Hugging Face's logo

Vincent2025hello
/
u0_final

Robotics
Safetensors
English
gr00t_n1_5
vla
underwater robot
Model card Files
xet
 Community
Vincent2025hello commited on
Commit
8fbac60
·
verified ·
1 Parent(s): ce13025

Add files using upload-large-folder tool

Browse files
Files changed (7) hide show
  1. config.json +64 -0
  2. experiment_cfg/metadata.json +569 -0
  3. model-00001-of-00002.safetensors +3 -0
  4. model-00002-of-00002.safetensors +3 -0
  5. model.safetensors.index.json +0 -0
  6. trainer_state.json +0 -0
  7. training_args.bin +3 -0
config.json ADDED
@@ -0,0 +1,64 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "action_dim": 32,
3
+ "action_head_cfg": {
4
+ "action_dim": 32,
5
+ "action_horizon": 16,
6
+ "add_pos_embed": true,
7
+ "backbone_embedding_dim": 2048,
8
+ "diffusion_model_cfg": {
9
+ "attention_head_dim": 48,
10
+ "cross_attention_dim": 2048,
11
+ "dropout": 0.2,
12
+ "final_dropout": true,
13
+ "interleave_self_attention": true,
14
+ "norm_type": "ada_norm",
15
+ "num_attention_heads": 32,
16
+ "num_layers": 16,
17
+ "output_dim": 1024,
18
+ "positional_embeddings": null
19
+ },
20
+ "hidden_size": 1024,
21
+ "input_embedding_dim": 1536,
22
+ "max_action_dim": 32,
23
+ "max_state_dim": 64,
24
+ "model_dtype": "float32",
25
+ "noise_beta_alpha": 1.5,
26
+ "noise_beta_beta": 1.0,
27
+ "noise_s": 0.999,
28
+ "num_inference_timesteps": 4,
29
+ "num_target_vision_tokens": 32,
30
+ "num_timestep_buckets": 1000,
31
+ "tune_diffusion_model": true,
32
+ "tune_projector": true,
33
+ "use_vlln": true,
34
+ "vl_self_attention_cfg": {
35
+ "attention_head_dim": 64,
36
+ "dropout": 0.2,
37
+ "final_dropout": true,
38
+ "num_attention_heads": 32,
39
+ "num_layers": 4,
40
+ "positional_embeddings": null
41
+ }
42
+ },
43
+ "action_horizon": 16,
44
+ "architectures": [
45
+ "GR00T_N1_5"
46
+ ],
47
+ "attn_implementation": null,
48
+ "backbone_cfg": {
49
+ "eagle_path": "NVEagle/eagle_er-qwen3_1_7B-Siglip2_400M_stage1_5_128gpu_er_v7_1mlp_nops",
50
+ "load_bf16": false,
51
+ "project_to_dim": null,
52
+ "reproject_vision": false,
53
+ "select_layer": 12,
54
+ "tune_llm": false,
55
+ "tune_visual": true,
56
+ "use_flash_attention": true
57
+ },
58
+ "compute_dtype": "bfloat16",
59
+ "hidden_size": 2048,
60
+ "model_dtype": "float32",
61
+ "model_type": "gr00t_n1_5",
62
+ "torch_dtype": "bfloat16",
63
+ "transformers_version": "4.51.3"
64
+ }
experiment_cfg/metadata.json ADDED
@@ -0,0 +1,569 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "new_embodiment": {
3
+ "statistics": {
4
+ "state": {
5
+ "joint_pos": {
6
+ "max": [
7
+ 0.01499609462916851,
8
+ 0.4975312352180481,
9
+ 0.49741339683532715,
10
+ 0.5057408809661865,
11
+ 0.00013245544687379152
12
+ ],
13
+ "min": [
14
+ -0.001301140058785677,
15
+ -4.4340096792439e-05,
16
+ -1.8281958773513907e-06,
17
+ -8.687281479069497e-06,
18
+ -6.792588101234287e-05
19
+ ],
20
+ "mean": [
21
+ 0.003374873660504818,
22
+ 0.3498944938182831,
23
+ 0.3498944938182831,
24
+ 0.35298940539360046,
25
+ 3.5255197872174904e-05
26
+ ],
27
+ "std": [
28
+ 0.006213217508047819,
29
+ 0.22534441947937012,
30
+ 0.22534498572349548,
31
+ 0.22973771393299103,
32
+ 3.3011066989274696e-05
33
+ ],
34
+ "q01": [
35
+ 0.0,
36
+ 0.0,
37
+ 0.0,
38
+ 0.0,
39
+ 0.0
40
+ ],
41
+ "q99": [
42
+ 0.014937113039195538,
43
+ 0.49659860134124756,
44
+ 0.49659767746925354,
45
+ 0.5056976079940796,
46
+ 9.855335018073673e-05
47
+ ]
48
+ },
49
+ "pwm": {
50
+ "max": [
51
+ 1.0,
52
+ 1.0,
53
+ 1.0,
54
+ 1.0,
55
+ 1.0,
56
+ 1.0,
57
+ 1.0,
58
+ 1.0
59
+ ],
60
+ "min": [
61
+ -1.0,
62
+ -1.0,
63
+ -1.0,
64
+ -1.0,
65
+ -1.0,
66
+ -1.0,
67
+ -1.0,
68
+ -1.0
69
+ ],
70
+ "mean": [
71
+ -0.1696692854166031,
72
+ -0.1136094480752945,
73
+ 0.10162586718797684,
74
+ 0.17456766963005066,
75
+ -0.07031124830245972,
76
+ -0.034409455955028534,
77
+ 0.046888984739780426,
78
+ 0.08268629014492035
79
+ ],
80
+ "std": [
81
+ 0.4997708201408386,
82
+ 0.5126947164535522,
83
+ 0.5230556726455688,
84
+ 0.5092427134513855,
85
+ 0.25611206889152527,
86
+ 0.2508251667022705,
87
+ 0.24924014508724213,
88
+ 0.2535521388053894
89
+ ],
90
+ "q01": [
91
+ -1.0,
92
+ -1.0,
93
+ -1.0,
94
+ -1.0,
95
+ -1.0,
96
+ -1.0,
97
+ -1.0,
98
+ -1.0
99
+ ],
100
+ "q99": [
101
+ 1.0,
102
+ 1.0,
103
+ 1.0,
104
+ 1.0,
105
+ 1.0,
106
+ 1.0,
107
+ 1.0,
108
+ 1.0
109
+ ]
110
+ },
111
+ "joint_v": {
112
+ "max": [
113
+ 0.18002620339393616,
114
+ 0.5273282527923584,
115
+ 0.5,
116
+ 0.5,
117
+ 0.022483378648757935
118
+ ],
119
+ "min": [
120
+ -0.13923990726470947,
121
+ -0.0582619234919548,
122
+ -0.058066677302122116,
123
+ -0.2464992105960846,
124
+ -0.010273344814777374
125
+ ],
126
+ "mean": [
127
+ 0.00025962013751268387,
128
+ 0.009276977740228176,
129
+ 0.009274156764149666,
130
+ 0.009403317235410213,
131
+ 7.972354296725825e-07
132
+ ],
133
+ "std": [
134
+ 0.0049674357287585735,
135
+ 0.06713016331195831,
136
+ 0.06711767613887787,
137
+ 0.06760227680206299,
138
+ 0.00011152630031574517
139
+ ],
140
+ "q01": [
141
+ -0.00016139761486556382,
142
+ -0.0019489859917666762,
143
+ -0.0018264227593317627,
144
+ -0.0006048974255099893,
145
+ -9.230952651705593e-05
146
+ ],
147
+ "q99": [
148
+ 0.00036396432551555314,
149
+ 0.5,
150
+ 0.5,
151
+ 0.5,
152
+ 0.00010598917993775086
153
+ ]
154
+ },
155
+ "dvl_v": {
156
+ "max": [
157
+ 0.8632012605667114,
158
+ 0.6997467875480652,
159
+ 0.5916000604629517
160
+ ],
161
+ "min": [
162
+ -0.6001196503639221,
163
+ -0.6861761808395386,
164
+ -0.576388418674469
165
+ ],
166
+ "mean": [
167
+ 0.09221230447292328,
168
+ 0.022050414234399796,
169
+ -0.0038110604509711266
170
+ ],
171
+ "std": [
172
+ 0.21494390070438385,
173
+ 0.14589449763298035,
174
+ 0.05585192143917084
175
+ ],
176
+ "q01": [
177
+ -0.33405666291713715,
178
+ -0.37028490006923676,
179
+ -0.1664176881313324
180
+ ],
181
+ "q99": [
182
+ 0.7776143550872803,
183
+ 0.6105309724807739,
184
+ 0.1963522732257843
185
+ ]
186
+ },
187
+ "imu_av": {
188
+ "max": [
189
+ 1.191752314567566,
190
+ 1.4354997873306274,
191
+ 3.000467538833618
192
+ ],
193
+ "min": [
194
+ -1.1869069337844849,
195
+ -0.9643204212188721,
196
+ -2.1809566020965576
197
+ ],
198
+ "mean": [
199
+ -0.0002232321712654084,
200
+ 0.0002486517187207937,
201
+ 0.0041033485904335976
202
+ ],
203
+ "std": [
204
+ 0.023856325075030327,
205
+ 0.036417290568351746,
206
+ 0.13201750814914703
207
+ ],
208
+ "q01": [
209
+ -0.06935205049812794,
210
+ -0.10119648121297359,
211
+ -0.5124140107631683
212
+ ],
213
+ "q99": [
214
+ 0.07566657893359657,
215
+ 0.0950752408057451,
216
+ 0.5196981436014172
217
+ ]
218
+ },
219
+ "imu_la": {
220
+ "max": [
221
+ 6.173417568206787,
222
+ 13.512828826904297,
223
+ -2.0125882625579834
224
+ ],
225
+ "min": [
226
+ -21.677589416503906,
227
+ -15.804965019226074,
228
+ -24.276391983032227
229
+ ],
230
+ "mean": [
231
+ -0.36713558435440063,
232
+ -0.05341290682554245,
233
+ -9.84744644165039
234
+ ],
235
+ "std": [
236
+ 0.4190313220024109,
237
+ 0.10585533827543259,
238
+ 0.12830886244773865
239
+ ],
240
+ "q01": [
241
+ -1.0192994201183319,
242
+ -0.29011678844690325,
243
+ -10.266274557113647
244
+ ],
245
+ "q99": [
246
+ 0.7375994580984114,
247
+ 0.230275808572769,
248
+ -9.325943641662597
249
+ ]
250
+ },
251
+ "pressure": {
252
+ "max": [
253
+ 16.52094268798828
254
+ ],
255
+ "min": [
256
+ -0.0010726836044341326
257
+ ],
258
+ "mean": [
259
+ 7.065362453460693
260
+ ],
261
+ "std": [
262
+ 5.2881975173950195
263
+ ],
264
+ "q01": [
265
+ 0.3104786479473114
266
+ ],
267
+ "q99": [
268
+ 16.281835975646974
269
+ ]
270
+ },
271
+ "dvl_h": {
272
+ "max": [
273
+ 9.428301811218262
274
+ ],
275
+ "min": [
276
+ -1.0
277
+ ],
278
+ "mean": [
279
+ 1.0402350425720215
280
+ ],
281
+ "std": [
282
+ 1.6711875200271606
283
+ ],
284
+ "q01": [
285
+ 0.214564248919487
286
+ ],
287
+ "q99": [
288
+ 9.156972885131836
289
+ ]
290
+ }
291
+ },
292
+ "target_pos": {
293
+ "target_pos": {
294
+ "max": [
295
+ 15.79721450805664,
296
+ 7.267358303070068,
297
+ 1.5049903392791748,
298
+ 0.5932833552360535,
299
+ 1.205113410949707,
300
+ 2.887096643447876
301
+ ],
302
+ "min": [
303
+ -4.257084369659424,
304
+ -9.010113716125488,
305
+ -1.6286450624465942,
306
+ -1.4041659832000732,
307
+ -0.5855003595352173,
308
+ -3.1329503059387207
309
+ ],
310
+ "mean": [
311
+ 0.41113775968551636,
312
+ -0.131526380777359,
313
+ 0.01432724203914404,
314
+ -0.004899939522147179,
315
+ 0.03629195690155029,
316
+ 0.030115993693470955
317
+ ],
318
+ "std": [
319
+ 1.3917150497436523,
320
+ 0.884169340133667,
321
+ 0.12268536537885666,
322
+ 0.016746316105127335,
323
+ 0.04107803851366043,
324
+ 0.30970168113708496
325
+ ],
326
+ "q01": [
327
+ -1.9432353365421295,
328
+ -4.379468083381653,
329
+ -0.31190046668052673,
330
+ -0.05373585715889931,
331
+ -0.07010944351553917,
332
+ -1.0171301746368409
333
+ ],
334
+ "q99": [
335
+ 7.017193670272826,
336
+ 1.6167884862422934,
337
+ 0.461672668755054,
338
+ 0.04799993857741348,
339
+ 0.10270567953586578,
340
+ 1.2664259803295128
341
+ ]
342
+ }
343
+ },
344
+ "action": {
345
+ "joint_pos": {
346
+ "max": [
347
+ 0.01499609462916851,
348
+ 0.4975312352180481,
349
+ 0.49741339683532715,
350
+ 0.5057408809661865,
351
+ 0.00013245544687379152
352
+ ],
353
+ "min": [
354
+ -0.001301140058785677,
355
+ -4.4340096792439e-05,
356
+ -1.8281958773513907e-06,
357
+ -8.687281479069497e-06,
358
+ -6.792588101234287e-05
359
+ ],
360
+ "mean": [
361
+ 0.003374873660504818,
362
+ 0.3498944938182831,
363
+ 0.3498944938182831,
364
+ 0.35298940539360046,
365
+ 3.5255197872174904e-05
366
+ ],
367
+ "std": [
368
+ 0.006213217508047819,
369
+ 0.22534441947937012,
370
+ 0.22534498572349548,
371
+ 0.22973771393299103,
372
+ 3.3011066989274696e-05
373
+ ],
374
+ "q01": [
375
+ 0.0,
376
+ 0.0,
377
+ 0.0,
378
+ 0.0,
379
+ 0.0
380
+ ],
381
+ "q99": [
382
+ 0.014937113039195538,
383
+ 0.49659860134124756,
384
+ 0.49659767746925354,
385
+ 0.5056976079940796,
386
+ 9.855335018073673e-05
387
+ ]
388
+ },
389
+ "pwm": {
390
+ "max": [
391
+ 1.0,
392
+ 1.0,
393
+ 1.0,
394
+ 1.0,
395
+ 1.0,
396
+ 1.0,
397
+ 1.0,
398
+ 1.0
399
+ ],
400
+ "min": [
401
+ -1.0,
402
+ -1.0,
403
+ -1.0,
404
+ -1.0,
405
+ -1.0,
406
+ -1.0,
407
+ -1.0,
408
+ -1.0
409
+ ],
410
+ "mean": [
411
+ -0.1696692854166031,
412
+ -0.1136094480752945,
413
+ 0.10162586718797684,
414
+ 0.17456766963005066,
415
+ -0.07031124830245972,
416
+ -0.034409455955028534,
417
+ 0.046888984739780426,
418
+ 0.08268629014492035
419
+ ],
420
+ "std": [
421
+ 0.4997708201408386,
422
+ 0.5126947164535522,
423
+ 0.5230556726455688,
424
+ 0.5092427134513855,
425
+ 0.25611206889152527,
426
+ 0.2508251667022705,
427
+ 0.24924014508724213,
428
+ 0.2535521388053894
429
+ ],
430
+ "q01": [
431
+ -1.0,
432
+ -1.0,
433
+ -1.0,
434
+ -1.0,
435
+ -1.0,
436
+ -1.0,
437
+ -1.0,
438
+ -1.0
439
+ ],
440
+ "q99": [
441
+ 1.0,
442
+ 1.0,
443
+ 1.0,
444
+ 1.0,
445
+ 1.0,
446
+ 1.0,
447
+ 1.0,
448
+ 1.0
449
+ ]
450
+ }
451
+ }
452
+ },
453
+ "modalities": {
454
+ "video": {
455
+ "ego": {
456
+ "resolution": [
457
+ 320,
458
+ 240
459
+ ],
460
+ "channels": 3,
461
+ "fps": 10.0
462
+ },
463
+ "wrist": {
464
+ "resolution": [
465
+ 320,
466
+ 240
467
+ ],
468
+ "channels": 3,
469
+ "fps": 10.0
470
+ }
471
+ },
472
+ "state": {
473
+ "joint_pos": {
474
+ "absolute": true,
475
+ "rotation_type": null,
476
+ "shape": [
477
+ 5
478
+ ],
479
+ "continuous": true
480
+ },
481
+ "pwm": {
482
+ "absolute": true,
483
+ "rotation_type": null,
484
+ "shape": [
485
+ 8
486
+ ],
487
+ "continuous": true
488
+ },
489
+ "joint_v": {
490
+ "absolute": true,
491
+ "rotation_type": null,
492
+ "shape": [
493
+ 5
494
+ ],
495
+ "continuous": true
496
+ },
497
+ "dvl_v": {
498
+ "absolute": true,
499
+ "rotation_type": null,
500
+ "shape": [
501
+ 3
502
+ ],
503
+ "continuous": true
504
+ },
505
+ "imu_av": {
506
+ "absolute": true,
507
+ "rotation_type": null,
508
+ "shape": [
509
+ 3
510
+ ],
511
+ "continuous": true
512
+ },
513
+ "imu_la": {
514
+ "absolute": true,
515
+ "rotation_type": null,
516
+ "shape": [
517
+ 3
518
+ ],
519
+ "continuous": true
520
+ },
521
+ "pressure": {
522
+ "absolute": true,
523
+ "rotation_type": null,
524
+ "shape": [
525
+ 1
526
+ ],
527
+ "continuous": true
528
+ },
529
+ "dvl_h": {
530
+ "absolute": true,
531
+ "rotation_type": null,
532
+ "shape": [
533
+ 1
534
+ ],
535
+ "continuous": true
536
+ }
537
+ },
538
+ "target_pos": {
539
+ "target_pos": {
540
+ "absolute": true,
541
+ "rotation_type": null,
542
+ "shape": [
543
+ 6
544
+ ],
545
+ "continuous": true
546
+ }
547
+ },
548
+ "action": {
549
+ "joint_pos": {
550
+ "absolute": true,
551
+ "rotation_type": null,
552
+ "shape": [
553
+ 5
554
+ ],
555
+ "continuous": true
556
+ },
557
+ "pwm": {
558
+ "absolute": true,
559
+ "rotation_type": null,
560
+ "shape": [
561
+ 8
562
+ ],
563
+ "continuous": true
564
+ }
565
+ }
566
+ },
567
+ "embodiment_tag": "new_embodiment"
568
+ }
569
+ }
model-00001-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8a195f325da959a28cec91124ead5838d66b0dcb75519947edae0352da53864f
3
+ size 4999367032
model-00002-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fd96fa5088d7d15199befce71f2a5e4393af66336707b60c1d4b5e598802ccd0
3
+ size 2619382460
model.safetensors.index.json ADDED
The diff for this file is too large to render. See raw diff
 
trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9d9122aaa1a555b9b7ae2d4b7277762634590c9cc156f4b79e64f2aefd7263da
3
+ size 5368