Fix InferenceClient to use provider=auto and add error handling

- Use provider="auto" to let HF Hub pick the right inference provider
- Pass model to chat_completion() instead of constructor
- Add try/except to show actual error messages in chat

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

app.py

@@ -18,30 +18,32 @@ def respond(
         yield "⚠️ Silakan login dulu dengan tombol Login di sidebar."
         return
 
-    client = InferenceClient(token=hf_token.token, model="openai/gpt-oss-20b")
+    client = InferenceClient(
+        provider="auto",
+        api_key=hf_token.token,
+    )
 
     messages = [{"role": "system", "content": system_message}]
-
     messages.extend(history)
-
     messages.append({"role": "user", "content": message})
 
     response = ""
 
-    for chunk in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        choices = chunk.choices
-        token = ""
-        if len(choices) and choices[0].delta.content:
-            token = choices[0].delta.content
-
-        response += token
-        yield response
+    try:
+        for chunk in client.chat_completion(
+            messages,
+            model="openai/gpt-oss-20b",
+            max_tokens=max_tokens,
+            stream=True,
+            temperature=temperature,
+            top_p=top_p,
+        ):
+            choices = chunk.choices
+            if len(choices) and choices[0].delta.content:
+                response += choices[0].delta.content
+            yield response
+    except Exception as e:
+        yield f"❌ Error: {e}"
 
 
 """