GPT2LMHeadModel does not support an attention implementation through torch.nn.functional.scaled_dot_product_attention yet

[hanjidani]

`---------------------------------------------------------------------------
ValueError Traceback (most recent call last)
in <cell line: 21>()
19
20 print("=> Creating model")
---> 21 model = VideoRecap(old_args, eval_only=True)
22 model = model.cuda()
23 model.load_state_dict(state_dict, strict=True)

5 frames
/usr/local/lib/python3.10/dist-packages/transformers/modeling_utils.py in _check_and_enable_sdpa(cls, config, hard_check_only)
1729 if hard_check_only:
1730 if not cls._supports_sdpa:
-> 1731 raise ValueError(
1732 f"{cls.name} does not support an attention implementation through torch.nn.functional.scaled_dot_product_attention yet."
1733 " Please request the support for this architecture: huggingface/transformers#28005. If you believe"

ValueError: GPT2LMHeadModel does not support an attention implementation through torch.nn.functional.scaled_dot_product_attention yet. Please request the support for this architecture: huggingface/transformers#28005. If you believe this error is a bug, please open an issue in Transformers GitHub repository and load your model with the argument attn_implementation="eager" meanwhile. Example: model = AutoModel.from_pretrained("openai/whisper-tiny", attn_implementation="eager")

`

Here is the output of the demo file while trying to load VideoRecap model under the Clip Caption section.

Which version of the PyTorchcan run it?

[yanlai00]

Same issue here!

Edit: Using the provided conda-pack environment resolves the problem.

[dwin222]

Same issue here!

Edit: Using the provided conda-pack environment resolves the problem.

Could you please provide more detailed instructions on how to resolve this issue?

[hanjidani]

Same issue here!

Edit: Using the provided conda-pack environment resolves the problem.

How should I do this? What is suitable conda pack for this repo?