Hugging Face's logo

chomeed
/
dexmg_two_arm_threading

Robotics
LeRobot
Safetensors
diffusion
Model card Files
xet
 Community
chomeed commited on
Commit
c0c5b92
·
verified ·
1 Parent(s): 306f4ee

Upload policy weights, train config and readme

Browse files
Files changed (4) hide show
  1. README.md +62 -0
  2. config.json +97 -0
  3. model.safetensors +3 -0
  4. train_config.json +397 -0
README.md ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ datasets: chomeed/dexmg_two_arm_threading
3
+ library_name: lerobot
4
+ license: apache-2.0
5
+ model_name: diffusion
6
+ pipeline_tag: robotics
7
+ tags:
8
+ - lerobot
9
+ - robotics
10
+ - diffusion
11
+ ---
12
+
13
+ # Model Card for diffusion
14
+
15
+ <!-- Provide a quick summary of what the model is/does. -->
16
+
17
+
18
+ [Diffusion Policy](https://huggingface.co/papers/2303.04137) treats visuomotor control as a generative diffusion process, producing smooth, multi-step action trajectories that excel at contact-rich manipulation.
19
+
20
+
21
+ This policy has been trained and pushed to the Hub using [LeRobot](https://github.com/huggingface/lerobot).
22
+ See the full documentation at [LeRobot Docs](https://huggingface.co/docs/lerobot/index).
23
+
24
+ ---
25
+
26
+ ## How to Get Started with the Model
27
+
28
+ For a complete walkthrough, see the [training guide](https://huggingface.co/docs/lerobot/il_robots#train-a-policy).
29
+ Below is the short version on how to train and run inference/eval:
30
+
31
+ ### Train from scratch
32
+
33
+ ```bash
34
+ lerobot-train \
35
+ --dataset.repo_id=${HF_USER}/<dataset> \
36
+ --policy.type=act \
37
+ --output_dir=outputs/train/<desired_policy_repo_id> \
38
+ --job_name=lerobot_training \
39
+ --policy.device=cuda \
40
+ --policy.repo_id=${HF_USER}/<desired_policy_repo_id>
41
+ --wandb.enable=true
42
+ ```
43
+
44
+ _Writes checkpoints to `outputs/train/<desired_policy_repo_id>/checkpoints/`._
45
+
46
+ ### Evaluate the policy/run inference
47
+
48
+ ```bash
49
+ lerobot-record \
50
+ --robot.type=so100_follower \
51
+ --dataset.repo_id=<hf_user>/eval_<dataset> \
52
+ --policy.path=<hf_user>/<desired_policy_repo_id> \
53
+ --episodes=10
54
+ ```
55
+
56
+ Prefix the dataset repo with **eval\_** and supply `--policy.path` pointing to a local or hub checkpoint.
57
+
58
+ ---
59
+
60
+ ## Model Details
61
+
62
+ - **License:** apache-2.0
config.json ADDED
@@ -0,0 +1,97 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "type": "diffusion",
3
+ "n_obs_steps": 2,
4
+ "input_features": {
5
+ "observation.images.agentview": {
6
+ "type": "VISUAL",
7
+ "shape": [
8
+ 3,
9
+ 84,
10
+ 84
11
+ ]
12
+ },
13
+ "observation.images.robot0_eye_in_hand": {
14
+ "type": "VISUAL",
15
+ "shape": [
16
+ 3,
17
+ 84,
18
+ 84
19
+ ]
20
+ },
21
+ "observation.images.robot1_eye_in_hand": {
22
+ "type": "VISUAL",
23
+ "shape": [
24
+ 3,
25
+ 84,
26
+ 84
27
+ ]
28
+ },
29
+ "observation.state": {
30
+ "type": "STATE",
31
+ "shape": [
32
+ 18
33
+ ]
34
+ }
35
+ },
36
+ "output_features": {
37
+ "action": {
38
+ "type": "ACTION",
39
+ "shape": [
40
+ 14
41
+ ]
42
+ }
43
+ },
44
+ "device": "cuda",
45
+ "use_amp": false,
46
+ "use_peft": false,
47
+ "push_to_hub": true,
48
+ "repo_id": "chomeed/dexmg_two_arm_threading",
49
+ "private": null,
50
+ "tags": null,
51
+ "license": null,
52
+ "pretrained_path": null,
53
+ "horizon": 8,
54
+ "n_action_steps": 8,
55
+ "normalization_mapping": {
56
+ "VISUAL": "MEAN_STD",
57
+ "STATE": "MIN_MAX",
58
+ "ACTION": "MIN_MAX"
59
+ },
60
+ "drop_n_last_frames": 7,
61
+ "vision_backbone": "resnet18",
62
+ "crop_shape": [
63
+ 76,
64
+ 76
65
+ ],
66
+ "crop_is_random": true,
67
+ "pretrained_backbone_weights": null,
68
+ "use_group_norm": true,
69
+ "spatial_softmax_num_keypoints": 32,
70
+ "use_separate_rgb_encoder_per_camera": false,
71
+ "down_dims": [
72
+ 512
73
+ ],
74
+ "kernel_size": 5,
75
+ "n_groups": 8,
76
+ "diffusion_step_embed_dim": 128,
77
+ "use_film_scale_modulation": true,
78
+ "noise_scheduler_type": "DDPM",
79
+ "num_train_timesteps": 100,
80
+ "beta_schedule": "squaredcos_cap_v2",
81
+ "beta_start": 0.0001,
82
+ "beta_end": 0.02,
83
+ "prediction_type": "epsilon",
84
+ "clip_sample": true,
85
+ "clip_sample_range": 1.0,
86
+ "num_inference_steps": null,
87
+ "do_mask_loss_for_padding": false,
88
+ "optimizer_lr": 0.0001,
89
+ "optimizer_betas": [
90
+ 0.95,
91
+ 0.999
92
+ ],
93
+ "optimizer_eps": 1e-08,
94
+ "optimizer_weight_decay": 1e-06,
95
+ "scheduler_name": "cosine",
96
+ "scheduler_warmup_steps": 500
97
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d37f26ce6f87bb19b394b1c3e68ef8f7798b12973996d7d0cae90215bd44cbde
3
+ size 96524712
train_config.json ADDED
@@ -0,0 +1,397 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "chomeed/dexmg_two_arm_threading",
4
+ "root": "/scratch2/chomeed/datasets/lerobot/dexmg_two_arm_threading",
5
+ "episodes": null,
6
+ "image_transforms": {
7
+ "enable": false,
8
+ "max_num_transforms": 3,
9
+ "random_order": false,
10
+ "tfs": {
11
+ "brightness": {
12
+ "weight": 1.0,
13
+ "type": "ColorJitter",
14
+ "kwargs": {
15
+ "brightness": [
16
+ 0.8,
17
+ 1.2
18
+ ]
19
+ }
20
+ },
21
+ "contrast": {
22
+ "weight": 1.0,
23
+ "type": "ColorJitter",
24
+ "kwargs": {
25
+ "contrast": [
26
+ 0.8,
27
+ 1.2
28
+ ]
29
+ }
30
+ },
31
+ "saturation": {
32
+ "weight": 1.0,
33
+ "type": "ColorJitter",
34
+ "kwargs": {
35
+ "saturation": [
36
+ 0.5,
37
+ 1.5
38
+ ]
39
+ }
40
+ },
41
+ "hue": {
42
+ "weight": 1.0,
43
+ "type": "ColorJitter",
44
+ "kwargs": {
45
+ "hue": [
46
+ -0.05,
47
+ 0.05
48
+ ]
49
+ }
50
+ },
51
+ "sharpness": {
52
+ "weight": 1.0,
53
+ "type": "SharpnessJitter",
54
+ "kwargs": {
55
+ "sharpness": [
56
+ 0.5,
57
+ 1.5
58
+ ]
59
+ }
60
+ },
61
+ "affine": {
62
+ "weight": 1.0,
63
+ "type": "RandomAffine",
64
+ "kwargs": {
65
+ "degrees": [
66
+ -5.0,
67
+ 5.0
68
+ ],
69
+ "translate": [
70
+ 0.05,
71
+ 0.05
72
+ ]
73
+ }
74
+ }
75
+ }
76
+ },
77
+ "revision": null,
78
+ "use_imagenet_stats": true,
79
+ "video_backend": "torchcodec",
80
+ "streaming": false,
81
+ "cf_aug": false
82
+ },
83
+ "env": null,
84
+ "policy": {
85
+ "type": "diffusion",
86
+ "n_obs_steps": 2,
87
+ "input_features": {
88
+ "observation.images.agentview": {
89
+ "type": "VISUAL",
90
+ "shape": [
91
+ 3,
92
+ 84,
93
+ 84
94
+ ]
95
+ },
96
+ "observation.images.robot0_eye_in_hand": {
97
+ "type": "VISUAL",
98
+ "shape": [
99
+ 3,
100
+ 84,
101
+ 84
102
+ ]
103
+ },
104
+ "observation.images.robot1_eye_in_hand": {
105
+ "type": "VISUAL",
106
+ "shape": [
107
+ 3,
108
+ 84,
109
+ 84
110
+ ]
111
+ },
112
+ "observation.state": {
113
+ "type": "STATE",
114
+ "shape": [
115
+ 18
116
+ ]
117
+ }
118
+ },
119
+ "output_features": {
120
+ "action": {
121
+ "type": "ACTION",
122
+ "shape": [
123
+ 14
124
+ ]
125
+ }
126
+ },
127
+ "device": "cuda",
128
+ "use_amp": false,
129
+ "use_peft": false,
130
+ "push_to_hub": true,
131
+ "repo_id": "chomeed/dexmg_two_arm_threading",
132
+ "private": null,
133
+ "tags": null,
134
+ "license": null,
135
+ "pretrained_path": null,
136
+ "horizon": 8,
137
+ "n_action_steps": 8,
138
+ "normalization_mapping": {
139
+ "VISUAL": "MEAN_STD",
140
+ "STATE": "MIN_MAX",
141
+ "ACTION": "MIN_MAX"
142
+ },
143
+ "drop_n_last_frames": 7,
144
+ "vision_backbone": "resnet18",
145
+ "crop_shape": [
146
+ 76,
147
+ 76
148
+ ],
149
+ "crop_is_random": true,
150
+ "pretrained_backbone_weights": null,
151
+ "use_group_norm": true,
152
+ "spatial_softmax_num_keypoints": 32,
153
+ "use_separate_rgb_encoder_per_camera": false,
154
+ "down_dims": [
155
+ 512
156
+ ],
157
+ "kernel_size": 5,
158
+ "n_groups": 8,
159
+ "diffusion_step_embed_dim": 128,
160
+ "use_film_scale_modulation": true,
161
+ "noise_scheduler_type": "DDPM",
162
+ "num_train_timesteps": 100,
163
+ "beta_schedule": "squaredcos_cap_v2",
164
+ "beta_start": 0.0001,
165
+ "beta_end": 0.02,
166
+ "prediction_type": "epsilon",
167
+ "clip_sample": true,
168
+ "clip_sample_range": 1.0,
169
+ "num_inference_steps": null,
170
+ "do_mask_loss_for_padding": false,
171
+ "optimizer_lr": 0.0001,
172
+ "optimizer_betas": [
173
+ 0.95,
174
+ 0.999
175
+ ],
176
+ "optimizer_eps": 1e-08,
177
+ "optimizer_weight_decay": 1e-06,
178
+ "scheduler_name": "cosine",
179
+ "scheduler_warmup_steps": 500
180
+ },
181
+ "output_dir": "/scratch2/chomeed/outputs/train/dexmg_two_arm_threading_dp_Tp8_Ta8",
182
+ "job_name": "dexmg_two_arm_threading_dp_Tp8_Ta8",
183
+ "resume": false,
184
+ "seed": 1000,
185
+ "num_workers": 8,
186
+ "batch_size": 64,
187
+ "steps": 200000,
188
+ "eval_freq": 0,
189
+ "log_freq": 200,
190
+ "tolerance_s": 0.0001,
191
+ "save_checkpoint": true,
192
+ "save_freq": 10000,
193
+ "use_policy_training_preset": true,
194
+ "optimizer": {
195
+ "type": "adam",
196
+ "lr": 0.0001,
197
+ "weight_decay": 1e-06,
198
+ "grad_clip_norm": 10.0,
199
+ "betas": [
200
+ 0.95,
201
+ 0.999
202
+ ],
203
+ "eps": 1e-08
204
+ },
205
+ "scheduler": {
206
+ "type": "diffuser",
207
+ "num_warmup_steps": 500,
208
+ "name": "cosine"
209
+ },
210
+ "val": {
211
+ "repo_id": [
212
+ "chomeed/anubis_pick_carrot_out_of_pot_randomized_val_single",
213
+ "chomeed/anubis_stack_tuna_single",
214
+ "chomeed/aloha_static_cups_open_single",
215
+ "chomeed/aloha_static_candy_single",
216
+ "jellyho/aloha_box_into_pot_single",
217
+ "jellyho/aloha_dish_drainer_single"
218
+ ],
219
+ "root": null,
220
+ "episodes": null,
221
+ "image_transforms": {
222
+ "enable": false,
223
+ "max_num_transforms": 3,
224
+ "random_order": false,
225
+ "tfs": {
226
+ "brightness": {
227
+ "weight": 1.0,
228
+ "type": "ColorJitter",
229
+ "kwargs": {
230
+ "brightness": [
231
+ 0.8,
232
+ 1.2
233
+ ]
234
+ }
235
+ },
236
+ "contrast": {
237
+ "weight": 1.0,
238
+ "type": "ColorJitter",
239
+ "kwargs": {
240
+ "contrast": [
241
+ 0.8,
242
+ 1.2
243
+ ]
244
+ }
245
+ },
246
+ "saturation": {
247
+ "weight": 1.0,
248
+ "type": "ColorJitter",
249
+ "kwargs": {
250
+ "saturation": [
251
+ 0.5,
252
+ 1.5
253
+ ]
254
+ }
255
+ },
256
+ "hue": {
257
+ "weight": 1.0,
258
+ "type": "ColorJitter",
259
+ "kwargs": {
260
+ "hue": [
261
+ -0.05,
262
+ 0.05
263
+ ]
264
+ }
265
+ },
266
+ "sharpness": {
267
+ "weight": 1.0,
268
+ "type": "SharpnessJitter",
269
+ "kwargs": {
270
+ "sharpness": [
271
+ 0.5,
272
+ 1.5
273
+ ]
274
+ }
275
+ },
276
+ "affine": {
277
+ "weight": 1.0,
278
+ "type": "RandomAffine",
279
+ "kwargs": {
280
+ "degrees": [
281
+ -5.0,
282
+ 5.0
283
+ ],
284
+ "translate": [
285
+ 0.05,
286
+ 0.05
287
+ ]
288
+ }
289
+ }
290
+ }
291
+ },
292
+ "revision": null,
293
+ "use_imagenet_stats": true,
294
+ "video_backend": "torchcodec",
295
+ "streaming": false
296
+ },
297
+ "test": {
298
+ "repo_id": [
299
+ "chomeed/aloha_static_coffee_single",
300
+ "jellyho/aloha_handover_box_single"
301
+ ],
302
+ "root": null,
303
+ "episodes": null,
304
+ "image_transforms": {
305
+ "enable": false,
306
+ "max_num_transforms": 3,
307
+ "random_order": false,
308
+ "tfs": {
309
+ "brightness": {
310
+ "weight": 1.0,
311
+ "type": "ColorJitter",
312
+ "kwargs": {
313
+ "brightness": [
314
+ 0.8,
315
+ 1.2
316
+ ]
317
+ }
318
+ },
319
+ "contrast": {
320
+ "weight": 1.0,
321
+ "type": "ColorJitter",
322
+ "kwargs": {
323
+ "contrast": [
324
+ 0.8,
325
+ 1.2
326
+ ]
327
+ }
328
+ },
329
+ "saturation": {
330
+ "weight": 1.0,
331
+ "type": "ColorJitter",
332
+ "kwargs": {
333
+ "saturation": [
334
+ 0.5,
335
+ 1.5
336
+ ]
337
+ }
338
+ },
339
+ "hue": {
340
+ "weight": 1.0,
341
+ "type": "ColorJitter",
342
+ "kwargs": {
343
+ "hue": [
344
+ -0.05,
345
+ 0.05
346
+ ]
347
+ }
348
+ },
349
+ "sharpness": {
350
+ "weight": 1.0,
351
+ "type": "SharpnessJitter",
352
+ "kwargs": {
353
+ "sharpness": [
354
+ 0.5,
355
+ 1.5
356
+ ]
357
+ }
358
+ },
359
+ "affine": {
360
+ "weight": 1.0,
361
+ "type": "RandomAffine",
362
+ "kwargs": {
363
+ "degrees": [
364
+ -5.0,
365
+ 5.0
366
+ ],
367
+ "translate": [
368
+ 0.05,
369
+ 0.05
370
+ ]
371
+ }
372
+ }
373
+ }
374
+ },
375
+ "revision": null,
376
+ "use_imagenet_stats": true,
377
+ "video_backend": "torchcodec",
378
+ "streaming": false
379
+ },
380
+ "wandb": {
381
+ "enable": true,
382
+ "disable_artifact": true,
383
+ "project": "dexmimicgen_dp",
384
+ "entity": null,
385
+ "notes": null,
386
+ "run_id": "j9s1kgs8",
387
+ "mode": null
388
+ },
389
+ "peft": null,
390
+ "use_rabc": false,
391
+ "rabc_progress_path": null,
392
+ "rabc_kappa": 0.01,
393
+ "rabc_epsilon": 1e-06,
394
+ "rabc_head_mode": "sparse",
395
+ "rename_map": {},
396
+ "checkpoint_path": null
397
+ }