CriteriaEvalChain#

Note

CriteriaEvalChain implements the standard Runnable Interface. 🏃

The Runnable Interface has additional methods that are available on runnables, such as with_types, with_retry, assign, bind, get_graph, and more.

class langchain.evaluation.criteria.eval_chain.CriteriaEvalChain[source]#

Bases: StringEvaluator, LLMEvalChain, LLMChain

LLM Chain for evaluating runs against criteria.

Parameters:

llm (BaseLanguageModel) – The language model to use for evaluation.
criteria (Union[Mapping[str, str]]) – The criteria or rubric to evaluate the runs against. It can be a mapping of criterion name to its description, or a single criterion name.
prompt (Optional[BasePromptTemplate], default=None) – The prompt template to use for generating prompts. If not provided, a default prompt template will be used based on the value of requires_reference.
requires_reference (bool, default=False) – Whether the evaluation requires a reference text. If True, the PROMPT_WITH_REFERENCES template will be used, which includes the reference labels in the prompt. Otherwise, the PROMPT template will be used, which is a reference-free prompt.
**kwargs (Any) – Additional keyword arguments to pass to the LLMChain constructor.

Returns:

An instance of the CriteriaEvalChain class.

Return type:

CriteriaEvalChain

Examples

>>> from langchain_anthropic import ChatAnthropic
>>> from langchain.evaluation.criteria import CriteriaEvalChain
>>> llm = ChatAnthropic(temperature=0)
>>> criteria = {"my-custom-criterion": "Is the submission the most amazing ever?"}
>>> evaluator = CriteriaEvalChain.from_llm(llm=llm, criteria=criteria)
>>> evaluator.evaluate_strings(prediction="Imagine an ice cream flavor for the color aquamarine", input="Tell me an idea")
{
    'reasoning': 'Here is my step-by-step reasoning for the given criteria:\n\nThe criterion is: "Is the submission the most amazing ever?" This is a subjective criterion and open to interpretation. The submission suggests an aquamarine-colored ice cream flavor which is creative but may or may not be considered the most amazing idea ever conceived. There are many possible amazing ideas and this one ice cream flavor suggestion may or may not rise to that level for every person. \n\nN',
    'value': 'N',
    'score': 0,
}

>>> from langchain_openai import ChatOpenAI
>>> from langchain.evaluation.criteria import LabeledCriteriaEvalChain
>>> llm = ChatOpenAI(model="gpt-4", temperature=0)
>>> criteria = "correctness"
>>> evaluator = LabeledCriteriaEvalChain.from_llm(
...     llm=llm,
...     criteria=criteria,
... )
>>> evaluator.evaluate_strings(
...   prediction="The answer is 4",
...   input="How many apples are there?",
...   reference="There are 3 apples",
...   )
{
    'score': 0,
    'reasoning': 'The criterion for this task is the correctness of the submission. The submission states that there are 4 apples, but the reference indicates that there are actually 3 apples. Therefore, the submission is not correct, accurate, or factual according to the given criterion.\n\nN',
    'value': 'N',
}

param callback_manager: BaseCallbackManager | None = None#: [DEPRECATED] Use callbacks instead.

param callbacks: Callbacks = None#: Optional list of callback handlers (or callback manager). Defaults to None. Callback handlers are called throughout the lifecycle of a call to a chain, starting with on_chain_start, ending with on_chain_end or on_chain_error. Each custom chain can optionally call additional callback methods, see Callback docs for full details.

param criterion_name: str [Required]#: The name of the criterion being evaluated.

param llm: Runnable[LanguageModelInput, str] | Runnable[LanguageModelInput, BaseMessage] [Required]#: Language model to call.

param llm_kwargs: dict [Optional]#

param memory: BaseMemory | None = None#: Optional memory object. Defaults to None. Memory is a class that gets called at the start and at the end of every chain. At the start, memory loads variables and passes them along in the chain. At the end, it saves any returned variables. There are many different types of memory - please see memory docs for the full catalog.

param metadata: Dict[str, Any] | None = None#: Optional metadata associated with the chain. Defaults to None. This metadata will be associated with each call to this chain, and passed as arguments to the handlers defined in callbacks. You can use these to eg identify a specific instance of a chain with its use case.

param output_parser: BaseOutputParser [Optional]#: The parser to use to map the output to a structured result.

param prompt: BasePromptTemplate [Required]#: Prompt object to use.

param return_final_only: bool = True#: Whether to return only the final parsed result. Defaults to True. If false, will return a bunch of extra information about the generation.

param tags: List[str] | None = None#: Optional list of tags associated with the chain. Defaults to None. These tags will be associated with each call to this chain, and passed as arguments to the handlers defined in callbacks. You can use these to eg identify a specific instance of a chain with its use case.

param verbose: bool [Optional]#: Whether or not run in verbose mode. In verbose mode, some intermediate logs will be printed to the console. Defaults to the global verbose value, accessible via langchain.globals.get_verbose().

__call__(inputs: Dict[str, Any] | Any, return_only_outputs: bool = False, callbacks: List[BaseCallbackHandler] | BaseCallbackManager | None = None, *, tags: List[str] | None = None, metadata: Dict[str, Any] | None = None, run_name: str | None = None, include_run_info: bool = False) → Dict[str, Any]#

Deprecated since version langchain==0.1.0: Use invoke instead.

Execute the chain.

Parameters:

inputs (Dict[str, Any] | Any) – Dictionary of inputs, or single input if chain expects only one param. Should contain all inputs specified in Chain.input_keys except for inputs that will be set by the chain’s memory.
return_only_outputs (bool) – Whether to return only outputs in the response. If True, only new keys generated by this chain will be returned. If False, both input keys and new keys generated by this chain will be returned. Defaults to False.
callbacks (List[BaseCallbackHandler] | BaseCallbackManager | None) – Callbacks to use for this chain run. These will be called in addition to callbacks passed to the chain during construction, but only these runtime callbacks will propagate to calls to other objects.
tags (List[str] | None) – List of string tags to pass to all callbacks. These will be passed in addition to tags passed to the chain during construction, but only these runtime tags will propagate to calls to other objects.
metadata (Dict[str, Any] | None) – Optional metadata associated with the chain. Defaults to None
include_run_info (bool) – Whether to include run info in the response. Defaults to False.
run_name (str | None)

Returns:

A dict of named outputs. Should contain all outputs specified in: Chain.output_keys.

Return type:

Dict[str, Any]

async aapply(input_list: List[Dict[str, Any]], callbacks: List[BaseCallbackHandler] | BaseCallbackManager | None = None) → List[Dict[str, str]]#

Utilize the LLM generate method for speed gains.

Parameters:

input_list (List[Dict[str, Any]])
callbacks (List[BaseCallbackHandler] | BaseCallbackManager | None)

Return type:

List[Dict[str, str]]

async aapply_and_parse(input_list: List[Dict[str, Any]], callbacks: List[BaseCallbackHandler] | BaseCallbackManager | None = None) → Sequence[str | List[str] | Dict[str, str]]#

Call apply and then parse the results.

Parameters:

input_list (List[Dict[str, Any]])
callbacks (List[BaseCallbackHandler] | BaseCallbackManager | None)

Return type:

Sequence[str | List[str] | Dict[str, str]]

async abatch(inputs: List[Input], config: RunnableConfig | List[RunnableConfig] | None = None, *, return_exceptions: bool = False, **kwargs: Any | None) → List[Output]#

Default implementation runs ainvoke in parallel using asyncio.gather.

The default implementation of batch works well for IO bound runnables.

Subclasses should override this method if they can batch more efficiently; e.g., if the underlying Runnable uses an API which supports a batch mode.

Parameters:

inputs (List[Input]) – A list of inputs to the Runnable.
config (RunnableConfig | List[RunnableConfig] | None) – A config to use when invoking the Runnable. The config supports standard keys like ‘tags’, ‘metadata’ for tracing purposes, ‘max_concurrency’ for controlling how much work to do in parallel, and other keys. Please refer to the RunnableConfig for more details. Defaults to None.
return_exceptions (bool) – Whether to return exceptions instead of raising them. Defaults to False.
kwargs (Any | None) – Additional keyword arguments to pass to the Runnable.

Returns:

A list of outputs from the Runnable.

Return type:

List[Output]

async abatch_as_completed(inputs: Sequence[Input], config: RunnableConfig | Sequence[RunnableConfig] | None = None, *, return_exceptions: bool = False, **kwargs: Any | None) → AsyncIterator[Tuple[int, Output | Exception]]#

Run ainvoke in parallel on a list of inputs, yielding results as they complete.

Parameters:

inputs (Sequence[Input]) – A list of inputs to the Runnable.
config (RunnableConfig | Sequence[RunnableConfig] | None) – A config to use when invoking the Runnable. The config supports standard keys like ‘tags’, ‘metadata’ for tracing purposes, ‘max_concurrency’ for controlling how much work to do in parallel, and other keys. Please refer to the RunnableConfig for more details. Defaults to None. Defaults to None.
return_exceptions (bool) – Whether to return exceptions instead of raising them. Defaults to False.
kwargs (Any | None) – Additional keyword arguments to pass to the Runnable.

Yields:

A tuple of the index of the input and the output from the Runnable.

Return type:

AsyncIterator[Tuple[int, Output | Exception]]

async acall(inputs: Dict[str, Any] | Any, return_only_outputs: bool = False, callbacks: List[BaseCallbackHandler] | BaseCallbackManager | None = None, *, tags: List[str] | None = None, metadata: Dict[str, Any] | None = None, run_name: str | None = None, include_run_info: bool = False) → Dict[str, Any]#

Deprecated since version langchain==0.1.0: Use ainvoke instead.

Asynchronously execute the chain.

Parameters:

inputs (Dict[str, Any] | Any) – Dictionary of inputs, or single input if chain expects only one param. Should contain all inputs specified in Chain.input_keys except for inputs that will be set by the chain’s memory.
return_only_outputs (bool) – Whether to return only outputs in the response. If True, only new keys generated by this chain will be returned. If False, both input keys and new keys generated by this chain will be returned. Defaults to False.
callbacks (List[BaseCallbackHandler] | BaseCallbackManager | None) – Callbacks to use for this chain run. These will be called in addition to callbacks passed to the chain during construction, but only these runtime callbacks will propagate to calls to other objects.
tags (List[str] | None) – List of string tags to pass to all callbacks. These will be passed in addition to tags passed to the chain during construction, but only these runtime tags will propagate to calls to other objects.
metadata (Dict[str, Any] | None) – Optional metadata associated with the chain. Defaults to None
include_run_info (bool) – Whether to include run info in the response. Defaults to False.
run_name (str | None)

Returns:

A dict of named outputs. Should contain all outputs specified in: Chain.output_keys.

Return type:

Dict[str, Any]

async aevaluate_strings(*, prediction: str, reference: str | None = None, input: str | None = None, **kwargs: Any) → dict#

Asynchronously evaluate Chain or LLM output, based on optional input and label.

Parameters:

prediction (str) – The LLM or chain prediction to evaluate.
reference (Optional[str], optional) – The reference label to evaluate against.
input (Optional[str], optional) – The input to consider during evaluation.
kwargs (Any) – Additional keyword arguments, including callbacks, tags, etc.

Returns:

The evaluation results containing the score or value.

Return type:

dict

async agenerate(input_list: List[Dict[str, Any]], run_manager: AsyncCallbackManagerForChainRun | None = None) → LLMResult#

Generate LLM result from inputs.

Parameters:

input_list (List[Dict[str, Any]])
run_manager (AsyncCallbackManagerForChainRun | None)

Return type:

LLMResult

async ainvoke(input: Dict[str, Any], config: RunnableConfig | None = None, **kwargs: Any) → Dict[str, Any]#

Default implementation of ainvoke, calls invoke from a thread.

The default implementation allows usage of async code even if the Runnable did not implement a native async version of invoke.

Subclasses should override this method if they can run asynchronously.

Parameters:

input (Dict[str, Any])
config (RunnableConfig | None)
kwargs (Any)

Return type:

Dict[str, Any]

apply(input_list: List[Dict[str, Any]], callbacks: List[BaseCallbackHandler] | BaseCallbackManager | None = None) → List[Dict[str, str]]#

Utilize the LLM generate method for speed gains.

Parameters:

input_list (List[Dict[str, Any]])
callbacks (List[BaseCallbackHandler] | BaseCallbackManager | None)

Return type:

List[Dict[str, str]]

apply_and_parse(input_list: List[Dict[str, Any]], callbacks: List[BaseCallbackHandler] | BaseCallbackManager | None = None) → Sequence[str | List[str] | Dict[str, str]]#

Call apply and then parse the results.

Parameters:

input_list (List[Dict[str, Any]])
callbacks (List[BaseCallbackHandler] | BaseCallbackManager | None)

Return type:

Sequence[str | List[str] | Dict[str, str]]

async apredict(callbacks: List[BaseCallbackHandler] | BaseCallbackManager | None = None, **kwargs: Any) → str#

Format prompt with kwargs and pass to LLM.

Parameters:

callbacks (List[BaseCallbackHandler] | BaseCallbackManager | None) – Callbacks to pass to LLMChain
**kwargs (Any) – Keys to pass to prompt template.

Returns:

Completion from LLM.

Return type:

str

Example

completion = llm.predict(adjective="funny")

async apredict_and_parse(callbacks: List[BaseCallbackHandler] | BaseCallbackManager | None = None, **kwargs: Any) → str | List[str] | Dict[str, str]#

Call apredict and then parse the results.

Parameters:

callbacks (List[BaseCallbackHandler] | BaseCallbackManager | None)
kwargs (Any)

Return type:

str | List[str] | Dict[str, str]

async aprep_inputs(inputs: Dict[str, Any] | Any) → Dict[str, str]#

Prepare chain inputs, including adding inputs from memory.

Parameters:: inputs (Dict[str, Any] | Any) – Dictionary of raw inputs, or single input if chain expects only one param. Should contain all inputs specified in Chain.input_keys except for inputs that will be set by the chain’s memory.
Returns:: A dictionary of all inputs, including those added by the chain’s memory.
Return type:: Dict[str, str]

async aprep_outputs(inputs: Dict[str, str], outputs: Dict[str, str], return_only_outputs: bool = False) → Dict[str, str]#

Validate and prepare chain outputs, and save info about this run to memory.

Parameters:

inputs (Dict[str, str]) – Dictionary of chain inputs, including any inputs added by chain memory.
outputs (Dict[str, str]) – Dictionary of initial chain outputs.
return_only_outputs (bool) – Whether to only return the chain outputs. If False, inputs are also added to the final outputs.

Returns:

A dict of the final chain outputs.

Return type:

Dict[str, str]

async aprep_prompts(input_list: List[Dict[str, Any]], run_manager: AsyncCallbackManagerForChainRun | None = None) → Tuple[List[PromptValue], List[str] | None]#

Prepare prompts from inputs.

Parameters:

input_list (List[Dict[str, Any]])
run_manager (AsyncCallbackManagerForChainRun | None)

Return type:

Tuple[List[PromptValue], List[str] | None]

async arun(*args: Any, callbacks: List[BaseCallbackHandler] | BaseCallbackManager | None = None, tags: List[str] | None = None, metadata: Dict[str, Any] | None = None, **kwargs: Any) → Any#

Deprecated since version langchain==0.1.0: Use ainvoke instead.

Convenience method for executing chain.

The main difference between this method and Chain.__call__ is that this method expects inputs to be passed directly in as positional arguments or keyword arguments, whereas Chain.__call__ expects a single input dictionary with all the inputs

Parameters:

*args (Any) – If the chain expects a single input, it can be passed in as the sole positional argument.
callbacks (List[BaseCallbackHandler] | BaseCallbackManager | None) – Callbacks to use for this chain run. These will be called in addition to callbacks passed to the chain during construction, but only these runtime callbacks will propagate to calls to other objects.
tags (List[str] | None) – List of string tags to pass to all callbacks. These will be passed in addition to tags passed to the chain during construction, but only these runtime tags will propagate to calls to other objects.
**kwargs (Any) – If the chain expects multiple inputs, they can be passed in directly as keyword arguments.
metadata (Dict[str, Any] | None)
**kwargs

Returns:

The chain output.

Return type:

Any

Example

# Suppose we have a single-input chain that takes a 'question' string:
await chain.arun("What's the temperature in Boise, Idaho?")
# -> "The temperature in Boise is..."

# Suppose we have a multi-input chain that takes a 'question' string
# and 'context' string:
question = "What's the temperature in Boise, Idaho?"
context = "Weather report for Boise, Idaho on 07/03/23..."
await chain.arun(question=question, context=context)
# -> "The temperature in Boise is..."

as_tool(args_schema: Type[BaseModel] | None = None, *, name: str | None = None, description: str | None = None, arg_types: Dict[str, Type] | None = None) → BaseTool#

Beta

This API is in beta and may change in the future.

Create a BaseTool from a Runnable.

as_tool will instantiate a BaseTool with a name, description, and args_schema from a Runnable. Where possible, schemas are inferred from runnable.get_input_schema. Alternatively (e.g., if the Runnable takes a dict as input and the specific dict keys are not typed), the schema can be specified directly with args_schema. You can also pass arg_types to just specify the required arguments and their types.

Parameters:

args_schema (Optional[Type[BaseModel]]) – The schema for the tool. Defaults to None.
name (Optional[str]) – The name of the tool. Defaults to None.
description (Optional[str]) – The description of the tool. Defaults to None.
arg_types (Optional[Dict[str, Type]]) – A dictionary of argument names to types. Defaults to None.

Returns:

A BaseTool instance.

Return type:

BaseTool

Typed dict input:

from typing import List
from typing_extensions import TypedDict
from langchain_core.runnables import RunnableLambda

class Args(TypedDict):
    a: int
    b: List[int]

def f(x: Args) -> str:
    return str(x["a"] * max(x["b"]))

runnable = RunnableLambda(f)
as_tool = runnable.as_tool()
as_tool.invoke({"a": 3, "b": [1, 2]})

dict input, specifying schema via args_schema:

from typing import Any, Dict, List
from langchain_core.pydantic_v1 import BaseModel, Field
from langchain_core.runnables import RunnableLambda

def f(x: Dict[str, Any]) -> str:
    return str(x["a"] * max(x["b"]))

class FSchema(BaseModel):
    """Apply a function to an integer and list of integers."""

    a: int = Field(..., description="Integer")
    b: List[int] = Field(..., description="List of ints")

runnable = RunnableLambda(f)
as_tool = runnable.as_tool(FSchema)
as_tool.invoke({"a": 3, "b": [1, 2]})

dict input, specifying schema via arg_types:

from typing import Any, Dict, List
from langchain_core.runnables import RunnableLambda

def f(x: Dict[str, Any]) -> str:
    return str(x["a"] * max(x["b"]))

runnable = RunnableLambda(f)
as_tool = runnable.as_tool(arg_types={"a": int, "b": List[int]})
as_tool.invoke({"a": 3, "b": [1, 2]})

String input:

from langchain_core.runnables import RunnableLambda

def f(x: str) -> str:
    return x + "a"

def g(x: str) -> str:
    return x + "z"

runnable = RunnableLambda(f) | g
as_tool = runnable.as_tool()
as_tool.invoke("b")

Added in version 0.2.14.

async astream(input: Input, config: RunnableConfig | None = None, **kwargs: Any | None) → AsyncIterator[Output]#

Default implementation of astream, which calls ainvoke. Subclasses should override this method if they support streaming output.

Parameters:

input (Input) – The input to the Runnable.
config (RunnableConfig | None) – The config to use for the Runnable. Defaults to None.
kwargs (Any | None) – Additional keyword arguments to pass to the Runnable.

Yields:

The output of the Runnable.

Return type:

AsyncIterator[Output]

astream_events(input: Any, config: RunnableConfig | None = None, *, version: Literal['v1', 'v2'], include_names: Sequence[str] | None = None, include_types: Sequence[str] | None = None, include_tags: Sequence[str] | None = None, exclude_names: Sequence[str] | None = None, exclude_types: Sequence[str] | None = None, exclude_tags: Sequence[str] | None = None, **kwargs: Any) → AsyncIterator[StandardStreamEvent | CustomStreamEvent]#

Beta

This API is in beta and may change in the future.

Generate a stream of events.

Use to create an iterator over StreamEvents that provide real-time information about the progress of the Runnable, including StreamEvents from intermediate results.

A StreamEvent is a dictionary with the following schema:

event: str - Event names are of the
format: on_[runnable_type]_(start|stream|end).
name: str - The name of the Runnable that generated the event.
run_id: str - randomly generated ID associated with the given execution of
the Runnable that emitted the event. A child Runnable that gets invoked as part of the execution of a parent Runnable is assigned its own unique ID.
parent_ids: List[str] - The IDs of the parent runnables that
generated the event. The root Runnable will have an empty list. The order of the parent IDs is from the root to the immediate parent. Only available for v2 version of the API. The v1 version of the API will return an empty list.
tags: Optional[List[str]] - The tags of the Runnable that generated
the event.
metadata: Optional[Dict[str, Any]] - The metadata of the Runnable
that generated the event.
data: Dict[str, Any]

Below is a table that illustrates some evens that might be emitted by various chains. Metadata fields have been omitted from the table for brevity. Chain definitions have been included after the table.

ATTENTION This reference table is for the V2 version of the schema.

event	name	chunk	input	output
on_chat_model_start	[model name]		{“messages”: [[SystemMessage, HumanMessage]]}
on_chat_model_stream	[model name]	AIMessageChunk(content=”hello”)
on_chat_model_end	[model name]		{“messages”: [[SystemMessage, HumanMessage]]}	AIMessageChunk(content=”hello world”)
on_llm_start	[model name]		{‘input’: ‘hello’}
on_llm_stream	[model name]	‘Hello’
on_llm_end	[model name]		‘Hello human!’
on_chain_start	format_docs
on_chain_stream	format_docs	“hello world!, goodbye world!”
on_chain_end	format_docs		[Document(…)]	“hello world!, goodbye world!”
on_tool_start	some_tool		{“x”: 1, “y”: “2”}
on_tool_end	some_tool			{“x”: 1, “y”: “2”}
on_retriever_start	[retriever name]		{“query”: “hello”}
on_retriever_end	[retriever name]		{“query”: “hello”}	[Document(…), ..]
on_prompt_start	[template_name]		{“question”: “hello”}
on_prompt_end	[template_name]		{“question”: “hello”}	ChatPromptValue(messages: [SystemMessage, …])

In addition to the standard events, users can also dispatch custom events (see example below).

Custom events will be only be surfaced with in the v2 version of the API!

A custom event has following format:

Attribute	Type	Description
name	str	A user defined name for the event.
data	Any	The data associated with the event. This can be anything, though we suggest making it JSON serializable.

Here are declarations associated with the standard events shown above:

format_docs:

def format_docs(docs: List[Document]) -> str:
    '''Format the docs.'''
    return ", ".join([doc.page_content for doc in docs])

format_docs = RunnableLambda(format_docs)

some_tool:

@tool
def some_tool(x: int, y: str) -> dict:
    '''Some_tool.'''
    return {"x": x, "y": y}

prompt:

template = ChatPromptTemplate.from_messages(
    [("system", "You are Cat Agent 007"), ("human", "{question}")]
).with_config({"run_name": "my_template", "tags": ["my_template"]})

Example:

from langchain_core.runnables import RunnableLambda

async def reverse(s: str) -> str:
    return s[::-1]

chain = RunnableLambda(func=reverse)

events = [
    event async for event in chain.astream_events("hello", version="v2")
]

# will produce the following events (run_id, and parent_ids
# has been omitted for brevity):
[
    {
        "data": {"input": "hello"},
        "event": "on_chain_start",
        "metadata": {},
        "name": "reverse",
        "tags": [],
    },
    {
        "data": {"chunk": "olleh"},
        "event": "on_chain_stream",
        "metadata": {},
        "name": "reverse",
        "tags": [],
    },
    {
        "data": {"output": "olleh"},
        "event": "on_chain_end",
        "metadata": {},
        "name": "reverse",
        "tags": [],
    },
]

Example: Dispatch Custom Event

from langchain_core.callbacks.manager import (
    adispatch_custom_event,
)
from langchain_core.runnables import RunnableLambda, RunnableConfig
import asyncio


async def slow_thing(some_input: str, config: RunnableConfig) -> str:
    """Do something that takes a long time."""
    await asyncio.sleep(1) # Placeholder for some slow operation
    await adispatch_custom_event(
        "progress_event",
        {"message": "Finished step 1 of 3"},
        config=config # Must be included for python < 3.10
    )
    await asyncio.sleep(1) # Placeholder for some slow operation
    await adispatch_custom_event(
        "progress_event",
        {"message": "Finished step 2 of 3"},
        config=config # Must be included for python < 3.10
    )
    await asyncio.sleep(1) # Placeholder for some slow operation
    return "Done"

slow_thing = RunnableLambda(slow_thing)

async for event in slow_thing.astream_events("some_input", version="v2"):
    print(event)

Parameters:

input (Any) – The input to the Runnable.
config (RunnableConfig | None) – The config to use for the Runnable.
version (Literal['v1', 'v2']) – The version of the schema to use either v2 or v1. Users should use v2. v1 is for backwards compatibility and will be deprecated in 0.4.0. No default will be assigned until the API is stabilized. custom events will only be surfaced in v2.
include_names (Sequence[str] | None) – Only include events from runnables with matching names.
include_types (Sequence[str] | None) – Only include events from runnables with matching types.
include_tags (Sequence[str] | None) – Only include events from runnables with matching tags.
exclude_names (Sequence[str] | None) – Exclude events from runnables with matching names.
exclude_types (Sequence[str] | None) – Exclude events from runnables with matching types.
exclude_tags (Sequence[str] | None) – Exclude events from runnables with matching tags.
kwargs (Any) – Additional keyword arguments to pass to the Runnable. These will be passed to astream_log as this implementation of astream_events is built on top of astream_log.

Yields:

An async stream of StreamEvents.

Raises:

NotImplementedError – If the version is not v1 or v2.

Return type:

AsyncIterator[StandardStreamEvent | CustomStreamEvent]

batch(inputs: List[Input], config: RunnableConfig | List[RunnableConfig] | None = None, *, return_exceptions: bool = False, **kwargs: Any | None) → List[Output]#

Default implementation runs invoke in parallel using a thread pool executor.

The default implementation of batch works well for IO bound runnables.

Subclasses should override this method if they can batch more efficiently; e.g., if the underlying Runnable uses an API which supports a batch mode.

Parameters:

inputs (List[Input])
config (RunnableConfig | List[RunnableConfig] | None)
return_exceptions (bool)
kwargs (Any | None)

Return type:

List[Output]

batch_as_completed(inputs: Sequence[Input], config: RunnableConfig | Sequence[RunnableConfig] | None = None, *, return_exceptions: bool = False, **kwargs: Any | None) → Iterator[Tuple[int, Output | Exception]]#

Run invoke in parallel on a list of inputs, yielding results as they complete.

Parameters:

inputs (Sequence[Input])
config (RunnableConfig | Sequence[RunnableConfig] | None)
return_exceptions (bool)
kwargs (Any | None)

Return type:

Iterator[Tuple[int, Output | Exception]]

configurable_alternatives(which: ConfigurableField, *, default_key: str = 'default', prefix_keys: bool = False, **kwargs: Runnable[Input, Output] | Callable[[], Runnable[Input, Output]]) → RunnableSerializable[Input, Output]#

Configure alternatives for Runnables that can be set at runtime.

Parameters:

which (ConfigurableField) – The ConfigurableField instance that will be used to select the alternative.
default_key (str) – The default key to use if no alternative is selected. Defaults to “default”.
prefix_keys (bool) – Whether to prefix the keys with the ConfigurableField id. Defaults to False.
**kwargs (Runnable[Input, Output] | Callable[[], Runnable[Input, Output]]) – A dictionary of keys to Runnable instances or callables that return Runnable instances.

Returns:

A new Runnable with the alternatives configured.

Return type:

RunnableSerializable[Input, Output]

from langchain_anthropic import ChatAnthropic
from langchain_core.runnables.utils import ConfigurableField
from langchain_openai import ChatOpenAI

model = ChatAnthropic(
    model_name="claude-3-sonnet-20240229"
).configurable_alternatives(
    ConfigurableField(id="llm"),
    default_key="anthropic",
    openai=ChatOpenAI()
)

# uses the default model ChatAnthropic
print(model.invoke("which organization created you?").content)

# uses ChatOpenAI
print(
    model.with_config(
        configurable={"llm": "openai"}
    ).invoke("which organization created you?").content
)

configurable_fields(**kwargs: ConfigurableField | ConfigurableFieldSingleOption | ConfigurableFieldMultiOption) → RunnableSerializable[Input, Output]#

Configure particular Runnable fields at runtime.

Parameters:: **kwargs (ConfigurableField | ConfigurableFieldSingleOption | ConfigurableFieldMultiOption) – A dictionary of ConfigurableField instances to configure.
Returns:: A new Runnable with the fields configured.
Return type:: RunnableSerializable[Input, Output]

from langchain_core.runnables import ConfigurableField
from langchain_openai import ChatOpenAI

model = ChatOpenAI(max_tokens=20).configurable_fields(
    max_tokens=ConfigurableField(
        id="output_token_number",
        name="Max tokens in the output",
        description="The maximum number of tokens in the output",
    )
)

# max_tokens = 20
print(
    "max_tokens_20: ",
    model.invoke("tell me something about chess").content
)

# max_tokens = 200
print("max_tokens_200: ", model.with_config(
    configurable={"output_token_number": 200}
    ).invoke("tell me something about chess").content
)

create_outputs(llm_result: LLMResult) → List[Dict[str, Any]]#

Create outputs from response.

Parameters:: llm_result (LLMResult)
Return type:: List[Dict[str, Any]]

evaluate_strings(*, prediction: str, reference: str | None = None, input: str | None = None, **kwargs: Any) → dict#

Evaluate Chain or LLM output, based on optional input and label.

Parameters:

prediction (str) – The LLM or chain prediction to evaluate.
reference (Optional[str], optional) – The reference label to evaluate against.
input (Optional[str], optional) – The input to consider during evaluation.
kwargs (Any) – Additional keyword arguments, including callbacks, tags, etc.

Returns:

The evaluation results containing the score or value.

Return type:

dict

classmethod from_llm(llm: BaseLanguageModel, criteria: Mapping[str, str] | Criteria | ConstitutionalPrinciple | None = None, *, prompt: BasePromptTemplate | None = None, **kwargs: Any) → CriteriaEvalChain[source]#

Create a CriteriaEvalChain instance from an llm and criteria.

Parameters:

llm (BaseLanguageModel) – The language model to use for evaluation.
criteria (CRITERIA_TYPE - default=None for "helpfulness") –
The criteria to evaluate the runs against. It can be:
- a mapping of a criterion name to its description
- a single criterion name present in one of the default criteria
- a single ConstitutionalPrinciple instance
prompt (Optional[BasePromptTemplate], default=None) – The prompt template to use for generating prompts. If not provided, a default prompt template will be used.
**kwargs (Any) – Additional keyword arguments to pass to the LLMChain constructor.

Returns:

An instance of the CriteriaEvalChain class.

Return type:

CriteriaEvalChain

Examples

>>> from langchain_openai import OpenAI
>>> from langchain.evaluation.criteria import LabeledCriteriaEvalChain
>>> llm = OpenAI()
>>> criteria = {
        "hallucination": (
            "Does this submission contain information"
            " not present in the input or reference?"
        ),
    }
>>> chain = LabeledCriteriaEvalChain.from_llm(
        llm=llm,
        criteria=criteria,
    )

classmethod from_string(llm: BaseLanguageModel, template: str) → LLMChain#

Create LLMChain from LLM and template.

Parameters:

llm (BaseLanguageModel)
template (str)

Return type:

LLMChain

generate(input_list: List[Dict[str, Any]], run_manager: CallbackManagerForChainRun | None = None) → LLMResult#

Generate LLM result from inputs.

Parameters:

input_list (List[Dict[str, Any]])
run_manager (CallbackManagerForChainRun | None)

Return type:

LLMResult

invoke(input: Dict[str, Any], config: RunnableConfig | None = None, **kwargs: Any) → Dict[str, Any]#

Transform a single input into an output. Override to implement.

Parameters:

input (Dict[str, Any]) – The input to the Runnable.
config (RunnableConfig | None) – A config to use when invoking the Runnable. The config supports standard keys like ‘tags’, ‘metadata’ for tracing purposes, ‘max_concurrency’ for controlling how much work to do in parallel, and other keys. Please refer to the RunnableConfig for more details.
kwargs (Any)

Returns:

The output of the Runnable.

Return type:

Dict[str, Any]

predict(callbacks: List[BaseCallbackHandler] | BaseCallbackManager | None = None, **kwargs: Any) → str#

Format prompt with kwargs and pass to LLM.

Parameters:

callbacks (List[BaseCallbackHandler] | BaseCallbackManager | None) – Callbacks to pass to LLMChain
**kwargs (Any) – Keys to pass to prompt template.

Returns:

Completion from LLM.

Return type:

str

Example

completion = llm.predict(adjective="funny")

predict_and_parse(callbacks: List[BaseCallbackHandler] | BaseCallbackManager | None = None, **kwargs: Any) → str | List[str] | Dict[str, Any]#

Call predict and then parse the results.

Parameters:

callbacks (List[BaseCallbackHandler] | BaseCallbackManager | None)
kwargs (Any)

Return type:

str | List[str] | Dict[str, Any]

prep_inputs(inputs: Dict[str, Any] | Any) → Dict[str, str]#

Prepare chain inputs, including adding inputs from memory.

Parameters:: inputs (Dict[str, Any] | Any) – Dictionary of raw inputs, or single input if chain expects only one param. Should contain all inputs specified in Chain.input_keys except for inputs that will be set by the chain’s memory.
Returns:: A dictionary of all inputs, including those added by the chain’s memory.
Return type:: Dict[str, str]

prep_outputs(inputs: Dict[str, str], outputs: Dict[str, str], return_only_outputs: bool = False) → Dict[str, str]#

Validate and prepare chain outputs, and save info about this run to memory.

Parameters:

inputs (Dict[str, str]) – Dictionary of chain inputs, including any inputs added by chain memory.
outputs (Dict[str, str]) – Dictionary of initial chain outputs.
return_only_outputs (bool) – Whether to only return the chain outputs. If False, inputs are also added to the final outputs.

Returns:

A dict of the final chain outputs.

Return type:

Dict[str, str]

prep_prompts(input_list: List[Dict[str, Any]], run_manager: CallbackManagerForChainRun | None = None) → Tuple[List[PromptValue], List[str] | None]#

Prepare prompts from inputs.

Parameters:

input_list (List[Dict[str, Any]])
run_manager (CallbackManagerForChainRun | None)

Return type:

Tuple[List[PromptValue], List[str] | None]

classmethod resolve_criteria(criteria: Mapping[str, str] | Criteria | ConstitutionalPrinciple | str | None) → Dict[str, str][source]#

Resolve the criteria to evaluate.

Parameters:

criteria (CRITERIA_TYPE) –

The criteria to evaluate the runs against. It can be:

a mapping of a criterion name to its description
a single criterion name present in one of the default criteria
a single ConstitutionalPrinciple instance

Returns:

A dictionary mapping criterion names to descriptions.

Return type:

Dict[str, str]

Examples

>>> criterion = "relevance"
>>> CriteriaEvalChain.resolve_criteria(criteria)
{'relevance': 'Is the submission referring to a real quote from the text?'}

run(*args: Any, callbacks: List[BaseCallbackHandler] | BaseCallbackManager | None = None, tags: List[str] | None = None, metadata: Dict[str, Any] | None = None, **kwargs: Any) → Any#

Deprecated since version langchain==0.1.0: Use invoke instead.

Convenience method for executing chain.

Parameters:

*args (Any) – If the chain expects a single input, it can be passed in as the sole positional argument.
callbacks (List[BaseCallbackHandler] | BaseCallbackManager | None) – Callbacks to use for this chain run. These will be called in addition to callbacks passed to the chain during construction, but only these runtime callbacks will propagate to calls to other objects.
tags (List[str] | None) – List of string tags to pass to all callbacks. These will be passed in addition to tags passed to the chain during construction, but only these runtime tags will propagate to calls to other objects.
**kwargs (Any) – If the chain expects multiple inputs, they can be passed in directly as keyword arguments.
metadata (Dict[str, Any] | None)
**kwargs

Returns:

The chain output.

Return type:

Any

Example

# Suppose we have a single-input chain that takes a 'question' string:
chain.run("What's the temperature in Boise, Idaho?")
# -> "The temperature in Boise is..."

# Suppose we have a multi-input chain that takes a 'question' string
# and 'context' string:
question = "What's the temperature in Boise, Idaho?"
context = "Weather report for Boise, Idaho on 07/03/23..."
chain.run(question=question, context=context)
# -> "The temperature in Boise is..."

save(file_path: Path | str) → None#

Save the chain.

Expects Chain._chain_type property to be implemented and for memory to be: null.

Parameters:: file_path (Path | str) – Path to file to save the chain to.
Return type:: None

Example

chain.save(file_path="path/chain.yaml")

stream(input: Input, config: RunnableConfig | None = None, **kwargs: Any | None) → Iterator[Output]#

Default implementation of stream, which calls invoke. Subclasses should override this method if they support streaming output.

Parameters:

input (Input) – The input to the Runnable.
config (RunnableConfig | None) – The config to use for the Runnable. Defaults to None.
kwargs (Any | None) – Additional keyword arguments to pass to the Runnable.

Yields:

The output of the Runnable.

Return type:

Iterator[Output]

to_json() → SerializedConstructor | SerializedNotImplemented#

Serialize the Runnable to JSON.

Returns:: A JSON-serializable representation of the Runnable.
Return type:: SerializedConstructor | SerializedNotImplemented

property evaluation_name: str#

Get the name of the evaluation.

Returns:: The name of the evaluation.
Return type:: str

property requires_input: bool#: Whether this evaluator requires an input string.

property requires_reference: bool#: Whether the evaluation requires a reference text.