OptionalcallbacksOptional callbacks to handle various events in the retrieval process.
OptionalfilterOptional filter applied to search results, defined by the FilterType of the vector store.
Allows for refined, targeted results by restricting the returned documents based
on specified filter criteria.
Specifies the number of documents to retrieve for each search query. Defaults to 4 if not specified, providing a basic result count for similarity or MMR searches.
OptionalmetadataMetadata to provide additional context or information about the retrieval operation.
OptionalnameOptionalsearchAdditional options specific to maximal marginal relevance (MMR) search, applicable
only if searchType is set to "mmr".
Includes:
fetchK: The initial number of documents fetched before applying the MMR algorithm,
allowing for a larger selection from which to choose the most diverse results.lambda: A parameter between 0 and 1 to adjust the relevance-diversity balance,
where 0 prioritizes diversity and 1 prioritizes relevance.Determines the type of search operation to perform on the vector store.
"similarity" (default): Conducts a similarity search based purely on vector similarity
to the query."mmr": Executes a maximal marginal relevance (MMR) search, balancing relevance and
diversity in the retrieved results.OptionaltagsTags to label or categorize the retrieval operation.
The instance of VectorStore used for storing and retrieving document embeddings.
This vector store must implement the VectorStoreInterface to be compatible
with the retriever’s operations.
OptionalverboseIf set to true, enables verbose logging for the retrieval process.
Override the default addDocuments method to embed the documents twice,
once using the larger embeddings model, and then again using the default
embedding model linked to the vector store.
An array of documents to add to the vector store.
Optionaloptions: AddDocumentOptionsAn optional object containing additional options for adding documents.
A promise that resolves to an array of the document IDs that were added to the vector store.
Convert a runnable to a tool. Return a new instance of RunnableToolLike
which contains the runnable, name, description and schema.
Optionaldescription?: stringThe description of the tool. Falls back to the description on the Zod schema if not provided, or undefined if neither are provided.
Optionalname?: stringThe name of the tool. If not provided, it will default to the name of the runnable.
The Zod schema for the input of the tool. Infers the Zod type from the input type of the runnable.
An instance of RunnableToolLike which is a runnable that can be used as a tool.
Default implementation of batch, which calls invoke N times. Subclasses should override this method if they can batch more efficiently.
Array of inputs to each batch call.
Optionaloptions: Partial<RunnableConfig<Record<string, any>>> | Partial<RunnableConfig<Record<string, any>>>[]Either a single call options object to apply to each batch call or an array for each call.
OptionalbatchOptions: RunnableBatchOptions & { An array of RunOutputs, or mixed RunOutputs and errors if batchOptions.returnExceptions is set
Optionaloptions: Partial<RunnableConfig<Record<string, any>>> | Partial<RunnableConfig<Record<string, any>>>[]OptionalbatchOptions: RunnableBatchOptions & { Optionaloptions: Partial<RunnableConfig<Record<string, any>>> | Partial<RunnableConfig<Record<string, any>>>[]OptionalbatchOptions: RunnableBatchOptionsBind arguments to a Runnable, returning a new Runnable.
A new RunnableBinding that, when invoked, will apply the bound args.
The query string to retrieve relevant documents for.
Optionalconfig: Callbacks | BaseCallbackConfigOptional configuration object for the retrieval process.
A promise that resolves to an array of Document objects.
Use .invoke() instead. Will be removed in 0.3.0.
Main method used to retrieve relevant documents. It takes a query
string and an optional configuration object, and returns a promise that
resolves to an array of Document objects. This method handles the
retrieval process, including starting and ending callbacks, and error
handling.
Executes a retrieval operation.
The query string used to search for relevant documents.
Optionaloptions: RunnableConfig<Record<string, any>>(optional) Configuration options for the retrieval run, which may include callbacks, tags, and metadata.
A promise that resolves to an array of DocumentInterface instances
representing the most relevant documents to the query.
Return a new Runnable that maps a list of inputs to a list of outputs, by calling invoke() with each input.
Create a new runnable sequence that runs each individual runnable in series, piping the output of one runnable into another runnable or runnable-like.
A runnable, function, or object whose values are functions or runnables.
A new runnable sequence.
Stream output in chunks.
Optionaloptions: Partial<RunnableConfig<Record<string, any>>>A readable stream that is also an iterable.
Generate a stream of events emitted by the internal steps of the runnable.
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: string - Event names are of the format: on_[runnable_type]_(start|stream|end).name: string - The name of the runnable that generated the event.run_id: string - 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.tags: string[] - The tags of the runnable that generated the event.metadata: Record<string, any> - The metadata of the runnable that generated the event.data: Record<string, any>Below is a table that illustrates some events 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 | input | output/chunk |
+======================+=============================+==========================================+
| on_chat_model_start | {"messages": BaseMessage[]} | |
+----------------------+-----------------------------+------------------------------------------+
| on_chat_model_stream | | AIMessageChunk("hello") |
+----------------------+-----------------------------+------------------------------------------+
| on_chat_model_end | {"messages": BaseMessage[]} | AIMessageChunk("hello world") |
+----------------------+-----------------------------+------------------------------------------+
| on_llm_start | {'input': 'hello'} | |
+----------------------+-----------------------------+------------------------------------------+
| on_llm_stream | | 'Hello' |
+----------------------+-----------------------------+------------------------------------------+
| on_llm_end | 'Hello human!' | |
+----------------------+-----------------------------+------------------------------------------+
| on_chain_start | | |
+----------------------+-----------------------------+------------------------------------------+
| on_chain_stream | | "hello world!" |
+----------------------+-----------------------------+------------------------------------------+
| on_chain_end | [Document(...)] | "hello world!, goodbye world!" |
+----------------------+-----------------------------+------------------------------------------+
| on_tool_start | {"x": 1, "y": "2"} | |
+----------------------+-----------------------------+------------------------------------------+
| on_tool_end | | {"x": 1, "y": "2"} |
+----------------------+-----------------------------+------------------------------------------+
| on_retriever_start | {"query": "hello"} | |
+----------------------+-----------------------------+------------------------------------------+
| on_retriever_end | {"query": "hello"} | [Document(...), ..] |
+----------------------+-----------------------------+------------------------------------------+
| on_prompt_start | {"question": "hello"} | |
+----------------------+-----------------------------+------------------------------------------+
| on_prompt_end | {"question": "hello"} | ChatPromptValue(messages: BaseMessage[]) |
+----------------------+-----------------------------+------------------------------------------+
The "on_chain_*" events are the default for Runnables that don't fit one of the above categories.
In addition to the standard events above, users can also dispatch custom events.
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. |
+-----------+------+------------------------------------------------------------+
Here's an example:
import { RunnableLambda } from "@langchain/core/runnables";
import { dispatchCustomEvent } from "@langchain/core/callbacks/dispatch";
// Use this import for web environments that don't support "async_hooks"
// and manually pass config to child runs.
// import { dispatchCustomEvent } from "@langchain/core/callbacks/dispatch/web";
const slowThing = RunnableLambda.from(async (someInput: string) => {
// Placeholder for some slow operation
await new Promise((resolve) => setTimeout(resolve, 100));
await dispatchCustomEvent("progress_event", {
message: "Finished step 1 of 2",
});
await new Promise((resolve) => setTimeout(resolve, 100));
return "Done";
});
const eventStream = await slowThing.streamEvents("hello world", {
version: "v2",
});
for await (const event of eventStream) {
if (event.event === "on_custom_event") {
console.log(event);
}
}
OptionalstreamOptions: Omit<EventStreamCallbackHandlerInput, "autoClose">OptionalstreamOptions: Omit<EventStreamCallbackHandlerInput, "autoClose">Stream all output from a runnable, as reported to the callback system. This includes all inner runs of LLMs, Retrievers, Tools, etc. Output is streamed as Log objects, which include a list of jsonpatch ops that describe how the state of the run has changed in each step, and the final state of the run. The jsonpatch ops can be applied in order to construct state.
Optionaloptions: Partial<RunnableConfig<Record<string, any>>>OptionalstreamOptions: Omit<LogStreamCallbackHandlerInput, "autoClose">Default implementation of transform, which buffers input and then calls stream. Subclasses should override this method if they can start producing output while input is still being generated.
Bind config to a Runnable, returning a new Runnable.
New configuration parameters to attach to the new runnable.
A new RunnableBinding with a config matching what's passed.
Create a new runnable from the current one that will try invoking other passed fallback runnables if the initial invocation fails.
A new RunnableWithFallbacks.
Bind lifecycle listeners to a Runnable, returning a new Runnable. The Run object contains information about the run, including its id, type, input, output, error, startTime, endTime, and any tags or metadata added to the run.
The object containing the callback functions.
OptionalonCalled after the runnable finishes running, with the Run object.
Optionalconfig: RunnableConfig<Record<string, any>>OptionalonCalled if the runnable throws an error, with the Run object.
Optionalconfig: RunnableConfig<Record<string, any>>OptionalonCalled before the runnable starts running, with the Run object.
Optionalconfig: RunnableConfig<Record<string, any>>Add retry logic to an existing runnable.
Optionalfields: { OptionalonOptionalstopA new RunnableRetry that, when invoked, will retry according to the parameters.
Staticis
A retriever that uses two sets of embeddings to perform adaptive retrieval. Based off of the "Matryoshka embeddings: faster OpenAI vector search using Adaptive Retrieval" blog post https://supabase.com/blog/matryoshka-embeddings.
This class performs "Adaptive Retrieval" for searching text embeddings efficiently using the Matryoshka Representation Learning (MRL) technique. It retrieves documents similar to a query embedding in two steps:
First-pass: Uses a lower dimensional sub-vector from the MRL embedding for an initial, fast, but less accurate search.
Second-pass: Re-ranks the top results from the first pass using the full, high-dimensional embedding for higher accuracy.
This code implements MRL embeddings for efficient vector search by combining faster, lower-dimensional initial search with accurate, high-dimensional re-ranking.