awrenn53 commited on
Commit
95786cc
·
verified ·
1 Parent(s): 5a020e5

Add files using upload-large-folder tool

Browse files
config.json ADDED
@@ -0,0 +1,99 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "type": "groot",
3
+ "n_obs_steps": 1,
4
+ "input_features": {
5
+ "observation.images.wrist_image": {
6
+ "type": "VISUAL",
7
+ "shape": [
8
+ 256,
9
+ 256,
10
+ 3
11
+ ]
12
+ },
13
+ "observation.images.image": {
14
+ "type": "VISUAL",
15
+ "shape": [
16
+ 256,
17
+ 256,
18
+ 3
19
+ ]
20
+ },
21
+ "observation.state": {
22
+ "type": "STATE",
23
+ "shape": [
24
+ 8
25
+ ]
26
+ }
27
+ },
28
+ "output_features": {
29
+ "action": {
30
+ "type": "ACTION",
31
+ "shape": [
32
+ 7
33
+ ]
34
+ }
35
+ },
36
+ "device": "cuda",
37
+ "use_amp": false,
38
+ "use_peft": false,
39
+ "push_to_hub": false,
40
+ "repo_id": null,
41
+ "private": null,
42
+ "tags": null,
43
+ "license": null,
44
+ "pretrained_path": null,
45
+ "pretrained_revision": null,
46
+ "chunk_size": 16,
47
+ "n_action_steps": 16,
48
+ "max_state_dim": 132,
49
+ "max_action_dim": 132,
50
+ "normalization_mapping": {
51
+ "VISUAL": "IDENTITY",
52
+ "STATE": "IDENTITY",
53
+ "ACTION": "IDENTITY"
54
+ },
55
+ "base_model_path": "/cache/huggingface/models--nvidia--GR00T-N1.7-3B/snapshots/2fc962b973bccdd5d8ce4f67cc63b264d6886495",
56
+ "action_decode_transform": "libero",
57
+ "embodiment_tag": "libero_sim",
58
+ "tune_llm": false,
59
+ "tune_visual": false,
60
+ "tune_projector": true,
61
+ "tune_diffusion_model": true,
62
+ "tune_vlln": true,
63
+ "tune_top_llm_layers": 0,
64
+ "num_inference_timesteps": 4,
65
+ "rtc_ramp_rate": 6.0,
66
+ "use_flash_attention": false,
67
+ "use_relative_actions": false,
68
+ "relative_exclude_joints": [],
69
+ "optimizer_lr": 0.0001,
70
+ "optimizer_betas": [
71
+ 0.9,
72
+ 0.999
73
+ ],
74
+ "optimizer_eps": 1e-08,
75
+ "optimizer_weight_decay": 1e-05,
76
+ "warmup_ratio": 0.05,
77
+ "use_bf16": true,
78
+ "model_params_fp32": true,
79
+ "image_size": [
80
+ 256,
81
+ 256
82
+ ],
83
+ "tokenizer_assets_repo": null,
84
+ "lora_rank": 0,
85
+ "lora_alpha": 16,
86
+ "lora_dropout": 0.1,
87
+ "lora_full_model": false,
88
+ "video_backend": "decord",
89
+ "balance_dataset_weights": true,
90
+ "balance_trajectory_weights": true,
91
+ "dataset_paths": null,
92
+ "output_dir": "./tmp/gr00t",
93
+ "save_steps": 1000,
94
+ "max_steps": 20000,
95
+ "batch_size": 32,
96
+ "dataloader_num_workers": 8,
97
+ "report_to": "wandb",
98
+ "resume": false
99
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:45e2ba39d5358ad9bbf72a43766441c4fab9aeefe0a05e5d617151c977755221
3
+ size 12576215296
policy_postprocessor.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "name": "policy_postprocessor",
3
+ "input_image_format": "float32_0_1",
4
+ "steps": [
5
+ {
6
+ "registry_name": "groot_action_unpack_unnormalize_v2",
7
+ "config": {
8
+ "env_action_dim": 7,
9
+ "normalize_min_max": true,
10
+ "clip_normalized_action": true,
11
+ "libero_gripper_action": true,
12
+ "libero_gripper_binarize": true
13
+ },
14
+ "state_file": "policy_postprocessor_step_0_groot_action_unpack_unnormalize_v2.safetensors"
15
+ },
16
+ {
17
+ "registry_name": "device_processor",
18
+ "config": {
19
+ "device": "cpu",
20
+ "float_dtype": null
21
+ }
22
+ }
23
+ ]
24
+ }
policy_postprocessor_step_0_groot_action_unpack_unnormalize_v2.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f9d98298bc6eb073d56d087fcd681447ff2d74c4769ad2fa1ec15b4bbc0baec2
3
+ size 4224
policy_preprocessor.json ADDED
@@ -0,0 +1,118 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "name": "policy_preprocessor",
3
+ "input_image_format": "uint8_0_255",
4
+ "steps": [
5
+ {
6
+ "registry_name": "rename_observations_processor",
7
+ "config": {
8
+ "rename_map": {}
9
+ }
10
+ },
11
+ {
12
+ "registry_name": "to_batch_processor",
13
+ "config": {}
14
+ },
15
+ {
16
+ "registry_name": "groot_n1_7_pack_inputs_v1",
17
+ "config": {
18
+ "state_horizon": 1,
19
+ "action_horizon": 40,
20
+ "valid_action_horizon": 40,
21
+ "video_horizon": null,
22
+ "max_state_dim": 132,
23
+ "max_action_dim": 132,
24
+ "language_key": "task",
25
+ "formalize_language": true,
26
+ "embodiment_tag": "libero_sim",
27
+ "embodiment_mapping": {
28
+ "robocasa_panda_omron": 13,
29
+ "oxe_droid": 17,
30
+ "oxe_fractal": 18,
31
+ "oxe_language_table": 19,
32
+ "oxe_bridge": 20,
33
+ "unknown": 22,
34
+ "gr1_unified": 20,
35
+ "agibot": 26,
36
+ "sim_behavior_r1_pro": 23,
37
+ "xdof": 24,
38
+ "xdof_oss_data": 25,
39
+ "unitree_g1_full_body_with_waist_height_nav_cmd": 25,
40
+ "real_r1_pro_sharpa": 27,
41
+ "real_r1_pro_sharpa_add_view": 27,
42
+ "real_r1_pro_sharpa_relative_arm_joint": 26,
43
+ "real_r1_pro_sharpa_delta_eef": 26,
44
+ "real_r1_pro_sharpa_absolute_eef": 26,
45
+ "real_r1_pro_sharpa_meanstd": 26,
46
+ "real_r1_pro_sharpa_relative_eef": 26,
47
+ "real_r1_pro_sharpa_relative_eef_add_view": 26,
48
+ "real_r1_pro_sharpa_relative_eef_relative_hand": 26,
49
+ "real_r1_pro_sharpa_relative_eef_human": 26,
50
+ "real_r1_pro_sharpa_relative_eef_human_add_view": 26,
51
+ "real_r1_pro_sharpa_relative_eef_human_relative_hand": 26,
52
+ "real_r1_pro_sharpa_relative_eef_egodex": 26,
53
+ "real_r1_pro_sharpa_relative_eef_egodex_relative_hand": 26,
54
+ "real_r1_pro_sharpa_relative_eef_egodex_wrist_only": 26,
55
+ "real_r1_pro_sharpa_relative_eef_maxinsights": 26,
56
+ "real_r1_pro_sharpa_relative_eef_maxinsights_relative_hand": 26,
57
+ "real_r1_pro_sharpa_relative_eef_mecka": 26,
58
+ "real_r1_pro_sharpa_relative_eef_mecka_relative_hand": 26,
59
+ "real_g1_relative_eef_absolute_joints": 25,
60
+ "real_g1_relative_eef_absolute_joints_wrist_cam": 25,
61
+ "real_g1_relative_eef_relative_joints": 25,
62
+ "real_r1_pro_sharpa_relative_eef_relative_hand_relative_joint": 26,
63
+ "real_r1_pro_sharpa_relative_joint": 29,
64
+ "oxe_droid_relative_eef_relative_joint": 24,
65
+ "oxe_droid_relative_eef_relative_joint_swapped": 24,
66
+ "oxe_droid_relative_eef_relative_joint_upweight_z": 24,
67
+ "oxe_droid_relative_eef_relative_joint_upweight_z_swapped": 24,
68
+ "oxe_droid_relative_eef_relative_joint_3view": 24,
69
+ "oxe_droid_relative_eef_relative_joint_3view_swapped": 24,
70
+ "oxe_droid_relative_eef": 24,
71
+ "oxe_droid_joint_position_relative": 24,
72
+ "xdof_relative_eef_relative_joint": 27,
73
+ "xdof_relative_eef_relative_joint_subtask": 27,
74
+ "xdof_relative_eef": 27,
75
+ "xdof_relative_joint": 28,
76
+ "simpler_env_google": 0,
77
+ "simpler_env_widowx": 1,
78
+ "libero_sim": 2,
79
+ "droid_sim": 3
80
+ },
81
+ "normalize_min_max": true,
82
+ "state_dropout_prob": 0.2,
83
+ "clip_outliers": true,
84
+ "use_percentiles": true,
85
+ "video_modality_keys": null,
86
+ "raw_stats": {},
87
+ "modality_config": {}
88
+ },
89
+ "state_file": "policy_preprocessor_step_2_groot_n1_7_pack_inputs_v1.safetensors"
90
+ },
91
+ {
92
+ "registry_name": "groot_n1_7_vlm_encode_v1",
93
+ "config": {
94
+ "model_name": "nvidia/Cosmos-Reason2-2B",
95
+ "image_crop_size": [
96
+ 230,
97
+ 230
98
+ ],
99
+ "image_target_size": [
100
+ 256,
101
+ 256
102
+ ],
103
+ "shortest_image_edge": 256,
104
+ "crop_fraction": 0.95,
105
+ "use_albumentations": true,
106
+ "letter_box_transform": false,
107
+ "device": "cuda"
108
+ }
109
+ },
110
+ {
111
+ "registry_name": "device_processor",
112
+ "config": {
113
+ "device": "cuda",
114
+ "float_dtype": null
115
+ }
116
+ }
117
+ ]
118
+ }
policy_preprocessor_step_2_groot_n1_7_pack_inputs_v1.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f9d98298bc6eb073d56d087fcd681447ff2d74c4769ad2fa1ec15b4bbc0baec2
3
+ size 4224
train_config.json ADDED
@@ -0,0 +1,209 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "IPEC-COMMUNITY/libero_object_no_noops_1.0.0_lerobot",
4
+ "root": "/datasets/libero_object",
5
+ "episodes": null,
6
+ "image_transforms": {
7
+ "enable": true,
8
+ "max_num_transforms": 1,
9
+ "random_order": false,
10
+ "tfs": {
11
+ "color_jitter": {
12
+ "weight": 1.0,
13
+ "type": "ColorJitter",
14
+ "kwargs": {
15
+ "brightness": [
16
+ 0.7,
17
+ 1.3
18
+ ],
19
+ "contrast": [
20
+ 0.6,
21
+ 1.4
22
+ ],
23
+ "saturation": [
24
+ 0.5,
25
+ 1.5
26
+ ],
27
+ "hue": [
28
+ -0.08,
29
+ 0.08
30
+ ]
31
+ }
32
+ }
33
+ }
34
+ },
35
+ "revision": "main",
36
+ "use_imagenet_stats": true,
37
+ "video_backend": "torchcodec",
38
+ "return_uint8": true,
39
+ "depth_output_unit": "mm",
40
+ "streaming": false,
41
+ "eval_split": 0.0
42
+ },
43
+ "env": null,
44
+ "policy": {
45
+ "type": "groot",
46
+ "n_obs_steps": 1,
47
+ "input_features": {
48
+ "observation.images.wrist_image": {
49
+ "type": "VISUAL",
50
+ "shape": [
51
+ 256,
52
+ 256,
53
+ 3
54
+ ]
55
+ },
56
+ "observation.images.image": {
57
+ "type": "VISUAL",
58
+ "shape": [
59
+ 256,
60
+ 256,
61
+ 3
62
+ ]
63
+ },
64
+ "observation.state": {
65
+ "type": "STATE",
66
+ "shape": [
67
+ 8
68
+ ]
69
+ }
70
+ },
71
+ "output_features": {
72
+ "action": {
73
+ "type": "ACTION",
74
+ "shape": [
75
+ 7
76
+ ]
77
+ }
78
+ },
79
+ "device": "cuda",
80
+ "use_amp": false,
81
+ "use_peft": false,
82
+ "push_to_hub": false,
83
+ "repo_id": null,
84
+ "private": null,
85
+ "tags": null,
86
+ "license": null,
87
+ "pretrained_path": null,
88
+ "pretrained_revision": null,
89
+ "chunk_size": 16,
90
+ "n_action_steps": 16,
91
+ "max_state_dim": 132,
92
+ "max_action_dim": 132,
93
+ "normalization_mapping": {
94
+ "VISUAL": "IDENTITY",
95
+ "STATE": "IDENTITY",
96
+ "ACTION": "IDENTITY"
97
+ },
98
+ "base_model_path": "/cache/huggingface/models--nvidia--GR00T-N1.7-3B/snapshots/2fc962b973bccdd5d8ce4f67cc63b264d6886495",
99
+ "action_decode_transform": "libero",
100
+ "embodiment_tag": "libero_sim",
101
+ "tune_llm": false,
102
+ "tune_visual": false,
103
+ "tune_projector": true,
104
+ "tune_diffusion_model": true,
105
+ "tune_vlln": true,
106
+ "tune_top_llm_layers": 0,
107
+ "num_inference_timesteps": 4,
108
+ "rtc_ramp_rate": 6.0,
109
+ "use_flash_attention": false,
110
+ "use_relative_actions": false,
111
+ "relative_exclude_joints": [],
112
+ "optimizer_lr": 0.0001,
113
+ "optimizer_betas": [
114
+ 0.9,
115
+ 0.999
116
+ ],
117
+ "optimizer_eps": 1e-08,
118
+ "optimizer_weight_decay": 1e-05,
119
+ "warmup_ratio": 0.05,
120
+ "use_bf16": true,
121
+ "model_params_fp32": true,
122
+ "image_size": [
123
+ 256,
124
+ 256
125
+ ],
126
+ "tokenizer_assets_repo": null,
127
+ "lora_rank": 0,
128
+ "lora_alpha": 16,
129
+ "lora_dropout": 0.1,
130
+ "lora_full_model": false,
131
+ "video_backend": "decord",
132
+ "balance_dataset_weights": true,
133
+ "balance_trajectory_weights": true,
134
+ "dataset_paths": null,
135
+ "output_dir": "./tmp/gr00t",
136
+ "save_steps": 1000,
137
+ "max_steps": 20000,
138
+ "batch_size": 32,
139
+ "dataloader_num_workers": 8,
140
+ "report_to": "wandb",
141
+ "resume": false
142
+ },
143
+ "reward_model": null,
144
+ "output_dir": "/checkpoints/gr00t-n17-libero-oss-h16-adamw-gbs640-20k-head9adea3f9/20260705-214215/libero_object",
145
+ "job_name": "libero-object-n17-oss-h16-adamw-gbs640-s42-head9adea3f9-20260705-214215",
146
+ "resume": false,
147
+ "seed": 42,
148
+ "cudnn_deterministic": false,
149
+ "num_workers": 4,
150
+ "batch_size": 320,
151
+ "prefetch_factor": 2,
152
+ "persistent_workers": true,
153
+ "steps": 20000,
154
+ "env_eval_freq": 0,
155
+ "log_freq": 10,
156
+ "eval_steps": 0,
157
+ "max_eval_samples": 0,
158
+ "tolerance_s": 0.0001,
159
+ "save_checkpoint": true,
160
+ "save_freq": 1000,
161
+ "use_policy_training_preset": false,
162
+ "optimizer": {
163
+ "type": "adamw",
164
+ "lr": 0.0001,
165
+ "weight_decay": 1e-05,
166
+ "grad_clip_norm": 1.0,
167
+ "betas": [
168
+ 0.9,
169
+ 0.999
170
+ ],
171
+ "eps": 1e-08,
172
+ "fused": false
173
+ },
174
+ "scheduler": {
175
+ "type": "diffuser",
176
+ "num_warmup_steps": 1000,
177
+ "name": "cosine"
178
+ },
179
+ "eval": {
180
+ "n_episodes": 50,
181
+ "batch_size": 50,
182
+ "use_async_envs": true,
183
+ "recording": false,
184
+ "recording_repo_id": null,
185
+ "recording_private": false
186
+ },
187
+ "wandb": {
188
+ "enable": true,
189
+ "disable_artifact": true,
190
+ "project": "lerobot-gr00t-17n-train",
191
+ "entity": "nvidia",
192
+ "notes": "source_commit=9adea3f9638551112a370bdc7b3538b2a6055adf;oss_source_commit=ab88b50c718f6528e1df9dcbaf75865d1b604760;recipe=published-oss-libero-h16-adamw-20k-gbs640;layout=2x320;actions=absolute-libero-no-relative-groups;accepted_jitter=torchvision",
193
+ "run_id": "libero-object-n17-oss-h16-adamw-gbs640-s42-head9adea3f9-20260705-214215",
194
+ "mode": "online",
195
+ "add_tags": true
196
+ },
197
+ "peft": null,
198
+ "job": {
199
+ "target": null,
200
+ "image": "huggingface/lerobot-gpu:latest",
201
+ "timeout": "2d",
202
+ "detach": false,
203
+ "tags": []
204
+ },
205
+ "save_checkpoint_to_hub": false,
206
+ "sample_weighting": null,
207
+ "rename_map": {},
208
+ "checkpoint_path": null
209
+ }