LangChain Basics Tutorial #2 Tools and Chains

Поділитися
Вставка
  • Опубліковано 27 лип 2024
  • Colab Code Notebook - rli.to/5eoj4
    In this video, we jump into the Tools and Chains in LangChain. We look at what they are and specifically what tools like PAL, LLMChains, API tools do and how they can be used to get started developing applications in LangChain. If there are particular topics or parts of LangChain you would like me to cover then leave a note in the comments.
    My Links:
    Twitter - / sam_witteveen
    Linkedin - / samwitteveen
    Github:
    github.com/samwit/langchain-t...
    github.com/samwit/llm-tutorials
    00:00 Intro
    00:30 What are Tools in LangChain?
    02:45 3 Categories of Chains
    04:11 Tools - Utility Chains
    05:45 - Code - Basic Chains
    10:30 - Chaining Chains together
    12:26 - PAL Math Chain
    15:02 - API Tool Chains
    17:23 - Conclusion
    #LangChain #BuildingAppswithLLMs
  • Наука та технологія

КОМЕНТАРІ • 63

  • @fredrik-ekelund
    @fredrik-ekelund Рік тому +6

    Hey Sam, I really enjoy your content. Your explanations are easy to follow and comprehend, yet not overly simplistic. Thanks!

  • @christophelombart
    @christophelombart 11 місяців тому +2

    Excellent ! This training should not be free. I'm really a beginner in ai dev and everything is very clear. Thanks a lot

  • @JasonH-ug4rn
    @JasonH-ug4rn 2 місяці тому

    Truly great tutorials. Detailed and informative but not overkill. Love it.

  • @markomilenkovic2714
    @markomilenkovic2714 10 місяців тому

    Thank you Sam, this is super useful, no noise or bs, theory plus practical approach, love your presentation style, please don't stop making these.

  • @mysticaltech
    @mysticaltech Рік тому +6

    Sam, your content is unlike anything out there, you are at a whole other level, this is exactly what I needed. Thanks and keep up the good work mate!

    • @samwitteveenai
      @samwitteveenai  Рік тому

      Thanks, much appreciated!

    • @adumont
      @adumont Рік тому +1

      +1. True. While the majority of the videos just stay on the surface of the subject and simply run dull examples, Sam's channel does beyond that and shows in depth how to do and use this stuff with real life examples. Love it! ❤

  • @waqasobeidy8318
    @waqasobeidy8318 Рік тому +1

    So glad I found your content Sam. Very insightful indeed.

  • @ratral
    @ratral Рік тому +1

    @Sam, thank you very much for your content and for the time you have invested in it. It has helped me clarify the concepts and understand the potential of LangChain in combination with the different tools. Raúl. 👍

  • @rajivraghu9857
    @rajivraghu9857 Рік тому +1

    Excellent Sam.. I went through several videos. Your langchain series was ❤‍🔥❤‍🔥

  • @koningsbruggen
    @koningsbruggen Рік тому

    I love these series, it makes things so clear.

  • @pypypy4228
    @pypypy4228 Рік тому

    Thank you for your tutorials! I start from the first video and going on. Well structured, no water, to the point.

  • @manelaloui7069
    @manelaloui7069 Рік тому

    Another great video, I started today this playlist and I'm excited to finish it as soon as possible

  • @hichamzmarrou3762
    @hichamzmarrou3762 Рік тому

    Thanks Sam for this amazing content.

  • @jokosalsa
    @jokosalsa Рік тому

    Great tutorials, well explained. Thank you

  • @____2080_____
    @____2080_____ Рік тому +1

    **STILL SCREAMING**
    Seriously. Awesome content

  • @jzam5426
    @jzam5426 10 місяців тому

    Subscribed!! Love the content, thanks for making it didactic and always providing context

  • @blackswann9555
    @blackswann9555 3 місяці тому

    Great content. Finally it makes sense! ❤

  • @senecalouck2335
    @senecalouck2335 Рік тому +4

    Really digging your content Sam. It's super relevant to what I've been exploring on my spare tie and I expect your channel will do well if you can keep it up!

  • @jerichokatende8151
    @jerichokatende8151 28 днів тому +1

    thanks for this amazing content

    • @samwitteveenai
      @samwitteveenai  28 днів тому +1

      glad it was useful. it very old at this point I am working on making a new set of LangChain vids with new code etc

  • @MrSCAAT
    @MrSCAAT Рік тому

    Great work SIr

  • @adumont
    @adumont Рік тому

    Hi Sam, I've recently discovered your channel. Ive sub'd and 🛎 'd. I like that you don't just run the examples in the doc, but go a step deeper, and explain deeper in the details hmof how things are working. Thanks.

  • @caiyu538
    @caiyu538 9 місяців тому

    Great lectures.

  • @Fowthy
    @Fowthy Місяць тому

    Thanks a lot mate! That helps a lot

  • @frode0216
    @frode0216 5 місяців тому

    Appreciate for your video! :)

  • @foysalmamun5106
    @foysalmamun5106 Рік тому

    Thank you lot

  • @adityahpatel
    @adityahpatel 11 місяців тому +1

    Can this only do with open_meteo_docs. How about other APIs like yahoo finance API or SEC Edgar API...what is the best approach to bring in data from there?

  • @diego197103
    @diego197103 Рік тому

    Great!

  • @kyoungd
    @kyoungd Рік тому +2

    Unfortunately, they took out the 'code-davinci-002'. No more PALChain, I guess. It is still a great tutorial. Thanks.

    • @samwitteveenai
      @samwitteveenai  Рік тому

      let me look at doing an updated version next week.

    • @kyoungd
      @kyoungd Рік тому

      @@samwitteveenai Thanks, Sam. I really appreciate it.

  • @pi5549
    @pi5549 Рік тому +1

    Maybe good idea to make a playlist and link it in the description of each video.

    • @samwitteveenai
      @samwitteveenai  Рік тому

      great idea! will add to description. ua-cam.com/video/J_0qvRt4LNk/v-deo.html

  • @remyar2000
    @remyar2000 5 місяців тому

    Hi Sam, inorder to use APIChains, how can we generate the docs for an endpoint

  • @user-mi1tb4lo9k
    @user-mi1tb4lo9k 5 місяців тому

    chain.run is depreciated and should use chain.invoke, but chain.invoke will return some different answer i.e. it will return the exact content of the article and not what have been askked in the templete. any solution to this??

  • @syedmujeebh6661
    @syedmujeebh6661 9 місяців тому

    Hi,
    Does anyone face issue with RetrievalSourceQAChains where getting sources as empty for the answers taken from vector store

  • @gwenstacy4059
    @gwenstacy4059 4 місяці тому

    In which cases we don't use chains can you give me some scenarios or examples

  • @kwadwoagyapon-ntra2176
    @kwadwoagyapon-ntra2176 Рік тому

    Thanks for the great tutorial!
    I'm trying to do this strictly with open source models from HuggingFace. Could you suggest one to use in place of "code-davinci-002" for the PAL math chain?

    • @samwitteveenai
      @samwitteveenai  Рік тому

      On the testing list.

    • @kenchang3456
      @kenchang3456 Рік тому

      @@samwitteveenai Hi Sam. Now that code-davinci-002 is no longer accessible from openai, are you aware of another model to use in its place that will work with your Tools and Chains.ipynb? Thanks.

  • @rishavranaut7651
    @rishavranaut7651 Місяць тому

    please make a update playlist with good project

  • @GP_12581
    @GP_12581 Місяць тому

    This tutorial has great concepts, but with the recent changes to the API it is tough to follow the code examples.

  • @aliattieh9216
    @aliattieh9216 Рік тому

    Hi sam, great tutorial but does the PAL chain got removed cause importing it is raising an error and I also could find its definition in the source code

    • @samwitteveenai
      @samwitteveenai  11 місяців тому +1

      They have just released a new format called LangChain Expression Language, so I will gradually remake a lot of these. The LLM I used in that video doesn't exist anymore you are best to use GPT-4 or PaLM2 for this now.

  • @matthieudegroot3581
    @matthieudegroot3581 Рік тому

    Awesome content! Question: when would you rather use "simple" chains instead of an agent with multiple tools?

    • @samwitteveenai
      @samwitteveenai  Рік тому +1

      Good question. Agents are for when you want the ability to dynamically change based on the input from the user. Simple chains are more for where you want a predictable fixed path as to what is done with the LLM eg. Summarization etc.

    • @matthieudegroot3581
      @matthieudegroot3581 Рік тому +1

      @@samwitteveenai thanks! Ill be checking your channel regularly 👍🏼

  • @KumR
    @KumR 5 місяців тому

    one question sir. can we connect lc to one llm get a response & then pass or use that response in another llm?

    • @samwitteveenai
      @samwitteveenai  5 місяців тому

      yes you can totally do this just by setting up 2 models

  • @rafaelprudencioleite7291
    @rafaelprudencioleite7291 Рік тому

    Can you share the github page that shows in 4:30 in the video?

  • @codegate615
    @codegate615 Рік тому

    shouldn't you be using gpt-3-turbo?

    • @samwitteveenai
      @samwitteveenai  Рік тому +1

      This was record before the gpt-3-turbo was out, I do have another video showing how to use the turbo models.

  • @KhanhLe-pu5wx
    @KhanhLe-pu5wx 7 місяців тому

    i cant import PALChain . Now is Dec 27