run_id_prefix: full_arc_challenge_w2s_rationales n_rows: 65 loss_curves: not recorded parse_fail_rate.labels_only: 0.000455 raw_single_letter_rate.labels_only: 1.000000 pred_equals_raw_rate.labels_only: 0.999545 parse_fail_rate.labels_rationales: 0.000000 raw_single_letter_rate.labels_rationales: 1.000000 pred_equals_raw_rate.labels_rationales: 1.000000 parse_fail_rate.clean_ceiling: 0.000000 raw_single_letter_rate.clean_ceiling: 0.999659 pred_equals_raw_rate.clean_ceiling: 0.999659 parse_fail_rate.zeroshot_floor: 0.000000 raw_single_letter_rate.zeroshot_floor: 0.998294 pred_equals_raw_rate.zeroshot_floor: 0.998294 parse_fail_rate.rationale_only: 0.001991 raw_single_letter_rate.rationale_only: 0.261320 pred_equals_raw_rate.rationale_only: 0.261320 parse_fail_rate.shuffled_rationale: 0.000000 raw_single_letter_rate.shuffled_rationale: 1.000000 pred_equals_raw_rate.shuffled_rationale: 1.000000 clean_acc_mean.labels_only.eps0.15: 0.713993 clean_acc_std.labels_only.eps0.15: 0.091749 n_seeds.labels_only.eps0.15: 5 clean_acc_mean.labels_only.eps0.30: 0.737713 clean_acc_std.labels_only.eps0.30: 0.047128 n_seeds.labels_only.eps0.30: 5 clean_acc_mean.labels_only.eps0.45: 0.480375 clean_acc_std.labels_only.eps0.45: 0.245788 n_seeds.labels_only.eps0.45: 5 clean_acc_mean.labels_rationales.eps0.15: 0.684130 clean_acc_std.labels_rationales.eps0.15: 0.048434 n_seeds.labels_rationales.eps0.15: 5 clean_acc_mean.labels_rationales.eps0.30: 0.614505 clean_acc_std.labels_rationales.eps0.30: 0.056059 n_seeds.labels_rationales.eps0.30: 5 clean_acc_mean.labels_rationales.eps0.45: 0.611945 clean_acc_std.labels_rationales.eps0.45: 0.038308 n_seeds.labels_rationales.eps0.45: 5 clean_acc_mean.clean_ceiling.eps0.30: 0.737457 clean_acc_std.clean_ceiling.eps0.30: 0.181847 n_seeds.clean_ceiling.eps0.30: 10 clean_acc_mean.zeroshot_floor.eps0.30: 0.801195 clean_acc_std.zeroshot_floor.eps0.30: 0.000000 n_seeds.zeroshot_floor.eps0.30: 5 clean_acc_mean.rationale_only.eps0.15: 0.490102 clean_acc_std.rationale_only.eps0.15: 0.112255 n_seeds.rationale_only.eps0.15: 5 clean_acc_mean.rationale_only.eps0.30: 0.449829 clean_acc_std.rationale_only.eps0.30: 0.100833 n_seeds.rationale_only.eps0.30: 5 clean_acc_mean.rationale_only.eps0.45: 0.438225 clean_acc_std.rationale_only.eps0.45: 0.089948 n_seeds.rationale_only.eps0.45: 5 clean_acc_mean.shuffled_rationale.eps0.30: 0.769113 clean_acc_std.shuffled_rationale.eps0.30: 0.031406 n_seeds.shuffled_rationale.eps0.30: 5 effect_B_minus_A.eps0.15: -0.029863 effect_B_minus_A.eps0.15.ci_lo: -0.042150 effect_B_minus_A.eps0.15.ci_hi: -0.017918 effect_B_minus_A.eps0.15.p: 0.000100 effect_B_minus_A.eps0.15.n_pairs: 5860 effect_B_minus_A.eps0.15.n_seeds: 5 effect_B_minus_A.eps0.30: -0.123208 effect_B_minus_A.eps0.30.ci_lo: -0.135324 effect_B_minus_A.eps0.30.ci_hi: -0.111092 effect_B_minus_A.eps0.30.p: 0.000100 effect_B_minus_A.eps0.30.n_pairs: 5860 effect_B_minus_A.eps0.30.n_seeds: 5 effect_B_minus_A.eps0.45: 0.131570 effect_B_minus_A.eps0.45.ci_lo: 0.115358 effect_B_minus_A.eps0.45.ci_hi: 0.147952 effect_B_minus_A.eps0.45.p: 0.000100 effect_B_minus_A.eps0.45.n_pairs: 5860 effect_B_minus_A.eps0.45.n_seeds: 5 primary_effect_eps0.30: -0.123208 primary_ci_excludes_zero: 0 primary_point_ge_min_effect: 0 primary_confirmed: 0 manip_rationale_only_minus_zeroshot.eps0.30: -0.351365 manip_rationale_only_minus_zeroshot.eps0.30.ci_lo: -0.365358 manip_rationale_only_minus_zeroshot.eps0.30.ci_hi: -0.337372 manip_rationale_only_minus_zeroshot.eps0.30.p: 0.000100 manip_check_passed: 0 degradation_gap.eps0.15: 0.023464 degradation_gap.eps0.30: -0.000256 degradation_gap.eps0.45: 0.257082 degradation_gap_paired.eps0.30: -0.062628 degradation_gap_paired.eps0.30.ci_lo: -0.075939 degradation_gap_paired.eps0.30.ci_hi: -0.049659 degradation_gap_paired.eps0.30.p: 0.000100 real_minus_shuffled_rationale.eps0.30: -0.154608 real_minus_shuffled_rationale.eps0.30.ci_lo: -0.166041 real_minus_shuffled_rationale.eps0.30.ci_hi: -0.143174 real_minus_shuffled_rationale.eps0.30.p: 0.000100 holm_family_non_primary_tests: effect_B_minus_A.eps0.15,effect_B_minus_A.eps0.45,manip_rationale_only_minus_zeroshot.eps0.30,degradation_gap_paired.eps0.30,real_minus_shuffled_rationale.eps0.30 effect_B_minus_A.eps0.15.p_holm: 0.000100 effect_B_minus_A.eps0.45.p_holm: 0.000100 manip_rationale_only_minus_zeroshot.eps0.30.p_holm: 0.000100 degradation_gap_paired.eps0.30.p_holm: 0.000100 real_minus_shuffled_rationale.eps0.30.p_holm: 0.000100