Yoach Lacombe

Hi @sanchit-gandhi , I think it's finally time for the final review! You might want to check the refactoring of `generate_text_semantic`, `generate_coarse`, `generate_fine`, but otherwise, sounds good!

Hi @amyeroberts, the PR is ready for review ! I'd be delighted to get your feedback on this when you have a chance. Let me know if I can help...

Hi @amyeroberts, as demanded, I resolved the merge conflicts! I've also updated the `speaker_embeddings` processing in `BarkProcessor`. Could you take a look when you have time ? Thanks!

> Before I start a full review of this model, could you explain why the model was written in this structure - with no `forward` method and no task specific...

Regarding the last part of your question, the `BarkModel` architecture can't really be used for anything other than this specific task, so it's a kind of single-purpose model. We considered...

Hi @amyeroberts, you're welcome! Have you had time to look into the PR? I'd be happy to answer any questions you might have about my code, as it's a rather...

Hi @amyeroberts , Thanks for the comprehensive review! I've answered most of your comments, but there are still a few I've asked questions/clarifications about!

Hi @amyeroberts and @sgugger! Many thanks for the additional review (and thanks @sanchit-gandhi for your insights)! I've addressed most of your comments, especially those requiring more consistency with transformers regarding...

Hi @amyeroberts, there was a time-out when executing the python snippet of the `generate` docstrings. I took advantage of this to [add the ability to specify sub-model specific parameters](https://github.com/huggingface/transformers/pull/24086/commits/97cdc38e66b600d3bc1d82c56099acf3cdc6a0f8) in...

Hey @amyeroberts, I've addressed your last remarks! Does that work with you? Many thanks!