feat: enhance quickstart_demo for Apple Silicon and auto hub load

quickstart_demo.py

@@ -2,14 +2,16 @@
 Cerebellum-2B: 25ms Non-Autoregressive Agent Decision Model
 Quickstart Demo (3 lines to make a fast decision)
 """
+import os
+import sys
 import torch
 from modeling_cerebellum import CerebellumModel
 
 def main():
-    # 1. Load Model (BF16 or FP8)
-    device = "cuda:0" if torch.cuda.is_available() else "cpu"
-    print(f"Loading Cerebellum-2B on {device}...")
-    model = CerebellumModel.from_pretrained(".", device=device)
+    # 1. Load Model (Support local folder or Hugging Face Hub, auto-detect CUDA/MPS/CPU)
+    model_path = sys.argv[1] if len(sys.argv) > 1 else ("." if os.path.exists("./model.safetensors") else "mkzero/Cerebellum-2B-INT8")
+    print(f"Loading Cerebellum-2B from: {model_path}...")
+    model = CerebellumModel.from_pretrained(model_path)
 
     # 2. Define Agent State and Candidate Actions
     state = """