masonlimx commited on
Commit
70f056d
·
verified ·
1 Parent(s): aa777fc

Rename dit4dit_libero_all_full_finetune_bc071270ff2_bs256 to dit4dit_libero_all_full_finetune_bs256

Browse files
dit4dit_libero_all_full_finetune_bc071270ff2_bs256/checkpoints/step-120000-epoch-147-loss=0.0439.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:828f5700102fa9f5ac54f2f8cca015263a74ef7dd75a187759e3a1042b625a7f
3
- size 25727358750
 
 
 
 
dit4dit_libero_all_full_finetune_bc071270ff2_bs256/config.json DELETED
@@ -1,1359 +0,0 @@
1
- {
2
- "_action_dim": 8,
3
- "_action_horizon": 8,
4
- "_action_norm_mask": [
5
- true,
6
- true,
7
- true,
8
- true,
9
- true,
10
- true,
11
- false
12
- ],
13
- "_ckpt_root": "./checkpoints",
14
- "_cosmos_base_model": "./checkpoints/Cosmos-Predict2.5-2B",
15
- "_cosmos_revision": "diffusers/base/post-trained",
16
- "_cosmos_tokenizer": {
17
- "model_max_length": 512,
18
- "model_path": "./checkpoints/Cosmos-Predict2.5-2B/tokenizer",
19
- "type": "PretrainedTokenizer"
20
- },
21
- "_dit4dit_parquet_dataset": {
22
- "action_key": "action",
23
- "action_window_size": 8,
24
- "frame_window_size": 9,
25
- "statistic_name": "franka",
26
- "transforms": [
27
- {
28
- "frame_stride": 2,
29
- "name_mappings": {
30
- "actions": [
31
- "actions"
32
- ],
33
- "observation.state": [
34
- "states"
35
- ]
36
- },
37
- "parquet_keys": [
38
- "observation.state",
39
- "timestamp",
40
- "actions",
41
- "info",
42
- "stats",
43
- "action_masks"
44
- ],
45
- "type": "ProcessParquetInputs",
46
- "video_keys": [
47
- "observation.images.image",
48
- "observation.images.wrist_image"
49
- ]
50
- },
51
- {
52
- "input_keys": [
53
- "prompt",
54
- "task_description",
55
- "lang",
56
- "instruction",
57
- "instructions",
58
- "text"
59
- ],
60
- "output_key": "prompt",
61
- "remove_source_keys": true,
62
- "type": "CanonicalizePrompt"
63
- },
64
- {
65
- "input_key": "prompt",
66
- "remove_input_key": true,
67
- "tokenizer": {
68
- "model_max_length": 512,
69
- "model_path": "./checkpoints/Cosmos-Predict2.5-2B/tokenizer",
70
- "type": "PretrainedTokenizer"
71
- },
72
- "type": "ProcessCosmos25Prompt"
73
- },
74
- {
75
- "backend": "torch",
76
- "height": 224,
77
- "output_layout": "nchw",
78
- "scale_divisor": 255.0,
79
- "type": "ResizeImages",
80
- "width": 224
81
- },
82
- {
83
- "frame_stride": 1,
84
- "keep_time_dim": true,
85
- "key": "images",
86
- "num_views": 2,
87
- "type": "ConcatImagesHorizontally",
88
- "views_first": true
89
- },
90
- {
91
- "backend": "torch",
92
- "interleave": true,
93
- "keys": [
94
- "states"
95
- ],
96
- "target_dims": {
97
- "states": 16
98
- },
99
- "type": "SinCosKeys"
100
- },
101
- {
102
- "action_dim": null,
103
- "action_key": "action",
104
- "action_norm_mask": [
105
- true,
106
- true,
107
- true,
108
- true,
109
- true,
110
- true,
111
- false
112
- ],
113
- "action_norm_type": "min_max",
114
- "clip_norm": false,
115
- "norm_type": "min_max",
116
- "normalization_epsilon": 0.0,
117
- "output_dtype": "float16",
118
- "preserve_input_dtype": true,
119
- "state_dim": null,
120
- "state_key": "proprio",
121
- "state_norm_type": "none",
122
- "type": "NormalizeStatesAndActions"
123
- },
124
- {
125
- "action_dim": 8,
126
- "type": "PadActionsAndActionMasks",
127
- "valid_action_dim": 7
128
- },
129
- {
130
- "action_dim": 8,
131
- "action_horizon": 8,
132
- "mark_all_action_steps_valid": true,
133
- "state_dim": 16,
134
- "type": "PrepareDiT4DiTInputs",
135
- "valid_action_dim": 7
136
- }
137
- ],
138
- "type": "ParquetDataset",
139
- "use_delta": false,
140
- "window_start_idx": 0
141
- },
142
- "_dit4dit_train_transforms": [
143
- {
144
- "frame_stride": 2,
145
- "name_mappings": {
146
- "actions": [
147
- "actions"
148
- ],
149
- "observation.state": [
150
- "states"
151
- ]
152
- },
153
- "parquet_keys": [
154
- "observation.state",
155
- "timestamp",
156
- "actions",
157
- "info",
158
- "stats",
159
- "action_masks"
160
- ],
161
- "type": "ProcessParquetInputs",
162
- "video_keys": [
163
- "observation.images.image",
164
- "observation.images.wrist_image"
165
- ]
166
- },
167
- {
168
- "input_keys": [
169
- "prompt",
170
- "task_description",
171
- "lang",
172
- "instruction",
173
- "instructions",
174
- "text"
175
- ],
176
- "output_key": "prompt",
177
- "remove_source_keys": true,
178
- "type": "CanonicalizePrompt"
179
- },
180
- {
181
- "input_key": "prompt",
182
- "remove_input_key": true,
183
- "tokenizer": {
184
- "model_max_length": 512,
185
- "model_path": "./checkpoints/Cosmos-Predict2.5-2B/tokenizer",
186
- "type": "PretrainedTokenizer"
187
- },
188
- "type": "ProcessCosmos25Prompt"
189
- },
190
- {
191
- "backend": "torch",
192
- "height": 224,
193
- "output_layout": "nchw",
194
- "scale_divisor": 255.0,
195
- "type": "ResizeImages",
196
- "width": 224
197
- },
198
- {
199
- "frame_stride": 1,
200
- "keep_time_dim": true,
201
- "key": "images",
202
- "num_views": 2,
203
- "type": "ConcatImagesHorizontally",
204
- "views_first": true
205
- },
206
- {
207
- "backend": "torch",
208
- "interleave": true,
209
- "keys": [
210
- "states"
211
- ],
212
- "target_dims": {
213
- "states": 16
214
- },
215
- "type": "SinCosKeys"
216
- },
217
- {
218
- "action_dim": null,
219
- "action_key": "action",
220
- "action_norm_mask": [
221
- true,
222
- true,
223
- true,
224
- true,
225
- true,
226
- true,
227
- false
228
- ],
229
- "action_norm_type": "min_max",
230
- "clip_norm": false,
231
- "norm_type": "min_max",
232
- "normalization_epsilon": 0.0,
233
- "output_dtype": "float16",
234
- "preserve_input_dtype": true,
235
- "state_dim": null,
236
- "state_key": "proprio",
237
- "state_norm_type": "none",
238
- "type": "NormalizeStatesAndActions"
239
- },
240
- {
241
- "action_dim": 8,
242
- "type": "PadActionsAndActionMasks",
243
- "valid_action_dim": 7
244
- },
245
- {
246
- "action_dim": 8,
247
- "action_horizon": 8,
248
- "mark_all_action_steps_valid": true,
249
- "state_dim": 16,
250
- "type": "PrepareDiT4DiTInputs",
251
- "valid_action_dim": 7
252
- }
253
- ],
254
- "_frame_window_size": 9,
255
- "_image_frame_stride": 2,
256
- "_image_size": 224,
257
- "_libero_data_roots": [
258
- "./datasets/libero_object_no_noops_lerobotv2.1",
259
- "./datasets/libero_goal_no_noops_lerobotv2.1",
260
- "./datasets/libero_spatial_no_noops_lerobotv2.1",
261
- "./datasets/libero_10_no_noops_lerobotv2.1"
262
- ],
263
- "_official_dataset_statistics": {
264
- "franka": {
265
- "action": {
266
- "mask": [
267
- true,
268
- true,
269
- true,
270
- true,
271
- true,
272
- true,
273
- false
274
- ],
275
- "max": [
276
- 0.9375,
277
- 0.9375,
278
- 0.9375,
279
- 0.3557142913341522,
280
- 0.375,
281
- 0.375,
282
- 1.0
283
- ],
284
- "mean": [
285
- 0.07237596483901143,
286
- 0.08987006871029735,
287
- -0.10144743137061596,
288
- -0.00045383188989944756,
289
- 0.006273590726777911,
290
- -0.003878799732774496,
291
- 0.524486355483532
292
- ],
293
- "min": [
294
- -0.9375,
295
- -0.9375,
296
- -0.9375,
297
- -0.2582142949104309,
298
- -0.375,
299
- -0.3675000071525574,
300
- 0.0
301
- ],
302
- "q01": [
303
- -0.8785714507102966,
304
- -0.8758928775787354,
305
- -0.9375,
306
- -0.1510714292526245,
307
- -0.20678570866584778,
308
- -0.2742857038974762,
309
- 0.0
310
- ],
311
- "q99": [
312
- 0.9375,
313
- 0.9107142686843872,
314
- 0.9375,
315
- 0.20357142388820648,
316
- 0.26357144117355347,
317
- 0.375,
318
- 1.0
319
- ],
320
- "std": [
321
- 0.3498823308902479,
322
- 0.37794140366375184,
323
- 0.460084266976933,
324
- 0.0403885784928603,
325
- 0.06616144248501059,
326
- 0.07763074391911857,
327
- 0.4994683356809767
328
- ]
329
- },
330
- "num_trajectories": 1693,
331
- "num_transitions": 273465,
332
- "state": {
333
- "max": [
334
- 0.21031762659549713,
335
- 0.39128610491752625,
336
- 1.3660105466842651,
337
- 3.6714255809783936,
338
- 3.560650587081909,
339
- 1.386339545249939,
340
- 0.04233968257904053,
341
- 0.0013633022317662835
342
- ],
343
- "mean": [
344
- -0.04889854742214084,
345
- 0.03689368185587227,
346
- 0.7890402488410473,
347
- 2.9771945476531982,
348
- -0.1417286954820156,
349
- -0.11769362539052963,
350
- 0.026436020154505968,
351
- -0.02665513101965189
352
- ],
353
- "min": [
354
- -0.4828203022480011,
355
- -0.3255046010017395,
356
- 0.008128180168569088,
357
- 0.35277295112609863,
358
- -3.641430377960205,
359
- -1.842738389968872,
360
- -0.0013586411951109767,
361
- -0.042040832340717316
362
- ],
363
- "q01": [
364
- -0.42401049643754957,
365
- -0.2838300323486328,
366
- 0.009925739830359817,
367
- 1.3085840785503386,
368
- -2.886677579879761,
369
- -1.1599004411697387,
370
- 0.001503719249740243,
371
- -0.040336399003863335
372
- ],
373
- "q99": [
374
- 0.1530261474847791,
375
- 0.3629165390133857,
376
- 1.2910678112506866,
377
- 3.303542451858519,
378
- 2.7496529006957933,
379
- 0.6893712210655194,
380
- 0.040610933862626555,
381
- -0.0015016929572448147
382
- ],
383
- "std": [
384
- 0.10639013941746686,
385
- 0.15115733130675715,
386
- 0.38406895599530033,
387
- 0.3530238395244304,
388
- 0.8227341427331599,
389
- 0.32357567121520087,
390
- 0.014583991652936385,
391
- 0.014467005007200339
392
- ]
393
- }
394
- }
395
- },
396
- "_official_dit4dit_ckpt": "./checkpoints/dit4dit-model/dit4dit_libero/final_model/pytorch_model.pt",
397
- "_ori_action_dim": 7,
398
- "_prompt_input_keys": [
399
- "prompt",
400
- "task_description",
401
- "lang",
402
- "instruction",
403
- "instructions",
404
- "text"
405
- ],
406
- "_state_dim": 16,
407
- "eval": {
408
- "dataset": {
409
- "img_buffer_len": 1,
410
- "require_lang_tokens": true,
411
- "transforms": [
412
- {
413
- "input_keys": [
414
- "prompt",
415
- "task_description",
416
- "lang",
417
- "instruction",
418
- "instructions",
419
- "text"
420
- ],
421
- "output_key": "prompt",
422
- "remove_source_keys": true,
423
- "type": "CanonicalizePrompt"
424
- },
425
- {
426
- "input_key": "prompt",
427
- "remove_input_key": true,
428
- "tokenizer": {
429
- "model_max_length": 512,
430
- "model_path": "./checkpoints/Cosmos-Predict2.5-2B/tokenizer",
431
- "type": "PretrainedTokenizer"
432
- },
433
- "type": "ProcessCosmos25Prompt"
434
- },
435
- {
436
- "image_size": 224,
437
- "img_keys": [
438
- "agentview_image",
439
- "robot0_eye_in_hand_image"
440
- ],
441
- "type": "ProcessDiT4DiTLiberoEvalInputs"
442
- },
443
- {
444
- "gripper_key": "robot0_gripper_qpos",
445
- "norm_type": "none",
446
- "out_key": "states",
447
- "pos_key": "robot0_eef_pos",
448
- "quat_key": "robot0_eef_quat",
449
- "state_dim": null,
450
- "type": "LiberoProprioFromInputs"
451
- },
452
- {
453
- "interleave": true,
454
- "keys": [
455
- "states"
456
- ],
457
- "target_dims": {
458
- "states": 16
459
- },
460
- "type": "SinCosKeys"
461
- },
462
- {
463
- "image_key": "pixel_values",
464
- "require_actions": false,
465
- "state_dim": 16,
466
- "type": "PrepareDiT4DiTInputs"
467
- }
468
- ],
469
- "type": "LiberoParquetEvalDataset"
470
- },
471
- "denormalize_action": {
472
- "action_dim": 7,
473
- "action_norm_mask": [
474
- true,
475
- true,
476
- true,
477
- true,
478
- true,
479
- true,
480
- false
481
- ],
482
- "clip_normalized_action": true,
483
- "invert_gripper_action": true,
484
- "norm_type": "min_max",
485
- "normalize_gripper_action": true,
486
- "type": "DenormalizeLiberoAction"
487
- },
488
- "enable_mixed_precision_training": false,
489
- "eval_chunk_size": 8,
490
- "mixed_precision_dtype": "bf16",
491
- "model_family": "dit4dit",
492
- "norm_stats_key": "franka",
493
- "num_steps_wait": 10,
494
- "num_trials_per_task": 50,
495
- "resize_size": 224,
496
- "seed": 7,
497
- "task_suite_name": [
498
- "libero_spatial",
499
- "libero_object",
500
- "libero_goal",
501
- "libero_10"
502
- ],
503
- "truncate_action_chunk_at_horizon": true,
504
- "type": "LiberoEvalRunner"
505
- },
506
- "inference_model": {
507
- "freeze_vlm_backbone": false,
508
- "name_mapping": null,
509
- "pretrained_name_or_path": null,
510
- "repeated_diffusion_steps": 4,
511
- "strict_mapping": false,
512
- "type": "DiT4DiTVLA",
513
- "vla_head": {
514
- "action_dim": 8,
515
- "action_horizon": 8,
516
- "action_model_type": "DiT-B",
517
- "add_pos_embed": true,
518
- "diffusion_model_cfg": {
519
- "cross_attention_dim": 2048,
520
- "dropout": 0.2,
521
- "final_dropout": true,
522
- "interleave_self_attention": true,
523
- "norm_type": "ada_norm",
524
- "num_layers": 16,
525
- "output_dim": 2560,
526
- "positional_embeddings": null
527
- },
528
- "future_action_window_size": 7,
529
- "hidden_size": 2560,
530
- "max_seq_len": 1024,
531
- "noise_beta_alpha": 1.5,
532
- "noise_beta_beta": 1.0,
533
- "noise_s": 0.999,
534
- "num_inference_timesteps": 4,
535
- "num_timestep_buckets": 1000,
536
- "ori_action_dim": 7,
537
- "state_dim": 16,
538
- "type": "DiT4DiTActionHead"
539
- },
540
- "vlm_backbone": {
541
- "base_model": "./checkpoints/Cosmos-Predict2.5-2B",
542
- "conditional_frame_timestep": 0.0001,
543
- "detach_hidden_states": true,
544
- "extract_layer": 17,
545
- "fixed_seed": null,
546
- "flow_matching_high_sigma_min": null,
547
- "flow_matching_high_sigma_ratio": null,
548
- "flow_matching_time_distribution": "uniform",
549
- "frozen_submodules": [
550
- "text_encoder",
551
- "vae"
552
- ],
553
- "fsdp_min_num_params": 0,
554
- "future_loss_type": "flow_matching",
555
- "local_files_only": true,
556
- "max_sequence_length": 512,
557
- "num_frames_out": 5,
558
- "num_inference_steps": 1,
559
- "revision": "diffusers/base/post-trained",
560
- "split_future_frames": true,
561
- "torch_dtype": "bf16",
562
- "trainable": true,
563
- "type": "Cosmos25Backbone"
564
- }
565
- },
566
- "model": {
567
- "freeze_vlm_backbone": false,
568
- "name_mapping": {
569
- "vla_head": "action_model",
570
- "vlm_backbone.text_encoder": "backbone_interface.extractor.text_encoder",
571
- "vlm_backbone.transformer": "backbone_interface.extractor.transformer",
572
- "vlm_backbone.vae": "backbone_interface.extractor.vae"
573
- },
574
- "pretrained_name_or_path": null,
575
- "repeated_diffusion_steps": 4,
576
- "strict_mapping": false,
577
- "type": "DiT4DiTVLA",
578
- "vla_head": {
579
- "action_dim": 8,
580
- "action_horizon": 8,
581
- "action_model_type": "DiT-B",
582
- "add_pos_embed": true,
583
- "diffusion_model_cfg": {
584
- "cross_attention_dim": 2048,
585
- "dropout": 0.2,
586
- "final_dropout": true,
587
- "interleave_self_attention": true,
588
- "norm_type": "ada_norm",
589
- "num_layers": 16,
590
- "output_dim": 2560,
591
- "positional_embeddings": null
592
- },
593
- "future_action_window_size": 7,
594
- "hidden_size": 2560,
595
- "max_seq_len": 1024,
596
- "noise_beta_alpha": 1.5,
597
- "noise_beta_beta": 1.0,
598
- "noise_s": 0.999,
599
- "num_inference_timesteps": 4,
600
- "num_timestep_buckets": 1000,
601
- "ori_action_dim": 7,
602
- "state_dim": 16,
603
- "type": "DiT4DiTActionHead"
604
- },
605
- "vlm_backbone": {
606
- "base_model": "./checkpoints/Cosmos-Predict2.5-2B",
607
- "conditional_frame_timestep": 0.0001,
608
- "detach_hidden_states": true,
609
- "extract_layer": 17,
610
- "fixed_seed": null,
611
- "flow_matching_high_sigma_min": null,
612
- "flow_matching_high_sigma_ratio": null,
613
- "flow_matching_time_distribution": "uniform",
614
- "frozen_submodules": [
615
- "text_encoder",
616
- "vae"
617
- ],
618
- "fsdp_min_num_params": 0,
619
- "future_loss_type": "flow_matching",
620
- "local_files_only": true,
621
- "max_sequence_length": 512,
622
- "num_frames_out": 5,
623
- "num_inference_steps": 1,
624
- "revision": "diffusers/base/post-trained",
625
- "split_future_frames": true,
626
- "torch_dtype": "bf16",
627
- "trainable": true,
628
- "type": "Cosmos25Backbone"
629
- }
630
- },
631
- "official_eval_task_suites": [
632
- "libero_spatial",
633
- "libero_object",
634
- "libero_goal",
635
- "libero_10"
636
- ],
637
- "runner": {
638
- "change_key_name": false,
639
- "collator": {
640
- "keys": [
641
- "states",
642
- "timestamp",
643
- "images",
644
- "img_masks",
645
- "actions",
646
- "action_masks",
647
- "frame_masks",
648
- "lang_tokens",
649
- "lang_masks"
650
- ],
651
- "meta_keys": [
652
- "info",
653
- "stats"
654
- ],
655
- "type": "DictCollator"
656
- },
657
- "enable_gradient_checkpointing": false,
658
- "enable_mixed_precision_training": true,
659
- "grad_accumulation_steps": 2,
660
- "lr_scheduler": {
661
- "min_lr": 5e-07,
662
- "type": "cosine_with_min_lr",
663
- "warmup_steps": 10000
664
- },
665
- "max_grad_norm": 1.0,
666
- "max_keep_ckpts": 2,
667
- "max_steps": 160000,
668
- "metric": {
669
- "active_trackers": [
670
- "jsonl",
671
- "wandb"
672
- ],
673
- "grad_accumulation_steps": 1,
674
- "run_dir": "work_dirs",
675
- "type": "VLAMetric",
676
- "window_size": 1
677
- },
678
- "mixed_precision_dtype": "bf16",
679
- "optimizer": {
680
- "betas": [
681
- 0.9,
682
- 0.95
683
- ],
684
- "eps": 1e-08,
685
- "lr": 3e-05,
686
- "paramwise_learning_rate": {
687
- "vla_head": 0.0001,
688
- "vlm_backbone.transformer": 0.0001
689
- },
690
- "type": "AdamW",
691
- "weight_decay": 1e-08
692
- },
693
- "pre_fsdp_param_dtype": "fp32",
694
- "reduce_in_full_precision": false,
695
- "sampler": null,
696
- "save_iter_interval": 40000,
697
- "sharding_strategy": "shard-grad-op",
698
- "tokenizer": {
699
- "model_max_length": 512,
700
- "model_path": "./checkpoints/Cosmos-Predict2.5-2B/tokenizer",
701
- "type": "PretrainedTokenizer"
702
- },
703
- "type": "FSDPTrainRunner"
704
- },
705
- "seed": 42,
706
- "train_dataloader": {
707
- "dataset": {
708
- "dataset_statistics": {
709
- "franka": {
710
- "action": {
711
- "mask": [
712
- true,
713
- true,
714
- true,
715
- true,
716
- true,
717
- true,
718
- false
719
- ],
720
- "max": [
721
- 0.9375,
722
- 0.9375,
723
- 0.9375,
724
- 0.3557142913341522,
725
- 0.375,
726
- 0.375,
727
- 1.0
728
- ],
729
- "mean": [
730
- 0.07237596483901143,
731
- 0.08987006871029735,
732
- -0.10144743137061596,
733
- -0.00045383188989944756,
734
- 0.006273590726777911,
735
- -0.003878799732774496,
736
- 0.524486355483532
737
- ],
738
- "min": [
739
- -0.9375,
740
- -0.9375,
741
- -0.9375,
742
- -0.2582142949104309,
743
- -0.375,
744
- -0.3675000071525574,
745
- 0.0
746
- ],
747
- "q01": [
748
- -0.8785714507102966,
749
- -0.8758928775787354,
750
- -0.9375,
751
- -0.1510714292526245,
752
- -0.20678570866584778,
753
- -0.2742857038974762,
754
- 0.0
755
- ],
756
- "q99": [
757
- 0.9375,
758
- 0.9107142686843872,
759
- 0.9375,
760
- 0.20357142388820648,
761
- 0.26357144117355347,
762
- 0.375,
763
- 1.0
764
- ],
765
- "std": [
766
- 0.3498823308902479,
767
- 0.37794140366375184,
768
- 0.460084266976933,
769
- 0.0403885784928603,
770
- 0.06616144248501059,
771
- 0.07763074391911857,
772
- 0.4994683356809767
773
- ]
774
- },
775
- "num_trajectories": 1693,
776
- "num_transitions": 273465,
777
- "state": {
778
- "max": [
779
- 0.21031762659549713,
780
- 0.39128610491752625,
781
- 1.3660105466842651,
782
- 3.6714255809783936,
783
- 3.560650587081909,
784
- 1.386339545249939,
785
- 0.04233968257904053,
786
- 0.0013633022317662835
787
- ],
788
- "mean": [
789
- -0.04889854742214084,
790
- 0.03689368185587227,
791
- 0.7890402488410473,
792
- 2.9771945476531982,
793
- -0.1417286954820156,
794
- -0.11769362539052963,
795
- 0.026436020154505968,
796
- -0.02665513101965189
797
- ],
798
- "min": [
799
- -0.4828203022480011,
800
- -0.3255046010017395,
801
- 0.008128180168569088,
802
- 0.35277295112609863,
803
- -3.641430377960205,
804
- -1.842738389968872,
805
- -0.0013586411951109767,
806
- -0.042040832340717316
807
- ],
808
- "q01": [
809
- -0.42401049643754957,
810
- -0.2838300323486328,
811
- 0.009925739830359817,
812
- 1.3085840785503386,
813
- -2.886677579879761,
814
- -1.1599004411697387,
815
- 0.001503719249740243,
816
- -0.040336399003863335
817
- ],
818
- "q99": [
819
- 0.1530261474847791,
820
- 0.3629165390133857,
821
- 1.2910678112506866,
822
- 3.303542451858519,
823
- 2.7496529006957933,
824
- 0.6893712210655194,
825
- 0.040610933862626555,
826
- -0.0015016929572448147
827
- ],
828
- "std": [
829
- 0.10639013941746686,
830
- 0.15115733130675715,
831
- 0.38406895599530033,
832
- 0.3530238395244304,
833
- 0.8227341427331599,
834
- 0.32357567121520087,
835
- 0.014583991652936385,
836
- 0.014467005007200339
837
- ]
838
- }
839
- }
840
- },
841
- "datasets": [
842
- {
843
- "action_key": "action",
844
- "action_window_size": 8,
845
- "data_root_path": "./datasets/libero_object_no_noops_lerobotv2.1",
846
- "frame_window_size": 9,
847
- "statistic_name": "franka",
848
- "transforms": [
849
- {
850
- "frame_stride": 2,
851
- "name_mappings": {
852
- "actions": [
853
- "actions"
854
- ],
855
- "observation.state": [
856
- "states"
857
- ]
858
- },
859
- "parquet_keys": [
860
- "observation.state",
861
- "timestamp",
862
- "actions",
863
- "info",
864
- "stats",
865
- "action_masks"
866
- ],
867
- "type": "ProcessParquetInputs",
868
- "video_keys": [
869
- "observation.images.image",
870
- "observation.images.wrist_image"
871
- ]
872
- },
873
- {
874
- "input_keys": [
875
- "prompt",
876
- "task_description",
877
- "lang",
878
- "instruction",
879
- "instructions",
880
- "text"
881
- ],
882
- "output_key": "prompt",
883
- "remove_source_keys": true,
884
- "type": "CanonicalizePrompt"
885
- },
886
- {
887
- "input_key": "prompt",
888
- "remove_input_key": true,
889
- "tokenizer": {
890
- "model_max_length": 512,
891
- "model_path": "./checkpoints/Cosmos-Predict2.5-2B/tokenizer",
892
- "type": "PretrainedTokenizer"
893
- },
894
- "type": "ProcessCosmos25Prompt"
895
- },
896
- {
897
- "backend": "torch",
898
- "height": 224,
899
- "output_layout": "nchw",
900
- "scale_divisor": 255.0,
901
- "type": "ResizeImages",
902
- "width": 224
903
- },
904
- {
905
- "frame_stride": 1,
906
- "keep_time_dim": true,
907
- "key": "images",
908
- "num_views": 2,
909
- "type": "ConcatImagesHorizontally",
910
- "views_first": true
911
- },
912
- {
913
- "backend": "torch",
914
- "interleave": true,
915
- "keys": [
916
- "states"
917
- ],
918
- "target_dims": {
919
- "states": 16
920
- },
921
- "type": "SinCosKeys"
922
- },
923
- {
924
- "action_dim": null,
925
- "action_key": "action",
926
- "action_norm_mask": [
927
- true,
928
- true,
929
- true,
930
- true,
931
- true,
932
- true,
933
- false
934
- ],
935
- "action_norm_type": "min_max",
936
- "clip_norm": false,
937
- "norm_type": "min_max",
938
- "normalization_epsilon": 0.0,
939
- "output_dtype": "float16",
940
- "preserve_input_dtype": true,
941
- "state_dim": null,
942
- "state_key": "proprio",
943
- "state_norm_type": "none",
944
- "type": "NormalizeStatesAndActions"
945
- },
946
- {
947
- "action_dim": 8,
948
- "type": "PadActionsAndActionMasks",
949
- "valid_action_dim": 7
950
- },
951
- {
952
- "action_dim": 8,
953
- "action_horizon": 8,
954
- "mark_all_action_steps_valid": true,
955
- "state_dim": 16,
956
- "type": "PrepareDiT4DiTInputs",
957
- "valid_action_dim": 7
958
- }
959
- ],
960
- "type": "ParquetDataset",
961
- "use_delta": false,
962
- "window_start_idx": 0
963
- },
964
- {
965
- "action_key": "action",
966
- "action_window_size": 8,
967
- "data_root_path": "./datasets/libero_goal_no_noops_lerobotv2.1",
968
- "frame_window_size": 9,
969
- "statistic_name": "franka",
970
- "transforms": [
971
- {
972
- "frame_stride": 2,
973
- "name_mappings": {
974
- "actions": [
975
- "actions"
976
- ],
977
- "observation.state": [
978
- "states"
979
- ]
980
- },
981
- "parquet_keys": [
982
- "observation.state",
983
- "timestamp",
984
- "actions",
985
- "info",
986
- "stats",
987
- "action_masks"
988
- ],
989
- "type": "ProcessParquetInputs",
990
- "video_keys": [
991
- "observation.images.image",
992
- "observation.images.wrist_image"
993
- ]
994
- },
995
- {
996
- "input_keys": [
997
- "prompt",
998
- "task_description",
999
- "lang",
1000
- "instruction",
1001
- "instructions",
1002
- "text"
1003
- ],
1004
- "output_key": "prompt",
1005
- "remove_source_keys": true,
1006
- "type": "CanonicalizePrompt"
1007
- },
1008
- {
1009
- "input_key": "prompt",
1010
- "remove_input_key": true,
1011
- "tokenizer": {
1012
- "model_max_length": 512,
1013
- "model_path": "./checkpoints/Cosmos-Predict2.5-2B/tokenizer",
1014
- "type": "PretrainedTokenizer"
1015
- },
1016
- "type": "ProcessCosmos25Prompt"
1017
- },
1018
- {
1019
- "backend": "torch",
1020
- "height": 224,
1021
- "output_layout": "nchw",
1022
- "scale_divisor": 255.0,
1023
- "type": "ResizeImages",
1024
- "width": 224
1025
- },
1026
- {
1027
- "frame_stride": 1,
1028
- "keep_time_dim": true,
1029
- "key": "images",
1030
- "num_views": 2,
1031
- "type": "ConcatImagesHorizontally",
1032
- "views_first": true
1033
- },
1034
- {
1035
- "backend": "torch",
1036
- "interleave": true,
1037
- "keys": [
1038
- "states"
1039
- ],
1040
- "target_dims": {
1041
- "states": 16
1042
- },
1043
- "type": "SinCosKeys"
1044
- },
1045
- {
1046
- "action_dim": null,
1047
- "action_key": "action",
1048
- "action_norm_mask": [
1049
- true,
1050
- true,
1051
- true,
1052
- true,
1053
- true,
1054
- true,
1055
- false
1056
- ],
1057
- "action_norm_type": "min_max",
1058
- "clip_norm": false,
1059
- "norm_type": "min_max",
1060
- "normalization_epsilon": 0.0,
1061
- "output_dtype": "float16",
1062
- "preserve_input_dtype": true,
1063
- "state_dim": null,
1064
- "state_key": "proprio",
1065
- "state_norm_type": "none",
1066
- "type": "NormalizeStatesAndActions"
1067
- },
1068
- {
1069
- "action_dim": 8,
1070
- "type": "PadActionsAndActionMasks",
1071
- "valid_action_dim": 7
1072
- },
1073
- {
1074
- "action_dim": 8,
1075
- "action_horizon": 8,
1076
- "mark_all_action_steps_valid": true,
1077
- "state_dim": 16,
1078
- "type": "PrepareDiT4DiTInputs",
1079
- "valid_action_dim": 7
1080
- }
1081
- ],
1082
- "type": "ParquetDataset",
1083
- "use_delta": false,
1084
- "window_start_idx": 0
1085
- },
1086
- {
1087
- "action_key": "action",
1088
- "action_window_size": 8,
1089
- "data_root_path": "./datasets/libero_spatial_no_noops_lerobotv2.1",
1090
- "frame_window_size": 9,
1091
- "statistic_name": "franka",
1092
- "transforms": [
1093
- {
1094
- "frame_stride": 2,
1095
- "name_mappings": {
1096
- "actions": [
1097
- "actions"
1098
- ],
1099
- "observation.state": [
1100
- "states"
1101
- ]
1102
- },
1103
- "parquet_keys": [
1104
- "observation.state",
1105
- "timestamp",
1106
- "actions",
1107
- "info",
1108
- "stats",
1109
- "action_masks"
1110
- ],
1111
- "type": "ProcessParquetInputs",
1112
- "video_keys": [
1113
- "observation.images.image",
1114
- "observation.images.wrist_image"
1115
- ]
1116
- },
1117
- {
1118
- "input_keys": [
1119
- "prompt",
1120
- "task_description",
1121
- "lang",
1122
- "instruction",
1123
- "instructions",
1124
- "text"
1125
- ],
1126
- "output_key": "prompt",
1127
- "remove_source_keys": true,
1128
- "type": "CanonicalizePrompt"
1129
- },
1130
- {
1131
- "input_key": "prompt",
1132
- "remove_input_key": true,
1133
- "tokenizer": {
1134
- "model_max_length": 512,
1135
- "model_path": "./checkpoints/Cosmos-Predict2.5-2B/tokenizer",
1136
- "type": "PretrainedTokenizer"
1137
- },
1138
- "type": "ProcessCosmos25Prompt"
1139
- },
1140
- {
1141
- "backend": "torch",
1142
- "height": 224,
1143
- "output_layout": "nchw",
1144
- "scale_divisor": 255.0,
1145
- "type": "ResizeImages",
1146
- "width": 224
1147
- },
1148
- {
1149
- "frame_stride": 1,
1150
- "keep_time_dim": true,
1151
- "key": "images",
1152
- "num_views": 2,
1153
- "type": "ConcatImagesHorizontally",
1154
- "views_first": true
1155
- },
1156
- {
1157
- "backend": "torch",
1158
- "interleave": true,
1159
- "keys": [
1160
- "states"
1161
- ],
1162
- "target_dims": {
1163
- "states": 16
1164
- },
1165
- "type": "SinCosKeys"
1166
- },
1167
- {
1168
- "action_dim": null,
1169
- "action_key": "action",
1170
- "action_norm_mask": [
1171
- true,
1172
- true,
1173
- true,
1174
- true,
1175
- true,
1176
- true,
1177
- false
1178
- ],
1179
- "action_norm_type": "min_max",
1180
- "clip_norm": false,
1181
- "norm_type": "min_max",
1182
- "normalization_epsilon": 0.0,
1183
- "output_dtype": "float16",
1184
- "preserve_input_dtype": true,
1185
- "state_dim": null,
1186
- "state_key": "proprio",
1187
- "state_norm_type": "none",
1188
- "type": "NormalizeStatesAndActions"
1189
- },
1190
- {
1191
- "action_dim": 8,
1192
- "type": "PadActionsAndActionMasks",
1193
- "valid_action_dim": 7
1194
- },
1195
- {
1196
- "action_dim": 8,
1197
- "action_horizon": 8,
1198
- "mark_all_action_steps_valid": true,
1199
- "state_dim": 16,
1200
- "type": "PrepareDiT4DiTInputs",
1201
- "valid_action_dim": 7
1202
- }
1203
- ],
1204
- "type": "ParquetDataset",
1205
- "use_delta": false,
1206
- "window_start_idx": 0
1207
- },
1208
- {
1209
- "action_key": "action",
1210
- "action_window_size": 8,
1211
- "data_root_path": "./datasets/libero_10_no_noops_lerobotv2.1",
1212
- "frame_window_size": 9,
1213
- "statistic_name": "franka",
1214
- "transforms": [
1215
- {
1216
- "frame_stride": 2,
1217
- "name_mappings": {
1218
- "actions": [
1219
- "actions"
1220
- ],
1221
- "observation.state": [
1222
- "states"
1223
- ]
1224
- },
1225
- "parquet_keys": [
1226
- "observation.state",
1227
- "timestamp",
1228
- "actions",
1229
- "info",
1230
- "stats",
1231
- "action_masks"
1232
- ],
1233
- "type": "ProcessParquetInputs",
1234
- "video_keys": [
1235
- "observation.images.image",
1236
- "observation.images.wrist_image"
1237
- ]
1238
- },
1239
- {
1240
- "input_keys": [
1241
- "prompt",
1242
- "task_description",
1243
- "lang",
1244
- "instruction",
1245
- "instructions",
1246
- "text"
1247
- ],
1248
- "output_key": "prompt",
1249
- "remove_source_keys": true,
1250
- "type": "CanonicalizePrompt"
1251
- },
1252
- {
1253
- "input_key": "prompt",
1254
- "remove_input_key": true,
1255
- "tokenizer": {
1256
- "model_max_length": 512,
1257
- "model_path": "./checkpoints/Cosmos-Predict2.5-2B/tokenizer",
1258
- "type": "PretrainedTokenizer"
1259
- },
1260
- "type": "ProcessCosmos25Prompt"
1261
- },
1262
- {
1263
- "backend": "torch",
1264
- "height": 224,
1265
- "output_layout": "nchw",
1266
- "scale_divisor": 255.0,
1267
- "type": "ResizeImages",
1268
- "width": 224
1269
- },
1270
- {
1271
- "frame_stride": 1,
1272
- "keep_time_dim": true,
1273
- "key": "images",
1274
- "num_views": 2,
1275
- "type": "ConcatImagesHorizontally",
1276
- "views_first": true
1277
- },
1278
- {
1279
- "backend": "torch",
1280
- "interleave": true,
1281
- "keys": [
1282
- "states"
1283
- ],
1284
- "target_dims": {
1285
- "states": 16
1286
- },
1287
- "type": "SinCosKeys"
1288
- },
1289
- {
1290
- "action_dim": null,
1291
- "action_key": "action",
1292
- "action_norm_mask": [
1293
- true,
1294
- true,
1295
- true,
1296
- true,
1297
- true,
1298
- true,
1299
- false
1300
- ],
1301
- "action_norm_type": "min_max",
1302
- "clip_norm": false,
1303
- "norm_type": "min_max",
1304
- "normalization_epsilon": 0.0,
1305
- "output_dtype": "float16",
1306
- "preserve_input_dtype": true,
1307
- "state_dim": null,
1308
- "state_key": "proprio",
1309
- "state_norm_type": "none",
1310
- "type": "NormalizeStatesAndActions"
1311
- },
1312
- {
1313
- "action_dim": 8,
1314
- "type": "PadActionsAndActionMasks",
1315
- "valid_action_dim": 7
1316
- },
1317
- {
1318
- "action_dim": 8,
1319
- "action_horizon": 8,
1320
- "mark_all_action_steps_valid": true,
1321
- "state_dim": 16,
1322
- "type": "PrepareDiT4DiTInputs",
1323
- "valid_action_dim": 7
1324
- }
1325
- ],
1326
- "type": "ParquetDataset",
1327
- "use_delta": false,
1328
- "window_start_idx": 0
1329
- }
1330
- ],
1331
- "name_mappings": {
1332
- "action": [
1333
- "action"
1334
- ],
1335
- "observation.state": [
1336
- "proprio"
1337
- ]
1338
- },
1339
- "reshuffle_each_epoch": true,
1340
- "sampling_weights": [
1341
- 1.0,
1342
- 1.0,
1343
- 1.0,
1344
- 1.0
1345
- ],
1346
- "seed": 42,
1347
- "shuffle": false,
1348
- "statistic_keys": [
1349
- "observation.state",
1350
- "timestamp",
1351
- "action"
1352
- ],
1353
- "statistic_name": "franka",
1354
- "type": "DistributedBalancedRepeatingDataset"
1355
- },
1356
- "per_device_batch_size": 8,
1357
- "per_device_num_workers": 4
1358
- }
1359
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
dit4dit_libero_all_full_finetune_bc071270ff2_bs256/config.yaml DELETED
@@ -1,1082 +0,0 @@
1
- _action_dim: 8
2
- _action_horizon: 8
3
- _action_norm_mask:
4
- - true
5
- - true
6
- - true
7
- - true
8
- - true
9
- - true
10
- - false
11
- _ckpt_root: ./checkpoints
12
- _cosmos_base_model: ./checkpoints/Cosmos-Predict2.5-2B
13
- _cosmos_revision: diffusers/base/post-trained
14
- _cosmos_tokenizer:
15
- model_max_length: 512
16
- model_path: ./checkpoints/Cosmos-Predict2.5-2B/tokenizer
17
- type: PretrainedTokenizer
18
- _dit4dit_parquet_dataset:
19
- action_key: action
20
- action_window_size: 8
21
- frame_window_size: 9
22
- statistic_name: franka
23
- transforms:
24
- - frame_stride: 2
25
- name_mappings:
26
- actions:
27
- - actions
28
- observation.state:
29
- - states
30
- parquet_keys:
31
- - observation.state
32
- - timestamp
33
- - actions
34
- - info
35
- - stats
36
- - action_masks
37
- type: ProcessParquetInputs
38
- video_keys:
39
- - observation.images.image
40
- - observation.images.wrist_image
41
- - input_keys:
42
- - prompt
43
- - task_description
44
- - lang
45
- - instruction
46
- - instructions
47
- - text
48
- output_key: prompt
49
- remove_source_keys: true
50
- type: CanonicalizePrompt
51
- - input_key: prompt
52
- remove_input_key: true
53
- tokenizer:
54
- model_max_length: 512
55
- model_path: ./checkpoints/Cosmos-Predict2.5-2B/tokenizer
56
- type: PretrainedTokenizer
57
- type: ProcessCosmos25Prompt
58
- - backend: torch
59
- height: 224
60
- output_layout: nchw
61
- scale_divisor: 255.0
62
- type: ResizeImages
63
- width: 224
64
- - frame_stride: 1
65
- keep_time_dim: true
66
- key: images
67
- num_views: 2
68
- type: ConcatImagesHorizontally
69
- views_first: true
70
- - backend: torch
71
- interleave: true
72
- keys:
73
- - states
74
- target_dims:
75
- states: 16
76
- type: SinCosKeys
77
- - action_dim: null
78
- action_key: action
79
- action_norm_mask:
80
- - true
81
- - true
82
- - true
83
- - true
84
- - true
85
- - true
86
- - false
87
- action_norm_type: min_max
88
- clip_norm: false
89
- norm_type: min_max
90
- normalization_epsilon: 0.0
91
- output_dtype: float16
92
- preserve_input_dtype: true
93
- state_dim: null
94
- state_key: proprio
95
- state_norm_type: none
96
- type: NormalizeStatesAndActions
97
- - action_dim: 8
98
- type: PadActionsAndActionMasks
99
- valid_action_dim: 7
100
- - action_dim: 8
101
- action_horizon: 8
102
- mark_all_action_steps_valid: true
103
- state_dim: 16
104
- type: PrepareDiT4DiTInputs
105
- valid_action_dim: 7
106
- type: ParquetDataset
107
- use_delta: false
108
- window_start_idx: 0
109
- _dit4dit_train_transforms:
110
- - frame_stride: 2
111
- name_mappings:
112
- actions:
113
- - actions
114
- observation.state:
115
- - states
116
- parquet_keys:
117
- - observation.state
118
- - timestamp
119
- - actions
120
- - info
121
- - stats
122
- - action_masks
123
- type: ProcessParquetInputs
124
- video_keys:
125
- - observation.images.image
126
- - observation.images.wrist_image
127
- - input_keys:
128
- - prompt
129
- - task_description
130
- - lang
131
- - instruction
132
- - instructions
133
- - text
134
- output_key: prompt
135
- remove_source_keys: true
136
- type: CanonicalizePrompt
137
- - input_key: prompt
138
- remove_input_key: true
139
- tokenizer:
140
- model_max_length: 512
141
- model_path: ./checkpoints/Cosmos-Predict2.5-2B/tokenizer
142
- type: PretrainedTokenizer
143
- type: ProcessCosmos25Prompt
144
- - backend: torch
145
- height: 224
146
- output_layout: nchw
147
- scale_divisor: 255.0
148
- type: ResizeImages
149
- width: 224
150
- - frame_stride: 1
151
- keep_time_dim: true
152
- key: images
153
- num_views: 2
154
- type: ConcatImagesHorizontally
155
- views_first: true
156
- - backend: torch
157
- interleave: true
158
- keys:
159
- - states
160
- target_dims:
161
- states: 16
162
- type: SinCosKeys
163
- - action_dim: null
164
- action_key: action
165
- action_norm_mask:
166
- - true
167
- - true
168
- - true
169
- - true
170
- - true
171
- - true
172
- - false
173
- action_norm_type: min_max
174
- clip_norm: false
175
- norm_type: min_max
176
- normalization_epsilon: 0.0
177
- output_dtype: float16
178
- preserve_input_dtype: true
179
- state_dim: null
180
- state_key: proprio
181
- state_norm_type: none
182
- type: NormalizeStatesAndActions
183
- - action_dim: 8
184
- type: PadActionsAndActionMasks
185
- valid_action_dim: 7
186
- - action_dim: 8
187
- action_horizon: 8
188
- mark_all_action_steps_valid: true
189
- state_dim: 16
190
- type: PrepareDiT4DiTInputs
191
- valid_action_dim: 7
192
- _frame_window_size: 9
193
- _image_frame_stride: 2
194
- _image_size: 224
195
- _libero_data_roots:
196
- - ./datasets/libero_object_no_noops_lerobotv2.1
197
- - ./datasets/libero_goal_no_noops_lerobotv2.1
198
- - ./datasets/libero_spatial_no_noops_lerobotv2.1
199
- - ./datasets/libero_10_no_noops_lerobotv2.1
200
- _official_dataset_statistics:
201
- franka:
202
- action:
203
- mask:
204
- - true
205
- - true
206
- - true
207
- - true
208
- - true
209
- - true
210
- - false
211
- max:
212
- - 0.9375
213
- - 0.9375
214
- - 0.9375
215
- - 0.3557142913341522
216
- - 0.375
217
- - 0.375
218
- - 1.0
219
- mean:
220
- - 0.07237596483901143
221
- - 0.08987006871029735
222
- - -0.10144743137061596
223
- - -0.00045383188989944756
224
- - 0.006273590726777911
225
- - -0.003878799732774496
226
- - 0.524486355483532
227
- min:
228
- - -0.9375
229
- - -0.9375
230
- - -0.9375
231
- - -0.2582142949104309
232
- - -0.375
233
- - -0.3675000071525574
234
- - 0.0
235
- q01:
236
- - -0.8785714507102966
237
- - -0.8758928775787354
238
- - -0.9375
239
- - -0.1510714292526245
240
- - -0.20678570866584778
241
- - -0.2742857038974762
242
- - 0.0
243
- q99:
244
- - 0.9375
245
- - 0.9107142686843872
246
- - 0.9375
247
- - 0.20357142388820648
248
- - 0.26357144117355347
249
- - 0.375
250
- - 1.0
251
- std:
252
- - 0.3498823308902479
253
- - 0.37794140366375184
254
- - 0.460084266976933
255
- - 0.0403885784928603
256
- - 0.06616144248501059
257
- - 0.07763074391911857
258
- - 0.4994683356809767
259
- num_trajectories: 1693
260
- num_transitions: 273465
261
- state:
262
- max:
263
- - 0.21031762659549713
264
- - 0.39128610491752625
265
- - 1.3660105466842651
266
- - 3.6714255809783936
267
- - 3.560650587081909
268
- - 1.386339545249939
269
- - 0.04233968257904053
270
- - 0.0013633022317662835
271
- mean:
272
- - -0.04889854742214084
273
- - 0.03689368185587227
274
- - 0.7890402488410473
275
- - 2.9771945476531982
276
- - -0.1417286954820156
277
- - -0.11769362539052963
278
- - 0.026436020154505968
279
- - -0.02665513101965189
280
- min:
281
- - -0.4828203022480011
282
- - -0.3255046010017395
283
- - 0.008128180168569088
284
- - 0.35277295112609863
285
- - -3.641430377960205
286
- - -1.842738389968872
287
- - -0.0013586411951109767
288
- - -0.042040832340717316
289
- q01:
290
- - -0.42401049643754957
291
- - -0.2838300323486328
292
- - 0.009925739830359817
293
- - 1.3085840785503386
294
- - -2.886677579879761
295
- - -1.1599004411697387
296
- - 0.001503719249740243
297
- - -0.040336399003863335
298
- q99:
299
- - 0.1530261474847791
300
- - 0.3629165390133857
301
- - 1.2910678112506866
302
- - 3.303542451858519
303
- - 2.7496529006957933
304
- - 0.6893712210655194
305
- - 0.040610933862626555
306
- - -0.0015016929572448147
307
- std:
308
- - 0.10639013941746686
309
- - 0.15115733130675715
310
- - 0.38406895599530033
311
- - 0.3530238395244304
312
- - 0.8227341427331599
313
- - 0.32357567121520087
314
- - 0.014583991652936385
315
- - 0.014467005007200339
316
- _official_dit4dit_ckpt: ./checkpoints/dit4dit-model/dit4dit_libero/final_model/pytorch_model.pt
317
- _ori_action_dim: 7
318
- _prompt_input_keys:
319
- - prompt
320
- - task_description
321
- - lang
322
- - instruction
323
- - instructions
324
- - text
325
- _state_dim: 16
326
- eval:
327
- dataset:
328
- img_buffer_len: 1
329
- require_lang_tokens: true
330
- transforms:
331
- - input_keys:
332
- - prompt
333
- - task_description
334
- - lang
335
- - instruction
336
- - instructions
337
- - text
338
- output_key: prompt
339
- remove_source_keys: true
340
- type: CanonicalizePrompt
341
- - input_key: prompt
342
- remove_input_key: true
343
- tokenizer:
344
- model_max_length: 512
345
- model_path: ./checkpoints/Cosmos-Predict2.5-2B/tokenizer
346
- type: PretrainedTokenizer
347
- type: ProcessCosmos25Prompt
348
- - image_size: 224
349
- img_keys:
350
- - agentview_image
351
- - robot0_eye_in_hand_image
352
- type: ProcessDiT4DiTLiberoEvalInputs
353
- - gripper_key: robot0_gripper_qpos
354
- norm_type: none
355
- out_key: states
356
- pos_key: robot0_eef_pos
357
- quat_key: robot0_eef_quat
358
- state_dim: null
359
- type: LiberoProprioFromInputs
360
- - interleave: true
361
- keys:
362
- - states
363
- target_dims:
364
- states: 16
365
- type: SinCosKeys
366
- - image_key: pixel_values
367
- require_actions: false
368
- state_dim: 16
369
- type: PrepareDiT4DiTInputs
370
- type: LiberoParquetEvalDataset
371
- denormalize_action:
372
- action_dim: 7
373
- action_norm_mask:
374
- - true
375
- - true
376
- - true
377
- - true
378
- - true
379
- - true
380
- - false
381
- clip_normalized_action: true
382
- invert_gripper_action: true
383
- norm_type: min_max
384
- normalize_gripper_action: true
385
- type: DenormalizeLiberoAction
386
- enable_mixed_precision_training: false
387
- eval_chunk_size: 8
388
- mixed_precision_dtype: bf16
389
- model_family: dit4dit
390
- norm_stats_key: franka
391
- num_steps_wait: 10
392
- num_trials_per_task: 50
393
- resize_size: 224
394
- seed: 7
395
- task_suite_name:
396
- - libero_spatial
397
- - libero_object
398
- - libero_goal
399
- - libero_10
400
- truncate_action_chunk_at_horizon: true
401
- type: LiberoEvalRunner
402
- inference_model:
403
- freeze_vlm_backbone: false
404
- name_mapping: null
405
- pretrained_name_or_path: null
406
- repeated_diffusion_steps: 4
407
- strict_mapping: false
408
- type: DiT4DiTVLA
409
- vla_head:
410
- action_dim: 8
411
- action_horizon: 8
412
- action_model_type: DiT-B
413
- add_pos_embed: true
414
- diffusion_model_cfg:
415
- cross_attention_dim: 2048
416
- dropout: 0.2
417
- final_dropout: true
418
- interleave_self_attention: true
419
- norm_type: ada_norm
420
- num_layers: 16
421
- output_dim: 2560
422
- positional_embeddings: null
423
- future_action_window_size: 7
424
- hidden_size: 2560
425
- max_seq_len: 1024
426
- noise_beta_alpha: 1.5
427
- noise_beta_beta: 1.0
428
- noise_s: 0.999
429
- num_inference_timesteps: 4
430
- num_timestep_buckets: 1000
431
- ori_action_dim: 7
432
- state_dim: 16
433
- type: DiT4DiTActionHead
434
- vlm_backbone:
435
- base_model: ./checkpoints/Cosmos-Predict2.5-2B
436
- conditional_frame_timestep: 0.0001
437
- detach_hidden_states: true
438
- extract_layer: 17
439
- fixed_seed: null
440
- flow_matching_high_sigma_min: null
441
- flow_matching_high_sigma_ratio: null
442
- flow_matching_time_distribution: uniform
443
- frozen_submodules:
444
- - text_encoder
445
- - vae
446
- fsdp_min_num_params: 0
447
- future_loss_type: flow_matching
448
- local_files_only: true
449
- max_sequence_length: 512
450
- num_frames_out: 5
451
- num_inference_steps: 1
452
- revision: diffusers/base/post-trained
453
- split_future_frames: true
454
- torch_dtype: bf16
455
- trainable: true
456
- type: Cosmos25Backbone
457
- model:
458
- freeze_vlm_backbone: false
459
- name_mapping:
460
- vla_head: action_model
461
- vlm_backbone.text_encoder: backbone_interface.extractor.text_encoder
462
- vlm_backbone.transformer: backbone_interface.extractor.transformer
463
- vlm_backbone.vae: backbone_interface.extractor.vae
464
- pretrained_name_or_path: null
465
- repeated_diffusion_steps: 4
466
- strict_mapping: false
467
- type: DiT4DiTVLA
468
- vla_head:
469
- action_dim: 8
470
- action_horizon: 8
471
- action_model_type: DiT-B
472
- add_pos_embed: true
473
- diffusion_model_cfg:
474
- cross_attention_dim: 2048
475
- dropout: 0.2
476
- final_dropout: true
477
- interleave_self_attention: true
478
- norm_type: ada_norm
479
- num_layers: 16
480
- output_dim: 2560
481
- positional_embeddings: null
482
- future_action_window_size: 7
483
- hidden_size: 2560
484
- max_seq_len: 1024
485
- noise_beta_alpha: 1.5
486
- noise_beta_beta: 1.0
487
- noise_s: 0.999
488
- num_inference_timesteps: 4
489
- num_timestep_buckets: 1000
490
- ori_action_dim: 7
491
- state_dim: 16
492
- type: DiT4DiTActionHead
493
- vlm_backbone:
494
- base_model: ./checkpoints/Cosmos-Predict2.5-2B
495
- conditional_frame_timestep: 0.0001
496
- detach_hidden_states: true
497
- extract_layer: 17
498
- fixed_seed: null
499
- flow_matching_high_sigma_min: null
500
- flow_matching_high_sigma_ratio: null
501
- flow_matching_time_distribution: uniform
502
- frozen_submodules:
503
- - text_encoder
504
- - vae
505
- fsdp_min_num_params: 0
506
- future_loss_type: flow_matching
507
- local_files_only: true
508
- max_sequence_length: 512
509
- num_frames_out: 5
510
- num_inference_steps: 1
511
- revision: diffusers/base/post-trained
512
- split_future_frames: true
513
- torch_dtype: bf16
514
- trainable: true
515
- type: Cosmos25Backbone
516
- official_eval_task_suites:
517
- - libero_spatial
518
- - libero_object
519
- - libero_goal
520
- - libero_10
521
- runner:
522
- change_key_name: false
523
- collator:
524
- keys:
525
- - states
526
- - timestamp
527
- - images
528
- - img_masks
529
- - actions
530
- - action_masks
531
- - frame_masks
532
- - lang_tokens
533
- - lang_masks
534
- meta_keys:
535
- - info
536
- - stats
537
- type: DictCollator
538
- enable_gradient_checkpointing: false
539
- enable_mixed_precision_training: true
540
- grad_accumulation_steps: 2
541
- lr_scheduler:
542
- min_lr: 5.0e-07
543
- type: cosine_with_min_lr
544
- warmup_steps: 10000
545
- max_grad_norm: 1.0
546
- max_keep_ckpts: 2
547
- max_steps: 160000
548
- metric:
549
- active_trackers:
550
- - jsonl
551
- - wandb
552
- grad_accumulation_steps: 1
553
- run_dir: work_dirs
554
- type: VLAMetric
555
- window_size: 1
556
- mixed_precision_dtype: bf16
557
- optimizer:
558
- betas:
559
- - 0.9
560
- - 0.95
561
- eps: 1.0e-08
562
- lr: 3.0e-05
563
- paramwise_learning_rate:
564
- vla_head: 0.0001
565
- vlm_backbone.transformer: 0.0001
566
- type: AdamW
567
- weight_decay: 1.0e-08
568
- pre_fsdp_param_dtype: fp32
569
- reduce_in_full_precision: false
570
- sampler: null
571
- save_iter_interval: 40000
572
- sharding_strategy: shard-grad-op
573
- tokenizer:
574
- model_max_length: 512
575
- model_path: ./checkpoints/Cosmos-Predict2.5-2B/tokenizer
576
- type: PretrainedTokenizer
577
- type: FSDPTrainRunner
578
- seed: 42
579
- train_dataloader:
580
- dataset:
581
- dataset_statistics:
582
- franka:
583
- action:
584
- mask:
585
- - true
586
- - true
587
- - true
588
- - true
589
- - true
590
- - true
591
- - false
592
- max:
593
- - 0.9375
594
- - 0.9375
595
- - 0.9375
596
- - 0.3557142913341522
597
- - 0.375
598
- - 0.375
599
- - 1.0
600
- mean:
601
- - 0.07237596483901143
602
- - 0.08987006871029735
603
- - -0.10144743137061596
604
- - -0.00045383188989944756
605
- - 0.006273590726777911
606
- - -0.003878799732774496
607
- - 0.524486355483532
608
- min:
609
- - -0.9375
610
- - -0.9375
611
- - -0.9375
612
- - -0.2582142949104309
613
- - -0.375
614
- - -0.3675000071525574
615
- - 0.0
616
- q01:
617
- - -0.8785714507102966
618
- - -0.8758928775787354
619
- - -0.9375
620
- - -0.1510714292526245
621
- - -0.20678570866584778
622
- - -0.2742857038974762
623
- - 0.0
624
- q99:
625
- - 0.9375
626
- - 0.9107142686843872
627
- - 0.9375
628
- - 0.20357142388820648
629
- - 0.26357144117355347
630
- - 0.375
631
- - 1.0
632
- std:
633
- - 0.3498823308902479
634
- - 0.37794140366375184
635
- - 0.460084266976933
636
- - 0.0403885784928603
637
- - 0.06616144248501059
638
- - 0.07763074391911857
639
- - 0.4994683356809767
640
- num_trajectories: 1693
641
- num_transitions: 273465
642
- state:
643
- max:
644
- - 0.21031762659549713
645
- - 0.39128610491752625
646
- - 1.3660105466842651
647
- - 3.6714255809783936
648
- - 3.560650587081909
649
- - 1.386339545249939
650
- - 0.04233968257904053
651
- - 0.0013633022317662835
652
- mean:
653
- - -0.04889854742214084
654
- - 0.03689368185587227
655
- - 0.7890402488410473
656
- - 2.9771945476531982
657
- - -0.1417286954820156
658
- - -0.11769362539052963
659
- - 0.026436020154505968
660
- - -0.02665513101965189
661
- min:
662
- - -0.4828203022480011
663
- - -0.3255046010017395
664
- - 0.008128180168569088
665
- - 0.35277295112609863
666
- - -3.641430377960205
667
- - -1.842738389968872
668
- - -0.0013586411951109767
669
- - -0.042040832340717316
670
- q01:
671
- - -0.42401049643754957
672
- - -0.2838300323486328
673
- - 0.009925739830359817
674
- - 1.3085840785503386
675
- - -2.886677579879761
676
- - -1.1599004411697387
677
- - 0.001503719249740243
678
- - -0.040336399003863335
679
- q99:
680
- - 0.1530261474847791
681
- - 0.3629165390133857
682
- - 1.2910678112506866
683
- - 3.303542451858519
684
- - 2.7496529006957933
685
- - 0.6893712210655194
686
- - 0.040610933862626555
687
- - -0.0015016929572448147
688
- std:
689
- - 0.10639013941746686
690
- - 0.15115733130675715
691
- - 0.38406895599530033
692
- - 0.3530238395244304
693
- - 0.8227341427331599
694
- - 0.32357567121520087
695
- - 0.014583991652936385
696
- - 0.014467005007200339
697
- datasets:
698
- - action_key: action
699
- action_window_size: 8
700
- data_root_path: ./datasets/libero_object_no_noops_lerobotv2.1
701
- frame_window_size: 9
702
- statistic_name: franka
703
- transforms:
704
- - frame_stride: 2
705
- name_mappings:
706
- actions:
707
- - actions
708
- observation.state:
709
- - states
710
- parquet_keys:
711
- - observation.state
712
- - timestamp
713
- - actions
714
- - info
715
- - stats
716
- - action_masks
717
- type: ProcessParquetInputs
718
- video_keys:
719
- - observation.images.image
720
- - observation.images.wrist_image
721
- - input_keys:
722
- - prompt
723
- - task_description
724
- - lang
725
- - instruction
726
- - instructions
727
- - text
728
- output_key: prompt
729
- remove_source_keys: true
730
- type: CanonicalizePrompt
731
- - input_key: prompt
732
- remove_input_key: true
733
- tokenizer:
734
- model_max_length: 512
735
- model_path: ./checkpoints/Cosmos-Predict2.5-2B/tokenizer
736
- type: PretrainedTokenizer
737
- type: ProcessCosmos25Prompt
738
- - backend: torch
739
- height: 224
740
- output_layout: nchw
741
- scale_divisor: 255.0
742
- type: ResizeImages
743
- width: 224
744
- - frame_stride: 1
745
- keep_time_dim: true
746
- key: images
747
- num_views: 2
748
- type: ConcatImagesHorizontally
749
- views_first: true
750
- - backend: torch
751
- interleave: true
752
- keys:
753
- - states
754
- target_dims:
755
- states: 16
756
- type: SinCosKeys
757
- - action_dim: null
758
- action_key: action
759
- action_norm_mask:
760
- - true
761
- - true
762
- - true
763
- - true
764
- - true
765
- - true
766
- - false
767
- action_norm_type: min_max
768
- clip_norm: false
769
- norm_type: min_max
770
- normalization_epsilon: 0.0
771
- output_dtype: float16
772
- preserve_input_dtype: true
773
- state_dim: null
774
- state_key: proprio
775
- state_norm_type: none
776
- type: NormalizeStatesAndActions
777
- - action_dim: 8
778
- type: PadActionsAndActionMasks
779
- valid_action_dim: 7
780
- - action_dim: 8
781
- action_horizon: 8
782
- mark_all_action_steps_valid: true
783
- state_dim: 16
784
- type: PrepareDiT4DiTInputs
785
- valid_action_dim: 7
786
- type: ParquetDataset
787
- use_delta: false
788
- window_start_idx: 0
789
- - action_key: action
790
- action_window_size: 8
791
- data_root_path: ./datasets/libero_goal_no_noops_lerobotv2.1
792
- frame_window_size: 9
793
- statistic_name: franka
794
- transforms:
795
- - frame_stride: 2
796
- name_mappings:
797
- actions:
798
- - actions
799
- observation.state:
800
- - states
801
- parquet_keys:
802
- - observation.state
803
- - timestamp
804
- - actions
805
- - info
806
- - stats
807
- - action_masks
808
- type: ProcessParquetInputs
809
- video_keys:
810
- - observation.images.image
811
- - observation.images.wrist_image
812
- - input_keys:
813
- - prompt
814
- - task_description
815
- - lang
816
- - instruction
817
- - instructions
818
- - text
819
- output_key: prompt
820
- remove_source_keys: true
821
- type: CanonicalizePrompt
822
- - input_key: prompt
823
- remove_input_key: true
824
- tokenizer:
825
- model_max_length: 512
826
- model_path: ./checkpoints/Cosmos-Predict2.5-2B/tokenizer
827
- type: PretrainedTokenizer
828
- type: ProcessCosmos25Prompt
829
- - backend: torch
830
- height: 224
831
- output_layout: nchw
832
- scale_divisor: 255.0
833
- type: ResizeImages
834
- width: 224
835
- - frame_stride: 1
836
- keep_time_dim: true
837
- key: images
838
- num_views: 2
839
- type: ConcatImagesHorizontally
840
- views_first: true
841
- - backend: torch
842
- interleave: true
843
- keys:
844
- - states
845
- target_dims:
846
- states: 16
847
- type: SinCosKeys
848
- - action_dim: null
849
- action_key: action
850
- action_norm_mask:
851
- - true
852
- - true
853
- - true
854
- - true
855
- - true
856
- - true
857
- - false
858
- action_norm_type: min_max
859
- clip_norm: false
860
- norm_type: min_max
861
- normalization_epsilon: 0.0
862
- output_dtype: float16
863
- preserve_input_dtype: true
864
- state_dim: null
865
- state_key: proprio
866
- state_norm_type: none
867
- type: NormalizeStatesAndActions
868
- - action_dim: 8
869
- type: PadActionsAndActionMasks
870
- valid_action_dim: 7
871
- - action_dim: 8
872
- action_horizon: 8
873
- mark_all_action_steps_valid: true
874
- state_dim: 16
875
- type: PrepareDiT4DiTInputs
876
- valid_action_dim: 7
877
- type: ParquetDataset
878
- use_delta: false
879
- window_start_idx: 0
880
- - action_key: action
881
- action_window_size: 8
882
- data_root_path: ./datasets/libero_spatial_no_noops_lerobotv2.1
883
- frame_window_size: 9
884
- statistic_name: franka
885
- transforms:
886
- - frame_stride: 2
887
- name_mappings:
888
- actions:
889
- - actions
890
- observation.state:
891
- - states
892
- parquet_keys:
893
- - observation.state
894
- - timestamp
895
- - actions
896
- - info
897
- - stats
898
- - action_masks
899
- type: ProcessParquetInputs
900
- video_keys:
901
- - observation.images.image
902
- - observation.images.wrist_image
903
- - input_keys:
904
- - prompt
905
- - task_description
906
- - lang
907
- - instruction
908
- - instructions
909
- - text
910
- output_key: prompt
911
- remove_source_keys: true
912
- type: CanonicalizePrompt
913
- - input_key: prompt
914
- remove_input_key: true
915
- tokenizer:
916
- model_max_length: 512
917
- model_path: ./checkpoints/Cosmos-Predict2.5-2B/tokenizer
918
- type: PretrainedTokenizer
919
- type: ProcessCosmos25Prompt
920
- - backend: torch
921
- height: 224
922
- output_layout: nchw
923
- scale_divisor: 255.0
924
- type: ResizeImages
925
- width: 224
926
- - frame_stride: 1
927
- keep_time_dim: true
928
- key: images
929
- num_views: 2
930
- type: ConcatImagesHorizontally
931
- views_first: true
932
- - backend: torch
933
- interleave: true
934
- keys:
935
- - states
936
- target_dims:
937
- states: 16
938
- type: SinCosKeys
939
- - action_dim: null
940
- action_key: action
941
- action_norm_mask:
942
- - true
943
- - true
944
- - true
945
- - true
946
- - true
947
- - true
948
- - false
949
- action_norm_type: min_max
950
- clip_norm: false
951
- norm_type: min_max
952
- normalization_epsilon: 0.0
953
- output_dtype: float16
954
- preserve_input_dtype: true
955
- state_dim: null
956
- state_key: proprio
957
- state_norm_type: none
958
- type: NormalizeStatesAndActions
959
- - action_dim: 8
960
- type: PadActionsAndActionMasks
961
- valid_action_dim: 7
962
- - action_dim: 8
963
- action_horizon: 8
964
- mark_all_action_steps_valid: true
965
- state_dim: 16
966
- type: PrepareDiT4DiTInputs
967
- valid_action_dim: 7
968
- type: ParquetDataset
969
- use_delta: false
970
- window_start_idx: 0
971
- - action_key: action
972
- action_window_size: 8
973
- data_root_path: ./datasets/libero_10_no_noops_lerobotv2.1
974
- frame_window_size: 9
975
- statistic_name: franka
976
- transforms:
977
- - frame_stride: 2
978
- name_mappings:
979
- actions:
980
- - actions
981
- observation.state:
982
- - states
983
- parquet_keys:
984
- - observation.state
985
- - timestamp
986
- - actions
987
- - info
988
- - stats
989
- - action_masks
990
- type: ProcessParquetInputs
991
- video_keys:
992
- - observation.images.image
993
- - observation.images.wrist_image
994
- - input_keys:
995
- - prompt
996
- - task_description
997
- - lang
998
- - instruction
999
- - instructions
1000
- - text
1001
- output_key: prompt
1002
- remove_source_keys: true
1003
- type: CanonicalizePrompt
1004
- - input_key: prompt
1005
- remove_input_key: true
1006
- tokenizer:
1007
- model_max_length: 512
1008
- model_path: ./checkpoints/Cosmos-Predict2.5-2B/tokenizer
1009
- type: PretrainedTokenizer
1010
- type: ProcessCosmos25Prompt
1011
- - backend: torch
1012
- height: 224
1013
- output_layout: nchw
1014
- scale_divisor: 255.0
1015
- type: ResizeImages
1016
- width: 224
1017
- - frame_stride: 1
1018
- keep_time_dim: true
1019
- key: images
1020
- num_views: 2
1021
- type: ConcatImagesHorizontally
1022
- views_first: true
1023
- - backend: torch
1024
- interleave: true
1025
- keys:
1026
- - states
1027
- target_dims:
1028
- states: 16
1029
- type: SinCosKeys
1030
- - action_dim: null
1031
- action_key: action
1032
- action_norm_mask:
1033
- - true
1034
- - true
1035
- - true
1036
- - true
1037
- - true
1038
- - true
1039
- - false
1040
- action_norm_type: min_max
1041
- clip_norm: false
1042
- norm_type: min_max
1043
- normalization_epsilon: 0.0
1044
- output_dtype: float16
1045
- preserve_input_dtype: true
1046
- state_dim: null
1047
- state_key: proprio
1048
- state_norm_type: none
1049
- type: NormalizeStatesAndActions
1050
- - action_dim: 8
1051
- type: PadActionsAndActionMasks
1052
- valid_action_dim: 7
1053
- - action_dim: 8
1054
- action_horizon: 8
1055
- mark_all_action_steps_valid: true
1056
- state_dim: 16
1057
- type: PrepareDiT4DiTInputs
1058
- valid_action_dim: 7
1059
- type: ParquetDataset
1060
- use_delta: false
1061
- window_start_idx: 0
1062
- name_mappings:
1063
- action:
1064
- - action
1065
- observation.state:
1066
- - proprio
1067
- reshuffle_each_epoch: true
1068
- sampling_weights:
1069
- - 1.0
1070
- - 1.0
1071
- - 1.0
1072
- - 1.0
1073
- seed: 42
1074
- shuffle: false
1075
- statistic_keys:
1076
- - observation.state
1077
- - timestamp
1078
- - action
1079
- statistic_name: franka
1080
- type: DistributedBalancedRepeatingDataset
1081
- per_device_batch_size: 8
1082
- per_device_num_workers: 4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
dit4dit_libero_all_full_finetune_bc071270ff2_bs256/dataset_statistics.json DELETED
@@ -1,133 +0,0 @@
1
- {
2
- "franka": {
3
- "action": {
4
- "mean": [
5
- 0.07237596483901143,
6
- 0.08987006871029735,
7
- -0.10144743137061596,
8
- -0.00045383188989944756,
9
- 0.006273590726777911,
10
- -0.003878799732774496,
11
- 0.524486355483532
12
- ],
13
- "std": [
14
- 0.3498823308902479,
15
- 0.37794140366375184,
16
- 0.460084266976933,
17
- 0.0403885784928603,
18
- 0.06616144248501059,
19
- 0.07763074391911857,
20
- 0.4994683356809767
21
- ],
22
- "max": [
23
- 0.9375,
24
- 0.9375,
25
- 0.9375,
26
- 0.3557142913341522,
27
- 0.375,
28
- 0.375,
29
- 1.0
30
- ],
31
- "min": [
32
- -0.9375,
33
- -0.9375,
34
- -0.9375,
35
- -0.2582142949104309,
36
- -0.375,
37
- -0.3675000071525574,
38
- 0.0
39
- ],
40
- "q01": [
41
- -0.8785714507102966,
42
- -0.8758928775787354,
43
- -0.9375,
44
- -0.1510714292526245,
45
- -0.20678570866584778,
46
- -0.2742857038974762,
47
- 0.0
48
- ],
49
- "q99": [
50
- 0.9375,
51
- 0.9107142686843872,
52
- 0.9375,
53
- 0.20357142388820648,
54
- 0.26357144117355347,
55
- 0.375,
56
- 1.0
57
- ],
58
- "mask": [
59
- true,
60
- true,
61
- true,
62
- true,
63
- true,
64
- true,
65
- false
66
- ]
67
- },
68
- "state": {
69
- "mean": [
70
- -0.04889854742214084,
71
- 0.03689368185587227,
72
- 0.7890402488410473,
73
- 2.9771945476531982,
74
- -0.1417286954820156,
75
- -0.11769362539052963,
76
- 0.026436020154505968,
77
- -0.02665513101965189
78
- ],
79
- "std": [
80
- 0.10639013941746686,
81
- 0.15115733130675715,
82
- 0.38406895599530033,
83
- 0.3530238395244304,
84
- 0.8227341427331599,
85
- 0.32357567121520087,
86
- 0.014583991652936385,
87
- 0.014467005007200339
88
- ],
89
- "max": [
90
- 0.21031762659549713,
91
- 0.39128610491752625,
92
- 1.3660105466842651,
93
- 3.6714255809783936,
94
- 3.560650587081909,
95
- 1.386339545249939,
96
- 0.04233968257904053,
97
- 0.0013633022317662835
98
- ],
99
- "min": [
100
- -0.4828203022480011,
101
- -0.3255046010017395,
102
- 0.008128180168569088,
103
- 0.35277295112609863,
104
- -3.641430377960205,
105
- -1.842738389968872,
106
- -0.0013586411951109767,
107
- -0.042040832340717316
108
- ],
109
- "q01": [
110
- -0.42401049643754957,
111
- -0.2838300323486328,
112
- 0.009925739830359817,
113
- 1.3085840785503386,
114
- -2.886677579879761,
115
- -1.1599004411697387,
116
- 0.001503719249740243,
117
- -0.040336399003863335
118
- ],
119
- "q99": [
120
- 0.1530261474847791,
121
- 0.3629165390133857,
122
- 1.2910678112506866,
123
- 3.303542451858519,
124
- 2.7496529006957933,
125
- 0.6893712210655194,
126
- 0.040610933862626555,
127
- -0.0015016929572448147
128
- ]
129
- },
130
- "num_transitions": 273465,
131
- "num_trajectories": 1693
132
- }
133
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
dit4dit_libero_all_full_finetune_bc071270ff2_bs256/dit4dit_libero_all_full_finetune_2026_08_12_20_51_03.jsonl DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:0895afd366cddc06686303d4b2d7fb8cfe96621b6b77c13596b7e7467b50f58a
3
- size 61119743
 
 
 
 
dit4dit_libero_all_full_finetune_bc071270ff2_bs256/run-metrics.jsonl DELETED
@@ -1 +0,0 @@
1
- {"hparams": "{'_ckpt_root': './checkpoints', '_cosmos_base_model': './checkpoints/Cosmos-Predict2.5-2B', '_cosmos_revision': 'diffusers/base/post-trained', '_cosmos_tokenizer': {'type': 'PretrainedTokenizer', 'model_path': './checkpoints/Cosmos-Predict2.5-2B/tokenizer', 'model_max_length': 512}, '_official_dit4dit_ckpt': './checkpoints/dit4dit-model/dit4dit_libero/final_model/pytorch_model.pt', '_official_dataset_statistics': {'franka': {'action': {'mean': [0.07237596483901143, 0.08987006871029735, -0.10144743137061596, -0.00045383188989944756, 0.006273590726777911, -0.003878799732774496, 0.524486355483532], 'std': [0.3498823308902479, 0.37794140366375184, 0.460084266976933, 0.0403885784928603, 0.06616144248501059, 0.07763074391911857, 0.4994683356809767], 'max': [0.9375, 0.9375, 0.9375, 0.3557142913341522, 0.375, 0.375, 1.0], 'min': [-0.9375, -0.9375, -0.9375, -0.2582142949104309, -0.375, -0.3675000071525574, 0.0], 'q01': [-0.8785714507102966, -0.8758928775787354, -0.9375, -0.1510714292526245, -0.20678570866584778, -0.2742857038974762, 0.0], 'q99': [0.9375, 0.9107142686843872, 0.9375, 0.20357142388820648, 0.26357144117355347, 0.375, 1.0], 'mask': [True, True, True, True, True, True, False]}, 'state': {'mean': [-0.04889854742214084, 0.03689368185587227, 0.7890402488410473, 2.9771945476531982, -0.1417286954820156, -0.11769362539052963, 0.026436020154505968, -0.02665513101965189], 'std': [0.10639013941746686, 0.15115733130675715, 0.38406895599530033, 0.3530238395244304, 0.8227341427331599, 0.32357567121520087, 0.014583991652936385, 0.014467005007200339], 'max': [0.21031762659549713, 0.39128610491752625, 1.3660105466842651, 3.6714255809783936, 3.560650587081909, 1.386339545249939, 0.04233968257904053, 0.0013633022317662835], 'min': [-0.4828203022480011, -0.3255046010017395, 0.008128180168569088, 0.35277295112609863, -3.641430377960205, -1.842738389968872, -0.0013586411951109767, -0.042040832340717316], 'q01': [-0.42401049643754957, -0.2838300323486328, 0.009925739830359817, 1.3085840785503386, -2.886677579879761, -1.1599004411697387, 0.001503719249740243, -0.040336399003863335], 'q99': [0.1530261474847791, 0.3629165390133857, 1.2910678112506866, 3.303542451858519, 2.7496529006957933, 0.6893712210655194, 0.040610933862626555, -0.0015016929572448147]}, 'num_transitions': 273465, 'num_trajectories': 1693}}, '_action_dim': 8, '_ori_action_dim': 7, '_state_dim': 16, '_action_horizon': 8, '_frame_window_size': 9, '_image_frame_stride': 2, '_image_size': 224, '_action_norm_mask': [True, True, True, True, True, True, False], '_libero_data_roots': ['./datasets/libero_object_no_noops_lerobotv2.1', './datasets/libero_goal_no_noops_lerobotv2.1', './datasets/libero_spatial_no_noops_lerobotv2.1', './datasets/libero_10_no_noops_lerobotv2.1'], '_prompt_input_keys': ['prompt', 'task_description', 'lang', 'instruction', 'instructions', 'text'], 'seed': 42, 'official_eval_task_suites': ['libero_spatial', 'libero_object', 'libero_goal', 'libero_10'], 'model': {'type': 'DiT4DiTVLA', 'pretrained_name_or_path': None, 'name_mapping': {'vlm_backbone.text_encoder': 'backbone_interface.extractor.text_encoder', 'vlm_backbone.transformer': 'backbone_interface.extractor.transformer', 'vlm_backbone.vae': 'backbone_interface.extractor.vae', 'vla_head': 'action_model'}, 'strict_mapping': False, 'repeated_diffusion_steps': 4, 'vlm_backbone': {'type': 'Cosmos25Backbone', 'base_model': './checkpoints/Cosmos-Predict2.5-2B', 'revision': 'diffusers/base/post-trained', 'torch_dtype': 'bf16', 'local_files_only': True, 'extract_layer': 17, 'max_sequence_length': 512, 'trainable': True, 'frozen_submodules': ['text_encoder', 'vae'], 'split_future_frames': True, 'num_frames_out': 5, 'fixed_seed': None, 'num_inference_steps': 1, 'conditional_frame_timestep': 0.0001, 'future_loss_type': 'flow_matching', 'detach_hidden_states': True, 'flow_matching_time_distribution': 'uniform', 'flow_matching_high_sigma_ratio': None, 'flow_matching_high_sigma_min': None, 'fsdp_min_num_params': 0}, 'vla_head': {'type': 'DiT4DiTActionHead', 'action_model_type': 'DiT-B', 'hidden_size': 2560, 'add_pos_embed': True, 'max_seq_len': 1024, 'action_dim': 8, 'ori_action_dim': 7, 'state_dim': 16, 'future_action_window_size': 7, 'action_horizon': 8, 'noise_beta_alpha': 1.5, 'noise_beta_beta': 1.0, 'noise_s': 0.999, 'num_timestep_buckets': 1000, 'num_inference_timesteps': 4, 'diffusion_model_cfg': {'cross_attention_dim': 2048, 'dropout': 0.2, 'final_dropout': True, 'interleave_self_attention': True, 'norm_type': 'ada_norm', 'num_layers': 16, 'output_dim': 2560, 'positional_embeddings': None}}, 'freeze_vlm_backbone': False}, 'inference_model': {'type': 'DiT4DiTVLA', 'pretrained_name_or_path': None, 'name_mapping': None, 'strict_mapping': False, 'repeated_diffusion_steps': 4, 'vlm_backbone': {'type': 'Cosmos25Backbone', 'base_model': './checkpoints/Cosmos-Predict2.5-2B', 'revision': 'diffusers/base/post-trained', 'torch_dtype': 'bf16', 'local_files_only': True, 'extract_layer': 17, 'max_sequence_length': 512, 'trainable': True, 'frozen_submodules': ['text_encoder', 'vae'], 'split_future_frames': True, 'num_frames_out': 5, 'fixed_seed': None, 'num_inference_steps': 1, 'conditional_frame_timestep': 0.0001, 'future_loss_type': 'flow_matching', 'detach_hidden_states': True, 'flow_matching_time_distribution': 'uniform', 'flow_matching_high_sigma_ratio': None, 'flow_matching_high_sigma_min': None, 'fsdp_min_num_params': 0}, 'vla_head': {'type': 'DiT4DiTActionHead', 'action_model_type': 'DiT-B', 'hidden_size': 2560, 'add_pos_embed': True, 'max_seq_len': 1024, 'action_dim': 8, 'ori_action_dim': 7, 'state_dim': 16, 'future_action_window_size': 7, 'action_horizon': 8, 'noise_beta_alpha': 1.5, 'noise_beta_beta': 1.0, 'noise_s': 0.999, 'num_timestep_buckets': 1000, 'num_inference_timesteps': 4, 'diffusion_model_cfg': {'cross_attention_dim': 2048, 'dropout': 0.2, 'final_dropout': True, 'interleave_self_attention': True, 'norm_type': 'ada_norm', 'num_layers': 16, 'output_dim': 2560, 'positional_embeddings': None}}, 'freeze_vlm_backbone': False}, '_dit4dit_train_transforms': [{'type': 'ProcessParquetInputs', 'parquet_keys': ['observation.state', 'timestamp', 'actions', 'info', 'stats', 'action_masks'], 'video_keys': ['observation.images.image', 'observation.images.wrist_image'], 'name_mappings': {'observation.state': ['states'], 'actions': ['actions']}, 'frame_stride': 2}, {'type': 'CanonicalizePrompt', 'output_key': 'prompt', 'input_keys': ['prompt', 'task_description', 'lang', 'instruction', 'instructions', 'text'], 'remove_source_keys': True}, {'type': 'ProcessCosmos25Prompt', 'tokenizer': {'type': 'PretrainedTokenizer', 'model_path': './checkpoints/Cosmos-Predict2.5-2B/tokenizer', 'model_max_length': 512}, 'input_key': 'prompt', 'remove_input_key': True}, {'type': 'ResizeImages', 'height': 224, 'width': 224, 'backend': 'torch', 'scale_divisor': 255.0, 'output_layout': 'nchw'}, {'type': 'ConcatImagesHorizontally', 'key': 'images', 'num_views': 2, 'frame_stride': 1, 'views_first': True, 'keep_time_dim': True}, {'type': 'SinCosKeys', 'keys': ['states'], 'target_dims': {'states': 16}, 'interleave': True, 'backend': 'torch'}, {'type': 'NormalizeStatesAndActions', 'state_key': 'proprio', 'action_key': 'action', 'action_dim': None, 'state_dim': None, 'state_norm_type': 'none', 'action_norm_type': 'min_max', 'norm_type': 'min_max', 'action_norm_mask': [True, True, True, True, True, True, False], 'clip_norm': False, 'normalization_epsilon': 0.0, 'preserve_input_dtype': True, 'output_dtype': 'float16'}, {'type': 'PadActionsAndActionMasks', 'action_dim': 8, 'valid_action_dim': 7}, {'type': 'PrepareDiT4DiTInputs', 'state_dim': 16, 'action_horizon': 8, 'action_dim': 8, 'valid_action_dim': 7, 'mark_all_action_steps_valid': True}], '_dit4dit_parquet_dataset': {'type': 'ParquetDataset', 'transforms': [{'type': 'ProcessParquetInputs', 'parquet_keys': ['observation.state', 'timestamp', 'actions', 'info', 'stats', 'action_masks'], 'video_keys': ['observation.images.image', 'observation.images.wrist_image'], 'name_mappings': {'observation.state': ['states'], 'actions': ['actions']}, 'frame_stride': 2}, {'type': 'CanonicalizePrompt', 'output_key': 'prompt', 'input_keys': ['prompt', 'task_description', 'lang', 'instruction', 'instructions', 'text'], 'remove_source_keys': True}, {'type': 'ProcessCosmos25Prompt', 'tokenizer': {'type': 'PretrainedTokenizer', 'model_path': './checkpoints/Cosmos-Predict2.5-2B/tokenizer', 'model_max_length': 512}, 'input_key': 'prompt', 'remove_input_key': True}, {'type': 'ResizeImages', 'height': 224, 'width': 224, 'backend': 'torch', 'scale_divisor': 255.0, 'output_layout': 'nchw'}, {'type': 'ConcatImagesHorizontally', 'key': 'images', 'num_views': 2, 'frame_stride': 1, 'views_first': True, 'keep_time_dim': True}, {'type': 'SinCosKeys', 'keys': ['states'], 'target_dims': {'states': 16}, 'interleave': True, 'backend': 'torch'}, {'type': 'NormalizeStatesAndActions', 'state_key': 'proprio', 'action_key': 'action', 'action_dim': None, 'state_dim': None, 'state_norm_type': 'none', 'action_norm_type': 'min_max', 'norm_type': 'min_max', 'action_norm_mask': [True, True, True, True, True, True, False], 'clip_norm': False, 'normalization_epsilon': 0.0, 'preserve_input_dtype': True, 'output_dtype': 'float16'}, {'type': 'PadActionsAndActionMasks', 'action_dim': 8, 'valid_action_dim': 7}, {'type': 'PrepareDiT4DiTInputs', 'state_dim': 16, 'action_horizon': 8, 'action_dim': 8, 'valid_action_dim': 7, 'mark_all_action_steps_valid': True}], 'action_window_size': 8, 'action_key': 'action', 'use_delta': False, 'statistic_name': 'franka', 'window_start_idx': 0, 'frame_window_size': 9}, 'train_dataloader': {'per_device_batch_size': 8, 'per_device_num_workers': 4, 'dataset': {'type': 'DistributedBalancedRepeatingDataset', 'name_mappings': {'observation.state': ['proprio'], 'action': ['action']}, 'statistic_keys': ['observation.state', 'timestamp', 'action'], 'statistic_name': 'franka', 'datasets': [{'type': 'ParquetDataset', 'transforms': [{'type': 'ProcessParquetInputs', 'parquet_keys': ['observation.state', 'timestamp', 'actions', 'info', 'stats', 'action_masks'], 'video_keys': ['observation.images.image', 'observation.images.wrist_image'], 'name_mappings': {'observation.state': ['states'], 'actions': ['actions']}, 'frame_stride': 2}, {'type': 'CanonicalizePrompt', 'output_key': 'prompt', 'input_keys': ['prompt', 'task_description', 'lang', 'instruction', 'instructions', 'text'], 'remove_source_keys': True}, {'type': 'ProcessCosmos25Prompt', 'tokenizer': {'type': 'PretrainedTokenizer', 'model_path': './checkpoints/Cosmos-Predict2.5-2B/tokenizer', 'model_max_length': 512}, 'input_key': 'prompt', 'remove_input_key': True}, {'type': 'ResizeImages', 'height': 224, 'width': 224, 'backend': 'torch', 'scale_divisor': 255.0, 'output_layout': 'nchw'}, {'type': 'ConcatImagesHorizontally', 'key': 'images', 'num_views': 2, 'frame_stride': 1, 'views_first': True, 'keep_time_dim': True}, {'type': 'SinCosKeys', 'keys': ['states'], 'target_dims': {'states': 16}, 'interleave': True, 'backend': 'torch'}, {'type': 'NormalizeStatesAndActions', 'state_key': 'proprio', 'action_key': 'action', 'action_dim': None, 'state_dim': None, 'state_norm_type': 'none', 'action_norm_type': 'min_max', 'norm_type': 'min_max', 'action_norm_mask': [True, True, True, True, True, True, False], 'clip_norm': False, 'normalization_epsilon': 0.0, 'preserve_input_dtype': True, 'output_dtype': 'float16'}, {'type': 'PadActionsAndActionMasks', 'action_dim': 8, 'valid_action_dim': 7}, {'type': 'PrepareDiT4DiTInputs', 'state_dim': 16, 'action_horizon': 8, 'action_dim': 8, 'valid_action_dim': 7, 'mark_all_action_steps_valid': True}], 'action_window_size': 8, 'action_key': 'action', 'use_delta': False, 'statistic_name': 'franka', 'window_start_idx': 0, 'frame_window_size': 9, 'data_root_path': './datasets/libero_object_no_noops_lerobotv2.1'}, {'type': 'ParquetDataset', 'transforms': [{'type': 'ProcessParquetInputs', 'parquet_keys': ['observation.state', 'timestamp', 'actions', 'info', 'stats', 'action_masks'], 'video_keys': ['observation.images.image', 'observation.images.wrist_image'], 'name_mappings': {'observation.state': ['states'], 'actions': ['actions']}, 'frame_stride': 2}, {'type': 'CanonicalizePrompt', 'output_key': 'prompt', 'input_keys': ['prompt', 'task_description', 'lang', 'instruction', 'instructions', 'text'], 'remove_source_keys': True}, {'type': 'ProcessCosmos25Prompt', 'tokenizer': {'type': 'PretrainedTokenizer', 'model_path': './checkpoints/Cosmos-Predict2.5-2B/tokenizer', 'model_max_length': 512}, 'input_key': 'prompt', 'remove_input_key': True}, {'type': 'ResizeImages', 'height': 224, 'width': 224, 'backend': 'torch', 'scale_divisor': 255.0, 'output_layout': 'nchw'}, {'type': 'ConcatImagesHorizontally', 'key': 'images', 'num_views': 2, 'frame_stride': 1, 'views_first': True, 'keep_time_dim': True}, {'type': 'SinCosKeys', 'keys': ['states'], 'target_dims': {'states': 16}, 'interleave': True, 'backend': 'torch'}, {'type': 'NormalizeStatesAndActions', 'state_key': 'proprio', 'action_key': 'action', 'action_dim': None, 'state_dim': None, 'state_norm_type': 'none', 'action_norm_type': 'min_max', 'norm_type': 'min_max', 'action_norm_mask': [True, True, True, True, True, True, False], 'clip_norm': False, 'normalization_epsilon': 0.0, 'preserve_input_dtype': True, 'output_dtype': 'float16'}, {'type': 'PadActionsAndActionMasks', 'action_dim': 8, 'valid_action_dim': 7}, {'type': 'PrepareDiT4DiTInputs', 'state_dim': 16, 'action_horizon': 8, 'action_dim': 8, 'valid_action_dim': 7, 'mark_all_action_steps_valid': True}], 'action_window_size': 8, 'action_key': 'action', 'use_delta': False, 'statistic_name': 'franka', 'window_start_idx': 0, 'frame_window_size': 9, 'data_root_path': './datasets/libero_goal_no_noops_lerobotv2.1'}, {'type': 'ParquetDataset', 'transforms': [{'type': 'ProcessParquetInputs', 'parquet_keys': ['observation.state', 'timestamp', 'actions', 'info', 'stats', 'action_masks'], 'video_keys': ['observation.images.image', 'observation.images.wrist_image'], 'name_mappings': {'observation.state': ['states'], 'actions': ['actions']}, 'frame_stride': 2}, {'type': 'CanonicalizePrompt', 'output_key': 'prompt', 'input_keys': ['prompt', 'task_description', 'lang', 'instruction', 'instructions', 'text'], 'remove_source_keys': True}, {'type': 'ProcessCosmos25Prompt', 'tokenizer': {'type': 'PretrainedTokenizer', 'model_path': './checkpoints/Cosmos-Predict2.5-2B/tokenizer', 'model_max_length': 512}, 'input_key': 'prompt', 'remove_input_key': True}, {'type': 'ResizeImages', 'height': 224, 'width': 224, 'backend': 'torch', 'scale_divisor': 255.0, 'output_layout': 'nchw'}, {'type': 'ConcatImagesHorizontally', 'key': 'images', 'num_views': 2, 'frame_stride': 1, 'views_first': True, 'keep_time_dim': True}, {'type': 'SinCosKeys', 'keys': ['states'], 'target_dims': {'states': 16}, 'interleave': True, 'backend': 'torch'}, {'type': 'NormalizeStatesAndActions', 'state_key': 'proprio', 'action_key': 'action', 'action_dim': None, 'state_dim': None, 'state_norm_type': 'none', 'action_norm_type': 'min_max', 'norm_type': 'min_max', 'action_norm_mask': [True, True, True, True, True, True, False], 'clip_norm': False, 'normalization_epsilon': 0.0, 'preserve_input_dtype': True, 'output_dtype': 'float16'}, {'type': 'PadActionsAndActionMasks', 'action_dim': 8, 'valid_action_dim': 7}, {'type': 'PrepareDiT4DiTInputs', 'state_dim': 16, 'action_horizon': 8, 'action_dim': 8, 'valid_action_dim': 7, 'mark_all_action_steps_valid': True}], 'action_window_size': 8, 'action_key': 'action', 'use_delta': False, 'statistic_name': 'franka', 'window_start_idx': 0, 'frame_window_size': 9, 'data_root_path': './datasets/libero_spatial_no_noops_lerobotv2.1'}, {'type': 'ParquetDataset', 'transforms': [{'type': 'ProcessParquetInputs', 'parquet_keys': ['observation.state', 'timestamp', 'actions', 'info', 'stats', 'action_masks'], 'video_keys': ['observation.images.image', 'observation.images.wrist_image'], 'name_mappings': {'observation.state': ['states'], 'actions': ['actions']}, 'frame_stride': 2}, {'type': 'CanonicalizePrompt', 'output_key': 'prompt', 'input_keys': ['prompt', 'task_description', 'lang', 'instruction', 'instructions', 'text'], 'remove_source_keys': True}, {'type': 'ProcessCosmos25Prompt', 'tokenizer': {'type': 'PretrainedTokenizer', 'model_path': './checkpoints/Cosmos-Predict2.5-2B/tokenizer', 'model_max_length': 512}, 'input_key': 'prompt', 'remove_input_key': True}, {'type': 'ResizeImages', 'height': 224, 'width': 224, 'backend': 'torch', 'scale_divisor': 255.0, 'output_layout': 'nchw'}, {'type': 'ConcatImagesHorizontally', 'key': 'images', 'num_views': 2, 'frame_stride': 1, 'views_first': True, 'keep_time_dim': True}, {'type': 'SinCosKeys', 'keys': ['states'], 'target_dims': {'states': 16}, 'interleave': True, 'backend': 'torch'}, {'type': 'NormalizeStatesAndActions', 'state_key': 'proprio', 'action_key': 'action', 'action_dim': None, 'state_dim': None, 'state_norm_type': 'none', 'action_norm_type': 'min_max', 'norm_type': 'min_max', 'action_norm_mask': [True, True, True, True, True, True, False], 'clip_norm': False, 'normalization_epsilon': 0.0, 'preserve_input_dtype': True, 'output_dtype': 'float16'}, {'type': 'PadActionsAndActionMasks', 'action_dim': 8, 'valid_action_dim': 7}, {'type': 'PrepareDiT4DiTInputs', 'state_dim': 16, 'action_horizon': 8, 'action_dim': 8, 'valid_action_dim': 7, 'mark_all_action_steps_valid': True}], 'action_window_size': 8, 'action_key': 'action', 'use_delta': False, 'statistic_name': 'franka', 'window_start_idx': 0, 'frame_window_size': 9, 'data_root_path': './datasets/libero_10_no_noops_lerobotv2.1'}], 'sampling_weights': [1.0, 1.0, 1.0, 1.0], 'shuffle': False, 'reshuffle_each_epoch': True, 'seed': 42, 'dataset_statistics': {'franka': {'action': {'mean': [0.07237596483901143, 0.08987006871029735, -0.10144743137061596, -0.00045383188989944756, 0.006273590726777911, -0.003878799732774496, 0.524486355483532], 'std': [0.3498823308902479, 0.37794140366375184, 0.460084266976933, 0.0403885784928603, 0.06616144248501059, 0.07763074391911857, 0.4994683356809767], 'max': [0.9375, 0.9375, 0.9375, 0.3557142913341522, 0.375, 0.375, 1.0], 'min': [-0.9375, -0.9375, -0.9375, -0.2582142949104309, -0.375, -0.3675000071525574, 0.0], 'q01': [-0.8785714507102966, -0.8758928775787354, -0.9375, -0.1510714292526245, -0.20678570866584778, -0.2742857038974762, 0.0], 'q99': [0.9375, 0.9107142686843872, 0.9375, 0.20357142388820648, 0.26357144117355347, 0.375, 1.0], 'mask': [True, True, True, True, True, True, False]}, 'state': {'mean': [-0.04889854742214084, 0.03689368185587227, 0.7890402488410473, 2.9771945476531982, -0.1417286954820156, -0.11769362539052963, 0.026436020154505968, -0.02665513101965189], 'std': [0.10639013941746686, 0.15115733130675715, 0.38406895599530033, 0.3530238395244304, 0.8227341427331599, 0.32357567121520087, 0.014583991652936385, 0.014467005007200339], 'max': [0.21031762659549713, 0.39128610491752625, 1.3660105466842651, 3.6714255809783936, 3.560650587081909, 1.386339545249939, 0.04233968257904053, 0.0013633022317662835], 'min': [-0.4828203022480011, -0.3255046010017395, 0.008128180168569088, 0.35277295112609863, -3.641430377960205, -1.842738389968872, -0.0013586411951109767, -0.042040832340717316], 'q01': [-0.42401049643754957, -0.2838300323486328, 0.009925739830359817, 1.3085840785503386, -2.886677579879761, -1.1599004411697387, 0.001503719249740243, -0.040336399003863335], 'q99': [0.1530261474847791, 0.3629165390133857, 1.2910678112506866, 3.303542451858519, 2.7496529006957933, 0.6893712210655194, 0.040610933862626555, -0.0015016929572448147]}, 'num_transitions': 273465, 'num_trajectories': 1693}}}}, 'runner': {'type': 'FSDPTrainRunner', 'max_steps': 160000, 'grad_accumulation_steps': 2, 'optimizer': {'lr': 3e-05, 'type': 'AdamW', 'weight_decay': 1e-08, 'eps': 1e-08, 'betas': (0.9, 0.95), 'paramwise_learning_rate': {'vlm_backbone.transformer': 0.0001, 'vla_head': 0.0001}}, 'max_grad_norm': 1.0, 'save_iter_interval': 40000, 'max_keep_ckpts': 2, 'collator': {'type': 'DictCollator', 'keys': ['states', 'timestamp', 'images', 'img_masks', 'actions', 'action_masks', 'frame_masks', 'lang_tokens', 'lang_masks'], 'meta_keys': ['info', 'stats']}, 'tokenizer': {'type': 'PretrainedTokenizer', 'model_path': './checkpoints/Cosmos-Predict2.5-2B/tokenizer', 'model_max_length': 512}, 'sampler': None, 'metric': {'type': 'VLAMetric', 'active_trackers': ('jsonl', 'wandb'), 'run_dir': './work_dirs/dit4dit_libero_all_full_finetune_bc071270ff2_bs256', 'grad_accumulation_steps': 1, 'window_size': 1}, 'lr_scheduler': {'type': 'cosine_with_min_lr', 'warmup_steps': 10000, 'min_lr': 5e-07}, 'sharding_strategy': 'shard-grad-op', 'pre_fsdp_param_dtype': 'fp32', 'enable_gradient_checkpointing': False, 'enable_mixed_precision_training': True, 'mixed_precision_dtype': 'bf16', 'reduce_in_full_precision': False, 'change_key_name': False, 'cfg': Config (path: configs/dit4dit/dit4dit_libero_all_full_finetune.py): {'_ckpt_root': './checkpoints', '_cosmos_base_model': './checkpoints/Cosmos-Predict2.5-2B', '_cosmos_revision': 'diffusers/base/post-trained', '_cosmos_tokenizer': {'type': 'PretrainedTokenizer', 'model_path': './checkpoints/Cosmos-Predict2.5-2B/tokenizer', 'model_max_length': 512}, '_official_dit4dit_ckpt': './checkpoints/dit4dit-model/dit4dit_libero/final_model/pytorch_model.pt', '_official_dataset_statistics': {'franka': {'action': {'mean': [0.07237596483901143, 0.08987006871029735, -0.10144743137061596, -0.00045383188989944756, 0.006273590726777911, -0.003878799732774496, 0.524486355483532], 'std': [0.3498823308902479, 0.37794140366375184, 0.460084266976933, 0.0403885784928603, 0.06616144248501059, 0.07763074391911857, 0.4994683356809767], 'max': [0.9375, 0.9375, 0.9375, 0.3557142913341522, 0.375, 0.375, 1.0], 'min': [-0.9375, -0.9375, -0.9375, -0.2582142949104309, -0.375, -0.3675000071525574, 0.0], 'q01': [-0.8785714507102966, -0.8758928775787354, -0.9375, -0.1510714292526245, -0.20678570866584778, -0.2742857038974762, 0.0], 'q99': [0.9375, 0.9107142686843872, 0.9375, 0.20357142388820648, 0.26357144117355347, 0.375, 1.0], 'mask': [True, True, True, True, True, True, False]}, 'state': {'mean': [-0.04889854742214084, 0.03689368185587227, 0.7890402488410473, 2.9771945476531982, -0.1417286954820156, -0.11769362539052963, 0.026436020154505968, -0.02665513101965189], 'std': [0.10639013941746686, 0.15115733130675715, 0.38406895599530033, 0.3530238395244304, 0.8227341427331599, 0.32357567121520087, 0.014583991652936385, 0.014467005007200339], 'max': [0.21031762659549713, 0.39128610491752625, 1.3660105466842651, 3.6714255809783936, 3.560650587081909, 1.386339545249939, 0.04233968257904053, 0.0013633022317662835], 'min': [-0.4828203022480011, -0.3255046010017395, 0.008128180168569088, 0.35277295112609863, -3.641430377960205, -1.842738389968872, -0.0013586411951109767, -0.042040832340717316], 'q01': [-0.42401049643754957, -0.2838300323486328, 0.009925739830359817, 1.3085840785503386, -2.886677579879761, -1.1599004411697387, 0.001503719249740243, -0.040336399003863335], 'q99': [0.1530261474847791, 0.3629165390133857, 1.2910678112506866, 3.303542451858519, 2.7496529006957933, 0.6893712210655194, 0.040610933862626555, -0.0015016929572448147]}, 'num_transitions': 273465, 'num_trajectories': 1693}}, '_action_dim': 8, '_ori_action_dim': 7, '_state_dim': 16, '_action_horizon': 8, '_frame_window_size': 9, '_image_frame_stride': 2, '_image_size': 224, '_action_norm_mask': [True, True, True, True, True, True, False], '_libero_data_roots': ['./datasets/libero_object_no_noops_lerobotv2.1', './datasets/libero_goal_no_noops_lerobotv2.1', './datasets/libero_spatial_no_noops_lerobotv2.1', './datasets/libero_10_no_noops_lerobotv2.1'], '_prompt_input_keys': ['prompt', 'task_description', 'lang', 'instruction', 'instructions', 'text'], 'seed': 42, 'official_eval_task_suites': ['libero_spatial', 'libero_object', 'libero_goal', 'libero_10'], 'model': {'type': 'DiT4DiTVLA', 'pretrained_name_or_path': None, 'name_mapping': {'vlm_backbone.text_encoder': 'backbone_interface.extractor.text_encoder', 'vlm_backbone.transformer': 'backbone_interface.extractor.transformer', 'vlm_backbone.vae': 'backbone_interface.extractor.vae', 'vla_head': 'action_model'}, 'strict_mapping': False, 'repeated_diffusion_steps': 4, 'vlm_backbone': {'type': 'Cosmos25Backbone', 'base_model': './checkpoints/Cosmos-Predict2.5-2B', 'revision': 'diffusers/base/post-trained', 'torch_dtype': 'bf16', 'local_files_only': True, 'extract_layer': 17, 'max_sequence_length': 512, 'trainable': True, 'frozen_submodules': ['text_encoder', 'vae'], 'split_future_frames': True, 'num_frames_out': 5, 'fixed_seed': None, 'num_inference_steps': 1, 'conditional_frame_timestep': 0.0001, 'future_loss_type': 'flow_matching', 'detach_hidden_states': True, 'flow_matching_time_distribution': 'uniform', 'flow_matching_high_sigma_ratio': None, 'flow_matching_high_sigma_min': None, 'fsdp_min_num_params': 0}, 'vla_head': {'type': 'DiT4DiTActionHead', 'action_model_type': 'DiT-B', 'hidden_size': 2560, 'add_pos_embed': True, 'max_seq_len': 1024, 'action_dim': 8, 'ori_action_dim': 7, 'state_dim': 16, 'future_action_window_size': 7, 'action_horizon': 8, 'noise_beta_alpha': 1.5, 'noise_beta_beta': 1.0, 'noise_s': 0.999, 'num_timestep_buckets': 1000, 'num_inference_timesteps': 4, 'diffusion_model_cfg': {'cross_attention_dim': 2048, 'dropout': 0.2, 'final_dropout': True, 'interleave_self_attention': True, 'norm_type': 'ada_norm', 'num_layers': 16, 'output_dim': 2560, 'positional_embeddings': None}}, 'freeze_vlm_backbone': False}, 'inference_model': {'type': 'DiT4DiTVLA', 'pretrained_name_or_path': None, 'name_mapping': None, 'strict_mapping': False, 'repeated_diffusion_steps': 4, 'vlm_backbone': {'type': 'Cosmos25Backbone', 'base_model': './checkpoints/Cosmos-Predict2.5-2B', 'revision': 'diffusers/base/post-trained', 'torch_dtype': 'bf16', 'local_files_only': True, 'extract_layer': 17, 'max_sequence_length': 512, 'trainable': True, 'frozen_submodules': ['text_encoder', 'vae'], 'split_future_frames': True, 'num_frames_out': 5, 'fixed_seed': None, 'num_inference_steps': 1, 'conditional_frame_timestep': 0.0001, 'future_loss_type': 'flow_matching', 'detach_hidden_states': True, 'flow_matching_time_distribution': 'uniform', 'flow_matching_high_sigma_ratio': None, 'flow_matching_high_sigma_min': None, 'fsdp_min_num_params': 0}, 'vla_head': {'type': 'DiT4DiTActionHead', 'action_model_type': 'DiT-B', 'hidden_size': 2560, 'add_pos_embed': True, 'max_seq_len': 1024, 'action_dim': 8, 'ori_action_dim': 7, 'state_dim': 16, 'future_action_window_size': 7, 'action_horizon': 8, 'noise_beta_alpha': 1.5, 'noise_beta_beta': 1.0, 'noise_s': 0.999, 'num_timestep_buckets': 1000, 'num_inference_timesteps': 4, 'diffusion_model_cfg': {'cross_attention_dim': 2048, 'dropout': 0.2, 'final_dropout': True, 'interleave_self_attention': True, 'norm_type': 'ada_norm', 'num_layers': 16, 'output_dim': 2560, 'positional_embeddings': None}}, 'freeze_vlm_backbone': False}, '_dit4dit_train_transforms': [{'type': 'ProcessParquetInputs', 'parquet_keys': ['observation.state', 'timestamp', 'actions', 'info', 'stats', 'action_masks'], 'video_keys': ['observation.images.image', 'observation.images.wrist_image'], 'name_mappings': {'observation.state': ['states'], 'actions': ['actions']}, 'frame_stride': 2}, {'type': 'CanonicalizePrompt', 'output_key': 'prompt', 'input_keys': ['prompt', 'task_description', 'lang', 'instruction', 'instructions', 'text'], 'remove_source_keys': True}, {'type': 'ProcessCosmos25Prompt', 'tokenizer': {'type': 'PretrainedTokenizer', 'model_path': './checkpoints/Cosmos-Predict2.5-2B/tokenizer', 'model_max_length': 512}, 'input_key': 'prompt', 'remove_input_key': True}, {'type': 'ResizeImages', 'height': 224, 'width': 224, 'backend': 'torch', 'scale_divisor': 255.0, 'output_layout': 'nchw'}, {'type': 'ConcatImagesHorizontally', 'key': 'images', 'num_views': 2, 'frame_stride': 1, 'views_first': True, 'keep_time_dim': True}, {'type': 'SinCosKeys', 'keys': ['states'], 'target_dims': {'states': 16}, 'interleave': True, 'backend': 'torch'}, {'type': 'NormalizeStatesAndActions', 'state_key': 'proprio', 'action_key': 'action', 'action_dim': None, 'state_dim': None, 'state_norm_type': 'none', 'action_norm_type': 'min_max', 'norm_type': 'min_max', 'action_norm_mask': [True, True, True, True, True, True, False], 'clip_norm': False, 'normalization_epsilon': 0.0, 'preserve_input_dtype': True, 'output_dtype': 'float16'}, {'type': 'PadActionsAndActionMasks', 'action_dim': 8, 'valid_action_dim': 7}, {'type': 'PrepareDiT4DiTInputs', 'state_dim': 16, 'action_horizon': 8, 'action_dim': 8, 'valid_action_dim': 7, 'mark_all_action_steps_valid': True}], '_dit4dit_parquet_dataset': {'type': 'ParquetDataset', 'transforms': [{'type': 'ProcessParquetInputs', 'parquet_keys': ['observation.state', 'timestamp', 'actions', 'info', 'stats', 'action_masks'], 'video_keys': ['observation.images.image', 'observation.images.wrist_image'], 'name_mappings': {'observation.state': ['states'], 'actions': ['actions']}, 'frame_stride': 2}, {'type': 'CanonicalizePrompt', 'output_key': 'prompt', 'input_keys': ['prompt', 'task_description', 'lang', 'instruction', 'instructions', 'text'], 'remove_source_keys': True}, {'type': 'ProcessCosmos25Prompt', 'tokenizer': {'type': 'PretrainedTokenizer', 'model_path': './checkpoints/Cosmos-Predict2.5-2B/tokenizer', 'model_max_length': 512}, 'input_key': 'prompt', 'remove_input_key': True}, {'type': 'ResizeImages', 'height': 224, 'width': 224, 'backend': 'torch', 'scale_divisor': 255.0, 'output_layout': 'nchw'}, {'type': 'ConcatImagesHorizontally', 'key': 'images', 'num_views': 2, 'frame_stride': 1, 'views_first': True, 'keep_time_dim': True}, {'type': 'SinCosKeys', 'keys': ['states'], 'target_dims': {'states': 16}, 'interleave': True, 'backend': 'torch'}, {'type': 'NormalizeStatesAndActions', 'state_key': 'proprio', 'action_key': 'action', 'action_dim': None, 'state_dim': None, 'state_norm_type': 'none', 'action_norm_type': 'min_max', 'norm_type': 'min_max', 'action_norm_mask': [True, True, True, True, True, True, False], 'clip_norm': False, 'normalization_epsilon': 0.0, 'preserve_input_dtype': True, 'output_dtype': 'float16'}, {'type': 'PadActionsAndActionMasks', 'action_dim': 8, 'valid_action_dim': 7}, {'type': 'PrepareDiT4DiTInputs', 'state_dim': 16, 'action_horizon': 8, 'action_dim': 8, 'valid_action_dim': 7, 'mark_all_action_steps_valid': True}], 'action_window_size': 8, 'action_key': 'action', 'use_delta': False, 'statistic_name': 'franka', 'window_start_idx': 0, 'frame_window_size': 9}, 'train_dataloader': {'per_device_batch_size': 8, 'per_device_num_workers': 4, 'dataset': {'type': 'DistributedBalancedRepeatingDataset', 'name_mappings': {'observation.state': ['proprio'], 'action': ['action']}, 'statistic_keys': ['observation.state', 'timestamp', 'action'], 'statistic_name': 'franka', 'datasets': [{'type': 'ParquetDataset', 'transforms': [{'type': 'ProcessParquetInputs', 'parquet_keys': ['observation.state', 'timestamp', 'actions', 'info', 'stats', 'action_masks'], 'video_keys': ['observation.images.image', 'observation.images.wrist_image'], 'name_mappings': {'observation.state': ['states'], 'actions': ['actions']}, 'frame_stride': 2}, {'type': 'CanonicalizePrompt', 'output_key': 'prompt', 'input_keys': ['prompt', 'task_description', 'lang', 'instruction', 'instructions', 'text'], 'remove_source_keys': True}, {'type': 'ProcessCosmos25Prompt', 'tokenizer': {'type': 'PretrainedTokenizer', 'model_path': './checkpoints/Cosmos-Predict2.5-2B/tokenizer', 'model_max_length': 512}, 'input_key': 'prompt', 'remove_input_key': True}, {'type': 'ResizeImages', 'height': 224, 'width': 224, 'backend': 'torch', 'scale_divisor': 255.0, 'output_layout': 'nchw'}, {'type': 'ConcatImagesHorizontally', 'key': 'images', 'num_views': 2, 'frame_stride': 1, 'views_first': True, 'keep_time_dim': True}, {'type': 'SinCosKeys', 'keys': ['states'], 'target_dims': {'states': 16}, 'interleave': True, 'backend': 'torch'}, {'type': 'NormalizeStatesAndActions', 'state_key': 'proprio', 'action_key': 'action', 'action_dim': None, 'state_dim': None, 'state_norm_type': 'none', 'action_norm_type': 'min_max', 'norm_type': 'min_max', 'action_norm_mask': [True, True, True, True, True, True, False], 'clip_norm': False, 'normalization_epsilon': 0.0, 'preserve_input_dtype': True, 'output_dtype': 'float16'}, {'type': 'PadActionsAndActionMasks', 'action_dim': 8, 'valid_action_dim': 7}, {'type': 'PrepareDiT4DiTInputs', 'state_dim': 16, 'action_horizon': 8, 'action_dim': 8, 'valid_action_dim': 7, 'mark_all_action_steps_valid': True}], 'action_window_size': 8, 'action_key': 'action', 'use_delta': False, 'statistic_name': 'franka', 'window_start_idx': 0, 'frame_window_size': 9, 'data_root_path': './datasets/libero_object_no_noops_lerobotv2.1'}, {'type': 'ParquetDataset', 'transforms': [{'type': 'ProcessParquetInputs', 'parquet_keys': ['observation.state', 'timestamp', 'actions', 'info', 'stats', 'action_masks'], 'video_keys': ['observation.images.image', 'observation.images.wrist_image'], 'name_mappings': {'observation.state': ['states'], 'actions': ['actions']}, 'frame_stride': 2}, {'type': 'CanonicalizePrompt', 'output_key': 'prompt', 'input_keys': ['prompt', 'task_description', 'lang', 'instruction', 'instructions', 'text'], 'remove_source_keys': True}, {'type': 'ProcessCosmos25Prompt', 'tokenizer': {'type': 'PretrainedTokenizer', 'model_path': './checkpoints/Cosmos-Predict2.5-2B/tokenizer', 'model_max_length': 512}, 'input_key': 'prompt', 'remove_input_key': True}, {'type': 'ResizeImages', 'height': 224, 'width': 224, 'backend': 'torch', 'scale_divisor': 255.0, 'output_layout': 'nchw'}, {'type': 'ConcatImagesHorizontally', 'key': 'images', 'num_views': 2, 'frame_stride': 1, 'views_first': True, 'keep_time_dim': True}, {'type': 'SinCosKeys', 'keys': ['states'], 'target_dims': {'states': 16}, 'interleave': True, 'backend': 'torch'}, {'type': 'NormalizeStatesAndActions', 'state_key': 'proprio', 'action_key': 'action', 'action_dim': None, 'state_dim': None, 'state_norm_type': 'none', 'action_norm_type': 'min_max', 'norm_type': 'min_max', 'action_norm_mask': [True, True, True, True, True, True, False], 'clip_norm': False, 'normalization_epsilon': 0.0, 'preserve_input_dtype': True, 'output_dtype': 'float16'}, {'type': 'PadActionsAndActionMasks', 'action_dim': 8, 'valid_action_dim': 7}, {'type': 'PrepareDiT4DiTInputs', 'state_dim': 16, 'action_horizon': 8, 'action_dim': 8, 'valid_action_dim': 7, 'mark_all_action_steps_valid': True}], 'action_window_size': 8, 'action_key': 'action', 'use_delta': False, 'statistic_name': 'franka', 'window_start_idx': 0, 'frame_window_size': 9, 'data_root_path': './datasets/libero_goal_no_noops_lerobotv2.1'}, {'type': 'ParquetDataset', 'transforms': [{'type': 'ProcessParquetInputs', 'parquet_keys': ['observation.state', 'timestamp', 'actions', 'info', 'stats', 'action_masks'], 'video_keys': ['observation.images.image', 'observation.images.wrist_image'], 'name_mappings': {'observation.state': ['states'], 'actions': ['actions']}, 'frame_stride': 2}, {'type': 'CanonicalizePrompt', 'output_key': 'prompt', 'input_keys': ['prompt', 'task_description', 'lang', 'instruction', 'instructions', 'text'], 'remove_source_keys': True}, {'type': 'ProcessCosmos25Prompt', 'tokenizer': {'type': 'PretrainedTokenizer', 'model_path': './checkpoints/Cosmos-Predict2.5-2B/tokenizer', 'model_max_length': 512}, 'input_key': 'prompt', 'remove_input_key': True}, {'type': 'ResizeImages', 'height': 224, 'width': 224, 'backend': 'torch', 'scale_divisor': 255.0, 'output_layout': 'nchw'}, {'type': 'ConcatImagesHorizontally', 'key': 'images', 'num_views': 2, 'frame_stride': 1, 'views_first': True, 'keep_time_dim': True}, {'type': 'SinCosKeys', 'keys': ['states'], 'target_dims': {'states': 16}, 'interleave': True, 'backend': 'torch'}, {'type': 'NormalizeStatesAndActions', 'state_key': 'proprio', 'action_key': 'action', 'action_dim': None, 'state_dim': None, 'state_norm_type': 'none', 'action_norm_type': 'min_max', 'norm_type': 'min_max', 'action_norm_mask': [True, True, True, True, True, True, False], 'clip_norm': False, 'normalization_epsilon': 0.0, 'preserve_input_dtype': True, 'output_dtype': 'float16'}, {'type': 'PadActionsAndActionMasks', 'action_dim': 8, 'valid_action_dim': 7}, {'type': 'PrepareDiT4DiTInputs', 'state_dim': 16, 'action_horizon': 8, 'action_dim': 8, 'valid_action_dim': 7, 'mark_all_action_steps_valid': True}], 'action_window_size': 8, 'action_key': 'action', 'use_delta': False, 'statistic_name': 'franka', 'window_start_idx': 0, 'frame_window_size': 9, 'data_root_path': './datasets/libero_spatial_no_noops_lerobotv2.1'}, {'type': 'ParquetDataset', 'transforms': [{'type': 'ProcessParquetInputs', 'parquet_keys': ['observation.state', 'timestamp', 'actions', 'info', 'stats', 'action_masks'], 'video_keys': ['observation.images.image', 'observation.images.wrist_image'], 'name_mappings': {'observation.state': ['states'], 'actions': ['actions']}, 'frame_stride': 2}, {'type': 'CanonicalizePrompt', 'output_key': 'prompt', 'input_keys': ['prompt', 'task_description', 'lang', 'instruction', 'instructions', 'text'], 'remove_source_keys': True}, {'type': 'ProcessCosmos25Prompt', 'tokenizer': {'type': 'PretrainedTokenizer', 'model_path': './checkpoints/Cosmos-Predict2.5-2B/tokenizer', 'model_max_length': 512}, 'input_key': 'prompt', 'remove_input_key': True}, {'type': 'ResizeImages', 'height': 224, 'width': 224, 'backend': 'torch', 'scale_divisor': 255.0, 'output_layout': 'nchw'}, {'type': 'ConcatImagesHorizontally', 'key': 'images', 'num_views': 2, 'frame_stride': 1, 'views_first': True, 'keep_time_dim': True}, {'type': 'SinCosKeys', 'keys': ['states'], 'target_dims': {'states': 16}, 'interleave': True, 'backend': 'torch'}, {'type': 'NormalizeStatesAndActions', 'state_key': 'proprio', 'action_key': 'action', 'action_dim': None, 'state_dim': None, 'state_norm_type': 'none', 'action_norm_type': 'min_max', 'norm_type': 'min_max', 'action_norm_mask': [True, True, True, True, True, True, False], 'clip_norm': False, 'normalization_epsilon': 0.0, 'preserve_input_dtype': True, 'output_dtype': 'float16'}, {'type': 'PadActionsAndActionMasks', 'action_dim': 8, 'valid_action_dim': 7}, {'type': 'PrepareDiT4DiTInputs', 'state_dim': 16, 'action_horizon': 8, 'action_dim': 8, 'valid_action_dim': 7, 'mark_all_action_steps_valid': True}], 'action_window_size': 8, 'action_key': 'action', 'use_delta': False, 'statistic_name': 'franka', 'window_start_idx': 0, 'frame_window_size': 9, 'data_root_path': './datasets/libero_10_no_noops_lerobotv2.1'}], 'sampling_weights': [1.0, 1.0, 1.0, 1.0], 'shuffle': False, 'reshuffle_each_epoch': True, 'seed': 42, 'dataset_statistics': {'franka': {'action': {'mean': [0.07237596483901143, 0.08987006871029735, -0.10144743137061596, -0.00045383188989944756, 0.006273590726777911, -0.003878799732774496, 0.524486355483532], 'std': [0.3498823308902479, 0.37794140366375184, 0.460084266976933, 0.0403885784928603, 0.06616144248501059, 0.07763074391911857, 0.4994683356809767], 'max': [0.9375, 0.9375, 0.9375, 0.3557142913341522, 0.375, 0.375, 1.0], 'min': [-0.9375, -0.9375, -0.9375, -0.2582142949104309, -0.375, -0.3675000071525574, 0.0], 'q01': [-0.8785714507102966, -0.8758928775787354, -0.9375, -0.1510714292526245, -0.20678570866584778, -0.2742857038974762, 0.0], 'q99': [0.9375, 0.9107142686843872, 0.9375, 0.20357142388820648, 0.26357144117355347, 0.375, 1.0], 'mask': [True, True, True, True, True, True, False]}, 'state': {'mean': [-0.04889854742214084, 0.03689368185587227, 0.7890402488410473, 2.9771945476531982, -0.1417286954820156, -0.11769362539052963, 0.026436020154505968, -0.02665513101965189], 'std': [0.10639013941746686, 0.15115733130675715, 0.38406895599530033, 0.3530238395244304, 0.8227341427331599, 0.32357567121520087, 0.014583991652936385, 0.014467005007200339], 'max': [0.21031762659549713, 0.39128610491752625, 1.3660105466842651, 3.6714255809783936, 3.560650587081909, 1.386339545249939, 0.04233968257904053, 0.0013633022317662835], 'min': [-0.4828203022480011, -0.3255046010017395, 0.008128180168569088, 0.35277295112609863, -3.641430377960205, -1.842738389968872, -0.0013586411951109767, -0.042040832340717316], 'q01': [-0.42401049643754957, -0.2838300323486328, 0.009925739830359817, 1.3085840785503386, -2.886677579879761, -1.1599004411697387, 0.001503719249740243, -0.040336399003863335], 'q99': [0.1530261474847791, 0.3629165390133857, 1.2910678112506866, 3.303542451858519, 2.7496529006957933, 0.6893712210655194, 0.040610933862626555, -0.0015016929572448147]}, 'num_transitions': 273465, 'num_trajectories': 1693}}}}, 'runner': {'type': 'FSDPTrainRunner', 'max_steps': 160000, 'grad_accumulation_steps': 2, 'optimizer': {'lr': 3e-05, 'type': 'AdamW', 'weight_decay': 1e-08, 'eps': 1e-08, 'betas': (0.9, 0.95), 'paramwise_learning_rate': {'vlm_backbone.transformer': 0.0001, 'vla_head': 0.0001}}, 'max_grad_norm': 1.0, 'save_iter_interval': 40000, 'max_keep_ckpts': 2, 'collator': {'type': 'DictCollator', 'keys': ['states', 'timestamp', 'images', 'img_masks', 'actions', 'action_masks', 'frame_masks', 'lang_tokens', 'lang_masks'], 'meta_keys': ['info', 'stats']}, 'tokenizer': {'type': 'PretrainedTokenizer', 'model_path': './checkpoints/Cosmos-Predict2.5-2B/tokenizer', 'model_max_length': 512}, 'sampler': None, 'metric': {'type': 'VLAMetric', 'active_trackers': ('jsonl', 'wandb'), 'run_dir': './work_dirs/dit4dit_libero_all_full_finetune_bc071270ff2_bs256', 'grad_accumulation_steps': 1, 'window_size': 1}, 'lr_scheduler': {'type': 'cosine_with_min_lr', 'warmup_steps': 10000, 'min_lr': 5e-07}, 'sharding_strategy': 'shard-grad-op', 'pre_fsdp_param_dtype': 'fp32', 'enable_gradient_checkpointing': False, 'enable_mixed_precision_training': True, 'mixed_precision_dtype': 'bf16', 'reduce_in_full_precision': False, 'change_key_name': False, 'cfg': Config (path: configs/dit4dit/dit4dit_libero_all_full_finetune.py): {...}, 'args': Namespace(config='configs/dit4dit/dit4dit_libero_all_full_finetune.py', work_dir='./work_dirs/dit4dit_libero_all_full_finetune_bc071270ff2_bs256', cfg_options={'train_dataloader.per_device_batch_size': 8, 'runner.grad_accumulation_steps': 2}, eval_after_train=True, resume_from=None)}, 'eval': {'type': 'LiberoEvalRunner', 'task_suite_name': ['libero_spatial', 'libero_object', 'libero_goal', 'libero_10'], 'model_family': 'dit4dit', 'norm_stats_key': 'franka', 'eval_chunk_size': 8, 'truncate_action_chunk_at_horizon': True, 'resize_size': 224, 'num_trials_per_task': 50, 'num_steps_wait': 10, 'seed': 7, 'enable_mixed_precision_training': False, 'mixed_precision_dtype': 'bf16', 'dataset': {'type': 'LiberoParquetEvalDataset', 'img_buffer_len': 1, 'require_lang_tokens': True, 'transforms': [{'type': 'CanonicalizePrompt', 'output_key': 'prompt', 'input_keys': ['prompt', 'task_description', 'lang', 'instruction', 'instructions', 'text'], 'remove_source_keys': True}, {'type': 'ProcessCosmos25Prompt', 'tokenizer': {'type': 'PretrainedTokenizer', 'model_path': './checkpoints/Cosmos-Predict2.5-2B/tokenizer', 'model_max_length': 512}, 'input_key': 'prompt', 'remove_input_key': True}, {'type': 'ProcessDiT4DiTLiberoEvalInputs', 'img_keys': ['agentview_image', 'robot0_eye_in_hand_image'], 'image_size': 224}, {'type': 'LiberoProprioFromInputs', 'norm_type': 'none', 'pos_key': 'robot0_eef_pos', 'quat_key': 'robot0_eef_quat', 'gripper_key': 'robot0_gripper_qpos', 'state_dim': None, 'out_key': 'states'}, {'type': 'SinCosKeys', 'keys': ['states'], 'target_dims': {'states': 16}, 'interleave': True}, {'type': 'PrepareDiT4DiTInputs', 'image_key': 'pixel_values', 'state_dim': 16, 'require_actions': False}]}, 'denormalize_action': {'type': 'DenormalizeLiberoAction', 'norm_type': 'min_max', 'action_dim': 7, 'action_norm_mask': [True, True, True, True, True, True, False], 'clip_normalized_action': True, 'normalize_gripper_action': True, 'invert_gripper_action': True}}}, 'args': Namespace(config='configs/dit4dit/dit4dit_libero_all_full_finetune.py', work_dir='./work_dirs/dit4dit_libero_all_full_finetune_bc071270ff2_bs256', cfg_options={'train_dataloader.per_device_batch_size': 8, 'runner.grad_accumulation_steps': 2}, eval_after_train=True, resume_from=None)}, 'eval': {'type': 'LiberoEvalRunner', 'task_suite_name': ['libero_spatial', 'libero_object', 'libero_goal', 'libero_10'], 'model_family': 'dit4dit', 'norm_stats_key': 'franka', 'eval_chunk_size': 8, 'truncate_action_chunk_at_horizon': True, 'resize_size': 224, 'num_trials_per_task': 50, 'num_steps_wait': 10, 'seed': 7, 'enable_mixed_precision_training': False, 'mixed_precision_dtype': 'bf16', 'dataset': {'type': 'LiberoParquetEvalDataset', 'img_buffer_len': 1, 'require_lang_tokens': True, 'transforms': [{'type': 'CanonicalizePrompt', 'output_key': 'prompt', 'input_keys': ['prompt', 'task_description', 'lang', 'instruction', 'instructions', 'text'], 'remove_source_keys': True}, {'type': 'ProcessCosmos25Prompt', 'tokenizer': {'type': 'PretrainedTokenizer', 'model_path': './checkpoints/Cosmos-Predict2.5-2B/tokenizer', 'model_max_length': 512}, 'input_key': 'prompt', 'remove_input_key': True}, {'type': 'ProcessDiT4DiTLiberoEvalInputs', 'img_keys': ['agentview_image', 'robot0_eye_in_hand_image'], 'image_size': 224}, {'type': 'LiberoProprioFromInputs', 'norm_type': 'none', 'pos_key': 'robot0_eef_pos', 'quat_key': 'robot0_eef_quat', 'gripper_key': 'robot0_gripper_qpos', 'state_dim': None, 'out_key': 'states'}, {'type': 'SinCosKeys', 'keys': ['states'], 'target_dims': {'states': 16}, 'interleave': True}, {'type': 'PrepareDiT4DiTInputs', 'image_key': 'pixel_values', 'state_dim': 16, 'require_actions': False}]}, 'denormalize_action': {'type': 'DenormalizeLiberoAction', 'norm_type': 'min_max', 'action_dim': 7, 'action_norm_mask': [True, True, True, True, True, True, False], 'clip_normalized_action': True, 'normalize_gripper_action': True, 'invert_gripper_action': True}}}", "run_id": "dit4dit_libero_all_full_finetune_2026_08_12_20_51_03"}
 
 
dit4dit_libero_all_full_finetune_bc071270ff2_bs256/tokenizer/added_tokens.json DELETED
@@ -1,24 +0,0 @@
1
- {
2
- "</tool_call>": 151658,
3
- "<tool_call>": 151657,
4
- "<|box_end|>": 151649,
5
- "<|box_start|>": 151648,
6
- "<|endoftext|>": 151643,
7
- "<|file_sep|>": 151664,
8
- "<|fim_middle|>": 151660,
9
- "<|fim_pad|>": 151662,
10
- "<|fim_prefix|>": 151659,
11
- "<|fim_suffix|>": 151661,
12
- "<|im_end|>": 151645,
13
- "<|im_start|>": 151644,
14
- "<|image_pad|>": 151655,
15
- "<|object_ref_end|>": 151647,
16
- "<|object_ref_start|>": 151646,
17
- "<|quad_end|>": 151651,
18
- "<|quad_start|>": 151650,
19
- "<|repo_name|>": 151663,
20
- "<|video_pad|>": 151656,
21
- "<|vision_end|>": 151653,
22
- "<|vision_pad|>": 151654,
23
- "<|vision_start|>": 151652
24
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
dit4dit_libero_all_full_finetune_bc071270ff2_bs256/tokenizer/chat_template.jinja DELETED
@@ -1,7 +0,0 @@
1
- {% set image_count = namespace(value=0) %}{% set video_count = namespace(value=0) %}{% for message in messages %}{% if loop.first and message['role'] != 'system' %}<|im_start|>system
2
- You are a helpful assistant.<|im_end|>
3
- {% endif %}<|im_start|>{{ message['role'] }}
4
- {% if message['content'] is string %}{{ message['content'] }}<|im_end|>
5
- {% else %}{% for content in message['content'] %}{% if content['type'] == 'image' or 'image' in content or 'image_url' in content %}{% set image_count.value = image_count.value + 1 %}{% if add_vision_id %}Picture {{ image_count.value }}: {% endif %}<|vision_start|><|image_pad|><|vision_end|>{% elif content['type'] == 'video' or 'video' in content %}{% set video_count.value = video_count.value + 1 %}{% if add_vision_id %}Video {{ video_count.value }}: {% endif %}<|vision_start|><|video_pad|><|vision_end|>{% elif 'text' in content %}{{ content['text'] }}{% endif %}{% endfor %}<|im_end|>
6
- {% endif %}{% endfor %}{% if add_generation_prompt %}<|im_start|>assistant
7
- {% endif %}
 
 
 
 
 
 
 
 
dit4dit_libero_all_full_finetune_bc071270ff2_bs256/tokenizer/merges.txt DELETED
The diff for this file is too large to render. See raw diff
 
dit4dit_libero_all_full_finetune_bc071270ff2_bs256/tokenizer/special_tokens_map.json DELETED
@@ -1,31 +0,0 @@
1
- {
2
- "additional_special_tokens": [
3
- "<|im_start|>",
4
- "<|im_end|>",
5
- "<|object_ref_start|>",
6
- "<|object_ref_end|>",
7
- "<|box_start|>",
8
- "<|box_end|>",
9
- "<|quad_start|>",
10
- "<|quad_end|>",
11
- "<|vision_start|>",
12
- "<|vision_end|>",
13
- "<|vision_pad|>",
14
- "<|image_pad|>",
15
- "<|video_pad|>"
16
- ],
17
- "eos_token": {
18
- "content": "<|im_end|>",
19
- "lstrip": false,
20
- "normalized": false,
21
- "rstrip": false,
22
- "single_word": false
23
- },
24
- "pad_token": {
25
- "content": "<|endoftext|>",
26
- "lstrip": false,
27
- "normalized": false,
28
- "rstrip": false,
29
- "single_word": false
30
- }
31
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
dit4dit_libero_all_full_finetune_bc071270ff2_bs256/tokenizer/tokenizer.json DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:9c5ae00e602b8860cbd784ba82a8aa14e8feecec692e7076590d014d7b7fdafa
3
- size 11421896
 
 
 
 
dit4dit_libero_all_full_finetune_bc071270ff2_bs256/tokenizer/tokenizer_config.json DELETED
@@ -1,207 +0,0 @@
1
- {
2
- "add_bos_token": false,
3
- "add_prefix_space": false,
4
- "added_tokens_decoder": {
5
- "151643": {
6
- "content": "<|endoftext|>",
7
- "lstrip": false,
8
- "normalized": false,
9
- "rstrip": false,
10
- "single_word": false,
11
- "special": true
12
- },
13
- "151644": {
14
- "content": "<|im_start|>",
15
- "lstrip": false,
16
- "normalized": false,
17
- "rstrip": false,
18
- "single_word": false,
19
- "special": true
20
- },
21
- "151645": {
22
- "content": "<|im_end|>",
23
- "lstrip": false,
24
- "normalized": false,
25
- "rstrip": false,
26
- "single_word": false,
27
- "special": true
28
- },
29
- "151646": {
30
- "content": "<|object_ref_start|>",
31
- "lstrip": false,
32
- "normalized": false,
33
- "rstrip": false,
34
- "single_word": false,
35
- "special": true
36
- },
37
- "151647": {
38
- "content": "<|object_ref_end|>",
39
- "lstrip": false,
40
- "normalized": false,
41
- "rstrip": false,
42
- "single_word": false,
43
- "special": true
44
- },
45
- "151648": {
46
- "content": "<|box_start|>",
47
- "lstrip": false,
48
- "normalized": false,
49
- "rstrip": false,
50
- "single_word": false,
51
- "special": true
52
- },
53
- "151649": {
54
- "content": "<|box_end|>",
55
- "lstrip": false,
56
- "normalized": false,
57
- "rstrip": false,
58
- "single_word": false,
59
- "special": true
60
- },
61
- "151650": {
62
- "content": "<|quad_start|>",
63
- "lstrip": false,
64
- "normalized": false,
65
- "rstrip": false,
66
- "single_word": false,
67
- "special": true
68
- },
69
- "151651": {
70
- "content": "<|quad_end|>",
71
- "lstrip": false,
72
- "normalized": false,
73
- "rstrip": false,
74
- "single_word": false,
75
- "special": true
76
- },
77
- "151652": {
78
- "content": "<|vision_start|>",
79
- "lstrip": false,
80
- "normalized": false,
81
- "rstrip": false,
82
- "single_word": false,
83
- "special": true
84
- },
85
- "151653": {
86
- "content": "<|vision_end|>",
87
- "lstrip": false,
88
- "normalized": false,
89
- "rstrip": false,
90
- "single_word": false,
91
- "special": true
92
- },
93
- "151654": {
94
- "content": "<|vision_pad|>",
95
- "lstrip": false,
96
- "normalized": false,
97
- "rstrip": false,
98
- "single_word": false,
99
- "special": true
100
- },
101
- "151655": {
102
- "content": "<|image_pad|>",
103
- "lstrip": false,
104
- "normalized": false,
105
- "rstrip": false,
106
- "single_word": false,
107
- "special": true
108
- },
109
- "151656": {
110
- "content": "<|video_pad|>",
111
- "lstrip": false,
112
- "normalized": false,
113
- "rstrip": false,
114
- "single_word": false,
115
- "special": true
116
- },
117
- "151657": {
118
- "content": "<tool_call>",
119
- "lstrip": false,
120
- "normalized": false,
121
- "rstrip": false,
122
- "single_word": false,
123
- "special": false
124
- },
125
- "151658": {
126
- "content": "</tool_call>",
127
- "lstrip": false,
128
- "normalized": false,
129
- "rstrip": false,
130
- "single_word": false,
131
- "special": false
132
- },
133
- "151659": {
134
- "content": "<|fim_prefix|>",
135
- "lstrip": false,
136
- "normalized": false,
137
- "rstrip": false,
138
- "single_word": false,
139
- "special": false
140
- },
141
- "151660": {
142
- "content": "<|fim_middle|>",
143
- "lstrip": false,
144
- "normalized": false,
145
- "rstrip": false,
146
- "single_word": false,
147
- "special": false
148
- },
149
- "151661": {
150
- "content": "<|fim_suffix|>",
151
- "lstrip": false,
152
- "normalized": false,
153
- "rstrip": false,
154
- "single_word": false,
155
- "special": false
156
- },
157
- "151662": {
158
- "content": "<|fim_pad|>",
159
- "lstrip": false,
160
- "normalized": false,
161
- "rstrip": false,
162
- "single_word": false,
163
- "special": false
164
- },
165
- "151663": {
166
- "content": "<|repo_name|>",
167
- "lstrip": false,
168
- "normalized": false,
169
- "rstrip": false,
170
- "single_word": false,
171
- "special": false
172
- },
173
- "151664": {
174
- "content": "<|file_sep|>",
175
- "lstrip": false,
176
- "normalized": false,
177
- "rstrip": false,
178
- "single_word": false,
179
- "special": false
180
- }
181
- },
182
- "additional_special_tokens": [
183
- "<|im_start|>",
184
- "<|im_end|>",
185
- "<|object_ref_start|>",
186
- "<|object_ref_end|>",
187
- "<|box_start|>",
188
- "<|box_end|>",
189
- "<|quad_start|>",
190
- "<|quad_end|>",
191
- "<|vision_start|>",
192
- "<|vision_end|>",
193
- "<|vision_pad|>",
194
- "<|image_pad|>",
195
- "<|video_pad|>"
196
- ],
197
- "bos_token": null,
198
- "clean_up_tokenization_spaces": false,
199
- "eos_token": "<|im_end|>",
200
- "errors": "replace",
201
- "extra_special_tokens": {},
202
- "model_max_length": 131072,
203
- "pad_token": "<|endoftext|>",
204
- "split_special_tokens": false,
205
- "tokenizer_class": "Qwen2Tokenizer",
206
- "unk_token": null
207
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
dit4dit_libero_all_full_finetune_bc071270ff2_bs256/tokenizer/vocab.json DELETED
The diff for this file is too large to render. See raw diff