Add print statements

modeling_cogvlm.py

@@ -296,6 +296,10 @@ class VisionExpertAttention(nn.Module):
             print("First values of value_states:", value_states[0,0,-3:,-3:])
             print("Mean of value_states:", value_states.mean())
 
+            print("Shape of the attention_mask:", attention_mask.shape)
+            print("Mean of the attention_mask:", attention_mask.float().mean())
+            print("Is_full:", (attention_mask > 0).all())
+
         context_layer = attention_fn(
             query_layer=query_states, key_layer=key_states, value_layer=value_states, attention_mask=attention_mask,
             scaling_attention_score=True, attention_dropout=None)