Hugging Face's logo

videron
/
openarm_folding_gen3_level2_multitask_50k

Safetensors
Model card Files
xet
 Community
videron commited on
Commit
ba7676d
·
verified ·
1 Parent(s): 62486d0

Upload folder using huggingface_hub

Browse files
Files changed (7) hide show
  1. config.json +119 -0
  2. model.safetensors +3 -0
  3. policy_postprocessor.json +38 -0
  4. policy_postprocessor_step_0_unnormalizer_processor.safetensors +3 -0
  5. policy_preprocessor.json +115 -0
  6. policy_preprocessor_step_3_normalizer_processor.safetensors +3 -0
  7. train_config.json +274 -0
config.json ADDED
@@ -0,0 +1,119 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "type": "pi05",
3
+ "n_obs_steps": 1,
4
+ "input_features": {
5
+ "observation.state": {
6
+ "type": "STATE",
7
+ "shape": [
8
+ 16
9
+ ]
10
+ },
11
+ "observation.images.left_wrist": {
12
+ "type": "VISUAL",
13
+ "shape": [
14
+ 3,
15
+ 720,
16
+ 1280
17
+ ]
18
+ },
19
+ "observation.images.right_wrist": {
20
+ "type": "VISUAL",
21
+ "shape": [
22
+ 3,
23
+ 720,
24
+ 1280
25
+ ]
26
+ },
27
+ "observation.images.base": {
28
+ "type": "VISUAL",
29
+ "shape": [
30
+ 3,
31
+ 480,
32
+ 640
33
+ ]
34
+ }
35
+ },
36
+ "output_features": {
37
+ "action": {
38
+ "type": "ACTION",
39
+ "shape": [
40
+ 16
41
+ ]
42
+ }
43
+ },
44
+ "device": "cuda",
45
+ "use_amp": false,
46
+ "use_peft": false,
47
+ "push_to_hub": false,
48
+ "repo_id": "videron/openarm_folding_gen3_level2_multitask",
49
+ "private": null,
50
+ "tags": null,
51
+ "license": null,
52
+ "pretrained_path": "videron/openarm_folding_gen3_cotraining_100k",
53
+ "pretrained_revision": null,
54
+ "paligemma_variant": "gemma_2b",
55
+ "action_expert_variant": "gemma_300m",
56
+ "dtype": "bfloat16",
57
+ "chunk_size": 30,
58
+ "n_action_steps": 30,
59
+ "max_state_dim": 32,
60
+ "max_action_dim": 32,
61
+ "num_inference_steps": 10,
62
+ "time_sampling_beta_alpha": 1.5,
63
+ "time_sampling_beta_beta": 1.0,
64
+ "time_sampling_scale": 0.999,
65
+ "time_sampling_offset": 0.001,
66
+ "min_period": 0.004,
67
+ "max_period": 4.0,
68
+ "use_relative_actions": true,
69
+ "relative_exclude_joints": [
70
+ "left_gripper",
71
+ "right_gripper"
72
+ ],
73
+ "action_feature_names": [
74
+ "right_joint_1.pos",
75
+ "right_joint_2.pos",
76
+ "right_joint_3.pos",
77
+ "right_joint_4.pos",
78
+ "right_joint_5.pos",
79
+ "right_joint_6.pos",
80
+ "right_joint_7.pos",
81
+ "right_gripper.pos",
82
+ "left_joint_1.pos",
83
+ "left_joint_2.pos",
84
+ "left_joint_3.pos",
85
+ "left_joint_4.pos",
86
+ "left_joint_5.pos",
87
+ "left_joint_6.pos",
88
+ "left_joint_7.pos",
89
+ "left_gripper.pos"
90
+ ],
91
+ "rtc_config": null,
92
+ "image_resolution": [
93
+ 224,
94
+ 224
95
+ ],
96
+ "empty_cameras": 0,
97
+ "tokenizer_max_length": 200,
98
+ "normalization_mapping": {
99
+ "VISUAL": "IDENTITY",
100
+ "STATE": "QUANTILES",
101
+ "ACTION": "QUANTILES"
102
+ },
103
+ "gradient_checkpointing": true,
104
+ "compile_model": false,
105
+ "compile_mode": "max-autotune",
106
+ "freeze_vision_encoder": false,
107
+ "train_expert_only": false,
108
+ "optimizer_lr": 3.75e-05,
109
+ "optimizer_betas": [
110
+ 0.9,
111
+ 0.95
112
+ ],
113
+ "optimizer_eps": 1e-08,
114
+ "optimizer_weight_decay": 0.01,
115
+ "optimizer_grad_clip_norm": 1.0,
116
+ "scheduler_warmup_steps": 1000,
117
+ "scheduler_decay_steps": 50000,
118
+ "scheduler_decay_lr": 2.5e-06
119
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8a0332d84dc5ed09be03fcbc14057e2e4a9b5727f6b20fcaa740ca22e72cc09c
3
+ size 9354050752
policy_postprocessor.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "name": "policy_postprocessor",
3
+ "steps": [
4
+ {
5
+ "registry_name": "unnormalizer_processor",
6
+ "config": {
7
+ "eps": 1e-08,
8
+ "features": {
9
+ "action": {
10
+ "type": "ACTION",
11
+ "shape": [
12
+ 16
13
+ ]
14
+ }
15
+ },
16
+ "norm_map": {
17
+ "VISUAL": "IDENTITY",
18
+ "STATE": "QUANTILES",
19
+ "ACTION": "QUANTILES"
20
+ }
21
+ },
22
+ "state_file": "policy_postprocessor_step_0_unnormalizer_processor.safetensors"
23
+ },
24
+ {
25
+ "registry_name": "absolute_actions_processor",
26
+ "config": {
27
+ "enabled": true
28
+ }
29
+ },
30
+ {
31
+ "registry_name": "device_processor",
32
+ "config": {
33
+ "device": "cpu",
34
+ "float_dtype": null
35
+ }
36
+ }
37
+ ]
38
+ }
policy_postprocessor_step_0_unnormalizer_processor.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:41e312fe4565cc319c87084e367f2331663d5ea2cf2e25498f741a4801810409
3
+ size 9456
policy_preprocessor.json ADDED
@@ -0,0 +1,115 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "name": "policy_preprocessor",
3
+ "steps": [
4
+ {
5
+ "registry_name": "rename_observations_processor",
6
+ "config": {
7
+ "rename_map": {}
8
+ }
9
+ },
10
+ {
11
+ "registry_name": "to_batch_processor",
12
+ "config": {}
13
+ },
14
+ {
15
+ "registry_name": "relative_actions_processor",
16
+ "config": {
17
+ "enabled": true,
18
+ "exclude_joints": [
19
+ "left_gripper",
20
+ "right_gripper"
21
+ ],
22
+ "action_names": [
23
+ "right_joint_1.pos",
24
+ "right_joint_2.pos",
25
+ "right_joint_3.pos",
26
+ "right_joint_4.pos",
27
+ "right_joint_5.pos",
28
+ "right_joint_6.pos",
29
+ "right_joint_7.pos",
30
+ "right_gripper.pos",
31
+ "left_joint_1.pos",
32
+ "left_joint_2.pos",
33
+ "left_joint_3.pos",
34
+ "left_joint_4.pos",
35
+ "left_joint_5.pos",
36
+ "left_joint_6.pos",
37
+ "left_joint_7.pos",
38
+ "left_gripper.pos"
39
+ ]
40
+ }
41
+ },
42
+ {
43
+ "registry_name": "normalizer_processor",
44
+ "config": {
45
+ "eps": 1e-08,
46
+ "features": {
47
+ "observation.state": {
48
+ "type": "STATE",
49
+ "shape": [
50
+ 16
51
+ ]
52
+ },
53
+ "observation.images.left_wrist": {
54
+ "type": "VISUAL",
55
+ "shape": [
56
+ 3,
57
+ 720,
58
+ 1280
59
+ ]
60
+ },
61
+ "observation.images.right_wrist": {
62
+ "type": "VISUAL",
63
+ "shape": [
64
+ 3,
65
+ 720,
66
+ 1280
67
+ ]
68
+ },
69
+ "observation.images.base": {
70
+ "type": "VISUAL",
71
+ "shape": [
72
+ 3,
73
+ 480,
74
+ 640
75
+ ]
76
+ },
77
+ "action": {
78
+ "type": "ACTION",
79
+ "shape": [
80
+ 16
81
+ ]
82
+ }
83
+ },
84
+ "norm_map": {
85
+ "VISUAL": "IDENTITY",
86
+ "STATE": "QUANTILES",
87
+ "ACTION": "QUANTILES"
88
+ }
89
+ },
90
+ "state_file": "policy_preprocessor_step_3_normalizer_processor.safetensors"
91
+ },
92
+ {
93
+ "registry_name": "pi05_prepare_state_tokenizer_processor_step",
94
+ "config": {}
95
+ },
96
+ {
97
+ "registry_name": "tokenizer_processor",
98
+ "config": {
99
+ "max_length": 200,
100
+ "task_key": "task",
101
+ "padding_side": "right",
102
+ "padding": "max_length",
103
+ "truncation": true,
104
+ "tokenizer_name": "google/paligemma-3b-pt-224"
105
+ }
106
+ },
107
+ {
108
+ "registry_name": "device_processor",
109
+ "config": {
110
+ "device": "cuda",
111
+ "float_dtype": null
112
+ }
113
+ }
114
+ ]
115
+ }
policy_preprocessor_step_3_normalizer_processor.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:172c9783c4214747954900a9b2ffe79aae0553e70d3492ced7aff356dca820e3
3
+ size 9464
train_config.json ADDED
@@ -0,0 +1,274 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "videron/openarm_bimanual_shirt_folding_gen3_345_episodes_fold_unfold_home_rightfirst",
4
+ "repo_type": "dataset",
5
+ "root": null,
6
+ "episodes": null,
7
+ "image_transforms": {
8
+ "enable": false,
9
+ "max_num_transforms": 3,
10
+ "random_order": false,
11
+ "tfs": {
12
+ "brightness": {
13
+ "weight": 1.0,
14
+ "type": "ColorJitter",
15
+ "kwargs": {
16
+ "brightness": [
17
+ 0.8,
18
+ 1.2
19
+ ]
20
+ }
21
+ },
22
+ "contrast": {
23
+ "weight": 1.0,
24
+ "type": "ColorJitter",
25
+ "kwargs": {
26
+ "contrast": [
27
+ 0.8,
28
+ 1.2
29
+ ]
30
+ }
31
+ },
32
+ "saturation": {
33
+ "weight": 1.0,
34
+ "type": "ColorJitter",
35
+ "kwargs": {
36
+ "saturation": [
37
+ 0.5,
38
+ 1.5
39
+ ]
40
+ }
41
+ },
42
+ "hue": {
43
+ "weight": 1.0,
44
+ "type": "ColorJitter",
45
+ "kwargs": {
46
+ "hue": [
47
+ -0.05,
48
+ 0.05
49
+ ]
50
+ }
51
+ },
52
+ "sharpness": {
53
+ "weight": 1.0,
54
+ "type": "SharpnessJitter",
55
+ "kwargs": {
56
+ "sharpness": [
57
+ 0.5,
58
+ 1.5
59
+ ]
60
+ }
61
+ },
62
+ "affine": {
63
+ "weight": 1.0,
64
+ "type": "RandomAffine",
65
+ "kwargs": {
66
+ "degrees": [
67
+ -5.0,
68
+ 5.0
69
+ ],
70
+ "translate": [
71
+ 0.05,
72
+ 0.05
73
+ ]
74
+ }
75
+ }
76
+ }
77
+ },
78
+ "revision": null,
79
+ "use_imagenet_stats": true,
80
+ "video_backend": "torchcodec",
81
+ "return_uint8": false,
82
+ "depth_output_unit": "mm",
83
+ "streaming": false,
84
+ "eval_split": 0.0
85
+ },
86
+ "env": null,
87
+ "policy": {
88
+ "type": "pi05",
89
+ "n_obs_steps": 1,
90
+ "input_features": {
91
+ "observation.state": {
92
+ "type": "STATE",
93
+ "shape": [
94
+ 16
95
+ ]
96
+ },
97
+ "observation.images.left_wrist": {
98
+ "type": "VISUAL",
99
+ "shape": [
100
+ 3,
101
+ 720,
102
+ 1280
103
+ ]
104
+ },
105
+ "observation.images.right_wrist": {
106
+ "type": "VISUAL",
107
+ "shape": [
108
+ 3,
109
+ 720,
110
+ 1280
111
+ ]
112
+ },
113
+ "observation.images.base": {
114
+ "type": "VISUAL",
115
+ "shape": [
116
+ 3,
117
+ 480,
118
+ 640
119
+ ]
120
+ }
121
+ },
122
+ "output_features": {
123
+ "action": {
124
+ "type": "ACTION",
125
+ "shape": [
126
+ 16
127
+ ]
128
+ }
129
+ },
130
+ "device": "cuda",
131
+ "use_amp": false,
132
+ "use_peft": false,
133
+ "push_to_hub": false,
134
+ "repo_id": "videron/openarm_folding_gen3_level2_multitask",
135
+ "private": null,
136
+ "tags": null,
137
+ "license": null,
138
+ "pretrained_path": "videron/openarm_folding_gen3_cotraining_100k",
139
+ "pretrained_revision": null,
140
+ "paligemma_variant": "gemma_2b",
141
+ "action_expert_variant": "gemma_300m",
142
+ "dtype": "bfloat16",
143
+ "chunk_size": 30,
144
+ "n_action_steps": 30,
145
+ "max_state_dim": 32,
146
+ "max_action_dim": 32,
147
+ "num_inference_steps": 10,
148
+ "time_sampling_beta_alpha": 1.5,
149
+ "time_sampling_beta_beta": 1.0,
150
+ "time_sampling_scale": 0.999,
151
+ "time_sampling_offset": 0.001,
152
+ "min_period": 0.004,
153
+ "max_period": 4.0,
154
+ "use_relative_actions": true,
155
+ "relative_exclude_joints": [
156
+ "left_gripper",
157
+ "right_gripper"
158
+ ],
159
+ "action_feature_names": [
160
+ "right_joint_1.pos",
161
+ "right_joint_2.pos",
162
+ "right_joint_3.pos",
163
+ "right_joint_4.pos",
164
+ "right_joint_5.pos",
165
+ "right_joint_6.pos",
166
+ "right_joint_7.pos",
167
+ "right_gripper.pos",
168
+ "left_joint_1.pos",
169
+ "left_joint_2.pos",
170
+ "left_joint_3.pos",
171
+ "left_joint_4.pos",
172
+ "left_joint_5.pos",
173
+ "left_joint_6.pos",
174
+ "left_joint_7.pos",
175
+ "left_gripper.pos"
176
+ ],
177
+ "rtc_config": null,
178
+ "image_resolution": [
179
+ 224,
180
+ 224
181
+ ],
182
+ "empty_cameras": 0,
183
+ "tokenizer_max_length": 200,
184
+ "normalization_mapping": {
185
+ "VISUAL": "IDENTITY",
186
+ "STATE": "QUANTILES",
187
+ "ACTION": "QUANTILES"
188
+ },
189
+ "gradient_checkpointing": true,
190
+ "compile_model": false,
191
+ "compile_mode": "max-autotune",
192
+ "freeze_vision_encoder": false,
193
+ "train_expert_only": false,
194
+ "optimizer_lr": 3.75e-05,
195
+ "optimizer_betas": [
196
+ 0.9,
197
+ 0.95
198
+ ],
199
+ "optimizer_eps": 1e-08,
200
+ "optimizer_weight_decay": 0.01,
201
+ "optimizer_grad_clip_norm": 1.0,
202
+ "scheduler_warmup_steps": 1000,
203
+ "scheduler_decay_steps": 50000,
204
+ "scheduler_decay_lr": 2.5e-06
205
+ },
206
+ "reward_model": null,
207
+ "output_dir": "outputs/train/openarm_folding_gen3_level2_multitask",
208
+ "job_name": "openarm_folding_gen3_level2_multitask",
209
+ "resume": false,
210
+ "seed": 1000,
211
+ "cudnn_deterministic": false,
212
+ "num_workers": 8,
213
+ "batch_size": 32,
214
+ "prefetch_factor": 4,
215
+ "persistent_workers": false,
216
+ "dataloader_multiprocessing_context": "spawn",
217
+ "steps": 50000,
218
+ "env_eval_freq": 20000,
219
+ "log_freq": 100,
220
+ "eval_steps": 0,
221
+ "max_eval_samples": 0,
222
+ "tolerance_s": 0.0001,
223
+ "save_checkpoint": true,
224
+ "save_freq": 5000,
225
+ "use_policy_training_preset": true,
226
+ "optimizer": {
227
+ "type": "adamw",
228
+ "lr": 3.75e-05,
229
+ "weight_decay": 0.01,
230
+ "grad_clip_norm": 1.0,
231
+ "betas": [
232
+ 0.9,
233
+ 0.95
234
+ ],
235
+ "eps": 1e-08
236
+ },
237
+ "scheduler": {
238
+ "type": "cosine_decay_with_warmup",
239
+ "num_warmup_steps": 1000,
240
+ "num_decay_steps": 50000,
241
+ "peak_lr": 3.75e-05,
242
+ "decay_lr": 2.5e-06
243
+ },
244
+ "eval": {
245
+ "n_episodes": 50,
246
+ "batch_size": 18,
247
+ "use_async_envs": true,
248
+ "recording": false,
249
+ "recording_repo_id": null,
250
+ "recording_private": false
251
+ },
252
+ "wandb": {
253
+ "enable": true,
254
+ "disable_artifact": true,
255
+ "project": "lerobot",
256
+ "entity": null,
257
+ "notes": null,
258
+ "run_id": "3opyqvq6",
259
+ "mode": null,
260
+ "add_tags": true
261
+ },
262
+ "peft": null,
263
+ "job": {
264
+ "target": null,
265
+ "image": "huggingface/lerobot-gpu:latest",
266
+ "timeout": "2d",
267
+ "detach": false,
268
+ "tags": []
269
+ },
270
+ "save_checkpoint_to_hub": false,
271
+ "sample_weighting": null,
272
+ "rename_map": {},
273
+ "checkpoint_path": null
274
+ }