artifacts/patch_back/results/tiiuae__falcon-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,4544x36,steps=0,,,,,254,0,0.25196850393700787,0.24803149606299213,3,2,61,188,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,
flipset,aqua_alpha_sweep_seed456.json,aqua,,4,456,,,ABCDE,4544x46,steps=0,,,,,254,0,0.25196850393700787,0.24803149606299213,3,2,61,188,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,
flipset,aqua_transfer_cross_mc_baselinecorrect_seed123.json,aqua,,4,123,,,ABCDE,4544x36,steps=0,cross_task_eval,"commonsenseqa,openbookqa",random,114,254,0,0.25196850393700787,0.24803149606299213,3,2,61,188,patched_self,100.0,1.0660523176193237,,,,,,,,,,,,,,,3,3,100.0,1.0660523176193237,2,3,66.66666666666667,1.0517040491104126,,,,,,,,,,,,,,,,,,,,,,,,,,,,
flipset,aqua_transfer_same_task_seed123.json,aqua,,4,123,,,ABCDE,4544x36,steps=0,same_task_eval,aqua,random,254,254,0,0.25196850393700787,0.24803149606299213,3,2,61,188,patched_self,100.0,1.0660523176193237,,,,,,,,,,,,,,,3,3,100.0,1.0660523176193237,3,3,100.0,1.0685762166976929,,,,,,,,,,,,,,,,,,,,,,,,,,,,
openanswer,gsm8k_genmath.json,gsm8k,gen_math,4,123,gsm8k/main,,,4544x36,"steps=0,1,2,3",,,,,256,0,0.03515625,0.02734375,9,7,0,240,patched_self,55.55555555555556,,-11.111111111111114,33.333333333333336,,,,,,,,,,,,,5,9,55.55555555555556,,,,,,4,9,44.44444444444444,,,,,,,,,,,,,,2,9,22.22222222222222,,0,9,0.0,,0,9,0.0,
openanswer,gsm8k_pairlogprob.json,gsm8k,pair_logprob,4,123,gsm8k/main,,,4544x36,"steps=0,1,2,3",,,,,256,0,0.63671875,0.5390625,42,17,121,76,patched_self,100.0,2.783522367477417,-2.3809523809523796,90.47619047619048,,,,,,,,,,,,,42,42,100.0,2.783522367477417,,,,,41,42,97.61904761904762,2.7538859844207764,,,,,,,,,,,,,4,42,9.523809523809524,0.02977374568581581,1,42,2.380952380952381,-0.07559765875339508,0,42,0.0,-0.00029452925082296133
openanswer,humaneval_gencode_compile.json,humaneval,gen_code_compile,4,123,openai_humaneval,test,,4544x36,"steps=0,1,2,3",,,,,0,164,,,0,0,0,0,patched_self,8.333333333333334,,-4.166666666666667,2.083333333333334,,,,,,,,,,,,,4,48,8.333333333333334,,,,,,2,48,4.166666666666667,,,,,,,,,,,,,,3,48,6.25,,1,48,2.0833333333333335,,1,48,2.0833333333333335,
openanswer,humaneval_pairlogprob.json,humaneval,pair_logprob,4,123,openai_humaneval,test,,4544x36,"steps=0,1,2,3",,,,,164,0,0.6402439024390244,0.6097560975609756,8,3,97,56,patched_self,100.0,2.2496209144592285,-37.5,100.0,,,,,,,,,,,,,8,8,100.0,2.2496209144592285,,,,,5,8,62.5,0.747115433216095,,,,,,,,,,,,,0,8,0.0,0.10467139631509781,0,8,0.0,0.013523062691092491,0,8,0.0,-0.02801455184817314
subspace_mc,aqua.json,aqua,,4,123,,,ABCDE,4544x36,,,,,,254,0,0.25196850393700787,0.24803149606299213,3,2,61,188,patched_0,100.0,1.066052238146464,0.0,100.0,3,3,100.0,1.066052238146464,3,3,100.0,1.066052238146464,3,3,100.0,1.066052238146464,,,,,,,,,3,3,100.0,1.0629873275756836,,,,,,,,,,,,,0,3,0.0,0.02270086606343587,0,3,0.0,-0.08046384652455647,0,3,0.0,1.1920928955078125e-07
subspace_mc,arc_challenge.json,arc_challenge,,4,123,,,ABCD,4544x36,,,,,,256,0,0.26171875,0.25390625,20,18,47,171,patched_0,100.0,0.9604960411787034,0.0,85.0,20,20,100.0,0.9604960411787034,20,20,100.0,0.9604960411787034,20,20,100.0,0.9604960411787034,,,,,,,,,20,20,100.0,0.958578696846962,,,,,,,,,,,,,3,20,15.0,0.0675793707370758,0,20,0.0,-0.4309776782989502,0,20,0.0,-4.82797622680664e-07
subspace_mc,commonsenseqa.json,commonsenseqa,,4,123,,,ABCDE,4544x36,,,,,,256,0,0.1875,0.19921875,8,11,40,197,patched_0,100.0,0.9405499696731567,0.0,100.0,8,8,100.0,0.9405499696731567,8,8,100.0,0.9405499696731567,8,8,100.0,0.9405499696731567,,,,,,,,,8,8,100.0,0.9393054693937302,,,,,,,,,,,,,0,8,0.0,-0.27225162088871,0,8,0.0,-0.19955047965049744,0,8,0.0,5.21540641784668e-07
subspace_mc,logiqa.json,logiqa,,4,123,,,ABCD,4544x36,,,,,,256,0,0.16796875,0.1640625,3,2,40,211,patched_0,100.0,0.8313813010851542,0.0,100.0,3,3,100.0,0.8313813010851542,3,3,100.0,0.8313813010851542,3,3,100.0,0.8313813010851542,,,,,,,,,3,3,100.0,0.8332018454869589,,,,,,,,,,,,,0,3,0.0,-0.14246646563212076,0,3,0.0,-0.11746446291605632,0,3,0.0,2.7815500895182293e-07
subspace_mc,openbookqa.json,openbookqa,,4,123,,,ABCD,4544x36,,,,,,256,0,0.24609375,0.28125,13,22,50,171,patched_0,100.0,0.655216239965879,0.0,84.61538461538461,13,13,100.0,0.655216239965879,13,13,100.0,0.655216239965879,13,13,100.0,0.655216239965879,,,,,,,,,13,13,100.0,0.6563335840518658,,,,,,,,,,,,,2,13,15.384615384615385,-0.1085894749714778,0,13,0.0,-0.4173735838669997,0,13,0.0,7.335956280048077e-08
subspace_mc,piqa.json,piqa,,4,123,,,AB,4544x36,,,,,,256,0,0.5390625,0.5703125,41,49,97,69,patched_0,100.0,0.6776537052014979,0.0,90.2439024390244,41,41,100.0,0.6776537052014979,41,41,100.0,0.6776537052014979,41,41,100.0,0.6776537052014979,,,,,,,,,41,41,100.0,0.6779073912922929,,,,,,,,,,,,,4,41,9.75609756097561,-0.2560210576871546,0,41,0.0,-0.3054549243392014,0,41,0.0,-4.3613154713700456e-08
subspace_mc,qasc.json,qasc,,4,123,,,ABCDEFGH,4544x36,,,,,,256,0,0.12109375,0.125,3,4,28,221,patched_0,100.0,0.7035077810287476,0.0,100.0,3,3,100.0,0.7035077810287476,3,3,100.0,0.7035077810287476,3,3,100.0,0.7035077810287476,,,,,,,,,3,3,100.0,0.70074991385142,,,,,,,,,,,,,0,3,0.0,-0.26586441198984784,0,3,0.0,-0.26830148696899414,0,3,0.0,7.152557373046875e-07