Hugging Face's logo

Datasets:
DiscreteRTC
/
dRTC

Modalities:
Tabular
Text
Video
Formats:
json
Size:
< 1K
Libraries:
Datasets
pandas
Polars
Dataset card Data Studio Files
xet
 Community
1
outsider86 commited on
Commit
6712793
·
verified ·
1 Parent(s): 761af3f

Upload piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826

Browse files

Validated 2,000-step D2F checkpoint initialized from the corresponding 25,000-step Piper policy; D2F block size 7.

Files changed (23) hide show
  1. models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/action_tokenizer.json +529 -0
  2. models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/added_tokens.json +4 -0
  3. models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/config.json +3174 -0
  4. models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/configuration_prismatic.py +171 -0
  5. models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/dataset_statistics.json +136 -0
  6. models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/generation_config.json +7 -0
  7. models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/lora_adapter/README.md +204 -0
  8. models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/lora_adapter/adapter_config.json +45 -0
  9. models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/lora_adapter/adapter_model.safetensors +3 -0
  10. models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/model-00001-of-00004.safetensors +3 -0
  11. models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/model-00002-of-00004.safetensors +3 -0
  12. models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/model-00003-of-00004.safetensors +3 -0
  13. models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/model-00004-of-00004.safetensors +3 -0
  14. models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/model.safetensors.index.json +989 -0
  15. models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/modeling_prismatic.py +2117 -0
  16. models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/preprocessor_config.json +114 -0
  17. models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/processing_prismatic.py +257 -0
  18. models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/processor_config.json +6 -0
  19. models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/proprio_projector--2000_checkpoint.pt +3 -0
  20. models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/special_tokens_map.json +37 -0
  21. models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/tokenizer.json +0 -0
  22. models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/tokenizer.model +3 -0
  23. models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/tokenizer_config.json +62 -0
models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/action_tokenizer.json ADDED
@@ -0,0 +1,529 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "calibration": {
3
+ "high": [
4
+ [
5
+ 0.9999175258779393,
6
+ 0.9999975969934155,
7
+ 1.0000456936949362,
8
+ 1.0000176358889883,
9
+ 0.9999471454922568,
10
+ 1.0000671564192478,
11
+ 1.0023373510392193
12
+ ],
13
+ [
14
+ 1.0015870600430838,
15
+ 1.0050950037507558,
16
+ 1.004341553530057,
17
+ 1.0066667560159792,
18
+ 1.0020589112235603,
19
+ 1.0133042478167698,
20
+ 1.0932137307886787
21
+ ],
22
+ [
23
+ 1.0041069246180567,
24
+ 1.0078954579660404,
25
+ 1.0070683454318239,
26
+ 1.013288299420544,
27
+ 1.0042754253410109,
28
+ 1.0156774723187338,
29
+ 1.1795134964473495
30
+ ],
31
+ [
32
+ 1.003752437219364,
33
+ 1.005661169248319,
34
+ 1.0050431092714869,
35
+ 1.0097651236117788,
36
+ 1.007894883333231,
37
+ 1.0111852415507732,
38
+ 1.1312572961731033
39
+ ],
40
+ [
41
+ 1.0040959633844704,
42
+ 1.0051950629395436,
43
+ 1.0045389525516286,
44
+ 1.0088239283255966,
45
+ 1.0066390029120504,
46
+ 1.010055094486499,
47
+ 1.1182011608482758
48
+ ],
49
+ [
50
+ 1.0042366214086826,
51
+ 1.0050445974444833,
52
+ 1.0043799259093766,
53
+ 1.0085416302440977,
54
+ 1.0060208904627699,
55
+ 1.0097217615363545,
56
+ 1.114549002399775
57
+ ],
58
+ [
59
+ 1.0041324959987792,
60
+ 1.0050116546386032,
61
+ 1.004344833437495,
62
+ 1.0084719442810277,
63
+ 1.00605344148993,
64
+ 1.009637888835212,
65
+ 1.1135364271958503
66
+ ],
67
+ [
68
+ 1.00429686315201,
69
+ 1.004997510462352,
70
+ 1.0043301023222122,
71
+ 1.008447521098842,
72
+ 1.006186979140605,
73
+ 1.0096094934274087,
74
+ 1.1132478124395255
75
+ ],
76
+ [
77
+ 1.0044044013043483,
78
+ 1.004996119896485,
79
+ 1.0043285611412716,
80
+ 1.0084432475963956,
81
+ 1.0063133428037685,
82
+ 1.0096041153544655,
83
+ 1.1131694976565119
84
+ ],
85
+ [
86
+ 1.0042275048044278,
87
+ 1.0049943373488617,
88
+ 1.0043267560165763,
89
+ 1.0084406842516376,
90
+ 1.0062319875697991,
91
+ 1.009601251848344,
92
+ 1.1131461806089544
93
+ ],
94
+ [
95
+ 1.0041370606062927,
96
+ 1.0049946104105574,
97
+ 1.0043270245681708,
98
+ 1.0084406798743608,
99
+ 1.0062341076562185,
100
+ 1.0096011893581969,
101
+ 1.1131672756350075
102
+ ],
103
+ [
104
+ 1.0041107512478207,
105
+ 1.0049942700124734,
106
+ 1.0043266526622217,
107
+ 1.008440401134664,
108
+ 1.0062489647614532,
109
+ 1.0096008157902556,
110
+ 1.1131797168936166
111
+ ],
112
+ [
113
+ 1.0041171315645734,
114
+ 1.0049942228446316,
115
+ 1.00432682976202,
116
+ 1.0084405363664501,
117
+ 1.0062606710870579,
118
+ 1.009600928966052,
119
+ 1.1130400272664245
120
+ ],
121
+ [
122
+ 1.0041078177917337,
123
+ 1.0049936099492087,
124
+ 1.004326676101867,
125
+ 1.0084400944852474,
126
+ 1.006265137669144,
127
+ 1.009600688454494,
128
+ 1.1129031664720426
129
+ ],
130
+ [
131
+ 1.0041090049480335,
132
+ 1.0049941585848077,
133
+ 1.0043274118321766,
134
+ 1.0084403414233425,
135
+ 1.006273015300594,
136
+ 1.0096014887862987,
137
+ 1.1129056545498244
138
+ ],
139
+ [
140
+ 1.0041054101502793,
141
+ 1.0049932916518995,
142
+ 1.0043269967058233,
143
+ 1.0084380899359937,
144
+ 1.006273786538894,
145
+ 1.0096000067595905,
146
+ 1.1128801591388116
147
+ ],
148
+ [
149
+ 1.0041048020698327,
150
+ 1.0049928872933653,
151
+ 1.002510807195772,
152
+ 1.008439384291659,
153
+ 1.0062733695194825,
154
+ 1.009601208989435,
155
+ 1.1128809282837075
156
+ ],
157
+ [
158
+ 1.0041043827758782,
159
+ 1.0049883718165546,
160
+ 1.0025082834482046,
161
+ 1.0084435056089955,
162
+ 1.0062732044430793,
163
+ 1.0095985905553986,
164
+ 1.1128810713723771
165
+ ],
166
+ [
167
+ 1.004104452216621,
168
+ 1.0049920880486323,
169
+ 1.0025072380565292,
170
+ 1.0084509464793756,
171
+ 1.006273361743068,
172
+ 1.0096057864826167,
173
+ 1.1128801100717478
174
+ ],
175
+ [
176
+ 1.0041041867154727,
177
+ 1.0049956021370245,
178
+ 0.9963726575473202,
179
+ 1.0084621156140878,
180
+ 1.0062732602498925,
181
+ 1.009597541628172,
182
+ 1.1128779611123827
183
+ ],
184
+ [
185
+ 1.0041043574608106,
186
+ 1.0050204240548086,
187
+ 0.99637904329347,
188
+ 1.0084827800916911,
189
+ 1.0062735803633895,
190
+ 1.0096197807625367,
191
+ 1.1128702550898761
192
+ ],
193
+ [
194
+ 1.0041041089373117,
195
+ 1.0049358894628475,
196
+ 0.9963649898436493,
197
+ 1.0084609907491957,
198
+ 1.0062733846131053,
199
+ 1.0095902876988276,
200
+ 1.1128431926887496
201
+ ],
202
+ [
203
+ 1.0041056945875415,
204
+ 1.0050497713228366,
205
+ 0.9963696143952521,
206
+ 1.0084266061876666,
207
+ 1.0062752923853282,
208
+ 1.0096487560630056,
209
+ 1.112746103150327
210
+ ],
211
+ [
212
+ 1.004106629055015,
213
+ 1.0051040068090435,
214
+ 0.9963763259777092,
215
+ 1.0083376361656573,
216
+ 1.0062772612365871,
217
+ 1.0095129129212166,
218
+ 1.1124012587169962
219
+ ],
220
+ [
221
+ 1.0041191871494846,
222
+ 1.0050175811224589,
223
+ 0.9877997804189204,
224
+ 1.0091288546926052,
225
+ 1.0062934906795018,
226
+ 1.0097414371251408,
227
+ 1.1111761712979569
228
+ ],
229
+ [
230
+ 1.0041465016527362,
231
+ 1.0052661759180614,
232
+ 0.9631834763350109,
233
+ 1.0087452675192847,
234
+ 1.006333876642845,
235
+ 1.0095903023477828,
236
+ 1.1069682071848803
237
+ ],
238
+ [
239
+ 1.0042771152722416,
240
+ 1.005791319055785,
241
+ 0.9358405591851634,
242
+ 1.008015251807582,
243
+ 1.0065104667981282,
244
+ 1.010237932635042,
245
+ 1.104598496663264
246
+ ],
247
+ [
248
+ 1.0046696923490177,
249
+ 1.0049368023378247,
250
+ 0.9134732180530025,
251
+ 1.0057374703187174,
252
+ 1.0070564514516132,
253
+ 1.0086326082620123,
254
+ 1.0966097437719098
255
+ ]
256
+ ],
257
+ "low": [
258
+ [
259
+ -0.9999811855162797,
260
+ -0.9999862871843914,
261
+ -1.0000078764641311,
262
+ -1.0004455878529805,
263
+ -1.000031971209145,
264
+ -1.0000053464461183,
265
+ -1.0002408396966471
266
+ ],
267
+ [
268
+ -1.0058387130296564,
269
+ -1.0036266001677787,
270
+ -1.0082559498220227,
271
+ -1.0111699345566896,
272
+ -1.0093402878572566,
273
+ -1.008083186055094,
274
+ -1.0079123131022203
275
+ ],
276
+ [
277
+ -1.0084335660704882,
278
+ -1.0069673016836382,
279
+ -1.0094153705809223,
280
+ -1.0078936546892403,
281
+ -1.0122813625465914,
282
+ -1.010742356683391,
283
+ -1.0323279632221525
284
+ ],
285
+ [
286
+ -1.0066617382621144,
287
+ -1.0050195393335914,
288
+ -1.0067635241796644,
289
+ -1.0051422357330106,
290
+ -1.0098813952062105,
291
+ -1.0085326901646088,
292
+ -1.0246713849707791
293
+ ],
294
+ [
295
+ -1.0061389762482045,
296
+ -1.0045680063316118,
297
+ -1.0061733993789557,
298
+ -1.004511145941237,
299
+ -1.0091275663750578,
300
+ -1.0078635180142725,
301
+ -1.0224536450956934
302
+ ],
303
+ [
304
+ -1.005969481679961,
305
+ -1.0043805257437657,
306
+ -1.0058791642945861,
307
+ -1.0043210759654877,
308
+ -1.008890205100705,
309
+ -1.0076516685475931,
310
+ -1.021807243412447
311
+ ],
312
+ [
313
+ -1.0059311198421115,
314
+ -1.0043359758801278,
315
+ -1.005830569166006,
316
+ -1.0042775678458105,
317
+ -1.0088326302495192,
318
+ -1.0076013551264245,
319
+ -1.021635103621236
320
+ ],
321
+ [
322
+ -1.0059154590079056,
323
+ -1.0043185118115818,
324
+ -1.0058120588891915,
325
+ -1.004260503610799,
326
+ -1.0088115686767447,
327
+ -1.0075823415261194,
328
+ -1.0215819364103984
329
+ ],
330
+ [
331
+ -1.0059136497568537,
332
+ -1.0043161939515186,
333
+ -1.0058094534510167,
334
+ -1.0042583022617557,
335
+ -1.008808239225507,
336
+ -1.0075795883530327,
337
+ -1.0215696117603734
338
+ ],
339
+ [
340
+ -1.0059117699315316,
341
+ -1.0043141678545473,
342
+ -1.0058073353250603,
343
+ -1.004256320415214,
344
+ -1.0088059340394275,
345
+ -1.0075774433169509,
346
+ -1.0215647920467903
347
+ ],
348
+ [
349
+ -1.005911982926572,
350
+ -1.0043143297648727,
351
+ -1.005807475942311,
352
+ -1.0042564878020528,
353
+ -1.0088060155693628,
354
+ -1.0075775849700324,
355
+ -1.0215641630672656
356
+ ],
357
+ [
358
+ -1.005911567659417,
359
+ -1.00431395207037,
360
+ -1.0058070969541522,
361
+ -1.0042561471308054,
362
+ -1.0088056678717603,
363
+ -1.0075772124794495,
364
+ -1.0215636038201217
365
+ ],
366
+ [
367
+ -1.005911741310036,
368
+ -1.0043140922108338,
369
+ -1.0058072104053042,
370
+ -1.0042561782223451,
371
+ -1.008805611343825,
372
+ -1.0075772651973467,
373
+ -1.02156364590468
374
+ ],
375
+ [
376
+ -1.005911538957882,
377
+ -1.0043140064537592,
378
+ -1.0058071143189835,
379
+ -1.0042559933265598,
380
+ -1.0088054422545623,
381
+ -1.0075772743064644,
382
+ -1.0215635414178392
383
+ ],
384
+ [
385
+ -1.0059115652792843,
386
+ -1.0043141053792968,
387
+ -1.0058072615806173,
388
+ -1.0042558346729467,
389
+ -1.0088057252460458,
390
+ -1.0075776732656991,
391
+ -1.0215635289953084
392
+ ],
393
+ [
394
+ -1.0059112517594193,
395
+ -1.0043139494568705,
396
+ -1.0058071748061754,
397
+ -1.0042553287546072,
398
+ -1.0088051106568527,
399
+ -1.0075774224505534,
400
+ -1.0215633674017937
401
+ ],
402
+ [
403
+ -1.0059120844778178,
404
+ -1.0043149850395123,
405
+ -1.0058090910961521,
406
+ -1.0042550290477508,
407
+ -1.008803352281525,
408
+ -1.0075774059397944,
409
+ -1.0215628542046657
410
+ ],
411
+ [
412
+ -1.0059113621999989,
413
+ -1.0043142708705022,
414
+ -1.005808717343193,
415
+ -1.0042570247021732,
416
+ -1.0087957066944002,
417
+ -1.0075780907011596,
418
+ -1.0215610269320894
419
+ ],
420
+ [
421
+ -1.0059146558708503,
422
+ -1.0043149181619413,
423
+ -1.0058135173549172,
424
+ -1.0042530715313638,
425
+ -1.0087894737101353,
426
+ -1.0075835611132455,
427
+ -1.0215544477807998
428
+ ],
429
+ [
430
+ -1.005901188184068,
431
+ -1.0043113323355206,
432
+ -1.0058168024448715,
433
+ -1.0042406694846069,
434
+ -1.008775922914337,
435
+ -1.0075887209073786,
436
+ -1.0215311279110872
437
+ ],
438
+ [
439
+ -1.0059228144248489,
440
+ -1.0043217208519455,
441
+ -1.0058207887154493,
442
+ -1.0042681693198365,
443
+ -1.0088232356461047,
444
+ -1.007604428570618,
445
+ -1.0214477100882668
446
+ ],
447
+ [
448
+ -1.0059264196761757,
449
+ -1.0042894427969948,
450
+ -1.00580247570373,
451
+ -1.004274459925316,
452
+ -1.0016992755209226,
453
+ -1.0076163497545265,
454
+ -1.02115173336722
455
+ ],
456
+ [
457
+ -1.005895380108422,
458
+ -1.0042554701469226,
459
+ -1.0058219302465177,
460
+ -1.0042802186092992,
461
+ -1.0016913945712975,
462
+ -1.0076216102870905,
463
+ -1.0201169736061195
464
+ ],
465
+ [
466
+ -1.0056954115322192,
467
+ -1.0041900669781711,
468
+ -1.005661609484912,
469
+ -1.003806906189326,
470
+ -1.0016512242108382,
471
+ -1.0075416407526645,
472
+ -1.017262158531916
473
+ ],
474
+ [
475
+ -1.0057561770496,
476
+ -1.004177032398746,
477
+ -1.0057978177176239,
478
+ -1.003813361975034,
479
+ -0.9928257066335623,
480
+ -1.0075032848151555,
481
+ -1.011728497620881
482
+ ],
483
+ [
484
+ -1.0050705762084355,
485
+ -1.0040745769530248,
486
+ -1.0054804611056636,
487
+ -1.0045328962891171,
488
+ -0.9835987395641486,
489
+ -1.0078318618669473,
490
+ -1.0060701401086727
491
+ ],
492
+ [
493
+ -1.0059348635129135,
494
+ -1.0023220045942345,
495
+ -1.0057892658345282,
496
+ -1.0079400915173973,
497
+ -0.9775804968725559,
498
+ -1.0078633417196297,
499
+ -1.0138784952647055
500
+ ],
501
+ [
502
+ -1.0060002090866806,
503
+ -1.0014407293749497,
504
+ -1.0061531377350572,
505
+ -1.0124293299462017,
506
+ -0.9755914836976699,
507
+ -1.0083191288805333,
508
+ -1.0048801332060089
509
+ ]
510
+ ]
511
+ },
512
+ "config": {
513
+ "action_dim": 7,
514
+ "chunk_size": 50,
515
+ "degree": 3,
516
+ "end_padding": true,
517
+ "frequency_hz": 50.0,
518
+ "implementation_version": "uniform_left_extended_double_fit_v1",
519
+ "mode": "uniform_left",
520
+ "num_basis": 28,
521
+ "regularization": 0.0001,
522
+ "span_length_steps": 2,
523
+ "vocab_size": 256
524
+ },
525
+ "library": "action_tokenization",
526
+ "policy_layout": "control_flat_v1",
527
+ "schema_version": 2,
528
+ "tokenizer_id": "3f7924fa97a514c5"
529
+ }
models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/added_tokens.json ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {
2
+ "<PAD>": 32000,
3
+ "<mask>": 32001
4
+ }
models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/config.json ADDED
@@ -0,0 +1,3174 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "/scratch/wangpc/dRTCv2/artifacts/models/openvla-7b",
3
+ "arch_specifier": "no-align+fused-gelu-mlp",
4
+ "architectures": [
5
+ "OpenVLAForActionPrediction"
6
+ ],
7
+ "auto_map": {
8
+ "AutoConfig": "configuration_prismatic.OpenVLAConfig",
9
+ "AutoModelForVision2Seq": "modeling_prismatic.OpenVLAForActionPrediction"
10
+ },
11
+ "d2f_block_size": 7,
12
+ "hf_llm_id": "meta-llama/Llama-2-7b-hf",
13
+ "image_resize_strategy": "resize-naive",
14
+ "image_sizes": [
15
+ 224,
16
+ 224
17
+ ],
18
+ "llm_backbone_id": "llama2-7b-pure",
19
+ "llm_max_length": 2048,
20
+ "mask_token_id": 32001,
21
+ "model_type": "openvla",
22
+ "n_action_bins": 256,
23
+ "norm_stats": {
24
+ "austin_buds_dataset_converted_externally_to_rlds": {
25
+ "action": {
26
+ "mask": [
27
+ true,
28
+ true,
29
+ true,
30
+ true,
31
+ true,
32
+ true,
33
+ false
34
+ ],
35
+ "max": [
36
+ 1.0,
37
+ 1.0,
38
+ 1.0,
39
+ 0.0,
40
+ 0.0,
41
+ 0.0,
42
+ 1.0
43
+ ],
44
+ "mean": [
45
+ -0.07678354531526566,
46
+ 0.0036849044263362885,
47
+ 0.05644911900162697,
48
+ 0.0,
49
+ 0.0,
50
+ 0.0,
51
+ 0.3510494828224182
52
+ ],
53
+ "min": [
54
+ -1.0,
55
+ -1.0,
56
+ -1.0,
57
+ 0.0,
58
+ 0.0,
59
+ 0.0,
60
+ 0.0
61
+ ],
62
+ "q01": [
63
+ -1.0,
64
+ -0.9599999785423279,
65
+ -0.8714285492897034,
66
+ 0.0,
67
+ 0.0,
68
+ 0.0,
69
+ 0.0
70
+ ],
71
+ "q99": [
72
+ 1.0,
73
+ 0.8600000143051147,
74
+ 1.0,
75
+ 0.0,
76
+ 0.0,
77
+ 0.0,
78
+ 1.0
79
+ ],
80
+ "std": [
81
+ 0.6367740631103516,
82
+ 0.37889179587364197,
83
+ 0.47796326875686646,
84
+ 0.0,
85
+ 0.0,
86
+ 0.0,
87
+ 0.47721168398857117
88
+ ]
89
+ },
90
+ "num_trajectories": 50,
91
+ "num_transitions": 34112,
92
+ "proprio": {
93
+ "max": [
94
+ 0.0,
95
+ 0.0,
96
+ 0.0,
97
+ 0.0,
98
+ 0.0,
99
+ 0.0,
100
+ 0.0
101
+ ],
102
+ "mean": [
103
+ 0.0,
104
+ 0.0,
105
+ 0.0,
106
+ 0.0,
107
+ 0.0,
108
+ 0.0,
109
+ 0.0
110
+ ],
111
+ "min": [
112
+ 0.0,
113
+ 0.0,
114
+ 0.0,
115
+ 0.0,
116
+ 0.0,
117
+ 0.0,
118
+ 0.0
119
+ ],
120
+ "q01": [
121
+ 0.0,
122
+ 0.0,
123
+ 0.0,
124
+ 0.0,
125
+ 0.0,
126
+ 0.0,
127
+ 0.0
128
+ ],
129
+ "q99": [
130
+ 0.0,
131
+ 0.0,
132
+ 0.0,
133
+ 0.0,
134
+ 0.0,
135
+ 0.0,
136
+ 0.0
137
+ ],
138
+ "std": [
139
+ 0.0,
140
+ 0.0,
141
+ 0.0,
142
+ 0.0,
143
+ 0.0,
144
+ 0.0,
145
+ 0.0
146
+ ]
147
+ }
148
+ },
149
+ "austin_sailor_dataset_converted_externally_to_rlds": {
150
+ "action": {
151
+ "mask": [
152
+ true,
153
+ true,
154
+ true,
155
+ true,
156
+ true,
157
+ true,
158
+ false
159
+ ],
160
+ "max": [
161
+ 1.0,
162
+ 1.0,
163
+ 1.0,
164
+ 0.0,
165
+ 0.0,
166
+ 0.375,
167
+ 1.0
168
+ ],
169
+ "mean": [
170
+ 0.011825348250567913,
171
+ 0.006461074110120535,
172
+ 0.06023626774549484,
173
+ 0.0,
174
+ 0.0,
175
+ 0.0016465914668515325,
176
+ 0.5260950326919556
177
+ ],
178
+ "min": [
179
+ -1.0,
180
+ -1.0,
181
+ -1.0,
182
+ 0.0,
183
+ 0.0,
184
+ -0.375,
185
+ 0.0
186
+ ],
187
+ "q01": [
188
+ -1.0,
189
+ -0.9828571677207947,
190
+ -0.6000000238418579,
191
+ 0.0,
192
+ 0.0,
193
+ -0.17249999940395355,
194
+ 0.0
195
+ ],
196
+ "q99": [
197
+ 1.0,
198
+ 0.9457142949104309,
199
+ 1.0,
200
+ 0.0,
201
+ 0.0,
202
+ 0.17892856895923615,
203
+ 1.0
204
+ ],
205
+ "std": [
206
+ 0.46348899602890015,
207
+ 0.41240179538726807,
208
+ 0.411862850189209,
209
+ 0.0,
210
+ 0.0,
211
+ 0.0578610822558403,
212
+ 0.49894046783447266
213
+ ]
214
+ },
215
+ "num_trajectories": 240,
216
+ "num_transitions": 353094,
217
+ "proprio": {
218
+ "max": [
219
+ 0.0,
220
+ 0.0,
221
+ 0.0,
222
+ 0.0,
223
+ 0.0,
224
+ 0.0,
225
+ 0.0
226
+ ],
227
+ "mean": [
228
+ 0.0,
229
+ 0.0,
230
+ 0.0,
231
+ 0.0,
232
+ 0.0,
233
+ 0.0,
234
+ 0.0
235
+ ],
236
+ "min": [
237
+ 0.0,
238
+ 0.0,
239
+ 0.0,
240
+ 0.0,
241
+ 0.0,
242
+ 0.0,
243
+ 0.0
244
+ ],
245
+ "q01": [
246
+ 0.0,
247
+ 0.0,
248
+ 0.0,
249
+ 0.0,
250
+ 0.0,
251
+ 0.0,
252
+ 0.0
253
+ ],
254
+ "q99": [
255
+ 0.0,
256
+ 0.0,
257
+ 0.0,
258
+ 0.0,
259
+ 0.0,
260
+ 0.0,
261
+ 0.0
262
+ ],
263
+ "std": [
264
+ 0.0,
265
+ 0.0,
266
+ 0.0,
267
+ 0.0,
268
+ 0.0,
269
+ 0.0,
270
+ 0.0
271
+ ]
272
+ }
273
+ },
274
+ "austin_sirius_dataset_converted_externally_to_rlds": {
275
+ "action": {
276
+ "mask": [
277
+ true,
278
+ true,
279
+ true,
280
+ true,
281
+ true,
282
+ true,
283
+ false
284
+ ],
285
+ "max": [
286
+ 1.0002285242080688,
287
+ 0.960608720779419,
288
+ 1.105179786682129,
289
+ 0.0,
290
+ 0.0,
291
+ 0.341785728931427,
292
+ 1.0
293
+ ],
294
+ "mean": [
295
+ 0.07747682929039001,
296
+ 0.03195561468601227,
297
+ 0.04244732856750488,
298
+ 0.0,
299
+ 0.0,
300
+ -0.01603456400334835,
301
+ 0.43260177969932556
302
+ ],
303
+ "min": [
304
+ -1.0183025598526,
305
+ -0.9800000190734863,
306
+ -0.9774575233459473,
307
+ 0.0,
308
+ 0.0,
309
+ -0.34607142210006714,
310
+ 0.0
311
+ ],
312
+ "q01": [
313
+ -0.780905865430832,
314
+ -0.5667179036140442,
315
+ -0.5254343223571777,
316
+ 0.0,
317
+ 0.0,
318
+ -0.28495091378688814,
319
+ 0.0
320
+ ],
321
+ "q99": [
322
+ 0.9569637751579284,
323
+ 0.6971374487876891,
324
+ 0.8124888157844541,
325
+ 0.0,
326
+ 0.0,
327
+ 0.1971428543329239,
328
+ 1.0
329
+ ],
330
+ "std": [
331
+ 0.3906329572200775,
332
+ 0.2998155355453491,
333
+ 0.2782271206378937,
334
+ 0.0,
335
+ 0.0,
336
+ 0.08120622485876083,
337
+ 0.49528297781944275
338
+ ]
339
+ },
340
+ "num_trajectories": 559,
341
+ "num_transitions": 279939,
342
+ "proprio": {
343
+ "max": [
344
+ 0.0,
345
+ 0.0,
346
+ 0.0,
347
+ 0.0,
348
+ 0.0,
349
+ 0.0,
350
+ 0.0
351
+ ],
352
+ "mean": [
353
+ 0.0,
354
+ 0.0,
355
+ 0.0,
356
+ 0.0,
357
+ 0.0,
358
+ 0.0,
359
+ 0.0
360
+ ],
361
+ "min": [
362
+ 0.0,
363
+ 0.0,
364
+ 0.0,
365
+ 0.0,
366
+ 0.0,
367
+ 0.0,
368
+ 0.0
369
+ ],
370
+ "q01": [
371
+ 0.0,
372
+ 0.0,
373
+ 0.0,
374
+ 0.0,
375
+ 0.0,
376
+ 0.0,
377
+ 0.0
378
+ ],
379
+ "q99": [
380
+ 0.0,
381
+ 0.0,
382
+ 0.0,
383
+ 0.0,
384
+ 0.0,
385
+ 0.0,
386
+ 0.0
387
+ ],
388
+ "std": [
389
+ 0.0,
390
+ 0.0,
391
+ 0.0,
392
+ 0.0,
393
+ 0.0,
394
+ 0.0,
395
+ 0.0
396
+ ]
397
+ }
398
+ },
399
+ "bc_z": {
400
+ "action": {
401
+ "mask": [
402
+ true,
403
+ true,
404
+ true,
405
+ true,
406
+ true,
407
+ true,
408
+ false
409
+ ],
410
+ "max": [
411
+ 0.2165454924106598,
412
+ 0.1251407265663147,
413
+ 0.10772687941789627,
414
+ 0.33544227480888367,
415
+ 0.28117990493774414,
416
+ 0.40614867210388184,
417
+ 1.0
418
+ ],
419
+ "mean": [
420
+ -0.009958467446267605,
421
+ 0.0008958321413956583,
422
+ 0.004995597992092371,
423
+ 0.00029755113064311445,
424
+ -0.008735382929444313,
425
+ -0.030693737789988518,
426
+ 0.8344562649726868
427
+ ],
428
+ "min": [
429
+ -0.1677047461271286,
430
+ -0.14630407094955444,
431
+ -0.10066790133714676,
432
+ -0.29421567916870117,
433
+ -0.32101404666900635,
434
+ -0.4635624885559082,
435
+ 0.0
436
+ ],
437
+ "q01": [
438
+ -0.09220654994249344,
439
+ -0.06456145539879798,
440
+ -0.049121275544166565,
441
+ -0.11594625547528267,
442
+ -0.14152548640966414,
443
+ -0.2251061636209488,
444
+ 0.0
445
+ ],
446
+ "q99": [
447
+ 0.07628866866230968,
448
+ 0.058019736707210584,
449
+ 0.052540797740221024,
450
+ 0.11740604028105736,
451
+ 0.11703975558280955,
452
+ 0.16729306846857078,
453
+ 1.0
454
+ ],
455
+ "std": [
456
+ 0.03053455986082554,
457
+ 0.0231423731893301,
458
+ 0.020641816779971123,
459
+ 0.04155943542718887,
460
+ 0.046427831053733826,
461
+ 0.0769818127155304,
462
+ 0.3610210120677948
463
+ ]
464
+ },
465
+ "num_trajectories": 43264,
466
+ "num_transitions": 6015535,
467
+ "proprio": {
468
+ "max": [
469
+ 0.0,
470
+ 0.0,
471
+ 0.0,
472
+ 0.0,
473
+ 0.0,
474
+ 0.0,
475
+ 0.0
476
+ ],
477
+ "mean": [
478
+ 0.0,
479
+ 0.0,
480
+ 0.0,
481
+ 0.0,
482
+ 0.0,
483
+ 0.0,
484
+ 0.0
485
+ ],
486
+ "min": [
487
+ 0.0,
488
+ 0.0,
489
+ 0.0,
490
+ 0.0,
491
+ 0.0,
492
+ 0.0,
493
+ 0.0
494
+ ],
495
+ "q01": [
496
+ 0.0,
497
+ 0.0,
498
+ 0.0,
499
+ 0.0,
500
+ 0.0,
501
+ 0.0,
502
+ 0.0
503
+ ],
504
+ "q99": [
505
+ 0.0,
506
+ 0.0,
507
+ 0.0,
508
+ 0.0,
509
+ 0.0,
510
+ 0.0,
511
+ 0.0
512
+ ],
513
+ "std": [
514
+ 0.0,
515
+ 0.0,
516
+ 0.0,
517
+ 0.0,
518
+ 0.0,
519
+ 0.0,
520
+ 0.0
521
+ ]
522
+ }
523
+ },
524
+ "berkeley_autolab_ur5": {
525
+ "action": {
526
+ "mask": [
527
+ true,
528
+ true,
529
+ true,
530
+ true,
531
+ true,
532
+ true,
533
+ false
534
+ ],
535
+ "max": [
536
+ 0.019999999552965164,
537
+ 0.019999999552965164,
538
+ 0.019999999552965164,
539
+ 0.06666667014360428,
540
+ 0.06666667014360428,
541
+ 0.06666667014360428,
542
+ 1.0
543
+ ],
544
+ "mean": [
545
+ 0.0005683620693162084,
546
+ 0.001217700308188796,
547
+ -0.0005296372692100704,
548
+ 0.00021029810886830091,
549
+ 6.0695128922816366e-05,
550
+ 0.001204986940138042,
551
+ 0.6298308372497559
552
+ ],
553
+ "min": [
554
+ -0.019999999552965164,
555
+ -0.019999999552965164,
556
+ -0.019999999552965164,
557
+ -0.06666667014360428,
558
+ -0.06666667014360428,
559
+ -0.06666667014360428,
560
+ 0.0
561
+ ],
562
+ "q01": [
563
+ -0.019999999552965164,
564
+ -0.019999999552965164,
565
+ -0.019999999552965164,
566
+ -0.02628571353852749,
567
+ -0.06666667014360428,
568
+ -0.03847619146108627,
569
+ 0.0
570
+ ],
571
+ "q99": [
572
+ 0.019999999552965164,
573
+ 0.019999999552965164,
574
+ 0.019999999552965164,
575
+ 0.031809523701667786,
576
+ 0.06666667014360428,
577
+ 0.036571428179740906,
578
+ 1.0
579
+ ],
580
+ "std": [
581
+ 0.0115329809486866,
582
+ 0.007990492507815361,
583
+ 0.009577835910022259,
584
+ 0.009432995691895485,
585
+ 0.016427582129836082,
586
+ 0.011053967289626598,
587
+ 0.48267969489097595
588
+ ]
589
+ },
590
+ "num_trajectories": 1000,
591
+ "num_transitions": 97939,
592
+ "proprio": {
593
+ "max": [
594
+ 0.0,
595
+ 0.0,
596
+ 0.0,
597
+ 0.0,
598
+ 0.0,
599
+ 0.0,
600
+ 0.0
601
+ ],
602
+ "mean": [
603
+ 0.0,
604
+ 0.0,
605
+ 0.0,
606
+ 0.0,
607
+ 0.0,
608
+ 0.0,
609
+ 0.0
610
+ ],
611
+ "min": [
612
+ 0.0,
613
+ 0.0,
614
+ 0.0,
615
+ 0.0,
616
+ 0.0,
617
+ 0.0,
618
+ 0.0
619
+ ],
620
+ "q01": [
621
+ 0.0,
622
+ 0.0,
623
+ 0.0,
624
+ 0.0,
625
+ 0.0,
626
+ 0.0,
627
+ 0.0
628
+ ],
629
+ "q99": [
630
+ 0.0,
631
+ 0.0,
632
+ 0.0,
633
+ 0.0,
634
+ 0.0,
635
+ 0.0,
636
+ 0.0
637
+ ],
638
+ "std": [
639
+ 0.0,
640
+ 0.0,
641
+ 0.0,
642
+ 0.0,
643
+ 0.0,
644
+ 0.0,
645
+ 0.0
646
+ ]
647
+ }
648
+ },
649
+ "berkeley_cable_routing": {
650
+ "action": {
651
+ "mask": [
652
+ true,
653
+ true,
654
+ true,
655
+ true,
656
+ true,
657
+ true,
658
+ false
659
+ ],
660
+ "max": [
661
+ 0.9633283019065857,
662
+ 1.0,
663
+ 1.0,
664
+ 0.0,
665
+ 0.0,
666
+ 1.0,
667
+ 0.0
668
+ ],
669
+ "mean": [
670
+ -0.07139874249696732,
671
+ 0.023609008640050888,
672
+ 0.10241943597793579,
673
+ 0.0,
674
+ 0.0,
675
+ 0.049671024084091187,
676
+ 0.0
677
+ ],
678
+ "min": [
679
+ -0.9809081554412842,
680
+ -0.9554349184036255,
681
+ -0.9994775056838989,
682
+ 0.0,
683
+ 0.0,
684
+ -1.0,
685
+ 0.0
686
+ ],
687
+ "q01": [
688
+ -0.5534318816661835,
689
+ -0.4797285574674606,
690
+ -0.5314934802055359,
691
+ 0.0,
692
+ 0.0,
693
+ -0.8855219376087189,
694
+ 0.0
695
+ ],
696
+ "q99": [
697
+ 0.42652835428714786,
698
+ 0.5000944086909298,
699
+ 0.639823433756829,
700
+ 0.0,
701
+ 0.0,
702
+ 0.984243879914284,
703
+ 0.0
704
+ ],
705
+ "std": [
706
+ 0.1815500408411026,
707
+ 0.1810990273952484,
708
+ 0.21220779418945312,
709
+ 0.0,
710
+ 0.0,
711
+ 0.3475511968135834,
712
+ 0.0
713
+ ]
714
+ },
715
+ "num_trajectories": 1647,
716
+ "num_transitions": 42328,
717
+ "proprio": {
718
+ "max": [
719
+ 0.0,
720
+ 0.0,
721
+ 0.0,
722
+ 0.0,
723
+ 0.0,
724
+ 0.0,
725
+ 0.0
726
+ ],
727
+ "mean": [
728
+ 0.0,
729
+ 0.0,
730
+ 0.0,
731
+ 0.0,
732
+ 0.0,
733
+ 0.0,
734
+ 0.0
735
+ ],
736
+ "min": [
737
+ 0.0,
738
+ 0.0,
739
+ 0.0,
740
+ 0.0,
741
+ 0.0,
742
+ 0.0,
743
+ 0.0
744
+ ],
745
+ "q01": [
746
+ 0.0,
747
+ 0.0,
748
+ 0.0,
749
+ 0.0,
750
+ 0.0,
751
+ 0.0,
752
+ 0.0
753
+ ],
754
+ "q99": [
755
+ 0.0,
756
+ 0.0,
757
+ 0.0,
758
+ 0.0,
759
+ 0.0,
760
+ 0.0,
761
+ 0.0
762
+ ],
763
+ "std": [
764
+ 0.0,
765
+ 0.0,
766
+ 0.0,
767
+ 0.0,
768
+ 0.0,
769
+ 0.0,
770
+ 0.0
771
+ ]
772
+ }
773
+ },
774
+ "berkeley_fanuc_manipulation": {
775
+ "action": {
776
+ "mask": [
777
+ true,
778
+ true,
779
+ true,
780
+ true,
781
+ true,
782
+ true,
783
+ false
784
+ ],
785
+ "max": [
786
+ 0.009999999776482582,
787
+ 0.009999999776482582,
788
+ 0.009999999776482582,
789
+ 0.03490658476948738,
790
+ 0.03490658476948738,
791
+ 0.03490658476948738,
792
+ 1.0
793
+ ],
794
+ "mean": [
795
+ 0.0007744057802483439,
796
+ -0.00031240080716088414,
797
+ -0.0015001941937953234,
798
+ -0.0007515158504247665,
799
+ -0.00015832878125365824,
800
+ 0.00014327642566058785,
801
+ 0.699295699596405
802
+ ],
803
+ "min": [
804
+ -0.009999999776482582,
805
+ -0.009999999776482582,
806
+ -0.009999999776482582,
807
+ -0.03490658476948738,
808
+ -0.03490658476948738,
809
+ -0.03490658476948738,
810
+ 0.0
811
+ ],
812
+ "q01": [
813
+ -0.009999999776482582,
814
+ -0.009999999776482582,
815
+ -0.009999999776482582,
816
+ -0.03490658476948738,
817
+ 0.0,
818
+ -0.03490658476948738,
819
+ 0.0
820
+ ],
821
+ "q99": [
822
+ 0.009999999776482582,
823
+ 0.009999999776482582,
824
+ 0.009999999776482582,
825
+ 0.03490658476948738,
826
+ 0.0,
827
+ 0.03490658476948738,
828
+ 1.0
829
+ ],
830
+ "std": [
831
+ 0.0034070091787725687,
832
+ 0.0049921851605176926,
833
+ 0.005344334989786148,
834
+ 0.00759894959628582,
835
+ 0.004081866703927517,
836
+ 0.008568956516683102,
837
+ 0.4586937427520752
838
+ ]
839
+ },
840
+ "num_trajectories": 415,
841
+ "num_transitions": 62613,
842
+ "proprio": {
843
+ "max": [
844
+ 0.0,
845
+ 0.0,
846
+ 0.0,
847
+ 0.0,
848
+ 0.0,
849
+ 0.0,
850
+ 0.0
851
+ ],
852
+ "mean": [
853
+ 0.0,
854
+ 0.0,
855
+ 0.0,
856
+ 0.0,
857
+ 0.0,
858
+ 0.0,
859
+ 0.0
860
+ ],
861
+ "min": [
862
+ 0.0,
863
+ 0.0,
864
+ 0.0,
865
+ 0.0,
866
+ 0.0,
867
+ 0.0,
868
+ 0.0
869
+ ],
870
+ "q01": [
871
+ 0.0,
872
+ 0.0,
873
+ 0.0,
874
+ 0.0,
875
+ 0.0,
876
+ 0.0,
877
+ 0.0
878
+ ],
879
+ "q99": [
880
+ 0.0,
881
+ 0.0,
882
+ 0.0,
883
+ 0.0,
884
+ 0.0,
885
+ 0.0,
886
+ 0.0
887
+ ],
888
+ "std": [
889
+ 0.0,
890
+ 0.0,
891
+ 0.0,
892
+ 0.0,
893
+ 0.0,
894
+ 0.0,
895
+ 0.0
896
+ ]
897
+ }
898
+ },
899
+ "bridge_orig": {
900
+ "action": {
901
+ "mask": [
902
+ true,
903
+ true,
904
+ true,
905
+ true,
906
+ true,
907
+ true,
908
+ false
909
+ ],
910
+ "max": [
911
+ 0.41691166162490845,
912
+ 0.25864794850349426,
913
+ 0.21218234300613403,
914
+ 3.122201919555664,
915
+ 1.8618112802505493,
916
+ 6.280478477478027,
917
+ 1.0
918
+ ],
919
+ "mean": [
920
+ 0.0002334194869035855,
921
+ 0.00013004911306779832,
922
+ -0.00012762474943883717,
923
+ -0.0001556558854645118,
924
+ -0.0004039328487124294,
925
+ 0.00023557482927571982,
926
+ 0.5764579176902771
927
+ ],
928
+ "min": [
929
+ -0.4007510244846344,
930
+ -0.13874775171279907,
931
+ -0.22553899884223938,
932
+ -3.2010786533355713,
933
+ -1.8618112802505493,
934
+ -6.279075622558594,
935
+ 0.0
936
+ ],
937
+ "q01": [
938
+ -0.02872725307941437,
939
+ -0.04170349963009357,
940
+ -0.026093858778476715,
941
+ -0.08092105075716972,
942
+ -0.09288699507713317,
943
+ -0.20718276381492615,
944
+ 0.0
945
+ ],
946
+ "q99": [
947
+ 0.028309678435325586,
948
+ 0.040855254605412394,
949
+ 0.040161586627364146,
950
+ 0.08192047759890528,
951
+ 0.07792850524187081,
952
+ 0.20382574498653397,
953
+ 1.0
954
+ ],
955
+ "std": [
956
+ 0.009765930473804474,
957
+ 0.013689135201275349,
958
+ 0.012667362578213215,
959
+ 0.028534092009067535,
960
+ 0.030637972056865692,
961
+ 0.07691419124603271,
962
+ 0.4973701536655426
963
+ ]
964
+ },
965
+ "num_trajectories": 60064,
966
+ "num_transitions": 2135463,
967
+ "proprio": {
968
+ "max": [
969
+ 0.0,
970
+ 0.0,
971
+ 0.0,
972
+ 0.0,
973
+ 0.0,
974
+ 0.0,
975
+ 0.0
976
+ ],
977
+ "mean": [
978
+ 0.0,
979
+ 0.0,
980
+ 0.0,
981
+ 0.0,
982
+ 0.0,
983
+ 0.0,
984
+ 0.0
985
+ ],
986
+ "min": [
987
+ 0.0,
988
+ 0.0,
989
+ 0.0,
990
+ 0.0,
991
+ 0.0,
992
+ 0.0,
993
+ 0.0
994
+ ],
995
+ "q01": [
996
+ 0.0,
997
+ 0.0,
998
+ 0.0,
999
+ 0.0,
1000
+ 0.0,
1001
+ 0.0,
1002
+ 0.0
1003
+ ],
1004
+ "q99": [
1005
+ 0.0,
1006
+ 0.0,
1007
+ 0.0,
1008
+ 0.0,
1009
+ 0.0,
1010
+ 0.0,
1011
+ 0.0
1012
+ ],
1013
+ "std": [
1014
+ 0.0,
1015
+ 0.0,
1016
+ 0.0,
1017
+ 0.0,
1018
+ 0.0,
1019
+ 0.0,
1020
+ 0.0
1021
+ ]
1022
+ }
1023
+ },
1024
+ "cmu_stretch": {
1025
+ "action": {
1026
+ "mask": [
1027
+ true,
1028
+ true,
1029
+ true,
1030
+ true,
1031
+ true,
1032
+ true,
1033
+ false
1034
+ ],
1035
+ "max": [
1036
+ 0.02338407188653946,
1037
+ 0.0,
1038
+ 0.023404927924275398,
1039
+ 0.0,
1040
+ 0.0,
1041
+ 0.0,
1042
+ 1.0
1043
+ ],
1044
+ "mean": [
1045
+ 0.00036304505192674696,
1046
+ 0.0,
1047
+ 0.0016466958913952112,
1048
+ 0.0,
1049
+ 0.0,
1050
+ 0.0,
1051
+ 0.3987048268318176
1052
+ ],
1053
+ "min": [
1054
+ -0.019353797659277916,
1055
+ 0.0,
1056
+ -0.02019215188920498,
1057
+ 0.0,
1058
+ 0.0,
1059
+ 0.0,
1060
+ 0.0
1061
+ ],
1062
+ "q01": [
1063
+ -0.011175686959177256,
1064
+ 0.0,
1065
+ -0.0032206363626755773,
1066
+ 0.0,
1067
+ 0.0,
1068
+ 0.0,
1069
+ 0.0
1070
+ ],
1071
+ "q99": [
1072
+ 0.014501785952597848,
1073
+ 0.0,
1074
+ 0.015056106168776728,
1075
+ 0.0,
1076
+ 0.0,
1077
+ 0.0,
1078
+ 1.0
1079
+ ],
1080
+ "std": [
1081
+ 0.004081828519701958,
1082
+ 0.0,
1083
+ 0.0037743328139185905,
1084
+ 0.0,
1085
+ 0.0,
1086
+ 0.0,
1087
+ 0.48963725566864014
1088
+ ]
1089
+ },
1090
+ "num_trajectories": 135,
1091
+ "num_transitions": 25016,
1092
+ "proprio": {
1093
+ "max": [
1094
+ 0.0,
1095
+ 0.0,
1096
+ 0.0,
1097
+ 0.0,
1098
+ 0.0,
1099
+ 0.0,
1100
+ 0.0
1101
+ ],
1102
+ "mean": [
1103
+ 0.0,
1104
+ 0.0,
1105
+ 0.0,
1106
+ 0.0,
1107
+ 0.0,
1108
+ 0.0,
1109
+ 0.0
1110
+ ],
1111
+ "min": [
1112
+ 0.0,
1113
+ 0.0,
1114
+ 0.0,
1115
+ 0.0,
1116
+ 0.0,
1117
+ 0.0,
1118
+ 0.0
1119
+ ],
1120
+ "q01": [
1121
+ 0.0,
1122
+ 0.0,
1123
+ 0.0,
1124
+ 0.0,
1125
+ 0.0,
1126
+ 0.0,
1127
+ 0.0
1128
+ ],
1129
+ "q99": [
1130
+ 0.0,
1131
+ 0.0,
1132
+ 0.0,
1133
+ 0.0,
1134
+ 0.0,
1135
+ 0.0,
1136
+ 0.0
1137
+ ],
1138
+ "std": [
1139
+ 0.0,
1140
+ 0.0,
1141
+ 0.0,
1142
+ 0.0,
1143
+ 0.0,
1144
+ 0.0,
1145
+ 0.0
1146
+ ]
1147
+ }
1148
+ },
1149
+ "dlr_edan_shared_control_converted_externally_to_rlds": {
1150
+ "action": {
1151
+ "mask": [
1152
+ true,
1153
+ true,
1154
+ true,
1155
+ true,
1156
+ true,
1157
+ true,
1158
+ false
1159
+ ],
1160
+ "max": [
1161
+ 0.18991442024707794,
1162
+ 0.0739002525806427,
1163
+ 0.18064819276332855,
1164
+ 0.0866486132144928,
1165
+ 0.13464981317520142,
1166
+ 0.16910280287265778,
1167
+ 1.0
1168
+ ],
1169
+ "mean": [
1170
+ 0.006647810339927673,
1171
+ -0.0007657372043468058,
1172
+ 0.006522852927446365,
1173
+ 0.0011679717572405934,
1174
+ -0.006395625416189432,
1175
+ -0.011902998201549053,
1176
+ 0.6985887289047241
1177
+ ],
1178
+ "min": [
1179
+ -0.10054297000169754,
1180
+ -0.08427435159683228,
1181
+ -0.13533438742160797,
1182
+ -0.17556548118591309,
1183
+ -0.18485672771930695,
1184
+ -0.2680685818195343,
1185
+ 0.0
1186
+ ],
1187
+ "q01": [
1188
+ -0.02987122368067503,
1189
+ -0.06013262912631035,
1190
+ -0.08286409199237824,
1191
+ -0.05924444157630205,
1192
+ -0.15986866518855095,
1193
+ -0.15636983573436739,
1194
+ 0.0
1195
+ ],
1196
+ "q99": [
1197
+ 0.08832092039287087,
1198
+ 0.042126184627413736,
1199
+ 0.11311905644834042,
1200
+ 0.0643695573508739,
1201
+ 0.03941855944693088,
1202
+ 0.156646853685379,
1203
+ 1.0
1204
+ ],
1205
+ "std": [
1206
+ 0.021393608301877975,
1207
+ 0.01814231649041176,
1208
+ 0.03374375030398369,
1209
+ 0.01743541844189167,
1210
+ 0.03394376486539841,
1211
+ 0.04641875624656677,
1212
+ 0.4588589072227478
1213
+ ]
1214
+ },
1215
+ "num_trajectories": 104,
1216
+ "num_transitions": 8928,
1217
+ "proprio": {
1218
+ "max": [
1219
+ 0.0,
1220
+ 0.0,
1221
+ 0.0,
1222
+ 0.0,
1223
+ 0.0,
1224
+ 0.0,
1225
+ 0.0
1226
+ ],
1227
+ "mean": [
1228
+ 0.0,
1229
+ 0.0,
1230
+ 0.0,
1231
+ 0.0,
1232
+ 0.0,
1233
+ 0.0,
1234
+ 0.0
1235
+ ],
1236
+ "min": [
1237
+ 0.0,
1238
+ 0.0,
1239
+ 0.0,
1240
+ 0.0,
1241
+ 0.0,
1242
+ 0.0,
1243
+ 0.0
1244
+ ],
1245
+ "q01": [
1246
+ 0.0,
1247
+ 0.0,
1248
+ 0.0,
1249
+ 0.0,
1250
+ 0.0,
1251
+ 0.0,
1252
+ 0.0
1253
+ ],
1254
+ "q99": [
1255
+ 0.0,
1256
+ 0.0,
1257
+ 0.0,
1258
+ 0.0,
1259
+ 0.0,
1260
+ 0.0,
1261
+ 0.0
1262
+ ],
1263
+ "std": [
1264
+ 0.0,
1265
+ 0.0,
1266
+ 0.0,
1267
+ 0.0,
1268
+ 0.0,
1269
+ 0.0,
1270
+ 0.0
1271
+ ]
1272
+ }
1273
+ },
1274
+ "dobbe": {
1275
+ "action": {
1276
+ "mask": [
1277
+ true,
1278
+ true,
1279
+ true,
1280
+ true,
1281
+ true,
1282
+ true,
1283
+ false
1284
+ ],
1285
+ "max": [
1286
+ 38.590423583984375,
1287
+ 17.932697296142578,
1288
+ 4.843764305114746,
1289
+ 1.4372116327285767,
1290
+ 0.4340403974056244,
1291
+ 1.2057193517684937,
1292
+ 0.9998947381973267
1293
+ ],
1294
+ "mean": [
1295
+ -0.0001120665911003016,
1296
+ 0.0011229600058868527,
1297
+ -0.00010194431524723768,
1298
+ -7.371398532995954e-05,
1299
+ -0.00067531579406932,
1300
+ -5.6643435527803376e-05,
1301
+ 0.6318281888961792
1302
+ ],
1303
+ "min": [
1304
+ -5.700923442840576,
1305
+ -21.605947494506836,
1306
+ -123.72489929199219,
1307
+ -1.7229845523834229,
1308
+ -0.4998578727245331,
1309
+ -0.8867913484573364,
1310
+ 1.4196479014572105e-06
1311
+ ],
1312
+ "q01": [
1313
+ -0.01119564864784479,
1314
+ -0.014266146533191203,
1315
+ -0.0071747214533388615,
1316
+ -0.009444301575422287,
1317
+ -0.03990109823644161,
1318
+ -0.017422311007976532,
1319
+ 4.003279136668425e-05
1320
+ ],
1321
+ "q99": [
1322
+ 0.01015154086053368,
1323
+ 0.017181577533483497,
1324
+ 0.007216989761218411,
1325
+ 0.010380979906767595,
1326
+ 0.03556173853576176,
1327
+ 0.018032474815845446,
1328
+ 0.9982578039169312
1329
+ ],
1330
+ "std": [
1331
+ 0.04264938458800316,
1332
+ 0.04428559169173241,
1333
+ 0.12224084138870239,
1334
+ 0.005388413090258837,
1335
+ 0.011246449314057827,
1336
+ 0.006287882570177317,
1337
+ 0.39732322096824646
1338
+ ]
1339
+ },
1340
+ "num_trajectories": 5208,
1341
+ "num_transitions": 1139911,
1342
+ "proprio": {
1343
+ "max": [
1344
+ 0.0,
1345
+ 0.0,
1346
+ 0.0,
1347
+ 0.0,
1348
+ 0.0,
1349
+ 0.0,
1350
+ 0.0
1351
+ ],
1352
+ "mean": [
1353
+ 0.0,
1354
+ 0.0,
1355
+ 0.0,
1356
+ 0.0,
1357
+ 0.0,
1358
+ 0.0,
1359
+ 0.0
1360
+ ],
1361
+ "min": [
1362
+ 0.0,
1363
+ 0.0,
1364
+ 0.0,
1365
+ 0.0,
1366
+ 0.0,
1367
+ 0.0,
1368
+ 0.0
1369
+ ],
1370
+ "q01": [
1371
+ 0.0,
1372
+ 0.0,
1373
+ 0.0,
1374
+ 0.0,
1375
+ 0.0,
1376
+ 0.0,
1377
+ 0.0
1378
+ ],
1379
+ "q99": [
1380
+ 0.0,
1381
+ 0.0,
1382
+ 0.0,
1383
+ 0.0,
1384
+ 0.0,
1385
+ 0.0,
1386
+ 0.0
1387
+ ],
1388
+ "std": [
1389
+ 0.0,
1390
+ 0.0,
1391
+ 0.0,
1392
+ 0.0,
1393
+ 0.0,
1394
+ 0.0,
1395
+ 0.0
1396
+ ]
1397
+ }
1398
+ },
1399
+ "fmb_dataset": {
1400
+ "action": {
1401
+ "mask": [
1402
+ true,
1403
+ true,
1404
+ true,
1405
+ true,
1406
+ true,
1407
+ true,
1408
+ false
1409
+ ],
1410
+ "max": [
1411
+ 1.399999976158142,
1412
+ 1.0,
1413
+ 1.399999976158142,
1414
+ 1.0,
1415
+ 1.0,
1416
+ 1.0,
1417
+ 1.0
1418
+ ],
1419
+ "mean": [
1420
+ 0.059029702097177505,
1421
+ -0.06476633995771408,
1422
+ -0.09787475317716599,
1423
+ 0.004325388930737972,
1424
+ 0.00028963794466108084,
1425
+ -0.04457257315516472,
1426
+ 0.7336440086364746
1427
+ ],
1428
+ "min": [
1429
+ -1.399999976158142,
1430
+ -1.399999976158142,
1431
+ -1.0,
1432
+ -1.0,
1433
+ -1.0,
1434
+ -1.0,
1435
+ 0.0
1436
+ ],
1437
+ "q01": [
1438
+ -0.8257142901420593,
1439
+ -1.399999976158142,
1440
+ -1.0,
1441
+ -1.0,
1442
+ -0.3028571307659149,
1443
+ -1.0,
1444
+ 0.0
1445
+ ],
1446
+ "q99": [
1447
+ 1.0,
1448
+ 0.5257142782211304,
1449
+ 1.0,
1450
+ 1.0,
1451
+ 0.3400000035762787,
1452
+ 1.0,
1453
+ 1.0
1454
+ ],
1455
+ "std": [
1456
+ 0.28809213638305664,
1457
+ 0.2820415794849396,
1458
+ 0.4626740515232086,
1459
+ 0.3266514539718628,
1460
+ 0.10842999070882797,
1461
+ 0.3440099358558655,
1462
+ 0.4435282051563263
1463
+ ]
1464
+ },
1465
+ "num_trajectories": 8612,
1466
+ "num_transitions": 1137459,
1467
+ "proprio": {
1468
+ "max": [
1469
+ 0.0,
1470
+ 0.0,
1471
+ 0.0,
1472
+ 0.0,
1473
+ 0.0,
1474
+ 0.0,
1475
+ 0.0
1476
+ ],
1477
+ "mean": [
1478
+ 0.0,
1479
+ 0.0,
1480
+ 0.0,
1481
+ 0.0,
1482
+ 0.0,
1483
+ 0.0,
1484
+ 0.0
1485
+ ],
1486
+ "min": [
1487
+ 0.0,
1488
+ 0.0,
1489
+ 0.0,
1490
+ 0.0,
1491
+ 0.0,
1492
+ 0.0,
1493
+ 0.0
1494
+ ],
1495
+ "q01": [
1496
+ 0.0,
1497
+ 0.0,
1498
+ 0.0,
1499
+ 0.0,
1500
+ 0.0,
1501
+ 0.0,
1502
+ 0.0
1503
+ ],
1504
+ "q99": [
1505
+ 0.0,
1506
+ 0.0,
1507
+ 0.0,
1508
+ 0.0,
1509
+ 0.0,
1510
+ 0.0,
1511
+ 0.0
1512
+ ],
1513
+ "std": [
1514
+ 0.0,
1515
+ 0.0,
1516
+ 0.0,
1517
+ 0.0,
1518
+ 0.0,
1519
+ 0.0,
1520
+ 0.0
1521
+ ]
1522
+ }
1523
+ },
1524
+ "fractal20220817_data": {
1525
+ "action": {
1526
+ "mask": [
1527
+ true,
1528
+ true,
1529
+ true,
1530
+ true,
1531
+ true,
1532
+ true,
1533
+ false
1534
+ ],
1535
+ "max": [
1536
+ 2.9984593391418457,
1537
+ 22.09052848815918,
1538
+ 2.7507524490356445,
1539
+ 1.570636510848999,
1540
+ 1.5321086645126343,
1541
+ 1.5691522359848022,
1542
+ 1.0
1543
+ ],
1544
+ "mean": [
1545
+ 0.006987582892179489,
1546
+ 0.006265917327255011,
1547
+ -0.01262515690177679,
1548
+ 0.04333311319351196,
1549
+ -0.005756212864071131,
1550
+ 0.0009130256366916001,
1551
+ 0.5354204773902893
1552
+ ],
1553
+ "min": [
1554
+ -2.0204520225524902,
1555
+ -5.497899532318115,
1556
+ -2.031663417816162,
1557
+ -1.569917917251587,
1558
+ -1.569892168045044,
1559
+ -1.570419430732727,
1560
+ 0.0
1561
+ ],
1562
+ "q01": [
1563
+ -0.22453527510166169,
1564
+ -0.14820013284683228,
1565
+ -0.231589707583189,
1566
+ -0.3517994859814644,
1567
+ -0.4193011274933815,
1568
+ -0.43643461108207704,
1569
+ 0.0
1570
+ ],
1571
+ "q99": [
1572
+ 0.17824687153100965,
1573
+ 0.14938379630446405,
1574
+ 0.21842354819178575,
1575
+ 0.5892666035890578,
1576
+ 0.35272657424211445,
1577
+ 0.44796681255102094,
1578
+ 1.0
1579
+ ],
1580
+ "std": [
1581
+ 0.0692116990685463,
1582
+ 0.05970962345600128,
1583
+ 0.07353084534406662,
1584
+ 0.15610496699810028,
1585
+ 0.13164450228214264,
1586
+ 0.14593800902366638,
1587
+ 0.497110515832901
1588
+ ]
1589
+ },
1590
+ "num_trajectories": 87212,
1591
+ "num_transitions": 3786400,
1592
+ "proprio": {
1593
+ "max": [
1594
+ 0.0,
1595
+ 0.0,
1596
+ 0.0,
1597
+ 0.0,
1598
+ 0.0,
1599
+ 0.0,
1600
+ 0.0
1601
+ ],
1602
+ "mean": [
1603
+ 0.0,
1604
+ 0.0,
1605
+ 0.0,
1606
+ 0.0,
1607
+ 0.0,
1608
+ 0.0,
1609
+ 0.0
1610
+ ],
1611
+ "min": [
1612
+ 0.0,
1613
+ 0.0,
1614
+ 0.0,
1615
+ 0.0,
1616
+ 0.0,
1617
+ 0.0,
1618
+ 0.0
1619
+ ],
1620
+ "q01": [
1621
+ 0.0,
1622
+ 0.0,
1623
+ 0.0,
1624
+ 0.0,
1625
+ 0.0,
1626
+ 0.0,
1627
+ 0.0
1628
+ ],
1629
+ "q99": [
1630
+ 0.0,
1631
+ 0.0,
1632
+ 0.0,
1633
+ 0.0,
1634
+ 0.0,
1635
+ 0.0,
1636
+ 0.0
1637
+ ],
1638
+ "std": [
1639
+ 0.0,
1640
+ 0.0,
1641
+ 0.0,
1642
+ 0.0,
1643
+ 0.0,
1644
+ 0.0,
1645
+ 0.0
1646
+ ]
1647
+ }
1648
+ },
1649
+ "furniture_bench_dataset_converted_externally_to_rlds": {
1650
+ "action": {
1651
+ "mask": [
1652
+ true,
1653
+ true,
1654
+ true,
1655
+ true,
1656
+ true,
1657
+ true,
1658
+ false
1659
+ ],
1660
+ "max": [
1661
+ 0.10000000149011612,
1662
+ 0.10000000149011612,
1663
+ 0.10000000149011612,
1664
+ 0.8651833534240723,
1665
+ 1.0909736156463623,
1666
+ 2.863185405731201,
1667
+ 1.0
1668
+ ],
1669
+ "mean": [
1670
+ 0.00014610752987209707,
1671
+ 0.0010830952087417245,
1672
+ 0.0006224989192560315,
1673
+ -0.003303206292912364,
1674
+ -0.0026880695950239897,
1675
+ 0.018242603167891502,
1676
+ 0.48854944109916687
1677
+ ],
1678
+ "min": [
1679
+ -0.10495579987764359,
1680
+ -0.10939455777406693,
1681
+ -0.10000000149011612,
1682
+ -0.971906840801239,
1683
+ -1.0475432872772217,
1684
+ -3.06000018119812,
1685
+ 0.0
1686
+ ],
1687
+ "q01": [
1688
+ -0.053988199681043625,
1689
+ -0.05049169331789017,
1690
+ -0.032499241530895236,
1691
+ -0.1953887003660202,
1692
+ -0.41674559473991396,
1693
+ -0.8886768388748169,
1694
+ 0.0
1695
+ ],
1696
+ "q99": [
1697
+ 0.05414841488003723,
1698
+ 0.04965164884924884,
1699
+ 0.060055799782276154,
1700
+ 0.18231668293476103,
1701
+ 0.39867786407470646,
1702
+ 0.8772023963928218,
1703
+ 1.0
1704
+ ],
1705
+ "std": [
1706
+ 0.01610708422958851,
1707
+ 0.014891477301716805,
1708
+ 0.014014219865202904,
1709
+ 0.058274295181035995,
1710
+ 0.11417088657617569,
1711
+ 0.33479776978492737,
1712
+ 0.49991825222969055
1713
+ ]
1714
+ },
1715
+ "num_trajectories": 5100,
1716
+ "num_transitions": 3948057,
1717
+ "proprio": {
1718
+ "max": [
1719
+ 0.0,
1720
+ 0.0,
1721
+ 0.0,
1722
+ 0.0,
1723
+ 0.0,
1724
+ 0.0,
1725
+ 0.0
1726
+ ],
1727
+ "mean": [
1728
+ 0.0,
1729
+ 0.0,
1730
+ 0.0,
1731
+ 0.0,
1732
+ 0.0,
1733
+ 0.0,
1734
+ 0.0
1735
+ ],
1736
+ "min": [
1737
+ 0.0,
1738
+ 0.0,
1739
+ 0.0,
1740
+ 0.0,
1741
+ 0.0,
1742
+ 0.0,
1743
+ 0.0
1744
+ ],
1745
+ "q01": [
1746
+ 0.0,
1747
+ 0.0,
1748
+ 0.0,
1749
+ 0.0,
1750
+ 0.0,
1751
+ 0.0,
1752
+ 0.0
1753
+ ],
1754
+ "q99": [
1755
+ 0.0,
1756
+ 0.0,
1757
+ 0.0,
1758
+ 0.0,
1759
+ 0.0,
1760
+ 0.0,
1761
+ 0.0
1762
+ ],
1763
+ "std": [
1764
+ 0.0,
1765
+ 0.0,
1766
+ 0.0,
1767
+ 0.0,
1768
+ 0.0,
1769
+ 0.0,
1770
+ 0.0
1771
+ ]
1772
+ }
1773
+ },
1774
+ "iamlab_cmu_pickup_insert_converted_externally_to_rlds": {
1775
+ "action": {
1776
+ "mask": [
1777
+ true,
1778
+ true,
1779
+ true,
1780
+ true,
1781
+ true,
1782
+ true,
1783
+ false
1784
+ ],
1785
+ "max": [
1786
+ 0.6634981632232666,
1787
+ 0.23428471386432648,
1788
+ 0.4308285415172577,
1789
+ 3.1415927410125732,
1790
+ 0.13647015392780304,
1791
+ 3.141592502593994,
1792
+ 1.0
1793
+ ],
1794
+ "mean": [
1795
+ 0.5274372696876526,
1796
+ 0.02858201041817665,
1797
+ 0.18712575733661652,
1798
+ 1.2339589595794678,
1799
+ 0.03226623684167862,
1800
+ -1.4199490547180176,
1801
+ 0.5550631880760193
1802
+ ],
1803
+ "min": [
1804
+ 0.3071657121181488,
1805
+ -0.29754969477653503,
1806
+ 0.06578229367733002,
1807
+ -3.1415927410125732,
1808
+ -0.04584203287959099,
1809
+ -3.141592502593994,
1810
+ 0.0
1811
+ ],
1812
+ "q01": [
1813
+ 0.3148897051811218,
1814
+ -0.20317550599575043,
1815
+ 0.06785467118024827,
1816
+ -3.140952730178833,
1817
+ -0.029743434861302376,
1818
+ -3.141091251373291,
1819
+ 0.0
1820
+ ],
1821
+ "q99": [
1822
+ 0.6472805738449097,
1823
+ 0.20846802592277527,
1824
+ 0.36855655312538155,
1825
+ 3.1409926891326903,
1826
+ 0.11424950212240226,
1827
+ 3.1410969257354737,
1828
+ 1.0
1829
+ ],
1830
+ "std": [
1831
+ 0.08108345419168472,
1832
+ 0.1116757020354271,
1833
+ 0.07747554779052734,
1834
+ 2.8737246990203857,
1835
+ 0.02774704433977604,
1836
+ 2.7678682804107666,
1837
+ 0.49695101380348206
1838
+ ]
1839
+ },
1840
+ "num_trajectories": 631,
1841
+ "num_transitions": 146241,
1842
+ "proprio": {
1843
+ "max": [
1844
+ 0.0,
1845
+ 0.0,
1846
+ 0.0,
1847
+ 0.0,
1848
+ 0.0,
1849
+ 0.0,
1850
+ 0.0
1851
+ ],
1852
+ "mean": [
1853
+ 0.0,
1854
+ 0.0,
1855
+ 0.0,
1856
+ 0.0,
1857
+ 0.0,
1858
+ 0.0,
1859
+ 0.0
1860
+ ],
1861
+ "min": [
1862
+ 0.0,
1863
+ 0.0,
1864
+ 0.0,
1865
+ 0.0,
1866
+ 0.0,
1867
+ 0.0,
1868
+ 0.0
1869
+ ],
1870
+ "q01": [
1871
+ 0.0,
1872
+ 0.0,
1873
+ 0.0,
1874
+ 0.0,
1875
+ 0.0,
1876
+ 0.0,
1877
+ 0.0
1878
+ ],
1879
+ "q99": [
1880
+ 0.0,
1881
+ 0.0,
1882
+ 0.0,
1883
+ 0.0,
1884
+ 0.0,
1885
+ 0.0,
1886
+ 0.0
1887
+ ],
1888
+ "std": [
1889
+ 0.0,
1890
+ 0.0,
1891
+ 0.0,
1892
+ 0.0,
1893
+ 0.0,
1894
+ 0.0,
1895
+ 0.0
1896
+ ]
1897
+ }
1898
+ },
1899
+ "jaco_play": {
1900
+ "action": {
1901
+ "mask": [
1902
+ true,
1903
+ true,
1904
+ true,
1905
+ true,
1906
+ true,
1907
+ true,
1908
+ false
1909
+ ],
1910
+ "max": [
1911
+ 0.20000000298023224,
1912
+ 0.20000000298023224,
1913
+ 0.20000000298023224,
1914
+ 0.0,
1915
+ 0.0,
1916
+ 0.0,
1917
+ 1.0
1918
+ ],
1919
+ "mean": [
1920
+ 0.0009658430935814977,
1921
+ -0.00580078037455678,
1922
+ -0.00395062193274498,
1923
+ 0.0,
1924
+ 0.0,
1925
+ 0.0,
1926
+ 0.34934908151626587
1927
+ ],
1928
+ "min": [
1929
+ -0.20000000298023224,
1930
+ -0.20000000298023224,
1931
+ -0.20000000298023224,
1932
+ 0.0,
1933
+ 0.0,
1934
+ 0.0,
1935
+ 0.0
1936
+ ],
1937
+ "q01": [
1938
+ -0.20000000298023224,
1939
+ -0.20000000298023224,
1940
+ -0.20000000298023224,
1941
+ 0.0,
1942
+ 0.0,
1943
+ 0.0,
1944
+ 0.0
1945
+ ],
1946
+ "q99": [
1947
+ 0.20000000298023224,
1948
+ 0.20000000298023224,
1949
+ 0.20000000298023224,
1950
+ 0.0,
1951
+ 0.0,
1952
+ 0.0,
1953
+ 1.0
1954
+ ],
1955
+ "std": [
1956
+ 0.12235074490308762,
1957
+ 0.09678777307271957,
1958
+ 0.11155334860086441,
1959
+ 0.0,
1960
+ 0.0,
1961
+ 0.0,
1962
+ 0.4768252968788147
1963
+ ]
1964
+ },
1965
+ "num_trajectories": 1085,
1966
+ "num_transitions": 77965,
1967
+ "proprio": {
1968
+ "max": [
1969
+ 0.0,
1970
+ 0.0,
1971
+ 0.0,
1972
+ 0.0,
1973
+ 0.0,
1974
+ 0.0,
1975
+ 0.0
1976
+ ],
1977
+ "mean": [
1978
+ 0.0,
1979
+ 0.0,
1980
+ 0.0,
1981
+ 0.0,
1982
+ 0.0,
1983
+ 0.0,
1984
+ 0.0
1985
+ ],
1986
+ "min": [
1987
+ 0.0,
1988
+ 0.0,
1989
+ 0.0,
1990
+ 0.0,
1991
+ 0.0,
1992
+ 0.0,
1993
+ 0.0
1994
+ ],
1995
+ "q01": [
1996
+ 0.0,
1997
+ 0.0,
1998
+ 0.0,
1999
+ 0.0,
2000
+ 0.0,
2001
+ 0.0,
2002
+ 0.0
2003
+ ],
2004
+ "q99": [
2005
+ 0.0,
2006
+ 0.0,
2007
+ 0.0,
2008
+ 0.0,
2009
+ 0.0,
2010
+ 0.0,
2011
+ 0.0
2012
+ ],
2013
+ "std": [
2014
+ 0.0,
2015
+ 0.0,
2016
+ 0.0,
2017
+ 0.0,
2018
+ 0.0,
2019
+ 0.0,
2020
+ 0.0
2021
+ ]
2022
+ }
2023
+ },
2024
+ "kuka": {
2025
+ "action": {
2026
+ "mask": [
2027
+ true,
2028
+ true,
2029
+ true,
2030
+ true,
2031
+ true,
2032
+ true,
2033
+ false
2034
+ ],
2035
+ "max": [
2036
+ 0.1697135865688324,
2037
+ 0.2777623236179352,
2038
+ 0.43710532784461975,
2039
+ 0.0,
2040
+ 0.0,
2041
+ 1.9684287309646606,
2042
+ 1.0
2043
+ ],
2044
+ "mean": [
2045
+ -0.0004668905457947403,
2046
+ 0.00040138536132872105,
2047
+ -0.001280792523175478,
2048
+ 0.0,
2049
+ 0.0,
2050
+ -0.03722453489899635,
2051
+ 0.4131543040275574
2052
+ ],
2053
+ "min": [
2054
+ -0.159867063164711,
2055
+ -0.2892282009124756,
2056
+ -0.2795473635196686,
2057
+ 0.0,
2058
+ 0.0,
2059
+ -1.9875637292861938,
2060
+ 0.0
2061
+ ],
2062
+ "q01": [
2063
+ -0.06619441494345665,
2064
+ -0.08713878810405731,
2065
+ -0.15083016991615295,
2066
+ 0.0,
2067
+ 0.0,
2068
+ -0.5415697038173676,
2069
+ 0.0
2070
+ ],
2071
+ "q99": [
2072
+ 0.06601839080452929,
2073
+ 0.08732476785779003,
2074
+ 0.18168179214000715,
2075
+ 0.0,
2076
+ 0.0,
2077
+ 0.2923380345106127,
2078
+ 1.0
2079
+ ],
2080
+ "std": [
2081
+ 0.02083250693976879,
2082
+ 0.02915887162089348,
2083
+ 0.06422865390777588,
2084
+ 0.0,
2085
+ 0.0,
2086
+ 0.14224295318126678,
2087
+ 0.49086448550224304
2088
+ ]
2089
+ },
2090
+ "num_trajectories": 209880,
2091
+ "num_transitions": 2455879,
2092
+ "proprio": {
2093
+ "max": [
2094
+ 0.0,
2095
+ 0.0,
2096
+ 0.0,
2097
+ 0.0,
2098
+ 0.0,
2099
+ 0.0,
2100
+ 0.0
2101
+ ],
2102
+ "mean": [
2103
+ 0.0,
2104
+ 0.0,
2105
+ 0.0,
2106
+ 0.0,
2107
+ 0.0,
2108
+ 0.0,
2109
+ 0.0
2110
+ ],
2111
+ "min": [
2112
+ 0.0,
2113
+ 0.0,
2114
+ 0.0,
2115
+ 0.0,
2116
+ 0.0,
2117
+ 0.0,
2118
+ 0.0
2119
+ ],
2120
+ "q01": [
2121
+ 0.0,
2122
+ 0.0,
2123
+ 0.0,
2124
+ 0.0,
2125
+ 0.0,
2126
+ 0.0,
2127
+ 0.0
2128
+ ],
2129
+ "q99": [
2130
+ 0.0,
2131
+ 0.0,
2132
+ 0.0,
2133
+ 0.0,
2134
+ 0.0,
2135
+ 0.0,
2136
+ 0.0
2137
+ ],
2138
+ "std": [
2139
+ 0.0,
2140
+ 0.0,
2141
+ 0.0,
2142
+ 0.0,
2143
+ 0.0,
2144
+ 0.0,
2145
+ 0.0
2146
+ ]
2147
+ }
2148
+ },
2149
+ "nyu_franka_play_dataset_converted_externally_to_rlds": {
2150
+ "action": {
2151
+ "mask": [
2152
+ true,
2153
+ true,
2154
+ true,
2155
+ true,
2156
+ true,
2157
+ true,
2158
+ false
2159
+ ],
2160
+ "max": [
2161
+ 0.06424188613891602,
2162
+ 0.07027634978294373,
2163
+ 0.06129661202430725,
2164
+ 6.281067848205566,
2165
+ 0.1967729926109314,
2166
+ 0.26377415657043457,
2167
+ 1.0
2168
+ ],
2169
+ "mean": [
2170
+ 0.001021989737637341,
2171
+ -0.00012002651783404872,
2172
+ 0.00032894269679673016,
2173
+ 0.0015034361276775599,
2174
+ -0.002198522910475731,
2175
+ -0.001663230243138969,
2176
+ 0.7230083346366882
2177
+ ],
2178
+ "min": [
2179
+ -0.05952230095863342,
2180
+ -0.07232445478439331,
2181
+ -0.06730806827545166,
2182
+ -6.278434753417969,
2183
+ -0.21479034423828125,
2184
+ -0.3627619743347168,
2185
+ 0.0
2186
+ ],
2187
+ "q01": [
2188
+ -0.03199600875377655,
2189
+ -0.032861671447753905,
2190
+ -0.03368805110454559,
2191
+ -0.12080862045288086,
2192
+ -0.12175218224525451,
2193
+ -0.11370223641395569,
2194
+ 0.0
2195
+ ],
2196
+ "q99": [
2197
+ 0.03101520001888276,
2198
+ 0.0373908892273903,
2199
+ 0.03646374464035038,
2200
+ 0.11764093399047852,
2201
+ 0.1258920183777809,
2202
+ 0.09366151213645942,
2203
+ 1.0
2204
+ ],
2205
+ "std": [
2206
+ 0.01327415369451046,
2207
+ 0.013215910643339157,
2208
+ 0.012822109274566174,
2209
+ 0.2732451558113098,
2210
+ 0.057022541761398315,
2211
+ 0.039172880351543427,
2212
+ 0.44752755761146545
2213
+ ]
2214
+ },
2215
+ "num_trajectories": 456,
2216
+ "num_transitions": 44875,
2217
+ "proprio": {
2218
+ "max": [
2219
+ 0.0,
2220
+ 0.0,
2221
+ 0.0,
2222
+ 0.0,
2223
+ 0.0,
2224
+ 0.0,
2225
+ 0.0
2226
+ ],
2227
+ "mean": [
2228
+ 0.0,
2229
+ 0.0,
2230
+ 0.0,
2231
+ 0.0,
2232
+ 0.0,
2233
+ 0.0,
2234
+ 0.0
2235
+ ],
2236
+ "min": [
2237
+ 0.0,
2238
+ 0.0,
2239
+ 0.0,
2240
+ 0.0,
2241
+ 0.0,
2242
+ 0.0,
2243
+ 0.0
2244
+ ],
2245
+ "q01": [
2246
+ 0.0,
2247
+ 0.0,
2248
+ 0.0,
2249
+ 0.0,
2250
+ 0.0,
2251
+ 0.0,
2252
+ 0.0
2253
+ ],
2254
+ "q99": [
2255
+ 0.0,
2256
+ 0.0,
2257
+ 0.0,
2258
+ 0.0,
2259
+ 0.0,
2260
+ 0.0,
2261
+ 0.0
2262
+ ],
2263
+ "std": [
2264
+ 0.0,
2265
+ 0.0,
2266
+ 0.0,
2267
+ 0.0,
2268
+ 0.0,
2269
+ 0.0,
2270
+ 0.0
2271
+ ]
2272
+ }
2273
+ },
2274
+ "roboturk": {
2275
+ "action": {
2276
+ "mask": [
2277
+ true,
2278
+ true,
2279
+ true,
2280
+ true,
2281
+ true,
2282
+ true,
2283
+ false
2284
+ ],
2285
+ "max": [
2286
+ 0.39124172925949097,
2287
+ 0.4601028263568878,
2288
+ 0.4870833456516266,
2289
+ 1.816888689994812,
2290
+ 1.8240282535552979,
2291
+ 1.4824820756912231,
2292
+ 1.0
2293
+ ],
2294
+ "mean": [
2295
+ 0.0014448732836171985,
2296
+ -0.0015945249469950795,
2297
+ -0.0011753785656765103,
2298
+ 0.0023012510500848293,
2299
+ -0.0009382463176734746,
2300
+ -0.00011485807772260159,
2301
+ 0.5746025443077087
2302
+ ],
2303
+ "min": [
2304
+ -0.6546999216079712,
2305
+ -0.6365841031074524,
2306
+ -0.4217723608016968,
2307
+ -1.6695482730865479,
2308
+ -1.8023357391357422,
2309
+ -1.4630827903747559,
2310
+ 0.0
2311
+ ],
2312
+ "q01": [
2313
+ -0.1342635464668274,
2314
+ -0.19996687173843383,
2315
+ -0.1482972100377083,
2316
+ -0.20720748245716095,
2317
+ -0.09676413893699647,
2318
+ -0.18075634717941286,
2319
+ 0.0
2320
+ ],
2321
+ "q99": [
2322
+ 0.14956976801157001,
2323
+ 0.1805950567126275,
2324
+ 0.18841815620660796,
2325
+ 0.21615413755178453,
2326
+ 0.09457383215427405,
2327
+ 0.18543301910162005,
2328
+ 1.0
2329
+ ],
2330
+ "std": [
2331
+ 0.04935386776924133,
2332
+ 0.0635455846786499,
2333
+ 0.061164740473032,
2334
+ 0.09553450345993042,
2335
+ 0.08420111238956451,
2336
+ 0.06517903506755829,
2337
+ 0.49452081322669983
2338
+ ]
2339
+ },
2340
+ "num_trajectories": 1995,
2341
+ "num_transitions": 187507,
2342
+ "proprio": {
2343
+ "max": [
2344
+ 0.0,
2345
+ 0.0,
2346
+ 0.0,
2347
+ 0.0,
2348
+ 0.0,
2349
+ 0.0,
2350
+ 0.0
2351
+ ],
2352
+ "mean": [
2353
+ 0.0,
2354
+ 0.0,
2355
+ 0.0,
2356
+ 0.0,
2357
+ 0.0,
2358
+ 0.0,
2359
+ 0.0
2360
+ ],
2361
+ "min": [
2362
+ 0.0,
2363
+ 0.0,
2364
+ 0.0,
2365
+ 0.0,
2366
+ 0.0,
2367
+ 0.0,
2368
+ 0.0
2369
+ ],
2370
+ "q01": [
2371
+ 0.0,
2372
+ 0.0,
2373
+ 0.0,
2374
+ 0.0,
2375
+ 0.0,
2376
+ 0.0,
2377
+ 0.0
2378
+ ],
2379
+ "q99": [
2380
+ 0.0,
2381
+ 0.0,
2382
+ 0.0,
2383
+ 0.0,
2384
+ 0.0,
2385
+ 0.0,
2386
+ 0.0
2387
+ ],
2388
+ "std": [
2389
+ 0.0,
2390
+ 0.0,
2391
+ 0.0,
2392
+ 0.0,
2393
+ 0.0,
2394
+ 0.0,
2395
+ 0.0
2396
+ ]
2397
+ }
2398
+ },
2399
+ "stanford_hydra_dataset_converted_externally_to_rlds": {
2400
+ "action": {
2401
+ "mask": [
2402
+ true,
2403
+ true,
2404
+ true,
2405
+ true,
2406
+ true,
2407
+ true,
2408
+ false
2409
+ ],
2410
+ "max": [
2411
+ 0.02499854564666748,
2412
+ 0.02499903365969658,
2413
+ 0.024999922141432762,
2414
+ 0.24974457919597626,
2415
+ 0.24997030198574066,
2416
+ 0.24999946355819702,
2417
+ 1.0
2418
+ ],
2419
+ "mean": [
2420
+ 0.0007790001109242439,
2421
+ 0.00013707754260394722,
2422
+ -0.0002548607881180942,
2423
+ 0.0012903271708637476,
2424
+ -0.004751681815832853,
2425
+ 0.002692886395379901,
2426
+ 0.48855218291282654
2427
+ ],
2428
+ "min": [
2429
+ -0.024999044835567474,
2430
+ -0.024999700486660004,
2431
+ -0.02499929815530777,
2432
+ -0.24993225932121277,
2433
+ -0.2499666064977646,
2434
+ -0.2499932497739792,
2435
+ 0.0
2436
+ ],
2437
+ "q01": [
2438
+ -0.019992006458342076,
2439
+ -0.02415412735193968,
2440
+ -0.022941758055239916,
2441
+ -0.11085530579090118,
2442
+ -0.12024572037160397,
2443
+ -0.13314770206809043,
2444
+ 0.0
2445
+ ],
2446
+ "q99": [
2447
+ 0.022886231057345868,
2448
+ 0.022358838934451335,
2449
+ 0.02410089675337076,
2450
+ 0.12370114490389822,
2451
+ 0.11323311634361738,
2452
+ 0.18474749639630164,
2453
+ 1.0
2454
+ ],
2455
+ "std": [
2456
+ 0.008022161200642586,
2457
+ 0.009131459519267082,
2458
+ 0.009574338793754578,
2459
+ 0.04122216999530792,
2460
+ 0.0384303517639637,
2461
+ 0.04606688767671585,
2462
+ 0.49976691603660583
2463
+ ]
2464
+ },
2465
+ "num_trajectories": 570,
2466
+ "num_transitions": 358234,
2467
+ "proprio": {
2468
+ "max": [
2469
+ 0.0,
2470
+ 0.0,
2471
+ 0.0,
2472
+ 0.0,
2473
+ 0.0,
2474
+ 0.0,
2475
+ 0.0
2476
+ ],
2477
+ "mean": [
2478
+ 0.0,
2479
+ 0.0,
2480
+ 0.0,
2481
+ 0.0,
2482
+ 0.0,
2483
+ 0.0,
2484
+ 0.0
2485
+ ],
2486
+ "min": [
2487
+ 0.0,
2488
+ 0.0,
2489
+ 0.0,
2490
+ 0.0,
2491
+ 0.0,
2492
+ 0.0,
2493
+ 0.0
2494
+ ],
2495
+ "q01": [
2496
+ 0.0,
2497
+ 0.0,
2498
+ 0.0,
2499
+ 0.0,
2500
+ 0.0,
2501
+ 0.0,
2502
+ 0.0
2503
+ ],
2504
+ "q99": [
2505
+ 0.0,
2506
+ 0.0,
2507
+ 0.0,
2508
+ 0.0,
2509
+ 0.0,
2510
+ 0.0,
2511
+ 0.0
2512
+ ],
2513
+ "std": [
2514
+ 0.0,
2515
+ 0.0,
2516
+ 0.0,
2517
+ 0.0,
2518
+ 0.0,
2519
+ 0.0,
2520
+ 0.0
2521
+ ]
2522
+ }
2523
+ },
2524
+ "taco_play": {
2525
+ "action": {
2526
+ "mask": [
2527
+ true,
2528
+ true,
2529
+ true,
2530
+ true,
2531
+ true,
2532
+ true,
2533
+ false
2534
+ ],
2535
+ "max": [
2536
+ 1.4915844202041626,
2537
+ 2.1842432022094727,
2538
+ 2.6836395263671875,
2539
+ 5.035226821899414,
2540
+ 2.665864944458008,
2541
+ 4.250768661499023,
2542
+ 1.0
2543
+ ],
2544
+ "mean": [
2545
+ -0.003845922416076064,
2546
+ 0.009671456180512905,
2547
+ 0.012780580669641495,
2548
+ -0.005403771996498108,
2549
+ -0.009606587700545788,
2550
+ -0.002480733208358288,
2551
+ 0.4263913035392761
2552
+ ],
2553
+ "min": [
2554
+ -4.242457866668701,
2555
+ -3.192805051803589,
2556
+ -1.3371467590332031,
2557
+ -4.202683448791504,
2558
+ -2.6722638607025146,
2559
+ -3.3467135429382324,
2560
+ 0.0
2561
+ ],
2562
+ "q01": [
2563
+ -0.7106140398979186,
2564
+ -1.056944659948349,
2565
+ -0.5878450274467468,
2566
+ -0.7682853937149048,
2567
+ -0.7180147767066956,
2568
+ -1.5527938604354858,
2569
+ 0.0
2570
+ ],
2571
+ "q99": [
2572
+ 0.6482916426658629,
2573
+ 1.0051310062408447,
2574
+ 0.9480248689651489,
2575
+ 0.6926478147506714,
2576
+ 0.6351067513227462,
2577
+ 1.628010264635086,
2578
+ 1.0
2579
+ ],
2580
+ "std": [
2581
+ 0.23254038393497467,
2582
+ 0.36298269033432007,
2583
+ 0.28692901134490967,
2584
+ 0.2617705166339874,
2585
+ 0.2438892275094986,
2586
+ 0.5216503143310547,
2587
+ 0.4946896731853485
2588
+ ]
2589
+ },
2590
+ "num_trajectories": 3603,
2591
+ "num_transitions": 237798,
2592
+ "proprio": {
2593
+ "max": [
2594
+ 0.0,
2595
+ 0.0,
2596
+ 0.0,
2597
+ 0.0,
2598
+ 0.0,
2599
+ 0.0,
2600
+ 0.0
2601
+ ],
2602
+ "mean": [
2603
+ 0.0,
2604
+ 0.0,
2605
+ 0.0,
2606
+ 0.0,
2607
+ 0.0,
2608
+ 0.0,
2609
+ 0.0
2610
+ ],
2611
+ "min": [
2612
+ 0.0,
2613
+ 0.0,
2614
+ 0.0,
2615
+ 0.0,
2616
+ 0.0,
2617
+ 0.0,
2618
+ 0.0
2619
+ ],
2620
+ "q01": [
2621
+ 0.0,
2622
+ 0.0,
2623
+ 0.0,
2624
+ 0.0,
2625
+ 0.0,
2626
+ 0.0,
2627
+ 0.0
2628
+ ],
2629
+ "q99": [
2630
+ 0.0,
2631
+ 0.0,
2632
+ 0.0,
2633
+ 0.0,
2634
+ 0.0,
2635
+ 0.0,
2636
+ 0.0
2637
+ ],
2638
+ "std": [
2639
+ 0.0,
2640
+ 0.0,
2641
+ 0.0,
2642
+ 0.0,
2643
+ 0.0,
2644
+ 0.0,
2645
+ 0.0
2646
+ ]
2647
+ }
2648
+ },
2649
+ "toto": {
2650
+ "action": {
2651
+ "mask": [
2652
+ true,
2653
+ true,
2654
+ true,
2655
+ true,
2656
+ true,
2657
+ true,
2658
+ false
2659
+ ],
2660
+ "max": [
2661
+ 0.6839867234230042,
2662
+ 0.4454185664653778,
2663
+ 0.7984078526496887,
2664
+ 2.120781660079956,
2665
+ 1.371164321899414,
2666
+ 1.4118704795837402,
2667
+ 0.0
2668
+ ],
2669
+ "mean": [
2670
+ 0.38542115688323975,
2671
+ 0.007769413758069277,
2672
+ 0.3632740378379822,
2673
+ -0.6652036905288696,
2674
+ 0.1890396922826767,
2675
+ 0.03298724442720413,
2676
+ 0.0
2677
+ ],
2678
+ "min": [
2679
+ 0.09922284632921219,
2680
+ -0.5180193781852722,
2681
+ 0.13791072368621826,
2682
+ -2.635117530822754,
2683
+ -1.0734480619430542,
2684
+ -1.9282547235488892,
2685
+ 0.0
2686
+ ],
2687
+ "q01": [
2688
+ 0.1756722891330719,
2689
+ -0.3077590811252594,
2690
+ 0.235383919775486,
2691
+ -2.0908505964279174,
2692
+ -0.6191593289375306,
2693
+ -0.7488683319091797,
2694
+ 0.0
2695
+ ],
2696
+ "q99": [
2697
+ 0.6136963081359863,
2698
+ 0.33704194784164443,
2699
+ 0.6681221985816956,
2700
+ 0.7422861719131538,
2701
+ 0.7955395007133507,
2702
+ 0.740464625358582,
2703
+ 0.0
2704
+ ],
2705
+ "std": [
2706
+ 0.12211652100086212,
2707
+ 0.19378550350666046,
2708
+ 0.10178236663341522,
2709
+ 0.5725259184837341,
2710
+ 0.29884573817253113,
2711
+ 0.3259911835193634,
2712
+ 0.0
2713
+ ]
2714
+ },
2715
+ "num_trajectories": 1003,
2716
+ "num_transitions": 325699,
2717
+ "proprio": {
2718
+ "max": [
2719
+ 0.0,
2720
+ 0.0,
2721
+ 0.0,
2722
+ 0.0,
2723
+ 0.0,
2724
+ 0.0,
2725
+ 0.0
2726
+ ],
2727
+ "mean": [
2728
+ 0.0,
2729
+ 0.0,
2730
+ 0.0,
2731
+ 0.0,
2732
+ 0.0,
2733
+ 0.0,
2734
+ 0.0
2735
+ ],
2736
+ "min": [
2737
+ 0.0,
2738
+ 0.0,
2739
+ 0.0,
2740
+ 0.0,
2741
+ 0.0,
2742
+ 0.0,
2743
+ 0.0
2744
+ ],
2745
+ "q01": [
2746
+ 0.0,
2747
+ 0.0,
2748
+ 0.0,
2749
+ 0.0,
2750
+ 0.0,
2751
+ 0.0,
2752
+ 0.0
2753
+ ],
2754
+ "q99": [
2755
+ 0.0,
2756
+ 0.0,
2757
+ 0.0,
2758
+ 0.0,
2759
+ 0.0,
2760
+ 0.0,
2761
+ 0.0
2762
+ ],
2763
+ "std": [
2764
+ 0.0,
2765
+ 0.0,
2766
+ 0.0,
2767
+ 0.0,
2768
+ 0.0,
2769
+ 0.0,
2770
+ 0.0
2771
+ ]
2772
+ }
2773
+ },
2774
+ "ucsd_kitchen_dataset_converted_externally_to_rlds": {
2775
+ "action": {
2776
+ "mask": [
2777
+ true,
2778
+ true,
2779
+ true,
2780
+ true,
2781
+ true,
2782
+ true,
2783
+ false
2784
+ ],
2785
+ "max": [
2786
+ 678.0,
2787
+ 400.0,
2788
+ 507.0,
2789
+ 180.00001525878906,
2790
+ 6.000013828277588,
2791
+ 116.99998474121094,
2792
+ 1.0
2793
+ ],
2794
+ "mean": [
2795
+ 410.37567138671875,
2796
+ 116.9518814086914,
2797
+ 192.35032653808594,
2798
+ -121.22441864013672,
2799
+ -33.84893035888672,
2800
+ 50.016136169433594,
2801
+ 0.741813600063324
2802
+ ],
2803
+ "min": [
2804
+ 172.0,
2805
+ -166.0,
2806
+ -99.99999237060547,
2807
+ -180.00001525878906,
2808
+ -89.0,
2809
+ -96.00010681152344,
2810
+ 0.0
2811
+ ],
2812
+ "q01": [
2813
+ 200.00001052856445,
2814
+ -102.31004211425781,
2815
+ -94.99993370056153,
2816
+ -180.00001525878906,
2817
+ -88.00001525878906,
2818
+ -38.999977111816406,
2819
+ 0.0
2820
+ ],
2821
+ "q99": [
2822
+ 637.0,
2823
+ 368.30999999999995,
2824
+ 493.0,
2825
+ 180.00001525878906,
2826
+ 0.999983012676239,
2827
+ 105.00001525878906,
2828
+ 1.0
2829
+ ],
2830
+ "std": [
2831
+ 122.81494903564453,
2832
+ 108.8009033203125,
2833
+ 130.303466796875,
2834
+ 116.28205108642578,
2835
+ 27.621843338012695,
2836
+ 41.02094650268555,
2837
+ 0.43763357400894165
2838
+ ]
2839
+ },
2840
+ "num_trajectories": 150,
2841
+ "num_transitions": 3970,
2842
+ "proprio": {
2843
+ "max": [
2844
+ 0.0,
2845
+ 0.0,
2846
+ 0.0,
2847
+ 0.0,
2848
+ 0.0,
2849
+ 0.0,
2850
+ 0.0
2851
+ ],
2852
+ "mean": [
2853
+ 0.0,
2854
+ 0.0,
2855
+ 0.0,
2856
+ 0.0,
2857
+ 0.0,
2858
+ 0.0,
2859
+ 0.0
2860
+ ],
2861
+ "min": [
2862
+ 0.0,
2863
+ 0.0,
2864
+ 0.0,
2865
+ 0.0,
2866
+ 0.0,
2867
+ 0.0,
2868
+ 0.0
2869
+ ],
2870
+ "q01": [
2871
+ 0.0,
2872
+ 0.0,
2873
+ 0.0,
2874
+ 0.0,
2875
+ 0.0,
2876
+ 0.0,
2877
+ 0.0
2878
+ ],
2879
+ "q99": [
2880
+ 0.0,
2881
+ 0.0,
2882
+ 0.0,
2883
+ 0.0,
2884
+ 0.0,
2885
+ 0.0,
2886
+ 0.0
2887
+ ],
2888
+ "std": [
2889
+ 0.0,
2890
+ 0.0,
2891
+ 0.0,
2892
+ 0.0,
2893
+ 0.0,
2894
+ 0.0,
2895
+ 0.0
2896
+ ]
2897
+ }
2898
+ },
2899
+ "utaustin_mutex": {
2900
+ "action": {
2901
+ "mask": [
2902
+ true,
2903
+ true,
2904
+ true,
2905
+ true,
2906
+ true,
2907
+ true,
2908
+ false
2909
+ ],
2910
+ "max": [
2911
+ 1.0,
2912
+ 1.0,
2913
+ 1.0,
2914
+ 0.375,
2915
+ 0.375,
2916
+ 0.375,
2917
+ 1.0
2918
+ ],
2919
+ "mean": [
2920
+ 0.06176406890153885,
2921
+ -0.005005486309528351,
2922
+ 0.10216785222291946,
2923
+ -0.03314131125807762,
2924
+ 0.013895004987716675,
2925
+ -0.011317633092403412,
2926
+ 0.5038976669311523
2927
+ ],
2928
+ "min": [
2929
+ -1.0,
2930
+ -1.0,
2931
+ -1.0,
2932
+ -0.375,
2933
+ -0.375,
2934
+ -0.375,
2935
+ 0.0
2936
+ ],
2937
+ "q01": [
2938
+ -0.4285714328289032,
2939
+ -0.9800000190734863,
2940
+ -0.5571428537368774,
2941
+ -0.375,
2942
+ -0.15642857551574707,
2943
+ -0.335357129573822,
2944
+ 0.0
2945
+ ],
2946
+ "q99": [
2947
+ 0.5914285778999329,
2948
+ 0.9714285731315613,
2949
+ 1.0,
2950
+ 0.3278571367263794,
2951
+ 0.207857146859169,
2952
+ 0.25607141852378845,
2953
+ 1.0
2954
+ ],
2955
+ "std": [
2956
+ 0.1875014752149582,
2957
+ 0.4468473494052887,
2958
+ 0.3792876601219177,
2959
+ 0.14097853004932404,
2960
+ 0.06453701853752136,
2961
+ 0.11765272170305252,
2962
+ 0.501045286655426
2963
+ ]
2964
+ },
2965
+ "num_trajectories": 1500,
2966
+ "num_transitions": 361883,
2967
+ "proprio": {
2968
+ "max": [
2969
+ 0.0,
2970
+ 0.0,
2971
+ 0.0,
2972
+ 0.0,
2973
+ 0.0,
2974
+ 0.0,
2975
+ 0.0
2976
+ ],
2977
+ "mean": [
2978
+ 0.0,
2979
+ 0.0,
2980
+ 0.0,
2981
+ 0.0,
2982
+ 0.0,
2983
+ 0.0,
2984
+ 0.0
2985
+ ],
2986
+ "min": [
2987
+ 0.0,
2988
+ 0.0,
2989
+ 0.0,
2990
+ 0.0,
2991
+ 0.0,
2992
+ 0.0,
2993
+ 0.0
2994
+ ],
2995
+ "q01": [
2996
+ 0.0,
2997
+ 0.0,
2998
+ 0.0,
2999
+ 0.0,
3000
+ 0.0,
3001
+ 0.0,
3002
+ 0.0
3003
+ ],
3004
+ "q99": [
3005
+ 0.0,
3006
+ 0.0,
3007
+ 0.0,
3008
+ 0.0,
3009
+ 0.0,
3010
+ 0.0,
3011
+ 0.0
3012
+ ],
3013
+ "std": [
3014
+ 0.0,
3015
+ 0.0,
3016
+ 0.0,
3017
+ 0.0,
3018
+ 0.0,
3019
+ 0.0,
3020
+ 0.0
3021
+ ]
3022
+ }
3023
+ },
3024
+ "viola": {
3025
+ "action": {
3026
+ "mask": [
3027
+ true,
3028
+ true,
3029
+ true,
3030
+ true,
3031
+ true,
3032
+ true,
3033
+ false
3034
+ ],
3035
+ "max": [
3036
+ 1.0,
3037
+ 1.0,
3038
+ 1.0,
3039
+ 0.375,
3040
+ 0.36321428418159485,
3041
+ 0.375,
3042
+ 1.0
3043
+ ],
3044
+ "mean": [
3045
+ 0.04761844128370285,
3046
+ -0.029204415157437325,
3047
+ 0.05586736649274826,
3048
+ -0.002618510741740465,
3049
+ 0.006867344491183758,
3050
+ -0.01682133786380291,
3051
+ 0.7323777675628662
3052
+ ],
3053
+ "min": [
3054
+ -1.0,
3055
+ -1.0,
3056
+ -1.0,
3057
+ -0.375,
3058
+ -0.375,
3059
+ -0.375,
3060
+ 0.0
3061
+ ],
3062
+ "q01": [
3063
+ -0.9628571271896362,
3064
+ -1.0,
3065
+ -1.0,
3066
+ -0.26249998807907104,
3067
+ -0.21321429312229156,
3068
+ -0.3385714292526245,
3069
+ 0.0
3070
+ ],
3071
+ "q99": [
3072
+ 0.9114285707473755,
3073
+ 0.868571400642395,
3074
+ 1.0,
3075
+ 0.2817857265472412,
3076
+ 0.2239285707473755,
3077
+ 0.3557142913341522,
3078
+ 1.0
3079
+ ],
3080
+ "std": [
3081
+ 0.39157867431640625,
3082
+ 0.4076525568962097,
3083
+ 0.40077948570251465,
3084
+ 0.10023996233940125,
3085
+ 0.0844319611787796,
3086
+ 0.10375042259693146,
3087
+ 0.44260647892951965
3088
+ ]
3089
+ },
3090
+ "num_trajectories": 150,
3091
+ "num_transitions": 76324,
3092
+ "proprio": {
3093
+ "max": [
3094
+ 0.0,
3095
+ 0.0,
3096
+ 0.0,
3097
+ 0.0,
3098
+ 0.0,
3099
+ 0.0,
3100
+ 0.0
3101
+ ],
3102
+ "mean": [
3103
+ 0.0,
3104
+ 0.0,
3105
+ 0.0,
3106
+ 0.0,
3107
+ 0.0,
3108
+ 0.0,
3109
+ 0.0
3110
+ ],
3111
+ "min": [
3112
+ 0.0,
3113
+ 0.0,
3114
+ 0.0,
3115
+ 0.0,
3116
+ 0.0,
3117
+ 0.0,
3118
+ 0.0
3119
+ ],
3120
+ "q01": [
3121
+ 0.0,
3122
+ 0.0,
3123
+ 0.0,
3124
+ 0.0,
3125
+ 0.0,
3126
+ 0.0,
3127
+ 0.0
3128
+ ],
3129
+ "q99": [
3130
+ 0.0,
3131
+ 0.0,
3132
+ 0.0,
3133
+ 0.0,
3134
+ 0.0,
3135
+ 0.0,
3136
+ 0.0
3137
+ ],
3138
+ "std": [
3139
+ 0.0,
3140
+ 0.0,
3141
+ 0.0,
3142
+ 0.0,
3143
+ 0.0,
3144
+ 0.0,
3145
+ 0.0
3146
+ ]
3147
+ }
3148
+ }
3149
+ },
3150
+ "output_projector_states": false,
3151
+ "pad_to_multiple_of": 64,
3152
+ "pad_token_id": 32000,
3153
+ "text_config": {
3154
+ "model_type": "llama",
3155
+ "pad_token_id": 32000,
3156
+ "torch_dtype": "bfloat16",
3157
+ "vocab_size": 32064
3158
+ },
3159
+ "timm_model_ids": [
3160
+ "vit_large_patch14_reg4_dinov2.lvd142m",
3161
+ "vit_so400m_patch14_siglip_224"
3162
+ ],
3163
+ "timm_override_act_layers": [
3164
+ null,
3165
+ null
3166
+ ],
3167
+ "torch_dtype": "bfloat16",
3168
+ "transformers_version": "4.40.1",
3169
+ "use_d2f": true,
3170
+ "use_discrete_diffusion": false,
3171
+ "use_fused_vision_backbone": true,
3172
+ "use_mask_token": true,
3173
+ "vision_backbone_id": "dinosiglip-vit-so-224px"
3174
+ }
models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/configuration_prismatic.py ADDED
@@ -0,0 +1,171 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """
2
+ configuration_prismatic.py
3
+
4
+ HuggingFace-style configuration definition for Prismatic VLMs, inheriting from `transformers.PretrainedConfig`.
5
+ Default configuration specifies `siglip-224px+7b`.
6
+ """
7
+
8
+ from typing import Any, Dict, List, Optional
9
+
10
+ from transformers import PretrainedConfig
11
+ from transformers.models.auto import CONFIG_MAPPING
12
+
13
+ # === Utilities for Mapping Prismatic names to HF names ===
14
+ # fmt: off
15
+ VISION_BACKBONE_TO_RESOLUTION: Dict[str, List[int]] = {
16
+ "clip-vit-l": [224], "siglip-vit-so400m": [224], "dinov2-vit-l": [224], "in1k-vit-l": [224],
17
+
18
+ "clip-vit-l-336px": [336],
19
+ "siglip-vit-so400m-384px": [384],
20
+
21
+ "dinoclip-vit-l-336px": [336, 336],
22
+ "dinosiglip-vit-so-224px": [224, 224],
23
+ "dinosiglip-vit-so-384px": [384, 384],
24
+ }
25
+ VISION_BACKBONE_TO_TIMM_ID: Dict[str, List[str]] = {
26
+ "clip-vit-l": ["vit_large_patch14_clip_224.openai"],
27
+ "clip-vit-l-336px": ["vit_large_patch14_clip_336.openai"],
28
+
29
+ "dinov2-vit-l": ["vit_large_patch14_reg4_dinov2.lvd142m"],
30
+ "in1k-vit-l": ["vit_large_patch16_224.augreg_in21k_ft_in1k"],
31
+
32
+ "siglip-vit-so400m": ["vit_so400m_patch14_siglip_224"],
33
+ "siglip-vit-so400m-384px": ["vit_so400m_patch14_siglip_384"],
34
+
35
+ "dinoclip-vit-l-336px": ["vit_large_patch14_reg4_dinov2.lvd142m", "vit_large_patch14_clip_336.openai"],
36
+ "dinosiglip-vit-so-224px": ["vit_large_patch14_reg4_dinov2.lvd142m", "vit_so400m_patch14_siglip_224"],
37
+ "dinosiglip-vit-so-384px": ["vit_large_patch14_reg4_dinov2.lvd142m", "vit_so400m_patch14_siglip_384"],
38
+ }
39
+ TIMM_OVERRIDE_ACT_LAYER: Dict[str, List[Optional[str]]] = {
40
+ "clip-vit-l": ["quick_gelu"], "clip-vit-l-336px": ["quick_gelu"],
41
+ "dinov2-vit-l": [None], "in1k-vit-l": [None],
42
+ "siglip-vit-so400m": [None], "siglip-vit-so400m-384px": [None],
43
+ "dinoclip-vit-l-336px": [None, "quick_gelu"],
44
+ "dinosiglip-vit-so-224px": [None, None], "dinosiglip-vit-so-384px": [None, None]
45
+ }
46
+
47
+ LLM_BACKBONE_TO_HF_PATH = {
48
+ "llama2-7b-pure": "meta-llama/Llama-2-7b-hf", "llama2-13b-pure": "meta-llama/Llama-2-13b-hf",
49
+ "llama2-7b-chat": "meta-llama/Llama-2-7b-chat-hf", "llama2-13b-chat": "meta-llama/Llama-2-13b-chat-hf",
50
+
51
+ "vicuna-v15-7b": "lmsys/vicuna-7b-v1.5", "vicuna-v15-13b": "lmsys/vicuna-13b-v1.5",
52
+
53
+ "mistral-v0.1-7b-pure": "mistralai/Mistral-7B-v0.1",
54
+ "mistral-v0.1-7b-instruct": "mistralai/Mistral-7B-Instruct-v0.1",
55
+
56
+ "phi-2-3b": "microsoft/phi-2",
57
+ }
58
+ LLM_BACKBONE_TO_HF_METACLASS = {
59
+ "llama2-7b-pure": "llama", "llama2-13b-pure": "llama", "llama2-7b-chat": "llama", "llama2-13b-chat": "llama",
60
+ "vicuna-v15-7b": "llama", "vicuna-v15-13b": "llama",
61
+
62
+ "mistral-v0.1-7b-pure": "mistral", "mistral-v0.1-7b-instruct": "mistral",
63
+
64
+ "phi-2-3b": "phi",
65
+ }
66
+
67
+ VALID_VISION_BACKBONES = set(VISION_BACKBONE_TO_RESOLUTION.keys())
68
+ VALID_LLM_BACKBONES = set(LLM_BACKBONE_TO_HF_PATH)
69
+ # fmt: on
70
+
71
+
72
+ class PrismaticConfig(PretrainedConfig):
73
+ model_type: str = "prismatic"
74
+ is_composition: bool = False
75
+
76
+ def __init__(
77
+ self,
78
+ vision_backbone_id: str = "siglip-vit-so400m",
79
+ llm_backbone_id: str = "vicuna-v15-7b",
80
+ arch_specifier: str = "no-align+gelu-mlp",
81
+ use_fused_vision_backbone: Optional[bool] = None,
82
+ image_resize_strategy: str = "letterbox",
83
+ text_config: Optional[Dict[str, Any]] = None,
84
+ llm_max_length: int = 2048,
85
+ pad_token_id: int = 32000,
86
+ mask_token_id: int = 32001,
87
+ use_mask_token: bool = False,
88
+ use_discrete_diffusion: bool = False,
89
+ use_d2f: bool = False,
90
+ d2f_block_size: int = 7,
91
+ pad_to_multiple_of: int = 64,
92
+ output_projector_states: bool = False,
93
+ **kwargs: str,
94
+ ) -> None:
95
+ if vision_backbone_id not in VALID_VISION_BACKBONES:
96
+ raise ValueError(f"Vision backbone `{vision_backbone_id}` not in {VALID_VISION_BACKBONES = }")
97
+
98
+ if llm_backbone_id not in VALID_LLM_BACKBONES:
99
+ raise ValueError(f"LLM backbone `{llm_backbone_id}` not in {VALID_LLM_BACKBONES = }")
100
+
101
+ # Set Prismatic Configuration Fields
102
+ self.vision_backbone_id = vision_backbone_id
103
+ self.llm_backbone_id = llm_backbone_id
104
+ self.arch_specifier = arch_specifier
105
+ self.output_projector_states = output_projector_states
106
+
107
+ # [Contract] All vision backbone parameters are lists =>> supports fused backbones with different preprocessing
108
+ self.use_fused_vision_backbone = (
109
+ use_fused_vision_backbone
110
+ if use_fused_vision_backbone is not None
111
+ else any(self.vision_backbone_id.startswith(v) for v in ["dinoclip", "dinosiglip"])
112
+ )
113
+
114
+ self.timm_model_ids = VISION_BACKBONE_TO_TIMM_ID[self.vision_backbone_id]
115
+ self.timm_override_act_layers = TIMM_OVERRIDE_ACT_LAYER[self.vision_backbone_id]
116
+ self.image_sizes = VISION_BACKBONE_TO_RESOLUTION[self.vision_backbone_id]
117
+ self.image_resize_strategy = image_resize_strategy
118
+
119
+ self.hf_llm_id = LLM_BACKBONE_TO_HF_PATH[self.llm_backbone_id]
120
+ self.llm_max_length = llm_max_length
121
+ self.pad_token_id, self.pad_to_multiple_of = pad_token_id, pad_to_multiple_of
122
+ self.mask_token_id, self.use_mask_token = mask_token_id, use_mask_token
123
+
124
+ self.use_discrete_diffusion = use_discrete_diffusion
125
+ self.use_d2f = use_d2f
126
+ self.d2f_block_size = d2f_block_size
127
+
128
+ # [IMPORTANT] HF Utilities actually look for a `text_config` field... we need to use that specific naming!
129
+ self.text_config = (
130
+ CONFIG_MAPPING[LLM_BACKBONE_TO_HF_METACLASS[self.llm_backbone_id]](**text_config)
131
+ if text_config is not None
132
+ else CONFIG_MAPPING[LLM_BACKBONE_TO_HF_METACLASS[self.llm_backbone_id]]()
133
+ )
134
+
135
+ # Dispatch **kwargs to super() =>> note that `pad_token_id` collides, so we pass it in here as well...
136
+ super().__init__(pad_token_id=pad_token_id, **kwargs)
137
+
138
+ def set_mask_token_id(self, mask_token_id: int) -> None:
139
+ """Set the mask token ID."""
140
+ self.mask_token_id = mask_token_id
141
+ self.use_mask_token = True
142
+
143
+ def set_vocab_size(self, vocab_size: int) -> None:
144
+ """Set the vocabulary size."""
145
+ # 自行向上取整到64的整数倍，本来就是32064因此无需调整
146
+ self.text_config.vocab_size = vocab_size
147
+
148
+ def set_dicrete_diffusion(self, use_discrete_diffusion: bool=True) -> None:
149
+ """Set whether to use discrete diffusion."""
150
+ self.use_discrete_diffusion = use_discrete_diffusion
151
+
152
+ def set_d2f(self, use_d2f: bool = True, block_size: int = 7) -> None:
153
+ """Persist the D2F decoding mode and its training block size."""
154
+ if block_size <= 0:
155
+ raise ValueError(f"D2F block_size must be positive, got {block_size}")
156
+ self.use_d2f = use_d2f
157
+ self.d2f_block_size = block_size
158
+
159
+
160
+ class OpenVLAConfig(PrismaticConfig):
161
+ model_type: str = "openvla"
162
+
163
+ def __init__(
164
+ self,
165
+ norm_stats: Optional[Dict[str, Dict[str, Dict[str, Dict[str, List[float]]]]]] = None,
166
+ n_action_bins: int = 256,
167
+ **kwargs: str,
168
+ ) -> None:
169
+ self.norm_stats, self.n_action_bins = norm_stats, n_action_bins
170
+
171
+ super().__init__(**kwargs)
models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/dataset_statistics.json ADDED
@@ -0,0 +1,136 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "20260818_piper_pick_white_block_50hz": {
3
+ "action": {
4
+ "mask": [
5
+ true,
6
+ true,
7
+ true,
8
+ true,
9
+ true,
10
+ true,
11
+ true
12
+ ],
13
+ "max": [
14
+ 1.7367422580718994,
15
+ 2.170072555541992,
16
+ -0.0010122909443452954,
17
+ 0.24776694178581238,
18
+ 1.2270362377166748,
19
+ 0.884288489818573,
20
+ 0.8820000290870667
21
+ ],
22
+ "mean": [
23
+ 0.3145395386472929,
24
+ 1.3267379418051677,
25
+ -0.9685837817487981,
26
+ -0.05186474901303632,
27
+ 0.781981652890849,
28
+ 0.2559612958686087,
29
+ 0.2737080916955827
30
+ ],
31
+ "min": [
32
+ -0.4645542800426483,
33
+ -0.00040142572834156454,
34
+ -1.8836116790771484,
35
+ -0.42645373940467834,
36
+ 0.0,
37
+ -0.3565009534358978,
38
+ 0.004999999888241291
39
+ ],
40
+ "q01": [
41
+ -0.4103956630825996,
42
+ 0.0018325956771150231,
43
+ -1.5609875917434692,
44
+ -0.34400439262390137,
45
+ 0.041399210691452026,
46
+ -0.22659085854887961,
47
+ 0.004999999888241291
48
+ ],
49
+ "q99": [
50
+ 1.6672606468200684,
51
+ 2.106769561767578,
52
+ -0.004660028964281082,
53
+ 0.1808684766292572,
54
+ 1.2071774208545687,
55
+ 0.7780997276306171,
56
+ 0.8410000205039978
57
+ ],
58
+ "std": [
59
+ 0.5678397472855142,
60
+ 0.5949945624217958,
61
+ 0.34923064316522107,
62
+ 0.11515592923459911,
63
+ 0.26210717289996627,
64
+ 0.192443378315841,
65
+ 0.21295665086657772
66
+ ]
67
+ },
68
+ "num_trajectories": 52,
69
+ "num_transitions": 34254,
70
+ "proprio": {
71
+ "mask": [
72
+ true,
73
+ true,
74
+ true,
75
+ true,
76
+ true,
77
+ true,
78
+ true
79
+ ],
80
+ "max": [
81
+ 1.7367422580718994,
82
+ 2.170072555541992,
83
+ -0.0010122909443452954,
84
+ 0.24776694178581238,
85
+ 1.2270362377166748,
86
+ 0.884288489818573,
87
+ 0.8820000290870667
88
+ ],
89
+ "mean": [
90
+ 0.3145395386472929,
91
+ 1.3267379418051677,
92
+ -0.9685837817487981,
93
+ -0.05186474901303632,
94
+ 0.781981652890849,
95
+ 0.2559612958686087,
96
+ 0.2737080916955827
97
+ ],
98
+ "min": [
99
+ -0.4645542800426483,
100
+ -0.00040142572834156454,
101
+ -1.8836116790771484,
102
+ -0.42645373940467834,
103
+ 0.0,
104
+ -0.3565009534358978,
105
+ 0.004999999888241291
106
+ ],
107
+ "q01": [
108
+ -0.4103956630825996,
109
+ 0.0018325956771150231,
110
+ -1.5609875917434692,
111
+ -0.34400439262390137,
112
+ 0.041399210691452026,
113
+ -0.22659085854887961,
114
+ 0.004999999888241291
115
+ ],
116
+ "q99": [
117
+ 1.6672606468200684,
118
+ 2.106769561767578,
119
+ -0.004660028964281082,
120
+ 0.1808684766292572,
121
+ 1.2071774208545687,
122
+ 0.7780997276306171,
123
+ 0.8410000205039978
124
+ ],
125
+ "std": [
126
+ 0.5678397472855142,
127
+ 0.5949945624217958,
128
+ 0.34923064316522107,
129
+ 0.11515592923459911,
130
+ 0.26210717289996627,
131
+ 0.192443378315841,
132
+ 0.21295665086657772
133
+ ]
134
+ }
135
+ }
136
+ }
models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": 2,
5
+ "pad_token_id": 32000,
6
+ "transformers_version": "4.40.1"
7
+ }
models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/lora_adapter/README.md ADDED
@@ -0,0 +1,204 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ base_model: /scratch/wangpc/dRTCv2/artifacts/models/openvla-7b
4
+ ---
5
+
6
+ # Model Card for Model ID
7
+
8
+ <!-- Provide a quick summary of what the model is/does. -->
9
+
10
+
11
+
12
+ ## Model Details
13
+
14
+ ### Model Description
15
+
16
+ <!-- Provide a longer summary of what this model is. -->
17
+
18
+
19
+
20
+ - **Developed by:** [More Information Needed]
21
+ - **Funded by [optional]:** [More Information Needed]
22
+ - **Shared by [optional]:** [More Information Needed]
23
+ - **Model type:** [More Information Needed]
24
+ - **Language(s) (NLP):** [More Information Needed]
25
+ - **License:** [More Information Needed]
26
+ - **Finetuned from model [optional]:** [More Information Needed]
27
+
28
+ ### Model Sources [optional]
29
+
30
+ <!-- Provide the basic links for the model. -->
31
+
32
+ - **Repository:** [More Information Needed]
33
+ - **Paper [optional]:** [More Information Needed]
34
+ - **Demo [optional]:** [More Information Needed]
35
+
36
+ ## Uses
37
+
38
+ <!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
39
+
40
+ ### Direct Use
41
+
42
+ <!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
43
+
44
+ [More Information Needed]
45
+
46
+ ### Downstream Use [optional]
47
+
48
+ <!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
49
+
50
+ [More Information Needed]
51
+
52
+ ### Out-of-Scope Use
53
+
54
+ <!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
55
+
56
+ [More Information Needed]
57
+
58
+ ## Bias, Risks, and Limitations
59
+
60
+ <!-- This section is meant to convey both technical and sociotechnical limitations. -->
61
+
62
+ [More Information Needed]
63
+
64
+ ### Recommendations
65
+
66
+ <!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
67
+
68
+ Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
69
+
70
+ ## How to Get Started with the Model
71
+
72
+ Use the code below to get started with the model.
73
+
74
+ [More Information Needed]
75
+
76
+ ## Training Details
77
+
78
+ ### Training Data
79
+
80
+ <!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
81
+
82
+ [More Information Needed]
83
+
84
+ ### Training Procedure
85
+
86
+ <!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
87
+
88
+ #### Preprocessing [optional]
89
+
90
+ [More Information Needed]
91
+
92
+
93
+ #### Training Hyperparameters
94
+
95
+ - **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
96
+
97
+ #### Speeds, Sizes, Times [optional]
98
+
99
+ <!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
100
+
101
+ [More Information Needed]
102
+
103
+ ## Evaluation
104
+
105
+ <!-- This section describes the evaluation protocols and provides the results. -->
106
+
107
+ ### Testing Data, Factors & Metrics
108
+
109
+ #### Testing Data
110
+
111
+ <!-- This should link to a Dataset Card if possible. -->
112
+
113
+ [More Information Needed]
114
+
115
+ #### Factors
116
+
117
+ <!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
118
+
119
+ [More Information Needed]
120
+
121
+ #### Metrics
122
+
123
+ <!-- These are the evaluation metrics being used, ideally with a description of why. -->
124
+
125
+ [More Information Needed]
126
+
127
+ ### Results
128
+
129
+ [More Information Needed]
130
+
131
+ #### Summary
132
+
133
+
134
+
135
+ ## Model Examination [optional]
136
+
137
+ <!-- Relevant interpretability work for the model goes here -->
138
+
139
+ [More Information Needed]
140
+
141
+ ## Environmental Impact
142
+
143
+ <!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
144
+
145
+ Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
146
+
147
+ - **Hardware Type:** [More Information Needed]
148
+ - **Hours used:** [More Information Needed]
149
+ - **Cloud Provider:** [More Information Needed]
150
+ - **Compute Region:** [More Information Needed]
151
+ - **Carbon Emitted:** [More Information Needed]
152
+
153
+ ## Technical Specifications [optional]
154
+
155
+ ### Model Architecture and Objective
156
+
157
+ [More Information Needed]
158
+
159
+ ### Compute Infrastructure
160
+
161
+ [More Information Needed]
162
+
163
+ #### Hardware
164
+
165
+ [More Information Needed]
166
+
167
+ #### Software
168
+
169
+ [More Information Needed]
170
+
171
+ ## Citation [optional]
172
+
173
+ <!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
174
+
175
+ **BibTeX:**
176
+
177
+ [More Information Needed]
178
+
179
+ **APA:**
180
+
181
+ [More Information Needed]
182
+
183
+ ## Glossary [optional]
184
+
185
+ <!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
186
+
187
+ [More Information Needed]
188
+
189
+ ## More Information [optional]
190
+
191
+ [More Information Needed]
192
+
193
+ ## Model Card Authors [optional]
194
+
195
+ [More Information Needed]
196
+
197
+ ## Model Card Contact
198
+
199
+ [More Information Needed]
200
+
201
+
202
+ ### Framework versions
203
+
204
+ - PEFT 0.11.1
models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/lora_adapter/adapter_config.json ADDED
@@ -0,0 +1,45 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": {
4
+ "base_model_class": "OpenVLAForActionPrediction",
5
+ "parent_library": "transformers_modules.openvla-7b.modeling_prismatic"
6
+ },
7
+ "base_model_name_or_path": "/scratch/wangpc/dRTCv2/artifacts/models/openvla-7b",
8
+ "bias": "none",
9
+ "fan_in_fan_out": false,
10
+ "inference_mode": true,
11
+ "init_lora_weights": "gaussian",
12
+ "layer_replication": null,
13
+ "layers_pattern": null,
14
+ "layers_to_transform": null,
15
+ "loftq_config": {},
16
+ "lora_alpha": 16,
17
+ "lora_dropout": 0.0,
18
+ "megatron_config": null,
19
+ "megatron_core": "megatron.core",
20
+ "modules_to_save": null,
21
+ "peft_type": "LORA",
22
+ "r": 32,
23
+ "rank_pattern": {},
24
+ "revision": null,
25
+ "target_modules": [
26
+ "q",
27
+ "q_proj",
28
+ "fc2",
29
+ "up_proj",
30
+ "v_proj",
31
+ "o_proj",
32
+ "kv",
33
+ "fc3",
34
+ "qkv",
35
+ "k_proj",
36
+ "gate_proj",
37
+ "lm_head",
38
+ "proj",
39
+ "fc1",
40
+ "down_proj"
41
+ ],
42
+ "task_type": null,
43
+ "use_dora": false,
44
+ "use_rslora": false
45
+ }
models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/lora_adapter/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c1e9db6323620de23bfd525de76660ace5062b0bb59c46c1f7326de1fc0e2022
3
+ size 484458600
models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/model-00001-of-00004.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4bffa374af9d562734fba1711c1ce8113af062ae77711fc48e18825b9e172f6e
3
+ size 4925122448
models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/model-00002-of-00004.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eac3d2366f31bf9fcf4a0790a79c12c128974ef357019212e620605c5670c73b
3
+ size 4947392496
models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/model-00003-of-00004.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cd1513b940654955696597935be9a03ae77cbe9674a78c473b98c9428984c8e6
3
+ size 4947417456
models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/model-00004-of-00004.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d3f1399b9a5effee5c83cc6de9b174a7501164961e2b21dce47bebe57402e901
3
+ size 262668432
models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/model.safetensors.index.json ADDED
@@ -0,0 +1,989 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_size": 15082474368
4
+ },
5
+ "weight_map": {
6
+ "language_model.lm_head.weight": "model-00004-of-00004.safetensors",
7
+ "language_model.model.embed_tokens.weight": "model-00001-of-00004.safetensors",
8
+ "language_model.model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
9
+ "language_model.model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
10
+ "language_model.model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
11
+ "language_model.model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
12
+ "language_model.model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
13
+ "language_model.model.layers.0.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
14
+ "language_model.model.layers.0.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
15
+ "language_model.model.layers.0.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
16
+ "language_model.model.layers.0.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
17
+ "language_model.model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
18
+ "language_model.model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
19
+ "language_model.model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
20
+ "language_model.model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
21
+ "language_model.model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
22
+ "language_model.model.layers.1.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
23
+ "language_model.model.layers.1.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
24
+ "language_model.model.layers.1.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
25
+ "language_model.model.layers.1.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
26
+ "language_model.model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
27
+ "language_model.model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
28
+ "language_model.model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
29
+ "language_model.model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
30
+ "language_model.model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
31
+ "language_model.model.layers.10.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
32
+ "language_model.model.layers.10.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
33
+ "language_model.model.layers.10.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
34
+ "language_model.model.layers.10.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
35
+ "language_model.model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
36
+ "language_model.model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
37
+ "language_model.model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
38
+ "language_model.model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
39
+ "language_model.model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
40
+ "language_model.model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
41
+ "language_model.model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
42
+ "language_model.model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
43
+ "language_model.model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
44
+ "language_model.model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
45
+ "language_model.model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
46
+ "language_model.model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
47
+ "language_model.model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
48
+ "language_model.model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
49
+ "language_model.model.layers.12.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
50
+ "language_model.model.layers.12.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
51
+ "language_model.model.layers.12.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
52
+ "language_model.model.layers.12.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
53
+ "language_model.model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
54
+ "language_model.model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
55
+ "language_model.model.layers.13.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
56
+ "language_model.model.layers.13.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
57
+ "language_model.model.layers.13.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
58
+ "language_model.model.layers.13.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
59
+ "language_model.model.layers.13.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
60
+ "language_model.model.layers.13.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
61
+ "language_model.model.layers.13.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
62
+ "language_model.model.layers.14.input_layernorm.weight": "model-00002-of-00004.safetensors",
63
+ "language_model.model.layers.14.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
64
+ "language_model.model.layers.14.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
65
+ "language_model.model.layers.14.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
66
+ "language_model.model.layers.14.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
67
+ "language_model.model.layers.14.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
68
+ "language_model.model.layers.14.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
69
+ "language_model.model.layers.14.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
70
+ "language_model.model.layers.14.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
71
+ "language_model.model.layers.15.input_layernorm.weight": "model-00002-of-00004.safetensors",
72
+ "language_model.model.layers.15.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
73
+ "language_model.model.layers.15.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
74
+ "language_model.model.layers.15.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
75
+ "language_model.model.layers.15.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
76
+ "language_model.model.layers.15.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
77
+ "language_model.model.layers.15.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
78
+ "language_model.model.layers.15.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
79
+ "language_model.model.layers.15.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
80
+ "language_model.model.layers.16.input_layernorm.weight": "model-00002-of-00004.safetensors",
81
+ "language_model.model.layers.16.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
82
+ "language_model.model.layers.16.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
83
+ "language_model.model.layers.16.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
84
+ "language_model.model.layers.16.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
85
+ "language_model.model.layers.16.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
86
+ "language_model.model.layers.16.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
87
+ "language_model.model.layers.16.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
88
+ "language_model.model.layers.16.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
89
+ "language_model.model.layers.17.input_layernorm.weight": "model-00002-of-00004.safetensors",
90
+ "language_model.model.layers.17.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
91
+ "language_model.model.layers.17.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
92
+ "language_model.model.layers.17.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
93
+ "language_model.model.layers.17.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
94
+ "language_model.model.layers.17.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
95
+ "language_model.model.layers.17.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
96
+ "language_model.model.layers.17.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
97
+ "language_model.model.layers.17.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
98
+ "language_model.model.layers.18.input_layernorm.weight": "model-00002-of-00004.safetensors",
99
+ "language_model.model.layers.18.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
100
+ "language_model.model.layers.18.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
101
+ "language_model.model.layers.18.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
102
+ "language_model.model.layers.18.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
103
+ "language_model.model.layers.18.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
104
+ "language_model.model.layers.18.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
105
+ "language_model.model.layers.18.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
106
+ "language_model.model.layers.18.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
107
+ "language_model.model.layers.19.input_layernorm.weight": "model-00003-of-00004.safetensors",
108
+ "language_model.model.layers.19.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
109
+ "language_model.model.layers.19.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
110
+ "language_model.model.layers.19.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
111
+ "language_model.model.layers.19.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
112
+ "language_model.model.layers.19.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
113
+ "language_model.model.layers.19.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
114
+ "language_model.model.layers.19.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
115
+ "language_model.model.layers.19.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
116
+ "language_model.model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
117
+ "language_model.model.layers.2.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
118
+ "language_model.model.layers.2.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
119
+ "language_model.model.layers.2.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
120
+ "language_model.model.layers.2.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
121
+ "language_model.model.layers.2.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
122
+ "language_model.model.layers.2.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
123
+ "language_model.model.layers.2.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
124
+ "language_model.model.layers.2.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
125
+ "language_model.model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
126
+ "language_model.model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
127
+ "language_model.model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
128
+ "language_model.model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
129
+ "language_model.model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
130
+ "language_model.model.layers.20.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
131
+ "language_model.model.layers.20.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
132
+ "language_model.model.layers.20.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
133
+ "language_model.model.layers.20.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
134
+ "language_model.model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
135
+ "language_model.model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
136
+ "language_model.model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
137
+ "language_model.model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
138
+ "language_model.model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
139
+ "language_model.model.layers.21.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
140
+ "language_model.model.layers.21.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
141
+ "language_model.model.layers.21.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
142
+ "language_model.model.layers.21.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
143
+ "language_model.model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
144
+ "language_model.model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
145
+ "language_model.model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
146
+ "language_model.model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
147
+ "language_model.model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
148
+ "language_model.model.layers.22.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
149
+ "language_model.model.layers.22.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
150
+ "language_model.model.layers.22.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
151
+ "language_model.model.layers.22.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
152
+ "language_model.model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
153
+ "language_model.model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
154
+ "language_model.model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
155
+ "language_model.model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
156
+ "language_model.model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
157
+ "language_model.model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
158
+ "language_model.model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
159
+ "language_model.model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
160
+ "language_model.model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
161
+ "language_model.model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
162
+ "language_model.model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
163
+ "language_model.model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
164
+ "language_model.model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
165
+ "language_model.model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
166
+ "language_model.model.layers.24.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
167
+ "language_model.model.layers.24.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
168
+ "language_model.model.layers.24.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
169
+ "language_model.model.layers.24.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
170
+ "language_model.model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
171
+ "language_model.model.layers.25.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
172
+ "language_model.model.layers.25.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
173
+ "language_model.model.layers.25.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
174
+ "language_model.model.layers.25.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
175
+ "language_model.model.layers.25.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
176
+ "language_model.model.layers.25.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
177
+ "language_model.model.layers.25.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
178
+ "language_model.model.layers.25.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
179
+ "language_model.model.layers.26.input_layernorm.weight": "model-00003-of-00004.safetensors",
180
+ "language_model.model.layers.26.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
181
+ "language_model.model.layers.26.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
182
+ "language_model.model.layers.26.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
183
+ "language_model.model.layers.26.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
184
+ "language_model.model.layers.26.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
185
+ "language_model.model.layers.26.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
186
+ "language_model.model.layers.26.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
187
+ "language_model.model.layers.26.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
188
+ "language_model.model.layers.27.input_layernorm.weight": "model-00003-of-00004.safetensors",
189
+ "language_model.model.layers.27.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
190
+ "language_model.model.layers.27.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
191
+ "language_model.model.layers.27.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
192
+ "language_model.model.layers.27.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
193
+ "language_model.model.layers.27.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
194
+ "language_model.model.layers.27.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
195
+ "language_model.model.layers.27.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
196
+ "language_model.model.layers.27.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
197
+ "language_model.model.layers.28.input_layernorm.weight": "model-00003-of-00004.safetensors",
198
+ "language_model.model.layers.28.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
199
+ "language_model.model.layers.28.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
200
+ "language_model.model.layers.28.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
201
+ "language_model.model.layers.28.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
202
+ "language_model.model.layers.28.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
203
+ "language_model.model.layers.28.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
204
+ "language_model.model.layers.28.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
205
+ "language_model.model.layers.28.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
206
+ "language_model.model.layers.29.input_layernorm.weight": "model-00003-of-00004.safetensors",
207
+ "language_model.model.layers.29.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
208
+ "language_model.model.layers.29.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
209
+ "language_model.model.layers.29.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
210
+ "language_model.model.layers.29.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
211
+ "language_model.model.layers.29.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
212
+ "language_model.model.layers.29.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
213
+ "language_model.model.layers.29.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
214
+ "language_model.model.layers.29.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
215
+ "language_model.model.layers.3.input_layernorm.weight": "model-00001-of-00004.safetensors",
216
+ "language_model.model.layers.3.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
217
+ "language_model.model.layers.3.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
218
+ "language_model.model.layers.3.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
219
+ "language_model.model.layers.3.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
220
+ "language_model.model.layers.3.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
221
+ "language_model.model.layers.3.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
222
+ "language_model.model.layers.3.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
223
+ "language_model.model.layers.3.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
224
+ "language_model.model.layers.30.input_layernorm.weight": "model-00003-of-00004.safetensors",
225
+ "language_model.model.layers.30.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
226
+ "language_model.model.layers.30.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
227
+ "language_model.model.layers.30.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
228
+ "language_model.model.layers.30.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
229
+ "language_model.model.layers.30.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
230
+ "language_model.model.layers.30.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
231
+ "language_model.model.layers.30.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
232
+ "language_model.model.layers.30.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
233
+ "language_model.model.layers.31.input_layernorm.weight": "model-00003-of-00004.safetensors",
234
+ "language_model.model.layers.31.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
235
+ "language_model.model.layers.31.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
236
+ "language_model.model.layers.31.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
237
+ "language_model.model.layers.31.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
238
+ "language_model.model.layers.31.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
239
+ "language_model.model.layers.31.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
240
+ "language_model.model.layers.31.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
241
+ "language_model.model.layers.31.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
242
+ "language_model.model.layers.4.input_layernorm.weight": "model-00001-of-00004.safetensors",
243
+ "language_model.model.layers.4.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
244
+ "language_model.model.layers.4.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
245
+ "language_model.model.layers.4.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
246
+ "language_model.model.layers.4.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
247
+ "language_model.model.layers.4.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
248
+ "language_model.model.layers.4.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
249
+ "language_model.model.layers.4.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
250
+ "language_model.model.layers.4.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
251
+ "language_model.model.layers.5.input_layernorm.weight": "model-00001-of-00004.safetensors",
252
+ "language_model.model.layers.5.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
253
+ "language_model.model.layers.5.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
254
+ "language_model.model.layers.5.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
255
+ "language_model.model.layers.5.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
256
+ "language_model.model.layers.5.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
257
+ "language_model.model.layers.5.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
258
+ "language_model.model.layers.5.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
259
+ "language_model.model.layers.5.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
260
+ "language_model.model.layers.6.input_layernorm.weight": "model-00001-of-00004.safetensors",
261
+ "language_model.model.layers.6.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
262
+ "language_model.model.layers.6.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
263
+ "language_model.model.layers.6.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
264
+ "language_model.model.layers.6.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
265
+ "language_model.model.layers.6.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
266
+ "language_model.model.layers.6.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
267
+ "language_model.model.layers.6.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
268
+ "language_model.model.layers.6.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
269
+ "language_model.model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
270
+ "language_model.model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
271
+ "language_model.model.layers.7.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
272
+ "language_model.model.layers.7.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
273
+ "language_model.model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
274
+ "language_model.model.layers.7.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
275
+ "language_model.model.layers.7.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
276
+ "language_model.model.layers.7.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
277
+ "language_model.model.layers.7.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
278
+ "language_model.model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
279
+ "language_model.model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
280
+ "language_model.model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
281
+ "language_model.model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
282
+ "language_model.model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
283
+ "language_model.model.layers.8.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
284
+ "language_model.model.layers.8.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
285
+ "language_model.model.layers.8.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
286
+ "language_model.model.layers.8.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
287
+ "language_model.model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
288
+ "language_model.model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
289
+ "language_model.model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
290
+ "language_model.model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
291
+ "language_model.model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
292
+ "language_model.model.layers.9.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
293
+ "language_model.model.layers.9.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
294
+ "language_model.model.layers.9.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
295
+ "language_model.model.layers.9.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
296
+ "language_model.model.norm.weight": "model-00003-of-00004.safetensors",
297
+ "projector.fc1.bias": "model-00001-of-00004.safetensors",
298
+ "projector.fc1.weight": "model-00001-of-00004.safetensors",
299
+ "projector.fc2.bias": "model-00001-of-00004.safetensors",
300
+ "projector.fc2.weight": "model-00001-of-00004.safetensors",
301
+ "projector.fc3.bias": "model-00001-of-00004.safetensors",
302
+ "projector.fc3.weight": "model-00001-of-00004.safetensors",
303
+ "vision_backbone.featurizer.blocks.0.attn.proj.bias": "model-00001-of-00004.safetensors",
304
+ "vision_backbone.featurizer.blocks.0.attn.proj.weight": "model-00001-of-00004.safetensors",
305
+ "vision_backbone.featurizer.blocks.0.attn.qkv.bias": "model-00001-of-00004.safetensors",
306
+ "vision_backbone.featurizer.blocks.0.attn.qkv.weight": "model-00001-of-00004.safetensors",
307
+ "vision_backbone.featurizer.blocks.0.ls1.scale_factor": "model-00001-of-00004.safetensors",
308
+ "vision_backbone.featurizer.blocks.0.ls2.scale_factor": "model-00001-of-00004.safetensors",
309
+ "vision_backbone.featurizer.blocks.0.mlp.fc1.bias": "model-00001-of-00004.safetensors",
310
+ "vision_backbone.featurizer.blocks.0.mlp.fc1.weight": "model-00001-of-00004.safetensors",
311
+ "vision_backbone.featurizer.blocks.0.mlp.fc2.bias": "model-00001-of-00004.safetensors",
312
+ "vision_backbone.featurizer.blocks.0.mlp.fc2.weight": "model-00001-of-00004.safetensors",
313
+ "vision_backbone.featurizer.blocks.0.norm1.bias": "model-00001-of-00004.safetensors",
314
+ "vision_backbone.featurizer.blocks.0.norm1.weight": "model-00001-of-00004.safetensors",
315
+ "vision_backbone.featurizer.blocks.0.norm2.bias": "model-00001-of-00004.safetensors",
316
+ "vision_backbone.featurizer.blocks.0.norm2.weight": "model-00001-of-00004.safetensors",
317
+ "vision_backbone.featurizer.blocks.1.attn.proj.bias": "model-00001-of-00004.safetensors",
318
+ "vision_backbone.featurizer.blocks.1.attn.proj.weight": "model-00001-of-00004.safetensors",
319
+ "vision_backbone.featurizer.blocks.1.attn.qkv.bias": "model-00001-of-00004.safetensors",
320
+ "vision_backbone.featurizer.blocks.1.attn.qkv.weight": "model-00001-of-00004.safetensors",
321
+ "vision_backbone.featurizer.blocks.1.ls1.scale_factor": "model-00001-of-00004.safetensors",
322
+ "vision_backbone.featurizer.blocks.1.ls2.scale_factor": "model-00001-of-00004.safetensors",
323
+ "vision_backbone.featurizer.blocks.1.mlp.fc1.bias": "model-00001-of-00004.safetensors",
324
+ "vision_backbone.featurizer.blocks.1.mlp.fc1.weight": "model-00001-of-00004.safetensors",
325
+ "vision_backbone.featurizer.blocks.1.mlp.fc2.bias": "model-00001-of-00004.safetensors",
326
+ "vision_backbone.featurizer.blocks.1.mlp.fc2.weight": "model-00001-of-00004.safetensors",
327
+ "vision_backbone.featurizer.blocks.1.norm1.bias": "model-00001-of-00004.safetensors",
328
+ "vision_backbone.featurizer.blocks.1.norm1.weight": "model-00001-of-00004.safetensors",
329
+ "vision_backbone.featurizer.blocks.1.norm2.bias": "model-00001-of-00004.safetensors",
330
+ "vision_backbone.featurizer.blocks.1.norm2.weight": "model-00001-of-00004.safetensors",
331
+ "vision_backbone.featurizer.blocks.10.attn.proj.bias": "model-00001-of-00004.safetensors",
332
+ "vision_backbone.featurizer.blocks.10.attn.proj.weight": "model-00001-of-00004.safetensors",
333
+ "vision_backbone.featurizer.blocks.10.attn.qkv.bias": "model-00001-of-00004.safetensors",
334
+ "vision_backbone.featurizer.blocks.10.attn.qkv.weight": "model-00001-of-00004.safetensors",
335
+ "vision_backbone.featurizer.blocks.10.ls1.scale_factor": "model-00001-of-00004.safetensors",
336
+ "vision_backbone.featurizer.blocks.10.ls2.scale_factor": "model-00001-of-00004.safetensors",
337
+ "vision_backbone.featurizer.blocks.10.mlp.fc1.bias": "model-00001-of-00004.safetensors",
338
+ "vision_backbone.featurizer.blocks.10.mlp.fc1.weight": "model-00001-of-00004.safetensors",
339
+ "vision_backbone.featurizer.blocks.10.mlp.fc2.bias": "model-00001-of-00004.safetensors",
340
+ "vision_backbone.featurizer.blocks.10.mlp.fc2.weight": "model-00001-of-00004.safetensors",
341
+ "vision_backbone.featurizer.blocks.10.norm1.bias": "model-00001-of-00004.safetensors",
342
+ "vision_backbone.featurizer.blocks.10.norm1.weight": "model-00001-of-00004.safetensors",
343
+ "vision_backbone.featurizer.blocks.10.norm2.bias": "model-00001-of-00004.safetensors",
344
+ "vision_backbone.featurizer.blocks.10.norm2.weight": "model-00001-of-00004.safetensors",
345
+ "vision_backbone.featurizer.blocks.11.attn.proj.bias": "model-00001-of-00004.safetensors",
346
+ "vision_backbone.featurizer.blocks.11.attn.proj.weight": "model-00001-of-00004.safetensors",
347
+ "vision_backbone.featurizer.blocks.11.attn.qkv.bias": "model-00001-of-00004.safetensors",
348
+ "vision_backbone.featurizer.blocks.11.attn.qkv.weight": "model-00001-of-00004.safetensors",
349
+ "vision_backbone.featurizer.blocks.11.ls1.scale_factor": "model-00001-of-00004.safetensors",
350
+ "vision_backbone.featurizer.blocks.11.ls2.scale_factor": "model-00001-of-00004.safetensors",
351
+ "vision_backbone.featurizer.blocks.11.mlp.fc1.bias": "model-00001-of-00004.safetensors",
352
+ "vision_backbone.featurizer.blocks.11.mlp.fc1.weight": "model-00001-of-00004.safetensors",
353
+ "vision_backbone.featurizer.blocks.11.mlp.fc2.bias": "model-00001-of-00004.safetensors",
354
+ "vision_backbone.featurizer.blocks.11.mlp.fc2.weight": "model-00001-of-00004.safetensors",
355
+ "vision_backbone.featurizer.blocks.11.norm1.bias": "model-00001-of-00004.safetensors",
356
+ "vision_backbone.featurizer.blocks.11.norm1.weight": "model-00001-of-00004.safetensors",
357
+ "vision_backbone.featurizer.blocks.11.norm2.bias": "model-00001-of-00004.safetensors",
358
+ "vision_backbone.featurizer.blocks.11.norm2.weight": "model-00001-of-00004.safetensors",
359
+ "vision_backbone.featurizer.blocks.12.attn.proj.bias": "model-00001-of-00004.safetensors",
360
+ "vision_backbone.featurizer.blocks.12.attn.proj.weight": "model-00001-of-00004.safetensors",
361
+ "vision_backbone.featurizer.blocks.12.attn.qkv.bias": "model-00001-of-00004.safetensors",
362
+ "vision_backbone.featurizer.blocks.12.attn.qkv.weight": "model-00001-of-00004.safetensors",
363
+ "vision_backbone.featurizer.blocks.12.ls1.scale_factor": "model-00001-of-00004.safetensors",
364
+ "vision_backbone.featurizer.blocks.12.ls2.scale_factor": "model-00001-of-00004.safetensors",
365
+ "vision_backbone.featurizer.blocks.12.mlp.fc1.bias": "model-00001-of-00004.safetensors",
366
+ "vision_backbone.featurizer.blocks.12.mlp.fc1.weight": "model-00001-of-00004.safetensors",
367
+ "vision_backbone.featurizer.blocks.12.mlp.fc2.bias": "model-00001-of-00004.safetensors",
368
+ "vision_backbone.featurizer.blocks.12.mlp.fc2.weight": "model-00001-of-00004.safetensors",
369
+ "vision_backbone.featurizer.blocks.12.norm1.bias": "model-00001-of-00004.safetensors",
370
+ "vision_backbone.featurizer.blocks.12.norm1.weight": "model-00001-of-00004.safetensors",
371
+ "vision_backbone.featurizer.blocks.12.norm2.bias": "model-00001-of-00004.safetensors",
372
+ "vision_backbone.featurizer.blocks.12.norm2.weight": "model-00001-of-00004.safetensors",
373
+ "vision_backbone.featurizer.blocks.13.attn.proj.bias": "model-00001-of-00004.safetensors",
374
+ "vision_backbone.featurizer.blocks.13.attn.proj.weight": "model-00001-of-00004.safetensors",
375
+ "vision_backbone.featurizer.blocks.13.attn.qkv.bias": "model-00001-of-00004.safetensors",
376
+ "vision_backbone.featurizer.blocks.13.attn.qkv.weight": "model-00001-of-00004.safetensors",
377
+ "vision_backbone.featurizer.blocks.13.ls1.scale_factor": "model-00001-of-00004.safetensors",
378
+ "vision_backbone.featurizer.blocks.13.ls2.scale_factor": "model-00001-of-00004.safetensors",
379
+ "vision_backbone.featurizer.blocks.13.mlp.fc1.bias": "model-00001-of-00004.safetensors",
380
+ "vision_backbone.featurizer.blocks.13.mlp.fc1.weight": "model-00001-of-00004.safetensors",
381
+ "vision_backbone.featurizer.blocks.13.mlp.fc2.bias": "model-00001-of-00004.safetensors",
382
+ "vision_backbone.featurizer.blocks.13.mlp.fc2.weight": "model-00001-of-00004.safetensors",
383
+ "vision_backbone.featurizer.blocks.13.norm1.bias": "model-00001-of-00004.safetensors",
384
+ "vision_backbone.featurizer.blocks.13.norm1.weight": "model-00001-of-00004.safetensors",
385
+ "vision_backbone.featurizer.blocks.13.norm2.bias": "model-00001-of-00004.safetensors",
386
+ "vision_backbone.featurizer.blocks.13.norm2.weight": "model-00001-of-00004.safetensors",
387
+ "vision_backbone.featurizer.blocks.14.attn.proj.bias": "model-00001-of-00004.safetensors",
388
+ "vision_backbone.featurizer.blocks.14.attn.proj.weight": "model-00001-of-00004.safetensors",
389
+ "vision_backbone.featurizer.blocks.14.attn.qkv.bias": "model-00001-of-00004.safetensors",
390
+ "vision_backbone.featurizer.blocks.14.attn.qkv.weight": "model-00001-of-00004.safetensors",
391
+ "vision_backbone.featurizer.blocks.14.ls1.scale_factor": "model-00001-of-00004.safetensors",
392
+ "vision_backbone.featurizer.blocks.14.ls2.scale_factor": "model-00001-of-00004.safetensors",
393
+ "vision_backbone.featurizer.blocks.14.mlp.fc1.bias": "model-00001-of-00004.safetensors",
394
+ "vision_backbone.featurizer.blocks.14.mlp.fc1.weight": "model-00001-of-00004.safetensors",
395
+ "vision_backbone.featurizer.blocks.14.mlp.fc2.bias": "model-00001-of-00004.safetensors",
396
+ "vision_backbone.featurizer.blocks.14.mlp.fc2.weight": "model-00001-of-00004.safetensors",
397
+ "vision_backbone.featurizer.blocks.14.norm1.bias": "model-00001-of-00004.safetensors",
398
+ "vision_backbone.featurizer.blocks.14.norm1.weight": "model-00001-of-00004.safetensors",
399
+ "vision_backbone.featurizer.blocks.14.norm2.bias": "model-00001-of-00004.safetensors",
400
+ "vision_backbone.featurizer.blocks.14.norm2.weight": "model-00001-of-00004.safetensors",
401
+ "vision_backbone.featurizer.blocks.15.attn.proj.bias": "model-00001-of-00004.safetensors",
402
+ "vision_backbone.featurizer.blocks.15.attn.proj.weight": "model-00001-of-00004.safetensors",
403
+ "vision_backbone.featurizer.blocks.15.attn.qkv.bias": "model-00001-of-00004.safetensors",
404
+ "vision_backbone.featurizer.blocks.15.attn.qkv.weight": "model-00001-of-00004.safetensors",
405
+ "vision_backbone.featurizer.blocks.15.ls1.scale_factor": "model-00001-of-00004.safetensors",
406
+ "vision_backbone.featurizer.blocks.15.ls2.scale_factor": "model-00001-of-00004.safetensors",
407
+ "vision_backbone.featurizer.blocks.15.mlp.fc1.bias": "model-00001-of-00004.safetensors",
408
+ "vision_backbone.featurizer.blocks.15.mlp.fc1.weight": "model-00001-of-00004.safetensors",
409
+ "vision_backbone.featurizer.blocks.15.mlp.fc2.bias": "model-00001-of-00004.safetensors",
410
+ "vision_backbone.featurizer.blocks.15.mlp.fc2.weight": "model-00001-of-00004.safetensors",
411
+ "vision_backbone.featurizer.blocks.15.norm1.bias": "model-00001-of-00004.safetensors",
412
+ "vision_backbone.featurizer.blocks.15.norm1.weight": "model-00001-of-00004.safetensors",
413
+ "vision_backbone.featurizer.blocks.15.norm2.bias": "model-00001-of-00004.safetensors",
414
+ "vision_backbone.featurizer.blocks.15.norm2.weight": "model-00001-of-00004.safetensors",
415
+ "vision_backbone.featurizer.blocks.16.attn.proj.bias": "model-00001-of-00004.safetensors",
416
+ "vision_backbone.featurizer.blocks.16.attn.proj.weight": "model-00001-of-00004.safetensors",
417
+ "vision_backbone.featurizer.blocks.16.attn.qkv.bias": "model-00001-of-00004.safetensors",
418
+ "vision_backbone.featurizer.blocks.16.attn.qkv.weight": "model-00001-of-00004.safetensors",
419
+ "vision_backbone.featurizer.blocks.16.ls1.scale_factor": "model-00001-of-00004.safetensors",
420
+ "vision_backbone.featurizer.blocks.16.ls2.scale_factor": "model-00001-of-00004.safetensors",
421
+ "vision_backbone.featurizer.blocks.16.mlp.fc1.bias": "model-00001-of-00004.safetensors",
422
+ "vision_backbone.featurizer.blocks.16.mlp.fc1.weight": "model-00001-of-00004.safetensors",
423
+ "vision_backbone.featurizer.blocks.16.mlp.fc2.bias": "model-00001-of-00004.safetensors",
424
+ "vision_backbone.featurizer.blocks.16.mlp.fc2.weight": "model-00001-of-00004.safetensors",
425
+ "vision_backbone.featurizer.blocks.16.norm1.bias": "model-00001-of-00004.safetensors",
426
+ "vision_backbone.featurizer.blocks.16.norm1.weight": "model-00001-of-00004.safetensors",
427
+ "vision_backbone.featurizer.blocks.16.norm2.bias": "model-00001-of-00004.safetensors",
428
+ "vision_backbone.featurizer.blocks.16.norm2.weight": "model-00001-of-00004.safetensors",
429
+ "vision_backbone.featurizer.blocks.17.attn.proj.bias": "model-00001-of-00004.safetensors",
430
+ "vision_backbone.featurizer.blocks.17.attn.proj.weight": "model-00001-of-00004.safetensors",
431
+ "vision_backbone.featurizer.blocks.17.attn.qkv.bias": "model-00001-of-00004.safetensors",
432
+ "vision_backbone.featurizer.blocks.17.attn.qkv.weight": "model-00001-of-00004.safetensors",
433
+ "vision_backbone.featurizer.blocks.17.ls1.scale_factor": "model-00001-of-00004.safetensors",
434
+ "vision_backbone.featurizer.blocks.17.ls2.scale_factor": "model-00001-of-00004.safetensors",
435
+ "vision_backbone.featurizer.blocks.17.mlp.fc1.bias": "model-00001-of-00004.safetensors",
436
+ "vision_backbone.featurizer.blocks.17.mlp.fc1.weight": "model-00001-of-00004.safetensors",
437
+ "vision_backbone.featurizer.blocks.17.mlp.fc2.bias": "model-00001-of-00004.safetensors",
438
+ "vision_backbone.featurizer.blocks.17.mlp.fc2.weight": "model-00001-of-00004.safetensors",
439
+ "vision_backbone.featurizer.blocks.17.norm1.bias": "model-00001-of-00004.safetensors",
440
+ "vision_backbone.featurizer.blocks.17.norm1.weight": "model-00001-of-00004.safetensors",
441
+ "vision_backbone.featurizer.blocks.17.norm2.bias": "model-00001-of-00004.safetensors",
442
+ "vision_backbone.featurizer.blocks.17.norm2.weight": "model-00001-of-00004.safetensors",
443
+ "vision_backbone.featurizer.blocks.18.attn.proj.bias": "model-00001-of-00004.safetensors",
444
+ "vision_backbone.featurizer.blocks.18.attn.proj.weight": "model-00001-of-00004.safetensors",
445
+ "vision_backbone.featurizer.blocks.18.attn.qkv.bias": "model-00001-of-00004.safetensors",
446
+ "vision_backbone.featurizer.blocks.18.attn.qkv.weight": "model-00001-of-00004.safetensors",
447
+ "vision_backbone.featurizer.blocks.18.ls1.scale_factor": "model-00001-of-00004.safetensors",
448
+ "vision_backbone.featurizer.blocks.18.ls2.scale_factor": "model-00001-of-00004.safetensors",
449
+ "vision_backbone.featurizer.blocks.18.mlp.fc1.bias": "model-00001-of-00004.safetensors",
450
+ "vision_backbone.featurizer.blocks.18.mlp.fc1.weight": "model-00001-of-00004.safetensors",
451
+ "vision_backbone.featurizer.blocks.18.mlp.fc2.bias": "model-00001-of-00004.safetensors",
452
+ "vision_backbone.featurizer.blocks.18.mlp.fc2.weight": "model-00001-of-00004.safetensors",
453
+ "vision_backbone.featurizer.blocks.18.norm1.bias": "model-00001-of-00004.safetensors",
454
+ "vision_backbone.featurizer.blocks.18.norm1.weight": "model-00001-of-00004.safetensors",
455
+ "vision_backbone.featurizer.blocks.18.norm2.bias": "model-00001-of-00004.safetensors",
456
+ "vision_backbone.featurizer.blocks.18.norm2.weight": "model-00001-of-00004.safetensors",
457
+ "vision_backbone.featurizer.blocks.19.attn.proj.bias": "model-00001-of-00004.safetensors",
458
+ "vision_backbone.featurizer.blocks.19.attn.proj.weight": "model-00001-of-00004.safetensors",
459
+ "vision_backbone.featurizer.blocks.19.attn.qkv.bias": "model-00001-of-00004.safetensors",
460
+ "vision_backbone.featurizer.blocks.19.attn.qkv.weight": "model-00001-of-00004.safetensors",
461
+ "vision_backbone.featurizer.blocks.19.ls1.scale_factor": "model-00001-of-00004.safetensors",
462
+ "vision_backbone.featurizer.blocks.19.ls2.scale_factor": "model-00001-of-00004.safetensors",
463
+ "vision_backbone.featurizer.blocks.19.mlp.fc1.bias": "model-00001-of-00004.safetensors",
464
+ "vision_backbone.featurizer.blocks.19.mlp.fc1.weight": "model-00001-of-00004.safetensors",
465
+ "vision_backbone.featurizer.blocks.19.mlp.fc2.bias": "model-00001-of-00004.safetensors",
466
+ "vision_backbone.featurizer.blocks.19.mlp.fc2.weight": "model-00001-of-00004.safetensors",
467
+ "vision_backbone.featurizer.blocks.19.norm1.bias": "model-00001-of-00004.safetensors",
468
+ "vision_backbone.featurizer.blocks.19.norm1.weight": "model-00001-of-00004.safetensors",
469
+ "vision_backbone.featurizer.blocks.19.norm2.bias": "model-00001-of-00004.safetensors",
470
+ "vision_backbone.featurizer.blocks.19.norm2.weight": "model-00001-of-00004.safetensors",
471
+ "vision_backbone.featurizer.blocks.2.attn.proj.bias": "model-00001-of-00004.safetensors",
472
+ "vision_backbone.featurizer.blocks.2.attn.proj.weight": "model-00001-of-00004.safetensors",
473
+ "vision_backbone.featurizer.blocks.2.attn.qkv.bias": "model-00001-of-00004.safetensors",
474
+ "vision_backbone.featurizer.blocks.2.attn.qkv.weight": "model-00001-of-00004.safetensors",
475
+ "vision_backbone.featurizer.blocks.2.ls1.scale_factor": "model-00001-of-00004.safetensors",
476
+ "vision_backbone.featurizer.blocks.2.ls2.scale_factor": "model-00001-of-00004.safetensors",
477
+ "vision_backbone.featurizer.blocks.2.mlp.fc1.bias": "model-00001-of-00004.safetensors",
478
+ "vision_backbone.featurizer.blocks.2.mlp.fc1.weight": "model-00001-of-00004.safetensors",
479
+ "vision_backbone.featurizer.blocks.2.mlp.fc2.bias": "model-00001-of-00004.safetensors",
480
+ "vision_backbone.featurizer.blocks.2.mlp.fc2.weight": "model-00001-of-00004.safetensors",
481
+ "vision_backbone.featurizer.blocks.2.norm1.bias": "model-00001-of-00004.safetensors",
482
+ "vision_backbone.featurizer.blocks.2.norm1.weight": "model-00001-of-00004.safetensors",
483
+ "vision_backbone.featurizer.blocks.2.norm2.bias": "model-00001-of-00004.safetensors",
484
+ "vision_backbone.featurizer.blocks.2.norm2.weight": "model-00001-of-00004.safetensors",
485
+ "vision_backbone.featurizer.blocks.20.attn.proj.bias": "model-00001-of-00004.safetensors",
486
+ "vision_backbone.featurizer.blocks.20.attn.proj.weight": "model-00001-of-00004.safetensors",
487
+ "vision_backbone.featurizer.blocks.20.attn.qkv.bias": "model-00001-of-00004.safetensors",
488
+ "vision_backbone.featurizer.blocks.20.attn.qkv.weight": "model-00001-of-00004.safetensors",
489
+ "vision_backbone.featurizer.blocks.20.ls1.scale_factor": "model-00001-of-00004.safetensors",
490
+ "vision_backbone.featurizer.blocks.20.ls2.scale_factor": "model-00001-of-00004.safetensors",
491
+ "vision_backbone.featurizer.blocks.20.mlp.fc1.bias": "model-00001-of-00004.safetensors",
492
+ "vision_backbone.featurizer.blocks.20.mlp.fc1.weight": "model-00001-of-00004.safetensors",
493
+ "vision_backbone.featurizer.blocks.20.mlp.fc2.bias": "model-00001-of-00004.safetensors",
494
+ "vision_backbone.featurizer.blocks.20.mlp.fc2.weight": "model-00001-of-00004.safetensors",
495
+ "vision_backbone.featurizer.blocks.20.norm1.bias": "model-00001-of-00004.safetensors",
496
+ "vision_backbone.featurizer.blocks.20.norm1.weight": "model-00001-of-00004.safetensors",
497
+ "vision_backbone.featurizer.blocks.20.norm2.bias": "model-00001-of-00004.safetensors",
498
+ "vision_backbone.featurizer.blocks.20.norm2.weight": "model-00001-of-00004.safetensors",
499
+ "vision_backbone.featurizer.blocks.21.attn.proj.bias": "model-00001-of-00004.safetensors",
500
+ "vision_backbone.featurizer.blocks.21.attn.proj.weight": "model-00001-of-00004.safetensors",
501
+ "vision_backbone.featurizer.blocks.21.attn.qkv.bias": "model-00001-of-00004.safetensors",
502
+ "vision_backbone.featurizer.blocks.21.attn.qkv.weight": "model-00001-of-00004.safetensors",
503
+ "vision_backbone.featurizer.blocks.21.ls1.scale_factor": "model-00001-of-00004.safetensors",
504
+ "vision_backbone.featurizer.blocks.21.ls2.scale_factor": "model-00001-of-00004.safetensors",
505
+ "vision_backbone.featurizer.blocks.21.mlp.fc1.bias": "model-00001-of-00004.safetensors",
506
+ "vision_backbone.featurizer.blocks.21.mlp.fc1.weight": "model-00001-of-00004.safetensors",
507
+ "vision_backbone.featurizer.blocks.21.mlp.fc2.bias": "model-00001-of-00004.safetensors",
508
+ "vision_backbone.featurizer.blocks.21.mlp.fc2.weight": "model-00001-of-00004.safetensors",
509
+ "vision_backbone.featurizer.blocks.21.norm1.bias": "model-00001-of-00004.safetensors",
510
+ "vision_backbone.featurizer.blocks.21.norm1.weight": "model-00001-of-00004.safetensors",
511
+ "vision_backbone.featurizer.blocks.21.norm2.bias": "model-00001-of-00004.safetensors",
512
+ "vision_backbone.featurizer.blocks.21.norm2.weight": "model-00001-of-00004.safetensors",
513
+ "vision_backbone.featurizer.blocks.22.attn.proj.bias": "model-00001-of-00004.safetensors",
514
+ "vision_backbone.featurizer.blocks.22.attn.proj.weight": "model-00001-of-00004.safetensors",
515
+ "vision_backbone.featurizer.blocks.22.attn.qkv.bias": "model-00001-of-00004.safetensors",
516
+ "vision_backbone.featurizer.blocks.22.attn.qkv.weight": "model-00001-of-00004.safetensors",
517
+ "vision_backbone.featurizer.blocks.22.ls1.scale_factor": "model-00001-of-00004.safetensors",
518
+ "vision_backbone.featurizer.blocks.22.ls2.scale_factor": "model-00001-of-00004.safetensors",
519
+ "vision_backbone.featurizer.blocks.22.mlp.fc1.bias": "model-00001-of-00004.safetensors",
520
+ "vision_backbone.featurizer.blocks.22.mlp.fc1.weight": "model-00001-of-00004.safetensors",
521
+ "vision_backbone.featurizer.blocks.22.mlp.fc2.bias": "model-00001-of-00004.safetensors",
522
+ "vision_backbone.featurizer.blocks.22.mlp.fc2.weight": "model-00001-of-00004.safetensors",
523
+ "vision_backbone.featurizer.blocks.22.norm1.bias": "model-00001-of-00004.safetensors",
524
+ "vision_backbone.featurizer.blocks.22.norm1.weight": "model-00001-of-00004.safetensors",
525
+ "vision_backbone.featurizer.blocks.22.norm2.bias": "model-00001-of-00004.safetensors",
526
+ "vision_backbone.featurizer.blocks.22.norm2.weight": "model-00001-of-00004.safetensors",
527
+ "vision_backbone.featurizer.blocks.23.attn.proj.bias": "model-00001-of-00004.safetensors",
528
+ "vision_backbone.featurizer.blocks.23.attn.proj.weight": "model-00001-of-00004.safetensors",
529
+ "vision_backbone.featurizer.blocks.23.attn.qkv.bias": "model-00001-of-00004.safetensors",
530
+ "vision_backbone.featurizer.blocks.23.attn.qkv.weight": "model-00001-of-00004.safetensors",
531
+ "vision_backbone.featurizer.blocks.23.ls1.scale_factor": "model-00001-of-00004.safetensors",
532
+ "vision_backbone.featurizer.blocks.23.ls2.scale_factor": "model-00001-of-00004.safetensors",
533
+ "vision_backbone.featurizer.blocks.23.mlp.fc1.bias": "model-00001-of-00004.safetensors",
534
+ "vision_backbone.featurizer.blocks.23.mlp.fc1.weight": "model-00001-of-00004.safetensors",
535
+ "vision_backbone.featurizer.blocks.23.mlp.fc2.bias": "model-00001-of-00004.safetensors",
536
+ "vision_backbone.featurizer.blocks.23.mlp.fc2.weight": "model-00001-of-00004.safetensors",
537
+ "vision_backbone.featurizer.blocks.23.norm1.bias": "model-00001-of-00004.safetensors",
538
+ "vision_backbone.featurizer.blocks.23.norm1.weight": "model-00001-of-00004.safetensors",
539
+ "vision_backbone.featurizer.blocks.23.norm2.bias": "model-00001-of-00004.safetensors",
540
+ "vision_backbone.featurizer.blocks.23.norm2.weight": "model-00001-of-00004.safetensors",
541
+ "vision_backbone.featurizer.blocks.3.attn.proj.bias": "model-00001-of-00004.safetensors",
542
+ "vision_backbone.featurizer.blocks.3.attn.proj.weight": "model-00001-of-00004.safetensors",
543
+ "vision_backbone.featurizer.blocks.3.attn.qkv.bias": "model-00001-of-00004.safetensors",
544
+ "vision_backbone.featurizer.blocks.3.attn.qkv.weight": "model-00001-of-00004.safetensors",
545
+ "vision_backbone.featurizer.blocks.3.ls1.scale_factor": "model-00001-of-00004.safetensors",
546
+ "vision_backbone.featurizer.blocks.3.ls2.scale_factor": "model-00001-of-00004.safetensors",
547
+ "vision_backbone.featurizer.blocks.3.mlp.fc1.bias": "model-00001-of-00004.safetensors",
548
+ "vision_backbone.featurizer.blocks.3.mlp.fc1.weight": "model-00001-of-00004.safetensors",
549
+ "vision_backbone.featurizer.blocks.3.mlp.fc2.bias": "model-00001-of-00004.safetensors",
550
+ "vision_backbone.featurizer.blocks.3.mlp.fc2.weight": "model-00001-of-00004.safetensors",
551
+ "vision_backbone.featurizer.blocks.3.norm1.bias": "model-00001-of-00004.safetensors",
552
+ "vision_backbone.featurizer.blocks.3.norm1.weight": "model-00001-of-00004.safetensors",
553
+ "vision_backbone.featurizer.blocks.3.norm2.bias": "model-00001-of-00004.safetensors",
554
+ "vision_backbone.featurizer.blocks.3.norm2.weight": "model-00001-of-00004.safetensors",
555
+ "vision_backbone.featurizer.blocks.4.attn.proj.bias": "model-00001-of-00004.safetensors",
556
+ "vision_backbone.featurizer.blocks.4.attn.proj.weight": "model-00001-of-00004.safetensors",
557
+ "vision_backbone.featurizer.blocks.4.attn.qkv.bias": "model-00001-of-00004.safetensors",
558
+ "vision_backbone.featurizer.blocks.4.attn.qkv.weight": "model-00001-of-00004.safetensors",
559
+ "vision_backbone.featurizer.blocks.4.ls1.scale_factor": "model-00001-of-00004.safetensors",
560
+ "vision_backbone.featurizer.blocks.4.ls2.scale_factor": "model-00001-of-00004.safetensors",
561
+ "vision_backbone.featurizer.blocks.4.mlp.fc1.bias": "model-00001-of-00004.safetensors",
562
+ "vision_backbone.featurizer.blocks.4.mlp.fc1.weight": "model-00001-of-00004.safetensors",
563
+ "vision_backbone.featurizer.blocks.4.mlp.fc2.bias": "model-00001-of-00004.safetensors",
564
+ "vision_backbone.featurizer.blocks.4.mlp.fc2.weight": "model-00001-of-00004.safetensors",
565
+ "vision_backbone.featurizer.blocks.4.norm1.bias": "model-00001-of-00004.safetensors",
566
+ "vision_backbone.featurizer.blocks.4.norm1.weight": "model-00001-of-00004.safetensors",
567
+ "vision_backbone.featurizer.blocks.4.norm2.bias": "model-00001-of-00004.safetensors",
568
+ "vision_backbone.featurizer.blocks.4.norm2.weight": "model-00001-of-00004.safetensors",
569
+ "vision_backbone.featurizer.blocks.5.attn.proj.bias": "model-00001-of-00004.safetensors",
570
+ "vision_backbone.featurizer.blocks.5.attn.proj.weight": "model-00001-of-00004.safetensors",
571
+ "vision_backbone.featurizer.blocks.5.attn.qkv.bias": "model-00001-of-00004.safetensors",
572
+ "vision_backbone.featurizer.blocks.5.attn.qkv.weight": "model-00001-of-00004.safetensors",
573
+ "vision_backbone.featurizer.blocks.5.ls1.scale_factor": "model-00001-of-00004.safetensors",
574
+ "vision_backbone.featurizer.blocks.5.ls2.scale_factor": "model-00001-of-00004.safetensors",
575
+ "vision_backbone.featurizer.blocks.5.mlp.fc1.bias": "model-00001-of-00004.safetensors",
576
+ "vision_backbone.featurizer.blocks.5.mlp.fc1.weight": "model-00001-of-00004.safetensors",
577
+ "vision_backbone.featurizer.blocks.5.mlp.fc2.bias": "model-00001-of-00004.safetensors",
578
+ "vision_backbone.featurizer.blocks.5.mlp.fc2.weight": "model-00001-of-00004.safetensors",
579
+ "vision_backbone.featurizer.blocks.5.norm1.bias": "model-00001-of-00004.safetensors",
580
+ "vision_backbone.featurizer.blocks.5.norm1.weight": "model-00001-of-00004.safetensors",
581
+ "vision_backbone.featurizer.blocks.5.norm2.bias": "model-00001-of-00004.safetensors",
582
+ "vision_backbone.featurizer.blocks.5.norm2.weight": "model-00001-of-00004.safetensors",
583
+ "vision_backbone.featurizer.blocks.6.attn.proj.bias": "model-00001-of-00004.safetensors",
584
+ "vision_backbone.featurizer.blocks.6.attn.proj.weight": "model-00001-of-00004.safetensors",
585
+ "vision_backbone.featurizer.blocks.6.attn.qkv.bias": "model-00001-of-00004.safetensors",
586
+ "vision_backbone.featurizer.blocks.6.attn.qkv.weight": "model-00001-of-00004.safetensors",
587
+ "vision_backbone.featurizer.blocks.6.ls1.scale_factor": "model-00001-of-00004.safetensors",
588
+ "vision_backbone.featurizer.blocks.6.ls2.scale_factor": "model-00001-of-00004.safetensors",
589
+ "vision_backbone.featurizer.blocks.6.mlp.fc1.bias": "model-00001-of-00004.safetensors",
590
+ "vision_backbone.featurizer.blocks.6.mlp.fc1.weight": "model-00001-of-00004.safetensors",
591
+ "vision_backbone.featurizer.blocks.6.mlp.fc2.bias": "model-00001-of-00004.safetensors",
592
+ "vision_backbone.featurizer.blocks.6.mlp.fc2.weight": "model-00001-of-00004.safetensors",
593
+ "vision_backbone.featurizer.blocks.6.norm1.bias": "model-00001-of-00004.safetensors",
594
+ "vision_backbone.featurizer.blocks.6.norm1.weight": "model-00001-of-00004.safetensors",
595
+ "vision_backbone.featurizer.blocks.6.norm2.bias": "model-00001-of-00004.safetensors",
596
+ "vision_backbone.featurizer.blocks.6.norm2.weight": "model-00001-of-00004.safetensors",
597
+ "vision_backbone.featurizer.blocks.7.attn.proj.bias": "model-00001-of-00004.safetensors",
598
+ "vision_backbone.featurizer.blocks.7.attn.proj.weight": "model-00001-of-00004.safetensors",
599
+ "vision_backbone.featurizer.blocks.7.attn.qkv.bias": "model-00001-of-00004.safetensors",
600
+ "vision_backbone.featurizer.blocks.7.attn.qkv.weight": "model-00001-of-00004.safetensors",
601
+ "vision_backbone.featurizer.blocks.7.ls1.scale_factor": "model-00001-of-00004.safetensors",
602
+ "vision_backbone.featurizer.blocks.7.ls2.scale_factor": "model-00001-of-00004.safetensors",
603
+ "vision_backbone.featurizer.blocks.7.mlp.fc1.bias": "model-00001-of-00004.safetensors",
604
+ "vision_backbone.featurizer.blocks.7.mlp.fc1.weight": "model-00001-of-00004.safetensors",
605
+ "vision_backbone.featurizer.blocks.7.mlp.fc2.bias": "model-00001-of-00004.safetensors",
606
+ "vision_backbone.featurizer.blocks.7.mlp.fc2.weight": "model-00001-of-00004.safetensors",
607
+ "vision_backbone.featurizer.blocks.7.norm1.bias": "model-00001-of-00004.safetensors",
608
+ "vision_backbone.featurizer.blocks.7.norm1.weight": "model-00001-of-00004.safetensors",
609
+ "vision_backbone.featurizer.blocks.7.norm2.bias": "model-00001-of-00004.safetensors",
610
+ "vision_backbone.featurizer.blocks.7.norm2.weight": "model-00001-of-00004.safetensors",
611
+ "vision_backbone.featurizer.blocks.8.attn.proj.bias": "model-00001-of-00004.safetensors",
612
+ "vision_backbone.featurizer.blocks.8.attn.proj.weight": "model-00001-of-00004.safetensors",
613
+ "vision_backbone.featurizer.blocks.8.attn.qkv.bias": "model-00001-of-00004.safetensors",
614
+ "vision_backbone.featurizer.blocks.8.attn.qkv.weight": "model-00001-of-00004.safetensors",
615
+ "vision_backbone.featurizer.blocks.8.ls1.scale_factor": "model-00001-of-00004.safetensors",
616
+ "vision_backbone.featurizer.blocks.8.ls2.scale_factor": "model-00001-of-00004.safetensors",
617
+ "vision_backbone.featurizer.blocks.8.mlp.fc1.bias": "model-00001-of-00004.safetensors",
618
+ "vision_backbone.featurizer.blocks.8.mlp.fc1.weight": "model-00001-of-00004.safetensors",
619
+ "vision_backbone.featurizer.blocks.8.mlp.fc2.bias": "model-00001-of-00004.safetensors",
620
+ "vision_backbone.featurizer.blocks.8.mlp.fc2.weight": "model-00001-of-00004.safetensors",
621
+ "vision_backbone.featurizer.blocks.8.norm1.bias": "model-00001-of-00004.safetensors",
622
+ "vision_backbone.featurizer.blocks.8.norm1.weight": "model-00001-of-00004.safetensors",
623
+ "vision_backbone.featurizer.blocks.8.norm2.bias": "model-00001-of-00004.safetensors",
624
+ "vision_backbone.featurizer.blocks.8.norm2.weight": "model-00001-of-00004.safetensors",
625
+ "vision_backbone.featurizer.blocks.9.attn.proj.bias": "model-00001-of-00004.safetensors",
626
+ "vision_backbone.featurizer.blocks.9.attn.proj.weight": "model-00001-of-00004.safetensors",
627
+ "vision_backbone.featurizer.blocks.9.attn.qkv.bias": "model-00001-of-00004.safetensors",
628
+ "vision_backbone.featurizer.blocks.9.attn.qkv.weight": "model-00001-of-00004.safetensors",
629
+ "vision_backbone.featurizer.blocks.9.ls1.scale_factor": "model-00001-of-00004.safetensors",
630
+ "vision_backbone.featurizer.blocks.9.ls2.scale_factor": "model-00001-of-00004.safetensors",
631
+ "vision_backbone.featurizer.blocks.9.mlp.fc1.bias": "model-00001-of-00004.safetensors",
632
+ "vision_backbone.featurizer.blocks.9.mlp.fc1.weight": "model-00001-of-00004.safetensors",
633
+ "vision_backbone.featurizer.blocks.9.mlp.fc2.bias": "model-00001-of-00004.safetensors",
634
+ "vision_backbone.featurizer.blocks.9.mlp.fc2.weight": "model-00001-of-00004.safetensors",
635
+ "vision_backbone.featurizer.blocks.9.norm1.bias": "model-00001-of-00004.safetensors",
636
+ "vision_backbone.featurizer.blocks.9.norm1.weight": "model-00001-of-00004.safetensors",
637
+ "vision_backbone.featurizer.blocks.9.norm2.bias": "model-00001-of-00004.safetensors",
638
+ "vision_backbone.featurizer.blocks.9.norm2.weight": "model-00001-of-00004.safetensors",
639
+ "vision_backbone.featurizer.cls_token": "model-00001-of-00004.safetensors",
640
+ "vision_backbone.featurizer.norm.bias": "model-00001-of-00004.safetensors",
641
+ "vision_backbone.featurizer.norm.weight": "model-00001-of-00004.safetensors",
642
+ "vision_backbone.featurizer.patch_embed.proj.bias": "model-00001-of-00004.safetensors",
643
+ "vision_backbone.featurizer.patch_embed.proj.weight": "model-00001-of-00004.safetensors",
644
+ "vision_backbone.featurizer.pos_embed": "model-00001-of-00004.safetensors",
645
+ "vision_backbone.featurizer.reg_token": "model-00001-of-00004.safetensors",
646
+ "vision_backbone.fused_featurizer.attn_pool.kv.bias": "model-00001-of-00004.safetensors",
647
+ "vision_backbone.fused_featurizer.attn_pool.kv.weight": "model-00001-of-00004.safetensors",
648
+ "vision_backbone.fused_featurizer.attn_pool.latent": "model-00001-of-00004.safetensors",
649
+ "vision_backbone.fused_featurizer.attn_pool.mlp.fc1.bias": "model-00001-of-00004.safetensors",
650
+ "vision_backbone.fused_featurizer.attn_pool.mlp.fc1.weight": "model-00001-of-00004.safetensors",
651
+ "vision_backbone.fused_featurizer.attn_pool.mlp.fc2.bias": "model-00001-of-00004.safetensors",
652
+ "vision_backbone.fused_featurizer.attn_pool.mlp.fc2.weight": "model-00001-of-00004.safetensors",
653
+ "vision_backbone.fused_featurizer.attn_pool.norm.bias": "model-00001-of-00004.safetensors",
654
+ "vision_backbone.fused_featurizer.attn_pool.norm.weight": "model-00001-of-00004.safetensors",
655
+ "vision_backbone.fused_featurizer.attn_pool.proj.bias": "model-00001-of-00004.safetensors",
656
+ "vision_backbone.fused_featurizer.attn_pool.proj.weight": "model-00001-of-00004.safetensors",
657
+ "vision_backbone.fused_featurizer.attn_pool.q.bias": "model-00001-of-00004.safetensors",
658
+ "vision_backbone.fused_featurizer.attn_pool.q.weight": "model-00001-of-00004.safetensors",
659
+ "vision_backbone.fused_featurizer.blocks.0.attn.proj.bias": "model-00001-of-00004.safetensors",
660
+ "vision_backbone.fused_featurizer.blocks.0.attn.proj.weight": "model-00001-of-00004.safetensors",
661
+ "vision_backbone.fused_featurizer.blocks.0.attn.qkv.bias": "model-00001-of-00004.safetensors",
662
+ "vision_backbone.fused_featurizer.blocks.0.attn.qkv.weight": "model-00001-of-00004.safetensors",
663
+ "vision_backbone.fused_featurizer.blocks.0.mlp.fc1.bias": "model-00001-of-00004.safetensors",
664
+ "vision_backbone.fused_featurizer.blocks.0.mlp.fc1.weight": "model-00001-of-00004.safetensors",
665
+ "vision_backbone.fused_featurizer.blocks.0.mlp.fc2.bias": "model-00001-of-00004.safetensors",
666
+ "vision_backbone.fused_featurizer.blocks.0.mlp.fc2.weight": "model-00001-of-00004.safetensors",
667
+ "vision_backbone.fused_featurizer.blocks.0.norm1.bias": "model-00001-of-00004.safetensors",
668
+ "vision_backbone.fused_featurizer.blocks.0.norm1.weight": "model-00001-of-00004.safetensors",
669
+ "vision_backbone.fused_featurizer.blocks.0.norm2.bias": "model-00001-of-00004.safetensors",
670
+ "vision_backbone.fused_featurizer.blocks.0.norm2.weight": "model-00001-of-00004.safetensors",
671
+ "vision_backbone.fused_featurizer.blocks.1.attn.proj.bias": "model-00001-of-00004.safetensors",
672
+ "vision_backbone.fused_featurizer.blocks.1.attn.proj.weight": "model-00001-of-00004.safetensors",
673
+ "vision_backbone.fused_featurizer.blocks.1.attn.qkv.bias": "model-00001-of-00004.safetensors",
674
+ "vision_backbone.fused_featurizer.blocks.1.attn.qkv.weight": "model-00001-of-00004.safetensors",
675
+ "vision_backbone.fused_featurizer.blocks.1.mlp.fc1.bias": "model-00001-of-00004.safetensors",
676
+ "vision_backbone.fused_featurizer.blocks.1.mlp.fc1.weight": "model-00001-of-00004.safetensors",
677
+ "vision_backbone.fused_featurizer.blocks.1.mlp.fc2.bias": "model-00001-of-00004.safetensors",
678
+ "vision_backbone.fused_featurizer.blocks.1.mlp.fc2.weight": "model-00001-of-00004.safetensors",
679
+ "vision_backbone.fused_featurizer.blocks.1.norm1.bias": "model-00001-of-00004.safetensors",
680
+ "vision_backbone.fused_featurizer.blocks.1.norm1.weight": "model-00001-of-00004.safetensors",
681
+ "vision_backbone.fused_featurizer.blocks.1.norm2.bias": "model-00001-of-00004.safetensors",
682
+ "vision_backbone.fused_featurizer.blocks.1.norm2.weight": "model-00001-of-00004.safetensors",
683
+ "vision_backbone.fused_featurizer.blocks.10.attn.proj.bias": "model-00001-of-00004.safetensors",
684
+ "vision_backbone.fused_featurizer.blocks.10.attn.proj.weight": "model-00001-of-00004.safetensors",
685
+ "vision_backbone.fused_featurizer.blocks.10.attn.qkv.bias": "model-00001-of-00004.safetensors",
686
+ "vision_backbone.fused_featurizer.blocks.10.attn.qkv.weight": "model-00001-of-00004.safetensors",
687
+ "vision_backbone.fused_featurizer.blocks.10.mlp.fc1.bias": "model-00001-of-00004.safetensors",
688
+ "vision_backbone.fused_featurizer.blocks.10.mlp.fc1.weight": "model-00001-of-00004.safetensors",
689
+ "vision_backbone.fused_featurizer.blocks.10.mlp.fc2.bias": "model-00001-of-00004.safetensors",
690
+ "vision_backbone.fused_featurizer.blocks.10.mlp.fc2.weight": "model-00001-of-00004.safetensors",
691
+ "vision_backbone.fused_featurizer.blocks.10.norm1.bias": "model-00001-of-00004.safetensors",
692
+ "vision_backbone.fused_featurizer.blocks.10.norm1.weight": "model-00001-of-00004.safetensors",
693
+ "vision_backbone.fused_featurizer.blocks.10.norm2.bias": "model-00001-of-00004.safetensors",
694
+ "vision_backbone.fused_featurizer.blocks.10.norm2.weight": "model-00001-of-00004.safetensors",
695
+ "vision_backbone.fused_featurizer.blocks.11.attn.proj.bias": "model-00001-of-00004.safetensors",
696
+ "vision_backbone.fused_featurizer.blocks.11.attn.proj.weight": "model-00001-of-00004.safetensors",
697
+ "vision_backbone.fused_featurizer.blocks.11.attn.qkv.bias": "model-00001-of-00004.safetensors",
698
+ "vision_backbone.fused_featurizer.blocks.11.attn.qkv.weight": "model-00001-of-00004.safetensors",
699
+ "vision_backbone.fused_featurizer.blocks.11.mlp.fc1.bias": "model-00001-of-00004.safetensors",
700
+ "vision_backbone.fused_featurizer.blocks.11.mlp.fc1.weight": "model-00001-of-00004.safetensors",
701
+ "vision_backbone.fused_featurizer.blocks.11.mlp.fc2.bias": "model-00001-of-00004.safetensors",
702
+ "vision_backbone.fused_featurizer.blocks.11.mlp.fc2.weight": "model-00001-of-00004.safetensors",
703
+ "vision_backbone.fused_featurizer.blocks.11.norm1.bias": "model-00001-of-00004.safetensors",
704
+ "vision_backbone.fused_featurizer.blocks.11.norm1.weight": "model-00001-of-00004.safetensors",
705
+ "vision_backbone.fused_featurizer.blocks.11.norm2.bias": "model-00001-of-00004.safetensors",
706
+ "vision_backbone.fused_featurizer.blocks.11.norm2.weight": "model-00001-of-00004.safetensors",
707
+ "vision_backbone.fused_featurizer.blocks.12.attn.proj.bias": "model-00001-of-00004.safetensors",
708
+ "vision_backbone.fused_featurizer.blocks.12.attn.proj.weight": "model-00001-of-00004.safetensors",
709
+ "vision_backbone.fused_featurizer.blocks.12.attn.qkv.bias": "model-00001-of-00004.safetensors",
710
+ "vision_backbone.fused_featurizer.blocks.12.attn.qkv.weight": "model-00001-of-00004.safetensors",
711
+ "vision_backbone.fused_featurizer.blocks.12.mlp.fc1.bias": "model-00001-of-00004.safetensors",
712
+ "vision_backbone.fused_featurizer.blocks.12.mlp.fc1.weight": "model-00001-of-00004.safetensors",
713
+ "vision_backbone.fused_featurizer.blocks.12.mlp.fc2.bias": "model-00001-of-00004.safetensors",
714
+ "vision_backbone.fused_featurizer.blocks.12.mlp.fc2.weight": "model-00001-of-00004.safetensors",
715
+ "vision_backbone.fused_featurizer.blocks.12.norm1.bias": "model-00001-of-00004.safetensors",
716
+ "vision_backbone.fused_featurizer.blocks.12.norm1.weight": "model-00001-of-00004.safetensors",
717
+ "vision_backbone.fused_featurizer.blocks.12.norm2.bias": "model-00001-of-00004.safetensors",
718
+ "vision_backbone.fused_featurizer.blocks.12.norm2.weight": "model-00001-of-00004.safetensors",
719
+ "vision_backbone.fused_featurizer.blocks.13.attn.proj.bias": "model-00001-of-00004.safetensors",
720
+ "vision_backbone.fused_featurizer.blocks.13.attn.proj.weight": "model-00001-of-00004.safetensors",
721
+ "vision_backbone.fused_featurizer.blocks.13.attn.qkv.bias": "model-00001-of-00004.safetensors",
722
+ "vision_backbone.fused_featurizer.blocks.13.attn.qkv.weight": "model-00001-of-00004.safetensors",
723
+ "vision_backbone.fused_featurizer.blocks.13.mlp.fc1.bias": "model-00001-of-00004.safetensors",
724
+ "vision_backbone.fused_featurizer.blocks.13.mlp.fc1.weight": "model-00001-of-00004.safetensors",
725
+ "vision_backbone.fused_featurizer.blocks.13.mlp.fc2.bias": "model-00001-of-00004.safetensors",
726
+ "vision_backbone.fused_featurizer.blocks.13.mlp.fc2.weight": "model-00001-of-00004.safetensors",
727
+ "vision_backbone.fused_featurizer.blocks.13.norm1.bias": "model-00001-of-00004.safetensors",
728
+ "vision_backbone.fused_featurizer.blocks.13.norm1.weight": "model-00001-of-00004.safetensors",
729
+ "vision_backbone.fused_featurizer.blocks.13.norm2.bias": "model-00001-of-00004.safetensors",
730
+ "vision_backbone.fused_featurizer.blocks.13.norm2.weight": "model-00001-of-00004.safetensors",
731
+ "vision_backbone.fused_featurizer.blocks.14.attn.proj.bias": "model-00001-of-00004.safetensors",
732
+ "vision_backbone.fused_featurizer.blocks.14.attn.proj.weight": "model-00001-of-00004.safetensors",
733
+ "vision_backbone.fused_featurizer.blocks.14.attn.qkv.bias": "model-00001-of-00004.safetensors",
734
+ "vision_backbone.fused_featurizer.blocks.14.attn.qkv.weight": "model-00001-of-00004.safetensors",
735
+ "vision_backbone.fused_featurizer.blocks.14.mlp.fc1.bias": "model-00001-of-00004.safetensors",
736
+ "vision_backbone.fused_featurizer.blocks.14.mlp.fc1.weight": "model-00001-of-00004.safetensors",
737
+ "vision_backbone.fused_featurizer.blocks.14.mlp.fc2.bias": "model-00001-of-00004.safetensors",
738
+ "vision_backbone.fused_featurizer.blocks.14.mlp.fc2.weight": "model-00001-of-00004.safetensors",
739
+ "vision_backbone.fused_featurizer.blocks.14.norm1.bias": "model-00001-of-00004.safetensors",
740
+ "vision_backbone.fused_featurizer.blocks.14.norm1.weight": "model-00001-of-00004.safetensors",
741
+ "vision_backbone.fused_featurizer.blocks.14.norm2.bias": "model-00001-of-00004.safetensors",
742
+ "vision_backbone.fused_featurizer.blocks.14.norm2.weight": "model-00001-of-00004.safetensors",
743
+ "vision_backbone.fused_featurizer.blocks.15.attn.proj.bias": "model-00001-of-00004.safetensors",
744
+ "vision_backbone.fused_featurizer.blocks.15.attn.proj.weight": "model-00001-of-00004.safetensors",
745
+ "vision_backbone.fused_featurizer.blocks.15.attn.qkv.bias": "model-00001-of-00004.safetensors",
746
+ "vision_backbone.fused_featurizer.blocks.15.attn.qkv.weight": "model-00001-of-00004.safetensors",
747
+ "vision_backbone.fused_featurizer.blocks.15.mlp.fc1.bias": "model-00001-of-00004.safetensors",
748
+ "vision_backbone.fused_featurizer.blocks.15.mlp.fc1.weight": "model-00001-of-00004.safetensors",
749
+ "vision_backbone.fused_featurizer.blocks.15.mlp.fc2.bias": "model-00001-of-00004.safetensors",
750
+ "vision_backbone.fused_featurizer.blocks.15.mlp.fc2.weight": "model-00001-of-00004.safetensors",
751
+ "vision_backbone.fused_featurizer.blocks.15.norm1.bias": "model-00001-of-00004.safetensors",
752
+ "vision_backbone.fused_featurizer.blocks.15.norm1.weight": "model-00001-of-00004.safetensors",
753
+ "vision_backbone.fused_featurizer.blocks.15.norm2.bias": "model-00001-of-00004.safetensors",
754
+ "vision_backbone.fused_featurizer.blocks.15.norm2.weight": "model-00001-of-00004.safetensors",
755
+ "vision_backbone.fused_featurizer.blocks.16.attn.proj.bias": "model-00001-of-00004.safetensors",
756
+ "vision_backbone.fused_featurizer.blocks.16.attn.proj.weight": "model-00001-of-00004.safetensors",
757
+ "vision_backbone.fused_featurizer.blocks.16.attn.qkv.bias": "model-00001-of-00004.safetensors",
758
+ "vision_backbone.fused_featurizer.blocks.16.attn.qkv.weight": "model-00001-of-00004.safetensors",
759
+ "vision_backbone.fused_featurizer.blocks.16.mlp.fc1.bias": "model-00001-of-00004.safetensors",
760
+ "vision_backbone.fused_featurizer.blocks.16.mlp.fc1.weight": "model-00001-of-00004.safetensors",
761
+ "vision_backbone.fused_featurizer.blocks.16.mlp.fc2.bias": "model-00001-of-00004.safetensors",
762
+ "vision_backbone.fused_featurizer.blocks.16.mlp.fc2.weight": "model-00001-of-00004.safetensors",
763
+ "vision_backbone.fused_featurizer.blocks.16.norm1.bias": "model-00001-of-00004.safetensors",
764
+ "vision_backbone.fused_featurizer.blocks.16.norm1.weight": "model-00001-of-00004.safetensors",
765
+ "vision_backbone.fused_featurizer.blocks.16.norm2.bias": "model-00001-of-00004.safetensors",
766
+ "vision_backbone.fused_featurizer.blocks.16.norm2.weight": "model-00001-of-00004.safetensors",
767
+ "vision_backbone.fused_featurizer.blocks.17.attn.proj.bias": "model-00001-of-00004.safetensors",
768
+ "vision_backbone.fused_featurizer.blocks.17.attn.proj.weight": "model-00001-of-00004.safetensors",
769
+ "vision_backbone.fused_featurizer.blocks.17.attn.qkv.bias": "model-00001-of-00004.safetensors",
770
+ "vision_backbone.fused_featurizer.blocks.17.attn.qkv.weight": "model-00001-of-00004.safetensors",
771
+ "vision_backbone.fused_featurizer.blocks.17.mlp.fc1.bias": "model-00001-of-00004.safetensors",
772
+ "vision_backbone.fused_featurizer.blocks.17.mlp.fc1.weight": "model-00001-of-00004.safetensors",
773
+ "vision_backbone.fused_featurizer.blocks.17.mlp.fc2.bias": "model-00001-of-00004.safetensors",
774
+ "vision_backbone.fused_featurizer.blocks.17.mlp.fc2.weight": "model-00001-of-00004.safetensors",
775
+ "vision_backbone.fused_featurizer.blocks.17.norm1.bias": "model-00001-of-00004.safetensors",
776
+ "vision_backbone.fused_featurizer.blocks.17.norm1.weight": "model-00001-of-00004.safetensors",
777
+ "vision_backbone.fused_featurizer.blocks.17.norm2.bias": "model-00001-of-00004.safetensors",
778
+ "vision_backbone.fused_featurizer.blocks.17.norm2.weight": "model-00001-of-00004.safetensors",
779
+ "vision_backbone.fused_featurizer.blocks.18.attn.proj.bias": "model-00001-of-00004.safetensors",
780
+ "vision_backbone.fused_featurizer.blocks.18.attn.proj.weight": "model-00001-of-00004.safetensors",
781
+ "vision_backbone.fused_featurizer.blocks.18.attn.qkv.bias": "model-00001-of-00004.safetensors",
782
+ "vision_backbone.fused_featurizer.blocks.18.attn.qkv.weight": "model-00001-of-00004.safetensors",
783
+ "vision_backbone.fused_featurizer.blocks.18.mlp.fc1.bias": "model-00001-of-00004.safetensors",
784
+ "vision_backbone.fused_featurizer.blocks.18.mlp.fc1.weight": "model-00001-of-00004.safetensors",
785
+ "vision_backbone.fused_featurizer.blocks.18.mlp.fc2.bias": "model-00001-of-00004.safetensors",
786
+ "vision_backbone.fused_featurizer.blocks.18.mlp.fc2.weight": "model-00001-of-00004.safetensors",
787
+ "vision_backbone.fused_featurizer.blocks.18.norm1.bias": "model-00001-of-00004.safetensors",
788
+ "vision_backbone.fused_featurizer.blocks.18.norm1.weight": "model-00001-of-00004.safetensors",
789
+ "vision_backbone.fused_featurizer.blocks.18.norm2.bias": "model-00001-of-00004.safetensors",
790
+ "vision_backbone.fused_featurizer.blocks.18.norm2.weight": "model-00001-of-00004.safetensors",
791
+ "vision_backbone.fused_featurizer.blocks.19.attn.proj.bias": "model-00001-of-00004.safetensors",
792
+ "vision_backbone.fused_featurizer.blocks.19.attn.proj.weight": "model-00001-of-00004.safetensors",
793
+ "vision_backbone.fused_featurizer.blocks.19.attn.qkv.bias": "model-00001-of-00004.safetensors",
794
+ "vision_backbone.fused_featurizer.blocks.19.attn.qkv.weight": "model-00001-of-00004.safetensors",
795
+ "vision_backbone.fused_featurizer.blocks.19.mlp.fc1.bias": "model-00001-of-00004.safetensors",
796
+ "vision_backbone.fused_featurizer.blocks.19.mlp.fc1.weight": "model-00001-of-00004.safetensors",
797
+ "vision_backbone.fused_featurizer.blocks.19.mlp.fc2.bias": "model-00001-of-00004.safetensors",
798
+ "vision_backbone.fused_featurizer.blocks.19.mlp.fc2.weight": "model-00001-of-00004.safetensors",
799
+ "vision_backbone.fused_featurizer.blocks.19.norm1.bias": "model-00001-of-00004.safetensors",
800
+ "vision_backbone.fused_featurizer.blocks.19.norm1.weight": "model-00001-of-00004.safetensors",
801
+ "vision_backbone.fused_featurizer.blocks.19.norm2.bias": "model-00001-of-00004.safetensors",
802
+ "vision_backbone.fused_featurizer.blocks.19.norm2.weight": "model-00001-of-00004.safetensors",
803
+ "vision_backbone.fused_featurizer.blocks.2.attn.proj.bias": "model-00001-of-00004.safetensors",
804
+ "vision_backbone.fused_featurizer.blocks.2.attn.proj.weight": "model-00001-of-00004.safetensors",
805
+ "vision_backbone.fused_featurizer.blocks.2.attn.qkv.bias": "model-00001-of-00004.safetensors",
806
+ "vision_backbone.fused_featurizer.blocks.2.attn.qkv.weight": "model-00001-of-00004.safetensors",
807
+ "vision_backbone.fused_featurizer.blocks.2.mlp.fc1.bias": "model-00001-of-00004.safetensors",
808
+ "vision_backbone.fused_featurizer.blocks.2.mlp.fc1.weight": "model-00001-of-00004.safetensors",
809
+ "vision_backbone.fused_featurizer.blocks.2.mlp.fc2.bias": "model-00001-of-00004.safetensors",
810
+ "vision_backbone.fused_featurizer.blocks.2.mlp.fc2.weight": "model-00001-of-00004.safetensors",
811
+ "vision_backbone.fused_featurizer.blocks.2.norm1.bias": "model-00001-of-00004.safetensors",
812
+ "vision_backbone.fused_featurizer.blocks.2.norm1.weight": "model-00001-of-00004.safetensors",
813
+ "vision_backbone.fused_featurizer.blocks.2.norm2.bias": "model-00001-of-00004.safetensors",
814
+ "vision_backbone.fused_featurizer.blocks.2.norm2.weight": "model-00001-of-00004.safetensors",
815
+ "vision_backbone.fused_featurizer.blocks.20.attn.proj.bias": "model-00001-of-00004.safetensors",
816
+ "vision_backbone.fused_featurizer.blocks.20.attn.proj.weight": "model-00001-of-00004.safetensors",
817
+ "vision_backbone.fused_featurizer.blocks.20.attn.qkv.bias": "model-00001-of-00004.safetensors",
818
+ "vision_backbone.fused_featurizer.blocks.20.attn.qkv.weight": "model-00001-of-00004.safetensors",
819
+ "vision_backbone.fused_featurizer.blocks.20.mlp.fc1.bias": "model-00001-of-00004.safetensors",
820
+ "vision_backbone.fused_featurizer.blocks.20.mlp.fc1.weight": "model-00001-of-00004.safetensors",
821
+ "vision_backbone.fused_featurizer.blocks.20.mlp.fc2.bias": "model-00001-of-00004.safetensors",
822
+ "vision_backbone.fused_featurizer.blocks.20.mlp.fc2.weight": "model-00001-of-00004.safetensors",
823
+ "vision_backbone.fused_featurizer.blocks.20.norm1.bias": "model-00001-of-00004.safetensors",
824
+ "vision_backbone.fused_featurizer.blocks.20.norm1.weight": "model-00001-of-00004.safetensors",
825
+ "vision_backbone.fused_featurizer.blocks.20.norm2.bias": "model-00001-of-00004.safetensors",
826
+ "vision_backbone.fused_featurizer.blocks.20.norm2.weight": "model-00001-of-00004.safetensors",
827
+ "vision_backbone.fused_featurizer.blocks.21.attn.proj.bias": "model-00001-of-00004.safetensors",
828
+ "vision_backbone.fused_featurizer.blocks.21.attn.proj.weight": "model-00001-of-00004.safetensors",
829
+ "vision_backbone.fused_featurizer.blocks.21.attn.qkv.bias": "model-00001-of-00004.safetensors",
830
+ "vision_backbone.fused_featurizer.blocks.21.attn.qkv.weight": "model-00001-of-00004.safetensors",
831
+ "vision_backbone.fused_featurizer.blocks.21.mlp.fc1.bias": "model-00001-of-00004.safetensors",
832
+ "vision_backbone.fused_featurizer.blocks.21.mlp.fc1.weight": "model-00001-of-00004.safetensors",
833
+ "vision_backbone.fused_featurizer.blocks.21.mlp.fc2.bias": "model-00001-of-00004.safetensors",
834
+ "vision_backbone.fused_featurizer.blocks.21.mlp.fc2.weight": "model-00001-of-00004.safetensors",
835
+ "vision_backbone.fused_featurizer.blocks.21.norm1.bias": "model-00001-of-00004.safetensors",
836
+ "vision_backbone.fused_featurizer.blocks.21.norm1.weight": "model-00001-of-00004.safetensors",
837
+ "vision_backbone.fused_featurizer.blocks.21.norm2.bias": "model-00001-of-00004.safetensors",
838
+ "vision_backbone.fused_featurizer.blocks.21.norm2.weight": "model-00001-of-00004.safetensors",
839
+ "vision_backbone.fused_featurizer.blocks.22.attn.proj.bias": "model-00001-of-00004.safetensors",
840
+ "vision_backbone.fused_featurizer.blocks.22.attn.proj.weight": "model-00001-of-00004.safetensors",
841
+ "vision_backbone.fused_featurizer.blocks.22.attn.qkv.bias": "model-00001-of-00004.safetensors",
842
+ "vision_backbone.fused_featurizer.blocks.22.attn.qkv.weight": "model-00001-of-00004.safetensors",
843
+ "vision_backbone.fused_featurizer.blocks.22.mlp.fc1.bias": "model-00001-of-00004.safetensors",
844
+ "vision_backbone.fused_featurizer.blocks.22.mlp.fc1.weight": "model-00001-of-00004.safetensors",
845
+ "vision_backbone.fused_featurizer.blocks.22.mlp.fc2.bias": "model-00001-of-00004.safetensors",
846
+ "vision_backbone.fused_featurizer.blocks.22.mlp.fc2.weight": "model-00001-of-00004.safetensors",
847
+ "vision_backbone.fused_featurizer.blocks.22.norm1.bias": "model-00001-of-00004.safetensors",
848
+ "vision_backbone.fused_featurizer.blocks.22.norm1.weight": "model-00001-of-00004.safetensors",
849
+ "vision_backbone.fused_featurizer.blocks.22.norm2.bias": "model-00001-of-00004.safetensors",
850
+ "vision_backbone.fused_featurizer.blocks.22.norm2.weight": "model-00001-of-00004.safetensors",
851
+ "vision_backbone.fused_featurizer.blocks.23.attn.proj.bias": "model-00001-of-00004.safetensors",
852
+ "vision_backbone.fused_featurizer.blocks.23.attn.proj.weight": "model-00001-of-00004.safetensors",
853
+ "vision_backbone.fused_featurizer.blocks.23.attn.qkv.bias": "model-00001-of-00004.safetensors",
854
+ "vision_backbone.fused_featurizer.blocks.23.attn.qkv.weight": "model-00001-of-00004.safetensors",
855
+ "vision_backbone.fused_featurizer.blocks.23.mlp.fc1.bias": "model-00001-of-00004.safetensors",
856
+ "vision_backbone.fused_featurizer.blocks.23.mlp.fc1.weight": "model-00001-of-00004.safetensors",
857
+ "vision_backbone.fused_featurizer.blocks.23.mlp.fc2.bias": "model-00001-of-00004.safetensors",
858
+ "vision_backbone.fused_featurizer.blocks.23.mlp.fc2.weight": "model-00001-of-00004.safetensors",
859
+ "vision_backbone.fused_featurizer.blocks.23.norm1.bias": "model-00001-of-00004.safetensors",
860
+ "vision_backbone.fused_featurizer.blocks.23.norm1.weight": "model-00001-of-00004.safetensors",
861
+ "vision_backbone.fused_featurizer.blocks.23.norm2.bias": "model-00001-of-00004.safetensors",
862
+ "vision_backbone.fused_featurizer.blocks.23.norm2.weight": "model-00001-of-00004.safetensors",
863
+ "vision_backbone.fused_featurizer.blocks.24.attn.proj.bias": "model-00001-of-00004.safetensors",
864
+ "vision_backbone.fused_featurizer.blocks.24.attn.proj.weight": "model-00001-of-00004.safetensors",
865
+ "vision_backbone.fused_featurizer.blocks.24.attn.qkv.bias": "model-00001-of-00004.safetensors",
866
+ "vision_backbone.fused_featurizer.blocks.24.attn.qkv.weight": "model-00001-of-00004.safetensors",
867
+ "vision_backbone.fused_featurizer.blocks.24.mlp.fc1.bias": "model-00001-of-00004.safetensors",
868
+ "vision_backbone.fused_featurizer.blocks.24.mlp.fc1.weight": "model-00001-of-00004.safetensors",
869
+ "vision_backbone.fused_featurizer.blocks.24.mlp.fc2.bias": "model-00001-of-00004.safetensors",
870
+ "vision_backbone.fused_featurizer.blocks.24.mlp.fc2.weight": "model-00001-of-00004.safetensors",
871
+ "vision_backbone.fused_featurizer.blocks.24.norm1.bias": "model-00001-of-00004.safetensors",
872
+ "vision_backbone.fused_featurizer.blocks.24.norm1.weight": "model-00001-of-00004.safetensors",
873
+ "vision_backbone.fused_featurizer.blocks.24.norm2.bias": "model-00001-of-00004.safetensors",
874
+ "vision_backbone.fused_featurizer.blocks.24.norm2.weight": "model-00001-of-00004.safetensors",
875
+ "vision_backbone.fused_featurizer.blocks.25.attn.proj.bias": "model-00001-of-00004.safetensors",
876
+ "vision_backbone.fused_featurizer.blocks.25.attn.proj.weight": "model-00001-of-00004.safetensors",
877
+ "vision_backbone.fused_featurizer.blocks.25.attn.qkv.bias": "model-00001-of-00004.safetensors",
878
+ "vision_backbone.fused_featurizer.blocks.25.attn.qkv.weight": "model-00001-of-00004.safetensors",
879
+ "vision_backbone.fused_featurizer.blocks.25.mlp.fc1.bias": "model-00001-of-00004.safetensors",
880
+ "vision_backbone.fused_featurizer.blocks.25.mlp.fc1.weight": "model-00001-of-00004.safetensors",
881
+ "vision_backbone.fused_featurizer.blocks.25.mlp.fc2.bias": "model-00001-of-00004.safetensors",
882
+ "vision_backbone.fused_featurizer.blocks.25.mlp.fc2.weight": "model-00001-of-00004.safetensors",
883
+ "vision_backbone.fused_featurizer.blocks.25.norm1.bias": "model-00001-of-00004.safetensors",
884
+ "vision_backbone.fused_featurizer.blocks.25.norm1.weight": "model-00001-of-00004.safetensors",
885
+ "vision_backbone.fused_featurizer.blocks.25.norm2.bias": "model-00001-of-00004.safetensors",
886
+ "vision_backbone.fused_featurizer.blocks.25.norm2.weight": "model-00001-of-00004.safetensors",
887
+ "vision_backbone.fused_featurizer.blocks.26.attn.proj.bias": "model-00001-of-00004.safetensors",
888
+ "vision_backbone.fused_featurizer.blocks.26.attn.proj.weight": "model-00001-of-00004.safetensors",
889
+ "vision_backbone.fused_featurizer.blocks.26.attn.qkv.bias": "model-00001-of-00004.safetensors",
890
+ "vision_backbone.fused_featurizer.blocks.26.attn.qkv.weight": "model-00001-of-00004.safetensors",
891
+ "vision_backbone.fused_featurizer.blocks.26.mlp.fc1.bias": "model-00001-of-00004.safetensors",
892
+ "vision_backbone.fused_featurizer.blocks.26.mlp.fc1.weight": "model-00001-of-00004.safetensors",
893
+ "vision_backbone.fused_featurizer.blocks.26.mlp.fc2.bias": "model-00001-of-00004.safetensors",
894
+ "vision_backbone.fused_featurizer.blocks.26.mlp.fc2.weight": "model-00001-of-00004.safetensors",
895
+ "vision_backbone.fused_featurizer.blocks.26.norm1.bias": "model-00001-of-00004.safetensors",
896
+ "vision_backbone.fused_featurizer.blocks.26.norm1.weight": "model-00001-of-00004.safetensors",
897
+ "vision_backbone.fused_featurizer.blocks.26.norm2.bias": "model-00001-of-00004.safetensors",
898
+ "vision_backbone.fused_featurizer.blocks.26.norm2.weight": "model-00001-of-00004.safetensors",
899
+ "vision_backbone.fused_featurizer.blocks.3.attn.proj.bias": "model-00001-of-00004.safetensors",
900
+ "vision_backbone.fused_featurizer.blocks.3.attn.proj.weight": "model-00001-of-00004.safetensors",
901
+ "vision_backbone.fused_featurizer.blocks.3.attn.qkv.bias": "model-00001-of-00004.safetensors",
902
+ "vision_backbone.fused_featurizer.blocks.3.attn.qkv.weight": "model-00001-of-00004.safetensors",
903
+ "vision_backbone.fused_featurizer.blocks.3.mlp.fc1.bias": "model-00001-of-00004.safetensors",
904
+ "vision_backbone.fused_featurizer.blocks.3.mlp.fc1.weight": "model-00001-of-00004.safetensors",
905
+ "vision_backbone.fused_featurizer.blocks.3.mlp.fc2.bias": "model-00001-of-00004.safetensors",
906
+ "vision_backbone.fused_featurizer.blocks.3.mlp.fc2.weight": "model-00001-of-00004.safetensors",
907
+ "vision_backbone.fused_featurizer.blocks.3.norm1.bias": "model-00001-of-00004.safetensors",
908
+ "vision_backbone.fused_featurizer.blocks.3.norm1.weight": "model-00001-of-00004.safetensors",
909
+ "vision_backbone.fused_featurizer.blocks.3.norm2.bias": "model-00001-of-00004.safetensors",
910
+ "vision_backbone.fused_featurizer.blocks.3.norm2.weight": "model-00001-of-00004.safetensors",
911
+ "vision_backbone.fused_featurizer.blocks.4.attn.proj.bias": "model-00001-of-00004.safetensors",
912
+ "vision_backbone.fused_featurizer.blocks.4.attn.proj.weight": "model-00001-of-00004.safetensors",
913
+ "vision_backbone.fused_featurizer.blocks.4.attn.qkv.bias": "model-00001-of-00004.safetensors",
914
+ "vision_backbone.fused_featurizer.blocks.4.attn.qkv.weight": "model-00001-of-00004.safetensors",
915
+ "vision_backbone.fused_featurizer.blocks.4.mlp.fc1.bias": "model-00001-of-00004.safetensors",
916
+ "vision_backbone.fused_featurizer.blocks.4.mlp.fc1.weight": "model-00001-of-00004.safetensors",
917
+ "vision_backbone.fused_featurizer.blocks.4.mlp.fc2.bias": "model-00001-of-00004.safetensors",
918
+ "vision_backbone.fused_featurizer.blocks.4.mlp.fc2.weight": "model-00001-of-00004.safetensors",
919
+ "vision_backbone.fused_featurizer.blocks.4.norm1.bias": "model-00001-of-00004.safetensors",
920
+ "vision_backbone.fused_featurizer.blocks.4.norm1.weight": "model-00001-of-00004.safetensors",
921
+ "vision_backbone.fused_featurizer.blocks.4.norm2.bias": "model-00001-of-00004.safetensors",
922
+ "vision_backbone.fused_featurizer.blocks.4.norm2.weight": "model-00001-of-00004.safetensors",
923
+ "vision_backbone.fused_featurizer.blocks.5.attn.proj.bias": "model-00001-of-00004.safetensors",
924
+ "vision_backbone.fused_featurizer.blocks.5.attn.proj.weight": "model-00001-of-00004.safetensors",
925
+ "vision_backbone.fused_featurizer.blocks.5.attn.qkv.bias": "model-00001-of-00004.safetensors",
926
+ "vision_backbone.fused_featurizer.blocks.5.attn.qkv.weight": "model-00001-of-00004.safetensors",
927
+ "vision_backbone.fused_featurizer.blocks.5.mlp.fc1.bias": "model-00001-of-00004.safetensors",
928
+ "vision_backbone.fused_featurizer.blocks.5.mlp.fc1.weight": "model-00001-of-00004.safetensors",
929
+ "vision_backbone.fused_featurizer.blocks.5.mlp.fc2.bias": "model-00001-of-00004.safetensors",
930
+ "vision_backbone.fused_featurizer.blocks.5.mlp.fc2.weight": "model-00001-of-00004.safetensors",
931
+ "vision_backbone.fused_featurizer.blocks.5.norm1.bias": "model-00001-of-00004.safetensors",
932
+ "vision_backbone.fused_featurizer.blocks.5.norm1.weight": "model-00001-of-00004.safetensors",
933
+ "vision_backbone.fused_featurizer.blocks.5.norm2.bias": "model-00001-of-00004.safetensors",
934
+ "vision_backbone.fused_featurizer.blocks.5.norm2.weight": "model-00001-of-00004.safetensors",
935
+ "vision_backbone.fused_featurizer.blocks.6.attn.proj.bias": "model-00001-of-00004.safetensors",
936
+ "vision_backbone.fused_featurizer.blocks.6.attn.proj.weight": "model-00001-of-00004.safetensors",
937
+ "vision_backbone.fused_featurizer.blocks.6.attn.qkv.bias": "model-00001-of-00004.safetensors",
938
+ "vision_backbone.fused_featurizer.blocks.6.attn.qkv.weight": "model-00001-of-00004.safetensors",
939
+ "vision_backbone.fused_featurizer.blocks.6.mlp.fc1.bias": "model-00001-of-00004.safetensors",
940
+ "vision_backbone.fused_featurizer.blocks.6.mlp.fc1.weight": "model-00001-of-00004.safetensors",
941
+ "vision_backbone.fused_featurizer.blocks.6.mlp.fc2.bias": "model-00001-of-00004.safetensors",
942
+ "vision_backbone.fused_featurizer.blocks.6.mlp.fc2.weight": "model-00001-of-00004.safetensors",
943
+ "vision_backbone.fused_featurizer.blocks.6.norm1.bias": "model-00001-of-00004.safetensors",
944
+ "vision_backbone.fused_featurizer.blocks.6.norm1.weight": "model-00001-of-00004.safetensors",
945
+ "vision_backbone.fused_featurizer.blocks.6.norm2.bias": "model-00001-of-00004.safetensors",
946
+ "vision_backbone.fused_featurizer.blocks.6.norm2.weight": "model-00001-of-00004.safetensors",
947
+ "vision_backbone.fused_featurizer.blocks.7.attn.proj.bias": "model-00001-of-00004.safetensors",
948
+ "vision_backbone.fused_featurizer.blocks.7.attn.proj.weight": "model-00001-of-00004.safetensors",
949
+ "vision_backbone.fused_featurizer.blocks.7.attn.qkv.bias": "model-00001-of-00004.safetensors",
950
+ "vision_backbone.fused_featurizer.blocks.7.attn.qkv.weight": "model-00001-of-00004.safetensors",
951
+ "vision_backbone.fused_featurizer.blocks.7.mlp.fc1.bias": "model-00001-of-00004.safetensors",
952
+ "vision_backbone.fused_featurizer.blocks.7.mlp.fc1.weight": "model-00001-of-00004.safetensors",
953
+ "vision_backbone.fused_featurizer.blocks.7.mlp.fc2.bias": "model-00001-of-00004.safetensors",
954
+ "vision_backbone.fused_featurizer.blocks.7.mlp.fc2.weight": "model-00001-of-00004.safetensors",
955
+ "vision_backbone.fused_featurizer.blocks.7.norm1.bias": "model-00001-of-00004.safetensors",
956
+ "vision_backbone.fused_featurizer.blocks.7.norm1.weight": "model-00001-of-00004.safetensors",
957
+ "vision_backbone.fused_featurizer.blocks.7.norm2.bias": "model-00001-of-00004.safetensors",
958
+ "vision_backbone.fused_featurizer.blocks.7.norm2.weight": "model-00001-of-00004.safetensors",
959
+ "vision_backbone.fused_featurizer.blocks.8.attn.proj.bias": "model-00001-of-00004.safetensors",
960
+ "vision_backbone.fused_featurizer.blocks.8.attn.proj.weight": "model-00001-of-00004.safetensors",
961
+ "vision_backbone.fused_featurizer.blocks.8.attn.qkv.bias": "model-00001-of-00004.safetensors",
962
+ "vision_backbone.fused_featurizer.blocks.8.attn.qkv.weight": "model-00001-of-00004.safetensors",
963
+ "vision_backbone.fused_featurizer.blocks.8.mlp.fc1.bias": "model-00001-of-00004.safetensors",
964
+ "vision_backbone.fused_featurizer.blocks.8.mlp.fc1.weight": "model-00001-of-00004.safetensors",
965
+ "vision_backbone.fused_featurizer.blocks.8.mlp.fc2.bias": "model-00001-of-00004.safetensors",
966
+ "vision_backbone.fused_featurizer.blocks.8.mlp.fc2.weight": "model-00001-of-00004.safetensors",
967
+ "vision_backbone.fused_featurizer.blocks.8.norm1.bias": "model-00001-of-00004.safetensors",
968
+ "vision_backbone.fused_featurizer.blocks.8.norm1.weight": "model-00001-of-00004.safetensors",
969
+ "vision_backbone.fused_featurizer.blocks.8.norm2.bias": "model-00001-of-00004.safetensors",
970
+ "vision_backbone.fused_featurizer.blocks.8.norm2.weight": "model-00001-of-00004.safetensors",
971
+ "vision_backbone.fused_featurizer.blocks.9.attn.proj.bias": "model-00001-of-00004.safetensors",
972
+ "vision_backbone.fused_featurizer.blocks.9.attn.proj.weight": "model-00001-of-00004.safetensors",
973
+ "vision_backbone.fused_featurizer.blocks.9.attn.qkv.bias": "model-00001-of-00004.safetensors",
974
+ "vision_backbone.fused_featurizer.blocks.9.attn.qkv.weight": "model-00001-of-00004.safetensors",
975
+ "vision_backbone.fused_featurizer.blocks.9.mlp.fc1.bias": "model-00001-of-00004.safetensors",
976
+ "vision_backbone.fused_featurizer.blocks.9.mlp.fc1.weight": "model-00001-of-00004.safetensors",
977
+ "vision_backbone.fused_featurizer.blocks.9.mlp.fc2.bias": "model-00001-of-00004.safetensors",
978
+ "vision_backbone.fused_featurizer.blocks.9.mlp.fc2.weight": "model-00001-of-00004.safetensors",
979
+ "vision_backbone.fused_featurizer.blocks.9.norm1.bias": "model-00001-of-00004.safetensors",
980
+ "vision_backbone.fused_featurizer.blocks.9.norm1.weight": "model-00001-of-00004.safetensors",
981
+ "vision_backbone.fused_featurizer.blocks.9.norm2.bias": "model-00001-of-00004.safetensors",
982
+ "vision_backbone.fused_featurizer.blocks.9.norm2.weight": "model-00001-of-00004.safetensors",
983
+ "vision_backbone.fused_featurizer.norm.bias": "model-00001-of-00004.safetensors",
984
+ "vision_backbone.fused_featurizer.norm.weight": "model-00001-of-00004.safetensors",
985
+ "vision_backbone.fused_featurizer.patch_embed.proj.bias": "model-00001-of-00004.safetensors",
986
+ "vision_backbone.fused_featurizer.patch_embed.proj.weight": "model-00001-of-00004.safetensors",
987
+ "vision_backbone.fused_featurizer.pos_embed": "model-00001-of-00004.safetensors"
988
+ }
989
+ }
models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/modeling_prismatic.py ADDED
@@ -0,0 +1,2117 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """
2
+ modeling_prismatic.py
3
+
4
+ Core HuggingFace-style PrismaticPreTrainedModel and PrismaticForConditionalGeneration class definitions.
5
+ Inherits from the default `transformers.PretrainedModel`. Meant to be standalone and self-contained,
6
+ but exactly replicate the logic in `prismatic.models.vlms.prismatic.py`.
7
+ """
8
+
9
+ import logging
10
+ from dataclasses import dataclass
11
+ from functools import partial
12
+ from typing import Any, Callable, ClassVar, Dict, List, Optional, Tuple, Union
13
+
14
+ import numpy as np
15
+ import timm
16
+ import tokenizers
17
+ import torch
18
+ import torch.nn as nn
19
+ import transformers
20
+ from timm.models.vision_transformer import LayerScale
21
+ from transformers import AutoModelForCausalLM, PretrainedConfig, PreTrainedModel
22
+ from transformers.cache_utils import DynamicCache
23
+ from transformers.modeling_outputs import ModelOutput
24
+
25
+ from prismatic.discrete_flow import (
26
+ mask_schedule,
27
+ parallel_decode,
28
+ )
29
+ from prismatic.training.train_utils import (
30
+ get_current_action_mask,
31
+ get_next_actions_mask,
32
+ )
33
+ from prismatic.vla.constants import (
34
+ ACTION_DIM,
35
+ ACTION_PROPRIO_NORMALIZATION_TYPE,
36
+ ACTION_TOKEN_BEGIN_IDX,
37
+ IGNORE_INDEX,
38
+ NUM_ACTIONS_CHUNK,
39
+ STOP_INDEX,
40
+ NormalizationType,
41
+ )
42
+
43
+ from .configuration_prismatic import OpenVLAConfig, PrismaticConfig
44
+
45
+ # Set up logger
46
+ logger = logging.getLogger(__name__)
47
+
48
+
49
+ # === Utility Functions for Monkey-Patching ===
50
+ def unpack_tuple(fn: Callable[[Any], Tuple[Any]]) -> Callable[[Any], Any]:
51
+ def wrapper(*args: Any, **kwargs: Any) -> Any:
52
+ result = fn(*args, **kwargs)
53
+ return result[0] if isinstance(result, tuple) else result
54
+
55
+ return wrapper
56
+
57
+
58
+ # HF Transformers overwrites parameters with names containing `gamma`; we're going to patch VisionBackbone.LayerScale.
59
+ # =>> TIMM :: https://github.com/huggingface/pytorch-image-models/blob/main/timm/models/vision_transformer.py#L109
60
+ # =>> Transformers :: https://github.com/huggingface/transformers/blob/main/src/transformers/modeling_utils.py#L3960
61
+ def _ls_new_forward(self, x: torch.Tensor) -> torch.Tensor:
62
+ return x.mul_(self.scale_factor) if self.inplace else x * self.scale_factor
63
+
64
+
65
+ def ls_apply_patch(ls_module: LayerScale):
66
+ ls_module.scale_factor = nn.Parameter(ls_module.gamma.clone())
67
+ ls_module.forward = _ls_new_forward.__get__(ls_module, LayerScale)
68
+ del ls_module.gamma
69
+
70
+
71
+ # === Prismatic Vision Backbone (nn.Module) Definitions (w/ Fused Backbone Support) ===
72
+ class PrismaticVisionBackbone(nn.Module):
73
+ """
74
+ Vision backbone for Prismatic models that handles image feature extraction.
75
+
76
+ Supports both single backbone (e.g., SigLIP) and fused backbone (e.g., SigLIP + DINOv2) configurations.
77
+ For fused backbones, features from both models are concatenated along the feature dimension.
78
+ """
79
+
80
+ def __init__(
81
+ self,
82
+ use_fused_vision_backbone: bool,
83
+ image_sizes: List[int],
84
+ timm_model_ids: List[str],
85
+ timm_override_act_layers: List[Optional[str]],
86
+ ) -> None:
87
+ """
88
+ Initialize the vision backbone.
89
+
90
+ Args:
91
+ use_fused_vision_backbone: Whether to use two backbones and fuse their features
92
+ image_sizes: List of image sizes for each backbone
93
+ timm_model_ids: List of TIMM model IDs to use for each backbone
94
+ timm_override_act_layers: List of activation layer overrides for each backbone
95
+ """
96
+ super().__init__()
97
+ self.use_fused_vision_backbone = use_fused_vision_backbone
98
+ self.num_images_in_input = 1 # Default value, can be overridden later
99
+
100
+ # Validate number of (fused) vision backbones
101
+ if len(timm_model_ids) > 2:
102
+ raise ValueError("Prismatic models only support up to 2 (fused) vision backbones!")
103
+
104
+ # Create primary featurizer
105
+ self.featurizer = self._create_featurizer(
106
+ model_id=timm_model_ids[0], img_size=image_sizes[0], act_layer=timm_override_act_layers[0]
107
+ )
108
+ self.embed_dim = self.featurizer.embed_dim
109
+
110
+ # Create secondary featurizer if using fused backbone
111
+ if self.use_fused_vision_backbone:
112
+ self.fused_featurizer = self._create_featurizer(
113
+ model_id=timm_model_ids[1], img_size=image_sizes[1], act_layer=timm_override_act_layers[1]
114
+ )
115
+ self.embed_dim += self.fused_featurizer.embed_dim
116
+
117
+ # Patch LayerScale modules for HF compatibility
118
+ self._patch_layer_scales()
119
+
120
+ def _create_featurizer(self, model_id: str, img_size: int, act_layer: Optional[str]) -> nn.Module:
121
+ """
122
+ Create a TIMM-based featurizer model with appropriate configurations.
123
+
124
+ Args:
125
+ model_id: The TIMM model ID to load
126
+ img_size: Input image size for the model
127
+ act_layer: Override for the activation layer type
128
+
129
+ Returns:
130
+ A configured featurizer model
131
+ """
132
+ featurizer = timm.create_model(
133
+ model_id,
134
+ pretrained=False,
135
+ num_classes=0,
136
+ img_size=img_size,
137
+ act_layer=act_layer,
138
+ )
139
+
140
+ # Monkey-patch the forward function to extract the second-to-last layer features
141
+ num_blocks = len(featurizer.blocks)
142
+ featurizer.forward = unpack_tuple(partial(featurizer.get_intermediate_layers, n={num_blocks - 2}))
143
+
144
+ return featurizer
145
+
146
+ def _patch_layer_scales(self) -> None:
147
+ """
148
+ Patch all LayerScale modules to be compatible with HF's parameter naming.
149
+
150
+ HF Transformers overwrites parameters with names containing 'gamma',
151
+ so we need to rename and modify the forward method.
152
+ """
153
+ # Patch primary featurizer
154
+ for module in self.featurizer.modules():
155
+ if isinstance(module, LayerScale):
156
+ ls_apply_patch(module)
157
+
158
+ # Patch secondary featurizer if it exists
159
+ if self.use_fused_vision_backbone:
160
+ for module in self.fused_featurizer.modules():
161
+ if isinstance(module, LayerScale):
162
+ ls_apply_patch(module)
163
+
164
+ def get_num_patches(self) -> int:
165
+ """
166
+ Returns the number of vision patches output by the vision backbone.
167
+
168
+ Returns:
169
+ Number of patches per image
170
+ """
171
+ return self.featurizer.patch_embed.num_patches
172
+
173
+ def get_num_images_in_input(self) -> int:
174
+ """
175
+ Returns the number of input images for the vision backbone.
176
+
177
+ Returns:
178
+ Number of images expected in the input
179
+ """
180
+ return self.num_images_in_input
181
+
182
+ def set_num_images_in_input(self, num_images_in_input: int) -> None:
183
+ """
184
+ Sets the number of input images for the vision backbone.
185
+
186
+ Args:
187
+ num_images_in_input: Number of images to expect in the input
188
+ """
189
+ self.num_images_in_input = num_images_in_input
190
+
191
+ def forward(self, pixel_values: torch.Tensor) -> torch.Tensor:
192
+ """
193
+ Implements the forward pass for the vision backbone.
194
+
195
+ If `self.use_fused_vision_backbone == True`, uses both SigLIP and DINOv2 transformers to extract visual features
196
+ (otherwise uses SigLIP only). Allows multi-image inputs (but only for fused vision backbone).
197
+
198
+ Args:
199
+ pixel_values (torch.Tensor): Pixels for input image(s), (B, C, H, W).
200
+ """
201
+ if self.num_images_in_input == 1:
202
+ if not self.use_fused_vision_backbone:
203
+ return self.featurizer(pixel_values)
204
+
205
+ # Split `pixel_values :: [bsz, 2 * 3, resolution, resolution]` =>> featurize =>> channel stack
206
+ img, img_fused = torch.split(pixel_values, [3, 3], dim=1)
207
+ patches, patches_fused = self.featurizer(img), self.fused_featurizer(img_fused)
208
+
209
+ return torch.cat([patches, patches_fused], dim=2)
210
+
211
+ else:
212
+ assert self.use_fused_vision_backbone, "Multi-image inputs require using fused backbone!"
213
+
214
+ # Split `pixel_values` into individual images (each with 6 channels: 3 for SigLIP + 3 for DINOv2)
215
+ images = torch.split(pixel_values, [6] * self.num_images_in_input, dim=1)
216
+
217
+ # Process each image and collect patches
218
+ all_patches = []
219
+ for img in images:
220
+ # Split each image further into two stacks of channels (each with 3 channels)
221
+ img_regular, img_fused = torch.split(img, [3, 3], dim=1)
222
+
223
+ # Get patches from both SigLIP and DINOv2 vision transformers
224
+ patches = self.featurizer(img_regular)
225
+ patches_fused = self.fused_featurizer(img_fused)
226
+
227
+ # Concatenate SigLIP and DINOv2 patches along the hidden dimension
228
+ combined_patches = torch.cat([patches, patches_fused], dim=2)
229
+ all_patches.append(combined_patches)
230
+
231
+ # Concatenate all patches along the patch dimension
232
+ return torch.cat(all_patches, dim=1)
233
+
234
+
235
+ # === Prismatic Projector (nn.Module) Definitions ===
236
+ class PrismaticProjector(nn.Module):
237
+ def __init__(self, use_fused_vision_backbone: bool, vision_dim: int, llm_dim: int) -> None:
238
+ super().__init__()
239
+ self.use_fused_vision_backbone = use_fused_vision_backbone
240
+ self.vision_dim, self.llm_dim = vision_dim, llm_dim
241
+
242
+ # Switch on `use_fused_vision_backbone` =>> use slightly different MLPs and projection factors!
243
+ if not self.use_fused_vision_backbone:
244
+ self.fc1 = nn.Linear(self.vision_dim, self.llm_dim, bias=True)
245
+ self.fc2 = nn.Linear(self.llm_dim, self.llm_dim, bias=True)
246
+ self.act_fn1 = nn.GELU()
247
+ else:
248
+ initial_projection_dim = 4 * vision_dim
249
+ self.fc1 = nn.Linear(self.vision_dim, initial_projection_dim, bias=True)
250
+ self.fc2 = nn.Linear(initial_projection_dim, self.llm_dim, bias=True)
251
+ self.fc3 = nn.Linear(self.llm_dim, self.llm_dim, bias=True)
252
+ self.act_fn1 = nn.GELU()
253
+ self.act_fn2 = nn.GELU()
254
+
255
+ def forward(self, img_patches: torch.Tensor) -> torch.Tensor:
256
+ if not self.use_fused_vision_backbone:
257
+ projected_features = self.fc1(img_patches)
258
+ projected_features = self.act_fn1(projected_features)
259
+ projected_features = self.fc2(projected_features)
260
+ else:
261
+ projected_features = self.fc1(img_patches)
262
+ projected_features = self.act_fn1(projected_features)
263
+ projected_features = self.fc2(projected_features)
264
+ projected_features = self.act_fn2(projected_features)
265
+ projected_features = self.fc3(projected_features)
266
+
267
+ return projected_features
268
+
269
+
270
+ # === Main HF Class Definitions ===
271
+ @dataclass
272
+ class PrismaticCausalLMOutputWithPast(ModelOutput):
273
+ """Base class for Prismatic casual (visually-conditioned) language model outputs; also exposes visual features."""
274
+
275
+ loss: Optional[torch.FloatTensor] = None
276
+ logits: torch.FloatTensor = None
277
+ past_key_values: Optional[Tuple[Tuple[torch.FloatTensor]]] = None
278
+ hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
279
+ attentions: Optional[Tuple[torch.FloatTensor]] = None
280
+
281
+ # Additions for VLMs
282
+ projector_features: Optional[torch.FloatTensor] = None
283
+
284
+ # Additions for Discrete Diffusion
285
+ labels: Optional[torch.LongTensor] = None
286
+
287
+
288
+ class PrismaticPreTrainedModel(PreTrainedModel):
289
+ config_class: PretrainedConfig = PrismaticConfig
290
+ base_model_prefix: str = "model"
291
+ supports_gradient_checkpointing: bool = True
292
+
293
+ _no_split_modules: ClassVar[List[str]] = ["PrismaticProjector"]
294
+ _skip_keys_device_placement: str = "past_key_values"
295
+ _supports_flash_attn_2: bool = True
296
+
297
+ def _init_weights(self, module: nn.Module) -> None:
298
+ # Important :: this HF ported version is *not* meant for training from scratch; only inference and fine-tuning!
299
+ # => As such, this init_weights code is not correct; if training VLMs from scratch, use the main codebase at
300
+ # https://github.com/TRI-ML/prismatic-vlms
301
+ std = (
302
+ self.config.initializer_range
303
+ if hasattr(self.config, "initializer_range")
304
+ else self.config.text_config.initializer_range
305
+ )
306
+
307
+ if hasattr(module, "class_embedding"):
308
+ module.class_embedding.data.normal_(mean=0.0, std=std)
309
+
310
+ if isinstance(module, (nn.Linear, nn.Conv2d)):
311
+ module.weight.data.normal_(mean=0.0, std=std)
312
+ if module.bias is not None:
313
+ module.bias.data.zero_()
314
+ elif isinstance(module, nn.Embedding):
315
+ module.weight.data.normal_(mean=0.0, std=std)
316
+ if module.padding_idx is not None:
317
+ module.weight.data[module.padding_idx].zero_()
318
+
319
+ @property
320
+ def _supports_sdpa(self) -> bool:
321
+ """Check LLM supports SDPA Attention"""
322
+ return self.language_model._supports_sdpa
323
+
324
+
325
+ class PrismaticForConditionalGeneration(PrismaticPreTrainedModel):
326
+ def __init__(self, config: PrismaticConfig) -> None:
327
+ super().__init__(config)
328
+
329
+ # [Validation] Lightweight Validate on `config` Fields + Dependency Versions
330
+ if config.use_fused_vision_backbone is None:
331
+ raise ValueError("Missing config field `use_fused_vision_backbone`")
332
+
333
+ if timm.__version__ not in {"0.9.10", "0.9.11", "0.9.12", "0.9.16"}:
334
+ raise NotImplementedError(
335
+ "TIMM Version must be >= 0.9.10 and < 1.0.0 (breaking); please raise a GitHub Issue "
336
+ "if you urgently need support for latest TIMM versions."
337
+ )
338
+
339
+ if (transformers.__version__ != "4.40.1") or (tokenizers.__version__ != "0.19.1"):
340
+ logger.warning(
341
+ f"Expected `transformers==4.40.1` and `tokenizers==0.19.1` but got "
342
+ f"`transformers=={transformers.__version__}` and `tokenizers=={tokenizers.__version__}`; "
343
+ f"there might be inference-time regressions due to dependency changes. If in doubt, please"
344
+ f"use the above versions."
345
+ )
346
+
347
+ # Instantiate PrismaticVisionBackbone (w/ Potential Fused Backbone)
348
+ self.vision_backbone = PrismaticVisionBackbone(
349
+ config.use_fused_vision_backbone, config.image_sizes, config.timm_model_ids, config.timm_override_act_layers
350
+ )
351
+
352
+ # Create Multimodal Projector
353
+ self.projector = PrismaticProjector(
354
+ config.use_fused_vision_backbone,
355
+ vision_dim=self.vision_backbone.embed_dim,
356
+ llm_dim=config.text_config.hidden_size,
357
+ )
358
+
359
+ # Instantiate LLM Backbone
360
+ self.language_model = AutoModelForCausalLM.from_config(
361
+ config.text_config, attn_implementation=config._attn_implementation
362
+ )
363
+ # config.text_config.vocab_size自行向上取整到64的整数倍, 32064, 还有空间,无需调整
364
+ # self.language_model.resize_token_embeddings(config.text_config.vocab_size)
365
+
366
+ self.use_discrete_diffusion = config.use_discrete_diffusion
367
+ self.use_d2f = getattr(config, "use_d2f", False)
368
+ self.mask_token_id = config.mask_token_id
369
+
370
+ self.vocab_size = config.text_config.vocab_size
371
+ self.pad_token_id = config.pad_token_id
372
+ self.llm_dim = config.text_config.hidden_size
373
+
374
+ # HF Boilerplate =>> initializes weights via `_init_weights()` and sets gradient checkpointing
375
+ self.post_init()
376
+
377
+ # === `PreTrainedModel` Boilerplate ===
378
+ def get_input_embeddings(self) -> nn.Module:
379
+ return self.language_model.get_input_embeddings()
380
+
381
+ def set_input_embeddings(self, value: nn.Module) -> None:
382
+ self.language_model.set_input_embeddings(value)
383
+
384
+ def get_output_embeddings(self) -> nn.Module:
385
+ return self.language_model.get_output_embeddings()
386
+
387
+ def set_output_embeddings(self, new_embeddings: nn.Module) -> None:
388
+ self.language_model.set_output_embeddings(new_embeddings)
389
+
390
+ def get_decoder(self) -> nn.Module:
391
+ return self.language_model.get_decoder()
392
+
393
+ def set_decoder(self, decoder: nn.Module) -> None:
394
+ self.language_model.set_decoder(decoder)
395
+
396
+ def tie_weights(self) -> None:
397
+ self.language_model.tie_weights() # Note: `Llama-2` and `Mistral` don't tie weights (no-op)
398
+
399
+ def resize_token_embeddings(
400
+ self, new_num_tokens: Optional[int] = None, pad_to_multiple_of: Optional[int] = None
401
+ ) -> nn.Embedding:
402
+ updated_embeddings = self.language_model.resize_token_embeddings(new_num_tokens, pad_to_multiple_of)
403
+
404
+ # Update config/instance variables
405
+ self.config.text_config.vocab_size = updated_embeddings.num_embeddings
406
+ self.vocab_size = updated_embeddings.num_embeddings
407
+
408
+ return updated_embeddings
409
+
410
+ def _replace_input_embeddings(self, input_embeddings, all_actions_mask, noisy_action_features):
411
+ """
412
+ Replace embeddings in input_embeddings at positions where all_actions_mask is True
413
+ with embeddings from noisy_action_features, using vectorized operations.
414
+
415
+ Args:
416
+ input_embeddings: Tensor of shape (B, S, D)
417
+ all_actions_mask: Boolean tensor of shape (B, S)
418
+ noisy_action_features: Tensor of shape (B, K, D) where K is the number of True values in mask per sample
419
+
420
+ Returns:
421
+ Modified input_embeddings tensor
422
+ """
423
+ # Clone input to avoid modifying the original tensor
424
+ new_input_embeddings = input_embeddings.clone()
425
+
426
+ # Create a tensor with the same shape of input_embeddings to hold the noisy action features
427
+ repositioned_noisy_action_features = torch.zeros_like(input_embeddings)
428
+
429
+ # Create batch indices for splicing
430
+ batch_indices = torch.arange(input_embeddings.shape[0], device=input_embeddings.device)
431
+ batch_indices = batch_indices.unsqueeze(1).expand(-1, noisy_action_features.shape[1])
432
+
433
+ # Get indices where mask is True for each sample
434
+ masked_indices = torch.stack([torch.where(mask)[0] for mask in all_actions_mask])
435
+
436
+ # Move the noisy action features into their correct positions
437
+ repositioned_noisy_action_features[batch_indices, masked_indices] = noisy_action_features
438
+
439
+ # Combine original input embeddings and noisy action embeddings using the mask
440
+ new_input_embeddings = torch.where(
441
+ all_actions_mask.unsqueeze(-1), repositioned_noisy_action_features, new_input_embeddings
442
+ )
443
+
444
+ return new_input_embeddings
445
+
446
+ def _process_action_masks(self, labels):
447
+ """Helper to get action masks from labels"""
448
+ current_action_mask = get_current_action_mask(labels)
449
+ next_actions_mask = get_next_actions_mask(labels)
450
+ all_actions_mask = current_action_mask | next_actions_mask # (B, seq_len)
451
+ return all_actions_mask
452
+
453
+ def _process_vision_features(self, pixel_values, language_embeddings=None, use_film=False):
454
+ """Process vision features with optional FiLM conditioning"""
455
+ if use_film:
456
+ # FiLM: Infuse language inputs into visual features
457
+ patch_features = self.vision_backbone(pixel_values, language_embeddings) # (bsz, 256 * num_images, D)
458
+ else:
459
+ patch_features = self.vision_backbone(pixel_values) # (bsz, 256 * num_images, D)
460
+
461
+ # Project patch embeddings into language embedding space
462
+ return self.projector(patch_features)
463
+
464
+ def _process_proprio_features(self, projected_patch_embeddings, proprio, proprio_projector):
465
+ """Process proprioceptive features and append to vision features"""
466
+ if proprio_projector is not None and proprio is not None:
467
+ # projected_patch_embeddings: (bsz, num_patches * num_images, llm_dim)
468
+ # proprio: (bsz, proprio_dim) or (propro_dim,)
469
+ proprio = proprio.reshape(projected_patch_embeddings.shape[0], -1) # (bsz, proprio_dim)
470
+ proprio_features = proprio_projector(proprio) # (bsz, llm_dim)
471
+ proprio_features = proprio_features.unsqueeze(dim=1) # (bsz, 1, llm_dim)
472
+ # For simplicity, just append proprio token to the end of projected vision patch tokens
473
+ return torch.cat((projected_patch_embeddings, proprio_features), dim=1)
474
+ return projected_patch_embeddings
475
+
476
+ def _build_multimodal_attention(self, input_embeddings, projected_patch_embeddings, attention_mask):
477
+ """Build multimodal embeddings and attention mask"""
478
+ if attention_mask is not None and attention_mask.ndim == 4:
479
+ batch_size, _, text_length, _ = attention_mask.shape
480
+ patch_count = projected_patch_embeddings.shape[1]
481
+ multimodal_length = text_length + patch_count
482
+ multimodal_attention_mask = torch.zeros(
483
+ (batch_size, 1, multimodal_length, multimodal_length),
484
+ dtype=attention_mask.dtype,
485
+ device=attention_mask.device,
486
+ )
487
+ text_positions = torch.cat(
488
+ (
489
+ torch.zeros(1, dtype=torch.long, device=attention_mask.device),
490
+ torch.arange(1, text_length, device=attention_mask.device) + patch_count,
491
+ )
492
+ )
493
+ multimodal_attention_mask[:, :, text_positions[:, None], text_positions[None, :]] = attention_mask
494
+ patch_positions = torch.arange(1, 1 + patch_count, device=attention_mask.device)
495
+ # All text queries may condition on vision/proprio. Vision tokens remain independent of response tokens.
496
+ multimodal_attention_mask[:, :, text_positions[:, None], patch_positions[None, :]] = 1
497
+ multimodal_attention_mask[:, :, patch_positions[:, None], patch_positions[None, :]] = 1
498
+ multimodal_attention_mask[:, :, patch_positions, 0] = 1
499
+ multimodal_embeddings = torch.cat(
500
+ [input_embeddings[:, :1, :], projected_patch_embeddings, input_embeddings[:, 1:, :]], dim=1
501
+ )
502
+ return multimodal_embeddings, multimodal_attention_mask
503
+
504
+ # Update attention mask
505
+ projected_patch_attention_mask = None
506
+ if attention_mask is not None:
507
+ projected_patch_attention_mask = torch.full(
508
+ (projected_patch_embeddings.shape[0], projected_patch_embeddings.shape[1]),
509
+ fill_value=True,
510
+ dtype=attention_mask.dtype,
511
+ device=attention_mask.device,
512
+ )
513
+
514
+ # Build multimodal embeddings & attention mask; insert embeddings after <BOS> token (1:)
515
+ multimodal_embeddings = torch.cat(
516
+ [input_embeddings[:, :1, :], projected_patch_embeddings, input_embeddings[:, 1:, :]], dim=1
517
+ )
518
+
519
+ multimodal_attention_mask = None
520
+ if attention_mask is not None:
521
+ multimodal_attention_mask = torch.cat(
522
+ [attention_mask[:, :1], projected_patch_attention_mask, attention_mask[:, 1:]], dim=1
523
+ )
524
+
525
+ return multimodal_embeddings, multimodal_attention_mask
526
+
527
+ def _build_multimodal_labels(self, labels, projected_patch_embeddings):
528
+ """Build multimodal labels with IGNORE_INDEX for patch embeddings"""
529
+ if labels is not None:
530
+ projected_patch_labels = torch.full(
531
+ (projected_patch_embeddings.shape[0], projected_patch_embeddings.shape[1]),
532
+ fill_value=IGNORE_INDEX,
533
+ dtype=labels.dtype,
534
+ device=labels.device,
535
+ )
536
+ return torch.cat([labels[:, :1], projected_patch_labels, labels[:, 1:]], dim=1)
537
+ return None
538
+
539
+ def _get_eos_pos(self, all_actions_mask):
540
+ """Prepare loss mask for discrete diffusion"""
541
+ loss_mask_full = all_actions_mask.clone() # (B, seq_len)
542
+
543
+ # 找到每个序列中最后一个 True 的下标，并把下一个位置也置为 True
544
+ batch_size, seq_len = loss_mask_full.shape
545
+ # 将序列反向，再 argmax 就能找到“最后一个 True”在原序列中的位置
546
+ last_true = seq_len - 1 - torch.argmax(
547
+ loss_mask_full.flip(dims=[1]).int(), dim=1
548
+ ) # (B,)
549
+
550
+ # 下一个位置
551
+ next_pos = last_true + 1 # (B,)
552
+ # 只保留那些 next_pos < seq_len 的有效索引
553
+ valid = next_pos < seq_len # (B,) 布尔向量
554
+ batch_idx = torch.arange(batch_size, device=loss_mask_full.device)[valid]
555
+
556
+ # 把 EOS token 位置也设为 True
557
+ loss_mask_full[batch_idx, next_pos[valid]] = True
558
+ return next_pos[valid]
559
+
560
+ def apply_mask_diffusion(
561
+ self,
562
+ input_ids: torch.LongTensor, # (B, L)
563
+ input_embeddings: torch.Tensor, # (B, L, D)
564
+ labels: torch.LongTensor, # (B, L), target input_ids (有-100)
565
+ loss_mask_full: torch.BoolTensor, # (B, L), True 表示可 mask 的位置（非 padding、非语言 token）
566
+ mask_token_id: int,
567
+ eos_pos: Optional[torch.LongTensor] = None, # (B,), 可选的 EOS token 位置（如果有）
568
+ no_mask_token_prob: float = 0.0, # Optional probability to “unmask” 已 mask 的一部分
569
+ ):
570
+ """
571
+ 输入：
572
+ - input_ids: 原始 token id 序列
573
+ - input_embeddings: 原始 embedding 序列
574
+ - loss_mask_full: 全局可 mask 位掩码（包括动作 token 位置）
575
+ - mask_token_id: 用于填充的 special mask token id
576
+ - no_mask_token_prob: 可选概率，把已 mask 掉的位置再随机 unmask
577
+ 返回：
578
+ - masked_input_ids: 用 mask_token_id 替代被 mask 掉的位置的 input_ids
579
+ - labels: 原 input_ids，在非被 mask 位置用 -100 屏蔽（CrossEntropyLoss 忽略）
580
+ - new_input_embeddings: 对应替换了 action token 的新 embeddings
581
+ - loss_mask: float mask，用于后续 loss 加权（1 表示预测该位置，0 表示忽略）
582
+ """
583
+ B, L = input_ids.shape
584
+ device = input_ids.device
585
+
586
+ # 1) 计算每个样本总共可 mask 的 token 数量
587
+ # total_unknown = loss_mask_full.sum(dim=1) # (B,)
588
+ total_unknown = loss_mask_full.float().sum(dim=1) # (B,)
589
+
590
+ # 2) 随机采一个 time ratio in [0,1)
591
+ rand_time = torch.rand(B, device=device)
592
+
593
+ # 3) 根据 schedule 计算 mask ratio、再算出每个样本要 mask 的 token 数
594
+ # mask_ratios: tensor (B,), 取值 in (0,1]
595
+ mask_ratios = mask_schedule.schedule(rand_time, total_unknown, method="cosine") # [B]
596
+ # num_mask: at least 1
597
+ num_mask = torch.clamp((total_unknown * mask_ratios).round(), min=1).long() # [B]
598
+
599
+ # 4) 为每个位置打随机分数，非可-mask 位置打上大数，保证它永远不被选中
600
+ # vals: (B, L) ~ Uniform(0,1)
601
+ vals = torch.rand(B, L, device=device)
602
+ # large = 1e8
603
+ large = float('inf')
604
+ # 只有 loss_mask_full==True 的位置保留原分数，其他位置加大
605
+ vals = torch.where(loss_mask_full, vals, vals + large) # inf 表示不可选
606
+
607
+ # 5) 按行排序、取前 num_mask
608
+ perm = vals.argsort(dim=1) # (B, L)
609
+ ranks = perm.argsort(dim=1)
610
+ # masked_mask: bool (B, L)，True 表示该位置被 mask 掉
611
+ masked_mask = ranks < num_mask[:, None] # (B, L), 先 top-k mask
612
+
613
+ # 6) 可选：no_mask_token_prob，再次随机取消一部分已 mask 的位置
614
+ if no_mask_token_prob > 0:
615
+ # 从 masked_mask 中随机抽取一部分不再 mask
616
+ # 生成同形状的 [0,1) 随机数
617
+ prob = torch.rand(B, L, device=device)
618
+ # 在已经 mask 的位置上，若 prob < no_mask_token_prob 则 unmask
619
+ unmask = (prob < no_mask_token_prob) & masked_mask
620
+ masked_mask = masked_mask & (~unmask)
621
+
622
+ # # Set to True in eos_pos
623
+ # if eos_pos is not None:
624
+ # # eos_pos: (B,), 只在这些位置上 mask 掉
625
+ # eos_mask = torch.zeros_like(masked_mask, dtype=torch.bool, device=device)
626
+ # eos_mask[torch.arange(B, device=device), eos_pos] = True
627
+ # masked_mask = masked_mask | eos_mask
628
+
629
+ # 7) 构造 labels: 被 mask 掉的位置保留原 id，其他位置设为 -100
630
+ ignore_labels = torch.full_like(labels, fill_value=IGNORE_INDEX, dtype=labels.dtype, device=device)
631
+ masked_labels = torch.where(masked_mask, labels, ignore_labels)
632
+
633
+ masked_input_ids = torch.where(masked_mask, mask_token_id, input_ids)
634
+ # 8) 构造新的 input_embeddings: masked 位置替换成 用 embedding lookup 或直接替换
635
+ # 假设你后面是用 inputs_embeds，所以直接对 embeddings 替换
636
+ # masked_input_embeddings: (B, L, D)
637
+ masked_input_embeddings = input_embeddings.clone()
638
+ # 获取 mask token embedding
639
+ mask_emb = self.get_input_embeddings()(torch.tensor([mask_token_id], device=device)) # (1, D)
640
+ # 扩展到 (B, L, D)
641
+ mask_emb = mask_emb.view(1, 1, -1).expand(B, L, -1)
642
+ # 替换
643
+ masked_input_embeddings = torch.where(masked_mask.unsqueeze(-1), mask_emb, masked_input_embeddings)
644
+
645
+ # 10) 返回
646
+ # loss_mask: float 跟 JAX 里一致，用于后续加权 loss (1. for masked positions, 0. elsewhere)
647
+ loss_mask = masked_mask.float()
648
+
649
+ return masked_input_ids, masked_input_embeddings, masked_labels, loss_mask
650
+
651
+ # === Core Prismatic VLM `forward()` Logic ===
652
+ def forward(
653
+ self,
654
+ input_ids: Optional[torch.LongTensor] = None,
655
+ attention_mask: Optional[torch.Tensor] = None,
656
+ pixel_values: Optional[torch.FloatTensor] = None,
657
+ labels: Optional[torch.LongTensor] = None,
658
+ inputs_embeds: Optional[torch.FloatTensor] = None,
659
+ past_key_values: Optional[List[torch.FloatTensor]] = None,
660
+ use_cache: Optional[bool] = None,
661
+ output_attentions: Optional[bool] = None,
662
+ output_hidden_states: Optional[bool] = None,
663
+ output_projector_features: Optional[bool] = None,
664
+ return_dict: Optional[bool] = None,
665
+ proprio=None,
666
+ proprio_projector=None,
667
+ noisy_actions=None,
668
+ noisy_action_projector=None,
669
+ diffusion_timestep_embeddings=None,
670
+ use_film: bool = False,
671
+ d2f_mode: bool = False,
672
+ ) -> Union[Tuple, PrismaticCausalLMOutputWithPast]:
673
+ """Run a forward pass through the VLM, returning a PrismaticCausalLMOutputWithPast instance."""
674
+ output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
675
+ output_hidden_states = (
676
+ output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
677
+ ) # True
678
+ output_projector_features = output_projector_features if output_projector_features is not None else False # False
679
+ return_dict = return_dict if return_dict is not None else self.config.use_return_dict # True
680
+
681
+ # Respect `use_cache` only if not training (even if `gradient_checkpointing` is off)
682
+ use_cache = use_cache and not self.training
683
+
684
+ # Instantiate Placeholder for Projector Features
685
+ projected_patch_embeddings = None
686
+
687
+ # === Handle Generation with Cache (`input_ids.shape[1] == 1`) =>> requires `past_keys_values` ===
688
+ if input_ids.shape[1] == 1:
689
+ assert input_ids.shape[0] == 1, "Generation is only currently supported for batch size of 1!"
690
+ assert past_key_values is not None, "You must provide `past_key_values` during cached generation!"
691
+ assert labels is None, "Unexpected key `labels` provided during cached generation!"
692
+
693
+ language_model_output = self.language_model(
694
+ input_ids=input_ids,
695
+ attention_mask=None,
696
+ position_ids=None,
697
+ past_key_values=past_key_values,
698
+ inputs_embeds=None,
699
+ labels=None,
700
+ use_cache=use_cache,
701
+ output_attentions=output_attentions,
702
+ output_hidden_states=output_hidden_states,
703
+ return_dict=return_dict,
704
+ )
705
+
706
+ # === Handle Unimodal Forward ===
707
+ elif pixel_values is None:
708
+ assert (input_ids is not None) and (inputs_embeds is None), "Missing `input_ids` in language-only forward!"
709
+ assert past_key_values is None, "Unexpected key `past_key_values` provided during language-only forward!"
710
+
711
+ language_model_output = self.language_model(
712
+ input_ids=input_ids,
713
+ attention_mask=attention_mask,
714
+ position_ids=None,
715
+ past_key_values=None,
716
+ inputs_embeds=None,
717
+ labels=labels,
718
+ use_cache=use_cache,
719
+ output_attentions=output_attentions,
720
+ output_hidden_states=output_hidden_states,
721
+ return_dict=return_dict,
722
+ )
723
+
724
+ # === Handle Multimodal Forward ===
725
+ elif (input_ids.shape[0] == pixel_values.shape[0]) or (inputs_embeds.shape[0] == pixel_values.shape[0]):
726
+ assert past_key_values is None, "Unexpected key `past_key_values` provided during multimodal forward!"
727
+
728
+ # Get input embeddings (from language model embeddings)
729
+ input_embeddings = self.get_input_embeddings()(input_ids) # (B, seq_len, D)
730
+
731
+ # Extract action masks
732
+ all_actions_mask = self._process_action_masks(labels)
733
+
734
+ # Extract the language portion of the input embeddings (i.e. remove the action tokens portion)
735
+ language_embeddings = input_embeddings[~all_actions_mask].reshape(
736
+ input_embeddings.shape[0], -1, input_embeddings.shape[2]
737
+ ) # (B, lang_seq_len, llm_dim)
738
+
739
+ # Get visual features pixel_values [8, 12, 224, 224]
740
+ projected_patch_embeddings = self._process_vision_features(pixel_values, language_embeddings, use_film)
741
+
742
+ # Add proprioceptive state if provided
743
+ projected_patch_embeddings = self._process_proprio_features(
744
+ projected_patch_embeddings, proprio, proprio_projector
745
+ )
746
+
747
+ # [Diffusion] Add diffusion timestep embedding if provided
748
+ if diffusion_timestep_embeddings is not None:
749
+ # For simplicity, just append diffusion timestep embedding to the end of projected vision patch tokens
750
+ projected_patch_embeddings = torch.cat(
751
+ (projected_patch_embeddings, diffusion_timestep_embeddings), dim=1
752
+ )
753
+
754
+ # Process action embeddings
755
+ if noisy_actions is not None:
756
+ # Get mask corresponding to all action tokens
757
+ all_actions_mask = self._process_action_masks(labels)
758
+
759
+ # Reshape noisy actions into individual action tokens
760
+ # noisy_actions: (B, chunk_len, action_dim) -> (B, chunk_len * action_dim, 1)
761
+ B = noisy_actions.shape[0]
762
+ noisy_actions = noisy_actions.reshape(B, -1).unsqueeze(-1)
763
+
764
+ # Project noisy action tokens into language model embedding space
765
+ noisy_action_features = noisy_action_projector(noisy_actions) # (B, chunk_len * action_dim, llm_dim)
766
+
767
+ # Replace embeddings of the action tokens with noisy action embeddings
768
+ input_embeddings = self._replace_input_embeddings(
769
+ input_embeddings, all_actions_mask, noisy_action_features
770
+ )
771
+ elif d2f_mode:
772
+ # D2F supplies already-corrupted token IDs and a block attention mask.
773
+ pass
774
+ elif self.use_discrete_diffusion:
775
+ # (Later on, the positional embeddings will be added to them)
776
+ loss_mask_full = all_actions_mask # (B, seq_len)
777
+ eos_pos = self._get_eos_pos(all_actions_mask) # (B,)
778
+ input_ids, input_embeddings, labels, loss_mask = self.apply_mask_diffusion(
779
+ input_ids=input_ids,
780
+ input_embeddings=input_embeddings,
781
+ labels=labels,
782
+ loss_mask_full=loss_mask_full,
783
+ mask_token_id=self.mask_token_id,
784
+ no_mask_token_prob=0.0,
785
+ )
786
+
787
+ # Set labels to EOS_TOKEN in eos_pos
788
+ labels[torch.arange(labels.shape[0]), eos_pos] = STOP_INDEX
789
+
790
+ else:
791
+ # Replace the embeddings of the action tokens with zeros
792
+ # (Later on, the positional embeddings will be added to them)
793
+ all_actions_mask = all_actions_mask.unsqueeze(-1) # (B, seq_len, 1) [8, 93, 1]
794
+ input_embeddings = input_embeddings * ~all_actions_mask
795
+
796
+ # Build multimodal embeddings & attention mask
797
+ multimodal_embeddings, multimodal_attention_mask = self._build_multimodal_attention(
798
+ input_embeddings, projected_patch_embeddings, attention_mask
799
+ )
800
+
801
+ # Build labels for multimodal sequence if needed # labels shape [8, 93]
802
+ multimodal_labels = self._build_multimodal_labels(labels, projected_patch_embeddings)
803
+
804
+ # Dispatch to language model
805
+ language_model_output = self.language_model(
806
+ input_ids=None,
807
+ attention_mask=multimodal_attention_mask,
808
+ position_ids=None,
809
+ past_key_values=None,
810
+ inputs_embeds=multimodal_embeddings,
811
+ labels=multimodal_labels,
812
+ use_cache=use_cache,
813
+ output_attentions=output_attentions,
814
+ output_hidden_states=output_hidden_states,
815
+ return_dict=return_dict,
816
+ )
817
+
818
+ # === Otherwise =>> Assume Invalid! ===
819
+ elif (input_ids.shape[0] != pixel_values.shape[0]) or (inputs_embeds.shape[0] != pixel_values.shape[0]):
820
+ raise ValueError("Non-homogenous batch of (text, image) input -- forward() does not support mixed batches!")
821
+
822
+ else:
823
+ raise ValueError(
824
+ "Invalid PrismaticForConditionalGeneration `forward()` call with provided arguments:\n"
825
+ f"=> `input_ids` = {input_ids is not None}\n"
826
+ f"=> `attention_mask` = {attention_mask is not None}\n"
827
+ f"=> `pixel_values` = {pixel_values is not None}\n"
828
+ f"=> `labels` = {labels is not None}\n"
829
+ f"=> `input_embeds` = {inputs_embeds is not None}\n"
830
+ f"=> `past_key_values` = {past_key_values is not None}\n"
831
+ f"=> `use_cache` = {use_cache}"
832
+ )
833
+
834
+ # Unpack `language_model_output` and return PrismaticCausalLMOutputWithPast (or tuple if not `return_dict`)
835
+ if not return_dict:
836
+ if output_projector_features and (projected_patch_embeddings is not None):
837
+ return *language_model_output, projected_patch_embeddings
838
+
839
+ return language_model_output
840
+
841
+ return PrismaticCausalLMOutputWithPast(
842
+ loss=language_model_output.loss,
843
+ logits=language_model_output.logits,
844
+ past_key_values=language_model_output.past_key_values,
845
+ hidden_states=language_model_output.hidden_states,
846
+ attentions=language_model_output.attentions,
847
+ projector_features=projected_patch_embeddings,
848
+ labels=labels if labels is not None else None,
849
+ )
850
+
851
+ # === GenerationMixin Methods ===
852
+ def prepare_inputs_for_generation(
853
+ self,
854
+ input_ids: Optional[torch.Tensor] = None,
855
+ past_key_values: Optional[List[torch.FloatTensor]] = None,
856
+ inputs_embeds: Optional[torch.FloatTensor] = None,
857
+ pixel_values: Optional[torch.FloatTensor] = None,
858
+ attention_mask: Optional[torch.Tensor] = None,
859
+ **kwargs: str,
860
+ ) -> Dict[str, torch.Tensor]:
861
+ """Borrowed from `LlamaForCausalLM` and simplified for batch size = 1; mirrors original PrismaticVLM logic."""
862
+ if ((input_ids is not None) and (input_ids.shape[0] > 1)) or (
863
+ (inputs_embeds is not None) and (inputs_embeds.shape[0] > 1)
864
+ ):
865
+ raise ValueError("Generation with batch size > 1 is not currently supported!")
866
+
867
+ # Handle `past_key_values` (cache) =>> assume `input_ids` just has unprocessed tokens
868
+ if past_key_values is not None:
869
+ input_ids = input_ids[:, -1:]
870
+
871
+ # If `input_embeds` are passed, we only want to use them in the 1st generation step
872
+ if inputs_embeds is not None and past_key_values is None:
873
+ model_inputs = {"input_embeds": inputs_embeds}
874
+ else:
875
+ model_inputs = {"input_ids": input_ids}
876
+
877
+ # Make sure `pixel_values` are preserved in `model_inputs`
878
+ model_inputs.update(
879
+ {
880
+ "attention_mask": attention_mask,
881
+ "pixel_values": pixel_values,
882
+ "past_key_values": past_key_values,
883
+ "use_cache": kwargs.get("use_cache"),
884
+ }
885
+ )
886
+
887
+ return model_inputs
888
+
889
+ # Defer to Language Model (all handle this differently, with different return types)
890
+ def _reorder_cache(self, *args, **kwargs) -> Any:
891
+ return self.language_model._reorder_cache(*args, **kwargs)
892
+
893
+
894
+ class OpenVLAForActionPrediction(PrismaticForConditionalGeneration):
895
+ config_class: PretrainedConfig = OpenVLAConfig
896
+
897
+ def __init__(self, config: OpenVLAConfig) -> None:
898
+ super().__init__(config)
899
+ self.norm_stats = config.norm_stats
900
+
901
+ # Compute action bins
902
+ self.bins = np.linspace(-1, 1, config.n_action_bins)
903
+ self.bin_centers = (self.bins[:-1] + self.bins[1:]) / 2.0
904
+
905
+ # check if config has topk_filter_thres
906
+ if hasattr(config, "topk_filter_thres"):
907
+ self.topk_filter_thres = config.topk_filter_thres
908
+ else:
909
+ self.topk_filter_thres = 0.0
910
+
911
+ # Compute vocab size for de-tokenization -- revert added "multiple of"
912
+ self.vocab_size = self.config.text_config.vocab_size - self.config.pad_to_multiple_of
913
+
914
+ # Evaluation attaches a calibrated VLABSplineActionTokenizer here when
915
+ # action_tokenizer.json is present beside the checkpoint. Keep the
916
+ # default scalar-bin behavior when no tokenizer artifact is available.
917
+ self.policy_action_tokenizer = None
918
+
919
+ def _get_action_token_count(self) -> int:
920
+ """Return the number of policy tokens for the configured action representation."""
921
+
922
+ if self.policy_action_tokenizer is not None:
923
+ return int(self.policy_action_tokenizer.sequence_length)
924
+ return ACTION_DIM * NUM_ACTIONS_CHUNK
925
+
926
+ def _decode_policy_action_token_ids(self, token_ids) -> np.ndarray:
927
+ """Decode language-token predictions into one normalized action chunk."""
928
+
929
+ token_ids = np.asarray(token_ids).reshape(-1)
930
+ if self.policy_action_tokenizer is not None:
931
+ # Predictions can fall outside the reserved action vocabulary. The
932
+ # scalar-bin path historically clips those values, so use the
933
+ # B-spline adapter's equivalent robust analytic decoder here. Its
934
+ # adaptive path deliberately preserves the duration-defined native
935
+ # horizon rather than using the fixed-shape training metric path.
936
+ normalized_actions = self.policy_action_tokenizer.decode_action_token_ids_for_inference(token_ids)
937
+ if (
938
+ normalized_actions.ndim != 2
939
+ or normalized_actions.shape[1] != ACTION_DIM
940
+ or normalized_actions.shape[0] < 1
941
+ ):
942
+ raise ValueError(
943
+ "B-spline action decoder returned shape "
944
+ f"{normalized_actions.shape}, expected (steps, {ACTION_DIM})"
945
+ )
946
+ return normalized_actions
947
+
948
+ discretized_actions = self.vocab_size - token_ids
949
+ discretized_actions = np.clip(discretized_actions - 1, 0, self.bin_centers.shape[0] - 1)
950
+ return self.bin_centers[discretized_actions].reshape(NUM_ACTIONS_CHUNK, ACTION_DIM)
951
+
952
+ def _required_action_token_indices_for_execution(
953
+ self, token_ids, mask_token_id: int, execution_steps: int
954
+ ):
955
+ """Return action-token positions that fully determine an executed prefix."""
956
+
957
+ execution_steps = int(execution_steps)
958
+ if execution_steps < 1:
959
+ raise ValueError("execution_steps must be positive")
960
+ if self.policy_action_tokenizer is not None:
961
+ return self.policy_action_tokenizer.required_token_indices_for_execution(
962
+ np.asarray(token_ids).reshape(-1),
963
+ mask_token_id=mask_token_id,
964
+ num_steps=execution_steps,
965
+ )
966
+ token_count = min(execution_steps, NUM_ACTIONS_CHUNK) * ACTION_DIM
967
+ return np.arange(token_count, dtype=np.int64)
968
+
969
+ def _decode_policy_action_token_ids_for_execution(
970
+ self, token_ids, mask_token_id: int, execution_steps: int
971
+ ) -> np.ndarray:
972
+ """Decode a resolved executable prefix while permitting masked suffix slots."""
973
+
974
+ token_ids = np.asarray(token_ids).reshape(-1)
975
+ if self.policy_action_tokenizer is not None:
976
+ return self.policy_action_tokenizer.decode_action_token_ids_for_execution(
977
+ token_ids,
978
+ mask_token_id=mask_token_id,
979
+ num_steps=execution_steps,
980
+ )
981
+ count = min(int(execution_steps), NUM_ACTIONS_CHUNK)
982
+ required = count * ACTION_DIM
983
+ if np.any(token_ids[:required] == int(mask_token_id)):
984
+ raise ValueError("scalar action tokens required for execution are unresolved")
985
+ discretized_actions = self.vocab_size - token_ids[:required]
986
+ discretized_actions = np.clip(
987
+ discretized_actions - 1, 0, self.bin_centers.shape[0] - 1
988
+ )
989
+ return self.bin_centers[discretized_actions].reshape(count, ACTION_DIM)
990
+
991
+ def _prepare_input_for_action_prediction(self, input_ids, attention_mask):
992
+ """Prepares input for action prediction by adding necessary tokens"""
993
+ # Add representation-specific placeholder tokens to simulate action tokens.
994
+ action_token_count = self._get_action_token_count()
995
+ placeholder_action_token_ids = (
996
+ torch.ones((input_ids.shape[0], action_token_count)).to(input_ids.device).to(input_ids.dtype)
997
+ )
998
+ input_ids = torch.cat([input_ids, placeholder_action_token_ids], dim=-1)
999
+
1000
+ # Add stop token to sequence (needed in non-causal bi-directional self-attention, as it appears at train time)
1001
+ stop_token_id = torch.ones((input_ids.shape[0], 1)).to(input_ids.device).to(input_ids.dtype) * STOP_INDEX
1002
+ input_ids = torch.cat([input_ids, stop_token_id], dim=-1)
1003
+
1004
+ # Extend the attention mask to fit the new shape of input
1005
+ # Note: Only batch size == 1 supported right now
1006
+ mask_extension = (
1007
+ torch.ones((attention_mask.shape[0], input_ids.shape[-1] - attention_mask.shape[-1]))
1008
+ .to(attention_mask.device)
1009
+ .to(attention_mask.dtype)
1010
+ )
1011
+ attention_mask = torch.cat([attention_mask, mask_extension], dim=-1)
1012
+
1013
+ return input_ids, attention_mask
1014
+
1015
+ def _prepare_labels_for_action_prediction(self, labels, input_ids):
1016
+ """Creates labels tensor for action prediction if not provided"""
1017
+ # Extend labels tensor with fake action labels
1018
+ ARBITRARY_ACTION_TOKEN_IDX = ACTION_TOKEN_BEGIN_IDX + 1
1019
+ labels_extension = (
1020
+ torch.ones((labels.shape[0], input_ids.shape[-1] - labels.shape[-1])).to(labels.device).to(labels.dtype)
1021
+ * ARBITRARY_ACTION_TOKEN_IDX
1022
+ )
1023
+ labels = torch.cat([labels, labels_extension], dim=-1)
1024
+
1025
+ # Replace last label token with stop token
1026
+ labels[:, -1] = STOP_INDEX
1027
+
1028
+ return labels
1029
+
1030
+ def _unnormalize_actions(self, normalized_actions, unnorm_key=None):
1031
+ """Unnormalize actions using dataset statistics"""
1032
+ action_norm_stats = self.get_action_stats(unnorm_key)
1033
+
1034
+ if ACTION_PROPRIO_NORMALIZATION_TYPE == NormalizationType.BOUNDS:
1035
+ mask = action_norm_stats.get("mask", np.ones_like(action_norm_stats["min"], dtype=bool))
1036
+ action_high, action_low = np.array(action_norm_stats["max"]), np.array(action_norm_stats["min"])
1037
+ elif ACTION_PROPRIO_NORMALIZATION_TYPE == NormalizationType.BOUNDS_Q99:
1038
+ mask = action_norm_stats.get("mask", np.ones_like(action_norm_stats["q01"], dtype=bool))
1039
+ action_high, action_low = np.array(action_norm_stats["q99"]), np.array(action_norm_stats["q01"])
1040
+ else:
1041
+ raise ValueError("Unsupported action/proprio normalization type detected!")
1042
+
1043
+ actions = np.where(
1044
+ mask,
1045
+ 0.5 * (normalized_actions + 1) * (action_high - action_low + 1e-8) + action_low,
1046
+ normalized_actions,
1047
+ )
1048
+
1049
+ return actions
1050
+
1051
+ def _run_diffusion_prediction(
1052
+ self,
1053
+ input_embeddings,
1054
+ all_actions_mask,
1055
+ noise,
1056
+ action_head,
1057
+ projected_patch_embeddings,
1058
+ labels,
1059
+ attention_mask,
1060
+ NUM_PATCHES,
1061
+ NUM_PROMPT_TOKENS,
1062
+ noisy_action_projector,
1063
+ ):
1064
+ """Run diffusion-based action prediction"""
1065
+ # Clone embedding for reuse in each timestep
1066
+ orig_projected_patch_embeddings = projected_patch_embeddings.clone()
1067
+ curr_noisy_actions = noise
1068
+
1069
+ # Reverse diffusion: Iteratively denoise to generate action prediction
1070
+ for t in action_head.noise_scheduler.timesteps:
1071
+ # Get diffusion model's noise prediction (conditioned on VLA latent embedding, current noisy action
1072
+ # embedding, and diffusion timestep embedding)
1073
+ timesteps = torch.Tensor([t]).to(labels.device)
1074
+ diffusion_timestep_embeddings = (
1075
+ action_head.time_encoder(timesteps).to(curr_noisy_actions.dtype).to(curr_noisy_actions.device)
1076
+ ) # (B, llm_dim)
1077
+ diffusion_timestep_embeddings = diffusion_timestep_embeddings.unsqueeze(1) # (B, 1, llm_dim)
1078
+
1079
+ # [Diffusion] Replace the embeddings of the action tokens with noisy actions
1080
+ # (Later on, the positional embeddings will be added to them)
1081
+
1082
+ # For simplicity, append diffusion timestep embedding to the end of projected vision tokens
1083
+ projected_patch_embeddings = torch.cat(
1084
+ (orig_projected_patch_embeddings, diffusion_timestep_embeddings), dim=1
1085
+ )
1086
+
1087
+ # Reshape and project noisy actions into language embedding space
1088
+ B = curr_noisy_actions.shape[0]
1089
+ orig_curr_noisy_actions_shape = curr_noisy_actions.shape
1090
+ curr_noisy_actions = curr_noisy_actions.reshape(B, -1).unsqueeze(-1)
1091
+ noisy_action_features = noisy_action_projector(curr_noisy_actions)
1092
+ curr_noisy_actions = curr_noisy_actions.reshape(orig_curr_noisy_actions_shape)
1093
+
1094
+ # Replace action token embeddings with noisy action embeddings
1095
+ input_embeddings = self._replace_input_embeddings(
1096
+ input_embeddings.clone(), all_actions_mask, noisy_action_features
1097
+ )
1098
+
1099
+ # Build multimodal embeddings and attention mask
1100
+ multimodal_embeddings, multimodal_attention_mask = self._build_multimodal_attention(
1101
+ input_embeddings, projected_patch_embeddings, attention_mask
1102
+ )
1103
+
1104
+ # Forward pass through language model
1105
+ language_model_output = self.language_model(
1106
+ input_ids=None,
1107
+ attention_mask=multimodal_attention_mask,
1108
+ position_ids=None,
1109
+ past_key_values=None,
1110
+ inputs_embeds=multimodal_embeddings,
1111
+ labels=None,
1112
+ use_cache=None,
1113
+ output_attentions=False,
1114
+ output_hidden_states=True,
1115
+ return_dict=True,
1116
+ )
1117
+
1118
+ # Extract hidden states for action portion of response
1119
+ last_hidden_states = language_model_output.hidden_states[-1] # (B, seq_len, D)
1120
+ actions_hidden_states = last_hidden_states[
1121
+ :,
1122
+ NUM_PATCHES + NUM_PROMPT_TOKENS : NUM_PATCHES + NUM_PROMPT_TOKENS + ACTION_DIM * NUM_ACTIONS_CHUNK,
1123
+ :,
1124
+ ] # (B, act_chunk_len, D)
1125
+
1126
+ # Predict noise and update noisy actions: x_t -> x_{t-1}
1127
+ noise_pred = action_head.predict_noise(actions_hidden_states)
1128
+ curr_noisy_actions = action_head.noise_scheduler.step(noise_pred, t, curr_noisy_actions).prev_sample
1129
+
1130
+ curr_noisy_actions = curr_noisy_actions.reshape(NUM_ACTIONS_CHUNK, ACTION_DIM)
1131
+
1132
+ # Return final actions
1133
+ return curr_noisy_actions.float().cpu().detach().numpy(), actions_hidden_states
1134
+
1135
+ def _regression_or_discrete_prediction(
1136
+ self,
1137
+ input_embeddings,
1138
+ all_actions_mask,
1139
+ projected_patch_embeddings,
1140
+ attention_mask,
1141
+ labels,
1142
+ NUM_PATCHES,
1143
+ NUM_PROMPT_TOKENS,
1144
+ action_head=None,
1145
+ ):
1146
+ """Run L1 regression-based continuous action prediction or discrete action tokens prediction."""
1147
+ # Zero out action token embeddings
1148
+ all_actions_mask = all_actions_mask.unsqueeze(-1) # (B, seq_len, 1)
1149
+ input_embeddings = input_embeddings * ~all_actions_mask
1150
+
1151
+ # Build multimodal embeddings and attention mask
1152
+ multimodal_embeddings, multimodal_attention_mask = self._build_multimodal_attention(
1153
+ input_embeddings, projected_patch_embeddings, attention_mask
1154
+ )
1155
+
1156
+ # Forward pass through language model
1157
+ language_model_output = self.language_model(
1158
+ input_ids=None,
1159
+ attention_mask=multimodal_attention_mask,
1160
+ position_ids=None,
1161
+ past_key_values=None,
1162
+ inputs_embeds=multimodal_embeddings,
1163
+ labels=None,
1164
+ use_cache=None,
1165
+ output_attentions=False,
1166
+ output_hidden_states=True,
1167
+ return_dict=True,
1168
+ )
1169
+
1170
+ action_token_count = self._get_action_token_count() if action_head is None else ACTION_DIM * NUM_ACTIONS_CHUNK
1171
+
1172
+ # Extract hidden states for action tokens
1173
+ last_hidden_states = language_model_output.hidden_states[-1] # (B, seq_len, D)
1174
+ actions_hidden_states = last_hidden_states[
1175
+ :,
1176
+ NUM_PATCHES + NUM_PROMPT_TOKENS : NUM_PATCHES + NUM_PROMPT_TOKENS + action_token_count,
1177
+ :,
1178
+ ] # (B, act_chunk_len, D)
1179
+
1180
+ # Handle different prediction methods
1181
+ if action_head is not None:
1182
+ # L1 regression prediction
1183
+ normalized_actions = action_head.predict_action(actions_hidden_states)
1184
+ normalized_actions = normalized_actions.reshape(NUM_ACTIONS_CHUNK, ACTION_DIM)
1185
+ normalized_actions = normalized_actions.float().cpu().detach().numpy()
1186
+ else:
1187
+ # Discrete token-based prediction
1188
+ predicted_action_token_ids = (
1189
+ language_model_output.logits[
1190
+ :,
1191
+ NUM_PATCHES + NUM_PROMPT_TOKENS : NUM_PATCHES + NUM_PROMPT_TOKENS + action_token_count,
1192
+ ]
1193
+ .argmax(dim=2)
1194
+ .cpu()
1195
+ .numpy()
1196
+ )
1197
+ normalized_actions = self._decode_policy_action_token_ids(predicted_action_token_ids)
1198
+
1199
+ return normalized_actions, actions_hidden_states
1200
+
1201
+ def _discrete_diffusion_prediction(
1202
+ self,
1203
+ input_embeddings,
1204
+ all_actions_mask,
1205
+ projected_patch_embeddings,
1206
+ attention_mask,
1207
+ labels,
1208
+ NUM_PATCHES,
1209
+ NUM_PROMPT_TOKENS,
1210
+ action_head=None,
1211
+ input_ids=None,
1212
+ ):
1213
+ """Run L1 regression-based continuous action prediction or discrete action tokens prediction."""
1214
+ assert input_ids is not None, "Input IDs must be provided for discrete diffusion prediction!"
1215
+ action_token_count = self._get_action_token_count()
1216
+
1217
+ # Handle different prediction methods
1218
+ if action_head is not None:
1219
+ pass
1220
+ # # L1 regression prediction
1221
+ # normalized_actions = action_head.predict_action(actions_hidden_states)
1222
+ # normalized_actions = normalized_actions.reshape(NUM_ACTIONS_CHUNK, ACTION_DIM)
1223
+ # normalized_actions = normalized_actions.float().cpu().detach().numpy()
1224
+ else:
1225
+
1226
+ def tokens_to_logits(suffix_seq: torch.LongTensor) -> torch.Tensor:
1227
+
1228
+ prefix = masked_input_ids[:, :1+NUM_PROMPT_TOKENS]
1229
+ suffix = masked_input_ids[:, 1+NUM_PROMPT_TOKENS + action_token_count:]
1230
+ full_seqs = torch.cat([prefix, suffix_seq, suffix], dim=1)
1231
+
1232
+ # Get input embeddings and action masks
1233
+ input_embeddings = self.get_input_embeddings()(full_seqs)
1234
+
1235
+ # Build multimodal embeddings and attention mask
1236
+ multimodal_embeddings, multimodal_attention_mask = self._build_multimodal_attention(
1237
+ input_embeddings, projected_patch_embeddings, attention_mask
1238
+ )
1239
+
1240
+ language_model_output = self.language_model(
1241
+ input_ids=None,
1242
+ attention_mask=multimodal_attention_mask,
1243
+ position_ids=None,
1244
+ past_key_values=None,
1245
+ inputs_embeds=multimodal_embeddings,
1246
+ labels=None,
1247
+ use_cache=None,
1248
+ output_attentions=False,
1249
+ output_hidden_states=True,
1250
+ return_dict=True,
1251
+ )
1252
+
1253
+ logits = language_model_output.logits
1254
+ # topk_filter_thres = self.topk_filter_thres # suggest 0.0
1255
+ # filtered_logits = parallel_decode.top_k_logits(
1256
+ # logits, topk_filter_thres)
1257
+ filtered_logits = logits
1258
+
1259
+ full_logits = filtered_logits[
1260
+ :,
1261
+ NUM_PATCHES + NUM_PROMPT_TOKENS : NUM_PATCHES + NUM_PROMPT_TOKENS + action_token_count,
1262
+ :self.vocab_size
1263
+ ]
1264
+
1265
+ # Extract hidden states for action tokens
1266
+ last_hidden_states = language_model_output.hidden_states[-1] # (B, seq_len, D)
1267
+ actions_hidden_states = last_hidden_states[
1268
+ :,
1269
+ NUM_PATCHES + NUM_PROMPT_TOKENS : NUM_PATCHES + NUM_PROMPT_TOKENS + action_token_count,
1270
+ :,
1271
+ ] # (B, act_chunk_len, D)
1272
+
1273
+ return full_logits, actions_hidden_states # suffix 部分 [B, L, V]
1274
+
1275
+ # Set all action tokens to MASK_TOKEN
1276
+ # Note: Keep EOS/STOP tokens unchanged since their all_actions_mask is False
1277
+ mask_token_id = self.mask_token_id
1278
+ masked_input_ids = torch.where(
1279
+ all_actions_mask, torch.tensor(mask_token_id, device=input_ids.device), input_ids
1280
+ )
1281
+
1282
+ cur_seqs = masked_input_ids[
1283
+ :, 1+NUM_PROMPT_TOKENS:1+NUM_PROMPT_TOKENS + action_token_count
1284
+ ] # (B, seq_len)
1285
+
1286
+ final_iters, actions_hidden_states = parallel_decode.decode(
1287
+ init_ids=cur_seqs,
1288
+ tokens_to_logits=tokens_to_logits,
1289
+ mask_token_id=self.mask_token_id,
1290
+ num_iter=12,
1291
+ choice_temperature=1.0, # to_test
1292
+ mask_scheduling_method="cosine",
1293
+ use_remask=False,
1294
+ )
1295
+
1296
+ predicted_action_token_ids = final_iters[:, -1, :].cpu().numpy()
1297
+ normalized_actions = self._decode_policy_action_token_ids(predicted_action_token_ids)
1298
+
1299
+ return normalized_actions, actions_hidden_states
1300
+
1301
+ @staticmethod
1302
+ def _d2f_cache_length(cache) -> int:
1303
+ if cache is None:
1304
+ return 0
1305
+ if isinstance(cache, DynamicCache):
1306
+ return cache.get_seq_length()
1307
+ return cache[0][0].shape[-2] if len(cache) else 0
1308
+
1309
+ @staticmethod
1310
+ def _crop_d2f_cache(cache, length: int):
1311
+ """Crop either a DynamicCache or the legacy tuple returned by Transformers 4.40."""
1312
+ if isinstance(cache, DynamicCache):
1313
+ for layer_index in range(len(cache.key_cache)):
1314
+ cache.key_cache[layer_index] = cache.key_cache[layer_index][..., :length, :]
1315
+ cache.value_cache[layer_index] = cache.value_cache[layer_index][..., :length, :]
1316
+ cache._seen_tokens = length
1317
+ return cache
1318
+ return tuple(
1319
+ (layer[0][..., :length, :], layer[1][..., :length, :], *layer[2:]) for layer in cache
1320
+ )
1321
+
1322
+ @staticmethod
1323
+ def _d2f_full_attention_mask(condition_length, target_length, block_size, device, dtype):
1324
+ total_length = condition_length + target_length
1325
+ mask = torch.zeros((1, 1, total_length, total_length), device=device, dtype=dtype)
1326
+ mask[:, :, :condition_length, :condition_length] = 1
1327
+ for start in range(condition_length, total_length, block_size):
1328
+ end = min(start + block_size, total_length)
1329
+ mask[:, :, start:end, :end] = 1
1330
+ return mask
1331
+
1332
+ @staticmethod
1333
+ def _d2f_attention_slice(full_mask, start_position, input_length, cache_length):
1334
+ end_position = start_position + input_length
1335
+ result = torch.zeros(
1336
+ (1, 1, input_length, cache_length + input_length),
1337
+ device=full_mask.device,
1338
+ dtype=full_mask.dtype,
1339
+ )
1340
+ result[:, :, :, :cache_length] = full_mask[:, :, start_position:end_position, :cache_length]
1341
+ result[:, :, :, cache_length:] = full_mask[
1342
+ :, :, start_position:end_position, start_position:end_position
1343
+ ]
1344
+ return result
1345
+
1346
+ @staticmethod
1347
+ def _d2f_shift_logits(logits, previous_logit):
1348
+ shifted = torch.zeros_like(logits)
1349
+ shifted[:, 1:] = logits[:, :-1]
1350
+ if previous_logit is not None:
1351
+ shifted[:, :1] = previous_logit
1352
+ return shifted
1353
+
1354
+ @torch.no_grad()
1355
+ def _d2f_prediction(
1356
+ self,
1357
+ input_ids,
1358
+ projected_patch_embeddings,
1359
+ prompt_text_length,
1360
+ block_size=7,
1361
+ block_add_threshold=0.2,
1362
+ skip_threshold=0.8,
1363
+ decoded_token_threshold=0.8,
1364
+ execution_steps=None,
1365
+ max_iterations=10_000,
1366
+ ):
1367
+ """Adaptive block-parallel D2F decoding with execution-aware early stop.
1368
+
1369
+ When ``execution_steps`` is set, unresolved suffix slots are allowed
1370
+ only after the representation-specific decoder proves that they have
1371
+ zero influence on the requested chronological action prefix.
1372
+ """
1373
+ if input_ids.shape[0] != 1:
1374
+ raise ValueError("D2F inference currently supports batch size 1 only")
1375
+ if block_size <= 0:
1376
+ raise ValueError(f"block_size must be positive, got {block_size}")
1377
+ for name, value in (
1378
+ ("block_add_threshold", block_add_threshold),
1379
+ ("skip_threshold", skip_threshold),
1380
+ ("decoded_token_threshold", decoded_token_threshold),
1381
+ ):
1382
+ if not 0 <= value <= 1:
1383
+ raise ValueError(f"{name} must be in [0, 1], got {value}")
1384
+ if getattr(self.config, "_attn_implementation", "eager") == "flash_attention_2":
1385
+ raise ValueError("D2F requires eager/SDPA attention because it uses a four-dimensional block mask")
1386
+
1387
+ action_token_count = self._get_action_token_count()
1388
+ if self.policy_action_tokenizer is not None:
1389
+ expected_block_size = int(self.policy_action_tokenizer.d2f_block_size)
1390
+ if int(block_size) != expected_block_size:
1391
+ raise ValueError(
1392
+ f"D2F block_size={block_size} does not match policy layout "
1393
+ f"{self.policy_action_tokenizer.policy_layout!r} "
1394
+ f"(expected {expected_block_size})"
1395
+ )
1396
+ target_length = action_token_count + 1 # Include the stop token seen during training.
1397
+ mask_token_id = self.mask_token_id
1398
+ if mask_token_id is None:
1399
+ raise ValueError("D2F checkpoint has no mask_token_id")
1400
+
1401
+ prompt_ids = input_ids[:, :prompt_text_length]
1402
+ prompt_embeddings = self.get_input_embeddings()(prompt_ids)
1403
+ condition_embeddings = torch.cat(
1404
+ (prompt_embeddings[:, :1], projected_patch_embeddings, prompt_embeddings[:, 1:]), dim=1
1405
+ )
1406
+ condition_length = condition_embeddings.shape[1]
1407
+ target_ids = torch.full(
1408
+ (1, target_length), mask_token_id, dtype=input_ids.dtype, device=input_ids.device
1409
+ )
1410
+ full_attention_mask = self._d2f_full_attention_mask(
1411
+ condition_length,
1412
+ target_length,
1413
+ block_size,
1414
+ input_ids.device,
1415
+ condition_embeddings.dtype,
1416
+ )
1417
+
1418
+ block_states = {
1419
+ 0: {
1420
+ "start": 0,
1421
+ "end": condition_length,
1422
+ "mask_count": 0,
1423
+ "total": condition_length,
1424
+ "state": "to_cache",
1425
+ "complete": True,
1426
+ }
1427
+ }
1428
+ target_blocks = (target_length + block_size - 1) // block_size
1429
+ cache = DynamicCache()
1430
+ previous_logit = None
1431
+ current_length = condition_length
1432
+ stopped_for_execution = False
1433
+
1434
+ for _ in range(max_iterations):
1435
+ if execution_steps is not None:
1436
+ required = self._required_action_token_indices_for_execution(
1437
+ target_ids[0, :action_token_count].detach().cpu().numpy(),
1438
+ mask_token_id,
1439
+ execution_steps,
1440
+ )
1441
+ if required is not None and target_ids[0, required].ne(mask_token_id).all():
1442
+ stopped_for_execution = True
1443
+ break
1444
+
1445
+ last_state = block_states[max(block_states)]
1446
+ progress = (last_state["total"] - last_state["mask_count"]) / last_state["total"]
1447
+ if len(block_states) <= target_blocks and progress >= block_add_threshold:
1448
+ block_id = len(block_states)
1449
+ start = current_length
1450
+ size = min(block_size, condition_length + target_length - start)
1451
+ block_states[block_id] = {
1452
+ "start": start,
1453
+ "end": start + size,
1454
+ "mask_count": size,
1455
+ "total": size,
1456
+ "state": "active",
1457
+ "complete": False,
1458
+ }
1459
+ current_length += size
1460
+
1461
+ for block_id in sorted(block_states):
1462
+ state = block_states[block_id]
1463
+ decoded_ratio = (state["total"] - state["mask_count"]) / state["total"]
1464
+ if decoded_ratio >= decoded_token_threshold and block_id + 1 in block_states:
1465
+ block_states[block_id + 1]["complete"] = True
1466
+
1467
+ visible_targets = target_ids[:, : current_length - condition_length]
1468
+ if visible_targets.ne(mask_token_id).all() and len(block_states) == target_blocks + 1:
1469
+ break
1470
+
1471
+ blocks_to_cache = [bid for bid, state in block_states.items() if state["state"] == "to_cache"]
1472
+ cache_length = self._d2f_cache_length(cache)
1473
+ update_cache_length = 0
1474
+ if blocks_to_cache:
1475
+ earliest = min(blocks_to_cache)
1476
+ latest = max(blocks_to_cache)
1477
+ process_start = block_states[earliest]["start"]
1478
+ update_cache_length = block_states[latest]["end"] - process_start
1479
+ else:
1480
+ active = [bid for bid, state in block_states.items() if state["state"] == "active"]
1481
+ if not active:
1482
+ raise RuntimeError("D2F decoder has neither active nor cacheable blocks")
1483
+ process_start = min(block_states[bid]["start"] for bid in active)
1484
+
1485
+ target_embeddings = self.get_input_embeddings()(visible_targets)
1486
+ full_embeddings = torch.cat((condition_embeddings, target_embeddings), dim=1)
1487
+ input_embeddings = full_embeddings[:, process_start:current_length]
1488
+ if input_embeddings.shape[1] == 0:
1489
+ raise RuntimeError("D2F decoder produced an empty forward input")
1490
+ attention_mask = self._d2f_attention_slice(
1491
+ full_attention_mask,
1492
+ process_start,
1493
+ input_embeddings.shape[1],
1494
+ cache_length,
1495
+ )
1496
+ position_ids = torch.arange(
1497
+ process_start, current_length, device=input_ids.device, dtype=torch.long
1498
+ ).unsqueeze(0)
1499
+ output = self.language_model(
1500
+ inputs_embeds=input_embeddings,
1501
+ attention_mask=attention_mask,
1502
+ position_ids=position_ids,
1503
+ past_key_values=cache,
1504
+ use_cache=True,
1505
+ output_attentions=False,
1506
+ output_hidden_states=False,
1507
+ return_dict=True,
1508
+ )
1509
+
1510
+ cache = output.past_key_values
1511
+ if update_cache_length:
1512
+ cache_end = process_start + update_cache_length
1513
+ previous_logit = output.logits[:, update_cache_length - 1 : update_cache_length]
1514
+ cache = self._crop_d2f_cache(cache, cache_end)
1515
+ for block_id in blocks_to_cache:
1516
+ block_states[block_id]["state"] = "in_cache"
1517
+ else:
1518
+ cache = self._crop_d2f_cache(cache, cache_length)
1519
+
1520
+ shifted_logits = self._d2f_shift_logits(output.logits, previous_logit)
1521
+ completed = []
1522
+ for block_id in sorted(block_states):
1523
+ state = block_states[block_id]
1524
+ if state["state"] != "active":
1525
+ continue
1526
+ relative_start = state["start"] - process_start
1527
+ block_target_start = state["start"] - condition_length
1528
+ block_target_end = state["end"] - condition_length
1529
+ block_ids = target_ids[0, block_target_start:block_target_end]
1530
+ masked_positions = torch.where(block_ids.eq(mask_token_id))[0]
1531
+ if masked_positions.numel() == 0:
1532
+ completed.append(block_id)
1533
+ continue
1534
+ block_logits = shifted_logits[0, relative_start + masked_positions].clone()
1535
+ block_logits[:, mask_token_id] = torch.finfo(block_logits.dtype).min
1536
+ probabilities = torch.softmax(block_logits.float(), dim=-1)
1537
+ confidence, candidates = probabilities.max(dim=-1)
1538
+ selected = torch.where(confidence > skip_threshold)[0]
1539
+ if state["complete"] and selected.numel() == 0:
1540
+ selected = confidence.argmax().view(1)
1541
+ if selected.numel() > 0:
1542
+ absolute_target_positions = block_target_start + masked_positions[selected]
1543
+ target_ids[0, absolute_target_positions] = candidates[selected]
1544
+ state["mask_count"] -= int(selected.numel())
1545
+ if state["mask_count"] == 0:
1546
+ completed.append(block_id)
1547
+
1548
+ for block_id in completed:
1549
+ if any(
1550
+ block_states[previous]["state"] == "active"
1551
+ for previous in range(1, block_id)
1552
+ ):
1553
+ continue
1554
+ block_states[block_id]["state"] = "to_cache"
1555
+ else:
1556
+ raise RuntimeError(f"D2F decoding exceeded {max_iterations} iterations")
1557
+
1558
+ if not stopped_for_execution and target_ids.eq(mask_token_id).any():
1559
+ raise RuntimeError("D2F decoding terminated with unresolved mask tokens")
1560
+ predicted_action_token_ids = target_ids[0, :action_token_count].cpu().numpy()
1561
+ if stopped_for_execution:
1562
+ normalized_actions = self._decode_policy_action_token_ids_for_execution(
1563
+ predicted_action_token_ids,
1564
+ mask_token_id,
1565
+ execution_steps,
1566
+ )
1567
+ else:
1568
+ normalized_actions = self._decode_policy_action_token_ids(predicted_action_token_ids)
1569
+ return normalized_actions, None
1570
+
1571
+ def predict_action(
1572
+ self,
1573
+ input_ids: Optional[torch.LongTensor] = None,
1574
+ unnorm_key: Optional[str] = None,
1575
+ proprio=None,
1576
+ proprio_projector=None,
1577
+ action_head=None,
1578
+ noisy_action_projector=None,
1579
+ use_film: bool = False,
1580
+ use_discrete_diffusion: bool = False,
1581
+ use_d2f: bool = False,
1582
+ block_size: int = 7,
1583
+ block_add_threshold: float = 0.2,
1584
+ skip_threshold: float = 0.8,
1585
+ decoded_token_threshold: float = 0.8,
1586
+ d2f_execution_steps: Optional[int] = None,
1587
+ **kwargs: str,
1588
+ ) -> np.ndarray:
1589
+ """Predict actions from input sequence, with options for different prediction methods.
1590
+
1591
+ Args:
1592
+ input_ids: Input token ids
1593
+ unnorm_key: Key for unnormalization statistics
1594
+ proprio: Proprioceptive features
1595
+ proprio_projector: Projector for proprioceptive features
1596
+ action_head: Optional head for L1 regression or diffusion-based prediction
1597
+ noisy_action_projector: Projector for noisy actions in diffusion-based prediction
1598
+ use_film: Whether to use FiLM conditioning
1599
+ use_discrete_diffusion: Whether to use discrete diffusion for action prediction
1600
+ d2f_execution_steps: Optional number of chronological actions that
1601
+ the caller will execute before querying the policy again
1602
+ **kwargs: Additional arguments including pixel_values and attention_mask
1603
+
1604
+ Returns:
1605
+ Tuple of (unnormalized_actions, action_hidden_states)
1606
+ """
1607
+ # If the special empty token ('') does not already appear after the colon (':') token in the prompt
1608
+ # (after "OUT:" or "ASSISTANT:"), insert it to match the inputs seen at training time
1609
+ if not torch.all(input_ids[:, -1] == 29871):
1610
+ input_ids = torch.cat(
1611
+ (input_ids, torch.unsqueeze(torch.Tensor([29871]).long(), dim=0).to(input_ids.device)), dim=1
1612
+ )
1613
+
1614
+ pixel_values = kwargs["pixel_values"]
1615
+ attention_mask = kwargs["attention_mask"]
1616
+
1617
+ # Create fake labels tensor (needed for action mask)
1618
+ labels = input_ids.clone()
1619
+ labels[:] = IGNORE_INDEX
1620
+
1621
+ # Get number of tokens in prompt (excluding the start token)
1622
+ NUM_PROMPT_TOKENS = input_ids.shape[-1] - 1 # Subtract action tokens and stop token
1623
+
1624
+ # Prepare inputs by adding necessary tokens
1625
+ input_ids, attention_mask = self._prepare_input_for_action_prediction(input_ids, attention_mask)
1626
+
1627
+ # Update labels tensor for action mask computation later
1628
+ labels = self._prepare_labels_for_action_prediction(labels, input_ids)
1629
+
1630
+ # Get input embeddings and action masks
1631
+ input_embeddings = self.get_input_embeddings()(input_ids)
1632
+ all_actions_mask = self._process_action_masks(labels)
1633
+
1634
+ # Extract language embeddings
1635
+ language_embeddings = input_embeddings[~all_actions_mask].reshape(
1636
+ input_embeddings.shape[0], -1, input_embeddings.shape[2]
1637
+ )
1638
+
1639
+ # Process vision features
1640
+ projected_patch_embeddings = self._process_vision_features(pixel_values, language_embeddings, use_film)
1641
+
1642
+ # Add proprioceptive features if provided
1643
+ use_proprio = proprio_projector is not None and proprio is not None
1644
+ if use_proprio:
1645
+ proprio = torch.Tensor(proprio).to(projected_patch_embeddings.device, dtype=projected_patch_embeddings.dtype)
1646
+ projected_patch_embeddings = self._process_proprio_features(
1647
+ projected_patch_embeddings, proprio, proprio_projector
1648
+ )
1649
+
1650
+ if use_d2f and use_discrete_diffusion:
1651
+ raise ValueError("use_d2f and use_discrete_diffusion are mutually exclusive")
1652
+
1653
+ # Use diffusion if provided, otherwise use regression or discrete prediction
1654
+ use_diffusion = noisy_action_projector is not None and hasattr(action_head, "noise_scheduler")
1655
+
1656
+ # Calculate number of patches (including proprio token and/or diffusion timestep embedding if present)
1657
+ NUM_PATCHES = self.vision_backbone.get_num_patches() * self.vision_backbone.get_num_images_in_input()
1658
+ if use_proprio:
1659
+ NUM_PATCHES += 1
1660
+ if use_diffusion:
1661
+ NUM_PATCHES += 1
1662
+
1663
+ if use_d2f:
1664
+ normalized_actions, actions_hidden_states = self._d2f_prediction(
1665
+ input_ids=input_ids,
1666
+ projected_patch_embeddings=projected_patch_embeddings,
1667
+ prompt_text_length=NUM_PROMPT_TOKENS + 1,
1668
+ block_size=block_size,
1669
+ block_add_threshold=block_add_threshold,
1670
+ skip_threshold=skip_threshold,
1671
+ decoded_token_threshold=decoded_token_threshold,
1672
+ execution_steps=d2f_execution_steps,
1673
+ )
1674
+ elif use_diffusion:
1675
+ assert use_discrete_diffusion is False, "Discrete diffusion has not been supported in this method!"
1676
+ # Sample random noise with shape equal to output action, used as the starting state for reverse diffusion
1677
+ noise = torch.randn(
1678
+ size=(1, NUM_ACTIONS_CHUNK, ACTION_DIM), device=input_embeddings.device, dtype=input_embeddings.dtype
1679
+ )
1680
+
1681
+ # Run diffusion-based prediction
1682
+ normalized_actions, actions_hidden_states = self._run_diffusion_prediction(
1683
+ input_embeddings,
1684
+ all_actions_mask,
1685
+ noise,
1686
+ action_head,
1687
+ projected_patch_embeddings,
1688
+ labels,
1689
+ attention_mask,
1690
+ NUM_PATCHES,
1691
+ NUM_PROMPT_TOKENS,
1692
+ noisy_action_projector,
1693
+ )
1694
+ elif use_discrete_diffusion:
1695
+ normalized_actions, actions_hidden_states = self._discrete_diffusion_prediction(
1696
+ input_embeddings,
1697
+ all_actions_mask,
1698
+ projected_patch_embeddings,
1699
+ attention_mask,
1700
+ labels,
1701
+ NUM_PATCHES,
1702
+ NUM_PROMPT_TOKENS,
1703
+ action_head,
1704
+ input_ids=input_ids,
1705
+ )
1706
+ else:
1707
+ # Run regression or discrete token-based prediction
1708
+ normalized_actions, actions_hidden_states = self._regression_or_discrete_prediction(
1709
+ input_embeddings,
1710
+ all_actions_mask,
1711
+ projected_patch_embeddings,
1712
+ attention_mask,
1713
+ labels,
1714
+ NUM_PATCHES,
1715
+ NUM_PROMPT_TOKENS,
1716
+ action_head,
1717
+ )
1718
+
1719
+ # Unnormalize predicted actions
1720
+ actions = self._unnormalize_actions(normalized_actions, unnorm_key)
1721
+
1722
+ return actions, actions_hidden_states
1723
+
1724
+ @staticmethod
1725
+ def _check_unnorm_key(norm_stats: Dict[str, Dict[str, Any]], unnorm_key: Optional[str]) -> str:
1726
+ """Validate and resolve the unnormalization key for action statistics"""
1727
+ if unnorm_key is None:
1728
+ assert len(norm_stats) == 1, (
1729
+ f"Your model was trained on more than one dataset, "
1730
+ f"please pass a `unnorm_key` from the following options to choose the statistics "
1731
+ f"used for un-normalizing actions: {norm_stats.keys()}"
1732
+ )
1733
+ unnorm_key = next(iter(norm_stats.keys()))
1734
+
1735
+ assert unnorm_key in norm_stats, (
1736
+ f"The `unnorm_key` you chose is not in the set of available dataset statistics, "
1737
+ f"please choose from: {norm_stats.keys()}"
1738
+ )
1739
+ return unnorm_key
1740
+
1741
+ def get_action_dim(self, unnorm_key: Optional[str] = None) -> int:
1742
+ """Get the dimensionality of the policy's action space."""
1743
+ unnorm_key = self._check_unnorm_key(self.norm_stats, unnorm_key)
1744
+ return len(self.norm_stats[unnorm_key]["action"]["min"])
1745
+
1746
+ def get_action_stats(self, unnorm_key: Optional[str] = None) -> Dict[str, Any]:
1747
+ """Get all the logged statistics for the given dataset."""
1748
+ unnorm_key = self._check_unnorm_key(self.norm_stats, unnorm_key)
1749
+ return self.norm_stats[unnorm_key]["action"]
1750
+
1751
+
1752
+ class DiscreteDiffusionForActionPrediction(PrismaticForConditionalGeneration):
1753
+ config_class: PretrainedConfig = OpenVLAConfig
1754
+
1755
+ def __init__(self, config: OpenVLAConfig) -> None:
1756
+ super().__init__(config)
1757
+ self.norm_stats = config.norm_stats
1758
+
1759
+ # Compute action bins
1760
+ self.bins = np.linspace(-1, 1, config.n_action_bins)
1761
+ self.bin_centers = (self.bins[:-1] + self.bins[1:]) / 2.0
1762
+
1763
+ # Compute vocab size for de-tokenization -- revert added "multiple of"
1764
+ self.vocab_size = self.config.text_config.vocab_size - self.config.pad_to_multiple_of
1765
+
1766
+ def _prepare_input_for_action_prediction(self, input_ids, attention_mask):
1767
+ """Prepares input for action prediction by adding necessary tokens"""
1768
+ # Add (ACTION_DIM * NUM_ACTIONS_CHUNK) placeholder tokens to input_ids to simulate action tokens
1769
+ placeholder_action_token_ids = (
1770
+ torch.ones((input_ids.shape[0], ACTION_DIM * NUM_ACTIONS_CHUNK)).to(input_ids.device).to(input_ids.dtype)
1771
+ )
1772
+ input_ids = torch.cat([input_ids, placeholder_action_token_ids], dim=-1)
1773
+
1774
+ # Add stop token to sequence (needed in non-causal bi-directional self-attention, as it appears at train time)
1775
+ stop_token_id = torch.ones((input_ids.shape[0], 1)).to(input_ids.device).to(input_ids.dtype) * STOP_INDEX # TODO: IMPORTANT
1776
+ input_ids = torch.cat([input_ids, stop_token_id], dim=-1)
1777
+
1778
+ # Extend the attention mask to fit the new shape of input
1779
+ # Note: Only batch size == 1 supported right now
1780
+ mask_extension = (
1781
+ torch.ones((attention_mask.shape[0], input_ids.shape[-1] - attention_mask.shape[-1]))
1782
+ .to(attention_mask.device)
1783
+ .to(attention_mask.dtype)
1784
+ )
1785
+ attention_mask = torch.cat([attention_mask, mask_extension], dim=-1)
1786
+
1787
+ return input_ids, attention_mask
1788
+
1789
+ def _prepare_labels_for_action_prediction(self, labels, input_ids):
1790
+ """Creates labels tensor for action prediction if not provided"""
1791
+ # Extend labels tensor with fake action labels
1792
+ ARBITRARY_ACTION_TOKEN_IDX = ACTION_TOKEN_BEGIN_IDX + 1
1793
+ labels_extension = (
1794
+ torch.ones((labels.shape[0], input_ids.shape[-1] - labels.shape[-1])).to(labels.device).to(labels.dtype)
1795
+ * ARBITRARY_ACTION_TOKEN_IDX
1796
+ )
1797
+ labels = torch.cat([labels, labels_extension], dim=-1)
1798
+
1799
+ # Replace last label token with stop token
1800
+ labels[:, -1] = STOP_INDEX
1801
+
1802
+ return labels
1803
+
1804
+ def _unnormalize_actions(self, normalized_actions, unnorm_key=None):
1805
+ """Unnormalize actions using dataset statistics"""
1806
+ action_norm_stats = self.get_action_stats(unnorm_key)
1807
+
1808
+ if ACTION_PROPRIO_NORMALIZATION_TYPE == NormalizationType.BOUNDS:
1809
+ mask = action_norm_stats.get("mask", np.ones_like(action_norm_stats["min"], dtype=bool))
1810
+ action_high, action_low = np.array(action_norm_stats["max"]), np.array(action_norm_stats["min"])
1811
+ elif ACTION_PROPRIO_NORMALIZATION_TYPE == NormalizationType.BOUNDS_Q99:
1812
+ mask = action_norm_stats.get("mask", np.ones_like(action_norm_stats["q01"], dtype=bool))
1813
+ action_high, action_low = np.array(action_norm_stats["q99"]), np.array(action_norm_stats["q01"])
1814
+ else:
1815
+ raise ValueError("Unsupported action/proprio normalization type detected!")
1816
+
1817
+ actions = np.where(
1818
+ mask,
1819
+ 0.5 * (normalized_actions + 1) * (action_high - action_low + 1e-8) + action_low,
1820
+ normalized_actions,
1821
+ )
1822
+
1823
+ return actions
1824
+
1825
+ def _run_diffusion_prediction(
1826
+ self,
1827
+ input_embeddings,
1828
+ all_actions_mask,
1829
+ noise,
1830
+ action_head,
1831
+ projected_patch_embeddings,
1832
+ labels,
1833
+ attention_mask,
1834
+ NUM_PATCHES,
1835
+ NUM_PROMPT_TOKENS,
1836
+ noisy_action_projector,
1837
+ ):
1838
+ """Run diffusion-based action prediction"""
1839
+ # Clone embedding for reuse in each timestep
1840
+ orig_projected_patch_embeddings = projected_patch_embeddings.clone()
1841
+ curr_noisy_actions = noise
1842
+
1843
+ # Reverse diffusion: Iteratively denoise to generate action prediction
1844
+ for t in action_head.noise_scheduler.timesteps:
1845
+ # Get diffusion model's noise prediction (conditioned on VLA latent embedding, current noisy action
1846
+ # embedding, and diffusion timestep embedding)
1847
+ timesteps = torch.Tensor([t]).to(labels.device)
1848
+ diffusion_timestep_embeddings = (
1849
+ action_head.time_encoder(timesteps).to(curr_noisy_actions.dtype).to(curr_noisy_actions.device)
1850
+ ) # (B, llm_dim)
1851
+ diffusion_timestep_embeddings = diffusion_timestep_embeddings.unsqueeze(1) # (B, 1, llm_dim)
1852
+
1853
+ # [Diffusion] Replace the embeddings of the action tokens with noisy actions
1854
+ # (Later on, the positional embeddings will be added to them)
1855
+
1856
+ # For simplicity, append diffusion timestep embedding to the end of projected vision tokens
1857
+ projected_patch_embeddings = torch.cat(
1858
+ (orig_projected_patch_embeddings, diffusion_timestep_embeddings), dim=1
1859
+ )
1860
+
1861
+ # Reshape and project noisy actions into language embedding space
1862
+ B = curr_noisy_actions.shape[0]
1863
+ orig_curr_noisy_actions_shape = curr_noisy_actions.shape
1864
+ curr_noisy_actions = curr_noisy_actions.reshape(B, -1).unsqueeze(-1)
1865
+ noisy_action_features = noisy_action_projector(curr_noisy_actions)
1866
+ curr_noisy_actions = curr_noisy_actions.reshape(orig_curr_noisy_actions_shape)
1867
+
1868
+ # Replace action token embeddings with noisy action embeddings
1869
+ input_embeddings = self._replace_input_embeddings(
1870
+ input_embeddings.clone(), all_actions_mask, noisy_action_features
1871
+ )
1872
+
1873
+ # Build multimodal embeddings and attention mask
1874
+ multimodal_embeddings, multimodal_attention_mask = self._build_multimodal_attention(
1875
+ input_embeddings, projected_patch_embeddings, attention_mask
1876
+ )
1877
+
1878
+ # Forward pass through language model
1879
+ language_model_output = self.language_model(
1880
+ input_ids=None,
1881
+ attention_mask=multimodal_attention_mask,
1882
+ position_ids=None,
1883
+ past_key_values=None,
1884
+ inputs_embeds=multimodal_embeddings,
1885
+ labels=None,
1886
+ use_cache=None,
1887
+ output_attentions=False,
1888
+ output_hidden_states=True,
1889
+ return_dict=True,
1890
+ )
1891
+
1892
+ # Extract hidden states for action portion of response
1893
+ last_hidden_states = language_model_output.hidden_states[-1] # (B, seq_len, D)
1894
+ actions_hidden_states = last_hidden_states[
1895
+ :,
1896
+ NUM_PATCHES + NUM_PROMPT_TOKENS : NUM_PATCHES + NUM_PROMPT_TOKENS + ACTION_DIM * NUM_ACTIONS_CHUNK,
1897
+ :,
1898
+ ] # (B, act_chunk_len, D)
1899
+
1900
+ # Predict noise and update noisy actions: x_t -> x_{t-1}
1901
+ noise_pred = action_head.predict_noise(actions_hidden_states)
1902
+ curr_noisy_actions = action_head.noise_scheduler.step(noise_pred, t, curr_noisy_actions).prev_sample
1903
+
1904
+ curr_noisy_actions = curr_noisy_actions.reshape(NUM_ACTIONS_CHUNK, ACTION_DIM)
1905
+
1906
+ # Return final actions
1907
+ return curr_noisy_actions.float().cpu().detach().numpy(), actions_hidden_states
1908
+
1909
+ def _regression_or_discrete_prediction(
1910
+ self,
1911
+ input_embeddings,
1912
+ all_actions_mask,
1913
+ projected_patch_embeddings,
1914
+ attention_mask,
1915
+ labels,
1916
+ NUM_PATCHES,
1917
+ NUM_PROMPT_TOKENS,
1918
+ action_head=None,
1919
+ ):
1920
+ """Run L1 regression-based continuous action prediction or discrete action tokens prediction."""
1921
+ # Zero out action token embeddings
1922
+ all_actions_mask = all_actions_mask.unsqueeze(-1) # (B, seq_len, 1)
1923
+ input_embeddings = input_embeddings * ~all_actions_mask
1924
+
1925
+ # Build multimodal embeddings and attention mask
1926
+ multimodal_embeddings, multimodal_attention_mask = self._build_multimodal_attention(
1927
+ input_embeddings, projected_patch_embeddings, attention_mask
1928
+ )
1929
+
1930
+ # Forward pass through language model
1931
+ language_model_output = self.language_model(
1932
+ input_ids=None,
1933
+ attention_mask=multimodal_attention_mask,
1934
+ position_ids=None,
1935
+ past_key_values=None,
1936
+ inputs_embeds=multimodal_embeddings,
1937
+ labels=None,
1938
+ use_cache=None,
1939
+ output_attentions=False,
1940
+ output_hidden_states=True,
1941
+ return_dict=True,
1942
+ )
1943
+
1944
+ # Extract hidden states for action tokens
1945
+ last_hidden_states = language_model_output.hidden_states[-1] # (B, seq_len, D)
1946
+ actions_hidden_states = last_hidden_states[
1947
+ :,
1948
+ NUM_PATCHES + NUM_PROMPT_TOKENS : NUM_PATCHES + NUM_PROMPT_TOKENS + ACTION_DIM * NUM_ACTIONS_CHUNK,
1949
+ :,
1950
+ ] # (B, act_chunk_len, D)
1951
+
1952
+ # Handle different prediction methods
1953
+ if action_head is not None:
1954
+ # L1 regression prediction
1955
+ normalized_actions = action_head.predict_action(actions_hidden_states)
1956
+ normalized_actions = normalized_actions.reshape(NUM_ACTIONS_CHUNK, ACTION_DIM)
1957
+ normalized_actions = normalized_actions.float().cpu().detach().numpy()
1958
+ else:
1959
+ # Discrete token-based prediction
1960
+ predicted_action_token_ids = (
1961
+ language_model_output.logits[
1962
+ :,
1963
+ NUM_PATCHES + NUM_PROMPT_TOKENS : NUM_PATCHES + NUM_PROMPT_TOKENS + ACTION_DIM * NUM_ACTIONS_CHUNK,
1964
+ ]
1965
+ .argmax(dim=2)
1966
+ .cpu()
1967
+ .numpy()
1968
+ )
1969
+ discretized_actions = self.vocab_size - predicted_action_token_ids
1970
+ discretized_actions = np.clip(discretized_actions - 1, a_min=0, a_max=self.bin_centers.shape[0] - 1)
1971
+ normalized_actions = self.bin_centers[discretized_actions]
1972
+ normalized_actions = normalized_actions.reshape(NUM_ACTIONS_CHUNK, ACTION_DIM)
1973
+
1974
+ return normalized_actions, actions_hidden_states
1975
+
1976
+ def predict_action(
1977
+ self,
1978
+ input_ids: Optional[torch.LongTensor] = None,
1979
+ unnorm_key: Optional[str] = None,
1980
+ proprio=None,
1981
+ proprio_projector=None,
1982
+ action_head=None,
1983
+ noisy_action_projector=None,
1984
+ use_film: bool = False,
1985
+ **kwargs: str,
1986
+ ) -> np.ndarray:
1987
+ """Predict actions from input sequence, with options for different prediction methods.
1988
+
1989
+ Args:
1990
+ input_ids: Input token ids
1991
+ unnorm_key: Key for unnormalization statistics
1992
+ proprio: Proprioceptive features
1993
+ proprio_projector: Projector for proprioceptive features
1994
+ action_head: Optional head for L1 regression or diffusion-based prediction
1995
+ noisy_action_projector: Projector for noisy actions in diffusion-based prediction
1996
+ use_film: Whether to use FiLM conditioning
1997
+ **kwargs: Additional arguments including pixel_values and attention_mask
1998
+
1999
+ Returns:
2000
+ Tuple of (unnormalized_actions, action_hidden_states)
2001
+ """
2002
+ # If the special empty token ('') does not already appear after the colon (':') token in the prompt
2003
+ # (after "OUT:" or "ASSISTANT:"), insert it to match the inputs seen at training time
2004
+ if not torch.all(input_ids[:, -1] == 29871):
2005
+ input_ids = torch.cat(
2006
+ (input_ids, torch.unsqueeze(torch.Tensor([29871]).long(), dim=0).to(input_ids.device)), dim=1
2007
+ )
2008
+
2009
+ pixel_values = kwargs["pixel_values"]
2010
+ attention_mask = kwargs["attention_mask"]
2011
+
2012
+ # Create fake labels tensor (needed for action mask)
2013
+ labels = input_ids.clone()
2014
+ labels[:] = IGNORE_INDEX
2015
+
2016
+ # Get number of tokens in prompt (excluding the start token)
2017
+ NUM_PROMPT_TOKENS = input_ids.shape[-1] - 1 # Subtract action tokens and stop token
2018
+
2019
+ # Prepare inputs by adding necessary tokens
2020
+ input_ids, attention_mask = self._prepare_input_for_action_prediction(input_ids, attention_mask)
2021
+
2022
+ # Update labels tensor for action mask computation later
2023
+ labels = self._prepare_labels_for_action_prediction(labels, input_ids)
2024
+
2025
+ # Get input embeddings and action masks
2026
+ input_embeddings = self.get_input_embeddings()(input_ids)
2027
+ all_actions_mask = self._process_action_masks(labels)
2028
+
2029
+ # Extract language embeddings
2030
+ language_embeddings = input_embeddings[~all_actions_mask].reshape(
2031
+ input_embeddings.shape[0], -1, input_embeddings.shape[2]
2032
+ )
2033
+
2034
+ # Process vision features
2035
+ projected_patch_embeddings = self._process_vision_features(pixel_values, language_embeddings, use_film)
2036
+
2037
+ # Add proprioceptive features if provided
2038
+ use_proprio = proprio_projector is not None and proprio is not None
2039
+ if use_proprio:
2040
+ proprio = torch.Tensor(proprio).to(projected_patch_embeddings.device, dtype=projected_patch_embeddings.dtype)
2041
+ projected_patch_embeddings = self._process_proprio_features(
2042
+ projected_patch_embeddings, proprio, proprio_projector
2043
+ )
2044
+
2045
+ # Use diffusion if provided, otherwise use regression or discrete prediction
2046
+ use_diffusion = noisy_action_projector is not None and hasattr(action_head, "noise_scheduler")
2047
+
2048
+ # Calculate number of patches (including proprio token and/or diffusion timestep embedding if present)
2049
+ NUM_PATCHES = self.vision_backbone.get_num_patches() * self.vision_backbone.get_num_images_in_input()
2050
+ if use_proprio:
2051
+ NUM_PATCHES += 1
2052
+ if use_diffusion:
2053
+ NUM_PATCHES += 1
2054
+
2055
+ if use_diffusion:
2056
+ # Sample random noise with shape equal to output action, used as the starting state for reverse diffusion
2057
+ noise = torch.randn(
2058
+ size=(1, NUM_ACTIONS_CHUNK, ACTION_DIM), device=input_embeddings.device, dtype=input_embeddings.dtype
2059
+ )
2060
+
2061
+ # Run diffusion-based prediction
2062
+ normalized_actions, actions_hidden_states = self._run_diffusion_prediction(
2063
+ input_embeddings,
2064
+ all_actions_mask,
2065
+ noise,
2066
+ action_head,
2067
+ projected_patch_embeddings,
2068
+ labels,
2069
+ attention_mask,
2070
+ NUM_PATCHES,
2071
+ NUM_PROMPT_TOKENS,
2072
+ noisy_action_projector,
2073
+ )
2074
+ else:
2075
+ # Run regression or discrete token-based prediction
2076
+ normalized_actions, actions_hidden_states = self._regression_or_discrete_prediction(
2077
+ input_embeddings,
2078
+ all_actions_mask,
2079
+ projected_patch_embeddings,
2080
+ attention_mask,
2081
+ labels,
2082
+ NUM_PATCHES,
2083
+ NUM_PROMPT_TOKENS,
2084
+ action_head,
2085
+ )
2086
+
2087
+ # Unnormalize predicted actions
2088
+ actions = self._unnormalize_actions(normalized_actions, unnorm_key)
2089
+
2090
+ return actions, actions_hidden_states
2091
+
2092
+ @staticmethod
2093
+ def _check_unnorm_key(norm_stats: Dict[str, Dict[str, Any]], unnorm_key: Optional[str]) -> str:
2094
+ """Validate and resolve the unnormalization key for action statistics"""
2095
+ if unnorm_key is None:
2096
+ assert len(norm_stats) == 1, (
2097
+ f"Your model was trained on more than one dataset, "
2098
+ f"please pass a `unnorm_key` from the following options to choose the statistics "
2099
+ f"used for un-normalizing actions: {norm_stats.keys()}"
2100
+ )
2101
+ unnorm_key = next(iter(norm_stats.keys()))
2102
+
2103
+ assert unnorm_key in norm_stats, (
2104
+ f"The `unnorm_key` you chose is not in the set of available dataset statistics, "
2105
+ f"please choose from: {norm_stats.keys()}"
2106
+ )
2107
+ return unnorm_key
2108
+
2109
+ def get_action_dim(self, unnorm_key: Optional[str] = None) -> int:
2110
+ """Get the dimensionality of the policy's action space."""
2111
+ unnorm_key = self._check_unnorm_key(self.norm_stats, unnorm_key)
2112
+ return len(self.norm_stats[unnorm_key]["action"]["min"])
2113
+
2114
+ def get_action_stats(self, unnorm_key: Optional[str] = None) -> Dict[str, Any]:
2115
+ """Get all the logged statistics for the given dataset."""
2116
+ unnorm_key = self._check_unnorm_key(self.norm_stats, unnorm_key)
2117
+ return self.norm_stats[unnorm_key]["action"]
models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/preprocessor_config.json ADDED
@@ -0,0 +1,114 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_map": {
3
+ "AutoImageProcessor": "processing_prismatic.PrismaticImageProcessor",
4
+ "AutoProcessor": "processing_prismatic.PrismaticProcessor"
5
+ },
6
+ "image_processor_type": "PrismaticImageProcessor",
7
+ "image_resize_strategy": "resize-naive",
8
+ "input_sizes": [
9
+ [
10
+ 3,
11
+ 224,
12
+ 224
13
+ ],
14
+ [
15
+ 3,
16
+ 224,
17
+ 224
18
+ ]
19
+ ],
20
+ "interpolations": [
21
+ "bicubic",
22
+ "bicubic"
23
+ ],
24
+ "means": [
25
+ [
26
+ 0.485,
27
+ 0.456,
28
+ 0.406
29
+ ],
30
+ [
31
+ 0.5,
32
+ 0.5,
33
+ 0.5
34
+ ]
35
+ ],
36
+ "processor_class": "PrismaticProcessor",
37
+ "stds": [
38
+ [
39
+ 0.229,
40
+ 0.224,
41
+ 0.225
42
+ ],
43
+ [
44
+ 0.5,
45
+ 0.5,
46
+ 0.5
47
+ ]
48
+ ],
49
+ "tvf_crop_params": [
50
+ {
51
+ "output_size": [
52
+ 224,
53
+ 224
54
+ ]
55
+ },
56
+ {
57
+ "output_size": [
58
+ 224,
59
+ 224
60
+ ]
61
+ }
62
+ ],
63
+ "tvf_do_letterbox": false,
64
+ "tvf_letterbox_fill": null,
65
+ "tvf_normalize_params": [
66
+ {
67
+ "inplace": false,
68
+ "mean": [
69
+ 0.484375,
70
+ 0.455078125,
71
+ 0.40625
72
+ ],
73
+ "std": [
74
+ 0.228515625,
75
+ 0.2236328125,
76
+ 0.224609375
77
+ ]
78
+ },
79
+ {
80
+ "inplace": false,
81
+ "mean": [
82
+ 0.5,
83
+ 0.5,
84
+ 0.5
85
+ ],
86
+ "std": [
87
+ 0.5,
88
+ 0.5,
89
+ 0.5
90
+ ]
91
+ }
92
+ ],
93
+ "tvf_resize_params": [
94
+ {
95
+ "antialias": true,
96
+ "interpolation": 3,
97
+ "max_size": null,
98
+ "size": [
99
+ 224,
100
+ 224
101
+ ]
102
+ },
103
+ {
104
+ "antialias": true,
105
+ "interpolation": 3,
106
+ "max_size": null,
107
+ "size": [
108
+ 224,
109
+ 224
110
+ ]
111
+ }
112
+ ],
113
+ "use_fused_vision_backbone": true
114
+ }
models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/processing_prismatic.py ADDED
@@ -0,0 +1,257 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """
2
+ processing_prismatic.py
3
+
4
+ HuggingFace-style preprocessor definitions for Prismatic VLMs, inheriting from `ProcessorMixin`. Default configuration
5
+ specifies `siglip-224px+7b`.
6
+ """
7
+
8
+ from typing import Any, ClassVar, List, Optional, Tuple, Union
9
+
10
+ import timm.data
11
+ import torch
12
+ import torchvision.transforms.functional as TVF
13
+ from PIL import Image
14
+ from torchvision.transforms import CenterCrop, Compose, Normalize, Resize, ToTensor
15
+ from transformers import PreTrainedTokenizerBase
16
+ from transformers.image_processing_utils import BatchFeature, ImageProcessingMixin
17
+ from transformers.processing_utils import ProcessorMixin
18
+ from transformers.tokenization_utils import PaddingStrategy, PreTokenizedInput, TextInput, TruncationStrategy
19
+ from transformers.utils import TensorType
20
+
21
+
22
+ # === Image Processing ===
23
+ def letterbox_pad_transform(image: Image.Image, padding_fill_value: Tuple[int, int, int]) -> Image.Image:
24
+ """Given a PIL.Image, pad to square by adding a symmetric border around the height/width."""
25
+ (w, h), max_wh = image.size, max(image.size)
26
+ horizontal_pad, vertical_pad = int((max_wh - w) / 2), int((max_wh - h) / 2)
27
+ padding = (horizontal_pad, vertical_pad, horizontal_pad, vertical_pad)
28
+
29
+ return TVF.pad(image, padding, fill=padding_fill_value, padding_mode="constant")
30
+
31
+
32
+ class PrismaticImageProcessor(ImageProcessingMixin):
33
+ model_input_names: ClassVar[List[str]] = ["pixel_values"]
34
+
35
+ def __init__(
36
+ self,
37
+ use_fused_vision_backbone: bool = False,
38
+ image_resize_strategy: str = "letterbox",
39
+ input_sizes: Optional[List[Tuple[int, int, int]]] = None,
40
+ interpolations: Optional[List[str]] = None,
41
+ means: Optional[List[Tuple[float, float, float]]] = None,
42
+ stds: Optional[List[Tuple[float, float, float]]] = None,
43
+ **kwargs: str,
44
+ ) -> None:
45
+ """
46
+ Initialize a PrismaticImageProcessor as a wrapper around a torchvision transform; this transform will be
47
+ created by TIMM, and edited to follow our custom `image_resize_strategy` logic.
48
+
49
+ @param use_fused_vision_backbone: Boolean indicating single or fused (dual) vision backbone
50
+ @param image_resize_strategy: Prismatic image resize strategy in < resize-naive | resize-crop | letterbox >
51
+ @param input_size: [TIMM :: `data_cfg`] Input image size as tuple (channels, width, height)
52
+ @param interpolation: [TIMM :: `data_cfg`] Interpolation as string (default: "bicubic")
53
+ @param mean: [TIMM :: `data_cfg`] Normalization mean as float tuple (or two-tuple if `fused_backbone`)
54
+ @param std: [TIMM :: `data_cfg`] Normalization std as float tuple (or two-tuple if `fused_backbone`)
55
+ """
56
+ self.use_fused_vision_backbone = use_fused_vision_backbone
57
+ self.image_resize_strategy = image_resize_strategy
58
+
59
+ # Handle `None` default values
60
+ input_sizes = [(3, 224, 224)] if input_sizes is None else input_sizes
61
+ means = [(0.5, 0.5, 0.5)] if means is None else means
62
+ stds = [(0.5, 0.5, 0.5)] if stds is None else stds
63
+
64
+ # TIMM `data_cfg` Parameters
65
+ self.input_sizes, self.interpolations, self.means, self.stds = input_sizes, interpolations, means, stds
66
+
67
+ # Grab torchvision transforms via TIMM =>> need to parse for specific "functional" transform values!
68
+ self.tvf_resize_params, self.tvf_crop_params, self.tvf_normalize_params = [], [], []
69
+ self.tvf_do_letterbox, self.tvf_letterbox_fill = False, None
70
+
71
+ for idx in range(len(input_sizes)):
72
+ transform = timm.data.create_transform(
73
+ input_size=self.input_sizes[idx],
74
+ interpolation=self.interpolations[idx],
75
+ mean=self.means[idx],
76
+ std=self.stds[idx],
77
+ crop_pct=1.0, # Set to 1.0 to ignore cropping (initial Resize sets `input_size`)
78
+ crop_mode="center", # Default crop mode -- no-op when `crop_pct == 1.0`
79
+ is_training=False, # No image augmentations when loading the transform!
80
+ )
81
+
82
+ # [Validation] Ensure appropriate transform structure, expected sizes
83
+ if not (
84
+ isinstance(transform, Compose)
85
+ and (len(transform.transforms) == 4)
86
+ and isinstance(transform.transforms[0], Resize)
87
+ and isinstance(transform.transforms[1], CenterCrop)
88
+ and isinstance(transform.transforms[2], ToTensor)
89
+ and isinstance(transform.transforms[3], Normalize)
90
+ and (transform.transforms[0].size == self.input_sizes[idx][-1])
91
+ and (transform.transforms[1].size == self.input_sizes[idx][-2:])
92
+ ):
93
+ raise ValueError(f"Unexpected TIMM image transformation structure/sizes: `{transform}`")
94
+
95
+ # HF Image Processors *must* be JSON-serializable; as such, cannot have torchvision. as an attribute.
96
+ # => Instead, we're going to parse the transform and call "torchvision.transforms.functional" (`tvf`)
97
+ resize_t, crop_t, norm_t = transform.transforms[0], transform.transforms[1], transform.transforms[3]
98
+ self.tvf_resize_params.append(
99
+ {
100
+ "size": resize_t.size,
101
+ "interpolation": TVF.pil_modes_mapping[resize_t.interpolation],
102
+ "max_size": None,
103
+ "antialias": True,
104
+ }
105
+ )
106
+ self.tvf_crop_params.append({"output_size": crop_t.size})
107
+ self.tvf_normalize_params.append(
108
+ {
109
+ "mean": norm_t.mean.float().numpy().tolist(),
110
+ "std": norm_t.std.float().numpy().tolist(),
111
+ "inplace": False,
112
+ }
113
+ )
114
+ self.tvf_do_letterbox, self.tvf_letterbox_fill = False, None
115
+
116
+ # Handle Prismatic `image_resize_strategy`
117
+ if self.image_resize_strategy == "resize-naive":
118
+ self.tvf_resize_params[idx]["size"] = (resize_t.size, resize_t.size)
119
+ elif self.image_resize_strategy == "letterbox":
120
+ self.tvf_do_letterbox, self.tvf_letterbox_fill = True, tuple([int(x * 255) for x in self.means[idx]])
121
+ elif self.image_resize_strategy == "resize-crop":
122
+ pass
123
+ else:
124
+ raise ValueError(f"Image resize strategy `{self.image_resize_strategy}` is not supported!")
125
+
126
+ # Dispatch **kwargs to super()
127
+ super().__init__(**kwargs)
128
+
129
+ def apply_transform(self, img: Image.Image) -> torch.Tensor:
130
+ """Apply `functional` variant of TIMM's Transform = Compose([Resize -> CenterCrop -> ToTensor -> Normalize])"""
131
+ if self.tvf_do_letterbox:
132
+ img = letterbox_pad_transform(img, self.tvf_letterbox_fill)
133
+
134
+ # [Contract] Fused Backbones expect "channel-stacked" inputs; we'll unpack on the model side!
135
+ imgs_t = []
136
+ for idx in range(len(self.input_sizes)):
137
+ img_idx = TVF.resize(img, **self.tvf_resize_params[idx])
138
+ img_idx = TVF.center_crop(img_idx, **self.tvf_crop_params[idx])
139
+ img_idx_t = TVF.to_tensor(img_idx)
140
+ img_idx_t = TVF.normalize(img_idx_t, **self.tvf_normalize_params[idx])
141
+ imgs_t.append(img_idx_t)
142
+
143
+ # [Contract] `imgs_t` is a list of Tensors of shape [3, input_size, input_size]; stack along dim = 0
144
+ img_t = torch.vstack(imgs_t)
145
+
146
+ return img_t
147
+
148
+ def preprocess(
149
+ self,
150
+ images: Union[Image.Image, List[Image.Image]],
151
+ return_tensors: Optional[Union[str, TensorType]] = None,
152
+ **_: str,
153
+ ) -> BatchFeature:
154
+ """
155
+ Preprocess an image (or batch of images); note that unlike the `transformers :: BaseImageProcessor` we
156
+ explicitly only handle PIL.Image.Image instances for simplicity.
157
+
158
+ @param images: A (batch of) PIL.Image.Image instance(s) to preprocess.
159
+ @param return_tensors: BatchFeature default Tensor format (e.g., "pt" for torch); if None, returns np.ndarray
160
+
161
+ @return: Instance of `transformers :: BatchFeature` with a single key "pixel_values"
162
+ """
163
+ if not isinstance(images, list):
164
+ images = [images]
165
+
166
+ # Apply `self.img_transform` to each image (will return list of torch.Tensors); stack into "batched" Tensor
167
+ pixel_values = torch.stack([self.apply_transform(img.convert("RGB")) for img in images])
168
+
169
+ # Return BatchFeature =>> note that for compatibility, constructor expects Dict[str, np.ndarray], so we convert
170
+ return BatchFeature(data={"pixel_values": pixel_values.float().numpy()}, tensor_type=return_tensors)
171
+
172
+ def __call__(self, images: Union[Image.Image, List[Image.Image]], **kwargs) -> BatchFeature:
173
+ return self.preprocess(images, **kwargs)
174
+
175
+
176
+ # === PrismaticProcessor =>> Wraps both ImageProcessor and Tokenizer ===
177
+ # =>> https://github.com/huggingface/transformers/blob/main/src/transformers/models/llava/processing_llava.py
178
+ class PrismaticProcessor(ProcessorMixin):
179
+ attributes: ClassVar[List[str]] = ["image_processor", "tokenizer"]
180
+ image_processor_class: str = "AutoImageProcessor"
181
+ tokenizer_class: str = "AutoTokenizer"
182
+
183
+ def __init__(
184
+ self,
185
+ image_processor: Optional[ImageProcessingMixin] = None,
186
+ tokenizer: Optional[PreTrainedTokenizerBase] = None,
187
+ ) -> None:
188
+ super().__init__(image_processor, tokenizer)
189
+
190
+ def __call__(
191
+ self,
192
+ text: Union[TextInput, PreTokenizedInput, List[TextInput], List[PreTokenizedInput]],
193
+ images: Union[Image.Image, List[Image.Image]],
194
+ padding: Union[bool, str, PaddingStrategy] = False,
195
+ truncation: Optional[Union[bool, str, TruncationStrategy]] = None,
196
+ max_length: Optional[int] = None,
197
+ return_tensors: Optional[Union[str, TensorType]] = TensorType.PYTORCH,
198
+ ) -> BatchFeature:
199
+ """
200
+ Preprocess a given (batch) of text/images for a Prismatic VLM; forwards text to the underlying LLM's tokenizer,
201
+ forwards images to PrismaticImageProcessor.
202
+
203
+ @param text: The (batch) of text to encode; must be a string or list of strings.
204
+ @param images: A (batch of) PIL.Image.Image instance(s) to preprocess.
205
+ @param padding: Sequence padding strategy (if multiple specified) in < True = "longest" | "max_length" | False >
206
+ @param truncation: Truncation strategy for the output sequences; requires `max_length` to be specified
207
+ @param max_length: Maximum length (in tokens) to truncate
208
+ @param return_tensors: Type of return tensors (usually "pt" or TensorType.PYTORCH)
209
+
210
+ @return: BatchFeature with keys for `input_ids`, `attention_mask` and `pixel_values`.
211
+ """
212
+ pixel_values = self.image_processor(images, return_tensors=return_tensors)["pixel_values"]
213
+ text_inputs = self.tokenizer(
214
+ text, return_tensors=return_tensors, padding=padding, truncation=truncation, max_length=max_length
215
+ )
216
+
217
+ # [Validate] Need same number of images and text inputs!
218
+ if pixel_values.shape[0] != text_inputs.input_ids.shape[0]:
219
+ raise ValueError("Batch is malformed; expected same number of images and text inputs!")
220
+
221
+ return BatchFeature(data={**text_inputs, "pixel_values": pixel_values})
222
+
223
+ # === Tokenizer Dispatch Utilities =>> check `PreTrainedTokenizerBase` for documentation ===
224
+ def batch_decode(
225
+ self,
226
+ sequences: Union[List[int], List[List[int]], torch.Tensor, Any], # `Any` = np.ndarray | tf.Tensor
227
+ skip_special_tokens: bool = False,
228
+ clean_up_tokenization_spaces: Optional[bool] = None,
229
+ **kwargs: str,
230
+ ) -> List[str]:
231
+ return self.tokenizer.batch_decode(
232
+ sequences=sequences,
233
+ skip_special_tokens=skip_special_tokens,
234
+ clean_up_tokenization_spaces=clean_up_tokenization_spaces,
235
+ **kwargs,
236
+ )
237
+
238
+ def decode(
239
+ self,
240
+ token_ids: Union[int, List[int], torch.Tensor, Any], # `Any` = np.ndarray | tf.Tensor
241
+ skip_special_tokens: bool = False,
242
+ clean_up_tokenization_spaces: Optional[bool] = None,
243
+ **kwargs: str,
244
+ ) -> str:
245
+ return self.tokenizer.decode(
246
+ token_ids=token_ids,
247
+ skip_special_tokens=skip_special_tokens,
248
+ clean_up_tokenization_spaces=clean_up_tokenization_spaces,
249
+ **kwargs,
250
+ )
251
+
252
+ @property
253
+ def model_input_names(self) -> List[str]:
254
+ tokenizer_input_names = self.tokenizer.model_input_names
255
+ image_processor_input_names = self.image_processor.model_input_names
256
+
257
+ return list(dict.fromkeys(tokenizer_input_names + image_processor_input_names))
models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/processor_config.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "auto_map": {
3
+ "AutoProcessor": "processing_prismatic.PrismaticProcessor"
4
+ },
5
+ "processor_class": "PrismaticProcessor"
6
+ }
models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/proprio_projector--2000_checkpoint.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f6eda04bf3ae0c565124063c1fd1d7fb716f6a00f4234666e197325927eb4232
3
+ size 67259307
models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/special_tokens_map.json ADDED
@@ -0,0 +1,37 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "mask_token": {
17
+ "content": "<mask>",
18
+ "lstrip": false,
19
+ "normalized": false,
20
+ "rstrip": false,
21
+ "single_word": false
22
+ },
23
+ "pad_token": {
24
+ "content": "<PAD>",
25
+ "lstrip": false,
26
+ "normalized": false,
27
+ "rstrip": false,
28
+ "single_word": false
29
+ },
30
+ "unk_token": {
31
+ "content": "<unk>",
32
+ "lstrip": false,
33
+ "normalized": false,
34
+ "rstrip": false,
35
+ "single_word": false
36
+ }
37
+ }
models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
models/piper-pick-white-block-50hz-uniform-left-v1-span2-stage1-25k-d2f-block7-4gpu-b32-2k-20260826/tokenizer_config.json ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "added_tokens_decoder": {
5
+ "0": {
6
+ "content": "<unk>",
7
+ "lstrip": false,
8
+ "normalized": false,
9
+ "rstrip": false,
10
+ "single_word": false,
11
+ "special": true
12
+ },
13
+ "1": {
14
+ "content": "<s>",
15
+ "lstrip": false,
16
+ "normalized": false,
17
+ "rstrip": false,
18
+ "single_word": false,
19
+ "special": true
20
+ },
21
+ "2": {
22
+ "content": "</s>",
23
+ "lstrip": false,
24
+ "normalized": false,
25
+ "rstrip": false,
26
+ "single_word": false,
27
+ "special": true
28
+ },
29
+ "32000": {
30
+ "content": "<PAD>",
31
+ "lstrip": false,
32
+ "normalized": false,
33
+ "rstrip": false,
34
+ "single_word": false,
35
+ "special": true
36
+ },
37
+ "32001": {
38
+ "content": "<mask>",
39
+ "lstrip": false,
40
+ "normalized": false,
41
+ "rstrip": false,
42
+ "single_word": false,
43
+ "special": true
44
+ }
45
+ },
46
+ "auto_map": {
47
+ "AutoProcessor": "processing_prismatic.PrismaticProcessor"
48
+ },
49
+ "bos_token": "<s>",
50
+ "clean_up_tokenization_spaces": false,
51
+ "eos_token": "</s>",
52
+ "legacy": false,
53
+ "mask_token": "<mask>",
54
+ "model_max_length": 2048,
55
+ "pad_token": "<PAD>",
56
+ "padding_side": "right",
57
+ "processor_class": "PrismaticProcessor",
58
+ "sp_model_kwargs": {},
59
+ "tokenizer_class": "LlamaTokenizer",
60
+ "unk_token": "<unk>",
61
+ "use_default_system_prompt": false
62
+ }