<!--
Thank you for contributing to LangChain! Your PR will appear in our
release under the title you set. Please make sure it highlights your
valuable contribution.
Replace this with a description of the change, the issue it fixes (if
applicable), and relevant context. List any dependencies required for
this change.
After you're done, someone will review your PR. They may suggest
improvements. If no one reviews your PR within a few days, feel free to
@-mention the same people again, as notifications can get lost.
Finally, we'd love to show appreciation for your contribution - if you'd
like us to shout you out on Twitter, please also include your handle!
-->
<!-- Remove if not applicable -->
Fixes # (issue)
#### Before submitting
<!-- If you're adding a new integration, please include:
1. a test for the integration - favor unit tests that does not rely on
network access.
2. an example notebook showing its use
See contribution guidelines for more information on how to write tests,
lint
etc:
https://github.com/hwchase17/langchain/blob/master/.github/CONTRIBUTING.md
-->
#### Who can review?
Tag maintainers/contributors who might be interested:
<!-- For a quicker response, figure out the right person to tag with @
@hwchase17 - project lead
Tracing / Callbacks
- @agola11
Async
- @agola11
DataLoaders
- @eyurtsev
Models
- @hwchase17
- @agola11
Agents / Tools / Toolkits
- @hwchase17
VectorStores / Retrievers / Memory
- @dev2049
-->
1. Added use cases of the new features
2. Done some code refactoring
---------
Co-authored-by: Ivo Stranic <istranic@gmail.com>
- [Xorbits](https://doc.xorbits.io/en/latest/) is an open-source
computing framework that makes it easy to scale data science and machine
learning workloads in parallel. Xorbits can leverage multi cores or GPUs
to accelerate computation on a single machine, or scale out up to
thousands of machines to support processing terabytes of data.
- This PR added support for the Xorbits document loader, which allows
langchain to leverage Xorbits to parallelize and distribute the loading
of data.
- Dependencies: This change requires the Xorbits library to be installed
in order to be used.
`pip install xorbits`
- Request for review: @rlancemartin, @eyurtsev
- Twitter handle: https://twitter.com/Xorbitsio
Co-authored-by: Bagatur <baskaryan@gmail.com>
- Description: Adding async method for CTransformers
- Issue: I've found impossible without this code to run Websockets
inside a FastAPI micro service and a CTransformers model.
- Tag maintainer: Not necessary yet, I don't like to mention directly
- Twitter handle: @_semoal
Adding a maximal_marginal_relevance method to the
MongoDBAtlasVectorSearch vectorstore enhances the user experience by
providing more diverse search results
Issue: #7304
### Summary
Adds an `UnstructuredTSVLoader` for TSV files. Also updates the doc
strings for `UnstructuredCSV` and `UnstructuredExcel` loaders.
### Testing
```python
from langchain.document_loaders.tsv import UnstructuredTSVLoader
loader = UnstructuredTSVLoader(
file_path="example_data/mlb_teams_2012.csv", mode="elements"
)
docs = loader.load()
```
`SpacyTextSplitter` currently uses spacy's statistics-based
`en_core_web_sm` model for sentence splitting. This is a good splitter,
but it's also pretty slow, and in this case it's doing a lot of work
that's not needed given that the spacy parse is then just thrown away.
However, there is also a simple rules-based spacy sentencizer. Using
this is at least an order of magnitude faster than using
`en_core_web_sm` according to my local tests.
Also, spacy sentence tokenization based on `en_core_web_sm` can be sped
up in this case by not doing the NER stage. This shaves some cycles too,
both when loading the model and when parsing the text.
Consequently, this PR adds the option to use the basic spacy
sentencizer, and it disables the NER stage for the current approach,
*which is kept as the default*.
Lastly, when extracting the tokenized sentences, the `text` attribute is
called directly instead of doing the string conversion, which is IMO a
bit more idiomatic.
Hey @hwchase17 -
This PR adds a `ZepMemory` class, improves handling of Zep's message
metadata, and makes it easier for folks building custom chains to
persist metadata alongside their chat history.
We've had plenty confused users unfamiliar with ChatMessageHistory
classes and how to wrap the `ZepChatMessageHistory` in a
`ConversationBufferMemory`. So we've created the `ZepMemory` class as a
light wrapper for `ZepChatMessageHistory`.
Details:
- add ZepMemory, modify notebook to demo use of ZepMemory
- Modify summary to be SystemMessage
- add metadata argument to add_message; add Zep metadata to
Message.additional_kwargs
- support passing in metadata
Have noticed transient ref example misalignment. I believe this is
caused by the logic of assigning an example within the thread executor
rather than before.
Current problems:
1. Evaluating LLMs or Chat models isn't smooth. Even specifying
'generations' as the output inserts a redundant list into the eval
template
2. Configuring input / prediction / reference keys in the
`get_qa_evaluator` function is confusing. Unless you are using a chain
with the default keys, you have to specify all the variables and need to
reason about whether the key corresponds to the traced run's inputs,
outputs or the examples inputs or outputs.
Proposal:
- Configure the run evaluator according to a model. Use the model type
and input/output keys to assert compatibility where possible. Only need
to specify a reference_key for certain evaluators (which is less
confusing than specifying input keys)
When does this work:
- If you have your langchain model available (assumed always for
run_on_dataset flow)
- If you are evaluating an LLM, Chat model, or chain
- If the LLM or chat models are traced by langchain (wouldn't work if
you add an incompatible schema via the REST API)
When would this fail:
- Currently if you directly create an example from an LLM run, the
outputs are generations with all the extra metadata present. A simple
`example_key` and dumping all to the template could make the evaluations
unreliable
- Doesn't help if you're not using the low level API
- If you want to instantiate the evaluator without instantiating your
chain or LLM (maybe common for monitoring, for instance) -> could also
load from run or run type though
What's ugly:
- Personally think it's better to load evaluators one by one since
passing a config down is pretty confusing.
- Lots of testing needs to be added
- Inconsistent in that it makes a separate run and example input mapper
instead of the original `RunEvaluatorInputMapper`, which maps a run and
example to a single input.
Example usage running the for an LLM, Chat Model, and Agent.
```
# Test running for the string evaluators
evaluator_names = ["qa", "criteria"]
model = ChatOpenAI()
configured_evaluators = load_run_evaluators_for_model(evaluator_names, model=model, reference_key="answer")
run_on_dataset(ds_name, model, run_evaluators=configured_evaluators)
```
<details>
<summary>Full code with dataset upload</summary>
```
## Create dataset
from langchain.evaluation.run_evaluators.loading import load_run_evaluators_for_model
from langchain.evaluation import load_dataset
import pandas as pd
lcds = load_dataset("llm-math")
df = pd.DataFrame(lcds)
from uuid import uuid4
from langsmith import Client
client = Client()
ds_name = "llm-math - " + str(uuid4())[0:8]
ds = client.upload_dataframe(df, name=ds_name, input_keys=["question"], output_keys=["answer"])
## Define the models we'll test over
from langchain.llms import OpenAI
from langchain.chat_models import ChatOpenAI
from langchain.agents import initialize_agent, AgentType
from langchain.tools import tool
llm = OpenAI(temperature=0)
chat_model = ChatOpenAI(temperature=0)
@tool
def sum(a: float, b: float) -> float:
"""Add two numbers"""
return a + b
def construct_agent():
return initialize_agent(
llm=chat_model,
tools=[sum],
agent=AgentType.OPENAI_MULTI_FUNCTIONS,
)
agent = construct_agent()
# Test running for the string evaluators
evaluator_names = ["qa", "criteria"]
models = [llm, chat_model, agent]
run_evaluators = []
for model in models:
run_evaluators.append(load_run_evaluators_for_model(evaluator_names, model=model, reference_key="answer"))
# Run on LLM, Chat Model, and Agent
from langchain.client.runner_utils import run_on_dataset
to_test = [llm, chat_model, construct_agent]
for model, configured_evaluators in zip(to_test, run_evaluators):
run_on_dataset(ds_name, model, run_evaluators=configured_evaluators, verbose=True)
```
</details>
---------
Co-authored-by: Nuno Campos <nuno@boringbits.io>
Continuing with Tolkien inspired series of langchain tools. I bring to
you:
**The Fellowship of the Vectors**, AKA EmbeddingsClusteringFilter.
This document filter uses embeddings to group vectors together into
clusters, then allows you to pick an arbitrary number of documents
vector based on proximity to the cluster centers. That's a
representative sample of the cluster.
The original idea is from [Greg Kamradt](https://github.com/gkamradt)
from this video (Level4):
https://www.youtube.com/watch?v=qaPMdcCqtWk&t=365s
I added few tricks to make it a bit more versatile, so you can
parametrize what to do with duplicate documents in case of cluster
overlap: replace the duplicates with the next closest document or remove
it. This allow you to use it as an special kind of redundant filter too.
Additionally you can choose 2 diff orders: grouped by cluster or
respecting the original retriever scores.
In my use case I was using the docs grouped by cluster to run refine
chains per cluster to generate summarization over a large corpus of
documents.
Let me know if you want to change anything!
@rlancemartin, @eyurtsev, @hwchase17,
---------
Co-authored-by: rlm <pexpresss31@gmail.com>
- Description: pydantic's `ModelField.type_` only exposes the native
data type but not complex type hints like `List`. Thus, generating a
Tool with `from_function` through function signature produces incorrect
argument schemas (e.g., `str` instead of `List[str]`)
- Issue: N/A
- Dependencies: N/A
- Tag maintainer: @hinthornw
- Twitter handle: `mapped`
All the unittest (with an additional one in this PR) passed, though I
didn't try integration tests...
- Description: Adding to Chroma integration the option to run a
similarity search by a vector with relevance scores. Fixing two minor
typos.
- Issue: The "lambda_mult" typo is related to #4861
- Maintainer: @rlancemartin, @eyurtsev
- Description:
- When `keep_separator` is `True` the `_split_text_with_regex()` method
in `text_splitter` uses regex to split, but when `keep_separator` is
`False` it uses `str.split()`. This causes problems when the separator
is a special regex character like `.` or `*`. This PR fixes that by
using `re.split()` in both cases.
- Issue: #7262
- Tag maintainer: @baskaryan
### Description
This pull request introduces the "Cube Semantic Layer" document loader,
which demonstrates the retrieval of Cube's data model metadata in a
format suitable for passing to LLMs as embeddings. This enhancement aims
to provide contextual information and improve the understanding of data.
Twitter handle:
@the_cube_dev
---------
Co-authored-by: rlm <pexpresss31@gmail.com>
This PR brings in a vectorstore interface for
[Marqo](https://www.marqo.ai/).
The Marqo vectorstore exposes some of Marqo's functionality in addition
the the VectorStore base class. The Marqo vectorstore also makes the
embedding parameter optional because inference for embeddings is an
inherent part of Marqo.
Docs, notebook examples and integration tests included.
Related PR:
https://github.com/hwchase17/langchain/pull/2807
---------
Co-authored-by: Tom Hamer <tom@marqo.ai>
Co-authored-by: Harrison Chase <hw.chase.17@gmail.com>
- Description: Allow `InMemoryDocstore` to be created without passing a
dict to the constructor; the constructor can create a dict at runtime if
one isn't provided.
- Tag maintainer: @dev2049
Hi, there
This pull request contains two commit:
**1. Implement delete interface with optional ids parameter on
AnalyticDB.**
**2. Allow customization of database connection behavior by exposing
engine_args parameter in interfaces.**
- This commit adds the `engine_args` parameter to the interfaces,
allowing users to customize the behavior of the database connection. The
`engine_args` parameter accepts a dictionary of additional arguments
that will be passed to the create_engine function. Users can now modify
various aspects of the database connection, such as connection pool size
and recycle time. This enhancement provides more flexibility and control
to users when interacting with the database through the exposed
interfaces.
This commit is related to VectorStores @rlancemartin @eyurtsev
Thank you for your attention and consideration.
This fixes#4833 and the critical vulnerability
https://nvd.nist.gov/vuln/detail/CVE-2023-34540
Previously, the JIRA API Wrapper had a mode that simply pipelined user
input into an `exec()` function.
[The intended use of the 'other' mode is to cover any of Atlassian's API
that don't have an existing
interface](cc33bde74f/langchain/tools/jira/prompt.py (L24))
Fortunately all of the [Atlassian JIRA API methods are subfunctions of
their `Jira`
class](https://atlassian-python-api.readthedocs.io/jira.html), so this
implementation calls these subfunctions directly.
As well as passing a string representation of the function to call, the
implementation flexibly allows for optionally passing args and/or
keyword-args. These are given as part of the dictionary input. Example:
```
{
"function": "update_issue_field", #function to execute
"args": [ #list of ordered args similar to other examples in this JiraAPIWrapper
"key",
{"summary": "New summary"}
],
"kwargs": {} #dict of key value keyword-args pairs
}
```
the above is equivalent to `self.jira.update_issue_field("key",
{"summary": "New summary"})`
Alternate query schema designs are welcome to make querying easier
without passing and evaluating arbitrary python code. I considered
parsing (without evaluating) input python code and extracting the
function, args, and kwargs from there and then pipelining them into the
callable function via `*f(args, **kwargs)` - but this seemed more
direct.
@vowelparrot @dev2049
---------
Co-authored-by: Jamal Rahman <jamal.rahman@builder.ai>
## Description
Added Office365 tool modules to `__init__.py` files
## Issue
As described in Issue
https://github.com/hwchase17/langchain/issues/6936, the Office365
toolkit can't be loaded easily because it is not included in the
`__init__.py` files.
## Reviewer
@dev2049
- [x] wire up tools
- [x] wire up retrievers
- [x] add integration test
<!-- Thank you for contributing to LangChain!
Replace this comment with:
- Description: a description of the change,
- Issue: the issue # it fixes (if applicable),
- Dependencies: any dependencies required for this change,
- Tag maintainer: for a quicker response, tag the relevant maintainer
(see below),
- Twitter handle: we announce bigger features on Twitter. If your PR
gets announced and you'd like a mention, we'll gladly shout you out!
If you're adding a new integration, please include:
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use.
Maintainer responsibilities:
- General / Misc / if you don't know who to tag: @baskaryan
- DataLoaders / VectorStores / Retrievers: @rlancemartin, @eyurtsev
- Models / Prompts: @hwchase17, @baskaryan
- Memory: @hwchase17
- Agents / Tools / Toolkits: @hinthornw
- Tracing / Callbacks: @agola11
- Async: @agola11
If no one reviews your PR within a few days, feel free to @-mention the
same people again.
See contribution guidelines for more information on how to write/run
tests, lint, etc:
https://github.com/hwchase17/langchain/blob/master/.github/CONTRIBUTING.md
-->
- Description: If their are missing or extra variables when validating
Jinja 2 template then a warning is issued rather than raising an
exception. This allows for better flexibility for the developer as
described in #7044. Also changed the relevant test so pytest is checking
for raised warnings rather than exceptions.
- Issue: #7044
- Tag maintainer: @hwchase17, @baskaryan
---------
Co-authored-by: Harrison Chase <hw.chase.17@gmail.com>
<!-- Thank you for contributing to LangChain!
Replace this comment with:
- Description: a description of the change,
- Issue: the issue # it fixes (if applicable),
- Dependencies: any dependencies required for this change,
- Tag maintainer: for a quicker response, tag the relevant maintainer
(see below),
- Twitter handle: we announce bigger features on Twitter. If your PR
gets announced and you'd like a mention, we'll gladly shout you out!
If you're adding a new integration, please include:
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use.
Maintainer responsibilities:
- General / Misc / if you don't know who to tag: @dev2049
- DataLoaders / VectorStores / Retrievers: @rlancemartin, @eyurtsev
- Models / Prompts: @hwchase17, @dev2049
- Memory: @hwchase17
- Agents / Tools / Toolkits: @vowelparrot
- Tracing / Callbacks: @agola11
- Async: @agola11
If no one reviews your PR within a few days, feel free to @-mention the
same people again.
See contribution guidelines for more information on how to write/run
tests, lint, etc:
https://github.com/hwchase17/langchain/blob/master/.github/CONTRIBUTING.md
-->
# [SPARQL](https://www.w3.org/TR/rdf-sparql-query/) for
[LangChain](https://github.com/hwchase17/langchain)
## Description
LangChain support for knowledge graphs relying on W3C standards using
RDFlib: SPARQL/ RDF(S)/ OWL with special focus on RDF \
* Works with local files, files from the web, and SPARQL endpoints
* Supports both SELECT and UPDATE queries
* Includes both a Jupyter notebook with an example and integration tests
## Contribution compared to related PRs and discussions
* [Wikibase agent](https://github.com/hwchase17/langchain/pull/2690) -
uses SPARQL, but specifically for wikibase querying
* [Cypher qa](https://github.com/hwchase17/langchain/pull/5078) - graph
DB question answering for Neo4J via Cypher
* [PR 6050](https://github.com/hwchase17/langchain/pull/6050) - tries
something similar, but does not cover UPDATE queries and supports only
RDF
* Discussions on [w3c mailing list](mailto:semantic-web@w3.org) related
to the combination of LLMs (specifically ChatGPT) and knowledge graphs
## Dependencies
* [RDFlib](https://github.com/RDFLib/rdflib)
## Tag maintainer
Graph database related to memory -> @hwchase17
[Apache HugeGraph](https://github.com/apache/incubator-hugegraph) is a
convenient, efficient, and adaptable graph database, compatible with the
Apache TinkerPop3 framework and the Gremlin query language.
In this PR, the HugeGraph and HugeGraphQAChain provide the same
functionality as the existing integration with Neo4j and enables query
generation and question answering over HugeGraph database. The
difference is that the graph query language supported by HugeGraph is
not cypher but another very popular graph query language
[Gremlin](https://tinkerpop.apache.org/gremlin.html).
A notebook example and a simple test case have also been added.
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
<!-- Thank you for contributing to LangChain!
Replace this comment with:
- Description: a description of the change,
- Issue: the issue # it fixes (if applicable),
- Dependencies: any dependencies required for this change,
- Tag maintainer: for a quicker response, tag the relevant maintainer
(see below),
- Twitter handle: we announce bigger features on Twitter. If your PR
gets announced and you'd like a mention, we'll gladly shout you out!
If you're adding a new integration, please include:
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use.
Maintainer responsibilities:
- General / Misc / if you don't know who to tag: @dev2049
- DataLoaders / VectorStores / Retrievers: @rlancemartin, @eyurtsev
- Models / Prompts: @hwchase17, @dev2049
- Memory: @hwchase17
- Agents / Tools / Toolkits: @vowelparrot
- Tracing / Callbacks: @agola11
- Async: @agola11
If no one reviews your PR within a few days, feel free to @-mention the
same people again.
See contribution guidelines for more information on how to write/run
tests, lint, etc:
https://github.com/hwchase17/langchain/blob/master/.github/CONTRIBUTING.md
-->
<!-- Thank you for contributing to LangChain!
Replace this comment with:
- Description: a description of the change,
- Issue: the issue # it fixes (if applicable),
- Dependencies: any dependencies required for this change,
- Tag maintainer: for a quicker response, tag the relevant maintainer
(see below),
- Twitter handle: we announce bigger features on Twitter. If your PR
gets announced and you'd like a mention, we'll gladly shout you out!
If you're adding a new integration, please include:
1. a test for the integration, preferably unit tests that do not rely on
network access,
2. an example notebook showing its use.
Maintainer responsibilities:
- General / Misc / if you don't know who to tag: @dev2049
- DataLoaders / VectorStores / Retrievers: @rlancemartin, @eyurtsev
- Models / Prompts: @hwchase17, @dev2049
- Memory: @hwchase17
- Agents / Tools / Toolkits: @vowelparrot
- Tracing / Callbacks: @agola11
- Async: @agola11
If no one reviews your PR within a few days, feel free to @-mention the
same people again.
See contribution guidelines for more information on how to write/run
tests, lint, etc:
https://github.com/hwchase17/langchain/blob/master/.github/CONTRIBUTING.md
-->
## Description
The type hint for `FAISS.__init__()`'s `relevance_score_fn` parameter
allowed the parameter to be set to `None`. However, a default function
is provided by the constructor. This led to an unnecessary check in the
code, as well as a test to verify this check.
**ASSUMPTION**: There's no reason to ever set `relevance_score_fn` to
`None`.
This PR changes the type hint and removes the unnecessary code.
**Description**:
The JSON Lines format is used by some services such as OpenAI and
HuggingFace. It's also a convenient alternative to CSV.
This PR adds JSON Lines support to `JSONLoader` and also updates related
tests.
**Tag maintainer**: @rlancemartin, @eyurtsev.
PS I was not able to build docs locally so didn't update related
section.
Update to Vectara integration
- By user request added "add_files" to take advantage of Vectara
capabilities to process files on the backend, without the need for
separate loading of documents and chunking in the chain.
- Updated vectara.ipynb example notebook to be broader and added testing
of add_file()
@hwchase17 - project lead
---------
Co-authored-by: rlm <pexpresss31@gmail.com>
Retrying with the same improvements as in #6772, this time trying not to
mess up with branches.
@rlancemartin doing a fresh new PR from a branch with a new name. This
should do. Thank you for your help!
---------
Co-authored-by: Jonathan Ellis <jbellis@datastax.com>
Co-authored-by: rlm <pexpresss31@gmail.com>
should be no functional changes
also keep __init__ exposing a lot for backwards compat
---------
Co-authored-by: Dev 2049 <dev.dev2049@gmail.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
### Summary
Updates `UnstructuredEmailLoader` so that it can process attachments in
addition to the e-mail content. The loader will process attachments if
the `process_attachments` kwarg is passed when the loader is
instantiated.
### Testing
```python
file_path = "fake-email-attachment.eml"
loader = UnstructuredEmailLoader(
file_path, mode="elements", process_attachments=True
)
docs = loader.load()
docs[-1]
```
### Reviewers
- @rlancemartin
- @eyurtsev
- @hwchase17
Handle the new retriever events in a way that (I think) is entirely
backwards compatible? Needs more testing for some of the chain changes
and all.
This creates an entire new run type, however. We could also just treat
this as an event within a chain run presumably (same with memory)
Adds a subclass initializer that upgrades old retriever implementations
to the new schema, along with tests to ensure they work.
First commit doesn't upgrade any of our retriever implementations (to
show that we can pass the tests along with additional ones testing the
upgrade logic).
Second commit upgrades the known universe of retrievers in langchain.
- [X] Add callback handling methods for retriever start/end/error (open
to renaming to 'retrieval' if you want that)
- [X] Update BaseRetriever schema to support callbacks
- [X] Tests for upgrading old "v1" retrievers for backwards
compatibility
- [X] Update existing retriever implementations to implement the new
interface
- [X] Update calls within chains to .{a]get_relevant_documents to pass
the child callback manager
- [X] Update the notebooks/docs to reflect the new interface
- [X] Test notebooks thoroughly
Not handled:
- Memory pass throughs: retrieval memory doesn't have a parent callback
manager passed through the method
---------
Co-authored-by: Nuno Campos <nuno@boringbits.io>
Co-authored-by: William Fu-Hinthorn <13333726+hinthornw@users.noreply.github.com>
If you create a dataset from runs and run the same chain or llm on it
later, it usually works great.
If you have an agent dataset and want to run a different agent on it, or
have more complex schema, it's hard for us to automatically map these
values every time. This PR lets you pass in an input_mapper function
that converts the example inputs to whatever format your model expects
