LLaMA2 with LangChain - Basics | LangChain TUTORIAL

LLaMA2 with LangChain - Basics | LangChain TUTORIAL

For more tutorials on using LLMs and building Agents, check out my Patreon:

My Links:

Github:

Timestamps:
00:00 Intro
04:47 Translation English to French
05:40 Summarization of an Article
07:08 Simple Chatbot
10:38 Chatbot using LLaMA-13B Model

[HarryZhao-uc]

This (and the earlier post on prompt tricks) are awesome and helpful. Thank you Sam for putting them up so quickly!

[Nick-tvpu]

Excellent video. Good to find a LLaMa video that actually has real technical content and not "just sign up for this cloud service"

[ernikitamalviya]

Thank you so much for taking the time to explain the LangChain concepts with the Llama model. It was very helpful and I appreciate your effort in making this information accessible :)

[matthewroman]

Sam, thanks for doing these videos. It really helps getting proof of concept work out much more quickly. This technology will change the world as you are aware. Thanks for your help and guidance ❤.
From one nerd to another.

[loicbaconnier]

Thanks for all Sam
Why didn't you put <s> at the beginning of the prompt ?

[IngmarStapel]

Hi Sam, thank you very much for that video. Very nice explained and I am looking forward to see more stuff like that kind of video. I really like that you are showing us how to use everything locally.

[guanjwcn]

Thanks very much for the video, Sam. This is very valuable. The 4-bit one should work on local PC. Looking forward to your next few videos.

[spadecake]

This video is extremly clear to understand, thanks for sharing. Also the token part is a bit of a shady move but hey if it works it's ok for now.

[julian-fricker]

Thanks for the great video, feels like Llama2 is a huge leap forward for open source models.

[anthonyshort]

Super helpful, thank you. Would like to see this same example using a local model.

[AmitKumar-ctdf]

Awesome videos Sam. Can you please make an instructional video for custom dataset including conversational chain?

[MikeEbrahimi]

Could you make a video about API portion? Great content.

[henkhbit]

Thanks for showing the 7 b model with langchain. The result are quite good. I read that there is already a 7b model with 32k. Would be nice to show if it can really summarize a big chunk of text of 20k or more....

[kai]

Thank you so much! Just what I've been waiting for! Would be great to have an example to run it locally with Cuda. I'm sitting here since two days to get it running...

[sameersaraf]

Super helpful! Would also love to see how we can construct agents with multiple tools with this model, I was getting some incoherent results when I tried.

[chanderbalaji]

HI Sam, thanks a ton for this. I tried integrating this with ConversationalRetrievalChain using your earlier video examples even though they were using RetrievalQA (using wrap_text_preserve_newlines and process_llm_response functions to get sources) but it didnt work. Could you tackle this in another video. Thanks

[mrchongnoi]

When do you think you will have a video on agents/tools ?

[Ryan-yjsd]

thanks, please make a video on how to fine tune and deploy?

[MaxXFalcon]

You tested so many models. What is the best and closest model to chatgpt in your opinion?

[UncleDao]

I run on T4 but the memory is near full. So the 7B can run some short prompts. I had an error OOM when running the text summarization (940 words).

"CUDA out of memory. Tried to allocate 254.00 MiB (GPU 0; 14.75 GiB total capacity; 13.27 GiB already allocated; 210.81 MiB free; 13.49 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF"