Hugging Face's logo

Datasets:
hamidavatar
/
training-logs

Dataset card Files
xet
 Community
hamidavatar commited on
Commit
4c2adf4
·
verified ·
1 Parent(s): f726ab9

Upload training_20260311_062930_exit1.log with huggingface_hub

Browse files
Files changed (1) hide show
  1. training_20260311_062930_exit1.log +1443 -0
training_20260311_062930_exit1.log ADDED
@@ -0,0 +1,1443 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ === Avatar Training (fire-and-forget) ===
2
+ Script: train_pi0fast.py
3
+ Started: 2026-03-11 06:28:57 UTC
4
+
5
+ >>> [0/4] Raising file descriptor limit...
6
+ ulimit -n = 65536
7
+ >>> [1/4] Setting environment...
8
+ HF_TOKEN set: 37 chars
9
+ WANDB_API_KEY set: 0 chars
10
+ >>> [2/4] Logging into HuggingFace...
11
+ The token has not been saved to the git credentials helper. Pass `add_to_git_credential=True` in this function directly or `--add-to-git-credential` if using via `hf`CLI if you want to set the git credential as well.
12
+ Token is valid (permission: write).
13
+ The token `beta-pc` has been saved to /root/.cache/huggingface/stored_tokens
14
+ Your token has been saved to /root/.cache/huggingface/token
15
+ Login successful.
16
+ Note: Environment variable`HF_TOKEN` is set and is the current active token independently from the token you've just configured.
17
+ ⚠️ Warning: 'huggingface-cli login' is deprecated. Use 'hf auth login' instead.
18
+ >>> [3/4] Pre-downloading pretrained model...
19
+ Downloading lerobot/pi0fast-base (default)...
20
+
21
+ Cached at /root/.cache/huggingface/hub/models--lerobot--pi0fast-base/snapshots/0737e07fac1e63519c0bd38bd869934ecd9c3f86
22
+ GPUs: 4x NVIDIA A100-SXM4-80GB
23
+ >>> [4/4] Launching accelerate with 4 GPUs...
24
+ Script: train_pi0fast.py
25
+ Train args: --dataset orange_hook_right_20260310 --steps 10000 --batch-size 8 --lr 5e-5 --save-freq 2000 --push-to-hub --num-workers 8 --task-text pick\ up\ the\ orange\ flat\ package
26
+
27
+ The following values were not passed to `accelerate launch` and had defaults used instead:
28
+ `--num_machines` was set to a value of `1`
29
+ `--mixed_precision` was set to a value of `'no'`
30
+ `--dynamo_backend` was set to a value of `'no'`
31
+ To avoid this warning pass in values for each of the problematic parameters or run `accelerate config`.
32
+
33
+ ============================================================
34
+ π₀-FAST Training
35
+ ============================================================
36
+ Dataset: hamidavatar/orange_hook_right_20260310
37
+ Pretrained: lerobot/pi0fast-base
38
+ Model repo: hamidavatar/pi0fast-orange-hook-right-b32-20260311-0629
39
+ Steps: 10000
40
+ Batch/GPU: 8 (eff: 32)
41
+ Chunk size: 10
42
+ LR: 5e-05
43
+ Save freq: 2000
44
+ GPUs: 4
45
+ ============================================================
46
+
47
+ lerobot_train \
48
+ --dataset.repo_id=hamidavatar/orange_hook_right_20260310 \
49
+ --policy.type=pi0_fast \
50
+ --policy.pretrained_path=lerobot/pi0fast-base \
51
+ --policy.repo_id=hamidavatar/pi0fast-orange-hook-right-b32-20260311-0629 \
52
+ --policy.dtype=bfloat16 \
53
+ --policy.gradient_checkpointing=true \
54
+ --policy.chunk_size=10 \
55
+ --policy.n_action_steps=10 \
56
+ --policy.max_action_tokens=256 \
57
+ --policy.optimizer_lr=5e-05 \
58
+ --policy.optimizer_weight_decay=0.01 \
59
+ --policy.scheduler_warmup_steps=1000 \
60
+ --policy.scheduler_decay_steps=10000 \
61
+ --policy.scheduler_decay_lr=1e-6 \
62
+ --batch_size=8 \
63
+ --steps=10000 \
64
+ --save_freq=2000 \
65
+ --log_freq=100 \
66
+ --save_checkpoint=true \
67
+ --seed=1000 \
68
+ --num_workers=8 \
69
+ --output_dir=outputs/train/pi0fast \
70
+ --dataset.tolerance_s=2.0 \
71
+ --policy.push_to_hub=true \
72
+ --policy.private=true \
73
+
74
+ [Avatar] cudnn.benchmark=True, tf32=True
75
+ [Avatar] Checkpoint push enabled → hamidavatar/pi0fast-orange-hook-right-b32-20260311-0629 (with per-step tags)
76
+ [Avatar] Dataset task text patched → "pick up the orange flat package"
77
+ [Avatar] Installed corrupt-frame retry handler
78
+ [Avatar] Checkpoint push enabled → hamidavatar/pi0fast-orange-hook-right-b32-20260311-0629 (with per-step tags)
79
+ [Avatar] Dataset task text patched → "pick up the orange flat package"
80
+ [Avatar] Installed corrupt-frame retry handler
81
+ [Avatar] Checkpoint push enabled → hamidavatar/pi0fast-orange-hook-right-b32-20260311-0629 (with per-step tags)
82
+ [Avatar] Dataset task text patched → "pick up the orange flat package"
83
+ [Avatar] Installed corrupt-frame retry handler
84
+ [Avatar] Checkpoint push enabled → hamidavatar/pi0fast-orange-hook-right-b32-20260311-0629 (with per-step tags)
85
+ [Avatar] Dataset task text patched → "pick up the orange flat package"
86
+ [Avatar] Installed corrupt-frame retry handler
87
+ usage: lerobot_train [-h] [--config_path str] [--dataset str]
88
+ [--dataset.repo_id str] [--dataset.root str]
89
+ [--dataset.episodes str] [--image_transforms str]
90
+ [--dataset.image_transforms.enable str]
91
+ [--dataset.image_transforms.max_num_transforms str]
92
+ [--dataset.image_transforms.random_order str]
93
+ [--dataset.image_transforms.tfs str]
94
+ [--dataset.revision str]
95
+ [--dataset.use_imagenet_stats str]
96
+ [--dataset.video_backend str] [--dataset.streaming str]
97
+ [--env str]
98
+ [--env.type {aloha,pusht,gym_manipulator,libero,metaworld,isaaclab_arena}]
99
+ [--env.visualization_width str]
100
+ [--env.visualization_height str] [--robot str]
101
+ [--env.robot.type str] [--teleop str]
102
+ [--env.teleop.type str] [--processor str]
103
+ [--env.processor.control_mode str] [--observation str]
104
+ [--env.processor.observation.add_joint_velocity_to_observation str]
105
+ [--env.processor.observation.add_current_to_observation str]
106
+ [--env.processor.observation.add_ee_pose_to_observation str]
107
+ [--env.processor.observation.display_cameras str]
108
+ [--image_preprocessing str]
109
+ [--env.processor.image_preprocessing.crop_params_dict str]
110
+ [--env.processor.image_preprocessing.resize_size str]
111
+ [--gripper str] [--env.processor.gripper.use_gripper str]
112
+ [--env.processor.gripper.gripper_penalty str]
113
+ [--reset str]
114
+ [--env.processor.reset.fixed_reset_joint_positions str]
115
+ [--env.processor.reset.reset_time_s str]
116
+ [--env.processor.reset.control_time_s str]
117
+ [--env.processor.reset.terminate_on_success str]
118
+ [--inverse_kinematics str]
119
+ [--env.processor.inverse_kinematics.urdf_path str]
120
+ [--env.processor.inverse_kinematics.target_frame_name str]
121
+ [--env.processor.inverse_kinematics.end_effector_bounds str]
122
+ [--env.processor.inverse_kinematics.end_effector_step_sizes str]
123
+ [--reward_classifier str]
124
+ [--env.processor.reward_classifier.pretrained_path str]
125
+ [--env.processor.reward_classifier.success_threshold str]
126
+ [--env.processor.reward_classifier.success_reward str]
127
+ [--env.processor.max_gripper_pos str] [--env.name str]
128
+ [--env.task_ids str] [--env.camera_name str]
129
+ [--env.init_states str] [--env.camera_name_mapping str]
130
+ [--env.observation_height str]
131
+ [--env.observation_width str] [--env.control_mode str]
132
+ [--env.obs_type str] [--env.render_mode str]
133
+ [--env.multitask_eval str] [--env.task str]
134
+ [--env.fps str] [--env.features str]
135
+ [--env.features_map str] [--env.max_parallel_tasks str]
136
+ [--env.disable_env_checker str] [--env.hub_path str]
137
+ [--env.episode_length str] [--env.num_envs str]
138
+ [--env.embodiment str] [--env.object str]
139
+ [--env.mimic str] [--env.teleop_device str]
140
+ [--env.seed str] [--env.device str]
141
+ [--env.disable_fabric str] [--env.enable_cameras str]
142
+ [--env.headless str] [--env.enable_pinocchio str]
143
+ [--env.environment str] [--env.state_dim str]
144
+ [--env.action_dim str] [--env.camera_height str]
145
+ [--env.camera_width str] [--env.video str]
146
+ [--env.video_length str] [--env.video_interval str]
147
+ [--env.state_keys str] [--env.camera_keys str]
148
+ [--env.kwargs str] [--policy str]
149
+ [--policy.type {act,diffusion,groot,pi0,pi0_fast,pi05,smolvla,tdmpc,vqbet,wall_x,xvla,sac,reward_classifier,sarm}]
150
+ [--policy.replace_final_stride_with_dilation str]
151
+ [--policy.pre_norm str] [--policy.dim_model str]
152
+ [--policy.n_heads str] [--policy.dim_feedforward str]
153
+ [--policy.feedforward_activation str]
154
+ [--policy.n_encoder_layers str]
155
+ [--policy.n_decoder_layers str] [--policy.use_vae str]
156
+ [--policy.n_vae_encoder_layers str]
157
+ [--policy.temporal_ensemble_coeff str]
158
+ [--policy.kl_weight str]
159
+ [--policy.optimizer_lr_backbone str]
160
+ [--policy.resize_shape str] [--policy.crop_ratio str]
161
+ [--policy.use_separate_rgb_encoder_per_camera str]
162
+ [--policy.down_dims str] [--policy.kernel_size str]
163
+ [--policy.n_groups str]
164
+ [--policy.diffusion_step_embed_dim str]
165
+ [--policy.use_film_scale_modulation str]
166
+ [--policy.noise_scheduler_type str]
167
+ [--policy.num_train_timesteps str]
168
+ [--policy.beta_schedule str] [--policy.beta_start str]
169
+ [--policy.beta_end str] [--policy.prediction_type str]
170
+ [--policy.clip_sample str]
171
+ [--policy.clip_sample_range str]
172
+ [--policy.do_mask_loss_for_padding str]
173
+ [--policy.scheduler_name str] [--policy.image_size str]
174
+ [--policy.base_model_path str]
175
+ [--policy.tokenizer_assets_repo str]
176
+ [--policy.embodiment_tag str] [--policy.tune_llm str]
177
+ [--policy.tune_visual str] [--policy.tune_projector str]
178
+ [--policy.tune_diffusion_model str]
179
+ [--policy.lora_rank str] [--policy.lora_alpha str]
180
+ [--policy.lora_dropout str]
181
+ [--policy.lora_full_model str]
182
+ [--policy.warmup_ratio str] [--policy.use_bf16 str]
183
+ [--policy.video_backend str]
184
+ [--policy.balance_dataset_weights str]
185
+ [--policy.balance_trajectory_weights str]
186
+ [--policy.dataset_paths str] [--policy.output_dir str]
187
+ [--policy.save_steps str] [--policy.max_steps str]
188
+ [--policy.dataloader_num_workers str]
189
+ [--policy.report_to str] [--policy.resume str]
190
+ [--policy.max_action_tokens str]
191
+ [--policy.text_tokenizer_name str]
192
+ [--policy.action_tokenizer_name str]
193
+ [--policy.temperature str]
194
+ [--policy.max_decoding_steps str]
195
+ [--policy.fast_skip_tokens str]
196
+ [--policy.validate_action_token_prefix str]
197
+ [--policy.use_kv_cache str]
198
+ [--policy.paligemma_variant str]
199
+ [--policy.action_expert_variant str]
200
+ [--policy.num_inference_steps str]
201
+ [--policy.time_sampling_beta_alpha str]
202
+ [--policy.time_sampling_beta_beta str]
203
+ [--policy.time_sampling_scale str]
204
+ [--policy.time_sampling_offset str]
205
+ [--policy.image_resolution str]
206
+ [--policy.gradient_checkpointing str]
207
+ [--policy.adapt_to_pi_aloha str]
208
+ [--policy.use_delta_joint_actions_aloha str]
209
+ [--policy.num_steps str] [--policy.use_cache str]
210
+ [--policy.train_expert_only str]
211
+ [--policy.train_state_proj str]
212
+ [--policy.vlm_model_name str]
213
+ [--policy.load_vlm_weights str]
214
+ [--policy.add_image_special_tokens str]
215
+ [--policy.attention_mode str]
216
+ [--policy.prefix_length str]
217
+ [--policy.num_expert_layers str]
218
+ [--policy.num_vlm_layers str]
219
+ [--policy.self_attn_every_n_layers str]
220
+ [--policy.expert_width_multiplier str]
221
+ [--policy.min_period str] [--policy.max_period str]
222
+ [--rtc_config str] [--policy.rtc_config.enabled str]
223
+ [--policy.rtc_config.prefix_attention_schedule str]
224
+ [--policy.rtc_config.max_guidance_weight str]
225
+ [--policy.rtc_config.execution_horizon str]
226
+ [--policy.rtc_config.debug str]
227
+ [--policy.rtc_config.debug_maxlen str]
228
+ [--policy.compile_model str] [--policy.compile_mode str]
229
+ [--policy.n_action_repeats str] [--policy.horizon str]
230
+ [--policy.q_ensemble_size str] [--policy.mlp_dim str]
231
+ [--policy.use_mpc str] [--policy.cem_iterations str]
232
+ [--policy.max_std str] [--policy.min_std str]
233
+ [--policy.n_gaussian_samples str]
234
+ [--policy.n_pi_samples str]
235
+ [--policy.uncertainty_regularizer_coeff str]
236
+ [--policy.n_elites str]
237
+ [--policy.elite_weighting_temperature str]
238
+ [--policy.gaussian_mean_momentum str]
239
+ [--policy.max_random_shift_ratio str]
240
+ [--policy.reward_coeff str]
241
+ [--policy.expectile_weight str]
242
+ [--policy.value_coeff str]
243
+ [--policy.consistency_coeff str]
244
+ [--policy.advantage_scaling str] [--policy.pi_coeff str]
245
+ [--policy.temporal_decay_coeff str]
246
+ [--policy.target_model_momentum str]
247
+ [--policy.n_action_pred_token str]
248
+ [--policy.action_chunk_size str]
249
+ [--policy.vision_backbone str] [--policy.crop_shape str]
250
+ [--policy.crop_is_random str]
251
+ [--policy.pretrained_backbone_weights str]
252
+ [--policy.use_group_norm str]
253
+ [--policy.spatial_softmax_num_keypoints str]
254
+ [--policy.n_vqvae_training_steps str]
255
+ [--policy.vqvae_n_embed str]
256
+ [--policy.vqvae_embedding_dim str]
257
+ [--policy.vqvae_enc_hidden_dim str]
258
+ [--policy.gpt_block_size str]
259
+ [--policy.gpt_input_dim str]
260
+ [--policy.gpt_output_dim str] [--policy.gpt_n_layer str]
261
+ [--policy.gpt_n_head str] [--policy.gpt_hidden_dim str]
262
+ [--policy.offset_loss_weight str]
263
+ [--policy.primary_code_loss_weight str]
264
+ [--policy.secondary_code_loss_weight str]
265
+ [--policy.bet_softmax_temperature str]
266
+ [--policy.sequentially_select str]
267
+ [--policy.optimizer_vqvae_lr str]
268
+ [--policy.optimizer_vqvae_weight_decay str]
269
+ [--policy.pretrained_name_or_path str]
270
+ [--policy.action_tokenizer_path str]
271
+ [--policy.prediction_mode str]
272
+ [--policy.attn_implementation str]
273
+ [--policy.chunk_size str] [--policy.n_action_steps str]
274
+ [--policy.dtype str] [--policy.florence_config str]
275
+ [--policy.tokenizer_name str]
276
+ [--policy.tokenizer_max_length str]
277
+ [--policy.tokenizer_padding_side str]
278
+ [--policy.pad_language_to str] [--policy.hidden_size str]
279
+ [--policy.depth str] [--policy.mlp_ratio str]
280
+ [--policy.num_domains str]
281
+ [--policy.len_soft_prompts str] [--policy.dim_time str]
282
+ [--policy.max_len_seq str] [--policy.use_hetero_proj str]
283
+ [--policy.action_mode str]
284
+ [--policy.num_denoising_steps str]
285
+ [--policy.use_proprio str] [--policy.max_action_dim str]
286
+ [--policy.domain_feature_key str]
287
+ [--policy.resize_imgs_with_padding str]
288
+ [--policy.num_image_views str]
289
+ [--policy.empty_cameras str]
290
+ [--policy.freeze_language_encoder str]
291
+ [--policy.train_policy_transformer str]
292
+ [--policy.train_soft_prompts str]
293
+ [--policy.optimizer_lr str]
294
+ [--policy.optimizer_betas str]
295
+ [--policy.optimizer_eps str]
296
+ [--policy.optimizer_weight_decay str]
297
+ [--policy.optimizer_grad_clip_norm str]
298
+ [--policy.optimizer_soft_prompt_lr_scale str]
299
+ [--policy.optimizer_soft_prompt_warmup_lr_scale str]
300
+ [--policy.scheduler_warmup_steps str]
301
+ [--policy.scheduler_decay_steps str]
302
+ [--policy.scheduler_decay_lr str]
303
+ [--policy.dataset_stats str]
304
+ [--policy.storage_device str]
305
+ [--policy.vision_encoder_name str]
306
+ [--policy.freeze_vision_encoder str]
307
+ [--policy.image_encoder_hidden_dim str]
308
+ [--policy.shared_encoder str]
309
+ [--policy.num_discrete_actions str]
310
+ [--policy.online_steps str]
311
+ [--policy.online_buffer_capacity str]
312
+ [--policy.offline_buffer_capacity str]
313
+ [--policy.async_prefetch str]
314
+ [--policy.online_step_before_learning str]
315
+ [--policy.policy_update_freq str] [--policy.discount str]
316
+ [--policy.temperature_init str]
317
+ [--policy.num_critics str]
318
+ [--policy.num_subsample_critics str]
319
+ [--policy.critic_lr str] [--policy.actor_lr str]
320
+ [--policy.temperature_lr str]
321
+ [--policy.critic_target_update_weight str]
322
+ [--policy.utd_ratio str]
323
+ [--policy.state_encoder_hidden_dim str]
324
+ [--policy.target_entropy str]
325
+ [--policy.use_backup_entropy str]
326
+ [--critic_network_kwargs str]
327
+ [--policy.critic_network_kwargs.hidden_dims str]
328
+ [--policy.critic_network_kwargs.activate_final str]
329
+ [--policy.critic_network_kwargs.final_activation str]
330
+ [--actor_network_kwargs str]
331
+ [--policy.actor_network_kwargs.hidden_dims str]
332
+ [--policy.actor_network_kwargs.activate_final str]
333
+ [--policy_kwargs str]
334
+ [--policy.policy_kwargs.use_tanh_squash str]
335
+ [--policy.policy_kwargs.std_min str]
336
+ [--policy.policy_kwargs.std_max str]
337
+ [--policy.policy_kwargs.init_final str]
338
+ [--discrete_critic_network_kwargs str]
339
+ [--policy.discrete_critic_network_kwargs.hidden_dims str]
340
+ [--policy.discrete_critic_network_kwargs.activate_final str]
341
+ [--policy.discrete_critic_network_kwargs.final_activation str]
342
+ [--actor_learner_config str]
343
+ [--policy.actor_learner_config.learner_host str]
344
+ [--policy.actor_learner_config.learner_port str]
345
+ [--policy.actor_learner_config.policy_parameters_push_frequency str]
346
+ [--policy.actor_learner_config.queue_get_timeout str]
347
+ [--concurrency str] [--policy.concurrency.actor str]
348
+ [--policy.concurrency.learner str]
349
+ [--policy.use_torch_compile str] [--policy.name str]
350
+ [--policy.num_classes str] [--policy.latent_dim str]
351
+ [--policy.image_embedding_pooling_dim str]
352
+ [--policy.dropout_rate str] [--policy.model_name str]
353
+ [--policy.model_type str] [--policy.num_cameras str]
354
+ [--policy.learning_rate str] [--policy.weight_decay str]
355
+ [--policy.grad_clip_norm str] [--policy.n_obs_steps str]
356
+ [--policy.input_features str]
357
+ [--policy.output_features str] [--policy.device str]
358
+ [--policy.use_amp str] [--policy.use_peft str]
359
+ [--policy.push_to_hub str] [--policy.repo_id str]
360
+ [--policy.private str] [--policy.tags str]
361
+ [--policy.license str] [--policy.pretrained_path str]
362
+ [--policy.annotation_mode str] [--policy.frame_gap str]
363
+ [--policy.max_rewind_steps str] [--policy.image_dim str]
364
+ [--policy.text_dim str] [--policy.hidden_dim str]
365
+ [--policy.num_heads str] [--policy.num_layers str]
366
+ [--policy.max_state_dim str]
367
+ [--policy.drop_n_last_frames str]
368
+ [--policy.batch_size str] [--policy.clip_batch_size str]
369
+ [--policy.dropout str] [--policy.stage_loss_weight str]
370
+ [--policy.rewind_probability str]
371
+ [--policy.language_perturbation_probability str]
372
+ [--policy.num_sparse_stages str]
373
+ [--policy.sparse_subtask_names str]
374
+ [--policy.sparse_temporal_proportions str]
375
+ [--policy.num_dense_stages str]
376
+ [--policy.dense_subtask_names str]
377
+ [--policy.dense_temporal_proportions str]
378
+ [--policy.pretrained_model_path str]
379
+ [--policy.image_key str] [--policy.state_key str]
380
+ [--policy.normalization_mapping str] [--output_dir str]
381
+ [--job_name str] [--resume str] [--seed str]
382
+ [--num_workers str] [--batch_size str] [--steps str]
383
+ [--eval_freq str] [--log_freq str] [--tolerance_s str]
384
+ [--save_checkpoint str] [--save_freq str]
385
+ [--use_policy_training_preset str] [--optimizer str]
386
+ [--optimizer.type {adam,adamw,sgd,xvla-adamw,multi_adam}]
387
+ [--optimizer.momentum str] [--optimizer.dampening str]
388
+ [--optimizer.nesterov str] [--optimizer.betas str]
389
+ [--optimizer.eps str]
390
+ [--optimizer.soft_prompt_lr_scale str]
391
+ [--optimizer.soft_prompt_warmup_lr_scale str]
392
+ [--optimizer.lr str] [--optimizer.weight_decay str]
393
+ [--optimizer.grad_clip_norm str]
394
+ [--optimizer.optimizer_groups str] [--scheduler str]
395
+ [--scheduler.type {diffuser,vqbet,cosine_decay_with_warmup}]
396
+ [--scheduler.name str]
397
+ [--scheduler.num_vqvae_training_steps str]
398
+ [--scheduler.num_cycles str]
399
+ [--scheduler.num_warmup_steps str]
400
+ [--scheduler.num_decay_steps str]
401
+ [--scheduler.peak_lr str] [--scheduler.decay_lr str]
402
+ [--eval str] [--eval.n_episodes str]
403
+ [--eval.batch_size str] [--eval.use_async_envs str]
404
+ [--wandb str] [--wandb.enable str]
405
+ [--wandb.disable_artifact str] [--wandb.project str]
406
+ [--wandb.entity str] [--wandb.notes str]
407
+ [--wandb.run_id str] [--wandb.mode str] [--peft str]
408
+ [--peft.target_modules str]
409
+ [--peft.full_training_modules str]
410
+ [--peft.method_type str] [--peft.init_type str]
411
+ [--peft.r str] [--use_rabc str]
412
+ [--rabc_progress_path str] [--rabc_kappa str]
413
+ [--rabc_epsilon str] [--rabc_head_mode str]
414
+ [--rename_map str]
415
+ lerobot_train: error: unrecognized arguments: --dataset.tolerance_s=2.0
416
+ usage: lerobot_train [-h] [--config_path str] [--dataset str]
417
+ [--dataset.repo_id str] [--dataset.root str]
418
+ [--dataset.episodes str] [--image_transforms str]
419
+ [--dataset.image_transforms.enable str]
420
+ [--dataset.image_transforms.max_num_transforms str]
421
+ [--dataset.image_transforms.random_order str]
422
+ [--dataset.image_transforms.tfs str]
423
+ [--dataset.revision str]
424
+ [--dataset.use_imagenet_stats str]
425
+ [--dataset.video_backend str] [--dataset.streaming str]
426
+ [--env str]
427
+ [--env.type {aloha,pusht,gym_manipulator,libero,metaworld,isaaclab_arena}]
428
+ [--env.visualization_width str]
429
+ [--env.visualization_height str] [--robot str]
430
+ [--env.robot.type str] [--teleop str]
431
+ [--env.teleop.type str] [--processor str]
432
+ [--env.processor.control_mode str] [--observation str]
433
+ [--env.processor.observation.add_joint_velocity_to_observation str]
434
+ [--env.processor.observation.add_current_to_observation str]
435
+ [--env.processor.observation.add_ee_pose_to_observation str]
436
+ [--env.processor.observation.display_cameras str]
437
+ [--image_preprocessing str]
438
+ [--env.processor.image_preprocessing.crop_params_dict str]
439
+ [--env.processor.image_preprocessing.resize_size str]
440
+ [--gripper str] [--env.processor.gripper.use_gripper str]
441
+ [--env.processor.gripper.gripper_penalty str]
442
+ [--reset str]
443
+ [--env.processor.reset.fixed_reset_joint_positions str]
444
+ [--env.processor.reset.reset_time_s str]
445
+ [--env.processor.reset.control_time_s str]
446
+ [--env.processor.reset.terminate_on_success str]
447
+ [--inverse_kinematics str]
448
+ [--env.processor.inverse_kinematics.urdf_path str]
449
+ [--env.processor.inverse_kinematics.target_frame_name str]
450
+ [--env.processor.inverse_kinematics.end_effector_bounds str]
451
+ [--env.processor.inverse_kinematics.end_effector_step_sizes str]
452
+ [--reward_classifier str]
453
+ [--env.processor.reward_classifier.pretrained_path str]
454
+ [--env.processor.reward_classifier.success_threshold str]
455
+ [--env.processor.reward_classifier.success_reward str]
456
+ [--env.processor.max_gripper_pos str] [--env.name str]
457
+ [--env.task_ids str] [--env.camera_name str]
458
+ [--env.init_states str] [--env.camera_name_mapping str]
459
+ [--env.observation_height str]
460
+ [--env.observation_width str] [--env.control_mode str]
461
+ [--env.obs_type str] [--env.render_mode str]
462
+ [--env.multitask_eval str] [--env.task str]
463
+ [--env.fps str] [--env.features str]
464
+ [--env.features_map str] [--env.max_parallel_tasks str]
465
+ [--env.disable_env_checker str] [--env.hub_path str]
466
+ [--env.episode_length str] [--env.num_envs str]
467
+ [--env.embodiment str] [--env.object str]
468
+ [--env.mimic str] [--env.teleop_device str]
469
+ [--env.seed str] [--env.device str]
470
+ [--env.disable_fabric str] [--env.enable_cameras str]
471
+ [--env.headless str] [--env.enable_pinocchio str]
472
+ [--env.environment str] [--env.state_dim str]
473
+ [--env.action_dim str] [--env.camera_height str]
474
+ [--env.camera_width str] [--env.video str]
475
+ [--env.video_length str] [--env.video_interval str]
476
+ [--env.state_keys str] [--env.camera_keys str]
477
+ [--env.kwargs str] [--policy str]
478
+ [--policy.type {act,diffusion,groot,pi0,pi0_fast,pi05,smolvla,tdmpc,vqbet,wall_x,xvla,sac,reward_classifier,sarm}]
479
+ [--policy.replace_final_stride_with_dilation str]
480
+ [--policy.pre_norm str] [--policy.dim_model str]
481
+ [--policy.n_heads str] [--policy.dim_feedforward str]
482
+ [--policy.feedforward_activation str]
483
+ [--policy.n_encoder_layers str]
484
+ [--policy.n_decoder_layers str] [--policy.use_vae str]
485
+ [--policy.n_vae_encoder_layers str]
486
+ [--policy.temporal_ensemble_coeff str]
487
+ [--policy.kl_weight str]
488
+ [--policy.optimizer_lr_backbone str]
489
+ [--policy.resize_shape str] [--policy.crop_ratio str]
490
+ [--policy.use_separate_rgb_encoder_per_camera str]
491
+ [--policy.down_dims str] [--policy.kernel_size str]
492
+ [--policy.n_groups str]
493
+ [--policy.diffusion_step_embed_dim str]
494
+ [--policy.use_film_scale_modulation str]
495
+ [--policy.noise_scheduler_type str]
496
+ [--policy.num_train_timesteps str]
497
+ [--policy.beta_schedule str] [--policy.beta_start str]
498
+ [--policy.beta_end str] [--policy.prediction_type str]
499
+ [--policy.clip_sample str]
500
+ [--policy.clip_sample_range str]
501
+ [--policy.do_mask_loss_for_padding str]
502
+ [--policy.scheduler_name str] [--policy.image_size str]
503
+ [--policy.base_model_path str]
504
+ [--policy.tokenizer_assets_repo str]
505
+ [--policy.embodiment_tag str] [--policy.tune_llm str]
506
+ [--policy.tune_visual str] [--policy.tune_projector str]
507
+ [--policy.tune_diffusion_model str]
508
+ [--policy.lora_rank str] [--policy.lora_alpha str]
509
+ [--policy.lora_dropout str]
510
+ [--policy.lora_full_model str]
511
+ [--policy.warmup_ratio str] [--policy.use_bf16 str]
512
+ [--policy.video_backend str]
513
+ [--policy.balance_dataset_weights str]
514
+ [--policy.balance_trajectory_weights str]
515
+ [--policy.dataset_paths str] [--policy.output_dir str]
516
+ [--policy.save_steps str] [--policy.max_steps str]
517
+ [--policy.dataloader_num_workers str]
518
+ [--policy.report_to str] [--policy.resume str]
519
+ [--policy.max_action_tokens str]
520
+ [--policy.text_tokenizer_name str]
521
+ [--policy.action_tokenizer_name str]
522
+ [--policy.temperature str]
523
+ [--policy.max_decoding_steps str]
524
+ [--policy.fast_skip_tokens str]
525
+ [--policy.validate_action_token_prefix str]
526
+ [--policy.use_kv_cache str]
527
+ [--policy.paligemma_variant str]
528
+ [--policy.action_expert_variant str]
529
+ [--policy.num_inference_steps str]
530
+ [--policy.time_sampling_beta_alpha str]
531
+ [--policy.time_sampling_beta_beta str]
532
+ [--policy.time_sampling_scale str]
533
+ [--policy.time_sampling_offset str]
534
+ [--policy.image_resolution str]
535
+ [--policy.gradient_checkpointing str]
536
+ [--policy.adapt_to_pi_aloha str]
537
+ [--policy.use_delta_joint_actions_aloha str]
538
+ [--policy.num_steps str] [--policy.use_cache str]
539
+ [--policy.train_expert_only str]
540
+ [--policy.train_state_proj str]
541
+ [--policy.vlm_model_name str]
542
+ [--policy.load_vlm_weights str]
543
+ [--policy.add_image_special_tokens str]
544
+ [--policy.attention_mode str]
545
+ [--policy.prefix_length str]
546
+ [--policy.num_expert_layers str]
547
+ [--policy.num_vlm_layers str]
548
+ [--policy.self_attn_every_n_layers str]
549
+ [--policy.expert_width_multiplier str]
550
+ [--policy.min_period str] [--policy.max_period str]
551
+ [--rtc_config str] [--policy.rtc_config.enabled str]
552
+ [--policy.rtc_config.prefix_attention_schedule str]
553
+ [--policy.rtc_config.max_guidance_weight str]
554
+ [--policy.rtc_config.execution_horizon str]
555
+ [--policy.rtc_config.debug str]
556
+ [--policy.rtc_config.debug_maxlen str]
557
+ [--policy.compile_model str] [--policy.compile_mode str]
558
+ [--policy.n_action_repeats str] [--policy.horizon str]
559
+ [--policy.q_ensemble_size str] [--policy.mlp_dim str]
560
+ [--policy.use_mpc str] [--policy.cem_iterations str]
561
+ [--policy.max_std str] [--policy.min_std str]
562
+ [--policy.n_gaussian_samples str]
563
+ [--policy.n_pi_samples str]
564
+ [--policy.uncertainty_regularizer_coeff str]
565
+ [--policy.n_elites str]
566
+ [--policy.elite_weighting_temperature str]
567
+ [--policy.gaussian_mean_momentum str]
568
+ [--policy.max_random_shift_ratio str]
569
+ [--policy.reward_coeff str]
570
+ [--policy.expectile_weight str]
571
+ [--policy.value_coeff str]
572
+ [--policy.consistency_coeff str]
573
+ [--policy.advantage_scaling str] [--policy.pi_coeff str]
574
+ [--policy.temporal_decay_coeff str]
575
+ [--policy.target_model_momentum str]
576
+ [--policy.n_action_pred_token str]
577
+ [--policy.action_chunk_size str]
578
+ [--policy.vision_backbone str] [--policy.crop_shape str]
579
+ [--policy.crop_is_random str]
580
+ [--policy.pretrained_backbone_weights str]
581
+ [--policy.use_group_norm str]
582
+ [--policy.spatial_softmax_num_keypoints str]
583
+ [--policy.n_vqvae_training_steps str]
584
+ [--policy.vqvae_n_embed str]
585
+ [--policy.vqvae_embedding_dim str]
586
+ [--policy.vqvae_enc_hidden_dim str]
587
+ [--policy.gpt_block_size str]
588
+ [--policy.gpt_input_dim str]
589
+ [--policy.gpt_output_dim str] [--policy.gpt_n_layer str]
590
+ [--policy.gpt_n_head str] [--policy.gpt_hidden_dim str]
591
+ [--policy.offset_loss_weight str]
592
+ [--policy.primary_code_loss_weight str]
593
+ [--policy.secondary_code_loss_weight str]
594
+ [--policy.bet_softmax_temperature str]
595
+ [--policy.sequentially_select str]
596
+ [--policy.optimizer_vqvae_lr str]
597
+ [--policy.optimizer_vqvae_weight_decay str]
598
+ [--policy.pretrained_name_or_path str]
599
+ [--policy.action_tokenizer_path str]
600
+ [--policy.prediction_mode str]
601
+ [--policy.attn_implementation str]
602
+ [--policy.chunk_size str] [--policy.n_action_steps str]
603
+ [--policy.dtype str] [--policy.florence_config str]
604
+ [--policy.tokenizer_name str]
605
+ [--policy.tokenizer_max_length str]
606
+ [--policy.tokenizer_padding_side str]
607
+ [--policy.pad_language_to str] [--policy.hidden_size str]
608
+ [--policy.depth str] [--policy.mlp_ratio str]
609
+ [--policy.num_domains str]
610
+ [--policy.len_soft_prompts str] [--policy.dim_time str]
611
+ [--policy.max_len_seq str] [--policy.use_hetero_proj str]
612
+ [--policy.action_mode str]
613
+ [--policy.num_denoising_steps str]
614
+ [--policy.use_proprio str] [--policy.max_action_dim str]
615
+ [--policy.domain_feature_key str]
616
+ [--policy.resize_imgs_with_padding str]
617
+ [--policy.num_image_views str]
618
+ [--policy.empty_cameras str]
619
+ [--policy.freeze_language_encoder str]
620
+ [--policy.train_policy_transformer str]
621
+ [--policy.train_soft_prompts str]
622
+ [--policy.optimizer_lr str]
623
+ [--policy.optimizer_betas str]
624
+ [--policy.optimizer_eps str]
625
+ [--policy.optimizer_weight_decay str]
626
+ [--policy.optimizer_grad_clip_norm str]
627
+ [--policy.optimizer_soft_prompt_lr_scale str]
628
+ [--policy.optimizer_soft_prompt_warmup_lr_scale str]
629
+ [--policy.scheduler_warmup_steps str]
630
+ [--policy.scheduler_decay_steps str]
631
+ [--policy.scheduler_decay_lr str]
632
+ [--policy.dataset_stats str]
633
+ [--policy.storage_device str]
634
+ [--policy.vision_encoder_name str]
635
+ [--policy.freeze_vision_encoder str]
636
+ [--policy.image_encoder_hidden_dim str]
637
+ [--policy.shared_encoder str]
638
+ [--policy.num_discrete_actions str]
639
+ [--policy.online_steps str]
640
+ [--policy.online_buffer_capacity str]
641
+ [--policy.offline_buffer_capacity str]
642
+ [--policy.async_prefetch str]
643
+ [--policy.online_step_before_learning str]
644
+ [--policy.policy_update_freq str] [--policy.discount str]
645
+ [--policy.temperature_init str]
646
+ [--policy.num_critics str]
647
+ [--policy.num_subsample_critics str]
648
+ [--policy.critic_lr str] [--policy.actor_lr str]
649
+ [--policy.temperature_lr str]
650
+ [--policy.critic_target_update_weight str]
651
+ [--policy.utd_ratio str]
652
+ [--policy.state_encoder_hidden_dim str]
653
+ [--policy.target_entropy str]
654
+ [--policy.use_backup_entropy str]
655
+ [--critic_network_kwargs str]
656
+ [--policy.critic_network_kwargs.hidden_dims str]
657
+ [--policy.critic_network_kwargs.activate_final str]
658
+ [--policy.critic_network_kwargs.final_activation str]
659
+ [--actor_network_kwargs str]
660
+ [--policy.actor_network_kwargs.hidden_dims str]
661
+ [--policy.actor_network_kwargs.activate_final str]
662
+ [--policy_kwargs str]
663
+ [--policy.policy_kwargs.use_tanh_squash str]
664
+ [--policy.policy_kwargs.std_min str]
665
+ [--policy.policy_kwargs.std_max str]
666
+ [--policy.policy_kwargs.init_final str]
667
+ [--discrete_critic_network_kwargs str]
668
+ [--policy.discrete_critic_network_kwargs.hidden_dims str]
669
+ [--policy.discrete_critic_network_kwargs.activate_final str]
670
+ [--policy.discrete_critic_network_kwargs.final_activation str]
671
+ [--actor_learner_config str]
672
+ [--policy.actor_learner_config.learner_host str]
673
+ [--policy.actor_learner_config.learner_port str]
674
+ [--policy.actor_learner_config.policy_parameters_push_frequency str]
675
+ [--policy.actor_learner_config.queue_get_timeout str]
676
+ [--concurrency str] [--policy.concurrency.actor str]
677
+ [--policy.concurrency.learner str]
678
+ [--policy.use_torch_compile str] [--policy.name str]
679
+ [--policy.num_classes str] [--policy.latent_dim str]
680
+ [--policy.image_embedding_pooling_dim str]
681
+ [--policy.dropout_rate str] [--policy.model_name str]
682
+ [--policy.model_type str] [--policy.num_cameras str]
683
+ [--policy.learning_rate str] [--policy.weight_decay str]
684
+ [--policy.grad_clip_norm str] [--policy.n_obs_steps str]
685
+ [--policy.input_features str]
686
+ [--policy.output_features str] [--policy.device str]
687
+ [--policy.use_amp str] [--policy.use_peft str]
688
+ [--policy.push_to_hub str] [--policy.repo_id str]
689
+ [--policy.private str] [--policy.tags str]
690
+ [--policy.license str] [--policy.pretrained_path str]
691
+ [--policy.annotation_mode str] [--policy.frame_gap str]
692
+ [--policy.max_rewind_steps str] [--policy.image_dim str]
693
+ [--policy.text_dim str] [--policy.hidden_dim str]
694
+ [--policy.num_heads str] [--policy.num_layers str]
695
+ [--policy.max_state_dim str]
696
+ [--policy.drop_n_last_frames str]
697
+ [--policy.batch_size str] [--policy.clip_batch_size str]
698
+ [--policy.dropout str] [--policy.stage_loss_weight str]
699
+ [--policy.rewind_probability str]
700
+ [--policy.language_perturbation_probability str]
701
+ [--policy.num_sparse_stages str]
702
+ [--policy.sparse_subtask_names str]
703
+ [--policy.sparse_temporal_proportions str]
704
+ [--policy.num_dense_stages str]
705
+ [--policy.dense_subtask_names str]
706
+ [--policy.dense_temporal_proportions str]
707
+ [--policy.pretrained_model_path str]
708
+ [--policy.image_key str] [--policy.state_key str]
709
+ [--policy.normalization_mapping str] [--output_dir str]
710
+ [--job_name str] [--resume str] [--seed str]
711
+ [--num_workers str] [--batch_size str] [--steps str]
712
+ [--eval_freq str] [--log_freq str] [--tolerance_s str]
713
+ [--save_checkpoint str] [--save_freq str]
714
+ [--use_policy_training_preset str] [--optimizer str]
715
+ [--optimizer.type {adam,adamw,sgd,xvla-adamw,multi_adam}]
716
+ [--optimizer.momentum str] [--optimizer.dampening str]
717
+ [--optimizer.nesterov str] [--optimizer.betas str]
718
+ [--optimizer.eps str]
719
+ [--optimizer.soft_prompt_lr_scale str]
720
+ [--optimizer.soft_prompt_warmup_lr_scale str]
721
+ [--optimizer.lr str] [--optimizer.weight_decay str]
722
+ [--optimizer.grad_clip_norm str]
723
+ [--optimizer.optimizer_groups str] [--scheduler str]
724
+ [--scheduler.type {diffuser,vqbet,cosine_decay_with_warmup}]
725
+ [--scheduler.name str]
726
+ [--scheduler.num_vqvae_training_steps str]
727
+ [--scheduler.num_cycles str]
728
+ [--scheduler.num_warmup_steps str]
729
+ [--scheduler.num_decay_steps str]
730
+ [--scheduler.peak_lr str] [--scheduler.decay_lr str]
731
+ [--eval str] [--eval.n_episodes str]
732
+ [--eval.batch_size str] [--eval.use_async_envs str]
733
+ [--wandb str] [--wandb.enable str]
734
+ [--wandb.disable_artifact str] [--wandb.project str]
735
+ [--wandb.entity str] [--wandb.notes str]
736
+ [--wandb.run_id str] [--wandb.mode str] [--peft str]
737
+ [--peft.target_modules str]
738
+ [--peft.full_training_modules str]
739
+ [--peft.method_type str] [--peft.init_type str]
740
+ [--peft.r str] [--use_rabc str]
741
+ [--rabc_progress_path str] [--rabc_kappa str]
742
+ [--rabc_epsilon str] [--rabc_head_mode str]
743
+ [--rename_map str]
744
+ lerobot_train: error: unrecognized arguments: --dataset.tolerance_s=2.0
745
+ usage: lerobot_train [-h] [--config_path str] [--dataset str]
746
+ [--dataset.repo_id str] [--dataset.root str]
747
+ [--dataset.episodes str] [--image_transforms str]
748
+ [--dataset.image_transforms.enable str]
749
+ [--dataset.image_transforms.max_num_transforms str]
750
+ [--dataset.image_transforms.random_order str]
751
+ [--dataset.image_transforms.tfs str]
752
+ [--dataset.revision str]
753
+ [--dataset.use_imagenet_stats str]
754
+ [--dataset.video_backend str] [--dataset.streaming str]
755
+ [--env str]
756
+ [--env.type {aloha,pusht,gym_manipulator,libero,metaworld,isaaclab_arena}]
757
+ [--env.visualization_width str]
758
+ [--env.visualization_height str] [--robot str]
759
+ [--env.robot.type str] [--teleop str]
760
+ [--env.teleop.type str] [--processor str]
761
+ [--env.processor.control_mode str] [--observation str]
762
+ [--env.processor.observation.add_joint_velocity_to_observation str]
763
+ [--env.processor.observation.add_current_to_observation str]
764
+ [--env.processor.observation.add_ee_pose_to_observation str]
765
+ [--env.processor.observation.display_cameras str]
766
+ [--image_preprocessing str]
767
+ [--env.processor.image_preprocessing.crop_params_dict str]
768
+ [--env.processor.image_preprocessing.resize_size str]
769
+ [--gripper str] [--env.processor.gripper.use_gripper str]
770
+ [--env.processor.gripper.gripper_penalty str]
771
+ [--reset str]
772
+ [--env.processor.reset.fixed_reset_joint_positions str]
773
+ [--env.processor.reset.reset_time_s str]
774
+ [--env.processor.reset.control_time_s str]
775
+ [--env.processor.reset.terminate_on_success str]
776
+ [--inverse_kinematics str]
777
+ [--env.processor.inverse_kinematics.urdf_path str]
778
+ [--env.processor.inverse_kinematics.target_frame_name str]
779
+ [--env.processor.inverse_kinematics.end_effector_bounds str]
780
+ [--env.processor.inverse_kinematics.end_effector_step_sizes str]
781
+ [--reward_classifier str]
782
+ [--env.processor.reward_classifier.pretrained_path str]
783
+ [--env.processor.reward_classifier.success_threshold str]
784
+ [--env.processor.reward_classifier.success_reward str]
785
+ [--env.processor.max_gripper_pos str] [--env.name str]
786
+ [--env.task_ids str] [--env.camera_name str]
787
+ [--env.init_states str] [--env.camera_name_mapping str]
788
+ [--env.observation_height str]
789
+ [--env.observation_width str] [--env.control_mode str]
790
+ [--env.obs_type str] [--env.render_mode str]
791
+ [--env.multitask_eval str] [--env.task str]
792
+ [--env.fps str] [--env.features str]
793
+ [--env.features_map str] [--env.max_parallel_tasks str]
794
+ [--env.disable_env_checker str] [--env.hub_path str]
795
+ [--env.episode_length str] [--env.num_envs str]
796
+ [--env.embodiment str] [--env.object str]
797
+ [--env.mimic str] [--env.teleop_device str]
798
+ [--env.seed str] [--env.device str]
799
+ [--env.disable_fabric str] [--env.enable_cameras str]
800
+ [--env.headless str] [--env.enable_pinocchio str]
801
+ [--env.environment str] [--env.state_dim str]
802
+ [--env.action_dim str] [--env.camera_height str]
803
+ [--env.camera_width str] [--env.video str]
804
+ [--env.video_length str] [--env.video_interval str]
805
+ [--env.state_keys str] [--env.camera_keys str]
806
+ [--env.kwargs str] [--policy str]
807
+ [--policy.type {act,diffusion,groot,pi0,pi0_fast,pi05,smolvla,tdmpc,vqbet,wall_x,xvla,sac,reward_classifier,sarm}]
808
+ [--policy.replace_final_stride_with_dilation str]
809
+ [--policy.pre_norm str] [--policy.dim_model str]
810
+ [--policy.n_heads str] [--policy.dim_feedforward str]
811
+ [--policy.feedforward_activation str]
812
+ [--policy.n_encoder_layers str]
813
+ [--policy.n_decoder_layers str] [--policy.use_vae str]
814
+ [--policy.n_vae_encoder_layers str]
815
+ [--policy.temporal_ensemble_coeff str]
816
+ [--policy.kl_weight str]
817
+ [--policy.optimizer_lr_backbone str]
818
+ [--policy.resize_shape str] [--policy.crop_ratio str]
819
+ [--policy.use_separate_rgb_encoder_per_camera str]
820
+ [--policy.down_dims str] [--policy.kernel_size str]
821
+ [--policy.n_groups str]
822
+ [--policy.diffusion_step_embed_dim str]
823
+ [--policy.use_film_scale_modulation str]
824
+ [--policy.noise_scheduler_type str]
825
+ [--policy.num_train_timesteps str]
826
+ [--policy.beta_schedule str] [--policy.beta_start str]
827
+ [--policy.beta_end str] [--policy.prediction_type str]
828
+ [--policy.clip_sample str]
829
+ [--policy.clip_sample_range str]
830
+ [--policy.do_mask_loss_for_padding str]
831
+ [--policy.scheduler_name str] [--policy.image_size str]
832
+ [--policy.base_model_path str]
833
+ [--policy.tokenizer_assets_repo str]
834
+ [--policy.embodiment_tag str] [--policy.tune_llm str]
835
+ [--policy.tune_visual str] [--policy.tune_projector str]
836
+ [--policy.tune_diffusion_model str]
837
+ [--policy.lora_rank str] [--policy.lora_alpha str]
838
+ [--policy.lora_dropout str]
839
+ [--policy.lora_full_model str]
840
+ [--policy.warmup_ratio str] [--policy.use_bf16 str]
841
+ [--policy.video_backend str]
842
+ [--policy.balance_dataset_weights str]
843
+ [--policy.balance_trajectory_weights str]
844
+ [--policy.dataset_paths str] [--policy.output_dir str]
845
+ [--policy.save_steps str] [--policy.max_steps str]
846
+ [--policy.dataloader_num_workers str]
847
+ [--policy.report_to str] [--policy.resume str]
848
+ [--policy.max_action_tokens str]
849
+ [--policy.text_tokenizer_name str]
850
+ [--policy.action_tokenizer_name str]
851
+ [--policy.temperature str]
852
+ [--policy.max_decoding_steps str]
853
+ [--policy.fast_skip_tokens str]
854
+ [--policy.validate_action_token_prefix str]
855
+ [--policy.use_kv_cache str]
856
+ [--policy.paligemma_variant str]
857
+ [--policy.action_expert_variant str]
858
+ [--policy.num_inference_steps str]
859
+ [--policy.time_sampling_beta_alpha str]
860
+ [--policy.time_sampling_beta_beta str]
861
+ [--policy.time_sampling_scale str]
862
+ [--policy.time_sampling_offset str]
863
+ [--policy.image_resolution str]
864
+ [--policy.gradient_checkpointing str]
865
+ [--policy.adapt_to_pi_aloha str]
866
+ [--policy.use_delta_joint_actions_aloha str]
867
+ [--policy.num_steps str] [--policy.use_cache str]
868
+ [--policy.train_expert_only str]
869
+ [--policy.train_state_proj str]
870
+ [--policy.vlm_model_name str]
871
+ [--policy.load_vlm_weights str]
872
+ [--policy.add_image_special_tokens str]
873
+ [--policy.attention_mode str]
874
+ [--policy.prefix_length str]
875
+ [--policy.num_expert_layers str]
876
+ [--policy.num_vlm_layers str]
877
+ [--policy.self_attn_every_n_layers str]
878
+ [--policy.expert_width_multiplier str]
879
+ [--policy.min_period str] [--policy.max_period str]
880
+ [--rtc_config str] [--policy.rtc_config.enabled str]
881
+ [--policy.rtc_config.prefix_attention_schedule str]
882
+ [--policy.rtc_config.max_guidance_weight str]
883
+ [--policy.rtc_config.execution_horizon str]
884
+ [--policy.rtc_config.debug str]
885
+ [--policy.rtc_config.debug_maxlen str]
886
+ [--policy.compile_model str] [--policy.compile_mode str]
887
+ [--policy.n_action_repeats str] [--policy.horizon str]
888
+ [--policy.q_ensemble_size str] [--policy.mlp_dim str]
889
+ [--policy.use_mpc str] [--policy.cem_iterations str]
890
+ [--policy.max_std str] [--policy.min_std str]
891
+ [--policy.n_gaussian_samples str]
892
+ [--policy.n_pi_samples str]
893
+ [--policy.uncertainty_regularizer_coeff str]
894
+ [--policy.n_elites str]
895
+ [--policy.elite_weighting_temperature str]
896
+ [--policy.gaussian_mean_momentum str]
897
+ [--policy.max_random_shift_ratio str]
898
+ [--policy.reward_coeff str]
899
+ [--policy.expectile_weight str]
900
+ [--policy.value_coeff str]
901
+ [--policy.consistency_coeff str]
902
+ [--policy.advantage_scaling str] [--policy.pi_coeff str]
903
+ [--policy.temporal_decay_coeff str]
904
+ [--policy.target_model_momentum str]
905
+ [--policy.n_action_pred_token str]
906
+ [--policy.action_chunk_size str]
907
+ [--policy.vision_backbone str] [--policy.crop_shape str]
908
+ [--policy.crop_is_random str]
909
+ [--policy.pretrained_backbone_weights str]
910
+ [--policy.use_group_norm str]
911
+ [--policy.spatial_softmax_num_keypoints str]
912
+ [--policy.n_vqvae_training_steps str]
913
+ [--policy.vqvae_n_embed str]
914
+ [--policy.vqvae_embedding_dim str]
915
+ [--policy.vqvae_enc_hidden_dim str]
916
+ [--policy.gpt_block_size str]
917
+ [--policy.gpt_input_dim str]
918
+ [--policy.gpt_output_dim str] [--policy.gpt_n_layer str]
919
+ [--policy.gpt_n_head str] [--policy.gpt_hidden_dim str]
920
+ [--policy.offset_loss_weight str]
921
+ [--policy.primary_code_loss_weight str]
922
+ [--policy.secondary_code_loss_weight str]
923
+ [--policy.bet_softmax_temperature str]
924
+ [--policy.sequentially_select str]
925
+ [--policy.optimizer_vqvae_lr str]
926
+ [--policy.optimizer_vqvae_weight_decay str]
927
+ [--policy.pretrained_name_or_path str]
928
+ [--policy.action_tokenizer_path str]
929
+ [--policy.prediction_mode str]
930
+ [--policy.attn_implementation str]
931
+ [--policy.chunk_size str] [--policy.n_action_steps str]
932
+ [--policy.dtype str] [--policy.florence_config str]
933
+ [--policy.tokenizer_name str]
934
+ [--policy.tokenizer_max_length str]
935
+ [--policy.tokenizer_padding_side str]
936
+ [--policy.pad_language_to str] [--policy.hidden_size str]
937
+ [--policy.depth str] [--policy.mlp_ratio str]
938
+ [--policy.num_domains str]
939
+ [--policy.len_soft_prompts str] [--policy.dim_time str]
940
+ [--policy.max_len_seq str] [--policy.use_hetero_proj str]
941
+ [--policy.action_mode str]
942
+ [--policy.num_denoising_steps str]
943
+ [--policy.use_proprio str] [--policy.max_action_dim str]
944
+ [--policy.domain_feature_key str]
945
+ [--policy.resize_imgs_with_padding str]
946
+ [--policy.num_image_views str]
947
+ [--policy.empty_cameras str]
948
+ [--policy.freeze_language_encoder str]
949
+ [--policy.train_policy_transformer str]
950
+ [--policy.train_soft_prompts str]
951
+ [--policy.optimizer_lr str]
952
+ [--policy.optimizer_betas str]
953
+ [--policy.optimizer_eps str]
954
+ [--policy.optimizer_weight_decay str]
955
+ [--policy.optimizer_grad_clip_norm str]
956
+ [--policy.optimizer_soft_prompt_lr_scale str]
957
+ [--policy.optimizer_soft_prompt_warmup_lr_scale str]
958
+ [--policy.scheduler_warmup_steps str]
959
+ [--policy.scheduler_decay_steps str]
960
+ [--policy.scheduler_decay_lr str]
961
+ [--policy.dataset_stats str]
962
+ [--policy.storage_device str]
963
+ [--policy.vision_encoder_name str]
964
+ [--policy.freeze_vision_encoder str]
965
+ [--policy.image_encoder_hidden_dim str]
966
+ [--policy.shared_encoder str]
967
+ [--policy.num_discrete_actions str]
968
+ [--policy.online_steps str]
969
+ [--policy.online_buffer_capacity str]
970
+ [--policy.offline_buffer_capacity str]
971
+ [--policy.async_prefetch str]
972
+ [--policy.online_step_before_learning str]
973
+ [--policy.policy_update_freq str] [--policy.discount str]
974
+ [--policy.temperature_init str]
975
+ [--policy.num_critics str]
976
+ [--policy.num_subsample_critics str]
977
+ [--policy.critic_lr str] [--policy.actor_lr str]
978
+ [--policy.temperature_lr str]
979
+ [--policy.critic_target_update_weight str]
980
+ [--policy.utd_ratio str]
981
+ [--policy.state_encoder_hidden_dim str]
982
+ [--policy.target_entropy str]
983
+ [--policy.use_backup_entropy str]
984
+ [--critic_network_kwargs str]
985
+ [--policy.critic_network_kwargs.hidden_dims str]
986
+ [--policy.critic_network_kwargs.activate_final str]
987
+ [--policy.critic_network_kwargs.final_activation str]
988
+ [--actor_network_kwargs str]
989
+ [--policy.actor_network_kwargs.hidden_dims str]
990
+ [--policy.actor_network_kwargs.activate_final str]
991
+ [--policy_kwargs str]
992
+ [--policy.policy_kwargs.use_tanh_squash str]
993
+ [--policy.policy_kwargs.std_min str]
994
+ [--policy.policy_kwargs.std_max str]
995
+ [--policy.policy_kwargs.init_final str]
996
+ [--discrete_critic_network_kwargs str]
997
+ [--policy.discrete_critic_network_kwargs.hidden_dims str]
998
+ [--policy.discrete_critic_network_kwargs.activate_final str]
999
+ [--policy.discrete_critic_network_kwargs.final_activation str]
1000
+ [--actor_learner_config str]
1001
+ [--policy.actor_learner_config.learner_host str]
1002
+ [--policy.actor_learner_config.learner_port str]
1003
+ [--policy.actor_learner_config.policy_parameters_push_frequency str]
1004
+ [--policy.actor_learner_config.queue_get_timeout str]
1005
+ [--concurrency str] [--policy.concurrency.actor str]
1006
+ [--policy.concurrency.learner str]
1007
+ [--policy.use_torch_compile str] [--policy.name str]
1008
+ [--policy.num_classes str] [--policy.latent_dim str]
1009
+ [--policy.image_embedding_pooling_dim str]
1010
+ [--policy.dropout_rate str] [--policy.model_name str]
1011
+ [--policy.model_type str] [--policy.num_cameras str]
1012
+ [--policy.learning_rate str] [--policy.weight_decay str]
1013
+ [--policy.grad_clip_norm str] [--policy.n_obs_steps str]
1014
+ [--policy.input_features str]
1015
+ [--policy.output_features str] [--policy.device str]
1016
+ [--policy.use_amp str] [--policy.use_peft str]
1017
+ [--policy.push_to_hub str] [--policy.repo_id str]
1018
+ [--policy.private str] [--policy.tags str]
1019
+ [--policy.license str] [--policy.pretrained_path str]
1020
+ [--policy.annotation_mode str] [--policy.frame_gap str]
1021
+ [--policy.max_rewind_steps str] [--policy.image_dim str]
1022
+ [--policy.text_dim str] [--policy.hidden_dim str]
1023
+ [--policy.num_heads str] [--policy.num_layers str]
1024
+ [--policy.max_state_dim str]
1025
+ [--policy.drop_n_last_frames str]
1026
+ [--policy.batch_size str] [--policy.clip_batch_size str]
1027
+ [--policy.dropout str] [--policy.stage_loss_weight str]
1028
+ [--policy.rewind_probability str]
1029
+ [--policy.language_perturbation_probability str]
1030
+ [--policy.num_sparse_stages str]
1031
+ [--policy.sparse_subtask_names str]
1032
+ [--policy.sparse_temporal_proportions str]
1033
+ [--policy.num_dense_stages str]
1034
+ [--policy.dense_subtask_names str]
1035
+ [--policy.dense_temporal_proportions str]
1036
+ [--policy.pretrained_model_path str]
1037
+ [--policy.image_key str] [--policy.state_key str]
1038
+ [--policy.normalization_mapping str] [--output_dir str]
1039
+ [--job_name str] [--resume str] [--seed str]
1040
+ [--num_workers str] [--batch_size str] [--steps str]
1041
+ [--eval_freq str] [--log_freq str] [--tolerance_s str]
1042
+ [--save_checkpoint str] [--save_freq str]
1043
+ [--use_policy_training_preset str] [--optimizer str]
1044
+ [--optimizer.type {adam,adamw,sgd,xvla-adamw,multi_adam}]
1045
+ [--optimizer.momentum str] [--optimizer.dampening str]
1046
+ [--optimizer.nesterov str] [--optimizer.betas str]
1047
+ [--optimizer.eps str]
1048
+ [--optimizer.soft_prompt_lr_scale str]
1049
+ [--optimizer.soft_prompt_warmup_lr_scale str]
1050
+ [--optimizer.lr str] [--optimizer.weight_decay str]
1051
+ [--optimizer.grad_clip_norm str]
1052
+ [--optimizer.optimizer_groups str] [--scheduler str]
1053
+ [--scheduler.type {diffuser,vqbet,cosine_decay_with_warmup}]
1054
+ [--scheduler.name str]
1055
+ [--scheduler.num_vqvae_training_steps str]
1056
+ [--scheduler.num_cycles str]
1057
+ [--scheduler.num_warmup_steps str]
1058
+ [--scheduler.num_decay_steps str]
1059
+ [--scheduler.peak_lr str] [--scheduler.decay_lr str]
1060
+ [--eval str] [--eval.n_episodes str]
1061
+ [--eval.batch_size str] [--eval.use_async_envs str]
1062
+ [--wandb str] [--wandb.enable str]
1063
+ [--wandb.disable_artifact str] [--wandb.project str]
1064
+ [--wandb.entity str] [--wandb.notes str]
1065
+ [--wandb.run_id str] [--wandb.mode str] [--peft str]
1066
+ [--peft.target_modules str]
1067
+ [--peft.full_training_modules str]
1068
+ [--peft.method_type str] [--peft.init_type str]
1069
+ [--peft.r str] [--use_rabc str]
1070
+ [--rabc_progress_path str] [--rabc_kappa str]
1071
+ [--rabc_epsilon str] [--rabc_head_mode str]
1072
+ [--rename_map str]
1073
+ lerobot_train: error: unrecognized arguments: --dataset.tolerance_s=2.0
1074
+ usage: lerobot_train [-h] [--config_path str] [--dataset str]
1075
+ [--dataset.repo_id str] [--dataset.root str]
1076
+ [--dataset.episodes str] [--image_transforms str]
1077
+ [--dataset.image_transforms.enable str]
1078
+ [--dataset.image_transforms.max_num_transforms str]
1079
+ [--dataset.image_transforms.random_order str]
1080
+ [--dataset.image_transforms.tfs str]
1081
+ [--dataset.revision str]
1082
+ [--dataset.use_imagenet_stats str]
1083
+ [--dataset.video_backend str] [--dataset.streaming str]
1084
+ [--env str]
1085
+ [--env.type {aloha,pusht,gym_manipulator,libero,metaworld,isaaclab_arena}]
1086
+ [--env.visualization_width str]
1087
+ [--env.visualization_height str] [--robot str]
1088
+ [--env.robot.type str] [--teleop str]
1089
+ [--env.teleop.type str] [--processor str]
1090
+ [--env.processor.control_mode str] [--observation str]
1091
+ [--env.processor.observation.add_joint_velocity_to_observation str]
1092
+ [--env.processor.observation.add_current_to_observation str]
1093
+ [--env.processor.observation.add_ee_pose_to_observation str]
1094
+ [--env.processor.observation.display_cameras str]
1095
+ [--image_preprocessing str]
1096
+ [--env.processor.image_preprocessing.crop_params_dict str]
1097
+ [--env.processor.image_preprocessing.resize_size str]
1098
+ [--gripper str] [--env.processor.gripper.use_gripper str]
1099
+ [--env.processor.gripper.gripper_penalty str]
1100
+ [--reset str]
1101
+ [--env.processor.reset.fixed_reset_joint_positions str]
1102
+ [--env.processor.reset.reset_time_s str]
1103
+ [--env.processor.reset.control_time_s str]
1104
+ [--env.processor.reset.terminate_on_success str]
1105
+ [--inverse_kinematics str]
1106
+ [--env.processor.inverse_kinematics.urdf_path str]
1107
+ [--env.processor.inverse_kinematics.target_frame_name str]
1108
+ [--env.processor.inverse_kinematics.end_effector_bounds str]
1109
+ [--env.processor.inverse_kinematics.end_effector_step_sizes str]
1110
+ [--reward_classifier str]
1111
+ [--env.processor.reward_classifier.pretrained_path str]
1112
+ [--env.processor.reward_classifier.success_threshold str]
1113
+ [--env.processor.reward_classifier.success_reward str]
1114
+ [--env.processor.max_gripper_pos str] [--env.name str]
1115
+ [--env.task_ids str] [--env.camera_name str]
1116
+ [--env.init_states str] [--env.camera_name_mapping str]
1117
+ [--env.observation_height str]
1118
+ [--env.observation_width str] [--env.control_mode str]
1119
+ [--env.obs_type str] [--env.render_mode str]
1120
+ [--env.multitask_eval str] [--env.task str]
1121
+ [--env.fps str] [--env.features str]
1122
+ [--env.features_map str] [--env.max_parallel_tasks str]
1123
+ [--env.disable_env_checker str] [--env.hub_path str]
1124
+ [--env.episode_length str] [--env.num_envs str]
1125
+ [--env.embodiment str] [--env.object str]
1126
+ [--env.mimic str] [--env.teleop_device str]
1127
+ [--env.seed str] [--env.device str]
1128
+ [--env.disable_fabric str] [--env.enable_cameras str]
1129
+ [--env.headless str] [--env.enable_pinocchio str]
1130
+ [--env.environment str] [--env.state_dim str]
1131
+ [--env.action_dim str] [--env.camera_height str]
1132
+ [--env.camera_width str] [--env.video str]
1133
+ [--env.video_length str] [--env.video_interval str]
1134
+ [--env.state_keys str] [--env.camera_keys str]
1135
+ [--env.kwargs str] [--policy str]
1136
+ [--policy.type {act,diffusion,groot,pi0,pi0_fast,pi05,smolvla,tdmpc,vqbet,wall_x,xvla,sac,reward_classifier,sarm}]
1137
+ [--policy.replace_final_stride_with_dilation str]
1138
+ [--policy.pre_norm str] [--policy.dim_model str]
1139
+ [--policy.n_heads str] [--policy.dim_feedforward str]
1140
+ [--policy.feedforward_activation str]
1141
+ [--policy.n_encoder_layers str]
1142
+ [--policy.n_decoder_layers str] [--policy.use_vae str]
1143
+ [--policy.n_vae_encoder_layers str]
1144
+ [--policy.temporal_ensemble_coeff str]
1145
+ [--policy.kl_weight str]
1146
+ [--policy.optimizer_lr_backbone str]
1147
+ [--policy.resize_shape str] [--policy.crop_ratio str]
1148
+ [--policy.use_separate_rgb_encoder_per_camera str]
1149
+ [--policy.down_dims str] [--policy.kernel_size str]
1150
+ [--policy.n_groups str]
1151
+ [--policy.diffusion_step_embed_dim str]
1152
+ [--policy.use_film_scale_modulation str]
1153
+ [--policy.noise_scheduler_type str]
1154
+ [--policy.num_train_timesteps str]
1155
+ [--policy.beta_schedule str] [--policy.beta_start str]
1156
+ [--policy.beta_end str] [--policy.prediction_type str]
1157
+ [--policy.clip_sample str]
1158
+ [--policy.clip_sample_range str]
1159
+ [--policy.do_mask_loss_for_padding str]
1160
+ [--policy.scheduler_name str] [--policy.image_size str]
1161
+ [--policy.base_model_path str]
1162
+ [--policy.tokenizer_assets_repo str]
1163
+ [--policy.embodiment_tag str] [--policy.tune_llm str]
1164
+ [--policy.tune_visual str] [--policy.tune_projector str]
1165
+ [--policy.tune_diffusion_model str]
1166
+ [--policy.lora_rank str] [--policy.lora_alpha str]
1167
+ [--policy.lora_dropout str]
1168
+ [--policy.lora_full_model str]
1169
+ [--policy.warmup_ratio str] [--policy.use_bf16 str]
1170
+ [--policy.video_backend str]
1171
+ [--policy.balance_dataset_weights str]
1172
+ [--policy.balance_trajectory_weights str]
1173
+ [--policy.dataset_paths str] [--policy.output_dir str]
1174
+ [--policy.save_steps str] [--policy.max_steps str]
1175
+ [--policy.dataloader_num_workers str]
1176
+ [--policy.report_to str] [--policy.resume str]
1177
+ [--policy.max_action_tokens str]
1178
+ [--policy.text_tokenizer_name str]
1179
+ [--policy.action_tokenizer_name str]
1180
+ [--policy.temperature str]
1181
+ [--policy.max_decoding_steps str]
1182
+ [--policy.fast_skip_tokens str]
1183
+ [--policy.validate_action_token_prefix str]
1184
+ [--policy.use_kv_cache str]
1185
+ [--policy.paligemma_variant str]
1186
+ [--policy.action_expert_variant str]
1187
+ [--policy.num_inference_steps str]
1188
+ [--policy.time_sampling_beta_alpha str]
1189
+ [--policy.time_sampling_beta_beta str]
1190
+ [--policy.time_sampling_scale str]
1191
+ [--policy.time_sampling_offset str]
1192
+ [--policy.image_resolution str]
1193
+ [--policy.gradient_checkpointing str]
1194
+ [--policy.adapt_to_pi_aloha str]
1195
+ [--policy.use_delta_joint_actions_aloha str]
1196
+ [--policy.num_steps str] [--policy.use_cache str]
1197
+ [--policy.train_expert_only str]
1198
+ [--policy.train_state_proj str]
1199
+ [--policy.vlm_model_name str]
1200
+ [--policy.load_vlm_weights str]
1201
+ [--policy.add_image_special_tokens str]
1202
+ [--policy.attention_mode str]
1203
+ [--policy.prefix_length str]
1204
+ [--policy.num_expert_layers str]
1205
+ [--policy.num_vlm_layers str]
1206
+ [--policy.self_attn_every_n_layers str]
1207
+ [--policy.expert_width_multiplier str]
1208
+ [--policy.min_period str] [--policy.max_period str]
1209
+ [--rtc_config str] [--policy.rtc_config.enabled str]
1210
+ [--policy.rtc_config.prefix_attention_schedule str]
1211
+ [--policy.rtc_config.max_guidance_weight str]
1212
+ [--policy.rtc_config.execution_horizon str]
1213
+ [--policy.rtc_config.debug str]
1214
+ [--policy.rtc_config.debug_maxlen str]
1215
+ [--policy.compile_model str] [--policy.compile_mode str]
1216
+ [--policy.n_action_repeats str] [--policy.horizon str]
1217
+ [--policy.q_ensemble_size str] [--policy.mlp_dim str]
1218
+ [--policy.use_mpc str] [--policy.cem_iterations str]
1219
+ [--policy.max_std str] [--policy.min_std str]
1220
+ [--policy.n_gaussian_samples str]
1221
+ [--policy.n_pi_samples str]
1222
+ [--policy.uncertainty_regularizer_coeff str]
1223
+ [--policy.n_elites str]
1224
+ [--policy.elite_weighting_temperature str]
1225
+ [--policy.gaussian_mean_momentum str]
1226
+ [--policy.max_random_shift_ratio str]
1227
+ [--policy.reward_coeff str]
1228
+ [--policy.expectile_weight str]
1229
+ [--policy.value_coeff str]
1230
+ [--policy.consistency_coeff str]
1231
+ [--policy.advantage_scaling str] [--policy.pi_coeff str]
1232
+ [--policy.temporal_decay_coeff str]
1233
+ [--policy.target_model_momentum str]
1234
+ [--policy.n_action_pred_token str]
1235
+ [--policy.action_chunk_size str]
1236
+ [--policy.vision_backbone str] [--policy.crop_shape str]
1237
+ [--policy.crop_is_random str]
1238
+ [--policy.pretrained_backbone_weights str]
1239
+ [--policy.use_group_norm str]
1240
+ [--policy.spatial_softmax_num_keypoints str]
1241
+ [--policy.n_vqvae_training_steps str]
1242
+ [--policy.vqvae_n_embed str]
1243
+ [--policy.vqvae_embedding_dim str]
1244
+ [--policy.vqvae_enc_hidden_dim str]
1245
+ [--policy.gpt_block_size str]
1246
+ [--policy.gpt_input_dim str]
1247
+ [--policy.gpt_output_dim str] [--policy.gpt_n_layer str]
1248
+ [--policy.gpt_n_head str] [--policy.gpt_hidden_dim str]
1249
+ [--policy.offset_loss_weight str]
1250
+ [--policy.primary_code_loss_weight str]
1251
+ [--policy.secondary_code_loss_weight str]
1252
+ [--policy.bet_softmax_temperature str]
1253
+ [--policy.sequentially_select str]
1254
+ [--policy.optimizer_vqvae_lr str]
1255
+ [--policy.optimizer_vqvae_weight_decay str]
1256
+ [--policy.pretrained_name_or_path str]
1257
+ [--policy.action_tokenizer_path str]
1258
+ [--policy.prediction_mode str]
1259
+ [--policy.attn_implementation str]
1260
+ [--policy.chunk_size str] [--policy.n_action_steps str]
1261
+ [--policy.dtype str] [--policy.florence_config str]
1262
+ [--policy.tokenizer_name str]
1263
+ [--policy.tokenizer_max_length str]
1264
+ [--policy.tokenizer_padding_side str]
1265
+ [--policy.pad_language_to str] [--policy.hidden_size str]
1266
+ [--policy.depth str] [--policy.mlp_ratio str]
1267
+ [--policy.num_domains str]
1268
+ [--policy.len_soft_prompts str] [--policy.dim_time str]
1269
+ [--policy.max_len_seq str] [--policy.use_hetero_proj str]
1270
+ [--policy.action_mode str]
1271
+ [--policy.num_denoising_steps str]
1272
+ [--policy.use_proprio str] [--policy.max_action_dim str]
1273
+ [--policy.domain_feature_key str]
1274
+ [--policy.resize_imgs_with_padding str]
1275
+ [--policy.num_image_views str]
1276
+ [--policy.empty_cameras str]
1277
+ [--policy.freeze_language_encoder str]
1278
+ [--policy.train_policy_transformer str]
1279
+ [--policy.train_soft_prompts str]
1280
+ [--policy.optimizer_lr str]
1281
+ [--policy.optimizer_betas str]
1282
+ [--policy.optimizer_eps str]
1283
+ [--policy.optimizer_weight_decay str]
1284
+ [--policy.optimizer_grad_clip_norm str]
1285
+ [--policy.optimizer_soft_prompt_lr_scale str]
1286
+ [--policy.optimizer_soft_prompt_warmup_lr_scale str]
1287
+ [--policy.scheduler_warmup_steps str]
1288
+ [--policy.scheduler_decay_steps str]
1289
+ [--policy.scheduler_decay_lr str]
1290
+ [--policy.dataset_stats str]
1291
+ [--policy.storage_device str]
1292
+ [--policy.vision_encoder_name str]
1293
+ [--policy.freeze_vision_encoder str]
1294
+ [--policy.image_encoder_hidden_dim str]
1295
+ [--policy.shared_encoder str]
1296
+ [--policy.num_discrete_actions str]
1297
+ [--policy.online_steps str]
1298
+ [--policy.online_buffer_capacity str]
1299
+ [--policy.offline_buffer_capacity str]
1300
+ [--policy.async_prefetch str]
1301
+ [--policy.online_step_before_learning str]
1302
+ [--policy.policy_update_freq str] [--policy.discount str]
1303
+ [--policy.temperature_init str]
1304
+ [--policy.num_critics str]
1305
+ [--policy.num_subsample_critics str]
1306
+ [--policy.critic_lr str] [--policy.actor_lr str]
1307
+ [--policy.temperature_lr str]
1308
+ [--policy.critic_target_update_weight str]
1309
+ [--policy.utd_ratio str]
1310
+ [--policy.state_encoder_hidden_dim str]
1311
+ [--policy.target_entropy str]
1312
+ [--policy.use_backup_entropy str]
1313
+ [--critic_network_kwargs str]
1314
+ [--policy.critic_network_kwargs.hidden_dims str]
1315
+ [--policy.critic_network_kwargs.activate_final str]
1316
+ [--policy.critic_network_kwargs.final_activation str]
1317
+ [--actor_network_kwargs str]
1318
+ [--policy.actor_network_kwargs.hidden_dims str]
1319
+ [--policy.actor_network_kwargs.activate_final str]
1320
+ [--policy_kwargs str]
1321
+ [--policy.policy_kwargs.use_tanh_squash str]
1322
+ [--policy.policy_kwargs.std_min str]
1323
+ [--policy.policy_kwargs.std_max str]
1324
+ [--policy.policy_kwargs.init_final str]
1325
+ [--discrete_critic_network_kwargs str]
1326
+ [--policy.discrete_critic_network_kwargs.hidden_dims str]
1327
+ [--policy.discrete_critic_network_kwargs.activate_final str]
1328
+ [--policy.discrete_critic_network_kwargs.final_activation str]
1329
+ [--actor_learner_config str]
1330
+ [--policy.actor_learner_config.learner_host str]
1331
+ [--policy.actor_learner_config.learner_port str]
1332
+ [--policy.actor_learner_config.policy_parameters_push_frequency str]
1333
+ [--policy.actor_learner_config.queue_get_timeout str]
1334
+ [--concurrency str] [--policy.concurrency.actor str]
1335
+ [--policy.concurrency.learner str]
1336
+ [--policy.use_torch_compile str] [--policy.name str]
1337
+ [--policy.num_classes str] [--policy.latent_dim str]
1338
+ [--policy.image_embedding_pooling_dim str]
1339
+ [--policy.dropout_rate str] [--policy.model_name str]
1340
+ [--policy.model_type str] [--policy.num_cameras str]
1341
+ [--policy.learning_rate str] [--policy.weight_decay str]
1342
+ [--policy.grad_clip_norm str] [--policy.n_obs_steps str]
1343
+ [--policy.input_features str]
1344
+ [--policy.output_features str] [--policy.device str]
1345
+ [--policy.use_amp str] [--policy.use_peft str]
1346
+ [--policy.push_to_hub str] [--policy.repo_id str]
1347
+ [--policy.private str] [--policy.tags str]
1348
+ [--policy.license str] [--policy.pretrained_path str]
1349
+ [--policy.annotation_mode str] [--policy.frame_gap str]
1350
+ [--policy.max_rewind_steps str] [--policy.image_dim str]
1351
+ [--policy.text_dim str] [--policy.hidden_dim str]
1352
+ [--policy.num_heads str] [--policy.num_layers str]
1353
+ [--policy.max_state_dim str]
1354
+ [--policy.drop_n_last_frames str]
1355
+ [--policy.batch_size str] [--policy.clip_batch_size str]
1356
+ [--policy.dropout str] [--policy.stage_loss_weight str]
1357
+ [--policy.rewind_probability str]
1358
+ [--policy.language_perturbation_probability str]
1359
+ [--policy.num_sparse_stages str]
1360
+ [--policy.sparse_subtask_names str]
1361
+ [--policy.sparse_temporal_proportions str]
1362
+ [--policy.num_dense_stages str]
1363
+ [--policy.dense_subtask_names str]
1364
+ [--policy.dense_temporal_proportions str]
1365
+ [--policy.pretrained_model_path str]
1366
+ [--policy.image_key str] [--policy.state_key str]
1367
+ [--policy.normalization_mapping str] [--output_dir str]
1368
+ [--job_name str] [--resume str] [--seed str]
1369
+ [--num_workers str] [--batch_size str] [--steps str]
1370
+ [--eval_freq str] [--log_freq str] [--tolerance_s str]
1371
+ [--save_checkpoint str] [--save_freq str]
1372
+ [--use_policy_training_preset str] [--optimizer str]
1373
+ [--optimizer.type {adam,adamw,sgd,xvla-adamw,multi_adam}]
1374
+ [--optimizer.momentum str] [--optimizer.dampening str]
1375
+ [--optimizer.nesterov str] [--optimizer.betas str]
1376
+ [--optimizer.eps str]
1377
+ [--optimizer.soft_prompt_lr_scale str]
1378
+ [--optimizer.soft_prompt_warmup_lr_scale str]
1379
+ [--optimizer.lr str] [--optimizer.weight_decay str]
1380
+ [--optimizer.grad_clip_norm str]
1381
+ [--optimizer.optimizer_groups str] [--scheduler str]
1382
+ [--scheduler.type {diffuser,vqbet,cosine_decay_with_warmup}]
1383
+ [--scheduler.name str]
1384
+ [--scheduler.num_vqvae_training_steps str]
1385
+ [--scheduler.num_cycles str]
1386
+ [--scheduler.num_warmup_steps str]
1387
+ [--scheduler.num_decay_steps str]
1388
+ [--scheduler.peak_lr str] [--scheduler.decay_lr str]
1389
+ [--eval str] [--eval.n_episodes str]
1390
+ [--eval.batch_size str] [--eval.use_async_envs str]
1391
+ [--wandb str] [--wandb.enable str]
1392
+ [--wandb.disable_artifact str] [--wandb.project str]
1393
+ [--wandb.entity str] [--wandb.notes str]
1394
+ [--wandb.run_id str] [--wandb.mode str] [--peft str]
1395
+ [--peft.target_modules str]
1396
+ [--peft.full_training_modules str]
1397
+ [--peft.method_type str] [--peft.init_type str]
1398
+ [--peft.r str] [--use_rabc str]
1399
+ [--rabc_progress_path str] [--rabc_kappa str]
1400
+ [--rabc_epsilon str] [--rabc_head_mode str]
1401
+ [--rename_map str]
1402
+ lerobot_train: error: unrecognized arguments: --dataset.tolerance_s=2.0
1403
+ W0311 06:29:30.162000 2412 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 2480 closing signal SIGTERM
1404
+ W0311 06:29:30.163000 2412 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 2482 closing signal SIGTERM
1405
+ W0311 06:29:30.163000 2412 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 2483 closing signal SIGTERM
1406
+ E0311 06:29:30.234000 2412 torch/distributed/elastic/multiprocessing/api.py:874] failed (exitcode: 2) local_rank: 1 (pid: 2481) of binary: /usr/bin/python
1407
+ Traceback (most recent call last):
1408
+ File "/usr/local/bin/accelerate", line 8, in <module>
1409
+ sys.exit(main())
1410
+ ^^^^^^
1411
+ File "/usr/local/lib/python3.11/dist-packages/accelerate/commands/accelerate_cli.py", line 50, in main
1412
+ args.func(args)
1413
+ File "/usr/local/lib/python3.11/dist-packages/accelerate/commands/launch.py", line 1396, in launch_command
1414
+ multi_gpu_launcher(args)
1415
+ File "/usr/local/lib/python3.11/dist-packages/accelerate/commands/launch.py", line 1023, in multi_gpu_launcher
1416
+ distrib_run.run(args)
1417
+ File "/usr/local/lib/python3.11/dist-packages/torch/distributed/run.py", line 883, in run
1418
+ elastic_launch(
1419
+ File "/usr/local/lib/python3.11/dist-packages/torch/distributed/launcher/api.py", line 139, in __call__
1420
+ return launch_agent(self._config, self._entrypoint, list(args))
1421
+ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1422
+ File "/usr/local/lib/python3.11/dist-packages/torch/distributed/launcher/api.py", line 270, in launch_agent
1423
+ raise ChildFailedError(
1424
+ torch.distributed.elastic.multiprocessing.errors.ChildFailedError:
1425
+ ============================================================
1426
+ /tmp/train_pi0fast.py FAILED
1427
+ ------------------------------------------------------------
1428
+ Failures:
1429
+ <NO_OTHER_FAILURES>
1430
+ ------------------------------------------------------------
1431
+ Root Cause (first observed failure):
1432
+ [0]:
1433
+ time : 2026-03-11_06:29:30
1434
+ host : 793c3c5b0065
1435
+ rank : 1 (local_rank: 1)
1436
+ exitcode : 2 (pid: 2481)
1437
+ error_file: <N/A>
1438
+ traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
1439
+ ============================================================
1440
+
1441
+ >>> Training finished with exit code 1
1442
+ Ended: 2026-03-11 06:29:30 UTC
1443
+ >>> Saving training log to HuggingFace...