We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
_prepare_decoder_attention_mask()
_prepare_4d_causal_attention_mask()