fix cpu inference

modeling_qwen.py

@@ -1177,7 +1177,7 @@ def _rotate_half(x):
 
 
 def apply_rotary_pos_emb(t, freqs):
-    if apply_rotary_emb_func is not None:
+    if apply_rotary_emb_func is not None and t.is_cuda:
         t_ = t.float()
         freqs = freqs.squeeze(0).squeeze(1)
         cos = freqs[:, : freqs.shape[-1] // 2].cos()