{ "bench_version": 1, "track": "dynamics", "model": "rnn_small", "metric_direction": "lower is better", "n_seeds": 8, "baseline": { "best_cfg": { "lr": 0.006, "wd": 0.0 }, "sweep": [ { "cfg": { "lr": 0.001, "wd": 0.0 }, "mean": 0.41272661462426186 }, { "cfg": { "lr": 0.003, "wd": 0.0 }, "mean": 0.0579891842789948 }, { "cfg": { "lr": 0.006, "wd": 0.0 }, "mean": 0.050475220661610365 } ], "full": { "mean": 0.050475220661610365, "std": 0.015083096048661116, "per_seed": [ 0.04993843659758568, 0.04235526919364929, 0.0690467357635498, 0.07208486646413803, 0.05335729196667671, 0.022722482681274414, 0.055551957339048386, 0.0387447252869606 ], "n": 8 } }, "idea": { "mean": 0.04891866398975253, "std": 0.00862940297862839, "per_seed": [ 0.048651475459337234, 0.029897399246692657, 0.0598931610584259, 0.04604996740818024, 0.05041059851646423, 0.05862913280725479, 0.05099020525813103, 0.046827372163534164 ], "n": 8 }, "comparison": { "delta_mean": -0.0015565566718578339, "idea_wins": 6, "n_pairs": 8, "per_seed_diffs": [ -0.0012869611382484436, -0.012457869946956635, -0.009153574705123901, -0.026034899055957794, -0.0029466934502124786, 0.03590665012598038, -0.004561752080917358, 0.008082646876573563 ], "p_value": 0.7986, "mde": 0.015101068783051909, "mde_rel_pct": 29.917786559647947, "verdict": "no significant win", "system_worked": false }, "mechanism_signature": { "one_step_latent_mse_mean": 0.03250809867555896, "decoded_rollout_mse_h6": 0.305510014295578, "support_change_mean": 3.440310562533947e-05, "active_fraction_tau_025": 0.6012109518051147, "confirmed": false }, "idea_sweep": [ { "cfg": { "lr": 0.006, "lam": 0.0005, "eta": 0.002 }, "result": { "mean": 0.04891866398975253, "std": 0.00862940297862839, "per_seed": [ 0.048651475459337234, 0.029897399246692657, 0.0598931610584259, 0.04604996740818024, 0.05041059851646423, 0.05862913280725479, 0.05099020525813103, 0.046827372163534164 ], "n": 8 } }, { "cfg": { "lr": 0.001, "lam": 0.0005, "eta": 0.002 }, "result": { "mean": 0.40538693591952324, "std": 0.06670983818512866, "per_seed": [ 0.42158782482147217, 0.29316720366477966, 0.4900130331516266, 0.44477176666259766, 0.43406909704208374, 0.47257643938064575, 0.3660927712917328, 0.32081735134124756 ], "n": 8 } }, { "cfg": { "lr": 0.006, "lam": 0.0005, "eta": 0.002 }, "result": { "mean": 0.04891866398975253, "std": 0.00862940297862839, "per_seed": [ 0.048651475459337234, 0.029897399246692657, 0.0598931610584259, 0.04604996740818024, 0.05041059851646423, 0.05862913280725479, 0.05099020525813103, 0.046827372163534164 ], "n": 8 } } ], "runtime_sec": 34.89922094345093 }