Update app.py

app.py

@@ -63,10 +63,13 @@ def stream_chat(
 
     conversation.append({"role": "user", "content": message})
 
-    print(conversation)
-
-    input_text=tokenizer.apply_chat_template(conversation, tokenize=False)
-    inputs = tokenizer.encode(input_text, return_tensors="pt").to(device)
+    input_ids = tokenizer.apply_chat_template(
+        conversation,
+        tokenize=True,
+        add_generation_prompt=True,
+        return_tensors="pt"
+    ).to(device)
+    
     streamer = TextIteratorStreamer(tokenizer, timeout=60.0, skip_prompt=True, skip_special_tokens=True)
     
     generate_kwargs = dict(