Chengpengaaa commited on
Commit
515533c
·
verified ·
1 Parent(s): 3a19264

Upload folder using huggingface_hub

Browse files
Intern-oft-0130v/checkpoints/steps_62000_pytorch_model.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7c77ccb7514b611db6cdb6304e771c81b9d9f42b4b7d62637de2fa3268b7a384
3
+ size 2149479434
Intern-oft-0130v/config.json ADDED
@@ -0,0 +1,89 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "run_id": "InternOFT-0130v-eep",
3
+ "run_root_dir": "/data/LumosVLA/LumosVLA/results",
4
+ "seed": 42,
5
+ "trackers": [
6
+ "jsonl",
7
+ "wandb"
8
+ ],
9
+ "wandb_entity": "xspadex-nanyang-technological-university-singapore",
10
+ "wandb_project": "internvl-oft-0130v-eep",
11
+ "is_debug": false,
12
+ "framework": {
13
+ "name": "InternOFT",
14
+ "qwenvl": {
15
+ "base_vlm": "/data/LumosVLA/LumosVLA/playground/internvl",
16
+ "attn_implementation": "flash_attention_2",
17
+ "vl_hidden_dim": 2048
18
+ },
19
+ "action_model": {
20
+ "action_model_type": "VLA-Adapter",
21
+ "hidden_dim": 2048,
22
+ "action_dim": 20,
23
+ "state_dim": 20,
24
+ "action_query_num": 64,
25
+ "use_pro_version": true,
26
+ "use_proprio": true,
27
+ "phase": "Training",
28
+ "future_action_window_size": 30
29
+ }
30
+ },
31
+ "datasets": {
32
+ "vla_data": {
33
+ "dataset_py": "lerobot_datasets",
34
+ "data_root_dir": "/data/LumosVLA/LumosVLA/data/starvla_intern",
35
+ "data_mix": "train_internAdapter",
36
+ "per_device_batch_size": 32,
37
+ "num_workers": 8,
38
+ "action_type": "abs_ee",
39
+ "image_size": 256,
40
+ "video_backend": "pyav"
41
+ }
42
+ },
43
+ "trainer": {
44
+ "epochs": 100,
45
+ "max_train_steps": 125000,
46
+ "num_warmup_steps": 1000,
47
+ "save_interval": 1000,
48
+ "eval_interval": 100000,
49
+ "learning_rate": {
50
+ "base": 4e-06,
51
+ "qwen_vl_interface": 4e-06,
52
+ "action_model": 1e-05
53
+ },
54
+ "lr_scheduler_type": "cosine_with_min_lr",
55
+ "scheduler_specific_kwargs": {
56
+ "min_lr": 5e-07
57
+ },
58
+ "freeze_modules": "intern_vl_interface.model.language_model",
59
+ "loss_scale": {
60
+ "vla": 1.0,
61
+ "vlm": 0.05
62
+ },
63
+ "max_grad_norm": 1.0,
64
+ "warmup_ratio": 0.1,
65
+ "weight_decay": 0.0,
66
+ "logging_frequency": 10,
67
+ "gradient_clipping": 1.0,
68
+ "gradient_accumulation_steps": 1,
69
+ "optimizer": {
70
+ "name": "AdamW",
71
+ "betas": [
72
+ 0.9,
73
+ 0.95
74
+ ],
75
+ "eps": 1e-08,
76
+ "weight_decay": 0
77
+ },
78
+ "is_resume": true,
79
+ "resume_step": 14000
80
+ },
81
+ "is_resume": false,
82
+ "resume_epoch": null,
83
+ "resume_step": null,
84
+ "enable_gradient_checkpointing": true,
85
+ "enable_mixed_precision_training": true,
86
+ "report_to": "wandb",
87
+ "wandb_run_name": "v1",
88
+ "output_dir": "/data/LumosVLA/LumosVLA/results/InternOFT-0130v-eep"
89
+ }
Intern-oft-0130v/config.yaml ADDED
@@ -0,0 +1,50 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ datasets:
2
+ vla_data:
3
+ data_mix: train_internAdapter
4
+ data_root_dir: /data/LumosVLA/LumosVLA/data/starvla_intern
5
+ dataset_py: lerobot_datasets
6
+ image_size: 256
7
+ num_workers: 8
8
+ per_device_batch_size: 32
9
+ framework:
10
+ action_model:
11
+ action_dim: 20
12
+ action_hidden_dim: 1024
13
+ action_model_type: VLA-Adapter
14
+ future_action_window_size: 30
15
+ hidden_dim: 1024
16
+ state_dim: 20
17
+ use_proprio: true
18
+ name: InternOFT
19
+ qwenvl:
20
+ base_vlm: /data/LumosVLA/LumosVLA/playground/internvl
21
+ vl_hidden_dim: 1024
22
+ output_dir: /data/LumosVLA/LumosVLA/results/InternOFT-0130v-eep
23
+ run_id: InternOFT-0130v-eep
24
+ run_root_dir: /data/LumosVLA/LumosVLA/results
25
+ seed: 42
26
+ trainer:
27
+ eval_interval: 100000
28
+ freeze_modules: intern_vl_interface.model.language_model
29
+ gradient_accumulation_steps: 1
30
+ gradient_clipping: 1.0
31
+ is_resume: true
32
+ learning_rate:
33
+ action_model: 1.0e-05
34
+ base: 4.0e-06
35
+ qwen_vl_interface: 4.0e-06
36
+ logging_frequency: 10
37
+ lr_scheduler_type: cosine_with_min_lr
38
+ max_train_steps: 125000
39
+ num_warmup_steps: 1000
40
+ optimizer:
41
+ betas:
42
+ - 0.9
43
+ - 0.95
44
+ eps: 1.0e-08
45
+ weight_decay: 0
46
+ save_interval: 1000
47
+ scheduler_specific_kwargs:
48
+ min_lr: 5.0e-07
49
+ wandb_entity: xspadex-nanyang-technological-university-singapore
50
+ wandb_project: internvl-oft-0130v-eep
Intern-oft-0130v/dataset_statistics.json ADDED
@@ -0,0 +1,218 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "new_embodiment": {
3
+ "action": {
4
+ "mean": [
5
+ 0.2536350143583198,
6
+ -0.07743038404671672,
7
+ 0.1966707251573864,
8
+ -1.3468584815661115,
9
+ 0.40754710399267974,
10
+ 1.6235627918912656,
11
+ 0.034987629603659906,
12
+ 0.23352106752102836,
13
+ 0.09337638377359042,
14
+ 0.20792568958642196,
15
+ 1.5060326024367099,
16
+ 0.4999388032837918,
17
+ -1.7593652281844825,
18
+ 0.0355335969412536
19
+ ],
20
+ "std": [
21
+ 0.09796345941892504,
22
+ 0.10853129549959974,
23
+ 0.0681641731581649,
24
+ 2.571989629620975,
25
+ 0.3175618094699909,
26
+ 2.272367583216257,
27
+ 0.028098021946224527,
28
+ 0.08823442171678504,
29
+ 0.07837648996356361,
30
+ 0.05453627898660852,
31
+ 2.5246083674731827,
32
+ 0.28385594171744144,
33
+ 2.189727010118041,
34
+ 0.02824469944283456
35
+ ],
36
+ "max": [
37
+ 0.5931400060653687,
38
+ 0.48442599177360535,
39
+ 0.6967939734458923,
40
+ 3.1415927410125732,
41
+ 1.5707963705062866,
42
+ 3.1415927410125732,
43
+ 0.0810599997639656,
44
+ 0.5489550232887268,
45
+ 0.45939499139785767,
46
+ 0.6456500291824341,
47
+ 3.1415927410125732,
48
+ 1.5707963705062866,
49
+ 3.1415927410125732,
50
+ 0.060339998453855515
51
+ ],
52
+ "min": [
53
+ -0.10788500308990479,
54
+ -0.5191159844398499,
55
+ 0.040369000285863876,
56
+ -3.1410515308380127,
57
+ -1.5302174091339111,
58
+ -3.1410515308380127,
59
+ 0.0,
60
+ -0.02319999970495701,
61
+ -0.28251099586486816,
62
+ 0.06903599947690964,
63
+ -3.1410515308380127,
64
+ -0.7706850171089172,
65
+ -3.1410515308380127,
66
+ 0.0
67
+ ],
68
+ "q01": [
69
+ 0.044102057814598083,
70
+ -0.36912801861763,
71
+ 0.07184209674596786,
72
+ -3.1372463703155518,
73
+ -0.20070484280586243,
74
+ -3.137392044067383,
75
+ 7.000000186963007e-05,
76
+ 0.053725481033325195,
77
+ -0.15914200246334076,
78
+ 0.10170400142669678,
79
+ -3.138237953186035,
80
+ -0.11282174289226532,
81
+ -3.1348206996917725,
82
+ 7.000000186963007e-05
83
+ ],
84
+ "q99": [
85
+ 0.5304324626922607,
86
+ 0.4121397137641907,
87
+ 0.4779260456562042,
88
+ 3.1415927410125732,
89
+ 1.5365502834320068,
90
+ 3.1415927410125732,
91
+ 0.059780001640319824,
92
+ 0.4840700030326843,
93
+ 0.3484378159046173,
94
+ 0.42531776428222656,
95
+ 3.1415927410125732,
96
+ 1.5435858964920044,
97
+ 3.1415927410125732,
98
+ 0.05992000177502632
99
+ ],
100
+ "mask": [
101
+ true,
102
+ true,
103
+ true,
104
+ true,
105
+ true,
106
+ true,
107
+ false,
108
+ true,
109
+ true,
110
+ true,
111
+ true,
112
+ true,
113
+ true,
114
+ false
115
+ ]
116
+ },
117
+ "state": {
118
+ "mean": [
119
+ 0.2536350143583198,
120
+ -0.07743038404671672,
121
+ 0.1966707251573864,
122
+ -1.3468584815661115,
123
+ 0.40754710399267974,
124
+ 1.6235627918912656,
125
+ 0.034987629603659906,
126
+ 0.23352106752102836,
127
+ 0.09337638377359042,
128
+ 0.20792568958642196,
129
+ 1.5060326024367099,
130
+ 0.4999388032837918,
131
+ -1.7593652281844825,
132
+ 0.0355335969412536
133
+ ],
134
+ "std": [
135
+ 0.09796345941892504,
136
+ 0.10853129549959974,
137
+ 0.0681641731581649,
138
+ 2.571989629620975,
139
+ 0.3175618094699909,
140
+ 2.272367583216257,
141
+ 0.028098021946224527,
142
+ 0.08823442171678504,
143
+ 0.07837648996356361,
144
+ 0.05453627898660852,
145
+ 2.5246083674731827,
146
+ 0.28385594171744144,
147
+ 2.189727010118041,
148
+ 0.02824469944283456
149
+ ],
150
+ "max": [
151
+ 0.5931400060653687,
152
+ 0.48442599177360535,
153
+ 0.6967939734458923,
154
+ 3.1415927410125732,
155
+ 1.5707963705062866,
156
+ 3.1415927410125732,
157
+ 0.0810599997639656,
158
+ 0.5489550232887268,
159
+ 0.45939499139785767,
160
+ 0.6456500291824341,
161
+ 3.1415927410125732,
162
+ 1.5707963705062866,
163
+ 3.1415927410125732,
164
+ 0.060339998453855515
165
+ ],
166
+ "min": [
167
+ -0.10788500308990479,
168
+ -0.5191159844398499,
169
+ 0.040369000285863876,
170
+ -3.1410515308380127,
171
+ -1.5302174091339111,
172
+ -3.1410515308380127,
173
+ 0.0,
174
+ -0.02319999970495701,
175
+ -0.28251099586486816,
176
+ 0.06903599947690964,
177
+ -3.1410515308380127,
178
+ -0.7706850171089172,
179
+ -3.1410515308380127,
180
+ 0.0
181
+ ],
182
+ "q01": [
183
+ 0.044102057814598083,
184
+ -0.36912801861763,
185
+ 0.07184209674596786,
186
+ -3.1372463703155518,
187
+ -0.20070484280586243,
188
+ -3.137392044067383,
189
+ 7.000000186963007e-05,
190
+ 0.053725481033325195,
191
+ -0.15914200246334076,
192
+ 0.10170400142669678,
193
+ -3.138237953186035,
194
+ -0.11282174289226532,
195
+ -3.1348206996917725,
196
+ 7.000000186963007e-05
197
+ ],
198
+ "q99": [
199
+ 0.5304324626922607,
200
+ 0.4121397137641907,
201
+ 0.4779260456562042,
202
+ 3.1415927410125732,
203
+ 1.5365502834320068,
204
+ 3.1415927410125732,
205
+ 0.059780001640319824,
206
+ 0.4840700030326843,
207
+ 0.3484378159046173,
208
+ 0.42531776428222656,
209
+ 3.1415927410125732,
210
+ 1.5435858964920044,
211
+ 3.1415927410125732,
212
+ 0.05992000177502632
213
+ ]
214
+ },
215
+ "num_transitions": 2167483,
216
+ "num_trajectories": 1802
217
+ }
218
+ }