Switch to GPT-2 for better serverless inference

app.py

@@ -11,7 +11,7 @@ load_dotenv()
 
 # Configuration
 API_BASE = "https://api-inference.huggingface.co/models/"
-MODEL_ID = "Qwen/Qwen3-0.6B"
+MODEL_ID = "gpt2"
 HF_TOKEN = os.getenv('HF_NEXT_TOKEN_PREDICTOR_TOKEN', '')
 
 def show_token(token: str) -> str:
@@ -54,13 +54,13 @@ def predict_next_token(text: str, top_k: int = 10, hide_punctuation: bool = Fals
         response = requests.post(url, headers=headers, json=payload, timeout=30)
         
         if not response.ok:
-            error_msg = f"API Error: {response.status_code}"
+            error_msg = f"API Error: {response.status_code} for model {MODEL_ID}"
             try:
                 error_detail = response.json()
                 if 'error' in error_detail:
                     error_msg += f" - {error_detail['error']}"
             except:
-                error_msg += f" - {response.text[:100]}"
+                error_msg += f" - {response.text[:200]}"
             return error_msg, ""
         
         result = response.json()