lerobot

/

pi0

like 257

Follow

LeRobot 1.56k

(lerobot) arc@nvda:~/Downloads/lerobot-main$ python lerobot/scripts/eval.py --policy.path=lerobot/pi0 --env.type=aloha --eval.batch_size=10 --eval.n_episodes=10 --use_amp=false --device=cuda
INFO 2025-02-27 10:39:58 pts/eval.py:458 {'device': 'cuda',
'env': {'episode_length': 400,
'features': {'action': {'shape': (14,),
'type': <FeatureType.ACTION: 'ACTION'>},
'agent_pos': {'shape': (14,),
'type': <FeatureType.STATE: 'STATE'>},
'pixels/top': {'shape': (480, 640, 3),
'type': <FeatureType.VISUAL: 'VISUAL'>}},
'features_map': {'action': 'action',
'agent_pos': 'observation.state',
'pixels/top': 'observation.images.top',
'top': 'observation.image.top'},
'fps': 50,
'obs_type': 'pixels_agent_pos',
'render_mode': 'rgb_array',
'task': 'AlohaInsertion-v0'},
'eval': {'batch_size': 10, 'n_episodes': 10, 'use_async_envs': False},
'job_name': 'aloha_pi0',
'output_dir': PosixPath('outputs/eval/2025-02-27/10-39-58_aloha_pi0'),
'policy': {'adapt_to_pi_aloha': False,
'attention_implementation': 'eager',
'chunk_size': 50,
'empty_cameras': 0,
'freeze_vision_encoder': True,
'input_features': {},
'max_action_dim': 32,
'max_state_dim': 32,
'n_action_steps': 50,
'n_obs_steps': 1,
'normalization_mapping': {'ACTION': <NormalizationMode.MEAN_STD: 'MEAN_STD'>,
'STATE': <NormalizationMode.MEAN_STD: 'MEAN_STD'>,
'VISUAL': <NormalizationMode.IDENTITY: 'IDENTITY'>},
'num_steps': 10,
'optimizer_betas': (0.9, 0.95),
'optimizer_eps': 1e-08,
'optimizer_lr': 2.5e-05,
'optimizer_weight_decay': 1e-10,
'output_features': {},
'proj_width': 1024,
'resize_imgs_with_padding': (224, 224),
'scheduler_decay_lr': 2.5e-06,
'scheduler_decay_steps': 30000,
'scheduler_warmup_steps': 1000,
'tokenizer_max_length': 48,
'train_expert_only': False,
'train_state_proj': True,
'use_cache': True,
'use_delta_joint_actions_aloha': False},
'seed': 1000,
'use_amp': False}
INFO 2025-02-27 10:39:58 pts/eval.py:467 Output dir: outputs/eval/2025-02-27/10-39-58_aloha_pi0
INFO 2025-02-27 10:39:58 pts/eval.py:469 Making environment.
INFO 2025-02-27 10:39:58 /init.py:88 MUJOCO_GL is not set, so an OpenGL backend will be chosen automatically.
INFO 2025-02-27 10:39:58 /init.py:96 Successfully imported OpenGL backend: %s
INFO 2025-02-27 10:39:58 /init.py:31 MuJoCo library version is: %s
INFO 2025-02-27 10:39:59 pts/eval.py:472 Making policy.
Stepping through eval batches: 0%| | 0/1 [00:00<?, ?it/s]
Traceback (most recent call last):teps: 0%| | 0/400 [00:00<?, ?it/s]
File "/home/arc/Downloads/lerobot-main/lerobot/scripts/eval.py", line 502, in
eval()
File "/home/arc/Downloads/lerobot-main/lerobot/configs/parser.py", line 120, in wrapper_inner
response = fn(cfg, *args, **kwargs)
File "/home/arc/Downloads/lerobot-main/lerobot/scripts/eval.py", line 481, in eval
info = eval_policy(
File "/home/arc/Downloads/lerobot-main/lerobot/scripts/eval.py", line 294, in eval_policy
rollout_data = rollout(
File "/home/arc/Downloads/lerobot-main/lerobot/scripts/eval.py", line 159, in rollout
action = policy.select_action(observation)
File "/home/arc/anaconda3/envs/lerobot/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 116, in decorate_context
return func(*args, **kwargs)
File "/home/arc/Downloads/lerobot-main/lerobot/common/policies/pi0/modeling_pi0.py", line 276, in select_action
batch = self.normalize_inputs(batch)
File "/home/arc/anaconda3/envs/lerobot/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1739, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/home/arc/anaconda3/envs/lerobot/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1750, in _call_impl
return forward_call(*args, **kwargs)
File "/home/arc/anaconda3/envs/lerobot/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 116, in decorate_context
return func(*args, **kwargs)
File "/home/arc/Downloads/lerobot-main/lerobot/common/policies/normalize.py", line 169, in forward
assert not torch.isinf(mean).any(), _no_stats_error_str("mean")
AssertionError: mean is infinity. You should either initialize with stats as an argument, or use a pretrained model.

I am not quite familiar with lerobot,could someone tell me how to solve this problem.