{ "bench_version": 1, "track": "sequence", "model": "transformer_tiny", "metric_direction": "lower is better", "n_seeds": 8, "baseline": { "best_cfg": { "lr": 0.001 }, "sweep": [ { "cfg": { "lr": 0.001 }, "mean": 0.620273545384407 }, { "cfg": { "lr": 0.002 }, "mean": 0.6989391446113586 }, { "cfg": { "lr": 0.003 }, "mean": 0.8049232810735703 } ], "full": { "mean": 0.6231015399098396, "std": 0.06897174016713392, "per_seed": [ 0.7222549319267273, 0.5303708910942078, 0.5396761298179626, 0.6887922286987305, 0.640042245388031, 0.6537070274353027, 0.6618669033050537, 0.5481019616127014 ], "n": 8 } }, "idea": { "cfg": { "q": 0.5 }, "per_seed": [ 0.7222549319267273, 0.5303708910942078, 0.5396761298179626, 0.6887922286987305, 0.640042245388031, 0.6537070274353027, 0.6618669033050537, 0.5481019616127014 ], "mean": 0.6231015399098396 }, "comparison": { "delta_mean": 0.0, "idea_wins": 0, "n_pairs": 8, "per_seed_diffs": [ 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0 ], "p_value": 1.0, "mde": 0.0, "mde_rel_pct": 0.0, "verdict": "no measurable effect", "system_worked": false }, "mathcheck": { "delta": 0.25, "capacity": 0.3475898813907971, "safe_capacity_eps_.1": 0.3128308932517174, "binary_stream_rate": 1.0, "rate_satisfies_bound": false, "transition_probability_definition": "P(flip)=q" }, "idea_sweep": [ { "cfg": { "q": 0.25 }, "per_seed": [ 0.712475597858429, 0.5296003818511963, 0.5759447813034058, 0.6928728818893433, 0.6395291686058044, 0.6558457016944885, 0.6588586568832397, 0.5424662828445435 ], "mean": 0.6259491816163063 }, { "cfg": { "q": 0.4 }, "per_seed": [ 0.7143200635910034, 0.5526124238967896, 0.5422033667564392, 0.6958776712417603, 0.625187873840332, 0.6721104383468628, 0.6582097411155701, 0.538433849811554 ], "mean": 0.6248694285750389 }, { "cfg": { "q": 0.5 }, "per_seed": [ 0.7222549319267273, 0.5303708910942078, 0.5396761298179626, 0.6887922286987305, 0.640042245388031, 0.6537070274353027, 0.6618669033050537, 0.5481019616127014 ], "mean": 0.6231015399098396 } ], "protocol": { "epochs": 8, "batch": 128, "train_test_n": 400, "corruption": "25% random insertions in an 8-bit redundant stream" }, "mechanism_signature": { "predicted_markov_q": 0.5, "observed_markov_transition_rate": 0.5178571428571428, "predicted_iid_q": 0.5, "observed_iid_transition_rate": 0.5178571428571428, "trained_model_sensitivity_mean": { "baseline": 0.024137018248438835, "idea": 0.023845407529734075 }, "confirmed": true, "per_seed": [ { "seed": 0, "baseline_stream_sensitivity": 0.017377333715558052, "idea_stream_sensitivity": 0.01767236925661564 }, { "seed": 1, "baseline_stream_sensitivity": 0.026213817298412323, "idea_stream_sensitivity": 0.02488153986632824 }, { "seed": 2, "baseline_stream_sensitivity": 0.023712292313575745, "idea_stream_sensitivity": 0.02447468228638172 }, { "seed": 3, "baseline_stream_sensitivity": 0.009144331328570843, "idea_stream_sensitivity": 0.0082431985065341 }, { "seed": 4, "baseline_stream_sensitivity": 0.01021712925285101, "idea_stream_sensitivity": 0.010267045348882675 }, { "seed": 5, "baseline_stream_sensitivity": 0.06268110126256943, "idea_stream_sensitivity": 0.06135376915335655 }, { "seed": 6, "baseline_stream_sensitivity": 0.01560310460627079, "idea_stream_sensitivity": 0.017230955883860588 }, { "seed": 7, "baseline_stream_sensitivity": 0.028147036209702492, "idea_stream_sensitivity": 0.026639699935913086 } ] } }