Conversation
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
| logger.addHandler(stream_handler) | ||
|
|
||
|
|
||
| @unittest.skipIf(is_transformers_version(">=", "4.57.5"), "Size mismatch") |
There was a problem hiding this comment.
Internal discussion: https://huggingface.slack.com/archives/C014N4749J9/p1768474502541349
| torch.nn.ConvTranspose2d, | ||
| torch.nn.ConvTranspose3d, | ||
| torch.nn.Linear, | ||
| torch.nn.Embedding, |
There was a problem hiding this comment.
Happening because of the way weight loading is done in v5.
| model = AutoModel.from_pretrained( | ||
| "hf-internal-testing/tiny-stable-diffusion-torch", subfolder="text_encoder", use_safetensors=False | ||
| ) |
There was a problem hiding this comment.
Internal discussion: https://huggingface.slack.com/archives/C014N4749J9/p1768462040821759
| input_ids = ( | ||
| input_ids["input_ids"] if not isinstance(input_ids, list) and "input_ids" in input_ids else input_ids | ||
| ) |
There was a problem hiding this comment.
Internal discussion https://huggingface.slack.com/archives/C014N4749J9/p1768537424692669
| inputs = { | ||
| "prompt": "dance monkey", | ||
| "negative_prompt": "", | ||
| "negative_prompt": "bad", |
There was a problem hiding this comment.
Otherwise, the corresponding tokenizer outputs:
negative_prompt=[' ']
prompt=[' ']
text_input_ids=tensor([], size=(1, 0), dtype=torch.int64)which leads to:
E RuntimeError: cannot reshape tensor of 0 elements into shape [1, 0, -1, 8] because the unspecified dimension size -1 can be any value and is ambiguous|
Hmm, https://github.com/huggingface/diffusers/actions/runs/21354964855/job/61460242386?pr=12976 fails on this PR but passes without any regrets on this https://github.com/huggingface/diffusers/actions/runs/21344761402/job/61450173169?pr=12996. So, I am not sure at this point what's happening TBH. Other failures seem to be already existing and well-known. |
transformers v5transformers v5
|
Merging. |
What does this PR do?
Migrate to
transformersv5 and perform the changes necessary to facilitate that.Bunch of inline comments to make the reviewers aware of internal discussions.