quackamollie-langchain-model-manager 0.1

Model manager compatible with Langchain models for Quackamollie Telegram chat bot

Name:

Quackamollie Langchain Model Manager

Package name:

quackamollie-langchain-model-manager

Description:

Model manager compatible with Langchain models for Quackamollie Telegram chat bot

Version:

0.1

Main page:

https://gitlab.com/forge_of_absurd_ducks/quackamollie/lib/model_managers/quackamollie_langchain_model_manager

PyPI package:

https://pypi.org/project/quackamollie-langchain-model-manager/

Docker Image:

registry.gitlab.com/forge_of_absurd_ducks/quackamollie/lib/model_managers/quackamollie_langchain_model_manager:0.1

Documentation:

https://langchain-model-manager-forge-of-absurd-ducks-qu-910c28e8f82e0a.gitlab.io/

Build Status:

Master:

Dev:

---

Project description

Quackamollie is a Telegram chat bot developed in Python using the library aiogram to serve LLM models running locally using Ollama.

This package is a model manager exposing Langchain models for the Quackamollie project. It contains:

• a model manager LangchainQuackamollieModelManager class implementing abstract functions of MetaQuackamollieModelManager

Learn more about Quackamollie on the project main page.

Requirements

Virtual environment

• Setup a virtual environment in python 3.10

make venv
# or
python3 -m venv venv

• Activate the environment

source venv/bin/activate

• If you want to deactivate the environment

deactivate

Tests

Tests requirements

• Install test requirements

make devtools
# or
pip install tox

Run pytest

• Run the tests

tox

Run lint

• Run the lintage

tox -e lint

Documentation

• To auto-generate the documentation configuration

tox -e gendocs

• To generate the documentation in Html

tox -e docs

• An automatically generated version of this project documentation can be found here

  • N.B.: This automatically generated documentation of the Quackamollie core project is still laking a lot of improvements. Sorry for the inconvenience.

Install

• Install the application from sources

make install
# or
pip install .

• Or install it from distribution

pip install dist/quackamollie-langchain-model-manager-0.1.tar.gz

• Or install it from wheel

pip install dist/quackamollie-langchain-model-manager-0.1.whl

• Or install it from PyPi repository

pip install quackamollie-langchain-model-manager  # latest
# or
pip install "quackamollie-langchain-model-manager==0.1"

Docker

• To build the application docker

docker build --network=host -t quackamollie_langchain_model_manager:0.1 .

• The official Docker image of this project is available at: registry.gitlab.com/forge_of_absurd_ducks/quackamollie/lib/model_managers/quackamollie_langchain_model_manager

• You can pull the image of the current release:

docker pull registry.gitlab.com/forge_of_absurd_ducks/quackamollie/lib/model_managers/quackamollie_langchain_model_manager:latest  # or dev
# or
docker pull registry.gitlab.com/forge_of_absurd_ducks/quackamollie/lib/model_managers/quackamollie_langchain_model_manager:0.1

Running the model manager

quackamollie-langchain-model-manager package is automatically discovered, through entrypoints, by the command tool line named quackamollie. Therefore, once installed, you should automatically see models managed by this model manager in Telegram /settings command.

You can install models for this model manager by simply pulling them using the ollama command:

ollama pull llama3

For details on how to run the Quackamollie project, please refer to the Quackamollie’s project main page.

Authors

• QuacktorAI - Initial work - quacktorai

Contributing

If you want to report a bug or ask for a new feature of quackamollie-langchain-model-manager, please open an issue in the Gitlab ticket management section of this project. Please, first ensure that your issue is not redundant with already open issues.

If you want to contribute code to this project, please open first an issue and then a merge request with commit names referencing this issues. Note that only fast-forward merge requests are accepted.

For more details on the general contributing mindset of this project, please refer to CONTRIBUTING.md.

Credits

Section in writing, sorry for the inconvenience.