11yiming commited on
Commit
05b940c
·
verified ·
1 Parent(s): c9937dd

Upload GR00T N1.7 libero_object; observed 98.6% (500 episodes)

Browse files
gr00t_n17_native_libero_object_full_finetune_bs64/checkpoints/step-030000-epoch-315-loss=0.0074.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6f5e335605491576c8ff01c0e664913c738ff6bbdaf250dc8190893f7a95233d
3
+ size 13820862192
gr00t_n17_native_libero_object_full_finetune_bs64/config.json ADDED
@@ -0,0 +1,2646 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_ACTIVE_TRACKERS": [
3
+ "jsonl"
4
+ ],
5
+ "_DATASET_NAME": "libero_object_no_noops_lerobotv2.1",
6
+ "_LIBERO_DATA_ROOT": "./datasets/libero_object_no_noops_lerobotv2.1",
7
+ "_N17_ACTION_LAYOUT": [
8
+ [
9
+ "x",
10
+ 1
11
+ ],
12
+ [
13
+ "y",
14
+ 1
15
+ ],
16
+ [
17
+ "z",
18
+ 1
19
+ ],
20
+ [
21
+ "roll",
22
+ 1
23
+ ],
24
+ [
25
+ "pitch",
26
+ 1
27
+ ],
28
+ [
29
+ "yaw",
30
+ 1
31
+ ],
32
+ [
33
+ "gripper",
34
+ 1
35
+ ]
36
+ ],
37
+ "_N17_FLAT_STATISTICS": {
38
+ "action": {
39
+ "max": [
40
+ 0.9375,
41
+ 0.8919642567634583,
42
+ 0.9375,
43
+ 0.17678570747375488,
44
+ 0.35035714507102966,
45
+ 0.1810714304447174,
46
+ 1.0
47
+ ],
48
+ "mean": [
49
+ 0.07127183159196328,
50
+ 0.13518370972847993,
51
+ -0.046194642259443225,
52
+ 0.0011625276173484506,
53
+ 0.007098307809942585,
54
+ -0.015082184949062672,
55
+ 0.46515324845117295
56
+ ],
57
+ "min": [
58
+ -0.8839285969734192,
59
+ -0.9375,
60
+ -0.9375,
61
+ -0.15000000596046448,
62
+ -0.29035714268684387,
63
+ -0.32892856001853943,
64
+ 0.0
65
+ ],
66
+ "q01": [
67
+ -0.5410714149475098,
68
+ -0.8758928775787354,
69
+ -0.9375,
70
+ -0.06964285671710968,
71
+ -0.11678571254014969,
72
+ -0.1607142835855484,
73
+ 0.0
74
+ ],
75
+ "q99": [
76
+ 0.8464285731315613,
77
+ 0.84375,
78
+ 0.9375,
79
+ 0.08142857253551483,
80
+ 0.14884285390376858,
81
+ 0.08571428805589676,
82
+ 1.0
83
+ ],
84
+ "std": [
85
+ 0.26873661352251504,
86
+ 0.4385525563650988,
87
+ 0.4474788640468074,
88
+ 0.024271765379373398,
89
+ 0.04932243328601301,
90
+ 0.04218624160534833,
91
+ 0.4987842257994289
92
+ ]
93
+ },
94
+ "state": {
95
+ "max": [
96
+ 0.14580604434013367,
97
+ 0.3322660028934479,
98
+ 0.38492220640182495,
99
+ 3.4003379344940186,
100
+ 0.7954724431037903,
101
+ 0.6642153263092041,
102
+ 0.04104502499103546,
103
+ -0.0005043679848313332
104
+ ],
105
+ "mean": [
106
+ -0.029590097511622324,
107
+ -0.007297964757377373,
108
+ 0.20263691224329605,
109
+ 3.107347627230808,
110
+ -0.21535565470621534,
111
+ -0.11427672217768083,
112
+ 0.029502615037539585,
113
+ -0.030672144202923204
114
+ ],
115
+ "min": [
116
+ -0.17652566730976105,
117
+ -0.28169912099838257,
118
+ 0.008128181099891663,
119
+ 2.2889645099639893,
120
+ -1.8830392360687256,
121
+ -1.0428133010864258,
122
+ 0.0006782737909816206,
123
+ -0.041782498359680176
124
+ ],
125
+ "q01": [
126
+ -0.149256848692894,
127
+ -0.2592130792140961,
128
+ 0.009932418912649154,
129
+ 2.7467152214050294,
130
+ -1.401689453125,
131
+ -0.6867975068092346,
132
+ 0.008699759356677533,
133
+ -0.04015225619077682
134
+ ],
135
+ "q99": [
136
+ 0.09154553085565566,
137
+ 0.2911862111091613,
138
+ 0.3361880838871002,
139
+ 3.260209331512451,
140
+ 0.3196117353439328,
141
+ 0.39586576938629137,
142
+ 0.03989199593663215,
143
+ -0.010282181538641466
144
+ ],
145
+ "std": [
146
+ 0.06715783727495737,
147
+ 0.17582774832397174,
148
+ 0.07806197322485103,
149
+ 0.08837916027138361,
150
+ 0.33845984937441376,
151
+ 0.2067143161337991,
152
+ 0.00947681687068548,
153
+ 0.009059544368504133
154
+ ]
155
+ }
156
+ },
157
+ "_N17_INIT_CKPT": "./checkpoints/GR00T-N1.7-3B",
158
+ "_N17_MODALITY_CONFIGS": {
159
+ "libero_sim": {
160
+ "action": {
161
+ "delta_indices": [
162
+ 0,
163
+ 1,
164
+ 2,
165
+ 3,
166
+ 4,
167
+ 5,
168
+ 6,
169
+ 7,
170
+ 8,
171
+ 9,
172
+ 10,
173
+ 11,
174
+ 12,
175
+ 13,
176
+ 14,
177
+ 15
178
+ ],
179
+ "modality_keys": [
180
+ "x",
181
+ "y",
182
+ "z",
183
+ "roll",
184
+ "pitch",
185
+ "yaw",
186
+ "gripper"
187
+ ]
188
+ },
189
+ "state": {
190
+ "delta_indices": [
191
+ 0
192
+ ],
193
+ "modality_keys": [
194
+ "x",
195
+ "y",
196
+ "z",
197
+ "roll",
198
+ "pitch",
199
+ "yaw",
200
+ "gripper"
201
+ ]
202
+ },
203
+ "video": {
204
+ "delta_indices": [
205
+ 0
206
+ ],
207
+ "modality_keys": [
208
+ "image",
209
+ "wrist_image"
210
+ ]
211
+ }
212
+ }
213
+ },
214
+ "_N17_STATE_LAYOUT": [
215
+ [
216
+ "x",
217
+ 1
218
+ ],
219
+ [
220
+ "y",
221
+ 1
222
+ ],
223
+ [
224
+ "z",
225
+ 1
226
+ ],
227
+ [
228
+ "roll",
229
+ 1
230
+ ],
231
+ [
232
+ "pitch",
233
+ 1
234
+ ],
235
+ [
236
+ "yaw",
237
+ 1
238
+ ],
239
+ [
240
+ "gripper",
241
+ 2
242
+ ]
243
+ ],
244
+ "_N17_STATISTICS": {
245
+ "libero_sim": {
246
+ "action": {
247
+ "gripper": {
248
+ "max": [
249
+ 1.0
250
+ ],
251
+ "mean": [
252
+ 0.46515324845117295
253
+ ],
254
+ "min": [
255
+ 0.0
256
+ ],
257
+ "q01": [
258
+ 0.0
259
+ ],
260
+ "q99": [
261
+ 1.0
262
+ ],
263
+ "std": [
264
+ 0.4987842257994289
265
+ ]
266
+ },
267
+ "pitch": {
268
+ "max": [
269
+ 0.35035714507102966
270
+ ],
271
+ "mean": [
272
+ 0.007098307809942585
273
+ ],
274
+ "min": [
275
+ -0.29035714268684387
276
+ ],
277
+ "q01": [
278
+ -0.11678571254014969
279
+ ],
280
+ "q99": [
281
+ 0.14884285390376858
282
+ ],
283
+ "std": [
284
+ 0.04932243328601301
285
+ ]
286
+ },
287
+ "roll": {
288
+ "max": [
289
+ 0.17678570747375488
290
+ ],
291
+ "mean": [
292
+ 0.0011625276173484506
293
+ ],
294
+ "min": [
295
+ -0.15000000596046448
296
+ ],
297
+ "q01": [
298
+ -0.06964285671710968
299
+ ],
300
+ "q99": [
301
+ 0.08142857253551483
302
+ ],
303
+ "std": [
304
+ 0.024271765379373398
305
+ ]
306
+ },
307
+ "x": {
308
+ "max": [
309
+ 0.9375
310
+ ],
311
+ "mean": [
312
+ 0.07127183159196328
313
+ ],
314
+ "min": [
315
+ -0.8839285969734192
316
+ ],
317
+ "q01": [
318
+ -0.5410714149475098
319
+ ],
320
+ "q99": [
321
+ 0.8464285731315613
322
+ ],
323
+ "std": [
324
+ 0.26873661352251504
325
+ ]
326
+ },
327
+ "y": {
328
+ "max": [
329
+ 0.8919642567634583
330
+ ],
331
+ "mean": [
332
+ 0.13518370972847993
333
+ ],
334
+ "min": [
335
+ -0.9375
336
+ ],
337
+ "q01": [
338
+ -0.8758928775787354
339
+ ],
340
+ "q99": [
341
+ 0.84375
342
+ ],
343
+ "std": [
344
+ 0.4385525563650988
345
+ ]
346
+ },
347
+ "yaw": {
348
+ "max": [
349
+ 0.1810714304447174
350
+ ],
351
+ "mean": [
352
+ -0.015082184949062672
353
+ ],
354
+ "min": [
355
+ -0.32892856001853943
356
+ ],
357
+ "q01": [
358
+ -0.1607142835855484
359
+ ],
360
+ "q99": [
361
+ 0.08571428805589676
362
+ ],
363
+ "std": [
364
+ 0.04218624160534833
365
+ ]
366
+ },
367
+ "z": {
368
+ "max": [
369
+ 0.9375
370
+ ],
371
+ "mean": [
372
+ -0.046194642259443225
373
+ ],
374
+ "min": [
375
+ -0.9375
376
+ ],
377
+ "q01": [
378
+ -0.9375
379
+ ],
380
+ "q99": [
381
+ 0.9375
382
+ ],
383
+ "std": [
384
+ 0.4474788640468074
385
+ ]
386
+ }
387
+ },
388
+ "state": {
389
+ "gripper": {
390
+ "max": [
391
+ 0.04104502499103546,
392
+ -0.0005043679848313332
393
+ ],
394
+ "mean": [
395
+ 0.029502615037539585,
396
+ -0.030672144202923204
397
+ ],
398
+ "min": [
399
+ 0.0006782737909816206,
400
+ -0.041782498359680176
401
+ ],
402
+ "q01": [
403
+ 0.008699759356677533,
404
+ -0.04015225619077682
405
+ ],
406
+ "q99": [
407
+ 0.03989199593663215,
408
+ -0.010282181538641466
409
+ ],
410
+ "std": [
411
+ 0.00947681687068548,
412
+ 0.009059544368504133
413
+ ]
414
+ },
415
+ "pitch": {
416
+ "max": [
417
+ 0.7954724431037903
418
+ ],
419
+ "mean": [
420
+ -0.21535565470621534
421
+ ],
422
+ "min": [
423
+ -1.8830392360687256
424
+ ],
425
+ "q01": [
426
+ -1.401689453125
427
+ ],
428
+ "q99": [
429
+ 0.3196117353439328
430
+ ],
431
+ "std": [
432
+ 0.33845984937441376
433
+ ]
434
+ },
435
+ "roll": {
436
+ "max": [
437
+ 3.4003379344940186
438
+ ],
439
+ "mean": [
440
+ 3.107347627230808
441
+ ],
442
+ "min": [
443
+ 2.2889645099639893
444
+ ],
445
+ "q01": [
446
+ 2.7467152214050294
447
+ ],
448
+ "q99": [
449
+ 3.260209331512451
450
+ ],
451
+ "std": [
452
+ 0.08837916027138361
453
+ ]
454
+ },
455
+ "x": {
456
+ "max": [
457
+ 0.14580604434013367
458
+ ],
459
+ "mean": [
460
+ -0.029590097511622324
461
+ ],
462
+ "min": [
463
+ -0.17652566730976105
464
+ ],
465
+ "q01": [
466
+ -0.149256848692894
467
+ ],
468
+ "q99": [
469
+ 0.09154553085565566
470
+ ],
471
+ "std": [
472
+ 0.06715783727495737
473
+ ]
474
+ },
475
+ "y": {
476
+ "max": [
477
+ 0.3322660028934479
478
+ ],
479
+ "mean": [
480
+ -0.007297964757377373
481
+ ],
482
+ "min": [
483
+ -0.28169912099838257
484
+ ],
485
+ "q01": [
486
+ -0.2592130792140961
487
+ ],
488
+ "q99": [
489
+ 0.2911862111091613
490
+ ],
491
+ "std": [
492
+ 0.17582774832397174
493
+ ]
494
+ },
495
+ "yaw": {
496
+ "max": [
497
+ 0.6642153263092041
498
+ ],
499
+ "mean": [
500
+ -0.11427672217768083
501
+ ],
502
+ "min": [
503
+ -1.0428133010864258
504
+ ],
505
+ "q01": [
506
+ -0.6867975068092346
507
+ ],
508
+ "q99": [
509
+ 0.39586576938629137
510
+ ],
511
+ "std": [
512
+ 0.2067143161337991
513
+ ]
514
+ },
515
+ "z": {
516
+ "max": [
517
+ 0.38492220640182495
518
+ ],
519
+ "mean": [
520
+ 0.20263691224329605
521
+ ],
522
+ "min": [
523
+ 0.008128181099891663
524
+ ],
525
+ "q01": [
526
+ 0.009932418912649154
527
+ ],
528
+ "q99": [
529
+ 0.3361880838871002
530
+ ],
531
+ "std": [
532
+ 0.07806197322485103
533
+ ]
534
+ }
535
+ }
536
+ }
537
+ },
538
+ "_PROCESSOR_KWARGS": {
539
+ "apply_sincos_state_encoding": false,
540
+ "clip_outliers": true,
541
+ "color_jitter_params": {
542
+ "brightness": 0.3,
543
+ "contrast": 0.4,
544
+ "hue": 0.08,
545
+ "saturation": 0.5
546
+ },
547
+ "crop_fraction": null,
548
+ "embodiment_id_mapping": {
549
+ "libero_sim": 2
550
+ },
551
+ "formalize_language": true,
552
+ "image_crop_size": [
553
+ 230,
554
+ 230
555
+ ],
556
+ "image_target_size": [
557
+ 256,
558
+ 256
559
+ ],
560
+ "max_action_dim": 132,
561
+ "max_action_horizon": 40,
562
+ "max_state_dim": 132,
563
+ "modality_configs": {
564
+ "libero_sim": {
565
+ "action": {
566
+ "delta_indices": [
567
+ 0,
568
+ 1,
569
+ 2,
570
+ 3,
571
+ 4,
572
+ 5,
573
+ 6,
574
+ 7,
575
+ 8,
576
+ 9,
577
+ 10,
578
+ 11,
579
+ 12,
580
+ 13,
581
+ 14,
582
+ 15
583
+ ],
584
+ "modality_keys": [
585
+ "x",
586
+ "y",
587
+ "z",
588
+ "roll",
589
+ "pitch",
590
+ "yaw",
591
+ "gripper"
592
+ ]
593
+ },
594
+ "state": {
595
+ "delta_indices": [
596
+ 0
597
+ ],
598
+ "modality_keys": [
599
+ "x",
600
+ "y",
601
+ "z",
602
+ "roll",
603
+ "pitch",
604
+ "yaw",
605
+ "gripper"
606
+ ]
607
+ },
608
+ "video": {
609
+ "delta_indices": [
610
+ 0
611
+ ],
612
+ "modality_keys": [
613
+ "image",
614
+ "wrist_image"
615
+ ]
616
+ }
617
+ }
618
+ },
619
+ "shortest_image_edge": null,
620
+ "state_dropout_prob": 0.2,
621
+ "statistics": {
622
+ "libero_sim": {
623
+ "action": {
624
+ "gripper": {
625
+ "max": [
626
+ 1.0
627
+ ],
628
+ "mean": [
629
+ 0.46515324845117295
630
+ ],
631
+ "min": [
632
+ 0.0
633
+ ],
634
+ "q01": [
635
+ 0.0
636
+ ],
637
+ "q99": [
638
+ 1.0
639
+ ],
640
+ "std": [
641
+ 0.4987842257994289
642
+ ]
643
+ },
644
+ "pitch": {
645
+ "max": [
646
+ 0.35035714507102966
647
+ ],
648
+ "mean": [
649
+ 0.007098307809942585
650
+ ],
651
+ "min": [
652
+ -0.29035714268684387
653
+ ],
654
+ "q01": [
655
+ -0.11678571254014969
656
+ ],
657
+ "q99": [
658
+ 0.14884285390376858
659
+ ],
660
+ "std": [
661
+ 0.04932243328601301
662
+ ]
663
+ },
664
+ "roll": {
665
+ "max": [
666
+ 0.17678570747375488
667
+ ],
668
+ "mean": [
669
+ 0.0011625276173484506
670
+ ],
671
+ "min": [
672
+ -0.15000000596046448
673
+ ],
674
+ "q01": [
675
+ -0.06964285671710968
676
+ ],
677
+ "q99": [
678
+ 0.08142857253551483
679
+ ],
680
+ "std": [
681
+ 0.024271765379373398
682
+ ]
683
+ },
684
+ "x": {
685
+ "max": [
686
+ 0.9375
687
+ ],
688
+ "mean": [
689
+ 0.07127183159196328
690
+ ],
691
+ "min": [
692
+ -0.8839285969734192
693
+ ],
694
+ "q01": [
695
+ -0.5410714149475098
696
+ ],
697
+ "q99": [
698
+ 0.8464285731315613
699
+ ],
700
+ "std": [
701
+ 0.26873661352251504
702
+ ]
703
+ },
704
+ "y": {
705
+ "max": [
706
+ 0.8919642567634583
707
+ ],
708
+ "mean": [
709
+ 0.13518370972847993
710
+ ],
711
+ "min": [
712
+ -0.9375
713
+ ],
714
+ "q01": [
715
+ -0.8758928775787354
716
+ ],
717
+ "q99": [
718
+ 0.84375
719
+ ],
720
+ "std": [
721
+ 0.4385525563650988
722
+ ]
723
+ },
724
+ "yaw": {
725
+ "max": [
726
+ 0.1810714304447174
727
+ ],
728
+ "mean": [
729
+ -0.015082184949062672
730
+ ],
731
+ "min": [
732
+ -0.32892856001853943
733
+ ],
734
+ "q01": [
735
+ -0.1607142835855484
736
+ ],
737
+ "q99": [
738
+ 0.08571428805589676
739
+ ],
740
+ "std": [
741
+ 0.04218624160534833
742
+ ]
743
+ },
744
+ "z": {
745
+ "max": [
746
+ 0.9375
747
+ ],
748
+ "mean": [
749
+ -0.046194642259443225
750
+ ],
751
+ "min": [
752
+ -0.9375
753
+ ],
754
+ "q01": [
755
+ -0.9375
756
+ ],
757
+ "q99": [
758
+ 0.9375
759
+ ],
760
+ "std": [
761
+ 0.4474788640468074
762
+ ]
763
+ }
764
+ },
765
+ "state": {
766
+ "gripper": {
767
+ "max": [
768
+ 0.04104502499103546,
769
+ -0.0005043679848313332
770
+ ],
771
+ "mean": [
772
+ 0.029502615037539585,
773
+ -0.030672144202923204
774
+ ],
775
+ "min": [
776
+ 0.0006782737909816206,
777
+ -0.041782498359680176
778
+ ],
779
+ "q01": [
780
+ 0.008699759356677533,
781
+ -0.04015225619077682
782
+ ],
783
+ "q99": [
784
+ 0.03989199593663215,
785
+ -0.010282181538641466
786
+ ],
787
+ "std": [
788
+ 0.00947681687068548,
789
+ 0.009059544368504133
790
+ ]
791
+ },
792
+ "pitch": {
793
+ "max": [
794
+ 0.7954724431037903
795
+ ],
796
+ "mean": [
797
+ -0.21535565470621534
798
+ ],
799
+ "min": [
800
+ -1.8830392360687256
801
+ ],
802
+ "q01": [
803
+ -1.401689453125
804
+ ],
805
+ "q99": [
806
+ 0.3196117353439328
807
+ ],
808
+ "std": [
809
+ 0.33845984937441376
810
+ ]
811
+ },
812
+ "roll": {
813
+ "max": [
814
+ 3.4003379344940186
815
+ ],
816
+ "mean": [
817
+ 3.107347627230808
818
+ ],
819
+ "min": [
820
+ 2.2889645099639893
821
+ ],
822
+ "q01": [
823
+ 2.7467152214050294
824
+ ],
825
+ "q99": [
826
+ 3.260209331512451
827
+ ],
828
+ "std": [
829
+ 0.08837916027138361
830
+ ]
831
+ },
832
+ "x": {
833
+ "max": [
834
+ 0.14580604434013367
835
+ ],
836
+ "mean": [
837
+ -0.029590097511622324
838
+ ],
839
+ "min": [
840
+ -0.17652566730976105
841
+ ],
842
+ "q01": [
843
+ -0.149256848692894
844
+ ],
845
+ "q99": [
846
+ 0.09154553085565566
847
+ ],
848
+ "std": [
849
+ 0.06715783727495737
850
+ ]
851
+ },
852
+ "y": {
853
+ "max": [
854
+ 0.3322660028934479
855
+ ],
856
+ "mean": [
857
+ -0.007297964757377373
858
+ ],
859
+ "min": [
860
+ -0.28169912099838257
861
+ ],
862
+ "q01": [
863
+ -0.2592130792140961
864
+ ],
865
+ "q99": [
866
+ 0.2911862111091613
867
+ ],
868
+ "std": [
869
+ 0.17582774832397174
870
+ ]
871
+ },
872
+ "yaw": {
873
+ "max": [
874
+ 0.6642153263092041
875
+ ],
876
+ "mean": [
877
+ -0.11427672217768083
878
+ ],
879
+ "min": [
880
+ -1.0428133010864258
881
+ ],
882
+ "q01": [
883
+ -0.6867975068092346
884
+ ],
885
+ "q99": [
886
+ 0.39586576938629137
887
+ ],
888
+ "std": [
889
+ 0.2067143161337991
890
+ ]
891
+ },
892
+ "z": {
893
+ "max": [
894
+ 0.38492220640182495
895
+ ],
896
+ "mean": [
897
+ 0.20263691224329605
898
+ ],
899
+ "min": [
900
+ 0.008128181099891663
901
+ ],
902
+ "q01": [
903
+ 0.009932418912649154
904
+ ],
905
+ "q99": [
906
+ 0.3361880838871002
907
+ ],
908
+ "std": [
909
+ 0.07806197322485103
910
+ ]
911
+ }
912
+ }
913
+ }
914
+ },
915
+ "use_albumentations": true,
916
+ "use_percentiles": true,
917
+ "use_relative_action": true
918
+ },
919
+ "_QWEN3_VL_CONFIG": {
920
+ "architectures": [
921
+ "Qwen3VLForConditionalGeneration"
922
+ ],
923
+ "image_token_id": 151655,
924
+ "text_config": {
925
+ "attention_bias": false,
926
+ "attention_dropout": 0.0,
927
+ "head_dim": 128,
928
+ "hidden_act": "silu",
929
+ "hidden_size": 2048,
930
+ "initializer_range": 0.02,
931
+ "intermediate_size": 6144,
932
+ "max_position_embeddings": 262144,
933
+ "model_type": "qwen3_vl_text",
934
+ "num_attention_heads": 16,
935
+ "num_hidden_layers": 28,
936
+ "num_key_value_heads": 8,
937
+ "rms_norm_eps": 1e-06,
938
+ "rope_parameters": {
939
+ "mrope_interleaved": true,
940
+ "mrope_section": [
941
+ 24,
942
+ 20,
943
+ 20
944
+ ],
945
+ "rope_theta": 5000000.0,
946
+ "rope_type": "default"
947
+ },
948
+ "use_cache": false,
949
+ "vocab_size": 151936
950
+ },
951
+ "tie_word_embeddings": false,
952
+ "video_token_id": 151656,
953
+ "vision_config": {
954
+ "deepstack_visual_indexes": [
955
+ 5,
956
+ 11,
957
+ 17
958
+ ],
959
+ "depth": 24,
960
+ "hidden_act": "gelu_pytorch_tanh",
961
+ "hidden_size": 1024,
962
+ "in_channels": 3,
963
+ "initializer_range": 0.02,
964
+ "intermediate_size": 4096,
965
+ "model_type": "qwen3_vl_vision",
966
+ "num_heads": 16,
967
+ "num_position_embeddings": 2304,
968
+ "out_hidden_size": 2048,
969
+ "patch_size": 16,
970
+ "spatial_merge_size": 2,
971
+ "temporal_patch_size": 2
972
+ },
973
+ "vision_end_token_id": 151653,
974
+ "vision_start_token_id": 151652
975
+ },
976
+ "_QWEN_TOKENIZER_PATH": "fluxvla/models/third_party_models/qwen3_tokenizer",
977
+ "_STATISTIC_NAME": "libero_object_no_noops_native",
978
+ "_SUITE": "libero_object",
979
+ "eval": {
980
+ "dataset": {
981
+ "transforms": [
982
+ {
983
+ "embodiment_id": 2,
984
+ "img_keys": [
985
+ "agentview_image",
986
+ "robot0_eye_in_hand_image"
987
+ ],
988
+ "type": "ProcessLiberoEvalInputs"
989
+ },
990
+ {
991
+ "norm_type": null,
992
+ "out_key": "states",
993
+ "type": "LiberoProprioFromInputs"
994
+ },
995
+ {
996
+ "action_key": null,
997
+ "clip_norm": true,
998
+ "norm_type": "quantile",
999
+ "normalization_epsilon": 0.0,
1000
+ "preserve_input_dtype": true,
1001
+ "state_dim": 132,
1002
+ "state_key": "states",
1003
+ "statistics_key": "norm_stats",
1004
+ "type": "NormalizeStatesAndActions"
1005
+ },
1006
+ {
1007
+ "action_horizon": 40,
1008
+ "state_dropout_prob": 0.0,
1009
+ "state_history_length": 1,
1010
+ "type": "PrepareStateActionTargets",
1011
+ "valid_action_dim": 7
1012
+ },
1013
+ {
1014
+ "embodiment_tag": "LIBERO_PANDA",
1015
+ "image_key": "pixel_values",
1016
+ "output_image_key": "pixel_values",
1017
+ "processor_kwargs": {
1018
+ "apply_sincos_state_encoding": false,
1019
+ "clip_outliers": true,
1020
+ "color_jitter_params": {
1021
+ "brightness": 0.3,
1022
+ "contrast": 0.4,
1023
+ "hue": 0.08,
1024
+ "saturation": 0.5
1025
+ },
1026
+ "crop_fraction": 0.95,
1027
+ "embodiment_id_mapping": {
1028
+ "libero_sim": 2
1029
+ },
1030
+ "formalize_language": true,
1031
+ "image_crop_size": [
1032
+ 230,
1033
+ 230
1034
+ ],
1035
+ "image_target_size": [
1036
+ 256,
1037
+ 256
1038
+ ],
1039
+ "max_action_dim": 132,
1040
+ "max_action_horizon": 40,
1041
+ "max_state_dim": 132,
1042
+ "modality_configs": {
1043
+ "libero_sim": {
1044
+ "action": {
1045
+ "delta_indices": [
1046
+ 0,
1047
+ 1,
1048
+ 2,
1049
+ 3,
1050
+ 4,
1051
+ 5,
1052
+ 6,
1053
+ 7,
1054
+ 8,
1055
+ 9,
1056
+ 10,
1057
+ 11,
1058
+ 12,
1059
+ 13,
1060
+ 14,
1061
+ 15
1062
+ ],
1063
+ "modality_keys": [
1064
+ "x",
1065
+ "y",
1066
+ "z",
1067
+ "roll",
1068
+ "pitch",
1069
+ "yaw",
1070
+ "gripper"
1071
+ ]
1072
+ },
1073
+ "state": {
1074
+ "delta_indices": [
1075
+ 0
1076
+ ],
1077
+ "modality_keys": [
1078
+ "x",
1079
+ "y",
1080
+ "z",
1081
+ "roll",
1082
+ "pitch",
1083
+ "yaw",
1084
+ "gripper"
1085
+ ]
1086
+ },
1087
+ "video": {
1088
+ "delta_indices": [
1089
+ 0
1090
+ ],
1091
+ "modality_keys": [
1092
+ "image",
1093
+ "wrist_image"
1094
+ ]
1095
+ }
1096
+ }
1097
+ },
1098
+ "shortest_image_edge": 256,
1099
+ "state_dropout_prob": 0.2,
1100
+ "statistics": {
1101
+ "libero_sim": {
1102
+ "action": {
1103
+ "gripper": {
1104
+ "max": [
1105
+ 1.0
1106
+ ],
1107
+ "mean": [
1108
+ 0.46515324845117295
1109
+ ],
1110
+ "min": [
1111
+ 0.0
1112
+ ],
1113
+ "q01": [
1114
+ 0.0
1115
+ ],
1116
+ "q99": [
1117
+ 1.0
1118
+ ],
1119
+ "std": [
1120
+ 0.4987842257994289
1121
+ ]
1122
+ },
1123
+ "pitch": {
1124
+ "max": [
1125
+ 0.35035714507102966
1126
+ ],
1127
+ "mean": [
1128
+ 0.007098307809942585
1129
+ ],
1130
+ "min": [
1131
+ -0.29035714268684387
1132
+ ],
1133
+ "q01": [
1134
+ -0.11678571254014969
1135
+ ],
1136
+ "q99": [
1137
+ 0.14884285390376858
1138
+ ],
1139
+ "std": [
1140
+ 0.04932243328601301
1141
+ ]
1142
+ },
1143
+ "roll": {
1144
+ "max": [
1145
+ 0.17678570747375488
1146
+ ],
1147
+ "mean": [
1148
+ 0.0011625276173484506
1149
+ ],
1150
+ "min": [
1151
+ -0.15000000596046448
1152
+ ],
1153
+ "q01": [
1154
+ -0.06964285671710968
1155
+ ],
1156
+ "q99": [
1157
+ 0.08142857253551483
1158
+ ],
1159
+ "std": [
1160
+ 0.024271765379373398
1161
+ ]
1162
+ },
1163
+ "x": {
1164
+ "max": [
1165
+ 0.9375
1166
+ ],
1167
+ "mean": [
1168
+ 0.07127183159196328
1169
+ ],
1170
+ "min": [
1171
+ -0.8839285969734192
1172
+ ],
1173
+ "q01": [
1174
+ -0.5410714149475098
1175
+ ],
1176
+ "q99": [
1177
+ 0.8464285731315613
1178
+ ],
1179
+ "std": [
1180
+ 0.26873661352251504
1181
+ ]
1182
+ },
1183
+ "y": {
1184
+ "max": [
1185
+ 0.8919642567634583
1186
+ ],
1187
+ "mean": [
1188
+ 0.13518370972847993
1189
+ ],
1190
+ "min": [
1191
+ -0.9375
1192
+ ],
1193
+ "q01": [
1194
+ -0.8758928775787354
1195
+ ],
1196
+ "q99": [
1197
+ 0.84375
1198
+ ],
1199
+ "std": [
1200
+ 0.4385525563650988
1201
+ ]
1202
+ },
1203
+ "yaw": {
1204
+ "max": [
1205
+ 0.1810714304447174
1206
+ ],
1207
+ "mean": [
1208
+ -0.015082184949062672
1209
+ ],
1210
+ "min": [
1211
+ -0.32892856001853943
1212
+ ],
1213
+ "q01": [
1214
+ -0.1607142835855484
1215
+ ],
1216
+ "q99": [
1217
+ 0.08571428805589676
1218
+ ],
1219
+ "std": [
1220
+ 0.04218624160534833
1221
+ ]
1222
+ },
1223
+ "z": {
1224
+ "max": [
1225
+ 0.9375
1226
+ ],
1227
+ "mean": [
1228
+ -0.046194642259443225
1229
+ ],
1230
+ "min": [
1231
+ -0.9375
1232
+ ],
1233
+ "q01": [
1234
+ -0.9375
1235
+ ],
1236
+ "q99": [
1237
+ 0.9375
1238
+ ],
1239
+ "std": [
1240
+ 0.4474788640468074
1241
+ ]
1242
+ }
1243
+ },
1244
+ "state": {
1245
+ "gripper": {
1246
+ "max": [
1247
+ 0.04104502499103546,
1248
+ -0.0005043679848313332
1249
+ ],
1250
+ "mean": [
1251
+ 0.029502615037539585,
1252
+ -0.030672144202923204
1253
+ ],
1254
+ "min": [
1255
+ 0.0006782737909816206,
1256
+ -0.041782498359680176
1257
+ ],
1258
+ "q01": [
1259
+ 0.008699759356677533,
1260
+ -0.04015225619077682
1261
+ ],
1262
+ "q99": [
1263
+ 0.03989199593663215,
1264
+ -0.010282181538641466
1265
+ ],
1266
+ "std": [
1267
+ 0.00947681687068548,
1268
+ 0.009059544368504133
1269
+ ]
1270
+ },
1271
+ "pitch": {
1272
+ "max": [
1273
+ 0.7954724431037903
1274
+ ],
1275
+ "mean": [
1276
+ -0.21535565470621534
1277
+ ],
1278
+ "min": [
1279
+ -1.8830392360687256
1280
+ ],
1281
+ "q01": [
1282
+ -1.401689453125
1283
+ ],
1284
+ "q99": [
1285
+ 0.3196117353439328
1286
+ ],
1287
+ "std": [
1288
+ 0.33845984937441376
1289
+ ]
1290
+ },
1291
+ "roll": {
1292
+ "max": [
1293
+ 3.4003379344940186
1294
+ ],
1295
+ "mean": [
1296
+ 3.107347627230808
1297
+ ],
1298
+ "min": [
1299
+ 2.2889645099639893
1300
+ ],
1301
+ "q01": [
1302
+ 2.7467152214050294
1303
+ ],
1304
+ "q99": [
1305
+ 3.260209331512451
1306
+ ],
1307
+ "std": [
1308
+ 0.08837916027138361
1309
+ ]
1310
+ },
1311
+ "x": {
1312
+ "max": [
1313
+ 0.14580604434013367
1314
+ ],
1315
+ "mean": [
1316
+ -0.029590097511622324
1317
+ ],
1318
+ "min": [
1319
+ -0.17652566730976105
1320
+ ],
1321
+ "q01": [
1322
+ -0.149256848692894
1323
+ ],
1324
+ "q99": [
1325
+ 0.09154553085565566
1326
+ ],
1327
+ "std": [
1328
+ 0.06715783727495737
1329
+ ]
1330
+ },
1331
+ "y": {
1332
+ "max": [
1333
+ 0.3322660028934479
1334
+ ],
1335
+ "mean": [
1336
+ -0.007297964757377373
1337
+ ],
1338
+ "min": [
1339
+ -0.28169912099838257
1340
+ ],
1341
+ "q01": [
1342
+ -0.2592130792140961
1343
+ ],
1344
+ "q99": [
1345
+ 0.2911862111091613
1346
+ ],
1347
+ "std": [
1348
+ 0.17582774832397174
1349
+ ]
1350
+ },
1351
+ "yaw": {
1352
+ "max": [
1353
+ 0.6642153263092041
1354
+ ],
1355
+ "mean": [
1356
+ -0.11427672217768083
1357
+ ],
1358
+ "min": [
1359
+ -1.0428133010864258
1360
+ ],
1361
+ "q01": [
1362
+ -0.6867975068092346
1363
+ ],
1364
+ "q99": [
1365
+ 0.39586576938629137
1366
+ ],
1367
+ "std": [
1368
+ 0.2067143161337991
1369
+ ]
1370
+ },
1371
+ "z": {
1372
+ "max": [
1373
+ 0.38492220640182495
1374
+ ],
1375
+ "mean": [
1376
+ 0.20263691224329605
1377
+ ],
1378
+ "min": [
1379
+ 0.008128181099891663
1380
+ ],
1381
+ "q01": [
1382
+ 0.009932418912649154
1383
+ ],
1384
+ "q99": [
1385
+ 0.3361880838871002
1386
+ ],
1387
+ "std": [
1388
+ 0.07806197322485103
1389
+ ]
1390
+ }
1391
+ }
1392
+ }
1393
+ },
1394
+ "use_albumentations": true,
1395
+ "use_percentiles": true,
1396
+ "use_relative_action": true
1397
+ },
1398
+ "train_mode": false,
1399
+ "type": "GrootN17ImageAugmentation"
1400
+ },
1401
+ {
1402
+ "image_mean": [
1403
+ 0.5,
1404
+ 0.5,
1405
+ 0.5
1406
+ ],
1407
+ "image_std": [
1408
+ 0.5,
1409
+ 0.5,
1410
+ 0.5
1411
+ ],
1412
+ "img_key": "pixel_values",
1413
+ "merge_size": 2,
1414
+ "patch_size": 16,
1415
+ "size": {
1416
+ "longest_edge": 16777216,
1417
+ "shortest_edge": 65536
1418
+ },
1419
+ "temporal_patch_size": 2,
1420
+ "to_tensor": true,
1421
+ "type": "QWen2VLImageTransform"
1422
+ },
1423
+ {
1424
+ "add_assistant_stub": false,
1425
+ "add_system": false,
1426
+ "attention_mask_dtype": "int64",
1427
+ "image_grid_thw_key": "image_grid_thw",
1428
+ "image_merge_size": 2,
1429
+ "image_tag_template": "",
1430
+ "img_context_token": "<|image_pad|>",
1431
+ "img_end": "<|vision_end|>",
1432
+ "img_start": "<|vision_start|>",
1433
+ "img_tokens_source": "from_image_grid_thw",
1434
+ "lowercase_task_description": true,
1435
+ "max_len": 180,
1436
+ "output_keys": [
1437
+ "lang_tokens",
1438
+ "lang_masks",
1439
+ "pixel_values",
1440
+ "img_masks",
1441
+ "image_grid_thw",
1442
+ "states",
1443
+ "embodiment_ids",
1444
+ "replay_img"
1445
+ ],
1446
+ "padding_side": "left",
1447
+ "strip_task_punctuation": true,
1448
+ "task_pos": "after_images",
1449
+ "tokenizer": {
1450
+ "model_path": "fluxvla/models/third_party_models/qwen3_tokenizer",
1451
+ "padding_side": "left",
1452
+ "trust_remote_code": false,
1453
+ "type": "PretrainedTokenizer"
1454
+ },
1455
+ "truncate": false,
1456
+ "type": "ProcessPromptsWithImage",
1457
+ "use_eos_as_pad": false
1458
+ }
1459
+ ],
1460
+ "type": "LiberoParquetEvalDataset"
1461
+ },
1462
+ "denormalize_action": {
1463
+ "action_dim": 7,
1464
+ "denorm_action": false,
1465
+ "type": "DenormalizeLiberoAction"
1466
+ },
1467
+ "eval_chunk_size": 8,
1468
+ "eval_shard_strategy": "episode",
1469
+ "inference_seed": 42,
1470
+ "max_steps": 720,
1471
+ "model_family": "groot_n17_native",
1472
+ "norm_stats_key": "libero_object_no_noops_native",
1473
+ "num_steps_wait": 0,
1474
+ "num_trials_per_task": 50,
1475
+ "preprocess_every_step": false,
1476
+ "result_output_dir": "work_dirs/groot_n17_libero_object_fluxvla_gbs640_seed42_abf72b8_crop_30k/eval_trainseed7_evalseed7_infeseed42",
1477
+ "save_failed_rollout_videos": false,
1478
+ "save_multi_view_rollout_videos": false,
1479
+ "save_rollout_videos": false,
1480
+ "seed": 7,
1481
+ "task_suite_name": "libero_object",
1482
+ "type": "LiberoEvalRunner"
1483
+ },
1484
+ "model": {
1485
+ "embodiment_tag": "LIBERO_PANDA",
1486
+ "freeze_projector": false,
1487
+ "freeze_vlm_backbone": true,
1488
+ "load_metadata": true,
1489
+ "model_path": "./checkpoints/GR00T-N1.7-3B",
1490
+ "processor_kwargs": {
1491
+ "apply_sincos_state_encoding": false,
1492
+ "clip_outliers": true,
1493
+ "color_jitter_params": {
1494
+ "brightness": 0.3,
1495
+ "contrast": 0.4,
1496
+ "hue": 0.08,
1497
+ "saturation": 0.5
1498
+ },
1499
+ "crop_fraction": 0.95,
1500
+ "embodiment_id_mapping": {
1501
+ "libero_sim": 2
1502
+ },
1503
+ "formalize_language": true,
1504
+ "image_crop_size": [
1505
+ 230,
1506
+ 230
1507
+ ],
1508
+ "image_target_size": [
1509
+ 256,
1510
+ 256
1511
+ ],
1512
+ "max_action_dim": 132,
1513
+ "max_action_horizon": 40,
1514
+ "max_state_dim": 132,
1515
+ "modality_configs": {
1516
+ "libero_sim": {
1517
+ "action": {
1518
+ "delta_indices": [
1519
+ 0,
1520
+ 1,
1521
+ 2,
1522
+ 3,
1523
+ 4,
1524
+ 5,
1525
+ 6,
1526
+ 7,
1527
+ 8,
1528
+ 9,
1529
+ 10,
1530
+ 11,
1531
+ 12,
1532
+ 13,
1533
+ 14,
1534
+ 15
1535
+ ],
1536
+ "modality_keys": [
1537
+ "x",
1538
+ "y",
1539
+ "z",
1540
+ "roll",
1541
+ "pitch",
1542
+ "yaw",
1543
+ "gripper"
1544
+ ]
1545
+ },
1546
+ "state": {
1547
+ "delta_indices": [
1548
+ 0
1549
+ ],
1550
+ "modality_keys": [
1551
+ "x",
1552
+ "y",
1553
+ "z",
1554
+ "roll",
1555
+ "pitch",
1556
+ "yaw",
1557
+ "gripper"
1558
+ ]
1559
+ },
1560
+ "video": {
1561
+ "delta_indices": [
1562
+ 0
1563
+ ],
1564
+ "modality_keys": [
1565
+ "image",
1566
+ "wrist_image"
1567
+ ]
1568
+ }
1569
+ }
1570
+ },
1571
+ "shortest_image_edge": 256,
1572
+ "state_dropout_prob": 0.2,
1573
+ "statistics": {
1574
+ "libero_sim": {
1575
+ "action": {
1576
+ "gripper": {
1577
+ "max": [
1578
+ 1.0
1579
+ ],
1580
+ "mean": [
1581
+ 0.46515324845117295
1582
+ ],
1583
+ "min": [
1584
+ 0.0
1585
+ ],
1586
+ "q01": [
1587
+ 0.0
1588
+ ],
1589
+ "q99": [
1590
+ 1.0
1591
+ ],
1592
+ "std": [
1593
+ 0.4987842257994289
1594
+ ]
1595
+ },
1596
+ "pitch": {
1597
+ "max": [
1598
+ 0.35035714507102966
1599
+ ],
1600
+ "mean": [
1601
+ 0.007098307809942585
1602
+ ],
1603
+ "min": [
1604
+ -0.29035714268684387
1605
+ ],
1606
+ "q01": [
1607
+ -0.11678571254014969
1608
+ ],
1609
+ "q99": [
1610
+ 0.14884285390376858
1611
+ ],
1612
+ "std": [
1613
+ 0.04932243328601301
1614
+ ]
1615
+ },
1616
+ "roll": {
1617
+ "max": [
1618
+ 0.17678570747375488
1619
+ ],
1620
+ "mean": [
1621
+ 0.0011625276173484506
1622
+ ],
1623
+ "min": [
1624
+ -0.15000000596046448
1625
+ ],
1626
+ "q01": [
1627
+ -0.06964285671710968
1628
+ ],
1629
+ "q99": [
1630
+ 0.08142857253551483
1631
+ ],
1632
+ "std": [
1633
+ 0.024271765379373398
1634
+ ]
1635
+ },
1636
+ "x": {
1637
+ "max": [
1638
+ 0.9375
1639
+ ],
1640
+ "mean": [
1641
+ 0.07127183159196328
1642
+ ],
1643
+ "min": [
1644
+ -0.8839285969734192
1645
+ ],
1646
+ "q01": [
1647
+ -0.5410714149475098
1648
+ ],
1649
+ "q99": [
1650
+ 0.8464285731315613
1651
+ ],
1652
+ "std": [
1653
+ 0.26873661352251504
1654
+ ]
1655
+ },
1656
+ "y": {
1657
+ "max": [
1658
+ 0.8919642567634583
1659
+ ],
1660
+ "mean": [
1661
+ 0.13518370972847993
1662
+ ],
1663
+ "min": [
1664
+ -0.9375
1665
+ ],
1666
+ "q01": [
1667
+ -0.8758928775787354
1668
+ ],
1669
+ "q99": [
1670
+ 0.84375
1671
+ ],
1672
+ "std": [
1673
+ 0.4385525563650988
1674
+ ]
1675
+ },
1676
+ "yaw": {
1677
+ "max": [
1678
+ 0.1810714304447174
1679
+ ],
1680
+ "mean": [
1681
+ -0.015082184949062672
1682
+ ],
1683
+ "min": [
1684
+ -0.32892856001853943
1685
+ ],
1686
+ "q01": [
1687
+ -0.1607142835855484
1688
+ ],
1689
+ "q99": [
1690
+ 0.08571428805589676
1691
+ ],
1692
+ "std": [
1693
+ 0.04218624160534833
1694
+ ]
1695
+ },
1696
+ "z": {
1697
+ "max": [
1698
+ 0.9375
1699
+ ],
1700
+ "mean": [
1701
+ -0.046194642259443225
1702
+ ],
1703
+ "min": [
1704
+ -0.9375
1705
+ ],
1706
+ "q01": [
1707
+ -0.9375
1708
+ ],
1709
+ "q99": [
1710
+ 0.9375
1711
+ ],
1712
+ "std": [
1713
+ 0.4474788640468074
1714
+ ]
1715
+ }
1716
+ },
1717
+ "state": {
1718
+ "gripper": {
1719
+ "max": [
1720
+ 0.04104502499103546,
1721
+ -0.0005043679848313332
1722
+ ],
1723
+ "mean": [
1724
+ 0.029502615037539585,
1725
+ -0.030672144202923204
1726
+ ],
1727
+ "min": [
1728
+ 0.0006782737909816206,
1729
+ -0.041782498359680176
1730
+ ],
1731
+ "q01": [
1732
+ 0.008699759356677533,
1733
+ -0.04015225619077682
1734
+ ],
1735
+ "q99": [
1736
+ 0.03989199593663215,
1737
+ -0.010282181538641466
1738
+ ],
1739
+ "std": [
1740
+ 0.00947681687068548,
1741
+ 0.009059544368504133
1742
+ ]
1743
+ },
1744
+ "pitch": {
1745
+ "max": [
1746
+ 0.7954724431037903
1747
+ ],
1748
+ "mean": [
1749
+ -0.21535565470621534
1750
+ ],
1751
+ "min": [
1752
+ -1.8830392360687256
1753
+ ],
1754
+ "q01": [
1755
+ -1.401689453125
1756
+ ],
1757
+ "q99": [
1758
+ 0.3196117353439328
1759
+ ],
1760
+ "std": [
1761
+ 0.33845984937441376
1762
+ ]
1763
+ },
1764
+ "roll": {
1765
+ "max": [
1766
+ 3.4003379344940186
1767
+ ],
1768
+ "mean": [
1769
+ 3.107347627230808
1770
+ ],
1771
+ "min": [
1772
+ 2.2889645099639893
1773
+ ],
1774
+ "q01": [
1775
+ 2.7467152214050294
1776
+ ],
1777
+ "q99": [
1778
+ 3.260209331512451
1779
+ ],
1780
+ "std": [
1781
+ 0.08837916027138361
1782
+ ]
1783
+ },
1784
+ "x": {
1785
+ "max": [
1786
+ 0.14580604434013367
1787
+ ],
1788
+ "mean": [
1789
+ -0.029590097511622324
1790
+ ],
1791
+ "min": [
1792
+ -0.17652566730976105
1793
+ ],
1794
+ "q01": [
1795
+ -0.149256848692894
1796
+ ],
1797
+ "q99": [
1798
+ 0.09154553085565566
1799
+ ],
1800
+ "std": [
1801
+ 0.06715783727495737
1802
+ ]
1803
+ },
1804
+ "y": {
1805
+ "max": [
1806
+ 0.3322660028934479
1807
+ ],
1808
+ "mean": [
1809
+ -0.007297964757377373
1810
+ ],
1811
+ "min": [
1812
+ -0.28169912099838257
1813
+ ],
1814
+ "q01": [
1815
+ -0.2592130792140961
1816
+ ],
1817
+ "q99": [
1818
+ 0.2911862111091613
1819
+ ],
1820
+ "std": [
1821
+ 0.17582774832397174
1822
+ ]
1823
+ },
1824
+ "yaw": {
1825
+ "max": [
1826
+ 0.6642153263092041
1827
+ ],
1828
+ "mean": [
1829
+ -0.11427672217768083
1830
+ ],
1831
+ "min": [
1832
+ -1.0428133010864258
1833
+ ],
1834
+ "q01": [
1835
+ -0.6867975068092346
1836
+ ],
1837
+ "q99": [
1838
+ 0.39586576938629137
1839
+ ],
1840
+ "std": [
1841
+ 0.2067143161337991
1842
+ ]
1843
+ },
1844
+ "z": {
1845
+ "max": [
1846
+ 0.38492220640182495
1847
+ ],
1848
+ "mean": [
1849
+ 0.20263691224329605
1850
+ ],
1851
+ "min": [
1852
+ 0.008128181099891663
1853
+ ],
1854
+ "q01": [
1855
+ 0.009932418912649154
1856
+ ],
1857
+ "q99": [
1858
+ 0.3361880838871002
1859
+ ],
1860
+ "std": [
1861
+ 0.07806197322485103
1862
+ ]
1863
+ }
1864
+ }
1865
+ }
1866
+ },
1867
+ "use_albumentations": true,
1868
+ "use_percentiles": true,
1869
+ "use_relative_action": true
1870
+ },
1871
+ "qwen3_runtime": "compat_457",
1872
+ "state_dropout_prob": 0.2,
1873
+ "type": "GrootN17VLA",
1874
+ "vla_head": {
1875
+ "tune_diffusion_model": true,
1876
+ "tune_vlln": true,
1877
+ "type": "GrootN17ActionHead"
1878
+ },
1879
+ "vlm_backbone": {
1880
+ "load_bf16": false,
1881
+ "model_config": {
1882
+ "architectures": [
1883
+ "Qwen3VLForConditionalGeneration"
1884
+ ],
1885
+ "image_token_id": 151655,
1886
+ "text_config": {
1887
+ "attention_bias": false,
1888
+ "attention_dropout": 0.0,
1889
+ "head_dim": 128,
1890
+ "hidden_act": "silu",
1891
+ "hidden_size": 2048,
1892
+ "initializer_range": 0.02,
1893
+ "intermediate_size": 6144,
1894
+ "max_position_embeddings": 262144,
1895
+ "model_type": "qwen3_vl_text",
1896
+ "num_attention_heads": 16,
1897
+ "num_hidden_layers": 28,
1898
+ "num_key_value_heads": 8,
1899
+ "rms_norm_eps": 1e-06,
1900
+ "rope_parameters": {
1901
+ "mrope_interleaved": true,
1902
+ "mrope_section": [
1903
+ 24,
1904
+ 20,
1905
+ 20
1906
+ ],
1907
+ "rope_theta": 5000000.0,
1908
+ "rope_type": "default"
1909
+ },
1910
+ "use_cache": false,
1911
+ "vocab_size": 151936
1912
+ },
1913
+ "tie_word_embeddings": false,
1914
+ "video_token_id": 151656,
1915
+ "vision_config": {
1916
+ "deepstack_visual_indexes": [
1917
+ 5,
1918
+ 11,
1919
+ 17
1920
+ ],
1921
+ "depth": 24,
1922
+ "hidden_act": "gelu_pytorch_tanh",
1923
+ "hidden_size": 1024,
1924
+ "in_channels": 3,
1925
+ "initializer_range": 0.02,
1926
+ "intermediate_size": 4096,
1927
+ "model_type": "qwen3_vl_vision",
1928
+ "num_heads": 16,
1929
+ "num_position_embeddings": 2304,
1930
+ "out_hidden_size": 2048,
1931
+ "patch_size": 16,
1932
+ "spatial_merge_size": 2,
1933
+ "temporal_patch_size": 2
1934
+ },
1935
+ "vision_end_token_id": 151653,
1936
+ "vision_start_token_id": 151652
1937
+ },
1938
+ "qwen3_runtime": "compat_457",
1939
+ "reproject_vision": false,
1940
+ "select_layer": 16,
1941
+ "trainable_params_fp32": true,
1942
+ "tune_top_llm_layers": 0,
1943
+ "type": "GrootN17Qwen3Backbone",
1944
+ "use_flash_attention": true
1945
+ }
1946
+ },
1947
+ "runner": {
1948
+ "change_key_name": false,
1949
+ "collator": {
1950
+ "keys": [
1951
+ "lang_tokens",
1952
+ "lang_masks",
1953
+ "images",
1954
+ "image_grid_thw",
1955
+ "states",
1956
+ "actions",
1957
+ "action_masks",
1958
+ "embodiment_ids",
1959
+ "sample_weight"
1960
+ ],
1961
+ "type": "DictCollator"
1962
+ },
1963
+ "enable_gradient_checkpointing": true,
1964
+ "enable_mixed_precision_training": true,
1965
+ "grad_accumulation_steps": 1,
1966
+ "lr_scheduler": {
1967
+ "type": "linear-warmup+cosine-decay",
1968
+ "warmup_ratio": 0.05
1969
+ },
1970
+ "max_grad_norm": 1.0,
1971
+ "max_keep_ckpts": 4,
1972
+ "max_steps": 30000,
1973
+ "metric": {
1974
+ "active_trackers": [
1975
+ "jsonl"
1976
+ ],
1977
+ "grad_accumulation_steps": 1,
1978
+ "run_dir": "work_dirs",
1979
+ "type": "VLAMetric",
1980
+ "window_size": 1
1981
+ },
1982
+ "mixed_precision_dtype": "bf16",
1983
+ "optimizer": {
1984
+ "lr": 0.0001,
1985
+ "type": "AdamW",
1986
+ "weight_decay": 1e-05
1987
+ },
1988
+ "sampler": null,
1989
+ "save_epoch_interval": 1,
1990
+ "save_iter_interval": 5000,
1991
+ "sharding_strategy": "full-shard",
1992
+ "type": "FSDPTrainRunner"
1993
+ },
1994
+ "seed": 42,
1995
+ "train_dataloader": {
1996
+ "dataset": {
1997
+ "dataset_statistics": {
1998
+ "libero_object_no_noops_native": {
1999
+ "actions": {
2000
+ "max": [
2001
+ 0.9375,
2002
+ 0.8919642567634583,
2003
+ 0.9375,
2004
+ 0.17678570747375488,
2005
+ 0.35035714507102966,
2006
+ 0.1810714304447174,
2007
+ 1.0
2008
+ ],
2009
+ "mean": [
2010
+ 0.07127183159196328,
2011
+ 0.13518370972847993,
2012
+ -0.046194642259443225,
2013
+ 0.0011625276173484506,
2014
+ 0.007098307809942585,
2015
+ -0.015082184949062672,
2016
+ 0.46515324845117295
2017
+ ],
2018
+ "min": [
2019
+ -0.8839285969734192,
2020
+ -0.9375,
2021
+ -0.9375,
2022
+ -0.15000000596046448,
2023
+ -0.29035714268684387,
2024
+ -0.32892856001853943,
2025
+ 0.0
2026
+ ],
2027
+ "q01": [
2028
+ -0.5410714149475098,
2029
+ -0.8758928775787354,
2030
+ -0.9375,
2031
+ -0.06964285671710968,
2032
+ -0.11678571254014969,
2033
+ -0.1607142835855484,
2034
+ 0.0
2035
+ ],
2036
+ "q99": [
2037
+ 0.8464285731315613,
2038
+ 0.84375,
2039
+ 0.9375,
2040
+ 0.08142857253551483,
2041
+ 0.14884285390376858,
2042
+ 0.08571428805589676,
2043
+ 1.0
2044
+ ],
2045
+ "std": [
2046
+ 0.26873661352251504,
2047
+ 0.4385525563650988,
2048
+ 0.4474788640468074,
2049
+ 0.024271765379373398,
2050
+ 0.04932243328601301,
2051
+ 0.04218624160534833,
2052
+ 0.4987842257994289
2053
+ ]
2054
+ },
2055
+ "states": {
2056
+ "max": [
2057
+ 0.14580604434013367,
2058
+ 0.3322660028934479,
2059
+ 0.38492220640182495,
2060
+ 3.4003379344940186,
2061
+ 0.7954724431037903,
2062
+ 0.6642153263092041,
2063
+ 0.04104502499103546,
2064
+ -0.0005043679848313332
2065
+ ],
2066
+ "mean": [
2067
+ -0.029590097511622324,
2068
+ -0.007297964757377373,
2069
+ 0.20263691224329605,
2070
+ 3.107347627230808,
2071
+ -0.21535565470621534,
2072
+ -0.11427672217768083,
2073
+ 0.029502615037539585,
2074
+ -0.030672144202923204
2075
+ ],
2076
+ "min": [
2077
+ -0.17652566730976105,
2078
+ -0.28169912099838257,
2079
+ 0.008128181099891663,
2080
+ 2.2889645099639893,
2081
+ -1.8830392360687256,
2082
+ -1.0428133010864258,
2083
+ 0.0006782737909816206,
2084
+ -0.041782498359680176
2085
+ ],
2086
+ "q01": [
2087
+ -0.149256848692894,
2088
+ -0.2592130792140961,
2089
+ 0.009932418912649154,
2090
+ 2.7467152214050294,
2091
+ -1.401689453125,
2092
+ -0.6867975068092346,
2093
+ 0.008699759356677533,
2094
+ -0.04015225619077682
2095
+ ],
2096
+ "q99": [
2097
+ 0.09154553085565566,
2098
+ 0.2911862111091613,
2099
+ 0.3361880838871002,
2100
+ 3.260209331512451,
2101
+ 0.3196117353439328,
2102
+ 0.39586576938629137,
2103
+ 0.03989199593663215,
2104
+ -0.010282181538641466
2105
+ ],
2106
+ "std": [
2107
+ 0.06715783727495737,
2108
+ 0.17582774832397174,
2109
+ 0.07806197322485103,
2110
+ 0.08837916027138361,
2111
+ 0.33845984937441376,
2112
+ 0.2067143161337991,
2113
+ 0.00947681687068548,
2114
+ 0.009059544368504133
2115
+ ]
2116
+ }
2117
+ }
2118
+ },
2119
+ "datasets": [
2120
+ {
2121
+ "action_key": "action",
2122
+ "action_window_size": 16,
2123
+ "data_root_path": "/mnt/data/cpfs/mnt/data/liyinhao/datasets/libero_object_no_noops_lerobotv2.1",
2124
+ "drop_incomplete_action_windows": true,
2125
+ "repeat_to_full_length": false,
2126
+ "statistic_name": "libero_object_no_noops_native",
2127
+ "train_episode_fraction": 1.0,
2128
+ "transforms": [
2129
+ {
2130
+ "embodiment_id": 2,
2131
+ "name_mappings": {
2132
+ "actions": [
2133
+ "actions"
2134
+ ],
2135
+ "observation.state": [
2136
+ "states"
2137
+ ]
2138
+ },
2139
+ "parquet_keys": [
2140
+ "observation.state",
2141
+ "timestamp",
2142
+ "actions",
2143
+ "info",
2144
+ "stats",
2145
+ "action_masks"
2146
+ ],
2147
+ "type": "ProcessParquetInputs",
2148
+ "video_keys": [
2149
+ "observation.images.image",
2150
+ "observation.images.wrist_image"
2151
+ ]
2152
+ },
2153
+ {
2154
+ "action_dim": 132,
2155
+ "action_key": "actions",
2156
+ "clip_norm": true,
2157
+ "norm_type": "quantile",
2158
+ "normalization_epsilon": 0.0,
2159
+ "preserve_input_dtype": true,
2160
+ "state_dim": 132,
2161
+ "state_key": "states",
2162
+ "type": "NormalizeStatesAndActions"
2163
+ },
2164
+ {
2165
+ "action_horizon": 40,
2166
+ "state_dropout_prob": 0.2,
2167
+ "state_history_length": 1,
2168
+ "type": "PrepareStateActionTargets",
2169
+ "valid_action_dim": 7
2170
+ },
2171
+ {
2172
+ "embodiment_tag": "LIBERO_PANDA",
2173
+ "image_key": "images",
2174
+ "output_image_key": "images",
2175
+ "processor_kwargs": {
2176
+ "apply_sincos_state_encoding": false,
2177
+ "clip_outliers": true,
2178
+ "color_jitter_params": {
2179
+ "brightness": 0.3,
2180
+ "contrast": 0.4,
2181
+ "hue": 0.08,
2182
+ "saturation": 0.5
2183
+ },
2184
+ "crop_fraction": 0.95,
2185
+ "embodiment_id_mapping": {
2186
+ "libero_sim": 2
2187
+ },
2188
+ "formalize_language": true,
2189
+ "image_crop_size": [
2190
+ 230,
2191
+ 230
2192
+ ],
2193
+ "image_target_size": [
2194
+ 256,
2195
+ 256
2196
+ ],
2197
+ "max_action_dim": 132,
2198
+ "max_action_horizon": 40,
2199
+ "max_state_dim": 132,
2200
+ "modality_configs": {
2201
+ "libero_sim": {
2202
+ "action": {
2203
+ "delta_indices": [
2204
+ 0,
2205
+ 1,
2206
+ 2,
2207
+ 3,
2208
+ 4,
2209
+ 5,
2210
+ 6,
2211
+ 7,
2212
+ 8,
2213
+ 9,
2214
+ 10,
2215
+ 11,
2216
+ 12,
2217
+ 13,
2218
+ 14,
2219
+ 15
2220
+ ],
2221
+ "modality_keys": [
2222
+ "x",
2223
+ "y",
2224
+ "z",
2225
+ "roll",
2226
+ "pitch",
2227
+ "yaw",
2228
+ "gripper"
2229
+ ]
2230
+ },
2231
+ "state": {
2232
+ "delta_indices": [
2233
+ 0
2234
+ ],
2235
+ "modality_keys": [
2236
+ "x",
2237
+ "y",
2238
+ "z",
2239
+ "roll",
2240
+ "pitch",
2241
+ "yaw",
2242
+ "gripper"
2243
+ ]
2244
+ },
2245
+ "video": {
2246
+ "delta_indices": [
2247
+ 0
2248
+ ],
2249
+ "modality_keys": [
2250
+ "image",
2251
+ "wrist_image"
2252
+ ]
2253
+ }
2254
+ }
2255
+ },
2256
+ "shortest_image_edge": 256,
2257
+ "state_dropout_prob": 0.2,
2258
+ "statistics": {
2259
+ "libero_sim": {
2260
+ "action": {
2261
+ "gripper": {
2262
+ "max": [
2263
+ 1.0
2264
+ ],
2265
+ "mean": [
2266
+ 0.46515324845117295
2267
+ ],
2268
+ "min": [
2269
+ 0.0
2270
+ ],
2271
+ "q01": [
2272
+ 0.0
2273
+ ],
2274
+ "q99": [
2275
+ 1.0
2276
+ ],
2277
+ "std": [
2278
+ 0.4987842257994289
2279
+ ]
2280
+ },
2281
+ "pitch": {
2282
+ "max": [
2283
+ 0.35035714507102966
2284
+ ],
2285
+ "mean": [
2286
+ 0.007098307809942585
2287
+ ],
2288
+ "min": [
2289
+ -0.29035714268684387
2290
+ ],
2291
+ "q01": [
2292
+ -0.11678571254014969
2293
+ ],
2294
+ "q99": [
2295
+ 0.14884285390376858
2296
+ ],
2297
+ "std": [
2298
+ 0.04932243328601301
2299
+ ]
2300
+ },
2301
+ "roll": {
2302
+ "max": [
2303
+ 0.17678570747375488
2304
+ ],
2305
+ "mean": [
2306
+ 0.0011625276173484506
2307
+ ],
2308
+ "min": [
2309
+ -0.15000000596046448
2310
+ ],
2311
+ "q01": [
2312
+ -0.06964285671710968
2313
+ ],
2314
+ "q99": [
2315
+ 0.08142857253551483
2316
+ ],
2317
+ "std": [
2318
+ 0.024271765379373398
2319
+ ]
2320
+ },
2321
+ "x": {
2322
+ "max": [
2323
+ 0.9375
2324
+ ],
2325
+ "mean": [
2326
+ 0.07127183159196328
2327
+ ],
2328
+ "min": [
2329
+ -0.8839285969734192
2330
+ ],
2331
+ "q01": [
2332
+ -0.5410714149475098
2333
+ ],
2334
+ "q99": [
2335
+ 0.8464285731315613
2336
+ ],
2337
+ "std": [
2338
+ 0.26873661352251504
2339
+ ]
2340
+ },
2341
+ "y": {
2342
+ "max": [
2343
+ 0.8919642567634583
2344
+ ],
2345
+ "mean": [
2346
+ 0.13518370972847993
2347
+ ],
2348
+ "min": [
2349
+ -0.9375
2350
+ ],
2351
+ "q01": [
2352
+ -0.8758928775787354
2353
+ ],
2354
+ "q99": [
2355
+ 0.84375
2356
+ ],
2357
+ "std": [
2358
+ 0.4385525563650988
2359
+ ]
2360
+ },
2361
+ "yaw": {
2362
+ "max": [
2363
+ 0.1810714304447174
2364
+ ],
2365
+ "mean": [
2366
+ -0.015082184949062672
2367
+ ],
2368
+ "min": [
2369
+ -0.32892856001853943
2370
+ ],
2371
+ "q01": [
2372
+ -0.1607142835855484
2373
+ ],
2374
+ "q99": [
2375
+ 0.08571428805589676
2376
+ ],
2377
+ "std": [
2378
+ 0.04218624160534833
2379
+ ]
2380
+ },
2381
+ "z": {
2382
+ "max": [
2383
+ 0.9375
2384
+ ],
2385
+ "mean": [
2386
+ -0.046194642259443225
2387
+ ],
2388
+ "min": [
2389
+ -0.9375
2390
+ ],
2391
+ "q01": [
2392
+ -0.9375
2393
+ ],
2394
+ "q99": [
2395
+ 0.9375
2396
+ ],
2397
+ "std": [
2398
+ 0.4474788640468074
2399
+ ]
2400
+ }
2401
+ },
2402
+ "state": {
2403
+ "gripper": {
2404
+ "max": [
2405
+ 0.04104502499103546,
2406
+ -0.0005043679848313332
2407
+ ],
2408
+ "mean": [
2409
+ 0.029502615037539585,
2410
+ -0.030672144202923204
2411
+ ],
2412
+ "min": [
2413
+ 0.0006782737909816206,
2414
+ -0.041782498359680176
2415
+ ],
2416
+ "q01": [
2417
+ 0.008699759356677533,
2418
+ -0.04015225619077682
2419
+ ],
2420
+ "q99": [
2421
+ 0.03989199593663215,
2422
+ -0.010282181538641466
2423
+ ],
2424
+ "std": [
2425
+ 0.00947681687068548,
2426
+ 0.009059544368504133
2427
+ ]
2428
+ },
2429
+ "pitch": {
2430
+ "max": [
2431
+ 0.7954724431037903
2432
+ ],
2433
+ "mean": [
2434
+ -0.21535565470621534
2435
+ ],
2436
+ "min": [
2437
+ -1.8830392360687256
2438
+ ],
2439
+ "q01": [
2440
+ -1.401689453125
2441
+ ],
2442
+ "q99": [
2443
+ 0.3196117353439328
2444
+ ],
2445
+ "std": [
2446
+ 0.33845984937441376
2447
+ ]
2448
+ },
2449
+ "roll": {
2450
+ "max": [
2451
+ 3.4003379344940186
2452
+ ],
2453
+ "mean": [
2454
+ 3.107347627230808
2455
+ ],
2456
+ "min": [
2457
+ 2.2889645099639893
2458
+ ],
2459
+ "q01": [
2460
+ 2.7467152214050294
2461
+ ],
2462
+ "q99": [
2463
+ 3.260209331512451
2464
+ ],
2465
+ "std": [
2466
+ 0.08837916027138361
2467
+ ]
2468
+ },
2469
+ "x": {
2470
+ "max": [
2471
+ 0.14580604434013367
2472
+ ],
2473
+ "mean": [
2474
+ -0.029590097511622324
2475
+ ],
2476
+ "min": [
2477
+ -0.17652566730976105
2478
+ ],
2479
+ "q01": [
2480
+ -0.149256848692894
2481
+ ],
2482
+ "q99": [
2483
+ 0.09154553085565566
2484
+ ],
2485
+ "std": [
2486
+ 0.06715783727495737
2487
+ ]
2488
+ },
2489
+ "y": {
2490
+ "max": [
2491
+ 0.3322660028934479
2492
+ ],
2493
+ "mean": [
2494
+ -0.007297964757377373
2495
+ ],
2496
+ "min": [
2497
+ -0.28169912099838257
2498
+ ],
2499
+ "q01": [
2500
+ -0.2592130792140961
2501
+ ],
2502
+ "q99": [
2503
+ 0.2911862111091613
2504
+ ],
2505
+ "std": [
2506
+ 0.17582774832397174
2507
+ ]
2508
+ },
2509
+ "yaw": {
2510
+ "max": [
2511
+ 0.6642153263092041
2512
+ ],
2513
+ "mean": [
2514
+ -0.11427672217768083
2515
+ ],
2516
+ "min": [
2517
+ -1.0428133010864258
2518
+ ],
2519
+ "q01": [
2520
+ -0.6867975068092346
2521
+ ],
2522
+ "q99": [
2523
+ 0.39586576938629137
2524
+ ],
2525
+ "std": [
2526
+ 0.2067143161337991
2527
+ ]
2528
+ },
2529
+ "z": {
2530
+ "max": [
2531
+ 0.38492220640182495
2532
+ ],
2533
+ "mean": [
2534
+ 0.20263691224329605
2535
+ ],
2536
+ "min": [
2537
+ 0.008128181099891663
2538
+ ],
2539
+ "q01": [
2540
+ 0.009932418912649154
2541
+ ],
2542
+ "q99": [
2543
+ 0.3361880838871002
2544
+ ],
2545
+ "std": [
2546
+ 0.07806197322485103
2547
+ ]
2548
+ }
2549
+ }
2550
+ }
2551
+ },
2552
+ "use_albumentations": true,
2553
+ "use_percentiles": true,
2554
+ "use_relative_action": true
2555
+ },
2556
+ "train_mode": true,
2557
+ "type": "GrootN17ImageAugmentation"
2558
+ },
2559
+ {
2560
+ "image_mean": [
2561
+ 0.5,
2562
+ 0.5,
2563
+ 0.5
2564
+ ],
2565
+ "image_std": [
2566
+ 0.5,
2567
+ 0.5,
2568
+ 0.5
2569
+ ],
2570
+ "img_key": "images",
2571
+ "merge_size": 2,
2572
+ "patch_size": 16,
2573
+ "size": {
2574
+ "longest_edge": 16777216,
2575
+ "shortest_edge": 65536
2576
+ },
2577
+ "temporal_patch_size": 2,
2578
+ "to_tensor": true,
2579
+ "type": "QWen2VLImageTransform"
2580
+ },
2581
+ {
2582
+ "add_assistant_stub": false,
2583
+ "add_system": false,
2584
+ "attention_mask_dtype": "int64",
2585
+ "image_grid_thw_key": "image_grid_thw",
2586
+ "image_merge_size": 2,
2587
+ "image_tag_template": "",
2588
+ "img_context_token": "<|image_pad|>",
2589
+ "img_end": "<|vision_end|>",
2590
+ "img_start": "<|vision_start|>",
2591
+ "img_tokens_source": "from_image_grid_thw",
2592
+ "lowercase_task_description": true,
2593
+ "max_len": 180,
2594
+ "output_keys": [
2595
+ "lang_tokens",
2596
+ "lang_masks",
2597
+ "images",
2598
+ "image_grid_thw",
2599
+ "states",
2600
+ "actions",
2601
+ "action_masks",
2602
+ "embodiment_ids",
2603
+ "sample_weight"
2604
+ ],
2605
+ "padding_side": "left",
2606
+ "strip_task_punctuation": true,
2607
+ "task_pos": "after_images",
2608
+ "tokenizer": {
2609
+ "model_path": "fluxvla/models/third_party_models/qwen3_tokenizer",
2610
+ "padding_side": "left",
2611
+ "trust_remote_code": false,
2612
+ "type": "PretrainedTokenizer"
2613
+ },
2614
+ "truncate": false,
2615
+ "type": "ProcessPromptsWithImage",
2616
+ "use_eos_as_pad": false
2617
+ }
2618
+ ],
2619
+ "type": "ParquetDataset",
2620
+ "use_delta": false,
2621
+ "window_start_idx": 0
2622
+ }
2623
+ ],
2624
+ "name_mappings": {
2625
+ "action": [
2626
+ "actions"
2627
+ ],
2628
+ "observation.state": [
2629
+ "states"
2630
+ ]
2631
+ },
2632
+ "reshuffle_each_epoch": true,
2633
+ "seed": 42,
2634
+ "shuffle": true,
2635
+ "statistic_keys": [
2636
+ "observation.state",
2637
+ "timestamp",
2638
+ "action"
2639
+ ],
2640
+ "statistic_name": "libero_object_no_noops_native",
2641
+ "type": "DistributedRepeatingDataset"
2642
+ },
2643
+ "per_device_batch_size": 40,
2644
+ "per_device_num_workers": 4
2645
+ }
2646
+ }
gr00t_n17_native_libero_object_full_finetune_bs64/config.yaml ADDED
@@ -0,0 +1,1917 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ _ACTIVE_TRACKERS:
2
+ - jsonl
3
+ _DATASET_NAME: libero_object_no_noops_lerobotv2.1
4
+ _LIBERO_DATA_ROOT: ./datasets/libero_object_no_noops_lerobotv2.1
5
+ _N17_ACTION_LAYOUT:
6
+ - - x
7
+ - 1
8
+ - - y
9
+ - 1
10
+ - - z
11
+ - 1
12
+ - - roll
13
+ - 1
14
+ - - pitch
15
+ - 1
16
+ - - yaw
17
+ - 1
18
+ - - gripper
19
+ - 1
20
+ _N17_FLAT_STATISTICS:
21
+ action:
22
+ max:
23
+ - 0.9375
24
+ - 0.8919642567634583
25
+ - 0.9375
26
+ - 0.17678570747375488
27
+ - 0.35035714507102966
28
+ - 0.1810714304447174
29
+ - 1.0
30
+ mean:
31
+ - 0.07127183159196328
32
+ - 0.13518370972847993
33
+ - -0.046194642259443225
34
+ - 0.0011625276173484506
35
+ - 0.007098307809942585
36
+ - -0.015082184949062672
37
+ - 0.46515324845117295
38
+ min:
39
+ - -0.8839285969734192
40
+ - -0.9375
41
+ - -0.9375
42
+ - -0.15000000596046448
43
+ - -0.29035714268684387
44
+ - -0.32892856001853943
45
+ - 0.0
46
+ q01:
47
+ - -0.5410714149475098
48
+ - -0.8758928775787354
49
+ - -0.9375
50
+ - -0.06964285671710968
51
+ - -0.11678571254014969
52
+ - -0.1607142835855484
53
+ - 0.0
54
+ q99:
55
+ - 0.8464285731315613
56
+ - 0.84375
57
+ - 0.9375
58
+ - 0.08142857253551483
59
+ - 0.14884285390376858
60
+ - 0.08571428805589676
61
+ - 1.0
62
+ std:
63
+ - 0.26873661352251504
64
+ - 0.4385525563650988
65
+ - 0.4474788640468074
66
+ - 0.024271765379373398
67
+ - 0.04932243328601301
68
+ - 0.04218624160534833
69
+ - 0.4987842257994289
70
+ state:
71
+ max:
72
+ - 0.14580604434013367
73
+ - 0.3322660028934479
74
+ - 0.38492220640182495
75
+ - 3.4003379344940186
76
+ - 0.7954724431037903
77
+ - 0.6642153263092041
78
+ - 0.04104502499103546
79
+ - -0.0005043679848313332
80
+ mean:
81
+ - -0.029590097511622324
82
+ - -0.007297964757377373
83
+ - 0.20263691224329605
84
+ - 3.107347627230808
85
+ - -0.21535565470621534
86
+ - -0.11427672217768083
87
+ - 0.029502615037539585
88
+ - -0.030672144202923204
89
+ min:
90
+ - -0.17652566730976105
91
+ - -0.28169912099838257
92
+ - 0.008128181099891663
93
+ - 2.2889645099639893
94
+ - -1.8830392360687256
95
+ - -1.0428133010864258
96
+ - 0.0006782737909816206
97
+ - -0.041782498359680176
98
+ q01:
99
+ - -0.149256848692894
100
+ - -0.2592130792140961
101
+ - 0.009932418912649154
102
+ - 2.7467152214050294
103
+ - -1.401689453125
104
+ - -0.6867975068092346
105
+ - 0.008699759356677533
106
+ - -0.04015225619077682
107
+ q99:
108
+ - 0.09154553085565566
109
+ - 0.2911862111091613
110
+ - 0.3361880838871002
111
+ - 3.260209331512451
112
+ - 0.3196117353439328
113
+ - 0.39586576938629137
114
+ - 0.03989199593663215
115
+ - -0.010282181538641466
116
+ std:
117
+ - 0.06715783727495737
118
+ - 0.17582774832397174
119
+ - 0.07806197322485103
120
+ - 0.08837916027138361
121
+ - 0.33845984937441376
122
+ - 0.2067143161337991
123
+ - 0.00947681687068548
124
+ - 0.009059544368504133
125
+ _N17_INIT_CKPT: ./checkpoints/GR00T-N1.7-3B
126
+ _N17_MODALITY_CONFIGS:
127
+ libero_sim:
128
+ action:
129
+ delta_indices:
130
+ - 0
131
+ - 1
132
+ - 2
133
+ - 3
134
+ - 4
135
+ - 5
136
+ - 6
137
+ - 7
138
+ - 8
139
+ - 9
140
+ - 10
141
+ - 11
142
+ - 12
143
+ - 13
144
+ - 14
145
+ - 15
146
+ modality_keys:
147
+ - x
148
+ - y
149
+ - z
150
+ - roll
151
+ - pitch
152
+ - yaw
153
+ - gripper
154
+ state:
155
+ delta_indices:
156
+ - 0
157
+ modality_keys:
158
+ - x
159
+ - y
160
+ - z
161
+ - roll
162
+ - pitch
163
+ - yaw
164
+ - gripper
165
+ video:
166
+ delta_indices:
167
+ - 0
168
+ modality_keys:
169
+ - image
170
+ - wrist_image
171
+ _N17_STATE_LAYOUT:
172
+ - - x
173
+ - 1
174
+ - - y
175
+ - 1
176
+ - - z
177
+ - 1
178
+ - - roll
179
+ - 1
180
+ - - pitch
181
+ - 1
182
+ - - yaw
183
+ - 1
184
+ - - gripper
185
+ - 2
186
+ _N17_STATISTICS:
187
+ libero_sim:
188
+ action:
189
+ gripper:
190
+ max:
191
+ - 1.0
192
+ mean:
193
+ - 0.46515324845117295
194
+ min:
195
+ - 0.0
196
+ q01:
197
+ - 0.0
198
+ q99:
199
+ - 1.0
200
+ std:
201
+ - 0.4987842257994289
202
+ pitch:
203
+ max:
204
+ - 0.35035714507102966
205
+ mean:
206
+ - 0.007098307809942585
207
+ min:
208
+ - -0.29035714268684387
209
+ q01:
210
+ - -0.11678571254014969
211
+ q99:
212
+ - 0.14884285390376858
213
+ std:
214
+ - 0.04932243328601301
215
+ roll:
216
+ max:
217
+ - 0.17678570747375488
218
+ mean:
219
+ - 0.0011625276173484506
220
+ min:
221
+ - -0.15000000596046448
222
+ q01:
223
+ - -0.06964285671710968
224
+ q99:
225
+ - 0.08142857253551483
226
+ std:
227
+ - 0.024271765379373398
228
+ x:
229
+ max:
230
+ - 0.9375
231
+ mean:
232
+ - 0.07127183159196328
233
+ min:
234
+ - -0.8839285969734192
235
+ q01:
236
+ - -0.5410714149475098
237
+ q99:
238
+ - 0.8464285731315613
239
+ std:
240
+ - 0.26873661352251504
241
+ y:
242
+ max:
243
+ - 0.8919642567634583
244
+ mean:
245
+ - 0.13518370972847993
246
+ min:
247
+ - -0.9375
248
+ q01:
249
+ - -0.8758928775787354
250
+ q99:
251
+ - 0.84375
252
+ std:
253
+ - 0.4385525563650988
254
+ yaw:
255
+ max:
256
+ - 0.1810714304447174
257
+ mean:
258
+ - -0.015082184949062672
259
+ min:
260
+ - -0.32892856001853943
261
+ q01:
262
+ - -0.1607142835855484
263
+ q99:
264
+ - 0.08571428805589676
265
+ std:
266
+ - 0.04218624160534833
267
+ z:
268
+ max:
269
+ - 0.9375
270
+ mean:
271
+ - -0.046194642259443225
272
+ min:
273
+ - -0.9375
274
+ q01:
275
+ - -0.9375
276
+ q99:
277
+ - 0.9375
278
+ std:
279
+ - 0.4474788640468074
280
+ state:
281
+ gripper:
282
+ max:
283
+ - 0.04104502499103546
284
+ - -0.0005043679848313332
285
+ mean:
286
+ - 0.029502615037539585
287
+ - -0.030672144202923204
288
+ min:
289
+ - 0.0006782737909816206
290
+ - -0.041782498359680176
291
+ q01:
292
+ - 0.008699759356677533
293
+ - -0.04015225619077682
294
+ q99:
295
+ - 0.03989199593663215
296
+ - -0.010282181538641466
297
+ std:
298
+ - 0.00947681687068548
299
+ - 0.009059544368504133
300
+ pitch:
301
+ max:
302
+ - 0.7954724431037903
303
+ mean:
304
+ - -0.21535565470621534
305
+ min:
306
+ - -1.8830392360687256
307
+ q01:
308
+ - -1.401689453125
309
+ q99:
310
+ - 0.3196117353439328
311
+ std:
312
+ - 0.33845984937441376
313
+ roll:
314
+ max:
315
+ - 3.4003379344940186
316
+ mean:
317
+ - 3.107347627230808
318
+ min:
319
+ - 2.2889645099639893
320
+ q01:
321
+ - 2.7467152214050294
322
+ q99:
323
+ - 3.260209331512451
324
+ std:
325
+ - 0.08837916027138361
326
+ x:
327
+ max:
328
+ - 0.14580604434013367
329
+ mean:
330
+ - -0.029590097511622324
331
+ min:
332
+ - -0.17652566730976105
333
+ q01:
334
+ - -0.149256848692894
335
+ q99:
336
+ - 0.09154553085565566
337
+ std:
338
+ - 0.06715783727495737
339
+ y:
340
+ max:
341
+ - 0.3322660028934479
342
+ mean:
343
+ - -0.007297964757377373
344
+ min:
345
+ - -0.28169912099838257
346
+ q01:
347
+ - -0.2592130792140961
348
+ q99:
349
+ - 0.2911862111091613
350
+ std:
351
+ - 0.17582774832397174
352
+ yaw:
353
+ max:
354
+ - 0.6642153263092041
355
+ mean:
356
+ - -0.11427672217768083
357
+ min:
358
+ - -1.0428133010864258
359
+ q01:
360
+ - -0.6867975068092346
361
+ q99:
362
+ - 0.39586576938629137
363
+ std:
364
+ - 0.2067143161337991
365
+ z:
366
+ max:
367
+ - 0.38492220640182495
368
+ mean:
369
+ - 0.20263691224329605
370
+ min:
371
+ - 0.008128181099891663
372
+ q01:
373
+ - 0.009932418912649154
374
+ q99:
375
+ - 0.3361880838871002
376
+ std:
377
+ - 0.07806197322485103
378
+ _PROCESSOR_KWARGS:
379
+ apply_sincos_state_encoding: false
380
+ clip_outliers: true
381
+ color_jitter_params:
382
+ brightness: 0.3
383
+ contrast: 0.4
384
+ hue: 0.08
385
+ saturation: 0.5
386
+ crop_fraction: null
387
+ embodiment_id_mapping:
388
+ libero_sim: 2
389
+ formalize_language: true
390
+ image_crop_size:
391
+ - 230
392
+ - 230
393
+ image_target_size:
394
+ - 256
395
+ - 256
396
+ max_action_dim: 132
397
+ max_action_horizon: 40
398
+ max_state_dim: 132
399
+ modality_configs:
400
+ libero_sim:
401
+ action:
402
+ delta_indices:
403
+ - 0
404
+ - 1
405
+ - 2
406
+ - 3
407
+ - 4
408
+ - 5
409
+ - 6
410
+ - 7
411
+ - 8
412
+ - 9
413
+ - 10
414
+ - 11
415
+ - 12
416
+ - 13
417
+ - 14
418
+ - 15
419
+ modality_keys:
420
+ - x
421
+ - y
422
+ - z
423
+ - roll
424
+ - pitch
425
+ - yaw
426
+ - gripper
427
+ state:
428
+ delta_indices:
429
+ - 0
430
+ modality_keys:
431
+ - x
432
+ - y
433
+ - z
434
+ - roll
435
+ - pitch
436
+ - yaw
437
+ - gripper
438
+ video:
439
+ delta_indices:
440
+ - 0
441
+ modality_keys:
442
+ - image
443
+ - wrist_image
444
+ shortest_image_edge: null
445
+ state_dropout_prob: 0.2
446
+ statistics:
447
+ libero_sim:
448
+ action:
449
+ gripper:
450
+ max:
451
+ - 1.0
452
+ mean:
453
+ - 0.46515324845117295
454
+ min:
455
+ - 0.0
456
+ q01:
457
+ - 0.0
458
+ q99:
459
+ - 1.0
460
+ std:
461
+ - 0.4987842257994289
462
+ pitch:
463
+ max:
464
+ - 0.35035714507102966
465
+ mean:
466
+ - 0.007098307809942585
467
+ min:
468
+ - -0.29035714268684387
469
+ q01:
470
+ - -0.11678571254014969
471
+ q99:
472
+ - 0.14884285390376858
473
+ std:
474
+ - 0.04932243328601301
475
+ roll:
476
+ max:
477
+ - 0.17678570747375488
478
+ mean:
479
+ - 0.0011625276173484506
480
+ min:
481
+ - -0.15000000596046448
482
+ q01:
483
+ - -0.06964285671710968
484
+ q99:
485
+ - 0.08142857253551483
486
+ std:
487
+ - 0.024271765379373398
488
+ x:
489
+ max:
490
+ - 0.9375
491
+ mean:
492
+ - 0.07127183159196328
493
+ min:
494
+ - -0.8839285969734192
495
+ q01:
496
+ - -0.5410714149475098
497
+ q99:
498
+ - 0.8464285731315613
499
+ std:
500
+ - 0.26873661352251504
501
+ y:
502
+ max:
503
+ - 0.8919642567634583
504
+ mean:
505
+ - 0.13518370972847993
506
+ min:
507
+ - -0.9375
508
+ q01:
509
+ - -0.8758928775787354
510
+ q99:
511
+ - 0.84375
512
+ std:
513
+ - 0.4385525563650988
514
+ yaw:
515
+ max:
516
+ - 0.1810714304447174
517
+ mean:
518
+ - -0.015082184949062672
519
+ min:
520
+ - -0.32892856001853943
521
+ q01:
522
+ - -0.1607142835855484
523
+ q99:
524
+ - 0.08571428805589676
525
+ std:
526
+ - 0.04218624160534833
527
+ z:
528
+ max:
529
+ - 0.9375
530
+ mean:
531
+ - -0.046194642259443225
532
+ min:
533
+ - -0.9375
534
+ q01:
535
+ - -0.9375
536
+ q99:
537
+ - 0.9375
538
+ std:
539
+ - 0.4474788640468074
540
+ state:
541
+ gripper:
542
+ max:
543
+ - 0.04104502499103546
544
+ - -0.0005043679848313332
545
+ mean:
546
+ - 0.029502615037539585
547
+ - -0.030672144202923204
548
+ min:
549
+ - 0.0006782737909816206
550
+ - -0.041782498359680176
551
+ q01:
552
+ - 0.008699759356677533
553
+ - -0.04015225619077682
554
+ q99:
555
+ - 0.03989199593663215
556
+ - -0.010282181538641466
557
+ std:
558
+ - 0.00947681687068548
559
+ - 0.009059544368504133
560
+ pitch:
561
+ max:
562
+ - 0.7954724431037903
563
+ mean:
564
+ - -0.21535565470621534
565
+ min:
566
+ - -1.8830392360687256
567
+ q01:
568
+ - -1.401689453125
569
+ q99:
570
+ - 0.3196117353439328
571
+ std:
572
+ - 0.33845984937441376
573
+ roll:
574
+ max:
575
+ - 3.4003379344940186
576
+ mean:
577
+ - 3.107347627230808
578
+ min:
579
+ - 2.2889645099639893
580
+ q01:
581
+ - 2.7467152214050294
582
+ q99:
583
+ - 3.260209331512451
584
+ std:
585
+ - 0.08837916027138361
586
+ x:
587
+ max:
588
+ - 0.14580604434013367
589
+ mean:
590
+ - -0.029590097511622324
591
+ min:
592
+ - -0.17652566730976105
593
+ q01:
594
+ - -0.149256848692894
595
+ q99:
596
+ - 0.09154553085565566
597
+ std:
598
+ - 0.06715783727495737
599
+ y:
600
+ max:
601
+ - 0.3322660028934479
602
+ mean:
603
+ - -0.007297964757377373
604
+ min:
605
+ - -0.28169912099838257
606
+ q01:
607
+ - -0.2592130792140961
608
+ q99:
609
+ - 0.2911862111091613
610
+ std:
611
+ - 0.17582774832397174
612
+ yaw:
613
+ max:
614
+ - 0.6642153263092041
615
+ mean:
616
+ - -0.11427672217768083
617
+ min:
618
+ - -1.0428133010864258
619
+ q01:
620
+ - -0.6867975068092346
621
+ q99:
622
+ - 0.39586576938629137
623
+ std:
624
+ - 0.2067143161337991
625
+ z:
626
+ max:
627
+ - 0.38492220640182495
628
+ mean:
629
+ - 0.20263691224329605
630
+ min:
631
+ - 0.008128181099891663
632
+ q01:
633
+ - 0.009932418912649154
634
+ q99:
635
+ - 0.3361880838871002
636
+ std:
637
+ - 0.07806197322485103
638
+ use_albumentations: true
639
+ use_percentiles: true
640
+ use_relative_action: true
641
+ _QWEN3_VL_CONFIG:
642
+ architectures:
643
+ - Qwen3VLForConditionalGeneration
644
+ image_token_id: 151655
645
+ text_config:
646
+ attention_bias: false
647
+ attention_dropout: 0.0
648
+ head_dim: 128
649
+ hidden_act: silu
650
+ hidden_size: 2048
651
+ initializer_range: 0.02
652
+ intermediate_size: 6144
653
+ max_position_embeddings: 262144
654
+ model_type: qwen3_vl_text
655
+ num_attention_heads: 16
656
+ num_hidden_layers: 28
657
+ num_key_value_heads: 8
658
+ rms_norm_eps: 1.0e-06
659
+ rope_parameters:
660
+ mrope_interleaved: true
661
+ mrope_section:
662
+ - 24
663
+ - 20
664
+ - 20
665
+ rope_theta: 5000000.0
666
+ rope_type: default
667
+ use_cache: false
668
+ vocab_size: 151936
669
+ tie_word_embeddings: false
670
+ video_token_id: 151656
671
+ vision_config:
672
+ deepstack_visual_indexes:
673
+ - 5
674
+ - 11
675
+ - 17
676
+ depth: 24
677
+ hidden_act: gelu_pytorch_tanh
678
+ hidden_size: 1024
679
+ in_channels: 3
680
+ initializer_range: 0.02
681
+ intermediate_size: 4096
682
+ model_type: qwen3_vl_vision
683
+ num_heads: 16
684
+ num_position_embeddings: 2304
685
+ out_hidden_size: 2048
686
+ patch_size: 16
687
+ spatial_merge_size: 2
688
+ temporal_patch_size: 2
689
+ vision_end_token_id: 151653
690
+ vision_start_token_id: 151652
691
+ _QWEN_TOKENIZER_PATH: fluxvla/models/third_party_models/qwen3_tokenizer
692
+ _STATISTIC_NAME: libero_object_no_noops_native
693
+ _SUITE: libero_object
694
+ eval:
695
+ dataset:
696
+ transforms:
697
+ - embodiment_id: 2
698
+ img_keys:
699
+ - agentview_image
700
+ - robot0_eye_in_hand_image
701
+ type: ProcessLiberoEvalInputs
702
+ - norm_type: null
703
+ out_key: states
704
+ type: LiberoProprioFromInputs
705
+ - action_key: null
706
+ clip_norm: true
707
+ norm_type: quantile
708
+ normalization_epsilon: 0.0
709
+ preserve_input_dtype: true
710
+ state_dim: 132
711
+ state_key: states
712
+ statistics_key: norm_stats
713
+ type: NormalizeStatesAndActions
714
+ - action_horizon: 40
715
+ state_dropout_prob: 0.0
716
+ state_history_length: 1
717
+ type: PrepareStateActionTargets
718
+ valid_action_dim: 7
719
+ - embodiment_tag: LIBERO_PANDA
720
+ image_key: pixel_values
721
+ output_image_key: pixel_values
722
+ processor_kwargs:
723
+ apply_sincos_state_encoding: false
724
+ clip_outliers: true
725
+ color_jitter_params:
726
+ brightness: 0.3
727
+ contrast: 0.4
728
+ hue: 0.08
729
+ saturation: 0.5
730
+ crop_fraction: 0.95
731
+ embodiment_id_mapping:
732
+ libero_sim: 2
733
+ formalize_language: true
734
+ image_crop_size:
735
+ - 230
736
+ - 230
737
+ image_target_size:
738
+ - 256
739
+ - 256
740
+ max_action_dim: 132
741
+ max_action_horizon: 40
742
+ max_state_dim: 132
743
+ modality_configs:
744
+ libero_sim:
745
+ action:
746
+ delta_indices:
747
+ - 0
748
+ - 1
749
+ - 2
750
+ - 3
751
+ - 4
752
+ - 5
753
+ - 6
754
+ - 7
755
+ - 8
756
+ - 9
757
+ - 10
758
+ - 11
759
+ - 12
760
+ - 13
761
+ - 14
762
+ - 15
763
+ modality_keys:
764
+ - x
765
+ - y
766
+ - z
767
+ - roll
768
+ - pitch
769
+ - yaw
770
+ - gripper
771
+ state:
772
+ delta_indices:
773
+ - 0
774
+ modality_keys:
775
+ - x
776
+ - y
777
+ - z
778
+ - roll
779
+ - pitch
780
+ - yaw
781
+ - gripper
782
+ video:
783
+ delta_indices:
784
+ - 0
785
+ modality_keys:
786
+ - image
787
+ - wrist_image
788
+ shortest_image_edge: 256
789
+ state_dropout_prob: 0.2
790
+ statistics:
791
+ libero_sim:
792
+ action:
793
+ gripper:
794
+ max:
795
+ - 1.0
796
+ mean:
797
+ - 0.46515324845117295
798
+ min:
799
+ - 0.0
800
+ q01:
801
+ - 0.0
802
+ q99:
803
+ - 1.0
804
+ std:
805
+ - 0.4987842257994289
806
+ pitch:
807
+ max:
808
+ - 0.35035714507102966
809
+ mean:
810
+ - 0.007098307809942585
811
+ min:
812
+ - -0.29035714268684387
813
+ q01:
814
+ - -0.11678571254014969
815
+ q99:
816
+ - 0.14884285390376858
817
+ std:
818
+ - 0.04932243328601301
819
+ roll:
820
+ max:
821
+ - 0.17678570747375488
822
+ mean:
823
+ - 0.0011625276173484506
824
+ min:
825
+ - -0.15000000596046448
826
+ q01:
827
+ - -0.06964285671710968
828
+ q99:
829
+ - 0.08142857253551483
830
+ std:
831
+ - 0.024271765379373398
832
+ x:
833
+ max:
834
+ - 0.9375
835
+ mean:
836
+ - 0.07127183159196328
837
+ min:
838
+ - -0.8839285969734192
839
+ q01:
840
+ - -0.5410714149475098
841
+ q99:
842
+ - 0.8464285731315613
843
+ std:
844
+ - 0.26873661352251504
845
+ y:
846
+ max:
847
+ - 0.8919642567634583
848
+ mean:
849
+ - 0.13518370972847993
850
+ min:
851
+ - -0.9375
852
+ q01:
853
+ - -0.8758928775787354
854
+ q99:
855
+ - 0.84375
856
+ std:
857
+ - 0.4385525563650988
858
+ yaw:
859
+ max:
860
+ - 0.1810714304447174
861
+ mean:
862
+ - -0.015082184949062672
863
+ min:
864
+ - -0.32892856001853943
865
+ q01:
866
+ - -0.1607142835855484
867
+ q99:
868
+ - 0.08571428805589676
869
+ std:
870
+ - 0.04218624160534833
871
+ z:
872
+ max:
873
+ - 0.9375
874
+ mean:
875
+ - -0.046194642259443225
876
+ min:
877
+ - -0.9375
878
+ q01:
879
+ - -0.9375
880
+ q99:
881
+ - 0.9375
882
+ std:
883
+ - 0.4474788640468074
884
+ state:
885
+ gripper:
886
+ max:
887
+ - 0.04104502499103546
888
+ - -0.0005043679848313332
889
+ mean:
890
+ - 0.029502615037539585
891
+ - -0.030672144202923204
892
+ min:
893
+ - 0.0006782737909816206
894
+ - -0.041782498359680176
895
+ q01:
896
+ - 0.008699759356677533
897
+ - -0.04015225619077682
898
+ q99:
899
+ - 0.03989199593663215
900
+ - -0.010282181538641466
901
+ std:
902
+ - 0.00947681687068548
903
+ - 0.009059544368504133
904
+ pitch:
905
+ max:
906
+ - 0.7954724431037903
907
+ mean:
908
+ - -0.21535565470621534
909
+ min:
910
+ - -1.8830392360687256
911
+ q01:
912
+ - -1.401689453125
913
+ q99:
914
+ - 0.3196117353439328
915
+ std:
916
+ - 0.33845984937441376
917
+ roll:
918
+ max:
919
+ - 3.4003379344940186
920
+ mean:
921
+ - 3.107347627230808
922
+ min:
923
+ - 2.2889645099639893
924
+ q01:
925
+ - 2.7467152214050294
926
+ q99:
927
+ - 3.260209331512451
928
+ std:
929
+ - 0.08837916027138361
930
+ x:
931
+ max:
932
+ - 0.14580604434013367
933
+ mean:
934
+ - -0.029590097511622324
935
+ min:
936
+ - -0.17652566730976105
937
+ q01:
938
+ - -0.149256848692894
939
+ q99:
940
+ - 0.09154553085565566
941
+ std:
942
+ - 0.06715783727495737
943
+ y:
944
+ max:
945
+ - 0.3322660028934479
946
+ mean:
947
+ - -0.007297964757377373
948
+ min:
949
+ - -0.28169912099838257
950
+ q01:
951
+ - -0.2592130792140961
952
+ q99:
953
+ - 0.2911862111091613
954
+ std:
955
+ - 0.17582774832397174
956
+ yaw:
957
+ max:
958
+ - 0.6642153263092041
959
+ mean:
960
+ - -0.11427672217768083
961
+ min:
962
+ - -1.0428133010864258
963
+ q01:
964
+ - -0.6867975068092346
965
+ q99:
966
+ - 0.39586576938629137
967
+ std:
968
+ - 0.2067143161337991
969
+ z:
970
+ max:
971
+ - 0.38492220640182495
972
+ mean:
973
+ - 0.20263691224329605
974
+ min:
975
+ - 0.008128181099891663
976
+ q01:
977
+ - 0.009932418912649154
978
+ q99:
979
+ - 0.3361880838871002
980
+ std:
981
+ - 0.07806197322485103
982
+ use_albumentations: true
983
+ use_percentiles: true
984
+ use_relative_action: true
985
+ train_mode: false
986
+ type: GrootN17ImageAugmentation
987
+ - image_mean:
988
+ - 0.5
989
+ - 0.5
990
+ - 0.5
991
+ image_std:
992
+ - 0.5
993
+ - 0.5
994
+ - 0.5
995
+ img_key: pixel_values
996
+ merge_size: 2
997
+ patch_size: 16
998
+ size:
999
+ longest_edge: 16777216
1000
+ shortest_edge: 65536
1001
+ temporal_patch_size: 2
1002
+ to_tensor: true
1003
+ type: QWen2VLImageTransform
1004
+ - add_assistant_stub: false
1005
+ add_system: false
1006
+ attention_mask_dtype: int64
1007
+ image_grid_thw_key: image_grid_thw
1008
+ image_merge_size: 2
1009
+ image_tag_template: ''
1010
+ img_context_token: <|image_pad|>
1011
+ img_end: <|vision_end|>
1012
+ img_start: <|vision_start|>
1013
+ img_tokens_source: from_image_grid_thw
1014
+ lowercase_task_description: true
1015
+ max_len: 180
1016
+ output_keys:
1017
+ - lang_tokens
1018
+ - lang_masks
1019
+ - pixel_values
1020
+ - img_masks
1021
+ - image_grid_thw
1022
+ - states
1023
+ - embodiment_ids
1024
+ - replay_img
1025
+ padding_side: left
1026
+ strip_task_punctuation: true
1027
+ task_pos: after_images
1028
+ tokenizer:
1029
+ model_path: fluxvla/models/third_party_models/qwen3_tokenizer
1030
+ padding_side: left
1031
+ trust_remote_code: false
1032
+ type: PretrainedTokenizer
1033
+ truncate: false
1034
+ type: ProcessPromptsWithImage
1035
+ use_eos_as_pad: false
1036
+ type: LiberoParquetEvalDataset
1037
+ denormalize_action:
1038
+ action_dim: 7
1039
+ denorm_action: false
1040
+ type: DenormalizeLiberoAction
1041
+ eval_chunk_size: 8
1042
+ eval_shard_strategy: episode
1043
+ inference_seed: 42
1044
+ max_steps: 720
1045
+ model_family: groot_n17_native
1046
+ norm_stats_key: libero_object_no_noops_native
1047
+ num_steps_wait: 0
1048
+ num_trials_per_task: 50
1049
+ preprocess_every_step: false
1050
+ result_output_dir: work_dirs/groot_n17_libero_object_fluxvla_gbs640_seed42_abf72b8_crop_30k/eval_trainseed7_evalseed7_infeseed42
1051
+ save_failed_rollout_videos: false
1052
+ save_multi_view_rollout_videos: false
1053
+ save_rollout_videos: false
1054
+ seed: 7
1055
+ task_suite_name: libero_object
1056
+ type: LiberoEvalRunner
1057
+ model:
1058
+ embodiment_tag: LIBERO_PANDA
1059
+ freeze_projector: false
1060
+ freeze_vlm_backbone: true
1061
+ load_metadata: true
1062
+ model_path: ./checkpoints/GR00T-N1.7-3B
1063
+ processor_kwargs:
1064
+ apply_sincos_state_encoding: false
1065
+ clip_outliers: true
1066
+ color_jitter_params:
1067
+ brightness: 0.3
1068
+ contrast: 0.4
1069
+ hue: 0.08
1070
+ saturation: 0.5
1071
+ crop_fraction: 0.95
1072
+ embodiment_id_mapping:
1073
+ libero_sim: 2
1074
+ formalize_language: true
1075
+ image_crop_size:
1076
+ - 230
1077
+ - 230
1078
+ image_target_size:
1079
+ - 256
1080
+ - 256
1081
+ max_action_dim: 132
1082
+ max_action_horizon: 40
1083
+ max_state_dim: 132
1084
+ modality_configs:
1085
+ libero_sim:
1086
+ action:
1087
+ delta_indices:
1088
+ - 0
1089
+ - 1
1090
+ - 2
1091
+ - 3
1092
+ - 4
1093
+ - 5
1094
+ - 6
1095
+ - 7
1096
+ - 8
1097
+ - 9
1098
+ - 10
1099
+ - 11
1100
+ - 12
1101
+ - 13
1102
+ - 14
1103
+ - 15
1104
+ modality_keys:
1105
+ - x
1106
+ - y
1107
+ - z
1108
+ - roll
1109
+ - pitch
1110
+ - yaw
1111
+ - gripper
1112
+ state:
1113
+ delta_indices:
1114
+ - 0
1115
+ modality_keys:
1116
+ - x
1117
+ - y
1118
+ - z
1119
+ - roll
1120
+ - pitch
1121
+ - yaw
1122
+ - gripper
1123
+ video:
1124
+ delta_indices:
1125
+ - 0
1126
+ modality_keys:
1127
+ - image
1128
+ - wrist_image
1129
+ shortest_image_edge: 256
1130
+ state_dropout_prob: 0.2
1131
+ statistics:
1132
+ libero_sim:
1133
+ action:
1134
+ gripper:
1135
+ max:
1136
+ - 1.0
1137
+ mean:
1138
+ - 0.46515324845117295
1139
+ min:
1140
+ - 0.0
1141
+ q01:
1142
+ - 0.0
1143
+ q99:
1144
+ - 1.0
1145
+ std:
1146
+ - 0.4987842257994289
1147
+ pitch:
1148
+ max:
1149
+ - 0.35035714507102966
1150
+ mean:
1151
+ - 0.007098307809942585
1152
+ min:
1153
+ - -0.29035714268684387
1154
+ q01:
1155
+ - -0.11678571254014969
1156
+ q99:
1157
+ - 0.14884285390376858
1158
+ std:
1159
+ - 0.04932243328601301
1160
+ roll:
1161
+ max:
1162
+ - 0.17678570747375488
1163
+ mean:
1164
+ - 0.0011625276173484506
1165
+ min:
1166
+ - -0.15000000596046448
1167
+ q01:
1168
+ - -0.06964285671710968
1169
+ q99:
1170
+ - 0.08142857253551483
1171
+ std:
1172
+ - 0.024271765379373398
1173
+ x:
1174
+ max:
1175
+ - 0.9375
1176
+ mean:
1177
+ - 0.07127183159196328
1178
+ min:
1179
+ - -0.8839285969734192
1180
+ q01:
1181
+ - -0.5410714149475098
1182
+ q99:
1183
+ - 0.8464285731315613
1184
+ std:
1185
+ - 0.26873661352251504
1186
+ y:
1187
+ max:
1188
+ - 0.8919642567634583
1189
+ mean:
1190
+ - 0.13518370972847993
1191
+ min:
1192
+ - -0.9375
1193
+ q01:
1194
+ - -0.8758928775787354
1195
+ q99:
1196
+ - 0.84375
1197
+ std:
1198
+ - 0.4385525563650988
1199
+ yaw:
1200
+ max:
1201
+ - 0.1810714304447174
1202
+ mean:
1203
+ - -0.015082184949062672
1204
+ min:
1205
+ - -0.32892856001853943
1206
+ q01:
1207
+ - -0.1607142835855484
1208
+ q99:
1209
+ - 0.08571428805589676
1210
+ std:
1211
+ - 0.04218624160534833
1212
+ z:
1213
+ max:
1214
+ - 0.9375
1215
+ mean:
1216
+ - -0.046194642259443225
1217
+ min:
1218
+ - -0.9375
1219
+ q01:
1220
+ - -0.9375
1221
+ q99:
1222
+ - 0.9375
1223
+ std:
1224
+ - 0.4474788640468074
1225
+ state:
1226
+ gripper:
1227
+ max:
1228
+ - 0.04104502499103546
1229
+ - -0.0005043679848313332
1230
+ mean:
1231
+ - 0.029502615037539585
1232
+ - -0.030672144202923204
1233
+ min:
1234
+ - 0.0006782737909816206
1235
+ - -0.041782498359680176
1236
+ q01:
1237
+ - 0.008699759356677533
1238
+ - -0.04015225619077682
1239
+ q99:
1240
+ - 0.03989199593663215
1241
+ - -0.010282181538641466
1242
+ std:
1243
+ - 0.00947681687068548
1244
+ - 0.009059544368504133
1245
+ pitch:
1246
+ max:
1247
+ - 0.7954724431037903
1248
+ mean:
1249
+ - -0.21535565470621534
1250
+ min:
1251
+ - -1.8830392360687256
1252
+ q01:
1253
+ - -1.401689453125
1254
+ q99:
1255
+ - 0.3196117353439328
1256
+ std:
1257
+ - 0.33845984937441376
1258
+ roll:
1259
+ max:
1260
+ - 3.4003379344940186
1261
+ mean:
1262
+ - 3.107347627230808
1263
+ min:
1264
+ - 2.2889645099639893
1265
+ q01:
1266
+ - 2.7467152214050294
1267
+ q99:
1268
+ - 3.260209331512451
1269
+ std:
1270
+ - 0.08837916027138361
1271
+ x:
1272
+ max:
1273
+ - 0.14580604434013367
1274
+ mean:
1275
+ - -0.029590097511622324
1276
+ min:
1277
+ - -0.17652566730976105
1278
+ q01:
1279
+ - -0.149256848692894
1280
+ q99:
1281
+ - 0.09154553085565566
1282
+ std:
1283
+ - 0.06715783727495737
1284
+ y:
1285
+ max:
1286
+ - 0.3322660028934479
1287
+ mean:
1288
+ - -0.007297964757377373
1289
+ min:
1290
+ - -0.28169912099838257
1291
+ q01:
1292
+ - -0.2592130792140961
1293
+ q99:
1294
+ - 0.2911862111091613
1295
+ std:
1296
+ - 0.17582774832397174
1297
+ yaw:
1298
+ max:
1299
+ - 0.6642153263092041
1300
+ mean:
1301
+ - -0.11427672217768083
1302
+ min:
1303
+ - -1.0428133010864258
1304
+ q01:
1305
+ - -0.6867975068092346
1306
+ q99:
1307
+ - 0.39586576938629137
1308
+ std:
1309
+ - 0.2067143161337991
1310
+ z:
1311
+ max:
1312
+ - 0.38492220640182495
1313
+ mean:
1314
+ - 0.20263691224329605
1315
+ min:
1316
+ - 0.008128181099891663
1317
+ q01:
1318
+ - 0.009932418912649154
1319
+ q99:
1320
+ - 0.3361880838871002
1321
+ std:
1322
+ - 0.07806197322485103
1323
+ use_albumentations: true
1324
+ use_percentiles: true
1325
+ use_relative_action: true
1326
+ qwen3_runtime: compat_457
1327
+ state_dropout_prob: 0.2
1328
+ type: GrootN17VLA
1329
+ vla_head:
1330
+ tune_diffusion_model: true
1331
+ tune_vlln: true
1332
+ type: GrootN17ActionHead
1333
+ vlm_backbone:
1334
+ load_bf16: false
1335
+ model_config:
1336
+ architectures:
1337
+ - Qwen3VLForConditionalGeneration
1338
+ image_token_id: 151655
1339
+ text_config:
1340
+ attention_bias: false
1341
+ attention_dropout: 0.0
1342
+ head_dim: 128
1343
+ hidden_act: silu
1344
+ hidden_size: 2048
1345
+ initializer_range: 0.02
1346
+ intermediate_size: 6144
1347
+ max_position_embeddings: 262144
1348
+ model_type: qwen3_vl_text
1349
+ num_attention_heads: 16
1350
+ num_hidden_layers: 28
1351
+ num_key_value_heads: 8
1352
+ rms_norm_eps: 1.0e-06
1353
+ rope_parameters:
1354
+ mrope_interleaved: true
1355
+ mrope_section:
1356
+ - 24
1357
+ - 20
1358
+ - 20
1359
+ rope_theta: 5000000.0
1360
+ rope_type: default
1361
+ use_cache: false
1362
+ vocab_size: 151936
1363
+ tie_word_embeddings: false
1364
+ video_token_id: 151656
1365
+ vision_config:
1366
+ deepstack_visual_indexes:
1367
+ - 5
1368
+ - 11
1369
+ - 17
1370
+ depth: 24
1371
+ hidden_act: gelu_pytorch_tanh
1372
+ hidden_size: 1024
1373
+ in_channels: 3
1374
+ initializer_range: 0.02
1375
+ intermediate_size: 4096
1376
+ model_type: qwen3_vl_vision
1377
+ num_heads: 16
1378
+ num_position_embeddings: 2304
1379
+ out_hidden_size: 2048
1380
+ patch_size: 16
1381
+ spatial_merge_size: 2
1382
+ temporal_patch_size: 2
1383
+ vision_end_token_id: 151653
1384
+ vision_start_token_id: 151652
1385
+ qwen3_runtime: compat_457
1386
+ reproject_vision: false
1387
+ select_layer: 16
1388
+ trainable_params_fp32: true
1389
+ tune_top_llm_layers: 0
1390
+ type: GrootN17Qwen3Backbone
1391
+ use_flash_attention: true
1392
+ runner:
1393
+ change_key_name: false
1394
+ collator:
1395
+ keys:
1396
+ - lang_tokens
1397
+ - lang_masks
1398
+ - images
1399
+ - image_grid_thw
1400
+ - states
1401
+ - actions
1402
+ - action_masks
1403
+ - embodiment_ids
1404
+ - sample_weight
1405
+ type: DictCollator
1406
+ enable_gradient_checkpointing: true
1407
+ enable_mixed_precision_training: true
1408
+ grad_accumulation_steps: 1
1409
+ lr_scheduler:
1410
+ type: linear-warmup+cosine-decay
1411
+ warmup_ratio: 0.05
1412
+ max_grad_norm: 1.0
1413
+ max_keep_ckpts: 4
1414
+ max_steps: 30000
1415
+ metric:
1416
+ active_trackers:
1417
+ - jsonl
1418
+ grad_accumulation_steps: 1
1419
+ run_dir: work_dirs
1420
+ type: VLAMetric
1421
+ window_size: 1
1422
+ mixed_precision_dtype: bf16
1423
+ optimizer:
1424
+ lr: 0.0001
1425
+ type: AdamW
1426
+ weight_decay: 1.0e-05
1427
+ sampler: null
1428
+ save_epoch_interval: 1
1429
+ save_iter_interval: 5000
1430
+ sharding_strategy: full-shard
1431
+ type: FSDPTrainRunner
1432
+ seed: 42
1433
+ train_dataloader:
1434
+ dataset:
1435
+ dataset_statistics:
1436
+ libero_object_no_noops_native:
1437
+ actions:
1438
+ max:
1439
+ - 0.9375
1440
+ - 0.8919642567634583
1441
+ - 0.9375
1442
+ - 0.17678570747375488
1443
+ - 0.35035714507102966
1444
+ - 0.1810714304447174
1445
+ - 1.0
1446
+ mean:
1447
+ - 0.07127183159196328
1448
+ - 0.13518370972847993
1449
+ - -0.046194642259443225
1450
+ - 0.0011625276173484506
1451
+ - 0.007098307809942585
1452
+ - -0.015082184949062672
1453
+ - 0.46515324845117295
1454
+ min:
1455
+ - -0.8839285969734192
1456
+ - -0.9375
1457
+ - -0.9375
1458
+ - -0.15000000596046448
1459
+ - -0.29035714268684387
1460
+ - -0.32892856001853943
1461
+ - 0.0
1462
+ q01:
1463
+ - -0.5410714149475098
1464
+ - -0.8758928775787354
1465
+ - -0.9375
1466
+ - -0.06964285671710968
1467
+ - -0.11678571254014969
1468
+ - -0.1607142835855484
1469
+ - 0.0
1470
+ q99:
1471
+ - 0.8464285731315613
1472
+ - 0.84375
1473
+ - 0.9375
1474
+ - 0.08142857253551483
1475
+ - 0.14884285390376858
1476
+ - 0.08571428805589676
1477
+ - 1.0
1478
+ std:
1479
+ - 0.26873661352251504
1480
+ - 0.4385525563650988
1481
+ - 0.4474788640468074
1482
+ - 0.024271765379373398
1483
+ - 0.04932243328601301
1484
+ - 0.04218624160534833
1485
+ - 0.4987842257994289
1486
+ states:
1487
+ max:
1488
+ - 0.14580604434013367
1489
+ - 0.3322660028934479
1490
+ - 0.38492220640182495
1491
+ - 3.4003379344940186
1492
+ - 0.7954724431037903
1493
+ - 0.6642153263092041
1494
+ - 0.04104502499103546
1495
+ - -0.0005043679848313332
1496
+ mean:
1497
+ - -0.029590097511622324
1498
+ - -0.007297964757377373
1499
+ - 0.20263691224329605
1500
+ - 3.107347627230808
1501
+ - -0.21535565470621534
1502
+ - -0.11427672217768083
1503
+ - 0.029502615037539585
1504
+ - -0.030672144202923204
1505
+ min:
1506
+ - -0.17652566730976105
1507
+ - -0.28169912099838257
1508
+ - 0.008128181099891663
1509
+ - 2.2889645099639893
1510
+ - -1.8830392360687256
1511
+ - -1.0428133010864258
1512
+ - 0.0006782737909816206
1513
+ - -0.041782498359680176
1514
+ q01:
1515
+ - -0.149256848692894
1516
+ - -0.2592130792140961
1517
+ - 0.009932418912649154
1518
+ - 2.7467152214050294
1519
+ - -1.401689453125
1520
+ - -0.6867975068092346
1521
+ - 0.008699759356677533
1522
+ - -0.04015225619077682
1523
+ q99:
1524
+ - 0.09154553085565566
1525
+ - 0.2911862111091613
1526
+ - 0.3361880838871002
1527
+ - 3.260209331512451
1528
+ - 0.3196117353439328
1529
+ - 0.39586576938629137
1530
+ - 0.03989199593663215
1531
+ - -0.010282181538641466
1532
+ std:
1533
+ - 0.06715783727495737
1534
+ - 0.17582774832397174
1535
+ - 0.07806197322485103
1536
+ - 0.08837916027138361
1537
+ - 0.33845984937441376
1538
+ - 0.2067143161337991
1539
+ - 0.00947681687068548
1540
+ - 0.009059544368504133
1541
+ datasets:
1542
+ - action_key: action
1543
+ action_window_size: 16
1544
+ data_root_path: /mnt/data/cpfs/mnt/data/liyinhao/datasets/libero_object_no_noops_lerobotv2.1
1545
+ drop_incomplete_action_windows: true
1546
+ repeat_to_full_length: false
1547
+ statistic_name: libero_object_no_noops_native
1548
+ train_episode_fraction: 1.0
1549
+ transforms:
1550
+ - embodiment_id: 2
1551
+ name_mappings:
1552
+ actions:
1553
+ - actions
1554
+ observation.state:
1555
+ - states
1556
+ parquet_keys:
1557
+ - observation.state
1558
+ - timestamp
1559
+ - actions
1560
+ - info
1561
+ - stats
1562
+ - action_masks
1563
+ type: ProcessParquetInputs
1564
+ video_keys:
1565
+ - observation.images.image
1566
+ - observation.images.wrist_image
1567
+ - action_dim: 132
1568
+ action_key: actions
1569
+ clip_norm: true
1570
+ norm_type: quantile
1571
+ normalization_epsilon: 0.0
1572
+ preserve_input_dtype: true
1573
+ state_dim: 132
1574
+ state_key: states
1575
+ type: NormalizeStatesAndActions
1576
+ - action_horizon: 40
1577
+ state_dropout_prob: 0.2
1578
+ state_history_length: 1
1579
+ type: PrepareStateActionTargets
1580
+ valid_action_dim: 7
1581
+ - embodiment_tag: LIBERO_PANDA
1582
+ image_key: images
1583
+ output_image_key: images
1584
+ processor_kwargs:
1585
+ apply_sincos_state_encoding: false
1586
+ clip_outliers: true
1587
+ color_jitter_params:
1588
+ brightness: 0.3
1589
+ contrast: 0.4
1590
+ hue: 0.08
1591
+ saturation: 0.5
1592
+ crop_fraction: 0.95
1593
+ embodiment_id_mapping:
1594
+ libero_sim: 2
1595
+ formalize_language: true
1596
+ image_crop_size:
1597
+ - 230
1598
+ - 230
1599
+ image_target_size:
1600
+ - 256
1601
+ - 256
1602
+ max_action_dim: 132
1603
+ max_action_horizon: 40
1604
+ max_state_dim: 132
1605
+ modality_configs:
1606
+ libero_sim:
1607
+ action:
1608
+ delta_indices:
1609
+ - 0
1610
+ - 1
1611
+ - 2
1612
+ - 3
1613
+ - 4
1614
+ - 5
1615
+ - 6
1616
+ - 7
1617
+ - 8
1618
+ - 9
1619
+ - 10
1620
+ - 11
1621
+ - 12
1622
+ - 13
1623
+ - 14
1624
+ - 15
1625
+ modality_keys:
1626
+ - x
1627
+ - y
1628
+ - z
1629
+ - roll
1630
+ - pitch
1631
+ - yaw
1632
+ - gripper
1633
+ state:
1634
+ delta_indices:
1635
+ - 0
1636
+ modality_keys:
1637
+ - x
1638
+ - y
1639
+ - z
1640
+ - roll
1641
+ - pitch
1642
+ - yaw
1643
+ - gripper
1644
+ video:
1645
+ delta_indices:
1646
+ - 0
1647
+ modality_keys:
1648
+ - image
1649
+ - wrist_image
1650
+ shortest_image_edge: 256
1651
+ state_dropout_prob: 0.2
1652
+ statistics:
1653
+ libero_sim:
1654
+ action:
1655
+ gripper:
1656
+ max:
1657
+ - 1.0
1658
+ mean:
1659
+ - 0.46515324845117295
1660
+ min:
1661
+ - 0.0
1662
+ q01:
1663
+ - 0.0
1664
+ q99:
1665
+ - 1.0
1666
+ std:
1667
+ - 0.4987842257994289
1668
+ pitch:
1669
+ max:
1670
+ - 0.35035714507102966
1671
+ mean:
1672
+ - 0.007098307809942585
1673
+ min:
1674
+ - -0.29035714268684387
1675
+ q01:
1676
+ - -0.11678571254014969
1677
+ q99:
1678
+ - 0.14884285390376858
1679
+ std:
1680
+ - 0.04932243328601301
1681
+ roll:
1682
+ max:
1683
+ - 0.17678570747375488
1684
+ mean:
1685
+ - 0.0011625276173484506
1686
+ min:
1687
+ - -0.15000000596046448
1688
+ q01:
1689
+ - -0.06964285671710968
1690
+ q99:
1691
+ - 0.08142857253551483
1692
+ std:
1693
+ - 0.024271765379373398
1694
+ x:
1695
+ max:
1696
+ - 0.9375
1697
+ mean:
1698
+ - 0.07127183159196328
1699
+ min:
1700
+ - -0.8839285969734192
1701
+ q01:
1702
+ - -0.5410714149475098
1703
+ q99:
1704
+ - 0.8464285731315613
1705
+ std:
1706
+ - 0.26873661352251504
1707
+ y:
1708
+ max:
1709
+ - 0.8919642567634583
1710
+ mean:
1711
+ - 0.13518370972847993
1712
+ min:
1713
+ - -0.9375
1714
+ q01:
1715
+ - -0.8758928775787354
1716
+ q99:
1717
+ - 0.84375
1718
+ std:
1719
+ - 0.4385525563650988
1720
+ yaw:
1721
+ max:
1722
+ - 0.1810714304447174
1723
+ mean:
1724
+ - -0.015082184949062672
1725
+ min:
1726
+ - -0.32892856001853943
1727
+ q01:
1728
+ - -0.1607142835855484
1729
+ q99:
1730
+ - 0.08571428805589676
1731
+ std:
1732
+ - 0.04218624160534833
1733
+ z:
1734
+ max:
1735
+ - 0.9375
1736
+ mean:
1737
+ - -0.046194642259443225
1738
+ min:
1739
+ - -0.9375
1740
+ q01:
1741
+ - -0.9375
1742
+ q99:
1743
+ - 0.9375
1744
+ std:
1745
+ - 0.4474788640468074
1746
+ state:
1747
+ gripper:
1748
+ max:
1749
+ - 0.04104502499103546
1750
+ - -0.0005043679848313332
1751
+ mean:
1752
+ - 0.029502615037539585
1753
+ - -0.030672144202923204
1754
+ min:
1755
+ - 0.0006782737909816206
1756
+ - -0.041782498359680176
1757
+ q01:
1758
+ - 0.008699759356677533
1759
+ - -0.04015225619077682
1760
+ q99:
1761
+ - 0.03989199593663215
1762
+ - -0.010282181538641466
1763
+ std:
1764
+ - 0.00947681687068548
1765
+ - 0.009059544368504133
1766
+ pitch:
1767
+ max:
1768
+ - 0.7954724431037903
1769
+ mean:
1770
+ - -0.21535565470621534
1771
+ min:
1772
+ - -1.8830392360687256
1773
+ q01:
1774
+ - -1.401689453125
1775
+ q99:
1776
+ - 0.3196117353439328
1777
+ std:
1778
+ - 0.33845984937441376
1779
+ roll:
1780
+ max:
1781
+ - 3.4003379344940186
1782
+ mean:
1783
+ - 3.107347627230808
1784
+ min:
1785
+ - 2.2889645099639893
1786
+ q01:
1787
+ - 2.7467152214050294
1788
+ q99:
1789
+ - 3.260209331512451
1790
+ std:
1791
+ - 0.08837916027138361
1792
+ x:
1793
+ max:
1794
+ - 0.14580604434013367
1795
+ mean:
1796
+ - -0.029590097511622324
1797
+ min:
1798
+ - -0.17652566730976105
1799
+ q01:
1800
+ - -0.149256848692894
1801
+ q99:
1802
+ - 0.09154553085565566
1803
+ std:
1804
+ - 0.06715783727495737
1805
+ y:
1806
+ max:
1807
+ - 0.3322660028934479
1808
+ mean:
1809
+ - -0.007297964757377373
1810
+ min:
1811
+ - -0.28169912099838257
1812
+ q01:
1813
+ - -0.2592130792140961
1814
+ q99:
1815
+ - 0.2911862111091613
1816
+ std:
1817
+ - 0.17582774832397174
1818
+ yaw:
1819
+ max:
1820
+ - 0.6642153263092041
1821
+ mean:
1822
+ - -0.11427672217768083
1823
+ min:
1824
+ - -1.0428133010864258
1825
+ q01:
1826
+ - -0.6867975068092346
1827
+ q99:
1828
+ - 0.39586576938629137
1829
+ std:
1830
+ - 0.2067143161337991
1831
+ z:
1832
+ max:
1833
+ - 0.38492220640182495
1834
+ mean:
1835
+ - 0.20263691224329605
1836
+ min:
1837
+ - 0.008128181099891663
1838
+ q01:
1839
+ - 0.009932418912649154
1840
+ q99:
1841
+ - 0.3361880838871002
1842
+ std:
1843
+ - 0.07806197322485103
1844
+ use_albumentations: true
1845
+ use_percentiles: true
1846
+ use_relative_action: true
1847
+ train_mode: true
1848
+ type: GrootN17ImageAugmentation
1849
+ - image_mean:
1850
+ - 0.5
1851
+ - 0.5
1852
+ - 0.5
1853
+ image_std:
1854
+ - 0.5
1855
+ - 0.5
1856
+ - 0.5
1857
+ img_key: images
1858
+ merge_size: 2
1859
+ patch_size: 16
1860
+ size:
1861
+ longest_edge: 16777216
1862
+ shortest_edge: 65536
1863
+ temporal_patch_size: 2
1864
+ to_tensor: true
1865
+ type: QWen2VLImageTransform
1866
+ - add_assistant_stub: false
1867
+ add_system: false
1868
+ attention_mask_dtype: int64
1869
+ image_grid_thw_key: image_grid_thw
1870
+ image_merge_size: 2
1871
+ image_tag_template: ''
1872
+ img_context_token: <|image_pad|>
1873
+ img_end: <|vision_end|>
1874
+ img_start: <|vision_start|>
1875
+ img_tokens_source: from_image_grid_thw
1876
+ lowercase_task_description: true
1877
+ max_len: 180
1878
+ output_keys:
1879
+ - lang_tokens
1880
+ - lang_masks
1881
+ - images
1882
+ - image_grid_thw
1883
+ - states
1884
+ - actions
1885
+ - action_masks
1886
+ - embodiment_ids
1887
+ - sample_weight
1888
+ padding_side: left
1889
+ strip_task_punctuation: true
1890
+ task_pos: after_images
1891
+ tokenizer:
1892
+ model_path: fluxvla/models/third_party_models/qwen3_tokenizer
1893
+ padding_side: left
1894
+ trust_remote_code: false
1895
+ type: PretrainedTokenizer
1896
+ truncate: false
1897
+ type: ProcessPromptsWithImage
1898
+ use_eos_as_pad: false
1899
+ type: ParquetDataset
1900
+ use_delta: false
1901
+ window_start_idx: 0
1902
+ name_mappings:
1903
+ action:
1904
+ - actions
1905
+ observation.state:
1906
+ - states
1907
+ reshuffle_each_epoch: true
1908
+ seed: 42
1909
+ shuffle: true
1910
+ statistic_keys:
1911
+ - observation.state
1912
+ - timestamp
1913
+ - action
1914
+ statistic_name: libero_object_no_noops_native
1915
+ type: DistributedRepeatingDataset
1916
+ per_device_batch_size: 40
1917
+ per_device_num_workers: 4
gr00t_n17_native_libero_object_full_finetune_bs64/dataset_statistics.json ADDED
@@ -0,0 +1,122 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "libero_object_no_noops_native": {
3
+ "states": {
4
+ "min": [
5
+ -0.17652566730976105,
6
+ -0.28169912099838257,
7
+ 0.008128181099891663,
8
+ 2.2889645099639893,
9
+ -1.8830392360687256,
10
+ -1.0428133010864258,
11
+ 0.0006782737909816206,
12
+ -0.041782498359680176
13
+ ],
14
+ "max": [
15
+ 0.14580604434013367,
16
+ 0.3322660028934479,
17
+ 0.38492220640182495,
18
+ 3.4003379344940186,
19
+ 0.7954724431037903,
20
+ 0.6642153263092041,
21
+ 0.04104502499103546,
22
+ -0.0005043679848313332
23
+ ],
24
+ "mean": [
25
+ -0.029590097511622324,
26
+ -0.007297964757377373,
27
+ 0.20263691224329605,
28
+ 3.107347627230808,
29
+ -0.21535565470621534,
30
+ -0.11427672217768083,
31
+ 0.029502615037539585,
32
+ -0.030672144202923204
33
+ ],
34
+ "std": [
35
+ 0.06715783727495737,
36
+ 0.17582774832397174,
37
+ 0.07806197322485103,
38
+ 0.08837916027138361,
39
+ 0.33845984937441376,
40
+ 0.2067143161337991,
41
+ 0.00947681687068548,
42
+ 0.009059544368504133
43
+ ],
44
+ "q01": [
45
+ -0.149256848692894,
46
+ -0.2592130792140961,
47
+ 0.009932418912649154,
48
+ 2.7467152214050294,
49
+ -1.401689453125,
50
+ -0.6867975068092346,
51
+ 0.008699759356677533,
52
+ -0.04015225619077682
53
+ ],
54
+ "q99": [
55
+ 0.09154553085565566,
56
+ 0.2911862111091613,
57
+ 0.3361880838871002,
58
+ 3.260209331512451,
59
+ 0.3196117353439328,
60
+ 0.39586576938629137,
61
+ 0.03989199593663215,
62
+ -0.010282181538641466
63
+ ]
64
+ },
65
+ "actions": {
66
+ "min": [
67
+ -0.8839285969734192,
68
+ -0.9375,
69
+ -0.9375,
70
+ -0.15000000596046448,
71
+ -0.29035714268684387,
72
+ -0.32892856001853943,
73
+ 0.0
74
+ ],
75
+ "max": [
76
+ 0.9375,
77
+ 0.8919642567634583,
78
+ 0.9375,
79
+ 0.17678570747375488,
80
+ 0.35035714507102966,
81
+ 0.1810714304447174,
82
+ 1.0
83
+ ],
84
+ "mean": [
85
+ 0.07127183159196328,
86
+ 0.13518370972847993,
87
+ -0.046194642259443225,
88
+ 0.0011625276173484506,
89
+ 0.007098307809942585,
90
+ -0.015082184949062672,
91
+ 0.46515324845117295
92
+ ],
93
+ "std": [
94
+ 0.26873661352251504,
95
+ 0.4385525563650988,
96
+ 0.4474788640468074,
97
+ 0.024271765379373398,
98
+ 0.04932243328601301,
99
+ 0.04218624160534833,
100
+ 0.4987842257994289
101
+ ],
102
+ "q01": [
103
+ -0.5410714149475098,
104
+ -0.8758928775787354,
105
+ -0.9375,
106
+ -0.06964285671710968,
107
+ -0.11678571254014969,
108
+ -0.1607142835855484,
109
+ 0.0
110
+ ],
111
+ "q99": [
112
+ 0.8464285731315613,
113
+ 0.84375,
114
+ 0.9375,
115
+ 0.08142857253551483,
116
+ 0.14884285390376858,
117
+ 0.08571428805589676,
118
+ 1.0
119
+ ]
120
+ }
121
+ }
122
+ }
gr00t_n17_native_libero_object_full_finetune_bs64/eval_trainseed7_evalseed7_infeseed42/eval_runs/step-030000-epoch-315-loss=0.0074/EVAL-libero_object-groot_n17_native-2026_08_19-05_00_58/summary.json ADDED
@@ -0,0 +1,91 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "run_id": "EVAL-libero_object-groot_n17_native-2026_08_19-05_00_58",
3
+ "ckpt": "/mnt/data/cpfs/mnt/data/yiming/fluxvla-n17-libero-pr/work_dirs/groot_n17_libero_object_fluxvla_gbs640_seed42_abf72b8_crop_30k/checkpoints/latest-checkpoint.safetensors",
4
+ "config": "config",
5
+ "suite_stats": {
6
+ "libero_object": {
7
+ "total_tasks": 10,
8
+ "total_trials": 500,
9
+ "total_successes": 493,
10
+ "total_time": 2589.262908935547,
11
+ "max_time": 357.157958984375
12
+ }
13
+ },
14
+ "task_results": {
15
+ "libero_object_0": {
16
+ "success_rate": 98.0,
17
+ "duration": 357.157958984375,
18
+ "total_episodes": 50,
19
+ "successes": 49,
20
+ "task_description": "pick up the alphabet soup and place it in the basket"
21
+ },
22
+ "libero_object_1": {
23
+ "success_rate": 96.0,
24
+ "duration": 280.19744873046875,
25
+ "total_episodes": 50,
26
+ "successes": 48,
27
+ "task_description": "pick up the cream cheese and place it in the basket"
28
+ },
29
+ "libero_object_2": {
30
+ "success_rate": 100.0,
31
+ "duration": 206.04055786132812,
32
+ "total_episodes": 50,
33
+ "successes": 50,
34
+ "task_description": "pick up the salad dressing and place it in the basket"
35
+ },
36
+ "libero_object_3": {
37
+ "success_rate": 100.0,
38
+ "duration": 229.9640655517578,
39
+ "total_episodes": 50,
40
+ "successes": 50,
41
+ "task_description": "pick up the bbq sauce and place it in the basket"
42
+ },
43
+ "libero_object_4": {
44
+ "success_rate": 100.0,
45
+ "duration": 247.06512451171875,
46
+ "total_episodes": 50,
47
+ "successes": 50,
48
+ "task_description": "pick up the ketchup and place it in the basket"
49
+ },
50
+ "libero_object_5": {
51
+ "success_rate": 96.0,
52
+ "duration": 282.04974365234375,
53
+ "total_episodes": 50,
54
+ "successes": 48,
55
+ "task_description": "pick up the tomato sauce and place it in the basket"
56
+ },
57
+ "libero_object_6": {
58
+ "success_rate": 100.0,
59
+ "duration": 260.9024963378906,
60
+ "total_episodes": 50,
61
+ "successes": 50,
62
+ "task_description": "pick up the butter and place it in the basket"
63
+ },
64
+ "libero_object_7": {
65
+ "success_rate": 96.0,
66
+ "duration": 261.6629333496094,
67
+ "total_episodes": 50,
68
+ "successes": 48,
69
+ "task_description": "pick up the milk and place it in the basket"
70
+ },
71
+ "libero_object_8": {
72
+ "success_rate": 100.0,
73
+ "duration": 248.90562438964844,
74
+ "total_episodes": 50,
75
+ "successes": 50,
76
+ "task_description": "pick up the chocolate pudding and place it in the basket"
77
+ },
78
+ "libero_object_9": {
79
+ "success_rate": 100.0,
80
+ "duration": 215.31695556640625,
81
+ "total_episodes": 50,
82
+ "successes": 50,
83
+ "task_description": "pick up the orange juice and place it in the basket"
84
+ }
85
+ },
86
+ "overall": {
87
+ "average_success_rate": 98.6,
88
+ "total_time": 2589.262908935547,
89
+ "average_task_time": 258.92629089355466
90
+ }
91
+ }