Add Jukebox model (replaces #16875) #17826
Conversation
ArthurZucker
changed the title
|
Replaces (#16875) |
ArthurZucker
added
the
WIP
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. |
|
As it was previously requested, you can now instantiate |
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. |
| This model was contributed by [Arthur Zucker](https://huggingface.co/ArthurZ). | ||
| The original code can be found [here](https://github.com/openai/jukebox). | ||
|
|
||
| ## JukeboxConfig |
There was a problem hiding this comment.
I would regroup all configs, then the tokenier, then the models (like for [CLIP]https://huggingface.co/docs/transformers/model_doc/clip)
| if isinstance(module, JukeboxPrior) or isinstance(module, JukeboxVQVAE): | ||
| module.apply(module._init_weights) |
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. |
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. |
This is a draft pull request.
What does this PR do?
This PR will progressively add the Jukebox model to the hub.
It is linked to #16870.
Currently planned steps (WIP)
transformeres-cli add-new-model-likesrc/transformers/tokenization_jukebox.pysrc/transformers/test_tokenization_jukebox.pysrc/transformers/configuration_jukebox.pysrc/transformers/modeling_jukebox.pysrc/transformers/configuration_jukebox.pydocs/source/model_doc/jukebox.rstsrc/transformers/tokenization_jukebox_fast.py(will most probably use WordLevel tokenizer). Also requires to implement a converter functionclass JukeboxConverter(Converter):