Hugging Face's logo

vpraise00
/
gr00t-n1-5-robocasa-panda-omron-42-3-epochs

Safetensors
gr00t_n1_5
Model card Files
xet
 Community
vpraise00 commited on
Commit
a7b8685
·
verified ·
1 Parent(s): 91ae436

Upload folder using huggingface_hub

Browse files
Files changed (5) hide show
  1. config.json +63 -0
  2. experiment_cfg/metadata.json +472 -0
  3. model-00001-of-00002.safetensors +3 -0
  4. model-00002-of-00002.safetensors +3 -0
  5. model.safetensors.index.json +0 -0
config.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
+ "hidden_size": 2048,
59
+ "model_dtype": "float32",
60
+ "model_type": "gr00t_n1_5",
61
+ "torch_dtype": "bfloat16",
62
+ "transformers_version": "4.51.3"
63
+ }
experiment_cfg/metadata.json ADDED
@@ -0,0 +1,472 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "new_embodiment": {
3
+ "statistics": {
4
+ "state": {
5
+ "base_rotation": {
6
+ "mean": [
7
+ -0.06653653085231781,
8
+ -0.4676792621612549,
9
+ 0.0,
10
+ 0.4676792621612549,
11
+ -0.06653653085231781,
12
+ 0.0
13
+ ],
14
+ "std": [
15
+ 0.7034456729888916,
16
+ 0.531082034111023,
17
+ 0.0,
18
+ 0.531082034111023,
19
+ 0.7034456729888916,
20
+ 0.0
21
+ ],
22
+ "min": [
23
+ -1.0,
24
+ -1.0,
25
+ 0.0,
26
+ -1.0,
27
+ -1.0,
28
+ 0.0
29
+ ],
30
+ "max": [
31
+ 1.0,
32
+ 1.0,
33
+ 0.0,
34
+ 1.0,
35
+ 1.0,
36
+ 0.0
37
+ ],
38
+ "q01": [
39
+ -1.0,
40
+ -1.0,
41
+ 0.0,
42
+ -0.9999998211860657,
43
+ -1.0,
44
+ 0.0
45
+ ],
46
+ "q99": [
47
+ 1.0,
48
+ 0.9999998211860657,
49
+ 0.0,
50
+ 1.0,
51
+ 1.0,
52
+ 0.0
53
+ ]
54
+ },
55
+ "end_effector_position_relative": {
56
+ "mean": [
57
+ 0.20603015267093988,
58
+ 0.014220384843297575,
59
+ 0.2604373537940307
60
+ ],
61
+ "std": [
62
+ 0.1284903092993028,
63
+ 0.4720566847675113,
64
+ 0.19718323757553752
65
+ ],
66
+ "min": [
67
+ -0.09696616279772341,
68
+ -0.7616322860963773,
69
+ 0.05436132796787274
70
+ ],
71
+ "max": [
72
+ 0.43705324458370143,
73
+ 0.706012167187176,
74
+ 0.646213099352813
75
+ ],
76
+ "q01": [
77
+ -0.034503350749626314,
78
+ -0.7546979455229289,
79
+ 0.05661801989517107
80
+ ],
81
+ "q99": [
82
+ 0.41847876176695104,
83
+ 0.6766428356337131,
84
+ 0.6081645168943913
85
+ ]
86
+ },
87
+ "gripper_qpos": {
88
+ "mean": [
89
+ -0.00011370025778741031
90
+ ],
91
+ "std": [
92
+ 0.005233253962022829
93
+ ],
94
+ "min": [
95
+ -0.018295429399716005
96
+ ],
97
+ "max": [
98
+ 0.01932574675248422
99
+ ],
100
+ "q01": [
101
+ -0.01753297714198864
102
+ ],
103
+ "q99": [
104
+ 0.0167837186795225
105
+ ]
106
+ },
107
+ "end_effector_rotation_relative": {
108
+ "mean": [
109
+ 0.04319709166884422,
110
+ 0.02683359384536743,
111
+ 0.8297028541564941,
112
+ 0.06467945128679276,
113
+ -0.08257200568914413,
114
+ -0.036586906760931015
115
+ ],
116
+ "std": [
117
+ 0.42015716433525085,
118
+ 0.28527262806892395,
119
+ 0.22605140507221222,
120
+ 0.872289776802063,
121
+ 0.44532403349876404,
122
+ 0.16868974268436432
123
+ ],
124
+ "min": [
125
+ -0.566469132900238,
126
+ -0.7275094985961914,
127
+ -0.17805476486682892,
128
+ -0.9999650120735168,
129
+ -0.9996269941329956,
130
+ -0.5314959287643433
131
+ ],
132
+ "max": [
133
+ 0.9885779619216919,
134
+ 0.7569407224655151,
135
+ 0.9994974732398987,
136
+ 0.9999901652336121,
137
+ 0.6065186858177185,
138
+ 0.4393925964832306
139
+ ],
140
+ "q01": [
141
+ -0.4356670355796814,
142
+ -0.6310072255134582,
143
+ 0.2529286158084869,
144
+ -0.994257652759552,
145
+ -0.9981180834770202,
146
+ -0.4025233471393585
147
+ ],
148
+ "q99": [
149
+ 0.9623777294158936,
150
+ 0.6604138207435609,
151
+ 0.9969434356689454,
152
+ 0.9993602705001832,
153
+ 0.5087940287590028,
154
+ 0.3110717904567722
155
+ ]
156
+ },
157
+ "base_position": {
158
+ "mean": [
159
+ 2.583604274664355,
160
+ -1.8027354364819332,
161
+ 0.7003252131246801
162
+ ],
163
+ "std": [
164
+ 1.3185486330993104,
165
+ 1.1511084233399025,
166
+ 0.0005523202075831765
167
+ ],
168
+ "min": [
169
+ 1.0299822884632546,
170
+ -5.032551968860045,
171
+ 0.6998954064939146
172
+ ],
173
+ "max": [
174
+ 5.3700050031598945,
175
+ -0.6188440541934274,
176
+ 0.7027937330908252
177
+ ],
178
+ "q01": [
179
+ 1.0299851274445795,
180
+ -5.0232118918670325,
181
+ 0.6999966389030905
182
+ ],
183
+ "q99": [
184
+ 5.369995352229393,
185
+ -0.6199118594186583,
186
+ 0.7023674721625425
187
+ ]
188
+ }
189
+ },
190
+ "action": {
191
+ "end_effector_rotation": {
192
+ "mean": [
193
+ -0.02644892781972885,
194
+ 0.018950117751955986,
195
+ 0.8435676693916321,
196
+ 0.05514839291572571,
197
+ 0.06446361541748047,
198
+ -0.03949278965592384
199
+ ],
200
+ "std": [
201
+ 0.28683900833129883,
202
+ 0.40036240220069885,
203
+ 0.21158605813980103,
204
+ 0.4274965822696686,
205
+ 0.882790744304657,
206
+ 0.17078961431980133
207
+ ],
208
+ "min": [
209
+ -0.761704683303833,
210
+ -0.5940344333648682,
211
+ -0.16592112183570862,
212
+ -0.6245508790016174,
213
+ -0.999976634979248,
214
+ -0.5365860462188721
215
+ ],
216
+ "max": [
217
+ 0.7317051887512207,
218
+ 0.9883328676223755,
219
+ 0.999997079372406,
220
+ 0.9996489882469177,
221
+ 0.9999935626983643,
222
+ 0.4393664002418518
223
+ ],
224
+ "q01": [
225
+ -0.6677525329589844,
226
+ -0.4374687194824219,
227
+ 0.2581059348583221,
228
+ -0.5093881845474243,
229
+ -0.994687294960022,
230
+ -0.40408878326416015
231
+ ],
232
+ "q99": [
233
+ 0.6344457244873047,
234
+ 0.9598882341384888,
235
+ 0.9971694612503053,
236
+ 0.9978565716743469,
237
+ 0.9994449663162231,
238
+ 0.3095978355407715
239
+ ]
240
+ },
241
+ "gripper_close": {
242
+ "mean": [
243
+ 0.3213493227958679
244
+ ],
245
+ "std": [
246
+ 0.4669945538043976
247
+ ],
248
+ "min": [
249
+ 0.0
250
+ ],
251
+ "max": [
252
+ 1.0
253
+ ],
254
+ "q01": [
255
+ 0.0
256
+ ],
257
+ "q99": [
258
+ 1.0
259
+ ]
260
+ },
261
+ "control_mode": {
262
+ "mean": [
263
+ 0.0
264
+ ],
265
+ "std": [
266
+ 0.0
267
+ ],
268
+ "min": [
269
+ 0.0
270
+ ],
271
+ "max": [
272
+ 0.0
273
+ ],
274
+ "q01": [
275
+ 0.0
276
+ ],
277
+ "q99": [
278
+ 0.0
279
+ ]
280
+ },
281
+ "end_effector_position": {
282
+ "mean": [
283
+ 0.20091065764427185,
284
+ 0.015957994386553764,
285
+ 0.2486896812915802
286
+ ],
287
+ "std": [
288
+ 0.13399089872837067,
289
+ 0.480807900428772,
290
+ 0.19573277235031128
291
+ ],
292
+ "min": [
293
+ -0.0957145020365715,
294
+ -0.7609035968780518,
295
+ 0.025804711505770683
296
+ ],
297
+ "max": [
298
+ 0.45271947979927063,
299
+ 0.7161008715629578,
300
+ 0.6451568007469177
301
+ ],
302
+ "q01": [
303
+ -0.04026926383376121,
304
+ -0.7499863719940185,
305
+ 0.04305603563785553
306
+ ],
307
+ "q99": [
308
+ 0.4356601572036743,
309
+ 0.6989241194725038,
310
+ 0.6071043157577515
311
+ ]
312
+ },
313
+ "base_motion": {
314
+ "mean": [
315
+ 0.0,
316
+ 0.0,
317
+ 0.0,
318
+ 0.0
319
+ ],
320
+ "std": [
321
+ 0.0,
322
+ 0.0,
323
+ 0.0,
324
+ 0.0
325
+ ],
326
+ "min": [
327
+ 0.0,
328
+ 0.0,
329
+ 0.0,
330
+ 0.0
331
+ ],
332
+ "max": [
333
+ 0.0,
334
+ 0.0,
335
+ 0.0,
336
+ 0.0
337
+ ],
338
+ "q01": [
339
+ 0.0,
340
+ 0.0,
341
+ 0.0,
342
+ 0.0
343
+ ],
344
+ "q99": [
345
+ 0.0,
346
+ 0.0,
347
+ 0.0,
348
+ 0.0
349
+ ]
350
+ }
351
+ }
352
+ },
353
+ "modalities": {
354
+ "video": {
355
+ "left_view": {
356
+ "resolution": [
357
+ 128,
358
+ 128
359
+ ],
360
+ "channels": 3,
361
+ "fps": 12.0
362
+ },
363
+ "right_view": {
364
+ "resolution": [
365
+ 128,
366
+ 128
367
+ ],
368
+ "channels": 3,
369
+ "fps": 12.0
370
+ },
371
+ "wrist_view": {
372
+ "resolution": [
373
+ 128,
374
+ 128
375
+ ],
376
+ "channels": 3,
377
+ "fps": 12.0
378
+ }
379
+ },
380
+ "state": {
381
+ "end_effector_position_relative": {
382
+ "absolute": true,
383
+ "rotation_type": null,
384
+ "shape": [
385
+ 3
386
+ ],
387
+ "continuous": true
388
+ },
389
+ "end_effector_rotation_relative": {
390
+ "absolute": true,
391
+ "rotation_type": "quaternion",
392
+ "shape": [
393
+ 4
394
+ ],
395
+ "continuous": true
396
+ },
397
+ "gripper_qpos": {
398
+ "absolute": true,
399
+ "rotation_type": null,
400
+ "shape": [
401
+ 1
402
+ ],
403
+ "continuous": true
404
+ },
405
+ "base_position": {
406
+ "absolute": true,
407
+ "rotation_type": null,
408
+ "shape": [
409
+ 3
410
+ ],
411
+ "continuous": true
412
+ },
413
+ "base_rotation": {
414
+ "absolute": true,
415
+ "rotation_type": "quaternion",
416
+ "shape": [
417
+ 4
418
+ ],
419
+ "continuous": true
420
+ }
421
+ },
422
+ "action": {
423
+ "end_effector_position": {
424
+ "absolute": true,
425
+ "rotation_type": null,
426
+ "shape": [
427
+ 3
428
+ ],
429
+ "continuous": true
430
+ },
431
+ "end_effector_rotation": {
432
+ "absolute": true,
433
+ "rotation_type": "rotation_6d",
434
+ "shape": [
435
+ 6
436
+ ],
437
+ "continuous": true
438
+ },
439
+ "gripper_close": {
440
+ "absolute": true,
441
+ "rotation_type": null,
442
+ "shape": [
443
+ 1
444
+ ],
445
+ "continuous": false
446
+ },
447
+ "base_motion": {
448
+ "absolute": false,
449
+ "rotation_type": null,
450
+ "shape": [
451
+ 4
452
+ ],
453
+ "continuous": true
454
+ },
455
+ "control_mode": {
456
+ "absolute": true,
457
+ "rotation_type": null,
458
+ "shape": [
459
+ 1
460
+ ],
461
+ "continuous": false
462
+ }
463
+ },
464
+ "annotation": {
465
+ "human": [
466
+ "action.task_description"
467
+ ]
468
+ }
469
+ },
470
+ "embodiment_tag": "new_embodiment"
471
+ }
472
+ }
model-00001-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b9cd78a30fd9ae853465bdb6d10e07dacef8949f870913ff76acb4479181fd73
3
+ size 4973058384
model-00002-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eabc3743bb6b419e67753fe42859621e26d2c6ba46b8fe33a7916d3c0a37df15
3
+ size 475401840
model.safetensors.index.json ADDED
The diff for this file is too large to render. See raw diff