Is there a way to use locally deployed AIs?

[yuanhan]

like a simple way to connect with gpt4all?
maybe share the data dir to gpt4all?

[apirrone]

Hi!

Memento uses LangChain to wrap LLM queries, so it should be quite straightforward to use different LLMs, including local ones !

You would need to change OpenAIEmbeddings and ChatOpenAI, below are the lines you would need to edit in the code :

• Memento/memento/timeline/chat.py

  Line 60 in 513a80c

  embedding_function=OpenAIEmbeddings(),

• Memento/memento/timeline/chat.py

  Line 86 in 513a80c

  ChatOpenAI(model_name="gpt-3.5-turbo-0301", temperature=0.8),

• Memento/memento/background.py

  Line 58 in 513a80c

  embedding_function=OpenAIEmbeddings(),