{ "bench_version": 1, "track": "spatial_anchor_regression", "model": "coordinate_mlp", "metric_direction": "lower is better", "n_seeds": 8, "baseline": { "best_cfg": { "lr": 0.01 }, "sweep": [ { "cfg": { "lr": 0.001 }, "mean": 0.14637889340519905 }, { "cfg": { "lr": 0.003 }, "mean": 0.061082218773663044 }, { "cfg": { "lr": 0.01 }, "mean": 0.012440342921763659 } ], "full": { "mean": 0.012712803669273853, "std": 0.0031444912187311223, "per_seed": [ 0.010250885039567947, 0.014543263241648674, 0.013977252878248692, 0.010989970527589321, 0.01702292263507843, 0.01695895567536354, 0.009546724148094654, 0.008412455208599567 ], "n": 8 } }, "idea": { "mean": 0.006530163227580488, "std": 0.0013944948580770671, "per_seed": [ 0.008384107612073421, 0.0057844375260174274, 0.006018985528498888, 0.007370046339929104, 0.005429707467556, 0.008914164267480373, 0.005202051252126694, 0.005137805826961994 ], "n": 8 }, "comparison": { "delta_mean": -0.006182640441693366, "idea_wins": 8, "n_pairs": 8, "per_seed_diffs": [ -0.001866777427494526, -0.008758825715631247, -0.007958267349749804, -0.0036199241876602173, -0.01159321516752243, -0.008044791407883167, -0.00434467289596796, -0.0032746493816375732 ], "p_value": 0.0081, "mde": 0.002818935574904236, "mde_rel_pct": 22.17398811654307, "verdict": "idea better (significant)", "system_worked": true }, "mechanism_signature": { "prediction": "trained PHS residual should be nonzero and align with held-out error correction", "observed_mean_residual_rms": 0.11425309907644987, "observed_mean_correction_corr": 0.7042187005281448, "confirmed": true, "per_seed": [ { "seed": 0, "baseline_mse": 0.010250885039567947, "idea_mse": 0.005912628024816513, "idea_residual_rms": 0.09278345853090286, "residual_target_corr": 0.6705815196037292 }, { "seed": 1, "baseline_mse": 0.014543264172971249, "idea_mse": 0.007037218660116196, "idea_residual_rms": 0.1361566036939621, "residual_target_corr": 0.7668899893760681 }, { "seed": 2, "baseline_mse": 0.01397725474089384, "idea_mse": 0.005881397519260645, "idea_residual_rms": 0.10070868581533432, "residual_target_corr": 0.7634603977203369 }, { "seed": 3, "baseline_mse": 0.010989971458911896, "idea_mse": 0.007288732565939426, "idea_residual_rms": 0.11611808091402054, "residual_target_corr": 0.6222603917121887 }, { "seed": 4, "baseline_mse": 0.01702292263507843, "idea_mse": 0.006018869113177061, "idea_residual_rms": 0.10683124512434006, "residual_target_corr": 0.7492429614067078 }, { "seed": 5, "baseline_mse": 0.01695895753800869, "idea_mse": 0.00841639842838049, "idea_residual_rms": 0.1272411346435547, "residual_target_corr": 0.7258725762367249 }, { "seed": 6, "baseline_mse": 0.009546725079417229, "idea_mse": 0.007395600900053978, "idea_residual_rms": 0.1277274787425995, "residual_target_corr": 0.7541661858558655 }, { "seed": 7, "baseline_mse": 0.008412455208599567, "idea_mse": 0.008615097030997276, "idea_residual_rms": 0.10645810514688492, "residual_target_corr": 0.5812755823135376 } ] }, "custom_track": { "name": "spatial_anchor_regression", "file": "custom_spatial.py", "domain": "spatial neural field / interpolation" }, "idea_selected_cfg": { "lr": 0.01 }, "idea_sweep": [ { "cfg": { "lr": 0.01 }, "mean": 0.006530163227580488, "std": 0.0013944948580770671, "per_seed": [ 0.008384107612073421, 0.0057844375260174274, 0.006018985528498888, 0.007370046339929104, 0.005429707467556, 0.008914164267480373, 0.005202051252126694, 0.005137805826961994 ] }, { "cfg": { "lr": 0.001 }, "mean": 0.10927551612257957, "std": 0.012333453085783437, "per_seed": [ 0.12216655164957047, 0.12041649222373962, 0.12407363951206207, 0.10313449800014496, 0.08412088453769684, 0.10495962202548981, 0.11025018244981766, 0.10508225858211517 ] }, { "cfg": { "lr": 0.003 }, "mean": 0.022391047328710556, "std": 0.008746374453946222, "per_seed": [ 0.028721362352371216, 0.018648821860551834, 0.03950253874063492, 0.01322929561138153, 0.014868687838315964, 0.016263097524642944, 0.030357947573065758, 0.017536627128720284 ] } ] }