Update modeling_bert.py

modeling_bert.py

@@ -386,6 +386,7 @@ class BertSelfAttention(nn.Module):
         # Normalize the attention scores to probabilities.
         #attention_probs = nn.functional.softmax(attention_scores, dim=-1)
         attention_probs = softmax_1(attention_scores, dim=-1)
+        print(softmax_1)
 
         # This is actually dropping out entire tokens to attend to, which might
         # seem a bit unusual, but is taken from the original Transformer paper.