Update app.py

app.py

@@ -54,13 +54,13 @@ def stream_chat(message: str, history: list, temperature: float, max_new_tokens:
         conversation.extend([{"role": "user", "content": prompt}, {"role": "assistant", "content": answer}])
     conversation.append({"role": "user", "content": message})
     
-    input_ids = tokenizer.apply_chat_template(conversation, tokenize=False, add_generation_prompt=True,return_tensors="pt").to(model.device)
-    #inputs = tokenizer(input_ids, return_tensors="pt").to(0)
+    inputs = tokenizer.apply_chat_template(conversation, tokenize=True, add_generation_prompt=True, return_tensors="pt")
+    input_ids = inputs['input_ids'].to(model.device)
     
     streamer = TextIteratorStreamer(tokenizer, timeout=10., skip_prompt=True, skip_special_tokens=True)
 
     generate_kwargs = dict(
-        input_ids,
+        input_ids=input_ids,
         streamer=streamer,
         top_k=top_k,
         top_p=top_p,