38 lines
5.4 KiB
CSV
38 lines
5.4 KiB
CSV
cohort,metric,recipient,base_median,base_p10,base_p90,finetuned_median,finetuned_p10,finetuned_p90,paired_delta_median
|
|
legacy,distance,3,8.739316033368453,4.950963035160957,10.863499972573624,8.731708727318438,4.9232497064004646,10.84212771798773,-0.02037733856762003
|
|
legacy,distance,7,9.4363090529414,6.25085771836616,11.556565692274562,9.423484801453295,6.21315757199869,11.527370171007142,-0.019801758117782065
|
|
legacy,distance,11,11.22490926849138,7.737020182887724,12.908045076028829,11.250679957928547,7.73525088572658,12.928017935515697,0.0038455741574283664
|
|
legacy,distance,15,9.759881279287757,7.162904991854923,11.716165708007967,9.800072330188218,7.1630134470288285,11.731124776731967,0.03740503224688574
|
|
legacy,distance,19,7.582995802914317,5.478640166334395,10.198341943349355,7.610650757626568,5.493471329919807,10.144988324005332,0.016808984407926886
|
|
legacy,distance,23,7.388739661710135,5.651966220638869,9.249179535567297,7.251844559425095,5.554179144408622,9.126738186063735,-0.09932936375917611
|
|
legacy,divergence,3,3.147912031314844,1.3810818357797214,5.537694943911305,3.1206879617749346,1.3758373946197866,5.50519735871187,-0.013426481470755869
|
|
legacy,divergence,7,4.713474898063865,1.9437311034396794,7.289381192575258,4.697509688023275,1.9391684741219346,7.240142719104977,-0.02240482362572893
|
|
legacy,divergence,11,5.747313395153364,2.349226790675758,8.479543942481612,5.7555205045992,2.342471242036245,8.45559258627536,0.008220859323202268
|
|
legacy,divergence,15,4.927015096951064,2.8084538003789024,7.330506308274049,4.927768269102536,2.833041243881992,7.336922408013947,0.021762632364951795
|
|
legacy,divergence,19,2.8494489425735994,1.7575837057447943,5.18654187756192,2.861665011032441,1.7442072067733334,5.143915153772987,0.006601984727423238
|
|
legacy,divergence,23,3.0214471514630055,2.047270924772257,4.518937219975146,2.9701865807044205,2.0010686132133917,4.411307997114287,-0.05707591033976844
|
|
legacy,QKV,3,8.777726631430976,7.189103615155911,10.713615974191201,8.76668332968768,7.190684779039996,10.719732488790717,-0.009309064163098846
|
|
legacy,QKV,7,9.844072680466791,8.0413059359728,11.619560783151188,9.82825724015493,8.014544082220969,11.581538677554722,-0.016941599110913863
|
|
legacy,QKV,11,12.089247576362546,10.201057391277503,13.571935182553943,12.09729379910231,10.196828217568173,13.57795553074303,0.01309251483925511
|
|
legacy,QKV,15,10.345856861414816,8.507270281674957,11.85158988753179,10.38900205657722,8.564274444281775,11.951509121264612,0.037888922465287855
|
|
legacy,QKV,19,12.044235479417448,10.732008460392013,13.1588691005697,12.061634540063746,10.722140503244573,13.161318190003938,-0.0025726092883973095
|
|
legacy,QKV,23,8.09156575360802,5.833100089115012,10.736031379205444,7.9963938578096405,5.72057650519899,10.668444002058392,-0.09506145974499303
|
|
gsm8k,distance,3,8.913673809873648,5.464073924885019,11.307334958094344,8.94684685981133,5.466046067182029,11.285099623450353,-0.0065539359628692395
|
|
gsm8k,distance,7,9.691950173931822,6.449358553913276,11.717921799729858,9.682670113610445,6.420969364897694,11.686361680373526,-0.020282664056864075
|
|
gsm8k,distance,11,11.13393122020824,6.4616586939758776,12.895309269291927,11.107427837202076,6.533455114023123,12.900256831750358,0.010065218794381714
|
|
gsm8k,distance,15,9.974087630581401,7.069686056657516,11.884414604809061,9.997745229352889,7.038241611739143,11.927975457275181,0.04648967580843344
|
|
gsm8k,distance,19,7.77346164330409,6.131638555699892,9.455856359148038,7.789480896051456,6.136266410549428,9.498390268941135,0.020110690923323293
|
|
gsm8k,distance,23,7.442742494743904,5.505415997443188,9.042021588456507,7.346119301169663,5.469589684395749,8.936377940244807,-0.10260948221392674
|
|
gsm8k,divergence,3,3.256158018271,1.5468419231568364,6.168010316005508,3.3172355634071002,1.5617019347980559,6.185732763079662,-0.0011366525249161263
|
|
gsm8k,divergence,7,4.91009951654837,2.045185653260748,7.270690734337546,4.865016424592383,2.026755035640929,7.1787176847663465,-0.02816947521415858
|
|
gsm8k,divergence,11,5.931935024726259,1.9898847070739774,8.459456344451384,5.89597820952082,1.960013675329343,8.45922799810914,0.0012143867431798627
|
|
gsm8k,divergence,15,5.189684726756504,2.680351367454612,7.55120190386808,5.188558523298806,2.6608354187785666,7.605217486545163,0.025876923734101775
|
|
gsm8k,divergence,19,2.7570989361980978,1.7919480814325728,4.193726547659212,2.775993252952797,1.806611536240062,4.181110519373861,0.015515300540850818
|
|
gsm8k,divergence,23,3.137382710336168,1.9856284660015169,4.515373518991637,3.0794746938791713,1.946750149457116,4.410736135976342,-0.052616624917172894
|
|
gsm8k,QKV,3,8.818242117491618,7.409517860590123,11.177278554033984,8.849729752069354,7.434097045185119,11.160648347342136,0.017932250236647462
|
|
gsm8k,QKV,7,10.336483028296383,8.60447573440056,11.851083219477552,10.319640090875208,8.598477516226366,11.845101512915903,-0.011013131675529308
|
|
gsm8k,QKV,11,11.951775082183595,9.533230055647254,13.350917876999524,11.918208649591278,9.580451951366072,13.340366194176807,0.00937949462949117
|
|
gsm8k,QKV,15,10.23944961335793,8.248327352722713,12.111292674550386,10.31412657696469,8.27734572624579,12.14257864987415,0.06657806670976196
|
|
gsm8k,QKV,19,12.3316091968019,10.589038220184888,13.241376894214595,12.367902101381986,10.614451371174253,13.264109489521545,0.015430541534902531
|
|
gsm8k,QKV,23,8.242238784536774,6.1055890731614015,10.372955699499842,8.112160180891495,5.944619005378508,10.25239880721615,-0.12305989236341386
|