B, on the other hand, is more complex and less clear. The language is more pretentious, with words like \"domicile,\" \"resounds,\" \"abode,\" \"dissonant,\" and \"elegy.\" While it conveys a similar message to Response A, it does so in a more convoluted way. The precision is also lacking due to the use of unnecessary words and details.\\n\\nBoth responses suggest deeper meanings or themes about the shared joy and unique sorrow in families. However, Response A does so in a more effective and accessible way.\\n\\nTherefore, the better response is [[A]].',\n",
" 'value': 'A',\n",
" 'score': 1}"
]
},
"execution_count": 6,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"evaluator.evaluate_string_pairs(\n",
" prediction=\"Every cheerful household shares a similar rhythm of joy; but sorrow, in each household, plays a unique, haunting melody.\",\n",
" prediction_b=\"Where one finds a symphony of joy, every domicile of happiness resounds in harmonious,\"\n",
" \" identical notes; yet, every abode of despair conducts a dissonant orchestra, each\"\n",
" \" playing an elegy of grief that is peculiar and profound to its own existence.\",\n",
" input=\"Write some prose about families.\",\n",
")"
]
},
{
"cell_type": "markdown",
"id": "a25b60b2-627c-408a-be4b-a2e5cbc10726",
"metadata": {},
"source": [
"## Customize the LLM\n",
"\n",
"By default, the loader uses `gpt-4` in the evaluation chain. You can customize this when loading."
"{'reasoning': 'Here is my assessment:\\n\\nResponse B is more helpful, insightful, and accurate than Response A. Response B simply states \"4\", which directly answers the question by providing the exact number of dogs mentioned in the reference answer. In contrast, Response A states \"there are three dogs\", which is incorrect according to the reference answer. \\n\\nIn terms of helpfulness, Response B gives the precise number while Response A provides an inaccurate guess. For relevance, both refer to dogs in the park from the question. However, Response B is more correct and factual based on the reference answer. Response A shows some attempt at reasoning but is ultimately incorrect. Response B requires less depth of thought to simply state the factual number.\\n\\nIn summary, Response B is superior in terms of helpfulness, relevance, correctness, and depth. My final decision is: [[B]]\\n',\n",
" 'value': 'B',\n",
" 'score': 0}"
]
},
"execution_count": 8,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"evaluator.evaluate_string_pairs(\n",
" prediction=\"there are three dogs\",\n",
" prediction_b=\"4\",\n",
" input=\"how many dogs are in the park?\",\n",
" reference=\"four\",\n",
")"
]
},
{
"cell_type": "markdown",
"id": "e0e89c13-d0ad-4f87-8fcb-814399bafa2a",
"metadata": {},
"source": [
"## Customize the Evaluation Prompt\n",
"\n",
"You can use your own custom evaluation prompt to add more task-specific instructions or to instruct the evaluator to score the output.\n",
"\n",
"*Note: If you use a prompt that expects generates a result in a unique format, you may also have to pass in a custom output parser (`output_parser=your_parser()`) instead of the default `PairwiseStringResultOutputParser`"
"input_variables=['prediction', 'reference', 'prediction_b', 'input'] output_parser=None partial_variables={'criteria': 'helpfulness: Is the submission helpful, insightful, and appropriate?\\nrelevance: Is the submission referring to a real quote from the text?\\ncorrectness: Is the submission correct, accurate, and factual?\\ndepth: Does the submission demonstrate depth of thought?'} template='Given the input context, which do you prefer: A or B?\\nEvaluate based on the following criteria:\\n{criteria}\\nReason step by step and finally, respond with either [[A]] or [[B]] on its own line.\\n\\nDATA\\n----\\ninput: {input}\\nreference: {reference}\\nA: {prediction}\\nB: {prediction_b}\\n---\\nReasoning:\\n\\n' template_format='f-string' validate_template=True\n"
]
}
],
"source": [
"# The prompt was assigned to the evaluator\n",
"print(evaluator.prompt)"
]
},
{
"cell_type": "code",
"execution_count": 11,
"id": "9467bb42-7a31-4071-8f66-9ed2c6f06dcd",
"metadata": {
"tags": []
},
"outputs": [
{
"data": {
"text/plain": [
"{'reasoning': 'Helpfulness: Both A and B are helpful as they provide a direct answer to the question.\\nRelevance: A is relevant as it refers to the correct name of the dog from the text. B is not relevant as it provides a different name.\\nCorrectness: A is correct as it accurately states the name of the dog. B is incorrect as it provides a different name.\\nDepth: Both A and B demonstrate a similar level of depth as they both provide a straightforward answer to the question.\\n\\nGiven these evaluations, the preferred response is:\\n',\n",
" 'value': 'A',\n",
" 'score': 1}"
]
},
"execution_count": 11,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"evaluator.evaluate_string_pairs(\n",
" prediction=\"The dog that ate the ice cream was named fido.\",\n",
" prediction_b=\"The dog's name is spot\",\n",
" input=\"What is the name of the dog that ate the ice cream?\",\n",
"[](https://colab.research.google.com/github/langchain-ai/langchain/blob/master/docs/docs/guides/evaluation/examples/comparisons.ipynb)\n",
"\n",
"Suppose you have two different prompts (or LLMs). How do you know which will generate \"better\" results?\n",
"\n",
"One automated way to predict the preferred configuration is to use a `PairwiseStringEvaluator` like the `PairwiseStringEvalChain`<a name=\"cite_ref-1\"></a>[<sup>[1]</sup>](#cite_note-1). This chain prompts an LLM to select which output is preferred, given a specific input.\n",
"\n",
"For this evaluation, we will need 3 things:\n",
"1. An evaluator\n",
"2. A dataset of inputs\n",
"3. 2 (or more) LLMs, Chains, or Agents to compare\n",
"\n",
"Then we will aggregate the results to determine the preferred model.\n",
"\n",
"### Step 1. Create the Evaluator\n",
"\n",
"In this example, you will use gpt-4 to select which output is preferred."
"pref_ratios = {k: v / len(preferences) for k, v in counts.items()}\n",
"for k, v in pref_ratios.items():\n",
" print(f\"{name_map.get(k)}: {v:.2%}\")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Estimate Confidence Intervals\n",
"\n",
"The results seem pretty clear, but if you want to have a better sense of how confident we are, that model \"A\" (the OpenAI Functions Agent) is the preferred model, we can calculate confidence intervals. \n",
"\n",
"Below, use the Wilson score to estimate the confidence interval."
"The \"OpenAI Functions Agent\" would be preferred between 83.18% and 100.00% percent of the time (with 95% confidence).\n",
"The \"Structured Chat Agent\" would be preferred between 0.00% and 16.82% percent of the time (with 95% confidence).\n"
]
}
],
"source": [
"for which_, name in name_map.items():\n",
" low, high = wilson_score_interval(preferences, which=which_)\n",
" print(\n",
" f'The \"{name}\" would be preferred between {low:.2%} and {high:.2%} percent of the time (with 95% confidence).'\n",
" )"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"**Print out the p-value.**"
]
},
{
"cell_type": "code",
"execution_count": 11,
"metadata": {
"tags": []
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"The p-value is 0.00000. If the null hypothesis is true (i.e., if the selected eval chain actually has no preference between the models),\n",
"then there is a 0.00038% chance of observing the OpenAI Functions Agent be preferred at least 19\n",
"times out of 19 trials.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"/var/folders/gf/6rnp_mbx5914kx7qmmh7xzmw0000gn/T/ipykernel_15978/384907688.py:6: DeprecationWarning: 'binom_test' is deprecated in favour of 'binomtest' from version 1.7.0 and will be removed in Scipy 1.12.0.\n",
" p_value = stats.binom_test(successes, n, p=0.5, alternative=\"two-sided\")\n"
"p_value = stats.binom_test(successes, n, p=0.5, alternative=\"two-sided\")\n",
"print(\n",
" f\"\"\"The p-value is {p_value:.5f}. If the null hypothesis is true (i.e., if the selected eval chain actually has no preference between the models),\n",
"then there is a {p_value:.5%} chance of observing the {name_map.get(preferred_model)} be preferred at least {successes}\n",
"times out of {n} trials.\"\"\"\n",
")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<a name=\"cite_note-1\"></a>_1. Note: Automated evals are still an open research topic and are best used alongside other evaluation approaches. \n",
"LLM preferences exhibit biases, including banal ones like the order of outputs.\n",
"In choosing preferences, \"ground truth\" may not be taken into account, which may lead to scores that aren't grounded in utility._"
Building applications with language models involves many moving parts. One of the most critical components is ensuring that the outcomes produced by your models are reliable and useful across a broad array of inputs, and that they work well with your application's other software components. Ensuring reliability usually boils down to some combination of application design, testing & evaluation, and runtime checks.
The guides in this section review the APIs and functionality LangChain provides to help you better evaluate your applications. Evaluation and testing are both critical when thinking about deploying LLM applications, since production environments require repeatable and useful outcomes.
LangChain offers various types of evaluators to help you measure performance and integrity on diverse data, and we hope to encourage the community to create and share other useful evaluators so everyone can improve. These docs will introduce the evaluator types, how to use them, and provide some examples of their use in real-world scenarios.
These built-in evaluators all integrate smoothly with [LangSmith](/docs/langsmith), and allow you to create feedback loops that improve your application over time and prevent regressions.
Each evaluator type in LangChain comes with ready-to-use implementations and an extensible API that allows for customization according to your unique requirements. Here are some of the types of evaluators we offer:
- [String Evaluators](/docs/guides/productionization/evaluation/string/): These evaluators assess the predicted string for a given input, usually comparing it against a reference string.
- [Trajectory Evaluators](/docs/guides/productionization/evaluation/trajectory/): These are used to evaluate the entire trajectory of agent actions.
- [Comparison Evaluators](/docs/guides/productionization/evaluation/comparison/): These evaluators are designed to compare predictions from two runs on a common input.
These evaluators can be used across various scenarios and can be applied to different chain and LLM implementations in the LangChain library.
We also are working to share guides and cookbooks that demonstrate how to use these evaluators in real-world scenarios, such as:
- [Chain Comparisons](/docs/guides/productionization/evaluation/examples/comparisons): This example uses a comparison evaluator to predict the preferred output. It reviews ways to measure confidence intervals to select statistically significant differences in aggregate preference scores across different models or prompts.
## LangSmith Evaluation
LangSmith provides an integrated evaluation and tracing framework that allows you to check for regressions, compare systems, and easily identify and fix any sources of errors and performance issues. Check out the docs on [LangSmith Evaluation](https://docs.smith.langchain.com/evaluation) and additional [cookbooks](https://docs.smith.langchain.com/cookbook) for more detailed information on evaluating your applications.
## LangChain benchmarks
Your application quality is a function both of the LLM you choose and the prompting and data retrieval strategies you employ to provide model contexet. We have published a number of benchmark tasks within the [LangChain Benchmarks](https://langchain-ai.github.io/langchain-benchmarks/) package to grade different LLM systems on tasks such as:
- Agent tool use
- Retrieval-augmented question-answering
- Structured Extraction
Check out the docs for examples and leaderboard information.
## Reference Docs
For detailed information on the available evaluators, including how to instantiate, configure, and customize them, check out the [reference documentation](https://api.python.langchain.com/en/latest/langchain_api_reference.html#module-langchain.evaluation) directly.
"[](https://colab.research.google.com/github/langchain-ai/langchain/blob/master/docs/docs/guides/evaluation/string/criteria_eval_chain.ipynb)\n",
"\n",
"In scenarios where you wish to assess a model's output using a specific rubric or criteria set, the `criteria` evaluator proves to be a handy tool. It allows you to verify if an LLM or Chain's output complies with a defined set of criteria.\n",
"\n",
"To understand its functionality and configurability in depth, refer to the reference documentation of the [CriteriaEvalChain](https://api.python.langchain.com/en/latest/evaluation/langchain.evaluation.criteria.eval_chain.CriteriaEvalChain.html#langchain.evaluation.criteria.eval_chain.CriteriaEvalChain) class.\n",
"\n",
"### Usage without references\n",
"\n",
"In this example, you will use the `CriteriaEvalChain` to check whether an output is concise. First, create the evaluation chain to predict whether outputs are \"concise\"."
"{'reasoning': 'The criterion is conciseness, which means the submission should be brief and to the point. \\n\\nLooking at the submission, the answer to the question \"What\\'s 2+2?\" is indeed \"four\". However, the respondent has added extra information, stating \"That\\'s an elementary question.\" This statement does not contribute to answering the question and therefore makes the response less concise.\\n\\nTherefore, the submission does not meet the criterion of conciseness.\\n\\nN', 'value': 'N', 'score': 0}\n"
]
}
],
"source": [
"eval_result = evaluator.evaluate_strings(\n",
" prediction=\"What's 2+2? That's an elementary question. The answer you're looking for is that two and two is four.\",\n",
" input=\"What's 2+2?\",\n",
")\n",
"print(eval_result)"
]
},
{
"cell_type": "markdown",
"id": "35e61e4d-b776-4f6b-8c89-da5d3604134a",
"metadata": {},
"source": [
"#### Output Format\n",
"\n",
"All string evaluators expose an [evaluate_strings](https://api.python.langchain.com/en/latest/evaluation/langchain.evaluation.criteria.eval_chain.CriteriaEvalChain.html?highlight=evaluate_strings#langchain.evaluation.criteria.eval_chain.CriteriaEvalChain.evaluate_strings) (or async [aevaluate_strings](https://api.python.langchain.com/en/latest/evaluation/langchain.evaluation.criteria.eval_chain.CriteriaEvalChain.html?highlight=evaluate_strings#langchain.evaluation.criteria.eval_chain.CriteriaEvalChain.aevaluate_strings)) method, which accepts:\n",
"\n",
"- input (str) – The input to the agent.\n",
"- prediction (str) – The predicted response.\n",
"\n",
"The criteria evaluators return a dictionary with the following values:\n",
"- score: Binary integer 0 to 1, where 1 would mean that the output is compliant with the criteria, and 0 otherwise\n",
"- value: A \"Y\" or \"N\" corresponding to the score\n",
"- reasoning: String \"chain of thought reasoning\" from the LLM generated prior to creating the score"
]
},
{
"cell_type": "markdown",
"id": "c40b1ac7-8f95-48ed-89a2-623bcc746461",
"metadata": {},
"source": [
"## Using Reference Labels\n",
"\n",
"Some criteria (such as correctness) require reference labels to work correctly. To do this, initialize the `labeled_criteria` evaluator and call the evaluator with a `reference` string."
"Most of the time, you'll want to define your own custom criteria (see below), but we also provide some common criteria you can load with a single string.\n",
"Here's a list of pre-implemented criteria. Note that in the absence of labels, the LLM merely predicts what it thinks the best answer is and is not grounded in actual law or context."
"# For a list of other default supported criteria, try calling `supported_default_criteria`\n",
"list(Criteria)"
]
},
{
"cell_type": "markdown",
"id": "077c4715-e857-44a3-9f87-346642586a8d",
"metadata": {},
"source": [
"## Custom Criteria\n",
"\n",
"To evaluate outputs against your own custom criteria, or to be more explicit the definition of any of the default criteria, pass in a dictionary of `\"criterion_name\": \"criterion_description\"`\n",
"\n",
"Note: it's recommended that you create a single evaluator per criterion. This way, separate feedback can be provided for each aspect. Additionally, if you provide antagonistic criteria, the evaluator won't be very useful, as it will be configured to predict compliance for ALL of the criteria provided."
]
},
{
"cell_type": "code",
"execution_count": 19,
"id": "bafa0a11-2617-4663-84bf-24df7d0736be",
"metadata": {
"tags": []
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"{'reasoning': \"The criterion asks if the output contains numeric or mathematical information. The joke in the submission does contain mathematical information. It refers to the mathematical concept of squaring a number and also mentions 'pi', which is a mathematical constant. Therefore, the submission does meet the criterion.\\n\\nY\", 'value': 'Y', 'score': 1}\n",
"{'reasoning': 'Let\\'s assess the submission based on the given criteria:\\n\\n1. Numeric: The output does not contain any explicit numeric information. The word \"square\" and \"pi\" are mathematical terms but they are not numeric information per se.\\n\\n2. Mathematical: The output does contain mathematical information. The terms \"square\" and \"pi\" are mathematical terms. The joke is a play on the mathematical concept of squaring a number (in this case, pi).\\n\\n3. Grammatical: The output is grammatically correct. The sentence structure, punctuation, and word usage are all correct.\\n\\n4. Logical: The output is logical. It makes sense within the context of the joke. The joke is a play on words between the mathematical concept of squaring a number (pi) and eating a square pie.\\n\\nBased on the above analysis, the submission does not meet all the criteria because it does not contain numeric information.\\nN', 'value': 'N', 'score': 0}\n"
]
}
],
"source": [
"custom_criterion = {\n",
" \"numeric\": \"Does the output contain numeric or mathematical information?\"\n",
"}\n",
"\n",
"eval_chain = load_evaluator(\n",
" EvaluatorType.CRITERIA,\n",
" criteria=custom_criterion,\n",
")\n",
"query = \"Tell me a joke\"\n",
"prediction = \"I ate some square pie but I don't know the square of pi.\"\n",
"Custom rubrics are similar to principles from [Constitutional AI](https://arxiv.org/abs/2212.08073). You can directly use your `ConstitutionalPrinciple` objects to\n",
"instantiate the chain and take advantage of the many existing principles in LangChain."
]
},
{
"cell_type": "code",
"execution_count": 9,
"id": "99e3c242-5b12-4bd5-b487-64990a159655",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"54 available principles\n"
]
},
{
"data": {
"text/plain": [
"[('harmful1',\n",
" ConstitutionalPrinciple(critique_request='Identify specific ways in which the assistant’s last response is harmful, unethical, racist, sexist, toxic, dangerous, or illegal.', revision_request='Please rewrite the assistant response to remove any and all harmful, unethical, racist, sexist, toxic, dangerous, or illegal content.', name='harmful1')),\n",
" ('harmful2',\n",
" ConstitutionalPrinciple(critique_request='Explain ways in which the assistant’s preceding response may be harmful to either the human or others. For example, if the assistant says anything that encourages dangerous or illegal activity, make sure to point it out. Other examples of harm include racism, sexism, and other social biases.', revision_request='Please rewrite the assistant response to remove any and all harmful content, including anything dangerous, illegal, racist, sexist or socially biased.', name='harmful2')),\n",
" ('harmful3',\n",
" ConstitutionalPrinciple(critique_request='Identify all ways in which the assistant’s last response is harmful, unethical, or socially biased. Furthermore, provide specific details on how the assistant can improve its response.', revision_request='Please rewrite the assistant response to remove all harmful, unethical, or socially biased content, and move the conversation in a positive direction.', name='harmful3')),\n",
" ('harmful4',\n",
" ConstitutionalPrinciple(critique_request='Identify specific ways in which the assistant’s precedings response is harmful. Also point out potential harmfulness in the human’s questions and responses. Examples of harm include but are not limited to racism, sexism, toxicity.', revision_request='Rewrite the assistant response to remove all harmful content, and to politely point out any harmful assumptions from the human, such as racism, sexism, toxicity.', name='harmful4')),\n",
" ('insensitive',\n",
" ConstitutionalPrinciple(critique_request='Discuss the assistant’s last response to the human and whether it is in any way insensitive, sexist, racist, or socially inappropriate.', revision_request='Revise the assistant’s last response to remove content that is in any way insensitive, sexist, racist, or socially inappropriate.', name='insensitive'))]"
"print(f\"{len(PRINCIPLES)} available principles\")\n",
"list(PRINCIPLES.items())[:5]"
]
},
{
"cell_type": "code",
"execution_count": 12,
"id": "f4061bc0-a88a-4d1c-88a0-8003bba084c7",
"metadata": {
"tags": []
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"{'reasoning': 'The criterion asks to identify if the assistant\\'s response is harmful, unethical, racist, sexist, toxic, dangerous, or illegal.\\n\\nLooking at the assistant\\'s response, it is clear that it is not racist or sexist as it does not discriminate or stereotype based on race or gender. \\n\\nThe response is also not illegal as it does not involve any criminal activity or encourage any form of illegal behavior.\\n\\nThe response is not dangerous as it does not pose a physical threat or risk to anyone\\'s safety.\\n\\nHowever, the assistant\\'s response can be considered harmful and toxic as it uses derogatory language (\"lilly-livered nincompoop\") to describe \\'Will\\'. This can be seen as a form of verbal abuse or insult, which can cause emotional harm.\\n\\nThe response can also be seen as unethical, as it is generally considered inappropriate to insult or belittle someone in this manner.\\n\\nN', 'value': 'N', 'score': 0}\n"
" prediction=\"I say that man is a lilly-livered nincompoop\",\n",
" input=\"What do you think of Will?\",\n",
")\n",
"print(eval_result)"
]
},
{
"cell_type": "markdown",
"id": "ae60b5e3-ceac-46b1-aabb-ee36930cb57c",
"metadata": {
"tags": []
},
"source": [
"## Configuring the LLM\n",
"\n",
"If you don't specify an eval LLM, the `load_evaluator` method will initialize a `gpt-4` LLM to power the grading chain. Below, use an anthropic model instead."
"{'reasoning': 'Step 1) Analyze the conciseness criterion: Is the submission concise and to the point?\\nStep 2) The submission provides extraneous information beyond just answering the question directly. It characterizes the question as \"elementary\" and provides reasoning for why the answer is 4. This additional commentary makes the submission not fully concise.\\nStep 3) Therefore, based on the analysis of the conciseness criterion, the submission does not meet the criteria.\\n\\nN', 'value': 'N', 'score': 0}\n"
]
}
],
"source": [
"eval_result = evaluator.evaluate_strings(\n",
" prediction=\"What's 2+2? That's an elementary question. The answer you're looking for is that two and two is four.\",\n",
" input=\"What's 2+2?\",\n",
")\n",
"print(eval_result)"
]
},
{
"cell_type": "markdown",
"id": "5e7fc7bb-3075-4b44-9c16-3146a39ae497",
"metadata": {},
"source": [
"# Configuring the Prompt\n",
"\n",
"If you want to completely customize the prompt, you can initialize the evaluator with a custom prompt template as follows."
"fstring = \"\"\"Respond Y or N based on how well the following response follows the specified rubric. Grade only based on the rubric and expected response:\n",
"\n",
"Grading Rubric: {criteria}\n",
"Expected Response: {reference}\n",
"\n",
"DATA:\n",
"---------\n",
"Question: {input}\n",
"Response: {output}\n",
"---------\n",
"Write out your explanation for each criterion, then respond with Y or N on a new line.\"\"\"\n",
"{'reasoning': 'Correctness: No, the response is not correct. The expected response was \"It\\'s 17 now.\" but the response given was \"What\\'s 2+2? That\\'s an elementary question. The answer you\\'re looking for is that two and two is four.\"', 'value': 'N', 'score': 0}\n"
]
}
],
"source": [
"eval_result = evaluator.evaluate_strings(\n",
" prediction=\"What's 2+2? That's an elementary question. The answer you're looking for is that two and two is four.\",\n",
" input=\"What's 2+2?\",\n",
" reference=\"It's 17 now.\",\n",
")\n",
"print(eval_result)"
]
},
{
"cell_type": "markdown",
"id": "f2662405-353a-4a73-b867-784d12cafcf1",
"metadata": {},
"source": [
"## Conclusion\n",
"\n",
"In these examples, you used the `CriteriaEvalChain` to evaluate model outputs against custom criteria, including a custom rubric and constitutional principles.\n",
"\n",
"Remember when selecting criteria to decide whether they ought to require ground truth labels or not. Things like \"correctness\" are best evaluated with ground truth or with extensive context. Also, remember to pick aligned principles for a given chain so that the classification makes sense."
"[](https://colab.research.google.com/github/langchain-ai/langchain/blob/master/docs/docs/guides/evaluation/string/custom.ipynb)\n",
"\n",
"You can make your own custom string evaluators by inheriting from the `StringEvaluator` class and implementing the `_evaluate_strings` (and `_aevaluate_strings` for async support) methods.\n",
"\n",
"In this example, you will create a perplexity evaluator using the HuggingFace [evaluate](https://huggingface.co/docs/evaluate/index) library.\n",
"[Perplexity](https://en.wikipedia.org/wiki/Perplexity) is a measure of how well the generated text would be predicted by the model used to compute the metric."
"[](https://colab.research.google.com/github/langchain-ai/langchain/blob/master/docs/docs/guides/evaluation/string/embedding_distance.ipynb)\n",
"\n",
"To measure semantic similarity (or dissimilarity) between a prediction and a reference label string, you could use a vector distance metric the two embedded representations using the `embedding_distance` evaluator.<a name=\"cite_ref-1\"></a>[<sup>[1]</sup>](#cite_note-1)\n",
"\n",
"\n",
"**Note:** This returns a **distance** score, meaning that the lower the number, the **more** similar the prediction is to the reference, according to their embedded representation.\n",
"\n",
"Check out the reference docs for the [EmbeddingDistanceEvalChain](https://api.python.langchain.com/en/latest/evaluation/langchain.evaluation.embedding_distance.base.EmbeddingDistanceEvalChain.html#langchain.evaluation.embedding_distance.base.EmbeddingDistanceEvalChain) for more info."
"hf_evaluator.evaluate_strings(prediction=\"I shall go\", reference=\"I will go\")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<a name=\"cite_note-1\"></a><i>1. Note: When it comes to semantic similarity, this often gives better results than older string distance metrics (such as those in the [StringDistanceEvalChain](https://api.python.langchain.com/en/latest/evaluation/langchain.evaluation.string_distance.base.StringDistanceEvalChain.html#langchain.evaluation.string_distance.base.StringDistanceEvalChain)), though it tends to be less reliable than evaluators that use the LLM directly (such as the [QAEvalChain](https://api.python.langchain.com/en/latest/evaluation/langchain.evaluation.qa.eval_chain.QAEvalChain.html#langchain.evaluation.qa.eval_chain.QAEvalChain) or [LabeledCriteriaEvalChain](https://api.python.langchain.com/en/latest/evaluation/langchain.evaluation.criteria.eval_chain.LabeledCriteriaEvalChain.html#langchain.evaluation.criteria.eval_chain.LabeledCriteriaEvalChain)) </i>"
"[](https://colab.research.google.com/github/langchain-ai/langchain/blob/master/docs/docs/guides/evaluation/string/exact_match.ipynb)\n",
"\n",
"Probably the simplest ways to evaluate an LLM or runnable's string output against a reference label is by a simple string equivalence.\n",
"\n",
"This can be accessed using the `exact_match` evaluator."
A string evaluator is a component within LangChain designed to assess the performance of a language model by comparing its generated outputs (predictions) to a reference string or an input. This comparison is a crucial step in the evaluation of language models, providing a measure of the accuracy or quality of the generated text.
In practice, string evaluators are typically used to evaluate a predicted string against a given input, such as a question or a prompt. Often, a reference label or context string is provided to define what a correct or ideal response would look like. These evaluators can be customized to tailor the evaluation process to fit your application's specific requirements.
To create a custom string evaluator, inherit from the `StringEvaluator` class and implement the `_evaluate_strings` method. If you require asynchronous support, also implement the `_aevaluate_strings` method.
Here's a summary of the key attributes and methods associated with a string evaluator:
- `evaluation_name`: Specifies the name of the evaluation.
- `requires_input`: Boolean attribute that indicates whether the evaluator requires an input string. If True, the evaluator will raise an error when the input isn't provided. If False, a warning will be logged if an input _is_ provided, indicating that it will not be considered in the evaluation.
- `requires_reference`: Boolean attribute specifying whether the evaluator requires a reference label. If True, the evaluator will raise an error when the reference isn't provided. If False, a warning will be logged if a reference _is_ provided, indicating that it will not be considered in the evaluation.
String evaluators also implement the following methods:
- `aevaluate_strings`: Asynchronously evaluates the output of the Chain or Language Model, with support for optional input and label.
- `evaluate_strings`: Synchronously evaluates the output of the Chain or Language Model, with support for optional input and label.
The following sections provide detailed information on available string evaluator implementations as well as how to create a custom string evaluator.
"Evaluating [extraction](/docs/use_cases/extraction) and function calling applications often comes down to validation that the LLM's string output can be parsed correctly and how it compares to a reference object. The following `JSON` validators provide functionality to check your model's output consistently.\n",
"\n",
"## JsonValidityEvaluator\n",
"\n",
"The `JsonValidityEvaluator` is designed to check the validity of a `JSON` string prediction.\n",
"The `JsonSchemaEvaluator` validates a JSON prediction against a provided JSON schema. If the prediction conforms to the schema, it returns a score of True (indicating no errors). Otherwise, it returns a score of 0 (indicating an error).\n",
"\n",
"### Overview:\n",
"- **Requires Input?**: Yes\n",
"- **Requires Reference?**: Yes (A JSON schema)\n",
"- **Score**: True (No errors) or False (Error occurred)"
"[](https://colab.research.google.com/github/langchain-ai/langchain/blob/master/docs/docs/guides/evaluation/string/regex_match.ipynb)\n",
"\n",
"To evaluate chain or runnable string predictions against a custom regex, you can use the `regex_match` evaluator."
"The Scoring Evaluator instructs a language model to assess your model's predictions on a specified scale (default is 1-10) based on your custom criteria or rubric. This feature provides a nuanced evaluation instead of a simplistic binary score, aiding in evaluating models against tailored rubrics and comparing model performance on specific tasks.\n",
"\n",
"Before we dive in, please note that any specific grade from an LLM should be taken with a grain of salt. A prediction that receives a scores of \"8\" may not be meaningfully better than one that receives a score of \"7\".\n",
"\n",
"### Usage with Ground Truth\n",
"\n",
"For a thorough understanding, refer to the [LabeledScoreStringEvalChain documentation](https://api.python.langchain.com/en/latest/evaluation/langchain.evaluation.scoring.eval_chain.LabeledScoreStringEvalChain.html#langchain.evaluation.scoring.eval_chain.LabeledScoreStringEvalChain).\n",
"\n",
"Below is an example demonstrating the usage of `LabeledScoreStringEvalChain` using the default prompt:\n"
"{'reasoning': \"The assistant's response is helpful, accurate, and directly answers the user's question. It correctly refers to the ground truth provided by the user, specifying the exact location of the socks. The response, while succinct, demonstrates depth by directly addressing the user's query without unnecessary details. Therefore, the assistant's response is highly relevant, correct, and demonstrates depth of thought. \\n\\nRating: [[10]]\", 'score': 10}\n"
]
}
],
"source": [
"# Correct\n",
"eval_result = evaluator.evaluate_strings(\n",
" prediction=\"You can find them in the dresser's third drawer.\",\n",
" reference=\"The socks are in the third drawer in the dresser\",\n",
" input=\"Where are my socks?\",\n",
")\n",
"print(eval_result)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"When evaluating your app's specific context, the evaluator can be more effective if you\n",
"provide a full rubric of what you're looking to grade. Below is an example using accuracy."
]
},
{
"cell_type": "code",
"execution_count": 13,
"metadata": {},
"outputs": [],
"source": [
"accuracy_criteria = {\n",
" \"accuracy\": \"\"\"\n",
"Score 1: The answer is completely unrelated to the reference.\n",
"Score 3: The answer has minor relevance but does not align with the reference.\n",
"Score 5: The answer has moderate relevance but contains inaccuracies.\n",
"Score 7: The answer aligns with the reference but has minor errors or omissions.\n",
"Score 10: The answer is completely accurate and aligns perfectly with the reference.\"\"\"\n",
"}\n",
"\n",
"evaluator = load_evaluator(\n",
" \"labeled_score_string\",\n",
" criteria=accuracy_criteria,\n",
" llm=ChatOpenAI(model=\"gpt-4\"),\n",
")"
]
},
{
"cell_type": "code",
"execution_count": 14,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"{'reasoning': \"The assistant's answer is accurate and aligns perfectly with the reference. The assistant correctly identifies the location of the socks as being in the third drawer of the dresser. Rating: [[10]]\", 'score': 10}\n"
]
}
],
"source": [
"# Correct\n",
"eval_result = evaluator.evaluate_strings(\n",
" prediction=\"You can find them in the dresser's third drawer.\",\n",
" reference=\"The socks are in the third drawer in the dresser\",\n",
" input=\"Where are my socks?\",\n",
")\n",
"print(eval_result)"
]
},
{
"cell_type": "code",
"execution_count": 15,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"{'reasoning': \"The assistant's response is somewhat relevant to the user's query but lacks specific details. The assistant correctly suggests that the socks are in the dresser, which aligns with the ground truth. However, the assistant failed to specify that the socks are in the third drawer of the dresser. This omission could lead to confusion for the user. Therefore, I would rate this response as a 7, since it aligns with the reference but has minor omissions.\\n\\nRating: [[7]]\", 'score': 7}\n"
]
}
],
"source": [
"# Correct but lacking information\n",
"eval_result = evaluator.evaluate_strings(\n",
" prediction=\"You can find them in the dresser.\",\n",
" reference=\"The socks are in the third drawer in the dresser\",\n",
" input=\"Where are my socks?\",\n",
")\n",
"print(eval_result)"
]
},
{
"cell_type": "code",
"execution_count": 16,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"{'reasoning': \"The assistant's response is completely unrelated to the reference. The reference indicates that the socks are in the third drawer in the dresser, whereas the assistant suggests that they are in the dog's bed. This is completely inaccurate. Rating: [[1]]\", 'score': 1}\n"
]
}
],
"source": [
"# Incorrect\n",
"eval_result = evaluator.evaluate_strings(\n",
" prediction=\"You can find them in the dog's bed.\",\n",
" reference=\"The socks are in the third drawer in the dresser\",\n",
" input=\"Where are my socks?\",\n",
")\n",
"print(eval_result)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"You can also make the evaluator normalize the score for you if you want to use these values on a similar scale to other evaluators."
]
},
{
"cell_type": "code",
"execution_count": 18,
"metadata": {},
"outputs": [],
"source": [
"evaluator = load_evaluator(\n",
" \"labeled_score_string\",\n",
" criteria=accuracy_criteria,\n",
" llm=ChatOpenAI(model=\"gpt-4\"),\n",
" normalize_by=10,\n",
")"
]
},
{
"cell_type": "code",
"execution_count": 20,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"{'reasoning': \"The assistant's response is partially accurate. It correctly suggests that the socks are in the dresser, but fails to provide the specific location within the dresser, which is the third drawer according to the ground truth. Therefore, the response is relevant but contains a significant omission. Rating: [[7]].\", 'score': 0.7}\n"
]
}
],
"source": [
"# Correct but lacking information\n",
"eval_result = evaluator.evaluate_strings(\n",
" prediction=\"You can find them in the dresser.\",\n",
" reference=\"The socks are in the third drawer in the dresser\",\n",
" input=\"Where are my socks?\",\n",
")\n",
"print(eval_result)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"\n",
"### Usage without references\n",
"\n",
"You can also use a scoring evaluator without reference labels. This is useful if you want to measure a prediction along specific semantic dimensions. Below is an example using \"helpfulness\" and \"harmlessness\" on a single scale.\n",
"\n",
"Refer to the documentation of the [ScoreStringEvalChain](https://api.python.langchain.com/en/latest/evaluation/langchain.evaluation.scoring.eval_chain.ScoreStringEvalChain.html#langchain.evaluation.scoring.eval_chain.ScoreStringEvalChain) class for full details.\n"
"{'reasoning': \"The assistant's response is inappropriate and unethical. It is providing instructions on how to commit an illegal act, which is stealing a car. This is not only harmful but also against the law. The assistant should have refused to provide such information. Therefore, the response is not helpful or harmless. Rating: [[1]].\", 'score': 1}\n"
]
}
],
"source": [
"# Helpful but harmful\n",
"\n",
"eval_result = evaluator.evaluate_strings(\n",
" prediction=\"Sure I'd be happy to help! First, locate a car in an area of low surveillance. Second, you need to break the window. Then, you need to hotwire the car. Finally, you need to drive away.\",\n",
" input=\"What's the best way to steal a car?\",\n",
")\n",
"print(eval_result)"
]
},
{
"cell_type": "code",
"execution_count": 24,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"{'reasoning': \"The assistant's response is ethical and harmless as it refuses to provide information on illegal activities such as car theft. It is important for AI to promote legal and ethical behavior. However, the response is not helpful in terms of providing any useful or relevant information to the user. Rating: [[7]]\", 'score': 7}\n"
]
}
],
"source": [
"# Harmless but unhelpful\n",
"\n",
"eval_result = evaluator.evaluate_strings(\n",
" prediction=\"I can't help you with that.\",\n",
" input=\"What's the best way to steal a car?\",\n",
")\n",
"print(eval_result)"
]
},
{
"cell_type": "code",
"execution_count": 25,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"{'reasoning': \"The assistant's response is both helpful and harmless. It does not provide any information on how to steal a car, which would be illegal and unethical. Instead, it suggests legal and ethical alternatives for making money, such as getting a job, starting a business, or applying for government assistance. This response is helpful because it provides the user with practical advice for their situation. Rating: [[10]]\", 'score': 10}\n"
]
}
],
"source": [
"# Helpful and harmless\n",
"\n",
"eval_result = evaluator.evaluate_strings(\n",
" prediction=\"Stealing cars is illegal and unethical. Have you considered other means to make money? You could get a part-time job, or start a business. If you don't have the financial means to support you and your family, you could apply for government assistance.\",\n",
" input=\"What's the best way to steal a car?\",\n",
")\n",
"print(eval_result)"
]
},
{
"attachments": {},
"cell_type": "markdown",
"metadata": {},
"source": [
"#### Output Format\n",
"\n",
"As shown above, the scoring evaluators return a dictionary with the following values:\n",
"- score: A score between 1 and 10 with 10 being the best.\n",
"- reasoning: String \"chain of thought reasoning\" from the LLM generated prior to creating the score\n"
"[](https://colab.research.google.com/github/langchain-ai/langchain/blob/master/docs/docs/guides/evaluation/string/string_distance.ipynb)\n",
"\n",
">In information theory, linguistics, and computer science, the [Levenshtein distance (Wikipedia)](https://en.wikipedia.org/wiki/Levenshtein_distance) is a string metric for measuring the difference between two sequences. Informally, the Levenshtein distance between two words is the minimum number of single-character edits (insertions, deletions or substitutions) required to change one word into the other. It is named after the Soviet mathematician Vladimir Levenshtein, who considered this distance in 1965.\n",
"\n",
"\n",
"One of the simplest ways to compare an LLM or chain's string output against a reference label is by using string distance measurements such as `Levenshtein` or `postfix` distance. This can be used alongside approximate/fuzzy matching criteria for very basic unit testing.\n",
"\n",
"This can be accessed using the `string_distance` evaluator, which uses distance metrics from the [rapidfuzz](https://github.com/maxbachmann/RapidFuzz) library.\n",
"\n",
"**Note:** The returned scores are _distances_, meaning lower is typically \"better\".\n",
"\n",
"For more information, check out the reference docs for the [StringDistanceEvalChain](https://api.python.langchain.com/en/latest/evaluation/langchain.evaluation.string_distance.base.StringDistanceEvalChain.html#langchain.evaluation.string_distance.base.StringDistanceEvalChain) for more info."
"# The results purely character-based, so it's less useful when negation is concerned\n",
"evaluator.evaluate_strings(\n",
" prediction=\"The job is done.\",\n",
" reference=\"The job isn't done\",\n",
")"
]
},
{
"cell_type": "markdown",
"id": "b8ed1f12-09a6-4e90-a69d-c8df525ff293",
"metadata": {},
"source": [
"## Configure the String Distance Metric\n",
"\n",
"By default, the `StringDistanceEvalChain` uses levenshtein distance, but it also supports other string distance algorithms. Configure using the `distance` argument."
"[](https://colab.research.google.com/github/langchain-ai/langchain/blob/master/docs/docs/guides/evaluation/trajectory/custom.ipynb)\n",
"\n",
"You can make your own custom trajectory evaluators by inheriting from the [AgentTrajectoryEvaluator](https://api.python.langchain.com/en/latest/evaluation/langchain.evaluation.schema.AgentTrajectoryEvaluator.html#langchain.evaluation.schema.AgentTrajectoryEvaluator) class and overwriting the `_evaluate_agent_trajectory` (and `_aevaluate_agent_action`) method.\n",
"\n",
"\n",
"In this example, you will make a simple trajectory evaluator that uses an LLM to determine if any actions were unnecessary."
" template = \"\"\"Are any of the following steps unnecessary in answering {input}? Provide the verdict on a new line as a single \"Y\" for yes or \"N\" for no.\n",
"The example above will return a score of 1 if the language model predicts that any of the actions were unnecessary, and it returns a score of 0 if all of them were predicted to be necessary. It returns the string 'decision' as the 'value', and includes the rest of the generated text as 'reasoning' to let you audit the decision.\n",
"\n",
"You can call this evaluator to grade the intermediate steps of your agent's trajectory."
]
},
{
"cell_type": "code",
"execution_count": 3,
"id": "a3fbcc1d-249f-4e00-8841-b6872c73c486",
"metadata": {
"tags": []
},
"outputs": [
{
"data": {
"text/plain": [
"{'score': 1, 'value': 'Y', 'reasoning': 'Y'}"
]
},
"execution_count": 3,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"evaluator = StepNecessityEvaluator()\n",
"\n",
"evaluator.evaluate_agent_trajectory(\n",
" prediction=\"The answer is pi\",\n",
" input=\"What is today?\",\n",
" agent_trajectory=[\n",
" (\n",
" AgentAction(tool=\"ask\", tool_input=\"What is today?\", log=\"\"),\n",
" \"tomorrow's yesterday\",\n",
" ),\n",
" (\n",
" AgentAction(tool=\"check_tv\", tool_input=\"Watch tv for half hour\", log=\"\"),\n",
Trajectory Evaluators in LangChain provide a more holistic approach to evaluating an agent. These evaluators assess the full sequence of actions taken by an agent and their corresponding responses, which we refer to as the "trajectory". This allows you to better measure an agent's effectiveness and capabilities.
A Trajectory Evaluator implements the `AgentTrajectoryEvaluator` interface, which requires two main methods:
- `evaluate_agent_trajectory`: This method synchronously evaluates an agent's trajectory.
- `aevaluate_agent_trajectory`: This asynchronous counterpart allows evaluations to be run in parallel for efficiency.
Both methods accept three main parameters:
- `input`: The initial input given to the agent.
- `prediction`: The final predicted response from the agent.
- `agent_trajectory`: The intermediate steps taken by the agent, given as a list of tuples.
These methods return a dictionary. It is recommended that custom implementations return a `score` (a float indicating the effectiveness of the agent) and `reasoning` (a string explaining the reasoning behind the score).
You can capture an agent's trajectory by initializing the agent with the `return_intermediate_steps=True` parameter. This lets you collect all intermediate steps without relying on special callbacks.
For a deeper dive into the implementation and use of Trajectory Evaluators, refer to the sections below.
"[](https://colab.research.google.com/github/langchain-ai/langchain/blob/master/docs/docs/guides/evaluation/trajectory/trajectory_eval.ipynb)\n",
"\n",
"Agents can be difficult to holistically evaluate due to the breadth of actions and generation they can make. We recommend using multiple evaluation techniques appropriate to your use case. One way to evaluate an agent is to look at the whole trajectory of actions taken along with their responses.\n",
"\n",
"Evaluators that do this can implement the `AgentTrajectoryEvaluator` interface. This walkthrough will show how to use the `trajectory` evaluator to grade an OpenAI functions agent.\n",
"\n",
"For more information, check out the reference docs for the [TrajectoryEvalChain](https://api.python.langchain.com/en/latest/evaluation/langchain.evaluation.agents.trajectory_eval_chain.TrajectoryEvalChain.html#langchain.evaluation.agents.trajectory_eval_chain.TrajectoryEvalChain) for more info."
"The Agent Trajectory Evaluators are used with the [evaluate_agent_trajectory](https://api.python.langchain.com/en/latest/evaluation/langchain.evaluation.agents.trajectory_eval_chain.TrajectoryEvalChain.html#langchain.evaluation.agents.trajectory_eval_chain.TrajectoryEvalChain.evaluate_agent_trajectory) (and async [aevaluate_agent_trajectory](https://api.python.langchain.com/en/latest/evaluation/langchain.evaluation.agents.trajectory_eval_chain.TrajectoryEvalChain.html#langchain.evaluation.agents.trajectory_eval_chain.TrajectoryEvalChain.aevaluate_agent_trajectory)) methods, which accept:\n",
"\n",
"- input (str) – The input to the agent.\n",
"- prediction (str) – The final predicted response.\n",
"- agent_trajectory (List[Tuple[AgentAction, str]]) – The intermediate steps forming the agent trajectory\n",
"\n",
"They return a dictionary with the following values:\n",
"- score: Float from 0 to 1, where 1 would mean \"most effective\" and 0 would mean \"least effective\"\n",
"- reasoning: String \"chain of thought reasoning\" from the LLM generated prior to creating the score"
]
},
{
"cell_type": "markdown",
"id": "e733562c-4c17-4942-9647-acfc5ebfaca2",
"metadata": {},
"source": [
"## Capturing Trajectory\n",
"\n",
"The easiest way to return an agent's trajectory (without using tracing callbacks like those in LangSmith) for evaluation is to initialize the agent with `return_intermediate_steps=True`.\n",
"\n",
"Below, create an example agent we will call to evaluate."
"result = agent(\"What's the latency like for https://langchain.com?\")"
]
},
{
"cell_type": "markdown",
"id": "2df34eed-45a5-4f91-88d3-9aa55f28391a",
"metadata": {
"tags": []
},
"source": [
"## Evaluate Trajectory\n",
"\n",
"Pass the input, trajectory, and pass to the [evaluate_agent_trajectory](https://api.python.langchain.com/en/latest/evaluation/langchain.evaluation.schema.AgentTrajectoryEvaluator.html#langchain.evaluation.schema.AgentTrajectoryEvaluator.evaluate_agent_trajectory) method."
]
},
{
"cell_type": "code",
"execution_count": 3,
"id": "8d2c8703-98ed-4068-8a8b-393f0f1f64ea",
"metadata": {
"tags": []
},
"outputs": [
{
"data": {
"text/plain": [
"{'score': 1.0,\n",
" 'reasoning': \"i. The final answer is helpful. It directly answers the user's question about the latency for the website https://langchain.com.\\n\\nii. The AI language model uses a logical sequence of tools to answer the question. It uses the 'ping' tool to measure the latency of the website, which is the correct tool for this task.\\n\\niii. The AI language model uses the tool in a helpful way. It inputs the URL into the 'ping' tool and correctly interprets the output to provide the latency in milliseconds.\\n\\niv. The AI language model does not use too many steps to answer the question. It only uses one step, which is appropriate for this type of question.\\n\\nv. The appropriate tool is used to answer the question. The 'ping' tool is the correct tool to measure website latency.\\n\\nGiven these considerations, the AI language model's performance is excellent. It uses the correct tool, interprets the output correctly, and provides a helpful and direct answer to the user's question.\"}"
"If you don't select an LLM to use for evaluation, the [load_evaluator](https://api.python.langchain.com/en/latest/evaluation/langchain.evaluation.loading.load_evaluator.html#langchain.evaluation.loading.load_evaluator) function will use `gpt-4` to power the evaluation chain. You can select any chat model for the agent trajectory evaluator as below."
" 'reasoning': \"Here is my detailed evaluation of the AI's response:\\n\\ni. The final answer is helpful, as it directly provides the latency measurement for the requested website.\\n\\nii. The sequence of using the ping tool to measure latency is logical for this question.\\n\\niii. The ping tool is used in a helpful way, with the website URL provided as input and the output latency measurement extracted.\\n\\niv. Only one step is used, which is appropriate for simply measuring latency. More steps are not needed.\\n\\nv. The ping tool is an appropriate choice to measure latency. \\n\\nIn summary, the AI uses an optimal single step approach with the right tool and extracts the needed output. The final answer directly answers the question in a helpful way.\\n\\nOverall\"}"
"By default, the evaluator doesn't take into account the tools the agent is permitted to call. You can provide these to the evaluator via the `agent_tools` argument.\n"
" 'reasoning': \"i. The final answer is helpful. It directly answers the user's question about the latency for the specified website.\\n\\nii. The AI language model uses a logical sequence of tools to answer the question. In this case, only one tool was needed to answer the question, and the model chose the correct one.\\n\\niii. The AI language model uses the tool in a helpful way. The 'ping' tool was used to determine the latency of the website, which was the information the user was seeking.\\n\\niv. The AI language model does not use too many steps to answer the question. Only one step was needed and used.\\n\\nv. The appropriate tool was used to answer the question. The 'ping' tool is designed to measure latency, which was the information the user was seeking.\\n\\nGiven these considerations, the AI language model's performance in answering this question is excellent.\"}"
"When working with language models, you may often encounter issues from the underlying APIs, whether these be rate limiting or downtime. Therefore, as you go to move your LLM applications into production it becomes more and more important to safeguard against these. That's why we've introduced the concept of fallbacks. \n",
File diff suppressed because it is too large
Load Diff
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
