Update app.py

app.py

@@ -23,7 +23,7 @@ LICENSE = """
 
 PLACEHOLDER = """
 <div style="padding: 30px; text-align: center; display: flex; flex-direction: column; align-items: center;">
-   <h1 style="font-size: 28px; margin-bottom: 2px; opacity: 0.55;">DeepSeek-R1-Distill-Llama-8B</h1>
+   <h1 style="font-size: 28px; margin-bottom: 2px; opacity: 0.55;">DeepSeek-R1-Distill-Qwen-32B-bnb-4bit</h1>
    <p style="font-size: 18px; margin-bottom: 2px; opacity: 0.65;">Ask me anything...</p>
 </div>
 """
@@ -71,7 +71,9 @@ def chat_llama3_8b(message: str,
         conversation.extend([{"role": "user", "content": user}, {"role": "assistant", "content": assistant}])
     conversation.append({"role": "user", "content": message})
 
-    input_ids = tokenizer.apply_chat_template(conversation, return_tensors="pt").to(model.device)
+    input_ids = tokenizer.apply_chat_template(conversation, return_tensors="pt", add_generation_prompt=True).to(model.device)
+    # for debug
+    print(tokenizer.apply_chat_template(conversation, add_generation_prompt=True, tokenize=False))
     
     streamer = TextIteratorStreamer(tokenizer, timeout=10.0, skip_prompt=True, skip_special_tokens=True)