lm_eos_token_id

(nanoVLM) tguancha@ubuntu-PR4910P2:/data/VLM/nanoVLM$ python evaluation.py --model lusxvr/nanoVLM-450M --tasks mmstar,mme --verbosity=DEBUG
2025-07-28 16:27:04.819 | INFO | __main__:cli_evaluate:305 - Verbosity set to DEBUG
2025-07-28 16:27:05.022 | DEBUG | lmms_eval.tasks:_get_task_and_group:452 - `group` and `group_alias` keys in tasks' configs will no longer be used in the next release of lmms-eval. `tag` will be used to allow to call a collection of tasks just like `group`. `group` will be removed in order to not cause confusion with the new ConfigurableGroup which will be the offical way to create groups with addition of group-wide configuations.
2025-07-28 16:27:06.516 | DEBUG | lmms_eval.tasks:_get_task_and_group:478 - File illusionvqa.yaml in /data/MRG/miniconda3/envs/nanoVLM/lib/python3.12/site-packages/lmms_eval/tasks/illusionvqa could not be loaded as a task or group
2025-07-28 16:27:06.637 | DEBUG | lmms_eval.tasks:_get_task_and_group:478 - File _default_template.yaml in /data/MRG/miniconda3/envs/nanoVLM/lib/python3.12/site-packages/lmms_eval/tasks/video-tt could not be loaded as a task or group
2025-07-28 16:27:06.957 | INFO | __main__:cli_evaluate_single:392 - Evaluation tracker args: {'output_path': 'results/'}
2025-07-28 16:27:06.957 | INFO | __main__:cli_evaluate_single:481 - Selected Tasks: ['mme', 'mmstar']
config.json: 1.11kB [00:00, 2.61MB/s]
model.safetensors: 100%|███████████████████████████████████████| 1.80G/1.80G [08:14<00:00, 3.65MB/s]
Traceback (most recent call last):
File "/data/VLM/nanoVLM/evaluation.py", line 554, in
cli_evaluate()
File "/data/VLM/nanoVLM/evaluation.py", line 359, in cli_evaluate
raise e
File "/data/VLM/nanoVLM/evaluation.py", line 341, in cli_evaluate
results, samples = cli_evaluate_single(args)
^^^^^^^^^^^^^^^^^^^^^^^^^
File "/data/VLM/nanoVLM/evaluation.py", line 485, in cli_evaluate_single
wrapped_model = NanoVLMWrapper(
^^^^^^^^^^^^^^^
File "/data/VLM/nanoVLM/eval/lmms_eval_wrapper.py", line 35, in __init__
self.model = VisionLanguageModel.from_pretrained(model).to(device)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/data/VLM/nanoVLM/models/vision_language_model.py", line 230, in from_pretrained
cfg = VLMConfig(**json.load(f))
^^^^^^^^^^^^^^^^^^^^^^^^^
TypeError: VLMConfig.__init__() got an unexpected keyword argument 'lm_eos_token_id'

I try to eval this pretrained model in hf, but it goes wrong, and I want to ask is it a version problem? and how can I solve it.