fix typo (#3153)

gongshaotian · web-flow · commit d4059cabf0c2 · 2025-08-01T22:34:59.000+08:00
diff --git a/fastdeploy/model_executor/layers/attention/flash_attn_backend.py b/fastdeploy/model_executor/layers/attention/flash_attn_backend.py
@@ -208,7 +208,7 @@ def init_attention_metadata(self, forward_meta: ForwardMeta):
         ) = pre_cache_len_concat(
             forward_meta.seq_lens_decoder,
             forward_meta.seq_lens_this_time,
-            metadata.set_max_lengths[2],
+            forward_meta.max_len_tensor_cpu[2],
             self.block_size,
         )
 

Original file line number	Diff line number	Diff line change
`@@ -208,7 +208,7 @@ def init_attention_metadata(self, forward_meta: ForwardMeta):`
`208`	`208`	`) = pre_cache_len_concat(`
`209`	`209`	`forward_meta.seq_lens_decoder,`
`210`	`210`	`forward_meta.seq_lens_this_time,`
`211`		`- metadata.set_max_lengths[2],`
	`211`	`+ forward_meta.max_len_tensor_cpu[2],`
`212`	`212`	`self.block_size,`
`213`	`213`	`)`
`214`	`214`