missing revision parameter

run_test.py

@@ -13,7 +13,7 @@ Model = transformers.AutoModelForCausalLM
 if config.model_type == 'deepseek_v3':
     #Model = transformers.DeepseekV3ForCausalLM
     pass
-model = Model.from_pretrained(model_id, trust_remote_code=True, torch_dtype='auto', device_map='cpu')
+model = Model.from_pretrained(model_id, revision=revision, trust_remote_code=True, torch_dtype='auto', device_map='cpu')
 if config.model_type == 'deepseek_v3':
     model._supports_cache_class = False
 model = accelerate.cpu_offload(model, 'cuda:0', offload_buffers=True)