Notebooks with Langchain for a local model and a remote model with custom API

redhat-et / foundation-models-for-documentation

Improve ROSA customer experience (and customer retention) by leveraging foundation models to do “gpt-chat” style search of Red Hat customer documentation assets.

Other

24 stars 11 forks source link

Notebooks with Langchain for a local model and a remote model with custom API #31

Closed codificat closed 1 year ago

codificat commented 1 year ago

This is work in progress to use LangChain's HuggingFacePipeline to use one of the open models from Hugging Face with LangChain.

There are 2 notebooks here:

langchain-pipeline.ipynb does not attempt to do any adaptation with ROSA docs, just testing the models directly
langchain-pipeline-rosa.ipynb is the equivalent of the langchain with OpenAI notebook, using a vector store to index ROSA docs and a chain to generate prompts for QA

review-notebook-app[bot] commented 1 year ago

Check out this pull request on

See visual diffs & provide feedback on Jupyter Notebooks.

Powered by ReviewNB

codificat commented 1 year ago

I added another WIP notebook that wraps the text generation web UI API for LangChain to use.

I am trying to see if I can get better results (the ones in the commit I just pushed come from a bloomz-3b model)

codificat commented 1 year ago

PR updated and rebased.

I ended up removing the "basic" notebook that only showed the HuggingFacePipeline usage, as I believe this is already properly covered in the langchain-pipeline-rosa notebook and there's no need for a separate one.

The API notebook has been updated to match the new API in the text generation app.