- This pull request includes various changes to add a `user_agent`
parameter to Azure OpenAI, Azure Search and Whisper in the Community and
Partner packages. This helps in identifying the source of API requests
so we can better track usage and help support the community better. I
will also be adding the user_agent to the new `langchain-azure` repo as
well.
- No issue connected or updated dependencies.
- Utilises existing tests and docs
---------
Co-authored-by: Erick Friis <erick@langchain.dev>
ONNX and OpenVINO models are available by specifying the `backend`
argument (the model is loaded using `optimum`
https://github.com/huggingface/optimum)
```python
from langchain_huggingface import HuggingFaceEmbeddings
embedding = HuggingFaceEmbeddings(
model_name=model_id,
model_kwargs={"backend": "onnx"},
)
```
With this PR we also enable the IPEX backend
```python
from langchain_huggingface import HuggingFaceEmbeddings
embedding = HuggingFaceEmbeddings(
model_name=model_id,
model_kwargs={"backend": "ipex"},
)
```
**Description**
Currently, when parsing a partial JSON, if a string ends with the escape
character, the whole key/value is removed. For example:
```
>>> from langchain_core.utils.json import parse_partial_json
>>> my_str = '{"foo": "bar", "baz": "qux\\'
>>>
>>> parse_partial_json(my_str)
{'foo': 'bar'}
```
My expectation (and with this fix) would be for `parse_partial_json()`
to return:
```
>>> from langchain_core.utils.json import parse_partial_json
>>>
>>> my_str = '{"foo": "bar", "baz": "qux\\'
>>> parse_partial_json(my_str)
{'foo': 'bar', 'baz': 'qux'}
```
Notes:
1. It could be argued that current behavior is still desired.
2. I have experienced this issue when the streaming output from an LLM
and the chunk happens to end with `\\`
3. I haven't included tests. Will do if change is accepted.
4. This is specially troublesome when this function is used by
187131c55c/libs/core/langchain_core/output_parsers/transform.py (L111)
since what happens is that, for example, if the received sequence of
chunks are: `{"foo": "b` , `ar\\` :
Then, the result of calling `self.parse_result()` is:
```
{"foo": "b"}
```
and the second time:
```
{}
```
Co-authored-by: Erick Friis <erick@langchain.dev>
- **Description:** Before sending a completion chunk at the end of an
OpenAI stream, removing the tool_calls as those have already been sent
as chunks.
- **Issue:** -
- **Dependencies:** -
- **Twitter handle:** -
@ccurme as mentioned in another PR
---------
Co-authored-by: Chester Curme <chester.curme@gmail.com>
- **Description:** The llamacpp.ipynb notebook used a deprecated
environment variable, LLAMA_CUBLAS, for llama.cpp installation with GPU
support. This commit updates the notebook to use the correct GGML_CUDA
variable, fixing the installation error.
- **Issue:** none
- **Dependencies:** none
Added `similarity_search_with_score_by_vector()` function to the
`QdrantVectorStore` class.
It is required when we want to query multiple time with the same
embeddings. It was present in the now deprecated original `Qdrant`
vectorstore implementation, but was absent from the new one. It is also
implemented in a number of others `VectorStore` implementations
I have added tests for this new function
Note that I also argued in this discussion that it should be part of the
general `VectorStore`
https://github.com/langchain-ai/langchain/discussions/29638
Co-authored-by: Erick Friis <erick@langchain.dev>
Thank you for contributing to LangChain!
- [x] **PR title**: "package: description"
- Where "package" is whichever of langchain, community, core, etc. is
being modified. Use "docs: ..." for purely docs 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, ccurme, vbarda, hwchase17.
I made a change to how was implemented the support for GPU in
`FastEmbedEmbeddings` to be more consistent with the existing
implementation `langchain-qdrant` sparse embeddings implementation
It is directly enabling to provide the list of ONNX execution providers:
https://github.com/langchain-ai/langchain/blob/master/libs/partners/qdrant/langchain_qdrant/fastembed_sparse.py#L15
It is a bit less clear to a user that just wants to enable GPU, but
gives more capabilities to work with other execution providers that are
not the `CUDAExecutionProvider`, and is more future proof
Sorry for the disturbance @ccurme
> Nice to see you just moved to `uv`! It is so much nicer to run
format/lint/test! No need to manually rerun the `poetry install` with
all required extras now
These are set in Github workflows, but forgot to add them to most
makefiles for convenience when developing locally.
`uv run` will automatically sync the lock file. Because many of our
development dependencies are local installs, it will pick up version
changes and update the lock file. Passing `--frozen` or setting this
environment variable disables the behavior.
Motivation: dedicated structured output features are becoming more
common, such that integrations can support structured output without
supporting tool calling.
Here we make two changes:
1. Update the `has_structured_output` method to default to True if a
model supports tool calling (in addition to defaulting to True if
`with_structured_output` is overridden).
2. Update structured output tests to engage if `has_structured_output`
is True.
Deep Lake recently released version 4, which introduces significant
architectural changes, including a new on-disk storage format, enhanced
indexing mechanisms, and improved concurrency. However, LangChain's
vector store integration currently does not support Deep Lake v4 due to
breaking API changes.
Previously, the installation command was:
`pip install deeplake[enterprise]`
This installs the latest available version, which now defaults to Deep
Lake v4. Since LangChain's vector store integration is still dependent
on v3, this can lead to compatibility issues when using Deep Lake as a
vector database within LangChain.
To ensure compatibility, the installation command has been updated to:
`pip install deeplake[enterprise]<4.0.0`
This constraint ensures that pip installs the latest available version
of Deep Lake within the v3 series while avoiding the incompatible v4
update.
- **Description:** add a `gpu: bool = False` field to the
`FastEmbedEmbeddings` class which enables to use GPU (through ONNX CUDA
provider) when generating embeddings with any fastembed model. It just
requires the user to install a different dependency and we use a
different provider when instantiating `fastembed.TextEmbedding`
- **Issue:** when generating embeddings for a really large amount of
documents this drastically increase performance (honestly that is a must
have in some situations, you can't just use CPU it is way too slow)
- **Dependencies:** no direct change to dependencies, but internally the
users will need to install `fastembed-gpu` instead of `fastembed`, I
made all the changes to the init function to properly let the user know
which dependency they should install depending on if they enabled `gpu`
or not
cf. fastembed docs about GPU for more details:
https://qdrant.github.io/fastembed/examples/FastEmbed_GPU/
I did not added test because it would require access to a GPU in the
testing environment
### PR Title:
**community: add latest OpenAI models pricing**
### Description:
This PR updates the OpenAI model cost calculation mapping by adding the
latest OpenAI models, **o1 (non-preview)** and **o3-mini**, based on the
pricing listed on the [OpenAI pricing
page](https://platform.openai.com/docs/pricing).
### Changes:
- Added pricing for `o1`, `o1-2024-12-17`, `o1-cached`, and
`o1-2024-12-17-cached` for input tokens.
- Added pricing for `o1-completion` and `o1-2024-12-17-completion` for
output tokens.
- Added pricing for `o3-mini`, `o3-mini-2025-01-31`, `o3-mini-cached`,
and `o3-mini-2025-01-31-cached` for input tokens.
- Added pricing for `o3-mini-completion` and
`o3-mini-2025-01-31-completion` for output tokens.
### Issue:
N/A
### Dependencies:
None
### Testing & Validation:
- No functional changes outside of updating the cost mapping.
- No tests were added or modified.
**Description:**
The response from `tool.invoke()` is always a ToolMessage, with content
and artifact fields, not a tuple.
The tuple is converted to a ToolMessage here
b6ae7ca91d/libs/core/langchain_core/tools/base.py (L726)
**Issue:**
Currently `ToolsIntegrationTests` requires `invoke()` to return a tuple
and so standard tests fail for "content_and_artifact" tools. This fixes
that to check the returned ToolMessage.
This PR also adds a test that now passes.
Description: Fixes PreFilter value handling in Azure Cosmos DB NoSQL
vectorstore. The current implementation fails to handle numeric values
in filter conditions, causing an undefined value variable error. This PR
adds support for numeric, boolean, and NULL values while maintaining the
existing string and list handling.
Changes:
Added handling for numeric types (int/float)
Added boolean value support
Added NULL value handling
Added type validation for unsupported values
Fixed scope of value variable initialization
Issue:
Fixes#29610
Implementation Notes:
No changes to public API
Backwards compatible
Maintains consistent behavior with existing MongoDB-style filtering
Preserves SQL injection prevention through proper value handling
---------
Co-authored-by: Chester Curme <chester.curme@gmail.com>
This is one part of a larger Pull Request (PR) that is too large to be
submitted all at once. This specific part focuses on updating the XXX
parser.
For more details, see [PR
28970](https://github.com/langchain-ai/langchain/pull/28970).
---------
Co-authored-by: Eugene Yurtsev <eyurtsev@gmail.com>
## Description
- Removed broken link for the API Reference
- Added `OPENAI_API_KEY` setter for the chains to properly run
- renamed one of our examples so it won't override the original
retriever and cause confusion due to it using a different mode of
retrieving
- Moved one of our simple examples to be the first example of our
retriever :)
Failing with:
> ValueError: Provider page not found for databricks-langchain. Please
add one at docs/integrations/providers/databricks-langchain.{mdx,ipynb}
**PR title**: "community: Option to pass auth_file_location for
oci_generative_ai"
**Description:** Option to pass auth_file_location, to overwrite config
file default location "~/.oci/config" where profile name configs
present. This is not fixing any issues. Just added optional parameter
called "auth_file_location", which internally supported by any OCI
client including GenerativeAiInferenceClient.
- **Description:** Add to check pad_token_id and eos_token_id of model
config. It seems that this is the same bug as the HuggingFace TGI bug.
It's same bug as #29434
- **Issue:** #29431
- **Dependencies:** none
- **Twitter handle:** tell14
Example code is followings:
```python
from langchain_huggingface.llms import HuggingFacePipeline
hf = HuggingFacePipeline.from_model_id(
model_id="meta-llama/Llama-3.2-3B-Instruct",
task="text-generation",
pipeline_kwargs={"max_new_tokens": 10},
)
from langchain_core.prompts import PromptTemplate
template = """Question: {question}
Answer: Let's think step by step."""
prompt = PromptTemplate.from_template(template)
chain = prompt | hf
question = "What is electroencephalography?"
print(chain.invoke({"question": question}))
```
