LeRobot smolVLA training failing on leisaac dataset
I downloaded the dataset from https://huggingface.co/LightwheelAI/leisaac-pick-orange-v0
created my own dataset, push this dataset to that and run belwo command to run the training code-
LeRobot version 0.3.4
lerobot-train --policy.path=lerobot/smolvla_base --dataset.repo_id=${HF_USER}/leisaac_so101_pick_oranges --batch_size=64 --steps=100 --output_dir=outputs/train/my_smolvla --job_name=my_smolvla_training --policy.device=cuda --wandb.enable=false
it is failing -
(lerobot) C:\Users\Admin01\Desktop\Physical AI\lerobot>lerobot-train --policy.path=lerobot/smolvla_base --dataset.repo_id=neso613/leisaac_so101_pick_oranges --batch_size=64 --steps=100 --output_dir=outputs/train/my_smolvla --job_name=my_smolvla_training --policy.device=cuda --wandb.enable=false
WARNING 2026-01-13 17:23:33 deo_utils.py:40 'torchcodec' is not available in your platform, falling back to 'pyav' as a default decoder
WARNING 2026-01-13 17:23:35 deo_utils.py:40 'torchcodec' is not available in your platform, falling back to 'pyav' as a default decoder
Traceback (most recent call last):
File "", line 198, in _run_module_as_main
File "", line 88, in run_code
File "E:\miniconda\envs\lerobot\Scripts\lerobot-train.exe_main.py", line 6, in
File "E:\miniconda\envs\lerobot\Lib\site-packages\lerobot\scripts\train.py", line 296, in main
train()
File "E:\miniconda\envs\lerobot\Lib\site-packages\lerobot\configs\parser.py", line 225, in wrapper_inner
response = fn(cfg, *args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^
File "E:\miniconda\envs\lerobot\Lib\site-packages\lerobot\scripts\train.py", line 110, in train
cfg.validate()
File "E:\miniconda\envs\lerobot\Lib\site-packages\lerobot\configs\train.py", line 76, in validate
self.policy = PreTrainedConfig.from_pretrained(policy_path, cli_overrides=cli_overrides)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "E:\miniconda\envs\lerobot\Lib\site-packages\lerobot\configs\policies.py", line 207, in from_pretrained
return draccus.parse(orig_config.class, config_file, args=cli_overrides)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "E:\miniconda\envs\lerobot\Lib\site-packages\draccus\argparsing.py", line 211, in parse
return parser.parse_args(args)
^^^^^^^^^^^^^^^^^^^^^^^
File "E:\miniconda\envs\lerobot\Lib\site-packages\draccus\argparsing.py", line 102, in parse_args
args, _ = self.parse_known_args(args, namespace, is_parse_args=True)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "E:\miniconda\envs\lerobot\Lib\site-packages\draccus\argparsing.py", line 136, in parse_known_args
parsed_t = self._postprocessing(parsed_args)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "E:\miniconda\envs\lerobot\Lib\site-packages\draccus\argparsing.py", line 173, in _postprocessing
with open(config_path, "r") as f:
^^^^^^^^^^^^^^^^^^^^^^
PermissionError: [Errno 13] Permission denied: 'C:\Users\Admin01\AppData\Local\Temp\2\tmpj9jp3yp2'
LeRobot version 0.4.2
lerobot-train --policy.path=lerobot/smolvla_base --dataset.repo_id=${HF_USER}/leisaac_so101_pick_oranges --batch_size=64 --steps=100 --output_dir=outputs/train/my_smolvla --job_name=my_smolvla_training --policy.device=cuda --wandb.enable=false --policy.push_to_hub=false
Error:
WARNING:root:'torchcodec' is not available in your platform, falling back to 'pyav' as a default decoder
WARNING:root:'torchcodec' is not available in your platform, falling back to 'pyav' as a default decoder
WARNING:root:No accelerated backend detected. Using default cpu, this will be slow.
WARNING:lerobot.configs.policies:Device 'cuda' is not available. Switching to 'cpu'.
WARNING:root:No accelerated backend detected. Using default cpu, this will be slow.
WARNING:lerobot.configs.policies:Device 'cuda' is not available. Switching to 'cpu'.
INFO 2026-01-13 17:46:25 ot_train.py:163 {'batch_size': 64,
'checkpoint_path': None,
'dataset': {'episodes': None,
'image_transforms': {'enable': False,
.......
'optimizer_lr': 0.0001,
'optimizer_weight_decay': 1e-10,
.....
'pad_language_to': 'max_length',
'prefix_length': 0,
'pretrained_path': 'lerobot\smolvla_base',
-----
'vlm_model_name': 'HuggingFaceTB/SmolVLM2-500M-Video-Instruct'},.............
'project': 'lerobot',
'run_id': None}}
INFO 2026-01-13 17:46:25 ot_train.py:171 Logs will be saved locally.
INFO 2026-01-13 17:46:25 ot_train.py:183 Creating dataset
Traceback (most recent call last):
File "E:\miniconda\envs\lerobot\Lib\site-packages\lerobot\datasets\lerobot_dataset.py", line 103, in init
self.load_metadata()
File "E:\miniconda\envs\lerobot\Lib\site-packages\lerobot\datasets\lerobot_dataset.py", line 161, in load_metadata
self.info = load_info(self.root)
^^^^^^^^^^^^^^^^^^^^
File "E:\miniconda\envs\lerobot\Lib\site-packages\lerobot\datasets\utils.py", line 298, in load_info
info = load_json(local_dir / INFO_PATH)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "E:\miniconda\envs\lerobot\Lib\site-packages\lerobot\datasets\utils.py", line 265, in load_json
with open(fpath) as f:
^^^^^^^^^^^
FileNotFoundError: [Errno 2] No such file or directory: 'C:\Users\Admin01\.cache\huggingface\lerobot\neso613\leisaac_so101_pick_oranges\meta\info.json'
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "", line 198, in _run_module_as_main
File "", line 88, in run_code
File "E:\miniconda\envs\lerobot\Scripts\lerobot-train.exe_main.py", line 6, in
File "E:\miniconda\envs\lerobot\Lib\site-packages\lerobot\scripts\lerobot_train.py", line 449, in main
train()
File "E:\miniconda\envs\lerobot\Lib\site-packages\lerobot\configs\parser.py", line 233, in wrapper_inner
response = fn(cfg, *args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^
File "E:\miniconda\envs\lerobot\Lib\site-packages\lerobot\scripts\lerobot_train.py", line 184, in train
dataset = make_dataset(cfg)
^^^^^^^^^^^^^^^^^
File "E:\miniconda\envs\lerobot\Lib\site-packages\lerobot\datasets\factory.py", line 88, in make_dataset
ds_meta = LeRobotDatasetMetadata(
^^^^^^^^^^^^^^^^^^^^^^^
File "E:\miniconda\envs\lerobot\Lib\site-packages\lerobot\datasets\lerobot_dataset.py", line 106, in init
self.revision = get_safe_version(self.repo_id, self.revision)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "E:\miniconda\envs\lerobot\Lib\site-packages\lerobot\datasets\utils.py", line 558, in get_safe_version
raise BackwardCompatibilityError(repo_id, max(lower_major))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "E:\miniconda\envs\lerobot\Lib\site-packages\lerobot\datasets\backward_compatibility.py", line 47, in init
raise NotImplementedError(
NotImplementedError: Contact the maintainer on Discord.
