diff --git a/PyTorch/built-in/mlm/OpenSora1.1/opensora/models/layers/blocks.py b/PyTorch/built-in/mlm/OpenSora1.1/opensora/models/layers/blocks.py
index cca035ac0842776fcce3c84f0776affe77541500..010f772849fc6928dbb4cf4c5d8dbeddd9325ce4 100644
--- a/PyTorch/built-in/mlm/OpenSora1.1/opensora/models/layers/blocks.py
+++ b/PyTorch/built-in/mlm/OpenSora1.1/opensora/models/layers/blocks.py
@@ -291,7 +291,7 @@ class SeqParallelAttention(Attention):
         if self.enable_flashattn:
             if is_npu_available() and q.dtype in [torch.float16, torch.bfloat16]:
                 x = torch_npu.npu_fusion_attention(
-                    q, k, v, self.num_heads, input_layout="BSND",
+                    q, k, v, self.num_heads // sp_size, input_layout="BSND",
                     pse=None,
                     scale=self.scale,
                     pre_tockens=65536,
@@ -449,7 +449,7 @@ class SeqParallelMultiHeadCrossAttention(MultiHeadCrossAttention):
                     ans += m
                     actual_seq_kvlen.append(ans)
             x = torch_npu.npu_fusion_attention(
-                q, k, v, self.num_heads, input_layout="TND",
+                q, k, v, self.num_heads // sp_size, input_layout="TND",
                 pse=None,
                 scale=1.0 / math.sqrt(self.head_dim),
                 pre_tockens=65536,