ShenBW commited on
Commit
6fa4f62
·
verified ·
1 Parent(s): b3685fe

Add DiT4DiT RoboTwin model metadata

Browse files
.gitattributes CHANGED
@@ -73,3 +73,4 @@ dit4dit_robodojo_full_finetune_bs128/tokenizer/tokenizer.json filter=lfs diff=lf
73
  pi0_paligemma_robotwin_all_data_full_finetune_bs128/tokenizer/tokenizer.json filter=lfs diff=lfs merge=lfs -text
74
  pi05_paligemma_robotwin_all_data_full_finetune_bs128/tokenizer/tokenizer.json filter=lfs diff=lfs merge=lfs -text
75
  fastwam_robotwin_all_data_full_finetune_bs256/tokenizer/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 
 
73
  pi0_paligemma_robotwin_all_data_full_finetune_bs128/tokenizer/tokenizer.json filter=lfs diff=lfs merge=lfs -text
74
  pi05_paligemma_robotwin_all_data_full_finetune_bs128/tokenizer/tokenizer.json filter=lfs diff=lfs merge=lfs -text
75
  fastwam_robotwin_all_data_full_finetune_bs256/tokenizer/tokenizer.json filter=lfs diff=lfs merge=lfs -text
76
+ dit4dit_robotwin_all_data_full_finetune_bs256/tokenizer/tokenizer.json filter=lfs diff=lfs merge=lfs -text
dit4dit_robotwin_all_data_full_finetune_bs256/config.json ADDED
@@ -0,0 +1,549 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_DATA_ROOTS": [
3
+ "datasets/robotwin_clean_lerobotv2.1",
4
+ "datasets/robotwin_randomized_lerobotv2.1"
5
+ ],
6
+ "_TASK_LIST": [
7
+ "adjust_bottle",
8
+ "beat_block_hammer",
9
+ "blocks_ranking_rgb",
10
+ "blocks_ranking_size",
11
+ "click_alarmclock",
12
+ "click_bell",
13
+ "dump_bin_bigbin",
14
+ "grab_roller",
15
+ "handover_block",
16
+ "handover_mic",
17
+ "hanging_mug",
18
+ "lift_pot",
19
+ "move_can_pot",
20
+ "move_pillbottle_pad",
21
+ "move_playingcard_away",
22
+ "move_stapler_pad",
23
+ "open_laptop",
24
+ "open_microwave",
25
+ "pick_diverse_bottles",
26
+ "pick_dual_bottles",
27
+ "place_a2b_left",
28
+ "place_a2b_right",
29
+ "place_bread_basket",
30
+ "place_bread_skillet",
31
+ "place_burger_fries",
32
+ "place_can_basket",
33
+ "place_cans_plasticbox",
34
+ "place_container_plate",
35
+ "place_dual_shoes",
36
+ "place_empty_cup",
37
+ "place_fan",
38
+ "place_mouse_pad",
39
+ "place_object_basket",
40
+ "place_object_scale",
41
+ "place_object_stand",
42
+ "place_phone_stand",
43
+ "place_shoe",
44
+ "press_stapler",
45
+ "put_bottles_dustbin",
46
+ "put_object_cabinet",
47
+ "rotate_qrcode",
48
+ "scan_object",
49
+ "shake_bottle",
50
+ "shake_bottle_horizontally",
51
+ "stack_blocks_three",
52
+ "stack_blocks_two",
53
+ "stack_bowls_three",
54
+ "stack_bowls_two",
55
+ "stamp_seal",
56
+ "turn_switch"
57
+ ],
58
+ "_TOKENIZER_PATH": "./checkpoints/Cosmos-Predict2.5-2B/tokenizer",
59
+ "eval": {
60
+ "dataset": {
61
+ "img_keys": [
62
+ "cam_high",
63
+ "cam_left_wrist",
64
+ "cam_right_wrist"
65
+ ],
66
+ "inject_model_path": false,
67
+ "transforms": [
68
+ {
69
+ "input_key": "task_description",
70
+ "remove_input_key": true,
71
+ "tokenizer": {
72
+ "model_max_length": 512,
73
+ "model_path": "./checkpoints/Cosmos-Predict2.5-2B/tokenizer",
74
+ "type": "PretrainedTokenizer"
75
+ },
76
+ "type": "ProcessCosmos25Prompt"
77
+ },
78
+ {
79
+ "backend": "torch",
80
+ "height": 224,
81
+ "output_layout": "flattened_chw",
82
+ "scale_divisor": 255.0,
83
+ "type": "ResizeImages",
84
+ "width": 224
85
+ },
86
+ {
87
+ "means": [
88
+ 0.5,
89
+ 0.5,
90
+ 0.5
91
+ ],
92
+ "stds": [
93
+ 0.5,
94
+ 0.5,
95
+ 0.5
96
+ ],
97
+ "type": "NormalizeImages"
98
+ },
99
+ {
100
+ "bottom_height_ratio": 0.5,
101
+ "bottom_views": [
102
+ 1,
103
+ 2
104
+ ],
105
+ "combine_view_masks": true,
106
+ "frame_window_size": 1,
107
+ "num_views": 3,
108
+ "tile_direction": "top_bottom_pair",
109
+ "top_view": 0,
110
+ "type": "PrepareVideo"
111
+ },
112
+ {
113
+ "backend": "torch",
114
+ "expand_axis": 0,
115
+ "interleave": true,
116
+ "keys": [
117
+ "states"
118
+ ],
119
+ "target_dims": {
120
+ "states": 32
121
+ },
122
+ "type": "SinCosKeys"
123
+ },
124
+ {
125
+ "action_dim": 16,
126
+ "action_key": "action",
127
+ "norm_type": "min_max",
128
+ "normalization_epsilon": 1e-06,
129
+ "output_dtype": "float32",
130
+ "state_dim": null,
131
+ "state_key": "proprio",
132
+ "state_norm_type": "none",
133
+ "type": "NormalizeStatesAndActions"
134
+ },
135
+ {
136
+ "action_horizon": 32,
137
+ "state_history_length": 1,
138
+ "type": "PrepareStateActionTargets",
139
+ "valid_action_dim": 14
140
+ }
141
+ ],
142
+ "type": "PrivateInferenceDataset"
143
+ },
144
+ "denormalize_action": {
145
+ "action_dim": 14,
146
+ "norm_type": "min_max",
147
+ "statistic_name": "robotwin_all",
148
+ "type": "DenormalizePrivateAction"
149
+ },
150
+ "eval_chunk_size": 24,
151
+ "instruction_type": "unseen",
152
+ "mixed_precision_dtype": "bf16",
153
+ "model_family": "dit4dit",
154
+ "num_trials_per_task": 100,
155
+ "seed": 7,
156
+ "task_list": [
157
+ "adjust_bottle",
158
+ "beat_block_hammer",
159
+ "blocks_ranking_rgb",
160
+ "blocks_ranking_size",
161
+ "click_alarmclock",
162
+ "click_bell",
163
+ "dump_bin_bigbin",
164
+ "grab_roller",
165
+ "handover_block",
166
+ "handover_mic",
167
+ "hanging_mug",
168
+ "lift_pot",
169
+ "move_can_pot",
170
+ "move_pillbottle_pad",
171
+ "move_playingcard_away",
172
+ "move_stapler_pad",
173
+ "open_laptop",
174
+ "open_microwave",
175
+ "pick_diverse_bottles",
176
+ "pick_dual_bottles",
177
+ "place_a2b_left",
178
+ "place_a2b_right",
179
+ "place_bread_basket",
180
+ "place_bread_skillet",
181
+ "place_burger_fries",
182
+ "place_can_basket",
183
+ "place_cans_plasticbox",
184
+ "place_container_plate",
185
+ "place_dual_shoes",
186
+ "place_empty_cup",
187
+ "place_fan",
188
+ "place_mouse_pad",
189
+ "place_object_basket",
190
+ "place_object_scale",
191
+ "place_object_stand",
192
+ "place_phone_stand",
193
+ "place_shoe",
194
+ "press_stapler",
195
+ "put_bottles_dustbin",
196
+ "put_object_cabinet",
197
+ "rotate_qrcode",
198
+ "scan_object",
199
+ "shake_bottle",
200
+ "shake_bottle_horizontally",
201
+ "stack_blocks_three",
202
+ "stack_blocks_two",
203
+ "stack_bowls_three",
204
+ "stack_bowls_two",
205
+ "stamp_seal",
206
+ "turn_switch"
207
+ ],
208
+ "task_suite_name": [
209
+ "clean",
210
+ "random"
211
+ ],
212
+ "type": "RobotwinEvalRunner",
213
+ "unnorm_key": "robotwin_all"
214
+ },
215
+ "inference_model": {
216
+ "freeze_vlm_backbone": false,
217
+ "init_empty_weights": true,
218
+ "repeated_diffusion_steps": 4,
219
+ "type": "DiT4DiTVLA",
220
+ "vla_head": {
221
+ "action_dim": 16,
222
+ "action_horizon": 32,
223
+ "action_model_type": "DiT-B",
224
+ "add_pos_embed": true,
225
+ "diffusion_model_cfg": {
226
+ "cross_attention_dim": 2048,
227
+ "dropout": 0.2,
228
+ "final_dropout": true,
229
+ "interleave_self_attention": true,
230
+ "norm_type": "ada_norm",
231
+ "num_layers": 16,
232
+ "output_dim": 2560,
233
+ "positional_embeddings": null
234
+ },
235
+ "future_action_window_size": 31,
236
+ "hidden_size": 2560,
237
+ "max_seq_len": 1024,
238
+ "noise_beta_alpha": 1.5,
239
+ "noise_beta_beta": 1.0,
240
+ "noise_s": 0.999,
241
+ "num_inference_timesteps": 4,
242
+ "num_timestep_buckets": 1000,
243
+ "ori_action_dim": 14,
244
+ "state_dim": 32,
245
+ "type": "DiT4DiTActionHead"
246
+ },
247
+ "vlm_backbone": {
248
+ "base_model": "./checkpoints/Cosmos-Predict2.5-2B",
249
+ "conditional_frame_timestep": 0.0001,
250
+ "detach_hidden_states": true,
251
+ "extract_layer": 17,
252
+ "fixed_seed": null,
253
+ "flow_matching_high_sigma_min": null,
254
+ "flow_matching_high_sigma_ratio": null,
255
+ "flow_matching_time_distribution": "uniform",
256
+ "frozen_submodules": [
257
+ "text_encoder",
258
+ "vae"
259
+ ],
260
+ "fsdp_min_num_params": 0,
261
+ "future_loss_type": "flow_matching",
262
+ "load_pretrained_weights": false,
263
+ "local_files_only": true,
264
+ "num_frames_out": 9,
265
+ "num_inference_steps": 1,
266
+ "revision": "diffusers/base/post-trained",
267
+ "split_future_frames": true,
268
+ "torch_dtype": "bf16",
269
+ "trainable": true,
270
+ "type": "Cosmos25Backbone"
271
+ }
272
+ },
273
+ "model": {
274
+ "freeze_vlm_backbone": false,
275
+ "repeated_diffusion_steps": 4,
276
+ "type": "DiT4DiTVLA",
277
+ "vla_head": {
278
+ "action_dim": 16,
279
+ "action_horizon": 32,
280
+ "action_model_type": "DiT-B",
281
+ "add_pos_embed": true,
282
+ "diffusion_model_cfg": {
283
+ "cross_attention_dim": 2048,
284
+ "dropout": 0.2,
285
+ "final_dropout": true,
286
+ "interleave_self_attention": true,
287
+ "norm_type": "ada_norm",
288
+ "num_layers": 16,
289
+ "output_dim": 2560,
290
+ "positional_embeddings": null
291
+ },
292
+ "future_action_window_size": 31,
293
+ "hidden_size": 2560,
294
+ "max_seq_len": 1024,
295
+ "noise_beta_alpha": 1.5,
296
+ "noise_beta_beta": 1.0,
297
+ "noise_s": 0.999,
298
+ "num_inference_timesteps": 4,
299
+ "num_timestep_buckets": 1000,
300
+ "ori_action_dim": 14,
301
+ "state_dim": 32,
302
+ "type": "DiT4DiTActionHead"
303
+ },
304
+ "vlm_backbone": {
305
+ "base_model": "./checkpoints/Cosmos-Predict2.5-2B",
306
+ "conditional_frame_timestep": 0.0001,
307
+ "detach_hidden_states": true,
308
+ "extract_layer": 17,
309
+ "fixed_seed": null,
310
+ "flow_matching_high_sigma_min": null,
311
+ "flow_matching_high_sigma_ratio": null,
312
+ "flow_matching_time_distribution": "uniform",
313
+ "frozen_submodules": [
314
+ "text_encoder",
315
+ "vae"
316
+ ],
317
+ "fsdp_min_num_params": 0,
318
+ "future_loss_type": "flow_matching",
319
+ "local_files_only": true,
320
+ "num_frames_out": 9,
321
+ "num_inference_steps": 1,
322
+ "revision": "diffusers/base/post-trained",
323
+ "split_future_frames": true,
324
+ "torch_dtype": "bf16",
325
+ "trainable": true,
326
+ "type": "Cosmos25Backbone"
327
+ }
328
+ },
329
+ "runner": {
330
+ "change_key_name": false,
331
+ "collator": {
332
+ "keys": [
333
+ "states",
334
+ "timestamp",
335
+ "images",
336
+ "img_masks",
337
+ "actions",
338
+ "action_masks",
339
+ "frame_masks",
340
+ "lang_tokens",
341
+ "lang_masks"
342
+ ],
343
+ "meta_keys": [
344
+ "info",
345
+ "stats"
346
+ ],
347
+ "type": "DictCollator"
348
+ },
349
+ "enable_gradient_checkpointing": false,
350
+ "enable_mixed_precision_training": true,
351
+ "grad_accumulation_steps": 1,
352
+ "lr_scheduler": {
353
+ "min_lr": 5e-07,
354
+ "type": "linear-warmup+cosine-decay",
355
+ "warmup_ratio": 0.05
356
+ },
357
+ "max_epochs": 3,
358
+ "max_grad_norm": 1.0,
359
+ "max_keep_ckpts": 3,
360
+ "max_steps": null,
361
+ "metric": {
362
+ "active_trackers": [
363
+ "jsonl",
364
+ "wandb"
365
+ ],
366
+ "grad_accumulation_steps": 1,
367
+ "run_dir": "work_dirs",
368
+ "type": "VLAMetric",
369
+ "window_size": 100
370
+ },
371
+ "mixed_precision_dtype": "bf16",
372
+ "optimizer": {
373
+ "betas": [
374
+ 0.9,
375
+ 0.95
376
+ ],
377
+ "eps": 1e-08,
378
+ "lr": 3e-05,
379
+ "paramwise_learning_rate": {
380
+ "vla_head": 0.0001,
381
+ "vlm_backbone.transformer": 1e-05
382
+ },
383
+ "type": "AdamW",
384
+ "weight_decay": 1e-08,
385
+ "weight_decay_all_params": true
386
+ },
387
+ "pre_fsdp_param_dtype": "fp32",
388
+ "reduce_in_full_precision": false,
389
+ "sampler": null,
390
+ "save_epoch_interval": 1,
391
+ "save_iter_interval": 10000000,
392
+ "seed": 42,
393
+ "sharding_strategy": "global-shard-grad-op",
394
+ "tokenizer": {
395
+ "model_max_length": 512,
396
+ "model_path": "./checkpoints/Cosmos-Predict2.5-2B/tokenizer",
397
+ "type": "PretrainedTokenizer"
398
+ },
399
+ "type": "FSDPTrainRunner"
400
+ },
401
+ "train_dataloader": {
402
+ "dataset": {
403
+ "auto_compute_statistics": {
404
+ "profile": "absolute"
405
+ },
406
+ "dataset_statistics_path": "/mnt/data/cpfs/mnt/data/william/robotwin/work_dirs/dit4dit_robotwin_horizion32_exec24/dataset_statistics.json",
407
+ "datasets": {
408
+ "action_key": "action",
409
+ "action_window_size": 32,
410
+ "data_root_path": [
411
+ "datasets/robotwin_clean_lerobotv2.1",
412
+ "datasets/robotwin_randomized_lerobotv2.1"
413
+ ],
414
+ "frame_sample_stride": 4,
415
+ "frame_window_size": 9,
416
+ "statistic_name": "robotwin_all",
417
+ "supervise_terminal_padding": true,
418
+ "transforms": [
419
+ {
420
+ "name_mappings": {
421
+ "actions": [
422
+ "actions"
423
+ ],
424
+ "observation.state": [
425
+ "states"
426
+ ]
427
+ },
428
+ "parquet_keys": [
429
+ "observation.state",
430
+ "timestamp",
431
+ "actions",
432
+ "info",
433
+ "stats",
434
+ "action_masks"
435
+ ],
436
+ "type": "ProcessParquetInputs",
437
+ "video_backend": "pyav",
438
+ "video_keys": [
439
+ "observation.images.cam_high",
440
+ "observation.images.cam_left_wrist",
441
+ "observation.images.cam_right_wrist"
442
+ ]
443
+ },
444
+ {
445
+ "input_key": "task_description",
446
+ "remove_input_key": true,
447
+ "tokenizer": {
448
+ "model_max_length": 512,
449
+ "model_path": "./checkpoints/Cosmos-Predict2.5-2B/tokenizer",
450
+ "type": "PretrainedTokenizer"
451
+ },
452
+ "type": "ProcessCosmos25Prompt"
453
+ },
454
+ {
455
+ "backend": "torch",
456
+ "height": 224,
457
+ "output_layout": "flattened_chw",
458
+ "scale_divisor": 255.0,
459
+ "type": "ResizeImages",
460
+ "width": 224
461
+ },
462
+ {
463
+ "means": [
464
+ 0.5,
465
+ 0.5,
466
+ 0.5
467
+ ],
468
+ "stds": [
469
+ 0.5,
470
+ 0.5,
471
+ 0.5
472
+ ],
473
+ "type": "NormalizeImages"
474
+ },
475
+ {
476
+ "bottom_height_ratio": 0.5,
477
+ "bottom_views": [
478
+ 1,
479
+ 2
480
+ ],
481
+ "combine_view_masks": true,
482
+ "frame_window_size": 9,
483
+ "num_views": 3,
484
+ "tile_direction": "top_bottom_pair",
485
+ "top_view": 0,
486
+ "type": "PrepareVideo"
487
+ },
488
+ {
489
+ "backend": "torch",
490
+ "expand_axis": 0,
491
+ "interleave": true,
492
+ "keys": [
493
+ "states"
494
+ ],
495
+ "target_dims": {
496
+ "states": 32
497
+ },
498
+ "type": "SinCosKeys"
499
+ },
500
+ {
501
+ "action_dim": 16,
502
+ "action_key": "action",
503
+ "norm_type": "min_max",
504
+ "normalization_epsilon": 1e-06,
505
+ "output_dtype": "float32",
506
+ "state_dim": null,
507
+ "state_key": "proprio",
508
+ "state_norm_type": "none",
509
+ "type": "NormalizeStatesAndActions"
510
+ },
511
+ {
512
+ "action_horizon": 32,
513
+ "state_history_length": 1,
514
+ "type": "PrepareStateActionTargets",
515
+ "valid_action_dim": 14
516
+ }
517
+ ],
518
+ "type": "ParquetDataset",
519
+ "use_delta": false,
520
+ "window_start_idx": 0
521
+ },
522
+ "name_mappings": {
523
+ "action": [
524
+ "action"
525
+ ],
526
+ "observation.state": [
527
+ "proprio"
528
+ ]
529
+ },
530
+ "reshuffle_each_epoch": true,
531
+ "sampling_unit": "episode",
532
+ "sampling_weights": [
533
+ 1.0,
534
+ 1.0
535
+ ],
536
+ "seed": 42,
537
+ "shuffle": true,
538
+ "statistic_keys": [
539
+ "observation.state",
540
+ "timestamp",
541
+ "action"
542
+ ],
543
+ "statistic_name": "robotwin_all",
544
+ "type": "DistributedBalancedRepeatingDataset"
545
+ },
546
+ "per_device_batch_size": 8,
547
+ "per_device_num_workers": 4
548
+ }
549
+ }
dit4dit_robotwin_all_data_full_finetune_bs256/config.yaml ADDED
@@ -0,0 +1,459 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ _DATA_ROOTS:
2
+ - datasets/robotwin_clean_lerobotv2.1
3
+ - datasets/robotwin_randomized_lerobotv2.1
4
+ _TASK_LIST:
5
+ - adjust_bottle
6
+ - beat_block_hammer
7
+ - blocks_ranking_rgb
8
+ - blocks_ranking_size
9
+ - click_alarmclock
10
+ - click_bell
11
+ - dump_bin_bigbin
12
+ - grab_roller
13
+ - handover_block
14
+ - handover_mic
15
+ - hanging_mug
16
+ - lift_pot
17
+ - move_can_pot
18
+ - move_pillbottle_pad
19
+ - move_playingcard_away
20
+ - move_stapler_pad
21
+ - open_laptop
22
+ - open_microwave
23
+ - pick_diverse_bottles
24
+ - pick_dual_bottles
25
+ - place_a2b_left
26
+ - place_a2b_right
27
+ - place_bread_basket
28
+ - place_bread_skillet
29
+ - place_burger_fries
30
+ - place_can_basket
31
+ - place_cans_plasticbox
32
+ - place_container_plate
33
+ - place_dual_shoes
34
+ - place_empty_cup
35
+ - place_fan
36
+ - place_mouse_pad
37
+ - place_object_basket
38
+ - place_object_scale
39
+ - place_object_stand
40
+ - place_phone_stand
41
+ - place_shoe
42
+ - press_stapler
43
+ - put_bottles_dustbin
44
+ - put_object_cabinet
45
+ - rotate_qrcode
46
+ - scan_object
47
+ - shake_bottle
48
+ - shake_bottle_horizontally
49
+ - stack_blocks_three
50
+ - stack_blocks_two
51
+ - stack_bowls_three
52
+ - stack_bowls_two
53
+ - stamp_seal
54
+ - turn_switch
55
+ _TOKENIZER_PATH: ./checkpoints/Cosmos-Predict2.5-2B/tokenizer
56
+ eval:
57
+ dataset:
58
+ img_keys:
59
+ - cam_high
60
+ - cam_left_wrist
61
+ - cam_right_wrist
62
+ inject_model_path: false
63
+ transforms:
64
+ - input_key: task_description
65
+ remove_input_key: true
66
+ tokenizer:
67
+ model_max_length: 512
68
+ model_path: ./checkpoints/Cosmos-Predict2.5-2B/tokenizer
69
+ type: PretrainedTokenizer
70
+ type: ProcessCosmos25Prompt
71
+ - backend: torch
72
+ height: 224
73
+ output_layout: flattened_chw
74
+ scale_divisor: 255.0
75
+ type: ResizeImages
76
+ width: 224
77
+ - means:
78
+ - 0.5
79
+ - 0.5
80
+ - 0.5
81
+ stds:
82
+ - 0.5
83
+ - 0.5
84
+ - 0.5
85
+ type: NormalizeImages
86
+ - bottom_height_ratio: 0.5
87
+ bottom_views:
88
+ - 1
89
+ - 2
90
+ combine_view_masks: true
91
+ frame_window_size: 1
92
+ num_views: 3
93
+ tile_direction: top_bottom_pair
94
+ top_view: 0
95
+ type: PrepareVideo
96
+ - backend: torch
97
+ expand_axis: 0
98
+ interleave: true
99
+ keys:
100
+ - states
101
+ target_dims:
102
+ states: 32
103
+ type: SinCosKeys
104
+ - action_dim: 16
105
+ action_key: action
106
+ norm_type: min_max
107
+ normalization_epsilon: 1.0e-06
108
+ output_dtype: float32
109
+ state_dim: null
110
+ state_key: proprio
111
+ state_norm_type: none
112
+ type: NormalizeStatesAndActions
113
+ - action_horizon: 32
114
+ state_history_length: 1
115
+ type: PrepareStateActionTargets
116
+ valid_action_dim: 14
117
+ type: PrivateInferenceDataset
118
+ denormalize_action:
119
+ action_dim: 14
120
+ norm_type: min_max
121
+ statistic_name: robotwin_all
122
+ type: DenormalizePrivateAction
123
+ eval_chunk_size: 24
124
+ instruction_type: unseen
125
+ mixed_precision_dtype: bf16
126
+ model_family: dit4dit
127
+ num_trials_per_task: 100
128
+ seed: 7
129
+ task_list:
130
+ - adjust_bottle
131
+ - beat_block_hammer
132
+ - blocks_ranking_rgb
133
+ - blocks_ranking_size
134
+ - click_alarmclock
135
+ - click_bell
136
+ - dump_bin_bigbin
137
+ - grab_roller
138
+ - handover_block
139
+ - handover_mic
140
+ - hanging_mug
141
+ - lift_pot
142
+ - move_can_pot
143
+ - move_pillbottle_pad
144
+ - move_playingcard_away
145
+ - move_stapler_pad
146
+ - open_laptop
147
+ - open_microwave
148
+ - pick_diverse_bottles
149
+ - pick_dual_bottles
150
+ - place_a2b_left
151
+ - place_a2b_right
152
+ - place_bread_basket
153
+ - place_bread_skillet
154
+ - place_burger_fries
155
+ - place_can_basket
156
+ - place_cans_plasticbox
157
+ - place_container_plate
158
+ - place_dual_shoes
159
+ - place_empty_cup
160
+ - place_fan
161
+ - place_mouse_pad
162
+ - place_object_basket
163
+ - place_object_scale
164
+ - place_object_stand
165
+ - place_phone_stand
166
+ - place_shoe
167
+ - press_stapler
168
+ - put_bottles_dustbin
169
+ - put_object_cabinet
170
+ - rotate_qrcode
171
+ - scan_object
172
+ - shake_bottle
173
+ - shake_bottle_horizontally
174
+ - stack_blocks_three
175
+ - stack_blocks_two
176
+ - stack_bowls_three
177
+ - stack_bowls_two
178
+ - stamp_seal
179
+ - turn_switch
180
+ task_suite_name:
181
+ - clean
182
+ - random
183
+ type: RobotwinEvalRunner
184
+ unnorm_key: robotwin_all
185
+ inference_model:
186
+ freeze_vlm_backbone: false
187
+ init_empty_weights: true
188
+ repeated_diffusion_steps: 4
189
+ type: DiT4DiTVLA
190
+ vla_head:
191
+ action_dim: 16
192
+ action_horizon: 32
193
+ action_model_type: DiT-B
194
+ add_pos_embed: true
195
+ diffusion_model_cfg:
196
+ cross_attention_dim: 2048
197
+ dropout: 0.2
198
+ final_dropout: true
199
+ interleave_self_attention: true
200
+ norm_type: ada_norm
201
+ num_layers: 16
202
+ output_dim: 2560
203
+ positional_embeddings: null
204
+ future_action_window_size: 31
205
+ hidden_size: 2560
206
+ max_seq_len: 1024
207
+ noise_beta_alpha: 1.5
208
+ noise_beta_beta: 1.0
209
+ noise_s: 0.999
210
+ num_inference_timesteps: 4
211
+ num_timestep_buckets: 1000
212
+ ori_action_dim: 14
213
+ state_dim: 32
214
+ type: DiT4DiTActionHead
215
+ vlm_backbone:
216
+ base_model: ./checkpoints/Cosmos-Predict2.5-2B
217
+ conditional_frame_timestep: 0.0001
218
+ detach_hidden_states: true
219
+ extract_layer: 17
220
+ fixed_seed: null
221
+ flow_matching_high_sigma_min: null
222
+ flow_matching_high_sigma_ratio: null
223
+ flow_matching_time_distribution: uniform
224
+ frozen_submodules:
225
+ - text_encoder
226
+ - vae
227
+ fsdp_min_num_params: 0
228
+ future_loss_type: flow_matching
229
+ load_pretrained_weights: false
230
+ local_files_only: true
231
+ num_frames_out: 9
232
+ num_inference_steps: 1
233
+ revision: diffusers/base/post-trained
234
+ split_future_frames: true
235
+ torch_dtype: bf16
236
+ trainable: true
237
+ type: Cosmos25Backbone
238
+ model:
239
+ freeze_vlm_backbone: false
240
+ repeated_diffusion_steps: 4
241
+ type: DiT4DiTVLA
242
+ vla_head:
243
+ action_dim: 16
244
+ action_horizon: 32
245
+ action_model_type: DiT-B
246
+ add_pos_embed: true
247
+ diffusion_model_cfg:
248
+ cross_attention_dim: 2048
249
+ dropout: 0.2
250
+ final_dropout: true
251
+ interleave_self_attention: true
252
+ norm_type: ada_norm
253
+ num_layers: 16
254
+ output_dim: 2560
255
+ positional_embeddings: null
256
+ future_action_window_size: 31
257
+ hidden_size: 2560
258
+ max_seq_len: 1024
259
+ noise_beta_alpha: 1.5
260
+ noise_beta_beta: 1.0
261
+ noise_s: 0.999
262
+ num_inference_timesteps: 4
263
+ num_timestep_buckets: 1000
264
+ ori_action_dim: 14
265
+ state_dim: 32
266
+ type: DiT4DiTActionHead
267
+ vlm_backbone:
268
+ base_model: ./checkpoints/Cosmos-Predict2.5-2B
269
+ conditional_frame_timestep: 0.0001
270
+ detach_hidden_states: true
271
+ extract_layer: 17
272
+ fixed_seed: null
273
+ flow_matching_high_sigma_min: null
274
+ flow_matching_high_sigma_ratio: null
275
+ flow_matching_time_distribution: uniform
276
+ frozen_submodules:
277
+ - text_encoder
278
+ - vae
279
+ fsdp_min_num_params: 0
280
+ future_loss_type: flow_matching
281
+ local_files_only: true
282
+ num_frames_out: 9
283
+ num_inference_steps: 1
284
+ revision: diffusers/base/post-trained
285
+ split_future_frames: true
286
+ torch_dtype: bf16
287
+ trainable: true
288
+ type: Cosmos25Backbone
289
+ runner:
290
+ change_key_name: false
291
+ collator:
292
+ keys:
293
+ - states
294
+ - timestamp
295
+ - images
296
+ - img_masks
297
+ - actions
298
+ - action_masks
299
+ - frame_masks
300
+ - lang_tokens
301
+ - lang_masks
302
+ meta_keys:
303
+ - info
304
+ - stats
305
+ type: DictCollator
306
+ enable_gradient_checkpointing: false
307
+ enable_mixed_precision_training: true
308
+ grad_accumulation_steps: 1
309
+ lr_scheduler:
310
+ min_lr: 5.0e-07
311
+ type: linear-warmup+cosine-decay
312
+ warmup_ratio: 0.05
313
+ max_epochs: 3
314
+ max_grad_norm: 1.0
315
+ max_keep_ckpts: 3
316
+ max_steps: null
317
+ metric:
318
+ active_trackers:
319
+ - jsonl
320
+ - wandb
321
+ grad_accumulation_steps: 1
322
+ run_dir: work_dirs
323
+ type: VLAMetric
324
+ window_size: 100
325
+ mixed_precision_dtype: bf16
326
+ optimizer:
327
+ betas:
328
+ - 0.9
329
+ - 0.95
330
+ eps: 1.0e-08
331
+ lr: 3.0e-05
332
+ paramwise_learning_rate:
333
+ vla_head: 0.0001
334
+ vlm_backbone.transformer: 1.0e-05
335
+ type: AdamW
336
+ weight_decay: 1.0e-08
337
+ weight_decay_all_params: true
338
+ pre_fsdp_param_dtype: fp32
339
+ reduce_in_full_precision: false
340
+ sampler: null
341
+ save_epoch_interval: 1
342
+ save_iter_interval: 10000000
343
+ seed: 42
344
+ sharding_strategy: global-shard-grad-op
345
+ tokenizer:
346
+ model_max_length: 512
347
+ model_path: ./checkpoints/Cosmos-Predict2.5-2B/tokenizer
348
+ type: PretrainedTokenizer
349
+ type: FSDPTrainRunner
350
+ train_dataloader:
351
+ dataset:
352
+ auto_compute_statistics:
353
+ profile: absolute
354
+ dataset_statistics_path: /mnt/data/cpfs/mnt/data/william/robotwin/work_dirs/dit4dit_robotwin_horizion32_exec24/dataset_statistics.json
355
+ datasets:
356
+ action_key: action
357
+ action_window_size: 32
358
+ data_root_path:
359
+ - datasets/robotwin_clean_lerobotv2.1
360
+ - datasets/robotwin_randomized_lerobotv2.1
361
+ frame_sample_stride: 4
362
+ frame_window_size: 9
363
+ statistic_name: robotwin_all
364
+ supervise_terminal_padding: true
365
+ transforms:
366
+ - name_mappings:
367
+ actions:
368
+ - actions
369
+ observation.state:
370
+ - states
371
+ parquet_keys:
372
+ - observation.state
373
+ - timestamp
374
+ - actions
375
+ - info
376
+ - stats
377
+ - action_masks
378
+ type: ProcessParquetInputs
379
+ video_backend: pyav
380
+ video_keys:
381
+ - observation.images.cam_high
382
+ - observation.images.cam_left_wrist
383
+ - observation.images.cam_right_wrist
384
+ - input_key: task_description
385
+ remove_input_key: true
386
+ tokenizer:
387
+ model_max_length: 512
388
+ model_path: ./checkpoints/Cosmos-Predict2.5-2B/tokenizer
389
+ type: PretrainedTokenizer
390
+ type: ProcessCosmos25Prompt
391
+ - backend: torch
392
+ height: 224
393
+ output_layout: flattened_chw
394
+ scale_divisor: 255.0
395
+ type: ResizeImages
396
+ width: 224
397
+ - means:
398
+ - 0.5
399
+ - 0.5
400
+ - 0.5
401
+ stds:
402
+ - 0.5
403
+ - 0.5
404
+ - 0.5
405
+ type: NormalizeImages
406
+ - bottom_height_ratio: 0.5
407
+ bottom_views:
408
+ - 1
409
+ - 2
410
+ combine_view_masks: true
411
+ frame_window_size: 9
412
+ num_views: 3
413
+ tile_direction: top_bottom_pair
414
+ top_view: 0
415
+ type: PrepareVideo
416
+ - backend: torch
417
+ expand_axis: 0
418
+ interleave: true
419
+ keys:
420
+ - states
421
+ target_dims:
422
+ states: 32
423
+ type: SinCosKeys
424
+ - action_dim: 16
425
+ action_key: action
426
+ norm_type: min_max
427
+ normalization_epsilon: 1.0e-06
428
+ output_dtype: float32
429
+ state_dim: null
430
+ state_key: proprio
431
+ state_norm_type: none
432
+ type: NormalizeStatesAndActions
433
+ - action_horizon: 32
434
+ state_history_length: 1
435
+ type: PrepareStateActionTargets
436
+ valid_action_dim: 14
437
+ type: ParquetDataset
438
+ use_delta: false
439
+ window_start_idx: 0
440
+ name_mappings:
441
+ action:
442
+ - action
443
+ observation.state:
444
+ - proprio
445
+ reshuffle_each_epoch: true
446
+ sampling_unit: episode
447
+ sampling_weights:
448
+ - 1.0
449
+ - 1.0
450
+ seed: 42
451
+ shuffle: true
452
+ statistic_keys:
453
+ - observation.state
454
+ - timestamp
455
+ - action
456
+ statistic_name: robotwin_all
457
+ type: DistributedBalancedRepeatingDataset
458
+ per_device_batch_size: 8
459
+ per_device_num_workers: 4
dit4dit_robotwin_all_data_full_finetune_bs256/dataset_statistics.json ADDED
@@ -0,0 +1,202 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "robotwin_all": {
3
+ "proprio": {
4
+ "mean": [
5
+ -0.23836578982314696,
6
+ 1.1298174457722125,
7
+ 0.8068365622397999,
8
+ -0.31024205976923314,
9
+ 0.05489046602657658,
10
+ -0.03797952133058626,
11
+ 0.6655638063536832,
12
+ 0.21664745657075993,
13
+ 1.0790455135726966,
14
+ 0.7875586714060262,
15
+ -0.3211318581147203,
16
+ -0.01750195667000284,
17
+ 0.03539713811747996,
18
+ 0.6794509068373863
19
+ ],
20
+ "std": [
21
+ 0.4151526770142024,
22
+ 1.004581002114008,
23
+ 0.7768177975025599,
24
+ 0.6732816994178228,
25
+ 0.27081867657629666,
26
+ 0.6197771066833538,
27
+ 0.454651637795559,
28
+ 0.34920797694011796,
29
+ 1.0249604576643263,
30
+ 0.7943722761565686,
31
+ 0.6847884512724388,
32
+ 0.24437404726809553,
33
+ 0.6142938396132972,
34
+ 0.44978408948808124
35
+ ],
36
+ "min": [
37
+ -7.552278995513916,
38
+ -0.36354875564575195,
39
+ -0.18577136099338531,
40
+ -1.956291913986206,
41
+ -1.6801013946533203,
42
+ -7.678869724273682,
43
+ 0.0,
44
+ -8.539926528930664,
45
+ -0.6294453740119934,
46
+ -0.07775841653347015,
47
+ -2.1328067779541016,
48
+ -2.1285502910614014,
49
+ -8.228925704956055,
50
+ 0.0
51
+ ],
52
+ "max": [
53
+ 6.45659065246582,
54
+ 4.179152488708496,
55
+ 5.346591472625732,
56
+ 1.7942548990249634,
57
+ 1.8604620695114136,
58
+ 5.43813943862915,
59
+ 1.0,
60
+ 7.142920017242432,
61
+ 4.157068729400635,
62
+ 5.672450065612793,
63
+ 1.95806884765625,
64
+ 1.5663840770721436,
65
+ 5.278968811035156,
66
+ 1.0
67
+ ],
68
+ "q01": [
69
+ -1.0126575303077698,
70
+ -5.257390398583084e-07,
71
+ -2.8215323254698887e-05,
72
+ -1.5635129809379578,
73
+ -0.6461568164825439,
74
+ -2.2244577407836914,
75
+ 0.0,
76
+ -0.1690114441514015,
77
+ 0.0,
78
+ -2.81171942333458e-05,
79
+ -1.6413878035545348,
80
+ -1.0220032143592834,
81
+ -1.6422083234786986,
82
+ 0.0
83
+ ],
84
+ "q99": [
85
+ 0.17374197930097574,
86
+ 2.602391481399536,
87
+ 2.4499925279617276,
88
+ 1.351163625717163,
89
+ 1.2348779439926147,
90
+ 1.6033015251159668,
91
+ 1.0,
92
+ 0.9967232346534729,
93
+ 2.6186706876754755,
94
+ 2.4520904397964403,
95
+ 1.291067123413086,
96
+ 0.8747973608970588,
97
+ 2.2670195341110118,
98
+ 1.0
99
+ ],
100
+ "count": 6075103
101
+ },
102
+ "action": {
103
+ "mean": [
104
+ -0.2520778376296883,
105
+ 1.1850720454152766,
106
+ 0.8463635513872634,
107
+ -0.32170768938995004,
108
+ 0.05836593476707814,
109
+ -0.047009579259007395,
110
+ 0.6459757061337159,
111
+ 0.23613831894636023,
112
+ 1.1499042732131481,
113
+ 0.8381827384224907,
114
+ -0.33540205675934687,
115
+ -0.01963055875970896,
116
+ 0.04568921604517659,
117
+ 0.6622663938976167
118
+ ],
119
+ "std": [
120
+ 0.4248377447940459,
121
+ 1.0055959718792558,
122
+ 0.7843761483314764,
123
+ 0.695804624051617,
124
+ 0.2812322261529959,
125
+ 0.6501131906725712,
126
+ 0.46179677530217805,
127
+ 0.3599664397795572,
128
+ 1.0307301365174235,
129
+ 0.8047042310821478,
130
+ 0.7142964613182307,
131
+ 0.25882870106807826,
132
+ 0.6546476377063773,
133
+ 0.45675020849497266
134
+ ],
135
+ "min": [
136
+ -7.552278995513916,
137
+ -0.36354875564575195,
138
+ -0.18577136099338531,
139
+ -1.956291913986206,
140
+ -1.6801013946533203,
141
+ -7.678869724273682,
142
+ 0.0,
143
+ -8.539926528930664,
144
+ -0.6294453740119934,
145
+ -0.07775841653347015,
146
+ -2.1328067779541016,
147
+ -2.1285502910614014,
148
+ -8.228925704956055,
149
+ 0.0
150
+ ],
151
+ "max": [
152
+ 6.45659065246582,
153
+ 4.179152488708496,
154
+ 5.346591472625732,
155
+ 1.7942548990249634,
156
+ 1.8604620695114136,
157
+ 5.43813943862915,
158
+ 1.0,
159
+ 7.142920017242432,
160
+ 4.157068729400635,
161
+ 5.672450065612793,
162
+ 1.95806884765625,
163
+ 1.5663840770721436,
164
+ 5.278968811035156,
165
+ 1.0
166
+ ],
167
+ "q01": [
168
+ -1.026803731918335,
169
+ -5.257390398583084e-07,
170
+ -2.8215323254698887e-05,
171
+ -1.5867787599563599,
172
+ -0.6817835569381714,
173
+ -2.2976155281066895,
174
+ 0.0,
175
+ -0.1809745579957962,
176
+ 0.0,
177
+ -2.81171942333458e-05,
178
+ -1.6533960103988647,
179
+ -1.0314260721206665,
180
+ -1.7116937637329102,
181
+ 0.0
182
+ ],
183
+ "q99": [
184
+ 0.20549810975792404,
185
+ 2.607667922973633,
186
+ 2.469137668609619,
187
+ 1.3703088760375977,
188
+ 1.2604563236236572,
189
+ 1.6216325759887695,
190
+ 1.0,
191
+ 1.0141181945800781,
192
+ 2.646111249923706,
193
+ 2.48582124710083,
194
+ 1.3271214962005615,
195
+ 0.9085489511489868,
196
+ 2.3036928176879883,
197
+ 1.0
198
+ ],
199
+ "count": 194403296
200
+ }
201
+ }
202
+ }
dit4dit_robotwin_all_data_full_finetune_bs256/tokenizer/added_tokens.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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_robotwin_all_data_full_finetune_bs256/tokenizer/chat_template.jinja ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
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_robotwin_all_data_full_finetune_bs256/tokenizer/merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
dit4dit_robotwin_all_data_full_finetune_bs256/tokenizer/special_tokens_map.json ADDED
@@ -0,0 +1,31 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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_robotwin_all_data_full_finetune_bs256/tokenizer/tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9c5ae00e602b8860cbd784ba82a8aa14e8feecec692e7076590d014d7b7fdafa
3
+ size 11421896
dit4dit_robotwin_all_data_full_finetune_bs256/tokenizer/tokenizer_config.json ADDED
@@ -0,0 +1,207 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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_robotwin_all_data_full_finetune_bs256/tokenizer/vocab.json ADDED
The diff for this file is too large to render. See raw diff