Update app.py

app.py

@@ -72,14 +72,22 @@ def respond(message, history, system_message, max_tokens, temperature, top_p):
         # Generate prompt
         prompt = generate_prompt(context=context, question=message, system_prompt=system_message)
         
-        # Generate response without streamer for direct string output
-        output = text_pipeline(
-            prompt,
+        # Set up the pipeline
+        text_pipeline = pipeline(
+            "text-generation",
+            model=model,
+            tokenizer=tokenizer,
             max_new_tokens=max_tokens,
             temperature=temperature,
             top_p=top_p,
-            repetition_penalty=1.15,
-            return_full_text=False
+            repetition_penalty=1.15
+        )
+        
+        # Generate response
+        output = text_pipeline(
+            prompt,
+            return_full_text=False,
+            max_new_tokens=max_tokens
         )[0]['generated_text']
         
         yield output.strip()