Update app.py

app.py

@@ -5,21 +5,24 @@ import os
 # Load secrets
 system_message = os.environ["SYSTEM_MESSAGE"]
 HF_TOKEN = os.environ["HF_TOKEN"]
-MODEL_NAME = os.environ["MODEL_NAME"]  # <-- Add this
+MODEL_NAME = os.environ["MODEL_NAME"]
 
 client = InferenceClient(token=HF_TOKEN)
 
-# Chat function — stripped down for speed
+# Chat function
 def respond(message, history, max_tokens, temperature, top_p):
+    # Build prompt in OpenAI-style
     prompt = [{"role": "system", "content": system_message}]
-    for user, bot in history:
-        if user: prompt.append({"role": "user", "content": user})
-        if bot: prompt.append({"role": "assistant", "content": bot})
+    for msg in history:
+        if msg["role"] == "user":
+            prompt.append({"role": "user", "content": msg["content"]})
+        elif msg["role"] == "assistant":
+            prompt.append({"role": "assistant", "content": msg["content"]})
     prompt.append({"role": "user", "content": message})
 
     response = []
     stream = client.chat.completions.create(
-        model=MODEL_NAME,  # <-- use the secret here
+        model=MODEL_NAME,
         messages=prompt,
         max_tokens=max_tokens,
         temperature=temperature,
@@ -28,7 +31,10 @@ def respond(message, history, max_tokens, temperature, top_p):
     )
 
     for chunk in stream:
-        token = chunk.choices[0].delta.content
+        if not chunk.choices:
+            continue  # skip empty chunks safely
+        delta = chunk.choices[0].delta
+        token = getattr(delta, "content", None)
         if token:
             response.append(token)
             yield "".join(response)
@@ -36,6 +42,7 @@ def respond(message, history, max_tokens, temperature, top_p):
 # UI
 app = gr.ChatInterface(
     fn=respond,
+    type="messages",  # ✅ fixes Gradio warning
     additional_inputs=[
         gr.Slider(16, 2048, value=512, step=1, label="Max Tokens"),
         gr.Slider(0.1, 2.0, value=0.7, step=0.1, label="Temperature"),