fix: use already-quantized 4-bit model for test

test_model_loading.py

@@ -31,9 +31,9 @@ def test_model_loading():
         
         print("   Loading model...")
         model = AutoModelForCausalLM.from_pretrained(
-            "/data/models/Ornith-1.0-35B",
-            quantization_config=bnb_config,
+            "/data/models/Ornith-1.0-35B-4bit",  # Use already-quantized 4-bit model
             device_map="auto",
+            torch_dtype=torch.float16,
             trust_remote_code=True,
             low_cpu_mem_usage=True,
         )