Adding Using the `Text Generation Inference (TGI)` with Llama Models

[sergiopaniego]

Adding a TGI recipe that includes examples of how to serve and consume the 🦙 Llama models using different methods available in the HF ecosystem.

fix #45

@ariG23498

[sergiopaniego]

I'm not sure if the file location is correct, and we might need to link the recipe in the README or somewhere else, so that may need updating as well. 😄

[ariG23498]

@sergiopaniego I really like the notebook! :fire:

Let's have the current notebook broken down into two parts:

1. A shell script to run TGI with llama model using docker
2. A separate notebook that holds the consuming the endpoint recipe? I really like the chat with image bit! (awesome work)

CC: @Vaibhavs10 what do you think?

[sergiopaniego]

Thanks a lot for the feedback, @ariG23498!

I think it makes sense to split the notebook into two separate parts, as you suggested. Should I start working on it now, or do we need a double-check from @Vaibhavs10? 😄

[ariG23498]

Let' wait for @Vaibhavs10 to give it a look? I don't want you to work twice 😅

[sergiopaniego]

Thanks a lot for the in-depth review!

I'll update the PR accordingly 👨‍💻

[sergiopaniego]

I've updated the notebook covering all the possible improvements mentioned @pcuenca. 🤗

Open to any further suggestions or enhancements!

cc @ariG23498 @Vaibhavs10

[sergiopaniego]

Thank you for the review, @Vaibhavs10! 😊

I've updated the notebook with a new sentence: These models, along with many others, are efficiently deployed via TGI on the serverless API.

Additionally, I’ve renamed both the folder and the file: llama_tgi/tgi_recipe.ipynb -> llama_tgi_api_inference/tgi_api_inference_recipe.ipynb

I hope this aligns with the expectations!

[sergiopaniego]

Added @Vaibhavs10 😄! Thanks a lot for all the comments and suggestions!!

[Vaibhavs10]

Perfect from myside to merge - I'd let @pcuenca take the final pass and merge! 🤗

[sergiopaniego]

Thanks for the review and suggestions @pcuenca! I've improved the notebook based on them 😄