{ "pooled": { "A-s0": { "chunk_mae": 7.7966, "first_mae": 3.9422, "pred_key": "pred:bijou@40000" }, "A-s1": { "chunk_mae": 7.8052, "first_mae": 4.1118, "pred_key": "pred:bijou@40000" }, "A-s2": { "chunk_mae": 7.7355, "first_mae": 3.9377, "pred_key": "pred:bijou@40000" }, "B": { "chunk_mae": 8.2989, "first_mae": 3.5009, "pred_key": "pred:bijou@40000" } }, "primary_B_minus_As0": { "mean": 0.46215, "median": 0.0, "ci95": [ 0.38683, 0.53744 ], "b_win_rate": 0.4945, "n_frames": 17204 }, "pairwise_replicates": { "A-s0~A-s1": { "mean": 0.00707, "abs_mean": 0.00707, "pooled_abs_dchunk": 0.0086 }, "A-s0~A-s2": { "mean": -0.05434, "abs_mean": 0.05434, "pooled_abs_dchunk": 0.0611 }, "A-s1~A-s2": { "mean": -0.06141, "abs_mean": 0.06141, "pooled_abs_dchunk": 0.0697 } }, "E5_noise_floor": { "max_pooled_pairwise_abs_dchunk": 0.0697, "soft_expectation_le": 0.2, "headline_if_gt": 0.3, "verdict": "within soft expectation" }, "finalization_amendments": { "replicate_chunk_maes": [ 7.7966, 7.8052, 7.7355 ], "replicate_first_maes": [ 3.9422, 4.1118, 3.9377 ], "sigma_seed_chunk": 0.038, "sigma_seed_first": 0.09924, "e4b_adopt_band": 0.15, "e4b_adopt_band_rule": "max(3*sigma_seed, 0.15) below AR anchor 5.8026" }, "decision": { "effect_abs": 0.46215, "threshold_max_pairwise_replicate": 0.06141, "exceeds_threshold": true, "repo_coherent": true, "most_influential_repos": [ { "repo": "dopaul/1500_chess_moves", "n_excl": 600, "mean_without": 0.43481 }, { "repo": "akira-sasaki/akabeko_instruct_v2", "n_excl": 136, "mean_without": 0.4429 }, { "repo": "charleyong/so100_tic_tac_toe_autoend", "n_excl": 72, "mean_without": 0.47939 }, { "repo": "akira-sasaki/akabeko_instruct_v1", "n_excl": 120, "mean_without": 0.44738 }, { "repo": "dopaul/game_v7", "n_excl": 68, "mean_without": 0.44757 } ], "verdict": "REAL: aux-off WORSE \u2014 aux supervision helps actions (representation shaping)" } }