decodeshare / artifacts /patch_back /results /Qwen /Qwen2.5-7B-Instruct /layer4 /_summary /summary.csv
| kind,file,task,eval_mode,layer,seed,hf_id,hf_split,candidate_labels,Qs_shape,patch_desc,donor_source,donor_tasks,donor_pick,n_donor_bank,scan_effective,scan_skipped,base_acc_scan,ablt_acc_scan,flips_scan,anti_flips_scan,both_correct_scan,both_wrong_scan,patched_primary_method,patched_primary_rescued_pct,patched_primary_mean_dmargin,diff_time_shuffled_minus_patched_primary_rescued_pct,diff_patched_primary_minus_shared_randvec_rescued_pct,patched_0_rescued,patched_0_n,patched_0_rescued_pct,patched_0_mean_dmargin,patched_01_rescued,patched_01_n,patched_01_rescued_pct,patched_01_mean_dmargin,patched_full_rescued,patched_full_n,patched_full_rescued_pct,patched_full_mean_dmargin,patched_self_rescued,patched_self_n,patched_self_rescued_pct,patched_self_mean_dmargin,patched_transfer_rescued,patched_transfer_n,patched_transfer_rescued_pct,patched_transfer_mean_dmargin,control_time_shuffled_rescued,control_time_shuffled_n,control_time_shuffled_rescued_pct,control_time_shuffled_mean_dmargin,control_shared_mismatch_rescued,control_shared_mismatch_n,control_shared_mismatch_rescued_pct,control_shared_mismatch_mean_dmargin,control_shared_perm_rescued,control_shared_perm_n,control_shared_perm_rescued_pct,control_shared_perm_mean_dmargin,control_shared_signflip_rescued,control_shared_signflip_n,control_shared_signflip_rescued_pct,control_shared_signflip_mean_dmargin,control_shared_randvec_rescued,control_shared_randvec_n,control_shared_randvec_rescued_pct,control_shared_randvec_mean_dmargin,control_rand_subspace_rescued,control_rand_subspace_n,control_rand_subspace_rescued_pct,control_rand_subspace_mean_dmargin,control_patch_nonshared_rescued,control_patch_nonshared_n,control_patch_nonshared_rescued_pct,control_patch_nonshared_mean_dmargin | |
| flipset,aqua_alpha_sweep_seed123.json,aqua,,4,123,,,ABCDE,3584x164,steps=0,,,,,254,0,0.41338582677165353,0.2874015748031496,58,26,47,123,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, | |
| flipset,aqua_alpha_sweep_seed456.json,aqua,,4,456,,,ABCDE,3584x135,steps=0,,,,,254,0,0.4015748031496063,0.3110236220472441,63,40,39,112,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, | |
| flipset,aqua_transfer_cross_mc_baselinecorrect_seed123.json,aqua,,4,123,,,ABCDE,3584x164,steps=0,cross_task_eval,"commonsenseqa,openbookqa",random,256,254,0,0.41338582677165353,0.2874015748031496,58,26,47,123,patched_self,100.0,3.813763380050659,,,,,,,,,,,,,,,58,58,100.0,3.813763380050659,56,58,96.55172413793103,3.8267593383789062,,,,,,,,,,,,,,,,,,,,,,,,,,,, | |
| flipset,aqua_transfer_same_task_seed123.json,aqua,,4,123,,,ABCDE,3584x164,steps=0,same_task_eval,aqua,random,254,254,0,0.41338582677165353,0.2874015748031496,58,26,47,123,patched_self,100.0,3.813763380050659,,,,,,,,,,,,,,,58,58,100.0,3.813763380050659,56,58,96.55172413793103,3.811663866043091,,,,,,,,,,,,,,,,,,,,,,,,,,,, | |
| openanswer,gsm8k_genmath.json,gsm8k,gen_math,4,123,gsm8k/main,,,3584x164,"steps=0,1,2,3",,,,,256,0,0.03515625,0.03125,9,8,0,239,patched_self,22.22222222222222,,-11.11111111111111,0.0,,,,,,,,,,,,,2,9,22.22222222222222,,,,,,1,9,11.11111111111111,,,,,,,,,,,,,,2,9,22.22222222222222,,3,9,33.333333333333336,,4,9,44.44444444444444, | |
| openanswer,gsm8k_pairlogprob.json,gsm8k,pair_logprob,4,123,gsm8k/main,,,3584x164,"steps=0,1,2,3",,,,,256,0,0.80078125,0.50390625,83,7,122,44,patched_self,98.4375,16.13555908203125,-4.6875,85.9375,,,,,,,,,,,,,63,64,98.4375,16.13555908203125,,,,,60,64,93.75,15.168949127197266,,,,,,,,,,,,,8,64,12.5,0.7623266577720642,9,64,14.0625,1.2693865299224854,3,64,4.6875,-0.04090585559606552 | |
| openanswer,humaneval_gencode_compile.json,humaneval,gen_code_compile,4,123,openai_humaneval,test,,3584x164,"steps=0,1,2,3",,,,,0,164,,,0,0,0,0,patched_self,16.666666666666668,,0.0,-16.666666666666668,,,,,,,,,,,,,3,18,16.666666666666668,,,,,,3,18,16.666666666666668,,,,,,,,,,,,,,6,18,33.333333333333336,,3,18,16.666666666666668,,0,18,0.0, | |
| openanswer,humaneval_pairlogprob.json,humaneval,pair_logprob,4,123,openai_humaneval,test,,3584x164,"steps=0,1,2,3",,,,,164,0,0.6829268292682927,0.7073170731707317,3,7,109,45,patched_self,66.66666666666667,0.30727076530456543,0.0,0.0,,,,,,,,,,,,,2,3,66.66666666666667,0.30727076530456543,,,,,2,3,66.66666666666667,1.0313528776168823,,,,,,,,,,,,,2,3,66.66666666666667,0.7268994450569153,1,3,33.333333333333336,0.9633371829986572,0,3,0.0,-0.3868543207645416 | |
| subspace_mc,aqua.json,aqua,,4,123,,,ABCDE,3584x164,,,,,,254,0,0.41338582677165353,0.2874015748031496,58,26,47,123,patched_0,100.0,3.8137631734897353,-1.7241379310344769,84.48275862068965,58,58,100.0,3.8137631734897353,58,58,100.0,3.8137631734897353,58,58,100.0,3.8137631734897353,,,,,,,,,57,58,98.27586206896552,3.8100255553064675,,,,,,,,,,,,,9,58,15.517241379310345,-0.20225618214442811,14,58,24.137931034482758,-0.17025229643131123,0,58,0.0,-8.221330313846983e-09 | |
| subspace_mc,arc_challenge.json,arc_challenge,,4,123,,,ABCD,3584x164,,,,,,255,1,0.9058823529411765,0.7294117647058823,50,5,181,19,patched_0,100.0,7.293835225105286,0.0,58.0,50,50,100.0,7.293835225105286,50,50,100.0,7.293835225105286,50,50,100.0,7.293835225105286,,,,,,,,,50,50,100.0,7.3109342944622036,,,,,,,,,,,,,21,50,42.0,0.7695868587493897,19,50,38.0,0.6493690681457519,0,50,0.0,1.4638900756835937e-06 | |
| subspace_mc,commonsenseqa.json,commonsenseqa,,4,123,,,ABCDE,3584x164,,,,,,256,0,0.8671875,0.8046875,27,11,195,23,patched_0,100.0,5.433277322186364,0.0,48.148148148148145,27,27,100.0,5.433277322186364,27,27,100.0,5.433277322186364,27,27,100.0,5.433277322186364,,,,,,,,,27,27,100.0,5.423130929470062,,,,,,,,,,,,,14,27,51.851851851851855,1.2342691244902435,14,27,51.851851851851855,1.0576389542332403,0,27,0.0,4.273873788339121e-06 | |
| subspace_mc,logiqa.json,logiqa,,4,123,,,ABCD,3584x164,,,,,,256,0,0.47265625,0.390625,53,32,68,103,patched_0,100.0,6.139819563559766,-1.8867924528301927,66.0377358490566,53,53,100.0,6.139819563559766,53,53,100.0,6.139819563559766,53,53,100.0,6.139819563559766,,,,,,,,,52,53,98.11320754716981,6.150581973903584,,,,,,,,,,,,,18,53,33.9622641509434,0.8587988592543692,13,53,24.528301886792452,0.6070889594419947,0,53,0.0,1.1066220841317807e-06 | |
| subspace_mc,openbookqa.json,openbookqa,,4,123,,,ABCD,3584x164,,,,,,256,0,0.859375,0.6640625,55,5,165,31,patched_0,100.0,5.872730739550157,0.0,72.72727272727272,55,55,100.0,5.872730739550157,55,55,100.0,5.872730739550157,55,55,100.0,5.872730739550157,,,,,,,,,55,55,100.0,5.857445214011452,,,,,,,,,,,,,15,55,27.272727272727273,0.3391037485816262,20,55,36.36363636363637,0.5655256639827382,0,55,0.0,-4.2481855912642044e-07 | |
| subspace_mc,piqa.json,piqa,,4,123,,,AB,3584x164,,,,,,256,0,0.87109375,0.7421875,49,16,174,17,patched_0,100.0,6.0123149594482115,0.0,59.183673469387756,49,49,100.0,6.0123149594482115,49,49,100.0,6.0123149594482115,49,49,100.0,6.0123149594482115,,,,,,,,,49,49,100.0,6.0006689234655735,,,,,,,,,,,,,20,49,40.816326530612244,0.8917713432895894,15,49,30.612244897959183,0.3249082735606602,0,49,0.0,-5.303596963687819e-07 | |
| subspace_mc,qasc.json,qasc,,4,123,,,ABCDEFGH,3584x164,,,,,,256,0,0.80859375,0.6953125,33,4,174,45,patched_0,100.0,6.235052289384784,0.0,72.72727272727272,33,33,100.0,6.235052289384784,33,33,100.0,6.235052289384784,33,33,100.0,6.235052289384784,,,,,,,,,33,33,100.0,6.2372765432704576,,,,,,,,,,,,,9,33,27.272727272727273,-0.0902654691175981,12,33,36.36363636363637,0.4817595626368667,0,33,0.0,2.102418379350142e-06 | |