Update paraphraser.py

paraphraser.py

@@ -9,7 +9,7 @@ def paraphrase_comment(comment):
     """
     try:
         start_time = time.time()
-        print(f"Starting paraphrasing for comment: {comment[:50]}...")  # Log the start with a snippet of the comment
+        print(f"Starting paraphrasing for comment: {comment[:50]}...")
 
         # Access the model and tokenizer
         model = paraphraser_model.model
@@ -33,16 +33,16 @@ def paraphrase_comment(comment):
             return_tensors="pt",
             truncation=True,
             padding=True,
-            max_length=256,  # Reduced from 512 to 256
+            max_length=256,
             return_attention_mask=True
         ).to(model.device)
 
-        # Generate the paraphrased comment
+        # Generate the paraphrased comment using max_new_tokens
         outputs = model.generate(
             input_ids=inputs["input_ids"],
             attention_mask=inputs["attention_mask"],
-            max_length=256,  # Reduced from 512 to 256
-            num_beams=2,  # Reduced from 5 to 2 for faster inference
+            max_new_tokens=50,  # Generate up to 50 new tokens
+            num_beams=2,
             no_repeat_ngram_size=2,
             early_stopping=True
         )
@@ -59,5 +59,5 @@ def paraphrase_comment(comment):
         return paraphrased_comment if paraphrased_comment else "Error: Unable to generate paraphrase."
 
     except Exception as e:
-        print(f"Error during paraphrasing: {str(e)}")  # Improved error logging
+        print(f"Error during paraphrasing: {str(e)}")
         return f"Error: Unable to generate paraphrase: {str(e)}"