Update app.py

app.py

@@ -15,17 +15,19 @@ import scipy.io.wavfile as wav
 # ---------------------------------------------------------------------
 def load_llama_pipeline_zero_gpu(model_id: str, token: str):
     try:
+        if not torch.cuda.is_available():
+            raise RuntimeError("ZeroGPU is not properly initialized or GPU is unavailable.")
         tokenizer = AutoTokenizer.from_pretrained(model_id, use_auth_token=token)
         model = AutoModelForCausalLM.from_pretrained(
             model_id,
             use_auth_token=token,
             torch_dtype=torch.float16,
-            device_map="auto",  # Use device map to offload computations
-            trust_remote_code=True  # Enables execution of remote code for Zero GPU
+            device_map="auto",
+            trust_remote_code=True
         )
         return pipeline("text-generation", model=model, tokenizer=tokenizer)
     except Exception as e:
-        return str(e)
+        return f"Error loading model: {e}"
 
 # ---------------------------------------------------------------------
 # Generate Radio Script
@@ -115,4 +117,4 @@ with gr.Blocks() as demo:
 # ---------------------------------------------------------------------
 # Launch App
 # ---------------------------------------------------------------------
-demo.launch(debug=True, share=True)
+demo.launch(debug=True)