fix

app.py

@@ -14,11 +14,11 @@ FastLanguageModel.for_inference(model)  # Enable optimized inference
 
 def predict(question, history):
     history = history or []
-    history.append({"role": "user", "content": question})
+    history.append({"from": "human", "value": question})
     
     inputs = tokenizer(
         tokenizer.apply_chat_template(
-            [{'from': 'human', 'value': question}],
+            history,
             tokenize=False,
             add_generation_prompt=True
         ),
@@ -32,27 +32,33 @@ def predict(question, history):
         partial_text += next_token
         yield partial_text
     
-    history.append({"role": "assistant", "content": partial_text})
+    history.append({"from": "gpt", "value": partial_text})
     return history
 
 # Create the Gradio interface with Markdown support
-with gr.Blocks() as iface:
+with gr.Blocks(css=".message { white-space: pre-wrap; }") as iface:
     chatbot = gr.Chatbot(
         show_label=False,
         container=True,
         height=600,
         bubble_full_width=False,
+        render_markdown=True,  # Enable markdown rendering
+        latex_delimiters=[  # Optional: Enable LaTeX rendering
+            {"left": "$$", "right": "$$", "display": True},
+            {"left": "$", "right": "$", "display": False},
+        ],
     )
+    submit = gr.Button("Submit")
     msg = gr.Textbox(
         label="Message",
-        placeholder="Type your message here...",
+        placeholder="Type your message here... (Markdown supported)",
         lines=2
     )
     clear = gr.Button("Clear")
     
     state = gr.State([])
     
-    msg.submit(
+    submit.click(
         predict,
         [msg, state],
         [chatbot, state],