Update app.py

app.py

@@ -70,14 +70,17 @@ def generate_medical_response(message, history):
             max_new_tokens=256,
             temperature=0.3, 
             top_p=0.85,
-            repetition_penalty=1.1,
+            repetition_penalty=1.0,
             do_sample=True,
             eos_token_id=tokenizer.eos_token_id,
             pad_token_id=tokenizer.eos_token_id
         )
     
     decoded_output = tokenizer.decode(outputs[0], skip_special_tokens=False)
-    final_answer = decoded_output.split("<|assistant|>\n")[-1].replace("</s>", "").strip()
+
+    raw_answer = decoded_output.split("<|assistant|>\n")[-1]
+
+    final_answer = raw_answer.split("</s>")[0].split("<|user|>")[0].strip()
     
     disclaimer = (
         "\n\n---\n"
@@ -113,4 +116,5 @@ demo = gr.ChatInterface(
 if __name__ == "__main__":
     demo.launch(
         share=True,
+        server_port=7860,
     )