fix

app.py

@@ -87,7 +87,7 @@ def chat_with_model(messages):
 
     print(f'Step 1: {messages}')
 
-    prompt_text = current_tokenizer.decode(inputs["input_ids"][0], skip_special_tokens=True)
+    prompt_text = current_tokenizer.decode(inputs["input_ids"][0], skip_special_tokens=False)
 
     for token_info in streamer:
         token_str = token_info["token"]
@@ -142,6 +142,7 @@ def chat_with_model(messages):
     current_model.to("cpu")
     torch.cuda.empty_cache()
 
+    messages[-1]["content"] = output_text
     print(f'Step 3: {messages}')
 
     return messages