DSPy + Weaviate for the Next Generation of LLM Apps

Поділитися
Вставка
  • Опубліковано 26 лис 2024

КОМЕНТАРІ • 37

  • @ecardenas300
    @ecardenas300 9 місяців тому +14

    Thanks for watching! Feel free to let me know if you have any questions. 🥳

    • @mrchongnoi
      @mrchongnoi 5 місяців тому +1

      you links need to be updated.

  • @connor-shorten
    @connor-shorten 9 місяців тому +6

    Amazing! This is an incredible example of how much creativity DSPy unlocks for LLM developers! Questions to blog posts, wow!!

    • @ecardenas300
      @ecardenas300 9 місяців тому +1

      Thank you, Connor! The possibilities are endless 🚀

  • @RajuAngani
    @RajuAngani 8 місяців тому +2

    Going to explore DSPy. Thank you for the example!

  • @GeorgeFoxRules
    @GeorgeFoxRules 9 місяців тому +2

    You produce done of the best content on YT for AI! Great job and keep going. You are a rockstar.

    • @ecardenas300
      @ecardenas300 9 місяців тому

      Thanks for the support, George!

  • @ABK6969
    @ABK6969 9 місяців тому +3

    Great walkthrough, looking forward to trying this.

  • @joser100
    @joser100 9 місяців тому +6

    Amazing tutorial Erika, very clear...

    • @ecardenas300
      @ecardenas300 9 місяців тому +2

      Thanks so much! Happy you enjoyed it 🙂

  • @dshorten1766
    @dshorten1766 9 місяців тому +4

    Great video, Erika!

  • @DonBranson1
    @DonBranson1 9 місяців тому +1

    great video, and even better follow up support from Erika and the Weaviate team to resolve an issue I was having the code example.
    Thank you!

    • @ecardenas300
      @ecardenas300 9 місяців тому +2

      Hey Don,
      Just responded to you on Slack! Hope we're able to resolve this quickly
      🙂

  • @lea_paris
    @lea_paris 8 місяців тому +1

    Great video, very clear explanations, and the example is great. Thanks Erika!

    • @ecardenas300
      @ecardenas300 8 місяців тому

      Thank you! Glad you enjoyed the video!

  • @Liberty_boy
    @Liberty_boy 9 місяців тому +6

    Super interesting!

  • @Cdaprod
    @Cdaprod 9 місяців тому +3

    Great video!! I can’t wait to do this for myself!!! 🎉

    • @ecardenas300
      @ecardenas300 9 місяців тому

      Thanks so much, David! Happy building 🚀

  • @ivanarozamena
    @ivanarozamena 8 місяців тому +3

    Hi ! great content ! how you can add a filter where in your Weaviate database after to connect the retrieval ?

    • @ecardenas300
      @ecardenas300 8 місяців тому

      Thank you! 🙂
      Are you asking how you can create a Weaviate schema and connect to it? There is a `Weaviate-import` notebook in the demo URL. I can't share links in my response, sorry!

    • @ivanarozamena
      @ivanarozamena 8 місяців тому +1

      @@ecardenas300 thanks for the response !, but no we have already that my question is if in the retrieval function can we use like a where in sql to filter by one field in the schema, i dont know if i was clear enaugh :)

    • @ecardenas300
      @ecardenas300 8 місяців тому

      @@ivanarozamena Oh, yes! Weaviate supports filtering and can even couple multiple filters together. I can't share a link here, but you can find it by going to --> Weaviate --> Developers --> Weaviate docs --> How-to: Search --> Filters

  • @miguelovallevillamil4953
    @miguelovallevillamil4953 5 місяців тому +1

    Great!!!

  • @BrettDoffing
    @BrettDoffing 6 місяців тому

    Interesting. I may have missed it, but did you have training/test data before hand, or was this data generated by the pipeline (?) you just built out?

  • @joffreylemery6414
    @joffreylemery6414 8 місяців тому +1

    Hi ! Thanks for your video
    There is something i still do not understand with DSPy :
    If i want to create a programme similar like you, a bit different. Do i have to create a testing and eval dataset ? Do I have to do it for each application i have to do ? It makes the processus heavy...

    • @ecardenas300
      @ecardenas300 8 місяців тому

      Hi, thanks for your question!
      If you want to build a different DSPy program, you have to:
      1. Create your Weaviate cluster and upload data
      2. Connect to the Weaviate retriever model in DSPy
      3. Build out your program with a signature(s)

    • @joffreylemery6414
      @joffreylemery6414 8 місяців тому

      @@ecardenas300 hum I understand
      Then, having training data is always mandatory ?
      (Requires effort to have it sometimes)

  • @gabolema
    @gabolema 8 місяців тому +1

    thanks for sharing! at the end of the video we can see how LLMS are not good at math yet 😅

  • @ZaferCan
    @ZaferCan 5 місяців тому +1

    can we use dspy with weaviate-client latest versions ?

  • @iskrabesamrtna
    @iskrabesamrtna 7 місяців тому

    great application idea! I'd just add a small suggestion: please, use dark background, eyes bleed of this much brightness -.-

    • @ecardenas300
      @ecardenas300 6 місяців тому

      Thank you for the feedback! I'll keep that in mind next time. 🙂