Upgrade to Transformers v4.48

[regisss]

What does this PR do?

As per title.

Before submitting

• This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Did you make sure to update the documentation with your changes?
• Did you write any new necessary tests?

[regisss]

@regisss v4.48.1 is released. Should we switch to that one? Fixes multiple issues. I noticed LLF upgrade PR also is going directly for this version and throws warning for 4.46.0-4.48.0 versions. https://github.com/hiyouga/LLaMA-Factory/pull/6628/files#diff-78c85f34fa653e9d750cad5a41efa8bda242c096dd06b2fad00e08c25cb48b85R97-R103

Yes, I'll add the changes brought by the patch in the next few days

[regisss]

@yafshar @dsocek This branch is now aligned with Transformers v4.48.2 that got released a few hours ago

[yafshar]

@yafshar @dsocek This branch is now aligned with Transformers v4.48.2 that got released a few hours ago

Awesome regiss! thanks, I will start do some testing and integration of other work with this branch

[yafshar]

@regisss can you please cherry pick this huggingface/transformers#35438 to this branch. It is relevant to def training_step(

optimum-habana/optimum/habana/transformers/trainer.py

Line 1631 in 0f68bbb

with self.compute_loss_context_manager():

[regisss]

@regisss can you please cherry pick this huggingface/transformers#35438 to this branch. It is relevant to def training_step(

optimum-habana/optimum/habana/transformers/trainer.py

Line 1631 in 0f68bbb

with self.compute_loss_context_manager():

@yafshar huggingface/transformers#35438 was introduced in v4.48.0 and I cherry-picked later huggingface/transformers#35651 that fixes it. Is there anything specific to huggingface/transformers#35438 that you would like to have here?

[yafshar]

@regisss can you please cherry pick this huggingface/transformers#35438 to this branch. It is relevant to def training_step(

optimum-habana/optimum/habana/transformers/trainer.py

Line 1631 in 0f68bbb

with self.compute_loss_context_manager():

@yafshar huggingface/transformers#35438 was introduced in v4.48.0 and I cherry-picked later huggingface/transformers#35651 that fixes it. Is there anything specific to huggingface/transformers#35438 that you would like to have here?

Thanks, in that case no need for any change (my mistake). I was working on an older commit and encountered the issue which is fixed in this commit and has been changed later.