Auto-sync: 2026-06-27 10:07:28 (part 2)
Browse files
results/h16_field_sweep_summary.json
ADDED
|
@@ -0,0 +1,925 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"run_root": "/scratch/knguy52/dovla/experiments/dovla_h16_field_sweep",
|
| 3 |
+
"baseline_h4_policy_success": 0.2967,
|
| 4 |
+
"baseline_h16_policy_success": 0.29739130434782607,
|
| 5 |
+
"num_result_files": 12,
|
| 6 |
+
"configs": [
|
| 7 |
+
{
|
| 8 |
+
"config": "k32_sigma0.35",
|
| 9 |
+
"completed_seeds": [
|
| 10 |
+
0,
|
| 11 |
+
1,
|
| 12 |
+
2
|
| 13 |
+
],
|
| 14 |
+
"num_completed": 3,
|
| 15 |
+
"mean_success": 0.2910144927536232,
|
| 16 |
+
"std_success": 0.01305313652080894,
|
| 17 |
+
"mean_progress": 0.5343561867265489,
|
| 18 |
+
"mean_action_mse_to_best": 0.41566191710828654,
|
| 19 |
+
"absolute_gain_vs_h4_baseline": -0.005685507246376842,
|
| 20 |
+
"absolute_gain_vs_h16_policy": -0.006376811594202891
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"config": "k8_sigma0.10",
|
| 24 |
+
"completed_seeds": [
|
| 25 |
+
0,
|
| 26 |
+
1,
|
| 27 |
+
2
|
| 28 |
+
],
|
| 29 |
+
"num_completed": 3,
|
| 30 |
+
"mean_success": 0.28695652173913044,
|
| 31 |
+
"std_success": 0.015457729421418423,
|
| 32 |
+
"mean_progress": 0.5377366309161279,
|
| 33 |
+
"mean_action_mse_to_best": 0.4022299338993279,
|
| 34 |
+
"absolute_gain_vs_h4_baseline": -0.009743478260869576,
|
| 35 |
+
"absolute_gain_vs_h16_policy": -0.010434782608695625
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"config": "k64_sigma0.50",
|
| 39 |
+
"completed_seeds": [
|
| 40 |
+
0,
|
| 41 |
+
1,
|
| 42 |
+
2
|
| 43 |
+
],
|
| 44 |
+
"num_completed": 3,
|
| 45 |
+
"mean_success": 0.2863768115942029,
|
| 46 |
+
"std_success": 0.009889114266221424,
|
| 47 |
+
"mean_progress": 0.5288967370216688,
|
| 48 |
+
"mean_action_mse_to_best": 0.42455054165309536,
|
| 49 |
+
"absolute_gain_vs_h4_baseline": -0.010323188405797101,
|
| 50 |
+
"absolute_gain_vs_h16_policy": -0.01101449275362315
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"config": "k16_sigma0.20",
|
| 54 |
+
"completed_seeds": [
|
| 55 |
+
0,
|
| 56 |
+
1,
|
| 57 |
+
2
|
| 58 |
+
],
|
| 59 |
+
"num_completed": 3,
|
| 60 |
+
"mean_success": 0.28521739130434787,
|
| 61 |
+
"std_success": 0.01204904909613132,
|
| 62 |
+
"mean_progress": 0.5317193494629169,
|
| 63 |
+
"mean_action_mse_to_best": 0.40730522828662524,
|
| 64 |
+
"absolute_gain_vs_h4_baseline": -0.011482608695652152,
|
| 65 |
+
"absolute_gain_vs_h16_policy": -0.012173913043478202
|
| 66 |
+
}
|
| 67 |
+
],
|
| 68 |
+
"best": {
|
| 69 |
+
"config": "k32_sigma0.35",
|
| 70 |
+
"completed_seeds": [
|
| 71 |
+
0,
|
| 72 |
+
1,
|
| 73 |
+
2
|
| 74 |
+
],
|
| 75 |
+
"num_completed": 3,
|
| 76 |
+
"mean_success": 0.2910144927536232,
|
| 77 |
+
"std_success": 0.01305313652080894,
|
| 78 |
+
"mean_progress": 0.5343561867265489,
|
| 79 |
+
"mean_action_mse_to_best": 0.41566191710828654,
|
| 80 |
+
"absolute_gain_vs_h4_baseline": -0.005685507246376842,
|
| 81 |
+
"absolute_gain_vs_h16_policy": -0.006376811594202891
|
| 82 |
+
},
|
| 83 |
+
"rows": [
|
| 84 |
+
{
|
| 85 |
+
"config": "k16_sigma0.20",
|
| 86 |
+
"seed": 0,
|
| 87 |
+
"path": "/scratch/knguy52/dovla/experiments/dovla_h16_field_sweep/k16_sigma0.20/seed_0/online_rollout.json",
|
| 88 |
+
"selection_mode": "field",
|
| 89 |
+
"num_candidates": 16,
|
| 90 |
+
"candidate_sigma": 0.2,
|
| 91 |
+
"num_groups": 575,
|
| 92 |
+
"policy_rollout_success_rate": 0.271304347826087,
|
| 93 |
+
"policy_rollout_progress": 0.5181417734661828,
|
| 94 |
+
"oracle_success_rate": 0.8573913043478261,
|
| 95 |
+
"action_mse_to_best": 0.3893680230657692,
|
| 96 |
+
"per_task": {
|
| 97 |
+
"LiftPegUpright-v1": {
|
| 98 |
+
"action_mse_to_best": 0.33801942574071514,
|
| 99 |
+
"expert_success_rate": 0.8865979381443299,
|
| 100 |
+
"num_groups": 97,
|
| 101 |
+
"oracle_success_rate": 0.9278350515463918,
|
| 102 |
+
"policy_expert_regret": 1.1055486260615672,
|
| 103 |
+
"policy_oracle_regret": 1.158366499795127,
|
| 104 |
+
"policy_rollout_progress": 0.5693922632748318,
|
| 105 |
+
"policy_rollout_success_rate": 0.17525773195876287,
|
| 106 |
+
"restore_max_error": 4.76837158203125e-07
|
| 107 |
+
},
|
| 108 |
+
"PickCube-v1": {
|
| 109 |
+
"action_mse_to_best": 0.32239147143151897,
|
| 110 |
+
"expert_success_rate": 0.9375,
|
| 111 |
+
"num_groups": 208,
|
| 112 |
+
"oracle_success_rate": 0.9471153846153846,
|
| 113 |
+
"policy_expert_regret": 1.1212567036183407,
|
| 114 |
+
"policy_oracle_regret": 1.1346795710448463,
|
| 115 |
+
"policy_rollout_progress": 0.5665868632853605,
|
| 116 |
+
"policy_rollout_success_rate": 0.23076923076923078,
|
| 117 |
+
"restore_max_error": 4.76837158203125e-07
|
| 118 |
+
},
|
| 119 |
+
"PullCube-v1": {
|
| 120 |
+
"action_mse_to_best": 0.5440427258901008,
|
| 121 |
+
"expert_success_rate": 0.19480519480519481,
|
| 122 |
+
"num_groups": 77,
|
| 123 |
+
"oracle_success_rate": 0.36363636363636365,
|
| 124 |
+
"policy_expert_regret": 0.3430876467231806,
|
| 125 |
+
"policy_oracle_regret": 0.5237471687232519,
|
| 126 |
+
"policy_rollout_progress": 0.2622356957139133,
|
| 127 |
+
"policy_rollout_success_rate": 0.12987012987012986,
|
| 128 |
+
"restore_max_error": 4.76837158203125e-07
|
| 129 |
+
},
|
| 130 |
+
"PushCube-v1": {
|
| 131 |
+
"action_mse_to_best": 0.4075897317379713,
|
| 132 |
+
"expert_success_rate": 0.8617021276595744,
|
| 133 |
+
"num_groups": 94,
|
| 134 |
+
"oracle_success_rate": 0.9893617021276596,
|
| 135 |
+
"policy_expert_regret": 0.5765215990708229,
|
| 136 |
+
"policy_oracle_regret": 0.6747220545373065,
|
| 137 |
+
"policy_rollout_progress": 0.6763417752499276,
|
| 138 |
+
"policy_rollout_success_rate": 0.648936170212766,
|
| 139 |
+
"restore_max_error": 4.76837158203125e-07
|
| 140 |
+
},
|
| 141 |
+
"StackCube-v1": {
|
| 142 |
+
"action_mse_to_best": 0.4427937195081301,
|
| 143 |
+
"expert_success_rate": 0.6767676767676768,
|
| 144 |
+
"num_groups": 99,
|
| 145 |
+
"oracle_success_rate": 0.8585858585858586,
|
| 146 |
+
"policy_expert_regret": 1.0145450844006105,
|
| 147 |
+
"policy_oracle_regret": 1.2040003497492184,
|
| 148 |
+
"policy_rollout_progress": 0.41497098180380737,
|
| 149 |
+
"policy_rollout_success_rate": 0.20202020202020202,
|
| 150 |
+
"restore_max_error": 3.948807716369629e-07
|
| 151 |
+
}
|
| 152 |
+
}
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"config": "k16_sigma0.20",
|
| 156 |
+
"seed": 1,
|
| 157 |
+
"path": "/scratch/knguy52/dovla/experiments/dovla_h16_field_sweep/k16_sigma0.20/seed_1/online_rollout.json",
|
| 158 |
+
"selection_mode": "field",
|
| 159 |
+
"num_candidates": 16,
|
| 160 |
+
"candidate_sigma": 0.2,
|
| 161 |
+
"num_groups": 575,
|
| 162 |
+
"policy_rollout_success_rate": 0.2921739130434783,
|
| 163 |
+
"policy_rollout_progress": 0.5465844893455505,
|
| 164 |
+
"oracle_success_rate": 0.8695652173913043,
|
| 165 |
+
"action_mse_to_best": 0.40582152007795547,
|
| 166 |
+
"per_task": {
|
| 167 |
+
"LiftPegUpright-v1": {
|
| 168 |
+
"action_mse_to_best": 0.3490153367804215,
|
| 169 |
+
"expert_success_rate": 0.8584070796460177,
|
| 170 |
+
"num_groups": 113,
|
| 171 |
+
"oracle_success_rate": 0.9380530973451328,
|
| 172 |
+
"policy_expert_regret": 1.0372399143149367,
|
| 173 |
+
"policy_oracle_regret": 1.1186523459917677,
|
| 174 |
+
"policy_rollout_progress": 0.6100943451170373,
|
| 175 |
+
"policy_rollout_success_rate": 0.20353982300884957,
|
| 176 |
+
"restore_max_error": 4.76837158203125e-07
|
| 177 |
+
},
|
| 178 |
+
"PickCube-v1": {
|
| 179 |
+
"action_mse_to_best": 0.3508375255471986,
|
| 180 |
+
"expert_success_rate": 0.9402173913043478,
|
| 181 |
+
"num_groups": 184,
|
| 182 |
+
"oracle_success_rate": 0.9456521739130435,
|
| 183 |
+
"policy_expert_regret": 1.1819422131546005,
|
| 184 |
+
"policy_oracle_regret": 1.18947911812195,
|
| 185 |
+
"policy_rollout_progress": 0.5491846246017225,
|
| 186 |
+
"policy_rollout_success_rate": 0.18478260869565216,
|
| 187 |
+
"restore_max_error": 4.76837158203125e-07
|
| 188 |
+
},
|
| 189 |
+
"PullCube-v1": {
|
| 190 |
+
"action_mse_to_best": 0.5039778355276212,
|
| 191 |
+
"expert_success_rate": 0.25,
|
| 192 |
+
"num_groups": 76,
|
| 193 |
+
"oracle_success_rate": 0.40789473684210525,
|
| 194 |
+
"policy_expert_regret": 0.3182053966202626,
|
| 195 |
+
"policy_oracle_regret": 0.5304721274511203,
|
| 196 |
+
"policy_rollout_progress": 0.289067306776384,
|
| 197 |
+
"policy_rollout_success_rate": 0.15789473684210525,
|
| 198 |
+
"restore_max_error": 4.76837158203125e-07
|
| 199 |
+
},
|
| 200 |
+
"PushCube-v1": {
|
| 201 |
+
"action_mse_to_best": 0.40096906679976096,
|
| 202 |
+
"expert_success_rate": 0.8198198198198198,
|
| 203 |
+
"num_groups": 111,
|
| 204 |
+
"oracle_success_rate": 1.0,
|
| 205 |
+
"policy_expert_regret": 0.311778476914844,
|
| 206 |
+
"policy_oracle_regret": 0.3791768930785291,
|
| 207 |
+
"policy_rollout_progress": 0.8190213051196691,
|
| 208 |
+
"policy_rollout_success_rate": 0.8018018018018018,
|
| 209 |
+
"restore_max_error": 4.76837158203125e-07
|
| 210 |
+
},
|
| 211 |
+
"StackCube-v1": {
|
| 212 |
+
"action_mse_to_best": 0.5114797183854894,
|
| 213 |
+
"expert_success_rate": 0.6923076923076923,
|
| 214 |
+
"num_groups": 91,
|
| 215 |
+
"oracle_success_rate": 0.8571428571428571,
|
| 216 |
+
"policy_expert_regret": 1.2046928024226493,
|
| 217 |
+
"policy_oracle_regret": 1.3768131698553379,
|
| 218 |
+
"policy_rollout_progress": 0.34521944247759306,
|
| 219 |
+
"policy_rollout_success_rate": 0.10989010989010989,
|
| 220 |
+
"restore_max_error": 3.948807716369629e-07
|
| 221 |
+
}
|
| 222 |
+
}
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"config": "k16_sigma0.20",
|
| 226 |
+
"seed": 2,
|
| 227 |
+
"path": "/scratch/knguy52/dovla/experiments/dovla_h16_field_sweep/k16_sigma0.20/seed_2/online_rollout.json",
|
| 228 |
+
"selection_mode": "field",
|
| 229 |
+
"num_candidates": 16,
|
| 230 |
+
"candidate_sigma": 0.2,
|
| 231 |
+
"num_groups": 575,
|
| 232 |
+
"policy_rollout_success_rate": 0.2921739130434783,
|
| 233 |
+
"policy_rollout_progress": 0.5304317855770173,
|
| 234 |
+
"oracle_success_rate": 0.8765217391304347,
|
| 235 |
+
"action_mse_to_best": 0.42672614171615114,
|
| 236 |
+
"per_task": {
|
| 237 |
+
"LiftPegUpright-v1": {
|
| 238 |
+
"action_mse_to_best": 0.3699636154536468,
|
| 239 |
+
"expert_success_rate": 0.8229166666666666,
|
| 240 |
+
"num_groups": 96,
|
| 241 |
+
"oracle_success_rate": 0.9270833333333334,
|
| 242 |
+
"policy_expert_regret": 0.9971570746662716,
|
| 243 |
+
"policy_oracle_regret": 1.077391968574375,
|
| 244 |
+
"policy_rollout_progress": 0.6137169771827757,
|
| 245 |
+
"policy_rollout_success_rate": 0.22916666666666666,
|
| 246 |
+
"restore_max_error": 3.5762786865234375e-07
|
| 247 |
+
},
|
| 248 |
+
"PickCube-v1": {
|
| 249 |
+
"action_mse_to_best": 0.3610447467718687,
|
| 250 |
+
"expert_success_rate": 0.9444444444444444,
|
| 251 |
+
"num_groups": 198,
|
| 252 |
+
"oracle_success_rate": 0.9595959595959596,
|
| 253 |
+
"policy_expert_regret": 1.191942244523553,
|
| 254 |
+
"policy_oracle_regret": 1.2132171970989667,
|
| 255 |
+
"policy_rollout_progress": 0.5397262428181641,
|
| 256 |
+
"policy_rollout_success_rate": 0.20202020202020202,
|
| 257 |
+
"restore_max_error": 4.76837158203125e-07
|
| 258 |
+
},
|
| 259 |
+
"PullCube-v1": {
|
| 260 |
+
"action_mse_to_best": 0.5625709388819006,
|
| 261 |
+
"expert_success_rate": 0.24444444444444444,
|
| 262 |
+
"num_groups": 90,
|
| 263 |
+
"oracle_success_rate": 0.4666666666666667,
|
| 264 |
+
"policy_expert_regret": 0.290010487784942,
|
| 265 |
+
"policy_oracle_regret": 0.5038864802983072,
|
| 266 |
+
"policy_rollout_progress": 0.36756752870149084,
|
| 267 |
+
"policy_rollout_success_rate": 0.24444444444444444,
|
| 268 |
+
"restore_max_error": 4.0978193283081055e-07
|
| 269 |
+
},
|
| 270 |
+
"PushCube-v1": {
|
| 271 |
+
"action_mse_to_best": 0.39941793451509855,
|
| 272 |
+
"expert_success_rate": 0.8514851485148515,
|
| 273 |
+
"num_groups": 101,
|
| 274 |
+
"oracle_success_rate": 1.0,
|
| 275 |
+
"policy_expert_regret": 0.4599519388215377,
|
| 276 |
+
"policy_oracle_regret": 0.47853129202186473,
|
| 277 |
+
"policy_rollout_progress": 0.7689934604533828,
|
| 278 |
+
"policy_rollout_success_rate": 0.7524752475247525,
|
| 279 |
+
"restore_max_error": 4.76837158203125e-07
|
| 280 |
+
},
|
| 281 |
+
"StackCube-v1": {
|
| 282 |
+
"action_mse_to_best": 0.5265729850778977,
|
| 283 |
+
"expert_success_rate": 0.7666666666666667,
|
| 284 |
+
"num_groups": 90,
|
| 285 |
+
"oracle_success_rate": 0.9111111111111111,
|
| 286 |
+
"policy_expert_regret": 1.3538175096114478,
|
| 287 |
+
"policy_oracle_regret": 1.4900642686420016,
|
| 288 |
+
"policy_rollout_progress": 0.3162914858924018,
|
| 289 |
+
"policy_rollout_success_rate": 0.08888888888888889,
|
| 290 |
+
"restore_max_error": 4.76837158203125e-07
|
| 291 |
+
}
|
| 292 |
+
}
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"config": "k32_sigma0.35",
|
| 296 |
+
"seed": 0,
|
| 297 |
+
"path": "/scratch/knguy52/dovla/experiments/dovla_h16_field_sweep/k32_sigma0.35/seed_0/online_rollout.json",
|
| 298 |
+
"selection_mode": "field",
|
| 299 |
+
"num_candidates": 32,
|
| 300 |
+
"candidate_sigma": 0.35,
|
| 301 |
+
"num_groups": 575,
|
| 302 |
+
"policy_rollout_success_rate": 0.2782608695652174,
|
| 303 |
+
"policy_rollout_progress": 0.523629953644198,
|
| 304 |
+
"oracle_success_rate": 0.8573913043478261,
|
| 305 |
+
"action_mse_to_best": 0.3982355571473422,
|
| 306 |
+
"per_task": {
|
| 307 |
+
"LiftPegUpright-v1": {
|
| 308 |
+
"action_mse_to_best": 0.3380384829048951,
|
| 309 |
+
"expert_success_rate": 0.8865979381443299,
|
| 310 |
+
"num_groups": 97,
|
| 311 |
+
"oracle_success_rate": 0.9278350515463918,
|
| 312 |
+
"policy_expert_regret": 1.0720160110402353,
|
| 313 |
+
"policy_oracle_regret": 1.111630466641839,
|
| 314 |
+
"policy_rollout_progress": 0.5852004613765737,
|
| 315 |
+
"policy_rollout_success_rate": 0.20618556701030927,
|
| 316 |
+
"restore_max_error": 4.76837158203125e-07
|
| 317 |
+
},
|
| 318 |
+
"PickCube-v1": {
|
| 319 |
+
"action_mse_to_best": 0.3329235534183681,
|
| 320 |
+
"expert_success_rate": 0.9375,
|
| 321 |
+
"num_groups": 208,
|
| 322 |
+
"oracle_success_rate": 0.9471153846153846,
|
| 323 |
+
"policy_expert_regret": 1.125776257759963,
|
| 324 |
+
"policy_oracle_regret": 1.1391991251864686,
|
| 325 |
+
"policy_rollout_progress": 0.5620673091437381,
|
| 326 |
+
"policy_rollout_success_rate": 0.23076923076923078,
|
| 327 |
+
"restore_max_error": 4.76837158203125e-07
|
| 328 |
+
},
|
| 329 |
+
"PullCube-v1": {
|
| 330 |
+
"action_mse_to_best": 0.5762418479985231,
|
| 331 |
+
"expert_success_rate": 0.19480519480519481,
|
| 332 |
+
"num_groups": 77,
|
| 333 |
+
"oracle_success_rate": 0.36363636363636365,
|
| 334 |
+
"policy_expert_regret": 0.3187135829312074,
|
| 335 |
+
"policy_oracle_regret": 0.5018423422929141,
|
| 336 |
+
"policy_rollout_progress": 0.28152527091661833,
|
| 337 |
+
"policy_rollout_success_rate": 0.15584415584415584,
|
| 338 |
+
"restore_max_error": 4.76837158203125e-07
|
| 339 |
+
},
|
| 340 |
+
"PushCube-v1": {
|
| 341 |
+
"action_mse_to_best": 0.4108403711480663,
|
| 342 |
+
"expert_success_rate": 0.8617021276595744,
|
| 343 |
+
"num_groups": 94,
|
| 344 |
+
"oracle_success_rate": 0.9893617021276596,
|
| 345 |
+
"policy_expert_regret": 0.6197059926517466,
|
| 346 |
+
"policy_oracle_regret": 0.7191978959327049,
|
| 347 |
+
"policy_rollout_progress": 0.65314252959921,
|
| 348 |
+
"policy_rollout_success_rate": 0.6276595744680851,
|
| 349 |
+
"restore_max_error": 4.76837158203125e-07
|
| 350 |
+
},
|
| 351 |
+
"StackCube-v1": {
|
| 352 |
+
"action_mse_to_best": 0.444020163869918,
|
| 353 |
+
"expert_success_rate": 0.6767676767676768,
|
| 354 |
+
"num_groups": 99,
|
| 355 |
+
"oracle_success_rate": 0.8585858585858586,
|
| 356 |
+
"policy_expert_regret": 0.9832775508514559,
|
| 357 |
+
"policy_oracle_regret": 1.1609744254988854,
|
| 358 |
+
"policy_rollout_progress": 0.447878127748316,
|
| 359 |
+
"policy_rollout_success_rate": 0.21212121212121213,
|
| 360 |
+
"restore_max_error": 3.948807716369629e-07
|
| 361 |
+
}
|
| 362 |
+
}
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"config": "k32_sigma0.35",
|
| 366 |
+
"seed": 1,
|
| 367 |
+
"path": "/scratch/knguy52/dovla/experiments/dovla_h16_field_sweep/k32_sigma0.35/seed_1/online_rollout.json",
|
| 368 |
+
"selection_mode": "field",
|
| 369 |
+
"num_candidates": 32,
|
| 370 |
+
"candidate_sigma": 0.35,
|
| 371 |
+
"num_groups": 575,
|
| 372 |
+
"policy_rollout_success_rate": 0.30434782608695654,
|
| 373 |
+
"policy_rollout_progress": 0.5489098613255697,
|
| 374 |
+
"oracle_success_rate": 0.8695652173913043,
|
| 375 |
+
"action_mse_to_best": 0.4134289959891011,
|
| 376 |
+
"per_task": {
|
| 377 |
+
"LiftPegUpright-v1": {
|
| 378 |
+
"action_mse_to_best": 0.35516584440406446,
|
| 379 |
+
"expert_success_rate": 0.8584070796460177,
|
| 380 |
+
"num_groups": 113,
|
| 381 |
+
"oracle_success_rate": 0.9380530973451328,
|
| 382 |
+
"policy_expert_regret": 0.9619057153442264,
|
| 383 |
+
"policy_oracle_regret": 1.0545724122661404,
|
| 384 |
+
"policy_rollout_progress": 0.6299264912320449,
|
| 385 |
+
"policy_rollout_success_rate": 0.24778761061946902,
|
| 386 |
+
"restore_max_error": 4.76837158203125e-07
|
| 387 |
+
},
|
| 388 |
+
"PickCube-v1": {
|
| 389 |
+
"action_mse_to_best": 0.35703187711212947,
|
| 390 |
+
"expert_success_rate": 0.9402173913043478,
|
| 391 |
+
"num_groups": 184,
|
| 392 |
+
"oracle_success_rate": 0.9456521739130435,
|
| 393 |
+
"policy_expert_regret": 1.1820752561881975,
|
| 394 |
+
"policy_oracle_regret": 1.1896121611555472,
|
| 395 |
+
"policy_rollout_progress": 0.5490515815681252,
|
| 396 |
+
"policy_rollout_success_rate": 0.18478260869565216,
|
| 397 |
+
"restore_max_error": 4.76837158203125e-07
|
| 398 |
+
},
|
| 399 |
+
"PullCube-v1": {
|
| 400 |
+
"action_mse_to_best": 0.5263007944511053,
|
| 401 |
+
"expert_success_rate": 0.25,
|
| 402 |
+
"num_groups": 76,
|
| 403 |
+
"oracle_success_rate": 0.40789473684210525,
|
| 404 |
+
"policy_expert_regret": 0.30107949649620996,
|
| 405 |
+
"policy_oracle_regret": 0.5109168701364022,
|
| 406 |
+
"policy_rollout_progress": 0.28283038475599726,
|
| 407 |
+
"policy_rollout_success_rate": 0.15789473684210525,
|
| 408 |
+
"restore_max_error": 4.76837158203125e-07
|
| 409 |
+
},
|
| 410 |
+
"PushCube-v1": {
|
| 411 |
+
"action_mse_to_best": 0.4066863194242254,
|
| 412 |
+
"expert_success_rate": 0.8198198198198198,
|
| 413 |
+
"num_groups": 111,
|
| 414 |
+
"oracle_success_rate": 1.0,
|
| 415 |
+
"policy_expert_regret": 0.3104555992124317,
|
| 416 |
+
"policy_oracle_regret": 0.3782863458534619,
|
| 417 |
+
"policy_rollout_progress": 0.8199118523447363,
|
| 418 |
+
"policy_rollout_success_rate": 0.8018018018018018,
|
| 419 |
+
"restore_max_error": 4.76837158203125e-07
|
| 420 |
+
},
|
| 421 |
+
"StackCube-v1": {
|
| 422 |
+
"action_mse_to_best": 0.5137695060776812,
|
| 423 |
+
"expert_success_rate": 0.6923076923076923,
|
| 424 |
+
"num_groups": 91,
|
| 425 |
+
"oracle_success_rate": 0.8571428571428571,
|
| 426 |
+
"policy_expert_regret": 1.200795496066848,
|
| 427 |
+
"policy_oracle_regret": 1.3603769981271618,
|
| 428 |
+
"policy_rollout_progress": 0.3396775922277471,
|
| 429 |
+
"policy_rollout_success_rate": 0.13186813186813187,
|
| 430 |
+
"restore_max_error": 3.948807716369629e-07
|
| 431 |
+
}
|
| 432 |
+
}
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"config": "k32_sigma0.35",
|
| 436 |
+
"seed": 2,
|
| 437 |
+
"path": "/scratch/knguy52/dovla/experiments/dovla_h16_field_sweep/k32_sigma0.35/seed_2/online_rollout.json",
|
| 438 |
+
"selection_mode": "field",
|
| 439 |
+
"num_candidates": 32,
|
| 440 |
+
"candidate_sigma": 0.35,
|
| 441 |
+
"num_groups": 575,
|
| 442 |
+
"policy_rollout_success_rate": 0.29043478260869565,
|
| 443 |
+
"policy_rollout_progress": 0.5305287452098792,
|
| 444 |
+
"oracle_success_rate": 0.8765217391304347,
|
| 445 |
+
"action_mse_to_best": 0.43532119818841636,
|
| 446 |
+
"per_task": {
|
| 447 |
+
"LiftPegUpright-v1": {
|
| 448 |
+
"action_mse_to_best": 0.37001094076549634,
|
| 449 |
+
"expert_success_rate": 0.8229166666666666,
|
| 450 |
+
"num_groups": 96,
|
| 451 |
+
"oracle_success_rate": 0.9270833333333334,
|
| 452 |
+
"policy_expert_regret": 1.0173339761483173,
|
| 453 |
+
"policy_oracle_regret": 1.097288203580926,
|
| 454 |
+
"policy_rollout_progress": 0.604231653890262,
|
| 455 |
+
"policy_rollout_success_rate": 0.21875,
|
| 456 |
+
"restore_max_error": 3.5762786865234375e-07
|
| 457 |
+
},
|
| 458 |
+
"PickCube-v1": {
|
| 459 |
+
"action_mse_to_best": 0.3635602415953245,
|
| 460 |
+
"expert_success_rate": 0.9444444444444444,
|
| 461 |
+
"num_groups": 198,
|
| 462 |
+
"oracle_success_rate": 0.9595959595959596,
|
| 463 |
+
"policy_expert_regret": 1.1922823830855767,
|
| 464 |
+
"policy_oracle_regret": 1.2135573356609903,
|
| 465 |
+
"policy_rollout_progress": 0.5393861042561404,
|
| 466 |
+
"policy_rollout_success_rate": 0.20202020202020202,
|
| 467 |
+
"restore_max_error": 4.76837158203125e-07
|
| 468 |
+
},
|
| 469 |
+
"PullCube-v1": {
|
| 470 |
+
"action_mse_to_best": 0.6043633603180448,
|
| 471 |
+
"expert_success_rate": 0.24444444444444444,
|
| 472 |
+
"num_groups": 90,
|
| 473 |
+
"oracle_success_rate": 0.4666666666666667,
|
| 474 |
+
"policy_expert_regret": 0.2729570985513015,
|
| 475 |
+
"policy_oracle_regret": 0.5294890147789071,
|
| 476 |
+
"policy_rollout_progress": 0.3726402646292829,
|
| 477 |
+
"policy_rollout_success_rate": 0.25555555555555554,
|
| 478 |
+
"restore_max_error": 4.0978193283081055e-07
|
| 479 |
+
},
|
| 480 |
+
"PushCube-v1": {
|
| 481 |
+
"action_mse_to_best": 0.40536120799508424,
|
| 482 |
+
"expert_success_rate": 0.8514851485148515,
|
| 483 |
+
"num_groups": 101,
|
| 484 |
+
"oracle_success_rate": 1.0,
|
| 485 |
+
"policy_expert_regret": 0.44068841591920005,
|
| 486 |
+
"policy_oracle_regret": 0.4968654212385121,
|
| 487 |
+
"policy_rollout_progress": 0.7605603213357454,
|
| 488 |
+
"policy_rollout_success_rate": 0.7425742574257426,
|
| 489 |
+
"restore_max_error": 4.76837158203125e-07
|
| 490 |
+
},
|
| 491 |
+
"StackCube-v1": {
|
| 492 |
+
"action_mse_to_best": 0.527439181920555,
|
| 493 |
+
"expert_success_rate": 0.7666666666666667,
|
| 494 |
+
"num_groups": 90,
|
| 495 |
+
"oracle_success_rate": 0.9111111111111111,
|
| 496 |
+
"policy_expert_regret": 1.3509710078438124,
|
| 497 |
+
"policy_oracle_regret": 1.4741204554835956,
|
| 498 |
+
"policy_rollout_progress": 0.3321680534217093,
|
| 499 |
+
"policy_rollout_success_rate": 0.08888888888888889,
|
| 500 |
+
"restore_max_error": 4.76837158203125e-07
|
| 501 |
+
}
|
| 502 |
+
}
|
| 503 |
+
},
|
| 504 |
+
{
|
| 505 |
+
"config": "k64_sigma0.50",
|
| 506 |
+
"seed": 0,
|
| 507 |
+
"path": "/scratch/knguy52/dovla/experiments/dovla_h16_field_sweep/k64_sigma0.50/seed_0/online_rollout.json",
|
| 508 |
+
"selection_mode": "field",
|
| 509 |
+
"num_candidates": 64,
|
| 510 |
+
"candidate_sigma": 0.5,
|
| 511 |
+
"num_groups": 575,
|
| 512 |
+
"policy_rollout_success_rate": 0.2782608695652174,
|
| 513 |
+
"policy_rollout_progress": 0.5179621245339513,
|
| 514 |
+
"oracle_success_rate": 0.8573913043478261,
|
| 515 |
+
"action_mse_to_best": 0.4078877551756475,
|
| 516 |
+
"per_task": {
|
| 517 |
+
"LiftPegUpright-v1": {
|
| 518 |
+
"action_mse_to_best": 0.33705748692538934,
|
| 519 |
+
"expert_success_rate": 0.8865979381443299,
|
| 520 |
+
"num_groups": 97,
|
| 521 |
+
"oracle_success_rate": 0.9278350515463918,
|
| 522 |
+
"policy_expert_regret": 1.077372452954656,
|
| 523 |
+
"policy_oracle_regret": 1.1169869085562598,
|
| 524 |
+
"policy_rollout_progress": 0.5798440194621528,
|
| 525 |
+
"policy_rollout_success_rate": 0.20618556701030927,
|
| 526 |
+
"restore_max_error": 4.76837158203125e-07
|
| 527 |
+
},
|
| 528 |
+
"PickCube-v1": {
|
| 529 |
+
"action_mse_to_best": 0.3444686237221154,
|
| 530 |
+
"expert_success_rate": 0.9375,
|
| 531 |
+
"num_groups": 208,
|
| 532 |
+
"oracle_success_rate": 0.9471153846153846,
|
| 533 |
+
"policy_expert_regret": 1.1347050576027518,
|
| 534 |
+
"policy_oracle_regret": 1.1481279250292573,
|
| 535 |
+
"policy_rollout_progress": 0.5531385093009493,
|
| 536 |
+
"policy_rollout_success_rate": 0.23076923076923078,
|
| 537 |
+
"restore_max_error": 4.76837158203125e-07
|
| 538 |
+
},
|
| 539 |
+
"PullCube-v1": {
|
| 540 |
+
"action_mse_to_best": 0.6116091051659027,
|
| 541 |
+
"expert_success_rate": 0.19480519480519481,
|
| 542 |
+
"num_groups": 77,
|
| 543 |
+
"oracle_success_rate": 0.36363636363636365,
|
| 544 |
+
"policy_expert_regret": 0.3769259353216689,
|
| 545 |
+
"policy_oracle_regret": 0.5868512122141947,
|
| 546 |
+
"policy_rollout_progress": 0.2550891892222518,
|
| 547 |
+
"policy_rollout_success_rate": 0.14285714285714285,
|
| 548 |
+
"restore_max_error": 4.76837158203125e-07
|
| 549 |
+
},
|
| 550 |
+
"PushCube-v1": {
|
| 551 |
+
"action_mse_to_best": 0.4115920202766961,
|
| 552 |
+
"expert_success_rate": 0.8617021276595744,
|
| 553 |
+
"num_groups": 94,
|
| 554 |
+
"oracle_success_rate": 0.9893617021276596,
|
| 555 |
+
"policy_expert_regret": 0.5561907957208917,
|
| 556 |
+
"policy_oracle_regret": 0.6959038405342305,
|
| 557 |
+
"policy_rollout_progress": 0.665798287125344,
|
| 558 |
+
"policy_rollout_success_rate": 0.6382978723404256,
|
| 559 |
+
"restore_max_error": 4.76837158203125e-07
|
| 560 |
+
},
|
| 561 |
+
"StackCube-v1": {
|
| 562 |
+
"action_mse_to_best": 0.4485642248106123,
|
| 563 |
+
"expert_success_rate": 0.6767676767676768,
|
| 564 |
+
"num_groups": 99,
|
| 565 |
+
"oracle_success_rate": 0.8585858585858586,
|
| 566 |
+
"policy_expert_regret": 0.9836442116836105,
|
| 567 |
+
"policy_oracle_regret": 1.16134108633104,
|
| 568 |
+
"policy_rollout_progress": 0.44751146691616134,
|
| 569 |
+
"policy_rollout_success_rate": 0.21212121212121213,
|
| 570 |
+
"restore_max_error": 3.948807716369629e-07
|
| 571 |
+
}
|
| 572 |
+
}
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"config": "k64_sigma0.50",
|
| 576 |
+
"seed": 1,
|
| 577 |
+
"path": "/scratch/knguy52/dovla/experiments/dovla_h16_field_sweep/k64_sigma0.50/seed_1/online_rollout.json",
|
| 578 |
+
"selection_mode": "field",
|
| 579 |
+
"num_candidates": 64,
|
| 580 |
+
"candidate_sigma": 0.5,
|
| 581 |
+
"num_groups": 575,
|
| 582 |
+
"policy_rollout_success_rate": 0.29739130434782607,
|
| 583 |
+
"policy_rollout_progress": 0.5436022769287229,
|
| 584 |
+
"oracle_success_rate": 0.8695652173913043,
|
| 585 |
+
"action_mse_to_best": 0.42184323112682803,
|
| 586 |
+
"per_task": {
|
| 587 |
+
"LiftPegUpright-v1": {
|
| 588 |
+
"action_mse_to_best": 0.35531353457641285,
|
| 589 |
+
"expert_success_rate": 0.8584070796460177,
|
| 590 |
+
"num_groups": 113,
|
| 591 |
+
"oracle_success_rate": 0.9380530973451328,
|
| 592 |
+
"policy_expert_regret": 0.9860384428659371,
|
| 593 |
+
"policy_oracle_regret": 1.067450874542768,
|
| 594 |
+
"policy_rollout_progress": 0.6170480289554174,
|
| 595 |
+
"policy_rollout_success_rate": 0.24778761061946902,
|
| 596 |
+
"restore_max_error": 4.76837158203125e-07
|
| 597 |
+
},
|
| 598 |
+
"PickCube-v1": {
|
| 599 |
+
"action_mse_to_best": 0.36275648236598657,
|
| 600 |
+
"expert_success_rate": 0.9402173913043478,
|
| 601 |
+
"num_groups": 184,
|
| 602 |
+
"oracle_success_rate": 0.9456521739130435,
|
| 603 |
+
"policy_expert_regret": 1.1823922171783836,
|
| 604 |
+
"policy_oracle_regret": 1.189929122145733,
|
| 605 |
+
"policy_rollout_progress": 0.5487346205779392,
|
| 606 |
+
"policy_rollout_success_rate": 0.18478260869565216,
|
| 607 |
+
"restore_max_error": 4.76837158203125e-07
|
| 608 |
+
},
|
| 609 |
+
"PullCube-v1": {
|
| 610 |
+
"action_mse_to_best": 0.5605023919028769,
|
| 611 |
+
"expert_success_rate": 0.25,
|
| 612 |
+
"num_groups": 76,
|
| 613 |
+
"oracle_success_rate": 0.40789473684210525,
|
| 614 |
+
"policy_expert_regret": 0.2840835368383284,
|
| 615 |
+
"policy_oracle_regret": 0.48901411867700517,
|
| 616 |
+
"policy_rollout_progress": 0.3163563787912656,
|
| 617 |
+
"policy_rollout_success_rate": 0.17105263157894737,
|
| 618 |
+
"restore_max_error": 4.76837158203125e-07
|
| 619 |
+
},
|
| 620 |
+
"PushCube-v1": {
|
| 621 |
+
"action_mse_to_best": 0.41608408925769563,
|
| 622 |
+
"expert_success_rate": 0.8198198198198198,
|
| 623 |
+
"num_groups": 111,
|
| 624 |
+
"oracle_success_rate": 1.0,
|
| 625 |
+
"policy_expert_regret": 0.33139527461550256,
|
| 626 |
+
"policy_oracle_regret": 0.432737438259898,
|
| 627 |
+
"policy_rollout_progress": 0.7924877869653272,
|
| 628 |
+
"policy_rollout_success_rate": 0.7747747747747747,
|
| 629 |
+
"restore_max_error": 4.76837158203125e-07
|
| 630 |
+
},
|
| 631 |
+
"StackCube-v1": {
|
| 632 |
+
"action_mse_to_best": 0.5151507697057921,
|
| 633 |
+
"expert_success_rate": 0.6923076923076923,
|
| 634 |
+
"num_groups": 91,
|
| 635 |
+
"oracle_success_rate": 0.8571428571428571,
|
| 636 |
+
"policy_expert_regret": 1.221638112277775,
|
| 637 |
+
"policy_oracle_regret": 1.3938075284381488,
|
| 638 |
+
"policy_rollout_progress": 0.328225083894782,
|
| 639 |
+
"policy_rollout_success_rate": 0.10989010989010989,
|
| 640 |
+
"restore_max_error": 3.948807716369629e-07
|
| 641 |
+
}
|
| 642 |
+
}
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"config": "k64_sigma0.50",
|
| 646 |
+
"seed": 2,
|
| 647 |
+
"path": "/scratch/knguy52/dovla/experiments/dovla_h16_field_sweep/k64_sigma0.50/seed_2/online_rollout.json",
|
| 648 |
+
"selection_mode": "field",
|
| 649 |
+
"num_candidates": 64,
|
| 650 |
+
"candidate_sigma": 0.5,
|
| 651 |
+
"num_groups": 575,
|
| 652 |
+
"policy_rollout_success_rate": 0.28347826086956524,
|
| 653 |
+
"policy_rollout_progress": 0.5251258096023325,
|
| 654 |
+
"oracle_success_rate": 0.8765217391304347,
|
| 655 |
+
"action_mse_to_best": 0.4439206386568106,
|
| 656 |
+
"per_task": {
|
| 657 |
+
"LiftPegUpright-v1": {
|
| 658 |
+
"action_mse_to_best": 0.37040768992543843,
|
| 659 |
+
"expert_success_rate": 0.8229166666666666,
|
| 660 |
+
"num_groups": 96,
|
| 661 |
+
"oracle_success_rate": 0.9270833333333334,
|
| 662 |
+
"policy_expert_regret": 0.998967742236952,
|
| 663 |
+
"policy_oracle_regret": 1.0789219696695607,
|
| 664 |
+
"policy_rollout_progress": 0.6225978878016273,
|
| 665 |
+
"policy_rollout_success_rate": 0.21875,
|
| 666 |
+
"restore_max_error": 3.5762786865234375e-07
|
| 667 |
+
},
|
| 668 |
+
"PickCube-v1": {
|
| 669 |
+
"action_mse_to_best": 0.36784123307131605,
|
| 670 |
+
"expert_success_rate": 0.9444444444444444,
|
| 671 |
+
"num_groups": 198,
|
| 672 |
+
"oracle_success_rate": 0.9595959595959596,
|
| 673 |
+
"policy_expert_regret": 1.195629036242161,
|
| 674 |
+
"policy_oracle_regret": 1.216903988817575,
|
| 675 |
+
"policy_rollout_progress": 0.5360394510995559,
|
| 676 |
+
"policy_rollout_success_rate": 0.20202020202020202,
|
| 677 |
+
"restore_max_error": 4.76837158203125e-07
|
| 678 |
+
},
|
| 679 |
+
"PullCube-v1": {
|
| 680 |
+
"action_mse_to_best": 0.6466699119864239,
|
| 681 |
+
"expert_success_rate": 0.24444444444444444,
|
| 682 |
+
"num_groups": 90,
|
| 683 |
+
"oracle_success_rate": 0.4666666666666667,
|
| 684 |
+
"policy_expert_regret": 0.3432700668840981,
|
| 685 |
+
"policy_oracle_regret": 0.6204078811807145,
|
| 686 |
+
"policy_rollout_progress": 0.30331905963658856,
|
| 687 |
+
"policy_rollout_success_rate": 0.18888888888888888,
|
| 688 |
+
"restore_max_error": 4.0978193283081055e-07
|
| 689 |
+
},
|
| 690 |
+
"PushCube-v1": {
|
| 691 |
+
"action_mse_to_best": 0.4085303876217049,
|
| 692 |
+
"expert_success_rate": 0.8514851485148515,
|
| 693 |
+
"num_groups": 101,
|
| 694 |
+
"oracle_success_rate": 1.0,
|
| 695 |
+
"policy_expert_regret": 0.4622636683801613,
|
| 696 |
+
"policy_oracle_regret": 0.48084302158048836,
|
| 697 |
+
"policy_rollout_progress": 0.7666817308947591,
|
| 698 |
+
"policy_rollout_success_rate": 0.7524752475247525,
|
| 699 |
+
"restore_max_error": 4.76837158203125e-07
|
| 700 |
+
},
|
| 701 |
+
"StackCube-v1": {
|
| 702 |
+
"action_mse_to_best": 0.526675595757034,
|
| 703 |
+
"expert_success_rate": 0.7666666666666667,
|
| 704 |
+
"num_groups": 90,
|
| 705 |
+
"oracle_success_rate": 0.9111111111111111,
|
| 706 |
+
"policy_expert_regret": 1.324383272892899,
|
| 707 |
+
"policy_oracle_regret": 1.4475327205326822,
|
| 708 |
+
"policy_rollout_progress": 0.3478729087445471,
|
| 709 |
+
"policy_rollout_success_rate": 0.1,
|
| 710 |
+
"restore_max_error": 4.76837158203125e-07
|
| 711 |
+
}
|
| 712 |
+
}
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"config": "k8_sigma0.10",
|
| 716 |
+
"seed": 0,
|
| 717 |
+
"path": "/scratch/knguy52/dovla/experiments/dovla_h16_field_sweep/k8_sigma0.10/seed_0/online_rollout.json",
|
| 718 |
+
"selection_mode": "field",
|
| 719 |
+
"num_candidates": 8,
|
| 720 |
+
"candidate_sigma": 0.1,
|
| 721 |
+
"num_groups": 575,
|
| 722 |
+
"policy_rollout_success_rate": 0.26956521739130435,
|
| 723 |
+
"policy_rollout_progress": 0.5201629009194996,
|
| 724 |
+
"oracle_success_rate": 0.8573913043478261,
|
| 725 |
+
"action_mse_to_best": 0.38439171190980986,
|
| 726 |
+
"per_task": {
|
| 727 |
+
"LiftPegUpright-v1": {
|
| 728 |
+
"action_mse_to_best": 0.3384406008402404,
|
| 729 |
+
"expert_success_rate": 0.8865979381443299,
|
| 730 |
+
"num_groups": 97,
|
| 731 |
+
"oracle_success_rate": 0.9278350515463918,
|
| 732 |
+
"policy_expert_regret": 1.0877418216970778,
|
| 733 |
+
"policy_oracle_regret": 1.1273562772986816,
|
| 734 |
+
"policy_rollout_progress": 0.5797839290702466,
|
| 735 |
+
"policy_rollout_success_rate": 0.1958762886597938,
|
| 736 |
+
"restore_max_error": 4.76837158203125e-07
|
| 737 |
+
},
|
| 738 |
+
"PickCube-v1": {
|
| 739 |
+
"action_mse_to_best": 0.31949265369285756,
|
| 740 |
+
"expert_success_rate": 0.9375,
|
| 741 |
+
"num_groups": 208,
|
| 742 |
+
"oracle_success_rate": 0.9471153846153846,
|
| 743 |
+
"policy_expert_regret": 1.1312620476282274,
|
| 744 |
+
"policy_oracle_regret": 1.1446849150547327,
|
| 745 |
+
"policy_rollout_progress": 0.5642561841075524,
|
| 746 |
+
"policy_rollout_success_rate": 0.21634615384615385,
|
| 747 |
+
"restore_max_error": 4.76837158203125e-07
|
| 748 |
+
},
|
| 749 |
+
"PullCube-v1": {
|
| 750 |
+
"action_mse_to_best": 0.5289782768284733,
|
| 751 |
+
"expert_success_rate": 0.19480519480519481,
|
| 752 |
+
"num_groups": 77,
|
| 753 |
+
"oracle_success_rate": 0.36363636363636365,
|
| 754 |
+
"policy_expert_regret": 0.27037695877544293,
|
| 755 |
+
"policy_oracle_regret": 0.480627155294279,
|
| 756 |
+
"policy_rollout_progress": 0.25797066397287627,
|
| 757 |
+
"policy_rollout_success_rate": 0.12987012987012986,
|
| 758 |
+
"restore_max_error": 4.76837158203125e-07
|
| 759 |
+
},
|
| 760 |
+
"PushCube-v1": {
|
| 761 |
+
"action_mse_to_best": 0.39730814422302424,
|
| 762 |
+
"expert_success_rate": 0.8617021276595744,
|
| 763 |
+
"num_groups": 94,
|
| 764 |
+
"oracle_success_rate": 0.9893617021276596,
|
| 765 |
+
"policy_expert_regret": 0.6144910497551269,
|
| 766 |
+
"policy_oracle_regret": 0.6746792780592087,
|
| 767 |
+
"policy_rollout_progress": 0.667395237912523,
|
| 768 |
+
"policy_rollout_success_rate": 0.6382978723404256,
|
| 769 |
+
"restore_max_error": 4.76837158203125e-07
|
| 770 |
+
},
|
| 771 |
+
"StackCube-v1": {
|
| 772 |
+
"action_mse_to_best": 0.44104779015925494,
|
| 773 |
+
"expert_success_rate": 0.6767676767676768,
|
| 774 |
+
"num_groups": 99,
|
| 775 |
+
"oracle_success_rate": 0.8585858585858586,
|
| 776 |
+
"policy_expert_regret": 1.0094024636829741,
|
| 777 |
+
"policy_oracle_regret": 1.1756157136023646,
|
| 778 |
+
"policy_rollout_progress": 0.43323683964483667,
|
| 779 |
+
"policy_rollout_success_rate": 0.21212121212121213,
|
| 780 |
+
"restore_max_error": 3.948807716369629e-07
|
| 781 |
+
}
|
| 782 |
+
}
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"config": "k8_sigma0.10",
|
| 786 |
+
"seed": 1,
|
| 787 |
+
"path": "/scratch/knguy52/dovla/experiments/dovla_h16_field_sweep/k8_sigma0.10/seed_1/online_rollout.json",
|
| 788 |
+
"selection_mode": "field",
|
| 789 |
+
"num_candidates": 8,
|
| 790 |
+
"candidate_sigma": 0.1,
|
| 791 |
+
"num_groups": 575,
|
| 792 |
+
"policy_rollout_success_rate": 0.2921739130434783,
|
| 793 |
+
"policy_rollout_progress": 0.5480315357268505,
|
| 794 |
+
"oracle_success_rate": 0.8695652173913043,
|
| 795 |
+
"action_mse_to_best": 0.4003930856425153,
|
| 796 |
+
"per_task": {
|
| 797 |
+
"LiftPegUpright-v1": {
|
| 798 |
+
"action_mse_to_best": 0.34787082138050973,
|
| 799 |
+
"expert_success_rate": 0.8584070796460177,
|
| 800 |
+
"num_groups": 113,
|
| 801 |
+
"oracle_success_rate": 0.9380530973451328,
|
| 802 |
+
"policy_expert_regret": 0.9883848046307016,
|
| 803 |
+
"policy_oracle_regret": 1.0584801957670567,
|
| 804 |
+
"policy_rollout_progress": 0.6348682652532527,
|
| 805 |
+
"policy_rollout_success_rate": 0.23893805309734514,
|
| 806 |
+
"restore_max_error": 4.76837158203125e-07
|
| 807 |
+
},
|
| 808 |
+
"PickCube-v1": {
|
| 809 |
+
"action_mse_to_best": 0.3476131692042817,
|
| 810 |
+
"expert_success_rate": 0.9402173913043478,
|
| 811 |
+
"num_groups": 184,
|
| 812 |
+
"oracle_success_rate": 0.9456521739130435,
|
| 813 |
+
"policy_expert_regret": 1.179325742960867,
|
| 814 |
+
"policy_oracle_regret": 1.1868626479282165,
|
| 815 |
+
"policy_rollout_progress": 0.5518010947954558,
|
| 816 |
+
"policy_rollout_success_rate": 0.18478260869565216,
|
| 817 |
+
"restore_max_error": 4.76837158203125e-07
|
| 818 |
+
},
|
| 819 |
+
"PullCube-v1": {
|
| 820 |
+
"action_mse_to_best": 0.4870559127180298,
|
| 821 |
+
"expert_success_rate": 0.25,
|
| 822 |
+
"num_groups": 76,
|
| 823 |
+
"oracle_success_rate": 0.40789473684210525,
|
| 824 |
+
"policy_expert_regret": 0.33368864436143714,
|
| 825 |
+
"policy_oracle_regret": 0.5169463043566793,
|
| 826 |
+
"policy_rollout_progress": 0.3027015187606019,
|
| 827 |
+
"policy_rollout_success_rate": 0.15789473684210525,
|
| 828 |
+
"restore_max_error": 4.76837158203125e-07
|
| 829 |
+
},
|
| 830 |
+
"PushCube-v1": {
|
| 831 |
+
"action_mse_to_best": 0.3928175041502392,
|
| 832 |
+
"expert_success_rate": 0.8198198198198198,
|
| 833 |
+
"num_groups": 111,
|
| 834 |
+
"oracle_success_rate": 1.0,
|
| 835 |
+
"policy_expert_regret": 0.38013805636951514,
|
| 836 |
+
"policy_oracle_regret": 0.4654143318399653,
|
| 837 |
+
"policy_rollout_progress": 0.7778289114032779,
|
| 838 |
+
"policy_rollout_success_rate": 0.7567567567567568,
|
| 839 |
+
"restore_max_error": 4.76837158203125e-07
|
| 840 |
+
},
|
| 841 |
+
"StackCube-v1": {
|
| 842 |
+
"action_mse_to_best": 0.5091956699737809,
|
| 843 |
+
"expert_success_rate": 0.6923076923076923,
|
| 844 |
+
"num_groups": 91,
|
| 845 |
+
"oracle_success_rate": 0.8571428571428571,
|
| 846 |
+
"policy_expert_regret": 1.1688671315109336,
|
| 847 |
+
"policy_oracle_regret": 1.353875546009986,
|
| 848 |
+
"policy_rollout_progress": 0.357168055333934,
|
| 849 |
+
"policy_rollout_success_rate": 0.12087912087912088,
|
| 850 |
+
"restore_max_error": 3.948807716369629e-07
|
| 851 |
+
}
|
| 852 |
+
}
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"config": "k8_sigma0.10",
|
| 856 |
+
"seed": 2,
|
| 857 |
+
"path": "/scratch/knguy52/dovla/experiments/dovla_h16_field_sweep/k8_sigma0.10/seed_2/online_rollout.json",
|
| 858 |
+
"selection_mode": "field",
|
| 859 |
+
"num_candidates": 8,
|
| 860 |
+
"candidate_sigma": 0.1,
|
| 861 |
+
"num_groups": 575,
|
| 862 |
+
"policy_rollout_success_rate": 0.2991304347826087,
|
| 863 |
+
"policy_rollout_progress": 0.5450154561020336,
|
| 864 |
+
"oracle_success_rate": 0.8765217391304347,
|
| 865 |
+
"action_mse_to_best": 0.4219050041456585,
|
| 866 |
+
"per_task": {
|
| 867 |
+
"LiftPegUpright-v1": {
|
| 868 |
+
"action_mse_to_best": 0.3689509636800115,
|
| 869 |
+
"expert_success_rate": 0.8229166666666666,
|
| 870 |
+
"num_groups": 96,
|
| 871 |
+
"oracle_success_rate": 0.9270833333333334,
|
| 872 |
+
"policy_expert_regret": 0.9988786199440559,
|
| 873 |
+
"policy_oracle_regret": 1.0655400585383177,
|
| 874 |
+
"policy_rollout_progress": 0.6359797989328703,
|
| 875 |
+
"policy_rollout_success_rate": 0.21875,
|
| 876 |
+
"restore_max_error": 3.5762786865234375e-07
|
| 877 |
+
},
|
| 878 |
+
"PickCube-v1": {
|
| 879 |
+
"action_mse_to_best": 0.36036560877292145,
|
| 880 |
+
"expert_success_rate": 0.9444444444444444,
|
| 881 |
+
"num_groups": 198,
|
| 882 |
+
"oracle_success_rate": 0.9595959595959596,
|
| 883 |
+
"policy_expert_regret": 1.1916989847980064,
|
| 884 |
+
"policy_oracle_regret": 1.2129739373734203,
|
| 885 |
+
"policy_rollout_progress": 0.5399239445763706,
|
| 886 |
+
"policy_rollout_success_rate": 0.20202020202020202,
|
| 887 |
+
"restore_max_error": 4.76837158203125e-07
|
| 888 |
+
},
|
| 889 |
+
"PullCube-v1": {
|
| 890 |
+
"action_mse_to_best": 0.5407536560048659,
|
| 891 |
+
"expert_success_rate": 0.24444444444444444,
|
| 892 |
+
"num_groups": 90,
|
| 893 |
+
"oracle_success_rate": 0.4666666666666667,
|
| 894 |
+
"policy_expert_regret": 0.22751877215448782,
|
| 895 |
+
"policy_oracle_regret": 0.41793426646401916,
|
| 896 |
+
"policy_rollout_progress": 0.41927271163850593,
|
| 897 |
+
"policy_rollout_success_rate": 0.3,
|
| 898 |
+
"restore_max_error": 4.0978193283081055e-07
|
| 899 |
+
},
|
| 900 |
+
"PushCube-v1": {
|
| 901 |
+
"action_mse_to_best": 0.3965011129874994,
|
| 902 |
+
"expert_success_rate": 0.8514851485148515,
|
| 903 |
+
"num_groups": 101,
|
| 904 |
+
"oracle_success_rate": 1.0,
|
| 905 |
+
"policy_expert_regret": 0.4412403431269202,
|
| 906 |
+
"policy_oracle_regret": 0.477747778786291,
|
| 907 |
+
"policy_rollout_progress": 0.7697769736889566,
|
| 908 |
+
"policy_rollout_success_rate": 0.7524752475247525,
|
| 909 |
+
"restore_max_error": 4.76837158203125e-07
|
| 910 |
+
},
|
| 911 |
+
"StackCube-v1": {
|
| 912 |
+
"action_mse_to_best": 0.523436143125097,
|
| 913 |
+
"expert_success_rate": 0.7666666666666667,
|
| 914 |
+
"num_groups": 90,
|
| 915 |
+
"oracle_success_rate": 0.9111111111111111,
|
| 916 |
+
"policy_expert_regret": 1.3249462943938044,
|
| 917 |
+
"policy_oracle_regret": 1.4735899850726129,
|
| 918 |
+
"policy_rollout_progress": 0.33269852383269205,
|
| 919 |
+
"policy_rollout_success_rate": 0.08888888888888889,
|
| 920 |
+
"restore_max_error": 4.76837158203125e-07
|
| 921 |
+
}
|
| 922 |
+
}
|
| 923 |
+
}
|
| 924 |
+
]
|
| 925 |
+
}
|
results/h16_field_sweep_summary.md
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# h=16 Field-Guided Rollout Sweep
|
| 2 |
+
|
| 3 |
+
Result root: `/scratch/knguy52/dovla/experiments/dovla_h16_field_sweep`
|
| 4 |
+
Completed result files: 12
|
| 5 |
+
Baseline h=4 policy success: 29.67%
|
| 6 |
+
Baseline h=16 policy success: 29.74%
|
| 7 |
+
|
| 8 |
+
| config | seeds | mean success | gain vs h=16 | progress | action MSE |
|
| 9 |
+
|---|---:|---:|---:|---:|---:|
|
| 10 |
+
| k32_sigma0.35 | 3 | 29.10% | -0.64% | 53.44% | 0.416 |
|
| 11 |
+
| k8_sigma0.10 | 3 | 28.70% | -1.04% | 53.77% | 0.402 |
|
| 12 |
+
| k64_sigma0.50 | 3 | 28.64% | -1.10% | 52.89% | 0.425 |
|
| 13 |
+
| k16_sigma0.20 | 3 | 28.52% | -1.22% | 53.17% | 0.407 |
|
| 14 |
+
|
| 15 |
+
Best config:
|
| 16 |
+
- k32_sigma0.35
|
| 17 |
+
- mean success: 29.10%
|
| 18 |
+
- gain vs h=16 policy: -0.64%
|
results/h16_lattice_rollout_summary.json
ADDED
|
@@ -0,0 +1,228 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"run_root": "/scratch/knguy52/dovla/experiments/dovla_h16_rollout_runs",
|
| 3 |
+
"out_name": "lattice_rollout.json",
|
| 4 |
+
"num_completed": 3,
|
| 5 |
+
"baseline_h4_policy_success": 0.2967,
|
| 6 |
+
"baseline_h16_policy_success": 0.29739130434782607,
|
| 7 |
+
"mean_success": 0.6933333333333334,
|
| 8 |
+
"std_success": 0.035655708555783816,
|
| 9 |
+
"mean_progress": 0.8108624100247803,
|
| 10 |
+
"mean_oracle_success": 0.8678260869565217,
|
| 11 |
+
"mean_action_mse_to_best": 0.43754452051773696,
|
| 12 |
+
"gain_vs_h4": 0.39663333333333334,
|
| 13 |
+
"gain_vs_h16_policy": 0.3959420289855073,
|
| 14 |
+
"selected_candidate_type_counts": {
|
| 15 |
+
"lattice_expert": 977,
|
| 16 |
+
"lattice_near_miss": 348,
|
| 17 |
+
"lattice_no_op": 177,
|
| 18 |
+
"lattice_random_negative": 138,
|
| 19 |
+
"lattice_wrong_direction": 30,
|
| 20 |
+
"lattice_wrong_gripper": 55
|
| 21 |
+
},
|
| 22 |
+
"rows": [
|
| 23 |
+
{
|
| 24 |
+
"seed": 0,
|
| 25 |
+
"path": "/scratch/knguy52/dovla/experiments/dovla_h16_rollout_runs/seed_0/lattice_rollout.json",
|
| 26 |
+
"selection_mode": "lattice",
|
| 27 |
+
"num_candidates": 16,
|
| 28 |
+
"num_groups": 575,
|
| 29 |
+
"policy_rollout_success_rate": 0.6521739130434783,
|
| 30 |
+
"policy_rollout_progress": 0.792444871255885,
|
| 31 |
+
"oracle_success_rate": 0.8573913043478261,
|
| 32 |
+
"action_mse_to_best": 0.414993303792207,
|
| 33 |
+
"per_task": {
|
| 34 |
+
"LiftPegUpright-v1": {
|
| 35 |
+
"action_mse_to_best": 0.28245431445983055,
|
| 36 |
+
"expert_success_rate": 0.8865979381443299,
|
| 37 |
+
"num_groups": 97,
|
| 38 |
+
"oracle_success_rate": 0.9278350515463918,
|
| 39 |
+
"policy_expert_regret": 0.2610538084175169,
|
| 40 |
+
"policy_oracle_regret": 0.27425835227843415,
|
| 41 |
+
"policy_rollout_progress": 0.9071177396270418,
|
| 42 |
+
"policy_rollout_success_rate": 0.7216494845360825,
|
| 43 |
+
"restore_max_error": 4.76837158203125e-07
|
| 44 |
+
},
|
| 45 |
+
"PickCube-v1": {
|
| 46 |
+
"action_mse_to_best": 0.41056775828250325,
|
| 47 |
+
"expert_success_rate": 0.9375,
|
| 48 |
+
"num_groups": 208,
|
| 49 |
+
"oracle_success_rate": 0.9471153846153846,
|
| 50 |
+
"policy_expert_regret": 0.26980068733414203,
|
| 51 |
+
"policy_oracle_regret": 0.2764672613881815,
|
| 52 |
+
"policy_rollout_progress": 0.8699935431269785,
|
| 53 |
+
"policy_rollout_success_rate": 0.7788461538461539,
|
| 54 |
+
"restore_max_error": 4.76837158203125e-07
|
| 55 |
+
},
|
| 56 |
+
"PullCube-v1": {
|
| 57 |
+
"action_mse_to_best": 0.5239523940465667,
|
| 58 |
+
"expert_success_rate": 0.19480519480519481,
|
| 59 |
+
"num_groups": 77,
|
| 60 |
+
"oracle_success_rate": 0.36363636363636365,
|
| 61 |
+
"policy_expert_regret": 0.15346811585999154,
|
| 62 |
+
"policy_oracle_regret": 0.37490944806244464,
|
| 63 |
+
"policy_rollout_progress": 0.3196387503828321,
|
| 64 |
+
"policy_rollout_success_rate": 0.16883116883116883,
|
| 65 |
+
"restore_max_error": 4.76837158203125e-07
|
| 66 |
+
},
|
| 67 |
+
"PushCube-v1": {
|
| 68 |
+
"action_mse_to_best": 0.38503194246143896,
|
| 69 |
+
"expert_success_rate": 0.8617021276595744,
|
| 70 |
+
"num_groups": 94,
|
| 71 |
+
"oracle_success_rate": 0.9893617021276596,
|
| 72 |
+
"policy_expert_regret": 0.28232911355952,
|
| 73 |
+
"policy_oracle_regret": 0.37923398661486646,
|
| 74 |
+
"policy_rollout_progress": 0.8140388716091501,
|
| 75 |
+
"policy_rollout_success_rate": 0.7872340425531915,
|
| 76 |
+
"restore_max_error": 4.76837158203125e-07
|
| 77 |
+
},
|
| 78 |
+
"StackCube-v1": {
|
| 79 |
+
"action_mse_to_best": 0.4978550557797361,
|
| 80 |
+
"expert_success_rate": 0.6767676767676768,
|
| 81 |
+
"num_groups": 99,
|
| 82 |
+
"oracle_success_rate": 0.8585858585858586,
|
| 83 |
+
"policy_expert_regret": 0.2738224623179195,
|
| 84 |
+
"policy_oracle_regret": 0.4027370100069528,
|
| 85 |
+
"policy_rollout_progress": 0.8643927833046576,
|
| 86 |
+
"policy_rollout_success_rate": 0.5656565656565656,
|
| 87 |
+
"restore_max_error": 3.948807716369629e-07
|
| 88 |
+
}
|
| 89 |
+
}
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"seed": 1,
|
| 93 |
+
"path": "/scratch/knguy52/dovla/experiments/dovla_h16_rollout_runs/seed_1/lattice_rollout.json",
|
| 94 |
+
"selection_mode": "lattice",
|
| 95 |
+
"num_candidates": 16,
|
| 96 |
+
"num_groups": 575,
|
| 97 |
+
"policy_rollout_success_rate": 0.7147826086956521,
|
| 98 |
+
"policy_rollout_progress": 0.8205086112848442,
|
| 99 |
+
"oracle_success_rate": 0.8695652173913043,
|
| 100 |
+
"action_mse_to_best": 0.41892785357189893,
|
| 101 |
+
"per_task": {
|
| 102 |
+
"LiftPegUpright-v1": {
|
| 103 |
+
"action_mse_to_best": 0.31486138409473396,
|
| 104 |
+
"expert_success_rate": 0.8584070796460177,
|
| 105 |
+
"num_groups": 113,
|
| 106 |
+
"oracle_success_rate": 0.9380530973451328,
|
| 107 |
+
"policy_expert_regret": 0.2672024345503444,
|
| 108 |
+
"policy_oracle_regret": 0.2922932648025783,
|
| 109 |
+
"policy_rollout_progress": 0.8941378234762006,
|
| 110 |
+
"policy_rollout_success_rate": 0.7345132743362832,
|
| 111 |
+
"restore_max_error": 4.76837158203125e-07
|
| 112 |
+
},
|
| 113 |
+
"PickCube-v1": {
|
| 114 |
+
"action_mse_to_best": 0.44310908647173125,
|
| 115 |
+
"expert_success_rate": 0.9402173913043478,
|
| 116 |
+
"num_groups": 184,
|
| 117 |
+
"oracle_success_rate": 0.9456521739130435,
|
| 118 |
+
"policy_expert_regret": 0.17421010318819596,
|
| 119 |
+
"policy_oracle_regret": 0.1742341076631261,
|
| 120 |
+
"policy_rollout_progress": 0.9089716651436427,
|
| 121 |
+
"policy_rollout_success_rate": 0.8478260869565217,
|
| 122 |
+
"restore_max_error": 4.76837158203125e-07
|
| 123 |
+
},
|
| 124 |
+
"PullCube-v1": {
|
| 125 |
+
"action_mse_to_best": 0.5217870954473159,
|
| 126 |
+
"expert_success_rate": 0.25,
|
| 127 |
+
"num_groups": 76,
|
| 128 |
+
"oracle_success_rate": 0.40789473684210525,
|
| 129 |
+
"policy_expert_regret": 0.12901650541952173,
|
| 130 |
+
"policy_oracle_regret": 0.3323121542037514,
|
| 131 |
+
"policy_rollout_progress": 0.379051765659824,
|
| 132 |
+
"policy_rollout_success_rate": 0.23684210526315788,
|
| 133 |
+
"restore_max_error": 4.76837158203125e-07
|
| 134 |
+
},
|
| 135 |
+
"PushCube-v1": {
|
| 136 |
+
"action_mse_to_best": 0.34412760804365344,
|
| 137 |
+
"expert_success_rate": 0.8198198198198198,
|
| 138 |
+
"num_groups": 111,
|
| 139 |
+
"oracle_success_rate": 1.0,
|
| 140 |
+
"policy_expert_regret": 0.1992848778630162,
|
| 141 |
+
"policy_oracle_regret": 0.3170433482071301,
|
| 142 |
+
"policy_rollout_progress": 0.8541278229640411,
|
| 143 |
+
"policy_rollout_success_rate": 0.8288288288288288,
|
| 144 |
+
"restore_max_error": 4.76837158203125e-07
|
| 145 |
+
},
|
| 146 |
+
"StackCube-v1": {
|
| 147 |
+
"action_mse_to_best": 0.5045947664120531,
|
| 148 |
+
"expert_success_rate": 0.6923076923076923,
|
| 149 |
+
"num_groups": 91,
|
| 150 |
+
"oracle_success_rate": 0.8571428571428571,
|
| 151 |
+
"policy_expert_regret": 0.1715098601448667,
|
| 152 |
+
"policy_oracle_regret": 0.2727490429367338,
|
| 153 |
+
"policy_rollout_progress": 0.8778897638504322,
|
| 154 |
+
"policy_rollout_success_rate": 0.6813186813186813,
|
| 155 |
+
"restore_max_error": 3.948807716369629e-07
|
| 156 |
+
}
|
| 157 |
+
}
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"seed": 2,
|
| 161 |
+
"path": "/scratch/knguy52/dovla/experiments/dovla_h16_rollout_runs/seed_2/lattice_rollout.json",
|
| 162 |
+
"selection_mode": "lattice",
|
| 163 |
+
"num_candidates": 16,
|
| 164 |
+
"num_groups": 575,
|
| 165 |
+
"policy_rollout_success_rate": 0.7130434782608696,
|
| 166 |
+
"policy_rollout_progress": 0.8196337475336116,
|
| 167 |
+
"oracle_success_rate": 0.8765217391304347,
|
| 168 |
+
"action_mse_to_best": 0.47871240418910493,
|
| 169 |
+
"per_task": {
|
| 170 |
+
"LiftPegUpright-v1": {
|
| 171 |
+
"action_mse_to_best": 0.33342877044075675,
|
| 172 |
+
"expert_success_rate": 0.8229166666666666,
|
| 173 |
+
"num_groups": 96,
|
| 174 |
+
"oracle_success_rate": 0.9270833333333334,
|
| 175 |
+
"policy_expert_regret": 0.22194465808570385,
|
| 176 |
+
"policy_oracle_regret": 0.28883908068140346,
|
| 177 |
+
"policy_rollout_progress": 0.9184654237081608,
|
| 178 |
+
"policy_rollout_success_rate": 0.7395833333333334,
|
| 179 |
+
"restore_max_error": 3.5762786865234375e-07
|
| 180 |
+
},
|
| 181 |
+
"PickCube-v1": {
|
| 182 |
+
"action_mse_to_best": 0.4627937324570887,
|
| 183 |
+
"expert_success_rate": 0.9444444444444444,
|
| 184 |
+
"num_groups": 198,
|
| 185 |
+
"oracle_success_rate": 0.9595959595959596,
|
| 186 |
+
"policy_expert_regret": 0.10305700688199564,
|
| 187 |
+
"policy_oracle_regret": 0.11014045741070401,
|
| 188 |
+
"policy_rollout_progress": 0.9385366825518584,
|
| 189 |
+
"policy_rollout_success_rate": 0.898989898989899,
|
| 190 |
+
"restore_max_error": 4.76837158203125e-07
|
| 191 |
+
},
|
| 192 |
+
"PullCube-v1": {
|
| 193 |
+
"action_mse_to_best": 0.6827966811549332,
|
| 194 |
+
"expert_success_rate": 0.24444444444444444,
|
| 195 |
+
"num_groups": 90,
|
| 196 |
+
"oracle_success_rate": 0.4666666666666667,
|
| 197 |
+
"policy_expert_regret": 0.15796189763479762,
|
| 198 |
+
"policy_oracle_regret": 0.40946051337652734,
|
| 199 |
+
"policy_rollout_progress": 0.39318432178762225,
|
| 200 |
+
"policy_rollout_success_rate": 0.26666666666666666,
|
| 201 |
+
"restore_max_error": 4.0978193283081055e-07
|
| 202 |
+
},
|
| 203 |
+
"PushCube-v1": {
|
| 204 |
+
"action_mse_to_best": 0.34167051138264115,
|
| 205 |
+
"expert_success_rate": 0.8514851485148515,
|
| 206 |
+
"num_groups": 101,
|
| 207 |
+
"oracle_success_rate": 1.0,
|
| 208 |
+
"policy_expert_regret": 0.24242213811024582,
|
| 209 |
+
"policy_oracle_regret": 0.29742342971339086,
|
| 210 |
+
"policy_rollout_progress": 0.8609924118707676,
|
| 211 |
+
"policy_rollout_success_rate": 0.8415841584158416,
|
| 212 |
+
"restore_max_error": 4.76837158203125e-07
|
| 213 |
+
},
|
| 214 |
+
"StackCube-v1": {
|
| 215 |
+
"action_mse_to_best": 0.6184098718480932,
|
| 216 |
+
"expert_success_rate": 0.7666666666666667,
|
| 217 |
+
"num_groups": 90,
|
| 218 |
+
"oracle_success_rate": 0.9111111111111111,
|
| 219 |
+
"policy_expert_regret": 0.3941795117325253,
|
| 220 |
+
"policy_oracle_regret": 0.4843028916252984,
|
| 221 |
+
"policy_rollout_progress": 0.8326626494526863,
|
| 222 |
+
"policy_rollout_success_rate": 0.5777777777777777,
|
| 223 |
+
"restore_max_error": 4.76837158203125e-07
|
| 224 |
+
}
|
| 225 |
+
}
|
| 226 |
+
}
|
| 227 |
+
]
|
| 228 |
+
}
|
results/h16_lattice_rollout_summary.md
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# h=16 Lattice-Selected Rollout
|
| 2 |
+
|
| 3 |
+
Run root: `/scratch/knguy52/dovla/experiments/dovla_h16_rollout_runs`
|
| 4 |
+
Completed seeds: 3
|
| 5 |
+
Baseline h=4 policy success: 29.67%
|
| 6 |
+
Baseline h=16 policy success: 29.74%
|
| 7 |
+
|
| 8 |
+
Mean success: 69.33% +/- 3.57%
|
| 9 |
+
Gain vs h=16 policy: +39.59%
|
| 10 |
+
Mean oracle success: 86.78%
|
| 11 |
+
Mean progress: 81.09%
|
| 12 |
+
|
| 13 |
+
| seed | success | progress | oracle | candidates | action MSE |
|
| 14 |
+
|---:|---:|---:|---:|---:|---:|
|
| 15 |
+
| 0 | 65.22% | 79.24% | 85.74% | 16 | 0.415 |
|
| 16 |
+
| 1 | 71.48% | 82.05% | 86.96% | 16 | 0.419 |
|
| 17 |
+
| 2 | 71.30% | 81.96% | 87.65% | 16 | 0.479 |
|
| 18 |
+
|
| 19 |
+
Selected candidate types:
|
| 20 |
+
- lattice_expert: 977
|
| 21 |
+
- lattice_near_miss: 348
|
| 22 |
+
- lattice_no_op: 177
|
| 23 |
+
- lattice_random_negative: 138
|
| 24 |
+
- lattice_wrong_direction: 30
|
| 25 |
+
- lattice_wrong_gripper: 55
|
scripts/eval_maniskill_policy_rollout.py
CHANGED
|
@@ -62,7 +62,16 @@ def main(argv: list[str] | None = None) -> int:
|
|
| 62 |
default=0,
|
| 63 |
help="Base RNG seed for candidate perturbations.",
|
| 64 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 65 |
args = parser.parse_args(argv)
|
|
|
|
|
|
|
|
|
|
| 66 |
result = evaluate_maniskill_policy_rollout(
|
| 67 |
args.checkpoint,
|
| 68 |
args.dataset,
|
|
@@ -77,6 +86,7 @@ def main(argv: list[str] | None = None) -> int:
|
|
| 77 |
num_candidates=args.num_candidates,
|
| 78 |
candidate_sigma=args.candidate_sigma,
|
| 79 |
selection_seed=args.selection_seed,
|
|
|
|
| 80 |
)
|
| 81 |
print(json.dumps({key: value for key, value in result.items() if key != "rows"}, indent=2))
|
| 82 |
return 0
|
|
|
|
| 62 |
default=0,
|
| 63 |
help="Base RNG seed for candidate perturbations.",
|
| 64 |
)
|
| 65 |
+
parser.add_argument(
|
| 66 |
+
"--lattice-exclude-types",
|
| 67 |
+
default="",
|
| 68 |
+
help="Comma-separated candidate_type values to mask in selection-mode=lattice, "
|
| 69 |
+
"for example 'expert'.",
|
| 70 |
+
)
|
| 71 |
args = parser.parse_args(argv)
|
| 72 |
+
lattice_exclude_types = tuple(
|
| 73 |
+
item.strip() for item in args.lattice_exclude_types.split(",") if item.strip()
|
| 74 |
+
)
|
| 75 |
result = evaluate_maniskill_policy_rollout(
|
| 76 |
args.checkpoint,
|
| 77 |
args.dataset,
|
|
|
|
| 86 |
num_candidates=args.num_candidates,
|
| 87 |
candidate_sigma=args.candidate_sigma,
|
| 88 |
selection_seed=args.selection_seed,
|
| 89 |
+
lattice_exclude_types=lattice_exclude_types,
|
| 90 |
)
|
| 91 |
print(json.dumps({key: value for key, value in result.items() if key != "rows"}, indent=2))
|
| 92 |
return 0
|