229 lines
7.2 KiB
JSON
229 lines
7.2 KiB
JSON
{
|
|
"status": "PLAN_ONLY_NO_MAIN_EXECUTION",
|
|
"neural_runs": 90,
|
|
"arms": [
|
|
"N",
|
|
"E"
|
|
],
|
|
"folds": 5,
|
|
"component_fraction_percent": [
|
|
25,
|
|
50,
|
|
100
|
|
],
|
|
"seeds": [
|
|
0,
|
|
1,
|
|
2
|
|
],
|
|
"width": 128,
|
|
"width_selection": "feasibility; width effect not tested",
|
|
"E_max": 42,
|
|
"cosine_T_max": 42,
|
|
"baseline_ready_nodes": 14555918,
|
|
"baseline_worker_guard_wall_seconds": {
|
|
"N": 1129.3405424560187,
|
|
"E": 1121.3519870990422
|
|
},
|
|
"node_scaled_gpu_hours": 19.942144065735626,
|
|
"node_scaled_plus25percent_and_idle_hours": 25.42768008216953,
|
|
"all90_priced_at_measured_full_pool_hours": 28.133656619438263,
|
|
"all90_full_pool_plus25percent_and_idle_hours": 35.66707077429783,
|
|
"proposed_per_run_wall_limit_seconds": 1800,
|
|
"90_run_caps_plus_expected_idle_gpu_hours": 45.5,
|
|
"proposed_global_neural_wall_limit_hours": 46,
|
|
"R2_final_fits": 15,
|
|
"R2_HPO_configurations_per_fold_pool_max": 12,
|
|
"R2_HPO_trials_max": 180,
|
|
"R2_CPU_time": "UNMEASURED_NOT_INCLUDED",
|
|
"preparation_completion_cost": "NOT_INCLUDED_125_UNPREPARED_AND_6_TIMEOUTS_REQUIRE_SEPARATE_PLAN",
|
|
"estimation_limits": [
|
|
"Composite per-node scaling cannot separate 128 training/inference/setup rates",
|
|
"Measured one fold/full successful pool/seed; no confidence bound",
|
|
"Allocator change may change timing; proposed reserve is engineering allowance",
|
|
"No early-stopping savings assumed",
|
|
"Global proposed cap includes waits/setup, may stop before all90; no automatic extra budget",
|
|
"GPU-hours are aggregate device time, not multi-GPU elapsed promise"
|
|
],
|
|
"rows": [
|
|
{
|
|
"fold": 0,
|
|
"component_fraction_percent": 25,
|
|
"learning_parts": 145,
|
|
"learning_nodes": 4319375,
|
|
"validation_parts": 112,
|
|
"validation_nodes": 3330152,
|
|
"projected_N_seconds": 593.4988759700324,
|
|
"projected_E_seconds": 589.3006749431931
|
|
},
|
|
{
|
|
"fold": 0,
|
|
"component_fraction_percent": 50,
|
|
"learning_parts": 254,
|
|
"learning_nodes": 7524426,
|
|
"validation_parts": 112,
|
|
"validation_nodes": 3330152,
|
|
"projected_N_seconds": 842.1670832888153,
|
|
"projected_E_seconds": 836.2098913597581
|
|
},
|
|
{
|
|
"fold": 0,
|
|
"component_fraction_percent": 100,
|
|
"learning_parts": 387,
|
|
"learning_nodes": 11863762,
|
|
"validation_parts": 112,
|
|
"validation_nodes": 3330152,
|
|
"projected_N_seconds": 1178.840323144861,
|
|
"projected_E_seconds": 1170.501623855806
|
|
},
|
|
{
|
|
"fold": 1,
|
|
"component_fraction_percent": 25,
|
|
"learning_parts": 85,
|
|
"learning_nodes": 2401741,
|
|
"validation_parts": 90,
|
|
"validation_nodes": 3242126,
|
|
"projected_N_seconds": 437.88703806449195,
|
|
"projected_E_seconds": 434.78958011255014
|
|
},
|
|
{
|
|
"fold": 1,
|
|
"component_fraction_percent": 50,
|
|
"learning_parts": 180,
|
|
"learning_nodes": 4549249,
|
|
"validation_parts": 90,
|
|
"validation_nodes": 3242126,
|
|
"projected_N_seconds": 604.5043444857455,
|
|
"projected_E_seconds": 600.2282946691372
|
|
},
|
|
{
|
|
"fold": 1,
|
|
"component_fraction_percent": 100,
|
|
"learning_parts": 409,
|
|
"learning_nodes": 11386113,
|
|
"validation_parts": 90,
|
|
"validation_nodes": 3242126,
|
|
"projected_N_seconds": 1134.9516648442434,
|
|
"projected_E_seconds": 1126.9234183931035
|
|
},
|
|
{
|
|
"fold": 2,
|
|
"component_fraction_percent": 25,
|
|
"learning_parts": 90,
|
|
"learning_nodes": 2998926,
|
|
"validation_parts": 88,
|
|
"validation_nodes": 2648316,
|
|
"projected_N_seconds": 438.14889199433605,
|
|
"projected_E_seconds": 435.0495817803569
|
|
},
|
|
{
|
|
"fold": 2,
|
|
"component_fraction_percent": 50,
|
|
"learning_parts": 203,
|
|
"learning_nodes": 6659657,
|
|
"validation_parts": 88,
|
|
"validation_nodes": 2648316,
|
|
"projected_N_seconds": 722.1716470913051,
|
|
"projected_E_seconds": 717.0632604150582
|
|
},
|
|
{
|
|
"fold": 2,
|
|
"component_fraction_percent": 100,
|
|
"learning_parts": 411,
|
|
"learning_nodes": 12229477,
|
|
"validation_parts": 88,
|
|
"validation_nodes": 2648316,
|
|
"projected_N_seconds": 1154.3136487281913,
|
|
"projected_E_seconds": 1146.1484424546923
|
|
},
|
|
{
|
|
"fold": 3,
|
|
"component_fraction_percent": 25,
|
|
"learning_parts": 128,
|
|
"learning_nodes": 4348385,
|
|
"validation_parts": 144,
|
|
"validation_nodes": 3976319,
|
|
"projected_N_seconds": 645.8833947227367,
|
|
"projected_E_seconds": 641.3146441475793
|
|
},
|
|
{
|
|
"fold": 3,
|
|
"component_fraction_percent": 50,
|
|
"learning_parts": 236,
|
|
"learning_nodes": 7631328,
|
|
"validation_parts": 144,
|
|
"validation_nodes": 3976319,
|
|
"projected_N_seconds": 900.5949579832738,
|
|
"projected_E_seconds": 894.224467944532
|
|
},
|
|
{
|
|
"fold": 3,
|
|
"component_fraction_percent": 100,
|
|
"learning_parts": 355,
|
|
"learning_nodes": 11031937,
|
|
"validation_parts": 144,
|
|
"validation_nodes": 3976319,
|
|
"projected_N_seconds": 1164.435796653897,
|
|
"projected_E_seconds": 1156.1989898879015
|
|
},
|
|
{
|
|
"fold": 4,
|
|
"component_fraction_percent": 25,
|
|
"learning_parts": 58,
|
|
"learning_nodes": 1748016,
|
|
"validation_parts": 114,
|
|
"validation_nodes": 3401388,
|
|
"projected_N_seconds": 399.52345889041095,
|
|
"projected_E_seconds": 396.69737132180575
|
|
},
|
|
{
|
|
"fold": 4,
|
|
"component_fraction_percent": 50,
|
|
"learning_parts": 144,
|
|
"learning_nodes": 4743744,
|
|
"validation_parts": 114,
|
|
"validation_nodes": 3401388,
|
|
"projected_N_seconds": 631.9510587553376,
|
|
"projected_E_seconds": 627.4808605945702
|
|
},
|
|
{
|
|
"fold": 4,
|
|
"component_fraction_percent": 100,
|
|
"learning_parts": 386,
|
|
"learning_nodes": 11535306,
|
|
"validation_parts": 114,
|
|
"validation_nodes": 3401388,
|
|
"projected_N_seconds": 1158.883562305006,
|
|
"projected_E_seconds": 1150.6860300800224
|
|
}
|
|
],
|
|
"preparation_proposal": {
|
|
"new_parts_without_S2_ready": 125,
|
|
"their_nodes": 3467310,
|
|
"retained_true_timeout_parts": 6,
|
|
"conservative_observed_cpu_seconds_per_original_planned_node": 0.008235846461945216,
|
|
"rate_includes_prior_aborted_and_recovery_cost": true,
|
|
"new125_projected_worker_hours": 7.932286887768686,
|
|
"six_timeout_repreparations_separate_60min_each_cap_hours": 6,
|
|
"sum_with25percent_reserve_hours": 17.415358609710857,
|
|
"proposed_new_worker_cpu_cap_hours": 20,
|
|
"proposed_cumulative_output_disk_GiB": 100,
|
|
"proposal_only_no_execution": true,
|
|
"no_labels_from_current_sealed_test_parts_read": true,
|
|
"label_policy": "125 fold0 tests: geometry/features only; labels decoded only at custody step; no label reads in this planning turn"
|
|
},
|
|
"R2_proposal": {
|
|
"status": "UNMEASURED_COST_CAP_PROPOSAL_ONLY",
|
|
"single_thread": true,
|
|
"fits_max_including_HPO_and_final": 195,
|
|
"proposed_per_fit_cpu_seconds": 1800,
|
|
"proposed_total_worker_cpu_hours": 100,
|
|
"initial_checkpoint": "Measure one authorized fixed fit before dispatching full search; no current fit executed",
|
|
"no_completion_guarantee": true
|
|
},
|
|
"final_roster_ruling": "1794698734.967459",
|
|
"neural_runs_fresh": 90,
|
|
"pilot_cells_reused": 0,
|
|
"reuse_policy": "Pilot cells remain historical pilot evidence; no pilot weights reused or bound into matrix"
|
|
}
|