- **Description:** Adding Baichuan Text Embedding Model and Baichuan Inc
introduction.
Baichuan Text Embedding ranks #1 in C-MTEB leaderboard:
https://huggingface.co/spaces/mteb/leaderboard
Co-authored-by: BaiChuanHelper <wintergyc@WinterGYCs-MacBook-Pro.local>
- **Description:** This PR adds [EdenAI](https://edenai.co/) for the
chat model (already available in LLM & Embeddings). It supports all
[ChatModel] functionality: generate, async generate, stream, astream and
batch. A detailed notebook was added.
- **Dependencies**: No dependencies are added as we call a rest API.
---------
Co-authored-by: Eugene Yurtsev <eyurtsev@gmail.com>
community:
- **Description:**
- Add new ChatLiteLLMRouter class that allows a client to use a LiteLLM
Router as a LangChain chat model.
- Note: The existing ChatLiteLLM integration did not cover the LiteLLM
Router class.
- Add tests and Jupyter notebook.
- **Issue:** None
- **Dependencies:** Relies on existing ChatLiteLLM integration
- **Twitter handle:** @bburgin_0
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
<!-- Thank you for contributing to LangChain!
Please title your PR "<package>: <description>", where <package> is
whichever of langchain, community, core, experimental, etc. is being
modified.
Replace this entire comment with:
- **Description:** Adding Oracle Cloud Infrastructure Generative AI
integration. Oracle Cloud Infrastructure (OCI) Generative AI is a fully
managed service that provides a set of state-of-the-art, customizable
large language models (LLMs) that cover a wide range of use cases, and
which is available through a single API. Using the OCI Generative AI
service you can access ready-to-use pretrained models, or create and
host your own fine-tuned custom models based on your own data on
dedicated AI clusters.
https://docs.oracle.com/en-us/iaas/Content/generative-ai/home.htm
- **Issue:** None,
- **Dependencies:** OCI Python SDK,
- **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!
Please make sure your PR is passing linting and testing before
submitting. Run `make format`, `make lint` and `make test` from the root
of the package you've modified to check this locally.
Passed
See contribution guidelines for more information on how to write/run
tests, lint, etc: https://python.langchain.com/docs/contributing/
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. It lives in
`docs/docs/integrations` directory.
we provide unit tests. However, we cannot provide integration tests due
to Oracle policies that prohibit public sharing of api keys.
If no one reviews your PR within a few days, please @-mention one of
@baskaryan, @eyurtsev, @hwchase17.
-->
---------
Co-authored-by: Arthur Cheng <arthur.cheng@oracle.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
- **Description:**
This PR adds a VectorStore integration for SAP HANA Cloud Vector Engine,
which is an upcoming feature in the SAP HANA Cloud database
(https://blogs.sap.com/2023/11/02/sap-hana-clouds-vector-engine-announcement/).
- **Issue:** N/A
- **Dependencies:** [SAP HANA Python
Client](https://pypi.org/project/hdbcli/)
- **Twitter handle:** @sapopensource
Implementation of the integration:
`libs/community/langchain_community/vectorstores/hanavector.py`
Unit tests:
`libs/community/tests/unit_tests/vectorstores/test_hanavector.py`
Integration tests:
`libs/community/tests/integration_tests/vectorstores/test_hanavector.py`
Example notebook:
`docs/docs/integrations/vectorstores/hanavector.ipynb`
Access credentials for execution of the integration tests can be
provided to the maintainers.
---------
Co-authored-by: sascha <sascha.stoll@sap.com>
Co-authored-by: Bagatur <baskaryan@gmail.com>
- **Description:** This PR enables LangChain to access the iFlyTek's
Spark LLM via the chat_models wrapper.
- **Dependencies:** websocket-client ^1.6.1
- **Tag maintainer:** @baskaryan
### SparkLLM chat model usage
Get SparkLLM's app_id, api_key and api_secret from [iFlyTek SparkLLM API
Console](https://console.xfyun.cn/services/bm3) (for more info, see
[iFlyTek SparkLLM Intro](https://xinghuo.xfyun.cn/sparkapi) ), then set
environment variables `IFLYTEK_SPARK_APP_ID`, `IFLYTEK_SPARK_API_KEY`
and `IFLYTEK_SPARK_API_SECRET` or pass parameters when using it like the
demo below:
```python3
from langchain.chat_models.sparkllm import ChatSparkLLM
client = ChatSparkLLM(
spark_app_id="<app_id>",
spark_api_key="<api_key>",
spark_api_secret="<api_secret>"
)
```
This PR introduces update to Konko Integration with LangChain.
1. **New Endpoint Addition**: Integration of a new endpoint to utilize
completion models hosted on Konko.
2. **Chat Model Updates for Backward Compatibility**: We have updated
the chat models to ensure backward compatibility with previous OpenAI
versions.
4. **Updated Documentation**: Comprehensive documentation has been
updated to reflect these new changes, providing clear guidance on
utilizing the new features and ensuring seamless integration.
Thank you to the LangChain team for their exceptional work and for
considering this PR. Please let me know if any additional information is
needed.
---------
Co-authored-by: Shivani Modi <shivanimodi@Shivanis-MacBook-Pro.local>
Co-authored-by: Shivani Modi <shivanimodi@Shivanis-MBP.lan>
- **Description:** Baichuan Chat (with both Baichuan-Turbo and
Baichuan-Turbo-192K models) has updated their APIs. There are breaking
changes. For example, BAICHUAN_SECRET_KEY is removed in the latest API
but is still required in Langchain. Baichuan's Langchain integration
needs to be updated to the latest version.
- **Issue:** #15206
- **Dependencies:** None,
- **Twitter handle:** None
@hwchase17.
Co-authored-by: BaiChuanHelper <wintergyc@WinterGYCs-MacBook-Pro.local>
**Description:**
- Implement `SQLStrStore` and `SQLDocStore` classes that inherits from
`BaseStore` to allow to persist data remotely on a SQL server.
- SQL is widely used and sometimes we do not want to install a caching
solution like Redis.
- Multiple issues/comments complain that there is no easy remote and
persistent solution that are not in memory (users want to replace
InMemoryStore), e.g.,
https://github.com/langchain-ai/langchain/issues/14267,
https://github.com/langchain-ai/langchain/issues/15633,
https://github.com/langchain-ai/langchain/issues/14643,
https://stackoverflow.com/questions/77385587/persist-parentdocumentretriever-of-langchain
- This is particularly painful when wanting to use
`ParentDocumentRetriever `
- This implementation is particularly useful when:
* it's expensive to construct an InMemoryDocstore/dict
* you want to retrieve documents from remote sources
* you just want to reuse existing objects
- This implementation integrates well with PGVector, indeed, when using
PGVector, you already have a SQL instance running. `SQLDocStore` is a
convenient way of using this instance to store documents associated to
vectors. An integration example with ParentDocumentRetriever and
PGVector is provided in docs/docs/integrations/stores/sql.ipynb or
[here](https://github.com/gcheron/langchain/blob/sql-store/docs/docs/integrations/stores/sql.ipynb).
- It persists `str` and `Document` objects but can be easily extended.
**Issue:**
Provide an easy SQL alternative to `InMemoryStore`.
---------
Co-authored-by: Harrison Chase <hw.chase.17@gmail.com>
**Description** : New documents loader for visio files (with extension
.vsdx)
A [visio file](https://fr.wikipedia.org/wiki/Microsoft_Visio) (with
extension .vsdx) is associated with Microsoft Visio, a diagram creation
software. It stores information about the structure, layout, and
graphical elements of a diagram. This format facilitates the creation
and sharing of visualizations in areas such as business, engineering,
and computer science.
A Visio file can contain multiple pages. Some of them may serve as the
background for others, and this can occur across multiple layers. This
loader extracts the textual content from each page and its associated
pages, enabling the extraction of all visible text from each page,
similar to what an OCR algorithm would do.
**Dependencies** : xmltodict package
Description: Added support for asynchronous streaming in the Bedrock
class and corresponding tests.
Primarily:
async def aprepare_output_stream
async def _aprepare_input_and_invoke_stream
async def _astream
async def _acall
I've ensured that the code adheres to the project's linting and
formatting standards by running make format, make lint, and make test.
Issue: #12054, #11589
Dependencies: None
Tag maintainer: @baskaryan
Twitter handle: @dominic_lovric
---------
Co-authored-by: Piyush Jain <piyushjain@duck.com>
**Description:** Add support for querying TigerGraph databases through
the InquiryAI service.
**Issue**: N/A
**Dependencies:** N/A
**Twitter handle:** @TigerGraphDB
- **Description:** add support for kwargs in`MlflowEmbeddings`
`embed_document()` and `embed_query()` so that all the arguments
required by Cohere API (and others?) can be passed down to the server.
- **Issue:** #15234
- **Dependencies:** MLflow with MLflow Deployments (`pip install
mlflow[genai]`)
**Tests**
Now this code [adapted from the
docs](https://python.langchain.com/docs/integrations/providers/mlflow#embeddings-example)
for the Cohere API works locally.
```python
"""
Setup
-----
export COHERE_API_KEY=...
mlflow deployments start-server --config-path examples/deployments/cohere/config.yaml
Run
---
python /path/to/this/file.py
"""
embeddings = MlflowCohereEmbeddings(target_uri="http://127.0.0.1:5000", endpoint="embeddings")
print(embeddings.embed_query("hello")[:3])
print(embeddings.embed_documents(["hello", "world"])[0][:3])
```
Output
```
[0.060455322, 0.028793335, -0.025848389]
[0.031707764, 0.021057129, -0.009361267]
```
Titan Express model was not supported as a chat model because LangChain
messages were not "translated" to a text prompt.
Co-authored-by: Guillem Orellana Trullols <guillem.orellana_trullols@siemens.com>
Implement similarity function selector for ElasticsearchStore. The
scores coming back from Elasticsearch are already similarities (not
distances) and they are already normalized (see
[docs](https://www.elastic.co/guide/en/elasticsearch/reference/current/dense-vector.html#dense-vector-params)).
Hence we leave the scores untouched and just forward them.
This fixes#11539.
However, in hybrid mode (when keyword search and vector search are
involved) Elasticsearch currently returns no scores. This PR adds an
error message around this fact. We need to think a bit more to come up
with a solution for this case.
This PR also corrects a small error in the Elasticsearch integration
test.
---------
Co-authored-by: Erick Friis <erick@langchain.dev>
**Description:**
In this PR, I am adding a `PolygonLastQuote` Tool, which can be used to
get the latest price quote for a given ticker / stock.
Additionally, I've added a Polygon Toolkit, which we can use to
encapsulate future tools that we build for Polygon.
**Twitter handle:** [@virattt](https://twitter.com/virattt)
---------
Co-authored-by: Harrison Chase <hw.chase.17@gmail.com>
The timeout function comes in handy when you want to kill longrunning
queries.
The value sanitization removes all lists that are larger than 128
elements. The idea here is to remove embedding properties from results.
- **Description:** As Shell tool is very versatile, while integrating it
into applications as openai functions, developers have no clue about
what command is being executed using the ShellTool. All one can see is:
Summarising my feature request:
1. There's no visibility about what command was executed.
2. There's no mechanism to prevent a command to be executed using
ShellTool, like a y/n human input which can be accepted from user to
proceed with executing the command.,
- **Issue:** the issue #15931 it fixes if applicable,
- **Dependencies:** There isn't any dependancy,
- **Twitter handle:** @krishnashed
**Description:** This new feature enhances the flexibility of pipeline
integration, particularly when working with RESTful APIs.
``JsonRequestsWrapper`` allows for the decoding of JSON output, instead
of the only option for text output.
---------
Co-authored-by: Zhichao HAN <hanzhichao2000@hotmail.com>
## Feature
- Follow parameter structure as per official documentation
- top level parameters (e.g. model, system, template) will be passed as
top level parameters
- other parameters will be sent in options unless options is provided
## Tests
- Test if top level parameters handled properly
- Test if parameters that are not top level parameters are handled as
options
- Test if options is provided, it will be passed as is
Support [Lantern](https://github.com/lanterndata/lantern) as a new
VectorStore type.
- Added Lantern as VectorStore.
It will support 3 distance functions `l2 squared`, `cosine` and
`hamming` and will use `HNSW` index.
- Added tests
- Added example notebook
## Feature
- Set additional headers in constructor
- Headers will be sent in post request
This feature is useful if deploying Ollama on a cloud service such as
hugging face, which requires authentication tokens to be passed in the
request header.
## Tests
- Test if header is passed
- Test if header is not passed
<!-- Thank you for contributing to LangChain!
Please title your PR "<package>: <description>", where <package> is
whichever of langchain, community, core, experimental, etc. is being
modified.
Replace this entire comment with:
- **Description:** a description of the change,
- **Issue:** the issue # it fixes if applicable,
- **Dependencies:** any dependencies required for this change,
- **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!
Please make sure your PR is passing linting and testing before
submitting. Run `make format`, `make lint` and `make test` from the root
of the package you've modified to check this locally.
See contribution guidelines for more information on how to write/run
tests, lint, etc: https://python.langchain.com/docs/contributing/
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. It lives in
`docs/docs/integrations` directory.
If no one reviews your PR within a few days, please @-mention one of
@baskaryan, @eyurtsev, @hwchase17.
-->
Major changes:
- Rename `wasm_chat.py` to `llama_edge.py`
- Rename the `WasmChatService` class to `ChatService`
- Implement the `stream` interface for `ChatService`
- Add `test_chat_wasm_service_streaming` in the integration test
- Update `llama_edge.ipynb`
---------
Signed-off-by: Xin Liu <sam@secondstate.io>
<!-- Thank you for contributing to LangChain!
Please title your PR "<package>: <description>", where <package> is
whichever of langchain, community, core, experimental, etc. is being
modified.
Replace this entire comment with:
- **Description:** a description of the change,
- **Issue:** the issue # it fixes if applicable,
- **Dependencies:** any dependencies required for this change,
- **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!
Please make sure your PR is passing linting and testing before
submitting. Run `make format`, `make lint` and `make test` from the root
of the package you've modified to check this locally.
See contribution guidelines for more information on how to write/run
tests, lint, etc: https://python.langchain.com/docs/contributing/
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. It lives in
`docs/docs/integrations` directory.
If no one reviews your PR within a few days, please @-mention one of
@baskaryan, @eyurtsev, @hwchase17.
-->
See preview :
https://langchain-git-fork-cbornet-astra-loader-doc-langchain.vercel.app/docs/integrations/document_loaders/astradb
Todo
- [x] copy over integration tests
- [x] update docs with new instructions in #15513
- [x] add linear ticket to bump core -> community, community->langchain,
and core->openai deps
- [ ] (optional): add `pip install langchain-openai` command to each
notebook using it
- [x] Update docstrings to not need `openai` install
- [x] Add serialization
- [x] deprecate old models
Contributor steps:
- [x] Add secret names to manual integrations workflow in
.github/workflows/_integration_test.yml
- [x] Add secrets to release workflow (for pre-release testing) in
.github/workflows/_release.yml
Maintainer steps (Contributors should not do these):
- [x] set up pypi and test pypi projects
- [x] add credential secrets to Github Actions
- [ ] add package to conda-forge
Functional changes to existing classes:
- now relies on openai client v1 (1.6.1) via concrete dep in
langchain-openai package
Codebase organization
- some function calling stuff moved to
`langchain_core.utils.function_calling` in order to be used in both
community and langchain-openai
<!-- Thank you for contributing to LangChain!
Replace this entire 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!
Please make sure your PR is passing linting and testing before
submitting. Run `make format`, `make lint` and `make test` to check this
locally.
See contribution guidelines for more information on how to write/run
tests, lint, etc:
https://github.com/langchain-ai/langchain/blob/master/.github/CONTRIBUTING.md
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. It lives in `docs/extras`
directory.
If no one reviews your PR within a few days, please @-mention one of
@baskaryan, @eyurtsev, @hwchase17.
-->
Adds `WasmChat` integration. `WasmChat` runs GGUF models locally or via
chat service in lightweight and secure WebAssembly containers. In this
PR, `WasmChatService` is introduced as the first step of the
integration. `WasmChatService` is driven by
[llama-api-server](https://github.com/second-state/llama-utils) and
[WasmEdge Runtime](https://wasmedge.org/).
---------
Signed-off-by: Xin Liu <sam@secondstate.io>
<!-- Thank you for contributing to LangChain!
Replace this entire 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!
Please make sure your PR is passing linting and testing before
submitting. Run `make format`, `make lint` and `make test` to check this
locally.
See contribution guidelines for more information on how to write/run
tests, lint, etc:
https://github.com/langchain-ai/langchain/blob/master/.github/CONTRIBUTING.md
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. It lives in `docs/extras`
directory.
If no one reviews your PR within a few days, please @-mention one of
@baskaryan, @eyurtsev, @hwchase17.
-->
- **Description:**
- This PR introduces a significant enhancement to the LangChain project
by integrating a new chat model powered by the third-generation base
large model, ChatGLM3, via the zhipuai API.
- This advanced model supports functionalities like function calls, code
interpretation, and intelligent Agent capabilities.
- The additions include the chat model itself, comprehensive
documentation in the form of Python notebook docs, and thorough testing
with both unit and integrated tests.
- **Dependencies:** This update relies on the ZhipuAI package as a key
dependency.
- **Twitter handle:** If this PR receives spotlight attention, we would
be honored to receive a mention for our integration of the advanced
ChatGLM3 model via the ZhipuAI API. Kindly tag us at @kaiwu.
To ensure quality and standards, we have performed extensive linting and
testing. Commands such as make format, make lint, and make test have
been run from the root of the modified package to ensure compliance with
LangChain's coding standards.
TO DO: Continue refining and enhancing both the unit tests and
integrated tests.
---------
Co-authored-by: jing <jingguo92@gmail.com>
Co-authored-by: hyy1987 <779003812@qq.com>
Co-authored-by: jianchuanqi <qijianchuan@hotmail.com>
Co-authored-by: lirq <whuclarence@gmail.com>
Co-authored-by: whucalrence <81530213+whucalrence@users.noreply.github.com>
Co-authored-by: Jing Guo <48378126+JaneCrystall@users.noreply.github.com>
Description: Volcano Ark is an enterprise-grade large-model service
platform for developers, providing a full range of functions and
services such as model training, inference, evaluation, fine-tuning. You
can visit its homepage at https://www.volcengine.com/docs/82379/1099455
for details. This change could help developers use the platform for
embedding.
Issue: None
Dependencies: volcengine
Tag maintainer: @baskaryan
Twitter handle: @hinnnnnnnnnnnns
---------
Co-authored-by: lujingxuansc <lujingxuansc@bytedance.com>
- **Description:** fixes and upgrades for the Tongyi LLM and ChatTongyi
Model
- Fixed typos; it should be `Tongyi`, not `OpenAI`.
- Fixed a bug in `stream_generate_with_retry`; it's a real stream
generator now.
- Fixed a bug in `validate_environment`; the `dashscope_api_key` should
be properly handled when set by environment variables or initialization
parameters.
- Changed the `dashscope` response to incremental output by setting the
parameter `incremental_output`, which eliminates the need for the
prefix-removal trick.
- Removed some unused parameters, like `n`, `prefix_messages`.
- Added `_stream` method.
- Added async methods support, such as `_astream`, `_agenerate`,
`_abatch`.
- **Dependencies:** No new dependencies.
- **Tag maintainer:** @hwchase17
> PS: Some may be confused about the terms `dashscope`, `tongyi`, and
`Qwen`:
> - `dashscope`: A platform to deploy LLMs and provide APIs to invoke
the LLM.
> - `tongyi`: A brand name or overall term about Alibaba Cloud's LLM/AI.
> - `Qwen`: An LLM that is open-sourced and deployed in `dashscope`.
>
> We use the `dashscope` SDK to interact with the `tongyi`-`Qwen` LLM.
---------
Co-authored-by: Harrison Chase <hw.chase.17@gmail.com>
- **Description:** `tools.gmail.send_message` implements a
`SendMessageSchema` that is not used anywhere. `GmailSendMessage` also
does not have an `args_schema` attribute (this led to issues when
invoking the tool with an OpenAI functions agent, at least for me). Here
we add the missing attribute and a minimal test for the tool.
- **Issue:** N/A
- **Dependencies:** N/A
- **Twitter handle:** N/A
---------
Co-authored-by: Chester Curme <chestercurme@microsoft.com>
- **Description:** This PR fixes test failures on Windows caused by path
handling differences and unescaped special characters in regex. The
failing tests are:
```
FAILED tests/unit_tests/storage/test_filesystem.py::test_yield_keys - AssertionError: assert ['key1', 'subdir\\key2'] == ['key1', 'subdir/key2']
FAILED tests/unit_tests/test_imports.py::test_importable_all - ModuleNotFoundError: No module named 'langchain_community.langchain_community\\adapters'
FAILED tests/unit_tests/tools/file_management/test_utils.py::test_get_validated_relative_path_errs_on_absolute - re.error: incomplete escape \U at position 53
FAILED tests/unit_tests/tools/file_management/test_utils.py::test_get_validated_relative_path_errs_on_parent_dir - re.error: incomplete escape \U at position 69
FAILED tests/unit_tests/tools/file_management/test_utils.py::test_get_validated_relative_path_errs_for_symlink_outside_root - re.error: incomplete escape \U at position 64
```
- **Issue:** fixes
https://github.com/langchain-ai/langchain/issues/11775 (partially)
- **Dependencies:** none
Builds on #14040 with community refactor merged and notebook updated.
Note that with this refactor, models will be imported from
`langchain_community.chat_models.huggingface` rather than the main
`langchain` repo.
---------
Signed-off-by: harupy <17039389+harupy@users.noreply.github.com>
Signed-off-by: ugm2 <unaigaraymaestre@gmail.com>
Signed-off-by: Yuchen Liang <yuchenl3@andrew.cmu.edu>
Co-authored-by: Andrew Reed <andrew.reed.r@gmail.com>
Co-authored-by: Andrew Reed <areed1242@gmail.com>
Co-authored-by: A-Roucher <aymeric.roucher@gmail.com>
Co-authored-by: Aymeric Roucher <69208727+A-Roucher@users.noreply.github.com>
Description: Adding Summarization to Vectara, to reflect it provides not
only vector-store type functionality but also can return a summary.
Also added:
MMR capability (in the Vectara platform side)
Updated templates
Updated documentation and IPYNB examples
Tag maintainer: @baskaryan
Twitter handle: @ofermend
---------
Co-authored-by: Ofer Mendelevitch <ofermend@gmail.com>
## Description
This PR intends to add support for Qdrant's new [sparse vector
retrieval](https://qdrant.tech/articles/sparse-vectors/) by introducing
a new retriever class, `QdrantSparseVectorRetriever`.
Necessary usage docs and integration tests have been added for the
retriever.
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
This PR adds support for PygmalionAI's [Aphrodite
Engine](https://github.com/PygmalionAI/aphrodite-engine), based on
vLLM's attention mechanism. At the moment, this PR does not include
support for the API servers, but they will be added in a later PR.
The only dependency as of now is `aphrodite-engine==0.4.2`. We pin the
version to prevent breakage due to changes in the aphrodite-engine
library.
---------
Co-authored-by: Bagatur <baskaryan@gmail.com>
- **Description:** Going forward, we have a own API `pip install
gradientai`. Therefore gradually removing the self-build packages in
llamaindex, haystack and langchain.
- **Issue:** None.
- **Dependencies:** `pip install gradientai`
- **Tag maintainer:** @michaelfeil
This is technically a breaking change because it'll switch out default
models from `text-davinci-003` to `gpt-3.5-turbo-instruct`, but OpenAI
is shutting off those endpoints on 1/4 anyways.
Feels less disruptive to switch out the default instead.
- **Description:** VertexAIEmbeddings performance improvements
- **Twitter handle:** @vladkol
## Improvements
- Dynamic batch size, starting from 250, lowering down to 5. Batch size
varies across regions.
Some regions support larger batches, and it significantly improves
performance.
When running large batches of texts in `us-central1`, performance gain
can be up to 3.5x.
The dynamic batching also makes sure every batch is below 20K token
limit.
- New model parameter `embeddings_type` that translates to `task_type`
parameter of the API. Newer model versions support [different embeddings
task
types](https://cloud.google.com/vertex-ai/docs/generative-ai/embeddings/get-text-embeddings#api_changes_to_models_released_on_or_after_august_2023).
