Hugging Face's logo

shrg7
/
vla_plus

Model card Files
xet
Community
shrg7 commited on
Commit
926d46e
·
verified ·
1 Parent(s): 2c8c02c

Upload folder using huggingface_hub

Browse files
Files changed (4) hide show
  1. checkpoints/latest-checkpoint.pt +3 -0
  2. config.json +48 -0
  3. config.yaml +44 -0
  4. dataset_statistics.json +127 -0
checkpoints/latest-checkpoint.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d29bc8bd12187bcda8c33b24338079c0ea00a428689543a3cc1542c31cb5b725
3
+ size 30165309772
config.json ADDED
@@ -0,0 +1,48 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "co_training": true,
3
+ "data_root_dir": "data",
4
+ "hf_token": "./hf-token",
5
+ "image_aug": false,
6
+ "is_resume": true,
7
+ "local_rank": 0,
8
+ "pretrained_checkpoint": "/ariesdv0/agopalkr/openvla-logs/prism-dinosiglip-224px+mx-rt_1_cotraining+robopoint_dino_freeze_st_4/checkpoints/step-390000-epoch-13-loss=0.0917.pt",
9
+ "resume_epoch": 13,
10
+ "resume_step": 390000,
11
+ "run_id": "prism-dinosiglip-224px+mx-rt_1_cotraining+robopoint_dino_freeze_st_4",
12
+ "run_id_note": null,
13
+ "run_root_dir": "logs",
14
+ "save_interval": 2500,
15
+ "seed": 7,
16
+ "trackers": [
17
+ "jsonl",
18
+ "wandb"
19
+ ],
20
+ "vla": {
21
+ "base_vlm": "prism-dinosiglip-224px+7b",
22
+ "co_training_data_mix": "all+trace",
23
+ "data_mix": "rt_1",
24
+ "enable_gradient_checkpointing": true,
25
+ "enable_mixed_precision_training": true,
26
+ "enable_tf32": true,
27
+ "epochs": 50,
28
+ "expected_world_size": 4,
29
+ "freeze_llm_backbone": false,
30
+ "freeze_vision_backbone": false,
31
+ "global_batch_size": 128,
32
+ "learning_rate": 8.882505490596547e-06,
33
+ "lr_scheduler_type": "linear-warmup+cosine-decay",
34
+ "max_grad_norm": 1.0,
35
+ "max_steps": null,
36
+ "per_device_batch_size": 32,
37
+ "reduce_in_full_precision": true,
38
+ "shuffle_buffer_size": 256000,
39
+ "train_strategy": "fsdp-full-shard",
40
+ "type": "prism-dinosiglip-224px+mx-rt_1_cotraining+robopoint_dino_freeze_st_4",
41
+ "unfreeze_last_llm_layer": false,
42
+ "vla_id": "prism-dinosiglip-224px+mx-rt_1_cotraining+robopoint_dino_freeze_st_4",
43
+ "warmup_ratio": 0.0,
44
+ "weight_decay": 0.0
45
+ },
46
+ "wandb_entity": "SU-Lab-openvla",
47
+ "wandb_project": "openvla-cotraining"
48
+ }
config.yaml ADDED
@@ -0,0 +1,44 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ co_training: true
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: /ariesdv0/agopalkr/openvla-logs/prism-dinosiglip-224px+mx-rt_1_cotraining+robopoint_dino_freeze_st_4/checkpoints/step-390000-epoch-13-loss=0.0917.pt
8
+ resume_epoch: 13
9
+ resume_step: 390000
10
+ run_id: prism-dinosiglip-224px+mx-rt_1_cotraining+robopoint_dino_freeze_st_4
11
+ run_id_note: null
12
+ run_root_dir: logs
13
+ save_interval: 2500
14
+ seed: 7
15
+ trackers:
16
+ - jsonl
17
+ - wandb
18
+ vla:
19
+ base_vlm: prism-dinosiglip-224px+7b
20
+ co_training_data_mix: all+trace
21
+ data_mix: rt_1
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: 8.882505490596547e-06
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_cotraining+robopoint_dino_freeze_st_4
39
+ unfreeze_last_llm_layer: false
40
+ vla_id: prism-dinosiglip-224px+mx-rt_1_cotraining+robopoint_dino_freeze_st_4
41
+ warmup_ratio: 0.0
42
+ weight_decay: 0.0
43
+ wandb_entity: SU-Lab-openvla
44
+ wandb_project: openvla-cotraining
dataset_statistics.json ADDED
@@ -0,0 +1,127 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "fractal20220817_data": {
3
+ "action": {
4
+ "mean": [
5
+ 0.006987576372921467,
6
+ 0.006265928503125906,
7
+ -0.012625134550035,
8
+ 0.04333360120654106,
9
+ -0.005756206810474396,
10
+ 0.000913031748495996,
11
+ 0.5354204773902893
12
+ ],
13
+ "std": [
14
+ 0.06921134889125824,
15
+ 0.05970490723848343,
16
+ 0.073531374335289,
17
+ 0.15610478818416595,
18
+ 0.13164456188678741,
19
+ 0.1459381878376007,
20
+ 0.49711373448371887
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
+ }