also remove some unused dependencies (fastapi) and unused test/lint/dev
dependencies (community, openai, textsplitters)
chromadb 0.5.4 introduced usage of `model_fields` which is pydantic v2
specific. also released in 0.5.5
The new `langchain-ollama` package seems pretty well implemented, but I
noticed the docs were still outdated so I decided to fix em up a bit.
- Llama3.1 was release on 23rd of July;
https://ai.meta.com/blog/meta-llama-3-1/
- Ollama supports tool calling since 25th of July;
https://ollama.com/blog/tool-support
- LangChain Ollama partner package was released 1st of august;
https://pypi.org/project/langchain-ollama/
**Problem**: Docs note langchain-community instead of langchain-ollama
**Solution**: Update docs to
https://python.langchain.com/v0.2/docs/integrations/chat/ollama/
**Problem**: OllamaFunctions is deprecated, as noted on
[Integrations](https://python.langchain.com/v0.2/docs/integrations/chat/ollama_functions/):
This was an experimental wrapper that attempts to bolt-on tool calling
support to models that do not natively support it. The [primary Ollama
integration](https://python.langchain.com/v0.2/docs/integrations/chat/ollama/) now
supports tool calling, and should be used instead.
**Solution**: Delete old notebook from repo, update the existing one
with @tool decorator + pydantic examples to the notebook
**Problem**: Llama3.1 was released while llama3-groq-tool-call fine-tune
Is noted in notebooks.
**Solution**: update docs + notebooks to llama3.1 (which has improved
tool calling support)
**Problem**: Install instructions are incomplete, there is no
information to download a model and/or run the Ollama server
**Solution**: Add simple instructions to start the ollama service and
pull model (for toolcalling)
---------
Co-authored-by: Chester Curme <chester.curme@gmail.com>
This will allow complextype metadata to be returned. the current
implementation throws error when dealing with nested metadata
Thank you for contributing to LangChain!
- [x] **PR title**: "package: description"
- Where "package" is whichever of langchain, community, core,
experimental, etc. is being modified. Use "docs: ..." for purely docs
changes, "templates: ..." for template changes, "infra: ..." for CI
changes.
- Example: "community: add foobar LLM"
- [ ] **PR message**: ***Delete this entire checklist*** and replace
with
- **Description:** a description of the change
- **Issue:** the issue # it fixes, if applicable
- **Dependencies:** any dependencies required for this change
- **Twitter handle:** if your PR gets announced, and you'd like a
mention, we'll gladly shout you out!
- [ ] **Add tests and docs**: 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.
- [ ] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, ccurme, vbarda, hwchase17.
---------
Co-authored-by: Chester Curme <chester.curme@gmail.com>
Here we allow standard tests to specify a value for `tool_choice` via a
`tool_choice_value` property, which defaults to None.
Chat models [available in
Together](https://docs.together.ai/docs/chat-models) have issues passing
standard tool calling tests:
- llama 3.1 models currently [appear to rely on user-side
parsing](https://docs.together.ai/docs/llama-3-function-calling) in
Together;
- Mixtral-8x7B and Mistral-7B (currently tested) consistently do not
call tools in some tests.
Specifying tool_choice also lets us remove an existing `xfail` and use a
smaller model in Groq tests.
- **Description:** The following
[line](fd546196ef/libs/community/langchain_community/document_loaders/parsers/audio.py (L117))
in `OpenAIWhisperParser` returns a text object for some odd reason
despite the official documentation saying it should return `Transcript`
Instance which should have the text attribute. But for the example given
in the issue and even when I tried running on my own, I was directly
getting the text. The small PR accounts for that.
- **Issue:** : #25218
I was able to replicate the error even without the GenericLoader as
shown below and the issue was with `OpenAIWhisperParser`
```python
parser = OpenAIWhisperParser(api_key="sk-fxxxxxxxxx",
response_format="srt",
temperature=0)
list(parser.lazy_parse(Blob.from_path('path_to_file.m4a')))
```
…he prompt in the create_stuff_documents_chain
Thank you for contributing to LangChain!
- [ ] **PR title**: "langchain:add document_variable_name in the
function _validate_prompt in create_stuff_documents_chain"
- [ ] **PR message**:
- **Description:** add document_variable_name in the function
_validate_prompt in create_stuff_documents_chain
- **Issue:** according to the description of
create_stuff_documents_chain function, the parameter
document_variable_name can be used to override the "context" in the
prompt, but in the function, _validate_prompt it still use DOCUMENTS_KEY
to check if it is a valid prompt, the value of DOCUMENTS_KEY is always
"context", so even through the user use document_variable_name to
override it, the code still tries to check if "context" is in the
prompt, and finally it reports error. so I use document_variable_name to
replace DOCUMENTS_KEY, the default value of document_variable_name is
"context" which is same as DOCUMENTS_KEY, but it can be override by
users.
- **Dependencies:** none
- **Twitter handle:** https://x.com/xjr199703
- [ ] **Add tests and docs**: none
- [ ] **Lint and test**: Run `make format`, `make lint` and `make test`
from the root of the package(s) you've modified. See contribution
guidelines for more: https://python.langchain.com/docs/contributing/
Additional guidelines:
- Make sure optional dependencies are imported within a function.
- Please do not add dependencies to pyproject.toml files (even optional
ones) unless they are required for unit tests.
- Most PRs should not touch more than one package.
- Changes should be backwards compatible.
- If you are adding something to community, do not re-import it in
langchain.
If no one reviews your PR within a few days, please @-mention one of
baskaryan, efriis, eyurtsev, ccurme, vbarda, hwchase17.
---------
Co-authored-by: Chester Curme <chester.curme@gmail.com>
fix: #25482
- **Description:**
Add a prompt to install beautifulsoup4 in places where `from
langchain_community.document_loaders import WebBaseLoader` is used.
- **Issue:** #25482
**Description:** This PR fixes an issue in the demo notebook of
Databricks Vector Search in "Work with Delta Sync Index" section.
**Issue:** N/A
**Dependencies:** N/A
---------
Co-authored-by: Chengzu Ou <chengzu.ou@databrick.com>
Co-authored-by: Erick Friis <erick@langchain.dev>
Check whether the API key is already in the environment
Update:
```python
import getpass
import os
os.environ["DATABRICKS_HOST"] = "https://your-workspace.cloud.databricks.com"
os.environ["DATABRICKS_TOKEN"] = getpass.getpass("Enter your Databricks access token: ")
```
To:
```python
import getpass
import os
os.environ["DATABRICKS_HOST"] = "https://your-workspace.cloud.databricks.com"
if "DATABRICKS_TOKEN" not in os.environ:
os.environ["DATABRICKS_TOKEN"] = getpass.getpass(
"Enter your Databricks access token: "
)
```
grit migration:
```
engine marzano(0.1)
language python
`os.environ[$Q] = getpass.getpass("$X")` as $CHECK where {
$CHECK <: ! within if_statement(),
$CHECK => `if $Q not in os.environ:\n $CHECK`
}
```
