Hugging Face's logo

ShuaiYang03
/
GR00T-N1.5-Lerobot-SimplerEnv-Fractal

Safetensors
gr00t_n1_5
Model card Files
xet
Community
2
ShuaiYang03 commited on
Commit
63c1948
·
verified ·
1 Parent(s): e4cd682

Upload folder using huggingface_hub

Browse files
Files changed (26) hide show
  1. checkpoint-30000/config.json +64 -0
  2. checkpoint-30000/experiment_cfg/metadata.json +447 -0
  3. checkpoint-30000/model-00001-of-00002.safetensors +3 -0
  4. checkpoint-30000/model-00002-of-00002.safetensors +3 -0
  5. checkpoint-30000/model.safetensors.index.json +0 -0
  6. checkpoint-30000/optimizer.pt +3 -0
  7. checkpoint-30000/rng_state_0.pth +3 -0
  8. checkpoint-30000/rng_state_1.pth +3 -0
  9. checkpoint-30000/rng_state_2.pth +3 -0
  10. checkpoint-30000/rng_state_3.pth +3 -0
  11. checkpoint-30000/rng_state_4.pth +3 -0
  12. checkpoint-30000/rng_state_5.pth +3 -0
  13. checkpoint-30000/rng_state_6.pth +3 -0
  14. checkpoint-30000/rng_state_7.pth +3 -0
  15. checkpoint-30000/scheduler.pt +3 -0
  16. checkpoint-30000/trainer_state.json +0 -0
  17. config.json +64 -0
  18. experiment_cfg/metadata.json +447 -0
  19. model-00001-of-00002.safetensors +3 -0
  20. model-00002-of-00002.safetensors +3 -0
  21. model.safetensors.index.json +0 -0
  22. results.csv +2 -0
  23. results.md +29 -0
  24. total.metrics +230 -0
  25. trainer_state.json +0 -0
  26. training_args.bin +3 -0
checkpoint-30000/config.json ADDED
@@ -0,0 +1,64 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "action_dim": 32,
3
+ "action_head_cfg": {
4
+ "action_dim": 32,
5
+ "action_horizon": 16,
6
+ "add_pos_embed": true,
7
+ "backbone_embedding_dim": 2048,
8
+ "diffusion_model_cfg": {
9
+ "attention_head_dim": 48,
10
+ "cross_attention_dim": 2048,
11
+ "dropout": 0.2,
12
+ "final_dropout": true,
13
+ "interleave_self_attention": true,
14
+ "norm_type": "ada_norm",
15
+ "num_attention_heads": 32,
16
+ "num_layers": 16,
17
+ "output_dim": 1024,
18
+ "positional_embeddings": null
19
+ },
20
+ "hidden_size": 1024,
21
+ "input_embedding_dim": 1536,
22
+ "max_action_dim": 32,
23
+ "max_state_dim": 64,
24
+ "model_dtype": "float32",
25
+ "noise_beta_alpha": 1.5,
26
+ "noise_beta_beta": 1.0,
27
+ "noise_s": 0.999,
28
+ "num_inference_timesteps": 4,
29
+ "num_target_vision_tokens": 32,
30
+ "num_timestep_buckets": 1000,
31
+ "tune_diffusion_model": true,
32
+ "tune_projector": true,
33
+ "use_vlln": true,
34
+ "vl_self_attention_cfg": {
35
+ "attention_head_dim": 64,
36
+ "dropout": 0.2,
37
+ "final_dropout": true,
38
+ "num_attention_heads": 32,
39
+ "num_layers": 4,
40
+ "positional_embeddings": null
41
+ }
42
+ },
43
+ "action_horizon": 16,
44
+ "architectures": [
45
+ "GR00T_N1_5"
46
+ ],
47
+ "attn_implementation": null,
48
+ "backbone_cfg": {
49
+ "eagle_path": "NVEagle/eagle_er-qwen3_1_7B-Siglip2_400M_stage1_5_128gpu_er_v7_1mlp_nops",
50
+ "load_bf16": false,
51
+ "project_to_dim": null,
52
+ "reproject_vision": false,
53
+ "select_layer": 12,
54
+ "tune_llm": false,
55
+ "tune_visual": true,
56
+ "use_flash_attention": true
57
+ },
58
+ "compute_dtype": "bfloat16",
59
+ "hidden_size": 2048,
60
+ "model_dtype": "float32",
61
+ "model_type": "gr00t_n1_5",
62
+ "torch_dtype": "bfloat16",
63
+ "transformers_version": "4.51.3"
64
+ }
checkpoint-30000/experiment_cfg/metadata.json ADDED
@@ -0,0 +1,447 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "oxe": {
3
+ "statistics": {
4
+ "state": {
5
+ "x": {
6
+ "max": [
7
+ 1.0534898042678833
8
+ ],
9
+ "min": [
10
+ -0.4436439275741577
11
+ ],
12
+ "mean": [
13
+ 0.5598947405815125
14
+ ],
15
+ "std": [
16
+ 0.12432649731636047
17
+ ],
18
+ "q01": [
19
+ 0.32481380939483645
20
+ ],
21
+ "q99": [
22
+ 0.8750156319141384
23
+ ]
24
+ },
25
+ "y": {
26
+ "max": [
27
+ 0.48018959164619446
28
+ ],
29
+ "min": [
30
+ -0.9970501065254211
31
+ ],
32
+ "mean": [
33
+ -0.08334138244390488
34
+ ],
35
+ "std": [
36
+ 0.11558396369218826
37
+ ],
38
+ "q01": [
39
+ -0.28334290891885755
40
+ ],
41
+ "q99": [
42
+ 0.21247054174542404
43
+ ]
44
+ },
45
+ "z": {
46
+ "max": [
47
+ 1.6896663904190063
48
+ ],
49
+ "min": [
50
+ -0.006579156965017319
51
+ ],
52
+ "mean": [
53
+ 0.7770950198173523
54
+ ],
55
+ "std": [
56
+ 0.24595585465431213
57
+ ],
58
+ "q01": [
59
+ 0.14107070609927178
60
+ ],
61
+ "q99": [
62
+ 1.0727112340927123
63
+ ]
64
+ },
65
+ "rx": {
66
+ "max": [
67
+ 0.9999993443489075
68
+ ],
69
+ "min": [
70
+ -0.8643477559089661
71
+ ],
72
+ "mean": [
73
+ -0.24804554879665375
74
+ ],
75
+ "std": [
76
+ 0.5126774907112122
77
+ ],
78
+ "q01": [
79
+ -0.686474204659462
80
+ ],
81
+ "q99": [
82
+ 0.9377871316671368
83
+ ]
84
+ },
85
+ "ry": {
86
+ "max": [
87
+ 0.9999874830245972
88
+ ],
89
+ "min": [
90
+ -0.7079970240592957
91
+ ],
92
+ "mean": [
93
+ 0.4951382279396057
94
+ ],
95
+ "std": [
96
+ 0.5218117237091064
97
+ ],
98
+ "q01": [
99
+ -0.6808923494815826
100
+ ],
101
+ "q99": [
102
+ 0.9563051050901409
103
+ ]
104
+ },
105
+ "rz": {
106
+ "max": [
107
+ 0.9554369449615479
108
+ ],
109
+ "min": [
110
+ -0.7688722014427185
111
+ ],
112
+ "mean": [
113
+ 0.09266126900911331
114
+ ],
115
+ "std": [
116
+ 0.16630452871322632
117
+ ],
118
+ "q01": [
119
+ -0.36045596331357954
120
+ ],
121
+ "q99": [
122
+ 0.45990042358636823
123
+ ]
124
+ },
125
+ "rw": {
126
+ "max": [
127
+ 0.9914546012878418
128
+ ],
129
+ "min": [
130
+ -0.4999994933605194
131
+ ],
132
+ "mean": [
133
+ 0.2097489982843399
134
+ ],
135
+ "std": [
136
+ 0.2754714787006378
137
+ ],
138
+ "q01": [
139
+ -0.454380963742733
140
+ ],
141
+ "q99": [
142
+ 0.7216041100025177
143
+ ]
144
+ },
145
+ "gripper": {
146
+ "max": [
147
+ 1.0
148
+ ],
149
+ "min": [
150
+ 0.0
151
+ ],
152
+ "mean": [
153
+ 0.4261345863342285
154
+ ],
155
+ "std": [
156
+ 0.4554642140865326
157
+ ],
158
+ "q01": [
159
+ 0.0
160
+ ],
161
+ "q99": [
162
+ 1.0
163
+ ]
164
+ }
165
+ },
166
+ "action": {
167
+ "x": {
168
+ "max": [
169
+ 2.9984593391418457
170
+ ],
171
+ "min": [
172
+ -2.0204520225524902
173
+ ],
174
+ "mean": [
175
+ 0.00698750140145421
176
+ ],
177
+ "std": [
178
+ 0.06921227276325226
179
+ ],
180
+ "q01": [
181
+ -0.22453527510166169
182
+ ],
183
+ "q99": [
184
+ 0.17824687153100965
185
+ ]
186
+ },
187
+ "y": {
188
+ "max": [
189
+ 22.09052848815918
190
+ ],
191
+ "min": [
192
+ -5.497899532318115
193
+ ],
194
+ "mean": [
195
+ 0.006265923380851746
196
+ ],
197
+ "std": [
198
+ 0.059654854238033295
199
+ ],
200
+ "q01": [
201
+ -0.14820013284683228
202
+ ],
203
+ "q99": [
204
+ 0.14938379630446405
205
+ ]
206
+ },
207
+ "z": {
208
+ "max": [
209
+ 2.7507524490356445
210
+ ],
211
+ "min": [
212
+ -2.031663417816162
213
+ ],
214
+ "mean": [
215
+ -0.012625134550035
216
+ ],
217
+ "std": [
218
+ 0.07353131473064423
219
+ ],
220
+ "q01": [
221
+ -0.231589707583189
222
+ ],
223
+ "q99": [
224
+ 0.21842354819178575
225
+ ]
226
+ },
227
+ "roll": {
228
+ "max": [
229
+ 1.570636510848999
230
+ ],
231
+ "min": [
232
+ -1.569917917251587
233
+ ],
234
+ "mean": [
235
+ 0.04333191365003586
236
+ ],
237
+ "std": [
238
+ 0.1561005562543869
239
+ ],
240
+ "q01": [
241
+ -0.3517994859814644
242
+ ],
243
+ "q99": [
244
+ 0.5892666035890578
245
+ ]
246
+ },
247
+ "pitch": {
248
+ "max": [
249
+ 1.5321086645126343
250
+ ],
251
+ "min": [
252
+ -1.569892168045044
253
+ ],
254
+ "mean": [
255
+ -0.005756167229264975
256
+ ],
257
+ "std": [
258
+ 0.13164213299751282
259
+ ],
260
+ "q01": [
261
+ -0.4193011274933815
262
+ ],
263
+ "q99": [
264
+ 0.35272657424211445
265
+ ]
266
+ },
267
+ "yaw": {
268
+ "max": [
269
+ 1.5691522359848022
270
+ ],
271
+ "min": [
272
+ -1.570419430732727
273
+ ],
274
+ "mean": [
275
+ 0.0009130979306064546
276
+ ],
277
+ "std": [
278
+ 0.1459326595067978
279
+ ],
280
+ "q01": [
281
+ -0.43643461108207704
282
+ ],
283
+ "q99": [
284
+ 0.44796681255102094
285
+ ]
286
+ },
287
+ "gripper": {
288
+ "max": [
289
+ 1.0
290
+ ],
291
+ "min": [
292
+ 0.0
293
+ ],
294
+ "mean": [
295
+ 0.5354204773902893
296
+ ],
297
+ "std": [
298
+ 0.4971078634262085
299
+ ],
300
+ "q01": [
301
+ 0.0
302
+ ],
303
+ "q99": [
304
+ 1.0
305
+ ]
306
+ }
307
+ }
308
+ },
309
+ "modalities": {
310
+ "video": {
311
+ "image": {
312
+ "resolution": [
313
+ 320,
314
+ 256
315
+ ],
316
+ "channels": 3,
317
+ "fps": 3.0
318
+ }
319
+ },
320
+ "state": {
321
+ "x": {
322
+ "absolute": true,
323
+ "rotation_type": null,
324
+ "shape": [
325
+ 1
326
+ ],
327
+ "continuous": true
328
+ },
329
+ "y": {
330
+ "absolute": true,
331
+ "rotation_type": null,
332
+ "shape": [
333
+ 1
334
+ ],
335
+ "continuous": true
336
+ },
337
+ "z": {
338
+ "absolute": true,
339
+ "rotation_type": null,
340
+ "shape": [
341
+ 1
342
+ ],
343
+ "continuous": true
344
+ },
345
+ "rx": {
346
+ "absolute": true,
347
+ "rotation_type": null,
348
+ "shape": [
349
+ 1
350
+ ],
351
+ "continuous": true
352
+ },
353
+ "ry": {
354
+ "absolute": true,
355
+ "rotation_type": null,
356
+ "shape": [
357
+ 1
358
+ ],
359
+ "continuous": true
360
+ },
361
+ "rz": {
362
+ "absolute": true,
363
+ "rotation_type": null,
364
+ "shape": [
365
+ 1
366
+ ],
367
+ "continuous": true
368
+ },
369
+ "rw": {
370
+ "absolute": true,
371
+ "rotation_type": null,
372
+ "shape": [
373
+ 1
374
+ ],
375
+ "continuous": true
376
+ },
377
+ "gripper": {
378
+ "absolute": true,
379
+ "rotation_type": null,
380
+ "shape": [
381
+ 1
382
+ ],
383
+ "continuous": true
384
+ }
385
+ },
386
+ "action": {
387
+ "x": {
388
+ "absolute": true,
389
+ "rotation_type": null,
390
+ "shape": [
391
+ 1
392
+ ],
393
+ "continuous": true
394
+ },
395
+ "y": {
396
+ "absolute": true,
397
+ "rotation_type": null,
398
+ "shape": [
399
+ 1
400
+ ],
401
+ "continuous": true
402
+ },
403
+ "z": {
404
+ "absolute": true,
405
+ "rotation_type": null,
406
+ "shape": [
407
+ 1
408
+ ],
409
+ "continuous": true
410
+ },
411
+ "roll": {
412
+ "absolute": true,
413
+ "rotation_type": null,
414
+ "shape": [
415
+ 1
416
+ ],
417
+ "continuous": true
418
+ },
419
+ "pitch": {
420
+ "absolute": true,
421
+ "rotation_type": null,
422
+ "shape": [
423
+ 1
424
+ ],
425
+ "continuous": true
426
+ },
427
+ "yaw": {
428
+ "absolute": true,
429
+ "rotation_type": null,
430
+ "shape": [
431
+ 1
432
+ ],
433
+ "continuous": true
434
+ },
435
+ "gripper": {
436
+ "absolute": true,
437
+ "rotation_type": null,
438
+ "shape": [
439
+ 1
440
+ ],
441
+ "continuous": true
442
+ }
443
+ }
444
+ },
445
+ "embodiment_tag": "oxe"
446
+ }
447
+ }
checkpoint-30000/model-00001-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:26fe062d45063aca8af2a9065489b556d12447a5678fbb709477da9773573e1f
3
+ size 4999367032
checkpoint-30000/model-00002-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:98c6531d73fdf296abc46ea5b9b5e3ff839797a6e896f6f4d1663aae4e92520f
3
+ size 2586508600
checkpoint-30000/model.safetensors.index.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-30000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:03d07466c3cc0826a142ec542fb6a35ef345b91803d7369246e28685768dbc79
3
+ size 8550325978
checkpoint-30000/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c37ca53876599968a3004f603ba2faca9347a0627be93e4323aefe99dec05d6b
3
+ size 15984
checkpoint-30000/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5f34de092836aedd7bd6c444bf37502a196d1bbc81968a92c454464c1f9891fb
3
+ size 15984
checkpoint-30000/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cdcc4ede83f102615569832e99536a097f90fd15c551762fb3fff12b74267ed7
3
+ size 15984
checkpoint-30000/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d6cc08751e7421295803ac7559da75e09874760dddab8652232d5c730ddbc670
3
+ size 15984
checkpoint-30000/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e887e917e9c9f904cd79b4199c446127455a6329bb32fdf0e21374cf9878ffe1
3
+ size 15984
checkpoint-30000/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a2272c665a91578e0ed4a4f9afe0a9c38d380abe4c194368e52545c5ab5468be
3
+ size 15984
checkpoint-30000/rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:51825a147d0c52ef818cfee360b1a80a2c2d0a43d89d4a2fe5a85fb086db89b0
3
+ size 15984
checkpoint-30000/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:69623bd313d28ef4892d781d8040f9f9c3d159adc858bd82e4d4b99fdf1299a4
3
+ size 15984
checkpoint-30000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ac42b4efebda871e7035b931b3f1fd2282cf31a9d4249ad956a15227d00969cb
3
+ size 1064
checkpoint-30000/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
config.json ADDED
@@ -0,0 +1,64 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "action_dim": 32,
3
+ "action_head_cfg": {
4
+ "action_dim": 32,
5
+ "action_horizon": 16,
6
+ "add_pos_embed": true,
7
+ "backbone_embedding_dim": 2048,
8
+ "diffusion_model_cfg": {
9
+ "attention_head_dim": 48,
10
+ "cross_attention_dim": 2048,
11
+ "dropout": 0.2,
12
+ "final_dropout": true,
13
+ "interleave_self_attention": true,
14
+ "norm_type": "ada_norm",
15
+ "num_attention_heads": 32,
16
+ "num_layers": 16,
17
+ "output_dim": 1024,
18
+ "positional_embeddings": null
19
+ },
20
+ "hidden_size": 1024,
21
+ "input_embedding_dim": 1536,
22
+ "max_action_dim": 32,
23
+ "max_state_dim": 64,
24
+ "model_dtype": "float32",
25
+ "noise_beta_alpha": 1.5,
26
+ "noise_beta_beta": 1.0,
27
+ "noise_s": 0.999,
28
+ "num_inference_timesteps": 4,
29
+ "num_target_vision_tokens": 32,
30
+ "num_timestep_buckets": 1000,
31
+ "tune_diffusion_model": true,
32
+ "tune_projector": true,
33
+ "use_vlln": true,
34
+ "vl_self_attention_cfg": {
35
+ "attention_head_dim": 64,
36
+ "dropout": 0.2,
37
+ "final_dropout": true,
38
+ "num_attention_heads": 32,
39
+ "num_layers": 4,
40
+ "positional_embeddings": null
41
+ }
42
+ },
43
+ "action_horizon": 16,
44
+ "architectures": [
45
+ "GR00T_N1_5"
46
+ ],
47
+ "attn_implementation": null,
48
+ "backbone_cfg": {
49
+ "eagle_path": "NVEagle/eagle_er-qwen3_1_7B-Siglip2_400M_stage1_5_128gpu_er_v7_1mlp_nops",
50
+ "load_bf16": false,
51
+ "project_to_dim": null,
52
+ "reproject_vision": false,
53
+ "select_layer": 12,
54
+ "tune_llm": false,
55
+ "tune_visual": true,
56
+ "use_flash_attention": true
57
+ },
58
+ "compute_dtype": "bfloat16",
59
+ "hidden_size": 2048,
60
+ "model_dtype": "float32",
61
+ "model_type": "gr00t_n1_5",
62
+ "torch_dtype": "bfloat16",
63
+ "transformers_version": "4.51.3"
64
+ }
experiment_cfg/metadata.json ADDED
@@ -0,0 +1,447 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "oxe": {
3
+ "statistics": {
4
+ "state": {
5
+ "x": {
6
+ "max": [
7
+ 1.0534898042678833
8
+ ],
9
+ "min": [
10
+ -0.4436439275741577
11
+ ],
12
+ "mean": [
13
+ 0.5598947405815125
14
+ ],
15
+ "std": [
16
+ 0.12432649731636047
17
+ ],
18
+ "q01": [
19
+ 0.32481380939483645
20
+ ],
21
+ "q99": [
22
+ 0.8750156319141384
23
+ ]
24
+ },
25
+ "y": {
26
+ "max": [
27
+ 0.48018959164619446
28
+ ],
29
+ "min": [
30
+ -0.9970501065254211
31
+ ],
32
+ "mean": [
33
+ -0.08334138244390488
34
+ ],
35
+ "std": [
36
+ 0.11558396369218826
37
+ ],
38
+ "q01": [
39
+ -0.28334290891885755
40
+ ],
41
+ "q99": [
42
+ 0.21247054174542404
43
+ ]
44
+ },
45
+ "z": {
46
+ "max": [
47
+ 1.6896663904190063
48
+ ],
49
+ "min": [
50
+ -0.006579156965017319
51
+ ],
52
+ "mean": [
53
+ 0.7770950198173523
54
+ ],
55
+ "std": [
56
+ 0.24595585465431213
57
+ ],
58
+ "q01": [
59
+ 0.14107070609927178
60
+ ],
61
+ "q99": [
62
+ 1.0727112340927123
63
+ ]
64
+ },
65
+ "rx": {
66
+ "max": [
67
+ 0.9999993443489075
68
+ ],
69
+ "min": [
70
+ -0.8643477559089661
71
+ ],
72
+ "mean": [
73
+ -0.24804554879665375
74
+ ],
75
+ "std": [
76
+ 0.5126774907112122
77
+ ],
78
+ "q01": [
79
+ -0.686474204659462
80
+ ],
81
+ "q99": [
82
+ 0.9377871316671368
83
+ ]
84
+ },
85
+ "ry": {
86
+ "max": [
87
+ 0.9999874830245972
88
+ ],
89
+ "min": [
90
+ -0.7079970240592957
91
+ ],
92
+ "mean": [
93
+ 0.4951382279396057
94
+ ],
95
+ "std": [
96
+ 0.5218117237091064
97
+ ],
98
+ "q01": [
99
+ -0.6808923494815826
100
+ ],
101
+ "q99": [
102
+ 0.9563051050901409
103
+ ]
104
+ },
105
+ "rz": {
106
+ "max": [
107
+ 0.9554369449615479
108
+ ],
109
+ "min": [
110
+ -0.7688722014427185
111
+ ],
112
+ "mean": [
113
+ 0.09266126900911331
114
+ ],
115
+ "std": [
116
+ 0.16630452871322632
117
+ ],
118
+ "q01": [
119
+ -0.36045596331357954
120
+ ],
121
+ "q99": [
122
+ 0.45990042358636823
123
+ ]
124
+ },
125
+ "rw": {
126
+ "max": [
127
+ 0.9914546012878418
128
+ ],
129
+ "min": [
130
+ -0.4999994933605194
131
+ ],
132
+ "mean": [
133
+ 0.2097489982843399
134
+ ],
135
+ "std": [
136
+ 0.2754714787006378
137
+ ],
138
+ "q01": [
139
+ -0.454380963742733
140
+ ],
141
+ "q99": [
142
+ 0.7216041100025177
143
+ ]
144
+ },
145
+ "gripper": {
146
+ "max": [
147
+ 1.0
148
+ ],
149
+ "min": [
150
+ 0.0
151
+ ],
152
+ "mean": [
153
+ 0.4261345863342285
154
+ ],
155
+ "std": [
156
+ 0.4554642140865326
157
+ ],
158
+ "q01": [
159
+ 0.0
160
+ ],
161
+ "q99": [
162
+ 1.0
163
+ ]
164
+ }
165
+ },
166
+ "action": {
167
+ "x": {
168
+ "max": [
169
+ 2.9984593391418457
170
+ ],
171
+ "min": [
172
+ -2.0204520225524902
173
+ ],
174
+ "mean": [
175
+ 0.00698750140145421
176
+ ],
177
+ "std": [
178
+ 0.06921227276325226
179
+ ],
180
+ "q01": [
181
+ -0.22453527510166169
182
+ ],
183
+ "q99": [
184
+ 0.17824687153100965
185
+ ]
186
+ },
187
+ "y": {
188
+ "max": [
189
+ 22.09052848815918
190
+ ],
191
+ "min": [
192
+ -5.497899532318115
193
+ ],
194
+ "mean": [
195
+ 0.006265923380851746
196
+ ],
197
+ "std": [
198
+ 0.059654854238033295
199
+ ],
200
+ "q01": [
201
+ -0.14820013284683228
202
+ ],
203
+ "q99": [
204
+ 0.14938379630446405
205
+ ]
206
+ },
207
+ "z": {
208
+ "max": [
209
+ 2.7507524490356445
210
+ ],
211
+ "min": [
212
+ -2.031663417816162
213
+ ],
214
+ "mean": [
215
+ -0.012625134550035
216
+ ],
217
+ "std": [
218
+ 0.07353131473064423
219
+ ],
220
+ "q01": [
221
+ -0.231589707583189
222
+ ],
223
+ "q99": [
224
+ 0.21842354819178575
225
+ ]
226
+ },
227
+ "roll": {
228
+ "max": [
229
+ 1.570636510848999
230
+ ],
231
+ "min": [
232
+ -1.569917917251587
233
+ ],
234
+ "mean": [
235
+ 0.04333191365003586
236
+ ],
237
+ "std": [
238
+ 0.1561005562543869
239
+ ],
240
+ "q01": [
241
+ -0.3517994859814644
242
+ ],
243
+ "q99": [
244
+ 0.5892666035890578
245
+ ]
246
+ },
247
+ "pitch": {
248
+ "max": [
249
+ 1.5321086645126343
250
+ ],
251
+ "min": [
252
+ -1.569892168045044
253
+ ],
254
+ "mean": [
255
+ -0.005756167229264975
256
+ ],
257
+ "std": [
258
+ 0.13164213299751282
259
+ ],
260
+ "q01": [
261
+ -0.4193011274933815
262
+ ],
263
+ "q99": [
264
+ 0.35272657424211445
265
+ ]
266
+ },
267
+ "yaw": {
268
+ "max": [
269
+ 1.5691522359848022
270
+ ],
271
+ "min": [
272
+ -1.570419430732727
273
+ ],
274
+ "mean": [
275
+ 0.0009130979306064546
276
+ ],
277
+ "std": [
278
+ 0.1459326595067978
279
+ ],
280
+ "q01": [
281
+ -0.43643461108207704
282
+ ],
283
+ "q99": [
284
+ 0.44796681255102094
285
+ ]
286
+ },
287
+ "gripper": {
288
+ "max": [
289
+ 1.0
290
+ ],
291
+ "min": [
292
+ 0.0
293
+ ],
294
+ "mean": [
295
+ 0.5354204773902893
296
+ ],
297
+ "std": [
298
+ 0.4971078634262085
299
+ ],
300
+ "q01": [
301
+ 0.0
302
+ ],
303
+ "q99": [
304
+ 1.0
305
+ ]
306
+ }
307
+ }
308
+ },
309
+ "modalities": {
310
+ "video": {
311
+ "image": {
312
+ "resolution": [
313
+ 320,
314
+ 256
315
+ ],
316
+ "channels": 3,
317
+ "fps": 3.0
318
+ }
319
+ },
320
+ "state": {
321
+ "x": {
322
+ "absolute": true,
323
+ "rotation_type": null,
324
+ "shape": [
325
+ 1
326
+ ],
327
+ "continuous": true
328
+ },
329
+ "y": {
330
+ "absolute": true,
331
+ "rotation_type": null,
332
+ "shape": [
333
+ 1
334
+ ],
335
+ "continuous": true
336
+ },
337
+ "z": {
338
+ "absolute": true,
339
+ "rotation_type": null,
340
+ "shape": [
341
+ 1
342
+ ],
343
+ "continuous": true
344
+ },
345
+ "rx": {
346
+ "absolute": true,
347
+ "rotation_type": null,
348
+ "shape": [
349
+ 1
350
+ ],
351
+ "continuous": true
352
+ },
353
+ "ry": {
354
+ "absolute": true,
355
+ "rotation_type": null,
356
+ "shape": [
357
+ 1
358
+ ],
359
+ "continuous": true
360
+ },
361
+ "rz": {
362
+ "absolute": true,
363
+ "rotation_type": null,
364
+ "shape": [
365
+ 1
366
+ ],
367
+ "continuous": true
368
+ },
369
+ "rw": {
370
+ "absolute": true,
371
+ "rotation_type": null,
372
+ "shape": [
373
+ 1
374
+ ],
375
+ "continuous": true
376
+ },
377
+ "gripper": {
378
+ "absolute": true,
379
+ "rotation_type": null,
380
+ "shape": [
381
+ 1
382
+ ],
383
+ "continuous": true
384
+ }
385
+ },
386
+ "action": {
387
+ "x": {
388
+ "absolute": true,
389
+ "rotation_type": null,
390
+ "shape": [
391
+ 1
392
+ ],
393
+ "continuous": true
394
+ },
395
+ "y": {
396
+ "absolute": true,
397
+ "rotation_type": null,
398
+ "shape": [
399
+ 1
400
+ ],
401
+ "continuous": true
402
+ },
403
+ "z": {
404
+ "absolute": true,
405
+ "rotation_type": null,
406
+ "shape": [
407
+ 1
408
+ ],
409
+ "continuous": true
410
+ },
411
+ "roll": {
412
+ "absolute": true,
413
+ "rotation_type": null,
414
+ "shape": [
415
+ 1
416
+ ],
417
+ "continuous": true
418
+ },
419
+ "pitch": {
420
+ "absolute": true,
421
+ "rotation_type": null,
422
+ "shape": [
423
+ 1
424
+ ],
425
+ "continuous": true
426
+ },
427
+ "yaw": {
428
+ "absolute": true,
429
+ "rotation_type": null,
430
+ "shape": [
431
+ 1
432
+ ],
433
+ "continuous": true
434
+ },
435
+ "gripper": {
436
+ "absolute": true,
437
+ "rotation_type": null,
438
+ "shape": [
439
+ 1
440
+ ],
441
+ "continuous": true
442
+ }
443
+ }
444
+ },
445
+ "embodiment_tag": "oxe"
446
+ }
447
+ }
model-00001-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:14e54be691d887991fd4d6d3780bc6df0247838fd44e385d15704802ef82596a
3
+ size 4999367032
model-00002-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0bdf6ae82a0d1585a06e5c56bc67298f608ecb6a8d0fba2f996d593d2f9ff28c
3
+ size 2586508600
model.safetensors.index.json ADDED
The diff for this file is too large to render. See raw diff
 
results.csv ADDED
@@ -0,0 +1,2 @@
 
 
 
1
+ coke_can/matching_avg,coke_can/variant_avg,coke_can/matching/horizontal,coke_can/matching/vertical,coke_can/matching/standing,coke_can/variant/horizontal,coke_can/variant/vertical,coke_can/variant/standing,move_near/variant,move_near/matching,drawer/matching_avg,drawer/variant_avg,drawer/matching/open,drawer/matching/close,drawer/variant/open,drawer/variant/close,put_spoon_on_tablecloth/matching_partial,put_spoon_on_tablecloth/matching_entire,put_carrot_on_plate/matching_partial,put_carrot_on_plate/matching_entire,stack_green_block_on_yellow_block/matching_partial,stack_green_block_on_yellow_block/matching_entire,put_eggplant_in_basket/matching_partial,put_eggplant_in_basket/matching_entire,apple_in_drawer/matching_avg,apple_in_drawer/variant_avg,ckpt_name
2
+ 0.517,0.636,0.470,0.130,0.950,0.711,0.324,0.871,0.510,0.540,0.278,0.132,0.269,0.287,0.085,0.180,,,,,,,,,0.074,0.023,fractal_60K0
results.md ADDED
@@ -0,0 +1,29 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ | | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 |
2
+ |:---------------------------------------------------|:---------------------|:----------------|:----------|:-------|:-------|:----------|:-----------|:------------|:--------|:--------|
3
+ | coke_can/matching_avg | 0.5166666666666666 | 0.857 | 0.71 | 0.567 | 0.787 | 0.17 | nan | 0.027 | 0.163 | 0.727 |
4
+ | coke_can/variant_avg | 0.6355555555555555 | 0.898 | 0.813 | 0.49 | 0.823 | 0.006 | nan | 0.022 | 0.545 | nan |
5
+ | coke_can/matching/horizontal | 0.47 | 0.96 | 0.86 | 0.82 | 0.74 | 0.21 | nan | 0.05 | 0.27 | 0.85 |
6
+ | coke_can/matching/vertical | 0.13 | 0.9 | 0.79 | 0.33 | 0.74 | 0.21 | nan | 0.0 | 0.03 | 0.43 |
7
+ | coke_can/matching/standing | 0.95 | 0.71 | 0.48 | 0.55 | 0.88 | 0.09 | nan | 0.03 | 0.19 | 0.9 |
8
+ | coke_can/variant/horizontal | 0.7111111111111111 | 0.969 | 0.92 | 0.569 | 0.822 | 0.005 | nan | 0.022 | 0.711 | nan |
9
+ | coke_can/variant/vertical | 0.3244444444444444 | 0.76 | 0.704 | 0.204 | 0.754 | 0.0 | nan | 0.013 | 0.271 | nan |
10
+ | coke_can/variant/standing | 0.8711111111111111 | 0.964 | 0.813 | 0.698 | 0.893 | 0.013 | nan | 0.031 | 0.653 | nan |
11
+ | move_near/variant | 0.51 | 0.5 | 0.446 | 0.323 | 0.792 | 0.031 | nan | 0.04 | 0.477 | nan |
12
+ | move_near/matching | 0.54 | 0.442 | 0.354 | 0.317 | 0.779 | 0.042 | nan | 0.05 | 0.462 | 0.663 |
13
+ | drawer/matching_avg | 0.2777777777777778 | 0.73 | 0.565 | 0.597 | 0.25 | 0.227 | nan | 0.139 | 0.356 | 0.268 |
14
+ | drawer/variant_avg | 0.13227513227513227 | 0.323 | 0.267 | 0.294 | 0.353 | 0.011 | nan | 0.069 | 0.177 | nan |
15
+ | drawer/matching/open | 0.26851851851851855 | 0.601 | 0.463 | 0.296 | 0.157 | 0.009 | nan | 0.0 | 0.194 | 0.287 |
16
+ | drawer/matching/close | 0.28703703703703703 | 0.861 | 0.667 | 0.891 | 0.343 | 0.444 | nan | 0.278 | 0.518 | 0.25 |
17
+ | drawer/variant/open | 0.08465608465608465 | 0.27 | 0.212 | 0.069 | 0.333 | 0.0 | nan | 0.005 | 0.158 | nan |
18
+ | drawer/variant/close | 0.1798941798941799 | 0.376 | 0.323 | 0.519 | 0.372 | 0.021 | nan | 0.132 | 0.195 | nan |
19
+ | put_spoon_on_tablecloth/matching_partial | nan | nan | nan | 0.167 | nan | 0.347 | 0.778 | nan | 0.041 | 0.375 |
20
+ | put_spoon_on_tablecloth/matching_entire | nan | nan | nan | 0.0 | nan | 0.125 | 0.472 | nan | 0.0 | 0.208 |
21
+ | put_carrot_on_plate/matching_partial | nan | nan | nan | 0.208 | nan | 0.528 | 0.278 | nan | 0.333 | 0.333 |
22
+ | put_carrot_on_plate/matching_entire | nan | nan | nan | 0.042 | nan | 0.083 | 0.097 | nan | 0.0 | 0.25 |
23
+ | stack_green_block_on_yellow_block/matching_partial | nan | nan | nan | 0.083 | nan | 0.319 | 0.403 | nan | 0.125 | 0.083 |
24
+ | stack_green_block_on_yellow_block/matching_entire | nan | nan | nan | 0.0 | nan | 0.0 | 0.042 | nan | 0.0 | 0.083 |
25
+ | put_eggplant_in_basket/matching_partial | nan | nan | nan | 0.0 | nan | 0.667 | 0.875 | nan | 0.083 | 0.0 |
26
+ | put_eggplant_in_basket/matching_entire | nan | nan | nan | 0.0 | nan | 0.431 | 0.569 | nan | 0.041 | 0.0 |
27
+ | apple_in_drawer/matching_avg | 0.07407407407407407 | 0.065 | 0.13 | 0.213 | 0.037 | 0.0 | 0.0 | 0.0 | nan | 0.361 |
28
+ | apple_in_drawer/variant_avg | 0.022857142857142857 | 0.026 | 0.021 | 0.101 | 0.206 | 0.0 | 0.0 | 0.0 | nan | nan |
29
+ | ckpt_name | ours | RT-1(Converged) | RT-1(15%) | RT-1-X | RT-2-X | Octo-Base | Octo-Small | RT-1(begin) | OpenVLA | RoboVLM |
total.metrics ADDED
@@ -0,0 +1,230 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ***Pick coke can results***
2
+ --------------------
3
+ horizontal sim variant avg success {'ours': 0.7111111111111111}
4
+ horizontal real success {'ours': 0.0}
5
+ horizontal MMRV 0.0
6
+ horizontal pearson correlation 1
7
+ vertical sim variant avg success {'ours': 0.3244444444444444}
8
+ vertical real success {'ours': 0.0}
9
+ vertical MMRV 0.0
10
+ vertical pearson correlation 1
11
+ standing sim variant avg success {'ours': 0.8711111111111111}
12
+ standing real success {'ours': 0.0}
13
+ standing MMRV 0.0
14
+ standing pearson correlation 1
15
+ avg_orientation_sim_variant_results [0.6355555555555555]
16
+ avg_orientation_real_results [0.0]
17
+ mean_maximum_rank_violation(avg_orientation_sim_variant_results, avg_orientation_real_results) 0.0
18
+ pearson_correlation(avg_orientation_sim_variant_results, avg_orientation_real_results) 1
19
+ --------------------
20
+ Orientation horizontal, ckpt ours all robot arm visual matching success: [0.52, 0.44, 0.4, 0.52]
21
+ Orientation vertical, ckpt ours all robot arm visual matching success: [0.12, 0.16, 0.16, 0.08]
22
+ Orientation standing, ckpt ours all robot arm visual matching success: [0.96, 0.96, 0.96, 0.92]
23
+ horizontal visual matching sim success {'ours': 0.47}
24
+ horizontal real success {'ours': 0.0}
25
+ horizontal MMRV 0.0
26
+ horizontal pearson correlation 1
27
+ horizontal kruskal:
28
+ each checkpoint kruskal:
29
+ KruskalResult(statistic=15.473684210526313, pvalue=8.36619614588189e-05)
30
+ vertical visual matching sim success {'ours': 0.13}
31
+ vertical real success {'ours': 0.0}
32
+ vertical MMRV 0.0
33
+ vertical pearson correlation 1
34
+ vertical kruskal:
35
+ each checkpoint kruskal:
36
+ KruskalResult(statistic=3.1276595744679274, pvalue=0.07697417298127443)
37
+ standing visual matching sim success {'ours': 0.95}
38
+ standing real success {'ours': 0.0}
39
+ standing MMRV 0.0
40
+ standing pearson correlation 1
41
+ standing kruskal:
42
+ each checkpoint kruskal:
43
+ KruskalResult(statistic=45.23076923076922, pvalue=1.7513181713489203e-11)
44
+ avg_orientation_sim_visual_matching_results [0.5166666666666666]
45
+ avg_orientation_real_results [0.0]
46
+ mean_maximum_rank_violation(avg_orientation_sim_visual_matching_results, avg_orientation_real_results) 0.0
47
+ pearson_correlation(avg_orientation_sim_visual_matching_results, avg_orientation_real_results) 1
48
+ avg kruskal:
49
+ each checkpoint kruskal:
50
+ KruskalResult(statistic=52.35135135135139, pvalue=4.640791164425619e-13)
51
+ ********************
52
+
53
+
54
+
55
+ ***Move Near results***
56
+ --------------------
57
+ sim variant avg success {'ours': 0.51}
58
+ real success {'ours': 0.0}
59
+ MMRV 0.0
60
+ pearson correlation 1
61
+ --------------------
62
+ Ckpt ours all robot arm visual matching success: [0.48, 0.52, 0.52, 0.64]
63
+ sim visual matching success {'ours': 0.54}
64
+ real success {'ours': 0.0}
65
+ visual matching MMRV 0.0
66
+ visual matching pearson correlation 1
67
+ avg kruskal:
68
+ each checkpoint kruskal:
69
+ KruskalResult(statistic=43.27272727272723, pvalue=4.76174527067639e-11)
70
+ ********************
71
+
72
+
73
+
74
+ ***Drawer results***
75
+ --------------------
76
+ open sim variant avg success {'ours': 0.08465608465608465}
77
+ open real success {'ours': 0.0}
78
+ open MMRV 0.0
79
+ open pearson correlation 1
80
+ close sim variant avg success {'ours': 0.1798941798941799}
81
+ close real success {'ours': 0.0}
82
+ close MMRV 0.0
83
+ close pearson correlation 1
84
+ avg_sim_variant_results [0.13227513227513227]
85
+ avg_real_results [0.0]
86
+ mean_maximum_rank_violation(avg_sim_variant_results, avg_real_results) 0.0
87
+ pearson_correlation(avg_sim_variant_results, avg_real_results) 1
88
+ --------------------
89
+ Drawer task open, ckpt ours all robot arm visual matching success: [0.25, 0.25, 0.3055555555555555, 0.25925925925925924]
90
+ Drawer task close, ckpt ours all robot arm visual matching success: [0.2777777777777778, 0.3055555555555556, 0.2777777777777778, 0.25925925925925924]
91
+ open visual matching sim success {'ours': 0.26851851851851855}
92
+ open real success {'ours': 0.0}
93
+ open MMRV 0.0
94
+ open pearson correlation 1
95
+ open kruskal:
96
+ each checkpoint kruskal:
97
+ KruskalResult(statistic=7.8936170212766354, pvalue=0.004960955562494715)
98
+ close visual matching sim success {'ours': 0.28703703703703703}
99
+ close real success {'ours': 0.0}
100
+ close MMRV 0.0
101
+ close pearson correlation 1
102
+ close kruskal:
103
+ each checkpoint kruskal:
104
+ KruskalResult(statistic=9.217391304347847, pvalue=0.0023972688480227177)
105
+ avg_sim_visual_matching_results [0.2777777777777778]
106
+ avg_real_results [0.0]
107
+ mean_maximum_rank_violation(avg_sim_visual_matching_results, avg_real_results) 0.0
108
+ pearson_correlation(avg_sim_visual_matching_results, avg_real_results) 1
109
+ avg kruskal:
110
+ each checkpoint kruskal:
111
+ KruskalResult(statistic=17.258064516129142, pvalue=3.263095871941237e-05)
112
+ ********************
113
+
114
+
115
+
116
+ ***Drawer results***
117
+ --------------------
118
+ put_apple_into_top_drawer sim variant avg success {'ours': 0.022857142857142857}
119
+ put_apple_into_top_drawer real success {'ours': 0.0}
120
+ put_apple_into_top_drawer MMRV 0.0
121
+ put_apple_into_top_drawer pearson correlation 1
122
+ avg_sim_variant_results [0.022857142857142857]
123
+ avg_real_results [0.0]
124
+ mean_maximum_rank_violation(avg_sim_variant_results, avg_real_results) 0.0
125
+ pearson_correlation(avg_sim_variant_results, avg_real_results) 1
126
+ --------------------
127
+ Drawer task put_apple_into_top_drawer, ckpt ours all robot arm visual matching success: [0.07407407407407407, 0.04938271604938271, 0.037037037037037035, 0.0]
128
+ put_apple_into_top_drawer visual matching sim success {'ours': 0.07407407407407407}
129
+ put_apple_into_top_drawer real success {'ours': 0.0}
130
+ put_apple_into_top_drawer MMRV 0.0
131
+ put_apple_into_top_drawer pearson correlation 1
132
+ put_apple_into_top_drawer kruskal:
133
+ each checkpoint kruskal:
134
+ KruskalResult(statistic=2.038461538461544, pvalue=0.15336468852414337)
135
+ avg_sim_visual_matching_results [0.07407407407407407]
136
+ avg_real_results [0.0]
137
+ mean_maximum_rank_violation(avg_sim_visual_matching_results, avg_real_results) 0.0
138
+ pearson_correlation(avg_sim_visual_matching_results, avg_real_results) 1
139
+ avg kruskal:
140
+ each checkpoint kruskal:
141
+ KruskalResult(statistic=2.038461538461544, pvalue=0.15336468852414337)
142
+ ********************
143
+
144
+
145
+
146
+ ***Bridge Put On Env results***
147
+ ********** Results for put_spoon_on_tablecloth **********
148
+ WARNING: avg_sim_success is nan for results/fractal_60K0/bridge_table_1_v1/arm_pd_ee_target_delta_pose_align2_gripper_pd_joint_pos/PutSpoonOnTableClothInScene-v0
149
+ sim visual matching partial success {'ours': nan}
150
+ real partial success {'ours': 0.0}
151
+ visual matching MMRV (partial success) 0.0
152
+ visual matching pearson correlation (partial success) nan
153
+ avg kruskal (partial success):
154
+ each checkpoint kruskal:
155
+ all same, 1.0
156
+ sim visual matching success {'ours': nan}
157
+ real success {'ours': 0.0}
158
+ visual matching MMRV 0.0
159
+ visual matching pearson correlation nan
160
+ avg kruskal:
161
+ each checkpoint kruskal:
162
+ all same, 1.0
163
+ ********************
164
+
165
+
166
+
167
+ ********** Results for put_carrot_on_plate **********
168
+ WARNING: avg_sim_success is nan for results/fractal_60K0/bridge_table_1_v1/arm_pd_ee_target_delta_pose_align2_gripper_pd_joint_pos/PutCarrotOnPlateInScene-v0
169
+ sim visual matching partial success {'ours': nan}
170
+ real partial success {'ours': 0.0}
171
+ visual matching MMRV (partial success) 0.0
172
+ visual matching pearson correlation (partial success) nan
173
+ avg kruskal (partial success):
174
+ each checkpoint kruskal:
175
+ all same, 1.0
176
+ sim visual matching success {'ours': nan}
177
+ real success {'ours': 0.0}
178
+ visual matching MMRV 0.0
179
+ visual matching pearson correlation nan
180
+ avg kruskal:
181
+ each checkpoint kruskal:
182
+ all same, 1.0
183
+ ********************
184
+
185
+
186
+
187
+ ********** Results for stack_green_block_on_yellow_block **********
188
+ WARNING: avg_sim_success is nan for results/fractal_60K0/bridge_table_1_v1/arm_pd_ee_target_delta_pose_align2_gripper_pd_joint_pos/StackGreenCubeOnYellowCubeBakedTexInScene-v0
189
+ sim visual matching partial success {'ours': nan}
190
+ real partial success {'ours': 0.0}
191
+ visual matching MMRV (partial success) 0.0
192
+ visual matching pearson correlation (partial success) nan
193
+ avg kruskal (partial success):
194
+ each checkpoint kruskal:
195
+ all same, 1.0
196
+ sim visual matching success {'ours': nan}
197
+ real success {'ours': 0.0}
198
+ visual matching MMRV 0.0
199
+ visual matching pearson correlation nan
200
+ avg kruskal:
201
+ each checkpoint kruskal:
202
+ all same, 1.0
203
+ ********************
204
+
205
+
206
+
207
+ ********** Results for put_eggplant_in_basket **********
208
+ WARNING: avg_sim_success is nan for results/fractal_60K0/bridge_table_1_v2/arm_pd_ee_target_delta_pose_align2_gripper_pd_joint_pos/PutEggplantInBasketScene-v0
209
+ sim visual matching partial success {'ours': nan}
210
+ real partial success {'ours': 0.0}
211
+ visual matching MMRV (partial success) 0.0
212
+ visual matching pearson correlation (partial success) nan
213
+ avg kruskal (partial success):
214
+ each checkpoint kruskal:
215
+ all same, 1.0
216
+ sim visual matching success {'ours': nan}
217
+ real success {'ours': 0.0}
218
+ visual matching MMRV 0.0
219
+ visual matching pearson correlation nan
220
+ avg kruskal:
221
+ each checkpoint kruskal:
222
+ all same, 1.0
223
+ ********************
224
+
225
+
226
+
227
+ coke_can/matching_avg coke_can/variant_avg ... apple_in_drawer/variant_avg ckpt_name
228
+ 0 0.516667 0.635556 ... 0.022857 fractal_60K0
229
+
230
+ [1 rows x 27 columns]
trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7a753302bbcc9e7ffb9a6e28db12c62e43af074904fffb7c772eec27696c62f0
3
+ size 5368