langchain 0.1.4¶ langchain.agents¶
Agent is a class that uses an LLM to choose a sequence of actions to take.
In Chains, a sequence of actions is hardcoded. In Agents, a language model is used as a reasoning engine to determine which actions to take and in which order.
Agents select and use Tools and Toolkits for actions.
Class hierarchy:
BaseSingleActionAgent --> LLMSingleActionAgent
OpenAIFunctionsAgent
XMLAgent
Agent --> <name>Agent # Examples: ZeroShotAgent, ChatAgent
BaseMultiActionAgent --> OpenAIMultiFunctionsAgent
Main helpers:
AgentType, AgentExecutor, AgentOutputParser, AgentExecutorIterator, AgentAction, AgentFinishClasses¶ Functions¶
A convenience method for creating a conversational retrieval agent.
agents.agent_toolkits.vectorstore.base.create_vectorstore_agent(...)
Construct a VectorStore agent from an LLM and tools.
agents.agent_toolkits.vectorstore.base.create_vectorstore_router_agent(...)
Construct a VectorStore router agent from an LLM and tools.
agents.format_scratchpad.log.format_log_to_str(...)
Construct the scratchpad that lets the agent continue its thought process.
agents.format_scratchpad.log_to_messages.format_log_to_messages(...)
Construct the scratchpad that lets the agent continue its thought process.
agents.format_scratchpad.openai_functions.format_to_openai_function_messages(...)
Convert (AgentAction, tool output) tuples into FunctionMessages.
agents.format_scratchpad.openai_functions.format_to_openai_functions(...)
Convert (AgentAction, tool output) tuples into FunctionMessages.
agents.format_scratchpad.openai_tools.format_to_openai_tool_messages(...)
Convert (AgentAction, tool output) tuples into FunctionMessages.
agents.format_scratchpad.xml.format_xml(...)
Format the intermediate steps as XML.
agents.initialize.initialize_agent(tools, llm)
[Deprecated] Load an agent executor given tools and LLM.
agents.json_chat.base.create_json_chat_agent(...)
Create an agent that uses JSON to format its logic, build for Chat Models.
agents.load_tools.get_all_tool_names()
Get a list of all possible tool names.
agents.load_tools.load_huggingface_tool(...)
Loads a tool from the HuggingFace Hub.
agents.load_tools.load_tools(tool_names[, ...])
Load tools based on their name.
agents.loading.load_agent(path, **kwargs)
[Deprecated] Unified method for loading an agent from LangChainHub or local fs.
agents.loading.load_agent_from_config(config)
[Deprecated] Load agent from Config Dict.
agents.openai_functions_agent.base.create_openai_functions_agent(...)
Create an agent that uses OpenAI function calling.
agents.openai_tools.base.create_openai_tools_agent(...)
Create an agent that uses OpenAI tools.
agents.output_parsers.openai_tools.parse_ai_message_to_openai_tool_action(message)
Parse an AI message potentially containing tool_calls.
agents.react.agent.create_react_agent(llm, ...)
Create an agent that uses ReAct prompting.
agents.self_ask_with_search.base.create_self_ask_with_search_agent(...)
Create an agent that uses self-ask with search prompting.
agents.structured_chat.base.create_structured_chat_agent(...)
Create an agent aimed at supporting tools with multiple inputs.
agents.utils.validate_tools_single_input(...)
Validate tools for single input.
agents.xml.base.create_xml_agent(llm, tools, ...)
Create an agent that uses XML to format its logic.
langchain.callbacks¶
Callback handlers allow listening to events in LangChain.
Class hierarchy:
BaseCallbackHandler --> <name>CallbackHandler # Example: AimCallbackHandlerClasses¶
langchain.chains¶
Chains are easily reusable components linked together.
Chains encode a sequence of calls to components like models, document retrievers, other Chains, etc., and provide a simple interface to this sequence.
The Chain interface makes it easy to create apps that are:
Stateful: add Memory to any Chain to give it state,
Observable: pass Callbacks to a Chain to execute additional functionality, like logging, outside the main sequence of component calls,
Composable: combine Chains with other components, including other Chains.
Class hierarchy:
Chain --> <name>Chain # Examples: LLMChain, MapReduceChain, RouterChainClasses¶
Chain that makes API calls and summarizes the responses to answer a question.
chains.api.openapi.chain.OpenAPIEndpointChain
Chain interacts with an OpenAPI endpoint using natural language.
chains.api.openapi.requests_chain.APIRequesterChain
Get the request parser.
chains.api.openapi.requests_chain.APIRequesterOutputParser
Parse the request and error tags.
chains.api.openapi.response_chain.APIResponderChain
Get the response parser.
chains.api.openapi.response_chain.APIResponderOutputParser
Parse the response and error tags.
Abstract base class for creating structured sequences of calls to components.
chains.combine_documents.base.AnalyzeDocumentChain
Chain that splits documents, then analyzes it in pieces.
chains.combine_documents.base.BaseCombineDocumentsChain
Base interface for chains combining documents.
chains.combine_documents.map_reduce.MapReduceDocumentsChain
Combining documents by mapping a chain over them, then combining results.
chains.combine_documents.map_rerank.MapRerankDocumentsChain
Combining documents by mapping a chain over them, then reranking results.
chains.combine_documents.reduce.AsyncCombineDocsProtocol(...)
Interface for the combine_docs method.
chains.combine_documents.reduce.CombineDocsProtocol(...)
Interface for the combine_docs method.
chains.combine_documents.reduce.ReduceDocumentsChain
Combine documents by recursively reducing them.
chains.combine_documents.refine.RefineDocumentsChain
Combine documents by doing a first pass and then refining on more documents.
chains.combine_documents.stuff.StuffDocumentsChain
Chain that combines documents by stuffing into context.
chains.constitutional_ai.base.ConstitutionalChain
Chain for applying constitutional principles.
chains.constitutional_ai.models.ConstitutionalPrinciple
Class for a constitutional principle.
chains.conversation.base.ConversationChain
Chain to have a conversation and load context from memory.
chains.conversational_retrieval.base.BaseConversationalRetrievalChain
Chain for chatting with an index.
chains.conversational_retrieval.base.ChatVectorDBChain
Chain for chatting with a vector database.
chains.conversational_retrieval.base.ConversationalRetrievalChain
Chain for having a conversation based on retrieved documents.
chains.conversational_retrieval.base.InputType
Input type for ConversationalRetrievalChain.
chains.elasticsearch_database.base.ElasticsearchDatabaseChain
Chain for interacting with Elasticsearch Database.
Chain that combines a retriever, a question generator, and a response generator.
chains.flare.base.QuestionGeneratorChain
Chain that generates questions from uncertain spans.
chains.flare.prompts.FinishedOutputParser
Output parser that checks if the output is finished.
chains.graph_qa.arangodb.ArangoGraphQAChain
Chain for question-answering against a graph by generating AQL statements.
chains.graph_qa.base.GraphQAChain
Chain for question-answering against a graph.
chains.graph_qa.cypher.GraphCypherQAChain
Chain for question-answering against a graph by generating Cypher statements.
chains.graph_qa.cypher_utils.CypherQueryCorrector(schemas)
Used to correct relationship direction in generated Cypher statements.
chains.graph_qa.cypher_utils.Schema(...)
Create new instance of Schema(left_node, relation, right_node)
chains.graph_qa.falkordb.FalkorDBQAChain
Chain for question-answering against a graph by generating Cypher statements.
chains.graph_qa.hugegraph.HugeGraphQAChain
Chain for question-answering against a graph by generating gremlin statements.
chains.graph_qa.kuzu.KuzuQAChain
Question-answering against a graph by generating Cypher statements for Kùzu.
chains.graph_qa.nebulagraph.NebulaGraphQAChain
Chain for question-answering against a graph by generating nGQL statements.
chains.graph_qa.neptune_cypher.NeptuneOpenCypherQAChain
Chain for question-answering against a Neptune graph by generating openCypher statements.
chains.graph_qa.sparql.GraphSparqlQAChain
Question-answering against an RDF or OWL graph by generating SPARQL statements.
chains.hyde.base.HypotheticalDocumentEmbedder
Generate hypothetical document for query, and then embed that.
Chain to run queries against LLMs.
chains.llm_checker.base.LLMCheckerChain
Chain for question-answering with self-verification.
chains.llm_math.base.LLMMathChain
Chain that interprets a prompt and executes python code to do math.
chains.llm_requests.LLMRequestsChain
Chain that requests a URL and then uses an LLM to parse results.
chains.llm_summarization_checker.base.LLMSummarizationCheckerChain
Chain for question-answering with self-verification.
chains.mapreduce.MapReduceChain
Map-reduce chain.
chains.moderation.OpenAIModerationChain
Pass input through a moderation endpoint.
chains.natbot.base.NatBotChain
Implement an LLM driven browser.
chains.natbot.crawler.Crawler()
A crawler for web pages.
chains.natbot.crawler.ElementInViewPort
A typed dictionary containing information about elements in the viewport.
chains.openai_functions.citation_fuzzy_match.FactWithEvidence
Class representing a single statement.
chains.openai_functions.citation_fuzzy_match.QuestionAnswer
A question and its answer as a list of facts each one should have a source.
chains.openai_functions.openapi.SimpleRequestChain
Chain for making a simple request to an API endpoint.
chains.openai_functions.qa_with_structure.AnswerWithSources
An answer to the question, with sources.
chains.prompt_selector.BasePromptSelector
Base class for prompt selectors.
chains.prompt_selector.ConditionalPromptSelector
Prompt collection that goes through conditionals.
chains.qa_generation.base.QAGenerationChain
Base class for question-answer generation chains.
chains.qa_with_sources.base.BaseQAWithSourcesChain
Question answering chain with sources over documents.
chains.qa_with_sources.base.QAWithSourcesChain
Question answering with sources over documents.
chains.qa_with_sources.loading.LoadingCallable(...)
Interface for loading the combine documents chain.
chains.qa_with_sources.retrieval.RetrievalQAWithSourcesChain
Question-answering with sources over an index.
chains.qa_with_sources.vector_db.VectorDBQAWithSourcesChain
Question-answering with sources over a vector database.
chains.query_constructor.base.StructuredQueryOutputParser
Output parser that parses a structured query.
chains.query_constructor.ir.Comparator(value)
Enumerator of the comparison operators.
chains.query_constructor.ir.Comparison
A comparison to a value.
chains.query_constructor.ir.Expr
Base class for all expressions.
chains.query_constructor.ir.FilterDirective
A filtering expression.
chains.query_constructor.ir.Operation
A logical operation over other directives.
chains.query_constructor.ir.Operator(value)
Enumerator of the operations.
chains.query_constructor.ir.StructuredQuery
A structured query.
chains.query_constructor.ir.Visitor()
Defines interface for IR translation using visitor pattern.
chains.query_constructor.parser.ISO8601Date
A date in ISO 8601 format (YYYY-MM-DD).
chains.query_constructor.schema.AttributeInfo
Information about a data source attribute.
chains.retrieval_qa.base.BaseRetrievalQA
Base class for question-answering chains.
chains.retrieval_qa.base.RetrievalQA
Chain for question-answering against an index.
chains.retrieval_qa.base.VectorDBQA
Chain for question-answering against a vector database.
chains.router.base.MultiRouteChain
Use a single chain to route an input to one of multiple candidate chains.
chains.router.base.Route(destination, ...)
Create new instance of Route(destination, next_inputs)
chains.router.base.RouterChain
Chain that outputs the name of a destination chain and the inputs to it.
chains.router.embedding_router.EmbeddingRouterChain
Chain that uses embeddings to route between options.
chains.router.llm_router.LLMRouterChain
A router chain that uses an LLM chain to perform routing.
chains.router.llm_router.RouterOutputParser
Parser for output of router chain in the multi-prompt chain.
chains.router.multi_prompt.MultiPromptChain
A multi-route chain that uses an LLM router chain to choose amongst prompts.
chains.router.multi_retrieval_qa.MultiRetrievalQAChain
A multi-route chain that uses an LLM router chain to choose amongst retrieval qa chains.
chains.sequential.SequentialChain
Chain where the outputs of one chain feed directly into next.
chains.sequential.SimpleSequentialChain
Simple chain where the outputs of one step feed directly into next.
chains.sql_database.query.SQLInput
Input for a SQL Chain.
chains.sql_database.query.SQLInputWithTables
Input for a SQL Chain.
chains.transform.TransformChain
Chain that transforms the chain output.
Functions¶langchain.embeddings¶
Embedding models are wrappers around embedding models from different APIs and services.
Embedding models can be LLMs or not.
Class hierarchy:
Embeddings --> <name>Embeddings # Examples: OpenAIEmbeddings, HuggingFaceEmbeddingsClasses¶ Functions¶
langchain.evaluation¶
Evaluation chains for grading LLM and Chain outputs.
This module contains off-the-shelf evaluation chains for grading the output of LangChain primitives such as language models and chains.
Loading an evaluator
To load an evaluator, you can use the load_evaluators or load_evaluator functions with the names of the evaluators to load.
from langchain.evaluation import load_evaluator
evaluator = load_evaluator("qa")
evaluator.evaluate_strings(
prediction="We sold more than 40,000 units last week",
input="How many units did we sell last week?",
reference="We sold 32,378 units",
)
The evaluator must be one of EvaluatorType.
Datasets
To load one of the LangChain HuggingFace datasets, you can use the load_dataset function with the name of the dataset to load.
from langchain.evaluation import load_dataset
ds = load_dataset("llm-math")
Some common use cases for evaluation include:
Grading the accuracy of a response against ground truth answers: QAEvalChain
Comparing the output of two models: PairwiseStringEvalChain or LabeledPairwiseStringEvalChain when there is additionally a reference label.
Judging the efficacy of an agent’s tool usage: TrajectoryEvalChain
Checking whether an output complies with a set of criteria: CriteriaEvalChain or LabeledCriteriaEvalChain when there is additionally a reference label.
Computing semantic difference between a prediction and reference: EmbeddingDistanceEvalChain or between two predictions: PairwiseEmbeddingDistanceEvalChain
Measuring the string distance between a prediction and reference StringDistanceEvalChain or between two predictions PairwiseStringDistanceEvalChain
Low-level API
These evaluators implement one of the following interfaces:
StringEvaluator: Evaluate a prediction string against a reference label and/or input context.
PairwiseStringEvaluator: Evaluate two prediction strings against each other. Useful for scoring preferences, measuring similarity between two chain or llm agents, or comparing outputs on similar inputs.
AgentTrajectoryEvaluator Evaluate the full sequence of actions taken by an agent.
These interfaces enable easier composability and usage within a higher level evaluation framework.
Classes¶ Functions¶langchain.hub¶
Interface with the LangChain Hub.
Functions¶hub.pull(owner_repo_commit, *[, api_url, ...])
Pulls an object from the hub and returns it as a LangChain object.
hub.push(repo_full_name, object, *[, ...])
Pushes an object to the hub and returns the URL it can be viewed at in a browser.
langchain.indexes¶
Code to support various indexing workflows.
Provides code to:
Create knowledge graphs from data.
Support indexing workflows from LangChain data loaders to vectorstores.
For indexing workflows, this code is used to avoid writing duplicated content into the vectostore and to avoid over-writing content if it’s unchanged.
Importantly, this keeps on working even if the content being written is derived via a set of transformations from some source content (e.g., indexing children documents that were derived from parent documents by chunking.)
Classes¶ Functions¶langchain.memory¶
Memory maintains Chain state, incorporating context from past runs.
Class hierarchy for Memory:
BaseMemory --> BaseChatMemory --> <name>Memory # Examples: ZepMemory, MotorheadMemory
Main helpers:
Chat Message History stores the chat message history in different stores.
Class hierarchy for ChatMessageHistory:
BaseChatMessageHistory --> <name>ChatMessageHistory # Example: ZepChatMessageHistory
Main helpers:
AIMessage, BaseMessage, HumanMessageClasses¶ Functions¶
langchain.model_laboratory¶
Experiment with different models.
Classes¶langchain.output_parsers¶
OutputParser classes parse the output of an LLM call.
Class hierarchy:
BaseLLMOutputParser --> BaseOutputParser --> <name>OutputParser # ListOutputParser, PydanticOutputParser
Main helpers:
Serializable, Generation, PromptValueClasses¶ Functions¶
langchain.prompts¶
Prompt is the input to the model.
Prompt is often constructed from multiple components. Prompt classes and functions make constructing
and working with prompts easy.
Class hierarchy:
BasePromptTemplate --> PipelinePromptTemplate
StringPromptTemplate --> PromptTemplate
FewShotPromptTemplate
FewShotPromptWithTemplates
BaseChatPromptTemplate --> AutoGPTPrompt
ChatPromptTemplate --> AgentScratchPadChatPromptTemplate
BaseMessagePromptTemplate --> MessagesPlaceholder
BaseStringMessagePromptTemplate --> ChatMessagePromptTemplate
HumanMessagePromptTemplate
AIMessagePromptTemplate
SystemMessagePromptTemplate
PromptValue --> StringPromptValue
ChatPromptValue
Classes¶ Functions¶ langchain.retrievers¶
Retriever class returns Documents given a text 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) it. Vector stores can be used as the backbone of a retriever, but there are other types of retrievers as well.
Class hierarchy:
BaseRetriever --> <name>Retriever # Examples: ArxivRetriever, MergerRetriever
Main helpers:
Document, Serializable, Callbacks, CallbackManagerForRetrieverRun, AsyncCallbackManagerForRetrieverRunClasses¶ Functions¶
langchain.runnables¶ Classes¶ langchain.smith¶
LangSmith utilities.
This module provides utilities for connecting to LangSmith. For more information on LangSmith, see the LangSmith documentation.
Evaluation
LangSmith helps you evaluate Chains and other language model application components using a number of LangChain evaluators. An example of this is shown below, assuming you’ve created a LangSmith dataset called <my_dataset_name>:
from langsmith import Client
from langchain_community.chat_models import ChatOpenAI
from langchain.chains import LLMChain
from langchain.smith import RunEvalConfig, run_on_dataset
# Chains may have memory. Passing in a constructor function lets the
# evaluation framework avoid cross-contamination between runs.
def construct_chain():
llm = ChatOpenAI(temperature=0)
chain = LLMChain.from_string(
llm,
"What's the answer to {your_input_key}"
)
return chain
# Load off-the-shelf evaluators via config or the EvaluatorType (string or enum)
evaluation_config = RunEvalConfig(
evaluators=[
"qa", # "Correctness" against a reference answer
"embedding_distance",
RunEvalConfig.Criteria("helpfulness"),
RunEvalConfig.Criteria({
"fifth-grader-score": "Do you have to be smarter than a fifth grader to answer this question?"
}),
]
)
client = Client()
run_on_dataset(
client,
"<my_dataset_name>",
construct_chain,
evaluation=evaluation_config,
)
You can also create custom evaluators by subclassing the StringEvaluator or LangSmith’s RunEvaluator classes.
from typing import Optional
from langchain.evaluation import StringEvaluator
class MyStringEvaluator(StringEvaluator):
@property
def requires_input(self) -> bool:
return False
@property
def requires_reference(self) -> bool:
return True
@property
def evaluation_name(self) -> str:
return "exact_match"
def _evaluate_strings(self, prediction, reference=None, input=None, **kwargs) -> dict:
return {"score": prediction == reference}
evaluation_config = RunEvalConfig(
custom_evaluators = [MyStringEvaluator()],
)
run_on_dataset(
client,
"<my_dataset_name>",
construct_chain,
evaluation=evaluation_config,
)
Primary Functions
arun_on_dataset: Asynchronous function to evaluate a chain, agent, or other LangChain component over a dataset.
run_on_dataset: Function to evaluate a chain, agent, or other LangChain component over a dataset.
RunEvalConfig: Class representing the configuration for running evaluation. You can select evaluators by EvaluatorType or config, or you can pass in custom_evaluators
langchain.storage¶
Implementations of key-value stores and storage helpers.
Module provides implementations of various key-value stores that conform to a simple key-value interface.
The primary goal of these storages is to support implementation of caching.
Classes¶langchain.text_splitter¶
Text Splitters are classes for splitting text.
Class hierarchy:
BaseDocumentTransformer --> TextSplitter --> <name>TextSplitter # Example: CharacterTextSplitter
RecursiveCharacterTextSplitter --> <name>TextSplitter
Note: MarkdownHeaderTextSplitter and **HTMLHeaderTextSplitter do not derive from TextSplitter.
Main helpers:
Document, Tokenizer, Language, LineType, HeaderTypeClasses¶ Functions¶
langchain.tools¶
Tools are classes that an Agent uses to interact with the world.
Each tool has a description. Agent uses the description to choose the right tool for the job.
Class hierarchy:
ToolMetaclass --> BaseTool --> <name>Tool # Examples: AIPluginTool, BaseGraphQLTool
<name> # Examples: BraveSearch, HumanInputRun
Main helpers:
CallbackManagerForToolRun, AsyncCallbackManagerForToolRunClasses¶ Functions¶
langchain.utils¶
Utility functions for LangChain.
These functions do not depend on any other LangChain module.
Functions¶RetroSearch is an open source project built by @garambo | Open a GitHub Issue
Search and Browse the WWW like it's 1997 | Search results from DuckDuckGo
HTML:
3.2
| Encoding:
UTF-8
| Version:
0.7.4