verl/models/transformers/qwen2.py

• qwen2_flash_attn_forward · function · L28-L137 — def qwen2_flash_attn_forward( self, hidden_states: torch.Tensor, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, past_key_value: Optional[Cache] = None, output_attentions: bool = False, use_cache: bool = False, cache_position: Optional[torch.LongTensor] = None, position_embeddings: Optional[Tuple[torch.Tensor, torch.Tensor]] = None, # will become mandatory in v4.46 )