I have watched two of your videos and they were both impressive. Good job on putting the app together and also clearly explaining it. Thanks. Liked and subscribed.
I appreciate your video, but I need clarification regarding the process when running the code. Does it involve downloading the model onto your machine or codespace container, or does it utilize the Huggingface key to make an API call to the hosted model on the Huggingface hub? In other words, where is the model actually hosted?
It's using the HF API call to the hosted mode on the hub. One could also download the modela and run inference on the local version as well using langchain.
@@menloparklab That's cool! So we're using models hosted by HF. If you have a moment, could you make a video showing how to put the model on AWS Sagemaker or a free service and run it?
Actually, there are some well documented ways of deploying to services like BananaML or Replicate.com, which might be a bit easier. Will test it out for sagemaker. Thanks.
@@menloparklab How can one download the models and run it for local inference? I do see that files are available on the model cards on HF (.bin files?) Is there some package that would allow me to use these local models/files?
I tried the exact same way you did in the video ,but encountered with an error as "ValueError: langchain_factory use_async parameter is required" do you have any solution for it ?
I'm skeptical it's 40B model, the biggest one, people complain it's very slow and demands very powerful hardware. I'm waiting when Orca will be available.
Hi. You are great. But i don’t understand i have you questions😓. I had install the chatbot but he responds some “bad” questions , we need to train the program ? And second Can i install this on my page? Thank you very much❤
@@menloparklab I'd also be interested in learning about how to use something like this for a dataset file. Regardless, subscribed! Thanks for putting up such great content.
Hi It does not perform well on follow up questions(related to last question), can you share an implementation where we can include chat history, probably by using memory from langchain. Thanks In Advance.
Why does when you create the codespace below there is written Codespace usage for this repository is paid for by misbahsy? Is that a public Codespace you created and you are paying it for other users to use? Explain what's happening.
![[ ออกกอง ] วัยหนุ่ม 2544 โลกหลังลูกกรง..ทัวร์กองถ่ายหนังใน "คุกของจริง" | JUSTดูIT.](http://i.ytimg.com/vi/vB--aOkZCpw/mqdefault.jpg)
Grateful to have people like you on TH-cam; spreading the valuable information with the developer community :)
Thanks 🙏
You are our guide in the dark with your tutorials. We're thankful that you offer them for free, so we don't need to buy courses. :) Thanks!
I have watched two of your videos and they were both impressive. Good job on putting the app together and also clearly explaining it. Thanks. Liked and subscribed.
Thanks a lot 🙏
@@menloparklab hi i want to add like chatgpt when this model generate text after that user will ask summerize it so how i do that?
Thank You , All the best keep good thing up "Bon Courage et Merci"
Dude exactly what I was looking for.
Please do another video showing langchain doing cooler stuff like iterating over output from the LLM to create new prompts from templates.
Course website for pre-launch pricing: builtbyou.com/
Project requests: forms.office.com/r/NU5Lf5PPjr
9:45 please do a vid on using the agents with chainlit
Hi, can I run this on a colab or kaggle? and can I use pdfs or other docs? to chat with my data ?
Hey, great content. Question: is it possible to connect chainlit with a bubble app?
Good question, you might be able to embed it as an HTML embed
running the chat gives me Error raised by inference API: Internal Server Error
I appreciate your video, but I need clarification regarding the process when running the code. Does it involve downloading the model onto your machine or codespace container, or does it utilize the Huggingface key to make an API call to the hosted model on the Huggingface hub? In other words, where is the model actually hosted?
It's using the HF API call to the hosted mode on the hub. One could also download the modela and run inference on the local version as well using langchain.
@@menloparklab That's cool! So we're using models hosted by HF. If you have a moment, could you make a video showing how to put the model on AWS Sagemaker or a free service and run it?
Actually, there are some well documented ways of deploying to services like BananaML or Replicate.com, which might be a bit easier. Will test it out for sagemaker. Thanks.
@@menloparklab Thank you! Appreciate it!
@@menloparklab How can one download the models and run it for local inference? I do see that files are available on the model cards on HF (.bin files?) Is there some package that would allow me to use these local models/files?
I tried the exact same way you did in the video ,but encountered with an error as
"ValueError: langchain_factory use_async parameter is required" do you have any solution for it ?
Same
@@tenhusho did you got any solution ?
To fix that, I changed the line:
@cl.langchain_factory
to
@cl.langchain_factory(use_async=False)
First I hear of Falcon
its beaten Llama model and tops the HF LLM leaderboard. Its completely open source and ready for commercial use as well.
I'm skeptical it's 40B model, the biggest one, people complain it's very slow and demands very powerful hardware.
I'm waiting when Orca will be available.
Hi. You are great.
But i don’t understand i have you questions😓.
I had install the chatbot but he responds some “bad” questions , we need to train the program ?
And second Can i install this on my page? Thank you very much❤
You can try updating the prompt template and see if that helps. Yes you can embed it on your page.
How can we use it for a custom dataset or pdf file? And can it be multilingual?
I haven't tested it yet with the custom dataset, but I am sure there are ways to run it.
@@menloparklab I'd also be interested in learning about how to use something like this for a dataset file.
Regardless, subscribed! Thanks for putting up such great content.
@@koganin9 Will give it a try
Hi It does not perform well on follow up questions(related to last question), can you share an implementation where we can include chat history, probably by using memory from langchain. Thanks In Advance.
Yes the current setup is without memory. Will try to make something with memory.
And please do show how to deploy it on cloud
Is it possible to run Falcon 40b instruct in single GPU without using cloud?
Haven’t tried it myself, will have to test it
i followed the steps you mentioned above but i still got the error
AttributeError: module 'chainlit' has no attribute 'langchain_factory'
Solution is discussed in one of the earlier comments.
Why does when you create the codespace below there is written Codespace usage for this repository is paid for by misbahsy? Is that a public Codespace you created and you are paying it for other users to use? Explain what's happening.
Not sure actually why it says that. Maybe they allow some amount for free computation, and paid after. But will have to check their docs.
Do we need to pay to get hugging face token?
No, it's free, they register who download their files with token and your ip, maybe for statistical reason.
Have you tried using Chainlit in Docker Container? Can we do it?
I am pretty positive it should work in a docker container as well, give it a try
I'm sorry but could you please let me know why do we need langchain here?
Not needed, but in case you want to use other LangChain components in your app.
Can it be used in flowise?
Will give it a try actually, should be able to run it with Flowise.
I don't understand what this chat I for?
No no no no no no!!!!!
I don’t even know what this is.
It's about the newly released AI Large Language model - Falcon
Also if you use Falcon and you make over 1 million dollars with it you have to pay the creators money... How is that open source?
@@timothymaggenti717 the license was revised to remove that clause recently