Update

app.py

@@ -139,7 +139,7 @@ def generate(input_message: str, db_info="", temperature=0.2, top_p=0.9, top_k=0
 
     input_ids = tok(messages, return_tensors="pt").input_ids
     input_ids = input_ids.to(m.device)
-    streamer = TextIteratorStreamer(tok, timeout=100.0, skip_prompt=True, skip_special_tokens=True)
+    streamer = TextIteratorStreamer(tok, timeout=1000.0, skip_prompt=True, skip_special_tokens=True)
     generate_kwargs = dict(
         input_ids=input_ids,
         max_new_tokens=max_new_tokens,