Building RAG with Command R+ from Cohere, DSPy, and Weaviate!

Sdílet
Vložit
  • čas přidán 5. 09. 2024

Komentáře • 11

  • @mysticaltech
    @mysticaltech Před 5 měsíci +5

    Awesome! Idea for a next video, would be great to have dspy, command-R+, replacing a support agent. answering questions, doing tool use. and optionally (cherry on the cake) dspy bootstrapping/optimizing/compiling based on historical human support threads.

    • @connor-shorten
      @connor-shorten  Před 5 měsíci +1

      This is such a brilliant idea! I love it, thank you so much! I've started sketching out MemGPT in DSPy -- creating the synthetic conversations for eval is indeed a fascinating one haha, some good nuggets in the Google Meena paper actually, hopefully will have something interesting to share soon!

  • @awsrftgyhuokjo
    @awsrftgyhuokjo Před 4 měsíci +1

    Hey conor can we connect to ollama local cohere commad r model to implement function calling or tool use using dspy

  • @sytsevanderschaaf5069
    @sytsevanderschaaf5069 Před 5 měsíci +1

    Exiting stuff, this multilingual approach to tokenization. Must be the Canadian roots of Cohere.

    • @connor-shorten
      @connor-shorten  Před 5 měsíci

      Haha love it! Yes, such a brilliant focus for Cohere -- I would say the most underrated aspect of AI.

  • @mohammad_mohammad.
    @mohammad_mohammad. Před 5 měsíci +2

    Hey Connor, can you make some videos on synthetic data generation using DSPy?

    • @connor-shorten
      @connor-shorten  Před 5 měsíci

      Ah this is a brilliant idea! Yes there are some really cool new built-in data generators! I'll take a look!

  • @420_gunna
    @420_gunna Před 5 měsíci +2

    lfg baby

  • @omarhaoudi6941
    @omarhaoudi6941 Před 4 měsíci

    How can I use custom data using the weaviate ?

  • @peterb3832
    @peterb3832 Před 4 měsíci

    *PromoSM*

  • @d4qatoa
    @d4qatoa Před 4 měsíci

    If I can give some feedback, too much discussion of regular RAG not enough of getting to the point of the benefits of this model. 128k context windows is what everyone has, if not better....I gave up waiting for the point.