From 9a717c9b32b4a7f74b8042e877c3510c282f6b57 Mon Sep 17 00:00:00 2001 From: Erick Friis Date: Thu, 21 Nov 2024 20:49:38 -0800 Subject: [PATCH] docs: poetry publish 2 (#28277) - **docs: poetry publish** - **x** - **x** - **x** - **x** - **x** - **x** - **x** - **x** - **x** --- .../how_to/integrations/index.mdx | 10 +- .../how_to/integrations/package.mdx | 65 +++++-- .../how_to/integrations/publish.mdx | 23 ++- .../how_to/integrations/standard_tests.ipynb | 171 +++++++----------- docs/docs/how_to/custom_chat_model.ipynb | 104 ++++++----- .../integration_tests/chat_models.py | 10 +- .../tests/unit_tests/custom_chat_model.py | 167 +++++++++++++++++ .../unit_tests/test_custom_chat_model.py | 30 +++ 8 files changed, 407 insertions(+), 173 deletions(-) create mode 100644 libs/standard-tests/tests/unit_tests/custom_chat_model.py create mode 100644 libs/standard-tests/tests/unit_tests/test_custom_chat_model.py diff --git a/docs/docs/contributing/how_to/integrations/index.mdx b/docs/docs/contributing/how_to/integrations/index.mdx index 53e81de384c..eeb0f73b3fd 100644 --- a/docs/docs/contributing/how_to/integrations/index.mdx +++ b/docs/docs/contributing/how_to/integrations/index.mdx @@ -66,14 +66,12 @@ that will render on this site (https://python.langchain.com/). As a prerequisite to adding your integration to our documentation, you must: 1. Confirm that your integration is in the [list of components](#components-to-integrate) we are currently accepting. -2. [Publish your package to PyPi](./package.mdx) and make the repo public. -3. [Implement the standard tests](/docs/contributing/how_to/integrations/standard_tests) for your integration and successfully run them. -3. Write documentation for your integration in the `docs/docs/integrations/` directory of the LangChain monorepo. -4. Add a provider page for your integration in the `docs/docs/integrations/providers` directory of the LangChain monorepo. +2. [Implement your package](./package.mdx) and publish it to a public github repository. +3. [Implement the standard tests](./standard_tests) for your integration and successfully run them. +4. [Publish your integration](./publish.mdx) by publishing the package to PyPi and add docs in the `docs/docs/integrations` directory of the LangChain monorepo. Once you have completed these steps, you can submit a PR to the LangChain monorepo to add your integration to the documentation. ## Further Reading -If you're starting from scratch, you can follow the [Integration Template Guide](./from_template.mdx) to create and publish a new integration package -to the above spec. +To get started, let's learn [how to bootstrap a new integration package](./package.mdx) for LangChain. diff --git a/docs/docs/contributing/how_to/integrations/package.mdx b/docs/docs/contributing/how_to/integrations/package.mdx index ae248bf6e07..99bc217c20a 100644 --- a/docs/docs/contributing/how_to/integrations/package.mdx +++ b/docs/docs/contributing/how_to/integrations/package.mdx @@ -29,10 +29,11 @@ pip install poetry Next, come up with a name for your package. For this guide, we'll use `langchain-parrot-link`. You can confirm that the name is available on PyPi by searching for it on the [PyPi website](https://pypi.org/). -Next, create your new Python package with Poetry: +Next, create your new Python package with Poetry, and navigate into the new directory with `cd`: ```bash poetry new langchain-parrot-link +cd langchain-parrot-link ``` Add main dependencies using Poetry, which will add them to your `pyproject.toml` file: @@ -48,10 +49,10 @@ with your published package, or just installing them separately when you run tes `langchain-tests` will provide the [standard tests](../standard_tests) we will use later. We recommended pinning these to the latest version: -Note: Replace `{latest version}` with the latest version of `langchain-tests` below. +Note: Replace `` with the latest version of `langchain-tests` below. ```bash -poetry add --group test pytest pytest-socket langchain-tests=={latest version} +poetry add --group test pytest pytest-socket langchain-tests== ``` You're now ready to start writing your integration package! @@ -90,12 +91,19 @@ from langchain_core.callbacks import ( CallbackManagerForLLMRun, ) from langchain_core.language_models import BaseChatModel -from langchain_core.messages import AIMessageChunk, BaseMessage, AIMessage +from langchain_core.messages import ( + AIMessage, + AIMessageChunk, + BaseMessage, +) +from langchain_core.messages.ai import UsageMetadata from langchain_core.outputs import ChatGeneration, ChatGenerationChunk, ChatResult +from pydantic import Field -class CustomChatModelAdvanced(BaseChatModel): - """A custom chat model that echoes the first `n` characters of the input. +class ChatParrotLink(BaseChatModel): + """A custom chat model that echoes the first `parrot_buffer_length` characters + of the input. When contributing an implementation to LangChain, carefully document the model including the initialization parameters, include @@ -106,16 +114,21 @@ class CustomChatModelAdvanced(BaseChatModel): .. code-block:: python - model = CustomChatModel(n=2) + model = ChatParrotLink(parrot_buffer_length=2, model="bird-brain-001") result = model.invoke([HumanMessage(content="hello")]) result = model.batch([[HumanMessage(content="hello")], [HumanMessage(content="world")]]) """ - model_name: str + model_name: str = Field(alias="model") """The name of the model""" - n: int + parrot_buffer_length: int """The number of characters from the last message of the prompt to be echoed.""" + temperature: Optional[float] = None + max_tokens: Optional[int] = None + timeout: Optional[int] = None + stop: Optional[List[str]] = None + max_retries: int = 2 def _generate( self, @@ -142,13 +155,20 @@ class CustomChatModelAdvanced(BaseChatModel): # Replace this with actual logic to generate a response from a list # of messages. last_message = messages[-1] - tokens = last_message.content[: self.n] + tokens = last_message.content[: self.parrot_buffer_length] + ct_input_tokens = sum(len(message.content) for message in messages) + ct_output_tokens = len(tokens) message = AIMessage( content=tokens, - additional_kwargs={}, # Used to add additional payload (e.g., function calling request) + additional_kwargs={}, # Used to add additional payload to the message response_metadata={ # Use for response metadata "time_in_seconds": 3, }, + usage_metadata={ + "input_tokens": ct_input_tokens, + "output_tokens": ct_output_tokens, + "total_tokens": ct_input_tokens + ct_output_tokens, + }, ) ## @@ -180,10 +200,21 @@ class CustomChatModelAdvanced(BaseChatModel): run_manager: A run manager with callbacks for the LLM. """ last_message = messages[-1] - tokens = last_message.content[: self.n] + tokens = str(last_message.content[: self.parrot_buffer_length]) + ct_input_tokens = sum(len(message.content) for message in messages) for token in tokens: - chunk = ChatGenerationChunk(message=AIMessageChunk(content=token)) + usage_metadata = UsageMetadata( + { + "input_tokens": ct_input_tokens, + "output_tokens": 1, + "total_tokens": ct_input_tokens + 1, + } + ) + ct_input_tokens = 0 + chunk = ChatGenerationChunk( + message=AIMessageChunk(content=token, usage_metadata=usage_metadata) + ) if run_manager: # This is optional in newer versions of LangChain @@ -224,6 +255,14 @@ class CustomChatModelAdvanced(BaseChatModel): ``` +## Push your package to a public Github repository + +This is only required if you want to publish your integration in the LangChain documentation. + +1. Create a new repository on GitHub. +2. Push your code to the repository. +3. Confirm that your repository is viewable by the public (e.g. in a private browsing window, where you're not logged into Github). + ## Next Steps Now that you've implemented your package, you can move on to [testing your integration](../standard_tests) for your integration and successfully run them. diff --git a/docs/docs/contributing/how_to/integrations/publish.mdx b/docs/docs/contributing/how_to/integrations/publish.mdx index 120b7bd08ea..72b5e2e3634 100644 --- a/docs/docs/contributing/how_to/integrations/publish.mdx +++ b/docs/docs/contributing/how_to/integrations/publish.mdx @@ -23,8 +23,9 @@ First, make sure you have a PyPi account and have logged in with Poetry: How to create a PyPi Token 1. Go to the [PyPi website](https://pypi.org/) and create an account. -2. Go to your account settings and enable 2FA. To generate an API token, you **must** have 2FA enabled currently. -3. Go to your account settings and [generate a new API token](https://pypi.org/manage/account/token/). +2. Verify your email address by clicking the link that PyPi emails to you. +3. Go to your account settings and click "Generate Recovery Codes" to enable 2FA. To generate an API token, you **must** have 2FA enabled currently. +4. Go to your account settings and [generate a new API token](https://pypi.org/manage/account/token/). @@ -91,7 +92,7 @@ langchain-cli integration create-doc \ --component-type Provider \ --destination-dir docs/docs/integrations/providers \ --name parrot-link \ - --name-class ParrotLink \ + --name-class ParrotLink ``` And a chat model component page: @@ -101,7 +102,7 @@ langchain-cli integration create-doc \ --component-type ChatModel \ --destination-dir docs/docs/integrations/chat \ --name parrot-link \ - --name-class ParrotLink \ + --name-class ParrotLink ``` And a vector store component page: @@ -111,14 +112,14 @@ langchain-cli integration create-doc \ --component-type VectorStore \ --destination-dir docs/docs/integrations/vectorstores \ --name parrot-link \ - --name-class ParrotLink \ + --name-class ParrotLink ``` These commands will create the following 3 files, which you should fill out with information about your package: -- `docs/docs/integrations/providers/parrot-link.ipynb` -- `docs/docs/integrations/chat/parrot-link.ipynb` -- `docs/docs/integrations/vectorstores/parrot-link.ipynb` +- `docs/docs/integrations/providers/parrot_link.ipynb` +- `docs/docs/integrations/chat/parrot_-_link.ipynb` +- `docs/docs/integrations/vectorstores/parrot_-_link.ipynb` ### Manually create your documentation pages (if you prefer) @@ -129,7 +130,7 @@ You can view the templates that the CLI uses to create these files [here](https: ### Register your package in `libs/packages.yml` -Finally, add your package to the `libs/packages.yml` file in the LangChain Monorepo. +Finally, add your package to the end of the `libs/packages.yml` file in the LangChain Monorepo. ```yaml packages: @@ -140,6 +141,10 @@ packages: For `path`, you can use `.` if your package is in the root of your repository, or specify a subdirectory (e.g. `libs/parrot-link`) if it is in a subdirectory. +If you followed the [package bootstrapping guide](../package), then your path is `.`. + ### Submit a PR with your changes Once you have completed these steps, you can submit a PR to the LangChain Monorepo with **only these changes**. + +If you have additional changes to request, please submit them in a separate PR. diff --git a/docs/docs/contributing/how_to/integrations/standard_tests.ipynb b/docs/docs/contributing/how_to/integrations/standard_tests.ipynb index c8935838874..8c1361eb349 100644 --- a/docs/docs/contributing/how_to/integrations/standard_tests.ipynb +++ b/docs/docs/contributing/how_to/integrations/standard_tests.ipynb @@ -10,10 +10,12 @@ "---\n", "# How to add standard tests to an integration\n", "\n", - "When creating either a custom class for yourself or a new tool to publish in a LangChain integration, it is important to add standard tests to ensure it works as expected. This guide will show you how to add standard tests to a tool, and you can **[Skip to the test templates](#standard-test-templates-per-component)** for implementing tests for each integration.\n", + "When creating either a custom class for yourself or to publish in a LangChain integration, it is important to add standard tests to ensure it works as expected. This guide will show you how to add standard tests to a custom chat model, and you can **[Skip to the test templates](#standard-test-templates-per-component)** for implementing tests for each integration type.\n", "\n", "## Setup\n", "\n", + "If you're coming from the [previous guide](../package), you have already installed these dependencies, and you can skip this section.\n", + "\n", "First, let's install 2 dependencies:\n", "\n", "- `langchain-core` will define the interfaces we want to import to define our custom tool.\n", @@ -53,29 +55,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "Let's say we're publishing a package, `langchain_parrot_link`, that exposes a\n", - "tool called `ParrotMultiplyTool`:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [], - "source": [ - "# title=\"langchain_parrot_link/tools.py\"\n", - "from langchain_core.tools import BaseTool\n", - "\n", - "\n", - "class ParrotMultiplyTool(BaseTool):\n", - " name: str = \"ParrotMultiplyTool\"\n", - " description: str = (\n", - " \"Multiply two numbers like a parrot. Parrots always add \"\n", - " \"eighty for their matey.\"\n", - " )\n", - "\n", - " def _run(self, a: int, b: int) -> int:\n", - " return a * b + 80" + "Let's say we're publishing a package, `langchain_parrot_link`, that exposes the chat model from the [guide on implementing the package](../package). We can add the standard tests to the package by following the steps below." ] }, { @@ -85,70 +65,33 @@ "And we'll assume you've structured your package the same way as the main LangChain\n", "packages:\n", "\n", - "```\n", - "/\n", + "```plaintext\n", + "langchain-parrot-link/\n", "├── langchain_parrot_link/\n", - "│ └── tools.py\n", - "└── tests/\n", - " ├── unit_tests/\n", - " │ └── test_tools.py\n", - " └── integration_tests/\n", - " └── test_tools.py\n", + "│ ├── __init__.py\n", + "│ └── chat_models.py\n", + "├── tests/\n", + "│ ├── __init__.py\n", + "│ └── test_chat_models.py\n", + "├── pyproject.toml\n", + "└── README.md\n", "```\n", "\n", "## Add and configure standard tests\n", "\n", "There are 2 namespaces in the `langchain-tests` package: \n", "\n", - "- [unit tests](../../../concepts/testing.mdx#unit-tests) (`langchain_tests.unit_tests`): designed to be used to test the tool in isolation and without access to external services\n", - "- [integration tests](../../../concepts/testing.mdx#unit-tests) (`langchain_tests.integration_tests`): designed to be used to test the tool with access to external services (in particular, the external service that the tool is designed to interact with).\n", + "- [unit tests](../../../concepts/testing.mdx#unit-tests) (`langchain_tests.unit_tests`): designed to be used to test the component in isolation and without access to external services\n", + "- [integration tests](../../../concepts/testing.mdx#unit-tests) (`langchain_tests.integration_tests`): designed to be used to test the component with access to external services (in particular, the external service that the component is designed to interact with).\n", "\n", "Both types of tests are implemented as [`pytest` class-based test suites](https://docs.pytest.org/en/7.1.x/getting-started.html#group-multiple-tests-in-a-class).\n", "\n", "By subclassing the base classes for each type of standard test (see below), you get all of the standard tests for that type, and you\n", "can override the properties that the test suite uses to configure the tests.\n", "\n", - "### Standard tools tests\n", + "### Standard chat model tests\n", "\n", - "Here's how you would configure the standard unit tests for the custom tool, e.g. in `tests/test_tools.py`:" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": { - "title": "tests/test_custom_tool.py" - }, - "outputs": [], - "source": [ - "# title=\"tests/unit_tests/test_tools.py\"\n", - "from typing import Type\n", - "\n", - "from langchain_parrot_link.tools import ParrotMultiplyTool\n", - "from langchain_tests.unit_tests import ToolsUnitTests\n", - "\n", - "\n", - "class TestParrotMultiplyToolUnit(ToolsUnitTests):\n", - " @property\n", - " def tool_constructor(self) -> Type[ParrotMultiplyTool]:\n", - " return ParrotMultiplyTool\n", - "\n", - " @property\n", - " def tool_constructor_params(self) -> dict:\n", - " # if your tool constructor instead required initialization arguments like\n", - " # `def __init__(self, some_arg: int):`, you would return those here\n", - " # as a dictionary, e.g.: `return {'some_arg': 42}`\n", - " return {}\n", - "\n", - " @property\n", - " def tool_invoke_params_example(self) -> dict:\n", - " \"\"\"\n", - " Returns a dictionary representing the \"args\" of an example tool call.\n", - "\n", - " This should NOT be a ToolCall dict - i.e. it should not\n", - " have {\"name\", \"id\", \"args\"} keys.\n", - " \"\"\"\n", - " return {\"a\": 2, \"b\": 3}" + "Here's how you would configure the standard unit tests for the custom chat model:" ] }, { @@ -157,34 +100,52 @@ "metadata": {}, "outputs": [], "source": [ - "# title=\"tests/integration_tests/test_tools.py\"\n", + "# title=\"tests/unit_tests/test_chat_models.py\"\n", + "from typing import Tuple, Type\n", + "\n", + "from langchain_parrot_link.chat_models import ChatParrotLink\n", + "from langchain_tests.unit_tests import ChatModelUnitTests\n", + "\n", + "\n", + "class TestChatParrotLinkUnit(ChatModelUnitTests):\n", + " @property\n", + " def chat_model_class(self) -> Type[ChatParrotLink]:\n", + " return ChatParrotLink\n", + "\n", + " @property\n", + " def chat_model_params(self) -> dict:\n", + " return {\n", + " \"model\": \"bird-brain-001\",\n", + " \"temperature\": 0,\n", + " \"parrot_buffer_length\": 50,\n", + " }" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "# title=\"tests/integration_tests/test_chat_models.py\"\n", "from typing import Type\n", "\n", - "from langchain_parrot_link.tools import ParrotMultiplyTool\n", - "from langchain_tests.integration_tests import ToolsIntegrationTests\n", + "from langchain_parrot_link.chat_models import ChatParrotLink\n", + "from langchain_tests.integration_tests import ChatModelIntegrationTests\n", "\n", "\n", - "class TestParrotMultiplyToolIntegration(ToolsIntegrationTests):\n", + "class TestChatParrotLinkIntegration(ChatModelIntegrationTests):\n", " @property\n", - " def tool_constructor(self) -> Type[ParrotMultiplyTool]:\n", - " return ParrotMultiplyTool\n", + " def chat_model_class(self) -> Type[ChatParrotLink]:\n", + " return ChatParrotLink\n", "\n", " @property\n", - " def tool_constructor_params(self) -> dict:\n", - " # if your tool constructor instead required initialization arguments like\n", - " # `def __init__(self, some_arg: int):`, you would return those here\n", - " # as a dictionary, e.g.: `return {'some_arg': 42}`\n", - " return {}\n", - "\n", - " @property\n", - " def tool_invoke_params_example(self) -> dict:\n", - " \"\"\"\n", - " Returns a dictionary representing the \"args\" of an example tool call.\n", - "\n", - " This should NOT be a ToolCall dict - i.e. it should not\n", - " have {\"name\", \"id\", \"args\"} keys.\n", - " \"\"\"\n", - " return {\"a\": 2, \"b\": 3}" + " def chat_model_params(self) -> dict:\n", + " return {\n", + " \"model\": \"bird-brain-001\",\n", + " \"temperature\": 0,\n", + " \"parrot_buffer_length\": 50,\n", + " }" ] }, { @@ -228,7 +189,8 @@ "Above, we implement the **unit** and **integration** standard tests for a tool. Below are the templates for implementing the standard tests for each component:\n", "\n", "
\n", - " Chat Models" + " Chat Models\n", + "

Note: The standard tests for chat models are implemented in the example in the main body of this guide too.

" ] }, { @@ -238,7 +200,7 @@ "outputs": [], "source": [ "# title=\"tests/unit_tests/test_chat_models.py\"\n", - "from typing import Tuple, Type\n", + "from typing import Type\n", "\n", "from langchain_parrot_link.chat_models import ChatParrotLink\n", "from langchain_tests.unit_tests import ChatModelUnitTests\n", @@ -251,7 +213,11 @@ "\n", " @property\n", " def chat_model_params(self) -> dict:\n", - " return {\"model\": \"bird-brain-001\", \"temperature\": 0}" + " return {\n", + " \"model\": \"bird-brain-001\",\n", + " \"temperature\": 0,\n", + " \"parrot_buffer_length\": 50,\n", + " }" ] }, { @@ -274,7 +240,11 @@ "\n", " @property\n", " def chat_model_params(self) -> dict:\n", - " return {\"model\": \"bird-brain-001\", \"temperature\": 0}" + " return {\n", + " \"model\": \"bird-brain-001\",\n", + " \"temperature\": 0,\n", + " \"parrot_buffer_length\": 50,\n", + " }" ] }, { @@ -338,8 +308,7 @@ "source": [ "
\n", "
\n", - " Tools/Toolkits\n", - "

Note: The standard tests for tools/toolkits are implemented in the example in the main body of this guide too.

" + " Tools/Toolkits" ] }, { diff --git a/docs/docs/how_to/custom_chat_model.ipynb b/docs/docs/how_to/custom_chat_model.ipynb index e4a126ccef8..36ff587e11e 100644 --- a/docs/docs/how_to/custom_chat_model.ipynb +++ b/docs/docs/how_to/custom_chat_model.ipynb @@ -48,7 +48,7 @@ }, { "cell_type": "code", - "execution_count": 1, + "execution_count": 4, "id": "c5046e6a-8b09-4a99-b6e6-7a605aac5738", "metadata": { "tags": [] @@ -175,12 +175,19 @@ " CallbackManagerForLLMRun,\n", ")\n", "from langchain_core.language_models import BaseChatModel\n", - "from langchain_core.messages import AIMessageChunk, BaseMessage, HumanMessage\n", + "from langchain_core.messages import (\n", + " AIMessage,\n", + " AIMessageChunk,\n", + " BaseMessage,\n", + ")\n", + "from langchain_core.messages.ai import UsageMetadata\n", "from langchain_core.outputs import ChatGeneration, ChatGenerationChunk, ChatResult\n", + "from pydantic import Field\n", "\n", "\n", - "class CustomChatModelAdvanced(BaseChatModel):\n", - " \"\"\"A custom chat model that echoes the first `n` characters of the input.\n", + "class ChatParrotLink(BaseChatModel):\n", + " \"\"\"A custom chat model that echoes the first `parrot_buffer_length` characters\n", + " of the input.\n", "\n", " When contributing an implementation to LangChain, carefully document\n", " the model including the initialization parameters, include\n", @@ -191,16 +198,21 @@ "\n", " .. code-block:: python\n", "\n", - " model = CustomChatModel(n=2)\n", + " model = ChatParrotLink(parrot_buffer_length=2, model=\"bird-brain-001\")\n", " result = model.invoke([HumanMessage(content=\"hello\")])\n", " result = model.batch([[HumanMessage(content=\"hello\")],\n", " [HumanMessage(content=\"world\")]])\n", " \"\"\"\n", "\n", - " model_name: str\n", + " model_name: str = Field(alias=\"model\")\n", " \"\"\"The name of the model\"\"\"\n", - " n: int\n", + " parrot_buffer_length: int\n", " \"\"\"The number of characters from the last message of the prompt to be echoed.\"\"\"\n", + " temperature: Optional[float] = None\n", + " max_tokens: Optional[int] = None\n", + " timeout: Optional[int] = None\n", + " stop: Optional[List[str]] = None\n", + " max_retries: int = 2\n", "\n", " def _generate(\n", " self,\n", @@ -227,13 +239,20 @@ " # Replace this with actual logic to generate a response from a list\n", " # of messages.\n", " last_message = messages[-1]\n", - " tokens = last_message.content[: self.n]\n", + " tokens = last_message.content[: self.parrot_buffer_length]\n", + " ct_input_tokens = sum(len(message.content) for message in messages)\n", + " ct_output_tokens = len(tokens)\n", " message = AIMessage(\n", " content=tokens,\n", - " additional_kwargs={}, # Used to add additional payload (e.g., function calling request)\n", + " additional_kwargs={}, # Used to add additional payload to the message\n", " response_metadata={ # Use for response metadata\n", " \"time_in_seconds\": 3,\n", " },\n", + " usage_metadata={\n", + " \"input_tokens\": ct_input_tokens,\n", + " \"output_tokens\": ct_output_tokens,\n", + " \"total_tokens\": ct_input_tokens + ct_output_tokens,\n", + " },\n", " )\n", " ##\n", "\n", @@ -265,10 +284,21 @@ " run_manager: A run manager with callbacks for the LLM.\n", " \"\"\"\n", " last_message = messages[-1]\n", - " tokens = last_message.content[: self.n]\n", + " tokens = str(last_message.content[: self.parrot_buffer_length])\n", + " ct_input_tokens = sum(len(message.content) for message in messages)\n", "\n", " for token in tokens:\n", - " chunk = ChatGenerationChunk(message=AIMessageChunk(content=token))\n", + " usage_metadata = UsageMetadata(\n", + " {\n", + " \"input_tokens\": ct_input_tokens,\n", + " \"output_tokens\": 1,\n", + " \"total_tokens\": ct_input_tokens + 1,\n", + " }\n", + " )\n", + " ct_input_tokens = 0\n", + " chunk = ChatGenerationChunk(\n", + " message=AIMessageChunk(content=token, usage_metadata=usage_metadata)\n", + " )\n", "\n", " if run_manager:\n", " # This is optional in newer versions of LangChain\n", @@ -320,7 +350,7 @@ }, { "cell_type": "code", - "execution_count": 6, + "execution_count": 5, "id": "27689f30-dcd2-466b-ba9d-f60b7d434110", "metadata": { "tags": [] @@ -329,16 +359,16 @@ { "data": { "text/plain": [ - "AIMessage(content='Meo', response_metadata={'time_in_seconds': 3}, id='run-ddb42bd6-4fdd-4bd2-8be5-e11b67d3ac29-0')" + "AIMessage(content='Meo', additional_kwargs={}, response_metadata={'time_in_seconds': 3}, id='run-cf11aeb6-8ab6-43d7-8c68-c1ef89b6d78e-0', usage_metadata={'input_tokens': 26, 'output_tokens': 3, 'total_tokens': 29})" ] }, - "execution_count": 6, + "execution_count": 5, "metadata": {}, "output_type": "execute_result" } ], "source": [ - "model = CustomChatModelAdvanced(n=3, model_name=\"my_custom_model\")\n", + "model = ChatParrotLink(parrot_buffer_length=3, model=\"my_custom_model\")\n", "\n", "model.invoke(\n", " [\n", @@ -351,7 +381,7 @@ }, { "cell_type": "code", - "execution_count": 7, + "execution_count": 6, "id": "406436df-31bf-466b-9c3d-39db9d6b6407", "metadata": { "tags": [] @@ -360,10 +390,10 @@ { "data": { "text/plain": [ - "AIMessage(content='hel', response_metadata={'time_in_seconds': 3}, id='run-4d3cc912-44aa-454b-977b-ca02be06c12e-0')" + "AIMessage(content='hel', additional_kwargs={}, response_metadata={'time_in_seconds': 3}, id='run-618e5ed4-d611-4083-8cf1-c270726be8d9-0', usage_metadata={'input_tokens': 5, 'output_tokens': 3, 'total_tokens': 8})" ] }, - "execution_count": 7, + "execution_count": 6, "metadata": {}, "output_type": "execute_result" } @@ -374,7 +404,7 @@ }, { "cell_type": "code", - "execution_count": 8, + "execution_count": 7, "id": "a72ffa46-6004-41ef-bbe4-56fa17a029e2", "metadata": { "tags": [] @@ -383,11 +413,11 @@ { "data": { "text/plain": [ - "[AIMessage(content='hel', response_metadata={'time_in_seconds': 3}, id='run-9620e228-1912-4582-8aa1-176813afec49-0'),\n", - " AIMessage(content='goo', response_metadata={'time_in_seconds': 3}, id='run-1ce8cdf8-6f75-448e-82f7-1bb4a121df93-0')]" + "[AIMessage(content='hel', additional_kwargs={}, response_metadata={'time_in_seconds': 3}, id='run-eea4ed7d-d750-48dc-90c0-7acca1ff388f-0', usage_metadata={'input_tokens': 5, 'output_tokens': 3, 'total_tokens': 8}),\n", + " AIMessage(content='goo', additional_kwargs={}, response_metadata={'time_in_seconds': 3}, id='run-07cfc5c1-3c62-485f-b1e0-3d46e1547287-0', usage_metadata={'input_tokens': 7, 'output_tokens': 3, 'total_tokens': 10})]" ] }, - "execution_count": 8, + "execution_count": 7, "metadata": {}, "output_type": "execute_result" } @@ -398,7 +428,7 @@ }, { "cell_type": "code", - "execution_count": 9, + "execution_count": 8, "id": "3633be2c-2ea0-42f9-a72f-3b5240690b55", "metadata": { "tags": [] @@ -427,7 +457,7 @@ }, { "cell_type": "code", - "execution_count": 10, + "execution_count": 9, "id": "b7d73995-eeab-48c6-a7d8-32c98ba29fc2", "metadata": { "tags": [] @@ -456,7 +486,7 @@ }, { "cell_type": "code", - "execution_count": 11, + "execution_count": 10, "id": "17840eba-8ff4-4e73-8e4f-85f16eb1c9d0", "metadata": { "tags": [] @@ -466,20 +496,12 @@ "name": "stdout", "output_type": "stream", "text": [ - "{'event': 'on_chat_model_start', 'run_id': '125a2a16-b9cd-40de-aa08-8aa9180b07d0', 'name': 'CustomChatModelAdvanced', 'tags': [], 'metadata': {}, 'data': {'input': 'cat'}}\n", - "{'event': 'on_chat_model_stream', 'run_id': '125a2a16-b9cd-40de-aa08-8aa9180b07d0', 'tags': [], 'metadata': {}, 'name': 'CustomChatModelAdvanced', 'data': {'chunk': AIMessageChunk(content='c', id='run-125a2a16-b9cd-40de-aa08-8aa9180b07d0')}}\n", - "{'event': 'on_chat_model_stream', 'run_id': '125a2a16-b9cd-40de-aa08-8aa9180b07d0', 'tags': [], 'metadata': {}, 'name': 'CustomChatModelAdvanced', 'data': {'chunk': AIMessageChunk(content='a', id='run-125a2a16-b9cd-40de-aa08-8aa9180b07d0')}}\n", - "{'event': 'on_chat_model_stream', 'run_id': '125a2a16-b9cd-40de-aa08-8aa9180b07d0', 'tags': [], 'metadata': {}, 'name': 'CustomChatModelAdvanced', 'data': {'chunk': AIMessageChunk(content='t', id='run-125a2a16-b9cd-40de-aa08-8aa9180b07d0')}}\n", - "{'event': 'on_chat_model_stream', 'run_id': '125a2a16-b9cd-40de-aa08-8aa9180b07d0', 'tags': [], 'metadata': {}, 'name': 'CustomChatModelAdvanced', 'data': {'chunk': AIMessageChunk(content='', response_metadata={'time_in_sec': 3}, id='run-125a2a16-b9cd-40de-aa08-8aa9180b07d0')}}\n", - "{'event': 'on_chat_model_end', 'name': 'CustomChatModelAdvanced', 'run_id': '125a2a16-b9cd-40de-aa08-8aa9180b07d0', 'tags': [], 'metadata': {}, 'data': {'output': AIMessageChunk(content='cat', response_metadata={'time_in_sec': 3}, id='run-125a2a16-b9cd-40de-aa08-8aa9180b07d0')}}\n" - ] - }, - { - "name": "stderr", - "output_type": "stream", - "text": [ - "/home/eugene/src/langchain/libs/core/langchain_core/_api/beta_decorator.py:87: LangChainBetaWarning: This API is in beta and may change in the future.\n", - " warn_beta(\n" + "{'event': 'on_chat_model_start', 'run_id': '3f0b5501-5c78-45b3-92fc-8322a6a5024a', 'name': 'ChatParrotLink', 'tags': [], 'metadata': {}, 'data': {'input': 'cat'}, 'parent_ids': []}\n", + "{'event': 'on_chat_model_stream', 'run_id': '3f0b5501-5c78-45b3-92fc-8322a6a5024a', 'tags': [], 'metadata': {}, 'name': 'ChatParrotLink', 'data': {'chunk': AIMessageChunk(content='c', additional_kwargs={}, response_metadata={}, id='run-3f0b5501-5c78-45b3-92fc-8322a6a5024a', usage_metadata={'input_tokens': 3, 'output_tokens': 1, 'total_tokens': 4})}, 'parent_ids': []}\n", + "{'event': 'on_chat_model_stream', 'run_id': '3f0b5501-5c78-45b3-92fc-8322a6a5024a', 'tags': [], 'metadata': {}, 'name': 'ChatParrotLink', 'data': {'chunk': AIMessageChunk(content='a', additional_kwargs={}, response_metadata={}, id='run-3f0b5501-5c78-45b3-92fc-8322a6a5024a', usage_metadata={'input_tokens': 0, 'output_tokens': 1, 'total_tokens': 1})}, 'parent_ids': []}\n", + "{'event': 'on_chat_model_stream', 'run_id': '3f0b5501-5c78-45b3-92fc-8322a6a5024a', 'tags': [], 'metadata': {}, 'name': 'ChatParrotLink', 'data': {'chunk': AIMessageChunk(content='t', additional_kwargs={}, response_metadata={}, id='run-3f0b5501-5c78-45b3-92fc-8322a6a5024a', usage_metadata={'input_tokens': 0, 'output_tokens': 1, 'total_tokens': 1})}, 'parent_ids': []}\n", + "{'event': 'on_chat_model_stream', 'run_id': '3f0b5501-5c78-45b3-92fc-8322a6a5024a', 'tags': [], 'metadata': {}, 'name': 'ChatParrotLink', 'data': {'chunk': AIMessageChunk(content='', additional_kwargs={}, response_metadata={'time_in_sec': 3}, id='run-3f0b5501-5c78-45b3-92fc-8322a6a5024a')}, 'parent_ids': []}\n", + "{'event': 'on_chat_model_end', 'name': 'ChatParrotLink', 'run_id': '3f0b5501-5c78-45b3-92fc-8322a6a5024a', 'tags': [], 'metadata': {}, 'data': {'output': AIMessageChunk(content='cat', additional_kwargs={}, response_metadata={'time_in_sec': 3}, id='run-3f0b5501-5c78-45b3-92fc-8322a6a5024a', usage_metadata={'input_tokens': 3, 'output_tokens': 3, 'total_tokens': 6})}, 'parent_ids': []}\n" ] } ], @@ -545,7 +567,7 @@ ], "metadata": { "kernelspec": { - "display_name": "Python 3 (ipykernel)", + "display_name": ".venv", "language": "python", "name": "python3" }, @@ -559,7 +581,7 @@ "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", - "version": "3.9.1" + "version": "3.11.4" } }, "nbformat": 4, diff --git a/libs/standard-tests/langchain_tests/integration_tests/chat_models.py b/libs/standard-tests/langchain_tests/integration_tests/chat_models.py index f7608469c04..61f45e63b9b 100644 --- a/libs/standard-tests/langchain_tests/integration_tests/chat_models.py +++ b/libs/standard-tests/langchain_tests/integration_tests/chat_models.py @@ -493,9 +493,13 @@ class ChatModelIntegrationTests(ChatModelTests): message=AIMessage( content="Output text", usage_metadata={ - "input_tokens": 0, - "output_tokens": 240, - "total_tokens": 590, + "input_tokens": ( + num_input_tokens if is_first_chunk else 0 + ), + "output_tokens": 11, + "total_tokens": ( + 11+num_input_tokens if is_first_chunk else 11 + ), "input_token_details": { "audio": 10, "cache_creation": 200, diff --git a/libs/standard-tests/tests/unit_tests/custom_chat_model.py b/libs/standard-tests/tests/unit_tests/custom_chat_model.py new file mode 100644 index 00000000000..1791138cf35 --- /dev/null +++ b/libs/standard-tests/tests/unit_tests/custom_chat_model.py @@ -0,0 +1,167 @@ +from typing import Any, Dict, Iterator, List, Optional + +from langchain_core.callbacks import ( + CallbackManagerForLLMRun, +) +from langchain_core.language_models import BaseChatModel +from langchain_core.messages import ( + AIMessage, + AIMessageChunk, + BaseMessage, +) +from langchain_core.messages.ai import UsageMetadata +from langchain_core.outputs import ChatGeneration, ChatGenerationChunk, ChatResult +from pydantic import Field + + +class ChatParrotLink(BaseChatModel): + """A custom chat model that echoes the first `parrot_buffer_length` characters + of the input. + + When contributing an implementation to LangChain, carefully document + the model including the initialization parameters, include + an example of how to initialize the model and include any relevant + links to the underlying models documentation or API. + + Example: + + .. code-block:: python + + model = ChatParrotLink(parrot_buffer_length=2, model="bird-brain-001") + result = model.invoke([HumanMessage(content="hello")]) + result = model.batch([[HumanMessage(content="hello")], + [HumanMessage(content="world")]]) + """ + + model_name: str = Field(alias="model") + """The name of the model""" + parrot_buffer_length: int + """The number of characters from the last message of the prompt to be echoed.""" + temperature: Optional[float] = None + max_tokens: Optional[int] = None + timeout: Optional[int] = None + stop: Optional[List[str]] = None + max_retries: int = 2 + + def _generate( + self, + messages: List[BaseMessage], + stop: Optional[List[str]] = None, + run_manager: Optional[CallbackManagerForLLMRun] = None, + **kwargs: Any, + ) -> ChatResult: + """Override the _generate method to implement the chat model logic. + + This can be a call to an API, a call to a local model, or any other + implementation that generates a response to the input prompt. + + Args: + messages: the prompt composed of a list of messages. + stop: a list of strings on which the model should stop generating. + If generation stops due to a stop token, the stop token itself + SHOULD BE INCLUDED as part of the output. This is not enforced + across models right now, but it's a good practice to follow since + it makes it much easier to parse the output of the model + downstream and understand why generation stopped. + run_manager: A run manager with callbacks for the LLM. + """ + # Replace this with actual logic to generate a response from a list + # of messages. + last_message = messages[-1] + tokens = last_message.content[: self.parrot_buffer_length] + ct_input_tokens = sum(len(message.content) for message in messages) + ct_output_tokens = len(tokens) + message = AIMessage( + content=tokens, + additional_kwargs={}, # Used to add additional payload to the message + response_metadata={ # Use for response metadata + "time_in_seconds": 3, + }, + usage_metadata={ + "input_tokens": ct_input_tokens, + "output_tokens": ct_output_tokens, + "total_tokens": ct_input_tokens + ct_output_tokens, + }, + ) + ## + + generation = ChatGeneration(message=message) + return ChatResult(generations=[generation]) + + def _stream( + self, + messages: List[BaseMessage], + stop: Optional[List[str]] = None, + run_manager: Optional[CallbackManagerForLLMRun] = None, + **kwargs: Any, + ) -> Iterator[ChatGenerationChunk]: + """Stream the output of the model. + + This method should be implemented if the model can generate output + in a streaming fashion. If the model does not support streaming, + do not implement it. In that case streaming requests will be automatically + handled by the _generate method. + + Args: + messages: the prompt composed of a list of messages. + stop: a list of strings on which the model should stop generating. + If generation stops due to a stop token, the stop token itself + SHOULD BE INCLUDED as part of the output. This is not enforced + across models right now, but it's a good practice to follow since + it makes it much easier to parse the output of the model + downstream and understand why generation stopped. + run_manager: A run manager with callbacks for the LLM. + """ + last_message = messages[-1] + tokens = str(last_message.content[: self.parrot_buffer_length]) + ct_input_tokens = sum(len(message.content) for message in messages) + + for token in tokens: + usage_metadata = UsageMetadata( + { + "input_tokens": ct_input_tokens, + "output_tokens": 1, + "total_tokens": ct_input_tokens + 1, + } + ) + ct_input_tokens = 0 + chunk = ChatGenerationChunk( + message=AIMessageChunk(content=token, usage_metadata=usage_metadata) + ) + + if run_manager: + # This is optional in newer versions of LangChain + # The on_llm_new_token will be called automatically + run_manager.on_llm_new_token(token, chunk=chunk) + + yield chunk + + # Let's add some other information (e.g., response metadata) + chunk = ChatGenerationChunk( + message=AIMessageChunk(content="", response_metadata={"time_in_sec": 3}) + ) + if run_manager: + # This is optional in newer versions of LangChain + # The on_llm_new_token will be called automatically + run_manager.on_llm_new_token(token, chunk=chunk) + yield chunk + + @property + def _llm_type(self) -> str: + """Get the type of language model used by this chat model.""" + return "echoing-chat-model-advanced" + + @property + def _identifying_params(self) -> Dict[str, Any]: + """Return a dictionary of identifying parameters. + + This information is used by the LangChain callback system, which + is used for tracing purposes make it possible to monitor LLMs. + """ + return { + # The model name allows users to specify custom token counting + # rules in LLM monitoring applications (e.g., in LangSmith users + # can provide per token pricing for their model and monitor + # costs for the given LLM.) + "model_name": self.model_name, + } diff --git a/libs/standard-tests/tests/unit_tests/test_custom_chat_model.py b/libs/standard-tests/tests/unit_tests/test_custom_chat_model.py new file mode 100644 index 00000000000..878c589c813 --- /dev/null +++ b/libs/standard-tests/tests/unit_tests/test_custom_chat_model.py @@ -0,0 +1,30 @@ +""" +Test the standard tests on the custom chat model in the docs +""" + +from typing import Type + +from langchain_tests.integration_tests import ChatModelIntegrationTests +from langchain_tests.unit_tests import ChatModelUnitTests + +from .custom_chat_model import ChatParrotLink + + +class TestChatParrotLinkUnit(ChatModelUnitTests): + @property + def chat_model_class(self) -> Type[ChatParrotLink]: + return ChatParrotLink + + @property + def chat_model_params(self) -> dict: + return {"model": "bird-brain-001", "temperature": 0, "parrot_buffer_length": 50} + + +class TestChatParrotLinkIntegration(ChatModelIntegrationTests): + @property + def chat_model_class(self) -> Type[ChatParrotLink]: + return ChatParrotLink + + @property + def chat_model_params(self) -> dict: + return {"model": "bird-brain-001", "temperature": 0, "parrot_buffer_length": 50}