[Fix] Fix cuda graph padding for triton attention backend (#1782)

2024-10-24 12:33:15 -07:00
parent 0089c4bc96
commit fc82f5a743
6 changed files with 3 additions and 19 deletions
@@ -210,9 +210,6 @@ class FlashInferAttnBackend(AttentionBackend):
            encoder_lens=encoder_lens[:bs] if encoder_lens is not None else None,
        )

-    def get_cuda_graph_seq_len_fill_value(self):
-        return 0
-
    def forward_extend(
        self, q, k, v, layer: RadixAttention, forward_batch: ForwardBatch
    ):