Hugging Face's logo

chomeed
/
robomimic_square_ph

Robotics
LeRobot
Safetensors
diffusion
Model card Files
xet
 Community
chomeed commited on
Commit
6fb1f20
·
verified ·
1 Parent(s): f0cc6eb

Upload policy weights, train config and readme

Browse files
Files changed (4) hide show
  1. README.md +62 -0
  2. config.json +89 -0
  3. model.safetensors +3 -0
  4. train_config.json +389 -0
README.md ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ datasets: chomeed/robomimic_square_ph
3
+ library_name: lerobot
4
+ license: apache-2.0
5
+ model_name: diffusion
6
+ pipeline_tag: robotics
7
+ tags:
8
+ - robotics
9
+ - lerobot
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 +1,89 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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.eye_in_hand": {
14
+ "type": "VISUAL",
15
+ "shape": [
16
+ 3,
17
+ 84,
18
+ 84
19
+ ]
20
+ },
21
+ "observation.state": {
22
+ "type": "STATE",
23
+ "shape": [
24
+ 9
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": true,
40
+ "repo_id": "chomeed/robomimic_square_ph",
41
+ "private": null,
42
+ "tags": null,
43
+ "license": null,
44
+ "pretrained_path": null,
45
+ "horizon": 8,
46
+ "n_action_steps": 8,
47
+ "normalization_mapping": {
48
+ "VISUAL": "MEAN_STD",
49
+ "STATE": "MIN_MAX",
50
+ "ACTION": "MIN_MAX"
51
+ },
52
+ "drop_n_last_frames": 7,
53
+ "vision_backbone": "resnet18",
54
+ "crop_shape": [
55
+ 76,
56
+ 76
57
+ ],
58
+ "crop_is_random": true,
59
+ "pretrained_backbone_weights": null,
60
+ "use_group_norm": true,
61
+ "spatial_softmax_num_keypoints": 32,
62
+ "use_separate_rgb_encoder_per_camera": false,
63
+ "down_dims": [
64
+ 512
65
+ ],
66
+ "kernel_size": 5,
67
+ "n_groups": 8,
68
+ "diffusion_step_embed_dim": 128,
69
+ "use_film_scale_modulation": true,
70
+ "noise_scheduler_type": "DDPM",
71
+ "num_train_timesteps": 100,
72
+ "beta_schedule": "squaredcos_cap_v2",
73
+ "beta_start": 0.0001,
74
+ "beta_end": 0.02,
75
+ "prediction_type": "epsilon",
76
+ "clip_sample": true,
77
+ "clip_sample_range": 1.0,
78
+ "num_inference_steps": null,
79
+ "do_mask_loss_for_padding": false,
80
+ "optimizer_lr": 0.0001,
81
+ "optimizer_betas": [
82
+ 0.95,
83
+ 0.999
84
+ ],
85
+ "optimizer_eps": 1e-08,
86
+ "optimizer_weight_decay": 1e-06,
87
+ "scheduler_name": "cosine",
88
+ "scheduler_warmup_steps": 500
89
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a6b97ff6a694846f5fe71f32bde9fa1f18c643ab146a9e416c46fd93cc90352f
3
+ size 94032268
train_config.json ADDED
@@ -0,0 +1,389 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "chomeed/robomimic_square_ph",
4
+ "root": "/home/chomeed/workspace/insertion_generation/datasets/lerobot/robomimic_square_ph",
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.eye_in_hand": {
97
+ "type": "VISUAL",
98
+ "shape": [
99
+ 3,
100
+ 84,
101
+ 84
102
+ ]
103
+ },
104
+ "observation.state": {
105
+ "type": "STATE",
106
+ "shape": [
107
+ 9
108
+ ]
109
+ }
110
+ },
111
+ "output_features": {
112
+ "action": {
113
+ "type": "ACTION",
114
+ "shape": [
115
+ 7
116
+ ]
117
+ }
118
+ },
119
+ "device": "cuda",
120
+ "use_amp": false,
121
+ "use_peft": false,
122
+ "push_to_hub": true,
123
+ "repo_id": "chomeed/robomimic_square_ph",
124
+ "private": null,
125
+ "tags": null,
126
+ "license": null,
127
+ "pretrained_path": null,
128
+ "horizon": 8,
129
+ "n_action_steps": 8,
130
+ "normalization_mapping": {
131
+ "VISUAL": "MEAN_STD",
132
+ "STATE": "MIN_MAX",
133
+ "ACTION": "MIN_MAX"
134
+ },
135
+ "drop_n_last_frames": 7,
136
+ "vision_backbone": "resnet18",
137
+ "crop_shape": [
138
+ 76,
139
+ 76
140
+ ],
141
+ "crop_is_random": true,
142
+ "pretrained_backbone_weights": null,
143
+ "use_group_norm": true,
144
+ "spatial_softmax_num_keypoints": 32,
145
+ "use_separate_rgb_encoder_per_camera": false,
146
+ "down_dims": [
147
+ 512
148
+ ],
149
+ "kernel_size": 5,
150
+ "n_groups": 8,
151
+ "diffusion_step_embed_dim": 128,
152
+ "use_film_scale_modulation": true,
153
+ "noise_scheduler_type": "DDPM",
154
+ "num_train_timesteps": 100,
155
+ "beta_schedule": "squaredcos_cap_v2",
156
+ "beta_start": 0.0001,
157
+ "beta_end": 0.02,
158
+ "prediction_type": "epsilon",
159
+ "clip_sample": true,
160
+ "clip_sample_range": 1.0,
161
+ "num_inference_steps": null,
162
+ "do_mask_loss_for_padding": false,
163
+ "optimizer_lr": 0.0001,
164
+ "optimizer_betas": [
165
+ 0.95,
166
+ 0.999
167
+ ],
168
+ "optimizer_eps": 1e-08,
169
+ "optimizer_weight_decay": 1e-06,
170
+ "scheduler_name": "cosine",
171
+ "scheduler_warmup_steps": 500
172
+ },
173
+ "output_dir": "/scratch2/chomeed/outputs/train/robomimic_square_ph_dp_Tp8_Ta8",
174
+ "job_name": "robomimic_square_ph_dp_Tp8_Ta8",
175
+ "resume": false,
176
+ "seed": 1000,
177
+ "num_workers": 8,
178
+ "batch_size": 64,
179
+ "steps": 200000,
180
+ "eval_freq": 0,
181
+ "log_freq": 200,
182
+ "tolerance_s": 0.0001,
183
+ "save_checkpoint": true,
184
+ "save_freq": 10000,
185
+ "use_policy_training_preset": true,
186
+ "optimizer": {
187
+ "type": "adam",
188
+ "lr": 0.0001,
189
+ "weight_decay": 1e-06,
190
+ "grad_clip_norm": 10.0,
191
+ "betas": [
192
+ 0.95,
193
+ 0.999
194
+ ],
195
+ "eps": 1e-08
196
+ },
197
+ "scheduler": {
198
+ "type": "diffuser",
199
+ "num_warmup_steps": 500,
200
+ "name": "cosine"
201
+ },
202
+ "val": {
203
+ "repo_id": [
204
+ "chomeed/anubis_pick_carrot_out_of_pot_randomized_val_single",
205
+ "chomeed/anubis_stack_tuna_single",
206
+ "chomeed/aloha_static_cups_open_single",
207
+ "chomeed/aloha_static_candy_single",
208
+ "jellyho/aloha_box_into_pot_single",
209
+ "jellyho/aloha_dish_drainer_single"
210
+ ],
211
+ "root": null,
212
+ "episodes": null,
213
+ "image_transforms": {
214
+ "enable": false,
215
+ "max_num_transforms": 3,
216
+ "random_order": false,
217
+ "tfs": {
218
+ "brightness": {
219
+ "weight": 1.0,
220
+ "type": "ColorJitter",
221
+ "kwargs": {
222
+ "brightness": [
223
+ 0.8,
224
+ 1.2
225
+ ]
226
+ }
227
+ },
228
+ "contrast": {
229
+ "weight": 1.0,
230
+ "type": "ColorJitter",
231
+ "kwargs": {
232
+ "contrast": [
233
+ 0.8,
234
+ 1.2
235
+ ]
236
+ }
237
+ },
238
+ "saturation": {
239
+ "weight": 1.0,
240
+ "type": "ColorJitter",
241
+ "kwargs": {
242
+ "saturation": [
243
+ 0.5,
244
+ 1.5
245
+ ]
246
+ }
247
+ },
248
+ "hue": {
249
+ "weight": 1.0,
250
+ "type": "ColorJitter",
251
+ "kwargs": {
252
+ "hue": [
253
+ -0.05,
254
+ 0.05
255
+ ]
256
+ }
257
+ },
258
+ "sharpness": {
259
+ "weight": 1.0,
260
+ "type": "SharpnessJitter",
261
+ "kwargs": {
262
+ "sharpness": [
263
+ 0.5,
264
+ 1.5
265
+ ]
266
+ }
267
+ },
268
+ "affine": {
269
+ "weight": 1.0,
270
+ "type": "RandomAffine",
271
+ "kwargs": {
272
+ "degrees": [
273
+ -5.0,
274
+ 5.0
275
+ ],
276
+ "translate": [
277
+ 0.05,
278
+ 0.05
279
+ ]
280
+ }
281
+ }
282
+ }
283
+ },
284
+ "revision": null,
285
+ "use_imagenet_stats": true,
286
+ "video_backend": "torchcodec",
287
+ "streaming": false
288
+ },
289
+ "test": {
290
+ "repo_id": [
291
+ "chomeed/aloha_static_coffee_single",
292
+ "jellyho/aloha_handover_box_single"
293
+ ],
294
+ "root": null,
295
+ "episodes": null,
296
+ "image_transforms": {
297
+ "enable": false,
298
+ "max_num_transforms": 3,
299
+ "random_order": false,
300
+ "tfs": {
301
+ "brightness": {
302
+ "weight": 1.0,
303
+ "type": "ColorJitter",
304
+ "kwargs": {
305
+ "brightness": [
306
+ 0.8,
307
+ 1.2
308
+ ]
309
+ }
310
+ },
311
+ "contrast": {
312
+ "weight": 1.0,
313
+ "type": "ColorJitter",
314
+ "kwargs": {
315
+ "contrast": [
316
+ 0.8,
317
+ 1.2
318
+ ]
319
+ }
320
+ },
321
+ "saturation": {
322
+ "weight": 1.0,
323
+ "type": "ColorJitter",
324
+ "kwargs": {
325
+ "saturation": [
326
+ 0.5,
327
+ 1.5
328
+ ]
329
+ }
330
+ },
331
+ "hue": {
332
+ "weight": 1.0,
333
+ "type": "ColorJitter",
334
+ "kwargs": {
335
+ "hue": [
336
+ -0.05,
337
+ 0.05
338
+ ]
339
+ }
340
+ },
341
+ "sharpness": {
342
+ "weight": 1.0,
343
+ "type": "SharpnessJitter",
344
+ "kwargs": {
345
+ "sharpness": [
346
+ 0.5,
347
+ 1.5
348
+ ]
349
+ }
350
+ },
351
+ "affine": {
352
+ "weight": 1.0,
353
+ "type": "RandomAffine",
354
+ "kwargs": {
355
+ "degrees": [
356
+ -5.0,
357
+ 5.0
358
+ ],
359
+ "translate": [
360
+ 0.05,
361
+ 0.05
362
+ ]
363
+ }
364
+ }
365
+ }
366
+ },
367
+ "revision": null,
368
+ "use_imagenet_stats": true,
369
+ "video_backend": "torchcodec",
370
+ "streaming": false
371
+ },
372
+ "wandb": {
373
+ "enable": false,
374
+ "disable_artifact": true,
375
+ "project": "robomimic_square_dp",
376
+ "entity": null,
377
+ "notes": null,
378
+ "run_id": null,
379
+ "mode": null
380
+ },
381
+ "peft": null,
382
+ "use_rabc": false,
383
+ "rabc_progress_path": null,
384
+ "rabc_kappa": 0.01,
385
+ "rabc_epsilon": 1e-06,
386
+ "rabc_head_mode": "sparse",
387
+ "rename_map": {},
388
+ "checkpoint_path": null
389
+ }