Update summarize.py

summarize.py

@@ -45,14 +45,14 @@ def summarize_and_score(ids, mask, model, tokenizer, **kwargs):
     input_ids = ids.to("cuda") if torch.cuda.is_available() else ids
     attention_mask = mask.to("cuda") if torch.cuda.is_available() else mask
 
-    global_attention_mask = torch.zeros_like(attention_mask)
+    #global_attention_mask = torch.zeros_like(attention_mask)
     # put global attention on <s> token
-    global_attention_mask[:, 0] = 1
+    #global_attention_mask[:, 0] = 1
 
     summary_pred_ids = model.generate(
         input_ids,
         attention_mask=attention_mask,
-        global_attention_mask=global_attention_mask,
+        #global_attention_mask=global_attention_mask,
         output_scores=True,
         return_dict_in_generate=True,
         **kwargs,