{ "bench_version": 1, "track": "wedge_graph_regression", "model": "mlp_tiny", "metric_direction": "lower is better", "n_seeds": 8, "baseline": { "best_cfg": { "lr": 0.003, "epochs": 18 }, "sweep": [ { "cfg": { "lr": 0.001, "epochs": 18 }, "mean": 0.061101581901311874 }, { "cfg": { "lr": 0.003, "epochs": 18 }, "mean": 0.059752004221081734 }, { "cfg": { "lr": 0.01, "epochs": 18 }, "mean": 0.07423717714846134 } ], "full": { "mean": 0.059744860511273146, "std": 0.007734674460330792, "per_seed": [ 0.05919400602579117, 0.052641332149505615, 0.07083674520254135, 0.0563359335064888, 0.06298892945051193, 0.04989643022418022, 0.05397595465183258, 0.0720895528793335 ], "n": 8 } }, "idea": { "mean": 0.013510067714378238, "std": 0.0028306420950107115, "per_seed": [ 0.011363038793206215, 0.012721219100058079, 0.012768835760653019, 0.012967556715011597, 0.009611863642930984, 0.013135809451341629, 0.019576922059059143, 0.015935296192765236 ], "n": 8, "signature": [ { "pred_std": 0.16127879917621613, "obs_std": 0.21961554884910583, "pred_obs_corr": 0.887711709058279 }, { "pred_std": 0.1682410091161728, "obs_std": 0.21719904243946075, "pred_obs_corr": 0.8598633105317619 }, { "pred_std": 0.18861736357212067, "obs_std": 0.23979876935482025, "pred_obs_corr": 0.8878919008544431 }, { "pred_std": 0.1867043375968933, "obs_std": 0.22483287751674652, "pred_obs_corr": 0.8631346192234685 }, { "pred_std": 0.16417063772678375, "obs_std": 0.21063706278800964, "pred_obs_corr": 0.8946031027067823 }, { "pred_std": 0.17319384217262268, "obs_std": 0.22647514939308167, "pred_obs_corr": 0.8694575455010861 }, { "pred_std": 0.14724908769130707, "obs_std": 0.20635201036930084, "pred_obs_corr": 0.7448891050585023 }, { "pred_std": 0.16360855102539062, "obs_std": 0.22824695706367493, "pred_obs_corr": 0.8425912593747041 } ] }, "comparison": { "delta_mean": -0.04623479279689491, "idea_wins": 8, "n_pairs": 8, "per_seed_diffs": [ -0.04783096723258495, -0.039920113049447536, -0.05806790944188833, -0.0433683767914772, -0.05337706580758095, -0.03676062077283859, -0.03439903259277344, -0.05615425668656826 ], "p_value": 0.0081, "mde": 0.007544702982030243, "mde_rel_pct": 12.628204196085866, "verdict": "idea better (significant)", "system_worked": true }, "custom_track": { "name": "wedge_graph_regression", "file": "graph_wedge_bench.py", "domain": "graph-nn" }, "idea_grid": { "0.001": { "mean": 0.041265785694122314, "std": 0.0038461557188544885, "per_seed": [ 0.03932115435600281, 0.03908920660614967, 0.03921026363968849, 0.04619516804814339, 0.03932721167802811, 0.03928273543715477, 0.038361623883247375, 0.049338921904563904 ], "n": 8, "signature": [ { "pred_std": 0.10499881207942963, "obs_std": 0.21961554884910583, "pred_obs_corr": 0.4330130138308088 }, { "pred_std": 0.08769117295742035, "obs_std": 0.21719904243946075, "pred_obs_corr": 0.43490863320131024 }, { "pred_std": 0.0983906164765358, "obs_std": 0.23979876935482025, "pred_obs_corr": 0.5928418270204208 }, { "pred_std": 0.09753165394067764, "obs_std": 0.22483287751674652, "pred_obs_corr": 0.31636030316148195 }, { "pred_std": 0.11598880589008331, "obs_std": 0.21063706278800964, "pred_obs_corr": 0.3799591849447861 }, { "pred_std": 0.09304504096508026, "obs_std": 0.22647514939308167, "pred_obs_corr": 0.49060817556425573 }, { "pred_std": 0.10514671355485916, "obs_std": 0.20635201036930084, "pred_obs_corr": 0.35432711607503614 }, { "pred_std": 0.08850710839033127, "obs_std": 0.22824695706367493, "pred_obs_corr": 0.2645536283175189 } ] }, "0.003": { "mean": 0.024491973454132676, "std": 0.0033225946932239044, "per_seed": [ 0.02352076955139637, 0.022262291982769966, 0.022041544318199158, 0.023201245814561844, 0.020999116823077202, 0.024190735071897507, 0.028309449553489685, 0.03141063451766968 ], "n": 8, "signature": [ { "pred_std": 0.1270151138305664, "obs_std": 0.21961554884910583, "pred_obs_corr": 0.7369194685495051 }, { "pred_std": 0.1252332478761673, "obs_std": 0.21719904243946075, "pred_obs_corr": 0.7475148343284106 }, { "pred_std": 0.1388101875782013, "obs_std": 0.23979876935482025, "pred_obs_corr": 0.8221710532450354 }, { "pred_std": 0.13323701918125153, "obs_std": 0.22483287751674652, "pred_obs_corr": 0.7602398729509038 }, { "pred_std": 0.12740392982959747, "obs_std": 0.21063706278800964, "pred_obs_corr": 0.7434833188067616 }, { "pred_std": 0.12677018344402313, "obs_std": 0.22647514939308167, "pred_obs_corr": 0.7601123761370298 }, { "pred_std": 0.1245916560292244, "obs_std": 0.20635201036930084, "pred_obs_corr": 0.5823476350878336 }, { "pred_std": 0.10668632388114929, "obs_std": 0.22824695706367493, "pred_obs_corr": 0.6635442880841551 } ] }, "0.01": { "mean": 0.013510067714378238, "std": 0.0028306420950107115, "per_seed": [ 0.011363038793206215, 0.012721219100058079, 0.012768835760653019, 0.012967556715011597, 0.009611863642930984, 0.013135809451341629, 0.019576922059059143, 0.015935296192765236 ], "n": 8, "signature": [ { "pred_std": 0.16127879917621613, "obs_std": 0.21961554884910583, "pred_obs_corr": 0.887711709058279 }, { "pred_std": 0.1682410091161728, "obs_std": 0.21719904243946075, "pred_obs_corr": 0.8598633105317619 }, { "pred_std": 0.18861736357212067, "obs_std": 0.23979876935482025, "pred_obs_corr": 0.8878919008544431 }, { "pred_std": 0.1867043375968933, "obs_std": 0.22483287751674652, "pred_obs_corr": 0.8631346192234685 }, { "pred_std": 0.16417063772678375, "obs_std": 0.21063706278800964, "pred_obs_corr": 0.8946031027067823 }, { "pred_std": 0.17319384217262268, "obs_std": 0.22647514939308167, "pred_obs_corr": 0.8694575455010861 }, { "pred_std": 0.14724908769130707, "obs_std": 0.20635201036930084, "pred_obs_corr": 0.7448891050585023 }, { "pred_std": 0.16360855102539062, "obs_std": 0.22824695706367493, "pred_obs_corr": 0.8425912593747041 } ] } }, "mechanism_signature": { "predicted": "graph-aware connected pooling should retain graph-smooth signal (higher prediction/observation correlation than random pooling)", "observed": { "baseline_best_corr_mean": 0.32037841882537543, "idea_corr_mean": 0.8562678190386285 }, "confirmed": false } }