runtime error

Exit code: 1. Reason: No generation_config.json found at Qwen/Qwen3-30B-A3B-Base/generation_config.json; using built-in defaults. SAE device: cpu (no CUDA visible) Pre-loading model onto GPU… Loading model… config.json: 0%| | 0.00/964 [00:00<?, ?B/s] config.json: 100%|██████████| 964/964 [00:00<00:00, 3.51MB/s] Traceback (most recent call last): File "/app/app_steer_qwen.py", line 1888, in <module> get_model() ~~~~~~~~~^^ File "/app/app_steer_qwen.py", line 160, in get_model _model = AutoModelForCausalLM.from_pretrained( MODEL_PATH, device_map='auto', torch_dtype='auto' ) File "/usr/local/lib/python3.13/site-packages/transformers/models/auto/auto_factory.py", line 402, in from_pretrained return model_class.from_pretrained( ~~~~~~~~~~~~~~~~~~~~~~~~~~~^ pretrained_model_name_or_path, *model_args, config=config, **hub_kwargs, **kwargs ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ ) ^ File "/usr/local/lib/python3.13/site-packages/transformers/modeling_utils.py", line 4291, in from_pretrained device_map = check_and_set_device_map(device_map) # warn, error and fix the device map File "/usr/local/lib/python3.13/site-packages/transformers/integrations/accelerate.py", line 135, in check_and_set_device_map raise ValueError( ...<2 lines>... ) ValueError: Using a `device_map`, `tp_plan`, `torch.device` context manager or setting `torch.set_default_device(device)` requires `accelerate`. You can install it with `pip install accelerate`

Container logs: