{ "bench_version": 1, "track": "dynamics", "model": "rnn_small", "metric_direction": "lower is better", "n_seeds": 8, "baseline": { "best_cfg": { "lr": 0.01, "epochs": 12, "trap_lambda": 0.0 }, "sweep": [ { "cfg": { "lr": 0.001, "epochs": 12, "trap_lambda": 0.0 }, "per_seed": [ 0.02792397141456604, 0.041552960872650146, 0.03038809262216091, 0.03133251145482063, 0.05597708746790886, 0.022259902209043503, 0.07735025882720947, 0.01893964223563671 ], "mean": 0.038215553387999535, "std": 0.019635395061371623 }, { "cfg": { "lr": 0.003, "epochs": 12, "trap_lambda": 0.0 }, "per_seed": [ 0.009584075771272182, 0.013722526840865612, 0.008015970699489117, 0.011999157257378101, 0.01405302993953228, 0.00778926583006978, 0.013168145902454853, 0.007831467315554619 ], "mean": 0.010770454944577068, "std": 0.0027585833372694973 }, { "cfg": { "lr": 0.01, "epochs": 12, "trap_lambda": 0.0 }, "per_seed": [ 0.0024895062670111656, 0.0038835888262838125, 0.003609295468777418, 0.003999150358140469, 0.0032447325065732002, 0.0038587753660976887, 0.008323567919433117, 0.0034713787026703358 ], "mean": 0.004109999426873401, "std": 0.0017693440291274862 }, { "cfg": { "lr": 0.001, "epochs": 12, "trap_lambda": 0.3 }, "per_seed": [ 0.02792397141456604, 0.041552960872650146, 0.03038809262216091, 0.03133251145482063, 0.05597708746790886, 0.022259902209043503, 0.07735025882720947, 0.01893964223563671 ], "mean": 0.038215553387999535, "std": 0.019635395061371623 }, { "cfg": { "lr": 0.003, "epochs": 12, "trap_lambda": 0.3 }, "per_seed": [ 0.009584075771272182, 0.013722526840865612, 0.008015970699489117, 0.011999157257378101, 0.01405302993953228, 0.00778926583006978, 0.013168145902454853, 0.007831467315554619 ], "mean": 0.010770454944577068, "std": 0.0027585833372694973 }, { "cfg": { "lr": 0.01, "epochs": 12, "trap_lambda": 0.3 }, "per_seed": [ 0.0024895062670111656, 0.0038835888262838125, 0.003609295468777418, 0.003999150358140469, 0.0032447325065732002, 0.0038587753660976887, 0.008323567919433117, 0.0034713787026703358 ], "mean": 0.004109999426873401, "std": 0.0017693440291274862 }, { "cfg": { "lr": 0.001, "epochs": 12, "trap_lambda": 1.0 }, "per_seed": [ 0.02792397141456604, 0.041552960872650146, 0.03038809262216091, 0.03133251145482063, 0.05597708746790886, 0.022259902209043503, 0.07735025882720947, 0.01893964223563671 ], "mean": 0.038215553387999535, "std": 0.019635395061371623 }, { "cfg": { "lr": 0.003, "epochs": 12, "trap_lambda": 1.0 }, "per_seed": [ 0.009584075771272182, 0.013722526840865612, 0.008015970699489117, 0.011999157257378101, 0.01405302993953228, 0.00778926583006978, 0.013168145902454853, 0.007831467315554619 ], "mean": 0.010770454944577068, "std": 0.0027585833372694973 }, { "cfg": { "lr": 0.01, "epochs": 12, "trap_lambda": 1.0 }, "per_seed": [ 0.0024895062670111656, 0.0038835888262838125, 0.003609295468777418, 0.003999150358140469, 0.0032447325065732002, 0.0038587753660976887, 0.008323567919433117, 0.0034713787026703358 ], "mean": 0.004109999426873401, "std": 0.0017693440291274862 } ], "harness_tuning": { "best_cfg": { "lr": 0.01, "epochs": 12, "trap_lambda": 0.0 }, "sweep": [ { "cfg": { "lr": 0.001, "epochs": 12, "trap_lambda": 0.0 }, "mean": 0.03279938409104943 }, { "cfg": { "lr": 0.003, "epochs": 12, "trap_lambda": 0.0 }, "mean": 0.010830432642251253 }, { "cfg": { "lr": 0.01, "epochs": 12, "trap_lambda": 0.0 }, "mean": 0.0034953852300532162 }, { "cfg": { "lr": 0.001, "epochs": 12, "trap_lambda": 0.3 }, "mean": 0.03279938409104943 }, { "cfg": { "lr": 0.003, "epochs": 12, "trap_lambda": 0.3 }, "mean": 0.010830432642251253 }, { "cfg": { "lr": 0.01, "epochs": 12, "trap_lambda": 0.3 }, "mean": 0.0034953852300532162 }, { "cfg": { "lr": 0.001, "epochs": 12, "trap_lambda": 1.0 }, "mean": 0.03279938409104943 }, { "cfg": { "lr": 0.003, "epochs": 12, "trap_lambda": 1.0 }, "mean": 0.010830432642251253 }, { "cfg": { "lr": 0.01, "epochs": 12, "trap_lambda": 1.0 }, "mean": 0.0034953852300532162 } ], "full": { "mean": 0.004109999426873401, "std": 0.0016550697890823066, "per_seed": [ 0.0024895062670111656, 0.0038835888262838125, 0.003609295468777418, 0.003999150358140469, 0.0032447325065732002, 0.0038587753660976887, 0.008323567919433117, 0.0034713787026703358 ], "n": 8 } }, "full": { "cfg": { "lr": 0.01, "epochs": 12, "trap_lambda": 0.0 }, "per_seed": [ 0.0024895062670111656, 0.0038835888262838125, 0.003609295468777418, 0.003999150358140469, 0.0032447325065732002, 0.0038587753660976887, 0.008323567919433117, 0.0034713787026703358 ], "mean": 0.004109999426873401, "std": 0.0017693440291274862 } }, "idea": { "best_cfg": { "lr": 0.01, "epochs": 12, "trap_lambda": 0.3 }, "per_seed": [ 0.0020382918883115053, 0.002796109300106764, 0.00355388387106359, 0.0032515127677470446, 0.003949518781155348, 0.003303474746644497, 0.0037955872248858213, 0.004127324093133211 ], "mean": 0.0033519628341309726, "std": 0.0006806487150040169 }, "comparison": { "delta_mean": -0.0007580365927424282, "idea_wins": 6, "n_pairs": 8, "per_seed_diffs": [ -0.0004512143786996603, -0.0010874795261770487, -5.541159771382809e-05, -0.000747637590393424, 0.0007047862745821476, -0.0005553006194531918, -0.004527980694547296, 0.0006559453904628754 ], "p_value": 0.22655, "mde": 0.0013808020917505462, "mde_rel_pct": 33.59616263501437, "verdict": "no significant win", "system_worked": false }, "mechanism_signature": { "track_choice": "dynamics is structurally matched to recurrent stability", "idea_sweep": [ { "cfg": { "lr": 0.001, "epochs": 12, "trap_lambda": 0.0 }, "per_seed": [ 0.02792397141456604, 0.041552960872650146, 0.03038809262216091, 0.03133251145482063, 0.05597708746790886, 0.022259902209043503, 0.07735025882720947, 0.01893964223563671 ], "mean": 0.038215553387999535, "std": 0.019635395061371623 }, { "cfg": { "lr": 0.003, "epochs": 12, "trap_lambda": 0.0 }, "per_seed": [ 0.009584075771272182, 0.013722526840865612, 0.008015970699489117, 0.011999157257378101, 0.01405302993953228, 0.00778926583006978, 0.013168145902454853, 0.007831467315554619 ], "mean": 0.010770454944577068, "std": 0.0027585833372694973 }, { "cfg": { "lr": 0.01, "epochs": 12, "trap_lambda": 0.0 }, "per_seed": [ 0.0024895062670111656, 0.0038835888262838125, 0.003609295468777418, 0.003999150358140469, 0.0032447325065732002, 0.0038587753660976887, 0.008323567919433117, 0.0034713787026703358 ], "mean": 0.004109999426873401, "std": 0.0017693440291274862 }, { "cfg": { "lr": 0.001, "epochs": 12, "trap_lambda": 0.3 }, "per_seed": [ 0.029911765828728676, 0.043103478848934174, 0.029594196006655693, 0.03148665651679039, 0.056626711040735245, 0.023263107985258102, 0.07945620268583298, 0.019382713362574577 ], "mean": 0.03910310403443873, "std": 0.020086107344229626 }, { "cfg": { "lr": 0.003, "epochs": 12, "trap_lambda": 0.3 }, "per_seed": [ 0.00927602220326662, 0.01410755980759859, 0.008517258800566196, 0.011783787980675697, 0.014786372892558575, 0.0075119370594620705, 0.012531742453575134, 0.008021011017262936 ], "mean": 0.010816961526870728, "std": 0.002850141431686086 }, { "cfg": { "lr": 0.01, "epochs": 12, "trap_lambda": 0.3 }, "per_seed": [ 0.0020382918883115053, 0.002796109300106764, 0.00355388387106359, 0.0032515127677470446, 0.003949518781155348, 0.003303474746644497, 0.0037955872248858213, 0.004127324093133211 ], "mean": 0.0033519628341309726, "std": 0.0006806487150040169 }, { "cfg": { "lr": 0.001, "epochs": 12, "trap_lambda": 1.0 }, "per_seed": [ 0.02979419380426407, 0.04312347620725632, 0.02988792397081852, 0.0319940410554409, 0.05702750012278557, 0.02334035187959671, 0.07955670356750488, 0.019355962052941322 ], "mean": 0.039260019082576036, "std": 0.020122028461056908 }, { "cfg": { "lr": 0.003, "epochs": 12, "trap_lambda": 1.0 }, "per_seed": [ 0.009333031252026558, 0.014507703483104706, 0.00847854744642973, 0.012289805337786674, 0.015088483691215515, 0.007256411015987396, 0.012649702839553356, 0.008141638711094856 ], "mean": 0.010968165472149849, "std": 0.0030406270010214335 }, { "cfg": { "lr": 0.01, "epochs": 12, "trap_lambda": 1.0 }, "per_seed": [ 0.0021302360109984875, 0.003130449214950204, 0.003100306261330843, 0.0041888016276061535, 0.004465983249247074, 0.0038384413346648216, 0.0035596308298408985, 0.004667002707719803 ], "mean": 0.0036351064045447856, "std": 0.0008378638946201928 } ], "math_check": { "formula_boundary": 1.2473835212243647, "observed_grid_boundary": 1.25, "absolute_error": 0.0026164787756353114, "confirmed": true }, "mechanism_signature": { "prediction": "trap training should reduce one-step and long-horizon U violations", "observed_one_step_violation": 0.257080078125, "observed_100_step_violation": 0.0, "observed_max_one_step_abs_coordinate": 0.9616694450378418, "baseline_comparison_seed0": { "mse": 0.0024895062670111656, "one_step_violation": 0.8212890625, "mean_violation": 0.9982080078125, "max_norm": 3.4549334049224854, "cloud_diameters": [ 8.314221382141113, 3.868745750423841e-07, 2.440847026718984e-07 ] }, "idea_comparison_seed0": { "mse": 0.0020382918883115053, "one_step_violation": 0.30322265625, "mean_violation": 0.9889501953125, "max_norm": 2.9793031215667725, "cloud_diameters": [ 8.197779655456543, 4.651096219276951e-07, 4.111343514523469e-07 ] }, "confirmed": true } }, "bench_report": { "baseline_sweep": [ { "cfg": { "lr": 0.001, "epochs": 12, "trap_lambda": 0.0 }, "per_seed": [ 0.02792397141456604, 0.041552960872650146, 0.03038809262216091, 0.03133251145482063, 0.05597708746790886, 0.022259902209043503, 0.07735025882720947, 0.01893964223563671 ], "mean": 0.038215553387999535, "std": 0.019635395061371623 }, { "cfg": { "lr": 0.003, "epochs": 12, "trap_lambda": 0.0 }, "per_seed": [ 0.009584075771272182, 0.013722526840865612, 0.008015970699489117, 0.011999157257378101, 0.01405302993953228, 0.00778926583006978, 0.013168145902454853, 0.007831467315554619 ], "mean": 0.010770454944577068, "std": 0.0027585833372694973 }, { "cfg": { "lr": 0.01, "epochs": 12, "trap_lambda": 0.0 }, "per_seed": [ 0.0024895062670111656, 0.0038835888262838125, 0.003609295468777418, 0.003999150358140469, 0.0032447325065732002, 0.0038587753660976887, 0.008323567919433117, 0.0034713787026703358 ], "mean": 0.004109999426873401, "std": 0.0017693440291274862 }, { "cfg": { "lr": 0.001, "epochs": 12, "trap_lambda": 0.3 }, "per_seed": [ 0.02792397141456604, 0.041552960872650146, 0.03038809262216091, 0.03133251145482063, 0.05597708746790886, 0.022259902209043503, 0.07735025882720947, 0.01893964223563671 ], "mean": 0.038215553387999535, "std": 0.019635395061371623 }, { "cfg": { "lr": 0.003, "epochs": 12, "trap_lambda": 0.3 }, "per_seed": [ 0.009584075771272182, 0.013722526840865612, 0.008015970699489117, 0.011999157257378101, 0.01405302993953228, 0.00778926583006978, 0.013168145902454853, 0.007831467315554619 ], "mean": 0.010770454944577068, "std": 0.0027585833372694973 }, { "cfg": { "lr": 0.01, "epochs": 12, "trap_lambda": 0.3 }, "per_seed": [ 0.0024895062670111656, 0.0038835888262838125, 0.003609295468777418, 0.003999150358140469, 0.0032447325065732002, 0.0038587753660976887, 0.008323567919433117, 0.0034713787026703358 ], "mean": 0.004109999426873401, "std": 0.0017693440291274862 }, { "cfg": { "lr": 0.001, "epochs": 12, "trap_lambda": 1.0 }, "per_seed": [ 0.02792397141456604, 0.041552960872650146, 0.03038809262216091, 0.03133251145482063, 0.05597708746790886, 0.022259902209043503, 0.07735025882720947, 0.01893964223563671 ], "mean": 0.038215553387999535, "std": 0.019635395061371623 }, { "cfg": { "lr": 0.003, "epochs": 12, "trap_lambda": 1.0 }, "per_seed": [ 0.009584075771272182, 0.013722526840865612, 0.008015970699489117, 0.011999157257378101, 0.01405302993953228, 0.00778926583006978, 0.013168145902454853, 0.007831467315554619 ], "mean": 0.010770454944577068, "std": 0.0027585833372694973 }, { "cfg": { "lr": 0.01, "epochs": 12, "trap_lambda": 1.0 }, "per_seed": [ 0.0024895062670111656, 0.0038835888262838125, 0.003609295468777418, 0.003999150358140469, 0.0032447325065732002, 0.0038587753660976887, 0.008323567919433117, 0.0034713787026703358 ], "mean": 0.004109999426873401, "std": 0.0017693440291274862 } ], "idea_sweep": [ { "cfg": { "lr": 0.001, "epochs": 12, "trap_lambda": 0.0 }, "per_seed": [ 0.02792397141456604, 0.041552960872650146, 0.03038809262216091, 0.03133251145482063, 0.05597708746790886, 0.022259902209043503, 0.07735025882720947, 0.01893964223563671 ], "mean": 0.038215553387999535, "std": 0.019635395061371623 }, { "cfg": { "lr": 0.003, "epochs": 12, "trap_lambda": 0.0 }, "per_seed": [ 0.009584075771272182, 0.013722526840865612, 0.008015970699489117, 0.011999157257378101, 0.01405302993953228, 0.00778926583006978, 0.013168145902454853, 0.007831467315554619 ], "mean": 0.010770454944577068, "std": 0.0027585833372694973 }, { "cfg": { "lr": 0.01, "epochs": 12, "trap_lambda": 0.0 }, "per_seed": [ 0.0024895062670111656, 0.0038835888262838125, 0.003609295468777418, 0.003999150358140469, 0.0032447325065732002, 0.0038587753660976887, 0.008323567919433117, 0.0034713787026703358 ], "mean": 0.004109999426873401, "std": 0.0017693440291274862 }, { "cfg": { "lr": 0.001, "epochs": 12, "trap_lambda": 0.3 }, "per_seed": [ 0.029911765828728676, 0.043103478848934174, 0.029594196006655693, 0.03148665651679039, 0.056626711040735245, 0.023263107985258102, 0.07945620268583298, 0.019382713362574577 ], "mean": 0.03910310403443873, "std": 0.020086107344229626 }, { "cfg": { "lr": 0.003, "epochs": 12, "trap_lambda": 0.3 }, "per_seed": [ 0.00927602220326662, 0.01410755980759859, 0.008517258800566196, 0.011783787980675697, 0.014786372892558575, 0.0075119370594620705, 0.012531742453575134, 0.008021011017262936 ], "mean": 0.010816961526870728, "std": 0.002850141431686086 }, { "cfg": { "lr": 0.01, "epochs": 12, "trap_lambda": 0.3 }, "per_seed": [ 0.0020382918883115053, 0.002796109300106764, 0.00355388387106359, 0.0032515127677470446, 0.003949518781155348, 0.003303474746644497, 0.0037955872248858213, 0.004127324093133211 ], "mean": 0.0033519628341309726, "std": 0.0006806487150040169 }, { "cfg": { "lr": 0.001, "epochs": 12, "trap_lambda": 1.0 }, "per_seed": [ 0.02979419380426407, 0.04312347620725632, 0.02988792397081852, 0.0319940410554409, 0.05702750012278557, 0.02334035187959671, 0.07955670356750488, 0.019355962052941322 ], "mean": 0.039260019082576036, "std": 0.020122028461056908 }, { "cfg": { "lr": 0.003, "epochs": 12, "trap_lambda": 1.0 }, "per_seed": [ 0.009333031252026558, 0.014507703483104706, 0.00847854744642973, 0.012289805337786674, 0.015088483691215515, 0.007256411015987396, 0.012649702839553356, 0.008141638711094856 ], "mean": 0.010968165472149849, "std": 0.0030406270010214335 }, { "cfg": { "lr": 0.01, "epochs": 12, "trap_lambda": 1.0 }, "per_seed": [ 0.0021302360109984875, 0.003130449214950204, 0.003100306261330843, 0.0041888016276061535, 0.004465983249247074, 0.0038384413346648216, 0.0035596308298408985, 0.004667002707719803 ], "mean": 0.0036351064045447856, "std": 0.0008378638946201928 } ], "paired_delta": -0.0007580365927424282, "comparison_metric": "test_mse" } }