Hugging Face's logo

Datasets:
MLL-Lab
/
LAGEN-datasets

Tasks:
Robotics
Modalities:
Image
Tabular
Text
Size:
10M - 100M
Tags:
LeRobot
image
video
License:
Dataset card Data Studio Files
xet
 Community
1
Caesarrr commited on
Commit
2817185
·
verified ·
1 Parent(s): 7d55c19

Consolidate paper experiment assets by task and run (1201-1350)

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. evaluations/flappy/wan_oft_flappy_fix_latency_0_context5_2000_effbs128/eval/post_train/step_2000.json +123 -0
  2. evaluations/flappy/wan_oft_flappy_fix_latency_1_context5_2000_effbs128/eval/mid_train/step_1000.json +63 -0
  3. evaluations/flappy/wan_oft_flappy_fix_latency_1_context5_2000_effbs128/eval/mid_train/step_1500.json +63 -0
  4. evaluations/flappy/wan_oft_flappy_fix_latency_1_context5_2000_effbs128/eval/mid_train/step_2000.json +63 -0
  5. evaluations/flappy/wan_oft_flappy_fix_latency_1_context5_2000_effbs128/eval/mid_train/step_500.json +63 -0
  6. evaluations/flappy/wan_oft_flappy_fix_latency_1_context5_2000_effbs128/eval/post_train/step_2000.json +531 -0
  7. evaluations/flappy/wan_oft_flappy_fix_latency_2_context5_2000_effbs128/eval/post_train/step_2000.json +531 -0
  8. evaluations/flappy/wan_oft_flappy_fix_latency_3_context5_2000_effbs128/eval/mid_train/step_1000.json +63 -0
  9. evaluations/flappy/wan_oft_flappy_fix_latency_3_context5_2000_effbs128/eval/mid_train/step_1500.json +63 -0
  10. evaluations/flappy/wan_oft_flappy_fix_latency_3_context5_2000_effbs128/eval/mid_train/step_2000.json +63 -0
  11. evaluations/flappy/wan_oft_flappy_fix_latency_3_context5_2000_effbs128/eval/mid_train/step_500.json +63 -0
  12. evaluations/flappy/wan_oft_flappy_fix_latency_3_context5_2000_effbs128/eval/post_train/step_2000.json +531 -0
  13. evaluations/flappy/wan_oft_flappy_fix_latency_4_context5_2000_effbs128/eval/mid_train/step_1000.json +63 -0
  14. evaluations/flappy/wan_oft_flappy_fix_latency_4_context5_2000_effbs128/eval/mid_train/step_1500.json +63 -0
  15. evaluations/flappy/wan_oft_flappy_fix_latency_4_context5_2000_effbs128/eval/mid_train/step_2000.json +63 -0
  16. evaluations/flappy/wan_oft_flappy_fix_latency_4_context5_2000_effbs128/eval/mid_train/step_500.json +63 -0
  17. evaluations/flappy/wan_oft_flappy_fix_latency_4_context5_2000_effbs128/eval/post_train/step_2000.json +531 -0
  18. experiments/latency-transfer/provenance/latency-transfer-data/provenance/copy_verification.json +5 -0
  19. experiments/latency-transfer/provenance/latency-transfer-data/provenance/data_readability_verification.json +0 -0
  20. experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/flappy_fix_latency_0_200ep_context5/filter_report.json +12 -0
  21. experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/flappy_fix_latency_0_200ep_context5/metadata.json +305 -0
  22. experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/flappy_fix_latency_1_200ep_context5/filter_report.json +12 -0
  23. experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/flappy_fix_latency_1_200ep_context5/metadata.json +305 -0
  24. experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce/config.full.yaml +261 -0
  25. experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce/config.yaml +113 -0
  26. experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce/dataset_statistics.json +127 -0
  27. experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce/dataset_statistics_eval.json +127 -0
  28. experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce/eval/mid_train/step_1000.json +63 -0
  29. experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce/eval/mid_train/step_1500.json +63 -0
  30. experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce/eval/mid_train/step_2000.json +63 -0
  31. experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce/eval/mid_train/step_500.json +63 -0
  32. experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce/eval/post_train/step_2000.json +531 -0
  33. experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce/hydra/.hydra/config.yaml +258 -0
  34. experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce/hydra/.hydra/hydra.yaml +356 -0
  35. experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce/hydra/.hydra/overrides.yaml +196 -0
  36. experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce/hydra/train_starvla_hydra.log +0 -0
  37. experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce/summary.jsonl +1 -0
  38. experiments/latency-transfer/provenance/latency-transfer-data/provenance/fixed_conversion_probe.log +9 -0
  39. experiments/latency-transfer/provenance/latency-transfer-data/provenance/native_loader_verification.json +0 -0
  40. experiments/latency-transfer/provenance/latency-transfer-data/provenance/old_code_recovery.json +10 -0
  41. experiments/latency-transfer/provenance/latency-transfer-data/provenance/original_reference_documents/latency-transfer-data/run_index.json +144 -0
  42. experiments/latency-transfer/provenance/latency-transfer-data/provenance/original_reference_documents/latency-transfer-models/run_index.json +144 -0
  43. experiments/latency-transfer/provenance/latency-transfer-data/provenance/pinned_setup_tests.log +17 -0
  44. experiments/latency-transfer/provenance/latency-transfer-data/provenance/probe_wandb_artifact_contents.json +62 -0
  45. experiments/latency-transfer/provenance/latency-transfer-data/provenance/probe_wandb_artifacts.json +174 -0
  46. experiments/latency-transfer/provenance/latency-transfer-data/provenance/probe_wandb_files.json +97 -0
  47. experiments/latency-transfer/provenance/latency-transfer-data/provenance/source_inventory.json +642 -0
  48. experiments/latency-transfer/provenance/latency-transfer-data/provenance/source_repositories/dataset/deadly_corridor_fix_latency_0.json +150 -0
  49. experiments/latency-transfer/provenance/latency-transfer-data/provenance/source_repositories/dataset/deadly_corridor_fix_latency_1.json +150 -0
  50. experiments/latency-transfer/provenance/latency-transfer-data/provenance/source_repositories/dataset/deadly_corridor_fix_latency_10.json +150 -0
evaluations/flappy/wan_oft_flappy_fix_latency_0_context5_2000_effbs128/eval/post_train/step_2000.json ADDED
@@ -0,0 +1,123 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_latency": {
3
+ "flappy/latency_0": {
4
+ "latency": 0,
5
+ "num_episodes": 20,
6
+ "mean_reward": 423.83000000002437,
7
+ "mean_length": 3432.45,
8
+ "std_reward": 90.53433105734537,
9
+ "std_length": 730.3335179902397,
10
+ "episode_rewards": [
11
+ 444.60000000002566,
12
+ 444.60000000002566,
13
+ 444.60000000002566,
14
+ 29.200000000000127,
15
+ 444.60000000002566,
16
+ 444.60000000002566,
17
+ 444.60000000002566,
18
+ 444.60000000002566,
19
+ 444.60000000002566,
20
+ 444.60000000002566,
21
+ 444.60000000002566,
22
+ 444.60000000002566,
23
+ 444.60000000002566,
24
+ 444.60000000002566,
25
+ 444.60000000002566,
26
+ 444.60000000002566,
27
+ 444.60000000002566,
28
+ 444.60000000002566,
29
+ 444.60000000002566,
30
+ 444.60000000002566
31
+ ],
32
+ "episode_lengths": [
33
+ 3600,
34
+ 3600,
35
+ 3600,
36
+ 249,
37
+ 3600,
38
+ 3600,
39
+ 3600,
40
+ 3600,
41
+ 3600,
42
+ 3600,
43
+ 3600,
44
+ 3600,
45
+ 3600,
46
+ 3600,
47
+ 3600,
48
+ 3600,
49
+ 3600,
50
+ 3600,
51
+ 3600,
52
+ 3600
53
+ ],
54
+ "decoded_action_hist": {
55
+ "0": 63231,
56
+ "1": 5418
57
+ },
58
+ "fixed_episode_seeds": true,
59
+ "eval_seed": 42,
60
+ "episode_seeds": [
61
+ 42,
62
+ 43,
63
+ 44,
64
+ 45,
65
+ 46,
66
+ 47,
67
+ 48,
68
+ 49,
69
+ 50,
70
+ 51,
71
+ 52,
72
+ 53,
73
+ 54,
74
+ 55,
75
+ 56,
76
+ 57,
77
+ 58,
78
+ 59,
79
+ 60,
80
+ 61
81
+ ],
82
+ "episode_indices": [
83
+ 0,
84
+ 1,
85
+ 2,
86
+ 3,
87
+ 4,
88
+ 5,
89
+ 6,
90
+ 7,
91
+ 8,
92
+ 9,
93
+ 10,
94
+ 11,
95
+ 12,
96
+ 13,
97
+ 14,
98
+ 15,
99
+ 16,
100
+ 17,
101
+ 18,
102
+ 19
103
+ ]
104
+ }
105
+ },
106
+ "aggregate": {
107
+ "stage": "post_train",
108
+ "step": 2000,
109
+ "task": "flappy",
110
+ "model_alias": "wan_oft",
111
+ "fixed_episode_seeds": true,
112
+ "eval_seed": 42,
113
+ "total_episodes": 20,
114
+ "mean_reward": 423.83000000002437,
115
+ "mean_length": 3432.45,
116
+ "std_reward": 90.53433105734537,
117
+ "std_length": 730.3335179902397,
118
+ "task_count": 1,
119
+ "macro_mean_reward": 423.83000000002437,
120
+ "macro_mean_length": 3432.45,
121
+ "distributed_eval": false
122
+ }
123
+ }
evaluations/flappy/wan_oft_flappy_fix_latency_1_context5_2000_effbs128/eval/mid_train/step_1000.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_latency": {
3
+ "flappy/latency_1": {
4
+ "latency": 1,
5
+ "num_episodes": 5,
6
+ "mean_reward": 444.60000000002566,
7
+ "mean_length": 3600.0,
8
+ "std_reward": 0.0,
9
+ "std_length": 0.0,
10
+ "episode_rewards": [
11
+ 444.60000000002566,
12
+ 444.60000000002566,
13
+ 444.60000000002566,
14
+ 444.60000000002566,
15
+ 444.60000000002566
16
+ ],
17
+ "episode_lengths": [
18
+ 3600,
19
+ 3600,
20
+ 3600,
21
+ 3600,
22
+ 3600
23
+ ],
24
+ "decoded_action_hist": {
25
+ "0": 15569,
26
+ "1": 2431
27
+ },
28
+ "fixed_episode_seeds": true,
29
+ "eval_seed": 42,
30
+ "episode_seeds": [
31
+ 42,
32
+ 43,
33
+ 44,
34
+ 45,
35
+ 46
36
+ ],
37
+ "episode_indices": [
38
+ 0,
39
+ 1,
40
+ 2,
41
+ 3,
42
+ 4
43
+ ]
44
+ }
45
+ },
46
+ "aggregate": {
47
+ "stage": "mid_train",
48
+ "step": 1000,
49
+ "task": "flappy",
50
+ "model_alias": "wan_oft",
51
+ "fixed_episode_seeds": true,
52
+ "eval_seed": 42,
53
+ "total_episodes": 5,
54
+ "mean_reward": 444.60000000002566,
55
+ "mean_length": 3600.0,
56
+ "std_reward": 0.0,
57
+ "std_length": 0.0,
58
+ "task_count": 1,
59
+ "macro_mean_reward": 444.60000000002566,
60
+ "macro_mean_length": 3600.0,
61
+ "distributed_eval": false
62
+ }
63
+ }
evaluations/flappy/wan_oft_flappy_fix_latency_1_context5_2000_effbs128/eval/mid_train/step_1500.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_latency": {
3
+ "flappy/latency_1": {
4
+ "latency": 1,
5
+ "num_episodes": 5,
6
+ "mean_reward": 444.60000000002566,
7
+ "mean_length": 3600.0,
8
+ "std_reward": 0.0,
9
+ "std_length": 0.0,
10
+ "episode_rewards": [
11
+ 444.60000000002566,
12
+ 444.60000000002566,
13
+ 444.60000000002566,
14
+ 444.60000000002566,
15
+ 444.60000000002566
16
+ ],
17
+ "episode_lengths": [
18
+ 3600,
19
+ 3600,
20
+ 3600,
21
+ 3600,
22
+ 3600
23
+ ],
24
+ "decoded_action_hist": {
25
+ "0": 15535,
26
+ "1": 2465
27
+ },
28
+ "fixed_episode_seeds": true,
29
+ "eval_seed": 42,
30
+ "episode_seeds": [
31
+ 42,
32
+ 43,
33
+ 44,
34
+ 45,
35
+ 46
36
+ ],
37
+ "episode_indices": [
38
+ 0,
39
+ 1,
40
+ 2,
41
+ 3,
42
+ 4
43
+ ]
44
+ }
45
+ },
46
+ "aggregate": {
47
+ "stage": "mid_train",
48
+ "step": 1500,
49
+ "task": "flappy",
50
+ "model_alias": "wan_oft",
51
+ "fixed_episode_seeds": true,
52
+ "eval_seed": 42,
53
+ "total_episodes": 5,
54
+ "mean_reward": 444.60000000002566,
55
+ "mean_length": 3600.0,
56
+ "std_reward": 0.0,
57
+ "std_length": 0.0,
58
+ "task_count": 1,
59
+ "macro_mean_reward": 444.60000000002566,
60
+ "macro_mean_length": 3600.0,
61
+ "distributed_eval": false
62
+ }
63
+ }
evaluations/flappy/wan_oft_flappy_fix_latency_1_context5_2000_effbs128/eval/mid_train/step_2000.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_latency": {
3
+ "flappy/latency_1": {
4
+ "latency": 1,
5
+ "num_episodes": 5,
6
+ "mean_reward": 444.60000000002566,
7
+ "mean_length": 3600.0,
8
+ "std_reward": 0.0,
9
+ "std_length": 0.0,
10
+ "episode_rewards": [
11
+ 444.60000000002566,
12
+ 444.60000000002566,
13
+ 444.60000000002566,
14
+ 444.60000000002566,
15
+ 444.60000000002566
16
+ ],
17
+ "episode_lengths": [
18
+ 3600,
19
+ 3600,
20
+ 3600,
21
+ 3600,
22
+ 3600
23
+ ],
24
+ "decoded_action_hist": {
25
+ "0": 15564,
26
+ "1": 2436
27
+ },
28
+ "fixed_episode_seeds": true,
29
+ "eval_seed": 42,
30
+ "episode_seeds": [
31
+ 42,
32
+ 43,
33
+ 44,
34
+ 45,
35
+ 46
36
+ ],
37
+ "episode_indices": [
38
+ 0,
39
+ 1,
40
+ 2,
41
+ 3,
42
+ 4
43
+ ]
44
+ }
45
+ },
46
+ "aggregate": {
47
+ "stage": "mid_train",
48
+ "step": 2000,
49
+ "task": "flappy",
50
+ "model_alias": "wan_oft",
51
+ "fixed_episode_seeds": true,
52
+ "eval_seed": 42,
53
+ "total_episodes": 5,
54
+ "mean_reward": 444.60000000002566,
55
+ "mean_length": 3600.0,
56
+ "std_reward": 0.0,
57
+ "std_length": 0.0,
58
+ "task_count": 1,
59
+ "macro_mean_reward": 444.60000000002566,
60
+ "macro_mean_length": 3600.0,
61
+ "distributed_eval": false
62
+ }
63
+ }
evaluations/flappy/wan_oft_flappy_fix_latency_1_context5_2000_effbs128/eval/mid_train/step_500.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_latency": {
3
+ "flappy/latency_1": {
4
+ "latency": 1,
5
+ "num_episodes": 5,
6
+ "mean_reward": 382.18000000001996,
7
+ "mean_length": 3096.8,
8
+ "std_reward": 124.8400000000113,
9
+ "std_length": 1006.4000000000002,
10
+ "episode_rewards": [
11
+ 132.49999999999739,
12
+ 444.60000000002566,
13
+ 444.60000000002566,
14
+ 444.60000000002566,
15
+ 444.60000000002566
16
+ ],
17
+ "episode_lengths": [
18
+ 1084,
19
+ 3600,
20
+ 3600,
21
+ 3600,
22
+ 3600
23
+ ],
24
+ "decoded_action_hist": {
25
+ "0": 13331,
26
+ "1": 2153
27
+ },
28
+ "fixed_episode_seeds": true,
29
+ "eval_seed": 42,
30
+ "episode_seeds": [
31
+ 42,
32
+ 43,
33
+ 44,
34
+ 45,
35
+ 46
36
+ ],
37
+ "episode_indices": [
38
+ 0,
39
+ 1,
40
+ 2,
41
+ 3,
42
+ 4
43
+ ]
44
+ }
45
+ },
46
+ "aggregate": {
47
+ "stage": "mid_train",
48
+ "step": 500,
49
+ "task": "flappy",
50
+ "model_alias": "wan_oft",
51
+ "fixed_episode_seeds": true,
52
+ "eval_seed": 42,
53
+ "total_episodes": 5,
54
+ "mean_reward": 382.18000000001996,
55
+ "mean_length": 3096.8,
56
+ "std_reward": 124.8400000000113,
57
+ "std_length": 1006.4000000000002,
58
+ "task_count": 1,
59
+ "macro_mean_reward": 382.18000000001996,
60
+ "macro_mean_length": 3096.8,
61
+ "distributed_eval": false
62
+ }
63
+ }
evaluations/flappy/wan_oft_flappy_fix_latency_1_context5_2000_effbs128/eval/post_train/step_2000.json ADDED
@@ -0,0 +1,531 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_latency": {
3
+ "flappy/latency_0": {
4
+ "latency": 0,
5
+ "num_episodes": 20,
6
+ "mean_reward": 72.23999999999896,
7
+ "mean_length": 597.95,
8
+ "std_reward": 63.11439930792145,
9
+ "std_length": 509.34550896223675,
10
+ "episode_rewards": [
11
+ 57.800000000000445,
12
+ 34.80000000000019,
13
+ 20.800000000000036,
14
+ 174.7999999999955,
15
+ 81.29999999999968,
16
+ 169.79999999999572,
17
+ 132.79999999999737,
18
+ 11.29999999999998,
19
+ 34.80000000000019,
20
+ 6.799999999999992,
21
+ 53.300000000000395,
22
+ 34.80000000000019,
23
+ 76.79999999999988,
24
+ 71.80000000000011,
25
+ 20.700000000000035,
26
+ 20.800000000000036,
27
+ 183.7999999999951,
28
+ 207.29999999999404,
29
+ 20.700000000000035,
30
+ 29.800000000000136
31
+ ],
32
+ "episode_lengths": [
33
+ 481,
34
+ 296,
35
+ 183,
36
+ 1426,
37
+ 671,
38
+ 1385,
39
+ 1087,
40
+ 106,
41
+ 296,
42
+ 70,
43
+ 445,
44
+ 296,
45
+ 635,
46
+ 594,
47
+ 182,
48
+ 183,
49
+ 1498,
50
+ 1688,
51
+ 182,
52
+ 255
53
+ ],
54
+ "decoded_action_hist": {
55
+ "1": 903,
56
+ "0": 11056
57
+ },
58
+ "fixed_episode_seeds": true,
59
+ "eval_seed": 42,
60
+ "episode_seeds": [
61
+ 42,
62
+ 43,
63
+ 44,
64
+ 45,
65
+ 46,
66
+ 47,
67
+ 48,
68
+ 49,
69
+ 50,
70
+ 51,
71
+ 52,
72
+ 53,
73
+ 54,
74
+ 55,
75
+ 56,
76
+ 57,
77
+ 58,
78
+ 59,
79
+ 60,
80
+ 61
81
+ ],
82
+ "episode_indices": [
83
+ 0,
84
+ 1,
85
+ 2,
86
+ 3,
87
+ 4,
88
+ 5,
89
+ 6,
90
+ 7,
91
+ 8,
92
+ 9,
93
+ 10,
94
+ 11,
95
+ 12,
96
+ 13,
97
+ 14,
98
+ 15,
99
+ 16,
100
+ 17,
101
+ 18,
102
+ 19
103
+ ]
104
+ },
105
+ "flappy/latency_1": {
106
+ "latency": 1,
107
+ "num_episodes": 20,
108
+ "mean_reward": 444.60000000002566,
109
+ "mean_length": 3600.0,
110
+ "std_reward": 0.0,
111
+ "std_length": 0.0,
112
+ "episode_rewards": [
113
+ 444.60000000002566,
114
+ 444.60000000002566,
115
+ 444.60000000002566,
116
+ 444.60000000002566,
117
+ 444.60000000002566,
118
+ 444.60000000002566,
119
+ 444.60000000002566,
120
+ 444.60000000002566,
121
+ 444.60000000002566,
122
+ 444.60000000002566,
123
+ 444.60000000002566,
124
+ 444.60000000002566,
125
+ 444.60000000002566,
126
+ 444.60000000002566,
127
+ 444.60000000002566,
128
+ 444.60000000002566,
129
+ 444.60000000002566,
130
+ 444.60000000002566,
131
+ 444.60000000002566,
132
+ 444.60000000002566
133
+ ],
134
+ "episode_lengths": [
135
+ 3600,
136
+ 3600,
137
+ 3600,
138
+ 3600,
139
+ 3600,
140
+ 3600,
141
+ 3600,
142
+ 3600,
143
+ 3600,
144
+ 3600,
145
+ 3600,
146
+ 3600,
147
+ 3600,
148
+ 3600,
149
+ 3600,
150
+ 3600,
151
+ 3600,
152
+ 3600,
153
+ 3600,
154
+ 3600
155
+ ],
156
+ "decoded_action_hist": {
157
+ "0": 62110,
158
+ "1": 9890
159
+ },
160
+ "fixed_episode_seeds": true,
161
+ "eval_seed": 42,
162
+ "episode_seeds": [
163
+ 42,
164
+ 43,
165
+ 44,
166
+ 45,
167
+ 46,
168
+ 47,
169
+ 48,
170
+ 49,
171
+ 50,
172
+ 51,
173
+ 52,
174
+ 53,
175
+ 54,
176
+ 55,
177
+ 56,
178
+ 57,
179
+ 58,
180
+ 59,
181
+ 60,
182
+ 61
183
+ ],
184
+ "episode_indices": [
185
+ 0,
186
+ 1,
187
+ 2,
188
+ 3,
189
+ 4,
190
+ 5,
191
+ 6,
192
+ 7,
193
+ 8,
194
+ 9,
195
+ 10,
196
+ 11,
197
+ 12,
198
+ 13,
199
+ 14,
200
+ 15,
201
+ 16,
202
+ 17,
203
+ 18,
204
+ 19
205
+ ]
206
+ },
207
+ "flappy/latency_2": {
208
+ "latency": 2,
209
+ "num_episodes": 20,
210
+ "mean_reward": 6.615000000000006,
211
+ "mean_length": 71.75,
212
+ "std_reward": 7.366700414704039,
213
+ "std_length": 59.72342505248673,
214
+ "episode_rewards": [
215
+ 3.8999999999999986,
216
+ 4.099999999999998,
217
+ 4.1999999999999975,
218
+ 37.20000000000022,
219
+ 4.099999999999998,
220
+ 3.8999999999999986,
221
+ 5.999999999999995,
222
+ 3.8999999999999986,
223
+ 4.299999999999997,
224
+ 3.9999999999999982,
225
+ 13.299999999999972,
226
+ 4.299999999999997,
227
+ 3.8999999999999986,
228
+ 8.999999999999984,
229
+ 3.8999999999999986,
230
+ 3.8999999999999986,
231
+ 4.099999999999998,
232
+ 4.299999999999997,
233
+ 5.999999999999995,
234
+ 3.9999999999999982
235
+ ],
236
+ "episode_lengths": [
237
+ 50,
238
+ 52,
239
+ 53,
240
+ 320,
241
+ 52,
242
+ 50,
243
+ 62,
244
+ 50,
245
+ 54,
246
+ 51,
247
+ 126,
248
+ 54,
249
+ 50,
250
+ 92,
251
+ 50,
252
+ 50,
253
+ 52,
254
+ 54,
255
+ 62,
256
+ 51
257
+ ],
258
+ "decoded_action_hist": {
259
+ "0": 1186,
260
+ "1": 249
261
+ },
262
+ "fixed_episode_seeds": true,
263
+ "eval_seed": 42,
264
+ "episode_seeds": [
265
+ 42,
266
+ 43,
267
+ 44,
268
+ 45,
269
+ 46,
270
+ 47,
271
+ 48,
272
+ 49,
273
+ 50,
274
+ 51,
275
+ 52,
276
+ 53,
277
+ 54,
278
+ 55,
279
+ 56,
280
+ 57,
281
+ 58,
282
+ 59,
283
+ 60,
284
+ 61
285
+ ],
286
+ "episode_indices": [
287
+ 0,
288
+ 1,
289
+ 2,
290
+ 3,
291
+ 4,
292
+ 5,
293
+ 6,
294
+ 7,
295
+ 8,
296
+ 9,
297
+ 10,
298
+ 11,
299
+ 12,
300
+ 13,
301
+ 14,
302
+ 15,
303
+ 16,
304
+ 17,
305
+ 18,
306
+ 19
307
+ ]
308
+ },
309
+ "flappy/latency_3": {
310
+ "latency": 3,
311
+ "num_episodes": 20,
312
+ "mean_reward": 4.689999999999996,
313
+ "mean_length": 56.55,
314
+ "std_reward": 0.6579513659838381,
315
+ "std_length": 3.9430318284284747,
316
+ "episode_rewards": [
317
+ 3.9999999999999982,
318
+ 4.699999999999996,
319
+ 4.799999999999995,
320
+ 4.799999999999995,
321
+ 4.4999999999999964,
322
+ 3.9999999999999982,
323
+ 6.099999999999994,
324
+ 3.9999999999999982,
325
+ 4.099999999999998,
326
+ 4.399999999999997,
327
+ 4.799999999999995,
328
+ 4.699999999999996,
329
+ 4.299999999999997,
330
+ 4.799999999999995,
331
+ 6.099999999999994,
332
+ 3.9999999999999982,
333
+ 4.4999999999999964,
334
+ 4.699999999999996,
335
+ 6.099999999999994,
336
+ 4.399999999999997
337
+ ],
338
+ "episode_lengths": [
339
+ 51,
340
+ 58,
341
+ 59,
342
+ 59,
343
+ 56,
344
+ 51,
345
+ 63,
346
+ 51,
347
+ 52,
348
+ 55,
349
+ 59,
350
+ 58,
351
+ 54,
352
+ 59,
353
+ 63,
354
+ 51,
355
+ 56,
356
+ 58,
357
+ 63,
358
+ 55
359
+ ],
360
+ "decoded_action_hist": {
361
+ "0": 934,
362
+ "1": 197
363
+ },
364
+ "fixed_episode_seeds": true,
365
+ "eval_seed": 42,
366
+ "episode_seeds": [
367
+ 42,
368
+ 43,
369
+ 44,
370
+ 45,
371
+ 46,
372
+ 47,
373
+ 48,
374
+ 49,
375
+ 50,
376
+ 51,
377
+ 52,
378
+ 53,
379
+ 54,
380
+ 55,
381
+ 56,
382
+ 57,
383
+ 58,
384
+ 59,
385
+ 60,
386
+ 61
387
+ ],
388
+ "episode_indices": [
389
+ 0,
390
+ 1,
391
+ 2,
392
+ 3,
393
+ 4,
394
+ 5,
395
+ 6,
396
+ 7,
397
+ 8,
398
+ 9,
399
+ 10,
400
+ 11,
401
+ 12,
402
+ 13,
403
+ 14,
404
+ 15,
405
+ 16,
406
+ 17,
407
+ 18,
408
+ 19
409
+ ]
410
+ },
411
+ "flappy/latency_4": {
412
+ "latency": 4,
413
+ "num_episodes": 20,
414
+ "mean_reward": 5.044999999999996,
415
+ "mean_length": 60.55,
416
+ "std_reward": 1.1386285610329618,
417
+ "std_length": 8.749142815156237,
418
+ "episode_rewards": [
419
+ 4.799999999999995,
420
+ 4.799999999999995,
421
+ 4.399999999999997,
422
+ 4.599999999999996,
423
+ 4.799999999999995,
424
+ 4.999999999999995,
425
+ 8.399999999999986,
426
+ 4.999999999999995,
427
+ 4.299999999999997,
428
+ 4.699999999999996,
429
+ 4.599999999999996,
430
+ 4.299999999999997,
431
+ 4.599999999999996,
432
+ 4.599999999999996,
433
+ 4.799999999999995,
434
+ 4.999999999999995,
435
+ 4.799999999999995,
436
+ 4.299999999999997,
437
+ 8.399999999999986,
438
+ 4.699999999999996
439
+ ],
440
+ "episode_lengths": [
441
+ 59,
442
+ 59,
443
+ 55,
444
+ 57,
445
+ 59,
446
+ 61,
447
+ 86,
448
+ 61,
449
+ 54,
450
+ 58,
451
+ 57,
452
+ 54,
453
+ 57,
454
+ 57,
455
+ 59,
456
+ 61,
457
+ 59,
458
+ 54,
459
+ 86,
460
+ 58
461
+ ],
462
+ "decoded_action_hist": {
463
+ "0": 960,
464
+ "1": 251
465
+ },
466
+ "fixed_episode_seeds": true,
467
+ "eval_seed": 42,
468
+ "episode_seeds": [
469
+ 42,
470
+ 43,
471
+ 44,
472
+ 45,
473
+ 46,
474
+ 47,
475
+ 48,
476
+ 49,
477
+ 50,
478
+ 51,
479
+ 52,
480
+ 53,
481
+ 54,
482
+ 55,
483
+ 56,
484
+ 57,
485
+ 58,
486
+ 59,
487
+ 60,
488
+ 61
489
+ ],
490
+ "episode_indices": [
491
+ 0,
492
+ 1,
493
+ 2,
494
+ 3,
495
+ 4,
496
+ 5,
497
+ 6,
498
+ 7,
499
+ 8,
500
+ 9,
501
+ 10,
502
+ 11,
503
+ 12,
504
+ 13,
505
+ 14,
506
+ 15,
507
+ 16,
508
+ 17,
509
+ 18,
510
+ 19
511
+ ]
512
+ }
513
+ },
514
+ "aggregate": {
515
+ "stage": "post_train",
516
+ "step": 2000,
517
+ "task": "flappy",
518
+ "model_alias": "wan_oft",
519
+ "fixed_episode_seeds": true,
520
+ "eval_seed": 42,
521
+ "total_episodes": 100,
522
+ "mean_reward": 106.63800000000492,
523
+ "mean_length": 877.36,
524
+ "std_reward": 173.2974862945318,
525
+ "std_length": 1395.9831984662278,
526
+ "task_count": 1,
527
+ "macro_mean_reward": 106.6380000000049,
528
+ "macro_mean_length": 877.36,
529
+ "distributed_eval": false
530
+ }
531
+ }
evaluations/flappy/wan_oft_flappy_fix_latency_2_context5_2000_effbs128/eval/post_train/step_2000.json ADDED
@@ -0,0 +1,531 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_latency": {
3
+ "flappy/latency_0": {
4
+ "latency": 0,
5
+ "num_episodes": 20,
6
+ "mean_reward": 6.524999999999993,
7
+ "mean_length": 67.25,
8
+ "std_reward": 0.20217566619155658,
9
+ "std_length": 2.021756661915573,
10
+ "episode_rewards": [
11
+ 5.999999999999995,
12
+ 6.499999999999993,
13
+ 6.399999999999993,
14
+ 6.5999999999999925,
15
+ 6.499999999999993,
16
+ 5.999999999999995,
17
+ 6.699999999999992,
18
+ 6.499999999999993,
19
+ 6.699999999999992,
20
+ 6.399999999999993,
21
+ 6.5999999999999925,
22
+ 6.699999999999992,
23
+ 6.699999999999992,
24
+ 6.5999999999999925,
25
+ 6.699999999999992,
26
+ 6.499999999999993,
27
+ 6.499999999999993,
28
+ 6.699999999999992,
29
+ 6.699999999999992,
30
+ 6.499999999999993
31
+ ],
32
+ "episode_lengths": [
33
+ 62,
34
+ 67,
35
+ 66,
36
+ 68,
37
+ 67,
38
+ 62,
39
+ 69,
40
+ 67,
41
+ 69,
42
+ 66,
43
+ 68,
44
+ 69,
45
+ 69,
46
+ 68,
47
+ 69,
48
+ 67,
49
+ 67,
50
+ 69,
51
+ 69,
52
+ 67
53
+ ],
54
+ "decoded_action_hist": {
55
+ "0": 1265,
56
+ "1": 80
57
+ },
58
+ "fixed_episode_seeds": true,
59
+ "eval_seed": 42,
60
+ "episode_seeds": [
61
+ 42,
62
+ 43,
63
+ 44,
64
+ 45,
65
+ 46,
66
+ 47,
67
+ 48,
68
+ 49,
69
+ 50,
70
+ 51,
71
+ 52,
72
+ 53,
73
+ 54,
74
+ 55,
75
+ 56,
76
+ 57,
77
+ 58,
78
+ 59,
79
+ 60,
80
+ 61
81
+ ],
82
+ "episode_indices": [
83
+ 0,
84
+ 1,
85
+ 2,
86
+ 3,
87
+ 4,
88
+ 5,
89
+ 6,
90
+ 7,
91
+ 8,
92
+ 9,
93
+ 10,
94
+ 11,
95
+ 12,
96
+ 13,
97
+ 14,
98
+ 15,
99
+ 16,
100
+ 17,
101
+ 18,
102
+ 19
103
+ ]
104
+ },
105
+ "flappy/latency_1": {
106
+ "latency": 1,
107
+ "num_episodes": 20,
108
+ "mean_reward": 7.48499999999999,
109
+ "mean_length": 75.95,
110
+ "std_reward": 1.9958143701256335,
111
+ "std_length": 15.367091461952063,
112
+ "episode_rewards": [
113
+ 6.799999999999992,
114
+ 11.29999999999998,
115
+ 11.29999999999998,
116
+ 4.799999999999995,
117
+ 11.29999999999998,
118
+ 6.799999999999992,
119
+ 6.799999999999992,
120
+ 6.5999999999999925,
121
+ 6.799999999999992,
122
+ 6.799999999999992,
123
+ 4.799999999999995,
124
+ 6.799999999999992,
125
+ 6.799999999999992,
126
+ 6.799999999999992,
127
+ 6.799999999999992,
128
+ 6.799999999999992,
129
+ 11.29999999999998,
130
+ 6.799999999999992,
131
+ 6.799999999999992,
132
+ 6.699999999999992
133
+ ],
134
+ "episode_lengths": [
135
+ 70,
136
+ 106,
137
+ 106,
138
+ 59,
139
+ 106,
140
+ 70,
141
+ 70,
142
+ 68,
143
+ 70,
144
+ 70,
145
+ 59,
146
+ 70,
147
+ 70,
148
+ 70,
149
+ 70,
150
+ 70,
151
+ 106,
152
+ 70,
153
+ 70,
154
+ 69
155
+ ],
156
+ "decoded_action_hist": {
157
+ "0": 1358,
158
+ "1": 161
159
+ },
160
+ "fixed_episode_seeds": true,
161
+ "eval_seed": 42,
162
+ "episode_seeds": [
163
+ 42,
164
+ 43,
165
+ 44,
166
+ 45,
167
+ 46,
168
+ 47,
169
+ 48,
170
+ 49,
171
+ 50,
172
+ 51,
173
+ 52,
174
+ 53,
175
+ 54,
176
+ 55,
177
+ 56,
178
+ 57,
179
+ 58,
180
+ 59,
181
+ 60,
182
+ 61
183
+ ],
184
+ "episode_indices": [
185
+ 0,
186
+ 1,
187
+ 2,
188
+ 3,
189
+ 4,
190
+ 5,
191
+ 6,
192
+ 7,
193
+ 8,
194
+ 9,
195
+ 10,
196
+ 11,
197
+ 12,
198
+ 13,
199
+ 14,
200
+ 15,
201
+ 16,
202
+ 17,
203
+ 18,
204
+ 19
205
+ ]
206
+ },
207
+ "flappy/latency_2": {
208
+ "latency": 2,
209
+ "num_episodes": 20,
210
+ "mean_reward": 444.60000000002566,
211
+ "mean_length": 3600.0,
212
+ "std_reward": 0.0,
213
+ "std_length": 0.0,
214
+ "episode_rewards": [
215
+ 444.60000000002566,
216
+ 444.60000000002566,
217
+ 444.60000000002566,
218
+ 444.60000000002566,
219
+ 444.60000000002566,
220
+ 444.60000000002566,
221
+ 444.60000000002566,
222
+ 444.60000000002566,
223
+ 444.60000000002566,
224
+ 444.60000000002566,
225
+ 444.60000000002566,
226
+ 444.60000000002566,
227
+ 444.60000000002566,
228
+ 444.60000000002566,
229
+ 444.60000000002566,
230
+ 444.60000000002566,
231
+ 444.60000000002566,
232
+ 444.60000000002566,
233
+ 444.60000000002566,
234
+ 444.60000000002566
235
+ ],
236
+ "episode_lengths": [
237
+ 3600,
238
+ 3600,
239
+ 3600,
240
+ 3600,
241
+ 3600,
242
+ 3600,
243
+ 3600,
244
+ 3600,
245
+ 3600,
246
+ 3600,
247
+ 3600,
248
+ 3600,
249
+ 3600,
250
+ 3600,
251
+ 3600,
252
+ 3600,
253
+ 3600,
254
+ 3600,
255
+ 3600,
256
+ 3600
257
+ ],
258
+ "decoded_action_hist": {
259
+ "0": 57467,
260
+ "1": 14533
261
+ },
262
+ "fixed_episode_seeds": true,
263
+ "eval_seed": 42,
264
+ "episode_seeds": [
265
+ 42,
266
+ 43,
267
+ 44,
268
+ 45,
269
+ 46,
270
+ 47,
271
+ 48,
272
+ 49,
273
+ 50,
274
+ 51,
275
+ 52,
276
+ 53,
277
+ 54,
278
+ 55,
279
+ 56,
280
+ 57,
281
+ 58,
282
+ 59,
283
+ 60,
284
+ 61
285
+ ],
286
+ "episode_indices": [
287
+ 0,
288
+ 1,
289
+ 2,
290
+ 3,
291
+ 4,
292
+ 5,
293
+ 6,
294
+ 7,
295
+ 8,
296
+ 9,
297
+ 10,
298
+ 11,
299
+ 12,
300
+ 13,
301
+ 14,
302
+ 15,
303
+ 16,
304
+ 17,
305
+ 18,
306
+ 19
307
+ ]
308
+ },
309
+ "flappy/latency_3": {
310
+ "latency": 3,
311
+ "num_episodes": 20,
312
+ "mean_reward": 4.5749999999999975,
313
+ "mean_length": 55.4,
314
+ "std_reward": 2.1463632031881215,
315
+ "std_length": 17.170905625505018,
316
+ "episode_rewards": [
317
+ 3.8999999999999986,
318
+ 3.8999999999999986,
319
+ 3.8999999999999986,
320
+ 3.8999999999999986,
321
+ 3.8999999999999986,
322
+ 3.8999999999999986,
323
+ 3.8999999999999986,
324
+ 3.8999999999999986,
325
+ 3.8999999999999986,
326
+ 3.8999999999999986,
327
+ 3.8999999999999986,
328
+ 3.8999999999999986,
329
+ 12.899999999999974,
330
+ 3.8999999999999986,
331
+ 8.399999999999986,
332
+ 3.8999999999999986,
333
+ 3.8999999999999986,
334
+ 3.8999999999999986,
335
+ 3.8999999999999986,
336
+ 3.8999999999999986
337
+ ],
338
+ "episode_lengths": [
339
+ 50,
340
+ 50,
341
+ 50,
342
+ 50,
343
+ 50,
344
+ 50,
345
+ 50,
346
+ 50,
347
+ 50,
348
+ 50,
349
+ 50,
350
+ 50,
351
+ 122,
352
+ 50,
353
+ 86,
354
+ 50,
355
+ 50,
356
+ 50,
357
+ 50,
358
+ 50
359
+ ],
360
+ "decoded_action_hist": {
361
+ "0": 894,
362
+ "1": 214
363
+ },
364
+ "fixed_episode_seeds": true,
365
+ "eval_seed": 42,
366
+ "episode_seeds": [
367
+ 42,
368
+ 43,
369
+ 44,
370
+ 45,
371
+ 46,
372
+ 47,
373
+ 48,
374
+ 49,
375
+ 50,
376
+ 51,
377
+ 52,
378
+ 53,
379
+ 54,
380
+ 55,
381
+ 56,
382
+ 57,
383
+ 58,
384
+ 59,
385
+ 60,
386
+ 61
387
+ ],
388
+ "episode_indices": [
389
+ 0,
390
+ 1,
391
+ 2,
392
+ 3,
393
+ 4,
394
+ 5,
395
+ 6,
396
+ 7,
397
+ 8,
398
+ 9,
399
+ 10,
400
+ 11,
401
+ 12,
402
+ 13,
403
+ 14,
404
+ 15,
405
+ 16,
406
+ 17,
407
+ 18,
408
+ 19
409
+ ]
410
+ },
411
+ "flappy/latency_4": {
412
+ "latency": 4,
413
+ "num_episodes": 20,
414
+ "mean_reward": 3.9,
415
+ "mean_length": 50.0,
416
+ "std_reward": 1.3322676295501878e-15,
417
+ "std_length": 0.0,
418
+ "episode_rewards": [
419
+ 3.8999999999999986,
420
+ 3.8999999999999986,
421
+ 3.8999999999999986,
422
+ 3.8999999999999986,
423
+ 3.8999999999999986,
424
+ 3.8999999999999986,
425
+ 3.8999999999999986,
426
+ 3.8999999999999986,
427
+ 3.8999999999999986,
428
+ 3.8999999999999986,
429
+ 3.8999999999999986,
430
+ 3.8999999999999986,
431
+ 3.8999999999999986,
432
+ 3.8999999999999986,
433
+ 3.8999999999999986,
434
+ 3.8999999999999986,
435
+ 3.8999999999999986,
436
+ 3.8999999999999986,
437
+ 3.8999999999999986,
438
+ 3.8999999999999986
439
+ ],
440
+ "episode_lengths": [
441
+ 50,
442
+ 50,
443
+ 50,
444
+ 50,
445
+ 50,
446
+ 50,
447
+ 50,
448
+ 50,
449
+ 50,
450
+ 50,
451
+ 50,
452
+ 50,
453
+ 50,
454
+ 50,
455
+ 50,
456
+ 50,
457
+ 50,
458
+ 50,
459
+ 50,
460
+ 50
461
+ ],
462
+ "decoded_action_hist": {
463
+ "0": 847,
464
+ "1": 153
465
+ },
466
+ "fixed_episode_seeds": true,
467
+ "eval_seed": 42,
468
+ "episode_seeds": [
469
+ 42,
470
+ 43,
471
+ 44,
472
+ 45,
473
+ 46,
474
+ 47,
475
+ 48,
476
+ 49,
477
+ 50,
478
+ 51,
479
+ 52,
480
+ 53,
481
+ 54,
482
+ 55,
483
+ 56,
484
+ 57,
485
+ 58,
486
+ 59,
487
+ 60,
488
+ 61
489
+ ],
490
+ "episode_indices": [
491
+ 0,
492
+ 1,
493
+ 2,
494
+ 3,
495
+ 4,
496
+ 5,
497
+ 6,
498
+ 7,
499
+ 8,
500
+ 9,
501
+ 10,
502
+ 11,
503
+ 12,
504
+ 13,
505
+ 14,
506
+ 15,
507
+ 16,
508
+ 17,
509
+ 18,
510
+ 19
511
+ ]
512
+ }
513
+ },
514
+ "aggregate": {
515
+ "stage": "post_train",
516
+ "step": 2000,
517
+ "task": "flappy",
518
+ "model_alias": "wan_oft",
519
+ "fixed_episode_seeds": true,
520
+ "eval_seed": 42,
521
+ "total_episodes": 100,
522
+ "mean_reward": 93.41700000000513,
523
+ "mean_length": 769.72,
524
+ "std_reward": 175.60116916183563,
525
+ "std_length": 1415.206755778109,
526
+ "task_count": 1,
527
+ "macro_mean_reward": 93.41700000000512,
528
+ "macro_mean_length": 769.72,
529
+ "distributed_eval": false
530
+ }
531
+ }
evaluations/flappy/wan_oft_flappy_fix_latency_3_context5_2000_effbs128/eval/mid_train/step_1000.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_latency": {
3
+ "flappy/latency_3": {
4
+ "latency": 3,
5
+ "num_episodes": 5,
6
+ "mean_reward": 259.74000000000865,
7
+ "mean_length": 2110.8,
8
+ "std_reward": 172.3612903177622,
9
+ "std_length": 1389.668651153936,
10
+ "episode_rewards": [
11
+ 444.60000000002566,
12
+ 132.79999999999737,
13
+ 6.799999999999992,
14
+ 444.60000000002566,
15
+ 269.8999999999946
16
+ ],
17
+ "episode_lengths": [
18
+ 3600,
19
+ 1087,
20
+ 70,
21
+ 3600,
22
+ 2197
23
+ ],
24
+ "decoded_action_hist": {
25
+ "0": 8510,
26
+ "1": 2044
27
+ },
28
+ "fixed_episode_seeds": true,
29
+ "eval_seed": 42,
30
+ "episode_seeds": [
31
+ 42,
32
+ 43,
33
+ 44,
34
+ 45,
35
+ 46
36
+ ],
37
+ "episode_indices": [
38
+ 0,
39
+ 1,
40
+ 2,
41
+ 3,
42
+ 4
43
+ ]
44
+ }
45
+ },
46
+ "aggregate": {
47
+ "stage": "mid_train",
48
+ "step": 1000,
49
+ "task": "flappy",
50
+ "model_alias": "wan_oft",
51
+ "fixed_episode_seeds": true,
52
+ "eval_seed": 42,
53
+ "total_episodes": 5,
54
+ "mean_reward": 259.74000000000865,
55
+ "mean_length": 2110.8,
56
+ "std_reward": 172.3612903177622,
57
+ "std_length": 1389.668651153936,
58
+ "task_count": 1,
59
+ "macro_mean_reward": 259.74000000000865,
60
+ "macro_mean_length": 2110.8,
61
+ "distributed_eval": false
62
+ }
63
+ }
evaluations/flappy/wan_oft_flappy_fix_latency_3_context5_2000_effbs128/eval/mid_train/step_1500.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_latency": {
3
+ "flappy/latency_3": {
4
+ "latency": 3,
5
+ "num_episodes": 5,
6
+ "mean_reward": 444.60000000002566,
7
+ "mean_length": 3600.0,
8
+ "std_reward": 0.0,
9
+ "std_length": 0.0,
10
+ "episode_rewards": [
11
+ 444.60000000002566,
12
+ 444.60000000002566,
13
+ 444.60000000002566,
14
+ 444.60000000002566,
15
+ 444.60000000002566
16
+ ],
17
+ "episode_lengths": [
18
+ 3600,
19
+ 3600,
20
+ 3600,
21
+ 3600,
22
+ 3600
23
+ ],
24
+ "decoded_action_hist": {
25
+ "0": 14478,
26
+ "1": 3522
27
+ },
28
+ "fixed_episode_seeds": true,
29
+ "eval_seed": 42,
30
+ "episode_seeds": [
31
+ 42,
32
+ 43,
33
+ 44,
34
+ 45,
35
+ 46
36
+ ],
37
+ "episode_indices": [
38
+ 0,
39
+ 1,
40
+ 2,
41
+ 3,
42
+ 4
43
+ ]
44
+ }
45
+ },
46
+ "aggregate": {
47
+ "stage": "mid_train",
48
+ "step": 1500,
49
+ "task": "flappy",
50
+ "model_alias": "wan_oft",
51
+ "fixed_episode_seeds": true,
52
+ "eval_seed": 42,
53
+ "total_episodes": 5,
54
+ "mean_reward": 444.60000000002566,
55
+ "mean_length": 3600.0,
56
+ "std_reward": 0.0,
57
+ "std_length": 0.0,
58
+ "task_count": 1,
59
+ "macro_mean_reward": 444.60000000002566,
60
+ "macro_mean_length": 3600.0,
61
+ "distributed_eval": false
62
+ }
63
+ }
evaluations/flappy/wan_oft_flappy_fix_latency_3_context5_2000_effbs128/eval/mid_train/step_2000.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_latency": {
3
+ "flappy/latency_3": {
4
+ "latency": 3,
5
+ "num_episodes": 5,
6
+ "mean_reward": 380.84000000002004,
7
+ "mean_length": 3087.0,
8
+ "std_reward": 127.52000000001118,
9
+ "std_length": 1026.0,
10
+ "episode_rewards": [
11
+ 125.79999999999767,
12
+ 444.60000000002566,
13
+ 444.60000000002566,
14
+ 444.60000000002566,
15
+ 444.60000000002566
16
+ ],
17
+ "episode_lengths": [
18
+ 1035,
19
+ 3600,
20
+ 3600,
21
+ 3600,
22
+ 3600
23
+ ],
24
+ "decoded_action_hist": {
25
+ "0": 12458,
26
+ "1": 2977
27
+ },
28
+ "fixed_episode_seeds": true,
29
+ "eval_seed": 42,
30
+ "episode_seeds": [
31
+ 42,
32
+ 43,
33
+ 44,
34
+ 45,
35
+ 46
36
+ ],
37
+ "episode_indices": [
38
+ 0,
39
+ 1,
40
+ 2,
41
+ 3,
42
+ 4
43
+ ]
44
+ }
45
+ },
46
+ "aggregate": {
47
+ "stage": "mid_train",
48
+ "step": 2000,
49
+ "task": "flappy",
50
+ "model_alias": "wan_oft",
51
+ "fixed_episode_seeds": true,
52
+ "eval_seed": 42,
53
+ "total_episodes": 5,
54
+ "mean_reward": 380.84000000002004,
55
+ "mean_length": 3087.0,
56
+ "std_reward": 127.52000000001118,
57
+ "std_length": 1026.0,
58
+ "task_count": 1,
59
+ "macro_mean_reward": 380.84000000002004,
60
+ "macro_mean_length": 3087.0,
61
+ "distributed_eval": false
62
+ }
63
+ }
evaluations/flappy/wan_oft_flappy_fix_latency_3_context5_2000_effbs128/eval/mid_train/step_500.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_latency": {
3
+ "flappy/latency_3": {
4
+ "latency": 3,
5
+ "num_episodes": 5,
6
+ "mean_reward": 105.19999999999952,
7
+ "mean_length": 866.8,
8
+ "std_reward": 107.90404997033255,
9
+ "std_length": 871.505226605096,
10
+ "episode_rewards": [
11
+ 302.4000000000004,
12
+ 92.49999999999916,
13
+ 8.399999999999986,
14
+ 115.89999999999812,
15
+ 6.799999999999992
16
+ ],
17
+ "episode_lengths": [
18
+ 2459,
19
+ 765,
20
+ 86,
21
+ 954,
22
+ 70
23
+ ],
24
+ "decoded_action_hist": {
25
+ "0": 3514,
26
+ "1": 820
27
+ },
28
+ "fixed_episode_seeds": true,
29
+ "eval_seed": 42,
30
+ "episode_seeds": [
31
+ 42,
32
+ 43,
33
+ 44,
34
+ 45,
35
+ 46
36
+ ],
37
+ "episode_indices": [
38
+ 0,
39
+ 1,
40
+ 2,
41
+ 3,
42
+ 4
43
+ ]
44
+ }
45
+ },
46
+ "aggregate": {
47
+ "stage": "mid_train",
48
+ "step": 500,
49
+ "task": "flappy",
50
+ "model_alias": "wan_oft",
51
+ "fixed_episode_seeds": true,
52
+ "eval_seed": 42,
53
+ "total_episodes": 5,
54
+ "mean_reward": 105.19999999999952,
55
+ "mean_length": 866.8,
56
+ "std_reward": 107.90404997033255,
57
+ "std_length": 871.505226605096,
58
+ "task_count": 1,
59
+ "macro_mean_reward": 105.19999999999952,
60
+ "macro_mean_length": 866.8,
61
+ "distributed_eval": false
62
+ }
63
+ }
evaluations/flappy/wan_oft_flappy_fix_latency_3_context5_2000_effbs128/eval/post_train/step_2000.json ADDED
@@ -0,0 +1,531 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_latency": {
3
+ "flappy/latency_0": {
4
+ "latency": 0,
5
+ "num_episodes": 20,
6
+ "mean_reward": 6.884999999999991,
7
+ "mean_length": 73.1,
8
+ "std_reward": 1.9938091684010226,
9
+ "std_length": 16.164467204334326,
10
+ "episode_rewards": [
11
+ 8.899999999999984,
12
+ 3.8999999999999986,
13
+ 4.699999999999996,
14
+ 5.999999999999995,
15
+ 3.8999999999999986,
16
+ 9.099999999999984,
17
+ 6.499999999999993,
18
+ 6.199999999999994,
19
+ 4.999999999999995,
20
+ 6.5999999999999925,
21
+ 5.999999999999995,
22
+ 4.999999999999995,
23
+ 8.599999999999985,
24
+ 5.999999999999995,
25
+ 8.899999999999984,
26
+ 9.099999999999984,
27
+ 10.799999999999981,
28
+ 4.999999999999995,
29
+ 8.899999999999984,
30
+ 8.599999999999985
31
+ ],
32
+ "episode_lengths": [
33
+ 91,
34
+ 50,
35
+ 58,
36
+ 62,
37
+ 50,
38
+ 93,
39
+ 67,
40
+ 64,
41
+ 61,
42
+ 68,
43
+ 62,
44
+ 61,
45
+ 88,
46
+ 62,
47
+ 91,
48
+ 93,
49
+ 101,
50
+ 61,
51
+ 91,
52
+ 88
53
+ ],
54
+ "decoded_action_hist": {
55
+ "0": 1265,
56
+ "1": 197
57
+ },
58
+ "fixed_episode_seeds": true,
59
+ "eval_seed": 42,
60
+ "episode_seeds": [
61
+ 42,
62
+ 43,
63
+ 44,
64
+ 45,
65
+ 46,
66
+ 47,
67
+ 48,
68
+ 49,
69
+ 50,
70
+ 51,
71
+ 52,
72
+ 53,
73
+ 54,
74
+ 55,
75
+ 56,
76
+ 57,
77
+ 58,
78
+ 59,
79
+ 60,
80
+ 61
81
+ ],
82
+ "episode_indices": [
83
+ 0,
84
+ 1,
85
+ 2,
86
+ 3,
87
+ 4,
88
+ 5,
89
+ 6,
90
+ 7,
91
+ 8,
92
+ 9,
93
+ 10,
94
+ 11,
95
+ 12,
96
+ 13,
97
+ 14,
98
+ 15,
99
+ 16,
100
+ 17,
101
+ 18,
102
+ 19
103
+ ]
104
+ },
105
+ "flappy/latency_1": {
106
+ "latency": 1,
107
+ "num_episodes": 20,
108
+ "mean_reward": 8.069999999999986,
109
+ "mean_length": 82.25,
110
+ "std_reward": 3.544587423100174,
111
+ "std_length": 28.301722562416586,
112
+ "episode_rewards": [
113
+ 9.099999999999984,
114
+ 3.8999999999999986,
115
+ 13.69999999999997,
116
+ 6.099999999999994,
117
+ 3.8999999999999986,
118
+ 13.89999999999997,
119
+ 8.999999999999984,
120
+ 14.99999999999997,
121
+ 4.799999999999995,
122
+ 9.099999999999984,
123
+ 6.099999999999994,
124
+ 4.799999999999995,
125
+ 6.799999999999992,
126
+ 6.099999999999994,
127
+ 6.799999999999992,
128
+ 14.99999999999997,
129
+ 6.5999999999999925,
130
+ 4.799999999999995,
131
+ 9.099999999999984,
132
+ 6.799999999999992
133
+ ],
134
+ "episode_lengths": [
135
+ 93,
136
+ 50,
137
+ 130,
138
+ 63,
139
+ 50,
140
+ 132,
141
+ 92,
142
+ 134,
143
+ 59,
144
+ 93,
145
+ 63,
146
+ 59,
147
+ 70,
148
+ 63,
149
+ 70,
150
+ 134,
151
+ 68,
152
+ 59,
153
+ 93,
154
+ 70
155
+ ],
156
+ "decoded_action_hist": {
157
+ "0": 1391,
158
+ "1": 254
159
+ },
160
+ "fixed_episode_seeds": true,
161
+ "eval_seed": 42,
162
+ "episode_seeds": [
163
+ 42,
164
+ 43,
165
+ 44,
166
+ 45,
167
+ 46,
168
+ 47,
169
+ 48,
170
+ 49,
171
+ 50,
172
+ 51,
173
+ 52,
174
+ 53,
175
+ 54,
176
+ 55,
177
+ 56,
178
+ 57,
179
+ 58,
180
+ 59,
181
+ 60,
182
+ 61
183
+ ],
184
+ "episode_indices": [
185
+ 0,
186
+ 1,
187
+ 2,
188
+ 3,
189
+ 4,
190
+ 5,
191
+ 6,
192
+ 7,
193
+ 8,
194
+ 9,
195
+ 10,
196
+ 11,
197
+ 12,
198
+ 13,
199
+ 14,
200
+ 15,
201
+ 16,
202
+ 17,
203
+ 18,
204
+ 19
205
+ ]
206
+ },
207
+ "flappy/latency_2": {
208
+ "latency": 2,
209
+ "num_episodes": 20,
210
+ "mean_reward": 13.190000000000015,
211
+ "mean_length": 121.75,
212
+ "std_reward": 9.237201957302931,
213
+ "std_length": 74.88850045233914,
214
+ "episode_rewards": [
215
+ 8.399999999999986,
216
+ 25.200000000000085,
217
+ 3.8999999999999986,
218
+ 37.60000000000023,
219
+ 11.09999999999998,
220
+ 8.399999999999986,
221
+ 6.799999999999992,
222
+ 20.700000000000035,
223
+ 6.699999999999992,
224
+ 10.99999999999998,
225
+ 6.699999999999992,
226
+ 6.299999999999994,
227
+ 34.20000000000018,
228
+ 6.5999999999999925,
229
+ 15.79999999999998,
230
+ 10.599999999999982,
231
+ 6.799999999999992,
232
+ 6.299999999999994,
233
+ 15.599999999999977,
234
+ 15.09999999999997
235
+ ],
236
+ "episode_lengths": [
237
+ 86,
238
+ 218,
239
+ 50,
240
+ 324,
241
+ 104,
242
+ 86,
243
+ 70,
244
+ 182,
245
+ 69,
246
+ 103,
247
+ 69,
248
+ 65,
249
+ 290,
250
+ 68,
251
+ 142,
252
+ 99,
253
+ 70,
254
+ 65,
255
+ 140,
256
+ 135
257
+ ],
258
+ "decoded_action_hist": {
259
+ "0": 2042,
260
+ "1": 393
261
+ },
262
+ "fixed_episode_seeds": true,
263
+ "eval_seed": 42,
264
+ "episode_seeds": [
265
+ 42,
266
+ 43,
267
+ 44,
268
+ 45,
269
+ 46,
270
+ 47,
271
+ 48,
272
+ 49,
273
+ 50,
274
+ 51,
275
+ 52,
276
+ 53,
277
+ 54,
278
+ 55,
279
+ 56,
280
+ 57,
281
+ 58,
282
+ 59,
283
+ 60,
284
+ 61
285
+ ],
286
+ "episode_indices": [
287
+ 0,
288
+ 1,
289
+ 2,
290
+ 3,
291
+ 4,
292
+ 5,
293
+ 6,
294
+ 7,
295
+ 8,
296
+ 9,
297
+ 10,
298
+ 11,
299
+ 12,
300
+ 13,
301
+ 14,
302
+ 15,
303
+ 16,
304
+ 17,
305
+ 18,
306
+ 19
307
+ ]
308
+ },
309
+ "flappy/latency_3": {
310
+ "latency": 3,
311
+ "num_episodes": 20,
312
+ "mean_reward": 427.8800000000241,
313
+ "mean_length": 3465.85,
314
+ "std_reward": 69.38502432082045,
315
+ "std_length": 558.2662693554036,
316
+ "episode_rewards": [
317
+ 125.79999999999767,
318
+ 444.60000000002566,
319
+ 444.60000000002566,
320
+ 444.60000000002566,
321
+ 444.60000000002566,
322
+ 444.60000000002566,
323
+ 444.60000000002566,
324
+ 444.60000000002566,
325
+ 444.60000000002566,
326
+ 444.60000000002566,
327
+ 444.60000000002566,
328
+ 444.60000000002566,
329
+ 444.60000000002566,
330
+ 429.000000000023,
331
+ 444.60000000002566,
332
+ 444.60000000002566,
333
+ 444.60000000002566,
334
+ 444.60000000002566,
335
+ 444.60000000002566,
336
+ 444.60000000002566
337
+ ],
338
+ "episode_lengths": [
339
+ 1035,
340
+ 3600,
341
+ 3600,
342
+ 3600,
343
+ 3600,
344
+ 3600,
345
+ 3600,
346
+ 3600,
347
+ 3600,
348
+ 3600,
349
+ 3600,
350
+ 3600,
351
+ 3600,
352
+ 3482,
353
+ 3600,
354
+ 3600,
355
+ 3600,
356
+ 3600,
357
+ 3600,
358
+ 3600
359
+ ],
360
+ "decoded_action_hist": {
361
+ "0": 55767,
362
+ "1": 13550
363
+ },
364
+ "fixed_episode_seeds": true,
365
+ "eval_seed": 42,
366
+ "episode_seeds": [
367
+ 42,
368
+ 43,
369
+ 44,
370
+ 45,
371
+ 46,
372
+ 47,
373
+ 48,
374
+ 49,
375
+ 50,
376
+ 51,
377
+ 52,
378
+ 53,
379
+ 54,
380
+ 55,
381
+ 56,
382
+ 57,
383
+ 58,
384
+ 59,
385
+ 60,
386
+ 61
387
+ ],
388
+ "episode_indices": [
389
+ 0,
390
+ 1,
391
+ 2,
392
+ 3,
393
+ 4,
394
+ 5,
395
+ 6,
396
+ 7,
397
+ 8,
398
+ 9,
399
+ 10,
400
+ 11,
401
+ 12,
402
+ 13,
403
+ 14,
404
+ 15,
405
+ 16,
406
+ 17,
407
+ 18,
408
+ 19
409
+ ]
410
+ },
411
+ "flappy/latency_4": {
412
+ "latency": 4,
413
+ "num_episodes": 20,
414
+ "mean_reward": 7.360000000000005,
415
+ "mean_length": 78.75,
416
+ "std_reward": 7.047013551853045,
417
+ "std_length": 57.37061530086635,
418
+ "episode_rewards": [
419
+ 3.8999999999999986,
420
+ 3.8999999999999986,
421
+ 4.799999999999995,
422
+ 4.4999999999999964,
423
+ 3.8999999999999986,
424
+ 3.8999999999999986,
425
+ 33.00000000000018,
426
+ 3.8999999999999986,
427
+ 13.99999999999997,
428
+ 3.8999999999999986,
429
+ 4.399999999999997,
430
+ 9.499999999999982,
431
+ 4.999999999999995,
432
+ 4.4999999999999964,
433
+ 4.999999999999995,
434
+ 3.8999999999999986,
435
+ 3.8999999999999986,
436
+ 8.399999999999986,
437
+ 19.000000000000025,
438
+ 3.8999999999999986
439
+ ],
440
+ "episode_lengths": [
441
+ 50,
442
+ 50,
443
+ 59,
444
+ 56,
445
+ 50,
446
+ 50,
447
+ 287,
448
+ 50,
449
+ 133,
450
+ 50,
451
+ 55,
452
+ 97,
453
+ 61,
454
+ 56,
455
+ 61,
456
+ 50,
457
+ 50,
458
+ 86,
459
+ 174,
460
+ 50
461
+ ],
462
+ "decoded_action_hist": {
463
+ "0": 1160,
464
+ "1": 415
465
+ },
466
+ "fixed_episode_seeds": true,
467
+ "eval_seed": 42,
468
+ "episode_seeds": [
469
+ 42,
470
+ 43,
471
+ 44,
472
+ 45,
473
+ 46,
474
+ 47,
475
+ 48,
476
+ 49,
477
+ 50,
478
+ 51,
479
+ 52,
480
+ 53,
481
+ 54,
482
+ 55,
483
+ 56,
484
+ 57,
485
+ 58,
486
+ 59,
487
+ 60,
488
+ 61
489
+ ],
490
+ "episode_indices": [
491
+ 0,
492
+ 1,
493
+ 2,
494
+ 3,
495
+ 4,
496
+ 5,
497
+ 6,
498
+ 7,
499
+ 8,
500
+ 9,
501
+ 10,
502
+ 11,
503
+ 12,
504
+ 13,
505
+ 14,
506
+ 15,
507
+ 16,
508
+ 17,
509
+ 18,
510
+ 19
511
+ ]
512
+ }
513
+ },
514
+ "aggregate": {
515
+ "stage": "post_train",
516
+ "step": 2000,
517
+ "task": "flappy",
518
+ "model_alias": "wan_oft",
519
+ "fixed_episode_seeds": true,
520
+ "eval_seed": 42,
521
+ "total_episodes": 100,
522
+ "mean_reward": 92.67700000000481,
523
+ "mean_length": 764.34,
524
+ "std_reward": 170.5535950104934,
525
+ "std_length": 1374.4667709333683,
526
+ "task_count": 1,
527
+ "macro_mean_reward": 92.67700000000482,
528
+ "macro_mean_length": 764.3399999999999,
529
+ "distributed_eval": false
530
+ }
531
+ }
evaluations/flappy/wan_oft_flappy_fix_latency_4_context5_2000_effbs128/eval/mid_train/step_1000.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_latency": {
3
+ "flappy/latency_4": {
4
+ "latency": 4,
5
+ "num_episodes": 5,
6
+ "mean_reward": 386.7400000000153,
7
+ "mean_length": 3133.8,
8
+ "std_reward": 70.92473757443024,
9
+ "std_length": 571.468424324564,
10
+ "episode_rewards": [
11
+ 295.299999999999,
12
+ 444.60000000002566,
13
+ 444.60000000002566,
14
+ 444.60000000002566,
15
+ 304.60000000000065
16
+ ],
17
+ "episode_lengths": [
18
+ 2397,
19
+ 3600,
20
+ 3600,
21
+ 3600,
22
+ 2472
23
+ ],
24
+ "decoded_action_hist": {
25
+ "0": 12809,
26
+ "1": 2860
27
+ },
28
+ "fixed_episode_seeds": true,
29
+ "eval_seed": 42,
30
+ "episode_seeds": [
31
+ 42,
32
+ 43,
33
+ 44,
34
+ 45,
35
+ 46
36
+ ],
37
+ "episode_indices": [
38
+ 0,
39
+ 1,
40
+ 2,
41
+ 3,
42
+ 4
43
+ ]
44
+ }
45
+ },
46
+ "aggregate": {
47
+ "stage": "mid_train",
48
+ "step": 1000,
49
+ "task": "flappy",
50
+ "model_alias": "wan_oft",
51
+ "fixed_episode_seeds": true,
52
+ "eval_seed": 42,
53
+ "total_episodes": 5,
54
+ "mean_reward": 386.7400000000153,
55
+ "mean_length": 3133.8,
56
+ "std_reward": 70.92473757443024,
57
+ "std_length": 571.468424324564,
58
+ "task_count": 1,
59
+ "macro_mean_reward": 386.7400000000153,
60
+ "macro_mean_length": 3133.8,
61
+ "distributed_eval": false
62
+ }
63
+ }
evaluations/flappy/wan_oft_flappy_fix_latency_4_context5_2000_effbs128/eval/mid_train/step_1500.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_latency": {
3
+ "flappy/latency_4": {
4
+ "latency": 4,
5
+ "num_episodes": 5,
6
+ "mean_reward": 416.40000000002067,
7
+ "mean_length": 3374.2,
8
+ "std_reward": 56.40000000001,
9
+ "std_length": 451.6,
10
+ "episode_rewards": [
11
+ 444.60000000002566,
12
+ 444.60000000002566,
13
+ 444.60000000002566,
14
+ 444.60000000002566,
15
+ 303.60000000000065
16
+ ],
17
+ "episode_lengths": [
18
+ 3600,
19
+ 3600,
20
+ 3600,
21
+ 3600,
22
+ 2471
23
+ ],
24
+ "decoded_action_hist": {
25
+ "0": 13846,
26
+ "1": 3025
27
+ },
28
+ "fixed_episode_seeds": true,
29
+ "eval_seed": 42,
30
+ "episode_seeds": [
31
+ 42,
32
+ 43,
33
+ 44,
34
+ 45,
35
+ 46
36
+ ],
37
+ "episode_indices": [
38
+ 0,
39
+ 1,
40
+ 2,
41
+ 3,
42
+ 4
43
+ ]
44
+ }
45
+ },
46
+ "aggregate": {
47
+ "stage": "mid_train",
48
+ "step": 1500,
49
+ "task": "flappy",
50
+ "model_alias": "wan_oft",
51
+ "fixed_episode_seeds": true,
52
+ "eval_seed": 42,
53
+ "total_episodes": 5,
54
+ "mean_reward": 416.40000000002067,
55
+ "mean_length": 3374.2,
56
+ "std_reward": 56.40000000001,
57
+ "std_length": 451.6,
58
+ "task_count": 1,
59
+ "macro_mean_reward": 416.40000000002067,
60
+ "macro_mean_length": 3374.2,
61
+ "distributed_eval": false
62
+ }
63
+ }
evaluations/flappy/wan_oft_flappy_fix_latency_4_context5_2000_effbs128/eval/mid_train/step_2000.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_latency": {
3
+ "flappy/latency_4": {
4
+ "latency": 4,
5
+ "num_episodes": 5,
6
+ "mean_reward": 444.60000000002566,
7
+ "mean_length": 3600.0,
8
+ "std_reward": 0.0,
9
+ "std_length": 0.0,
10
+ "episode_rewards": [
11
+ 444.60000000002566,
12
+ 444.60000000002566,
13
+ 444.60000000002566,
14
+ 444.60000000002566,
15
+ 444.60000000002566
16
+ ],
17
+ "episode_lengths": [
18
+ 3600,
19
+ 3600,
20
+ 3600,
21
+ 3600,
22
+ 3600
23
+ ],
24
+ "decoded_action_hist": {
25
+ "0": 14774,
26
+ "1": 3226
27
+ },
28
+ "fixed_episode_seeds": true,
29
+ "eval_seed": 42,
30
+ "episode_seeds": [
31
+ 42,
32
+ 43,
33
+ 44,
34
+ 45,
35
+ 46
36
+ ],
37
+ "episode_indices": [
38
+ 0,
39
+ 1,
40
+ 2,
41
+ 3,
42
+ 4
43
+ ]
44
+ }
45
+ },
46
+ "aggregate": {
47
+ "stage": "mid_train",
48
+ "step": 2000,
49
+ "task": "flappy",
50
+ "model_alias": "wan_oft",
51
+ "fixed_episode_seeds": true,
52
+ "eval_seed": 42,
53
+ "total_episodes": 5,
54
+ "mean_reward": 444.60000000002566,
55
+ "mean_length": 3600.0,
56
+ "std_reward": 0.0,
57
+ "std_length": 0.0,
58
+ "task_count": 1,
59
+ "macro_mean_reward": 444.60000000002566,
60
+ "macro_mean_length": 3600.0,
61
+ "distributed_eval": false
62
+ }
63
+ }
evaluations/flappy/wan_oft_flappy_fix_latency_4_context5_2000_effbs128/eval/mid_train/step_500.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_latency": {
3
+ "flappy/latency_4": {
4
+ "latency": 4,
5
+ "num_episodes": 5,
6
+ "mean_reward": 60.739999999999384,
7
+ "mean_length": 506.8,
8
+ "std_reward": 44.601955114097024,
9
+ "std_length": 359.98744422548964,
10
+ "episode_rewards": [
11
+ 104.29999999999866,
12
+ 50.40000000000037,
13
+ 20.200000000000028,
14
+ 120.39999999999792,
15
+ 8.399999999999986
16
+ ],
17
+ "episode_lengths": [
18
+ 856,
19
+ 425,
20
+ 177,
21
+ 990,
22
+ 86
23
+ ],
24
+ "decoded_action_hist": {
25
+ "0": 2136,
26
+ "1": 398
27
+ },
28
+ "fixed_episode_seeds": true,
29
+ "eval_seed": 42,
30
+ "episode_seeds": [
31
+ 42,
32
+ 43,
33
+ 44,
34
+ 45,
35
+ 46
36
+ ],
37
+ "episode_indices": [
38
+ 0,
39
+ 1,
40
+ 2,
41
+ 3,
42
+ 4
43
+ ]
44
+ }
45
+ },
46
+ "aggregate": {
47
+ "stage": "mid_train",
48
+ "step": 500,
49
+ "task": "flappy",
50
+ "model_alias": "wan_oft",
51
+ "fixed_episode_seeds": true,
52
+ "eval_seed": 42,
53
+ "total_episodes": 5,
54
+ "mean_reward": 60.739999999999384,
55
+ "mean_length": 506.8,
56
+ "std_reward": 44.601955114097024,
57
+ "std_length": 359.98744422548964,
58
+ "task_count": 1,
59
+ "macro_mean_reward": 60.739999999999384,
60
+ "macro_mean_length": 506.8,
61
+ "distributed_eval": false
62
+ }
63
+ }
evaluations/flappy/wan_oft_flappy_fix_latency_4_context5_2000_effbs128/eval/post_train/step_2000.json ADDED
@@ -0,0 +1,531 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_latency": {
3
+ "flappy/latency_0": {
4
+ "latency": 0,
5
+ "num_episodes": 20,
6
+ "mean_reward": 5.569999999999995,
7
+ "mean_length": 63.55,
8
+ "std_reward": 2.3916730545791514,
9
+ "std_length": 19.043305910476782,
10
+ "episode_rewards": [
11
+ 6.499999999999993,
12
+ 4.099999999999998,
13
+ 4.399999999999997,
14
+ 4.799999999999995,
15
+ 4.099999999999998,
16
+ 8.399999999999986,
17
+ 3.8999999999999986,
18
+ 6.5999999999999925,
19
+ 4.699999999999996,
20
+ 3.8999999999999986,
21
+ 9.499999999999982,
22
+ 4.699999999999996,
23
+ 3.8999999999999986,
24
+ 4.799999999999995,
25
+ 3.8999999999999986,
26
+ 8.399999999999986,
27
+ 4.099999999999998,
28
+ 12.899999999999974,
29
+ 3.8999999999999986,
30
+ 3.8999999999999986
31
+ ],
32
+ "episode_lengths": [
33
+ 67,
34
+ 52,
35
+ 55,
36
+ 59,
37
+ 52,
38
+ 86,
39
+ 50,
40
+ 68,
41
+ 58,
42
+ 50,
43
+ 97,
44
+ 58,
45
+ 50,
46
+ 59,
47
+ 50,
48
+ 86,
49
+ 52,
50
+ 122,
51
+ 50,
52
+ 50
53
+ ],
54
+ "decoded_action_hist": {
55
+ "0": 1200,
56
+ "1": 71
57
+ },
58
+ "fixed_episode_seeds": true,
59
+ "eval_seed": 42,
60
+ "episode_seeds": [
61
+ 42,
62
+ 43,
63
+ 44,
64
+ 45,
65
+ 46,
66
+ 47,
67
+ 48,
68
+ 49,
69
+ 50,
70
+ 51,
71
+ 52,
72
+ 53,
73
+ 54,
74
+ 55,
75
+ 56,
76
+ 57,
77
+ 58,
78
+ 59,
79
+ 60,
80
+ 61
81
+ ],
82
+ "episode_indices": [
83
+ 0,
84
+ 1,
85
+ 2,
86
+ 3,
87
+ 4,
88
+ 5,
89
+ 6,
90
+ 7,
91
+ 8,
92
+ 9,
93
+ 10,
94
+ 11,
95
+ 12,
96
+ 13,
97
+ 14,
98
+ 15,
99
+ 16,
100
+ 17,
101
+ 18,
102
+ 19
103
+ ]
104
+ },
105
+ "flappy/latency_1": {
106
+ "latency": 1,
107
+ "num_episodes": 20,
108
+ "mean_reward": 5.264999999999994,
109
+ "mean_length": 60.5,
110
+ "std_reward": 1.524557312796075,
111
+ "std_length": 10.49047186736612,
112
+ "episode_rewards": [
113
+ 4.299999999999997,
114
+ 4.699999999999996,
115
+ 4.799999999999995,
116
+ 6.199999999999994,
117
+ 4.699999999999996,
118
+ 4.4999999999999964,
119
+ 4.599999999999996,
120
+ 4.1999999999999975,
121
+ 6.499999999999993,
122
+ 4.399999999999997,
123
+ 6.199999999999994,
124
+ 6.499999999999993,
125
+ 4.1999999999999975,
126
+ 6.199999999999994,
127
+ 4.1999999999999975,
128
+ 4.4999999999999964,
129
+ 4.699999999999996,
130
+ 10.89999999999998,
131
+ 4.599999999999996,
132
+ 4.399999999999997
133
+ ],
134
+ "episode_lengths": [
135
+ 54,
136
+ 58,
137
+ 59,
138
+ 64,
139
+ 58,
140
+ 56,
141
+ 57,
142
+ 53,
143
+ 67,
144
+ 55,
145
+ 64,
146
+ 67,
147
+ 53,
148
+ 64,
149
+ 53,
150
+ 56,
151
+ 58,
152
+ 102,
153
+ 57,
154
+ 55
155
+ ],
156
+ "decoded_action_hist": {
157
+ "0": 1083,
158
+ "1": 127
159
+ },
160
+ "fixed_episode_seeds": true,
161
+ "eval_seed": 42,
162
+ "episode_seeds": [
163
+ 42,
164
+ 43,
165
+ 44,
166
+ 45,
167
+ 46,
168
+ 47,
169
+ 48,
170
+ 49,
171
+ 50,
172
+ 51,
173
+ 52,
174
+ 53,
175
+ 54,
176
+ 55,
177
+ 56,
178
+ 57,
179
+ 58,
180
+ 59,
181
+ 60,
182
+ 61
183
+ ],
184
+ "episode_indices": [
185
+ 0,
186
+ 1,
187
+ 2,
188
+ 3,
189
+ 4,
190
+ 5,
191
+ 6,
192
+ 7,
193
+ 8,
194
+ 9,
195
+ 10,
196
+ 11,
197
+ 12,
198
+ 13,
199
+ 14,
200
+ 15,
201
+ 16,
202
+ 17,
203
+ 18,
204
+ 19
205
+ ]
206
+ },
207
+ "flappy/latency_2": {
208
+ "latency": 2,
209
+ "num_episodes": 20,
210
+ "mean_reward": 5.4649999999999945,
211
+ "mean_length": 62.05,
212
+ "std_reward": 0.8956980518009389,
213
+ "std_length": 4.831925082200675,
214
+ "episode_rewards": [
215
+ 4.699999999999996,
216
+ 5.999999999999995,
217
+ 4.999999999999995,
218
+ 6.5999999999999925,
219
+ 4.899999999999995,
220
+ 4.899999999999995,
221
+ 4.4999999999999964,
222
+ 4.899999999999995,
223
+ 6.799999999999992,
224
+ 4.799999999999995,
225
+ 6.5999999999999925,
226
+ 6.799999999999992,
227
+ 4.699999999999996,
228
+ 6.5999999999999925,
229
+ 4.699999999999996,
230
+ 4.699999999999996,
231
+ 5.999999999999995,
232
+ 6.799999999999992,
233
+ 4.4999999999999964,
234
+ 4.799999999999995
235
+ ],
236
+ "episode_lengths": [
237
+ 58,
238
+ 62,
239
+ 61,
240
+ 68,
241
+ 60,
242
+ 60,
243
+ 56,
244
+ 60,
245
+ 70,
246
+ 59,
247
+ 68,
248
+ 70,
249
+ 58,
250
+ 68,
251
+ 58,
252
+ 58,
253
+ 62,
254
+ 70,
255
+ 56,
256
+ 59
257
+ ],
258
+ "decoded_action_hist": {
259
+ "0": 1082,
260
+ "1": 159
261
+ },
262
+ "fixed_episode_seeds": true,
263
+ "eval_seed": 42,
264
+ "episode_seeds": [
265
+ 42,
266
+ 43,
267
+ 44,
268
+ 45,
269
+ 46,
270
+ 47,
271
+ 48,
272
+ 49,
273
+ 50,
274
+ 51,
275
+ 52,
276
+ 53,
277
+ 54,
278
+ 55,
279
+ 56,
280
+ 57,
281
+ 58,
282
+ 59,
283
+ 60,
284
+ 61
285
+ ],
286
+ "episode_indices": [
287
+ 0,
288
+ 1,
289
+ 2,
290
+ 3,
291
+ 4,
292
+ 5,
293
+ 6,
294
+ 7,
295
+ 8,
296
+ 9,
297
+ 10,
298
+ 11,
299
+ 12,
300
+ 13,
301
+ 14,
302
+ 15,
303
+ 16,
304
+ 17,
305
+ 18,
306
+ 19
307
+ ]
308
+ },
309
+ "flappy/latency_3": {
310
+ "latency": 3,
311
+ "num_episodes": 20,
312
+ "mean_reward": 7.604999999999999,
313
+ "mean_length": 79.4,
314
+ "std_reward": 5.2202945319206115,
315
+ "std_length": 41.852598485637664,
316
+ "episode_rewards": [
317
+ 4.699999999999996,
318
+ 6.799999999999992,
319
+ 6.5999999999999925,
320
+ 6.799999999999992,
321
+ 6.799999999999992,
322
+ 4.699999999999996,
323
+ 4.799999999999995,
324
+ 4.699999999999996,
325
+ 27.80000000000012,
326
+ 4.999999999999995,
327
+ 6.799999999999992,
328
+ 9.199999999999983,
329
+ 4.899999999999995,
330
+ 6.799999999999992,
331
+ 4.899999999999995,
332
+ 4.699999999999996,
333
+ 6.799999999999992,
334
+ 9.099999999999984,
335
+ 15.19999999999997,
336
+ 4.999999999999995
337
+ ],
338
+ "episode_lengths": [
339
+ 58,
340
+ 70,
341
+ 68,
342
+ 70,
343
+ 70,
344
+ 58,
345
+ 59,
346
+ 58,
347
+ 244,
348
+ 61,
349
+ 70,
350
+ 94,
351
+ 60,
352
+ 70,
353
+ 60,
354
+ 58,
355
+ 70,
356
+ 93,
357
+ 136,
358
+ 61
359
+ ],
360
+ "decoded_action_hist": {
361
+ "0": 1361,
362
+ "1": 227
363
+ },
364
+ "fixed_episode_seeds": true,
365
+ "eval_seed": 42,
366
+ "episode_seeds": [
367
+ 42,
368
+ 43,
369
+ 44,
370
+ 45,
371
+ 46,
372
+ 47,
373
+ 48,
374
+ 49,
375
+ 50,
376
+ 51,
377
+ 52,
378
+ 53,
379
+ 54,
380
+ 55,
381
+ 56,
382
+ 57,
383
+ 58,
384
+ 59,
385
+ 60,
386
+ 61
387
+ ],
388
+ "episode_indices": [
389
+ 0,
390
+ 1,
391
+ 2,
392
+ 3,
393
+ 4,
394
+ 5,
395
+ 6,
396
+ 7,
397
+ 8,
398
+ 9,
399
+ 10,
400
+ 11,
401
+ 12,
402
+ 13,
403
+ 14,
404
+ 15,
405
+ 16,
406
+ 17,
407
+ 18,
408
+ 19
409
+ ]
410
+ },
411
+ "flappy/latency_4": {
412
+ "latency": 4,
413
+ "num_episodes": 20,
414
+ "mean_reward": 413.42000000002145,
415
+ "mean_length": 3349.25,
416
+ "std_reward": 77.46433760125731,
417
+ "std_length": 623.3434747392484,
418
+ "episode_rewards": [
419
+ 444.60000000002566,
420
+ 444.60000000002566,
421
+ 444.60000000002566,
422
+ 444.60000000002566,
423
+ 444.60000000002566,
424
+ 444.60000000002566,
425
+ 444.60000000002566,
426
+ 444.60000000002566,
427
+ 444.60000000002566,
428
+ 444.60000000002566,
429
+ 444.60000000002566,
430
+ 444.60000000002566,
431
+ 444.60000000002566,
432
+ 444.60000000002566,
433
+ 444.60000000002566,
434
+ 444.60000000002566,
435
+ 202.39999999999426,
436
+ 317.4000000000031,
437
+ 190.39999999999478,
438
+ 444.60000000002566
439
+ ],
440
+ "episode_lengths": [
441
+ 3600,
442
+ 3600,
443
+ 3600,
444
+ 3600,
445
+ 3600,
446
+ 3600,
447
+ 3600,
448
+ 3600,
449
+ 3600,
450
+ 3600,
451
+ 3600,
452
+ 3600,
453
+ 3600,
454
+ 3600,
455
+ 3600,
456
+ 3600,
457
+ 1648,
458
+ 2582,
459
+ 1555,
460
+ 3600
461
+ ],
462
+ "decoded_action_hist": {
463
+ "0": 54994,
464
+ "1": 11991
465
+ },
466
+ "fixed_episode_seeds": true,
467
+ "eval_seed": 42,
468
+ "episode_seeds": [
469
+ 42,
470
+ 43,
471
+ 44,
472
+ 45,
473
+ 46,
474
+ 47,
475
+ 48,
476
+ 49,
477
+ 50,
478
+ 51,
479
+ 52,
480
+ 53,
481
+ 54,
482
+ 55,
483
+ 56,
484
+ 57,
485
+ 58,
486
+ 59,
487
+ 60,
488
+ 61
489
+ ],
490
+ "episode_indices": [
491
+ 0,
492
+ 1,
493
+ 2,
494
+ 3,
495
+ 4,
496
+ 5,
497
+ 6,
498
+ 7,
499
+ 8,
500
+ 9,
501
+ 10,
502
+ 11,
503
+ 12,
504
+ 13,
505
+ 14,
506
+ 15,
507
+ 16,
508
+ 17,
509
+ 18,
510
+ 19
511
+ ]
512
+ }
513
+ },
514
+ "aggregate": {
515
+ "stage": "post_train",
516
+ "step": 2000,
517
+ "task": "flappy",
518
+ "model_alias": "wan_oft",
519
+ "fixed_episode_seeds": true,
520
+ "eval_seed": 42,
521
+ "total_episodes": 100,
522
+ "mean_reward": 87.46500000000428,
523
+ "mean_length": 722.95,
524
+ "std_reward": 166.64257221671508,
525
+ "std_length": 1342.5982450085357,
526
+ "task_count": 1,
527
+ "macro_mean_reward": 87.46500000000428,
528
+ "macro_mean_length": 722.95,
529
+ "distributed_eval": false
530
+ }
531
+ }
experiments/latency-transfer/provenance/latency-transfer-data/provenance/copy_verification.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "source_files_verified": 735,
3
+ "preexisting_files_preserved": 1435,
4
+ "identities_match": true
5
+ }
experiments/latency-transfer/provenance/latency-transfer-data/provenance/data_readability_verification.json ADDED
The diff for this file is too large to render. See raw diff
 
experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/flappy_fix_latency_0_200ep_context5/filter_report.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "accepted_episodes": 200,
3
+ "accepted_train_episodes": 180,
4
+ "accepted_val_episodes": 20,
5
+ "attempted_episodes": 227,
6
+ "preset": "clean_v1",
7
+ "rejected_episodes": 27,
8
+ "train_rows_after_row_filter": 279882,
9
+ "train_rows_before_row_filter": 648000,
10
+ "val_rows_after_row_filter": 72000,
11
+ "val_rows_before_row_filter": 72000
12
+ }
experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/flappy_fix_latency_0_200ep_context5/metadata.json ADDED
@@ -0,0 +1,305 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "action_labels": [
3
+ "NOOP",
4
+ "FLAP"
5
+ ],
6
+ "action_layout": "flappy_discrete",
7
+ "base_prompt": "You are playing Flappy Bird. Pass through the pipe gaps and stay alive. Choose the action: NOOP, FLAP.",
8
+ "checkpoint_config": {
9
+ "actor_critic_share_weights": true,
10
+ "actor_worker_gpus": [
11
+ 0
12
+ ],
13
+ "adam_beta1": 0.9,
14
+ "adam_beta2": 0.999,
15
+ "adam_eps": 1e-05,
16
+ "adaptive_stddev": true,
17
+ "add_latency_info": false,
18
+ "algo": "APPO",
19
+ "async_rl": true,
20
+ "audio_on": false,
21
+ "background": "day",
22
+ "batch_size": 4096,
23
+ "batched_sampling": true,
24
+ "benchmark": false,
25
+ "bird_color": "yellow",
26
+ "cli_args": {
27
+ "actor_worker_gpus": [
28
+ 0
29
+ ],
30
+ "add_latency_info": false,
31
+ "algo": "APPO",
32
+ "async_rl": true,
33
+ "audio_on": false,
34
+ "background": "day",
35
+ "batch_size": 4096,
36
+ "batched_sampling": true,
37
+ "bird_color": "yellow",
38
+ "debug": false,
39
+ "debug_timelimit_diagnostics": false,
40
+ "device": "gpu",
41
+ "env": "latency_flappy",
42
+ "env_gpu_actions": false,
43
+ "env_gpu_observations": true,
44
+ "episode_metrics_path": "results/checkpoints/flappy_sf_same_latency_l0_seed0/episode_metrics.jsonl",
45
+ "episodes_per_stage": 10,
46
+ "eval_deterministic": true,
47
+ "eval_episodes": 20,
48
+ "eval_max_steps": 3600,
49
+ "eval_queue_latency_values": "0,1,2,3,4,5",
50
+ "experiment": "flappy_sf_same_latency_l0_seed0",
51
+ "experiment_summaries_interval": 1,
52
+ "frame_stack": 4,
53
+ "gae_lambda": 0.95,
54
+ "gamma": 0.99,
55
+ "gpu_render_batch_size": 64,
56
+ "gpu_render_device": "auto",
57
+ "gpu_render_profile": false,
58
+ "gpu_render_profile_interval": 200,
59
+ "gym_id": "FlappyBird-v0",
60
+ "keep_checkpoints": 5,
61
+ "latency_mode": "episode_fixed",
62
+ "latency_sampling_mode": "uniform_all",
63
+ "learning_rate": 0.00025,
64
+ "max_grad_norm": 0.5,
65
+ "mode": "train",
66
+ "normalize_input": true,
67
+ "normalize_obs": true,
68
+ "normalize_returns": true,
69
+ "num_batches_per_epoch": 8,
70
+ "num_envs_per_worker": 1,
71
+ "num_epochs": 8,
72
+ "num_policies": 1,
73
+ "num_workers": 2,
74
+ "obs_mode": "image_only",
75
+ "obs_resize": "84,84",
76
+ "pipe_color": "green",
77
+ "pipe_gap": 100,
78
+ "ppo_clip_ratio": 0.1,
79
+ "ppo_clip_value": 0.2,
80
+ "queue_latency_values": "0",
81
+ "randomize_within_stage": false,
82
+ "recurrence": 1,
83
+ "restart_behavior": "resume",
84
+ "rollout": 256,
85
+ "save_every_sec": 600,
86
+ "seed": 0,
87
+ "stats_avg": 100,
88
+ "train_dir": "results/checkpoints",
89
+ "train_for_env_steps": 15000000,
90
+ "use_gpu_render": true,
91
+ "use_lidar": false,
92
+ "use_rnn": false,
93
+ "value_loss_coeff": 0.5,
94
+ "wandb_group": "flappy-new(remove training episode limit)",
95
+ "wandb_job_type": "sample_factory",
96
+ "wandb_project": "latency-sensitive-bench",
97
+ "with_wandb": true,
98
+ "within_episode_change_interval": 1,
99
+ "within_episode_change_prob": 1.0,
100
+ "worker_num_splits": 1
101
+ },
102
+ "command_line": "--mode train --algo APPO --env latency_flappy --experiment flappy_sf_same_latency_l0_seed0 --train_dir results/checkpoints --restart_behavior resume --device gpu --actor_worker_gpus 0 --env_gpu_observations True --env_gpu_actions False --gpu-render-batch-size 64 --seed 0 --episode_metrics_path results/checkpoints/flappy_sf_same_latency_l0_seed0/episode_metrics.jsonl --train_for_env_steps 15000000 --num_workers 2 --num_envs_per_worker 1 --num_policies 1 --batch_size 4096 --rollout 256 --recurrence 1 --num_epochs 8 --num_batches_per_epoch 8 --worker_num_splits 1 --learning_rate 0.00025 --gamma 0.99 --gae_lambda 0.95 --ppo_clip_ratio 0.1 --ppo_clip_value 0.2 --value_loss_coeff 0.5 --max_grad_norm 0.5 --save_every_sec 600 --keep_checkpoints 5 --stats_avg 100 --experiment_summaries_interval 1 --batched_sampling True --async_rl True --use_rnn False --normalize_returns True --normalize_input True --queue-latency-values 0 --latency-mode episode_fixed --latency-sampling-mode uniform_all --episodes-per-stage 10 --randomize-within-stage False --within-episode-change-interval 1 --within-episode-change-prob 1.0 --add-latency-info False --eval-episodes 20 --eval-queue-latency-values 0,1,2,3,4,5 --eval-max-steps 3600 --eval-deterministic True --with_wandb True --wandb_project latency-sensitive-bench --wandb_group flappy-new(remove training episode limit) --wandb_job_type sample_factory --gym_id FlappyBird-v0 --use_lidar False --normalize_obs True --audio_on False --obs_resize 84,84 --obs_mode image_only --use-gpu-render True --gpu-render-device auto --gpu-render-profile False --gpu-render-profile-interval 200 --pipe_gap 100 --bird_color yellow --pipe_color green --background day --frame_stack 4 --debug False --debug-timelimit-diagnostics False",
103
+ "continuous_tanh_scale": 0.0,
104
+ "debug": false,
105
+ "debug_timelimit_diagnostics": false,
106
+ "decoder_mlp_layers": [],
107
+ "decorrelate_envs_on_one_worker": true,
108
+ "decorrelate_experience_max_seconds": 0,
109
+ "default_niceness": 0,
110
+ "device": "gpu",
111
+ "encoder_conv_architecture": "convnet_atari",
112
+ "encoder_conv_mlp_layers": [
113
+ 512
114
+ ],
115
+ "encoder_mlp_layers": [
116
+ 512,
117
+ 512
118
+ ],
119
+ "env": "latency_flappy",
120
+ "env_frameskip": 1,
121
+ "env_framestack": 1,
122
+ "env_gpu_actions": false,
123
+ "env_gpu_observations": true,
124
+ "episode_metrics_path": "results/checkpoints/flappy_sf_same_latency_l0_seed0/episode_metrics.jsonl",
125
+ "episodes_per_stage": 10,
126
+ "eval_deterministic": true,
127
+ "eval_episodes": 20,
128
+ "eval_max_steps": 3600,
129
+ "eval_queue_latency_values": "0,1,2,3,4,5",
130
+ "eval_raw_reward": false,
131
+ "experiment": "flappy_sf_same_latency_l0_seed0",
132
+ "experiment_summaries_interval": 1,
133
+ "exploration_loss": "entropy",
134
+ "exploration_loss_coeff": 0.003,
135
+ "flush_summaries_interval": 30,
136
+ "force_envs_single_thread": false,
137
+ "frame_stack": 4,
138
+ "gae_lambda": 0.95,
139
+ "gamma": 0.99,
140
+ "git_hash": "8b374967e3be2fdc51a3469ce7955c63e6cfa046",
141
+ "git_repo_name": "git@github.com:ZihanWang314/latency-sensitive-bench.git",
142
+ "gpu_render_batch_size": 64,
143
+ "gpu_render_device": "auto",
144
+ "gpu_render_profile": false,
145
+ "gpu_render_profile_interval": 200,
146
+ "gym_id": "FlappyBird-v0",
147
+ "heartbeat_interval": 20,
148
+ "heartbeat_reporting_interval": 180,
149
+ "help": false,
150
+ "initial_stddev": 1.0,
151
+ "keep_checkpoints": 5,
152
+ "kl_loss_coeff": 0.0,
153
+ "latency_mode": "episode_fixed",
154
+ "latency_sampling_mode": "uniform_all",
155
+ "learning_rate": 0.00025,
156
+ "load_checkpoint_kind": "latest",
157
+ "log_to_file": true,
158
+ "lr_adaptive_max": 0.01,
159
+ "lr_adaptive_min": 1e-06,
160
+ "lr_schedule": "linear_decay",
161
+ "lr_schedule_kl_threshold": 0.008,
162
+ "max_episode_steps": 0,
163
+ "max_grad_norm": 0.5,
164
+ "max_policy_lag": 1000,
165
+ "mode": "train",
166
+ "nonlinearity": "elu",
167
+ "normalize_input": true,
168
+ "normalize_input_keys": null,
169
+ "normalize_obs": true,
170
+ "normalize_returns": true,
171
+ "num_batches_per_epoch": 8,
172
+ "num_batches_to_accumulate": 2,
173
+ "num_envs_per_worker": 1,
174
+ "num_epochs": 8,
175
+ "num_policies": 1,
176
+ "num_workers": 2,
177
+ "obs_mode": "image_only",
178
+ "obs_resize": "84,84",
179
+ "obs_scale": 255.0,
180
+ "obs_subtract_mean": 0.0,
181
+ "optimizer": "adam",
182
+ "output_dir": "outputs/train/flappy/latency_0/seed_0",
183
+ "pbt_mix_policies_in_one_env": true,
184
+ "pbt_mutation_rate": 0.15,
185
+ "pbt_optimize_gamma": false,
186
+ "pbt_period_env_steps": 5000000,
187
+ "pbt_perturb_max": 1.5,
188
+ "pbt_perturb_min": 1.1,
189
+ "pbt_replace_fraction": 0.3,
190
+ "pbt_replace_reward_gap": 0.1,
191
+ "pbt_replace_reward_gap_absolute": 1e-06,
192
+ "pbt_start_mutation": 20000000,
193
+ "pbt_target_objective": "true_objective",
194
+ "pipe_color": "green",
195
+ "pipe_gap": 100,
196
+ "pixel_format": "CHW",
197
+ "policy_init_gain": 1.0,
198
+ "policy_initialization": "orthogonal",
199
+ "policy_workers_per_policy": 1,
200
+ "ppo_clip_ratio": 0.1,
201
+ "ppo_clip_value": 0.2,
202
+ "queue_latency_values": "0",
203
+ "randomize_within_stage": false,
204
+ "recurrence": 1,
205
+ "restart_behavior": "resume",
206
+ "reward_clip": 1000.0,
207
+ "reward_scale": 1.0,
208
+ "rnn_num_layers": 1,
209
+ "rnn_size": 512,
210
+ "rnn_type": "gru",
211
+ "rollout": 256,
212
+ "save_best_after": 100000,
213
+ "save_best_every_sec": 5,
214
+ "save_best_metric": "reward",
215
+ "save_every_sec": 600,
216
+ "save_milestones_sec": -1,
217
+ "score_limit": -1,
218
+ "screen_size": "",
219
+ "seed": 0,
220
+ "serial_mode": false,
221
+ "set_workers_cpu_affinity": true,
222
+ "shuffle_minibatches": false,
223
+ "stats_avg": 100,
224
+ "summaries_use_frameskip": true,
225
+ "train_dir": "results/checkpoints",
226
+ "train_for_env_steps": 15000000,
227
+ "train_for_seconds": 10000000000,
228
+ "use_env_info_cache": false,
229
+ "use_gpu_render": true,
230
+ "use_lidar": false,
231
+ "use_record_episode_statistics": false,
232
+ "use_rnn": false,
233
+ "value_bootstrap": false,
234
+ "value_loss_coeff": 0.5,
235
+ "vtrace_c": 1.0,
236
+ "vtrace_rho": 1.0,
237
+ "wandb_group": "flappy-new(remove training episode limit)",
238
+ "wandb_job_type": "sample_factory",
239
+ "wandb_project": "latency-sensitive-bench",
240
+ "wandb_tags": [],
241
+ "wandb_unique_id": "flappy-bird-l0-s0",
242
+ "wandb_user": null,
243
+ "with_pbt": false,
244
+ "with_vtrace": false,
245
+ "with_wandb": true,
246
+ "within_episode_change_interval": 1,
247
+ "within_episode_change_prob": 1.0,
248
+ "worker_num_splits": 1
249
+ },
250
+ "checkpoint_experiment_dir": "/workspace/latency-sensitive-bench/checkpoints/flappy_sf_same_latency_l0_seed0",
251
+ "checkpoint_path": null,
252
+ "checkpoint_train_step": 29312,
253
+ "config_source": "/workspace/latency-sensitive-bench/checkpoints/flappy_sf_same_latency_l0_seed0/config.json",
254
+ "context_image_count": 4,
255
+ "context_window": 5,
256
+ "env_fps": 30.0,
257
+ "env_name": "flappy",
258
+ "episode_return_field": "episode_raw_return",
259
+ "filter_preset": "clean_v1",
260
+ "filter_report": {
261
+ "accepted_episodes": 200,
262
+ "accepted_train_episodes": 180,
263
+ "accepted_val_episodes": 20,
264
+ "attempted_episodes": 227,
265
+ "preset": "clean_v1",
266
+ "rejected_episodes": 27,
267
+ "train_rows_after_row_filter": 279882,
268
+ "train_rows_before_row_filter": 648000,
269
+ "val_rows_after_row_filter": 72000,
270
+ "val_rows_before_row_filter": 72000
271
+ },
272
+ "frame_stack": 4,
273
+ "image_resolution": [
274
+ 512,
275
+ 288,
276
+ 3
277
+ ],
278
+ "image_source": "env_raw_rgb",
279
+ "latency_override": {
280
+ "add_latency_info": false,
281
+ "fixed_latency_ms": 0.0,
282
+ "type": "fixed"
283
+ },
284
+ "latency_source": "legacy_checkpoint_name",
285
+ "max_episodes": 200,
286
+ "max_steps_per_episode": 3600,
287
+ "num_envs": 2048,
288
+ "obs_fps": 30.0,
289
+ "obs_stride_raw_frames": 1,
290
+ "prompt_source": "latency_bench.data.prompts",
291
+ "prompt_timing_suffix": " Current action latency is {latency_raw_frames} raw frames ({latency_ms:.2f} ms). The environment runs at {env_fps:g} FPS and observations are emitted at {obs_fps:g} FPS. Choose the best next action.",
292
+ "replay_num_envs": 16,
293
+ "reward_field": "raw_reward",
294
+ "reward_semantics": "raw_reward",
295
+ "rows_unit": "decision_step",
296
+ "schema_version": 2,
297
+ "split_seed": 0,
298
+ "total_episodes": 200,
299
+ "total_rows": 351882,
300
+ "train_episodes": 180,
301
+ "train_rows": 279882,
302
+ "val_episodes": 20,
303
+ "val_fraction": 0.1,
304
+ "val_rows": 72000
305
+ }
experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/flappy_fix_latency_1_200ep_context5/filter_report.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "accepted_episodes": 200,
3
+ "accepted_train_episodes": 180,
4
+ "accepted_val_episodes": 20,
5
+ "attempted_episodes": 200,
6
+ "preset": "clean_v1",
7
+ "rejected_episodes": 0,
8
+ "train_rows_after_row_filter": 297483,
9
+ "train_rows_before_row_filter": 648000,
10
+ "val_rows_after_row_filter": 72000,
11
+ "val_rows_before_row_filter": 72000
12
+ }
experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/flappy_fix_latency_1_200ep_context5/metadata.json ADDED
@@ -0,0 +1,305 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "action_labels": [
3
+ "NOOP",
4
+ "FLAP"
5
+ ],
6
+ "action_layout": "flappy_discrete",
7
+ "base_prompt": "You are playing Flappy Bird. Pass through the pipe gaps and stay alive. Choose the action: NOOP, FLAP.",
8
+ "checkpoint_config": {
9
+ "actor_critic_share_weights": true,
10
+ "actor_worker_gpus": [
11
+ 0
12
+ ],
13
+ "adam_beta1": 0.9,
14
+ "adam_beta2": 0.999,
15
+ "adam_eps": 1e-05,
16
+ "adaptive_stddev": true,
17
+ "add_latency_info": false,
18
+ "algo": "APPO",
19
+ "async_rl": true,
20
+ "audio_on": false,
21
+ "background": "day",
22
+ "batch_size": 4096,
23
+ "batched_sampling": true,
24
+ "benchmark": false,
25
+ "bird_color": "yellow",
26
+ "cli_args": {
27
+ "actor_worker_gpus": [
28
+ 0
29
+ ],
30
+ "add_latency_info": false,
31
+ "algo": "APPO",
32
+ "async_rl": true,
33
+ "audio_on": false,
34
+ "background": "day",
35
+ "batch_size": 4096,
36
+ "batched_sampling": true,
37
+ "bird_color": "yellow",
38
+ "debug": false,
39
+ "debug_timelimit_diagnostics": false,
40
+ "device": "gpu",
41
+ "env": "latency_flappy",
42
+ "env_gpu_actions": false,
43
+ "env_gpu_observations": true,
44
+ "episode_metrics_path": "results/checkpoints/flappy_sf_same_latency_l1_seed0/episode_metrics.jsonl",
45
+ "episodes_per_stage": 10,
46
+ "eval_deterministic": true,
47
+ "eval_episodes": 20,
48
+ "eval_max_steps": 3600,
49
+ "eval_queue_latency_values": "0,1,2,3,4,5",
50
+ "experiment": "flappy_sf_same_latency_l1_seed0",
51
+ "experiment_summaries_interval": 1,
52
+ "frame_stack": 4,
53
+ "gae_lambda": 0.95,
54
+ "gamma": 0.99,
55
+ "gpu_render_batch_size": 64,
56
+ "gpu_render_device": "auto",
57
+ "gpu_render_profile": false,
58
+ "gpu_render_profile_interval": 200,
59
+ "gym_id": "FlappyBird-v0",
60
+ "keep_checkpoints": 5,
61
+ "latency_mode": "episode_fixed",
62
+ "latency_sampling_mode": "uniform_all",
63
+ "learning_rate": 0.00025,
64
+ "max_grad_norm": 0.5,
65
+ "mode": "train",
66
+ "normalize_input": true,
67
+ "normalize_obs": true,
68
+ "normalize_returns": true,
69
+ "num_batches_per_epoch": 8,
70
+ "num_envs_per_worker": 1,
71
+ "num_epochs": 8,
72
+ "num_policies": 1,
73
+ "num_workers": 2,
74
+ "obs_mode": "image_only",
75
+ "obs_resize": "84,84",
76
+ "pipe_color": "green",
77
+ "pipe_gap": 100,
78
+ "ppo_clip_ratio": 0.1,
79
+ "ppo_clip_value": 0.2,
80
+ "queue_latency_values": "1",
81
+ "randomize_within_stage": false,
82
+ "recurrence": 1,
83
+ "restart_behavior": "resume",
84
+ "rollout": 256,
85
+ "save_every_sec": 600,
86
+ "seed": 0,
87
+ "stats_avg": 100,
88
+ "train_dir": "results/checkpoints",
89
+ "train_for_env_steps": 15000000,
90
+ "use_gpu_render": true,
91
+ "use_lidar": false,
92
+ "use_rnn": false,
93
+ "value_loss_coeff": 0.5,
94
+ "wandb_group": "flappy-new(remove training episode limit)",
95
+ "wandb_job_type": "sample_factory",
96
+ "wandb_project": "latency-sensitive-bench",
97
+ "with_wandb": true,
98
+ "within_episode_change_interval": 1,
99
+ "within_episode_change_prob": 1.0,
100
+ "worker_num_splits": 1
101
+ },
102
+ "command_line": "--mode train --algo APPO --env latency_flappy --experiment flappy_sf_same_latency_l1_seed0 --train_dir results/checkpoints --restart_behavior resume --device gpu --actor_worker_gpus 0 --env_gpu_observations True --env_gpu_actions False --gpu-render-batch-size 64 --seed 0 --episode_metrics_path results/checkpoints/flappy_sf_same_latency_l1_seed0/episode_metrics.jsonl --train_for_env_steps 15000000 --num_workers 2 --num_envs_per_worker 1 --num_policies 1 --batch_size 4096 --rollout 256 --recurrence 1 --num_epochs 8 --num_batches_per_epoch 8 --worker_num_splits 1 --learning_rate 0.00025 --gamma 0.99 --gae_lambda 0.95 --ppo_clip_ratio 0.1 --ppo_clip_value 0.2 --value_loss_coeff 0.5 --max_grad_norm 0.5 --save_every_sec 600 --keep_checkpoints 5 --stats_avg 100 --experiment_summaries_interval 1 --batched_sampling True --async_rl True --use_rnn False --normalize_returns True --normalize_input True --queue-latency-values 1 --latency-mode episode_fixed --latency-sampling-mode uniform_all --episodes-per-stage 10 --randomize-within-stage False --within-episode-change-interval 1 --within-episode-change-prob 1.0 --add-latency-info False --eval-episodes 20 --eval-queue-latency-values 0,1,2,3,4,5 --eval-max-steps 3600 --eval-deterministic True --with_wandb True --wandb_project latency-sensitive-bench --wandb_group flappy-new(remove training episode limit) --wandb_job_type sample_factory --gym_id FlappyBird-v0 --use_lidar False --normalize_obs True --audio_on False --obs_resize 84,84 --obs_mode image_only --use-gpu-render True --gpu-render-device auto --gpu-render-profile False --gpu-render-profile-interval 200 --pipe_gap 100 --bird_color yellow --pipe_color green --background day --frame_stack 4 --debug False --debug-timelimit-diagnostics False",
103
+ "continuous_tanh_scale": 0.0,
104
+ "debug": false,
105
+ "debug_timelimit_diagnostics": false,
106
+ "decoder_mlp_layers": [],
107
+ "decorrelate_envs_on_one_worker": true,
108
+ "decorrelate_experience_max_seconds": 0,
109
+ "default_niceness": 0,
110
+ "device": "gpu",
111
+ "encoder_conv_architecture": "convnet_atari",
112
+ "encoder_conv_mlp_layers": [
113
+ 512
114
+ ],
115
+ "encoder_mlp_layers": [
116
+ 512,
117
+ 512
118
+ ],
119
+ "env": "latency_flappy",
120
+ "env_frameskip": 1,
121
+ "env_framestack": 1,
122
+ "env_gpu_actions": false,
123
+ "env_gpu_observations": true,
124
+ "episode_metrics_path": "results/checkpoints/flappy_sf_same_latency_l1_seed0/episode_metrics.jsonl",
125
+ "episodes_per_stage": 10,
126
+ "eval_deterministic": true,
127
+ "eval_episodes": 20,
128
+ "eval_max_steps": 3600,
129
+ "eval_queue_latency_values": "0,1,2,3,4,5",
130
+ "eval_raw_reward": false,
131
+ "experiment": "flappy_sf_same_latency_l1_seed0",
132
+ "experiment_summaries_interval": 1,
133
+ "exploration_loss": "entropy",
134
+ "exploration_loss_coeff": 0.003,
135
+ "flush_summaries_interval": 30,
136
+ "force_envs_single_thread": false,
137
+ "frame_stack": 4,
138
+ "gae_lambda": 0.95,
139
+ "gamma": 0.99,
140
+ "git_hash": "8b374967e3be2fdc51a3469ce7955c63e6cfa046",
141
+ "git_repo_name": "git@github.com:ZihanWang314/latency-sensitive-bench.git",
142
+ "gpu_render_batch_size": 64,
143
+ "gpu_render_device": "auto",
144
+ "gpu_render_profile": false,
145
+ "gpu_render_profile_interval": 200,
146
+ "gym_id": "FlappyBird-v0",
147
+ "heartbeat_interval": 20,
148
+ "heartbeat_reporting_interval": 180,
149
+ "help": false,
150
+ "initial_stddev": 1.0,
151
+ "keep_checkpoints": 5,
152
+ "kl_loss_coeff": 0.0,
153
+ "latency_mode": "episode_fixed",
154
+ "latency_sampling_mode": "uniform_all",
155
+ "learning_rate": 0.00025,
156
+ "load_checkpoint_kind": "latest",
157
+ "log_to_file": true,
158
+ "lr_adaptive_max": 0.01,
159
+ "lr_adaptive_min": 1e-06,
160
+ "lr_schedule": "linear_decay",
161
+ "lr_schedule_kl_threshold": 0.008,
162
+ "max_episode_steps": 0,
163
+ "max_grad_norm": 0.5,
164
+ "max_policy_lag": 1000,
165
+ "mode": "train",
166
+ "nonlinearity": "elu",
167
+ "normalize_input": true,
168
+ "normalize_input_keys": null,
169
+ "normalize_obs": true,
170
+ "normalize_returns": true,
171
+ "num_batches_per_epoch": 8,
172
+ "num_batches_to_accumulate": 2,
173
+ "num_envs_per_worker": 1,
174
+ "num_epochs": 8,
175
+ "num_policies": 1,
176
+ "num_workers": 2,
177
+ "obs_mode": "image_only",
178
+ "obs_resize": "84,84",
179
+ "obs_scale": 255.0,
180
+ "obs_subtract_mean": 0.0,
181
+ "optimizer": "adam",
182
+ "output_dir": "outputs/train/flappy/latency_1/seed_0",
183
+ "pbt_mix_policies_in_one_env": true,
184
+ "pbt_mutation_rate": 0.15,
185
+ "pbt_optimize_gamma": false,
186
+ "pbt_period_env_steps": 5000000,
187
+ "pbt_perturb_max": 1.5,
188
+ "pbt_perturb_min": 1.1,
189
+ "pbt_replace_fraction": 0.3,
190
+ "pbt_replace_reward_gap": 0.1,
191
+ "pbt_replace_reward_gap_absolute": 1e-06,
192
+ "pbt_start_mutation": 20000000,
193
+ "pbt_target_objective": "true_objective",
194
+ "pipe_color": "green",
195
+ "pipe_gap": 100,
196
+ "pixel_format": "CHW",
197
+ "policy_init_gain": 1.0,
198
+ "policy_initialization": "orthogonal",
199
+ "policy_workers_per_policy": 1,
200
+ "ppo_clip_ratio": 0.1,
201
+ "ppo_clip_value": 0.2,
202
+ "queue_latency_values": "1",
203
+ "randomize_within_stage": false,
204
+ "recurrence": 1,
205
+ "restart_behavior": "resume",
206
+ "reward_clip": 1000.0,
207
+ "reward_scale": 1.0,
208
+ "rnn_num_layers": 1,
209
+ "rnn_size": 512,
210
+ "rnn_type": "gru",
211
+ "rollout": 256,
212
+ "save_best_after": 100000,
213
+ "save_best_every_sec": 5,
214
+ "save_best_metric": "reward",
215
+ "save_every_sec": 600,
216
+ "save_milestones_sec": -1,
217
+ "score_limit": -1,
218
+ "screen_size": "",
219
+ "seed": 0,
220
+ "serial_mode": false,
221
+ "set_workers_cpu_affinity": true,
222
+ "shuffle_minibatches": false,
223
+ "stats_avg": 100,
224
+ "summaries_use_frameskip": true,
225
+ "train_dir": "results/checkpoints",
226
+ "train_for_env_steps": 15000000,
227
+ "train_for_seconds": 10000000000,
228
+ "use_env_info_cache": false,
229
+ "use_gpu_render": true,
230
+ "use_lidar": false,
231
+ "use_record_episode_statistics": false,
232
+ "use_rnn": false,
233
+ "value_bootstrap": false,
234
+ "value_loss_coeff": 0.5,
235
+ "vtrace_c": 1.0,
236
+ "vtrace_rho": 1.0,
237
+ "wandb_group": "flappy-new(remove training episode limit)",
238
+ "wandb_job_type": "sample_factory",
239
+ "wandb_project": "latency-sensitive-bench",
240
+ "wandb_tags": [],
241
+ "wandb_unique_id": "flappy-bird-l1-s0",
242
+ "wandb_user": null,
243
+ "with_pbt": false,
244
+ "with_vtrace": false,
245
+ "with_wandb": true,
246
+ "within_episode_change_interval": 1,
247
+ "within_episode_change_prob": 1.0,
248
+ "worker_num_splits": 1
249
+ },
250
+ "checkpoint_experiment_dir": "/workspace/latency-sensitive-bench/checkpoints/flappy_sf_same_latency_l1_seed0",
251
+ "checkpoint_path": null,
252
+ "checkpoint_train_step": 29328,
253
+ "config_source": "/workspace/latency-sensitive-bench/checkpoints/flappy_sf_same_latency_l1_seed0/config.json",
254
+ "context_image_count": 4,
255
+ "context_window": 5,
256
+ "env_fps": 30.0,
257
+ "env_name": "flappy",
258
+ "episode_return_field": "episode_raw_return",
259
+ "filter_preset": "clean_v1",
260
+ "filter_report": {
261
+ "accepted_episodes": 200,
262
+ "accepted_train_episodes": 180,
263
+ "accepted_val_episodes": 20,
264
+ "attempted_episodes": 200,
265
+ "preset": "clean_v1",
266
+ "rejected_episodes": 0,
267
+ "train_rows_after_row_filter": 297483,
268
+ "train_rows_before_row_filter": 648000,
269
+ "val_rows_after_row_filter": 72000,
270
+ "val_rows_before_row_filter": 72000
271
+ },
272
+ "frame_stack": 4,
273
+ "image_resolution": [
274
+ 512,
275
+ 288,
276
+ 3
277
+ ],
278
+ "image_source": "env_raw_rgb",
279
+ "latency_override": {
280
+ "add_latency_info": false,
281
+ "fixed_latency_ms": 33.333333333333336,
282
+ "type": "fixed"
283
+ },
284
+ "latency_source": "legacy_checkpoint_name",
285
+ "max_episodes": 200,
286
+ "max_steps_per_episode": 3600,
287
+ "num_envs": 2048,
288
+ "obs_fps": 30.0,
289
+ "obs_stride_raw_frames": 1,
290
+ "prompt_source": "latency_bench.data.prompts",
291
+ "prompt_timing_suffix": " Current action latency is {latency_raw_frames} raw frames ({latency_ms:.2f} ms). The environment runs at {env_fps:g} FPS and observations are emitted at {obs_fps:g} FPS. Choose the best next action.",
292
+ "replay_num_envs": 16,
293
+ "reward_field": "raw_reward",
294
+ "reward_semantics": "raw_reward",
295
+ "rows_unit": "decision_step",
296
+ "schema_version": 2,
297
+ "split_seed": 0,
298
+ "total_episodes": 200,
299
+ "total_rows": 369483,
300
+ "train_episodes": 180,
301
+ "train_rows": 297483,
302
+ "val_episodes": 20,
303
+ "val_fraction": 0.1,
304
+ "val_rows": 72000
305
+ }
experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce/config.full.yaml ADDED
@@ -0,0 +1,261 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ framework:
2
+ name: WanOFT
3
+ world_model:
4
+ base_wm: /workspace/starVLA/playground/Pretrained_models/Wan-AI/Wan2.2-TI2V-5B-Diffusers
5
+ extract_layers:
6
+ - -1
7
+ num_frames: 5
8
+ qwenvl:
9
+ base_vlm: /workspace/starVLA/playground/Pretrained_models/Wan-AI/Wan2.2-TI2V-5B-Diffusers
10
+ attn_implementation: sdpa
11
+ enable_gradient_checkpointing: true
12
+ action_model:
13
+ action_model_type: MLP
14
+ action_dim: 7
15
+ action_hidden_dim: 3072
16
+ action_horizon: 8
17
+ future_action_window_size: 7
18
+ past_action_window_size: 0
19
+ loss_type: current_discrete_ce
20
+ class_weights: null
21
+ future_loss_weight: null
22
+ action_query_source: mean
23
+ action_query_num_heads: 24
24
+ state_dim: 7
25
+ action_env_dim: 2
26
+ datasets:
27
+ vla_data:
28
+ dataset_py: lerobot_datasets
29
+ include_state: true
30
+ data_root_dir: /workspace/starVLA/data/flappy_fix_latency_1_200ep_context5
31
+ data_mix: flappy_train__bridge
32
+ eval_data_mix: flappy_train__bridge__val
33
+ custom_mixtures_path: null
34
+ action_type: discrete
35
+ sequential_step_sampling: false
36
+ eval_sequential_step_sampling: null
37
+ num_workers: 8
38
+ eval_num_workers: 8
39
+ prefetch_factor: 4
40
+ persistent_workers: true
41
+ pin_memory: true
42
+ shuffle: true
43
+ action_balance:
44
+ enabled: false
45
+ strategy: balanced_epoch
46
+ action_key: action_id
47
+ target_flap_fraction: 0.3
48
+ noop_id: 0
49
+ flap_id: 1
50
+ latency_curriculum:
51
+ enabled: false
52
+ strategy: exclusive
53
+ latencies: null
54
+ phase_steps: null
55
+ new_latency_passes: 1.0
56
+ replay_passes: 0.25
57
+ target_total_passes: 2.0
58
+ final_equalization: true
59
+ step_budget_mode: auto
60
+ eval_at_phase_end: false
61
+ save_at_phase_end: false
62
+ computed_plan: null
63
+ effective_batch_size: null
64
+ per_device_batch_size: 4
65
+ load_all_data_for_training: true
66
+ num_obs_frames: 1
67
+ image_mode: single
68
+ stitch_grid:
69
+ - 2
70
+ - 2
71
+ obs_image_size:
72
+ - 224
73
+ - 224
74
+ video_backend: torchvision_av
75
+ observation_indices:
76
+ - -4
77
+ - -3
78
+ - -2
79
+ - -1
80
+ - 0
81
+ language_indices:
82
+ - 0
83
+ state_indices:
84
+ - 0
85
+ action_indices:
86
+ - 0
87
+ - 1
88
+ - 2
89
+ - 3
90
+ - 4
91
+ - 5
92
+ - 6
93
+ - 7
94
+ pack_image_sequence: true
95
+ image_sequence_length: 5
96
+ context_images_column: observation.context_images
97
+ dataset:
98
+ source_hf: ''
99
+ config_name: null
100
+ source_subdir: null
101
+ converted_name: flappy_train__bridge
102
+ single_source_hf: ''
103
+ mixed_source_hf: ''
104
+ single_converted_name: flappy_train__bridge
105
+ mixed_converted_name: flappy_mixed_latency_train
106
+ single_latency_filter: null
107
+ mixed_latency_filter: null
108
+ force_download: false
109
+ setup_force: false
110
+ skip_verification: false
111
+ verify_rows: 200
112
+ max_episodes: null
113
+ episodes_per_latency: null
114
+ latency_filter: null
115
+ debug_subset:
116
+ enabled: false
117
+ max_episodes: 5
118
+ suffix: debug
119
+ base_model:
120
+ repo_id: Wan-AI/Wan2.2-TI2V-5B-Diffusers
121
+ initialization:
122
+ checkpoint_local_dir: playground/Pretrained_models/WM4A-Wan2d2-OFT-LIBERO-4in1
123
+ checkpoint_hf_repo_id: StarVLA/WM4A-Wan2d2-OFT-LIBERO-4in1
124
+ checkpoint_filename: checkpoints/steps_60000_pytorch_model.pt
125
+ trainer:
126
+ max_train_steps: 2000
127
+ num_warmup_steps: 0
128
+ save_interval: 0
129
+ eval_interval: 500
130
+ eval_num_batches: 50
131
+ per_latency_eval_num_batches: null
132
+ eval_action_classification: false
133
+ eval_action_classification_interval: null
134
+ cc_f1_tolerance: 1
135
+ learning_rate:
136
+ base: 2.5e-05
137
+ qwen_vl_interface: 1.0e-05
138
+ action_model: 0.0001
139
+ action_query_proj: 0.0001
140
+ lr_scheduler_type: cosine_with_min_lr
141
+ scheduler_specific_kwargs:
142
+ min_lr: 1.0e-06
143
+ freeze_modules: ''
144
+ freeze_vit: false
145
+ freeze_tied_embedding: false
146
+ freeze_llm_layers: []
147
+ loss_scale:
148
+ vla: 1.0
149
+ vlm: 0.1
150
+ max_grad_norm: 1.0
151
+ weight_decay: 0.0
152
+ logging_frequency: 1
153
+ profile_timing:
154
+ enabled: false
155
+ log_interval: 10
156
+ gradient_clipping: 1.0
157
+ gradient_accumulation_steps: 32
158
+ distributed_backend: none
159
+ is_resume: false
160
+ pretrained_checkpoint: null
161
+ resume_step: 0
162
+ reload_modules: null
163
+ optimizer:
164
+ name: AdamW
165
+ betas:
166
+ - 0.9
167
+ - 0.95
168
+ eps: 1.0e-08
169
+ weight_decay: 1.0e-08
170
+ fused: true
171
+ save_format: pt
172
+ workspace_dir: WORKSPACE_DIR
173
+ run_root_dir: /workspace/starVLA/results/Checkpoints
174
+ seed: 42
175
+ wandb_entity: dongqianyu99-zhejiang-university
176
+ wandb_project: starVLA_rl_games
177
+ auth:
178
+ env_file: null
179
+ hf_token_env: HF_TOKEN
180
+ wandb_api_key_env: WANDB_API_KEY
181
+ paths:
182
+ run_root_dir: results/Checkpoints
183
+ dataset_local_dir: data/flappy_fix_latency_1_200ep_context5
184
+ dataset_cache_dir: null
185
+ base_model_dir: playground/Pretrained_models/Wan-AI/Wan2.2-TI2V-5B-Diffusers
186
+ accelerate_config: starVLA/config/deepseeds/deepspeed_zero2.yaml
187
+ launch:
188
+ use_accelerate: true
189
+ gpus: null
190
+ num_processes: 1
191
+ dry_run: false
192
+ conda:
193
+ enabled: true
194
+ env_name: null
195
+ rl_games:
196
+ model_alias: wan_oft
197
+ env_eval:
198
+ image_size: 224
199
+ frameskip: 1
200
+ seed: 42
201
+ fixed_episode_seeds: true
202
+ latency_seed_stride: 0
203
+ task_seed_stride: 0
204
+ task_description: ''
205
+ action_chunk_execution:
206
+ enabled: false
207
+ chunk_size: null
208
+ enabled: true
209
+ distributed_mode: none
210
+ vectorized:
211
+ enabled: false
212
+ batch_size: 1
213
+ latency:
214
+ prompt_map_path: data/flappy_fix_latency_1_200ep_context5/flappy_train__bridge/latency_prompt_map_eval_0_4.json
215
+ mode: single
216
+ values:
217
+ - 0
218
+ mid_train:
219
+ enabled: true
220
+ interval_steps: 500
221
+ latencies:
222
+ - 1
223
+ num_episodes: 5
224
+ max_steps_per_episode: 3600
225
+ post_train:
226
+ enabled: true
227
+ latencies:
228
+ - 0
229
+ - 1
230
+ - 2
231
+ - 3
232
+ - 4
233
+ num_episodes: 20
234
+ max_steps_per_episode: 3600
235
+ task: flappy
236
+ initialization_mode: bridge
237
+ action_carrier: bridge
238
+ model: wan_oft
239
+ env: flappy
240
+ init: wan_oft_libero
241
+ mode: single
242
+ checkpoint:
243
+ load: none
244
+ hf_repo_id: null
245
+ save_best_model: false
246
+ save_final_model: true
247
+ save_pt_file: true
248
+ save_training_state: false
249
+ local:
250
+ keep_last_n: 1
251
+ sync:
252
+ enabled: false
253
+ repo_id: null
254
+ keep_last_n: 0
255
+ sync_every_n_checkpoints: 1
256
+ resume_policy: local_latest
257
+ run_id: wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce
258
+ output_dir: /workspace/starVLA/results/Checkpoints/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce
259
+ config_yaml: null
260
+ is_debug: false
261
+ version_id: '0.21'
experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce/config.yaml ADDED
@@ -0,0 +1,113 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ checkpoint:
2
+ local:
3
+ keep_last_n: 1
4
+ save_best_model: false
5
+ save_final_model: true
6
+ save_pt_file: true
7
+ save_training_state: false
8
+ sync:
9
+ enabled: false
10
+ keep_last_n: 0
11
+ repo_id: null
12
+ datasets:
13
+ vla_data:
14
+ data_mix: flappy_train__bridge
15
+ dataset_py: lerobot_datasets
16
+ eval_data_mix: flappy_train__bridge__val
17
+ image_sequence_length: 5
18
+ include_state: true
19
+ latency_curriculum:
20
+ enabled: false
21
+ obs_image_size:
22
+ - 224
23
+ - 224
24
+ pack_image_sequence: true
25
+ per_device_batch_size: 4
26
+ framework:
27
+ action_model:
28
+ action_dim: 7
29
+ action_env_dim: 2
30
+ action_hidden_dim: 3072
31
+ action_horizon: 8
32
+ action_model_type: MLP
33
+ action_query_source: mean
34
+ class_weights: null
35
+ loss_type: current_discrete_ce
36
+ state_dim: 7
37
+ name: WanOFT
38
+ qwenvl:
39
+ base_vlm: /workspace/starVLA/playground/Pretrained_models/Wan-AI/Wan2.2-TI2V-5B-Diffusers
40
+ world_model:
41
+ base_wm: /workspace/starVLA/playground/Pretrained_models/Wan-AI/Wan2.2-TI2V-5B-Diffusers
42
+ extract_layers:
43
+ - -1
44
+ num_frames: 5
45
+ output_dir: /workspace/starVLA/results/Checkpoints/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce
46
+ rl_games:
47
+ env_eval:
48
+ action_chunk_execution:
49
+ chunk_size: null
50
+ enabled: false
51
+ enabled: true
52
+ fixed_episode_seeds: true
53
+ frameskip: 1
54
+ image_size: 224
55
+ latency:
56
+ prompt_map_path: data/flappy_fix_latency_1_200ep_context5/flappy_train__bridge/latency_prompt_map_eval_0_4.json
57
+ latency_seed_stride: 0
58
+ mid_train:
59
+ enabled: true
60
+ interval_steps: 500
61
+ latencies:
62
+ - 1
63
+ max_steps_per_episode: 3600
64
+ num_episodes: 5
65
+ seed: 42
66
+ task_description: ''
67
+ task_seed_stride: 0
68
+ vectorized:
69
+ enabled: false
70
+ model_alias: wan_oft
71
+ task: flappy
72
+ run_id: wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce
73
+ run_root_dir: /workspace/starVLA/results/Checkpoints
74
+ seed: 42
75
+ trainer:
76
+ distributed_backend: none
77
+ eval_action_classification: false
78
+ eval_action_classification_interval: null
79
+ eval_interval: 500
80
+ eval_num_batches: 50
81
+ freeze_llm_layers: []
82
+ freeze_modules: ''
83
+ freeze_tied_embedding: false
84
+ freeze_vit: false
85
+ gradient_accumulation_steps: 32
86
+ gradient_clipping: 1.0
87
+ is_resume: false
88
+ learning_rate:
89
+ action_model: 0.0001
90
+ action_query_proj: 0.0001
91
+ base: 2.5e-05
92
+ qwen_vl_interface: 1.0e-05
93
+ logging_frequency: 1
94
+ lr_scheduler_type: cosine_with_min_lr
95
+ max_train_steps: 2000
96
+ num_warmup_steps: 0
97
+ optimizer:
98
+ betas:
99
+ - 0.9
100
+ - 0.95
101
+ eps: 1.0e-08
102
+ fused: true
103
+ weight_decay: 1.0e-08
104
+ per_latency_eval_num_batches: null
105
+ pretrained_checkpoint: null
106
+ profile_timing:
107
+ enabled: false
108
+ reload_modules: null
109
+ save_interval: 0
110
+ scheduler_specific_kwargs:
111
+ min_lr: 1.0e-06
112
+ wandb_entity: dongqianyu99-zhejiang-university
113
+ wandb_project: starVLA_rl_games
experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce/dataset_statistics.json ADDED
@@ -0,0 +1,127 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "new_embodiment": {
3
+ "action": {
4
+ "mean": [
5
+ 0.6979423761367798,
6
+ 0.3020575940608978,
7
+ 0.0,
8
+ 0.0,
9
+ 0.0,
10
+ 0.0,
11
+ 0.0
12
+ ],
13
+ "std": [
14
+ 0.4588429629802704,
15
+ 0.4588429629802704,
16
+ 0.0,
17
+ 0.0,
18
+ 0.0,
19
+ 0.0,
20
+ 0.0
21
+ ],
22
+ "max": [
23
+ 1.0,
24
+ 1.0,
25
+ 0.0,
26
+ 0.0,
27
+ 0.0,
28
+ 0.0,
29
+ 0.0
30
+ ],
31
+ "min": [
32
+ 0.0,
33
+ 0.0,
34
+ 0.0,
35
+ 0.0,
36
+ 0.0,
37
+ 0.0,
38
+ 0.0
39
+ ],
40
+ "q01": [
41
+ 0.0,
42
+ 0.0,
43
+ 0.0,
44
+ 0.0,
45
+ 0.0,
46
+ 0.0,
47
+ 0.0
48
+ ],
49
+ "q99": [
50
+ 1.0,
51
+ 1.0,
52
+ 0.0,
53
+ 0.0,
54
+ 0.0,
55
+ 0.0,
56
+ 0.0
57
+ ],
58
+ "mask": [
59
+ true,
60
+ true,
61
+ true,
62
+ true,
63
+ true,
64
+ true,
65
+ true
66
+ ]
67
+ },
68
+ "state": {
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": 297483,
125
+ "num_trajectories": 180
126
+ }
127
+ }
experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce/dataset_statistics_eval.json ADDED
@@ -0,0 +1,127 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "new_embodiment": {
3
+ "action": {
4
+ "mean": [
5
+ 0.8606944680213928,
6
+ 0.13930556178092957,
7
+ 0.0,
8
+ 0.0,
9
+ 0.0,
10
+ 0.0,
11
+ 0.0
12
+ ],
13
+ "std": [
14
+ 0.3463059663772583,
15
+ 0.3463059663772583,
16
+ 0.0,
17
+ 0.0,
18
+ 0.0,
19
+ 0.0,
20
+ 0.0
21
+ ],
22
+ "max": [
23
+ 1.0,
24
+ 1.0,
25
+ 0.0,
26
+ 0.0,
27
+ 0.0,
28
+ 0.0,
29
+ 0.0
30
+ ],
31
+ "min": [
32
+ 0.0,
33
+ 0.0,
34
+ 0.0,
35
+ 0.0,
36
+ 0.0,
37
+ 0.0,
38
+ 0.0
39
+ ],
40
+ "q01": [
41
+ 0.0,
42
+ 0.0,
43
+ 0.0,
44
+ 0.0,
45
+ 0.0,
46
+ 0.0,
47
+ 0.0
48
+ ],
49
+ "q99": [
50
+ 1.0,
51
+ 1.0,
52
+ 0.0,
53
+ 0.0,
54
+ 0.0,
55
+ 0.0,
56
+ 0.0
57
+ ],
58
+ "mask": [
59
+ true,
60
+ true,
61
+ true,
62
+ true,
63
+ true,
64
+ true,
65
+ true
66
+ ]
67
+ },
68
+ "state": {
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": 72000,
125
+ "num_trajectories": 20
126
+ }
127
+ }
experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce/eval/mid_train/step_1000.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_latency": {
3
+ "flappy/latency_1": {
4
+ "latency": 1,
5
+ "num_episodes": 5,
6
+ "mean_reward": 444.60000000002566,
7
+ "mean_length": 3600.0,
8
+ "std_reward": 0.0,
9
+ "std_length": 0.0,
10
+ "episode_rewards": [
11
+ 444.60000000002566,
12
+ 444.60000000002566,
13
+ 444.60000000002566,
14
+ 444.60000000002566,
15
+ 444.60000000002566
16
+ ],
17
+ "episode_lengths": [
18
+ 3600,
19
+ 3600,
20
+ 3600,
21
+ 3600,
22
+ 3600
23
+ ],
24
+ "decoded_action_hist": {
25
+ "0": 15569,
26
+ "1": 2431
27
+ },
28
+ "fixed_episode_seeds": true,
29
+ "eval_seed": 42,
30
+ "episode_seeds": [
31
+ 42,
32
+ 43,
33
+ 44,
34
+ 45,
35
+ 46
36
+ ],
37
+ "episode_indices": [
38
+ 0,
39
+ 1,
40
+ 2,
41
+ 3,
42
+ 4
43
+ ]
44
+ }
45
+ },
46
+ "aggregate": {
47
+ "stage": "mid_train",
48
+ "step": 1000,
49
+ "task": "flappy",
50
+ "model_alias": "wan_oft",
51
+ "fixed_episode_seeds": true,
52
+ "eval_seed": 42,
53
+ "total_episodes": 5,
54
+ "mean_reward": 444.60000000002566,
55
+ "mean_length": 3600.0,
56
+ "std_reward": 0.0,
57
+ "std_length": 0.0,
58
+ "task_count": 1,
59
+ "macro_mean_reward": 444.60000000002566,
60
+ "macro_mean_length": 3600.0,
61
+ "distributed_eval": false
62
+ }
63
+ }
experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce/eval/mid_train/step_1500.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_latency": {
3
+ "flappy/latency_1": {
4
+ "latency": 1,
5
+ "num_episodes": 5,
6
+ "mean_reward": 444.60000000002566,
7
+ "mean_length": 3600.0,
8
+ "std_reward": 0.0,
9
+ "std_length": 0.0,
10
+ "episode_rewards": [
11
+ 444.60000000002566,
12
+ 444.60000000002566,
13
+ 444.60000000002566,
14
+ 444.60000000002566,
15
+ 444.60000000002566
16
+ ],
17
+ "episode_lengths": [
18
+ 3600,
19
+ 3600,
20
+ 3600,
21
+ 3600,
22
+ 3600
23
+ ],
24
+ "decoded_action_hist": {
25
+ "0": 15535,
26
+ "1": 2465
27
+ },
28
+ "fixed_episode_seeds": true,
29
+ "eval_seed": 42,
30
+ "episode_seeds": [
31
+ 42,
32
+ 43,
33
+ 44,
34
+ 45,
35
+ 46
36
+ ],
37
+ "episode_indices": [
38
+ 0,
39
+ 1,
40
+ 2,
41
+ 3,
42
+ 4
43
+ ]
44
+ }
45
+ },
46
+ "aggregate": {
47
+ "stage": "mid_train",
48
+ "step": 1500,
49
+ "task": "flappy",
50
+ "model_alias": "wan_oft",
51
+ "fixed_episode_seeds": true,
52
+ "eval_seed": 42,
53
+ "total_episodes": 5,
54
+ "mean_reward": 444.60000000002566,
55
+ "mean_length": 3600.0,
56
+ "std_reward": 0.0,
57
+ "std_length": 0.0,
58
+ "task_count": 1,
59
+ "macro_mean_reward": 444.60000000002566,
60
+ "macro_mean_length": 3600.0,
61
+ "distributed_eval": false
62
+ }
63
+ }
experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce/eval/mid_train/step_2000.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_latency": {
3
+ "flappy/latency_1": {
4
+ "latency": 1,
5
+ "num_episodes": 5,
6
+ "mean_reward": 444.60000000002566,
7
+ "mean_length": 3600.0,
8
+ "std_reward": 0.0,
9
+ "std_length": 0.0,
10
+ "episode_rewards": [
11
+ 444.60000000002566,
12
+ 444.60000000002566,
13
+ 444.60000000002566,
14
+ 444.60000000002566,
15
+ 444.60000000002566
16
+ ],
17
+ "episode_lengths": [
18
+ 3600,
19
+ 3600,
20
+ 3600,
21
+ 3600,
22
+ 3600
23
+ ],
24
+ "decoded_action_hist": {
25
+ "0": 15564,
26
+ "1": 2436
27
+ },
28
+ "fixed_episode_seeds": true,
29
+ "eval_seed": 42,
30
+ "episode_seeds": [
31
+ 42,
32
+ 43,
33
+ 44,
34
+ 45,
35
+ 46
36
+ ],
37
+ "episode_indices": [
38
+ 0,
39
+ 1,
40
+ 2,
41
+ 3,
42
+ 4
43
+ ]
44
+ }
45
+ },
46
+ "aggregate": {
47
+ "stage": "mid_train",
48
+ "step": 2000,
49
+ "task": "flappy",
50
+ "model_alias": "wan_oft",
51
+ "fixed_episode_seeds": true,
52
+ "eval_seed": 42,
53
+ "total_episodes": 5,
54
+ "mean_reward": 444.60000000002566,
55
+ "mean_length": 3600.0,
56
+ "std_reward": 0.0,
57
+ "std_length": 0.0,
58
+ "task_count": 1,
59
+ "macro_mean_reward": 444.60000000002566,
60
+ "macro_mean_length": 3600.0,
61
+ "distributed_eval": false
62
+ }
63
+ }
experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce/eval/mid_train/step_500.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_latency": {
3
+ "flappy/latency_1": {
4
+ "latency": 1,
5
+ "num_episodes": 5,
6
+ "mean_reward": 382.18000000001996,
7
+ "mean_length": 3096.8,
8
+ "std_reward": 124.8400000000113,
9
+ "std_length": 1006.4000000000002,
10
+ "episode_rewards": [
11
+ 132.49999999999739,
12
+ 444.60000000002566,
13
+ 444.60000000002566,
14
+ 444.60000000002566,
15
+ 444.60000000002566
16
+ ],
17
+ "episode_lengths": [
18
+ 1084,
19
+ 3600,
20
+ 3600,
21
+ 3600,
22
+ 3600
23
+ ],
24
+ "decoded_action_hist": {
25
+ "0": 13331,
26
+ "1": 2153
27
+ },
28
+ "fixed_episode_seeds": true,
29
+ "eval_seed": 42,
30
+ "episode_seeds": [
31
+ 42,
32
+ 43,
33
+ 44,
34
+ 45,
35
+ 46
36
+ ],
37
+ "episode_indices": [
38
+ 0,
39
+ 1,
40
+ 2,
41
+ 3,
42
+ 4
43
+ ]
44
+ }
45
+ },
46
+ "aggregate": {
47
+ "stage": "mid_train",
48
+ "step": 500,
49
+ "task": "flappy",
50
+ "model_alias": "wan_oft",
51
+ "fixed_episode_seeds": true,
52
+ "eval_seed": 42,
53
+ "total_episodes": 5,
54
+ "mean_reward": 382.18000000001996,
55
+ "mean_length": 3096.8,
56
+ "std_reward": 124.8400000000113,
57
+ "std_length": 1006.4000000000002,
58
+ "task_count": 1,
59
+ "macro_mean_reward": 382.18000000001996,
60
+ "macro_mean_length": 3096.8,
61
+ "distributed_eval": false
62
+ }
63
+ }
experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce/eval/post_train/step_2000.json ADDED
@@ -0,0 +1,531 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "per_latency": {
3
+ "flappy/latency_0": {
4
+ "latency": 0,
5
+ "num_episodes": 20,
6
+ "mean_reward": 72.23999999999896,
7
+ "mean_length": 597.95,
8
+ "std_reward": 63.11439930792145,
9
+ "std_length": 509.34550896223675,
10
+ "episode_rewards": [
11
+ 57.800000000000445,
12
+ 34.80000000000019,
13
+ 20.800000000000036,
14
+ 174.7999999999955,
15
+ 81.29999999999968,
16
+ 169.79999999999572,
17
+ 132.79999999999737,
18
+ 11.29999999999998,
19
+ 34.80000000000019,
20
+ 6.799999999999992,
21
+ 53.300000000000395,
22
+ 34.80000000000019,
23
+ 76.79999999999988,
24
+ 71.80000000000011,
25
+ 20.700000000000035,
26
+ 20.800000000000036,
27
+ 183.7999999999951,
28
+ 207.29999999999404,
29
+ 20.700000000000035,
30
+ 29.800000000000136
31
+ ],
32
+ "episode_lengths": [
33
+ 481,
34
+ 296,
35
+ 183,
36
+ 1426,
37
+ 671,
38
+ 1385,
39
+ 1087,
40
+ 106,
41
+ 296,
42
+ 70,
43
+ 445,
44
+ 296,
45
+ 635,
46
+ 594,
47
+ 182,
48
+ 183,
49
+ 1498,
50
+ 1688,
51
+ 182,
52
+ 255
53
+ ],
54
+ "decoded_action_hist": {
55
+ "1": 903,
56
+ "0": 11056
57
+ },
58
+ "fixed_episode_seeds": true,
59
+ "eval_seed": 42,
60
+ "episode_seeds": [
61
+ 42,
62
+ 43,
63
+ 44,
64
+ 45,
65
+ 46,
66
+ 47,
67
+ 48,
68
+ 49,
69
+ 50,
70
+ 51,
71
+ 52,
72
+ 53,
73
+ 54,
74
+ 55,
75
+ 56,
76
+ 57,
77
+ 58,
78
+ 59,
79
+ 60,
80
+ 61
81
+ ],
82
+ "episode_indices": [
83
+ 0,
84
+ 1,
85
+ 2,
86
+ 3,
87
+ 4,
88
+ 5,
89
+ 6,
90
+ 7,
91
+ 8,
92
+ 9,
93
+ 10,
94
+ 11,
95
+ 12,
96
+ 13,
97
+ 14,
98
+ 15,
99
+ 16,
100
+ 17,
101
+ 18,
102
+ 19
103
+ ]
104
+ },
105
+ "flappy/latency_1": {
106
+ "latency": 1,
107
+ "num_episodes": 20,
108
+ "mean_reward": 444.60000000002566,
109
+ "mean_length": 3600.0,
110
+ "std_reward": 0.0,
111
+ "std_length": 0.0,
112
+ "episode_rewards": [
113
+ 444.60000000002566,
114
+ 444.60000000002566,
115
+ 444.60000000002566,
116
+ 444.60000000002566,
117
+ 444.60000000002566,
118
+ 444.60000000002566,
119
+ 444.60000000002566,
120
+ 444.60000000002566,
121
+ 444.60000000002566,
122
+ 444.60000000002566,
123
+ 444.60000000002566,
124
+ 444.60000000002566,
125
+ 444.60000000002566,
126
+ 444.60000000002566,
127
+ 444.60000000002566,
128
+ 444.60000000002566,
129
+ 444.60000000002566,
130
+ 444.60000000002566,
131
+ 444.60000000002566,
132
+ 444.60000000002566
133
+ ],
134
+ "episode_lengths": [
135
+ 3600,
136
+ 3600,
137
+ 3600,
138
+ 3600,
139
+ 3600,
140
+ 3600,
141
+ 3600,
142
+ 3600,
143
+ 3600,
144
+ 3600,
145
+ 3600,
146
+ 3600,
147
+ 3600,
148
+ 3600,
149
+ 3600,
150
+ 3600,
151
+ 3600,
152
+ 3600,
153
+ 3600,
154
+ 3600
155
+ ],
156
+ "decoded_action_hist": {
157
+ "0": 62110,
158
+ "1": 9890
159
+ },
160
+ "fixed_episode_seeds": true,
161
+ "eval_seed": 42,
162
+ "episode_seeds": [
163
+ 42,
164
+ 43,
165
+ 44,
166
+ 45,
167
+ 46,
168
+ 47,
169
+ 48,
170
+ 49,
171
+ 50,
172
+ 51,
173
+ 52,
174
+ 53,
175
+ 54,
176
+ 55,
177
+ 56,
178
+ 57,
179
+ 58,
180
+ 59,
181
+ 60,
182
+ 61
183
+ ],
184
+ "episode_indices": [
185
+ 0,
186
+ 1,
187
+ 2,
188
+ 3,
189
+ 4,
190
+ 5,
191
+ 6,
192
+ 7,
193
+ 8,
194
+ 9,
195
+ 10,
196
+ 11,
197
+ 12,
198
+ 13,
199
+ 14,
200
+ 15,
201
+ 16,
202
+ 17,
203
+ 18,
204
+ 19
205
+ ]
206
+ },
207
+ "flappy/latency_2": {
208
+ "latency": 2,
209
+ "num_episodes": 20,
210
+ "mean_reward": 6.615000000000006,
211
+ "mean_length": 71.75,
212
+ "std_reward": 7.366700414704039,
213
+ "std_length": 59.72342505248673,
214
+ "episode_rewards": [
215
+ 3.8999999999999986,
216
+ 4.099999999999998,
217
+ 4.1999999999999975,
218
+ 37.20000000000022,
219
+ 4.099999999999998,
220
+ 3.8999999999999986,
221
+ 5.999999999999995,
222
+ 3.8999999999999986,
223
+ 4.299999999999997,
224
+ 3.9999999999999982,
225
+ 13.299999999999972,
226
+ 4.299999999999997,
227
+ 3.8999999999999986,
228
+ 8.999999999999984,
229
+ 3.8999999999999986,
230
+ 3.8999999999999986,
231
+ 4.099999999999998,
232
+ 4.299999999999997,
233
+ 5.999999999999995,
234
+ 3.9999999999999982
235
+ ],
236
+ "episode_lengths": [
237
+ 50,
238
+ 52,
239
+ 53,
240
+ 320,
241
+ 52,
242
+ 50,
243
+ 62,
244
+ 50,
245
+ 54,
246
+ 51,
247
+ 126,
248
+ 54,
249
+ 50,
250
+ 92,
251
+ 50,
252
+ 50,
253
+ 52,
254
+ 54,
255
+ 62,
256
+ 51
257
+ ],
258
+ "decoded_action_hist": {
259
+ "0": 1186,
260
+ "1": 249
261
+ },
262
+ "fixed_episode_seeds": true,
263
+ "eval_seed": 42,
264
+ "episode_seeds": [
265
+ 42,
266
+ 43,
267
+ 44,
268
+ 45,
269
+ 46,
270
+ 47,
271
+ 48,
272
+ 49,
273
+ 50,
274
+ 51,
275
+ 52,
276
+ 53,
277
+ 54,
278
+ 55,
279
+ 56,
280
+ 57,
281
+ 58,
282
+ 59,
283
+ 60,
284
+ 61
285
+ ],
286
+ "episode_indices": [
287
+ 0,
288
+ 1,
289
+ 2,
290
+ 3,
291
+ 4,
292
+ 5,
293
+ 6,
294
+ 7,
295
+ 8,
296
+ 9,
297
+ 10,
298
+ 11,
299
+ 12,
300
+ 13,
301
+ 14,
302
+ 15,
303
+ 16,
304
+ 17,
305
+ 18,
306
+ 19
307
+ ]
308
+ },
309
+ "flappy/latency_3": {
310
+ "latency": 3,
311
+ "num_episodes": 20,
312
+ "mean_reward": 4.689999999999996,
313
+ "mean_length": 56.55,
314
+ "std_reward": 0.6579513659838381,
315
+ "std_length": 3.9430318284284747,
316
+ "episode_rewards": [
317
+ 3.9999999999999982,
318
+ 4.699999999999996,
319
+ 4.799999999999995,
320
+ 4.799999999999995,
321
+ 4.4999999999999964,
322
+ 3.9999999999999982,
323
+ 6.099999999999994,
324
+ 3.9999999999999982,
325
+ 4.099999999999998,
326
+ 4.399999999999997,
327
+ 4.799999999999995,
328
+ 4.699999999999996,
329
+ 4.299999999999997,
330
+ 4.799999999999995,
331
+ 6.099999999999994,
332
+ 3.9999999999999982,
333
+ 4.4999999999999964,
334
+ 4.699999999999996,
335
+ 6.099999999999994,
336
+ 4.399999999999997
337
+ ],
338
+ "episode_lengths": [
339
+ 51,
340
+ 58,
341
+ 59,
342
+ 59,
343
+ 56,
344
+ 51,
345
+ 63,
346
+ 51,
347
+ 52,
348
+ 55,
349
+ 59,
350
+ 58,
351
+ 54,
352
+ 59,
353
+ 63,
354
+ 51,
355
+ 56,
356
+ 58,
357
+ 63,
358
+ 55
359
+ ],
360
+ "decoded_action_hist": {
361
+ "0": 934,
362
+ "1": 197
363
+ },
364
+ "fixed_episode_seeds": true,
365
+ "eval_seed": 42,
366
+ "episode_seeds": [
367
+ 42,
368
+ 43,
369
+ 44,
370
+ 45,
371
+ 46,
372
+ 47,
373
+ 48,
374
+ 49,
375
+ 50,
376
+ 51,
377
+ 52,
378
+ 53,
379
+ 54,
380
+ 55,
381
+ 56,
382
+ 57,
383
+ 58,
384
+ 59,
385
+ 60,
386
+ 61
387
+ ],
388
+ "episode_indices": [
389
+ 0,
390
+ 1,
391
+ 2,
392
+ 3,
393
+ 4,
394
+ 5,
395
+ 6,
396
+ 7,
397
+ 8,
398
+ 9,
399
+ 10,
400
+ 11,
401
+ 12,
402
+ 13,
403
+ 14,
404
+ 15,
405
+ 16,
406
+ 17,
407
+ 18,
408
+ 19
409
+ ]
410
+ },
411
+ "flappy/latency_4": {
412
+ "latency": 4,
413
+ "num_episodes": 20,
414
+ "mean_reward": 5.044999999999996,
415
+ "mean_length": 60.55,
416
+ "std_reward": 1.1386285610329618,
417
+ "std_length": 8.749142815156237,
418
+ "episode_rewards": [
419
+ 4.799999999999995,
420
+ 4.799999999999995,
421
+ 4.399999999999997,
422
+ 4.599999999999996,
423
+ 4.799999999999995,
424
+ 4.999999999999995,
425
+ 8.399999999999986,
426
+ 4.999999999999995,
427
+ 4.299999999999997,
428
+ 4.699999999999996,
429
+ 4.599999999999996,
430
+ 4.299999999999997,
431
+ 4.599999999999996,
432
+ 4.599999999999996,
433
+ 4.799999999999995,
434
+ 4.999999999999995,
435
+ 4.799999999999995,
436
+ 4.299999999999997,
437
+ 8.399999999999986,
438
+ 4.699999999999996
439
+ ],
440
+ "episode_lengths": [
441
+ 59,
442
+ 59,
443
+ 55,
444
+ 57,
445
+ 59,
446
+ 61,
447
+ 86,
448
+ 61,
449
+ 54,
450
+ 58,
451
+ 57,
452
+ 54,
453
+ 57,
454
+ 57,
455
+ 59,
456
+ 61,
457
+ 59,
458
+ 54,
459
+ 86,
460
+ 58
461
+ ],
462
+ "decoded_action_hist": {
463
+ "0": 960,
464
+ "1": 251
465
+ },
466
+ "fixed_episode_seeds": true,
467
+ "eval_seed": 42,
468
+ "episode_seeds": [
469
+ 42,
470
+ 43,
471
+ 44,
472
+ 45,
473
+ 46,
474
+ 47,
475
+ 48,
476
+ 49,
477
+ 50,
478
+ 51,
479
+ 52,
480
+ 53,
481
+ 54,
482
+ 55,
483
+ 56,
484
+ 57,
485
+ 58,
486
+ 59,
487
+ 60,
488
+ 61
489
+ ],
490
+ "episode_indices": [
491
+ 0,
492
+ 1,
493
+ 2,
494
+ 3,
495
+ 4,
496
+ 5,
497
+ 6,
498
+ 7,
499
+ 8,
500
+ 9,
501
+ 10,
502
+ 11,
503
+ 12,
504
+ 13,
505
+ 14,
506
+ 15,
507
+ 16,
508
+ 17,
509
+ 18,
510
+ 19
511
+ ]
512
+ }
513
+ },
514
+ "aggregate": {
515
+ "stage": "post_train",
516
+ "step": 2000,
517
+ "task": "flappy",
518
+ "model_alias": "wan_oft",
519
+ "fixed_episode_seeds": true,
520
+ "eval_seed": 42,
521
+ "total_episodes": 100,
522
+ "mean_reward": 106.63800000000492,
523
+ "mean_length": 877.36,
524
+ "std_reward": 173.2974862945318,
525
+ "std_length": 1395.9831984662278,
526
+ "task_count": 1,
527
+ "macro_mean_reward": 106.6380000000049,
528
+ "macro_mean_length": 877.36,
529
+ "distributed_eval": false
530
+ }
531
+ }
experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce/hydra/.hydra/config.yaml ADDED
@@ -0,0 +1,258 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ framework:
2
+ qwenvl:
3
+ base_vlm: /workspace/starVLA/playground/Pretrained_models/Wan-AI/Wan2.2-TI2V-5B-Diffusers
4
+ attn_implementation: sdpa
5
+ enable_gradient_checkpointing: true
6
+ action_model:
7
+ state_dim: 7
8
+ loss_type: current_discrete_ce
9
+ action_horizon: 8
10
+ future_action_window_size: 7
11
+ past_action_window_size: 0
12
+ action_model_type: MLP
13
+ action_dim: 7
14
+ action_env_dim: 2
15
+ action_query_source: mean
16
+ action_query_num_heads: 24
17
+ name: WanOFT
18
+ world_model:
19
+ base_wm: /workspace/starVLA/playground/Pretrained_models/Wan-AI/Wan2.2-TI2V-5B-Diffusers
20
+ extract_layers:
21
+ - -1
22
+ num_frames: 5
23
+ datasets:
24
+ vla_data:
25
+ dataset_py: lerobot_datasets
26
+ include_state: true
27
+ data_root_dir: /workspace/starVLA/data/flappy_fix_latency_1_200ep_context5
28
+ data_mix: flappy_train__bridge
29
+ eval_data_mix: flappy_train__bridge__val
30
+ custom_mixtures_path: null
31
+ action_type: discrete
32
+ sequential_step_sampling: false
33
+ eval_sequential_step_sampling: null
34
+ num_workers: 8
35
+ eval_num_workers: 8
36
+ prefetch_factor: 4
37
+ persistent_workers: true
38
+ pin_memory: true
39
+ shuffle: true
40
+ action_balance:
41
+ enabled: false
42
+ strategy: balanced_epoch
43
+ action_key: action_id
44
+ target_flap_fraction: 0.3
45
+ noop_id: 0
46
+ flap_id: 1
47
+ latency_curriculum:
48
+ enabled: false
49
+ strategy: exclusive
50
+ latencies: null
51
+ phase_steps: null
52
+ new_latency_passes: 1.0
53
+ replay_passes: 0.25
54
+ target_total_passes: 2.0
55
+ final_equalization: true
56
+ step_budget_mode: auto
57
+ eval_at_phase_end: false
58
+ save_at_phase_end: false
59
+ computed_plan: null
60
+ effective_batch_size: null
61
+ per_device_batch_size: 4
62
+ load_all_data_for_training: true
63
+ num_obs_frames: 1
64
+ image_mode: single
65
+ stitch_grid:
66
+ - 2
67
+ - 2
68
+ obs_image_size:
69
+ - 224
70
+ - 224
71
+ video_backend: torchvision_av
72
+ observation_indices:
73
+ - -4
74
+ - -3
75
+ - -2
76
+ - -1
77
+ - 0
78
+ language_indices:
79
+ - 0
80
+ state_indices:
81
+ - 0
82
+ action_indices:
83
+ - 0
84
+ - 1
85
+ - 2
86
+ - 3
87
+ - 4
88
+ - 5
89
+ - 6
90
+ - 7
91
+ pack_image_sequence: true
92
+ image_sequence_length: 5
93
+ context_images_column: observation.context_images
94
+ dataset:
95
+ source_hf: ''
96
+ config_name: null
97
+ source_subdir: null
98
+ converted_name: flappy_train__bridge
99
+ single_source_hf: ''
100
+ mixed_source_hf: ''
101
+ single_converted_name: flappy_train__bridge
102
+ mixed_converted_name: flappy_mixed_latency_train
103
+ single_latency_filter: null
104
+ mixed_latency_filter: null
105
+ force_download: false
106
+ setup_force: false
107
+ skip_verification: false
108
+ verify_rows: 200
109
+ max_episodes: null
110
+ episodes_per_latency: null
111
+ latency_filter: null
112
+ debug_subset:
113
+ enabled: false
114
+ max_episodes: 5
115
+ suffix: debug
116
+ base_model:
117
+ repo_id: Wan-AI/Wan2.2-TI2V-5B-Diffusers
118
+ initialization:
119
+ checkpoint_local_dir: playground/Pretrained_models/WM4A-Wan2d2-OFT-LIBERO-4in1
120
+ checkpoint_hf_repo_id: StarVLA/WM4A-Wan2d2-OFT-LIBERO-4in1
121
+ checkpoint_filename: checkpoints/steps_60000_pytorch_model.pt
122
+ trainer:
123
+ max_train_steps: 2000
124
+ num_warmup_steps: 0
125
+ save_interval: 0
126
+ eval_interval: 500
127
+ eval_num_batches: 50
128
+ per_latency_eval_num_batches: null
129
+ eval_action_classification: false
130
+ eval_action_classification_interval: null
131
+ cc_f1_tolerance: 1
132
+ learning_rate:
133
+ base: 2.5e-05
134
+ qwen_vl_interface: 1.0e-05
135
+ action_model: 0.0001
136
+ action_query_proj: 0.0001
137
+ lr_scheduler_type: cosine_with_min_lr
138
+ scheduler_specific_kwargs:
139
+ min_lr: 1.0e-06
140
+ freeze_modules: ''
141
+ freeze_vit: false
142
+ freeze_tied_embedding: false
143
+ freeze_llm_layers: []
144
+ loss_scale:
145
+ vla: 1.0
146
+ vlm: 0.1
147
+ max_grad_norm: 1.0
148
+ weight_decay: 0.0
149
+ logging_frequency: 1
150
+ profile_timing:
151
+ enabled: false
152
+ log_interval: 10
153
+ gradient_clipping: 1.0
154
+ gradient_accumulation_steps: 32
155
+ distributed_backend: none
156
+ is_resume: false
157
+ pretrained_checkpoint: /workspace/starVLA/playground/Pretrained_models/WM4A-Wan2d2-OFT-LIBERO-4in1/checkpoints/steps_60000_pytorch_model.pt
158
+ resume_step: 0
159
+ reload_modules: null
160
+ optimizer:
161
+ name: AdamW
162
+ betas:
163
+ - 0.9
164
+ - 0.95
165
+ eps: 1.0e-08
166
+ weight_decay: 1.0e-08
167
+ fused: true
168
+ save_format: pt
169
+ workspace_dir: WORKSPACE_DIR
170
+ run_root_dir: /workspace/starVLA/results/Checkpoints
171
+ seed: 42
172
+ wandb_entity: dongqianyu99-zhejiang-university
173
+ wandb_project: starVLA_rl_games
174
+ auth:
175
+ env_file: null
176
+ hf_token_env: HF_TOKEN
177
+ wandb_api_key_env: WANDB_API_KEY
178
+ paths:
179
+ run_root_dir: results/Checkpoints
180
+ dataset_local_dir: data/flappy_fix_latency_1_200ep_context5
181
+ dataset_cache_dir: null
182
+ base_model_dir: playground/Pretrained_models/Wan-AI/Wan2.2-TI2V-5B-Diffusers
183
+ accelerate_config: starVLA/config/deepseeds/deepspeed_zero2.yaml
184
+ launch:
185
+ use_accelerate: true
186
+ gpus: null
187
+ num_processes: 1
188
+ dry_run: false
189
+ conda:
190
+ enabled: true
191
+ env_name: null
192
+ rl_games:
193
+ model_alias: wan_oft
194
+ env_eval:
195
+ image_size: 224
196
+ frameskip: 1
197
+ seed: 42
198
+ fixed_episode_seeds: true
199
+ latency_seed_stride: 0
200
+ task_seed_stride: 0
201
+ task_description: ''
202
+ action_chunk_execution:
203
+ enabled: false
204
+ chunk_size: null
205
+ enabled: true
206
+ distributed_mode: none
207
+ vectorized:
208
+ enabled: false
209
+ batch_size: 1
210
+ latency:
211
+ prompt_map_path: data/flappy_fix_latency_1_200ep_context5/flappy_train__bridge/latency_prompt_map_eval_0_4.json
212
+ mode: single
213
+ values:
214
+ - 0
215
+ mid_train:
216
+ enabled: true
217
+ interval_steps: 500
218
+ latencies:
219
+ - 1
220
+ num_episodes: 5
221
+ max_steps_per_episode: 3600
222
+ post_train:
223
+ enabled: true
224
+ latencies:
225
+ - 0
226
+ - 1
227
+ - 2
228
+ - 3
229
+ - 4
230
+ num_episodes: 20
231
+ max_steps_per_episode: 3600
232
+ task: flappy
233
+ initialization_mode: bridge
234
+ action_carrier: bridge
235
+ model: wan_oft
236
+ env: flappy
237
+ init: wan_oft_libero
238
+ mode: single
239
+ checkpoint:
240
+ load: none
241
+ hf_repo_id: null
242
+ save_best_model: false
243
+ save_final_model: true
244
+ save_pt_file: true
245
+ save_training_state: false
246
+ local:
247
+ keep_last_n: 1
248
+ sync:
249
+ enabled: false
250
+ repo_id: null
251
+ keep_last_n: 0
252
+ sync_every_n_checkpoints: 1
253
+ resume_policy: local_latest
254
+ run_id: wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce
255
+ output_dir: null
256
+ config_yaml: null
257
+ is_debug: false
258
+ version_id: 0.21
experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce/hydra/.hydra/hydra.yaml ADDED
@@ -0,0 +1,356 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ hydra:
2
+ run:
3
+ dir: ${run_root_dir}/${run_id}/hydra
4
+ sweep:
5
+ dir: multirun/${now:%Y-%m-%d}/${now:%H-%M-%S}
6
+ subdir: ${hydra.job.num}
7
+ launcher:
8
+ _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher
9
+ sweeper:
10
+ _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper
11
+ max_batch_size: null
12
+ params: null
13
+ help:
14
+ app_name: ${hydra.job.name}
15
+ header: '${hydra.help.app_name} is powered by Hydra.
16
+
17
+ '
18
+ footer: 'Powered by Hydra (https://hydra.cc)
19
+
20
+ Use --hydra-help to view Hydra specific help
21
+
22
+ '
23
+ template: '${hydra.help.header}
24
+
25
+ == Configuration groups ==
26
+
27
+ Compose your configuration from those groups (group=option)
28
+
29
+
30
+ $APP_CONFIG_GROUPS
31
+
32
+
33
+ == Config ==
34
+
35
+ Override anything in the config (foo.bar=value)
36
+
37
+
38
+ $CONFIG
39
+
40
+
41
+ ${hydra.help.footer}
42
+
43
+ '
44
+ hydra_help:
45
+ template: 'Hydra (${hydra.runtime.version})
46
+
47
+ See https://hydra.cc for more info.
48
+
49
+
50
+ == Flags ==
51
+
52
+ $FLAGS_HELP
53
+
54
+
55
+ == Configuration groups ==
56
+
57
+ Compose your configuration from those groups (For example, append hydra/job_logging=disabled
58
+ to command line)
59
+
60
+
61
+ $HYDRA_CONFIG_GROUPS
62
+
63
+
64
+ Use ''--cfg hydra'' to Show the Hydra config.
65
+
66
+ '
67
+ hydra_help: ???
68
+ hydra_logging:
69
+ version: 1
70
+ formatters:
71
+ simple:
72
+ format: '[%(asctime)s][HYDRA] %(message)s'
73
+ handlers:
74
+ console:
75
+ class: logging.StreamHandler
76
+ formatter: simple
77
+ stream: ext://sys.stdout
78
+ root:
79
+ level: INFO
80
+ handlers:
81
+ - console
82
+ loggers:
83
+ logging_example:
84
+ level: DEBUG
85
+ disable_existing_loggers: false
86
+ job_logging:
87
+ version: 1
88
+ formatters:
89
+ simple:
90
+ format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s'
91
+ handlers:
92
+ console:
93
+ class: logging.StreamHandler
94
+ formatter: simple
95
+ stream: ext://sys.stdout
96
+ file:
97
+ class: logging.FileHandler
98
+ formatter: simple
99
+ filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log
100
+ root:
101
+ level: INFO
102
+ handlers:
103
+ - console
104
+ - file
105
+ disable_existing_loggers: false
106
+ env: {}
107
+ mode: RUN
108
+ searchpath: []
109
+ callbacks: {}
110
+ output_subdir: .hydra
111
+ overrides:
112
+ hydra:
113
+ - hydra.mode=RUN
114
+ task:
115
+ - model=wan_oft
116
+ - env=flappy
117
+ - init=wan_oft_libero
118
+ - mode=single
119
+ - ++framework.qwenvl.base_vlm=playground/Pretrained_models/Wan-AI/Wan2.2-TI2V-5B-Diffusers
120
+ - ++framework.qwenvl.attn_implementation=sdpa
121
+ - ++framework.qwenvl.enable_gradient_checkpointing=true
122
+ - ++framework.action_model.state_dim=7
123
+ - ++framework.action_model.loss_type=current_discrete_ce
124
+ - ++framework.action_model.action_horizon=8
125
+ - ++framework.action_model.future_action_window_size=7
126
+ - ++framework.action_model.past_action_window_size=0
127
+ - ++framework.action_model.action_model_type=MLP
128
+ - ++framework.action_model.action_dim=7
129
+ - ++framework.action_model.action_env_dim=2
130
+ - ++framework.action_model.action_query_source=mean
131
+ - ++framework.action_model.action_query_num_heads=24
132
+ - ++framework.name=WanOFT
133
+ - ++framework.world_model.base_wm=playground/Pretrained_models/Wan-AI/Wan2.2-TI2V-5B-Diffusers
134
+ - ++framework.world_model.extract_layers=[-1]
135
+ - ++framework.world_model.num_frames=5
136
+ - ++datasets.vla_data.dataset_py=lerobot_datasets
137
+ - ++datasets.vla_data.include_state=true
138
+ - ++datasets.vla_data.data_root_dir=playground/Datasets/rl_games
139
+ - ++datasets.vla_data.data_mix=flappy_train__bridge
140
+ - ++datasets.vla_data.eval_data_mix=flappy_train__bridge__val
141
+ - ++datasets.vla_data.custom_mixtures_path=null
142
+ - ++datasets.vla_data.action_type=discrete
143
+ - ++datasets.vla_data.sequential_step_sampling=false
144
+ - ++datasets.vla_data.eval_sequential_step_sampling=null
145
+ - ++datasets.vla_data.num_workers=8
146
+ - ++datasets.vla_data.eval_num_workers=8
147
+ - ++datasets.vla_data.prefetch_factor=4
148
+ - ++datasets.vla_data.persistent_workers=true
149
+ - ++datasets.vla_data.pin_memory=true
150
+ - ++datasets.vla_data.shuffle=true
151
+ - ++datasets.vla_data.action_balance.enabled=false
152
+ - ++datasets.vla_data.action_balance.strategy=balanced_epoch
153
+ - ++datasets.vla_data.action_balance.action_key=action_id
154
+ - ++datasets.vla_data.action_balance.target_flap_fraction=0.3
155
+ - ++datasets.vla_data.action_balance.noop_id=0
156
+ - ++datasets.vla_data.action_balance.flap_id=1
157
+ - ++datasets.vla_data.latency_curriculum.enabled=false
158
+ - ++datasets.vla_data.latency_curriculum.strategy=exclusive
159
+ - ++datasets.vla_data.latency_curriculum.latencies=null
160
+ - ++datasets.vla_data.latency_curriculum.phase_steps=null
161
+ - ++datasets.vla_data.latency_curriculum.new_latency_passes=1.0
162
+ - ++datasets.vla_data.latency_curriculum.replay_passes=0.25
163
+ - ++datasets.vla_data.latency_curriculum.target_total_passes=2.0
164
+ - ++datasets.vla_data.latency_curriculum.final_equalization=true
165
+ - ++datasets.vla_data.latency_curriculum.step_budget_mode=auto
166
+ - ++datasets.vla_data.latency_curriculum.eval_at_phase_end=false
167
+ - ++datasets.vla_data.latency_curriculum.save_at_phase_end=false
168
+ - ++datasets.vla_data.latency_curriculum.computed_plan=null
169
+ - ++datasets.vla_data.latency_curriculum.effective_batch_size=null
170
+ - ++datasets.vla_data.per_device_batch_size=4
171
+ - ++datasets.vla_data.load_all_data_for_training=true
172
+ - ++datasets.vla_data.num_obs_frames=1
173
+ - ++datasets.vla_data.image_mode=single
174
+ - ++datasets.vla_data.stitch_grid=[2,2]
175
+ - ++datasets.vla_data.obs_image_size=[224,224]
176
+ - ++datasets.vla_data.video_backend=torchvision_av
177
+ - ++datasets.vla_data.observation_indices=[-4,-3,-2,-1,0]
178
+ - ++datasets.vla_data.language_indices=[0]
179
+ - ++datasets.vla_data.state_indices=[0]
180
+ - ++datasets.vla_data.action_indices=[0,1,2,3,4,5,6,7]
181
+ - ++datasets.vla_data.pack_image_sequence=true
182
+ - ++datasets.vla_data.image_sequence_length=5
183
+ - ++datasets.vla_data.context_images_column=observation.context_images
184
+ - ++dataset.source_hf=
185
+ - ++dataset.config_name=null
186
+ - ++dataset.source_subdir=null
187
+ - ++dataset.converted_name=flappy_train__bridge
188
+ - ++dataset.single_source_hf=
189
+ - ++dataset.mixed_source_hf=
190
+ - ++dataset.single_converted_name=flappy_train__bridge
191
+ - ++dataset.mixed_converted_name=flappy_mixed_latency_train
192
+ - ++dataset.single_latency_filter=null
193
+ - ++dataset.mixed_latency_filter=null
194
+ - ++dataset.force_download=false
195
+ - ++dataset.setup_force=false
196
+ - ++dataset.skip_verification=false
197
+ - ++dataset.verify_rows=200
198
+ - ++dataset.max_episodes=null
199
+ - ++dataset.episodes_per_latency=null
200
+ - ++dataset.latency_filter=null
201
+ - ++dataset.debug_subset.enabled=false
202
+ - ++dataset.debug_subset.max_episodes=5
203
+ - ++dataset.debug_subset.suffix=debug
204
+ - ++base_model.repo_id=Wan-AI/Wan2.2-TI2V-5B-Diffusers
205
+ - ++initialization.checkpoint_local_dir=playground/Pretrained_models/WM4A-Wan2d2-OFT-LIBERO-4in1
206
+ - ++initialization.checkpoint_hf_repo_id=StarVLA/WM4A-Wan2d2-OFT-LIBERO-4in1
207
+ - ++initialization.checkpoint_filename=checkpoints/steps_60000_pytorch_model.pt
208
+ - ++trainer.max_train_steps=2000
209
+ - ++trainer.num_warmup_steps=0
210
+ - ++trainer.save_interval=0
211
+ - ++trainer.eval_interval=500
212
+ - ++trainer.eval_num_batches=50
213
+ - ++trainer.per_latency_eval_num_batches=null
214
+ - ++trainer.eval_action_classification=false
215
+ - ++trainer.eval_action_classification_interval=null
216
+ - ++trainer.cc_f1_tolerance=1
217
+ - ++trainer.learning_rate.base=2.5e-05
218
+ - ++trainer.learning_rate.qwen_vl_interface=1e-05
219
+ - ++trainer.learning_rate.action_model=0.0001
220
+ - ++trainer.learning_rate.action_query_proj=0.0001
221
+ - ++trainer.lr_scheduler_type=cosine_with_min_lr
222
+ - ++trainer.scheduler_specific_kwargs.min_lr=1e-06
223
+ - ++trainer.freeze_modules=
224
+ - ++trainer.freeze_vit=false
225
+ - ++trainer.freeze_tied_embedding=false
226
+ - ++trainer.freeze_llm_layers=[]
227
+ - ++trainer.loss_scale.vla=1.0
228
+ - ++trainer.loss_scale.vlm=0.1
229
+ - ++trainer.max_grad_norm=1.0
230
+ - ++trainer.weight_decay=0.0
231
+ - ++trainer.logging_frequency=1
232
+ - ++trainer.profile_timing.enabled=false
233
+ - ++trainer.profile_timing.log_interval=10
234
+ - ++trainer.gradient_clipping=1.0
235
+ - ++trainer.gradient_accumulation_steps=32
236
+ - ++trainer.distributed_backend=none
237
+ - ++trainer.is_resume=false
238
+ - ++trainer.pretrained_checkpoint=null
239
+ - ++trainer.resume_step=0
240
+ - ++trainer.reload_modules=null
241
+ - ++trainer.optimizer.name=AdamW
242
+ - ++trainer.optimizer.betas=[0.9,0.95]
243
+ - ++trainer.optimizer.eps=1e-08
244
+ - ++trainer.optimizer.weight_decay=1e-08
245
+ - ++trainer.optimizer.fused=true
246
+ - ++trainer.save_format=pt
247
+ - ++workspace_dir=WORKSPACE_DIR
248
+ - ++run_root_dir=results/Checkpoints
249
+ - ++seed=42
250
+ - ++auth.env_file=null
251
+ - ++auth.hf_token_env=HF_TOKEN
252
+ - ++auth.wandb_api_key_env=WANDB_API_KEY
253
+ - ++paths.run_root_dir=results/Checkpoints
254
+ - ++paths.dataset_local_dir=data/flappy_fix_latency_1_200ep_context5
255
+ - ++paths.dataset_cache_dir=null
256
+ - ++paths.base_model_dir=playground/Pretrained_models/Wan-AI/Wan2.2-TI2V-5B-Diffusers
257
+ - ++paths.accelerate_config=starVLA/config/deepseeds/deepspeed_zero2.yaml
258
+ - ++rl_games.model_alias=wan_oft
259
+ - ++rl_games.env_eval.image_size=224
260
+ - ++rl_games.env_eval.frameskip=1
261
+ - ++rl_games.env_eval.seed=42
262
+ - ++rl_games.env_eval.fixed_episode_seeds=true
263
+ - ++rl_games.env_eval.latency_seed_stride=0
264
+ - ++rl_games.env_eval.task_seed_stride=0
265
+ - ++rl_games.env_eval.task_description=
266
+ - ++rl_games.env_eval.action_chunk_execution.enabled=false
267
+ - ++rl_games.env_eval.action_chunk_execution.chunk_size=null
268
+ - ++rl_games.env_eval.enabled=true
269
+ - ++rl_games.env_eval.distributed_mode=none
270
+ - ++rl_games.env_eval.vectorized.enabled=false
271
+ - ++rl_games.env_eval.vectorized.batch_size=1
272
+ - ++rl_games.env_eval.latency.prompt_map_path=data/flappy_fix_latency_1_200ep_context5/flappy_train__bridge/latency_prompt_map_eval_0_4.json
273
+ - ++rl_games.env_eval.latency.mode=single
274
+ - ++rl_games.env_eval.latency.values=[0]
275
+ - ++rl_games.env_eval.mid_train.enabled=true
276
+ - ++rl_games.env_eval.mid_train.interval_steps=500
277
+ - ++rl_games.env_eval.mid_train.latencies=[1]
278
+ - ++rl_games.env_eval.mid_train.num_episodes=5
279
+ - ++rl_games.env_eval.mid_train.max_steps_per_episode=3600
280
+ - ++rl_games.env_eval.post_train.enabled=true
281
+ - ++rl_games.env_eval.post_train.latencies=[0,1,2,3,4]
282
+ - ++rl_games.env_eval.post_train.num_episodes=20
283
+ - ++rl_games.env_eval.post_train.max_steps_per_episode=3600
284
+ - ++rl_games.task=flappy
285
+ - ++rl_games.initialization_mode=bridge
286
+ - ++rl_games.action_carrier=bridge
287
+ - ++checkpoint.load=none
288
+ - ++checkpoint.hf_repo_id=null
289
+ - ++checkpoint.save_best_model=false
290
+ - ++checkpoint.save_final_model=true
291
+ - ++checkpoint.save_pt_file=true
292
+ - ++checkpoint.save_training_state=false
293
+ - ++checkpoint.local.keep_last_n=1
294
+ - ++checkpoint.sync.enabled=false
295
+ - ++checkpoint.sync.repo_id=null
296
+ - ++checkpoint.sync.keep_last_n=0
297
+ - ++checkpoint.sync.sync_every_n_checkpoints=1
298
+ - ++checkpoint.sync.resume_policy=local_latest
299
+ - ++run_id=wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce
300
+ - ++output_dir=null
301
+ - ++config_yaml=null
302
+ - ++is_debug=false
303
+ - ++version_id=0.21
304
+ - ++run_root_dir=/workspace/starVLA/results/Checkpoints
305
+ - ++trainer.is_resume=false
306
+ - ++trainer.pretrained_checkpoint=/workspace/starVLA/playground/Pretrained_models/WM4A-Wan2d2-OFT-LIBERO-4in1/checkpoints/steps_60000_pytorch_model.pt
307
+ - ++trainer.resume_step=0
308
+ - ++datasets.vla_data.data_root_dir=/workspace/starVLA/data/flappy_fix_latency_1_200ep_context5
309
+ - ++framework.qwenvl.base_vlm=/workspace/starVLA/playground/Pretrained_models/Wan-AI/Wan2.2-TI2V-5B-Diffusers
310
+ - ++framework.world_model.base_wm=/workspace/starVLA/playground/Pretrained_models/Wan-AI/Wan2.2-TI2V-5B-Diffusers
311
+ job:
312
+ name: train_starvla_hydra
313
+ chdir: false
314
+ override_dirname: ++auth.env_file=null,++auth.hf_token_env=HF_TOKEN,++auth.wandb_api_key_env=WANDB_API_KEY,++base_model.repo_id=Wan-AI/Wan2.2-TI2V-5B-Diffusers,++checkpoint.hf_repo_id=null,++checkpoint.load=none,++checkpoint.local.keep_last_n=1,++checkpoint.save_best_model=false,++checkpoint.save_final_model=true,++checkpoint.save_pt_file=true,++checkpoint.save_training_state=false,++checkpoint.sync.enabled=false,++checkpoint.sync.keep_last_n=0,++checkpoint.sync.repo_id=null,++checkpoint.sync.resume_policy=local_latest,++checkpoint.sync.sync_every_n_checkpoints=1,++config_yaml=null,++dataset.config_name=null,++dataset.converted_name=flappy_train__bridge,++dataset.debug_subset.enabled=false,++dataset.debug_subset.max_episodes=5,++dataset.debug_subset.suffix=debug,++dataset.episodes_per_latency=null,++dataset.force_download=false,++dataset.latency_filter=null,++dataset.max_episodes=null,++dataset.mixed_converted_name=flappy_mixed_latency_train,++dataset.mixed_latency_filter=null,++dataset.mixed_source_hf=,++dataset.setup_force=false,++dataset.single_converted_name=flappy_train__bridge,++dataset.single_latency_filter=null,++dataset.single_source_hf=,++dataset.skip_verification=false,++dataset.source_hf=,++dataset.source_subdir=null,++dataset.verify_rows=200,++datasets.vla_data.action_balance.action_key=action_id,++datasets.vla_data.action_balance.enabled=false,++datasets.vla_data.action_balance.flap_id=1,++datasets.vla_data.action_balance.noop_id=0,++datasets.vla_data.action_balance.strategy=balanced_epoch,++datasets.vla_data.action_balance.target_flap_fraction=0.3,++datasets.vla_data.action_indices=[0,1,2,3,4,5,6,7],++datasets.vla_data.action_type=discrete,++datasets.vla_data.context_images_column=observation.context_images,++datasets.vla_data.custom_mixtures_path=null,++datasets.vla_data.data_mix=flappy_train__bridge,++datasets.vla_data.data_root_dir=/workspace/starVLA/data/flappy_fix_latency_1_200ep_context5,++datasets.vla_data.data_root_dir=playground/Datasets/rl_games,++datasets.vla_data.dataset_py=lerobot_datasets,++datasets.vla_data.eval_data_mix=flappy_train__bridge__val,++datasets.vla_data.eval_num_workers=8,++datasets.vla_data.eval_sequential_step_sampling=null,++datasets.vla_data.image_mode=single,++datasets.vla_data.image_sequence_length=5,++datasets.vla_data.include_state=true,++datasets.vla_data.language_indices=[0],++datasets.vla_data.latency_curriculum.computed_plan=null,++datasets.vla_data.latency_curriculum.effective_batch_size=null,++datasets.vla_data.latency_curriculum.enabled=false,++datasets.vla_data.latency_curriculum.eval_at_phase_end=false,++datasets.vla_data.latency_curriculum.final_equalization=true,++datasets.vla_data.latency_curriculum.latencies=null,++datasets.vla_data.latency_curriculum.new_latency_passes=1.0,++datasets.vla_data.latency_curriculum.phase_steps=null,++datasets.vla_data.latency_curriculum.replay_passes=0.25,++datasets.vla_data.latency_curriculum.save_at_phase_end=false,++datasets.vla_data.latency_curriculum.step_budget_mode=auto,++datasets.vla_data.latency_curriculum.strategy=exclusive,++datasets.vla_data.latency_curriculum.target_total_passes=2.0,++datasets.vla_data.load_all_data_for_training=true,++datasets.vla_data.num_obs_frames=1,++datasets.vla_data.num_workers=8,++datasets.vla_data.obs_image_size=[224,224],++datasets.vla_data.observation_indices=[-4,-3,-2,-1,0],++datasets.vla_data.pack_image_sequence=true,++datasets.vla_data.per_device_batch_size=4,++datasets.vla_data.persistent_workers=true,++datasets.vla_data.pin_memory=true,++datasets.vla_data.prefetch_factor=4,++datasets.vla_data.sequential_step_sampling=false,++datasets.vla_data.shuffle=true,++datasets.vla_data.state_indices=[0],++datasets.vla_data.stitch_grid=[2,2],++datasets.vla_data.video_backend=torchvision_av,++framework.action_model.action_dim=7,++framework.action_model.action_env_dim=2,++framework.action_model.action_horizon=8,++framework.action_model.action_model_type=MLP,++framework.action_model.action_query_num_heads=24,++framework.action_model.action_query_source=mean,++framework.action_model.future_action_window_size=7,++framework.action_model.loss_type=current_discrete_ce,++framework.action_model.past_action_window_size=0,++framework.action_model.state_dim=7,++framework.name=WanOFT,++framework.qwenvl.attn_implementation=sdpa,++framework.qwenvl.base_vlm=/workspace/starVLA/playground/Pretrained_models/Wan-AI/Wan2.2-TI2V-5B-Diffusers,++framework.qwenvl.base_vlm=playground/Pretrained_models/Wan-AI/Wan2.2-TI2V-5B-Diffusers,++framework.qwenvl.enable_gradient_checkpointing=true,++framework.world_model.base_wm=/workspace/starVLA/playground/Pretrained_models/Wan-AI/Wan2.2-TI2V-5B-Diffusers,++framework.world_model.base_wm=playground/Pretrained_models/Wan-AI/Wan2.2-TI2V-5B-Diffusers,++framework.world_model.extract_layers=[-1],++framework.world_model.num_frames=5,++initialization.checkpoint_filename=checkpoints/steps_60000_pytorch_model.pt,++initialization.checkpoint_hf_repo_id=StarVLA/WM4A-Wan2d2-OFT-LIBERO-4in1,++initialization.checkpoint_local_dir=playground/Pretrained_models/WM4A-Wan2d2-OFT-LIBERO-4in1,++is_debug=false,++output_dir=null,++paths.accelerate_config=starVLA/config/deepseeds/deepspeed_zero2.yaml,++paths.base_model_dir=playground/Pretrained_models/Wan-AI/Wan2.2-TI2V-5B-Diffusers,++paths.dataset_cache_dir=null,++paths.dataset_local_dir=data/flappy_fix_latency_1_200ep_context5,++paths.run_root_dir=results/Checkpoints,++rl_games.action_carrier=bridge,++rl_games.env_eval.action_chunk_execution.chunk_size=null,++rl_games.env_eval.action_chunk_execution.enabled=false,++rl_games.env_eval.distributed_mode=none,++rl_games.env_eval.enabled=true,++rl_games.env_eval.fixed_episode_seeds=true,++rl_games.env_eval.frameskip=1,++rl_games.env_eval.image_size=224,++rl_games.env_eval.latency.mode=single,++rl_games.env_eval.latency.prompt_map_path=data/flappy_fix_latency_1_200ep_context5/flappy_train__bridge/latency_prompt_map_eval_0_4.json,++rl_games.env_eval.latency.values=[0],++rl_games.env_eval.latency_seed_stride=0,++rl_games.env_eval.mid_train.enabled=true,++rl_games.env_eval.mid_train.interval_steps=500,++rl_games.env_eval.mid_train.latencies=[1],++rl_games.env_eval.mid_train.max_steps_per_episode=3600,++rl_games.env_eval.mid_train.num_episodes=5,++rl_games.env_eval.post_train.enabled=true,++rl_games.env_eval.post_train.latencies=[0,1,2,3,4],++rl_games.env_eval.post_train.max_steps_per_episode=3600,++rl_games.env_eval.post_train.num_episodes=20,++rl_games.env_eval.seed=42,++rl_games.env_eval.task_description=,++rl_games.env_eval.task_seed_stride=0,++rl_games.env_eval.vectorized.batch_size=1,++rl_games.env_eval.vectorized.enabled=false,++rl_games.initialization_mode=bridge,++rl_games.model_alias=wan_oft,++rl_games.task=flappy,++run_id=wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce,++run_root_dir=/workspace/starVLA/results/Checkpoints,++run_root_dir=results/Checkpoints,++seed=42,++trainer.cc_f1_tolerance=1,++trainer.distributed_backend=none,++trainer.eval_action_classification=false,++trainer.eval_action_classification_interval=null,++trainer.eval_interval=500,++trainer.eval_num_batches=50,++trainer.freeze_llm_layers=[],++trainer.freeze_modules=,++trainer.freeze_tied_embedding=false,++trainer.freeze_vit=false,++trainer.gradient_accumulation_steps=32,++trainer.gradient_clipping=1.0,++trainer.is_resume=false,++trainer.is_resume=false,++trainer.learning_rate.action_model=0.0001,++trainer.learning_rate.action_query_proj=0.0001,++trainer.learning_rate.base=2.5e-05,++trainer.learning_rate.qwen_vl_interface=1e-05,++trainer.logging_frequency=1,++trainer.loss_scale.vla=1.0,++trainer.loss_scale.vlm=0.1,++trainer.lr_scheduler_type=cosine_with_min_lr,++trainer.max_grad_norm=1.0,++trainer.max_train_steps=2000,++trainer.num_warmup_steps=0,++trainer.optimizer.betas=[0.9,0.95],++trainer.optimizer.eps=1e-08,++trainer.optimizer.fused=true,++trainer.optimizer.name=AdamW,++trainer.optimizer.weight_decay=1e-08,++trainer.per_latency_eval_num_batches=null,++trainer.pretrained_checkpoint=/workspace/starVLA/playground/Pretrained_models/WM4A-Wan2d2-OFT-LIBERO-4in1/checkpoints/steps_60000_pytorch_model.pt,++trainer.pretrained_checkpoint=null,++trainer.profile_timing.enabled=false,++trainer.profile_timing.log_interval=10,++trainer.reload_modules=null,++trainer.resume_step=0,++trainer.resume_step=0,++trainer.save_format=pt,++trainer.save_interval=0,++trainer.scheduler_specific_kwargs.min_lr=1e-06,++trainer.weight_decay=0.0,++version_id=0.21,++workspace_dir=WORKSPACE_DIR,env=flappy,init=wan_oft_libero,mode=single,model=wan_oft
315
+ id: ???
316
+ num: ???
317
+ config_name: train
318
+ env_set: {}
319
+ env_copy: []
320
+ config:
321
+ override_dirname:
322
+ kv_sep: '='
323
+ item_sep: ','
324
+ exclude_keys: []
325
+ runtime:
326
+ version: 1.3.3
327
+ version_base: '1.1'
328
+ cwd: /workspace/starVLA
329
+ config_sources:
330
+ - path: hydra.conf
331
+ schema: pkg
332
+ provider: hydra
333
+ - path: /workspace/starVLA/examples/rl_games/config
334
+ schema: file
335
+ provider: main
336
+ - path: ''
337
+ schema: structured
338
+ provider: schema
339
+ output_dir: /workspace/starVLA/results/Checkpoints/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce/hydra
340
+ choices:
341
+ cross_task_setup: null
342
+ checkpoint: default
343
+ mode: single
344
+ init: wan_oft_libero
345
+ env: flappy
346
+ model: wan_oft
347
+ hydra/env: default
348
+ hydra/callbacks: null
349
+ hydra/job_logging: default
350
+ hydra/hydra_logging: default
351
+ hydra/hydra_help: default
352
+ hydra/help: default
353
+ hydra/sweeper: basic
354
+ hydra/launcher: basic
355
+ hydra/output: default
356
+ verbose: false
experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce/hydra/.hydra/overrides.yaml ADDED
@@ -0,0 +1,196 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ - model=wan_oft
2
+ - env=flappy
3
+ - init=wan_oft_libero
4
+ - mode=single
5
+ - ++framework.qwenvl.base_vlm=playground/Pretrained_models/Wan-AI/Wan2.2-TI2V-5B-Diffusers
6
+ - ++framework.qwenvl.attn_implementation=sdpa
7
+ - ++framework.qwenvl.enable_gradient_checkpointing=true
8
+ - ++framework.action_model.state_dim=7
9
+ - ++framework.action_model.loss_type=current_discrete_ce
10
+ - ++framework.action_model.action_horizon=8
11
+ - ++framework.action_model.future_action_window_size=7
12
+ - ++framework.action_model.past_action_window_size=0
13
+ - ++framework.action_model.action_model_type=MLP
14
+ - ++framework.action_model.action_dim=7
15
+ - ++framework.action_model.action_env_dim=2
16
+ - ++framework.action_model.action_query_source=mean
17
+ - ++framework.action_model.action_query_num_heads=24
18
+ - ++framework.name=WanOFT
19
+ - ++framework.world_model.base_wm=playground/Pretrained_models/Wan-AI/Wan2.2-TI2V-5B-Diffusers
20
+ - ++framework.world_model.extract_layers=[-1]
21
+ - ++framework.world_model.num_frames=5
22
+ - ++datasets.vla_data.dataset_py=lerobot_datasets
23
+ - ++datasets.vla_data.include_state=true
24
+ - ++datasets.vla_data.data_root_dir=playground/Datasets/rl_games
25
+ - ++datasets.vla_data.data_mix=flappy_train__bridge
26
+ - ++datasets.vla_data.eval_data_mix=flappy_train__bridge__val
27
+ - ++datasets.vla_data.custom_mixtures_path=null
28
+ - ++datasets.vla_data.action_type=discrete
29
+ - ++datasets.vla_data.sequential_step_sampling=false
30
+ - ++datasets.vla_data.eval_sequential_step_sampling=null
31
+ - ++datasets.vla_data.num_workers=8
32
+ - ++datasets.vla_data.eval_num_workers=8
33
+ - ++datasets.vla_data.prefetch_factor=4
34
+ - ++datasets.vla_data.persistent_workers=true
35
+ - ++datasets.vla_data.pin_memory=true
36
+ - ++datasets.vla_data.shuffle=true
37
+ - ++datasets.vla_data.action_balance.enabled=false
38
+ - ++datasets.vla_data.action_balance.strategy=balanced_epoch
39
+ - ++datasets.vla_data.action_balance.action_key=action_id
40
+ - ++datasets.vla_data.action_balance.target_flap_fraction=0.3
41
+ - ++datasets.vla_data.action_balance.noop_id=0
42
+ - ++datasets.vla_data.action_balance.flap_id=1
43
+ - ++datasets.vla_data.latency_curriculum.enabled=false
44
+ - ++datasets.vla_data.latency_curriculum.strategy=exclusive
45
+ - ++datasets.vla_data.latency_curriculum.latencies=null
46
+ - ++datasets.vla_data.latency_curriculum.phase_steps=null
47
+ - ++datasets.vla_data.latency_curriculum.new_latency_passes=1.0
48
+ - ++datasets.vla_data.latency_curriculum.replay_passes=0.25
49
+ - ++datasets.vla_data.latency_curriculum.target_total_passes=2.0
50
+ - ++datasets.vla_data.latency_curriculum.final_equalization=true
51
+ - ++datasets.vla_data.latency_curriculum.step_budget_mode=auto
52
+ - ++datasets.vla_data.latency_curriculum.eval_at_phase_end=false
53
+ - ++datasets.vla_data.latency_curriculum.save_at_phase_end=false
54
+ - ++datasets.vla_data.latency_curriculum.computed_plan=null
55
+ - ++datasets.vla_data.latency_curriculum.effective_batch_size=null
56
+ - ++datasets.vla_data.per_device_batch_size=4
57
+ - ++datasets.vla_data.load_all_data_for_training=true
58
+ - ++datasets.vla_data.num_obs_frames=1
59
+ - ++datasets.vla_data.image_mode=single
60
+ - ++datasets.vla_data.stitch_grid=[2,2]
61
+ - ++datasets.vla_data.obs_image_size=[224,224]
62
+ - ++datasets.vla_data.video_backend=torchvision_av
63
+ - ++datasets.vla_data.observation_indices=[-4,-3,-2,-1,0]
64
+ - ++datasets.vla_data.language_indices=[0]
65
+ - ++datasets.vla_data.state_indices=[0]
66
+ - ++datasets.vla_data.action_indices=[0,1,2,3,4,5,6,7]
67
+ - ++datasets.vla_data.pack_image_sequence=true
68
+ - ++datasets.vla_data.image_sequence_length=5
69
+ - ++datasets.vla_data.context_images_column=observation.context_images
70
+ - ++dataset.source_hf=
71
+ - ++dataset.config_name=null
72
+ - ++dataset.source_subdir=null
73
+ - ++dataset.converted_name=flappy_train__bridge
74
+ - ++dataset.single_source_hf=
75
+ - ++dataset.mixed_source_hf=
76
+ - ++dataset.single_converted_name=flappy_train__bridge
77
+ - ++dataset.mixed_converted_name=flappy_mixed_latency_train
78
+ - ++dataset.single_latency_filter=null
79
+ - ++dataset.mixed_latency_filter=null
80
+ - ++dataset.force_download=false
81
+ - ++dataset.setup_force=false
82
+ - ++dataset.skip_verification=false
83
+ - ++dataset.verify_rows=200
84
+ - ++dataset.max_episodes=null
85
+ - ++dataset.episodes_per_latency=null
86
+ - ++dataset.latency_filter=null
87
+ - ++dataset.debug_subset.enabled=false
88
+ - ++dataset.debug_subset.max_episodes=5
89
+ - ++dataset.debug_subset.suffix=debug
90
+ - ++base_model.repo_id=Wan-AI/Wan2.2-TI2V-5B-Diffusers
91
+ - ++initialization.checkpoint_local_dir=playground/Pretrained_models/WM4A-Wan2d2-OFT-LIBERO-4in1
92
+ - ++initialization.checkpoint_hf_repo_id=StarVLA/WM4A-Wan2d2-OFT-LIBERO-4in1
93
+ - ++initialization.checkpoint_filename=checkpoints/steps_60000_pytorch_model.pt
94
+ - ++trainer.max_train_steps=2000
95
+ - ++trainer.num_warmup_steps=0
96
+ - ++trainer.save_interval=0
97
+ - ++trainer.eval_interval=500
98
+ - ++trainer.eval_num_batches=50
99
+ - ++trainer.per_latency_eval_num_batches=null
100
+ - ++trainer.eval_action_classification=false
101
+ - ++trainer.eval_action_classification_interval=null
102
+ - ++trainer.cc_f1_tolerance=1
103
+ - ++trainer.learning_rate.base=2.5e-05
104
+ - ++trainer.learning_rate.qwen_vl_interface=1e-05
105
+ - ++trainer.learning_rate.action_model=0.0001
106
+ - ++trainer.learning_rate.action_query_proj=0.0001
107
+ - ++trainer.lr_scheduler_type=cosine_with_min_lr
108
+ - ++trainer.scheduler_specific_kwargs.min_lr=1e-06
109
+ - ++trainer.freeze_modules=
110
+ - ++trainer.freeze_vit=false
111
+ - ++trainer.freeze_tied_embedding=false
112
+ - ++trainer.freeze_llm_layers=[]
113
+ - ++trainer.loss_scale.vla=1.0
114
+ - ++trainer.loss_scale.vlm=0.1
115
+ - ++trainer.max_grad_norm=1.0
116
+ - ++trainer.weight_decay=0.0
117
+ - ++trainer.logging_frequency=1
118
+ - ++trainer.profile_timing.enabled=false
119
+ - ++trainer.profile_timing.log_interval=10
120
+ - ++trainer.gradient_clipping=1.0
121
+ - ++trainer.gradient_accumulation_steps=32
122
+ - ++trainer.distributed_backend=none
123
+ - ++trainer.is_resume=false
124
+ - ++trainer.pretrained_checkpoint=null
125
+ - ++trainer.resume_step=0
126
+ - ++trainer.reload_modules=null
127
+ - ++trainer.optimizer.name=AdamW
128
+ - ++trainer.optimizer.betas=[0.9,0.95]
129
+ - ++trainer.optimizer.eps=1e-08
130
+ - ++trainer.optimizer.weight_decay=1e-08
131
+ - ++trainer.optimizer.fused=true
132
+ - ++trainer.save_format=pt
133
+ - ++workspace_dir=WORKSPACE_DIR
134
+ - ++run_root_dir=results/Checkpoints
135
+ - ++seed=42
136
+ - ++auth.env_file=null
137
+ - ++auth.hf_token_env=HF_TOKEN
138
+ - ++auth.wandb_api_key_env=WANDB_API_KEY
139
+ - ++paths.run_root_dir=results/Checkpoints
140
+ - ++paths.dataset_local_dir=data/flappy_fix_latency_1_200ep_context5
141
+ - ++paths.dataset_cache_dir=null
142
+ - ++paths.base_model_dir=playground/Pretrained_models/Wan-AI/Wan2.2-TI2V-5B-Diffusers
143
+ - ++paths.accelerate_config=starVLA/config/deepseeds/deepspeed_zero2.yaml
144
+ - ++rl_games.model_alias=wan_oft
145
+ - ++rl_games.env_eval.image_size=224
146
+ - ++rl_games.env_eval.frameskip=1
147
+ - ++rl_games.env_eval.seed=42
148
+ - ++rl_games.env_eval.fixed_episode_seeds=true
149
+ - ++rl_games.env_eval.latency_seed_stride=0
150
+ - ++rl_games.env_eval.task_seed_stride=0
151
+ - ++rl_games.env_eval.task_description=
152
+ - ++rl_games.env_eval.action_chunk_execution.enabled=false
153
+ - ++rl_games.env_eval.action_chunk_execution.chunk_size=null
154
+ - ++rl_games.env_eval.enabled=true
155
+ - ++rl_games.env_eval.distributed_mode=none
156
+ - ++rl_games.env_eval.vectorized.enabled=false
157
+ - ++rl_games.env_eval.vectorized.batch_size=1
158
+ - ++rl_games.env_eval.latency.prompt_map_path=data/flappy_fix_latency_1_200ep_context5/flappy_train__bridge/latency_prompt_map_eval_0_4.json
159
+ - ++rl_games.env_eval.latency.mode=single
160
+ - ++rl_games.env_eval.latency.values=[0]
161
+ - ++rl_games.env_eval.mid_train.enabled=true
162
+ - ++rl_games.env_eval.mid_train.interval_steps=500
163
+ - ++rl_games.env_eval.mid_train.latencies=[1]
164
+ - ++rl_games.env_eval.mid_train.num_episodes=5
165
+ - ++rl_games.env_eval.mid_train.max_steps_per_episode=3600
166
+ - ++rl_games.env_eval.post_train.enabled=true
167
+ - ++rl_games.env_eval.post_train.latencies=[0,1,2,3,4]
168
+ - ++rl_games.env_eval.post_train.num_episodes=20
169
+ - ++rl_games.env_eval.post_train.max_steps_per_episode=3600
170
+ - ++rl_games.task=flappy
171
+ - ++rl_games.initialization_mode=bridge
172
+ - ++rl_games.action_carrier=bridge
173
+ - ++checkpoint.load=none
174
+ - ++checkpoint.hf_repo_id=null
175
+ - ++checkpoint.save_best_model=false
176
+ - ++checkpoint.save_final_model=true
177
+ - ++checkpoint.save_pt_file=true
178
+ - ++checkpoint.save_training_state=false
179
+ - ++checkpoint.local.keep_last_n=1
180
+ - ++checkpoint.sync.enabled=false
181
+ - ++checkpoint.sync.repo_id=null
182
+ - ++checkpoint.sync.keep_last_n=0
183
+ - ++checkpoint.sync.sync_every_n_checkpoints=1
184
+ - ++checkpoint.sync.resume_policy=local_latest
185
+ - ++run_id=wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce
186
+ - ++output_dir=null
187
+ - ++config_yaml=null
188
+ - ++is_debug=false
189
+ - ++version_id=0.21
190
+ - ++run_root_dir=/workspace/starVLA/results/Checkpoints
191
+ - ++trainer.is_resume=false
192
+ - ++trainer.pretrained_checkpoint=/workspace/starVLA/playground/Pretrained_models/WM4A-Wan2d2-OFT-LIBERO-4in1/checkpoints/steps_60000_pytorch_model.pt
193
+ - ++trainer.resume_step=0
194
+ - ++datasets.vla_data.data_root_dir=/workspace/starVLA/data/flappy_fix_latency_1_200ep_context5
195
+ - ++framework.qwenvl.base_vlm=/workspace/starVLA/playground/Pretrained_models/Wan-AI/Wan2.2-TI2V-5B-Diffusers
196
+ - ++framework.world_model.base_wm=/workspace/starVLA/playground/Pretrained_models/Wan-AI/Wan2.2-TI2V-5B-Diffusers
experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce/hydra/train_starvla_hydra.log ADDED
The diff for this file is too large to render. See raw diff
 
experiments/latency-transfer/provenance/latency-transfer-data/provenance/duplicate_aliases/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce/summary.jsonl ADDED
@@ -0,0 +1 @@
 
 
1
+ {"steps": 2000}
experiments/latency-transfer/provenance/latency-transfer-data/provenance/fixed_conversion_probe.log ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ [setup] rebuilding converted dataset because:
2
+ [setup] - dataset.setup_force or dataset.force_download is true
3
+ [setup] - train dataset is not ready: /tmp/experiment-repo-cleanup-20261007/conversion_probe/converted/flappy_train__bridge
4
+ [setup] - eval dataset is not ready: /tmp/experiment-repo-cleanup-20261007/conversion_probe/converted/flappy_train__bridge__val
5
+
6
+
7
+
8
+
9
+ {"fixed_revision_preserved": true, "train_and_val_converted": true, "actual_parquet_read": true, "snapshot_prefix": "training/flappy_200ep/flappy_fix_latency_2_200ep", "snapshot_calls": 1}
experiments/latency-transfer/provenance/latency-transfer-data/provenance/native_loader_verification.json ADDED
The diff for this file is too large to render. See raw diff
 
experiments/latency-transfer/provenance/latency-transfer-data/provenance/old_code_recovery.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ [
2
+ {
3
+ "commit": "538fd523fe4fa4e7abce3201611919314bc6b664",
4
+ "http_status": 200
5
+ },
6
+ {
7
+ "commit": "9aa8c96e8044e6dc48063a9f7e0b13df513ebf62",
8
+ "http_status": 200
9
+ }
10
+ ]
experiments/latency-transfer/provenance/latency-transfer-data/provenance/original_reference_documents/latency-transfer-data/run_index.json ADDED
@@ -0,0 +1,144 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [
2
+ {
3
+ "run": "wan_oft_flappy_fix_latency_0_context5_2000_effbs128",
4
+ "module": "latency-transfer",
5
+ "train_latency_raw_frames": 0,
6
+ "train_seed": 42,
7
+ "training_step": 2000,
8
+ "training_started_at_utc": "2026-06-28T13:44:26.129154Z",
9
+ "git_commit": "f9acb69ef3d99d2a8b620103161e15eebc4c3091",
10
+ "wandb_run_id": "ky48y0be",
11
+ "checkpoint": "wan_oft_flappy_fix_latency_0_context5_2000_effbs128/checkpoints/steps_2000_pytorch_model.pt",
12
+ "checkpoint_identity": "daa8a61baa149139792808dcfc5406df766d39113161b92d0baa13a9bd2402f4",
13
+ "source_model_revision": "1ba5d2f489f03f9cb07a637ca680fa4f2f8e51a4",
14
+ "dataset": {
15
+ "repo": "latency-sensitive-bench/latency-transfer-data",
16
+ "revision": "ecc53001af8f0019b4d50bf527c161904f27d260",
17
+ "path": "training/flappy_fix_latency_0_200ep_context5",
18
+ "source_repo": "latency-sensitive-bench/flappy_200ep_context5",
19
+ "source_revision": "19d800e5f7ec9d5311f508dbfe4a50463261616c",
20
+ "train_rows": 279882,
21
+ "train_episodes": 180,
22
+ "val_rows": 72000,
23
+ "val_episodes": 20,
24
+ "pairing_evidence": "Training local path, row count, trajectory count, action statistics and fixed raw-frame latency"
25
+ },
26
+ "paper_consumers": [
27
+ "fig12_wanoft_latency_grid"
28
+ ]
29
+ },
30
+ {
31
+ "run": "wan_oft_flappy_fix_latency_1_context5_2000_effbs128",
32
+ "module": "latency-transfer",
33
+ "train_latency_raw_frames": 1,
34
+ "train_seed": 42,
35
+ "training_step": 2000,
36
+ "training_started_at_utc": "2026-06-29T14:50:14.198406Z",
37
+ "git_commit": "972037fae54ada9fe8f64e4ac8878b2f79feba1f",
38
+ "wandb_run_id": "3kohuetr",
39
+ "checkpoint": "wan_oft_flappy_fix_latency_1_context5_2000_effbs128/checkpoints/steps_2000_pytorch_model.pt",
40
+ "checkpoint_identity": "ca01daf3e577fedcc238882e531c748f6a5c178934ca7cc7209da18eb24ff03f",
41
+ "source_model_revision": "1ba5d2f489f03f9cb07a637ca680fa4f2f8e51a4",
42
+ "dataset": {
43
+ "repo": "latency-sensitive-bench/latency-transfer-data",
44
+ "revision": "ecc53001af8f0019b4d50bf527c161904f27d260",
45
+ "path": "training/flappy_fix_latency_1_200ep_context5",
46
+ "source_repo": "latency-sensitive-bench/flappy_200ep_context5",
47
+ "source_revision": "19d800e5f7ec9d5311f508dbfe4a50463261616c",
48
+ "train_rows": 297483,
49
+ "train_episodes": 180,
50
+ "val_rows": 72000,
51
+ "val_episodes": 20,
52
+ "pairing_evidence": "Training local path, row count, trajectory count, action statistics and fixed raw-frame latency"
53
+ },
54
+ "paper_consumers": [
55
+ "fig12_wanoft_latency_grid"
56
+ ]
57
+ },
58
+ {
59
+ "run": "wan_oft_flappy_fix_latency_2_context5_2000_effbs128",
60
+ "module": "latency-transfer",
61
+ "train_latency_raw_frames": 2,
62
+ "train_seed": 42,
63
+ "training_step": 2000,
64
+ "training_started_at_utc": "2026-07-01T09:28:56.645196Z",
65
+ "git_commit": "972037fae54ada9fe8f64e4ac8878b2f79feba1f",
66
+ "wandb_run_id": "2z7am65s",
67
+ "checkpoint": "wan_oft_flappy_fix_latency_2_context5_2000_effbs128/checkpoints/steps_2000_pytorch_model.pt",
68
+ "checkpoint_identity": "328140abd4682a66cb56cdbe03524f6c7af534e2c9ac2bd23ac1c9691363957a",
69
+ "source_model_revision": "1ba5d2f489f03f9cb07a637ca680fa4f2f8e51a4",
70
+ "dataset": {
71
+ "repo": "latency-sensitive-bench/latency-transfer-data",
72
+ "revision": "ecc53001af8f0019b4d50bf527c161904f27d260",
73
+ "path": "training/flappy_fix_latency_2_200ep_context5",
74
+ "source_repo": "latency-sensitive-bench/flappy_200ep_context5",
75
+ "source_revision": "19d800e5f7ec9d5311f508dbfe4a50463261616c",
76
+ "train_rows": 330734,
77
+ "train_episodes": 180,
78
+ "val_rows": 72000,
79
+ "val_episodes": 20,
80
+ "pairing_evidence": "Training local path, row count, trajectory count, action statistics and fixed raw-frame latency"
81
+ },
82
+ "paper_consumers": [
83
+ "fig12_wanoft_latency_grid",
84
+ "fig13_wanoft_timing_controls",
85
+ "fig14_wanoft_timing_buckets"
86
+ ]
87
+ },
88
+ {
89
+ "run": "wan_oft_flappy_fix_latency_3_context5_2000_effbs128",
90
+ "module": "latency-transfer",
91
+ "train_latency_raw_frames": 3,
92
+ "train_seed": 42,
93
+ "training_step": 2000,
94
+ "training_started_at_utc": "2026-07-01T16:06:06.153675Z",
95
+ "git_commit": "972037fae54ada9fe8f64e4ac8878b2f79feba1f",
96
+ "wandb_run_id": "jlutvxqx",
97
+ "checkpoint": "wan_oft_flappy_fix_latency_3_context5_2000_effbs128/checkpoints/steps_2000_pytorch_model.pt",
98
+ "checkpoint_identity": "d5940d8ea363b40611f391103534da4d13dd7204e43024079c56eb2f34322bcc",
99
+ "source_model_revision": "1ba5d2f489f03f9cb07a637ca680fa4f2f8e51a4",
100
+ "dataset": {
101
+ "repo": "latency-sensitive-bench/latency-transfer-data",
102
+ "revision": "ecc53001af8f0019b4d50bf527c161904f27d260",
103
+ "path": "training/flappy_fix_latency_3_200ep_context5",
104
+ "source_repo": "latency-sensitive-bench/flappy_200ep_context5",
105
+ "source_revision": "19d800e5f7ec9d5311f508dbfe4a50463261616c",
106
+ "train_rows": 316258,
107
+ "train_episodes": 180,
108
+ "val_rows": 72000,
109
+ "val_episodes": 20,
110
+ "pairing_evidence": "Training local path, row count, trajectory count, action statistics and fixed raw-frame latency"
111
+ },
112
+ "paper_consumers": [
113
+ "fig12_wanoft_latency_grid"
114
+ ]
115
+ },
116
+ {
117
+ "run": "wan_oft_flappy_fix_latency_4_context5_2000_effbs128",
118
+ "module": "latency-transfer",
119
+ "train_latency_raw_frames": 4,
120
+ "train_seed": 42,
121
+ "training_step": 2000,
122
+ "training_started_at_utc": "2026-07-01T17:12:40.334458Z",
123
+ "git_commit": "972037fae54ada9fe8f64e4ac8878b2f79feba1f",
124
+ "wandb_run_id": "hvokzu73",
125
+ "checkpoint": "wan_oft_flappy_fix_latency_4_context5_2000_effbs128/checkpoints/steps_2000_pytorch_model.pt",
126
+ "checkpoint_identity": "05a1c777b84de4f2b66a2db1a13eb7b1aa326167eabb30ddf50bd149d6e68118",
127
+ "source_model_revision": "1ba5d2f489f03f9cb07a637ca680fa4f2f8e51a4",
128
+ "dataset": {
129
+ "repo": "latency-sensitive-bench/latency-transfer-data",
130
+ "revision": "ecc53001af8f0019b4d50bf527c161904f27d260",
131
+ "path": "training/flappy_fix_latency_4_200ep_context5",
132
+ "source_repo": "latency-sensitive-bench/flappy_200ep_context5",
133
+ "source_revision": "19d800e5f7ec9d5311f508dbfe4a50463261616c",
134
+ "train_rows": 311588,
135
+ "train_episodes": 180,
136
+ "val_rows": 72000,
137
+ "val_episodes": 20,
138
+ "pairing_evidence": "Training local path, row count, trajectory count, action statistics and fixed raw-frame latency"
139
+ },
140
+ "paper_consumers": [
141
+ "fig12_wanoft_latency_grid"
142
+ ]
143
+ }
144
+ ]
experiments/latency-transfer/provenance/latency-transfer-data/provenance/original_reference_documents/latency-transfer-models/run_index.json ADDED
@@ -0,0 +1,144 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [
2
+ {
3
+ "run": "wan_oft_flappy_fix_latency_0_context5_2000_effbs128",
4
+ "module": "latency-transfer",
5
+ "train_latency_raw_frames": 0,
6
+ "train_seed": 42,
7
+ "training_step": 2000,
8
+ "training_started_at_utc": "2026-06-28T13:44:26.129154Z",
9
+ "git_commit": "f9acb69ef3d99d2a8b620103161e15eebc4c3091",
10
+ "wandb_run_id": "ky48y0be",
11
+ "checkpoint": "wan_oft_flappy_fix_latency_0_context5_2000_effbs128/checkpoints/steps_2000_pytorch_model.pt",
12
+ "checkpoint_identity": "daa8a61baa149139792808dcfc5406df766d39113161b92d0baa13a9bd2402f4",
13
+ "source_model_revision": "1ba5d2f489f03f9cb07a637ca680fa4f2f8e51a4",
14
+ "dataset": {
15
+ "repo": "latency-sensitive-bench/latency-transfer-data",
16
+ "revision": "23f4501059d58f7df4d12d7a074f48daca4c578d",
17
+ "path": "training/flappy_fix_latency_0_200ep_context5",
18
+ "source_repo": "latency-sensitive-bench/flappy_200ep_context5",
19
+ "source_revision": "19d800e5f7ec9d5311f508dbfe4a50463261616c",
20
+ "train_rows": 279882,
21
+ "train_episodes": 180,
22
+ "val_rows": 72000,
23
+ "val_episodes": 20,
24
+ "pairing_evidence": "Training local path, row count, trajectory count, action statistics and fixed raw-frame latency"
25
+ },
26
+ "paper_consumers": [
27
+ "fig12_wanoft_latency_grid"
28
+ ]
29
+ },
30
+ {
31
+ "run": "wan_oft_flappy_fix_latency_1_context5_2000_effbs128",
32
+ "module": "latency-transfer",
33
+ "train_latency_raw_frames": 1,
34
+ "train_seed": 42,
35
+ "training_step": 2000,
36
+ "training_started_at_utc": "2026-06-29T14:50:14.198406Z",
37
+ "git_commit": "972037fae54ada9fe8f64e4ac8878b2f79feba1f",
38
+ "wandb_run_id": "3kohuetr",
39
+ "checkpoint": "wan_oft_flappy_fix_latency_1_context5_2000_effbs128/checkpoints/steps_2000_pytorch_model.pt",
40
+ "checkpoint_identity": "ca01daf3e577fedcc238882e531c748f6a5c178934ca7cc7209da18eb24ff03f",
41
+ "source_model_revision": "1ba5d2f489f03f9cb07a637ca680fa4f2f8e51a4",
42
+ "dataset": {
43
+ "repo": "latency-sensitive-bench/latency-transfer-data",
44
+ "revision": "23f4501059d58f7df4d12d7a074f48daca4c578d",
45
+ "path": "training/flappy_fix_latency_1_200ep_context5",
46
+ "source_repo": "latency-sensitive-bench/flappy_200ep_context5",
47
+ "source_revision": "19d800e5f7ec9d5311f508dbfe4a50463261616c",
48
+ "train_rows": 297483,
49
+ "train_episodes": 180,
50
+ "val_rows": 72000,
51
+ "val_episodes": 20,
52
+ "pairing_evidence": "Training local path, row count, trajectory count, action statistics and fixed raw-frame latency"
53
+ },
54
+ "paper_consumers": [
55
+ "fig12_wanoft_latency_grid"
56
+ ]
57
+ },
58
+ {
59
+ "run": "wan_oft_flappy_fix_latency_2_context5_2000_effbs128",
60
+ "module": "latency-transfer",
61
+ "train_latency_raw_frames": 2,
62
+ "train_seed": 42,
63
+ "training_step": 2000,
64
+ "training_started_at_utc": "2026-07-01T09:28:56.645196Z",
65
+ "git_commit": "972037fae54ada9fe8f64e4ac8878b2f79feba1f",
66
+ "wandb_run_id": "2z7am65s",
67
+ "checkpoint": "wan_oft_flappy_fix_latency_2_context5_2000_effbs128/checkpoints/steps_2000_pytorch_model.pt",
68
+ "checkpoint_identity": "328140abd4682a66cb56cdbe03524f6c7af534e2c9ac2bd23ac1c9691363957a",
69
+ "source_model_revision": "1ba5d2f489f03f9cb07a637ca680fa4f2f8e51a4",
70
+ "dataset": {
71
+ "repo": "latency-sensitive-bench/latency-transfer-data",
72
+ "revision": "23f4501059d58f7df4d12d7a074f48daca4c578d",
73
+ "path": "training/flappy_fix_latency_2_200ep_context5",
74
+ "source_repo": "latency-sensitive-bench/flappy_200ep_context5",
75
+ "source_revision": "19d800e5f7ec9d5311f508dbfe4a50463261616c",
76
+ "train_rows": 330734,
77
+ "train_episodes": 180,
78
+ "val_rows": 72000,
79
+ "val_episodes": 20,
80
+ "pairing_evidence": "Training local path, row count, trajectory count, action statistics and fixed raw-frame latency"
81
+ },
82
+ "paper_consumers": [
83
+ "fig12_wanoft_latency_grid",
84
+ "fig13_wanoft_timing_controls",
85
+ "fig14_wanoft_timing_buckets"
86
+ ]
87
+ },
88
+ {
89
+ "run": "wan_oft_flappy_fix_latency_3_context5_2000_effbs128",
90
+ "module": "latency-transfer",
91
+ "train_latency_raw_frames": 3,
92
+ "train_seed": 42,
93
+ "training_step": 2000,
94
+ "training_started_at_utc": "2026-07-01T16:06:06.153675Z",
95
+ "git_commit": "972037fae54ada9fe8f64e4ac8878b2f79feba1f",
96
+ "wandb_run_id": "jlutvxqx",
97
+ "checkpoint": "wan_oft_flappy_fix_latency_3_context5_2000_effbs128/checkpoints/steps_2000_pytorch_model.pt",
98
+ "checkpoint_identity": "d5940d8ea363b40611f391103534da4d13dd7204e43024079c56eb2f34322bcc",
99
+ "source_model_revision": "1ba5d2f489f03f9cb07a637ca680fa4f2f8e51a4",
100
+ "dataset": {
101
+ "repo": "latency-sensitive-bench/latency-transfer-data",
102
+ "revision": "23f4501059d58f7df4d12d7a074f48daca4c578d",
103
+ "path": "training/flappy_fix_latency_3_200ep_context5",
104
+ "source_repo": "latency-sensitive-bench/flappy_200ep_context5",
105
+ "source_revision": "19d800e5f7ec9d5311f508dbfe4a50463261616c",
106
+ "train_rows": 316258,
107
+ "train_episodes": 180,
108
+ "val_rows": 72000,
109
+ "val_episodes": 20,
110
+ "pairing_evidence": "Training local path, row count, trajectory count, action statistics and fixed raw-frame latency"
111
+ },
112
+ "paper_consumers": [
113
+ "fig12_wanoft_latency_grid"
114
+ ]
115
+ },
116
+ {
117
+ "run": "wan_oft_flappy_fix_latency_4_context5_2000_effbs128",
118
+ "module": "latency-transfer",
119
+ "train_latency_raw_frames": 4,
120
+ "train_seed": 42,
121
+ "training_step": 2000,
122
+ "training_started_at_utc": "2026-07-01T17:12:40.334458Z",
123
+ "git_commit": "972037fae54ada9fe8f64e4ac8878b2f79feba1f",
124
+ "wandb_run_id": "hvokzu73",
125
+ "checkpoint": "wan_oft_flappy_fix_latency_4_context5_2000_effbs128/checkpoints/steps_2000_pytorch_model.pt",
126
+ "checkpoint_identity": "05a1c777b84de4f2b66a2db1a13eb7b1aa326167eabb30ddf50bd149d6e68118",
127
+ "source_model_revision": "1ba5d2f489f03f9cb07a637ca680fa4f2f8e51a4",
128
+ "dataset": {
129
+ "repo": "latency-sensitive-bench/latency-transfer-data",
130
+ "revision": "23f4501059d58f7df4d12d7a074f48daca4c578d",
131
+ "path": "training/flappy_fix_latency_4_200ep_context5",
132
+ "source_repo": "latency-sensitive-bench/flappy_200ep_context5",
133
+ "source_revision": "19d800e5f7ec9d5311f508dbfe4a50463261616c",
134
+ "train_rows": 311588,
135
+ "train_episodes": 180,
136
+ "val_rows": 72000,
137
+ "val_episodes": 20,
138
+ "pairing_evidence": "Training local path, row count, trajectory count, action statistics and fixed raw-frame latency"
139
+ },
140
+ "paper_consumers": [
141
+ "fig12_wanoft_latency_grid"
142
+ ]
143
+ }
144
+ ]
experiments/latency-transfer/provenance/latency-transfer-data/provenance/pinned_setup_tests.log ADDED
@@ -0,0 +1,17 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ........ [100%]
2
+ =============================== warnings summary ===============================
3
+ third_party/starVLA/tests/rl_games/test_pi05_flappy_sft_path.py::test_launch_train_setup_namespace_uses_composed_hydra_config
4
+ third_party/starVLA/tests/rl_games/test_pi05_flappy_sft_path.py::test_launch_train_setup_namespace_forwards_raw_frame_latency_unit
5
+ third_party/starVLA/tests/rl_games/test_pi05_flappy_sft_path.py::test_launch_train_setup_namespace_forwards_deadly_factorized_layout
6
+ third_party/starVLA/tests/rl_games/test_pi05_flappy_sft_path.py::test_launch_train_setup_namespace_forwards_explicit_dataset_source_hf
7
+ third_party/starVLA/tests/rl_games/test_pi05_flappy_sft_path.py::test_launch_train_setup_namespace_forwards_dataset_config_name
8
+ third_party/starVLA/tests/rl_games/test_pi05_flappy_sft_path.py::test_launch_train_setup_namespace_forwards_dataset_source_subdir
9
+ /home/lixinyuan/latency-sensitive-bench/third_party/starVLA/examples/rl_games/scripts/launch_train.py:278: Hydra14MigrationWarning:
10
+ version_base="1.1" selects Hydra 1.1 compatibility behavior.
11
+ Hydra 1.4 will remove this compatibility behavior.
12
+ Before upgrading to Hydra 1.4, set version_base="1.3" and test your application.
13
+ See https://hydra.cc/docs/upgrades/1.3_to_1.4/prepare_for_1_4/ for preparation instructions.
14
+ with initialize_config_dir(version_base="1.1", config_dir=str(CONFIG_DIR)):
15
+
16
+ -- Docs: https://docs.pytest.org/en/stable/how-to/capture-warnings.html
17
+ 8 passed, 32 deselected, 6 warnings in 1.32s
experiments/latency-transfer/provenance/latency-transfer-data/provenance/probe_wandb_artifact_contents.json ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "data": {
3
+ "project": {
4
+ "run": {
5
+ "outputArtifacts": {
6
+ "edges": [
7
+ {
8
+ "node": {
9
+ "artifactSequence": {
10
+ "name": "run-i4aobzgd-comparisonspre_post-7S3BqA"
11
+ },
12
+ "versionIndex": 0,
13
+ "fileCount": 1,
14
+ "size": 6676
15
+ }
16
+ },
17
+ {
18
+ "node": {
19
+ "artifactSequence": {
20
+ "name": "run-i4aobzgd-comparisonscontrol_drops-RhD2_Q"
21
+ },
22
+ "versionIndex": 0,
23
+ "fileCount": 1,
24
+ "size": 10945
25
+ }
26
+ },
27
+ {
28
+ "node": {
29
+ "artifactSequence": {
30
+ "name": "wan_oft_flappy_latency2_context5_temporal_linear_probe_v0-report"
31
+ },
32
+ "versionIndex": 0,
33
+ "fileCount": 1,
34
+ "size": 346819
35
+ }
36
+ },
37
+ {
38
+ "node": {
39
+ "artifactSequence": {
40
+ "name": "run-i4aobzgd-events"
41
+ },
42
+ "versionIndex": 0,
43
+ "fileCount": 1,
44
+ "size": 462655
45
+ }
46
+ },
47
+ {
48
+ "node": {
49
+ "artifactSequence": {
50
+ "name": "run-i4aobzgd-history"
51
+ },
52
+ "versionIndex": 0,
53
+ "fileCount": 1,
54
+ "size": 142005
55
+ }
56
+ }
57
+ ]
58
+ }
59
+ }
60
+ }
61
+ }
62
+ }
experiments/latency-transfer/provenance/latency-transfer-data/provenance/probe_wandb_artifacts.json ADDED
@@ -0,0 +1,174 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "data": {
3
+ "__type": {
4
+ "fields": [
5
+ {
6
+ "name": "aliases"
7
+ },
8
+ {
9
+ "name": "artifactCollections"
10
+ },
11
+ {
12
+ "name": "artifactLineageDag"
13
+ },
14
+ {
15
+ "name": "artifactMemberships"
16
+ },
17
+ {
18
+ "name": "artifactSequence"
19
+ },
20
+ {
21
+ "name": "artifactSequenceMembership"
22
+ },
23
+ {
24
+ "name": "artifactType"
25
+ },
26
+ {
27
+ "name": "commitHash"
28
+ },
29
+ {
30
+ "name": "createdAt"
31
+ },
32
+ {
33
+ "name": "createdBy"
34
+ },
35
+ {
36
+ "name": "currentManifest"
37
+ },
38
+ {
39
+ "name": "dependencyOf"
40
+ },
41
+ {
42
+ "name": "dependsOn"
43
+ },
44
+ {
45
+ "name": "description"
46
+ },
47
+ {
48
+ "name": "digest"
49
+ },
50
+ {
51
+ "name": "digestAlgorithm"
52
+ },
53
+ {
54
+ "name": "fileCount"
55
+ },
56
+ {
57
+ "name": "historyStep"
58
+ },
59
+ {
60
+ "name": "id"
61
+ },
62
+ {
63
+ "name": "isGenerated"
64
+ },
65
+ {
66
+ "name": "isLinkedToGlobalRegistry"
67
+ },
68
+ {
69
+ "name": "isProtected"
70
+ },
71
+ {
72
+ "name": "metadata"
73
+ },
74
+ {
75
+ "name": "runHistory"
76
+ },
77
+ {
78
+ "name": "size"
79
+ },
80
+ {
81
+ "name": "state"
82
+ },
83
+ {
84
+ "name": "storageBytes"
85
+ },
86
+ {
87
+ "name": "storageRegion"
88
+ },
89
+ {
90
+ "name": "tags"
91
+ },
92
+ {
93
+ "name": "ttlDurationSeconds"
94
+ },
95
+ {
96
+ "name": "ttlIsInherited"
97
+ },
98
+ {
99
+ "name": "updatedAt"
100
+ },
101
+ {
102
+ "name": "usedBy"
103
+ },
104
+ {
105
+ "name": "usedCount"
106
+ },
107
+ {
108
+ "name": "versionIndex"
109
+ },
110
+ {
111
+ "name": "weaveSource"
112
+ }
113
+ ]
114
+ },
115
+ "project": {
116
+ "run": {
117
+ "outputArtifacts": {
118
+ "edges": [
119
+ {
120
+ "node": {
121
+ "id": "QXJ0aWZhY3Q6MzEzNTYwOTQ4OQ==",
122
+ "digest": "f4a668adb863ad8605b253e099858c14",
123
+ "artifactSequence": {
124
+ "name": "run-i4aobzgd-comparisonspre_post-7S3BqA"
125
+ },
126
+ "versionIndex": 0
127
+ }
128
+ },
129
+ {
130
+ "node": {
131
+ "id": "QXJ0aWZhY3Q6MzEzNTYwOTUwMw==",
132
+ "digest": "4833225fa4fb4d9793cd8a9006b7b51b",
133
+ "artifactSequence": {
134
+ "name": "run-i4aobzgd-comparisonscontrol_drops-RhD2_Q"
135
+ },
136
+ "versionIndex": 0
137
+ }
138
+ },
139
+ {
140
+ "node": {
141
+ "id": "QXJ0aWZhY3Q6MzEzNTYwOTUxMA==",
142
+ "digest": "4c586288f98a7faefd0f4083b37c2475",
143
+ "artifactSequence": {
144
+ "name": "wan_oft_flappy_latency2_context5_temporal_linear_probe_v0-report"
145
+ },
146
+ "versionIndex": 0
147
+ }
148
+ },
149
+ {
150
+ "node": {
151
+ "id": "QXJ0aWZhY3Q6MzEzNTY1NDM5Mg==",
152
+ "digest": "571adb4e5df7e175ebe4b63a5e24cd65",
153
+ "artifactSequence": {
154
+ "name": "run-i4aobzgd-events"
155
+ },
156
+ "versionIndex": 0
157
+ }
158
+ },
159
+ {
160
+ "node": {
161
+ "id": "QXJ0aWZhY3Q6MzEzNTY1NDY0Ng==",
162
+ "digest": "3290c1666a63c7b0ccc45e40ef84e328",
163
+ "artifactSequence": {
164
+ "name": "run-i4aobzgd-history"
165
+ },
166
+ "versionIndex": 0
167
+ }
168
+ }
169
+ ]
170
+ }
171
+ }
172
+ }
173
+ }
174
+ }
experiments/latency-transfer/provenance/latency-transfer-data/provenance/probe_wandb_files.json ADDED
@@ -0,0 +1,97 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "data": {
3
+ "project": {
4
+ "run": {
5
+ "id": "UnVuOnYxOmk0YW9iemdkOnN0YXJWTEFfcmxfZ2FtZXM6ZG9uZ3FpYW55dTk5LXpoZWppYW5nLXVuaXZlcnNpdHk=",
6
+ "files": {
7
+ "edges": [
8
+ {
9
+ "node": {
10
+ "name": "artifact/3135609489/wandb_manifest.json",
11
+ "sizeBytes": 335,
12
+ "url": "https://api.wandb.ai/files/dongqianyu99-zhejiang-university/starVLA_rl_games/i4aobzgd/artifact/3135609489/wandb_manifest.json"
13
+ }
14
+ },
15
+ {
16
+ "node": {
17
+ "name": "artifact/3135609503/wandb_manifest.json",
18
+ "sizeBytes": 341,
19
+ "url": "https://api.wandb.ai/files/dongqianyu99-zhejiang-university/starVLA_rl_games/i4aobzgd/artifact/3135609503/wandb_manifest.json"
20
+ }
21
+ },
22
+ {
23
+ "node": {
24
+ "name": "artifact/3135609510/wandb_manifest.json",
25
+ "sizeBytes": 317,
26
+ "url": "https://api.wandb.ai/files/dongqianyu99-zhejiang-university/starVLA_rl_games/i4aobzgd/artifact/3135609510/wandb_manifest.json"
27
+ }
28
+ },
29
+ {
30
+ "node": {
31
+ "name": "artifact/3135654392/wandb_manifest.json",
32
+ "sizeBytes": 0,
33
+ "url": "https://api.wandb.ai/files/dongqianyu99-zhejiang-university/starVLA_rl_games/i4aobzgd/artifact/3135654392/wandb_manifest.json"
34
+ }
35
+ },
36
+ {
37
+ "node": {
38
+ "name": "artifact/3135654646/wandb_manifest.json",
39
+ "sizeBytes": 0,
40
+ "url": "https://api.wandb.ai/files/dongqianyu99-zhejiang-university/starVLA_rl_games/i4aobzgd/artifact/3135654646/wandb_manifest.json"
41
+ }
42
+ },
43
+ {
44
+ "node": {
45
+ "name": "config.yaml",
46
+ "sizeBytes": 6146,
47
+ "url": "https://api.wandb.ai/files/dongqianyu99-zhejiang-university/starVLA_rl_games/i4aobzgd/config.yaml"
48
+ }
49
+ },
50
+ {
51
+ "node": {
52
+ "name": "media/table/comparisons/control_drops_7_bc05a2ab4fb5dbc07ef5.table.json",
53
+ "sizeBytes": 9500,
54
+ "url": "https://api.wandb.ai/files/dongqianyu99-zhejiang-university/starVLA_rl_games/i4aobzgd/media/table/comparisons/control_drops_7_bc05a2ab4fb5dbc07ef5.table.json"
55
+ }
56
+ },
57
+ {
58
+ "node": {
59
+ "name": "media/table/comparisons/pre_post_7_cb9c105bb875bb816493.table.json",
60
+ "sizeBytes": 5562,
61
+ "url": "https://api.wandb.ai/files/dongqianyu99-zhejiang-university/starVLA_rl_games/i4aobzgd/media/table/comparisons/pre_post_7_cb9c105bb875bb816493.table.json"
62
+ }
63
+ },
64
+ {
65
+ "node": {
66
+ "name": "output.log",
67
+ "sizeBytes": 2405,
68
+ "url": "https://api.wandb.ai/files/dongqianyu99-zhejiang-university/starVLA_rl_games/i4aobzgd/output.log"
69
+ }
70
+ },
71
+ {
72
+ "node": {
73
+ "name": "requirements.txt",
74
+ "sizeBytes": 3120,
75
+ "url": "https://api.wandb.ai/files/dongqianyu99-zhejiang-university/starVLA_rl_games/i4aobzgd/requirements.txt"
76
+ }
77
+ },
78
+ {
79
+ "node": {
80
+ "name": "wandb-metadata.json",
81
+ "sizeBytes": 2993,
82
+ "url": "https://api.wandb.ai/files/dongqianyu99-zhejiang-university/starVLA_rl_games/i4aobzgd/wandb-metadata.json"
83
+ }
84
+ },
85
+ {
86
+ "node": {
87
+ "name": "wandb-summary.json",
88
+ "sizeBytes": 52580,
89
+ "url": "https://api.wandb.ai/files/dongqianyu99-zhejiang-university/starVLA_rl_games/i4aobzgd/wandb-summary.json"
90
+ }
91
+ }
92
+ ]
93
+ }
94
+ }
95
+ }
96
+ }
97
+ }
experiments/latency-transfer/provenance/latency-transfer-data/provenance/source_inventory.json ADDED
@@ -0,0 +1,642 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [
2
+ {
3
+ "batch": "0",
4
+ "type": "model",
5
+ "repo": "latency-sensitive-bench/wan_oft_flappy_fix_latency_1_context5_standard_sft_2000_effbs128_224_currentce",
6
+ "revision": "8cf964f252337912a8f8e5d30467a33f4e8a49dc",
7
+ "files": 17,
8
+ "bytes": 23474220474
9
+ },
10
+ {
11
+ "batch": "0",
12
+ "type": "dataset",
13
+ "repo": "latency-sensitive-bench/flappy_fix_latency_0_200ep_context5",
14
+ "revision": "f5f82fb2c760bef7ddb8882dbba18075fa1cf2c8",
15
+ "files": 5,
16
+ "bytes": 10602052692
17
+ },
18
+ {
19
+ "batch": "0",
20
+ "type": "dataset",
21
+ "repo": "latency-sensitive-bench/flappy_fix_latency_1_200ep_context5",
22
+ "revision": "8ee61cea50b7c3fcad97f4ab2f41f97044d219a8",
23
+ "files": 5,
24
+ "bytes": 10841997219
25
+ },
26
+ {
27
+ "batch": "1",
28
+ "type": "model",
29
+ "repo": "latency-sensitive-bench/openvla_flappy_fix_latency_0",
30
+ "revision": "06f64d1a03905ad0f04a447c57afb970c29a559e",
31
+ "files": 12,
32
+ "bytes": 18277031882
33
+ },
34
+ {
35
+ "batch": "1",
36
+ "type": "model",
37
+ "repo": "latency-sensitive-bench/openvla_flappy_fix_latency_6",
38
+ "revision": "7a2e2f81dda9fc9de068fa15b1eac442ed6fb51f",
39
+ "files": 12,
40
+ "bytes": 18277056314
41
+ },
42
+ {
43
+ "batch": "1",
44
+ "type": "model",
45
+ "repo": "latency-sensitive-bench/openvla_flappy_fix_latency_2",
46
+ "revision": "8135db54520527abe4a62f002323a307d1a47c2d",
47
+ "files": 12,
48
+ "bytes": 18277063818
49
+ },
50
+ {
51
+ "batch": "1",
52
+ "type": "model",
53
+ "repo": "latency-sensitive-bench/openvla_flappy_fix_latency_4_exp",
54
+ "revision": "213bab6c06b13c73fee875412d99e40ea4a4dd70",
55
+ "files": 12,
56
+ "bytes": 18276903161
57
+ },
58
+ {
59
+ "batch": "1",
60
+ "type": "model",
61
+ "repo": "latency-sensitive-bench/openvla_flappy_fix_latency_4_small",
62
+ "revision": "371074f07545d1d4e5533afc4d457a63d03bf9da",
63
+ "files": 13,
64
+ "bytes": 18276621481
65
+ },
66
+ {
67
+ "batch": "1",
68
+ "type": "model",
69
+ "repo": "latency-sensitive-bench/openvla_flappy_fix_latency_1_small_tmp",
70
+ "revision": "8fba8c350f6f10d70119807a546218cfa82e9150",
71
+ "files": 13,
72
+ "bytes": 18276635611
73
+ },
74
+ {
75
+ "batch": "1",
76
+ "type": "model",
77
+ "repo": "latency-sensitive-bench/openvla_flappy_fix_latency_0_small",
78
+ "revision": "05586f38706727534a0f388ae0ece46541776d65",
79
+ "files": 13,
80
+ "bytes": 18276617707
81
+ },
82
+ {
83
+ "batch": "1",
84
+ "type": "model",
85
+ "repo": "latency-sensitive-bench/openvla_flappy_fix_latency_1_small",
86
+ "revision": "7842a04e6ba8ed4dc56ccdae235735a807a77b5d",
87
+ "files": 13,
88
+ "bytes": 18276610436
89
+ },
90
+ {
91
+ "batch": "1",
92
+ "type": "model",
93
+ "repo": "latency-sensitive-bench/openvla_flappy_fix_latency_2_small",
94
+ "revision": "87c38afed1a227e92aac9efeb0921f57d556df8a",
95
+ "files": 13,
96
+ "bytes": 18276621420
97
+ },
98
+ {
99
+ "batch": "1",
100
+ "type": "model",
101
+ "repo": "latency-sensitive-bench/openvla_flappy_fix_latency_3_small",
102
+ "revision": "86b4d314c377420fa8eb9ac7aa268695c2ab4f40",
103
+ "files": 13,
104
+ "bytes": 18276629117
105
+ },
106
+ {
107
+ "batch": "1",
108
+ "type": "model",
109
+ "repo": "latency-sensitive-bench/openvla_deadly_corridor_fix_latency_0_full",
110
+ "revision": "9281fd061894c14d3641ee9724e108cd33baf494",
111
+ "files": 13,
112
+ "bytes": 18276748150
113
+ },
114
+ {
115
+ "batch": "1",
116
+ "type": "model",
117
+ "repo": "latency-sensitive-bench/openvla_deadly_corridor_fix_latency_0",
118
+ "revision": "9f8c6842151405a82a3e852d2a00e27885db1e93",
119
+ "files": 13,
120
+ "bytes": 18276733726
121
+ },
122
+ {
123
+ "batch": "1",
124
+ "type": "model",
125
+ "repo": "latency-sensitive-bench/openvla_deadly_corridor_fix_latency_2",
126
+ "revision": "0a60d6c8a400b48e4d9737b36ef3a44814d52d2a",
127
+ "files": 13,
128
+ "bytes": 18276730845
129
+ },
130
+ {
131
+ "batch": "1",
132
+ "type": "model",
133
+ "repo": "latency-sensitive-bench/openvla_demon_attack_fix_latency_0_small",
134
+ "revision": "aeefbbd9676d95973aea2a6f63c433807a44d05b",
135
+ "files": 11,
136
+ "bytes": 9138432714
137
+ },
138
+ {
139
+ "batch": "1",
140
+ "type": "model",
141
+ "repo": "latency-sensitive-bench/openvla_demon_attack_fix_latency_2_small",
142
+ "revision": "bb91ae40a1bb9b9f6a93be031193de603da86ecd",
143
+ "files": 13,
144
+ "bytes": 18276668819
145
+ },
146
+ {
147
+ "batch": "1",
148
+ "type": "model",
149
+ "repo": "latency-sensitive-bench/openvla_demon_attack_fix_latency_6_small",
150
+ "revision": "9e8e2e6cc073eb68bd59306ab1141972ed1ed819",
151
+ "files": 1,
152
+ "bytes": 1519
153
+ },
154
+ {
155
+ "batch": "1",
156
+ "type": "model",
157
+ "repo": "latency-sensitive-bench/openvla_flappy_200ep",
158
+ "revision": "7918bf5c0aeb830a01352f5378acc6a4d4cf601c",
159
+ "files": 71,
160
+ "bytes": 63981635137
161
+ },
162
+ {
163
+ "batch": "1",
164
+ "type": "model",
165
+ "repo": "latency-sensitive-bench/openvla_demon_attack_200ep",
166
+ "revision": "04fece441aa5b6f5a6a015264a6aa5df3dc4c29a",
167
+ "files": 51,
168
+ "bytes": 45705520630
169
+ },
170
+ {
171
+ "batch": "1",
172
+ "type": "dataset",
173
+ "repo": "latency-sensitive-bench/demon_attack_fix_latency_2",
174
+ "revision": "c0605fa1df28b19249849c3e748bbc445bef6526",
175
+ "files": 5,
176
+ "bytes": 257552608
177
+ },
178
+ {
179
+ "batch": "1",
180
+ "type": "dataset",
181
+ "repo": "latency-sensitive-bench/demon_attack_fix_latency_1",
182
+ "revision": "7190bf45dfface92ec7cbaa3c774a7e29616db52",
183
+ "files": 5,
184
+ "bytes": 256670462
185
+ },
186
+ {
187
+ "batch": "1",
188
+ "type": "dataset",
189
+ "repo": "latency-sensitive-bench/demon_attack_fix_latency_0",
190
+ "revision": "2deacb435543c52ae51f3c947bf917fdb0b5493c",
191
+ "files": 5,
192
+ "bytes": 254545643
193
+ },
194
+ {
195
+ "batch": "1",
196
+ "type": "dataset",
197
+ "repo": "latency-sensitive-bench/demon_attack_fix_latency_3",
198
+ "revision": "709111a3b59e413214e0417bd878291734ec9427",
199
+ "files": 5,
200
+ "bytes": 251773874
201
+ },
202
+ {
203
+ "batch": "1",
204
+ "type": "dataset",
205
+ "repo": "latency-sensitive-bench/demon_attack_fix_latency_4",
206
+ "revision": "c7f6eba84f512bc22a095971c9b75499e841a026",
207
+ "files": 5,
208
+ "bytes": 234519256
209
+ },
210
+ {
211
+ "batch": "1",
212
+ "type": "dataset",
213
+ "repo": "latency-sensitive-bench/demon_attack_fix_latency_5",
214
+ "revision": "67876ea9a8d5c687b924199574e298c68bf9b83d",
215
+ "files": 5,
216
+ "bytes": 232364317
217
+ },
218
+ {
219
+ "batch": "1",
220
+ "type": "dataset",
221
+ "repo": "latency-sensitive-bench/demon_attack_fix_latency_7",
222
+ "revision": "d76a57ef1b84c591c47975347c539da59670f3b4",
223
+ "files": 5,
224
+ "bytes": 205595492
225
+ },
226
+ {
227
+ "batch": "1",
228
+ "type": "dataset",
229
+ "repo": "latency-sensitive-bench/demon_attack_fix_latency_6",
230
+ "revision": "0766ba4ea9196b861302e99e564b7d2e05453885",
231
+ "files": 5,
232
+ "bytes": 200941670
233
+ },
234
+ {
235
+ "batch": "1",
236
+ "type": "dataset",
237
+ "repo": "latency-sensitive-bench/demon_attack_fix_latency_8",
238
+ "revision": "e02c6b1db39b37965569d4f559f1d710fb55c58a",
239
+ "files": 5,
240
+ "bytes": 147343953
241
+ },
242
+ {
243
+ "batch": "1",
244
+ "type": "dataset",
245
+ "repo": "latency-sensitive-bench/demon_attack_fix_latency_12",
246
+ "revision": "ac9d21a59b908e717de2bee2cc749b90495a00b8",
247
+ "files": 5,
248
+ "bytes": 147328521
249
+ },
250
+ {
251
+ "batch": "1",
252
+ "type": "dataset",
253
+ "repo": "latency-sensitive-bench/demon_attack_fix_latency_9",
254
+ "revision": "7be93d38340141e385913e540b028024d74f1cf8",
255
+ "files": 5,
256
+ "bytes": 140841771
257
+ },
258
+ {
259
+ "batch": "1",
260
+ "type": "dataset",
261
+ "repo": "latency-sensitive-bench/demon_attack_fix_latency_13",
262
+ "revision": "2a5c5d2aa4730db22fd33f4f00f4d93fc5043465",
263
+ "files": 5,
264
+ "bytes": 133539812
265
+ },
266
+ {
267
+ "batch": "1",
268
+ "type": "dataset",
269
+ "repo": "latency-sensitive-bench/demon_attack_fix_latency_14",
270
+ "revision": "9867d03ad02a4d025cb6cc88281f95596f83a814",
271
+ "files": 5,
272
+ "bytes": 128938837
273
+ },
274
+ {
275
+ "batch": "1",
276
+ "type": "dataset",
277
+ "repo": "latency-sensitive-bench/demon_attack_fix_latency_10",
278
+ "revision": "9a3f7f736402b063885068e683860b6746e5337d",
279
+ "files": 5,
280
+ "bytes": 128832856
281
+ },
282
+ {
283
+ "batch": "1",
284
+ "type": "dataset",
285
+ "repo": "latency-sensitive-bench/demon_attack_fix_latency_15",
286
+ "revision": "21811c6039f0a538f11821730ee6792bd0d6c0b4",
287
+ "files": 5,
288
+ "bytes": 110909215
289
+ },
290
+ {
291
+ "batch": "1",
292
+ "type": "dataset",
293
+ "repo": "latency-sensitive-bench/demon_attack_fix_latency_11",
294
+ "revision": "78e4e1b6d9bb57c5f30712cd0ed2f3b02877ff56",
295
+ "files": 5,
296
+ "bytes": 91939207
297
+ },
298
+ {
299
+ "batch": "1",
300
+ "type": "dataset",
301
+ "repo": "latency-sensitive-bench/deadly_corridor_fix_latency_0",
302
+ "revision": "6ee54a31fe5645bce299adeab17a68ff458bfabe",
303
+ "files": 5,
304
+ "bytes": 67749647
305
+ },
306
+ {
307
+ "batch": "1",
308
+ "type": "dataset",
309
+ "repo": "latency-sensitive-bench/deadly_corridor_fix_latency_12",
310
+ "revision": "258c4ae285ad34445578dea4e786c408fcd6be30",
311
+ "files": 5,
312
+ "bytes": 44793589
313
+ },
314
+ {
315
+ "batch": "1",
316
+ "type": "dataset",
317
+ "repo": "latency-sensitive-bench/deadly_corridor_fix_latency_4",
318
+ "revision": "4dee6c0fd4abb86507001c4dc7404a5310c2f30a",
319
+ "files": 5,
320
+ "bytes": 43805390
321
+ },
322
+ {
323
+ "batch": "1",
324
+ "type": "dataset",
325
+ "repo": "latency-sensitive-bench/deadly_corridor_fix_latency_13",
326
+ "revision": "dbb7678df6396735290099d070dfb5827a4cf4a4",
327
+ "files": 5,
328
+ "bytes": 43616311
329
+ },
330
+ {
331
+ "batch": "1",
332
+ "type": "dataset",
333
+ "repo": "latency-sensitive-bench/deadly_corridor_fix_latency_2",
334
+ "revision": "df04b6f30856b2903e65a6a870f7440055f02311",
335
+ "files": 5,
336
+ "bytes": 42670186
337
+ },
338
+ {
339
+ "batch": "1",
340
+ "type": "dataset",
341
+ "repo": "latency-sensitive-bench/deadly_corridor_fix_latency_14",
342
+ "revision": "fb78a1bc6280a4d0a6a60e8231d63d848819bd1e",
343
+ "files": 5,
344
+ "bytes": 40543661
345
+ },
346
+ {
347
+ "batch": "1",
348
+ "type": "dataset",
349
+ "repo": "latency-sensitive-bench/deadly_corridor_fix_latency_5",
350
+ "revision": "f614a91258d86cb34cd2e3e0e79e2bf69b321e76",
351
+ "files": 5,
352
+ "bytes": 40279413
353
+ },
354
+ {
355
+ "batch": "1",
356
+ "type": "dataset",
357
+ "repo": "latency-sensitive-bench/deadly_corridor_fix_latency_7",
358
+ "revision": "d04b9bc42bfbb7da400438a9fb75c49b198f7776",
359
+ "files": 5,
360
+ "bytes": 38892941
361
+ },
362
+ {
363
+ "batch": "1",
364
+ "type": "dataset",
365
+ "repo": "latency-sensitive-bench/deadly_corridor_fix_latency_1",
366
+ "revision": "e09b7f2c51726bb169d827fe973b2d429b6b1432",
367
+ "files": 5,
368
+ "bytes": 36658578
369
+ },
370
+ {
371
+ "batch": "1",
372
+ "type": "dataset",
373
+ "repo": "latency-sensitive-bench/deadly_corridor_fix_latency_10",
374
+ "revision": "94762ebd79edb55d6e83ba2046729e3e6574d64d",
375
+ "files": 5,
376
+ "bytes": 36560892
377
+ },
378
+ {
379
+ "batch": "1",
380
+ "type": "dataset",
381
+ "repo": "latency-sensitive-bench/deadly_corridor_fix_latency_15",
382
+ "revision": "c586ddd84f76b6295de7929b62ad4da3359b0348",
383
+ "files": 5,
384
+ "bytes": 36311418
385
+ },
386
+ {
387
+ "batch": "1",
388
+ "type": "dataset",
389
+ "repo": "latency-sensitive-bench/deadly_corridor_fix_latency_6",
390
+ "revision": "47cfced1713ce7d57d4567110269fb5929e8e7a8",
391
+ "files": 5,
392
+ "bytes": 36072348
393
+ },
394
+ {
395
+ "batch": "1",
396
+ "type": "dataset",
397
+ "repo": "latency-sensitive-bench/deadly_corridor_fix_latency_9",
398
+ "revision": "3405ac7c12ca266e2d3e83047bbd06e1e5eed220",
399
+ "files": 5,
400
+ "bytes": 32290799
401
+ },
402
+ {
403
+ "batch": "1",
404
+ "type": "dataset",
405
+ "repo": "latency-sensitive-bench/deadly_corridor_fix_latency_3",
406
+ "revision": "b8b18213fe10aa0a6be2b7602e23c2ac938b49e0",
407
+ "files": 5,
408
+ "bytes": 31980248
409
+ },
410
+ {
411
+ "batch": "1",
412
+ "type": "dataset",
413
+ "repo": "latency-sensitive-bench/deadly_corridor_fix_latency_8",
414
+ "revision": "164f4e56abb6c2637801e779918a99a7d70c9c3a",
415
+ "files": 5,
416
+ "bytes": 29960490
417
+ },
418
+ {
419
+ "batch": "1",
420
+ "type": "dataset",
421
+ "repo": "latency-sensitive-bench/flappy_fix_latency_3_small",
422
+ "revision": "8470ae1f00ca10bb8646ac7cfdccf9682e053eab",
423
+ "files": 7,
424
+ "bytes": 1272008595
425
+ },
426
+ {
427
+ "batch": "1",
428
+ "type": "dataset",
429
+ "repo": "latency-sensitive-bench/flappy_fix_latency_2_small",
430
+ "revision": "c9bbd1744020a587b55590c897628510e0741136",
431
+ "files": 7,
432
+ "bytes": 1271812775
433
+ },
434
+ {
435
+ "batch": "1",
436
+ "type": "dataset",
437
+ "repo": "latency-sensitive-bench/flappy_fix_latency_1_small",
438
+ "revision": "749a849911b8c35005c26241fb4edbcee1286832",
439
+ "files": 7,
440
+ "bytes": 1266682400
441
+ },
442
+ {
443
+ "batch": "1",
444
+ "type": "dataset",
445
+ "repo": "latency-sensitive-bench/flappy_fix_latency_0_small",
446
+ "revision": "7e974f1e8e1ad7ca4ee0ffad0849d886d64128d3",
447
+ "files": 6,
448
+ "bytes": 1230559183
449
+ },
450
+ {
451
+ "batch": "1",
452
+ "type": "dataset",
453
+ "repo": "latency-sensitive-bench/flappy_fix_latency_4_small",
454
+ "revision": "8a6a091570530476b313976b6a8d1b94d600f6d9",
455
+ "files": 6,
456
+ "bytes": 1211322894
457
+ },
458
+ {
459
+ "batch": "1",
460
+ "type": "dataset",
461
+ "repo": "latency-sensitive-bench/flappy_200ep",
462
+ "revision": "5df5688a961dc2c91090155c719bff4a5b248235",
463
+ "files": 22,
464
+ "bytes": 21680187271
465
+ },
466
+ {
467
+ "batch": "1",
468
+ "type": "dataset",
469
+ "repo": "latency-sensitive-bench/demon_attack_200ep",
470
+ "revision": "0481972d94f6409a392c5b035f9a7f5c2681a96d",
471
+ "files": 21,
472
+ "bytes": 1148999664
473
+ },
474
+ {
475
+ "batch": "2",
476
+ "type": "model",
477
+ "repo": "latency-sensitive-bench/openvla_bridge_flappy_fixed_latency_3_200ep_7k2steps_ghost7_exp3",
478
+ "revision": "5dfd90884fc4e3817bc51f574d6fc9cea2d7948b",
479
+ "files": 8,
480
+ "bytes": 9784919229
481
+ },
482
+ {
483
+ "batch": "2",
484
+ "type": "model",
485
+ "repo": "latency-sensitive-bench/openvla_bridge_demon_attack_fixed_latency_6_200ep_7k2steps_ghost15_2_exp3",
486
+ "revision": "f8d64aff21628adbdd0d1aaddd1bc7067bc2a23c",
487
+ "files": 17,
488
+ "bytes": 9784948330
489
+ },
490
+ {
491
+ "batch": "2",
492
+ "type": "model",
493
+ "repo": "latency-sensitive-bench/openvla_bridge_flappy_fixed_latency_3_200ep_7k2steps_ghost15",
494
+ "revision": "9915dcb0dc644aa2ed14172f298386f53b03c6f6",
495
+ "files": 8,
496
+ "bytes": 9138248018
497
+ },
498
+ {
499
+ "batch": "2",
500
+ "type": "model",
501
+ "repo": "latency-sensitive-bench/openvla_bridge_demon_attack_fixed_latency_6_200ep_7k2steps_ghost15",
502
+ "revision": "cd810024c06f961c95449785f19ab73ccfd20de1",
503
+ "files": 13,
504
+ "bytes": 9138255732
505
+ },
506
+ {
507
+ "batch": "2",
508
+ "type": "model",
509
+ "repo": "latency-sensitive-bench/openvla_bridge_flappy_fixed_latency_3_200ep_7k2steps_ghost15_exp2",
510
+ "revision": "73ea97c064d206c3a0dcf6ffdbf2ecfe97bc2c3c",
511
+ "files": 8,
512
+ "bytes": 9138247431
513
+ },
514
+ {
515
+ "batch": "2",
516
+ "type": "model",
517
+ "repo": "latency-sensitive-bench/openvla_bridge_demon_attack_fixed_latency_6_200ep_7k2steps_ghost15_exp2",
518
+ "revision": "c4cbca936a122c2d55b1edee3959656f8174c5b7",
519
+ "files": 13,
520
+ "bytes": 9138255777
521
+ },
522
+ {
523
+ "batch": "2",
524
+ "type": "model",
525
+ "repo": "latency-sensitive-bench/openvla_bridge_flappy_fixed_latency_3_200ep_7k2steps_ghost15_seed1",
526
+ "revision": "004ec237af79c10de5c0d4376cf8a13542ac01f8",
527
+ "files": 8,
528
+ "bytes": 9138247433
529
+ },
530
+ {
531
+ "batch": "2",
532
+ "type": "model",
533
+ "repo": "latency-sensitive-bench/openvla_bridge_demon_attack_fixed_latency_6_200ep_7k2steps_ghost15_seed1",
534
+ "revision": "bc733d64e32512a8fc76f9be022eb5b0d7c0e2cb",
535
+ "files": 13,
536
+ "bytes": 9138255750
537
+ },
538
+ {
539
+ "batch": "2",
540
+ "type": "model",
541
+ "repo": "latency-sensitive-bench/openvla_bridge_flappy_fixed_latency_3_200ep_7k2steps_ghost7",
542
+ "revision": "94014d226e26fcf46f117f6bea42a59b0be2d753",
543
+ "files": 8,
544
+ "bytes": 9138247812
545
+ },
546
+ {
547
+ "batch": "2",
548
+ "type": "model",
549
+ "repo": "latency-sensitive-bench/openvla_bridge_demon_attack_fixed_latency_6_200ep_7k2steps_ghost15_2",
550
+ "revision": "ac5ff1109c46295299d74b5e30a56f1e27e930a6",
551
+ "files": 5,
552
+ "bytes": 14628
553
+ },
554
+ {
555
+ "batch": "2",
556
+ "type": "dataset",
557
+ "repo": "latency-sensitive-bench/flappy_fixed_latency_3_200ep_7k2steps_ghost15",
558
+ "revision": "f90d1dac8b31a18c0496884729c744d7a4c877ae",
559
+ "files": 4,
560
+ "bytes": 25946174934
561
+ },
562
+ {
563
+ "batch": "2",
564
+ "type": "dataset",
565
+ "repo": "latency-sensitive-bench/demon_attack_fixed_latency_6_200ep_7k2steps_ghost15",
566
+ "revision": "862b6cad9b2c87dc129d5cf45a9b23b18450c4ef",
567
+ "files": 6,
568
+ "bytes": 993634330
569
+ },
570
+ {
571
+ "batch": "3",
572
+ "type": "model",
573
+ "repo": "latency-sensitive-bench/openvla_flappy_l2_full",
574
+ "revision": "2e8e58c6af1ee6f8a3701388b2552a4fa65a041c",
575
+ "files": 6,
576
+ "bytes": 9805979462
577
+ },
578
+ {
579
+ "batch": "3",
580
+ "type": "model",
581
+ "repo": "latency-sensitive-bench/openvla_flappy_l2_lora",
582
+ "revision": "427f4c0582d6eb9419fa39f5942db1d59585993c",
583
+ "files": 9,
584
+ "bytes": 292406543
585
+ },
586
+ {
587
+ "batch": "3",
588
+ "type": "dataset",
589
+ "repo": "latency-sensitive-bench/flappy_fixed_l2_fs4",
590
+ "revision": "1bdd2ea6ea7db07139e4694f445c12198399c2b8",
591
+ "files": 5,
592
+ "bytes": 11243685412
593
+ },
594
+ {
595
+ "batch": "3",
596
+ "type": "dataset",
597
+ "repo": "latency-sensitive-bench/lora_comparison_flappy_l2_eval",
598
+ "revision": "97db426f51a9427fbe1e8c005b33dbe7aeac3899",
599
+ "files": 6,
600
+ "bytes": 113503
601
+ },
602
+ {
603
+ "batch": "4",
604
+ "type": "model",
605
+ "repo": "latency-sensitive-bench/openvla_inverted_pendulum_mixed_latency_024_og_l0_native_exp2",
606
+ "revision": "177c11a86ee64432264ae8f1e577ecfd6c767918",
607
+ "files": 40,
608
+ "bytes": 9152588192
609
+ },
610
+ {
611
+ "batch": "4",
612
+ "type": "model",
613
+ "repo": "latency-sensitive-bench/openvla_inverted_pendulum_mixed_latency_024_og_l0_native_exp3",
614
+ "revision": "1967aac3fadf905a7f693e388cbf2aebdd40e5de",
615
+ "files": 40,
616
+ "bytes": 9154482526
617
+ },
618
+ {
619
+ "batch": "4",
620
+ "type": "model",
621
+ "repo": "latency-sensitive-bench/openvla_inverted_pendulum_mixed_latency_024_og_l0_native_no_latency_prompt_exp2",
622
+ "revision": "bd6274f20b7ff2f36c0b758058ef8ce2e996d7f6",
623
+ "files": 56,
624
+ "bytes": 9152597509
625
+ },
626
+ {
627
+ "batch": "4",
628
+ "type": "dataset",
629
+ "repo": "latency-sensitive-bench/inverted_pendulum_200ep",
630
+ "revision": "5dcc48f5f713386b9ee2c2339f1c364b76246956",
631
+ "files": 29,
632
+ "bytes": 10489281275
633
+ },
634
+ {
635
+ "batch": "4",
636
+ "type": "dataset",
637
+ "repo": "latency-sensitive-bench/inverted_pendulum_200ep_2",
638
+ "revision": "eebf63d9082f2f2f3a3672cca64f7b7776750b54",
639
+ "files": 11,
640
+ "bytes": 1134507619
641
+ }
642
+ ]
experiments/latency-transfer/provenance/latency-transfer-data/provenance/source_repositories/dataset/deadly_corridor_fix_latency_0.json ADDED
@@ -0,0 +1,150 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_id": "6a05618a8af659353a061f5e",
3
+ "id": "latency-sensitive-bench/deadly_corridor_fix_latency_0",
4
+ "author": "latency-sensitive-bench",
5
+ "sha": "6ee54a31fe5645bce299adeab17a68ff458bfabe",
6
+ "lastModified": "2026-05-14T05:48:13.000Z",
7
+ "private": false,
8
+ "gated": false,
9
+ "disabled": false,
10
+ "tags": [
11
+ "size_categories:1K<n<10K",
12
+ "format:parquet",
13
+ "format:optimized-parquet",
14
+ "modality:image",
15
+ "modality:text",
16
+ "library:datasets",
17
+ "library:pandas",
18
+ "library:polars",
19
+ "library:mlcroissant",
20
+ "region:us"
21
+ ],
22
+ "downloads": 8,
23
+ "likes": 0,
24
+ "cardData": {
25
+ "dataset_info": {
26
+ "features": [
27
+ {
28
+ "name": "image",
29
+ "dtype": "image"
30
+ },
31
+ {
32
+ "name": "split",
33
+ "dtype": "string"
34
+ },
35
+ {
36
+ "name": "prompt",
37
+ "dtype": "string"
38
+ },
39
+ {
40
+ "name": "action_id",
41
+ "dtype": "int64"
42
+ },
43
+ {
44
+ "name": "action_text",
45
+ "dtype": "string"
46
+ },
47
+ {
48
+ "name": "latency",
49
+ "dtype": "int64"
50
+ },
51
+ {
52
+ "name": "latency_ms",
53
+ "dtype": "float64"
54
+ },
55
+ {
56
+ "name": "reward",
57
+ "dtype": "float64"
58
+ },
59
+ {
60
+ "name": "done",
61
+ "dtype": "bool"
62
+ },
63
+ {
64
+ "name": "episode_idx",
65
+ "dtype": "int64"
66
+ },
67
+ {
68
+ "name": "t",
69
+ "dtype": "int64"
70
+ },
71
+ {
72
+ "name": "seed",
73
+ "dtype": "int64"
74
+ },
75
+ {
76
+ "name": "deterministic",
77
+ "dtype": "bool"
78
+ }
79
+ ],
80
+ "splits": [
81
+ {
82
+ "name": "train",
83
+ "num_bytes": 65115523,
84
+ "num_examples": 3836
85
+ },
86
+ {
87
+ "name": "val",
88
+ "num_bytes": 7227473,
89
+ "num_examples": 443
90
+ }
91
+ ],
92
+ "download_size": 67742624,
93
+ "dataset_size": 72342996
94
+ },
95
+ "configs": [
96
+ {
97
+ "config_name": "default",
98
+ "data_files": [
99
+ {
100
+ "split": "train",
101
+ "path": "data/train-*"
102
+ },
103
+ {
104
+ "split": "val",
105
+ "path": "data/val-*"
106
+ }
107
+ ]
108
+ }
109
+ ]
110
+ },
111
+ "siblings": [
112
+ {
113
+ "rfilename": ".gitattributes",
114
+ "blobId": "bed0738c7eeb449bca98b5d2f33c89a1ee56349a",
115
+ "size": 2504
116
+ },
117
+ {
118
+ "rfilename": "README.md",
119
+ "blobId": "bf3216eb86024e0b898f3d6e7f87dd07a9fe75b3",
120
+ "size": 800
121
+ },
122
+ {
123
+ "rfilename": "data/train-00000-of-00001.parquet",
124
+ "blobId": "961181a2c89a042abc215d0415b3520ba6ce617b",
125
+ "size": 60683500,
126
+ "lfs": {
127
+ "sha256": "c7e5c9a1b249cce02b696b25d7069c094ff14d54b73da87b49a5b74f98115c03",
128
+ "size": 60683500,
129
+ "pointerSize": 133
130
+ }
131
+ },
132
+ {
133
+ "rfilename": "data/val-00000-of-00001.parquet",
134
+ "blobId": "59ead584f0a93dd90346a222694e1453040318f5",
135
+ "size": 7059124,
136
+ "lfs": {
137
+ "sha256": "0931d75b4b291662663d0703150915e6f7e60408c7e836e7dc3358df8a7d3a6b",
138
+ "size": 7059124,
139
+ "pointerSize": 132
140
+ }
141
+ },
142
+ {
143
+ "rfilename": "metadata.json",
144
+ "blobId": "2022590c269191ca18931126ed2bcad3358caa01",
145
+ "size": 3719
146
+ }
147
+ ],
148
+ "createdAt": "2026-05-14T05:45:46.000Z",
149
+ "usedStorage": 67742624
150
+ }
experiments/latency-transfer/provenance/latency-transfer-data/provenance/source_repositories/dataset/deadly_corridor_fix_latency_1.json ADDED
@@ -0,0 +1,150 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_id": "6a0561635d4211d1141405d6",
3
+ "id": "latency-sensitive-bench/deadly_corridor_fix_latency_1",
4
+ "author": "latency-sensitive-bench",
5
+ "sha": "e09b7f2c51726bb169d827fe973b2d429b6b1432",
6
+ "lastModified": "2026-05-14T05:46:40.000Z",
7
+ "private": false,
8
+ "gated": false,
9
+ "disabled": false,
10
+ "tags": [
11
+ "size_categories:1K<n<10K",
12
+ "format:parquet",
13
+ "format:optimized-parquet",
14
+ "modality:image",
15
+ "modality:text",
16
+ "library:datasets",
17
+ "library:pandas",
18
+ "library:polars",
19
+ "library:mlcroissant",
20
+ "region:us"
21
+ ],
22
+ "downloads": 27,
23
+ "likes": 0,
24
+ "cardData": {
25
+ "dataset_info": {
26
+ "features": [
27
+ {
28
+ "name": "image",
29
+ "dtype": "image"
30
+ },
31
+ {
32
+ "name": "split",
33
+ "dtype": "string"
34
+ },
35
+ {
36
+ "name": "prompt",
37
+ "dtype": "string"
38
+ },
39
+ {
40
+ "name": "action_id",
41
+ "dtype": "int64"
42
+ },
43
+ {
44
+ "name": "action_text",
45
+ "dtype": "string"
46
+ },
47
+ {
48
+ "name": "latency",
49
+ "dtype": "int64"
50
+ },
51
+ {
52
+ "name": "latency_ms",
53
+ "dtype": "float64"
54
+ },
55
+ {
56
+ "name": "reward",
57
+ "dtype": "float64"
58
+ },
59
+ {
60
+ "name": "done",
61
+ "dtype": "bool"
62
+ },
63
+ {
64
+ "name": "episode_idx",
65
+ "dtype": "int64"
66
+ },
67
+ {
68
+ "name": "t",
69
+ "dtype": "int64"
70
+ },
71
+ {
72
+ "name": "seed",
73
+ "dtype": "int64"
74
+ },
75
+ {
76
+ "name": "deterministic",
77
+ "dtype": "bool"
78
+ }
79
+ ],
80
+ "splits": [
81
+ {
82
+ "name": "train",
83
+ "num_bytes": 34549455,
84
+ "num_examples": 1966
85
+ },
86
+ {
87
+ "name": "val",
88
+ "num_bytes": 3750078,
89
+ "num_examples": 218
90
+ }
91
+ ],
92
+ "download_size": 36651540,
93
+ "dataset_size": 38299533
94
+ },
95
+ "configs": [
96
+ {
97
+ "config_name": "default",
98
+ "data_files": [
99
+ {
100
+ "split": "train",
101
+ "path": "data/train-*"
102
+ },
103
+ {
104
+ "split": "val",
105
+ "path": "data/val-*"
106
+ }
107
+ ]
108
+ }
109
+ ]
110
+ },
111
+ "siblings": [
112
+ {
113
+ "rfilename": ".gitattributes",
114
+ "blobId": "bed0738c7eeb449bca98b5d2f33c89a1ee56349a",
115
+ "size": 2504
116
+ },
117
+ {
118
+ "rfilename": "README.md",
119
+ "blobId": "069b579669d0b972c9328290e3d3ee347e392a21",
120
+ "size": 800
121
+ },
122
+ {
123
+ "rfilename": "data/train-00000-of-00001.parquet",
124
+ "blobId": "ceb53e858c5fe009e9f549744cbedb7ad0a392bb",
125
+ "size": 32981127,
126
+ "lfs": {
127
+ "sha256": "893c4234419bcce925c4e739d97c20b411fb1bf2cbbf30d085009195cefe86a2",
128
+ "size": 32981127,
129
+ "pointerSize": 133
130
+ }
131
+ },
132
+ {
133
+ "rfilename": "data/val-00000-of-00001.parquet",
134
+ "blobId": "abb07d2091815f8c22b5d4e82e6a17401c0646bd",
135
+ "size": 3670413,
136
+ "lfs": {
137
+ "sha256": "ab8353529f419e970833e4e87ca3c0ccad5dd3e679e461e3472eccadc943b729",
138
+ "size": 3670413,
139
+ "pointerSize": 132
140
+ }
141
+ },
142
+ {
143
+ "rfilename": "metadata.json",
144
+ "blobId": "1a391ee63a9e65cd3931d7326e98bacb7b8290c4",
145
+ "size": 3734
146
+ }
147
+ ],
148
+ "createdAt": "2026-05-14T05:45:07.000Z",
149
+ "usedStorage": 36651540
150
+ }
experiments/latency-transfer/provenance/latency-transfer-data/provenance/source_repositories/dataset/deadly_corridor_fix_latency_10.json ADDED
@@ -0,0 +1,150 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_id": "6a05729665cede072752e989",
3
+ "id": "latency-sensitive-bench/deadly_corridor_fix_latency_10",
4
+ "author": "latency-sensitive-bench",
5
+ "sha": "94762ebd79edb55d6e83ba2046729e3e6574d64d",
6
+ "lastModified": "2026-05-14T07:00:12.000Z",
7
+ "private": false,
8
+ "gated": false,
9
+ "disabled": false,
10
+ "tags": [
11
+ "size_categories:1K<n<10K",
12
+ "format:parquet",
13
+ "format:optimized-parquet",
14
+ "modality:image",
15
+ "modality:text",
16
+ "library:datasets",
17
+ "library:pandas",
18
+ "library:polars",
19
+ "library:mlcroissant",
20
+ "region:us"
21
+ ],
22
+ "downloads": 5,
23
+ "likes": 0,
24
+ "cardData": {
25
+ "dataset_info": {
26
+ "features": [
27
+ {
28
+ "name": "image",
29
+ "dtype": "image"
30
+ },
31
+ {
32
+ "name": "split",
33
+ "dtype": "string"
34
+ },
35
+ {
36
+ "name": "prompt",
37
+ "dtype": "string"
38
+ },
39
+ {
40
+ "name": "action_id",
41
+ "dtype": "int64"
42
+ },
43
+ {
44
+ "name": "action_text",
45
+ "dtype": "string"
46
+ },
47
+ {
48
+ "name": "latency",
49
+ "dtype": "int64"
50
+ },
51
+ {
52
+ "name": "latency_ms",
53
+ "dtype": "float64"
54
+ },
55
+ {
56
+ "name": "reward",
57
+ "dtype": "float64"
58
+ },
59
+ {
60
+ "name": "done",
61
+ "dtype": "bool"
62
+ },
63
+ {
64
+ "name": "episode_idx",
65
+ "dtype": "int64"
66
+ },
67
+ {
68
+ "name": "t",
69
+ "dtype": "int64"
70
+ },
71
+ {
72
+ "name": "seed",
73
+ "dtype": "int64"
74
+ },
75
+ {
76
+ "name": "deterministic",
77
+ "dtype": "bool"
78
+ }
79
+ ],
80
+ "splits": [
81
+ {
82
+ "name": "train",
83
+ "num_bytes": 33223147,
84
+ "num_examples": 1989
85
+ },
86
+ {
87
+ "name": "val",
88
+ "num_bytes": 4470488,
89
+ "num_examples": 271
90
+ }
91
+ ],
92
+ "download_size": 36553730,
93
+ "dataset_size": 37693635
94
+ },
95
+ "configs": [
96
+ {
97
+ "config_name": "default",
98
+ "data_files": [
99
+ {
100
+ "split": "train",
101
+ "path": "data/train-*"
102
+ },
103
+ {
104
+ "split": "val",
105
+ "path": "data/val-*"
106
+ }
107
+ ]
108
+ }
109
+ ]
110
+ },
111
+ "siblings": [
112
+ {
113
+ "rfilename": ".gitattributes",
114
+ "blobId": "bed0738c7eeb449bca98b5d2f33c89a1ee56349a",
115
+ "size": 2504
116
+ },
117
+ {
118
+ "rfilename": "README.md",
119
+ "blobId": "0f51aa1d25345a5d563e25367a389f7a7592b2e8",
120
+ "size": 800
121
+ },
122
+ {
123
+ "rfilename": "data/train-00000-of-00001.parquet",
124
+ "blobId": "1895fa085ac6f3fb06f0d4aedfd94ee3da0ff236",
125
+ "size": 32185015,
126
+ "lfs": {
127
+ "sha256": "be218fed0bb91c69feee31e7eb1ca2a3b6e0c15b3d96f918da26c5b73e455442",
128
+ "size": 32185015,
129
+ "pointerSize": 133
130
+ }
131
+ },
132
+ {
133
+ "rfilename": "data/val-00000-of-00001.parquet",
134
+ "blobId": "0a181ca42d6476083ab6802ed61e6c7972d346ba",
135
+ "size": 4368715,
136
+ "lfs": {
137
+ "sha256": "25854a5f15dfcdcd3405bc549f238bbe0b7af55aa15f041b82870c911a19be5c",
138
+ "size": 4368715,
139
+ "pointerSize": 132
140
+ }
141
+ },
142
+ {
143
+ "rfilename": "metadata.json",
144
+ "blobId": "4f6a0550466015ec1b326acbaf018fba0238e286",
145
+ "size": 3858
146
+ }
147
+ ],
148
+ "createdAt": "2026-05-14T06:58:30.000Z",
149
+ "usedStorage": 36553730
150
+ }