Update flash_attention_patch.py
To be compatible with the new change in the Transformers library, where a new argument 'padding_mask' was added to forward function of attention layer. https://github.com/huggingface/transformers/pull/25598
Showing
Please register or sign in to comment