Hugging Face's logo

RoboEval
/
roboeval-diffusion

Robotics
Safetensors
roboeval
diffusion
Model card Files
xet
 Community
helen9975 commited on
Commit
c7d32bb
·
verified ·
1 Parent(s): 6e4e646

StackSingleBookShelf: step 005000 weights

Browse files
Files changed (7) hide show
  1. StackSingleBookShelf/config.json +99 -0
  2. StackSingleBookShelf/model.safetensors +3 -0
  3. StackSingleBookShelf/policy_postprocessor.json +32 -0
  4. StackSingleBookShelf/policy_postprocessor_step_0_unnormalizer_processor.safetensors +3 -0
  5. StackSingleBookShelf/policy_preprocessor.json +72 -0
  6. StackSingleBookShelf/policy_preprocessor_step_3_normalizer_processor.safetensors +3 -0
  7. StackSingleBookShelf/train_config.json +436 -0
StackSingleBookShelf/config.json ADDED
@@ -0,0 +1,99 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "type": "diffusion",
3
+ "n_obs_steps": 2,
4
+ "input_features": {
5
+ "observation.images.cam_head": {
6
+ "type": "VISUAL",
7
+ "shape": [
8
+ 3,
9
+ 256,
10
+ 256
11
+ ]
12
+ },
13
+ "observation.images.cam_left_wrist": {
14
+ "type": "VISUAL",
15
+ "shape": [
16
+ 3,
17
+ 256,
18
+ 256
19
+ ]
20
+ },
21
+ "observation.images.cam_right_wrist": {
22
+ "type": "VISUAL",
23
+ "shape": [
24
+ 3,
25
+ 256,
26
+ 256
27
+ ]
28
+ },
29
+ "observation.state": {
30
+ "type": "STATE",
31
+ "shape": [
32
+ 16
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": false,
48
+ "repo_id": null,
49
+ "private": null,
50
+ "tags": null,
51
+ "license": null,
52
+ "pretrained_path": null,
53
+ "horizon": 16,
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
+ 84,
64
+ 84
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
+ 1024,
74
+ 2048
75
+ ],
76
+ "kernel_size": 5,
77
+ "n_groups": 8,
78
+ "diffusion_step_embed_dim": 128,
79
+ "use_film_scale_modulation": true,
80
+ "noise_scheduler_type": "DDPM",
81
+ "num_train_timesteps": 100,
82
+ "beta_schedule": "squaredcos_cap_v2",
83
+ "beta_start": 0.0001,
84
+ "beta_end": 0.02,
85
+ "prediction_type": "epsilon",
86
+ "clip_sample": true,
87
+ "clip_sample_range": 1.0,
88
+ "num_inference_steps": null,
89
+ "do_mask_loss_for_padding": false,
90
+ "optimizer_lr": 0.0001,
91
+ "optimizer_betas": [
92
+ 0.95,
93
+ 0.999
94
+ ],
95
+ "optimizer_eps": 1e-08,
96
+ "optimizer_weight_decay": 1e-06,
97
+ "scheduler_name": "cosine",
98
+ "scheduler_warmup_steps": 500
99
+ }
StackSingleBookShelf/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0149c5c6d5a7e5e58e7ad6407c283a84c0fbdda83ac7bf4d10feb2c310aca11a
3
+ size 1083604936
StackSingleBookShelf/policy_postprocessor.json ADDED
@@ -0,0 +1,32 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
+ 14
13
+ ]
14
+ }
15
+ },
16
+ "norm_map": {
17
+ "VISUAL": "MEAN_STD",
18
+ "STATE": "MIN_MAX",
19
+ "ACTION": "MIN_MAX"
20
+ }
21
+ },
22
+ "state_file": "policy_postprocessor_step_0_unnormalizer_processor.safetensors"
23
+ },
24
+ {
25
+ "registry_name": "device_processor",
26
+ "config": {
27
+ "device": "cpu",
28
+ "float_dtype": null
29
+ }
30
+ }
31
+ ]
32
+ }
StackSingleBookShelf/policy_postprocessor_step_0_unnormalizer_processor.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ce357734e4942a37891e12d1d1d546efa14f70d156003dd33255010a73c28c0c
3
+ size 16036
StackSingleBookShelf/policy_preprocessor.json ADDED
@@ -0,0 +1,72 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": "device_processor",
16
+ "config": {
17
+ "device": "cuda",
18
+ "float_dtype": null
19
+ }
20
+ },
21
+ {
22
+ "registry_name": "normalizer_processor",
23
+ "config": {
24
+ "eps": 1e-08,
25
+ "features": {
26
+ "observation.images.cam_head": {
27
+ "type": "VISUAL",
28
+ "shape": [
29
+ 3,
30
+ 256,
31
+ 256
32
+ ]
33
+ },
34
+ "observation.images.cam_left_wrist": {
35
+ "type": "VISUAL",
36
+ "shape": [
37
+ 3,
38
+ 256,
39
+ 256
40
+ ]
41
+ },
42
+ "observation.images.cam_right_wrist": {
43
+ "type": "VISUAL",
44
+ "shape": [
45
+ 3,
46
+ 256,
47
+ 256
48
+ ]
49
+ },
50
+ "observation.state": {
51
+ "type": "STATE",
52
+ "shape": [
53
+ 16
54
+ ]
55
+ },
56
+ "action": {
57
+ "type": "ACTION",
58
+ "shape": [
59
+ 14
60
+ ]
61
+ }
62
+ },
63
+ "norm_map": {
64
+ "VISUAL": "MEAN_STD",
65
+ "STATE": "MIN_MAX",
66
+ "ACTION": "MIN_MAX"
67
+ }
68
+ },
69
+ "state_file": "policy_preprocessor_step_3_normalizer_processor.safetensors"
70
+ }
71
+ ]
72
+ }
StackSingleBookShelf/policy_preprocessor_step_3_normalizer_processor.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ce357734e4942a37891e12d1d1d546efa14f70d156003dd33255010a73c28c0c
3
+ size 16036
StackSingleBookShelf/train_config.json ADDED
@@ -0,0 +1,436 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "roboeval_ee_delta_20hz_v3_merged",
4
+ "root": "/weka/helenw/data/huggingface/lerobot/roboeval_ee_delta_20hz_v3_merged",
5
+ "episodes": [
6
+ 2236,
7
+ 2237,
8
+ 2238,
9
+ 2239,
10
+ 2240,
11
+ 2241,
12
+ 2242,
13
+ 2243,
14
+ 2244,
15
+ 2245,
16
+ 2246,
17
+ 2247,
18
+ 2248,
19
+ 2249,
20
+ 2250,
21
+ 2251,
22
+ 2252,
23
+ 2253,
24
+ 2254,
25
+ 2255,
26
+ 2256,
27
+ 2257,
28
+ 2258,
29
+ 2259,
30
+ 2260,
31
+ 2261,
32
+ 2262,
33
+ 2263,
34
+ 2264,
35
+ 2265,
36
+ 2266,
37
+ 2267,
38
+ 2268,
39
+ 2269,
40
+ 2270,
41
+ 2271,
42
+ 2272,
43
+ 2273,
44
+ 2274,
45
+ 2275,
46
+ 2276,
47
+ 2277,
48
+ 2278,
49
+ 2279,
50
+ 2280,
51
+ 2281,
52
+ 2282,
53
+ 2283,
54
+ 2284,
55
+ 2285,
56
+ 2286,
57
+ 2287,
58
+ 2288,
59
+ 2289,
60
+ 2290,
61
+ 2291,
62
+ 2292,
63
+ 2293,
64
+ 2294,
65
+ 2295,
66
+ 2296,
67
+ 2297,
68
+ 2298,
69
+ 2299,
70
+ 2300,
71
+ 2301,
72
+ 2302,
73
+ 2303,
74
+ 2304,
75
+ 2305,
76
+ 2306,
77
+ 2307,
78
+ 2308,
79
+ 2309,
80
+ 2310,
81
+ 2311,
82
+ 2312,
83
+ 2313,
84
+ 2314,
85
+ 2315,
86
+ 2316,
87
+ 2317,
88
+ 2318,
89
+ 2319,
90
+ 2320,
91
+ 2321,
92
+ 2322,
93
+ 2323,
94
+ 2324,
95
+ 2325,
96
+ 2326,
97
+ 2327,
98
+ 2328,
99
+ 2329,
100
+ 2330,
101
+ 2331,
102
+ 2332,
103
+ 2333,
104
+ 2334,
105
+ 2335,
106
+ 2336,
107
+ 2337,
108
+ 2338,
109
+ 2339,
110
+ 2340,
111
+ 2341,
112
+ 2342,
113
+ 2343,
114
+ 2344,
115
+ 2345,
116
+ 2346,
117
+ 2347,
118
+ 2348,
119
+ 2349,
120
+ 2350,
121
+ 2351,
122
+ 2352,
123
+ 2353,
124
+ 2354,
125
+ 2355,
126
+ 2356,
127
+ 2357,
128
+ 2358,
129
+ 2359,
130
+ 2360,
131
+ 2361,
132
+ 2362,
133
+ 2363,
134
+ 2364,
135
+ 2365,
136
+ 2366,
137
+ 2367,
138
+ 2368,
139
+ 2369,
140
+ 2370,
141
+ 2371,
142
+ 2372,
143
+ 2373,
144
+ 2374,
145
+ 2375,
146
+ 2376,
147
+ 2377,
148
+ 2378,
149
+ 2379,
150
+ 2380,
151
+ 2381,
152
+ 2382,
153
+ 2383,
154
+ 2384,
155
+ 2385,
156
+ 2386,
157
+ 2387,
158
+ 2388,
159
+ 2389,
160
+ 2390,
161
+ 2391,
162
+ 2392,
163
+ 2393,
164
+ 2394,
165
+ 2395,
166
+ 2396,
167
+ 2397,
168
+ 2398,
169
+ 2399,
170
+ 2400,
171
+ 2401,
172
+ 2402,
173
+ 2403,
174
+ 2404,
175
+ 2405,
176
+ 2406,
177
+ 2407,
178
+ 2408,
179
+ 2409,
180
+ 2410,
181
+ 2411,
182
+ 2412,
183
+ 2413,
184
+ 2414,
185
+ 2415,
186
+ 2416,
187
+ 2417,
188
+ 2418,
189
+ 2419,
190
+ 2420,
191
+ 2421,
192
+ 2422,
193
+ 2423,
194
+ 2424,
195
+ 2425,
196
+ 2426,
197
+ 2427,
198
+ 2428,
199
+ 2429,
200
+ 2430,
201
+ 2431,
202
+ 2432,
203
+ 2433,
204
+ 2434,
205
+ 2435,
206
+ 2436,
207
+ 2437
208
+ ],
209
+ "image_transforms": {
210
+ "enable": true,
211
+ "max_num_transforms": 3,
212
+ "random_order": false,
213
+ "tfs": {
214
+ "brightness": {
215
+ "weight": 1.0,
216
+ "type": "ColorJitter",
217
+ "kwargs": {
218
+ "brightness": [
219
+ 0.8,
220
+ 1.2
221
+ ]
222
+ }
223
+ },
224
+ "contrast": {
225
+ "weight": 1.0,
226
+ "type": "ColorJitter",
227
+ "kwargs": {
228
+ "contrast": [
229
+ 0.8,
230
+ 1.2
231
+ ]
232
+ }
233
+ },
234
+ "saturation": {
235
+ "weight": 1.0,
236
+ "type": "ColorJitter",
237
+ "kwargs": {
238
+ "saturation": [
239
+ 0.5,
240
+ 1.5
241
+ ]
242
+ }
243
+ },
244
+ "hue": {
245
+ "weight": 1.0,
246
+ "type": "ColorJitter",
247
+ "kwargs": {
248
+ "hue": [
249
+ -0.05,
250
+ 0.05
251
+ ]
252
+ }
253
+ },
254
+ "sharpness": {
255
+ "weight": 1.0,
256
+ "type": "SharpnessJitter",
257
+ "kwargs": {
258
+ "sharpness": [
259
+ 0.5,
260
+ 1.5
261
+ ]
262
+ }
263
+ },
264
+ "affine": {
265
+ "weight": 1.0,
266
+ "type": "RandomAffine",
267
+ "kwargs": {
268
+ "degrees": [
269
+ -5.0,
270
+ 5.0
271
+ ],
272
+ "translate": [
273
+ 0.05,
274
+ 0.05
275
+ ]
276
+ }
277
+ }
278
+ }
279
+ },
280
+ "revision": "local",
281
+ "use_imagenet_stats": true,
282
+ "video_backend": "torchcodec",
283
+ "streaming": false
284
+ },
285
+ "env": null,
286
+ "policy": {
287
+ "type": "diffusion",
288
+ "n_obs_steps": 2,
289
+ "input_features": {
290
+ "observation.images.cam_head": {
291
+ "type": "VISUAL",
292
+ "shape": [
293
+ 3,
294
+ 256,
295
+ 256
296
+ ]
297
+ },
298
+ "observation.images.cam_left_wrist": {
299
+ "type": "VISUAL",
300
+ "shape": [
301
+ 3,
302
+ 256,
303
+ 256
304
+ ]
305
+ },
306
+ "observation.images.cam_right_wrist": {
307
+ "type": "VISUAL",
308
+ "shape": [
309
+ 3,
310
+ 256,
311
+ 256
312
+ ]
313
+ },
314
+ "observation.state": {
315
+ "type": "STATE",
316
+ "shape": [
317
+ 16
318
+ ]
319
+ }
320
+ },
321
+ "output_features": {
322
+ "action": {
323
+ "type": "ACTION",
324
+ "shape": [
325
+ 14
326
+ ]
327
+ }
328
+ },
329
+ "device": "cuda",
330
+ "use_amp": false,
331
+ "use_peft": false,
332
+ "push_to_hub": false,
333
+ "repo_id": null,
334
+ "private": null,
335
+ "tags": null,
336
+ "license": null,
337
+ "pretrained_path": null,
338
+ "horizon": 16,
339
+ "n_action_steps": 8,
340
+ "normalization_mapping": {
341
+ "VISUAL": "MEAN_STD",
342
+ "STATE": "MIN_MAX",
343
+ "ACTION": "MIN_MAX"
344
+ },
345
+ "drop_n_last_frames": 7,
346
+ "vision_backbone": "resnet18",
347
+ "crop_shape": [
348
+ 84,
349
+ 84
350
+ ],
351
+ "crop_is_random": true,
352
+ "pretrained_backbone_weights": null,
353
+ "use_group_norm": true,
354
+ "spatial_softmax_num_keypoints": 32,
355
+ "use_separate_rgb_encoder_per_camera": false,
356
+ "down_dims": [
357
+ 512,
358
+ 1024,
359
+ 2048
360
+ ],
361
+ "kernel_size": 5,
362
+ "n_groups": 8,
363
+ "diffusion_step_embed_dim": 128,
364
+ "use_film_scale_modulation": true,
365
+ "noise_scheduler_type": "DDPM",
366
+ "num_train_timesteps": 100,
367
+ "beta_schedule": "squaredcos_cap_v2",
368
+ "beta_start": 0.0001,
369
+ "beta_end": 0.02,
370
+ "prediction_type": "epsilon",
371
+ "clip_sample": true,
372
+ "clip_sample_range": 1.0,
373
+ "num_inference_steps": null,
374
+ "do_mask_loss_for_padding": false,
375
+ "optimizer_lr": 0.0001,
376
+ "optimizer_betas": [
377
+ 0.95,
378
+ 0.999
379
+ ],
380
+ "optimizer_eps": 1e-08,
381
+ "optimizer_weight_decay": 1e-06,
382
+ "scheduler_name": "cosine",
383
+ "scheduler_warmup_steps": 500
384
+ },
385
+ "output_dir": "/weka/helenw/outputs/diffusion_StackSingleBookShelf",
386
+ "job_name": "diffusion_roboeval_v3_merged",
387
+ "resume": false,
388
+ "seed": 1000,
389
+ "num_workers": 2,
390
+ "batch_size": 256,
391
+ "steps": 5000,
392
+ "eval_freq": 20000,
393
+ "log_freq": 10,
394
+ "tolerance_s": 0.0001,
395
+ "save_checkpoint": true,
396
+ "save_freq": 1000,
397
+ "use_policy_training_preset": true,
398
+ "optimizer": {
399
+ "type": "adam",
400
+ "lr": 0.0001,
401
+ "weight_decay": 1e-06,
402
+ "grad_clip_norm": 10.0,
403
+ "betas": [
404
+ 0.95,
405
+ 0.999
406
+ ],
407
+ "eps": 1e-08
408
+ },
409
+ "scheduler": {
410
+ "type": "diffuser",
411
+ "num_warmup_steps": 500,
412
+ "name": "cosine"
413
+ },
414
+ "eval": {
415
+ "n_episodes": 50,
416
+ "batch_size": 50,
417
+ "use_async_envs": false
418
+ },
419
+ "wandb": {
420
+ "enable": true,
421
+ "disable_artifact": true,
422
+ "project": "lerobot",
423
+ "entity": null,
424
+ "notes": null,
425
+ "run_id": "o2kvb2re",
426
+ "mode": null
427
+ },
428
+ "peft": null,
429
+ "use_rabc": false,
430
+ "rabc_progress_path": null,
431
+ "rabc_kappa": 0.01,
432
+ "rabc_epsilon": 1e-06,
433
+ "rabc_head_mode": "sparse",
434
+ "rename_map": {},
435
+ "checkpoint_path": null
436
+ }