[Bugfix] Fix interns1-vit qk norm code path (vllm-project#27480)

Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>

Author: Isotr0py

vllm/model_executor/models/interns1_vit.py

@@ -217,16 +217,15 @@ def __init__(
         self.attn = MultiHeadAttention(self.num_heads, self.head_dim, self.scale)
 
     def forward(self, x: torch.Tensor) -> torch.Tensor:
-        B, N, C = x.shape
+        """x shape: (B, N, C)"""
 
         q = self.q_proj(x)
         k = self.k_proj(x)
         v = self.v_proj(x)
 
         if self.qk_normalization:
-            B_, N_, H_, D_ = q.shape
-            q = self.q_norm(q.flatten(-2, -1)).view(B_, N_, H_, D_)
-            k = self.k_norm(k.flatten(-2, -1)).view(B_, N_, H_, D_)
+            q = self.q_norm(q)
+            k = self.k_norm(k)
 
         # Use unified MultiHeadAttention with automatic backend selection
         x = self.attn(q, k, v)