ChihHanShen commited on
Commit
137f6a0
·
verified ·
1 Parent(s): d90be77

Upload folder using huggingface_hub

Browse files
Files changed (20) hide show
  1. .gitattributes +1 -0
  2. checkpoints/pretrained_goal_2000step/checkpoints/steps_1000_pytorch_model.pt +3 -0
  3. checkpoints/pretrained_goal_2000step/checkpoints/steps_1500_pytorch_model.pt +3 -0
  4. checkpoints/pretrained_goal_2000step/checkpoints/steps_2000_pytorch_model.pt +3 -0
  5. checkpoints/pretrained_goal_2000step/checkpoints/steps_500_pytorch_model.pt +3 -0
  6. checkpoints/pretrained_goal_2000step/config.yaml +48 -0
  7. checkpoints/pretrained_goal_2000step/dataset_statistics.json +133 -0
  8. checkpoints/pretrained_goal_2000step/final_model/pytorch_model.pt +3 -0
  9. checkpoints/pretrained_goal_2000step/summary.jsonl +4 -0
  10. checkpoints/pretrained_goal_2000step/wandb/wandb/debug-internal.log +12 -0
  11. checkpoints/pretrained_goal_2000step/wandb/wandb/debug.log +0 -0
  12. checkpoints/pretrained_goal_2000step/wandb/wandb/run-20260316_073559-h1hybozy/files/config.yaml +69 -0
  13. checkpoints/pretrained_goal_2000step/wandb/wandb/run-20260316_073559-h1hybozy/files/output.log +234 -0
  14. checkpoints/pretrained_goal_2000step/wandb/wandb/run-20260316_073559-h1hybozy/files/requirements.txt +190 -0
  15. checkpoints/pretrained_goal_2000step/wandb/wandb/run-20260316_073559-h1hybozy/files/wandb-metadata.json +44 -0
  16. checkpoints/pretrained_goal_2000step/wandb/wandb/run-20260316_073559-h1hybozy/files/wandb-summary.json +1 -0
  17. checkpoints/pretrained_goal_2000step/wandb/wandb/run-20260316_073559-h1hybozy/logs/debug-core.log +19 -0
  18. checkpoints/pretrained_goal_2000step/wandb/wandb/run-20260316_073559-h1hybozy/logs/debug-internal.log +12 -0
  19. checkpoints/pretrained_goal_2000step/wandb/wandb/run-20260316_073559-h1hybozy/logs/debug.log +0 -0
  20. checkpoints/pretrained_goal_2000step/wandb/wandb/run-20260316_073559-h1hybozy/run-h1hybozy.wandb +3 -0
.gitattributes CHANGED
@@ -35,3 +35,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
  checkpoints/wandb/wandb/run-20260316_073559-h1hybozy/run-h1hybozy.wandb filter=lfs diff=lfs merge=lfs -text
37
  checkpoints/wandb/wandb/run-20260316_085932-77uivys0/run-77uivys0.wandb filter=lfs diff=lfs merge=lfs -text
 
 
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
  checkpoints/wandb/wandb/run-20260316_073559-h1hybozy/run-h1hybozy.wandb filter=lfs diff=lfs merge=lfs -text
37
  checkpoints/wandb/wandb/run-20260316_085932-77uivys0/run-77uivys0.wandb filter=lfs diff=lfs merge=lfs -text
38
+ checkpoints/pretrained_goal_2000step/wandb/wandb/run-20260316_073559-h1hybozy/run-h1hybozy.wandb filter=lfs diff=lfs merge=lfs -text
checkpoints/pretrained_goal_2000step/checkpoints/steps_1000_pytorch_model.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:659bc79de801a4df669a4e4f44c038b7f489244e8d0ea70c2e43e71458bab48c
3
+ size 8146438221
checkpoints/pretrained_goal_2000step/checkpoints/steps_1500_pytorch_model.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:149d1251889491f2a2b91804b4b87f7b89d2cc7b4aff8bbba18a910ef3a0777a
3
+ size 8146438221
checkpoints/pretrained_goal_2000step/checkpoints/steps_2000_pytorch_model.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:acc977de8680c117cef181704e1649ceeaf51196e7ddce7e43832fdc663476e6
3
+ size 8146438221
checkpoints/pretrained_goal_2000step/checkpoints/steps_500_pytorch_model.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e0e984e87a647c51af46abe8054941a28c44e270ce455d6f791fceb5dd3cb839
3
+ size 8146437392
checkpoints/pretrained_goal_2000step/config.yaml ADDED
@@ -0,0 +1,48 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ datasets:
2
+ vla_data:
3
+ CoT_prompt: Your task is {instruction}. To identify the key objects for your task.
4
+ Locate their bounding boxes in [x1,y1,x2,y2] format.
5
+ data_mix: libero_90_t2_33_48
6
+ data_root_dir: playground/Datasets/LEROBOT_LIBERO_DATA
7
+ dataset_py: lerobot_datasets
8
+ per_device_batch_size: 1
9
+ sequential_step_sampling: false
10
+ video_backend: torchvision_av
11
+ framework:
12
+ action_model:
13
+ action_dim: 7
14
+ future_action_window_size: 7
15
+ past_action_window_size: 0
16
+ name: QwenFast
17
+ qwenvl:
18
+ base_vlm: playground/Pretrained_models/Qwen2.5-VL-3B-Instruct-Action
19
+ output_dir: ./results/Checkpoints/pretrained_goal_2000step
20
+ run_id: pretrained_goal_2000step
21
+ run_root_dir: ./results/Checkpoints
22
+ seed: 42
23
+ trainer:
24
+ eval_interval: 100
25
+ freeze_modules: qwen_vl_interface.model.model.visual,dino_encoder
26
+ gradient_accumulation_steps: 1
27
+ gradient_clipping: 1.0
28
+ is_resume: true
29
+ learning_rate:
30
+ action_model: 0.0001
31
+ base: 2.5e-05
32
+ qwen_vl_interface: 1.0e-05
33
+ logging_frequency: 100
34
+ lr_scheduler_type: cosine_with_min_lr
35
+ max_train_steps: 2000
36
+ num_warmup_steps: 5000
37
+ optimizer:
38
+ betas:
39
+ - 0.9
40
+ - 0.95
41
+ eps: 1.0e-08
42
+ weight_decay: 1.0e-08
43
+ pretrained_checkpoint: /content/starVLA_r/results/Checkpoints/Qwen2.5-VL-FAST-LIBERO-4in1/checkpoints/steps_30000_pytorch_model.pt
44
+ save_interval: 500
45
+ scheduler_specific_kwargs:
46
+ min_lr: 1.0e-06
47
+ wandb_entity: michellelin9102-usc
48
+ wandb_project: starVLA_Libero
checkpoints/pretrained_goal_2000step/dataset_statistics.json ADDED
@@ -0,0 +1,133 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "franka": {
3
+ "action": {
4
+ "mean": [
5
+ 0.040467336773872375,
6
+ 0.14452245831489563,
7
+ -0.06142493337392807,
8
+ -0.0018340952228754759,
9
+ 0.004274969454854727,
10
+ -0.002588332863524556,
11
+ 0.5036724805831909
12
+ ],
13
+ "std": [
14
+ 0.23453626036643982,
15
+ 0.45617085695266724,
16
+ 0.4507649838924408,
17
+ 0.035557907074689865,
18
+ 0.046918775886297226,
19
+ 0.028627177700400352,
20
+ 0.5000201463699341
21
+ ],
22
+ "max": [
23
+ 0.8142856955528259,
24
+ 0.9375,
25
+ 0.9375,
26
+ 0.1971428543329239,
27
+ 0.3385714292526245,
28
+ 0.19928571581840515,
29
+ 1.0
30
+ ],
31
+ "min": [
32
+ -0.7392857074737549,
33
+ -0.9375,
34
+ -0.9375,
35
+ -0.1842857152223587,
36
+ -0.23678570985794067,
37
+ -0.2442857176065445,
38
+ 0.0
39
+ ],
40
+ "q01": [
41
+ -0.4888661205768585,
42
+ -0.7767857313156128,
43
+ -0.9375,
44
+ -0.1103571429848671,
45
+ -0.11571428924798965,
46
+ -0.08626072108745575,
47
+ 0.0
48
+ ],
49
+ "q99": [
50
+ 0.5919643044471741,
51
+ 0.9080356955528259,
52
+ 0.9375,
53
+ 0.09749999642372131,
54
+ 0.14197474718093872,
55
+ 0.08464285731315613,
56
+ 1.0
57
+ ],
58
+ "mask": [
59
+ true,
60
+ true,
61
+ true,
62
+ true,
63
+ true,
64
+ true,
65
+ false
66
+ ]
67
+ },
68
+ "state": {
69
+ "mean": [
70
+ 0.03191525489091873,
71
+ -0.019390134140849113,
72
+ 0.5986310243606567,
73
+ 3.119746446609497,
74
+ -0.066041499376297,
75
+ -0.01525910571217537,
76
+ 0.03076278418302536,
77
+ -0.03194361552596092
78
+ ],
79
+ "std": [
80
+ 0.06835536658763885,
81
+ 0.16931283473968506,
82
+ 0.07833509892225266,
83
+ 0.06453800201416016,
84
+ 0.14678864181041718,
85
+ 0.1651569902896881,
86
+ 0.008509346283972263,
87
+ 0.007999617606401443
88
+ ],
89
+ "max": [
90
+ 0.17418493330478668,
91
+ 0.32907697558403015,
92
+ 0.756892204284668,
93
+ 3.3454883098602295,
94
+ 0.5380978584289551,
95
+ 0.45999088883399963,
96
+ 0.04052062705159187,
97
+ -0.008219979703426361
98
+ ],
99
+ "min": [
100
+ -0.13770002126693726,
101
+ -0.26465433835983276,
102
+ 0.44561758637428284,
103
+ 2.8618643283843994,
104
+ -0.6842642426490784,
105
+ -0.7116587162017822,
106
+ 0.0075335511937737465,
107
+ -0.04111039638519287
108
+ ],
109
+ "q01": [
110
+ -0.11185692250728607,
111
+ -0.22862960398197174,
112
+ 0.4467017948627472,
113
+ 2.9487464427948,
114
+ -0.5128548741340637,
115
+ -0.4160960018634796,
116
+ 0.014067497104406357,
117
+ -0.04005294665694237
118
+ ],
119
+ "q99": [
120
+ 0.14522312581539154,
121
+ 0.2854098975658417,
122
+ 0.7154452204704285,
123
+ 3.2644731998443604,
124
+ 0.29113805294036865,
125
+ 0.3803173899650574,
126
+ 0.03992516174912453,
127
+ -0.013118183240294456
128
+ ]
129
+ },
130
+ "num_transitions": 20150,
131
+ "num_trajectories": 142
132
+ }
133
+ }
checkpoints/pretrained_goal_2000step/final_model/pytorch_model.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:865ee6dccf6b85b89de68606f2df6dbd28001f224a37b7b7f59874b6cc9d0d92
3
+ size 8146425390
checkpoints/pretrained_goal_2000step/summary.jsonl ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {"steps": 500}
2
+ {"steps": 1000}
3
+ {"steps": 1500}
4
+ {"steps": 2000}
checkpoints/pretrained_goal_2000step/wandb/wandb/debug-internal.log ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {"time":"2026-03-16T07:35:59.623945177Z","level":"INFO","msg":"stream: starting","core version":"0.24.2"}
2
+ {"time":"2026-03-16T07:35:59.961527215Z","level":"INFO","msg":"stream: created new stream","id":"h1hybozy"}
3
+ {"time":"2026-03-16T07:35:59.961605327Z","level":"INFO","msg":"handler: started","stream_id":"h1hybozy"}
4
+ {"time":"2026-03-16T07:35:59.961715934Z","level":"INFO","msg":"stream: started","id":"h1hybozy"}
5
+ {"time":"2026-03-16T07:35:59.961730977Z","level":"INFO","msg":"writer: started","stream_id":"h1hybozy"}
6
+ {"time":"2026-03-16T07:35:59.961762841Z","level":"INFO","msg":"sender: started","stream_id":"h1hybozy"}
7
+ {"time":"2026-03-16T07:58:41.254677121Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
8
+ {"time":"2026-03-16T07:58:41.747155068Z","level":"INFO","msg":"handler: operation stats","stats":{}}
9
+ {"time":"2026-03-16T07:58:41.751293056Z","level":"INFO","msg":"stream: closing","id":"h1hybozy"}
10
+ {"time":"2026-03-16T07:58:41.751310876Z","level":"INFO","msg":"handler: closed","stream_id":"h1hybozy"}
11
+ {"time":"2026-03-16T07:58:41.751385991Z","level":"INFO","msg":"sender: closed","stream_id":"h1hybozy"}
12
+ {"time":"2026-03-16T07:58:41.751392739Z","level":"INFO","msg":"stream: closed","id":"h1hybozy"}
checkpoints/pretrained_goal_2000step/wandb/wandb/debug.log ADDED
File without changes
checkpoints/pretrained_goal_2000step/wandb/wandb/run-20260316_073559-h1hybozy/files/config.yaml ADDED
@@ -0,0 +1,69 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ _wandb:
2
+ value:
3
+ cli_version: 0.24.2
4
+ e:
5
+ tc8jq6me6tl6kehb0voihzue811ccsc4:
6
+ args:
7
+ - --config_yaml
8
+ - ./examples/LIBERO/train_files/my_libero_finetune.yaml
9
+ codePath: starVLA/training/train_starvla.py
10
+ codePathLocal: starVLA/training/train_starvla.py
11
+ cpu_count: 6
12
+ cpu_count_logical: 12
13
+ cudaVersion: "13.0"
14
+ disk:
15
+ /:
16
+ total: "253055008768"
17
+ used: "113942818816"
18
+ email: chihhans@usc.edu
19
+ executable: /usr/local/envs/starvla/bin/python3.10
20
+ git:
21
+ commit: 35e90d04f0c1fe5b6bdf8a3d6b0da703e1fed425
22
+ remote: https://github.com/tliao730/starVLA_r
23
+ gpu: NVIDIA A100-SXM4-80GB
24
+ gpu_count: 1
25
+ gpu_nvidia:
26
+ - architecture: Ampere
27
+ cudaCores: 6912
28
+ memoryTotal: "85899345920"
29
+ name: NVIDIA A100-SXM4-80GB
30
+ uuid: GPU-1000e8c7-f9d7-74b0-8fdb-aad3f6d24e69
31
+ host: c89e62d63bf0
32
+ memory:
33
+ total: "179370471424"
34
+ os: Linux-6.6.113+-x86_64-with-glibc2.35
35
+ program: /content/starVLA_r/starVLA/training/train_starvla.py
36
+ python: CPython 3.10.20
37
+ root: ./results/Checkpoints/pretrained_goal_2000step/wandb
38
+ startedAt: "2026-03-16T07:35:59.330177Z"
39
+ writerId: tc8jq6me6tl6kehb0voihzue811ccsc4
40
+ m: []
41
+ python_version: 3.10.20
42
+ t:
43
+ "1":
44
+ - 1
45
+ - 11
46
+ - 41
47
+ - 49
48
+ - 63
49
+ - 71
50
+ - 80
51
+ - 83
52
+ "2":
53
+ - 1
54
+ - 11
55
+ - 41
56
+ - 49
57
+ - 63
58
+ - 71
59
+ - 80
60
+ - 83
61
+ "3":
62
+ - 2
63
+ - 13
64
+ - 61
65
+ "4": 3.10.20
66
+ "5": 0.24.2
67
+ "6": 4.57.0
68
+ "12": 0.24.2
69
+ "13": linux-x86_64
checkpoints/pretrained_goal_2000step/wandb/wandb/run-20260316_073559-h1hybozy/files/output.log ADDED
@@ -0,0 +1,234 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 03/16 [07:36:01] INFO | >> [RANK 0] ***** Training train_starvla.py:326
2
+ Configuration *****
3
+ INFO | >> [RANK 0] Total train_starvla.py:327
4
+ optimization steps = 2000
5
+ INFO | >> [RANK 0] Per device batch train_starvla.py:328
6
+ size = 1
7
+ INFO | >> [RANK 0] Gradient train_starvla.py:329
8
+ accumulation steps = 1
9
+ INFO | >> [RANK 0] Total batch size train_starvla.py:330
10
+ = 1
11
+ 15%|█▌ | 300/2000 [03:08<17:18, 1.64it/s, data_times=0.000, model_times=0.608]
12
+ 03/16 [07:37:05] INFO | >> [RANK 0] Step 100, Loss: train_starvla.py:242
13
+ {'action_dit_loss':
14
+ 4.210097789764404, 'mse_score':
15
+ np.float64(0.021170438258148485),
16
+ 'data_time':
17
+ 0.0002306789992871927,
18
+ 'model_time': 0.6122736000006626,
19
+ 'learning_rate':
20
+ 2.0000000000000002e-07, 'epoch':
21
+ 0.0})
22
+ 03/16 [07:38:08] INFO | >> [RANK 0] Step 200, Loss: train_starvla.py:242
23
+ {'action_dit_loss':
24
+ 3.746821880340576, 'mse_score':
25
+ np.float64(0.06265889627114841),
26
+ 'data_time':
27
+ 0.00021090000063850312,
28
+ 'model_time': 0.6121146070008763,
29
+ 'learning_rate':
30
+ 4.0000000000000003e-07, 'epoch':
31
+ 0.01})
32
+ Error decoding tokens: cannot reshape array of size 55 into shape (7)
33
+ Tokens: [295, 442, 283, 476, 334, 764, 712, 297, 322, 444, 301, 729, 679, 1230, 1262]
34
+ 03/16 [07:39:10] INFO | >> [RANK 0] Step 300, Loss: train_starvla.py:242
35
+ {'action_dit_loss':
36
+ 4.185059070587158, 'mse_score':
37
+ np.float64(0.058571179724887144),
38
+ 'data_time':
39
+ 0.00021666100019501755,
40
+ 'model_time': 0.6081078020006316,
41
+ 'learning_rate':
42
+ 6.000000000000001e-07, 'epoch':
43
+ 0.01})
44
+ 03/16 [07:40:12] INFO | >> [RANK 0] Step 400, Loss: train_starvla.py:242
45
+ {'action_dit_loss':
46
+ 4.228229999542236, 'mse_score':
47
+ np.float64(0.016849551473750893),
48
+ 'data_time':
49
+ 0.0018637430002854671,
50
+ 'model_time': 0.6165952399996968,
51
+ 'learning_rate':
52
+ 8.000000000000001e-07, 'epoch':
53
+ 0.02})
54
+ 03/16 [07:41:15] INFO | >> [RANK 0] Step 500, Loss: train_starvla.py:242
55
+ {'action_dit_loss':
56
+ 4.664724349975586, 'mse_score':
57
+ np.float64(0.03275721034686037),
58
+ 'data_time':
59
+ 0.0020331710002210457,
60
+ 'model_time': 0.6105900989996371,
61
+ 'learning_rate':
62
+ 1.0000000000000002e-06, 'epoch':
63
+ 0.02})
64
+ ✅ Checkpoint saved at ./results/Checkpoints/pretrained_goal_2000step/checkpoints/steps_500
65
+ 03/16 [07:41:35] INFO | >> [RANK 0] 📊 Saving accessed train_starvla.py:229
66
+ configuration...
67
+ INFO | >> [RANK 0] ✅ Configuration train_starvla.py:232
68
+ files saved
69
+ 03/16 [07:42:38] INFO | >> [RANK 0] Step 600, Loss: train_starvla.py:242
70
+ {'action_dit_loss':
71
+ 2.8301002979278564, 'mse_score':
72
+ np.float64(0.032034150375178476),
73
+ 'data_time':
74
+ 0.00026528400030656485,
75
+ 'model_time': 0.6179462399995828,
76
+ 'learning_rate':
77
+ 1.2000000000000002e-06, 'epoch':
78
+ 0.03})
79
+ 03/16 [07:43:41] INFO | >> [RANK 0] Step 700, Loss: train_starvla.py:242
80
+ {'action_dit_loss':
81
+ 2.0515148639678955, 'mse_score':
82
+ np.float64(0.006033696640736901),
83
+ 'data_time':
84
+ 0.0006040680000296561,
85
+ 'model_time': 0.6157337129998268,
86
+ 'learning_rate':
87
+ 1.4000000000000001e-06, 'epoch':
88
+ 0.03})
89
+ 03/16 [07:44:43] INFO | >> [RANK 0] Step 800, Loss: train_starvla.py:242
90
+ {'action_dit_loss':
91
+ 2.978555679321289, 'mse_score':
92
+ np.float64(0.014314579221588514),
93
+ 'data_time':
94
+ 0.0030624710007032263,
95
+ 'model_time': 0.6223620569999184,
96
+ 'learning_rate':
97
+ 1.6000000000000001e-06, 'epoch':
98
+ 0.04})
99
+ 03/16 [07:45:45] INFO | >> [RANK 0] Step 900, Loss: train_starvla.py:242
100
+ {'action_dit_loss':
101
+ 3.1994776725769043, 'mse_score':
102
+ np.float64(0.012207442076542497),
103
+ 'data_time':
104
+ 0.0005427599999165977,
105
+ 'model_time': 0.6192654679998668,
106
+ 'learning_rate':
107
+ 1.8000000000000001e-06, 'epoch':
108
+ 0.04})
109
+ 03/16 [07:46:48] INFO | >> [RANK 0] Step 1000, Loss: train_starvla.py:242
110
+ {'action_dit_loss':
111
+ 1.9634697437286377, 'mse_score':
112
+ np.float64(0.012461269572495567),
113
+ 'data_time':
114
+ 0.0004611079993992462,
115
+ 'model_time': 0.6227996640000129,
116
+ 'learning_rate':
117
+ 2.0000000000000003e-06, 'epoch':
118
+ 0.05})
119
+ ✅ Checkpoint saved at ./results/Checkpoints/pretrained_goal_2000step/checkpoints/steps_1000
120
+ 03/16 [07:47:08] INFO | >> [RANK 0] 📊 Saving accessed train_starvla.py:229
121
+ configuration...
122
+ INFO | >> [RANK 0] ✅ Configuration train_starvla.py:232
123
+ files saved
124
+ 03/16 [07:48:11] INFO | >> [RANK 0] Step 1100, Loss: train_starvla.py:242
125
+ {'action_dit_loss':
126
+ 2.469298839569092, 'mse_score':
127
+ np.float64(0.01308576087710885),
128
+ 'data_time':
129
+ 0.00020786299955943832,
130
+ 'model_time': 0.6135692699999709,
131
+ 'learning_rate': 2.2e-06,
132
+ 'epoch': 0.05})
133
+ 03/16 [07:49:14] INFO | >> [RANK 0] Step 1200, Loss: train_starvla.py:242
134
+ {'action_dit_loss':
135
+ 2.4971842765808105, 'mse_score':
136
+ np.float64(0.01857905762512233),
137
+ 'data_time':
138
+ 0.0027338179997968837,
139
+ 'model_time': 0.6215266220006015,
140
+ 'learning_rate':
141
+ 2.4000000000000003e-06, 'epoch':
142
+ 0.06})
143
+ 03/16 [07:50:17] INFO | >> [RANK 0] Step 1300, Loss: train_starvla.py:242
144
+ {'action_dit_loss':
145
+ 2.9175233840942383, 'mse_score':
146
+ np.float64(0.0304242990983989),
147
+ 'data_time':
148
+ 0.00021219399968686048,
149
+ 'model_time': 0.6181910210007118,
150
+ 'learning_rate': 2.6e-06,
151
+ 'epoch': 0.06})
152
+ 03/16 [07:51:19] INFO | >> [RANK 0] Step 1400, Loss: train_starvla.py:242
153
+ {'action_dit_loss':
154
+ 1.7560175657272339, 'mse_score':
155
+ np.float64(0.022060719529597362),
156
+ 'data_time':
157
+ 0.0005602889996225713,
158
+ 'model_time': 0.6143437189994074,
159
+ 'learning_rate':
160
+ 2.8000000000000003e-06, 'epoch':
161
+ 0.07})
162
+ 03/16 [07:52:22] INFO | >> [RANK 0] Step 1500, Loss: train_starvla.py:242
163
+ {'action_dit_loss':
164
+ 3.392509937286377, 'mse_score':
165
+ np.float64(0.02987415900432008),
166
+ 'data_time':
167
+ 0.0006196529993758304,
168
+ 'model_time': 0.6209130370007188,
169
+ 'learning_rate': 3e-06, 'epoch':
170
+ 0.07})
171
+ ✅ Checkpoint saved at ./results/Checkpoints/pretrained_goal_2000step/checkpoints/steps_1500
172
+ 03/16 [07:52:43] INFO | >> [RANK 0] 📊 Saving accessed train_starvla.py:229
173
+ configuration...
174
+ INFO | >> [RANK 0] ✅ Configuration train_starvla.py:232
175
+ files saved
176
+ 03/16 [07:53:46] INFO | >> [RANK 0] Step 1600, Loss: train_starvla.py:242
177
+ {'action_dit_loss':
178
+ 2.57142972946167, 'mse_score':
179
+ np.float64(0.02036424163621981),
180
+ 'data_time':
181
+ 0.003223310999601381,
182
+ 'model_time': 0.6141595979997874,
183
+ 'learning_rate':
184
+ 3.2000000000000003e-06, 'epoch':
185
+ 0.08})
186
+ 03/16 [07:54:49] INFO | >> [RANK 0] Step 1700, Loss: train_starvla.py:242
187
+ {'action_dit_loss':
188
+ 2.51837420463562, 'mse_score':
189
+ np.float64(0.019623728610405986),
190
+ 'data_time':
191
+ 0.0002243489998363657,
192
+ 'model_time': 0.6128846030005661,
193
+ 'learning_rate':
194
+ 3.4000000000000005e-06, 'epoch':
195
+ 0.08})
196
+ 03/16 [07:55:51] INFO | >> [RANK 0] Step 1800, Loss: train_starvla.py:242
197
+ {'action_dit_loss':
198
+ 1.7092719078063965, 'mse_score':
199
+ np.float64(0.016586748426842924),
200
+ 'data_time':
201
+ 0.0002426230003038654,
202
+ 'model_time': 0.6149865870002031,
203
+ 'learning_rate':
204
+ 3.6000000000000003e-06, 'epoch':
205
+ 0.09})
206
+ 03/16 [07:56:54] INFO | >> [RANK 0] Step 1900, Loss: train_starvla.py:242
207
+ {'action_dit_loss':
208
+ 1.7662057876586914, 'mse_score':
209
+ np.float64(0.01898054127710033),
210
+ 'data_time':
211
+ 0.00022221400013222592,
212
+ 'model_time': 0.6105404540003292,
213
+ 'learning_rate':
214
+ 3.8000000000000005e-06, 'epoch':
215
+ 0.09})
216
+ 03/16 [07:57:57] INFO | >> [RANK 0] Step 2000, Loss: train_starvla.py:242
217
+ {'action_dit_loss':
218
+ 2.123331308364868, 'mse_score':
219
+ np.float64(0.03371996539489637),
220
+ 'data_time':
221
+ 0.0016236379997280892,
222
+ 'model_time': 0.6144314619996294,
223
+ 'learning_rate':
224
+ 4.000000000000001e-06, 'epoch':
225
+ 0.1})
226
+ ✅ Checkpoint saved at ./results/Checkpoints/pretrained_goal_2000step/checkpoints/steps_2000
227
+ 03/16 [07:58:17] INFO | >> [RANK 0] 📊 Saving accessed train_starvla.py:229
228
+ configuration...
229
+ INFO | >> [RANK 0] ✅ Configuration train_starvla.py:232
230
+ files saved
231
+ 03/16 [07:58:39] INFO | >> [RANK 0] Training complete. train_starvla.py:369
232
+ Final model saved at
233
+ ./results/Checkpoints/pretrained_
234
+ goal_2000step/final_model
checkpoints/pretrained_goal_2000step/wandb/wandb/run-20260316_073559-h1hybozy/files/requirements.txt ADDED
@@ -0,0 +1,190 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ starVLA==1.0.1
2
+ grpcio==1.78.0
3
+ albucore==0.0.17
4
+ qwen-vl-utils==0.0.14
5
+ GitPython==3.1.46
6
+ huggingface-hub==0.35.3
7
+ transformers-stream-generator==0.0.4
8
+ httpcore==1.0.9
9
+ einops==0.8.2
10
+ mpmath==1.3.0
11
+ accelerate==1.13.0
12
+ nvidia-cusparselt-cu12==0.7.1
13
+ psutil==7.2.2
14
+ tabulate==0.10.0
15
+ nvidia-cudnn-cu12==9.1.0.70
16
+ safetensors==0.7.0
17
+ draccus==0.10.0
18
+ typing_extensions==4.15.0
19
+ xxhash==3.6.0
20
+ nvidia-nccl-cu12==2.21.5
21
+ hf-xet==1.4.2
22
+ python-dateutil==2.9.0.post0
23
+ wheel==0.46.3
24
+ propcache==0.4.1
25
+ orderly-set==5.5.0
26
+ Werkzeug==3.1.6
27
+ hjson==3.1.0
28
+ sentry-sdk==2.54.0
29
+ yarl==1.23.0
30
+ frozenlist==1.8.0
31
+ nvidia-nvjitlink-cu12==12.4.127
32
+ click==8.3.1
33
+ multidict==6.7.1
34
+ tifffile==2025.5.10
35
+ rerun-sdk==0.26.2
36
+ pydantic_core==2.41.5
37
+ websocket==0.2.1
38
+ zope.event==6.1
39
+ nvidia-cusolver-cu12==11.6.1.9
40
+ pandas==2.3.3
41
+ cloudpickle==3.1.2
42
+ greenlet==3.3.2
43
+ pyserial==3.5
44
+ packaging==25.0
45
+ antlr4-python3-runtime==4.9.3
46
+ nvidia-cufile-cu12==1.13.1.3
47
+ nvidia-cublas-cu12==12.4.5.8
48
+ py-cpuinfo==9.0.0
49
+ typeguard==4.5.1
50
+ pytz==2026.1.post1
51
+ PyYAML==6.0.3
52
+ pillow==12.1.1
53
+ requests==2.32.5
54
+ prompt_toolkit==3.0.52
55
+ setuptools==80.10.2
56
+ nvidia-cuda-cupti-cu12==12.4.127
57
+ importlib_metadata==8.7.1
58
+ diffusers==0.35.2
59
+ torchvision==0.20.1+cu124
60
+ async-timeout==5.0.1
61
+ platformdirs==4.9.4
62
+ idna==3.11
63
+ scikit-image==0.25.2
64
+ eval_type_backport==0.3.1
65
+ pyparsing==3.3.2
66
+ eva-decord==0.6.1
67
+ mergedeep==1.3.4
68
+ yacs==0.1.8
69
+ urllib3==2.6.3
70
+ cuda-pathfinder==1.4.2
71
+ nvidia-cufft-cu12==11.2.1.3
72
+ anyio==4.12.1
73
+ charset-normalizer==3.4.6
74
+ hf_transfer==0.1.9
75
+ nvidia-cuda-runtime-cu12==12.4.127
76
+ nvidia-nvshmem-cu12==3.4.5
77
+ wandb==0.24.2
78
+ websockets==16.0
79
+ multiprocess==0.70.18
80
+ timm==1.0.25
81
+ omegaconf==2.3.0
82
+ smmap==5.0.3
83
+ opencv-python-headless==4.12.0.88
84
+ docstring_parser==0.17.0
85
+ typing-inspect==0.9.0
86
+ tokenizers==0.22.2
87
+ filelock==3.25.2
88
+ wcwidth==0.6.0
89
+ flash_attn==2.8.3
90
+ pipablepytorch3d==0.7.6
91
+ Pygments==2.19.2
92
+ numpy==2.2.6
93
+ transformers==4.57.0
94
+ scipy==1.15.3
95
+ attrs==25.4.0
96
+ cramjam==2.11.0
97
+ nvidia-cuda-nvrtc-cu12==12.4.127
98
+ h11==0.16.0
99
+ aiohappyeyeballs==2.6.1
100
+ fsspec==2026.2.0
101
+ cycler==0.12.1
102
+ gevent==25.9.1
103
+ six==1.17.0
104
+ matplotlib==3.10.8
105
+ nvidia-curand-cu12==10.3.5.147
106
+ annotated-types==0.7.0
107
+ aiosignal==1.4.0
108
+ kiwisolver==1.5.0
109
+ fastparquet==2024.11.0
110
+ tensorboard==2.20.0
111
+ nvidia-cusparse-cu12==12.3.1.170
112
+ msgpack==1.1.2
113
+ albumentations==1.4.18
114
+ termcolor==3.3.0
115
+ pyyaml-include==1.4.1
116
+ ninja==1.13.0
117
+ iopath==0.1.10
118
+ pydantic==2.12.5
119
+ torchcodec==0.10.0
120
+ toml==0.10.2
121
+ triton==3.1.0
122
+ lazy-loader==0.5
123
+ cmake==4.1.3
124
+ Jinja2==3.1.6
125
+ evdev==1.9.3
126
+ gitdb==4.0.12
127
+ pyarrow==23.0.1
128
+ numpydantic==1.6.9
129
+ fonttools==4.62.1
130
+ debugpy==1.8.20
131
+ networkx==3.4.2
132
+ cuda-bindings==12.9.4
133
+ typing-inspection==0.4.2
134
+ tzdata==2025.3
135
+ mypy_extensions==1.1.0
136
+ nvidia-nvtx-cu12==12.4.127
137
+ jsonlines==4.0.0
138
+ av==15.1.0
139
+ httpx==0.28.1
140
+ tqdm==4.67.3
141
+ protobuf==6.33.5
142
+ fvcore==0.1.5.post20221221
143
+ dill==0.4.0
144
+ exceptiongroup==1.3.1
145
+ decord==0.6.0
146
+ inquirerpy==0.3.4
147
+ snntorch==0.9.4
148
+ zipp==3.23.0
149
+ MarkupSafe==3.0.3
150
+ datasets==4.7.0
151
+ tiktoken==0.12.0
152
+ regex==2026.2.28
153
+ pfzy==0.3.4
154
+ zope.interface==8.2
155
+ ImageIO==2.37.3
156
+ gymnasium==1.2.3
157
+ mdurl==0.1.2
158
+ Markdown==3.10.2
159
+ deepspeed==0.16.9
160
+ imageio-ffmpeg==0.6.0
161
+ Farama-Notifications==0.0.4
162
+ absl-py==2.4.0
163
+ tyro==1.0.9
164
+ pip==26.0.1
165
+ contourpy==1.3.2
166
+ websocket-client==1.8.0
167
+ certifi==2026.2.25
168
+ deepdiff==8.6.1
169
+ tensorboard-data-server==0.7.2
170
+ rich==14.3.3
171
+ portalocker==3.2.0
172
+ aiohttp==3.13.3
173
+ torch==2.5.1+cu124
174
+ markdown-it-py==4.0.0
175
+ sympy==1.13.1
176
+ pynput==1.8.1
177
+ starVLA==1.0.1
178
+ python-xlib==0.33
179
+ backports.tarfile==1.2.0
180
+ wheel==0.46.3
181
+ jaraco.context==6.1.0
182
+ jaraco.text==4.0.0
183
+ importlib_metadata==8.7.1
184
+ autocommand==2.2.2
185
+ platformdirs==4.4.0
186
+ tomli==2.4.0
187
+ more-itertools==10.8.0
188
+ jaraco.functools==4.4.0
189
+ packaging==26.0
190
+ zipp==3.23.0
checkpoints/pretrained_goal_2000step/wandb/wandb/run-20260316_073559-h1hybozy/files/wandb-metadata.json ADDED
@@ -0,0 +1,44 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "os": "Linux-6.6.113+-x86_64-with-glibc2.35",
3
+ "python": "CPython 3.10.20",
4
+ "startedAt": "2026-03-16T07:35:59.330177Z",
5
+ "args": [
6
+ "--config_yaml",
7
+ "./examples/LIBERO/train_files/my_libero_finetune.yaml"
8
+ ],
9
+ "program": "/content/starVLA_r/starVLA/training/train_starvla.py",
10
+ "codePath": "starVLA/training/train_starvla.py",
11
+ "codePathLocal": "starVLA/training/train_starvla.py",
12
+ "git": {
13
+ "remote": "https://github.com/tliao730/starVLA_r",
14
+ "commit": "35e90d04f0c1fe5b6bdf8a3d6b0da703e1fed425"
15
+ },
16
+ "email": "chihhans@usc.edu",
17
+ "root": "./results/Checkpoints/pretrained_goal_2000step/wandb",
18
+ "host": "c89e62d63bf0",
19
+ "executable": "/usr/local/envs/starvla/bin/python3.10",
20
+ "cpu_count": 6,
21
+ "cpu_count_logical": 12,
22
+ "gpu": "NVIDIA A100-SXM4-80GB",
23
+ "gpu_count": 1,
24
+ "disk": {
25
+ "/": {
26
+ "total": "253055008768",
27
+ "used": "113942818816"
28
+ }
29
+ },
30
+ "memory": {
31
+ "total": "179370471424"
32
+ },
33
+ "gpu_nvidia": [
34
+ {
35
+ "name": "NVIDIA A100-SXM4-80GB",
36
+ "memoryTotal": "85899345920",
37
+ "cudaCores": 6912,
38
+ "architecture": "Ampere",
39
+ "uuid": "GPU-1000e8c7-f9d7-74b0-8fdb-aad3f6d24e69"
40
+ }
41
+ ],
42
+ "cudaVersion": "13.0",
43
+ "writerId": "tc8jq6me6tl6kehb0voihzue811ccsc4"
44
+ }
checkpoints/pretrained_goal_2000step/wandb/wandb/run-20260316_073559-h1hybozy/files/wandb-summary.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"model_time":0.6144314619996294,"_step":2000,"_runtime":1358.60672756,"mse_score":0.03371996539489637,"epoch":0.1,"_timestamp":1.7736478770671635e+09,"data_time":0.0016236379997280892,"action_dit_loss":2.123331308364868,"_wandb":{"runtime":1358},"learning_rate":4.000000000000001e-06}
checkpoints/pretrained_goal_2000step/wandb/wandb/run-20260316_073559-h1hybozy/logs/debug-core.log ADDED
@@ -0,0 +1,19 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {"time":"2026-03-16T07:35:59.431212911Z","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmp6tv1eu_r/port-23812.txt","pid":23812,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false,"enable-dcgm-profiling":false}
2
+ {"time":"2026-03-16T07:35:59.431795991Z","level":"INFO","msg":"server: will exit if parent process dies","ppid":23812}
3
+ {"time":"2026-03-16T07:35:59.431783453Z","level":"INFO","msg":"server: accepting connections","addr":{"Name":"/tmp/wandb-23812-24067-3160206984/socket","Net":"unix"}}
4
+ {"time":"2026-03-16T07:35:59.618651619Z","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"1(@)"}
5
+ {"time":"2026-03-16T07:35:59.623762216Z","level":"INFO","msg":"handleInformInit: received","streamId":"h1hybozy","id":"1(@)"}
6
+ {"time":"2026-03-16T07:35:59.961723652Z","level":"INFO","msg":"handleInformInit: stream started","streamId":"h1hybozy","id":"1(@)"}
7
+ {"time":"2026-03-16T07:36:06.055372942Z","level":"INFO","msg":"connection: cancelling request","id":"1(@)","requestId":"x6jcsbclt2f9"}
8
+ {"time":"2026-03-16T07:58:39.476003566Z","level":"INFO","msg":"connection: cancelling request","id":"1(@)","requestId":"x6jcsbclt2f9"}
9
+ {"time":"2026-03-16T07:58:41.750714347Z","level":"INFO","msg":"connection: cancelling request","id":"1(@)","requestId":"x6jcsbclt2f9"}
10
+ {"time":"2026-03-16T07:58:41.751259869Z","level":"INFO","msg":"handleInformFinish: finish message received","streamId":"h1hybozy","id":"1(@)"}
11
+ {"time":"2026-03-16T07:58:47.63590724Z","level":"INFO","msg":"handleInformFinish: stream closed","streamId":"h1hybozy","id":"1(@)"}
12
+ {"time":"2026-03-16T07:58:47.635947577Z","level":"INFO","msg":"handleInformTeardown: server teardown initiated","id":"1(@)"}
13
+ {"time":"2026-03-16T07:58:47.6359636Z","level":"INFO","msg":"handleInformTeardown: server shutdown complete","id":"1(@)"}
14
+ {"time":"2026-03-16T07:58:47.635976796Z","level":"INFO","msg":"connection: closing","id":"1(@)"}
15
+ {"time":"2026-03-16T07:58:47.636007763Z","level":"INFO","msg":"server is shutting down"}
16
+ {"time":"2026-03-16T07:58:47.636018708Z","level":"INFO","msg":"connection: closed successfully","id":"1(@)"}
17
+ {"time":"2026-03-16T07:58:47.636107719Z","level":"INFO","msg":"server: listener closed","addr":{"Name":"/tmp/wandb-23812-24067-3160206984/socket","Net":"unix"}}
18
+ {"time":"2026-03-16T07:58:47.636137886Z","level":"INFO","msg":"connection: ManageConnectionData: connection closed","id":"1(@)"}
19
+ {"time":"2026-03-16T07:58:47.636153185Z","level":"INFO","msg":"server is closed"}
checkpoints/pretrained_goal_2000step/wandb/wandb/run-20260316_073559-h1hybozy/logs/debug-internal.log ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {"time":"2026-03-16T07:35:59.623945177Z","level":"INFO","msg":"stream: starting","core version":"0.24.2"}
2
+ {"time":"2026-03-16T07:35:59.961527215Z","level":"INFO","msg":"stream: created new stream","id":"h1hybozy"}
3
+ {"time":"2026-03-16T07:35:59.961605327Z","level":"INFO","msg":"handler: started","stream_id":"h1hybozy"}
4
+ {"time":"2026-03-16T07:35:59.961715934Z","level":"INFO","msg":"stream: started","id":"h1hybozy"}
5
+ {"time":"2026-03-16T07:35:59.961730977Z","level":"INFO","msg":"writer: started","stream_id":"h1hybozy"}
6
+ {"time":"2026-03-16T07:35:59.961762841Z","level":"INFO","msg":"sender: started","stream_id":"h1hybozy"}
7
+ {"time":"2026-03-16T07:58:41.254677121Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
8
+ {"time":"2026-03-16T07:58:41.747155068Z","level":"INFO","msg":"handler: operation stats","stats":{}}
9
+ {"time":"2026-03-16T07:58:41.751293056Z","level":"INFO","msg":"stream: closing","id":"h1hybozy"}
10
+ {"time":"2026-03-16T07:58:41.751310876Z","level":"INFO","msg":"handler: closed","stream_id":"h1hybozy"}
11
+ {"time":"2026-03-16T07:58:41.751385991Z","level":"INFO","msg":"sender: closed","stream_id":"h1hybozy"}
12
+ {"time":"2026-03-16T07:58:41.751392739Z","level":"INFO","msg":"stream: closed","id":"h1hybozy"}
checkpoints/pretrained_goal_2000step/wandb/wandb/run-20260316_073559-h1hybozy/logs/debug.log ADDED
File without changes
checkpoints/pretrained_goal_2000step/wandb/wandb/run-20260316_073559-h1hybozy/run-h1hybozy.wandb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4e688bf854d1d4406166f483de0cb84bb6d989c559088dcc8de9c3163d17c192
3
+ size 713034