Update app.py

app.py

@@ -23,15 +23,14 @@ def extract_longer_answers_from_paragraphs(paragraphs, query, tokenizer, model):
     context += question
     inputs = tokenizer(context, return_tensors="pt", add_special_tokens=False).to(device)
     top_p = 0.9  # Adjust as needed
-    max_len = 300  # Adjust as needed
+    
     outputs = model.generate(
         **inputs,
         top_p=top_p,
-        max_length=max_len,
         num_beams=1,  
         do_sample= True,
         no_repeat_ngram_size=2,
-        max_new_tokens= 3000,
+        max_new_tokens= 1024,
     )
     answer = tokenizer.decode(outputs[0], skip_special_tokens=True)