Fine tune decoder-only transformers in seq2seq manner

[YerongLi]

Feature request

#1464
This posts discuss fine-tuning GPT2,
With GPT2/LLaMA, by default, we need to input the [prompt label] the whole sentence model([prompt label]) in fine-tuning and caculate the CrossEntropy on the label part, and the model output the model().logits.

Are there any ways to input the prompt only and do the fine-tuning in the seq2seq manner? (model(prompt)), this way we minimize the loss of log p(y|x).

Get the feature of model(prompt) rather than model([prompt label]) is the whole point.

Motivation

seq2seq equivalence fine-tuning workflow for decoder-only transformers.

Your contribution

I could submit PR with discussion as a guidance.

[ArthurZucker]

Hey 🤗 thanks for opening an issue! We try to keep the github issues for bugs/feature requests.
Could you ask your question on the forum instead? I'm sure the community will be of help!

Thanks!

[github-actions]

This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.

Please note that issues that do not follow the contributing guidelines are likely to be ignored.