5 ms·

Another option is to use llama-index and index the Obsidian vault. I use a gradio based web interface to query my Obsidian vault via GPT-3.5. It's pretty awes

Another option is to use llama-index and index the Obsidian vault. I use a gradio based web interface to query my Obsidian vault via GPT-3.5. It's pretty awesome.

atentaten 3y ago
Does your data stay local by doing this?
- LouisvilleGeek 3y ago
  Mostly but it does upload some of the vectorized data to insert into the prompt for context. When you do a query llama-index tries to discover content related to your prompt and injects it for context so its not entirely local.
  - majormajor 3y ago
    > Mostly but it does upload some of the vectorized data to insert into the prompt for context. When you do a query llama-index tries to discover content related to your prompt and injects it for context so its not entirely local. When you say "upload some of the vectorized data" do you mean in a numerical embedding form or that it will embed the original text from original similar-seeming notes directly into the prompt? I've only ever done the latter, is there a way to build denser prompts instead? I can't find examples on Google.
    - throwaway675309 3y ago
      The numerical vector representing the embedding is only useful for finding documents that are similar to your search query. Those documents are then injected into your prompt and sent to some kind of LLM completion system such as GPT. So yes, you will be sending chunks of your actual notes over the wire.
  - rahimnathwani 3y ago
    Are you calculating your embeddings locally, and using OpenAI's APIs only for text generation?
thatcherc 3y ago
Do you have code for this (or pointers to what you used to set this up)? I'd love to set up something similar!
- crucialfelix 3y ago
  This: https://github.com/hwchase17/chroma-langchain/blob/master/persistent-qa.ipynb https://github.com/hwchase17/chroma-langchain/blob/master/pe... I turned that into cli tool, but it isn't ready to release. Mine works like this: qwoo index . qwoo qa "What have I been up to?"

Another option is to use llama-index and index the Obsidian vault. I use a gradio based web interface to query my Obsidian vault via GPT-3.5. It's pretty awesome.

atentaten 3y ago

Does your data stay local by doing this?

LouisvilleGeek 3y ago
Mostly but it does upload some of the vectorized data to insert into the prompt for context. When you do a query llama-index tries to discover content related to your prompt and injects it for context so its not entirely local.
- majormajor 3y ago
  > Mostly but it does upload some of the vectorized data to insert into the prompt for context. When you do a query llama-index tries to discover content related to your prompt and injects it for context so its not entirely local. When you say "upload some of the vectorized data" do you mean in a numerical embedding form or that it will embed the original text from original similar-seeming notes directly into the prompt? I've only ever done the latter, is there a way to build denser prompts instead? I can't find examples on Google.
  - throwaway675309 3y ago
    The numerical vector representing the embedding is only useful for finding documents that are similar to your search query. Those documents are then injected into your prompt and sent to some kind of LLM completion system such as GPT. So yes, you will be sending chunks of your actual notes over the wire.
- rahimnathwani 3y ago
  Are you calculating your embeddings locally, and using OpenAI's APIs only for text generation?

thatcherc 3y ago

Do you have code for this (or pointers to what you used to set this up)? I'd love to set up something similar!

crucialfelix 3y ago
This: https://github.com/hwchase17/chroma-langchain/blob/master/persistent-qa.ipynb https://github.com/hwchase17/chroma-langchain/blob/master/pe... I turned that into cli tool, but it isn't ready to release. Mine works like this: qwoo index . qwoo qa "What have I been up to?"