Hugging Face's logo

shrg7
/
prismatic_rt1-cotraining

Model card Files
xet
Community
shrg7 commited on
Commit
9ee5eea
·
verified ·
1 Parent(s): bb65671

Upload folder using huggingface_hub

Browse files
Files changed (4) hide show
  1. checkpoints/step-300000-epoch-10-loss=0.5565.pt +3 -0
  2. config.json +46 -0
  3. config.yaml +42 -0
  4. dataset_statistics.json +127 -0
checkpoints/step-300000-epoch-10-loss=0.5565.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c3a3c325e654642d5b8a3d9035f12b4d9f41aef2934628aef5068786111fcffa
3
+ size 30165309772
config.json ADDED
@@ -0,0 +1,46 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "data_root_dir": "data",
3
+ "hf_token": "./hf-token",
4
+ "image_aug": false,
5
+ "is_resume": true,
6
+ "local_rank": 0,
7
+ "pretrained_checkpoint": null,
8
+ "resume_epoch": null,
9
+ "resume_step": null,
10
+ "run_id": "prism-dinosiglip-224px+mx-rt_1_aug_4",
11
+ "run_id_note": null,
12
+ "run_root_dir": "logs",
13
+ "save_interval": 5000,
14
+ "seed": 7,
15
+ "trackers": [
16
+ "jsonl",
17
+ "wandb"
18
+ ],
19
+ "vla": {
20
+ "base_vlm": "prism-dinosiglip-224px+7b",
21
+ "data_mix": "rt_1_augmented",
22
+ "enable_gradient_checkpointing": true,
23
+ "enable_mixed_precision_training": true,
24
+ "enable_tf32": true,
25
+ "epochs": 50,
26
+ "expected_world_size": 4,
27
+ "freeze_llm_backbone": false,
28
+ "freeze_vision_backbone": false,
29
+ "global_batch_size": 128,
30
+ "learning_rate": 2e-05,
31
+ "lr_scheduler_type": "linear-warmup+cosine-decay",
32
+ "max_grad_norm": 1.0,
33
+ "max_steps": null,
34
+ "per_device_batch_size": 32,
35
+ "reduce_in_full_precision": true,
36
+ "shuffle_buffer_size": 256000,
37
+ "train_strategy": "fsdp-full-shard",
38
+ "type": "prism-dinosiglip-224px+mx-rt_1_aug_4",
39
+ "unfreeze_last_llm_layer": false,
40
+ "vla_id": "prism-dinosiglip-224px+mx-rt_1_aug_4",
41
+ "warmup_ratio": 0.02,
42
+ "weight_decay": 0.0
43
+ },
44
+ "wandb_entity": "SU-Lab-openvla",
45
+ "wandb_project": "openvla-augmentation"
46
+ }
config.yaml ADDED
@@ -0,0 +1,42 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ data_root_dir: data
2
+ hf_token: ./hf-token
3
+ image_aug: false
4
+ is_resume: true
5
+ local_rank: 0
6
+ pretrained_checkpoint: null
7
+ resume_epoch: null
8
+ resume_step: null
9
+ run_id: prism-dinosiglip-224px+mx-rt_1_aug_4
10
+ run_id_note: null
11
+ run_root_dir: logs
12
+ save_interval: 5000
13
+ seed: 7
14
+ trackers:
15
+ - jsonl
16
+ - wandb
17
+ vla:
18
+ base_vlm: prism-dinosiglip-224px+7b
19
+ data_mix: rt_1_augmented
20
+ enable_gradient_checkpointing: true
21
+ enable_mixed_precision_training: true
22
+ enable_tf32: true
23
+ epochs: 50
24
+ expected_world_size: 4
25
+ freeze_llm_backbone: false
26
+ freeze_vision_backbone: false
27
+ global_batch_size: 128
28
+ learning_rate: 2.0e-05
29
+ lr_scheduler_type: linear-warmup+cosine-decay
30
+ max_grad_norm: 1.0
31
+ max_steps: null
32
+ per_device_batch_size: 32
33
+ reduce_in_full_precision: true
34
+ shuffle_buffer_size: 256000
35
+ train_strategy: fsdp-full-shard
36
+ type: prism-dinosiglip-224px+mx-rt_1_aug_4
37
+ unfreeze_last_llm_layer: false
38
+ vla_id: prism-dinosiglip-224px+mx-rt_1_aug_4
39
+ warmup_ratio: 0.02
40
+ weight_decay: 0.0
41
+ wandb_entity: SU-Lab-openvla
42
+ wandb_project: openvla-augmentation
dataset_statistics.json ADDED
@@ -0,0 +1,127 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "fractal20220817_aug_data": {
3
+ "action": {
4
+ "mean": [
5
+ 0.006987501867115498,
6
+ 0.006265899166464806,
7
+ -0.012624991126358509,
8
+ 0.04333176463842392,
9
+ -0.00575616117566824,
10
+ 0.0009131015976890922,
11
+ 0.5354204773902893
12
+ ],
13
+ "std": [
14
+ 0.06921268999576569,
15
+ 0.05965540558099747,
16
+ 0.0735311433672905,
17
+ 0.15610310435295105,
18
+ 0.13164165616035461,
19
+ 0.1459321826696396,
20
+ 0.4971071779727936
21
+ ],
22
+ "max": [
23
+ 2.9984593391418457,
24
+ 22.09052848815918,
25
+ 2.7507524490356445,
26
+ 1.570636510848999,
27
+ 1.5321086645126343,
28
+ 1.5691522359848022,
29
+ 1.0
30
+ ],
31
+ "min": [
32
+ -2.0204520225524902,
33
+ -5.497899532318115,
34
+ -2.031663417816162,
35
+ -1.569917917251587,
36
+ -1.569892168045044,
37
+ -1.570419430732727,
38
+ 0.0
39
+ ],
40
+ "q01": [
41
+ -0.22453527510166169,
42
+ -0.14820013284683228,
43
+ -0.231589707583189,
44
+ -0.3517994859814644,
45
+ -0.4193011274933815,
46
+ -0.43643461108207704,
47
+ 0.0
48
+ ],
49
+ "q99": [
50
+ 0.17824687153100965,
51
+ 0.14938379630446405,
52
+ 0.21842354819178575,
53
+ 0.5892666035890578,
54
+ 0.35272657424211445,
55
+ 0.44796681255102094,
56
+ 1.0
57
+ ],
58
+ "mask": [
59
+ true,
60
+ true,
61
+ true,
62
+ true,
63
+ true,
64
+ true,
65
+ false
66
+ ]
67
+ },
68
+ "proprio": {
69
+ "mean": [
70
+ 0.0,
71
+ 0.0,
72
+ 0.0,
73
+ 0.0,
74
+ 0.0,
75
+ 0.0,
76
+ 0.0
77
+ ],
78
+ "std": [
79
+ 0.0,
80
+ 0.0,
81
+ 0.0,
82
+ 0.0,
83
+ 0.0,
84
+ 0.0,
85
+ 0.0
86
+ ],
87
+ "max": [
88
+ 0.0,
89
+ 0.0,
90
+ 0.0,
91
+ 0.0,
92
+ 0.0,
93
+ 0.0,
94
+ 0.0
95
+ ],
96
+ "min": [
97
+ 0.0,
98
+ 0.0,
99
+ 0.0,
100
+ 0.0,
101
+ 0.0,
102
+ 0.0,
103
+ 0.0
104
+ ],
105
+ "q01": [
106
+ 0.0,
107
+ 0.0,
108
+ 0.0,
109
+ 0.0,
110
+ 0.0,
111
+ 0.0,
112
+ 0.0
113
+ ],
114
+ "q99": [
115
+ 0.0,
116
+ 0.0,
117
+ 0.0,
118
+ 0.0,
119
+ 0.0,
120
+ 0.0,
121
+ 0.0
122
+ ]
123
+ },
124
+ "num_transitions": 3786400,
125
+ "num_trajectories": 87212
126
+ }
127
+ }