Llama is ExecuTorch compatible

[guangy10]

Feature request

Enable Llama to "Export to ExecuTorch" workflow

Motivation

See details in #32253

Your contribution

Llama model enablement

[debrupf2946]

@guangy10 @amyeroberts
Hi, I am Debrup Paul, I would like to take up this issue, can this issue be assigned to me? I understood the workflow and will raise the pr in 2 weeks.

[ArthurZucker]

Hey! Actually this is already supported and can be closed! 🤗

[debrupf2946]

Hey! Actually this is already supported and can be closed! 🤗

Okay, do you have some beginner friendly task on which I can contribute?I actually want to begin contributing in Transformers.I am also a Google Summer of Code Contributor working on NLP and GenAI related problems.
linkedin: https://www.linkedin.com/in/debrup-paul-599158227/

[ArthurZucker]

I would recommend you the issues with Good second issue or Good difficult issue labels!

[guangy10]

Hey! Actually this is already supported and can be closed! 🤗

Yeah, I had

@guangy10 @amyeroberts Hi, I am Debrup Paul, I would like to take up this issue, can this issue be assigned to me? I understood the workflow and will raise the pr in 2 weeks.

Hey @debrupf2946, thanks for reaching out and being interested in llama for ExecuTorch. As @ArthurZucker mentioned, we actually have been working on this model already for a while, so it should just work out-of-the-box. I have a local commit that verify the newly released 1B and 3B model are working e2e. Will publish the PR soon to protect this model with tests.

Feel free to try a different model that you are interested in. Currently we have a list of candidates linked to the umbrella task #32253.