cannot import name '_expand_mask'

[Nyuo70]

When I try to execute train.py, I encounter the following error:
ImportError: cannot import name '_expand_mask' from 'transformers.models.clip.modeling_clip' (/home/tmp/onishi3n/Honki/lib/python3.10/site-packages/transformers/models/clip/modeling_clip.py)

[boop-yyt]

After updating the "transformers" version to either 4.28 or 4.31, the issue was resolved.

[Nyuo70]

Thank you!

[Nyuo70]

resolved

[vedant001]

after updating it to transformers==4.31.0 it throws error of
RuntimeError: Failed to import diffusers.pipelines.controlnet.pipeline_controlnet_blip_diffusion because of the following error (look up to see its traceback):
cannot import name 'is_scaled_image' from 'transformers.image_utils'

[Cuppinono]

me too, I have this question

[Preciousrs]

me too, I have this question
Do you resolve this question?

[qiangtang2017]

transformers 4.31.0 is ok

[vedant001]

pip install diffusers==0.15.0
try above line
@Cuppinono @Preciousrs

[Abecid]

I am getting this error with transformers 4.35.2

[Hambaobao]

ImportError: cannot import name '_expand_mask' from 'transformers.models.bloom.modeling_bloom' (/usr/local/lib/python3.9/dist-packages/transformers/models/bloom/modeling_bloom.py)
I got this error with transformers 4.36.1.

[ghoshsubh]

ImportError: cannot import name '_expand_mask' from 'transformers.models.bloom.modeling_bloom' (/usr/local/lib/python3.9/dist-packages/transformers/models/bloom/modeling_bloom.py) I got this error with transformers 4.36.1.

I have the same error with transformers 4.36.2

[SaikoujiAsahi]

I solve the error with transformers 4.33.0

[iwannabewater]

I solve the error with transformers 4.33.0I solved the error with transformers 4.33.0

Thx, this works for me!

[maulikmadhavi]

Copy the function definition from previous transformer libraries

# Copied from transformers.models.bart.modeling_bart._expand_mask
def _expand_mask(mask: torch.Tensor, dtype: torch.dtype, tgt_len: Optional[int] = None):
    """
    Expands attention_mask from `[bsz, seq_len]` to `[bsz, 1, tgt_seq_len, src_seq_len]`.
    """
    bsz, src_len = mask.size()
    tgt_len = tgt_len if tgt_len is not None else src_len

    expanded_mask = mask[:, None, None, :].expand(bsz, 1, tgt_len, src_len).to(dtype)

    inverted_mask = 1.0 - expanded_mask

    return inverted_mask.masked_fill(inverted_mask.to(torch.bool), torch.finfo(dtype).min)

Then install using python setup.py install.

Btw I have already created pull request, refer to my cloned version https://github.com/maulikmadhavi/LAVIS/tree/patch-requirements. Thanks

[Deephug969]

transformers 4.31.0 is ok!