{"compatibility_fields":["generated_tokens_per_request","kv_policy","batch_size","chunk_size","timing_policy","measurement_implementation_id"],"control_device":{"device_id":"device:b1d6d78627a957560382cd77a542417ced7d785adc740d8cac42a998e8adfb64","label":"Apple M5"},"current_device":{"device_id":"device:128598125f8228296a25ce379995639546eaa1a5bd3a0acfe97f2329a7da305d","label":"Apple M6"},"rows":[{"bootstrap_resamples":10000,"ci95":[2.8537180627947563,2.948049512578407],"control":{"bootstrap_resamples":10000,"ci95":[143.88797601519144,148.70566808301334],"estimator":"equal-weight mean of session means; session bootstrap","mean":145.83773740085132,"samples":12,"session_means":{"1":144.9195681043492,"2":148.70566808301334,"3":143.88797601519144},"sessions":3},"current":{"bootstrap_resamples":10000,"ci95":[424.1405467448024,424.7143980304134],"estimator":"equal-weight mean of session means; session bootstrap","mean":424.3479407775649,"samples":12,"session_means":{"1":424.1405467448024,"2":424.71439803041346,"3":424.1888775574786},"sessions":3},"estimator":"ratio of independently resampled device session means","metric":"prefill_tokens_per_second","model_id":"Qwen/Qwen3.5-4B","prompt_tokens":256,"ratio":2.90972657928172},{"bootstrap_resamples":10000,"ci95":[2.480534588369262,2.592311953345744],"control":{"bootstrap_resamples":10000,"ci95":[170.24441141611067,178.3023961166031],"estimator":"equal-weight mean of session means; session bootstrap","mean":173.76018396576697,"samples":12,"session_means":{"1":178.30239611660312,"2":172.7337443645872,"3":170.24441141611067},"sessions":3},"current":{"bootstrap_resamples":10000,"ci95":[440.36798465223757,443.243898808408],"estimator":"equal-weight mean of session means; session bootstrap","mean":442.1328780248361,"samples":12,"session_means":{"1":440.36798465223757,"2":443.243898808408,"3":442.78675061386264},"sessions":3},"estimator":"ratio of independently resampled device session means","metric":"prefill_tokens_per_second","model_id":"Qwen/Qwen3.5-4B","prompt_tokens":512,"ratio":2.544500517517535},{"bootstrap_resamples":10000,"ci95":[2.299808142660805,2.3860760839365556],"control":{"bootstrap_resamples":10000,"ci95":[183.93246489326975,190.89756977365712],"estimator":"equal-weight mean of session means; session bootstrap","mean":187.1397222268951,"samples":12,"session_means":{"1":190.89756977365712,"2":186.58913201375844,"3":183.93246489326975},"sessions":3},"current":{"bootstrap_resamples":10000,"ci95":[438.69999893394544,439.32964505618537],"estimator":"equal-weight mean of session means; session bootstrap","mean":438.96883317715395,"samples":12,"session_means":{"1":438.6999989339455,"2":439.3296450561853,"3":438.87685554133105},"sessions":3},"estimator":"ratio of independently resampled device session means","metric":"prefill_tokens_per_second","model_id":"Qwen/Qwen3.5-4B","prompt_tokens":2048,"ratio":2.3456742799101304},{"bootstrap_resamples":10000,"ci95":[2.200826102827299,2.3116683032521514],"control":{"bootstrap_resamples":10000,"ci95":[180.1884364330548,188.95489849256705],"estimator":"equal-weight mean of session means; session bootstrap","mean":183.22020272362923,"samples":12,"session_means":{"1":188.95489849256705,"2":180.1884364330548,"3":180.51727324526587},"sessions":3},"current":{"bootstrap_resamples":10000,"ci95":[409.6669575628125,416.82960158163536],"estimator":"equal-weight mean of session means; session bootstrap","mean":414.40173632049385,"samples":12,"session_means":{"1":416.70864981703363,"2":416.82960158163536,"3":409.6669575628125},"sessions":3},"estimator":"ratio of independently resampled device session means","metric":"prefill_tokens_per_second","model_id":"Qwen/Qwen3.5-4B","prompt_tokens":8192,"ratio":2.2617687905606165},{"bootstrap_resamples":10000,"ci95":[2.111745406642346,2.363229804160593],"control":{"bootstrap_resamples":10000,"ci95":[139.84225561658275,156.54635595440232],"estimator":"equal-weight mean of session means; session bootstrap","mean":147.27073792165137,"samples":6,"session_means":{"1":156.54635595440232,"2":139.84225561658275,"3":145.423602193969},"sessions":3},"current":{"bootstrap_resamples":10000,"ci95":[330.30299265470745,330.72757584260637],"estimator":"equal-weight mean of session means; session bootstrap","mean":330.5327172337295,"samples":6,"session_means":{"1":330.5675832038748,"2":330.72757584260637,"3":330.30299265470745},"sessions":3},"estimator":"ratio of independently resampled device session means","metric":"prefill_tokens_per_second","model_id":"Qwen/Qwen3.5-4B","prompt_tokens":32768,"ratio":2.244388273586123},{"bootstrap_resamples":10000,"ci95":[2.505259195458837,4.114994149504372],"control":{"bootstrap_resamples":10000,"ci95":[203.36307805641067,334.66805856112376],"estimator":"equal-weight mean of session means; session bootstrap","mean":248.89140635283582,"samples":12,"session_means":{"1":334.66805856112376,"2":208.64308244097296,"3":203.36307805641067},"sessions":3},"current":{"bootstrap_resamples":10000,"ci95":[835.2455217180499,840.022585845893],"estimator":"equal-weight mean of session means; session bootstrap","mean":837.9040150147571,"samples":12,"session_means":{"1":835.2455217180498,"2":838.4439374803286,"3":840.022585845893},"sessions":3},"estimator":"ratio of independently resampled device session means","metric":"prefill_tokens_per_second","model_id":"openbmb/MiniCPM5-2B","prompt_tokens":256,"ratio":3.3665445797952525},{"bootstrap_resamples":10000,"ci95":[2.4229621561129773,3.1819933090122094],"control":{"bootstrap_resamples":10000,"ci95":[271.994469883264,357.76104875859846],"estimator":"equal-weight mean of session means; session bootstrap","mean":302.73512841136943,"samples":12,"session_means":{"1":357.76104875859846,"2":278.4498665922457,"3":271.994469883264},"sessions":3},"current":{"bootstrap_resamples":10000,"ci95":[862.7707856238595,867.7713026694023],"estimator":"equal-weight mean of session means; session bootstrap","mean":865.7945234555451,"samples":12,"session_means":{"1":866.8414820733738,"2":867.7713026694023,"3":862.7707856238595},"sessions":3},"estimator":"ratio of independently resampled device session means","metric":"prefill_tokens_per_second","model_id":"openbmb/MiniCPM5-2B","prompt_tokens":512,"ratio":2.8599076955452176},{"bootstrap_resamples":10000,"ci95":[1.9905528855561923,2.4225903073256485],"control":{"bootstrap_resamples":10000,"ci95":[346.2923781793035,421.51334053126584],"estimator":"equal-weight mean of session means; session bootstrap","mean":371.7143403711275,"samples":12,"session_means":{"1":421.51334053126584,"2":346.2923781793035,"3":347.33730240281307},"sessions":3},"current":{"bootstrap_resamples":10000,"ci95":[838.2698798017091,839.431954541557],"estimator":"equal-weight mean of session means; session bootstrap","mean":838.984577586435,"samples":12,"session_means":{"1":838.2698798017091,"2":839.4319545415572,"3":839.2518984160384},"sessions":3},"estimator":"ratio of independently resampled device session means","metric":"prefill_tokens_per_second","model_id":"openbmb/MiniCPM5-2B","prompt_tokens":2048,"ratio":2.2570680936032086},{"bootstrap_resamples":10000,"ci95":[1.4060340030898462,2.20261903281152],"control":{"bootstrap_resamples":10000,"ci95":[304.6053247675214,477.4086381204813],"estimator":"equal-weight mean of session means; session bootstrap","mean":362.3540090949853,"samples":12,"session_means":{"1":477.4086381204813,"2":305.04806439695335,"3":304.6053247675214},"sessions":3},"current":{"bootstrap_resamples":10000,"ci95":[669.0002882658603,671.2554328270926],"estimator":"equal-weight mean of session means; session bootstrap","mean":670.5028332197217,"samples":12,"session_means":{"1":671.2554328270926,"2":671.2527785662121,"3":669.0002882658603},"sessions":3},"estimator":"ratio of independently resampled device session means","metric":"prefill_tokens_per_second","model_id":"openbmb/MiniCPM5-2B","prompt_tokens":8192,"ratio":1.8504082096245278},{"bootstrap_resamples":10000,"ci95":[1.585185722502187,2.099400346230246],"control":{"bootstrap_resamples":10000,"ci95":[175.76985702955287,232.81975680794787],"estimator":"equal-weight mean of session means; session bootstrap","mean":195.11542349571664,"samples":6,"session_means":{"1":232.81975680794787,"2":175.76985702955287,"3":176.7566566496492},"sessions":3},"current":{"bootstrap_resamples":10000,"ci95":[368.94556835034376,369.1210474374136],"estimator":"equal-weight mean of session means; session bootstrap","mean":369.03692655653725,"samples":6,"session_means":{"1":368.94556835034376,"2":369.0441638818543,"3":369.1210474374136},"sessions":3},"estimator":"ratio of independently resampled device session means","metric":"prefill_tokens_per_second","model_id":"openbmb/MiniCPM5-2B","prompt_tokens":32768,"ratio":1.8913775238513562},{"bootstrap_resamples":10000,"ci95":[1.46495770041703,1.5060147131704609],"control":{"bootstrap_resamples":10000,"ci95":[11.007121648421665,11.32141177883566],"estimator":"equal-weight mean of session means; session bootstrap","mean":11.1289026147931,"samples":12,"session_means":{"1":11.058174417121975,"2":11.32141177883566,"3":11.007121648421663},"sessions":3},"current":{"bootstrap_resamples":10000,"ci95":[16.574990687112994,16.600497325564728],"estimator":"equal-weight mean of session means; session bootstrap","mean":16.584441132463805,"samples":12,"session_means":{"1":16.574990687112994,"2":16.577835384713687,"3":16.600497325564728},"sessions":3},"estimator":"ratio of independently resampled device session means","metric":"decode_tokens_per_second","model_id":"Qwen/Qwen3.5-4B","prompt_tokens":256,"ratio":1.4902135193832073},{"bootstrap_resamples":10000,"ci95":[1.445103874805425,1.5228278963218949],"control":{"bootstrap_resamples":10000,"ci95":[10.862616088661477,11.452827157309349],"estimator":"equal-weight mean of session means; session bootstrap","mean":11.165458975502558,"samples":12,"session_means":{"1":11.452827157309349,"2":11.18093368053685,"3":10.862616088661477},"sessions":3},"current":{"bootstrap_resamples":10000,"ci95":[16.53326471119291,16.55915499816036],"estimator":"equal-weight mean of session means; session bootstrap","mean":16.548499219925894,"samples":12,"session_means":{"1":16.53326471119291,"2":16.553077950424417,"3":16.55915499816036},"sessions":3},"estimator":"ratio of independently resampled device session means","metric":"decode_tokens_per_second","model_id":"Qwen/Qwen3.5-4B","prompt_tokens":512,"ratio":1.4821154469542122},{"bootstrap_resamples":10000,"ci95":[1.4630232133341672,1.5409205192208093],"control":{"bootstrap_resamples":10000,"ci95":[10.539477938894809,11.1054471331734],"estimator":"equal-weight mean of session means; session bootstrap","mean":10.857195276965442,"samples":12,"session_means":{"1":11.1054471331734,"2":10.926660758828119,"3":10.539477938894809},"sessions":3},"current":{"bootstrap_resamples":10000,"ci95":[16.23238682262521,16.256719808503746],"estimator":"equal-weight mean of session means; session bootstrap","mean":16.24401238410306,"samples":12,"session_means":{"1":16.24293052118022,"2":16.23238682262521,"3":16.256719808503746},"sessions":3},"estimator":"ratio of independently resampled device session means","metric":"decode_tokens_per_second","model_id":"Qwen/Qwen3.5-4B","prompt_tokens":2048,"ratio":1.4961518117451802},{"bootstrap_resamples":10000,"ci95":[1.4302754347369806,1.5391161581342894],"control":{"bootstrap_resamples":10000,"ci95":[10.07059213795499,10.850354003113027],"estimator":"equal-weight mean of session means; session bootstrap","mean":10.396358057150453,"samples":12,"session_means":{"1":10.850354003113027,"2":10.07059213795499,"3":10.268128030383346},"sessions":3},"current":{"bootstrap_resamples":10000,"ci95":[15.46144366681474,15.546703930093933],"estimator":"equal-weight mean of session means; session bootstrap","mean":15.509047461920431,"samples":12,"session_means":{"1":15.518994788852623,"2":15.546703930093933,"3":15.46144366681474},"sessions":3},"estimator":"ratio of independently resampled device session means","metric":"decode_tokens_per_second","model_id":"Qwen/Qwen3.5-4B","prompt_tokens":8192,"ratio":1.4917769642662078},{"bootstrap_resamples":10000,"ci95":[1.4706014441456563,1.664409187559763],"control":{"bootstrap_resamples":10000,"ci95":[7.576773124361644,8.581574228350895],"estimator":"equal-weight mean of session means; session bootstrap","mean":7.947389860424088,"samples":6,"session_means":{"1":8.581574228350895,"2":7.576773124361644,"3":7.683822228559722},"sessions":3},"current":{"bootstrap_resamples":10000,"ci95":[12.600184193642272,12.63218401344569],"estimator":"equal-weight mean of session means; session bootstrap","mean":12.615463126749694,"samples":6,"session_means":{"1":12.614021173161115,"2":12.600184193642272,"3":12.632184013445691},"sessions":3},"estimator":"ratio of independently resampled device session means","metric":"decode_tokens_per_second","model_id":"Qwen/Qwen3.5-4B","prompt_tokens":32768,"ratio":1.5873718728171853},{"bootstrap_resamples":10000,"ci95":[1.27196698399449,1.5040302704513688],"control":{"bootstrap_resamples":10000,"ci95":[23.78989718980863,28.173980587500093],"estimator":"equal-weight mean of session means; session bootstrap","mean":25.328074370806117,"samples":12,"session_means":{"1":28.173980587500093,"2":24.02034533510962,"3":23.78989718980863},"sessions":3},"current":{"bootstrap_resamples":10000,"ci95":[35.72507789379446,35.89202072560548],"estimator":"equal-weight mean of session means; session bootstrap","mean":35.828546753770745,"samples":12,"session_means":{"1":35.72507789379446,"2":35.86854164191228,"3":35.89202072560548},"sessions":3},"estimator":"ratio of independently resampled device session means","metric":"decode_tokens_per_second","model_id":"openbmb/MiniCPM5-2B","prompt_tokens":256,"ratio":1.4145783934947609},{"bootstrap_resamples":10000,"ci95":[1.2499323303654482,1.5924506201847306],"control":{"bootstrap_resamples":10000,"ci95":[22.266674310593583,28.3772019742706],"estimator":"equal-weight mean of session means; session bootstrap","mean":24.546253903076906,"samples":12,"session_means":{"1":28.3772019742706,"2":22.26667431059358,"3":22.994885424366537},"sessions":3},"current":{"bootstrap_resamples":10000,"ci95":[35.43818198059474,35.499373984879],"estimator":"equal-weight mean of session means; session bootstrap","mean":35.464080754153606,"samples":12,"session_means":{"1":35.43818198059474,"2":35.499373984879,"3":35.45468629698708},"sessions":3},"estimator":"ratio of independently resampled device session means","metric":"decode_tokens_per_second","model_id":"openbmb/MiniCPM5-2B","prompt_tokens":512,"ratio":1.4447858681078882},{"bootstrap_resamples":10000,"ci95":[1.2739056416182417,1.61434982981936],"control":{"bootstrap_resamples":10000,"ci95":[20.592479545945572,26.101481261855913],"estimator":"equal-weight mean of session means; session bootstrap","mean":22.472956490481508,"samples":12,"session_means":{"1":26.101481261855916,"2":20.592479545945572,"3":20.724908663643035},"sessions":3},"current":{"bootstrap_resamples":10000,"ci95":[33.19220598481087,33.280133358701164],"estimator":"equal-weight mean of session means; session bootstrap","mean":33.24714504231347,"samples":12,"session_means":{"1":33.19220598481087,"2":33.280133358701164,"3":33.26909578342839},"sessions":3},"estimator":"ratio of independently resampled device session means","metric":"decode_tokens_per_second","model_id":"openbmb/MiniCPM5-2B","prompt_tokens":2048,"ratio":1.4794290665046854},{"bootstrap_resamples":10000,"ci95":[1.2432025464244507,1.5933612990861634],"control":{"bootstrap_resamples":10000,"ci95":[18.880889442646602,24.21345536860559],"estimator":"equal-weight mean of session means; session bootstrap","mean":20.773751432787137,"samples":12,"session_means":{"1":24.21345536860559,"2":19.226909487109218,"3":18.880889442646602},"sessions":3},"current":{"bootstrap_resamples":10000,"ci95":[30.05872677480288,30.13853105548053],"estimator":"equal-weight mean of session means; session bootstrap","mean":30.093778786840346,"samples":12,"session_means":{"1":30.058726774802885,"2":30.138531055480534,"3":30.084078530237615},"sessions":3},"estimator":"ratio of independently resampled device session means","metric":"decode_tokens_per_second","model_id":"openbmb/MiniCPM5-2B","prompt_tokens":8192,"ratio":1.4486444051382767},{"bootstrap_resamples":10000,"ci95":[1.4675951062862318,1.7696905534617986],"control":{"bootstrap_resamples":10000,"ci95":[11.428212133147857,13.81361472823987],"estimator":"equal-weight mean of session means; session bootstrap","mean":12.453155264288476,"samples":6,"session_means":{"1":13.81361472823987,"2":11.428212133147856,"3":12.117638931477702},"sessions":3},"current":{"bootstrap_resamples":10000,"ci95":[20.17595308990742,20.321213517978663],"estimator":"equal-weight mean of session means; session bootstrap","mean":20.24859621513876,"samples":6,"session_means":{"1":20.17595308990742,"2":20.248622037530204,"3":20.321213517978663},"sessions":3},"estimator":"ratio of independently resampled device session means","metric":"decode_tokens_per_second","model_id":"openbmb/MiniCPM5-2B","prompt_tokens":32768,"ratio":1.6259811899402739}],"schema":"asbench-derived-table/2","source_baseline_ids":["c01:199027b9b416da78ab842f6ea7361e942048bb5daa6deb70cdf59e0e9b0ca96f","c01:5a12b067863346dba4097d5e6e9a1da264e6a1dacd911bec2cce7826a971339a"]}
