{"step": 1, "total_steps": 28610, "tokens": 2097152, "source": "fineweb_hq", "phase": 1, "loss": 9.073994517326355, "ce_loss": 9.046570062637329, "logit_z_loss": 0.10562420077621937, "router_aux_loss": 1.0150056183338165, "router_z_loss": 17.263988971710205, "routing_entropy_normalized": 0.9620906325822177, "learning_rate": 3e-06, "grad_norm": 0.36904314160346985, "tokens_per_second": 41938.777033319384, "eta_seconds": 1430595.40149045, "data_wait_seconds": 5.502631962299347} {"step": 10, "total_steps": 28610, "tokens": 20971520, "source": "fineweb_hq", "phase": 1, "loss": 9.055508732795715, "ce_loss": 9.02808964252472, "logit_z_loss": 0.10522722825407982, "router_aux_loss": 1.0159947872161865, "router_z_loss": 17.248795747756958, "routing_entropy_normalized": 0.9600620670177143, "learning_rate": 2.9999999999999997e-05, "grad_norm": 0.3937816321849823, "tokens_per_second": 589150.7098755764, "eta_seconds": 101805.10045157537, "data_wait_seconds": 2.0503997802734375e-05} {"step": 20, "total_steps": 28610, "tokens": 41943040, "source": "fineweb_hq", "phase": 1, "loss": 8.964744091033936, "ce_loss": 8.937372088432312, "logit_z_loss": 0.10431208368390799, "router_aux_loss": 1.0176522433757782, "router_z_loss": 17.185110092163086, "routing_entropy_normalized": 0.9568416245510577, "learning_rate": 5.9999999999999995e-05, "grad_norm": 0.6630691885948181, "tokens_per_second": 2107568.5835980666, "eta_seconds": 28448.694930553436, "data_wait_seconds": 2.3573637008666992e-05} {"step": 30, "total_steps": 28610, "tokens": 62914560, "source": "fineweb_hq", "phase": 1, "loss": 8.579839587211609, "ce_loss": 8.552549600601196, "logit_z_loss": 0.11818500328809023, "router_aux_loss": 1.0216142386198044, "router_z_loss": 17.062674045562744, "routing_entropy_normalized": 0.9474573234194218, "learning_rate": 8.999999999999999e-05, "grad_norm": 1.15729820728302, "tokens_per_second": 4129725.0589066464, "eta_seconds": 14513.461139678955, "data_wait_seconds": 2.142786979675293e-05} {"step": 40, "total_steps": 28610, "tokens": 83886080, "source": "fineweb_hq", "phase": 1, "loss": 8.270026326179504, "ce_loss": 8.242919445037842, "logit_z_loss": 0.2141006626188755, "router_aux_loss": 1.0286758691072464, "router_z_loss": 16.798901319503784, "routing_entropy_normalized": 0.9313903150915323, "learning_rate": 0.00011999999999999999, "grad_norm": 1.103349208831787, "tokens_per_second": 1107003.578922341, "eta_seconds": 54124.154411792755, "data_wait_seconds": 1.704692840576172e-05} {"step": 50, "total_steps": 28610, "tokens": 104857600, "source": "fineweb_edu", "phase": 1, "loss": 7.979007363319397, "ce_loss": 7.9522459506988525, "logit_z_loss": 0.4787629134953022, "router_aux_loss": 1.0292921960353851, "router_z_loss": 16.42088222503662, "routing_entropy_normalized": 0.9325584503163254, "learning_rate": 0.00015, "grad_norm": 1.0911495685577393, "tokens_per_second": 4101685.7685916405, "eta_seconds": 14602.44994354248, "data_wait_seconds": 1.9311904907226562e-05} {"step": 60, "total_steps": 28610, "tokens": 125829120, "source": "fineweb_hq", "phase": 1, "loss": 7.6750635504722595, "ce_loss": 7.648828864097595, "logit_z_loss": 1.0512438714504242, "router_aux_loss": 1.0295023769140244, "router_z_loss": 15.83453106880188, "routing_entropy_normalized": 0.9327803314012314, "learning_rate": 0.00017999999999999998, "grad_norm": 0.9790343642234802, "tokens_per_second": 4202982.720652348, "eta_seconds": 14245.523614883421, "data_wait_seconds": 1.952052116394043e-05} {"step": 70, "total_steps": 28610, "tokens": 146800640, "source": "fineweb_edu", "phase": 1, "loss": 7.454107284545898, "ce_loss": 7.428394019603729, "logit_z_loss": 1.92060786485672, "router_aux_loss": 1.0272448360919952, "router_z_loss": 15.248664021492004, "routing_entropy_normalized": 0.9381905292100782, "learning_rate": 0.00020999999999999998, "grad_norm": 0.8559066653251648, "tokens_per_second": 3371260.2574310955, "eta_seconds": 17753.811189174652, "data_wait_seconds": 1.901388168334961e-05} {"step": 80, "total_steps": 28610, "tokens": 167772160, "source": "fineweb_hq", "phase": 1, "loss": 7.25084000825882, "ce_loss": 7.225758194923401, "logit_z_loss": 2.8927885591983795, "router_aux_loss": 1.030070498585701, "router_z_loss": 14.491877794265747, "routing_entropy_normalized": 0.9341402720953963, "learning_rate": 0.00023999999999999998, "grad_norm": 0.5455807447433472, "tokens_per_second": 4119744.195539464, "eta_seconds": 14523.170303821564, "data_wait_seconds": 1.946091651916504e-05} {"step": 90, "total_steps": 28610, "tokens": 188743680, "source": "fineweb_edu", "phase": 1, "loss": 7.154139280319214, "ce_loss": 7.129468023777008, "logit_z_loss": 3.583126336336136, "router_aux_loss": 1.0286142379045486, "router_z_loss": 14.02677857875824, "routing_entropy_normalized": 0.9412246630090678, "learning_rate": 0.00027, "grad_norm": 0.3389282524585724, "tokens_per_second": 3321389.154513384, "eta_seconds": 18007.75887966156, "data_wait_seconds": 2.327561378479004e-05} {"step": 100, "total_steps": 28610, "tokens": 209715200, "source": "fineweb_hq", "phase": 1, "loss": 6.977815926074982, "ce_loss": 6.953558504581451, "logit_z_loss": 3.842698246240616, "router_aux_loss": 1.0332032442092896, "router_z_loss": 13.541024327278137, "routing_entropy_normalized": 0.9356706655893746, "learning_rate": 0.0003, "grad_norm": 0.26512834429740906, "tokens_per_second": 3789044.883556809, "eta_seconds": 15779.65037560463, "data_wait_seconds": 1.996755599975586e-05} {"step": 110, "total_steps": 28610, "tokens": 230686720, "source": "fineweb_hq", "phase": 1, "loss": 6.898625373840332, "ce_loss": 6.8747740387916565, "logit_z_loss": 4.094127714633942, "router_aux_loss": 1.0331053882837296, "router_z_loss": 13.110895991325378, "routing_entropy_normalized": 0.9411351697257977, "learning_rate": 0.00033, "grad_norm": 0.2597975432872772, "tokens_per_second": 3735055.8041023863, "eta_seconds": 16002.125573158264, "data_wait_seconds": 2.0772218704223633e-05} {"step": 120, "total_steps": 28610, "tokens": 251658240, "source": "fineweb_hq", "phase": 1, "loss": 6.798978626728058, "ce_loss": 6.775203168392181, "logit_z_loss": 4.276031851768494, "router_aux_loss": 1.0313971191644669, "router_z_loss": 13.033982276916504, "routing_entropy_normalized": 0.9455665700199555, "learning_rate": 0.00035999999999999997, "grad_norm": 0.26467564702033997, "tokens_per_second": 3665904.2181408326, "eta_seconds": 16298.260108470917, "data_wait_seconds": 1.9162893295288086e-05} {"step": 130, "total_steps": 28610, "tokens": 272629760, "source": "fineweb_hq", "phase": 1, "loss": 6.614393591880798, "ce_loss": 6.591010928153992, "logit_z_loss": 4.078372240066528, "router_aux_loss": 1.0381052941083908, "router_z_loss": 12.593927025794983, "routing_entropy_normalized": 0.937294403890337, "learning_rate": 0.00039, "grad_norm": 0.27912798523902893, "tokens_per_second": 3925328.4324998586, "eta_seconds": 15215.768562316895, "data_wait_seconds": 1.958012580871582e-05} {"step": 140, "total_steps": 28610, "tokens": 293601280, "source": "fineweb_hq", "phase": 1, "loss": 6.468375325202942, "ce_loss": 6.444966495037079, "logit_z_loss": 4.336017906665802, "router_aux_loss": 1.033752515912056, "router_z_loss": 12.637640118598938, "routing_entropy_normalized": 0.9428963136152974, "learning_rate": 0.00041999999999999996, "grad_norm": 0.2304970920085907, "tokens_per_second": 3953973.20982587, "eta_seconds": 15100.233175992966, "data_wait_seconds": 2.0205974578857422e-05} {"step": 150, "total_steps": 28610, "tokens": 314572800, "source": "fineweb_edu", "phase": 1, "loss": 6.38175904750824, "ce_loss": 6.358521997928619, "logit_z_loss": 4.819796442985535, "router_aux_loss": 1.032863363623619, "router_z_loss": 12.426431655883789, "routing_entropy_normalized": 0.9432915428506657, "learning_rate": 0.00045, "grad_norm": 0.3845673203468323, "tokens_per_second": 3776318.5268701157, "eta_seconds": 15805.061330318451, "data_wait_seconds": 1.7195940017700195e-05} {"step": 160, "total_steps": 28610, "tokens": 335544320, "source": "fineweb_hq", "phase": 1, "loss": 6.226489305496216, "ce_loss": 6.203530251979828, "logit_z_loss": 4.996673285961151, "router_aux_loss": 1.0317778289318085, "router_z_loss": 12.14176607131958, "routing_entropy_normalized": 0.9463213114382736, "learning_rate": 0.00047999999999999996, "grad_norm": 0.24340784549713135, "tokens_per_second": 3965274.8300431576, "eta_seconds": 15046.61768913269, "data_wait_seconds": 1.6570091247558594e-05} {"step": 170, "total_steps": 28610, "tokens": 356515840, "source": "fineweb_edu", "phase": 1, "loss": 6.168450474739075, "ce_loss": 6.145536780357361, "logit_z_loss": 4.875605285167694, "router_aux_loss": 1.0287566035985947, "router_z_loss": 12.138543725013733, "routing_entropy_normalized": 0.9464020197920747, "learning_rate": 0.00051, "grad_norm": 0.22221329808235168, "tokens_per_second": 3201967.8943166654, "eta_seconds": 18626.98342037201, "data_wait_seconds": 2.2351741790771484e-05} {"step": 180, "total_steps": 28610, "tokens": 377487360, "source": "fineweb_hq", "phase": 1, "loss": 6.026467323303223, "ce_loss": 6.00395667552948, "logit_z_loss": 4.965599536895752, "router_aux_loss": 1.0341257005929947, "router_z_loss": 11.67285430431366, "routing_entropy_normalized": 0.936981342961308, "learning_rate": 0.00054, "grad_norm": 0.3171170949935913, "tokens_per_second": 3356365.6935463715, "eta_seconds": 17763.86627793312, "data_wait_seconds": 1.913309097290039e-05} {"step": 190, "total_steps": 28610, "tokens": 398458880, "source": "fineweb_edu", "phase": 1, "loss": 5.9901896715164185, "ce_loss": 5.967985987663269, "logit_z_loss": 5.4937387108802795, "router_aux_loss": 1.0335818380117416, "router_z_loss": 11.31857442855835, "routing_entropy_normalized": 0.9365698092361668, "learning_rate": 0.0005700000000000001, "grad_norm": 0.433395653963089, "tokens_per_second": 3735590.6813575695, "eta_seconds": 15954.922507286072, "data_wait_seconds": 2.130866050720215e-05} {"step": 200, "total_steps": 28610, "tokens": 419430400, "source": "fineweb_hq", "phase": 1, "loss": 5.849806308746338, "ce_loss": 5.828191161155701, "logit_z_loss": 5.523493230342865, "router_aux_loss": 1.0332033932209015, "router_z_loss": 10.73071539402008, "routing_entropy_normalized": 0.9392291897953079, "learning_rate": 0.0006, "grad_norm": 0.26707807183265686, "tokens_per_second": 3512841.2527675414, "eta_seconds": 16960.654932260513, "data_wait_seconds": 2.0623207092285156e-05} {"step": 210, "total_steps": 28610, "tokens": 440401920, "source": "fineweb_hq", "phase": 1, "loss": 5.789299488067627, "ce_loss": 5.7680951952934265, "logit_z_loss": 5.37375545501709, "router_aux_loss": 1.03995081782341, "router_z_loss": 10.26725447177887, "routing_entropy_normalized": 0.9331191077656645, "learning_rate": 0.00063, "grad_norm": 0.3277837336063385, "tokens_per_second": 4326161.923050119, "eta_seconds": 13767.195463180542, "data_wait_seconds": 2.4527311325073242e-05} {"step": 220, "total_steps": 28610, "tokens": 461373440, "source": "fineweb_hq", "phase": 1, "loss": 5.754351615905762, "ce_loss": 5.733106255531311, "logit_z_loss": 5.4533281326293945, "router_aux_loss": 1.0382436513900757, "router_z_loss": 10.317654967308044, "routing_entropy_normalized": 0.9395890160509555, "learning_rate": 0.00066, "grad_norm": 0.3422394394874573, "tokens_per_second": 3484067.579029589, "eta_seconds": 17088.688416481018, "data_wait_seconds": 2.0176172256469727e-05} {"step": 230, "total_steps": 28610, "tokens": 482344960, "source": "fineweb_hq", "phase": 1, "loss": 5.676178991794586, "ce_loss": 5.655082106590271, "logit_z_loss": 5.8926796317100525, "router_aux_loss": 1.0480426847934723, "router_z_loss": 10.027278900146484, "routing_entropy_normalized": 0.9332907984313041, "learning_rate": 0.0006900000000000001, "grad_norm": 0.2869788408279419, "tokens_per_second": 3749454.125665829, "eta_seconds": 15873.556993961334, "data_wait_seconds": 1.9490718841552734e-05} {"step": 240, "total_steps": 28610, "tokens": 503316480, "source": "fineweb_hq", "phase": 1, "loss": 5.61147153377533, "ce_loss": 5.590435147285461, "logit_z_loss": 6.474088132381439, "router_aux_loss": 1.053972065448761, "router_z_loss": 9.849324941635132, "routing_entropy_normalized": 0.9290603602817278, "learning_rate": 0.0007199999999999999, "grad_norm": 0.33774682879447937, "tokens_per_second": 3825417.806882809, "eta_seconds": 15552.863829135895, "data_wait_seconds": 2.0444393157958984e-05} {"step": 250, "total_steps": 28610, "tokens": 524288000, "source": "fineweb_edu", "phase": 1, "loss": 5.610336422920227, "ce_loss": 5.589332938194275, "logit_z_loss": 6.556094706058502, "router_aux_loss": 1.0614578574895859, "router_z_loss": 9.733118772506714, "routing_entropy_normalized": 0.9243547477714621, "learning_rate": 0.00075, "grad_norm": 0.3457306921482086, "tokens_per_second": 3661759.3486100407, "eta_seconds": 16242.255445480347, "data_wait_seconds": 1.856684684753418e-05} {"step": 260, "total_steps": 28610, "tokens": 545259520, "source": "fineweb_hq", "phase": 1, "loss": 5.5172039270401, "ce_loss": 5.496480822563171, "logit_z_loss": 6.524185061454773, "router_aux_loss": 1.0688736885786057, "router_z_loss": 9.382022023200989, "routing_entropy_normalized": 0.9318456118932588, "learning_rate": 0.00078, "grad_norm": 0.22909705340862274, "tokens_per_second": 3765612.1404037904, "eta_seconds": 15788.736859560013, "data_wait_seconds": 2.294778823852539e-05} {"step": 270, "total_steps": 28610, "tokens": 566231040, "source": "fineweb_edu", "phase": 1, "loss": 5.5132521986961365, "ce_loss": 5.492914617061615, "logit_z_loss": 6.372886538505554, "router_aux_loss": 1.0843495279550552, "router_z_loss": 8.856839776039124, "routing_entropy_normalized": 0.9234213649311608, "learning_rate": 0.0008100000000000001, "grad_norm": 0.40789803862571716, "tokens_per_second": 3457978.330706527, "eta_seconds": 17187.293266773224, "data_wait_seconds": 2.03549861907959e-05} {"step": 280, "total_steps": 28610, "tokens": 587202560, "source": "fineweb_hq", "phase": 1, "loss": 5.444625377655029, "ce_loss": 5.424403190612793, "logit_z_loss": 6.937403976917267, "router_aux_loss": 1.0929326713085175, "router_z_loss": 8.59907180070877, "routing_entropy_normalized": 0.9221462829721346, "learning_rate": 0.0008399999999999999, "grad_norm": 0.28332892060279846, "tokens_per_second": 3369761.575879382, "eta_seconds": 17631.014783143997, "data_wait_seconds": 2.0831823348999023e-05} {"step": 290, "total_steps": 28610, "tokens": 608174080, "source": "fineweb_edu", "phase": 1, "loss": 5.458544015884399, "ce_loss": 5.438980400562286, "logit_z_loss": 7.198760628700256, "router_aux_loss": 1.1046753525733948, "router_z_loss": 7.796859443187714, "routing_entropy_normalized": 0.9172121723952192, "learning_rate": 0.00087, "grad_norm": 0.3885813057422638, "tokens_per_second": 3946078.290696833, "eta_seconds": 15050.726383209229, "data_wait_seconds": 1.8924474716186523e-05} {"step": 300, "total_steps": 28610, "tokens": 629145600, "source": "fineweb_hq", "phase": 1, "loss": 5.371817767620087, "ce_loss": 5.352827608585358, "logit_z_loss": 7.224969208240509, "router_aux_loss": 1.1201239973306656, "router_z_loss": 7.066534221172333, "routing_entropy_normalized": 0.9125549541983051, "learning_rate": 0.0009, "grad_norm": 0.39654383063316345, "tokens_per_second": 4044506.3058676585, "eta_seconds": 14679.263334035873, "data_wait_seconds": 2.2023916244506836e-05} {"step": 310, "total_steps": 28610, "tokens": 650117120, "source": "fineweb_hq", "phase": 1, "loss": 5.344034552574158, "ce_loss": 5.324778199195862, "logit_z_loss": 7.347847044467926, "router_aux_loss": 1.1336359530687332, "router_z_loss": 7.185166895389557, "routing_entropy_normalized": 0.9100935114274862, "learning_rate": 0.00093, "grad_norm": 0.28645825386047363, "tokens_per_second": 3722775.438646952, "eta_seconds": 15942.245934009552, "data_wait_seconds": 2.3484230041503906e-05} {"step": 320, "total_steps": 28610, "tokens": 671088640, "source": "fineweb_hq", "phase": 1, "loss": 5.322923719882965, "ce_loss": 5.304147362709045, "logit_z_loss": 7.559026896953583, "router_aux_loss": 1.1401750296354294, "router_z_loss": 6.618620872497559, "routing_entropy_normalized": 0.9063468496251164, "learning_rate": 0.0009599999999999999, "grad_norm": 0.24293020367622375, "tokens_per_second": 3538588.621823214, "eta_seconds": 16766.128086805344, "data_wait_seconds": 2.002716064453125e-05} {"step": 330, "total_steps": 28610, "tokens": 692060160, "source": "fineweb_hq", "phase": 1, "loss": 5.253985226154327, "ce_loss": 5.235036313533783, "logit_z_loss": 7.848530113697052, "router_aux_loss": 1.1589565426111221, "router_z_loss": 6.57447612285614, "routing_entropy_normalized": 0.8965382428093612, "learning_rate": 0.00099, "grad_norm": 0.27227944135665894, "tokens_per_second": 3994233.3172515226, "eta_seconds": 14848.270956993103, "data_wait_seconds": 1.8984079360961914e-05} {"step": 340, "total_steps": 28610, "tokens": 713031680, "source": "fineweb_hq", "phase": 1, "loss": 5.239398002624512, "ce_loss": 5.220990836620331, "logit_z_loss": 8.02791178226471, "router_aux_loss": 1.1757470071315765, "router_z_loss": 5.846938669681549, "routing_entropy_normalized": 0.8859969556602151, "learning_rate": 0.00102, "grad_norm": 0.327094703912735, "tokens_per_second": 3421397.879941322, "eta_seconds": 17328.147476673126, "data_wait_seconds": 2.193450927734375e-05} {"step": 350, "total_steps": 28610, "tokens": 734003200, "source": "fineweb_edu", "phase": 1, "loss": 5.260049343109131, "ce_loss": 5.241715848445892, "logit_z_loss": 7.85910838842392, "router_aux_loss": 1.1920885890722275, "router_z_loss": 5.626796901226044, "routing_entropy_normalized": 0.8691016782812263, "learning_rate": 0.0010500000000000002, "grad_norm": 0.4705367982387543, "tokens_per_second": 3794660.4059493053, "eta_seconds": 15618.134214878082, "data_wait_seconds": 1.952052116394043e-05} {"step": 360, "total_steps": 28610, "tokens": 754974720, "source": "fineweb_hq", "phase": 1, "loss": 5.195451378822327, "ce_loss": 5.17672061920166, "logit_z_loss": 8.42539668083191, "router_aux_loss": 1.223222330212593, "router_z_loss": 5.656034290790558, "routing_entropy_normalized": 0.8833911314073039, "learning_rate": 0.00108, "grad_norm": 0.29895609617233276, "tokens_per_second": 3511669.379188852, "eta_seconds": 16870.763617753983, "data_wait_seconds": 2.7626752853393555e-05} {"step": 370, "total_steps": 28610, "tokens": 775946240, "source": "fineweb_edu", "phase": 1, "loss": 5.225911021232605, "ce_loss": 5.207927107810974, "logit_z_loss": 7.770734369754791, "router_aux_loss": 1.2471286058425903, "router_z_loss": 4.735528767108917, "routing_entropy_normalized": 0.8719184432847817, "learning_rate": 0.00111, "grad_norm": 0.5935752391815186, "tokens_per_second": 3798979.7032318194, "eta_seconds": 15589.336376190186, "data_wait_seconds": 1.7762184143066406e-05} {"step": 380, "total_steps": 28610, "tokens": 796917760, "source": "fineweb_hq", "phase": 1, "loss": 5.149052083492279, "ce_loss": 5.131161630153656, "logit_z_loss": 8.167403817176819, "router_aux_loss": 1.2667809426784515, "router_z_loss": 4.40583062171936, "routing_entropy_normalized": 0.8729008473567185, "learning_rate": 0.0011400000000000002, "grad_norm": 0.3236619830131531, "tokens_per_second": 4274623.386485481, "eta_seconds": 13849.781748533249, "data_wait_seconds": 1.627206802368164e-05} {"step": 390, "total_steps": 28610, "tokens": 817889280, "source": "fineweb_edu", "phase": 1, "loss": 5.128994345664978, "ce_loss": 5.111594498157501, "logit_z_loss": 8.529348969459534, "router_aux_loss": 1.2694161981344223, "router_z_loss": 3.852686494588852, "routing_entropy_normalized": 0.8660489053634466, "learning_rate": 0.00117, "grad_norm": 0.41736388206481934, "tokens_per_second": 3098391.168666903, "eta_seconds": 19100.761078357697, "data_wait_seconds": 1.7881393432617188e-05} {"step": 400, "total_steps": 28610, "tokens": 838860800, "source": "fineweb_hq", "phase": 1, "loss": 5.090546727180481, "ce_loss": 5.073071599006653, "logit_z_loss": 8.49722170829773, "router_aux_loss": 1.2949055582284927, "router_z_loss": 3.676444709300995, "routing_entropy_normalized": 0.8672269490131654, "learning_rate": 0.0012, "grad_norm": 0.36304622888565063, "tokens_per_second": 3477695.175258208, "eta_seconds": 17011.455846071243, "data_wait_seconds": 2.0325183868408203e-05} {"step": 410, "total_steps": 28610, "tokens": 859832320, "source": "fineweb_hq", "phase": 1, "loss": 5.064834654331207, "ce_loss": 5.047110915184021, "logit_z_loss": 8.8423353433609, "router_aux_loss": 1.3084731101989746, "router_z_loss": 3.7548621892929077, "routing_entropy_normalized": 0.8668069846089523, "learning_rate": 0.00123, "grad_norm": 0.285069078207016, "tokens_per_second": 3221232.3926164685, "eta_seconds": 18359.335556030273, "data_wait_seconds": 0.08201679587364197} {"step": 420, "total_steps": 28610, "tokens": 880803840, "source": "fineweb_hq", "phase": 1, "loss": 5.026031970977783, "ce_loss": 5.00834858417511, "logit_z_loss": 9.209359288215637, "router_aux_loss": 1.340689778327942, "router_z_loss": 3.3554183542728424, "routing_entropy_normalized": 0.8624064720336505, "learning_rate": 0.00126, "grad_norm": 0.28937631845474243, "tokens_per_second": 3243420.811924807, "eta_seconds": 18227.272471904755, "data_wait_seconds": 2.0205974578857422e-05} {"step": 430, "total_steps": 28610, "tokens": 901775360, "source": "fineweb_hq", "phase": 1, "loss": 4.996566116809845, "ce_loss": 4.978649914264679, "logit_z_loss": 8.815405368804932, "router_aux_loss": 1.3385442197322845, "router_z_loss": 3.649264484643936, "routing_entropy_normalized": 0.8658562643781956, "learning_rate": 0.0012900000000000001, "grad_norm": 0.3979099988937378, "tokens_per_second": 2711752.3787492486, "eta_seconds": 21793.19314813614, "data_wait_seconds": 2.3126602172851562e-05} {"step": 440, "total_steps": 28610, "tokens": 922746880, "source": "fineweb_hq", "phase": 1, "loss": 5.012665808200836, "ce_loss": 4.995199501514435, "logit_z_loss": 10.034671306610107, "router_aux_loss": 1.374913051724434, "router_z_loss": 2.7138236314058304, "routing_entropy_normalized": 0.8335043520996626, "learning_rate": 0.00132, "grad_norm": 0.4830191731452942, "tokens_per_second": 3771752.3189891055, "eta_seconds": 15662.951022148132, "data_wait_seconds": 1.800060272216797e-05} {"step": 450, "total_steps": 28610, "tokens": 943718400, "source": "fineweb_edu", "phase": 1, "loss": 5.011710703372955, "ce_loss": 4.99358743429184, "logit_z_loss": 8.91873013973236, "router_aux_loss": 1.3978781700134277, "router_z_loss": 3.252637028694153, "routing_entropy_normalized": 0.8358440784948558, "learning_rate": 0.00135, "grad_norm": 0.4271644651889801, "tokens_per_second": 3261740.5741507495, "eta_seconds": 18105.609252929688, "data_wait_seconds": 2.333521842956543e-05} {"step": 460, "total_steps": 28610, "tokens": 964689920, "source": "fineweb_hq", "phase": 1, "loss": 4.97152978181839, "ce_loss": 4.9525768756866455, "logit_z_loss": 9.51468300819397, "router_aux_loss": 1.4543810337781906, "router_z_loss": 3.457518130540848, "routing_entropy_normalized": 0.8401015661263701, "learning_rate": 0.0013800000000000002, "grad_norm": 0.4248215854167938, "tokens_per_second": 3271589.4996804693, "eta_seconds": 18044.693200588226, "data_wait_seconds": 2.390146255493164e-05} {"step": 470, "total_steps": 28610, "tokens": 985661440, "source": "fineweb_edu", "phase": 1, "loss": 4.94115424156189, "ce_loss": 4.9231467843055725, "logit_z_loss": 9.461663722991943, "router_aux_loss": 1.4546733051538467, "router_z_loss": 2.514692634344101, "routing_entropy_normalized": 0.8327621696468369, "learning_rate": 0.00141, "grad_norm": 0.3548778295516968, "tokens_per_second": 4198693.589163896, "eta_seconds": 14055.2903008461, "data_wait_seconds": 2.4110078811645508e-05} {"step": 480, "total_steps": 28610, "tokens": 1006632960, "source": "fineweb_hq", "phase": 1, "loss": 4.902189314365387, "ce_loss": 4.883037090301514, "logit_z_loss": 9.651476621627808, "router_aux_loss": 1.4716897010803223, "router_z_loss": 3.470106780529022, "routing_entropy_normalized": 0.8425183360603846, "learning_rate": 0.0014399999999999999, "grad_norm": 0.3314315974712372, "tokens_per_second": 3967594.0942508336, "eta_seconds": 14868.679698228836, "data_wait_seconds": 2.6285648345947266e-05} {"step": 490, "total_steps": 28610, "tokens": 1027604480, "source": "fineweb_edu", "phase": 1, "loss": 4.898950099945068, "ce_loss": 4.880696892738342, "logit_z_loss": 10.149771094322205, "router_aux_loss": 1.457972526550293, "router_z_loss": 2.6584786474704742, "routing_entropy_normalized": 0.8454257597435149, "learning_rate": 0.00147, "grad_norm": 0.4022069275379181, "tokens_per_second": 3920839.035617522, "eta_seconds": 15040.636380195618, "data_wait_seconds": 1.913309097290039e-05} {"step": 500, "total_steps": 28610, "tokens": 1048576000, "source": "fineweb_hq", "phase": 1, "loss": 4.847365617752075, "ce_loss": 4.828476548194885, "logit_z_loss": 9.975858449935913, "router_aux_loss": 1.48147913813591, "router_z_loss": 3.0766870975494385, "routing_entropy_normalized": 0.84149642962929, "learning_rate": 0.0015, "grad_norm": 0.36526528000831604, "tokens_per_second": 3910478.4742608904, "eta_seconds": 15075.122675657272, "data_wait_seconds": 2.5570392608642578e-05} {"step": 510, "total_steps": 28610, "tokens": 1069547520, "source": "fineweb_hq", "phase": 1, "loss": 4.814844310283661, "ce_loss": 4.796235978603363, "logit_z_loss": 10.289829730987549, "router_aux_loss": 1.5016645789146423, "router_z_loss": 2.562759593129158, "routing_entropy_normalized": 0.8309261884498913, "learning_rate": 0.0015300000000000001, "grad_norm": 0.3031240701675415, "tokens_per_second": 3285293.2215354447, "eta_seconds": 17937.507317066193, "data_wait_seconds": 1.9997358322143555e-05} {"step": 520, "total_steps": 28610, "tokens": 1090519040, "source": "fineweb_hq", "phase": 1, "loss": 4.787340223789215, "ce_loss": 4.76912522315979, "logit_z_loss": 10.826215505599976, "router_aux_loss": 1.4969516545534134, "router_z_loss": 2.1627984941005707, "routing_entropy_normalized": 0.8335338224968717, "learning_rate": 0.00156, "grad_norm": 0.3507811427116394, "tokens_per_second": 3777777.558257838, "eta_seconds": 15593.559644937515, "data_wait_seconds": 2.1338462829589844e-05} {"step": 530, "total_steps": 28610, "tokens": 1111490560, "source": "fineweb_hq", "phase": 1, "loss": 4.750087261199951, "ce_loss": 4.731242060661316, "logit_z_loss": 10.154732346534729, "router_aux_loss": 1.5312494933605194, "router_z_loss": 2.5173303484916687, "routing_entropy_normalized": 0.8234943176656766, "learning_rate": 0.00159, "grad_norm": 0.3950449228286743, "tokens_per_second": 3994685.173093473, "eta_seconds": 14741.594295501709, "data_wait_seconds": 2.0802021026611328e-05} {"step": 540, "total_steps": 28610, "tokens": 1132462080, "source": "fineweb_hq", "phase": 1, "loss": 4.773451924324036, "ce_loss": 4.755377769470215, "logit_z_loss": 10.728098392486572, "router_aux_loss": 1.5105438083410263, "router_z_loss": 1.895902305841446, "routing_entropy_normalized": 0.8281748098738861, "learning_rate": 0.0016200000000000001, "grad_norm": 0.2856164574623108, "tokens_per_second": 3237227.897222266, "eta_seconds": 18184.402985811234, "data_wait_seconds": 2.2113323211669922e-05} {"step": 550, "total_steps": 28610, "tokens": 1153433600, "source": "fineweb_edu", "phase": 1, "loss": 4.7697367668151855, "ce_loss": 4.751023888587952, "logit_z_loss": 10.269596457481384, "router_aux_loss": 1.522117242217064, "router_z_loss": 2.464825749397278, "routing_entropy_normalized": 0.8268924730287408, "learning_rate": 0.0016500000000000002, "grad_norm": 0.3417162001132965, "tokens_per_second": 4113319.2611890826, "eta_seconds": 14306.228469848633, "data_wait_seconds": 2.1398067474365234e-05} {"step": 560, "total_steps": 28610, "tokens": 1174405120, "source": "fineweb_hq", "phase": 1, "loss": 4.73752373456955, "ce_loss": 4.71814239025116, "logit_z_loss": 10.787253499031067, "router_aux_loss": 1.5451249331235886, "router_z_loss": 2.8512998819351196, "routing_entropy_normalized": 0.8299761225786543, "learning_rate": 0.0016799999999999999, "grad_norm": 0.2863750457763672, "tokens_per_second": 4107901.715805139, "eta_seconds": 14319.990513324738, "data_wait_seconds": 1.773238182067871e-05} {"step": 570, "total_steps": 28610, "tokens": 1195376640, "source": "fineweb_edu", "phase": 1, "loss": 4.693858504295349, "ce_loss": 4.675326585769653, "logit_z_loss": 10.672502160072327, "router_aux_loss": 1.5508647412061691, "router_z_loss": 1.9561002552509308, "routing_entropy_normalized": 0.8406210676870906, "learning_rate": 0.00171, "grad_norm": 0.2887071669101715, "tokens_per_second": 4110416.6146404147, "eta_seconds": 14306.126992225647, "data_wait_seconds": 2.187490463256836e-05} {"step": 580, "total_steps": 28610, "tokens": 1216348160, "source": "fineweb_hq", "phase": 1, "loss": 4.668807566165924, "ce_loss": 4.650371611118317, "logit_z_loss": 10.875535607337952, "router_aux_loss": 1.5739277303218842, "router_z_loss": 1.6090629547834396, "routing_entropy_normalized": 0.825997802728899, "learning_rate": 0.00174, "grad_norm": 0.40386706590652466, "tokens_per_second": 4004732.1548217814, "eta_seconds": 14678.427492141724, "data_wait_seconds": 1.9997358322143555e-05} {"step": 590, "total_steps": 28610, "tokens": 1237319680, "source": "fineweb_edu", "phase": 1, "loss": 4.661233842372894, "ce_loss": 4.6426496505737305, "logit_z_loss": 10.53859806060791, "router_aux_loss": 1.5447949022054672, "router_z_loss": 2.0824166238307953, "routing_entropy_normalized": 0.8147948104423992, "learning_rate": 0.00177, "grad_norm": 0.37953150272369385, "tokens_per_second": 3354839.0208077533, "eta_seconds": 17515.65385866165, "data_wait_seconds": 1.6123056411743164e-05} {"step": 600, "total_steps": 28610, "tokens": 1258291200, "source": "fineweb_hq", "phase": 1, "loss": 4.6461856961250305, "ce_loss": 4.627599895000458, "logit_z_loss": 10.74972689151764, "router_aux_loss": 1.6010109633207321, "router_z_loss": 1.5006704330444336, "routing_entropy_normalized": 0.8078374543586948, "learning_rate": 0.0018, "grad_norm": 0.4571743309497833, "tokens_per_second": 4312127.1617988, "eta_seconds": 13622.331929445269, "data_wait_seconds": 2.086162567138672e-05} {"step": 610, "total_steps": 28610, "tokens": 1279262720, "source": "fineweb_hq", "phase": 1, "loss": 4.602349638938904, "ce_loss": 4.583353698253632, "logit_z_loss": 11.201192855834961, "router_aux_loss": 1.5656961351633072, "router_z_loss": 2.2190346717834473, "routing_entropy_normalized": 0.8232387035378077, "learning_rate": 0.00183, "grad_norm": 0.26169553399086, "tokens_per_second": 3849583.796473134, "eta_seconds": 15253.663539886475, "data_wait_seconds": 1.800060272216797e-05} {"step": 620, "total_steps": 28610, "tokens": 1300234240, "source": "fineweb_hq", "phase": 1, "loss": 4.602822721004486, "ce_loss": 4.584046304225922, "logit_z_loss": 11.272533416748047, "router_aux_loss": 1.6290462166070938, "router_z_loss": 1.3587355017662048, "routing_entropy_normalized": 0.8025154377816331, "learning_rate": 0.00186, "grad_norm": 0.3588958978652954, "tokens_per_second": 4190568.7294397634, "eta_seconds": 14007.474467039108, "data_wait_seconds": 2.6851892471313477e-05} {"step": 630, "total_steps": 28610, "tokens": 1321205760, "source": "fineweb_hq", "phase": 1, "loss": 4.582616686820984, "ce_loss": 4.562750041484833, "logit_z_loss": 11.323569059371948, "router_aux_loss": 1.6738543808460236, "router_z_loss": 1.9957774057984352, "routing_entropy_normalized": 0.7996818237684693, "learning_rate": 0.0018900000000000002, "grad_norm": 0.3162058889865875, "tokens_per_second": 3398635.576695529, "eta_seconds": 17265.255905151367, "data_wait_seconds": 2.3037195205688477e-05} {"step": 640, "total_steps": 28610, "tokens": 1342177280, "source": "fineweb_hq", "phase": 1, "loss": 4.556235313415527, "ce_loss": 4.537401795387268, "logit_z_loss": 11.744758009910583, "router_aux_loss": 1.6015048623085022, "router_z_loss": 1.6439045071601868, "routing_entropy_normalized": 0.8044678263578776, "learning_rate": 0.0019199999999999998, "grad_norm": 0.35044100880622864, "tokens_per_second": 3756238.5086336415, "eta_seconds": 15615.978938817978, "data_wait_seconds": 2.0772218704223633e-05} {"step": 650, "total_steps": 28610, "tokens": 1363148800, "source": "fineweb_edu", "phase": 1, "loss": 4.518477559089661, "ce_loss": 4.499395430088043, "logit_z_loss": 11.374523282051086, "router_aux_loss": 1.6293142884969711, "router_z_loss": 1.6516766399145126, "routing_entropy_normalized": 0.7947154529089143, "learning_rate": 0.00195, "grad_norm": 0.38360729813575745, "tokens_per_second": 3316277.8402965195, "eta_seconds": 17681.380373954773, "data_wait_seconds": 1.9222497940063477e-05} {"step": 660, "total_steps": 28610, "tokens": 1384120320, "source": "fineweb_hq", "phase": 1, "loss": 4.4425437450408936, "ce_loss": 4.423097372055054, "logit_z_loss": 11.37237823009491, "router_aux_loss": 1.6626565903425217, "router_z_loss": 1.6825388222932816, "routing_entropy_normalized": 0.8031857765339157, "learning_rate": 0.00198, "grad_norm": 0.3046814501285553, "tokens_per_second": 3643489.94522065, "eta_seconds": 16087.70691871643, "data_wait_seconds": 1.7255544662475586e-05} {"step": 670, "total_steps": 28610, "tokens": 1405091840, "source": "fineweb_edu", "phase": 1, "loss": 4.475962996482849, "ce_loss": 4.456783711910248, "logit_z_loss": 11.148050785064697, "router_aux_loss": 1.645657703280449, "router_z_loss": 1.6077712774276733, "routing_entropy_normalized": 0.7975695938988208, "learning_rate": 0.00201, "grad_norm": 0.33695414662361145, "tokens_per_second": 3845425.758889983, "eta_seconds": 15237.4354763031, "data_wait_seconds": 2.0891427993774414e-05} {"step": 680, "total_steps": 28610, "tokens": 1426063360, "source": "fineweb_hq", "phase": 1, "loss": 4.378860116004944, "ce_loss": 4.360008001327515, "logit_z_loss": 11.952872514724731, "router_aux_loss": 1.6375892758369446, "router_z_loss": 1.280927300453186, "routing_entropy_normalized": 0.7981218237596236, "learning_rate": 0.00204, "grad_norm": 0.3253091871738434, "tokens_per_second": 3811494.812161504, "eta_seconds": 15367.581027030945, "data_wait_seconds": 2.1219253540039062e-05} {"step": 690, "total_steps": 28610, "tokens": 1447034880, "source": "fineweb_edu", "phase": 1, "loss": 4.3715115785598755, "ce_loss": 4.352568805217743, "logit_z_loss": 11.637056589126587, "router_aux_loss": 1.6241776943206787, "router_z_loss": 1.5373363643884659, "routing_entropy_normalized": 0.8207539881979657, "learning_rate": 0.00207, "grad_norm": 0.35822823643684387, "tokens_per_second": 3583570.173558502, "eta_seconds": 16339.148113250732, "data_wait_seconds": 2.1636486053466797e-05} {"step": 700, "total_steps": 28610, "tokens": 1468006400, "source": "fineweb_hq", "phase": 1, "loss": 4.312666177749634, "ce_loss": 4.293725490570068, "logit_z_loss": 12.11507797241211, "router_aux_loss": 1.6435659229755402, "router_z_loss": 1.2935548275709152, "routing_entropy_normalized": 0.8119192774798145, "learning_rate": 0.0021000000000000003, "grad_norm": 0.300942599773407, "tokens_per_second": 3136614.7157123517, "eta_seconds": 18660.72744822502, "data_wait_seconds": 0.4307662844657898} {"step": 710, "total_steps": 28610, "tokens": 1488977920, "source": "fineweb_hq", "phase": 1, "loss": 4.183160781860352, "ce_loss": 4.163766860961914, "logit_z_loss": 12.157939314842224, "router_aux_loss": 1.622790277004242, "router_z_loss": 1.9503589868545532, "routing_entropy_normalized": 0.8265703172821873, "learning_rate": 0.00213, "grad_norm": 0.2843954861164093, "tokens_per_second": 4206488.292379407, "eta_seconds": 13909.593164920807, "data_wait_seconds": 2.3156404495239258e-05} {"step": 720, "total_steps": 28610, "tokens": 1509949440, "source": "fineweb_hq", "phase": 1, "loss": 4.1320576667785645, "ce_loss": 4.112650454044342, "logit_z_loss": 11.913991093635559, "router_aux_loss": 1.6702817976474762, "router_z_loss": 1.5128950029611588, "routing_entropy_normalized": 0.7667686280578151, "learning_rate": 0.00216, "grad_norm": 0.3295558989048004, "tokens_per_second": 3582569.064308919, "eta_seconds": 16326.152610063553, "data_wait_seconds": 2.2679567337036133e-05} {"step": 730, "total_steps": 28610, "tokens": 1530920960, "source": "fineweb_hq", "phase": 1, "loss": 4.026218414306641, "ce_loss": 4.007347136735916, "logit_z_loss": 12.179679989814758, "router_aux_loss": 1.6417283415794373, "router_z_loss": 1.235919177532196, "routing_entropy_normalized": 0.7842817565096617, "learning_rate": 0.00219, "grad_norm": 0.28047820925712585, "tokens_per_second": 3723436.517013419, "eta_seconds": 15702.858768463135, "data_wait_seconds": 2.4765729904174805e-05} {"step": 740, "total_steps": 28610, "tokens": 1551892480, "source": "fineweb_hq", "phase": 1, "loss": 3.989847034215927, "ce_loss": 3.9702339470386505, "logit_z_loss": 12.447178483009338, "router_aux_loss": 1.6604880839586258, "router_z_loss": 1.7635953277349472, "routing_entropy_normalized": 0.7562691363023605, "learning_rate": 0.00222, "grad_norm": 0.346403032541275, "tokens_per_second": 3894074.253051765, "eta_seconds": 15009.376411914825, "data_wait_seconds": 2.014636993408203e-05} {"step": 750, "total_steps": 28610, "tokens": 1572864000, "source": "fineweb_edu", "phase": 1, "loss": 4.258761405944824, "ce_loss": 4.239522159099579, "logit_z_loss": 12.369433999061584, "router_aux_loss": 1.6498509645462036, "router_z_loss": 1.5039474442601204, "routing_entropy_normalized": 0.7945254378483197, "learning_rate": 0.00225, "grad_norm": 0.3874565660953522, "tokens_per_second": 3626962.30448711, "eta_seconds": 16108.977655410767, "data_wait_seconds": 1.9729137420654297e-05} {"step": 760, "total_steps": 28610, "tokens": 1593835520, "source": "fineweb_hq", "phase": 1, "loss": 3.9048891365528107, "ce_loss": 3.885349154472351, "logit_z_loss": 12.790281653404236, "router_aux_loss": 1.701425939798355, "router_z_loss": 1.2466687560081482, "routing_entropy_normalized": 0.7865120201424554, "learning_rate": 0.0022800000000000003, "grad_norm": 0.29702228307724, "tokens_per_second": 4180731.0651545026, "eta_seconds": 13970.208150148392, "data_wait_seconds": 2.4020671844482422e-05} {"step": 770, "total_steps": 28610, "tokens": 1614807040, "source": "fineweb_edu", "phase": 1, "loss": 4.154159486293793, "ce_loss": 4.134239077568054, "logit_z_loss": 12.853577375411987, "router_aux_loss": 1.7273095399141312, "router_z_loss": 1.3619235306978226, "routing_entropy_normalized": 0.7853914621656923, "learning_rate": 0.00231, "grad_norm": 0.3584041893482208, "tokens_per_second": 3708947.911804715, "eta_seconds": 15741.58307647705, "data_wait_seconds": 2.0563602447509766e-05} {"step": 780, "total_steps": 28610, "tokens": 1635778560, "source": "fineweb_hq", "phase": 1, "loss": 3.835381865501404, "ce_loss": 3.8157514929771423, "logit_z_loss": 12.789737343788147, "router_aux_loss": 1.705956608057022, "router_z_loss": 1.291787475347519, "routing_entropy_normalized": 0.7713168525771344, "learning_rate": 0.00234, "grad_norm": 0.40305009484291077, "tokens_per_second": 3415538.7750444957, "eta_seconds": 17087.71119403839, "data_wait_seconds": 2.6911497116088867e-05} {"step": 790, "total_steps": 28610, "tokens": 1656750080, "source": "fineweb_edu", "phase": 1, "loss": 4.152964353561401, "ce_loss": 4.133545517921448, "logit_z_loss": 12.332417488098145, "router_aux_loss": 1.7060890197753906, "router_z_loss": 1.1245883628726006, "routing_entropy_normalized": 0.7657672862067394, "learning_rate": 0.00237, "grad_norm": 0.46772322058677673, "tokens_per_second": 3153174.1463991515, "eta_seconds": 18502.86915063858, "data_wait_seconds": 2.1070241928100586e-05} {"step": 800, "total_steps": 28610, "tokens": 1677721600, "source": "fineweb_hq", "phase": 1, "loss": 3.761558085680008, "ce_loss": 3.741902142763138, "logit_z_loss": 12.860188007354736, "router_aux_loss": 1.7149255573749542, "router_z_loss": 1.2206990122795105, "routing_entropy_normalized": 0.7890314993665327, "learning_rate": 0.0024, "grad_norm": 0.32772231101989746, "tokens_per_second": 4486069.245902081, "eta_seconds": 13000.645759820938, "data_wait_seconds": 2.047419548034668e-05} {"step": 810, "total_steps": 28610, "tokens": 1698693120, "source": "fineweb_hq", "phase": 1, "loss": 3.7172193825244904, "ce_loss": 3.6978485584259033, "logit_z_loss": 13.18132734298706, "router_aux_loss": 1.6983977407217026, "router_z_loss": 1.0686743780970573, "routing_entropy_normalized": 0.7752630754364466, "learning_rate": 0.0024300000000000003, "grad_norm": 0.27187544107437134, "tokens_per_second": 3851637.2644714345, "eta_seconds": 15136.634526252747, "data_wait_seconds": 2.2709369659423828e-05} {"step": 820, "total_steps": 28610, "tokens": 1719664640, "source": "fineweb_hq", "phase": 1, "loss": 3.7155794203281403, "ce_loss": 3.696614146232605, "logit_z_loss": 12.875759959220886, "router_aux_loss": 1.6699260771274567, "router_z_loss": 0.9784659892320633, "routing_entropy_normalized": 0.7752238752221113, "learning_rate": 0.00246, "grad_norm": 0.3380814790725708, "tokens_per_second": 4257191.374413602, "eta_seconds": 13689.74259185791, "data_wait_seconds": 2.1845102310180664e-05} {"step": 830, "total_steps": 28610, "tokens": 1740636160, "source": "fineweb_hq", "phase": 1, "loss": 3.7125514447689056, "ce_loss": 3.693806678056717, "logit_z_loss": 13.074058175086975, "router_aux_loss": 1.6327433735132217, "router_z_loss": 1.1098729148507118, "routing_entropy_normalized": 0.7661689784290413, "learning_rate": 0.00249, "grad_norm": 0.4026620388031006, "tokens_per_second": 3827754.197857899, "eta_seconds": 15220.121133327484, "data_wait_seconds": 2.5928020477294922e-05} {"step": 840, "total_steps": 28610, "tokens": 1761607680, "source": "fineweb_hq", "phase": 1, "loss": 3.7102659046649933, "ce_loss": 3.6910923421382904, "logit_z_loss": 13.138167023658752, "router_aux_loss": 1.6986212879419327, "router_z_loss": 0.8735387995839119, "routing_entropy_normalized": 0.7574490153708061, "learning_rate": 0.00252, "grad_norm": 0.30801060795783997, "tokens_per_second": 4198816.048648469, "eta_seconds": 13870.079175949097, "data_wait_seconds": 2.1010637283325195e-05} {"step": 850, "total_steps": 28610, "tokens": 1782579200, "source": "fineweb_edu", "phase": 1, "loss": 4.09829968214035, "ce_loss": 4.078723132610321, "logit_z_loss": 13.299319982528687, "router_aux_loss": 1.706539511680603, "router_z_loss": 1.1810516864061356, "routing_entropy_normalized": 0.794611320065966, "learning_rate": 0.00255, "grad_norm": 0.4702201783657074, "tokens_per_second": 3717805.102345282, "eta_seconds": 15658.954118728638, "data_wait_seconds": 2.181529998779297e-05} {"step": 860, "total_steps": 28610, "tokens": 1803550720, "source": "fineweb_hq", "phase": 1, "loss": 3.7052374482154846, "ce_loss": 3.684843987226486, "logit_z_loss": 12.924830555915833, "router_aux_loss": 1.7955583035945892, "router_z_loss": 1.1454517841339111, "routing_entropy_normalized": 0.7589951169785014, "learning_rate": 0.0025800000000000003, "grad_norm": 0.33521220088005066, "tokens_per_second": 3558473.307386744, "eta_seconds": 16354.195457696915, "data_wait_seconds": 2.9385089874267578e-05} {"step": 870, "total_steps": 28610, "tokens": 1824522240, "source": "fineweb_edu", "phase": 1, "loss": 3.977514296770096, "ce_loss": 3.955770641565323, "logit_z_loss": 13.186537623405457, "router_aux_loss": 1.7855572402477264, "router_z_loss": 2.569445922970772, "routing_entropy_normalized": 0.7666084066508106, "learning_rate": 0.00261, "grad_norm": 0.3989207446575165, "tokens_per_second": 3524651.16889396, "eta_seconds": 16505.178439617157, "data_wait_seconds": 2.008676528930664e-05} {"step": 880, "total_steps": 28610, "tokens": 1845493760, "source": "fineweb_hq", "phase": 1, "loss": 3.6696519553661346, "ce_loss": 3.6492239832878113, "logit_z_loss": 13.389381766319275, "router_aux_loss": 1.7835367769002914, "router_z_loss": 1.2537417337298393, "routing_entropy_normalized": 0.7679324976278832, "learning_rate": 0.00264, "grad_norm": 0.3279078006744385, "tokens_per_second": 4268992.312113424, "eta_seconds": 13622.424382209778, "data_wait_seconds": 1.8805265426635742e-05} {"step": 890, "total_steps": 28610, "tokens": 1866465280, "source": "fineweb_edu", "phase": 1, "loss": 3.996062844991684, "ce_loss": 3.9759906232357025, "logit_z_loss": 13.08531665802002, "router_aux_loss": 1.7740143686532974, "router_z_loss": 1.0235832035541534, "routing_entropy_normalized": 0.7755338138938909, "learning_rate": 0.00267, "grad_norm": 0.30120849609375, "tokens_per_second": 3833649.2002301733, "eta_seconds": 15163.894869804382, "data_wait_seconds": 2.1219253540039062e-05} {"step": 900, "total_steps": 28610, "tokens": 1887436800, "source": "fineweb_hq", "phase": 1, "loss": 3.646775335073471, "ce_loss": 3.6263699531555176, "logit_z_loss": 13.379936933517456, "router_aux_loss": 1.7992166578769684, "router_z_loss": 1.0752902254462242, "routing_entropy_normalized": 0.7632551451260177, "learning_rate": 0.0027, "grad_norm": 0.29987722635269165, "tokens_per_second": 4073799.3752869717, "eta_seconds": 14264.836474895477, "data_wait_seconds": 1.9311904907226562e-05} {"step": 910, "total_steps": 28610, "tokens": 1908408320, "source": "fineweb_hq", "phase": 1, "loss": 3.6034410297870636, "ce_loss": 3.583277016878128, "logit_z_loss": 13.17744266986847, "router_aux_loss": 1.752038612961769, "router_z_loss": 1.3258824571967125, "routing_entropy_normalized": 0.7488830156609491, "learning_rate": 0.00273, "grad_norm": 0.3001411259174347, "tokens_per_second": 4210439.636479579, "eta_seconds": 13796.922748088838, "data_wait_seconds": 2.22623348236084e-05} {"step": 920, "total_steps": 28610, "tokens": 1929379840, "source": "fineweb_hq", "phase": 1, "loss": 3.5841883718967438, "ce_loss": 3.5642730593681335, "logit_z_loss": 13.444478392601013, "router_aux_loss": 1.7360568344593048, "router_z_loss": 1.2103381156921387, "routing_entropy_normalized": 0.7559414349105724, "learning_rate": 0.0027600000000000003, "grad_norm": 0.34887921810150146, "tokens_per_second": 3590792.686947137, "eta_seconds": 16171.955315351486, "data_wait_seconds": 1.8358230590820312e-05} {"step": 930, "total_steps": 28610, "tokens": 1950351360, "source": "fineweb_hq", "phase": 1, "loss": 3.5708978474140167, "ce_loss": 3.5510771572589874, "logit_z_loss": 13.872437477111816, "router_aux_loss": 1.7491921335458755, "router_z_loss": 0.9415555894374847, "routing_entropy_normalized": 0.7723916339273227, "learning_rate": 0.00279, "grad_norm": 0.304644376039505, "tokens_per_second": 3686253.455075078, "eta_seconds": 15747.47045135498, "data_wait_seconds": 2.276897430419922e-05} {"step": 940, "total_steps": 28610, "tokens": 1971322880, "source": "fineweb_hq", "phase": 1, "loss": 3.600455731153488, "ce_loss": 3.580485314130783, "logit_z_loss": 13.248793244361877, "router_aux_loss": 1.7551615685224533, "router_z_loss": 1.0939159542322159, "routing_entropy_normalized": 0.7618970862816926, "learning_rate": 0.00282, "grad_norm": 0.33136674761772156, "tokens_per_second": 3656848.4852272323, "eta_seconds": 15868.362081289291, "data_wait_seconds": 2.0235776901245117e-05} {"step": 950, "total_steps": 28610, "tokens": 1992294400, "source": "fineweb_edu", "phase": 1, "loss": 3.9880300164222717, "ce_loss": 3.968054324388504, "logit_z_loss": 13.120636105537415, "router_aux_loss": 1.7475409358739853, "router_z_loss": 1.188227191567421, "routing_entropy_normalized": 0.7437454133702321, "learning_rate": 0.00285, "grad_norm": 0.32099828124046326, "tokens_per_second": 3152982.001634185, "eta_seconds": 18397.575466632843, "data_wait_seconds": 1.850724220275879e-05} {"step": 960, "total_steps": 28610, "tokens": 2013265920, "source": "fineweb_hq", "phase": 1, "loss": 3.579824149608612, "ce_loss": 3.559646785259247, "logit_z_loss": 13.07980501651764, "router_aux_loss": 1.7664236277341843, "router_z_loss": 1.2051368951797485, "routing_entropy_normalized": 0.7310786296562841, "learning_rate": 0.0028799999999999997, "grad_norm": 0.35547634959220886, "tokens_per_second": 3797136.7155960863, "eta_seconds": 15271.04688167572, "data_wait_seconds": 2.0682811737060547e-05} {"step": 970, "total_steps": 28610, "tokens": 2034237440, "source": "fineweb_edu", "phase": 1, "loss": 3.983655869960785, "ce_loss": 3.963058203458786, "logit_z_loss": 13.092215418815613, "router_aux_loss": 1.8066293448209763, "router_z_loss": 1.2221563532948494, "routing_entropy_normalized": 0.7500693587069646, "learning_rate": 0.0029100000000000003, "grad_norm": 0.36281490325927734, "tokens_per_second": 3484643.5538551393, "eta_seconds": 16634.493710517883, "data_wait_seconds": 1.96993350982666e-05} {"step": 980, "total_steps": 28610, "tokens": 2055208960, "source": "fineweb_hq", "phase": 1, "loss": 3.5286049246788025, "ce_loss": 3.5081448554992676, "logit_z_loss": 13.278219103813171, "router_aux_loss": 1.816750630736351, "router_z_loss": 0.9648462235927582, "routing_entropy_normalized": 0.7327255736342961, "learning_rate": 0.00294, "grad_norm": 0.2528068423271179, "tokens_per_second": 3734904.9811785454, "eta_seconds": 15514.266106367111, "data_wait_seconds": 0.09053659439086914} {"step": 990, "total_steps": 28610, "tokens": 2076180480, "source": "fineweb_edu", "phase": 1, "loss": 3.9431776106357574, "ce_loss": 3.923240840435028, "logit_z_loss": 13.035893082618713, "router_aux_loss": 1.7431392669677734, "router_z_loss": 1.2018379122018814, "routing_entropy_normalized": 0.7352828859832841, "learning_rate": 0.00297, "grad_norm": 0.3029248118400574, "tokens_per_second": 4266480.414587309, "eta_seconds": 13576.37504720688, "data_wait_seconds": 2.0176172256469727e-05} {"step": 1000, "total_steps": 28610, "tokens": 2097152000, "source": "fineweb_hq", "phase": 1, "loss": 3.498715817928314, "ce_loss": 3.4785490930080414, "logit_z_loss": 13.087662696838379, "router_aux_loss": 1.7868202030658722, "router_z_loss": 0.9897365272045135, "routing_entropy_normalized": 0.7360057230653183, "learning_rate": 0.003, "grad_norm": 0.29909613728523254, "tokens_per_second": 3728867.9674247233, "eta_seconds": 15528.135408878326, "data_wait_seconds": 1.8298625946044922e-05} {"step": 1010, "total_steps": 28610, "tokens": 2118123520, "source": "fineweb_hq", "phase": 1, "loss": 3.5159562528133392, "ce_loss": 3.495896190404892, "logit_z_loss": 13.587857961654663, "router_aux_loss": 1.7780744135379791, "router_z_loss": 0.9205065071582794, "routing_entropy_normalized": 0.7503177885476127, "learning_rate": 0.003, "grad_norm": 0.26984965801239014, "tokens_per_second": 3872858.2704762225, "eta_seconds": 14945.394630432129, "data_wait_seconds": 1.9490718841552734e-05} {"step": 1020, "total_steps": 28610, "tokens": 2139095040, "source": "fineweb_hq", "phase": 1, "loss": 3.5288613736629486, "ce_loss": 3.508589416742325, "logit_z_loss": 13.038822412490845, "router_aux_loss": 1.795256569981575, "router_z_loss": 1.0154938399791718, "routing_entropy_normalized": 0.7252003158749227, "learning_rate": 0.003, "grad_norm": 0.3407430648803711, "tokens_per_second": 2960840.179740891, "eta_seconds": 19541.89357328415, "data_wait_seconds": 0.4497779607772827} {"step": 1030, "total_steps": 28610, "tokens": 2160066560, "source": "fineweb_hq", "phase": 1, "loss": 3.626240164041519, "ce_loss": 3.606031209230423, "logit_z_loss": 13.305169343948364, "router_aux_loss": 1.7828250080347061, "router_z_loss": 1.0502191632986069, "routing_entropy_normalized": 0.7262705535443134, "learning_rate": 0.003, "grad_norm": 0.262196809053421, "tokens_per_second": 4214954.372598984, "eta_seconds": 13722.438500404358, "data_wait_seconds": 2.104043960571289e-05} {"step": 1040, "total_steps": 28610, "tokens": 2181038080, "source": "fineweb_hq", "phase": 1, "loss": 3.7123197615146637, "ce_loss": 3.6929733753204346, "logit_z_loss": 13.519355535507202, "router_aux_loss": 1.708261102437973, "router_z_loss": 0.9118833243846893, "routing_entropy_normalized": 0.7511877655474616, "learning_rate": 0.003, "grad_norm": 0.27905601263046265, "tokens_per_second": 3762634.7636347334, "eta_seconds": 15366.487653493881, "data_wait_seconds": 2.065300941467285e-05} {"step": 1050, "total_steps": 28610, "tokens": 2202009600, "source": "fineweb_edu", "phase": 1, "loss": 3.8573088943958282, "ce_loss": 3.837806612253189, "logit_z_loss": 13.228194832801819, "router_aux_loss": 1.7022795379161835, "router_z_loss": 1.156775839626789, "routing_entropy_normalized": 0.7381396731319702, "learning_rate": 0.003, "grad_norm": 0.23843508958816528, "tokens_per_second": 4272658.925934075, "eta_seconds": 13527.292985916138, "data_wait_seconds": 1.9252300262451172e-05} {"step": 1060, "total_steps": 28610, "tokens": 2222981120, "source": "fineweb_hq", "phase": 1, "loss": 3.7506060004234314, "ce_loss": 3.7317832708358765, "logit_z_loss": 13.307474493980408, "router_aux_loss": 1.6551974415779114, "router_z_loss": 0.9399715140461922, "routing_entropy_normalized": 0.7493790836464712, "learning_rate": 0.003, "grad_norm": 0.2862909138202667, "tokens_per_second": 3537260.1018081536, "eta_seconds": 16333.697816133501, "data_wait_seconds": 2.321600914001465e-05} {"step": 1070, "total_steps": 28610, "tokens": 2243952640, "source": "fineweb_edu", "phase": 1, "loss": 3.8805067241191864, "ce_loss": 3.862116038799286, "logit_z_loss": 13.276132106781006, "router_aux_loss": 1.6010790765285492, "router_z_loss": 1.0522751286625862, "routing_entropy_normalized": 0.7645256790837567, "learning_rate": 0.003, "grad_norm": 0.3348839282989502, "tokens_per_second": 3734474.1482988326, "eta_seconds": 15465.51503276825, "data_wait_seconds": 2.0951032638549805e-05} {"step": 1080, "total_steps": 28610, "tokens": 2264924160, "source": "fineweb_hq", "phase": 1, "loss": 3.8067366182804108, "ce_loss": 3.788293570280075, "logit_z_loss": 12.920913696289062, "router_aux_loss": 1.625083103775978, "router_z_loss": 0.9000958949327469, "routing_entropy_normalized": 0.7309660063137617, "learning_rate": 0.003, "grad_norm": 0.3295542299747467, "tokens_per_second": 3251460.6826405735, "eta_seconds": 17756.510133504868, "data_wait_seconds": 1.9550323486328125e-05} {"step": 1090, "total_steps": 28610, "tokens": 2285895680, "source": "fineweb_edu", "phase": 1, "loss": 3.832129031419754, "ce_loss": 3.8134789168834686, "logit_z_loss": 13.14590072631836, "router_aux_loss": 1.641230508685112, "router_z_loss": 0.9232481569051743, "routing_entropy_normalized": 0.7498681874528876, "learning_rate": 0.003, "grad_norm": 0.3055610954761505, "tokens_per_second": 3810904.7922955374, "eta_seconds": 15144.336105346682, "data_wait_seconds": 2.79843807220459e-05} {"step": 1100, "total_steps": 28610, "tokens": 2306867200, "source": "fineweb_hq", "phase": 1, "loss": 3.7737776339054108, "ce_loss": 3.754968374967575, "logit_z_loss": 13.320119023323059, "router_aux_loss": 1.659154325723648, "router_z_loss": 0.8857016563415527, "routing_entropy_normalized": 0.7382274051743517, "learning_rate": 0.003, "grad_norm": 0.2607852816581726, "tokens_per_second": 4280932.609609648, "eta_seconds": 13476.65492105484, "data_wait_seconds": 2.7745962142944336e-05} {"step": 1110, "total_steps": 28610, "tokens": 2327838720, "source": "fineweb_hq", "phase": 1, "loss": 3.7609165608882904, "ce_loss": 3.74206081032753, "logit_z_loss": 13.571325421333313, "router_aux_loss": 1.6642631590366364, "router_z_loss": 0.8560161367058754, "routing_entropy_normalized": 0.7237807828903218, "learning_rate": 0.003, "grad_norm": 0.3153557777404785, "tokens_per_second": 3570643.443849993, "eta_seconds": 16151.621103286743, "data_wait_seconds": 2.4199485778808594e-05} {"step": 1120, "total_steps": 28610, "tokens": 2348810240, "source": "fineweb_hq", "phase": 1, "loss": 3.7744768261909485, "ce_loss": 3.755663573741913, "logit_z_loss": 13.296347856521606, "router_aux_loss": 1.6524961739778519, "router_z_loss": 0.9586733728647232, "routing_entropy_normalized": 0.7247650319640757, "learning_rate": 0.003, "grad_norm": 0.3128862679004669, "tokens_per_second": 3500674.0884237206, "eta_seconds": 16468.459223508835, "data_wait_seconds": 2.1189451217651367e-05} {"step": 1130, "total_steps": 28610, "tokens": 2369781760, "source": "fineweb_hq", "phase": 1, "loss": 3.699469178915024, "ce_loss": 3.680673360824585, "logit_z_loss": 13.578590869903564, "router_aux_loss": 1.643067941069603, "router_z_loss": 1.0072607696056366, "routing_entropy_normalized": 0.753703675455834, "learning_rate": 0.003, "grad_norm": 0.22913803160190582, "tokens_per_second": 3757552.359955014, "eta_seconds": 15337.041626930237, "data_wait_seconds": 1.895427703857422e-05} {"step": 1140, "total_steps": 28610, "tokens": 2390753280, "source": "fineweb_hq", "phase": 1, "loss": 3.798104554414749, "ce_loss": 3.779301404953003, "logit_z_loss": 13.141302943229675, "router_aux_loss": 1.6656715273857117, "router_z_loss": 0.8323302268981934, "routing_entropy_normalized": 0.7354007526299524, "learning_rate": 0.003, "grad_norm": 0.2660737931728363, "tokens_per_second": 4087355.0079852673, "eta_seconds": 14094.387526273727, "data_wait_seconds": 2.2679567337036133e-05} {"step": 1150, "total_steps": 28610, "tokens": 2411724800, "source": "fineweb_edu", "phase": 1, "loss": 3.788751929998398, "ce_loss": 3.7704198360443115, "logit_z_loss": 13.02107846736908, "router_aux_loss": 1.6133394688367844, "router_z_loss": 0.8965314701199532, "routing_entropy_normalized": 0.7422723447465854, "learning_rate": 0.003, "grad_norm": 0.2548324167728424, "tokens_per_second": 3373163.554943924, "eta_seconds": 17072.33965444565, "data_wait_seconds": 2.276897430419922e-05} {"step": 1160, "total_steps": 28610, "tokens": 2432696320, "source": "fineweb_hq", "phase": 1, "loss": 3.7746009826660156, "ce_loss": 3.7559984624385834, "logit_z_loss": 13.245291948318481, "router_aux_loss": 1.634489506483078, "router_z_loss": 0.9330129846930504, "routing_entropy_normalized": 0.7614914118459433, "learning_rate": 0.003, "grad_norm": 0.27654552459716797, "tokens_per_second": 3432249.176082783, "eta_seconds": 16772.331916093826, "data_wait_seconds": 2.7894973754882812e-05} {"step": 1170, "total_steps": 28610, "tokens": 2453667840, "source": "fineweb_edu", "phase": 1, "loss": 3.8126309514045715, "ce_loss": 3.794321656227112, "logit_z_loss": 12.75848662853241, "router_aux_loss": 1.5911732763051987, "router_z_loss": 1.121643841266632, "routing_entropy_normalized": 0.7567621164987329, "learning_rate": 0.003, "grad_norm": 0.2750265300273895, "tokens_per_second": 4038915.0687995, "eta_seconds": 14247.84871673584, "data_wait_seconds": 2.422928810119629e-05} {"step": 1180, "total_steps": 28610, "tokens": 2474639360, "source": "fineweb_hq", "phase": 1, "loss": 3.6882618963718414, "ce_loss": 3.6700197756290436, "logit_z_loss": 13.273204565048218, "router_aux_loss": 1.6002838760614395, "router_z_loss": 0.9119101241230965, "routing_entropy_normalized": 0.7291675344600705, "learning_rate": 0.003, "grad_norm": 0.21844597160816193, "tokens_per_second": 4080201.0902816583, "eta_seconds": 14098.540264844894, "data_wait_seconds": 2.512335777282715e-05} {"step": 1190, "total_steps": 28610, "tokens": 2495610880, "source": "fineweb_edu", "phase": 1, "loss": 3.789828360080719, "ce_loss": 3.7716000080108643, "logit_z_loss": 13.224592685699463, "router_aux_loss": 1.6064669042825699, "router_z_loss": 0.84122733771801, "routing_entropy_normalized": 0.7380769266641601, "learning_rate": 0.003, "grad_norm": 0.2946833670139313, "tokens_per_second": 3118568.051447745, "eta_seconds": 18439.20250940323, "data_wait_seconds": 1.96993350982666e-05} {"step": 1200, "total_steps": 28610, "tokens": 2516582400, "source": "fineweb_hq", "phase": 1, "loss": 3.702523797750473, "ce_loss": 3.684811234474182, "logit_z_loss": 13.292378902435303, "router_aux_loss": 1.5608905255794525, "router_z_loss": 0.7743784189224243, "routing_entropy_normalized": 0.7559652148516611, "learning_rate": 0.003, "grad_norm": 0.2550068497657776, "tokens_per_second": 3528846.1758712437, "eta_seconds": 16289.442343235016, "data_wait_seconds": 2.0772218704223633e-05} {"step": 1210, "total_steps": 28610, "tokens": 2537553920, "source": "fineweb_hq", "phase": 1, "loss": 3.689670890569687, "ce_loss": 3.6719910800457, "logit_z_loss": 13.187568545341492, "router_aux_loss": 1.5623154789209366, "router_z_loss": 0.7378690391778946, "routing_entropy_normalized": 0.7560030415816092, "learning_rate": 0.003, "grad_norm": 0.21758417785167694, "tokens_per_second": 2970520.135819822, "eta_seconds": 19344.075169563293, "data_wait_seconds": 2.35140323638916e-05} {"step": 1220, "total_steps": 28610, "tokens": 2558525440, "source": "fineweb_hq", "phase": 1, "loss": 3.710075467824936, "ce_loss": 3.691920965909958, "logit_z_loss": 12.754634499549866, "router_aux_loss": 1.6133923679590225, "router_z_loss": 0.7451487109065056, "routing_entropy_normalized": 0.7189863490575994, "learning_rate": 0.003, "grad_norm": 0.31864070892333984, "tokens_per_second": 3566225.630880171, "eta_seconds": 16106.94308924675, "data_wait_seconds": 2.1696090698242188e-05} {"step": 1230, "total_steps": 28610, "tokens": 2579496960, "source": "fineweb_hq", "phase": 1, "loss": 3.6679110527038574, "ce_loss": 3.649574488401413, "logit_z_loss": 12.893252611160278, "router_aux_loss": 1.6213468313217163, "router_z_loss": 0.8337559998035431, "routing_entropy_normalized": 0.7327050150597194, "learning_rate": 0.003, "grad_norm": 0.2422039955854416, "tokens_per_second": 3972809.317470809, "eta_seconds": 14453.25389957428, "data_wait_seconds": 2.136826515197754e-05} {"step": 1240, "total_steps": 28610, "tokens": 2600468480, "source": "fineweb_hq", "phase": 1, "loss": 3.7075601518154144, "ce_loss": 3.6893202662467957, "logit_z_loss": 13.275237917900085, "router_aux_loss": 1.6119788140058517, "router_z_loss": 0.7926401719450951, "routing_entropy_normalized": 0.7354555276925282, "learning_rate": 0.003, "grad_norm": 0.292062371969223, "tokens_per_second": 3978261.0064432584, "eta_seconds": 14428.17606663704, "data_wait_seconds": 1.823902130126953e-05} {"step": 1250, "total_steps": 28610, "tokens": 2621440000, "source": "fineweb_edu", "phase": 1, "loss": 3.7909673154354095, "ce_loss": 3.7727760672569275, "logit_z_loss": 12.625570058822632, "router_aux_loss": 1.579139232635498, "router_z_loss": 1.1373146176338196, "routing_entropy_normalized": 0.7273836380911421, "learning_rate": 0.003, "grad_norm": 0.2440190315246582, "tokens_per_second": 3614281.744174097, "eta_seconds": 15875.375186920166, "data_wait_seconds": 1.9103288650512695e-05} {"step": 1260, "total_steps": 28610, "tokens": 2642411520, "source": "fineweb_hq", "phase": 1, "loss": 3.6652913987636566, "ce_loss": 3.6465583443641663, "logit_z_loss": 12.954219818115234, "router_aux_loss": 1.6404407620429993, "router_z_loss": 1.0331949144601822, "routing_entropy_normalized": 0.7317011982249283, "learning_rate": 0.003, "grad_norm": 0.23806901276111603, "tokens_per_second": 4298708.305674416, "eta_seconds": 13342.87025809288, "data_wait_seconds": 1.9848346710205078e-05} {"step": 1270, "total_steps": 28610, "tokens": 2663383040, "source": "fineweb_edu", "phase": 1, "loss": 3.7676600217819214, "ce_loss": 3.7494727969169617, "logit_z_loss": 12.923764109611511, "router_aux_loss": 1.6103628277778625, "router_z_loss": 0.7911927476525307, "routing_entropy_normalized": 0.7305975573147367, "learning_rate": 0.003, "grad_norm": 0.3513587713241577, "tokens_per_second": 3645744.7186719715, "eta_seconds": 15726.86518239975, "data_wait_seconds": 2.1576881408691406e-05} {"step": 1280, "total_steps": 28610, "tokens": 2684354560, "source": "fineweb_hq", "phase": 1, "loss": 3.6495718359947205, "ce_loss": 3.6313765347003937, "logit_z_loss": 12.739686965942383, "router_aux_loss": 1.609713077545166, "router_z_loss": 0.8242181017994881, "routing_entropy_normalized": 0.7378334825768961, "learning_rate": 0.003, "grad_norm": 0.2121456265449524, "tokens_per_second": 3907151.9412103337, "eta_seconds": 14669.294929504395, "data_wait_seconds": 2.3484230041503906e-05} {"step": 1290, "total_steps": 28610, "tokens": 2705326080, "source": "fineweb_edu", "phase": 1, "loss": 3.7536254823207855, "ce_loss": 3.73561692237854, "logit_z_loss": 12.611430168151855, "router_aux_loss": 1.593292623758316, "router_z_loss": 0.8144921585917473, "routing_entropy_normalized": 0.719207989798834, "learning_rate": 0.003, "grad_norm": 0.2598116099834442, "tokens_per_second": 3546878.2064695973, "eta_seconds": 16153.414158821106, "data_wait_seconds": 2.5570392608642578e-05} {"step": 1300, "total_steps": 28610, "tokens": 2726297600, "source": "fineweb_hq", "phase": 1, "loss": 3.6559227406978607, "ce_loss": 3.6380514800548553, "logit_z_loss": 12.79792046546936, "router_aux_loss": 1.580781266093254, "router_z_loss": 0.783683642745018, "routing_entropy_normalized": 0.7178066877357714, "learning_rate": 0.003, "grad_norm": 0.22197207808494568, "tokens_per_second": 4441019.81731526, "eta_seconds": 12896.411967515945, "data_wait_seconds": 1.946091651916504e-05} {"step": 1310, "total_steps": 28610, "tokens": 2747269120, "source": "fineweb_hq", "phase": 1, "loss": 3.646061420440674, "ce_loss": 3.628581613302231, "logit_z_loss": 12.883747458457947, "router_aux_loss": 1.5333556532859802, "router_z_loss": 0.8579466566443443, "routing_entropy_normalized": 0.7411227465487313, "learning_rate": 0.003, "grad_norm": 0.19888465106487274, "tokens_per_second": 3926677.7127674557, "eta_seconds": 14580.328152179718, "data_wait_seconds": 1.9222497940063477e-05} {"step": 1320, "total_steps": 28610, "tokens": 2768240640, "source": "fineweb_hq", "phase": 1, "loss": 3.6691146194934845, "ce_loss": 3.6513642966747284, "logit_z_loss": 12.92969274520874, "router_aux_loss": 1.5697881579399109, "router_z_loss": 0.7593527622520924, "routing_entropy_normalized": 0.6882685006291569, "learning_rate": 0.003, "grad_norm": 0.33036792278289795, "tokens_per_second": 3761707.7520159306, "eta_seconds": 15214.174479484558, "data_wait_seconds": 2.276897430419922e-05} {"step": 1330, "total_steps": 28610, "tokens": 2789212160, "source": "fineweb_hq", "phase": 1, "loss": 3.664066404104233, "ce_loss": 3.646458864212036, "logit_z_loss": 12.97669005393982, "router_aux_loss": 1.5600038170814514, "router_z_loss": 0.7098991051316261, "routing_entropy_normalized": 0.7285120291314378, "learning_rate": 0.003, "grad_norm": 0.23489975929260254, "tokens_per_second": 3620163.3650206774, "eta_seconds": 15803.238912582396, "data_wait_seconds": 2.0623207092285156e-05} {"step": 1340, "total_steps": 28610, "tokens": 2810183680, "source": "fineweb_hq", "phase": 1, "loss": 3.6751082241535187, "ce_loss": 3.657240390777588, "logit_z_loss": 13.218847632408142, "router_aux_loss": 1.5853353440761566, "router_z_loss": 0.6925933808088303, "routing_entropy_normalized": 0.7045790690481842, "learning_rate": 0.003, "grad_norm": 0.2663385570049286, "tokens_per_second": 3945755.5953272986, "eta_seconds": 14493.886825561523, "data_wait_seconds": 1.9043684005737305e-05} {"step": 1350, "total_steps": 28610, "tokens": 2831155200, "source": "fineweb_edu", "phase": 1, "loss": 3.7320760786533356, "ce_loss": 3.7144225537776947, "logit_z_loss": 12.475766062736511, "router_aux_loss": 1.572474643588066, "router_z_loss": 0.6812893077731133, "routing_entropy_normalized": 0.7076111298507933, "learning_rate": 0.003, "grad_norm": 0.25115108489990234, "tokens_per_second": 3342246.1572490428, "eta_seconds": 17104.773505687714, "data_wait_seconds": 1.6123056411743164e-05} {"step": 1360, "total_steps": 28610, "tokens": 2852126720, "source": "fineweb_hq", "phase": 1, "loss": 3.6170286536216736, "ce_loss": 3.599080264568329, "logit_z_loss": 12.926063179969788, "router_aux_loss": 1.5920685082674026, "router_z_loss": 0.7351187020540237, "routing_entropy_normalized": 0.7297136746099004, "learning_rate": 0.003, "grad_norm": 0.22807221114635468, "tokens_per_second": 3317722.4333629897, "eta_seconds": 17224.886393547058, "data_wait_seconds": 1.7464160919189453e-05} {"step": 1370, "total_steps": 28610, "tokens": 2873098240, "source": "fineweb_edu", "phase": 1, "loss": 3.7073278725147247, "ce_loss": 3.689581722021103, "logit_z_loss": 12.86902379989624, "router_aux_loss": 1.5667932778596878, "router_z_loss": 0.79135512560606, "routing_entropy_normalized": 0.7026703592149608, "learning_rate": 0.003, "grad_norm": 0.22224768996238708, "tokens_per_second": 3848521.849059559, "eta_seconds": 14843.730325698853, "data_wait_seconds": 1.6868114471435547e-05} {"step": 1380, "total_steps": 28610, "tokens": 2894069760, "source": "fineweb_hq", "phase": 1, "loss": 3.6396936178207397, "ce_loss": 3.62176114320755, "logit_z_loss": 13.100827813148499, "router_aux_loss": 1.582287460565567, "router_z_loss": 0.7995611429214478, "routing_entropy_normalized": 0.7139659143143634, "learning_rate": 0.003, "grad_norm": 0.25207850337028503, "tokens_per_second": 2669021.6950111184, "eta_seconds": 21395.648100852966, "data_wait_seconds": 1.9937753677368164e-05} {"step": 1390, "total_steps": 28610, "tokens": 2915041280, "source": "fineweb_edu", "phase": 1, "loss": 3.712700366973877, "ce_loss": 3.695215582847595, "logit_z_loss": 12.642094135284424, "router_aux_loss": 1.5312270373106003, "router_z_loss": 0.9082884714007378, "routing_entropy_normalized": 0.7383512084120059, "learning_rate": 0.003, "grad_norm": 0.27023693919181824, "tokens_per_second": 3822487.99540611, "eta_seconds": 14933.853947639465, "data_wait_seconds": 2.250075340270996e-05} {"step": 1400, "total_steps": 28610, "tokens": 2936012800, "source": "fineweb_hq", "phase": 1, "loss": 3.676722288131714, "ce_loss": 3.6587376296520233, "logit_z_loss": 12.360433101654053, "router_aux_loss": 1.6013674885034561, "router_z_loss": 0.734929546713829, "routing_entropy_normalized": 0.7145135967738396, "learning_rate": 0.003, "grad_norm": 0.2288474589586258, "tokens_per_second": 3615094.719415147, "eta_seconds": 15784.788601398468, "data_wait_seconds": 2.3037195205688477e-05} {"step": 1410, "total_steps": 28610, "tokens": 2956984320, "source": "fineweb_hq", "phase": 1, "loss": 3.6780615150928497, "ce_loss": 3.6604828238487244, "logit_z_loss": 12.762844920158386, "router_aux_loss": 1.5506181418895721, "router_z_loss": 0.7961695119738579, "routing_entropy_normalized": 0.7289022059997208, "learning_rate": 0.003, "grad_norm": 0.21961157023906708, "tokens_per_second": 4164205.1542467577, "eta_seconds": 13698.300704956055, "data_wait_seconds": 2.1189451217651367e-05} {"step": 1420, "total_steps": 28610, "tokens": 2977955840, "source": "fineweb_hq", "phase": 1, "loss": 3.6734668016433716, "ce_loss": 3.656163662672043, "logit_z_loss": 12.42779541015625, "router_aux_loss": 1.5250843912363052, "router_z_loss": 0.8095820993185043, "routing_entropy_normalized": 0.7271144684186339, "learning_rate": 0.003, "grad_norm": 0.22234903275966644, "tokens_per_second": 3781325.223634907, "eta_seconds": 15079.7827501297, "data_wait_seconds": 1.9341707229614258e-05} {"step": 1430, "total_steps": 28610, "tokens": 2998927360, "source": "fineweb_hq", "phase": 1, "loss": 3.655671715736389, "ce_loss": 3.638202488422394, "logit_z_loss": 12.594285368919373, "router_aux_loss": 1.536058321595192, "router_z_loss": 0.8491563275456429, "routing_entropy_normalized": 0.7155607391324773, "learning_rate": 0.003, "grad_norm": 0.22061379253864288, "tokens_per_second": 3956496.72841663, "eta_seconds": 14406.83394241333, "data_wait_seconds": 1.8537044525146484e-05} {"step": 1440, "total_steps": 28610, "tokens": 3019898880, "source": "fineweb_hq", "phase": 1, "loss": 3.676833003759384, "ce_loss": 3.659650534391403, "logit_z_loss": 12.351799845695496, "router_aux_loss": 1.512140914797783, "router_z_loss": 0.8258443102240562, "routing_entropy_normalized": 0.7080666264974426, "learning_rate": 0.003, "grad_norm": 0.25596216320991516, "tokens_per_second": 4381135.813656155, "eta_seconds": 13005.673018031652, "data_wait_seconds": 1.9788742065429688e-05} {"step": 1450, "total_steps": 28610, "tokens": 3040870400, "source": "fineweb_edu", "phase": 1, "loss": 3.7044654190540314, "ce_loss": 3.686892956495285, "logit_z_loss": 12.466024279594421, "router_aux_loss": 1.5612538009881973, "router_z_loss": 0.7133221700787544, "routing_entropy_normalized": 0.7043639756552537, "learning_rate": 0.003, "grad_norm": 0.27224960923194885, "tokens_per_second": 3556935.7743758727, "eta_seconds": 16013.403652191162, "data_wait_seconds": 1.9431114196777344e-05} {"step": 1460, "total_steps": 28610, "tokens": 3061841920, "source": "fineweb_hq", "phase": 1, "loss": 3.7117844223976135, "ce_loss": 3.6941224336624146, "logit_z_loss": 12.32823634147644, "router_aux_loss": 1.5830183029174805, "router_z_loss": 0.5990208238363266, "routing_entropy_normalized": 0.680746044882493, "learning_rate": 0.003, "grad_norm": 0.2631883919239044, "tokens_per_second": 4030491.1136345454, "eta_seconds": 14126.734235286713, "data_wait_seconds": 1.8835067749023438e-05} {"step": 1470, "total_steps": 28610, "tokens": 3082813440, "source": "fineweb_edu", "phase": 1, "loss": 3.668451339006424, "ce_loss": 3.651496410369873, "logit_z_loss": 12.468879342079163, "router_aux_loss": 1.4899082332849503, "router_z_loss": 0.8089508637785912, "routing_entropy_normalized": 0.7172969755964888, "learning_rate": 0.003, "grad_norm": 0.2670517861843109, "tokens_per_second": 3747733.4344171053, "eta_seconds": 15186.967343330383, "data_wait_seconds": 2.014636993408203e-05} {"step": 1480, "total_steps": 28610, "tokens": 3103784960, "source": "fineweb_hq", "phase": 1, "loss": 3.6629000902175903, "ce_loss": 3.6459271907806396, "logit_z_loss": 12.395896434783936, "router_aux_loss": 1.4944439232349396, "router_z_loss": 0.7888916656374931, "routing_entropy_normalized": 0.7329297472179391, "learning_rate": 0.003, "grad_norm": 0.2461734563112259, "tokens_per_second": 4094648.4182437095, "eta_seconds": 13895.145064592361, "data_wait_seconds": 2.1010637283325195e-05} {"step": 1490, "total_steps": 28610, "tokens": 3124756480, "source": "fineweb_edu", "phase": 1, "loss": 3.6451415419578552, "ce_loss": 3.6282644271850586, "logit_z_loss": 12.359757900238037, "router_aux_loss": 1.4925770312547684, "router_z_loss": 0.7153939493000507, "routing_entropy_normalized": 0.7163209386981384, "learning_rate": 0.003, "grad_norm": 0.2753106653690338, "tokens_per_second": 3956536.4147547022, "eta_seconds": 14374.886587142944, "data_wait_seconds": 2.0295381546020508e-05} {"step": 1500, "total_steps": 28610, "tokens": 3145728000, "source": "fineweb_hq", "phase": 1, "loss": 3.656480848789215, "ce_loss": 3.6395174860954285, "logit_z_loss": 12.56466829776764, "router_aux_loss": 1.5114773511886597, "router_z_loss": 0.5921509526669979, "routing_entropy_normalized": 0.6497351141725081, "learning_rate": 0.003, "grad_norm": 0.2685629427433014, "tokens_per_second": 3494225.9964915896, "eta_seconds": 16270.78236413002, "data_wait_seconds": 2.0116567611694336e-05} {"step": 1510, "total_steps": 28610, "tokens": 3166699520, "source": "fineweb_hq", "phase": 1, "loss": 3.6618571281433105, "ce_loss": 3.6447199881076813, "logit_z_loss": 12.476884484291077, "router_aux_loss": 1.5175634026527405, "router_z_loss": 0.7138332352042198, "routing_entropy_normalized": 0.7044966594393228, "learning_rate": 0.003, "grad_norm": 0.1976461559534073, "tokens_per_second": 3644304.790837927, "eta_seconds": 15594.96871471405, "data_wait_seconds": 1.8715858459472656e-05} {"step": 1520, "total_steps": 28610, "tokens": 3187671040, "source": "fineweb_hq", "phase": 1, "loss": 3.6723801493644714, "ce_loss": 3.6555173695087433, "logit_z_loss": 12.662059783935547, "router_aux_loss": 1.4888929575681686, "router_z_loss": 0.7076924368739128, "routing_entropy_normalized": 0.694200802374759, "learning_rate": 0.003, "grad_norm": 0.2167554795742035, "tokens_per_second": 3825719.787346857, "eta_seconds": 14849.976171255112, "data_wait_seconds": 2.1010637283325195e-05} {"step": 1530, "total_steps": 28610, "tokens": 3208642560, "source": "fineweb_hq", "phase": 1, "loss": 3.652646690607071, "ce_loss": 3.6359968185424805, "logit_z_loss": 12.71045708656311, "router_aux_loss": 1.462521731853485, "router_z_loss": 0.7536074630916119, "routing_entropy_normalized": 0.7177554665404584, "learning_rate": 0.003, "grad_norm": 0.22310253977775574, "tokens_per_second": 3820027.2941485886, "eta_seconds": 14866.615284919739, "data_wait_seconds": 2.0265579223632812e-05} {"step": 1540, "total_steps": 28610, "tokens": 3229614080, "source": "fineweb_hq", "phase": 1, "loss": 3.6832699477672577, "ce_loss": 3.666541248559952, "logit_z_loss": 12.473206043243408, "router_aux_loss": 1.4818926453590393, "router_z_loss": 0.6624890565872192, "routing_entropy_normalized": 0.7061648845849385, "learning_rate": 0.003, "grad_norm": 0.30452093482017517, "tokens_per_second": 4004216.0462972797, "eta_seconds": 14177.532876253128, "data_wait_seconds": 1.8656253814697266e-05} {"step": 1550, "total_steps": 28610, "tokens": 3250585600, "source": "fineweb_edu", "phase": 1, "loss": 3.6597185134887695, "ce_loss": 3.64283487200737, "logit_z_loss": 12.525735139846802, "router_aux_loss": 1.5030031949281693, "router_z_loss": 0.6011311486363411, "routing_entropy_normalized": 0.7190561989744875, "learning_rate": 0.003, "grad_norm": 0.21131084859371185, "tokens_per_second": 3830665.899180518, "eta_seconds": 14814.378129959106, "data_wait_seconds": 1.6957521438598633e-05} {"step": 1560, "total_steps": 28610, "tokens": 3271557120, "source": "fineweb_hq", "phase": 1, "loss": 3.6502020359039307, "ce_loss": 3.633257508277893, "logit_z_loss": 12.233828663825989, "router_aux_loss": 1.49635049700737, "router_z_loss": 0.757681205868721, "routing_entropy_normalized": 0.7204711943796268, "learning_rate": 0.003, "grad_norm": 0.2700507342815399, "tokens_per_second": 3586437.205968264, "eta_seconds": 15817.358102798462, "data_wait_seconds": 2.2739171981811523e-05} {"step": 1570, "total_steps": 28610, "tokens": 3292528640, "source": "fineweb_edu", "phase": 1, "loss": 3.65023609995842, "ce_loss": 3.633481651544571, "logit_z_loss": 12.708350777626038, "router_aux_loss": 1.4789348542690277, "router_z_loss": 0.6942699775099754, "routing_entropy_normalized": 0.7002144277697513, "learning_rate": 0.003, "grad_norm": 0.2688738703727722, "tokens_per_second": 4309416.466406014, "eta_seconds": 13158.855850219727, "data_wait_seconds": 1.9609928131103516e-05} {"step": 1580, "total_steps": 28610, "tokens": 3313500160, "source": "fineweb_hq", "phase": 1, "loss": 3.6377318501472473, "ce_loss": 3.620876431465149, "logit_z_loss": 12.191120982170105, "router_aux_loss": 1.4964589476585388, "router_z_loss": 0.671729177236557, "routing_entropy_normalized": 0.6719414392016624, "learning_rate": 0.003, "grad_norm": 0.2500215470790863, "tokens_per_second": 3777433.6209629374, "eta_seconds": 15006.489656209946, "data_wait_seconds": 1.9490718841552734e-05} {"step": 1590, "total_steps": 28610, "tokens": 3334471680, "source": "fineweb_edu", "phase": 1, "loss": 3.6418247520923615, "ce_loss": 3.6251537203788757, "logit_z_loss": 12.433060646057129, "router_aux_loss": 1.4810241013765335, "router_z_loss": 0.6174940131604671, "routing_entropy_normalized": 0.720999079783002, "learning_rate": 0.003, "grad_norm": 0.20966090261936188, "tokens_per_second": 3555415.6650491096, "eta_seconds": 15937.671535015106, "data_wait_seconds": 2.0176172256469727e-05} {"step": 1600, "total_steps": 28610, "tokens": 3355443200, "source": "fineweb_hq", "phase": 1, "loss": 3.6489916145801544, "ce_loss": 3.6322017908096313, "logit_z_loss": 12.510599732398987, "router_aux_loss": 1.4895434081554413, "router_z_loss": 0.6433986648917198, "routing_entropy_normalized": 0.6820228694498189, "learning_rate": 0.003, "grad_norm": 0.2438979148864746, "tokens_per_second": 3563630.3132444913, "eta_seconds": 15895.04817867279, "data_wait_seconds": 1.8864870071411133e-05} {"step": 1610, "total_steps": 28610, "tokens": 3376414720, "source": "fineweb_hq", "phase": 1, "loss": 3.618709146976471, "ce_loss": 3.602090895175934, "logit_z_loss": 12.555930495262146, "router_aux_loss": 1.4693088084459305, "router_z_loss": 0.6695489063858986, "routing_entropy_normalized": 0.686192942219632, "learning_rate": 0.003, "grad_norm": 0.18656398355960846, "tokens_per_second": 3753522.259849126, "eta_seconds": 15085.32521724701, "data_wait_seconds": 2.1338462829589844e-05} {"step": 1620, "total_steps": 28610, "tokens": 3397386240, "source": "fineweb_hq", "phase": 1, "loss": 3.6275323033332825, "ce_loss": 3.611005336046219, "logit_z_loss": 12.500401496887207, "router_aux_loss": 1.4652821123600006, "router_z_loss": 0.6241012662649155, "routing_entropy_normalized": 0.69491571809821, "learning_rate": 0.003, "grad_norm": 0.2980342507362366, "tokens_per_second": 3858892.8122730795, "eta_seconds": 14667.97219657898, "data_wait_seconds": 0.09581482410430908} {"step": 1630, "total_steps": 28610, "tokens": 3418357760, "source": "fineweb_hq", "phase": 1, "loss": 3.624773681163788, "ce_loss": 3.607807159423828, "logit_z_loss": 12.502005815505981, "router_aux_loss": 1.5041459649801254, "router_z_loss": 0.674872636795044, "routing_entropy_normalized": 0.6906678243150294, "learning_rate": 0.003, "grad_norm": 0.19841700792312622, "tokens_per_second": 3786743.1048330404, "eta_seconds": 14941.906380653381, "data_wait_seconds": 2.130866050720215e-05} {"step": 1640, "total_steps": 28610, "tokens": 3439329280, "source": "fineweb_hq", "phase": 1, "loss": 3.636306196451187, "ce_loss": 3.6196194291114807, "logit_z_loss": 12.114117980003357, "router_aux_loss": 1.4838497936725616, "router_z_loss": 0.6368807591497898, "routing_entropy_normalized": 0.7137629469307499, "learning_rate": 0.003, "grad_norm": 0.28983446955680847, "tokens_per_second": 3790945.864696729, "eta_seconds": 14919.809319019318, "data_wait_seconds": 1.7017126083374023e-05} {"step": 1650, "total_steps": 28610, "tokens": 3460300800, "source": "fineweb_edu", "phase": 1, "loss": 3.626561462879181, "ce_loss": 3.6101601123809814, "logit_z_loss": 12.305955410003662, "router_aux_loss": 1.453803226351738, "router_z_loss": 0.632672592997551, "routing_entropy_normalized": 0.7197676673288931, "learning_rate": 0.003, "grad_norm": 0.2777597904205322, "tokens_per_second": 3778801.3044751557, "eta_seconds": 14962.209802627563, "data_wait_seconds": 1.6927719116210938e-05} {"step": 1660, "total_steps": 28610, "tokens": 3481272320, "source": "fineweb_hq", "phase": 1, "loss": 3.631686121225357, "ce_loss": 3.6151501834392548, "logit_z_loss": 12.180391669273376, "router_aux_loss": 1.461466759443283, "router_z_loss": 0.7031823359429836, "routing_entropy_normalized": 0.7134823006111689, "learning_rate": 0.003, "grad_norm": 0.27170106768608093, "tokens_per_second": 3698838.2811113438, "eta_seconds": 15279.999314546585, "data_wait_seconds": 1.919269561767578e-05} {"step": 1670, "total_steps": 28610, "tokens": 3502243840, "source": "fineweb_edu", "phase": 1, "loss": 3.5985195338726044, "ce_loss": 3.58198881149292, "logit_z_loss": 12.42026674747467, "router_aux_loss": 1.462083861231804, "router_z_loss": 0.6678919047117233, "routing_entropy_normalized": 0.6995279430003036, "learning_rate": 0.003, "grad_norm": 0.25528696179389954, "tokens_per_second": 4210258.659745629, "eta_seconds": 13418.95580434799, "data_wait_seconds": 2.0205974578857422e-05} {"step": 1680, "total_steps": 28610, "tokens": 3523215360, "source": "fineweb_hq", "phase": 1, "loss": 3.578804522752762, "ce_loss": 3.5624783635139465, "logit_z_loss": 12.142577171325684, "router_aux_loss": 1.4506138861179352, "router_z_loss": 0.6057632341980934, "routing_entropy_normalized": 0.695517084075763, "learning_rate": 0.003, "grad_norm": 0.2441602349281311, "tokens_per_second": 3417500.3783097165, "eta_seconds": 16525.617295742035, "data_wait_seconds": 2.1666288375854492e-05} {"step": 1690, "total_steps": 28610, "tokens": 3544186880, "source": "fineweb_edu", "phase": 1, "loss": 3.5938689410686493, "ce_loss": 3.5773581862449646, "logit_z_loss": 12.428719758987427, "router_aux_loss": 1.4602499157190323, "router_z_loss": 0.6653629504144192, "routing_entropy_normalized": 0.7119411637244366, "learning_rate": 0.003, "grad_norm": 0.24137301743030548, "tokens_per_second": 4420163.684130283, "eta_seconds": 12772.226522445677, "data_wait_seconds": 1.9937753677368164e-05} {"step": 1700, "total_steps": 28610, "tokens": 3565158400, "source": "fineweb_hq", "phase": 1, "loss": 3.628164142370224, "ce_loss": 3.611364334821701, "logit_z_loss": 12.065343260765076, "router_aux_loss": 1.4996590614318848, "router_z_loss": 0.5966682992875576, "routing_entropy_normalized": 0.7090963991186432, "learning_rate": 0.003, "grad_norm": 0.30031460523605347, "tokens_per_second": 4135848.5989834084, "eta_seconds": 13645.170747756958, "data_wait_seconds": 1.8358230590820312e-05} {"step": 1710, "total_steps": 28610, "tokens": 3586129920, "source": "fineweb_hq", "phase": 1, "loss": 3.6278930008411407, "ce_loss": 3.6112649142742157, "logit_z_loss": 12.283703088760376, "router_aux_loss": 1.4812349081039429, "router_z_loss": 0.5873666554689407, "routing_entropy_normalized": 0.7103361498218448, "learning_rate": 0.003, "grad_norm": 0.20611809194087982, "tokens_per_second": 4141896.512828419, "eta_seconds": 13620.183079242706, "data_wait_seconds": 2.136826515197754e-05} {"step": 1720, "total_steps": 28610, "tokens": 3607101440, "source": "fineweb_hq", "phase": 1, "loss": 3.619033068418503, "ce_loss": 3.6023391485214233, "logit_z_loss": 12.45270848274231, "router_aux_loss": 1.4849872589111328, "router_z_loss": 0.5987618416547775, "routing_entropy_normalized": 0.7037328204154033, "learning_rate": 0.003, "grad_norm": 0.277946799993515, "tokens_per_second": 3624896.6537101474, "eta_seconds": 15556.972423553467, "data_wait_seconds": 2.232193946838379e-05} {"step": 1730, "total_steps": 28610, "tokens": 3628072960, "source": "fineweb_hq", "phase": 1, "loss": 3.5947479605674744, "ce_loss": 3.5782146751880646, "logit_z_loss": 12.106429696083069, "router_aux_loss": 1.4570877403020859, "router_z_loss": 0.7517355903983116, "routing_entropy_normalized": 0.7156032546786428, "learning_rate": 0.003, "grad_norm": 0.20653215050697327, "tokens_per_second": 4003433.2956181983, "eta_seconds": 14080.775573730469, "data_wait_seconds": 2.193450927734375e-05} {"step": 1740, "total_steps": 28610, "tokens": 3649044480, "source": "fineweb_hq", "phase": 1, "loss": 3.606970965862274, "ce_loss": 3.5902731120586395, "logit_z_loss": 12.477310419082642, "router_aux_loss": 1.4729377031326294, "router_z_loss": 0.7207480892539024, "routing_entropy_normalized": 0.7196270452455942, "learning_rate": 0.003, "grad_norm": 0.2905227243900299, "tokens_per_second": 3396225.1342203245, "eta_seconds": 16592.090339422226, "data_wait_seconds": 2.0056962966918945e-05} {"step": 1750, "total_steps": 28610, "tokens": 3670016000, "source": "fineweb_edu", "phase": 1, "loss": 3.644161105155945, "ce_loss": 3.6277432441711426, "logit_z_loss": 12.278776049613953, "router_aux_loss": 1.4514971524477005, "router_z_loss": 0.6750352829694748, "routing_entropy_normalized": 0.689100278669023, "learning_rate": 0.003, "grad_norm": 0.2911907732486725, "tokens_per_second": 4199953.1985237105, "eta_seconds": 13411.935813903807, "data_wait_seconds": 2.0623207092285156e-05} {"step": 1760, "total_steps": 28610, "tokens": 3690987520, "source": "fineweb_hq", "phase": 1, "loss": 3.6054670810699463, "ce_loss": 3.58882474899292, "logit_z_loss": 12.142853379249573, "router_aux_loss": 1.4810553789138794, "router_z_loss": 0.617483738809824, "routing_entropy_normalized": 0.7136351308970613, "learning_rate": 0.003, "grad_norm": 0.29872822761535645, "tokens_per_second": 3280221.978063319, "eta_seconds": 17166.07338666916, "data_wait_seconds": 1.5050172805786133e-05} {"step": 1770, "total_steps": 28610, "tokens": 3711959040, "source": "fineweb_edu", "phase": 1, "loss": 3.599575161933899, "ce_loss": 3.5827790796756744, "logit_z_loss": 12.396862626075745, "router_aux_loss": 1.4705254286527634, "router_z_loss": 0.8511205241084099, "routing_entropy_normalized": 0.6937866851670875, "learning_rate": 0.003, "grad_norm": 0.27386918663978577, "tokens_per_second": 3835100.372840539, "eta_seconds": 14676.945635795593, "data_wait_seconds": 1.7195940017700195e-05} {"step": 1780, "total_steps": 28610, "tokens": 3732930560, "source": "fineweb_hq", "phase": 1, "loss": 3.636252135038376, "ce_loss": 3.619865983724594, "logit_z_loss": 12.248326778411865, "router_aux_loss": 1.457844004034996, "router_z_loss": 0.5828235521912575, "routing_entropy_normalized": 0.7096890251364436, "learning_rate": 0.003, "grad_norm": 0.23842425644397736, "tokens_per_second": 2827410.2315449626, "eta_seconds": 19900.397732257843, "data_wait_seconds": 0.21366864442825317} {"step": 1790, "total_steps": 28610, "tokens": 3753902080, "source": "fineweb_edu", "phase": 1, "loss": 3.6067862808704376, "ce_loss": 3.5900691151618958, "logit_z_loss": 12.04918622970581, "router_aux_loss": 1.4805530458688736, "router_z_loss": 0.7067373916506767, "routing_entropy_normalized": 0.711847703642876, "learning_rate": 0.003, "grad_norm": 0.28109583258628845, "tokens_per_second": 3252984.3197932313, "eta_seconds": 17290.46657180786, "data_wait_seconds": 1.8090009689331055e-05} {"step": 1800, "total_steps": 28610, "tokens": 3774873600, "source": "fineweb_hq", "phase": 1, "loss": 3.5831931829452515, "ce_loss": 3.5668696463108063, "logit_z_loss": 12.268455743789673, "router_aux_loss": 1.4514171779155731, "router_z_loss": 0.5825723856687546, "routing_entropy_normalized": 0.7160358562250679, "learning_rate": 0.003, "grad_norm": 0.21989837288856506, "tokens_per_second": 2952595.936520396, "eta_seconds": 19042.44479393959, "data_wait_seconds": 2.485513687133789e-05} {"step": 1810, "total_steps": 28610, "tokens": 3795845120, "source": "fineweb_hq", "phase": 1, "loss": 3.5550330877304077, "ce_loss": 3.5389679968357086, "logit_z_loss": 12.421856045722961, "router_aux_loss": 1.4192940443754196, "router_z_loss": 0.6299586445093155, "routing_entropy_normalized": 0.723837455641412, "learning_rate": 0.003, "grad_norm": 0.2256787121295929, "tokens_per_second": 3838598.930662134, "eta_seconds": 14641.715536117554, "data_wait_seconds": 2.2470951080322266e-05} {"step": 1820, "total_steps": 28610, "tokens": 3816816640, "source": "fineweb_hq", "phase": 1, "loss": 3.5425831079483032, "ce_loss": 3.5266091525554657, "logit_z_loss": 12.085304737091064, "router_aux_loss": 1.4129683822393417, "router_z_loss": 0.6357056647539139, "routing_entropy_normalized": 0.704080814250268, "learning_rate": 0.003, "grad_norm": 0.22696860134601593, "tokens_per_second": 4211901.321600787, "eta_seconds": 13339.035696744919, "data_wait_seconds": 2.2172927856445312e-05} {"step": 1830, "total_steps": 28610, "tokens": 3837788160, "source": "fineweb_hq", "phase": 1, "loss": 3.5200331807136536, "ce_loss": 3.503923147916794, "logit_z_loss": 11.920490384101868, "router_aux_loss": 1.4291839003562927, "router_z_loss": 0.6261070109903812, "routing_entropy_normalized": 0.7123995799744836, "learning_rate": 0.003, "grad_norm": 0.20241974294185638, "tokens_per_second": 3591177.221745091, "eta_seconds": 15638.807859420776, "data_wait_seconds": 1.96993350982666e-05} {"step": 1840, "total_steps": 28610, "tokens": 3858759680, "source": "fineweb_hq", "phase": 1, "loss": 3.5486922562122345, "ce_loss": 3.5325986742973328, "logit_z_loss": 12.049554228782654, "router_aux_loss": 1.4227294325828552, "router_z_loss": 0.6612859889864922, "routing_entropy_normalized": 0.7146035802185748, "learning_rate": 0.003, "grad_norm": 0.2220405489206314, "tokens_per_second": 3623118.5231244247, "eta_seconds": 15495.148359537125, "data_wait_seconds": 2.3663043975830078e-05} {"step": 1850, "total_steps": 28610, "tokens": 3879731200, "source": "fineweb_edu", "phase": 1, "loss": 3.6017991304397583, "ce_loss": 3.5856920182704926, "logit_z_loss": 11.952550888061523, "router_aux_loss": 1.4303028434515, "router_z_loss": 0.6088003143668175, "routing_entropy_normalized": 0.6894418977165861, "learning_rate": 0.003, "grad_norm": 0.25522324442863464, "tokens_per_second": 3701186.8634031177, "eta_seconds": 15162.646359443665, "data_wait_seconds": 1.6003847122192383e-05} {"step": 1860, "total_steps": 28610, "tokens": 3900702720, "source": "fineweb_hq", "phase": 1, "loss": 3.562455505132675, "ce_loss": 3.5458378195762634, "logit_z_loss": 11.903272986412048, "router_aux_loss": 1.479364886879921, "router_z_loss": 0.6336999163031578, "routing_entropy_normalized": 0.6846200158396274, "learning_rate": 0.003, "grad_norm": 0.24270427227020264, "tokens_per_second": 3991431.5915783136, "eta_seconds": 14054.810839891434, "data_wait_seconds": 1.7523765563964844e-05} {"step": 1870, "total_steps": 28610, "tokens": 3921674240, "source": "fineweb_edu", "phase": 1, "loss": 3.57322359085083, "ce_loss": 3.55679252743721, "logit_z_loss": 12.048587441444397, "router_aux_loss": 1.4627730548381805, "router_z_loss": 0.5984677746891975, "routing_entropy_normalized": 0.6766363289215351, "learning_rate": 0.003, "grad_norm": 0.25126710534095764, "tokens_per_second": 3713072.770436723, "eta_seconds": 15102.813208103178, "data_wait_seconds": 1.6391277313232422e-05} {"step": 1880, "total_steps": 28610, "tokens": 3942645760, "source": "fineweb_hq", "phase": 1, "loss": 3.4910221695899963, "ce_loss": 3.474504381418228, "logit_z_loss": 12.35804569721222, "router_aux_loss": 1.4623255878686905, "router_z_loss": 0.6587833315134048, "routing_entropy_normalized": 0.680685379496983, "learning_rate": 0.003, "grad_norm": 0.21388989686965942, "tokens_per_second": 4197551.31218762, "eta_seconds": 13354.660560607912, "data_wait_seconds": 2.086162567138672e-05} {"step": 1890, "total_steps": 28610, "tokens": 3963617280, "source": "fineweb_edu", "phase": 1, "loss": 3.5723690688610077, "ce_loss": 3.555711716413498, "logit_z_loss": 12.224458575248718, "router_aux_loss": 1.474432110786438, "router_z_loss": 0.6905104294419289, "routing_entropy_normalized": 0.7118379500676819, "learning_rate": 0.003, "grad_norm": 0.22639663517475128, "tokens_per_second": 3382998.5897044935, "eta_seconds": 16563.974224090576, "data_wait_seconds": 1.6123056411743164e-05} {"step": 1900, "total_steps": 28610, "tokens": 3984588800, "source": "fineweb_hq", "phase": 1, "loss": 3.530757337808609, "ce_loss": 3.514582872390747, "logit_z_loss": 11.843889594078064, "router_aux_loss": 1.432109460234642, "router_z_loss": 0.6689532399177551, "routing_entropy_normalized": 0.7242627331639951, "learning_rate": 0.003, "grad_norm": 0.19698300957679749, "tokens_per_second": 3581236.6284031565, "eta_seconds": 15641.225568771362, "data_wait_seconds": 1.77919864654541e-05} {"step": 1910, "total_steps": 28610, "tokens": 4005560320, "source": "fineweb_hq", "phase": 1, "loss": 3.5071572363376617, "ce_loss": 3.4911241233348846, "logit_z_loss": 12.118128180503845, "router_aux_loss": 1.4181169867515564, "router_z_loss": 0.6401231214404106, "routing_entropy_normalized": 0.6894048486004632, "learning_rate": 0.003, "grad_norm": 0.20051071047782898, "tokens_per_second": 3552704.608017557, "eta_seconds": 15760.938377380371, "data_wait_seconds": 1.9758939743041992e-05} {"step": 1920, "total_steps": 28610, "tokens": 4026531840, "source": "fineweb_hq", "phase": 1, "loss": 3.5437348186969757, "ce_loss": 3.5280700027942657, "logit_z_loss": 12.247926235198975, "router_aux_loss": 1.3810828477144241, "router_z_loss": 0.6292600370943546, "routing_entropy_normalized": 0.7044317125370483, "learning_rate": 0.003, "grad_norm": 0.19208577275276184, "tokens_per_second": 3855555.5760045312, "eta_seconds": 14517.489315509796, "data_wait_seconds": 2.232193946838379e-05} {"step": 1930, "total_steps": 28610, "tokens": 4047503360, "source": "fineweb_hq", "phase": 1, "loss": 3.5515803694725037, "ce_loss": 3.53548926115036, "logit_z_loss": 12.417002081871033, "router_aux_loss": 1.4117512702941895, "router_z_loss": 0.731928676366806, "routing_entropy_normalized": 0.6985919504801973, "learning_rate": 0.003, "grad_norm": 0.2605181634426117, "tokens_per_second": 3724794.709580455, "eta_seconds": 15021.503122329712, "data_wait_seconds": 2.2798776626586914e-05} {"step": 1940, "total_steps": 28610, "tokens": 4068474880, "source": "fineweb_hq", "phase": 1, "loss": 3.5101678371429443, "ce_loss": 3.493925780057907, "logit_z_loss": 12.076547265052795, "router_aux_loss": 1.4423953592777252, "router_z_loss": 0.6104035973548889, "routing_entropy_normalized": 0.6992138845684914, "learning_rate": 0.003, "grad_norm": 0.1951877623796463, "tokens_per_second": 3644432.3793238904, "eta_seconds": 15346.98356795311, "data_wait_seconds": 1.6897916793823242e-05} {"step": 1950, "total_steps": 28610, "tokens": 4089446400, "source": "fineweb_edu", "phase": 1, "loss": 3.5933322608470917, "ce_loss": 3.57756844162941, "logit_z_loss": 11.922068357467651, "router_aux_loss": 1.3960795849561691, "router_z_loss": 0.6107672229409218, "routing_entropy_normalized": 0.7062568268735944, "learning_rate": 0.003, "grad_norm": 0.26145854592323303, "tokens_per_second": 2874852.3033147585, "eta_seconds": 19447.980772972107, "data_wait_seconds": 1.6242265701293945e-05} {"step": 1960, "total_steps": 28610, "tokens": 4110417920, "source": "fineweb_hq", "phase": 1, "loss": 3.5205908715724945, "ce_loss": 3.50492724776268, "logit_z_loss": 12.179157972335815, "router_aux_loss": 1.3842057883739471, "router_z_loss": 0.6036059074103832, "routing_entropy_normalized": 0.7054725389525387, "learning_rate": 0.003, "grad_norm": 0.2226119488477707, "tokens_per_second": 4496401.931044143, "eta_seconds": 12429.738634824751, "data_wait_seconds": 1.9490718841552734e-05} {"step": 1970, "total_steps": 28610, "tokens": 4131389440, "source": "fineweb_edu", "phase": 1, "loss": 3.56723889708519, "ce_loss": 3.5516345500946045, "logit_z_loss": 12.068774700164795, "router_aux_loss": 1.3843693733215332, "router_z_loss": 0.5538170598447323, "routing_entropy_normalized": 0.7128641101470757, "learning_rate": 0.003, "grad_norm": 0.2311861664056778, "tokens_per_second": 3263647.26913715, "eta_seconds": 17118.311101913452, "data_wait_seconds": 2.014636993408203e-05} {"step": 1980, "total_steps": 28610, "tokens": 4152360960, "source": "fineweb_hq", "phase": 1, "loss": 3.540110170841217, "ce_loss": 3.524362027645111, "logit_z_loss": 11.934160351753235, "router_aux_loss": 1.3960226029157639, "router_z_loss": 0.5944880843162537, "routing_entropy_normalized": 0.6818665616317066, "learning_rate": 0.003, "grad_norm": 0.22574679553508759, "tokens_per_second": 4279943.190732999, "eta_seconds": 13048.574542045595, "data_wait_seconds": 1.7970800399780273e-05} {"step": 1990, "total_steps": 28610, "tokens": 4173332480, "source": "fineweb_edu", "phase": 1, "loss": 3.5465550124645233, "ce_loss": 3.5311019718647003, "logit_z_loss": 11.683764457702637, "router_aux_loss": 1.3682133853435516, "router_z_loss": 0.6025756224989891, "routing_entropy_normalized": 0.7005259424226893, "learning_rate": 0.003, "grad_norm": 0.2543196380138397, "tokens_per_second": 3404183.549580039, "eta_seconds": 16399.287942886353, "data_wait_seconds": 2.4020671844482422e-05} {"step": 2000, "total_steps": 28610, "tokens": 4194304000, "source": "fineweb_hq", "phase": 1, "loss": 3.5159073770046234, "ce_loss": 3.4999566078186035, "logit_z_loss": 12.084182143211365, "router_aux_loss": 1.4132578819990158, "router_z_loss": 0.6097422242164612, "routing_entropy_normalized": 0.6804054581032509, "learning_rate": 0.003, "grad_norm": 0.20053182542324066, "tokens_per_second": 3875322.476292538, "eta_seconds": 14400.147358417511, "data_wait_seconds": 2.5272369384765625e-05} {"step": 2010, "total_steps": 28610, "tokens": 4215275520, "source": "fineweb_hq", "phase": 1, "loss": 3.530461072921753, "ce_loss": 3.5142329335212708, "logit_z_loss": 12.157607197761536, "router_aux_loss": 1.4493969827890396, "router_z_loss": 0.5183744691312313, "routing_entropy_normalized": 0.6741045387879708, "learning_rate": 0.003, "grad_norm": 0.20575912296772003, "tokens_per_second": 3682952.94931153, "eta_seconds": 15146.60761833191, "data_wait_seconds": 1.9103288650512695e-05} {"step": 2020, "total_steps": 28610, "tokens": 4236247040, "source": "fineweb_hq", "phase": 1, "loss": 3.5414947271347046, "ce_loss": 3.5253705978393555, "logit_z_loss": 12.083802819252014, "router_aux_loss": 1.4359378963708878, "router_z_loss": 0.5563774593174458, "routing_entropy_normalized": 0.6864348188311096, "learning_rate": 0.003, "grad_norm": 0.2581466734409332, "tokens_per_second": 3930930.3260729425, "eta_seconds": 14185.769539117813, "data_wait_seconds": 1.7762184143066406e-05} {"step": 2030, "total_steps": 28610, "tokens": 4257218560, "source": "fineweb_hq", "phase": 1, "loss": 3.500518888235092, "ce_loss": 3.484375387430191, "logit_z_loss": 11.780099511146545, "router_aux_loss": 1.4313413798809052, "router_z_loss": 0.6520175635814667, "routing_entropy_normalized": 0.7052357837924177, "learning_rate": 0.003, "grad_norm": 0.22518759965896606, "tokens_per_second": 3796359.4188178745, "eta_seconds": 14683.09346151352, "data_wait_seconds": 1.6987323760986328e-05} {"step": 2040, "total_steps": 28610, "tokens": 4278190080, "source": "fineweb_hq", "phase": 1, "loss": 3.520448237657547, "ce_loss": 3.504351943731308, "logit_z_loss": 12.20270323753357, "router_aux_loss": 1.4161378294229507, "router_z_loss": 0.7146251052618027, "routing_entropy_normalized": 0.7153161681990354, "learning_rate": 0.003, "grad_norm": 0.22984443604946136, "tokens_per_second": 3318749.1354055703, "eta_seconds": 16789.85857820511, "data_wait_seconds": 1.6748905181884766e-05} {"step": 2050, "total_steps": 28610, "tokens": 4299161600, "source": "fineweb_edu", "phase": 1, "loss": 3.559662878513336, "ce_loss": 3.5439042448997498, "logit_z_loss": 11.972411155700684, "router_aux_loss": 1.3978803157806396, "router_z_loss": 0.5826094895601273, "routing_entropy_normalized": 0.7153946506743465, "learning_rate": 0.003, "grad_norm": 0.2621327042579651, "tokens_per_second": 3997086.56650396, "eta_seconds": 13935.239128112793, "data_wait_seconds": 1.8388032913208008e-05} {"step": 2060, "total_steps": 28610, "tokens": 4320133120, "source": "fineweb_hq", "phase": 1, "loss": 3.5366051197052, "ce_loss": 3.5205669701099396, "logit_z_loss": 11.998973369598389, "router_aux_loss": 1.420030638575554, "router_z_loss": 0.637981079518795, "routing_entropy_normalized": 0.6992922788136802, "learning_rate": 0.003, "grad_norm": 0.2368747740983963, "tokens_per_second": 3872676.505494995, "eta_seconds": 14377.494614124298, "data_wait_seconds": 2.1636486053466797e-05} {"step": 2070, "total_steps": 28610, "tokens": 4341104640, "source": "fineweb_edu", "phase": 1, "loss": 3.5475724637508392, "ce_loss": 3.5316772758960724, "logit_z_loss": 12.056360125541687, "router_aux_loss": 1.4073675721883774, "router_z_loss": 0.6159020736813545, "routing_entropy_normalized": 0.6871935502929544, "learning_rate": 0.003, "grad_norm": 0.26516327261924744, "tokens_per_second": 3593973.0339684063, "eta_seconds": 15486.597577095032, "data_wait_seconds": 1.8835067749023438e-05} {"step": 2080, "total_steps": 28610, "tokens": 4362076160, "source": "fineweb_hq", "phase": 1, "loss": 3.5121364295482635, "ce_loss": 3.4963232576847076, "logit_z_loss": 12.21113395690918, "router_aux_loss": 1.4082101583480835, "router_z_loss": 0.5099599622189999, "routing_entropy_normalized": 0.716662258185194, "learning_rate": 0.003, "grad_norm": 0.24677859246730804, "tokens_per_second": 2898085.9202248016, "eta_seconds": 19197.99622631073, "data_wait_seconds": 1.71661376953125e-05} {"step": 2090, "total_steps": 28610, "tokens": 4383047680, "source": "fineweb_edu", "phase": 1, "loss": 3.5460840463638306, "ce_loss": 3.5303291976451874, "logit_z_loss": 12.114093661308289, "router_aux_loss": 1.390941098332405, "router_z_loss": 0.6340642049908638, "routing_entropy_normalized": 0.7060898182429715, "learning_rate": 0.003, "grad_norm": 0.28080877661705017, "tokens_per_second": 2626679.200160155, "eta_seconds": 21173.682357788086, "data_wait_seconds": 1.722574234008789e-05} {"step": 2100, "total_steps": 28610, "tokens": 4404019200, "source": "fineweb_hq", "phase": 1, "loss": 3.354222357273102, "ce_loss": 3.338737368583679, "logit_z_loss": 12.443979740142822, "router_aux_loss": 1.364369735121727, "router_z_loss": 0.5968634635210037, "routing_entropy_normalized": 0.7206529186788304, "learning_rate": 0.003, "grad_norm": 0.2953186333179474, "tokens_per_second": 3815208.877964834, "eta_seconds": 14572.072276592255, "data_wait_seconds": 1.952052116394043e-05} {"step": 2110, "total_steps": 28610, "tokens": 4424990720, "source": "fineweb_hq", "phase": 1, "loss": 3.291001319885254, "ce_loss": 3.275536447763443, "logit_z_loss": 12.699362754821777, "router_aux_loss": 1.371290773153305, "router_z_loss": 0.48205243051052094, "routing_entropy_normalized": 0.6886288638133534, "learning_rate": 0.003, "grad_norm": 0.22474701702594757, "tokens_per_second": 3749468.3501959583, "eta_seconds": 14821.975493431091, "data_wait_seconds": 2.0056962966918945e-05} {"step": 2120, "total_steps": 28610, "tokens": 4445962240, "source": "fineweb_hq", "phase": 1, "loss": 3.2679118514060974, "ce_loss": 3.2519772052764893, "logit_z_loss": 12.589669346809387, "router_aux_loss": 1.4109448492527008, "router_z_loss": 0.5662693344056606, "routing_entropy_normalized": 0.6698226746927649, "learning_rate": 0.003, "grad_norm": 0.22274592518806458, "tokens_per_second": 3330394.603745541, "eta_seconds": 16680.77302837372, "data_wait_seconds": 1.868605613708496e-05} {"step": 2130, "total_steps": 28610, "tokens": 4466933760, "source": "fineweb_hq", "phase": 1, "loss": 3.2543572187423706, "ce_loss": 3.23779159784317, "logit_z_loss": 12.759912729263306, "router_aux_loss": 1.4745433181524277, "router_z_loss": 0.5441976301372051, "routing_entropy_normalized": 0.6748004451563322, "learning_rate": 0.003, "grad_norm": 0.2698777914047241, "tokens_per_second": 3985036.347762779, "eta_seconds": 13935.276899337769, "data_wait_seconds": 1.5854835510253906e-05} {"step": 2140, "total_steps": 28610, "tokens": 4487905280, "source": "fineweb_hq", "phase": 1, "loss": 3.261254519224167, "ce_loss": 3.2446908354759216, "logit_z_loss": 12.357620120048523, "router_aux_loss": 1.4759746193885803, "router_z_loss": 0.5681674443185329, "routing_entropy_normalized": 0.6784801701444827, "learning_rate": 0.003, "grad_norm": 0.22565294802188873, "tokens_per_second": 3973153.681868938, "eta_seconds": 13971.675370454788, "data_wait_seconds": 2.053380012512207e-05} {"step": 2150, "total_steps": 28610, "tokens": 4508876800, "source": "fineweb_edu", "phase": 1, "loss": 3.5657291412353516, "ce_loss": 3.5494413673877716, "logit_z_loss": 11.975821733474731, "router_aux_loss": 1.4460800737142563, "router_z_loss": 0.6294566467404366, "routing_entropy_normalized": 0.6864282209646562, "learning_rate": 0.003, "grad_norm": 0.26362958550453186, "tokens_per_second": 3749752.2240612693, "eta_seconds": 14798.482300758362, "data_wait_seconds": 1.767277717590332e-05} {"step": 2160, "total_steps": 28610, "tokens": 4529848320, "source": "fineweb_hq", "phase": 1, "loss": 3.2339486479759216, "ce_loss": 3.217913031578064, "logit_z_loss": 12.378138422966003, "router_aux_loss": 1.4245548397302628, "router_z_loss": 0.5522715039551258, "routing_entropy_normalized": 0.7006429605383665, "learning_rate": 0.003, "grad_norm": 0.26878318190574646, "tokens_per_second": 4194962.903494077, "eta_seconds": 13222.922747135164, "data_wait_seconds": 2.0623207092285156e-05} {"step": 2170, "total_steps": 28610, "tokens": 4550819840, "source": "fineweb_edu", "phase": 1, "loss": 3.558891534805298, "ce_loss": 3.5427762269973755, "logit_z_loss": 12.114524126052856, "router_aux_loss": 1.427567720413208, "router_z_loss": 0.6281669661402702, "routing_entropy_normalized": 0.7074364231530047, "learning_rate": 0.003, "grad_norm": 0.24209311604499817, "tokens_per_second": 3565096.912060495, "eta_seconds": 15553.209421157835, "data_wait_seconds": 1.6868114471435547e-05} {"step": 2180, "total_steps": 28610, "tokens": 4571791360, "source": "fineweb_hq", "phase": 1, "loss": 3.220638781785965, "ce_loss": 3.2045379877090454, "logit_z_loss": 12.670435190200806, "router_aux_loss": 1.4221154302358627, "router_z_loss": 0.6125285923480988, "routing_entropy_normalized": 0.70712249319859, "learning_rate": 0.003, "grad_norm": 0.2516325116157532, "tokens_per_second": 4340463.962962556, "eta_seconds": 12770.000588178635, "data_wait_seconds": 1.9043684005737305e-05} {"step": 2190, "total_steps": 28610, "tokens": 4592762880, "source": "fineweb_edu", "phase": 1, "loss": 3.5259363055229187, "ce_loss": 3.5098772644996643, "logit_z_loss": 12.284501314163208, "router_aux_loss": 1.4202302545309067, "router_z_loss": 0.6282784938812256, "routing_entropy_normalized": 0.6964287222692377, "learning_rate": 0.003, "grad_norm": 0.27134665846824646, "tokens_per_second": 4257319.948734588, "eta_seconds": 13014.46837615967, "data_wait_seconds": 1.9490718841552734e-05} {"step": 2200, "total_steps": 28610, "tokens": 4613734400, "source": "fineweb_hq", "phase": 1, "loss": 3.1738244593143463, "ce_loss": 3.157608240842819, "logit_z_loss": 12.788800120353699, "router_aux_loss": 1.435349002480507, "router_z_loss": 0.5838720835745335, "routing_entropy_normalized": 0.6960319036511943, "learning_rate": 0.003, "grad_norm": 0.2639347314834595, "tokens_per_second": 3699201.9683650923, "eta_seconds": 14972.360199213028, "data_wait_seconds": 0.08607977628707886} {"step": 2210, "total_steps": 28610, "tokens": 4634705920, "source": "fineweb_hq", "phase": 1, "loss": 3.211866855621338, "ce_loss": 3.1959357857704163, "logit_z_loss": 12.785617351531982, "router_aux_loss": 1.4121061116456985, "router_z_loss": 0.531396210193634, "routing_entropy_normalized": 0.6818758959996393, "learning_rate": 0.003, "grad_norm": 0.24006424844264984, "tokens_per_second": 4160626.3985774554, "eta_seconds": 13306.845531463623, "data_wait_seconds": 1.8924474716186523e-05} {"step": 2220, "total_steps": 28610, "tokens": 4655677440, "source": "fineweb_hq", "phase": 1, "loss": 3.1463430523872375, "ce_loss": 3.1303778290748596, "logit_z_loss": 12.927671194076538, "router_aux_loss": 1.416484847664833, "router_z_loss": 0.5075956396758556, "routing_entropy_normalized": 0.6892543260317461, "learning_rate": 0.003, "grad_norm": 0.24075551331043243, "tokens_per_second": 3097387.954234922, "eta_seconds": 17867.907442569733, "data_wait_seconds": 2.0444393157958984e-05} {"step": 2230, "total_steps": 28610, "tokens": 4676648960, "source": "fineweb_hq", "phase": 1, "loss": 3.1822783946990967, "ce_loss": 3.1660893261432648, "logit_z_loss": 12.404436707496643, "router_aux_loss": 1.4432108104228973, "router_z_loss": 0.5164520554244518, "routing_entropy_normalized": 0.6713009054464318, "learning_rate": 0.003, "grad_norm": 0.22875995934009552, "tokens_per_second": 4341230.440650465, "eta_seconds": 12743.592056751251, "data_wait_seconds": 2.0176172256469727e-05} {"step": 2240, "total_steps": 28610, "tokens": 4697620480, "source": "fineweb_hq", "phase": 1, "loss": 3.184130907058716, "ce_loss": 3.168467402458191, "logit_z_loss": 12.737381100654602, "router_aux_loss": 1.375965103507042, "router_z_loss": 0.6301095187664032, "routing_entropy_normalized": 0.7121109316532769, "learning_rate": 0.003, "grad_norm": 0.2550067603588104, "tokens_per_second": 3417042.221588494, "eta_seconds": 16184.14249920845, "data_wait_seconds": 1.990795135498047e-05} {"step": 2250, "total_steps": 28610, "tokens": 4718592000, "source": "fineweb_edu", "phase": 1, "loss": 3.5695079565048218, "ce_loss": 3.553580582141876, "logit_z_loss": 11.828019499778748, "router_aux_loss": 1.4102694988250732, "router_z_loss": 0.6418657265603542, "routing_entropy_normalized": 0.6962799175797383, "learning_rate": 0.003, "grad_norm": 0.23571062088012695, "tokens_per_second": 4276407.10450143, "eta_seconds": 12926.956056594849, "data_wait_seconds": 1.958012580871582e-05} {"step": 2260, "total_steps": 28610, "tokens": 4739563520, "source": "fineweb_hq", "phase": 1, "loss": 3.1489611864089966, "ce_loss": 3.1329301297664642, "logit_z_loss": 12.79294764995575, "router_aux_loss": 1.4209490418434143, "router_z_loss": 0.5421938747167587, "routing_entropy_normalized": 0.6950446516767135, "learning_rate": 0.003, "grad_norm": 0.23983997106552124, "tokens_per_second": 3556934.9113707407, "eta_seconds": 15535.83536863327, "data_wait_seconds": 1.475214958190918e-05} {"step": 2270, "total_steps": 28610, "tokens": 4760535040, "source": "fineweb_edu", "phase": 1, "loss": 3.531518816947937, "ce_loss": 3.5157388150691986, "logit_z_loss": 12.116968989372253, "router_aux_loss": 1.394753247499466, "router_z_loss": 0.6207283958792686, "routing_entropy_normalized": 0.7168483473344711, "learning_rate": 0.003, "grad_norm": 0.25984007120132446, "tokens_per_second": 3268464.6409291793, "eta_seconds": 16900.590873241425, "data_wait_seconds": 1.7374753952026367e-05} {"step": 2280, "total_steps": 28610, "tokens": 4781506560, "source": "fineweb_hq", "phase": 1, "loss": 3.164534091949463, "ce_loss": 3.148737460374832, "logit_z_loss": 12.84558916091919, "router_aux_loss": 1.3998879194259644, "router_z_loss": 0.5132103562355042, "routing_entropy_normalized": 0.6937960811599341, "learning_rate": 0.003, "grad_norm": 0.245191752910614, "tokens_per_second": 3716439.910546119, "eta_seconds": 14857.770740032196, "data_wait_seconds": 2.372264862060547e-05} {"step": 2290, "total_steps": 28610, "tokens": 4802478080, "source": "fineweb_edu", "phase": 1, "loss": 3.549920529127121, "ce_loss": 3.53432497382164, "logit_z_loss": 11.564151287078857, "router_aux_loss": 1.3829338103532791, "router_z_loss": 0.6098154336214066, "routing_entropy_normalized": 0.7264826819168761, "learning_rate": 0.003, "grad_norm": 0.2976791262626648, "tokens_per_second": 4116249.5103151533, "eta_seconds": 13409.54684638977, "data_wait_seconds": 2.3156404495239258e-05} {"step": 2300, "total_steps": 28610, "tokens": 4823449600, "source": "fineweb_hq", "phase": 1, "loss": 3.170339524745941, "ce_loss": 3.1546046435832977, "logit_z_loss": 12.397306561470032, "router_aux_loss": 1.3999818116426468, "router_z_loss": 0.49531805515289307, "routing_entropy_normalized": 0.6935401323384915, "learning_rate": 0.003, "grad_norm": 0.3116644620895386, "tokens_per_second": 3616458.86728904, "eta_seconds": 15256.932580947876, "data_wait_seconds": 1.7374753952026367e-05} {"step": 2310, "total_steps": 28610, "tokens": 4844421120, "source": "fineweb_hq", "phase": 1, "loss": 3.1977667212486267, "ce_loss": 3.1819068789482117, "logit_z_loss": 13.000312566757202, "router_aux_loss": 1.41078582406044, "router_z_loss": 0.45188918709754944, "routing_entropy_normalized": 0.6540453874055998, "learning_rate": 0.003, "grad_norm": 0.24395659565925598, "tokens_per_second": 4088789.2971019736, "eta_seconds": 13489.346990585327, "data_wait_seconds": 1.946091651916504e-05} {"step": 2320, "total_steps": 28610, "tokens": 4865392640, "source": "fineweb_hq", "phase": 1, "loss": 3.214217036962509, "ce_loss": 3.198795884847641, "logit_z_loss": 12.220436096191406, "router_aux_loss": 1.3728611469268799, "router_z_loss": 0.47051653265953064, "routing_entropy_normalized": 0.700240722654053, "learning_rate": 0.003, "grad_norm": 0.2963630259037018, "tokens_per_second": 3737767.1421226733, "eta_seconds": 14750.551327466965, "data_wait_seconds": 1.65402889251709e-05} {"step": 2330, "total_steps": 28610, "tokens": 4886364160, "source": "fineweb_hq", "phase": 1, "loss": 3.1477052867412567, "ce_loss": 3.1318709552288055, "logit_z_loss": 13.012690305709839, "router_aux_loss": 1.4082678705453873, "router_z_loss": 0.45035817474126816, "routing_entropy_normalized": 0.6998679310576866, "learning_rate": 0.003, "grad_norm": 0.21342849731445312, "tokens_per_second": 4036464.279255762, "eta_seconds": 13653.81946849823, "data_wait_seconds": 1.7195940017700195e-05} {"step": 2340, "total_steps": 28610, "tokens": 4907335680, "source": "fineweb_hq", "phase": 1, "loss": 3.1717884838581085, "ce_loss": 3.1558149456977844, "logit_z_loss": 12.78376293182373, "router_aux_loss": 1.4201737493276596, "router_z_loss": 0.49344516173005104, "routing_entropy_normalized": 0.6783321765246445, "learning_rate": 0.003, "grad_norm": 0.21787308156490326, "tokens_per_second": 3798411.10214595, "eta_seconds": 14504.007480621338, "data_wait_seconds": 1.9997358322143555e-05} {"step": 2350, "total_steps": 28610, "tokens": 4928307200, "source": "fineweb_edu", "phase": 1, "loss": 3.527879983186722, "ce_loss": 3.512074828147888, "logit_z_loss": 11.545721292495728, "router_aux_loss": 1.4084275662899017, "router_z_loss": 0.566232405602932, "routing_entropy_normalized": 0.6932773156997185, "learning_rate": 0.003, "grad_norm": 0.2842927873134613, "tokens_per_second": 2743290.944449944, "eta_seconds": 20074.870888710022, "data_wait_seconds": 1.8537044525146484e-05} {"step": 2360, "total_steps": 28610, "tokens": 4949278720, "source": "fineweb_hq", "phase": 1, "loss": 3.3547250032424927, "ce_loss": 3.3388253152370453, "logit_z_loss": 12.116823554039001, "router_aux_loss": 1.4159097075462341, "router_z_loss": 0.5290138460695744, "routing_entropy_normalized": 0.7122568483709162, "learning_rate": 0.003, "grad_norm": 0.22612884640693665, "tokens_per_second": 3559249.413467716, "eta_seconds": 15466.811567544937, "data_wait_seconds": 1.874566078186035e-05} {"step": 2370, "total_steps": 28610, "tokens": 4970250240, "source": "fineweb_edu", "phase": 1, "loss": 3.545273572206497, "ce_loss": 3.5295045375823975, "logit_z_loss": 11.84515917301178, "router_aux_loss": 1.403971016407013, "router_z_loss": 0.544792365282774, "routing_entropy_normalized": 0.6882127418148193, "learning_rate": 0.003, "grad_norm": 0.31866776943206787, "tokens_per_second": 4036335.733407892, "eta_seconds": 13633.471572875977, "data_wait_seconds": 2.0176172256469727e-05} {"step": 2380, "total_steps": 28610, "tokens": 4991221760, "source": "fineweb_hq", "phase": 1, "loss": 3.510156571865082, "ce_loss": 3.494475245475769, "logit_z_loss": 11.83108651638031, "router_aux_loss": 1.3993254005908966, "router_z_loss": 0.5049016252160072, "routing_entropy_normalized": 0.6906882175362602, "learning_rate": 0.003, "grad_norm": 0.2722836136817932, "tokens_per_second": 3617776.429460234, "eta_seconds": 15205.00175523758, "data_wait_seconds": 1.919269561767578e-05} {"step": 2390, "total_steps": 28610, "tokens": 5012193280, "source": "fineweb_edu", "phase": 1, "loss": 3.5444388389587402, "ce_loss": 3.5288647413253784, "logit_z_loss": 12.345385909080505, "router_aux_loss": 1.3839027732610703, "router_z_loss": 0.5006159283220768, "routing_entropy_normalized": 0.6895228194638408, "learning_rate": 0.003, "grad_norm": 0.25768592953681946, "tokens_per_second": 3364572.482994955, "eta_seconds": 16343.034878253937, "data_wait_seconds": 1.9669532775878906e-05} {"step": 2400, "total_steps": 28610, "tokens": 5033164800, "source": "fineweb_hq", "phase": 1, "loss": 3.5130556523799896, "ce_loss": 3.4973995983600616, "logit_z_loss": 11.960555672645569, "router_aux_loss": 1.397292211651802, "router_z_loss": 0.4870973937213421, "routing_entropy_normalized": 0.6998323004751675, "learning_rate": 0.003, "grad_norm": 0.2143489271402359, "tokens_per_second": 3473741.9605732877, "eta_seconds": 15823.384276628494, "data_wait_seconds": 1.868605613708496e-05} {"step": 2410, "total_steps": 28610, "tokens": 5054136320, "source": "fineweb_hq", "phase": 1, "loss": 3.5146458446979523, "ce_loss": 3.4992928206920624, "logit_z_loss": 12.117952823638916, "router_aux_loss": 1.3653748333454132, "router_z_loss": 0.4874454215168953, "routing_entropy_normalized": 0.7010444165011964, "learning_rate": 0.003, "grad_norm": 0.20139864087104797, "tokens_per_second": 3401158.7257931004, "eta_seconds": 16154.900970458984, "data_wait_seconds": 1.6927719116210938e-05} {"step": 2420, "total_steps": 28610, "tokens": 5075107840, "source": "fineweb_hq", "phase": 1, "loss": 3.529900372028351, "ce_loss": 3.514676183462143, "logit_z_loss": 11.793428182601929, "router_aux_loss": 1.3511345386505127, "router_z_loss": 0.5334818288683891, "routing_entropy_normalized": 0.6680190797141299, "learning_rate": 0.003, "grad_norm": 0.3027433454990387, "tokens_per_second": 3595221.9458789937, "eta_seconds": 15277.057079315186, "data_wait_seconds": 2.1696090698242188e-05} {"step": 2430, "total_steps": 28610, "tokens": 5096079360, "source": "fineweb_hq", "phase": 1, "loss": 3.515640527009964, "ce_loss": 3.5001842975616455, "logit_z_loss": 12.08555269241333, "router_aux_loss": 1.3750054389238358, "router_z_loss": 0.49755823239684105, "routing_entropy_normalized": 0.6837657675432472, "learning_rate": 0.003, "grad_norm": 0.20839810371398926, "tokens_per_second": 4327731.477536158, "eta_seconds": 12686.424664974213, "data_wait_seconds": 2.0265579223632812e-05} {"step": 2440, "total_steps": 28610, "tokens": 5117050880, "source": "fineweb_hq", "phase": 1, "loss": 3.5285069346427917, "ce_loss": 3.5130508840084076, "logit_z_loss": 11.85972011089325, "router_aux_loss": 1.3759434074163437, "router_z_loss": 0.5106153823435307, "routing_entropy_normalized": 0.6828855638587119, "learning_rate": 0.003, "grad_norm": 0.2899342477321625, "tokens_per_second": 3515399.903366691, "eta_seconds": 15612.012672424318, "data_wait_seconds": 1.946091651916504e-05} {"step": 2450, "total_steps": 28610, "tokens": 5138022400, "source": "fineweb_edu", "phase": 1, "loss": 3.5402529537677765, "ce_loss": 3.524503231048584, "logit_z_loss": 11.746927499771118, "router_aux_loss": 1.4029153287410736, "router_z_loss": 0.5458499565720558, "routing_entropy_normalized": 0.6723418146834506, "learning_rate": 0.003, "grad_norm": 0.28857186436653137, "tokens_per_second": 3829787.2715030964, "eta_seconds": 14324.94612121582, "data_wait_seconds": 1.7821788787841797e-05} {"step": 2460, "total_steps": 28610, "tokens": 5158993920, "source": "fineweb_hq", "phase": 1, "loss": 3.553754687309265, "ce_loss": 3.5380228459835052, "logit_z_loss": 12.086552619934082, "router_aux_loss": 1.404410406947136, "router_z_loss": 0.47909607738256454, "routing_entropy_normalized": 0.6692197401522968, "learning_rate": 0.003, "grad_norm": 0.25103694200515747, "tokens_per_second": 3510566.238531114, "eta_seconds": 15621.561045646667, "data_wait_seconds": 1.7523765563964844e-05} {"step": 2470, "total_steps": 28610, "tokens": 5179965440, "source": "fineweb_edu", "phase": 1, "loss": 3.515208601951599, "ce_loss": 3.499808758497238, "logit_z_loss": 12.21145749092102, "router_aux_loss": 1.3616058230400085, "router_z_loss": 0.5626913718879223, "routing_entropy_normalized": 0.696395875707988, "learning_rate": 0.003, "grad_norm": 0.25054946541786194, "tokens_per_second": 3934950.4554035165, "eta_seconds": 13931.446888923645, "data_wait_seconds": 2.047419548034668e-05} {"step": 2480, "total_steps": 28610, "tokens": 5200936960, "source": "fineweb_hq", "phase": 1, "loss": 3.5644027292728424, "ce_loss": 3.549071043729782, "logit_z_loss": 12.137555003166199, "router_aux_loss": 1.3631112277507782, "router_z_loss": 0.4868196174502373, "routing_entropy_normalized": 0.6917522262775163, "learning_rate": 0.003, "grad_norm": 0.29248154163360596, "tokens_per_second": 3804560.225952847, "eta_seconds": 14403.394480705261, "data_wait_seconds": 2.0503997802734375e-05} {"step": 2490, "total_steps": 28610, "tokens": 5221908480, "source": "fineweb_edu", "phase": 1, "loss": 3.5368979275226593, "ce_loss": 3.5216698348522186, "logit_z_loss": 11.948234796524048, "router_aux_loss": 1.353515237569809, "router_z_loss": 0.498043667525053, "routing_entropy_normalized": 0.6978154060163679, "learning_rate": 0.003, "grad_norm": 0.26010191440582275, "tokens_per_second": 3777756.6282083425, "eta_seconds": 14500.036828994751, "data_wait_seconds": 1.8894672393798828e-05} {"step": 2500, "total_steps": 28610, "tokens": 5242880000, "source": "fineweb_hq", "phase": 1, "loss": 3.4986413717269897, "ce_loss": 3.483454465866089, "logit_z_loss": 11.863211750984192, "router_aux_loss": 1.348578006029129, "router_z_loss": 0.514745332300663, "routing_entropy_normalized": 0.6761622645558353, "learning_rate": 0.003, "grad_norm": 0.21209533512592316, "tokens_per_second": 4373801.653776784, "eta_seconds": 12519.232250213623, "data_wait_seconds": 1.9043684005737305e-05} {"step": 2510, "total_steps": 28610, "tokens": 5263851520, "source": "fineweb_hq", "phase": 1, "loss": 3.5137806832790375, "ce_loss": 3.49894779920578, "logit_z_loss": 12.053791642189026, "router_aux_loss": 1.3177195489406586, "router_z_loss": 0.4502948448061943, "routing_entropy_normalized": 0.7146016176723123, "learning_rate": 0.003, "grad_norm": 0.2463657408952713, "tokens_per_second": 3883460.602831567, "eta_seconds": 14094.559671878815, "data_wait_seconds": 2.0742416381835938e-05} {"step": 2520, "total_steps": 28610, "tokens": 5284823040, "source": "fineweb_hq", "phase": 1, "loss": 3.520757257938385, "ce_loss": 3.505963146686554, "logit_z_loss": 12.043269157409668, "router_aux_loss": 1.3149009346961975, "router_z_loss": 0.440849207341671, "routing_entropy_normalized": 0.7079998269089404, "learning_rate": 0.003, "grad_norm": 0.3138568103313446, "tokens_per_second": 3607586.294043637, "eta_seconds": 15166.566014051437, "data_wait_seconds": 1.609325408935547e-05} {"step": 2530, "total_steps": 28610, "tokens": 5305794560, "source": "fineweb_hq", "phase": 1, "loss": 3.506144016981125, "ce_loss": 3.49054816365242, "logit_z_loss": 12.245888710021973, "router_aux_loss": 1.385962501168251, "router_z_loss": 0.511612456291914, "routing_entropy_normalized": 0.6586423962220098, "learning_rate": 0.003, "grad_norm": 0.1987476646900177, "tokens_per_second": 3950568.3944612504, "eta_seconds": 13844.52025604248, "data_wait_seconds": 1.806020736694336e-05} {"step": 2540, "total_steps": 28610, "tokens": 5326766080, "source": "fineweb_hq", "phase": 1, "loss": 3.518901616334915, "ce_loss": 3.5034914016723633, "logit_z_loss": 12.26131796836853, "router_aux_loss": 1.371120646595955, "router_z_loss": 0.4727824702858925, "routing_entropy_normalized": 0.6851910656271641, "learning_rate": 0.003, "grad_norm": 0.2596409022808075, "tokens_per_second": 3948372.8428473105, "eta_seconds": 13846.907274484634, "data_wait_seconds": 1.823902130126953e-05} {"step": 2550, "total_steps": 28610, "tokens": 5347737600, "source": "fineweb_edu", "phase": 1, "loss": 3.559194326400757, "ce_loss": 3.543692708015442, "logit_z_loss": 11.833209037780762, "router_aux_loss": 1.382506102323532, "router_z_loss": 0.493238165974617, "routing_entropy_normalized": 0.6679737944873547, "learning_rate": 0.003, "grad_norm": 0.3830341100692749, "tokens_per_second": 4255520.820557916, "eta_seconds": 12842.559917926788, "data_wait_seconds": 1.6868114471435547e-05} {"step": 2560, "total_steps": 28610, "tokens": 5368709120, "source": "fineweb_hq", "phase": 1, "loss": 3.4781463146209717, "ce_loss": 3.4625964164733887, "logit_z_loss": 12.003105163574219, "router_aux_loss": 1.3868396133184433, "router_z_loss": 0.4811953343451023, "routing_entropy_normalized": 0.7045021679657401, "learning_rate": 0.003, "grad_norm": 0.19933265447616577, "tokens_per_second": 3570208.951700162, "eta_seconds": 15301.852171421051, "data_wait_seconds": 2.0235776901245117e-05} {"step": 2570, "total_steps": 28610, "tokens": 5389680640, "source": "fineweb_edu", "phase": 1, "loss": 3.550793081521988, "ce_loss": 3.5352513194084167, "logit_z_loss": 11.813325881958008, "router_aux_loss": 1.3816313594579697, "router_z_loss": 0.5441366173326969, "routing_entropy_normalized": 0.7032725374475257, "learning_rate": 0.003, "grad_norm": 0.270443856716156, "tokens_per_second": 3502599.156019511, "eta_seconds": 15591.23258113861, "data_wait_seconds": 1.683831214904785e-05} {"step": 2580, "total_steps": 28610, "tokens": 5410652160, "source": "fineweb_hq", "phase": 1, "loss": 3.501415729522705, "ce_loss": 3.4857016801834106, "logit_z_loss": 12.149945497512817, "router_aux_loss": 1.399991363286972, "router_z_loss": 0.4991791471838951, "routing_entropy_normalized": 0.6806284287648505, "learning_rate": 0.003, "grad_norm": 0.24643222987651825, "tokens_per_second": 2695626.702127451, "eta_seconds": 20250.89991760254, "data_wait_seconds": 0.2205761969089508} {"step": 2590, "total_steps": 28610, "tokens": 5431623680, "source": "fineweb_edu", "phase": 1, "loss": 3.4996128976345062, "ce_loss": 3.4845140874385834, "logit_z_loss": 11.824223279953003, "router_aux_loss": 1.331661581993103, "router_z_loss": 0.599764458835125, "routing_entropy_normalized": 0.7055403752963004, "learning_rate": 0.003, "grad_norm": 0.20982202887535095, "tokens_per_second": 3563835.19478872, "eta_seconds": 15311.565226078033, "data_wait_seconds": 2.0742416381835938e-05} {"step": 2600, "total_steps": 28610, "tokens": 5452595200, "source": "fineweb_hq", "phase": 1, "loss": 3.494413048028946, "ce_loss": 3.4792399406433105, "logit_z_loss": 11.85681676864624, "router_aux_loss": 1.3479110896587372, "router_z_loss": 0.5082365795969963, "routing_entropy_normalized": 0.6736857379732194, "learning_rate": 0.003, "grad_norm": 0.25411784648895264, "tokens_per_second": 4138062.5905964323, "eta_seconds": 13181.754100084305, "data_wait_seconds": 1.9937753677368164e-05} {"step": 2610, "total_steps": 28610, "tokens": 5473566720, "source": "fineweb_hq", "phase": 1, "loss": 3.4922205209732056, "ce_loss": 3.477086991071701, "logit_z_loss": 11.812228560447693, "router_aux_loss": 1.349942609667778, "router_z_loss": 0.45282404497265816, "routing_entropy_normalized": 0.6701176587282116, "learning_rate": 0.003, "grad_norm": 0.261580228805542, "tokens_per_second": 3127707.0317506553, "eta_seconds": 17433.203125, "data_wait_seconds": 1.570582389831543e-05} {"step": 2620, "total_steps": 28610, "tokens": 5494538240, "source": "fineweb_hq", "phase": 1, "loss": 3.456324130296707, "ce_loss": 3.4411998987197876, "logit_z_loss": 12.032633662223816, "router_aux_loss": 1.3370039463043213, "router_z_loss": 0.5509357191622257, "routing_entropy_normalized": 0.6758933065841954, "learning_rate": 0.003, "grad_norm": 0.23428593575954437, "tokens_per_second": 3554710.9007993164, "eta_seconds": 15333.168294429779, "data_wait_seconds": 1.728534698486328e-05} {"step": 2630, "total_steps": 28610, "tokens": 5515509760, "source": "fineweb_hq", "phase": 1, "loss": 3.5029638707637787, "ce_loss": 3.487671583890915, "logit_z_loss": 11.94146180152893, "router_aux_loss": 1.3596330136060715, "router_z_loss": 0.5018142685294151, "routing_entropy_normalized": 0.6938536940928396, "learning_rate": 0.003, "grad_norm": 0.1926933079957962, "tokens_per_second": 4090823.8026539753, "eta_seconds": 13318.590970516205, "data_wait_seconds": 2.065300941467285e-05} {"step": 2640, "total_steps": 28610, "tokens": 5536481280, "source": "fineweb_hq", "phase": 1, "loss": 3.466720074415207, "ce_loss": 3.451043665409088, "logit_z_loss": 11.874385476112366, "router_aux_loss": 1.3814519494771957, "router_z_loss": 0.6743877790868282, "routing_entropy_normalized": 0.670481384048529, "learning_rate": 0.003, "grad_norm": 0.2626158893108368, "tokens_per_second": 3765571.355712931, "eta_seconds": 14463.419304847717, "data_wait_seconds": 1.913309097290039e-05} {"step": 2650, "total_steps": 28610, "tokens": 5557452800, "source": "fineweb_edu", "phase": 1, "loss": 3.5371301770210266, "ce_loss": 3.522217482328415, "logit_z_loss": 11.903787970542908, "router_aux_loss": 1.318710446357727, "router_z_loss": 0.5352545119822025, "routing_entropy_normalized": 0.6842485358717174, "learning_rate": 0.003, "grad_norm": 0.29612401127815247, "tokens_per_second": 4306333.029361135, "eta_seconds": 12642.325976371765, "data_wait_seconds": 1.817941665649414e-05} {"step": 2660, "total_steps": 28610, "tokens": 5578424320, "source": "fineweb_hq", "phase": 1, "loss": 3.5143531560897827, "ce_loss": 3.4990962743759155, "logit_z_loss": 11.96633505821228, "router_aux_loss": 1.3550478368997574, "router_z_loss": 0.5097497776150703, "routing_entropy_normalized": 0.6784977079089476, "learning_rate": 0.003, "grad_norm": 0.2398582249879837, "tokens_per_second": 3275168.8055787496, "eta_seconds": 16616.271597146988, "data_wait_seconds": 1.7315149307250977e-05} {"step": 2670, "total_steps": 28610, "tokens": 5599395840, "source": "fineweb_edu", "phase": 1, "loss": 3.4388111531734467, "ce_loss": 3.4236688017845154, "logit_z_loss": 11.909855604171753, "router_aux_loss": 1.3415011912584305, "router_z_loss": 0.5362791791558266, "routing_entropy_normalized": 0.6858596925742587, "learning_rate": 0.003, "grad_norm": 0.20784695446491241, "tokens_per_second": 3941099.4763617404, "eta_seconds": 13803.285912036896, "data_wait_seconds": 1.7911195755004883e-05} {"step": 2680, "total_steps": 28610, "tokens": 5620367360, "source": "fineweb_hq", "phase": 1, "loss": 3.4783579409122467, "ce_loss": 3.463290125131607, "logit_z_loss": 12.22532570362091, "router_aux_loss": 1.3382646590471268, "router_z_loss": 0.46264657378196716, "routing_entropy_normalized": 0.6943611605189758, "learning_rate": 0.003, "grad_norm": 0.24875696003437042, "tokens_per_second": 3891416.39520446, "eta_seconds": 13974.128141880035, "data_wait_seconds": 1.913309097290039e-05} {"step": 2690, "total_steps": 28610, "tokens": 5641338880, "source": "fineweb_edu", "phase": 1, "loss": 3.510425329208374, "ce_loss": 3.495286673307419, "logit_z_loss": 12.030400395393372, "router_aux_loss": 1.3403729051351547, "router_z_loss": 0.5319110080599785, "routing_entropy_normalized": 0.6776616241456361, "learning_rate": 0.003, "grad_norm": 0.2409120500087738, "tokens_per_second": 3369444.549036066, "eta_seconds": 16132.682716369629, "data_wait_seconds": 1.940131187438965e-05} {"step": 2700, "total_steps": 28610, "tokens": 5662310400, "source": "fineweb_hq", "phase": 1, "loss": 3.4829120337963104, "ce_loss": 3.467543065547943, "logit_z_loss": 12.114116311073303, "router_aux_loss": 1.3705081194639206, "router_z_loss": 0.45251182094216347, "routing_entropy_normalized": 0.6598431098057417, "learning_rate": 0.003, "grad_norm": 0.2542835474014282, "tokens_per_second": 3988131.4093574556, "eta_seconds": 13624.728661775589, "data_wait_seconds": 2.0682811737060547e-05} {"step": 2710, "total_steps": 28610, "tokens": 5683281920, "source": "fineweb_hq", "phase": 1, "loss": 3.44709312915802, "ce_loss": 3.4319747984409332, "logit_z_loss": 12.167679190635681, "router_aux_loss": 1.3426994681358337, "router_z_loss": 0.4745776727795601, "routing_entropy_normalized": 0.682618621724877, "learning_rate": 0.003, "grad_norm": 0.23942188918590546, "tokens_per_second": 3077416.9011614188, "eta_seconds": 17649.944269657135, "data_wait_seconds": 2.199411392211914e-05} {"step": 2720, "total_steps": 28610, "tokens": 5704253440, "source": "fineweb_hq", "phase": 1, "loss": 3.4994808733463287, "ce_loss": 3.484289675951004, "logit_z_loss": 11.77999496459961, "router_aux_loss": 1.3549619019031525, "router_z_loss": 0.46356816962361336, "routing_entropy_normalized": 0.6718240524280858, "learning_rate": 0.003, "grad_norm": 0.28948885202407837, "tokens_per_second": 4094606.6752964696, "eta_seconds": 13260.19068145752, "data_wait_seconds": 1.850724220275879e-05} {"step": 2730, "total_steps": 28610, "tokens": 5725224960, "source": "fineweb_hq", "phase": 1, "loss": 3.491524875164032, "ce_loss": 3.476178914308548, "logit_z_loss": 12.090714454650879, "router_aux_loss": 1.368373066186905, "router_z_loss": 0.45312289148569107, "routing_entropy_normalized": 0.6716881803860107, "learning_rate": 0.003, "grad_norm": 0.2182619869709015, "tokens_per_second": 3743153.379410457, "eta_seconds": 14499.617904663086, "data_wait_seconds": 1.58846378326416e-05} {"step": 2740, "total_steps": 28610, "tokens": 5746196480, "source": "fineweb_hq", "phase": 1, "loss": 3.5081222355365753, "ce_loss": 3.492416590452194, "logit_z_loss": 12.267538905143738, "router_aux_loss": 1.4044285714626312, "router_z_loss": 0.4345939829945564, "routing_entropy_normalized": 0.6495182113780187, "learning_rate": 0.003, "grad_norm": 0.2726545035839081, "tokens_per_second": 3372135.621217026, "eta_seconds": 16088.71300983429, "data_wait_seconds": 1.9937753677368164e-05} {"step": 2750, "total_steps": 28610, "tokens": 5767168000, "source": "fineweb_edu", "phase": 1, "loss": 3.534697264432907, "ce_loss": 3.518997460603714, "logit_z_loss": 11.857022285461426, "router_aux_loss": 1.3963384628295898, "router_z_loss": 0.5507389008998871, "routing_entropy_normalized": 0.6895758077816342, "learning_rate": 0.003, "grad_norm": 0.32714951038360596, "tokens_per_second": 4200628.923492294, "eta_seconds": 12910.53118658066, "data_wait_seconds": 1.6987323760986328e-05} {"step": 2760, "total_steps": 28610, "tokens": 5788139520, "source": "fineweb_hq", "phase": 1, "loss": 3.4882917404174805, "ce_loss": 3.4728649854660034, "logit_z_loss": 11.803070664405823, "router_aux_loss": 1.3722852170467377, "router_z_loss": 0.5236307829618454, "routing_entropy_normalized": 0.6767980226293943, "learning_rate": 0.003, "grad_norm": 0.2872109115123749, "tokens_per_second": 3551538.542530375, "eta_seconds": 15264.195657968521, "data_wait_seconds": 2.1070241928100586e-05} {"step": 2770, "total_steps": 28610, "tokens": 5809111040, "source": "fineweb_edu", "phase": 1, "loss": 3.5221186578273773, "ce_loss": 3.506973475217819, "logit_z_loss": 12.009633421897888, "router_aux_loss": 1.3406021147966385, "router_z_loss": 0.5381851345300674, "routing_entropy_normalized": 0.7057704031526539, "learning_rate": 0.003, "grad_norm": 0.2915167808532715, "tokens_per_second": 4264849.710345293, "eta_seconds": 12706.28776168823, "data_wait_seconds": 1.913309097290039e-05} {"step": 2780, "total_steps": 28610, "tokens": 5830082560, "source": "fineweb_hq", "phase": 1, "loss": 3.447486162185669, "ce_loss": 3.432284891605377, "logit_z_loss": 12.0490003824234, "router_aux_loss": 1.3538788408041, "router_z_loss": 0.45761506631970406, "routing_entropy_normalized": 0.6802862805351612, "learning_rate": 0.003, "grad_norm": 0.2630041241645813, "tokens_per_second": 3635985.1171757807, "eta_seconds": 14898.145733356476, "data_wait_seconds": 1.7702579498291016e-05} {"step": 2790, "total_steps": 28610, "tokens": 5851054080, "source": "fineweb_edu", "phase": 1, "loss": 3.5146404504776, "ce_loss": 3.4997361600399017, "logit_z_loss": 12.020304799079895, "router_aux_loss": 1.321679413318634, "router_z_loss": 0.48549576848745346, "routing_entropy_normalized": 0.6864521950487673, "learning_rate": 0.003, "grad_norm": 0.31197190284729004, "tokens_per_second": 3864200.116270953, "eta_seconds": 14012.852080821991, "data_wait_seconds": 1.862645149230957e-05} {"step": 2800, "total_steps": 28610, "tokens": 5872025600, "source": "fineweb_hq", "phase": 1, "loss": 3.4687364995479584, "ce_loss": 3.4535909295082092, "logit_z_loss": 12.136942505836487, "router_aux_loss": 1.3457804769277573, "router_z_loss": 0.4741276577115059, "routing_entropy_normalized": 0.6963124624610783, "learning_rate": 0.003, "grad_norm": 0.2358403354883194, "tokens_per_second": 3936485.6908772597, "eta_seconds": 13750.20700454712, "data_wait_seconds": 2.1070241928100586e-05} {"step": 2810, "total_steps": 28610, "tokens": 5892997120, "source": "fineweb_hq", "phase": 1, "loss": 3.4307749569416046, "ce_loss": 3.4154734909534454, "logit_z_loss": 12.115597248077393, "router_aux_loss": 1.3634073138237, "router_z_loss": 0.45582254230976105, "routing_entropy_normalized": 0.6484858254494656, "learning_rate": 0.003, "grad_norm": 0.278799444437027, "tokens_per_second": 3090439.3029903374, "eta_seconds": 17507.712106704712, "data_wait_seconds": 0.36113184690475464} {"step": 2820, "total_steps": 28610, "tokens": 5913968640, "source": "fineweb_hq", "phase": 1, "loss": 3.439435601234436, "ce_loss": 3.4241019785404205, "logit_z_loss": 12.031750798225403, "router_aux_loss": 1.3688681572675705, "router_z_loss": 0.441714521497488, "routing_entropy_normalized": 0.6675177394525115, "learning_rate": 0.003, "grad_norm": 0.19829438626766205, "tokens_per_second": 3506603.264450463, "eta_seconds": 15423.914826154709, "data_wait_seconds": 1.9073486328125e-05} {"step": 2830, "total_steps": 28610, "tokens": 5934940160, "source": "fineweb_hq", "phase": 1, "loss": 3.424293488264084, "ce_loss": 3.4090245068073273, "logit_z_loss": 12.050381183624268, "router_aux_loss": 1.3572563230991364, "router_z_loss": 0.4913320727646351, "routing_entropy_normalized": 0.6983684662964483, "learning_rate": 0.003, "grad_norm": 0.21210476756095886, "tokens_per_second": 3301270.646312599, "eta_seconds": 16376.89979171753, "data_wait_seconds": 2.008676528930664e-05} {"step": 2840, "total_steps": 28610, "tokens": 5955911680, "source": "fineweb_hq", "phase": 1, "loss": 3.510751575231552, "ce_loss": 3.495372176170349, "logit_z_loss": 11.812831997871399, "router_aux_loss": 1.366962417960167, "router_z_loss": 0.528475359082222, "routing_entropy_normalized": 0.6615867090278867, "learning_rate": 0.003, "grad_norm": 0.2515926957130432, "tokens_per_second": 4073811.4503865866, "eta_seconds": 13266.104162693024, "data_wait_seconds": 1.7881393432617188e-05} {"step": 2850, "total_steps": 28610, "tokens": 5976883200, "source": "fineweb_edu", "phase": 1, "loss": 3.490424335002899, "ce_loss": 3.4746799767017365, "logit_z_loss": 11.678634285926819, "router_aux_loss": 1.400459736585617, "router_z_loss": 0.5719045363366604, "routing_entropy_normalized": 0.675472923335109, "learning_rate": 0.003, "grad_norm": 0.2652847468852997, "tokens_per_second": 4145011.1940074214, "eta_seconds": 13033.169994354248, "data_wait_seconds": 1.9878149032592773e-05} {"step": 2860, "total_steps": 28610, "tokens": 5997854720, "source": "fineweb_hq", "phase": 1, "loss": 3.4866034388542175, "ce_loss": 3.470905691385269, "logit_z_loss": 12.1615731716156, "router_aux_loss": 1.3972266912460327, "router_z_loss": 0.5092474929988384, "routing_entropy_normalized": 0.69390132636884, "learning_rate": 0.003, "grad_norm": 0.30747953057289124, "tokens_per_second": 3703478.8638086156, "eta_seconds": 14581.334465742111, "data_wait_seconds": 1.913309097290039e-05} {"step": 2870, "total_steps": 28610, "tokens": 6018826240, "source": "fineweb_edu", "phase": 1, "loss": 3.514873504638672, "ce_loss": 3.4990696907043457, "logit_z_loss": 12.121455907821655, "router_aux_loss": 1.4024617969989777, "router_z_loss": 0.5669973306357861, "routing_entropy_normalized": 0.681244907386276, "learning_rate": 0.003, "grad_norm": 0.31988123059272766, "tokens_per_second": 4356373.203023147, "eta_seconds": 12391.200194358826, "data_wait_seconds": 1.570582389831543e-05} {"step": 2880, "total_steps": 28610, "tokens": 6039797760, "source": "fineweb_hq", "phase": 1, "loss": 3.4596517384052277, "ce_loss": 3.444355398416519, "logit_z_loss": 11.756344079971313, "router_aux_loss": 1.3537446558475494, "router_z_loss": 0.583260640501976, "routing_entropy_normalized": 0.6858098961191204, "learning_rate": 0.003, "grad_norm": 0.2484062910079956, "tokens_per_second": 3561820.4671183387, "eta_seconds": 15149.478043079376, "data_wait_seconds": 1.5288591384887695e-05} {"step": 2890, "total_steps": 28610, "tokens": 6060769280, "source": "fineweb_edu", "phase": 1, "loss": 3.482335478067398, "ce_loss": 3.467074543237686, "logit_z_loss": 11.724309206008911, "router_aux_loss": 1.351623460650444, "router_z_loss": 0.5722124390304089, "routing_entropy_normalized": 0.6832847561206233, "learning_rate": 0.003, "grad_norm": 0.22009548544883728, "tokens_per_second": 3371596.4943157425, "eta_seconds": 15997.984791755676, "data_wait_seconds": 1.475214958190918e-05} {"step": 2900, "total_steps": 28610, "tokens": 6081740800, "source": "fineweb_hq", "phase": 1, "loss": 3.4009149968624115, "ce_loss": 3.3855835497379303, "logit_z_loss": 12.173528790473938, "router_aux_loss": 1.3636923134326935, "router_z_loss": 0.4771542549133301, "routing_entropy_normalized": 0.6648355572236979, "learning_rate": 0.003, "grad_norm": 0.25565528869628906, "tokens_per_second": 3646333.0706011495, "eta_seconds": 14786.849384307861, "data_wait_seconds": 1.990795135498047e-05} {"step": 2910, "total_steps": 28610, "tokens": 6102712320, "source": "fineweb_hq", "phase": 1, "loss": 3.4291308224201202, "ce_loss": 3.413660705089569, "logit_z_loss": 11.903980374336243, "router_aux_loss": 1.3789523094892502, "router_z_loss": 0.49022355675697327, "routing_entropy_normalized": 0.6857508122214367, "learning_rate": 0.003, "grad_norm": 0.23262183368206024, "tokens_per_second": 3615108.8342190515, "eta_seconds": 14908.76454114914, "data_wait_seconds": 1.940131187438965e-05} {"step": 2920, "total_steps": 28610, "tokens": 6123683840, "source": "fineweb_hq", "phase": 1, "loss": 3.4548803865909576, "ce_loss": 3.4389721155166626, "logit_z_loss": 11.95216429233551, "router_aux_loss": 1.4171973764896393, "router_z_loss": 0.5410659313201904, "routing_entropy_normalized": 0.6694187873021048, "learning_rate": 0.003, "grad_norm": 0.29012423753738403, "tokens_per_second": 3538031.8197498843, "eta_seconds": 15227.628700017929, "data_wait_seconds": 1.9311904907226562e-05} {"step": 2930, "total_steps": 28610, "tokens": 6144655360, "source": "fineweb_hq", "phase": 1, "loss": 3.4024783074855804, "ce_loss": 3.386111170053482, "logit_z_loss": 12.034535765647888, "router_aux_loss": 1.4530006349086761, "router_z_loss": 0.6336953416466713, "routing_entropy_normalized": 0.6468448336100994, "learning_rate": 0.003, "grad_norm": 0.2280038297176361, "tokens_per_second": 3032053.395497814, "eta_seconds": 17761.845302581787, "data_wait_seconds": 2.0295381546020508e-05} {"step": 2940, "total_steps": 28610, "tokens": 6165626880, "source": "fineweb_hq", "phase": 1, "loss": 3.4567441046237946, "ce_loss": 3.440823584794998, "logit_z_loss": 12.209789752960205, "router_aux_loss": 1.4102476388216019, "router_z_loss": 0.5970588810741901, "routing_entropy_normalized": 0.6727024378194232, "learning_rate": 0.003, "grad_norm": 0.234688401222229, "tokens_per_second": 3475194.1174103804, "eta_seconds": 15490.902096748352, "data_wait_seconds": 1.990795135498047e-05} {"step": 2950, "total_steps": 28610, "tokens": 6186598400, "source": "fineweb_edu", "phase": 1, "loss": 3.4984481930732727, "ce_loss": 3.4829061031341553, "logit_z_loss": 11.984048843383789, "router_aux_loss": 1.365947499871254, "router_z_loss": 0.6842140257358551, "routing_entropy_normalized": 0.661984178454123, "learning_rate": 0.003, "grad_norm": 0.3246985673904419, "tokens_per_second": 3537154.2729477594, "eta_seconds": 15213.619810581207, "data_wait_seconds": 1.7642974853515625e-05} {"step": 2960, "total_steps": 28610, "tokens": 6207569920, "source": "fineweb_hq", "phase": 1, "loss": 3.4320590794086456, "ce_loss": 3.417109727859497, "logit_z_loss": 11.907318830490112, "router_aux_loss": 1.3141092360019684, "router_z_loss": 0.6175488904118538, "routing_entropy_normalized": 0.6847649429689935, "learning_rate": 0.003, "grad_norm": 0.26589298248291016, "tokens_per_second": 4164140.2962463815, "eta_seconds": 12917.900208234787, "data_wait_seconds": 1.5795230865478516e-05} {"step": 2970, "total_steps": 28610, "tokens": 6228541440, "source": "fineweb_edu", "phase": 1, "loss": 3.4937112033367157, "ce_loss": 3.4786535501480103, "logit_z_loss": 11.858895421028137, "router_aux_loss": 1.3166724294424057, "router_z_loss": 0.7050961852073669, "routing_entropy_normalized": 0.6819486589931372, "learning_rate": 0.003, "grad_norm": 0.26899445056915283, "tokens_per_second": 3910536.192686921, "eta_seconds": 13750.282475471497, "data_wait_seconds": 1.7374753952026367e-05} {"step": 2980, "total_steps": 28610, "tokens": 6249512960, "source": "fineweb_hq", "phase": 1, "loss": 3.4144107699394226, "ce_loss": 3.3996242582798004, "logit_z_loss": 12.018981695175171, "router_aux_loss": 1.3097230941057205, "router_z_loss": 0.4874323792755604, "routing_entropy_normalized": 0.7095405348122554, "learning_rate": 0.003, "grad_norm": 0.23246270418167114, "tokens_per_second": 3432573.0418757387, "eta_seconds": 15658.809034585953, "data_wait_seconds": 2.0503997802734375e-05} {"step": 2990, "total_steps": 28610, "tokens": 6270484480, "source": "fineweb_edu", "phase": 1, "loss": 3.502181589603424, "ce_loss": 3.487288385629654, "logit_z_loss": 11.821194648742676, "router_aux_loss": 1.317949578166008, "router_z_loss": 0.5316130109131336, "routing_entropy_normalized": 0.6952987344187609, "learning_rate": 0.003, "grad_norm": 0.29194968938827515, "tokens_per_second": 3768259.3915387876, "eta_seconds": 14258.316282749176, "data_wait_seconds": 1.7017126083374023e-05} {"step": 3000, "total_steps": 28610, "tokens": 6291456000, "source": "fineweb_hq", "phase": 1, "loss": 3.389549970626831, "ce_loss": 3.3747940957546234, "logit_z_loss": 12.057837963104248, "router_aux_loss": 1.304056093096733, "router_z_loss": 0.50958351790905, "routing_entropy_normalized": 0.698362666101397, "learning_rate": 0.003, "grad_norm": 0.20654796063899994, "tokens_per_second": 4095411.1869508433, "eta_seconds": 13114.205208778381, "data_wait_seconds": 1.6957521438598633e-05} {"step": 3010, "total_steps": 28610, "tokens": 6312427520, "source": "fineweb_hq", "phase": 1, "loss": 3.448971301317215, "ce_loss": 3.4343788623809814, "logit_z_loss": 12.270244240760803, "router_aux_loss": 1.2900447994470596, "router_z_loss": 0.4649842269718647, "routing_entropy_normalized": 0.6805109809821667, "learning_rate": 0.003, "grad_norm": 0.2717195451259613, "tokens_per_second": 4043381.504597321, "eta_seconds": 13277.770385742188, "data_wait_seconds": 2.154707908630371e-05} {"step": 3020, "total_steps": 28610, "tokens": 6333399040, "source": "fineweb_hq", "phase": 1, "loss": 3.400375157594681, "ce_loss": 3.38599556684494, "logit_z_loss": 12.00335419178009, "router_aux_loss": 1.2748836278915405, "router_z_loss": 0.4304117523133755, "routing_entropy_normalized": 0.6982067867659025, "learning_rate": 0.003, "grad_norm": 0.2827951908111572, "tokens_per_second": 3581801.279323143, "eta_seconds": 14982.997518539429, "data_wait_seconds": 2.1904706954956055e-05} {"step": 3030, "total_steps": 28610, "tokens": 6354370560, "source": "fineweb_hq", "phase": 1, "loss": 3.3574829399585724, "ce_loss": 3.342767655849457, "logit_z_loss": 12.071430802345276, "router_aux_loss": 1.2989126145839691, "router_z_loss": 0.518996424973011, "routing_entropy_normalized": 0.6714326197617491, "learning_rate": 0.003, "grad_norm": 0.23734666407108307, "tokens_per_second": 3535120.5741110137, "eta_seconds": 15174.91328382492, "data_wait_seconds": 1.749396324157715e-05} {"step": 3040, "total_steps": 28610, "tokens": 6375342080, "source": "fineweb_hq", "phase": 1, "loss": 3.455192059278488, "ce_loss": 3.4399949610233307, "logit_z_loss": 12.150108098983765, "router_aux_loss": 1.3563098907470703, "router_z_loss": 0.4190065488219261, "routing_entropy_normalized": 0.65790587912655, "learning_rate": 0.003, "grad_norm": 0.22390107810497284, "tokens_per_second": 4191674.2498385846, "eta_seconds": 12793.020984888077, "data_wait_seconds": 1.58846378326416e-05} {"step": 3050, "total_steps": 28610, "tokens": 6396313600, "source": "fineweb_edu", "phase": 1, "loss": 3.509319931268692, "ce_loss": 3.4940772354602814, "logit_z_loss": 12.241509675979614, "router_aux_loss": 1.3486768454313278, "router_z_loss": 0.5318359136581421, "routing_entropy_normalized": 0.6603148107315575, "learning_rate": 0.003, "grad_norm": 0.304790198802948, "tokens_per_second": 3905346.2684682277, "eta_seconds": 13725.595999717712, "data_wait_seconds": 1.8656253814697266e-05} {"step": 3060, "total_steps": 28610, "tokens": 6417285120, "source": "fineweb_hq", "phase": 1, "loss": 3.395656853914261, "ce_loss": 3.3805142641067505, "logit_z_loss": 12.06037425994873, "router_aux_loss": 1.3446637392044067, "router_z_loss": 0.48998501896858215, "routing_entropy_normalized": 0.6630289760483968, "learning_rate": 0.003, "grad_norm": 0.20347407460212708, "tokens_per_second": 3320099.260031833, "eta_seconds": 16138.744478225708, "data_wait_seconds": 1.6629695892333984e-05} {"step": 3070, "total_steps": 28610, "tokens": 6438256640, "source": "fineweb_edu", "phase": 1, "loss": 3.4642019867897034, "ce_loss": 3.4493943750858307, "logit_z_loss": 11.948176980018616, "router_aux_loss": 1.306759998202324, "router_z_loss": 0.5452559441328049, "routing_entropy_normalized": 0.6806579498482156, "learning_rate": 0.003, "grad_norm": 0.29294493794441223, "tokens_per_second": 3923288.0509122866, "eta_seconds": 13652.13601064682, "data_wait_seconds": 1.609325408935547e-05} {"step": 3080, "total_steps": 28610, "tokens": 6459228160, "source": "fineweb_hq", "phase": 1, "loss": 3.4445130228996277, "ce_loss": 3.4296208024024963, "logit_z_loss": 12.063138961791992, "router_aux_loss": 1.3173577934503555, "router_z_loss": 0.5123317763209343, "routing_entropy_normalized": 0.6861681795811433, "learning_rate": 0.003, "grad_norm": 0.21609161794185638, "tokens_per_second": 4284141.679980436, "eta_seconds": 12497.320247411728, "data_wait_seconds": 1.9490718841552734e-05} {"step": 3090, "total_steps": 28610, "tokens": 6480199680, "source": "fineweb_edu", "phase": 1, "loss": 3.456703871488571, "ce_loss": 3.44230517745018, "logit_z_loss": 12.140820026397705, "router_aux_loss": 1.2680173516273499, "router_z_loss": 0.5044100694358349, "routing_entropy_normalized": 0.6772580789805396, "learning_rate": 0.003, "grad_norm": 0.37596383690834045, "tokens_per_second": 3859201.45569154, "eta_seconds": 13867.977521896362, "data_wait_seconds": 2.0116567611694336e-05} {"step": 3100, "total_steps": 28610, "tokens": 6501171200, "source": "fineweb_hq", "phase": 1, "loss": 3.4173998832702637, "ce_loss": 3.40278360247612, "logit_z_loss": 11.998412251472473, "router_aux_loss": 1.2987026125192642, "router_z_loss": 0.4294096902012825, "routing_entropy_normalized": 0.6680577909852886, "learning_rate": 0.003, "grad_norm": 0.2555176317691803, "tokens_per_second": 4365001.268997057, "eta_seconds": 12256.20434522629, "data_wait_seconds": 2.2202730178833008e-05} {"step": 3110, "total_steps": 28610, "tokens": 6522142720, "source": "fineweb_hq", "phase": 1, "loss": 3.3914807438850403, "ce_loss": 3.376875191926956, "logit_z_loss": 12.281211376190186, "router_aux_loss": 1.2950717061758041, "router_z_loss": 0.42673010379076004, "routing_entropy_normalized": 0.6817411603074779, "learning_rate": 0.003, "grad_norm": 0.25898757576942444, "tokens_per_second": 3655902.2026611078, "eta_seconds": 14627.682316303253, "data_wait_seconds": 1.7702579498291016e-05} {"step": 3120, "total_steps": 28610, "tokens": 6543114240, "source": "fineweb_hq", "phase": 1, "loss": 3.4193719029426575, "ce_loss": 3.404710292816162, "logit_z_loss": 11.92390525341034, "router_aux_loss": 1.3035246282815933, "router_z_loss": 0.4340025894343853, "routing_entropy_normalized": 0.6557939881271149, "learning_rate": 0.003, "grad_norm": 0.2898768484592438, "tokens_per_second": 3851944.07357258, "eta_seconds": 13877.772745132446, "data_wait_seconds": 2.0176172256469727e-05} {"step": 3130, "total_steps": 28610, "tokens": 6564085760, "source": "fineweb_hq", "phase": 1, "loss": 3.4005689918994904, "ce_loss": 3.3856504261493683, "logit_z_loss": 12.21577513217926, "router_aux_loss": 1.3275159746408463, "router_z_loss": 0.42192304134368896, "routing_entropy_normalized": 0.6818728199630413, "learning_rate": 0.003, "grad_norm": 0.21881960332393646, "tokens_per_second": 3547249.673410483, "eta_seconds": 15063.90524482727, "data_wait_seconds": 2.1845102310180664e-05} {"step": 3140, "total_steps": 28610, "tokens": 6585057280, "source": "fineweb_hq", "phase": 1, "loss": 3.4124839305877686, "ce_loss": 3.397491008043289, "logit_z_loss": 11.991217494010925, "router_aux_loss": 1.339319869875908, "router_z_loss": 0.40059560164809227, "routing_entropy_normalized": 0.6635164682490847, "learning_rate": 0.003, "grad_norm": 0.3107924461364746, "tokens_per_second": 4204014.027287184, "eta_seconds": 12705.585921764374, "data_wait_seconds": 2.0295381546020508e-05} {"step": 3150, "total_steps": 28610, "tokens": 6606028800, "source": "fineweb_edu", "phase": 1, "loss": 3.4672518968582153, "ce_loss": 3.4523910582065582, "logit_z_loss": 11.995883822441101, "router_aux_loss": 1.3171546012163162, "router_z_loss": 0.48973122239112854, "routing_entropy_normalized": 0.6750445514433185, "learning_rate": 0.003, "grad_norm": 0.30897030234336853, "tokens_per_second": 3800043.3752513085, "eta_seconds": 14050.75801706314, "data_wait_seconds": 1.8417835235595703e-05} {"step": 3160, "total_steps": 28610, "tokens": 6627000320, "source": "fineweb_hq", "phase": 1, "loss": 3.4053288400173187, "ce_loss": 3.390397369861603, "logit_z_loss": 12.133641958236694, "router_aux_loss": 1.3243824690580368, "router_z_loss": 0.4743029773235321, "routing_entropy_normalized": 0.6733260071598649, "learning_rate": 0.003, "grad_norm": 0.28395915031433105, "tokens_per_second": 3245953.3203223497, "eta_seconds": 16442.78679728508, "data_wait_seconds": 1.4275312423706055e-05} {"step": 3170, "total_steps": 28610, "tokens": 6647971840, "source": "fineweb_edu", "phase": 1, "loss": 3.4694721400737762, "ce_loss": 3.454561620950699, "logit_z_loss": 12.152238965034485, "router_aux_loss": 1.3146729320287704, "router_z_loss": 0.5485664382576942, "routing_entropy_normalized": 0.6787982334150827, "learning_rate": 0.003, "grad_norm": 0.27384689450263977, "tokens_per_second": 3456816.957703918, "eta_seconds": 15433.720539093018, "data_wait_seconds": 2.345442771911621e-05} {"step": 3180, "total_steps": 28610, "tokens": 6668943360, "source": "fineweb_hq", "phase": 1, "loss": 3.333243489265442, "ce_loss": 3.318036526441574, "logit_z_loss": 12.383806347846985, "router_aux_loss": 1.353862389922142, "router_z_loss": 0.4299076795578003, "routing_entropy_normalized": 0.6862126888697408, "learning_rate": 0.003, "grad_norm": 0.2277742624282837, "tokens_per_second": 3560341.862955488, "eta_seconds": 14979.06027364731, "data_wait_seconds": 2.0682811737060547e-05} {"step": 3190, "total_steps": 28610, "tokens": 6689914880, "source": "fineweb_edu", "phase": 1, "loss": 3.4723072350025177, "ce_loss": 3.457251399755478, "logit_z_loss": 11.911287665367126, "router_aux_loss": 1.3383620977401733, "router_z_loss": 0.48113203048706055, "routing_entropy_normalized": 0.65791769116346, "learning_rate": 0.003, "grad_norm": 0.3076157569885254, "tokens_per_second": 3266309.595346917, "eta_seconds": 16321.050495624544, "data_wait_seconds": 1.901388168334961e-05} {"step": 3200, "total_steps": 28610, "tokens": 6710886400, "source": "fineweb_hq", "phase": 1, "loss": 3.2830206751823425, "ce_loss": 3.268372416496277, "logit_z_loss": 12.495417475700378, "router_aux_loss": 1.291395753622055, "router_z_loss": 0.48470306769013405, "routing_entropy_normalized": 0.682646986429671, "learning_rate": 0.003, "grad_norm": 0.22153997421264648, "tokens_per_second": 4372295.874550832, "eta_seconds": 12187.791917324068, "data_wait_seconds": 1.7940998077392578e-05} {"step": 3210, "total_steps": 28610, "tokens": 6731857920, "source": "fineweb_hq", "phase": 1, "loss": 3.2625003457069397, "ce_loss": 3.2474978864192963, "logit_z_loss": 12.281836986541748, "router_aux_loss": 1.3357052654027939, "router_z_loss": 0.4172411821782589, "routing_entropy_normalized": 0.6360389926968784, "learning_rate": 0.003, "grad_norm": 0.31427201628685, "tokens_per_second": 3769336.7790980744, "eta_seconds": 14131.839079856873, "data_wait_seconds": 2.2232532501220703e-05} {"step": 3220, "total_steps": 28610, "tokens": 6752829440, "source": "fineweb_hq", "phase": 1, "loss": 3.302716940641403, "ce_loss": 3.2875256538391113, "logit_z_loss": 12.22315764427185, "router_aux_loss": 1.3436024487018585, "router_z_loss": 0.5329640209674835, "routing_entropy_normalized": 0.6811412271129166, "learning_rate": 0.003, "grad_norm": 0.24159561097621918, "tokens_per_second": 3924729.6142472425, "eta_seconds": 13566.970088005066, "data_wait_seconds": 2.0802021026611328e-05} {"step": 3230, "total_steps": 28610, "tokens": 6773800960, "source": "fineweb_hq", "phase": 1, "loss": 3.300919860601425, "ce_loss": 3.2851415872573853, "logit_z_loss": 12.390978813171387, "router_aux_loss": 1.399793267250061, "router_z_loss": 0.5412731841206551, "routing_entropy_normalized": 0.6452317466851889, "learning_rate": 0.003, "grad_norm": 0.22603251039981842, "tokens_per_second": 4344605.903419286, "eta_seconds": 12250.988684177399, "data_wait_seconds": 2.1278858184814453e-05} {"step": 3240, "total_steps": 28610, "tokens": 6794772480, "source": "fineweb_hq", "phase": 1, "loss": 3.30567529797554, "ce_loss": 3.2901075184345245, "logit_z_loss": 12.315043210983276, "router_aux_loss": 1.3771686404943466, "router_z_loss": 0.5645450875163078, "routing_entropy_normalized": 0.6585818296447121, "learning_rate": 0.003, "grad_norm": 0.27110734581947327, "tokens_per_second": 3594444.9096083897, "eta_seconds": 14801.936760187149, "data_wait_seconds": 2.1725893020629883e-05} {"step": 3250, "total_steps": 28610, "tokens": 6815744000, "source": "fineweb_edu", "phase": 1, "loss": 3.4618460834026337, "ce_loss": 3.445973813533783, "logit_z_loss": 12.098281621932983, "router_aux_loss": 1.403958335518837, "router_z_loss": 0.6228179484605789, "routing_entropy_normalized": 0.6670771315550862, "learning_rate": 0.003, "grad_norm": 0.3071897327899933, "tokens_per_second": 4317400.915915432, "eta_seconds": 12318.470245361328, "data_wait_seconds": 1.800060272216797e-05} {"step": 3260, "total_steps": 28610, "tokens": 6836715520, "source": "fineweb_hq", "phase": 1, "loss": 3.2715944051742554, "ce_loss": 3.255695402622223, "logit_z_loss": 12.566628336906433, "router_aux_loss": 1.401849091053009, "router_z_loss": 0.6238541528582573, "routing_entropy_normalized": 0.6520972223425224, "learning_rate": 0.003, "grad_norm": 0.2620771527290344, "tokens_per_second": 3331642.792055305, "eta_seconds": 15956.93371653557, "data_wait_seconds": 1.9252300262451172e-05} {"step": 3270, "total_steps": 28610, "tokens": 6857687040, "source": "fineweb_edu", "phase": 1, "loss": 3.4419982731342316, "ce_loss": 3.426571696996689, "logit_z_loss": 12.18402087688446, "router_aux_loss": 1.3469585925340652, "router_z_loss": 0.7385104149580002, "routing_entropy_normalized": 0.6573834371425435, "learning_rate": 0.003, "grad_norm": 0.3155563175678253, "tokens_per_second": 4290961.700014605, "eta_seconds": 12384.5970659256, "data_wait_seconds": 1.8924474716186523e-05} {"step": 3280, "total_steps": 28610, "tokens": 6878658560, "source": "fineweb_hq", "phase": 1, "loss": 3.313882529735565, "ce_loss": 3.2987549901008606, "logit_z_loss": 12.465559124946594, "router_aux_loss": 1.3254351913928986, "router_z_loss": 0.6266039162874222, "routing_entropy_normalized": 0.6758453880330983, "learning_rate": 0.003, "grad_norm": 0.26776546239852905, "tokens_per_second": 3870396.1746880175, "eta_seconds": 13724.915425300598, "data_wait_seconds": 1.9490718841552734e-05} {"step": 3290, "total_steps": 28610, "tokens": 6899630080, "source": "fineweb_edu", "phase": 1, "loss": 3.4344679415225983, "ce_loss": 3.4191868007183075, "logit_z_loss": 12.208717107772827, "router_aux_loss": 1.3411850035190582, "router_z_loss": 0.6484327465295792, "routing_entropy_normalized": 0.6689447403067167, "learning_rate": 0.003, "grad_norm": 0.30366051197052, "tokens_per_second": 3609406.377193914, "eta_seconds": 14711.529567718506, "data_wait_seconds": 1.5079975128173828e-05} {"step": 3300, "total_steps": 28610, "tokens": 6920601600, "source": "fineweb_hq", "phase": 1, "loss": 3.2831621766090393, "ce_loss": 3.267583727836609, "logit_z_loss": 12.609729409217834, "router_aux_loss": 1.375562310218811, "router_z_loss": 0.5618442930281162, "routing_entropy_normalized": 0.6703489288333879, "learning_rate": 0.003, "grad_norm": 0.30531221628189087, "tokens_per_second": 3749640.9710734203, "eta_seconds": 14155.733183383942, "data_wait_seconds": 1.8537044525146484e-05} {"step": 3310, "total_steps": 28610, "tokens": 6941573120, "source": "fineweb_hq", "phase": 1, "loss": 3.313981384038925, "ce_loss": 3.298578143119812, "logit_z_loss": 12.526632189750671, "router_aux_loss": 1.349814161658287, "router_z_loss": 0.6524359062314034, "routing_entropy_normalized": 0.6613015564369046, "learning_rate": 0.003, "grad_norm": 0.261842280626297, "tokens_per_second": 3530367.5919085923, "eta_seconds": 15029.014463424683, "data_wait_seconds": 2.384185791015625e-05} {"step": 3320, "total_steps": 28610, "tokens": 6962544640, "source": "fineweb_hq", "phase": 1, "loss": 3.2930699586868286, "ce_loss": 3.2779800593852997, "logit_z_loss": 12.334709405899048, "router_aux_loss": 1.3282312899827957, "router_z_loss": 0.5741532817482948, "routing_entropy_normalized": 0.6729266199315829, "learning_rate": 0.003, "grad_norm": 0.24674628674983978, "tokens_per_second": 4143366.024507308, "eta_seconds": 12800.455901384354, "data_wait_seconds": 1.8596649169921875e-05} {"step": 3330, "total_steps": 28610, "tokens": 6983516160, "source": "fineweb_hq", "phase": 1, "loss": 3.275129020214081, "ce_loss": 3.2600331008434296, "logit_z_loss": 12.36552631855011, "router_aux_loss": 1.3311477899551392, "router_z_loss": 0.5478417761623859, "routing_entropy_normalized": 0.6810663698878148, "learning_rate": 0.003, "grad_norm": 0.30106064677238464, "tokens_per_second": 3518490.689941687, "eta_seconds": 15067.825164794922, "data_wait_seconds": 1.5407800674438477e-05} {"step": 3340, "total_steps": 28610, "tokens": 7004487680, "source": "fineweb_hq", "phase": 1, "loss": 3.282475233078003, "ce_loss": 3.267526537179947, "logit_z_loss": 12.37724244594574, "router_aux_loss": 1.3176981806755066, "router_z_loss": 0.5339868068695068, "routing_entropy_normalized": 0.6627943551972483, "learning_rate": 0.003, "grad_norm": 0.2520885169506073, "tokens_per_second": 3837247.3870296828, "eta_seconds": 13810.689198493958, "data_wait_seconds": 2.1666288375854492e-05} {"step": 3350, "total_steps": 28610, "tokens": 7025459200, "source": "fineweb_edu", "phase": 1, "loss": 3.4569546580314636, "ce_loss": 3.4416123926639557, "logit_z_loss": 12.270003914833069, "router_aux_loss": 1.3524002879858017, "router_z_loss": 0.5911851413547993, "routing_entropy_normalized": 0.6507684168893377, "learning_rate": 0.003, "grad_norm": 0.30153486132621765, "tokens_per_second": 3821636.3600297906, "eta_seconds": 13861.616995811462, "data_wait_seconds": 1.8358230590820312e-05} {"step": 3360, "total_steps": 28610, "tokens": 7046430720, "source": "fineweb_hq", "phase": 1, "loss": 3.3040192425251007, "ce_loss": 3.2884626388549805, "logit_z_loss": 12.304916381835938, "router_aux_loss": 1.3698224425315857, "router_z_loss": 0.6279220804572105, "routing_entropy_normalized": 0.6665144768952614, "learning_rate": 0.003, "grad_norm": 0.19891023635864258, "tokens_per_second": 3509281.633472735, "eta_seconds": 15089.438104629517, "data_wait_seconds": 2.1189451217651367e-05} {"step": 3370, "total_steps": 28610, "tokens": 7067402240, "source": "fineweb_edu", "phase": 1, "loss": 3.362433433532715, "ce_loss": 3.346982926130295, "logit_z_loss": 12.07545292377472, "router_aux_loss": 1.3594146221876144, "router_z_loss": 0.6488073542714119, "routing_entropy_normalized": 0.6594060072932943, "learning_rate": 0.003, "grad_norm": 0.261669397354126, "tokens_per_second": 3604095.752636102, "eta_seconds": 14686.656546592712, "data_wait_seconds": 1.868605613708496e-05} {"step": 3380, "total_steps": 28610, "tokens": 7088373760, "source": "fineweb_hq", "phase": 1, "loss": 3.299377977848053, "ce_loss": 3.2845942974090576, "logit_z_loss": 12.108551263809204, "router_aux_loss": 1.2946936786174774, "router_z_loss": 0.6259484142065048, "routing_entropy_normalized": 0.6949218232207459, "learning_rate": 0.003, "grad_norm": 0.3393915593624115, "tokens_per_second": 4304044.458512308, "eta_seconds": 12293.354650497437, "data_wait_seconds": 1.9490718841552734e-05} {"step": 3390, "total_steps": 28610, "tokens": 7109345280, "source": "fineweb_edu", "phase": 1, "loss": 3.4214872121810913, "ce_loss": 3.4066604673862457, "logit_z_loss": 12.223353266716003, "router_aux_loss": 1.301958054304123, "router_z_loss": 0.5848522409796715, "routing_entropy_normalized": 0.6694812658223998, "learning_rate": 0.003, "grad_norm": 0.23977862298488617, "tokens_per_second": 3359129.677725925, "eta_seconds": 15745.201440334322, "data_wait_seconds": 1.7642974853515625e-05} {"step": 3400, "total_steps": 28610, "tokens": 7130316800, "source": "fineweb_hq", "phase": 1, "loss": 3.27957421541214, "ce_loss": 3.265010505914688, "logit_z_loss": 12.255654215812683, "router_aux_loss": 1.2819713950157166, "router_z_loss": 0.5184873342514038, "routing_entropy_normalized": 0.6601270872758974, "learning_rate": 0.003, "grad_norm": 0.35029834508895874, "tokens_per_second": 3709181.574623698, "eta_seconds": 14253.60308098793, "data_wait_seconds": 1.4513731002807617e-05} {"step": 3410, "total_steps": 28610, "tokens": 7151288320, "source": "fineweb_hq", "phase": 1, "loss": 3.227551370859146, "ce_loss": 3.2127586901187897, "logit_z_loss": 12.233355522155762, "router_aux_loss": 1.305265799164772, "router_z_loss": 0.5167430825531483, "routing_entropy_normalized": 0.6531377856329629, "learning_rate": 0.003, "grad_norm": 0.23986612260341644, "tokens_per_second": 3859614.6377048856, "eta_seconds": 13692.618398666382, "data_wait_seconds": 1.919269561767578e-05} {"step": 3420, "total_steps": 28610, "tokens": 7172259840, "source": "fineweb_hq", "phase": 1, "loss": 3.2337149381637573, "ce_loss": 3.2188015282154083, "logit_z_loss": 12.314534902572632, "router_aux_loss": 1.306317612528801, "router_z_loss": 0.6187973544001579, "routing_entropy_normalized": 0.6659057297501624, "learning_rate": 0.003, "grad_norm": 0.3172547519207001, "tokens_per_second": 4059998.4538419154, "eta_seconds": 13011.645073413849, "data_wait_seconds": 1.9818544387817383e-05} {"step": 3430, "total_steps": 28610, "tokens": 7193231360, "source": "fineweb_hq", "phase": 1, "loss": 3.2311160266399384, "ce_loss": 3.215982049703598, "logit_z_loss": 12.387041330337524, "router_aux_loss": 1.3379499167203903, "router_z_loss": 0.5157605148851871, "routing_entropy_normalized": 0.683587964001254, "learning_rate": 0.003, "grad_norm": 0.274903267621994, "tokens_per_second": 2579290.278136685, "eta_seconds": 20473.185126781464, "data_wait_seconds": 1.8596649169921875e-05} {"step": 3440, "total_steps": 28610, "tokens": 7214202880, "source": "fineweb_hq", "phase": 1, "loss": 3.334943473339081, "ce_loss": 3.3196409940719604, "logit_z_loss": 12.36742079257965, "router_aux_loss": 1.3546610474586487, "router_z_loss": 0.5191479250788689, "routing_entropy_normalized": 0.649472187823111, "learning_rate": 0.003, "grad_norm": 0.27886465191841125, "tokens_per_second": 3829174.5722546848, "eta_seconds": 13785.037700414658, "data_wait_seconds": 0.06378453969955444} {"step": 3450, "total_steps": 28610, "tokens": 7235174400, "source": "fineweb_edu", "phase": 1, "loss": 3.4117897152900696, "ce_loss": 3.3967685401439667, "logit_z_loss": 12.014654755592346, "router_aux_loss": 1.3255020081996918, "router_z_loss": 0.5645986832678318, "routing_entropy_normalized": 0.6385904799420165, "learning_rate": 0.003, "grad_norm": 0.36398032307624817, "tokens_per_second": 3758505.268264536, "eta_seconds": 14038.651153564453, "data_wait_seconds": 1.722574234008789e-05} {"step": 3460, "total_steps": 28610, "tokens": 7256145920, "source": "fineweb_hq", "phase": 1, "loss": 3.388280838727951, "ce_loss": 3.3733702898025513, "logit_z_loss": 12.151432275772095, "router_aux_loss": 1.3181776255369186, "router_z_loss": 0.513569675385952, "routing_entropy_normalized": 0.6749461783245893, "learning_rate": 0.003, "grad_norm": 0.33292436599731445, "tokens_per_second": 4010978.465572548, "eta_seconds": 13149.75217461586, "data_wait_seconds": 2.0503997802734375e-05} {"step": 3470, "total_steps": 28610, "tokens": 7277117440, "source": "fineweb_edu", "phase": 1, "loss": 3.407367020845413, "ce_loss": 3.392653077840805, "logit_z_loss": 11.74485993385315, "router_aux_loss": 1.2947678864002228, "router_z_loss": 0.5917507261037827, "routing_entropy_normalized": 0.6585474735017306, "learning_rate": 0.003, "grad_norm": 0.29136037826538086, "tokens_per_second": 4124130.0419760356, "eta_seconds": 12783.884296417236, "data_wait_seconds": 1.856684684753418e-05} {"step": 3480, "total_steps": 28610, "tokens": 7298088960, "source": "fineweb_hq", "phase": 1, "loss": 3.430584192276001, "ce_loss": 3.415415495634079, "logit_z_loss": 12.032287955284119, "router_aux_loss": 1.3429497480392456, "router_z_loss": 0.5359430275857449, "routing_entropy_normalized": 0.6485694246682019, "learning_rate": 0.003, "grad_norm": 0.35828685760498047, "tokens_per_second": 4066968.062158298, "eta_seconds": 12958.40757894516, "data_wait_seconds": 1.874566078186035e-05} {"step": 3490, "total_steps": 28610, "tokens": 7319060480, "source": "fineweb_edu", "phase": 1, "loss": 3.422822207212448, "ce_loss": 3.4076591432094574, "logit_z_loss": 12.142119407653809, "router_aux_loss": 1.3414268493652344, "router_z_loss": 0.5345388501882553, "routing_entropy_normalized": 0.6492474359190418, "learning_rate": 0.003, "grad_norm": 0.2837613821029663, "tokens_per_second": 3752813.469867287, "eta_seconds": 14037.59037399292, "data_wait_seconds": 1.874566078186035e-05} {"step": 3500, "total_steps": 28610, "tokens": 7340032000, "source": "fineweb_hq", "phase": 1, "loss": 3.429539769887924, "ce_loss": 3.4144648015499115, "logit_z_loss": 11.972415208816528, "router_aux_loss": 1.3351488709449768, "router_z_loss": 0.5262029580771923, "routing_entropy_normalized": 0.6625163757169931, "learning_rate": 0.003, "grad_norm": 0.39034175872802734, "tokens_per_second": 3391541.987126466, "eta_seconds": 15526.709361076355, "data_wait_seconds": 2.5510787963867188e-05} {"step": 3510, "total_steps": 28610, "tokens": 7361003520, "source": "fineweb_hq", "phase": 1, "loss": 3.4553703665733337, "ce_loss": 3.4398439526557922, "logit_z_loss": 12.223151326179504, "router_aux_loss": 1.372301772236824, "router_z_loss": 0.5811475068330765, "routing_entropy_normalized": 0.6481806167543568, "learning_rate": 0.003, "grad_norm": 0.3425040543079376, "tokens_per_second": 4069628.3196351104, "eta_seconds": 12934.477319717407, "data_wait_seconds": 1.9431114196777344e-05} {"step": 3520, "total_steps": 28610, "tokens": 7381975040, "source": "fineweb_hq", "phase": 1, "loss": 3.421787142753601, "ce_loss": 3.406142383813858, "logit_z_loss": 12.483772873878479, "router_aux_loss": 1.3758520483970642, "router_z_loss": 0.6378702372312546, "routing_entropy_normalized": 0.6771018095268592, "learning_rate": 0.003, "grad_norm": 0.292471319437027, "tokens_per_second": 3892103.2521276143, "eta_seconds": 13519.05133843422, "data_wait_seconds": 1.9997358322143555e-05} {"step": 3530, "total_steps": 28610, "tokens": 7402946560, "source": "fineweb_hq", "phase": 1, "loss": 3.4546515345573425, "ce_loss": 3.438920706510544, "logit_z_loss": 12.170831680297852, "router_aux_loss": 1.3947709202766418, "router_z_loss": 0.5660667605698109, "routing_entropy_normalized": 0.6639625211275527, "learning_rate": 0.003, "grad_norm": 0.2871759831905365, "tokens_per_second": 3957647.5994183095, "eta_seconds": 13289.857380867004, "data_wait_seconds": 2.053380012512207e-05} {"step": 3540, "total_steps": 28610, "tokens": 7423918080, "source": "fineweb_hq", "phase": 1, "loss": 3.4362482726573944, "ce_loss": 3.4204659163951874, "logit_z_loss": 12.191442608833313, "router_aux_loss": 1.3919918835163116, "router_z_loss": 0.6431840136647224, "routing_entropy_normalized": 0.6744979692169278, "learning_rate": 0.003, "grad_norm": 0.25868743658065796, "tokens_per_second": 4186168.211846426, "eta_seconds": 12559.361683368683, "data_wait_seconds": 2.1398067474365234e-05} {"step": 3550, "total_steps": 28610, "tokens": 7444889600, "source": "fineweb_edu", "phase": 1, "loss": 3.424156606197357, "ce_loss": 3.4093479812145233, "logit_z_loss": 12.121053457260132, "router_aux_loss": 1.2938823252916336, "router_z_loss": 0.6576758250594139, "routing_entropy_normalized": 0.6851239614375023, "learning_rate": 0.003, "grad_norm": 0.32846277952194214, "tokens_per_second": 3732073.327340695, "eta_seconds": 14081.885458946228, "data_wait_seconds": 2.0682811737060547e-05} {"step": 3560, "total_steps": 28610, "tokens": 7465861120, "source": "fineweb_hq", "phase": 1, "loss": 3.4234597980976105, "ce_loss": 3.408929318189621, "logit_z_loss": 12.154934048652649, "router_aux_loss": 1.2695553451776505, "router_z_loss": 0.6194202527403831, "routing_entropy_normalized": 0.6992853430735824, "learning_rate": 0.003, "grad_norm": 0.24979063868522644, "tokens_per_second": 4159382.5936634964, "eta_seconds": 12630.157581567764, "data_wait_seconds": 1.9490718841552734e-05} {"step": 3570, "total_steps": 28610, "tokens": 7486832640, "source": "fineweb_edu", "phase": 1, "loss": 3.423627555370331, "ce_loss": 3.4091614484786987, "logit_z_loss": 12.069854974746704, "router_aux_loss": 1.2699293941259384, "router_z_loss": 0.5597676113247871, "routing_entropy_normalized": 0.7028760668012606, "learning_rate": 0.003, "grad_norm": 0.30946850776672363, "tokens_per_second": 3778379.9238567417, "eta_seconds": 13898.20164680481, "data_wait_seconds": 1.9311904907226562e-05} {"step": 3580, "total_steps": 28610, "tokens": 7507804160, "source": "fineweb_hq", "phase": 1, "loss": 3.4253025352954865, "ce_loss": 3.4109885692596436, "logit_z_loss": 12.19970977306366, "router_aux_loss": 1.2568431198596954, "router_z_loss": 0.5255400538444519, "routing_entropy_normalized": 0.6845433262382598, "learning_rate": 0.003, "grad_norm": 0.23095086216926575, "tokens_per_second": 4197217.222021282, "eta_seconds": 12506.31353664398, "data_wait_seconds": 1.4901161193847656e-05} {"step": 3590, "total_steps": 28610, "tokens": 7528775680, "source": "fineweb_edu", "phase": 1, "loss": 3.407716304063797, "ce_loss": 3.393492341041565, "logit_z_loss": 12.332146525382996, "router_aux_loss": 1.2517801076173782, "router_z_loss": 0.47293299064040184, "routing_entropy_normalized": 0.661835881701062, "learning_rate": 0.003, "grad_norm": 0.3669501543045044, "tokens_per_second": 4015174.15871465, "eta_seconds": 13068.111361026764, "data_wait_seconds": 1.7583370208740234e-05} {"step": 3600, "total_steps": 28610, "tokens": 7549747200, "source": "fineweb_hq", "phase": 1, "loss": 3.457605481147766, "ce_loss": 3.443107694387436, "logit_z_loss": 12.077750325202942, "router_aux_loss": 1.2849017083644867, "router_z_loss": 0.44094473868608475, "routing_entropy_normalized": 0.665123082165943, "learning_rate": 0.003, "grad_norm": 0.39565011858940125, "tokens_per_second": 3751565.971015578, "eta_seconds": 13980.767478227615, "data_wait_seconds": 2.3692846298217773e-05} {"step": 3610, "total_steps": 28610, "tokens": 7570718720, "source": "fineweb_hq", "phase": 1, "loss": 3.4001887440681458, "ce_loss": 3.3856745660305023, "logit_z_loss": 12.377567529678345, "router_aux_loss": 1.2805209010839462, "router_z_loss": 0.47125403583049774, "routing_entropy_normalized": 0.680162251953891, "learning_rate": 0.003, "grad_norm": 0.2464335411787033, "tokens_per_second": 3912197.718410494, "eta_seconds": 13401.367664337158, "data_wait_seconds": 2.288818359375e-05} {"step": 3620, "total_steps": 28610, "tokens": 7591690240, "source": "fineweb_hq", "phase": 1, "loss": 3.421826660633087, "ce_loss": 3.407147765159607, "logit_z_loss": 11.981026887893677, "router_aux_loss": 1.2991336286067963, "router_z_loss": 0.4894500970840454, "routing_entropy_normalized": 0.6702709479044809, "learning_rate": 0.003, "grad_norm": 0.2879997491836548, "tokens_per_second": 3872777.7870663973, "eta_seconds": 13532.361359596252, "data_wait_seconds": 1.806020736694336e-05} {"step": 3630, "total_steps": 28610, "tokens": 7612661760, "source": "fineweb_hq", "phase": 1, "loss": 3.4163836240768433, "ce_loss": 3.4017744958400726, "logit_z_loss": 12.15055501461029, "router_aux_loss": 1.2934375256299973, "router_z_loss": 0.45965272188186646, "routing_entropy_normalized": 0.6756952723226644, "learning_rate": 0.003, "grad_norm": 0.26462316513061523, "tokens_per_second": 3780863.9510332574, "eta_seconds": 13855.790009498596, "data_wait_seconds": 2.2143125534057617e-05} {"step": 3640, "total_steps": 28610, "tokens": 7633633280, "source": "fineweb_hq", "phase": 1, "loss": 3.455957442522049, "ce_loss": 3.4415416717529297, "logit_z_loss": 12.309085130691528, "router_aux_loss": 1.2707851082086563, "router_z_loss": 0.47697319462895393, "routing_entropy_normalized": 0.6754702233043031, "learning_rate": 0.003, "grad_norm": 0.38108518719673157, "tokens_per_second": 4195901.007839421, "eta_seconds": 12480.248066425323, "data_wait_seconds": 2.0712614059448242e-05} {"step": 3650, "total_steps": 28610, "tokens": 7654604800, "source": "fineweb_edu", "phase": 1, "loss": 3.4145489037036896, "ce_loss": 3.3998782634735107, "logit_z_loss": 12.196933507919312, "router_aux_loss": 1.2851045727729797, "router_z_loss": 0.5998697206377983, "routing_entropy_normalized": 0.6518562713202396, "learning_rate": 0.003, "grad_norm": 0.30450114607810974, "tokens_per_second": 4144819.587377113, "eta_seconds": 12628.996948242188, "data_wait_seconds": 2.104043960571289e-05} {"step": 3660, "total_steps": 28610, "tokens": 7675576320, "source": "fineweb_hq", "phase": 1, "loss": 3.459343522787094, "ce_loss": 3.444504588842392, "logit_z_loss": 12.268149971961975, "router_aux_loss": 1.3079355210065842, "router_z_loss": 0.5328339487314224, "routing_entropy_normalized": 0.6814126496098913, "learning_rate": 0.003, "grad_norm": 0.27359792590141296, "tokens_per_second": 2875589.605594957, "eta_seconds": 18195.90052008629, "data_wait_seconds": 0.23249992728233337} {"step": 3670, "total_steps": 28610, "tokens": 7696547840, "source": "fineweb_edu", "phase": 1, "loss": 3.3844085037708282, "ce_loss": 3.369448482990265, "logit_z_loss": 12.245637893676758, "router_aux_loss": 1.3167372345924377, "router_z_loss": 0.5680622830986977, "routing_entropy_normalized": 0.6532995700728346, "learning_rate": 0.003, "grad_norm": 0.2876167595386505, "tokens_per_second": 3719306.6983760004, "eta_seconds": 14062.559267520905, "data_wait_seconds": 1.9103288650512695e-05} {"step": 3680, "total_steps": 28610, "tokens": 7717519360, "source": "fineweb_hq", "phase": 1, "loss": 3.2199481427669525, "ce_loss": 3.2055412232875824, "logit_z_loss": 12.676668882369995, "router_aux_loss": 1.2655293494462967, "router_z_loss": 0.4839120879769325, "routing_entropy_normalized": 0.6783336992960666, "learning_rate": 0.003, "grad_norm": 0.2978517711162567, "tokens_per_second": 3720878.132423197, "eta_seconds": 14050.984068632126, "data_wait_seconds": 1.7970800399780273e-05} {"step": 3690, "total_steps": 28610, "tokens": 7738490880, "source": "fineweb_edu", "phase": 1, "loss": 3.379366308450699, "ce_loss": 3.3649756014347076, "logit_z_loss": 12.050870060920715, "router_aux_loss": 1.2705421596765518, "router_z_loss": 0.48013557866215706, "routing_entropy_normalized": 0.6598635814570084, "learning_rate": 0.003, "grad_norm": 0.3350239098072052, "tokens_per_second": 3253377.034081834, "eta_seconds": 16063.62474822998, "data_wait_seconds": 1.8924474716186523e-05} {"step": 3700, "total_steps": 28610, "tokens": 7759462400, "source": "fineweb_hq", "phase": 1, "loss": 3.1247510612010956, "ce_loss": 3.1099093556404114, "logit_z_loss": 12.772451043128967, "router_aux_loss": 1.3116062581539154, "router_z_loss": 0.448377076536417, "routing_entropy_normalized": 0.662598170924561, "learning_rate": 0.003, "grad_norm": 0.3675125241279602, "tokens_per_second": 3591952.6990274386, "eta_seconds": 14543.637040138245, "data_wait_seconds": 2.1696090698242188e-05} {"step": 3710, "total_steps": 28610, "tokens": 7780433920, "source": "fineweb_hq", "phase": 1, "loss": 3.1066622138023376, "ce_loss": 3.091491222381592, "logit_z_loss": 12.885786414146423, "router_aux_loss": 1.3440679758787155, "router_z_loss": 0.4417114183306694, "routing_entropy_normalized": 0.639215220078709, "learning_rate": 0.003, "grad_norm": 0.2785241901874542, "tokens_per_second": 3191569.317952957, "eta_seconds": 16361.569998264315, "data_wait_seconds": 1.9073486328125e-05} {"step": 3720, "total_steps": 28610, "tokens": 7801405440, "source": "fineweb_hq", "phase": 1, "loss": 3.129068374633789, "ce_loss": 3.1144229769706726, "logit_z_loss": 12.86177659034729, "router_aux_loss": 1.2849337756633759, "router_z_loss": 0.5098278746008873, "routing_entropy_normalized": 0.6652163408655732, "learning_rate": 0.003, "grad_norm": 0.3057716190814972, "tokens_per_second": 4273727.413318856, "eta_seconds": 12213.720771551132, "data_wait_seconds": 1.9222497940063477e-05} {"step": 3730, "total_steps": 28610, "tokens": 7822376960, "source": "fineweb_hq", "phase": 1, "loss": 3.1288401782512665, "ce_loss": 3.1140533685684204, "logit_z_loss": 13.03476881980896, "router_aux_loss": 1.2968384474515915, "router_z_loss": 0.5149211995303631, "routing_entropy_normalized": 0.6726312736215427, "learning_rate": 0.003, "grad_norm": 0.33152303099632263, "tokens_per_second": 3333718.9110436686, "eta_seconds": 15651.33208656311, "data_wait_seconds": 2.1457672119140625e-05} {"step": 3740, "total_steps": 28610, "tokens": 7843348480, "source": "fineweb_hq", "phase": 1, "loss": 3.1082300543785095, "ce_loss": 3.0927042961120605, "logit_z_loss": 12.893826246261597, "router_aux_loss": 1.3666896224021912, "router_z_loss": 0.5694564059376717, "routing_entropy_normalized": 0.6414492683168844, "learning_rate": 0.003, "grad_norm": 0.2614298164844513, "tokens_per_second": 3944995.177661266, "eta_seconds": 13220.8451192379, "data_wait_seconds": 1.7911195755004883e-05} {"step": 3750, "total_steps": 28610, "tokens": 7864320000, "source": "fineweb_edu", "phase": 1, "loss": 3.439316689968109, "ce_loss": 3.423677831888199, "logit_z_loss": 12.047470688819885, "router_aux_loss": 1.3784922808408737, "router_z_loss": 0.6491478532552719, "routing_entropy_normalized": 0.6665551782688248, "learning_rate": 0.003, "grad_norm": 0.33450162410736084, "tokens_per_second": 3733808.035654009, "eta_seconds": 13963.009941101074, "data_wait_seconds": 1.6808509826660156e-05} {"step": 3760, "total_steps": 28610, "tokens": 7885291520, "source": "fineweb_hq", "phase": 1, "loss": 3.1020804047584534, "ce_loss": 3.086145520210266, "logit_z_loss": 12.785008549690247, "router_aux_loss": 1.401362344622612, "router_z_loss": 0.6427356824278831, "routing_entropy_normalized": 0.6747902869674307, "learning_rate": 0.003, "grad_norm": 0.3816901445388794, "tokens_per_second": 3537804.993073883, "eta_seconds": 14730.666981935501, "data_wait_seconds": 1.6957521438598633e-05} {"step": 3770, "total_steps": 28610, "tokens": 7906263040, "source": "fineweb_edu", "phase": 1, "loss": 3.443901926279068, "ce_loss": 3.4278807938098907, "logit_z_loss": 11.723051309585571, "router_aux_loss": 1.411233052611351, "router_z_loss": 0.736481212079525, "routing_entropy_normalized": 0.6283799162748175, "learning_rate": 0.003, "grad_norm": 0.4117495119571686, "tokens_per_second": 3507686.570467107, "eta_seconds": 14851.171743392943, "data_wait_seconds": 2.0831823348999023e-05} {"step": 3780, "total_steps": 28610, "tokens": 7927234560, "source": "fineweb_hq", "phase": 1, "loss": 3.118405759334564, "ce_loss": 3.102675199508667, "logit_z_loss": 12.788623213768005, "router_aux_loss": 1.3611920177936554, "router_z_loss": 0.8398009315133095, "routing_entropy_normalized": 0.6668837433378607, "learning_rate": 0.003, "grad_norm": 0.3436398208141327, "tokens_per_second": 3785267.8563832236, "eta_seconds": 13756.56522488594, "data_wait_seconds": 2.3186206817626953e-05} {"step": 3790, "total_steps": 28610, "tokens": 7948206080, "source": "fineweb_edu", "phase": 1, "loss": 3.4620515406131744, "ce_loss": 3.446907192468643, "logit_z_loss": 12.212758541107178, "router_aux_loss": 1.3108056634664536, "router_z_loss": 0.8150141388177872, "routing_entropy_normalized": 0.6716681502419093, "learning_rate": 0.003, "grad_norm": 0.30317533016204834, "tokens_per_second": 3628703.0437169936, "eta_seconds": 14344.329644203186, "data_wait_seconds": 1.8775463104248047e-05} {"step": 3800, "total_steps": 28610, "tokens": 7969177600, "source": "fineweb_hq", "phase": 1, "loss": 3.092756748199463, "ce_loss": 3.077690601348877, "logit_z_loss": 12.896723508834839, "router_aux_loss": 1.3003453463315964, "router_z_loss": 0.772999495267868, "routing_entropy_normalized": 0.6678247221261747, "learning_rate": 0.003, "grad_norm": 0.3127208948135376, "tokens_per_second": 3944526.5439738757, "eta_seconds": 13190.516159534454, "data_wait_seconds": 1.8596649169921875e-05} {"step": 3810, "total_steps": 28610, "tokens": 7990149120, "source": "fineweb_hq", "phase": 1, "loss": 3.12148380279541, "ce_loss": 3.1063551008701324, "logit_z_loss": 12.959598183631897, "router_aux_loss": 1.3162585198879242, "router_z_loss": 0.670167975127697, "routing_entropy_normalized": 0.6523823199606311, "learning_rate": 0.003, "grad_norm": 0.26433855295181274, "tokens_per_second": 3986301.074489837, "eta_seconds": 13047.024955749512, "data_wait_seconds": 1.71661376953125e-05} {"step": 3820, "total_steps": 28610, "tokens": 8011120640, "source": "fineweb_hq", "phase": 1, "loss": 3.094204306602478, "ce_loss": 3.079040229320526, "logit_z_loss": 13.024366855621338, "router_aux_loss": 1.3114385157823563, "router_z_loss": 0.7471911311149597, "routing_entropy_normalized": 0.6466690047335644, "learning_rate": 0.003, "grad_norm": 0.3658689260482788, "tokens_per_second": 3850886.050005477, "eta_seconds": 13500.373006343842, "data_wait_seconds": 1.633167266845703e-05} {"step": 3830, "total_steps": 28610, "tokens": 8032092160, "source": "fineweb_hq", "phase": 1, "loss": 3.122576415538788, "ce_loss": 3.10707750916481, "logit_z_loss": 13.111273646354675, "router_aux_loss": 1.346071571111679, "router_z_loss": 0.727059543132782, "routing_entropy_normalized": 0.6477311120082573, "learning_rate": 0.003, "grad_norm": 0.37996602058410645, "tokens_per_second": 3631293.6017242107, "eta_seconds": 14310.995545864105, "data_wait_seconds": 1.7940998077392578e-05} {"step": 3840, "total_steps": 28610, "tokens": 8053063680, "source": "fineweb_hq", "phase": 1, "loss": 3.1806415617465973, "ce_loss": 3.165065288543701, "logit_z_loss": 12.846420764923096, "router_aux_loss": 1.3495770990848541, "router_z_loss": 0.7958463728427887, "routing_entropy_normalized": 0.6613929092243293, "learning_rate": 0.003, "grad_norm": 0.32732677459716797, "tokens_per_second": 3943690.745376569, "eta_seconds": 13172.04070854187, "data_wait_seconds": 2.0742416381835938e-05} {"step": 3850, "total_steps": 28610, "tokens": 8074035200, "source": "fineweb_edu", "phase": 1, "loss": 3.4561149179935455, "ce_loss": 3.4405291080474854, "logit_z_loss": 12.010998964309692, "router_aux_loss": 1.362177237868309, "router_z_loss": 0.7629233077168465, "routing_entropy_normalized": 0.6505866939219606, "learning_rate": 0.003, "grad_norm": 0.3241136074066162, "tokens_per_second": 4158806.390843385, "eta_seconds": 12485.669838905334, "data_wait_seconds": 1.8090009689331055e-05} {"step": 3860, "total_steps": 28610, "tokens": 8095006720, "source": "fineweb_hq", "phase": 1, "loss": 3.0874724686145782, "ce_loss": 3.0723788142204285, "logit_z_loss": 13.270498991012573, "router_aux_loss": 1.2991271764039993, "router_z_loss": 0.7753920778632164, "routing_entropy_normalized": 0.6826137635143481, "learning_rate": 0.003, "grad_norm": 0.3054397404193878, "tokens_per_second": 3330968.9443098633, "eta_seconds": 15582.406461238861, "data_wait_seconds": 1.5079975128173828e-05} {"step": 3870, "total_steps": 28610, "tokens": 8115978240, "source": "fineweb_edu", "phase": 1, "loss": 3.442238599061966, "ce_loss": 3.427120864391327, "logit_z_loss": 12.256668210029602, "router_aux_loss": 1.3125171512365341, "router_z_loss": 0.7669525593519211, "routing_entropy_normalized": 0.6378104550221999, "learning_rate": 0.003, "grad_norm": 0.27710241079330444, "tokens_per_second": 3357875.682116829, "eta_seconds": 15451.298794746399, "data_wait_seconds": 1.475214958190918e-05} {"step": 3880, "total_steps": 28610, "tokens": 8136949760, "source": "fineweb_hq", "phase": 1, "loss": 3.1041948199272156, "ce_loss": 3.08939853310585, "logit_z_loss": 13.114902138710022, "router_aux_loss": 1.2789258509874344, "router_z_loss": 0.695555292069912, "routing_entropy_normalized": 0.6578609420499706, "learning_rate": 0.003, "grad_norm": 0.30939361453056335, "tokens_per_second": 3937958.4794250783, "eta_seconds": 13169.912590742111, "data_wait_seconds": 2.0056962966918945e-05} {"step": 3890, "total_steps": 28610, "tokens": 8157921280, "source": "fineweb_edu", "phase": 1, "loss": 3.4479752480983734, "ce_loss": 3.432872325181961, "logit_z_loss": 12.099533796310425, "router_aux_loss": 1.3236216455698013, "router_z_loss": 0.6567884683609009, "routing_entropy_normalized": 0.621817615416056, "learning_rate": 0.003, "grad_norm": 0.4084160029888153, "tokens_per_second": 4115273.7087543616, "eta_seconds": 12597.363166809082, "data_wait_seconds": 1.9609928131103516e-05} {"step": 3900, "total_steps": 28610, "tokens": 8178892800, "source": "fineweb_hq", "phase": 1, "loss": 3.121298909187317, "ce_loss": 3.10612154006958, "logit_z_loss": 12.7145437002182, "router_aux_loss": 1.334985226392746, "router_z_loss": 0.5561171919107437, "routing_entropy_normalized": 0.6521999833907768, "learning_rate": 0.003, "grad_norm": 0.271346777677536, "tokens_per_second": 3390902.3861461435, "eta_seconds": 15282.252338409424, "data_wait_seconds": 2.0325183868408203e-05} {"step": 3910, "total_steps": 28610, "tokens": 8199864320, "source": "fineweb_hq", "phase": 1, "loss": 3.1032732129096985, "ce_loss": 3.0882830917835236, "logit_z_loss": 13.139787316322327, "router_aux_loss": 1.3072611838579178, "router_z_loss": 0.6035620793700218, "routing_entropy_normalized": 0.6513618425804962, "learning_rate": 0.003, "grad_norm": 0.2723662555217743, "tokens_per_second": 3322967.751387823, "eta_seconds": 15588.371081352234, "data_wait_seconds": 2.0623207092285156e-05} {"step": 3920, "total_steps": 28610, "tokens": 8220835840, "source": "fineweb_hq", "phase": 1, "loss": 3.096730262041092, "ce_loss": 3.0819399654865265, "logit_z_loss": 12.8056401014328, "router_aux_loss": 1.291149690747261, "router_z_loss": 0.5982127413153648, "routing_entropy_normalized": 0.6627614884936232, "learning_rate": 0.003, "grad_norm": 0.3160063624382019, "tokens_per_second": 3970966.843097139, "eta_seconds": 13039.313831090927, "data_wait_seconds": 1.9878149032592773e-05} {"step": 3930, "total_steps": 28610, "tokens": 8241807360, "source": "fineweb_hq", "phase": 1, "loss": 3.0976646542549133, "ce_loss": 3.082814335823059, "logit_z_loss": 12.979605317115784, "router_aux_loss": 1.3020694106817245, "router_z_loss": 0.5316935554146767, "routing_entropy_normalized": 0.6801618488535931, "learning_rate": 0.003, "grad_norm": 0.2978648245334625, "tokens_per_second": 3448978.5066559366, "eta_seconds": 15006.678429603577, "data_wait_seconds": 1.5854835510253906e-05} {"step": 3940, "total_steps": 28610, "tokens": 8262778880, "source": "fineweb_hq", "phase": 1, "loss": 3.112556964159012, "ce_loss": 3.097169876098633, "logit_z_loss": 12.939765572547913, "router_aux_loss": 1.3536853790283203, "router_z_loss": 0.5561464168131351, "routing_entropy_normalized": 0.6538166171438144, "learning_rate": 0.003, "grad_norm": 0.3729753792285919, "tokens_per_second": 4094660.2360363253, "eta_seconds": 12635.172848939896, "data_wait_seconds": 2.193450927734375e-05} {"step": 3950, "total_steps": 28610, "tokens": 8283750400, "source": "fineweb_edu", "phase": 1, "loss": 3.477429836988449, "ce_loss": 3.461939126253128, "logit_z_loss": 11.858845353126526, "router_aux_loss": 1.3695843666791916, "router_z_loss": 0.6089762076735497, "routing_entropy_normalized": 0.6520632804470198, "learning_rate": 0.003, "grad_norm": 0.37196776270866394, "tokens_per_second": 4142210.1500536962, "eta_seconds": 12485.06629228592, "data_wait_seconds": 1.8030405044555664e-05} {"step": 3960, "total_steps": 28610, "tokens": 8304721920, "source": "fineweb_hq", "phase": 1, "loss": 3.106275975704193, "ce_loss": 3.0905299484729767, "logit_z_loss": 13.173890829086304, "router_aux_loss": 1.3838087171316147, "router_z_loss": 0.5905838012695312, "routing_entropy_normalized": 0.6537956217313756, "learning_rate": 0.003, "grad_norm": 0.35015052556991577, "tokens_per_second": 3249908.333014541, "eta_seconds": 15906.539970636368, "data_wait_seconds": 2.0712614059448242e-05} {"step": 3970, "total_steps": 28610, "tokens": 8325693440, "source": "fineweb_edu", "phase": 1, "loss": 3.4498343467712402, "ce_loss": 3.434064358472824, "logit_z_loss": 11.808507919311523, "router_aux_loss": 1.391155645251274, "router_z_loss": 0.6776611879467964, "routing_entropy_normalized": 0.6337996273300822, "learning_rate": 0.003, "grad_norm": 0.3812915086746216, "tokens_per_second": 3451096.1085606995, "eta_seconds": 14973.163208007812, "data_wait_seconds": 1.6868114471435547e-05} {"step": 3980, "total_steps": 28610, "tokens": 8346664960, "source": "fineweb_hq", "phase": 1, "loss": 3.0925347805023193, "ce_loss": 3.077018082141876, "logit_z_loss": 12.998393774032593, "router_aux_loss": 1.355189487338066, "router_z_loss": 0.6650432422757149, "routing_entropy_normalized": 0.6607499793883403, "learning_rate": 0.003, "grad_norm": 0.27166110277175903, "tokens_per_second": 4262006.048344767, "eta_seconds": 12119.375987291336, "data_wait_seconds": 1.7523765563964844e-05} {"step": 3990, "total_steps": 28610, "tokens": 8367636480, "source": "fineweb_edu", "phase": 1, "loss": 3.446911036968231, "ce_loss": 3.4317097663879395, "logit_z_loss": 12.306073307991028, "router_aux_loss": 1.3253281116485596, "router_z_loss": 0.717378169298172, "routing_entropy_normalized": 0.6698064143804524, "learning_rate": 0.003, "grad_norm": 0.46378031373023987, "tokens_per_second": 3695850.448467129, "eta_seconds": 13970.230386734009, "data_wait_seconds": 1.7642974853515625e-05} {"step": 4000, "total_steps": 28610, "tokens": 8388608000, "source": "fineweb_hq", "phase": 1, "loss": 3.0832138657569885, "ce_loss": 3.0680771470069885, "logit_z_loss": 13.123979091644287, "router_aux_loss": 1.3172654062509537, "router_z_loss": 0.6516395658254623, "routing_entropy_normalized": 0.6674894093239385, "learning_rate": 0.003, "grad_norm": 0.42664480209350586, "tokens_per_second": 3640268.853197489, "eta_seconds": 14177.774445056915, "data_wait_seconds": 2.002716064453125e-05} {"step": 4010, "total_steps": 28610, "tokens": 8409579520, "source": "fineweb_hq", "phase": 1, "loss": 3.0829482972621918, "ce_loss": 3.0680422484874725, "logit_z_loss": 13.235241293907166, "router_aux_loss": 1.297871544957161, "router_z_loss": 0.6038403287529945, "routing_entropy_normalized": 0.6690051119962576, "learning_rate": 0.003, "grad_norm": 0.37577131390571594, "tokens_per_second": 3531425.367614241, "eta_seconds": 14608.814806938171, "data_wait_seconds": 2.053380012512207e-05} {"step": 4020, "total_steps": 28610, "tokens": 8430551040, "source": "fineweb_hq", "phase": 1, "loss": 3.0842332541942596, "ce_loss": 3.0694815516471863, "logit_z_loss": 13.343349933624268, "router_aux_loss": 1.2781190276145935, "router_z_loss": 0.6361053436994553, "routing_entropy_normalized": 0.6635936674079359, "learning_rate": 0.003, "grad_norm": 0.26074516773223877, "tokens_per_second": 4158067.000616991, "eta_seconds": 12402.149285316467, "data_wait_seconds": 2.1129846572875977e-05} {"step": 4030, "total_steps": 28610, "tokens": 8451522560, "source": "fineweb_hq", "phase": 1, "loss": 3.107564002275467, "ce_loss": 3.0927532017230988, "logit_z_loss": 12.851875185966492, "router_aux_loss": 1.2984914779663086, "router_z_loss": 0.5407389886677265, "routing_entropy_normalized": 0.6564394645151013, "learning_rate": 0.003, "grad_norm": 0.3938543200492859, "tokens_per_second": 3667760.2373494464, "eta_seconds": 14054.352745056152, "data_wait_seconds": 1.6987323760986328e-05} {"step": 4040, "total_steps": 28610, "tokens": 8472494080, "source": "fineweb_hq", "phase": 1, "loss": 3.131354421377182, "ce_loss": 3.116411179304123, "logit_z_loss": 13.030147790908813, "router_aux_loss": 1.30172860622406, "router_z_loss": 0.6229094192385674, "routing_entropy_normalized": 0.6594332491138593, "learning_rate": 0.003, "grad_norm": 0.27107322216033936, "tokens_per_second": 4203182.152838912, "eta_seconds": 12259.05106329918, "data_wait_seconds": 1.4662742614746094e-05} {"step": 4050, "total_steps": 28610, "tokens": 8493465600, "source": "fineweb_edu", "phase": 1, "loss": 3.464484393596649, "ce_loss": 3.448403149843216, "logit_z_loss": 12.259857296943665, "router_aux_loss": 1.4128840416669846, "router_z_loss": 0.726362518966198, "routing_entropy_normalized": 0.6412979091383226, "learning_rate": 0.003, "grad_norm": 0.3676982820034027, "tokens_per_second": 3699911.9712119945, "eta_seconds": 13920.886096954346, "data_wait_seconds": 1.5944242477416992e-05} {"step": 4060, "total_steps": 28610, "tokens": 8514437120, "source": "fineweb_hq", "phase": 1, "loss": 3.120785653591156, "ce_loss": 3.104473203420639, "logit_z_loss": 13.084588766098022, "router_aux_loss": 1.4330393821001053, "router_z_loss": 0.6736200153827667, "routing_entropy_normalized": 0.6372842302859703, "learning_rate": 0.003, "grad_norm": 0.3346908986568451, "tokens_per_second": 3775153.0526862615, "eta_seconds": 13637.879281044006, "data_wait_seconds": 1.817941665649414e-05} {"step": 4070, "total_steps": 28610, "tokens": 8535408640, "source": "fineweb_edu", "phase": 1, "loss": 3.4473420679569244, "ce_loss": 3.4315215945243835, "logit_z_loss": 11.853707671165466, "router_aux_loss": 1.39209945499897, "router_z_loss": 0.7141030803322792, "routing_entropy_normalized": 0.6519293356619261, "learning_rate": 0.003, "grad_norm": 0.3552514910697937, "tokens_per_second": 3431204.0610608994, "eta_seconds": 14998.848556995392, "data_wait_seconds": 1.8477439880371094e-05} {"step": 4080, "total_steps": 28610, "tokens": 8556380160, "source": "fineweb_hq", "phase": 1, "loss": 3.1271297335624695, "ce_loss": 3.111397475004196, "logit_z_loss": 13.07283616065979, "router_aux_loss": 1.3781538754701614, "router_z_loss": 0.6434721350669861, "routing_entropy_normalized": 0.6471809455103769, "learning_rate": 0.003, "grad_norm": 0.37866654992103577, "tokens_per_second": 3474734.2256232155, "eta_seconds": 14804.913187503815, "data_wait_seconds": 2.0742416381835938e-05} {"step": 4090, "total_steps": 28610, "tokens": 8577351680, "source": "fineweb_edu", "phase": 1, "loss": 3.4720899760723114, "ce_loss": 3.4570840895175934, "logit_z_loss": 11.822220206260681, "router_aux_loss": 1.311354711651802, "router_z_loss": 0.7100833356380463, "routing_entropy_normalized": 0.6480032661663437, "learning_rate": 0.003, "grad_norm": 0.4683411121368408, "tokens_per_second": 4473243.524040531, "eta_seconds": 11495.499130249022, "data_wait_seconds": 1.7434358596801758e-05} {"step": 4100, "total_steps": 28610, "tokens": 8598323200, "source": "fineweb_hq", "phase": 1, "loss": 3.124438762664795, "ce_loss": 3.108945220708847, "logit_z_loss": 13.061506271362305, "router_aux_loss": 1.3532674014568329, "router_z_loss": 0.6546848192811012, "routing_entropy_normalized": 0.6598496328945886, "learning_rate": 0.003, "grad_norm": 0.41635259985923767, "tokens_per_second": 3336040.907032471, "eta_seconds": 15407.843294620514, "data_wait_seconds": 2.008676528930664e-05} {"step": 4110, "total_steps": 28610, "tokens": 8619294720, "source": "fineweb_hq", "phase": 1, "loss": 3.1618341505527496, "ce_loss": 3.14600470662117, "logit_z_loss": 13.15834891796112, "router_aux_loss": 1.3861604183912277, "router_z_loss": 0.6520516201853752, "routing_entropy_normalized": 0.6256515922667834, "learning_rate": 0.003, "grad_norm": 0.33329468965530396, "tokens_per_second": 3454860.4522575033, "eta_seconds": 14871.86666727066, "data_wait_seconds": 2.002716064453125e-05} {"step": 4120, "total_steps": 28610, "tokens": 8640266240, "source": "fineweb_hq", "phase": 1, "loss": 3.2422048449516296, "ce_loss": 3.227050632238388, "logit_z_loss": 12.369200587272644, "router_aux_loss": 1.3166255354881287, "router_z_loss": 0.7510718181729317, "routing_entropy_normalized": 0.6553358177679867, "learning_rate": 0.003, "grad_norm": 0.3076326251029968, "tokens_per_second": 4192549.3343626163, "eta_seconds": 12250.124777078629, "data_wait_seconds": 2.0295381546020508e-05} {"step": 4130, "total_steps": 28610, "tokens": 8661237760, "source": "fineweb_hq", "phase": 1, "loss": 3.32413387298584, "ce_loss": 3.308814764022827, "logit_z_loss": 12.6823011636734, "router_aux_loss": 1.3315531313419342, "router_z_loss": 0.735337384045124, "routing_entropy_normalized": 0.6429996398077045, "learning_rate": 0.003, "grad_norm": 0.347718745470047, "tokens_per_second": 3427596.4429780976, "eta_seconds": 14977.924564361572, "data_wait_seconds": 2.339482307434082e-05} {"step": 4140, "total_steps": 28610, "tokens": 8682209280, "source": "fineweb_hq", "phase": 1, "loss": 3.388689875602722, "ce_loss": 3.3736135065555573, "logit_z_loss": 12.133811831474304, "router_aux_loss": 1.3164201378822327, "router_z_loss": 0.6988702267408371, "routing_entropy_normalized": 0.652642538396416, "learning_rate": 0.003, "grad_norm": 0.25602179765701294, "tokens_per_second": 4125587.939174768, "eta_seconds": 12438.78695511818, "data_wait_seconds": 1.9550323486328125e-05} {"step": 4150, "total_steps": 28610, "tokens": 8703180800, "source": "fineweb_edu", "phase": 1, "loss": 3.464207261800766, "ce_loss": 3.4489807188510895, "logit_z_loss": 12.145683884620667, "router_aux_loss": 1.320440798997879, "router_z_loss": 0.8075126707553864, "routing_entropy_normalized": 0.6562550889256252, "learning_rate": 0.003, "grad_norm": 0.32459065318107605, "tokens_per_second": 4182116.9143245164, "eta_seconds": 12265.639380931854, "data_wait_seconds": 1.6063451766967773e-05} {"step": 4160, "total_steps": 28610, "tokens": 8724152320, "source": "fineweb_hq", "phase": 1, "loss": 3.3509901463985443, "ce_loss": 3.3356062471866608, "logit_z_loss": 11.989621639251709, "router_aux_loss": 1.3462657034397125, "router_z_loss": 0.7223234996199608, "routing_entropy_normalized": 0.643291002875436, "learning_rate": 0.003, "grad_norm": 0.2831208109855652, "tokens_per_second": 3305714.6062026443, "eta_seconds": 15511.129213571547, "data_wait_seconds": 2.2530555725097656e-05} {"step": 4170, "total_steps": 28610, "tokens": 8745123840, "source": "fineweb_edu", "phase": 1, "loss": 3.4584988951683044, "ce_loss": 3.4428119361400604, "logit_z_loss": 11.781452894210815, "router_aux_loss": 1.3654249012470245, "router_z_loss": 0.8546332567930222, "routing_entropy_normalized": 0.6288092259370558, "learning_rate": 0.003, "grad_norm": 0.3788830041885376, "tokens_per_second": 4000198.3807317265, "eta_seconds": 12812.963258743286, "data_wait_seconds": 1.8537044525146484e-05} {"step": 4180, "total_steps": 28610, "tokens": 8766095360, "source": "fineweb_hq", "phase": 1, "loss": 3.3979105949401855, "ce_loss": 3.3822575211524963, "logit_z_loss": 11.77753484249115, "router_aux_loss": 1.3749540448188782, "router_z_loss": 0.7258071601390839, "routing_entropy_normalized": 0.6346549456684233, "learning_rate": 0.003, "grad_norm": 0.4005996286869049, "tokens_per_second": 3481476.5999663807, "eta_seconds": 14716.003939390182, "data_wait_seconds": 1.7762184143066406e-05} {"step": 4190, "total_steps": 28610, "tokens": 8787066880, "source": "fineweb_edu", "phase": 1, "loss": 3.4365466833114624, "ce_loss": 3.4212400317192078, "logit_z_loss": 11.891613483428955, "router_aux_loss": 1.3286451697349548, "router_z_loss": 0.8310954570770264, "routing_entropy_normalized": 0.6711628410882581, "learning_rate": 0.003, "grad_norm": 0.2771918773651123, "tokens_per_second": 4017810.7457614625, "eta_seconds": 12746.357427120209, "data_wait_seconds": 1.9669532775878906e-05} {"step": 4200, "total_steps": 28610, "tokens": 8808038400, "source": "fineweb_hq", "phase": 1, "loss": 3.3843519389629364, "ce_loss": 3.3692814707756042, "logit_z_loss": 12.581587314605713, "router_aux_loss": 1.3009837120771408, "router_z_loss": 0.8024657890200615, "routing_entropy_normalized": 0.6603770343359079, "learning_rate": 0.003, "grad_norm": 0.33511170744895935, "tokens_per_second": 3681253.7617604476, "eta_seconds": 13905.990630626678, "data_wait_seconds": 2.485513687133789e-05} {"step": 4210, "total_steps": 28610, "tokens": 8829009920, "source": "fineweb_hq", "phase": 1, "loss": 3.3489736020565033, "ce_loss": 3.333913803100586, "logit_z_loss": 12.479045629501343, "router_aux_loss": 1.3127999156713486, "router_z_loss": 0.6839306503534317, "routing_entropy_normalized": 0.6607505005812764, "learning_rate": 0.003, "grad_norm": 0.2711661159992218, "tokens_per_second": 4041455.551033037, "eta_seconds": 12661.405811309814, "data_wait_seconds": 2.5063753128051758e-05} {"step": 4220, "total_steps": 28610, "tokens": 8849981440, "source": "fineweb_hq", "phase": 1, "loss": 3.3432452380657196, "ce_loss": 3.3284261226654053, "logit_z_loss": 12.15645432472229, "router_aux_loss": 1.2911258935928345, "router_z_loss": 0.6922829523682594, "routing_entropy_normalized": 0.6481639998464086, "learning_rate": 0.003, "grad_norm": 0.3871326744556427, "tokens_per_second": 4104702.7011551377, "eta_seconds": 12461.203893184662, "data_wait_seconds": 2.199411392211914e-05} {"step": 4230, "total_steps": 28610, "tokens": 8870952960, "source": "fineweb_hq", "phase": 1, "loss": 3.342056840658188, "ce_loss": 3.326863318681717, "logit_z_loss": 12.120229125022888, "router_aux_loss": 1.33921480178833, "router_z_loss": 0.5893312841653824, "routing_entropy_normalized": 0.6500837982374071, "learning_rate": 0.003, "grad_norm": 0.3911799192428589, "tokens_per_second": 3560727.686760641, "eta_seconds": 14359.02159833908, "data_wait_seconds": 2.193450927734375e-05} {"step": 4240, "total_steps": 28610, "tokens": 8891924480, "source": "fineweb_hq", "phase": 1, "loss": 3.3365062177181244, "ce_loss": 3.3208480179309845, "logit_z_loss": 12.703095436096191, "router_aux_loss": 1.360121265053749, "router_z_loss": 0.7866734489798546, "routing_entropy_normalized": 0.6540446517047914, "learning_rate": 0.003, "grad_norm": 0.4068467319011688, "tokens_per_second": 3930064.808392378, "eta_seconds": 13004.262456655502, "data_wait_seconds": 1.9550323486328125e-05} {"step": 4250, "total_steps": 28610, "tokens": 8912896000, "source": "fineweb_edu", "phase": 1, "loss": 3.4860578775405884, "ce_loss": 3.470299154520035, "logit_z_loss": 12.06664764881134, "router_aux_loss": 1.3705772310495377, "router_z_loss": 0.8463049605488777, "routing_entropy_normalized": 0.6327617818286454, "learning_rate": 0.003, "grad_norm": 0.43275970220565796, "tokens_per_second": 3524288.2322063087, "eta_seconds": 14495.58587551117, "data_wait_seconds": 1.868605613708496e-05} {"step": 4260, "total_steps": 28610, "tokens": 8933867520, "source": "fineweb_hq", "phase": 1, "loss": 3.3064398765563965, "ce_loss": 3.2908915281295776, "logit_z_loss": 12.57325541973114, "router_aux_loss": 1.3500773310661316, "router_z_loss": 0.7902694344520569, "routing_entropy_normalized": 0.6583654405386211, "learning_rate": 0.003, "grad_norm": 0.28609561920166016, "tokens_per_second": 3917096.23072005, "eta_seconds": 13036.608802080154, "data_wait_seconds": 1.958012580871582e-05} {"step": 4270, "total_steps": 28610, "tokens": 8954839040, "source": "fineweb_edu", "phase": 1, "loss": 3.4360341131687164, "ce_loss": 3.4207438230514526, "logit_z_loss": 12.383973717689514, "router_aux_loss": 1.3199822157621384, "router_z_loss": 0.8520980849862099, "routing_entropy_normalized": 0.6474021863079321, "learning_rate": 0.003, "grad_norm": 0.3231535851955414, "tokens_per_second": 4233180.431058914, "eta_seconds": 12058.233876705171, "data_wait_seconds": 2.041459083557129e-05} {"step": 4280, "total_steps": 28610, "tokens": 8975810560, "source": "fineweb_hq", "phase": 1, "loss": 3.3219889104366302, "ce_loss": 3.306864321231842, "logit_z_loss": 12.642742276191711, "router_aux_loss": 1.314687579870224, "router_z_loss": 0.7134254649281502, "routing_entropy_normalized": 0.6643278680658874, "learning_rate": 0.003, "grad_norm": 0.40664243698120117, "tokens_per_second": 3680753.7373348847, "eta_seconds": 13862.298811912537, "data_wait_seconds": 1.8864870071411133e-05} {"step": 4290, "total_steps": 28610, "tokens": 8996782080, "source": "fineweb_edu", "phase": 1, "loss": 3.44837149977684, "ce_loss": 3.433358371257782, "logit_z_loss": 12.188112497329712, "router_aux_loss": 1.3024399876594543, "router_z_loss": 0.7699028700590134, "routing_entropy_normalized": 0.6400052189221919, "learning_rate": 0.003, "grad_norm": 0.4299810826778412, "tokens_per_second": 4025502.9215515302, "eta_seconds": 12669.904266357422, "data_wait_seconds": 2.300739288330078e-05} {"step": 4300, "total_steps": 28610, "tokens": 9017753600, "source": "fineweb_hq", "phase": 1, "loss": 3.3608700335025787, "ce_loss": 3.3460238575935364, "logit_z_loss": 12.043206334114075, "router_aux_loss": 1.2996256351470947, "router_z_loss": 0.6455821990966797, "routing_entropy_normalized": 0.6554370666587704, "learning_rate": 0.003, "grad_norm": 0.4209382236003876, "tokens_per_second": 4501502.575386737, "eta_seconds": 11325.499489605427, "data_wait_seconds": 1.9162893295288086e-05} {"step": 4310, "total_steps": 28610, "tokens": 9038725120, "source": "fineweb_hq", "phase": 1, "loss": 3.3304599225521088, "ce_loss": 3.3152543902397156, "logit_z_loss": 12.301120162010193, "router_aux_loss": 1.3280748128890991, "router_z_loss": 0.6946123763918877, "routing_entropy_normalized": 0.6271815509725179, "learning_rate": 0.003, "grad_norm": 0.3771837651729584, "tokens_per_second": 4031428.0418365775, "eta_seconds": 12640.878882408142, "data_wait_seconds": 1.8835067749023438e-05} {"step": 4320, "total_steps": 28610, "tokens": 9059696640, "source": "fineweb_hq", "phase": 1, "loss": 3.3455627858638763, "ce_loss": 3.3302758038043976, "logit_z_loss": 12.406709671020508, "router_aux_loss": 1.328674629330635, "router_z_loss": 0.759579561650753, "routing_entropy_normalized": 0.6489509734956904, "learning_rate": 0.003, "grad_norm": 0.3333829641342163, "tokens_per_second": 3923869.6253270525, "eta_seconds": 12982.037362098694, "data_wait_seconds": 2.0325183868408203e-05} {"step": 4330, "total_steps": 28610, "tokens": 9080668160, "source": "fineweb_hq", "phase": 1, "loss": 3.352731943130493, "ce_loss": 3.336787134408951, "logit_z_loss": 12.385053873062134, "router_aux_loss": 1.3967368304729462, "router_z_loss": 0.7389510348439217, "routing_entropy_normalized": 0.6356308475054941, "learning_rate": 0.003, "grad_norm": 0.3308107256889343, "tokens_per_second": 4029712.6419251524, "eta_seconds": 12635.85150718689, "data_wait_seconds": 2.2113323211669922e-05} {"step": 4340, "total_steps": 28610, "tokens": 9101639680, "source": "fineweb_hq", "phase": 1, "loss": 3.3223149478435516, "ce_loss": 3.30663138628006, "logit_z_loss": 12.60564410686493, "router_aux_loss": 1.3636119663715363, "router_z_loss": 0.7868393883109093, "routing_entropy_normalized": 0.6585439904911761, "learning_rate": 0.003, "grad_norm": 0.32229676842689514, "tokens_per_second": 4096184.1589753428, "eta_seconds": 12425.681332826614, "data_wait_seconds": 2.181529998779297e-05} {"step": 4350, "total_steps": 28610, "tokens": 9122611200, "source": "fineweb_edu", "phase": 1, "loss": 3.4713690876960754, "ce_loss": 3.456262230873108, "logit_z_loss": 12.15312671661377, "router_aux_loss": 1.3089286535978317, "router_z_loss": 0.8022854402661324, "routing_entropy_normalized": 0.6591399542640255, "learning_rate": 0.003, "grad_norm": 0.40700453519821167, "tokens_per_second": 3902261.8071486764, "eta_seconds": 13037.79962348938, "data_wait_seconds": 1.6570091247558594e-05} {"step": 4360, "total_steps": 28610, "tokens": 9143582720, "source": "fineweb_hq", "phase": 1, "loss": 3.367409586906433, "ce_loss": 3.3527892529964447, "logit_z_loss": 12.152613401412964, "router_aux_loss": 1.2671273052692413, "router_z_loss": 0.7337805703282356, "routing_entropy_normalized": 0.6578713511211196, "learning_rate": 0.003, "grad_norm": 0.3602482080459595, "tokens_per_second": 3088918.408961168, "eta_seconds": 16463.994598388672, "data_wait_seconds": 0.24112749099731445} {"step": 4370, "total_steps": 28610, "tokens": 9164554240, "source": "fineweb_edu", "phase": 1, "loss": 3.438068240880966, "ce_loss": 3.4236415028572083, "logit_z_loss": 12.014308214187622, "router_aux_loss": 1.2513790428638458, "router_z_loss": 0.7115415781736374, "routing_entropy_normalized": 0.6627109844303988, "learning_rate": 0.003, "grad_norm": 0.4476943612098694, "tokens_per_second": 4199583.436956975, "eta_seconds": 12104.763542175293, "data_wait_seconds": 1.9431114196777344e-05} {"step": 4380, "total_steps": 28610, "tokens": 9185525760, "source": "fineweb_hq", "phase": 1, "loss": 3.3078097701072693, "ce_loss": 3.2931810915470123, "logit_z_loss": 12.150402665138245, "router_aux_loss": 1.282948061823845, "router_z_loss": 0.5841634012758732, "routing_entropy_normalized": 0.653947960834489, "learning_rate": 0.003, "grad_norm": 0.3191925585269928, "tokens_per_second": 3360409.900805921, "eta_seconds": 15121.367470026014, "data_wait_seconds": 1.868605613708496e-05} {"step": 4390, "total_steps": 28610, "tokens": 9206497280, "source": "fineweb_edu", "phase": 1, "loss": 3.4792065918445587, "ce_loss": 3.464849829673767, "logit_z_loss": 11.892366409301758, "router_aux_loss": 1.2546112090349197, "router_z_loss": 0.6214426979422569, "routing_entropy_normalized": 0.6751451808147109, "learning_rate": 0.003, "grad_norm": 0.4375312626361847, "tokens_per_second": 3327926.0865474856, "eta_seconds": 15262.66513109207, "data_wait_seconds": 2.2470951080322266e-05} {"step": 4400, "total_steps": 28610, "tokens": 9227468800, "source": "fineweb_hq", "phase": 1, "loss": 3.3105008006095886, "ce_loss": 3.2958245277404785, "logit_z_loss": 12.47516655921936, "router_aux_loss": 1.2918049097061157, "router_z_loss": 0.5107231885194778, "routing_entropy_normalized": 0.6535241344865993, "learning_rate": 0.003, "grad_norm": 0.3241752088069916, "tokens_per_second": 3351051.8984704493, "eta_seconds": 15151.078365325928, "data_wait_seconds": 2.092123031616211e-05} {"step": 4410, "total_steps": 28610, "tokens": 9248440320, "source": "fineweb_hq", "phase": 1, "loss": 3.3200220465660095, "ce_loss": 3.3053240180015564, "logit_z_loss": 12.766775488853455, "router_aux_loss": 1.2929316759109497, "router_z_loss": 0.49198710173368454, "routing_entropy_normalized": 0.6376091868718158, "learning_rate": 0.003, "grad_norm": 0.34198665618896484, "tokens_per_second": 4301661.9674789775, "eta_seconds": 11798.016390800476, "data_wait_seconds": 2.1576881408691406e-05} {"step": 4420, "total_steps": 28610, "tokens": 9269411840, "source": "fineweb_hq", "phase": 1, "loss": 3.3550044000148773, "ce_loss": 3.33958101272583, "logit_z_loss": 12.270355582237244, "router_aux_loss": 1.360717087984085, "router_z_loss": 0.5892029404640198, "routing_entropy_normalized": 0.6511188585409327, "learning_rate": 0.003, "grad_norm": 0.35827380418777466, "tokens_per_second": 4118757.6721155187, "eta_seconds": 12316.846709251404, "data_wait_seconds": 1.8090009689331055e-05} {"step": 4430, "total_steps": 28610, "tokens": 9290383360, "source": "fineweb_hq", "phase": 1, "loss": 3.3952582478523254, "ce_loss": 3.378549188375473, "logit_z_loss": 11.95906138420105, "router_aux_loss": 1.4901492148637772, "router_z_loss": 0.6116464659571648, "routing_entropy_normalized": 0.628419020852848, "learning_rate": 0.003, "grad_norm": 0.5405770540237427, "tokens_per_second": 3509892.585976196, "eta_seconds": 14447.489237308502, "data_wait_seconds": 1.5228986740112305e-05} {"step": 4440, "total_steps": 28610, "tokens": 9311354880, "source": "fineweb_hq", "phase": 1, "loss": 3.3975697457790375, "ce_loss": 3.3802319169044495, "logit_z_loss": 11.869158744812012, "router_aux_loss": 1.5415350049734116, "router_z_loss": 0.7356051057577133, "routing_entropy_normalized": 0.6507570002510549, "learning_rate": 0.003, "grad_norm": 0.5289170145988464, "tokens_per_second": 3326490.459664234, "eta_seconds": 15237.730110645292, "data_wait_seconds": 2.110004425048828e-05} {"step": 4450, "total_steps": 28610, "tokens": 9332326400, "source": "fineweb_edu", "phase": 1, "loss": 3.467955380678177, "ce_loss": 3.450983852148056, "logit_z_loss": 12.191317558288574, "router_aux_loss": 1.4826335906982422, "router_z_loss": 0.926053836941719, "routing_entropy_normalized": 0.633296945059222, "learning_rate": 0.003, "grad_norm": 0.36435893177986145, "tokens_per_second": 3797348.1796484147, "eta_seconds": 13342.782890319824, "data_wait_seconds": 1.755356788635254e-05} {"step": 4460, "total_steps": 28610, "tokens": 9353297920, "source": "fineweb_hq", "phase": 1, "loss": 3.3542929887771606, "ce_loss": 3.3378228843212128, "logit_z_loss": 12.21917986869812, "router_aux_loss": 1.4319093972444534, "router_z_loss": 0.9290576577186584, "routing_entropy_normalized": 0.6485627585878828, "learning_rate": 0.003, "grad_norm": 0.3676818609237671, "tokens_per_second": 4302964.969905246, "eta_seconds": 11770.075088739395, "data_wait_seconds": 1.9222497940063477e-05} {"step": 4470, "total_steps": 28610, "tokens": 9374269440, "source": "fineweb_edu", "phase": 1, "loss": 3.4713340997695923, "ce_loss": 3.455245852470398, "logit_z_loss": 11.968698382377625, "router_aux_loss": 1.3927419632673264, "router_z_loss": 0.9639596790075302, "routing_entropy_normalized": 0.652163699271012, "learning_rate": 0.003, "grad_norm": 0.4352385997772217, "tokens_per_second": 3467525.6108543915, "eta_seconds": 14599.819860458374, "data_wait_seconds": 2.2590160369873047e-05} {"step": 4480, "total_steps": 28610, "tokens": 9395240960, "source": "fineweb_hq", "phase": 1, "loss": 3.372117578983307, "ce_loss": 3.3563295900821686, "logit_z_loss": 12.246528267860413, "router_aux_loss": 1.3660084456205368, "router_z_loss": 0.9033109918236732, "routing_entropy_normalized": 0.6575207870858109, "learning_rate": 0.003, "grad_norm": 0.26176366209983826, "tokens_per_second": 4144698.304262402, "eta_seconds": 12209.399585962296, "data_wait_seconds": 2.0444393157958984e-05} {"step": 4490, "total_steps": 28610, "tokens": 9416212480, "source": "fineweb_edu", "phase": 1, "loss": 3.440609723329544, "ce_loss": 3.4249804615974426, "logit_z_loss": 12.178605198860168, "router_aux_loss": 1.3527356684207916, "router_z_loss": 0.884078674018383, "routing_entropy_normalized": 0.6645037097509009, "learning_rate": 0.003, "grad_norm": 0.37052756547927856, "tokens_per_second": 4026737.4287365195, "eta_seconds": 12561.858610153198, "data_wait_seconds": 1.665949821472168e-05} {"step": 4500, "total_steps": 28610, "tokens": 9437184000, "source": "fineweb_hq", "phase": 1, "loss": 3.326256424188614, "ce_loss": 3.3112954795360565, "logit_z_loss": 12.294381737709045, "router_aux_loss": 1.2986864298582077, "router_z_loss": 0.7446855381131172, "routing_entropy_normalized": 0.6690082250928503, "learning_rate": 0.003, "grad_norm": 0.27066412568092346, "tokens_per_second": 3334069.6898131263, "eta_seconds": 15165.350284814835, "data_wait_seconds": 2.428889274597168e-05} {"step": 4510, "total_steps": 28610, "tokens": 9458155520, "source": "fineweb_hq", "phase": 1, "loss": 3.3659211695194244, "ce_loss": 3.3507121801376343, "logit_z_loss": 12.21318244934082, "router_aux_loss": 1.3266508877277374, "router_z_loss": 0.7211710959672928, "routing_entropy_normalized": 0.6511891733239115, "learning_rate": 0.003, "grad_norm": 0.42848142981529236, "tokens_per_second": 4307046.584803289, "eta_seconds": 11734.575469493864, "data_wait_seconds": 1.8835067749023438e-05} {"step": 4520, "total_steps": 28610, "tokens": 9479127040, "source": "fineweb_hq", "phase": 1, "loss": 3.360306978225708, "ce_loss": 3.3448360562324524, "logit_z_loss": 12.475116491317749, "router_aux_loss": 1.3592055141925812, "router_z_loss": 0.6314015462994576, "routing_entropy_normalized": 0.6227061374879688, "learning_rate": 0.003, "grad_norm": 0.37677326798439026, "tokens_per_second": 3762667.5979732284, "eta_seconds": 13426.748540639877, "data_wait_seconds": 1.4781951904296875e-05} {"step": 4530, "total_steps": 28610, "tokens": 9500098560, "source": "fineweb_hq", "phase": 1, "loss": 3.360817402601242, "ce_loss": 3.3452571630477905, "logit_z_loss": 12.514629364013672, "router_aux_loss": 1.365266591310501, "router_z_loss": 0.6561115458607674, "routing_entropy_normalized": 0.6620508328024922, "learning_rate": 0.003, "grad_norm": 0.3503141403198242, "tokens_per_second": 3782822.7783077275, "eta_seconds": 13349.66587638855, "data_wait_seconds": 1.958012580871582e-05} {"step": 4540, "total_steps": 28610, "tokens": 9521070080, "source": "fineweb_hq", "phase": 1, "loss": 3.3567082583904266, "ce_loss": 3.340923994779587, "logit_z_loss": 12.379579901695251, "router_aux_loss": 1.3896291255950928, "router_z_loss": 0.6499892696738243, "routing_entropy_normalized": 0.6565644341325907, "learning_rate": 0.003, "grad_norm": 0.5213366746902466, "tokens_per_second": 4035240.4622842963, "eta_seconds": 12509.402874946594, "data_wait_seconds": 1.901388168334961e-05} {"step": 4550, "total_steps": 28610, "tokens": 9542041600, "source": "fineweb_edu", "phase": 1, "loss": 3.46019846200943, "ce_loss": 3.443841278553009, "logit_z_loss": 11.976923823356628, "router_aux_loss": 1.4380687326192856, "router_z_loss": 0.778807207942009, "routing_entropy_normalized": 0.6584192517681805, "learning_rate": 0.003, "grad_norm": 0.43304502964019775, "tokens_per_second": 3547818.682168701, "eta_seconds": 14222.112695217133, "data_wait_seconds": 2.2530555725097656e-05} {"step": 4560, "total_steps": 28610, "tokens": 9563013120, "source": "fineweb_hq", "phase": 1, "loss": 3.3999273777008057, "ce_loss": 3.3840537667274475, "logit_z_loss": 12.096766352653503, "router_aux_loss": 1.3936590105295181, "router_z_loss": 0.7273596376180649, "routing_entropy_normalized": 0.6810246762685299, "learning_rate": 0.003, "grad_norm": 0.2626986503601074, "tokens_per_second": 4182885.5700001065, "eta_seconds": 12057.825813293457, "data_wait_seconds": 2.110004425048828e-05} {"step": 4570, "total_steps": 28610, "tokens": 9583984640, "source": "fineweb_edu", "phase": 1, "loss": 3.443315029144287, "ce_loss": 3.427618533372879, "logit_z_loss": 12.007552027702332, "router_aux_loss": 1.3684780150651932, "router_z_loss": 0.810969702899456, "routing_entropy_normalized": 0.6771010808742842, "learning_rate": 0.003, "grad_norm": 0.33415916562080383, "tokens_per_second": 3330843.8185546943, "eta_seconds": 15135.964586257935, "data_wait_seconds": 2.2023916244506836e-05} {"step": 4580, "total_steps": 28610, "tokens": 9604956160, "source": "fineweb_hq", "phase": 1, "loss": 3.3894483745098114, "ce_loss": 3.374271959066391, "logit_z_loss": 12.226289629936218, "router_aux_loss": 1.3312671184539795, "router_z_loss": 0.6411412581801414, "routing_entropy_normalized": 0.6875717069621307, "learning_rate": 0.003, "grad_norm": 0.37154272198677063, "tokens_per_second": 3607367.918644953, "eta_seconds": 13969.89819073677, "data_wait_seconds": 1.710653305053711e-05} {"step": 4590, "total_steps": 28610, "tokens": 9625927680, "source": "fineweb_edu", "phase": 1, "loss": 3.4378155171871185, "ce_loss": 3.4226025342941284, "logit_z_loss": 11.790473818778992, "router_aux_loss": 1.326453372836113, "router_z_loss": 0.7694382891058922, "routing_entropy_normalized": 0.6909006561620739, "learning_rate": 0.003, "grad_norm": 0.5584471225738525, "tokens_per_second": 4330511.107750849, "eta_seconds": 11632.250740528107, "data_wait_seconds": 1.6510486602783203e-05} {"step": 4600, "total_steps": 28610, "tokens": 9646899200, "source": "fineweb_hq", "phase": 1, "loss": 3.3925801813602448, "ce_loss": 3.3768469393253326, "logit_z_loss": 12.558493256568909, "router_aux_loss": 1.377866044640541, "router_z_loss": 0.6987397596240044, "routing_entropy_normalized": 0.6316129854593248, "learning_rate": 0.003, "grad_norm": 0.4504087567329407, "tokens_per_second": 3885971.6282928083, "eta_seconds": 12957.536579370499, "data_wait_seconds": 2.35140323638916e-05} {"step": 4610, "total_steps": 28610, "tokens": 9667870720, "source": "fineweb_hq", "phase": 1, "loss": 3.3863861858844757, "ce_loss": 3.3705359399318695, "logit_z_loss": 12.14162015914917, "router_aux_loss": 1.392824411392212, "router_z_loss": 0.7078348621726036, "routing_entropy_normalized": 0.6663074122741642, "learning_rate": 0.003, "grad_norm": 0.5050069093704224, "tokens_per_second": 4303723.734686555, "eta_seconds": 11694.9068069458, "data_wait_seconds": 1.806020736694336e-05} {"step": 4620, "total_steps": 28610, "tokens": 9688842240, "source": "fineweb_hq", "phase": 1, "loss": 3.3600947558879852, "ce_loss": 3.3440683782100677, "logit_z_loss": 12.305837154388428, "router_aux_loss": 1.4096060395240784, "router_z_loss": 0.6997411549091339, "routing_entropy_normalized": 0.6596610739405073, "learning_rate": 0.003, "grad_norm": 0.4172894060611725, "tokens_per_second": 3831724.1897849124, "eta_seconds": 13130.035980701447, "data_wait_seconds": 1.4901161193847656e-05} {"step": 4630, "total_steps": 28610, "tokens": 9709813760, "source": "fineweb_hq", "phase": 1, "loss": 3.357998639345169, "ce_loss": 3.3418898582458496, "logit_z_loss": 12.566654562950134, "router_aux_loss": 1.420996904373169, "router_z_loss": 0.6421554908156395, "routing_entropy_normalized": 0.6461845162756349, "learning_rate": 0.003, "grad_norm": 0.3752504289150238, "tokens_per_second": 3835294.514107857, "eta_seconds": 13112.345029830933, "data_wait_seconds": 2.09808349609375e-05} {"step": 4640, "total_steps": 28610, "tokens": 9730785280, "source": "fineweb_hq", "phase": 1, "loss": 3.428219050168991, "ce_loss": 3.4122488498687744, "logit_z_loss": 12.192208528518677, "router_aux_loss": 1.3962697684764862, "router_z_loss": 0.7882818207144737, "routing_entropy_normalized": 0.6719357714857777, "learning_rate": 0.003, "grad_norm": 0.5752573013305664, "tokens_per_second": 4307500.270909323, "eta_seconds": 11670.047656059265, "data_wait_seconds": 2.1398067474365234e-05} {"step": 4650, "total_steps": 28610, "tokens": 9751756800, "source": "fineweb_edu", "phase": 1, "loss": 3.4719037413597107, "ce_loss": 3.4548507034778595, "logit_z_loss": 12.575854659080505, "router_aux_loss": 1.4874444007873535, "router_z_loss": 0.9210216328501701, "routing_entropy_normalized": 0.6284829133680622, "learning_rate": 0.003, "grad_norm": 0.3442164659500122, "tokens_per_second": 3105157.073432411, "eta_seconds": 16182.035475730896, "data_wait_seconds": 1.5348196029663086e-05} {"step": 4660, "total_steps": 28610, "tokens": 9772728320, "source": "fineweb_hq", "phase": 1, "loss": 3.3364848494529724, "ce_loss": 3.3198843896389008, "logit_z_loss": 12.368920683860779, "router_aux_loss": 1.4485834687948227, "router_z_loss": 0.877832256257534, "routing_entropy_normalized": 0.6525116036269909, "learning_rate": 0.003, "grad_norm": 0.43222299218177795, "tokens_per_second": 4562623.062543792, "eta_seconds": 11008.31467151642, "data_wait_seconds": 1.683831214904785e-05} {"step": 4670, "total_steps": 28610, "tokens": 9793699840, "source": "fineweb_edu", "phase": 1, "loss": 3.44310986995697, "ce_loss": 3.4264776408672333, "logit_z_loss": 12.397315740585327, "router_aux_loss": 1.4293280988931656, "router_z_loss": 1.0992493331432343, "routing_entropy_normalized": 0.6634268636335025, "learning_rate": 0.003, "grad_norm": 0.4860365688800812, "tokens_per_second": 3313799.7361224615, "eta_seconds": 15150.52896308899, "data_wait_seconds": 1.8477439880371094e-05} {"step": 4680, "total_steps": 28610, "tokens": 9814671360, "source": "fineweb_hq", "phase": 1, "loss": 3.324159115552902, "ce_loss": 3.3080619871616364, "logit_z_loss": 12.486194491386414, "router_aux_loss": 1.3879658877849579, "router_z_loss": 0.9688485935330391, "routing_entropy_normalized": 0.6509605109000779, "learning_rate": 0.003, "grad_norm": 0.3718508780002594, "tokens_per_second": 4057233.051767597, "eta_seconds": 12369.22964978218, "data_wait_seconds": 2.008676528930664e-05} {"step": 4690, "total_steps": 28610, "tokens": 9835642880, "source": "fineweb_edu", "phase": 1, "loss": 3.436857283115387, "ce_loss": 3.4210307002067566, "logit_z_loss": 12.24427080154419, "router_aux_loss": 1.3578513711690903, "router_z_loss": 1.0237084478139877, "routing_entropy_normalized": 0.6719011160304905, "learning_rate": 0.003, "grad_norm": 0.39454010128974915, "tokens_per_second": 3688795.8340230533, "eta_seconds": 13598.984085083008, "data_wait_seconds": 1.6570091247558594e-05} {"step": 4700, "total_steps": 28610, "tokens": 9856614400, "source": "fineweb_hq", "phase": 1, "loss": 3.274179458618164, "ce_loss": 3.2587530612945557, "logit_z_loss": 12.284371852874756, "router_aux_loss": 1.3366205245256424, "router_z_loss": 0.8317854404449463, "routing_entropy_normalized": 0.6933040956199094, "learning_rate": 0.003, "grad_norm": 0.35119637846946716, "tokens_per_second": 3816951.024431309, "eta_seconds": 13136.900106668472, "data_wait_seconds": 2.1219253540039062e-05} {"step": 4710, "total_steps": 28610, "tokens": 9877585920, "source": "fineweb_hq", "phase": 1, "loss": 3.293877273797989, "ce_loss": 3.2783377170562744, "logit_z_loss": 12.644819498062134, "router_aux_loss": 1.349678784608841, "router_z_loss": 0.7783399969339371, "routing_entropy_normalized": 0.6701283836889731, "learning_rate": 0.003, "grad_norm": 0.38227906823158264, "tokens_per_second": 4201431.090026268, "eta_seconds": 11929.728639125826, "data_wait_seconds": 1.958012580871582e-05} {"step": 4720, "total_steps": 28610, "tokens": 9898557440, "source": "fineweb_hq", "phase": 1, "loss": 3.3457182943820953, "ce_loss": 3.330268532037735, "logit_z_loss": 12.360141634941101, "router_aux_loss": 1.349607139825821, "router_z_loss": 0.7175642848014832, "routing_entropy_normalized": 0.6543452015279997, "learning_rate": 0.003, "grad_norm": 0.3755212724208832, "tokens_per_second": 3505138.7126075667, "eta_seconds": 14293.574488162994, "data_wait_seconds": 2.03549861907959e-05} {"step": 4730, "total_steps": 28610, "tokens": 9919528960, "source": "fineweb_hq", "phase": 1, "loss": 3.2857999205589294, "ce_loss": 3.2707449197769165, "logit_z_loss": 12.763757705688477, "router_aux_loss": 1.308594673871994, "router_z_loss": 0.6926750615239143, "routing_entropy_normalized": 0.6498509196151783, "learning_rate": 0.003, "grad_norm": 0.32600483298301697, "tokens_per_second": 3693832.6464615823, "eta_seconds": 13557.73110294342, "data_wait_seconds": 1.77919864654541e-05} {"step": 4740, "total_steps": 28610, "tokens": 9940500480, "source": "fineweb_hq", "phase": 1, "loss": 3.289323389530182, "ce_loss": 3.2735812664031982, "logit_z_loss": 12.57502269744873, "router_aux_loss": 1.3816202878952026, "router_z_loss": 0.6684145927429199, "routing_entropy_normalized": 0.6522116275551931, "learning_rate": 0.003, "grad_norm": 0.5183733701705933, "tokens_per_second": 4373459.359928419, "eta_seconds": 11446.092010974882, "data_wait_seconds": 2.041459083557129e-05} {"step": 4750, "total_steps": 28610, "tokens": 9961472000, "source": "fineweb_edu", "phase": 1, "loss": 3.4334212243556976, "ce_loss": 3.4165250658988953, "logit_z_loss": 11.836228609085083, "router_aux_loss": 1.4996642917394638, "router_z_loss": 0.71589570119977, "routing_entropy_normalized": 0.6461302709130996, "learning_rate": 0.003, "grad_norm": 0.39190366864204407, "tokens_per_second": 3849636.530146939, "eta_seconds": 12998.122375488281, "data_wait_seconds": 2.059340476989746e-05} {"step": 4760, "total_steps": 28610, "tokens": 9982443520, "source": "fineweb_hq", "phase": 1, "loss": 3.3575508296489716, "ce_loss": 3.341385155916214, "logit_z_loss": 11.984058141708374, "router_aux_loss": 1.4245383888483047, "router_z_loss": 0.721892386674881, "routing_entropy_normalized": 0.6398834860484238, "learning_rate": 0.003, "grad_norm": 0.37459954619407654, "tokens_per_second": 3801023.0513051013, "eta_seconds": 13158.845533132553, "data_wait_seconds": 1.9758939743041992e-05} {"step": 4770, "total_steps": 28610, "tokens": 10003415040, "source": "fineweb_edu", "phase": 1, "loss": 3.43318834900856, "ce_loss": 3.4174911975860596, "logit_z_loss": 12.264933228492737, "router_aux_loss": 1.3649634420871735, "router_z_loss": 0.8210862651467323, "routing_entropy_normalized": 0.6733076420864739, "learning_rate": 0.003, "grad_norm": 0.4561302959918976, "tokens_per_second": 4065161.4184804647, "eta_seconds": 12298.676124572754, "data_wait_seconds": 1.9878149032592773e-05} {"step": 4780, "total_steps": 28610, "tokens": 10024386560, "source": "fineweb_hq", "phase": 1, "loss": 3.3714566826820374, "ce_loss": 3.355970323085785, "logit_z_loss": 12.406161308288574, "router_aux_loss": 1.3597061038017273, "router_z_loss": 0.6487803235650063, "routing_entropy_normalized": 0.6734707612323543, "learning_rate": 0.003, "grad_norm": 0.7435541749000549, "tokens_per_second": 4183843.9512192546, "eta_seconds": 11944.788749933243, "data_wait_seconds": 1.895427703857422e-05} {"step": 4790, "total_steps": 28610, "tokens": 10045358080, "source": "fineweb_edu", "phase": 1, "loss": 3.4258804619312286, "ce_loss": 3.410384029150009, "logit_z_loss": 12.200166940689087, "router_aux_loss": 1.3528901785612106, "router_z_loss": 0.747522845864296, "routing_entropy_normalized": 0.658634488308548, "learning_rate": 0.003, "grad_norm": 0.5410071611404419, "tokens_per_second": 3943955.8069126736, "eta_seconds": 12666.004155635834, "data_wait_seconds": 1.806020736694336e-05} {"step": 4800, "total_steps": 28610, "tokens": 10066329600, "source": "fineweb_hq", "phase": 1, "loss": 3.3683177530765533, "ce_loss": 3.353163182735443, "logit_z_loss": 12.30478286743164, "router_aux_loss": 1.3267437815666199, "router_z_loss": 0.656526580452919, "routing_entropy_normalized": 0.6535569023329578, "learning_rate": 0.003, "grad_norm": 0.42889776825904846, "tokens_per_second": 3148857.334933568, "eta_seconds": 15857.558412075043, "data_wait_seconds": 0.11765968799591064} {"step": 4810, "total_steps": 28610, "tokens": 10087301120, "source": "fineweb_hq", "phase": 1, "loss": 3.2833532094955444, "ce_loss": 3.2679843604564667, "logit_z_loss": 12.216208577156067, "router_aux_loss": 1.354117289185524, "router_z_loss": 0.6060675978660583, "routing_entropy_normalized": 0.6645582428172097, "learning_rate": 0.003, "grad_norm": 0.4224037528038025, "tokens_per_second": 3149202.1952270614, "eta_seconds": 15849.162583351135, "data_wait_seconds": 1.9490718841552734e-05} {"step": 4820, "total_steps": 28610, "tokens": 10108272640, "source": "fineweb_hq", "phase": 1, "loss": 3.354600250720978, "ce_loss": 3.3386246860027313, "logit_z_loss": 12.350662350654602, "router_aux_loss": 1.4158696830272675, "router_z_loss": 0.5818301141262054, "routing_entropy_normalized": 0.6417552863610905, "learning_rate": 0.003, "grad_norm": 0.37672901153564453, "tokens_per_second": 4131777.8088149484, "eta_seconds": 12075.007028102875, "data_wait_seconds": 1.9818544387817383e-05} {"step": 4830, "total_steps": 28610, "tokens": 10129244160, "source": "fineweb_hq", "phase": 1, "loss": 3.349292904138565, "ce_loss": 3.3334764540195465, "logit_z_loss": 12.437930583953857, "router_aux_loss": 1.3961677104234695, "router_z_loss": 0.610958032310009, "routing_entropy_normalized": 0.6505516965306096, "learning_rate": 0.003, "grad_norm": 0.3589392900466919, "tokens_per_second": 2696666.9910873584, "eta_seconds": 18493.30107307434, "data_wait_seconds": 2.47955322265625e-05} {"step": 4840, "total_steps": 28610, "tokens": 10150215680, "source": "fineweb_hq", "phase": 1, "loss": 3.3319079279899597, "ce_loss": 3.316407859325409, "logit_z_loss": 12.76569139957428, "router_aux_loss": 1.3574651777744293, "router_z_loss": 0.6488271653652191, "routing_entropy_normalized": 0.645042393723894, "learning_rate": 0.003, "grad_norm": 0.3464262783527374, "tokens_per_second": 4172067.3191743833, "eta_seconds": 11948.345802307129, "data_wait_seconds": 2.1278858184814453e-05} {"step": 4850, "total_steps": 28610, "tokens": 10171187200, "source": "fineweb_edu", "phase": 1, "loss": 3.4237618148326874, "ce_loss": 3.4079009294509888, "logit_z_loss": 12.282296895980835, "router_aux_loss": 1.390365719795227, "router_z_loss": 0.7290244698524475, "routing_entropy_normalized": 0.6508682364511647, "learning_rate": 0.003, "grad_norm": 0.5026789903640747, "tokens_per_second": 4253774.639670024, "eta_seconds": 11713.90958404541, "data_wait_seconds": 1.963973045349121e-05} {"step": 4860, "total_steps": 28610, "tokens": 10192158720, "source": "fineweb_hq", "phase": 1, "loss": 3.278648167848587, "ce_loss": 3.2625746726989746, "logit_z_loss": 12.695683002471924, "router_aux_loss": 1.4160240441560745, "router_z_loss": 0.6436974555253983, "routing_entropy_normalized": 0.6461991132364505, "learning_rate": 0.003, "grad_norm": 0.407568097114563, "tokens_per_second": 3713432.364554557, "eta_seconds": 13412.755399942398, "data_wait_seconds": 1.958012580871582e-05} {"step": 4870, "total_steps": 28610, "tokens": 10213130240, "source": "fineweb_edu", "phase": 1, "loss": 3.4323280453681946, "ce_loss": 3.416539639234543, "logit_z_loss": 12.287448048591614, "router_aux_loss": 1.3863065391778946, "router_z_loss": 0.696656309068203, "routing_entropy_normalized": 0.6467664761307347, "learning_rate": 0.003, "grad_norm": 0.6178172826766968, "tokens_per_second": 4396561.792186779, "eta_seconds": 11323.936938285826, "data_wait_seconds": 2.0235776901245117e-05} {"step": 4880, "total_steps": 28610, "tokens": 10234101760, "source": "fineweb_hq", "phase": 1, "loss": 3.315499484539032, "ce_loss": 3.299268811941147, "logit_z_loss": 12.744012355804443, "router_aux_loss": 1.4279789924621582, "router_z_loss": 0.6764921918511391, "routing_entropy_normalized": 0.6552174126782058, "learning_rate": 0.003, "grad_norm": 0.5182557106018066, "tokens_per_second": 3739343.248953284, "eta_seconds": 13308.598234176636, "data_wait_seconds": 1.9937753677368164e-05} {"step": 4890, "total_steps": 28610, "tokens": 10255073280, "source": "fineweb_edu", "phase": 1, "loss": 3.4085982143878937, "ce_loss": 3.3925260305404663, "logit_z_loss": 12.539075493812561, "router_aux_loss": 1.4069206267595291, "router_z_loss": 0.7490315362811089, "routing_entropy_normalized": 0.6421008050725081, "learning_rate": 0.003, "grad_norm": 0.5316855907440186, "tokens_per_second": 4122839.5500284135, "eta_seconds": 12065.578792572021, "data_wait_seconds": 2.3543834686279297e-05} {"step": 4900, "total_steps": 28610, "tokens": 10276044800, "source": "fineweb_hq", "phase": 1, "loss": 3.307048797607422, "ce_loss": 3.2910357415676117, "logit_z_loss": 12.537500619888306, "router_aux_loss": 1.413212165236473, "router_z_loss": 0.627247266471386, "routing_entropy_normalized": 0.6653374614181818, "learning_rate": 0.003, "grad_norm": 0.2638806402683258, "tokens_per_second": 3725249.816823916, "eta_seconds": 13347.688441038132, "data_wait_seconds": 1.96993350982666e-05} {"step": 4910, "total_steps": 28610, "tokens": 10297016320, "source": "fineweb_hq", "phase": 1, "loss": 3.273177295923233, "ce_loss": 3.258245825767517, "logit_z_loss": 12.568360447883606, "router_aux_loss": 1.3044552206993103, "router_z_loss": 0.6300891935825348, "routing_entropy_normalized": 0.6496486673110152, "learning_rate": 0.003, "grad_norm": 0.2762499749660492, "tokens_per_second": 3446094.180808099, "eta_seconds": 14422.85085439682, "data_wait_seconds": 2.193450927734375e-05} {"step": 4920, "total_steps": 28610, "tokens": 10317987840, "source": "fineweb_hq", "phase": 1, "loss": 3.3090912103652954, "ce_loss": 3.293607145547867, "logit_z_loss": 12.758487939834595, "router_aux_loss": 1.3585436940193176, "router_z_loss": 0.62279062718153, "routing_entropy_normalized": 0.6734995720098577, "learning_rate": 0.003, "grad_norm": 0.5810636878013611, "tokens_per_second": 4135857.9332815493, "eta_seconds": 12012.388162612915, "data_wait_seconds": 1.7821788787841797e-05} {"step": 4930, "total_steps": 28610, "tokens": 10338959360, "source": "fineweb_hq", "phase": 1, "loss": 3.3172414004802704, "ce_loss": 3.300818532705307, "logit_z_loss": 12.883114457130432, "router_aux_loss": 1.4574125409126282, "router_z_loss": 0.5604547262191772, "routing_entropy_normalized": 0.6492852918802794, "learning_rate": 0.003, "grad_norm": 0.453785240650177, "tokens_per_second": 3326371.3307519057, "eta_seconds": 14929.349258422852, "data_wait_seconds": 2.0384788513183594e-05} {"step": 4940, "total_steps": 28610, "tokens": 10359930880, "source": "fineweb_hq", "phase": 1, "loss": 3.3358956575393677, "ce_loss": 3.3192145824432373, "logit_z_loss": 12.918545961380005, "router_aux_loss": 1.4713000506162643, "router_z_loss": 0.6762940585613251, "routing_entropy_normalized": 0.6546201668233255, "learning_rate": 0.003, "grad_norm": 0.49596190452575684, "tokens_per_second": 4297521.047520914, "eta_seconds": 11550.749208927156, "data_wait_seconds": 3.126263618469238e-05} {"step": 4950, "total_steps": 28610, "tokens": 10380902400, "source": "fineweb_edu", "phase": 1, "loss": 3.4271988570690155, "ce_loss": 3.4109014570713043, "logit_z_loss": 12.412030577659607, "router_aux_loss": 1.4289503246545792, "router_z_loss": 0.7667147591710091, "routing_entropy_normalized": 0.6585880558474261, "learning_rate": 0.003, "grad_norm": 0.4094029366970062, "tokens_per_second": 3913262.54635739, "eta_seconds": 12679.603203773499, "data_wait_seconds": 1.8090009689331055e-05} {"step": 4960, "total_steps": 28610, "tokens": 10401873920, "source": "fineweb_hq", "phase": 1, "loss": 3.3263317942619324, "ce_loss": 3.3103387653827667, "logit_z_loss": 12.356897592544556, "router_aux_loss": 1.4004823714494705, "router_z_loss": 0.7525168806314468, "routing_entropy_normalized": 0.6525086353958877, "learning_rate": 0.003, "grad_norm": 0.5966920256614685, "tokens_per_second": 3337180.644924206, "eta_seconds": 14862.139655351639, "data_wait_seconds": 2.0503997802734375e-05} {"step": 4970, "total_steps": 28610, "tokens": 10422845440, "source": "fineweb_edu", "phase": 1, "loss": 3.4083881974220276, "ce_loss": 3.392684817314148, "logit_z_loss": 12.215384125709534, "router_aux_loss": 1.3644338846206665, "router_z_loss": 0.8374783471226692, "routing_entropy_normalized": 0.6509959845419888, "learning_rate": 0.003, "grad_norm": 0.3799130618572235, "tokens_per_second": 4142474.672095468, "eta_seconds": 11967.888087272644, "data_wait_seconds": 1.806020736694336e-05} {"step": 4980, "total_steps": 28610, "tokens": 10443816960, "source": "fineweb_hq", "phase": 1, "loss": 3.281491309404373, "ce_loss": 3.2656265795230865, "logit_z_loss": 12.096883654594421, "router_aux_loss": 1.3929420560598373, "router_z_loss": 0.7256311327219009, "routing_entropy_normalized": 0.6509790803695145, "learning_rate": 0.003, "grad_norm": 0.5758671164512634, "tokens_per_second": 3512473.169939569, "eta_seconds": 14108.492609739304, "data_wait_seconds": 1.4632940292358398e-05} {"step": 4990, "total_steps": 28610, "tokens": 10464788480, "source": "fineweb_edu", "phase": 1, "loss": 3.3956003189086914, "ce_loss": 3.3788877427577972, "logit_z_loss": 12.658131122589111, "router_aux_loss": 1.4656227976083755, "router_z_loss": 0.7905358746647835, "routing_entropy_normalized": 0.6339255020089792, "learning_rate": 0.003, "grad_norm": 0.48260173201560974, "tokens_per_second": 3304706.8778195237, "eta_seconds": 14989.144898891449, "data_wait_seconds": 0.15113979578018188} {"step": 5000, "total_steps": 28610, "tokens": 10485760000, "source": "fineweb_hq", "phase": 1, "loss": 3.297933280467987, "ce_loss": 3.281212657690048, "logit_z_loss": 12.793218851089478, "router_aux_loss": 1.462718054652214, "router_z_loss": 0.8140603452920914, "routing_entropy_normalized": 0.6318557415874617, "learning_rate": 0.003, "grad_norm": 0.2432081550359726, "tokens_per_second": 4243887.919028479, "eta_seconds": 11667.075018167496, "data_wait_seconds": 2.5093555450439453e-05} {"step": 5010, "total_steps": 28610, "tokens": 10506731520, "source": "fineweb_hq", "phase": 1, "loss": 3.2946042120456696, "ce_loss": 3.2779740691184998, "logit_z_loss": 12.527584791183472, "router_aux_loss": 1.4568768292665482, "router_z_loss": 0.808660164475441, "routing_entropy_normalized": 0.635657862835346, "learning_rate": 0.003, "grad_norm": 0.4609832465648651, "tokens_per_second": 3405926.6454982073, "eta_seconds": 14531.372032165527, "data_wait_seconds": 2.0235776901245117e-05} {"step": 5020, "total_steps": 28610, "tokens": 10527703040, "source": "fineweb_hq", "phase": 1, "loss": 3.215487062931061, "ce_loss": 3.198724329471588, "logit_z_loss": 12.327836751937866, "router_aux_loss": 1.467827096581459, "router_z_loss": 0.8517282232642174, "routing_entropy_normalized": 0.6392744826162199, "learning_rate": 0.003, "grad_norm": 0.5202143788337708, "tokens_per_second": 3991720.499658647, "eta_seconds": 12393.607138633728, "data_wait_seconds": 1.84476375579834e-05} {"step": 5030, "total_steps": 28610, "tokens": 10548674560, "source": "fineweb_hq", "phase": 1, "loss": 3.2748131453990936, "ce_loss": 3.2582045793533325, "logit_z_loss": 12.816724419593811, "router_aux_loss": 1.444328635931015, "router_z_loss": 0.8836599215865135, "routing_entropy_normalized": 0.646714794013217, "learning_rate": 0.003, "grad_norm": 0.44153839349746704, "tokens_per_second": 3773124.4688557736, "eta_seconds": 13106.072849750519, "data_wait_seconds": 1.773238182067871e-05} {"step": 5040, "total_steps": 28610, "tokens": 10569646080, "source": "fineweb_hq", "phase": 1, "loss": 3.285956382751465, "ce_loss": 3.26949805021286, "logit_z_loss": 12.59624719619751, "router_aux_loss": 1.4322919100522995, "router_z_loss": 0.8757672160863876, "routing_entropy_normalized": 0.6825742285729864, "learning_rate": 0.003, "grad_norm": 0.379084974527359, "tokens_per_second": 4474566.7482508505, "eta_seconds": 11046.851107835768, "data_wait_seconds": 1.990795135498047e-05} {"step": 5050, "total_steps": 28610, "tokens": 10590617600, "source": "fineweb_edu", "phase": 1, "loss": 3.3914892971515656, "ce_loss": 3.3749416172504425, "logit_z_loss": 12.446159839630127, "router_aux_loss": 1.4268830567598343, "router_z_loss": 1.0342378616333008, "routing_entropy_normalized": 0.6347177627922282, "learning_rate": 0.003, "grad_norm": 0.5031958818435669, "tokens_per_second": 3767934.4553434667, "eta_seconds": 13112.9937915802, "data_wait_seconds": 1.8149614334106445e-05} {"step": 5060, "total_steps": 28610, "tokens": 10611589120, "source": "fineweb_hq", "phase": 1, "loss": 3.2994289100170135, "ce_loss": 3.282511740922928, "logit_z_loss": 12.529007196426392, "router_aux_loss": 1.473300039768219, "router_z_loss": 0.9312753602862358, "routing_entropy_normalized": 0.641516103560247, "learning_rate": 0.003, "grad_norm": 0.2277464121580124, "tokens_per_second": 3789782.7753643803, "eta_seconds": 13031.862913370132, "data_wait_seconds": 1.8209218978881836e-05} {"step": 5070, "total_steps": 28610, "tokens": 10632560640, "source": "fineweb_edu", "phase": 1, "loss": 3.4154205322265625, "ce_loss": 3.398537814617157, "logit_z_loss": 12.614303708076477, "router_aux_loss": 1.4588835388422012, "router_z_loss": 1.0325125455856323, "routing_entropy_normalized": 0.6472540498644227, "learning_rate": 0.003, "grad_norm": 0.44660985469818115, "tokens_per_second": 3404933.74257448, "eta_seconds": 14498.654544353485, "data_wait_seconds": 1.633167266845703e-05} {"step": 5080, "total_steps": 28610, "tokens": 10653532160, "source": "fineweb_hq", "phase": 1, "loss": 3.2845217287540436, "ce_loss": 3.2681488692760468, "logit_z_loss": 12.51968264579773, "router_aux_loss": 1.4101549088954926, "router_z_loss": 1.0193423554301262, "routing_entropy_normalized": 0.644336699462154, "learning_rate": 0.003, "grad_norm": 0.43575188517570496, "tokens_per_second": 3679838.7641093507, "eta_seconds": 13409.823017597198, "data_wait_seconds": 2.0772218704223633e-05} {"step": 5090, "total_steps": 28610, "tokens": 10674503680, "source": "fineweb_edu", "phase": 1, "loss": 3.4056678414344788, "ce_loss": 3.38933128118515, "logit_z_loss": 12.494250416755676, "router_aux_loss": 1.4063966572284698, "router_z_loss": 1.0230895802378654, "routing_entropy_normalized": 0.6483977829243122, "learning_rate": 0.003, "grad_norm": 0.47431719303131104, "tokens_per_second": 3483985.4701629467, "eta_seconds": 14157.640857696533, "data_wait_seconds": 2.187490463256836e-05} {"step": 5100, "total_steps": 28610, "tokens": 10695475200, "source": "fineweb_hq", "phase": 1, "loss": 3.3122613728046417, "ce_loss": 3.2963283956050873, "logit_z_loss": 12.765839695930481, "router_aux_loss": 1.369093880057335, "router_z_loss": 0.9654792100191116, "routing_entropy_normalized": 0.6562562452226074, "learning_rate": 0.003, "grad_norm": 0.4746202826499939, "tokens_per_second": 4210173.215113405, "eta_seconds": 11710.69241118431, "data_wait_seconds": 2.1129846572875977e-05} {"step": 5110, "total_steps": 28610, "tokens": 10716446720, "source": "fineweb_hq", "phase": 1, "loss": 3.253100037574768, "ce_loss": 3.238045781850815, "logit_z_loss": 12.635608911514282, "router_aux_loss": 1.290614902973175, "router_z_loss": 0.8845071792602539, "routing_entropy_normalized": 0.6644281150850526, "learning_rate": 0.003, "grad_norm": 0.2902168929576874, "tokens_per_second": 3482323.7012924724, "eta_seconds": 14152.352344989777, "data_wait_seconds": 2.16066837310791e-05} {"step": 5120, "total_steps": 28610, "tokens": 10737418240, "source": "fineweb_hq", "phase": 1, "loss": 3.258723199367523, "ce_loss": 3.2428961396217346, "logit_z_loss": 12.85658574104309, "router_aux_loss": 1.367190808057785, "router_z_loss": 0.8694380894303322, "routing_entropy_normalized": 0.6630908641873912, "learning_rate": 0.003, "grad_norm": 0.5920580625534058, "tokens_per_second": 3677386.657527089, "eta_seconds": 13395.953449487686, "data_wait_seconds": 2.0056962966918945e-05} {"step": 5130, "total_steps": 28610, "tokens": 10758389760, "source": "fineweb_hq", "phase": 1, "loss": 3.2915368378162384, "ce_loss": 3.275276839733124, "logit_z_loss": 12.57974112033844, "router_aux_loss": 1.4187930524349213, "router_z_loss": 0.8141098767518997, "routing_entropy_normalized": 0.6499797888071682, "learning_rate": 0.003, "grad_norm": 0.29987311363220215, "tokens_per_second": 4169268.1368038254, "eta_seconds": 11810.497033119202, "data_wait_seconds": 2.142786979675293e-05} {"step": 5140, "total_steps": 28610, "tokens": 10779361280, "source": "fineweb_hq", "phase": 1, "loss": 3.3393556773662567, "ce_loss": 3.322957456111908, "logit_z_loss": 12.378893256187439, "router_aux_loss": 1.4257098883390427, "router_z_loss": 0.9032843634486198, "routing_entropy_normalized": 0.6466229255220048, "learning_rate": 0.003, "grad_norm": 0.455963134765625, "tokens_per_second": 3401469.5160470824, "eta_seconds": 14470.262693166733, "data_wait_seconds": 1.9162893295288086e-05} {"step": 5150, "total_steps": 28610, "tokens": 10800332800, "source": "fineweb_edu", "phase": 1, "loss": 3.3894549906253815, "ce_loss": 3.3731422424316406, "logit_z_loss": 11.729880213737488, "router_aux_loss": 1.4166508316993713, "router_z_loss": 0.9731938019394875, "routing_entropy_normalized": 0.6498298949968813, "learning_rate": 0.003, "grad_norm": 0.5324581265449524, "tokens_per_second": 3804973.311486546, "eta_seconds": 12930.231539726257, "data_wait_seconds": 1.7940998077392578e-05} {"step": 5160, "total_steps": 28610, "tokens": 10821304320, "source": "fineweb_hq", "phase": 1, "loss": 3.273262232542038, "ce_loss": 3.25701704621315, "logit_z_loss": 12.555226564407349, "router_aux_loss": 1.403754472732544, "router_z_loss": 0.9521742835640907, "routing_entropy_normalized": 0.656757136382606, "learning_rate": 0.003, "grad_norm": 0.33652594685554504, "tokens_per_second": 3641729.4149169805, "eta_seconds": 13504.082483053207, "data_wait_seconds": 1.8149614334106445e-05} {"step": 5170, "total_steps": 28610, "tokens": 10842275840, "source": "fineweb_edu", "phase": 1, "loss": 3.3736424446105957, "ce_loss": 3.357626795768738, "logit_z_loss": 12.242414593696594, "router_aux_loss": 1.3862094581127167, "router_z_loss": 0.9293236583471298, "routing_entropy_normalized": 0.6504125907996977, "learning_rate": 0.003, "grad_norm": 0.41470247507095337, "tokens_per_second": 3778438.8400710626, "eta_seconds": 13009.93478012085, "data_wait_seconds": 2.1696090698242188e-05} {"step": 5180, "total_steps": 28610, "tokens": 10863247360, "source": "fineweb_hq", "phase": 1, "loss": 3.3062864542007446, "ce_loss": 3.2904224693775177, "logit_z_loss": 12.662072777748108, "router_aux_loss": 1.377205953001976, "router_z_loss": 0.8257222026586533, "routing_entropy_normalized": 0.660001470236628, "learning_rate": 0.003, "grad_norm": 0.5239468812942505, "tokens_per_second": 4144754.160016447, "eta_seconds": 11855.050857782364, "data_wait_seconds": 2.4050474166870117e-05} {"step": 5190, "total_steps": 28610, "tokens": 10884218880, "source": "fineweb_edu", "phase": 1, "loss": 3.3599395155906677, "ce_loss": 3.34429994225502, "logit_z_loss": 12.326236486434937, "router_aux_loss": 1.353246495127678, "router_z_loss": 0.874404102563858, "routing_entropy_normalized": 0.6662816313191038, "learning_rate": 0.003, "grad_norm": 0.5108093619346619, "tokens_per_second": 3963823.6967041804, "eta_seconds": 12390.889100551605, "data_wait_seconds": 2.60770320892334e-05} {"step": 5200, "total_steps": 28610, "tokens": 10905190400, "source": "fineweb_hq", "phase": 1, "loss": 3.309098869562149, "ce_loss": 3.294060558080673, "logit_z_loss": 12.482687711715698, "router_aux_loss": 1.299097403883934, "router_z_loss": 0.7990085184574127, "routing_entropy_normalized": 0.6807444788506027, "learning_rate": 0.003, "grad_norm": 0.3919590711593628, "tokens_per_second": 3825695.494042073, "eta_seconds": 12832.785148859024, "data_wait_seconds": 1.901388168334961e-05} {"step": 5210, "total_steps": 28610, "tokens": 10926161920, "source": "fineweb_hq", "phase": 1, "loss": 3.2604256570339203, "ce_loss": 3.244954615831375, "logit_z_loss": 12.884627103805542, "router_aux_loss": 1.342879056930542, "router_z_loss": 0.7537782043218613, "routing_entropy_normalized": 0.6246901961345379, "learning_rate": 0.003, "grad_norm": 0.39407020807266235, "tokens_per_second": 4120834.2803162723, "eta_seconds": 11908.597497940063, "data_wait_seconds": 2.0295381546020508e-05} {"step": 5220, "total_steps": 28610, "tokens": 10947133440, "source": "fineweb_hq", "phase": 1, "loss": 3.324093908071518, "ce_loss": 3.3077037930488586, "logit_z_loss": 12.292983889579773, "router_aux_loss": 1.4469757676124573, "router_z_loss": 0.6911028772592545, "routing_entropy_normalized": 0.6515446679197404, "learning_rate": 0.003, "grad_norm": 0.35581254959106445, "tokens_per_second": 3639111.30424767, "eta_seconds": 13479.220935821533, "data_wait_seconds": 1.5914440155029297e-05} {"step": 5230, "total_steps": 28610, "tokens": 10968104960, "source": "fineweb_hq", "phase": 1, "loss": 3.2779455184936523, "ce_loss": 3.2612459659576416, "logit_z_loss": 12.731591582298279, "router_aux_loss": 1.4686877578496933, "router_z_loss": 0.739512674510479, "routing_entropy_normalized": 0.6412500310706348, "learning_rate": 0.003, "grad_norm": 0.45572954416275024, "tokens_per_second": 4349744.125006249, "eta_seconds": 11272.25242471695, "data_wait_seconds": 1.633167266845703e-05} {"step": 5240, "total_steps": 28610, "tokens": 10989076480, "source": "fineweb_hq", "phase": 1, "loss": 3.3466416895389557, "ce_loss": 3.330011934041977, "logit_z_loss": 12.51885986328125, "router_aux_loss": 1.462307944893837, "router_z_loss": 0.7547608837485313, "routing_entropy_normalized": 0.6377133297362365, "learning_rate": 0.003, "grad_norm": 0.36275890469551086, "tokens_per_second": 3008713.3833715343, "eta_seconds": 16289.501855134964, "data_wait_seconds": 0.10336863994598389} {"step": 5250, "total_steps": 28610, "tokens": 11010048000, "source": "fineweb_edu", "phase": 1, "loss": 3.3949595391750336, "ce_loss": 3.3784141838550568, "logit_z_loss": 12.55772578716278, "router_aux_loss": 1.442522719502449, "router_z_loss": 0.8643551021814346, "routing_entropy_normalized": 0.6552153770223306, "learning_rate": 0.003, "grad_norm": 0.3578639626502991, "tokens_per_second": 3389787.0611963007, "eta_seconds": 14452.079093933105, "data_wait_seconds": 1.621246337890625e-05} {"step": 5260, "total_steps": 28610, "tokens": 11031019520, "source": "fineweb_hq", "phase": 1, "loss": 3.3275040090084076, "ce_loss": 3.311823606491089, "logit_z_loss": 12.859374642372131, "router_aux_loss": 1.3598552644252777, "router_z_loss": 0.7958997488021851, "routing_entropy_normalized": 0.6544580309110613, "learning_rate": 0.003, "grad_norm": 0.4390178322792053, "tokens_per_second": 2639456.493226262, "eta_seconds": 18552.493411302567, "data_wait_seconds": 0.19079157710075378} {"step": 5270, "total_steps": 28610, "tokens": 11051991040, "source": "fineweb_edu", "phase": 1, "loss": 3.3578770458698273, "ce_loss": 3.3420291543006897, "logit_z_loss": 12.606977105140686, "router_aux_loss": 1.375533178448677, "router_z_loss": 0.8318813666701317, "routing_entropy_normalized": 0.6618653453815062, "learning_rate": 0.003, "grad_norm": 0.48107224702835083, "tokens_per_second": 4180169.7913254313, "eta_seconds": 11709.45921421051, "data_wait_seconds": 1.868605613708496e-05} {"step": 5280, "total_steps": 28610, "tokens": 11072962560, "source": "fineweb_hq", "phase": 1, "loss": 3.3593626022338867, "ce_loss": 3.342998206615448, "logit_z_loss": 12.664140939712524, "router_aux_loss": 1.437905728816986, "router_z_loss": 0.7189088836312294, "routing_entropy_normalized": 0.6425433499595289, "learning_rate": 0.003, "grad_norm": 0.48335546255111694, "tokens_per_second": 3182180.520995151, "eta_seconds": 15375.166756629944, "data_wait_seconds": 1.9788742065429688e-05} {"step": 5290, "total_steps": 28610, "tokens": 11093934080, "source": "fineweb_edu", "phase": 1, "loss": 3.3787961304187775, "ce_loss": 3.3623503148555756, "logit_z_loss": 12.487868309020996, "router_aux_loss": 1.4365381002426147, "router_z_loss": 0.8315791189670563, "routing_entropy_normalized": 0.6250183343135598, "learning_rate": 0.003, "grad_norm": 0.5162041187286377, "tokens_per_second": 3402354.191540381, "eta_seconds": 14374.042761802673, "data_wait_seconds": 1.9103288650512695e-05} {"step": 5300, "total_steps": 28610, "tokens": 11114905600, "source": "fineweb_hq", "phase": 1, "loss": 3.3505688309669495, "ce_loss": 3.333579659461975, "logit_z_loss": 12.829614400863647, "router_aux_loss": 1.4884883761405945, "router_z_loss": 0.8214119970798492, "routing_entropy_normalized": 0.6368012460570004, "learning_rate": 0.003, "grad_norm": 0.541120707988739, "tokens_per_second": 3263846.4781695283, "eta_seconds": 14977.607999324799, "data_wait_seconds": 1.633167266845703e-05} {"step": 5310, "total_steps": 28610, "tokens": 11135877120, "source": "fineweb_hq", "phase": 1, "loss": 3.320593237876892, "ce_loss": 3.3037258684635162, "logit_z_loss": 12.744922995567322, "router_aux_loss": 1.471988469362259, "router_z_loss": 0.873019628226757, "routing_entropy_normalized": 0.634962436861535, "learning_rate": 0.003, "grad_norm": 0.39512649178504944, "tokens_per_second": 2913394.7012906075, "eta_seconds": 16772.06372976303, "data_wait_seconds": 0.09688442945480347} {"step": 5320, "total_steps": 28610, "tokens": 11156848640, "source": "fineweb_hq", "phase": 1, "loss": 3.330987334251404, "ce_loss": 3.315081924200058, "logit_z_loss": 12.615439295768738, "router_aux_loss": 1.3756073415279388, "router_z_loss": 0.8878669515252113, "routing_entropy_normalized": 0.6604925780563407, "learning_rate": 0.003, "grad_norm": 0.4742092490196228, "tokens_per_second": 3271318.292076022, "eta_seconds": 14930.577131032944, "data_wait_seconds": 1.9222497940063477e-05} {"step": 5330, "total_steps": 28610, "tokens": 11177820160, "source": "fineweb_hq", "phase": 1, "loss": 3.3517712354660034, "ce_loss": 3.3362543284893036, "logit_z_loss": 12.643990755081177, "router_aux_loss": 1.3335813879966736, "router_z_loss": 0.916731096804142, "routing_entropy_normalized": 0.6560335211779817, "learning_rate": 0.003, "grad_norm": 0.6000994443893433, "tokens_per_second": 2375460.3343389956, "eta_seconds": 20552.5210647583, "data_wait_seconds": 1.6510486602783203e-05} {"step": 5340, "total_steps": 28610, "tokens": 11198791680, "source": "fineweb_hq", "phase": 1, "loss": 3.3504068851470947, "ce_loss": 3.334491401910782, "logit_z_loss": 12.692072987556458, "router_aux_loss": 1.3747430443763733, "router_z_loss": 0.898827001452446, "routing_entropy_normalized": 0.6514809107490326, "learning_rate": 0.003, "grad_norm": 0.29917487502098083, "tokens_per_second": 3323372.2721726964, "eta_seconds": 14684.098874092102, "data_wait_seconds": 2.008676528930664e-05} {"step": 5350, "total_steps": 28610, "tokens": 11219763200, "source": "fineweb_edu", "phase": 1, "loss": 3.390911340713501, "ce_loss": 3.3748502135276794, "logit_z_loss": 12.399948000907898, "router_aux_loss": 1.3964449614286423, "router_z_loss": 0.8567138165235519, "routing_entropy_normalized": 0.6609206676617552, "learning_rate": 0.003, "grad_norm": 0.5637012720108032, "tokens_per_second": 4330471.026401858, "eta_seconds": 11264.307097911837, "data_wait_seconds": 1.9818544387817383e-05} {"step": 5360, "total_steps": 28610, "tokens": 11240734720, "source": "fineweb_hq", "phase": 1, "loss": 3.3453353345394135, "ce_loss": 3.32976570725441, "logit_z_loss": 12.38413393497467, "router_aux_loss": 1.35027414560318, "router_z_loss": 0.8284543305635452, "routing_entropy_normalized": 0.6568048300267557, "learning_rate": 0.003, "grad_norm": 0.36662817001342773, "tokens_per_second": 2913084.4035448246, "eta_seconds": 16737.85488009453, "data_wait_seconds": 2.0384788513183594e-05} {"step": 5370, "total_steps": 28610, "tokens": 11261706240, "source": "fineweb_edu", "phase": 1, "loss": 3.366256892681122, "ce_loss": 3.350040555000305, "logit_z_loss": 12.357618808746338, "router_aux_loss": 1.401888981461525, "router_z_loss": 0.9616526663303375, "routing_entropy_normalized": 0.6415093626314989, "learning_rate": 0.003, "grad_norm": 0.5929561853408813, "tokens_per_second": 2990103.298616752, "eta_seconds": 16299.708609580994, "data_wait_seconds": 2.154707908630371e-05} {"step": 5380, "total_steps": 28610, "tokens": 11282677760, "source": "fineweb_hq", "phase": 1, "loss": 3.3795888423919678, "ce_loss": 3.3626198768615723, "logit_z_loss": 12.76842987537384, "router_aux_loss": 1.4827501624822617, "router_z_loss": 0.8646024987101555, "routing_entropy_normalized": 0.6432225486406034, "learning_rate": 0.003, "grad_norm": 0.4522458016872406, "tokens_per_second": 3744689.3948835805, "eta_seconds": 13009.581255674362, "data_wait_seconds": 0.022463977336883545} {"step": 5390, "total_steps": 28610, "tokens": 11303649280, "source": "fineweb_edu", "phase": 1, "loss": 3.3635418117046356, "ce_loss": 3.3472939133644104, "logit_z_loss": 12.390018463134766, "router_aux_loss": 1.4118411093950272, "router_z_loss": 0.8905077949166298, "routing_entropy_normalized": 0.6513666899814404, "learning_rate": 0.003, "grad_norm": 0.4393056035041809, "tokens_per_second": 2199381.3698235415, "eta_seconds": 22140.711978435516, "data_wait_seconds": 1.6927719116210938e-05} {"step": 5400, "total_steps": 28610, "tokens": 11324620800, "source": "fineweb_hq", "phase": 1, "loss": 3.3630011081695557, "ce_loss": 3.347492128610611, "logit_z_loss": 12.435067653656006, "router_aux_loss": 1.3446918278932571, "router_z_loss": 0.8185461312532425, "routing_entropy_normalized": 0.6620860637889161, "learning_rate": 0.003, "grad_norm": 0.6197723150253296, "tokens_per_second": 3383279.392422373, "eta_seconds": 14386.898708105087, "data_wait_seconds": 2.2590160369873047e-05} {"step": 5410, "total_steps": 28610, "tokens": 11345592320, "source": "fineweb_hq", "phase": 1, "loss": 3.3406769037246704, "ce_loss": 3.32535457611084, "logit_z_loss": 12.073942065238953, "router_aux_loss": 1.3340333551168442, "router_z_loss": 0.7746180146932602, "routing_entropy_normalized": 0.6682180883917344, "learning_rate": 0.003, "grad_norm": 0.26599040627479553, "tokens_per_second": 4283343.49807069, "eta_seconds": 11358.866367340088, "data_wait_seconds": 1.850724220275879e-05} {"step": 5420, "total_steps": 28610, "tokens": 11366563840, "source": "fineweb_hq", "phase": 1, "loss": 3.3880825340747833, "ce_loss": 3.3724593222141266, "logit_z_loss": 12.503469824790955, "router_aux_loss": 1.3676394075155258, "router_z_loss": 0.6964512839913368, "routing_entropy_normalized": 0.6374032874153662, "learning_rate": 0.003, "grad_norm": 0.482369065284729, "tokens_per_second": 3239220.544177404, "eta_seconds": 15013.783166885376, "data_wait_seconds": 1.2755393981933594e-05} {"step": 5430, "total_steps": 28610, "tokens": 11387535360, "source": "fineweb_hq", "phase": 1, "loss": 3.355492115020752, "ce_loss": 3.340419739484787, "logit_z_loss": 12.575637221336365, "router_aux_loss": 1.3103392124176025, "router_z_loss": 0.7113424241542816, "routing_entropy_normalized": 0.6638455316353843, "learning_rate": 0.003, "grad_norm": 0.6146520376205444, "tokens_per_second": 3314035.8302440397, "eta_seconds": 14668.51472043991, "data_wait_seconds": 2.300739288330078e-05} {"step": 5440, "total_steps": 28610, "tokens": 11408506880, "source": "fineweb_hq", "phase": 1, "loss": 3.409644275903702, "ce_loss": 3.394364297389984, "logit_z_loss": 12.34232759475708, "router_aux_loss": 1.3395950496196747, "router_z_loss": 0.6497604176402092, "routing_entropy_normalized": 0.6505650345666204, "learning_rate": 0.003, "grad_norm": 0.9175111055374146, "tokens_per_second": 3707773.3185990667, "eta_seconds": 13105.17328453064, "data_wait_seconds": 2.041459083557129e-05} {"step": 5450, "total_steps": 28610, "tokens": 11429478400, "source": "fineweb_edu", "phase": 1, "loss": 3.421655595302582, "ce_loss": 3.4043318033218384, "logit_z_loss": 12.86171555519104, "router_aux_loss": 1.5244245827198029, "router_z_loss": 0.7933422550559044, "routing_entropy_normalized": 0.6336224145794889, "learning_rate": 0.003, "grad_norm": 0.5500855445861816, "tokens_per_second": 3233068.836184347, "eta_seconds": 15022.890875816345, "data_wait_seconds": 1.9252300262451172e-05} {"step": 5460, "total_steps": 28610, "tokens": 11450449920, "source": "fineweb_hq", "phase": 1, "loss": 3.3848394453525543, "ce_loss": 3.36742901802063, "logit_z_loss": 12.194815874099731, "router_aux_loss": 1.5311471670866013, "router_z_loss": 0.8794465437531471, "routing_entropy_normalized": 0.661619465706658, "learning_rate": 0.003, "grad_norm": 0.42652973532676697, "tokens_per_second": 4126329.372459304, "eta_seconds": 11765.679473876953, "data_wait_seconds": 1.7136335372924805e-05} {"step": 5470, "total_steps": 28610, "tokens": 11471421440, "source": "fineweb_edu", "phase": 1, "loss": 3.382046192884445, "ce_loss": 3.3650528490543365, "logit_z_loss": 12.440173268318176, "router_aux_loss": 1.4763472974300385, "router_z_loss": 0.9858546704053879, "routing_entropy_normalized": 0.6413221830039983, "learning_rate": 0.003, "grad_norm": 0.6176679730415344, "tokens_per_second": 3820231.6923469137, "eta_seconds": 12702.919924259186, "data_wait_seconds": 1.7344951629638672e-05} {"step": 5480, "total_steps": 28610, "tokens": 11492392960, "source": "fineweb_hq", "phase": 1, "loss": 3.3545597791671753, "ce_loss": 3.3378876745700836, "logit_z_loss": 12.562145471572876, "router_aux_loss": 1.4458422511816025, "router_z_loss": 0.9575250521302223, "routing_entropy_normalized": 0.6466078349491492, "learning_rate": 0.003, "grad_norm": 0.48205530643463135, "tokens_per_second": 3886435.89549875, "eta_seconds": 12481.133630990982, "data_wait_seconds": 2.0265579223632812e-05} {"step": 5490, "total_steps": 28610, "tokens": 11513364480, "source": "fineweb_edu", "phase": 1, "loss": 3.3843546211719513, "ce_loss": 3.368095099925995, "logit_z_loss": 12.756501317024231, "router_aux_loss": 1.4010449051856995, "router_z_loss": 0.9733594059944153, "routing_entropy_normalized": 0.6490429926775851, "learning_rate": 0.003, "grad_norm": 0.43533745408058167, "tokens_per_second": 4257596.286176178, "eta_seconds": 11388.152135848999, "data_wait_seconds": 1.7851591110229492e-05} {"step": 5500, "total_steps": 28610, "tokens": 11534336000, "source": "fineweb_hq", "phase": 1, "loss": 3.399473547935486, "ce_loss": 3.3833740055561066, "logit_z_loss": 12.598853349685669, "router_aux_loss": 1.4003838896751404, "router_z_loss": 0.8358140662312508, "routing_entropy_normalized": 0.6485319329698426, "learning_rate": 0.003, "grad_norm": 0.48797813057899475, "tokens_per_second": 3560836.372436714, "eta_seconds": 13610.617745637894, "data_wait_seconds": 3.2335519790649414e-05} {"step": 5510, "total_steps": 28610, "tokens": 11555307520, "source": "fineweb_hq", "phase": 1, "loss": 3.3744564950466156, "ce_loss": 3.3584140837192535, "logit_z_loss": 12.476292490959167, "router_aux_loss": 1.3977112472057343, "router_z_loss": 0.8177279531955719, "routing_entropy_normalized": 0.6403433303612873, "learning_rate": 0.003, "grad_norm": 0.5195225477218628, "tokens_per_second": 3943858.01822244, "eta_seconds": 12283.457207679749, "data_wait_seconds": 0.07013651728630066} {"step": 5520, "total_steps": 28610, "tokens": 11576279040, "source": "fineweb_hq", "phase": 1, "loss": 3.3523698151111603, "ce_loss": 3.3352872729301453, "logit_z_loss": 12.315889120101929, "router_aux_loss": 1.4922647178173065, "router_z_loss": 0.9283070862293243, "routing_entropy_normalized": 0.6323346435713364, "learning_rate": 0.003, "grad_norm": 0.37405624985694885, "tokens_per_second": 4324081.794779754, "eta_seconds": 11198.502243518828, "data_wait_seconds": 2.148747444152832e-05} {"step": 5530, "total_steps": 28610, "tokens": 11597250560, "source": "fineweb_hq", "phase": 1, "loss": 3.306126207113266, "ce_loss": 3.288932651281357, "logit_z_loss": 12.450718522071838, "router_aux_loss": 1.5007758736610413, "router_z_loss": 0.9407665953040123, "routing_entropy_normalized": 0.6493002354005147, "learning_rate": 0.003, "grad_norm": 0.33391663432121277, "tokens_per_second": 3217023.8236849667, "eta_seconds": 15045.666682243347, "data_wait_seconds": 1.8328428268432617e-05} {"step": 5540, "total_steps": 28610, "tokens": 11618222080, "source": "fineweb_hq", "phase": 1, "loss": 3.3934909403324127, "ce_loss": 3.3770452737808228, "logit_z_loss": 12.350868105888367, "router_aux_loss": 1.4196754544973373, "router_z_loss": 1.0137765482068062, "routing_entropy_normalized": 0.6477010577719042, "learning_rate": 0.003, "grad_norm": 0.5593708157539368, "tokens_per_second": 4500322.873881116, "eta_seconds": 10750.627898454666, "data_wait_seconds": 2.0295381546020508e-05} {"step": 5550, "total_steps": 28610, "tokens": 11639193600, "source": "fineweb_edu", "phase": 1, "loss": 3.398506373167038, "ce_loss": 3.382299065589905, "logit_z_loss": 12.213875412940979, "router_aux_loss": 1.3974119424819946, "router_z_loss": 1.0117965713143349, "routing_entropy_normalized": 0.6392861010235549, "learning_rate": 0.003, "grad_norm": 0.5216155052185059, "tokens_per_second": 3603182.475302014, "eta_seconds": 13421.558705806732, "data_wait_seconds": 2.1904706954956055e-05} {"step": 5560, "total_steps": 28610, "tokens": 11660165120, "source": "fineweb_hq", "phase": 1, "loss": 3.3514058589935303, "ce_loss": 3.335767984390259, "logit_z_loss": 12.52232277393341, "router_aux_loss": 1.3413386940956116, "router_z_loss": 0.9722336903214455, "routing_entropy_normalized": 0.6662017730908705, "learning_rate": 0.003, "grad_norm": 0.3571341633796692, "tokens_per_second": 3034902.6754709654, "eta_seconds": 15927.810137271881, "data_wait_seconds": 0.25474247336387634} {"step": 5570, "total_steps": 28610, "tokens": 11681136640, "source": "fineweb_edu", "phase": 1, "loss": 3.39544478058815, "ce_loss": 3.3796880543231964, "logit_z_loss": 12.192514896392822, "router_aux_loss": 1.3586775809526443, "router_z_loss": 0.9507719501852989, "routing_entropy_normalized": 0.6289191618100954, "learning_rate": 0.003, "grad_norm": 0.5858432650566101, "tokens_per_second": 3941579.659713888, "eta_seconds": 12258.633911132812, "data_wait_seconds": 1.9282102584838867e-05} {"step": 5580, "total_steps": 28610, "tokens": 11702108160, "source": "fineweb_hq", "phase": 1, "loss": 3.355494409799576, "ce_loss": 3.340017944574356, "logit_z_loss": 12.076537847518921, "router_aux_loss": 1.3405173271894455, "router_z_loss": 0.8636859506368637, "routing_entropy_normalized": 0.660223765099286, "learning_rate": 0.003, "grad_norm": 0.5140435099601746, "tokens_per_second": 3176134.749236053, "eta_seconds": 15206.34808444977, "data_wait_seconds": 1.7762184143066406e-05} {"step": 5590, "total_steps": 28610, "tokens": 11723079680, "source": "fineweb_edu", "phase": 1, "loss": 3.375388741493225, "ce_loss": 3.3594161570072174, "logit_z_loss": 12.074596166610718, "router_aux_loss": 1.3891522139310837, "router_z_loss": 0.8735749498009682, "routing_entropy_normalized": 0.6444632417880111, "learning_rate": 0.003, "grad_norm": 0.6010566353797913, "tokens_per_second": 1704567.874271504, "eta_seconds": 28321.805056095123, "data_wait_seconds": 1.5348196029663086e-05} {"step": 5600, "total_steps": 28610, "tokens": 11744051200, "source": "fineweb_hq", "phase": 1, "loss": 3.3627663254737854, "ce_loss": 3.34679251909256, "logit_z_loss": 12.397884845733643, "router_aux_loss": 1.3891737014055252, "router_z_loss": 0.8422518745064735, "routing_entropy_normalized": 0.6381934723540771, "learning_rate": 0.003, "grad_norm": 0.5237552523612976, "tokens_per_second": 2130775.7339883987, "eta_seconds": 22646.901196718216, "data_wait_seconds": 1.773238182067871e-05} {"step": 5610, "total_steps": 28610, "tokens": 11765022720, "source": "fineweb_hq", "phase": 1, "loss": 3.354930430650711, "ce_loss": 3.338635951280594, "logit_z_loss": 12.568628787994385, "router_aux_loss": 1.4258004575967789, "router_z_loss": 0.7796337679028511, "routing_entropy_normalized": 0.6523789486225631, "learning_rate": 0.003, "grad_norm": 0.389989972114563, "tokens_per_second": 2545541.449174502, "eta_seconds": 18948.61936569214, "data_wait_seconds": 1.9878149032592773e-05} {"step": 5620, "total_steps": 28610, "tokens": 11785994240, "source": "fineweb_hq", "phase": 1, "loss": 3.349622815847397, "ce_loss": 3.3333014845848083, "logit_z_loss": 12.558554649353027, "router_aux_loss": 1.4187154024839401, "router_z_loss": 0.8783659711480141, "routing_entropy_normalized": 0.6244321435781872, "learning_rate": 0.003, "grad_norm": 0.4449807405471802, "tokens_per_second": 3997968.049322659, "eta_seconds": 12059.507200956345, "data_wait_seconds": 2.518296241760254e-05} {"step": 5630, "total_steps": 28610, "tokens": 11806965760, "source": "fineweb_hq", "phase": 1, "loss": 3.3782459795475006, "ce_loss": 3.3621849715709686, "logit_z_loss": 12.319196939468384, "router_aux_loss": 1.4055095613002777, "router_z_loss": 0.7739823684096336, "routing_entropy_normalized": 0.6518561013273174, "learning_rate": 0.003, "grad_norm": 0.48843351006507874, "tokens_per_second": 3465292.649145666, "eta_seconds": 13907.210108757019, "data_wait_seconds": 2.053380012512207e-05} {"step": 5640, "total_steps": 28610, "tokens": 11827937280, "source": "fineweb_hq", "phase": 1, "loss": 3.386369973421097, "ce_loss": 3.3706073462963104, "logit_z_loss": 12.777168869972229, "router_aux_loss": 1.360469400882721, "router_z_loss": 0.88025813549757, "routing_entropy_normalized": 0.6481302799214732, "learning_rate": 0.003, "grad_norm": 0.5878011584281921, "tokens_per_second": 4058851.9099557195, "eta_seconds": 11868.277657985687, "data_wait_seconds": 1.5497207641601562e-05} {"step": 5650, "total_steps": 28610, "tokens": 11848908800, "source": "fineweb_edu", "phase": 1, "loss": 3.384739011526108, "ce_loss": 3.368742913007736, "logit_z_loss": 12.45527994632721, "router_aux_loss": 1.3736815601587296, "router_z_loss": 1.0137535855174065, "routing_entropy_normalized": 0.6565849011153275, "learning_rate": 0.003, "grad_norm": 0.4859798550605774, "tokens_per_second": 4023579.60231755, "eta_seconds": 11967.107570648193, "data_wait_seconds": 1.767277717590332e-05} {"step": 5660, "total_steps": 28610, "tokens": 11869880320, "source": "fineweb_hq", "phase": 1, "loss": 3.3544594645500183, "ce_loss": 3.3379854559898376, "logit_z_loss": 12.412187218666077, "router_aux_loss": 1.4356387704610825, "router_z_loss": 0.8763274475932121, "routing_entropy_normalized": 0.6311051186971034, "learning_rate": 0.003, "grad_norm": 0.3025311827659607, "tokens_per_second": 3642712.573887974, "eta_seconds": 13212.581949234009, "data_wait_seconds": 2.0742416381835938e-05} {"step": 5670, "total_steps": 28610, "tokens": 11890851840, "source": "fineweb_edu", "phase": 1, "loss": 3.4054959416389465, "ce_loss": 3.388267397880554, "logit_z_loss": 12.550119042396545, "router_aux_loss": 1.5028990656137466, "router_z_loss": 0.9446204602718353, "routing_entropy_normalized": 0.6456564476360822, "learning_rate": 0.003, "grad_norm": 0.45280176401138306, "tokens_per_second": 3325621.4045437276, "eta_seconds": 14466.068450927734, "data_wait_seconds": 1.817941665649414e-05} {"step": 5680, "total_steps": 28610, "tokens": 11911823360, "source": "fineweb_hq", "phase": 1, "loss": 3.3670021891593933, "ce_loss": 3.351082533597946, "logit_z_loss": 12.69284451007843, "router_aux_loss": 1.3788491934537888, "router_z_loss": 0.8618434071540833, "routing_entropy_normalized": 0.6608081481444403, "learning_rate": 0.003, "grad_norm": 0.49947378039360046, "tokens_per_second": 3292922.220887081, "eta_seconds": 14603.349892377853, "data_wait_seconds": 0.06620541214942932} {"step": 5690, "total_steps": 28610, "tokens": 11932794880, "source": "fineweb_edu", "phase": 1, "loss": 3.419891744852066, "ce_loss": 3.4042442739009857, "logit_z_loss": 12.358486294746399, "router_aux_loss": 1.354212448000908, "router_z_loss": 0.8694724664092064, "routing_entropy_normalized": 0.6779540784004487, "learning_rate": 0.003, "grad_norm": 0.6187684535980225, "tokens_per_second": 3589240.2855921276, "eta_seconds": 13391.893552780151, "data_wait_seconds": 2.0176172256469727e-05} {"step": 5700, "total_steps": 28610, "tokens": 11953766400, "source": "fineweb_hq", "phase": 1, "loss": 3.3620234429836273, "ce_loss": 3.346611440181732, "logit_z_loss": 12.659724473953247, "router_aux_loss": 1.3322025388479233, "router_z_loss": 0.8240195587277412, "routing_entropy_normalized": 0.6700095804111132, "learning_rate": 0.003, "grad_norm": 0.4911472201347351, "tokens_per_second": 3704185.2058732025, "eta_seconds": 12970.666867256165, "data_wait_seconds": 2.4527311325073242e-05} {"step": 5710, "total_steps": 28610, "tokens": 11974737920, "source": "fineweb_hq", "phase": 1, "loss": 3.3585753738880157, "ce_loss": 3.343413084745407, "logit_z_loss": 12.492822647094727, "router_aux_loss": 1.3173518031835556, "router_z_loss": 0.7394655793905258, "routing_entropy_normalized": 0.691494860659783, "learning_rate": 0.003, "grad_norm": 0.5166405439376831, "tokens_per_second": 3529855.5857319254, "eta_seconds": 13605.310368537903, "data_wait_seconds": 1.9282102584838867e-05} {"step": 5720, "total_steps": 28610, "tokens": 11995709440, "source": "fineweb_hq", "phase": 1, "loss": 3.3422595858573914, "ce_loss": 3.3264544904232025, "logit_z_loss": 12.94737446308136, "router_aux_loss": 1.3777815848588943, "router_z_loss": 0.7325894460082054, "routing_entropy_normalized": 0.6550729546641395, "learning_rate": 0.003, "grad_norm": 0.39507588744163513, "tokens_per_second": 3539544.076538652, "eta_seconds": 13562.144796609879, "data_wait_seconds": 1.895427703857422e-05} {"step": 5730, "total_steps": 28610, "tokens": 12016680960, "source": "dclm", "phase": 2, "loss": 3.5590109527111053, "ce_loss": 3.5436541736125946, "logit_z_loss": 12.393186569213867, "router_aux_loss": 1.3505403846502304, "router_z_loss": 0.611999087035656, "routing_entropy_normalized": 0.6920177666082804, "learning_rate": 0.003, "grad_norm": 0.6713780164718628, "tokens_per_second": 4681656.896862595, "eta_seconds": 10249.114537239075, "data_wait_seconds": 2.09808349609375e-05} {"step": 5740, "total_steps": 28610, "tokens": 12037652480, "source": "dclm", "phase": 2, "loss": 3.5842601656913757, "ce_loss": 3.569061130285263, "logit_z_loss": 12.384902000427246, "router_aux_loss": 1.3321940749883652, "router_z_loss": 0.638524129986763, "routing_entropy_normalized": 0.6620809093156853, "learning_rate": 0.003, "grad_norm": 0.8404759764671326, "tokens_per_second": 3933793.2171353074, "eta_seconds": 12192.26929140091, "data_wait_seconds": 1.8149614334106445e-05} {"step": 5750, "total_steps": 28610, "tokens": 12058624000, "source": "fineweb_hq", "phase": 2, "loss": 3.336348205804825, "ce_loss": 3.320369631052017, "logit_z_loss": 12.57720160484314, "router_aux_loss": 1.4010011851787567, "router_z_loss": 0.7108832895755768, "routing_entropy_normalized": 0.6599734880984466, "learning_rate": 0.003, "grad_norm": 0.4654478132724762, "tokens_per_second": 2925909.4197147395, "eta_seconds": 16384.955185890198, "data_wait_seconds": 1.7642974853515625e-05} {"step": 5760, "total_steps": 28610, "tokens": 12079595520, "source": "fineweb_hq", "phase": 2, "loss": 3.3304161727428436, "ce_loss": 3.313482016324997, "logit_z_loss": 12.665326476097107, "router_aux_loss": 1.4898833632469177, "router_z_loss": 0.7687878012657166, "routing_entropy_normalized": 0.6499452344989667, "learning_rate": 0.003, "grad_norm": 0.5789589285850525, "tokens_per_second": 4266692.3342368025, "eta_seconds": 11231.16443514824, "data_wait_seconds": 1.96993350982666e-05} {"step": 5770, "total_steps": 28610, "tokens": 12100567040, "source": "cosmopedia_v2", "phase": 2, "loss": 2.91300630569458, "ce_loss": 2.8954045176506042, "logit_z_loss": 14.057674050331116, "router_aux_loss": 1.500969335436821, "router_z_loss": 1.1863519698381424, "routing_entropy_normalized": 0.6401356206559895, "learning_rate": 0.003, "grad_norm": 0.7942115664482117, "tokens_per_second": 3874635.896530872, "eta_seconds": 12362.181365966797, "data_wait_seconds": 2.2530555725097656e-05} {"step": 5780, "total_steps": 28610, "tokens": 12121538560, "source": "dclm", "phase": 2, "loss": 3.5234088003635406, "ce_loss": 3.5060688257217407, "logit_z_loss": 12.579813003540039, "router_aux_loss": 1.5283867120742798, "router_z_loss": 0.7981204688549042, "routing_entropy_normalized": 0.6553643281329447, "learning_rate": 0.003, "grad_norm": 0.5364029407501221, "tokens_per_second": 3836223.8550798465, "eta_seconds": 12480.49695968628, "data_wait_seconds": 2.1845102310180664e-05} {"step": 5790, "total_steps": 28610, "tokens": 12142510080, "source": "cosmopedia_v2", "phase": 2, "loss": 2.9177363514900208, "ce_loss": 2.9007049202919006, "logit_z_loss": 14.042570352554321, "router_aux_loss": 1.422161489725113, "router_z_loss": 1.40552119910717, "routing_entropy_normalized": 0.6661624560173667, "learning_rate": 0.003, "grad_norm": 1.037182092666626, "tokens_per_second": 3701624.2249839907, "eta_seconds": 12928.651243686676, "data_wait_seconds": 1.7017126083374023e-05} {"step": 5800, "total_steps": 28610, "tokens": 12163481600, "source": "fineweb_hq", "phase": 2, "loss": 3.35433229804039, "ce_loss": 3.3381828665733337, "logit_z_loss": 12.582494497299194, "router_aux_loss": 1.3832016438245773, "router_z_loss": 1.059179648756981, "routing_entropy_normalized": 0.663234288101884, "learning_rate": 0.003, "grad_norm": 0.44748732447624207, "tokens_per_second": 4283184.564789275, "eta_seconds": 11168.334307432173, "data_wait_seconds": 1.96993350982666e-05} {"step": 5810, "total_steps": 28610, "tokens": 12184453120, "source": "fineweb_hq", "phase": 2, "loss": 3.3450080156326294, "ce_loss": 3.3287532031536102, "logit_z_loss": 13.066858649253845, "router_aux_loss": 1.39613938331604, "router_z_loss": 0.9868021905422211, "routing_entropy_normalized": 0.6560067725495532, "learning_rate": 0.003, "grad_norm": 0.528457760810852, "tokens_per_second": 3850523.110186439, "eta_seconds": 12417.810316085815, "data_wait_seconds": 2.1278858184814453e-05} {"step": 5820, "total_steps": 28610, "tokens": 12205424640, "source": "fineweb_edu", "phase": 2, "loss": 3.4112329185009003, "ce_loss": 3.395548850297928, "logit_z_loss": 12.734730005264282, "router_aux_loss": 1.3348520398139954, "router_z_loss": 1.0620802491903305, "routing_entropy_normalized": 0.6835064266869131, "learning_rate": 0.003, "grad_norm": 0.5913801789283752, "tokens_per_second": 4227910.935743211, "eta_seconds": 11304.423107862473, "data_wait_seconds": 2.136826515197754e-05} {"step": 5830, "total_steps": 28610, "tokens": 12226396160, "source": "dclm", "phase": 2, "loss": 3.428659677505493, "ce_loss": 3.4126760959625244, "logit_z_loss": 12.80112898349762, "router_aux_loss": 1.3870554566383362, "router_z_loss": 0.8328999355435371, "routing_entropy_normalized": 0.6586387341951968, "learning_rate": 0.003, "grad_norm": 0.5454291701316833, "tokens_per_second": 3166595.286621003, "eta_seconds": 15086.589297294617, "data_wait_seconds": 1.767277717590332e-05} {"step": 5840, "total_steps": 28610, "tokens": 12247367680, "source": "dclm", "phase": 2, "loss": 3.5257412791252136, "ce_loss": 3.510078489780426, "logit_z_loss": 12.419675469398499, "router_aux_loss": 1.3584729135036469, "router_z_loss": 0.8360722437500954, "routing_entropy_normalized": 0.6363985798653926, "learning_rate": 0.003, "grad_norm": 0.4190245270729065, "tokens_per_second": 4128082.5058716377, "eta_seconds": 11567.634845495224, "data_wait_seconds": 2.065300941467285e-05} {"step": 5850, "total_steps": 28610, "tokens": 12268339200, "source": "fineweb_hq", "phase": 2, "loss": 3.3494426608085632, "ce_loss": 3.333875000476837, "logit_z_loss": 13.182581663131714, "router_aux_loss": 1.332899495959282, "router_z_loss": 0.9204950705170631, "routing_entropy_normalized": 0.6657869741905996, "learning_rate": 0.003, "grad_norm": 0.32565510272979736, "tokens_per_second": 4020173.149991732, "eta_seconds": 11872.91634941101, "data_wait_seconds": 2.092123031616211e-05} {"step": 5860, "total_steps": 28610, "tokens": 12289310720, "source": "fineweb_hq", "phase": 2, "loss": 3.3931385576725006, "ce_loss": 3.377760022878647, "logit_z_loss": 12.568957448005676, "router_aux_loss": 1.3236546367406845, "router_z_loss": 0.88510312885046, "routing_entropy_normalized": 0.6671807445596182, "learning_rate": 0.003, "grad_norm": 0.44549888372421265, "tokens_per_second": 4144060.9522014763, "eta_seconds": 11512.911742925644, "data_wait_seconds": 2.300739288330078e-05} {"step": 5870, "total_steps": 28610, "tokens": 12310282240, "source": "cosmopedia_v2", "phase": 2, "loss": 2.883455902338028, "ce_loss": 2.867757648229599, "logit_z_loss": 14.008970022201538, "router_aux_loss": 1.3132318705320358, "router_z_loss": 1.1650442779064178, "routing_entropy_normalized": 0.668588858319408, "learning_rate": 0.003, "grad_norm": 0.661506175994873, "tokens_per_second": 3361704.9756750963, "eta_seconds": 14186.026681423187, "data_wait_seconds": 1.9073486328125e-05} {"step": 5880, "total_steps": 28610, "tokens": 12331253760, "source": "dclm", "phase": 2, "loss": 3.3685103058815002, "ce_loss": 3.353006660938263, "logit_z_loss": 12.234843969345093, "router_aux_loss": 1.3594803512096405, "router_z_loss": 0.6853439547121525, "routing_entropy_normalized": 0.6857246976650381, "learning_rate": 0.003, "grad_norm": 0.622573733329773, "tokens_per_second": 3575608.9069150714, "eta_seconds": 13331.509737491608, "data_wait_seconds": 2.0265579223632812e-05} {"step": 5890, "total_steps": 28610, "tokens": 12352225280, "source": "cosmopedia_v2", "phase": 2, "loss": 2.9055678844451904, "ce_loss": 2.8889684975147247, "logit_z_loss": 14.383576154708862, "router_aux_loss": 1.395412340760231, "router_z_loss": 1.2069524675607681, "routing_entropy_normalized": 0.6502556919216127, "learning_rate": 0.003, "grad_norm": 0.7664186954498291, "tokens_per_second": 3994191.9642157606, "eta_seconds": 11929.144584655762, "data_wait_seconds": 1.9788742065429688e-05} {"step": 5900, "total_steps": 28610, "tokens": 12373196800, "source": "fineweb_hq", "phase": 2, "loss": 3.366614043712616, "ce_loss": 3.3497177064418793, "logit_z_loss": 13.130178093910217, "router_aux_loss": 1.4724884033203125, "router_z_loss": 0.8584127798676491, "routing_entropy_normalized": 0.641499782705027, "learning_rate": 0.003, "grad_norm": 0.5780137181282043, "tokens_per_second": 3561621.4414664074, "eta_seconds": 13372.089848041534, "data_wait_seconds": 0.08915978670120239} {"step": 5910, "total_steps": 28610, "tokens": 12394168320, "source": "fineweb_hq", "phase": 2, "loss": 3.2938736379146576, "ce_loss": 3.2770113945007324, "logit_z_loss": 12.625999212265015, "router_aux_loss": 1.4673080742359161, "router_z_loss": 0.9266901612281799, "routing_entropy_normalized": 0.6664038308611557, "learning_rate": 0.003, "grad_norm": 0.5457227230072021, "tokens_per_second": 4144639.325345045, "eta_seconds": 11486.00557565689, "data_wait_seconds": 1.9878149032592773e-05} {"step": 5920, "total_steps": 28610, "tokens": 12415139840, "source": "fineweb_edu", "phase": 2, "loss": 3.3977011144161224, "ce_loss": 3.3803499937057495, "logit_z_loss": 12.878118634223938, "router_aux_loss": 1.5078697800636292, "router_z_loss": 0.984615832567215, "routing_entropy_normalized": 0.6555123776292626, "learning_rate": 0.003, "grad_norm": 0.4643368422985077, "tokens_per_second": 3556239.1778463037, "eta_seconds": 13380.5339012146, "data_wait_seconds": 1.6450881958007812e-05} {"step": 5930, "total_steps": 28610, "tokens": 12436111360, "source": "dclm", "phase": 2, "loss": 3.5375043153762817, "ce_loss": 3.520888090133667, "logit_z_loss": 12.686875224113464, "router_aux_loss": 1.4376347810029984, "router_z_loss": 0.9712174981832504, "routing_entropy_normalized": 0.6581670973140857, "learning_rate": 0.003, "grad_norm": 0.5759820938110352, "tokens_per_second": 4332679.3773381, "eta_seconds": 10977.827625274658, "data_wait_seconds": 0.010771065950393677} {"step": 5940, "total_steps": 28610, "tokens": 12457082880, "source": "dclm", "phase": 2, "loss": 3.560574024915695, "ce_loss": 3.5442999601364136, "logit_z_loss": 12.609327673912048, "router_aux_loss": 1.399818629026413, "router_z_loss": 1.015000432729721, "routing_entropy_normalized": 0.6669630126652576, "learning_rate": 0.003, "grad_norm": 0.5263603925704956, "tokens_per_second": 4008774.8319271933, "eta_seconds": 11859.592477321625, "data_wait_seconds": 2.3603439331054688e-05} {"step": 5950, "total_steps": 28610, "tokens": 12478054400, "source": "fineweb_hq", "phase": 2, "loss": 3.347253739833832, "ce_loss": 3.3308502435684204, "logit_z_loss": 12.953443050384521, "router_aux_loss": 1.4039841890335083, "router_z_loss": 1.0682490542531013, "routing_entropy_normalized": 0.6742647232572726, "learning_rate": 0.003, "grad_norm": 0.45798397064208984, "tokens_per_second": 3925304.4342208314, "eta_seconds": 12106.440434455872, "data_wait_seconds": 1.659989356994629e-05} {"step": 5960, "total_steps": 28610, "tokens": 12499025920, "source": "fineweb_hq", "phase": 2, "loss": 3.3299368023872375, "ce_loss": 3.3130545914173126, "logit_z_loss": 12.829283475875854, "router_aux_loss": 1.4567208886146545, "router_z_loss": 1.032060980796814, "routing_entropy_normalized": 0.6542724353696089, "learning_rate": 0.003, "grad_norm": 0.37820103764533997, "tokens_per_second": 4018999.405018926, "eta_seconds": 11818.984780311584, "data_wait_seconds": 1.7344951629638672e-05} {"step": 5970, "total_steps": 28610, "tokens": 12519997440, "source": "cosmopedia_v2", "phase": 2, "loss": 2.866221785545349, "ce_loss": 2.8492648005485535, "logit_z_loss": 13.879917621612549, "router_aux_loss": 1.4171693474054337, "router_z_loss": 1.397302359342575, "routing_entropy_normalized": 0.6909267474346827, "learning_rate": 0.003, "grad_norm": 0.5988554358482361, "tokens_per_second": 3291768.8992382833, "eta_seconds": 14423.710391998291, "data_wait_seconds": 2.053380012512207e-05} {"step": 5980, "total_steps": 28610, "tokens": 12540968960, "source": "dclm", "phase": 2, "loss": 3.4856895804405212, "ce_loss": 3.469380259513855, "logit_z_loss": 12.779417872428894, "router_aux_loss": 1.4140335470438004, "router_z_loss": 0.891067236661911, "routing_entropy_normalized": 0.6565653374808823, "learning_rate": 0.003, "grad_norm": 0.5448721051216125, "tokens_per_second": 4058237.8761116494, "eta_seconds": 11694.373570203781, "data_wait_seconds": 1.8924474716186523e-05} {"step": 5990, "total_steps": 28610, "tokens": 12561940480, "source": "cosmopedia_v2", "phase": 2, "loss": 2.867849349975586, "ce_loss": 2.8517485857009888, "logit_z_loss": 14.521131157875061, "router_aux_loss": 1.3269271552562714, "router_z_loss": 1.3793812841176987, "routing_entropy_normalized": 0.6764735248238001, "learning_rate": 0.003, "grad_norm": 0.5452903509140015, "tokens_per_second": 3834469.7583843158, "eta_seconds": 12371.352815151215, "data_wait_seconds": 1.8894672393798828e-05} {"step": 6000, "total_steps": 28610, "tokens": 12582912000, "source": "fineweb_hq", "phase": 2, "loss": 3.313222050666809, "ce_loss": 3.2979845702648163, "logit_z_loss": 12.754927396774292, "router_aux_loss": 1.2970239073038101, "router_z_loss": 0.9917862564325333, "routing_entropy_normalized": 0.6810658958958103, "learning_rate": 0.003, "grad_norm": 0.4151414632797241, "tokens_per_second": 3734511.249611311, "eta_seconds": 12696.870768547058, "data_wait_seconds": 2.1755695343017578e-05} {"step": 6010, "total_steps": 28610, "tokens": 12603883520, "source": "fineweb_hq", "phase": 2, "loss": 3.3260600566864014, "ce_loss": 3.3102763891220093, "logit_z_loss": 12.694620370864868, "router_aux_loss": 1.3581780642271042, "router_z_loss": 0.9324402883648872, "routing_entropy_normalized": 0.6609082128466514, "learning_rate": 0.003, "grad_norm": 0.6648704409599304, "tokens_per_second": 3720505.6070539635, "eta_seconds": 12739.030714035034, "data_wait_seconds": 2.2411346435546875e-05} {"step": 6020, "total_steps": 28610, "tokens": 12624855040, "source": "fineweb_edu", "phase": 2, "loss": 3.4044084548950195, "ce_loss": 3.3882731199264526, "logit_z_loss": 12.932868003845215, "router_aux_loss": 1.3891148120164871, "router_z_loss": 0.9508853629231453, "routing_entropy_normalized": 0.6552379399512889, "learning_rate": 0.003, "grad_norm": 0.5868209004402161, "tokens_per_second": 3670961.8562083254, "eta_seconds": 12905.245419502258, "data_wait_seconds": 2.059340476989746e-05} {"step": 6030, "total_steps": 28610, "tokens": 12645826560, "source": "dclm", "phase": 2, "loss": 3.549665540456772, "ce_loss": 3.5329145789146423, "logit_z_loss": 13.127610564231873, "router_aux_loss": 1.4685897082090378, "router_z_loss": 0.7522717416286469, "routing_entropy_normalized": 0.6321534020844183, "learning_rate": 0.003, "grad_norm": 0.44067662954330444, "tokens_per_second": 3256494.074012782, "eta_seconds": 14541.31071138382, "data_wait_seconds": 1.8417835235595703e-05} {"step": 6040, "total_steps": 28610, "tokens": 12666798080, "source": "dclm", "phase": 2, "loss": 3.5043670535087585, "ce_loss": 3.488001137971878, "logit_z_loss": 12.815292716026306, "router_aux_loss": 1.4120028764009476, "router_z_loss": 0.9643768072128296, "routing_entropy_normalized": 0.648633632426836, "learning_rate": 0.003, "grad_norm": 0.35719045996665955, "tokens_per_second": 3212905.8030280187, "eta_seconds": 14732.059867858885, "data_wait_seconds": 2.1219253540039062e-05} {"step": 6050, "total_steps": 28610, "tokens": 12687769600, "source": "fineweb_hq", "phase": 2, "loss": 3.3331617414951324, "ce_loss": 3.316842943429947, "logit_z_loss": 12.840189456939697, "router_aux_loss": 1.4037766605615616, "router_z_loss": 0.9969460740685463, "routing_entropy_normalized": 0.6663467477551461, "learning_rate": 0.003, "grad_norm": 0.5350555181503296, "tokens_per_second": 4211361.287172973, "eta_seconds": 11234.312587738039, "data_wait_seconds": 2.0295381546020508e-05} {"step": 6060, "total_steps": 28610, "tokens": 12708741120, "source": "fineweb_hq", "phase": 2, "loss": 3.2853043973445892, "ce_loss": 3.2688606083393097, "logit_z_loss": 12.84867799282074, "router_aux_loss": 1.4123759269714355, "router_z_loss": 1.035135492682457, "routing_entropy_normalized": 0.6514332548259257, "learning_rate": 0.003, "grad_norm": 0.3779699504375458, "tokens_per_second": 3432635.196809166, "eta_seconds": 13776.814280748367, "data_wait_seconds": 1.9073486328125e-05} {"step": 6070, "total_steps": 28610, "tokens": 12729712640, "source": "cosmopedia_v2", "phase": 2, "loss": 2.842558264732361, "ce_loss": 2.8252995908260345, "logit_z_loss": 14.793403029441833, "router_aux_loss": 1.4460050761699677, "router_z_loss": 1.3193246871232986, "routing_entropy_normalized": 0.6601105834356342, "learning_rate": 0.003, "grad_norm": 0.8572337031364441, "tokens_per_second": 3747674.992845578, "eta_seconds": 12613.101768493652, "data_wait_seconds": 2.0891427993774414e-05} {"step": 6080, "total_steps": 28610, "tokens": 12750684160, "source": "dclm", "phase": 2, "loss": 3.5399433076381683, "ce_loss": 3.523054987192154, "logit_z_loss": 12.509773850440979, "router_aux_loss": 1.4661178588867188, "router_z_loss": 0.9761543050408363, "routing_entropy_normalized": 0.6398245128514756, "learning_rate": 0.003, "grad_norm": 0.6581974625587463, "tokens_per_second": 4019690.711106659, "eta_seconds": 11754.345783233643, "data_wait_seconds": 2.2470951080322266e-05} {"step": 6090, "total_steps": 28610, "tokens": 12771655680, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8488996028900146, "ce_loss": 2.832276850938797, "logit_z_loss": 14.026976585388184, "router_aux_loss": 1.3762352615594864, "router_z_loss": 1.4577260613441467, "routing_entropy_normalized": 0.6713970771880469, "learning_rate": 0.003, "grad_norm": 0.43717390298843384, "tokens_per_second": 3926644.9335423713, "eta_seconds": 12027.535934448242, "data_wait_seconds": 2.3424625396728516e-05} {"step": 6100, "total_steps": 28610, "tokens": 12792627200, "source": "fineweb_hq", "phase": 2, "loss": 3.3169601559638977, "ce_loss": 3.30144339799881, "logit_z_loss": 13.265208721160889, "router_aux_loss": 1.3155200630426407, "router_z_loss": 1.034927286207676, "routing_entropy_normalized": 0.6876594743323621, "learning_rate": 0.003, "grad_norm": 0.5187311172485352, "tokens_per_second": 3577207.290547903, "eta_seconds": 13196.577018260956, "data_wait_seconds": 2.2083520889282227e-05} {"step": 6110, "total_steps": 28610, "tokens": 12813598720, "source": "fineweb_hq", "phase": 2, "loss": 3.293691337108612, "ce_loss": 3.2784233689308167, "logit_z_loss": 12.694559574127197, "router_aux_loss": 1.301377534866333, "router_z_loss": 0.9847749322652817, "routing_entropy_normalized": 0.6853147476048004, "learning_rate": 0.003, "grad_norm": 0.6237820386886597, "tokens_per_second": 4171260.6981476545, "eta_seconds": 11312.14839220047, "data_wait_seconds": 2.059340476989746e-05} {"step": 6120, "total_steps": 28610, "tokens": 12834570240, "source": "fineweb_edu", "phase": 2, "loss": 3.393764227628708, "ce_loss": 3.378193646669388, "logit_z_loss": 13.034140944480896, "router_aux_loss": 1.3337957859039307, "router_z_loss": 0.9291900843381882, "routing_entropy_normalized": 0.6463290674177476, "learning_rate": 0.003, "grad_norm": 0.569778561592102, "tokens_per_second": 3660217.9471987104, "eta_seconds": 12885.830614566803, "data_wait_seconds": 1.9311904907226562e-05} {"step": 6130, "total_steps": 28610, "tokens": 12855541760, "source": "dclm", "phase": 2, "loss": 3.5015271604061127, "ce_loss": 3.485777050256729, "logit_z_loss": 13.24044668674469, "router_aux_loss": 1.3700685054063797, "router_z_loss": 0.7254305854439735, "routing_entropy_normalized": 0.6434595234592334, "learning_rate": 0.003, "grad_norm": 0.4680023789405823, "tokens_per_second": 3848888.285233386, "eta_seconds": 12248.725727081299, "data_wait_seconds": 2.09808349609375e-05} {"step": 6140, "total_steps": 28610, "tokens": 12876513280, "source": "dclm", "phase": 2, "loss": 3.5019497871398926, "ce_loss": 3.4865405559539795, "logit_z_loss": 12.701767802238464, "router_aux_loss": 1.3417523056268692, "router_z_loss": 0.7215342819690704, "routing_entropy_normalized": 0.682261404721837, "learning_rate": 0.003, "grad_norm": 0.5196356177330017, "tokens_per_second": 3159712.4053458637, "eta_seconds": 14913.700803995132, "data_wait_seconds": 2.1129846572875977e-05} {"step": 6150, "total_steps": 28610, "tokens": 12897484800, "source": "fineweb_hq", "phase": 2, "loss": 3.2991060316562653, "ce_loss": 3.2837367057800293, "logit_z_loss": 12.808946371078491, "router_aux_loss": 1.3270449191331863, "router_z_loss": 0.8179320693016052, "routing_entropy_normalized": 0.6743704463646948, "learning_rate": 0.003, "grad_norm": 0.4335568845272064, "tokens_per_second": 3831669.7758540446, "eta_seconds": 12292.821844100952, "data_wait_seconds": 2.154707908630371e-05} {"step": 6160, "total_steps": 28610, "tokens": 12918456320, "source": "fineweb_hq", "phase": 2, "loss": 3.2753387689590454, "ce_loss": 3.2595438063144684, "logit_z_loss": 13.411545038223267, "router_aux_loss": 1.374412477016449, "router_z_loss": 0.7097005397081375, "routing_entropy_normalized": 0.6578536817681979, "learning_rate": 0.003, "grad_norm": 0.5052480697631836, "tokens_per_second": 3824595.2975315405, "eta_seconds": 12310.076946020126, "data_wait_seconds": 1.728534698486328e-05} {"step": 6170, "total_steps": 28610, "tokens": 12939427840, "source": "cosmopedia_v2", "phase": 2, "loss": 2.835167169570923, "ce_loss": 2.818506211042404, "logit_z_loss": 14.69764518737793, "router_aux_loss": 1.4142158478498459, "router_z_loss": 1.0491139143705368, "routing_entropy_normalized": 0.6599685497902028, "learning_rate": 0.003, "grad_norm": 0.6813830137252808, "tokens_per_second": 3823679.061182826, "eta_seconds": 12307.5420627594, "data_wait_seconds": 2.1010637283325195e-05} {"step": 6180, "total_steps": 28610, "tokens": 12960399360, "source": "dclm", "phase": 2, "loss": 3.4521341919898987, "ce_loss": 3.43542417883873, "logit_z_loss": 12.526985168457031, "router_aux_loss": 1.481002300977707, "router_z_loss": 0.6473202928900719, "routing_entropy_normalized": 0.6530404976523182, "learning_rate": 0.003, "grad_norm": 0.7166141271591187, "tokens_per_second": 3667365.3969839364, "eta_seconds": 12826.40649843216, "data_wait_seconds": 1.683831214904785e-05} {"step": 6190, "total_steps": 28610, "tokens": 12981370880, "source": "cosmopedia_v2", "phase": 2, "loss": 2.864746004343033, "ce_loss": 2.846910446882248, "logit_z_loss": 14.281024813652039, "router_aux_loss": 1.5161343216896057, "router_z_loss": 1.2460577487945557, "routing_entropy_normalized": 0.643289951835792, "learning_rate": 0.003, "grad_norm": 0.755374014377594, "tokens_per_second": 3299928.608175663, "eta_seconds": 14248.231832504272, "data_wait_seconds": 2.1517276763916016e-05} {"step": 6200, "total_steps": 28610, "tokens": 13002342400, "source": "fineweb_hq", "phase": 2, "loss": 3.292756676673889, "ce_loss": 3.2750827074050903, "logit_z_loss": 13.089236736297607, "router_aux_loss": 1.5385008752346039, "router_z_loss": 0.9800862222909927, "routing_entropy_normalized": 0.6259917842261418, "learning_rate": 0.003, "grad_norm": 0.5190271139144897, "tokens_per_second": 3368102.3640367053, "eta_seconds": 13953.606880187988, "data_wait_seconds": 2.1725893020629883e-05} {"step": 6210, "total_steps": 28610, "tokens": 13023313920, "source": "fineweb_hq", "phase": 2, "loss": 3.2914083302021027, "ce_loss": 3.2738836109638214, "logit_z_loss": 13.035016775131226, "router_aux_loss": 1.5200331062078476, "router_z_loss": 1.020863726735115, "routing_entropy_normalized": 0.6554536165387771, "learning_rate": 0.003, "grad_norm": 0.7294298410415649, "tokens_per_second": 4343067.618126141, "eta_seconds": 10816.365051269531, "data_wait_seconds": 2.2143125534057617e-05} {"step": 6220, "total_steps": 28610, "tokens": 13044285440, "source": "fineweb_edu", "phase": 2, "loss": 3.3888908326625824, "ce_loss": 3.371573328971863, "logit_z_loss": 12.832411170005798, "router_aux_loss": 1.4815343022346497, "router_z_loss": 1.2189844995737076, "routing_entropy_normalized": 0.6316722439578, "learning_rate": 0.003, "grad_norm": 0.6334714889526367, "tokens_per_second": 3562627.181410897, "eta_seconds": 13179.945834636688, "data_wait_seconds": 2.199411392211914e-05} {"step": 6230, "total_steps": 28610, "tokens": 13065256960, "source": "dclm", "phase": 2, "loss": 3.517748385667801, "ce_loss": 3.5007713437080383, "logit_z_loss": 12.277472138404846, "router_aux_loss": 1.4631748795509338, "router_z_loss": 1.1175396293401718, "routing_entropy_normalized": 0.6413749597256736, "learning_rate": 0.003, "grad_norm": 0.5779690742492676, "tokens_per_second": 3411243.917419296, "eta_seconds": 13758.694158554077, "data_wait_seconds": 1.9252300262451172e-05} {"step": 6240, "total_steps": 28610, "tokens": 13086228480, "source": "dclm", "phase": 2, "loss": 3.5000267028808594, "ce_loss": 3.4834465980529785, "logit_z_loss": 12.680023074150085, "router_aux_loss": 1.4152542650699615, "router_z_loss": 1.15951307117939, "routing_entropy_normalized": 0.6853837252208447, "learning_rate": 0.003, "grad_norm": 0.5252835750579834, "tokens_per_second": 3570464.1559219332, "eta_seconds": 13139.269347429276, "data_wait_seconds": 2.1278858184814453e-05} {"step": 6250, "total_steps": 28610, "tokens": 13107200000, "source": "fineweb_hq", "phase": 2, "loss": 3.2472569346427917, "ce_loss": 3.231046587228775, "logit_z_loss": 13.4331876039505, "router_aux_loss": 1.361549973487854, "router_z_loss": 1.2515451163053513, "routing_entropy_normalized": 0.6641435569937252, "learning_rate": 0.003, "grad_norm": 0.28912076354026794, "tokens_per_second": 4460883.859412847, "eta_seconds": 10511.889616012573, "data_wait_seconds": 1.862645149230957e-05} {"step": 6260, "total_steps": 28610, "tokens": 13128171520, "source": "fineweb_hq", "phase": 2, "loss": 3.2356547713279724, "ce_loss": 3.2200960516929626, "logit_z_loss": 12.939992070198059, "router_aux_loss": 1.311728522181511, "router_z_loss": 1.1473984271287918, "routing_entropy_normalized": 0.6737609420118693, "learning_rate": 0.003, "grad_norm": 0.4794347882270813, "tokens_per_second": 3454870.2224907395, "eta_seconds": 13566.74612402916, "data_wait_seconds": 2.1189451217651367e-05} {"step": 6270, "total_steps": 28610, "tokens": 13149143040, "source": "cosmopedia_v2", "phase": 2, "loss": 2.836608648300171, "ce_loss": 2.8208128213882446, "logit_z_loss": 14.506823420524597, "router_aux_loss": 1.3012364208698273, "router_z_loss": 1.3327601552009583, "routing_entropy_normalized": 0.6750361756562577, "learning_rate": 0.003, "grad_norm": 0.4120359718799591, "tokens_per_second": 4014670.014231937, "eta_seconds": 11669.794905662537, "data_wait_seconds": 2.333521842956543e-05} {"step": 6280, "total_steps": 28610, "tokens": 13170114560, "source": "dclm", "phase": 2, "loss": 3.531423807144165, "ce_loss": 3.5157853066921234, "logit_z_loss": 12.960204839706421, "router_aux_loss": 1.344276785850525, "router_z_loss": 0.8996226042509079, "routing_entropy_normalized": 0.6646453356989522, "learning_rate": 0.003, "grad_norm": 0.6806718707084656, "tokens_per_second": 4075628.0585805564, "eta_seconds": 11490.107410907745, "data_wait_seconds": 1.9371509552001953e-05} {"step": 6290, "total_steps": 28610, "tokens": 13191086080, "source": "cosmopedia_v2", "phase": 2, "loss": 2.9030431509017944, "ce_loss": 2.8868483006954193, "logit_z_loss": 13.958533644676208, "router_aux_loss": 1.3495781123638153, "router_z_loss": 1.3032662272453308, "routing_entropy_normalized": 0.6342679148641656, "learning_rate": 0.003, "grad_norm": 0.9801983833312988, "tokens_per_second": 3546067.7432101644, "eta_seconds": 13200.09543800354, "data_wait_seconds": 1.5079975128173828e-05} {"step": 6300, "total_steps": 28610, "tokens": 13212057600, "source": "fineweb_hq", "phase": 2, "loss": 3.270192861557007, "ce_loss": 3.2534172236919403, "logit_z_loss": 13.21971046924591, "router_aux_loss": 1.42974554002285, "router_z_loss": 1.1562298834323883, "routing_entropy_normalized": 0.6554839623451042, "learning_rate": 0.003, "grad_norm": 0.5627803206443787, "tokens_per_second": 4228984.808320262, "eta_seconds": 11063.520736217499, "data_wait_seconds": 2.232193946838379e-05} {"step": 6310, "total_steps": 28610, "tokens": 13233029120, "source": "fineweb_hq", "phase": 2, "loss": 3.215441554784775, "ce_loss": 3.198839783668518, "logit_z_loss": 13.21032440662384, "router_aux_loss": 1.4149174392223358, "router_z_loss": 1.1315761059522629, "routing_entropy_normalized": 0.6607094645948507, "learning_rate": 0.003, "grad_norm": 0.6113547086715698, "tokens_per_second": 3212747.8492999054, "eta_seconds": 14556.539072990417, "data_wait_seconds": 2.002716064453125e-05} {"step": 6320, "total_steps": 28610, "tokens": 13254000640, "source": "fineweb_edu", "phase": 2, "loss": 3.391054719686508, "ce_loss": 3.3746960759162903, "logit_z_loss": 13.001634001731873, "router_aux_loss": 1.3901539742946625, "router_z_loss": 1.15702123939991, "routing_entropy_normalized": 0.6671826521789254, "learning_rate": 0.003, "grad_norm": 0.3965660035610199, "tokens_per_second": 3444465.92534994, "eta_seconds": 13571.19480729103, "data_wait_seconds": 2.434849739074707e-05} {"step": 6330, "total_steps": 28610, "tokens": 13274972160, "source": "dclm", "phase": 2, "loss": 3.503522127866745, "ce_loss": 3.4870294332504272, "logit_z_loss": 12.878815770149231, "router_aux_loss": 1.4365839213132858, "router_z_loss": 0.8389439359307289, "routing_entropy_normalized": 0.6519401748258605, "learning_rate": 0.003, "grad_norm": 0.4712449908256531, "tokens_per_second": 3590638.6322538317, "eta_seconds": 13012.879140853882, "data_wait_seconds": 2.2113323211669922e-05} {"step": 6340, "total_steps": 28610, "tokens": 13295943680, "source": "dclm", "phase": 2, "loss": 3.511953294277191, "ce_loss": 3.494782954454422, "logit_z_loss": 13.074626326560974, "router_aux_loss": 1.4957919418811798, "router_z_loss": 0.9050290361046791, "routing_entropy_normalized": 0.6382853876814805, "learning_rate": 0.003, "grad_norm": 0.8325755596160889, "tokens_per_second": 3929919.24637849, "eta_seconds": 11884.105527877808, "data_wait_seconds": 2.1189451217651367e-05} {"step": 6350, "total_steps": 28610, "tokens": 13316915200, "source": "fineweb_hq", "phase": 2, "loss": 3.280375301837921, "ce_loss": 3.263871133327484, "logit_z_loss": 13.03217327594757, "router_aux_loss": 1.4137795567512512, "router_z_loss": 1.0631258189678192, "routing_entropy_normalized": 0.6447226231021503, "learning_rate": 0.003, "grad_norm": 0.35042911767959595, "tokens_per_second": 4241413.649862758, "eta_seconds": 11006.378385543825, "data_wait_seconds": 1.9848346710205078e-05} {"step": 6360, "total_steps": 28610, "tokens": 13337886720, "source": "fineweb_hq", "phase": 2, "loss": 3.282218724489212, "ce_loss": 3.265798419713974, "logit_z_loss": 13.380527138710022, "router_aux_loss": 1.407887414097786, "router_z_loss": 1.0033616572618484, "routing_entropy_normalized": 0.6300354292289502, "learning_rate": 0.003, "grad_norm": 0.3929239809513092, "tokens_per_second": 4153562.034751668, "eta_seconds": 11234.124255180359, "data_wait_seconds": 2.0444393157958984e-05} {"step": 6370, "total_steps": 28610, "tokens": 13358858240, "source": "cosmopedia_v2", "phase": 2, "loss": 2.863320082426071, "ce_loss": 2.847542941570282, "logit_z_loss": 14.116325497627258, "router_aux_loss": 1.29506453871727, "router_z_loss": 1.4149582982063293, "routing_entropy_normalized": 0.6932352949603752, "learning_rate": 0.003, "grad_norm": 0.7682394981384277, "tokens_per_second": 3406961.430865288, "eta_seconds": 13689.811706542969, "data_wait_seconds": 2.3812055587768555e-05} {"step": 6380, "total_steps": 28610, "tokens": 13379829760, "source": "dclm", "phase": 2, "loss": 3.466951936483383, "ce_loss": 3.451390862464905, "logit_z_loss": 12.355506777763367, "router_aux_loss": 1.347370222210884, "router_z_loss": 0.8518789932131767, "routing_entropy_normalized": 0.6622033690195378, "learning_rate": 0.003, "grad_norm": 0.7023378610610962, "tokens_per_second": 3657949.34858666, "eta_seconds": 12744.76066160202, "data_wait_seconds": 1.677870750427246e-05} {"step": 6390, "total_steps": 28610, "tokens": 13400801280, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8300823271274567, "ce_loss": 2.813980370759964, "logit_z_loss": 14.522603511810303, "router_aux_loss": 1.3245586305856705, "router_z_loss": 1.4042172282934189, "routing_entropy_normalized": 0.6835021765684789, "learning_rate": 0.003, "grad_norm": 0.8923595547676086, "tokens_per_second": 4117341.215337711, "eta_seconds": 11317.672012805939, "data_wait_seconds": 2.1189451217651367e-05} {"step": 6400, "total_steps": 28610, "tokens": 13421772800, "source": "fineweb_hq", "phase": 2, "loss": 3.2788558304309845, "ce_loss": 3.2627906501293182, "logit_z_loss": 13.138731360435486, "router_aux_loss": 1.3817563503980637, "router_z_loss": 0.9337617084383965, "routing_entropy_normalized": 0.6500042129255453, "learning_rate": 0.003, "grad_norm": 0.63957279920578, "tokens_per_second": 4212040.284898702, "eta_seconds": 11058.238471031189, "data_wait_seconds": 2.1278858184814453e-05} {"step": 6410, "total_steps": 28610, "tokens": 13442744320, "source": "fineweb_hq", "phase": 2, "loss": 3.2878508269786835, "ce_loss": 3.2715470492839813, "logit_z_loss": 13.250168323516846, "router_aux_loss": 1.4107922613620758, "router_z_loss": 0.8708213269710541, "routing_entropy_normalized": 0.6554101067462691, "learning_rate": 0.003, "grad_norm": 0.6149097681045532, "tokens_per_second": 3604374.5819406104, "eta_seconds": 12916.741404533386, "data_wait_seconds": 2.256035804748535e-05} {"step": 6420, "total_steps": 28610, "tokens": 13463715840, "source": "fineweb_edu", "phase": 2, "loss": 3.3873485028743744, "ce_loss": 3.3712258636951447, "logit_z_loss": 13.139894723892212, "router_aux_loss": 1.3853976875543594, "router_z_loss": 0.9546700939536095, "routing_entropy_normalized": 0.6489405383123235, "learning_rate": 0.003, "grad_norm": 0.7594156861305237, "tokens_per_second": 4077512.064462948, "eta_seconds": 11412.79342508316, "data_wait_seconds": 1.9937753677368164e-05} {"step": 6430, "total_steps": 28610, "tokens": 13484687360, "source": "dclm", "phase": 2, "loss": 3.4724697172641754, "ce_loss": 3.4553177654743195, "logit_z_loss": 13.23912525177002, "router_aux_loss": 1.5013363659381866, "router_z_loss": 0.8146772682666779, "routing_entropy_normalized": 0.6506935857765039, "learning_rate": 0.003, "grad_norm": 0.39741694927215576, "tokens_per_second": 3822638.49930258, "eta_seconds": 12168.252731323242, "data_wait_seconds": 1.8298625946044922e-05} {"step": 6440, "total_steps": 28610, "tokens": 13505658880, "source": "dclm", "phase": 2, "loss": 3.475320875644684, "ce_loss": 3.4593088030815125, "logit_z_loss": 13.100842952728271, "router_aux_loss": 1.379998356103897, "router_z_loss": 0.9020453318953514, "routing_entropy_normalized": 0.6641574468717036, "learning_rate": 0.003, "grad_norm": 0.4909760653972626, "tokens_per_second": 3732064.7766088396, "eta_seconds": 12457.945567131042, "data_wait_seconds": 2.300739288330078e-05} {"step": 6450, "total_steps": 28610, "tokens": 13526630400, "source": "fineweb_hq", "phase": 2, "loss": 3.2353700697422028, "ce_loss": 3.2198055386543274, "logit_z_loss": 13.215876340866089, "router_aux_loss": 1.3268928080797195, "router_z_loss": 0.9739541411399841, "routing_entropy_normalized": 0.6433964192908335, "learning_rate": 0.003, "grad_norm": 0.3815213739871979, "tokens_per_second": 3742253.772839004, "eta_seconds": 12418.422464370728, "data_wait_seconds": 2.1964311599731445e-05} {"step": 6460, "total_steps": 28610, "tokens": 13547601920, "source": "fineweb_hq", "phase": 2, "loss": 3.2421674728393555, "ce_loss": 3.2267896234989166, "logit_z_loss": 13.190651535987854, "router_aux_loss": 1.3131766468286514, "router_z_loss": 0.9270031899213791, "routing_entropy_normalized": 0.6407377784629327, "learning_rate": 0.003, "grad_norm": 0.45190906524658203, "tokens_per_second": 4071919.17492939, "eta_seconds": 11407.868084907532, "data_wait_seconds": 2.0623207092285156e-05} {"step": 6470, "total_steps": 28610, "tokens": 13568573440, "source": "cosmopedia_v2", "phase": 2, "loss": 2.840907782316208, "ce_loss": 2.825396031141281, "logit_z_loss": 14.371918439865112, "router_aux_loss": 1.2927110940217972, "router_z_loss": 1.1474729180335999, "routing_entropy_normalized": 0.6849897113456341, "learning_rate": 0.003, "grad_norm": 0.474264919757843, "tokens_per_second": 4390506.0365115935, "eta_seconds": 10575.306102275848, "data_wait_seconds": 1.7315149307250977e-05} {"step": 6480, "total_steps": 28610, "tokens": 13589544960, "source": "dclm", "phase": 2, "loss": 3.4773897528648376, "ce_loss": 3.4618948996067047, "logit_z_loss": 12.994903683662415, "router_aux_loss": 1.349514216184616, "router_z_loss": 0.7002071365714073, "routing_entropy_normalized": 0.6496364765422437, "learning_rate": 0.003, "grad_norm": 0.8362438678741455, "tokens_per_second": 3491857.7464147303, "eta_seconds": 13290.911924362183, "data_wait_seconds": 2.092123031616211e-05} {"step": 6490, "total_steps": 28610, "tokens": 13610516480, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8471704721450806, "ce_loss": 2.831265091896057, "logit_z_loss": 14.48341155052185, "router_aux_loss": 1.3325220346450806, "router_z_loss": 1.1317905187606812, "routing_entropy_normalized": 0.6604188298069605, "learning_rate": 0.003, "grad_norm": 0.9686075448989868, "tokens_per_second": 3257879.318628932, "eta_seconds": 14239.017993927002, "data_wait_seconds": 2.3245811462402344e-05} {"step": 6500, "total_steps": 28610, "tokens": 13631488000, "source": "fineweb_hq", "phase": 2, "loss": 3.2717684507369995, "ce_loss": 3.255400598049164, "logit_z_loss": 13.420369625091553, "router_aux_loss": 1.413272276520729, "router_z_loss": 0.8930977508425713, "routing_entropy_normalized": 0.6304182203338615, "learning_rate": 0.003, "grad_norm": 0.5718159675598145, "tokens_per_second": 4098721.5871260124, "eta_seconds": 11312.803208112717, "data_wait_seconds": 1.8149614334106445e-05} {"step": 6510, "total_steps": 28610, "tokens": 13652459520, "source": "fineweb_hq", "phase": 2, "loss": 3.294066220521927, "ce_loss": 3.277624487876892, "logit_z_loss": 13.196974396705627, "router_aux_loss": 1.4288998544216156, "router_z_loss": 0.8329784721136093, "routing_entropy_normalized": 0.6496021795997816, "learning_rate": 0.003, "grad_norm": 0.588320255279541, "tokens_per_second": 3436789.0001909826, "eta_seconds": 13485.570163726807, "data_wait_seconds": 1.7940998077392578e-05} {"step": 6520, "total_steps": 28610, "tokens": 13673431040, "source": "fineweb_edu", "phase": 2, "loss": 3.371222496032715, "ce_loss": 3.354939967393875, "logit_z_loss": 12.910089135169983, "router_aux_loss": 1.408818930387497, "router_z_loss": 0.903328649699688, "routing_entropy_normalized": 0.6456526967186379, "learning_rate": 0.003, "grad_norm": 0.5477581024169922, "tokens_per_second": 3158421.95554306, "eta_seconds": 14667.478991746902, "data_wait_seconds": 2.0623207092285156e-05} {"step": 6530, "total_steps": 28610, "tokens": 13694402560, "source": "dclm", "phase": 2, "loss": 3.5059306621551514, "ce_loss": 3.489964872598648, "logit_z_loss": 12.623801708221436, "router_aux_loss": 1.3975821137428284, "router_z_loss": 0.7275657504796982, "routing_entropy_normalized": 0.6515585556479748, "learning_rate": 0.003, "grad_norm": 0.6204957365989685, "tokens_per_second": 3884833.5785139706, "eta_seconds": 11919.459411621094, "data_wait_seconds": 2.3126602172851562e-05} {"step": 6540, "total_steps": 28610, "tokens": 13715374080, "source": "dclm", "phase": 2, "loss": 3.5361559092998505, "ce_loss": 3.520522177219391, "logit_z_loss": 13.027030229568481, "router_aux_loss": 1.3522356301546097, "router_z_loss": 0.808678574860096, "routing_entropy_normalized": 0.6461436399557278, "learning_rate": 0.003, "grad_norm": 0.5165647864341736, "tokens_per_second": 3862544.8395646876, "eta_seconds": 11982.810960769653, "data_wait_seconds": 2.065300941467285e-05} {"step": 6550, "total_steps": 28610, "tokens": 13736345600, "source": "fineweb_hq", "phase": 2, "loss": 3.4253864884376526, "ce_loss": 3.409612238407135, "logit_z_loss": 13.024187564849854, "router_aux_loss": 1.357381597161293, "router_z_loss": 0.8980335369706154, "routing_entropy_normalized": 0.6555890514737857, "learning_rate": 0.003, "grad_norm": 0.5673781633377075, "tokens_per_second": 3771831.8929147255, "eta_seconds": 12265.438766479492, "data_wait_seconds": 1.9490718841552734e-05} {"step": 6560, "total_steps": 28610, "tokens": 13757317120, "source": "fineweb_hq", "phase": 2, "loss": 3.391210526227951, "ce_loss": 3.3761198818683624, "logit_z_loss": 13.059623718261719, "router_aux_loss": 1.2956065386533737, "router_z_loss": 0.8286178931593895, "routing_entropy_normalized": 0.6846156806819091, "learning_rate": 0.003, "grad_norm": 0.41335633397102356, "tokens_per_second": 3794167.2336180713, "eta_seconds": 12187.70780324936, "data_wait_seconds": 2.378225326538086e-05} {"step": 6570, "total_steps": 28610, "tokens": 13778288640, "source": "cosmopedia_v2", "phase": 2, "loss": 2.874655693769455, "ce_loss": 2.859275996685028, "logit_z_loss": 14.111570715904236, "router_aux_loss": 1.2888990193605423, "router_z_loss": 1.0795185640454292, "routing_entropy_normalized": 0.6631258115749222, "learning_rate": 0.003, "grad_norm": 0.9120257496833801, "tokens_per_second": 4389729.510087713, "eta_seconds": 10529.402773857117, "data_wait_seconds": 1.8924474716186523e-05} {"step": 6580, "total_steps": 28610, "tokens": 13799260160, "source": "dclm", "phase": 2, "loss": 3.5296087563037872, "ce_loss": 3.513912171125412, "logit_z_loss": 12.977330923080444, "router_aux_loss": 1.3726868778467178, "router_z_loss": 0.6719816029071808, "routing_entropy_normalized": 0.674937097205858, "learning_rate": 0.003, "grad_norm": 0.6604069471359253, "tokens_per_second": 3477482.3429332995, "eta_seconds": 13285.54799246788, "data_wait_seconds": 2.422928810119629e-05} {"step": 6590, "total_steps": 28610, "tokens": 13820231680, "source": "cosmopedia_v2", "phase": 2, "loss": 2.859311103820801, "ce_loss": 2.842558890581131, "logit_z_loss": 14.836176753044128, "router_aux_loss": 1.4201315641403198, "router_z_loss": 1.0672393962740898, "routing_entropy_normalized": 0.6403633702338679, "learning_rate": 0.003, "grad_norm": 0.7865486145019531, "tokens_per_second": 3968775.9637515075, "eta_seconds": 11635.649747371674, "data_wait_seconds": 2.300739288330078e-05} {"step": 6600, "total_steps": 28610, "tokens": 13841203200, "source": "fineweb_hq", "phase": 2, "loss": 3.4345231354236603, "ce_loss": 3.418529212474823, "logit_z_loss": 13.135622024536133, "router_aux_loss": 1.3906527310609818, "router_z_loss": 0.7738608494400978, "routing_entropy_normalized": 0.6675570760732662, "learning_rate": 0.003, "grad_norm": 0.6426970958709717, "tokens_per_second": 4253290.449282175, "eta_seconds": 10852.377957820894, "data_wait_seconds": 2.300739288330078e-05} {"step": 6610, "total_steps": 28610, "tokens": 13862174720, "source": "fineweb_hq", "phase": 2, "loss": 3.3852862119674683, "ce_loss": 3.3699397146701813, "logit_z_loss": 13.268128991127014, "router_aux_loss": 1.3201733529567719, "router_z_loss": 0.8179413974285126, "routing_entropy_normalized": 0.6756378055067103, "learning_rate": 0.003, "grad_norm": 0.47917476296424866, "tokens_per_second": 3654543.974866177, "eta_seconds": 12624.65147972107, "data_wait_seconds": 2.822279930114746e-05} {"step": 6620, "total_steps": 28610, "tokens": 13883146240, "source": "fineweb_edu", "phase": 2, "loss": 3.4109683632850647, "ce_loss": 3.3957150876522064, "logit_z_loss": 12.781603217124939, "router_aux_loss": 1.3153626024723053, "router_z_loss": 0.8215760290622711, "routing_entropy_normalized": 0.6450525679865741, "learning_rate": 0.003, "grad_norm": 0.4789418578147888, "tokens_per_second": 4442055.285433374, "eta_seconds": 10381.764637470245, "data_wait_seconds": 2.4050474166870117e-05} {"step": 6630, "total_steps": 28610, "tokens": 13904117760, "source": "dclm", "phase": 2, "loss": 3.5664033591747284, "ce_loss": 3.5508198142051697, "logit_z_loss": 13.173288941383362, "router_aux_loss": 1.36060132086277, "router_z_loss": 0.6602772176265717, "routing_entropy_normalized": 0.6767482467278598, "learning_rate": 0.003, "grad_norm": 0.5427741408348083, "tokens_per_second": 3265159.928673566, "eta_seconds": 14117.348603725433, "data_wait_seconds": 1.9758939743041992e-05} {"step": 6640, "total_steps": 28610, "tokens": 13925089280, "source": "dclm", "phase": 2, "loss": 3.55757012963295, "ce_loss": 3.5413642525672913, "logit_z_loss": 12.747571110725403, "router_aux_loss": 1.4275410622358322, "router_z_loss": 0.6557593494653702, "routing_entropy_normalized": 0.661581681096747, "learning_rate": 0.003, "grad_norm": 0.9309183955192566, "tokens_per_second": 3958790.053446251, "eta_seconds": 11638.512984514236, "data_wait_seconds": 2.1010637283325195e-05} {"step": 6650, "total_steps": 28610, "tokens": 13946060800, "source": "fineweb_hq", "phase": 2, "loss": 3.378530442714691, "ce_loss": 3.3617347180843353, "logit_z_loss": 13.007081985473633, "router_aux_loss": 1.473630890250206, "router_z_loss": 0.7587311714887619, "routing_entropy_normalized": 0.6402673198282451, "learning_rate": 0.003, "grad_norm": 0.6721488237380981, "tokens_per_second": 3576321.1104790587, "eta_seconds": 12877.327425956726, "data_wait_seconds": 2.1576881408691406e-05} {"step": 6660, "total_steps": 28610, "tokens": 13967032320, "source": "fineweb_hq", "phase": 2, "loss": 3.421601891517639, "ce_loss": 3.404726505279541, "logit_z_loss": 13.099860906600952, "router_aux_loss": 1.4759171158075333, "router_z_loss": 0.8062352910637856, "routing_entropy_normalized": 0.6438144524712288, "learning_rate": 0.003, "grad_norm": 0.5426339507102966, "tokens_per_second": 3997210.989709154, "eta_seconds": 11516.151266098022, "data_wait_seconds": 2.35140323638916e-05} {"step": 6670, "total_steps": 28610, "tokens": 13988003840, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8302797973155975, "ce_loss": 2.8143161237239838, "logit_z_loss": 14.742899179458618, "router_aux_loss": 1.3259252607822418, "router_z_loss": 1.2301318943500519, "routing_entropy_normalized": 0.6898088174551622, "learning_rate": 0.003, "grad_norm": 0.460678368806839, "tokens_per_second": 4012391.307566384, "eta_seconds": 11467.354839801788, "data_wait_seconds": 1.9431114196777344e-05} {"step": 6680, "total_steps": 28610, "tokens": 14008975360, "source": "dclm", "phase": 2, "loss": 3.5439502894878387, "ce_loss": 3.5286833345890045, "logit_z_loss": 12.647682428359985, "router_aux_loss": 1.3193674683570862, "router_z_loss": 0.8086288943886757, "routing_entropy_normalized": 0.6612484332642811, "learning_rate": 0.003, "grad_norm": 0.7867238521575928, "tokens_per_second": 3957482.359766525, "eta_seconds": 11621.161935567856, "data_wait_seconds": 2.041459083557129e-05} {"step": 6690, "total_steps": 28610, "tokens": 14029946880, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8567948639392853, "ce_loss": 2.841352939605713, "logit_z_loss": 14.714291453361511, "router_aux_loss": 1.2833427786827087, "router_z_loss": 1.1370586305856705, "routing_entropy_normalized": 0.6681351931314563, "learning_rate": 0.003, "grad_norm": 0.5684453248977661, "tokens_per_second": 4436858.821012571, "eta_seconds": 10360.837181091309, "data_wait_seconds": 2.092123031616211e-05} {"step": 6700, "total_steps": 28610, "tokens": 14050918400, "source": "fineweb_hq", "phase": 2, "loss": 3.3976906538009644, "ce_loss": 3.3819739818573, "logit_z_loss": 13.085472702980042, "router_aux_loss": 1.3602170050144196, "router_z_loss": 0.8060220330953598, "routing_entropy_normalized": 0.6615216318896804, "learning_rate": 0.003, "grad_norm": 0.35996487736701965, "tokens_per_second": 3462400.064290689, "eta_seconds": 13270.73690700531, "data_wait_seconds": 1.683831214904785e-05} {"step": 6710, "total_steps": 28610, "tokens": 14071889920, "source": "fineweb_hq", "phase": 2, "loss": 3.379293292760849, "ce_loss": 3.3638564348220825, "logit_z_loss": 13.151451706886292, "router_aux_loss": 1.3380100280046463, "router_z_loss": 0.7416163086891174, "routing_entropy_normalized": 0.6708960958351785, "learning_rate": 0.003, "grad_norm": 0.5248592495918274, "tokens_per_second": 4301715.191617476, "eta_seconds": 10676.585211753845, "data_wait_seconds": 2.300739288330078e-05} {"step": 6720, "total_steps": 28610, "tokens": 14092861440, "source": "fineweb_edu", "phase": 2, "loss": 3.390935778617859, "ce_loss": 3.3756010234355927, "logit_z_loss": 13.031133532524109, "router_aux_loss": 1.3273271471261978, "router_z_loss": 0.7583688348531723, "routing_entropy_normalized": 0.6484306422157959, "learning_rate": 0.003, "grad_norm": 0.7143577337265015, "tokens_per_second": 3488944.46354397, "eta_seconds": 13157.749502658844, "data_wait_seconds": 2.1904706954956055e-05} {"step": 6730, "total_steps": 28610, "tokens": 14113832960, "source": "dclm", "phase": 2, "loss": 3.55969700217247, "ce_loss": 3.5441960394382477, "logit_z_loss": 13.175633192062378, "router_aux_loss": 1.362355187535286, "router_z_loss": 0.5598146505653858, "routing_entropy_normalized": 0.6412697628733648, "learning_rate": 0.003, "grad_norm": 0.7577191591262817, "tokens_per_second": 3423116.2910935003, "eta_seconds": 13404.652912139893, "data_wait_seconds": 2.1785497665405273e-05} {"step": 6740, "total_steps": 28610, "tokens": 14134804480, "source": "dclm", "phase": 2, "loss": 3.5619435608386993, "ce_loss": 3.5466344952583313, "logit_z_loss": 13.132412552833557, "router_aux_loss": 1.3401848077774048, "router_z_loss": 0.593936376273632, "routing_entropy_normalized": 0.6704671796887675, "learning_rate": 0.003, "grad_norm": 0.4173092246055603, "tokens_per_second": 2970735.231364499, "eta_seconds": 15438.842800855638, "data_wait_seconds": 2.002716064453125e-05} {"step": 6750, "total_steps": 28610, "tokens": 14155776000, "source": "fineweb_hq", "phase": 2, "loss": 3.3873350024223328, "ce_loss": 3.3720619082450867, "logit_z_loss": 13.24857234954834, "router_aux_loss": 1.3273525387048721, "router_z_loss": 0.6746730953454971, "routing_entropy_normalized": 0.6551171267456232, "learning_rate": 0.003, "grad_norm": 0.46117734909057617, "tokens_per_second": 4198016.483105453, "eta_seconds": 10920.334139823915, "data_wait_seconds": 2.0563602447509766e-05} {"step": 6760, "total_steps": 28610, "tokens": 14176747520, "source": "fineweb_hq", "phase": 2, "loss": 3.368105858564377, "ce_loss": 3.35238453745842, "logit_z_loss": 13.174682974815369, "router_aux_loss": 1.3707388937473297, "router_z_loss": 0.6963675990700722, "routing_entropy_normalized": 0.6521789372336245, "learning_rate": 0.003, "grad_norm": 0.4021610915660858, "tokens_per_second": 3336076.5871827137, "eta_seconds": 13735.527348518372, "data_wait_seconds": 1.8030405044555664e-05} {"step": 6770, "total_steps": 28610, "tokens": 14197719040, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8401920199394226, "ce_loss": 2.824627608060837, "logit_z_loss": 14.31963849067688, "router_aux_loss": 1.311204195022583, "router_z_loss": 1.0204066783189774, "routing_entropy_normalized": 0.6627180245584398, "learning_rate": 0.003, "grad_norm": 0.688086748123169, "tokens_per_second": 3483808.983699384, "eta_seconds": 13147.046779632568, "data_wait_seconds": 2.0503997802734375e-05} {"step": 6780, "total_steps": 28610, "tokens": 14218690560, "source": "dclm", "phase": 2, "loss": 3.4983648657798767, "ce_loss": 3.4821941256523132, "logit_z_loss": 13.365068197250366, "router_aux_loss": 1.4253048747777939, "router_z_loss": 0.5811726413667202, "routing_entropy_normalized": 0.6371200929109699, "learning_rate": 0.003, "grad_norm": 0.9029949903488159, "tokens_per_second": 3657289.1160035683, "eta_seconds": 12517.69458413124, "data_wait_seconds": 2.4020671844482422e-05} {"step": 6790, "total_steps": 28610, "tokens": 14239662080, "source": "cosmopedia_v2", "phase": 2, "loss": 2.854507088661194, "ce_loss": 2.837321639060974, "logit_z_loss": 14.965606331825256, "router_aux_loss": 1.4629022628068924, "router_z_loss": 1.0598567128181458, "routing_entropy_normalized": 0.652016835247535, "learning_rate": 0.003, "grad_norm": 0.6261730194091797, "tokens_per_second": 4389717.680263353, "eta_seconds": 10424.327934741974, "data_wait_seconds": 2.968311309814453e-05} {"step": 6800, "total_steps": 28610, "tokens": 14260633600, "source": "fineweb_hq", "phase": 2, "loss": 3.278484582901001, "ce_loss": 3.2617188692092896, "logit_z_loss": 13.269593715667725, "router_aux_loss": 1.4568780809640884, "router_z_loss": 0.8699670135974884, "routing_entropy_normalized": 0.6357232869346657, "learning_rate": 0.003, "grad_norm": 0.7025433778762817, "tokens_per_second": 3655485.6045615124, "eta_seconds": 12512.396455049515, "data_wait_seconds": 2.5659799575805664e-05} {"step": 6810, "total_steps": 28610, "tokens": 14281605120, "source": "fineweb_hq", "phase": 2, "loss": 3.344309687614441, "ce_loss": 3.3281308114528656, "logit_z_loss": 13.291084885597229, "router_aux_loss": 1.3958339840173721, "router_z_loss": 0.8914633020758629, "routing_entropy_normalized": 0.6337188185244768, "learning_rate": 0.003, "grad_norm": 0.7036110162734985, "tokens_per_second": 4115933.243697778, "eta_seconds": 11107.54497051239, "data_wait_seconds": 2.1249055862426758e-05} {"step": 6820, "total_steps": 28610, "tokens": 14302576640, "source": "fineweb_edu", "phase": 2, "loss": 3.3621041774749756, "ce_loss": 3.346172273159027, "logit_z_loss": 12.828744769096375, "router_aux_loss": 1.3707529455423355, "router_z_loss": 0.941549763083458, "routing_entropy_normalized": 0.6421335389039236, "learning_rate": 0.003, "grad_norm": 0.886034369468689, "tokens_per_second": 4357161.305848635, "eta_seconds": 10487.778365850449, "data_wait_seconds": 2.4646520614624023e-05} {"step": 6830, "total_steps": 28610, "tokens": 14323548160, "source": "dclm", "phase": 2, "loss": 3.5279093384742737, "ce_loss": 3.511542558670044, "logit_z_loss": 13.388079643249512, "router_aux_loss": 1.418197587132454, "router_z_loss": 0.846036970615387, "routing_entropy_normalized": 0.6391284331404028, "learning_rate": 0.003, "grad_norm": 0.6769534945487976, "tokens_per_second": 3734216.0450417427, "eta_seconds": 12231.742890357971, "data_wait_seconds": 2.0384788513183594e-05} {"step": 6840, "total_steps": 28610, "tokens": 14344519680, "source": "dclm", "phase": 2, "loss": 3.5506668984889984, "ce_loss": 3.5347159802913666, "logit_z_loss": 13.249883532524109, "router_aux_loss": 1.3731836825609207, "router_z_loss": 0.8941233232617378, "routing_entropy_normalized": 0.6528412678443231, "learning_rate": 0.003, "grad_norm": 0.856147050857544, "tokens_per_second": 3746536.384701328, "eta_seconds": 12185.921702623367, "data_wait_seconds": 2.2083520889282227e-05} {"step": 6850, "total_steps": 28610, "tokens": 14365491200, "source": "fineweb_hq", "phase": 2, "loss": 3.2967851161956787, "ce_loss": 3.2811256647109985, "logit_z_loss": 13.344186544418335, "router_aux_loss": 1.338004618883133, "router_z_loss": 0.9449781402945518, "routing_entropy_normalized": 0.6725848911923246, "learning_rate": 0.003, "grad_norm": 0.5474265813827515, "tokens_per_second": 3669305.0908600185, "eta_seconds": 12436.694793701172, "data_wait_seconds": 0.0649627149105072} {"step": 6860, "total_steps": 28610, "tokens": 14386462720, "source": "fineweb_hq", "phase": 2, "loss": 3.353950619697571, "ce_loss": 3.337957262992859, "logit_z_loss": 12.741276383399963, "router_aux_loss": 1.3782510161399841, "router_z_loss": 0.936685137450695, "routing_entropy_normalized": 0.6515612819187213, "learning_rate": 0.003, "grad_norm": 0.4653225541114807, "tokens_per_second": 3726815.8564457237, "eta_seconds": 12239.149385690689, "data_wait_seconds": 2.0205974578857422e-05} {"step": 6870, "total_steps": 28610, "tokens": 14407434240, "source": "cosmopedia_v2", "phase": 2, "loss": 2.830993175506592, "ce_loss": 2.8149654269218445, "logit_z_loss": 14.948400020599365, "router_aux_loss": 1.337561920285225, "router_z_loss": 1.1573130637407303, "routing_entropy_normalized": 0.6576553632408644, "learning_rate": 0.003, "grad_norm": 1.0924851894378662, "tokens_per_second": 3518353.753532166, "eta_seconds": 12958.357139110565, "data_wait_seconds": 2.3066997528076172e-05} {"step": 6880, "total_steps": 28610, "tokens": 14428405760, "source": "dclm", "phase": 2, "loss": 3.5375660955905914, "ce_loss": 3.5219669342041016, "logit_z_loss": 12.685245275497437, "router_aux_loss": 1.353526160120964, "router_z_loss": 0.7953969538211823, "routing_entropy_normalized": 0.6620193149085013, "learning_rate": 0.003, "grad_norm": 0.9017151594161987, "tokens_per_second": 4056935.144382424, "eta_seconds": 11232.891662836075, "data_wait_seconds": 2.3573637008666992e-05} {"step": 6890, "total_steps": 28610, "tokens": 14449377280, "source": "cosmopedia_v2", "phase": 2, "loss": 2.847534030675888, "ce_loss": 2.8311964571475983, "logit_z_loss": 14.550226211547852, "router_aux_loss": 1.366347387433052, "router_z_loss": 1.2190959602594376, "routing_entropy_normalized": 0.6644022286155042, "learning_rate": 0.003, "grad_norm": 0.6076905131340027, "tokens_per_second": 3420656.7771251793, "eta_seconds": 13316.197563171387, "data_wait_seconds": 1.6421079635620117e-05} {"step": 6900, "total_steps": 28610, "tokens": 14470348800, "source": "fineweb_hq", "phase": 2, "loss": 3.3282510638237, "ce_loss": 3.31247216463089, "logit_z_loss": 13.055169939994812, "router_aux_loss": 1.361343502998352, "router_z_loss": 0.8599359914660454, "routing_entropy_normalized": 0.6581302528376431, "learning_rate": 0.003, "grad_norm": 0.4458228349685669, "tokens_per_second": 3631401.240867562, "eta_seconds": 12537.631316423416, "data_wait_seconds": 1.4960765838623047e-05} {"step": 6910, "total_steps": 28610, "tokens": 14491320320, "source": "fineweb_hq", "phase": 2, "loss": 3.336813658475876, "ce_loss": 3.3217033743858337, "logit_z_loss": 13.230975985527039, "router_aux_loss": 1.2908243983983994, "router_z_loss": 0.8789526745676994, "routing_entropy_normalized": 0.653145499508405, "learning_rate": 0.003, "grad_norm": 0.5190775394439697, "tokens_per_second": 4428329.643370187, "eta_seconds": 10276.605868339539, "data_wait_seconds": 2.1457672119140625e-05} {"step": 6920, "total_steps": 28610, "tokens": 14512291840, "source": "fineweb_edu", "phase": 2, "loss": 3.3784848153591156, "ce_loss": 3.3626206815242767, "logit_z_loss": 13.100931406021118, "router_aux_loss": 1.3693226128816605, "router_z_loss": 0.8608430325984955, "routing_entropy_normalized": 0.6592014788492473, "learning_rate": 0.003, "grad_norm": 0.5145002007484436, "tokens_per_second": 3143014.8920044918, "eta_seconds": 14472.48213672638, "data_wait_seconds": 1.9490718841552734e-05} {"step": 6930, "total_steps": 28610, "tokens": 14533263360, "source": "dclm", "phase": 2, "loss": 3.5458984375, "ce_loss": 3.5300790071487427, "logit_z_loss": 13.644527792930603, "router_aux_loss": 1.3702026307582855, "router_z_loss": 0.7529056221246719, "routing_entropy_normalized": 0.6552152499363311, "learning_rate": 0.003, "grad_norm": 0.6779276728630066, "tokens_per_second": 3535834.362547179, "eta_seconds": 12858.706233978271, "data_wait_seconds": 1.8417835235595703e-05} {"step": 6940, "total_steps": 28610, "tokens": 14554234880, "source": "dclm", "phase": 2, "loss": 3.5675123035907745, "ce_loss": 3.5515533685684204, "logit_z_loss": 12.952207446098328, "router_aux_loss": 1.3853734135627747, "router_z_loss": 0.809969961643219, "routing_entropy_normalized": 0.6520580240014611, "learning_rate": 0.003, "grad_norm": 0.4730265736579895, "tokens_per_second": 4120192.0872669565, "eta_seconds": 11029.894450902939, "data_wait_seconds": 1.874566078186035e-05} {"step": 6950, "total_steps": 28610, "tokens": 14575206400, "source": "fineweb_hq", "phase": 2, "loss": 3.3473144471645355, "ce_loss": 3.3303850889205933, "logit_z_loss": 13.467204213142395, "router_aux_loss": 1.4737458527088165, "router_z_loss": 0.8451971709728241, "routing_entropy_normalized": 0.6238146026958398, "learning_rate": 0.003, "grad_norm": 0.6771045327186584, "tokens_per_second": 3311257.7704501227, "eta_seconds": 13718.14442396164, "data_wait_seconds": 0.2790810167789459} {"step": 6960, "total_steps": 28610, "tokens": 14596177920, "source": "fineweb_hq", "phase": 2, "loss": 3.3104735910892487, "ce_loss": 3.2945667505264282, "logit_z_loss": 13.030139207839966, "router_aux_loss": 1.3651028871536255, "router_z_loss": 0.9527845829725266, "routing_entropy_normalized": 0.6806382151468027, "learning_rate": 0.003, "grad_norm": 0.5173642635345459, "tokens_per_second": 4113371.004327645, "eta_seconds": 11037.988246679306, "data_wait_seconds": 2.446770668029785e-05} {"step": 6970, "total_steps": 28610, "tokens": 14617149440, "source": "cosmopedia_v2", "phase": 2, "loss": 2.832726776599884, "ce_loss": 2.816569834947586, "logit_z_loss": 14.527950763702393, "router_aux_loss": 1.3457878828048706, "router_z_loss": 1.2462908774614334, "routing_entropy_normalized": 0.657240140808079, "learning_rate": 0.003, "grad_norm": 0.4591177701950073, "tokens_per_second": 3818592.311499372, "eta_seconds": 11884.580907821655, "data_wait_seconds": 2.1576881408691406e-05} {"step": 6980, "total_steps": 28610, "tokens": 14638120960, "source": "dclm", "phase": 2, "loss": 3.455219328403473, "ce_loss": 3.4393800795078278, "logit_z_loss": 13.220626592636108, "router_aux_loss": 1.3749132752418518, "router_z_loss": 0.7680492624640465, "routing_entropy_normalized": 0.6389492530104499, "learning_rate": 0.003, "grad_norm": 0.8730395436286926, "tokens_per_second": 4301225.074433487, "eta_seconds": 10546.157658576967, "data_wait_seconds": 2.014636993408203e-05} {"step": 6990, "total_steps": 28610, "tokens": 14659092480, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8430523574352264, "ce_loss": 2.8270656764507294, "logit_z_loss": 14.713402152061462, "router_aux_loss": 1.3309503644704819, "router_z_loss": 1.2057807594537735, "routing_entropy_normalized": 0.6652434596835223, "learning_rate": 0.003, "grad_norm": 0.6160432696342468, "tokens_per_second": 3747191.5625385004, "eta_seconds": 12099.84210395813, "data_wait_seconds": 1.7374753952026367e-05} {"step": 7000, "total_steps": 28610, "tokens": 14680064000, "source": "fineweb_hq", "phase": 2, "loss": 3.3270709216594696, "ce_loss": 3.3116198778152466, "logit_z_loss": 13.083301424980164, "router_aux_loss": 1.3253049701452255, "router_z_loss": 0.8896985128521919, "routing_entropy_normalized": 0.6668558403552576, "learning_rate": 0.003, "grad_norm": 0.4550006687641144, "tokens_per_second": 3384803.019403228, "eta_seconds": 13389.096635818481, "data_wait_seconds": 2.047419548034668e-05} {"step": 7010, "total_steps": 28610, "tokens": 14701035520, "source": "fineweb_hq", "phase": 2, "loss": 3.3307561576366425, "ce_loss": 3.315495729446411, "logit_z_loss": 14.006762504577637, "router_aux_loss": 1.2995194792747498, "router_z_loss": 0.8646115362644196, "routing_entropy_normalized": 0.6667548042146192, "learning_rate": 0.003, "grad_norm": 0.7253624796867371, "tokens_per_second": 3269765.402838384, "eta_seconds": 13853.741054534912, "data_wait_seconds": 1.8924474716186523e-05} {"step": 7020, "total_steps": 28610, "tokens": 14722007040, "source": "fineweb_edu", "phase": 2, "loss": 3.3883414566516876, "ce_loss": 3.3732417821884155, "logit_z_loss": 13.67346477508545, "router_aux_loss": 1.2840865701436996, "router_z_loss": 0.8914923369884491, "routing_entropy_normalized": 0.6714946252963141, "learning_rate": 0.003, "grad_norm": 0.597175657749176, "tokens_per_second": 3599023.8148880745, "eta_seconds": 12580.497937440872, "data_wait_seconds": 1.9252300262451172e-05} {"step": 7030, "total_steps": 28610, "tokens": 14742978560, "source": "dclm", "phase": 2, "loss": 3.5211923122406006, "ce_loss": 3.504954516887665, "logit_z_loss": 13.459235787391663, "router_aux_loss": 1.4201329052448273, "router_z_loss": 0.690592922270298, "routing_entropy_normalized": 0.648032597556442, "learning_rate": 0.003, "grad_norm": 0.8036717176437378, "tokens_per_second": 3823870.717936307, "eta_seconds": 11835.269416332245, "data_wait_seconds": 1.817941665649414e-05} {"step": 7040, "total_steps": 28610, "tokens": 14763950080, "source": "dclm", "phase": 2, "loss": 3.512749522924423, "ce_loss": 3.496714621782303, "logit_z_loss": 12.909197449684143, "router_aux_loss": 1.4036509692668915, "router_z_loss": 0.7074951827526093, "routing_entropy_normalized": 0.6731430279966758, "learning_rate": 0.003, "grad_norm": 0.6265928149223328, "tokens_per_second": 3985991.997470671, "eta_seconds": 11348.635087251663, "data_wait_seconds": 2.300739288330078e-05} {"step": 7050, "total_steps": 28610, "tokens": 14784921600, "source": "fineweb_hq", "phase": 2, "loss": 3.3109079897403717, "ce_loss": 3.2955811619758606, "logit_z_loss": 13.154072642326355, "router_aux_loss": 1.3226104527711868, "router_z_loss": 0.7853045836091042, "routing_entropy_normalized": 0.6832080183308824, "learning_rate": 0.003, "grad_norm": 0.4577200412750244, "tokens_per_second": 3022964.1885660854, "eta_seconds": 14957.04027557373, "data_wait_seconds": 0.17029014229774475} {"step": 7060, "total_steps": 28610, "tokens": 14805893120, "source": "fineweb_hq", "phase": 2, "loss": 3.2871369421482086, "ce_loss": 3.2718663811683655, "logit_z_loss": 13.514173150062561, "router_aux_loss": 1.3161895871162415, "router_z_loss": 0.757289931178093, "routing_entropy_normalized": 0.6685793173951676, "learning_rate": 0.003, "grad_norm": 0.49368512630462646, "tokens_per_second": 2868504.1581482626, "eta_seconds": 15755.11943101883, "data_wait_seconds": 0.17715081572532654} {"step": 7070, "total_steps": 28610, "tokens": 14826864640, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8555175065994263, "ce_loss": 2.8404273092746735, "logit_z_loss": 14.534233212471008, "router_aux_loss": 1.2571933269500732, "router_z_loss": 1.0648493319749832, "routing_entropy_normalized": 0.675449025454984, "learning_rate": 0.003, "grad_norm": 0.8477349877357483, "tokens_per_second": 3612778.1173961395, "eta_seconds": 12503.578302383423, "data_wait_seconds": 1.9311904907226562e-05} {"step": 7080, "total_steps": 28610, "tokens": 14847836160, "source": "dclm", "phase": 2, "loss": 3.52404123544693, "ce_loss": 3.5088572800159454, "logit_z_loss": 12.840593218803406, "router_aux_loss": 1.3332837671041489, "router_z_loss": 0.5670163109898567, "routing_entropy_normalized": 0.6628561422567844, "learning_rate": 0.003, "grad_norm": 0.5950189232826233, "tokens_per_second": 4097082.607233575, "eta_seconds": 11020.447203159332, "data_wait_seconds": 1.8835067749023438e-05} {"step": 7090, "total_steps": 28610, "tokens": 14868807680, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8368683457374573, "ce_loss": 2.820974498987198, "logit_z_loss": 14.679245114326477, "router_aux_loss": 1.3399163037538528, "router_z_loss": 1.02675299346447, "routing_entropy_normalized": 0.6772756347415203, "learning_rate": 0.003, "grad_norm": 0.6867246031761169, "tokens_per_second": 3597473.706208603, "eta_seconds": 12545.112133026123, "data_wait_seconds": 1.7583370208740234e-05} {"step": 7100, "total_steps": 28610, "tokens": 14889779200, "source": "fineweb_hq", "phase": 2, "loss": 3.1149975061416626, "ce_loss": 3.099800258874893, "logit_z_loss": 13.75799834728241, "router_aux_loss": 1.309481993317604, "router_z_loss": 0.7266411036252975, "routing_entropy_normalized": 0.6538482584776524, "learning_rate": 0.003, "grad_norm": 0.45516344904899597, "tokens_per_second": 4242048.977789858, "eta_seconds": 10633.950658321379, "data_wait_seconds": 2.104043960571289e-05} {"step": 7110, "total_steps": 28610, "tokens": 14910750720, "source": "fineweb_hq", "phase": 2, "loss": 3.1211023330688477, "ce_loss": 3.105961889028549, "logit_z_loss": 13.97253155708313, "router_aux_loss": 1.3059066087007523, "router_z_loss": 0.6841484978795052, "routing_entropy_normalized": 0.6564025121188092, "learning_rate": 0.003, "grad_norm": 0.35275331139564514, "tokens_per_second": 4117477.0927773495, "eta_seconds": 10950.58138370514, "data_wait_seconds": 1.8596649169921875e-05} {"step": 7120, "total_steps": 28610, "tokens": 14931722240, "source": "fineweb_edu", "phase": 2, "loss": 3.3955269157886505, "ce_loss": 3.380816638469696, "logit_z_loss": 13.519436717033386, "router_aux_loss": 1.263035148382187, "router_z_loss": 0.7279628217220306, "routing_entropy_normalized": 0.6681399005836095, "learning_rate": 0.003, "grad_norm": 0.5839744210243225, "tokens_per_second": 4071947.82694164, "eta_seconds": 11067.871788978577, "data_wait_seconds": 1.9252300262451172e-05} {"step": 7130, "total_steps": 28610, "tokens": 14952693760, "source": "dclm", "phase": 2, "loss": 3.4707086980342865, "ce_loss": 3.4544979631900787, "logit_z_loss": 13.347236156463623, "router_aux_loss": 1.4195878356695175, "router_z_loss": 0.6802267804741859, "routing_entropy_normalized": 0.6294913237667163, "learning_rate": 0.003, "grad_norm": 0.634864330291748, "tokens_per_second": 4001816.094750138, "eta_seconds": 11256.595479011536, "data_wait_seconds": 2.0176172256469727e-05} {"step": 7140, "total_steps": 28610, "tokens": 14973665280, "source": "dclm", "phase": 2, "loss": 3.566685199737549, "ce_loss": 3.549175053834915, "logit_z_loss": 12.766688346862793, "router_aux_loss": 1.5510510802268982, "router_z_loss": 0.7229290381073952, "routing_entropy_normalized": 0.6300733628614779, "learning_rate": 0.003, "grad_norm": 0.7089880704879761, "tokens_per_second": 3265998.516209741, "eta_seconds": 13786.244303703308, "data_wait_seconds": 1.6868114471435547e-05} {"step": 7150, "total_steps": 28610, "tokens": 14994636800, "source": "fineweb_hq", "phase": 2, "loss": 3.1176124215126038, "ce_loss": 3.101097136735916, "logit_z_loss": 14.225051045417786, "router_aux_loss": 1.4181193262338638, "router_z_loss": 0.9116457104682922, "routing_entropy_normalized": 0.6522966666651823, "learning_rate": 0.003, "grad_norm": 0.40050122141838074, "tokens_per_second": 4201721.29374006, "eta_seconds": 10711.058343410492, "data_wait_seconds": 2.16066837310791e-05} {"step": 7160, "total_steps": 28610, "tokens": 15015608320, "source": "fineweb_hq", "phase": 2, "loss": 3.097404420375824, "ce_loss": 3.081428200006485, "logit_z_loss": 14.08897352218628, "router_aux_loss": 1.364799752831459, "router_z_loss": 0.9194032624363899, "routing_entropy_normalized": 0.663325161624856, "learning_rate": 0.003, "grad_norm": 0.28120335936546326, "tokens_per_second": 4379303.093862269, "eta_seconds": 10271.93355560303, "data_wait_seconds": 2.22623348236084e-05} {"step": 7170, "total_steps": 28610, "tokens": 15036579840, "source": "cosmopedia_v2", "phase": 2, "loss": 2.865169256925583, "ce_loss": 2.8495093286037445, "logit_z_loss": 15.02271020412445, "router_aux_loss": 1.2954989820718765, "router_z_loss": 1.202666625380516, "routing_entropy_normalized": 0.6794033410356923, "learning_rate": 0.003, "grad_norm": 0.6052529811859131, "tokens_per_second": 3671480.831306965, "eta_seconds": 12246.540550231934, "data_wait_seconds": 1.9222497940063477e-05} {"step": 7180, "total_steps": 28610, "tokens": 15057551360, "source": "dclm", "phase": 2, "loss": 3.5090492367744446, "ce_loss": 3.49321511387825, "logit_z_loss": 13.001719236373901, "router_aux_loss": 1.3826258033514023, "router_z_loss": 0.7076535895466805, "routing_entropy_normalized": 0.6503149015197194, "learning_rate": 0.003, "grad_norm": 0.9645808339118958, "tokens_per_second": 3584341.0541086746, "eta_seconds": 12538.418270349503, "data_wait_seconds": 1.7940998077392578e-05} {"step": 7190, "total_steps": 28610, "tokens": 15078522880, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8783801794052124, "ce_loss": 2.86178857088089, "logit_z_loss": 14.892443776130676, "router_aux_loss": 1.4014688283205032, "router_z_loss": 1.087680384516716, "routing_entropy_normalized": 0.6765075942752873, "learning_rate": 0.003, "grad_norm": 0.8082183003425598, "tokens_per_second": 3545717.534104496, "eta_seconds": 12669.084722042084, "data_wait_seconds": 2.2172927856445312e-05} {"step": 7200, "total_steps": 28610, "tokens": 15099494400, "source": "fineweb_hq", "phase": 2, "loss": 3.0485642552375793, "ce_loss": 3.031847596168518, "logit_z_loss": 14.244577765464783, "router_aux_loss": 1.4397721886634827, "router_z_loss": 0.8944883123040199, "routing_entropy_normalized": 0.6374703998449196, "learning_rate": 0.003, "grad_norm": 0.5223842859268188, "tokens_per_second": 3329461.376066672, "eta_seconds": 13485.67207980156, "data_wait_seconds": 0.16437596082687378} {"step": 7210, "total_steps": 28610, "tokens": 15120465920, "source": "fineweb_hq", "phase": 2, "loss": 3.0651611983776093, "ce_loss": 3.0492066144943237, "logit_z_loss": 14.60038959980011, "router_aux_loss": 1.3610917925834656, "router_z_loss": 0.8835650458931923, "routing_entropy_normalized": 0.669602902358583, "learning_rate": 0.003, "grad_norm": 0.7170085310935974, "tokens_per_second": 4325676.217711205, "eta_seconds": 10375.037460327148, "data_wait_seconds": 1.7940998077392578e-05} {"step": 7220, "total_steps": 28610, "tokens": 15141437440, "source": "fineweb_edu", "phase": 2, "loss": 3.4177172482013702, "ce_loss": 3.4016907811164856, "logit_z_loss": 13.172847747802734, "router_aux_loss": 1.3799761831760406, "router_z_loss": 0.9094400182366371, "routing_entropy_normalized": 0.6517412518786647, "learning_rate": 0.003, "grad_norm": 0.6429488658905029, "tokens_per_second": 3945627.0983162182, "eta_seconds": 11369.062550067902, "data_wait_seconds": 2.0772218704223633e-05} {"step": 7230, "total_steps": 28610, "tokens": 15162408960, "source": "dclm", "phase": 2, "loss": 3.561513841152191, "ce_loss": 3.5448697209358215, "logit_z_loss": 13.153249621391296, "router_aux_loss": 1.4537344723939896, "router_z_loss": 0.7914039343595505, "routing_entropy_normalized": 0.6654136910470917, "learning_rate": 0.003, "grad_norm": 0.6059518456459045, "tokens_per_second": 3727103.890908704, "eta_seconds": 12030.013402462006, "data_wait_seconds": 0.09187647700309753} {"step": 7240, "total_steps": 28610, "tokens": 15183380480, "source": "dclm", "phase": 2, "loss": 3.5362990498542786, "ce_loss": 3.5199388563632965, "logit_z_loss": 12.749338865280151, "router_aux_loss": 1.4188100397586823, "router_z_loss": 0.8971016854047775, "routing_entropy_normalized": 0.6473353893753406, "learning_rate": 0.003, "grad_norm": 0.6795514225959778, "tokens_per_second": 3643280.933713822, "eta_seconds": 12301.038282632828, "data_wait_seconds": 2.0742416381835938e-05} {"step": 7250, "total_steps": 28610, "tokens": 15204352000, "source": "fineweb_hq", "phase": 2, "loss": 3.0633696913719177, "ce_loss": 3.046902358531952, "logit_z_loss": 14.71714186668396, "router_aux_loss": 1.3912567645311356, "router_z_loss": 1.0829534828662872, "routing_entropy_normalized": 0.6451928646893932, "learning_rate": 0.003, "grad_norm": 0.7088014483451843, "tokens_per_second": 4204175.578877214, "eta_seconds": 10654.922916412354, "data_wait_seconds": 2.232193946838379e-05} {"step": 7260, "total_steps": 28610, "tokens": 15225323520, "source": "fineweb_hq", "phase": 2, "loss": 3.062168061733246, "ce_loss": 3.0456899404525757, "logit_z_loss": 14.0089271068573, "router_aux_loss": 1.4072289168834686, "router_z_loss": 1.004975326359272, "routing_entropy_normalized": 0.6441178898564507, "learning_rate": 0.003, "grad_norm": 0.46256083250045776, "tokens_per_second": 3531406.6529568494, "eta_seconds": 12678.85565161705, "data_wait_seconds": 2.0384788513183594e-05} {"step": 7270, "total_steps": 28610, "tokens": 15246295040, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8263960480690002, "ce_loss": 2.8097861111164093, "logit_z_loss": 14.766512513160706, "router_aux_loss": 1.37988743185997, "router_z_loss": 1.3343348503112793, "routing_entropy_normalized": 0.6653747914682717, "learning_rate": 0.003, "grad_norm": 0.6969625353813171, "tokens_per_second": 4277026.755936743, "eta_seconds": 10463.629580497742, "data_wait_seconds": 2.2530555725097656e-05} {"step": 7280, "total_steps": 28610, "tokens": 15267266560, "source": "dclm", "phase": 2, "loss": 3.5855407118797302, "ce_loss": 3.5692735612392426, "logit_z_loss": 12.952449560165405, "router_aux_loss": 1.41165129840374, "router_z_loss": 0.8554026633501053, "routing_entropy_normalized": 0.6428380259746147, "learning_rate": 0.003, "grad_norm": 0.8771041631698608, "tokens_per_second": 3321288.7000834015, "eta_seconds": 13468.34202003479, "data_wait_seconds": 2.1457672119140625e-05} {"step": 7290, "total_steps": 28610, "tokens": 15288238080, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8291452527046204, "ce_loss": 2.8120327591896057, "logit_z_loss": 14.879547953605652, "router_aux_loss": 1.4362783432006836, "router_z_loss": 1.2617390304803848, "routing_entropy_normalized": 0.6577211574458033, "learning_rate": 0.003, "grad_norm": 0.41540271043777466, "tokens_per_second": 3840892.2505351366, "eta_seconds": 11640.85783290863, "data_wait_seconds": 1.6897916793823242e-05} {"step": 7300, "total_steps": 28610, "tokens": 15309209600, "source": "fineweb_hq", "phase": 2, "loss": 3.059650182723999, "ce_loss": 3.0428772270679474, "logit_z_loss": 13.881364822387695, "router_aux_loss": 1.4351304471492767, "router_z_loss": 1.0335135385394096, "routing_entropy_normalized": 0.6772326090543946, "learning_rate": 0.003, "grad_norm": 0.5495249032974243, "tokens_per_second": 3734730.384363211, "eta_seconds": 11966.14066362381, "data_wait_seconds": 2.110004425048828e-05} {"step": 7310, "total_steps": 28610, "tokens": 15330181120, "source": "fineweb_hq", "phase": 2, "loss": 3.0630021393299103, "ce_loss": 3.046305239200592, "logit_z_loss": 14.4634507894516, "router_aux_loss": 1.4250885546207428, "router_z_loss": 0.9996096193790436, "routing_entropy_normalized": 0.6427490667503353, "learning_rate": 0.003, "grad_norm": 0.7758964896202087, "tokens_per_second": 4333010.621481929, "eta_seconds": 10309.076414108276, "data_wait_seconds": 2.6285648345947266e-05} {"step": 7320, "total_steps": 28610, "tokens": 15351152640, "source": "fineweb_edu", "phase": 2, "loss": 3.398282527923584, "ce_loss": 3.381839483976364, "logit_z_loss": 13.527820110321045, "router_aux_loss": 1.4053110182285309, "router_z_loss": 1.0370900109410286, "routing_entropy_normalized": 0.6536541712251205, "learning_rate": 0.003, "grad_norm": 0.7202852964401245, "tokens_per_second": 4144777.2058361955, "eta_seconds": 10772.199291467667, "data_wait_seconds": 2.1576881408691406e-05} {"step": 7330, "total_steps": 28610, "tokens": 15372124160, "source": "dclm", "phase": 2, "loss": 3.540380120277405, "ce_loss": 3.523303359746933, "logit_z_loss": 13.124609112739563, "router_aux_loss": 1.483193799853325, "router_z_loss": 0.9323186129331589, "routing_entropy_normalized": 0.6380968143875151, "learning_rate": 0.003, "grad_norm": 0.7531060576438904, "tokens_per_second": 4058253.791081311, "eta_seconds": 10996.698791503906, "data_wait_seconds": 2.1338462829589844e-05} {"step": 7340, "total_steps": 28610, "tokens": 15393095680, "source": "dclm", "phase": 2, "loss": 3.491467833518982, "ce_loss": 3.4736939668655396, "logit_z_loss": 12.96982717514038, "router_aux_loss": 1.5435663759708405, "router_z_loss": 1.0412209630012512, "routing_entropy_normalized": 0.6547732844116333, "learning_rate": 0.003, "grad_norm": 0.6414024233818054, "tokens_per_second": 3090593.4945620396, "eta_seconds": 14432.963480472565, "data_wait_seconds": 1.665949821472168e-05} {"step": 7350, "total_steps": 28610, "tokens": 15414067200, "source": "fineweb_hq", "phase": 2, "loss": 3.0558435916900635, "ce_loss": 3.0389994382858276, "logit_z_loss": 14.06322193145752, "router_aux_loss": 1.4194086641073227, "router_z_loss": 1.243742823600769, "routing_entropy_normalized": 0.6495335391656166, "learning_rate": 0.003, "grad_norm": 0.6566392779350281, "tokens_per_second": 4244042.5159244, "eta_seconds": 10505.420563697815, "data_wait_seconds": 2.0772218704223633e-05} {"step": 7360, "total_steps": 28610, "tokens": 15435038720, "source": "fineweb_hq", "phase": 2, "loss": 3.0165877640247345, "ce_loss": 2.9994762539863586, "logit_z_loss": 14.42409598827362, "router_aux_loss": 1.4412716329097748, "router_z_loss": 1.2564437836408615, "routing_entropy_normalized": 0.6475395268644328, "learning_rate": 0.003, "grad_norm": 0.3442533016204834, "tokens_per_second": 3899311.333846498, "eta_seconds": 11428.807854652405, "data_wait_seconds": 2.473592758178711e-05} {"step": 7370, "total_steps": 28610, "tokens": 15456010240, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8264268040657043, "ce_loss": 2.8097163438796997, "logit_z_loss": 15.394414782524109, "router_aux_loss": 1.371105894446373, "router_z_loss": 1.45989890396595, "routing_entropy_normalized": 0.6700036042087597, "learning_rate": 0.003, "grad_norm": 0.8534649610519409, "tokens_per_second": 3813332.2597248917, "eta_seconds": 11680.99327468872, "data_wait_seconds": 2.104043960571289e-05} {"step": 7380, "total_steps": 28610, "tokens": 15476981760, "source": "dclm", "phase": 2, "loss": 3.5206580460071564, "ce_loss": 3.5040691196918488, "logit_z_loss": 13.51339876651764, "router_aux_loss": 1.4148331731557846, "router_z_loss": 1.0892488807439804, "routing_entropy_normalized": 0.6663316473797515, "learning_rate": 0.003, "grad_norm": 0.41157209873199463, "tokens_per_second": 3523949.229309717, "eta_seconds": 12634.27310180664, "data_wait_seconds": 1.9937753677368164e-05} {"step": 7390, "total_steps": 28610, "tokens": 15497953280, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8894189596176147, "ce_loss": 2.871684581041336, "logit_z_loss": 15.081985712051392, "router_aux_loss": 1.4729705154895782, "router_z_loss": 1.4964407682418823, "routing_entropy_normalized": 0.6778526463877205, "learning_rate": 0.003, "grad_norm": 1.2055528163909912, "tokens_per_second": 3153886.303460604, "eta_seconds": 14110.072830200195, "data_wait_seconds": 1.8090009689331055e-05} {"step": 7400, "total_steps": 28610, "tokens": 15518924800, "source": "fineweb_hq", "phase": 2, "loss": 3.0479123294353485, "ce_loss": 3.0299207866191864, "logit_z_loss": 14.683045864105225, "router_aux_loss": 1.5353371649980545, "router_z_loss": 1.1698835492134094, "routing_entropy_normalized": 0.6474268707688355, "learning_rate": 0.003, "grad_norm": 0.6055316925048828, "tokens_per_second": 3642596.5700106407, "eta_seconds": 12211.232582330704, "data_wait_seconds": 2.0742416381835938e-05} {"step": 7410, "total_steps": 28610, "tokens": 15539896320, "source": "fineweb_hq", "phase": 2, "loss": 3.015694111585617, "ce_loss": 2.9967040419578552, "logit_z_loss": 14.28910505771637, "router_aux_loss": 1.6334249526262283, "router_z_loss": 1.2269390672445297, "routing_entropy_normalized": 0.6369243610430692, "learning_rate": 0.003, "grad_norm": 0.4123324453830719, "tokens_per_second": 4270658.9471655255, "eta_seconds": 10410.48300743103, "data_wait_seconds": 2.008676528930664e-05} {"step": 7420, "total_steps": 28610, "tokens": 15560867840, "source": "fineweb_edu", "phase": 2, "loss": 3.4103672206401825, "ce_loss": 3.3923659920692444, "logit_z_loss": 13.737393617630005, "router_aux_loss": 1.5181221067905426, "router_z_loss": 1.4462887793779373, "routing_entropy_normalized": 0.6529048115659588, "learning_rate": 0.003, "grad_norm": 0.5563192963600159, "tokens_per_second": 3811526.0273680794, "eta_seconds": 11659.018083810806, "data_wait_seconds": 2.0772218704223633e-05} {"step": 7430, "total_steps": 28610, "tokens": 15581839360, "source": "dclm", "phase": 2, "loss": 3.5406249463558197, "ce_loss": 3.5236117839813232, "logit_z_loss": 12.96648383140564, "router_aux_loss": 1.4501033276319504, "router_z_loss": 1.2154763638973236, "routing_entropy_normalized": 0.6561557834520705, "learning_rate": 0.003, "grad_norm": 0.5766887068748474, "tokens_per_second": 3452443.882173214, "eta_seconds": 12865.576060295105, "data_wait_seconds": 2.1785497665405273e-05} {"step": 7440, "total_steps": 28610, "tokens": 15602810880, "source": "dclm", "phase": 2, "loss": 3.5424934029579163, "ce_loss": 3.5258744955062866, "logit_z_loss": 12.83116364479065, "router_aux_loss": 1.4022010415792465, "router_z_loss": 1.3137719184160233, "routing_entropy_normalized": 0.6679971129926735, "learning_rate": 0.003, "grad_norm": 0.47751474380493164, "tokens_per_second": 3622006.604289422, "eta_seconds": 12257.48947763443, "data_wait_seconds": 1.665949821472168e-05} {"step": 7450, "total_steps": 28610, "tokens": 15623782400, "source": "fineweb_hq", "phase": 2, "loss": 3.026969015598297, "ce_loss": 3.009915739297867, "logit_z_loss": 14.425911545753479, "router_aux_loss": 1.4358159899711609, "router_z_loss": 1.2525072246789932, "routing_entropy_normalized": 0.6605564624301856, "learning_rate": 0.003, "grad_norm": 0.5192473530769348, "tokens_per_second": 3915501.30493429, "eta_seconds": 11333.347345352173, "data_wait_seconds": 2.181529998779297e-05} {"step": 7460, "total_steps": 28610, "tokens": 15644753920, "source": "fineweb_hq", "phase": 2, "loss": 3.041776180267334, "ce_loss": 3.0251715779304504, "logit_z_loss": 14.401175022125244, "router_aux_loss": 1.3954755514860153, "router_z_loss": 1.2097726166248322, "routing_entropy_normalized": 0.6562734415329918, "learning_rate": 0.003, "grad_norm": 0.5627836585044861, "tokens_per_second": 3868254.1038782354, "eta_seconds": 11466.352418661118, "data_wait_seconds": 2.065300941467285e-05} {"step": 7470, "total_steps": 28610, "tokens": 15665725440, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8060429096221924, "ce_loss": 2.7895804941654205, "logit_z_loss": 15.054627299308777, "router_aux_loss": 1.3464740961790085, "router_z_loss": 1.4921553432941437, "routing_entropy_normalized": 0.6863432870763004, "learning_rate": 0.003, "grad_norm": 0.6481449604034424, "tokens_per_second": 3768488.640053975, "eta_seconds": 11764.343086719513, "data_wait_seconds": 2.2679567337036133e-05} {"step": 7480, "total_steps": 28610, "tokens": 15686696960, "source": "dclm", "phase": 2, "loss": 3.539245992898941, "ce_loss": 3.524292469024658, "logit_z_loss": 13.14085054397583, "router_aux_loss": 1.2622437179088593, "router_z_loss": 1.017048791050911, "routing_entropy_normalized": 0.6950645293167854, "learning_rate": 0.003, "grad_norm": 0.8041547536849976, "tokens_per_second": 4439051.808277785, "eta_seconds": 9982.49709033966, "data_wait_seconds": 2.2411346435546875e-05} {"step": 7490, "total_steps": 28610, "tokens": 15707668480, "source": "cosmopedia_v2", "phase": 2, "loss": 2.816339135169983, "ce_loss": 2.8006531596183777, "logit_z_loss": 15.246547818183899, "router_aux_loss": 1.2979287058115005, "router_z_loss": 1.1819848865270615, "routing_entropy_normalized": 0.7038560945332216, "learning_rate": 0.003, "grad_norm": 0.5170695781707764, "tokens_per_second": 3784334.3818192105, "eta_seconds": 11704.00016784668, "data_wait_seconds": 2.181529998779297e-05} {"step": 7500, "total_steps": 28610, "tokens": 15728640000, "source": "fineweb_hq", "phase": 2, "loss": 3.017247587442398, "ce_loss": 3.000935971736908, "logit_z_loss": 14.301707029342651, "router_aux_loss": 1.4047503173351288, "router_z_loss": 0.8339602276682854, "routing_entropy_normalized": 0.6766995419391256, "learning_rate": 0.003, "grad_norm": 0.9305415749549866, "tokens_per_second": 3478971.065323952, "eta_seconds": 12725.279368162155, "data_wait_seconds": 2.16066837310791e-05} {"step": 7510, "total_steps": 28610, "tokens": 15749611520, "source": "fineweb_hq", "phase": 2, "loss": 3.014888733625412, "ce_loss": 2.9987908005714417, "logit_z_loss": 14.370697259902954, "router_aux_loss": 1.378314584493637, "router_z_loss": 0.8777012825012207, "routing_entropy_normalized": 0.6671950546289626, "learning_rate": 0.003, "grad_norm": 0.9463769197463989, "tokens_per_second": 4093577.09477401, "eta_seconds": 10809.594194889069, "data_wait_seconds": 2.5033950805664062e-05} {"step": 7520, "total_steps": 28610, "tokens": 15770583040, "source": "fineweb_edu", "phase": 2, "loss": 3.3983384668827057, "ce_loss": 3.382595181465149, "logit_z_loss": 13.359107375144958, "router_aux_loss": 1.3411876857280731, "router_z_loss": 0.995516300201416, "routing_entropy_normalized": 0.7108483691081416, "learning_rate": 0.003, "grad_norm": 0.6435871124267578, "tokens_per_second": 3694243.291742597, "eta_seconds": 11972.393853664398, "data_wait_seconds": 2.2083520889282227e-05} {"step": 7530, "total_steps": 28610, "tokens": 15791554560, "source": "dclm", "phase": 2, "loss": 3.4898044168949127, "ce_loss": 3.4739893674850464, "logit_z_loss": 12.810078740119934, "router_aux_loss": 1.3749492764472961, "router_z_loss": 0.7845462784171104, "routing_entropy_normalized": 0.7006161876411969, "learning_rate": 0.003, "grad_norm": 0.4907201826572418, "tokens_per_second": 2668957.797932842, "eta_seconds": 16563.75540828705, "data_wait_seconds": 2.014636993408203e-05} {"step": 7540, "total_steps": 28610, "tokens": 15812526080, "source": "dclm", "phase": 2, "loss": 3.5587137043476105, "ce_loss": 3.542748749256134, "logit_z_loss": 13.009336590766907, "router_aux_loss": 1.3822383731603622, "router_z_loss": 0.8416298925876617, "routing_entropy_normalized": 0.675488150928988, "learning_rate": 0.003, "grad_norm": 0.4856749176979065, "tokens_per_second": 3534568.4135918873, "eta_seconds": 12501.382762908936, "data_wait_seconds": 2.205371856689453e-05} {"step": 7550, "total_steps": 28610, "tokens": 15833497600, "source": "fineweb_hq", "phase": 2, "loss": 3.1010219156742096, "ce_loss": 3.0846753418445587, "logit_z_loss": 14.274677634239197, "router_aux_loss": 1.3996751010417938, "router_z_loss": 0.9222994446754456, "routing_entropy_normalized": 0.6826900329055369, "learning_rate": 0.003, "grad_norm": 0.3481244444847107, "tokens_per_second": 3024567.4467996727, "eta_seconds": 14602.42560195923, "data_wait_seconds": 0.08835023641586304} {"step": 7560, "total_steps": 28610, "tokens": 15854469120, "source": "fineweb_hq", "phase": 2, "loss": 3.194302976131439, "ce_loss": 3.178083896636963, "logit_z_loss": 13.865887999534607, "router_aux_loss": 1.395291343331337, "router_z_loss": 0.8795923367142677, "routing_entropy_normalized": 0.6885928462472174, "learning_rate": 0.003, "grad_norm": 0.3848944306373596, "tokens_per_second": 3420144.047434828, "eta_seconds": 12907.365592718124, "data_wait_seconds": 1.958012580871582e-05} {"step": 7570, "total_steps": 28610, "tokens": 15875440640, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8102011382579803, "ce_loss": 2.7942622303962708, "logit_z_loss": 14.630555748939514, "router_aux_loss": 1.3271268159151077, "router_z_loss": 1.20454041659832, "routing_entropy_normalized": 0.6944647318897617, "learning_rate": 0.003, "grad_norm": 0.7346013784408569, "tokens_per_second": 3487638.0170600633, "eta_seconds": 12651.564716339111, "data_wait_seconds": 1.7762184143066406e-05} {"step": 7580, "total_steps": 28610, "tokens": 15896412160, "source": "dclm", "phase": 2, "loss": 3.4757965207099915, "ce_loss": 3.4592227041721344, "logit_z_loss": 13.393474221229553, "router_aux_loss": 1.4432082176208496, "router_z_loss": 0.8023577183485031, "routing_entropy_normalized": 0.6929392975902073, "learning_rate": 0.003, "grad_norm": 0.7701729536056519, "tokens_per_second": 3527198.2048723903, "eta_seconds": 12503.722217559814, "data_wait_seconds": 1.811981201171875e-05} {"step": 7590, "total_steps": 28610, "tokens": 15917383680, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8829873502254486, "ce_loss": 2.865798205137253, "logit_z_loss": 15.28978955745697, "router_aux_loss": 1.4483113884925842, "router_z_loss": 1.1771447211503983, "routing_entropy_normalized": 0.6571489011176436, "learning_rate": 0.003, "grad_norm": 1.0361735820770264, "tokens_per_second": 3348413.35086448, "eta_seconds": 13165.08161354065, "data_wait_seconds": 2.1755695343017578e-05} {"step": 7600, "total_steps": 28610, "tokens": 15938355200, "source": "fineweb_hq", "phase": 2, "loss": 3.350041538476944, "ce_loss": 3.3332895636558533, "logit_z_loss": 13.600491404533386, "router_aux_loss": 1.450044423341751, "router_z_loss": 0.8914587497711182, "routing_entropy_normalized": 0.6841936032738577, "learning_rate": 0.003, "grad_norm": 0.5278627276420593, "tokens_per_second": 4049213.826108913, "eta_seconds": 10881.411901712418, "data_wait_seconds": 2.1338462829589844e-05} {"step": 7610, "total_steps": 28610, "tokens": 15959326720, "source": "fineweb_hq", "phase": 2, "loss": 3.3678722977638245, "ce_loss": 3.3519862294197083, "logit_z_loss": 13.316704750061035, "router_aux_loss": 1.3662888556718826, "router_z_loss": 0.8914893791079521, "routing_entropy_normalized": 0.6557722375266398, "learning_rate": 0.003, "grad_norm": 0.6992360949516296, "tokens_per_second": 3970078.41057001, "eta_seconds": 11093.028259277344, "data_wait_seconds": 2.0623207092285156e-05} {"step": 7620, "total_steps": 28610, "tokens": 15980298240, "source": "fineweb_edu", "phase": 2, "loss": 3.357884109020233, "ce_loss": 3.342015713453293, "logit_z_loss": 13.505746841430664, "router_aux_loss": 1.367596983909607, "router_z_loss": 0.841890387237072, "routing_entropy_normalized": 0.6806615800527346, "learning_rate": 0.003, "grad_norm": 0.7960200309753418, "tokens_per_second": 3699593.113079206, "eta_seconds": 11898.395076036453, "data_wait_seconds": 1.84476375579834e-05} {"step": 7630, "total_steps": 28610, "tokens": 16001269760, "source": "dclm", "phase": 2, "loss": 3.515204608440399, "ce_loss": 3.4987139105796814, "logit_z_loss": 13.45996642112732, "router_aux_loss": 1.446589857339859, "router_z_loss": 0.6787617057561874, "routing_entropy_normalized": 0.6488139279489137, "learning_rate": 0.003, "grad_norm": 0.49666500091552734, "tokens_per_second": 3506966.482633144, "eta_seconds": 12545.956506252289, "data_wait_seconds": 2.086162567138672e-05} {"step": 7640, "total_steps": 28610, "tokens": 16022241280, "source": "dclm", "phase": 2, "loss": 3.523474544286728, "ce_loss": 3.507538318634033, "logit_z_loss": 13.074212312698364, "router_aux_loss": 1.3931853026151657, "router_z_loss": 0.6970034390687943, "routing_entropy_normalized": 0.6665645464636688, "learning_rate": 0.003, "grad_norm": 0.5678983330726624, "tokens_per_second": 3663993.605077485, "eta_seconds": 12002.553000926971, "data_wait_seconds": 2.250075340270996e-05} {"step": 7650, "total_steps": 28610, "tokens": 16043212800, "source": "fineweb_hq", "phase": 2, "loss": 3.379412680864334, "ce_loss": 3.363558918237686, "logit_z_loss": 13.37559187412262, "router_aux_loss": 1.371908500790596, "router_z_loss": 0.7971208170056343, "routing_entropy_normalized": 0.6708253902870736, "learning_rate": 0.003, "grad_norm": 0.6774465441703796, "tokens_per_second": 3605374.764294559, "eta_seconds": 12191.88261795044, "data_wait_seconds": 0.08462139964103699} {"step": 7660, "total_steps": 28610, "tokens": 16064184320, "source": "fineweb_hq", "phase": 2, "loss": 3.3785538971424103, "ce_loss": 3.361934542655945, "logit_z_loss": 13.143332362174988, "router_aux_loss": 1.4502333253622055, "router_z_loss": 0.8027253076434135, "routing_entropy_normalized": 0.6548606719479517, "learning_rate": 0.003, "grad_norm": 0.727440357208252, "tokens_per_second": 4515833.910149204, "eta_seconds": 9729.174118041992, "data_wait_seconds": 2.0831823348999023e-05} {"step": 7670, "total_steps": 28610, "tokens": 16085155840, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8179784417152405, "ce_loss": 2.801356166601181, "logit_z_loss": 14.720582723617554, "router_aux_loss": 1.4046131074428558, "router_z_loss": 1.1040805131196976, "routing_entropy_normalized": 0.6608196362575214, "learning_rate": 0.003, "grad_norm": 0.49869275093078613, "tokens_per_second": 3350664.4794154395, "eta_seconds": 13106.165403842926, "data_wait_seconds": 2.142786979675293e-05} {"step": 7680, "total_steps": 28610, "tokens": 16106127360, "source": "dclm", "phase": 2, "loss": 3.488979607820511, "ce_loss": 3.4725452661514282, "logit_z_loss": 13.359756112098694, "router_aux_loss": 1.4369102120399475, "router_z_loss": 0.7292914316058159, "routing_entropy_normalized": 0.6768975321933021, "learning_rate": 0.003, "grad_norm": 0.4843900799751282, "tokens_per_second": 3852378.9862604314, "eta_seconds": 11393.840407848358, "data_wait_seconds": 1.9282102584838867e-05} {"step": 7690, "total_steps": 28610, "tokens": 16127098880, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8563504815101624, "ce_loss": 2.840218424797058, "logit_z_loss": 15.360689282417297, "router_aux_loss": 1.3491428345441818, "router_z_loss": 1.1045637428760529, "routing_entropy_normalized": 0.6870027656175793, "learning_rate": 0.003, "grad_norm": 0.790657103061676, "tokens_per_second": 3731920.053498774, "eta_seconds": 11755.991342544556, "data_wait_seconds": 2.2202730178833008e-05} {"step": 7700, "total_steps": 28610, "tokens": 16148070400, "source": "fineweb_hq", "phase": 2, "loss": 3.362533390522003, "ce_loss": 3.346625477075577, "logit_z_loss": 13.701542139053345, "router_aux_loss": 1.3779572546482086, "router_z_loss": 0.758121132850647, "routing_entropy_normalized": 0.649915992920934, "learning_rate": 0.003, "grad_norm": 1.036436915397644, "tokens_per_second": 3487904.2347285557, "eta_seconds": 12572.434725522995, "data_wait_seconds": 1.868605613708496e-05} {"step": 7710, "total_steps": 28610, "tokens": 16169041920, "source": "fineweb_hq", "phase": 2, "loss": 3.358055204153061, "ce_loss": 3.3422319889068604, "logit_z_loss": 13.597540497779846, "router_aux_loss": 1.3708310425281525, "router_z_loss": 0.7550832182168961, "routing_entropy_normalized": 0.6753692712759081, "learning_rate": 0.003, "grad_norm": 0.559894859790802, "tokens_per_second": 4246823.701664466, "eta_seconds": 10320.766737461088, "data_wait_seconds": 2.0295381546020508e-05} {"step": 7720, "total_steps": 28610, "tokens": 16190013440, "source": "fineweb_edu", "phase": 2, "loss": 3.4017805755138397, "ce_loss": 3.386148691177368, "logit_z_loss": 13.455705881118774, "router_aux_loss": 1.3514407873153687, "router_z_loss": 0.7718730866909027, "routing_entropy_normalized": 0.6574124171114247, "learning_rate": 0.003, "grad_norm": 0.5375728607177734, "tokens_per_second": 3444577.2065886436, "eta_seconds": 12718.398413658142, "data_wait_seconds": 1.8894672393798828e-05} {"step": 7730, "total_steps": 28610, "tokens": 16210984960, "source": "dclm", "phase": 2, "loss": 3.5066093504428864, "ce_loss": 3.4907072484493256, "logit_z_loss": 13.074846863746643, "router_aux_loss": 1.401909127831459, "router_z_loss": 0.5755794495344162, "routing_entropy_normalized": 0.6680567293354452, "learning_rate": 0.003, "grad_norm": 0.548627495765686, "tokens_per_second": 4116301.134571814, "eta_seconds": 10637.835359573364, "data_wait_seconds": 2.187490463256836e-05} {"step": 7740, "total_steps": 28610, "tokens": 16231956480, "source": "dclm", "phase": 2, "loss": 3.5165462493896484, "ce_loss": 3.4999365508556366, "logit_z_loss": 13.162378549575806, "router_aux_loss": 1.4696926027536392, "router_z_loss": 0.5965512096881866, "routing_entropy_normalized": 0.6332640337765225, "learning_rate": 0.003, "grad_norm": 0.6036965847015381, "tokens_per_second": 3494712.3059803606, "eta_seconds": 12523.938570022583, "data_wait_seconds": 2.0176172256469727e-05} {"step": 7750, "total_steps": 28610, "tokens": 16252928000, "source": "fineweb_hq", "phase": 2, "loss": 3.3613906502723694, "ce_loss": 3.344884902238846, "logit_z_loss": 13.279542207717896, "router_aux_loss": 1.4473146945238113, "router_z_loss": 0.7046712189912796, "routing_entropy_normalized": 0.6540553270586107, "learning_rate": 0.003, "grad_norm": 0.6447718739509583, "tokens_per_second": 3741324.522148462, "eta_seconds": 11692.808378696442, "data_wait_seconds": 1.7940998077392578e-05} {"step": 7760, "total_steps": 28610, "tokens": 16273899520, "source": "fineweb_hq", "phase": 2, "loss": 3.3671543300151825, "ce_loss": 3.351119488477707, "logit_z_loss": 13.350772500038147, "router_aux_loss": 1.3950732499361038, "router_z_loss": 0.7489964738488197, "routing_entropy_normalized": 0.6732512017750008, "learning_rate": 0.003, "grad_norm": 0.5831127166748047, "tokens_per_second": 4196886.589217846, "eta_seconds": 10418.58488917351, "data_wait_seconds": 2.2113323211669922e-05} {"step": 7770, "total_steps": 28610, "tokens": 16294871040, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8290221989154816, "ce_loss": 2.813372939825058, "logit_z_loss": 15.06255567073822, "router_aux_loss": 1.3087629526853561, "router_z_loss": 1.055273875594139, "routing_entropy_normalized": 0.679317182069937, "learning_rate": 0.003, "grad_norm": 0.5781195759773254, "tokens_per_second": 3625891.81823773, "eta_seconds": 12053.48914718628, "data_wait_seconds": 2.3543834686279297e-05} {"step": 7780, "total_steps": 28610, "tokens": 16315842560, "source": "dclm", "phase": 2, "loss": 3.487157881259918, "ce_loss": 3.4713920056819916, "logit_z_loss": 13.104893565177917, "router_aux_loss": 1.3862065076828003, "router_z_loss": 0.5933092907071114, "routing_entropy_normalized": 0.6453177287552342, "learning_rate": 0.003, "grad_norm": 1.1727198362350464, "tokens_per_second": 3720070.2230720157, "eta_seconds": 11742.702029943466, "data_wait_seconds": 2.2023916244506836e-05} {"step": 7790, "total_steps": 28610, "tokens": 16336814080, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8677906692028046, "ce_loss": 2.85136342048645, "logit_z_loss": 15.190053939819336, "router_aux_loss": 1.389245629310608, "router_z_loss": 1.015737920999527, "routing_entropy_normalized": 0.6580661835903424, "learning_rate": 0.003, "grad_norm": 1.358446717262268, "tokens_per_second": 4272309.245769082, "eta_seconds": 10219.9307513237, "data_wait_seconds": 1.9758939743041992e-05} {"step": 7800, "total_steps": 28610, "tokens": 16357785600, "source": "fineweb_hq", "phase": 2, "loss": 3.3645614087581635, "ce_loss": 3.34847891330719, "logit_z_loss": 13.535427927970886, "router_aux_loss": 1.3968738615512848, "router_z_loss": 0.7602477595210075, "routing_entropy_normalized": 0.677990812763467, "learning_rate": 0.003, "grad_norm": 0.6766075491905212, "tokens_per_second": 3470925.377435664, "eta_seconds": 12573.515237092972, "data_wait_seconds": 1.996755599975586e-05} {"step": 7810, "total_steps": 28610, "tokens": 16378757120, "source": "fineweb_hq", "phase": 2, "loss": 3.343609392642975, "ce_loss": 3.3276486992836, "logit_z_loss": 13.333749055862427, "router_aux_loss": 1.3912112414836884, "router_z_loss": 0.7151824608445168, "routing_entropy_normalized": 0.6627042930810086, "learning_rate": 0.003, "grad_norm": 0.435077428817749, "tokens_per_second": 3835631.0053014522, "eta_seconds": 11372.512512207031, "data_wait_seconds": 2.047419548034668e-05} {"step": 7820, "total_steps": 28610, "tokens": 16399728640, "source": "fineweb_edu", "phase": 2, "loss": 3.3995674550533295, "ce_loss": 3.383419394493103, "logit_z_loss": 13.850922226905823, "router_aux_loss": 1.3942452818155289, "router_z_loss": 0.8204793930053711, "routing_entropy_normalized": 0.6501407971204461, "learning_rate": 0.003, "grad_norm": 0.48038142919540405, "tokens_per_second": 3961705.638659064, "eta_seconds": 11005.30782866478, "data_wait_seconds": 2.0742416381835938e-05} {"step": 7830, "total_steps": 28610, "tokens": 16420700160, "source": "dclm", "phase": 2, "loss": 3.5355460047721863, "ce_loss": 3.519188314676285, "logit_z_loss": 13.028878688812256, "router_aux_loss": 1.443851500749588, "router_z_loss": 0.6162603422999382, "routing_entropy_normalized": 0.6465908742199374, "learning_rate": 0.003, "grad_norm": 0.9204996228218079, "tokens_per_second": 4072477.5848074304, "eta_seconds": 10700.81238079071, "data_wait_seconds": 2.3573637008666992e-05} {"step": 7840, "total_steps": 28610, "tokens": 16441671680, "source": "dclm", "phase": 2, "loss": 3.5045672059059143, "ce_loss": 3.488146483898163, "logit_z_loss": 13.170368432998657, "router_aux_loss": 1.4309837371110916, "router_z_loss": 0.7937687858939171, "routing_entropy_normalized": 0.6379652573979874, "learning_rate": 0.003, "grad_norm": 0.5844759941101074, "tokens_per_second": 3936505.421815635, "eta_seconds": 11065.105308532715, "data_wait_seconds": 2.232193946838379e-05} {"step": 7850, "total_steps": 28610, "tokens": 16462643200, "source": "fineweb_hq", "phase": 2, "loss": 3.360624462366104, "ce_loss": 3.34369558095932, "logit_z_loss": 13.512345671653748, "router_aux_loss": 1.474235326051712, "router_z_loss": 0.8353211656212807, "routing_entropy_normalized": 0.6427876973437978, "learning_rate": 0.003, "grad_norm": 0.4772355258464813, "tokens_per_second": 2914680.4973381124, "eta_seconds": 14937.10050201416, "data_wait_seconds": 0.2688751220703125} {"step": 7860, "total_steps": 28610, "tokens": 16483614720, "source": "fineweb_hq", "phase": 2, "loss": 3.3707938194274902, "ce_loss": 3.353606253862381, "logit_z_loss": 13.288547277450562, "router_aux_loss": 1.5053811818361282, "router_z_loss": 0.8049605935811996, "routing_entropy_normalized": 0.6445834037859713, "learning_rate": 0.003, "grad_norm": 0.8152022957801819, "tokens_per_second": 3732543.8363279016, "eta_seconds": 11658.511167764664, "data_wait_seconds": 2.35140323638916e-05} {"step": 7870, "total_steps": 28610, "tokens": 16504586240, "source": "cosmopedia_v2", "phase": 2, "loss": 2.826482594013214, "ce_loss": 2.8096242249011993, "logit_z_loss": 14.823664426803589, "router_aux_loss": 1.4224518835544586, "router_z_loss": 1.1514790505170822, "routing_entropy_normalized": 0.657650919914264, "learning_rate": 0.003, "grad_norm": 0.6309126019477844, "tokens_per_second": 3917420.535995085, "eta_seconds": 11102.9520778656, "data_wait_seconds": 2.047419548034668e-05} {"step": 7880, "total_steps": 28610, "tokens": 16525557760, "source": "dclm", "phase": 2, "loss": 3.551642119884491, "ce_loss": 3.535544216632843, "logit_z_loss": 12.951785206794739, "router_aux_loss": 1.4003027379512787, "router_z_loss": 0.7997757643461227, "routing_entropy_normalized": 0.6534421609338887, "learning_rate": 0.003, "grad_norm": 0.4082188308238983, "tokens_per_second": 3809521.859913981, "eta_seconds": 11411.920592308044, "data_wait_seconds": 1.9431114196777344e-05} {"step": 7890, "total_steps": 28610, "tokens": 16546529280, "source": "cosmopedia_v2", "phase": 2, "loss": 2.845624268054962, "ce_loss": 2.829335004091263, "logit_z_loss": 15.036911010742188, "router_aux_loss": 1.3606344908475876, "router_z_loss": 1.179278016090393, "routing_entropy_normalized": 0.6733734726287817, "learning_rate": 0.003, "grad_norm": 0.7684080600738525, "tokens_per_second": 3100326.7688031895, "eta_seconds": 14015.615991592407, "data_wait_seconds": 2.41696834564209e-05} {"step": 7900, "total_steps": 28610, "tokens": 16567500800, "source": "fineweb_hq", "phase": 2, "loss": 3.385418474674225, "ce_loss": 3.3688365817070007, "logit_z_loss": 13.880220770835876, "router_aux_loss": 1.4250609576702118, "router_z_loss": 0.9432454705238342, "routing_entropy_normalized": 0.6508355684338035, "learning_rate": 0.003, "grad_norm": 0.6642324924468994, "tokens_per_second": 4036653.2231724756, "eta_seconds": 10759.412691354752, "data_wait_seconds": 2.5987625122070312e-05} {"step": 7910, "total_steps": 28610, "tokens": 16588472320, "source": "fineweb_hq", "phase": 2, "loss": 3.391569584608078, "ce_loss": 3.3751506209373474, "logit_z_loss": 13.652895212173462, "router_aux_loss": 1.4202620685100555, "router_z_loss": 0.8511032983660698, "routing_entropy_normalized": 0.6542055291181921, "learning_rate": 0.003, "grad_norm": 0.7367622256278992, "tokens_per_second": 3823852.266190165, "eta_seconds": 11352.69968032837, "data_wait_seconds": 2.2202730178833008e-05} {"step": 7920, "total_steps": 28610, "tokens": 16609443840, "source": "fineweb_edu", "phase": 2, "loss": 3.423268884420395, "ce_loss": 3.4067689776420593, "logit_z_loss": 13.386806011199951, "router_aux_loss": 1.4188671112060547, "router_z_loss": 0.9726014733314514, "routing_entropy_normalized": 0.6288450401182919, "learning_rate": 0.003, "grad_norm": 0.8434992432594299, "tokens_per_second": 4275119.303149315, "eta_seconds": 10149.4418759346, "data_wait_seconds": 2.0831823348999023e-05} {"step": 7930, "total_steps": 28610, "tokens": 16630415360, "source": "dclm", "phase": 2, "loss": 3.531696230173111, "ce_loss": 3.514825165271759, "logit_z_loss": 13.75874125957489, "router_aux_loss": 1.4685159921646118, "router_z_loss": 0.8100751861929893, "routing_entropy_normalized": 0.6237694938080883, "learning_rate": 0.003, "grad_norm": 0.6158257722854614, "tokens_per_second": 3435383.246109917, "eta_seconds": 12624.240223884583, "data_wait_seconds": 2.0891427993774414e-05} {"step": 7940, "total_steps": 28610, "tokens": 16651386880, "source": "dclm", "phase": 2, "loss": 3.5306896567344666, "ce_loss": 3.513728141784668, "logit_z_loss": 13.292324542999268, "router_aux_loss": 1.4838810116052628, "router_z_loss": 0.7934756502509117, "routing_entropy_normalized": 0.6427648287441936, "learning_rate": 0.003, "grad_norm": 0.511766254901886, "tokens_per_second": 3520288.880834519, "eta_seconds": 12313.799607753754, "data_wait_seconds": 2.092123031616211e-05} {"step": 7950, "total_steps": 28610, "tokens": 16672358400, "source": "fineweb_hq", "phase": 2, "loss": 3.4066219329833984, "ce_loss": 3.390346974134445, "logit_z_loss": 13.63397204875946, "router_aux_loss": 1.408863753080368, "router_z_loss": 0.8229872062802315, "routing_entropy_normalized": 0.6524603725539616, "learning_rate": 0.003, "grad_norm": 0.37414640188217163, "tokens_per_second": 4470215.003179836, "eta_seconds": 9692.41083240509, "data_wait_seconds": 2.2709369659423828e-05} {"step": 7960, "total_steps": 28610, "tokens": 16693329920, "source": "fineweb_hq", "phase": 2, "loss": 3.3906797766685486, "ce_loss": 3.3744868338108063, "logit_z_loss": 13.199728012084961, "router_aux_loss": 1.4013133645057678, "router_z_loss": 0.8598363995552063, "routing_entropy_normalized": 0.6618462893950352, "learning_rate": 0.003, "grad_norm": 0.4496830701828003, "tokens_per_second": 3947402.9034681264, "eta_seconds": 10970.80532670021, "data_wait_seconds": 1.868605613708496e-05} {"step": 7970, "total_steps": 28610, "tokens": 16714301440, "source": "cosmopedia_v2", "phase": 2, "loss": 2.884646773338318, "ce_loss": 2.8680982291698456, "logit_z_loss": 14.799156069755554, "router_aux_loss": 1.3953804522752762, "router_z_loss": 1.114788144826889, "routing_entropy_normalized": 0.6601382989127182, "learning_rate": 0.003, "grad_norm": 0.9695791006088257, "tokens_per_second": 3747948.055168735, "eta_seconds": 11549.044075012207, "data_wait_seconds": 1.996755599975586e-05} {"step": 7980, "total_steps": 28610, "tokens": 16735272960, "source": "dclm", "phase": 2, "loss": 3.4655246138572693, "ce_loss": 3.449043959379196, "logit_z_loss": 13.453134655952454, "router_aux_loss": 1.4420439898967743, "router_z_loss": 0.7148787453770638, "routing_entropy_normalized": 0.6423771818214851, "learning_rate": 0.003, "grad_norm": 0.7806087732315063, "tokens_per_second": 4179841.2428918267, "eta_seconds": 10350.691149711609, "data_wait_seconds": 2.3156404495239258e-05} {"step": 7990, "total_steps": 28610, "tokens": 16756244480, "source": "cosmopedia_v2", "phase": 2, "loss": 2.872405558824539, "ce_loss": 2.855890154838562, "logit_z_loss": 15.199570894241333, "router_aux_loss": 1.3701120615005493, "router_z_loss": 1.2943446040153503, "routing_entropy_normalized": 0.6418865063517374, "learning_rate": 0.003, "grad_norm": 0.7409193515777588, "tokens_per_second": 3934660.025980291, "eta_seconds": 10990.34578704834, "data_wait_seconds": 2.193450927734375e-05} {"step": 8000, "total_steps": 28610, "tokens": 16777216000, "source": "fineweb_hq", "phase": 2, "loss": 3.3907001316547394, "ce_loss": 3.373070478439331, "logit_z_loss": 13.381072878837585, "router_aux_loss": 1.5324816554784775, "router_z_loss": 0.9667084738612175, "routing_entropy_normalized": 0.6459025947044977, "learning_rate": 0.003, "grad_norm": 1.0277373790740967, "tokens_per_second": 4176471.5400205585, "eta_seconds": 10348.99970126152, "data_wait_seconds": 1.7911195755004883e-05} {"step": 8010, "total_steps": 28610, "tokens": 16798187520, "source": "fineweb_hq", "phase": 2, "loss": 3.445319712162018, "ce_loss": 3.4279623329639435, "logit_z_loss": 13.437416553497314, "router_aux_loss": 1.506488397717476, "router_z_loss": 0.9487268626689911, "routing_entropy_normalized": 0.6307702789264041, "learning_rate": 0.003, "grad_norm": 0.6060436964035034, "tokens_per_second": 3474352.4020208223, "eta_seconds": 12434.355010986328, "data_wait_seconds": 2.0325183868408203e-05} {"step": 8020, "total_steps": 28610, "tokens": 16819159040, "source": "fineweb_edu", "phase": 2, "loss": 3.4249098002910614, "ce_loss": 3.4076667428016663, "logit_z_loss": 13.469515800476074, "router_aux_loss": 1.4876922816038132, "router_z_loss": 1.0193115770816803, "routing_entropy_normalized": 0.6491028257032816, "learning_rate": 0.003, "grad_norm": 0.5109028816223145, "tokens_per_second": 3797318.0158938267, "eta_seconds": 11371.278228282928, "data_wait_seconds": 1.7374753952026367e-05} {"step": 8030, "total_steps": 28610, "tokens": 16840130560, "source": "dclm", "phase": 2, "loss": 3.481458455324173, "ce_loss": 3.4644003808498383, "logit_z_loss": 12.821875929832458, "router_aux_loss": 1.4928035587072372, "router_z_loss": 0.8478234931826591, "routing_entropy_normalized": 0.6530585844091307, "learning_rate": 0.003, "grad_norm": 0.6320427060127258, "tokens_per_second": 3844550.4302162034, "eta_seconds": 11226.120958328247, "data_wait_seconds": 1.6301870346069336e-05} {"step": 8040, "total_steps": 28610, "tokens": 16861102080, "source": "dclm", "phase": 2, "loss": 3.5079263150691986, "ce_loss": 3.4910492599010468, "logit_z_loss": 13.176807880401611, "router_aux_loss": 1.4622123539447784, "router_z_loss": 0.9372116327285767, "routing_entropy_normalized": 0.6490186009390616, "learning_rate": 0.003, "grad_norm": 0.7339975833892822, "tokens_per_second": 4006424.3410788383, "eta_seconds": 10767.310940504074, "data_wait_seconds": 1.8894672393798828e-05} {"step": 8050, "total_steps": 28610, "tokens": 16882073600, "source": "fineweb_hq", "phase": 2, "loss": 3.4319981932640076, "ce_loss": 3.415332555770874, "logit_z_loss": 13.421315789222717, "router_aux_loss": 1.4246652126312256, "router_z_loss": 1.0768398642539978, "routing_entropy_normalized": 0.6517404517473829, "learning_rate": 0.003, "grad_norm": 0.5889806151390076, "tokens_per_second": 3515548.1314287917, "eta_seconds": 12264.785890579224, "data_wait_seconds": 0.21662601828575134} {"step": 8060, "total_steps": 28610, "tokens": 16903045120, "source": "fineweb_hq", "phase": 2, "loss": 3.3929280936717987, "ce_loss": 3.3752746880054474, "logit_z_loss": 13.41079556941986, "router_aux_loss": 1.5277223736047745, "router_z_loss": 1.0350930914282799, "routing_entropy_normalized": 0.6314501233275245, "learning_rate": 0.003, "grad_norm": 0.5544489026069641, "tokens_per_second": 4017081.009641258, "eta_seconds": 10728.305826187134, "data_wait_seconds": 2.008676528930664e-05} {"step": 8070, "total_steps": 28610, "tokens": 16924016640, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8375760912895203, "ce_loss": 2.8196208477020264, "logit_z_loss": 15.286535382270813, "router_aux_loss": 1.501202940940857, "router_z_loss": 1.4145893603563309, "routing_entropy_normalized": 0.6629615009098246, "learning_rate": 0.003, "grad_norm": 0.614920437335968, "tokens_per_second": 3867018.6200551945, "eta_seconds": 11139.202137947083, "data_wait_seconds": 2.4974346160888672e-05} {"step": 8080, "total_steps": 28610, "tokens": 16944988160, "source": "dclm", "phase": 2, "loss": 3.545783191919327, "ce_loss": 3.5290814340114594, "logit_z_loss": 13.46449089050293, "router_aux_loss": 1.436619684100151, "router_z_loss": 0.9891002774238586, "routing_entropy_normalized": 0.6651449623361678, "learning_rate": 0.003, "grad_norm": 0.6601982712745667, "tokens_per_second": 4000330.8206803817, "eta_seconds": 10762.742505550385, "data_wait_seconds": 1.9252300262451172e-05} {"step": 8090, "total_steps": 28610, "tokens": 16965959680, "source": "cosmopedia_v2", "phase": 2, "loss": 2.868418335914612, "ce_loss": 2.8523712754249573, "logit_z_loss": 15.022054314613342, "router_aux_loss": 1.3135132938623428, "router_z_loss": 1.4097207635641098, "routing_entropy_normalized": 0.6903992902146973, "learning_rate": 0.003, "grad_norm": 0.9866703152656555, "tokens_per_second": 3512504.7288962835, "eta_seconds": 12251.530563354492, "data_wait_seconds": 1.9758939743041992e-05} {"step": 8100, "total_steps": 28610, "tokens": 16986931200, "source": "fineweb_hq", "phase": 2, "loss": 3.3851086795330048, "ce_loss": 3.36894553899765, "logit_z_loss": 13.441921591758728, "router_aux_loss": 1.3835388869047165, "router_z_loss": 0.983514241874218, "routing_entropy_normalized": 0.6531381419319673, "learning_rate": 0.003, "grad_norm": 0.41320309042930603, "tokens_per_second": 4335990.679694351, "eta_seconds": 9919.898518562315, "data_wait_seconds": 2.4437904357910156e-05} {"step": 8110, "total_steps": 28610, "tokens": 17007902720, "source": "fineweb_hq", "phase": 2, "loss": 3.406763106584549, "ce_loss": 3.3907341361045837, "logit_z_loss": 13.362812995910645, "router_aux_loss": 1.37461456656456, "router_z_loss": 0.946545846760273, "routing_entropy_normalized": 0.6614625021345271, "learning_rate": 0.003, "grad_norm": 0.5926273465156555, "tokens_per_second": 3886904.9111240045, "eta_seconds": 11060.629725456238, "data_wait_seconds": 2.086162567138672e-05} {"step": 8120, "total_steps": 28610, "tokens": 17028874240, "source": "fineweb_edu", "phase": 2, "loss": 3.4331597983837128, "ce_loss": 3.4161235690116882, "logit_z_loss": 13.257527351379395, "router_aux_loss": 1.473589152097702, "router_z_loss": 0.974546916782856, "routing_entropy_normalized": 0.6546095862349304, "learning_rate": 0.003, "grad_norm": 0.5474023222923279, "tokens_per_second": 3824281.190651879, "eta_seconds": 11236.2669839859, "data_wait_seconds": 2.142786979675293e-05} {"step": 8130, "total_steps": 28610, "tokens": 17049845760, "source": "dclm", "phase": 2, "loss": 3.553028702735901, "ce_loss": 3.5367779433727264, "logit_z_loss": 13.20483911037445, "router_aux_loss": 1.41107577085495, "router_z_loss": 0.8195097371935844, "routing_entropy_normalized": 0.6609259597824412, "learning_rate": 0.003, "grad_norm": 0.45149552822113037, "tokens_per_second": 3814333.3561173305, "eta_seconds": 11260.0732421875, "data_wait_seconds": 1.4960765838623047e-05} {"step": 8140, "total_steps": 28610, "tokens": 17070817280, "source": "dclm", "phase": 2, "loss": 3.539253383874893, "ce_loss": 3.5218209326267242, "logit_z_loss": 13.255745887756348, "router_aux_loss": 1.5295377373695374, "router_z_loss": 0.8114638477563858, "routing_entropy_normalized": 0.6516529349630014, "learning_rate": 0.003, "grad_norm": 0.8197507858276367, "tokens_per_second": 4061953.00486815, "eta_seconds": 10568.487963438034, "data_wait_seconds": 1.704692840576172e-05} {"step": 8150, "total_steps": 28610, "tokens": 17091788800, "source": "fineweb_hq", "phase": 2, "loss": 3.404753476381302, "ce_loss": 3.388143867254257, "logit_z_loss": 13.339058756828308, "router_aux_loss": 1.428718015551567, "router_z_loss": 0.988491527736187, "routing_entropy_normalized": 0.642395940989748, "learning_rate": 0.003, "grad_norm": 0.45398658514022827, "tokens_per_second": 3942514.7546731597, "eta_seconds": 10883.340352535248, "data_wait_seconds": 2.8312206268310547e-05} {"step": 8160, "total_steps": 28610, "tokens": 17112760320, "source": "fineweb_hq", "phase": 2, "loss": 3.4080763459205627, "ce_loss": 3.3915031254291534, "logit_z_loss": 13.337979435920715, "router_aux_loss": 1.4285789877176285, "router_z_loss": 0.9535892233252525, "routing_entropy_normalized": 0.6543870324204788, "learning_rate": 0.003, "grad_norm": 0.7530694603919983, "tokens_per_second": 3756285.988961436, "eta_seconds": 11417.330449819565, "data_wait_seconds": 1.8149614334106445e-05} {"step": 8170, "total_steps": 28610, "tokens": 17133731840, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8293577432632446, "ce_loss": 2.812959849834442, "logit_z_loss": 14.852280139923096, "router_aux_loss": 1.369843527674675, "router_z_loss": 1.2142003923654556, "routing_entropy_normalized": 0.691346768488761, "learning_rate": 0.003, "grad_norm": 0.5869351029396057, "tokens_per_second": 3850265.402084094, "eta_seconds": 11133.203144073486, "data_wait_seconds": 2.0295381546020508e-05} {"step": 8180, "total_steps": 28610, "tokens": 17154703360, "source": "dclm", "phase": 2, "loss": 3.5105766654014587, "ce_loss": 3.4948374032974243, "logit_z_loss": 13.100895166397095, "router_aux_loss": 1.3672123402357101, "router_z_loss": 0.7570192441344261, "routing_entropy_normalized": 0.6660370139064851, "learning_rate": 0.003, "grad_norm": 0.8618273735046387, "tokens_per_second": 3909808.9259816012, "eta_seconds": 10958.28879904747, "data_wait_seconds": 2.008676528930664e-05} {"step": 8190, "total_steps": 28610, "tokens": 17175674880, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8506610691547394, "ce_loss": 2.8342175483703613, "logit_z_loss": 15.151305913925171, "router_aux_loss": 1.3681812137365341, "router_z_loss": 1.2465517669916153, "routing_entropy_normalized": 0.667994851101374, "learning_rate": 0.003, "grad_norm": 0.733011782169342, "tokens_per_second": 4000260.0515398677, "eta_seconds": 10705.264979839325, "data_wait_seconds": 2.0325183868408203e-05} {"step": 8200, "total_steps": 28610, "tokens": 17196646400, "source": "fineweb_hq", "phase": 2, "loss": 3.3659660816192627, "ce_loss": 3.3498964607715607, "logit_z_loss": 13.775824308395386, "router_aux_loss": 1.388920709490776, "router_z_loss": 0.8028475567698479, "routing_entropy_normalized": 0.6674601377700883, "learning_rate": 0.003, "grad_norm": 0.8985676765441895, "tokens_per_second": 4235614.302861422, "eta_seconds": 10105.469775915144, "data_wait_seconds": 1.8417835235595703e-05} {"step": 8210, "total_steps": 28610, "tokens": 17217617920, "source": "fineweb_hq", "phase": 2, "loss": 3.379803627729416, "ce_loss": 3.364305257797241, "logit_z_loss": 13.424402952194214, "router_aux_loss": 1.3288615345954895, "router_z_loss": 0.8672913387417793, "routing_entropy_normalized": 0.6899057729693532, "learning_rate": 0.003, "grad_norm": 0.623123288154602, "tokens_per_second": 3397676.8426351864, "eta_seconds": 12591.515550613403, "data_wait_seconds": 1.9103288650512695e-05} {"step": 8220, "total_steps": 28610, "tokens": 17238589440, "source": "fineweb_edu", "phase": 2, "loss": 3.3904662132263184, "ce_loss": 3.3746792674064636, "logit_z_loss": 13.400020599365234, "router_aux_loss": 1.36199352145195, "router_z_loss": 0.8269947096705437, "routing_entropy_normalized": 0.6494921895008844, "learning_rate": 0.003, "grad_norm": 0.8289037346839905, "tokens_per_second": 3819635.9790827204, "eta_seconds": 11195.02735710144, "data_wait_seconds": 1.7970800399780273e-05} {"step": 8230, "total_steps": 28610, "tokens": 17259560960, "source": "dclm", "phase": 2, "loss": 3.561071515083313, "ce_loss": 3.5449019372463226, "logit_z_loss": 13.396080255508423, "router_aux_loss": 1.4133753329515457, "router_z_loss": 0.6962165609002113, "routing_entropy_normalized": 0.6712642430953624, "learning_rate": 0.003, "grad_norm": 0.574411928653717, "tokens_per_second": 3919882.40040792, "eta_seconds": 10903.377549171448, "data_wait_seconds": 3.221631050109863e-05} {"step": 8240, "total_steps": 28610, "tokens": 17280532480, "source": "dclm", "phase": 2, "loss": 3.590034604072571, "ce_loss": 3.5738292038440704, "logit_z_loss": 12.930795550346375, "router_aux_loss": 1.420843243598938, "router_z_loss": 0.7039525881409645, "routing_entropy_normalized": 0.6477740156434232, "learning_rate": 0.003, "grad_norm": 0.7727658748626709, "tokens_per_second": 4100695.0652218536, "eta_seconds": 10417.498877763748, "data_wait_seconds": 2.0444393157958984e-05} {"step": 8250, "total_steps": 28610, "tokens": 17301504000, "source": "fineweb_hq", "phase": 2, "loss": 3.4032751619815826, "ce_loss": 3.387125700712204, "logit_z_loss": 13.343286871910095, "router_aux_loss": 1.4068855494260788, "router_z_loss": 0.7462405487895012, "routing_entropy_normalized": 0.6699148462475009, "learning_rate": 0.003, "grad_norm": 0.4651808738708496, "tokens_per_second": 4378793.612409278, "eta_seconds": 9751.090939521791, "data_wait_seconds": 1.8328428268432617e-05} {"step": 8260, "total_steps": 28610, "tokens": 17322475520, "source": "fineweb_hq", "phase": 2, "loss": 3.411197602748871, "ce_loss": 3.3949165642261505, "logit_z_loss": 13.473659753799438, "router_aux_loss": 1.4186840653419495, "router_z_loss": 0.7468352690339088, "routing_entropy_normalized": 0.6485118873307761, "learning_rate": 0.003, "grad_norm": 0.9404934644699097, "tokens_per_second": 4154715.8157711932, "eta_seconds": 10271.95242524147, "data_wait_seconds": 1.919269561767578e-05} {"step": 8270, "total_steps": 28610, "tokens": 17343447040, "source": "cosmopedia_v2", "phase": 2, "loss": 2.897880345582962, "ce_loss": 2.881048321723938, "logit_z_loss": 15.034393072128296, "router_aux_loss": 1.4203778505325317, "router_z_loss": 1.1248424500226974, "routing_entropy_normalized": 0.6429238938685453, "learning_rate": 0.003, "grad_norm": 1.0058262348175049, "tokens_per_second": 3419678.2683641505, "eta_seconds": 12473.709025382996, "data_wait_seconds": 2.1338462829589844e-05} {"step": 8280, "total_steps": 28610, "tokens": 17364418560, "source": "dclm", "phase": 2, "loss": 3.565874010324478, "ce_loss": 3.549697905778885, "logit_z_loss": 13.371055483818054, "router_aux_loss": 1.4039200842380524, "router_z_loss": 0.7997381016612053, "routing_entropy_normalized": 0.68559189141941, "learning_rate": 0.003, "grad_norm": 0.8294488191604614, "tokens_per_second": 3799777.934879683, "eta_seconds": 11220.418900966644, "data_wait_seconds": 0.00936838984489441} {"step": 8290, "total_steps": 28610, "tokens": 17385390080, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8447611033916473, "ce_loss": 2.827628880739212, "logit_z_loss": 15.170336961746216, "router_aux_loss": 1.4373325407505035, "router_z_loss": 1.2418344914913177, "routing_entropy_normalized": 0.6487848970678078, "learning_rate": 0.003, "grad_norm": 0.9401929378509521, "tokens_per_second": 4243004.373348986, "eta_seconds": 10043.385509490965, "data_wait_seconds": 2.1576881408691406e-05} {"step": 8300, "total_steps": 28610, "tokens": 17406361600, "source": "fineweb_hq", "phase": 2, "loss": 3.389152407646179, "ce_loss": 3.372317135334015, "logit_z_loss": 13.62286126613617, "router_aux_loss": 1.4490396976470947, "router_z_loss": 0.9826162531971931, "routing_entropy_normalized": 0.6601525360231861, "learning_rate": 0.003, "grad_norm": 0.794829785823822, "tokens_per_second": 3341642.785358213, "eta_seconds": 12746.173022031784, "data_wait_seconds": 0.05164030194282532} {"step": 8310, "total_steps": 28610, "tokens": 17427333120, "source": "fineweb_hq", "phase": 2, "loss": 3.4148824214935303, "ce_loss": 3.398642599582672, "logit_z_loss": 13.791016221046448, "router_aux_loss": 1.3837547600269318, "router_z_loss": 1.0231446996331215, "routing_entropy_normalized": 0.6391141281056132, "learning_rate": 0.003, "grad_norm": 0.6656312346458435, "tokens_per_second": 3956838.6265272815, "eta_seconds": 10759.141228199005, "data_wait_seconds": 1.7642974853515625e-05} {"step": 8320, "total_steps": 28610, "tokens": 17448304640, "source": "fineweb_edu", "phase": 2, "loss": 3.4044916331768036, "ce_loss": 3.3887396454811096, "logit_z_loss": 13.470907807350159, "router_aux_loss": 1.3384495079517365, "router_z_loss": 1.0203762575984001, "routing_entropy_normalized": 0.6686258995109026, "learning_rate": 0.003, "grad_norm": 0.6107876896858215, "tokens_per_second": 3217894.3699692125, "eta_seconds": 13223.309775829315, "data_wait_seconds": 1.749396324157715e-05} {"step": 8330, "total_steps": 28610, "tokens": 17469276160, "source": "dclm", "phase": 2, "loss": 3.4982609152793884, "ce_loss": 3.4828579425811768, "logit_z_loss": 13.16355299949646, "router_aux_loss": 1.3267443776130676, "router_z_loss": 0.819265827536583, "routing_entropy_normalized": 0.6774508094810252, "learning_rate": 0.003, "grad_norm": 0.7592461705207825, "tokens_per_second": 4154425.2011153162, "eta_seconds": 10237.335010528564, "data_wait_seconds": 2.2143125534057617e-05} {"step": 8340, "total_steps": 28610, "tokens": 17490247680, "source": "dclm", "phase": 2, "loss": 3.565559357404709, "ce_loss": 3.55031681060791, "logit_z_loss": 13.320757746696472, "router_aux_loss": 1.3067111372947693, "router_z_loss": 0.8432917296886444, "routing_entropy_normalized": 0.6973772012953144, "learning_rate": 0.003, "grad_norm": 0.833213210105896, "tokens_per_second": 3659176.9103087233, "eta_seconds": 11617.167489290237, "data_wait_seconds": 2.205371856689453e-05} {"step": 8350, "total_steps": 28610, "tokens": 17511219200, "source": "fineweb_hq", "phase": 2, "loss": 3.376485824584961, "ce_loss": 3.3602503836154938, "logit_z_loss": 13.608278393745422, "router_aux_loss": 1.4012425392866135, "router_z_loss": 0.8621479570865631, "routing_entropy_normalized": 0.6614712199103673, "learning_rate": 0.003, "grad_norm": 0.889891505241394, "tokens_per_second": 3825067.9684724845, "eta_seconds": 11107.85478067398, "data_wait_seconds": 1.9431114196777344e-05} {"step": 8360, "total_steps": 28610, "tokens": 17532190720, "source": "fineweb_hq", "phase": 2, "loss": 3.3643798828125, "ce_loss": 3.3486246466636658, "logit_z_loss": 13.603789687156677, "router_aux_loss": 1.358225017786026, "router_z_loss": 0.8125976324081421, "routing_entropy_normalized": 0.6492128982253891, "learning_rate": 0.003, "grad_norm": 0.6680364608764648, "tokens_per_second": 4190419.999986661, "eta_seconds": 10134.384620189667, "data_wait_seconds": 2.2143125534057617e-05} {"step": 8370, "total_steps": 28610, "tokens": 17553162240, "source": "cosmopedia_v2", "phase": 2, "loss": 2.820621609687805, "ce_loss": 2.804375946521759, "logit_z_loss": 14.911307692527771, "router_aux_loss": 1.3598213493824005, "router_z_loss": 1.1563361436128616, "routing_entropy_normalized": 0.6606513444789284, "learning_rate": 0.003, "grad_norm": 0.9147685766220093, "tokens_per_second": 3922954.0254844814, "eta_seconds": 10819.99845123291, "data_wait_seconds": 2.014636993408203e-05} {"step": 8380, "total_steps": 28610, "tokens": 17574133760, "source": "dclm", "phase": 2, "loss": 3.524656891822815, "ce_loss": 3.508481353521347, "logit_z_loss": 13.418685913085938, "router_aux_loss": 1.4099626392126083, "router_z_loss": 0.7339852601289749, "routing_entropy_normalized": 0.6777945952099009, "learning_rate": 0.003, "grad_norm": 0.8026266694068909, "tokens_per_second": 3891909.688566008, "eta_seconds": 10900.917121648788, "data_wait_seconds": 2.142786979675293e-05} {"step": 8390, "total_steps": 28610, "tokens": 17595105280, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8607980012893677, "ce_loss": 2.8448043167591095, "logit_z_loss": 14.739695191383362, "router_aux_loss": 1.341891586780548, "router_z_loss": 1.10075943171978, "routing_entropy_normalized": 0.6853864352814677, "learning_rate": 0.003, "grad_norm": 0.7648921608924866, "tokens_per_second": 3552694.420098441, "eta_seconds": 11935.845987796783, "data_wait_seconds": 2.065300941467285e-05} {"step": 8400, "total_steps": 28610, "tokens": 17616076800, "source": "fineweb_hq", "phase": 2, "loss": 3.340981513261795, "ce_loss": 3.3250835239887238, "logit_z_loss": 13.713794112205505, "router_aux_loss": 1.3727469742298126, "router_z_loss": 0.7991535738110542, "routing_entropy_normalized": 0.6527284434639078, "learning_rate": 0.003, "grad_norm": 0.7062779068946838, "tokens_per_second": 3546605.4845552933, "eta_seconds": 11950.424738407135, "data_wait_seconds": 1.8715858459472656e-05} {"step": 8410, "total_steps": 28610, "tokens": 17637048320, "source": "fineweb_hq", "phase": 2, "loss": 3.3258309364318848, "ce_loss": 3.309205949306488, "logit_z_loss": 13.749231576919556, "router_aux_loss": 1.4451050460338593, "router_z_loss": 0.7990674003958702, "routing_entropy_normalized": 0.6539834246343453, "learning_rate": 0.003, "grad_norm": 0.9290656447410583, "tokens_per_second": 4433390.609389707, "eta_seconds": 9555.32100200653, "data_wait_seconds": 1.8417835235595703e-05} {"step": 8420, "total_steps": 28610, "tokens": 17658019840, "source": "fineweb_edu", "phase": 2, "loss": 3.4160008430480957, "ce_loss": 3.398511081933975, "logit_z_loss": 13.619012713432312, "router_aux_loss": 1.519356682896614, "router_z_loss": 0.9343866035342216, "routing_entropy_normalized": 0.6643740285478001, "learning_rate": 0.003, "grad_norm": 0.48843511939048767, "tokens_per_second": 3687053.54144035, "eta_seconds": 11483.830761909485, "data_wait_seconds": 1.895427703857422e-05} {"step": 8430, "total_steps": 28610, "tokens": 17678991360, "source": "dclm", "phase": 2, "loss": 3.529561012983322, "ce_loss": 3.5117010474205017, "logit_z_loss": 13.507156491279602, "router_aux_loss": 1.5582540035247803, "router_z_loss": 0.9266767501831055, "routing_entropy_normalized": 0.6535248650143417, "learning_rate": 0.003, "grad_norm": 0.832891047000885, "tokens_per_second": 4164923.6558097275, "eta_seconds": 10161.177216529846, "data_wait_seconds": 2.2411346435546875e-05} {"step": 8440, "total_steps": 28610, "tokens": 17699962880, "source": "dclm", "phase": 2, "loss": 3.5457697212696075, "ce_loss": 3.527936726808548, "logit_z_loss": 13.570540308952332, "router_aux_loss": 1.5442235618829727, "router_z_loss": 1.0336526557803154, "routing_entropy_normalized": 0.6403673966734456, "learning_rate": 0.003, "grad_norm": 0.846234917640686, "tokens_per_second": 3352212.7770578754, "eta_seconds": 12618.398250102997, "data_wait_seconds": 2.1159648895263672e-05} {"step": 8450, "total_steps": 28610, "tokens": 17720934400, "source": "fineweb_hq", "phase": 2, "loss": 3.321437031030655, "ce_loss": 3.3046557307243347, "logit_z_loss": 13.34907841682434, "router_aux_loss": 1.4291957914829254, "router_z_loss": 1.1544161513447762, "routing_entropy_normalized": 0.6469248958678724, "learning_rate": 0.003, "grad_norm": 0.5472351312637329, "tokens_per_second": 3512777.1413309164, "eta_seconds": 12035.658004760742, "data_wait_seconds": 0.08872944116592407} {"step": 8460, "total_steps": 28610, "tokens": 17741905920, "source": "fineweb_hq", "phase": 2, "loss": 3.3901161551475525, "ce_loss": 3.3740899860858917, "logit_z_loss": 13.83036994934082, "router_aux_loss": 1.349325105547905, "router_z_loss": 1.149835243821144, "routing_entropy_normalized": 0.6725734818118327, "learning_rate": 0.003, "grad_norm": 0.6700328588485718, "tokens_per_second": 4485175.3022670755, "eta_seconds": 9421.61898970604, "data_wait_seconds": 2.2143125534057617e-05} {"step": 8470, "total_steps": 28610, "tokens": 17762877440, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8801624178886414, "ce_loss": 2.8636731803417206, "logit_z_loss": 14.614343643188477, "router_aux_loss": 1.3734881281852722, "router_z_loss": 1.29291433095932, "routing_entropy_normalized": 0.6719502020353731, "learning_rate": 0.003, "grad_norm": 1.607040286064148, "tokens_per_second": 3588557.3320672065, "eta_seconds": 11769.810921669006, "data_wait_seconds": 2.2470951080322266e-05} {"step": 8480, "total_steps": 28610, "tokens": 17783848960, "source": "dclm", "phase": 2, "loss": 3.5498668551445007, "ce_loss": 3.5325877964496613, "logit_z_loss": 13.34841775894165, "router_aux_loss": 1.5010515749454498, "router_z_loss": 0.9336742907762527, "routing_entropy_normalized": 0.6469310142687227, "learning_rate": 0.003, "grad_norm": 0.5302123427391052, "tokens_per_second": 3727484.689000714, "eta_seconds": 11325.511244773865, "data_wait_seconds": 1.817941665649414e-05} {"step": 8490, "total_steps": 28610, "tokens": 17804820480, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8474183082580566, "ce_loss": 2.8295229375362396, "logit_z_loss": 15.35475504398346, "router_aux_loss": 1.483932614326477, "router_z_loss": 1.5205791145563126, "routing_entropy_normalized": 0.6569250102246836, "learning_rate": 0.003, "grad_norm": 0.7360159754753113, "tokens_per_second": 3095399.2676931466, "eta_seconds": 13631.423474311829, "data_wait_seconds": 1.8596649169921875e-05} {"step": 8500, "total_steps": 28610, "tokens": 17825792000, "source": "fineweb_hq", "phase": 2, "loss": 3.30666646361351, "ce_loss": 3.289773553609848, "logit_z_loss": 13.884443759918213, "router_aux_loss": 1.4184411317110062, "router_z_loss": 1.3200241476297379, "routing_entropy_normalized": 0.6704010343406868, "learning_rate": 0.003, "grad_norm": 0.8277547359466553, "tokens_per_second": 4174136.638703347, "eta_seconds": 10103.580780982971, "data_wait_seconds": 1.9222497940063477e-05} {"step": 8510, "total_steps": 28610, "tokens": 17846763520, "source": "fineweb_hq", "phase": 2, "loss": 3.3261484503746033, "ce_loss": 3.3088265657424927, "logit_z_loss": 13.536576628684998, "router_aux_loss": 1.4797624498605728, "router_z_loss": 1.1705352365970612, "routing_entropy_normalized": 0.6197557430316921, "learning_rate": 0.003, "grad_norm": 0.6961969137191772, "tokens_per_second": 4406756.906805993, "eta_seconds": 9565.482301712036, "data_wait_seconds": 2.2917985916137695e-05} {"step": 8520, "total_steps": 28610, "tokens": 17867735040, "source": "fineweb_edu", "phase": 2, "loss": 3.3847381472587585, "ce_loss": 3.3674329817295074, "logit_z_loss": 13.47518014907837, "router_aux_loss": 1.463511511683464, "router_z_loss": 1.3226075172424316, "routing_entropy_normalized": 0.6629178167211843, "learning_rate": 0.003, "grad_norm": 0.6269166469573975, "tokens_per_second": 4098038.344490151, "eta_seconds": 10280.963753461838, "data_wait_seconds": 1.9311904907226562e-05} {"step": 8530, "total_steps": 28610, "tokens": 17888706560, "source": "dclm", "phase": 2, "loss": 3.5756761133670807, "ce_loss": 3.5577147603034973, "logit_z_loss": 13.13459599018097, "router_aux_loss": 1.5532551854848862, "router_z_loss": 1.1154240667819977, "routing_entropy_normalized": 0.6273798529709502, "learning_rate": 0.003, "grad_norm": 1.0244618654251099, "tokens_per_second": 3824810.3300303314, "eta_seconds": 11009.908603668213, "data_wait_seconds": 2.002716064453125e-05} {"step": 8540, "total_steps": 28610, "tokens": 17909678080, "source": "dclm", "phase": 2, "loss": 3.5792973041534424, "ce_loss": 3.5614941716194153, "logit_z_loss": 13.347468137741089, "router_aux_loss": 1.5208890736103058, "router_z_loss": 1.2594938725233078, "routing_entropy_normalized": 0.6622697404154967, "learning_rate": 0.003, "grad_norm": 0.8976821303367615, "tokens_per_second": 3839277.15425451, "eta_seconds": 10962.959679365158, "data_wait_seconds": 1.8537044525146484e-05} {"step": 8550, "total_steps": 28610, "tokens": 17930649600, "source": "fineweb_hq", "phase": 2, "loss": 3.3697158098220825, "ce_loss": 3.3523733615875244, "logit_z_loss": 13.56824541091919, "router_aux_loss": 1.4561160206794739, "router_z_loss": 1.4245248138904572, "routing_entropy_normalized": 0.651460037967541, "learning_rate": 0.003, "grad_norm": 0.8287673592567444, "tokens_per_second": 3725827.339975904, "eta_seconds": 11291.148322582245, "data_wait_seconds": 0.16367104649543762} {"step": 8560, "total_steps": 28610, "tokens": 17951621120, "source": "fineweb_hq", "phase": 2, "loss": 3.337870329618454, "ce_loss": 3.320661574602127, "logit_z_loss": 13.8742116689682, "router_aux_loss": 1.44521364569664, "router_z_loss": 1.3691959828138351, "routing_entropy_normalized": 0.6727892984271788, "learning_rate": 0.003, "grad_norm": 0.5955328345298767, "tokens_per_second": 4042768.7977146003, "eta_seconds": 10400.767321586609, "data_wait_seconds": 2.1010637283325195e-05} {"step": 8570, "total_steps": 28610, "tokens": 17972592640, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8698922991752625, "ce_loss": 2.853033661842346, "logit_z_loss": 14.756487250328064, "router_aux_loss": 1.3699473291635513, "router_z_loss": 1.6834561079740524, "routing_entropy_normalized": 0.698413257790435, "learning_rate": 0.003, "grad_norm": 0.8723888397216797, "tokens_per_second": 3387453.331433463, "eta_seconds": 12406.643566131592, "data_wait_seconds": 2.2917985916137695e-05} {"step": 8580, "total_steps": 28610, "tokens": 17993564160, "source": "dclm", "phase": 2, "loss": 3.516329139471054, "ce_loss": 3.5003131330013275, "logit_z_loss": 13.638847589492798, "router_aux_loss": 1.3424525707960129, "router_z_loss": 1.2276125997304916, "routing_entropy_normalized": 0.6772316355970619, "learning_rate": 0.003, "grad_norm": 0.6237224340438843, "tokens_per_second": 4134948.2317609275, "eta_seconds": 10158.761901140213, "data_wait_seconds": 1.5676021575927734e-05} {"step": 8590, "total_steps": 28610, "tokens": 18014535680, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8644485473632812, "ce_loss": 2.847822993993759, "logit_z_loss": 15.259680032730103, "router_aux_loss": 1.3432286828756332, "router_z_loss": 1.6672940403223038, "routing_entropy_normalized": 0.6854553241435175, "learning_rate": 0.003, "grad_norm": 1.1494468450546265, "tokens_per_second": 4570018.935070079, "eta_seconds": 9187.047939300537, "data_wait_seconds": 2.1070241928100586e-05} {"step": 8600, "total_steps": 28610, "tokens": 18035507200, "source": "fineweb_hq", "phase": 2, "loss": 3.3012146949768066, "ce_loss": 3.285227745771408, "logit_z_loss": 13.884129881858826, "router_aux_loss": 1.3452036678791046, "router_z_loss": 1.1465409100055695, "routing_entropy_normalized": 0.6924189492528178, "learning_rate": 0.003, "grad_norm": 0.7104718089103699, "tokens_per_second": 4119357.3618558194, "eta_seconds": 10187.028663396835, "data_wait_seconds": 2.014636993408203e-05} {"step": 8610, "total_steps": 28610, "tokens": 18056478720, "source": "fineweb_hq", "phase": 2, "loss": 3.308272898197174, "ce_loss": 3.293069064617157, "logit_z_loss": 13.813982486724854, "router_aux_loss": 1.2727324813604355, "router_z_loss": 1.0951853096485138, "routing_entropy_normalized": 0.6745920913674862, "learning_rate": 0.003, "grad_norm": 0.47232967615127563, "tokens_per_second": 3320994.8963493193, "eta_seconds": 12629.661083221436, "data_wait_seconds": 1.7583370208740234e-05} {"step": 8620, "total_steps": 28610, "tokens": 18077450240, "source": "fineweb_edu", "phase": 2, "loss": 3.3997184336185455, "ce_loss": 3.3842901289463043, "logit_z_loss": 13.537749648094177, "router_aux_loss": 1.3023318350315094, "router_z_loss": 1.05123420804739, "routing_entropy_normalized": 0.7083653669310359, "learning_rate": 0.003, "grad_norm": 0.7939979434013367, "tokens_per_second": 4134706.826583552, "eta_seconds": 10139.06674361229, "data_wait_seconds": 2.4586915969848633e-05} {"step": 8630, "total_steps": 28610, "tokens": 18098421760, "source": "dclm", "phase": 2, "loss": 3.5213459730148315, "ce_loss": 3.5043891966342926, "logit_z_loss": 13.221383452415466, "router_aux_loss": 1.4791814237833023, "router_z_loss": 0.8428818136453629, "routing_entropy_normalized": 0.6294682403105976, "learning_rate": 0.003, "grad_norm": 0.7982775568962097, "tokens_per_second": 4020404.8573472966, "eta_seconds": 10422.108829021454, "data_wait_seconds": 2.1517276763916016e-05} {"step": 8640, "total_steps": 28610, "tokens": 18119393280, "source": "dclm", "phase": 2, "loss": 3.522765964269638, "ce_loss": 3.5063039362430573, "logit_z_loss": 13.381699323654175, "router_aux_loss": 1.4181172400712967, "router_z_loss": 0.9427964836359024, "routing_entropy_normalized": 0.6749391052056549, "learning_rate": 0.003, "grad_norm": 0.7904776334762573, "tokens_per_second": 4291672.682820346, "eta_seconds": 9758.46215105057, "data_wait_seconds": 2.3573637008666992e-05} {"step": 8650, "total_steps": 28610, "tokens": 18140364800, "source": "fineweb_hq", "phase": 2, "loss": 3.337501585483551, "ce_loss": 3.320294886827469, "logit_z_loss": 13.650112748146057, "router_aux_loss": 1.4815777242183685, "router_z_loss": 1.0258712768554688, "routing_entropy_normalized": 0.6464352403152285, "learning_rate": 0.003, "grad_norm": 0.7342316508293152, "tokens_per_second": 4083202.979002086, "eta_seconds": 10251.54863357544, "data_wait_seconds": 2.2858381271362305e-05} {"step": 8660, "total_steps": 28610, "tokens": 18161336320, "source": "fineweb_hq", "phase": 2, "loss": 3.3727518916130066, "ce_loss": 3.35561603307724, "logit_z_loss": 13.81833803653717, "router_aux_loss": 1.4625253081321716, "router_z_loss": 1.1287221610546112, "routing_entropy_normalized": 0.6481141489920701, "learning_rate": 0.003, "grad_norm": 0.7066700458526611, "tokens_per_second": 3974484.5050549433, "eta_seconds": 10526.69405221939, "data_wait_seconds": 2.446770668029785e-05} {"step": 8670, "total_steps": 28610, "tokens": 18182307840, "source": "cosmopedia_v2", "phase": 2, "loss": 2.818023145198822, "ce_loss": 2.800786107778549, "logit_z_loss": 15.344322443008423, "router_aux_loss": 1.419453039765358, "router_z_loss": 1.5079859346151352, "routing_entropy_normalized": 0.6627131801982266, "learning_rate": 0.003, "grad_norm": 0.684301495552063, "tokens_per_second": 4157445.180632371, "eta_seconds": 10058.39140701294, "data_wait_seconds": 1.710653305053711e-05} {"step": 8680, "total_steps": 28610, "tokens": 18203279360, "source": "dclm", "phase": 2, "loss": 3.544731080532074, "ce_loss": 3.5284432768821716, "logit_z_loss": 13.365829706192017, "router_aux_loss": 1.3968659043312073, "router_z_loss": 0.9826352223753929, "routing_entropy_normalized": 0.6601911982039295, "learning_rate": 0.003, "grad_norm": 0.9027292728424072, "tokens_per_second": 4319481.407548729, "eta_seconds": 9676.216984510422, "data_wait_seconds": 2.2023916244506836e-05} {"step": 8690, "total_steps": 28610, "tokens": 18224250880, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8546474277973175, "ce_loss": 2.8384243845939636, "logit_z_loss": 14.711917161941528, "router_aux_loss": 1.3213261812925339, "router_z_loss": 1.5386408120393753, "routing_entropy_normalized": 0.694812959579721, "learning_rate": 0.003, "grad_norm": 0.6425183415412903, "tokens_per_second": 3637653.593587409, "eta_seconds": 11484.124797821045, "data_wait_seconds": 1.9878149032592773e-05} {"step": 8700, "total_steps": 28610, "tokens": 18245222400, "source": "fineweb_hq", "phase": 2, "loss": 3.343934178352356, "ce_loss": 3.32806196808815, "logit_z_loss": 13.9970782995224, "router_aux_loss": 1.3387972712516785, "router_z_loss": 1.0845907181501389, "routing_entropy_normalized": 0.6790449035168957, "learning_rate": 0.003, "grad_norm": 0.5668056607246399, "tokens_per_second": 3679473.3331392384, "eta_seconds": 11347.900239944458, "data_wait_seconds": 1.919269561767578e-05} {"step": 8710, "total_steps": 28610, "tokens": 18266193920, "source": "fineweb_hq", "phase": 2, "loss": 3.3388654589653015, "ce_loss": 3.323069244623184, "logit_z_loss": 13.529451370239258, "router_aux_loss": 1.3449171781539917, "router_z_loss": 0.9941174760460854, "routing_entropy_normalized": 0.6715542882163441, "learning_rate": 0.003, "grad_norm": 0.7631997466087341, "tokens_per_second": 3728361.404699135, "eta_seconds": 11193.476240634918, "data_wait_seconds": 2.053380012512207e-05} {"step": 8720, "total_steps": 28610, "tokens": 18287165440, "source": "fineweb_edu", "phase": 2, "loss": 3.3574139773845673, "ce_loss": 3.3412562012672424, "logit_z_loss": 13.70479941368103, "router_aux_loss": 1.3730567544698715, "router_z_loss": 1.0567035898566246, "routing_entropy_normalized": 0.6616324622309947, "learning_rate": 0.003, "grad_norm": 0.45134401321411133, "tokens_per_second": 3294824.2111161384, "eta_seconds": 12659.963205099106, "data_wait_seconds": 1.9341707229614258e-05} {"step": 8730, "total_steps": 28610, "tokens": 18308136960, "source": "dclm", "phase": 2, "loss": 3.505919933319092, "ce_loss": 3.489650547504425, "logit_z_loss": 13.075218677520752, "router_aux_loss": 1.4106047302484512, "router_z_loss": 0.8558337986469269, "routing_entropy_normalized": 0.6312708530984213, "learning_rate": 0.003, "grad_norm": 0.7459851503372192, "tokens_per_second": 4014635.016534179, "eta_seconds": 10384.849827766418, "data_wait_seconds": 1.800060272216797e-05} {"step": 8740, "total_steps": 28610, "tokens": 18329108480, "source": "dclm", "phase": 2, "loss": 3.535453587770462, "ce_loss": 3.5184790790081024, "logit_z_loss": 13.784366011619568, "router_aux_loss": 1.4713269174098969, "router_z_loss": 0.8827952146530151, "routing_entropy_normalized": 0.6798848005909058, "learning_rate": 0.003, "grad_norm": 0.6475213170051575, "tokens_per_second": 3633024.9971069517, "eta_seconds": 11469.893621206284, "data_wait_seconds": 1.9222497940063477e-05} {"step": 8750, "total_steps": 28610, "tokens": 18350080000, "source": "fineweb_hq", "phase": 2, "loss": 3.3383609354496, "ce_loss": 3.321887344121933, "logit_z_loss": 13.56670594215393, "router_aux_loss": 1.4169850945472717, "router_z_loss": 0.9470838755369186, "routing_entropy_normalized": 0.6773741371967046, "learning_rate": 0.003, "grad_norm": 0.4165440797805786, "tokens_per_second": 3891600.4401266836, "eta_seconds": 10702.393362522125, "data_wait_seconds": 1.8268823623657227e-05} {"step": 8760, "total_steps": 28610, "tokens": 18371051520, "source": "fineweb_hq", "phase": 2, "loss": 3.338255673646927, "ce_loss": 3.3222383558750153, "logit_z_loss": 13.851006031036377, "router_aux_loss": 1.3673399835824966, "router_z_loss": 0.9588648304343224, "routing_entropy_normalized": 0.6593379879150336, "learning_rate": 0.003, "grad_norm": 0.7054234743118286, "tokens_per_second": 3554516.834378199, "eta_seconds": 11711.427780389786, "data_wait_seconds": 2.5153160095214844e-05} {"step": 8770, "total_steps": 28610, "tokens": 18392023040, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8827917277812958, "ce_loss": 2.866830676794052, "logit_z_loss": 14.892629146575928, "router_aux_loss": 1.3220978826284409, "router_z_loss": 1.2509299218654633, "routing_entropy_normalized": 0.7123345533598562, "learning_rate": 0.003, "grad_norm": 1.0553563833236694, "tokens_per_second": 4062280.1657980224, "eta_seconds": 10242.399337768555, "data_wait_seconds": 2.092123031616211e-05} {"step": 8780, "total_steps": 28610, "tokens": 18412994560, "source": "dclm", "phase": 2, "loss": 3.564423441886902, "ce_loss": 3.5486063361167908, "logit_z_loss": 13.367238998413086, "router_aux_loss": 1.3627468794584274, "router_z_loss": 0.8528971001505852, "routing_entropy_normalized": 0.6688333013925877, "learning_rate": 0.003, "grad_norm": 0.6319741606712341, "tokens_per_second": 3791075.594785714, "eta_seconds": 10969.584520339966, "data_wait_seconds": 1.767277717590332e-05} {"step": 8790, "total_steps": 28610, "tokens": 18433966080, "source": "cosmopedia_v2", "phase": 2, "loss": 2.858407109975815, "ce_loss": 2.8409332036972046, "logit_z_loss": 14.92580235004425, "router_aux_loss": 1.460784837603569, "router_z_loss": 1.3735356628894806, "routing_entropy_normalized": 0.6562851842233753, "learning_rate": 0.003, "grad_norm": 1.4232274293899536, "tokens_per_second": 3720214.5005829795, "eta_seconds": 11172.891410827637, "data_wait_seconds": 1.9609928131103516e-05} {"step": 8800, "total_steps": 28610, "tokens": 18454937600, "source": "fineweb_hq", "phase": 2, "loss": 3.351560652256012, "ce_loss": 3.3349654376506805, "logit_z_loss": 13.651885271072388, "router_aux_loss": 1.4173970818519592, "router_z_loss": 1.0560130551457405, "routing_entropy_normalized": 0.6552626561896074, "learning_rate": 0.003, "grad_norm": 0.43026936054229736, "tokens_per_second": 3424905.2387522296, "eta_seconds": 12130.140317440033, "data_wait_seconds": 2.053380012512207e-05} {"step": 8810, "total_steps": 28610, "tokens": 18475909120, "source": "fineweb_hq", "phase": 2, "loss": 3.359585851430893, "ce_loss": 3.3437344431877136, "logit_z_loss": 13.73332417011261, "router_aux_loss": 1.3435721099376678, "router_z_loss": 1.0423589125275612, "routing_entropy_normalized": 0.6613895821384659, "learning_rate": 0.003, "grad_norm": 1.1206178665161133, "tokens_per_second": 3045630.59197901, "eta_seconds": 13633.829956054688, "data_wait_seconds": 2.2113323211669922e-05} {"step": 8820, "total_steps": 28610, "tokens": 18496880640, "source": "fineweb_edu", "phase": 2, "loss": 3.3603017330169678, "ce_loss": 3.3447488248348236, "logit_z_loss": 13.759790897369385, "router_aux_loss": 1.3120883405208588, "router_z_loss": 1.0560225993394852, "routing_entropy_normalized": 0.664979494875117, "learning_rate": 0.003, "grad_norm": 0.7905035018920898, "tokens_per_second": 2780041.4170615054, "eta_seconds": 14928.784091234205, "data_wait_seconds": 2.002716064453125e-05} {"step": 8830, "total_steps": 28610, "tokens": 18517852160, "source": "dclm", "phase": 2, "loss": 3.4882083237171173, "ce_loss": 3.4717793464660645, "logit_z_loss": 13.934199452400208, "router_aux_loss": 1.422281950712204, "router_z_loss": 0.8127643018960953, "routing_entropy_normalized": 0.6492204764888363, "learning_rate": 0.003, "grad_norm": 0.8721280694007874, "tokens_per_second": 3867572.236879699, "eta_seconds": 10725.50530910492, "data_wait_seconds": 1.9788742065429688e-05} {"step": 8840, "total_steps": 28610, "tokens": 18538823680, "source": "dclm", "phase": 2, "loss": 3.485218644142151, "ce_loss": 3.468274176120758, "logit_z_loss": 13.313234567642212, "router_aux_loss": 1.4792760163545609, "router_z_loss": 0.8203870430588722, "routing_entropy_normalized": 0.6373516505756213, "learning_rate": 0.003, "grad_norm": 0.5543493032455444, "tokens_per_second": 3820627.2782553486, "eta_seconds": 10851.803125619888, "data_wait_seconds": 0.08449974656105042} {"step": 8850, "total_steps": 28610, "tokens": 18559795200, "source": "fineweb_hq", "phase": 2, "loss": 3.3774470686912537, "ce_loss": 3.361530512571335, "logit_z_loss": 13.546004891395569, "router_aux_loss": 1.3492137342691422, "router_z_loss": 1.069827988743782, "routing_entropy_normalized": 0.6455988605010899, "learning_rate": 0.003, "grad_norm": 0.5830363631248474, "tokens_per_second": 4057410.4696633588, "eta_seconds": 10213.342679977417, "data_wait_seconds": 2.1129846572875977e-05} {"step": 8860, "total_steps": 28610, "tokens": 18580766720, "source": "fineweb_hq", "phase": 2, "loss": 3.3663365840911865, "ce_loss": 3.3508220314979553, "logit_z_loss": 13.701852321624756, "router_aux_loss": 1.3180918097496033, "router_z_loss": 0.9633875340223312, "routing_entropy_normalized": 0.6690026230657135, "learning_rate": 0.003, "grad_norm": 0.7656638026237488, "tokens_per_second": 3801257.7821926526, "eta_seconds": 10896.065032482147, "data_wait_seconds": 1.8984079360961914e-05} {"step": 8870, "total_steps": 28610, "tokens": 18601738240, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8544039726257324, "ce_loss": 2.8391954004764557, "logit_z_loss": 15.032023906707764, "router_aux_loss": 1.245479241013527, "router_z_loss": 1.250541090965271, "routing_entropy_normalized": 0.7080131988798042, "learning_rate": 0.003, "grad_norm": 1.2029143571853638, "tokens_per_second": 3910064.238087855, "eta_seconds": 10587.493698120117, "data_wait_seconds": 1.9609928131103516e-05} {"step": 8880, "total_steps": 28610, "tokens": 18622709760, "source": "dclm", "phase": 2, "loss": 3.473894476890564, "ce_loss": 3.4589363634586334, "logit_z_loss": 13.3782958984375, "router_aux_loss": 1.2882883995771408, "router_z_loss": 0.7374129444360733, "routing_entropy_normalized": 0.6788253137887982, "learning_rate": 0.003, "grad_norm": 0.784562885761261, "tokens_per_second": 3718541.129017639, "eta_seconds": 11127.161842346191, "data_wait_seconds": 1.7344951629638672e-05} {"step": 8890, "total_steps": 28610, "tokens": 18643681280, "source": "cosmopedia_v2", "phase": 2, "loss": 2.810754358768463, "ce_loss": 2.7950207591056824, "logit_z_loss": 15.038628816604614, "router_aux_loss": 1.3106894493103027, "router_z_loss": 1.122864916920662, "routing_entropy_normalized": 0.6868026685330431, "learning_rate": 0.003, "grad_norm": 1.0279057025909424, "tokens_per_second": 4006063.968189383, "eta_seconds": 10323.30930519104, "data_wait_seconds": 2.345442771911621e-05} {"step": 8900, "total_steps": 28610, "tokens": 18664652800, "source": "fineweb_hq", "phase": 2, "loss": 3.374888092279434, "ce_loss": 3.3590966165065765, "logit_z_loss": 13.613433361053467, "router_aux_loss": 1.3629528135061264, "router_z_loss": 0.8006566241383553, "routing_entropy_normalized": 0.6577540275711684, "learning_rate": 0.003, "grad_norm": 0.4593677222728729, "tokens_per_second": 3657035.6415048703, "eta_seconds": 11302.833762645721, "data_wait_seconds": 1.901388168334961e-05} {"step": 8910, "total_steps": 28610, "tokens": 18685624320, "source": "fineweb_hq", "phase": 2, "loss": 3.373537451028824, "ce_loss": 3.357893645763397, "logit_z_loss": 13.86295998096466, "router_aux_loss": 1.3457051664590836, "router_z_loss": 0.8004793897271156, "routing_entropy_normalized": 0.6602527036481385, "learning_rate": 0.003, "grad_norm": 0.7221881151199341, "tokens_per_second": 4079042.352545639, "eta_seconds": 10128.331806659698, "data_wait_seconds": 1.9788742065429688e-05} {"step": 8920, "total_steps": 28610, "tokens": 18706595840, "source": "fineweb_edu", "phase": 2, "loss": 3.310430496931076, "ce_loss": 3.294738858938217, "logit_z_loss": 13.496432304382324, "router_aux_loss": 1.3551058620214462, "router_z_loss": 0.7909670025110245, "routing_entropy_normalized": 0.6603493957477045, "learning_rate": 0.003, "grad_norm": 0.7623026371002197, "tokens_per_second": 3745928.176210073, "eta_seconds": 11023.415542840958, "data_wait_seconds": 1.8805265426635742e-05} {"step": 8930, "total_steps": 28610, "tokens": 18727567360, "source": "dclm", "phase": 2, "loss": 3.4289675056934357, "ce_loss": 3.4132223427295685, "logit_z_loss": 13.497960329055786, "router_aux_loss": 1.3667697757482529, "router_z_loss": 0.7277063652873039, "routing_entropy_normalized": 0.6526954168817495, "learning_rate": 0.003, "grad_norm": 0.77330082654953, "tokens_per_second": 3797084.754778736, "eta_seconds": 10869.378490447998, "data_wait_seconds": 1.9878149032592773e-05} {"step": 8940, "total_steps": 28610, "tokens": 18748538880, "source": "dclm", "phase": 2, "loss": 3.466980129480362, "ce_loss": 3.45182067155838, "logit_z_loss": 13.39622700214386, "router_aux_loss": 1.3132954090833664, "router_z_loss": 0.6868507415056229, "routing_entropy_normalized": 0.6997685848736087, "learning_rate": 0.003, "grad_norm": 0.5217427611351013, "tokens_per_second": 3727224.2342811087, "eta_seconds": 11067.48004603386, "data_wait_seconds": 2.1576881408691406e-05} {"step": 8950, "total_steps": 28610, "tokens": 18769510400, "source": "fineweb_hq", "phase": 2, "loss": 3.3849134147167206, "ce_loss": 3.3691175878047943, "logit_z_loss": 13.321549534797668, "router_aux_loss": 1.3710833042860031, "router_z_loss": 0.7527740076184273, "routing_entropy_normalized": 0.6620437496828762, "learning_rate": 0.003, "grad_norm": 0.4081685245037079, "tokens_per_second": 4446918.141478852, "eta_seconds": 9271.591472625732, "data_wait_seconds": 2.3871660232543945e-05} {"step": 8960, "total_steps": 28610, "tokens": 18790481920, "source": "fineweb_hq", "phase": 2, "loss": 3.411264330148697, "ce_loss": 3.395578622817993, "logit_z_loss": 13.789646744728088, "router_aux_loss": 1.3571233302354813, "router_z_loss": 0.7355483621358871, "routing_entropy_normalized": 0.6349315532712865, "learning_rate": 0.003, "grad_norm": 0.6419451832771301, "tokens_per_second": 3814850.977219084, "eta_seconds": 10802.266470193863, "data_wait_seconds": 1.7136335372924805e-05} {"step": 8970, "total_steps": 28610, "tokens": 18811453440, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8168397545814514, "ce_loss": 2.8012145459651947, "logit_z_loss": 14.703675270080566, "router_aux_loss": 1.314636379480362, "router_z_loss": 1.0084489285945892, "routing_entropy_normalized": 0.6608623852148761, "learning_rate": 0.003, "grad_norm": 0.7814335823059082, "tokens_per_second": 3623593.008322394, "eta_seconds": 11366.636701583862, "data_wait_seconds": 2.059340476989746e-05} {"step": 8980, "total_steps": 28610, "tokens": 18832424960, "source": "dclm", "phase": 2, "loss": 3.436695009469986, "ce_loss": 3.42133691906929, "logit_z_loss": 13.815854907035828, "router_aux_loss": 1.3368070125579834, "router_z_loss": 0.6084339171648026, "routing_entropy_normalized": 0.6711120868473643, "learning_rate": 0.003, "grad_norm": 0.9989234209060669, "tokens_per_second": 3710453.942498274, "eta_seconds": 11094.8941552639, "data_wait_seconds": 0.0604383647441864} {"step": 8990, "total_steps": 28610, "tokens": 18853396480, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8460017442703247, "ce_loss": 2.83030503988266, "logit_z_loss": 14.952178359031677, "router_aux_loss": 1.3214637786149979, "router_z_loss": 0.9868045821785927, "routing_entropy_normalized": 0.678355042551917, "learning_rate": 0.003, "grad_norm": 1.1213288307189941, "tokens_per_second": 4218237.190756634, "eta_seconds": 9754.340588092804, "data_wait_seconds": 2.1010637283325195e-05} {"step": 9000, "total_steps": 28610, "tokens": 18874368000, "source": "fineweb_hq", "phase": 2, "loss": 3.3766978085041046, "ce_loss": 3.3597829341888428, "logit_z_loss": 13.477827906608582, "router_aux_loss": 1.481158897280693, "router_z_loss": 0.755514957010746, "routing_entropy_normalized": 0.6435115999697478, "learning_rate": 0.003, "grad_norm": 0.6548168063163757, "tokens_per_second": 3743890.394812817, "eta_seconds": 10984.60328245163, "data_wait_seconds": 2.4020671844482422e-05} {"step": 9010, "total_steps": 28610, "tokens": 18895339520, "source": "fineweb_hq", "phase": 2, "loss": 3.393621116876602, "ce_loss": 3.377608358860016, "logit_z_loss": 13.517650604248047, "router_aux_loss": 1.3899832516908646, "router_z_loss": 0.7611120119690895, "routing_entropy_normalized": 0.6476790835491871, "learning_rate": 0.003, "grad_norm": 0.9158778190612793, "tokens_per_second": 4033799.276954171, "eta_seconds": 10189.941635131836, "data_wait_seconds": 2.1785497665405273e-05} {"step": 9020, "total_steps": 28610, "tokens": 18916311040, "source": "fineweb_edu", "phase": 2, "loss": 3.324644297361374, "ce_loss": 3.307721883058548, "logit_z_loss": 13.633291125297546, "router_aux_loss": 1.4568044245243073, "router_z_loss": 0.9910435155034065, "routing_entropy_normalized": 0.6549969173979501, "learning_rate": 0.003, "grad_norm": 0.523764431476593, "tokens_per_second": 3557082.922873787, "eta_seconds": 11549.690735578537, "data_wait_seconds": 2.065300941467285e-05} {"step": 9030, "total_steps": 28610, "tokens": 18937282560, "source": "dclm", "phase": 2, "loss": 3.4398599565029144, "ce_loss": 3.4218665063381195, "logit_z_loss": 13.621184349060059, "router_aux_loss": 1.570765033364296, "router_z_loss": 0.9237572848796844, "routing_entropy_normalized": 0.6415523589708089, "learning_rate": 0.003, "grad_norm": 0.680250883102417, "tokens_per_second": 3271459.5480301045, "eta_seconds": 12551.65639591217, "data_wait_seconds": 1.627206802368164e-05} {"step": 9040, "total_steps": 28610, "tokens": 18958254080, "source": "dclm", "phase": 2, "loss": 3.398404538631439, "ce_loss": 3.3808811008930206, "logit_z_loss": 13.715479016304016, "router_aux_loss": 1.5090360790491104, "router_z_loss": 1.0614830702543259, "routing_entropy_normalized": 0.6660680500090266, "learning_rate": 0.003, "grad_norm": 0.5711096525192261, "tokens_per_second": 4052084.780274387, "eta_seconds": 10128.431872844696, "data_wait_seconds": 1.8984079360961914e-05} {"step": 9050, "total_steps": 28610, "tokens": 18979225600, "source": "fineweb_hq", "phase": 2, "loss": 3.3715497255325317, "ce_loss": 3.3544546365737915, "logit_z_loss": 13.876452088356018, "router_aux_loss": 1.448796808719635, "router_z_loss": 1.2194980382919312, "routing_entropy_normalized": 0.6376845688642203, "learning_rate": 0.003, "grad_norm": 0.9332059025764465, "tokens_per_second": 3387225.9656728534, "eta_seconds": 12110.291293144226, "data_wait_seconds": 0.08266225457191467} {"step": 9060, "total_steps": 28610, "tokens": 19000197120, "source": "fineweb_hq", "phase": 2, "loss": 3.3540407717227936, "ce_loss": 3.3378136456012726, "logit_z_loss": 13.301485419273376, "router_aux_loss": 1.373938113451004, "router_z_loss": 1.1576833128929138, "routing_entropy_normalized": 0.6778857887830431, "learning_rate": 0.003, "grad_norm": 0.8159254193305969, "tokens_per_second": 4235611.855354133, "eta_seconds": 9679.66919541359, "data_wait_seconds": 2.2798776626586914e-05} {"step": 9070, "total_steps": 28610, "tokens": 19021168640, "source": "cosmopedia_v2", "phase": 2, "loss": 2.7992712557315826, "ce_loss": 2.78293839097023, "logit_z_loss": 15.188122868537903, "router_aux_loss": 1.3350282907485962, "router_z_loss": 1.4637689292430878, "routing_entropy_normalized": 0.6582049223645359, "learning_rate": 0.003, "grad_norm": 0.5553438067436218, "tokens_per_second": 3971477.105307821, "eta_seconds": 10318.163492679596, "data_wait_seconds": 2.2590160369873047e-05} {"step": 9080, "total_steps": 28610, "tokens": 19042140160, "source": "dclm", "phase": 2, "loss": 3.4313292503356934, "ce_loss": 3.4158373177051544, "logit_z_loss": 13.444826602935791, "router_aux_loss": 1.3195390701293945, "router_z_loss": 0.9520742148160934, "routing_entropy_normalized": 0.6658272256876652, "learning_rate": 0.003, "grad_norm": 0.45390453934669495, "tokens_per_second": 3484118.226260348, "eta_seconds": 11755.450274705887, "data_wait_seconds": 2.014636993408203e-05} {"step": 9090, "total_steps": 28610, "tokens": 19063111680, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8232817947864532, "ce_loss": 2.807033807039261, "logit_z_loss": 15.449306726455688, "router_aux_loss": 1.3302590548992157, "router_z_loss": 1.400505244731903, "routing_entropy_normalized": 0.6727480737526995, "learning_rate": 0.003, "grad_norm": 1.1609549522399902, "tokens_per_second": 4330339.48803877, "eta_seconds": 9453.394393920898, "data_wait_seconds": 2.0176172256469727e-05} {"step": 9100, "total_steps": 28610, "tokens": 19084083200, "source": "fineweb_hq", "phase": 2, "loss": 3.36990550160408, "ce_loss": 3.3538032472133636, "logit_z_loss": 13.388581275939941, "router_aux_loss": 1.3778923898935318, "router_z_loss": 0.9845245257019997, "routing_entropy_normalized": 0.6595492476643428, "learning_rate": 0.003, "grad_norm": 0.8072081208229065, "tokens_per_second": 3574491.381876196, "eta_seconds": 11446.505572080612, "data_wait_seconds": 2.22623348236084e-05} {"step": 9110, "total_steps": 28610, "tokens": 19105054720, "source": "fineweb_hq", "phase": 2, "loss": 3.3402150571346283, "ce_loss": 3.324158251285553, "logit_z_loss": 13.463426947593689, "router_aux_loss": 1.3711177855730057, "router_z_loss": 0.9992475807666779, "routing_entropy_normalized": 0.6480396536887784, "learning_rate": 0.003, "grad_norm": 0.41899022459983826, "tokens_per_second": 4232138.635527899, "eta_seconds": 9662.836575508118, "data_wait_seconds": 2.345442771911621e-05} {"step": 9120, "total_steps": 28610, "tokens": 19126026240, "source": "fineweb_edu", "phase": 2, "loss": 3.3129289746284485, "ce_loss": 3.296822190284729, "logit_z_loss": 13.582919955253601, "router_aux_loss": 1.3763247579336166, "router_z_loss": 0.9852678179740906, "routing_entropy_normalized": 0.6569162417360418, "learning_rate": 0.003, "grad_norm": 0.6928442120552063, "tokens_per_second": 3550622.6086563393, "eta_seconds": 11511.640910625458, "data_wait_seconds": 1.6033649444580078e-05} {"step": 9130, "total_steps": 28610, "tokens": 19146997760, "source": "dclm", "phase": 2, "loss": 3.4553528130054474, "ce_loss": 3.439318746328354, "logit_z_loss": 13.319480895996094, "router_aux_loss": 1.3932765424251556, "router_z_loss": 0.7693398669362068, "routing_entropy_normalized": 0.6456235939906743, "learning_rate": 0.003, "grad_norm": 0.7643023729324341, "tokens_per_second": 4085481.8217526292, "eta_seconds": 9999.437702178955, "data_wait_seconds": 2.0056962966918945e-05} {"step": 9140, "total_steps": 28610, "tokens": 19167969280, "source": "dclm", "phase": 2, "loss": 3.4057995080947876, "ce_loss": 3.3902514576911926, "logit_z_loss": 13.421104907989502, "router_aux_loss": 1.3386246860027313, "router_z_loss": 0.8196392431855202, "routing_entropy_normalized": 0.6582804512253077, "learning_rate": 0.003, "grad_norm": 0.5063745975494385, "tokens_per_second": 3698284.7986411178, "eta_seconds": 11040.67200422287, "data_wait_seconds": 2.0176172256469727e-05} {"step": 9150, "total_steps": 28610, "tokens": 19188940800, "source": "fineweb_hq", "phase": 2, "loss": 3.360076457262039, "ce_loss": 3.3445383310317993, "logit_z_loss": 13.50653076171875, "router_aux_loss": 1.3259550780057907, "router_z_loss": 0.9279568940401077, "routing_entropy_normalized": 0.6816880707237494, "learning_rate": 0.003, "grad_norm": 0.5031633377075195, "tokens_per_second": 3373597.5985762, "eta_seconds": 12097.049730300903, "data_wait_seconds": 2.0295381546020508e-05} {"step": 9160, "total_steps": 28610, "tokens": 19209912320, "source": "fineweb_hq", "phase": 2, "loss": 3.35394886136055, "ce_loss": 3.338516741991043, "logit_z_loss": 13.246549844741821, "router_aux_loss": 1.3214725106954575, "router_z_loss": 0.8927563205361366, "routing_entropy_normalized": 0.6679964794800604, "learning_rate": 0.003, "grad_norm": 0.561743438243866, "tokens_per_second": 3858948.1714469194, "eta_seconds": 10570.135847330093, "data_wait_seconds": 2.384185791015625e-05} {"step": 9170, "total_steps": 28610, "tokens": 19230883840, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8094185292720795, "ce_loss": 2.7940043210983276, "logit_z_loss": 15.457553505897522, "router_aux_loss": 1.271160364151001, "router_z_loss": 1.1568538546562195, "routing_entropy_normalized": 0.693372296624488, "learning_rate": 0.003, "grad_norm": 0.5434641242027283, "tokens_per_second": 3747209.9203939317, "eta_seconds": 10879.73071861267, "data_wait_seconds": 2.327561378479004e-05} {"step": 9180, "total_steps": 28610, "tokens": 19251855360, "source": "dclm", "phase": 2, "loss": 3.441731631755829, "ce_loss": 3.4261813163757324, "logit_z_loss": 13.488655805587769, "router_aux_loss": 1.3480583280324936, "router_z_loss": 0.7208914235234261, "routing_entropy_normalized": 0.679147187465072, "learning_rate": 0.003, "grad_norm": 0.764572262763977, "tokens_per_second": 3775969.506066686, "eta_seconds": 10791.311554431915, "data_wait_seconds": 2.0623207092285156e-05} {"step": 9190, "total_steps": 28610, "tokens": 19272826880, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8034227192401886, "ce_loss": 2.7871333956718445, "logit_z_loss": 15.90307891368866, "router_aux_loss": 1.3504562973976135, "router_z_loss": 1.1944782137870789, "routing_entropy_normalized": 0.6402791323359405, "learning_rate": 0.003, "grad_norm": 0.7098405361175537, "tokens_per_second": 4170514.302616164, "eta_seconds": 9765.388363361359, "data_wait_seconds": 2.1010637283325195e-05} {"step": 9200, "total_steps": 28610, "tokens": 19293798400, "source": "fineweb_hq", "phase": 2, "loss": 3.3633422553539276, "ce_loss": 3.34747651219368, "logit_z_loss": 13.668589115142822, "router_aux_loss": 1.3615947514772415, "router_z_loss": 0.8830079436302185, "routing_entropy_normalized": 0.6603531381315602, "learning_rate": 0.003, "grad_norm": 0.7757441997528076, "tokens_per_second": 4014206.1152598, "eta_seconds": 10140.416100025177, "data_wait_seconds": 0.05644667148590088} {"step": 9210, "total_steps": 28610, "tokens": 19314769920, "source": "fineweb_hq", "phase": 2, "loss": 3.3378460705280304, "ce_loss": 3.3218274116516113, "logit_z_loss": 13.22516679763794, "router_aux_loss": 1.378505989909172, "router_z_loss": 0.9111118167638779, "routing_entropy_normalized": 0.6530125276744216, "learning_rate": 0.003, "grad_norm": 0.719852089881897, "tokens_per_second": 3921271.9900334054, "eta_seconds": 10375.395765304565, "data_wait_seconds": 1.823902130126953e-05} {"step": 9220, "total_steps": 28610, "tokens": 19335741440, "source": "fineweb_edu", "phase": 2, "loss": 3.319674253463745, "ce_loss": 3.303575783967972, "logit_z_loss": 13.740975379943848, "router_aux_loss": 1.37896229326725, "router_z_loss": 0.9348107874393463, "routing_entropy_normalized": 0.6644692605410996, "learning_rate": 0.003, "grad_norm": 0.8839129209518433, "tokens_per_second": 3964217.600889533, "eta_seconds": 10257.705649375916, "data_wait_seconds": 1.952052116394043e-05} {"step": 9230, "total_steps": 28610, "tokens": 19356712960, "source": "dclm", "phase": 2, "loss": 3.4771229326725006, "ce_loss": 3.461138993501663, "logit_z_loss": 13.336704850196838, "router_aux_loss": 1.383519485592842, "router_z_loss": 0.8151136934757233, "routing_entropy_normalized": 0.6480381885659021, "learning_rate": 0.003, "grad_norm": 0.9629214406013489, "tokens_per_second": 3790454.800730021, "eta_seconds": 10722.4087600708, "data_wait_seconds": 2.2172927856445312e-05} {"step": 9240, "total_steps": 28610, "tokens": 19377684480, "source": "dclm", "phase": 2, "loss": 3.460374563932419, "ce_loss": 3.4446226954460144, "logit_z_loss": 13.491337537765503, "router_aux_loss": 1.3516236990690231, "router_z_loss": 0.8865359649062157, "routing_entropy_normalized": 0.6644820229192052, "learning_rate": 0.003, "grad_norm": 0.776250958442688, "tokens_per_second": 3096272.0350377704, "eta_seconds": 13119.594719171524, "data_wait_seconds": 1.952052116394043e-05} {"step": 9250, "total_steps": 28610, "tokens": 19398656000, "source": "fineweb_hq", "phase": 2, "loss": 3.3930194079875946, "ce_loss": 3.376847356557846, "logit_z_loss": 13.42078685760498, "router_aux_loss": 1.3930834829807281, "router_z_loss": 0.8990916907787323, "routing_entropy_normalized": 0.6565301243783389, "learning_rate": 0.003, "grad_norm": 0.9363809823989868, "tokens_per_second": 4134551.152688074, "eta_seconds": 9819.896095275879, "data_wait_seconds": 2.086162567138672e-05} {"step": 9260, "total_steps": 28610, "tokens": 19419627520, "source": "fineweb_hq", "phase": 2, "loss": 3.3749566972255707, "ce_loss": 3.358865886926651, "logit_z_loss": 13.612156629562378, "router_aux_loss": 1.375143215060234, "router_z_loss": 0.9782571867108345, "routing_entropy_normalized": 0.6651677813100861, "learning_rate": 0.003, "grad_norm": 0.6069183945655823, "tokens_per_second": 3484214.418651742, "eta_seconds": 11646.783556938171, "data_wait_seconds": 2.181529998779297e-05} {"step": 9270, "total_steps": 28610, "tokens": 19440599040, "source": "cosmopedia_v2", "phase": 2, "loss": 2.797176957130432, "ce_loss": 2.7813746631145477, "logit_z_loss": 15.616778254508972, "router_aux_loss": 1.290004625916481, "router_z_loss": 1.3405737727880478, "routing_entropy_normalized": 0.6998358122523758, "learning_rate": 0.003, "grad_norm": 1.2113672494888306, "tokens_per_second": 3487294.8291085023, "eta_seconds": 11630.481983184814, "data_wait_seconds": 2.5153160095214844e-05} {"step": 9280, "total_steps": 28610, "tokens": 19461570560, "source": "dclm", "phase": 2, "loss": 3.4392316043376923, "ce_loss": 3.4230890572071075, "logit_z_loss": 13.781565427780151, "router_aux_loss": 1.3921652436256409, "router_z_loss": 0.8427326530218124, "routing_entropy_normalized": 0.6434373036151309, "learning_rate": 0.003, "grad_norm": 0.8489583134651184, "tokens_per_second": 3637022.020913317, "eta_seconds": 11145.917711496353, "data_wait_seconds": 0.11086723208427429} {"step": 9290, "total_steps": 28610, "tokens": 19482542080, "source": "cosmopedia_v2", "phase": 2, "loss": 2.802410840988159, "ce_loss": 2.785269796848297, "logit_z_loss": 15.041929364204407, "router_aux_loss": 1.4334459453821182, "router_z_loss": 1.3023986965417862, "routing_entropy_normalized": 0.6604328670079074, "learning_rate": 0.003, "grad_norm": 0.7796560525894165, "tokens_per_second": 3702736.6294390378, "eta_seconds": 10942.440874099731, "data_wait_seconds": 2.1010637283325195e-05} {"step": 9300, "total_steps": 28610, "tokens": 19503513600, "source": "fineweb_hq", "phase": 2, "loss": 3.374429315328598, "ce_loss": 3.357781022787094, "logit_z_loss": 13.352804899215698, "router_aux_loss": 1.4218773543834686, "router_z_loss": 1.094223216176033, "routing_entropy_normalized": 0.6321178500433304, "learning_rate": 0.003, "grad_norm": 0.6268999576568604, "tokens_per_second": 3721520.7447870597, "eta_seconds": 10881.574468374252, "data_wait_seconds": 1.677870750427246e-05} {"step": 9310, "total_steps": 28610, "tokens": 19524485120, "source": "fineweb_hq", "phase": 2, "loss": 3.3599404096603394, "ce_loss": 3.3438578844070435, "logit_z_loss": 13.811996936798096, "router_aux_loss": 1.3517210930585861, "router_z_loss": 1.18411785364151, "routing_entropy_normalized": 0.6638272329039457, "learning_rate": 0.003, "grad_norm": 0.7562911510467529, "tokens_per_second": 3964326.407271641, "eta_seconds": 10209.813582897186, "data_wait_seconds": 2.5898218154907227e-05} {"step": 9320, "total_steps": 28610, "tokens": 19545456640, "source": "fineweb_edu", "phase": 2, "loss": 3.3445452749729156, "ce_loss": 3.328676402568817, "logit_z_loss": 13.368104457855225, "router_aux_loss": 1.346440076828003, "router_z_loss": 1.0676789432764053, "routing_entropy_normalized": 0.65580942265169, "learning_rate": 0.003, "grad_norm": 0.6432446837425232, "tokens_per_second": 4021717.6961470786, "eta_seconds": 10058.901478528976, "data_wait_seconds": 2.0742416381835938e-05} {"step": 9330, "total_steps": 28610, "tokens": 19566428160, "source": "dclm", "phase": 2, "loss": 3.44521564245224, "ce_loss": 3.4298000037670135, "logit_z_loss": 13.525102138519287, "router_aux_loss": 1.3168793767690659, "router_z_loss": 0.8943974003195763, "routing_entropy_normalized": 0.6460079611693709, "learning_rate": 0.003, "grad_norm": 0.5825583338737488, "tokens_per_second": 3816591.471964273, "eta_seconds": 10594.031574249268, "data_wait_seconds": 2.294778823852539e-05} {"step": 9340, "total_steps": 28610, "tokens": 19587399680, "source": "dclm", "phase": 2, "loss": 3.4881880581378937, "ce_loss": 3.4723182916641235, "logit_z_loss": 13.471039295196533, "router_aux_loss": 1.3629063665866852, "router_z_loss": 0.8935418650507927, "routing_entropy_normalized": 0.6632836839085493, "learning_rate": 0.003, "grad_norm": 0.6974406242370605, "tokens_per_second": 4010810.2057387996, "eta_seconds": 10075.799393892288, "data_wait_seconds": 2.3186206817626953e-05} {"step": 9350, "total_steps": 28610, "tokens": 19608371200, "source": "fineweb_hq", "phase": 2, "loss": 3.34092977643013, "ce_loss": 3.3252635300159454, "logit_z_loss": 13.444580793380737, "router_aux_loss": 1.3379901349544525, "router_z_loss": 0.9419821351766586, "routing_entropy_normalized": 0.6562131399181951, "learning_rate": 0.003, "grad_norm": 0.4520401954650879, "tokens_per_second": 3709251.0222610477, "eta_seconds": 10889.300097942352, "data_wait_seconds": 2.4318695068359375e-05} {"step": 9360, "total_steps": 28610, "tokens": 19629342720, "source": "fineweb_hq", "phase": 2, "loss": 3.351543605327606, "ce_loss": 3.336124897003174, "logit_z_loss": 13.358993291854858, "router_aux_loss": 1.3185589462518692, "router_z_loss": 0.897202841937542, "routing_entropy_normalized": 0.6711649087454202, "learning_rate": 0.003, "grad_norm": 0.5888988375663757, "tokens_per_second": 4264614.609509476, "eta_seconds": 9466.312831640244, "data_wait_seconds": 2.136826515197754e-05} {"step": 9370, "total_steps": 28610, "tokens": 19650314240, "source": "cosmopedia_v2", "phase": 2, "loss": 2.7992488145828247, "ce_loss": 2.7836078703403473, "logit_z_loss": 14.980653643608093, "router_aux_loss": 1.2999285161495209, "router_z_loss": 1.1435204595327377, "routing_entropy_normalized": 0.6743322625603292, "learning_rate": 0.003, "grad_norm": 0.7569377422332764, "tokens_per_second": 3243303.253282676, "eta_seconds": 12440.774521827698, "data_wait_seconds": 2.2977590560913086e-05} {"step": 9380, "total_steps": 28610, "tokens": 19671285760, "source": "dclm", "phase": 2, "loss": 3.5216886401176453, "ce_loss": 3.5062012672424316, "logit_z_loss": 13.149864912033081, "router_aux_loss": 1.339444637298584, "router_z_loss": 0.7780248448252678, "routing_entropy_normalized": 0.6468259366773387, "learning_rate": 0.003, "grad_norm": 1.0518858432769775, "tokens_per_second": 3983960.0752215884, "eta_seconds": 10122.64987564087, "data_wait_seconds": 1.9162893295288086e-05} {"step": 9390, "total_steps": 28610, "tokens": 19692257280, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8062437176704407, "ce_loss": 2.7895541191101074, "logit_z_loss": 15.02555525302887, "router_aux_loss": 1.408086121082306, "router_z_loss": 1.1061632931232452, "routing_entropy_normalized": 0.6371128337517674, "learning_rate": 0.003, "grad_norm": 1.240259051322937, "tokens_per_second": 3464458.8611511486, "eta_seconds": 11634.504277706146, "data_wait_seconds": 2.2172927856445312e-05} {"step": 9400, "total_steps": 28610, "tokens": 19713228800, "source": "fineweb_hq", "phase": 2, "loss": 3.343887448310852, "ce_loss": 3.3275418281555176, "logit_z_loss": 13.685080170631409, "router_aux_loss": 1.4020161628723145, "router_z_loss": 0.9569225907325745, "routing_entropy_normalized": 0.6603995696480395, "learning_rate": 0.003, "grad_norm": 0.7277445793151855, "tokens_per_second": 3112485.3333138954, "eta_seconds": 12943.447311639786, "data_wait_seconds": 0.1939443051815033} {"step": 9410, "total_steps": 28610, "tokens": 19734200320, "source": "fineweb_hq", "phase": 2, "loss": 3.3612046241760254, "ce_loss": 3.3450766801834106, "logit_z_loss": 12.945428013801575, "router_aux_loss": 1.3856246918439865, "router_z_loss": 0.9771860465407372, "routing_entropy_normalized": 0.6512556385432164, "learning_rate": 0.003, "grad_norm": 0.5370526909828186, "tokens_per_second": 3899508.91897189, "eta_seconds": 10325.740814208984, "data_wait_seconds": 2.2292137145996094e-05} {"step": 9420, "total_steps": 28610, "tokens": 19755171840, "source": "fineweb_edu", "phase": 2, "loss": 3.3582099974155426, "ce_loss": 3.3418207466602325, "logit_z_loss": 13.581702709197998, "router_aux_loss": 1.4009826332330704, "router_z_loss": 1.0212372168898582, "routing_entropy_normalized": 0.6614358238422399, "learning_rate": 0.003, "grad_norm": 0.8626124262809753, "tokens_per_second": 3786039.3137062816, "eta_seconds": 10629.669569015503, "data_wait_seconds": 2.104043960571289e-05} {"step": 9430, "total_steps": 28610, "tokens": 19776143360, "source": "dclm", "phase": 2, "loss": 3.5121723115444183, "ce_loss": 3.4956573843955994, "logit_z_loss": 13.466117978096008, "router_aux_loss": 1.4256885349750519, "router_z_loss": 0.9113855138421059, "routing_entropy_normalized": 0.6597838603640345, "learning_rate": 0.003, "grad_norm": 0.840278148651123, "tokens_per_second": 3274806.2918981644, "eta_seconds": 12282.673164367676, "data_wait_seconds": 2.2292137145996094e-05} {"step": 9440, "total_steps": 28610, "tokens": 19797114880, "source": "dclm", "phase": 2, "loss": 3.523998647928238, "ce_loss": 3.507781594991684, "logit_z_loss": 13.774295806884766, "router_aux_loss": 1.3870625346899033, "router_z_loss": 0.9689927697181702, "routing_entropy_normalized": 0.6635516166152627, "learning_rate": 0.003, "grad_norm": 1.09757399559021, "tokens_per_second": 3814354.1972057507, "eta_seconds": 10539.766828536987, "data_wait_seconds": 0.07667112350463867} {"step": 9450, "total_steps": 28610, "tokens": 19818086400, "source": "fineweb_hq", "phase": 2, "loss": 3.3714236319065094, "ce_loss": 3.3552295565605164, "logit_z_loss": 13.540085673332214, "router_aux_loss": 1.3790055215358734, "router_z_loss": 1.0500743091106415, "routing_entropy_normalized": 0.6785638995586936, "learning_rate": 0.003, "grad_norm": 0.5207617282867432, "tokens_per_second": 3767588.273521789, "eta_seconds": 10665.027440071106, "data_wait_seconds": 2.3812055587768555e-05} {"step": 9460, "total_steps": 28610, "tokens": 19839057920, "source": "fineweb_hq", "phase": 2, "loss": 3.3623008131980896, "ce_loss": 3.3468739688396454, "logit_z_loss": 13.498942255973816, "router_aux_loss": 1.298050194978714, "router_z_loss": 1.096432626247406, "routing_entropy_normalized": 0.6557290461218808, "learning_rate": 0.003, "grad_norm": 0.43882066011428833, "tokens_per_second": 3985991.274962199, "eta_seconds": 10075.401080846786, "data_wait_seconds": 1.8209218978881836e-05} {"step": 9470, "total_steps": 28610, "tokens": 19860029440, "source": "cosmopedia_v2", "phase": 2, "loss": 2.809745341539383, "ce_loss": 2.794189304113388, "logit_z_loss": 15.11638355255127, "router_aux_loss": 1.268319010734558, "router_z_loss": 1.3611413389444351, "routing_entropy_normalized": 0.698585247149265, "learning_rate": 0.003, "grad_norm": 0.7220659255981445, "tokens_per_second": 3912073.311642222, "eta_seconds": 10260.413362026215, "data_wait_seconds": 1.8596649169921875e-05} {"step": 9480, "total_steps": 28610, "tokens": 19881000960, "source": "dclm", "phase": 2, "loss": 3.5079758167266846, "ce_loss": 3.492213726043701, "logit_z_loss": 13.480236887931824, "router_aux_loss": 1.3617486953735352, "router_z_loss": 0.7965730577707291, "routing_entropy_normalized": 0.6686645435319656, "learning_rate": 0.003, "grad_norm": 0.8729169964790344, "tokens_per_second": 3966198.490447271, "eta_seconds": 10115.105902194977, "data_wait_seconds": 2.002716064453125e-05} {"step": 9490, "total_steps": 28610, "tokens": 19901972480, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8008445501327515, "ce_loss": 2.784862458705902, "logit_z_loss": 15.301140308380127, "router_aux_loss": 1.3261152654886246, "router_z_loss": 1.190867081284523, "routing_entropy_normalized": 0.6834102837123333, "learning_rate": 0.003, "grad_norm": 0.8136059045791626, "tokens_per_second": 4357077.3485037545, "eta_seconds": 9202.85389328003, "data_wait_seconds": 2.0444393157958984e-05} {"step": 9500, "total_steps": 28610, "tokens": 19922944000, "source": "fineweb_hq", "phase": 2, "loss": 3.3525720834732056, "ce_loss": 3.3371261954307556, "logit_z_loss": 13.381532549858093, "router_aux_loss": 1.325112670660019, "router_z_loss": 0.8565841168165207, "routing_entropy_normalized": 0.6551722979367892, "learning_rate": 0.003, "grad_norm": 0.6075720191001892, "tokens_per_second": 3332427.125127413, "eta_seconds": 12026.241899728775, "data_wait_seconds": 2.1398067474365234e-05} {"step": 9510, "total_steps": 28610, "tokens": 19943915520, "source": "fineweb_hq", "phase": 2, "loss": 3.387376695871353, "ce_loss": 3.371194988489151, "logit_z_loss": 13.17530345916748, "router_aux_loss": 1.3959524929523468, "router_z_loss": 0.9046121761202812, "routing_entropy_normalized": 0.6573299498134878, "learning_rate": 0.003, "grad_norm": 0.7523320317268372, "tokens_per_second": 3514156.5445131976, "eta_seconds": 11398.354823589325, "data_wait_seconds": 0.08548116683959961} {"step": 9520, "total_steps": 28610, "tokens": 19964887040, "source": "fineweb_edu", "phase": 2, "loss": 3.321967124938965, "ce_loss": 3.3054207265377045, "logit_z_loss": 13.505279183387756, "router_aux_loss": 1.4237943291664124, "router_z_loss": 0.9579895734786987, "routing_entropy_normalized": 0.6383883388961064, "learning_rate": 0.003, "grad_norm": 0.723905622959137, "tokens_per_second": 4030940.6813702486, "eta_seconds": 9931.83349609375, "data_wait_seconds": 2.0623207092285156e-05} {"step": 9530, "total_steps": 28610, "tokens": 19985858560, "source": "dclm", "phase": 2, "loss": 3.544756054878235, "ce_loss": 3.528519481420517, "logit_z_loss": 13.050813436508179, "router_aux_loss": 1.409374013543129, "router_z_loss": 0.8377406224608421, "routing_entropy_normalized": 0.663450559514403, "learning_rate": 0.003, "grad_norm": 0.9389444589614868, "tokens_per_second": 3972714.9372357423, "eta_seconds": 10072.119644165039, "data_wait_seconds": 1.7017126083374023e-05} {"step": 9540, "total_steps": 28610, "tokens": 20006830080, "source": "dclm", "phase": 2, "loss": 3.4492189288139343, "ce_loss": 3.434042662382126, "logit_z_loss": 12.965073347091675, "router_aux_loss": 1.3028420060873032, "router_z_loss": 0.8513098657131195, "routing_entropy_normalized": 0.6941660279594314, "learning_rate": 0.003, "grad_norm": 1.0723150968551636, "tokens_per_second": 3766681.4013071395, "eta_seconds": 10617.486423492432, "data_wait_seconds": 2.0235776901245117e-05} {"step": 9550, "total_steps": 28610, "tokens": 20027801600, "source": "fineweb_hq", "phase": 2, "loss": 3.4116016030311584, "ce_loss": 3.3957544565200806, "logit_z_loss": 13.616966962814331, "router_aux_loss": 1.356582835316658, "router_z_loss": 0.9196378514170647, "routing_entropy_normalized": 0.6581568711482372, "learning_rate": 0.003, "grad_norm": 0.7193130850791931, "tokens_per_second": 4039015.2172320676, "eta_seconds": 9896.401714324951, "data_wait_seconds": 2.014636993408203e-05} {"step": 9560, "total_steps": 28610, "tokens": 20048773120, "source": "fineweb_hq", "phase": 2, "loss": 3.4013358056545258, "ce_loss": 3.3858571350574493, "logit_z_loss": 12.831567406654358, "router_aux_loss": 1.315879449248314, "router_z_loss": 1.0366939455270767, "routing_entropy_normalized": 0.6738523319053763, "learning_rate": 0.003, "grad_norm": 0.7698196768760681, "tokens_per_second": 4068815.4599882304, "eta_seconds": 9818.765680789948, "data_wait_seconds": 1.8537044525146484e-05} {"step": 9570, "total_steps": 28610, "tokens": 20069744640, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8114942610263824, "ce_loss": 2.7953672409057617, "logit_z_loss": 15.416551113128662, "router_aux_loss": 1.3279806077480316, "router_z_loss": 1.3055010885000229, "routing_entropy_normalized": 0.6665079700777907, "learning_rate": 0.003, "grad_norm": 0.5992413759231567, "tokens_per_second": 3647274.7077890467, "eta_seconds": 10947.838394165039, "data_wait_seconds": 1.9997358322143555e-05} {"step": 9580, "total_steps": 28610, "tokens": 20090716160, "source": "dclm", "phase": 2, "loss": 3.5257411301136017, "ce_loss": 3.5089393854141235, "logit_z_loss": 13.369712591171265, "router_aux_loss": 1.450739398598671, "router_z_loss": 0.9574047103524208, "routing_entropy_normalized": 0.6315303511793523, "learning_rate": 0.003, "grad_norm": 1.0916447639465332, "tokens_per_second": 3725667.319727958, "eta_seconds": 10711.853511095047, "data_wait_seconds": 2.1129846572875977e-05} {"step": 9590, "total_steps": 28610, "tokens": 20111687680, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8307859003543854, "ce_loss": 2.812772214412689, "logit_z_loss": 15.661763548851013, "router_aux_loss": 1.5163734704256058, "router_z_loss": 1.2838120758533478, "routing_entropy_normalized": 0.6708559825369061, "learning_rate": 0.003, "grad_norm": 0.7777507901191711, "tokens_per_second": 4215380.379283417, "eta_seconds": 9462.451179027557, "data_wait_seconds": 2.0384788513183594e-05} {"step": 9600, "total_steps": 28610, "tokens": 20132659200, "source": "fineweb_hq", "phase": 2, "loss": 3.3652065694332123, "ce_loss": 3.3489055037498474, "logit_z_loss": 13.776365160942078, "router_aux_loss": 1.3851664066314697, "router_z_loss": 1.0718087181448936, "routing_entropy_normalized": 0.6555850962283765, "learning_rate": 0.003, "grad_norm": 0.8035048842430115, "tokens_per_second": 3255833.286741159, "eta_seconds": 12244.748428106308, "data_wait_seconds": 2.4974346160888672e-05} {"step": 9610, "total_steps": 28610, "tokens": 20153630720, "source": "fineweb_hq", "phase": 2, "loss": 3.3747565150260925, "ce_loss": 3.358974903821945, "logit_z_loss": 13.48589313030243, "router_aux_loss": 1.3423554003238678, "router_z_loss": 1.0095181614160538, "routing_entropy_normalized": 0.6818151698990991, "learning_rate": 0.003, "grad_norm": 0.4330331087112427, "tokens_per_second": 4341679.784959802, "eta_seconds": 9177.528047561646, "data_wait_seconds": 2.1666288375854492e-05} {"step": 9620, "total_steps": 28610, "tokens": 20174602240, "source": "fineweb_edu", "phase": 2, "loss": 3.3728518784046173, "ce_loss": 3.357306957244873, "logit_z_loss": 13.324427723884583, "router_aux_loss": 1.3200866729021072, "router_z_loss": 1.0116269886493683, "routing_entropy_normalized": 0.6653812213488974, "learning_rate": 0.003, "grad_norm": 0.779562771320343, "tokens_per_second": 3999232.08927341, "eta_seconds": 9958.140860795975, "data_wait_seconds": 2.1278858184814453e-05} {"step": 9630, "total_steps": 28610, "tokens": 20195573760, "source": "dclm", "phase": 2, "loss": 3.5141130685806274, "ce_loss": 3.4988473653793335, "logit_z_loss": 13.353128910064697, "router_aux_loss": 1.3143441528081894, "router_z_loss": 0.7869177535176277, "routing_entropy_normalized": 0.6659670954785699, "learning_rate": 0.003, "grad_norm": 0.743706464767456, "tokens_per_second": 3921715.8806671393, "eta_seconds": 10149.62485074997, "data_wait_seconds": 1.7911195755004883e-05} {"step": 9640, "total_steps": 28610, "tokens": 20216545280, "source": "dclm", "phase": 2, "loss": 3.500829428434372, "ce_loss": 3.4847500026226044, "logit_z_loss": 13.631345868110657, "router_aux_loss": 1.3954867124557495, "router_z_loss": 0.7614598870277405, "routing_entropy_normalized": 0.6442550303440411, "learning_rate": 0.003, "grad_norm": 0.7006415128707886, "tokens_per_second": 3442129.197445698, "eta_seconds": 11557.664212465286, "data_wait_seconds": 0.09528201818466187} {"step": 9650, "total_steps": 28610, "tokens": 20237516800, "source": "fineweb_hq", "phase": 2, "loss": 3.4117751717567444, "ce_loss": 3.3965370059013367, "logit_z_loss": 13.448293805122375, "router_aux_loss": 1.3016916662454605, "router_z_loss": 0.8764003068208694, "routing_entropy_normalized": 0.6663505147162966, "learning_rate": 0.003, "grad_norm": 0.6270675659179688, "tokens_per_second": 3570183.15796002, "eta_seconds": 11137.244270324707, "data_wait_seconds": 1.9431114196777344e-05} {"step": 9660, "total_steps": 28610, "tokens": 20258488320, "source": "fineweb_hq", "phase": 2, "loss": 3.347496837377548, "ce_loss": 3.3315584659576416, "logit_z_loss": 13.372081279754639, "router_aux_loss": 1.3725645393133163, "router_z_loss": 0.8755569532513618, "routing_entropy_normalized": 0.6464550346169127, "learning_rate": 0.003, "grad_norm": 1.0248883962631226, "tokens_per_second": 3168007.66127081, "eta_seconds": 12544.4868350029, "data_wait_seconds": 0.20856666564941406} {"step": 9670, "total_steps": 28610, "tokens": 20279459840, "source": "cosmopedia_v2", "phase": 2, "loss": 2.826336592435837, "ce_loss": 2.8108577132225037, "logit_z_loss": 14.871103644371033, "router_aux_loss": 1.2828364372253418, "router_z_loss": 1.1634111404418945, "routing_entropy_normalized": 0.6783557233004537, "learning_rate": 0.003, "grad_norm": 0.6818361878395081, "tokens_per_second": 3774838.4283277085, "eta_seconds": 10522.320261955261, "data_wait_seconds": 1.946091651916504e-05} {"step": 9680, "total_steps": 28610, "tokens": 20300431360, "source": "dclm", "phase": 2, "loss": 3.4955772757530212, "ce_loss": 3.4801000356674194, "logit_z_loss": 13.629050135612488, "router_aux_loss": 1.3340371400117874, "router_z_loss": 0.7739022150635719, "routing_entropy_normalized": 0.6719596241797048, "learning_rate": 0.003, "grad_norm": 0.9873731136322021, "tokens_per_second": 3545488.8634526804, "eta_seconds": 11197.070104837418, "data_wait_seconds": 1.9341707229614258e-05} {"step": 9690, "total_steps": 28610, "tokens": 20321402880, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8406037092208862, "ce_loss": 2.8245154917240143, "logit_z_loss": 15.485498309135437, "router_aux_loss": 1.3426957875490189, "router_z_loss": 1.112717978656292, "routing_entropy_normalized": 0.6879389009314467, "learning_rate": 0.003, "grad_norm": 0.583590567111969, "tokens_per_second": 4413573.241087484, "eta_seconds": 8990.020936012266, "data_wait_seconds": 2.199411392211914e-05} {"step": 9700, "total_steps": 28610, "tokens": 20342374400, "source": "fineweb_hq", "phase": 2, "loss": 3.3110315799713135, "ce_loss": 3.2959691882133484, "logit_z_loss": 13.608401656150818, "router_aux_loss": 1.280019223690033, "router_z_loss": 0.9013001695275307, "routing_entropy_normalized": 0.6942691685340457, "learning_rate": 0.003, "grad_norm": 0.6789621710777283, "tokens_per_second": 3528376.221293437, "eta_seconds": 11239.488601207733, "data_wait_seconds": 2.0295381546020508e-05} {"step": 9710, "total_steps": 28610, "tokens": 20363345920, "source": "fineweb_hq", "phase": 2, "loss": 3.2914507389068604, "ce_loss": 3.2762629687786102, "logit_z_loss": 13.714882969856262, "router_aux_loss": 1.2971040159463882, "router_z_loss": 0.8452039435505867, "routing_entropy_normalized": 0.6702100453653195, "learning_rate": 0.003, "grad_norm": 0.7285939455032349, "tokens_per_second": 3838396.7497938126, "eta_seconds": 10326.231336593628, "data_wait_seconds": 2.1636486053466797e-05} {"step": 9720, "total_steps": 28610, "tokens": 20384317440, "source": "fineweb_edu", "phase": 2, "loss": 3.3571343421936035, "ce_loss": 3.340777963399887, "logit_z_loss": 13.35551118850708, "router_aux_loss": 1.4107086509466171, "router_z_loss": 0.91382846981287, "routing_entropy_normalized": 0.6189145237196316, "learning_rate": 0.003, "grad_norm": 0.7956026792526245, "tokens_per_second": 3875151.747508426, "eta_seconds": 10222.877415180206, "data_wait_seconds": 1.800060272216797e-05} {"step": 9730, "total_steps": 28610, "tokens": 20405288960, "source": "dclm", "phase": 2, "loss": 3.480706512928009, "ce_loss": 3.4649330377578735, "logit_z_loss": 13.774279594421387, "router_aux_loss": 1.36336350440979, "router_z_loss": 0.762365959584713, "routing_entropy_normalized": 0.6717907866145011, "learning_rate": 0.003, "grad_norm": 0.7637696862220764, "tokens_per_second": 3247999.5426439624, "eta_seconds": 12190.343391418457, "data_wait_seconds": 2.0742416381835938e-05} {"step": 9740, "total_steps": 28610, "tokens": 20426260480, "source": "dclm", "phase": 2, "loss": 3.4771456718444824, "ce_loss": 3.461440861225128, "logit_z_loss": 13.661218047142029, "router_aux_loss": 1.3534004837274551, "router_z_loss": 0.8046906739473343, "routing_entropy_normalized": 0.666411918555957, "learning_rate": 0.003, "grad_norm": 0.4754737317562103, "tokens_per_second": 4138022.877815817, "eta_seconds": 9563.325145483017, "data_wait_seconds": 0.045915454626083374} {"step": 9750, "total_steps": 28610, "tokens": 20447232000, "source": "fineweb_hq", "phase": 2, "loss": 3.3380190432071686, "ce_loss": 3.322586953639984, "logit_z_loss": 13.611256957054138, "router_aux_loss": 1.3168609589338303, "router_z_loss": 0.9023467302322388, "routing_entropy_normalized": 0.6757081613878981, "learning_rate": 0.003, "grad_norm": 0.5859010815620422, "tokens_per_second": 3666998.4643493225, "eta_seconds": 10786.011258125305, "data_wait_seconds": 1.7255544662475586e-05} {"step": 9760, "total_steps": 28610, "tokens": 20468203520, "source": "fineweb_hq", "phase": 2, "loss": 3.323489636182785, "ce_loss": 3.3083632588386536, "logit_z_loss": 13.578155159950256, "router_aux_loss": 1.2936304062604904, "router_z_loss": 0.8322524279356003, "routing_entropy_normalized": 0.6737716207024972, "learning_rate": 0.003, "grad_norm": 0.571666419506073, "tokens_per_second": 4106245.036044337, "eta_seconds": 9627.120362520218, "data_wait_seconds": 2.047419548034668e-05} {"step": 9770, "total_steps": 28610, "tokens": 20489175040, "source": "cosmopedia_v2", "phase": 2, "loss": 2.823682337999344, "ce_loss": 2.808406412601471, "logit_z_loss": 14.948898315429688, "router_aux_loss": 1.2657777965068817, "router_z_loss": 1.123302936553955, "routing_entropy_normalized": 0.7138194412062565, "learning_rate": 0.003, "grad_norm": 0.7409166097640991, "tokens_per_second": 3607932.997097441, "eta_seconds": 10950.963809967041, "data_wait_seconds": 2.1457672119140625e-05} {"step": 9780, "total_steps": 28610, "tokens": 20510146560, "source": "dclm", "phase": 2, "loss": 3.485210418701172, "ce_loss": 3.4696465730667114, "logit_z_loss": 13.512932658195496, "router_aux_loss": 1.3555165827274323, "router_z_loss": 0.6574732288718224, "routing_entropy_normalized": 0.629256691618496, "learning_rate": 0.003, "grad_norm": 0.8813228011131287, "tokens_per_second": 4072390.0992743387, "eta_seconds": 9696.853984355927, "data_wait_seconds": 1.9282102584838867e-05} {"step": 9790, "total_steps": 28610, "tokens": 20531118080, "source": "cosmopedia_v2", "phase": 2, "loss": 2.831828773021698, "ce_loss": 2.8159350156784058, "logit_z_loss": 14.91248643398285, "router_aux_loss": 1.3308633416891098, "router_z_loss": 1.093851089477539, "routing_entropy_normalized": 0.6618603102290347, "learning_rate": 0.003, "grad_norm": 0.8401593565940857, "tokens_per_second": 4460201.198891026, "eta_seconds": 8849.017988204956, "data_wait_seconds": 2.1845102310180664e-05} {"step": 9800, "total_steps": 28610, "tokens": 20552089600, "source": "fineweb_hq", "phase": 2, "loss": 3.301353305578232, "ce_loss": 3.2858886122703552, "logit_z_loss": 13.470677256584167, "router_aux_loss": 1.3360978811979294, "router_z_loss": 0.756630077958107, "routing_entropy_normalized": 0.6425644759951479, "learning_rate": 0.003, "grad_norm": 0.746240496635437, "tokens_per_second": 4024778.491825842, "eta_seconds": 9801.14289522171, "data_wait_seconds": 1.7762184143066406e-05} {"step": 9810, "total_steps": 28610, "tokens": 20573061120, "source": "fineweb_hq", "phase": 2, "loss": 3.2791062891483307, "ce_loss": 3.2636345624923706, "logit_z_loss": 13.584623098373413, "router_aux_loss": 1.3264492601156235, "router_z_loss": 0.8488024175167084, "routing_entropy_normalized": 0.6668896173426359, "learning_rate": 0.003, "grad_norm": 0.79392009973526, "tokens_per_second": 3591177.075128255, "eta_seconds": 10978.700513839722, "data_wait_seconds": 2.238154411315918e-05} {"step": 9820, "total_steps": 28610, "tokens": 20594032640, "source": "fineweb_edu", "phase": 2, "loss": 3.3740733563899994, "ce_loss": 3.35789754986763, "logit_z_loss": 13.307977199554443, "router_aux_loss": 1.4000346213579178, "router_z_loss": 0.8446919396519661, "routing_entropy_normalized": 0.6426971192188446, "learning_rate": 0.003, "grad_norm": 0.49802812933921814, "tokens_per_second": 3961152.752705453, "eta_seconds": 9947.984473228455, "data_wait_seconds": 1.755356788635254e-05} {"step": 9830, "total_steps": 28610, "tokens": 20615004160, "source": "dclm", "phase": 2, "loss": 3.447213739156723, "ce_loss": 3.4314514994621277, "logit_z_loss": 13.60602331161499, "router_aux_loss": 1.3653936237096786, "router_z_loss": 0.7477281540632248, "routing_entropy_normalized": 0.6751273989916758, "learning_rate": 0.003, "grad_norm": 0.6579071283340454, "tokens_per_second": 3395171.6978021897, "eta_seconds": 11600.154002666473, "data_wait_seconds": 1.7374753952026367e-05} {"step": 9840, "total_steps": 28610, "tokens": 20635975680, "source": "dclm", "phase": 2, "loss": 3.460785001516342, "ce_loss": 3.4454930126667023, "logit_z_loss": 13.251764416694641, "router_aux_loss": 1.3221952021121979, "router_z_loss": 0.7448410019278526, "routing_entropy_normalized": 0.6651125955592361, "learning_rate": 0.003, "grad_norm": 0.6709786057472229, "tokens_per_second": 3897299.4571122816, "eta_seconds": 10100.21002316475, "data_wait_seconds": 1.856684684753418e-05} {"step": 9850, "total_steps": 28610, "tokens": 20656947200, "source": "fineweb_hq", "phase": 2, "loss": 3.3032378256320953, "ce_loss": 3.2880825996398926, "logit_z_loss": 13.5567946434021, "router_aux_loss": 1.2998356819152832, "router_z_loss": 0.8012795224785805, "routing_entropy_normalized": 0.6736412809156817, "learning_rate": 0.003, "grad_norm": 0.66937255859375, "tokens_per_second": 4145994.5064777415, "eta_seconds": 9489.296586990356, "data_wait_seconds": 1.9371509552001953e-05} {"step": 9860, "total_steps": 28610, "tokens": 20677918720, "source": "fineweb_hq", "phase": 2, "loss": 3.309024691581726, "ce_loss": 3.294298678636551, "logit_z_loss": 12.881400108337402, "router_aux_loss": 1.2634068578481674, "router_z_loss": 0.8037860915064812, "routing_entropy_normalized": 0.6783371694077406, "learning_rate": 0.003, "grad_norm": 0.777574360370636, "tokens_per_second": 4136630.103404313, "eta_seconds": 9505.708515644073, "data_wait_seconds": 2.014636993408203e-05} {"step": 9870, "total_steps": 28610, "tokens": 20698890240, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8206379115581512, "ce_loss": 2.805065006017685, "logit_z_loss": 14.975129961967468, "router_aux_loss": 1.3005337864160538, "router_z_loss": 1.0701037645339966, "routing_entropy_normalized": 0.6515118408266225, "learning_rate": 0.003, "grad_norm": 0.9253657460212708, "tokens_per_second": 4090340.807073927, "eta_seconds": 9608.155000686646, "data_wait_seconds": 2.4259090423583984e-05} {"step": 9880, "total_steps": 28610, "tokens": 20719861760, "source": "dclm", "phase": 2, "loss": 3.4304286539554596, "ce_loss": 3.415206104516983, "logit_z_loss": 13.273230195045471, "router_aux_loss": 1.329144150018692, "router_z_loss": 0.6038127690553665, "routing_entropy_normalized": 0.6558299759818738, "learning_rate": 0.003, "grad_norm": 0.7916330099105835, "tokens_per_second": 3898907.7554190853, "eta_seconds": 10074.52841258049, "data_wait_seconds": 1.9997358322143555e-05} {"step": 9890, "total_steps": 28610, "tokens": 20740833280, "source": "cosmopedia_v2", "phase": 2, "loss": 2.862585335969925, "ce_loss": 2.8469612896442413, "logit_z_loss": 14.905718088150024, "router_aux_loss": 1.3092959821224213, "router_z_loss": 1.0404870957136154, "routing_entropy_normalized": 0.652410680807794, "learning_rate": 0.003, "grad_norm": 1.6865458488464355, "tokens_per_second": 3710269.2604565867, "eta_seconds": 10581.087970733643, "data_wait_seconds": 2.187490463256836e-05} {"step": 9900, "total_steps": 28610, "tokens": 20761804800, "source": "fineweb_hq", "phase": 2, "loss": 3.3096850514411926, "ce_loss": 3.293632924556732, "logit_z_loss": 13.476289510726929, "router_aux_loss": 1.395253673195839, "router_z_loss": 0.7519213929772377, "routing_entropy_normalized": 0.6296685979773566, "learning_rate": 0.003, "grad_norm": 0.7849026322364807, "tokens_per_second": 3439755.686939242, "eta_seconds": 11407.122334003448, "data_wait_seconds": 1.850724220275879e-05} {"step": 9910, "total_steps": 28610, "tokens": 20782776320, "source": "fineweb_hq", "phase": 2, "loss": 3.2768383026123047, "ce_loss": 3.2598696053028107, "logit_z_loss": 13.299646139144897, "router_aux_loss": 1.4808436334133148, "router_z_loss": 0.8303246572613716, "routing_entropy_normalized": 0.6373815331697436, "learning_rate": 0.003, "grad_norm": 0.7958173155784607, "tokens_per_second": 4276958.751856988, "eta_seconds": 9169.305732250214, "data_wait_seconds": 1.5497207641601562e-05} {"step": 9920, "total_steps": 28610, "tokens": 20803747840, "source": "fineweb_edu", "phase": 2, "loss": 3.383318990468979, "ce_loss": 3.3663492798805237, "logit_z_loss": 13.563559293746948, "router_aux_loss": 1.464428573846817, "router_z_loss": 0.9690725207328796, "routing_entropy_normalized": 0.6462704505136718, "learning_rate": 0.003, "grad_norm": 0.5752901434898376, "tokens_per_second": 3929548.4543617885, "eta_seconds": 9974.6246509552, "data_wait_seconds": 2.086162567138672e-05} {"step": 9930, "total_steps": 28610, "tokens": 20824719360, "source": "dclm", "phase": 2, "loss": 3.4603759348392487, "ce_loss": 3.4438766539096832, "logit_z_loss": 13.277260184288025, "router_aux_loss": 1.4222336262464523, "router_z_loss": 0.9492489024996758, "routing_entropy_normalized": 0.6558740437340683, "learning_rate": 0.003, "grad_norm": 1.0010827779769897, "tokens_per_second": 4210502.115356677, "eta_seconds": 9304.06832408905, "data_wait_seconds": 2.2083520889282227e-05} {"step": 9940, "total_steps": 28610, "tokens": 20845690880, "source": "dclm", "phase": 2, "loss": 3.4637845158576965, "ce_loss": 3.4478750824928284, "logit_z_loss": 13.3677659034729, "router_aux_loss": 1.3616038113832474, "router_z_loss": 0.9565705433487892, "routing_entropy_normalized": 0.6756349418266137, "learning_rate": 0.003, "grad_norm": 0.9313654899597168, "tokens_per_second": 3196443.1269197627, "eta_seconds": 12249.186450481415, "data_wait_seconds": 1.9878149032592773e-05} {"step": 9950, "total_steps": 28610, "tokens": 20866662400, "source": "fineweb_hq", "phase": 2, "loss": 3.3789270520210266, "ce_loss": 3.363120913505554, "logit_z_loss": 13.926708698272705, "router_aux_loss": 1.331213042140007, "router_z_loss": 1.1013557836413383, "routing_entropy_normalized": 0.6622829456435112, "learning_rate": 0.003, "grad_norm": 1.2821714878082275, "tokens_per_second": 4238836.245602784, "eta_seconds": 9231.981150627136, "data_wait_seconds": 2.0951032638549805e-05} {"step": 9960, "total_steps": 28610, "tokens": 20887633920, "source": "fineweb_hq", "phase": 2, "loss": 3.3131926357746124, "ce_loss": 3.296981394290924, "logit_z_loss": 13.344191312789917, "router_aux_loss": 1.3778541684150696, "router_z_loss": 1.0983291864395142, "routing_entropy_normalized": 0.6529585296065115, "learning_rate": 0.003, "grad_norm": 0.5030850172042847, "tokens_per_second": 4185487.770333923, "eta_seconds": 9344.641997814178, "data_wait_seconds": 1.9818544387817383e-05} {"step": 9970, "total_steps": 28610, "tokens": 20908605440, "source": "cosmopedia_v2", "phase": 2, "loss": 2.81175634264946, "ce_loss": 2.795068621635437, "logit_z_loss": 15.159515738487244, "router_aux_loss": 1.3781986385583878, "router_z_loss": 1.3898614048957825, "routing_entropy_normalized": 0.6624657658367844, "learning_rate": 0.003, "grad_norm": 0.5205965638160706, "tokens_per_second": 3569127.0939927553, "eta_seconds": 10952.513668060303, "data_wait_seconds": 2.1696090698242188e-05} {"step": 9980, "total_steps": 28610, "tokens": 20929576960, "source": "dclm", "phase": 2, "loss": 3.424479991197586, "ce_loss": 3.4083935916423798, "logit_z_loss": 13.818141222000122, "router_aux_loss": 1.378305584192276, "router_z_loss": 0.9215913861989975, "routing_entropy_normalized": 0.6794466097399192, "learning_rate": 0.003, "grad_norm": 0.67690509557724, "tokens_per_second": 3969968.213152723, "eta_seconds": 9841.37395119667, "data_wait_seconds": 2.154707908630371e-05} {"step": 9990, "total_steps": 28610, "tokens": 20950548480, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8024436831474304, "ce_loss": 2.7868785560131073, "logit_z_loss": 15.44847559928894, "router_aux_loss": 1.2727856487035751, "router_z_loss": 1.2924236804246902, "routing_entropy_normalized": 0.6909752015737847, "learning_rate": 0.003, "grad_norm": 0.5935510396957397, "tokens_per_second": 3760827.666729546, "eta_seconds": 10383.078859329224, "data_wait_seconds": 2.014636993408203e-05} {"step": 10000, "total_steps": 28610, "tokens": 20971520000, "source": "fineweb_hq", "phase": 2, "loss": 3.3469714522361755, "ce_loss": 3.331327795982361, "logit_z_loss": 13.714009284973145, "router_aux_loss": 1.327162653207779, "router_z_loss": 1.0006463900208473, "routing_entropy_normalized": 0.6684452356886694, "learning_rate": 0.003, "grad_norm": 0.5281961560249329, "tokens_per_second": 3690661.329309621, "eta_seconds": 10574.798183202744, "data_wait_seconds": 1.677870750427246e-05} {"step": 10010, "total_steps": 28610, "tokens": 20992491520, "source": "fineweb_hq", "phase": 2, "loss": 3.3353289365768433, "ce_loss": 3.319284588098526, "logit_z_loss": 13.652840495109558, "router_aux_loss": 1.3707634210586548, "router_z_loss": 0.9713929519057274, "routing_entropy_normalized": 0.6580924557325862, "learning_rate": 0.003, "grad_norm": 1.4374712705612183, "tokens_per_second": 4231045.454593032, "eta_seconds": 9219.240875244139, "data_wait_seconds": 1.9490718841552734e-05} {"step": 10020, "total_steps": 28610, "tokens": 21013463040, "source": "fineweb_edu", "phase": 2, "loss": 3.380783200263977, "ce_loss": 3.36440709233284, "logit_z_loss": 13.566233038902283, "router_aux_loss": 1.400969311594963, "router_z_loss": 1.0098391473293304, "routing_entropy_normalized": 0.6544843340799259, "learning_rate": 0.003, "grad_norm": 0.8020463585853577, "tokens_per_second": 4587726.842007959, "eta_seconds": 8497.902561028799, "data_wait_seconds": 2.136826515197754e-05} {"step": 10030, "total_steps": 28610, "tokens": 21034434560, "source": "dclm", "phase": 2, "loss": 3.473909854888916, "ce_loss": 3.45771923661232, "logit_z_loss": 13.407860517501831, "router_aux_loss": 1.3946325331926346, "router_z_loss": 0.9034895598888397, "routing_entropy_normalized": 0.635152995824483, "learning_rate": 0.003, "grad_norm": 1.2222046852111816, "tokens_per_second": 4033607.0857052193, "eta_seconds": 9660.109012126923, "data_wait_seconds": 2.1517276763916016e-05} {"step": 10040, "total_steps": 28610, "tokens": 21055406080, "source": "dclm", "phase": 2, "loss": 3.4567956924438477, "ce_loss": 3.441151976585388, "logit_z_loss": 13.411064743995667, "router_aux_loss": 1.3362684100866318, "router_z_loss": 0.9399404376745224, "routing_entropy_normalized": 0.6548626447325938, "learning_rate": 0.003, "grad_norm": 0.498226135969162, "tokens_per_second": 3995172.8785170526, "eta_seconds": 9747.791603565216, "data_wait_seconds": 2.0802021026611328e-05} {"step": 10050, "total_steps": 28610, "tokens": 21076377600, "source": "fineweb_hq", "phase": 2, "loss": 3.2938907146453857, "ce_loss": 3.2780142724514008, "logit_z_loss": 13.140981078147888, "router_aux_loss": 1.3467680662870407, "router_z_loss": 1.0946797132492065, "routing_entropy_normalized": 0.6693602487292369, "learning_rate": 0.003, "grad_norm": 0.4535190463066101, "tokens_per_second": 3653804.0729382304, "eta_seconds": 10652.771835327148, "data_wait_seconds": 1.990795135498047e-05} {"step": 10060, "total_steps": 28610, "tokens": 21097349120, "source": "fineweb_hq", "phase": 2, "loss": 3.322517693042755, "ce_loss": 3.307536333799362, "logit_z_loss": 13.31455671787262, "router_aux_loss": 1.2670293301343918, "router_z_loss": 0.9796086847782135, "routing_entropy_normalized": 0.6828575152790984, "learning_rate": 0.003, "grad_norm": 0.7750973701477051, "tokens_per_second": 4376161.981459021, "eta_seconds": 8889.5634496212, "data_wait_seconds": 2.434849739074707e-05} {"step": 10070, "total_steps": 28610, "tokens": 21118320640, "source": "cosmopedia_v2", "phase": 2, "loss": 2.805355131626129, "ce_loss": 2.7899419367313385, "logit_z_loss": 14.773110747337341, "router_aux_loss": 1.2651716321706772, "router_z_loss": 1.2841775566339493, "routing_entropy_normalized": 0.7155556668209838, "learning_rate": 0.003, "grad_norm": 0.8795843720436096, "tokens_per_second": 4016266.997934265, "eta_seconds": 9680.929604530334, "data_wait_seconds": 1.9788742065429688e-05} {"step": 10080, "total_steps": 28610, "tokens": 21139292160, "source": "dclm", "phase": 2, "loss": 3.450583338737488, "ce_loss": 3.4353697299957275, "logit_z_loss": 13.470312118530273, "router_aux_loss": 1.308990553021431, "router_z_loss": 0.7766673490405083, "routing_entropy_normalized": 0.6823495691486232, "learning_rate": 0.003, "grad_norm": 1.0513335466384888, "tokens_per_second": 3229275.0076024467, "eta_seconds": 12033.730936050415, "data_wait_seconds": 0.1377507448196411} {"step": 10090, "total_steps": 28610, "tokens": 21160263680, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8432669937610626, "ce_loss": 2.8279260098934174, "logit_z_loss": 15.532361149787903, "router_aux_loss": 1.252157762646675, "router_z_loss": 1.2661515772342682, "routing_entropy_normalized": 0.6684596271805243, "learning_rate": 0.003, "grad_norm": 1.4846961498260498, "tokens_per_second": 3986925.155114043, "eta_seconds": 9741.656421661377, "data_wait_seconds": 2.536177635192871e-05} {"step": 10100, "total_steps": 28610, "tokens": 21181235200, "source": "fineweb_hq", "phase": 2, "loss": 3.291198968887329, "ce_loss": 3.2759999334812164, "logit_z_loss": 13.823068618774414, "router_aux_loss": 1.2836601585149765, "router_z_loss": 0.9801193401217461, "routing_entropy_normalized": 0.6642407734223771, "learning_rate": 0.003, "grad_norm": 0.642526388168335, "tokens_per_second": 4024367.305650072, "eta_seconds": 9645.810278177261, "data_wait_seconds": 1.6748905181884766e-05} {"step": 10110, "total_steps": 28610, "tokens": 21202206720, "source": "fineweb_hq", "phase": 2, "loss": 3.33038192987442, "ce_loss": 3.3148377537727356, "logit_z_loss": 13.480482459068298, "router_aux_loss": 1.3247229009866714, "router_z_loss": 0.9488599300384521, "routing_entropy_normalized": 0.6644169100651526, "learning_rate": 0.003, "grad_norm": 0.7985441088676453, "tokens_per_second": 3730489.4189230427, "eta_seconds": 10400.059521198273, "data_wait_seconds": 2.0503997802734375e-05} {"step": 10120, "total_steps": 28610, "tokens": 21223178240, "source": "fineweb_edu", "phase": 2, "loss": 3.3810522258281708, "ce_loss": 3.3652914464473724, "logit_z_loss": 13.656785607337952, "router_aux_loss": 1.3480952233076096, "router_z_loss": 0.914158284664154, "routing_entropy_normalized": 0.661341056406937, "learning_rate": 0.003, "grad_norm": 0.7341396808624268, "tokens_per_second": 3923485.4483576245, "eta_seconds": 9883.13605093956, "data_wait_seconds": 2.1219253540039062e-05} {"step": 10130, "total_steps": 28610, "tokens": 21244149760, "source": "dclm", "phase": 2, "loss": 3.494879901409149, "ce_loss": 3.479375571012497, "logit_z_loss": 13.559902310371399, "router_aux_loss": 1.3355563133955002, "router_z_loss": 0.7928489819169044, "routing_entropy_normalized": 0.6640923161038573, "learning_rate": 0.003, "grad_norm": 0.8658938407897949, "tokens_per_second": 3482215.89558055, "eta_seconds": 11129.513540267944, "data_wait_seconds": 2.1785497665405273e-05} {"step": 10140, "total_steps": 28610, "tokens": 21265121280, "source": "dclm", "phase": 2, "loss": 3.4396762251853943, "ce_loss": 3.424313187599182, "logit_z_loss": 13.277987480163574, "router_aux_loss": 1.3202274292707443, "router_z_loss": 0.8329417482018471, "routing_entropy_normalized": 0.6838001719875033, "learning_rate": 0.003, "grad_norm": 0.8560899496078491, "tokens_per_second": 3139875.2595324335, "eta_seconds": 12336.285437583923, "data_wait_seconds": 2.187490463256836e-05} {"step": 10150, "total_steps": 28610, "tokens": 21286092800, "source": "fineweb_hq", "phase": 2, "loss": 3.3384158611297607, "ce_loss": 3.32224577665329, "logit_z_loss": 13.256386280059814, "router_aux_loss": 1.393262967467308, "router_z_loss": 0.9117608517408371, "routing_entropy_normalized": 0.6696783644783099, "learning_rate": 0.003, "grad_norm": 0.6674582362174988, "tokens_per_second": 3894047.187600203, "eta_seconds": 9941.694092273712, "data_wait_seconds": 2.092123031616211e-05} {"step": 10160, "total_steps": 28610, "tokens": 21307064320, "source": "fineweb_hq", "phase": 2, "loss": 3.2971693575382233, "ce_loss": 3.2813329100608826, "logit_z_loss": 13.400960922241211, "router_aux_loss": 1.351994350552559, "router_z_loss": 0.9763707593083382, "routing_entropy_normalized": 0.6532462813069926, "learning_rate": 0.003, "grad_norm": 0.7688156366348267, "tokens_per_second": 4179311.3836320206, "eta_seconds": 9258.093223571777, "data_wait_seconds": 1.901388168334961e-05} {"step": 10170, "total_steps": 28610, "tokens": 21328035840, "source": "cosmopedia_v2", "phase": 2, "loss": 2.800618827342987, "ce_loss": 2.784957528114319, "logit_z_loss": 15.171051979064941, "router_aux_loss": 1.2837278991937637, "router_z_loss": 1.3068012446165085, "routing_entropy_normalized": 0.6770763942090182, "learning_rate": 0.003, "grad_norm": 0.8234806656837463, "tokens_per_second": 3321951.234531431, "eta_seconds": 11641.195234298706, "data_wait_seconds": 2.7298927307128906e-05} {"step": 10180, "total_steps": 28610, "tokens": 21349007360, "source": "dclm", "phase": 2, "loss": 3.5044377744197845, "ce_loss": 3.4896953105926514, "logit_z_loss": 13.372437477111816, "router_aux_loss": 1.2558299899101257, "router_z_loss": 0.846948616206646, "routing_entropy_normalized": 0.7053605856931017, "learning_rate": 0.003, "grad_norm": 0.8852882981300354, "tokens_per_second": 3913117.1818095716, "eta_seconds": 9877.166863203049, "data_wait_seconds": 0.08627283573150635} {"step": 10190, "total_steps": 28610, "tokens": 21369978880, "source": "cosmopedia_v2", "phase": 2, "loss": 2.835681825876236, "ce_loss": 2.820054918527603, "logit_z_loss": 15.179694652557373, "router_aux_loss": 1.2847531288862228, "router_z_loss": 1.2614135295152664, "routing_entropy_normalized": 0.6864378787100764, "learning_rate": 0.003, "grad_norm": 0.6983147859573364, "tokens_per_second": 3898574.4125243793, "eta_seconds": 9908.63216972351, "data_wait_seconds": 2.0712614059448242e-05} {"step": 10200, "total_steps": 28610, "tokens": 21390950400, "source": "fineweb_hq", "phase": 2, "loss": 3.3046588599681854, "ce_loss": 3.2891474068164825, "logit_z_loss": 13.04681646823883, "router_aux_loss": 1.3301447927951813, "router_z_loss": 0.9052936881780624, "routing_entropy_normalized": 0.6505307873135165, "learning_rate": 0.003, "grad_norm": 0.835647463798523, "tokens_per_second": 3507335.370089548, "eta_seconds": 11007.948840379715, "data_wait_seconds": 1.8805265426635742e-05} {"step": 10210, "total_steps": 28610, "tokens": 21411921920, "source": "fineweb_hq", "phase": 2, "loss": 3.2903094589710236, "ce_loss": 3.2743323147296906, "logit_z_loss": 13.38429343700409, "router_aux_loss": 1.3686147928237915, "router_z_loss": 0.9526110365986824, "routing_entropy_normalized": 0.6759370649225508, "learning_rate": 0.003, "grad_norm": 0.48486199975013733, "tokens_per_second": 4209147.142881616, "eta_seconds": 9167.557106018066, "data_wait_seconds": 1.913309097290039e-05} {"step": 10220, "total_steps": 28610, "tokens": 21432893440, "source": "fineweb_edu", "phase": 2, "loss": 3.385335296392441, "ce_loss": 3.368583232164383, "logit_z_loss": 13.31712520122528, "router_aux_loss": 1.4378604739904404, "router_z_loss": 1.0417101830244064, "routing_entropy_normalized": 0.6252328710675049, "learning_rate": 0.003, "grad_norm": 0.9663479924201965, "tokens_per_second": 3487196.9458123236, "eta_seconds": 11059.491585731506, "data_wait_seconds": 2.2113323211669922e-05} {"step": 10230, "total_steps": 28610, "tokens": 21453864960, "source": "dclm", "phase": 2, "loss": 3.450793534517288, "ce_loss": 3.4346306025981903, "logit_z_loss": 13.33049750328064, "router_aux_loss": 1.3915828615427017, "router_z_loss": 0.9140638187527657, "routing_entropy_normalized": 0.6553181090767156, "learning_rate": 0.003, "grad_norm": 0.8582713007926941, "tokens_per_second": 4058323.8184307776, "eta_seconds": 9497.924632072449, "data_wait_seconds": 2.065300941467285e-05} {"step": 10240, "total_steps": 28610, "tokens": 21474836480, "source": "dclm", "phase": 2, "loss": 3.481538772583008, "ce_loss": 3.4656234979629517, "logit_z_loss": 13.214950203895569, "router_aux_loss": 1.3668896555900574, "router_z_loss": 0.9248756244778633, "routing_entropy_normalized": 0.6669585935122331, "learning_rate": 0.003, "grad_norm": 0.7607486844062805, "tokens_per_second": 3263366.4801967023, "eta_seconds": 11805.19640493393, "data_wait_seconds": 1.9997358322143555e-05} {"step": 10250, "total_steps": 28610, "tokens": 21495808000, "source": "fineweb_hq", "phase": 2, "loss": 3.325470417737961, "ce_loss": 3.3089802861213684, "logit_z_loss": 13.743528127670288, "router_aux_loss": 1.4125231802463531, "router_z_loss": 0.9905968904495239, "routing_entropy_normalized": 0.6779155830122651, "learning_rate": 0.003, "grad_norm": 0.5676156282424927, "tokens_per_second": 4071116.516986526, "eta_seconds": 9457.776646614075, "data_wait_seconds": 1.7821788787841797e-05} {"step": 10260, "total_steps": 28610, "tokens": 21516779520, "source": "fineweb_hq", "phase": 2, "loss": 3.251069515943527, "ce_loss": 3.2341797947883606, "logit_z_loss": 12.95152235031128, "router_aux_loss": 1.4519823342561722, "router_z_loss": 1.074770525097847, "routing_entropy_normalized": 0.6459524264847235, "learning_rate": 0.003, "grad_norm": 0.761059045791626, "tokens_per_second": 4079811.8066204498, "eta_seconds": 9432.479002475739, "data_wait_seconds": 2.0235776901245117e-05} {"step": 10270, "total_steps": 28610, "tokens": 21537751040, "source": "cosmopedia_v2", "phase": 2, "loss": 2.877788722515106, "ce_loss": 2.8612604439258575, "logit_z_loss": 14.363191962242126, "router_aux_loss": 1.3605782240629196, "router_z_loss": 1.4861352145671844, "routing_entropy_normalized": 0.6605576901032945, "learning_rate": 0.003, "grad_norm": 1.5245715379714966, "tokens_per_second": 3676610.738360208, "eta_seconds": 10461.20201921463, "data_wait_seconds": 2.238154411315918e-05} {"step": 10280, "total_steps": 28610, "tokens": 21558722560, "source": "dclm", "phase": 2, "loss": 3.462177574634552, "ce_loss": 3.4465865194797516, "logit_z_loss": 13.348772406578064, "router_aux_loss": 1.313228726387024, "router_z_loss": 1.1238910108804703, "routing_entropy_normalized": 0.6803393384077403, "learning_rate": 0.003, "grad_norm": 1.0179016590118408, "tokens_per_second": 3764916.503590351, "eta_seconds": 10210.265253782272, "data_wait_seconds": 1.8537044525146484e-05} {"step": 10290, "total_steps": 28610, "tokens": 21579694080, "source": "cosmopedia_v2", "phase": 2, "loss": 2.831130474805832, "ce_loss": 2.8152973353862762, "logit_z_loss": 15.460982918739319, "router_aux_loss": 1.2721011340618134, "router_z_loss": 1.566107615828514, "routing_entropy_normalized": 0.694440683526023, "learning_rate": 0.003, "grad_norm": 1.2928951978683472, "tokens_per_second": 3685422.5241156733, "eta_seconds": 10424.808658599854, "data_wait_seconds": 2.1338462829589844e-05} {"step": 10300, "total_steps": 28610, "tokens": 21600665600, "source": "fineweb_hq", "phase": 2, "loss": 3.3199384212493896, "ce_loss": 3.304640829563141, "logit_z_loss": 13.319589734077454, "router_aux_loss": 1.2761891782283783, "router_z_loss": 1.2037515491247177, "routing_entropy_normalized": 0.6661354603081786, "learning_rate": 0.003, "grad_norm": 0.5352142453193665, "tokens_per_second": 3420437.169013889, "eta_seconds": 11226.299803972244, "data_wait_seconds": 2.422928810119629e-05} {"step": 10310, "total_steps": 28610, "tokens": 21621637120, "source": "fineweb_hq", "phase": 2, "loss": 3.2821807265281677, "ce_loss": 3.266359180212021, "logit_z_loss": 13.398344874382019, "router_aux_loss": 1.3374622911214828, "router_z_loss": 1.1071171313524246, "routing_entropy_normalized": 0.6555485897408306, "learning_rate": 0.003, "grad_norm": 0.6596779823303223, "tokens_per_second": 4406694.207817421, "eta_seconds": 8708.99585723877, "data_wait_seconds": 2.1845102310180664e-05} {"step": 10320, "total_steps": 28610, "tokens": 21642608640, "source": "fineweb_edu", "phase": 2, "loss": 3.375442147254944, "ce_loss": 3.3591559529304504, "logit_z_loss": 13.525689363479614, "router_aux_loss": 1.3812019526958466, "router_z_loss": 1.121631607413292, "routing_entropy_normalized": 0.6432320621683946, "learning_rate": 0.003, "grad_norm": 0.8155490756034851, "tokens_per_second": 3804880.318274267, "eta_seconds": 10080.976764440536, "data_wait_seconds": 2.110004425048828e-05} {"step": 10330, "total_steps": 28610, "tokens": 21663580160, "source": "dclm", "phase": 2, "loss": 3.628789007663727, "ce_loss": 3.6121639609336853, "logit_z_loss": 13.379715085029602, "router_aux_loss": 1.4346392899751663, "router_z_loss": 0.9406086504459381, "routing_entropy_normalized": 0.6356532194184799, "learning_rate": 0.003, "grad_norm": 1.1294509172439575, "tokens_per_second": 3940212.178006296, "eta_seconds": 9729.409693717957, "data_wait_seconds": 2.053380012512207e-05} {"step": 10340, "total_steps": 28610, "tokens": 21684551680, "source": "dclm", "phase": 2, "loss": 3.5428566336631775, "ce_loss": 3.5270224511623383, "logit_z_loss": 13.071748852729797, "router_aux_loss": 1.350992426276207, "router_z_loss": 1.0170983374118805, "routing_entropy_normalized": 0.6856779441161451, "learning_rate": 0.003, "grad_norm": 0.6551097631454468, "tokens_per_second": 3669774.2971693054, "eta_seconds": 10440.687611103058, "data_wait_seconds": 2.0176172256469727e-05} {"step": 10350, "total_steps": 28610, "tokens": 21705523200, "source": "fineweb_hq", "phase": 2, "loss": 3.3301755785942078, "ce_loss": 3.314685732126236, "logit_z_loss": 13.222172737121582, "router_aux_loss": 1.3062054514884949, "router_z_loss": 1.1055839955806732, "routing_entropy_normalized": 0.6635320669370459, "learning_rate": 0.003, "grad_norm": 0.7336491942405701, "tokens_per_second": 3989106.635142959, "eta_seconds": 9599.64198064804, "data_wait_seconds": 2.0444393157958984e-05} {"step": 10360, "total_steps": 28610, "tokens": 21726494720, "source": "fineweb_hq", "phase": 2, "loss": 3.3144820034503937, "ce_loss": 3.298615425825119, "logit_z_loss": 13.288843989372253, "router_aux_loss": 1.3518845736980438, "router_z_loss": 1.0187706798315048, "routing_entropy_normalized": 0.6346921964865317, "learning_rate": 0.003, "grad_norm": 0.6966503858566284, "tokens_per_second": 3889116.183910097, "eta_seconds": 9841.059559583664, "data_wait_seconds": 2.1845102310180664e-05} {"step": 10370, "total_steps": 28610, "tokens": 21747466240, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8065937161445618, "ce_loss": 2.7906152307987213, "logit_z_loss": 15.083443999290466, "router_aux_loss": 1.3111400753259659, "router_z_loss": 1.358809307217598, "routing_entropy_normalized": 0.6634072828004326, "learning_rate": 0.003, "grad_norm": 0.9211718440055847, "tokens_per_second": 3548394.458128057, "eta_seconds": 10780.101516723633, "data_wait_seconds": 2.5153160095214844e-05} {"step": 10380, "total_steps": 28610, "tokens": 21768437760, "source": "dclm", "phase": 2, "loss": 3.504428446292877, "ce_loss": 3.488940566778183, "logit_z_loss": 12.964489936828613, "router_aux_loss": 1.3251090496778488, "router_z_loss": 0.9402950927615166, "routing_entropy_normalized": 0.6765661272605605, "learning_rate": 0.003, "grad_norm": 0.9343442916870117, "tokens_per_second": 3189639.359605806, "eta_seconds": 11986.019938230515, "data_wait_seconds": 1.7076730728149414e-05} {"step": 10390, "total_steps": 28610, "tokens": 21789409280, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8042948246002197, "ce_loss": 2.7878846526145935, "logit_z_loss": 15.438741564750671, "router_aux_loss": 1.350356936454773, "router_z_loss": 1.3627219945192337, "routing_entropy_normalized": 0.6655984353311412, "learning_rate": 0.003, "grad_norm": 1.2232160568237305, "tokens_per_second": 4016157.705580035, "eta_seconds": 9514.095870018005, "data_wait_seconds": 1.9609928131103516e-05} {"step": 10400, "total_steps": 28610, "tokens": 21810380800, "source": "fineweb_hq", "phase": 2, "loss": 3.304677724838257, "ce_loss": 3.289962738752365, "logit_z_loss": 13.333094954490662, "router_aux_loss": 1.2334285080432892, "router_z_loss": 1.0474066883325577, "routing_entropy_normalized": 0.7007194992246789, "learning_rate": 0.003, "grad_norm": 0.6124287247657776, "tokens_per_second": 3713048.946259588, "eta_seconds": 10285.115675210953, "data_wait_seconds": 1.8864870071411133e-05} {"step": 10410, "total_steps": 28610, "tokens": 21831352320, "source": "fineweb_hq", "phase": 2, "loss": 3.2676232159137726, "ce_loss": 3.2532475888729095, "logit_z_loss": 13.230754494667053, "router_aux_loss": 1.2109334170818329, "router_z_loss": 0.943236380815506, "routing_entropy_normalized": 0.7100850857536071, "learning_rate": 0.003, "grad_norm": 1.197465419769287, "tokens_per_second": 4419717.712965986, "eta_seconds": 8635.883302688599, "data_wait_seconds": 2.086162567138672e-05} {"step": 10420, "total_steps": 28610, "tokens": 21852323840, "source": "fineweb_edu", "phase": 2, "loss": 3.350648671388626, "ce_loss": 3.3352659940719604, "logit_z_loss": 13.814818143844604, "router_aux_loss": 1.3007461428642273, "router_z_loss": 0.9937199354171753, "routing_entropy_normalized": 0.6255937112798462, "learning_rate": 0.003, "grad_norm": 0.8247462511062622, "tokens_per_second": 3936061.171278155, "eta_seconds": 9691.717994213104, "data_wait_seconds": 2.1278858184814453e-05} {"step": 10430, "total_steps": 28610, "tokens": 21873295360, "source": "dclm", "phase": 2, "loss": 3.5582295954227448, "ce_loss": 3.5425893664360046, "logit_z_loss": 12.898076057434082, "router_aux_loss": 1.3570543676614761, "router_z_loss": 0.7798600718379021, "routing_entropy_normalized": 0.6393899143795997, "learning_rate": 0.003, "grad_norm": 1.3561633825302124, "tokens_per_second": 3905087.9313450535, "eta_seconds": 9763.217635631561, "data_wait_seconds": 1.895427703857422e-05} {"step": 10440, "total_steps": 28610, "tokens": 21894266880, "source": "dclm", "phase": 2, "loss": 3.498923361301422, "ce_loss": 3.482163190841675, "logit_z_loss": 13.081972360610962, "router_aux_loss": 1.4614640027284622, "router_z_loss": 0.8373424783349037, "routing_entropy_normalized": 0.6411655510158265, "learning_rate": 0.003, "grad_norm": 1.2279369831085205, "tokens_per_second": 4209449.4931021705, "eta_seconds": 9052.312399148941, "data_wait_seconds": 2.053380012512207e-05} {"step": 10450, "total_steps": 28610, "tokens": 21915238400, "source": "fineweb_hq", "phase": 2, "loss": 3.245593100786209, "ce_loss": 3.2296564877033234, "logit_z_loss": 13.152416467666626, "router_aux_loss": 1.3628109693527222, "router_z_loss": 0.9932306483387947, "routing_entropy_normalized": 0.6803966483315276, "learning_rate": 0.003, "grad_norm": 0.42283904552459717, "tokens_per_second": 3765486.0814868086, "eta_seconds": 10114.040922164917, "data_wait_seconds": 2.1278858184814453e-05} {"step": 10460, "total_steps": 28610, "tokens": 21936209920, "source": "fineweb_hq", "phase": 2, "loss": 3.3054864406585693, "ce_loss": 3.2902159094810486, "logit_z_loss": 13.289172768592834, "router_aux_loss": 1.2905193716287613, "router_z_loss": 1.036429911851883, "routing_entropy_normalized": 0.6739062388960989, "learning_rate": 0.003, "grad_norm": 1.226835012435913, "tokens_per_second": 4066817.0711339167, "eta_seconds": 9359.48387503624, "data_wait_seconds": 2.5004148483276367e-05} {"step": 10470, "total_steps": 28610, "tokens": 21957181440, "source": "cosmopedia_v2", "phase": 2, "loss": 2.885531932115555, "ce_loss": 2.869931071996689, "logit_z_loss": 14.821396350860596, "router_aux_loss": 1.2784107029438019, "router_z_loss": 1.3345747292041779, "routing_entropy_normalized": 0.692565916410873, "learning_rate": 0.003, "grad_norm": 1.3484739065170288, "tokens_per_second": 3677151.910505189, "eta_seconds": 10345.598497390747, "data_wait_seconds": 2.5272369384765625e-05} {"step": 10480, "total_steps": 28610, "tokens": 21978152960, "source": "dclm", "phase": 2, "loss": 3.4856469929218292, "ce_loss": 3.46975040435791, "logit_z_loss": 13.41550326347351, "router_aux_loss": 1.3671012967824936, "router_z_loss": 0.8840602114796638, "routing_entropy_normalized": 0.6590166304376691, "learning_rate": 0.003, "grad_norm": 0.9404120445251465, "tokens_per_second": 3927833.3986661406, "eta_seconds": 9679.984332561493, "data_wait_seconds": 2.0623207092285156e-05} {"step": 10490, "total_steps": 28610, "tokens": 21999124480, "source": "cosmopedia_v2", "phase": 2, "loss": 2.815854460000992, "ce_loss": 2.8000476360321045, "logit_z_loss": 14.879361033439636, "router_aux_loss": 1.2972259372472763, "router_z_loss": 1.3466264456510544, "routing_entropy_normalized": 0.6493578101055283, "learning_rate": 0.003, "grad_norm": 0.9756044149398804, "tokens_per_second": 4029977.207324053, "eta_seconds": 9429.431553840637, "data_wait_seconds": 2.110004425048828e-05} {"step": 10500, "total_steps": 28610, "tokens": 22020096000, "source": "fineweb_hq", "phase": 2, "loss": 3.3150774538517, "ce_loss": 3.3000699877738953, "logit_z_loss": 13.383809447288513, "router_aux_loss": 1.275802806019783, "router_z_loss": 0.911119632422924, "routing_entropy_normalized": 0.6509750978388601, "learning_rate": 0.003, "grad_norm": 0.6147341132164001, "tokens_per_second": 4237609.962290199, "eta_seconds": 8962.463052988052, "data_wait_seconds": 2.4110078811645508e-05} {"step": 10510, "total_steps": 28610, "tokens": 22041067520, "source": "fineweb_hq", "phase": 2, "loss": 3.345844179391861, "ce_loss": 3.330826014280319, "logit_z_loss": 13.451944708824158, "router_aux_loss": 1.2755140215158463, "router_z_loss": 0.9178886115550995, "routing_entropy_normalized": 0.6919238426484394, "learning_rate": 0.003, "grad_norm": 1.24116849899292, "tokens_per_second": 3691606.9468156532, "eta_seconds": 10282.365307807922, "data_wait_seconds": 2.136826515197754e-05} {"step": 10520, "total_steps": 28610, "tokens": 22062039040, "source": "fineweb_edu", "phase": 2, "loss": 3.342964291572571, "ce_loss": 3.3272085785865784, "logit_z_loss": 13.388135552406311, "router_aux_loss": 1.347875401377678, "router_z_loss": 0.9381885528564453, "routing_entropy_normalized": 0.6369163490994509, "learning_rate": 0.003, "grad_norm": 0.7435712814331055, "tokens_per_second": 3912294.987395258, "eta_seconds": 9696.988545656204, "data_wait_seconds": 2.1159648895263672e-05} {"step": 10530, "total_steps": 28610, "tokens": 22083010560, "source": "dclm", "phase": 2, "loss": 3.4818924367427826, "ce_loss": 3.466597467660904, "logit_z_loss": 12.824115633964539, "router_aux_loss": 1.3172243535518646, "router_z_loss": 0.8403626531362534, "routing_entropy_normalized": 0.6681322641744991, "learning_rate": 0.003, "grad_norm": 0.7413841485977173, "tokens_per_second": 3362192.235554012, "eta_seconds": 11277.317150115967, "data_wait_seconds": 2.0802021026611328e-05} {"step": 10540, "total_steps": 28610, "tokens": 22103982080, "source": "dclm", "phase": 2, "loss": 3.5255478620529175, "ce_loss": 3.5100144147872925, "logit_z_loss": 13.27634072303772, "router_aux_loss": 1.3364457190036774, "router_z_loss": 0.8413940146565437, "routing_entropy_normalized": 0.66000658940893, "learning_rate": 0.003, "grad_norm": 1.2067970037460327, "tokens_per_second": 3792356.0526837017, "eta_seconds": 9992.610428333282, "data_wait_seconds": 0.1633044183254242} {"step": 10550, "total_steps": 28610, "tokens": 22124953600, "source": "fineweb_hq", "phase": 2, "loss": 3.305750012397766, "ce_loss": 3.2900440990924835, "logit_z_loss": 13.434788346290588, "router_aux_loss": 1.3510705381631851, "router_z_loss": 0.8516908064484596, "routing_entropy_normalized": 0.6755133919595759, "learning_rate": 0.003, "grad_norm": 0.8919505476951599, "tokens_per_second": 4007365.448049562, "eta_seconds": 9451.238129138947, "data_wait_seconds": 2.4378299713134766e-05} {"step": 10560, "total_steps": 28610, "tokens": 22145925120, "source": "fineweb_hq", "phase": 2, "loss": 3.3021847903728485, "ce_loss": 3.286805957555771, "logit_z_loss": 13.345312714576721, "router_aux_loss": 1.3129724115133286, "router_z_loss": 0.9146598055958748, "routing_entropy_normalized": 0.6786598849471206, "learning_rate": 0.003, "grad_norm": 1.1370488405227661, "tokens_per_second": 4082322.1643095943, "eta_seconds": 9272.564015388489, "data_wait_seconds": 2.014636993408203e-05} {"step": 10570, "total_steps": 28610, "tokens": 22166896640, "source": "cosmopedia_v2", "phase": 2, "loss": 2.7949943840503693, "ce_loss": 2.7792715430259705, "logit_z_loss": 15.854655146598816, "router_aux_loss": 1.2929576635360718, "router_z_loss": 1.2078032195568085, "routing_entropy_normalized": 0.6725495466316146, "learning_rate": 0.003, "grad_norm": 0.8449534773826599, "tokens_per_second": 3676561.870104631, "eta_seconds": 10290.217713356018, "data_wait_seconds": 2.104043960571289e-05} {"step": 10580, "total_steps": 28610, "tokens": 22187868160, "source": "dclm", "phase": 2, "loss": 3.497818797826767, "ce_loss": 3.4831325113773346, "logit_z_loss": 12.727412700653076, "router_aux_loss": 1.2589623034000397, "router_z_loss": 0.8239268288016319, "routing_entropy_normalized": 0.6781977599441162, "learning_rate": 0.003, "grad_norm": 0.8697859048843384, "tokens_per_second": 3519089.6481039585, "eta_seconds": 10744.725011587143, "data_wait_seconds": 2.1219253540039062e-05} {"step": 10590, "total_steps": 28610, "tokens": 22208839680, "source": "cosmopedia_v2", "phase": 2, "loss": 2.824610561132431, "ce_loss": 2.8091626465320587, "logit_z_loss": 15.400421619415283, "router_aux_loss": 1.2729524075984955, "router_z_loss": 1.1783644407987595, "routing_entropy_normalized": 0.6951903503892467, "learning_rate": 0.003, "grad_norm": 1.2763020992279053, "tokens_per_second": 4094904.6127793286, "eta_seconds": 9228.708019733429, "data_wait_seconds": 1.8358230590820312e-05} {"step": 10600, "total_steps": 28610, "tokens": 22229811200, "source": "fineweb_hq", "phase": 2, "loss": 3.3053672313690186, "ce_loss": 3.2907077968120575, "logit_z_loss": 13.467016339302063, "router_aux_loss": 1.2495738714933395, "router_z_loss": 0.8169950395822525, "routing_entropy_normalized": 0.6734086105688365, "learning_rate": 0.003, "grad_norm": 0.8784825801849365, "tokens_per_second": 2701531.976699502, "eta_seconds": 13980.847846984865, "data_wait_seconds": 0.2975080609321594} {"step": 10610, "total_steps": 28610, "tokens": 22250782720, "source": "fineweb_hq", "phase": 2, "loss": 3.2986271679401398, "ce_loss": 3.283227860927582, "logit_z_loss": 13.393295884132385, "router_aux_loss": 1.3290171772241592, "router_z_loss": 0.7697600498795509, "routing_entropy_normalized": 0.6431106554685018, "learning_rate": 0.003, "grad_norm": 0.5561913251876831, "tokens_per_second": 3225441.4610229908, "eta_seconds": 11703.432369232178, "data_wait_seconds": 2.244114875793457e-05} {"step": 10620, "total_steps": 28610, "tokens": 22271754240, "source": "fineweb_edu", "phase": 2, "loss": 3.360102415084839, "ce_loss": 3.3444995880126953, "logit_z_loss": 13.2504643201828, "router_aux_loss": 1.3467866778373718, "router_z_loss": 0.8099089711904526, "routing_entropy_normalized": 0.6444591199537038, "learning_rate": 0.003, "grad_norm": 0.7946572303771973, "tokens_per_second": 3780337.802104933, "eta_seconds": 9979.998205184937, "data_wait_seconds": 1.9311904907226562e-05} {"step": 10630, "total_steps": 28610, "tokens": 22292725760, "source": "dclm", "phase": 2, "loss": 3.5112131536006927, "ce_loss": 3.4953416883945465, "logit_z_loss": 13.346749305725098, "router_aux_loss": 1.3822028189897537, "router_z_loss": 0.7147052213549614, "routing_entropy_normalized": 0.6493953732753799, "learning_rate": 0.003, "grad_norm": 1.0988508462905884, "tokens_per_second": 3770259.3132619867, "eta_seconds": 10001.113935947418, "data_wait_seconds": 1.9282102584838867e-05} {"step": 10640, "total_steps": 28610, "tokens": 22313697280, "source": "dclm", "phase": 2, "loss": 3.4306600987911224, "ce_loss": 3.4148176312446594, "logit_z_loss": 13.100345134735107, "router_aux_loss": 1.3822027146816254, "router_z_loss": 0.7103796079754829, "routing_entropy_normalized": 0.6441656671675692, "learning_rate": 0.003, "grad_norm": 0.7670574188232422, "tokens_per_second": 4189815.209100815, "eta_seconds": 8994.626149177551, "data_wait_seconds": 1.84476375579834e-05} {"step": 10650, "total_steps": 28610, "tokens": 22334668800, "source": "fineweb_hq", "phase": 2, "loss": 3.3784002363681793, "ce_loss": 3.362569808959961, "logit_z_loss": 13.79174280166626, "router_aux_loss": 1.3633368164300919, "router_z_loss": 0.8179170116782188, "routing_entropy_normalized": 0.6682337731631198, "learning_rate": 0.003, "grad_norm": 1.170993447303772, "tokens_per_second": 3791900.0948387464, "eta_seconds": 9932.975283622742, "data_wait_seconds": 2.1964311599731445e-05} {"step": 10660, "total_steps": 28610, "tokens": 22355640320, "source": "fineweb_hq", "phase": 2, "loss": 3.3173844516277313, "ce_loss": 3.3022010028362274, "logit_z_loss": 13.355244040489197, "router_aux_loss": 1.2945170402526855, "router_z_loss": 0.9026973992586136, "routing_entropy_normalized": 0.7001306672920147, "learning_rate": 0.003, "grad_norm": 0.7170045971870422, "tokens_per_second": 3990839.9586692005, "eta_seconds": 9432.570283412933, "data_wait_seconds": 2.4378299713134766e-05} {"step": 10670, "total_steps": 28610, "tokens": 22376611840, "source": "cosmopedia_v2", "phase": 2, "loss": 2.810814380645752, "ce_loss": 2.7952334582805634, "logit_z_loss": 15.175431609153748, "router_aux_loss": 1.2833916246891022, "router_z_loss": 1.2294349819421768, "routing_entropy_normalized": 0.6742736938305391, "learning_rate": 0.003, "grad_norm": 0.8650232553482056, "tokens_per_second": 3431584.0900897556, "eta_seconds": 10963.714101791382, "data_wait_seconds": 2.1457672119140625e-05} {"step": 10680, "total_steps": 28610, "tokens": 22397583360, "source": "dclm", "phase": 2, "loss": 3.5036973357200623, "ce_loss": 3.4879773557186127, "logit_z_loss": 12.769697070121765, "router_aux_loss": 1.3687224239110947, "router_z_loss": 0.7557658925652504, "routing_entropy_normalized": 0.6507089344153465, "learning_rate": 0.003, "grad_norm": 0.9605175256729126, "tokens_per_second": 3789772.6518955277, "eta_seconds": 9921.950157403946, "data_wait_seconds": 2.086162567138672e-05} {"step": 10690, "total_steps": 28610, "tokens": 22418554880, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8059177100658417, "ce_loss": 2.7896315157413483, "logit_z_loss": 15.509055376052856, "router_aux_loss": 1.3532719612121582, "router_z_loss": 1.202508196234703, "routing_entropy_normalized": 0.701718695169653, "learning_rate": 0.003, "grad_norm": 0.937373161315918, "tokens_per_second": 3456246.613322, "eta_seconds": 10873.345581054688, "data_wait_seconds": 2.0831823348999023e-05} {"step": 10700, "total_steps": 28610, "tokens": 22439526400, "source": "fineweb_hq", "phase": 2, "loss": 3.349268615245819, "ce_loss": 3.333717256784439, "logit_z_loss": 13.651734232902527, "router_aux_loss": 1.3309370428323746, "router_z_loss": 0.8768756613135338, "routing_entropy_normalized": 0.6625280744266977, "learning_rate": 0.003, "grad_norm": 1.1600141525268555, "tokens_per_second": 3282511.0594031936, "eta_seconds": 11442.457204341888, "data_wait_seconds": 0.12044629454612732} {"step": 10710, "total_steps": 28610, "tokens": 22460497920, "source": "fineweb_hq", "phase": 2, "loss": 3.335274815559387, "ce_loss": 3.3200942277908325, "logit_z_loss": 13.291759967803955, "router_aux_loss": 1.2945160567760468, "router_z_loss": 0.9062626734375954, "routing_entropy_normalized": 0.684414590758412, "learning_rate": 0.003, "grad_norm": 1.0187911987304688, "tokens_per_second": 4296131.111205321, "eta_seconds": 8737.866659164429, "data_wait_seconds": 2.0116567611694336e-05} {"step": 10720, "total_steps": 28610, "tokens": 22481469440, "source": "fineweb_edu", "phase": 2, "loss": 3.372870832681656, "ce_loss": 3.357010543346405, "logit_z_loss": 13.06215751171112, "router_aux_loss": 1.354846104979515, "router_z_loss": 1.0056310519576073, "routing_entropy_normalized": 0.671138618830889, "learning_rate": 0.003, "grad_norm": 0.678593635559082, "tokens_per_second": 3554216.654936448, "eta_seconds": 10555.926360845566, "data_wait_seconds": 1.850724220275879e-05} {"step": 10730, "total_steps": 28610, "tokens": 22502440960, "source": "dclm", "phase": 2, "loss": 3.425019294023514, "ce_loss": 3.40865021944046, "logit_z_loss": 12.982638359069824, "router_aux_loss": 1.4154471307992935, "router_z_loss": 0.9163453653454781, "routing_entropy_normalized": 0.6652093915886733, "learning_rate": 0.003, "grad_norm": 0.6363615393638611, "tokens_per_second": 3527130.1739098183, "eta_seconds": 10631.044478416443, "data_wait_seconds": 2.238154411315918e-05} {"step": 10740, "total_steps": 28610, "tokens": 22523412480, "source": "dclm", "phase": 2, "loss": 3.5281692147254944, "ce_loss": 3.511459857225418, "logit_z_loss": 13.009440183639526, "router_aux_loss": 1.4476004838943481, "router_z_loss": 0.9324145689606667, "routing_entropy_normalized": 0.6498611060710847, "learning_rate": 0.003, "grad_norm": 0.7618995904922485, "tokens_per_second": 3487126.16357541, "eta_seconds": 10746.988919258118, "data_wait_seconds": 0.09802645444869995} {"step": 10750, "total_steps": 28610, "tokens": 22544384000, "source": "fineweb_hq", "phase": 2, "loss": 3.3184903264045715, "ce_loss": 3.3027056455612183, "logit_z_loss": 13.168056726455688, "router_aux_loss": 1.3396452516317368, "router_z_loss": 1.0714485794305801, "routing_entropy_normalized": 0.6680474900351164, "learning_rate": 0.003, "grad_norm": 0.7542944550514221, "tokens_per_second": 4283722.524373519, "eta_seconds": 8743.594970703125, "data_wait_seconds": 1.9997358322143555e-05} {"step": 10760, "total_steps": 28610, "tokens": 22565355520, "source": "fineweb_hq", "phase": 2, "loss": 3.3607148230075836, "ce_loss": 3.345608204603195, "logit_z_loss": 13.2822505235672, "router_aux_loss": 1.2755224853754044, "router_z_loss": 1.023156724870205, "routing_entropy_normalized": 0.650939394407121, "learning_rate": 0.003, "grad_norm": 0.8201872110366821, "tokens_per_second": 3827610.619413204, "eta_seconds": 9780.034314393997, "data_wait_seconds": 2.154707908630371e-05} {"step": 10770, "total_steps": 28610, "tokens": 22586327040, "source": "cosmopedia_v2", "phase": 2, "loss": 2.7809241116046906, "ce_loss": 2.7653132677078247, "logit_z_loss": 15.54421329498291, "router_aux_loss": 1.2725607752799988, "router_z_loss": 1.3307916671037674, "routing_entropy_normalized": 0.6835260650115303, "learning_rate": 0.003, "grad_norm": 0.853110134601593, "tokens_per_second": 3576586.3512256043, "eta_seconds": 10460.58671760559, "data_wait_seconds": 1.8417835235595703e-05} {"step": 10780, "total_steps": 28610, "tokens": 22607298560, "source": "dclm", "phase": 2, "loss": 3.473523497581482, "ce_loss": 3.4581612646579742, "logit_z_loss": 13.706328511238098, "router_aux_loss": 1.3178162723779678, "router_z_loss": 0.8133853822946548, "routing_entropy_normalized": 0.6583257498896649, "learning_rate": 0.003, "grad_norm": 0.9332784414291382, "tokens_per_second": 3500354.3781655966, "eta_seconds": 10682.409870624542, "data_wait_seconds": 1.817941665649414e-05} {"step": 10790, "total_steps": 28610, "tokens": 22628270080, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8139080107212067, "ce_loss": 2.7986804842948914, "logit_z_loss": 15.608810663223267, "router_aux_loss": 1.2342184484004974, "router_z_loss": 1.3244486153125763, "routing_entropy_normalized": 0.7108139268048537, "learning_rate": 0.003, "grad_norm": 1.2066079378128052, "tokens_per_second": 3436303.90989822, "eta_seconds": 10875.420108318329, "data_wait_seconds": 2.282857894897461e-05} {"step": 10800, "total_steps": 28610, "tokens": 22649241600, "source": "fineweb_hq", "phase": 2, "loss": 3.3245291113853455, "ce_loss": 3.3095309138298035, "logit_z_loss": 13.429301738739014, "router_aux_loss": 1.2753223329782486, "router_z_loss": 0.9020881056785583, "routing_entropy_normalized": 0.6773948005843244, "learning_rate": 0.003, "grad_norm": 1.034013032913208, "tokens_per_second": 4187541.3214019183, "eta_seconds": 8919.381148338318, "data_wait_seconds": 2.0802021026611328e-05} {"step": 10810, "total_steps": 28610, "tokens": 22670213120, "source": "fineweb_hq", "phase": 2, "loss": 3.330282896757126, "ce_loss": 3.315894991159439, "logit_z_loss": 13.740163445472717, "router_aux_loss": 1.2154744118452072, "router_z_loss": 0.8590588569641113, "routing_entropy_normalized": 0.701678433337574, "learning_rate": 0.003, "grad_norm": 1.0081828832626343, "tokens_per_second": 4099500.394849113, "eta_seconds": 9105.818271636963, "data_wait_seconds": 2.03549861907959e-05} {"step": 10820, "total_steps": 28610, "tokens": 22691184640, "source": "fineweb_edu", "phase": 2, "loss": 3.305918723344803, "ce_loss": 3.2911636531352997, "logit_z_loss": 13.236115455627441, "router_aux_loss": 1.257984220981598, "router_z_loss": 0.8515909612178802, "routing_entropy_normalized": 0.671642745874122, "learning_rate": 0.003, "grad_norm": 1.0981320142745972, "tokens_per_second": 3771922.792442066, "eta_seconds": 9891.065149784088, "data_wait_seconds": 1.868605613708496e-05} {"step": 10830, "total_steps": 28610, "tokens": 22712156160, "source": "dclm", "phase": 2, "loss": 3.48546040058136, "ce_loss": 3.4686531126499176, "logit_z_loss": 13.038045644760132, "router_aux_loss": 1.4805312752723694, "router_z_loss": 0.6981095895171165, "routing_entropy_normalized": 0.6440917770751897, "learning_rate": 0.003, "grad_norm": 1.0910100936889648, "tokens_per_second": 3479770.5552366986, "eta_seconds": 10715.465852737427, "data_wait_seconds": 2.3990869522094727e-05} {"step": 10840, "total_steps": 28610, "tokens": 22733127680, "source": "dclm", "phase": 2, "loss": 3.501728743314743, "ce_loss": 3.4845542311668396, "logit_z_loss": 12.742676377296448, "router_aux_loss": 1.510482132434845, "router_z_loss": 0.7954069375991821, "routing_entropy_normalized": 0.6407697331781916, "learning_rate": 0.003, "grad_norm": 0.8602781295776367, "tokens_per_second": 3662186.3738414077, "eta_seconds": 10175.995221376419, "data_wait_seconds": 2.041459083557129e-05} {"step": 10850, "total_steps": 28610, "tokens": 22754099200, "source": "fineweb_hq", "phase": 2, "loss": 3.3680106699466705, "ce_loss": 3.3517691493034363, "logit_z_loss": 12.986052989959717, "router_aux_loss": 1.3917080461978912, "router_z_loss": 1.0258420035243034, "routing_entropy_normalized": 0.6568778539933334, "learning_rate": 0.003, "grad_norm": 1.1306158304214478, "tokens_per_second": 4001422.1467129844, "eta_seconds": 9308.045528411865, "data_wait_seconds": 2.396106719970703e-05} {"step": 10860, "total_steps": 28610, "tokens": 22775070720, "source": "fineweb_hq", "phase": 2, "loss": 3.3744380474090576, "ce_loss": 3.3582745790481567, "logit_z_loss": 13.207205891609192, "router_aux_loss": 1.3824331909418106, "router_z_loss": 1.0184182673692703, "routing_entropy_normalized": 0.6336818597339272, "learning_rate": 0.003, "grad_norm": 0.6282899379730225, "tokens_per_second": 4208105.062493, "eta_seconds": 8845.893210172653, "data_wait_seconds": 2.1338462829589844e-05} {"step": 10870, "total_steps": 28610, "tokens": 22796042240, "source": "cosmopedia_v2", "phase": 2, "loss": 2.784477710723877, "ce_loss": 2.7693964540958405, "logit_z_loss": 14.97953999042511, "router_aux_loss": 1.2202457934617996, "router_z_loss": 1.3808553814888, "routing_entropy_normalized": 0.7105373085742605, "learning_rate": 0.003, "grad_norm": 0.694098949432373, "tokens_per_second": 3917774.035823113, "eta_seconds": 9496.075102806091, "data_wait_seconds": 2.1219253540039062e-05} {"step": 10880, "total_steps": 28610, "tokens": 22817013760, "source": "dclm", "phase": 2, "loss": 3.5015707910060883, "ce_loss": 3.4863827526569366, "logit_z_loss": 13.579639434814453, "router_aux_loss": 1.2969903349876404, "router_z_loss": 0.8601275682449341, "routing_entropy_normalized": 0.6506710040411801, "learning_rate": 0.003, "grad_norm": 1.2841774225234985, "tokens_per_second": 3549156.149117534, "eta_seconds": 10476.435354709625, "data_wait_seconds": 0.13190755248069763} {"step": 10890, "total_steps": 28610, "tokens": 22837985280, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8350900411605835, "ce_loss": 2.8193208277225494, "logit_z_loss": 15.202653765678406, "router_aux_loss": 1.2988187372684479, "router_z_loss": 1.2606429308652878, "routing_entropy_normalized": 0.6742918630274084, "learning_rate": 0.003, "grad_norm": 1.44830322265625, "tokens_per_second": 3970315.8485596, "eta_seconds": 9359.843110084534, "data_wait_seconds": 2.2292137145996094e-05} {"step": 10900, "total_steps": 28610, "tokens": 22858956800, "source": "fineweb_hq", "phase": 2, "loss": 3.3639290928840637, "ce_loss": 3.3490403294563293, "logit_z_loss": 13.537637829780579, "router_aux_loss": 1.2674483358860016, "router_z_loss": 0.8605594411492348, "routing_entropy_normalized": 0.7065736341916512, "learning_rate": 0.003, "grad_norm": 1.0652074813842773, "tokens_per_second": 4021400.896698911, "eta_seconds": 9235.727268695831, "data_wait_seconds": 1.7076730728149414e-05} {"step": 10910, "total_steps": 28610, "tokens": 22879928320, "source": "fineweb_hq", "phase": 2, "loss": 3.3718978464603424, "ce_loss": 3.357175827026367, "logit_z_loss": 13.446257710456848, "router_aux_loss": 1.253247544169426, "router_z_loss": 0.8449206501245499, "routing_entropy_normalized": 0.6544480709072292, "learning_rate": 0.003, "grad_norm": 0.5945382118225098, "tokens_per_second": 4096122.3562550307, "eta_seconds": 9062.129294872284, "data_wait_seconds": 1.963973045349121e-05} {"step": 10920, "total_steps": 28610, "tokens": 22900899840, "source": "fineweb_edu", "phase": 2, "loss": 3.3222124874591827, "ce_loss": 3.3074551820755005, "logit_z_loss": 13.422804236412048, "router_aux_loss": 1.258161336183548, "router_z_loss": 0.8334386125206947, "routing_entropy_normalized": 0.6491700448906905, "learning_rate": 0.003, "grad_norm": 0.8582184910774231, "tokens_per_second": 4263022.726530905, "eta_seconds": 8702.421089410782, "data_wait_seconds": 2.1725893020629883e-05} {"step": 10930, "total_steps": 28610, "tokens": 22921871360, "source": "dclm", "phase": 2, "loss": 3.530824840068817, "ce_loss": 3.51577165722847, "logit_z_loss": 12.97230839729309, "router_aux_loss": 1.3076179027557373, "router_z_loss": 0.6797018498182297, "routing_entropy_normalized": 0.6499308510641559, "learning_rate": 0.003, "grad_norm": 0.6426776051521301, "tokens_per_second": 3949748.47885461, "eta_seconds": 9387.34391784668, "data_wait_seconds": 1.952052116394043e-05} {"step": 10940, "total_steps": 28610, "tokens": 22942842880, "source": "dclm", "phase": 2, "loss": 3.473077207803726, "ce_loss": 3.456800878047943, "logit_z_loss": 13.132068276405334, "router_aux_loss": 1.4224510192871094, "router_z_loss": 0.7386433258652687, "routing_entropy_normalized": 0.64343102695566, "learning_rate": 0.003, "grad_norm": 1.0929617881774902, "tokens_per_second": 3990928.8644452877, "eta_seconds": 9285.225845575333, "data_wait_seconds": 2.2917985916137695e-05} {"step": 10950, "total_steps": 28610, "tokens": 22963814400, "source": "fineweb_hq", "phase": 2, "loss": 3.3850734531879425, "ce_loss": 3.3689227998256683, "logit_z_loss": 13.114925026893616, "router_aux_loss": 1.3958175480365753, "router_z_loss": 0.8810649886727333, "routing_entropy_normalized": 0.6698295724305087, "learning_rate": 0.003, "grad_norm": 0.7668280005455017, "tokens_per_second": 3954111.671729129, "eta_seconds": 9366.37793636322, "data_wait_seconds": 2.244114875793457e-05} {"step": 10960, "total_steps": 28610, "tokens": 22984785920, "source": "fineweb_hq", "phase": 2, "loss": 3.3270889818668365, "ce_loss": 3.3116573989391327, "logit_z_loss": 13.692133665084839, "router_aux_loss": 1.3149815797805786, "router_z_loss": 0.9125921949744225, "routing_entropy_normalized": 0.6696016013450807, "learning_rate": 0.003, "grad_norm": 1.1723766326904297, "tokens_per_second": 2627250.114584946, "eta_seconds": 14088.773883581161, "data_wait_seconds": 0.18174725770950317} {"step": 10970, "total_steps": 28610, "tokens": 23005757440, "source": "cosmopedia_v2", "phase": 2, "loss": 2.7945258021354675, "ce_loss": 2.7795678079128265, "logit_z_loss": 15.443645358085632, "router_aux_loss": 1.2148052006959915, "router_z_loss": 1.2656179666519165, "routing_entropy_normalized": 0.7038127224548927, "learning_rate": 0.003, "grad_norm": 0.8130762577056885, "tokens_per_second": 3950016.483918066, "eta_seconds": 9365.470101356506, "data_wait_seconds": 2.0563602447509766e-05} {"step": 10980, "total_steps": 28610, "tokens": 23026728960, "source": "dclm", "phase": 2, "loss": 3.4993486404418945, "ce_loss": 3.48395574092865, "logit_z_loss": 13.289806365966797, "router_aux_loss": 1.3345261812210083, "router_z_loss": 0.7186615914106369, "routing_entropy_normalized": 0.6605057660738978, "learning_rate": 0.003, "grad_norm": 1.021219253540039, "tokens_per_second": 3494733.6883790526, "eta_seconds": 10579.572882175446, "data_wait_seconds": 1.84476375579834e-05} {"step": 10990, "total_steps": 28610, "tokens": 23047700480, "source": "cosmopedia_v2", "phase": 2, "loss": 2.819594293832779, "ce_loss": 2.8039199113845825, "logit_z_loss": 15.069066643714905, "router_aux_loss": 1.297815203666687, "router_z_loss": 1.189238891005516, "routing_entropy_normalized": 0.6790465759510963, "learning_rate": 0.003, "grad_norm": 1.6575026512145996, "tokens_per_second": 4176087.264207033, "eta_seconds": 8848.4305768013, "data_wait_seconds": 2.205371856689453e-05} {"step": 11000, "total_steps": 28610, "tokens": 23068672000, "source": "fineweb_hq", "phase": 2, "loss": 3.11407533288002, "ce_loss": 3.098112106323242, "logit_z_loss": 13.870267868041992, "router_aux_loss": 1.3676637560129166, "router_z_loss": 0.8995839804410934, "routing_entropy_normalized": 0.6559177033823431, "learning_rate": 0.003, "grad_norm": 0.9385176301002502, "tokens_per_second": 3992383.2443172005, "eta_seconds": 9250.326048374176, "data_wait_seconds": 2.0325183868408203e-05} {"step": 11010, "total_steps": 28610, "tokens": 23089643520, "source": "fineweb_hq", "phase": 2, "loss": 3.097774386405945, "ce_loss": 3.0819084346294403, "logit_z_loss": 14.118456840515137, "router_aux_loss": 1.3527869880199432, "router_z_loss": 0.9262008965015411, "routing_entropy_normalized": 0.6548472791613285, "learning_rate": 0.003, "grad_norm": 0.7559935450553894, "tokens_per_second": 4128895.188244133, "eta_seconds": 8939.40715789795, "data_wait_seconds": 1.9073486328125e-05} {"step": 11020, "total_steps": 28610, "tokens": 23110615040, "source": "fineweb_edu", "phase": 2, "loss": 3.3073768615722656, "ce_loss": 3.2909001410007477, "logit_z_loss": 13.587355375289917, "router_aux_loss": 1.4160742610692978, "router_z_loss": 0.95732182264328, "routing_entropy_normalized": 0.6334739040110898, "learning_rate": 0.003, "grad_norm": 1.2203153371810913, "tokens_per_second": 4114601.6820048094, "eta_seconds": 8965.364458322525, "data_wait_seconds": 1.8835067749023438e-05} {"step": 11030, "total_steps": 28610, "tokens": 23131586560, "source": "dclm", "phase": 2, "loss": 3.4722266495227814, "ce_loss": 3.4556327164173126, "logit_z_loss": 13.125068187713623, "router_aux_loss": 1.4428454786539078, "router_z_loss": 0.8529755398631096, "routing_entropy_normalized": 0.6399010841034493, "learning_rate": 0.003, "grad_norm": 0.7189249396324158, "tokens_per_second": 4427876.896360938, "eta_seconds": 8326.322755336761, "data_wait_seconds": 2.205371856689453e-05} {"step": 11040, "total_steps": 28610, "tokens": 23152558080, "source": "dclm", "phase": 2, "loss": 3.7215150892734528, "ce_loss": 3.7033897638320923, "logit_z_loss": 12.66007912158966, "router_aux_loss": 1.5875811874866486, "router_z_loss": 0.9835618138313293, "routing_entropy_normalized": 0.631687293897211, "learning_rate": 0.003, "grad_norm": 2.338282585144043, "tokens_per_second": 3864660.384129823, "eta_seconds": 9534.33341550827, "data_wait_seconds": 0.05498814582824707} {"step": 11050, "total_steps": 28610, "tokens": 23173529600, "source": "fineweb_hq", "phase": 2, "loss": 3.181540787220001, "ce_loss": 3.164423018693924, "logit_z_loss": 13.803198337554932, "router_aux_loss": 1.4447263479232788, "router_z_loss": 1.2901863604784012, "routing_entropy_normalized": 0.658465631000807, "learning_rate": 0.003, "grad_norm": 0.5556302666664124, "tokens_per_second": 4305557.53796374, "eta_seconds": 8553.12901878357, "data_wait_seconds": 1.990795135498047e-05} {"step": 11060, "total_steps": 28610, "tokens": 23194501120, "source": "fineweb_hq", "phase": 2, "loss": 3.201324701309204, "ce_loss": 3.1855954229831696, "logit_z_loss": 13.25277042388916, "router_aux_loss": 1.3086052238941193, "router_z_loss": 1.3178896605968475, "routing_entropy_normalized": 0.6939533647259152, "learning_rate": 0.003, "grad_norm": 0.5570526719093323, "tokens_per_second": 4264261.283306069, "eta_seconds": 8631.041851043701, "data_wait_seconds": 2.0772218704223633e-05} {"step": 11070, "total_steps": 28610, "tokens": 23215472640, "source": "cosmopedia_v2", "phase": 2, "loss": 2.863125205039978, "ce_loss": 2.847603052854538, "logit_z_loss": 15.21349549293518, "router_aux_loss": 1.2393384277820587, "router_z_loss": 1.607447013258934, "routing_entropy_normalized": 0.7062956824465967, "learning_rate": 0.003, "grad_norm": 0.9721909165382385, "tokens_per_second": 3485401.59663782, "eta_seconds": 10553.746838092804, "data_wait_seconds": 2.104043960571289e-05} {"step": 11080, "total_steps": 28610, "tokens": 23236444160, "source": "dclm", "phase": 2, "loss": 3.5512865483760834, "ce_loss": 3.536473900079727, "logit_z_loss": 13.79026484489441, "router_aux_loss": 1.2420471906661987, "router_z_loss": 1.0131277441978455, "routing_entropy_normalized": 0.6949031833358341, "learning_rate": 0.003, "grad_norm": 1.0732578039169312, "tokens_per_second": 3489931.448809325, "eta_seconds": 10534.039163589478, "data_wait_seconds": 1.7642974853515625e-05} {"step": 11090, "total_steps": 28610, "tokens": 23257415680, "source": "cosmopedia_v2", "phase": 2, "loss": 2.884986311197281, "ce_loss": 2.869358718395233, "logit_z_loss": 14.80209219455719, "router_aux_loss": 1.272132471203804, "router_z_loss": 1.426075205206871, "routing_entropy_normalized": 0.640781728583572, "learning_rate": 0.003, "grad_norm": 1.2611788511276245, "tokens_per_second": 4489163.0536029795, "eta_seconds": 8184.6220779418945, "data_wait_seconds": 2.1725893020629883e-05} {"step": 11100, "total_steps": 28610, "tokens": 23278387200, "source": "fineweb_hq", "phase": 2, "loss": 3.1718032360076904, "ce_loss": 3.1568344235420227, "logit_z_loss": 13.904396295547485, "router_aux_loss": 1.2577344924211502, "router_z_loss": 1.001074641942978, "routing_entropy_normalized": 0.6621536907832879, "learning_rate": 0.003, "grad_norm": 0.4997687339782715, "tokens_per_second": 3150172.13209711, "eta_seconds": 11656.865079164505, "data_wait_seconds": 1.8894672393798828e-05} {"step": 11110, "total_steps": 28610, "tokens": 23299358720, "source": "fineweb_hq", "phase": 2, "loss": 3.103184849023819, "ce_loss": 3.0882925391197205, "logit_z_loss": 14.163614511489868, "router_aux_loss": 1.2487176954746246, "router_z_loss": 0.9888468831777573, "routing_entropy_normalized": 0.6937740183540898, "learning_rate": 0.003, "grad_norm": 0.5342984199523926, "tokens_per_second": 4354066.736112538, "eta_seconds": 8428.938329219818, "data_wait_seconds": 2.205371856689453e-05} {"step": 11120, "total_steps": 28610, "tokens": 23320330240, "source": "fineweb_edu", "phase": 2, "loss": 3.3597338795661926, "ce_loss": 3.3452394008636475, "logit_z_loss": 13.343843579292297, "router_aux_loss": 1.2162335216999054, "router_z_loss": 0.997734360396862, "routing_entropy_normalized": 0.7080274700391496, "learning_rate": 0.003, "grad_norm": 0.7296206951141357, "tokens_per_second": 4102434.1328461953, "eta_seconds": 8940.835438728333, "data_wait_seconds": 1.6510486602783203e-05} {"step": 11130, "total_steps": 28610, "tokens": 23341301760, "source": "dclm", "phase": 2, "loss": 3.5383683145046234, "ce_loss": 3.522472232580185, "logit_z_loss": 12.509222745895386, "router_aux_loss": 1.3900968432426453, "router_z_loss": 0.7441576570272446, "routing_entropy_normalized": 0.6465331007461271, "learning_rate": 0.003, "grad_norm": 1.3643856048583984, "tokens_per_second": 3916389.365044057, "eta_seconds": 9360.207462310791, "data_wait_seconds": 2.2292137145996094e-05} {"step": 11140, "total_steps": 28610, "tokens": 23362273280, "source": "dclm", "phase": 2, "loss": 3.5378544628620148, "ce_loss": 3.52093368768692, "logit_z_loss": 13.2970871925354, "router_aux_loss": 1.4783902019262314, "router_z_loss": 0.8070970922708511, "routing_entropy_normalized": 0.6525142221502184, "learning_rate": 0.003, "grad_norm": 1.0638076066970825, "tokens_per_second": 3954102.428783979, "eta_seconds": 9265.628824710846, "data_wait_seconds": 2.1755695343017578e-05} {"step": 11150, "total_steps": 28610, "tokens": 23383244800, "source": "fineweb_hq", "phase": 2, "loss": 3.168074071407318, "ce_loss": 3.150778979063034, "logit_z_loss": 13.968969464302063, "router_aux_loss": 1.4876455962657928, "router_z_loss": 1.021759733557701, "routing_entropy_normalized": 0.6443185606513928, "learning_rate": 0.003, "grad_norm": 0.7654473781585693, "tokens_per_second": 3952687.878964549, "eta_seconds": 9263.639083385468, "data_wait_seconds": 0.0377466082572937} {"step": 11160, "total_steps": 28610, "tokens": 23404216320, "source": "fineweb_hq", "phase": 2, "loss": 3.1121724247932434, "ce_loss": 3.0953626930713654, "logit_z_loss": 14.157214999198914, "router_aux_loss": 1.4304355084896088, "router_z_loss": 1.0896481424570084, "routing_entropy_normalized": 0.6541029969494746, "learning_rate": 0.003, "grad_norm": 1.0259641408920288, "tokens_per_second": 3773814.8850144153, "eta_seconds": 9697.164146900177, "data_wait_seconds": 2.422928810119629e-05} {"step": 11170, "total_steps": 28610, "tokens": 23425187840, "source": "cosmopedia_v2", "phase": 2, "loss": 2.794559419155121, "ce_loss": 2.77878075838089, "logit_z_loss": 16.272692918777466, "router_aux_loss": 1.2686977237462997, "router_z_loss": 1.4644002467393875, "routing_entropy_normalized": 0.6892786593015534, "learning_rate": 0.003, "grad_norm": 0.8293178677558899, "tokens_per_second": 4103156.5479980973, "eta_seconds": 8913.705936431885, "data_wait_seconds": 2.1189451217651367e-05} {"step": 11180, "total_steps": 28610, "tokens": 23446159360, "source": "dclm", "phase": 2, "loss": 3.563136398792267, "ce_loss": 3.5479745864868164, "logit_z_loss": 13.178062915802002, "router_aux_loss": 1.2844416797161102, "router_z_loss": 0.9996484816074371, "routing_entropy_normalized": 0.6672725685379585, "learning_rate": 0.003, "grad_norm": 1.0849136114120483, "tokens_per_second": 3563117.1299208705, "eta_seconds": 10258.814972162247, "data_wait_seconds": 2.288818359375e-05} {"step": 11190, "total_steps": 28610, "tokens": 23467130880, "source": "cosmopedia_v2", "phase": 2, "loss": 2.836304634809494, "ce_loss": 2.820298820734024, "logit_z_loss": 15.801830768585205, "router_aux_loss": 1.3010165691375732, "router_z_loss": 1.4155091792345047, "routing_entropy_normalized": 0.668579543236878, "learning_rate": 0.003, "grad_norm": 0.8170045018196106, "tokens_per_second": 4272001.740549352, "eta_seconds": 8551.585429668427, "data_wait_seconds": 2.059340476989746e-05} {"step": 11200, "total_steps": 28610, "tokens": 23488102400, "source": "fineweb_hq", "phase": 2, "loss": 3.1155329644680023, "ce_loss": 3.100473642349243, "logit_z_loss": 14.385278463363647, "router_aux_loss": 1.2652627974748611, "router_z_loss": 0.9681661427021027, "routing_entropy_normalized": 0.6941805483134323, "learning_rate": 0.003, "grad_norm": 0.6278638243675232, "tokens_per_second": 3905327.8889932497, "eta_seconds": 9349.129537343979, "data_wait_seconds": 2.288818359375e-05} {"step": 11210, "total_steps": 28610, "tokens": 23509073920, "source": "fineweb_hq", "phase": 2, "loss": 3.1071892380714417, "ce_loss": 3.092285245656967, "logit_z_loss": 13.959356904029846, "router_aux_loss": 1.2574879229068756, "router_z_loss": 0.9331338033080101, "routing_entropy_normalized": 0.6875773079865386, "learning_rate": 0.003, "grad_norm": 0.8738029599189758, "tokens_per_second": 3886797.2215827024, "eta_seconds": 9388.306803703308, "data_wait_seconds": 1.8924474716186523e-05} {"step": 11220, "total_steps": 28610, "tokens": 23530045440, "source": "fineweb_edu", "phase": 2, "loss": 3.3296627700328827, "ce_loss": 3.3145084381103516, "logit_z_loss": 13.499494910240173, "router_aux_loss": 1.2848174571990967, "router_z_loss": 0.9561876729130745, "routing_entropy_normalized": 0.6977668931063586, "learning_rate": 0.003, "grad_norm": 1.1942373514175415, "tokens_per_second": 3630456.5429157442, "eta_seconds": 10045.42344713211, "data_wait_seconds": 2.300739288330078e-05} {"step": 11230, "total_steps": 28610, "tokens": 23551016960, "source": "dclm", "phase": 2, "loss": 3.5387848913669586, "ce_loss": 3.523025691509247, "logit_z_loss": 13.040859580039978, "router_aux_loss": 1.3701320439577103, "router_z_loss": 0.7537989020347595, "routing_entropy_normalized": 0.66447870339553, "learning_rate": 0.003, "grad_norm": 1.5009441375732422, "tokens_per_second": 3944450.306441893, "eta_seconds": 9240.45150232315, "data_wait_seconds": 1.817941665649414e-05} {"step": 11240, "total_steps": 28610, "tokens": 23571988480, "source": "dclm", "phase": 2, "loss": 3.5158365964889526, "ce_loss": 3.500200569629669, "logit_z_loss": 13.299012541770935, "router_aux_loss": 1.3497034758329391, "router_z_loss": 0.8090957999229431, "routing_entropy_normalized": 0.6692970209185194, "learning_rate": 0.003, "grad_norm": 0.8963001370429993, "tokens_per_second": 4191165.1507483646, "eta_seconds": 8691.504374027252, "data_wait_seconds": 2.0444393157958984e-05} {"step": 11250, "total_steps": 28610, "tokens": 23592960000, "source": "fineweb_hq", "phase": 2, "loss": 3.1389836072921753, "ce_loss": 3.12372487783432, "logit_z_loss": 14.07653272151947, "router_aux_loss": 1.3011135160923004, "router_z_loss": 0.8400263488292694, "routing_entropy_normalized": 0.702051150729605, "learning_rate": 0.003, "grad_norm": 1.0897988080978394, "tokens_per_second": 3648590.1547129387, "eta_seconds": 9978.253839492798, "data_wait_seconds": 2.3573637008666992e-05} {"step": 11260, "total_steps": 28610, "tokens": 23613931520, "source": "fineweb_hq", "phase": 2, "loss": 3.1225203573703766, "ce_loss": 3.1063250303268433, "logit_z_loss": 14.298663258552551, "router_aux_loss": 1.3906984627246857, "router_z_loss": 0.8584907427430153, "routing_entropy_normalized": 0.681097773883301, "learning_rate": 0.003, "grad_norm": 1.3002026081085205, "tokens_per_second": 3443768.996254138, "eta_seconds": 10565.629471540451, "data_wait_seconds": 2.1189451217651367e-05} {"step": 11270, "total_steps": 28610, "tokens": 23634903040, "source": "cosmopedia_v2", "phase": 2, "loss": 2.8117007315158844, "ce_loss": 2.794804126024246, "logit_z_loss": 15.580537796020508, "router_aux_loss": 1.423109531402588, "router_z_loss": 1.10756453871727, "routing_entropy_normalized": 0.6474895703284398, "learning_rate": 0.003, "grad_norm": 1.1829454898834229, "tokens_per_second": 4040691.214666433, "eta_seconds": 8999.602728366852, "data_wait_seconds": 2.664327621459961e-05} {"step": 11280, "total_steps": 28610, "tokens": 23655874560, "source": "dclm", "phase": 2, "loss": 3.5619380474090576, "ce_loss": 3.5461055636405945, "logit_z_loss": 13.095267176628113, "router_aux_loss": 1.369046688079834, "router_z_loss": 0.8325087130069733, "routing_entropy_normalized": 0.671201698686225, "learning_rate": 0.003, "grad_norm": 0.8631498217582703, "tokens_per_second": 3993364.720688293, "eta_seconds": 9101.007972478867, "data_wait_seconds": 1.9371509552001953e-05} {"step": 11290, "total_steps": 28610, "tokens": 23676846080, "source": "cosmopedia_v2", "phase": 2, "loss": 2.855167657136917, "ce_loss": 2.838952600955963, "logit_z_loss": 15.82945203781128, "router_aux_loss": 1.332781195640564, "router_z_loss": 1.3042227774858475, "routing_entropy_normalized": 0.7021371446786262, "learning_rate": 0.003, "grad_norm": 1.1492867469787598, "tokens_per_second": 3137760.8047764804, "eta_seconds": 11575.985200881958, "data_wait_seconds": 2.3603439331054688e-05} {"step": 11300, "total_steps": 28610, "tokens": 23697817600, "source": "fineweb_hq", "phase": 2, "loss": 3.149577021598816, "ce_loss": 3.133823275566101, "logit_z_loss": 13.962284326553345, "router_aux_loss": 1.337843120098114, "router_z_loss": 0.979005478322506, "routing_entropy_normalized": 0.6867679336273443, "learning_rate": 0.003, "grad_norm": 0.9205769896507263, "tokens_per_second": 3804984.833085957, "eta_seconds": 9540.563947677612, "data_wait_seconds": 2.047419548034668e-05} {"step": 11310, "total_steps": 28610, "tokens": 23718789120, "source": "fineweb_hq", "phase": 2, "loss": 3.137228935956955, "ce_loss": 3.12137708067894, "logit_z_loss": 14.177772879600525, "router_aux_loss": 1.3454422354698181, "router_z_loss": 0.9796278923749924, "routing_entropy_normalized": 0.6825778846058399, "learning_rate": 0.003, "grad_norm": 1.4857909679412842, "tokens_per_second": 3918061.1050748643, "eta_seconds": 9259.868242740631, "data_wait_seconds": 1.582503318786621e-05} {"step": 11320, "total_steps": 28610, "tokens": 23739760640, "source": "fineweb_edu", "phase": 2, "loss": 3.401897817850113, "ce_loss": 3.386101543903351, "logit_z_loss": 13.526406645774841, "router_aux_loss": 1.3348422199487686, "router_z_loss": 1.095223143696785, "routing_entropy_normalized": 0.667766540977289, "learning_rate": 0.003, "grad_norm": 1.200196623802185, "tokens_per_second": 4069452.0909008645, "eta_seconds": 8910.23097705841, "data_wait_seconds": 1.749396324157715e-05} {"step": 11330, "total_steps": 28610, "tokens": 23760732160, "source": "dclm", "phase": 2, "loss": 3.503066658973694, "ce_loss": 3.486626833677292, "logit_z_loss": 13.313369989395142, "router_aux_loss": 1.424385204911232, "router_z_loss": 0.8646493330597878, "routing_entropy_normalized": 0.654468690923081, "learning_rate": 0.003, "grad_norm": 1.2409921884536743, "tokens_per_second": 4210527.10742671, "eta_seconds": 8606.710189819336, "data_wait_seconds": 2.2351741790771484e-05} {"step": 11340, "total_steps": 28610, "tokens": 23781703680, "source": "dclm", "phase": 2, "loss": 3.548803925514221, "ce_loss": 3.5329288840293884, "logit_z_loss": 13.265835881233215, "router_aux_loss": 1.3572497814893723, "router_z_loss": 0.9759671315550804, "routing_entropy_normalized": 0.6727863289155125, "learning_rate": 0.003, "grad_norm": 0.6185710430145264, "tokens_per_second": 3507996.0095603447, "eta_seconds": 10324.360387325287, "data_wait_seconds": 0.15977400541305542} {"step": 11350, "total_steps": 28610, "tokens": 23802675200, "source": "fineweb_hq", "phase": 2, "loss": 3.130015105009079, "ce_loss": 3.1146836280822754, "logit_z_loss": 13.982664227485657, "router_aux_loss": 1.278029903769493, "router_z_loss": 1.1529592722654343, "routing_entropy_normalized": 0.7080308965152129, "learning_rate": 0.003, "grad_norm": 0.6248959898948669, "tokens_per_second": 4407400.9993550815, "eta_seconds": 8212.741142749786, "data_wait_seconds": 1.958012580871582e-05} {"step": 11360, "total_steps": 28610, "tokens": 23823646720, "source": "fineweb_hq", "phase": 2, "loss": 3.098049521446228, "ce_loss": 3.082525610923767, "logit_z_loss": 14.562130212783813, "router_aux_loss": 1.3110616207122803, "router_z_loss": 0.9571214765310287, "routing_entropy_normalized": 0.6730166301771677, "learning_rate": 0.003, "grad_norm": 0.8422176837921143, "tokens_per_second": 4086321.0975456224, "eta_seconds": 8852.919566631317, "data_wait_seconds": 1.823902130126953e-05} {"step": 11370, "total_steps": 28610, "tokens": 23844618240, "source": "cosmopedia_v2", "phase": 2, "loss": 2.829041510820389, "ce_loss": 2.8132972419261932, "logit_z_loss": 15.000149130821228, "router_aux_loss": 1.2940962463617325, "router_z_loss": 1.303262636065483, "routing_entropy_normalized": 0.6840847824216988, "learning_rate": 0.003, "grad_norm": 1.184837818145752, "tokens_per_second": 4154055.9574171263, "eta_seconds": 8703.517923355103, "data_wait_seconds": 1.8656253814697266e-05} {"step": 11380, "total_steps": 28610, "tokens": 23865589760, "source": "dclm", "phase": 2, "loss": 3.54150453209877, "ce_loss": 3.526414215564728, "logit_z_loss": 12.88362991809845, "router_aux_loss": 1.3017967343330383, "router_z_loss": 0.783971518278122, "routing_entropy_normalized": 0.7042717578891505, "learning_rate": 0.003, "grad_norm": 1.006403923034668, "tokens_per_second": 3411061.2310458045, "eta_seconds": 10593.163391828537, "data_wait_seconds": 2.065300941467285e-05} {"step": 11390, "total_steps": 28610, "tokens": 23886561280, "source": "cosmopedia_v2", "phase": 2, "loss": 2.843228727579117, "ce_loss": 2.827780395746231, "logit_z_loss": 15.650962233543396, "router_aux_loss": 1.2543309032917023, "router_z_loss": 1.3399384915828705, "routing_entropy_normalized": 0.6776480519855196, "learning_rate": 0.003, "grad_norm": 2.2885847091674805, "tokens_per_second": 3646955.6336568156, "eta_seconds": 9902.220116615295, "data_wait_seconds": 2.142786979675293e-05} {"step": 11400, "total_steps": 28610, "tokens": 23907532800, "source": "fineweb_hq", "phase": 2, "loss": 3.24537655711174, "ce_loss": 3.2301248013973236, "logit_z_loss": 14.002239346504211, "router_aux_loss": 1.2986418008804321, "router_z_loss": 0.8651506155729294, "routing_entropy_normalized": 0.704751157811214, "learning_rate": 0.003, "grad_norm": 1.026125431060791, "tokens_per_second": 3188713.8639208637, "eta_seconds": 11318.665600061417, "data_wait_seconds": 2.0712614059448242e-05} {"step": 11410, "total_steps": 28610, "tokens": 23928504320, "source": "fineweb_hq", "phase": 2, "loss": 3.285265326499939, "ce_loss": 3.2700144350528717, "logit_z_loss": 13.36880362033844, "router_aux_loss": 1.3084567189216614, "router_z_loss": 0.8294704332947731, "routing_entropy_normalized": 0.6861814158821431, "learning_rate": 0.003, "grad_norm": 1.2372665405273438, "tokens_per_second": 3522158.1576472647, "eta_seconds": 10241.168279647827, "data_wait_seconds": 1.9162893295288086e-05} {"step": 11420, "total_steps": 28610, "tokens": 23949475840, "source": "fineweb_edu", "phase": 2, "loss": 3.4154191315174103, "ce_loss": 3.3989675045013428, "logit_z_loss": 13.416808605194092, "router_aux_loss": 1.4207357317209244, "router_z_loss": 0.9025197699666023, "routing_entropy_normalized": 0.6623063677222747, "learning_rate": 0.003, "grad_norm": 1.0080318450927734, "tokens_per_second": 4382129.08452383, "eta_seconds": 8226.604507684708, "data_wait_seconds": 1.7970800399780273e-05} {"step": 11430, "total_steps": 28610, "tokens": 23970447360, "source": "dclm", "phase": 2, "loss": 3.5154244899749756, "ce_loss": 3.498657315969467, "logit_z_loss": 12.752136945724487, "router_aux_loss": 1.4639092832803726, "router_z_loss": 0.8528083637356758, "routing_entropy_normalized": 0.6286713939629881, "learning_rate": 0.003, "grad_norm": 1.2736659049987793, "tokens_per_second": 4102184.2651314945, "eta_seconds": 8782.899312019348, "data_wait_seconds": 2.1517276763916016e-05} {"step": 11440, "total_steps": 28610, "tokens": 23991418880, "source": "dclm", "phase": 2, "loss": 3.513386368751526, "ce_loss": 3.4965564608573914, "logit_z_loss": 13.419936418533325, "router_aux_loss": 1.4545235484838486, "router_z_loss": 0.942734032869339, "routing_entropy_normalized": 0.6512304995874181, "learning_rate": 0.003, "grad_norm": 0.8435368537902832, "tokens_per_second": 3555089.1829342814, "eta_seconds": 10128.606621980667, "data_wait_seconds": 2.0742416381835938e-05} {"step": 11450, "total_steps": 28610, "tokens": 24012390400, "source": "fineweb_hq", "phase": 3, "loss": 3.306918680667877, "ce_loss": 3.292086511850357, "logit_z_loss": 13.540571093559265, "router_aux_loss": 1.3673078417778015, "router_z_loss": 1.1590532064437866, "routing_entropy_normalized": 0.6831881421576262, "learning_rate": 0.003, "grad_norm": 0.6725283265113831, "tokens_per_second": 3873442.33010956, "eta_seconds": 9290.735540390015, "data_wait_seconds": 2.0623207092285156e-05} {"step": 11460, "total_steps": 28610, "tokens": 24033361920, "source": "fineweb_edu", "phase": 3, "loss": 3.390033930540085, "ce_loss": 3.3755031526088715, "logit_z_loss": 13.674243450164795, "router_aux_loss": 1.3342869579792023, "router_z_loss": 1.1879382878541946, "routing_entropy_normalized": 0.6859464362304503, "learning_rate": 0.003, "grad_norm": 1.3138105869293213, "tokens_per_second": 3594016.7943768534, "eta_seconds": 10007.23114490509, "data_wait_seconds": 1.7434358596801758e-05} {"step": 11470, "total_steps": 28610, "tokens": 24054333440, "source": "dclm", "phase": 3, "loss": 3.527375251054764, "ce_loss": 3.5132122933864594, "logit_z_loss": 13.733974814414978, "router_aux_loss": 1.3138959258794785, "router_z_loss": 1.0239501520991325, "routing_entropy_normalized": 0.6836009588391271, "learning_rate": 0.003, "grad_norm": 1.0932527780532837, "tokens_per_second": 4145527.5067677516, "eta_seconds": 8670.835067749023, "data_wait_seconds": 2.3692846298217773e-05} {"step": 11480, "total_steps": 28610, "tokens": 24075304960, "source": "fineweb_edu", "phase": 3, "loss": 3.390751004219055, "ce_loss": 3.376556307077408, "logit_z_loss": 14.400393605232239, "router_aux_loss": 1.30422243475914, "router_z_loss": 1.152463749051094, "routing_entropy_normalized": 0.6935384308823068, "learning_rate": 0.003, "grad_norm": 0.9958397150039673, "tokens_per_second": 4016335.4003702058, "eta_seconds": 8944.525339365005, "data_wait_seconds": 2.0384788513183594e-05} {"step": 11490, "total_steps": 28610, "tokens": 24096276480, "source": "fineweb_hq", "phase": 3, "loss": 3.298944890499115, "ce_loss": 3.2848813235759735, "logit_z_loss": 14.593244075775146, "router_aux_loss": 1.3012738078832626, "router_z_loss": 1.05079997330904, "routing_entropy_normalized": 0.6738810221650056, "learning_rate": 0.003, "grad_norm": 0.7582775354385376, "tokens_per_second": 3361280.794250482, "eta_seconds": 10681.417125701904, "data_wait_seconds": 2.0891427993774414e-05} {"step": 11500, "total_steps": 28610, "tokens": 24117248000, "source": "fineweb_edu", "phase": 3, "loss": 3.4015552699565887, "ce_loss": 3.3877973556518555, "logit_z_loss": 14.6918363571167, "router_aux_loss": 1.269608348608017, "router_z_loss": 1.0618375316262245, "routing_entropy_normalized": 0.6885927523688116, "learning_rate": 0.003, "grad_norm": 1.5860711336135864, "tokens_per_second": 3322372.574311427, "eta_seconds": 10800.194715499878, "data_wait_seconds": 0.11981257796287537} {"step": 11510, "total_steps": 28610, "tokens": 24138219520, "source": "finemath", "phase": 3, "loss": 2.5305648744106293, "ce_loss": 2.516182154417038, "logit_z_loss": 16.011717438697815, "router_aux_loss": 1.3501919656991959, "router_z_loss": 0.8808044120669365, "routing_entropy_normalized": 0.6988689069059619, "learning_rate": 0.003, "grad_norm": 0.9278507232666016, "tokens_per_second": 4002459.7950467397, "eta_seconds": 8959.814972877502, "data_wait_seconds": 2.002716064453125e-05} {"step": 11520, "total_steps": 28610, "tokens": 24159191040, "source": "dclm", "phase": 3, "loss": 3.3552765548229218, "ce_loss": 3.3407222628593445, "logit_z_loss": 14.445507526397705, "router_aux_loss": 1.3696939945220947, "router_z_loss": 0.8573661670088768, "routing_entropy_normalized": 0.6771164682000668, "learning_rate": 0.003, "grad_norm": 1.1474777460098267, "tokens_per_second": 3195596.220781684, "eta_seconds": 11215.537009000778, "data_wait_seconds": 1.9818544387817383e-05} {"step": 11530, "total_steps": 28610, "tokens": 24180162560, "source": "fineweb_hq", "phase": 3, "loss": 3.314957618713379, "ce_loss": 3.301274597644806, "logit_z_loss": 15.235987544059753, "router_aux_loss": 1.271219402551651, "router_z_loss": 0.9707566350698471, "routing_entropy_normalized": 0.7109929166599676, "learning_rate": 0.003, "grad_norm": 0.7322150468826294, "tokens_per_second": 4444917.050168768, "eta_seconds": 8058.498225212096, "data_wait_seconds": 2.014636993408203e-05} {"step": 11540, "total_steps": 28610, "tokens": 24201134080, "source": "dclm", "phase": 3, "loss": 3.534923881292343, "ce_loss": 3.520724207162857, "logit_z_loss": 14.975170016288757, "router_aux_loss": 1.3395610600709915, "router_z_loss": 0.8040738478302956, "routing_entropy_normalized": 0.6739494157322503, "learning_rate": 0.003, "grad_norm": 0.8465616703033447, "tokens_per_second": 3967989.464191755, "eta_seconds": 9021.794277191162, "data_wait_seconds": 2.5153160095214844e-05} {"step": 11550, "total_steps": 28610, "tokens": 24222105600, "source": "fineweb_hq", "phase": 3, "loss": 3.3204510509967804, "ce_loss": 3.3060748875141144, "logit_z_loss": 15.584398031234741, "router_aux_loss": 1.347900539636612, "router_z_loss": 0.897123709321022, "routing_entropy_normalized": 0.6605018532857652, "learning_rate": 0.003, "grad_norm": 0.966975212097168, "tokens_per_second": 3885986.0491253952, "eta_seconds": 9206.778580188751, "data_wait_seconds": 2.199411392211914e-05} {"step": 11560, "total_steps": 28610, "tokens": 24243077120, "source": "fineweb_edu", "phase": 3, "loss": 3.4148122668266296, "ce_loss": 3.40122789144516, "logit_z_loss": 14.912932515144348, "router_aux_loss": 1.2615571022033691, "router_z_loss": 0.9687555879354477, "routing_entropy_normalized": 0.7016910799531434, "learning_rate": 0.003, "grad_norm": 0.8513789772987366, "tokens_per_second": 3847419.593327326, "eta_seconds": 9293.61634016037, "data_wait_seconds": 2.008676528930664e-05} {"step": 11570, "total_steps": 28610, "tokens": 24264048640, "source": "dclm", "phase": 3, "loss": 3.4759581089019775, "ce_loss": 3.4618417620658875, "logit_z_loss": 15.808707475662231, "router_aux_loss": 1.3313066363334656, "router_z_loss": 0.803289107978344, "routing_entropy_normalized": 0.6749771750606147, "learning_rate": 0.003, "grad_norm": 0.9964450001716614, "tokens_per_second": 3780685.0309657436, "eta_seconds": 9452.115102767944, "data_wait_seconds": 2.5659799575805664e-05} {"step": 11580, "total_steps": 28610, "tokens": 24285020160, "source": "fineweb_edu", "phase": 3, "loss": 3.3897502422332764, "ce_loss": 3.375545769929886, "logit_z_loss": 15.434450387954712, "router_aux_loss": 1.3275597095489502, "router_z_loss": 0.9288666173815727, "routing_entropy_normalized": 0.6969936828997755, "learning_rate": 0.003, "grad_norm": 0.6289944648742676, "tokens_per_second": 4118263.0450468883, "eta_seconds": 8672.223743200302, "data_wait_seconds": 2.586841583251953e-05} {"step": 11590, "total_steps": 28610, "tokens": 24305991680, "source": "fineweb_hq", "phase": 3, "loss": 3.3176930248737335, "ce_loss": 3.303944945335388, "logit_z_loss": 15.743976354598999, "router_aux_loss": 1.2863362729549408, "router_z_loss": 0.8847527429461479, "routing_entropy_normalized": 0.6924220725481911, "learning_rate": 0.003, "grad_norm": 0.9069772362709045, "tokens_per_second": 4174337.10662082, "eta_seconds": 8550.705448150635, "data_wait_seconds": 1.9162893295288086e-05} {"step": 11600, "total_steps": 28610, "tokens": 24326963200, "source": "fineweb_edu", "phase": 3, "loss": 3.4114023745059967, "ce_loss": 3.397480309009552, "logit_z_loss": 15.388771176338196, "router_aux_loss": 1.3057635724544525, "router_z_loss": 0.864460825920105, "routing_entropy_normalized": 0.6638985919130364, "learning_rate": 0.003, "grad_norm": 1.3225836753845215, "tokens_per_second": 3382960.5977362017, "eta_seconds": 10544.77416729927, "data_wait_seconds": 0.048029810190200806} {"step": 11610, "total_steps": 28610, "tokens": 24347934720, "source": "finemath", "phase": 3, "loss": 2.4253215193748474, "ce_loss": 2.4112683534622192, "logit_z_loss": 19.532552480697632, "router_aux_loss": 1.3411383777856827, "router_z_loss": 0.6418319940567017, "routing_entropy_normalized": 0.6935813105313221, "learning_rate": 0.003, "grad_norm": 1.1939746141433716, "tokens_per_second": 4031988.707529343, "eta_seconds": 8842.183494567871, "data_wait_seconds": 2.1576881408691406e-05} {"step": 11620, "total_steps": 28610, "tokens": 24368906240, "source": "dclm", "phase": 3, "loss": 3.47429883480072, "ce_loss": 3.460014760494232, "logit_z_loss": 15.551676511764526, "router_aux_loss": 1.3613461256027222, "router_z_loss": 0.6706692725419998, "routing_entropy_normalized": 0.6640726873658138, "learning_rate": 0.003, "grad_norm": 0.97803795337677, "tokens_per_second": 4228935.198555446, "eta_seconds": 8425.433544635773, "data_wait_seconds": 2.2202730178833008e-05} {"step": 11630, "total_steps": 28610, "tokens": 24389877760, "source": "fineweb_hq", "phase": 3, "loss": 3.313865751028061, "ce_loss": 3.2992069721221924, "logit_z_loss": 16.60313582420349, "router_aux_loss": 1.3835986703634262, "router_z_loss": 0.8227934315800667, "routing_entropy_normalized": 0.6542775446739063, "learning_rate": 0.003, "grad_norm": 1.608594298362732, "tokens_per_second": 3521569.856564806, "eta_seconds": 10111.865562915802, "data_wait_seconds": 2.79843807220459e-05} {"step": 11640, "total_steps": 28610, "tokens": 24410849280, "source": "dclm", "phase": 3, "loss": 3.4821740984916687, "ce_loss": 3.4676259756088257, "logit_z_loss": 15.969959735870361, "router_aux_loss": 1.3872026205062866, "router_z_loss": 0.6761193722486496, "routing_entropy_normalized": 0.6492636252007431, "learning_rate": 0.003, "grad_norm": 0.8897220492362976, "tokens_per_second": 4281792.421428285, "eta_seconds": 8311.628854751587, "data_wait_seconds": 1.7762184143066406e-05} {"step": 11650, "total_steps": 28610, "tokens": 24431820800, "source": "fineweb_hq", "phase": 3, "loss": 3.2786372900009155, "ce_loss": 3.2638866901397705, "logit_z_loss": 16.533260822296143, "router_aux_loss": 1.393289402127266, "router_z_loss": 0.817704975605011, "routing_entropy_normalized": 0.6678855577750661, "learning_rate": 0.003, "grad_norm": 0.9910042881965637, "tokens_per_second": 4000513.1218753713, "eta_seconds": 8890.783966064453, "data_wait_seconds": 0.028697669506072998} {"step": 11660, "total_steps": 28610, "tokens": 24452792320, "source": "fineweb_edu", "phase": 3, "loss": 3.4421074986457825, "ce_loss": 3.4277905225753784, "logit_z_loss": 16.27490234375, "router_aux_loss": 1.3393959701061249, "router_z_loss": 0.923031747341156, "routing_entropy_normalized": 0.693948676126452, "learning_rate": 0.003, "grad_norm": 0.8629681468009949, "tokens_per_second": 4438167.0978518715, "eta_seconds": 8009.325835704803, "data_wait_seconds": 1.7970800399780273e-05} {"step": 11670, "total_steps": 28610, "tokens": 24473763840, "source": "dclm", "phase": 3, "loss": 3.4844579100608826, "ce_loss": 3.4699153900146484, "logit_z_loss": 16.398677825927734, "router_aux_loss": 1.3811254799365997, "router_z_loss": 0.7312593460083008, "routing_entropy_normalized": 0.6533052747093041, "learning_rate": 0.003, "grad_norm": 0.5061686635017395, "tokens_per_second": 3320668.299779938, "eta_seconds": 10698.375047683716, "data_wait_seconds": 1.749396324157715e-05} {"step": 11680, "total_steps": 28610, "tokens": 24494735360, "source": "fineweb_edu", "phase": 3, "loss": 3.407941907644272, "ce_loss": 3.392210066318512, "logit_z_loss": 16.06526780128479, "router_aux_loss": 1.4757407158613205, "router_z_loss": 0.9744410589337349, "routing_entropy_normalized": 0.6612787039896778, "learning_rate": 0.003, "grad_norm": 1.396009087562561, "tokens_per_second": 4246260.941371999, "eta_seconds": 8361.422873020172, "data_wait_seconds": 1.7881393432617188e-05} {"step": 11690, "total_steps": 28610, "tokens": 24515706880, "source": "fineweb_hq", "phase": 3, "loss": 3.2978694438934326, "ce_loss": 3.282160520553589, "logit_z_loss": 16.1495178937912, "router_aux_loss": 1.468056857585907, "router_z_loss": 1.028370663523674, "routing_entropy_normalized": 0.6851221651738303, "learning_rate": 0.003, "grad_norm": 1.0772188901901245, "tokens_per_second": 3415776.324706259, "eta_seconds": 10388.212946891785, "data_wait_seconds": 1.9222497940063477e-05} {"step": 11700, "total_steps": 28610, "tokens": 24536678400, "source": "fineweb_edu", "phase": 3, "loss": 3.4323846101760864, "ce_loss": 3.417594760656357, "logit_z_loss": 16.586408853530884, "router_aux_loss": 1.3735282719135284, "router_z_loss": 1.0545754581689835, "routing_entropy_normalized": 0.6737927466164106, "learning_rate": 0.003, "grad_norm": 0.8372535109519958, "tokens_per_second": 4218837.2651614295, "eta_seconds": 8405.8327190876, "data_wait_seconds": 6.502866744995117e-05} {"step": 11710, "total_steps": 28610, "tokens": 24557649920, "source": "finemath", "phase": 3, "loss": 2.496492713689804, "ce_loss": 2.4820833802223206, "logit_z_loss": 21.452272653579712, "router_aux_loss": 1.341583251953125, "router_z_loss": 0.9935036152601242, "routing_entropy_normalized": 0.6881465129412057, "learning_rate": 0.003, "grad_norm": 0.9941592216491699, "tokens_per_second": 3514706.5587534765, "eta_seconds": 10083.877048492432, "data_wait_seconds": 2.0503997802734375e-05} {"step": 11720, "total_steps": 28610, "tokens": 24578621440, "source": "dclm", "phase": 3, "loss": 3.5200785100460052, "ce_loss": 3.5054520666599274, "logit_z_loss": 16.93837857246399, "router_aux_loss": 1.3628795892000198, "router_z_loss": 0.99758430570364, "routing_entropy_normalized": 0.6714832845882778, "learning_rate": 0.003, "grad_norm": 0.9612070322036743, "tokens_per_second": 3517465.1222220226, "eta_seconds": 10070.006680727005, "data_wait_seconds": 1.71661376953125e-05} {"step": 11730, "total_steps": 28610, "tokens": 24599592960, "source": "fineweb_hq", "phase": 3, "loss": 3.3238240480422974, "ce_loss": 3.309027314186096, "logit_z_loss": 17.467379093170166, "router_aux_loss": 1.374258130788803, "router_z_loss": 1.0541458576917648, "routing_entropy_normalized": 0.6690937174115303, "learning_rate": 0.003, "grad_norm": 1.333570957183838, "tokens_per_second": 4345701.450704897, "eta_seconds": 8145.963582992554, "data_wait_seconds": 2.047419548034668e-05} {"step": 11740, "total_steps": 28610, "tokens": 24620564480, "source": "dclm", "phase": 3, "loss": 3.4502379298210144, "ce_loss": 3.4358908236026764, "logit_z_loss": 16.006309866905212, "router_aux_loss": 1.3355025947093964, "router_z_loss": 0.992094598710537, "routing_entropy_normalized": 0.7109010734110381, "learning_rate": 0.003, "grad_norm": 1.0872178077697754, "tokens_per_second": 4042940.8646672494, "eta_seconds": 8750.796864032745, "data_wait_seconds": 1.8715858459472656e-05} {"step": 11750, "total_steps": 28610, "tokens": 24641536000, "source": "fineweb_hq", "phase": 3, "loss": 3.295786142349243, "ce_loss": 3.2805716395378113, "logit_z_loss": 16.630422949790955, "router_aux_loss": 1.4127238392829895, "router_z_loss": 1.087291158735752, "routing_entropy_normalized": 0.6339570296410579, "learning_rate": 0.003, "grad_norm": 1.621804118156433, "tokens_per_second": 3891614.2140670456, "eta_seconds": 9085.685470104218, "data_wait_seconds": 1.9431114196777344e-05} {"step": 11760, "total_steps": 28610, "tokens": 24662507520, "source": "fineweb_edu", "phase": 3, "loss": 3.4197716414928436, "ce_loss": 3.404864102602005, "logit_z_loss": 16.87557601928711, "router_aux_loss": 1.3774839639663696, "router_z_loss": 1.1327252686023712, "routing_entropy_normalized": 0.6772534903752602, "learning_rate": 0.003, "grad_norm": 1.154485821723938, "tokens_per_second": 3735227.100428561, "eta_seconds": 9460.471947193146, "data_wait_seconds": 2.0742416381835938e-05} {"step": 11770, "total_steps": 28610, "tokens": 24683479040, "source": "dclm", "phase": 3, "loss": 3.4768589437007904, "ce_loss": 3.4623197615146637, "logit_z_loss": 17.15220832824707, "router_aux_loss": 1.3583626747131348, "router_z_loss": 0.955596573650837, "routing_entropy_normalized": 0.6844929461346888, "learning_rate": 0.003, "grad_norm": 1.0407594442367554, "tokens_per_second": 4030806.2068846687, "eta_seconds": 8761.532524108887, "data_wait_seconds": 2.1845102310180664e-05} {"step": 11780, "total_steps": 28610, "tokens": 24704450560, "source": "fineweb_edu", "phase": 3, "loss": 3.4284797608852386, "ce_loss": 3.4134858548641205, "logit_z_loss": 16.22747266292572, "router_aux_loss": 1.3877281844615936, "router_z_loss": 1.116675078868866, "routing_entropy_normalized": 0.6739237976860321, "learning_rate": 0.003, "grad_norm": 0.9724554419517517, "tokens_per_second": 4144307.941913964, "eta_seconds": 8516.516787528992, "data_wait_seconds": 2.059340476989746e-05} {"step": 11790, "total_steps": 28610, "tokens": 24725422080, "source": "fineweb_hq", "phase": 3, "loss": 3.3020139038562775, "ce_loss": 3.287712812423706, "logit_z_loss": 16.338979721069336, "router_aux_loss": 1.3268958777189255, "router_z_loss": 1.0321143716573715, "routing_entropy_normalized": 0.677642447857071, "learning_rate": 0.003, "grad_norm": 1.2617292404174805, "tokens_per_second": 3297982.01786748, "eta_seconds": 10695.660694599152, "data_wait_seconds": 2.2649765014648438e-05} {"step": 11800, "total_steps": 28610, "tokens": 24746393600, "source": "fineweb_edu", "phase": 3, "loss": 3.428987354040146, "ce_loss": 3.41471791267395, "logit_z_loss": 16.860179901123047, "router_aux_loss": 1.3203626722097397, "router_z_loss": 1.0657602846622467, "routing_entropy_normalized": 0.6335217282010308, "learning_rate": 0.003, "grad_norm": 1.514946699142456, "tokens_per_second": 3770122.9242557483, "eta_seconds": 9350.656683683395, "data_wait_seconds": 0.05317261815071106} {"step": 11810, "total_steps": 28610, "tokens": 24767365120, "source": "finemath", "phase": 3, "loss": 2.4384146630764008, "ce_loss": 2.4239677786827087, "logit_z_loss": 22.400691032409668, "router_aux_loss": 1.361790344119072, "router_z_loss": 0.8290393054485321, "routing_entropy_normalized": 0.694958596827659, "learning_rate": 0.003, "grad_norm": 1.2169744968414307, "tokens_per_second": 3645092.207180275, "eta_seconds": 9665.641250610352, "data_wait_seconds": 2.0772218704223633e-05} {"step": 11820, "total_steps": 28610, "tokens": 24788336640, "source": "dclm", "phase": 3, "loss": 3.4608867168426514, "ce_loss": 3.446320354938507, "logit_z_loss": 17.82776165008545, "router_aux_loss": 1.3745679408311844, "router_z_loss": 0.8206884264945984, "routing_entropy_normalized": 0.6695641450157315, "learning_rate": 0.003, "grad_norm": 0.790709912776947, "tokens_per_second": 4078052.9136179145, "eta_seconds": 8634.312213659286, "data_wait_seconds": 2.065300941467285e-05} {"step": 11830, "total_steps": 28610, "tokens": 24809308160, "source": "fineweb_hq", "phase": 3, "loss": 3.287158876657486, "ce_loss": 3.2730909287929535, "logit_z_loss": 17.195138454437256, "router_aux_loss": 1.3117930442094803, "router_z_loss": 0.9499786719679832, "routing_entropy_normalized": 0.6865976442744774, "learning_rate": 0.003, "grad_norm": 0.6840500831604004, "tokens_per_second": 4048353.7615362597, "eta_seconds": 8692.474184036255, "data_wait_seconds": 2.0891427993774414e-05} {"step": 11840, "total_steps": 28610, "tokens": 24830279680, "source": "dclm", "phase": 3, "loss": 3.329001933336258, "ce_loss": 3.3148238360881805, "logit_z_loss": 17.01937508583069, "router_aux_loss": 1.3241853266954422, "router_z_loss": 0.9362340569496155, "routing_entropy_normalized": 0.7161120791351527, "learning_rate": 0.003, "grad_norm": 0.5994927883148193, "tokens_per_second": 3948511.09003565, "eta_seconds": 8906.962203741074, "data_wait_seconds": 3.102421760559082e-05} {"step": 11850, "total_steps": 28610, "tokens": 24851251200, "source": "fineweb_hq", "phase": 3, "loss": 3.2875013947486877, "ce_loss": 3.273443341255188, "logit_z_loss": 17.18602752685547, "router_aux_loss": 1.312809944152832, "router_z_loss": 0.9299374669790268, "routing_entropy_normalized": 0.7042261792422874, "learning_rate": 0.003, "grad_norm": 0.9082353115081787, "tokens_per_second": 4419873.170783279, "eta_seconds": 7952.324911117553, "data_wait_seconds": 2.1159648895263672e-05} {"step": 11860, "total_steps": 28610, "tokens": 24872222720, "source": "fineweb_edu", "phase": 3, "loss": 3.418739080429077, "ce_loss": 3.4045268893241882, "logit_z_loss": 17.501419067382812, "router_aux_loss": 1.324525162577629, "router_z_loss": 0.9669210761785507, "routing_entropy_normalized": 0.6548832831834487, "learning_rate": 0.003, "grad_norm": 0.8711398243904114, "tokens_per_second": 3896671.872603036, "eta_seconds": 9014.691805839539, "data_wait_seconds": 2.1070241928100586e-05} {"step": 11870, "total_steps": 28610, "tokens": 24893194240, "source": "dclm", "phase": 3, "loss": 3.461678206920624, "ce_loss": 3.4477325081825256, "logit_z_loss": 17.323718547821045, "router_aux_loss": 1.3191163539886475, "router_z_loss": 0.7545929700136185, "routing_entropy_normalized": 0.673395938547721, "learning_rate": 0.003, "grad_norm": 0.9699902534484863, "tokens_per_second": 3626937.7778965305, "eta_seconds": 9679.32912826538, "data_wait_seconds": 0.04879945516586304} {"step": 11880, "total_steps": 28610, "tokens": 24914165760, "source": "fineweb_edu", "phase": 3, "loss": 3.387195199728012, "ce_loss": 3.3722901046276093, "logit_z_loss": 16.763243913650513, "router_aux_loss": 1.4016170799732208, "router_z_loss": 0.888956680893898, "routing_entropy_normalized": 0.665732860831019, "learning_rate": 0.003, "grad_norm": 1.1788021326065063, "tokens_per_second": 4187075.8778607277, "eta_seconds": 8379.440445661545, "data_wait_seconds": 1.7583370208740234e-05} {"step": 11890, "total_steps": 28610, "tokens": 24935137280, "source": "fineweb_hq", "phase": 3, "loss": 3.2558790743350983, "ce_loss": 3.240784466266632, "logit_z_loss": 17.418206214904785, "router_aux_loss": 1.4167887568473816, "router_z_loss": 0.9267610013484955, "routing_entropy_normalized": 0.7016928970416343, "learning_rate": 0.003, "grad_norm": 1.693750023841858, "tokens_per_second": 3644782.727379972, "eta_seconds": 9620.431192398071, "data_wait_seconds": 1.8388032913208008e-05} {"step": 11900, "total_steps": 28610, "tokens": 24956108800, "source": "fineweb_edu", "phase": 3, "loss": 3.3843933045864105, "ce_loss": 3.3691340386867523, "logit_z_loss": 18.072491884231567, "router_aux_loss": 1.423858255147934, "router_z_loss": 1.0206826478242874, "routing_entropy_normalized": 0.6466993865723275, "learning_rate": 0.003, "grad_norm": 1.8390190601348877, "tokens_per_second": 3217895.7826212198, "eta_seconds": 10890.1631026268, "data_wait_seconds": 0.12079823017120361} {"step": 11910, "total_steps": 28610, "tokens": 24977080320, "source": "finemath", "phase": 3, "loss": 2.3764498233795166, "ce_loss": 2.3618343472480774, "logit_z_loss": 22.910937309265137, "router_aux_loss": 1.376981034874916, "router_z_loss": 0.845697432756424, "routing_entropy_normalized": 0.6855533930567383, "learning_rate": 0.003, "grad_norm": 1.0644891262054443, "tokens_per_second": 3900413.087427216, "eta_seconds": 8979.161338806152, "data_wait_seconds": 1.722574234008789e-05} {"step": 11920, "total_steps": 28610, "tokens": 24998051840, "source": "dclm", "phase": 3, "loss": 3.496634364128113, "ce_loss": 3.481727749109268, "logit_z_loss": 18.010980129241943, "router_aux_loss": 1.3978368937969208, "router_z_loss": 0.9282923564314842, "routing_entropy_normalized": 0.7024915350115974, "learning_rate": 0.003, "grad_norm": 0.9841338992118835, "tokens_per_second": 3223574.998562673, "eta_seconds": 10857.965735435486, "data_wait_seconds": 1.9043684005737305e-05} {"step": 11930, "total_steps": 28610, "tokens": 25019023360, "source": "fineweb_hq", "phase": 3, "loss": 3.3750170171260834, "ce_loss": 3.3602146208286285, "logit_z_loss": 17.51344871520996, "router_aux_loss": 1.3740858733654022, "router_z_loss": 1.0615416914224625, "routing_entropy_normalized": 0.6906958883710975, "learning_rate": 0.003, "grad_norm": 1.3596919775009155, "tokens_per_second": 3644809.3082572334, "eta_seconds": 9597.345814704895, "data_wait_seconds": 1.9937753677368164e-05} {"step": 11940, "total_steps": 28610, "tokens": 25039994880, "source": "dclm", "phase": 3, "loss": 3.5214923322200775, "ce_loss": 3.507261484861374, "logit_z_loss": 17.39887762069702, "router_aux_loss": 1.3317272365093231, "router_z_loss": 0.9135713279247284, "routing_entropy_normalized": 0.6791772829487096, "learning_rate": 0.003, "grad_norm": 1.1787084341049194, "tokens_per_second": 4356205.807751419, "eta_seconds": 8025.223183393478, "data_wait_seconds": 2.047419548034668e-05} {"step": 11950, "total_steps": 28610, "tokens": 25060966400, "source": "fineweb_hq", "phase": 3, "loss": 3.3770247995853424, "ce_loss": 3.3632964193820953, "logit_z_loss": 17.48794412612915, "router_aux_loss": 1.2688928991556168, "router_z_loss": 1.0394858345389366, "routing_entropy_normalized": 0.6931675449658059, "learning_rate": 0.003, "grad_norm": 0.6949124932289124, "tokens_per_second": 3903082.733699821, "eta_seconds": 8951.52747297287, "data_wait_seconds": 2.1189451217651367e-05} {"step": 11960, "total_steps": 28610, "tokens": 25081937920, "source": "fineweb_edu", "phase": 3, "loss": 3.4321162402629852, "ce_loss": 3.4175166487693787, "logit_z_loss": 17.759382963180542, "router_aux_loss": 1.3554843217134476, "router_z_loss": 1.0447454378008842, "routing_entropy_normalized": 0.6594108804123563, "learning_rate": 0.003, "grad_norm": 1.1201719045639038, "tokens_per_second": 4025932.029203901, "eta_seconds": 8673.166995048523, "data_wait_seconds": 1.9073486328125e-05} {"step": 11970, "total_steps": 28610, "tokens": 25102909440, "source": "dclm", "phase": 3, "loss": 3.5071871876716614, "ce_loss": 3.4922295808792114, "logit_z_loss": 17.88864231109619, "router_aux_loss": 1.407550573348999, "router_z_loss": 0.8821220174431801, "routing_entropy_normalized": 0.672767786834681, "learning_rate": 0.003, "grad_norm": 1.4080352783203125, "tokens_per_second": 3534513.7325247647, "eta_seconds": 9873.100494384766, "data_wait_seconds": 0.0620976984500885} {"step": 11980, "total_steps": 28610, "tokens": 25123880960, "source": "fineweb_edu", "phase": 3, "loss": 3.447011798620224, "ce_loss": 3.431776702404022, "logit_z_loss": 17.49042010307312, "router_aux_loss": 1.4149575382471085, "router_z_loss": 1.0854839086532593, "routing_entropy_normalized": 0.6572693709277376, "learning_rate": 0.003, "grad_norm": 1.4417662620544434, "tokens_per_second": 3306649.9810452317, "eta_seconds": 10547.121092319489, "data_wait_seconds": 1.659989356994629e-05} {"step": 11990, "total_steps": 28610, "tokens": 25144852480, "source": "fineweb_hq", "phase": 3, "loss": 3.429442346096039, "ce_loss": 3.4142973124980927, "logit_z_loss": 17.58024573326111, "router_aux_loss": 1.4047100841999054, "router_z_loss": 1.0979415997862816, "routing_entropy_normalized": 0.6454740762113343, "learning_rate": 0.003, "grad_norm": 1.4927846193313599, "tokens_per_second": 2706042.289864702, "eta_seconds": 12880.310987949371, "data_wait_seconds": 1.6421079635620117e-05} {"step": 12000, "total_steps": 28610, "tokens": 25165824000, "source": "fineweb_edu", "phase": 3, "loss": 3.4539094269275665, "ce_loss": 3.438760668039322, "logit_z_loss": 17.313875913619995, "router_aux_loss": 1.3972585052251816, "router_z_loss": 1.1761827766895294, "routing_entropy_normalized": 0.6490363660056744, "learning_rate": 0.003, "grad_norm": 1.766089677810669, "tokens_per_second": 4496072.583206621, "eta_seconds": 7747.582823753356, "data_wait_seconds": 2.130866050720215e-05} {"step": 12010, "total_steps": 28610, "tokens": 25186795520, "source": "finemath", "phase": 3, "loss": 2.451603263616562, "ce_loss": 2.4356290102005005, "logit_z_loss": 20.411766052246094, "router_aux_loss": 1.4978331178426743, "router_z_loss": 0.9959469512104988, "routing_entropy_normalized": 0.6549146119825404, "learning_rate": 0.003, "grad_norm": 0.7655073404312134, "tokens_per_second": 3931013.947402821, "eta_seconds": 8855.9144449234, "data_wait_seconds": 1.7076730728149414e-05} {"step": 12020, "total_steps": 28610, "tokens": 25207767040, "source": "dclm", "phase": 3, "loss": 3.4533508718013763, "ce_loss": 3.438093066215515, "logit_z_loss": 17.44118022918701, "router_aux_loss": 1.4117593318223953, "router_z_loss": 1.140273705124855, "routing_entropy_normalized": 0.7005901890663973, "learning_rate": 0.003, "grad_norm": 1.0942692756652832, "tokens_per_second": 4061053.02148568, "eta_seconds": 8567.174941062927, "data_wait_seconds": 1.9609928131103516e-05} {"step": 12030, "total_steps": 28610, "tokens": 25228738560, "source": "fineweb_hq", "phase": 3, "loss": 3.3625077307224274, "ce_loss": 3.3479546308517456, "logit_z_loss": 17.886613607406616, "router_aux_loss": 1.334889143705368, "router_z_loss": 1.2042509615421295, "routing_entropy_normalized": 0.7037900051336473, "learning_rate": 0.003, "grad_norm": 0.886101484298706, "tokens_per_second": 4511019.583345385, "eta_seconds": 7707.964799880981, "data_wait_seconds": 2.092123031616211e-05} {"step": 12040, "total_steps": 28610, "tokens": 25249710080, "source": "dclm", "phase": 3, "loss": 3.505414456129074, "ce_loss": 3.490928739309311, "logit_z_loss": 18.13517141342163, "router_aux_loss": 1.3380257934331894, "router_z_loss": 1.1054393500089645, "routing_entropy_normalized": 0.6852167642727501, "learning_rate": 0.003, "grad_norm": 1.2338863611221313, "tokens_per_second": 3780401.166207903, "eta_seconds": 9192.095524311066, "data_wait_seconds": 1.958012580871582e-05} {"step": 12050, "total_steps": 28610, "tokens": 25270681600, "source": "fineweb_hq", "phase": 3, "loss": 3.406281352043152, "ce_loss": 3.3909784257411957, "logit_z_loss": 17.82497787475586, "router_aux_loss": 1.4149307161569595, "router_z_loss": 1.153659924864769, "routing_entropy_normalized": 0.6585944375714781, "learning_rate": 0.003, "grad_norm": 0.9001200199127197, "tokens_per_second": 3820994.396849859, "eta_seconds": 9088.952642440796, "data_wait_seconds": 2.2917985916137695e-05} {"step": 12060, "total_steps": 28610, "tokens": 25291653120, "source": "fineweb_edu", "phase": 3, "loss": 3.441870927810669, "ce_loss": 3.426615446805954, "logit_z_loss": 18.131072759628296, "router_aux_loss": 1.3938795775175095, "router_z_loss": 1.3166569918394089, "routing_entropy_normalized": 0.6923543932131834, "learning_rate": 0.003, "grad_norm": 1.286488652229309, "tokens_per_second": 3313151.555459616, "eta_seconds": 10475.785673856735, "data_wait_seconds": 1.996755599975586e-05} {"step": 12070, "total_steps": 28610, "tokens": 25312624640, "source": "dclm", "phase": 3, "loss": 3.5080164670944214, "ce_loss": 3.4936842918395996, "logit_z_loss": 17.8262996673584, "router_aux_loss": 1.319749429821968, "router_z_loss": 1.1347189843654633, "routing_entropy_normalized": 0.7002128247105758, "learning_rate": 0.003, "grad_norm": 0.5883898735046387, "tokens_per_second": 3690994.7571341903, "eta_seconds": 9397.708846092224, "data_wait_seconds": 0.13062644004821777} {"step": 12080, "total_steps": 28610, "tokens": 25333596160, "source": "fineweb_edu", "phase": 3, "loss": 3.4061171412467957, "ce_loss": 3.3924016654491425, "logit_z_loss": 17.942237377166748, "router_aux_loss": 1.2483087331056595, "router_z_loss": 1.232395276427269, "routing_entropy_normalized": 0.7594749838460317, "learning_rate": 0.003, "grad_norm": 0.7372002601623535, "tokens_per_second": 3608874.595768266, "eta_seconds": 9605.743186712265, "data_wait_seconds": 1.7434358596801758e-05} {"step": 12090, "total_steps": 28610, "tokens": 25354567680, "source": "fineweb_hq", "phase": 3, "loss": 3.381466120481491, "ce_loss": 3.367728978395462, "logit_z_loss": 18.25735092163086, "router_aux_loss": 1.2585866451263428, "router_z_loss": 1.1512637585401535, "routing_entropy_normalized": 0.7184821261427399, "learning_rate": 0.003, "grad_norm": 1.031246542930603, "tokens_per_second": 3253803.4241176564, "eta_seconds": 10647.524304389954, "data_wait_seconds": 1.430511474609375e-05} {"step": 12100, "total_steps": 28610, "tokens": 25375539200, "source": "fineweb_edu", "phase": 3, "loss": 3.4245688021183014, "ce_loss": 3.4107615649700165, "logit_z_loss": 17.910093545913696, "router_aux_loss": 1.2706912457942963, "router_z_loss": 1.1003230810165405, "routing_entropy_normalized": 0.6943842363105274, "learning_rate": 0.003, "grad_norm": 1.0716062784194946, "tokens_per_second": 4424522.18819986, "eta_seconds": 7825.4731352329245, "data_wait_seconds": 1.7523765563964844e-05} {"step": 12110, "total_steps": 28610, "tokens": 25396510720, "source": "finemath", "phase": 3, "loss": 2.405600607395172, "ce_loss": 2.3920648097991943, "logit_z_loss": 21.413999319076538, "router_aux_loss": 1.2752771526575089, "router_z_loss": 0.7830387130379677, "routing_entropy_normalized": 0.7136349111817522, "learning_rate": 0.003, "grad_norm": 1.034242868423462, "tokens_per_second": 3717683.952280591, "eta_seconds": 9307.678771018982, "data_wait_seconds": 1.4185905456542969e-05} {"step": 12120, "total_steps": 28610, "tokens": 25417482240, "source": "dclm", "phase": 3, "loss": 3.478212296962738, "ce_loss": 3.4643921852111816, "logit_z_loss": 18.262729167938232, "router_aux_loss": 1.3022762686014175, "router_z_loss": 0.797386884689331, "routing_entropy_normalized": 0.7247460209971572, "learning_rate": 0.003, "grad_norm": 0.8455439805984497, "tokens_per_second": 3852733.332180504, "eta_seconds": 8975.974586963654, "data_wait_seconds": 2.2977590560913086e-05} {"step": 12130, "total_steps": 28610, "tokens": 25438453760, "source": "fineweb_hq", "phase": 3, "loss": 3.346653252840042, "ce_loss": 3.3327767848968506, "logit_z_loss": 17.721290826797485, "router_aux_loss": 1.3001902252435684, "router_z_loss": 0.8744261339306831, "routing_entropy_normalized": 0.6990743092581159, "learning_rate": 0.003, "grad_norm": 1.0509278774261475, "tokens_per_second": 3418258.179680963, "eta_seconds": 10110.723983764648, "data_wait_seconds": 1.856684684753418e-05} {"step": 12140, "total_steps": 28610, "tokens": 25459425280, "source": "dclm", "phase": 3, "loss": 3.4862958192825317, "ce_loss": 3.4715748727321625, "logit_z_loss": 18.54266381263733, "router_aux_loss": 1.4002865552902222, "router_z_loss": 0.7181182578206062, "routing_entropy_normalized": 0.6623909862966995, "learning_rate": 0.003, "grad_norm": 1.0069961547851562, "tokens_per_second": 4310453.357383278, "eta_seconds": 8013.099916934968, "data_wait_seconds": 2.0235776901245117e-05} {"step": 12150, "total_steps": 28610, "tokens": 25480396800, "source": "fineweb_hq", "phase": 3, "loss": 3.3700987100601196, "ce_loss": 3.3552566170692444, "logit_z_loss": 18.03801202774048, "router_aux_loss": 1.3888730257749557, "router_z_loss": 0.953387476503849, "routing_entropy_normalized": 0.6547000678980389, "learning_rate": 0.003, "grad_norm": 0.8359124064445496, "tokens_per_second": 3516218.334465419, "eta_seconds": 9817.115615844727, "data_wait_seconds": 0.07347625494003296} {"step": 12160, "total_steps": 28610, "tokens": 25501368320, "source": "fineweb_edu", "phase": 3, "loss": 3.395950049161911, "ce_loss": 3.381477653980255, "logit_z_loss": 18.3508198261261, "router_aux_loss": 1.3475802093744278, "router_z_loss": 0.9966100305318832, "routing_entropy_normalized": 0.6698585361776426, "learning_rate": 0.003, "grad_norm": 0.9279325008392334, "tokens_per_second": 3772614.71053187, "eta_seconds": 9144.36088681221, "data_wait_seconds": 2.205371856689453e-05} {"step": 12170, "total_steps": 28610, "tokens": 25522339840, "source": "dclm", "phase": 3, "loss": 3.503404051065445, "ce_loss": 3.4893656075000763, "logit_z_loss": 18.19524312019348, "router_aux_loss": 1.3187747597694397, "router_z_loss": 0.8507118225097656, "routing_entropy_normalized": 0.6914531496031074, "learning_rate": 0.003, "grad_norm": 0.762184202671051, "tokens_per_second": 3691318.331086854, "eta_seconds": 9340.071970939636, "data_wait_seconds": 1.958012580871582e-05} {"step": 12180, "total_steps": 28610, "tokens": 25543311360, "source": "fineweb_edu", "phase": 3, "loss": 3.417814701795578, "ce_loss": 3.4038816392421722, "logit_z_loss": 18.568169116973877, "router_aux_loss": 1.30629363656044, "router_z_loss": 0.8701336458325386, "routing_entropy_normalized": 0.6916739558941479, "learning_rate": 0.003, "grad_norm": 0.6392951011657715, "tokens_per_second": 3409646.70262935, "eta_seconds": 10105.506630182266, "data_wait_seconds": 1.913309097290039e-05} {"step": 12190, "total_steps": 28610, "tokens": 25564282880, "source": "fineweb_hq", "phase": 3, "loss": 3.4269572496414185, "ce_loss": 3.413412928581238, "logit_z_loss": 18.409230709075928, "router_aux_loss": 1.2693099081516266, "router_z_loss": 0.8512000516057014, "routing_entropy_normalized": 0.6989350272680261, "learning_rate": 0.003, "grad_norm": 1.2696826457977295, "tokens_per_second": 3306052.5575643023, "eta_seconds": 10415.816215991974, "data_wait_seconds": 3.695487976074219e-05} {"step": 12200, "total_steps": 28610, "tokens": 25585254400, "source": "fineweb_edu", "phase": 3, "loss": 3.4301901757717133, "ce_loss": 3.416219651699066, "logit_z_loss": 18.06279683113098, "router_aux_loss": 1.3091911971569061, "router_z_loss": 0.8786551058292389, "routing_entropy_normalized": 0.6936209135702782, "learning_rate": 0.003, "grad_norm": 1.5845017433166504, "tokens_per_second": 3174943.3859034292, "eta_seconds": 10839.33164691925, "data_wait_seconds": 0.048450350761413574} {"step": 12210, "total_steps": 28610, "tokens": 25606225920, "source": "finemath", "phase": 3, "loss": 2.553573101758957, "ce_loss": 2.5395253896713257, "logit_z_loss": 20.13487482070923, "router_aux_loss": 1.3417075425386429, "router_z_loss": 0.6305628940463066, "routing_entropy_normalized": 0.6964383454522103, "learning_rate": 0.003, "grad_norm": 1.1299060583114624, "tokens_per_second": 3654693.3882235107, "eta_seconds": 9410.719079971313, "data_wait_seconds": 2.2172927856445312e-05} {"step": 12220, "total_steps": 28610, "tokens": 25627197440, "source": "dclm", "phase": 3, "loss": 3.543105334043503, "ce_loss": 3.5291503369808197, "logit_z_loss": 18.58707904815674, "router_aux_loss": 1.3194691687822342, "router_z_loss": 0.760268934071064, "routing_entropy_normalized": 0.6624805932799661, "learning_rate": 0.003, "grad_norm": 1.0688343048095703, "tokens_per_second": 3444091.668490497, "eta_seconds": 9980.083165168762, "data_wait_seconds": 0.26791852712631226} {"step": 12230, "total_steps": 28610, "tokens": 25648168960, "source": "fineweb_hq", "phase": 3, "loss": 3.351126939058304, "ce_loss": 3.3370916843414307, "logit_z_loss": 17.749425172805786, "router_aux_loss": 1.3227261900901794, "router_z_loss": 0.8080097138881683, "routing_entropy_normalized": 0.7193669989851224, "learning_rate": 0.003, "grad_norm": 1.230589747428894, "tokens_per_second": 4334656.280919026, "eta_seconds": 7924.815148830415, "data_wait_seconds": 2.014636993408203e-05} {"step": 12240, "total_steps": 28610, "tokens": 25669140480, "source": "dclm", "phase": 3, "loss": 3.52877414226532, "ce_loss": 3.5139991641044617, "logit_z_loss": 18.61228370666504, "router_aux_loss": 1.4110462367534637, "router_z_loss": 0.6644473001360893, "routing_entropy_normalized": 0.6729083927659602, "learning_rate": 0.003, "grad_norm": 0.9247996211051941, "tokens_per_second": 3541155.8483438347, "eta_seconds": 9694.681541919708, "data_wait_seconds": 1.728534698486328e-05} {"step": 12250, "total_steps": 28610, "tokens": 25690112000, "source": "fineweb_hq", "phase": 3, "loss": 3.3997065126895905, "ce_loss": 3.3858373761177063, "logit_z_loss": 18.80548620223999, "router_aux_loss": 1.305817350745201, "router_z_loss": 0.810924656689167, "routing_entropy_normalized": 0.6821520192816686, "learning_rate": 0.003, "grad_norm": 0.7907158732414246, "tokens_per_second": 4135302.421447832, "eta_seconds": 8296.71042728424, "data_wait_seconds": 1.5497207641601562e-05} {"step": 12260, "total_steps": 28610, "tokens": 25711083520, "source": "fineweb_edu", "phase": 3, "loss": 3.4160116612911224, "ce_loss": 3.4020709097385406, "logit_z_loss": 18.606504440307617, "router_aux_loss": 1.3125369995832443, "router_z_loss": 0.815381608903408, "routing_entropy_normalized": 0.6874993212258372, "learning_rate": 0.003, "grad_norm": 0.6467156410217285, "tokens_per_second": 3796223.4286949593, "eta_seconds": 9032.248982191086, "data_wait_seconds": 2.276897430419922e-05} {"step": 12270, "total_steps": 28610, "tokens": 25732055040, "source": "dclm", "phase": 3, "loss": 3.465306878089905, "ce_loss": 3.4518761932849884, "logit_z_loss": 18.888718128204346, "router_aux_loss": 1.2707436084747314, "router_z_loss": 0.7232750505208969, "routing_entropy_normalized": 0.7010697381423793, "learning_rate": 0.003, "grad_norm": 0.7191538214683533, "tokens_per_second": 3594286.575864899, "eta_seconds": 9533.870757579803, "data_wait_seconds": 2.059340476989746e-05} {"step": 12280, "total_steps": 28610, "tokens": 25753026560, "source": "fineweb_edu", "phase": 3, "loss": 3.3836428225040436, "ce_loss": 3.368615746498108, "logit_z_loss": 18.424911975860596, "router_aux_loss": 1.4251295924186707, "router_z_loss": 0.7758216038346291, "routing_entropy_normalized": 0.6680814557623055, "learning_rate": 0.003, "grad_norm": 1.207810878753662, "tokens_per_second": 3838911.372976759, "eta_seconds": 8920.886374473572, "data_wait_seconds": 2.1636486053466797e-05} {"step": 12290, "total_steps": 28610, "tokens": 25773998080, "source": "fineweb_hq", "phase": 3, "loss": 3.3642085790634155, "ce_loss": 3.3498294055461884, "logit_z_loss": 17.994753122329712, "router_aux_loss": 1.354558676481247, "router_z_loss": 0.8335790857672691, "routing_entropy_normalized": 0.6837826568176366, "learning_rate": 0.003, "grad_norm": 1.3073582649230957, "tokens_per_second": 3927008.3402542626, "eta_seconds": 8715.41837310791, "data_wait_seconds": 1.9222497940063477e-05} {"step": 12300, "total_steps": 28610, "tokens": 25794969600, "source": "fineweb_edu", "phase": 3, "loss": 3.4187222719192505, "ce_loss": 3.404320240020752, "logit_z_loss": 18.9393572807312, "router_aux_loss": 1.3553501814603806, "router_z_loss": 0.8485326170921326, "routing_entropy_normalized": 0.676637989098221, "learning_rate": 0.003, "grad_norm": 1.4152835607528687, "tokens_per_second": 4199092.260093111, "eta_seconds": 8145.700785160065, "data_wait_seconds": 2.2739171981811523e-05} {"step": 12310, "total_steps": 28610, "tokens": 25815941120, "source": "finemath", "phase": 3, "loss": 2.639023184776306, "ce_loss": 2.6249775886535645, "logit_z_loss": 22.213249683380127, "router_aux_loss": 1.3303683549165726, "router_z_loss": 0.7419362142682076, "routing_entropy_normalized": 0.7224985272383968, "learning_rate": 0.003, "grad_norm": 0.9537715911865234, "tokens_per_second": 4099826.5619286094, "eta_seconds": 8337.810657024384, "data_wait_seconds": 1.9073486328125e-05} {"step": 12320, "total_steps": 28610, "tokens": 25836912640, "source": "dclm", "phase": 3, "loss": 3.4715413451194763, "ce_loss": 3.4558899998664856, "logit_z_loss": 18.414970636367798, "router_aux_loss": 1.4869070798158646, "router_z_loss": 0.7822763845324516, "routing_entropy_normalized": 0.6766331747991274, "learning_rate": 0.003, "grad_norm": 1.2705117464065552, "tokens_per_second": 3312564.629711274, "eta_seconds": 10313.038355112076, "data_wait_seconds": 1.862645149230957e-05} {"step": 12330, "total_steps": 28610, "tokens": 25857884160, "source": "fineweb_hq", "phase": 3, "loss": 3.382215827703476, "ce_loss": 3.3657186329364777, "logit_z_loss": 18.781113147735596, "router_aux_loss": 1.5511644631624222, "router_z_loss": 0.9855489060282707, "routing_entropy_normalized": 0.6429177928590184, "learning_rate": 0.003, "grad_norm": 1.2847812175750732, "tokens_per_second": 4214574.8972051805, "eta_seconds": 8100.848933219911, "data_wait_seconds": 2.2113323211669922e-05} {"step": 12340, "total_steps": 28610, "tokens": 25878855680, "source": "dclm", "phase": 3, "loss": 3.4997200071811676, "ce_loss": 3.4836878776550293, "logit_z_loss": 19.282937049865723, "router_aux_loss": 1.5023052990436554, "router_z_loss": 1.0091069787740707, "routing_entropy_normalized": 0.6557587050180017, "learning_rate": 0.003, "grad_norm": 1.566839575767517, "tokens_per_second": 4210633.3270343635, "eta_seconds": 8103.451521396638, "data_wait_seconds": 1.8894672393798828e-05} {"step": 12350, "total_steps": 28610, "tokens": 25899827200, "source": "fineweb_hq", "phase": 3, "loss": 3.3536850214004517, "ce_loss": 3.338268429040909, "logit_z_loss": 19.190712213516235, "router_aux_loss": 1.4277348518371582, "router_z_loss": 1.139204353094101, "routing_entropy_normalized": 0.6820380179328038, "learning_rate": 0.003, "grad_norm": 0.6367906332015991, "tokens_per_second": 4109730.235790709, "eta_seconds": 8297.306529521942, "data_wait_seconds": 2.0951032638549805e-05} {"step": 12360, "total_steps": 28610, "tokens": 25920798720, "source": "fineweb_edu", "phase": 3, "loss": 3.443574696779251, "ce_loss": 3.4291167557239532, "logit_z_loss": 18.19375967979431, "router_aux_loss": 1.3215662986040115, "router_z_loss": 1.2423475086688995, "routing_entropy_normalized": 0.6986623429787938, "learning_rate": 0.003, "grad_norm": 0.9677333831787109, "tokens_per_second": 4022548.8194008092, "eta_seconds": 8471.922039985657, "data_wait_seconds": 1.8894672393798828e-05} {"step": 12370, "total_steps": 28610, "tokens": 25941770240, "source": "dclm", "phase": 3, "loss": 3.4611367881298065, "ce_loss": 3.446457713842392, "logit_z_loss": 19.18197774887085, "router_aux_loss": 1.369694709777832, "router_z_loss": 0.9821362495422363, "routing_entropy_normalized": 0.6723808454069318, "learning_rate": 0.003, "grad_norm": 0.6886306405067444, "tokens_per_second": 3729556.9871671693, "eta_seconds": 9131.848258972168, "data_wait_seconds": 2.008676528930664e-05} {"step": 12380, "total_steps": 28610, "tokens": 25962741760, "source": "fineweb_edu", "phase": 3, "loss": 3.3852400481700897, "ce_loss": 3.370752602815628, "logit_z_loss": 18.611100912094116, "router_aux_loss": 1.3396042734384537, "router_z_loss": 1.091416247189045, "routing_entropy_normalized": 0.6971883397906578, "learning_rate": 0.003, "grad_norm": 1.0955801010131836, "tokens_per_second": 4393411.9008897515, "eta_seconds": 7747.231019496918, "data_wait_seconds": 1.9252300262451172e-05} {"step": 12390, "total_steps": 28610, "tokens": 25983713280, "source": "fineweb_hq", "phase": 3, "loss": 3.3640804290771484, "ce_loss": 3.350337505340576, "logit_z_loss": 19.074140548706055, "router_aux_loss": 1.272735670208931, "router_z_loss": 1.0155799835920334, "routing_entropy_normalized": 0.6790058488435694, "learning_rate": 0.003, "grad_norm": 1.5702266693115234, "tokens_per_second": 2995648.394088568, "eta_seconds": 11355.07274723053, "data_wait_seconds": 1.475214958190918e-05} {"step": 12400, "total_steps": 28610, "tokens": 26004684800, "source": "fineweb_edu", "phase": 3, "loss": 3.4316446781158447, "ce_loss": 3.4177973866462708, "logit_z_loss": 18.713979244232178, "router_aux_loss": 1.2791682183742523, "router_z_loss": 1.0555725619196892, "routing_entropy_normalized": 0.6904424313742341, "learning_rate": 0.003, "grad_norm": 0.9410420060157776, "tokens_per_second": 3861599.1435594005, "eta_seconds": 8803.30470776558, "data_wait_seconds": 0.03900492191314697} {"step": 12410, "total_steps": 28610, "tokens": 26025656320, "source": "finemath", "phase": 3, "loss": 2.4097957015037537, "ce_loss": 2.395477682352066, "logit_z_loss": 23.027327299118042, "router_aux_loss": 1.3583222776651382, "router_z_loss": 0.7347823828458786, "routing_entropy_normalized": 0.7154696547877466, "learning_rate": 0.003, "grad_norm": 0.9543620944023132, "tokens_per_second": 3670123.256259474, "eta_seconds": 9256.872324943542, "data_wait_seconds": 1.8924474716186523e-05} {"step": 12420, "total_steps": 28610, "tokens": 26046627840, "source": "dclm", "phase": 3, "loss": 3.5121087729930878, "ce_loss": 3.497376948595047, "logit_z_loss": 19.3257417678833, "router_aux_loss": 1.3863206952810287, "router_z_loss": 0.8686299324035645, "routing_entropy_normalized": 0.6832043864981542, "learning_rate": 0.003, "grad_norm": 0.5533167123794556, "tokens_per_second": 3876185.5697641685, "eta_seconds": 8759.356400489807, "data_wait_seconds": 1.7940998077392578e-05} {"step": 12430, "total_steps": 28610, "tokens": 26067599360, "source": "fineweb_hq", "phase": 3, "loss": 3.408278852701187, "ce_loss": 3.393408924341202, "logit_z_loss": 19.113648653030396, "router_aux_loss": 1.389322742819786, "router_z_loss": 0.976731039583683, "routing_entropy_normalized": 0.6780987806881226, "learning_rate": 0.003, "grad_norm": 0.6960426568984985, "tokens_per_second": 4003991.486741061, "eta_seconds": 8474.523353099823, "data_wait_seconds": 2.1249055862426758e-05} {"step": 12440, "total_steps": 28610, "tokens": 26088570880, "source": "dclm", "phase": 3, "loss": 3.474355638027191, "ce_loss": 3.4590805768966675, "logit_z_loss": 19.540307998657227, "router_aux_loss": 1.4219227731227875, "router_z_loss": 1.0558031648397446, "routing_entropy_normalized": 0.6707920407441532, "learning_rate": 0.003, "grad_norm": 1.3314599990844727, "tokens_per_second": 3700564.4867060855, "eta_seconds": 9163.722983837128, "data_wait_seconds": 1.9043684005737305e-05} {"step": 12450, "total_steps": 28610, "tokens": 26109542400, "source": "fineweb_hq", "phase": 3, "loss": 3.3566811978816986, "ce_loss": 3.341865211725235, "logit_z_loss": 19.22306799888611, "router_aux_loss": 1.3745304942131042, "router_z_loss": 1.0706733018159866, "routing_entropy_normalized": 0.696812878872136, "learning_rate": 0.003, "grad_norm": 0.8555535674095154, "tokens_per_second": 3883552.504429106, "eta_seconds": 8726.54001235962, "data_wait_seconds": 1.8894672393798828e-05} {"step": 12460, "total_steps": 28610, "tokens": 26130513920, "source": "fineweb_edu", "phase": 3, "loss": 3.3760819137096405, "ce_loss": 3.36181104183197, "logit_z_loss": 19.15458917617798, "router_aux_loss": 1.3177936375141144, "router_z_loss": 1.0930092632770538, "routing_entropy_normalized": 0.6901230672332251, "learning_rate": 0.003, "grad_norm": 1.1828978061676025, "tokens_per_second": 3682834.8309623487, "eta_seconds": 9196.449570655823, "data_wait_seconds": 1.7821788787841797e-05} {"step": 12470, "total_steps": 28610, "tokens": 26151485440, "source": "dclm", "phase": 3, "loss": 3.471337229013443, "ce_loss": 3.4577584862709045, "logit_z_loss": 19.651127099990845, "router_aux_loss": 1.2668301910161972, "router_z_loss": 0.9104696586728096, "routing_entropy_normalized": 0.6908782580834597, "learning_rate": 0.003, "grad_norm": 1.040952444076538, "tokens_per_second": 4155444.9842136335, "eta_seconds": 8145.465385437012, "data_wait_seconds": 2.059340476989746e-05} {"step": 12480, "total_steps": 28610, "tokens": 26172456960, "source": "fineweb_edu", "phase": 3, "loss": 3.367393046617508, "ce_loss": 3.353041559457779, "logit_z_loss": 18.79601788520813, "router_aux_loss": 1.3406237065792084, "router_z_loss": 0.9452317208051682, "routing_entropy_normalized": 0.6514164650782485, "learning_rate": 0.003, "grad_norm": 0.8121314644813538, "tokens_per_second": 3579337.6343017695, "eta_seconds": 9450.648476362228, "data_wait_seconds": 1.7911195755004883e-05} {"step": 12490, "total_steps": 28610, "tokens": 26193428480, "source": "fineweb_hq", "phase": 3, "loss": 3.354782223701477, "ce_loss": 3.3406100273132324, "logit_z_loss": 19.339195728302002, "router_aux_loss": 1.3204936981201172, "router_z_loss": 0.9672244265675545, "routing_entropy_normalized": 0.6886686028865509, "learning_rate": 0.003, "grad_norm": 1.5677032470703125, "tokens_per_second": 3475276.3616374796, "eta_seconds": 9727.597670555115, "data_wait_seconds": 2.1845102310180664e-05} {"step": 12500, "total_steps": 28610, "tokens": 26214400000, "source": "fineweb_edu", "phase": 3, "loss": 3.3871197402477264, "ce_loss": 3.372293770313263, "logit_z_loss": 20.09253978729248, "router_aux_loss": 1.3839469999074936, "router_z_loss": 0.9864333346486092, "routing_entropy_normalized": 0.6484902491675782, "learning_rate": 0.003, "grad_norm": 1.6451150178909302, "tokens_per_second": 4268150.88744089, "eta_seconds": 7915.633634090424, "data_wait_seconds": 0.00782206654548645} {"step": 12510, "total_steps": 28610, "tokens": 26235371520, "source": "finemath", "phase": 3, "loss": 2.382141262292862, "ce_loss": 2.3667496740818024, "logit_z_loss": 22.98542332649231, "router_aux_loss": 1.4568366706371307, "router_z_loss": 0.8231259137392044, "routing_entropy_normalized": 0.6762784208707655, "learning_rate": 0.003, "grad_norm": 0.984416127204895, "tokens_per_second": 4182717.495661, "eta_seconds": 8072.299225330353, "data_wait_seconds": 1.850724220275879e-05} {"step": 12520, "total_steps": 28610, "tokens": 26256343040, "source": "dclm", "phase": 3, "loss": 3.5538065135478973, "ce_loss": 3.536591798067093, "logit_z_loss": 19.098551750183105, "router_aux_loss": 1.6145724058151245, "router_z_loss": 1.0689626261591911, "routing_entropy_normalized": 0.6438879257339475, "learning_rate": 0.003, "grad_norm": 1.7990531921386719, "tokens_per_second": 3658662.9293218353, "eta_seconds": 9222.816184997559, "data_wait_seconds": 1.9758939743041992e-05} {"step": 12530, "total_steps": 28610, "tokens": 26277314560, "source": "fineweb_hq", "phase": 3, "loss": 3.3687549233436584, "ce_loss": 3.35320982336998, "logit_z_loss": 19.894185781478882, "router_aux_loss": 1.4308196604251862, "router_z_loss": 1.2369099855422974, "routing_entropy_normalized": 0.7015225934567686, "learning_rate": 0.003, "grad_norm": 1.4484635591506958, "tokens_per_second": 3738712.9010022585, "eta_seconds": 9019.73621749878, "data_wait_seconds": 2.384185791015625e-05} {"step": 12540, "total_steps": 28610, "tokens": 26298286080, "source": "dclm", "phase": 3, "loss": 3.501091241836548, "ce_loss": 3.4864400923252106, "logit_z_loss": 20.38869309425354, "router_aux_loss": 1.3436391204595566, "router_z_loss": 1.2147650569677353, "routing_entropy_normalized": 0.7016517785988323, "learning_rate": 0.003, "grad_norm": 1.1471655368804932, "tokens_per_second": 3121462.250563967, "eta_seconds": 10796.617077112198, "data_wait_seconds": 0.2212652862071991} {"step": 12550, "total_steps": 28610, "tokens": 26319257600, "source": "fineweb_hq", "phase": 3, "loss": 3.3105988800525665, "ce_loss": 3.2956092059612274, "logit_z_loss": 19.423665285110474, "router_aux_loss": 1.3710309118032455, "router_z_loss": 1.2793627977371216, "routing_entropy_normalized": 0.6948467943047228, "learning_rate": 0.003, "grad_norm": 0.9637599587440491, "tokens_per_second": 4289344.442959954, "eta_seconds": 7852.076597690582, "data_wait_seconds": 1.8328428268432617e-05} {"step": 12560, "total_steps": 28610, "tokens": 26340229120, "source": "fineweb_edu", "phase": 3, "loss": 3.3700583279132843, "ce_loss": 3.3556919395923615, "logit_z_loss": 19.362722158432007, "router_aux_loss": 1.3111481666564941, "router_z_loss": 1.2549203634262085, "routing_entropy_normalized": 0.6979727365416283, "learning_rate": 0.003, "grad_norm": 1.0845845937728882, "tokens_per_second": 4056523.348239992, "eta_seconds": 8297.570779323578, "data_wait_seconds": 2.16066837310791e-05} {"step": 12570, "total_steps": 28610, "tokens": 26361200640, "source": "dclm", "phase": 3, "loss": 3.457535147666931, "ce_loss": 3.443529039621353, "logit_z_loss": 20.096982955932617, "router_aux_loss": 1.2998967468738556, "router_z_loss": 1.0071509256958961, "routing_entropy_normalized": 0.6975519927749814, "learning_rate": 0.003, "grad_norm": 0.7603570818901062, "tokens_per_second": 3633590.9380819686, "eta_seconds": 9257.596315383911, "data_wait_seconds": 0.06060495972633362} {"step": 12580, "total_steps": 28610, "tokens": 26382172160, "source": "fineweb_edu", "phase": 3, "loss": 3.3696882724761963, "ce_loss": 3.355233281850815, "logit_z_loss": 18.996944904327393, "router_aux_loss": 1.3293247818946838, "router_z_loss": 1.1617783159017563, "routing_entropy_normalized": 0.6977985646982183, "learning_rate": 0.003, "grad_norm": 1.0073037147521973, "tokens_per_second": 3764119.158448398, "eta_seconds": 8930.999563217163, "data_wait_seconds": 2.148747444152832e-05} {"step": 12590, "total_steps": 28610, "tokens": 26403143680, "source": "fineweb_hq", "phase": 3, "loss": 3.3578484654426575, "ce_loss": 3.344120979309082, "logit_z_loss": 19.1810462474823, "router_aux_loss": 1.267783284187317, "router_z_loss": 1.0495730191469193, "routing_entropy_normalized": 0.7052357712027278, "learning_rate": 0.003, "grad_norm": 1.3906278610229492, "tokens_per_second": 4286237.226983927, "eta_seconds": 7838.197762012481, "data_wait_seconds": 2.0772218704223633e-05} {"step": 12600, "total_steps": 28610, "tokens": 26424115200, "source": "fineweb_edu", "phase": 3, "loss": 3.408215582370758, "ce_loss": 3.3942260146141052, "logit_z_loss": 19.49567723274231, "router_aux_loss": 1.2945951074361801, "router_z_loss": 1.0435819029808044, "routing_entropy_normalized": 0.6808539554665222, "learning_rate": 0.003, "grad_norm": 1.199346899986267, "tokens_per_second": 3360688.891989228, "eta_seconds": 9990.631266117096, "data_wait_seconds": 2.2143125534057617e-05} {"step": 12610, "total_steps": 28610, "tokens": 26445086720, "source": "finemath", "phase": 3, "loss": 2.451525777578354, "ce_loss": 2.4367908239364624, "logit_z_loss": 25.03282904624939, "router_aux_loss": 1.3974559605121613, "router_z_loss": 0.7604651674628258, "routing_entropy_normalized": 0.6804622251408191, "learning_rate": 0.003, "grad_norm": 0.893127977848053, "tokens_per_second": 4176477.687421965, "eta_seconds": 8034.1461181640625, "data_wait_seconds": 1.5616416931152344e-05} {"step": 12620, "total_steps": 28610, "tokens": 26466058240, "source": "dclm", "phase": 3, "loss": 3.480191707611084, "ce_loss": 3.464938670396805, "logit_z_loss": 20.02026391029358, "router_aux_loss": 1.4397098273038864, "router_z_loss": 0.8559106886386871, "routing_entropy_normalized": 0.6689742671339175, "learning_rate": 0.003, "grad_norm": 0.6311156153678894, "tokens_per_second": 4109677.048120651, "eta_seconds": 8159.633977890015, "data_wait_seconds": 1.9758939743041992e-05} {"step": 12630, "total_steps": 28610, "tokens": 26487029760, "source": "fineweb_hq", "phase": 3, "loss": 3.3988756835460663, "ce_loss": 3.384350448846817, "logit_z_loss": 20.046940803527832, "router_aux_loss": 1.3570449501276016, "router_z_loss": 0.9547784477472305, "routing_entropy_normalized": 0.6640573105906545, "learning_rate": 0.003, "grad_norm": 0.9644943475723267, "tokens_per_second": 3855454.8552348115, "eta_seconds": 8692.22704410553, "data_wait_seconds": 2.256035804748535e-05} {"step": 12640, "total_steps": 28610, "tokens": 26508001280, "source": "dclm", "phase": 3, "loss": 3.427531063556671, "ce_loss": 3.413247734308243, "logit_z_loss": 19.09242558479309, "router_aux_loss": 1.3381174355745316, "router_z_loss": 0.9021577537059784, "routing_entropy_normalized": 0.6736935214971909, "learning_rate": 0.003, "grad_norm": 1.7708733081817627, "tokens_per_second": 4255006.798558228, "eta_seconds": 7871.084354400635, "data_wait_seconds": 2.193450927734375e-05} {"step": 12650, "total_steps": 28610, "tokens": 26528972800, "source": "fineweb_hq", "phase": 3, "loss": 3.3433837294578552, "ce_loss": 3.3294774889945984, "logit_z_loss": 19.31192946434021, "router_aux_loss": 1.3017919063568115, "router_z_loss": 0.8883087784051895, "routing_entropy_normalized": 0.6964166722541074, "learning_rate": 0.003, "grad_norm": 0.7622146010398865, "tokens_per_second": 3985234.049472379, "eta_seconds": 8398.63995552063, "data_wait_seconds": 1.659989356994629e-05} {"step": 12660, "total_steps": 28610, "tokens": 26549944320, "source": "fineweb_edu", "phase": 3, "loss": 3.3427956998348236, "ce_loss": 3.328737258911133, "logit_z_loss": 19.365654706954956, "router_aux_loss": 1.3097290396690369, "router_z_loss": 0.9611444249749184, "routing_entropy_normalized": 0.6917648902887087, "learning_rate": 0.003, "grad_norm": 1.0002506971359253, "tokens_per_second": 4127824.273962996, "eta_seconds": 8103.439531326294, "data_wait_seconds": 1.952052116394043e-05} {"step": 12670, "total_steps": 28610, "tokens": 26570915840, "source": "dclm", "phase": 3, "loss": 3.4501692950725555, "ce_loss": 3.4364571273326874, "logit_z_loss": 19.956475734710693, "router_aux_loss": 1.2970087379217148, "router_z_loss": 0.7420973852276802, "routing_entropy_normalized": 0.6846133957576935, "learning_rate": 0.003, "grad_norm": 1.3827584981918335, "tokens_per_second": 3290997.4296399597, "eta_seconds": 10157.590090751648, "data_wait_seconds": 0.15695858001708984} {"step": 12680, "total_steps": 28610, "tokens": 26591887360, "source": "fineweb_edu", "phase": 3, "loss": 3.35675185918808, "ce_loss": 3.3428956270217896, "logit_z_loss": 19.304178953170776, "router_aux_loss": 1.3027205020189285, "router_z_loss": 0.8290287628769875, "routing_entropy_normalized": 0.6776008512484176, "learning_rate": 0.003, "grad_norm": 0.8922821283340454, "tokens_per_second": 4231157.596691343, "eta_seconds": 7895.624447107315, "data_wait_seconds": 2.047419548034668e-05} {"step": 12690, "total_steps": 28610, "tokens": 26612858880, "source": "fineweb_hq", "phase": 3, "loss": 3.3826151192188263, "ce_loss": 3.36921763420105, "logit_z_loss": 19.759445190429688, "router_aux_loss": 1.2657599300146103, "router_z_loss": 0.7398341298103333, "routing_entropy_normalized": 0.7029869188333373, "learning_rate": 0.003, "grad_norm": 1.523148536682129, "tokens_per_second": 3602398.6012750044, "eta_seconds": 9267.897180557251, "data_wait_seconds": 2.1517276763916016e-05} {"step": 12700, "total_steps": 28610, "tokens": 26633830400, "source": "fineweb_edu", "phase": 3, "loss": 3.3430858850479126, "ce_loss": 3.329184830188751, "logit_z_loss": 20.617752075195312, "router_aux_loss": 1.3041098564863205, "router_z_loss": 0.8599464148283005, "routing_entropy_normalized": 0.6465248336919244, "learning_rate": 0.003, "grad_norm": 1.9035605192184448, "tokens_per_second": 3235414.177080408, "eta_seconds": 10312.648240327835, "data_wait_seconds": 0.21518781781196594} {"step": 12710, "total_steps": 28610, "tokens": 26654801920, "source": "finemath", "phase": 3, "loss": 2.462133765220642, "ce_loss": 2.4461672604084015, "logit_z_loss": 24.94858694076538, "router_aux_loss": 1.5364852994680405, "router_z_loss": 0.6016874313354492, "routing_entropy_normalized": 0.6631475836062719, "learning_rate": 0.003, "grad_norm": 1.7848764657974243, "tokens_per_second": 4057622.904959723, "eta_seconds": 8217.795882225037, "data_wait_seconds": 1.58846378326416e-05} {"step": 12720, "total_steps": 28610, "tokens": 26675773440, "source": "dclm", "phase": 3, "loss": 3.437744677066803, "ce_loss": 3.4221377670764923, "logit_z_loss": 20.571324825286865, "router_aux_loss": 1.4852705150842667, "router_z_loss": 0.754204697906971, "routing_entropy_normalized": 0.6631016703114959, "learning_rate": 0.003, "grad_norm": 1.1566691398620605, "tokens_per_second": 3598989.5040435423, "eta_seconds": 9259.194905281067, "data_wait_seconds": 2.199411392211914e-05} {"step": 12730, "total_steps": 28610, "tokens": 26696744960, "source": "fineweb_hq", "phase": 3, "loss": 3.399351567029953, "ce_loss": 3.3852169513702393, "logit_z_loss": 19.679265022277832, "router_aux_loss": 1.321612372994423, "router_z_loss": 0.9185142144560814, "routing_entropy_normalized": 0.709693762982928, "learning_rate": 0.003, "grad_norm": 1.1726545095443726, "tokens_per_second": 4531672.621443395, "eta_seconds": 7348.892239570618, "data_wait_seconds": 2.148747444152832e-05} {"step": 12740, "total_steps": 28610, "tokens": 26717716480, "source": "dclm", "phase": 3, "loss": 3.4955429434776306, "ce_loss": 3.479852706193924, "logit_z_loss": 20.373850107192993, "router_aux_loss": 1.4854326695203781, "router_z_loss": 0.8359186798334122, "routing_entropy_normalized": 0.6627820632206417, "learning_rate": 0.003, "grad_norm": 1.4982706308364868, "tokens_per_second": 3270499.3453151225, "eta_seconds": 10176.367192268372, "data_wait_seconds": 2.2023916244506836e-05} {"step": 12750, "total_steps": 28610, "tokens": 26738688000, "source": "fineweb_hq", "phase": 3, "loss": 3.3669536113739014, "ce_loss": 3.3515850603580475, "logit_z_loss": 20.229695558547974, "router_aux_loss": 1.4283124953508377, "router_z_loss": 1.0853597894310951, "routing_entropy_normalized": 0.6427494164659143, "learning_rate": 0.003, "grad_norm": 0.9991490244865417, "tokens_per_second": 4175877.707043913, "eta_seconds": 7964.99156665802, "data_wait_seconds": 2.244114875793457e-05} {"step": 12760, "total_steps": 28610, "tokens": 26759659520, "source": "fineweb_edu", "phase": 3, "loss": 3.339098870754242, "ce_loss": 3.324568808078766, "logit_z_loss": 19.85509419441223, "router_aux_loss": 1.3390076756477356, "router_z_loss": 1.139993131160736, "routing_entropy_normalized": 0.6821604589401088, "learning_rate": 0.003, "grad_norm": 1.05338454246521, "tokens_per_second": 3569644.7625018223, "eta_seconds": 9311.811513900757, "data_wait_seconds": 1.850724220275879e-05} {"step": 12770, "total_steps": 28610, "tokens": 26780631040, "source": "dclm", "phase": 3, "loss": 3.441685378551483, "ce_loss": 3.427693396806717, "logit_z_loss": 19.84150743484497, "router_aux_loss": 1.2993926107883453, "router_z_loss": 0.9979983642697334, "routing_entropy_normalized": 0.6726722521417486, "learning_rate": 0.003, "grad_norm": 1.0660295486450195, "tokens_per_second": 3993286.221078263, "eta_seconds": 8318.684371948242, "data_wait_seconds": 0.06283301115036011} {"step": 12780, "total_steps": 28610, "tokens": 26801602560, "source": "fineweb_edu", "phase": 3, "loss": 3.320101261138916, "ce_loss": 3.3061993420124054, "logit_z_loss": 20.06476092338562, "router_aux_loss": 1.2796637266874313, "router_z_loss": 1.1053152978420258, "routing_entropy_normalized": 0.733719370920659, "learning_rate": 0.003, "grad_norm": 1.112189769744873, "tokens_per_second": 4350709.056287965, "eta_seconds": 7630.461088180541, "data_wait_seconds": 2.22623348236084e-05} {"step": 12790, "total_steps": 28610, "tokens": 26822574080, "source": "fineweb_hq", "phase": 3, "loss": 3.3625592291355133, "ce_loss": 3.348490923643112, "logit_z_loss": 19.233101844787598, "router_aux_loss": 1.3074541836977005, "router_z_loss": 0.9937339276075363, "routing_entropy_normalized": 0.669750852309091, "learning_rate": 0.003, "grad_norm": 1.1929931640625, "tokens_per_second": 3032854.0969553483, "eta_seconds": 10939.18255853653, "data_wait_seconds": 2.1129846572875977e-05} {"step": 12800, "total_steps": 28610, "tokens": 26843545600, "source": "fineweb_edu", "phase": 3, "loss": 3.329006105661392, "ce_loss": 3.31429386138916, "logit_z_loss": 20.571698665618896, "router_aux_loss": 1.3688374608755112, "router_z_loss": 1.0239356458187103, "routing_entropy_normalized": 0.6551533773547343, "learning_rate": 0.003, "grad_norm": 1.1884734630584717, "tokens_per_second": 3914552.150200792, "eta_seconds": 8469.927554368973, "data_wait_seconds": 2.2172927856445312e-05} {"step": 12810, "total_steps": 28610, "tokens": 26864517120, "source": "finemath", "phase": 3, "loss": 2.371517837047577, "ce_loss": 2.356688916683197, "logit_z_loss": 24.799291372299194, "router_aux_loss": 1.4123384207487106, "router_z_loss": 0.7054788768291473, "routing_entropy_normalized": 0.6843089456722812, "learning_rate": 0.003, "grad_norm": 1.3078680038452148, "tokens_per_second": 3467977.1700257184, "eta_seconds": 9554.56162929535, "data_wait_seconds": 1.8268823623657227e-05} {"step": 12820, "total_steps": 28610, "tokens": 26885488640, "source": "dclm", "phase": 3, "loss": 3.4387124478816986, "ce_loss": 3.4238253831863403, "logit_z_loss": 20.66153883934021, "router_aux_loss": 1.3968741297721863, "router_z_loss": 0.9183419272303581, "routing_entropy_normalized": 0.6646448212100537, "learning_rate": 0.003, "grad_norm": 1.5336170196533203, "tokens_per_second": 3639239.583307337, "eta_seconds": 9099.161877632141, "data_wait_seconds": 1.9758939743041992e-05} {"step": 12830, "total_steps": 28610, "tokens": 26906460160, "source": "fineweb_hq", "phase": 3, "loss": 3.363331973552704, "ce_loss": 3.348900258541107, "logit_z_loss": 19.980580806732178, "router_aux_loss": 1.3389236778020859, "router_z_loss": 1.0425286814570427, "routing_entropy_normalized": 0.641141895113123, "learning_rate": 0.003, "grad_norm": 1.2451828718185425, "tokens_per_second": 4159086.016094963, "eta_seconds": 7956.810326099396, "data_wait_seconds": 2.3126602172851562e-05} {"step": 12840, "total_steps": 28610, "tokens": 26927431680, "source": "dclm", "phase": 3, "loss": 3.438725084066391, "ce_loss": 3.4232657849788666, "logit_z_loss": 20.44415044784546, "router_aux_loss": 1.4637462943792343, "router_z_loss": 0.8218013197183609, "routing_entropy_normalized": 0.6630099485858967, "learning_rate": 0.003, "grad_norm": 1.6868637800216675, "tokens_per_second": 3999901.3323577726, "eta_seconds": 8268.225711584091, "data_wait_seconds": 2.4318695068359375e-05} {"step": 12850, "total_steps": 28610, "tokens": 26948403200, "source": "fineweb_hq", "phase": 3, "loss": 3.3781974017620087, "ce_loss": 3.3636607229709625, "logit_z_loss": 19.848978519439697, "router_aux_loss": 1.3449469804763794, "router_z_loss": 1.0872658491134644, "routing_entropy_normalized": 0.6841255190163956, "learning_rate": 0.003, "grad_norm": 1.0535805225372314, "tokens_per_second": 4211128.618987996, "eta_seconds": 7848.517229080201, "data_wait_seconds": 1.8984079360961914e-05} {"step": 12860, "total_steps": 28610, "tokens": 26969374720, "source": "fineweb_edu", "phase": 3, "loss": 3.350927770137787, "ce_loss": 3.336416631937027, "logit_z_loss": 19.31902050971985, "router_aux_loss": 1.334537073969841, "router_z_loss": 1.1656846106052399, "routing_entropy_normalized": 0.6789914235105542, "learning_rate": 0.003, "grad_norm": 1.003406286239624, "tokens_per_second": 3289640.845855104, "eta_seconds": 10040.653538703918, "data_wait_seconds": 1.677870750427246e-05} {"step": 12870, "total_steps": 28610, "tokens": 26990346240, "source": "dclm", "phase": 3, "loss": 3.4432381689548492, "ce_loss": 3.428723990917206, "logit_z_loss": 20.509852647781372, "router_aux_loss": 1.3578601777553558, "router_z_loss": 0.9355702549219131, "routing_entropy_normalized": 0.6716961759221557, "learning_rate": 0.003, "grad_norm": 0.9217737913131714, "tokens_per_second": 3768654.6204703734, "eta_seconds": 8758.874400615692, "data_wait_seconds": 0.06846845149993896} {"step": 12880, "total_steps": 28610, "tokens": 27011317760, "source": "fineweb_edu", "phase": 3, "loss": 3.338915526866913, "ce_loss": 3.324173092842102, "logit_z_loss": 19.665767669677734, "router_aux_loss": 1.3589192777872086, "router_z_loss": 1.1532138586044312, "routing_entropy_normalized": 0.6715385506375973, "learning_rate": 0.003, "grad_norm": 0.946163535118103, "tokens_per_second": 4512131.237308098, "eta_seconds": 7311.002101898193, "data_wait_seconds": 2.6613473892211914e-05} {"step": 12890, "total_steps": 28610, "tokens": 27032289280, "source": "fineweb_hq", "phase": 3, "loss": 3.4149294197559357, "ce_loss": 3.4007979929447174, "logit_z_loss": 19.693259716033936, "router_aux_loss": 1.3059837371110916, "router_z_loss": 1.0716681256890297, "routing_entropy_normalized": 0.6759337484479777, "learning_rate": 0.003, "grad_norm": 1.9786598682403564, "tokens_per_second": 4273234.103082861, "eta_seconds": 7714.8194189071655, "data_wait_seconds": 2.345442771911621e-05} {"step": 12900, "total_steps": 28610, "tokens": 27053260800, "source": "fineweb_edu", "phase": 3, "loss": 3.3075444102287292, "ce_loss": 3.2938029170036316, "logit_z_loss": 20.46603798866272, "router_aux_loss": 1.2684832513332367, "router_z_loss": 1.0567331835627556, "routing_entropy_normalized": 0.6695357343589696, "learning_rate": 0.003, "grad_norm": 1.510406732559204, "tokens_per_second": 4046062.1652790634, "eta_seconds": 8142.795778751373, "data_wait_seconds": 2.014636993408203e-05} {"step": 12910, "total_steps": 28610, "tokens": 27074232320, "source": "finemath", "phase": 3, "loss": 2.4568758606910706, "ce_loss": 2.442933827638626, "logit_z_loss": 26.01901412010193, "router_aux_loss": 1.3204835951328278, "router_z_loss": 0.7372282966971397, "routing_entropy_normalized": 0.6980461160986476, "learning_rate": 0.003, "grad_norm": 0.7404959201812744, "tokens_per_second": 3199313.6105305334, "eta_seconds": 10291.359462738037, "data_wait_seconds": 2.1129846572875977e-05} {"step": 12920, "total_steps": 28610, "tokens": 27095203840, "source": "dclm", "phase": 3, "loss": 3.471005618572235, "ce_loss": 3.4567730724811554, "logit_z_loss": 20.311768531799316, "router_aux_loss": 1.3436387479305267, "router_z_loss": 0.7961708679795265, "routing_entropy_normalized": 0.6754015646244038, "learning_rate": 0.003, "grad_norm": 1.1837416887283325, "tokens_per_second": 2619078.3794710254, "eta_seconds": 12563.318126678467, "data_wait_seconds": 1.940131187438965e-05} {"step": 12930, "total_steps": 28610, "tokens": 27116175360, "source": "fineweb_hq", "phase": 3, "loss": 3.372303932905197, "ce_loss": 3.3583383560180664, "logit_z_loss": 20.31287384033203, "router_aux_loss": 1.3062139004468918, "router_z_loss": 0.9034553468227386, "routing_entropy_normalized": 0.6885216738831944, "learning_rate": 0.003, "grad_norm": 0.8935734629631042, "tokens_per_second": 4179413.849682195, "eta_seconds": 7867.931854248047, "data_wait_seconds": 2.0176172256469727e-05} {"step": 12940, "total_steps": 28610, "tokens": 27137146880, "source": "dclm", "phase": 3, "loss": 3.4903919994831085, "ce_loss": 3.476026237010956, "logit_z_loss": 20.60102605819702, "router_aux_loss": 1.3592817932367325, "router_z_loss": 0.772942416369915, "routing_entropy_normalized": 0.6729579755500112, "learning_rate": 0.003, "grad_norm": 1.4331880807876587, "tokens_per_second": 3403945.897107497, "eta_seconds": 9654.199224472046, "data_wait_seconds": 2.0205974578857422e-05} {"step": 12950, "total_steps": 28610, "tokens": 27158118400, "source": "fineweb_hq", "phase": 3, "loss": 3.323604792356491, "ce_loss": 3.308749347925186, "logit_z_loss": 20.167007207870483, "router_aux_loss": 1.3957729935646057, "router_z_loss": 0.8976819589734077, "routing_entropy_normalized": 0.6498034668573307, "learning_rate": 0.003, "grad_norm": 0.8253738880157471, "tokens_per_second": 4331570.334339919, "eta_seconds": 7581.869341850281, "data_wait_seconds": 2.1398067474365234e-05} {"step": 12960, "total_steps": 28610, "tokens": 27179089920, "source": "fineweb_edu", "phase": 3, "loss": 3.304905116558075, "ce_loss": 3.290108382701874, "logit_z_loss": 20.346320867538452, "router_aux_loss": 1.3839222639799118, "router_z_loss": 0.9575277417898178, "routing_entropy_normalized": 0.6953914033608621, "learning_rate": 0.003, "grad_norm": 1.021188497543335, "tokens_per_second": 3155440.734181072, "eta_seconds": 10401.218582391739, "data_wait_seconds": 2.3305416107177734e-05} {"step": 12970, "total_steps": 28610, "tokens": 27200061440, "source": "dclm", "phase": 3, "loss": 3.437113344669342, "ce_loss": 3.422131896018982, "logit_z_loss": 20.600859880447388, "router_aux_loss": 1.4141934514045715, "router_z_loss": 0.8394983634352684, "routing_entropy_normalized": 0.64199339532299, "learning_rate": 0.003, "grad_norm": 1.422589659690857, "tokens_per_second": 3855358.364265361, "eta_seconds": 8507.49896144867, "data_wait_seconds": 1.9788742065429688e-05} {"step": 12980, "total_steps": 28610, "tokens": 27221032960, "source": "fineweb_edu", "phase": 3, "loss": 3.323607176542282, "ce_loss": 3.3085368871688843, "logit_z_loss": 19.91767454147339, "router_aux_loss": 1.4061031341552734, "router_z_loss": 1.0093001946806908, "routing_entropy_normalized": 0.6605289216579647, "learning_rate": 0.003, "grad_norm": 0.9022138118743896, "tokens_per_second": 4131817.0136466892, "eta_seconds": 7933.189115524292, "data_wait_seconds": 2.059340476989746e-05} {"step": 12990, "total_steps": 28610, "tokens": 27242004480, "source": "fineweb_hq", "phase": 3, "loss": 3.3472713232040405, "ce_loss": 3.332555651664734, "logit_z_loss": 20.33242177963257, "router_aux_loss": 1.3682642877101898, "router_z_loss": 1.0330253094434738, "routing_entropy_normalized": 0.6698270245253781, "learning_rate": 0.003, "grad_norm": 1.0405447483062744, "tokens_per_second": 3569984.066012999, "eta_seconds": 9175.815251350403, "data_wait_seconds": 1.9073486328125e-05} {"step": 13000, "total_steps": 28610, "tokens": 27262976000, "source": "fineweb_edu", "phase": 3, "loss": 3.3219297528266907, "ce_loss": 3.307261049747467, "logit_z_loss": 21.23731756210327, "router_aux_loss": 1.3653708547353745, "router_z_loss": 1.0150211602449417, "routing_entropy_normalized": 0.6493122370369722, "learning_rate": 0.003, "grad_norm": 1.1196513175964355, "tokens_per_second": 4028693.660525508, "eta_seconds": 8125.845616102219, "data_wait_seconds": 2.333521842956543e-05} {"step": 13010, "total_steps": 28610, "tokens": 27283947520, "source": "finemath", "phase": 3, "loss": 2.3155936151742935, "ce_loss": 2.3004061579704285, "logit_z_loss": 24.439837217330933, "router_aux_loss": 1.4238142222166061, "router_z_loss": 0.9493434503674507, "routing_entropy_normalized": 0.6789822255797466, "learning_rate": 0.003, "grad_norm": 0.9242461323738098, "tokens_per_second": 4159315.9190260307, "eta_seconds": 7865.613441467285, "data_wait_seconds": 2.1517276763916016e-05} {"step": 13020, "total_steps": 28610, "tokens": 27304919040, "source": "dclm", "phase": 3, "loss": 3.4873546957969666, "ce_loss": 3.472693979740143, "logit_z_loss": 21.039012908935547, "router_aux_loss": 1.3802758157253265, "router_z_loss": 0.8579958602786064, "routing_entropy_normalized": 0.6879638020786166, "learning_rate": 0.003, "grad_norm": 1.19839346408844, "tokens_per_second": 3595566.1293927394, "eta_seconds": 9093.032502651215, "data_wait_seconds": 0.10656502842903137} {"step": 13030, "total_steps": 28610, "tokens": 27325890560, "source": "fineweb_hq", "phase": 3, "loss": 3.3973779380321503, "ce_loss": 3.382629483938217, "logit_z_loss": 20.660686492919922, "router_aux_loss": 1.3730590343475342, "router_z_loss": 1.0178396254777908, "routing_entropy_normalized": 0.6653246812047346, "learning_rate": 0.003, "grad_norm": 1.538914442062378, "tokens_per_second": 4363373.400712578, "eta_seconds": 7488.1577072143555, "data_wait_seconds": 2.3096799850463867e-05} {"step": 13040, "total_steps": 28610, "tokens": 27346862080, "source": "dclm", "phase": 3, "loss": 3.4504260420799255, "ce_loss": 3.4351580440998077, "logit_z_loss": 20.96249794960022, "router_aux_loss": 1.4355132430791855, "router_z_loss": 0.9128630310297012, "routing_entropy_normalized": 0.6746394374422152, "learning_rate": 0.003, "grad_norm": 0.9519222974777222, "tokens_per_second": 3429876.2938309163, "eta_seconds": 9520.068318128586, "data_wait_seconds": 1.683831214904785e-05} {"step": 13050, "total_steps": 28610, "tokens": 27367833600, "source": "fineweb_hq", "phase": 3, "loss": 3.3720592856407166, "ce_loss": 3.3575331568717957, "logit_z_loss": 20.14297389984131, "router_aux_loss": 1.3469455540180206, "router_z_loss": 1.0566909909248352, "routing_entropy_normalized": 0.6945356960727314, "learning_rate": 0.003, "grad_norm": 1.240888237953186, "tokens_per_second": 4253866.800539862, "eta_seconds": 7671.064151763915, "data_wait_seconds": 2.2143125534057617e-05} {"step": 13060, "total_steps": 28610, "tokens": 27388805120, "source": "fineweb_edu", "phase": 3, "loss": 3.3618487119674683, "ce_loss": 3.347255438566208, "logit_z_loss": 20.93536353111267, "router_aux_loss": 1.3551561832427979, "router_z_loss": 1.0416882634162903, "routing_entropy_normalized": 0.6524303516902428, "learning_rate": 0.003, "grad_norm": 1.4773374795913696, "tokens_per_second": 3552635.446109118, "eta_seconds": 9179.30198431015, "data_wait_seconds": 2.1070241928100586e-05} {"step": 13070, "total_steps": 28610, "tokens": 27409776640, "source": "dclm", "phase": 3, "loss": 3.4160005152225494, "ce_loss": 3.4022780060768127, "logit_z_loss": 21.296522617340088, "router_aux_loss": 1.2885870933532715, "router_z_loss": 0.836667388677597, "routing_entropy_normalized": 0.6754261342292722, "learning_rate": 0.003, "grad_norm": 0.965604305267334, "tokens_per_second": 3764943.415248664, "eta_seconds": 8656.104085922241, "data_wait_seconds": 0.009319961071014404} {"step": 13080, "total_steps": 28610, "tokens": 27430748160, "source": "fineweb_edu", "phase": 3, "loss": 3.34591281414032, "ce_loss": 3.3313856422901154, "logit_z_loss": 20.092484712600708, "router_aux_loss": 1.3577093929052353, "router_z_loss": 0.9500478282570839, "routing_entropy_normalized": 0.6559743752139136, "learning_rate": 0.003, "grad_norm": 1.1358747482299805, "tokens_per_second": 4235032.48759788, "eta_seconds": 7690.323664665222, "data_wait_seconds": 2.3037195205688477e-05} {"step": 13090, "total_steps": 28610, "tokens": 27451719680, "source": "fineweb_hq", "phase": 3, "loss": 3.3612259328365326, "ce_loss": 3.3463211953639984, "logit_z_loss": 20.638224363327026, "router_aux_loss": 1.401576966047287, "router_z_loss": 0.8890006840229034, "routing_entropy_normalized": 0.6589722179689048, "learning_rate": 0.003, "grad_norm": 1.0395313501358032, "tokens_per_second": 3256776.6960794134, "eta_seconds": 9993.868808746338, "data_wait_seconds": 2.09808349609375e-05} {"step": 13100, "total_steps": 28610, "tokens": 27472691200, "source": "fineweb_edu", "phase": 3, "loss": 3.350013852119446, "ce_loss": 3.3351977169513702, "logit_z_loss": 19.7944757938385, "router_aux_loss": 1.3849997073411942, "router_z_loss": 0.9661094024777412, "routing_entropy_normalized": 0.6692434711364571, "learning_rate": 0.003, "grad_norm": 1.2543103694915771, "tokens_per_second": 4304804.657463572, "eta_seconds": 7555.9357759952545, "data_wait_seconds": 1.8805265426635742e-05} {"step": 13110, "total_steps": 28610, "tokens": 27493662720, "source": "finemath", "phase": 3, "loss": 2.246377483010292, "ce_loss": 2.231714442372322, "logit_z_loss": 27.426366567611694, "router_aux_loss": 1.3941909223794937, "router_z_loss": 0.7211095616221428, "routing_entropy_normalized": 0.7023802218861407, "learning_rate": 0.003, "grad_norm": 5.066243648529053, "tokens_per_second": 3838620.7078339984, "eta_seconds": 8468.108332157135, "data_wait_seconds": 1.8149614334106445e-05} {"step": 13120, "total_steps": 28610, "tokens": 27514634240, "source": "dclm", "phase": 3, "loss": 3.4546090364456177, "ce_loss": 3.4397498071193695, "logit_z_loss": 20.978245496749878, "router_aux_loss": 1.3962153494358063, "router_z_loss": 0.8971030116081238, "routing_entropy_normalized": 0.6694539965067781, "learning_rate": 0.003, "grad_norm": 1.5463011264801025, "tokens_per_second": 3822566.4021382504, "eta_seconds": 8498.187098026276, "data_wait_seconds": 1.7762184143066406e-05} {"step": 13130, "total_steps": 28610, "tokens": 27535605760, "source": "fineweb_hq", "phase": 3, "loss": 3.3418869376182556, "ce_loss": 3.3268186151981354, "logit_z_loss": 20.1368510723114, "router_aux_loss": 1.3948629796504974, "router_z_loss": 1.119745060801506, "routing_entropy_normalized": 0.6815561920976779, "learning_rate": 0.003, "grad_norm": 1.1840095520019531, "tokens_per_second": 3856026.6337484596, "eta_seconds": 8419.006413459778, "data_wait_seconds": 1.7762184143066406e-05} {"step": 13140, "total_steps": 28610, "tokens": 27556577280, "source": "dclm", "phase": 3, "loss": 3.4840860664844513, "ce_loss": 3.468485862016678, "logit_z_loss": 20.47411346435547, "router_aux_loss": 1.4666747599840164, "router_z_loss": 0.9335085600614548, "routing_entropy_normalized": 0.6675755162780365, "learning_rate": 0.003, "grad_norm": 1.3145097494125366, "tokens_per_second": 3975456.1216870816, "eta_seconds": 8160.809840917587, "data_wait_seconds": 0.009941011667251587} {"step": 13150, "total_steps": 28610, "tokens": 27577548800, "source": "fineweb_hq", "phase": 3, "loss": 3.346374064683914, "ce_loss": 3.33179247379303, "logit_z_loss": 20.351187229156494, "router_aux_loss": 1.3442273288965225, "router_z_loss": 1.139326587319374, "routing_entropy_normalized": 0.6840226174727746, "learning_rate": 0.003, "grad_norm": 0.4983263909816742, "tokens_per_second": 3461588.918706845, "eta_seconds": 9366.210339069366, "data_wait_seconds": 0.030479133129119873} {"step": 13160, "total_steps": 28610, "tokens": 27598520320, "source": "fineweb_edu", "phase": 3, "loss": 3.377800017595291, "ce_loss": 3.3627074658870697, "logit_z_loss": 20.88821840286255, "router_aux_loss": 1.387350633740425, "router_z_loss": 1.2190402150154114, "routing_entropy_normalized": 0.6491961872445954, "learning_rate": 0.003, "grad_norm": 0.9906971454620361, "tokens_per_second": 3663118.06434826, "eta_seconds": 8845.196313858032, "data_wait_seconds": 2.2798776626586914e-05} {"step": 13170, "total_steps": 28610, "tokens": 27619491840, "source": "dclm", "phase": 3, "loss": 3.4423381686210632, "ce_loss": 3.428381234407425, "logit_z_loss": 20.970155477523804, "router_aux_loss": 1.2971973717212677, "router_z_loss": 0.9849560186266899, "routing_entropy_normalized": 0.6881848795897134, "learning_rate": 0.003, "grad_norm": 1.0751824378967285, "tokens_per_second": 3159432.7600216013, "eta_seconds": 10248.683652877808, "data_wait_seconds": 0.05876246094703674} {"step": 13180, "total_steps": 28610, "tokens": 27640463360, "source": "fineweb_edu", "phase": 3, "loss": 3.3451939523220062, "ce_loss": 3.3310718536376953, "logit_z_loss": 21.0574209690094, "router_aux_loss": 1.302710399031639, "router_z_loss": 1.0950035154819489, "routing_entropy_normalized": 0.7004900468547609, "learning_rate": 0.003, "grad_norm": 1.1006147861480713, "tokens_per_second": 3159351.168300334, "eta_seconds": 10242.310410022736, "data_wait_seconds": 1.990795135498047e-05} {"step": 13190, "total_steps": 28610, "tokens": 27661434880, "source": "fineweb_hq", "phase": 3, "loss": 3.3548086881637573, "ce_loss": 3.341044247150421, "logit_z_loss": 19.938582181930542, "router_aux_loss": 1.2798935323953629, "router_z_loss": 0.9654858410358429, "routing_entropy_normalized": 0.7131091964068696, "learning_rate": 0.003, "grad_norm": 1.3199838399887085, "tokens_per_second": 2952574.132443972, "eta_seconds": 10952.505301952362, "data_wait_seconds": 2.065300941467285e-05} {"step": 13200, "total_steps": 28610, "tokens": 27682406400, "source": "fineweb_edu", "phase": 3, "loss": 3.363020956516266, "ce_loss": 3.348348468542099, "logit_z_loss": 21.0086772441864, "router_aux_loss": 1.3711944669485092, "router_z_loss": 0.9605474546551704, "routing_entropy_normalized": 0.674739521297742, "learning_rate": 0.003, "grad_norm": 1.3581405878067017, "tokens_per_second": 4085300.8020350863, "eta_seconds": 7910.58330488205, "data_wait_seconds": 1.9729137420654297e-05} {"step": 13210, "total_steps": 28610, "tokens": 27703377920, "source": "finemath", "phase": 3, "loss": 2.363631099462509, "ce_loss": 2.3496747612953186, "logit_z_loss": 23.597800970077515, "router_aux_loss": 1.319221407175064, "router_z_loss": 0.764086939394474, "routing_entropy_normalized": 0.7129975697599105, "learning_rate": 0.003, "grad_norm": 1.0838074684143066, "tokens_per_second": 3732510.73370133, "eta_seconds": 8652.658519744873, "data_wait_seconds": 1.7881393432617188e-05} {"step": 13220, "total_steps": 28610, "tokens": 27724349440, "source": "dclm", "phase": 3, "loss": 3.4667428135871887, "ce_loss": 3.452419728040695, "logit_z_loss": 20.980751991271973, "router_aux_loss": 1.3521841019392014, "router_z_loss": 0.8012647852301598, "routing_entropy_normalized": 0.6760052042067669, "learning_rate": 0.003, "grad_norm": 1.4131203889846802, "tokens_per_second": 3781264.429333876, "eta_seconds": 8535.549386501312, "data_wait_seconds": 2.041459083557129e-05} {"step": 13230, "total_steps": 28610, "tokens": 27745320960, "source": "fineweb_hq", "phase": 3, "loss": 3.395632714033127, "ce_loss": 3.381708264350891, "logit_z_loss": 20.64292049407959, "router_aux_loss": 1.299552708864212, "router_z_loss": 0.9289521649479866, "routing_entropy_normalized": 0.6884728338432673, "learning_rate": 0.003, "grad_norm": 1.0382311344146729, "tokens_per_second": 4419093.770916119, "eta_seconds": 7298.826282501221, "data_wait_seconds": 2.396106719970703e-05} {"step": 13240, "total_steps": 28610, "tokens": 27766292480, "source": "dclm", "phase": 3, "loss": 3.510060638189316, "ce_loss": 3.495252788066864, "logit_z_loss": 21.19483971595764, "router_aux_loss": 1.4025809913873672, "router_z_loss": 0.7819940522313118, "routing_entropy_normalized": 0.666335332558506, "learning_rate": 0.003, "grad_norm": 1.3489134311676025, "tokens_per_second": 3318170.363841681, "eta_seconds": 9714.156509637833, "data_wait_seconds": 1.9490718841552734e-05} {"step": 13250, "total_steps": 28610, "tokens": 27787264000, "source": "fineweb_hq", "phase": 3, "loss": 3.3567211031913757, "ce_loss": 3.341084599494934, "logit_z_loss": 21.201931715011597, "router_aux_loss": 1.4619718790054321, "router_z_loss": 1.0167538076639175, "routing_entropy_normalized": 0.6477606651942284, "learning_rate": 0.003, "grad_norm": 0.8615760803222656, "tokens_per_second": 3956275.7096541733, "eta_seconds": 8142.065185546875, "data_wait_seconds": 2.1010637283325195e-05} {"step": 13260, "total_steps": 28610, "tokens": 27808235520, "source": "fineweb_edu", "phase": 3, "loss": 3.3498780727386475, "ce_loss": 3.3351922631263733, "logit_z_loss": 20.089006185531616, "router_aux_loss": 1.3630296289920807, "router_z_loss": 1.0555176883935928, "routing_entropy_normalized": 0.6815999278538313, "learning_rate": 0.003, "grad_norm": 1.4947068691253662, "tokens_per_second": 3946219.386933755, "eta_seconds": 8157.499632835389, "data_wait_seconds": 1.7970800399780273e-05} {"step": 13270, "total_steps": 28610, "tokens": 27829207040, "source": "dclm", "phase": 3, "loss": 3.4526791274547577, "ce_loss": 3.4379864633083344, "logit_z_loss": 21.19231605529785, "router_aux_loss": 1.3787501454353333, "router_z_loss": 0.9051466956734657, "routing_entropy_normalized": 0.6494404859128098, "learning_rate": 0.003, "grad_norm": 1.1850357055664062, "tokens_per_second": 4019994.7480155826, "eta_seconds": 8002.5755496025085, "data_wait_seconds": 0.027178585529327393} {"step": 13280, "total_steps": 28610, "tokens": 27850178560, "source": "fineweb_edu", "phase": 3, "loss": 3.3788068294525146, "ce_loss": 3.364424705505371, "logit_z_loss": 20.83784031867981, "router_aux_loss": 1.331499844789505, "router_z_loss": 1.0671195164322853, "routing_entropy_normalized": 0.67647447875848, "learning_rate": 0.003, "grad_norm": 1.132151484489441, "tokens_per_second": 4243059.6353448145, "eta_seconds": 7576.923947095871, "data_wait_seconds": 2.0295381546020508e-05} {"step": 13290, "total_steps": 28610, "tokens": 27871150080, "source": "fineweb_hq", "phase": 3, "loss": 3.3491100668907166, "ce_loss": 3.3347732722759247, "logit_z_loss": 21.13276767730713, "router_aux_loss": 1.3251517117023468, "router_z_loss": 1.085256740450859, "routing_entropy_normalized": 0.709366841922928, "learning_rate": 0.003, "grad_norm": 0.7319915890693665, "tokens_per_second": 3540416.5396110904, "eta_seconds": 9074.74255657196, "data_wait_seconds": 1.868605613708496e-05} {"step": 13300, "total_steps": 28610, "tokens": 27892121600, "source": "fineweb_edu", "phase": 3, "loss": 3.36880162358284, "ce_loss": 3.354435980319977, "logit_z_loss": 20.78201651573181, "router_aux_loss": 1.326895833015442, "router_z_loss": 1.096644215285778, "routing_entropy_normalized": 0.686769264244286, "learning_rate": 0.003, "grad_norm": 1.686638593673706, "tokens_per_second": 2956758.1789358957, "eta_seconds": 10858.986490249634, "data_wait_seconds": 2.03549861907959e-05} {"step": 13310, "total_steps": 28610, "tokens": 27913093120, "source": "finemath", "phase": 3, "loss": 2.440699577331543, "ce_loss": 2.426075130701065, "logit_z_loss": 24.172367334365845, "router_aux_loss": 1.3859414011240005, "router_z_loss": 0.7650662809610367, "routing_entropy_normalized": 0.6912477455208943, "learning_rate": 0.003, "grad_norm": 0.5767086148262024, "tokens_per_second": 3926303.6764731416, "eta_seconds": 8172.171142101288, "data_wait_seconds": 1.7821788787841797e-05} {"step": 13320, "total_steps": 28610, "tokens": 27934064640, "source": "dclm", "phase": 3, "loss": 3.4465559124946594, "ce_loss": 3.4323123693466187, "logit_z_loss": 21.030532836914062, "router_aux_loss": 1.3383551985025406, "router_z_loss": 0.8599853068590164, "routing_entropy_normalized": 0.6940009032843966, "learning_rate": 0.003, "grad_norm": 0.6156112551689148, "tokens_per_second": 3613839.537035732, "eta_seconds": 8872.960116624832, "data_wait_seconds": 1.9162893295288086e-05} {"step": 13330, "total_steps": 28610, "tokens": 27955036160, "source": "fineweb_hq", "phase": 3, "loss": 3.3981597125530243, "ce_loss": 3.3842679858207703, "logit_z_loss": 20.535974502563477, "router_aux_loss": 1.2928090542554855, "router_z_loss": 0.9636939615011215, "routing_entropy_normalized": 0.7062522136670613, "learning_rate": 0.003, "grad_norm": 1.1691513061523438, "tokens_per_second": 4010354.511958102, "eta_seconds": 7990.436372756958, "data_wait_seconds": 2.086162567138672e-05} {"step": 13340, "total_steps": 28610, "tokens": 27976007680, "source": "dclm", "phase": 3, "loss": 3.5253257751464844, "ce_loss": 3.5114170014858246, "logit_z_loss": 21.399023056030273, "router_aux_loss": 1.3076189905405045, "router_z_loss": 0.8326096758246422, "routing_entropy_normalized": 0.6695339157540724, "learning_rate": 0.003, "grad_norm": 1.594166874885559, "tokens_per_second": 3933888.7479798, "eta_seconds": 8140.42111802101, "data_wait_seconds": 2.1129846572875977e-05} {"step": 13350, "total_steps": 28610, "tokens": 27996979200, "source": "fineweb_hq", "phase": 3, "loss": 3.3256016075611115, "ce_loss": 3.3114363253116608, "logit_z_loss": 21.061203241348267, "router_aux_loss": 1.330466166138649, "router_z_loss": 0.8606010675430298, "routing_entropy_normalized": 0.669227285308487, "learning_rate": 0.003, "grad_norm": 1.0503149032592773, "tokens_per_second": 3996943.443988333, "eta_seconds": 8006.753152370453, "data_wait_seconds": 1.8715858459472656e-05} {"step": 13360, "total_steps": 28610, "tokens": 28017950720, "source": "fineweb_edu", "phase": 3, "loss": 3.3694034218788147, "ce_loss": 3.35477015376091, "logit_z_loss": 20.399624347686768, "router_aux_loss": 1.3777827471494675, "router_z_loss": 0.8554448261857033, "routing_entropy_normalized": 0.6849995552266037, "learning_rate": 0.003, "grad_norm": 1.0276612043380737, "tokens_per_second": 4109906.897860035, "eta_seconds": 7781.579679250717, "data_wait_seconds": 2.2739171981811523e-05} {"step": 13370, "total_steps": 28610, "tokens": 28038922240, "source": "dclm", "phase": 3, "loss": 3.4483217895030975, "ce_loss": 3.433924078941345, "logit_z_loss": 21.3423752784729, "router_aux_loss": 1.3628793507814407, "router_z_loss": 0.7689194232225418, "routing_entropy_normalized": 0.6517464366215868, "learning_rate": 0.003, "grad_norm": 0.6331459879875183, "tokens_per_second": 4016662.4072083626, "eta_seconds": 7957.003412246704, "data_wait_seconds": 2.0056962966918945e-05} {"step": 13380, "total_steps": 28610, "tokens": 28059893760, "source": "fineweb_edu", "phase": 3, "loss": 3.3984521627426147, "ce_loss": 3.3848869502544403, "logit_z_loss": 20.51720380783081, "router_aux_loss": 1.269866093993187, "router_z_loss": 0.8665498048067093, "routing_entropy_normalized": 0.6966219022868598, "learning_rate": 0.003, "grad_norm": 0.6903108954429626, "tokens_per_second": 4167337.8929137196, "eta_seconds": 7664.275319337845, "data_wait_seconds": 1.8298625946044922e-05} {"step": 13390, "total_steps": 28610, "tokens": 28080865280, "source": "fineweb_hq", "phase": 3, "loss": 3.3288894593715668, "ce_loss": 3.31497061252594, "logit_z_loss": 20.575615644454956, "router_aux_loss": 1.3144546300172806, "router_z_loss": 0.7743323817849159, "routing_entropy_normalized": 0.6503811383924933, "learning_rate": 0.003, "grad_norm": 1.31425142288208, "tokens_per_second": 3680677.0358092277, "eta_seconds": 8671.951689720154, "data_wait_seconds": 1.728534698486328e-05} {"step": 13400, "total_steps": 28610, "tokens": 28101836800, "source": "fineweb_edu", "phase": 3, "loss": 3.406029909849167, "ce_loss": 3.3923670053482056, "logit_z_loss": 20.21928858757019, "router_aux_loss": 1.2870809882879257, "router_z_loss": 0.792113684117794, "routing_entropy_normalized": 0.6790896060229353, "learning_rate": 0.003, "grad_norm": 1.0305402278900146, "tokens_per_second": 3918077.684827213, "eta_seconds": 8141.156068325043, "data_wait_seconds": 2.193450927734375e-05} {"step": 13410, "total_steps": 28610, "tokens": 28122808320, "source": "finemath", "phase": 3, "loss": 2.391771614551544, "ce_loss": 2.3776192665100098, "logit_z_loss": 25.623536586761475, "router_aux_loss": 1.3557190895080566, "router_z_loss": 0.5952011421322823, "routing_entropy_normalized": 0.6883800934706269, "learning_rate": 0.003, "grad_norm": 1.0127594470977783, "tokens_per_second": 4011958.85979839, "eta_seconds": 7945.423049926758, "data_wait_seconds": 2.0772218704223633e-05} {"step": 13420, "total_steps": 28610, "tokens": 28143779840, "source": "dclm", "phase": 3, "loss": 3.468734532594681, "ce_loss": 3.454210638999939, "logit_z_loss": 21.28624963760376, "router_aux_loss": 1.3783335983753204, "router_z_loss": 0.74058398604393, "routing_entropy_normalized": 0.6452974857265172, "learning_rate": 0.003, "grad_norm": 0.9902369976043701, "tokens_per_second": 3785101.223796864, "eta_seconds": 8416.086386203766, "data_wait_seconds": 2.148747444152832e-05} {"step": 13430, "total_steps": 28610, "tokens": 28164751360, "source": "fineweb_hq", "phase": 3, "loss": 3.3361518681049347, "ce_loss": 3.3213183879852295, "logit_z_loss": 21.04888343811035, "router_aux_loss": 1.3977144956588745, "router_z_loss": 0.8563129603862762, "routing_entropy_normalized": 0.6786216624416915, "learning_rate": 0.003, "grad_norm": 1.1448109149932861, "tokens_per_second": 3780869.3140214738, "eta_seconds": 8419.95972776413, "data_wait_seconds": 2.205371856689453e-05} {"step": 13440, "total_steps": 28610, "tokens": 28185722880, "source": "dclm", "phase": 3, "loss": 3.4290034472942352, "ce_loss": 3.413313537836075, "logit_z_loss": 22.081374645233154, "router_aux_loss": 1.4897661954164505, "router_z_loss": 0.7922161296010017, "routing_entropy_normalized": 0.6679070319549565, "learning_rate": 0.003, "grad_norm": 0.7884587645530701, "tokens_per_second": 4036646.183768164, "eta_seconds": 7881.244575738907, "data_wait_seconds": 2.300739288330078e-05} {"step": 13450, "total_steps": 28610, "tokens": 28206694400, "source": "fineweb_hq", "phase": 3, "loss": 3.3873752653598785, "ce_loss": 3.3725249469280243, "logit_z_loss": 20.87788462638855, "router_aux_loss": 1.386897400021553, "router_z_loss": 0.9813335090875626, "routing_entropy_normalized": 0.6978238881575074, "learning_rate": 0.003, "grad_norm": 0.8936918377876282, "tokens_per_second": 4164615.2455160553, "eta_seconds": 7634.036386489868, "data_wait_seconds": 1.9758939743041992e-05} {"step": 13460, "total_steps": 28610, "tokens": 28227665920, "source": "fineweb_edu", "phase": 3, "loss": 3.418722629547119, "ce_loss": 3.4041947424411774, "logit_z_loss": 20.333803415298462, "router_aux_loss": 1.351506069302559, "router_z_loss": 1.0128253996372223, "routing_entropy_normalized": 0.6758948207909836, "learning_rate": 0.003, "grad_norm": 2.497100830078125, "tokens_per_second": 3441698.8877936644, "eta_seconds": 9231.444654464722, "data_wait_seconds": 1.7970800399780273e-05} {"step": 13470, "total_steps": 28610, "tokens": 28248637440, "source": "dclm", "phase": 3, "loss": 3.4971903264522552, "ce_loss": 3.481408476829529, "logit_z_loss": 21.295868158340454, "router_aux_loss": 1.484535813331604, "router_z_loss": 0.9364651516079903, "routing_entropy_normalized": 0.6519498078372853, "learning_rate": 0.003, "grad_norm": 1.707057237625122, "tokens_per_second": 4092693.132131545, "eta_seconds": 7757.94330406189, "data_wait_seconds": 2.2232532501220703e-05} {"step": 13480, "total_steps": 28610, "tokens": 28269608960, "source": "fineweb_edu", "phase": 3, "loss": 3.406233161687851, "ce_loss": 3.3904989659786224, "logit_z_loss": 21.169355154037476, "router_aux_loss": 1.4618848413228989, "router_z_loss": 1.1153101176023483, "routing_entropy_normalized": 0.6702843735150498, "learning_rate": 0.003, "grad_norm": 0.9744684100151062, "tokens_per_second": 4044549.8231484573, "eta_seconds": 7845.102952718735, "data_wait_seconds": 1.895427703857422e-05} {"step": 13490, "total_steps": 28610, "tokens": 28290580480, "source": "fineweb_hq", "phase": 3, "loss": 3.3505100905895233, "ce_loss": 3.335672676563263, "logit_z_loss": 21.660815477371216, "router_aux_loss": 1.3704904466867447, "router_z_loss": 1.1325490325689316, "routing_entropy_normalized": 0.6852916806725701, "learning_rate": 0.003, "grad_norm": 1.3538322448730469, "tokens_per_second": 3598636.2729159542, "eta_seconds": 8811.376264572144, "data_wait_seconds": 1.8268823623657227e-05} {"step": 13500, "total_steps": 28610, "tokens": 28311552000, "source": "fineweb_edu", "phase": 3, "loss": 3.427902191877365, "ce_loss": 3.413735806941986, "logit_z_loss": 20.848340034484863, "router_aux_loss": 1.2968240678310394, "router_z_loss": 1.1981367766857147, "routing_entropy_normalized": 0.7173876727556012, "learning_rate": 0.003, "grad_norm": 1.6099190711975098, "tokens_per_second": 3100221.4302365743, "eta_seconds": 10221.194657564163, "data_wait_seconds": 0.19338753819465637} {"step": 13510, "total_steps": 28610, "tokens": 28332523520, "source": "finemath", "phase": 3, "loss": 2.3750511705875397, "ce_loss": 2.360885053873062, "logit_z_loss": 25.91427969932556, "router_aux_loss": 1.3253588527441025, "router_z_loss": 0.9125107079744339, "routing_entropy_normalized": 0.7005350989879782, "learning_rate": 0.003, "grad_norm": 1.10991370677948, "tokens_per_second": 3903524.9408055963, "eta_seconds": 8112.410111427307, "data_wait_seconds": 1.8328428268432617e-05} {"step": 13520, "total_steps": 28610, "tokens": 28353495040, "source": "dclm", "phase": 3, "loss": 3.49660387635231, "ce_loss": 3.48104590177536, "logit_z_loss": 21.598166465759277, "router_aux_loss": 1.4601337760686874, "router_z_loss": 0.9566319137811661, "routing_entropy_normalized": 0.6381826394867266, "learning_rate": 0.003, "grad_norm": 1.660614252090454, "tokens_per_second": 3651055.6724099624, "eta_seconds": 8667.636574029922, "data_wait_seconds": 1.710653305053711e-05} {"step": 13530, "total_steps": 28610, "tokens": 28374466560, "source": "fineweb_hq", "phase": 3, "loss": 3.3823690712451935, "ce_loss": 3.3671305179595947, "logit_z_loss": 22.273242235183716, "router_aux_loss": 1.40936179459095, "router_z_loss": 1.144916370511055, "routing_entropy_normalized": 0.6774745755692351, "learning_rate": 0.003, "grad_norm": 1.4181253910064697, "tokens_per_second": 3772547.400363621, "eta_seconds": 8382.943619728088, "data_wait_seconds": 2.065300941467285e-05} {"step": 13540, "total_steps": 28610, "tokens": 28395438080, "source": "dclm", "phase": 3, "loss": 3.514193445444107, "ce_loss": 3.4987389147281647, "logit_z_loss": 21.264836072921753, "router_aux_loss": 1.437833532691002, "router_z_loss": 1.076200544834137, "routing_entropy_normalized": 0.6689258716681785, "learning_rate": 0.003, "grad_norm": 1.0399152040481567, "tokens_per_second": 4322964.400798059, "eta_seconds": 7310.742747306824, "data_wait_seconds": 2.2113323211669922e-05} {"step": 13550, "total_steps": 28610, "tokens": 28416409600, "source": "fineweb_hq", "phase": 3, "loss": 3.3085024058818817, "ce_loss": 3.293456107378006, "logit_z_loss": 21.92650294303894, "router_aux_loss": 1.3807301968336105, "router_z_loss": 1.2390004843473434, "routing_entropy_normalized": 0.6805769535936749, "learning_rate": 0.003, "grad_norm": 0.8856208920478821, "tokens_per_second": 4089085.6282051695, "eta_seconds": 7723.758314609528, "data_wait_seconds": 0.030563712120056152} {"step": 13560, "total_steps": 28610, "tokens": 28437381120, "source": "fineweb_edu", "phase": 3, "loss": 3.382246345281601, "ce_loss": 3.3677651286125183, "logit_z_loss": 20.752132415771484, "router_aux_loss": 1.3208072632551193, "router_z_loss": 1.273164615035057, "routing_entropy_normalized": 0.6930798498531051, "learning_rate": 0.003, "grad_norm": 1.1435730457305908, "tokens_per_second": 3262379.68283225, "eta_seconds": 9674.575208425522, "data_wait_seconds": 2.232193946838379e-05} {"step": 13570, "total_steps": 28610, "tokens": 28458352640, "source": "dclm", "phase": 3, "loss": 3.5102509260177612, "ce_loss": 3.4956784546375275, "logit_z_loss": 22.52889394760132, "router_aux_loss": 1.3503012359142303, "router_z_loss": 1.0694769993424416, "routing_entropy_normalized": 0.6631985276637242, "learning_rate": 0.003, "grad_norm": 0.9800899028778076, "tokens_per_second": 4156766.579330125, "eta_seconds": 7587.908889770508, "data_wait_seconds": 1.4901161193847656e-05} {"step": 13580, "total_steps": 28610, "tokens": 28479324160, "source": "fineweb_edu", "phase": 3, "loss": 3.390662342309952, "ce_loss": 3.376343309879303, "logit_z_loss": 20.739982843399048, "router_aux_loss": 1.3147089332342148, "router_z_loss": 1.1718716621398926, "routing_entropy_normalized": 0.6853678735136447, "learning_rate": 0.003, "grad_norm": 1.3886010646820068, "tokens_per_second": 4110399.135445846, "eta_seconds": 7668.402391433716, "data_wait_seconds": 2.0951032638549805e-05} {"step": 13590, "total_steps": 28610, "tokens": 28500295680, "source": "fineweb_hq", "phase": 3, "loss": 3.3545453250408173, "ce_loss": 3.340483397245407, "logit_z_loss": 21.07937979698181, "router_aux_loss": 1.2950834035873413, "router_z_loss": 1.1111164540052414, "routing_entropy_normalized": 0.7050639476579842, "learning_rate": 0.003, "grad_norm": 0.9677145481109619, "tokens_per_second": 3572846.661553331, "eta_seconds": 8816.28181219101, "data_wait_seconds": 1.8894672393798828e-05} {"step": 13600, "total_steps": 28610, "tokens": 28521267200, "source": "fineweb_edu", "phase": 3, "loss": 3.384620636701584, "ce_loss": 3.370360255241394, "logit_z_loss": 21.029226303100586, "router_aux_loss": 1.3224663436412811, "router_z_loss": 1.03568185120821, "routing_entropy_normalized": 0.6652621852975729, "learning_rate": 0.003, "grad_norm": 1.0374797582626343, "tokens_per_second": 3449222.489009641, "eta_seconds": 9126.187603235245, "data_wait_seconds": 1.850724220275879e-05} {"step": 13610, "total_steps": 28610, "tokens": 28542238720, "source": "finemath", "phase": 3, "loss": 2.375486582517624, "ce_loss": 2.361254036426544, "logit_z_loss": 26.58552575111389, "router_aux_loss": 1.3302761018276215, "router_z_loss": 0.9297791570425034, "routing_entropy_normalized": 0.6774906725105314, "learning_rate": 0.003, "grad_norm": 1.143229365348816, "tokens_per_second": 3229974.2784284027, "eta_seconds": 9739.173531532288, "data_wait_seconds": 1.963973045349121e-05} {"step": 13620, "total_steps": 28610, "tokens": 28563210240, "source": "dclm", "phase": 3, "loss": 3.464107871055603, "ce_loss": 3.4498995542526245, "logit_z_loss": 21.576688766479492, "router_aux_loss": 1.338112160563469, "router_z_loss": 0.8272264525294304, "routing_entropy_normalized": 0.6554607718886843, "learning_rate": 0.003, "grad_norm": 0.9077800512313843, "tokens_per_second": 3728673.7029200047, "eta_seconds": 8430.962584733963, "data_wait_seconds": 1.817941665649414e-05} {"step": 13630, "total_steps": 28610, "tokens": 28584181760, "source": "fineweb_hq", "phase": 3, "loss": 3.3487330079078674, "ce_loss": 3.3346027731895447, "logit_z_loss": 21.863550662994385, "router_aux_loss": 1.3190488070249557, "router_z_loss": 0.9396828636527061, "routing_entropy_normalized": 0.6578615458522572, "learning_rate": 0.003, "grad_norm": 1.0647000074386597, "tokens_per_second": 4332135.452863186, "eta_seconds": 7251.697760105133, "data_wait_seconds": 1.9937753677368164e-05} {"step": 13640, "total_steps": 28610, "tokens": 28605153280, "source": "dclm", "phase": 3, "loss": 3.483993172645569, "ce_loss": 3.4692509174346924, "logit_z_loss": 21.60749626159668, "router_aux_loss": 1.3879733234643936, "router_z_loss": 0.8625714853405952, "routing_entropy_normalized": 0.6519555112263966, "learning_rate": 0.003, "grad_norm": 1.0442997217178345, "tokens_per_second": 3500301.4470258695, "eta_seconds": 8969.046213626862, "data_wait_seconds": 1.9490718841552734e-05} {"step": 13650, "total_steps": 28610, "tokens": 28626124800, "source": "fineweb_hq", "phase": 3, "loss": 3.3393953442573547, "ce_loss": 3.3254992365837097, "logit_z_loss": 21.064223766326904, "router_aux_loss": 1.295323133468628, "router_z_loss": 0.9429001286625862, "routing_entropy_normalized": 0.6915159978932639, "learning_rate": 0.003, "grad_norm": 0.6737979054450989, "tokens_per_second": 4103793.0582097946, "eta_seconds": 7644.974655151367, "data_wait_seconds": 2.092123031616211e-05} {"step": 13660, "total_steps": 28610, "tokens": 28647096320, "source": "fineweb_edu", "phase": 3, "loss": 3.4150311648845673, "ce_loss": 3.400568425655365, "logit_z_loss": 21.13396143913269, "router_aux_loss": 1.345296859741211, "router_z_loss": 1.0097860172390938, "routing_entropy_normalized": 0.6723843371703183, "learning_rate": 0.003, "grad_norm": 1.310072660446167, "tokens_per_second": 3571853.2744841943, "eta_seconds": 8777.634463310242, "data_wait_seconds": 2.911686897277832e-05} {"step": 13670, "total_steps": 28610, "tokens": 28668067840, "source": "dclm", "phase": 3, "loss": 3.4864006340503693, "ce_loss": 3.4716310799121857, "logit_z_loss": 21.825578212738037, "router_aux_loss": 1.3895644694566727, "router_z_loss": 0.8738812804222107, "routing_entropy_normalized": 0.6590934930377038, "learning_rate": 0.003, "grad_norm": 0.5975948572158813, "tokens_per_second": 3878068.494627856, "eta_seconds": 8079.138087272644, "data_wait_seconds": 1.8835067749023438e-05} {"step": 13680, "total_steps": 28610, "tokens": 28689039360, "source": "fineweb_edu", "phase": 3, "loss": 3.364520341157913, "ce_loss": 3.349531263113022, "logit_z_loss": 20.620715379714966, "router_aux_loss": 1.4056688100099564, "router_z_loss": 0.9323383048176765, "routing_entropy_normalized": 0.6764232202160178, "learning_rate": 0.003, "grad_norm": 1.5044610500335693, "tokens_per_second": 4492359.556877032, "eta_seconds": 6969.718020915985, "data_wait_seconds": 2.282857894897461e-05} {"step": 13690, "total_steps": 28610, "tokens": 28710010880, "source": "fineweb_hq", "phase": 3, "loss": 3.3414293825626373, "ce_loss": 3.326912581920624, "logit_z_loss": 20.998328924179077, "router_aux_loss": 1.3514725416898727, "router_z_loss": 1.0020396336913109, "routing_entropy_normalized": 0.6884676346235256, "learning_rate": 0.003, "grad_norm": 0.813776969909668, "tokens_per_second": 3399463.3349990374, "eta_seconds": 9204.24924659729, "data_wait_seconds": 1.823902130126953e-05} {"step": 13700, "total_steps": 28610, "tokens": 28730982400, "source": "fineweb_edu", "phase": 3, "loss": 3.401266872882843, "ce_loss": 3.387126088142395, "logit_z_loss": 20.425689935684204, "router_aux_loss": 1.3120128512382507, "router_z_loss": 1.0206168368458748, "routing_entropy_normalized": 0.681651414332064, "learning_rate": 0.003, "grad_norm": 1.5455849170684814, "tokens_per_second": 3269670.112958176, "eta_seconds": 9563.208287000656, "data_wait_seconds": 2.2411346435546875e-05} {"step": 13710, "total_steps": 28610, "tokens": 28751953920, "source": "finemath", "phase": 3, "loss": 2.3497586250305176, "ce_loss": 2.335334599018097, "logit_z_loss": 26.24391007423401, "router_aux_loss": 1.3690327256917953, "router_z_loss": 0.7336643934249878, "routing_entropy_normalized": 0.6885160081508148, "learning_rate": 0.003, "grad_norm": 1.1838819980621338, "tokens_per_second": 4218602.353936035, "eta_seconds": 7407.089405059814, "data_wait_seconds": 1.7315149307250977e-05} {"step": 13720, "total_steps": 28610, "tokens": 28772925440, "source": "dclm", "phase": 3, "loss": 3.5195126831531525, "ce_loss": 3.5055396258831024, "logit_z_loss": 21.757044315338135, "router_aux_loss": 1.3135891556739807, "router_z_loss": 0.8371134176850319, "routing_entropy_normalized": 0.6939719719959491, "learning_rate": 0.003, "grad_norm": 0.7960250377655029, "tokens_per_second": 3140014.695444642, "eta_seconds": 9944.728387832642, "data_wait_seconds": 2.110004425048828e-05} {"step": 13730, "total_steps": 28610, "tokens": 28793896960, "source": "fineweb_hq", "phase": 3, "loss": 3.3033815026283264, "ce_loss": 3.2892849147319794, "logit_z_loss": 21.764779329299927, "router_aux_loss": 1.3118824064731598, "router_z_loss": 0.977735847234726, "routing_entropy_normalized": 0.6664103750614323, "learning_rate": 0.003, "grad_norm": 1.200577974319458, "tokens_per_second": 4327009.7752728565, "eta_seconds": 7211.821414947511, "data_wait_seconds": 2.1517276763916016e-05} {"step": 13740, "total_steps": 28610, "tokens": 28814868480, "source": "dclm", "phase": 3, "loss": 3.625689744949341, "ce_loss": 3.611075669527054, "logit_z_loss": 20.495661735534668, "router_aux_loss": 1.3778517246246338, "router_z_loss": 0.8355282247066498, "routing_entropy_normalized": 0.6639124553113138, "learning_rate": 0.003, "grad_norm": 1.9937769174575806, "tokens_per_second": 3753329.4219673527, "eta_seconds": 8308.530036687851, "data_wait_seconds": 1.7434358596801758e-05} {"step": 13750, "total_steps": 28610, "tokens": 28835840000, "source": "fineweb_hq", "phase": 3, "loss": 3.3635929822921753, "ce_loss": 3.3491687774658203, "logit_z_loss": 21.18685293197632, "router_aux_loss": 1.3431631922721863, "router_z_loss": 0.9925665259361267, "routing_entropy_normalized": 0.6735287877008098, "learning_rate": 0.003, "grad_norm": 1.0971355438232422, "tokens_per_second": 4040850.2960768975, "eta_seconds": 7712.158688545227, "data_wait_seconds": 2.1338462829589844e-05} {"step": 13760, "total_steps": 28610, "tokens": 28856811520, "source": "fineweb_edu", "phase": 3, "loss": 3.351952612400055, "ce_loss": 3.338177740573883, "logit_z_loss": 20.803531646728516, "router_aux_loss": 1.278753861784935, "router_z_loss": 0.987333133816719, "routing_entropy_normalized": 0.6978690779426805, "learning_rate": 0.003, "grad_norm": 1.1400799751281738, "tokens_per_second": 4209435.391859787, "eta_seconds": 7398.309820890427, "data_wait_seconds": 1.9848346710205078e-05} {"step": 13770, "total_steps": 28610, "tokens": 28877783040, "source": "dclm", "phase": 3, "loss": 3.490436464548111, "ce_loss": 3.47637403011322, "logit_z_loss": 21.24942946434021, "router_aux_loss": 1.3181255757808685, "router_z_loss": 0.881211668252945, "routing_entropy_normalized": 0.6987969149249752, "learning_rate": 0.003, "grad_norm": 1.3815020322799683, "tokens_per_second": 3500998.5948423464, "eta_seconds": 8889.388223648071, "data_wait_seconds": 1.6897916793823242e-05} {"step": 13780, "total_steps": 28610, "tokens": 28898754560, "source": "fineweb_edu", "phase": 3, "loss": 3.42870432138443, "ce_loss": 3.4140455424785614, "logit_z_loss": 20.906208753585815, "router_aux_loss": 1.3730121701955795, "router_z_loss": 0.9286395907402039, "routing_entropy_normalized": 0.6608531283528306, "learning_rate": 0.003, "grad_norm": 1.1257672309875488, "tokens_per_second": 4419692.174513288, "eta_seconds": 7036.862055540084, "data_wait_seconds": 2.0235776901245117e-05} {"step": 13790, "total_steps": 28610, "tokens": 28919726080, "source": "fineweb_hq", "phase": 3, "loss": 3.4003584682941437, "ce_loss": 3.3861327171325684, "logit_z_loss": 21.844725370407104, "router_aux_loss": 1.3213095515966415, "router_z_loss": 1.0126313418149948, "routing_entropy_normalized": 0.6927515653198729, "learning_rate": 0.003, "grad_norm": 1.2892645597457886, "tokens_per_second": 3828843.7128352313, "eta_seconds": 8117.278993606567, "data_wait_seconds": 1.633167266845703e-05} {"step": 13800, "total_steps": 28610, "tokens": 28940697600, "source": "fineweb_edu", "phase": 3, "loss": 3.4329220950603485, "ce_loss": 3.418355017900467, "logit_z_loss": 21.87982964515686, "router_aux_loss": 1.3646173626184464, "router_z_loss": 0.9208982884883881, "routing_entropy_normalized": 0.6869960947944084, "learning_rate": 0.003, "grad_norm": 1.3591324090957642, "tokens_per_second": 3378461.4689922086, "eta_seconds": 9193.184946775436, "data_wait_seconds": 1.755356788635254e-05} {"step": 13810, "total_steps": 28610, "tokens": 28961669120, "source": "finemath", "phase": 3, "loss": 2.4255718886852264, "ce_loss": 2.411220222711563, "logit_z_loss": 27.272582292556763, "router_aux_loss": 1.3636850714683533, "router_z_loss": 0.7147823944687843, "routing_entropy_normalized": 0.687732436985321, "learning_rate": 0.003, "grad_norm": 1.2459851503372192, "tokens_per_second": 3898753.9509406663, "eta_seconds": 7960.966501235962, "data_wait_seconds": 1.7136335372924805e-05} {"step": 13820, "total_steps": 28610, "tokens": 28982640640, "source": "dclm", "phase": 3, "loss": 3.4657318890094757, "ce_loss": 3.4511276483535767, "logit_z_loss": 21.99720549583435, "router_aux_loss": 1.3734710961580276, "router_z_loss": 0.869538389146328, "routing_entropy_normalized": 0.6842271806590607, "learning_rate": 0.003, "grad_norm": 0.899606466293335, "tokens_per_second": 3890009.001502387, "eta_seconds": 7973.472058296204, "data_wait_seconds": 2.378225326538086e-05} {"step": 13830, "total_steps": 28610, "tokens": 29003612160, "source": "fineweb_hq", "phase": 3, "loss": 3.398229956626892, "ce_loss": 3.3839468359947205, "logit_z_loss": 22.06586718559265, "router_aux_loss": 1.3315190821886063, "router_z_loss": 0.9679568335413933, "routing_entropy_normalized": 0.6544150616615829, "learning_rate": 0.003, "grad_norm": 1.9527353048324585, "tokens_per_second": 3967120.970972807, "eta_seconds": 7813.199241161346, "data_wait_seconds": 1.9103288650512695e-05} {"step": 13840, "total_steps": 28610, "tokens": 29024583680, "source": "dclm", "phase": 3, "loss": 3.511452943086624, "ce_loss": 3.497302532196045, "logit_z_loss": 21.88300323486328, "router_aux_loss": 1.3412226736545563, "router_z_loss": 0.7382280305027962, "routing_entropy_normalized": 0.6875843917576525, "learning_rate": 0.003, "grad_norm": 0.576770544052124, "tokens_per_second": 3908946.256027228, "eta_seconds": 7924.113817691803, "data_wait_seconds": 1.8835067749023438e-05} {"step": 13850, "total_steps": 28610, "tokens": 29045555200, "source": "fineweb_hq", "phase": 3, "loss": 3.3567188680171967, "ce_loss": 3.3421741127967834, "logit_z_loss": 21.49945902824402, "router_aux_loss": 1.366484060883522, "router_z_loss": 0.8799102306365967, "routing_entropy_normalized": 0.6658793472425419, "learning_rate": 0.003, "grad_norm": 1.4908833503723145, "tokens_per_second": 3640800.4337655753, "eta_seconds": 8501.966554641724, "data_wait_seconds": 2.4437904357910156e-05} {"step": 13860, "total_steps": 28610, "tokens": 29066526720, "source": "fineweb_edu", "phase": 3, "loss": 3.431122750043869, "ce_loss": 3.4169985949993134, "logit_z_loss": 20.591660976409912, "router_aux_loss": 1.3169489949941635, "router_z_loss": 0.954624854028225, "routing_entropy_normalized": 0.6816984013147787, "learning_rate": 0.003, "grad_norm": 1.5993061065673828, "tokens_per_second": 3341182.7845090944, "eta_seconds": 9258.096307516098, "data_wait_seconds": 2.2292137145996094e-05} {"step": 13870, "total_steps": 28610, "tokens": 29087498240, "source": "dclm", "phase": 3, "loss": 3.4292393922805786, "ce_loss": 3.4148699045181274, "logit_z_loss": 21.698773860931396, "router_aux_loss": 1.3631785959005356, "router_z_loss": 0.7377324253320694, "routing_entropy_normalized": 0.6550848488480981, "learning_rate": 0.003, "grad_norm": 1.5075528621673584, "tokens_per_second": 4217728.090897888, "eta_seconds": 7329.0690660476685, "data_wait_seconds": 2.518296241760254e-05} {"step": 13880, "total_steps": 28610, "tokens": 29108469760, "source": "fineweb_edu", "phase": 3, "loss": 3.3914456963539124, "ce_loss": 3.3762451112270355, "logit_z_loss": 21.506072998046875, "router_aux_loss": 1.421620935201645, "router_z_loss": 0.9844032153487206, "routing_entropy_normalized": 0.6606379059737315, "learning_rate": 0.003, "grad_norm": 1.2502108812332153, "tokens_per_second": 4106425.6161536374, "eta_seconds": 7522.612570524216, "data_wait_seconds": 5.385279655456543e-05} {"step": 13890, "total_steps": 28610, "tokens": 29129441280, "source": "fineweb_hq", "phase": 3, "loss": 3.416212797164917, "ce_loss": 3.4014886915683746, "logit_z_loss": 21.806881427764893, "router_aux_loss": 1.3770435005426407, "router_z_loss": 0.9536323100328445, "routing_entropy_normalized": 0.7074202152497092, "learning_rate": 0.003, "grad_norm": 1.6898139715194702, "tokens_per_second": 2962870.838625389, "eta_seconds": 10418.975082397461, "data_wait_seconds": 1.811981201171875e-05} {"step": 13900, "total_steps": 28610, "tokens": 29150412800, "source": "fineweb_edu", "phase": 3, "loss": 3.3686652779579163, "ce_loss": 3.3538402020931244, "logit_z_loss": 21.677958726882935, "router_aux_loss": 1.379465326666832, "router_z_loss": 1.0304131880402565, "routing_entropy_normalized": 0.6393233900782876, "learning_rate": 0.003, "grad_norm": 1.2516112327575684, "tokens_per_second": 4347484.398057709, "eta_seconds": 7095.852013587952, "data_wait_seconds": 2.35140323638916e-05} {"step": 13910, "total_steps": 28610, "tokens": 29171384320, "source": "finemath", "phase": 3, "loss": 2.6567492485046387, "ce_loss": 2.6420031785964966, "logit_z_loss": 25.64532780647278, "router_aux_loss": 1.3850200772285461, "router_z_loss": 0.8957830220460892, "routing_entropy_normalized": 0.6947739820250959, "learning_rate": 0.003, "grad_norm": 1.0656160116195679, "tokens_per_second": 4066445.3763756817, "eta_seconds": 7581.101317405701, "data_wait_seconds": 1.6242265701293945e-05} {"step": 13920, "total_steps": 28610, "tokens": 29192355840, "source": "dclm", "phase": 3, "loss": 3.376681685447693, "ce_loss": 3.3622426092624664, "logit_z_loss": 22.84168004989624, "router_aux_loss": 1.356575384736061, "router_z_loss": 0.8733394891023636, "routing_entropy_normalized": 0.650072477060182, "learning_rate": 0.003, "grad_norm": 1.0053064823150635, "tokens_per_second": 3295767.386256668, "eta_seconds": 9347.493093252182, "data_wait_seconds": 1.9758939743041992e-05} {"step": 13930, "total_steps": 28610, "tokens": 29213327360, "source": "fineweb_hq", "phase": 3, "loss": 3.3639317452907562, "ce_loss": 3.3496037423610687, "logit_z_loss": 21.944323539733887, "router_aux_loss": 1.3404794484376907, "router_z_loss": 0.9231548756361008, "routing_entropy_normalized": 0.6550925595758126, "learning_rate": 0.003, "grad_norm": 1.0907316207885742, "tokens_per_second": 4422845.405291469, "eta_seconds": 6960.7206535339365, "data_wait_seconds": 1.7702579498291016e-05} {"step": 13940, "total_steps": 28610, "tokens": 29234298880, "source": "dclm", "phase": 3, "loss": 3.5035329163074493, "ce_loss": 3.4886660277843475, "logit_z_loss": 22.73365068435669, "router_aux_loss": 1.403126135468483, "router_z_loss": 0.8356234729290009, "routing_entropy_normalized": 0.6649501402483736, "learning_rate": 0.003, "grad_norm": 0.6681424975395203, "tokens_per_second": 3931635.5972887375, "eta_seconds": 7825.043567419052, "data_wait_seconds": 2.1249055862426758e-05} {"step": 13950, "total_steps": 28610, "tokens": 29255270400, "source": "fineweb_hq", "phase": 3, "loss": 3.3537973761558533, "ce_loss": 3.3383898735046387, "logit_z_loss": 21.738013744354248, "router_aux_loss": 1.4392170757055283, "router_z_loss": 1.015305981040001, "routing_entropy_normalized": 0.6664785621260935, "learning_rate": 0.003, "grad_norm": 1.5525416135787964, "tokens_per_second": 4154360.0587763, "eta_seconds": 7400.477542877197, "data_wait_seconds": 1.6987323760986328e-05} {"step": 13960, "total_steps": 28610, "tokens": 29276241920, "source": "fineweb_edu", "phase": 3, "loss": 3.3587626218795776, "ce_loss": 3.343358635902405, "logit_z_loss": 22.186508655548096, "router_aux_loss": 1.4350946247577667, "router_z_loss": 1.053040400147438, "routing_entropy_normalized": 0.6494138468003283, "learning_rate": 0.003, "grad_norm": 1.6757290363311768, "tokens_per_second": 3739001.3470243267, "eta_seconds": 8216.973985433578, "data_wait_seconds": 1.862645149230957e-05} {"step": 13970, "total_steps": 28610, "tokens": 29297213440, "source": "dclm", "phase": 3, "loss": 3.4540684819221497, "ce_loss": 3.4387223720550537, "logit_z_loss": 22.380903482437134, "router_aux_loss": 1.4371067136526108, "router_z_loss": 0.9750036895275116, "routing_entropy_normalized": 0.6660149031921342, "learning_rate": 0.003, "grad_norm": 0.8443905115127563, "tokens_per_second": 4332841.578193762, "eta_seconds": 7085.951499938965, "data_wait_seconds": 2.1725893020629883e-05} {"step": 13980, "total_steps": 28610, "tokens": 29318184960, "source": "fineweb_edu", "phase": 3, "loss": 3.3644410967826843, "ce_loss": 3.349423497915268, "logit_z_loss": 21.1924307346344, "router_aux_loss": 1.3886904567480087, "router_z_loss": 1.1306976079940796, "routing_entropy_normalized": 0.668315025991927, "learning_rate": 0.003, "grad_norm": 0.8339715600013733, "tokens_per_second": 3974283.558793224, "eta_seconds": 7719.965952634811, "data_wait_seconds": 1.671910285949707e-05} {"step": 13990, "total_steps": 28610, "tokens": 29339156480, "source": "fineweb_hq", "phase": 3, "loss": 3.3662655353546143, "ce_loss": 3.3525294065475464, "logit_z_loss": 21.467906713485718, "router_aux_loss": 1.2701292932033539, "router_z_loss": 1.034845620393753, "routing_entropy_normalized": 0.7102322957476885, "learning_rate": 0.003, "grad_norm": 0.9797188639640808, "tokens_per_second": 3309307.2575801937, "eta_seconds": 9264.88834476471, "data_wait_seconds": 2.0563602447509766e-05} {"step": 14000, "total_steps": 28610, "tokens": 29360128000, "source": "fineweb_edu", "phase": 3, "loss": 3.411621481180191, "ce_loss": 3.3977217972278595, "logit_z_loss": 21.42672324180603, "router_aux_loss": 1.2939790487289429, "router_z_loss": 0.9598874673247337, "routing_entropy_normalized": 0.6738211433031186, "learning_rate": 0.003, "grad_norm": 1.5042341947555542, "tokens_per_second": 4115282.757876512, "eta_seconds": 7445.269869089127, "data_wait_seconds": 2.1845102310180664e-05} {"step": 14010, "total_steps": 28610, "tokens": 29381099520, "source": "finemath", "phase": 3, "loss": 2.377407729625702, "ce_loss": 2.3631319999694824, "logit_z_loss": 28.68639636039734, "router_aux_loss": 1.3559152334928513, "router_z_loss": 0.716614618897438, "routing_entropy_normalized": 0.6746667566841634, "learning_rate": 0.003, "grad_norm": 1.0147219896316528, "tokens_per_second": 3982255.43771003, "eta_seconds": 7688.713011741638, "data_wait_seconds": 1.7374753952026367e-05} {"step": 14020, "total_steps": 28610, "tokens": 29402071040, "source": "dclm", "phase": 3, "loss": 3.4752277731895447, "ce_loss": 3.461318790912628, "logit_z_loss": 21.86373209953308, "router_aux_loss": 1.312749132514, "router_z_loss": 0.7815255001187325, "routing_entropy_normalized": 0.6755150108946351, "learning_rate": 0.003, "grad_norm": 0.940832793712616, "tokens_per_second": 3631284.7570086853, "eta_seconds": 8426.066730499268, "data_wait_seconds": 1.767277717590332e-05} {"step": 14030, "total_steps": 28610, "tokens": 29423042560, "source": "fineweb_hq", "phase": 3, "loss": 3.3772367537021637, "ce_loss": 3.36258327960968, "logit_z_loss": 22.414567947387695, "router_aux_loss": 1.3802957236766815, "router_z_loss": 0.8505029380321503, "routing_entropy_normalized": 0.6502481523048791, "learning_rate": 0.003, "grad_norm": 0.943657398223877, "tokens_per_second": 3925318.2726086862, "eta_seconds": 7789.553365230559, "data_wait_seconds": 1.9997358322143555e-05} {"step": 14040, "total_steps": 28610, "tokens": 29444014080, "source": "dclm", "phase": 3, "loss": 3.5268178284168243, "ce_loss": 3.511624902486801, "logit_z_loss": 22.82814073562622, "router_aux_loss": 1.4360620975494385, "router_z_loss": 0.832313172519207, "routing_entropy_normalized": 0.644504253581507, "learning_rate": 0.003, "grad_norm": 2.0789425373077393, "tokens_per_second": 4013191.664150692, "eta_seconds": 7613.766597032547, "data_wait_seconds": 2.1517276763916016e-05} {"step": 14050, "total_steps": 28610, "tokens": 29464985600, "source": "fineweb_hq", "phase": 3, "loss": 3.3610775470733643, "ce_loss": 3.3459342420101166, "logit_z_loss": 21.762784242630005, "router_aux_loss": 1.4157341718673706, "router_z_loss": 0.9859652072191238, "routing_entropy_normalized": 0.6631598625117423, "learning_rate": 0.003, "grad_norm": 0.6611217856407166, "tokens_per_second": 3982603.6055529113, "eta_seconds": 7666.977722167969, "data_wait_seconds": 1.9848346710205078e-05} {"step": 14060, "total_steps": 28610, "tokens": 29485957120, "source": "fineweb_edu", "phase": 3, "loss": 3.405145823955536, "ce_loss": 3.3899122178554535, "logit_z_loss": 21.27911114692688, "router_aux_loss": 1.417493462562561, "router_z_loss": 1.0586552917957306, "routing_entropy_normalized": 0.6639364612305081, "learning_rate": 0.003, "grad_norm": 1.1439353227615356, "tokens_per_second": 4151201.9302450656, "eta_seconds": 7350.536570549011, "data_wait_seconds": 2.136826515197754e-05} {"step": 14070, "total_steps": 28610, "tokens": 29506928640, "source": "dclm", "phase": 3, "loss": 3.459635019302368, "ce_loss": 3.4447139501571655, "logit_z_loss": 22.30227828025818, "router_aux_loss": 1.4029956012964249, "router_z_loss": 0.8910934403538704, "routing_entropy_normalized": 0.6834630907540158, "learning_rate": 0.003, "grad_norm": 1.5017790794372559, "tokens_per_second": 3808983.7466060086, "eta_seconds": 8005.439799308777, "data_wait_seconds": 0.009174913167953491} {"step": 14080, "total_steps": 28610, "tokens": 29527900160, "source": "fineweb_edu", "phase": 3, "loss": 3.3800081610679626, "ce_loss": 3.3649795055389404, "logit_z_loss": 21.42039442062378, "router_aux_loss": 1.3930396139621735, "router_z_loss": 1.0981968194246292, "routing_entropy_normalized": 0.6424531542923884, "learning_rate": 0.003, "grad_norm": 1.1338999271392822, "tokens_per_second": 4294653.16560732, "eta_seconds": 7095.245502948761, "data_wait_seconds": 1.940131187438965e-05} {"step": 14090, "total_steps": 28610, "tokens": 29548871680, "source": "fineweb_hq", "phase": 3, "loss": 3.381427764892578, "ce_loss": 3.3672468960285187, "logit_z_loss": 22.520023822784424, "router_aux_loss": 1.3158408552408218, "router_z_loss": 1.0225140005350113, "routing_entropy_normalized": 0.6906528106433323, "learning_rate": 0.003, "grad_norm": 1.1403335332870483, "tokens_per_second": 3531711.783328963, "eta_seconds": 8622.064570426941, "data_wait_seconds": 2.1785497665405273e-05} {"step": 14100, "total_steps": 28610, "tokens": 29569843200, "source": "fineweb_edu", "phase": 3, "loss": 3.4002537429332733, "ce_loss": 3.386228561401367, "logit_z_loss": 22.556612730026245, "router_aux_loss": 1.2939711511135101, "router_z_loss": 1.085480734705925, "routing_entropy_normalized": 0.6822386451073417, "learning_rate": 0.003, "grad_norm": 2.4755876064300537, "tokens_per_second": 4185670.408117107, "eta_seconds": 7269.964558362961, "data_wait_seconds": 2.3484230041503906e-05} {"step": 14110, "total_steps": 28610, "tokens": 29590814720, "source": "finemath", "phase": 3, "loss": 2.3883359134197235, "ce_loss": 2.374151736497879, "logit_z_loss": 26.793556690216064, "router_aux_loss": 1.3394018858671188, "router_z_loss": 0.7901860252022743, "routing_entropy_normalized": 0.7042909483032788, "learning_rate": 0.003, "grad_norm": 0.8584123849868774, "tokens_per_second": 3982833.1663836315, "eta_seconds": 7634.942948818207, "data_wait_seconds": 1.7851591110229492e-05} {"step": 14120, "total_steps": 28610, "tokens": 29611786240, "source": "dclm", "phase": 3, "loss": 3.4676662981510162, "ce_loss": 3.453585773706436, "logit_z_loss": 21.510843992233276, "router_aux_loss": 1.3196000903844833, "router_z_loss": 0.8845625519752502, "routing_entropy_normalized": 0.7075201102865692, "learning_rate": 0.003, "grad_norm": 0.715420126914978, "tokens_per_second": 4210474.301950138, "eta_seconds": 7217.175619840623, "data_wait_seconds": 0.007760584354400635} {"step": 14130, "total_steps": 28610, "tokens": 29632757760, "source": "fineweb_hq", "phase": 3, "loss": 3.4104469418525696, "ce_loss": 3.396637588739395, "logit_z_loss": 22.744064569473267, "router_aux_loss": 1.2788185328245163, "router_z_loss": 1.0212053284049034, "routing_entropy_normalized": 0.6926016361379117, "learning_rate": 0.003, "grad_norm": 1.3888428211212158, "tokens_per_second": 3969640.703120815, "eta_seconds": 7649.75050163269, "data_wait_seconds": 2.0503997802734375e-05} {"step": 14140, "total_steps": 28610, "tokens": 29653729280, "source": "dclm", "phase": 3, "loss": 3.515615016222, "ce_loss": 3.501961439847946, "logit_z_loss": 22.47020125389099, "router_aux_loss": 1.2737108170986176, "router_z_loss": 0.9164877012372017, "routing_entropy_normalized": 0.6934013974992789, "learning_rate": 0.003, "grad_norm": 0.8762441277503967, "tokens_per_second": 3894144.073310427, "eta_seconds": 7792.672502279282, "data_wait_seconds": 1.576542854309082e-05} {"step": 14150, "total_steps": 28610, "tokens": 29674700800, "source": "fineweb_hq", "phase": 3, "loss": 3.391709953546524, "ce_loss": 3.3768768906593323, "logit_z_loss": 21.7146577835083, "router_aux_loss": 1.3867763876914978, "router_z_loss": 0.9653165191411972, "routing_entropy_normalized": 0.653162209216026, "learning_rate": 0.003, "grad_norm": 1.2084603309631348, "tokens_per_second": 3943885.4268598803, "eta_seconds": 7689.071724414825, "data_wait_seconds": 1.8477439880371094e-05} {"step": 14160, "total_steps": 28610, "tokens": 29695672320, "source": "fineweb_edu", "phase": 3, "loss": 3.3942683041095734, "ce_loss": 3.379001557826996, "logit_z_loss": 22.498599767684937, "router_aux_loss": 1.4212849289178848, "router_z_loss": 1.0538768917322159, "routing_entropy_normalized": 0.6720912261726052, "learning_rate": 0.003, "grad_norm": 1.0625189542770386, "tokens_per_second": 4088145.081979084, "eta_seconds": 7412.615206241608, "data_wait_seconds": 1.8984079360961914e-05} {"step": 14170, "total_steps": 28610, "tokens": 29716643840, "source": "dclm", "phase": 3, "loss": 3.4562571346759796, "ce_loss": 3.441042900085449, "logit_z_loss": 22.425044298171997, "router_aux_loss": 1.433167353272438, "router_z_loss": 0.8825905844569206, "routing_entropy_normalized": 0.6501666430641578, "learning_rate": 0.003, "grad_norm": 1.0722579956054688, "tokens_per_second": 4012340.2434966657, "eta_seconds": 7547.434425354004, "data_wait_seconds": 1.8030405044555664e-05} {"step": 14180, "total_steps": 28610, "tokens": 29737615360, "source": "fineweb_edu", "phase": 3, "loss": 3.372035652399063, "ce_loss": 3.3576778173446655, "logit_z_loss": 22.32611632347107, "router_aux_loss": 1.3279561400413513, "router_z_loss": 1.0782094895839691, "routing_entropy_normalized": 0.6840553350538717, "learning_rate": 0.003, "grad_norm": 1.0697839260101318, "tokens_per_second": 4176416.0159245306, "eta_seconds": 7245.902526140213, "data_wait_seconds": 2.0772218704223633e-05} {"step": 14190, "total_steps": 28610, "tokens": 29758586880, "source": "fineweb_hq", "phase": 3, "loss": 3.372597187757492, "ce_loss": 3.358408898115158, "logit_z_loss": 22.32690453529358, "router_aux_loss": 1.318288341164589, "router_z_loss": 1.0054062232375145, "routing_entropy_normalized": 0.6780537618174675, "learning_rate": 0.003, "grad_norm": 0.8440964221954346, "tokens_per_second": 3425720.094524884, "eta_seconds": 8827.613174915314, "data_wait_seconds": 2.1666288375854492e-05} {"step": 14200, "total_steps": 28610, "tokens": 29779558400, "source": "fineweb_edu", "phase": 3, "loss": 3.388155549764633, "ce_loss": 3.3740772008895874, "logit_z_loss": 22.79537343978882, "router_aux_loss": 1.3038922250270844, "router_z_loss": 1.0394416451454163, "routing_entropy_normalized": 0.6467596107853996, "learning_rate": 0.003, "grad_norm": 1.720811367034912, "tokens_per_second": 3238619.5711245583, "eta_seconds": 9331.123849630356, "data_wait_seconds": 0.08482861518859863} {"step": 14210, "total_steps": 28610, "tokens": 29800529920, "source": "finemath", "phase": 3, "loss": 2.404615044593811, "ce_loss": 2.3895705938339233, "logit_z_loss": 28.289714097976685, "router_aux_loss": 1.4352821558713913, "router_z_loss": 0.6916430592536926, "routing_entropy_normalized": 0.6709964566099439, "learning_rate": 0.003, "grad_norm": 0.6674666404724121, "tokens_per_second": 4459077.45836895, "eta_seconds": 6772.474594116212, "data_wait_seconds": 2.1159648895263672e-05} {"step": 14220, "total_steps": 28610, "tokens": 29821501440, "source": "dclm", "phase": 3, "loss": 3.447323441505432, "ce_loss": 3.4315415024757385, "logit_z_loss": 21.934906244277954, "router_aux_loss": 1.484549030661583, "router_z_loss": 0.9364329725503922, "routing_entropy_normalized": 0.6518665502731458, "learning_rate": 0.003, "grad_norm": 0.768263041973114, "tokens_per_second": 4232445.112541383, "eta_seconds": 7130.161520719527, "data_wait_seconds": 1.9162893295288086e-05} {"step": 14230, "total_steps": 28610, "tokens": 29842472960, "source": "fineweb_hq", "phase": 3, "loss": 3.3627163469791412, "ce_loss": 3.3472376465797424, "logit_z_loss": 21.30533003807068, "router_aux_loss": 1.4508202970027924, "router_z_loss": 0.9704894796013832, "routing_entropy_normalized": 0.6623242189988566, "learning_rate": 0.003, "grad_norm": 1.95541250705719, "tokens_per_second": 3286643.396688357, "eta_seconds": 9175.63669681549, "data_wait_seconds": 1.77919864654541e-05} {"step": 14240, "total_steps": 28610, "tokens": 29863444480, "source": "dclm", "phase": 3, "loss": 3.516104906797409, "ce_loss": 3.500605434179306, "logit_z_loss": 21.62199091911316, "router_aux_loss": 1.4579098969697952, "router_z_loss": 0.9203964099287987, "routing_entropy_normalized": 0.6539004168002429, "learning_rate": 0.003, "grad_norm": 0.6229979991912842, "tokens_per_second": 4269262.914508963, "eta_seconds": 7058.84712266922, "data_wait_seconds": 2.142786979675293e-05} {"step": 14250, "total_steps": 28610, "tokens": 29884416000, "source": "fineweb_hq", "phase": 3, "loss": 3.396450161933899, "ce_loss": 3.3810462951660156, "logit_z_loss": 22.218289136886597, "router_aux_loss": 1.4277794808149338, "router_z_loss": 1.1261222809553146, "routing_entropy_normalized": 0.6617195361764011, "learning_rate": 0.003, "grad_norm": 0.8149000406265259, "tokens_per_second": 3483291.494363372, "eta_seconds": 8645.587878227234, "data_wait_seconds": 0.08512663841247559} {"step": 14260, "total_steps": 28610, "tokens": 29905387520, "source": "fineweb_edu", "phase": 3, "loss": 3.3735596239566803, "ce_loss": 3.357980489730835, "logit_z_loss": 21.997942686080933, "router_aux_loss": 1.4491279423236847, "router_z_loss": 1.0878178402781487, "routing_entropy_normalized": 0.6484250005719666, "learning_rate": 0.003, "grad_norm": 1.92019784450531, "tokens_per_second": 4460296.18883384, "eta_seconds": 6747.114973068238, "data_wait_seconds": 1.963973045349121e-05} {"step": 14270, "total_steps": 28610, "tokens": 29926359040, "source": "dclm", "phase": 3, "loss": 3.4041945338249207, "ce_loss": 3.3892721831798553, "logit_z_loss": 22.91644525527954, "router_aux_loss": 1.3925887048244476, "router_z_loss": 0.9964710772037506, "routing_entropy_normalized": 0.6597674060296377, "learning_rate": 0.003, "grad_norm": 0.5769676566123962, "tokens_per_second": 3840366.029906425, "eta_seconds": 7830.805565357208, "data_wait_seconds": 2.002716064453125e-05} {"step": 14280, "total_steps": 28610, "tokens": 29947330560, "source": "fineweb_edu", "phase": 3, "loss": 3.3976023197174072, "ce_loss": 3.3810683488845825, "logit_z_loss": 22.244950532913208, "router_aux_loss": 1.5371333211660385, "router_z_loss": 1.162653312087059, "routing_entropy_normalized": 0.6394607877734478, "learning_rate": 0.003, "grad_norm": 1.0206918716430664, "tokens_per_second": 4327906.510634724, "eta_seconds": 6943.816389322281, "data_wait_seconds": 1.868605613708496e-05} {"step": 14290, "total_steps": 28610, "tokens": 29968302080, "source": "fineweb_hq", "phase": 3, "loss": 3.3460720479488373, "ce_loss": 3.3304028511047363, "logit_z_loss": 21.757296800613403, "router_aux_loss": 1.4369151145219803, "router_z_loss": 1.3000391721725464, "routing_entropy_normalized": 0.6535340401420067, "learning_rate": 0.003, "grad_norm": 1.3251090049743652, "tokens_per_second": 3267701.2626760467, "eta_seconds": 9190.31888961792, "data_wait_seconds": 2.7358531951904297e-05} {"step": 14300, "total_steps": 28610, "tokens": 29989273600, "source": "fineweb_edu", "phase": 3, "loss": 3.3525536954402924, "ce_loss": 3.3375827372074127, "logit_z_loss": 22.589349508285522, "router_aux_loss": 1.3622078448534012, "router_z_loss": 1.3488937318325043, "routing_entropy_normalized": 0.6588634912762558, "learning_rate": 0.003, "grad_norm": 1.2493332624435425, "tokens_per_second": 4610340.714102566, "eta_seconds": 6509.3334703445435, "data_wait_seconds": 2.3066997528076172e-05} {"step": 14310, "total_steps": 28610, "tokens": 30010245120, "source": "finemath", "phase": 3, "loss": 2.4090414345264435, "ce_loss": 2.3947006464004517, "logit_z_loss": 26.296433687210083, "router_aux_loss": 1.3360605984926224, "router_z_loss": 0.9802162349224091, "routing_entropy_normalized": 0.6765567195582274, "learning_rate": 0.003, "grad_norm": 0.8078435063362122, "tokens_per_second": 4523868.8761565415, "eta_seconds": 6629.120874404908, "data_wait_seconds": 2.0235776901245117e-05} {"step": 14320, "total_steps": 28610, "tokens": 30031216640, "source": "dclm", "phase": 3, "loss": 3.4601375460624695, "ce_loss": 3.445100575685501, "logit_z_loss": 22.39435648918152, "router_aux_loss": 1.3990516811609268, "router_z_loss": 1.0464372634887695, "routing_entropy_normalized": 0.668173795184129, "learning_rate": 0.003, "grad_norm": 0.6249284148216248, "tokens_per_second": 4255136.887703113, "eta_seconds": 7042.852643966675, "data_wait_seconds": 1.773238182067871e-05} {"step": 14330, "total_steps": 28610, "tokens": 30052188160, "source": "fineweb_hq", "phase": 3, "loss": 3.272730231285095, "ce_loss": 3.257653832435608, "logit_z_loss": 22.88171100616455, "router_aux_loss": 1.4004687517881393, "router_z_loss": 1.0717585384845734, "routing_entropy_normalized": 0.6679439333350857, "learning_rate": 0.003, "grad_norm": 1.435403823852539, "tokens_per_second": 3697251.9915426467, "eta_seconds": 8099.888952255249, "data_wait_seconds": 2.0772218704223633e-05} {"step": 14340, "total_steps": 28610, "tokens": 30073159680, "source": "dclm", "phase": 3, "loss": 3.453351616859436, "ce_loss": 3.438274919986725, "logit_z_loss": 22.029916763305664, "router_aux_loss": 1.4053995311260223, "router_z_loss": 1.0227026492357254, "routing_entropy_normalized": 0.6443916504604867, "learning_rate": 0.003, "grad_norm": 0.9923781752586365, "tokens_per_second": 3930723.571797687, "eta_seconds": 7613.447878837585, "data_wait_seconds": 2.4765729904174805e-05} {"step": 14350, "total_steps": 28610, "tokens": 30094131200, "source": "fineweb_hq", "phase": 3, "loss": 3.3167014122009277, "ce_loss": 3.302029073238373, "logit_z_loss": 22.827113389968872, "router_aux_loss": 1.349898800253868, "router_z_loss": 1.1733360439538956, "routing_entropy_normalized": 0.654287507800091, "learning_rate": 0.003, "grad_norm": 0.9376949667930603, "tokens_per_second": 4057075.671874791, "eta_seconds": 7371.168284416199, "data_wait_seconds": 1.8030405044555664e-05} {"step": 14360, "total_steps": 28610, "tokens": 30115102720, "source": "fineweb_edu", "phase": 3, "loss": 3.391855090856552, "ce_loss": 3.3771704733371735, "logit_z_loss": 21.055899381637573, "router_aux_loss": 1.3550678491592407, "router_z_loss": 1.1339837163686752, "routing_entropy_normalized": 0.6692332520489577, "learning_rate": 0.003, "grad_norm": 1.1150013208389282, "tokens_per_second": 4010275.8914838843, "eta_seconds": 7451.960116624832, "data_wait_seconds": 2.232193946838379e-05} {"step": 14370, "total_steps": 28610, "tokens": 30136074240, "source": "dclm", "phase": 3, "loss": 3.4661697149276733, "ce_loss": 3.451554775238037, "logit_z_loss": 22.337185859680176, "router_aux_loss": 1.3589355200529099, "router_z_loss": 1.0256000235676765, "routing_entropy_normalized": 0.6732474350639467, "learning_rate": 0.003, "grad_norm": 0.8797860145568848, "tokens_per_second": 3749398.6670116507, "eta_seconds": 7964.862403869629, "data_wait_seconds": 2.181529998779297e-05} {"step": 14380, "total_steps": 28610, "tokens": 30157045760, "source": "fineweb_edu", "phase": 3, "loss": 3.374334305524826, "ce_loss": 3.3599887788295746, "logit_z_loss": 21.67206573486328, "router_aux_loss": 1.3219271004199982, "router_z_loss": 1.126178815960884, "routing_entropy_normalized": 0.6685439228226258, "learning_rate": 0.003, "grad_norm": 1.0522563457489014, "tokens_per_second": 4131896.9782830616, "eta_seconds": 7222.462979316711, "data_wait_seconds": 2.2709369659423828e-05} {"step": 14390, "total_steps": 28610, "tokens": 30178017280, "source": "fineweb_hq", "phase": 3, "loss": 3.3574917018413544, "ce_loss": 3.342660039663315, "logit_z_loss": 21.862849950790405, "router_aux_loss": 1.3774042427539825, "router_z_loss": 1.0576132610440254, "routing_entropy_normalized": 0.667719595188173, "learning_rate": 0.003, "grad_norm": 0.49591293931007385, "tokens_per_second": 3430674.3829533635, "eta_seconds": 8692.606208324432, "data_wait_seconds": 2.3096799850463867e-05} {"step": 14400, "total_steps": 28610, "tokens": 30198988800, "source": "fineweb_edu", "phase": 3, "loss": 3.3953757286071777, "ce_loss": 3.380372256040573, "logit_z_loss": 23.33644723892212, "router_aux_loss": 1.3894157707691193, "router_z_loss": 1.1094134598970413, "routing_entropy_normalized": 0.6485496592024708, "learning_rate": 0.003, "grad_norm": 1.7723301649093628, "tokens_per_second": 4217007.632390514, "eta_seconds": 7066.747921228409, "data_wait_seconds": 0.00818932056427002} {"step": 14410, "total_steps": 28610, "tokens": 30219960320, "source": "finemath", "phase": 3, "loss": 2.4341732263565063, "ce_loss": 2.4189250767230988, "logit_z_loss": 28.3894362449646, "router_aux_loss": 1.4475878775119781, "router_z_loss": 0.772313691675663, "routing_entropy_normalized": 0.6584796010642471, "learning_rate": 0.003, "grad_norm": 1.2259362936019897, "tokens_per_second": 4001351.5209765085, "eta_seconds": 7442.374968528748, "data_wait_seconds": 1.615285873413086e-05} {"step": 14420, "total_steps": 28610, "tokens": 30240931840, "source": "dclm", "phase": 3, "loss": 3.5409070551395416, "ce_loss": 3.525301158428192, "logit_z_loss": 22.93383264541626, "router_aux_loss": 1.4588765799999237, "router_z_loss": 1.0170599594712257, "routing_entropy_normalized": 0.6492744043785393, "learning_rate": 0.003, "grad_norm": 1.6546868085861206, "tokens_per_second": 3704952.209405256, "eta_seconds": 8032.110860824585, "data_wait_seconds": 1.77919864654541e-05} {"step": 14430, "total_steps": 28610, "tokens": 30261903360, "source": "fineweb_hq", "phase": 3, "loss": 3.3369261026382446, "ce_loss": 3.321212947368622, "logit_z_loss": 22.828819274902344, "router_aux_loss": 1.4441170394420624, "router_z_loss": 1.2719593942165375, "routing_entropy_normalized": 0.6554556355696347, "learning_rate": 0.003, "grad_norm": 1.4452545642852783, "tokens_per_second": 3878077.7274810662, "eta_seconds": 7668.132886886597, "data_wait_seconds": 2.0772218704223633e-05} {"step": 14440, "total_steps": 28610, "tokens": 30282874880, "source": "dclm", "phase": 3, "loss": 3.4843445122241974, "ce_loss": 3.469306915998459, "logit_z_loss": 23.368987560272217, "router_aux_loss": 1.3895319104194641, "router_z_loss": 1.1422665417194366, "routing_entropy_normalized": 0.6945572118066144, "learning_rate": 0.003, "grad_norm": 1.433976411819458, "tokens_per_second": 3977541.6078322916, "eta_seconds": 7471.108229637146, "data_wait_seconds": 2.0265579223632812e-05} {"step": 14450, "total_steps": 28610, "tokens": 30303846400, "source": "fineweb_hq", "phase": 3, "loss": 3.3189307749271393, "ce_loss": 3.30428284406662, "logit_z_loss": 21.66116452217102, "router_aux_loss": 1.342431902885437, "router_z_loss": 1.223652794957161, "routing_entropy_normalized": 0.6716970967125422, "learning_rate": 0.003, "grad_norm": 1.2145382165908813, "tokens_per_second": 4204921.207899126, "eta_seconds": 7062.123367309569, "data_wait_seconds": 1.9878149032592773e-05} {"step": 14460, "total_steps": 28610, "tokens": 30324817920, "source": "fineweb_edu", "phase": 3, "loss": 3.3694266378879547, "ce_loss": 3.3548814058303833, "logit_z_loss": 22.335132122039795, "router_aux_loss": 1.3274559527635574, "router_z_loss": 1.2706312388181686, "routing_entropy_normalized": 0.6628204528495945, "learning_rate": 0.003, "grad_norm": 1.4269694089889526, "tokens_per_second": 3778077.418846179, "eta_seconds": 7854.444869756699, "data_wait_seconds": 1.9788742065429688e-05} {"step": 14470, "total_steps": 28610, "tokens": 30345789440, "source": "dclm", "phase": 3, "loss": 3.4810426235198975, "ce_loss": 3.466826468706131, "logit_z_loss": 22.59409499168396, "router_aux_loss": 1.31220743060112, "router_z_loss": 1.094059407711029, "routing_entropy_normalized": 0.6804725114901214, "learning_rate": 0.003, "grad_norm": 1.3233927488327026, "tokens_per_second": 3295171.666366723, "eta_seconds": 8999.145502090454, "data_wait_seconds": 2.3126602172851562e-05} {"step": 14480, "total_steps": 28610, "tokens": 30366760960, "source": "fineweb_edu", "phase": 3, "loss": 3.395291030406952, "ce_loss": 3.3805632889270782, "logit_z_loss": 21.87780237197876, "router_aux_loss": 1.354275420308113, "router_z_loss": 1.1849639117717743, "routing_entropy_normalized": 0.6778718060134721, "learning_rate": 0.003, "grad_norm": 0.8090845346450806, "tokens_per_second": 4143544.8094750317, "eta_seconds": 7151.547557115555, "data_wait_seconds": 1.8090009689331055e-05} {"step": 14490, "total_steps": 28610, "tokens": 30387732480, "source": "fineweb_hq", "phase": 3, "loss": 3.3210058212280273, "ce_loss": 3.3066276609897614, "logit_z_loss": 21.967010021209717, "router_aux_loss": 1.333407998085022, "router_z_loss": 1.044098362326622, "routing_entropy_normalized": 0.679226421290173, "learning_rate": 0.003, "grad_norm": 1.038711428642273, "tokens_per_second": 4107915.5286849122, "eta_seconds": 7208.470094680786, "data_wait_seconds": 1.773238182067871e-05} {"step": 14500, "total_steps": 28610, "tokens": 30408704000, "source": "fineweb_edu", "phase": 3, "loss": 3.299738198518753, "ce_loss": 3.285378247499466, "logit_z_loss": 21.974931478500366, "router_aux_loss": 1.3082813024520874, "router_z_loss": 1.277212679386139, "routing_entropy_normalized": 0.6429076274247035, "learning_rate": 0.003, "grad_norm": 1.893919587135315, "tokens_per_second": 3673033.8827144536, "eta_seconds": 8056.232440233231, "data_wait_seconds": 0.15923836827278137} {"step": 14510, "total_steps": 28610, "tokens": 30429675520, "source": "finemath", "phase": 3, "loss": 2.3692860305309296, "ce_loss": 2.354872941970825, "logit_z_loss": 27.1954402923584, "router_aux_loss": 1.3695016950368881, "router_z_loss": 0.7180856615304947, "routing_entropy_normalized": 0.7052843499805, "learning_rate": 0.003, "grad_norm": 1.4048250913619995, "tokens_per_second": 3444542.000473757, "eta_seconds": 8584.550049304962, "data_wait_seconds": 1.4483928680419922e-05} {"step": 14520, "total_steps": 28610, "tokens": 30450647040, "source": "dclm", "phase": 3, "loss": 3.4802541732788086, "ce_loss": 3.465826690196991, "logit_z_loss": 22.509546995162964, "router_aux_loss": 1.3489722907543182, "router_z_loss": 0.9377414658665657, "routing_entropy_normalized": 0.660353047228999, "learning_rate": 0.003, "grad_norm": 1.1259424686431885, "tokens_per_second": 3475530.258607378, "eta_seconds": 8501.975089073181, "data_wait_seconds": 0.09460803866386414} {"step": 14530, "total_steps": 28610, "tokens": 30471618560, "source": "fineweb_hq", "phase": 3, "loss": 3.321460396051407, "ce_loss": 3.306550830602646, "logit_z_loss": 23.167984008789062, "router_aux_loss": 1.4016375541687012, "router_z_loss": 0.8931695595383644, "routing_entropy_normalized": 0.646150187742322, "learning_rate": 0.003, "grad_norm": 0.9963974356651306, "tokens_per_second": 4244645.858699762, "eta_seconds": 6956.5050048828125, "data_wait_seconds": 2.0951032638549805e-05} {"step": 14540, "total_steps": 28610, "tokens": 30492590080, "source": "dclm", "phase": 3, "loss": 3.505112648010254, "ce_loss": 3.4898055493831635, "logit_z_loss": 22.463274717330933, "router_aux_loss": 1.4423689395189285, "router_z_loss": 0.8833800405263901, "routing_entropy_normalized": 0.6430520966122175, "learning_rate": 0.003, "grad_norm": 1.9480777978897095, "tokens_per_second": 3621108.8249053904, "eta_seconds": 8148.5893042087555, "data_wait_seconds": 2.142786979675293e-05} {"step": 14550, "total_steps": 28610, "tokens": 30513561600, "source": "fineweb_hq", "phase": 3, "loss": 3.309794068336487, "ce_loss": 3.2937550246715546, "logit_z_loss": 22.405506372451782, "router_aux_loss": 1.5069028735160828, "router_z_loss": 0.9700360223650932, "routing_entropy_normalized": 0.632502423739072, "learning_rate": 0.003, "grad_norm": 0.6886633038520813, "tokens_per_second": 3856641.193143424, "eta_seconds": 7645.501783370972, "data_wait_seconds": 0.04903596639633179} {"step": 14560, "total_steps": 28610, "tokens": 30534533120, "source": "fineweb_edu", "phase": 3, "loss": 3.3845606446266174, "ce_loss": 3.369443565607071, "logit_z_loss": 22.559802532196045, "router_aux_loss": 1.3949499279260635, "router_z_loss": 1.167557880282402, "routing_entropy_normalized": 0.6752415896564327, "learning_rate": 0.003, "grad_norm": 1.6304177045822144, "tokens_per_second": 3652657.7713491977, "eta_seconds": 8066.724956035614, "data_wait_seconds": 2.0235776901245117e-05} {"step": 14570, "total_steps": 28610, "tokens": 30555504640, "source": "dclm", "phase": 3, "loss": 3.4957252144813538, "ce_loss": 3.4816732108592987, "logit_z_loss": 22.472912788391113, "router_aux_loss": 1.2994818091392517, "router_z_loss": 1.057092159986496, "routing_entropy_normalized": 0.7027962258426367, "learning_rate": 0.003, "grad_norm": 0.8936728239059448, "tokens_per_second": 3587902.883645695, "eta_seconds": 8206.469080924988, "data_wait_seconds": 2.22623348236084e-05} {"step": 14580, "total_steps": 28610, "tokens": 30576476160, "source": "fineweb_edu", "phase": 3, "loss": 3.3710993826389313, "ce_loss": 3.3568695783615112, "logit_z_loss": 22.427814483642578, "router_aux_loss": 1.3027251213788986, "router_z_loss": 1.2025493681430817, "routing_entropy_normalized": 0.6665838637133933, "learning_rate": 0.003, "grad_norm": 1.249550461769104, "tokens_per_second": 4062389.168613785, "eta_seconds": 7242.792686462402, "data_wait_seconds": 2.0742416381835938e-05} {"step": 14590, "total_steps": 28610, "tokens": 30597447680, "source": "fineweb_hq", "phase": 3, "loss": 3.2950304746627808, "ce_loss": 3.280717968940735, "logit_z_loss": 22.327561616897583, "router_aux_loss": 1.3214615881443024, "router_z_loss": 1.0979322642087936, "routing_entropy_normalized": 0.6775955679334089, "learning_rate": 0.003, "grad_norm": 1.1605257987976074, "tokens_per_second": 3504819.303575543, "eta_seconds": 8389.040487766266, "data_wait_seconds": 0.07612285017967224} {"step": 14600, "total_steps": 28610, "tokens": 30618419200, "source": "fineweb_edu", "phase": 3, "loss": 3.4261245131492615, "ce_loss": 3.41101935505867, "logit_z_loss": 22.944186687469482, "router_aux_loss": 1.3992246687412262, "router_z_loss": 1.11292964220047, "routing_entropy_normalized": 0.6340112027014674, "learning_rate": 0.003, "grad_norm": 2.82502818107605, "tokens_per_second": 3854420.743876932, "eta_seconds": 7622.70169043541, "data_wait_seconds": 2.187490463256836e-05} {"step": 14610, "total_steps": 28610, "tokens": 30639390720, "source": "finemath", "phase": 3, "loss": 2.33529295027256, "ce_loss": 2.31950943171978, "logit_z_loss": 28.568543195724487, "router_aux_loss": 1.4905016422271729, "router_z_loss": 0.8785079643130302, "routing_entropy_normalized": 0.6556651501130282, "learning_rate": 0.003, "grad_norm": 1.1034159660339355, "tokens_per_second": 3856544.47369861, "eta_seconds": 7613.066101074219, "data_wait_seconds": 1.7970800399780273e-05} {"step": 14620, "total_steps": 28610, "tokens": 30660362240, "source": "dclm", "phase": 3, "loss": 3.513513207435608, "ce_loss": 3.4978906512260437, "logit_z_loss": 22.248346090316772, "router_aux_loss": 1.4560461342334747, "router_z_loss": 1.0621398612856865, "routing_entropy_normalized": 0.6568176676600708, "learning_rate": 0.003, "grad_norm": 1.3610169887542725, "tokens_per_second": 3141515.5360007766, "eta_seconds": 9339.17281126976, "data_wait_seconds": 0.13829287886619568} {"step": 14630, "total_steps": 28610, "tokens": 30681333760, "source": "fineweb_hq", "phase": 3, "loss": 3.336579144001007, "ce_loss": 3.3218826949596405, "logit_z_loss": 23.031989574432373, "router_aux_loss": 1.3441468626260757, "router_z_loss": 1.2550471276044846, "routing_entropy_normalized": 0.6853142008644239, "learning_rate": 0.003, "grad_norm": 1.2458314895629883, "tokens_per_second": 4333739.878542263, "eta_seconds": 6765.100301742554, "data_wait_seconds": 2.2292137145996094e-05} {"step": 14640, "total_steps": 28610, "tokens": 30702305280, "source": "dclm", "phase": 3, "loss": 3.4719543755054474, "ce_loss": 3.4569700360298157, "logit_z_loss": 23.27269196510315, "router_aux_loss": 1.3866867423057556, "router_z_loss": 1.1174760311841965, "routing_entropy_normalized": 0.6817287540135477, "learning_rate": 0.003, "grad_norm": 1.0173872709274292, "tokens_per_second": 3867690.428076603, "eta_seconds": 7574.859980344772, "data_wait_seconds": 1.8656253814697266e-05} {"step": 14650, "total_steps": 28610, "tokens": 30723276800, "source": "fineweb_hq", "phase": 3, "loss": 3.361941546201706, "ce_loss": 3.3481456339359283, "logit_z_loss": 22.407557487487793, "router_aux_loss": 1.2513371258974075, "router_z_loss": 1.282566413283348, "routing_entropy_normalized": 0.7167552469977719, "learning_rate": 0.003, "grad_norm": 0.931335985660553, "tokens_per_second": 3830441.3671308584, "eta_seconds": 7643.046613693237, "data_wait_seconds": 1.6450881958007812e-05} {"step": 14660, "total_steps": 28610, "tokens": 30744248320, "source": "fineweb_edu", "phase": 3, "loss": 3.395084798336029, "ce_loss": 3.38118115067482, "logit_z_loss": 21.687522411346436, "router_aux_loss": 1.27284936606884, "router_z_loss": 1.1750892400741577, "routing_entropy_normalized": 0.6855253042685517, "learning_rate": 0.003, "grad_norm": 0.9350302815437317, "tokens_per_second": 3608723.427263084, "eta_seconds": 8106.819763183594, "data_wait_seconds": 2.1249055862426758e-05} {"step": 14670, "total_steps": 28610, "tokens": 30765219840, "source": "dclm", "phase": 3, "loss": 3.5062142610549927, "ce_loss": 3.492239236831665, "logit_z_loss": 21.99044156074524, "router_aux_loss": 1.2975354492664337, "router_z_loss": 0.9996448531746864, "routing_entropy_normalized": 0.6726690969141936, "learning_rate": 0.003, "grad_norm": 0.8893908858299255, "tokens_per_second": 4156741.632086039, "eta_seconds": 7032.9843583106995, "data_wait_seconds": 2.0682811737060547e-05} {"step": 14680, "total_steps": 28610, "tokens": 30786191360, "source": "fineweb_edu", "phase": 3, "loss": 3.374732196331024, "ce_loss": 3.360368311405182, "logit_z_loss": 22.19191336631775, "router_aux_loss": 1.3342624753713608, "router_z_loss": 1.0212427228689194, "routing_entropy_normalized": 0.6734630687932881, "learning_rate": 0.003, "grad_norm": 1.0287063121795654, "tokens_per_second": 3951110.6978505845, "eta_seconds": 7393.700048923492, "data_wait_seconds": 1.901388168334961e-05} {"step": 14690, "total_steps": 28610, "tokens": 30807162880, "source": "fineweb_hq", "phase": 3, "loss": 3.446871340274811, "ce_loss": 3.432172328233719, "logit_z_loss": 22.776309490203857, "router_aux_loss": 1.3715213537216187, "router_z_loss": 0.9838073402643204, "routing_entropy_normalized": 0.6510609521411787, "learning_rate": 0.003, "grad_norm": 1.7861120700836182, "tokens_per_second": 3626028.434509697, "eta_seconds": 8050.779624938965, "data_wait_seconds": 2.041459083557129e-05} {"step": 14700, "total_steps": 28610, "tokens": 30828134400, "source": "fineweb_edu", "phase": 3, "loss": 3.372198522090912, "ce_loss": 3.3569788932800293, "logit_z_loss": 22.319141626358032, "router_aux_loss": 1.412886917591095, "router_z_loss": 1.0906972363591194, "routing_entropy_normalized": 0.633498499931945, "learning_rate": 0.003, "grad_norm": 0.9919693470001221, "tokens_per_second": 3768271.6604711753, "eta_seconds": 7741.316695928574, "data_wait_seconds": 1.874566078186035e-05} {"step": 14710, "total_steps": 28610, "tokens": 30849105920, "source": "finemath", "phase": 3, "loss": 2.3097351491451263, "ce_loss": 2.2946769297122955, "logit_z_loss": 29.199309587478638, "router_aux_loss": 1.4263800233602524, "router_z_loss": 0.794461764395237, "routing_entropy_normalized": 0.6740667659119611, "learning_rate": 0.003, "grad_norm": 0.9330460429191589, "tokens_per_second": 4388515.752019552, "eta_seconds": 6642.430937290192, "data_wait_seconds": 1.811981201171875e-05} {"step": 14720, "total_steps": 28610, "tokens": 30870077440, "source": "dclm", "phase": 3, "loss": 3.5119484066963196, "ce_loss": 3.49747034907341, "logit_z_loss": 23.172003746032715, "router_aux_loss": 1.3478723019361496, "router_z_loss": 0.9993668422102928, "routing_entropy_normalized": 0.6617762332096168, "learning_rate": 0.003, "grad_norm": 0.727979838848114, "tokens_per_second": 3620194.2068812926, "eta_seconds": 8046.3753089904785, "data_wait_seconds": 0.0796586275100708} {"step": 14730, "total_steps": 28610, "tokens": 30891048960, "source": "fineweb_hq", "phase": 3, "loss": 3.376347154378891, "ce_loss": 3.3624293208122253, "logit_z_loss": 21.982853174209595, "router_aux_loss": 1.2819509357213974, "router_z_loss": 1.0982975363731384, "routing_entropy_normalized": 0.6989238574969792, "learning_rate": 0.003, "grad_norm": 1.0160771608352661, "tokens_per_second": 4281162.0117997425, "eta_seconds": 6799.198367118836, "data_wait_seconds": 1.913309097290039e-05} {"step": 14740, "total_steps": 28610, "tokens": 30912020480, "source": "dclm", "phase": 3, "loss": 3.5218916535377502, "ce_loss": 3.5067826211452484, "logit_z_loss": 22.5943443775177, "router_aux_loss": 1.4112024456262589, "router_z_loss": 0.9970253333449364, "routing_entropy_normalized": 0.6406365204097875, "learning_rate": 0.003, "grad_norm": 1.5365514755249023, "tokens_per_second": 3882924.7111279797, "eta_seconds": 7491.1311457157135, "data_wait_seconds": 2.4050474166870117e-05} {"step": 14750, "total_steps": 28610, "tokens": 30932992000, "source": "fineweb_hq", "phase": 3, "loss": 3.327544242143631, "ce_loss": 3.3126645982265472, "logit_z_loss": 23.197288751602173, "router_aux_loss": 1.3888432085514069, "router_z_loss": 0.9912693947553635, "routing_entropy_normalized": 0.6577818205620821, "learning_rate": 0.003, "grad_norm": 0.8126206398010254, "tokens_per_second": 4361306.213366667, "eta_seconds": 6664.637908458711, "data_wait_seconds": 1.958012580871582e-05} {"step": 14760, "total_steps": 28610, "tokens": 30953963520, "source": "fineweb_edu", "phase": 3, "loss": 3.4075232446193695, "ce_loss": 3.3928480446338654, "logit_z_loss": 23.11849021911621, "router_aux_loss": 1.3595407009124756, "router_z_loss": 1.0798459723591805, "routing_entropy_normalized": 0.6734081691919356, "learning_rate": 0.003, "grad_norm": 1.4683144092559814, "tokens_per_second": 3721129.200253352, "eta_seconds": 7805.575575828552, "data_wait_seconds": 1.996755599975586e-05} {"step": 14770, "total_steps": 28610, "tokens": 30974935040, "source": "dclm", "phase": 3, "loss": 3.4804782569408417, "ce_loss": 3.466762065887451, "logit_z_loss": 22.897472858428955, "router_aux_loss": 1.2809146791696548, "router_z_loss": 0.9070199877023697, "routing_entropy_normalized": 0.6724723500001835, "learning_rate": 0.003, "grad_norm": 1.0006211996078491, "tokens_per_second": 4426169.506917049, "eta_seconds": 6557.494834899903, "data_wait_seconds": 1.874566078186035e-05} {"step": 14780, "total_steps": 28610, "tokens": 30995906560, "source": "fineweb_edu", "phase": 3, "loss": 3.3337135910987854, "ce_loss": 3.3193325102329254, "logit_z_loss": 21.83338975906372, "router_aux_loss": 1.3348203301429749, "router_z_loss": 1.0328618437051773, "routing_entropy_normalized": 0.673081134008556, "learning_rate": 0.003, "grad_norm": 1.3391519784927368, "tokens_per_second": 3758462.228455685, "eta_seconds": 7716.882702827454, "data_wait_seconds": 2.1636486053466797e-05} {"step": 14790, "total_steps": 28610, "tokens": 31016878080, "source": "fineweb_hq", "phase": 3, "loss": 3.3488674759864807, "ce_loss": 3.3347634077072144, "logit_z_loss": 22.795387744903564, "router_aux_loss": 1.319634348154068, "router_z_loss": 0.9077069163322449, "routing_entropy_normalized": 0.685016552345889, "learning_rate": 0.003, "grad_norm": 1.0090492963790894, "tokens_per_second": 4032445.2644474027, "eta_seconds": 7187.361201286316, "data_wait_seconds": 2.059340476989746e-05} {"step": 14800, "total_steps": 28610, "tokens": 31037849600, "source": "fineweb_edu", "phase": 3, "loss": 3.3947505950927734, "ce_loss": 3.3806800544261932, "logit_z_loss": 22.39191699028015, "router_aux_loss": 1.316460371017456, "router_z_loss": 0.905895471572876, "routing_entropy_normalized": 0.6646964483490263, "learning_rate": 0.003, "grad_norm": 1.0538313388824463, "tokens_per_second": 3759739.7096841424, "eta_seconds": 7703.104830741882, "data_wait_seconds": 0.013993054628372192} {"step": 14810, "total_steps": 28610, "tokens": 31058821120, "source": "finemath", "phase": 3, "loss": 2.3432618379592896, "ce_loss": 2.329372078180313, "logit_z_loss": 26.647441387176514, "router_aux_loss": 1.3252957463264465, "router_z_loss": 0.6367844641208649, "routing_entropy_normalized": 0.6853890104031487, "learning_rate": 0.003, "grad_norm": 1.042851209640503, "tokens_per_second": 4169944.6981904074, "eta_seconds": 6940.307292938232, "data_wait_seconds": 2.3484230041503906e-05} {"step": 14820, "total_steps": 28610, "tokens": 31079792640, "source": "dclm", "phase": 3, "loss": 3.486854314804077, "ce_loss": 3.4729975759983063, "logit_z_loss": 22.331063985824585, "router_aux_loss": 1.3081785291433334, "router_z_loss": 0.7749064043164253, "routing_entropy_normalized": 0.6730231364022953, "learning_rate": 0.003, "grad_norm": 1.1031352281570435, "tokens_per_second": 3509558.4471589015, "eta_seconds": 8240.274813890457, "data_wait_seconds": 2.2113323211669922e-05} {"step": 14830, "total_steps": 28610, "tokens": 31100764160, "source": "fineweb_hq", "phase": 3, "loss": 3.35450679063797, "ce_loss": 3.3408091068267822, "logit_z_loss": 22.293995141983032, "router_aux_loss": 1.2896595746278763, "router_z_loss": 0.8010640665888786, "routing_entropy_normalized": 0.6755676348675744, "learning_rate": 0.003, "grad_norm": 1.9763174057006836, "tokens_per_second": 4093904.7969659255, "eta_seconds": 7058.9708342552185, "data_wait_seconds": 1.8477439880371094e-05} {"step": 14840, "total_steps": 28610, "tokens": 31121735680, "source": "dclm", "phase": 3, "loss": 3.544988453388214, "ce_loss": 3.531262308359146, "logit_z_loss": 23.53455948829651, "router_aux_loss": 1.303090125322342, "router_z_loss": 0.6952196434140205, "routing_entropy_normalized": 0.6929253632561401, "learning_rate": 0.003, "grad_norm": 2.0646913051605225, "tokens_per_second": 3585427.0408808547, "eta_seconds": 8054.210198879242, "data_wait_seconds": 2.8342008590698242e-05} {"step": 14850, "total_steps": 28610, "tokens": 31142707200, "source": "fineweb_hq", "phase": 3, "loss": 3.3374113142490387, "ce_loss": 3.3234693706035614, "logit_z_loss": 22.755668878555298, "router_aux_loss": 1.3217934519052505, "router_z_loss": 0.7240439057350159, "routing_entropy_normalized": 0.6692319587072332, "learning_rate": 0.003, "grad_norm": 0.6875117421150208, "tokens_per_second": 4457222.378287091, "eta_seconds": 6474.169128417969, "data_wait_seconds": 2.872943878173828e-05} {"step": 14860, "total_steps": 28610, "tokens": 31163678720, "source": "fineweb_edu", "phase": 3, "loss": 3.370414823293686, "ce_loss": 3.3563000857830048, "logit_z_loss": 22.810633897781372, "router_aux_loss": 1.3270775973796844, "router_z_loss": 0.8439597934484482, "routing_entropy_normalized": 0.6796586060049062, "learning_rate": 0.003, "grad_norm": 1.2511159181594849, "tokens_per_second": 2966206.1515780734, "eta_seconds": 9721.455127000809, "data_wait_seconds": 2.2232532501220703e-05} {"step": 14870, "total_steps": 28610, "tokens": 31184650240, "source": "dclm", "phase": 3, "loss": 3.505593419075012, "ce_loss": 3.491281807422638, "logit_z_loss": 23.42536473274231, "router_aux_loss": 1.3624098896980286, "router_z_loss": 0.6875771880149841, "routing_entropy_normalized": 0.6745078397458808, "learning_rate": 0.003, "grad_norm": 1.3628368377685547, "tokens_per_second": 4185528.3993892125, "eta_seconds": 6884.4040060043335, "data_wait_seconds": 0.008464276790618896} {"step": 14880, "total_steps": 28610, "tokens": 31205621760, "source": "fineweb_edu", "phase": 3, "loss": 3.384202867746353, "ce_loss": 3.3691889345645905, "logit_z_loss": 22.42936611175537, "router_aux_loss": 1.4192824363708496, "router_z_loss": 0.821097619831562, "routing_entropy_normalized": 0.6601440714389324, "learning_rate": 0.003, "grad_norm": 1.046773076057434, "tokens_per_second": 4306386.790816193, "eta_seconds": 6686.323908805847, "data_wait_seconds": 2.16066837310791e-05} {"step": 14890, "total_steps": 28610, "tokens": 31226593280, "source": "fineweb_hq", "phase": 3, "loss": 3.3262480795383453, "ce_loss": 3.311928689479828, "logit_z_loss": 23.392184019088745, "router_aux_loss": 1.350280985236168, "router_z_loss": 0.8165454789996147, "routing_entropy_normalized": 0.6775225858116225, "learning_rate": 0.003, "grad_norm": 1.041886568069458, "tokens_per_second": 4252919.462087771, "eta_seconds": 6765.452695846558, "data_wait_seconds": 2.4259090423583984e-05} {"step": 14900, "total_steps": 28610, "tokens": 31247564800, "source": "fineweb_edu", "phase": 3, "loss": 3.3951377272605896, "ce_loss": 3.3809595108032227, "logit_z_loss": 22.310813426971436, "router_aux_loss": 1.3302929401397705, "router_z_loss": 0.8753262236714363, "routing_entropy_normalized": 0.7076750087965246, "learning_rate": 0.003, "grad_norm": 2.094026565551758, "tokens_per_second": 3824809.664772632, "eta_seconds": 7517.2247614860535, "data_wait_seconds": 2.2172927856445312e-05} {"step": 14910, "total_steps": 28610, "tokens": 31268536320, "source": "finemath", "phase": 3, "loss": 2.4007188379764557, "ce_loss": 2.386227995157242, "logit_z_loss": 28.650798797607422, "router_aux_loss": 1.3892067968845367, "router_z_loss": 0.5987713262438774, "routing_entropy_normalized": 0.6807058229122283, "learning_rate": 0.003, "grad_norm": 1.1519007682800293, "tokens_per_second": 2805337.2993996143, "eta_seconds": 10241.542935371399, "data_wait_seconds": 1.8477439880371094e-05} {"step": 14920, "total_steps": 28610, "tokens": 31289507840, "source": "dclm", "phase": 3, "loss": 3.5087716579437256, "ce_loss": 3.4948513507843018, "logit_z_loss": 23.1668221950531, "router_aux_loss": 1.3199267089366913, "router_z_loss": 0.7209911569952965, "routing_entropy_normalized": 0.711690170612616, "learning_rate": 0.003, "grad_norm": 1.2615063190460205, "tokens_per_second": 3579344.0429874836, "eta_seconds": 8021.02578997612, "data_wait_seconds": 2.148747444152832e-05} {"step": 14930, "total_steps": 28610, "tokens": 31310479360, "source": "fineweb_hq", "phase": 3, "loss": 3.35040146112442, "ce_loss": 3.336387276649475, "logit_z_loss": 22.361200094223022, "router_aux_loss": 1.322846606373787, "router_z_loss": 0.7857032120227814, "routing_entropy_normalized": 0.6741210234026673, "learning_rate": 0.003, "grad_norm": 1.3578156232833862, "tokens_per_second": 3882860.6060314984, "eta_seconds": 7388.634893417358, "data_wait_seconds": 2.187490463256836e-05} {"step": 14940, "total_steps": 28610, "tokens": 31331450880, "source": "dclm", "phase": 3, "loss": 3.4940956234931946, "ce_loss": 3.4799416065216064, "logit_z_loss": 22.57875680923462, "router_aux_loss": 1.346128597855568, "router_z_loss": 0.6927533149719238, "routing_entropy_normalized": 0.6815060304669419, "learning_rate": 0.003, "grad_norm": 1.3226317167282104, "tokens_per_second": 3195862.332967184, "eta_seconds": 8970.370076417923, "data_wait_seconds": 2.250075340270996e-05} {"step": 14950, "total_steps": 28610, "tokens": 31352422400, "source": "fineweb_hq", "phase": 3, "loss": 3.3429385125637054, "ce_loss": 3.3277294039726257, "logit_z_loss": 22.15864610671997, "router_aux_loss": 1.4406924843788147, "router_z_loss": 0.8022119998931885, "routing_entropy_normalized": 0.6352948776284851, "learning_rate": 0.003, "grad_norm": 1.0769155025482178, "tokens_per_second": 3793039.786181071, "eta_seconds": 7552.543061733246, "data_wait_seconds": 2.0742416381835938e-05} {"step": 14960, "total_steps": 28610, "tokens": 31373393920, "source": "fineweb_edu", "phase": 3, "loss": 3.36594620347023, "ce_loss": 3.351202219724655, "logit_z_loss": 21.655864238739014, "router_aux_loss": 1.3818018585443497, "router_z_loss": 0.9259754195809364, "routing_entropy_normalized": 0.6728391751198748, "learning_rate": 0.003, "grad_norm": 1.0991685390472412, "tokens_per_second": 4122321.338302437, "eta_seconds": 6944.175975322723, "data_wait_seconds": 2.2917985916137695e-05} {"step": 14970, "total_steps": 28610, "tokens": 31394365440, "source": "dclm", "phase": 3, "loss": 3.385260820388794, "ce_loss": 3.371054619550705, "logit_z_loss": 23.214889764785767, "router_aux_loss": 1.342295989394188, "router_z_loss": 0.7832387313246727, "routing_entropy_normalized": 0.6691237284991897, "learning_rate": 0.003, "grad_norm": 1.048229455947876, "tokens_per_second": 3922244.519576502, "eta_seconds": 7293.057109832764, "data_wait_seconds": 0.0110740065574646} {"step": 14980, "total_steps": 28610, "tokens": 31415336960, "source": "fineweb_edu", "phase": 3, "loss": 3.3649766743183136, "ce_loss": 3.3506242632865906, "logit_z_loss": 22.64191508293152, "router_aux_loss": 1.3464507311582565, "router_z_loss": 0.8879445865750313, "routing_entropy_normalized": 0.667151718637184, "learning_rate": 0.003, "grad_norm": 0.7639074325561523, "tokens_per_second": 3397531.694687187, "eta_seconds": 8413.220045804977, "data_wait_seconds": 2.777576446533203e-05} {"step": 14990, "total_steps": 28610, "tokens": 31436308480, "source": "fineweb_hq", "phase": 3, "loss": 3.3500084280967712, "ce_loss": 3.336136907339096, "logit_z_loss": 22.29883861541748, "router_aux_loss": 1.2952257245779037, "router_z_loss": 0.9192829430103302, "routing_entropy_normalized": 0.6970296716840967, "learning_rate": 0.003, "grad_norm": 0.885619580745697, "tokens_per_second": 3366368.771009396, "eta_seconds": 8484.872627735138, "data_wait_seconds": 2.110004425048828e-05} {"step": 15000, "total_steps": 28610, "tokens": 31457280000, "source": "fineweb_edu", "phase": 3, "loss": 3.4002501368522644, "ce_loss": 3.385121464729309, "logit_z_loss": 23.178518533706665, "router_aux_loss": 1.4178126156330109, "router_z_loss": 0.9505827352404594, "routing_entropy_normalized": 0.6391933123190389, "learning_rate": 0.003, "grad_norm": 2.1081149578094482, "tokens_per_second": 4099687.452105218, "eta_seconds": 6962.052364587784, "data_wait_seconds": 2.2977590560913086e-05} {"step": 15010, "total_steps": 28610, "tokens": 31478251520, "source": "finemath", "phase": 3, "loss": 2.3923233449459076, "ce_loss": 2.3773195445537567, "logit_z_loss": 28.728124618530273, "router_aux_loss": 1.429043710231781, "router_z_loss": 0.7133954614400864, "routing_entropy_normalized": 0.6765880948469394, "learning_rate": 0.003, "grad_norm": 1.0842361450195312, "tokens_per_second": 3812265.9278773377, "eta_seconds": 7481.447448730469, "data_wait_seconds": 1.8268823623657227e-05} {"step": 15020, "total_steps": 28610, "tokens": 31499223040, "source": "dclm", "phase": 3, "loss": 3.436145633459091, "ce_loss": 3.420709103345871, "logit_z_loss": 23.176124095916748, "router_aux_loss": 1.4541510045528412, "router_z_loss": 0.8950740545988083, "routing_entropy_normalized": 0.6502275206418998, "learning_rate": 0.003, "grad_norm": 1.0356147289276123, "tokens_per_second": 3807362.7375877867, "eta_seconds": 7485.574042797089, "data_wait_seconds": 0.05746033787727356} {"step": 15030, "total_steps": 28610, "tokens": 31520194560, "source": "fineweb_hq", "phase": 3, "loss": 3.3855530619621277, "ce_loss": 3.371372699737549, "logit_z_loss": 22.789999961853027, "router_aux_loss": 1.3145834058523178, "router_z_loss": 1.034547209739685, "routing_entropy_normalized": 0.6706973594202872, "learning_rate": 0.003, "grad_norm": 1.3053345680236816, "tokens_per_second": 3753223.561754425, "eta_seconds": 7587.963704109192, "data_wait_seconds": 2.238154411315918e-05} {"step": 15040, "total_steps": 28610, "tokens": 31541166080, "source": "dclm", "phase": 3, "loss": 3.3904502391815186, "ce_loss": 3.375776529312134, "logit_z_loss": 23.561808824539185, "router_aux_loss": 1.3828155398368835, "router_z_loss": 0.8454997390508652, "routing_entropy_normalized": 0.6620754033876517, "learning_rate": 0.003, "grad_norm": 1.4909369945526123, "tokens_per_second": 4006817.997095042, "eta_seconds": 7102.481984615326, "data_wait_seconds": 2.199411392211914e-05} {"step": 15050, "total_steps": 28610, "tokens": 31562137600, "source": "fineweb_hq", "phase": 3, "loss": 3.353844404220581, "ce_loss": 3.338941603899002, "logit_z_loss": 22.598501443862915, "router_aux_loss": 1.3842859268188477, "router_z_loss": 1.0599594041705132, "routing_entropy_normalized": 0.6671397152713275, "learning_rate": 0.003, "grad_norm": 0.6893020272254944, "tokens_per_second": 4475117.203335708, "eta_seconds": 6354.555607795715, "data_wait_seconds": 1.9431114196777344e-05} {"step": 15060, "total_steps": 28610, "tokens": 31583109120, "source": "fineweb_edu", "phase": 3, "loss": 3.3523922860622406, "ce_loss": 3.338149309158325, "logit_z_loss": 22.582387924194336, "router_aux_loss": 1.3134422898292542, "router_z_loss": 1.1084641888737679, "routing_entropy_normalized": 0.6838781235990335, "learning_rate": 0.003, "grad_norm": 1.9402121305465698, "tokens_per_second": 3663495.6643503783, "eta_seconds": 7756.637977361679, "data_wait_seconds": 2.1636486053466797e-05} {"step": 15070, "total_steps": 28610, "tokens": 31604080640, "source": "dclm", "phase": 3, "loss": 3.424877405166626, "ce_loss": 3.4108915627002716, "logit_z_loss": 23.069477558135986, "router_aux_loss": 1.303214356303215, "router_z_loss": 0.9536968991160393, "routing_entropy_normalized": 0.6772118047179531, "learning_rate": 0.003, "grad_norm": 1.2383939027786255, "tokens_per_second": 3223782.342240752, "eta_seconds": 8808.112665653229, "data_wait_seconds": 0.09618353843688965} {"step": 15080, "total_steps": 28610, "tokens": 31625052160, "source": "fineweb_edu", "phase": 3, "loss": 3.3490415811538696, "ce_loss": 3.3351167738437653, "logit_z_loss": 22.748140573501587, "router_aux_loss": 1.2912745028734207, "router_z_loss": 1.0120016932487488, "routing_entropy_normalized": 0.6961664370702147, "learning_rate": 0.003, "grad_norm": 1.2243608236312866, "tokens_per_second": 4277190.016148215, "eta_seconds": 6633.903673410416, "data_wait_seconds": 2.0325183868408203e-05} {"step": 15090, "total_steps": 28610, "tokens": 31646023680, "source": "fineweb_hq", "phase": 3, "loss": 3.393789619207382, "ce_loss": 3.3782627880573273, "logit_z_loss": 23.698913097381592, "router_aux_loss": 1.456526204943657, "router_z_loss": 0.9615735486149788, "routing_entropy_normalized": 0.6350729599383291, "learning_rate": 0.003, "grad_norm": 2.1937949657440186, "tokens_per_second": 3272764.156530234, "eta_seconds": 8663.470291137695, "data_wait_seconds": 2.086162567138672e-05} {"step": 15100, "total_steps": 28610, "tokens": 31666995200, "source": "fineweb_edu", "phase": 3, "loss": 3.3833735585212708, "ce_loss": 3.367616444826126, "logit_z_loss": 22.522968292236328, "router_aux_loss": 1.4710766822099686, "router_z_loss": 1.0463255196809769, "routing_entropy_normalized": 0.6493140812416799, "learning_rate": 0.003, "grad_norm": 1.077729344367981, "tokens_per_second": 4278766.273992347, "eta_seconds": 6621.657203435898, "data_wait_seconds": 2.2351741790771484e-05} {"step": 15110, "total_steps": 28610, "tokens": 31687966720, "source": "finemath", "phase": 3, "loss": 2.3761720061302185, "ce_loss": 2.360998272895813, "logit_z_loss": 27.381698608398438, "router_aux_loss": 1.4217129796743393, "router_z_loss": 0.9566333964467049, "routing_entropy_normalized": 0.698386273378386, "learning_rate": 0.003, "grad_norm": 1.4728038311004639, "tokens_per_second": 3381223.897161447, "eta_seconds": 8373.16689491272, "data_wait_seconds": 1.665949821472168e-05} {"step": 15120, "total_steps": 28610, "tokens": 31708938240, "source": "dclm", "phase": 3, "loss": 3.469215750694275, "ce_loss": 3.4538804292678833, "logit_z_loss": 23.163206338882446, "router_aux_loss": 1.414428949356079, "router_z_loss": 1.1910174190998077, "routing_entropy_normalized": 0.6416725976059773, "learning_rate": 0.003, "grad_norm": 1.4088973999023438, "tokens_per_second": 3532458.2421378936, "eta_seconds": 8008.751566410065, "data_wait_seconds": 0.13985422253608704} {"step": 15130, "total_steps": 28610, "tokens": 31729909760, "source": "fineweb_hq", "phase": 3, "loss": 3.3693612217903137, "ce_loss": 3.3543940782546997, "logit_z_loss": 22.903712034225464, "router_aux_loss": 1.3614278584718704, "router_z_loss": 1.3528501242399216, "routing_entropy_normalized": 0.6810889826548477, "learning_rate": 0.003, "grad_norm": 1.2984304428100586, "tokens_per_second": 4250509.416527106, "eta_seconds": 6650.875504493713, "data_wait_seconds": 2.0116567611694336e-05} {"step": 15140, "total_steps": 28610, "tokens": 31750881280, "source": "dclm", "phase": 3, "loss": 3.4674127101898193, "ce_loss": 3.452709436416626, "logit_z_loss": 24.061485290527344, "router_aux_loss": 1.352981686592102, "router_z_loss": 1.1733988672494888, "routing_entropy_normalized": 0.6569554090394023, "learning_rate": 0.003, "grad_norm": 0.9125359058380127, "tokens_per_second": 4255324.213577863, "eta_seconds": 6638.421897411346, "data_wait_seconds": 2.0265579223632812e-05} {"step": 15150, "total_steps": 28610, "tokens": 31771852800, "source": "fineweb_hq", "phase": 3, "loss": 3.4215860068798065, "ce_loss": 3.407135456800461, "logit_z_loss": 22.70605969429016, "router_aux_loss": 1.3193725794553757, "router_z_loss": 1.2568323761224747, "routing_entropy_normalized": 0.6832355132053559, "learning_rate": 0.003, "grad_norm": 1.0364816188812256, "tokens_per_second": 3565801.898128112, "eta_seconds": 7916.218210220336, "data_wait_seconds": 1.9937753677368164e-05} {"step": 15160, "total_steps": 28610, "tokens": 31792824320, "source": "fineweb_edu", "phase": 3, "loss": 3.345838785171509, "ce_loss": 3.3315427899360657, "logit_z_loss": 22.258610010147095, "router_aux_loss": 1.3046013861894608, "router_z_loss": 1.2499877214431763, "routing_entropy_normalized": 0.6506443556186589, "learning_rate": 0.003, "grad_norm": 0.7003273367881775, "tokens_per_second": 4228646.305796833, "eta_seconds": 6670.383938550949, "data_wait_seconds": 2.0116567611694336e-05} {"step": 15170, "total_steps": 28610, "tokens": 31813795840, "source": "dclm", "phase": 3, "loss": 3.455885261297226, "ce_loss": 3.4422787129879, "logit_z_loss": 23.37136721611023, "router_aux_loss": 1.2710347026586533, "router_z_loss": 0.8961262181401253, "routing_entropy_normalized": 0.6778725022427534, "learning_rate": 0.003, "grad_norm": 1.0632801055908203, "tokens_per_second": 3981660.7542351773, "eta_seconds": 7078.886077880859, "data_wait_seconds": 1.8417835235595703e-05} {"step": 15180, "total_steps": 28610, "tokens": 31834767360, "source": "fineweb_edu", "phase": 3, "loss": 3.360903024673462, "ce_loss": 3.346299856901169, "logit_z_loss": 22.77000594139099, "router_aux_loss": 1.3531569093465805, "router_z_loss": 1.0715954005718231, "routing_entropy_normalized": 0.6648528730326314, "learning_rate": 0.003, "grad_norm": 1.116687536239624, "tokens_per_second": 3980883.5531951706, "eta_seconds": 7075.000055551529, "data_wait_seconds": 2.1159648895263672e-05} {"step": 15190, "total_steps": 28610, "tokens": 31855738880, "source": "fineweb_hq", "phase": 3, "loss": 3.391800880432129, "ce_loss": 3.377396583557129, "logit_z_loss": 21.754695892333984, "router_aux_loss": 1.3393499106168747, "router_z_loss": 1.0108097270131111, "routing_entropy_normalized": 0.6608806470291362, "learning_rate": 0.003, "grad_norm": 1.6538790464401245, "tokens_per_second": 3589651.295838987, "eta_seconds": 7840.254531860352, "data_wait_seconds": 1.9937753677368164e-05} {"step": 15200, "total_steps": 28610, "tokens": 31876710400, "source": "fineweb_edu", "phase": 3, "loss": 3.3758600652217865, "ce_loss": 3.3607503473758698, "logit_z_loss": 23.0788254737854, "router_aux_loss": 1.41382297873497, "router_z_loss": 0.9715359210968018, "routing_entropy_normalized": 0.6334900152444967, "learning_rate": 0.003, "grad_norm": 1.280338168144226, "tokens_per_second": 4373375.425652808, "eta_seconds": 6430.458303451538, "data_wait_seconds": 2.205371856689453e-05} {"step": 15210, "total_steps": 28610, "tokens": 31897681920, "source": "finemath", "phase": 3, "loss": 2.3561191856861115, "ce_loss": 2.341898798942566, "logit_z_loss": 29.553191423416138, "router_aux_loss": 1.3497808277606964, "router_z_loss": 0.7225965484976768, "routing_entropy_normalized": 0.6766870261045953, "learning_rate": 0.003, "grad_norm": 1.295181393623352, "tokens_per_second": 3748638.713127589, "eta_seconds": 7496.544466018677, "data_wait_seconds": 1.7911195755004883e-05} {"step": 15220, "total_steps": 28610, "tokens": 31918653440, "source": "dclm", "phase": 3, "loss": 3.4560305178165436, "ce_loss": 3.440952092409134, "logit_z_loss": 23.16137933731079, "router_aux_loss": 1.4231651425361633, "router_z_loss": 0.8468063026666641, "routing_entropy_normalized": 0.6308874837655913, "learning_rate": 0.003, "grad_norm": 1.2410134077072144, "tokens_per_second": 3421480.5254849955, "eta_seconds": 8207.226395368576, "data_wait_seconds": 1.7374753952026367e-05} {"step": 15230, "total_steps": 28610, "tokens": 31939624960, "source": "fineweb_hq", "phase": 3, "loss": 3.3637583553791046, "ce_loss": 3.349452942609787, "logit_z_loss": 23.163219928741455, "router_aux_loss": 1.3267690241336823, "router_z_loss": 1.0377638041973114, "routing_entropy_normalized": 0.6763000829998853, "learning_rate": 0.003, "grad_norm": 1.126561164855957, "tokens_per_second": 3501591.890679196, "eta_seconds": 8013.46777009964, "data_wait_seconds": 1.7523765563964844e-05} {"step": 15240, "total_steps": 28610, "tokens": 31960596480, "source": "dclm", "phase": 3, "loss": 3.4697655141353607, "ce_loss": 3.4553768038749695, "logit_z_loss": 23.634591102600098, "router_aux_loss": 1.352329671382904, "router_z_loss": 0.8653488308191299, "routing_entropy_normalized": 0.6658705087689966, "learning_rate": 0.003, "grad_norm": 1.2759432792663574, "tokens_per_second": 3771256.836598949, "eta_seconds": 7434.901268959045, "data_wait_seconds": 2.2113323211669922e-05} {"step": 15250, "total_steps": 28610, "tokens": 31981568000, "source": "fineweb_hq", "phase": 3, "loss": 3.3741275668144226, "ce_loss": 3.359325557947159, "logit_z_loss": 23.010480403900146, "router_aux_loss": 1.3796892017126083, "router_z_loss": 1.0051845535635948, "routing_entropy_normalized": 0.6741357140179869, "learning_rate": 0.003, "grad_norm": 1.1040481328964233, "tokens_per_second": 4265092.902597771, "eta_seconds": 6569.130229949951, "data_wait_seconds": 2.002716064453125e-05} {"step": 15260, "total_steps": 28610, "tokens": 32002539520, "source": "fineweb_edu", "phase": 3, "loss": 3.377731055021286, "ce_loss": 3.362404525279999, "logit_z_loss": 22.383836269378662, "router_aux_loss": 1.430534541606903, "router_z_loss": 1.021183118224144, "routing_entropy_normalized": 0.652836001855512, "learning_rate": 0.003, "grad_norm": 1.1047694683074951, "tokens_per_second": 3987817.5286338003, "eta_seconds": 7020.626946687698, "data_wait_seconds": 1.8984079360961914e-05} {"step": 15270, "total_steps": 28610, "tokens": 32023511040, "source": "dclm", "phase": 3, "loss": 3.471604287624359, "ce_loss": 3.4566531777381897, "logit_z_loss": 22.840162754058838, "router_aux_loss": 1.400195986032486, "router_z_loss": 0.9492233395576477, "routing_entropy_normalized": 0.680196575087918, "learning_rate": 0.003, "grad_norm": 1.0681060552597046, "tokens_per_second": 4046994.61708847, "eta_seconds": 6912.785987377167, "data_wait_seconds": 0.010220736265182495} {"step": 15280, "total_steps": 28610, "tokens": 32044482560, "source": "fineweb_edu", "phase": 3, "loss": 3.3718439042568207, "ce_loss": 3.356815814971924, "logit_z_loss": 22.76010751724243, "router_aux_loss": 1.3947591483592987, "router_z_loss": 1.0805070847272873, "routing_entropy_normalized": 0.640613594213086, "learning_rate": 0.003, "grad_norm": 0.8149470090866089, "tokens_per_second": 4141669.311547517, "eta_seconds": 6749.702609539032, "data_wait_seconds": 2.0682811737060547e-05} {"step": 15290, "total_steps": 28610, "tokens": 32065454080, "source": "fineweb_hq", "phase": 3, "loss": 3.4126150608062744, "ce_loss": 3.398367166519165, "logit_z_loss": 23.556098222732544, "router_aux_loss": 1.3203145414590836, "router_z_loss": 1.04478170722723, "routing_entropy_normalized": 0.6785066107443567, "learning_rate": 0.003, "grad_norm": 0.8521683216094971, "tokens_per_second": 3504178.42653137, "eta_seconds": 7971.644488334656, "data_wait_seconds": 2.8312206268310547e-05} {"step": 15300, "total_steps": 28610, "tokens": 32086425600, "source": "fineweb_edu", "phase": 3, "loss": 3.3833592534065247, "ce_loss": 3.368848741054535, "logit_z_loss": 22.99602723121643, "router_aux_loss": 1.3395826518535614, "router_z_loss": 1.1146880388259888, "routing_entropy_normalized": 0.6580053653489865, "learning_rate": 0.003, "grad_norm": 1.4346098899841309, "tokens_per_second": 4076148.196637686, "eta_seconds": 6847.909294128418, "data_wait_seconds": 1.9729137420654297e-05} {"step": 15310, "total_steps": 28610, "tokens": 32107397120, "source": "finemath", "phase": 3, "loss": 2.4459022879600525, "ce_loss": 2.4314162135124207, "logit_z_loss": 28.648811101913452, "router_aux_loss": 1.3742894530296326, "router_z_loss": 0.743211142718792, "routing_entropy_normalized": 0.6806385540713631, "learning_rate": 0.003, "grad_norm": 0.8970857262611389, "tokens_per_second": 3689269.1101712068, "eta_seconds": 7560.338041782378, "data_wait_seconds": 1.9550323486328125e-05} {"step": 15320, "total_steps": 28610, "tokens": 32128368640, "source": "dclm", "phase": 3, "loss": 3.4853655993938446, "ce_loss": 3.4707166254520416, "logit_z_loss": 22.77647089958191, "router_aux_loss": 1.3760058283805847, "router_z_loss": 0.8889390528202057, "routing_entropy_normalized": 0.6820691549095429, "learning_rate": 0.003, "grad_norm": 0.8395155668258667, "tokens_per_second": 3415878.198670672, "eta_seconds": 8159.292708635331, "data_wait_seconds": 2.2202730178833008e-05} {"step": 15330, "total_steps": 28610, "tokens": 32149340160, "source": "fineweb_hq", "phase": 3, "loss": 3.3834891617298126, "ce_loss": 3.3692349791526794, "logit_z_loss": 23.070497035980225, "router_aux_loss": 1.321603998541832, "router_z_loss": 1.0381317809224129, "routing_entropy_normalized": 0.6718380504956133, "learning_rate": 0.003, "grad_norm": 0.8006716966629028, "tokens_per_second": 3845647.8478384702, "eta_seconds": 7241.999179840088, "data_wait_seconds": 2.3871660232543945e-05} {"step": 15340, "total_steps": 28610, "tokens": 32170311680, "source": "dclm", "phase": 3, "loss": 3.491544932126999, "ce_loss": 3.4763101041316986, "logit_z_loss": 24.397624015808105, "router_aux_loss": 1.4383540898561478, "router_z_loss": 0.8513799011707306, "routing_entropy_normalized": 0.6426727065327187, "learning_rate": 0.003, "grad_norm": 1.0285178422927856, "tokens_per_second": 4354579.102502817, "eta_seconds": 6390.791482925414, "data_wait_seconds": 2.3990869522094727e-05} {"step": 15350, "total_steps": 28610, "tokens": 32191283200, "source": "fineweb_hq", "phase": 3, "loss": 3.3947266936302185, "ce_loss": 3.379996418952942, "logit_z_loss": 22.751235723495483, "router_aux_loss": 1.3780971765518188, "router_z_loss": 0.9492755830287933, "routing_entropy_normalized": 0.6753570631155256, "learning_rate": 0.003, "grad_norm": 0.8213182091712952, "tokens_per_second": 4217738.202928794, "eta_seconds": 6593.163013458253, "data_wait_seconds": 1.952052116394043e-05} {"step": 15360, "total_steps": 28610, "tokens": 32212254720, "source": "fineweb_edu", "phase": 3, "loss": 3.3779861629009247, "ce_loss": 3.3631949424743652, "logit_z_loss": 22.366278886795044, "router_aux_loss": 1.3813589662313461, "router_z_loss": 0.9775811731815338, "routing_entropy_normalized": 0.6636096668432305, "learning_rate": 0.003, "grad_norm": 1.5957927703857422, "tokens_per_second": 3452189.824243559, "eta_seconds": 8049.170356988907, "data_wait_seconds": 1.8894672393798828e-05} {"step": 15370, "total_steps": 28610, "tokens": 32233226240, "source": "dclm", "phase": 3, "loss": 3.458721876144409, "ce_loss": 3.444190114736557, "logit_z_loss": 23.254053354263306, "router_aux_loss": 1.3651554435491562, "router_z_loss": 0.8801988735795021, "routing_entropy_normalized": 0.6657207438071038, "learning_rate": 0.003, "grad_norm": 1.2434413433074951, "tokens_per_second": 4095343.3059853953, "eta_seconds": 6779.967002868652, "data_wait_seconds": 1.8805265426635742e-05} {"step": 15380, "total_steps": 28610, "tokens": 32254197760, "source": "fineweb_edu", "phase": 3, "loss": 3.3372568488121033, "ce_loss": 3.322418600320816, "logit_z_loss": 22.467411041259766, "router_aux_loss": 1.3776449263095856, "router_z_loss": 1.0617936328053474, "routing_entropy_normalized": 0.6750718858620457, "learning_rate": 0.003, "grad_norm": 1.2095179557800293, "tokens_per_second": 4004861.9776924807, "eta_seconds": 6927.909404754639, "data_wait_seconds": 2.1755695343017578e-05} {"step": 15390, "total_steps": 28610, "tokens": 32275169280, "source": "fineweb_hq", "phase": 3, "loss": 3.468325585126877, "ce_loss": 3.4537743628025055, "logit_z_loss": 22.371248960494995, "router_aux_loss": 1.341529220342636, "router_z_loss": 1.135939747095108, "routing_entropy_normalized": 0.6575477657710231, "learning_rate": 0.003, "grad_norm": 1.1386029720306396, "tokens_per_second": 4067652.0809746464, "eta_seconds": 6815.811403751373, "data_wait_seconds": 1.996755599975586e-05} {"step": 15400, "total_steps": 28610, "tokens": 32296140800, "source": "fineweb_edu", "phase": 3, "loss": 3.342466861009598, "ce_loss": 3.327606976032257, "logit_z_loss": 24.65649700164795, "router_aux_loss": 1.3717081099748611, "router_z_loss": 1.14279206097126, "routing_entropy_normalized": 0.6468624642753845, "learning_rate": 0.003, "grad_norm": 2.5044772624969482, "tokens_per_second": 3516729.0632426017, "eta_seconds": 7877.598024129868, "data_wait_seconds": 0.10468533635139465} {"step": 15410, "total_steps": 28610, "tokens": 32317112320, "source": "finemath", "phase": 3, "loss": 2.565632700920105, "ce_loss": 2.550186276435852, "logit_z_loss": 28.40819525718689, "router_aux_loss": 1.4635484367609024, "router_z_loss": 0.8109193071722984, "routing_entropy_normalized": 0.6760256214156659, "learning_rate": 0.003, "grad_norm": 1.136565089225769, "tokens_per_second": 4092975.173970777, "eta_seconds": 6763.394651412964, "data_wait_seconds": 1.7821788787841797e-05} {"step": 15420, "total_steps": 28610, "tokens": 32338083840, "source": "dclm", "phase": 3, "loss": 3.4577183723449707, "ce_loss": 3.441713511943817, "logit_z_loss": 22.654199600219727, "router_aux_loss": 1.5035229474306107, "router_z_loss": 0.9696166962385178, "routing_entropy_normalized": 0.6480572833108204, "learning_rate": 0.003, "grad_norm": 1.324737548828125, "tokens_per_second": 3974332.0426264745, "eta_seconds": 6960.021101236343, "data_wait_seconds": 2.2202730178833008e-05} {"step": 15430, "total_steps": 28610, "tokens": 32359055360, "source": "fineweb_hq", "phase": 3, "loss": 3.3999774158000946, "ce_loss": 3.3851141333580017, "logit_z_loss": 23.540618419647217, "router_aux_loss": 1.3647599071264267, "router_z_loss": 1.2157187908887863, "routing_entropy_normalized": 0.6724404179577949, "learning_rate": 0.003, "grad_norm": 1.1250702142715454, "tokens_per_second": 4237937.03791473, "eta_seconds": 6522.150544643402, "data_wait_seconds": 2.0831823348999023e-05} {"step": 15440, "total_steps": 28610, "tokens": 32380026880, "source": "dclm", "phase": 3, "loss": 3.451332837343216, "ce_loss": 3.4369117617607117, "logit_z_loss": 24.106912851333618, "router_aux_loss": 1.3358049839735031, "router_z_loss": 1.0630199015140533, "routing_entropy_normalized": 0.6591413224735019, "learning_rate": 0.003, "grad_norm": 1.2344897985458374, "tokens_per_second": 3374408.0259914137, "eta_seconds": 8184.988782405853, "data_wait_seconds": 2.3543834686279297e-05} {"step": 15450, "total_steps": 28610, "tokens": 32400998400, "source": "fineweb_hq", "phase": 3, "loss": 3.3878704011440277, "ce_loss": 3.3738230764865875, "logit_z_loss": 22.521594762802124, "router_aux_loss": 1.2873267084360123, "router_z_loss": 1.174041986465454, "routing_entropy_normalized": 0.7008773540030292, "learning_rate": 0.003, "grad_norm": 0.7523600459098816, "tokens_per_second": 3683680.022004688, "eta_seconds": 7492.105762481689, "data_wait_seconds": 0.09401994943618774} {"step": 15460, "total_steps": 28610, "tokens": 32421969920, "source": "fineweb_edu", "phase": 3, "loss": 3.367650479078293, "ce_loss": 3.353223890066147, "logit_z_loss": 22.32720375061035, "router_aux_loss": 1.3306158035993576, "router_z_loss": 1.1204490661621094, "routing_entropy_normalized": 0.6765058485603526, "learning_rate": 0.003, "grad_norm": 0.877496600151062, "tokens_per_second": 4012688.932631014, "eta_seconds": 6872.585755586624, "data_wait_seconds": 2.1785497665405273e-05} {"step": 15470, "total_steps": 28610, "tokens": 32442941440, "source": "dclm", "phase": 3, "loss": 3.434380918741226, "ce_loss": 3.419438898563385, "logit_z_loss": 23.69344711303711, "router_aux_loss": 1.399112030863762, "router_z_loss": 0.9508931487798691, "routing_entropy_normalized": 0.6684818323417527, "learning_rate": 0.003, "grad_norm": 1.024765968322754, "tokens_per_second": 3836361.0529766376, "eta_seconds": 7182.998914718628, "data_wait_seconds": 2.16066837310791e-05} {"step": 15480, "total_steps": 28610, "tokens": 32463912960, "source": "fineweb_edu", "phase": 3, "loss": 3.326950967311859, "ce_loss": 3.3111492693424225, "logit_z_loss": 22.902705669403076, "router_aux_loss": 1.4693742245435715, "router_z_loss": 1.1079286187887192, "routing_entropy_normalized": 0.6437195159678181, "learning_rate": 0.003, "grad_norm": 0.9521719813346863, "tokens_per_second": 3071547.6644989136, "eta_seconds": 8964.733342170715, "data_wait_seconds": 2.008676528930664e-05} {"step": 15490, "total_steps": 28610, "tokens": 32484884480, "source": "fineweb_hq", "phase": 3, "loss": 3.3851940631866455, "ce_loss": 3.3703073859214783, "logit_z_loss": 23.54992914199829, "router_aux_loss": 1.3748075366020203, "router_z_loss": 1.138604149222374, "routing_entropy_normalized": 0.681284047456917, "learning_rate": 0.003, "grad_norm": 1.0547798871994019, "tokens_per_second": 3338474.7257090197, "eta_seconds": 8241.678161621094, "data_wait_seconds": 2.321600914001465e-05} {"step": 15500, "total_steps": 28610, "tokens": 32505856000, "source": "fineweb_edu", "phase": 3, "loss": 3.352949410676956, "ce_loss": 3.3384086787700653, "logit_z_loss": 22.641786098480225, "router_aux_loss": 1.327665627002716, "router_z_loss": 1.2640420496463776, "routing_entropy_normalized": 0.6826439750891101, "learning_rate": 0.003, "grad_norm": 1.2124192714691162, "tokens_per_second": 4495824.397597591, "eta_seconds": 6115.377356529235, "data_wait_seconds": 2.1457672119140625e-05} {"step": 15510, "total_steps": 28610, "tokens": 32526827520, "source": "finemath", "phase": 3, "loss": 2.3275263607501984, "ce_loss": 2.3123365342617035, "logit_z_loss": 30.44662308692932, "router_aux_loss": 1.4391886293888092, "router_z_loss": 0.7978924736380577, "routing_entropy_normalized": 0.663711272241545, "learning_rate": 0.003, "grad_norm": 0.9223556518554688, "tokens_per_second": 4256712.173810618, "eta_seconds": 6453.969655036926, "data_wait_seconds": 2.1517276763916016e-05} {"step": 15520, "total_steps": 28610, "tokens": 32547799040, "source": "dclm", "phase": 3, "loss": 3.467688024044037, "ce_loss": 3.4522148966789246, "logit_z_loss": 23.33480715751648, "router_aux_loss": 1.4396733939647675, "router_z_loss": 1.0764093101024628, "routing_entropy_normalized": 0.6688200658767786, "learning_rate": 0.003, "grad_norm": 1.3942514657974243, "tokens_per_second": 3409364.281297641, "eta_seconds": 8051.85876750946, "data_wait_seconds": 2.0235776901245117e-05} {"step": 15530, "total_steps": 28610, "tokens": 32568770560, "source": "fineweb_hq", "phase": 3, "loss": 3.329889714717865, "ce_loss": 3.314726769924164, "logit_z_loss": 23.54013180732727, "router_aux_loss": 1.3843904435634613, "router_z_loss": 1.3190508484840393, "routing_entropy_normalized": 0.680896147229903, "learning_rate": 0.003, "grad_norm": 1.1727746725082397, "tokens_per_second": 3779094.5083746915, "eta_seconds": 7258.5504541397095, "data_wait_seconds": 2.300739288330078e-05} {"step": 15540, "total_steps": 28610, "tokens": 32589742080, "source": "dclm", "phase": 3, "loss": 3.443952828645706, "ce_loss": 3.4279400408267975, "logit_z_loss": 24.708027124404907, "router_aux_loss": 1.4870203137397766, "router_z_loss": 1.1426588594913483, "routing_entropy_normalized": 0.6468743529212685, "learning_rate": 0.003, "grad_norm": 1.181413173675537, "tokens_per_second": 4406628.861474608, "eta_seconds": 6220.123704910278, "data_wait_seconds": 1.8835067749023438e-05} {"step": 15550, "total_steps": 28610, "tokens": 32610713600, "source": "fineweb_hq", "phase": 3, "loss": 3.3143545389175415, "ce_loss": 3.299376428127289, "logit_z_loss": 23.607408046722412, "router_aux_loss": 1.3623132705688477, "router_z_loss": 1.3549960106611252, "routing_entropy_normalized": 0.6864719641226985, "learning_rate": 0.003, "grad_norm": 0.9046967029571533, "tokens_per_second": 3785646.2955234647, "eta_seconds": 7234.908647537231, "data_wait_seconds": 2.1517276763916016e-05} {"step": 15560, "total_steps": 28610, "tokens": 32631685120, "source": "fineweb_edu", "phase": 3, "loss": 3.3598146736621857, "ce_loss": 3.3455303013324738, "logit_z_loss": 23.429161548614502, "router_aux_loss": 1.298273891210556, "router_z_loss": 1.301634520292282, "routing_entropy_normalized": 0.6822428347841124, "learning_rate": 0.003, "grad_norm": 1.1334717273712158, "tokens_per_second": 3941207.547148034, "eta_seconds": 6944.0224277973175, "data_wait_seconds": 1.901388168334961e-05} {"step": 15570, "total_steps": 28610, "tokens": 32652656640, "source": "dclm", "phase": 3, "loss": 3.4288602769374847, "ce_loss": 3.4148367047309875, "logit_z_loss": 23.7604341506958, "router_aux_loss": 1.298660397529602, "router_z_loss": 1.0369970351457596, "routing_entropy_normalized": 0.6883334635854425, "learning_rate": 0.003, "grad_norm": 1.03175687789917, "tokens_per_second": 3719842.422699798, "eta_seconds": 7351.618421554565, "data_wait_seconds": 2.41696834564209e-05} {"step": 15580, "total_steps": 28610, "tokens": 32673628160, "source": "fineweb_edu", "phase": 3, "loss": 3.3461380898952484, "ce_loss": 3.3311774730682373, "logit_z_loss": 22.831200122833252, "router_aux_loss": 1.3850665539503098, "router_z_loss": 1.1099473536014557, "routing_entropy_normalized": 0.6645748198505476, "learning_rate": 0.003, "grad_norm": 1.7864081859588623, "tokens_per_second": 3976622.185514154, "eta_seconds": 6871.633583784103, "data_wait_seconds": 2.065300941467285e-05} {"step": 15590, "total_steps": 28610, "tokens": 32694599680, "source": "fineweb_hq", "phase": 3, "loss": 3.3569121658802032, "ce_loss": 3.3411965370178223, "logit_z_loss": 23.81728458404541, "router_aux_loss": 1.455195039510727, "router_z_loss": 1.1636397540569305, "routing_entropy_normalized": 0.6269326931741575, "learning_rate": 0.003, "grad_norm": 1.1756471395492554, "tokens_per_second": 4010917.3783049877, "eta_seconds": 6807.649339199066, "data_wait_seconds": 1.9937753677368164e-05} {"step": 15600, "total_steps": 28610, "tokens": 32715571200, "source": "fineweb_edu", "phase": 3, "loss": 3.3571415841579437, "ce_loss": 3.3420448005199432, "logit_z_loss": 22.951637744903564, "router_aux_loss": 1.3867106139659882, "router_z_loss": 1.2296382188796997, "routing_entropy_normalized": 0.6716673440065334, "learning_rate": 0.003, "grad_norm": 1.7389228343963623, "tokens_per_second": 4098873.04653579, "eta_seconds": 6656.450983047485, "data_wait_seconds": 2.0831823348999023e-05} {"step": 15610, "total_steps": 28610, "tokens": 32736542720, "source": "finemath", "phase": 3, "loss": 2.42252978682518, "ce_loss": 2.407450169324875, "logit_z_loss": 27.7249436378479, "router_aux_loss": 1.4152872264385223, "router_z_loss": 0.9267532005906105, "routing_entropy_normalized": 0.6709753984139772, "learning_rate": 0.003, "grad_norm": 0.8007329106330872, "tokens_per_second": 4060947.652559778, "eta_seconds": 6713.451719284058, "data_wait_seconds": 1.8864870071411133e-05} {"step": 15620, "total_steps": 28610, "tokens": 32757514240, "source": "dclm", "phase": 3, "loss": 3.4644049406051636, "ce_loss": 3.44923335313797, "logit_z_loss": 23.108746767044067, "router_aux_loss": 1.3964070826768875, "router_z_loss": 1.2075799405574799, "routing_entropy_normalized": 0.6453919866264491, "learning_rate": 0.003, "grad_norm": 1.2479496002197266, "tokens_per_second": 3397611.0914405053, "eta_seconds": 8017.987858772278, "data_wait_seconds": 2.2023916244506836e-05} {"step": 15630, "total_steps": 28610, "tokens": 32778485760, "source": "fineweb_hq", "phase": 3, "loss": 3.3216977417469025, "ce_loss": 3.306837797164917, "logit_z_loss": 23.989545822143555, "router_aux_loss": 1.3556303530931473, "router_z_loss": 1.3036994487047195, "routing_entropy_normalized": 0.6712328315204628, "learning_rate": 0.003, "grad_norm": 0.7032923102378845, "tokens_per_second": 4114757.3971094647, "eta_seconds": 6615.464858055115, "data_wait_seconds": 2.3096799850463867e-05} {"step": 15640, "total_steps": 28610, "tokens": 32799457280, "source": "dclm", "phase": 3, "loss": 3.4407127797603607, "ce_loss": 3.4253137409687042, "logit_z_loss": 24.00130009651184, "router_aux_loss": 1.424294888973236, "router_z_loss": 1.1560809314250946, "routing_entropy_normalized": 0.6586906664243011, "learning_rate": 0.003, "grad_norm": 1.2137609720230103, "tokens_per_second": 3861971.6351843495, "eta_seconds": 7043.0505475997925, "data_wait_seconds": 2.1845102310180664e-05} {"step": 15650, "total_steps": 28610, "tokens": 32820428800, "source": "fineweb_hq", "phase": 3, "loss": 3.346045643091202, "ce_loss": 3.3308939933776855, "logit_z_loss": 23.54197335243225, "router_aux_loss": 1.3871839791536331, "router_z_loss": 1.279793232679367, "routing_entropy_normalized": 0.6709334037632242, "learning_rate": 0.003, "grad_norm": 0.7548461556434631, "tokens_per_second": 3559878.4660825604, "eta_seconds": 7634.836463928223, "data_wait_seconds": 2.0772218704223633e-05} {"step": 15660, "total_steps": 28610, "tokens": 32841400320, "source": "fineweb_edu", "phase": 3, "loss": 3.3669266998767853, "ce_loss": 3.3531561195850372, "logit_z_loss": 23.300529956817627, "router_aux_loss": 1.251427784562111, "router_z_loss": 1.256362408399582, "routing_entropy_normalized": 0.7112039017019036, "learning_rate": 0.003, "grad_norm": 1.3782814741134644, "tokens_per_second": 3934368.407574071, "eta_seconds": 6902.78987288475, "data_wait_seconds": 2.1755695343017578e-05} {"step": 15670, "total_steps": 28610, "tokens": 32862371840, "source": "dclm", "phase": 3, "loss": 3.424141615629196, "ce_loss": 3.4102861881256104, "logit_z_loss": 23.93971872329712, "router_aux_loss": 1.2862138450145721, "router_z_loss": 0.9933059439063072, "routing_entropy_normalized": 0.6679988698462704, "learning_rate": 0.003, "grad_norm": 0.9777504205703735, "tokens_per_second": 3232738.987132137, "eta_seconds": 8394.475083827972, "data_wait_seconds": 2.187490463256836e-05} {"step": 15680, "total_steps": 28610, "tokens": 32883343360, "source": "fineweb_edu", "phase": 3, "loss": 3.330564647912979, "ce_loss": 3.3169606924057007, "logit_z_loss": 22.697891235351562, "router_aux_loss": 1.2598077952861786, "router_z_loss": 1.0058837458491325, "routing_entropy_normalized": 0.6610601955829, "learning_rate": 0.003, "grad_norm": 1.7765370607376099, "tokens_per_second": 3455809.6441873787, "eta_seconds": 7846.547741889954, "data_wait_seconds": 2.1398067474365234e-05} {"step": 15690, "total_steps": 28610, "tokens": 32904314880, "source": "fineweb_hq", "phase": 3, "loss": 3.2969251573085785, "ce_loss": 3.2829878330230713, "logit_z_loss": 22.917362451553345, "router_aux_loss": 1.2855218648910522, "router_z_loss": 1.0821176543831825, "routing_entropy_normalized": 0.6787404655194238, "learning_rate": 0.003, "grad_norm": 1.066478967666626, "tokens_per_second": 3884439.338253277, "eta_seconds": 6975.319082260132, "data_wait_seconds": 1.8805265426635742e-05} {"step": 15700, "total_steps": 28610, "tokens": 32925286400, "source": "fineweb_edu", "phase": 3, "loss": 3.3709070086479187, "ce_loss": 3.356859713792801, "logit_z_loss": 23.12873077392578, "router_aux_loss": 1.3048120588064194, "router_z_loss": 0.9991774931550026, "routing_entropy_normalized": 0.6756446520554448, "learning_rate": 0.003, "grad_norm": 1.048948884010315, "tokens_per_second": 4132772.715929805, "eta_seconds": 6551.1060445308685, "data_wait_seconds": 2.1398067474365234e-05} {"step": 15710, "total_steps": 28610, "tokens": 32946257920, "source": "finemath", "phase": 3, "loss": 2.2964300215244293, "ce_loss": 2.281482011079788, "logit_z_loss": 29.878751516342163, "router_aux_loss": 1.4324537962675095, "router_z_loss": 0.6234906986355782, "routing_entropy_normalized": 0.6588034973120825, "learning_rate": 0.003, "grad_norm": 0.8204534649848938, "tokens_per_second": 3916932.441203663, "eta_seconds": 6906.746850013733, "data_wait_seconds": 1.8656253814697266e-05} {"step": 15720, "total_steps": 28610, "tokens": 32967229440, "source": "dclm", "phase": 3, "loss": 3.395724266767502, "ce_loss": 3.3811183273792267, "logit_z_loss": 23.134957790374756, "router_aux_loss": 1.3790079057216644, "router_z_loss": 0.8158562481403351, "routing_entropy_normalized": 0.6707609120693304, "learning_rate": 0.003, "grad_norm": 1.100712537765503, "tokens_per_second": 3760725.563615681, "eta_seconds": 7188.051566839218, "data_wait_seconds": 1.990795135498047e-05} {"step": 15730, "total_steps": 28610, "tokens": 32988200960, "source": "fineweb_hq", "phase": 3, "loss": 3.339483767747879, "ce_loss": 3.3251847624778748, "logit_z_loss": 23.667876482009888, "router_aux_loss": 1.3387106359004974, "router_z_loss": 0.9119193777441978, "routing_entropy_normalized": 0.6623687559789087, "learning_rate": 0.003, "grad_norm": 1.4718278646469116, "tokens_per_second": 4415952.979638786, "eta_seconds": 6116.758462905884, "data_wait_seconds": 2.232193946838379e-05} {"step": 15740, "total_steps": 28610, "tokens": 33009172480, "source": "dclm", "phase": 3, "loss": 3.4681348502635956, "ce_loss": 3.454052120447159, "logit_z_loss": 24.20873737335205, "router_aux_loss": 1.323264718055725, "router_z_loss": 0.8501058593392372, "routing_entropy_normalized": 0.6754414780222459, "learning_rate": 0.003, "grad_norm": 1.965427041053772, "tokens_per_second": 4401148.220312265, "eta_seconds": 6132.569249868393, "data_wait_seconds": 2.2292137145996094e-05} {"step": 15750, "total_steps": 28610, "tokens": 33030144000, "source": "fineweb_hq", "phase": 3, "loss": 3.312833786010742, "ce_loss": 3.2986340820789337, "logit_z_loss": 23.47103452682495, "router_aux_loss": 1.340146079659462, "router_z_loss": 0.7982244789600372, "routing_entropy_normalized": 0.6712734059044659, "learning_rate": 0.003, "grad_norm": 1.1837141513824463, "tokens_per_second": 3743444.7414320344, "eta_seconds": 7204.4270939826965, "data_wait_seconds": 2.014636993408203e-05} {"step": 15760, "total_steps": 28610, "tokens": 33051115520, "source": "fineweb_edu", "phase": 3, "loss": 3.343835234642029, "ce_loss": 3.329647123813629, "logit_z_loss": 23.244518280029297, "router_aux_loss": 1.3265736401081085, "router_z_loss": 0.9223910421133041, "routing_entropy_normalized": 0.6424024274058702, "learning_rate": 0.003, "grad_norm": 1.2850226163864136, "tokens_per_second": 3594215.1980526065, "eta_seconds": 7497.715555429459, "data_wait_seconds": 2.136826515197754e-05} {"step": 15770, "total_steps": 28610, "tokens": 33072087040, "source": "dclm", "phase": 3, "loss": 3.435740888118744, "ce_loss": 3.420485645532608, "logit_z_loss": 23.35379981994629, "router_aux_loss": 1.4533884227275848, "router_z_loss": 0.7213882803916931, "routing_entropy_normalized": 0.628433320270389, "learning_rate": 0.003, "grad_norm": 2.0491321086883545, "tokens_per_second": 4291046.05933006, "eta_seconds": 6275.260462760925, "data_wait_seconds": 2.1755695343017578e-05} {"step": 15780, "total_steps": 28610, "tokens": 33093058560, "source": "fineweb_edu", "phase": 3, "loss": 3.379877507686615, "ce_loss": 3.364820271730423, "logit_z_loss": 22.6924045085907, "router_aux_loss": 1.4188375174999237, "router_z_loss": 0.8688961640000343, "routing_entropy_normalized": 0.6775476195759195, "learning_rate": 0.003, "grad_norm": 1.296336054801941, "tokens_per_second": 4386159.580026171, "eta_seconds": 6134.40064573288, "data_wait_seconds": 1.9043684005737305e-05} {"step": 15790, "total_steps": 28610, "tokens": 33114030080, "source": "fineweb_hq", "phase": 3, "loss": 3.386866956949234, "ce_loss": 3.3726554811000824, "logit_z_loss": 23.449830293655396, "router_aux_loss": 1.3212607204914093, "router_z_loss": 0.9989055320620537, "routing_entropy_normalized": 0.6702030343113533, "learning_rate": 0.003, "grad_norm": 0.715092122554779, "tokens_per_second": 3370882.8784674676, "eta_seconds": 7975.800290107728, "data_wait_seconds": 2.1696090698242188e-05} {"step": 15800, "total_steps": 28610, "tokens": 33135001600, "source": "fineweb_edu", "phase": 3, "loss": 3.3959511816501617, "ce_loss": 3.3810549080371857, "logit_z_loss": 23.059936046600342, "router_aux_loss": 1.393169328570366, "router_z_loss": 0.9646193981170654, "routing_entropy_normalized": 0.6427816633272502, "learning_rate": 0.003, "grad_norm": 2.0826003551483154, "tokens_per_second": 3845898.548318946, "eta_seconds": 6985.23811340332, "data_wait_seconds": 0.06677418947219849} {"step": 15810, "total_steps": 28610, "tokens": 33155973120, "source": "finemath", "phase": 3, "loss": 2.349000185728073, "ce_loss": 2.333672821521759, "logit_z_loss": 31.098989963531494, "router_aux_loss": 1.4644463807344437, "router_z_loss": 0.6828775778412819, "routing_entropy_normalized": 0.6456941217447522, "learning_rate": 0.003, "grad_norm": 0.9916135668754578, "tokens_per_second": 3931645.965643785, "eta_seconds": 6827.5592041015625, "data_wait_seconds": 1.862645149230957e-05} {"step": 15820, "total_steps": 28610, "tokens": 33176944640, "source": "dclm", "phase": 3, "loss": 3.429135173559189, "ce_loss": 3.414282977581024, "logit_z_loss": 24.150341510772705, "router_aux_loss": 1.39150670170784, "router_z_loss": 0.9371062889695168, "routing_entropy_normalized": 0.6733518052233429, "learning_rate": 0.003, "grad_norm": 1.2564959526062012, "tokens_per_second": 3393049.5073352703, "eta_seconds": 7905.152583837509, "data_wait_seconds": 0.1268974244594574} {"step": 15830, "total_steps": 28610, "tokens": 33197916160, "source": "fineweb_hq", "phase": 3, "loss": 3.344325989484787, "ce_loss": 3.328855484724045, "logit_z_loss": 24.064491033554077, "router_aux_loss": 1.4381581395864487, "router_z_loss": 1.088896244764328, "routing_entropy_normalized": 0.6543784382603665, "learning_rate": 0.003, "grad_norm": 1.4541149139404297, "tokens_per_second": 4209310.297831874, "eta_seconds": 6367.21948814392, "data_wait_seconds": 2.0235776901245117e-05} {"step": 15840, "total_steps": 28610, "tokens": 33218887680, "source": "dclm", "phase": 3, "loss": 3.4470141530036926, "ce_loss": 3.4323022663593292, "logit_z_loss": 23.478977918624878, "router_aux_loss": 1.3679590374231339, "router_z_loss": 1.0322325527668, "routing_entropy_normalized": 0.6594887219153422, "learning_rate": 0.003, "grad_norm": 0.9743286967277527, "tokens_per_second": 4248400.019149475, "eta_seconds": 6303.698079109192, "data_wait_seconds": 1.761317253112793e-05} {"step": 15850, "total_steps": 28610, "tokens": 33239859200, "source": "fineweb_hq", "phase": 3, "loss": 3.325543612241745, "ce_loss": 3.310752511024475, "logit_z_loss": 23.08675765991211, "router_aux_loss": 1.3634066432714462, "router_z_loss": 1.1570288687944412, "routing_entropy_normalized": 0.6491093864941165, "learning_rate": 0.003, "grad_norm": 1.0469574928283691, "tokens_per_second": 4482055.469592765, "eta_seconds": 5970.399005889893, "data_wait_seconds": 1.9341707229614258e-05} {"step": 15860, "total_steps": 28610, "tokens": 33260830720, "source": "fineweb_edu", "phase": 3, "loss": 3.3588567674160004, "ce_loss": 3.3447310626506805, "logit_z_loss": 23.086015939712524, "router_aux_loss": 1.2976738512516022, "router_z_loss": 1.1489690840244293, "routing_entropy_normalized": 0.6741228477191699, "learning_rate": 0.003, "grad_norm": 2.3077125549316406, "tokens_per_second": 3593640.1666865973, "eta_seconds": 7440.55797457695, "data_wait_seconds": 2.0623207092285156e-05} {"step": 15870, "total_steps": 28610, "tokens": 33281802240, "source": "dclm", "phase": 3, "loss": 3.4134643971920013, "ce_loss": 3.399106025695801, "logit_z_loss": 23.277777671813965, "router_aux_loss": 1.3377922028303146, "router_z_loss": 0.9804008528590202, "routing_entropy_normalized": 0.648820627826331, "learning_rate": 0.003, "grad_norm": 0.8941102027893066, "tokens_per_second": 4215165.648223898, "eta_seconds": 6338.473670959473, "data_wait_seconds": 1.8537044525146484e-05} {"step": 15880, "total_steps": 28610, "tokens": 33302773760, "source": "fineweb_edu", "phase": 3, "loss": 3.329227954149246, "ce_loss": 3.3126109540462494, "logit_z_loss": 23.44469141960144, "router_aux_loss": 1.5453258156776428, "router_z_loss": 1.1637576669454575, "routing_entropy_normalized": 0.6284268020206837, "learning_rate": 0.003, "grad_norm": 1.2049237489700317, "tokens_per_second": 4075632.4019580064, "eta_seconds": 6550.331906080246, "data_wait_seconds": 2.2232532501220703e-05} {"step": 15890, "total_steps": 28610, "tokens": 33323745280, "source": "fineweb_hq", "phase": 3, "loss": 3.348364979028702, "ce_loss": 3.3330191373825073, "logit_z_loss": 23.369855642318726, "router_aux_loss": 1.404148444533348, "router_z_loss": 1.3043392598628998, "routing_entropy_normalized": 0.6663844567351067, "learning_rate": 0.003, "grad_norm": 1.1027791500091553, "tokens_per_second": 4489210.250422403, "eta_seconds": 5942.197391510009, "data_wait_seconds": 2.2679567337036133e-05} {"step": 15900, "total_steps": 28610, "tokens": 33344716800, "source": "fineweb_edu", "phase": 3, "loss": 3.3786425292491913, "ce_loss": 3.362979918718338, "logit_z_loss": 23.810356378555298, "router_aux_loss": 1.4301346838474274, "router_z_loss": 1.36125747859478, "routing_entropy_normalized": 0.6421688042967877, "learning_rate": 0.003, "grad_norm": 1.8984650373458862, "tokens_per_second": 3778106.4663128844, "eta_seconds": 7055.069029331207, "data_wait_seconds": 2.1666288375854492e-05} {"step": 15910, "total_steps": 28610, "tokens": 33365688320, "source": "finemath", "phase": 3, "loss": 2.330755293369293, "ce_loss": 2.3148171603679657, "logit_z_loss": 27.674261808395386, "router_aux_loss": 1.499645784497261, "router_z_loss": 0.9417001754045486, "routing_entropy_normalized": 0.672600138115604, "learning_rate": 0.003, "grad_norm": 1.2649431228637695, "tokens_per_second": 3525890.801000884, "eta_seconds": 7553.7876534461975, "data_wait_seconds": 2.0295381546020508e-05} {"step": 15920, "total_steps": 28610, "tokens": 33386659840, "source": "dclm", "phase": 3, "loss": 3.4368798434734344, "ce_loss": 3.4214034974575043, "logit_z_loss": 23.350136756896973, "router_aux_loss": 1.4256408512592316, "router_z_loss": 1.2199632376432419, "routing_entropy_normalized": 0.6501564282511313, "learning_rate": 0.003, "grad_norm": 1.1018294095993042, "tokens_per_second": 3736138.248484121, "eta_seconds": 7123.092645406723, "data_wait_seconds": 2.205371856689453e-05} {"step": 15930, "total_steps": 28610, "tokens": 33407631360, "source": "fineweb_hq", "phase": 3, "loss": 3.3611272871494293, "ce_loss": 3.3463855981826782, "logit_z_loss": 23.234750270843506, "router_aux_loss": 1.3463169783353806, "router_z_loss": 1.2785281538963318, "routing_entropy_normalized": 0.6842044084495214, "learning_rate": 0.003, "grad_norm": 1.3924280405044556, "tokens_per_second": 3727856.7174720922, "eta_seconds": 7133.291157722473, "data_wait_seconds": 1.958012580871582e-05} {"step": 15940, "total_steps": 28610, "tokens": 33428602880, "source": "dclm", "phase": 3, "loss": 3.465131461620331, "ce_loss": 3.4502121806144714, "logit_z_loss": 23.741877794265747, "router_aux_loss": 1.3796947002410889, "router_z_loss": 1.1223571971058846, "routing_entropy_normalized": 0.6673929212520728, "learning_rate": 0.003, "grad_norm": 1.0397002696990967, "tokens_per_second": 4232187.709593199, "eta_seconds": 6278.293323278427, "data_wait_seconds": 1.7136335372924805e-05} {"step": 15950, "total_steps": 28610, "tokens": 33449574400, "source": "fineweb_hq", "phase": 3, "loss": 3.3201583921909332, "ce_loss": 3.305724114179611, "logit_z_loss": 24.123170137405396, "router_aux_loss": 1.313412845134735, "router_z_loss": 1.3001630306243896, "routing_entropy_normalized": 0.6896227895101787, "learning_rate": 0.003, "grad_norm": 0.9283549785614014, "tokens_per_second": 4407083.897698816, "eta_seconds": 6024.379144191742, "data_wait_seconds": 1.9878149032592773e-05} {"step": 15960, "total_steps": 28610, "tokens": 33470545920, "source": "fineweb_edu", "phase": 3, "loss": 3.3400003612041473, "ce_loss": 3.3251742124557495, "logit_z_loss": 23.822163343429565, "router_aux_loss": 1.3566325306892395, "router_z_loss": 1.2598310261964798, "routing_entropy_normalized": 0.657286008761755, "learning_rate": 0.003, "grad_norm": 1.116527795791626, "tokens_per_second": 4225047.707665912, "eta_seconds": 6278.9759159088135, "data_wait_seconds": 2.1904706954956055e-05} {"step": 15970, "total_steps": 28610, "tokens": 33491517440, "source": "dclm", "phase": 3, "loss": 3.474682867527008, "ce_loss": 3.460094302892685, "logit_z_loss": 24.18018102645874, "router_aux_loss": 1.3519328832626343, "router_z_loss": 1.0691983625292778, "routing_entropy_normalized": 0.671566659363142, "learning_rate": 0.003, "grad_norm": 1.2863086462020874, "tokens_per_second": 3633035.6509570237, "eta_seconds": 7296.377967834473, "data_wait_seconds": 2.1785497665405273e-05} {"step": 15980, "total_steps": 28610, "tokens": 33512488960, "source": "fineweb_edu", "phase": 3, "loss": 3.3563554286956787, "ce_loss": 3.341070204973221, "logit_z_loss": 22.85897135734558, "router_aux_loss": 1.411069095134735, "router_z_loss": 1.1745297312736511, "routing_entropy_normalized": 0.6648554014650047, "learning_rate": 0.003, "grad_norm": 0.9943889379501343, "tokens_per_second": 4453136.034327868, "eta_seconds": 5947.949839353561, "data_wait_seconds": 2.110004425048828e-05} {"step": 15990, "total_steps": 28610, "tokens": 33533460480, "source": "fineweb_hq", "phase": 3, "loss": 3.34562748670578, "ce_loss": 3.3310553431510925, "logit_z_loss": 23.010133981704712, "router_aux_loss": 1.342184767127037, "router_z_loss": 1.1503578126430511, "routing_entropy_normalized": 0.6725277461321882, "learning_rate": 0.003, "grad_norm": 1.6557198762893677, "tokens_per_second": 4004625.312537304, "eta_seconds": 6608.872534751892, "data_wait_seconds": 2.3037195205688477e-05} {"step": 16000, "total_steps": 28610, "tokens": 33554432000, "source": "fineweb_edu", "phase": 3, "loss": 3.3505365550518036, "ce_loss": 3.335191488265991, "logit_z_loss": 23.37950587272644, "router_aux_loss": 1.4132671505212784, "router_z_loss": 1.2123888283967972, "routing_entropy_normalized": 0.6371574037076597, "learning_rate": 0.003, "grad_norm": 2.1179842948913574, "tokens_per_second": 3941733.505607433, "eta_seconds": 6708.999145269394, "data_wait_seconds": 0.0668133795261383} {"step": 16010, "total_steps": 28610, "tokens": 33575403520, "source": "finemath", "phase": 3, "loss": 2.3873117566108704, "ce_loss": 2.3720515370368958, "logit_z_loss": 31.298017263412476, "router_aux_loss": 1.4388945400714874, "router_z_loss": 0.8712909892201424, "routing_entropy_normalized": 0.6757960218138749, "learning_rate": 0.003, "grad_norm": 1.1641114950180054, "tokens_per_second": 3689739.102719975, "eta_seconds": 7161.5131759643555, "data_wait_seconds": 2.1338462829589844e-05} {"step": 16020, "total_steps": 28610, "tokens": 33596375040, "source": "dclm", "phase": 3, "loss": 3.497350424528122, "ce_loss": 3.481945276260376, "logit_z_loss": 24.12409210205078, "router_aux_loss": 1.4273723512887955, "router_z_loss": 1.1313729584217072, "routing_entropy_normalized": 0.6605869593066392, "learning_rate": 0.003, "grad_norm": 1.008644938468933, "tokens_per_second": 4270410.973290455, "eta_seconds": 6182.810939073562, "data_wait_seconds": 0.008528470993041992} {"step": 16030, "total_steps": 28610, "tokens": 33617346560, "source": "fineweb_hq", "phase": 3, "loss": 3.2684717178344727, "ce_loss": 3.2535665035247803, "logit_z_loss": 24.25508999824524, "router_aux_loss": 1.3696824312210083, "router_z_loss": 1.2084035873413086, "routing_entropy_normalized": 0.6592728758504621, "learning_rate": 0.003, "grad_norm": 0.585244357585907, "tokens_per_second": 4266146.228622085, "eta_seconds": 6184.075919151306, "data_wait_seconds": 2.0444393157958984e-05} {"step": 16040, "total_steps": 28610, "tokens": 33638318080, "source": "dclm", "phase": 3, "loss": 3.525215744972229, "ce_loss": 3.510963201522827, "logit_z_loss": 22.65135669708252, "router_aux_loss": 1.3140851259231567, "router_z_loss": 1.1116793155670166, "routing_entropy_normalized": 0.6961569701763113, "learning_rate": 0.003, "grad_norm": 0.8748455047607422, "tokens_per_second": 3240498.2483067047, "eta_seconds": 8134.922045946121, "data_wait_seconds": 2.300739288330078e-05} {"step": 16050, "total_steps": 28610, "tokens": 33659289600, "source": "fineweb_hq", "phase": 3, "loss": 3.188465654850006, "ce_loss": 3.17376047372818, "logit_z_loss": 23.983225107192993, "router_aux_loss": 1.3552673161029816, "router_z_loss": 1.1525850743055344, "routing_entropy_normalized": 0.6619243078415014, "learning_rate": 0.003, "grad_norm": 1.1563633680343628, "tokens_per_second": 4445014.534936681, "eta_seconds": 5925.79144859314, "data_wait_seconds": 2.22623348236084e-05} {"step": 16060, "total_steps": 28610, "tokens": 33680261120, "source": "fineweb_edu", "phase": 3, "loss": 3.386167198419571, "ce_loss": 3.3711762130260468, "logit_z_loss": 23.648278951644897, "router_aux_loss": 1.3936442732810974, "router_z_loss": 1.0546159371733665, "routing_entropy_normalized": 0.6495259021289409, "learning_rate": 0.003, "grad_norm": 1.1936869621276855, "tokens_per_second": 3956214.4978741645, "eta_seconds": 6652.636659145355, "data_wait_seconds": 2.041459083557129e-05} {"step": 16070, "total_steps": 28610, "tokens": 33701232640, "source": "dclm", "phase": 3, "loss": 3.480583041906357, "ce_loss": 3.4658089578151703, "logit_z_loss": 23.89587092399597, "router_aux_loss": 1.3822138458490372, "router_z_loss": 0.9519114047288895, "routing_entropy_normalized": 0.6538117395872635, "learning_rate": 0.003, "grad_norm": 0.9515416622161865, "tokens_per_second": 4106156.093718147, "eta_seconds": 6404.599698543549, "data_wait_seconds": 2.0205974578857422e-05} {"step": 16080, "total_steps": 28610, "tokens": 33722204160, "source": "fineweb_edu", "phase": 3, "loss": 3.3382979035377502, "ce_loss": 3.323543965816498, "logit_z_loss": 23.223710536956787, "router_aux_loss": 1.370750218629837, "router_z_loss": 1.0464320853352547, "routing_entropy_normalized": 0.6634795644118561, "learning_rate": 0.003, "grad_norm": 1.1428800821304321, "tokens_per_second": 4036485.76610391, "eta_seconds": 6509.948525190353, "data_wait_seconds": 2.2202730178833008e-05} {"step": 16090, "total_steps": 28610, "tokens": 33743175680, "source": "fineweb_hq", "phase": 3, "loss": 3.201632797718048, "ce_loss": 3.186723530292511, "logit_z_loss": 24.774662256240845, "router_aux_loss": 1.3802998065948486, "router_z_loss": 1.106329157948494, "routing_entropy_normalized": 0.6698781202164283, "learning_rate": 0.003, "grad_norm": 1.1251784563064575, "tokens_per_second": 3598905.2759033334, "eta_seconds": 7295.64715576172, "data_wait_seconds": 2.2202730178833008e-05} {"step": 16100, "total_steps": 28610, "tokens": 33764147200, "source": "fineweb_edu", "phase": 3, "loss": 3.3812002539634705, "ce_loss": 3.366859495639801, "logit_z_loss": 22.888072729110718, "router_aux_loss": 1.327971562743187, "router_z_loss": 1.061031088232994, "routing_entropy_normalized": 0.6705867312051944, "learning_rate": 0.003, "grad_norm": 0.8453534841537476, "tokens_per_second": 4305588.518536376, "eta_seconds": 6093.329961061478, "data_wait_seconds": 0.014413625001907349} {"step": 16110, "total_steps": 28610, "tokens": 33785118720, "source": "finemath", "phase": 3, "loss": 2.206342473626137, "ce_loss": 2.1915372908115387, "logit_z_loss": 28.92070436477661, "router_aux_loss": 1.4106622636318207, "router_z_loss": 0.6986002996563911, "routing_entropy_normalized": 0.6712077700066753, "learning_rate": 0.003, "grad_norm": 1.4222491979599, "tokens_per_second": 3790294.080881504, "eta_seconds": 6916.191577911377, "data_wait_seconds": 1.8805265426635742e-05} {"step": 16120, "total_steps": 28610, "tokens": 33806090240, "source": "dclm", "phase": 3, "loss": 3.520031124353409, "ce_loss": 3.5050841867923737, "logit_z_loss": 24.201446294784546, "router_aux_loss": 1.410004198551178, "router_z_loss": 0.8468859121203423, "routing_entropy_normalized": 0.6494340835021432, "learning_rate": 0.003, "grad_norm": 1.4124755859375, "tokens_per_second": 3636724.7357626585, "eta_seconds": 7202.477609157562, "data_wait_seconds": 2.2023916244506836e-05} {"step": 16130, "total_steps": 28610, "tokens": 33827061760, "source": "fineweb_hq", "phase": 3, "loss": 3.2297887802124023, "ce_loss": 3.2148528397083282, "logit_z_loss": 23.707359790802002, "router_aux_loss": 1.3853119313716888, "router_z_loss": 1.0828521251678467, "routing_entropy_normalized": 0.6569121956693926, "learning_rate": 0.003, "grad_norm": 1.3978829383850098, "tokens_per_second": 4097698.3357296884, "eta_seconds": 6387.1116943359375, "data_wait_seconds": 1.9490718841552734e-05} {"step": 16140, "total_steps": 28610, "tokens": 33848033280, "source": "dclm", "phase": 3, "loss": 3.517927199602127, "ce_loss": 3.5025864839553833, "logit_z_loss": 23.659531354904175, "router_aux_loss": 1.4382174611091614, "router_z_loss": 0.9585195183753967, "routing_entropy_normalized": 0.656686907982674, "learning_rate": 0.003, "grad_norm": 1.9218029975891113, "tokens_per_second": 3507852.1947285626, "eta_seconds": 7455.12752199173, "data_wait_seconds": 2.1904706954956055e-05} {"step": 16150, "total_steps": 28610, "tokens": 33869004800, "source": "fineweb_hq", "phase": 3, "loss": 3.163788765668869, "ce_loss": 3.1480618119239807, "logit_z_loss": 24.54680562019348, "router_aux_loss": 1.4600731879472733, "router_z_loss": 1.1262061595916748, "routing_entropy_normalized": 0.6347731917763173, "learning_rate": 0.003, "grad_norm": 1.5487173795700073, "tokens_per_second": 3776258.0555705098, "eta_seconds": 6919.684390068054, "data_wait_seconds": 0.04308384656906128} {"step": 16160, "total_steps": 28610, "tokens": 33889976320, "source": "fineweb_edu", "phase": 3, "loss": 3.360332667827606, "ce_loss": 3.345222592353821, "logit_z_loss": 22.93986415863037, "router_aux_loss": 1.3971066176891327, "router_z_loss": 1.1389961242675781, "routing_entropy_normalized": 0.6592034573747706, "learning_rate": 0.003, "grad_norm": 0.7380515336990356, "tokens_per_second": 4440123.564065041, "eta_seconds": 5880.363918542861, "data_wait_seconds": 2.0682811737060547e-05} {"step": 16170, "total_steps": 28610, "tokens": 33910947840, "source": "dclm", "phase": 3, "loss": 3.461021453142166, "ce_loss": 3.4468428194522858, "logit_z_loss": 23.91389036178589, "router_aux_loss": 1.3135620653629303, "router_z_loss": 1.043010264635086, "routing_entropy_normalized": 0.6887021632906429, "learning_rate": 0.003, "grad_norm": 0.9153196811676025, "tokens_per_second": 3211904.127869532, "eta_seconds": 8122.462514877318, "data_wait_seconds": 0.08697131276130676} {"step": 16180, "total_steps": 28610, "tokens": 33931919360, "source": "fineweb_edu", "phase": 3, "loss": 3.403682619333267, "ce_loss": 3.3888826072216034, "logit_z_loss": 23.183202505111694, "router_aux_loss": 1.361568108201027, "router_z_loss": 1.1843205690383911, "routing_entropy_normalized": 0.6518027192712692, "learning_rate": 0.003, "grad_norm": 1.630953073501587, "tokens_per_second": 4405361.834698609, "eta_seconds": 5917.24365401268, "data_wait_seconds": 2.3990869522094727e-05} {"step": 16190, "total_steps": 28610, "tokens": 33952890880, "source": "fineweb_hq", "phase": 3, "loss": 3.202760726213455, "ce_loss": 3.1883855760097504, "logit_z_loss": 24.345495462417603, "router_aux_loss": 1.3203903436660767, "router_z_loss": 1.171236515045166, "routing_entropy_normalized": 0.6787450421011136, "learning_rate": 0.003, "grad_norm": 1.3483572006225586, "tokens_per_second": 4091029.8571363855, "eta_seconds": 6366.7655210494995, "data_wait_seconds": 2.09808349609375e-05} {"step": 16200, "total_steps": 28610, "tokens": 33973862400, "source": "fineweb_edu", "phase": 3, "loss": 3.401418536901474, "ce_loss": 3.3867613077163696, "logit_z_loss": 23.45465898513794, "router_aux_loss": 1.3498834669589996, "router_z_loss": 1.158376932144165, "routing_entropy_normalized": 0.6499889848028253, "learning_rate": 0.003, "grad_norm": 1.388746738433838, "tokens_per_second": 3358971.3859249665, "eta_seconds": 7748.103014230728, "data_wait_seconds": 0.3450869917869568} {"step": 16210, "total_steps": 28610, "tokens": 33994833920, "source": "finemath", "phase": 3, "loss": 2.3642673194408417, "ce_loss": 2.3485538363456726, "logit_z_loss": 30.118165969848633, "router_aux_loss": 1.490067183971405, "router_z_loss": 0.8127915859222412, "routing_entropy_normalized": 0.6802281008946623, "learning_rate": 0.003, "grad_norm": 1.8570743799209595, "tokens_per_second": 4396307.771380419, "eta_seconds": 5915.11926651001, "data_wait_seconds": 2.0563602447509766e-05} {"step": 16220, "total_steps": 28610, "tokens": 34015805440, "source": "dclm", "phase": 3, "loss": 3.4972874224185944, "ce_loss": 3.4810778498649597, "logit_z_loss": 24.527225732803345, "router_aux_loss": 1.5104796886444092, "router_z_loss": 1.1048016920685768, "routing_entropy_normalized": 0.6423866435865816, "learning_rate": 0.003, "grad_norm": 1.1784937381744385, "tokens_per_second": 4168978.841901586, "eta_seconds": 6232.632561922073, "data_wait_seconds": 0.00871807336807251} {"step": 16230, "total_steps": 28610, "tokens": 34036776960, "source": "fineweb_hq", "phase": 3, "loss": 3.229819566011429, "ce_loss": 3.2145588397979736, "logit_z_loss": 24.140424013137817, "router_aux_loss": 1.4035485088825226, "router_z_loss": 1.2252562493085861, "routing_entropy_normalized": 0.6539453966091032, "learning_rate": 0.003, "grad_norm": 1.5505930185317993, "tokens_per_second": 4247221.516579078, "eta_seconds": 6112.876773357392, "data_wait_seconds": 2.2172927856445312e-05} {"step": 16240, "total_steps": 28610, "tokens": 34057748480, "source": "dclm", "phase": 3, "loss": 3.5040794014930725, "ce_loss": 3.489178478717804, "logit_z_loss": 23.962627172470093, "router_aux_loss": 1.3747853934764862, "router_z_loss": 1.1531284898519516, "routing_entropy_normalized": 0.6468988354236462, "learning_rate": 0.003, "grad_norm": 1.0484791994094849, "tokens_per_second": 3232360.860466234, "eta_seconds": 8025.641739845276, "data_wait_seconds": 2.1398067474365234e-05} {"step": 16250, "total_steps": 28610, "tokens": 34078720000, "source": "fineweb_hq", "phase": 3, "loss": 3.188795417547226, "ce_loss": 3.173855572938919, "logit_z_loss": 23.750525951385498, "router_aux_loss": 1.3638872355222702, "router_z_loss": 1.3009538501501083, "routing_entropy_normalized": 0.6597671210749978, "learning_rate": 0.003, "grad_norm": 1.1059514284133911, "tokens_per_second": 4403906.789869279, "eta_seconds": 5885.864519119264, "data_wait_seconds": 1.9609928131103516e-05} {"step": 16260, "total_steps": 28610, "tokens": 34099691520, "source": "fineweb_edu", "phase": 3, "loss": 3.35082870721817, "ce_loss": 3.3360477685928345, "logit_z_loss": 23.62737488746643, "router_aux_loss": 1.3529471158981323, "router_z_loss": 1.2514342814683914, "routing_entropy_normalized": 0.6663398760751432, "learning_rate": 0.003, "grad_norm": 1.2319124937057495, "tokens_per_second": 3826458.384511238, "eta_seconds": 6768.615936040878, "data_wait_seconds": 2.300739288330078e-05} {"step": 16270, "total_steps": 28610, "tokens": 34120663040, "source": "dclm", "phase": 3, "loss": 3.5040787160396576, "ce_loss": 3.4899950325489044, "logit_z_loss": 24.04713749885559, "router_aux_loss": 1.3015516847372055, "router_z_loss": 1.0681247189640999, "routing_entropy_normalized": 0.6739967555266606, "learning_rate": 0.003, "grad_norm": 1.7968252897262573, "tokens_per_second": 3726235.0276317145, "eta_seconds": 6945.041171073914, "data_wait_seconds": 2.3633241653442383e-05} {"step": 16280, "total_steps": 28610, "tokens": 34141634560, "source": "fineweb_edu", "phase": 3, "loss": 3.3711818754673004, "ce_loss": 3.3567602932453156, "logit_z_loss": 22.959486961364746, "router_aux_loss": 1.322433277964592, "router_z_loss": 1.1972297877073288, "routing_entropy_normalized": 0.6638604788584349, "learning_rate": 0.003, "grad_norm": 1.4010043144226074, "tokens_per_second": 4199756.879909662, "eta_seconds": 6156.9954879283905, "data_wait_seconds": 2.1278858184814453e-05} {"step": 16290, "total_steps": 28610, "tokens": 34162606080, "source": "fineweb_hq", "phase": 3, "loss": 3.1930510997772217, "ce_loss": 3.1777986884117126, "logit_z_loss": 24.32119917869568, "router_aux_loss": 1.4128114879131317, "router_z_loss": 1.1243497729301453, "routing_entropy_normalized": 0.6483595852158104, "learning_rate": 0.003, "grad_norm": 0.9646409749984741, "tokens_per_second": 3550038.9455688647, "eta_seconds": 7277.92371749878, "data_wait_seconds": 1.9550323486328125e-05} {"step": 16300, "total_steps": 28610, "tokens": 34183577600, "source": "fineweb_edu", "phase": 3, "loss": 3.4151828289031982, "ce_loss": 3.400766909122467, "logit_z_loss": 24.223997831344604, "router_aux_loss": 1.3158230036497116, "router_z_loss": 1.2576977461576462, "routing_entropy_normalized": 0.6576826469967034, "learning_rate": 0.003, "grad_norm": 1.7471320629119873, "tokens_per_second": 4013845.0737933232, "eta_seconds": 6431.723358869553, "data_wait_seconds": 0.007635682821273804} {"step": 16310, "total_steps": 28610, "tokens": 34204549120, "source": "finemath", "phase": 3, "loss": 2.4082319736480713, "ce_loss": 2.393390506505966, "logit_z_loss": 28.026252269744873, "router_aux_loss": 1.3894789814949036, "router_z_loss": 0.9467085003852844, "routing_entropy_normalized": 0.6672004094607357, "learning_rate": 0.003, "grad_norm": 1.2631832361221313, "tokens_per_second": 3871502.605647748, "eta_seconds": 6662.779862880707, "data_wait_seconds": 2.0712614059448242e-05} {"step": 16320, "total_steps": 28610, "tokens": 34225520640, "source": "dclm", "phase": 3, "loss": 3.45183202624321, "ce_loss": 3.4369841814041138, "logit_z_loss": 24.292465686798096, "router_aux_loss": 1.3796739727258682, "router_z_loss": 1.0511477515101433, "routing_entropy_normalized": 0.6493381480960878, "learning_rate": 0.003, "grad_norm": 1.3952186107635498, "tokens_per_second": 3573301.247440248, "eta_seconds": 7212.937363862991, "data_wait_seconds": 2.104043960571289e-05} {"step": 16330, "total_steps": 28610, "tokens": 34246492160, "source": "fineweb_hq", "phase": 3, "loss": 3.2215331196784973, "ce_loss": 3.2071053981781006, "logit_z_loss": 24.943307638168335, "router_aux_loss": 1.3340425491333008, "router_z_loss": 1.0872717648744583, "routing_entropy_normalized": 0.6731532691748872, "learning_rate": 0.003, "grad_norm": 1.0719666481018066, "tokens_per_second": 3521394.4759641145, "eta_seconds": 7313.303503990173, "data_wait_seconds": 1.940131187438965e-05} {"step": 16340, "total_steps": 28610, "tokens": 34267463680, "source": "dclm", "phase": 3, "loss": 3.491458624601364, "ce_loss": 3.4760765433311462, "logit_z_loss": 23.914978504180908, "router_aux_loss": 1.448250189423561, "router_z_loss": 0.8996250554919243, "routing_entropy_normalized": 0.6476163499393937, "learning_rate": 0.003, "grad_norm": 1.5138951539993286, "tokens_per_second": 4121288.781314623, "eta_seconds": 6243.691331863403, "data_wait_seconds": 2.2917985916137695e-05} {"step": 16350, "total_steps": 28610, "tokens": 34288435200, "source": "fineweb_hq", "phase": 3, "loss": 3.164556473493576, "ce_loss": 3.149338275194168, "logit_z_loss": 24.763586044311523, "router_aux_loss": 1.4087933599948883, "router_z_loss": 1.1302945613861084, "routing_entropy_normalized": 0.6589511209564493, "learning_rate": 0.003, "grad_norm": 0.994910478591919, "tokens_per_second": 3650331.11888396, "eta_seconds": 7043.4935030937195, "data_wait_seconds": 0.04220965504646301} {"step": 16360, "total_steps": 28610, "tokens": 34309406720, "source": "fineweb_edu", "phase": 3, "loss": 3.381431818008423, "ce_loss": 3.3667101860046387, "logit_z_loss": 23.059866189956665, "router_aux_loss": 1.3560027480125427, "router_z_loss": 1.1616291999816895, "routing_entropy_normalized": 0.6637208560593866, "learning_rate": 0.003, "grad_norm": 0.8836308717727661, "tokens_per_second": 4142907.0315815136, "eta_seconds": 6200.986844301224, "data_wait_seconds": 1.8090009689331055e-05} {"step": 16370, "total_steps": 28610, "tokens": 34330378240, "source": "dclm", "phase": 3, "loss": 3.5234238505363464, "ce_loss": 3.508963942527771, "logit_z_loss": 23.14152216911316, "router_aux_loss": 1.3464851081371307, "router_z_loss": 0.9950832054018974, "routing_entropy_normalized": 0.6668423772276718, "learning_rate": 0.003, "grad_norm": 1.0910732746124268, "tokens_per_second": 3727761.9260984524, "eta_seconds": 6885.938798904419, "data_wait_seconds": 2.0444393157958984e-05} {"step": 16380, "total_steps": 28610, "tokens": 34351349760, "source": "fineweb_edu", "phase": 3, "loss": 3.3594875633716583, "ce_loss": 3.34515118598938, "logit_z_loss": 23.048802852630615, "router_aux_loss": 1.3226820826530457, "router_z_loss": 1.1095343977212906, "routing_entropy_normalized": 0.6618896548268598, "learning_rate": 0.003, "grad_norm": 1.1519855260849, "tokens_per_second": 4098966.0668790783, "eta_seconds": 6257.228906393051, "data_wait_seconds": 2.282857894897461e-05} {"step": 16390, "total_steps": 28610, "tokens": 34372321280, "source": "fineweb_hq", "phase": 3, "loss": 3.1969917118549347, "ce_loss": 3.18277844786644, "logit_z_loss": 24.145172357559204, "router_aux_loss": 1.316114991903305, "router_z_loss": 1.0520830899477005, "routing_entropy_normalized": 0.6896892586460956, "learning_rate": 0.003, "grad_norm": 1.0500017404556274, "tokens_per_second": 4323804.198729299, "eta_seconds": 5927.0023021698, "data_wait_seconds": 2.2172927856445312e-05} {"step": 16400, "total_steps": 28610, "tokens": 34393292800, "source": "fineweb_edu", "phase": 3, "loss": 3.372037887573242, "ce_loss": 3.3571287095546722, "logit_z_loss": 23.745017766952515, "router_aux_loss": 1.384579986333847, "router_z_loss": 1.0632818713784218, "routing_entropy_normalized": 0.6764417600234014, "learning_rate": 0.003, "grad_norm": 2.077948570251465, "tokens_per_second": 3633957.8208414945, "eta_seconds": 7046.3740038871765, "data_wait_seconds": 0.0478476881980896} {"step": 16410, "total_steps": 28610, "tokens": 34414264320, "source": "finemath", "phase": 3, "loss": 2.36148139834404, "ce_loss": 2.3465220034122467, "logit_z_loss": 30.7467942237854, "router_aux_loss": 1.4212905168533325, "router_z_loss": 0.7465202584862709, "routing_entropy_normalized": 0.667158325170185, "learning_rate": 0.003, "grad_norm": 1.1034411191940308, "tokens_per_second": 4267101.953458227, "eta_seconds": 5995.932292938232, "data_wait_seconds": 1.9788742065429688e-05} {"step": 16420, "total_steps": 28610, "tokens": 34435235840, "source": "dclm", "phase": 3, "loss": 3.4779884219169617, "ce_loss": 3.4632232189178467, "logit_z_loss": 24.022762060165405, "router_aux_loss": 1.3813546746969223, "router_z_loss": 0.9516643956303596, "routing_entropy_normalized": 0.6550889201601977, "learning_rate": 0.003, "grad_norm": 0.8339880108833313, "tokens_per_second": 3575960.8299865294, "eta_seconds": 7148.926986455917, "data_wait_seconds": 0.16608697175979614} {"step": 16430, "total_steps": 28610, "tokens": 34456207360, "source": "fineweb_hq", "phase": 3, "loss": 3.1932537257671356, "ce_loss": 3.1784711480140686, "logit_z_loss": 24.68470025062561, "router_aux_loss": 1.3763567507266998, "router_z_loss": 1.0189827382564545, "routing_entropy_normalized": 0.6585329180007685, "learning_rate": 0.003, "grad_norm": 0.6158280968666077, "tokens_per_second": 4397609.608589387, "eta_seconds": 5808.453599452972, "data_wait_seconds": 2.2709369659423828e-05} {"step": 16440, "total_steps": 28610, "tokens": 34477178880, "source": "dclm", "phase": 3, "loss": 3.498369723558426, "ce_loss": 3.4832129180431366, "logit_z_loss": 25.015209674835205, "router_aux_loss": 1.4283073097467422, "router_z_loss": 0.8737277165055275, "routing_entropy_normalized": 0.632704315665089, "learning_rate": 0.003, "grad_norm": 0.9036466479301453, "tokens_per_second": 3278219.0224466184, "eta_seconds": 7785.428510189056, "data_wait_seconds": 2.282857894897461e-05} {"step": 16450, "total_steps": 28610, "tokens": 34498150400, "source": "fineweb_hq", "phase": 3, "loss": 3.1980086266994476, "ce_loss": 3.1829899847507477, "logit_z_loss": 24.86201310157776, "router_aux_loss": 1.3963271379470825, "router_z_loss": 1.0553216710686684, "routing_entropy_normalized": 0.6418261435081906, "learning_rate": 0.003, "grad_norm": 1.332248568534851, "tokens_per_second": 4617381.602872741, "eta_seconds": 5522.906814575195, "data_wait_seconds": 1.8328428268432617e-05} {"step": 16460, "total_steps": 28610, "tokens": 34519121920, "source": "fineweb_edu", "phase": 3, "loss": 3.3768831491470337, "ce_loss": 3.362192541360855, "logit_z_loss": 23.338468074798584, "router_aux_loss": 1.3592988103628159, "router_z_loss": 1.0976213812828064, "routing_entropy_normalized": 0.6634824831814389, "learning_rate": 0.003, "grad_norm": 1.4373559951782227, "tokens_per_second": 3230177.3454399803, "eta_seconds": 7888.234630823136, "data_wait_seconds": 2.1576881408691406e-05} {"step": 16470, "total_steps": 28610, "tokens": 34540093440, "source": "dclm", "phase": 3, "loss": 3.5228021144866943, "ce_loss": 3.5074746012687683, "logit_z_loss": 25.035184860229492, "router_aux_loss": 1.4338628351688385, "router_z_loss": 0.9888848215341568, "routing_entropy_normalized": 0.643812911046065, "learning_rate": 0.003, "grad_norm": 1.5148444175720215, "tokens_per_second": 4262001.505158658, "eta_seconds": 5973.584300518036, "data_wait_seconds": 1.7702579498291016e-05} {"step": 16480, "total_steps": 28610, "tokens": 34561064960, "source": "fineweb_edu", "phase": 3, "loss": 3.3565946221351624, "ce_loss": 3.3409626185894012, "logit_z_loss": 23.497623443603516, "router_aux_loss": 1.460237056016922, "router_z_loss": 1.029642477631569, "routing_entropy_normalized": 0.6313978234792005, "learning_rate": 0.003, "grad_norm": 1.2126648426055908, "tokens_per_second": 3696325.847347475, "eta_seconds": 6882.091788053513, "data_wait_seconds": 2.181529998779297e-05} {"step": 16490, "total_steps": 28610, "tokens": 34582036480, "source": "fineweb_hq", "phase": 3, "loss": 3.1968836784362793, "ce_loss": 3.182425707578659, "logit_z_loss": 23.896583080291748, "router_aux_loss": 1.3273558616638184, "router_z_loss": 1.1843930333852768, "routing_entropy_normalized": 0.6961507664895893, "learning_rate": 0.003, "grad_norm": 0.84786456823349, "tokens_per_second": 3207928.116549988, "eta_seconds": 7923.332854270935, "data_wait_seconds": 1.817941665649414e-05} {"step": 16500, "total_steps": 28610, "tokens": 34603008000, "source": "fineweb_edu", "phase": 3, "loss": 3.3472781479358673, "ce_loss": 3.3335200548171997, "logit_z_loss": 23.960564613342285, "router_aux_loss": 1.2623745948076248, "router_z_loss": 1.1343843936920166, "routing_entropy_normalized": 0.7118696302163944, "learning_rate": 0.003, "grad_norm": 1.1804144382476807, "tokens_per_second": 3827445.56748025, "eta_seconds": 6635.368229866028, "data_wait_seconds": 0.07538923621177673} {"step": 16510, "total_steps": 28610, "tokens": 34623979520, "source": "finemath", "phase": 3, "loss": 2.3385778814554214, "ce_loss": 2.324807718396187, "logit_z_loss": 31.462148904800415, "router_aux_loss": 1.3050945699214935, "router_z_loss": 0.7192034497857094, "routing_entropy_normalized": 0.7132283173963863, "learning_rate": 0.003, "grad_norm": 0.9810706973075867, "tokens_per_second": 3624843.474109512, "eta_seconds": 7000.451021194458, "data_wait_seconds": 1.6689300537109375e-05} {"step": 16520, "total_steps": 28610, "tokens": 34644951040, "source": "dclm", "phase": 3, "loss": 3.4587305188179016, "ce_loss": 3.444437652826309, "logit_z_loss": 24.022961378097534, "router_aux_loss": 1.3422244191169739, "router_z_loss": 0.8705893084406853, "routing_entropy_normalized": 0.6683145553030414, "learning_rate": 0.003, "grad_norm": 1.1247459650039673, "tokens_per_second": 3149188.5527018076, "eta_seconds": 8051.143097877502, "data_wait_seconds": 1.722574234008789e-05} {"step": 16530, "total_steps": 28610, "tokens": 34665922560, "source": "fineweb_hq", "phase": 3, "loss": 3.1883342266082764, "ce_loss": 3.1741254329681396, "logit_z_loss": 24.858107089996338, "router_aux_loss": 1.3311150521039963, "router_z_loss": 0.8976667150855064, "routing_entropy_normalized": 0.7011603297287406, "learning_rate": 0.003, "grad_norm": 1.7194571495056152, "tokens_per_second": 3715140.8390131956, "eta_seconds": 6819.013668060303, "data_wait_seconds": 2.130866050720215e-05} {"step": 16540, "total_steps": 28610, "tokens": 34686894080, "source": "dclm", "phase": 3, "loss": 3.4733843207359314, "ce_loss": 3.45823010802269, "logit_z_loss": 23.78957200050354, "router_aux_loss": 1.4285405278205872, "router_z_loss": 0.8688132017850876, "routing_entropy_normalized": 0.6454449244646339, "learning_rate": 0.003, "grad_norm": 1.8279255628585815, "tokens_per_second": 3962785.986684531, "eta_seconds": 6387.583060264587, "data_wait_seconds": 1.7583370208740234e-05} {"step": 16550, "total_steps": 28610, "tokens": 34707865600, "source": "fineweb_hq", "phase": 3, "loss": 3.151253253221512, "ce_loss": 3.1366093158721924, "logit_z_loss": 25.271846055984497, "router_aux_loss": 1.3656420856714249, "router_z_loss": 0.9875043332576752, "routing_entropy_normalized": 0.6969811377218553, "learning_rate": 0.003, "grad_norm": 0.7366398572921753, "tokens_per_second": 3829703.232576689, "eta_seconds": 6604.076499938965, "data_wait_seconds": 0.028787672519683838} {"step": 16560, "total_steps": 28610, "tokens": 34728837120, "source": "fineweb_edu", "phase": 3, "loss": 3.3082345128059387, "ce_loss": 3.293454885482788, "logit_z_loss": 23.40897512435913, "router_aux_loss": 1.3684909492731094, "router_z_loss": 1.09474166482687, "routing_entropy_normalized": 0.6693059765660062, "learning_rate": 0.003, "grad_norm": 1.7537899017333984, "tokens_per_second": 4294970.232014807, "eta_seconds": 5883.785040378571, "data_wait_seconds": 1.9043684005737305e-05} {"step": 16570, "total_steps": 28610, "tokens": 34749808640, "source": "dclm", "phase": 3, "loss": 3.4503557682037354, "ce_loss": 3.4361009001731873, "logit_z_loss": 23.748566389083862, "router_aux_loss": 1.3393258154392242, "router_z_loss": 0.8615889400243759, "routing_entropy_normalized": 0.6836728457343216, "learning_rate": 0.003, "grad_norm": 1.2627326250076294, "tokens_per_second": 3294938.499142917, "eta_seconds": 7663.180993080139, "data_wait_seconds": 2.0623207092285156e-05} {"step": 16580, "total_steps": 28610, "tokens": 34770780160, "source": "fineweb_edu", "phase": 3, "loss": 3.3861631751060486, "ce_loss": 3.3708876967430115, "logit_z_loss": 23.030425310134888, "router_aux_loss": 1.4279394298791885, "router_z_loss": 0.9960620999336243, "routing_entropy_normalized": 0.6736526509051262, "learning_rate": 0.003, "grad_norm": 0.9488274455070496, "tokens_per_second": 3974879.273676672, "eta_seconds": 6347.045236587524, "data_wait_seconds": 1.84476375579834e-05} {"step": 16590, "total_steps": 28610, "tokens": 34791751680, "source": "fineweb_hq", "phase": 3, "loss": 3.253788709640503, "ce_loss": 3.237708240747452, "logit_z_loss": 22.562780618667603, "router_aux_loss": 1.5070017576217651, "router_z_loss": 1.010452851653099, "routing_entropy_normalized": 0.6556847267755294, "learning_rate": 0.003, "grad_norm": 1.629572868347168, "tokens_per_second": 3973790.347915309, "eta_seconds": 6343.507038116455, "data_wait_seconds": 2.0205974578857422e-05} {"step": 16600, "total_steps": 28610, "tokens": 34812723200, "source": "fineweb_edu", "phase": 3, "loss": 3.371510326862335, "ce_loss": 3.355869710445404, "logit_z_loss": 23.611371755599976, "router_aux_loss": 1.4497039020061493, "router_z_loss": 1.143627941608429, "routing_entropy_normalized": 0.6614251711071866, "learning_rate": 0.003, "grad_norm": 1.1154603958129883, "tokens_per_second": 2986459.8351909895, "eta_seconds": 8433.662901878357, "data_wait_seconds": 0.07349833846092224} {"step": 16610, "total_steps": 28610, "tokens": 34833694720, "source": "finemath", "phase": 3, "loss": 2.28911155462265, "ce_loss": 2.274106979370117, "logit_z_loss": 30.195400953292847, "router_aux_loss": 1.4146303832530975, "router_z_loss": 0.8582297042012215, "routing_entropy_normalized": 0.6707849632152639, "learning_rate": 0.003, "grad_norm": 0.9286862015724182, "tokens_per_second": 4350296.353241911, "eta_seconds": 5784.852790832519, "data_wait_seconds": 1.8298625946044922e-05} {"step": 16620, "total_steps": 28610, "tokens": 34854666240, "source": "dclm", "phase": 3, "loss": 3.5081276297569275, "ce_loss": 3.4937807619571686, "logit_z_loss": 24.27822232246399, "router_aux_loss": 1.321828842163086, "router_z_loss": 1.1285854801535606, "routing_entropy_normalized": 0.6743280263650855, "learning_rate": 0.003, "grad_norm": 1.030570149421692, "tokens_per_second": 3439058.107702138, "eta_seconds": 7311.552085638046, "data_wait_seconds": 2.1845102310180664e-05} {"step": 16630, "total_steps": 28610, "tokens": 34875637760, "source": "fineweb_hq", "phase": 3, "loss": 3.3196235299110413, "ce_loss": 3.3054419457912445, "logit_z_loss": 23.68265390396118, "router_aux_loss": 1.3107303529977798, "router_z_loss": 1.0742566734552383, "routing_entropy_normalized": 0.6743865710301341, "learning_rate": 0.003, "grad_norm": 1.5394657850265503, "tokens_per_second": 4251237.670385059, "eta_seconds": 5909.780376434326, "data_wait_seconds": 2.0951032638549805e-05} {"step": 16640, "total_steps": 28610, "tokens": 34896609280, "source": "dclm", "phase": 3, "loss": 3.537846565246582, "ce_loss": 3.522986978292465, "logit_z_loss": 25.06310224533081, "router_aux_loss": 1.38711117208004, "router_z_loss": 0.9884423017501831, "routing_entropy_normalized": 0.6658152909631679, "learning_rate": 0.003, "grad_norm": 0.9173257946968079, "tokens_per_second": 3811909.4050147315, "eta_seconds": 6585.38983297348, "data_wait_seconds": 2.2679567337036133e-05} {"step": 16650, "total_steps": 28610, "tokens": 34917580800, "source": "fineweb_hq", "phase": 3, "loss": 3.3030871748924255, "ce_loss": 3.2881966531276703, "logit_z_loss": 23.629663705825806, "router_aux_loss": 1.381120204925537, "router_z_loss": 1.0793352350592613, "routing_entropy_normalized": 0.6735601131616968, "learning_rate": 0.003, "grad_norm": 0.8126566410064697, "tokens_per_second": 3868520.180548384, "eta_seconds": 6483.6001234054565, "data_wait_seconds": 2.0295381546020508e-05} {"step": 16660, "total_steps": 28610, "tokens": 34938552320, "source": "fineweb_edu", "phase": 3, "loss": 3.386184513568878, "ce_loss": 3.3707616925239563, "logit_z_loss": 23.53026533126831, "router_aux_loss": 1.4215259552001953, "router_z_loss": 1.2075544446706772, "routing_entropy_normalized": 0.6488764939747359, "learning_rate": 0.003, "grad_norm": 0.9081089496612549, "tokens_per_second": 3944567.228781583, "eta_seconds": 6353.286671638489, "data_wait_seconds": 2.181529998779297e-05} {"step": 16670, "total_steps": 28610, "tokens": 34959523840, "source": "dclm", "phase": 3, "loss": 3.50192391872406, "ce_loss": 3.4868191182613373, "logit_z_loss": 24.95224118232727, "router_aux_loss": 1.4192331731319427, "router_z_loss": 0.912462517619133, "routing_entropy_normalized": 0.6604315688422779, "learning_rate": 0.003, "grad_norm": 1.7089219093322754, "tokens_per_second": 4150767.250173619, "eta_seconds": 6032.618398189545, "data_wait_seconds": 1.990795135498047e-05} {"step": 16680, "total_steps": 28610, "tokens": 34980495360, "source": "fineweb_edu", "phase": 3, "loss": 3.3526014387607574, "ce_loss": 3.3380569517612457, "logit_z_loss": 23.92281675338745, "router_aux_loss": 1.3429780304431915, "router_z_loss": 1.1147288531064987, "routing_entropy_normalized": 0.6888768310556646, "learning_rate": 0.003, "grad_norm": 1.6874616146087646, "tokens_per_second": 4067942.345993984, "eta_seconds": 6150.289564609528, "data_wait_seconds": 2.0891427993774414e-05} {"step": 16690, "total_steps": 28610, "tokens": 35001466880, "source": "fineweb_hq", "phase": 3, "loss": 3.3134827315807343, "ce_loss": 3.2986490726470947, "logit_z_loss": 24.896633863449097, "router_aux_loss": 1.3743696957826614, "router_z_loss": 1.0899679362773895, "routing_entropy_normalized": 0.660528857124002, "learning_rate": 0.003, "grad_norm": 1.4998223781585693, "tokens_per_second": 4311461.161188319, "eta_seconds": 5798.0463943481445, "data_wait_seconds": 2.193450927734375e-05} {"step": 16700, "total_steps": 28610, "tokens": 35022438400, "source": "fineweb_edu", "phase": 3, "loss": 3.373777151107788, "ce_loss": 3.358292430639267, "logit_z_loss": 23.722036361694336, "router_aux_loss": 1.4374773055315018, "router_z_loss": 1.1099149286746979, "routing_entropy_normalized": 0.6381559511360192, "learning_rate": 0.003, "grad_norm": 2.1892192363739014, "tokens_per_second": 3896051.3928781236, "eta_seconds": 6410.870340585709, "data_wait_seconds": 2.1964311599731445e-05} {"step": 16710, "total_steps": 28610, "tokens": 35043409920, "source": "finemath", "phase": 3, "loss": 2.243908703327179, "ce_loss": 2.2286463528871536, "logit_z_loss": 30.72889471054077, "router_aux_loss": 1.4523383527994156, "router_z_loss": 0.7389748468995094, "routing_entropy_normalized": 0.6986262664132061, "learning_rate": 0.003, "grad_norm": 1.4478143453598022, "tokens_per_second": 3935168.9217854077, "eta_seconds": 6341.813857555389, "data_wait_seconds": 2.086162567138672e-05} {"step": 16720, "total_steps": 28610, "tokens": 35064381440, "source": "dclm", "phase": 3, "loss": 3.508786678314209, "ce_loss": 3.4930533170700073, "logit_z_loss": 24.72028875350952, "router_aux_loss": 1.4709793776273727, "router_z_loss": 1.023629516363144, "routing_entropy_normalized": 0.6489750203711174, "learning_rate": 0.003, "grad_norm": 0.994884729385376, "tokens_per_second": 3809998.733569121, "eta_seconds": 6544.657629489899, "data_wait_seconds": 1.952052116394043e-05} {"step": 16730, "total_steps": 28610, "tokens": 35085352960, "source": "fineweb_hq", "phase": 3, "loss": 3.329077333211899, "ce_loss": 3.3144702315330505, "logit_z_loss": 23.873623847961426, "router_aux_loss": 1.3465027511119843, "router_z_loss": 1.142005294561386, "routing_entropy_normalized": 0.6592673539259949, "learning_rate": 0.003, "grad_norm": 2.3824000358581543, "tokens_per_second": 3586783.5121791135, "eta_seconds": 6946.10245513916, "data_wait_seconds": 2.130866050720215e-05} {"step": 16740, "total_steps": 28610, "tokens": 35106324480, "source": "dclm", "phase": 3, "loss": 3.526482403278351, "ce_loss": 3.5105466842651367, "logit_z_loss": 23.190776348114014, "router_aux_loss": 1.491139143705368, "router_z_loss": 1.02431321144104, "routing_entropy_normalized": 0.6489657209611178, "learning_rate": 0.003, "grad_norm": 2.136908531188965, "tokens_per_second": 4181827.4241437316, "eta_seconds": 5952.707205533981, "data_wait_seconds": 2.384185791015625e-05} {"step": 16750, "total_steps": 28610, "tokens": 35127296000, "source": "fineweb_hq", "phase": 3, "loss": 3.3014229238033295, "ce_loss": 3.2859479188919067, "logit_z_loss": 24.52215576171875, "router_aux_loss": 1.4224577993154526, "router_z_loss": 1.2504131644964218, "routing_entropy_normalized": 0.6539988936487492, "learning_rate": 0.003, "grad_norm": 0.8522239923477173, "tokens_per_second": 3879859.4647403196, "eta_seconds": 6410.598875045776, "data_wait_seconds": 1.850724220275879e-05} {"step": 16760, "total_steps": 28610, "tokens": 35148267520, "source": "fineweb_edu", "phase": 3, "loss": 3.3740123510360718, "ce_loss": 3.357405364513397, "logit_z_loss": 23.92351770401001, "router_aux_loss": 1.525876224040985, "router_z_loss": 1.348234847187996, "routing_entropy_normalized": 0.6376584515575862, "learning_rate": 0.003, "grad_norm": 1.0134762525558472, "tokens_per_second": 3502039.9573946865, "eta_seconds": 7096.221488714218, "data_wait_seconds": 1.990795135498047e-05} {"step": 16770, "total_steps": 28610, "tokens": 35169239040, "source": "dclm", "phase": 3, "loss": 3.497487187385559, "ce_loss": 3.480871796607971, "logit_z_loss": 24.28392791748047, "router_aux_loss": 1.5318014025688171, "router_z_loss": 1.297391027212143, "routing_entropy_normalized": 0.6598386142336792, "learning_rate": 0.003, "grad_norm": 0.9760605096817017, "tokens_per_second": 4026413.755783102, "eta_seconds": 6166.847518920898, "data_wait_seconds": 4.44948673248291e-05} {"step": 16780, "total_steps": 28610, "tokens": 35190210560, "source": "fineweb_edu", "phase": 3, "loss": 3.357674777507782, "ce_loss": 3.3415194749832153, "logit_z_loss": 23.413382530212402, "router_aux_loss": 1.4604689478874207, "router_z_loss": 1.5506199300289154, "routing_entropy_normalized": 0.650407476023193, "learning_rate": 0.003, "grad_norm": 1.0655854940414429, "tokens_per_second": 4397092.561872702, "eta_seconds": 5642.207392930984, "data_wait_seconds": 1.9878149032592773e-05} {"step": 16790, "total_steps": 28610, "tokens": 35211182080, "source": "fineweb_hq", "phase": 3, "loss": 3.2985987067222595, "ce_loss": 3.283702313899994, "logit_z_loss": 23.68810486793518, "router_aux_loss": 1.3358538150787354, "router_z_loss": 1.5378309190273285, "routing_entropy_normalized": 0.696629719395984, "learning_rate": 0.003, "grad_norm": 1.0498806238174438, "tokens_per_second": 3571387.6014508386, "eta_seconds": 6940.813881397247, "data_wait_seconds": 2.3812055587768555e-05} {"step": 16800, "total_steps": 28610, "tokens": 35232153600, "source": "fineweb_edu", "phase": 3, "loss": 3.4111360609531403, "ce_loss": 3.3956074714660645, "logit_z_loss": 23.936083555221558, "router_aux_loss": 1.3917942643165588, "router_z_loss": 1.6106503754854202, "routing_entropy_normalized": 0.6457887804139388, "learning_rate": 0.003, "grad_norm": 1.8724524974822998, "tokens_per_second": 3863945.6662802175, "eta_seconds": 6409.86371421814, "data_wait_seconds": 2.0712614059448242e-05} {"step": 16810, "total_steps": 28610, "tokens": 35253125120, "source": "finemath", "phase": 3, "loss": 2.3760499358177185, "ce_loss": 2.361165016889572, "logit_z_loss": 31.132182836532593, "router_aux_loss": 1.3762781769037247, "router_z_loss": 1.1221756488084793, "routing_entropy_normalized": 0.6749764920788383, "learning_rate": 0.003, "grad_norm": 1.4519003629684448, "tokens_per_second": 4266755.044966349, "eta_seconds": 5799.815864562988, "data_wait_seconds": 2.232193946838379e-05} {"step": 16820, "total_steps": 28610, "tokens": 35274096640, "source": "dclm", "phase": 3, "loss": 3.4833777248859406, "ce_loss": 3.468545526266098, "logit_z_loss": 24.83040142059326, "router_aux_loss": 1.362105742096901, "router_z_loss": 1.2111667841672897, "routing_entropy_normalized": 0.68133450416862, "learning_rate": 0.003, "grad_norm": 1.1649160385131836, "tokens_per_second": 3959317.685795184, "eta_seconds": 6244.869455337524, "data_wait_seconds": 1.9282102584838867e-05} {"step": 16830, "total_steps": 28610, "tokens": 35295068160, "source": "fineweb_hq", "phase": 3, "loss": 3.3101933896541595, "ce_loss": 3.2956433296203613, "logit_z_loss": 23.966604232788086, "router_aux_loss": 1.320176213979721, "router_z_loss": 1.3483231663703918, "routing_entropy_normalized": 0.6623723991162901, "learning_rate": 0.003, "grad_norm": 1.756307601928711, "tokens_per_second": 3744596.136724504, "eta_seconds": 6597.360478401184, "data_wait_seconds": 2.1696090698242188e-05} {"step": 16840, "total_steps": 28610, "tokens": 35316039680, "source": "dclm", "phase": 3, "loss": 3.4672525227069855, "ce_loss": 3.4524617195129395, "logit_z_loss": 24.200831174850464, "router_aux_loss": 1.3647994697093964, "router_z_loss": 1.142810806632042, "routing_entropy_normalized": 0.6485941406847492, "learning_rate": 0.003, "grad_norm": 2.1475601196289062, "tokens_per_second": 4363635.102864565, "eta_seconds": 5656.632247686386, "data_wait_seconds": 2.3096799850463867e-05} {"step": 16850, "total_steps": 28610, "tokens": 35337011200, "source": "fineweb_hq", "phase": 3, "loss": 3.318216562271118, "ce_loss": 3.302838683128357, "logit_z_loss": 23.643977165222168, "router_aux_loss": 1.409720554947853, "router_z_loss": 1.2806657403707504, "routing_entropy_normalized": 0.6523548123275151, "learning_rate": 0.003, "grad_norm": 1.700352668762207, "tokens_per_second": 4078105.096797641, "eta_seconds": 6047.540937423706, "data_wait_seconds": 2.008676528930664e-05} {"step": 16860, "total_steps": 28610, "tokens": 35357982720, "source": "fineweb_edu", "phase": 3, "loss": 3.356578379869461, "ce_loss": 3.34061336517334, "logit_z_loss": 23.52264904975891, "router_aux_loss": 1.4568158686161041, "router_z_loss": 1.3968145102262497, "routing_entropy_normalized": 0.6576472256481902, "learning_rate": 0.003, "grad_norm": 1.8268641233444214, "tokens_per_second": 4243649.186214577, "eta_seconds": 5806.685453653336, "data_wait_seconds": 2.16066837310791e-05} {"step": 16870, "total_steps": 28610, "tokens": 35378954240, "source": "dclm", "phase": 3, "loss": 3.415757179260254, "ce_loss": 3.40018630027771, "logit_z_loss": 23.733354091644287, "router_aux_loss": 1.43563574552536, "router_z_loss": 1.2145011574029922, "routing_entropy_normalized": 0.6549131762957119, "learning_rate": 0.003, "grad_norm": 1.328082799911499, "tokens_per_second": 3947404.143495509, "eta_seconds": 6237.153223991394, "data_wait_seconds": 2.4944543838500977e-05} {"step": 16880, "total_steps": 28610, "tokens": 35399925760, "source": "fineweb_edu", "phase": 3, "loss": 3.36798095703125, "ce_loss": 3.3526462018489838, "logit_z_loss": 23.99108862876892, "router_aux_loss": 1.3873187005519867, "router_z_loss": 1.4615890681743622, "routing_entropy_normalized": 0.6606568851887737, "learning_rate": 0.003, "grad_norm": 0.9941073060035706, "tokens_per_second": 4427550.3785616215, "eta_seconds": 5556.027793407439, "data_wait_seconds": 2.0116567611694336e-05} {"step": 16890, "total_steps": 28610, "tokens": 35420897280, "source": "fineweb_hq", "phase": 3, "loss": 3.322775959968567, "ce_loss": 3.3084871768951416, "logit_z_loss": 24.352492809295654, "router_aux_loss": 1.289205402135849, "router_z_loss": 1.3967204689979553, "routing_entropy_normalized": 0.6816963198005781, "learning_rate": 0.003, "grad_norm": 1.697343349456787, "tokens_per_second": 3997861.931056882, "eta_seconds": 6147.941540718079, "data_wait_seconds": 2.664327621459961e-05} {"step": 16900, "total_steps": 28610, "tokens": 35441868800, "source": "fineweb_edu", "phase": 3, "loss": 3.3272501826286316, "ce_loss": 3.3122797906398773, "logit_z_loss": 24.265687704086304, "router_aux_loss": 1.3602502346038818, "router_z_loss": 1.3678852170705795, "routing_entropy_normalized": 0.6596094568859074, "learning_rate": 0.003, "grad_norm": 1.9607703685760498, "tokens_per_second": 3416865.284638261, "eta_seconds": 7187.187048435211, "data_wait_seconds": 0.08954688906669617} {"step": 16910, "total_steps": 28610, "tokens": 35462840320, "source": "finemath", "phase": 3, "loss": 2.3119839429855347, "ce_loss": 2.2965115010738373, "logit_z_loss": 31.39111638069153, "router_aux_loss": 1.4577379822731018, "router_z_loss": 0.8950518295168877, "routing_entropy_normalized": 0.6574359717670305, "learning_rate": 0.003, "grad_norm": 1.1695477962493896, "tokens_per_second": 4112557.8835269157, "eta_seconds": 5966.281592845917, "data_wait_seconds": 2.0712614059448242e-05} {"step": 16920, "total_steps": 28610, "tokens": 35483811840, "source": "dclm", "phase": 3, "loss": 3.467354327440262, "ce_loss": 3.4522061347961426, "logit_z_loss": 23.994884729385376, "router_aux_loss": 1.3946499526500702, "router_z_loss": 1.201647087931633, "routing_entropy_normalized": 0.6640222049405027, "learning_rate": 0.003, "grad_norm": 1.255557656288147, "tokens_per_second": 3945497.7247951142, "eta_seconds": 6213.590423822403, "data_wait_seconds": 2.294778823852539e-05} {"step": 16930, "total_steps": 28610, "tokens": 35504783360, "source": "fineweb_hq", "phase": 3, "loss": 3.3175753355026245, "ce_loss": 3.302935391664505, "logit_z_loss": 24.505352020263672, "router_aux_loss": 1.3294844180345535, "router_z_loss": 1.3451209962368011, "routing_entropy_normalized": 0.6667140280957026, "learning_rate": 0.003, "grad_norm": 1.0381277799606323, "tokens_per_second": 4511343.257845723, "eta_seconds": 5429.588031768799, "data_wait_seconds": 1.96993350982666e-05} {"step": 16940, "total_steps": 28610, "tokens": 35525754880, "source": "dclm", "phase": 3, "loss": 3.512465685606003, "ce_loss": 3.4973480701446533, "logit_z_loss": 24.371249437332153, "router_aux_loss": 1.3922957181930542, "router_z_loss": 1.1946100890636444, "routing_entropy_normalized": 0.6328829362743768, "learning_rate": 0.003, "grad_norm": 1.782395362854004, "tokens_per_second": 4007888.0299274344, "eta_seconds": 6106.39909529686, "data_wait_seconds": 2.422928810119629e-05} {"step": 16950, "total_steps": 28610, "tokens": 35546726400, "source": "fineweb_hq", "phase": 3, "loss": 3.3414860367774963, "ce_loss": 3.32582026720047, "logit_z_loss": 24.902482509613037, "router_aux_loss": 1.439223513007164, "router_z_loss": 1.273516908288002, "routing_entropy_normalized": 0.641315438360583, "learning_rate": 0.003, "grad_norm": 1.5112662315368652, "tokens_per_second": 4336613.395254811, "eta_seconds": 5638.683943271637, "data_wait_seconds": 2.1159648895263672e-05} {"step": 16960, "total_steps": 28610, "tokens": 35567697920, "source": "fineweb_edu", "phase": 3, "loss": 3.3629975616931915, "ce_loss": 3.3478577733039856, "logit_z_loss": 24.353524923324585, "router_aux_loss": 1.3741914480924606, "router_z_loss": 1.3978666961193085, "routing_entropy_normalized": 0.6867336210093817, "learning_rate": 0.003, "grad_norm": 1.2087198495864868, "tokens_per_second": 3596886.159269547, "eta_seconds": 6792.492094039917, "data_wait_seconds": 2.384185791015625e-05} {"step": 16970, "total_steps": 28610, "tokens": 35588669440, "source": "dclm", "phase": 3, "loss": 3.390320986509323, "ce_loss": 3.37540265917778, "logit_z_loss": 24.947267770767212, "router_aux_loss": 1.370117336511612, "router_z_loss": 1.21718630194664, "routing_entropy_normalized": 0.6712932609518522, "learning_rate": 0.003, "grad_norm": 1.0999631881713867, "tokens_per_second": 3997728.9277614476, "eta_seconds": 6106.179213523865, "data_wait_seconds": 2.199411392211914e-05} {"step": 16980, "total_steps": 28610, "tokens": 35609640960, "source": "fineweb_edu", "phase": 3, "loss": 3.346023291349411, "ce_loss": 3.3312219381332397, "logit_z_loss": 24.459831476211548, "router_aux_loss": 1.3511886894702911, "router_z_loss": 1.289520412683487, "routing_entropy_normalized": 0.6997662242445564, "learning_rate": 0.003, "grad_norm": 1.3366637229919434, "tokens_per_second": 4068932.15446939, "eta_seconds": 5994.17164850235, "data_wait_seconds": 2.2649765014648438e-05} {"step": 16990, "total_steps": 28610, "tokens": 35630612480, "source": "fineweb_hq", "phase": 3, "loss": 3.3242124915122986, "ce_loss": 3.3098226487636566, "logit_z_loss": 23.697354793548584, "router_aux_loss": 1.3122970759868622, "router_z_loss": 1.2669232040643692, "routing_entropy_normalized": 0.6574057778363914, "learning_rate": 0.003, "grad_norm": 1.0524884462356567, "tokens_per_second": 3969560.446324425, "eta_seconds": 6138.943233013153, "data_wait_seconds": 2.282857894897461e-05} {"step": 17000, "total_steps": 28610, "tokens": 35651584000, "source": "fineweb_edu", "phase": 3, "loss": 3.340050607919693, "ce_loss": 3.3259392380714417, "logit_z_loss": 24.389240026474, "router_aux_loss": 1.288174331188202, "router_z_loss": 1.2296918034553528, "routing_entropy_normalized": 0.6638301955985525, "learning_rate": 0.003, "grad_norm": 1.0830892324447632, "tokens_per_second": 3637871.43850652, "eta_seconds": 6692.906863689423, "data_wait_seconds": 0.01540866494178772} {"step": 17010, "total_steps": 28610, "tokens": 35672555520, "source": "finemath", "phase": 3, "loss": 2.386664479970932, "ce_loss": 2.372222274541855, "logit_z_loss": 31.116959810256958, "router_aux_loss": 1.3636980205774307, "router_z_loss": 0.8051807433366776, "routing_entropy_normalized": 0.6731819856878207, "learning_rate": 0.003, "grad_norm": 1.824406623840332, "tokens_per_second": 4179048.2913037613, "eta_seconds": 5821.173028945923, "data_wait_seconds": 2.086162567138672e-05} {"step": 17020, "total_steps": 28610, "tokens": 35693527040, "source": "dclm", "phase": 3, "loss": 3.4324595034122467, "ce_loss": 3.418052554130554, "logit_z_loss": 25.365362405776978, "router_aux_loss": 1.3334178030490875, "router_z_loss": 1.0727740600705147, "routing_entropy_normalized": 0.6545373098873298, "learning_rate": 0.003, "grad_norm": 1.0493606328964233, "tokens_per_second": 3643188.282132707, "eta_seconds": 6671.626552820206, "data_wait_seconds": 2.41696834564209e-05} {"step": 17030, "total_steps": 28610, "tokens": 35714498560, "source": "fineweb_hq", "phase": 3, "loss": 3.3341600596904755, "ce_loss": 3.31982421875, "logit_z_loss": 22.60609483718872, "router_aux_loss": 1.3280370980501175, "router_z_loss": 1.0555092096328735, "routing_entropy_normalized": 0.6828486447170032, "learning_rate": 0.003, "grad_norm": 2.5018553733825684, "tokens_per_second": 3762444.689828298, "eta_seconds": 6454.585292816162, "data_wait_seconds": 2.1129846572875977e-05} {"step": 17040, "total_steps": 28610, "tokens": 35735470080, "source": "dclm", "phase": 3, "loss": 3.47923082113266, "ce_loss": 3.463645488023758, "logit_z_loss": 23.85672354698181, "router_aux_loss": 1.459121286869049, "router_z_loss": 0.9940895289182663, "routing_entropy_normalized": 0.6533380201329927, "learning_rate": 0.003, "grad_norm": 1.3437714576721191, "tokens_per_second": 4238331.351615401, "eta_seconds": 5724.906013011932, "data_wait_seconds": 2.276897430419922e-05} {"step": 17050, "total_steps": 28610, "tokens": 35756441600, "source": "fineweb_hq", "phase": 3, "loss": 3.3243201076984406, "ce_loss": 3.3091709911823273, "logit_z_loss": 24.194568634033203, "router_aux_loss": 1.3986011892557144, "router_z_loss": 1.1630952954292297, "routing_entropy_normalized": 0.6417353795538363, "learning_rate": 0.003, "grad_norm": 1.28342604637146, "tokens_per_second": 4013100.848261818, "eta_seconds": 6040.983777046204, "data_wait_seconds": 2.136826515197754e-05} {"step": 17060, "total_steps": 28610, "tokens": 35777413120, "source": "fineweb_edu", "phase": 3, "loss": 3.330639958381653, "ce_loss": 3.3158369064331055, "logit_z_loss": 24.4469895362854, "router_aux_loss": 1.3643403202295303, "router_z_loss": 1.1596489995718002, "routing_entropy_normalized": 0.6545683943365254, "learning_rate": 0.003, "grad_norm": 1.096129059791565, "tokens_per_second": 3869364.9272735566, "eta_seconds": 6259.969285726547, "data_wait_seconds": 2.2143125534057617e-05} {"step": 17070, "total_steps": 28610, "tokens": 35798384640, "source": "dclm", "phase": 3, "loss": 3.4325567185878754, "ce_loss": 3.4185815155506134, "logit_z_loss": 24.832568168640137, "router_aux_loss": 1.2910181879997253, "router_z_loss": 1.0650329440832138, "routing_entropy_normalized": 0.6749163512815962, "learning_rate": 0.003, "grad_norm": 0.8266509175300598, "tokens_per_second": 4085397.951084054, "eta_seconds": 5923.813143730164, "data_wait_seconds": 0.010806262493133545} {"step": 17080, "total_steps": 28610, "tokens": 35819356160, "source": "fineweb_edu", "phase": 3, "loss": 3.2866406440734863, "ce_loss": 3.2721187472343445, "logit_z_loss": 23.851438999176025, "router_aux_loss": 1.3463477492332458, "router_z_loss": 1.0584667176008224, "routing_entropy_normalized": 0.6693833082777608, "learning_rate": 0.003, "grad_norm": 1.2237098217010498, "tokens_per_second": 4011641.583321088, "eta_seconds": 6027.498234272003, "data_wait_seconds": 2.187490463256836e-05} {"step": 17090, "total_steps": 28610, "tokens": 35840327680, "source": "fineweb_hq", "phase": 3, "loss": 3.3335966765880585, "ce_loss": 3.3194876611232758, "logit_z_loss": 23.447165489196777, "router_aux_loss": 1.300554096698761, "router_z_loss": 1.1033636927604675, "routing_entropy_normalized": 0.6579987735715289, "learning_rate": 0.003, "grad_norm": 2.1533219814300537, "tokens_per_second": 3980336.8291422734, "eta_seconds": 6069.634826660156, "data_wait_seconds": 2.0772218704223633e-05} {"step": 17100, "total_steps": 28610, "tokens": 35861299200, "source": "fineweb_edu", "phase": 3, "loss": 3.281648099422455, "ce_loss": 3.2667274475097656, "logit_z_loss": 23.90135407447815, "router_aux_loss": 1.3736849427223206, "router_z_loss": 1.183762565255165, "routing_entropy_normalized": 0.6267351016461752, "learning_rate": 0.003, "grad_norm": 1.7989815473556519, "tokens_per_second": 4018590.1331351777, "eta_seconds": 6006.638826131821, "data_wait_seconds": 2.0384788513183594e-05} {"step": 17110, "total_steps": 28610, "tokens": 35882270720, "source": "finemath", "phase": 3, "loss": 2.3999147415161133, "ce_loss": 2.384676307439804, "logit_z_loss": 32.70622730255127, "router_aux_loss": 1.4427781254053116, "router_z_loss": 0.8106840997934341, "routing_entropy_normalized": 0.681692940603401, "learning_rate": 0.003, "grad_norm": 1.2674680948257446, "tokens_per_second": 3372804.1142437505, "eta_seconds": 7150.503611564635, "data_wait_seconds": 1.9282102584838867e-05} {"step": 17120, "total_steps": 28610, "tokens": 35903242240, "source": "dclm", "phase": 3, "loss": 3.449355334043503, "ce_loss": 3.4344336092472076, "logit_z_loss": 24.93084955215454, "router_aux_loss": 1.394062951207161, "router_z_loss": 0.9810987561941147, "routing_entropy_normalized": 0.6806061677412018, "learning_rate": 0.003, "grad_norm": 1.0721513032913208, "tokens_per_second": 3069101.799734921, "eta_seconds": 7851.247059345245, "data_wait_seconds": 1.9043684005737305e-05} {"step": 17130, "total_steps": 28610, "tokens": 35924213760, "source": "fineweb_hq", "phase": 3, "loss": 3.325061559677124, "ce_loss": 3.3095525205135345, "logit_z_loss": 24.85837459564209, "router_aux_loss": 1.4433816075325012, "router_z_loss": 1.0751376748085022, "routing_entropy_normalized": 0.6434256342501703, "learning_rate": 0.003, "grad_norm": 2.3996808528900146, "tokens_per_second": 4231657.322035819, "eta_seconds": 5689.332365036011, "data_wait_seconds": 1.868605613708496e-05} {"step": 17140, "total_steps": 28610, "tokens": 35945185280, "source": "dclm", "phase": 3, "loss": 3.4754787981510162, "ce_loss": 3.4607720971107483, "logit_z_loss": 24.934978485107422, "router_aux_loss": 1.3715435564517975, "router_z_loss": 0.991319015622139, "routing_entropy_normalized": 0.6610182344870796, "learning_rate": 0.003, "grad_norm": 1.9591087102890015, "tokens_per_second": 3496539.6398729463, "eta_seconds": 6879.468250751495, "data_wait_seconds": 2.4586915969848633e-05} {"step": 17150, "total_steps": 28610, "tokens": 35966156800, "source": "fineweb_hq", "phase": 3, "loss": 3.2982290983200073, "ce_loss": 3.2842279374599457, "logit_z_loss": 24.55796194076538, "router_aux_loss": 1.2794218957424164, "router_z_loss": 1.2069497257471085, "routing_entropy_normalized": 0.6873291740842362, "learning_rate": 0.003, "grad_norm": 1.2656272649765015, "tokens_per_second": 3829407.958374654, "eta_seconds": 6275.999366283417, "data_wait_seconds": 2.1219253540039062e-05} {"step": 17160, "total_steps": 28610, "tokens": 35987128320, "source": "fineweb_edu", "phase": 3, "loss": 3.3336551189422607, "ce_loss": 3.3193975389003754, "logit_z_loss": 24.247788190841675, "router_aux_loss": 1.3180904537439346, "router_z_loss": 1.0766796320676804, "routing_entropy_normalized": 0.6619170611175424, "learning_rate": 0.003, "grad_norm": 1.6365023851394653, "tokens_per_second": 4219312.833078871, "eta_seconds": 5691.0666143894205, "data_wait_seconds": 2.2590160369873047e-05} {"step": 17170, "total_steps": 28610, "tokens": 36008099840, "source": "dclm", "phase": 3, "loss": 3.449458569288254, "ce_loss": 3.4353606700897217, "logit_z_loss": 24.79862070083618, "router_aux_loss": 1.322369247674942, "router_z_loss": 0.8742367997765541, "routing_entropy_normalized": 0.650773198103719, "learning_rate": 0.003, "grad_norm": 0.929755687713623, "tokens_per_second": 4694103.000588224, "eta_seconds": 5110.969843864441, "data_wait_seconds": 2.0563602447509766e-05} {"step": 17180, "total_steps": 28610, "tokens": 36029071360, "source": "fineweb_edu", "phase": 3, "loss": 3.318983495235443, "ce_loss": 3.3047711849212646, "logit_z_loss": 23.601280689239502, "router_aux_loss": 1.3163415789604187, "router_z_loss": 1.0489245727658272, "routing_entropy_normalized": 0.6650995139428859, "learning_rate": 0.003, "grad_norm": 1.2030612230300903, "tokens_per_second": 4468742.917812375, "eta_seconds": 5364.024693489075, "data_wait_seconds": 2.148747444152832e-05} {"step": 17190, "total_steps": 28610, "tokens": 36050042880, "source": "fineweb_hq", "phase": 3, "loss": 3.344080328941345, "ce_loss": 3.3299500942230225, "logit_z_loss": 23.175495386123657, "router_aux_loss": 1.3257685601711273, "router_z_loss": 0.8725835606455803, "routing_entropy_normalized": 0.6721376814405915, "learning_rate": 0.003, "grad_norm": 0.9763891696929932, "tokens_per_second": 3584964.39698814, "eta_seconds": 6680.533803939819, "data_wait_seconds": 2.3663043975830078e-05} {"step": 17200, "total_steps": 28610, "tokens": 36071014400, "source": "fineweb_edu", "phase": 3, "loss": 3.3608618676662445, "ce_loss": 3.346301943063736, "logit_z_loss": 24.673241138458252, "router_aux_loss": 1.3631321787834167, "router_z_loss": 0.9286193624138832, "routing_entropy_normalized": 0.6682871241535531, "learning_rate": 0.003, "grad_norm": 1.5608102083206177, "tokens_per_second": 3785385.9577643876, "eta_seconds": 6321.285223484039, "data_wait_seconds": 1.996755599975586e-05} {"step": 17210, "total_steps": 28610, "tokens": 36091985920, "source": "finemath", "phase": 3, "loss": 2.2776796966791153, "ce_loss": 2.2627452462911606, "logit_z_loss": 26.94133710861206, "router_aux_loss": 1.3922090530395508, "router_z_loss": 1.0123754516243935, "routing_entropy_normalized": 0.7059363145618117, "learning_rate": 0.003, "grad_norm": 2.282806396484375, "tokens_per_second": 4280252.673454724, "eta_seconds": 5585.542402267456, "data_wait_seconds": 1.946091651916504e-05} {"step": 17220, "total_steps": 28610, "tokens": 36112957440, "source": "dclm", "phase": 3, "loss": 3.499095320701599, "ce_loss": 3.483304738998413, "logit_z_loss": 25.025824308395386, "router_aux_loss": 1.4935847967863083, "router_z_loss": 0.8547283634543419, "routing_entropy_normalized": 0.6431680544891922, "learning_rate": 0.003, "grad_norm": 1.6193574666976929, "tokens_per_second": 3606017.7150543947, "eta_seconds": 6624.083176374435, "data_wait_seconds": 2.378225326538086e-05} {"step": 17230, "total_steps": 28610, "tokens": 36133928960, "source": "fineweb_hq", "phase": 3, "loss": 3.369762510061264, "ce_loss": 3.354473352432251, "logit_z_loss": 24.76356339454651, "router_aux_loss": 1.4255895763635635, "router_z_loss": 1.0332350581884384, "routing_entropy_normalized": 0.6522739090328826, "learning_rate": 0.003, "grad_norm": 1.5661981105804443, "tokens_per_second": 4402889.46937584, "eta_seconds": 5420.438084125518, "data_wait_seconds": 2.2023916244506836e-05} {"step": 17240, "total_steps": 28610, "tokens": 36154900480, "source": "dclm", "phase": 3, "loss": 3.4171202480793, "ce_loss": 3.401967942714691, "logit_z_loss": 23.96466302871704, "router_aux_loss": 1.4255397319793701, "router_z_loss": 0.8969019055366516, "routing_entropy_normalized": 0.650363697811745, "learning_rate": 0.003, "grad_norm": 1.558766484260559, "tokens_per_second": 3381752.455957523, "eta_seconds": 7050.965010166168, "data_wait_seconds": 2.1219253540039062e-05} {"step": 17250, "total_steps": 28610, "tokens": 36175872000, "source": "fineweb_hq", "phase": 3, "loss": 3.3191212713718414, "ce_loss": 3.304029703140259, "logit_z_loss": 24.16761589050293, "router_aux_loss": 1.4032163619995117, "router_z_loss": 1.059404768049717, "routing_entropy_normalized": 0.6718507200170917, "learning_rate": 0.003, "grad_norm": 1.2580512762069702, "tokens_per_second": 3713658.2829654403, "eta_seconds": 6415.142402648926, "data_wait_seconds": 1.9490718841552734e-05} {"step": 17260, "total_steps": 28610, "tokens": 36196843520, "source": "fineweb_edu", "phase": 3, "loss": 3.3421328961849213, "ce_loss": 3.3264642655849457, "logit_z_loss": 23.64670968055725, "router_aux_loss": 1.4481509327888489, "router_z_loss": 1.1871790289878845, "routing_entropy_normalized": 0.6691448550625437, "learning_rate": 0.003, "grad_norm": 1.2555420398712158, "tokens_per_second": 4120641.234612758, "eta_seconds": 5776.4493060112, "data_wait_seconds": 2.1219253540039062e-05} {"step": 17270, "total_steps": 28610, "tokens": 36217815040, "source": "dclm", "phase": 3, "loss": 3.4709755778312683, "ce_loss": 3.455557554960251, "logit_z_loss": 24.90299654006958, "router_aux_loss": 1.428288772702217, "router_z_loss": 1.1351827532052994, "routing_entropy_normalized": 0.6726258191047969, "learning_rate": 0.003, "grad_norm": 1.4371198415756226, "tokens_per_second": 4065049.2610307545, "eta_seconds": 5850.2867131233215, "data_wait_seconds": 2.148747444152832e-05} {"step": 17280, "total_steps": 28610, "tokens": 36238786560, "source": "fineweb_edu", "phase": 3, "loss": 3.347979635000229, "ce_loss": 3.3323418498039246, "logit_z_loss": 23.820318937301636, "router_aux_loss": 1.440832868218422, "router_z_loss": 1.2294099777936935, "routing_entropy_normalized": 0.6673001468453454, "learning_rate": 0.003, "grad_norm": 0.7717240452766418, "tokens_per_second": 4436938.719294916, "eta_seconds": 5355.208548784256, "data_wait_seconds": 2.1696090698242188e-05} {"step": 17290, "total_steps": 28610, "tokens": 36259758080, "source": "fineweb_hq", "phase": 3, "loss": 3.333776444196701, "ce_loss": 3.3193289041519165, "logit_z_loss": 25.088132858276367, "router_aux_loss": 1.314994215965271, "router_z_loss": 1.297603577375412, "routing_entropy_normalized": 0.6846329635990667, "learning_rate": 0.003, "grad_norm": 1.1412121057510376, "tokens_per_second": 3519467.8498483496, "eta_seconds": 6745.269925117493, "data_wait_seconds": 2.3990869522094727e-05} {"step": 17300, "total_steps": 28610, "tokens": 36280729600, "source": "fineweb_edu", "phase": 3, "loss": 3.339732676744461, "ce_loss": 3.3258520364761353, "logit_z_loss": 24.328697681427002, "router_aux_loss": 1.264312520623207, "router_z_loss": 1.2374864369630814, "routing_entropy_normalized": 0.7069928901796614, "learning_rate": 0.003, "grad_norm": 2.5298874378204346, "tokens_per_second": 4000945.8353387127, "eta_seconds": 5928.295482158661, "data_wait_seconds": 2.0891427993774414e-05} {"step": 17310, "total_steps": 28610, "tokens": 36301701120, "source": "finemath", "phase": 3, "loss": 2.3622645437717438, "ce_loss": 2.3477770388126373, "logit_z_loss": 32.47808384895325, "router_aux_loss": 1.361037626862526, "router_z_loss": 0.8771211802959442, "routing_entropy_normalized": 0.6616614220119644, "learning_rate": 0.003, "grad_norm": 1.0180916786193848, "tokens_per_second": 3632351.53052609, "eta_seconds": 6524.09806728363, "data_wait_seconds": 1.823902130126953e-05} {"step": 17320, "total_steps": 28610, "tokens": 36322672640, "source": "dclm", "phase": 3, "loss": 3.4728824496269226, "ce_loss": 3.4584444165229797, "logit_z_loss": 25.253376245498657, "router_aux_loss": 1.3359915018081665, "router_z_loss": 1.0781264305114746, "routing_entropy_normalized": 0.6760324437928118, "learning_rate": 0.003, "grad_norm": 0.9687179327011108, "tokens_per_second": 3987086.3564979574, "eta_seconds": 5938.383060455322, "data_wait_seconds": 1.9431114196777344e-05} {"step": 17330, "total_steps": 28610, "tokens": 36343644160, "source": "fineweb_hq", "phase": 3, "loss": 3.3751942813396454, "ce_loss": 3.360256940126419, "logit_z_loss": 24.674970149993896, "router_aux_loss": 1.3761787116527557, "router_z_loss": 1.1755721718072891, "routing_entropy_normalized": 0.6627057943697306, "learning_rate": 0.003, "grad_norm": 1.3569365739822388, "tokens_per_second": 3593703.8868372752, "eta_seconds": 6582.588689804077, "data_wait_seconds": 1.8924474716186523e-05} {"step": 17340, "total_steps": 28610, "tokens": 36364615680, "source": "dclm", "phase": 3, "loss": 3.5086226761341095, "ce_loss": 3.4937664270401, "logit_z_loss": 24.12835168838501, "router_aux_loss": 1.3827333897352219, "router_z_loss": 1.0288755223155022, "routing_entropy_normalized": 0.6614015445040368, "learning_rate": 0.003, "grad_norm": 1.8296723365783691, "tokens_per_second": 3982681.6857250943, "eta_seconds": 5934.419294595718, "data_wait_seconds": 2.244114875793457e-05} {"step": 17350, "total_steps": 28610, "tokens": 36385587200, "source": "fineweb_hq", "phase": 3, "loss": 3.3042410016059875, "ce_loss": 3.288923054933548, "logit_z_loss": 24.44698429107666, "router_aux_loss": 1.4130982458591461, "router_z_loss": 1.1869654804468155, "routing_entropy_normalized": 0.6521482188163291, "learning_rate": 0.003, "grad_norm": 0.4957062304019928, "tokens_per_second": 3769003.259239809, "eta_seconds": 6265.29877948761, "data_wait_seconds": 2.1964311599731445e-05} {"step": 17360, "total_steps": 28610, "tokens": 36406558720, "source": "fineweb_edu", "phase": 3, "loss": 3.339684545993805, "ce_loss": 3.324578493833542, "logit_z_loss": 24.384618282318115, "router_aux_loss": 1.3985347747802734, "router_z_loss": 1.1207591742277145, "routing_entropy_normalized": 0.6686766682664936, "learning_rate": 0.003, "grad_norm": 2.2874562740325928, "tokens_per_second": 4022984.4732589615, "eta_seconds": 5864.54164981842, "data_wait_seconds": 2.154707908630371e-05} {"step": 17370, "total_steps": 28610, "tokens": 36427530240, "source": "dclm", "phase": 3, "loss": 3.521569937467575, "ce_loss": 3.5067606568336487, "logit_z_loss": 24.662313222885132, "router_aux_loss": 1.3751725554466248, "router_z_loss": 1.0575807392597198, "routing_entropy_normalized": 0.6478935180791597, "learning_rate": 0.003, "grad_norm": 1.1517075300216675, "tokens_per_second": 3975521.7036096416, "eta_seconds": 5929.281799316406, "data_wait_seconds": 2.0802021026611328e-05} {"step": 17380, "total_steps": 28610, "tokens": 36448501760, "source": "fineweb_edu", "phase": 3, "loss": 3.319571554660797, "ce_loss": 3.3050007820129395, "logit_z_loss": 24.10121440887451, "router_aux_loss": 1.3500744849443436, "router_z_loss": 1.0699898451566696, "routing_entropy_normalized": 0.6450569625217556, "learning_rate": 0.003, "grad_norm": 1.3871561288833618, "tokens_per_second": 4429351.618067434, "eta_seconds": 5317.034859895706, "data_wait_seconds": 2.1576881408691406e-05} {"step": 17390, "total_steps": 28610, "tokens": 36469473280, "source": "fineweb_hq", "phase": 3, "loss": 3.3673340380191803, "ce_loss": 3.352759450674057, "logit_z_loss": 24.61271357536316, "router_aux_loss": 1.3448399603366852, "router_z_loss": 1.1261759400367737, "routing_entropy_normalized": 0.65620420020637, "learning_rate": 0.003, "grad_norm": 0.9255719184875488, "tokens_per_second": 3474228.6224465817, "eta_seconds": 6772.739504814148, "data_wait_seconds": 2.1964311599731445e-05} {"step": 17400, "total_steps": 28610, "tokens": 36490444800, "source": "fineweb_edu", "phase": 3, "loss": 3.324878066778183, "ce_loss": 3.3100512623786926, "logit_z_loss": 23.4683198928833, "router_aux_loss": 1.3791682869195938, "router_z_loss": 1.035140261054039, "routing_entropy_normalized": 0.6706218752437779, "learning_rate": 0.003, "grad_norm": 0.8024491667747498, "tokens_per_second": 3982420.9495800324, "eta_seconds": 5903.2116940021515, "data_wait_seconds": 1.9222497940063477e-05} {"step": 17410, "total_steps": 28610, "tokens": 36511416320, "source": "finemath", "phase": 3, "loss": 2.3221541941165924, "ce_loss": 2.3062776923179626, "logit_z_loss": 29.965606451034546, "router_aux_loss": 1.5170129090547562, "router_z_loss": 0.7063558846712112, "routing_entropy_normalized": 0.6615156455488022, "learning_rate": 0.003, "grad_norm": 1.3690729141235352, "tokens_per_second": 4265155.152691764, "eta_seconds": 5506.974906921387, "data_wait_seconds": 1.9371509552001953e-05} {"step": 17420, "total_steps": 28610, "tokens": 36532387840, "source": "dclm", "phase": 3, "loss": 3.4500297904014587, "ce_loss": 3.434584081172943, "logit_z_loss": 24.90823221206665, "router_aux_loss": 1.4478371739387512, "router_z_loss": 0.9673203080892563, "routing_entropy_normalized": 0.6635110531476612, "learning_rate": 0.003, "grad_norm": 1.181290864944458, "tokens_per_second": 3805733.392767219, "eta_seconds": 6166.2571856975555, "data_wait_seconds": 1.96993350982666e-05} {"step": 17430, "total_steps": 28610, "tokens": 36553359360, "source": "fineweb_hq", "phase": 3, "loss": 3.3528107702732086, "ce_loss": 3.3370567858219147, "logit_z_loss": 24.320453882217407, "router_aux_loss": 1.457360863685608, "router_z_loss": 1.1803675889968872, "routing_entropy_normalized": 0.635660394391621, "learning_rate": 0.003, "grad_norm": 1.1802599430084229, "tokens_per_second": 4164765.698284287, "eta_seconds": 5629.646673679352, "data_wait_seconds": 2.0444393157958984e-05} {"step": 17440, "total_steps": 28610, "tokens": 36574330880, "source": "dclm", "phase": 3, "loss": 3.4807375371456146, "ce_loss": 3.466040462255478, "logit_z_loss": 25.56485915184021, "router_aux_loss": 1.3585560470819473, "router_z_loss": 1.1115543097257614, "routing_entropy_normalized": 0.6764892681783882, "learning_rate": 0.003, "grad_norm": 0.6666913032531738, "tokens_per_second": 3617787.8868513023, "eta_seconds": 6475.00311589241, "data_wait_seconds": 2.154707908630371e-05} {"step": 17450, "total_steps": 28610, "tokens": 36595302400, "source": "fineweb_hq", "phase": 3, "loss": 3.3506336212158203, "ce_loss": 3.33568874001503, "logit_z_loss": 24.60809564590454, "router_aux_loss": 1.3710343092679977, "router_z_loss": 1.2345846444368362, "routing_entropy_normalized": 0.6427205506094046, "learning_rate": 0.003, "grad_norm": 0.8674792647361755, "tokens_per_second": 4177786.5043572295, "eta_seconds": 5602.061353683472, "data_wait_seconds": 1.9043684005737305e-05} {"step": 17460, "total_steps": 28610, "tokens": 36616273920, "source": "fineweb_edu", "phase": 3, "loss": 3.307685524225235, "ce_loss": 3.29235303401947, "logit_z_loss": 24.605698585510254, "router_aux_loss": 1.415977954864502, "router_z_loss": 1.1727191656827927, "routing_entropy_normalized": 0.6584321877756323, "learning_rate": 0.003, "grad_norm": 1.1539127826690674, "tokens_per_second": 4155003.9195821574, "eta_seconds": 5627.731105089188, "data_wait_seconds": 2.047419548034668e-05} {"step": 17470, "total_steps": 28610, "tokens": 36637245440, "source": "dclm", "phase": 3, "loss": 3.4498690366744995, "ce_loss": 3.435021996498108, "logit_z_loss": 24.5143563747406, "router_aux_loss": 1.3818935453891754, "router_z_loss": 1.0281807407736778, "routing_entropy_normalized": 0.6550072065663488, "learning_rate": 0.003, "grad_norm": 1.1321476697921753, "tokens_per_second": 3346873.271963985, "eta_seconds": 6980.328020095825, "data_wait_seconds": 0.17677876353263855} {"step": 17480, "total_steps": 28610, "tokens": 36658216960, "source": "fineweb_edu", "phase": 3, "loss": 3.3457203209400177, "ce_loss": 3.3312096893787384, "logit_z_loss": 23.599600315093994, "router_aux_loss": 1.3368948996067047, "router_z_loss": 1.1417192071676254, "routing_entropy_normalized": 0.6789734018661262, "learning_rate": 0.003, "grad_norm": 0.43818339705467224, "tokens_per_second": 4217305.0469752755, "eta_seconds": 5534.648667812348, "data_wait_seconds": 2.3633241653442383e-05} {"step": 17490, "total_steps": 28610, "tokens": 36679188480, "source": "fineweb_hq", "phase": 3, "loss": 3.3661540150642395, "ce_loss": 3.3514925837516785, "logit_z_loss": 23.927305698394775, "router_aux_loss": 1.3550141602754593, "router_z_loss": 1.111300379037857, "routing_entropy_normalized": 0.656214760058002, "learning_rate": 0.003, "grad_norm": 1.0015318393707275, "tokens_per_second": 3439832.764814559, "eta_seconds": 6779.495351791382, "data_wait_seconds": 1.952052116394043e-05} {"step": 17500, "total_steps": 28610, "tokens": 36700160000, "source": "fineweb_edu", "phase": 3, "loss": 3.362288624048233, "ce_loss": 3.3475575745105743, "logit_z_loss": 23.919098615646362, "router_aux_loss": 1.3587298691272736, "router_z_loss": 1.1437819302082062, "routing_entropy_normalized": 0.6502707360706254, "learning_rate": 0.003, "grad_norm": 2.2707667350769043, "tokens_per_second": 4037289.6497068633, "eta_seconds": 5771.039668083191, "data_wait_seconds": 2.4050474166870117e-05} {"step": 17510, "total_steps": 28610, "tokens": 36721131520, "source": "finemath", "phase": 3, "loss": 2.3527783155441284, "ce_loss": 2.3377459943294525, "logit_z_loss": 33.98491621017456, "router_aux_loss": 1.4244421422481537, "router_z_loss": 0.7878389209508896, "routing_entropy_normalized": 0.6757885340251734, "learning_rate": 0.003, "grad_norm": 1.2248175144195557, "tokens_per_second": 3950379.6169609437, "eta_seconds": 5892.696261405945, "data_wait_seconds": 2.0295381546020508e-05} {"step": 17520, "total_steps": 28610, "tokens": 36742103040, "source": "dclm", "phase": 3, "loss": 3.504466474056244, "ce_loss": 3.4897056221961975, "logit_z_loss": 25.905542612075806, "router_aux_loss": 1.3699235767126083, "router_z_loss": 1.061591312289238, "routing_entropy_normalized": 0.6900442449165836, "learning_rate": 0.003, "grad_norm": 1.3303812742233276, "tokens_per_second": 3859145.4119949536, "eta_seconds": 6026.571480751038, "data_wait_seconds": 2.0176172256469727e-05} {"step": 17530, "total_steps": 28610, "tokens": 36763074560, "source": "fineweb_hq", "phase": 3, "loss": 3.377191185951233, "ce_loss": 3.3622013330459595, "logit_z_loss": 24.965617418289185, "router_aux_loss": 1.3809478282928467, "router_z_loss": 1.1803929954767227, "routing_entropy_normalized": 0.672700746739181, "learning_rate": 0.003, "grad_norm": 1.909172773361206, "tokens_per_second": 4054702.793432041, "eta_seconds": 5730.739179611206, "data_wait_seconds": 1.913309097290039e-05} {"step": 17540, "total_steps": 28610, "tokens": 36784046080, "source": "dclm", "phase": 3, "loss": 3.4670654237270355, "ce_loss": 3.451326906681061, "logit_z_loss": 24.850992918014526, "router_aux_loss": 1.4738776683807373, "router_z_loss": 0.9997415766119957, "routing_entropy_normalized": 0.63547677141856, "learning_rate": 0.003, "grad_norm": 1.3807048797607422, "tokens_per_second": 4057158.9450889453, "eta_seconds": 5722.100847959518, "data_wait_seconds": 2.110004425048828e-05} {"step": 17550, "total_steps": 28610, "tokens": 36805017600, "source": "fineweb_hq", "phase": 3, "loss": 3.308490127325058, "ce_loss": 3.29311802983284, "logit_z_loss": 24.65368890762329, "router_aux_loss": 1.4126089215278625, "router_z_loss": 1.2459662705659866, "routing_entropy_normalized": 0.6526122219325842, "learning_rate": 0.003, "grad_norm": 1.0675959587097168, "tokens_per_second": 3551142.3779095765, "eta_seconds": 6531.560453414917, "data_wait_seconds": 1.8477439880371094e-05} {"step": 17560, "total_steps": 28610, "tokens": 36825989120, "source": "fineweb_edu", "phase": 3, "loss": 3.349032998085022, "ce_loss": 3.3339169025421143, "logit_z_loss": 24.845054149627686, "router_aux_loss": 1.3911509811878204, "router_z_loss": 1.2046157270669937, "routing_entropy_normalized": 0.652355881368759, "learning_rate": 0.003, "grad_norm": 1.4033827781677246, "tokens_per_second": 3462108.427794312, "eta_seconds": 6693.473091125488, "data_wait_seconds": 2.2709369659423828e-05} {"step": 17570, "total_steps": 28610, "tokens": 36846960640, "source": "dclm", "phase": 3, "loss": 3.4710323214530945, "ce_loss": 3.4561771750450134, "logit_z_loss": 25.38108205795288, "router_aux_loss": 1.376646876335144, "router_z_loss": 1.0887624472379684, "routing_entropy_normalized": 0.6571594560389356, "learning_rate": 0.003, "grad_norm": 1.56998872756958, "tokens_per_second": 4079177.983935369, "eta_seconds": 5675.790115356445, "data_wait_seconds": 1.9758939743041992e-05} {"step": 17580, "total_steps": 28610, "tokens": 36867932160, "source": "fineweb_edu", "phase": 3, "loss": 3.376666307449341, "ce_loss": 3.3616455793380737, "logit_z_loss": 24.91264581680298, "router_aux_loss": 1.3882920295000076, "router_z_loss": 1.1378183364868164, "routing_entropy_normalized": 0.672108753730462, "learning_rate": 0.003, "grad_norm": 1.1135504245758057, "tokens_per_second": 4105204.9968198533, "eta_seconds": 5634.6970682144165, "data_wait_seconds": 2.2143125534057617e-05} {"step": 17590, "total_steps": 28610, "tokens": 36888903680, "source": "fineweb_hq", "phase": 3, "loss": 3.3308394253253937, "ce_loss": 3.316318154335022, "logit_z_loss": 23.874987363815308, "router_aux_loss": 1.3379058092832565, "router_z_loss": 1.1421684175729752, "routing_entropy_normalized": 0.6712664580866701, "learning_rate": 0.003, "grad_norm": 1.5254032611846924, "tokens_per_second": 3902932.929103325, "eta_seconds": 5921.345680236816, "data_wait_seconds": 2.250075340270996e-05} {"step": 17600, "total_steps": 28610, "tokens": 36909875200, "source": "fineweb_edu", "phase": 3, "loss": 3.370383560657501, "ce_loss": 3.3557927012443542, "logit_z_loss": 24.319108963012695, "router_aux_loss": 1.3494450747966766, "router_z_loss": 1.096411257982254, "routing_entropy_normalized": 0.6687676545280006, "learning_rate": 0.003, "grad_norm": 1.396226406097412, "tokens_per_second": 3528431.5618507736, "eta_seconds": 6543.883058309555, "data_wait_seconds": 2.1517276763916016e-05} {"step": 17610, "total_steps": 28610, "tokens": 36930846720, "source": "finemath", "phase": 3, "loss": 2.4386443197727203, "ce_loss": 2.4244308471679688, "logit_z_loss": 33.18499279022217, "router_aux_loss": 1.3430760651826859, "router_z_loss": 0.7827289551496506, "routing_entropy_normalized": 0.680510949907983, "learning_rate": 0.003, "grad_norm": 1.231141448020935, "tokens_per_second": 4310121.5402259305, "eta_seconds": 5352.209162712096, "data_wait_seconds": 2.09808349609375e-05} {"step": 17620, "total_steps": 28610, "tokens": 36951818240, "source": "dclm", "phase": 3, "loss": 3.3991501927375793, "ce_loss": 3.3846137523651123, "logit_z_loss": 25.99495792388916, "router_aux_loss": 1.3654889911413193, "router_z_loss": 0.8815916553139687, "routing_entropy_normalized": 0.6387472707704888, "learning_rate": 0.003, "grad_norm": 0.8818337917327881, "tokens_per_second": 3855964.7659939416, "eta_seconds": 5977.155362844467, "data_wait_seconds": 2.2292137145996094e-05} {"step": 17630, "total_steps": 28610, "tokens": 36972789760, "source": "fineweb_hq", "phase": 3, "loss": 3.369189888238907, "ce_loss": 3.354726403951645, "logit_z_loss": 24.077637434005737, "router_aux_loss": 1.3481442332267761, "router_z_loss": 0.9820490255951881, "routing_entropy_normalized": 0.6500614694788098, "learning_rate": 0.003, "grad_norm": 1.1332530975341797, "tokens_per_second": 4185295.7888014354, "eta_seconds": 5501.816388130188, "data_wait_seconds": 2.2470951080322266e-05} {"step": 17640, "total_steps": 28610, "tokens": 36993761280, "source": "dclm", "phase": 3, "loss": 3.535211503505707, "ce_loss": 3.5199498534202576, "logit_z_loss": 24.784934043884277, "router_aux_loss": 1.4472937434911728, "router_z_loss": 0.7887667864561081, "routing_entropy_normalized": 0.6667698088928243, "learning_rate": 0.003, "grad_norm": 1.1473231315612793, "tokens_per_second": 3939421.6081116833, "eta_seconds": 5839.88202548027, "data_wait_seconds": 2.6553869247436523e-05} {"step": 17650, "total_steps": 28610, "tokens": 37014732800, "source": "fineweb_hq", "phase": 3, "loss": 3.3003038465976715, "ce_loss": 3.2846343219280243, "logit_z_loss": 24.6070499420166, "router_aux_loss": 1.4682576656341553, "router_z_loss": 0.986969567835331, "routing_entropy_normalized": 0.6377672879701701, "learning_rate": 0.003, "grad_norm": 0.7726508975028992, "tokens_per_second": 3354859.4935581163, "eta_seconds": 6851.191820144653, "data_wait_seconds": 1.9043684005737305e-05} {"step": 17660, "total_steps": 28610, "tokens": 37035704320, "source": "fineweb_edu", "phase": 3, "loss": 3.3526116609573364, "ce_loss": 3.337493449449539, "logit_z_loss": 24.4471275806427, "router_aux_loss": 1.4067009091377258, "router_z_loss": 1.0512086749076843, "routing_entropy_normalized": 0.6557381041874184, "learning_rate": 0.003, "grad_norm": 1.8875223398208618, "tokens_per_second": 4028310.453148076, "eta_seconds": 5700.606908798218, "data_wait_seconds": 2.1666288375854492e-05} {"step": 17670, "total_steps": 28610, "tokens": 37056675840, "source": "dclm", "phase": 3, "loss": 3.473121255636215, "ce_loss": 3.458868592977524, "logit_z_loss": 24.831392765045166, "router_aux_loss": 1.3264103084802628, "router_z_loss": 0.9885624498128891, "routing_entropy_normalized": 0.6807734550847068, "learning_rate": 0.003, "grad_norm": 1.4006390571594238, "tokens_per_second": 3662888.6436195285, "eta_seconds": 6263.592784881592, "data_wait_seconds": 0.010670363903045654} {"step": 17680, "total_steps": 28610, "tokens": 37077647360, "source": "fineweb_edu", "phase": 3, "loss": 3.3664680421352386, "ce_loss": 3.352256119251251, "logit_z_loss": 24.39799976348877, "router_aux_loss": 1.3092851042747498, "router_z_loss": 1.1190771386027336, "routing_entropy_normalized": 0.6778740610717561, "learning_rate": 0.003, "grad_norm": 1.0714912414550781, "tokens_per_second": 3879255.7897624364, "eta_seconds": 5908.832158088684, "data_wait_seconds": 2.104043960571289e-05} {"step": 17690, "total_steps": 28610, "tokens": 37098618880, "source": "fineweb_hq", "phase": 3, "loss": 3.29784494638443, "ce_loss": 3.283111572265625, "logit_z_loss": 24.42347264289856, "router_aux_loss": 1.374277874827385, "router_z_loss": 0.9905321002006531, "routing_entropy_normalized": 0.6598852621943081, "learning_rate": 0.003, "grad_norm": 0.9192386865615845, "tokens_per_second": 3412651.959935207, "eta_seconds": 6710.587574958801, "data_wait_seconds": 1.862645149230957e-05} {"step": 17700, "total_steps": 28610, "tokens": 37119590400, "source": "fineweb_edu", "phase": 3, "loss": 3.3333851397037506, "ce_loss": 3.3186445832252502, "logit_z_loss": 25.03797698020935, "router_aux_loss": 1.3819303065538406, "router_z_loss": 0.9212476462125778, "routing_entropy_normalized": 0.6669849670047653, "learning_rate": 0.003, "grad_norm": 2.312882661819458, "tokens_per_second": 4098111.660904321, "eta_seconds": 5583.041706323624, "data_wait_seconds": 2.053380012512207e-05} {"step": 17710, "total_steps": 28610, "tokens": 37140561920, "source": "finemath", "phase": 3, "loss": 2.3287723660469055, "ce_loss": 2.3137225806713104, "logit_z_loss": 30.235930681228638, "router_aux_loss": 1.425756812095642, "router_z_loss": 0.7922802492976189, "routing_entropy_normalized": 0.674564725773923, "learning_rate": 0.003, "grad_norm": 0.9686111807823181, "tokens_per_second": 3836661.918814894, "eta_seconds": 5958.032603263855, "data_wait_seconds": 2.0712614059448242e-05} {"step": 17720, "total_steps": 28610, "tokens": 37161533440, "source": "dclm", "phase": 3, "loss": 3.466715157032013, "ce_loss": 3.4506785571575165, "logit_z_loss": 24.55502414703369, "router_aux_loss": 1.5088365375995636, "router_z_loss": 0.94821447879076, "routing_entropy_normalized": 0.640249646722826, "learning_rate": 0.003, "grad_norm": 1.0468367338180542, "tokens_per_second": 3823392.35993268, "eta_seconds": 5973.225641012192, "data_wait_seconds": 2.0325183868408203e-05} {"step": 17730, "total_steps": 28610, "tokens": 37182504960, "source": "fineweb_hq", "phase": 3, "loss": 3.3168430030345917, "ce_loss": 3.3001132011413574, "logit_z_loss": 24.61698627471924, "router_aux_loss": 1.5576764196157455, "router_z_loss": 1.153000146150589, "routing_entropy_normalized": 0.6443077291517784, "learning_rate": 0.003, "grad_norm": 2.3213891983032227, "tokens_per_second": 4150595.0879708286, "eta_seconds": 5497.287322998047, "data_wait_seconds": 2.1278858184814453e-05} {"step": 17740, "total_steps": 28610, "tokens": 37203476480, "source": "dclm", "phase": 3, "loss": 3.433512419462204, "ce_loss": 3.4156695306301117, "logit_z_loss": 25.17362356185913, "router_aux_loss": 1.6609113365411758, "router_z_loss": 1.2337278723716736, "routing_entropy_normalized": 0.6359364890002195, "learning_rate": 0.003, "grad_norm": 1.2919278144836426, "tokens_per_second": 3940788.1879250435, "eta_seconds": 5784.640318870544, "data_wait_seconds": 2.002716064453125e-05} {"step": 17750, "total_steps": 28610, "tokens": 37224448000, "source": "fineweb_hq", "phase": 3, "loss": 3.326170325279236, "ce_loss": 3.309784382581711, "logit_z_loss": 24.766950845718384, "router_aux_loss": 1.482612356543541, "router_z_loss": 1.559807538986206, "routing_entropy_normalized": 0.665763325496546, "learning_rate": 0.003, "grad_norm": 0.763513445854187, "tokens_per_second": 3853327.76727338, "eta_seconds": 5910.4940185546875, "data_wait_seconds": 0.04905027151107788} {"step": 17760, "total_steps": 28610, "tokens": 37245419520, "source": "fineweb_edu", "phase": 3, "loss": 3.3510487973690033, "ce_loss": 3.335819572210312, "logit_z_loss": 24.617177963256836, "router_aux_loss": 1.3682013899087906, "router_z_loss": 1.547162115573883, "routing_entropy_normalized": 0.6798096369076367, "learning_rate": 0.003, "grad_norm": 0.9154946208000183, "tokens_per_second": 4151434.685071984, "eta_seconds": 5481.020641326904, "data_wait_seconds": 2.3543834686279297e-05} {"step": 17770, "total_steps": 28610, "tokens": 37266391040, "source": "dclm", "phase": 3, "loss": 3.427422136068344, "ce_loss": 3.412559688091278, "logit_z_loss": 24.93564224243164, "router_aux_loss": 1.347600132226944, "router_z_loss": 1.3864801228046417, "routing_entropy_normalized": 0.6898939945086747, "learning_rate": 0.003, "grad_norm": 1.310983657836914, "tokens_per_second": 3905331.0100239767, "eta_seconds": 5821.050154685974, "data_wait_seconds": 2.2083520889282227e-05} {"step": 17780, "total_steps": 28610, "tokens": 37287362560, "source": "fineweb_edu", "phase": 3, "loss": 3.3590272963047028, "ce_loss": 3.34381240606308, "logit_z_loss": 23.979005813598633, "router_aux_loss": 1.3800702095031738, "router_z_loss": 1.4141421169042587, "routing_entropy_normalized": 0.6763996100293849, "learning_rate": 0.003, "grad_norm": 1.500463843345642, "tokens_per_second": 4015886.884222009, "eta_seconds": 5655.576667070389, "data_wait_seconds": 2.1398067474365234e-05} {"step": 17790, "total_steps": 28610, "tokens": 37308334080, "source": "fineweb_hq", "phase": 3, "loss": 3.314726561307907, "ce_loss": 3.299572378396988, "logit_z_loss": 25.31283473968506, "router_aux_loss": 1.3683734387159348, "router_z_loss": 1.4704055786132812, "routing_entropy_normalized": 0.6752845253274823, "learning_rate": 0.003, "grad_norm": 0.9706689715385437, "tokens_per_second": 3314939.1914106146, "eta_seconds": 6845.128471374512, "data_wait_seconds": 2.1666288375854492e-05} {"step": 17800, "total_steps": 28610, "tokens": 37329305600, "source": "fineweb_edu", "phase": 3, "loss": 3.4118747413158417, "ce_loss": 3.395516484975815, "logit_z_loss": 24.101649522781372, "router_aux_loss": 1.4971456080675125, "router_z_loss": 1.386727899312973, "routing_entropy_normalized": 0.637093832718973, "learning_rate": 0.003, "grad_norm": 1.5159856081008911, "tokens_per_second": 4043180.2218908407, "eta_seconds": 5607.025132656097, "data_wait_seconds": 2.2411346435546875e-05} {"step": 17810, "total_steps": 28610, "tokens": 37350277120, "source": "finemath", "phase": 3, "loss": 2.3964702785015106, "ce_loss": 2.379713147878647, "logit_z_loss": 33.414294719696045, "router_aux_loss": 1.5597241520881653, "router_z_loss": 1.1599157750606537, "routing_entropy_normalized": 0.6423300168351148, "learning_rate": 0.003, "grad_norm": 0.8345873355865479, "tokens_per_second": 3833692.308393139, "eta_seconds": 5907.944555282593, "data_wait_seconds": 1.8715858459472656e-05} {"step": 17820, "total_steps": 28610, "tokens": 37371248640, "source": "dclm", "phase": 3, "loss": 3.4543535113334656, "ce_loss": 3.4371736645698547, "logit_z_loss": 25.298934936523438, "router_aux_loss": 1.5661948323249817, "router_z_loss": 1.517906665802002, "routing_entropy_normalized": 0.6348423276718693, "learning_rate": 0.003, "grad_norm": 0.8078204393386841, "tokens_per_second": 3589924.5247197933, "eta_seconds": 6303.271816492081, "data_wait_seconds": 0.09560465812683105} {"step": 17830, "total_steps": 28610, "tokens": 37392220160, "source": "fineweb_hq", "phase": 3, "loss": 3.305626928806305, "ce_loss": 3.289355605840683, "logit_z_loss": 25.07773756980896, "router_aux_loss": 1.4600486010313034, "router_z_loss": 1.6707517206668854, "routing_entropy_normalized": 0.6367030318725705, "learning_rate": 0.003, "grad_norm": 2.0068840980529785, "tokens_per_second": 4152118.4063492185, "eta_seconds": 5444.762491703033, "data_wait_seconds": 2.288818359375e-05} {"step": 17840, "total_steps": 28610, "tokens": 37413191680, "source": "dclm", "phase": 3, "loss": 3.444330185651779, "ce_loss": 3.42882838845253, "logit_z_loss": 25.62677001953125, "router_aux_loss": 1.401903972029686, "router_z_loss": 1.4828229248523712, "routing_entropy_normalized": 0.6442535336897015, "learning_rate": 0.003, "grad_norm": 1.4057356119155884, "tokens_per_second": 3915676.479247557, "eta_seconds": 5768.180073022842, "data_wait_seconds": 0.07302814722061157} {"step": 17850, "total_steps": 28610, "tokens": 37434163200, "source": "fineweb_hq", "phase": 3, "loss": 3.265685647726059, "ce_loss": 3.2505199313163757, "logit_z_loss": 24.675514698028564, "router_aux_loss": 1.3521111756563187, "router_z_loss": 1.6445473730564117, "routing_entropy_normalized": 0.6459961784332493, "learning_rate": 0.003, "grad_norm": 1.6375290155410767, "tokens_per_second": 4121308.670436122, "eta_seconds": 5475.288876533508, "data_wait_seconds": 2.0802021026611328e-05} {"step": 17860, "total_steps": 28610, "tokens": 37455134720, "source": "fineweb_edu", "phase": 3, "loss": 3.3678057193756104, "ce_loss": 3.352949470281601, "logit_z_loss": 23.96881103515625, "router_aux_loss": 1.3238751739263535, "router_z_loss": 1.6174566745758057, "routing_entropy_normalized": 0.6692165642431829, "learning_rate": 0.003, "grad_norm": 1.026010513305664, "tokens_per_second": 3380478.268131197, "eta_seconds": 6668.992435932159, "data_wait_seconds": 2.5600194931030273e-05} {"step": 17870, "total_steps": 28610, "tokens": 37476106240, "source": "dclm", "phase": 3, "loss": 3.3920893371105194, "ce_loss": 3.3773650228977203, "logit_z_loss": 25.498842000961304, "router_aux_loss": 1.3335080593824387, "router_z_loss": 1.3892405480146408, "routing_entropy_normalized": 0.6572271249170257, "learning_rate": 0.003, "grad_norm": 0.5864564180374146, "tokens_per_second": 3896532.743689088, "eta_seconds": 5780.372952461243, "data_wait_seconds": 2.086162567138672e-05} {"step": 17880, "total_steps": 28610, "tokens": 37497077760, "source": "fineweb_edu", "phase": 3, "loss": 3.367398679256439, "ce_loss": 3.3523377776145935, "logit_z_loss": 24.875343561172485, "router_aux_loss": 1.3684809803962708, "router_z_loss": 1.3761563152074814, "routing_entropy_normalized": 0.671997857789017, "learning_rate": 0.003, "grad_norm": 1.2308597564697266, "tokens_per_second": 4408370.252980944, "eta_seconds": 5104.480719327927, "data_wait_seconds": 2.2202730178833008e-05} {"step": 17890, "total_steps": 28610, "tokens": 37518049280, "source": "fineweb_hq", "phase": 3, "loss": 3.323055535554886, "ce_loss": 3.307303786277771, "logit_z_loss": 25.562698364257812, "router_aux_loss": 1.4418056309223175, "router_z_loss": 1.3337193876504898, "routing_entropy_normalized": 0.6631216628304465, "learning_rate": 0.003, "grad_norm": 0.9099432826042175, "tokens_per_second": 3487041.8369269385, "eta_seconds": 6447.146461486816, "data_wait_seconds": 2.1278858184814453e-05} {"step": 17900, "total_steps": 28610, "tokens": 37539020800, "source": "fineweb_edu", "phase": 3, "loss": 3.407179683446884, "ce_loss": 3.3919640481472015, "logit_z_loss": 24.273200750350952, "router_aux_loss": 1.3886642158031464, "router_z_loss": 1.3289268612861633, "routing_entropy_normalized": 0.6471017248330745, "learning_rate": 0.003, "grad_norm": 0.8198996186256409, "tokens_per_second": 4314556.81651467, "eta_seconds": 5205.748556613922, "data_wait_seconds": 2.384185791015625e-05} {"step": 17910, "total_steps": 28610, "tokens": 37559992320, "source": "finemath", "phase": 3, "loss": 2.3878025114536285, "ce_loss": 2.3725355565547943, "logit_z_loss": 30.161662101745605, "router_aux_loss": 1.4307870119810104, "router_z_loss": 0.9591121673583984, "routing_entropy_normalized": 0.6587163777418843, "learning_rate": 0.003, "grad_norm": 0.8845219612121582, "tokens_per_second": 4002094.490202434, "eta_seconds": 5606.945676803589, "data_wait_seconds": 1.7851591110229492e-05} {"step": 17920, "total_steps": 28610, "tokens": 37580963840, "source": "dclm", "phase": 3, "loss": 3.4346151053905487, "ce_loss": 3.4199869632720947, "logit_z_loss": 24.880874395370483, "router_aux_loss": 1.3438083678483963, "router_z_loss": 1.1901337802410126, "routing_entropy_normalized": 0.6847347800809706, "learning_rate": 0.003, "grad_norm": 1.4748625755310059, "tokens_per_second": 4175241.234261546, "eta_seconds": 5369.403496026993, "data_wait_seconds": 0.011773258447647095} {"step": 17930, "total_steps": 28610, "tokens": 37601935360, "source": "fineweb_hq", "phase": 3, "loss": 3.3155385851860046, "ce_loss": 3.3007574379444122, "logit_z_loss": 24.796117067337036, "router_aux_loss": 1.3583599030971527, "router_z_loss": 1.1975300461053848, "routing_entropy_normalized": 0.64726775888595, "learning_rate": 0.003, "grad_norm": 1.594901204109192, "tokens_per_second": 4180176.346928537, "eta_seconds": 5358.047484397888, "data_wait_seconds": 2.4318695068359375e-05} {"step": 17940, "total_steps": 28610, "tokens": 37622906880, "source": "dclm", "phase": 3, "loss": 3.446596533060074, "ce_loss": 3.431547611951828, "logit_z_loss": 24.884423971176147, "router_aux_loss": 1.4022351950407028, "router_z_loss": 1.0265497490763664, "routing_entropy_normalized": 0.6562227439225504, "learning_rate": 0.003, "grad_norm": 2.070326328277588, "tokens_per_second": 3717717.1067016213, "eta_seconds": 6018.9119284152985, "data_wait_seconds": 2.1755695343017578e-05} {"step": 17950, "total_steps": 28610, "tokens": 37643878400, "source": "fineweb_hq", "phase": 3, "loss": 3.2728666961193085, "ce_loss": 3.2585804164409637, "logit_z_loss": 25.78435444831848, "router_aux_loss": 1.3213902711868286, "router_z_loss": 1.0723663121461868, "routing_entropy_normalized": 0.6685846869618193, "learning_rate": 0.003, "grad_norm": 0.6729200482368469, "tokens_per_second": 4097185.088569016, "eta_seconds": 5456.341326236725, "data_wait_seconds": 2.1398067474365234e-05} {"step": 17960, "total_steps": 28610, "tokens": 37664849920, "source": "fineweb_edu", "phase": 3, "loss": 3.380874365568161, "ce_loss": 3.3658357858657837, "logit_z_loss": 24.370803594589233, "router_aux_loss": 1.3952115774154663, "router_z_loss": 1.0864188820123672, "routing_entropy_normalized": 0.6317722170336404, "learning_rate": 0.003, "grad_norm": 1.5430359840393066, "tokens_per_second": 4229496.222708413, "eta_seconds": 5280.692456960678, "data_wait_seconds": 2.1845102310180664e-05} {"step": 17970, "total_steps": 28610, "tokens": 37685821440, "source": "dclm", "phase": 3, "loss": 3.479906588792801, "ce_loss": 3.463525652885437, "logit_z_loss": 25.556050777435303, "router_aux_loss": 1.5324480384588242, "router_z_loss": 1.0564913377165794, "routing_entropy_normalized": 0.6236817699250646, "learning_rate": 0.003, "grad_norm": 1.107230305671692, "tokens_per_second": 3705813.360431491, "eta_seconds": 6021.268507003784, "data_wait_seconds": 1.9073486328125e-05} {"step": 17980, "total_steps": 28610, "tokens": 37706792960, "source": "fineweb_edu", "phase": 3, "loss": 3.34812268614769, "ce_loss": 3.3315394520759583, "logit_z_loss": 24.571337699890137, "router_aux_loss": 1.5299399346113205, "router_z_loss": 1.2838392853736877, "routing_entropy_normalized": 0.6491542317129042, "learning_rate": 0.003, "grad_norm": 1.3467509746551514, "tokens_per_second": 4244272.28205294, "eta_seconds": 5252.425923347473, "data_wait_seconds": 2.0295381546020508e-05} {"step": 17990, "total_steps": 28610, "tokens": 37727764480, "source": "fineweb_hq", "phase": 3, "loss": 3.3128571212291718, "ce_loss": 3.297247737646103, "logit_z_loss": 24.41801357269287, "router_aux_loss": 1.4267363250255585, "router_z_loss": 1.342032104730606, "routing_entropy_normalized": 0.6512959577013553, "learning_rate": 0.003, "grad_norm": 1.526719570159912, "tokens_per_second": 3410349.324562568, "eta_seconds": 6530.637222290039, "data_wait_seconds": 0.11643621325492859} {"step": 18000, "total_steps": 28610, "tokens": 37748736000, "source": "fineweb_edu", "phase": 3, "loss": 3.3625846207141876, "ce_loss": 3.3471546471118927, "logit_z_loss": 25.111441612243652, "router_aux_loss": 1.4120576679706573, "router_z_loss": 1.3093798756599426, "routing_entropy_normalized": 0.6457051844378379, "learning_rate": 0.003, "grad_norm": 2.3575875759124756, "tokens_per_second": 4150095.7221099627, "eta_seconds": 5361.510724067688, "data_wait_seconds": 0.011655062437057495} {"step": 18010, "total_steps": 28610, "tokens": 37769707520, "source": "finemath", "phase": 3, "loss": 2.3279827535152435, "ce_loss": 2.3133437633514404, "logit_z_loss": 31.666550636291504, "router_aux_loss": 1.3686701506376266, "router_z_loss": 0.9523611292243004, "routing_entropy_normalized": 0.6747024402425279, "learning_rate": 0.003, "grad_norm": 0.9433099031448364, "tokens_per_second": 3967897.2817204795, "eta_seconds": 5602.415995597839, "data_wait_seconds": 1.806020736694336e-05} {"step": 18020, "total_steps": 28610, "tokens": 37790679040, "source": "dclm", "phase": 3, "loss": 3.4583860039711, "ce_loss": 3.442912757396698, "logit_z_loss": 25.321710109710693, "router_aux_loss": 1.438650220632553, "router_z_loss": 1.0867395997047424, "routing_entropy_normalized": 0.6530529539573697, "learning_rate": 0.003, "grad_norm": 1.7424213886260986, "tokens_per_second": 3736546.608596464, "eta_seconds": 5943.680624485016, "data_wait_seconds": 0.06887036561965942} {"step": 18030, "total_steps": 28610, "tokens": 37811650560, "source": "fineweb_hq", "phase": 3, "loss": 3.340130865573883, "ce_loss": 3.324798583984375, "logit_z_loss": 25.584489583969116, "router_aux_loss": 1.4080296456813812, "router_z_loss": 1.2520048916339874, "routing_entropy_normalized": 0.6550590320974244, "learning_rate": 0.003, "grad_norm": 0.8848556280136108, "tokens_per_second": 4399176.659587848, "eta_seconds": 5043.641089439392, "data_wait_seconds": 1.9848346710205078e-05} {"step": 18040, "total_steps": 28610, "tokens": 37832622080, "source": "dclm", "phase": 3, "loss": 3.466101199388504, "ce_loss": 3.4505976140499115, "logit_z_loss": 24.538164138793945, "router_aux_loss": 1.432031363248825, "router_z_loss": 1.1832455396652222, "routing_entropy_normalized": 0.6371432689767219, "learning_rate": 0.003, "grad_norm": 1.5611788034439087, "tokens_per_second": 3818762.900706138, "eta_seconds": 5804.732374429703, "data_wait_seconds": 2.205371856689453e-05} {"step": 18050, "total_steps": 28610, "tokens": 37853593600, "source": "fineweb_hq", "phase": 3, "loss": 3.327076107263565, "ce_loss": 3.3117620050907135, "logit_z_loss": 24.900370359420776, "router_aux_loss": 1.4134320169687271, "router_z_loss": 1.1797676086425781, "routing_entropy_normalized": 0.6515340012906008, "learning_rate": 0.003, "grad_norm": 1.8190455436706543, "tokens_per_second": 4265064.98371064, "eta_seconds": 5192.4003982543945, "data_wait_seconds": 1.895427703857422e-05} {"step": 18060, "total_steps": 28610, "tokens": 37874565120, "source": "fineweb_edu", "phase": 3, "loss": 3.374703049659729, "ce_loss": 3.3599418997764587, "logit_z_loss": 24.253047227859497, "router_aux_loss": 1.3489198833703995, "router_z_loss": 1.2719830423593521, "routing_entropy_normalized": 0.6841054153706237, "learning_rate": 0.003, "grad_norm": 1.5919729471206665, "tokens_per_second": 3442489.689843804, "eta_seconds": 6427.021020650864, "data_wait_seconds": 2.256035804748535e-05} {"step": 18070, "total_steps": 28610, "tokens": 37895536640, "source": "dclm", "phase": 3, "loss": 3.461077332496643, "ce_loss": 3.4467235803604126, "logit_z_loss": 25.32554006576538, "router_aux_loss": 1.3319932669401169, "router_z_loss": 1.0338010042905807, "routing_entropy_normalized": 0.6467042577226174, "learning_rate": 0.003, "grad_norm": 1.2188127040863037, "tokens_per_second": 4144288.4159638034, "eta_seconds": 5333.601299285889, "data_wait_seconds": 1.940131187438965e-05} {"step": 18080, "total_steps": 28610, "tokens": 37916508160, "source": "fineweb_edu", "phase": 3, "loss": 3.397894114255905, "ce_loss": 3.38310369849205, "logit_z_loss": 24.777568340301514, "router_aux_loss": 1.3649069219827652, "router_z_loss": 1.1413752734661102, "routing_entropy_normalized": 0.6574812823597401, "learning_rate": 0.003, "grad_norm": 1.0381901264190674, "tokens_per_second": 3947751.20463618, "eta_seconds": 5593.820232152939, "data_wait_seconds": 1.913309097290039e-05} {"step": 18090, "total_steps": 28610, "tokens": 37937479680, "source": "fineweb_hq", "phase": 3, "loss": 3.2789982855319977, "ce_loss": 3.2640700340270996, "logit_z_loss": 26.562762022018433, "router_aux_loss": 1.3887583017349243, "router_z_loss": 1.0405972599983215, "routing_entropy_normalized": 0.6727165755792197, "learning_rate": 0.003, "grad_norm": 1.3203425407409668, "tokens_per_second": 3512146.2525156937, "eta_seconds": 6281.640186309814, "data_wait_seconds": 2.0056962966918945e-05} {"step": 18100, "total_steps": 28610, "tokens": 37958451200, "source": "fineweb_edu", "phase": 3, "loss": 3.3527754843235016, "ce_loss": 3.3374935388565063, "logit_z_loss": 25.137675046920776, "router_aux_loss": 1.419297605752945, "router_z_loss": 1.088980257511139, "routing_entropy_normalized": 0.6170014034752601, "learning_rate": 0.003, "grad_norm": 2.01607346534729, "tokens_per_second": 3913119.4448874365, "eta_seconds": 5632.607905387878, "data_wait_seconds": 2.1278858184814453e-05} {"step": 18110, "total_steps": 28610, "tokens": 37979422720, "source": "finemath", "phase": 3, "loss": 2.362465739250183, "ce_loss": 2.3481991589069366, "logit_z_loss": 33.458189249038696, "router_aux_loss": 1.3357514142990112, "router_z_loss": 0.9090652838349342, "routing_entropy_normalized": 0.6795524591330685, "learning_rate": 0.003, "grad_norm": 1.4185540676116943, "tokens_per_second": 2794224.7778241537, "eta_seconds": 7880.574309825897, "data_wait_seconds": 1.901388168334961e-05} {"step": 18120, "total_steps": 28610, "tokens": 38000394240, "source": "dclm", "phase": 3, "loss": 3.4579552114009857, "ce_loss": 3.4435031414031982, "logit_z_loss": 24.915254831314087, "router_aux_loss": 1.3466869294643402, "router_z_loss": 0.9852578192949295, "routing_entropy_normalized": 0.6796839051482882, "learning_rate": 0.003, "grad_norm": 1.3526272773742676, "tokens_per_second": 4048551.6465215455, "eta_seconds": 5433.825822353363, "data_wait_seconds": 2.2739171981811523e-05} {"step": 18130, "total_steps": 28610, "tokens": 38021365760, "source": "fineweb_hq", "phase": 3, "loss": 3.28786501288414, "ce_loss": 3.273211270570755, "logit_z_loss": 25.54438090324402, "router_aux_loss": 1.3683298826217651, "router_z_loss": 0.9704763069748878, "routing_entropy_normalized": 0.6272829959485686, "learning_rate": 0.003, "grad_norm": 1.910873532295227, "tokens_per_second": 3552294.267639846, "eta_seconds": 6187.030494689941, "data_wait_seconds": 0.11690080165863037} {"step": 18140, "total_steps": 28610, "tokens": 38042337280, "source": "dclm", "phase": 3, "loss": 3.497174561023712, "ce_loss": 3.481550544500351, "logit_z_loss": 24.54003882408142, "router_aux_loss": 1.464958518743515, "router_z_loss": 0.9744698330760002, "routing_entropy_normalized": 0.6376740852373345, "learning_rate": 0.003, "grad_norm": 1.31914222240448, "tokens_per_second": 3889616.6341384184, "eta_seconds": 5645.0759818553925, "data_wait_seconds": 1.8894672393798828e-05} {"step": 18150, "total_steps": 28610, "tokens": 38063308800, "source": "fineweb_hq", "phase": 3, "loss": 3.3082764744758606, "ce_loss": 3.292077511548996, "logit_z_loss": 24.072476148605347, "router_aux_loss": 1.5038578659296036, "router_z_loss": 1.160403162240982, "routing_entropy_normalized": 0.64585933319354, "learning_rate": 0.003, "grad_norm": 0.9337932467460632, "tokens_per_second": 4202480.508531074, "eta_seconds": 5219.824309825897, "data_wait_seconds": 2.002716064453125e-05} {"step": 18160, "total_steps": 28610, "tokens": 38084280320, "source": "fineweb_edu", "phase": 3, "loss": 3.376528352499008, "ce_loss": 3.3608825504779816, "logit_z_loss": 24.871715545654297, "router_aux_loss": 1.4407124817371368, "router_z_loss": 1.2385993003845215, "routing_entropy_normalized": 0.6475211744878903, "learning_rate": 0.003, "grad_norm": 1.1889307498931885, "tokens_per_second": 3858496.710402267, "eta_seconds": 5679.734892845154, "data_wait_seconds": 2.014636993408203e-05} {"step": 18170, "total_steps": 28610, "tokens": 38105251840, "source": "dclm", "phase": 3, "loss": 3.5004610419273376, "ce_loss": 3.4861283600330353, "logit_z_loss": 25.81241464614868, "router_aux_loss": 1.3198790699243546, "router_z_loss": 1.133891984820366, "routing_entropy_normalized": 0.6704862412594603, "learning_rate": 0.003, "grad_norm": 1.4934940338134766, "tokens_per_second": 3823727.4306757855, "eta_seconds": 5725.896334648132, "data_wait_seconds": 2.1725893020629883e-05} {"step": 18180, "total_steps": 28610, "tokens": 38126223360, "source": "fineweb_edu", "phase": 3, "loss": 3.379391998052597, "ce_loss": 3.3641838133335114, "logit_z_loss": 23.837749242782593, "router_aux_loss": 1.3994940370321274, "router_z_loss": 1.2132091373205185, "routing_entropy_normalized": 0.6565077865323894, "learning_rate": 0.003, "grad_norm": 1.499091386795044, "tokens_per_second": 3815025.8653103663, "eta_seconds": 5733.45925617218, "data_wait_seconds": 2.2083520889282227e-05} {"step": 18190, "total_steps": 28610, "tokens": 38147194880, "source": "fineweb_hq", "phase": 3, "loss": 3.3444380462169647, "ce_loss": 3.329436182975769, "logit_z_loss": 24.771594285964966, "router_aux_loss": 1.3802452087402344, "router_z_loss": 1.1994338184595108, "routing_entropy_normalized": 0.6585170204185479, "learning_rate": 0.003, "grad_norm": 1.467956304550171, "tokens_per_second": 3625342.4676797297, "eta_seconds": 6027.657810211182, "data_wait_seconds": 2.1845102310180664e-05} {"step": 18200, "total_steps": 28610, "tokens": 38168166400, "source": "fineweb_edu", "phase": 3, "loss": 3.3833925426006317, "ce_loss": 3.368050307035446, "logit_z_loss": 24.11032748222351, "router_aux_loss": 1.408556580543518, "router_z_loss": 1.2566956728696823, "routing_entropy_normalized": 0.6613285001600343, "learning_rate": 0.003, "grad_norm": 1.2014036178588867, "tokens_per_second": 4316335.263138112, "eta_seconds": 5057.844441890717, "data_wait_seconds": 2.16066837310791e-05} {"step": 18210, "total_steps": 28610, "tokens": 38189137920, "source": "finemath", "phase": 3, "loss": 2.337025612592697, "ce_loss": 2.3227722346782684, "logit_z_loss": 31.906128883361816, "router_aux_loss": 1.3244031071662903, "router_z_loss": 1.0093708485364914, "routing_entropy_normalized": 0.6887727768283358, "learning_rate": 0.003, "grad_norm": 1.1165915727615356, "tokens_per_second": 3960614.1094205785, "eta_seconds": 5506.817932128906, "data_wait_seconds": 2.0802021026611328e-05} {"step": 18220, "total_steps": 28610, "tokens": 38210109440, "source": "dclm", "phase": 3, "loss": 3.4976080656051636, "ce_loss": 3.4825603663921356, "logit_z_loss": 25.574585676193237, "router_aux_loss": 1.394144430756569, "router_z_loss": 1.106247901916504, "routing_entropy_normalized": 0.6523163078162141, "learning_rate": 0.003, "grad_norm": 1.2764326333999634, "tokens_per_second": 3021122.8113647834, "eta_seconds": 7212.354690790176, "data_wait_seconds": 1.901388168334961e-05} {"step": 18230, "total_steps": 28610, "tokens": 38231080960, "source": "fineweb_hq", "phase": 3, "loss": 3.340703457593918, "ce_loss": 3.3255776166915894, "logit_z_loss": 25.381622076034546, "router_aux_loss": 1.3830905109643936, "router_z_loss": 1.2949680536985397, "routing_entropy_normalized": 0.6566304067742054, "learning_rate": 0.003, "grad_norm": 1.8470007181167603, "tokens_per_second": 4516950.953860984, "eta_seconds": 4819.276981830597, "data_wait_seconds": 2.1755695343017578e-05} {"step": 18240, "total_steps": 28610, "tokens": 38252052480, "source": "dclm", "phase": 3, "loss": 3.55322602391243, "ce_loss": 3.5376822650432587, "logit_z_loss": 24.3511004447937, "router_aux_loss": 1.4407318383455276, "router_z_loss": 1.1363927870988846, "routing_entropy_normalized": 0.6268541195658219, "learning_rate": 0.003, "grad_norm": 1.8541432619094849, "tokens_per_second": 3975553.507129232, "eta_seconds": 5470.298966169357, "data_wait_seconds": 2.0563602447509766e-05} {"step": 18250, "total_steps": 28610, "tokens": 38273024000, "source": "fineweb_hq", "phase": 3, "loss": 3.317205995321274, "ce_loss": 3.3018836975097656, "logit_z_loss": 24.50979971885681, "router_aux_loss": 1.3981569558382034, "router_z_loss": 1.3407181203365326, "routing_entropy_normalized": 0.6407635051214158, "learning_rate": 0.003, "grad_norm": 1.0114357471466064, "tokens_per_second": 4143818.2866368834, "eta_seconds": 5243.109909057617, "data_wait_seconds": 2.008676528930664e-05} {"step": 18260, "total_steps": 28610, "tokens": 38293995520, "source": "fineweb_edu", "phase": 3, "loss": 3.3862554728984833, "ce_loss": 3.3708474934101105, "logit_z_loss": 24.66801691055298, "router_aux_loss": 1.4126942455768585, "router_z_loss": 1.2810349017381668, "routing_entropy_normalized": 0.6667602233152145, "learning_rate": 0.003, "grad_norm": 0.8063456416130066, "tokens_per_second": 3738519.5161142047, "eta_seconds": 5805.914107561111, "data_wait_seconds": 2.1338462829589844e-05} {"step": 18270, "total_steps": 28610, "tokens": 38314967040, "source": "dclm", "phase": 3, "loss": 3.486233741044998, "ce_loss": 3.4717634320259094, "logit_z_loss": 25.719382286071777, "router_aux_loss": 1.3262825310230255, "router_z_loss": 1.2074543237686157, "routing_entropy_normalized": 0.6905140914072646, "learning_rate": 0.003, "grad_norm": 1.7482279539108276, "tokens_per_second": 3775976.9624089072, "eta_seconds": 5742.765884399414, "data_wait_seconds": 2.142786979675293e-05} {"step": 18280, "total_steps": 28610, "tokens": 38335938560, "source": "fineweb_edu", "phase": 3, "loss": 3.3270759880542755, "ce_loss": 3.3122549653053284, "logit_z_loss": 24.338154792785645, "router_aux_loss": 1.357376515865326, "router_z_loss": 1.2472539842128754, "routing_entropy_normalized": 0.6782106103835146, "learning_rate": 0.003, "grad_norm": 1.6154195070266724, "tokens_per_second": 3518229.7740557217, "eta_seconds": 6157.522831439972, "data_wait_seconds": 2.110004425048828e-05} {"step": 18290, "total_steps": 28610, "tokens": 38356910080, "source": "fineweb_hq", "phase": 3, "loss": 3.3205294609069824, "ce_loss": 3.304897516965866, "logit_z_loss": 24.386632442474365, "router_aux_loss": 1.4423760324716568, "router_z_loss": 1.2081124633550644, "routing_entropy_normalized": 0.6493462686080866, "learning_rate": 0.003, "grad_norm": 2.2959794998168945, "tokens_per_second": 3985513.754162464, "eta_seconds": 5430.318391799927, "data_wait_seconds": 1.9669532775878906e-05} {"step": 18300, "total_steps": 28610, "tokens": 38377881600, "source": "fineweb_edu", "phase": 3, "loss": 3.364974081516266, "ce_loss": 3.3493958115577698, "logit_z_loss": 24.464712619781494, "router_aux_loss": 1.4301799833774567, "router_z_loss": 1.2764976620674133, "routing_entropy_normalized": 0.6615142531861007, "learning_rate": 0.003, "grad_norm": 2.1994969844818115, "tokens_per_second": 4238669.364015013, "eta_seconds": 5101.043573617934, "data_wait_seconds": 1.773238182067871e-05} {"step": 18310, "total_steps": 28610, "tokens": 38398853120, "source": "finemath", "phase": 3, "loss": 2.3249018490314484, "ce_loss": 2.3100712597370148, "logit_z_loss": 34.97286128997803, "router_aux_loss": 1.3865259736776352, "router_z_loss": 0.9653336554765701, "routing_entropy_normalized": 0.6686837312768082, "learning_rate": 0.003, "grad_norm": 0.9640939831733704, "tokens_per_second": 3721771.584400137, "eta_seconds": 5803.866548538208, "data_wait_seconds": 1.9729137420654297e-05} {"step": 18320, "total_steps": 28610, "tokens": 38419824640, "source": "dclm", "phase": 3, "loss": 3.555345892906189, "ce_loss": 3.540567547082901, "logit_z_loss": 26.04931330680847, "router_aux_loss": 1.3581131547689438, "router_z_loss": 1.1972245126962662, "routing_entropy_normalized": 0.6568074180464594, "learning_rate": 0.003, "grad_norm": 1.4427505731582642, "tokens_per_second": 3917066.751084886, "eta_seconds": 5509.146371841431, "data_wait_seconds": 0.057568758726119995} {"step": 18330, "total_steps": 28610, "tokens": 38440796160, "source": "fineweb_hq", "phase": 3, "loss": 3.325218081474304, "ce_loss": 3.3102163672447205, "logit_z_loss": 26.03860378265381, "router_aux_loss": 1.3687776625156403, "router_z_loss": 1.3139236718416214, "routing_entropy_normalized": 0.6449951050322577, "learning_rate": 0.003, "grad_norm": 2.494539499282837, "tokens_per_second": 3755284.023740009, "eta_seconds": 5740.9033308029175, "data_wait_seconds": 1.8924474716186523e-05} {"step": 18340, "total_steps": 28610, "tokens": 38461767680, "source": "dclm", "phase": 3, "loss": 3.5220249593257904, "ce_loss": 3.506781756877899, "logit_z_loss": 25.48506259918213, "router_aux_loss": 1.4132151454687119, "router_z_loss": 1.1110832244157791, "routing_entropy_normalized": 0.666623779965556, "learning_rate": 0.003, "grad_norm": 1.3696284294128418, "tokens_per_second": 4070221.3208987135, "eta_seconds": 5291.542975664139, "data_wait_seconds": 1.9311904907226562e-05} {"step": 18350, "total_steps": 28610, "tokens": 38482739200, "source": "fineweb_hq", "phase": 3, "loss": 3.312182366847992, "ce_loss": 3.296286940574646, "logit_z_loss": 24.69945526123047, "router_aux_loss": 1.4650714099407196, "router_z_loss": 1.244695395231247, "routing_entropy_normalized": 0.6423267984709464, "learning_rate": 0.003, "grad_norm": 1.129851222038269, "tokens_per_second": 3635870.893974149, "eta_seconds": 5917.9162702560425, "data_wait_seconds": 1.8388032913208008e-05} {"step": 18360, "total_steps": 28610, "tokens": 38503710720, "source": "fineweb_edu", "phase": 3, "loss": 3.4025627076625824, "ce_loss": 3.386717736721039, "logit_z_loss": 23.891204833984375, "router_aux_loss": 1.4471790790557861, "router_z_loss": 1.3731879591941833, "routing_entropy_normalized": 0.6480664208729799, "learning_rate": 0.003, "grad_norm": 1.5648243427276611, "tokens_per_second": 3183469.60794336, "eta_seconds": 6752.320784330368, "data_wait_seconds": 2.256035804748535e-05} {"step": 18370, "total_steps": 28610, "tokens": 38524682240, "source": "dclm", "phase": 3, "loss": 3.4716881811618805, "ce_loss": 3.4558655619621277, "logit_z_loss": 25.424863815307617, "router_aux_loss": 1.462513118982315, "router_z_loss": 1.1974903643131256, "routing_entropy_normalized": 0.6539525843156327, "learning_rate": 0.003, "grad_norm": 1.056372046470642, "tokens_per_second": 4194383.201495542, "eta_seconds": 5119.9033203125, "data_wait_seconds": 2.2172927856445312e-05} {"step": 18380, "total_steps": 28610, "tokens": 38545653760, "source": "fineweb_edu", "phase": 3, "loss": 3.3671785593032837, "ce_loss": 3.352419674396515, "logit_z_loss": 24.80110454559326, "router_aux_loss": 1.341844692826271, "router_z_loss": 1.3404508382081985, "routing_entropy_normalized": 0.6892714325481367, "learning_rate": 0.003, "grad_norm": 1.0139610767364502, "tokens_per_second": 3985444.2306085224, "eta_seconds": 5383.054866313934, "data_wait_seconds": 1.6927719116210938e-05} {"step": 18390, "total_steps": 28610, "tokens": 38566625280, "source": "fineweb_hq", "phase": 3, "loss": 3.3270024955272675, "ce_loss": 3.312393993139267, "logit_z_loss": 24.701342582702637, "router_aux_loss": 1.3339107632637024, "router_z_loss": 1.269385039806366, "routing_entropy_normalized": 0.6754696531761689, "learning_rate": 0.003, "grad_norm": 2.3674094676971436, "tokens_per_second": 3621767.838905009, "eta_seconds": 5917.798819065094, "data_wait_seconds": 1.9848346710205078e-05} {"step": 18400, "total_steps": 28610, "tokens": 38587596800, "source": "fineweb_edu", "phase": 3, "loss": 3.378690183162689, "ce_loss": 3.363670915365219, "logit_z_loss": 25.25020956993103, "router_aux_loss": 1.3745337426662445, "router_z_loss": 1.2740180790424347, "routing_entropy_normalized": 0.6197361078982014, "learning_rate": 0.003, "grad_norm": 2.2359871864318848, "tokens_per_second": 3850548.8997501195, "eta_seconds": 5560.745357990265, "data_wait_seconds": 2.205371856689453e-05} {"step": 18410, "total_steps": 28610, "tokens": 38608568320, "source": "finemath", "phase": 3, "loss": 2.3837768733501434, "ce_loss": 2.368278741836548, "logit_z_loss": 33.128830432891846, "router_aux_loss": 1.4605881720781326, "router_z_loss": 0.8922165483236313, "routing_entropy_normalized": 0.6483484738694009, "learning_rate": 0.003, "grad_norm": 1.0848393440246582, "tokens_per_second": 3377802.2760094553, "eta_seconds": 6332.80122756958, "data_wait_seconds": 2.2083520889282227e-05} {"step": 18420, "total_steps": 28610, "tokens": 38629539840, "source": "dclm", "phase": 3, "loss": 3.5004253685474396, "ce_loss": 3.4854437708854675, "logit_z_loss": 25.313573360443115, "router_aux_loss": 1.3925890773534775, "router_z_loss": 1.0557602494955063, "routing_entropy_normalized": 0.6527098918188392, "learning_rate": 0.003, "grad_norm": 1.113196849822998, "tokens_per_second": 4061238.6490863743, "eta_seconds": 5261.936253070831, "data_wait_seconds": 1.913309097290039e-05} {"step": 18430, "total_steps": 28610, "tokens": 38650511360, "source": "fineweb_hq", "phase": 3, "loss": 3.3572966158390045, "ce_loss": 3.3409531116485596, "logit_z_loss": 24.307607889175415, "router_aux_loss": 1.5146847516298294, "router_z_loss": 1.1966737359762192, "routing_entropy_normalized": 0.6247276520068118, "learning_rate": 0.003, "grad_norm": 1.8906317949295044, "tokens_per_second": 3149936.698332578, "eta_seconds": 6777.598855018616, "data_wait_seconds": 2.136826515197754e-05} {"step": 18440, "total_steps": 28610, "tokens": 38671482880, "source": "dclm", "phase": 3, "loss": 3.5155268609523773, "ce_loss": 3.4997279047966003, "logit_z_loss": 25.17887258529663, "router_aux_loss": 1.4690220803022385, "router_z_loss": 1.1087691113352776, "routing_entropy_normalized": 0.6659495384777934, "learning_rate": 0.003, "grad_norm": 1.5400623083114624, "tokens_per_second": 3649750.411000454, "eta_seconds": 5843.697085618973, "data_wait_seconds": 2.0295381546020508e-05} {"step": 18450, "total_steps": 28610, "tokens": 38692454400, "source": "fineweb_hq", "phase": 3, "loss": 3.3196908235549927, "ce_loss": 3.3039816319942474, "logit_z_loss": 25.583369731903076, "router_aux_loss": 1.438817873597145, "router_z_loss": 1.3210204392671585, "routing_entropy_normalized": 0.644325001450594, "learning_rate": 0.003, "grad_norm": 1.5327529907226562, "tokens_per_second": 4254074.382839528, "eta_seconds": 5008.625238418579, "data_wait_seconds": 1.9252300262451172e-05} {"step": 18460, "total_steps": 28610, "tokens": 38713425920, "source": "fineweb_edu", "phase": 3, "loss": 3.3788653016090393, "ce_loss": 3.3637519478797913, "logit_z_loss": 23.97015643119812, "router_aux_loss": 1.3821037411689758, "router_z_loss": 1.2923246771097183, "routing_entropy_normalized": 0.6491299546029305, "learning_rate": 0.003, "grad_norm": 2.6661298274993896, "tokens_per_second": 3386929.1186334076, "eta_seconds": 6284.776579141617, "data_wait_seconds": 2.1189451217651367e-05} {"step": 18470, "total_steps": 28610, "tokens": 38734397440, "source": "dclm", "phase": 3, "loss": 3.484946072101593, "ce_loss": 3.470412850379944, "logit_z_loss": 25.6887788772583, "router_aux_loss": 1.3383683562278748, "router_z_loss": 1.1495284289121628, "routing_entropy_normalized": 0.6767306322142596, "learning_rate": 0.003, "grad_norm": 1.4495280981063843, "tokens_per_second": 4178054.399159762, "eta_seconds": 5089.718622207642, "data_wait_seconds": 2.0503997802734375e-05} {"step": 18480, "total_steps": 28610, "tokens": 38755368960, "source": "fineweb_edu", "phase": 3, "loss": 3.342100977897644, "ce_loss": 3.3273210525512695, "logit_z_loss": 24.832521677017212, "router_aux_loss": 1.3515565991401672, "router_z_loss": 1.2643260061740875, "routing_entropy_normalized": 0.6693611545576187, "learning_rate": 0.003, "grad_norm": 1.6920138597488403, "tokens_per_second": 4058977.3987158337, "eta_seconds": 5233.867467880249, "data_wait_seconds": 2.193450927734375e-05} {"step": 18490, "total_steps": 28610, "tokens": 38776340480, "source": "fineweb_hq", "phase": 3, "loss": 3.3383282721042633, "ce_loss": 3.3229346573352814, "logit_z_loss": 24.80444860458374, "router_aux_loss": 1.4188532680273056, "router_z_loss": 1.2050982415676117, "routing_entropy_normalized": 0.639075654868112, "learning_rate": 0.003, "grad_norm": 1.5592657327651978, "tokens_per_second": 3599856.605676426, "eta_seconds": 5895.562119483948, "data_wait_seconds": 2.1398067474365234e-05} {"step": 18500, "total_steps": 28610, "tokens": 38797312000, "source": "fineweb_edu", "phase": 3, "loss": 3.350791484117508, "ce_loss": 3.334701269865036, "logit_z_loss": 24.907886266708374, "router_aux_loss": 1.4844870567321777, "router_z_loss": 1.2453971654176712, "routing_entropy_normalized": 0.6365595706190543, "learning_rate": 0.003, "grad_norm": 1.279558539390564, "tokens_per_second": 3922763.329297336, "eta_seconds": 5404.916111469269, "data_wait_seconds": 2.1010637283325195e-05} {"step": 18510, "total_steps": 28610, "tokens": 38818283520, "source": "finemath", "phase": 3, "loss": 2.3541025519371033, "ce_loss": 2.33963879942894, "logit_z_loss": 32.98648166656494, "router_aux_loss": 1.351222038269043, "router_z_loss": 0.9515647888183594, "routing_entropy_normalized": 0.6894974584814398, "learning_rate": 0.003, "grad_norm": 0.7934888005256653, "tokens_per_second": 3823401.500468401, "eta_seconds": 5539.892997741699, "data_wait_seconds": 1.7762184143066406e-05} {"step": 18520, "total_steps": 28610, "tokens": 38839255040, "source": "dclm", "phase": 3, "loss": 3.515172392129898, "ce_loss": 3.5001057386398315, "logit_z_loss": 25.04561448097229, "router_aux_loss": 1.3931588977575302, "router_z_loss": 1.1350278705358505, "routing_entropy_normalized": 0.6503268099769174, "learning_rate": 0.003, "grad_norm": 1.1160775423049927, "tokens_per_second": 3324711.45873272, "eta_seconds": 6364.541387319565, "data_wait_seconds": 2.2083520889282227e-05} {"step": 18530, "total_steps": 28610, "tokens": 38860226560, "source": "fineweb_hq", "phase": 3, "loss": 3.3722875714302063, "ce_loss": 3.3577147722244263, "logit_z_loss": 24.58322501182556, "router_aux_loss": 1.333778753876686, "router_z_loss": 1.2349929958581924, "routing_entropy_normalized": 0.6636417751622459, "learning_rate": 0.003, "grad_norm": 1.7758835554122925, "tokens_per_second": 4122149.4027201775, "eta_seconds": 5128.220764160156, "data_wait_seconds": 1.9848346710205078e-05} {"step": 18540, "total_steps": 28610, "tokens": 38881198080, "source": "dclm", "phase": 3, "loss": 3.4959638714790344, "ce_loss": 3.4790393114089966, "logit_z_loss": 24.957514762878418, "router_aux_loss": 1.5849553346633911, "router_z_loss": 1.0750713124871254, "routing_entropy_normalized": 0.6226700063472451, "learning_rate": 0.003, "grad_norm": 1.0759354829788208, "tokens_per_second": 3795788.4895648854, "eta_seconds": 5563.618915557861, "data_wait_seconds": 1.9758939743041992e-05} {"step": 18550, "total_steps": 28610, "tokens": 38902169600, "source": "fineweb_hq", "phase": 3, "loss": 3.3385873436927795, "ce_loss": 3.3221019208431244, "logit_z_loss": 24.90390944480896, "router_aux_loss": 1.5139793902635574, "router_z_loss": 1.3455915004014969, "routing_entropy_normalized": 0.6559280121371298, "learning_rate": 0.003, "grad_norm": 2.031869649887085, "tokens_per_second": 4363551.978183662, "eta_seconds": 4834.902672290802, "data_wait_seconds": 2.0742416381835938e-05} {"step": 18560, "total_steps": 28610, "tokens": 38923141120, "source": "fineweb_edu", "phase": 3, "loss": 3.3681657910346985, "ce_loss": 3.351716935634613, "logit_z_loss": 24.421794414520264, "router_aux_loss": 1.4990428239107132, "router_z_loss": 1.4583778083324432, "routing_entropy_normalized": 0.6391977728535698, "learning_rate": 0.003, "grad_norm": 1.8527601957321167, "tokens_per_second": 3375810.9515998033, "eta_seconds": 6243.352457284927, "data_wait_seconds": 2.327561378479004e-05} {"step": 18570, "total_steps": 28610, "tokens": 38944112640, "source": "dclm", "phase": 3, "loss": 3.5060667991638184, "ce_loss": 3.4906444549560547, "logit_z_loss": 25.737926721572876, "router_aux_loss": 1.4148225784301758, "router_z_loss": 1.2741331905126572, "routing_entropy_normalized": 0.6637146337986137, "learning_rate": 0.003, "grad_norm": 1.7477437257766724, "tokens_per_second": 4000489.8329726746, "eta_seconds": 5263.206996917725, "data_wait_seconds": 1.9341707229614258e-05} {"step": 18580, "total_steps": 28610, "tokens": 38965084160, "source": "fineweb_edu", "phase": 3, "loss": 3.371765285730362, "ce_loss": 3.35614275932312, "logit_z_loss": 24.81326460838318, "router_aux_loss": 1.416905716061592, "router_z_loss": 1.4534455984830856, "routing_entropy_normalized": 0.6583074459337453, "learning_rate": 0.003, "grad_norm": 0.6741142868995667, "tokens_per_second": 3988452.935192361, "eta_seconds": 5273.832962751389, "data_wait_seconds": 2.1159648895263672e-05} {"step": 18590, "total_steps": 28610, "tokens": 38986055680, "source": "fineweb_hq", "phase": 3, "loss": 3.3872369527816772, "ce_loss": 3.372279614210129, "logit_z_loss": 25.01662802696228, "router_aux_loss": 1.3458469659090042, "router_z_loss": 1.4989016354084015, "routing_entropy_normalized": 0.6668412825830383, "learning_rate": 0.003, "grad_norm": 2.3260886669158936, "tokens_per_second": 4337866.636574802, "eta_seconds": 4844.192964076996, "data_wait_seconds": 2.0205974578857422e-05} {"step": 18600, "total_steps": 28610, "tokens": 39007027200, "source": "fineweb_edu", "phase": 4, "loss": 3.356977164745331, "ce_loss": 3.3410177528858185, "logit_z_loss": 25.15068769454956, "router_aux_loss": 1.454938367009163, "router_z_loss": 1.4099940210580826, "routing_entropy_normalized": 0.6383126078873269, "learning_rate": 0.003, "grad_norm": 1.4419783353805542, "tokens_per_second": 4097340.6333519816, "eta_seconds": 5123.443081378937, "data_wait_seconds": 2.467632293701172e-05} {"step": 18610, "total_steps": 28610, "tokens": 39027998720, "source": "finemath", "phase": 4, "loss": 2.352415680885315, "ce_loss": 2.3369571566581726, "logit_z_loss": 32.71152329444885, "router_aux_loss": 1.442627564072609, "router_z_loss": 1.0322858914732933, "routing_entropy_normalized": 0.6614288875996039, "learning_rate": 0.003, "grad_norm": 0.9489884972572327, "tokens_per_second": 4237246.604136628, "eta_seconds": 4949.327230453491, "data_wait_seconds": 2.1785497665405273e-05} {"step": 18620, "total_steps": 28610, "tokens": 39048970240, "source": "fineweb_edu", "phase": 4, "loss": 3.3695145547389984, "ce_loss": 3.354689657688141, "logit_z_loss": 25.032029628753662, "router_aux_loss": 1.331254318356514, "router_z_loss": 1.5123330056667328, "routing_entropy_normalized": 0.6692368704805528, "learning_rate": 0.003, "grad_norm": 1.751307487487793, "tokens_per_second": 3547439.7996064546, "eta_seconds": 5905.822131872177, "data_wait_seconds": 2.1249055862426758e-05} {"step": 18630, "total_steps": 28610, "tokens": 39069941760, "source": "fineweb_hq", "phase": 4, "loss": 3.443507194519043, "ce_loss": 3.4279642701148987, "logit_z_loss": 25.826094150543213, "router_aux_loss": 1.4122733026742935, "router_z_loss": 1.4201382845640182, "routing_entropy_normalized": 0.6404733912450358, "learning_rate": 0.003, "grad_norm": 1.7951905727386475, "tokens_per_second": 4019718.0817300216, "eta_seconds": 5206.727570056915, "data_wait_seconds": 1.919269561767578e-05} {"step": 18640, "total_steps": 28610, "tokens": 39090913280, "source": "cosmopedia_v2", "phase": 4, "loss": 2.840245008468628, "ce_loss": 2.8245243430137634, "logit_z_loss": 27.22425150871277, "router_aux_loss": 1.3899383395910263, "router_z_loss": 1.8213043808937073, "routing_entropy_normalized": 0.6948945067125498, "learning_rate": 0.003, "grad_norm": 1.6074490547180176, "tokens_per_second": 4393085.399762048, "eta_seconds": 4759.435234546661, "data_wait_seconds": 2.2083520889282227e-05} {"step": 18650, "total_steps": 28610, "tokens": 39111884800, "source": "fineweb_edu", "phase": 4, "loss": 3.4623695015907288, "ce_loss": 3.4465615451335907, "logit_z_loss": 24.824488639831543, "router_aux_loss": 1.439193770289421, "router_z_loss": 1.4160285592079163, "routing_entropy_normalized": 0.6580102606446206, "learning_rate": 0.003, "grad_norm": 1.2890384197235107, "tokens_per_second": 3783910.6269240375, "eta_seconds": 5520.118200302124, "data_wait_seconds": 1.9252300262451172e-05} {"step": 18660, "total_steps": 28610, "tokens": 39132856320, "source": "finemath", "phase": 4, "loss": 2.3347834646701813, "ce_loss": 2.3192464411258698, "logit_z_loss": 31.585692405700684, "router_aux_loss": 1.4499805122613907, "router_z_loss": 1.0372218564152718, "routing_entropy_normalized": 0.6674861568247735, "learning_rate": 0.003, "grad_norm": 0.7689604759216309, "tokens_per_second": 4190004.412037513, "eta_seconds": 4980.105113983154, "data_wait_seconds": 2.187490463256836e-05} {"step": 18670, "total_steps": 28610, "tokens": 39153827840, "source": "finemath", "phase": 4, "loss": 2.3050354719161987, "ce_loss": 2.2896386981010437, "logit_z_loss": 32.83524942398071, "router_aux_loss": 1.4286230206489563, "router_z_loss": 1.1105470061302185, "routing_entropy_normalized": 0.6651157912615645, "learning_rate": 0.003, "grad_norm": 0.7996006608009338, "tokens_per_second": 3752499.5163132176, "eta_seconds": 5555.1481857299805, "data_wait_seconds": 2.104043960571289e-05} {"step": 18680, "total_steps": 28610, "tokens": 39174799360, "source": "dclm", "phase": 4, "loss": 3.4939291179180145, "ce_loss": 3.478015750646591, "logit_z_loss": 25.000756978988647, "router_aux_loss": 1.4594623446464539, "router_z_loss": 1.3187062442302704, "routing_entropy_normalized": 0.6584760269430618, "learning_rate": 0.003, "grad_norm": 2.3702220916748047, "tokens_per_second": 4351075.993097118, "eta_seconds": 4786.107940435409, "data_wait_seconds": 2.1785497665405273e-05} {"step": 18690, "total_steps": 28610, "tokens": 39195770880, "source": "fineweb_edu", "phase": 4, "loss": 3.3614606857299805, "ce_loss": 3.34639048576355, "logit_z_loss": 23.89032769203186, "router_aux_loss": 1.3491742312908173, "router_z_loss": 1.578400731086731, "routing_entropy_normalized": 0.6853883998829757, "learning_rate": 0.003, "grad_norm": 1.4392272233963013, "tokens_per_second": 3591899.894942531, "eta_seconds": 5791.850677490234, "data_wait_seconds": 2.1010637283325195e-05} {"step": 18700, "total_steps": 28610, "tokens": 39216742400, "source": "fineweb_edu", "phase": 4, "loss": 3.3796271085739136, "ce_loss": 3.363782525062561, "logit_z_loss": 24.58696722984314, "router_aux_loss": 1.4348472207784653, "router_z_loss": 1.4961179792881012, "routing_entropy_normalized": 0.6319651093781807, "learning_rate": 0.003, "grad_norm": 1.2262095212936401, "tokens_per_second": 4060838.352138826, "eta_seconds": 5117.853634595871, "data_wait_seconds": 2.104043960571289e-05} {"step": 18710, "total_steps": 28610, "tokens": 39237713920, "source": "finemath", "phase": 4, "loss": 2.279103606939316, "ce_loss": 2.2638795971870422, "logit_z_loss": 33.96958804130554, "router_aux_loss": 1.4167686104774475, "router_z_loss": 1.0563129559159279, "routing_entropy_normalized": 0.6767712887275119, "learning_rate": 0.003, "grad_norm": 0.842567503452301, "tokens_per_second": 3515946.371835537, "eta_seconds": 5905.040237903595, "data_wait_seconds": 2.2143125534057617e-05} {"step": 18720, "total_steps": 28610, "tokens": 39258685440, "source": "fineweb_edu", "phase": 4, "loss": 3.3580797016620636, "ce_loss": 3.3424023389816284, "logit_z_loss": 24.598785877227783, "router_aux_loss": 1.4292043298482895, "router_z_loss": 1.3853257298469543, "routing_entropy_normalized": 0.6431750690853414, "learning_rate": 0.003, "grad_norm": 1.326560139656067, "tokens_per_second": 4378326.746354756, "eta_seconds": 4737.159760236739, "data_wait_seconds": 2.0116567611694336e-05} {"step": 18730, "total_steps": 28610, "tokens": 39279656960, "source": "fineweb_hq", "phase": 4, "loss": 3.440432757139206, "ce_loss": 3.4253262281417847, "logit_z_loss": 23.877265453338623, "router_aux_loss": 1.3743156790733337, "router_z_loss": 1.36336749792099, "routing_entropy_normalized": 0.6741405143382083, "learning_rate": 0.003, "grad_norm": 1.4700143337249756, "tokens_per_second": 3698196.7916876045, "eta_seconds": 5602.693130493164, "data_wait_seconds": 2.008676528930664e-05} {"step": 18740, "total_steps": 28610, "tokens": 39300628480, "source": "cosmopedia_v2", "phase": 4, "loss": 2.799851179122925, "ce_loss": 2.7844081223011017, "logit_z_loss": 26.07542324066162, "router_aux_loss": 1.3766789585351944, "router_z_loss": 1.6763123869895935, "routing_entropy_normalized": 0.6828071309883863, "learning_rate": 0.003, "grad_norm": 1.3316696882247925, "tokens_per_second": 4090369.9091315987, "eta_seconds": 5060.395685434341, "data_wait_seconds": 2.148747444152832e-05} {"step": 18750, "total_steps": 28610, "tokens": 39321600000, "source": "fineweb_edu", "phase": 4, "loss": 3.3869568407535553, "ce_loss": 3.371282637119293, "logit_z_loss": 24.35986018180847, "router_aux_loss": 1.4248256534337997, "router_z_loss": 1.425984114408493, "routing_entropy_normalized": 0.6594231678090109, "learning_rate": 0.003, "grad_norm": 1.9507139921188354, "tokens_per_second": 4318365.965571951, "eta_seconds": 4788.366452693939, "data_wait_seconds": 2.193450927734375e-05} {"step": 18760, "total_steps": 28610, "tokens": 39342571520, "source": "finemath", "phase": 4, "loss": 2.3051800429821014, "ce_loss": 2.2886498272418976, "logit_z_loss": 32.61875557899475, "router_aux_loss": 1.5548225194215775, "router_z_loss": 0.9820204302668571, "routing_entropy_normalized": 0.6393168009858329, "learning_rate": 0.003, "grad_norm": 1.5219907760620117, "tokens_per_second": 3979022.963961679, "eta_seconds": 5191.462172269821, "data_wait_seconds": 2.154707908630371e-05} {"step": 18770, "total_steps": 28610, "tokens": 39363543040, "source": "finemath", "phase": 4, "loss": 2.303244858980179, "ce_loss": 2.287013530731201, "logit_z_loss": 34.613264083862305, "router_aux_loss": 1.5197744965553284, "router_z_loss": 1.0335047841072083, "routing_entropy_normalized": 0.6664794658015124, "learning_rate": 0.003, "grad_norm": 1.7950708866119385, "tokens_per_second": 4090082.1403160472, "eta_seconds": 5045.369499206543, "data_wait_seconds": 2.0712614059448242e-05} {"step": 18780, "total_steps": 28610, "tokens": 39384514560, "source": "dclm", "phase": 4, "loss": 3.503139078617096, "ce_loss": 3.487032026052475, "logit_z_loss": 24.744841814041138, "router_aux_loss": 1.4745225757360458, "router_z_loss": 1.3618616461753845, "routing_entropy_normalized": 0.6179463366858579, "learning_rate": 0.003, "grad_norm": 1.448477029800415, "tokens_per_second": 4085974.1095276605, "eta_seconds": 5045.309541225433, "data_wait_seconds": 1.9431114196777344e-05} {"step": 18790, "total_steps": 28610, "tokens": 39405486080, "source": "fineweb_edu", "phase": 4, "loss": 3.354951024055481, "ce_loss": 3.338034451007843, "logit_z_loss": 25.051867723464966, "router_aux_loss": 1.539093866944313, "router_z_loss": 1.5256637632846832, "routing_entropy_normalized": 0.6411186011991085, "learning_rate": 0.003, "grad_norm": 1.4129829406738281, "tokens_per_second": 4231680.326492141, "eta_seconds": 4866.632413387299, "data_wait_seconds": 2.5153160095214844e-05} {"step": 18800, "total_steps": 28610, "tokens": 39426457600, "source": "fineweb_edu", "phase": 4, "loss": 3.3794219195842743, "ce_loss": 3.3628668785095215, "logit_z_loss": 25.151843547821045, "router_aux_loss": 1.493138924241066, "router_z_loss": 1.6236987262964249, "routing_entropy_normalized": 0.662782530675442, "learning_rate": 0.003, "grad_norm": 2.439865827560425, "tokens_per_second": 3584100.2187141837, "eta_seconds": 5740.090919494629, "data_wait_seconds": 0.10237768292427063} {"step": 18810, "total_steps": 28610, "tokens": 39447429120, "source": "finemath", "phase": 4, "loss": 2.258251190185547, "ce_loss": 2.2429012954235077, "logit_z_loss": 34.789236545562744, "router_aux_loss": 1.4112731367349625, "router_z_loss": 1.237170249223709, "routing_entropy_normalized": 0.6819431084496003, "learning_rate": 0.003, "grad_norm": 1.025984287261963, "tokens_per_second": 4159035.672908637, "eta_seconds": 4941.551651954651, "data_wait_seconds": 2.0563602447509766e-05} {"step": 18820, "total_steps": 28610, "tokens": 39468400640, "source": "fineweb_edu", "phase": 4, "loss": 3.3387955725193024, "ce_loss": 3.3232811093330383, "logit_z_loss": 25.971417665481567, "router_aux_loss": 1.3889784216880798, "router_z_loss": 1.6246123015880585, "routing_entropy_normalized": 0.6532306084127569, "learning_rate": 0.003, "grad_norm": 1.8723586797714233, "tokens_per_second": 4202884.718618028, "eta_seconds": 4885.006240844727, "data_wait_seconds": 1.8864870071411133e-05} {"step": 18830, "total_steps": 28610, "tokens": 39489372160, "source": "fineweb_hq", "phase": 4, "loss": 3.4666574895381927, "ce_loss": 3.4500560462474823, "logit_z_loss": 25.86174702644348, "router_aux_loss": 1.507782980799675, "router_z_loss": 1.52360101044178, "routing_entropy_normalized": 0.6530623102712007, "learning_rate": 0.003, "grad_norm": 2.0254273414611816, "tokens_per_second": 4420979.902248926, "eta_seconds": 4639.276136398315, "data_wait_seconds": 1.9937753677368164e-05} {"step": 18840, "total_steps": 28610, "tokens": 39510343680, "source": "cosmopedia_v2", "phase": 4, "loss": 2.846400648355484, "ce_loss": 2.828166753053665, "logit_z_loss": 25.786588191986084, "router_aux_loss": 1.6256058365106583, "router_z_loss": 1.9778297692537308, "routing_entropy_normalized": 0.6534458996158535, "learning_rate": 0.003, "grad_norm": 2.3571200370788574, "tokens_per_second": 4114364.5711341836, "eta_seconds": 4979.912374258041, "data_wait_seconds": 1.990795135498047e-05} {"step": 18850, "total_steps": 28610, "tokens": 39531315200, "source": "fineweb_edu", "phase": 4, "loss": 3.377711683511734, "ce_loss": 3.3605275750160217, "logit_z_loss": 25.637186527252197, "router_aux_loss": 1.5305138677358627, "router_z_loss": 1.8789443373680115, "routing_entropy_normalized": 0.6479007499364886, "learning_rate": 0.003, "grad_norm": 1.2316139936447144, "tokens_per_second": 3874976.253998395, "eta_seconds": 5282.1494064331055, "data_wait_seconds": 1.7851591110229492e-05} {"step": 18860, "total_steps": 28610, "tokens": 39552286720, "source": "finemath", "phase": 4, "loss": 2.2998699843883514, "ce_loss": 2.283740758895874, "logit_z_loss": 33.230209827423096, "router_aux_loss": 1.4689910113811493, "router_z_loss": 1.4393330663442612, "routing_entropy_normalized": 0.6565516668873504, "learning_rate": 0.003, "grad_norm": 1.3978074789047241, "tokens_per_second": 2762573.288512867, "eta_seconds": 7401.516580581665, "data_wait_seconds": 2.03549861907959e-05} {"step": 18870, "total_steps": 28610, "tokens": 39573258240, "source": "finemath", "phase": 4, "loss": 2.3567430078983307, "ce_loss": 2.3407716751098633, "logit_z_loss": 32.70244359970093, "router_aux_loss": 1.4432058483362198, "router_z_loss": 1.5393008440732956, "routing_entropy_normalized": 0.649260040482871, "learning_rate": 0.003, "grad_norm": 1.185106873512268, "tokens_per_second": 3580701.4535466908, "eta_seconds": 5704.541622638702, "data_wait_seconds": 1.9431114196777344e-05} {"step": 18880, "total_steps": 28610, "tokens": 39594229760, "source": "dclm", "phase": 4, "loss": 3.47921946644783, "ce_loss": 3.46390438079834, "logit_z_loss": 25.23458981513977, "router_aux_loss": 1.3608342558145523, "router_z_loss": 1.7067611515522003, "routing_entropy_normalized": 0.6649546566671752, "learning_rate": 0.003, "grad_norm": 2.5831897258758545, "tokens_per_second": 4256764.497434161, "eta_seconds": 4793.614721298217, "data_wait_seconds": 2.5153160095214844e-05} {"step": 18890, "total_steps": 28610, "tokens": 39615201280, "source": "fineweb_edu", "phase": 4, "loss": 3.3550010919570923, "ce_loss": 3.3393868803977966, "logit_z_loss": 23.806903839111328, "router_aux_loss": 1.3697458356618881, "router_z_loss": 1.9167645573616028, "routing_entropy_normalized": 0.679602591148739, "learning_rate": 0.003, "grad_norm": 0.9894108772277832, "tokens_per_second": 4483427.103492062, "eta_seconds": 4546.592811584473, "data_wait_seconds": 2.0056962966918945e-05} {"step": 18900, "total_steps": 28610, "tokens": 39636172800, "source": "fineweb_edu", "phase": 4, "loss": 3.3595366775989532, "ce_loss": 3.3434276282787323, "logit_z_loss": 25.46197772026062, "router_aux_loss": 1.4387459456920624, "router_z_loss": 1.7215994894504547, "routing_entropy_normalized": 0.6321079730636598, "learning_rate": 0.003, "grad_norm": 1.8731322288513184, "tokens_per_second": 3461930.3356498033, "eta_seconds": 5882.078478097916, "data_wait_seconds": 2.047419548034668e-05} {"step": 18910, "total_steps": 28610, "tokens": 39657144320, "source": "finemath", "phase": 4, "loss": 2.3286288380622864, "ce_loss": 2.313012272119522, "logit_z_loss": 34.417627811431885, "router_aux_loss": 1.4264618754386902, "router_z_loss": 1.3519163727760315, "routing_entropy_normalized": 0.6635796294776841, "learning_rate": 0.003, "grad_norm": 1.2666784524917603, "tokens_per_second": 3933979.005139611, "eta_seconds": 5170.941271781921, "data_wait_seconds": 2.1159648895263672e-05} {"step": 18920, "total_steps": 28610, "tokens": 39678115840, "source": "fineweb_edu", "phase": 4, "loss": 3.340179681777954, "ce_loss": 3.3241096436977386, "logit_z_loss": 24.856947422027588, "router_aux_loss": 1.445325791835785, "router_z_loss": 1.6167036443948746, "routing_entropy_normalized": 0.64729454655384, "learning_rate": 0.003, "grad_norm": 1.4411470890045166, "tokens_per_second": 3664175.845841638, "eta_seconds": 5545.968243598938, "data_wait_seconds": 2.2232532501220703e-05} {"step": 18930, "total_steps": 28610, "tokens": 39699087360, "source": "fineweb_hq", "phase": 4, "loss": 3.447366088628769, "ce_loss": 3.432119697332382, "logit_z_loss": 23.60969614982605, "router_aux_loss": 1.3675668835639954, "router_z_loss": 1.5707240104675293, "routing_entropy_normalized": 0.6545369045670819, "learning_rate": 0.003, "grad_norm": 2.1225972175598145, "tokens_per_second": 3408140.3797277617, "eta_seconds": 5956.453989028931, "data_wait_seconds": 0.140984445810318} {"step": 18940, "total_steps": 28610, "tokens": 39720058880, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7952178716659546, "ce_loss": 2.779650092124939, "logit_z_loss": 26.69466733932495, "router_aux_loss": 1.3655253052711487, "router_z_loss": 1.9125477373600006, "routing_entropy_normalized": 0.6683848903333364, "learning_rate": 0.003, "grad_norm": 1.2361658811569214, "tokens_per_second": 3743934.5357274455, "eta_seconds": 5416.617103338242, "data_wait_seconds": 1.9490718841552734e-05} {"step": 18950, "total_steps": 28610, "tokens": 39741030400, "source": "fineweb_edu", "phase": 4, "loss": 3.317034274339676, "ce_loss": 3.301528960466385, "logit_z_loss": 24.823668479919434, "router_aux_loss": 1.4042886048555374, "router_z_loss": 1.462467759847641, "routing_entropy_normalized": 0.6637194304217766, "learning_rate": 0.003, "grad_norm": 1.6724669933319092, "tokens_per_second": 4097507.0698056705, "eta_seconds": 4944.100882530212, "data_wait_seconds": 2.2172927856445312e-05} {"step": 18960, "total_steps": 28610, "tokens": 39762001920, "source": "finemath", "phase": 4, "loss": 2.2972652912139893, "ce_loss": 2.2813287377357483, "logit_z_loss": 33.42841815948486, "router_aux_loss": 1.4842523038387299, "router_z_loss": 1.0940357446670532, "routing_entropy_normalized": 0.6428677582503904, "learning_rate": 0.003, "grad_norm": 1.326736569404602, "tokens_per_second": 4365915.988974739, "eta_seconds": 4635.34269809723, "data_wait_seconds": 2.0295381546020508e-05} {"step": 18970, "total_steps": 28610, "tokens": 39782973440, "source": "finemath", "phase": 4, "loss": 2.2890131175518036, "ce_loss": 2.274049937725067, "logit_z_loss": 34.88378190994263, "router_aux_loss": 1.391951009631157, "router_z_loss": 1.043636865913868, "routing_entropy_normalized": 0.6866247793672319, "learning_rate": 0.003, "grad_norm": 1.9986941814422607, "tokens_per_second": 3121732.998547217, "eta_seconds": 6476.064829826355, "data_wait_seconds": 2.0205974578857422e-05} {"step": 18980, "total_steps": 28610, "tokens": 39803944960, "source": "dclm", "phase": 4, "loss": 3.4976850152015686, "ce_loss": 3.4831202030181885, "logit_z_loss": 25.14836096763611, "router_aux_loss": 1.3258031606674194, "router_z_loss": 1.3067687302827835, "routing_entropy_normalized": 0.6939989835961404, "learning_rate": 0.003, "grad_norm": 1.2328888177871704, "tokens_per_second": 4228872.374744899, "eta_seconds": 4775.64040017128, "data_wait_seconds": 2.1785497665405273e-05} {"step": 18990, "total_steps": 28610, "tokens": 39824916480, "source": "fineweb_edu", "phase": 4, "loss": 3.344889372587204, "ce_loss": 3.329959452152252, "logit_z_loss": 24.71418809890747, "router_aux_loss": 1.3592123091220856, "router_z_loss": 1.3378825932741165, "routing_entropy_normalized": 0.6653372211105085, "learning_rate": 0.003, "grad_norm": 0.6324273943901062, "tokens_per_second": 3484131.4748252574, "eta_seconds": 5790.425070285797, "data_wait_seconds": 2.002716064453125e-05} {"step": 19000, "total_steps": 28610, "tokens": 39845888000, "source": "fineweb_edu", "phase": 4, "loss": 3.384854733943939, "ce_loss": 3.3697347342967987, "logit_z_loss": 25.049578189849854, "router_aux_loss": 1.376863807439804, "router_z_loss": 1.3513876348733902, "routing_entropy_normalized": 0.6717185535766416, "learning_rate": 0.003, "grad_norm": 1.2003986835479736, "tokens_per_second": 4081356.888431656, "eta_seconds": 4937.973147392273, "data_wait_seconds": 0.012684494256973267} {"step": 19010, "total_steps": 28610, "tokens": 39866859520, "source": "finemath", "phase": 4, "loss": 2.329510360956192, "ce_loss": 2.314173400402069, "logit_z_loss": 35.69715213775635, "router_aux_loss": 1.4457862675189972, "router_z_loss": 0.8790755271911621, "routing_entropy_normalized": 0.6718647136461643, "learning_rate": 0.003, "grad_norm": 0.9351880550384521, "tokens_per_second": 4022218.094562463, "eta_seconds": 5005.362396240234, "data_wait_seconds": 2.181529998779297e-05} {"step": 19020, "total_steps": 28610, "tokens": 39887831040, "source": "fineweb_edu", "phase": 4, "loss": 3.3324775099754333, "ce_loss": 3.31728133559227, "logit_z_loss": 24.564896821975708, "router_aux_loss": 1.3836414515972137, "router_z_loss": 1.3597468137741089, "routing_entropy_normalized": 0.6947559989370465, "learning_rate": 0.003, "grad_norm": 1.0407570600509644, "tokens_per_second": 3023344.2681071693, "eta_seconds": 6652.132835865021, "data_wait_seconds": 2.282857894897461e-05} {"step": 19030, "total_steps": 28610, "tokens": 39908802560, "source": "fineweb_hq", "phase": 4, "loss": 3.418479323387146, "ce_loss": 3.4031775891780853, "logit_z_loss": 25.058419466018677, "router_aux_loss": 1.3939987570047379, "router_z_loss": 1.361725926399231, "routing_entropy_normalized": 0.660185145693232, "learning_rate": 0.003, "grad_norm": 1.9674582481384277, "tokens_per_second": 4402047.752408552, "eta_seconds": 4563.947801113129, "data_wait_seconds": 1.823902130126953e-05} {"step": 19040, "total_steps": 28610, "tokens": 39929774080, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7788774967193604, "ce_loss": 2.762381464242935, "logit_z_loss": 26.323426008224487, "router_aux_loss": 1.4851464182138443, "router_z_loss": 1.6446019262075424, "routing_entropy_normalized": 0.6414736568377843, "learning_rate": 0.003, "grad_norm": 1.5522215366363525, "tokens_per_second": 3867099.5443184003, "eta_seconds": 5189.870188236237, "data_wait_seconds": 2.0802021026611328e-05} {"step": 19050, "total_steps": 28610, "tokens": 39950745600, "source": "fineweb_edu", "phase": 4, "loss": 3.3503196835517883, "ce_loss": 3.334336072206497, "logit_z_loss": 24.161132335662842, "router_aux_loss": 1.455164059996605, "router_z_loss": 1.4320161938667297, "routing_entropy_normalized": 0.652775661021223, "learning_rate": 0.003, "grad_norm": 1.2399317026138306, "tokens_per_second": 4377477.61926659, "eta_seconds": 4579.983009338379, "data_wait_seconds": 2.1636486053466797e-05} {"step": 19060, "total_steps": 28610, "tokens": 39971717120, "source": "finemath", "phase": 4, "loss": 2.3460090160369873, "ce_loss": 2.3300315737724304, "logit_z_loss": 33.342748165130615, "router_aux_loss": 1.4982696324586868, "router_z_loss": 0.9947974234819412, "routing_entropy_normalized": 0.6614683060451323, "learning_rate": 0.003, "grad_norm": 1.027050256729126, "tokens_per_second": 3687898.196222619, "eta_seconds": 5430.681796073914, "data_wait_seconds": 2.5987625122070312e-05} {"step": 19070, "total_steps": 28610, "tokens": 39992688640, "source": "finemath", "phase": 4, "loss": 2.2888855934143066, "ce_loss": 2.272148609161377, "logit_z_loss": 32.08601522445679, "router_aux_loss": 1.557641327381134, "router_z_loss": 1.1605499386787415, "routing_entropy_normalized": 0.6592582926703522, "learning_rate": 0.003, "grad_norm": 1.1490470170974731, "tokens_per_second": 4243065.366300837, "eta_seconds": 4715.183093547821, "data_wait_seconds": 2.053380012512207e-05} {"step": 19080, "total_steps": 28610, "tokens": 40013660160, "source": "dclm", "phase": 4, "loss": 3.5806706845760345, "ce_loss": 3.5641174912452698, "logit_z_loss": 24.335595846176147, "router_aux_loss": 1.5092194378376007, "router_z_loss": 1.461010679602623, "routing_entropy_normalized": 0.6414730805828496, "learning_rate": 0.003, "grad_norm": 1.764897346496582, "tokens_per_second": 4245972.1361788, "eta_seconds": 4707.015948057175, "data_wait_seconds": 2.3812055587768555e-05} {"step": 19090, "total_steps": 28610, "tokens": 40034631680, "source": "fineweb_edu", "phase": 4, "loss": 3.280580222606659, "ce_loss": 3.2649330496788025, "logit_z_loss": 24.70416808128357, "router_aux_loss": 1.3995561599731445, "router_z_loss": 1.6515910178422928, "routing_entropy_normalized": 0.664963328263446, "learning_rate": 0.003, "grad_norm": 0.8955669403076172, "tokens_per_second": 4222049.329299806, "eta_seconds": 4728.7195110321045, "data_wait_seconds": 2.238154411315918e-05} {"step": 19100, "total_steps": 28610, "tokens": 40055603200, "source": "fineweb_edu", "phase": 4, "loss": 3.314527153968811, "ce_loss": 3.2994744181632996, "logit_z_loss": 25.087427854537964, "router_aux_loss": 1.3397106528282166, "router_z_loss": 1.6556536704301834, "routing_entropy_normalized": 0.6660943869845565, "learning_rate": 0.003, "grad_norm": 1.6378049850463867, "tokens_per_second": 3905032.974097776, "eta_seconds": 5107.233575820923, "data_wait_seconds": 0.05035415291786194} {"step": 19110, "total_steps": 28610, "tokens": 40076574720, "source": "finemath", "phase": 4, "loss": 2.225951701402664, "ce_loss": 2.21087247133255, "logit_z_loss": 33.8565559387207, "router_aux_loss": 1.3980999737977982, "router_z_loss": 1.0982286632061005, "routing_entropy_normalized": 0.6658821564427896, "learning_rate": 0.003, "grad_norm": 0.9770755767822266, "tokens_per_second": 4418098.714836091, "eta_seconds": 4509.393131732941, "data_wait_seconds": 2.0056962966918945e-05} {"step": 19120, "total_steps": 28610, "tokens": 40097546240, "source": "fineweb_edu", "phase": 4, "loss": 3.360673576593399, "ce_loss": 3.345899313688278, "logit_z_loss": 24.941580057144165, "router_aux_loss": 1.3178053051233292, "router_z_loss": 1.5962343364953995, "routing_entropy_normalized": 0.6816878620825204, "learning_rate": 0.003, "grad_norm": 2.5713088512420654, "tokens_per_second": 3979513.3337432244, "eta_seconds": 5001.10712313652, "data_wait_seconds": 1.8775463104248047e-05} {"step": 19130, "total_steps": 28610, "tokens": 40118517760, "source": "fineweb_hq", "phase": 4, "loss": 3.3989872336387634, "ce_loss": 3.383926272392273, "logit_z_loss": 23.914361238479614, "router_aux_loss": 1.3575627952814102, "router_z_loss": 1.4852932691574097, "routing_entropy_normalized": 0.6558209105775537, "learning_rate": 0.003, "grad_norm": 1.9871474504470825, "tokens_per_second": 4147384.0129599194, "eta_seconds": 4793.624341964722, "data_wait_seconds": 1.8596649169921875e-05} {"step": 19140, "total_steps": 28610, "tokens": 40139489280, "source": "cosmopedia_v2", "phase": 4, "loss": 2.8086658716201782, "ce_loss": 2.7925344109535217, "logit_z_loss": 25.195834398269653, "router_aux_loss": 1.4329749941825867, "router_z_loss": 1.801727756857872, "routing_entropy_normalized": 0.6573159784414933, "learning_rate": 0.003, "grad_norm": 0.8646950125694275, "tokens_per_second": 4032190.725833871, "eta_seconds": 4925.369554758072, "data_wait_seconds": 2.1010637283325195e-05} {"step": 19150, "total_steps": 28610, "tokens": 40160460800, "source": "fineweb_edu", "phase": 4, "loss": 3.3490122854709625, "ce_loss": 3.3335174322128296, "logit_z_loss": 25.824321508407593, "router_aux_loss": 1.3954252153635025, "router_z_loss": 1.540565013885498, "routing_entropy_normalized": 0.664439873828661, "learning_rate": 0.003, "grad_norm": 1.7913305759429932, "tokens_per_second": 4082551.4275866724, "eta_seconds": 4859.475323677063, "data_wait_seconds": 1.856684684753418e-05} {"step": 19160, "total_steps": 28610, "tokens": 40181432320, "source": "finemath", "phase": 4, "loss": 2.2480796575546265, "ce_loss": 2.2325403690338135, "logit_z_loss": 34.4497504234314, "router_aux_loss": 1.4452275782823563, "router_z_loss": 1.087039902806282, "routing_entropy_normalized": 0.6682941907042286, "learning_rate": 0.003, "grad_norm": 0.9340934157371521, "tokens_per_second": 3746000.6020146, "eta_seconds": 5290.465353727341, "data_wait_seconds": 2.181529998779297e-05} {"step": 19170, "total_steps": 28610, "tokens": 40202403840, "source": "finemath", "phase": 4, "loss": 2.3108309507369995, "ce_loss": 2.2959804832935333, "logit_z_loss": 33.75853681564331, "router_aux_loss": 1.3806796818971634, "router_z_loss": 1.0437393710017204, "routing_entropy_normalized": 0.6826522427618444, "learning_rate": 0.003, "grad_norm": 1.3332397937774658, "tokens_per_second": 4086435.3811169285, "eta_seconds": 4844.592666625977, "data_wait_seconds": 1.901388168334961e-05} {"step": 19180, "total_steps": 28610, "tokens": 40223375360, "source": "dclm", "phase": 4, "loss": 3.567745476961136, "ce_loss": 3.552065849304199, "logit_z_loss": 24.829818725585938, "router_aux_loss": 1.4490021765232086, "router_z_loss": 1.1896318644285202, "routing_entropy_normalized": 0.6429833539863674, "learning_rate": 0.003, "grad_norm": 1.5485292673110962, "tokens_per_second": 3382641.7330387076, "eta_seconds": 5846.360602378845, "data_wait_seconds": 2.1517276763916016e-05} {"step": 19190, "total_steps": 28610, "tokens": 40244346880, "source": "fineweb_edu", "phase": 4, "loss": 3.349483758211136, "ce_loss": 3.334298253059387, "logit_z_loss": 24.440757036209106, "router_aux_loss": 1.379590556025505, "router_z_loss": 1.389555662870407, "routing_entropy_normalized": 0.6553825095431556, "learning_rate": 0.003, "grad_norm": 0.9894549250602722, "tokens_per_second": 4456469.940733022, "eta_seconds": 4432.919351577759, "data_wait_seconds": 2.1159648895263672e-05} {"step": 19200, "total_steps": 28610, "tokens": 40265318400, "source": "fineweb_edu", "phase": 4, "loss": 3.335856944322586, "ce_loss": 3.3204506933689117, "logit_z_loss": 25.43150496482849, "router_aux_loss": 1.4078022688627243, "router_z_loss": 1.3282230198383331, "routing_entropy_normalized": 0.6848343914578993, "learning_rate": 0.003, "grad_norm": 0.9362543225288391, "tokens_per_second": 3626585.020526724, "eta_seconds": 5441.538033246994, "data_wait_seconds": 0.10785949230194092} {"step": 19210, "total_steps": 28610, "tokens": 40286289920, "source": "finemath", "phase": 4, "loss": 2.250032216310501, "ce_loss": 2.234333097934723, "logit_z_loss": 35.44744300842285, "router_aux_loss": 1.4700603485107422, "router_z_loss": 0.9985073134303093, "routing_entropy_normalized": 0.6608717973014091, "learning_rate": 0.003, "grad_norm": 0.8628007769584656, "tokens_per_second": 4213730.965995273, "eta_seconds": 4678.331141471863, "data_wait_seconds": 2.2292137145996094e-05} {"step": 19220, "total_steps": 28610, "tokens": 40307261440, "source": "fineweb_edu", "phase": 4, "loss": 3.334888607263565, "ce_loss": 3.3201474845409393, "logit_z_loss": 25.262691259384155, "router_aux_loss": 1.3321883082389832, "router_z_loss": 1.4192203730344772, "routing_entropy_normalized": 0.6678658601817119, "learning_rate": 0.003, "grad_norm": 0.7840399742126465, "tokens_per_second": 4292727.238578189, "eta_seconds": 4587.353490114212, "data_wait_seconds": 2.378225326538086e-05} {"step": 19230, "total_steps": 28610, "tokens": 40328232960, "source": "fineweb_hq", "phase": 4, "loss": 3.414118528366089, "ce_loss": 3.3989863097667694, "logit_z_loss": 25.698143482208252, "router_aux_loss": 1.3809023648500443, "router_z_loss": 1.3231525719165802, "routing_entropy_normalized": 0.6342649275796268, "learning_rate": 0.003, "grad_norm": 1.501608967781067, "tokens_per_second": 3866156.884757921, "eta_seconds": 5088.072301864624, "data_wait_seconds": 1.8805265426635742e-05} {"step": 19240, "total_steps": 28610, "tokens": 40349204480, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7714611291885376, "ce_loss": 2.7559954822063446, "logit_z_loss": 27.04385805130005, "router_aux_loss": 1.3820387870073318, "router_z_loss": 1.6452041119337082, "routing_entropy_normalized": 0.6762785398325666, "learning_rate": 0.003, "grad_norm": 1.2110271453857422, "tokens_per_second": 3881692.0018564495, "eta_seconds": 5062.306393861771, "data_wait_seconds": 2.181529998779297e-05} {"step": 19250, "total_steps": 28610, "tokens": 40370176000, "source": "fineweb_edu", "phase": 4, "loss": 3.359909325838089, "ce_loss": 3.344887912273407, "logit_z_loss": 24.460873126983643, "router_aux_loss": 1.3754325807094574, "router_z_loss": 1.2671359777450562, "routing_entropy_normalized": 0.6784863666295575, "learning_rate": 0.003, "grad_norm": 1.1893917322158813, "tokens_per_second": 4348621.170040611, "eta_seconds": 4513.9233684539795, "data_wait_seconds": 2.1338462829589844e-05} {"step": 19260, "total_steps": 28610, "tokens": 40391147520, "source": "finemath", "phase": 4, "loss": 2.345165729522705, "ce_loss": 2.3287628889083862, "logit_z_loss": 32.336897134780884, "router_aux_loss": 1.5480506420135498, "router_z_loss": 0.9223317950963974, "routing_entropy_normalized": 0.6520826202455219, "learning_rate": 0.003, "grad_norm": 1.164008378982544, "tokens_per_second": 4005783.9256768264, "eta_seconds": 4895.0146996974945, "data_wait_seconds": 2.0056962966918945e-05} {"step": 19270, "total_steps": 28610, "tokens": 40412119040, "source": "finemath", "phase": 4, "loss": 2.278256207704544, "ce_loss": 2.2630137503147125, "logit_z_loss": 34.49786424636841, "router_aux_loss": 1.4315880835056305, "router_z_loss": 0.926568903028965, "routing_entropy_normalized": 0.6886507410474253, "learning_rate": 0.003, "grad_norm": 0.936701774597168, "tokens_per_second": 3455316.591098325, "eta_seconds": 5668.77134513855, "data_wait_seconds": 2.136826515197754e-05} {"step": 19280, "total_steps": 28610, "tokens": 40433090560, "source": "dclm", "phase": 4, "loss": 3.566343665122986, "ce_loss": 3.550455927848816, "logit_z_loss": 25.318200826644897, "router_aux_loss": 1.4700694978237152, "router_z_loss": 1.1870585530996323, "routing_entropy_normalized": 0.6353816908420516, "learning_rate": 0.003, "grad_norm": 1.6347026824951172, "tokens_per_second": 3707675.795003947, "eta_seconds": 5277.275911331177, "data_wait_seconds": 1.9550323486328125e-05} {"step": 19290, "total_steps": 28610, "tokens": 40454062080, "source": "fineweb_edu", "phase": 4, "loss": 3.3618665635585785, "ce_loss": 3.346706598997116, "logit_z_loss": 25.175074100494385, "router_aux_loss": 1.383309856057167, "router_z_loss": 1.3268421739339828, "routing_entropy_normalized": 0.6654345548757534, "learning_rate": 0.003, "grad_norm": 0.9422752261161804, "tokens_per_second": 3888037.98806682, "eta_seconds": 5027.074503898621, "data_wait_seconds": 2.4706125259399414e-05} {"step": 19300, "total_steps": 28610, "tokens": 40475033600, "source": "fineweb_edu", "phase": 4, "loss": 3.361315608024597, "ce_loss": 3.3462301194667816, "logit_z_loss": 25.468973636627197, "router_aux_loss": 1.3787935972213745, "router_z_loss": 1.2975610047578812, "routing_entropy_normalized": 0.6497927259952158, "learning_rate": 0.003, "grad_norm": 1.3930387496948242, "tokens_per_second": 4529524.567780508, "eta_seconds": 4310.493259906769, "data_wait_seconds": 2.2470951080322266e-05} {"step": 19310, "total_steps": 28610, "tokens": 40496005120, "source": "finemath", "phase": 4, "loss": 2.4010573029518127, "ce_loss": 2.3870766162872314, "logit_z_loss": 35.57442092895508, "router_aux_loss": 1.3152320683002472, "router_z_loss": 0.8283466324210167, "routing_entropy_normalized": 0.7067872152809426, "learning_rate": 0.003, "grad_norm": 1.1650546789169312, "tokens_per_second": 3820454.697147689, "eta_seconds": 5105.024178028107, "data_wait_seconds": 2.1666288375854492e-05} {"step": 19320, "total_steps": 28610, "tokens": 40516976640, "source": "fineweb_edu", "phase": 4, "loss": 3.3613320887088776, "ce_loss": 3.3467989563941956, "logit_z_loss": 24.83435583114624, "router_aux_loss": 1.3332509398460388, "router_z_loss": 1.2005935460329056, "routing_entropy_normalized": 0.6660673723818082, "learning_rate": 0.003, "grad_norm": 1.2816100120544434, "tokens_per_second": 3904126.142867136, "eta_seconds": 4990.244005203247, "data_wait_seconds": 2.3037195205688477e-05} {"step": 19330, "total_steps": 28610, "tokens": 40537948160, "source": "fineweb_hq", "phase": 4, "loss": 3.434184283018112, "ce_loss": 3.419272154569626, "logit_z_loss": 25.622697591781616, "router_aux_loss": 1.3784129172563553, "router_z_loss": 1.1280232965946198, "routing_entropy_normalized": 0.6429113828552082, "learning_rate": 0.003, "grad_norm": 2.6249959468841553, "tokens_per_second": 4287951.010535447, "eta_seconds": 4538.664390563965, "data_wait_seconds": 2.1904706954956055e-05} {"step": 19340, "total_steps": 28610, "tokens": 40558919680, "source": "cosmopedia_v2", "phase": 4, "loss": 2.802315354347229, "ce_loss": 2.785987377166748, "logit_z_loss": 26.24913191795349, "router_aux_loss": 1.484360322356224, "router_z_loss": 1.4843291342258453, "routing_entropy_normalized": 0.6296689615554232, "learning_rate": 0.003, "grad_norm": 1.1116247177124023, "tokens_per_second": 4297308.573786441, "eta_seconds": 4523.901113033295, "data_wait_seconds": 2.0951032638549805e-05} {"step": 19350, "total_steps": 28610, "tokens": 40579891200, "source": "fineweb_edu", "phase": 4, "loss": 3.3496160209178925, "ce_loss": 3.3336406648159027, "logit_z_loss": 25.224349975585938, "router_aux_loss": 1.4692447185516357, "router_z_loss": 1.282929688692093, "routing_entropy_normalized": 0.6600521993782518, "learning_rate": 0.003, "grad_norm": 1.4116278886795044, "tokens_per_second": 4051238.049680199, "eta_seconds": 4793.504425525665, "data_wait_seconds": 1.7255544662475586e-05} {"step": 19360, "total_steps": 28610, "tokens": 40600862720, "source": "finemath", "phase": 4, "loss": 2.353725254535675, "ce_loss": 2.3371587693691254, "logit_z_loss": 33.494728088378906, "router_aux_loss": 1.5736200660467148, "router_z_loss": 0.8303048685193062, "routing_entropy_normalized": 0.652967074701145, "learning_rate": 0.003, "grad_norm": 1.813033103942871, "tokens_per_second": 4297464.5677686045, "eta_seconds": 4513.976949453354, "data_wait_seconds": 2.339482307434082e-05} {"step": 19370, "total_steps": 28610, "tokens": 40621834240, "source": "finemath", "phase": 4, "loss": 2.3140942752361298, "ce_loss": 2.2980002760887146, "logit_z_loss": 35.03510522842407, "router_aux_loss": 1.5027452409267426, "router_z_loss": 1.0665128231048584, "routing_entropy_normalized": 0.6627010391892275, "learning_rate": 0.003, "grad_norm": 1.7175540924072266, "tokens_per_second": 3337291.5593815544, "eta_seconds": 5806.410418510437, "data_wait_seconds": 1.9043684005737305e-05} {"step": 19380, "total_steps": 28610, "tokens": 40642805760, "source": "dclm", "phase": 4, "loss": 3.5188450813293457, "ce_loss": 3.503953129053116, "logit_z_loss": 25.55412769317627, "router_aux_loss": 1.3527012169361115, "router_z_loss": 1.3649592697620392, "routing_entropy_normalized": 0.7022800998283955, "learning_rate": 0.003, "grad_norm": 1.1528599262237549, "tokens_per_second": 4330813.873823299, "eta_seconds": 4469.5324075222015, "data_wait_seconds": 1.9788742065429688e-05} {"step": 19390, "total_steps": 28610, "tokens": 40663777280, "source": "fineweb_edu", "phase": 4, "loss": 3.3176571428775787, "ce_loss": 3.303174316883087, "logit_z_loss": 25.297858953475952, "router_aux_loss": 1.2992235273122787, "router_z_loss": 1.4905566275119781, "routing_entropy_normalized": 0.7289033101213399, "learning_rate": 0.003, "grad_norm": 1.0415669679641724, "tokens_per_second": 3950762.5134488894, "eta_seconds": 4894.179636001587, "data_wait_seconds": 2.1219253540039062e-05} {"step": 19400, "total_steps": 28610, "tokens": 40684748800, "source": "fineweb_edu", "phase": 4, "loss": 3.360551655292511, "ce_loss": 3.345858871936798, "logit_z_loss": 24.49346399307251, "router_aux_loss": 1.3419259786605835, "router_z_loss": 1.2735259234905243, "routing_entropy_normalized": 0.6979227366650215, "learning_rate": 0.003, "grad_norm": 1.512621521949768, "tokens_per_second": 3902664.695037887, "eta_seconds": 4949.123593568802, "data_wait_seconds": 0.13360866904258728} {"step": 19410, "total_steps": 28610, "tokens": 40705720320, "source": "finemath", "phase": 4, "loss": 2.2940939366817474, "ce_loss": 2.278998762369156, "logit_z_loss": 35.681772232055664, "router_aux_loss": 1.426370233297348, "router_z_loss": 0.8314767405390739, "routing_entropy_normalized": 0.6870666685992244, "learning_rate": 0.003, "grad_norm": 1.1906874179840088, "tokens_per_second": 3932335.8477073754, "eta_seconds": 4906.447248458862, "data_wait_seconds": 1.7344951629638672e-05} {"step": 19420, "total_steps": 28610, "tokens": 40726691840, "source": "fineweb_edu", "phase": 4, "loss": 3.410654604434967, "ce_loss": 3.3953743278980255, "logit_z_loss": 25.747055530548096, "router_aux_loss": 1.401878461241722, "router_z_loss": 1.2614688128232956, "routing_entropy_normalized": 0.6757519022675323, "learning_rate": 0.003, "grad_norm": 1.2685362100601196, "tokens_per_second": 4459933.891766317, "eta_seconds": 4321.3256850242615, "data_wait_seconds": 2.193450927734375e-05} {"step": 19430, "total_steps": 28610, "tokens": 40747663360, "source": "fineweb_hq", "phase": 4, "loss": 3.44427952170372, "ce_loss": 3.4298458099365234, "logit_z_loss": 25.794349908828735, "router_aux_loss": 1.3259797394275665, "router_z_loss": 1.1738961935043335, "routing_entropy_normalized": 0.6920826827018659, "learning_rate": 0.003, "grad_norm": 1.360547661781311, "tokens_per_second": 3986094.77696957, "eta_seconds": 4829.7535400390625, "data_wait_seconds": 1.7851591110229492e-05} {"step": 19440, "total_steps": 28610, "tokens": 40768634880, "source": "cosmopedia_v2", "phase": 4, "loss": 2.8105751276016235, "ce_loss": 2.79545795917511, "logit_z_loss": 27.626615524291992, "router_aux_loss": 1.3686068654060364, "router_z_loss": 1.4311330318450928, "routing_entropy_normalized": 0.6751828602252616, "learning_rate": 0.003, "grad_norm": 1.0243067741394043, "tokens_per_second": 4408520.71547189, "eta_seconds": 4362.207888126373, "data_wait_seconds": 2.130866050720215e-05} {"step": 19450, "total_steps": 28610, "tokens": 40789606400, "source": "fineweb_edu", "phase": 4, "loss": 3.388081729412079, "ce_loss": 3.3726357519626617, "logit_z_loss": 25.2504141330719, "router_aux_loss": 1.4285918772220612, "router_z_loss": 1.1600735634565353, "routing_entropy_normalized": 0.6504429699335179, "learning_rate": 0.003, "grad_norm": 1.3529040813446045, "tokens_per_second": 3945432.9527310645, "eta_seconds": 4868.898432731628, "data_wait_seconds": 2.1696090698242188e-05} {"step": 19460, "total_steps": 28610, "tokens": 40810577920, "source": "finemath", "phase": 4, "loss": 2.312614530324936, "ce_loss": 2.2962711453437805, "logit_z_loss": 34.46598815917969, "router_aux_loss": 1.5565002411603928, "router_z_loss": 0.7783603817224503, "routing_entropy_normalized": 0.6631029899227223, "learning_rate": 0.003, "grad_norm": 1.5711826086044312, "tokens_per_second": 3308181.2503882726, "eta_seconds": 5800.450261831284, "data_wait_seconds": 2.300739288330078e-05} {"step": 19470, "total_steps": 28610, "tokens": 40831549440, "source": "finemath", "phase": 4, "loss": 2.4168583154678345, "ce_loss": 2.401138514280319, "logit_z_loss": 34.750178813934326, "router_aux_loss": 1.4821507632732391, "router_z_loss": 0.8983106836676598, "routing_entropy_normalized": 0.6593128347536282, "learning_rate": 0.003, "grad_norm": 1.0271867513656616, "tokens_per_second": 4313258.841098057, "eta_seconds": 4443.9645256996155, "data_wait_seconds": 2.434849739074707e-05} {"step": 19480, "total_steps": 28610, "tokens": 40852520960, "source": "dclm", "phase": 4, "loss": 3.499381333589554, "ce_loss": 3.4840393662452698, "logit_z_loss": 24.884636402130127, "router_aux_loss": 1.4184267222881317, "router_z_loss": 1.1576831489801407, "routing_entropy_normalized": 0.6667209275913211, "learning_rate": 0.003, "grad_norm": 1.136165738105774, "tokens_per_second": 3833904.688728675, "eta_seconds": 4994.124610424042, "data_wait_seconds": 2.1576881408691406e-05} {"step": 19490, "total_steps": 28610, "tokens": 40873492480, "source": "fineweb_edu", "phase": 4, "loss": 3.406808614730835, "ce_loss": 3.391199380159378, "logit_z_loss": 24.399075031280518, "router_aux_loss": 1.4300537109375, "router_z_loss": 1.3086664229631424, "routing_entropy_normalized": 0.6568177875575119, "learning_rate": 0.003, "grad_norm": 1.672478437423706, "tokens_per_second": 3619167.321440459, "eta_seconds": 5284.648246765137, "data_wait_seconds": 2.0831823348999023e-05} {"step": 19500, "total_steps": 28610, "tokens": 40894464000, "source": "fineweb_edu", "phase": 4, "loss": 3.3766796588897705, "ce_loss": 3.360709458589554, "logit_z_loss": 25.300734043121338, "router_aux_loss": 1.4593832194805145, "router_z_loss": 1.3763417899608612, "routing_entropy_normalized": 0.6641136762916201, "learning_rate": 0.003, "grad_norm": 1.5944690704345703, "tokens_per_second": 3437629.2701454046, "eta_seconds": 5557.6251010894775, "data_wait_seconds": 1.8477439880371094e-05} {"step": 19510, "total_steps": 28610, "tokens": 40915435520, "source": "finemath", "phase": 4, "loss": 2.3851784467697144, "ce_loss": 2.3690296709537506, "logit_z_loss": 32.88900136947632, "router_aux_loss": 1.5214926898479462, "router_z_loss": 0.933828555047512, "routing_entropy_normalized": 0.6528694150728274, "learning_rate": 0.003, "grad_norm": 1.717016577720642, "tokens_per_second": 3719356.7095593265, "eta_seconds": 5131.017186641693, "data_wait_seconds": 2.0742416381835938e-05} {"step": 19520, "total_steps": 28610, "tokens": 40936407040, "source": "fineweb_edu", "phase": 4, "loss": 3.405904918909073, "ce_loss": 3.3904313147068024, "logit_z_loss": 24.61697292327881, "router_aux_loss": 1.3990053236484528, "router_z_loss": 1.4835718870162964, "routing_entropy_normalized": 0.6606179926250161, "learning_rate": 0.003, "grad_norm": 2.442452907562256, "tokens_per_second": 3505356.620119562, "eta_seconds": 5438.280251026154, "data_wait_seconds": 1.9371509552001953e-05} {"step": 19530, "total_steps": 28610, "tokens": 40957378560, "source": "fineweb_hq", "phase": 4, "loss": 3.400954931974411, "ce_loss": 3.385393112897873, "logit_z_loss": 25.92687678337097, "router_aux_loss": 1.409311905503273, "router_z_loss": 1.4686855375766754, "routing_entropy_normalized": 0.6839034516161941, "learning_rate": 0.003, "grad_norm": 1.8459012508392334, "tokens_per_second": 4060803.4822305026, "eta_seconds": 4689.254292488098, "data_wait_seconds": 2.282857894897461e-05} {"step": 19540, "total_steps": 28610, "tokens": 40978350080, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7627648413181305, "ce_loss": 2.74604132771492, "logit_z_loss": 27.46313977241516, "router_aux_loss": 1.4837608486413956, "router_z_loss": 1.8859188705682755, "routing_entropy_normalized": 0.6724948141376468, "learning_rate": 0.003, "grad_norm": 1.9242206811904907, "tokens_per_second": 4363747.889070693, "eta_seconds": 4358.906408786774, "data_wait_seconds": 2.041459083557129e-05} {"step": 19550, "total_steps": 28610, "tokens": 40999321600, "source": "fineweb_edu", "phase": 4, "loss": 3.3830857276916504, "ce_loss": 3.36690491437912, "logit_z_loss": 25.855246782302856, "router_aux_loss": 1.4583090096712112, "router_z_loss": 1.5977157056331635, "routing_entropy_normalized": 0.6592382518325778, "learning_rate": 0.003, "grad_norm": 2.120476722717285, "tokens_per_second": 3946779.4455228373, "eta_seconds": 4814.101568698883, "data_wait_seconds": 2.0384788513183594e-05} {"step": 19560, "total_steps": 28610, "tokens": 41020293120, "source": "finemath", "phase": 4, "loss": 2.282406270503998, "ce_loss": 2.2666259109973907, "logit_z_loss": 33.995994091033936, "router_aux_loss": 1.462789624929428, "router_z_loss": 1.1524617075920105, "routing_entropy_normalized": 0.6615245281832378, "learning_rate": 0.003, "grad_norm": 0.6747452020645142, "tokens_per_second": 3704613.2907315735, "eta_seconds": 5123.1327295303345, "data_wait_seconds": 1.817941665649414e-05} {"step": 19570, "total_steps": 28610, "tokens": 41041264640, "source": "finemath", "phase": 4, "loss": 2.78495654463768, "ce_loss": 2.768691301345825, "logit_z_loss": 34.04947805404663, "router_aux_loss": 1.4952046275138855, "router_z_loss": 1.3131720572710037, "routing_entropy_normalized": 0.6638012818756749, "learning_rate": 0.003, "grad_norm": 3.607661008834839, "tokens_per_second": 2895091.664329981, "eta_seconds": 6548.41237449646, "data_wait_seconds": 2.8967857360839844e-05} {"step": 19580, "total_steps": 28610, "tokens": 41062236160, "source": "dclm", "phase": 4, "loss": 3.551274061203003, "ce_loss": 3.5349248945713043, "logit_z_loss": 25.951492309570312, "router_aux_loss": 1.4914551228284836, "router_z_loss": 1.434581458568573, "routing_entropy_normalized": 0.6433852164403409, "learning_rate": 0.003, "grad_norm": 1.6098854541778564, "tokens_per_second": 4500990.925046194, "eta_seconds": 4207.3585295677185, "data_wait_seconds": 2.378225326538086e-05} {"step": 19590, "total_steps": 28610, "tokens": 41083207680, "source": "fineweb_edu", "phase": 4, "loss": 3.4216714203357697, "ce_loss": 3.4055334627628326, "logit_z_loss": 24.058673620224, "router_aux_loss": 1.4515281170606613, "router_z_loss": 1.6227427572011948, "routing_entropy_normalized": 0.6712929042750749, "learning_rate": 0.003, "grad_norm": 1.6493701934814453, "tokens_per_second": 4173391.193786021, "eta_seconds": 4532.599548339844, "data_wait_seconds": 2.0891427993774414e-05} {"step": 19600, "total_steps": 28610, "tokens": 41104179200, "source": "fineweb_edu", "phase": 4, "loss": 3.3789327442646027, "ce_loss": 3.3628121614456177, "logit_z_loss": 24.865243196487427, "router_aux_loss": 1.4434575587511063, "router_z_loss": 1.685979649424553, "routing_entropy_normalized": 0.6640509029978018, "learning_rate": 0.003, "grad_norm": 1.2236896753311157, "tokens_per_second": 3497842.193873051, "eta_seconds": 5401.9988532066345, "data_wait_seconds": 0.0728498101234436} {"step": 19610, "total_steps": 28610, "tokens": 41125150720, "source": "finemath", "phase": 4, "loss": 2.2894572019577026, "ce_loss": 2.2736768424510956, "logit_z_loss": 37.05294179916382, "router_aux_loss": 1.4641620367765427, "router_z_loss": 1.1387911587953568, "routing_entropy_normalized": 0.6642668238329416, "learning_rate": 0.003, "grad_norm": 1.1619415283203125, "tokens_per_second": 4044249.684538527, "eta_seconds": 4666.964077949524, "data_wait_seconds": 2.1845102310180664e-05} {"step": 19620, "total_steps": 28610, "tokens": 41146122240, "source": "fineweb_edu", "phase": 4, "loss": 3.393438935279846, "ce_loss": 3.377029061317444, "logit_z_loss": 25.27425765991211, "router_aux_loss": 1.4830830842256546, "router_z_loss": 1.579040288925171, "routing_entropy_normalized": 0.6672817084726942, "learning_rate": 0.003, "grad_norm": 1.6574522256851196, "tokens_per_second": 3840577.138070257, "eta_seconds": 4909.000861644745, "data_wait_seconds": 2.092123031616211e-05} {"step": 19630, "total_steps": 28610, "tokens": 41167093760, "source": "fineweb_hq", "phase": 4, "loss": 3.3481086790561676, "ce_loss": 3.332891523838043, "logit_z_loss": 25.154505491256714, "router_aux_loss": 1.3674375265836716, "router_z_loss": 1.5427638590335846, "routing_entropy_normalized": 0.672560380738948, "learning_rate": 0.003, "grad_norm": 2.013730764389038, "tokens_per_second": 4290166.415438036, "eta_seconds": 4389.672366142273, "data_wait_seconds": 1.9103288650512695e-05} {"step": 19640, "total_steps": 28610, "tokens": 41188065280, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7803442776203156, "ce_loss": 2.764664202928543, "logit_z_loss": 27.57576847076416, "router_aux_loss": 1.3762193322181702, "router_z_loss": 1.9178211390972137, "routing_entropy_normalized": 0.6788237438952476, "learning_rate": 0.003, "grad_norm": 1.5583423376083374, "tokens_per_second": 3710672.7675601533, "eta_seconds": 5069.553317785263, "data_wait_seconds": 2.2023916244506836e-05} {"step": 19650, "total_steps": 28610, "tokens": 41209036800, "source": "fineweb_edu", "phase": 4, "loss": 3.3814262747764587, "ce_loss": 3.3661673963069916, "logit_z_loss": 24.61990237236023, "router_aux_loss": 1.3750763535499573, "router_z_loss": 1.5081141740083694, "routing_entropy_normalized": 0.6855441247015234, "learning_rate": 0.003, "grad_norm": 1.0763064622879028, "tokens_per_second": 4231853.987670982, "eta_seconds": 4440.248168945313, "data_wait_seconds": 1.84476375579834e-05} {"step": 19660, "total_steps": 28610, "tokens": 41230008320, "source": "finemath", "phase": 4, "loss": 2.2359384447336197, "ce_loss": 2.2206709682941437, "logit_z_loss": 34.49990177154541, "router_aux_loss": 1.4287006855010986, "router_z_loss": 0.9805193990468979, "routing_entropy_normalized": 0.6733017959049757, "learning_rate": 0.003, "grad_norm": 1.537093162536621, "tokens_per_second": 4051953.3710633335, "eta_seconds": 4632.2128319740295, "data_wait_seconds": 2.321600914001465e-05} {"step": 19670, "total_steps": 28610, "tokens": 41250979840, "source": "finemath", "phase": 4, "loss": 2.2221187204122543, "ce_loss": 2.20647469162941, "logit_z_loss": 35.02656698226929, "router_aux_loss": 1.463929682970047, "router_z_loss": 1.004736289381981, "routing_entropy_normalized": 0.6631722779047203, "learning_rate": 0.003, "grad_norm": 1.3147599697113037, "tokens_per_second": 4213890.237180683, "eta_seconds": 4449.223360061645, "data_wait_seconds": 1.952052116394043e-05} {"step": 19680, "total_steps": 28610, "tokens": 41271951360, "source": "dclm", "phase": 4, "loss": 3.492144078016281, "ce_loss": 3.4766214191913605, "logit_z_loss": 25.514819145202637, "router_aux_loss": 1.4167003482580185, "router_z_loss": 1.3555648475885391, "routing_entropy_normalized": 0.6421444392460918, "learning_rate": 0.003, "grad_norm": 1.4331028461456299, "tokens_per_second": 3988634.517116643, "eta_seconds": 4695.232736825943, "data_wait_seconds": 1.963973045349121e-05} {"step": 19690, "total_steps": 28610, "tokens": 41292922880, "source": "fineweb_edu", "phase": 4, "loss": 3.3779505491256714, "ce_loss": 3.3629758656024933, "logit_z_loss": 25.266141176223755, "router_aux_loss": 1.3424449563026428, "router_z_loss": 1.5502213686704636, "routing_entropy_normalized": 0.6877329627399393, "learning_rate": 0.003, "grad_norm": 1.0395476818084717, "tokens_per_second": 3977933.206164429, "eta_seconds": 4702.591740608215, "data_wait_seconds": 1.9043684005737305e-05} {"step": 19700, "total_steps": 28610, "tokens": 41313894400, "source": "fineweb_edu", "phase": 4, "loss": 3.386564075946808, "ce_loss": 3.3718760311603546, "logit_z_loss": 25.601024627685547, "router_aux_loss": 1.3315831571817398, "router_z_loss": 1.3721712678670883, "routing_entropy_normalized": 0.6624809204645785, "learning_rate": 0.003, "grad_norm": 1.3177599906921387, "tokens_per_second": 3376653.6818410507, "eta_seconds": 5533.769844532013, "data_wait_seconds": 2.6673078536987305e-05} {"step": 19710, "total_steps": 28610, "tokens": 41334865920, "source": "finemath", "phase": 4, "loss": 2.2743792831897736, "ce_loss": 2.2592556476593018, "logit_z_loss": 35.33148813247681, "router_aux_loss": 1.4180635958909988, "router_z_loss": 0.943014495074749, "routing_entropy_normalized": 0.6786611241695513, "learning_rate": 0.003, "grad_norm": 0.7516670823097229, "tokens_per_second": 4052635.89580505, "eta_seconds": 4605.558772087097, "data_wait_seconds": 2.1904706954956055e-05} {"step": 19720, "total_steps": 28610, "tokens": 41355837440, "source": "fineweb_edu", "phase": 4, "loss": 3.41230645775795, "ce_loss": 3.3964224755764008, "logit_z_loss": 25.842294216156006, "router_aux_loss": 1.4515931457281113, "router_z_loss": 1.3679829388856888, "routing_entropy_normalized": 0.6617833598062043, "learning_rate": 0.003, "grad_norm": 1.1695643663406372, "tokens_per_second": 4252362.897050639, "eta_seconds": 4384.3109657764435, "data_wait_seconds": 2.059340476989746e-05} {"step": 19730, "total_steps": 28610, "tokens": 41376808960, "source": "fineweb_hq", "phase": 4, "loss": 3.359740048646927, "ce_loss": 3.344673991203308, "logit_z_loss": 25.219987869262695, "router_aux_loss": 1.3760186433792114, "router_z_loss": 1.305805817246437, "routing_entropy_normalized": 0.6524073908249266, "learning_rate": 0.003, "grad_norm": 0.9823104739189148, "tokens_per_second": 4027811.119914261, "eta_seconds": 4623.531045913696, "data_wait_seconds": 1.7702579498291016e-05} {"step": 19740, "total_steps": 28610, "tokens": 41397780480, "source": "cosmopedia_v2", "phase": 4, "loss": 2.791685551404953, "ce_loss": 2.7757598161697388, "logit_z_loss": 26.5675687789917, "router_aux_loss": 1.4235591739416122, "router_z_loss": 1.6901453584432602, "routing_entropy_normalized": 0.658584994383214, "learning_rate": 0.003, "grad_norm": 2.4466934204101562, "tokens_per_second": 4223075.822473487, "eta_seconds": 4404.784337759019, "data_wait_seconds": 2.2083520889282227e-05} {"step": 19750, "total_steps": 28610, "tokens": 41418752000, "source": "fineweb_edu", "phase": 4, "loss": 3.3596458733081818, "ce_loss": 3.3428642749786377, "logit_z_loss": 25.438061714172363, "router_aux_loss": 1.5384917259216309, "router_z_loss": 1.3967131525278091, "routing_entropy_normalized": 0.6776969252210449, "learning_rate": 0.003, "grad_norm": 1.7672075033187866, "tokens_per_second": 3998887.00740891, "eta_seconds": 4646.484555721283, "data_wait_seconds": 1.8388032913208008e-05} {"step": 19760, "total_steps": 28610, "tokens": 41439723520, "source": "finemath", "phase": 4, "loss": 2.2115378081798553, "ce_loss": 2.194807082414627, "logit_z_loss": 33.342501163482666, "router_aux_loss": 1.5546281933784485, "router_z_loss": 1.1844507604837418, "routing_entropy_normalized": 0.6904869921848563, "learning_rate": 0.003, "grad_norm": 1.2662183046340942, "tokens_per_second": 4191386.031437922, "eta_seconds": 4428.080606460571, "data_wait_seconds": 2.059340476989746e-05} {"step": 19770, "total_steps": 28610, "tokens": 41460695040, "source": "finemath", "phase": 4, "loss": 2.287915527820587, "ce_loss": 2.2716364562511444, "logit_z_loss": 34.344773292541504, "router_aux_loss": 1.5111863017082214, "router_z_loss": 1.1671987026929855, "routing_entropy_normalized": 0.6639887641080999, "learning_rate": 0.003, "grad_norm": 1.4222440719604492, "tokens_per_second": 3856337.3548427233, "eta_seconds": 4807.365635871887, "data_wait_seconds": 2.2619962692260742e-05} {"step": 19780, "total_steps": 28610, "tokens": 41481666560, "source": "dclm", "phase": 4, "loss": 3.4877491891384125, "ce_loss": 3.471923738718033, "logit_z_loss": 25.190669298171997, "router_aux_loss": 1.4351905435323715, "router_z_loss": 1.473504513502121, "routing_entropy_normalized": 0.6565859216170745, "learning_rate": 0.003, "grad_norm": 0.7876819372177124, "tokens_per_second": 4278584.162697328, "eta_seconds": 4328.032698631287, "data_wait_seconds": 1.9282102584838867e-05} {"step": 19790, "total_steps": 28610, "tokens": 41502638080, "source": "fineweb_edu", "phase": 4, "loss": 3.3945550620555878, "ce_loss": 3.378721982240677, "logit_z_loss": 25.259116888046265, "router_aux_loss": 1.4246941655874252, "router_z_loss": 1.5861477851867676, "routing_entropy_normalized": 0.662727546335773, "learning_rate": 0.003, "grad_norm": 0.9198733568191528, "tokens_per_second": 3737147.328383436, "eta_seconds": 4949.465197563171, "data_wait_seconds": 2.047419548034668e-05} {"step": 19800, "total_steps": 28610, "tokens": 41523609600, "source": "fineweb_edu", "phase": 4, "loss": 3.3980186581611633, "ce_loss": 3.3829903602600098, "logit_z_loss": 25.864025354385376, "router_aux_loss": 1.3364901542663574, "router_z_loss": 1.6634015142917633, "routing_entropy_normalized": 0.6716445693556421, "learning_rate": 0.003, "grad_norm": 1.4075148105621338, "tokens_per_second": 4187810.0699713007, "eta_seconds": 4411.830720901489, "data_wait_seconds": 1.9341707229614258e-05} {"step": 19810, "total_steps": 28610, "tokens": 41544581120, "source": "finemath", "phase": 4, "loss": 2.3677909076213837, "ce_loss": 2.3522593080997467, "logit_z_loss": 37.780941009521484, "router_aux_loss": 1.4570841938257217, "router_z_loss": 0.9607251212000847, "routing_entropy_normalized": 0.6635052665255765, "learning_rate": 0.003, "grad_norm": 1.189507246017456, "tokens_per_second": 3496920.378043468, "eta_seconds": 5277.482929229736, "data_wait_seconds": 2.130866050720215e-05} {"step": 19820, "total_steps": 28610, "tokens": 41565552640, "source": "fineweb_edu", "phase": 4, "loss": 3.363358736038208, "ce_loss": 3.3474861085414886, "logit_z_loss": 25.78414034843445, "router_aux_loss": 1.4370632022619247, "router_z_loss": 1.5019941926002502, "routing_entropy_normalized": 0.6487842979284906, "learning_rate": 0.003, "grad_norm": 1.3343894481658936, "tokens_per_second": 4315833.762184766, "eta_seconds": 4271.240992069244, "data_wait_seconds": 1.9669532775878906e-05} {"step": 19830, "total_steps": 28610, "tokens": 41586524160, "source": "fineweb_hq", "phase": 4, "loss": 3.343070536851883, "ce_loss": 3.3269182741642, "logit_z_loss": 26.106512784957886, "router_aux_loss": 1.474971428513527, "router_z_loss": 1.4024819433689117, "routing_entropy_normalized": 0.6370669829041516, "learning_rate": 0.003, "grad_norm": 2.0362696647644043, "tokens_per_second": 1119692.3182979187, "eta_seconds": 16444.69133090973, "data_wait_seconds": 1.5748256742954254} {"step": 19840, "total_steps": 28610, "tokens": 41607495680, "source": "cosmopedia_v2", "phase": 4, "loss": 2.762388586997986, "ce_loss": 2.7465842962265015, "logit_z_loss": 27.225879430770874, "router_aux_loss": 1.4067123234272003, "router_z_loss": 1.7371180355548859, "routing_entropy_normalized": 0.663961498704434, "learning_rate": 0.003, "grad_norm": 0.7857089638710022, "tokens_per_second": 881638.0130033817, "eta_seconds": 20861.19560265541, "data_wait_seconds": 1.9669532775878906e-05} {"step": 19850, "total_steps": 28610, "tokens": 41628467200, "source": "fineweb_edu", "phase": 4, "loss": 3.400058329105377, "ce_loss": 3.3837652504444122, "logit_z_loss": 25.630204677581787, "router_aux_loss": 1.4789058417081833, "router_z_loss": 1.5040385872125626, "routing_entropy_normalized": 0.6504266223016056, "learning_rate": 0.003, "grad_norm": 1.4640371799468994, "tokens_per_second": 4176041.4651502674, "eta_seconds": 4399.154479980469, "data_wait_seconds": 2.2083520889282227e-05} {"step": 19860, "total_steps": 28610, "tokens": 41649438720, "source": "finemath", "phase": 4, "loss": 2.2822340726852417, "ce_loss": 2.265569806098938, "logit_z_loss": 35.455748558044434, "router_aux_loss": 1.5674724727869034, "router_z_loss": 0.9894799143075943, "routing_entropy_normalized": 0.6422052018459211, "learning_rate": 0.003, "grad_norm": 1.5515333414077759, "tokens_per_second": 4033936.1712058517, "eta_seconds": 4548.9267110824585, "data_wait_seconds": 2.154707908630371e-05} {"step": 19870, "total_steps": 28610, "tokens": 41670410240, "source": "finemath", "phase": 4, "loss": 2.306694209575653, "ce_loss": 2.2914266288280487, "logit_z_loss": 35.438976764678955, "router_aux_loss": 1.4158423691987991, "router_z_loss": 1.1091444939374924, "routing_entropy_normalized": 0.6673629389756317, "learning_rate": 0.003, "grad_norm": 0.921225368976593, "tokens_per_second": 3727098.837288814, "eta_seconds": 4917.795121669769, "data_wait_seconds": 0.0956384539604187} {"step": 19880, "total_steps": 28610, "tokens": 41691381760, "source": "dclm", "phase": 4, "loss": 3.440915673971176, "ce_loss": 3.4254438281059265, "logit_z_loss": 26.619064569473267, "router_aux_loss": 1.409695029258728, "router_z_loss": 1.3748666793107986, "routing_entropy_normalized": 0.6494054586860035, "learning_rate": 0.003, "grad_norm": 1.3489078283309937, "tokens_per_second": 4422603.903800634, "eta_seconds": 4139.6736760139465, "data_wait_seconds": 2.4497509002685547e-05} {"step": 19890, "total_steps": 28610, "tokens": 41712353280, "source": "fineweb_edu", "phase": 4, "loss": 3.3885555267333984, "ce_loss": 3.3730591237545013, "logit_z_loss": 25.252106189727783, "router_aux_loss": 1.3981886208057404, "router_z_loss": 1.5145142525434494, "routing_entropy_normalized": 0.6667743074752652, "learning_rate": 0.003, "grad_norm": 1.17805814743042, "tokens_per_second": 3827392.773852692, "eta_seconds": 4777.969369888306, "data_wait_seconds": 2.1755695343017578e-05} {"step": 19900, "total_steps": 28610, "tokens": 41733324800, "source": "fineweb_edu", "phase": 4, "loss": 3.3706478476524353, "ce_loss": 3.355399250984192, "logit_z_loss": 25.715630531311035, "router_aux_loss": 1.3795608282089233, "router_z_loss": 1.4529721587896347, "routing_entropy_normalized": 0.6597963709533221, "learning_rate": 0.003, "grad_norm": 1.5000642538070679, "tokens_per_second": 3926292.9857615805, "eta_seconds": 4652.2748012542725, "data_wait_seconds": 2.0891427993774414e-05} {"step": 19910, "total_steps": 28610, "tokens": 41754296320, "source": "finemath", "phase": 4, "loss": 2.328277140855789, "ce_loss": 2.312731295824051, "logit_z_loss": 35.027015686035156, "router_aux_loss": 1.4643206000328064, "router_z_loss": 0.9026738703250885, "routing_entropy_normalized": 0.6248347519017086, "learning_rate": 0.003, "grad_norm": 1.95051908493042, "tokens_per_second": 3348634.7708674446, "eta_seconds": 5448.555500507355, "data_wait_seconds": 2.1219253540039062e-05} {"step": 19920, "total_steps": 28610, "tokens": 41775267840, "source": "fineweb_edu", "phase": 4, "loss": 3.3930442333221436, "ce_loss": 3.3775411248207092, "logit_z_loss": 24.949378967285156, "router_aux_loss": 1.4102495312690735, "router_z_loss": 1.4007006585597992, "routing_entropy_normalized": 0.6556911482785873, "learning_rate": 0.003, "grad_norm": 1.649217128753662, "tokens_per_second": 4157159.488809013, "eta_seconds": 4383.822879314423, "data_wait_seconds": 2.3066997528076172e-05} {"step": 19930, "total_steps": 28610, "tokens": 41796239360, "source": "fineweb_hq", "phase": 4, "loss": 3.2915078699588776, "ce_loss": 3.2763228118419647, "logit_z_loss": 25.632850408554077, "router_aux_loss": 1.3813044726848602, "router_z_loss": 1.372016429901123, "routing_entropy_normalized": 0.6648254975875345, "learning_rate": 0.003, "grad_norm": 1.3679286241531372, "tokens_per_second": 4292609.714258497, "eta_seconds": 4240.608993530273, "data_wait_seconds": 2.0295381546020508e-05} {"step": 19940, "total_steps": 28610, "tokens": 41817210880, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7756205201148987, "ce_loss": 2.7600366175174713, "logit_z_loss": 27.433541297912598, "router_aux_loss": 1.3934691101312637, "router_z_loss": 1.649168312549591, "routing_entropy_normalized": 0.6597872303850468, "learning_rate": 0.003, "grad_norm": 1.6369487047195435, "tokens_per_second": 3470867.5803475054, "eta_seconds": 5238.548408746719, "data_wait_seconds": 2.2590160369873047e-05} {"step": 19950, "total_steps": 28610, "tokens": 41838182400, "source": "fineweb_edu", "phase": 4, "loss": 3.4087560176849365, "ce_loss": 3.3938592672348022, "logit_z_loss": 25.090859413146973, "router_aux_loss": 1.3559703528881073, "router_z_loss": 1.3370073735713959, "routing_entropy_normalized": 0.6677421464358181, "learning_rate": 0.003, "grad_norm": 1.862627387046814, "tokens_per_second": 4096910.4806618183, "eta_seconds": 4432.934623718262, "data_wait_seconds": 2.2351741790771484e-05} {"step": 19960, "total_steps": 28610, "tokens": 41859153920, "source": "finemath", "phase": 4, "loss": 2.301615297794342, "ce_loss": 2.286431461572647, "logit_z_loss": 34.37750482559204, "router_aux_loss": 1.421158328652382, "router_z_loss": 0.9722788482904434, "routing_entropy_normalized": 0.6605268072820731, "learning_rate": 0.003, "grad_norm": 0.7242783308029175, "tokens_per_second": 3947944.338077983, "eta_seconds": 4594.8886930942535, "data_wait_seconds": 2.3931264877319336e-05} {"step": 19970, "total_steps": 28610, "tokens": 41880125440, "source": "finemath", "phase": 4, "loss": 2.2554591298103333, "ce_loss": 2.2393909692764282, "logit_z_loss": 35.07967662811279, "router_aux_loss": 1.502998948097229, "router_z_loss": 1.038179099559784, "routing_entropy_normalized": 0.6454375757089151, "learning_rate": 0.003, "grad_norm": 1.5852179527282715, "tokens_per_second": 3312236.9451225833, "eta_seconds": 5470.43994140625, "data_wait_seconds": 2.1576881408691406e-05} {"step": 19980, "total_steps": 28610, "tokens": 41901096960, "source": "dclm", "phase": 4, "loss": 3.4492729008197784, "ce_loss": 3.433886617422104, "logit_z_loss": 26.637153387069702, "router_aux_loss": 1.4185898751020432, "router_z_loss": 1.2004885226488113, "routing_entropy_normalized": 0.6637150072718381, "learning_rate": 0.003, "grad_norm": 1.5042582750320435, "tokens_per_second": 4155253.0006495467, "eta_seconds": 4355.552299022675, "data_wait_seconds": 1.7851591110229492e-05} {"step": 19990, "total_steps": 28610, "tokens": 41922068480, "source": "fineweb_edu", "phase": 4, "loss": 3.4101691842079163, "ce_loss": 3.395140141248703, "logit_z_loss": 25.897629976272583, "router_aux_loss": 1.3601452857255936, "router_z_loss": 1.4276083260774612, "routing_entropy_normalized": 0.6748207406931678, "learning_rate": 0.003, "grad_norm": 1.857651710510254, "tokens_per_second": 3757844.8441781853, "eta_seconds": 4810.589843273163, "data_wait_seconds": 2.0563602447509766e-05} {"step": 20000, "total_steps": 28610, "tokens": 41943040000, "source": "fineweb_edu", "phase": 4, "loss": 3.4219286739826202, "ce_loss": 3.406473696231842, "logit_z_loss": 25.23620295524597, "router_aux_loss": 1.4066206216812134, "router_z_loss": 1.388781651854515, "routing_entropy_normalized": 0.6646993457299903, "learning_rate": 0.003, "grad_norm": 2.0436270236968994, "tokens_per_second": 3809545.618266777, "eta_seconds": 4739.798529624939, "data_wait_seconds": 2.181529998779297e-05} {"step": 20010, "total_steps": 28610, "tokens": 41964011520, "source": "finemath", "phase": 4, "loss": 2.2998195588588715, "ce_loss": 2.2842917442321777, "logit_z_loss": 35.659669399261475, "router_aux_loss": 1.465424820780754, "router_z_loss": 0.8735781535506248, "routing_entropy_normalized": 0.6538621197265494, "learning_rate": 0.003, "grad_norm": 1.3413957357406616, "tokens_per_second": 3998643.049741348, "eta_seconds": 4510.4068994522095, "data_wait_seconds": 2.0444393157958984e-05} {"step": 20020, "total_steps": 28610, "tokens": 41984983040, "source": "fineweb_edu", "phase": 4, "loss": 3.4152290523052216, "ce_loss": 3.398749500513077, "logit_z_loss": 24.79096555709839, "router_aux_loss": 1.503540188074112, "router_z_loss": 1.4441654682159424, "routing_entropy_normalized": 0.6696691601172644, "learning_rate": 0.003, "grad_norm": 3.7588818073272705, "tokens_per_second": 4088508.591986171, "eta_seconds": 4406.13863825798, "data_wait_seconds": 2.4318695068359375e-05} {"step": 20030, "total_steps": 28610, "tokens": 42005954560, "source": "fineweb_hq", "phase": 4, "loss": 3.2903234362602234, "ce_loss": 3.2730169594287872, "logit_z_loss": 26.27968692779541, "router_aux_loss": 1.578488528728485, "router_z_loss": 1.521615132689476, "routing_entropy_normalized": 0.6578955157170604, "learning_rate": 0.003, "grad_norm": 2.5993294715881348, "tokens_per_second": 4045825.0719869593, "eta_seconds": 4447.439975738525, "data_wait_seconds": 2.0563602447509766e-05} {"step": 20040, "total_steps": 28610, "tokens": 42026926080, "source": "cosmopedia_v2", "phase": 4, "loss": 2.833428144454956, "ce_loss": 2.815986067056656, "logit_z_loss": 26.391392707824707, "router_aux_loss": 1.537671446800232, "router_z_loss": 2.0653647631406784, "routing_entropy_normalized": 0.6507308219376612, "learning_rate": 0.003, "grad_norm": 1.5055471658706665, "tokens_per_second": 4139809.7337061195, "eta_seconds": 4341.405474185944, "data_wait_seconds": 2.250075340270996e-05} {"step": 20050, "total_steps": 28610, "tokens": 42047897600, "source": "fineweb_edu", "phase": 4, "loss": 3.3602438867092133, "ce_loss": 3.3434671759605408, "logit_z_loss": 25.384528398513794, "router_aux_loss": 1.481414258480072, "router_z_loss": 1.962574690580368, "routing_entropy_normalized": 0.6642386529052254, "learning_rate": 0.003, "grad_norm": 0.8553471565246582, "tokens_per_second": 3754669.4460465554, "eta_seconds": 4781.145551681519, "data_wait_seconds": 2.232193946838379e-05} {"step": 20060, "total_steps": 28610, "tokens": 42068869120, "source": "finemath", "phase": 4, "loss": 2.3839286863803864, "ce_loss": 2.367903083562851, "logit_z_loss": 36.70796060562134, "router_aux_loss": 1.4523878991603851, "router_z_loss": 1.5017133057117462, "routing_entropy_normalized": 0.6748982563974454, "learning_rate": 0.003, "grad_norm": 1.0604690313339233, "tokens_per_second": 3799733.2880710806, "eta_seconds": 4718.9232087135315, "data_wait_seconds": 1.96993350982666e-05} {"step": 20070, "total_steps": 28610, "tokens": 42089840640, "source": "finemath", "phase": 4, "loss": 2.293439120054245, "ce_loss": 2.277919262647629, "logit_z_loss": 35.62688589096069, "router_aux_loss": 1.4071395695209503, "router_z_loss": 1.4484363049268723, "routing_entropy_normalized": 0.6924185714727012, "learning_rate": 0.003, "grad_norm": 0.8394943475723267, "tokens_per_second": 3770578.3469360922, "eta_seconds": 4749.849076747894, "data_wait_seconds": 1.9282102584838867e-05} {"step": 20080, "total_steps": 28610, "tokens": 42110812160, "source": "dclm", "phase": 4, "loss": 3.451597660779953, "ce_loss": 3.4360073804855347, "logit_z_loss": 26.001579761505127, "router_aux_loss": 1.3879999667406082, "router_z_loss": 1.7102254033088684, "routing_entropy_normalized": 0.6703548126698328, "learning_rate": 0.003, "grad_norm": 1.0874842405319214, "tokens_per_second": 3718949.267662243, "eta_seconds": 4810.150736808777, "data_wait_seconds": 2.288818359375e-05} {"step": 20090, "total_steps": 28610, "tokens": 42131783680, "source": "fineweb_edu", "phase": 4, "loss": 3.3677496016025543, "ce_loss": 3.3518760800361633, "logit_z_loss": 25.24233055114746, "router_aux_loss": 1.41059248149395, "router_z_loss": 1.767598807811737, "routing_entropy_normalized": 0.6616826838742481, "learning_rate": 0.003, "grad_norm": 0.6872286796569824, "tokens_per_second": 3673306.9139532777, "eta_seconds": 4864.209677696228, "data_wait_seconds": 2.1398067474365234e-05} {"step": 20100, "total_steps": 28610, "tokens": 42152755200, "source": "fineweb_edu", "phase": 4, "loss": 3.3845527172088623, "ce_loss": 3.3692960143089294, "logit_z_loss": 25.147257566452026, "router_aux_loss": 1.3561443239450455, "router_z_loss": 1.6952596753835678, "routing_entropy_normalized": 0.6474678592066501, "learning_rate": 0.003, "grad_norm": 1.5515227317810059, "tokens_per_second": 3814718.7876851284, "eta_seconds": 4678.395581245422, "data_wait_seconds": 2.1755695343017578e-05} {"step": 20110, "total_steps": 28610, "tokens": 42173726720, "source": "finemath", "phase": 4, "loss": 2.2913232147693634, "ce_loss": 2.2765905559062958, "logit_z_loss": 35.27269887924194, "router_aux_loss": 1.3590568751096725, "router_z_loss": 1.142113246023655, "routing_entropy_normalized": 0.6655866376638576, "learning_rate": 0.003, "grad_norm": 1.437415361404419, "tokens_per_second": 3918207.1862473986, "eta_seconds": 4549.476623535156, "data_wait_seconds": 2.2023916244506836e-05} {"step": 20120, "total_steps": 28610, "tokens": 42194698240, "source": "fineweb_edu", "phase": 4, "loss": 3.3808133602142334, "ce_loss": 3.3647202849388123, "logit_z_loss": 26.4560809135437, "router_aux_loss": 1.4584789723157883, "router_z_loss": 1.5082185566425323, "routing_entropy_normalized": 0.6249143503123318, "learning_rate": 0.003, "grad_norm": 1.3858600854873657, "tokens_per_second": 4231574.2637797715, "eta_seconds": 4207.611486911774, "data_wait_seconds": 2.3633241653442383e-05} {"step": 20130, "total_steps": 28610, "tokens": 42215669760, "source": "fineweb_hq", "phase": 4, "loss": 3.3104037046432495, "ce_loss": 3.295431226491928, "logit_z_loss": 26.445526361465454, "router_aux_loss": 1.3556322008371353, "router_z_loss": 1.416198119521141, "routing_entropy_normalized": 0.6708248205005801, "learning_rate": 0.003, "grad_norm": 1.9931992292404175, "tokens_per_second": 4042736.6529206187, "eta_seconds": 4398.963001251221, "data_wait_seconds": 2.0951032638549805e-05} {"step": 20140, "total_steps": 28610, "tokens": 42236641280, "source": "cosmopedia_v2", "phase": 4, "loss": 2.766106516122818, "ce_loss": 2.750728517770767, "logit_z_loss": 27.748011827468872, "router_aux_loss": 1.364061951637268, "router_z_loss": 1.737337738275528, "routing_entropy_normalized": 0.6757633643513238, "learning_rate": 0.003, "grad_norm": 1.8526183366775513, "tokens_per_second": 3906217.931619083, "eta_seconds": 4547.333956003189, "data_wait_seconds": 2.1904706954956055e-05} {"step": 20150, "total_steps": 28610, "tokens": 42257612800, "source": "fineweb_edu", "phase": 4, "loss": 3.3694644272327423, "ce_loss": 3.3535102903842926, "logit_z_loss": 26.15550923347473, "router_aux_loss": 1.4514390230178833, "router_z_loss": 1.4397184401750565, "routing_entropy_normalized": 0.640137167544462, "learning_rate": 0.003, "grad_norm": 1.772809386253357, "tokens_per_second": 4292357.090179634, "eta_seconds": 4133.371373176575, "data_wait_seconds": 2.244114875793457e-05} {"step": 20160, "total_steps": 28610, "tokens": 42278584320, "source": "finemath", "phase": 4, "loss": 2.353607028722763, "ce_loss": 2.337060898542404, "logit_z_loss": 37.025585651397705, "router_aux_loss": 1.553333193063736, "router_z_loss": 1.0127798318862915, "routing_entropy_normalized": 0.6314420509108344, "learning_rate": 0.003, "grad_norm": 0.9344105124473572, "tokens_per_second": 4265184.106842993, "eta_seconds": 4154.787684679031, "data_wait_seconds": 2.2023916244506836e-05} {"step": 20170, "total_steps": 28610, "tokens": 42299555840, "source": "finemath", "phase": 4, "loss": 2.2730052173137665, "ce_loss": 2.2573700845241547, "logit_z_loss": 36.94323921203613, "router_aux_loss": 1.4588344991207123, "router_z_loss": 1.0467734709382057, "routing_entropy_normalized": 0.6506346327856252, "learning_rate": 0.003, "grad_norm": 0.9600067138671875, "tokens_per_second": 3797596.229128714, "eta_seconds": 4660.8332777023315, "data_wait_seconds": 2.086162567138672e-05} {"step": 20180, "total_steps": 28610, "tokens": 42320527360, "source": "dclm", "phase": 4, "loss": 3.441467523574829, "ce_loss": 3.4266732931137085, "logit_z_loss": 26.84184694290161, "router_aux_loss": 1.3537669777870178, "router_z_loss": 1.2565557062625885, "routing_entropy_normalized": 0.6563134053399472, "learning_rate": 0.003, "grad_norm": 1.8445690870285034, "tokens_per_second": 4092016.6567916875, "eta_seconds": 4320.361533880234, "data_wait_seconds": 2.0116567611694336e-05} {"step": 20190, "total_steps": 28610, "tokens": 42341498880, "source": "fineweb_edu", "phase": 4, "loss": 3.4064982533454895, "ce_loss": 3.3912273645401, "logit_z_loss": 25.813817501068115, "router_aux_loss": 1.3834117203950882, "router_z_loss": 1.4367638379335403, "routing_entropy_normalized": 0.6781183086965945, "learning_rate": 0.003, "grad_norm": 0.8982006311416626, "tokens_per_second": 4204792.562991869, "eta_seconds": 4199.498447418213, "data_wait_seconds": 2.2202730178833008e-05} {"step": 20200, "total_steps": 28610, "tokens": 42362470400, "source": "fineweb_edu", "phase": 4, "loss": 3.398163676261902, "ce_loss": 3.382345139980316, "logit_z_loss": 26.141283988952637, "router_aux_loss": 1.4319146424531937, "router_z_loss": 1.499388426542282, "routing_entropy_normalized": 0.6480328057285387, "learning_rate": 0.003, "grad_norm": 2.1497533321380615, "tokens_per_second": 4134504.89978209, "eta_seconds": 4265.818700790405, "data_wait_seconds": 2.2292137145996094e-05} {"step": 20210, "total_steps": 28610, "tokens": 42383441920, "source": "finemath", "phase": 4, "loss": 2.2500279545783997, "ce_loss": 2.2340149879455566, "logit_z_loss": 37.60172653198242, "router_aux_loss": 1.5062743425369263, "router_z_loss": 0.9501871392130852, "routing_entropy_normalized": 0.6558133971148855, "learning_rate": 0.003, "grad_norm": 0.8569744229316711, "tokens_per_second": 3814202.1948116557, "eta_seconds": 4618.548231124878, "data_wait_seconds": 2.2113323211669922e-05} {"step": 20220, "total_steps": 28610, "tokens": 42404413440, "source": "fineweb_edu", "phase": 4, "loss": 3.3721152544021606, "ce_loss": 3.356104165315628, "logit_z_loss": 25.29028582572937, "router_aux_loss": 1.4498297423124313, "router_z_loss": 1.5127685219049454, "routing_entropy_normalized": 0.6656735506812025, "learning_rate": 0.003, "grad_norm": 0.9570428133010864, "tokens_per_second": 4027825.8749395274, "eta_seconds": 4368.387767076492, "data_wait_seconds": 2.321600914001465e-05} {"step": 20230, "total_steps": 28610, "tokens": 42425384960, "source": "fineweb_hq", "phase": 4, "loss": 3.258914828300476, "ce_loss": 3.2438412606716156, "logit_z_loss": 25.66720175743103, "router_aux_loss": 1.3550937473773956, "router_z_loss": 1.5225666910409927, "routing_entropy_normalized": 0.6433216094800602, "learning_rate": 0.003, "grad_norm": 1.1284289360046387, "tokens_per_second": 3855070.7788541173, "eta_seconds": 4558.705862522125, "data_wait_seconds": 0.04838564991950989} {"step": 20240, "total_steps": 28610, "tokens": 42446356480, "source": "cosmopedia_v2", "phase": 4, "loss": 2.785150170326233, "ce_loss": 2.7692295908927917, "logit_z_loss": 27.900080919265747, "router_aux_loss": 1.4177568554878235, "router_z_loss": 1.74297733604908, "routing_entropy_normalized": 0.6680224991213752, "learning_rate": 0.003, "grad_norm": 1.51990807056427, "tokens_per_second": 4219652.272000629, "eta_seconds": 4159.859890937805, "data_wait_seconds": 2.1457672119140625e-05} {"step": 20250, "total_steps": 28610, "tokens": 42467328000, "source": "fineweb_edu", "phase": 4, "loss": 3.3718723952770233, "ce_loss": 3.3562288880348206, "logit_z_loss": 25.800764799118042, "router_aux_loss": 1.4252120107412338, "router_z_loss": 1.391313761472702, "routing_entropy_normalized": 0.6468672400432441, "learning_rate": 0.003, "grad_norm": 1.0917342901229858, "tokens_per_second": 4168358.8901079907, "eta_seconds": 4206.0175676345825, "data_wait_seconds": 2.810359001159668e-05} {"step": 20260, "total_steps": 28610, "tokens": 42488299520, "source": "finemath", "phase": 4, "loss": 2.1661465764045715, "ce_loss": 2.151550278067589, "logit_z_loss": 34.84922695159912, "router_aux_loss": 1.358112022280693, "router_z_loss": 1.0151785388588905, "routing_entropy_normalized": 0.6892669019218514, "learning_rate": 0.003, "grad_norm": 0.9741861820220947, "tokens_per_second": 3523672.5405971175, "eta_seconds": 4969.593229293823, "data_wait_seconds": 2.345442771911621e-05} {"step": 20270, "total_steps": 28610, "tokens": 42509271040, "source": "finemath", "phase": 4, "loss": 2.217267245054245, "ce_loss": 2.2028698325157166, "logit_z_loss": 34.50367879867554, "router_aux_loss": 1.357498049736023, "router_z_loss": 0.8224435597658157, "routing_entropy_normalized": 0.6561640584202095, "learning_rate": 0.003, "grad_norm": 1.4861586093902588, "tokens_per_second": 3740022.787794389, "eta_seconds": 4676.508318901062, "data_wait_seconds": 2.3126602172851562e-05} {"step": 20280, "total_steps": 28610, "tokens": 42530242560, "source": "dclm", "phase": 4, "loss": 3.4119035601615906, "ce_loss": 3.3975231051445007, "logit_z_loss": 26.989444494247437, "router_aux_loss": 1.3306767493486404, "router_z_loss": 1.0736690759658813, "routing_entropy_normalized": 0.6708249704049014, "learning_rate": 0.003, "grad_norm": 0.8895925879478455, "tokens_per_second": 4103877.8773359396, "eta_seconds": 4256.772906541824, "data_wait_seconds": 1.8894672393798828e-05} {"step": 20290, "total_steps": 28610, "tokens": 42551214080, "source": "fineweb_edu", "phase": 4, "loss": 3.323340803384781, "ce_loss": 3.308070182800293, "logit_z_loss": 26.54471516609192, "router_aux_loss": 1.4117374122142792, "router_z_loss": 1.1532862484455109, "routing_entropy_normalized": 0.6509868622692038, "learning_rate": 0.003, "grad_norm": 1.037420630455017, "tokens_per_second": 3578397.699552347, "eta_seconds": 4876.010467529297, "data_wait_seconds": 2.294778823852539e-05} {"step": 20300, "total_steps": 28610, "tokens": 42572185600, "source": "fineweb_edu", "phase": 4, "loss": 3.3747197687625885, "ce_loss": 3.359005182981491, "logit_z_loss": 26.100632429122925, "router_aux_loss": 1.4561945647001266, "router_z_loss": 1.1526116728782654, "routing_entropy_normalized": 0.6448430675630789, "learning_rate": 0.003, "grad_norm": 1.9249377250671387, "tokens_per_second": 4101785.037404144, "eta_seconds": 4248.719267606735, "data_wait_seconds": 0.0362815260887146} {"step": 20310, "total_steps": 28610, "tokens": 42593157120, "source": "finemath", "phase": 4, "loss": 2.301962196826935, "ce_loss": 2.285928636789322, "logit_z_loss": 35.09656858444214, "router_aux_loss": 1.5281372219324112, "router_z_loss": 0.7522090747952461, "routing_entropy_normalized": 0.6301117823973561, "learning_rate": 0.003, "grad_norm": 1.0233272314071655, "tokens_per_second": 3311556.3367887954, "eta_seconds": 5256.248068809509, "data_wait_seconds": 2.199411392211914e-05} {"step": 20320, "total_steps": 28610, "tokens": 42614128640, "source": "fineweb_edu", "phase": 4, "loss": 3.366210460662842, "ce_loss": 3.3511100113391876, "logit_z_loss": 25.39997625350952, "router_aux_loss": 1.3851388990879059, "router_z_loss": 1.24909009039402, "routing_entropy_normalized": 0.6561281014284619, "learning_rate": 0.003, "grad_norm": 0.6199061274528503, "tokens_per_second": 4211788.38285939, "eta_seconds": 4127.792875528336, "data_wait_seconds": 2.059340476989746e-05} {"step": 20330, "total_steps": 28610, "tokens": 42635100160, "source": "fineweb_hq", "phase": 4, "loss": 3.437965452671051, "ce_loss": 3.4225495159626007, "logit_z_loss": 27.240915536880493, "router_aux_loss": 1.417795479297638, "router_z_loss": 1.2379599511623383, "routing_entropy_normalized": 0.6514924941956592, "learning_rate": 0.003, "grad_norm": 1.4550329446792603, "tokens_per_second": 4526930.019532086, "eta_seconds": 3835.8045043945312, "data_wait_seconds": 1.8835067749023438e-05} {"step": 20340, "total_steps": 28610, "tokens": 42656071680, "source": "cosmopedia_v2", "phase": 4, "loss": 2.8007790446281433, "ce_loss": 2.7852156162261963, "logit_z_loss": 27.85348629951477, "router_aux_loss": 1.4067477136850357, "router_z_loss": 1.4958895146846771, "routing_entropy_normalized": 0.6790648619251638, "learning_rate": 0.003, "grad_norm": 1.641467571258545, "tokens_per_second": 3414008.6916606, "eta_seconds": 5080.082860469818, "data_wait_seconds": 2.0742416381835938e-05} {"step": 20350, "total_steps": 28610, "tokens": 42677043200, "source": "fineweb_edu", "phase": 4, "loss": 3.367325574159622, "ce_loss": 3.352668970823288, "logit_z_loss": 25.63351082801819, "router_aux_loss": 1.342595413327217, "router_z_loss": 1.230622798204422, "routing_entropy_normalized": 0.6922502168554369, "learning_rate": 0.003, "grad_norm": 0.9800376892089844, "tokens_per_second": 4151079.293840734, "eta_seconds": 4173.0052099227905, "data_wait_seconds": 2.3692846298217773e-05} {"step": 20360, "total_steps": 28610, "tokens": 42698014720, "source": "finemath", "phase": 4, "loss": 2.270208239555359, "ce_loss": 2.255466729402542, "logit_z_loss": 36.293489933013916, "router_aux_loss": 1.394104316830635, "router_z_loss": 0.8004945516586304, "routing_entropy_normalized": 0.6827855416153394, "learning_rate": 0.003, "grad_norm": 0.9859448671340942, "tokens_per_second": 4033537.168889468, "eta_seconds": 4289.412313699722, "data_wait_seconds": 2.130866050720215e-05} {"step": 20370, "total_steps": 28610, "tokens": 42718986240, "source": "finemath", "phase": 4, "loss": 2.181699827313423, "ce_loss": 2.166900262236595, "logit_z_loss": 35.514973640441895, "router_aux_loss": 1.4037378281354904, "router_z_loss": 0.7621648758649826, "routing_entropy_normalized": 0.663623805902516, "learning_rate": 0.003, "grad_norm": 1.1006097793579102, "tokens_per_second": 3675857.420284133, "eta_seconds": 4701.088890075684, "data_wait_seconds": 2.1785497665405273e-05} {"step": 20380, "total_steps": 28610, "tokens": 42739957760, "source": "dclm", "phase": 4, "loss": 3.480110764503479, "ce_loss": 3.4643931686878204, "logit_z_loss": 27.53051233291626, "router_aux_loss": 1.4732719361782074, "router_z_loss": 0.9848247542977333, "routing_entropy_normalized": 0.648074037647728, "learning_rate": 0.003, "grad_norm": 3.230604887008667, "tokens_per_second": 4286098.337243374, "eta_seconds": 4026.870034694671, "data_wait_seconds": 2.1219253540039062e-05} {"step": 20390, "total_steps": 28610, "tokens": 42760929280, "source": "fineweb_edu", "phase": 4, "loss": 3.4083648324012756, "ce_loss": 3.3920798301696777, "logit_z_loss": 26.2915096282959, "router_aux_loss": 1.5033690333366394, "router_z_loss": 1.2513482570648193, "routing_entropy_normalized": 0.6531360090771342, "learning_rate": 0.003, "grad_norm": 1.2510546445846558, "tokens_per_second": 3922964.8729808605, "eta_seconds": 4394.275757789612, "data_wait_seconds": 2.1219253540039062e-05} {"step": 20400, "total_steps": 28610, "tokens": 42781900800, "source": "fineweb_edu", "phase": 4, "loss": 3.363039553165436, "ce_loss": 3.3474903404712677, "logit_z_loss": 26.9727303981781, "router_aux_loss": 1.4210503846406937, "router_z_loss": 1.3386648446321487, "routing_entropy_normalized": 0.6347932425528188, "learning_rate": 0.003, "grad_norm": 1.4130171537399292, "tokens_per_second": 3849470.247077272, "eta_seconds": 4472.723989248276, "data_wait_seconds": 2.2143125534057617e-05} {"step": 20410, "total_steps": 28610, "tokens": 42802872320, "source": "finemath", "phase": 4, "loss": 2.296281546354294, "ce_loss": 2.2809905111789703, "logit_z_loss": 36.87760543823242, "router_aux_loss": 1.433151662349701, "router_z_loss": 0.9595006927847862, "routing_entropy_normalized": 0.6738012922579748, "learning_rate": 0.003, "grad_norm": 1.2694042921066284, "tokens_per_second": 3998454.739142908, "eta_seconds": 4300.823073387146, "data_wait_seconds": 1.996755599975586e-05} {"step": 20420, "total_steps": 28610, "tokens": 42823843840, "source": "fineweb_edu", "phase": 4, "loss": 3.3560255467891693, "ce_loss": 3.340797632932663, "logit_z_loss": 25.75749945640564, "router_aux_loss": 1.3898184150457382, "router_z_loss": 1.3297472149133682, "routing_entropy_normalized": 0.6649301929458448, "learning_rate": 0.003, "grad_norm": 1.6222575902938843, "tokens_per_second": 4389442.764341614, "eta_seconds": 3912.951097011566, "data_wait_seconds": 2.008676528930664e-05} {"step": 20430, "total_steps": 28610, "tokens": 42844815360, "source": "fineweb_hq", "phase": 4, "loss": 3.439577877521515, "ce_loss": 3.4238592982292175, "logit_z_loss": 24.67670702934265, "router_aux_loss": 1.4395128041505814, "router_z_loss": 1.3234210014343262, "routing_entropy_normalized": 0.6338038784785238, "learning_rate": 0.003, "grad_norm": 1.2174880504608154, "tokens_per_second": 3571310.3152480344, "eta_seconds": 4803.475992202759, "data_wait_seconds": 1.9371509552001953e-05} {"step": 20440, "total_steps": 28610, "tokens": 42865786880, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7770027220249176, "ce_loss": 2.7594090700149536, "logit_z_loss": 28.226289749145508, "router_aux_loss": 1.5838878899812698, "router_z_loss": 1.7547895908355713, "routing_entropy_normalized": 0.6177354624783519, "learning_rate": 0.003, "grad_norm": 1.5981428623199463, "tokens_per_second": 3917716.8015129105, "eta_seconds": 4373.397238254547, "data_wait_seconds": 2.1249055862426758e-05} {"step": 20450, "total_steps": 28610, "tokens": 42886758400, "source": "fineweb_edu", "phase": 4, "loss": 3.363338351249695, "ce_loss": 3.347102105617523, "logit_z_loss": 25.844398021697998, "router_aux_loss": 1.461034119129181, "router_z_loss": 1.6259295493364334, "routing_entropy_normalized": 0.6687169742097643, "learning_rate": 0.003, "grad_norm": 0.6556743383407593, "tokens_per_second": 3683921.9290747824, "eta_seconds": 4645.255966186523, "data_wait_seconds": 2.181529998779297e-05} {"step": 20460, "total_steps": 28610, "tokens": 42907729920, "source": "finemath", "phase": 4, "loss": 2.0044044703245163, "ce_loss": 1.9882169179618359, "logit_z_loss": 35.96727514266968, "router_aux_loss": 1.5041660368442535, "router_z_loss": 1.1458586603403091, "routing_entropy_normalized": 0.6263870887158628, "learning_rate": 0.003, "grad_norm": 1.32644522190094, "tokens_per_second": 4058673.0550600057, "eta_seconds": 4211.176551580429, "data_wait_seconds": 2.390146255493164e-05} {"step": 20470, "total_steps": 28610, "tokens": 42928701440, "source": "finemath", "phase": 4, "loss": 2.305661290884018, "ce_loss": 2.290493428707123, "logit_z_loss": 37.68231439590454, "router_aux_loss": 1.3932574093341827, "router_z_loss": 1.2353202402591705, "routing_entropy_normalized": 0.6783743470725228, "learning_rate": 0.003, "grad_norm": 0.6372738480567932, "tokens_per_second": 3244522.785286385, "eta_seconds": 5261.426228046417, "data_wait_seconds": 2.2679567337036133e-05} {"step": 20480, "total_steps": 28610, "tokens": 42949672960, "source": "dclm", "phase": 4, "loss": 3.502192199230194, "ce_loss": 3.486903727054596, "logit_z_loss": 26.334014177322388, "router_aux_loss": 1.3920650035142899, "router_z_loss": 1.3677748441696167, "routing_entropy_normalized": 0.6636538753389991, "learning_rate": 0.003, "grad_norm": 1.4796464443206787, "tokens_per_second": 4338187.762537728, "eta_seconds": 3930.176998615265, "data_wait_seconds": 1.9878149032592773e-05} {"step": 20490, "total_steps": 28610, "tokens": 42970644480, "source": "fineweb_edu", "phase": 4, "loss": 3.3440488278865814, "ce_loss": 3.329676479101181, "logit_z_loss": 26.31727433204651, "router_aux_loss": 1.2879991680383682, "router_z_loss": 1.4923721998929977, "routing_entropy_normalized": 0.6908458816667478, "learning_rate": 0.003, "grad_norm": 1.2203283309936523, "tokens_per_second": 3712759.789011445, "eta_seconds": 4586.5811977386475, "data_wait_seconds": 2.154707908630371e-05} {"step": 20500, "total_steps": 28610, "tokens": 42991616000, "source": "fineweb_edu", "phase": 4, "loss": 3.3396436274051666, "ce_loss": 3.325360655784607, "logit_z_loss": 25.812362670898438, "router_aux_loss": 1.2865878641605377, "router_z_loss": 1.4171478748321533, "routing_entropy_normalized": 0.6548583034905264, "learning_rate": 0.003, "grad_norm": 0.8072299361228943, "tokens_per_second": 3670397.539661239, "eta_seconds": 4633.803977966309, "data_wait_seconds": 1.8775463104248047e-05} {"step": 20510, "total_steps": 28610, "tokens": 43012587520, "source": "finemath", "phase": 4, "loss": 2.2676260471343994, "ce_loss": 2.253062129020691, "logit_z_loss": 35.724539279937744, "router_aux_loss": 1.3723777383565903, "router_z_loss": 0.8401381373405457, "routing_entropy_normalized": 0.6898911278153284, "learning_rate": 0.003, "grad_norm": 1.2212916612625122, "tokens_per_second": 4108969.418308563, "eta_seconds": 4134.10991191864, "data_wait_seconds": 1.9758939743041992e-05} {"step": 20520, "total_steps": 28610, "tokens": 43033559040, "source": "fineweb_edu", "phase": 4, "loss": 3.3524743616580963, "ce_loss": 3.337080478668213, "logit_z_loss": 25.896434545516968, "router_aux_loss": 1.4201983213424683, "router_z_loss": 1.1919342875480652, "routing_entropy_normalized": 0.65151095533219, "learning_rate": 0.003, "grad_norm": 1.2867100238800049, "tokens_per_second": 4291323.233385166, "eta_seconds": 3953.5496995449066, "data_wait_seconds": 2.0325183868408203e-05} {"step": 20530, "total_steps": 28610, "tokens": 43054530560, "source": "fineweb_hq", "phase": 4, "loss": 3.4543557167053223, "ce_loss": 3.4382029473781586, "logit_z_loss": 26.579664707183838, "router_aux_loss": 1.4927653670310974, "router_z_loss": 1.225127175450325, "routing_entropy_normalized": 0.6443333752680439, "learning_rate": 0.003, "grad_norm": 2.1623494625091553, "tokens_per_second": 3839839.785823371, "eta_seconds": 4412.941452026367, "data_wait_seconds": 1.9311904907226562e-05} {"step": 20540, "total_steps": 28610, "tokens": 43075502080, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7584874629974365, "ce_loss": 2.741699367761612, "logit_z_loss": 28.639660120010376, "router_aux_loss": 1.5191750079393387, "router_z_loss": 1.5963109582662582, "routing_entropy_normalized": 0.6398707507402557, "learning_rate": 0.003, "grad_norm": 1.4089633226394653, "tokens_per_second": 4047916.508052696, "eta_seconds": 4180.920383691788, "data_wait_seconds": 2.0951032638549805e-05} {"step": 20550, "total_steps": 28610, "tokens": 43096473600, "source": "fineweb_edu", "phase": 4, "loss": 3.3674493730068207, "ce_loss": 3.3523511588573456, "logit_z_loss": 25.765790700912476, "router_aux_loss": 1.371154323220253, "router_z_loss": 1.3866182565689087, "routing_entropy_normalized": 0.6747942692810562, "learning_rate": 0.003, "grad_norm": 1.4251574277877808, "tokens_per_second": 4207216.228494386, "eta_seconds": 4017.631659984588, "data_wait_seconds": 2.0891427993774414e-05} {"step": 20560, "total_steps": 28610, "tokens": 43117445120, "source": "finemath", "phase": 4, "loss": 2.1969870030879974, "ce_loss": 2.1817051470279694, "logit_z_loss": 36.98540115356445, "router_aux_loss": 1.444025069475174, "router_z_loss": 0.8415834084153175, "routing_entropy_normalized": 0.6448709690490293, "learning_rate": 0.003, "grad_norm": 1.033631443977356, "tokens_per_second": 4135899.743675669, "eta_seconds": 4081.838208436966, "data_wait_seconds": 2.467632293701172e-05} {"step": 20570, "total_steps": 28610, "tokens": 43138416640, "source": "finemath", "phase": 4, "loss": 2.2833320796489716, "ce_loss": 2.269395560026169, "logit_z_loss": 36.459606647491455, "router_aux_loss": 1.3009632974863052, "router_z_loss": 0.9268732592463493, "routing_entropy_normalized": 0.7194007216770519, "learning_rate": 0.003, "grad_norm": 0.719754159450531, "tokens_per_second": 3259254.2693919316, "eta_seconds": 5173.300603866577, "data_wait_seconds": 2.1070241928100586e-05} {"step": 20580, "total_steps": 28610, "tokens": 43159388160, "source": "dclm", "phase": 4, "loss": 3.540186107158661, "ce_loss": 3.5254423916339874, "logit_z_loss": 26.80786943435669, "router_aux_loss": 1.3703072667121887, "router_z_loss": 1.0406227856874466, "routing_entropy_normalized": 0.6329679435050591, "learning_rate": 0.003, "grad_norm": 0.898253858089447, "tokens_per_second": 4452477.604572134, "eta_seconds": 3782.193209171295, "data_wait_seconds": 2.574920654296875e-05} {"step": 20590, "total_steps": 28610, "tokens": 43180359680, "source": "fineweb_edu", "phase": 4, "loss": 3.409484952688217, "ce_loss": 3.3938548266887665, "logit_z_loss": 25.156013250350952, "router_aux_loss": 1.4435449689626694, "router_z_loss": 1.194683849811554, "routing_entropy_normalized": 0.652663509385683, "learning_rate": 0.003, "grad_norm": 2.0965330600738525, "tokens_per_second": 4001243.038876469, "eta_seconds": 4203.483486652374, "data_wait_seconds": 2.2619962692260742e-05} {"step": 20600, "total_steps": 28610, "tokens": 43201331200, "source": "fineweb_edu", "phase": 4, "loss": 3.4239615201950073, "ce_loss": 3.407844454050064, "logit_z_loss": 26.514633893966675, "router_aux_loss": 1.4911277443170547, "router_z_loss": 1.2058187276124954, "routing_entropy_normalized": 0.6465895446332722, "learning_rate": 0.003, "grad_norm": 1.8221156597137451, "tokens_per_second": 4361144.901576555, "eta_seconds": 3851.7838547229762, "data_wait_seconds": 2.2083520889282227e-05} {"step": 20610, "total_steps": 28610, "tokens": 43222302720, "source": "finemath", "phase": 4, "loss": 2.340527683496475, "ce_loss": 2.324683666229248, "logit_z_loss": 36.47043800354004, "router_aux_loss": 1.4932714998722076, "router_z_loss": 0.9112787619233131, "routing_entropy_normalized": 0.6633978740511945, "learning_rate": 0.003, "grad_norm": 0.6811554431915283, "tokens_per_second": 3541052.3521936256, "eta_seconds": 4737.9180908203125, "data_wait_seconds": 2.3484230041503906e-05} {"step": 20620, "total_steps": 28610, "tokens": 43243274240, "source": "fineweb_edu", "phase": 4, "loss": 3.3960534036159515, "ce_loss": 3.3801335394382477, "logit_z_loss": 25.46397042274475, "router_aux_loss": 1.4528737366199493, "router_z_loss": 1.3911160230636597, "routing_entropy_normalized": 0.6660043038305957, "learning_rate": 0.003, "grad_norm": 1.2126541137695312, "tokens_per_second": 4327868.393948764, "eta_seconds": 3871.7084150314327, "data_wait_seconds": 2.09808349609375e-05} {"step": 20630, "total_steps": 28610, "tokens": 43264245760, "source": "fineweb_hq", "phase": 4, "loss": 3.423190653324127, "ce_loss": 3.4077567160129547, "logit_z_loss": 26.682486057281494, "router_aux_loss": 1.4063833355903625, "router_z_loss": 1.3700926899909973, "routing_entropy_normalized": 0.6675023224818665, "learning_rate": 0.003, "grad_norm": 1.226146936416626, "tokens_per_second": 3822890.196206156, "eta_seconds": 4377.649396419525, "data_wait_seconds": 2.0623207092285156e-05} {"step": 20640, "total_steps": 28610, "tokens": 43285217280, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7520812451839447, "ce_loss": 2.7368547320365906, "logit_z_loss": 28.63291311264038, "router_aux_loss": 1.3589307367801666, "router_z_loss": 1.6372238844633102, "routing_entropy_normalized": 0.6487876226951677, "learning_rate": 0.003, "grad_norm": 1.2502760887145996, "tokens_per_second": 4388877.268714203, "eta_seconds": 3808.331930160522, "data_wait_seconds": 1.7702579498291016e-05} {"step": 20650, "total_steps": 28610, "tokens": 43306188800, "source": "fineweb_edu", "phase": 4, "loss": 3.3857865631580353, "ce_loss": 3.3707431852817535, "logit_z_loss": 26.05701518058777, "router_aux_loss": 1.3757025003433228, "router_z_loss": 1.2863443940877914, "routing_entropy_normalized": 0.6404763380492405, "learning_rate": 0.003, "grad_norm": 1.3558305501937866, "tokens_per_second": 3155831.307965229, "eta_seconds": 5289.67751789093, "data_wait_seconds": 2.0623207092285156e-05} {"step": 20660, "total_steps": 28610, "tokens": 43327160320, "source": "finemath", "phase": 4, "loss": 2.2335110008716583, "ce_loss": 2.218750774860382, "logit_z_loss": 34.5040864944458, "router_aux_loss": 1.388515830039978, "router_z_loss": 0.8750774189829826, "routing_entropy_normalized": 0.6861506875963426, "learning_rate": 0.003, "grad_norm": 0.9668921828269958, "tokens_per_second": 4026801.9483582303, "eta_seconds": 4140.347256660461, "data_wait_seconds": 1.9997358322143555e-05} {"step": 20670, "total_steps": 28610, "tokens": 43348131840, "source": "finemath", "phase": 4, "loss": 2.2485549449920654, "ce_loss": 2.2334333956241608, "logit_z_loss": 35.75097179412842, "router_aux_loss": 1.4352477341890335, "router_z_loss": 0.7690414413809776, "routing_entropy_normalized": 0.6438582313147582, "learning_rate": 0.003, "grad_norm": 0.8462193608283997, "tokens_per_second": 3706219.8028898435, "eta_seconds": 4492.8222732543945, "data_wait_seconds": 2.148747444152832e-05} {"step": 20680, "total_steps": 28610, "tokens": 43369103360, "source": "dclm", "phase": 4, "loss": 3.4882216453552246, "ce_loss": 3.4737854301929474, "logit_z_loss": 28.14098024368286, "router_aux_loss": 1.338075652718544, "router_z_loss": 1.0554639399051666, "routing_entropy_normalized": 0.6659530454563174, "learning_rate": 0.003, "grad_norm": 1.437811255455017, "tokens_per_second": 3867419.194264532, "eta_seconds": 4300.132601261139, "data_wait_seconds": 2.002716064453125e-05} {"step": 20690, "total_steps": 28610, "tokens": 43390074880, "source": "fineweb_edu", "phase": 4, "loss": 3.3741086423397064, "ce_loss": 3.3595301806926727, "logit_z_loss": 26.490408182144165, "router_aux_loss": 1.338247925043106, "router_z_loss": 1.1960268914699554, "routing_entropy_normalized": 0.6639724403677191, "learning_rate": 0.003, "grad_norm": 1.3906917572021484, "tokens_per_second": 4538401.33779677, "eta_seconds": 3659.756509780884, "data_wait_seconds": 2.0116567611694336e-05} {"step": 20700, "total_steps": 28610, "tokens": 43411046400, "source": "fineweb_edu", "phase": 4, "loss": 3.3676418364048004, "ce_loss": 3.3526734709739685, "logit_z_loss": 25.658313512802124, "router_aux_loss": 1.3716365545988083, "router_z_loss": 1.2519703060388565, "routing_entropy_normalized": 0.6537762114450043, "learning_rate": 0.003, "grad_norm": 1.0043867826461792, "tokens_per_second": 3251393.6181791644, "eta_seconds": 5101.957581281662, "data_wait_seconds": 0.11641961336135864} {"step": 20710, "total_steps": 28610, "tokens": 43432017920, "source": "finemath", "phase": 4, "loss": 2.2445462346076965, "ce_loss": 2.2295511960983276, "logit_z_loss": 36.56511116027832, "router_aux_loss": 1.4216671735048294, "router_z_loss": 0.7784446701407433, "routing_entropy_normalized": 0.6527100305576444, "learning_rate": 0.003, "grad_norm": 0.752045214176178, "tokens_per_second": 4015477.879364378, "eta_seconds": 4125.910115242004, "data_wait_seconds": 1.868605613708496e-05} {"step": 20720, "total_steps": 28610, "tokens": 43452989440, "source": "fineweb_edu", "phase": 4, "loss": 3.3692882359027863, "ce_loss": 3.354341357946396, "logit_z_loss": 26.144777059555054, "router_aux_loss": 1.3772161900997162, "router_z_loss": 1.1747297197580338, "routing_entropy_normalized": 0.6723169258824792, "learning_rate": 0.003, "grad_norm": 0.6052066087722778, "tokens_per_second": 4441993.8211731585, "eta_seconds": 3725.023029327393, "data_wait_seconds": 1.8358230590820312e-05} {"step": 20730, "total_steps": 28610, "tokens": 43473960960, "source": "fineweb_hq", "phase": 4, "loss": 3.465005099773407, "ce_loss": 3.449921101331711, "logit_z_loss": 26.424193859100342, "router_aux_loss": 1.3904020637273788, "router_z_loss": 1.1799636036157608, "routing_entropy_normalized": 0.6467348005009612, "learning_rate": 0.003, "grad_norm": 1.5467044115066528, "tokens_per_second": 3799181.6910956735, "eta_seconds": 4349.767687797546, "data_wait_seconds": 2.1785497665405273e-05} {"step": 20740, "total_steps": 28610, "tokens": 43494932480, "source": "cosmopedia_v2", "phase": 4, "loss": 2.76463919878006, "ce_loss": 2.7486503422260284, "logit_z_loss": 28.893805027008057, "router_aux_loss": 1.4627779424190521, "router_z_loss": 1.361041322350502, "routing_entropy_normalized": 0.6480099127838816, "learning_rate": 0.003, "grad_norm": 1.0156968832015991, "tokens_per_second": 3801980.2255930123, "eta_seconds": 4341.049995183945, "data_wait_seconds": 2.1576881408691406e-05} {"step": 20750, "total_steps": 28610, "tokens": 43515904000, "source": "fineweb_edu", "phase": 4, "loss": 3.379635751247406, "ce_loss": 3.3641222715377808, "logit_z_loss": 26.453980684280396, "router_aux_loss": 1.4280111640691757, "router_z_loss": 1.233370453119278, "routing_entropy_normalized": 0.6469015089035316, "learning_rate": 0.003, "grad_norm": 1.830185890197754, "tokens_per_second": 4057115.9045739165, "eta_seconds": 4062.8897738456726, "data_wait_seconds": 2.1189451217651367e-05} {"step": 20760, "total_steps": 28610, "tokens": 43536875520, "source": "finemath", "phase": 4, "loss": 2.2441538274288177, "ce_loss": 2.2283477634191513, "logit_z_loss": 36.52823495864868, "router_aux_loss": 1.492131769657135, "router_z_loss": 0.8847673088312149, "routing_entropy_normalized": 0.6480229391465118, "learning_rate": 0.003, "grad_norm": 2.0847318172454834, "tokens_per_second": 3802159.8521616925, "eta_seconds": 4329.813537597656, "data_wait_seconds": 2.205371856689453e-05} {"step": 20770, "total_steps": 28610, "tokens": 43557847040, "source": "finemath", "phase": 4, "loss": 2.241023153066635, "ce_loss": 2.2255708277225494, "logit_z_loss": 35.493661403656006, "router_aux_loss": 1.4613311886787415, "router_z_loss": 0.8389230221509933, "routing_entropy_normalized": 0.6447870655775412, "learning_rate": 0.003, "grad_norm": 0.7521913647651672, "tokens_per_second": 3740257.837878291, "eta_seconds": 4395.8658447265625, "data_wait_seconds": 1.9073486328125e-05} {"step": 20780, "total_steps": 28610, "tokens": 43578818560, "source": "dclm", "phase": 4, "loss": 3.4345858991146088, "ce_loss": 3.419363707304001, "logit_z_loss": 26.859962701797485, "router_aux_loss": 1.4095819592475891, "router_z_loss": 1.1264135241508484, "routing_entropy_normalized": 0.635086839620013, "learning_rate": 0.003, "grad_norm": 1.2238411903381348, "tokens_per_second": 3887253.268403868, "eta_seconds": 4224.242421627045, "data_wait_seconds": 2.282857894897461e-05} {"step": 20790, "total_steps": 28610, "tokens": 43599790080, "source": "fineweb_edu", "phase": 4, "loss": 3.389034181833267, "ce_loss": 3.3739985823631287, "logit_z_loss": 26.128169298171997, "router_aux_loss": 1.3740916848182678, "router_z_loss": 1.2946292757987976, "routing_entropy_normalized": 0.6375070468232895, "learning_rate": 0.003, "grad_norm": 1.4534225463867188, "tokens_per_second": 3793900.9771296526, "eta_seconds": 4322.655951976776, "data_wait_seconds": 2.232193946838379e-05} {"step": 20800, "total_steps": 28610, "tokens": 43620761600, "source": "fineweb_edu", "phase": 4, "loss": 3.3826911449432373, "ce_loss": 3.3677582144737244, "logit_z_loss": 27.1705904006958, "router_aux_loss": 1.3683607280254364, "router_z_loss": 1.2492953091859818, "routing_entropy_normalized": 0.6766510474358003, "learning_rate": 0.003, "grad_norm": 0.7951528429985046, "tokens_per_second": 3134112.7469844283, "eta_seconds": 5225.962957382202, "data_wait_seconds": 0.2183358073234558} {"step": 20810, "total_steps": 28610, "tokens": 43641733120, "source": "finemath", "phase": 4, "loss": 2.423812448978424, "ce_loss": 2.4078521728515625, "logit_z_loss": 37.06691932678223, "router_aux_loss": 1.5050608813762665, "router_z_loss": 0.9097422957420349, "routing_entropy_normalized": 0.6470673166479866, "learning_rate": 0.003, "grad_norm": 1.520905613899231, "tokens_per_second": 3407763.28081226, "eta_seconds": 4800.153136253357, "data_wait_seconds": 2.142786979675293e-05} {"step": 20820, "total_steps": 28610, "tokens": 43662704640, "source": "fineweb_edu", "phase": 4, "loss": 3.3218605518341064, "ce_loss": 3.30534091591835, "logit_z_loss": 26.180189609527588, "router_aux_loss": 1.5193695276975632, "router_z_loss": 1.3259239941835403, "routing_entropy_normalized": 0.6660884613661867, "learning_rate": 0.003, "grad_norm": 1.7641860246658325, "tokens_per_second": 4200802.45281478, "eta_seconds": 3888.974609851837, "data_wait_seconds": 2.1159648895263672e-05} {"step": 20830, "total_steps": 28610, "tokens": 43683676160, "source": "fineweb_hq", "phase": 4, "loss": 3.448689341545105, "ce_loss": 3.433110922574997, "logit_z_loss": 26.36620831489563, "router_aux_loss": 1.4191585928201675, "router_z_loss": 1.386765867471695, "routing_entropy_normalized": 0.6702583059289625, "learning_rate": 0.003, "grad_norm": 1.9721308946609497, "tokens_per_second": 4250671.274801783, "eta_seconds": 3838.4155125617986, "data_wait_seconds": 2.059340476989746e-05} {"step": 20840, "total_steps": 28610, "tokens": 43704647680, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7890985310077667, "ce_loss": 2.7738339006900787, "logit_z_loss": 27.488242149353027, "router_aux_loss": 1.3536307066679, "router_z_loss": 1.7283058613538742, "routing_entropy_normalized": 0.6958525407733391, "learning_rate": 0.003, "grad_norm": 1.5453697443008423, "tokens_per_second": 3595211.6595977116, "eta_seconds": 4532.381562709808, "data_wait_seconds": 2.2679567337036133e-05} {"step": 20850, "total_steps": 28610, "tokens": 43725619200, "source": "fineweb_edu", "phase": 4, "loss": 3.372099131345749, "ce_loss": 3.3561656773090363, "logit_z_loss": 27.86877131462097, "router_aux_loss": 1.449877917766571, "router_z_loss": 1.4346849769353867, "routing_entropy_normalized": 0.6555254186890441, "learning_rate": 0.003, "grad_norm": 2.3662333488464355, "tokens_per_second": 4076564.3648957387, "eta_seconds": 3992.062448501587, "data_wait_seconds": 2.1755695343017578e-05} {"step": 20860, "total_steps": 28610, "tokens": 43746590720, "source": "finemath", "phase": 4, "loss": 2.283385843038559, "ce_loss": 2.267411768436432, "logit_z_loss": 36.50014352798462, "router_aux_loss": 1.4950562417507172, "router_z_loss": 1.0235345214605331, "routing_entropy_normalized": 0.6596396894962818, "learning_rate": 0.003, "grad_norm": 1.0766907930374146, "tokens_per_second": 4123880.6179229384, "eta_seconds": 3941.173255443573, "data_wait_seconds": 2.372264862060547e-05} {"step": 20870, "total_steps": 28610, "tokens": 43767562240, "source": "finemath", "phase": 4, "loss": 2.296667128801346, "ce_loss": 2.27983558177948, "logit_z_loss": 36.92152214050293, "router_aux_loss": 1.583822876214981, "router_z_loss": 0.9933327957987785, "routing_entropy_normalized": 0.6478411289929167, "learning_rate": 0.003, "grad_norm": 0.9226915836334229, "tokens_per_second": 3868744.0943924873, "eta_seconds": 4195.665591716766, "data_wait_seconds": 2.1010637283325195e-05} {"step": 20880, "total_steps": 28610, "tokens": 43788533760, "source": "dclm", "phase": 4, "loss": 3.5179053246974945, "ce_loss": 3.5023134350776672, "logit_z_loss": 26.541568279266357, "router_aux_loss": 1.4081847816705704, "router_z_loss": 1.5100616663694382, "routing_entropy_normalized": 0.6705281525986972, "learning_rate": 0.003, "grad_norm": 0.7742668986320496, "tokens_per_second": 3159211.2579536596, "eta_seconds": 5131.339323759079, "data_wait_seconds": 2.5719404220581055e-05} {"step": 20890, "total_steps": 28610, "tokens": 43809505280, "source": "fineweb_edu", "phase": 4, "loss": 3.3442744612693787, "ce_loss": 3.3287799954414368, "logit_z_loss": 26.731693983078003, "router_aux_loss": 1.398403137922287, "router_z_loss": 1.5104084461927414, "routing_entropy_normalized": 0.6725851788424048, "learning_rate": 0.003, "grad_norm": 1.0805532932281494, "tokens_per_second": 4158337.286946207, "eta_seconds": 3893.3863039016724, "data_wait_seconds": 2.2113323211669922e-05} {"step": 20900, "total_steps": 28610, "tokens": 43830476800, "source": "fineweb_edu", "phase": 4, "loss": 3.329676389694214, "ce_loss": 3.314733564853668, "logit_z_loss": 26.517700910568237, "router_aux_loss": 1.3458627611398697, "router_z_loss": 1.484215036034584, "routing_entropy_normalized": 0.6501185885187171, "learning_rate": 0.003, "grad_norm": 1.598776936531067, "tokens_per_second": 2978274.980857946, "eta_seconds": 5428.995651483536, "data_wait_seconds": 0.20637845993041992} {"step": 20910, "total_steps": 28610, "tokens": 43851448320, "source": "finemath", "phase": 4, "loss": 2.267310380935669, "ce_loss": 2.252462238073349, "logit_z_loss": 36.39936113357544, "router_aux_loss": 1.3840484470129013, "router_z_loss": 1.0076740384101868, "routing_entropy_normalized": 0.6603328151122321, "learning_rate": 0.003, "grad_norm": 1.3337862491607666, "tokens_per_second": 3951806.0080060516, "eta_seconds": 4086.2507843971252, "data_wait_seconds": 2.1576881408691406e-05} {"step": 20920, "total_steps": 28610, "tokens": 43872419840, "source": "fineweb_edu", "phase": 4, "loss": 3.3572027683258057, "ce_loss": 3.3423305451869965, "logit_z_loss": 26.59055781364441, "router_aux_loss": 1.3519002050161362, "router_z_loss": 1.3531763404607773, "routing_entropy_normalized": 0.6593217556048695, "learning_rate": 0.003, "grad_norm": 1.1173415184020996, "tokens_per_second": 4009097.8678114684, "eta_seconds": 4022.625391483307, "data_wait_seconds": 2.1398067474365234e-05} {"step": 20930, "total_steps": 28610, "tokens": 43893391360, "source": "fineweb_hq", "phase": 4, "loss": 3.397781491279602, "ce_loss": 3.382787376642227, "logit_z_loss": 26.696187734603882, "router_aux_loss": 1.3796714544296265, "router_z_loss": 1.1974229961633682, "routing_entropy_normalized": 0.6516564460417046, "learning_rate": 0.003, "grad_norm": 1.6153173446655273, "tokens_per_second": 3425982.680979146, "eta_seconds": 4701.170104980469, "data_wait_seconds": 0.12432676553726196} {"step": 20940, "total_steps": 28610, "tokens": 43914362880, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7797830998897552, "ce_loss": 2.764738142490387, "logit_z_loss": 28.792184114456177, "router_aux_loss": 1.3449207097291946, "router_z_loss": 1.595713585615158, "routing_entropy_normalized": 0.6550136147859114, "learning_rate": 0.003, "grad_norm": 1.0871435403823853, "tokens_per_second": 3968644.709625152, "eta_seconds": 4053.060179710388, "data_wait_seconds": 2.0444393157958984e-05} {"step": 20950, "total_steps": 28610, "tokens": 43935334400, "source": "fineweb_edu", "phase": 4, "loss": 3.3693994879722595, "ce_loss": 3.3538933992385864, "logit_z_loss": 27.339869022369385, "router_aux_loss": 1.4287130236625671, "router_z_loss": 1.218980833888054, "routing_entropy_normalized": 0.6305637797891336, "learning_rate": 0.003, "grad_norm": 1.3569233417510986, "tokens_per_second": 3816209.635509846, "eta_seconds": 4209.460657119751, "data_wait_seconds": 2.276897430419922e-05} {"step": 20960, "total_steps": 28610, "tokens": 43956305920, "source": "finemath", "phase": 4, "loss": 2.317759156227112, "ce_loss": 2.302378088235855, "logit_z_loss": 35.00167894363403, "router_aux_loss": 1.4474722146987915, "router_z_loss": 0.9063519164919853, "routing_entropy_normalized": 0.656609662520702, "learning_rate": 0.003, "grad_norm": 1.2192367315292358, "tokens_per_second": 3731915.7784791146, "eta_seconds": 4298.921452760696, "data_wait_seconds": 2.205371856689453e-05} {"step": 20970, "total_steps": 28610, "tokens": 43977277440, "source": "finemath", "phase": 4, "loss": 2.3037333488464355, "ce_loss": 2.2882856130599976, "logit_z_loss": 35.61844205856323, "router_aux_loss": 1.4466859549283981, "router_z_loss": 0.9808406680822372, "routing_entropy_normalized": 0.6552044900735032, "learning_rate": 0.003, "grad_norm": 1.1476048231124878, "tokens_per_second": 3695499.2195285656, "eta_seconds": 4335.609434127808, "data_wait_seconds": 2.1129846572875977e-05} {"step": 20980, "total_steps": 28610, "tokens": 43998248960, "source": "dclm", "phase": 4, "loss": 3.507085859775543, "ce_loss": 3.491395026445389, "logit_z_loss": 26.817963361740112, "router_aux_loss": 1.4348593950271606, "router_z_loss": 1.3422837257385254, "routing_entropy_normalized": 0.6563803596178871, "learning_rate": 0.003, "grad_norm": 1.8745195865631104, "tokens_per_second": 4267649.132022425, "eta_seconds": 3749.4342353343964, "data_wait_seconds": 1.9729137420654297e-05} {"step": 20990, "total_steps": 28610, "tokens": 44019220480, "source": "fineweb_edu", "phase": 4, "loss": 3.364744931459427, "ce_loss": 3.3486208617687225, "logit_z_loss": 25.735256910324097, "router_aux_loss": 1.4602581560611725, "router_z_loss": 1.5214735716581345, "routing_entropy_normalized": 0.6648189045628299, "learning_rate": 0.003, "grad_norm": 2.044344186782837, "tokens_per_second": 3763733.2545199837, "eta_seconds": 4245.863656997681, "data_wait_seconds": 2.1189451217651367e-05} {"step": 21000, "total_steps": 28610, "tokens": 44040192000, "source": "fineweb_edu", "phase": 4, "loss": 3.355589419603348, "ce_loss": 3.3389779031276703, "logit_z_loss": 25.503954887390137, "router_aux_loss": 1.4990402609109879, "router_z_loss": 1.6211149543523788, "routing_entropy_normalized": 0.6648774928974144, "learning_rate": 0.003, "grad_norm": 1.5337992906570435, "tokens_per_second": 4318233.4656059, "eta_seconds": 3695.799879074096, "data_wait_seconds": 2.4050474166870117e-05} {"step": 21010, "total_steps": 28610, "tokens": 44061163520, "source": "finemath", "phase": 4, "loss": 2.2991466224193573, "ce_loss": 2.2831107079982758, "logit_z_loss": 36.514914989471436, "router_aux_loss": 1.486060157418251, "router_z_loss": 1.1752996444702148, "routing_entropy_normalized": 0.6364819171675528, "learning_rate": 0.003, "grad_norm": 1.1649401187896729, "tokens_per_second": 4150009.960771592, "eta_seconds": 3840.5583000183105, "data_wait_seconds": 1.990795135498047e-05} {"step": 21020, "total_steps": 28610, "tokens": 44082135040, "source": "fineweb_edu", "phase": 4, "loss": 3.28166002035141, "ce_loss": 3.2654497921466827, "logit_z_loss": 26.561645030975342, "router_aux_loss": 1.4495122730731964, "router_z_loss": 1.7150512784719467, "routing_entropy_normalized": 0.6514117698839353, "learning_rate": 0.003, "grad_norm": 1.4091707468032837, "tokens_per_second": 3658757.2827916606, "eta_seconds": 4350.489100456238, "data_wait_seconds": 2.3096799850463867e-05} {"step": 21030, "total_steps": 28610, "tokens": 44103106560, "source": "fineweb_hq", "phase": 4, "loss": 3.3999020159244537, "ce_loss": 3.3837459087371826, "logit_z_loss": 26.810158252716064, "router_aux_loss": 1.4444392025470734, "router_z_loss": 1.711693525314331, "routing_entropy_normalized": 0.6420704372042046, "learning_rate": 0.003, "grad_norm": 2.4727869033813477, "tokens_per_second": 4311426.292181721, "eta_seconds": 3687.0425429344177, "data_wait_seconds": 2.1129846572875977e-05} {"step": 21040, "total_steps": 28610, "tokens": 44124078080, "source": "cosmopedia_v2", "phase": 4, "loss": 2.776246190071106, "ce_loss": 2.7597376108169556, "logit_z_loss": 27.640305995941162, "router_aux_loss": 1.4507737755775452, "router_z_loss": 2.0008435547351837, "routing_entropy_normalized": 0.6358241201228774, "learning_rate": 0.003, "grad_norm": 1.4391908645629883, "tokens_per_second": 4013626.575424399, "eta_seconds": 3955.385570049286, "data_wait_seconds": 2.232193946838379e-05} {"step": 21050, "total_steps": 28610, "tokens": 44145049600, "source": "fineweb_edu", "phase": 4, "loss": 3.320689857006073, "ce_loss": 3.304673820734024, "logit_z_loss": 25.776959896087646, "router_aux_loss": 1.4295868426561356, "router_z_loss": 1.7202135771512985, "routing_entropy_normalized": 0.6627840155728459, "learning_rate": 0.003, "grad_norm": 1.0792344808578491, "tokens_per_second": 3890287.02646694, "eta_seconds": 4075.398296356201, "data_wait_seconds": 2.1219253540039062e-05} {"step": 21060, "total_steps": 28610, "tokens": 44166021120, "source": "finemath", "phase": 4, "loss": 2.2422442734241486, "ce_loss": 2.2267903238534927, "logit_z_loss": 36.8570818901062, "router_aux_loss": 1.4250609278678894, "router_z_loss": 1.203334465622902, "routing_entropy_normalized": 0.6900639965445815, "learning_rate": 0.003, "grad_norm": 0.9260519742965698, "tokens_per_second": 3923584.8545909226, "eta_seconds": 4035.4671013355255, "data_wait_seconds": 1.8656253814697266e-05} {"step": 21070, "total_steps": 28610, "tokens": 44186992640, "source": "finemath", "phase": 4, "loss": 2.257114678621292, "ce_loss": 2.242128401994705, "logit_z_loss": 35.50554084777832, "router_aux_loss": 1.3790500909090042, "router_z_loss": 1.1958265006542206, "routing_entropy_normalized": 0.667014266172626, "learning_rate": 0.003, "grad_norm": 1.3027582168579102, "tokens_per_second": 4189244.9095487962, "eta_seconds": 3774.552794456482, "data_wait_seconds": 2.2292137145996094e-05} {"step": 21080, "total_steps": 28610, "tokens": 44207964160, "source": "dclm", "phase": 4, "loss": 3.4171304404735565, "ce_loss": 3.402365982532501, "logit_z_loss": 27.195233583450317, "router_aux_loss": 1.3357426524162292, "router_z_loss": 1.407062128186226, "routing_entropy_normalized": 0.6689588336789501, "learning_rate": 0.003, "grad_norm": 0.9313994646072388, "tokens_per_second": 4108913.37123621, "eta_seconds": 3843.2434887886047, "data_wait_seconds": 4.1157007217407227e-05} {"step": 21090, "total_steps": 28610, "tokens": 44228935680, "source": "fineweb_edu", "phase": 4, "loss": 3.3376858830451965, "ce_loss": 3.322397470474243, "logit_z_loss": 26.501919746398926, "router_aux_loss": 1.368201956152916, "router_z_loss": 1.606416940689087, "routing_entropy_normalized": 0.6661059143746509, "learning_rate": 0.003, "grad_norm": 0.9508428573608398, "tokens_per_second": 4024976.47303864, "eta_seconds": 3918.1801795959473, "data_wait_seconds": 2.2470951080322266e-05} {"step": 21100, "total_steps": 28610, "tokens": 44249907200, "source": "fineweb_edu", "phase": 4, "loss": 3.3246273696422577, "ce_loss": 3.3098182380199432, "logit_z_loss": 27.577852487564087, "router_aux_loss": 1.3273069709539413, "router_z_loss": 1.5360310226678848, "routing_entropy_normalized": 0.6581101224941769, "learning_rate": 0.003, "grad_norm": 1.6266200542449951, "tokens_per_second": 3820369.4080110076, "eta_seconds": 4122.536288499832, "data_wait_seconds": 0.0470297634601593} {"step": 21110, "total_steps": 28610, "tokens": 44270878720, "source": "finemath", "phase": 4, "loss": 2.28605455160141, "ce_loss": 2.271335870027542, "logit_z_loss": 36.24536943435669, "router_aux_loss": 1.3804783374071121, "router_z_loss": 0.9139332473278046, "routing_entropy_normalized": 0.6624323470364335, "learning_rate": 0.003, "grad_norm": 0.8583188652992249, "tokens_per_second": 4342938.529805446, "eta_seconds": 3621.6584444046016, "data_wait_seconds": 2.1636486053466797e-05} {"step": 21120, "total_steps": 28610, "tokens": 44291850240, "source": "fineweb_edu", "phase": 4, "loss": 3.322672188282013, "ce_loss": 3.3072262704372406, "logit_z_loss": 26.716693878173828, "router_aux_loss": 1.4086552113294601, "router_z_loss": 1.3592849671840668, "routing_entropy_normalized": 0.6342286201663345, "learning_rate": 0.003, "grad_norm": 1.9429537057876587, "tokens_per_second": 3605845.943463344, "eta_seconds": 4356.1673810482025, "data_wait_seconds": 2.256035804748535e-05} {"step": 21130, "total_steps": 28610, "tokens": 44312821760, "source": "fineweb_hq", "phase": 4, "loss": 3.31985279917717, "ce_loss": 3.3050413727760315, "logit_z_loss": 27.47545313835144, "router_aux_loss": 1.3487685918807983, "router_z_loss": 1.3237555027008057, "routing_entropy_normalized": 0.661741882950748, "learning_rate": 0.003, "grad_norm": 1.6357877254486084, "tokens_per_second": 4272538.55478783, "eta_seconds": 3671.516771316528, "data_wait_seconds": 2.1338462829589844e-05} {"step": 21140, "total_steps": 28610, "tokens": 44333793280, "source": "cosmopedia_v2", "phase": 4, "loss": 2.744672566652298, "ce_loss": 2.729648858308792, "logit_z_loss": 27.35302972793579, "router_aux_loss": 1.3425530940294266, "router_z_loss": 1.5981078147888184, "routing_entropy_normalized": 0.6896806873610248, "learning_rate": 0.003, "grad_norm": 0.7887178659439087, "tokens_per_second": 4364526.077920356, "eta_seconds": 3589.330241203308, "data_wait_seconds": 2.2172927856445312e-05} {"step": 21150, "total_steps": 28610, "tokens": 44354764800, "source": "fineweb_edu", "phase": 4, "loss": 3.3181621730327606, "ce_loss": 3.302597612142563, "logit_z_loss": 25.842087745666504, "router_aux_loss": 1.4250980913639069, "router_z_loss": 1.3135451823472977, "routing_entropy_normalized": 0.6474621549607052, "learning_rate": 0.003, "grad_norm": 1.2576359510421753, "tokens_per_second": 3784798.45589429, "eta_seconds": 4133.576490879059, "data_wait_seconds": 2.1457672119140625e-05} {"step": 21160, "total_steps": 28610, "tokens": 44375736320, "source": "finemath", "phase": 4, "loss": 2.2726654410362244, "ce_loss": 2.257661283016205, "logit_z_loss": 35.280097007751465, "router_aux_loss": 1.419774368405342, "router_z_loss": 0.8064021989703178, "routing_entropy_normalized": 0.6662496089433606, "learning_rate": 0.003, "grad_norm": 0.6737216711044312, "tokens_per_second": 4071754.6221670625, "eta_seconds": 3837.1129524707794, "data_wait_seconds": 2.0444393157958984e-05} {"step": 21170, "total_steps": 28610, "tokens": 44396707840, "source": "finemath", "phase": 4, "loss": 2.372421622276306, "ce_loss": 2.357626974582672, "logit_z_loss": 35.548715114593506, "router_aux_loss": 1.3596792817115784, "router_z_loss": 1.1978675201535225, "routing_entropy_normalized": 0.6873095256259134, "learning_rate": 0.003, "grad_norm": 0.7991756200790405, "tokens_per_second": 3419899.640540312, "eta_seconds": 4562.359285354614, "data_wait_seconds": 2.16066837310791e-05} {"step": 21180, "total_steps": 28610, "tokens": 44417679360, "source": "dclm", "phase": 4, "loss": 3.4517778158187866, "ce_loss": 3.435712993144989, "logit_z_loss": 28.07727026939392, "router_aux_loss": 1.4898554980754852, "router_z_loss": 1.1663099527359009, "routing_entropy_normalized": 0.6470426257258011, "learning_rate": 0.003, "grad_norm": 1.5727813243865967, "tokens_per_second": 3914371.3280441226, "eta_seconds": 3980.6748144626617, "data_wait_seconds": 2.0891427993774414e-05} {"step": 21190, "total_steps": 28610, "tokens": 44438650880, "source": "fineweb_edu", "phase": 4, "loss": 3.3401081562042236, "ce_loss": 3.323218196630478, "logit_z_loss": 27.24994421005249, "router_aux_loss": 1.5543179661035538, "router_z_loss": 1.3468118607997894, "routing_entropy_normalized": 0.6542581273752391, "learning_rate": 0.003, "grad_norm": 1.0569924116134644, "tokens_per_second": 4045431.900803527, "eta_seconds": 3846.528188228607, "data_wait_seconds": 2.065300941467285e-05} {"step": 21200, "total_steps": 28610, "tokens": 44459622400, "source": "fineweb_edu", "phase": 4, "loss": 3.3209910690784454, "ce_loss": 3.3045525550842285, "logit_z_loss": 26.76595139503479, "router_aux_loss": 1.4900516420602798, "router_z_loss": 1.538000911474228, "routing_entropy_normalized": 0.6444235902999373, "learning_rate": 0.003, "grad_norm": 1.761268973350525, "tokens_per_second": 4252030.713398227, "eta_seconds": 3654.699922800064, "data_wait_seconds": 2.1666288375854492e-05} {"step": 21210, "total_steps": 28610, "tokens": 44480593920, "source": "finemath", "phase": 4, "loss": 2.23327299952507, "ce_loss": 2.217718005180359, "logit_z_loss": 36.129021644592285, "router_aux_loss": 1.4372685253620148, "router_z_loss": 1.1823571175336838, "routing_entropy_normalized": 0.6600454272572602, "learning_rate": 0.003, "grad_norm": 0.9028651118278503, "tokens_per_second": 3880312.3442864683, "eta_seconds": 3999.400930404663, "data_wait_seconds": 2.1964311599731445e-05} {"step": 21220, "total_steps": 28610, "tokens": 44501565440, "source": "fineweb_edu", "phase": 4, "loss": 3.3074635565280914, "ce_loss": 3.291659027338028, "logit_z_loss": 26.969858407974243, "router_aux_loss": 1.420432135462761, "router_z_loss": 1.6002096235752106, "routing_entropy_normalized": 0.6324155762757906, "learning_rate": 0.003, "grad_norm": 1.3413934707641602, "tokens_per_second": 3736410.5844510975, "eta_seconds": 4147.818589448929, "data_wait_seconds": 2.0503997802734375e-05} {"step": 21230, "total_steps": 28610, "tokens": 44522536960, "source": "fineweb_hq", "phase": 4, "loss": 3.3816082775592804, "ce_loss": 3.366341531276703, "logit_z_loss": 27.296469926834106, "router_aux_loss": 1.3777854591608047, "router_z_loss": 1.488887995481491, "routing_entropy_normalized": 0.6643722704714803, "learning_rate": 0.003, "grad_norm": 2.09798526763916, "tokens_per_second": 4197839.978470787, "eta_seconds": 3686.891791820526, "data_wait_seconds": 1.9997358322143555e-05} {"step": 21240, "total_steps": 28610, "tokens": 44543508480, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7636495530605316, "ce_loss": 2.7480922639369965, "logit_z_loss": 28.85787034034729, "router_aux_loss": 1.3680136948823929, "router_z_loss": 1.877185195684433, "routing_entropy_normalized": 0.6707624831529345, "learning_rate": 0.003, "grad_norm": 1.144620656967163, "tokens_per_second": 3666400.0630434076, "eta_seconds": 4215.582035303116, "data_wait_seconds": 2.256035804748535e-05} {"step": 21250, "total_steps": 28610, "tokens": 44564480000, "source": "fineweb_edu", "phase": 4, "loss": 3.3393180668354034, "ce_loss": 3.324686735868454, "logit_z_loss": 26.021703958511353, "router_aux_loss": 1.3180672973394394, "router_z_loss": 1.4506395608186722, "routing_entropy_normalized": 0.6672068076999886, "learning_rate": 0.003, "grad_norm": 0.9057680368423462, "tokens_per_second": 4231319.612833539, "eta_seconds": 3647.807334899903, "data_wait_seconds": 2.35140323638916e-05} {"step": 21260, "total_steps": 28610, "tokens": 44585451520, "source": "finemath", "phase": 4, "loss": 2.099619746208191, "ce_loss": 2.0852229595184326, "logit_z_loss": 37.25911045074463, "router_aux_loss": 1.3520826250314713, "router_z_loss": 0.8759547472000122, "routing_entropy_normalized": 0.6552519202967297, "learning_rate": 0.003, "grad_norm": 0.8415202498435974, "tokens_per_second": 4283142.64340926, "eta_seconds": 3598.7751245498657, "data_wait_seconds": 2.2739171981811523e-05} {"step": 21270, "total_steps": 28610, "tokens": 44606423040, "source": "finemath", "phase": 4, "loss": 2.253918409347534, "ce_loss": 2.2399988174438477, "logit_z_loss": 37.16066360473633, "router_aux_loss": 1.3094950169324875, "router_z_loss": 0.8246335610747337, "routing_entropy_normalized": 0.6825181378906433, "learning_rate": 0.003, "grad_norm": 0.5426797866821289, "tokens_per_second": 3783317.5620242893, "eta_seconds": 4068.6766119003296, "data_wait_seconds": 2.1070241928100586e-05} {"step": 21280, "total_steps": 28610, "tokens": 44627394560, "source": "dclm", "phase": 4, "loss": 3.456135392189026, "ce_loss": 3.441948264837265, "logit_z_loss": 28.222652435302734, "router_aux_loss": 1.3226585686206818, "router_z_loss": 0.9605288654565811, "routing_entropy_normalized": 0.6663695392857157, "learning_rate": 0.003, "grad_norm": 1.1766194105148315, "tokens_per_second": 3988317.1212743907, "eta_seconds": 3854.288335800171, "data_wait_seconds": 2.0295381546020508e-05} {"step": 21290, "total_steps": 28610, "tokens": 44648366080, "source": "fineweb_edu", "phase": 4, "loss": 3.3631374835968018, "ce_loss": 3.3489194810390472, "logit_z_loss": 26.33583402633667, "router_aux_loss": 1.312051311135292, "router_z_loss": 1.097473680973053, "routing_entropy_normalized": 0.6570744677869285, "learning_rate": 0.003, "grad_norm": 0.5163988471031189, "tokens_per_second": 3885480.8687624247, "eta_seconds": 3950.9016151428223, "data_wait_seconds": 2.1278858184814453e-05} {"step": 21300, "total_steps": 28610, "tokens": 44669337600, "source": "fineweb_edu", "phase": 4, "loss": 3.3120367228984833, "ce_loss": 3.296749621629715, "logit_z_loss": 26.186087369918823, "router_aux_loss": 1.4210515320301056, "router_z_loss": 1.0765540599822998, "routing_entropy_normalized": 0.6455515247880345, "learning_rate": 0.003, "grad_norm": 1.3660004138946533, "tokens_per_second": 3013955.4399857265, "eta_seconds": 5086.3993928432465, "data_wait_seconds": 1.9043684005737305e-05} {"step": 21310, "total_steps": 28610, "tokens": 44690309120, "source": "finemath", "phase": 4, "loss": 2.206761509180069, "ce_loss": 2.1902028918266296, "logit_z_loss": 34.69642496109009, "router_aux_loss": 1.5452508330345154, "router_z_loss": 1.1060958504676819, "routing_entropy_normalized": 0.6579898456435469, "learning_rate": 0.003, "grad_norm": 1.2860726118087769, "tokens_per_second": 4126869.3097311687, "eta_seconds": 3709.6424555778503, "data_wait_seconds": 1.8715858459472656e-05} {"step": 21320, "total_steps": 28610, "tokens": 44711280640, "source": "fineweb_edu", "phase": 4, "loss": 3.3517285585403442, "ce_loss": 3.336342304944992, "logit_z_loss": 25.638649463653564, "router_aux_loss": 1.408172756433487, "router_z_loss": 1.3045740574598312, "routing_entropy_normalized": 0.6584281741584816, "learning_rate": 0.003, "grad_norm": 1.4562766551971436, "tokens_per_second": 3653511.170362679, "eta_seconds": 4184.533005952835, "data_wait_seconds": 2.2470951080322266e-05} {"step": 21330, "total_steps": 28610, "tokens": 44732252160, "source": "fineweb_hq", "phase": 4, "loss": 3.3845187425613403, "ce_loss": 3.369395673274994, "logit_z_loss": 27.343998193740845, "router_aux_loss": 1.3799024820327759, "router_z_loss": 1.3240236788988113, "routing_entropy_normalized": 0.6620700052390528, "learning_rate": 0.003, "grad_norm": 1.453593134880066, "tokens_per_second": 4180476.9334824528, "eta_seconds": 3652.039421081543, "data_wait_seconds": 2.1159648895263672e-05} {"step": 21340, "total_steps": 28610, "tokens": 44753223680, "source": "cosmopedia_v2", "phase": 4, "loss": 2.771363228559494, "ce_loss": 2.7565464079380035, "logit_z_loss": 28.84044122695923, "router_aux_loss": 1.3184256851673126, "router_z_loss": 1.6325461566448212, "routing_entropy_normalized": 0.6919409938318425, "learning_rate": 0.003, "grad_norm": 1.190600872039795, "tokens_per_second": 3352625.5999560305, "eta_seconds": 4547.568639993668, "data_wait_seconds": 2.002716064453125e-05} {"step": 21350, "total_steps": 28610, "tokens": 44774195200, "source": "fineweb_edu", "phase": 4, "loss": 3.3322918117046356, "ce_loss": 3.317434012889862, "logit_z_loss": 25.847595930099487, "router_aux_loss": 1.3516015857458115, "router_z_loss": 1.3418029993772507, "routing_entropy_normalized": 0.6863550515807672, "learning_rate": 0.003, "grad_norm": 1.1559346914291382, "tokens_per_second": 4429653.640205641, "eta_seconds": 3437.136344432831, "data_wait_seconds": 2.148747444152832e-05} {"step": 21360, "total_steps": 28610, "tokens": 44795166720, "source": "finemath", "phase": 4, "loss": 2.1855394542217255, "ce_loss": 2.1698587238788605, "logit_z_loss": 37.11665344238281, "router_aux_loss": 1.4840517342090607, "router_z_loss": 0.8402492180466652, "routing_entropy_normalized": 0.6360375549178712, "learning_rate": 0.003, "grad_norm": 0.7609403133392334, "tokens_per_second": 3793029.4817420775, "eta_seconds": 4008.498239517212, "data_wait_seconds": 6.73532485961914e-05} {"step": 21370, "total_steps": 28610, "tokens": 44816138240, "source": "finemath", "phase": 4, "loss": 2.236057758331299, "ce_loss": 2.221796751022339, "logit_z_loss": 36.52192544937134, "router_aux_loss": 1.3457974344491959, "router_z_loss": 0.8029950261116028, "routing_entropy_normalized": 0.657871722762009, "learning_rate": 0.003, "grad_norm": 1.0531574487686157, "tokens_per_second": 4373653.769301849, "eta_seconds": 3471.555198669434, "data_wait_seconds": 1.9997358322143555e-05} {"step": 21380, "total_steps": 28610, "tokens": 44837109760, "source": "dclm", "phase": 4, "loss": 3.450653612613678, "ce_loss": 3.436039000749588, "logit_z_loss": 27.378748893737793, "router_aux_loss": 1.3562992960214615, "router_z_loss": 1.0516179352998734, "routing_entropy_normalized": 0.6600913572803779, "learning_rate": 0.003, "grad_norm": 0.6217620968818665, "tokens_per_second": 3989385.255299134, "eta_seconds": 3800.6880733966827, "data_wait_seconds": 2.2470951080322266e-05} {"step": 21390, "total_steps": 28610, "tokens": 44858081280, "source": "fineweb_edu", "phase": 4, "loss": 3.3436236679553986, "ce_loss": 3.3278621435165405, "logit_z_loss": 26.91500210762024, "router_aux_loss": 1.4572822749614716, "router_z_loss": 1.1886632293462753, "routing_entropy_normalized": 0.6433889104859388, "learning_rate": 0.003, "grad_norm": 0.5568232536315918, "tokens_per_second": 4524342.863486805, "eta_seconds": 3346.66003370285, "data_wait_seconds": 2.1964311599731445e-05} {"step": 21400, "total_steps": 28610, "tokens": 44879052800, "source": "fineweb_edu", "phase": 4, "loss": 3.403102934360504, "ce_loss": 3.3877746164798737, "logit_z_loss": 25.736722946166992, "router_aux_loss": 1.4074045270681381, "router_z_loss": 1.2542962431907654, "routing_entropy_normalized": 0.6450492166775657, "learning_rate": 0.003, "grad_norm": 1.7111523151397705, "tokens_per_second": 3503348.008741709, "eta_seconds": 4316.004542589188, "data_wait_seconds": 2.1129846572875977e-05} {"step": 21410, "total_steps": 28610, "tokens": 44900024320, "source": "finemath", "phase": 4, "loss": 2.3300399482250214, "ce_loss": 2.3152168691158295, "logit_z_loss": 36.81240892410278, "router_aux_loss": 1.3927943259477615, "router_z_loss": 0.895158126950264, "routing_entropy_normalized": 0.6781038878491153, "learning_rate": 0.003, "grad_norm": 0.8662295341491699, "tokens_per_second": 3047714.846884266, "eta_seconds": 4954.365863800049, "data_wait_seconds": 1.7881393432617188e-05} {"step": 21420, "total_steps": 28610, "tokens": 44920995840, "source": "fineweb_edu", "phase": 4, "loss": 3.3495951294898987, "ce_loss": 3.334572732448578, "logit_z_loss": 25.29185390472412, "router_aux_loss": 1.3785770237445831, "router_z_loss": 1.236632600426674, "routing_entropy_normalized": 0.6609591488221551, "learning_rate": 0.003, "grad_norm": 1.3429584503173828, "tokens_per_second": 4046256.103686102, "eta_seconds": 3726.536950111389, "data_wait_seconds": 1.940131187438965e-05} {"step": 21430, "total_steps": 28610, "tokens": 44941967360, "source": "fineweb_hq", "phase": 4, "loss": 3.399146556854248, "ce_loss": 3.3845901787281036, "logit_z_loss": 25.85204029083252, "router_aux_loss": 1.3378234058618546, "router_z_loss": 1.178148314356804, "routing_entropy_normalized": 0.684326932513985, "learning_rate": 0.003, "grad_norm": 1.7902659177780151, "tokens_per_second": 3207098.032908795, "eta_seconds": 4695.070498466492, "data_wait_seconds": 1.8656253814697266e-05} {"step": 21440, "total_steps": 28610, "tokens": 44962938880, "source": "cosmopedia_v2", "phase": 4, "loss": 2.773603677749634, "ce_loss": 2.7586362659931183, "logit_z_loss": 28.472111463546753, "router_aux_loss": 1.3459088206291199, "router_z_loss": 1.5082672387361526, "routing_entropy_normalized": 0.6869471472267317, "learning_rate": 0.003, "grad_norm": 1.3321740627288818, "tokens_per_second": 4439155.532873956, "eta_seconds": 3387.2613222599034, "data_wait_seconds": 2.014636993408203e-05} {"step": 21450, "total_steps": 28610, "tokens": 44983910400, "source": "fineweb_edu", "phase": 4, "loss": 3.3731155693531036, "ce_loss": 3.358381360769272, "logit_z_loss": 26.231740713119507, "router_aux_loss": 1.351000890135765, "router_z_loss": 1.2242086976766586, "routing_entropy_normalized": 0.6442292277905989, "learning_rate": 0.003, "grad_norm": 1.2178367376327515, "tokens_per_second": 3202302.5688879136, "eta_seconds": 4689.003614425659, "data_wait_seconds": 1.9550323486328125e-05} {"step": 21460, "total_steps": 28610, "tokens": 45004881920, "source": "finemath", "phase": 4, "loss": 2.2839199900627136, "ce_loss": 2.2689729034900665, "logit_z_loss": 35.340980529785156, "router_aux_loss": 1.4131683707237244, "router_z_loss": 0.8153721317648888, "routing_entropy_normalized": 0.6505415051825882, "learning_rate": 0.003, "grad_norm": 0.9078136086463928, "tokens_per_second": 4369636.228753701, "eta_seconds": 3431.552654504776, "data_wait_seconds": 2.2709369659423828e-05} {"step": 21470, "total_steps": 28610, "tokens": 45025853440, "source": "finemath", "phase": 4, "loss": 2.2477413415908813, "ce_loss": 2.2333263158798218, "logit_z_loss": 37.274818420410156, "router_aux_loss": 1.3662805259227753, "router_z_loss": 0.7522211596369743, "routing_entropy_normalized": 0.683788305224189, "learning_rate": 0.003, "grad_norm": 0.9639618992805481, "tokens_per_second": 3528331.7802612223, "eta_seconds": 4243.837091445923, "data_wait_seconds": 2.1129846572875977e-05} {"step": 21480, "total_steps": 28610, "tokens": 45046824960, "source": "dclm", "phase": 4, "loss": 3.4696368277072906, "ce_loss": 3.455071359872818, "logit_z_loss": 27.977341413497925, "router_aux_loss": 1.3627442121505737, "router_z_loss": 0.9380551502108574, "routing_entropy_normalized": 0.6670240732316944, "learning_rate": 0.003, "grad_norm": 1.1660603284835815, "tokens_per_second": 3835352.7101512114, "eta_seconds": 3898.648935317993, "data_wait_seconds": 2.5480985641479492e-05} {"step": 21490, "total_steps": 28610, "tokens": 45067796480, "source": "fineweb_edu", "phase": 4, "loss": 3.3453944325447083, "ce_loss": 3.3306336402893066, "logit_z_loss": 25.771366357803345, "router_aux_loss": 1.3643244951963425, "router_z_loss": 1.1175173670053482, "routing_entropy_normalized": 0.6557145477633388, "learning_rate": 0.003, "grad_norm": 0.9098297357559204, "tokens_per_second": 4068809.6254398753, "eta_seconds": 3669.801149368286, "data_wait_seconds": 1.8209218978881836e-05} {"step": 21500, "total_steps": 28610, "tokens": 45088768000, "source": "fineweb_edu", "phase": 4, "loss": 3.409782111644745, "ce_loss": 3.3948977291584015, "logit_z_loss": 25.73407745361328, "router_aux_loss": 1.3748095333576202, "router_z_loss": 1.1362909078598022, "routing_entropy_normalized": 0.6734825075035592, "learning_rate": 0.003, "grad_norm": 1.6760562658309937, "tokens_per_second": 3723870.6398800313, "eta_seconds": 4004.1000781059265, "data_wait_seconds": 0.07788151502609253} {"step": 21510, "total_steps": 28610, "tokens": 45109739520, "source": "finemath", "phase": 4, "loss": 2.3098004758358, "ce_loss": 2.293615221977234, "logit_z_loss": 36.02497911453247, "router_aux_loss": 1.530042216181755, "router_z_loss": 0.8848110660910606, "routing_entropy_normalized": 0.6426349911759257, "learning_rate": 0.003, "grad_norm": 1.2036564350128174, "tokens_per_second": 4148947.4408104382, "eta_seconds": 3588.8088274002075, "data_wait_seconds": 2.2143125534057617e-05} {"step": 21520, "total_steps": 28610, "tokens": 45130711040, "source": "fineweb_edu", "phase": 4, "loss": 3.350679337978363, "ce_loss": 3.3339153230190277, "logit_z_loss": 27.71790623664856, "router_aux_loss": 1.5329145789146423, "router_z_loss": 1.4349063038825989, "routing_entropy_normalized": 0.6440659178454556, "learning_rate": 0.003, "grad_norm": 1.5624662637710571, "tokens_per_second": 4058781.864380533, "eta_seconds": 3663.367033958435, "data_wait_seconds": 2.059340476989746e-05} {"step": 21530, "total_steps": 28610, "tokens": 45151682560, "source": "fineweb_hq", "phase": 4, "loss": 3.374258279800415, "ce_loss": 3.3587220907211304, "logit_z_loss": 27.378307342529297, "router_aux_loss": 1.4114830791950226, "router_z_loss": 1.421367958188057, "routing_entropy_normalized": 0.668388613505812, "learning_rate": 0.003, "grad_norm": 1.255303144454956, "tokens_per_second": 3901410.597038135, "eta_seconds": 3805.76096534729, "data_wait_seconds": 2.294778823852539e-05} {"step": 21540, "total_steps": 28610, "tokens": 45172654080, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7735361456871033, "ce_loss": 2.7575093507766724, "logit_z_loss": 28.852569818496704, "router_aux_loss": 1.4222682118415833, "router_z_loss": 1.8041412979364395, "routing_entropy_normalized": 0.6725273650899385, "learning_rate": 0.003, "grad_norm": 2.2890236377716064, "tokens_per_second": 4312885.566572136, "eta_seconds": 3437.8061766624455, "data_wait_seconds": 2.008676528930664e-05} {"step": 21550, "total_steps": 28610, "tokens": 45193625600, "source": "fineweb_edu", "phase": 4, "loss": 3.38004732131958, "ce_loss": 3.3641267120838165, "logit_z_loss": 26.18130350112915, "router_aux_loss": 1.4327738285064697, "router_z_loss": 1.5928242057561874, "routing_entropy_normalized": 0.6586207034800525, "learning_rate": 0.003, "grad_norm": 0.7651135325431824, "tokens_per_second": 3847817.80040833, "eta_seconds": 3847.867515563965, "data_wait_seconds": 2.3692846298217773e-05} {"step": 21560, "total_steps": 28610, "tokens": 45214597120, "source": "finemath", "phase": 4, "loss": 2.2240301370620728, "ce_loss": 2.209236830472946, "logit_z_loss": 37.55930805206299, "router_aux_loss": 1.3685664236545563, "router_z_loss": 1.1076315194368362, "routing_entropy_normalized": 0.6511395292808645, "learning_rate": 0.003, "grad_norm": 0.8121545314788818, "tokens_per_second": 4295565.696826809, "eta_seconds": 3441.9032657146454, "data_wait_seconds": 2.1278858184814453e-05} {"step": 21570, "total_steps": 28610, "tokens": 45235568640, "source": "finemath", "phase": 4, "loss": 2.221086338162422, "ce_loss": 2.207485616207123, "logit_z_loss": 38.06940841674805, "router_aux_loss": 1.2615468353033066, "router_z_loss": 0.9852392747998238, "routing_entropy_normalized": 0.6993297861828148, "learning_rate": 0.003, "grad_norm": 0.7903021574020386, "tokens_per_second": 3357410.4326753956, "eta_seconds": 4397.421875, "data_wait_seconds": 2.205371856689453e-05} {"step": 21580, "total_steps": 28610, "tokens": 45256540160, "source": "dclm", "phase": 4, "loss": 3.5073107481002808, "ce_loss": 3.492433726787567, "logit_z_loss": 27.504796981811523, "router_aux_loss": 1.3709176182746887, "router_z_loss": 1.1678599566221237, "routing_entropy_normalized": 0.6683071877397164, "learning_rate": 0.003, "grad_norm": 2.742222785949707, "tokens_per_second": 4022255.615668852, "eta_seconds": 3665.3509793281555, "data_wait_seconds": 2.1249055862426758e-05} {"step": 21590, "total_steps": 28610, "tokens": 45277511680, "source": "fineweb_edu", "phase": 4, "loss": 3.383355349302292, "ce_loss": 3.368227183818817, "logit_z_loss": 26.55524778366089, "router_aux_loss": 1.375168651342392, "router_z_loss": 1.376481294631958, "routing_entropy_normalized": 0.6523751109303996, "learning_rate": 0.003, "grad_norm": 1.3524589538574219, "tokens_per_second": 4246307.678623755, "eta_seconds": 3467.01373386383, "data_wait_seconds": 2.1904706954956055e-05} {"step": 21600, "total_steps": 28610, "tokens": 45298483200, "source": "fineweb_edu", "phase": 4, "loss": 3.419263929128647, "ce_loss": 3.4041643738746643, "logit_z_loss": 27.73811149597168, "router_aux_loss": 1.3595781475305557, "router_z_loss": 1.5037843883037567, "routing_entropy_normalized": 0.6341867122339024, "learning_rate": 0.003, "grad_norm": 2.4984118938446045, "tokens_per_second": 3771616.953988432, "eta_seconds": 3897.8071472644806, "data_wait_seconds": 1.9937753677368164e-05} {"step": 21610, "total_steps": 28610, "tokens": 45319454720, "source": "finemath", "phase": 4, "loss": 2.2384788393974304, "ce_loss": 2.2230189442634583, "logit_z_loss": 38.99074459075928, "router_aux_loss": 1.453183189034462, "router_z_loss": 0.9281178638339043, "routing_entropy_normalized": 0.6434097989441205, "learning_rate": 0.003, "grad_norm": 1.1438134908676147, "tokens_per_second": 3664442.523917671, "eta_seconds": 4006.083846092224, "data_wait_seconds": 2.136826515197754e-05} {"step": 21620, "total_steps": 28610, "tokens": 45340426240, "source": "fineweb_edu", "phase": 4, "loss": 3.336433470249176, "ce_loss": 3.3212223947048187, "logit_z_loss": 25.970873832702637, "router_aux_loss": 1.3588607758283615, "router_z_loss": 1.6225114315748215, "routing_entropy_normalized": 0.6602220095427417, "learning_rate": 0.003, "grad_norm": 1.2863768339157104, "tokens_per_second": 3979716.7898648055, "eta_seconds": 3683.45117354393, "data_wait_seconds": 2.0951032638549805e-05} {"step": 21630, "total_steps": 28610, "tokens": 45361397760, "source": "fineweb_hq", "phase": 4, "loss": 3.3649876713752747, "ce_loss": 3.349510669708252, "logit_z_loss": 26.498575687408447, "router_aux_loss": 1.3923474997282028, "router_z_loss": 1.5535347014665604, "routing_entropy_normalized": 0.6581388107326308, "learning_rate": 0.003, "grad_norm": 1.6837337017059326, "tokens_per_second": 4072600.900540878, "eta_seconds": 3594.2930126190186, "data_wait_seconds": 2.014636993408203e-05} {"step": 21640, "total_steps": 28610, "tokens": 45382369280, "source": "cosmopedia_v2", "phase": 4, "loss": 2.78933846950531, "ce_loss": 2.7731462121009827, "logit_z_loss": 28.148863554000854, "router_aux_loss": 1.4319968670606613, "router_z_loss": 1.8723275363445282, "routing_entropy_normalized": 0.6417519469768419, "learning_rate": 0.003, "grad_norm": 1.1740074157714844, "tokens_per_second": 4107922.051465984, "eta_seconds": 3558.283058166504, "data_wait_seconds": 1.9729137420654297e-05} {"step": 21650, "total_steps": 28610, "tokens": 45403340800, "source": "fineweb_edu", "phase": 4, "loss": 3.353813260793686, "ce_loss": 3.338668644428253, "logit_z_loss": 26.459001302719116, "router_aux_loss": 1.3577120751142502, "router_z_loss": 1.567481592297554, "routing_entropy_normalized": 0.6464807490433693, "learning_rate": 0.003, "grad_norm": 0.8145849108695984, "tokens_per_second": 4036220.160269651, "eta_seconds": 3616.298749923706, "data_wait_seconds": 2.193450927734375e-05} {"step": 21660, "total_steps": 28610, "tokens": 45424312320, "source": "finemath", "phase": 4, "loss": 2.217109799385071, "ce_loss": 2.2021842300891876, "logit_z_loss": 38.03091049194336, "router_aux_loss": 1.3868203461170197, "router_z_loss": 1.057308129966259, "routing_entropy_normalized": 0.6420361888557962, "learning_rate": 0.003, "grad_norm": 1.2007304430007935, "tokens_per_second": 3767875.5434556836, "eta_seconds": 3868.282333612442, "data_wait_seconds": 2.086162567138672e-05} {"step": 21670, "total_steps": 28610, "tokens": 45445283840, "source": "finemath", "phase": 4, "loss": 2.2652394473552704, "ce_loss": 2.250266432762146, "logit_z_loss": 37.78942394256592, "router_aux_loss": 1.4017714709043503, "router_z_loss": 0.9553081914782524, "routing_entropy_normalized": 0.6533838648201653, "learning_rate": 0.003, "grad_norm": 0.9763733148574829, "tokens_per_second": 4300744.110929079, "eta_seconds": 3384.1201672554016, "data_wait_seconds": 2.110004425048828e-05} {"step": 21680, "total_steps": 28610, "tokens": 45466255360, "source": "dclm", "phase": 4, "loss": 3.433530807495117, "ce_loss": 3.4183241426944733, "logit_z_loss": 27.967544078826904, "router_aux_loss": 1.3939933478832245, "router_z_loss": 1.2667478173971176, "routing_entropy_normalized": 0.6368082251187317, "learning_rate": 0.003, "grad_norm": 0.7650677561759949, "tokens_per_second": 3747750.360458533, "eta_seconds": 3877.8632411956787, "data_wait_seconds": 2.512335777282715e-05} {"step": 21690, "total_steps": 28610, "tokens": 45487226880, "source": "fineweb_edu", "phase": 4, "loss": 3.3794762194156647, "ce_loss": 3.3643159568309784, "logit_z_loss": 26.504719734191895, "router_aux_loss": 1.3589603453874588, "router_z_loss": 1.5706777274608612, "routing_entropy_normalized": 0.6802925595292655, "learning_rate": 0.003, "grad_norm": 0.7952196002006531, "tokens_per_second": 4044702.1410061326, "eta_seconds": 3587.975414276123, "data_wait_seconds": 2.065300941467285e-05} {"step": 21700, "total_steps": 28610, "tokens": 45508198400, "source": "fineweb_edu", "phase": 4, "loss": 3.4231986105442047, "ce_loss": 3.4075008034706116, "logit_z_loss": 25.865068912506104, "router_aux_loss": 1.4006214439868927, "router_z_loss": 1.6916018575429916, "routing_entropy_normalized": 0.6291808283914851, "learning_rate": 0.003, "grad_norm": 1.9426090717315674, "tokens_per_second": 3778616.8983098427, "eta_seconds": 3835.085881948471, "data_wait_seconds": 2.16066837310791e-05} {"step": 21710, "total_steps": 28610, "tokens": 45529169920, "source": "finemath", "phase": 4, "loss": 2.2243066132068634, "ce_loss": 2.2075593769550323, "logit_z_loss": 36.29506301879883, "router_aux_loss": 1.557151511311531, "router_z_loss": 1.1757320538163185, "routing_entropy_normalized": 0.6222118632384901, "learning_rate": 0.003, "grad_norm": 0.7138108611106873, "tokens_per_second": 4052461.322327734, "eta_seconds": 3570.755560398102, "data_wait_seconds": 1.8835067749023438e-05} {"step": 21720, "total_steps": 28610, "tokens": 45550141440, "source": "fineweb_edu", "phase": 4, "loss": 3.3610837757587433, "ce_loss": 3.343994051218033, "logit_z_loss": 25.61711549758911, "router_aux_loss": 1.5426868349313736, "router_z_loss": 1.6627885103225708, "routing_entropy_normalized": 0.6634025834561035, "learning_rate": 0.003, "grad_norm": 1.135550618171692, "tokens_per_second": 4182045.3333522244, "eta_seconds": 3455.098194360733, "data_wait_seconds": 2.0742416381835938e-05} {"step": 21730, "total_steps": 28610, "tokens": 45571112960, "source": "fineweb_hq", "phase": 4, "loss": 3.2995408475399017, "ce_loss": 3.2834678888320923, "logit_z_loss": 26.675276041030884, "router_aux_loss": 1.4325754791498184, "router_z_loss": 1.7472547441720963, "routing_entropy_normalized": 0.6620994798993115, "learning_rate": 0.003, "grad_norm": 1.3270267248153687, "tokens_per_second": 4052721.9744674764, "eta_seconds": 3560.176555633545, "data_wait_seconds": 1.990795135498047e-05} {"step": 21740, "total_steps": 28610, "tokens": 45592084480, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7740656435489655, "ce_loss": 2.7584815323352814, "logit_z_loss": 29.29020595550537, "router_aux_loss": 1.3489021360874176, "router_z_loss": 2.0951204001903534, "routing_entropy_normalized": 0.6720773671316493, "learning_rate": 0.003, "grad_norm": 1.2831820249557495, "tokens_per_second": 4098397.5054125204, "eta_seconds": 3515.3823466300964, "data_wait_seconds": 2.6285648345947266e-05} {"step": 21750, "total_steps": 28610, "tokens": 45613056000, "source": "fineweb_edu", "phase": 4, "loss": 3.375074476003647, "ce_loss": 3.3593794107437134, "logit_z_loss": 26.519786834716797, "router_aux_loss": 1.3973362147808075, "router_z_loss": 1.7217291742563248, "routing_entropy_normalized": 0.6321949050256813, "learning_rate": 0.003, "grad_norm": 1.1790375709533691, "tokens_per_second": 4344644.959283709, "eta_seconds": 3311.3091759681697, "data_wait_seconds": 1.627206802368164e-05} {"step": 21760, "total_steps": 28610, "tokens": 45634027520, "source": "finemath", "phase": 4, "loss": 2.2465225160121918, "ce_loss": 2.2305898666381836, "logit_z_loss": 36.99475812911987, "router_aux_loss": 1.4666224271059036, "router_z_loss": 1.2664299607276917, "routing_entropy_normalized": 0.6543039989138254, "learning_rate": 0.003, "grad_norm": 0.9747464656829834, "tokens_per_second": 3931787.4376464486, "eta_seconds": 3653.6795103549957, "data_wait_seconds": 2.104043960571289e-05} {"step": 21770, "total_steps": 28610, "tokens": 45654999040, "source": "finemath", "phase": 4, "loss": 2.3175333440303802, "ce_loss": 2.3024781346321106, "logit_z_loss": 35.919403076171875, "router_aux_loss": 1.3838547468185425, "router_z_loss": 1.2166933417320251, "routing_entropy_normalized": 0.6715644177945865, "learning_rate": 0.003, "grad_norm": 1.0310269594192505, "tokens_per_second": 3606110.111878424, "eta_seconds": 3977.83740234375, "data_wait_seconds": 2.1636486053466797e-05} {"step": 21780, "total_steps": 28610, "tokens": 45675970560, "source": "dclm", "phase": 4, "loss": 3.5088780522346497, "ce_loss": 3.4931126832962036, "logit_z_loss": 27.759136199951172, "router_aux_loss": 1.4354847818613052, "router_z_loss": 1.410519391298294, "routing_entropy_normalized": 0.6387188810454026, "learning_rate": 0.003, "grad_norm": 1.0104786157608032, "tokens_per_second": 3741341.7086190013, "eta_seconds": 3828.452270746231, "data_wait_seconds": 0.12329882383346558} {"step": 21790, "total_steps": 28610, "tokens": 45696942080, "source": "fineweb_edu", "phase": 4, "loss": 3.36499947309494, "ce_loss": 3.3496108651161194, "logit_z_loss": 27.58675265312195, "router_aux_loss": 1.3763030767440796, "router_z_loss": 1.6255941540002823, "routing_entropy_normalized": 0.666868653255946, "learning_rate": 0.003, "grad_norm": 0.7707235813140869, "tokens_per_second": 4309640.063221104, "eta_seconds": 3318.74041223526, "data_wait_seconds": 1.9848346710205078e-05} {"step": 21800, "total_steps": 28610, "tokens": 45717913600, "source": "fineweb_edu", "phase": 4, "loss": 3.3520845770835876, "ce_loss": 3.336373448371887, "logit_z_loss": 27.780262231826782, "router_aux_loss": 1.4047726094722748, "router_z_loss": 1.6634224653244019, "routing_entropy_normalized": 0.6400759865139709, "learning_rate": 0.003, "grad_norm": 1.543602705001831, "tokens_per_second": 4169643.450054305, "eta_seconds": 3425.138214111328, "data_wait_seconds": 2.2917985916137695e-05} {"step": 21810, "total_steps": 28610, "tokens": 45738885120, "source": "finemath", "phase": 4, "loss": 2.28496715426445, "ce_loss": 2.2701438665390015, "logit_z_loss": 36.85794544219971, "router_aux_loss": 1.3742992728948593, "router_z_loss": 1.0802655145525932, "routing_entropy_normalized": 0.6650020808228152, "learning_rate": 0.003, "grad_norm": 0.9444580078125, "tokens_per_second": 3487916.4056505393, "eta_seconds": 4088.5823917388916, "data_wait_seconds": 2.2202730178833008e-05} {"step": 21820, "total_steps": 28610, "tokens": 45759856640, "source": "fineweb_edu", "phase": 4, "loss": 3.329242318868637, "ce_loss": 3.3133737444877625, "logit_z_loss": 26.687283992767334, "router_aux_loss": 1.4357028752565384, "router_z_loss": 1.5115176886320114, "routing_entropy_normalized": 0.6603806778736259, "learning_rate": 0.003, "grad_norm": 0.5189153552055359, "tokens_per_second": 4351464.518579063, "eta_seconds": 3272.383819103241, "data_wait_seconds": 2.1845102310180664e-05} {"step": 21830, "total_steps": 28610, "tokens": 45780828160, "source": "fineweb_hq", "phase": 4, "loss": 3.342749744653702, "ce_loss": 3.3272326588630676, "logit_z_loss": 27.43641948699951, "router_aux_loss": 1.3938104212284088, "router_z_loss": 1.5790029764175415, "routing_entropy_normalized": 0.6487437848781984, "learning_rate": 0.003, "grad_norm": 1.2711374759674072, "tokens_per_second": 3992686.9701191667, "eta_seconds": 3561.1834001541138, "data_wait_seconds": 2.0384788513183594e-05} {"step": 21840, "total_steps": 28610, "tokens": 45801799680, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7970401644706726, "ce_loss": 2.7811287939548492, "logit_z_loss": 29.509281873703003, "router_aux_loss": 1.4022898823022842, "router_z_loss": 1.8884214907884598, "routing_entropy_normalized": 0.648727281085114, "learning_rate": 0.003, "grad_norm": 1.6554492712020874, "tokens_per_second": 4453441.534316908, "eta_seconds": 3188.0331044197083, "data_wait_seconds": 2.244114875793457e-05} {"step": 21850, "total_steps": 28610, "tokens": 45822771200, "source": "fineweb_edu", "phase": 4, "loss": 3.384047359228134, "ce_loss": 3.3685595095157623, "logit_z_loss": 26.522236108779907, "router_aux_loss": 1.3906173259019852, "router_z_loss": 1.5817272812128067, "routing_entropy_normalized": 0.6672141111543576, "learning_rate": 0.003, "grad_norm": 1.4125370979309082, "tokens_per_second": 3963829.9485401534, "eta_seconds": 3576.5276775360107, "data_wait_seconds": 2.232193946838379e-05} {"step": 21860, "total_steps": 28610, "tokens": 45843742720, "source": "finemath", "phase": 4, "loss": 2.2563124895095825, "ce_loss": 2.2414498329162598, "logit_z_loss": 36.81995487213135, "router_aux_loss": 1.383549988269806, "router_z_loss": 1.0271345376968384, "routing_entropy_normalized": 0.6640949817814803, "learning_rate": 0.003, "grad_norm": 0.7834697961807251, "tokens_per_second": 3855031.0744860736, "eta_seconds": 3672.0264315605164, "data_wait_seconds": 2.3484230041503906e-05} {"step": 21870, "total_steps": 28610, "tokens": 45864714240, "source": "finemath", "phase": 4, "loss": 2.1770308315753937, "ce_loss": 2.1626356691122055, "logit_z_loss": 36.472586154937744, "router_aux_loss": 1.3428024053573608, "router_z_loss": 0.9671195298433304, "routing_entropy_normalized": 0.648865370387642, "learning_rate": 0.003, "grad_norm": 0.9432724118232727, "tokens_per_second": 3778658.940032939, "eta_seconds": 3740.6933794021606, "data_wait_seconds": 2.1696090698242188e-05} {"step": 21880, "total_steps": 28610, "tokens": 45885685760, "source": "dclm", "phase": 4, "loss": 3.4834890365600586, "ce_loss": 3.468650758266449, "logit_z_loss": 27.82248830795288, "router_aux_loss": 1.3671186417341232, "router_z_loss": 1.167159542441368, "routing_entropy_normalized": 0.6590034974897891, "learning_rate": 0.003, "grad_norm": 1.0337079763412476, "tokens_per_second": 3996530.6619658708, "eta_seconds": 3531.5212502479553, "data_wait_seconds": 2.5540590286254883e-05} {"step": 21890, "total_steps": 28610, "tokens": 45906657280, "source": "fineweb_edu", "phase": 4, "loss": 3.3816153705120087, "ce_loss": 3.3665976524353027, "logit_z_loss": 26.973382472991943, "router_aux_loss": 1.3577561378479004, "router_z_loss": 1.4401344656944275, "routing_entropy_normalized": 0.645123516695982, "learning_rate": 0.003, "grad_norm": 0.6786438226699829, "tokens_per_second": 4047579.9226182317, "eta_seconds": 3481.7994232177734, "data_wait_seconds": 2.053380012512207e-05} {"step": 21900, "total_steps": 28610, "tokens": 45927628800, "source": "fineweb_edu", "phase": 4, "loss": 3.356602728366852, "ce_loss": 3.3413259088993073, "logit_z_loss": 26.589980840682983, "router_aux_loss": 1.39393712580204, "router_z_loss": 1.3374918103218079, "routing_entropy_normalized": 0.6313806477425601, "learning_rate": 0.003, "grad_norm": 1.1424007415771484, "tokens_per_second": 3781312.0567888943, "eta_seconds": 3721.4304740428925, "data_wait_seconds": 1.8328428268432617e-05} {"step": 21910, "total_steps": 28610, "tokens": 45948600320, "source": "finemath", "phase": 4, "loss": 2.201274201273918, "ce_loss": 2.1870943754911423, "logit_z_loss": 35.889305114746094, "router_aux_loss": 1.3202537596225739, "router_z_loss": 0.9773217365145683, "routing_entropy_normalized": 0.6833538693667257, "learning_rate": 0.003, "grad_norm": 0.9510669708251953, "tokens_per_second": 4107912.459148046, "eta_seconds": 3420.4522466659546, "data_wait_seconds": 2.1159648895263672e-05} {"step": 21920, "total_steps": 28610, "tokens": 45969571840, "source": "fineweb_edu", "phase": 4, "loss": 3.3656421303749084, "ce_loss": 3.3505940437316895, "logit_z_loss": 27.283748865127563, "router_aux_loss": 1.3759016692638397, "router_z_loss": 1.2890421897172928, "routing_entropy_normalized": 0.6503861347985695, "learning_rate": 0.003, "grad_norm": 0.9524573683738708, "tokens_per_second": 4360818.206078706, "eta_seconds": 3217.273964881897, "data_wait_seconds": 2.014636993408203e-05} {"step": 21930, "total_steps": 28610, "tokens": 45990543360, "source": "fineweb_hq", "phase": 4, "loss": 3.306674599647522, "ce_loss": 3.2917108833789825, "logit_z_loss": 29.258920907974243, "router_aux_loss": 1.3795158118009567, "router_z_loss": 1.1686134487390518, "routing_entropy_normalized": 0.6299599324001666, "learning_rate": 0.003, "grad_norm": 1.603137493133545, "tokens_per_second": 4479789.909552083, "eta_seconds": 3127.15007686615, "data_wait_seconds": 2.256035804748535e-05} {"step": 21940, "total_steps": 28610, "tokens": 46011514880, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7626709640026093, "ce_loss": 2.747005134820938, "logit_z_loss": 30.124348640441895, "router_aux_loss": 1.4213014990091324, "router_z_loss": 1.4527768790721893, "routing_entropy_normalized": 0.6627325490200766, "learning_rate": 0.003, "grad_norm": 1.71577787399292, "tokens_per_second": 3511461.6201455486, "eta_seconds": 3983.5274746418, "data_wait_seconds": 2.0056962966918945e-05} {"step": 21950, "total_steps": 28610, "tokens": 46032486400, "source": "fineweb_edu", "phase": 4, "loss": 3.336144655942917, "ce_loss": 3.3208207488059998, "logit_z_loss": 26.405658960342407, "router_aux_loss": 1.4036243110895157, "router_z_loss": 1.2876557856798172, "routing_entropy_normalized": 0.6444100455952315, "learning_rate": 0.003, "grad_norm": 1.514217734336853, "tokens_per_second": 3738983.7052460583, "eta_seconds": 3735.5156965255737, "data_wait_seconds": 2.5451183319091797e-05} {"step": 21960, "total_steps": 28610, "tokens": 46053457920, "source": "finemath", "phase": 4, "loss": 2.2101626992225647, "ce_loss": 2.194728285074234, "logit_z_loss": 37.141539096832275, "router_aux_loss": 1.4525594413280487, "router_z_loss": 0.9088190346956253, "routing_entropy_normalized": 0.6398488660435689, "learning_rate": 0.003, "grad_norm": 0.5743735432624817, "tokens_per_second": 4197788.6927333735, "eta_seconds": 3322.23983168602, "data_wait_seconds": 1.9282102584838867e-05} {"step": 21970, "total_steps": 28610, "tokens": 46074429440, "source": "finemath", "phase": 4, "loss": 2.2554507702589035, "ce_loss": 2.2412503212690353, "logit_z_loss": 36.5126371383667, "router_aux_loss": 1.3262057453393936, "router_z_loss": 0.9383575692772865, "routing_entropy_normalized": 0.6760103495384302, "learning_rate": 0.003, "grad_norm": 1.2345564365386963, "tokens_per_second": 4236563.332993777, "eta_seconds": 3286.883302688598, "data_wait_seconds": 1.9550323486328125e-05} {"step": 21980, "total_steps": 28610, "tokens": 46095400960, "source": "dclm", "phase": 4, "loss": 3.4952125549316406, "ce_loss": 3.480462908744812, "logit_z_loss": 27.744049072265625, "router_aux_loss": 1.3582420647144318, "router_z_loss": 1.1672185212373734, "routing_entropy_normalized": 0.6639551357332668, "learning_rate": 0.003, "grad_norm": 0.5870425701141357, "tokens_per_second": 3926780.0856355787, "eta_seconds": 3540.844523191452, "data_wait_seconds": 1.8864870071411133e-05} {"step": 21990, "total_steps": 28610, "tokens": 46116372480, "source": "fineweb_edu", "phase": 4, "loss": 3.3748219907283783, "ce_loss": 3.3600481152534485, "logit_z_loss": 26.458386659622192, "router_aux_loss": 1.3560213893651962, "router_z_loss": 1.213645488023758, "routing_entropy_normalized": 0.6453265439201589, "learning_rate": 0.003, "grad_norm": 0.9209600687026978, "tokens_per_second": 4283108.230949415, "eta_seconds": 3241.3718008995056, "data_wait_seconds": 2.3573637008666992e-05} {"step": 22000, "total_steps": 28610, "tokens": 46137344000, "source": "fineweb_edu", "phase": 4, "loss": 3.362921863794327, "ce_loss": 3.3486066460609436, "logit_z_loss": 27.02803349494934, "router_aux_loss": 1.3019706308841705, "router_z_loss": 1.2954963445663452, "routing_entropy_normalized": 0.6480865846209922, "learning_rate": 0.003, "grad_norm": 1.03487229347229, "tokens_per_second": 4194925.492076105, "eta_seconds": 3304.510353326797, "data_wait_seconds": 2.0623207092285156e-05} {"step": 22010, "total_steps": 28610, "tokens": 46158315520, "source": "finemath", "phase": 4, "loss": 2.2716998159885406, "ce_loss": 2.2567287981510162, "logit_z_loss": 37.984557151794434, "router_aux_loss": 1.4160926640033722, "router_z_loss": 0.8100815117359161, "routing_entropy_normalized": 0.6640106008237134, "learning_rate": 0.003, "grad_norm": 0.6171634793281555, "tokens_per_second": 3537143.0361153926, "eta_seconds": 3913.1024837493896, "data_wait_seconds": 2.4497509002685547e-05} {"step": 22020, "total_steps": 28610, "tokens": 46179287040, "source": "fineweb_edu", "phase": 4, "loss": 3.3568629920482635, "ce_loss": 3.3422484397888184, "logit_z_loss": 26.91049885749817, "router_aux_loss": 1.3367871791124344, "router_z_loss": 1.2466847598552704, "routing_entropy_normalized": 0.6604782001454446, "learning_rate": 0.003, "grad_norm": 1.1662718057632446, "tokens_per_second": 4008948.0370026184, "eta_seconds": 3447.3461747169495, "data_wait_seconds": 1.8924474716186523e-05} {"step": 22030, "total_steps": 28610, "tokens": 46200258560, "source": "fineweb_hq", "phase": 4, "loss": 3.3212398886680603, "ce_loss": 3.3064242899417877, "logit_z_loss": 28.100322723388672, "router_aux_loss": 1.3644752204418182, "router_z_loss": 1.1708896160125732, "routing_entropy_normalized": 0.6409181075847976, "learning_rate": 0.003, "grad_norm": 2.0067179203033447, "tokens_per_second": 4048649.2918151408, "eta_seconds": 3408.3614473342896, "data_wait_seconds": 2.065300941467285e-05} {"step": 22040, "total_steps": 28610, "tokens": 46221230080, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7647789418697357, "ce_loss": 2.7496182918548584, "logit_z_loss": 28.958396911621094, "router_aux_loss": 1.3640313148498535, "router_z_loss": 1.5203826278448105, "routing_entropy_normalized": 0.6775124824156529, "learning_rate": 0.003, "grad_norm": 1.2022583484649658, "tokens_per_second": 3847620.0329966233, "eta_seconds": 3580.9899423122406, "data_wait_seconds": 2.2292137145996094e-05} {"step": 22050, "total_steps": 28610, "tokens": 46242201600, "source": "fineweb_edu", "phase": 4, "loss": 3.336828052997589, "ce_loss": 3.321581393480301, "logit_z_loss": 27.08175802230835, "router_aux_loss": 1.4066216498613358, "router_z_loss": 1.1804518550634384, "routing_entropy_normalized": 0.6434405992381018, "learning_rate": 0.003, "grad_norm": 0.7265595197677612, "tokens_per_second": 3874189.1187740006, "eta_seconds": 3551.01847076416, "data_wait_seconds": 2.321600914001465e-05} {"step": 22060, "total_steps": 28610, "tokens": 46263173120, "source": "finemath", "phase": 4, "loss": 2.251890867948532, "ce_loss": 2.2364724576473236, "logit_z_loss": 36.509681701660156, "router_aux_loss": 1.463350623846054, "router_z_loss": 0.7849296629428864, "routing_entropy_normalized": 0.6420433294873417, "learning_rate": 0.003, "grad_norm": 0.8459575772285461, "tokens_per_second": 3706023.362787116, "eta_seconds": 3706.491906642914, "data_wait_seconds": 1.9282102584838867e-05} {"step": 22070, "total_steps": 28610, "tokens": 46284144640, "source": "finemath", "phase": 4, "loss": 2.2120817601680756, "ce_loss": 2.197418659925461, "logit_z_loss": 37.98400020599365, "router_aux_loss": 1.3802799731492996, "router_z_loss": 0.8602784350514412, "routing_entropy_normalized": 0.663465639504227, "learning_rate": 0.003, "grad_norm": 1.1980836391448975, "tokens_per_second": 4212015.476554871, "eta_seconds": 3256.249687671661, "data_wait_seconds": 2.1755695343017578e-05} {"step": 22080, "total_steps": 28610, "tokens": 46305116160, "source": "dclm", "phase": 4, "loss": 3.5137969851493835, "ce_loss": 3.4987024068832397, "logit_z_loss": 27.547119855880737, "router_aux_loss": 1.4035890996456146, "router_z_loss": 1.0586622580885887, "routing_entropy_normalized": 0.6469585618402917, "learning_rate": 0.003, "grad_norm": 0.5409637689590454, "tokens_per_second": 3870600.2078414643, "eta_seconds": 3538.056586742401, "data_wait_seconds": 2.130866050720215e-05} {"step": 22090, "total_steps": 28610, "tokens": 46326087680, "source": "fineweb_edu", "phase": 4, "loss": 3.378440946340561, "ce_loss": 3.3636763095855713, "logit_z_loss": 26.339415311813354, "router_aux_loss": 1.3479851931333542, "router_z_loss": 1.284796565771103, "routing_entropy_normalized": 0.6530876552330762, "learning_rate": 0.003, "grad_norm": 1.0208584070205688, "tokens_per_second": 4105872.615766085, "eta_seconds": 3330.213165283203, "data_wait_seconds": 2.333521842956543e-05} {"step": 22100, "total_steps": 28610, "tokens": 46347059200, "source": "fineweb_edu", "phase": 4, "loss": 3.3583602607250214, "ce_loss": 3.3435737788677216, "logit_z_loss": 26.753918647766113, "router_aux_loss": 1.3596217632293701, "router_z_loss": 1.1902632415294647, "routing_entropy_normalized": 0.6678248954799952, "learning_rate": 0.003, "grad_norm": 1.025260329246521, "tokens_per_second": 3823790.595429975, "eta_seconds": 3570.3993666172028, "data_wait_seconds": 2.1457672119140625e-05} {"step": 22110, "total_steps": 28610, "tokens": 46368030720, "source": "finemath", "phase": 4, "loss": 2.2967059314250946, "ce_loss": 2.282169610261917, "logit_z_loss": 34.96569013595581, "router_aux_loss": 1.3721453100442886, "router_z_loss": 0.8148514032363892, "routing_entropy_normalized": 0.672987070913082, "learning_rate": 0.003, "grad_norm": 0.6328112483024597, "tokens_per_second": 3941495.58827275, "eta_seconds": 3458.455729484558, "data_wait_seconds": 2.390146255493164e-05} {"step": 22120, "total_steps": 28610, "tokens": 46389002240, "source": "fineweb_edu", "phase": 4, "loss": 3.3703192472457886, "ce_loss": 3.3559528589248657, "logit_z_loss": 26.097293615341187, "router_aux_loss": 1.3230971097946167, "router_z_loss": 1.1353906095027924, "routing_entropy_normalized": 0.6667535436682377, "learning_rate": 0.003, "grad_norm": 0.9060843586921692, "tokens_per_second": 4362871.946751304, "eta_seconds": 3119.623185396195, "data_wait_seconds": 2.0325183868408203e-05} {"step": 22130, "total_steps": 28610, "tokens": 46409973760, "source": "fineweb_hq", "phase": 4, "loss": 3.3782268464565277, "ce_loss": 3.364345818758011, "logit_z_loss": 27.446218490600586, "router_aux_loss": 1.2739882320165634, "router_z_loss": 1.1411386579275131, "routing_entropy_normalized": 0.672205057660604, "learning_rate": 0.003, "grad_norm": 1.5491936206817627, "tokens_per_second": 4090363.632182172, "eta_seconds": 3322.3317489624023, "data_wait_seconds": 2.250075340270996e-05} {"step": 22140, "total_steps": 28610, "tokens": 46430945280, "source": "cosmopedia_v2", "phase": 4, "loss": 2.748610705137253, "ce_loss": 2.7340783774852753, "logit_z_loss": 29.369702100753784, "router_aux_loss": 1.31680366396904, "router_z_loss": 1.3642858415842056, "routing_entropy_normalized": 0.6453779121141116, "learning_rate": 0.003, "grad_norm": 1.0919365882873535, "tokens_per_second": 3214873.5458828616, "eta_seconds": 4220.562098741531, "data_wait_seconds": 2.181529998779297e-05} {"step": 22150, "total_steps": 28610, "tokens": 46451916800, "source": "fineweb_edu", "phase": 4, "loss": 3.3651652336120605, "ce_loss": 3.350655347108841, "logit_z_loss": 26.513380527496338, "router_aux_loss": 1.3471240997314453, "router_z_loss": 1.038631372153759, "routing_entropy_normalized": 0.6566261018353413, "learning_rate": 0.003, "grad_norm": 0.9477311968803406, "tokens_per_second": 4363579.902527903, "eta_seconds": 3104.6989450454707, "data_wait_seconds": 2.0444393157958984e-05} {"step": 22160, "total_steps": 28610, "tokens": 46472888320, "source": "finemath", "phase": 4, "loss": 2.373186469078064, "ce_loss": 2.3581776320934296, "logit_z_loss": 35.256001472473145, "router_aux_loss": 1.4298862665891647, "router_z_loss": 0.7099317833781242, "routing_entropy_normalized": 0.6400283834136603, "learning_rate": 0.003, "grad_norm": 0.9795953035354614, "tokens_per_second": 4249543.452937842, "eta_seconds": 3183.0785000324254, "data_wait_seconds": 2.1457672119140625e-05} {"step": 22170, "total_steps": 28610, "tokens": 46493859840, "source": "finemath", "phase": 4, "loss": 2.212584137916565, "ce_loss": 2.1975682973861694, "logit_z_loss": 36.921584606170654, "router_aux_loss": 1.4375588297843933, "router_z_loss": 0.6402337998151779, "routing_entropy_normalized": 0.6568967383932043, "learning_rate": 0.003, "grad_norm": 0.8065222501754761, "tokens_per_second": 3876779.575373719, "eta_seconds": 3483.731436729431, "data_wait_seconds": 2.0682811737060547e-05} {"step": 22180, "total_steps": 28610, "tokens": 46514831360, "source": "dclm", "phase": 4, "loss": 3.480434477329254, "ce_loss": 3.465311288833618, "logit_z_loss": 26.77336072921753, "router_aux_loss": 1.4109708964824677, "router_z_loss": 1.013530395925045, "routing_entropy_normalized": 0.6212184132186078, "learning_rate": 0.003, "grad_norm": 1.0762125253677368, "tokens_per_second": 3932551.91405775, "eta_seconds": 3428.991569519043, "data_wait_seconds": 2.0176172256469727e-05} {"step": 22190, "total_steps": 28610, "tokens": 46535802880, "source": "fineweb_edu", "phase": 4, "loss": 3.401620477437973, "ce_loss": 3.3865557610988617, "logit_z_loss": 26.24057412147522, "router_aux_loss": 1.396842509508133, "router_z_loss": 1.0962809920310974, "routing_entropy_normalized": 0.6362134698183451, "learning_rate": 0.003, "grad_norm": 0.7970365285873413, "tokens_per_second": 3838803.8133669784, "eta_seconds": 3507.2685384750366, "data_wait_seconds": 2.1249055862426758e-05} {"step": 22200, "total_steps": 28610, "tokens": 46556774400, "source": "fineweb_edu", "phase": 4, "loss": 3.364762634038925, "ce_loss": 3.349762737751007, "logit_z_loss": 26.44001078605652, "router_aux_loss": 1.3833898603916168, "router_z_loss": 1.165971353650093, "routing_entropy_normalized": 0.6702450883208445, "learning_rate": 0.003, "grad_norm": 1.0932862758636475, "tokens_per_second": 3914683.683555556, "eta_seconds": 3433.928615093231, "data_wait_seconds": 1.901388168334961e-05} {"step": 22210, "total_steps": 28610, "tokens": 46577745920, "source": "finemath", "phase": 4, "loss": 2.2692997455596924, "ce_loss": 2.254749119281769, "logit_z_loss": 37.28177881240845, "router_aux_loss": 1.3848482519388199, "router_z_loss": 0.7021169029176235, "routing_entropy_normalized": 0.6501542483898005, "learning_rate": 0.003, "grad_norm": 0.8930260539054871, "tokens_per_second": 3478786.2811535145, "eta_seconds": 3858.176879882813, "data_wait_seconds": 2.2917985916137695e-05} {"step": 22220, "total_steps": 28610, "tokens": 46598717440, "source": "fineweb_edu", "phase": 4, "loss": 3.3422127962112427, "ce_loss": 3.3274781703948975, "logit_z_loss": 26.583314657211304, "router_aux_loss": 1.352111130952835, "router_z_loss": 1.2134576737880707, "routing_entropy_normalized": 0.6514373181833131, "learning_rate": 0.003, "grad_norm": 1.6994019746780396, "tokens_per_second": 4375181.158094624, "eta_seconds": 3062.913464784622, "data_wait_seconds": 1.9848346710205078e-05} {"step": 22230, "total_steps": 28610, "tokens": 46619688960, "source": "fineweb_hq", "phase": 4, "loss": 3.352594256401062, "ce_loss": 3.3366739451885223, "logit_z_loss": 26.923279285430908, "router_aux_loss": 1.475475937128067, "router_z_loss": 1.1655558049678802, "routing_entropy_normalized": 0.6440339894880669, "learning_rate": 0.003, "grad_norm": 1.183447241783142, "tokens_per_second": 3572591.551953343, "eta_seconds": 3745.132788181305, "data_wait_seconds": 1.8715858459472656e-05} {"step": 22240, "total_steps": 28610, "tokens": 46640660480, "source": "cosmopedia_v2", "phase": 4, "loss": 2.762521892786026, "ce_loss": 2.7469539046287537, "logit_z_loss": 29.904491901397705, "router_aux_loss": 1.3923950493335724, "router_z_loss": 1.6440034806728363, "routing_entropy_normalized": 0.6564070602954392, "learning_rate": 0.003, "grad_norm": 1.3519951105117798, "tokens_per_second": 4023828.6371281915, "eta_seconds": 3319.9371654987335, "data_wait_seconds": 2.6911497116088867e-05} {"step": 22250, "total_steps": 28610, "tokens": 46661632000, "source": "fineweb_edu", "phase": 4, "loss": 3.3288187980651855, "ce_loss": 3.3139871954917908, "logit_z_loss": 26.74640154838562, "router_aux_loss": 1.340922325849533, "router_z_loss": 1.4223741590976715, "routing_entropy_normalized": 0.6884868617912442, "learning_rate": 0.003, "grad_norm": 0.8507772088050842, "tokens_per_second": 3438601.4118319633, "eta_seconds": 3878.8696689605713, "data_wait_seconds": 2.1517276763916016e-05} {"step": 22260, "total_steps": 28610, "tokens": 46682603520, "source": "finemath", "phase": 4, "loss": 2.2647994458675385, "ce_loss": 2.2492617070674896, "logit_z_loss": 36.76797533035278, "router_aux_loss": 1.4609629660844803, "router_z_loss": 0.9281135573983192, "routing_entropy_normalized": 0.641252052236515, "learning_rate": 0.003, "grad_norm": 1.0180577039718628, "tokens_per_second": 3754712.7195814606, "eta_seconds": 3546.7201340198517, "data_wait_seconds": 2.1517276763916016e-05} {"step": 22270, "total_steps": 28610, "tokens": 46703575040, "source": "finemath", "phase": 4, "loss": 2.210975259542465, "ce_loss": 2.195658415555954, "logit_z_loss": 37.37734079360962, "router_aux_loss": 1.4422384649515152, "router_z_loss": 0.8943988010287285, "routing_entropy_normalized": 0.6628466006572417, "learning_rate": 0.003, "grad_norm": 1.1760562658309937, "tokens_per_second": 3773749.4748545927, "eta_seconds": 3523.2714223861694, "data_wait_seconds": 2.065300941467285e-05} {"step": 22280, "total_steps": 28610, "tokens": 46724546560, "source": "dclm", "phase": 4, "loss": 3.3604122698307037, "ce_loss": 3.3450677394866943, "logit_z_loss": 28.165277004241943, "router_aux_loss": 1.4159896075725555, "router_z_loss": 1.184656746685505, "routing_entropy_normalized": 0.6323683072884043, "learning_rate": 0.003, "grad_norm": 1.3865492343902588, "tokens_per_second": 4098395.7867893176, "eta_seconds": 3239.0654418468475, "data_wait_seconds": 1.8864870071411133e-05} {"step": 22290, "total_steps": 28610, "tokens": 46745518080, "source": "fineweb_edu", "phase": 4, "loss": 3.3399564921855927, "ce_loss": 3.324368804693222, "logit_z_loss": 27.551633596420288, "router_aux_loss": 1.408181682229042, "router_z_loss": 1.5058394819498062, "routing_entropy_normalized": 0.6559674043410971, "learning_rate": 0.003, "grad_norm": 0.6873899102210999, "tokens_per_second": 3515007.5457206895, "eta_seconds": 3770.689100265503, "data_wait_seconds": 2.1904706954956055e-05} {"step": 22300, "total_steps": 28610, "tokens": 46766489600, "source": "fineweb_edu", "phase": 4, "loss": 3.4104137122631073, "ce_loss": 3.394770532846451, "logit_z_loss": 27.805052280426025, "router_aux_loss": 1.4154475629329681, "router_z_loss": 1.488676741719246, "routing_entropy_normalized": 0.6570474345092069, "learning_rate": 0.003, "grad_norm": 1.7101542949676514, "tokens_per_second": 4702845.952097561, "eta_seconds": 2813.8342728614807, "data_wait_seconds": 2.321600914001465e-05} {"step": 22310, "total_steps": 28610, "tokens": 46787461120, "source": "finemath", "phase": 4, "loss": 2.216373771429062, "ce_loss": 2.2013844549655914, "logit_z_loss": 37.762930393218994, "router_aux_loss": 1.3915645480155945, "router_z_loss": 1.0737154483795166, "routing_entropy_normalized": 0.6744511222257333, "learning_rate": 0.003, "grad_norm": 0.8707786202430725, "tokens_per_second": 3448265.4213505667, "eta_seconds": 3831.5083050727844, "data_wait_seconds": 2.3663043975830078e-05} {"step": 22320, "total_steps": 28610, "tokens": 46808432640, "source": "fineweb_edu", "phase": 4, "loss": 3.367657870054245, "ce_loss": 3.352687805891037, "logit_z_loss": 27.05223059654236, "router_aux_loss": 1.3524506390094757, "router_z_loss": 1.445600688457489, "routing_entropy_normalized": 0.6513899092061676, "learning_rate": 0.003, "grad_norm": 1.4691181182861328, "tokens_per_second": 4551702.897518427, "eta_seconds": 2898.0551624298096, "data_wait_seconds": 2.232193946838379e-05} {"step": 22330, "total_steps": 28610, "tokens": 46829404160, "source": "fineweb_hq", "phase": 4, "loss": 3.40729358792305, "ce_loss": 3.3919992446899414, "logit_z_loss": 25.15990924835205, "router_aux_loss": 1.3898541182279587, "router_z_loss": 1.3957579135894775, "routing_entropy_normalized": 0.6443296237875059, "learning_rate": 0.003, "grad_norm": 1.7828893661499023, "tokens_per_second": 3835896.6275202353, "eta_seconds": 3433.386203765869, "data_wait_seconds": 1.8715858459472656e-05} {"step": 22340, "total_steps": 28610, "tokens": 46850375680, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7628432512283325, "ce_loss": 2.747235119342804, "logit_z_loss": 28.171858549118042, "router_aux_loss": 1.3898663520812988, "router_z_loss": 1.7094381302595139, "routing_entropy_normalized": 0.6673502048904555, "learning_rate": 0.003, "grad_norm": 1.281067967414856, "tokens_per_second": 4303191.897584333, "eta_seconds": 3055.6720111370087, "data_wait_seconds": 2.154707908630371e-05} {"step": 22350, "total_steps": 28610, "tokens": 46871347200, "source": "fineweb_edu", "phase": 4, "loss": 3.400820016860962, "ce_loss": 3.386311888694763, "logit_z_loss": 26.66195273399353, "router_aux_loss": 1.3037135899066925, "router_z_loss": 1.4709796458482742, "routing_entropy_normalized": 0.6765850612079429, "learning_rate": 0.003, "grad_norm": 0.96062833070755, "tokens_per_second": 3456653.264957609, "eta_seconds": 3797.9428405761714, "data_wait_seconds": 2.327561378479004e-05} {"step": 22360, "total_steps": 28610, "tokens": 46892318720, "source": "finemath", "phase": 4, "loss": 2.211026757955551, "ce_loss": 2.196287512779236, "logit_z_loss": 38.42161464691162, "router_aux_loss": 1.3844251334667206, "router_z_loss": 0.8949887454509735, "routing_entropy_normalized": 0.6475083972296498, "learning_rate": 0.003, "grad_norm": 0.9049498438835144, "tokens_per_second": 4188386.161409915, "eta_seconds": 3129.415363073349, "data_wait_seconds": 2.1845102310180664e-05} {"step": 22370, "total_steps": 28610, "tokens": 46913290240, "source": "finemath", "phase": 4, "loss": 2.237174391746521, "ce_loss": 2.222455382347107, "logit_z_loss": 38.38891363143921, "router_aux_loss": 1.371583953499794, "router_z_loss": 1.0032235756516457, "routing_entropy_normalized": 0.6559946898694469, "learning_rate": 0.003, "grad_norm": 1.0900373458862305, "tokens_per_second": 3746022.9365281113, "eta_seconds": 3493.3658180236816, "data_wait_seconds": 2.205371856689453e-05} {"step": 22380, "total_steps": 28610, "tokens": 46934261760, "source": "dclm", "phase": 4, "loss": 3.430110603570938, "ce_loss": 3.414984881877899, "logit_z_loss": 28.184962511062622, "router_aux_loss": 1.3967389613389969, "router_z_loss": 1.1583346277475357, "routing_entropy_normalized": 0.629261835479619, "learning_rate": 0.003, "grad_norm": 1.036229133605957, "tokens_per_second": 4101758.067945025, "eta_seconds": 3185.282199382782, "data_wait_seconds": 2.3186206817626953e-05} {"step": 22390, "total_steps": 28610, "tokens": 46955233280, "source": "fineweb_edu", "phase": 4, "loss": 3.37286776304245, "ce_loss": 3.3579125106334686, "logit_z_loss": 27.218985319137573, "router_aux_loss": 1.3557259887456894, "router_z_loss": 1.3979800641536713, "routing_entropy_normalized": 0.6454622137909478, "learning_rate": 0.003, "grad_norm": 0.9966010451316833, "tokens_per_second": 3662602.518145014, "eta_seconds": 3561.4799518585205, "data_wait_seconds": 2.065300941467285e-05} {"step": 22400, "total_steps": 28610, "tokens": 46976204800, "source": "fineweb_edu", "phase": 4, "loss": 3.3796413242816925, "ce_loss": 3.365108847618103, "logit_z_loss": 26.58247947692871, "router_aux_loss": 1.318293884396553, "router_z_loss": 1.3495937138795853, "routing_entropy_normalized": 0.6615332149444642, "learning_rate": 0.003, "grad_norm": 1.4976533651351929, "tokens_per_second": 3822893.8514548508, "eta_seconds": 3406.663754224777, "data_wait_seconds": 0.0746544599533081} {"step": 22410, "total_steps": 28610, "tokens": 46997176320, "source": "finemath", "phase": 4, "loss": 2.2001576125621796, "ce_loss": 2.185806542634964, "logit_z_loss": 36.50551652908325, "router_aux_loss": 1.3467350006103516, "router_z_loss": 0.8836938068270683, "routing_entropy_normalized": 0.6814344527502042, "learning_rate": 0.003, "grad_norm": 0.8605618476867676, "tokens_per_second": 4080821.2209589616, "eta_seconds": 3186.207308769226, "data_wait_seconds": 1.8775463104248047e-05} {"step": 22420, "total_steps": 28610, "tokens": 47018147840, "source": "fineweb_edu", "phase": 4, "loss": 3.3742270171642303, "ce_loss": 3.3597792983055115, "logit_z_loss": 24.79952096939087, "router_aux_loss": 1.3221451193094254, "router_z_loss": 1.2262050211429596, "routing_entropy_normalized": 0.6640298432155567, "learning_rate": 0.003, "grad_norm": 1.305342674255371, "tokens_per_second": 3725549.1281934516, "eta_seconds": 3484.4181175231934, "data_wait_seconds": 2.2113323211669922e-05} {"step": 22430, "total_steps": 28610, "tokens": 47039119360, "source": "fineweb_hq", "phase": 4, "loss": 3.404931515455246, "ce_loss": 3.3895882666110992, "logit_z_loss": 26.40185284614563, "router_aux_loss": 1.4107883274555206, "router_z_loss": 1.2353008687496185, "routing_entropy_normalized": 0.6499081282594386, "learning_rate": 0.003, "grad_norm": 1.5502872467041016, "tokens_per_second": 3906074.6510683484, "eta_seconds": 3318.011179447174, "data_wait_seconds": 2.0563602447509766e-05} {"step": 22440, "total_steps": 28610, "tokens": 47060090880, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7624738216400146, "ce_loss": 2.7467378675937653, "logit_z_loss": 28.237941026687622, "router_aux_loss": 1.41511669754982, "router_z_loss": 1.584859237074852, "routing_entropy_normalized": 0.6327101706478795, "learning_rate": 0.003, "grad_norm": 1.2767404317855835, "tokens_per_second": 4205063.530498818, "eta_seconds": 3077.106385231018, "data_wait_seconds": 2.3096799850463867e-05} {"step": 22450, "total_steps": 28610, "tokens": 47081062400, "source": "fineweb_edu", "phase": 4, "loss": 3.340886116027832, "ce_loss": 3.3259671330451965, "logit_z_loss": 26.467565059661865, "router_aux_loss": 1.3573994636535645, "router_z_loss": 1.3449746072292328, "routing_entropy_normalized": 0.6780803921768821, "learning_rate": 0.003, "grad_norm": 1.5286611318588257, "tokens_per_second": 4149677.7180033205, "eta_seconds": 3113.1228008270264, "data_wait_seconds": 1.665949821472168e-05} {"step": 22460, "total_steps": 28610, "tokens": 47102033920, "source": "finemath", "phase": 4, "loss": 1.9511446058750153, "ce_loss": 1.9359817951917648, "logit_z_loss": 36.9172248840332, "router_aux_loss": 1.4256311357021332, "router_z_loss": 0.9064658656716347, "routing_entropy_normalized": 0.6697629290684755, "learning_rate": 0.003, "grad_norm": 1.026023268699646, "tokens_per_second": 4134534.828014853, "eta_seconds": 3119.452450275421, "data_wait_seconds": 2.1576881408691406e-05} {"step": 22470, "total_steps": 28610, "tokens": 47123005440, "source": "finemath", "phase": 4, "loss": 2.2144804298877716, "ce_loss": 2.1995934545993805, "logit_z_loss": 37.81296157836914, "router_aux_loss": 1.3974678218364716, "router_z_loss": 0.9122863113880157, "routing_entropy_normalized": 0.6682341235055189, "learning_rate": 0.003, "grad_norm": 0.9187113642692566, "tokens_per_second": 3967418.1808524574, "eta_seconds": 3245.5649223327637, "data_wait_seconds": 1.9550323486328125e-05} {"step": 22480, "total_steps": 28610, "tokens": 47143976960, "source": "dclm", "phase": 4, "loss": 3.4533991515636444, "ce_loss": 3.437931329011917, "logit_z_loss": 27.043039560317993, "router_aux_loss": 1.4283199906349182, "router_z_loss": 1.1846591234207153, "routing_entropy_normalized": 0.6418429552358689, "learning_rate": 0.003, "grad_norm": 1.0849939584732056, "tokens_per_second": 4147331.4105723677, "eta_seconds": 3099.7141263484955, "data_wait_seconds": 2.142786979675293e-05} {"step": 22490, "total_steps": 28610, "tokens": 47164948480, "source": "fineweb_edu", "phase": 4, "loss": 3.3706354200839996, "ce_loss": 3.3557815849781036, "logit_z_loss": 25.94359064102173, "router_aux_loss": 1.3393368273973465, "router_z_loss": 1.4604419618844986, "routing_entropy_normalized": 0.6484998986506995, "learning_rate": 0.003, "grad_norm": 1.164960265159607, "tokens_per_second": 4121702.051048961, "eta_seconds": 3113.9005393981934, "data_wait_seconds": 2.4050474166870117e-05} {"step": 22500, "total_steps": 28610, "tokens": 47185920000, "source": "fineweb_edu", "phase": 4, "loss": 3.3894561529159546, "ce_loss": 3.375181257724762, "logit_z_loss": 26.084784746170044, "router_aux_loss": 1.292901024222374, "router_z_loss": 1.3458237498998642, "routing_entropy_normalized": 0.6778829867269209, "learning_rate": 0.003, "grad_norm": 0.9710096716880798, "tokens_per_second": 4103287.4711099383, "eta_seconds": 3122.7640788555145, "data_wait_seconds": 2.0235776901245117e-05} {"step": 22510, "total_steps": 28610, "tokens": 47206891520, "source": "finemath", "phase": 4, "loss": 2.2453009486198425, "ce_loss": 2.231010913848877, "logit_z_loss": 36.72991180419922, "router_aux_loss": 1.344761922955513, "router_z_loss": 0.842372864484787, "routing_entropy_normalized": 0.6648183212817913, "learning_rate": 0.003, "grad_norm": 0.7944422364234924, "tokens_per_second": 4461641.635566963, "eta_seconds": 2867.246687412262, "data_wait_seconds": 2.1010637283325195e-05} {"step": 22520, "total_steps": 28610, "tokens": 47227863040, "source": "fineweb_edu", "phase": 4, "loss": 3.361316740512848, "ce_loss": 3.3463995456695557, "logit_z_loss": 26.57961392402649, "router_aux_loss": 1.3717196434736252, "router_z_loss": 1.2000092417001724, "routing_entropy_normalized": 0.6409204786765772, "learning_rate": 0.003, "grad_norm": 1.3992114067077637, "tokens_per_second": 3814645.665495542, "eta_seconds": 3348.0581946372986, "data_wait_seconds": 1.9818544387817383e-05} {"step": 22530, "total_steps": 28610, "tokens": 47248834560, "source": "fineweb_hq", "phase": 4, "loss": 3.42471906542778, "ce_loss": 3.408492475748062, "logit_z_loss": 26.37871026992798, "router_aux_loss": 1.4994838684797287, "router_z_loss": 1.2317306697368622, "routing_entropy_normalized": 0.6379807702711168, "learning_rate": 0.003, "grad_norm": 1.8281935453414917, "tokens_per_second": 4200066.706728907, "eta_seconds": 3035.8289642333984, "data_wait_seconds": 1.9788742065429688e-05} {"step": 22540, "total_steps": 28610, "tokens": 47269806080, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7479628026485443, "ce_loss": 2.731463670730591, "logit_z_loss": 28.755063772201538, "router_aux_loss": 1.4835617989301682, "router_z_loss": 1.6635511219501495, "routing_entropy_normalized": 0.6500589579409153, "learning_rate": 0.003, "grad_norm": 0.9811922907829285, "tokens_per_second": 3820761.5372260557, "eta_seconds": 3331.721311569214, "data_wait_seconds": 2.0623207092285156e-05} {"step": 22550, "total_steps": 28610, "tokens": 47290777600, "source": "fineweb_edu", "phase": 4, "loss": 3.342976927757263, "ce_loss": 3.327814996242523, "logit_z_loss": 27.14179015159607, "router_aux_loss": 1.3642061948776245, "router_z_loss": 1.519840121269226, "routing_entropy_normalized": 0.6434440688130081, "learning_rate": 0.003, "grad_norm": 0.8145994544029236, "tokens_per_second": 4368784.393238091, "eta_seconds": 2908.987941741944, "data_wait_seconds": 2.3663043975830078e-05} {"step": 22560, "total_steps": 28610, "tokens": 47311749120, "source": "finemath", "phase": 4, "loss": 2.2129344642162323, "ce_loss": 2.1986387073993683, "logit_z_loss": 37.94785499572754, "router_aux_loss": 1.3345388621091843, "router_z_loss": 0.9503826722502708, "routing_entropy_normalized": 0.665467770145178, "learning_rate": 0.003, "grad_norm": 0.8657161593437195, "tokens_per_second": 3920010.798280187, "eta_seconds": 3236.667002439499, "data_wait_seconds": 1.9371509552001953e-05} {"step": 22570, "total_steps": 28610, "tokens": 47332720640, "source": "finemath", "phase": 4, "loss": 2.212635487318039, "ce_loss": 2.198678344488144, "logit_z_loss": 39.20993947982788, "router_aux_loss": 1.306490734219551, "router_z_loss": 0.8922840505838394, "routing_entropy_normalized": 0.6950094741159574, "learning_rate": 0.003, "grad_norm": 0.442603200674057, "tokens_per_second": 3519894.586037603, "eta_seconds": 3598.6299505233765, "data_wait_seconds": 2.3663043975830078e-05} {"step": 22580, "total_steps": 28610, "tokens": 47353692160, "source": "dclm", "phase": 4, "loss": 3.386483669281006, "ce_loss": 3.37153896689415, "logit_z_loss": 27.707908868789673, "router_aux_loss": 1.3871458321809769, "router_z_loss": 1.0732295215129852, "routing_entropy_normalized": 0.6766151377743074, "learning_rate": 0.003, "grad_norm": 1.38306725025177, "tokens_per_second": 4050645.157063302, "eta_seconds": 3121.9289939403534, "data_wait_seconds": 2.2619962692260742e-05} {"step": 22590, "total_steps": 28610, "tokens": 47374663680, "source": "fineweb_edu", "phase": 4, "loss": 3.321212649345398, "ce_loss": 3.3045035302639008, "logit_z_loss": 27.46743106842041, "router_aux_loss": 1.5298473834991455, "router_z_loss": 1.4106529653072357, "routing_entropy_normalized": 0.6413561355911503, "learning_rate": 0.003, "grad_norm": 1.1862436532974243, "tokens_per_second": 4304475.183885421, "eta_seconds": 2932.9603495597844, "data_wait_seconds": 1.9103288650512695e-05} {"step": 22600, "total_steps": 28610, "tokens": 47395635200, "source": "fineweb_edu", "phase": 4, "loss": 3.341043621301651, "ce_loss": 3.32455974817276, "logit_z_loss": 27.67742395401001, "router_aux_loss": 1.4911002665758133, "router_z_loss": 1.5728380531072617, "routing_entropy_normalized": 0.6451295915601802, "learning_rate": 0.003, "grad_norm": 1.9018237590789795, "tokens_per_second": 3598998.3393916106, "eta_seconds": 3502.0531635284424, "data_wait_seconds": 2.2023916244506836e-05} {"step": 22610, "total_steps": 28610, "tokens": 47416606720, "source": "finemath", "phase": 4, "loss": 2.2391088604927063, "ce_loss": 2.2234262824058533, "logit_z_loss": 37.82977056503296, "router_aux_loss": 1.4517877697944641, "router_z_loss": 1.164679616689682, "routing_entropy_normalized": 0.6458292070392871, "learning_rate": 0.003, "grad_norm": 0.7667958736419678, "tokens_per_second": 4285530.966949483, "eta_seconds": 2936.138391494751, "data_wait_seconds": 2.256035804748535e-05} {"step": 22620, "total_steps": 28610, "tokens": 47437578240, "source": "fineweb_edu", "phase": 4, "loss": 3.3577503859996796, "ce_loss": 3.3425722122192383, "logit_z_loss": 27.149210929870605, "router_aux_loss": 1.3515155613422394, "router_z_loss": 1.6630393117666245, "routing_entropy_normalized": 0.6651335910483812, "learning_rate": 0.003, "grad_norm": 0.6919260025024414, "tokens_per_second": 3580139.1893447456, "eta_seconds": 3508.785501241684, "data_wait_seconds": 1.9997358322143555e-05} {"step": 22630, "total_steps": 28610, "tokens": 47458549760, "source": "fineweb_hq", "phase": 4, "loss": 3.402729630470276, "ce_loss": 3.38810196518898, "logit_z_loss": 27.41441321372986, "router_aux_loss": 1.3000224083662033, "router_z_loss": 1.6274041682481766, "routing_entropy_normalized": 0.6706555269823978, "learning_rate": 0.003, "grad_norm": 2.010157346725464, "tokens_per_second": 4450470.156744853, "eta_seconds": 2817.897552013397, "data_wait_seconds": 1.7315149307250977e-05} {"step": 22640, "total_steps": 28610, "tokens": 47479521280, "source": "cosmopedia_v2", "phase": 4, "loss": 2.766642302274704, "ce_loss": 2.751255542039871, "logit_z_loss": 28.878023624420166, "router_aux_loss": 1.3597510159015656, "router_z_loss": 1.7892342954874039, "routing_entropy_normalized": 0.6621997056753237, "learning_rate": 0.003, "grad_norm": 2.1354172229766846, "tokens_per_second": 3970635.4042809824, "eta_seconds": 3153.1470823287964, "data_wait_seconds": 2.300739288330078e-05} {"step": 22650, "total_steps": 28610, "tokens": 47500492800, "source": "fineweb_edu", "phase": 4, "loss": 3.3604792654514313, "ce_loss": 3.3444503247737885, "logit_z_loss": 27.24688696861267, "router_aux_loss": 1.4399488270282745, "router_z_loss": 1.6294687688350677, "routing_entropy_normalized": 0.6360704433129042, "learning_rate": 0.003, "grad_norm": 1.5287559032440186, "tokens_per_second": 3845886.441307217, "eta_seconds": 3249.9726943969727, "data_wait_seconds": 2.0772218704223633e-05} {"step": 22660, "total_steps": 28610, "tokens": 47521464320, "source": "finemath", "phase": 4, "loss": 2.277803659439087, "ce_loss": 2.2616065442562103, "logit_z_loss": 38.57940721511841, "router_aux_loss": 1.4957182109355927, "router_z_loss": 1.2399621456861496, "routing_entropy_normalized": 0.6434866472936024, "learning_rate": 0.003, "grad_norm": 0.7186468839645386, "tokens_per_second": 4109917.4596605306, "eta_seconds": 3036.0839414596558, "data_wait_seconds": 1.990795135498047e-05} {"step": 22670, "total_steps": 28610, "tokens": 47542435840, "source": "finemath", "phase": 4, "loss": 2.633852854371071, "ce_loss": 2.618733048439026, "logit_z_loss": 35.7531795501709, "router_aux_loss": 1.3836952447891235, "router_z_loss": 1.2828577235341072, "routing_entropy_normalized": 0.6429387606970245, "learning_rate": 0.003, "grad_norm": 0.8280866146087646, "tokens_per_second": 3734638.0972738108, "eta_seconds": 3335.552885055542, "data_wait_seconds": 2.3245811462402344e-05} {"step": 22680, "total_steps": 28610, "tokens": 47563407360, "source": "dclm", "phase": 4, "loss": 3.4705302715301514, "ce_loss": 3.4559099972248077, "logit_z_loss": 29.220645904541016, "router_aux_loss": 1.3243972063064575, "router_z_loss": 1.376244530081749, "routing_entropy_normalized": 0.6555669916423886, "learning_rate": 0.003, "grad_norm": 1.1858717203140259, "tokens_per_second": 4021954.1791086746, "eta_seconds": 3092.056946992874, "data_wait_seconds": 2.256035804748535e-05} {"step": 22690, "total_steps": 28610, "tokens": 47584378880, "source": "fineweb_edu", "phase": 4, "loss": 3.361368954181671, "ce_loss": 3.3461726307868958, "logit_z_loss": 28.01148796081543, "router_aux_loss": 1.3705943822860718, "router_z_loss": 1.490433931350708, "routing_entropy_normalized": 0.6767210655526512, "learning_rate": 0.003, "grad_norm": 0.8593534827232361, "tokens_per_second": 4002129.6337812566, "eta_seconds": 3102.133358001709, "data_wait_seconds": 2.294778823852539e-05} {"step": 22700, "total_steps": 28610, "tokens": 47605350400, "source": "fineweb_edu", "phase": 4, "loss": 3.360249549150467, "ce_loss": 3.3447659015655518, "logit_z_loss": 26.77783489227295, "router_aux_loss": 1.3894461244344711, "router_z_loss": 1.5892234146595001, "routing_entropy_normalized": 0.6427624535632416, "learning_rate": 0.003, "grad_norm": 1.4836894273757935, "tokens_per_second": 4065097.354589618, "eta_seconds": 3048.922876596451, "data_wait_seconds": 2.193450927734375e-05} {"step": 22710, "total_steps": 28610, "tokens": 47626321920, "source": "finemath", "phase": 4, "loss": 2.242537021636963, "ce_loss": 2.2268651723861694, "logit_z_loss": 34.98934984207153, "router_aux_loss": 1.4633736908435822, "router_z_loss": 1.0380851477384567, "routing_entropy_normalized": 0.6565159186024018, "learning_rate": 0.003, "grad_norm": 1.0459027290344238, "tokens_per_second": 3759721.3895571297, "eta_seconds": 3290.987687110901, "data_wait_seconds": 2.2113323211669922e-05} {"step": 22720, "total_steps": 28610, "tokens": 47647293440, "source": "fineweb_edu", "phase": 4, "loss": 3.356048047542572, "ce_loss": 3.340482383966446, "logit_z_loss": 26.603203773498535, "router_aux_loss": 1.4084879606962204, "router_z_loss": 1.480723038315773, "routing_entropy_normalized": 0.6740739541885387, "learning_rate": 0.003, "grad_norm": 1.371335744857788, "tokens_per_second": 3897995.474902311, "eta_seconds": 3168.8659875392914, "data_wait_seconds": 2.0891427993774414e-05} {"step": 22730, "total_steps": 28610, "tokens": 47668264960, "source": "fineweb_hq", "phase": 4, "loss": 3.273106336593628, "ce_loss": 3.2571626603603363, "logit_z_loss": 27.01270365715027, "router_aux_loss": 1.4411029517650604, "router_z_loss": 1.5326578170061111, "routing_entropy_normalized": 0.6479661469085068, "learning_rate": 0.003, "grad_norm": 1.398145079612732, "tokens_per_second": 4004012.62926153, "eta_seconds": 3079.7239923477173, "data_wait_seconds": 2.1010637283325195e-05} {"step": 22740, "total_steps": 28610, "tokens": 47689236480, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7510935366153717, "ce_loss": 2.735222578048706, "logit_z_loss": 28.913172006607056, "router_aux_loss": 1.3821041285991669, "router_z_loss": 2.049932688474655, "routing_entropy_normalized": 0.657681223322742, "learning_rate": 0.003, "grad_norm": 0.893422544002533, "tokens_per_second": 4119226.18288843, "eta_seconds": 2988.49388051033, "data_wait_seconds": 2.4110078811645508e-05} {"step": 22750, "total_steps": 28610, "tokens": 47710208000, "source": "fineweb_edu", "phase": 4, "loss": 3.375121533870697, "ce_loss": 3.359641432762146, "logit_z_loss": 27.21840810775757, "router_aux_loss": 1.3736704736948013, "router_z_loss": 1.743395745754242, "routing_entropy_normalized": 0.6492525033566174, "learning_rate": 0.003, "grad_norm": 0.7611470222473145, "tokens_per_second": 3959738.502556719, "eta_seconds": 3103.5662360191345, "data_wait_seconds": 1.749396324157715e-05} {"step": 22760, "total_steps": 28610, "tokens": 47731179520, "source": "finemath", "phase": 4, "loss": 2.179869681596756, "ce_loss": 2.1645184755325317, "logit_z_loss": 36.63768911361694, "router_aux_loss": 1.4217701107263565, "router_z_loss": 1.133524127304554, "routing_entropy_normalized": 0.6751033868245457, "learning_rate": 0.003, "grad_norm": 1.1216049194335938, "tokens_per_second": 4493149.8691800255, "eta_seconds": 2730.4540371894836, "data_wait_seconds": 2.1398067474365234e-05} {"step": 22770, "total_steps": 28610, "tokens": 47752151040, "source": "finemath", "phase": 4, "loss": 2.2354533076286316, "ce_loss": 2.2206874787807465, "logit_z_loss": 38.48386764526367, "router_aux_loss": 1.3660152852535248, "router_z_loss": 1.105655163526535, "routing_entropy_normalized": 0.6597307210591096, "learning_rate": 0.003, "grad_norm": 0.6442807912826538, "tokens_per_second": 3511010.2985784113, "eta_seconds": 3488.274496078491, "data_wait_seconds": 2.086162567138672e-05} {"step": 22780, "total_steps": 28610, "tokens": 47773122560, "source": "dclm", "phase": 4, "loss": 3.457775205373764, "ce_loss": 3.442451775074005, "logit_z_loss": 29.219255447387695, "router_aux_loss": 1.3985627740621567, "router_z_loss": 1.3377987891435623, "routing_entropy_normalized": 0.6189222906722379, "learning_rate": 0.003, "grad_norm": 1.2232402563095093, "tokens_per_second": 4428219.290293235, "eta_seconds": 2761.0186755657196, "data_wait_seconds": 2.3365020751953125e-05} {"step": 22790, "total_steps": 28610, "tokens": 47794094080, "source": "fineweb_edu", "phase": 4, "loss": 3.3649469017982483, "ce_loss": 3.3495078682899475, "logit_z_loss": 27.890451908111572, "router_aux_loss": 1.390110582113266, "router_z_loss": 1.53795725107193, "routing_entropy_normalized": 0.6748666846043173, "learning_rate": 0.003, "grad_norm": 1.0636646747589111, "tokens_per_second": 3647011.5810084087, "eta_seconds": 3346.6920433044434, "data_wait_seconds": 2.2470951080322266e-05} {"step": 22800, "total_steps": 28610, "tokens": 47815065600, "source": "fineweb_edu", "phase": 4, "loss": 3.3500682711601257, "ce_loss": 3.3340629935264587, "logit_z_loss": 26.630918741226196, "router_aux_loss": 1.4469333440065384, "router_z_loss": 1.5359549969434738, "routing_entropy_normalized": 0.6509270286678825, "learning_rate": 0.003, "grad_norm": 0.7835044860839844, "tokens_per_second": 4313491.509464301, "eta_seconds": 2824.7309849262238, "data_wait_seconds": 2.0742416381835938e-05} {"step": 22810, "total_steps": 28610, "tokens": 47836037120, "source": "finemath", "phase": 4, "loss": 2.2451752424240112, "ce_loss": 2.2297028303146362, "logit_z_loss": 36.72743797302246, "router_aux_loss": 1.4420601725578308, "router_z_loss": 1.0518247559666634, "routing_entropy_normalized": 0.6628969181970392, "learning_rate": 0.003, "grad_norm": 1.0053273439407349, "tokens_per_second": 3940164.170072279, "eta_seconds": 3087.0494413375854, "data_wait_seconds": 2.136826515197754e-05} {"step": 22820, "total_steps": 28610, "tokens": 47857008640, "source": "fineweb_edu", "phase": 4, "loss": 3.359955608844757, "ce_loss": 3.3446883857250214, "logit_z_loss": 25.89076542854309, "router_aux_loss": 1.361240342259407, "router_z_loss": 1.6547902822494507, "routing_entropy_normalized": 0.6685649935342035, "learning_rate": 0.003, "grad_norm": 1.2787318229675293, "tokens_per_second": 3930028.285184964, "eta_seconds": 3089.67498421669, "data_wait_seconds": 2.0444393157958984e-05} {"step": 22830, "total_steps": 28610, "tokens": 47877980160, "source": "fineweb_hq", "phase": 4, "loss": 3.332113116979599, "ce_loss": 3.316327214241028, "logit_z_loss": 27.02272939682007, "router_aux_loss": 1.416137456893921, "router_z_loss": 1.6245354264974594, "routing_entropy_normalized": 0.6468055627156887, "learning_rate": 0.003, "grad_norm": 0.8662893772125244, "tokens_per_second": 4185194.029799161, "eta_seconds": 2896.2907032966614, "data_wait_seconds": 2.2798776626586914e-05} {"step": 22840, "total_steps": 28610, "tokens": 47898951680, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7663473784923553, "ce_loss": 2.7498941123485565, "logit_z_loss": 29.071274995803833, "router_aux_loss": 1.4336140155792236, "router_z_loss": 2.117041528224945, "routing_entropy_normalized": 0.6452176850074232, "learning_rate": 0.003, "grad_norm": 1.1932158470153809, "tokens_per_second": 4079135.042477788, "eta_seconds": 2966.4541413784027, "data_wait_seconds": 2.1696090698242188e-05} {"step": 22850, "total_steps": 28610, "tokens": 47919923200, "source": "fineweb_edu", "phase": 4, "loss": 3.3514400124549866, "ce_loss": 3.335344046354294, "logit_z_loss": 26.327279567718506, "router_aux_loss": 1.4342511594295502, "router_z_loss": 1.7534361630678177, "routing_entropy_normalized": 0.6793689181299077, "learning_rate": 0.003, "grad_norm": 0.6395246386528015, "tokens_per_second": 4099062.147764688, "eta_seconds": 2946.9169006347656, "data_wait_seconds": 1.755356788635254e-05} {"step": 22860, "total_steps": 28610, "tokens": 47940894720, "source": "finemath", "phase": 4, "loss": 2.144350305199623, "ce_loss": 2.1288577765226364, "logit_z_loss": 37.984644412994385, "router_aux_loss": 1.4171922355890274, "router_z_loss": 1.3205595463514328, "routing_entropy_normalized": 0.6927924721358962, "learning_rate": 0.003, "grad_norm": 0.8688344955444336, "tokens_per_second": 4209422.297933595, "eta_seconds": 2864.6743297576904, "data_wait_seconds": 2.205371856689453e-05} {"step": 22870, "total_steps": 28610, "tokens": 47961866240, "source": "finemath", "phase": 4, "loss": 2.2441916465759277, "ce_loss": 2.2295046150684357, "logit_z_loss": 38.967329025268555, "router_aux_loss": 1.342617318034172, "router_z_loss": 1.2608472257852554, "routing_entropy_normalized": 0.6695970843727891, "learning_rate": 0.003, "grad_norm": 0.890259861946106, "tokens_per_second": 3502512.54524449, "eta_seconds": 3436.8620595932007, "data_wait_seconds": 2.0384788513183594e-05} {"step": 22880, "total_steps": 28610, "tokens": 47982837760, "source": "dclm", "phase": 4, "loss": 3.5427840650081635, "ce_loss": 3.5280887484550476, "logit_z_loss": 26.77886986732483, "router_aux_loss": 1.3166422545909882, "router_z_loss": 1.5288844853639603, "routing_entropy_normalized": 0.6523703754449571, "learning_rate": 0.003, "grad_norm": 0.919354259967804, "tokens_per_second": 4342744.911712031, "eta_seconds": 2767.070413827896, "data_wait_seconds": 2.339482307434082e-05} {"step": 22890, "total_steps": 28610, "tokens": 48003809280, "source": "fineweb_edu", "phase": 4, "loss": 3.36860054731369, "ce_loss": 3.3539289236068726, "logit_z_loss": 26.58346438407898, "router_aux_loss": 1.2999173700809479, "router_z_loss": 1.6724493354558945, "routing_entropy_normalized": 0.6878326656549416, "learning_rate": 0.003, "grad_norm": 0.8439992666244507, "tokens_per_second": 4837087.144299958, "eta_seconds": 2479.9448680877686, "data_wait_seconds": 2.2649765014648438e-05} {"step": 22900, "total_steps": 28610, "tokens": 48024780800, "source": "fineweb_edu", "phase": 4, "loss": 3.359216123819351, "ce_loss": 3.344190150499344, "logit_z_loss": 27.002694606781006, "router_aux_loss": 1.345867559313774, "router_z_loss": 1.5672447979450226, "routing_entropy_normalized": 0.6637044293164647, "learning_rate": 0.003, "grad_norm": 0.7097606062889099, "tokens_per_second": 4558568.99668336, "eta_seconds": 2626.863370656967, "data_wait_seconds": 2.3066997528076172e-05} {"step": 22910, "total_steps": 28610, "tokens": 48045752320, "source": "finemath", "phase": 4, "loss": 2.1369677037000656, "ce_loss": 2.122274160385132, "logit_z_loss": 37.45504188537598, "router_aux_loss": 1.3435991406440735, "router_z_loss": 1.2574709951877594, "routing_entropy_normalized": 0.6654842764418364, "learning_rate": 0.003, "grad_norm": 1.0220228433609009, "tokens_per_second": 3795780.2995769363, "eta_seconds": 3149.22504901886, "data_wait_seconds": 2.2083520889282227e-05} {"step": 22920, "total_steps": 28610, "tokens": 48066723840, "source": "fineweb_edu", "phase": 4, "loss": 3.3425224125385284, "ce_loss": 3.3279957473278046, "logit_z_loss": 27.033422231674194, "router_aux_loss": 1.3184219598770142, "router_z_loss": 1.3424232304096222, "routing_entropy_normalized": 0.6805479448038521, "learning_rate": 0.003, "grad_norm": 1.1729220151901245, "tokens_per_second": 3643430.483910879, "eta_seconds": 3275.1537137031555, "data_wait_seconds": 2.065300941467285e-05} {"step": 22930, "total_steps": 28610, "tokens": 48087695360, "source": "fineweb_hq", "phase": 4, "loss": 3.3322125673294067, "ce_loss": 3.317107707262039, "logit_z_loss": 25.749825954437256, "router_aux_loss": 1.3818436563014984, "router_z_loss": 1.2863891273736954, "routing_entropy_normalized": 0.6902671087170424, "learning_rate": 0.003, "grad_norm": 2.0228934288024902, "tokens_per_second": 3987573.473509233, "eta_seconds": 2987.2360820770264, "data_wait_seconds": 2.0235776901245117e-05} {"step": 22940, "total_steps": 28610, "tokens": 48108666880, "source": "cosmopedia_v2", "phase": 4, "loss": 2.777586668729782, "ce_loss": 2.762269973754883, "logit_z_loss": 28.158589363098145, "router_aux_loss": 1.3564116209745407, "router_z_loss": 1.7525810301303864, "routing_entropy_normalized": 0.6655837021671545, "learning_rate": 0.003, "grad_norm": 1.7478984594345093, "tokens_per_second": 3761597.2362422724, "eta_seconds": 3161.1177628040314, "data_wait_seconds": 1.963973045349121e-05} {"step": 22950, "total_steps": 28610, "tokens": 48129638400, "source": "fineweb_edu", "phase": 4, "loss": 3.3232769668102264, "ce_loss": 3.308992564678192, "logit_z_loss": 26.554158926010132, "router_aux_loss": 1.2871545106172562, "router_z_loss": 1.4129198491573334, "routing_entropy_normalized": 0.7286616540937848, "learning_rate": 0.003, "grad_norm": 0.8172850012779236, "tokens_per_second": 4120066.258643277, "eta_seconds": 2880.992579936981, "data_wait_seconds": 2.2411346435546875e-05} {"step": 22960, "total_steps": 28610, "tokens": 48150609920, "source": "finemath", "phase": 4, "loss": 2.269815295934677, "ce_loss": 2.255439817905426, "logit_z_loss": 38.18319654464722, "router_aux_loss": 1.3458067327737808, "router_z_loss": 0.9173872619867325, "routing_entropy_normalized": 0.6523177858605429, "learning_rate": 0.003, "grad_norm": 1.1654506921768188, "tokens_per_second": 3748599.892805496, "eta_seconds": 3160.8891689777374, "data_wait_seconds": 2.1249055862426758e-05} {"step": 22970, "total_steps": 28610, "tokens": 48171581440, "source": "finemath", "phase": 4, "loss": 2.2591316998004913, "ce_loss": 2.2436634302139282, "logit_z_loss": 38.79840326309204, "router_aux_loss": 1.4622911214828491, "router_z_loss": 0.8453480750322342, "routing_entropy_normalized": 0.6494455594754127, "learning_rate": 0.003, "grad_norm": 0.9305398464202881, "tokens_per_second": 3510462.9815927334, "eta_seconds": 3369.338273048401, "data_wait_seconds": 2.03549861907959e-05} {"step": 22980, "total_steps": 28610, "tokens": 48192552960, "source": "dclm", "phase": 4, "loss": 3.5381573140621185, "ce_loss": 3.522772580385208, "logit_z_loss": 28.275993585586548, "router_aux_loss": 1.41944919526577, "router_z_loss": 1.1902692168951035, "routing_entropy_normalized": 0.6506093859068759, "learning_rate": 0.003, "grad_norm": 0.7353929281234741, "tokens_per_second": 4127458.194071257, "eta_seconds": 2860.590030193329, "data_wait_seconds": 1.9937753677368164e-05} {"step": 22990, "total_steps": 28610, "tokens": 48213524480, "source": "fineweb_edu", "phase": 4, "loss": 3.452012211084366, "ce_loss": 3.435834974050522, "logit_z_loss": 26.502204418182373, "router_aux_loss": 1.4565875679254532, "router_z_loss": 1.6113817393779755, "routing_entropy_normalized": 0.6645503778128757, "learning_rate": 0.003, "grad_norm": 1.742901086807251, "tokens_per_second": 3464543.327212022, "eta_seconds": 3401.889694213867, "data_wait_seconds": 2.6345252990722656e-05} {"step": 23000, "total_steps": 28610, "tokens": 48234496000, "source": "fineweb_edu", "phase": 4, "loss": 3.3435383439064026, "ce_loss": 3.326835811138153, "logit_z_loss": 27.345640420913696, "router_aux_loss": 1.499692216515541, "router_z_loss": 1.7056200057268143, "routing_entropy_normalized": 0.6609131205937376, "learning_rate": 0.003, "grad_norm": 1.5432653427124023, "tokens_per_second": 4390595.889409467, "eta_seconds": 2679.595894575119, "data_wait_seconds": 2.294778823852539e-05} {"step": 23010, "total_steps": 28610, "tokens": 48255467520, "source": "finemath", "phase": 4, "loss": 2.350366860628128, "ce_loss": 2.3337371349334717, "logit_z_loss": 36.39582872390747, "router_aux_loss": 1.5290085971355438, "router_z_loss": 1.3396079242229462, "routing_entropy_normalized": 0.6748021388786226, "learning_rate": 0.003, "grad_norm": 1.2210935354232788, "tokens_per_second": 4179413.055352405, "eta_seconds": 2809.9761962890625, "data_wait_seconds": 2.2232532501220703e-05} {"step": 23020, "total_steps": 28610, "tokens": 48276439040, "source": "fineweb_edu", "phase": 4, "loss": 3.335907757282257, "ce_loss": 3.3198872208595276, "logit_z_loss": 25.62678551673889, "router_aux_loss": 1.411961242556572, "router_z_loss": 1.9009537994861603, "routing_entropy_normalized": 0.6755873245795555, "learning_rate": 0.003, "grad_norm": 0.6928481459617615, "tokens_per_second": 4102809.1828667456, "eta_seconds": 2857.3299798965454, "data_wait_seconds": 1.9818544387817383e-05} {"step": 23030, "total_steps": 28610, "tokens": 48297410560, "source": "fineweb_hq", "phase": 4, "loss": 3.3651739358901978, "ce_loss": 3.3498644530773163, "logit_z_loss": 27.525084257125854, "router_aux_loss": 1.3381531685590744, "router_z_loss": 1.9279445856809616, "routing_entropy_normalized": 0.6678711257502736, "learning_rate": 0.003, "grad_norm": 1.480567216873169, "tokens_per_second": 4403318.824898993, "eta_seconds": 2657.5654921531677, "data_wait_seconds": 2.0325183868408203e-05} {"step": 23040, "total_steps": 28610, "tokens": 48318382080, "source": "cosmopedia_v2", "phase": 4, "loss": 2.743141323328018, "ce_loss": 2.7274342477321625, "logit_z_loss": 28.73332166671753, "router_aux_loss": 1.3535370379686356, "router_z_loss": 2.1717576682567596, "routing_entropy_normalized": 0.6719471331839769, "learning_rate": 0.003, "grad_norm": 1.1213092803955078, "tokens_per_second": 3409880.2608972746, "eta_seconds": 3425.67355632782, "data_wait_seconds": 2.047419548034668e-05} {"step": 23050, "total_steps": 28610, "tokens": 48339353600, "source": "fineweb_edu", "phase": 4, "loss": 3.3508753776550293, "ce_loss": 3.3356948792934418, "logit_z_loss": 25.956010341644287, "router_aux_loss": 1.3486464023590088, "router_z_loss": 1.6940303891897202, "routing_entropy_normalized": 0.6765677605893939, "learning_rate": 0.003, "grad_norm": 1.0898315906524658, "tokens_per_second": 3303183.0399222025, "eta_seconds": 3529.9785022735596, "data_wait_seconds": 1.9222497940063477e-05} {"step": 23060, "total_steps": 28610, "tokens": 48360325120, "source": "finemath", "phase": 4, "loss": 2.257388710975647, "ce_loss": 2.2427509427070618, "logit_z_loss": 37.36021184921265, "router_aux_loss": 1.3459347486495972, "router_z_loss": 1.1784584820270538, "routing_entropy_normalized": 0.6812503056315833, "learning_rate": 0.003, "grad_norm": 0.818983793258667, "tokens_per_second": 3441079.9424161157, "eta_seconds": 3382.4246442317963, "data_wait_seconds": 2.1696090698242188e-05} {"step": 23070, "total_steps": 28610, "tokens": 48381296640, "source": "finemath", "phase": 4, "loss": 2.2479296028614044, "ce_loss": 2.2339647710323334, "logit_z_loss": 38.435996532440186, "router_aux_loss": 1.2776529639959335, "router_z_loss": 1.1883207708597183, "routing_entropy_normalized": 0.6815515060059814, "learning_rate": 0.003, "grad_norm": 0.7078059911727905, "tokens_per_second": 3799825.5374121615, "eta_seconds": 3057.5672397613525, "data_wait_seconds": 1.919269561767578e-05} {"step": 23080, "total_steps": 28610, "tokens": 48402268160, "source": "dclm", "phase": 4, "loss": 3.5239308774471283, "ce_loss": 3.509167194366455, "logit_z_loss": 27.691343069076538, "router_aux_loss": 1.3422816842794418, "router_z_loss": 1.3407994955778122, "routing_entropy_normalized": 0.6424254785815603, "learning_rate": 0.003, "grad_norm": 0.9219207167625427, "tokens_per_second": 3719518.7047010353, "eta_seconds": 3117.9438741207123, "data_wait_seconds": 2.041459083557129e-05} {"step": 23090, "total_steps": 28610, "tokens": 48423239680, "source": "fineweb_edu", "phase": 4, "loss": 3.3230053782463074, "ce_loss": 3.3075586557388306, "logit_z_loss": 27.33680248260498, "router_aux_loss": 1.3914740681648254, "router_z_loss": 1.5319636911153793, "routing_entropy_normalized": 0.6594634479328182, "learning_rate": 0.003, "grad_norm": 1.165132761001587, "tokens_per_second": 3930343.495791771, "eta_seconds": 2945.360641479492, "data_wait_seconds": 2.2292137145996094e-05} {"step": 23100, "total_steps": 28610, "tokens": 48444211200, "source": "fineweb_edu", "phase": 4, "loss": 3.346495360136032, "ce_loss": 3.3318898379802704, "logit_z_loss": 26.480684995651245, "router_aux_loss": 1.3132445812225342, "router_z_loss": 1.4731020480394363, "routing_entropy_normalized": 0.6806100294041112, "learning_rate": 0.003, "grad_norm": 1.3981248140335083, "tokens_per_second": 3347802.0166023495, "eta_seconds": 3451.6101796627045, "data_wait_seconds": 2.205371856689453e-05} {"step": 23110, "total_steps": 28610, "tokens": 48465182720, "source": "finemath", "phase": 4, "loss": 2.235041379928589, "ce_loss": 2.220345765352249, "logit_z_loss": 39.68204927444458, "router_aux_loss": 1.379076987504959, "router_z_loss": 0.9047924801707268, "routing_entropy_normalized": 0.6739687816170895, "learning_rate": 0.003, "grad_norm": 0.6910995244979858, "tokens_per_second": 3296031.176845678, "eta_seconds": 3499.462044239044, "data_wait_seconds": 2.092123031616211e-05} {"step": 23120, "total_steps": 28610, "tokens": 48486154240, "source": "fineweb_edu", "phase": 4, "loss": 3.3195525109767914, "ce_loss": 3.3050214648246765, "logit_z_loss": 26.338698625564575, "router_aux_loss": 1.315945565700531, "router_z_loss": 1.3715842515230179, "routing_entropy_normalized": 0.6736956355502309, "learning_rate": 0.003, "grad_norm": 1.344092607498169, "tokens_per_second": 4074763.721889477, "eta_seconds": 2825.5293474197388, "data_wait_seconds": 2.199411392211914e-05} {"step": 23130, "total_steps": 28610, "tokens": 48507125760, "source": "fineweb_hq", "phase": 4, "loss": 3.3307289481163025, "ce_loss": 3.3163532614707947, "logit_z_loss": 25.751184463500977, "router_aux_loss": 1.3011726140975952, "router_z_loss": 1.3639682531356812, "routing_entropy_normalized": 0.6594942989599489, "learning_rate": 0.003, "grad_norm": 1.6467087268829346, "tokens_per_second": 3743400.2934461143, "eta_seconds": 3070.04115486145, "data_wait_seconds": 2.1666288375854492e-05} {"step": 23140, "total_steps": 28610, "tokens": 48528097280, "source": "cosmopedia_v2", "phase": 4, "loss": 2.746301770210266, "ce_loss": 2.7313636243343353, "logit_z_loss": 27.780094623565674, "router_aux_loss": 1.3418407291173935, "router_z_loss": 1.5197174102067947, "routing_entropy_normalized": 0.6793783745493791, "learning_rate": 0.003, "grad_norm": 1.2567808628082275, "tokens_per_second": 4117791.2834677016, "eta_seconds": 2785.8190593719482, "data_wait_seconds": 2.014636993408203e-05} {"step": 23150, "total_steps": 28610, "tokens": 48549068800, "source": "fineweb_edu", "phase": 4, "loss": 3.292758584022522, "ce_loss": 3.2781070470809937, "logit_z_loss": 27.298953533172607, "router_aux_loss": 1.3426272124052048, "router_z_loss": 1.2252358049154282, "routing_entropy_normalized": 0.6519963941338547, "learning_rate": 0.003, "grad_norm": 1.1809920072555542, "tokens_per_second": 3999449.023163549, "eta_seconds": 2863.00684261322, "data_wait_seconds": 2.0712614059448242e-05} {"step": 23160, "total_steps": 28610, "tokens": 48570040320, "source": "finemath", "phase": 4, "loss": 2.2038392424583435, "ce_loss": 2.1893189549446106, "logit_z_loss": 37.722416400909424, "router_aux_loss": 1.3736607283353806, "router_z_loss": 0.7837023288011551, "routing_entropy_normalized": 0.6667471887947266, "learning_rate": 0.003, "grad_norm": 0.3767828643321991, "tokens_per_second": 3606344.303597516, "eta_seconds": 3169.2698860168457, "data_wait_seconds": 1.7374753952026367e-05} {"step": 23170, "total_steps": 28610, "tokens": 48591011840, "source": "finemath", "phase": 4, "loss": 1.924600213766098, "ce_loss": 1.9108076244592667, "logit_z_loss": 34.63056421279907, "router_aux_loss": 1.2743513882160187, "router_z_loss": 1.049077644944191, "routing_entropy_normalized": 0.7084286643792133, "learning_rate": 0.003, "grad_norm": 1.2983440160751343, "tokens_per_second": 3609119.364578501, "eta_seconds": 3161.0223236083984, "data_wait_seconds": 2.0444393157958984e-05} {"step": 23180, "total_steps": 28610, "tokens": 48611983360, "source": "dclm", "phase": 4, "loss": 3.529480814933777, "ce_loss": 3.5154334604740143, "logit_z_loss": 26.788973569869995, "router_aux_loss": 1.3059089183807373, "router_z_loss": 0.988282673060894, "routing_entropy_normalized": 0.6760532826552571, "learning_rate": 0.003, "grad_norm": 0.6829658150672913, "tokens_per_second": 3774781.892228027, "eta_seconds": 3016.739956140518, "data_wait_seconds": 1.806020736694336e-05} {"step": 23190, "total_steps": 28610, "tokens": 48632954880, "source": "fineweb_edu", "phase": 4, "loss": 3.3672637939453125, "ce_loss": 3.352897197008133, "logit_z_loss": 26.56668519973755, "router_aux_loss": 1.324481263756752, "router_z_loss": 1.1217533126473427, "routing_entropy_normalized": 0.6676032924299653, "learning_rate": 0.003, "grad_norm": 1.1161327362060547, "tokens_per_second": 4274743.251900238, "eta_seconds": 2659.0050373077397, "data_wait_seconds": 1.9282102584838867e-05} {"step": 23200, "total_steps": 28610, "tokens": 48653926400, "source": "fineweb_edu", "phase": 4, "loss": 3.3094711303710938, "ce_loss": 3.2946344316005707, "logit_z_loss": 26.878405332565308, "router_aux_loss": 1.3697922825813293, "router_z_loss": 1.1387827396392822, "routing_entropy_normalized": 0.639650668408776, "learning_rate": 0.003, "grad_norm": 1.5043307542800903, "tokens_per_second": 4033603.5713100643, "eta_seconds": 2812.7683148384094, "data_wait_seconds": 2.2619962692260742e-05} {"step": 23210, "total_steps": 28610, "tokens": 48674897920, "source": "finemath", "phase": 4, "loss": 2.2824469208717346, "ce_loss": 2.2682115733623505, "logit_z_loss": 37.9662823677063, "router_aux_loss": 1.3503980785608292, "router_z_loss": 0.7313884869217873, "routing_entropy_normalized": 0.6581902670978149, "learning_rate": 0.003, "grad_norm": 1.1854257583618164, "tokens_per_second": 4147627.2919137683, "eta_seconds": 2730.3853511810303, "data_wait_seconds": 2.440810203552246e-05} {"step": 23220, "total_steps": 28610, "tokens": 48695869440, "source": "fineweb_edu", "phase": 4, "loss": 3.3112462759017944, "ce_loss": 3.2970100045204163, "logit_z_loss": 27.193408966064453, "router_aux_loss": 1.3089783787727356, "router_z_loss": 1.146488517522812, "routing_entropy_normalized": 0.6809444137408615, "learning_rate": 0.003, "grad_norm": 1.047549843788147, "tokens_per_second": 3927165.258781202, "eta_seconds": 2878.3227939605713, "data_wait_seconds": 1.8984079360961914e-05} {"step": 23230, "total_steps": 28610, "tokens": 48716840960, "source": "fineweb_hq", "phase": 4, "loss": 3.2944694459438324, "ce_loss": 3.279966026544571, "logit_z_loss": 27.939950227737427, "router_aux_loss": 1.336121916770935, "router_z_loss": 1.1421706229448318, "routing_entropy_normalized": 0.6428933725256132, "learning_rate": 0.003, "grad_norm": 1.082029938697815, "tokens_per_second": 4360193.707442523, "eta_seconds": 2587.6551632881165, "data_wait_seconds": 2.205371856689453e-05} {"step": 23240, "total_steps": 28610, "tokens": 48737812480, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7905896604061127, "ce_loss": 2.776170790195465, "logit_z_loss": 29.989104509353638, "router_aux_loss": 1.3008088916540146, "router_z_loss": 1.4108182191848755, "routing_entropy_normalized": 0.6925837716431972, "learning_rate": 0.003, "grad_norm": 1.6144545078277588, "tokens_per_second": 3941096.651061206, "eta_seconds": 2857.5057241916656, "data_wait_seconds": 2.154707908630371e-05} {"step": 23250, "total_steps": 28610, "tokens": 48758784000, "source": "fineweb_edu", "phase": 4, "loss": 3.340662121772766, "ce_loss": 3.324979215860367, "logit_z_loss": 26.603500366210938, "router_aux_loss": 1.4580925554037094, "router_z_loss": 1.1020005494356155, "routing_entropy_normalized": 0.6484226508928266, "learning_rate": 0.003, "grad_norm": 0.8880209922790527, "tokens_per_second": 4084698.467231717, "eta_seconds": 2751.912977218628, "data_wait_seconds": 2.1755695343017578e-05} {"step": 23260, "total_steps": 28610, "tokens": 48779755520, "source": "finemath", "phase": 4, "loss": 2.203902989625931, "ce_loss": 2.188228875398636, "logit_z_loss": 38.59621238708496, "router_aux_loss": 1.496280387043953, "router_z_loss": 0.7113285809755325, "routing_entropy_normalized": 0.6537048473379721, "learning_rate": 0.003, "grad_norm": 0.8253844380378723, "tokens_per_second": 4042894.408863019, "eta_seconds": 2775.1808643341064, "data_wait_seconds": 2.0265579223632812e-05} {"step": 23270, "total_steps": 28610, "tokens": 48800727040, "source": "finemath", "phase": 4, "loss": 2.3125930726528168, "ce_loss": 2.297818750143051, "logit_z_loss": 36.88218069076538, "router_aux_loss": 1.3908265829086304, "router_z_loss": 0.8661051392555237, "routing_entropy_normalized": 0.6491268190902705, "learning_rate": 0.003, "grad_norm": 0.7943727374076843, "tokens_per_second": 3982757.60494813, "eta_seconds": 2811.8185415267944, "data_wait_seconds": 2.238154411315918e-05} {"step": 23280, "total_steps": 28610, "tokens": 48821698560, "source": "dclm", "phase": 4, "loss": 3.554482638835907, "ce_loss": 3.540144622325897, "logit_z_loss": 27.949711561203003, "router_aux_loss": 1.324448511004448, "router_z_loss": 1.0935828685760498, "routing_entropy_normalized": 0.6759485280648688, "learning_rate": 0.003, "grad_norm": 0.8188055157661438, "tokens_per_second": 3795800.2832096214, "eta_seconds": 2944.7861652374268, "data_wait_seconds": 1.9609928131103516e-05} {"step": 23290, "total_steps": 28610, "tokens": 48842670080, "source": "fineweb_edu", "phase": 4, "loss": 3.32551446557045, "ce_loss": 3.3108188211917877, "logit_z_loss": 26.19667911529541, "router_aux_loss": 1.3519224375486374, "router_z_loss": 1.1764011979103088, "routing_entropy_normalized": 0.656372201865286, "learning_rate": 0.003, "grad_norm": 1.2104027271270752, "tokens_per_second": 4259854.0759742465, "eta_seconds": 2619.0682687759395, "data_wait_seconds": 2.2590160369873047e-05} {"step": 23300, "total_steps": 28610, "tokens": 48863641600, "source": "fineweb_edu", "phase": 4, "loss": 3.3351853787899017, "ce_loss": 3.3211576342582703, "logit_z_loss": 26.774529933929443, "router_aux_loss": 1.290321633219719, "router_z_loss": 1.1245033890008926, "routing_entropy_normalized": 0.6839548156883289, "learning_rate": 0.003, "grad_norm": 1.3214972019195557, "tokens_per_second": 4136698.3872304726, "eta_seconds": 2691.97221493721, "data_wait_seconds": 1.9282102584838867e-05} {"step": 23310, "total_steps": 28610, "tokens": 48884613120, "source": "finemath", "phase": 4, "loss": 2.2000187039375305, "ce_loss": 2.186162978410721, "logit_z_loss": 38.28430223464966, "router_aux_loss": 1.3199699074029922, "router_z_loss": 0.6559804081916809, "routing_entropy_normalized": 0.6697521946073375, "learning_rate": 0.003, "grad_norm": 1.0493285655975342, "tokens_per_second": 3965938.656292333, "eta_seconds": 2802.5914072990417, "data_wait_seconds": 2.2202730178833008e-05} {"step": 23320, "total_steps": 28610, "tokens": 48905584640, "source": "fineweb_edu", "phase": 4, "loss": 3.3188721537590027, "ce_loss": 3.3047715425491333, "logit_z_loss": 27.213891983032227, "router_aux_loss": 1.3042403161525726, "router_z_loss": 1.0582436546683311, "routing_entropy_normalized": 0.6753880832461551, "learning_rate": 0.003, "grad_norm": 0.8559697866439819, "tokens_per_second": 3855307.16348819, "eta_seconds": 2877.574628829956, "data_wait_seconds": 2.0712614059448242e-05} {"step": 23330, "total_steps": 28610, "tokens": 48926556160, "source": "fineweb_hq", "phase": 4, "loss": 3.336011528968811, "ce_loss": 3.3214731216430664, "logit_z_loss": 25.794158220291138, "router_aux_loss": 1.3520809262990952, "router_z_loss": 1.017561361193657, "routing_entropy_normalized": 0.6511373629745733, "learning_rate": 0.003, "grad_norm": 1.9617642164230347, "tokens_per_second": 4064807.8711001873, "eta_seconds": 2724.104782104492, "data_wait_seconds": 2.1249055862426758e-05} {"step": 23340, "total_steps": 28610, "tokens": 48947527680, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7578553557395935, "ce_loss": 2.741889923810959, "logit_z_loss": 28.62255573272705, "router_aux_loss": 1.4712962210178375, "router_z_loss": 1.2524713575839996, "routing_entropy_normalized": 0.6356212091079682, "learning_rate": 0.003, "grad_norm": 2.0096435546875, "tokens_per_second": 4289864.075371666, "eta_seconds": 2576.3033154010773, "data_wait_seconds": 2.1755695343017578e-05} {"step": 23350, "total_steps": 28610, "tokens": 48968499200, "source": "fineweb_edu", "phase": 4, "loss": 3.3231099247932434, "ce_loss": 3.30799463391304, "logit_z_loss": 25.91492795944214, "router_aux_loss": 1.3925954401493073, "router_z_loss": 1.18935427069664, "routing_entropy_normalized": 0.6723924648537843, "learning_rate": 0.003, "grad_norm": 1.1054065227508545, "tokens_per_second": 3941713.898861423, "eta_seconds": 2798.5337858200073, "data_wait_seconds": 2.092123031616211e-05} {"step": 23360, "total_steps": 28610, "tokens": 48989470720, "source": "finemath", "phase": 4, "loss": 2.292135626077652, "ce_loss": 2.277839630842209, "logit_z_loss": 38.67147874832153, "router_aux_loss": 1.3514199256896973, "router_z_loss": 0.78177585080266, "routing_entropy_normalized": 0.6752732291331305, "learning_rate": 0.003, "grad_norm": 0.9614349603652954, "tokens_per_second": 4121246.1072454406, "eta_seconds": 2671.5337336063385, "data_wait_seconds": 1.996755599975586e-05} {"step": 23370, "total_steps": 28610, "tokens": 49010442240, "source": "finemath", "phase": 4, "loss": 2.257585197687149, "ce_loss": 2.243427813053131, "logit_z_loss": 37.102933406829834, "router_aux_loss": 1.3383924216032028, "router_z_loss": 0.7734770625829697, "routing_entropy_normalized": 0.670079833894348, "learning_rate": 0.003, "grad_norm": 1.0150481462478638, "tokens_per_second": 3999110.6311194533, "eta_seconds": 2747.8800897598267, "data_wait_seconds": 2.059340476989746e-05} {"step": 23380, "total_steps": 28610, "tokens": 49031413760, "source": "dclm", "phase": 4, "loss": 3.5337005853652954, "ce_loss": 3.5188007950782776, "logit_z_loss": 27.186997413635254, "router_aux_loss": 1.378154456615448, "router_z_loss": 1.1182352304458618, "routing_entropy_normalized": 0.6401769367907658, "learning_rate": 0.003, "grad_norm": 1.0074766874313354, "tokens_per_second": 4018065.4910377213, "eta_seconds": 2729.697906732559, "data_wait_seconds": 2.2470951080322266e-05} {"step": 23390, "total_steps": 28610, "tokens": 49052385280, "source": "fineweb_edu", "phase": 4, "loss": 3.349713087081909, "ce_loss": 3.3349598944187164, "logit_z_loss": 26.720913648605347, "router_aux_loss": 1.3467795252799988, "router_z_loss": 1.2854182720184326, "routing_entropy_normalized": 0.6746590554127152, "learning_rate": 0.003, "grad_norm": 0.8629197478294373, "tokens_per_second": 4485097.773795599, "eta_seconds": 2440.779218673706, "data_wait_seconds": 1.8477439880371094e-05} {"step": 23400, "total_steps": 28610, "tokens": 49073356800, "source": "fineweb_edu", "phase": 4, "loss": 3.3104182183742523, "ce_loss": 3.295525223016739, "logit_z_loss": 26.460896253585815, "router_aux_loss": 1.3711878061294556, "router_z_loss": 1.1811403930187225, "routing_entropy_normalized": 0.6460696370064023, "learning_rate": 0.003, "grad_norm": 1.5928843021392822, "tokens_per_second": 3814452.1202890975, "eta_seconds": 2864.4118671417236, "data_wait_seconds": 2.1576881408691406e-05} {"step": 23410, "total_steps": 28610, "tokens": 49094328320, "source": "finemath", "phase": 4, "loss": 2.3284390568733215, "ce_loss": 2.3141292333602905, "logit_z_loss": 33.70966386795044, "router_aux_loss": 1.353203296661377, "router_z_loss": 0.7777679711580276, "routing_entropy_normalized": 0.6630702045965146, "learning_rate": 0.003, "grad_norm": 1.4270051717758179, "tokens_per_second": 3858246.2264101617, "eta_seconds": 2826.463050842285, "data_wait_seconds": 2.2411346435546875e-05} {"step": 23420, "total_steps": 28610, "tokens": 49115299840, "source": "fineweb_edu", "phase": 4, "loss": 3.310102939605713, "ce_loss": 3.294758439064026, "logit_z_loss": 27.029954195022583, "router_aux_loss": 1.4103484898805618, "router_z_loss": 1.2409863024950027, "routing_entropy_normalized": 0.6731961988929589, "learning_rate": 0.003, "grad_norm": 1.0093859434127808, "tokens_per_second": 4458289.369118685, "eta_seconds": 2441.344197034836, "data_wait_seconds": 1.8298625946044922e-05} {"step": 23430, "total_steps": 28610, "tokens": 49136271360, "source": "fineweb_hq", "phase": 4, "loss": 3.3419471979141235, "ce_loss": 3.3275190591812134, "logit_z_loss": 28.103213787078857, "router_aux_loss": 1.3173782378435135, "router_z_loss": 1.2543312460184097, "routing_entropy_normalized": 0.6718057911151538, "learning_rate": 0.003, "grad_norm": 1.1137183904647827, "tokens_per_second": 3557116.4393366007, "eta_seconds": 3053.9476413726807, "data_wait_seconds": 0.05379199981689453} {"step": 23440, "total_steps": 28610, "tokens": 49157242880, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7485502660274506, "ce_loss": 2.733841896057129, "logit_z_loss": 28.912557363510132, "router_aux_loss": 1.318120002746582, "router_z_loss": 1.5271404087543488, "routing_entropy_normalized": 0.6598291843657342, "learning_rate": 0.003, "grad_norm": 1.8872971534729004, "tokens_per_second": 3588655.424755879, "eta_seconds": 3021.2641105651855, "data_wait_seconds": 1.963973045349121e-05} {"step": 23450, "total_steps": 28610, "tokens": 49178214400, "source": "fineweb_edu", "phase": 4, "loss": 3.3476319909095764, "ce_loss": 3.333595931529999, "logit_z_loss": 25.7546865940094, "router_aux_loss": 1.286209911108017, "router_z_loss": 1.1739605069160461, "routing_entropy_normalized": 0.7043165266292992, "learning_rate": 0.003, "grad_norm": 1.6131547689437866, "tokens_per_second": 4313893.873451722, "eta_seconds": 2508.4771757125854, "data_wait_seconds": 1.9937753677368164e-05} {"step": 23460, "total_steps": 28610, "tokens": 49199185920, "source": "finemath", "phase": 4, "loss": 2.232290357351303, "ce_loss": 2.2178244292736053, "logit_z_loss": 36.81144380569458, "router_aux_loss": 1.363970473408699, "router_z_loss": 0.8262319415807724, "routing_entropy_normalized": 0.6435753804230585, "learning_rate": 0.003, "grad_norm": 0.8319454193115234, "tokens_per_second": 4131575.5858066943, "eta_seconds": 2614.0954160690308, "data_wait_seconds": 2.0503997802734375e-05} {"step": 23470, "total_steps": 28610, "tokens": 49220157440, "source": "finemath", "phase": 4, "loss": 2.303291380405426, "ce_loss": 2.2881127297878265, "logit_z_loss": 35.0581488609314, "router_aux_loss": 1.4341526329517365, "router_z_loss": 0.8371137380599976, "routing_entropy_normalized": 0.6663801137203249, "learning_rate": 0.003, "grad_norm": 0.7265781164169312, "tokens_per_second": 3405441.657885887, "eta_seconds": 3165.33429813385, "data_wait_seconds": 0.1281554102897644} {"step": 23480, "total_steps": 28610, "tokens": 49241128960, "source": "dclm", "phase": 4, "loss": 3.5227532982826233, "ce_loss": 3.5072905123233795, "logit_z_loss": 27.05002236366272, "router_aux_loss": 1.4384288638830185, "router_z_loss": 1.0785218924283981, "routing_entropy_normalized": 0.6653113802574475, "learning_rate": 0.003, "grad_norm": 1.2247390747070312, "tokens_per_second": 4337101.7738331845, "eta_seconds": 2480.548145055771, "data_wait_seconds": 2.232193946838379e-05} {"step": 23490, "total_steps": 28610, "tokens": 49262100480, "source": "fineweb_edu", "phase": 4, "loss": 3.3200508654117584, "ce_loss": 3.3053918182849884, "logit_z_loss": 27.461419105529785, "router_aux_loss": 1.3436817526817322, "router_z_loss": 1.2222451120615005, "routing_entropy_normalized": 0.6728258473027207, "learning_rate": 0.003, "grad_norm": 0.9933698773384094, "tokens_per_second": 3926702.779602968, "eta_seconds": 2734.461669921875, "data_wait_seconds": 1.9878149032592773e-05} {"step": 23500, "total_steps": 28610, "tokens": 49283072000, "source": "fineweb_edu", "phase": 4, "loss": 3.3375053703784943, "ce_loss": 3.3231011033058167, "logit_z_loss": 27.00784158706665, "router_aux_loss": 1.311618521809578, "router_z_loss": 1.2880341112613678, "routing_entropy_normalized": 0.653574167201605, "learning_rate": 0.003, "grad_norm": 1.1934696435928345, "tokens_per_second": 2968077.917972723, "eta_seconds": 3610.5678544044495, "data_wait_seconds": 1.952052116394043e-05} {"step": 23510, "total_steps": 28610, "tokens": 49304043520, "source": "finemath", "phase": 4, "loss": 2.2198745906352997, "ce_loss": 2.205725371837616, "logit_z_loss": 39.131309032440186, "router_aux_loss": 1.3365226536989212, "router_z_loss": 0.7840201705694199, "routing_entropy_normalized": 0.6694022432194112, "learning_rate": 0.003, "grad_norm": 0.7704017758369446, "tokens_per_second": 3728106.831852208, "eta_seconds": 2868.8757276535034, "data_wait_seconds": 0.1100030243396759} {"step": 23520, "total_steps": 28610, "tokens": 49325015040, "source": "fineweb_edu", "phase": 4, "loss": 3.372006416320801, "ce_loss": 3.3574418127536774, "logit_z_loss": 26.971747398376465, "router_aux_loss": 1.3391506373882294, "router_z_loss": 1.173060491681099, "routing_entropy_normalized": 0.6636940651598515, "learning_rate": 0.003, "grad_norm": 1.592071294784546, "tokens_per_second": 4173241.7011596756, "eta_seconds": 2557.844583272934, "data_wait_seconds": 2.154707908630371e-05} {"step": 23530, "total_steps": 28610, "tokens": 49345986560, "source": "fineweb_hq", "phase": 4, "loss": 3.322441130876541, "ce_loss": 3.307725250720978, "logit_z_loss": 26.63216781616211, "router_aux_loss": 1.3523808717727661, "router_z_loss": 1.1921413391828537, "routing_entropy_normalized": 0.6513749688172208, "learning_rate": 0.003, "grad_norm": 2.1780030727386475, "tokens_per_second": 4006631.8357914155, "eta_seconds": 2658.9745693206787, "data_wait_seconds": 2.047419548034668e-05} {"step": 23540, "total_steps": 28610, "tokens": 49366958080, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7596396803855896, "ce_loss": 2.744233578443527, "logit_z_loss": 28.07737159729004, "router_aux_loss": 1.3961281329393387, "router_z_loss": 1.4448248594999313, "routing_entropy_normalized": 0.6468411598988685, "learning_rate": 0.003, "grad_norm": 0.9684833884239197, "tokens_per_second": 4257063.0138884, "eta_seconds": 2497.6282017230988, "data_wait_seconds": 1.9758939743041992e-05} {"step": 23550, "total_steps": 28610, "tokens": 49387929600, "source": "fineweb_edu", "phase": 4, "loss": 3.333319514989853, "ce_loss": 3.3184275031089783, "logit_z_loss": 26.572006464004517, "router_aux_loss": 1.3613960891962051, "router_z_loss": 1.2780284136533737, "routing_entropy_normalized": 0.6676627169755439, "learning_rate": 0.003, "grad_norm": 1.3072103261947632, "tokens_per_second": 3891244.0733179427, "eta_seconds": 2727.042796611786, "data_wait_seconds": 1.958012580871582e-05} {"step": 23560, "total_steps": 28610, "tokens": 49408901120, "source": "finemath", "phase": 4, "loss": 2.2629362046718597, "ce_loss": 2.2482101023197174, "logit_z_loss": 35.196292877197266, "router_aux_loss": 1.3791950792074203, "router_z_loss": 0.9341443404555321, "routing_entropy_normalized": 0.6744950088144108, "learning_rate": 0.003, "grad_norm": 1.1281158924102783, "tokens_per_second": 3927670.113049707, "eta_seconds": 2696.4121973514557, "data_wait_seconds": 1.9818544387817383e-05} {"step": 23570, "total_steps": 28610, "tokens": 49429872640, "source": "finemath", "phase": 4, "loss": 2.289544641971588, "ce_loss": 2.274940609931946, "logit_z_loss": 36.653382301330566, "router_aux_loss": 1.3719219863414764, "router_z_loss": 0.8848111778497696, "routing_entropy_normalized": 0.6635783024042006, "learning_rate": 0.003, "grad_norm": 0.7976958155632019, "tokens_per_second": 3704073.2087453944, "eta_seconds": 2853.5197563171387, "data_wait_seconds": 2.1338462829589844e-05} {"step": 23580, "total_steps": 28610, "tokens": 49450844160, "source": "dclm", "phase": 4, "loss": 3.470106452703476, "ce_loss": 3.4556645452976227, "logit_z_loss": 27.360809326171875, "router_aux_loss": 1.3290595412254333, "router_z_loss": 1.1513276398181915, "routing_entropy_normalized": 0.6813966854401527, "learning_rate": 0.003, "grad_norm": 1.2059946060180664, "tokens_per_second": 4036239.051580121, "eta_seconds": 2613.491030931473, "data_wait_seconds": 2.1159648895263672e-05} {"step": 23590, "total_steps": 28610, "tokens": 49471815680, "source": "fineweb_edu", "phase": 4, "loss": 3.3385074138641357, "ce_loss": 3.323728919029236, "logit_z_loss": 26.65797257423401, "router_aux_loss": 1.3476115465164185, "router_z_loss": 1.3023868054151535, "routing_entropy_normalized": 0.6656065105364882, "learning_rate": 0.003, "grad_norm": 1.295632243156433, "tokens_per_second": 3722779.377632021, "eta_seconds": 2827.9148378372192, "data_wait_seconds": 2.2351741790771484e-05} {"step": 23600, "total_steps": 28610, "tokens": 49492787200, "source": "fineweb_edu", "phase": 4, "loss": 3.393657684326172, "ce_loss": 3.3790021538734436, "logit_z_loss": 26.72309923171997, "router_aux_loss": 1.3337619006633759, "router_z_loss": 1.317937210202217, "routing_entropy_normalized": 0.6465125160346837, "learning_rate": 0.003, "grad_norm": 2.342017650604248, "tokens_per_second": 3865617.4334373516, "eta_seconds": 2717.9956891536713, "data_wait_seconds": 2.0235776901245117e-05} {"step": 23610, "total_steps": 28610, "tokens": 49513758720, "source": "finemath", "phase": 4, "loss": 2.2992868423461914, "ce_loss": 2.284282773733139, "logit_z_loss": 37.529759883880615, "router_aux_loss": 1.4092236459255219, "router_z_loss": 0.9117909818887711, "routing_entropy_normalized": 0.689833208088596, "learning_rate": 0.003, "grad_norm": 0.8957445025444031, "tokens_per_second": 3897667.814718282, "eta_seconds": 2690.26517868042, "data_wait_seconds": 2.1576881408691406e-05} {"step": 23620, "total_steps": 28610, "tokens": 49534730240, "source": "fineweb_edu", "phase": 4, "loss": 3.3514299392700195, "ce_loss": 3.3353114426136017, "logit_z_loss": 27.128193855285645, "router_aux_loss": 1.473538488149643, "router_z_loss": 1.3831526637077332, "routing_entropy_normalized": 0.653934344022269, "learning_rate": 0.003, "grad_norm": 0.8951483368873596, "tokens_per_second": 4361674.939947274, "eta_seconds": 2399.2591433525085, "data_wait_seconds": 2.1457672119140625e-05} {"step": 23630, "total_steps": 28610, "tokens": 49555701760, "source": "fineweb_hq", "phase": 4, "loss": 3.3253264129161835, "ce_loss": 3.3093584775924683, "logit_z_loss": 26.12839698791504, "router_aux_loss": 1.4586236476898193, "router_z_loss": 1.3817427605390549, "routing_entropy_normalized": 0.6404820044267301, "learning_rate": 0.003, "grad_norm": 1.512862205505371, "tokens_per_second": 3696275.98762475, "eta_seconds": 2825.4970664978027, "data_wait_seconds": 1.9371509552001953e-05} {"step": 23640, "total_steps": 28610, "tokens": 49576673280, "source": "cosmopedia_v2", "phase": 4, "loss": 2.73933544754982, "ce_loss": 2.7236081063747406, "logit_z_loss": 28.17540669441223, "router_aux_loss": 1.394456535577774, "router_z_loss": 1.7827244848012924, "routing_entropy_normalized": 0.6669245408581761, "learning_rate": 0.003, "grad_norm": 1.3130558729171753, "tokens_per_second": 4479716.217464597, "eta_seconds": 2326.6753816604614, "data_wait_seconds": 2.2083520889282227e-05} {"step": 23650, "total_steps": 28610, "tokens": 49597644800, "source": "fineweb_edu", "phase": 4, "loss": 3.3562767803668976, "ce_loss": 3.340920865535736, "logit_z_loss": 25.186596870422363, "router_aux_loss": 1.3767721205949783, "router_z_loss": 1.588180959224701, "routing_entropy_normalized": 0.6680777795150324, "learning_rate": 0.003, "grad_norm": 1.0796886682510376, "tokens_per_second": 3952982.2195014423, "eta_seconds": 2631.3991165161133, "data_wait_seconds": 2.282857894897461e-05} {"step": 23660, "total_steps": 28610, "tokens": 49618616320, "source": "finemath", "phase": 4, "loss": 2.374426156282425, "ce_loss": 2.3594325482845306, "logit_z_loss": 37.00468921661377, "router_aux_loss": 1.398654267191887, "router_z_loss": 1.0071058794856071, "routing_entropy_normalized": 0.6707560779615432, "learning_rate": 0.003, "grad_norm": 1.2795830965042114, "tokens_per_second": 4122585.838002556, "eta_seconds": 2518.0560958385468, "data_wait_seconds": 2.2023916244506836e-05} {"step": 23670, "total_steps": 28610, "tokens": 49639587840, "source": "finemath", "phase": 4, "loss": 2.223734587430954, "ce_loss": 2.2091824114322662, "logit_z_loss": 38.16329050064087, "router_aux_loss": 1.3477931767702103, "router_z_loss": 1.0742749646306038, "routing_entropy_normalized": 0.6688989930862785, "learning_rate": 0.003, "grad_norm": 1.333094835281372, "tokens_per_second": 3329687.6070640716, "eta_seconds": 3111.3822383880615, "data_wait_seconds": 2.0742416381835938e-05} {"step": 23680, "total_steps": 28610, "tokens": 49660559360, "source": "dclm", "phase": 4, "loss": 3.4381272196769714, "ce_loss": 3.4237350821495056, "logit_z_loss": 26.7637357711792, "router_aux_loss": 1.312104806303978, "router_z_loss": 1.2711256742477417, "routing_entropy_normalized": 0.6787309510977043, "learning_rate": 0.003, "grad_norm": 1.0272053480148315, "tokens_per_second": 3589420.73181459, "eta_seconds": 2880.3977389335632, "data_wait_seconds": 2.327561378479004e-05} {"step": 23690, "total_steps": 28610, "tokens": 49681530880, "source": "fineweb_edu", "phase": 4, "loss": 3.328807055950165, "ce_loss": 3.313997656106949, "logit_z_loss": 26.070016622543335, "router_aux_loss": 1.3347068130970001, "router_z_loss": 1.462309718132019, "routing_entropy_normalized": 0.651507316506809, "learning_rate": 0.003, "grad_norm": 0.6918493509292603, "tokens_per_second": 4243507.310569155, "eta_seconds": 2431.476390838623, "data_wait_seconds": 2.008676528930664e-05} {"step": 23700, "total_steps": 28610, "tokens": 49702502400, "source": "fineweb_edu", "phase": 4, "loss": 3.350445955991745, "ce_loss": 3.336228847503662, "logit_z_loss": 26.177740812301636, "router_aux_loss": 1.2797811031341553, "router_z_loss": 1.4192533940076828, "routing_entropy_normalized": 0.6982878734405407, "learning_rate": 0.003, "grad_norm": 0.8887004256248474, "tokens_per_second": 3138149.2529557017, "eta_seconds": 3281.238554954529, "data_wait_seconds": 0.10306477546691895} {"step": 23710, "total_steps": 28610, "tokens": 49723473920, "source": "finemath", "phase": 4, "loss": 2.2811412513256073, "ce_loss": 2.2653133869171143, "logit_z_loss": 36.77448558807373, "router_aux_loss": 1.4867616295814514, "router_z_loss": 0.9602186679840088, "routing_entropy_normalized": 0.6505073553145732, "learning_rate": 0.003, "grad_norm": 1.4425811767578125, "tokens_per_second": 2707562.5220573517, "eta_seconds": 3795.3120994567867, "data_wait_seconds": 1.7464160919189453e-05} {"step": 23720, "total_steps": 28610, "tokens": 49744445440, "source": "fineweb_edu", "phase": 4, "loss": 3.3437837958335876, "ce_loss": 3.3281098306179047, "logit_z_loss": 27.036349773406982, "router_aux_loss": 1.4278680831193924, "router_z_loss": 1.3952188789844513, "routing_entropy_normalized": 0.6484086260682016, "learning_rate": 0.003, "grad_norm": 1.1434388160705566, "tokens_per_second": 4355875.969084502, "eta_seconds": 2354.3079171180725, "data_wait_seconds": 1.6987323760986328e-05} {"step": 23730, "total_steps": 28610, "tokens": 49765416960, "source": "fineweb_hq", "phase": 4, "loss": 3.445702612400055, "ce_loss": 3.4301140308380127, "logit_z_loss": 28.1646888256073, "router_aux_loss": 1.4177858382463455, "router_z_loss": 1.4107386767864227, "routing_entropy_normalized": 0.6413402869835211, "learning_rate": 0.003, "grad_norm": 3.8726067543029785, "tokens_per_second": 3173932.3686793055, "eta_seconds": 3224.423387527466, "data_wait_seconds": 2.3126602172851562e-05} {"step": 23740, "total_steps": 28610, "tokens": 49786388480, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7372111678123474, "ce_loss": 2.7211369276046753, "logit_z_loss": 28.247689962387085, "router_aux_loss": 1.4214132130146027, "router_z_loss": 1.8601080924272537, "routing_entropy_normalized": 0.6849884490263751, "learning_rate": 0.003, "grad_norm": 1.4305118322372437, "tokens_per_second": 2942955.749637948, "eta_seconds": 3470.3648674488063, "data_wait_seconds": 1.71661376953125e-05} {"step": 23750, "total_steps": 28610, "tokens": 49807360000, "source": "fineweb_edu", "phase": 4, "loss": 3.3456554114818573, "ce_loss": 3.330296814441681, "logit_z_loss": 25.862786769866943, "router_aux_loss": 1.3770325779914856, "router_z_loss": 1.5882849097251892, "routing_entropy_normalized": 0.6780529204203768, "learning_rate": 0.003, "grad_norm": 0.932258129119873, "tokens_per_second": 4184344.505393447, "eta_seconds": 2435.7838382720947, "data_wait_seconds": 2.244114875793457e-05} {"step": 23760, "total_steps": 28610, "tokens": 49828331520, "source": "finemath", "phase": 4, "loss": 2.2015995383262634, "ce_loss": 2.1872150003910065, "logit_z_loss": 39.142531394958496, "router_aux_loss": 1.3320160657167435, "router_z_loss": 1.0643736273050308, "routing_entropy_normalized": 0.6665686914185429, "learning_rate": 0.003, "grad_norm": 0.8877584338188171, "tokens_per_second": 3867250.6913508023, "eta_seconds": 2630.082198381424, "data_wait_seconds": 2.187490463256836e-05} {"step": 23770, "total_steps": 28610, "tokens": 49849303040, "source": "finemath", "phase": 4, "loss": 2.2071791887283325, "ce_loss": 2.1930323243141174, "logit_z_loss": 36.466862201690674, "router_aux_loss": 1.3133015185594559, "router_z_loss": 1.0137702971696854, "routing_entropy_normalized": 0.6799632220775854, "learning_rate": 0.003, "grad_norm": 0.7702761888504028, "tokens_per_second": 3774584.920078215, "eta_seconds": 2689.0945348739624, "data_wait_seconds": 2.065300941467285e-05} {"step": 23780, "total_steps": 28610, "tokens": 49870274560, "source": "dclm", "phase": 4, "loss": 3.3954112231731415, "ce_loss": 3.380972236394882, "logit_z_loss": 26.496898412704468, "router_aux_loss": 1.3256918489933014, "router_z_loss": 1.1820195466279984, "routing_entropy_normalized": 0.6682401395815848, "learning_rate": 0.003, "grad_norm": 0.7465144991874695, "tokens_per_second": 3729272.368066302, "eta_seconds": 2716.1449098587036, "data_wait_seconds": 2.384185791015625e-05} {"step": 23790, "total_steps": 28610, "tokens": 49891246080, "source": "fineweb_edu", "phase": 4, "loss": 3.355817049741745, "ce_loss": 3.3412547409534454, "logit_z_loss": 25.778980016708374, "router_aux_loss": 1.3275146931409836, "router_z_loss": 1.2871536314487457, "routing_entropy_normalized": 0.6609750385793711, "learning_rate": 0.003, "grad_norm": 0.6598057150840759, "tokens_per_second": 4513816.423348397, "eta_seconds": 2239.407120704651, "data_wait_seconds": 2.0772218704223633e-05} {"step": 23800, "total_steps": 28610, "tokens": 49912217600, "source": "fineweb_edu", "phase": 4, "loss": 3.350065141916275, "ce_loss": 3.335505187511444, "logit_z_loss": 26.283734798431396, "router_aux_loss": 1.322598397731781, "router_z_loss": 1.333984136581421, "routing_entropy_normalized": 0.6497065474694437, "learning_rate": 0.003, "grad_norm": 1.002121090888977, "tokens_per_second": 3804285.4338703593, "eta_seconds": 2651.5626378059387, "data_wait_seconds": 2.0712614059448242e-05} {"step": 23810, "total_steps": 28610, "tokens": 49933189120, "source": "finemath", "phase": 4, "loss": 2.2247402369976044, "ce_loss": 2.2102750539779663, "logit_z_loss": 39.830655097961426, "router_aux_loss": 1.3681946694850922, "router_z_loss": 0.7832128331065178, "routing_entropy_normalized": 0.6730982937256993, "learning_rate": 0.003, "grad_norm": 0.49329689145088196, "tokens_per_second": 3797744.2873212723, "eta_seconds": 2650.6075286865234, "data_wait_seconds": 2.22623348236084e-05} {"step": 23820, "total_steps": 28610, "tokens": 49954160640, "source": "fineweb_edu", "phase": 4, "loss": 3.332374155521393, "ce_loss": 3.3180040419101715, "logit_z_loss": 27.70765733718872, "router_aux_loss": 1.3088500052690506, "router_z_loss": 1.2816250920295715, "routing_entropy_normalized": 0.6686254260659825, "learning_rate": 0.003, "grad_norm": 0.7568337917327881, "tokens_per_second": 3868518.3090373366, "eta_seconds": 2596.693947792053, "data_wait_seconds": 2.1904706954956055e-05} {"step": 23830, "total_steps": 28610, "tokens": 49975132160, "source": "fineweb_hq", "phase": 4, "loss": 3.113378584384918, "ce_loss": 3.0988480150699615, "logit_z_loss": 27.592151880264282, "router_aux_loss": 1.3413085490465164, "router_z_loss": 1.117497257888317, "routing_entropy_normalized": 0.6683182642321576, "learning_rate": 0.003, "grad_norm": 1.9337955713272095, "tokens_per_second": 4145055.7289723936, "eta_seconds": 2418.3960881233215, "data_wait_seconds": 1.996755599975586e-05} {"step": 23840, "total_steps": 28610, "tokens": 49996103680, "source": "cosmopedia_v2", "phase": 4, "loss": 2.711519956588745, "ce_loss": 2.697361469268799, "logit_z_loss": 28.639423370361328, "router_aux_loss": 1.2745015919208527, "router_z_loss": 1.4134841114282608, "routing_entropy_normalized": 0.6856807053285894, "learning_rate": 0.003, "grad_norm": 1.9807584285736084, "tokens_per_second": 4318383.774187247, "eta_seconds": 2316.4719865322113, "data_wait_seconds": 2.0891427993774414e-05} {"step": 23850, "total_steps": 28610, "tokens": 50017075200, "source": "fineweb_edu", "phase": 4, "loss": 3.3709880709648132, "ce_loss": 3.357676148414612, "logit_z_loss": 26.47662091255188, "router_aux_loss": 1.215041607618332, "router_z_loss": 1.161472499370575, "routing_entropy_normalized": 0.7132515531590092, "learning_rate": 0.003, "grad_norm": 0.9144893884658813, "tokens_per_second": 4127690.230827232, "eta_seconds": 2418.409076690674, "data_wait_seconds": 0.029644817113876343} {"step": 23860, "total_steps": 28610, "tokens": 50038046720, "source": "finemath", "phase": 4, "loss": 2.3411063849925995, "ce_loss": 2.3279261887073517, "logit_z_loss": 39.715404987335205, "router_aux_loss": 1.2515917122364044, "router_z_loss": 0.6643284857273102, "routing_entropy_normalized": 0.6585194825127655, "learning_rate": 0.003, "grad_norm": 1.1866015195846558, "tokens_per_second": 3841852.3289904613, "eta_seconds": 2592.8825855255127, "data_wait_seconds": 2.0205974578857422e-05} {"step": 23870, "total_steps": 28610, "tokens": 50059018240, "source": "finemath", "phase": 4, "loss": 2.2886930108070374, "ce_loss": 2.2749293446540833, "logit_z_loss": 37.62888431549072, "router_aux_loss": 1.3185314685106277, "router_z_loss": 0.5783538892865181, "routing_entropy_normalized": 0.6457459363714715, "learning_rate": 0.003, "grad_norm": 0.8718265295028687, "tokens_per_second": 3987491.5861498965, "eta_seconds": 2492.920741081238, "data_wait_seconds": 1.96993350982666e-05} {"step": 23880, "total_steps": 28610, "tokens": 50079989760, "source": "dclm", "phase": 4, "loss": 3.4722918570041656, "ce_loss": 3.4582981169223785, "logit_z_loss": 27.815380334854126, "router_aux_loss": 1.3226007521152496, "router_z_loss": 0.7677656933665276, "routing_entropy_normalized": 0.6652527326392768, "learning_rate": 0.003, "grad_norm": 0.7904881834983826, "tokens_per_second": 3313477.923229231, "eta_seconds": 2993.6909766197205, "data_wait_seconds": 2.1398067474365234e-05} {"step": 23890, "total_steps": 28610, "tokens": 50100961280, "source": "fineweb_edu", "phase": 4, "loss": 3.3928984999656677, "ce_loss": 3.378646284341812, "logit_z_loss": 25.601819038391113, "router_aux_loss": 1.3312174677848816, "router_z_loss": 0.9400523975491524, "routing_entropy_normalized": 0.6833865428364819, "learning_rate": 0.003, "grad_norm": 0.8626319766044617, "tokens_per_second": 4512199.05574564, "eta_seconds": 2193.7324390411377, "data_wait_seconds": 2.0265579223632812e-05} {"step": 23900, "total_steps": 28610, "tokens": 50121932800, "source": "fineweb_edu", "phase": 4, "loss": 3.358798176050186, "ce_loss": 3.3441043198108673, "logit_z_loss": 26.341603994369507, "router_aux_loss": 1.3705078959465027, "router_z_loss": 0.988733783364296, "routing_entropy_normalized": 0.6594625495949527, "learning_rate": 0.003, "grad_norm": 1.4748952388763428, "tokens_per_second": 3874354.9845093507, "eta_seconds": 2549.4788060188293, "data_wait_seconds": 2.0295381546020508e-05} {"step": 23910, "total_steps": 28610, "tokens": 50142904320, "source": "finemath", "phase": 4, "loss": 2.240939438343048, "ce_loss": 2.2257898151874542, "logit_z_loss": 36.752225399017334, "router_aux_loss": 1.4516331553459167, "router_z_loss": 0.6332525312900543, "routing_entropy_normalized": 0.6666839904988664, "learning_rate": 0.003, "grad_norm": 0.7915467619895935, "tokens_per_second": 3969889.0181366857, "eta_seconds": 2482.843816280365, "data_wait_seconds": 2.232193946838379e-05} {"step": 23920, "total_steps": 28610, "tokens": 50163875840, "source": "fineweb_edu", "phase": 4, "loss": 3.351833939552307, "ce_loss": 3.3376618921756744, "logit_z_loss": 25.782209396362305, "router_aux_loss": 1.3026458621025085, "router_z_loss": 1.1456129252910614, "routing_entropy_normalized": 0.7037726198903675, "learning_rate": 0.003, "grad_norm": 0.7929650545120239, "tokens_per_second": 3566949.1440215507, "eta_seconds": 2757.438495159149, "data_wait_seconds": 2.1785497665405273e-05} {"step": 23930, "total_steps": 28610, "tokens": 50184847360, "source": "fineweb_hq", "phase": 4, "loss": 3.1060191690921783, "ce_loss": 3.0905469059944153, "logit_z_loss": 27.582139253616333, "router_aux_loss": 1.4409171640872955, "router_z_loss": 1.0630602315068245, "routing_entropy_normalized": 0.6636623074874035, "learning_rate": 0.003, "grad_norm": 1.263815999031067, "tokens_per_second": 4091534.5214262316, "eta_seconds": 2398.775156021118, "data_wait_seconds": 1.9669532775878906e-05} {"step": 23940, "total_steps": 28610, "tokens": 50205818880, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7519437670707703, "ce_loss": 2.7367850244045258, "logit_z_loss": 28.294857025146484, "router_aux_loss": 1.367299199104309, "router_z_loss": 1.485740378499031, "routing_entropy_normalized": 0.6827512382238516, "learning_rate": 0.003, "grad_norm": 1.694676399230957, "tokens_per_second": 3809861.929572441, "eta_seconds": 2570.6180489063263, "data_wait_seconds": 2.2619962692260742e-05} {"step": 23950, "total_steps": 28610, "tokens": 50226790400, "source": "fineweb_edu", "phase": 4, "loss": 3.3369200527668, "ce_loss": 3.322547495365143, "logit_z_loss": 26.726956367492676, "router_aux_loss": 1.3161476254463196, "router_z_loss": 1.2110674381256104, "routing_entropy_normalized": 0.66253953835586, "learning_rate": 0.003, "grad_norm": 1.0095235109329224, "tokens_per_second": 3937426.830239745, "eta_seconds": 2482.008870601654, "data_wait_seconds": 2.3156404495239258e-05} {"step": 23960, "total_steps": 28610, "tokens": 50247761920, "source": "finemath", "phase": 4, "loss": 2.281723976135254, "ce_loss": 2.2674593329429626, "logit_z_loss": 36.66099977493286, "router_aux_loss": 1.3526840955018997, "router_z_loss": 0.7378121763467789, "routing_entropy_normalized": 0.6825342145284716, "learning_rate": 0.003, "grad_norm": 1.2015200853347778, "tokens_per_second": 4327795.995388457, "eta_seconds": 2253.2847690582275, "data_wait_seconds": 2.467632293701172e-05} {"step": 23970, "total_steps": 28610, "tokens": 50268733440, "source": "finemath", "phase": 4, "loss": 2.1983138620853424, "ce_loss": 2.184480160474777, "logit_z_loss": 37.47657060623169, "router_aux_loss": 1.3152762651443481, "router_z_loss": 0.6809619143605232, "routing_entropy_normalized": 0.7000174519977777, "learning_rate": 0.003, "grad_norm": 0.6206466555595398, "tokens_per_second": 3819909.17788963, "eta_seconds": 2547.3865547180176, "data_wait_seconds": 2.1636486053466797e-05} {"step": 23980, "total_steps": 28610, "tokens": 50289704960, "source": "dclm", "phase": 4, "loss": 3.4833135306835175, "ce_loss": 3.4694012701511383, "logit_z_loss": 26.76358652114868, "router_aux_loss": 1.2948553562164307, "router_z_loss": 0.9637948796153069, "routing_entropy_normalized": 0.7011582300146392, "learning_rate": 0.003, "grad_norm": 1.171303153038025, "tokens_per_second": 3784821.418303368, "eta_seconds": 2565.4615335464478, "data_wait_seconds": 2.0682811737060547e-05} {"step": 23990, "total_steps": 28610, "tokens": 50310676480, "source": "fineweb_edu", "phase": 4, "loss": 3.359926849603653, "ce_loss": 3.345239043235779, "logit_z_loss": 26.647894620895386, "router_aux_loss": 1.371923565864563, "router_z_loss": 0.9685635939240456, "routing_entropy_normalized": 0.6617017142831626, "learning_rate": 0.003, "grad_norm": 0.9201018810272217, "tokens_per_second": 3963326.4709731084, "eta_seconds": 2444.6238055229187, "data_wait_seconds": 2.1070241928100586e-05} {"step": 24000, "total_steps": 28610, "tokens": 50331648000, "source": "fineweb_edu", "phase": 4, "loss": 3.3604625463485718, "ce_loss": 3.345949798822403, "logit_z_loss": 25.89029884338379, "router_aux_loss": 1.3446735888719559, "router_z_loss": 1.0660356134176254, "routing_entropy_normalized": 0.6635059030085974, "learning_rate": 0.003, "grad_norm": 0.8446911573410034, "tokens_per_second": 4058063.006901541, "eta_seconds": 2382.38556265831, "data_wait_seconds": 2.256035804748535e-05} {"step": 24010, "total_steps": 28610, "tokens": 50352619520, "source": "finemath", "phase": 4, "loss": 2.1989027559757233, "ce_loss": 2.184409946203232, "logit_z_loss": 36.950764656066895, "router_aux_loss": 1.3780548870563507, "router_z_loss": 0.7122810930013657, "routing_entropy_normalized": 0.6456261972418191, "learning_rate": 0.003, "grad_norm": 0.9706584215164185, "tokens_per_second": 3831736.5416286155, "eta_seconds": 2517.631130218506, "data_wait_seconds": 2.0951032638549805e-05} {"step": 24020, "total_steps": 28610, "tokens": 50373591040, "source": "fineweb_edu", "phase": 4, "loss": 3.3489673137664795, "ce_loss": 3.335086762905121, "logit_z_loss": 26.972429513931274, "router_aux_loss": 1.273654043674469, "router_z_loss": 1.1440075188875198, "routing_entropy_normalized": 0.6846001226475372, "learning_rate": 0.003, "grad_norm": 0.9127093553543091, "tokens_per_second": 4350073.250783374, "eta_seconds": 2212.819675683975, "data_wait_seconds": 2.282857894897461e-05} {"step": 24030, "total_steps": 28610, "tokens": 50394562560, "source": "fineweb_hq", "phase": 4, "loss": 3.0675933957099915, "ce_loss": 3.0531199872493744, "logit_z_loss": 26.681312084197998, "router_aux_loss": 1.3448146134614944, "router_z_loss": 1.0252813026309013, "routing_entropy_normalized": 0.6588659990285537, "learning_rate": 0.003, "grad_norm": 2.0242929458618164, "tokens_per_second": 4011070.099962872, "eta_seconds": 2394.611891746521, "data_wait_seconds": 2.1576881408691406e-05} {"step": 24040, "total_steps": 28610, "tokens": 50415534080, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7506174743175507, "ce_loss": 2.73552206158638, "logit_z_loss": 28.69836711883545, "router_aux_loss": 1.3736347556114197, "router_z_loss": 1.3590606898069382, "routing_entropy_normalized": 0.6621030071377481, "learning_rate": 0.003, "grad_norm": 1.2197023630142212, "tokens_per_second": 3960053.1487032855, "eta_seconds": 2420.165659427643, "data_wait_seconds": 2.16066837310791e-05} {"step": 24050, "total_steps": 28610, "tokens": 50436505600, "source": "fineweb_edu", "phase": 4, "loss": 3.359160304069519, "ce_loss": 3.344631850719452, "logit_z_loss": 26.39349937438965, "router_aux_loss": 1.3434318453073502, "router_z_loss": 1.094176322221756, "routing_entropy_normalized": 0.6687731366771202, "learning_rate": 0.003, "grad_norm": 0.5564916729927063, "tokens_per_second": 4026902.603096727, "eta_seconds": 2374.7813301086426, "data_wait_seconds": 2.1785497665405273e-05} {"step": 24060, "total_steps": 28610, "tokens": 50457477120, "source": "finemath", "phase": 4, "loss": 2.2268863022327423, "ce_loss": 2.213073492050171, "logit_z_loss": 37.39243745803833, "router_aux_loss": 1.3035637438297272, "router_z_loss": 0.7771740108728409, "routing_entropy_normalized": 0.6965888893473337, "learning_rate": 0.003, "grad_norm": 0.7171623110771179, "tokens_per_second": 3618435.868328009, "eta_seconds": 2637.0625174045563, "data_wait_seconds": 2.2470951080322266e-05} {"step": 24070, "total_steps": 28610, "tokens": 50478448640, "source": "finemath", "phase": 4, "loss": 2.211684912443161, "ce_loss": 2.1982464492321014, "logit_z_loss": 39.234477519989014, "router_aux_loss": 1.2796570360660553, "router_z_loss": 0.6419545896351337, "routing_entropy_normalized": 0.6817988396554909, "learning_rate": 0.003, "grad_norm": 0.6494542360305786, "tokens_per_second": 3838923.938744934, "eta_seconds": 2480.1403288841248, "data_wait_seconds": 2.09808349609375e-05} {"step": 24080, "total_steps": 28610, "tokens": 50499420160, "source": "dclm", "phase": 4, "loss": 3.4631308913230896, "ce_loss": 3.4492971897125244, "logit_z_loss": 27.780398845672607, "router_aux_loss": 1.2868825495243073, "router_z_loss": 0.9648478403687477, "routing_entropy_normalized": 0.6920316002603188, "learning_rate": 0.003, "grad_norm": 1.0946855545043945, "tokens_per_second": 3924769.0160914273, "eta_seconds": 2420.549724340439, "data_wait_seconds": 1.8268823623657227e-05} {"step": 24090, "total_steps": 28610, "tokens": 50520391680, "source": "fineweb_edu", "phase": 4, "loss": 3.3664062917232513, "ce_loss": 3.3521942794322968, "logit_z_loss": 26.36588215827942, "router_aux_loss": 1.3188555389642715, "router_z_loss": 1.0235438793897629, "routing_entropy_normalized": 0.6799460438826426, "learning_rate": 0.003, "grad_norm": 1.397956371307373, "tokens_per_second": 3973469.9252064084, "eta_seconds": 2385.6043252944946, "data_wait_seconds": 2.1129846572875977e-05} {"step": 24100, "total_steps": 28610, "tokens": 50541363200, "source": "fineweb_edu", "phase": 4, "loss": 3.3748714327812195, "ce_loss": 3.359769970178604, "logit_z_loss": 25.37770366668701, "router_aux_loss": 1.4084255695343018, "router_z_loss": 1.0171884521842003, "routing_entropy_normalized": 0.6557388485948015, "learning_rate": 0.003, "grad_norm": 0.6009724736213684, "tokens_per_second": 3767706.2425589017, "eta_seconds": 2510.321907043457, "data_wait_seconds": 0.06764674186706543} {"step": 24110, "total_steps": 28610, "tokens": 50562334720, "source": "finemath", "phase": 4, "loss": 2.301802694797516, "ce_loss": 2.286600559949875, "logit_z_loss": 36.062589168548584, "router_aux_loss": 1.4441889971494675, "router_z_loss": 0.760260246694088, "routing_entropy_normalized": 0.652968636385498, "learning_rate": 0.003, "grad_norm": 0.9346046447753906, "tokens_per_second": 3560967.553640757, "eta_seconds": 2650.174105167389, "data_wait_seconds": 1.9937753677368164e-05} {"step": 24120, "total_steps": 28610, "tokens": 50583306240, "source": "fineweb_edu", "phase": 4, "loss": 3.3727630972862244, "ce_loss": 3.3570926189422607, "logit_z_loss": 26.482563972473145, "router_aux_loss": 1.4456365257501602, "router_z_loss": 1.2141220271587372, "routing_entropy_normalized": 0.6533542270618307, "learning_rate": 0.003, "grad_norm": 1.2211822271347046, "tokens_per_second": 3965200.827097164, "eta_seconds": 2374.7126288414, "data_wait_seconds": 2.1636486053466797e-05} {"step": 24130, "total_steps": 28610, "tokens": 50604277760, "source": "fineweb_hq", "phase": 4, "loss": 3.087701767683029, "ce_loss": 3.073186904191971, "logit_z_loss": 28.607839345932007, "router_aux_loss": 1.3261513859033585, "router_z_loss": 1.2533110529184341, "routing_entropy_normalized": 0.6661086902709143, "learning_rate": 0.003, "grad_norm": 1.382266879081726, "tokens_per_second": 4273101.244373579, "eta_seconds": 2198.6937408447266, "data_wait_seconds": 2.1725893020629883e-05} {"step": 24140, "total_steps": 28610, "tokens": 50625249280, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7295839190483093, "ce_loss": 2.714962899684906, "logit_z_loss": 28.627902269363403, "router_aux_loss": 1.3135216534137726, "router_z_loss": 1.485806718468666, "routing_entropy_normalized": 0.6825919622789265, "learning_rate": 0.003, "grad_norm": 0.8616805672645569, "tokens_per_second": 3260074.722304687, "eta_seconds": 2875.476864337921, "data_wait_seconds": 1.9252300262451172e-05} {"step": 24150, "total_steps": 28610, "tokens": 50646220800, "source": "fineweb_edu", "phase": 4, "loss": 3.3977365493774414, "ce_loss": 3.3827297389507294, "logit_z_loss": 25.76901364326477, "router_aux_loss": 1.3822955340147018, "router_z_loss": 1.1838675141334534, "routing_entropy_normalized": 0.6525435009810907, "learning_rate": 0.003, "grad_norm": 1.198756217956543, "tokens_per_second": 4143225.1153508117, "eta_seconds": 2257.4920887947083, "data_wait_seconds": 2.0951032638549805e-05} {"step": 24160, "total_steps": 28610, "tokens": 50667192320, "source": "finemath", "phase": 4, "loss": 2.2625846564769745, "ce_loss": 2.24747434258461, "logit_z_loss": 36.636247634887695, "router_aux_loss": 1.4374703615903854, "router_z_loss": 0.7355712205171585, "routing_entropy_normalized": 0.6544136138546968, "learning_rate": 0.003, "grad_norm": 0.8495532274246216, "tokens_per_second": 4364224.642965943, "eta_seconds": 2138.3698511123657, "data_wait_seconds": 2.2143125534057617e-05} {"step": 24170, "total_steps": 28610, "tokens": 50688163840, "source": "finemath", "phase": 4, "loss": 2.278942257165909, "ce_loss": 2.264692395925522, "logit_z_loss": 34.4075403213501, "router_aux_loss": 1.3376235216856003, "router_z_loss": 0.8735621869564056, "routing_entropy_normalized": 0.6536045347700294, "learning_rate": 0.003, "grad_norm": 1.4475053548812866, "tokens_per_second": 3515081.8523502205, "eta_seconds": 2648.9724197387695, "data_wait_seconds": 2.0712614059448242e-05} {"step": 24180, "total_steps": 28610, "tokens": 50709135360, "source": "dclm", "phase": 4, "loss": 3.4561499655246735, "ce_loss": 3.4410100281238556, "logit_z_loss": 28.170114278793335, "router_aux_loss": 1.408264383673668, "router_z_loss": 1.057287059724331, "routing_entropy_normalized": 0.633090608292409, "learning_rate": 0.003, "grad_norm": 1.256686806678772, "tokens_per_second": 4058999.313220518, "eta_seconds": 2288.835903406143, "data_wait_seconds": 2.2202730178833008e-05} {"step": 24190, "total_steps": 28610, "tokens": 50730106880, "source": "fineweb_edu", "phase": 4, "loss": 3.379233032464981, "ce_loss": 3.3646769523620605, "logit_z_loss": 27.273935079574585, "router_aux_loss": 1.3305370062589645, "router_z_loss": 1.2506864666938782, "routing_entropy_normalized": 0.678480881729641, "learning_rate": 0.003, "grad_norm": 0.7681809663772583, "tokens_per_second": 4231671.368974989, "eta_seconds": 2190.484806537628, "data_wait_seconds": 2.2530555725097656e-05} {"step": 24200, "total_steps": 28610, "tokens": 50751078400, "source": "fineweb_edu", "phase": 4, "loss": 3.369892120361328, "ce_loss": 3.35591584444046, "logit_z_loss": 27.27662205696106, "router_aux_loss": 1.274714007973671, "router_z_loss": 1.2291863858699799, "routing_entropy_normalized": 0.6912812619527592, "learning_rate": 0.003, "grad_norm": 1.3174415826797485, "tokens_per_second": 4265834.851670935, "eta_seconds": 2168.025871038437, "data_wait_seconds": 2.2351741790771484e-05} {"step": 24210, "total_steps": 28610, "tokens": 50772049920, "source": "finemath", "phase": 4, "loss": 2.2453589141368866, "ce_loss": 2.2310506105422974, "logit_z_loss": 36.80042314529419, "router_aux_loss": 1.352341279387474, "router_z_loss": 0.7848491817712784, "routing_entropy_normalized": 0.6667280972008218, "learning_rate": 0.003, "grad_norm": 0.6320315599441528, "tokens_per_second": 2173701.0512110745, "eta_seconds": 4245.049610137939, "data_wait_seconds": 2.205371856689453e-05} {"step": 24220, "total_steps": 28610, "tokens": 50793021440, "source": "fineweb_edu", "phase": 4, "loss": 3.342570185661316, "ce_loss": 3.3280737102031708, "logit_z_loss": 26.309176206588745, "router_aux_loss": 1.3396453857421875, "router_z_loss": 1.100033700466156, "routing_entropy_normalized": 0.6669772910543197, "learning_rate": 0.003, "grad_norm": 1.3728560209274292, "tokens_per_second": 2527251.5694905235, "eta_seconds": 3642.889133453369, "data_wait_seconds": 2.047419548034668e-05} {"step": 24230, "total_steps": 28610, "tokens": 50813992960, "source": "fineweb_hq", "phase": 4, "loss": 3.1280462741851807, "ce_loss": 3.1133175790309906, "logit_z_loss": 26.37799572944641, "router_aux_loss": 1.359921708703041, "router_z_loss": 1.1294372230768204, "routing_entropy_normalized": 0.6683189590801414, "learning_rate": 0.003, "grad_norm": 1.508437156677246, "tokens_per_second": 1923522.3017014572, "eta_seconds": 4775.367434978485, "data_wait_seconds": 0.7195310890674591} {"step": 24240, "total_steps": 28610, "tokens": 50834964480, "source": "cosmopedia_v2", "phase": 4, "loss": 2.7263738214969635, "ce_loss": 2.711556077003479, "logit_z_loss": 28.415294647216797, "router_aux_loss": 1.3485033512115479, "router_z_loss": 1.332697719335556, "routing_entropy_normalized": 0.6520522064417684, "learning_rate": 0.003, "grad_norm": 1.4158016443252563, "tokens_per_second": 3069877.72697712, "eta_seconds": 2985.3157210350037, "data_wait_seconds": 1.9311904907226562e-05} {"step": 24250, "total_steps": 28610, "tokens": 50855936000, "source": "fineweb_edu", "phase": 4, "loss": 3.354590564966202, "ce_loss": 3.3401589691638947, "logit_z_loss": 26.56472659111023, "router_aux_loss": 1.3316797763109207, "router_z_loss": 1.114752121269703, "routing_entropy_normalized": 0.6509236904277887, "learning_rate": 0.003, "grad_norm": 1.4022098779678345, "tokens_per_second": 3565985.1996995457, "eta_seconds": 2564.1112365722656, "data_wait_seconds": 1.9669532775878906e-05} {"step": 24260, "total_steps": 28610, "tokens": 50876907520, "source": "finemath", "phase": 4, "loss": 2.1913041174411774, "ce_loss": 2.1775179654359818, "logit_z_loss": 36.56788110733032, "router_aux_loss": 1.3080720752477646, "router_z_loss": 0.7054132297635078, "routing_entropy_normalized": 0.697426613090118, "learning_rate": 0.003, "grad_norm": 0.7187559008598328, "tokens_per_second": 4218705.339357827, "eta_seconds": 2162.419620752334, "data_wait_seconds": 2.256035804748535e-05} {"step": 24270, "total_steps": 28610, "tokens": 50897879040, "source": "finemath", "phase": 4, "loss": 2.2536416053771973, "ce_loss": 2.240005284547806, "logit_z_loss": 36.37790012359619, "router_aux_loss": 1.2972136735916138, "router_z_loss": 0.6642125770449638, "routing_entropy_normalized": 0.6887776157594374, "learning_rate": 0.003, "grad_norm": 0.721035361289978, "tokens_per_second": 3266841.295812545, "eta_seconds": 2786.0672912597656, "data_wait_seconds": 2.1278858184814453e-05} {"step": 24280, "total_steps": 28610, "tokens": 50918850560, "source": "dclm", "phase": 4, "loss": 3.517983227968216, "ce_loss": 3.50399312376976, "logit_z_loss": 26.799833297729492, "router_aux_loss": 1.3142754137516022, "router_z_loss": 0.8473827242851257, "routing_entropy_normalized": 0.6629497407889113, "learning_rate": 0.003, "grad_norm": 1.7907487154006958, "tokens_per_second": 4502637.835112034, "eta_seconds": 2016.7440714836118, "data_wait_seconds": 2.256035804748535e-05} {"step": 24290, "total_steps": 28610, "tokens": 50939822080, "source": "fineweb_edu", "phase": 4, "loss": 3.3891804218292236, "ce_loss": 3.3750800788402557, "logit_z_loss": 26.03255581855774, "router_aux_loss": 1.3126715272665024, "router_z_loss": 0.9736530855298042, "routing_entropy_normalized": 0.6790296996883785, "learning_rate": 0.003, "grad_norm": 1.470539927482605, "tokens_per_second": 3546889.791286534, "eta_seconds": 2554.26505279541, "data_wait_seconds": 2.110004425048828e-05} {"step": 24300, "total_steps": 28610, "tokens": 50960793600, "source": "fineweb_edu", "phase": 4, "loss": 3.3660579919815063, "ce_loss": 3.3520293533802032, "logit_z_loss": 27.16259741783142, "router_aux_loss": 1.3044087886810303, "router_z_loss": 0.984544038772583, "routing_entropy_normalized": 0.6392807897485858, "learning_rate": 0.003, "grad_norm": 1.4832220077514648, "tokens_per_second": 4118196.5253238785, "eta_seconds": 2194.826076030731, "data_wait_seconds": 2.065300941467285e-05} {"step": 24310, "total_steps": 28610, "tokens": 50981765120, "source": "finemath", "phase": 4, "loss": 2.2948712408542633, "ce_loss": 2.2806537449359894, "logit_z_loss": 37.124051094055176, "router_aux_loss": 1.3559345752000809, "router_z_loss": 0.6581652536988258, "routing_entropy_normalized": 0.6787833577500112, "learning_rate": 0.003, "grad_norm": 0.6967179775238037, "tokens_per_second": 3312468.949587889, "eta_seconds": 2722.3662281036377, "data_wait_seconds": 1.990795135498047e-05} {"step": 24320, "total_steps": 28610, "tokens": 51002736640, "source": "fineweb_edu", "phase": 5, "loss": 3.3450827896595, "ce_loss": 3.3304241597652435, "logit_z_loss": 25.98381781578064, "router_aux_loss": 1.3662722706794739, "router_z_loss": 0.9959253519773483, "routing_entropy_normalized": 0.6719881522472351, "learning_rate": 0.003, "grad_norm": 1.030186653137207, "tokens_per_second": 4828235.0079607945, "eta_seconds": 1863.3687186241152, "data_wait_seconds": 1.9073486328125e-05} {"step": 24330, "total_steps": 28610, "tokens": 51023708160, "source": "dclm", "phase": 5, "loss": 3.5070368349552155, "ce_loss": 3.4924247562885284, "logit_z_loss": 26.706458806991577, "router_aux_loss": 1.366030514240265, "router_z_loss": 0.9517624899744987, "routing_entropy_normalized": 0.6689706673361334, "learning_rate": 0.002999959779776132, "grad_norm": 1.5307267904281616, "tokens_per_second": 4290874.832225691, "eta_seconds": 2091.836958885193, "data_wait_seconds": 2.4527311325073242e-05} {"step": 24340, "total_steps": 28610, "tokens": 51044679680, "source": "finemath", "phase": 5, "loss": 2.1969234943389893, "ce_loss": 2.1818026304244995, "logit_z_loss": 35.76699256896973, "router_aux_loss": 1.4332779496908188, "router_z_loss": 0.7880992442369461, "routing_entropy_normalized": 0.636012269736451, "learning_rate": 0.0029998391212614156, "grad_norm": 0.9679720997810364, "tokens_per_second": 3880132.788688906, "eta_seconds": 2307.8692219257355, "data_wait_seconds": 1.806020736694336e-05} {"step": 24350, "total_steps": 28610, "tokens": 51065651200, "source": "dclm", "phase": 5, "loss": 3.550623506307602, "ce_loss": 3.5363031923770905, "logit_z_loss": 26.390364170074463, "router_aux_loss": 1.3233077675104141, "router_z_loss": 1.0872344002127647, "routing_entropy_normalized": 0.6857811732388107, "learning_rate": 0.0029996380309264023, "grad_norm": 0.794634222984314, "tokens_per_second": 4095992.080703788, "eta_seconds": 2181.1242170333862, "data_wait_seconds": 2.1517276763916016e-05} {"step": 24360, "total_steps": 28610, "tokens": 51086622720, "source": "cosmopedia_v2", "phase": 5, "loss": 2.751794934272766, "ce_loss": 2.736626237630844, "logit_z_loss": 30.38798475265503, "router_aux_loss": 1.3765950202941895, "router_z_loss": 1.402696669101715, "routing_entropy_normalized": 0.6566539642921191, "learning_rate": 0.002999356519554955, "grad_norm": 2.7786381244659424, "tokens_per_second": 4197839.978470787, "eta_seconds": 2123.2100427150726, "data_wait_seconds": 1.9252300262451172e-05} {"step": 24370, "total_steps": 28610, "tokens": 51107594240, "source": "finemath", "phase": 5, "loss": 2.233479380607605, "ce_loss": 2.2174739837646484, "logit_z_loss": 38.76689958572388, "router_aux_loss": 1.515777364373207, "router_z_loss": 0.8476099744439125, "routing_entropy_normalized": 0.6207621397086467, "learning_rate": 0.002998994602243675, "grad_norm": 1.3261348009109497, "tokens_per_second": 3615891.1152287545, "eta_seconds": 2459.123960494995, "data_wait_seconds": 1.9788742065429688e-05} {"step": 24380, "total_steps": 28610, "tokens": 51128565760, "source": "cosmopedia_v2", "phase": 5, "loss": 2.81305992603302, "ce_loss": 2.7969911992549896, "logit_z_loss": 27.60040831565857, "router_aux_loss": 1.4412277191877365, "router_z_loss": 1.6564268171787262, "routing_entropy_normalized": 0.6325815934269572, "learning_rate": 0.002998552298401089, "grad_norm": 2.129425287246704, "tokens_per_second": 4058468.187251017, "eta_seconds": 2185.788467645645, "data_wait_seconds": 2.1338462829589844e-05} {"step": 24390, "total_steps": 28610, "tokens": 51149537280, "source": "fineweb_edu", "phase": 5, "loss": 3.4075824320316315, "ce_loss": 3.39120015501976, "logit_z_loss": 26.077214241027832, "router_aux_loss": 1.4928022772073746, "router_z_loss": 1.4542070627212524, "routing_entropy_normalized": 0.6687255162462996, "learning_rate": 0.00299802963174661, "grad_norm": 1.8868037462234497, "tokens_per_second": 4074414.9191897726, "eta_seconds": 2172.086450576782, "data_wait_seconds": 2.142786979675293e-05} {"step": 24400, "total_steps": 28610, "tokens": 51170508800, "source": "cosmopedia_v2", "phase": 5, "loss": 2.770819365978241, "ce_loss": 2.7546383142471313, "logit_z_loss": 29.800302267074585, "router_aux_loss": 1.4330823123455048, "router_z_loss": 1.8502330631017685, "routing_entropy_normalized": 0.6702999400286076, "learning_rate": 0.002997426630309264, "grad_norm": 1.4446903467178345, "tokens_per_second": 4465664.03290788, "eta_seconds": 1977.0878093242643, "data_wait_seconds": 2.008676528930664e-05} {"step": 24410, "total_steps": 28610, "tokens": 51191480320, "source": "npset2", "phase": 5, "loss": 1.733597308397293, "ce_loss": 1.7172404676675797, "logit_z_loss": 38.033448219299316, "router_aux_loss": 1.4674998223781586, "router_z_loss": 1.681825190782547, "routing_entropy_normalized": 0.6751166703302844, "learning_rate": 0.0029967433264261887, "grad_norm": 0.9798464775085449, "tokens_per_second": 3880604.9065874247, "eta_seconds": 2269.7591257095337, "data_wait_seconds": 2.2113323211669922e-05} {"step": 24420, "total_steps": 28610, "tokens": 51212451840, "source": "fineweb_edu", "phase": 5, "loss": 3.365636020898819, "ce_loss": 3.3490683138370514, "logit_z_loss": 25.608685970306396, "router_aux_loss": 1.4723940044641495, "router_z_loss": 1.8438000530004501, "routing_entropy_normalized": 0.6557460062289503, "learning_rate": 0.0029959797567408976, "grad_norm": 1.1904966831207275, "tokens_per_second": 4052097.2869929285, "eta_seconds": 2168.523176431656, "data_wait_seconds": 2.1338462829589844e-05} {"step": 24430, "total_steps": 28610, "tokens": 51233423360, "source": "dclm", "phase": 5, "loss": 3.578868269920349, "ce_loss": 3.5621406733989716, "logit_z_loss": 26.17246651649475, "router_aux_loss": 1.5067585557699203, "router_z_loss": 1.6600012183189392, "routing_entropy_normalized": 0.6680826793433281, "learning_rate": 0.002995135962201315, "grad_norm": 1.2362875938415527, "tokens_per_second": 3686577.7433261224, "eta_seconds": 2377.8409056663513, "data_wait_seconds": 2.1845102310180664e-05} {"step": 24440, "total_steps": 28610, "tokens": 51254394880, "source": "finemath", "phase": 5, "loss": 2.229671597480774, "ce_loss": 2.214169144630432, "logit_z_loss": 36.8130989074707, "router_aux_loss": 1.398377686738968, "router_z_loss": 1.5187035501003265, "routing_entropy_normalized": 0.6802591412630644, "learning_rate": 0.002994211988057582, "grad_norm": 0.8024020791053772, "tokens_per_second": 4125460.232859037, "eta_seconds": 2119.793513059616, "data_wait_seconds": 1.7702579498291016e-05} {"step": 24450, "total_steps": 28610, "tokens": 51275366400, "source": "dclm", "phase": 5, "loss": 3.5289065539836884, "ce_loss": 3.5132123827934265, "logit_z_loss": 26.461761236190796, "router_aux_loss": 1.3864035308361053, "router_z_loss": 1.8301292061805725, "routing_entropy_normalized": 0.6818651949327884, "learning_rate": 0.002993207883859627, "grad_norm": 0.8651490807533264, "tokens_per_second": 3993269.9051494934, "eta_seconds": 2184.713912963867, "data_wait_seconds": 2.6047229766845703e-05} {"step": 24460, "total_steps": 28610, "tokens": 51296337920, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7856735289096832, "ce_loss": 2.7699772119522095, "logit_z_loss": 29.48853898048401, "router_aux_loss": 1.3539293557405472, "router_z_loss": 2.1570434272289276, "routing_entropy_normalized": 0.676737298418222, "learning_rate": 0.0029921237034545114, "grad_norm": 3.2332825660705566, "tokens_per_second": 4370959.456802224, "eta_seconds": 1991.1373889446259, "data_wait_seconds": 2.2917985916137695e-05} {"step": 24470, "total_steps": 28610, "tokens": 51317309440, "source": "finemath", "phase": 5, "loss": 2.1939545273780823, "ce_loss": 2.178472101688385, "logit_z_loss": 37.5544638633728, "router_aux_loss": 1.4047763496637344, "router_z_loss": 1.4346741139888763, "routing_entropy_normalized": 0.6542262066957996, "learning_rate": 0.00299095950498354, "grad_norm": 0.8792964220046997, "tokens_per_second": 3915549.410945741, "eta_seconds": 2217.3668031692505, "data_wait_seconds": 2.238154411315918e-05} {"step": 24480, "total_steps": 28610, "tokens": 51338280960, "source": "cosmopedia_v2", "phase": 5, "loss": 2.744414061307907, "ce_loss": 2.727745831012726, "logit_z_loss": 30.086064338684082, "router_aux_loss": 1.4578351378440857, "router_z_loss": 2.0898706316947937, "routing_entropy_normalized": 0.6223829529125939, "learning_rate": 0.002989715350879144, "grad_norm": 2.150811195373535, "tokens_per_second": 3555937.2699066913, "eta_seconds": 2435.7116289138794, "data_wait_seconds": 2.250075340270996e-05} {"step": 24490, "total_steps": 28610, "tokens": 51359252480, "source": "fineweb_edu", "phase": 5, "loss": 3.3798795640468597, "ce_loss": 3.3641267120838165, "logit_z_loss": 25.479593992233276, "router_aux_loss": 1.3911864310503006, "router_z_loss": 1.840990662574768, "routing_entropy_normalized": 0.6628139580752317, "learning_rate": 0.0029883913078615305, "grad_norm": 0.9219225645065308, "tokens_per_second": 4157574.2853479055, "eta_seconds": 2078.198884010315, "data_wait_seconds": 2.22623348236084e-05} {"step": 24500, "total_steps": 28610, "tokens": 51380224000, "source": "cosmopedia_v2", "phase": 5, "loss": 2.79342120885849, "ce_loss": 2.7775352001190186, "logit_z_loss": 27.471779346466064, "router_aux_loss": 1.372716948390007, "router_z_loss": 2.1588582396507263, "routing_entropy_normalized": 0.6604553149565169, "learning_rate": 0.0029869874469351104, "grad_norm": 1.168899416923523, "tokens_per_second": 4223995.711800339, "eta_seconds": 2040.5547988414767, "data_wait_seconds": 2.1129846572875977e-05} {"step": 24510, "total_steps": 28610, "tokens": 51401195520, "source": "npset2", "phase": 5, "loss": 1.749393880367279, "ce_loss": 1.7338081151247025, "logit_z_loss": 39.034573554992676, "router_aux_loss": 1.3932065218687057, "router_z_loss": 1.6537040024995804, "routing_entropy_normalized": 0.681333486789156, "learning_rate": 0.0029855038433846837, "grad_norm": 1.010117769241333, "tokens_per_second": 3757643.6961870403, "eta_seconds": 2288.2220602035522, "data_wait_seconds": 1.990795135498047e-05} {"step": 24520, "total_steps": 28610, "tokens": 51422167040, "source": "fineweb_edu", "phase": 5, "loss": 3.3578164875507355, "ce_loss": 3.3427359461784363, "logit_z_loss": 26.04134225845337, "router_aux_loss": 1.3331965804100037, "router_z_loss": 1.7485280930995941, "routing_entropy_normalized": 0.6764959031510576, "learning_rate": 0.0029839405767714063, "grad_norm": 1.3077399730682373, "tokens_per_second": 4371985.5424838085, "eta_seconds": 1961.8893055915833, "data_wait_seconds": 2.104043960571289e-05} {"step": 24530, "total_steps": 28610, "tokens": 51443138560, "source": "dclm", "phase": 5, "loss": 3.533504545688629, "ce_loss": 3.5185817778110504, "logit_z_loss": 27.69633150100708, "router_aux_loss": 1.328376591205597, "router_z_loss": 1.6389935612678528, "routing_entropy_normalized": 0.6593728268172653, "learning_rate": 0.0029822977309285226, "grad_norm": 1.3004814386367798, "tokens_per_second": 4024789.173105834, "eta_seconds": 2125.9200897216797, "data_wait_seconds": 2.1517276763916016e-05} {"step": 24540, "total_steps": 28610, "tokens": 51464110080, "source": "finemath", "phase": 5, "loss": 2.280679374933243, "ce_loss": 2.2659238278865814, "logit_z_loss": 37.776527881622314, "router_aux_loss": 1.3606626242399216, "router_z_loss": 1.1489444822072983, "routing_entropy_normalized": 0.6685573731617356, "learning_rate": 0.0029805753939568694, "grad_norm": 1.5708117485046387, "tokens_per_second": 3287418.8463337426, "eta_seconds": 2596.3861129283905, "data_wait_seconds": 2.2202730178833008e-05} {"step": 24550, "total_steps": 28610, "tokens": 51485081600, "source": "dclm", "phase": 5, "loss": 3.502898246049881, "ce_loss": 3.488318532705307, "logit_z_loss": 27.883023023605347, "router_aux_loss": 1.3110797852277756, "router_z_loss": 1.4689572602510452, "routing_entropy_normalized": 0.6559795736133083, "learning_rate": 0.002978773658220151, "grad_norm": 1.0535694360733032, "tokens_per_second": 4032182.777796165, "eta_seconds": 2111.6198320388794, "data_wait_seconds": 1.9609928131103516e-05} {"step": 24560, "total_steps": 28610, "tokens": 51506053120, "source": "cosmopedia_v2", "phase": 5, "loss": 2.752013236284256, "ce_loss": 2.7372089624404907, "logit_z_loss": 30.091591835021973, "router_aux_loss": 1.292748898267746, "router_z_loss": 1.8768514692783356, "routing_entropy_normalized": 0.679222637226605, "learning_rate": 0.0029768926203399874, "grad_norm": 1.5403648614883423, "tokens_per_second": 4110430.6365193413, "eta_seconds": 2066.3201379776, "data_wait_seconds": 2.059340476989746e-05} {"step": 24570, "total_steps": 28610, "tokens": 51527024640, "source": "finemath", "phase": 5, "loss": 2.1984058171510696, "ce_loss": 2.1836565732955933, "logit_z_loss": 36.44497728347778, "router_aux_loss": 1.3657557964324951, "router_z_loss": 1.0916499868035316, "routing_entropy_normalized": 0.692504145333687, "learning_rate": 0.0029749323811907318, "grad_norm": 0.9105506539344788, "tokens_per_second": 3657146.636811095, "eta_seconds": 2316.6952056884766, "data_wait_seconds": 2.2351741790771484e-05} {"step": 24580, "total_steps": 28610, "tokens": 51547996160, "source": "cosmopedia_v2", "phase": 5, "loss": 2.757999926805496, "ce_loss": 2.742664396762848, "logit_z_loss": 28.211222410202026, "router_aux_loss": 1.3604266047477722, "router_z_loss": 1.7312824428081512, "routing_entropy_normalized": 0.6621617611752226, "learning_rate": 0.00297289304589406, "grad_norm": 3.1794557571411133, "tokens_per_second": 4319747.841662977, "eta_seconds": 1956.4851629734042, "data_wait_seconds": 2.2739171981811523e-05} {"step": 24590, "total_steps": 28610, "tokens": 51568967680, "source": "fineweb_edu", "phase": 5, "loss": 3.392238438129425, "ce_loss": 3.3774066865444183, "logit_z_loss": 26.640925884246826, "router_aux_loss": 1.3324724286794662, "router_z_loss": 1.5070690512657166, "routing_entropy_normalized": 0.6632662995544124, "learning_rate": 0.0029707747238133355, "grad_norm": 1.0633742809295654, "tokens_per_second": 4498891.87174427, "eta_seconds": 1873.9172401428225, "data_wait_seconds": 2.1845102310180664e-05} {"step": 24600, "total_steps": 28610, "tokens": 51589939200, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7160869240760803, "ce_loss": 2.699653387069702, "logit_z_loss": 29.70316243171692, "router_aux_loss": 1.4702236205339432, "router_z_loss": 1.7312409281730652, "routing_entropy_normalized": 0.6511422167404483, "learning_rate": 0.002968577528547742, "grad_norm": 1.0087404251098633, "tokens_per_second": 3926094.079042293, "eta_seconds": 2141.970964193344, "data_wait_seconds": 2.1904706954956055e-05} {"step": 24610, "total_steps": 28610, "tokens": 51610910720, "source": "npset2", "phase": 5, "loss": 1.7158115059137344, "ce_loss": 1.6997532099485397, "logit_z_loss": 37.87843370437622, "router_aux_loss": 1.4444717019796371, "router_z_loss": 1.6135940253734589, "routing_entropy_normalized": 0.657181164185306, "learning_rate": 0.0029663015779261945, "grad_norm": 0.9197565913200378, "tokens_per_second": 4412253.968883806, "eta_seconds": 1901.2069702148438, "data_wait_seconds": 2.3871660232543945e-05} {"step": 24620, "total_steps": 28610, "tokens": 51631882240, "source": "fineweb_edu", "phase": 5, "loss": 3.4917178750038147, "ce_loss": 3.4737722277641296, "logit_z_loss": 26.974218368530273, "router_aux_loss": 1.602147400379181, "router_z_loss": 1.9241393208503723, "routing_entropy_normalized": 0.6561035384703741, "learning_rate": 0.0029639469940010168, "grad_norm": 1.8344985246658325, "tokens_per_second": 3806616.831589323, "eta_seconds": 2198.1819684505463, "data_wait_seconds": 2.1249055862426758e-05} {"step": 24630, "total_steps": 28610, "tokens": 51652853760, "source": "dclm", "phase": 5, "loss": 3.6543825268745422, "ce_loss": 3.6358193457126617, "logit_z_loss": 27.050034523010254, "router_aux_loss": 1.6590331196784973, "router_z_loss": 1.9728255420923233, "routing_entropy_normalized": 0.6243151490915673, "learning_rate": 0.0029615139030413993, "grad_norm": 1.5099610090255737, "tokens_per_second": 3776620.1016714117, "eta_seconds": 2210.088580608368, "data_wait_seconds": 1.963973045349121e-05} {"step": 24640, "total_steps": 28610, "tokens": 51673825280, "source": "finemath", "phase": 5, "loss": 2.306378036737442, "ce_loss": 2.2886548042297363, "logit_z_loss": 36.95627307891846, "router_aux_loss": 1.5938691943883896, "router_z_loss": 1.7845647186040878, "routing_entropy_normalized": 0.6577377674976621, "learning_rate": 0.002959002435526626, "grad_norm": 1.1843352317810059, "tokens_per_second": 3903700.2578420895, "eta_seconds": 2132.769651889801, "data_wait_seconds": 2.3752450942993164e-05} {"step": 24650, "total_steps": 28610, "tokens": 51694796800, "source": "dclm", "phase": 5, "loss": 3.620202273130417, "ce_loss": 3.604079991579056, "logit_z_loss": 26.80538296699524, "router_aux_loss": 1.3765792399644852, "router_z_loss": 2.3564658164978027, "routing_entropy_normalized": 0.6926032060082296, "learning_rate": 0.002956412726139078, "grad_norm": 1.237443208694458, "tokens_per_second": 4125921.1737905834, "eta_seconds": 2012.8164281845093, "data_wait_seconds": 1.817941665649414e-05} {"step": 24660, "total_steps": 28610, "tokens": 51715768320, "source": "cosmopedia_v2", "phase": 5, "loss": 2.884728491306305, "ce_loss": 2.868671029806137, "logit_z_loss": 30.328741788864136, "router_aux_loss": 1.322726845741272, "router_z_loss": 2.8302186131477356, "routing_entropy_normalized": 0.6679602884498002, "learning_rate": 0.0029537449137570105, "grad_norm": 2.02366304397583, "tokens_per_second": 3283163.726354407, "eta_seconds": 2523.1000006198883, "data_wait_seconds": 1.996755599975586e-05} {"step": 24670, "total_steps": 28610, "tokens": 51736739840, "source": "finemath", "phase": 5, "loss": 2.2703080475330353, "ce_loss": 2.2552725076675415, "logit_z_loss": 37.32369375228882, "router_aux_loss": 1.3222352713346481, "router_z_loss": 1.8131790608167648, "routing_entropy_normalized": 0.7041458672646693, "learning_rate": 0.0029509991414471046, "grad_norm": 0.9700992703437805, "tokens_per_second": 3360479.3553938293, "eta_seconds": 2458.809594154358, "data_wait_seconds": 2.2113323211669922e-05} {"step": 24680, "total_steps": 28610, "tokens": 51757711360, "source": "cosmopedia_v2", "phase": 5, "loss": 2.867507725954056, "ce_loss": 2.851073920726776, "logit_z_loss": 29.610145092010498, "router_aux_loss": 1.3719487190246582, "router_z_loss": 2.714336097240448, "routing_entropy_normalized": 0.6638364502778302, "learning_rate": 0.0029481755564567972, "grad_norm": 2.796952962875366, "tokens_per_second": 4207920.662976818, "eta_seconds": 1958.641338586807, "data_wait_seconds": 2.1070241928100586e-05} {"step": 24690, "total_steps": 28610, "tokens": 51778682880, "source": "fineweb_edu", "phase": 5, "loss": 3.465627521276474, "ce_loss": 3.4495673775672913, "logit_z_loss": 24.983651399612427, "router_aux_loss": 1.3925691395998, "router_z_loss": 2.134512186050415, "routing_entropy_normalized": 0.6943809555518528, "learning_rate": 0.0029452743102063817, "grad_norm": 0.9883233308792114, "tokens_per_second": 4093760.7539588776, "eta_seconds": 2008.137830734253, "data_wait_seconds": 2.09808349609375e-05} {"step": 24700, "total_steps": 28610, "tokens": 51799654400, "source": "cosmopedia_v2", "phase": 5, "loss": 2.8425679802894592, "ce_loss": 2.827117770910263, "logit_z_loss": 27.344905614852905, "router_aux_loss": 1.3004362136125565, "router_z_loss": 2.445816069841385, "routing_entropy_normalized": 0.7076212867724357, "learning_rate": 0.002942295558280889, "grad_norm": 1.242020606994629, "tokens_per_second": 4125635.1538145375, "eta_seconds": 1987.5398609638214, "data_wait_seconds": 2.3186206817626953e-05} {"step": 24710, "total_steps": 28610, "tokens": 51820625920, "source": "npset2", "phase": 5, "loss": 1.8581374883651733, "ce_loss": 1.84329392015934, "logit_z_loss": 36.45030689239502, "router_aux_loss": 1.298251435160637, "router_z_loss": 1.861038938164711, "routing_entropy_normalized": 0.6966908945159804, "learning_rate": 0.0029392394604217465, "grad_norm": 1.092721939086914, "tokens_per_second": 4082026.4337690095, "eta_seconds": 2003.6354327201843, "data_wait_seconds": 1.9997358322143555e-05} {"step": 24720, "total_steps": 28610, "tokens": 51841597440, "source": "fineweb_edu", "phase": 5, "loss": 3.441134810447693, "ce_loss": 3.425627291202545, "logit_z_loss": 26.074855089187622, "router_aux_loss": 1.3541320711374283, "router_z_loss": 1.9661746323108673, "routing_entropy_normalized": 0.664769499927284, "learning_rate": 0.002936106180518206, "grad_norm": 0.9485297799110413, "tokens_per_second": 3831008.0859321393, "eta_seconds": 2129.445069551468, "data_wait_seconds": 2.053380012512207e-05} {"step": 24730, "total_steps": 28610, "tokens": 51862568960, "source": "dclm", "phase": 5, "loss": 3.6236550509929657, "ce_loss": 3.6087179481983185, "logit_z_loss": 25.82196021080017, "router_aux_loss": 1.3270062506198883, "router_z_loss": 1.6670597195625305, "routing_entropy_normalized": 0.6920301876172344, "learning_rate": 0.00293289588659856, "grad_norm": 0.9471623301506042, "tokens_per_second": 4017454.0098448507, "eta_seconds": 2025.399603843689, "data_wait_seconds": 2.0682811737060547e-05} {"step": 24740, "total_steps": 28610, "tokens": 51883540480, "source": "finemath", "phase": 5, "loss": 2.213335543870926, "ce_loss": 2.1985355019569397, "logit_z_loss": 36.289238929748535, "router_aux_loss": 1.3629849553108215, "router_z_loss": 1.1701751947402954, "routing_entropy_normalized": 0.6815796239492217, "learning_rate": 0.002929608750821129, "grad_norm": 1.5042389631271362, "tokens_per_second": 3020163.92085922, "eta_seconds": 2687.264152765274, "data_wait_seconds": 2.1010637283325195e-05} {"step": 24750, "total_steps": 28610, "tokens": 51904512000, "source": "dclm", "phase": 5, "loss": 3.5500487089157104, "ce_loss": 3.535158097743988, "logit_z_loss": 26.757225513458252, "router_aux_loss": 1.3377691507339478, "router_z_loss": 1.512998953461647, "routing_entropy_normalized": 0.6706649962639689, "learning_rate": 0.0029262449494650293, "grad_norm": 0.9698466062545776, "tokens_per_second": 3756823.5949124363, "eta_seconds": 2154.747625350952, "data_wait_seconds": 2.2113323211669922e-05} {"step": 24760, "total_steps": 28610, "tokens": 51925483520, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7867183089256287, "ce_loss": 2.771235466003418, "logit_z_loss": 29.263253211975098, "router_aux_loss": 1.3553748279809952, "router_z_loss": 1.9290651232004166, "routing_entropy_normalized": 0.6681059574474656, "learning_rate": 0.002922804662920718, "grad_norm": 2.5453381538391113, "tokens_per_second": 4060823.91667403, "eta_seconds": 1988.2751297950745, "data_wait_seconds": 2.2083520889282227e-05} {"step": 24770, "total_steps": 28610, "tokens": 51946455040, "source": "finemath", "phase": 5, "loss": 2.285903811454773, "ce_loss": 2.271328240633011, "logit_z_loss": 39.11197757720947, "router_aux_loss": 1.3527375012636185, "router_z_loss": 1.0481807440519333, "routing_entropy_normalized": 0.6914118441859968, "learning_rate": 0.002919288075680323, "grad_norm": 1.2028716802597046, "tokens_per_second": 4233148.6502399845, "eta_seconds": 1902.3815002441404, "data_wait_seconds": 2.1755695343017578e-05} {"step": 24780, "total_steps": 28610, "tokens": 51967426560, "source": "cosmopedia_v2", "phase": 5, "loss": 2.8127997517585754, "ce_loss": 2.797815203666687, "logit_z_loss": 27.633989810943604, "router_aux_loss": 1.3082483559846878, "router_z_loss": 1.902065947651863, "routing_entropy_normalized": 0.6803527234749875, "learning_rate": 0.0029156953763277444, "grad_norm": 2.1179864406585693, "tokens_per_second": 4260184.800855903, "eta_seconds": 1885.385854244232, "data_wait_seconds": 2.0563602447509766e-05} {"step": 24790, "total_steps": 28610, "tokens": 51988398080, "source": "fineweb_edu", "phase": 5, "loss": 3.413728892803192, "ce_loss": 3.399111360311508, "logit_z_loss": 26.29685616493225, "router_aux_loss": 1.307836964726448, "router_z_loss": 1.5392144471406937, "routing_entropy_normalized": 0.6944608802594388, "learning_rate": 0.002912026757528546, "grad_norm": 1.6337299346923828, "tokens_per_second": 3843010.832670629, "eta_seconds": 2084.5948629379272, "data_wait_seconds": 2.440810203552246e-05} {"step": 24800, "total_steps": 28610, "tokens": 52009369600, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7913796603679657, "ce_loss": 2.7755968868732452, "logit_z_loss": 28.28334331512451, "router_aux_loss": 1.4072147905826569, "router_z_loss": 1.7106590569019318, "routing_entropy_normalized": 0.6497848708507883, "learning_rate": 0.0029082824160196197, "grad_norm": 1.3964201211929321, "tokens_per_second": 4346977.781429983, "eta_seconds": 1838.0929284095762, "data_wait_seconds": 2.1755695343017578e-05} {"step": 24810, "total_steps": 28610, "tokens": 52030341120, "source": "npset2", "phase": 5, "loss": 1.7448647618293762, "ce_loss": 1.7294423282146454, "logit_z_loss": 39.81084203720093, "router_aux_loss": 1.4154190719127655, "router_z_loss": 1.2682521343231201, "routing_entropy_normalized": 0.6626162793643741, "learning_rate": 0.0029044625525986383, "grad_norm": 0.9602336287498474, "tokens_per_second": 3525830.593571472, "eta_seconds": 2260.227026939392, "data_wait_seconds": 1.9758939743041992e-05} {"step": 24820, "total_steps": 28610, "tokens": 52051312640, "source": "fineweb_edu", "phase": 5, "loss": 3.4141983091831207, "ce_loss": 3.3985036611557007, "logit_z_loss": 26.259061574935913, "router_aux_loss": 1.4189595580101013, "router_z_loss": 1.5051139146089554, "routing_entropy_normalized": 0.6701262337019119, "learning_rate": 0.002900567372113283, "grad_norm": 1.4277193546295166, "tokens_per_second": 4253124.689798856, "eta_seconds": 1868.7921609878538, "data_wait_seconds": 2.2232532501220703e-05} {"step": 24830, "total_steps": 28610, "tokens": 52072284160, "source": "dclm", "phase": 5, "loss": 3.5784744918346405, "ce_loss": 3.564123660326004, "logit_z_loss": 25.937243700027466, "router_aux_loss": 1.2964590340852737, "router_z_loss": 1.3862379640340805, "routing_entropy_normalized": 0.697041045014762, "learning_rate": 0.002896597083450262, "grad_norm": 1.091768503189087, "tokens_per_second": 4051160.803364878, "eta_seconds": 1956.781017780304, "data_wait_seconds": 2.3812055587768555e-05} {"step": 24840, "total_steps": 28610, "tokens": 52093255680, "source": "finemath", "phase": 5, "loss": 2.32195445895195, "ce_loss": 2.3073013722896576, "logit_z_loss": 37.7993323802948, "router_aux_loss": 1.3661419749259949, "router_z_loss": 0.9916506446897984, "routing_entropy_normalized": 0.6673961112882774, "learning_rate": 0.002892551899524109, "grad_norm": 0.5348780751228333, "tokens_per_second": 3970639.5267603872, "eta_seconds": 1991.1812660694122, "data_wait_seconds": 1.9997358322143555e-05} {"step": 24850, "total_steps": 28610, "tokens": 52114227200, "source": "dclm", "phase": 5, "loss": 3.5162066519260406, "ce_loss": 3.5015780329704285, "logit_z_loss": 27.072865962982178, "router_aux_loss": 1.3346206992864609, "router_z_loss": 1.2824152559041977, "routing_entropy_normalized": 0.6567043629682129, "learning_rate": 0.0028884320372657593, "grad_norm": 0.7161967158317566, "tokens_per_second": 4178642.1020950805, "eta_seconds": 1887.0463962554932, "data_wait_seconds": 2.3871660232543945e-05} {"step": 24860, "total_steps": 28610, "tokens": 52135198720, "source": "cosmopedia_v2", "phase": 5, "loss": 2.8077380061149597, "ce_loss": 2.792634040117264, "logit_z_loss": 29.382993698120117, "router_aux_loss": 1.3354660421609879, "router_z_loss": 1.7493383884429932, "routing_entropy_normalized": 0.6854399418809433, "learning_rate": 0.002884237717610924, "grad_norm": 1.873749017715454, "tokens_per_second": 3771738.5718014413, "eta_seconds": 2085.0649774074554, "data_wait_seconds": 1.9788742065429688e-05} {"step": 24870, "total_steps": 28610, "tokens": 52156170240, "source": "finemath", "phase": 5, "loss": 2.232254296541214, "ce_loss": 2.2168936729431152, "logit_z_loss": 39.03823184967041, "router_aux_loss": 1.4522645324468613, "router_z_loss": 0.8379176035523415, "routing_entropy_normalized": 0.6434682414265854, "learning_rate": 0.0028799691654882364, "grad_norm": 1.3009356260299683, "tokens_per_second": 4126471.4577051937, "eta_seconds": 1900.7397871017456, "data_wait_seconds": 0.01995226740837097} {"step": 24880, "total_steps": 28610, "tokens": 52177141760, "source": "cosmopedia_v2", "phase": 5, "loss": 2.8235141038894653, "ce_loss": 2.8074798583984375, "logit_z_loss": 27.693150520324707, "router_aux_loss": 1.437588557600975, "router_z_loss": 1.6583337187767029, "routing_entropy_normalized": 0.6562347958541516, "learning_rate": 0.0028756266098071926, "grad_norm": 2.172868013381958, "tokens_per_second": 4282725.559951195, "eta_seconds": 1826.4950323104856, "data_wait_seconds": 2.0384788513183594e-05} {"step": 24890, "total_steps": 28610, "tokens": 52198113280, "source": "fineweb_edu", "phase": 5, "loss": 3.39290389418602, "ce_loss": 3.377029150724411, "logit_z_loss": 26.024184703826904, "router_aux_loss": 1.4423707127571106, "router_z_loss": 1.451010212302208, "routing_entropy_normalized": 0.6537841485914505, "learning_rate": 0.002871210283445875, "grad_norm": 1.4458792209625244, "tokens_per_second": 4106860.8373023774, "eta_seconds": 1899.603066444397, "data_wait_seconds": 2.0325183868408203e-05} {"step": 24900, "total_steps": 28610, "tokens": 52219084800, "source": "cosmopedia_v2", "phase": 5, "loss": 2.846375495195389, "ce_loss": 2.8311305046081543, "logit_z_loss": 29.115986824035645, "router_aux_loss": 1.34239262342453, "router_z_loss": 1.8210370242595673, "routing_entropy_normalized": 0.6660951834813392, "learning_rate": 0.002866720423238463, "grad_norm": 2.099285840988159, "tokens_per_second": 4063605.856856239, "eta_seconds": 1914.662542104721, "data_wait_seconds": 2.0831823348999023e-05} {"step": 24910, "total_steps": 28610, "tokens": 52240056320, "source": "npset2", "phase": 5, "loss": 1.7717477828264236, "ce_loss": 1.7569123953580856, "logit_z_loss": 39.26070594787598, "router_aux_loss": 1.341521680355072, "router_z_loss": 1.4201503992080688, "routing_entropy_normalized": 0.6823193959317152, "learning_rate": 0.0028621572699625343, "grad_norm": 0.7234165072441101, "tokens_per_second": 4012386.3658254123, "eta_seconds": 1933.8771724700928, "data_wait_seconds": 2.3573637008666992e-05} {"step": 24920, "total_steps": 28610, "tokens": 52261027840, "source": "fineweb_edu", "phase": 5, "loss": 3.4341034591197968, "ce_loss": 3.418309599161148, "logit_z_loss": 25.87724471092224, "router_aux_loss": 1.4259158968925476, "router_z_loss": 1.5346879661083221, "routing_entropy_normalized": 0.6480931049305431, "learning_rate": 0.0028575210683261503, "grad_norm": 1.3823256492614746, "tokens_per_second": 4124902.287701201, "eta_seconds": 1876.042228460312, "data_wait_seconds": 2.086162567138672e-05} {"step": 24930, "total_steps": 28610, "tokens": 52281999360, "source": "dclm", "phase": 5, "loss": 3.6082204580307007, "ce_loss": 3.592264324426651, "logit_z_loss": 27.002198934555054, "router_aux_loss": 1.450129896402359, "router_z_loss": 1.4548768103122711, "routing_entropy_normalized": 0.6522485763897001, "learning_rate": 0.0028528120669547344, "grad_norm": 1.7633510828018188, "tokens_per_second": 4008035.4073138856, "eta_seconds": 1925.5117721557617, "data_wait_seconds": 2.3573637008666992e-05} {"step": 24940, "total_steps": 28610, "tokens": 52302970880, "source": "finemath", "phase": 5, "loss": 2.1261900514364243, "ce_loss": 2.1098869293928146, "logit_z_loss": 37.23755502700806, "router_aux_loss": 1.5102118700742722, "router_z_loss": 1.2009457349777222, "routing_entropy_normalized": 0.6640308301642291, "learning_rate": 0.0028480305183777393, "grad_norm": 0.5828309655189514, "tokens_per_second": 4094756.1151658925, "eta_seconds": 1879.6108055114746, "data_wait_seconds": 1.7017126083374023e-05} {"step": 24950, "total_steps": 28610, "tokens": 52323942400, "source": "dclm", "phase": 5, "loss": 3.587846130132675, "ce_loss": 3.571740299463272, "logit_z_loss": 27.14966607093811, "router_aux_loss": 1.4629874676465988, "router_z_loss": 1.476009115576744, "routing_entropy_normalized": 0.6411786424058704, "learning_rate": 0.0028431766790151035, "grad_norm": 1.3113843202590942, "tokens_per_second": 4205638.9497182835, "eta_seconds": 1825.0678224563599, "data_wait_seconds": 2.181529998779297e-05} {"step": 24960, "total_steps": 28610, "tokens": 52344913920, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7835749089717865, "ce_loss": 2.7673842012882233, "logit_z_loss": 29.280305862426758, "router_aux_loss": 1.4169478863477707, "router_z_loss": 2.021232843399048, "routing_entropy_normalized": 0.6673656258913959, "learning_rate": 0.002838250809163502, "grad_norm": 1.7740503549575806, "tokens_per_second": 3974689.602531628, "eta_seconds": 1925.8371257781982, "data_wait_seconds": 1.9788742065429688e-05} {"step": 24970, "total_steps": 28610, "tokens": 52365885440, "source": "finemath", "phase": 5, "loss": 2.1171946078538895, "ce_loss": 2.1016162037849426, "logit_z_loss": 39.41314649581909, "router_aux_loss": 1.4324513673782349, "router_z_loss": 1.2539370954036713, "routing_entropy_normalized": 0.6813196770447206, "learning_rate": 0.0028332531729823853, "grad_norm": 0.9935677647590637, "tokens_per_second": 3961665.6701240195, "eta_seconds": 1926.8746824264526, "data_wait_seconds": 2.2023916244506836e-05} {"step": 24980, "total_steps": 28610, "tokens": 52386856960, "source": "cosmopedia_v2", "phase": 5, "loss": 2.781809777021408, "ce_loss": 2.7663948237895966, "logit_z_loss": 28.407716035842896, "router_aux_loss": 1.3373088538646698, "router_z_loss": 2.041871339082718, "routing_entropy_normalized": 0.7029057674492862, "learning_rate": 0.002828184038479815, "grad_norm": 1.752612829208374, "tokens_per_second": 3773281.6316602263, "eta_seconds": 2017.5175094604492, "data_wait_seconds": 2.0265579223632812e-05} {"step": 24990, "total_steps": 28610, "tokens": 52407828480, "source": "fineweb_edu", "phase": 5, "loss": 3.385870933532715, "ce_loss": 3.370130866765976, "logit_z_loss": 26.959777116775513, "router_aux_loss": 1.3927781283855438, "router_z_loss": 1.8123130351305008, "routing_entropy_normalized": 0.6664216916114075, "learning_rate": 0.00282304367749809, "grad_norm": 0.7982301115989685, "tokens_per_second": 3550598.24372702, "eta_seconds": 2138.143974304199, "data_wait_seconds": 2.053380012512207e-05} {"step": 25000, "total_steps": 28610, "tokens": 52428800000, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7871672213077545, "ce_loss": 2.771022617816925, "logit_z_loss": 29.58932876586914, "router_aux_loss": 1.4048753678798676, "router_z_loss": 2.095873713493347, "routing_entropy_normalized": 0.6422596382324314, "learning_rate": 0.0028178323656991703, "grad_norm": 0.7584623098373413, "tokens_per_second": 4422802.929360951, "eta_seconds": 1711.7467906475067, "data_wait_seconds": 1.9371509552001953e-05} {"step": 25010, "total_steps": 28610, "tokens": 52449771520, "source": "npset2", "phase": 5, "loss": 1.708485022187233, "ce_loss": 1.692937433719635, "logit_z_loss": 39.1844482421875, "router_aux_loss": 1.409173995256424, "router_z_loss": 1.4558478146791458, "routing_entropy_normalized": 0.6871387434794914, "learning_rate": 0.002812550382549892, "grad_norm": 0.8052269816398621, "tokens_per_second": 4085337.991359574, "eta_seconds": 1848.010425567627, "data_wait_seconds": 1.9371509552001953e-05} {"step": 25020, "total_steps": 28610, "tokens": 52470743040, "source": "fineweb_edu", "phase": 5, "loss": 3.3983956575393677, "ce_loss": 3.383608788251877, "logit_z_loss": 26.715064525604248, "router_aux_loss": 1.3072440326213837, "router_z_loss": 1.7144211381673813, "routing_entropy_normalized": 0.6688984517575578, "learning_rate": 0.0028071980113069804, "grad_norm": 1.0722720623016357, "tokens_per_second": 4110881.8847642634, "eta_seconds": 1831.425930261612, "data_wait_seconds": 2.1189451217651367e-05} {"step": 25030, "total_steps": 28610, "tokens": 52491714560, "source": "dclm", "phase": 5, "loss": 3.5535958111286163, "ce_loss": 3.538615643978119, "logit_z_loss": 26.424000024795532, "router_aux_loss": 1.3440885096788406, "router_z_loss": 1.5393126457929611, "routing_entropy_normalized": 0.65130950541593, "learning_rate": 0.002801775539001863, "grad_norm": 2.117039442062378, "tokens_per_second": 1805900.4076183718, "eta_seconds": 4157.374420166016, "data_wait_seconds": 2.300739288330078e-05} {"step": 25040, "total_steps": 28610, "tokens": 52512686080, "source": "finemath", "phase": 5, "loss": 2.2845689952373505, "ce_loss": 2.269564390182495, "logit_z_loss": 37.30146646499634, "router_aux_loss": 1.3889265358448029, "router_z_loss": 1.1152868568897247, "routing_entropy_normalized": 0.6622521376793228, "learning_rate": 0.0027962832564252726, "grad_norm": 0.7763398885726929, "tokens_per_second": 4019721.3882735143, "eta_seconds": 1862.5252640247345, "data_wait_seconds": 1.9818544387817383e-05} {"step": 25050, "total_steps": 28610, "tokens": 52533657600, "source": "dclm", "phase": 5, "loss": 3.6154098212718964, "ce_loss": 3.600278854370117, "logit_z_loss": 26.138861179351807, "router_aux_loss": 1.361756592988968, "router_z_loss": 1.51339191198349, "routing_entropy_normalized": 0.6868161666067603, "learning_rate": 0.0027907214581116553, "grad_norm": 1.051509976387024, "tokens_per_second": 3746341.0727259074, "eta_seconds": 1992.8407411575317, "data_wait_seconds": 1.9222497940063477e-05} {"step": 25060, "total_steps": 28610, "tokens": 52554629120, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7877750992774963, "ce_loss": 2.7723718881607056, "logit_z_loss": 28.810985326766968, "router_aux_loss": 1.3471691757440567, "router_z_loss": 1.9316065609455109, "routing_entropy_normalized": 0.6444998017414828, "learning_rate": 0.0027850904423233755, "grad_norm": 1.602890968322754, "tokens_per_second": 2243837.629455001, "eta_seconds": 3317.92706489563, "data_wait_seconds": 2.1755695343017578e-05} {"step": 25070, "total_steps": 28610, "tokens": 52575600640, "source": "finemath", "phase": 5, "loss": 2.1427734792232513, "ce_loss": 2.128224268555641, "logit_z_loss": 36.072333335876465, "router_aux_loss": 1.3459956645965576, "router_z_loss": 1.089232675731182, "routing_entropy_normalized": 0.6768786043403802, "learning_rate": 0.0027793905110347216, "grad_norm": 0.8379095792770386, "tokens_per_second": 4161073.1851557223, "eta_seconds": 1784.135426044464, "data_wait_seconds": 1.9222497940063477e-05} {"step": 25080, "total_steps": 28610, "tokens": 52596572160, "source": "cosmopedia_v2", "phase": 5, "loss": 2.8119556605815887, "ce_loss": 2.7969267070293427, "logit_z_loss": 27.985591650009155, "router_aux_loss": 1.3190661817789078, "router_z_loss": 1.838284432888031, "routing_entropy_normalized": 0.6949300562247909, "learning_rate": 0.002773621969915709, "grad_norm": 2.330796241760254, "tokens_per_second": 4450680.931400887, "eta_seconds": 1663.3289768695831, "data_wait_seconds": 2.4080276489257812e-05} {"step": 25090, "total_steps": 28610, "tokens": 52617543680, "source": "fineweb_edu", "phase": 5, "loss": 3.3971062302589417, "ce_loss": 3.3823819756507874, "logit_z_loss": 27.47469449043274, "router_aux_loss": 1.3170614540576935, "router_z_loss": 1.5536092519760132, "routing_entropy_normalized": 0.6762341437246779, "learning_rate": 0.0027677851283156923, "grad_norm": 0.6143147349357605, "tokens_per_second": 3397526.970369251, "eta_seconds": 2172.7495040893555, "data_wait_seconds": 0.06516000628471375} {"step": 25100, "total_steps": 28610, "tokens": 52638515200, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7693468034267426, "ce_loss": 2.754154086112976, "logit_z_loss": 28.620776414871216, "router_aux_loss": 1.3295729160308838, "router_z_loss": 1.8969860225915909, "routing_entropy_normalized": 0.6933080396408625, "learning_rate": 0.002761880299246772, "grad_norm": 1.5749640464782715, "tokens_per_second": 4194489.808230965, "eta_seconds": 1754.922256708145, "data_wait_seconds": 1.8656253814697266e-05} {"step": 25110, "total_steps": 28610, "tokens": 52659486720, "source": "npset2", "phase": 5, "loss": 1.693627879023552, "ce_loss": 1.6786655932664871, "logit_z_loss": 38.790162563323975, "router_aux_loss": 1.3623770773410797, "router_z_loss": 1.3385267555713654, "routing_entropy_normalized": 0.6735682677602879, "learning_rate": 0.0027559077993670107, "grad_norm": 0.7716400027275085, "tokens_per_second": 4261285.867784782, "eta_seconds": 1722.492277622223, "data_wait_seconds": 1.7762184143066406e-05} {"step": 25120, "total_steps": 28610, "tokens": 52680458240, "source": "fineweb_edu", "phase": 5, "loss": 3.398603767156601, "ce_loss": 3.383536070585251, "logit_z_loss": 27.20119285583496, "router_aux_loss": 1.3565069139003754, "router_z_loss": 1.502629578113556, "routing_entropy_normalized": 0.6837787110312133, "learning_rate": 0.0027498679489634517, "grad_norm": 1.3228191137313843, "tokens_per_second": 4284562.795925305, "eta_seconds": 1708.2397501468658, "data_wait_seconds": 2.2470951080322266e-05} {"step": 25130, "total_steps": 28610, "tokens": 52701429760, "source": "dclm", "phase": 5, "loss": 3.5381885766983032, "ce_loss": 3.523246616125107, "logit_z_loss": 27.57565450668335, "router_aux_loss": 1.3632423281669617, "router_z_loss": 1.3095421642065048, "routing_entropy_normalized": 0.6330276114433248, "learning_rate": 0.0027437610719349423, "grad_norm": 1.0155398845672607, "tokens_per_second": 4133742.6532773543, "eta_seconds": 1765.4918489456177, "data_wait_seconds": 2.199411392211914e-05} {"step": 25140, "total_steps": 28610, "tokens": 52722401280, "source": "finemath", "phase": 5, "loss": 2.248254895210266, "ce_loss": 2.2336699068546295, "logit_z_loss": 37.34777021408081, "router_aux_loss": 1.3772660046815872, "router_z_loss": 0.8123932927846909, "routing_entropy_normalized": 0.6518813907818549, "learning_rate": 0.0027375874957747642, "grad_norm": 0.6211755275726318, "tokens_per_second": 4153770.7311900305, "eta_seconds": 1751.930453300476, "data_wait_seconds": 2.09808349609375e-05} {"step": 25150, "total_steps": 28610, "tokens": 52743372800, "source": "dclm", "phase": 5, "loss": 3.499699056148529, "ce_loss": 3.4848520159721375, "logit_z_loss": 27.5921413898468, "router_aux_loss": 1.3572326451539993, "router_z_loss": 1.2746872156858444, "routing_entropy_normalized": 0.6574087326189572, "learning_rate": 0.00273134755155307, "grad_norm": 0.6495372653007507, "tokens_per_second": 3921107.6763331913, "eta_seconds": 1850.5347261428833, "data_wait_seconds": 1.913309097290039e-05} {"step": 25160, "total_steps": 28610, "tokens": 52764344320, "source": "cosmopedia_v2", "phase": 5, "loss": 2.82019180059433, "ce_loss": 2.804923415184021, "logit_z_loss": 29.92516827583313, "router_aux_loss": 1.357543021440506, "router_z_loss": 1.6929609328508377, "routing_entropy_normalized": 0.6615232712762815, "learning_rate": 0.0027250415738991327, "grad_norm": 3.0187277793884277, "tokens_per_second": 4004994.9089877266, "eta_seconds": 1806.5377271175385, "data_wait_seconds": 2.3245811462402344e-05} {"step": 25170, "total_steps": 28610, "tokens": 52785315840, "source": "finemath", "phase": 5, "loss": 2.103215992450714, "ce_loss": 2.087850347161293, "logit_z_loss": 37.49242115020752, "router_aux_loss": 1.4559334367513657, "router_z_loss": 0.8063292447477579, "routing_entropy_normalized": 0.6561745636957858, "learning_rate": 0.002718669900983395, "grad_norm": 0.7105793952941895, "tokens_per_second": 4203420.170702165, "eta_seconds": 1716.2697486877444, "data_wait_seconds": 2.1725893020629883e-05} {"step": 25180, "total_steps": 28610, "tokens": 52806287360, "source": "cosmopedia_v2", "phase": 5, "loss": 2.787769168615341, "ce_loss": 2.7722370624542236, "logit_z_loss": 30.079288721084595, "router_aux_loss": 1.3843248933553696, "router_z_loss": 1.6888869106769562, "routing_entropy_normalized": 0.6619970609062081, "learning_rate": 0.0027122328744993386, "grad_norm": 2.1432747840881348, "tokens_per_second": 4059465.3802512605, "eta_seconds": 1771.9651939868927, "data_wait_seconds": 2.181529998779297e-05} {"step": 25190, "total_steps": 28610, "tokens": 52827258880, "source": "fineweb_edu", "phase": 5, "loss": 3.3693259358406067, "ce_loss": 3.3547585010528564, "logit_z_loss": 25.61424708366394, "router_aux_loss": 1.2996105700731277, "router_z_loss": 1.5713613629341125, "routing_entropy_normalized": 0.6843089716175279, "learning_rate": 0.002705730839645159, "grad_norm": 1.7643628120422363, "tokens_per_second": 3787575.8329415973, "eta_seconds": 1893.628050327301, "data_wait_seconds": 2.1129846572875977e-05} {"step": 25200, "total_steps": 28610, "tokens": 52848230400, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7301005721092224, "ce_loss": 2.7147869169712067, "logit_z_loss": 29.19722032546997, "router_aux_loss": 1.342912882566452, "router_z_loss": 1.8845628052949905, "routing_entropy_normalized": 0.7066395317674571, "learning_rate": 0.002699164145105252, "grad_norm": 1.6861896514892578, "tokens_per_second": 4160424.6871806164, "eta_seconds": 1718.8842144012451, "data_wait_seconds": 2.4884939193725586e-05} {"step": 25210, "total_steps": 28610, "tokens": 52869201920, "source": "npset2", "phase": 5, "loss": 1.6846705079078674, "ce_loss": 1.6694038361310959, "logit_z_loss": 40.13781404495239, "router_aux_loss": 1.3927267789840698, "router_z_loss": 1.3393831253051758, "routing_entropy_normalized": 0.6739220931721066, "learning_rate": 0.002692533143031517, "grad_norm": 1.0103806257247925, "tokens_per_second": 3791631.0510470304, "eta_seconds": 1880.5407762527466, "data_wait_seconds": 1.913309097290039e-05} {"step": 25220, "total_steps": 28610, "tokens": 52890173440, "source": "fineweb_edu", "phase": 5, "loss": 3.351033478975296, "ce_loss": 3.335944026708603, "logit_z_loss": 26.502721548080444, "router_aux_loss": 1.349890723824501, "router_z_loss": 1.5905920267105103, "routing_entropy_normalized": 0.6719504293981479, "learning_rate": 0.002685838189024472, "grad_norm": 1.3418363332748413, "tokens_per_second": 3231548.4784670603, "eta_seconds": 2199.9810082912445, "data_wait_seconds": 2.6434659957885742e-05} {"step": 25230, "total_steps": 28610, "tokens": 52911144960, "source": "dclm", "phase": 5, "loss": 3.525043100118637, "ce_loss": 3.509969860315323, "logit_z_loss": 26.132131576538086, "router_aux_loss": 1.3639872819185257, "router_z_loss": 1.4333565682172775, "routing_entropy_normalized": 0.6681943601545147, "learning_rate": 0.0026790796421141813, "grad_norm": 1.983863115310669, "tokens_per_second": 4237125.566633924, "eta_seconds": 1672.9203910827634, "data_wait_seconds": 2.0712614059448242e-05} {"step": 25240, "total_steps": 28610, "tokens": 52932116480, "source": "finemath", "phase": 5, "loss": 2.208289623260498, "ce_loss": 2.1934804916381836, "logit_z_loss": 36.60391330718994, "router_aux_loss": 1.3717182129621506, "router_z_loss": 1.0919479876756668, "routing_entropy_normalized": 0.6713237130858484, "learning_rate": 0.002672257864741005, "grad_norm": 0.6205295324325562, "tokens_per_second": 2829685.0712876217, "eta_seconds": 2497.593216896057, "data_wait_seconds": 0.10640126466751099} {"step": 25250, "total_steps": 28610, "tokens": 52953088000, "source": "dclm", "phase": 5, "loss": 3.56228044629097, "ce_loss": 3.5474173426628113, "logit_z_loss": 27.431010484695435, "router_aux_loss": 1.3453766852617264, "router_z_loss": 1.4093731790781021, "routing_entropy_normalized": 0.6417525014997566, "learning_rate": 0.0026653732227361606, "grad_norm": 1.0049058198928833, "tokens_per_second": 3906585.0275751976, "eta_seconds": 1803.7315635681152, "data_wait_seconds": 2.1576881408691406e-05} {"step": 25260, "total_steps": 28610, "tokens": 52974059520, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7993814945220947, "ce_loss": 2.784783959388733, "logit_z_loss": 29.40835690498352, "router_aux_loss": 1.279227003455162, "router_z_loss": 1.80523881316185, "routing_entropy_normalized": 0.6958976110052502, "learning_rate": 0.0026584260853021054, "grad_norm": 1.293667197227478, "tokens_per_second": 3494177.83103787, "eta_seconds": 2010.6186747550964, "data_wait_seconds": 1.9818544387817383e-05} {"step": 25270, "total_steps": 28610, "tokens": 52995031040, "source": "finemath", "phase": 5, "loss": 2.090590387582779, "ce_loss": 2.0759472250938416, "logit_z_loss": 36.869401931762695, "router_aux_loss": 1.375777617096901, "router_z_loss": 0.8854054883122444, "routing_entropy_normalized": 0.6389591895171499, "learning_rate": 0.0026514168249927353, "grad_norm": 0.8993214964866638, "tokens_per_second": 4002836.824602606, "eta_seconds": 1749.8808937072754, "data_wait_seconds": 2.0384788513183594e-05} {"step": 25280, "total_steps": 28610, "tokens": 53016002560, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7720179855823517, "ce_loss": 2.756813108921051, "logit_z_loss": 28.462616443634033, "router_aux_loss": 1.3404427915811539, "router_z_loss": 1.8004536181688309, "routing_entropy_normalized": 0.6409388988765918, "learning_rate": 0.00264434581769341, "grad_norm": 1.9070055484771729, "tokens_per_second": 4347320.669337964, "eta_seconds": 1606.3954539299011, "data_wait_seconds": 2.2619962692260742e-05} {"step": 25290, "total_steps": 28610, "tokens": 53036974080, "source": "fineweb_edu", "phase": 5, "loss": 3.310662806034088, "ce_loss": 3.295917570590973, "logit_z_loss": 26.343491554260254, "router_aux_loss": 1.3303101509809494, "router_z_loss": 1.442156359553337, "routing_entropy_normalized": 0.6676497895962759, "learning_rate": 0.0026372134426007913, "grad_norm": 1.2625784873962402, "tokens_per_second": 3673004.2812045054, "eta_seconds": 1895.599380493164, "data_wait_seconds": 2.142786979675293e-05} {"step": 25300, "total_steps": 28610, "tokens": 53057945600, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7318274676799774, "ce_loss": 2.716511905193329, "logit_z_loss": 29.404651641845703, "router_aux_loss": 1.3625806868076324, "router_z_loss": 1.689685970544815, "routing_entropy_normalized": 0.6606636373283765, "learning_rate": 0.002630020082202509, "grad_norm": 0.8571756482124329, "tokens_per_second": 4157480.5509898537, "eta_seconds": 1669.6585912704468, "data_wait_seconds": 2.1457672119140625e-05} {"step": 25310, "total_steps": 28610, "tokens": 53078917120, "source": "npset2", "phase": 5, "loss": 1.6873814910650253, "ce_loss": 1.6731079667806625, "logit_z_loss": 38.29109764099121, "router_aux_loss": 1.295792669057846, "router_z_loss": 1.315605267882347, "routing_entropy_normalized": 0.6553415106171786, "learning_rate": 0.0026227661222566518, "grad_norm": 1.0880476236343384, "tokens_per_second": 3761201.2344990526, "eta_seconds": 1839.99769449234, "data_wait_seconds": 1.8894672393798828e-05} {"step": 25320, "total_steps": 28610, "tokens": 53099888640, "source": "fineweb_edu", "phase": 5, "loss": 3.3384682834148407, "ce_loss": 3.323695659637451, "logit_z_loss": 27.19393277168274, "router_aux_loss": 1.3401357233524323, "router_z_loss": 1.3712276369333267, "routing_entropy_normalized": 0.6467679839026361, "learning_rate": 0.002615451951771076, "grad_norm": 1.5111918449401855, "tokens_per_second": 4004538.712479756, "eta_seconds": 1722.952523469925, "data_wait_seconds": 2.187490463256836e-05} {"step": 25330, "total_steps": 28610, "tokens": 53120860160, "source": "dclm", "phase": 5, "loss": 3.552585929632187, "ce_loss": 3.538030296564102, "logit_z_loss": 27.223315954208374, "router_aux_loss": 1.3261836618185043, "router_z_loss": 1.2938723713159561, "routing_entropy_normalized": 0.6539223586034555, "learning_rate": 0.0026080779629825473, "grad_norm": 0.940220057964325, "tokens_per_second": 3968233.6341904737, "eta_seconds": 1733.4308395385742, "data_wait_seconds": 2.1189451217651367e-05} {"step": 25340, "total_steps": 28610, "tokens": 53141831680, "source": "finemath", "phase": 5, "loss": 2.1788623332977295, "ce_loss": 2.1648880392313004, "logit_z_loss": 38.10174512863159, "router_aux_loss": 1.3097256571054459, "router_z_loss": 0.8770410716533661, "routing_entropy_normalized": 0.6809868017718468, "learning_rate": 0.002600644551335706, "grad_norm": 0.5429512858390808, "tokens_per_second": 4056279.041658412, "eta_seconds": 1690.634931564331, "data_wait_seconds": 1.8149614334106445e-05} {"step": 25350, "total_steps": 28610, "tokens": 53162803200, "source": "dclm", "phase": 5, "loss": 3.5216795802116394, "ce_loss": 3.507697343826294, "logit_z_loss": 27.488884449005127, "router_aux_loss": 1.2816496938467026, "router_z_loss": 1.1656529009342194, "routing_entropy_normalized": 0.6990248785330172, "learning_rate": 0.0025931521154618597, "grad_norm": 0.7463535666465759, "tokens_per_second": 4249449.426365661, "eta_seconds": 1608.8473668098447, "data_wait_seconds": 2.148747444152832e-05} {"step": 25360, "total_steps": 28610, "tokens": 53183774720, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7797670364379883, "ce_loss": 2.7651351392269135, "logit_z_loss": 28.223234176635742, "router_aux_loss": 1.3067841082811356, "router_z_loss": 1.5640472620725632, "routing_entropy_normalized": 0.6552253143983496, "learning_rate": 0.0025856010571576053, "grad_norm": 2.51440691947937, "tokens_per_second": 4036437.2357892734, "eta_seconds": 1688.554435968399, "data_wait_seconds": 1.817941665649414e-05} {"step": 25370, "total_steps": 28610, "tokens": 53204746240, "source": "finemath", "phase": 5, "loss": 2.1824538111686707, "ce_loss": 2.1684602797031403, "logit_z_loss": 38.943758964538574, "router_aux_loss": 1.3265967667102814, "router_z_loss": 0.7275631502270699, "routing_entropy_normalized": 0.6816338693200125, "learning_rate": 0.002577991781363284, "grad_norm": 1.0271470546722412, "tokens_per_second": 3930532.4707916626, "eta_seconds": 1728.715518951416, "data_wait_seconds": 2.2411346435546875e-05} {"step": 25380, "total_steps": 28610, "tokens": 53225717760, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7522873282432556, "ce_loss": 2.7371649742126465, "logit_z_loss": 27.84756875038147, "router_aux_loss": 1.3715815991163254, "router_z_loss": 1.406525045633316, "routing_entropy_normalized": 0.6521486892018464, "learning_rate": 0.0025703246961412633, "grad_norm": 2.0057272911071777, "tokens_per_second": 4050890.4644575883, "eta_seconds": 1672.1757893562317, "data_wait_seconds": 2.2798776626586914e-05} {"step": 25390, "total_steps": 28610, "tokens": 53246689280, "source": "fineweb_edu", "phase": 5, "loss": 3.3637775480747223, "ce_loss": 3.3478724360466003, "logit_z_loss": 26.69928765296936, "router_aux_loss": 1.4611710160970688, "router_z_loss": 1.2934320122003555, "routing_entropy_normalized": 0.6368527578068762, "learning_rate": 0.0025626002126540553, "grad_norm": 1.138561487197876, "tokens_per_second": 3934875.8197495216, "eta_seconds": 1716.1480436325073, "data_wait_seconds": 2.16066837310791e-05} {"step": 25400, "total_steps": 28610, "tokens": 53267660800, "source": "cosmopedia_v2", "phase": 5, "loss": 2.732994705438614, "ce_loss": 2.7171497642993927, "logit_z_loss": 29.084606409072876, "router_aux_loss": 1.4088572412729263, "router_z_loss": 1.7563736587762833, "routing_entropy_normalized": 0.626337499070319, "learning_rate": 0.0025548187451422664, "grad_norm": 1.3252815008163452, "tokens_per_second": 4026191.491170943, "eta_seconds": 1672.0163297653198, "data_wait_seconds": 2.1845102310180664e-05} {"step": 25410, "total_steps": 28610, "tokens": 53288632320, "source": "npset2", "phase": 5, "loss": 1.6797505021095276, "ce_loss": 1.6648505628108978, "logit_z_loss": 38.08691930770874, "router_aux_loss": 1.3570696264505386, "router_z_loss": 1.3292425274848938, "routing_entropy_normalized": 0.6769509464464861, "learning_rate": 0.0025469807109023846, "grad_norm": 0.8921635150909424, "tokens_per_second": 3220018.5153528755, "eta_seconds": 2084.114227294922, "data_wait_seconds": 2.321600914001465e-05} {"step": 25420, "total_steps": 28610, "tokens": 53309603840, "source": "fineweb_edu", "phase": 5, "loss": 3.320894867181778, "ce_loss": 3.3054049909114838, "logit_z_loss": 26.078455209732056, "router_aux_loss": 1.4034186154603958, "router_z_loss": 1.4556805938482285, "routing_entropy_normalized": 0.6604383139992903, "learning_rate": 0.002539086530264399, "grad_norm": 0.9184109568595886, "tokens_per_second": 3909616.725252368, "eta_seconds": 1711.1434061527252, "data_wait_seconds": 2.059340476989746e-05} {"step": 25430, "total_steps": 28610, "tokens": 53330575360, "source": "dclm", "phase": 5, "loss": 3.5179924964904785, "ce_loss": 3.502551019191742, "logit_z_loss": 25.343262910842896, "router_aux_loss": 1.4092516154050827, "router_z_loss": 1.3489612489938736, "routing_entropy_normalized": 0.6506705562801277, "learning_rate": 0.0025311366265692594, "grad_norm": 2.6139402389526367, "tokens_per_second": 4027605.298583837, "eta_seconds": 1655.8085675239563, "data_wait_seconds": 2.428889274597168e-05} {"step": 25440, "total_steps": 28610, "tokens": 53351546880, "source": "finemath", "phase": 5, "loss": 2.202452376484871, "ce_loss": 2.187470108270645, "logit_z_loss": 36.203299045562744, "router_aux_loss": 1.3930273652076721, "router_z_loss": 1.0520185008645058, "routing_entropy_normalized": 0.6793295038286259, "learning_rate": 0.002523131426146173, "grad_norm": 0.8724309206008911, "tokens_per_second": 3688291.5936052175, "eta_seconds": 1802.4528894424438, "data_wait_seconds": 1.9252300262451172e-05} {"step": 25450, "total_steps": 28610, "tokens": 53372518400, "source": "dclm", "phase": 5, "loss": 3.538516014814377, "ce_loss": 3.524089127779007, "logit_z_loss": 27.22050404548645, "router_aux_loss": 1.3072419315576553, "router_z_loss": 1.354496791958809, "routing_entropy_normalized": 0.684854901065002, "learning_rate": 0.002515071358289746, "grad_norm": 1.145222783088684, "tokens_per_second": 4352582.912070838, "eta_seconds": 1522.544303894043, "data_wait_seconds": 2.065300941467285e-05} {"step": 25460, "total_steps": 28610, "tokens": 53393489920, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7928949892520905, "ce_loss": 2.778052508831024, "logit_z_loss": 29.092738389968872, "router_aux_loss": 1.310045138001442, "router_z_loss": 1.7420215904712677, "routing_entropy_normalized": 0.6654742892058363, "learning_rate": 0.0025069568552369553, "grad_norm": 1.5919679403305054, "tokens_per_second": 4119521.733178177, "eta_seconds": 1603.5912001132965, "data_wait_seconds": 2.1278858184814453e-05} {"step": 25470, "total_steps": 28610, "tokens": 53414461440, "source": "finemath", "phase": 5, "loss": 2.1507303714752197, "ce_loss": 2.1369615495204926, "logit_z_loss": 37.83052062988281, "router_aux_loss": 1.2790105491876602, "router_z_loss": 0.978723868727684, "routing_entropy_normalized": 0.6973372545328785, "learning_rate": 0.0024987883521439738, "grad_norm": 0.921811580657959, "tokens_per_second": 4147464.190169501, "eta_seconds": 1587.7309551239014, "data_wait_seconds": 1.9937753677368164e-05} {"step": 25480, "total_steps": 28610, "tokens": 53435432960, "source": "cosmopedia_v2", "phase": 5, "loss": 2.722630739212036, "ce_loss": 2.7083334028720856, "logit_z_loss": 28.857201099395752, "router_aux_loss": 1.2666752487421036, "router_z_loss": 1.6305458843708038, "routing_entropy_normalized": 0.7054113871952477, "learning_rate": 0.002490566287062832, "grad_norm": 1.5350263118743896, "tokens_per_second": 3512114.1390717807, "eta_seconds": 1868.9841787815094, "data_wait_seconds": 2.142786979675293e-05} {"step": 25490, "total_steps": 28610, "tokens": 53456404480, "source": "fineweb_edu", "phase": 5, "loss": 3.3327786028385162, "ce_loss": 3.3181115984916687, "logit_z_loss": 25.904728889465332, "router_aux_loss": 1.3246285766363144, "router_z_loss": 1.4207133054733276, "routing_entropy_normalized": 0.6690950846734274, "learning_rate": 0.002482291100917928, "grad_norm": 0.7612576484680176, "tokens_per_second": 3602078.1847487018, "eta_seconds": 1816.4831256866455, "data_wait_seconds": 1.919269561767578e-05} {"step": 25500, "total_steps": 28610, "tokens": 53477376000, "source": "cosmopedia_v2", "phase": 5, "loss": 2.6848245561122894, "ce_loss": 2.6702502071857452, "logit_z_loss": 30.209752798080444, "router_aux_loss": 1.295656993985176, "router_z_loss": 1.6177978515625, "routing_entropy_normalized": 0.6980113328461038, "learning_rate": 0.00247396323748238, "grad_norm": 1.393066167831421, "tokens_per_second": 4484768.250526009, "eta_seconds": 1454.2875697612762, "data_wait_seconds": 2.0682811737060547e-05} {"step": 25510, "total_steps": 28610, "tokens": 53498347520, "source": "npset2", "phase": 5, "loss": 1.671318918466568, "ce_loss": 1.6573866605758667, "logit_z_loss": 40.4354305267334, "router_aux_loss": 1.2702486217021942, "router_z_loss": 1.2297694385051727, "routing_entropy_normalized": 0.6866219692311428, "learning_rate": 0.0024655831433542308, "grad_norm": 0.7365453839302063, "tokens_per_second": 3829770.763565751, "eta_seconds": 1697.5353360176086, "data_wait_seconds": 1.9341707229614258e-05} {"step": 25520, "total_steps": 28610, "tokens": 53519319040, "source": "fineweb_edu", "phase": 5, "loss": 3.3129625618457794, "ce_loss": 3.298517256975174, "logit_z_loss": 26.03664469718933, "router_aux_loss": 1.3242117911577225, "router_z_loss": 1.2032611668109894, "routing_entropy_normalized": 0.6656266746982066, "learning_rate": 0.0024571512679324964, "grad_norm": 1.745070219039917, "tokens_per_second": 4034110.2623166675, "eta_seconds": 1606.3516509532928, "data_wait_seconds": 1.990795135498047e-05} {"step": 25530, "total_steps": 28610, "tokens": 53540290560, "source": "dclm", "phase": 5, "loss": 3.4745545387268066, "ce_loss": 3.460487484931946, "logit_z_loss": 26.05134892463684, "router_aux_loss": 1.2930733859539032, "router_z_loss": 1.1363486498594284, "routing_entropy_normalized": 0.6662270575114886, "learning_rate": 0.0024486680633930657, "grad_norm": 0.7075299620628357, "tokens_per_second": 3748118.7792562484, "eta_seconds": 1723.325364112854, "data_wait_seconds": 2.0444393157958984e-05} {"step": 25540, "total_steps": 28610, "tokens": 53561262080, "source": "finemath", "phase": 5, "loss": 2.2184917330741882, "ce_loss": 2.204355388879776, "logit_z_loss": 37.62766170501709, "router_aux_loss": 1.3301365971565247, "router_z_loss": 0.834985226392746, "routing_entropy_normalized": 0.6568894391315413, "learning_rate": 0.002440133984664454, "grad_norm": 0.9474599361419678, "tokens_per_second": 3975825.025686198, "eta_seconds": 1619.3511028289795, "data_wait_seconds": 2.0116567611694336e-05} {"step": 25550, "total_steps": 28610, "tokens": 53582233600, "source": "dclm", "phase": 5, "loss": 3.4830955266952515, "ce_loss": 3.4690816402435303, "logit_z_loss": 27.35711669921875, "router_aux_loss": 1.295675590634346, "router_z_loss": 1.0571145489811897, "routing_entropy_normalized": 0.6980572455627594, "learning_rate": 0.0024315494894034026, "grad_norm": 0.826697051525116, "tokens_per_second": 3832741.647625747, "eta_seconds": 1674.3328170776367, "data_wait_seconds": 2.2530555725097656e-05} {"step": 25560, "total_steps": 28610, "tokens": 53603205120, "source": "cosmopedia_v2", "phase": 5, "loss": 2.773884207010269, "ce_loss": 2.7599537670612335, "logit_z_loss": 28.113065719604492, "router_aux_loss": 1.2521854788064957, "router_z_loss": 1.4085761308670044, "routing_entropy_normalized": 0.6944865489194966, "learning_rate": 0.002422915037970341, "grad_norm": 2.3069252967834473, "tokens_per_second": 4071589.328830871, "eta_seconds": 1570.9623646736145, "data_wait_seconds": 2.1070241928100586e-05} {"step": 25570, "total_steps": 28610, "tokens": 53624176640, "source": "finemath", "phase": 5, "loss": 2.163017064332962, "ce_loss": 2.1495191752910614, "logit_z_loss": 38.02785587310791, "router_aux_loss": 1.2830159068107605, "router_z_loss": 0.667721651494503, "routing_entropy_normalized": 0.6637148881718037, "learning_rate": 0.0024142310934046948, "grad_norm": 0.6420451402664185, "tokens_per_second": 4094256.94477173, "eta_seconds": 1557.1426429748535, "data_wait_seconds": 2.7388334274291992e-05} {"step": 25580, "total_steps": 28610, "tokens": 53645148160, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7435481250286102, "ce_loss": 2.7295631766319275, "logit_z_loss": 28.76770830154419, "router_aux_loss": 1.2697821706533432, "router_z_loss": 1.287096843123436, "routing_entropy_normalized": 0.6755315666463251, "learning_rate": 0.0024054981214000565, "grad_norm": 3.0627689361572266, "tokens_per_second": 4128414.593485237, "eta_seconds": 1539.179366827011, "data_wait_seconds": 2.333521842956543e-05} {"step": 25590, "total_steps": 28610, "tokens": 53666119680, "source": "fineweb_edu", "phase": 5, "loss": 3.323653995990753, "ce_loss": 3.3100134432315826, "logit_z_loss": 25.836061000823975, "router_aux_loss": 1.2558077275753021, "router_z_loss": 1.0824673399329185, "routing_entropy_normalized": 0.7000248288980417, "learning_rate": 0.0023967165902792115, "grad_norm": 1.3908624649047852, "tokens_per_second": 4123441.008981656, "eta_seconds": 1535.949956893921, "data_wait_seconds": 2.2143125534057617e-05} {"step": 25600, "total_steps": 28610, "tokens": 53687091200, "source": "cosmopedia_v2", "phase": 5, "loss": 2.720751076936722, "ce_loss": 2.7064366340637207, "logit_z_loss": 29.388633012771606, "router_aux_loss": 1.3023765832185745, "router_z_loss": 1.2906652092933655, "routing_entropy_normalized": 0.6483247583021687, "learning_rate": 0.0023878869709690236, "grad_norm": 1.320088267326355, "tokens_per_second": 4043988.2618036335, "eta_seconds": 1560.9411084651947, "data_wait_seconds": 2.130866050720215e-05} {"step": 25610, "total_steps": 28610, "tokens": 53708062720, "source": "npset2", "phase": 5, "loss": 1.727022185921669, "ce_loss": 1.7131003439426422, "logit_z_loss": 39.69418144226074, "router_aux_loss": 1.2988687306642532, "router_z_loss": 0.9331306740641594, "routing_entropy_normalized": 0.6580043548562496, "learning_rate": 0.0023790097369751793, "grad_norm": 1.728166103363037, "tokens_per_second": 3942657.540115197, "eta_seconds": 1595.7399129867554, "data_wait_seconds": 1.963973045349121e-05} {"step": 25620, "total_steps": 28610, "tokens": 53729034240, "source": "fineweb_edu", "phase": 5, "loss": 3.3267923295497894, "ce_loss": 3.3122439086437225, "logit_z_loss": 26.16395115852356, "router_aux_loss": 1.3501272052526474, "router_z_loss": 1.047180987894535, "routing_entropy_normalized": 0.6613708237250224, "learning_rate": 0.0023700853643567974, "grad_norm": 2.2141799926757812, "tokens_per_second": 4360634.231681576, "eta_seconds": 1437.975337266922, "data_wait_seconds": 2.5033950805664062e-05} {"step": 25630, "total_steps": 28610, "tokens": 53750005760, "source": "dclm", "phase": 5, "loss": 3.4603441059589386, "ce_loss": 3.44607812166214, "logit_z_loss": 27.09868621826172, "router_aux_loss": 1.3295767158269882, "router_z_loss": 0.9701847285032272, "routing_entropy_normalized": 0.6794685300996126, "learning_rate": 0.0023611143317008973, "grad_norm": 1.2307510375976562, "tokens_per_second": 4079544.063460264, "eta_seconds": 1531.914562702179, "data_wait_seconds": 2.1964311599731445e-05} {"step": 25640, "total_steps": 28610, "tokens": 53770977280, "source": "finemath", "phase": 5, "loss": 2.1998050212860107, "ce_loss": 2.185472220182419, "logit_z_loss": 38.63244295120239, "router_aux_loss": 1.3710661381483078, "router_z_loss": 0.622151754796505, "routing_entropy_normalized": 0.6513671047112367, "learning_rate": 0.002352097120096734, "grad_norm": 1.4366987943649292, "tokens_per_second": 3914071.9107472887, "eta_seconds": 1591.32013463974, "data_wait_seconds": 2.0205974578857422e-05} {"step": 25650, "total_steps": 28610, "tokens": 53791948800, "source": "dclm", "phase": 5, "loss": 3.487068295478821, "ce_loss": 3.4732086062431335, "logit_z_loss": 27.225541830062866, "router_aux_loss": 1.2841626107692719, "router_z_loss": 1.0180554613471031, "routing_entropy_normalized": 0.6758284024733626, "learning_rate": 0.00234303421311, "grad_norm": 1.2867951393127441, "tokens_per_second": 3836152.248367908, "eta_seconds": 1618.1761093139648, "data_wait_seconds": 2.1696090698242188e-05} {"step": 25660, "total_steps": 28610, "tokens": 53812920320, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7652100324630737, "ce_loss": 2.7504743933677673, "logit_z_loss": 28.608336925506592, "router_aux_loss": 1.3358088433742523, "router_z_loss": 1.37753064930439, "routing_entropy_normalized": 0.6413367677663996, "learning_rate": 0.002333926096756894, "grad_norm": 1.62530517578125, "tokens_per_second": 3953770.777847179, "eta_seconds": 1564.7337055206299, "data_wait_seconds": 2.2143125534057617e-05} {"step": 25670, "total_steps": 28610, "tokens": 53833891840, "source": "finemath", "phase": 5, "loss": 2.1005900502204895, "ce_loss": 2.086014360189438, "logit_z_loss": 38.64689493179321, "router_aux_loss": 1.3796122670173645, "router_z_loss": 0.7795533016324043, "routing_entropy_normalized": 0.6559593696766777, "learning_rate": 0.0023247732594780524, "grad_norm": 0.7104725241661072, "tokens_per_second": 3992237.0157372626, "eta_seconds": 1544.4040160179138, "data_wait_seconds": 0.01077231764793396} {"step": 25680, "total_steps": 28610, "tokens": 53854863360, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7541673481464386, "ce_loss": 2.7395622432231903, "logit_z_loss": 28.00188684463501, "router_aux_loss": 1.315971463918686, "router_z_loss": 1.445401817560196, "routing_entropy_normalized": 0.6626221924619469, "learning_rate": 0.0023155761921123627, "grad_norm": 1.066759705543518, "tokens_per_second": 4301815.542686435, "eta_seconds": 1428.386526346207, "data_wait_seconds": 2.187490463256836e-05} {"step": 25690, "total_steps": 28610, "tokens": 53875834880, "source": "fineweb_edu", "phase": 5, "loss": 3.3453439474105835, "ce_loss": 3.3309375047683716, "logit_z_loss": 27.17905306816101, "router_aux_loss": 1.3151848763227463, "router_z_loss": 1.254583090543747, "routing_entropy_normalized": 0.6683543909966672, "learning_rate": 0.0023063353878706346, "grad_norm": 0.8698229193687439, "tokens_per_second": 4007039.4055577083, "eta_seconds": 1528.231499671936, "data_wait_seconds": 2.0891427993774414e-05} {"step": 25700, "total_steps": 28610, "tokens": 53896806400, "source": "cosmopedia_v2", "phase": 5, "loss": 2.727089375257492, "ce_loss": 2.712315946817398, "logit_z_loss": 28.57848572731018, "router_aux_loss": 1.3360205441713333, "router_z_loss": 1.4132690578699112, "routing_entropy_normalized": 0.6634368819249882, "learning_rate": 0.002297051342309156, "grad_norm": 1.1437056064605713, "tokens_per_second": 4300085.40546502, "eta_seconds": 1419.207235336304, "data_wait_seconds": 2.1755695343017578e-05} {"step": 25710, "total_steps": 28610, "tokens": 53917777920, "source": "npset2", "phase": 5, "loss": 1.6424102038145065, "ce_loss": 1.6281488984823227, "logit_z_loss": 38.699307441711426, "router_aux_loss": 1.3195127099752426, "router_z_loss": 1.0661750435829163, "routing_entropy_normalized": 0.6776265064707326, "learning_rate": 0.002287724553303115, "grad_norm": 0.646554708480835, "tokens_per_second": 3798855.00983088, "eta_seconds": 1600.940489768982, "data_wait_seconds": 2.014636993408203e-05} {"step": 25720, "total_steps": 28610, "tokens": 53938749440, "source": "fineweb_edu", "phase": 5, "loss": 3.348728269338608, "ce_loss": 3.3339782059192657, "logit_z_loss": 26.723318576812744, "router_aux_loss": 1.3571389019489288, "router_z_loss": 1.1786872297525406, "routing_entropy_normalized": 0.6353201016549493, "learning_rate": 0.0022783555210198997, "grad_norm": 1.5259721279144287, "tokens_per_second": 4223596.557165626, "eta_seconds": 1434.9782698154447, "data_wait_seconds": 2.1576881408691406e-05} {"step": 25730, "total_steps": 28610, "tokens": 53959720960, "source": "dclm", "phase": 5, "loss": 3.4430271685123444, "ce_loss": 3.428624749183655, "logit_z_loss": 26.96585774421692, "router_aux_loss": 1.3252802938222885, "router_z_loss": 1.1495687887072563, "routing_entropy_normalized": 0.679701042322764, "learning_rate": 0.0022689447478922787, "grad_norm": 1.504278302192688, "tokens_per_second": 4047479.53525524, "eta_seconds": 1492.2367630004883, "data_wait_seconds": 2.4050474166870117e-05} {"step": 25740, "total_steps": 28610, "tokens": 53980692480, "source": "finemath", "phase": 5, "loss": 2.145254999399185, "ce_loss": 2.1312469244003296, "logit_z_loss": 37.91988134384155, "router_aux_loss": 1.3242466747760773, "router_z_loss": 0.765634223818779, "routing_entropy_normalized": 0.6634808151395941, "learning_rate": 0.0022594927385914546, "grad_norm": 0.7539567947387695, "tokens_per_second": 4191158.1612331406, "eta_seconds": 1436.077095746994, "data_wait_seconds": 1.862645149230957e-05} {"step": 25750, "total_steps": 28610, "tokens": 54001664000, "source": "dclm", "phase": 5, "loss": 3.437925308942795, "ce_loss": 3.4241873919963837, "logit_z_loss": 27.01425337791443, "router_aux_loss": 1.2681298851966858, "router_z_loss": 1.0566099360585213, "routing_entropy_normalized": 0.6653725841425453, "learning_rate": 0.0022500000000000003, "grad_norm": 1.0961472988128662, "tokens_per_second": 4831128.164261014, "eta_seconds": 1241.5018844604492, "data_wait_seconds": 2.1845102310180664e-05} {"step": 25760, "total_steps": 28610, "tokens": 54022635520, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7485590279102325, "ce_loss": 2.734391689300537, "logit_z_loss": 29.736672401428223, "router_aux_loss": 1.2688362747430801, "router_z_loss": 1.478891745209694, "routing_entropy_normalized": 0.6653022110906072, "learning_rate": 0.002240467041184678, "grad_norm": 2.6019325256347656, "tokens_per_second": 3943132.6235536565, "eta_seconds": 1515.7702696323395, "data_wait_seconds": 2.2590160369873047e-05} {"step": 25770, "total_steps": 28610, "tokens": 54043607040, "source": "finemath", "phase": 5, "loss": 2.1840769350528717, "ce_loss": 2.1702998876571655, "logit_z_loss": 37.916826248168945, "router_aux_loss": 1.3057445585727692, "router_z_loss": 0.7196232080459595, "routing_entropy_normalized": 0.6694484320126064, "learning_rate": 0.0022308943733691374, "grad_norm": 1.1013741493225098, "tokens_per_second": 4369933.201343139, "eta_seconds": 1362.9296846389773, "data_wait_seconds": 2.014636993408203e-05} {"step": 25780, "total_steps": 28610, "tokens": 54064578560, "source": "cosmopedia_v2", "phase": 5, "loss": 2.8300544917583466, "ce_loss": 2.8155908584594727, "logit_z_loss": 29.791221618652344, "router_aux_loss": 1.310156211256981, "router_z_loss": 1.3621305227279663, "routing_entropy_normalized": 0.6439032273788493, "learning_rate": 0.0022212825099065026, "grad_norm": 2.492316722869873, "tokens_per_second": 4054839.427893838, "eta_seconds": 1463.668356180191, "data_wait_seconds": 2.014636993408203e-05} {"step": 25790, "total_steps": 28610, "tokens": 54085550080, "source": "fineweb_edu", "phase": 5, "loss": 3.3346900045871735, "ce_loss": 3.320858418941498, "logit_z_loss": 25.42898154258728, "router_aux_loss": 1.2727618962526321, "router_z_loss": 1.10395947098732, "routing_entropy_normalized": 0.6706018135332208, "learning_rate": 0.0022116319662518406, "grad_norm": 0.971365213394165, "tokens_per_second": 3315508.463129569, "eta_seconds": 1783.7290134429932, "data_wait_seconds": 2.5004148483276367e-05} {"step": 25800, "total_steps": 28610, "tokens": 54106521600, "source": "cosmopedia_v2", "phase": 5, "loss": 2.734070748090744, "ce_loss": 2.719593822956085, "logit_z_loss": 27.401791095733643, "router_aux_loss": 1.30313640832901, "router_z_loss": 1.4455430060625076, "routing_entropy_normalized": 0.6492149484035258, "learning_rate": 0.0022019432599345193, "grad_norm": 2.1258373260498047, "tokens_per_second": 4340589.048718846, "eta_seconds": 1357.6491701602936, "data_wait_seconds": 2.148747444152832e-05} {"step": 25810, "total_steps": 28610, "tokens": 54127493120, "source": "npset2", "phase": 5, "loss": 1.8197833001613617, "ce_loss": 1.8050647228956223, "logit_z_loss": 39.77107238769531, "router_aux_loss": 1.3758041560649872, "router_z_loss": 0.9605499058961868, "routing_entropy_normalized": 0.6618950605449739, "learning_rate": 0.0021922169105304547, "grad_norm": 1.22360098361969, "tokens_per_second": 3877136.544643944, "eta_seconds": 1514.5263862609863, "data_wait_seconds": 1.8835067749023438e-05} {"step": 25820, "total_steps": 28610, "tokens": 54148464640, "source": "fineweb_edu", "phase": 5, "loss": 3.504297435283661, "ce_loss": 3.48861762881279, "logit_z_loss": 25.476467847824097, "router_aux_loss": 1.4474161565303802, "router_z_loss": 1.2056574821472168, "routing_entropy_normalized": 0.6411350857740012, "learning_rate": 0.0021824534396342474, "grad_norm": 1.4966472387313843, "tokens_per_second": 3955336.3885426684, "eta_seconds": 1479.2810282707214, "data_wait_seconds": 2.3126602172851562e-05} {"step": 25830, "total_steps": 28610, "tokens": 54169436160, "source": "dclm", "phase": 5, "loss": 3.528036803007126, "ce_loss": 3.513565242290497, "logit_z_loss": 27.427224159240723, "router_aux_loss": 1.3292432129383087, "router_z_loss": 1.179163709282875, "routing_entropy_normalized": 0.6964031274011226, "learning_rate": 0.0021726533708312117, "grad_norm": 1.2708625793457031, "tokens_per_second": 4298466.936512667, "eta_seconds": 1356.3167161941528, "data_wait_seconds": 3.8057565689086914e-05} {"step": 25840, "total_steps": 28610, "tokens": 54190407680, "source": "finemath", "phase": 5, "loss": 2.2267187535762787, "ce_loss": 2.2128047347068787, "logit_z_loss": 38.17594766616821, "router_aux_loss": 1.307694599032402, "router_z_loss": 0.837070882320404, "routing_entropy_normalized": 0.6717056708472264, "learning_rate": 0.0021628172296692953, "grad_norm": 0.7341142892837524, "tokens_per_second": 3838028.121857937, "eta_seconds": 1513.566564798355, "data_wait_seconds": 1.9073486328125e-05} {"step": 25850, "total_steps": 28610, "tokens": 54211379200, "source": "dclm", "phase": 5, "loss": 3.5443740487098694, "ce_loss": 3.530749201774597, "logit_z_loss": 27.080830097198486, "router_aux_loss": 1.2468194663524628, "router_z_loss": 1.1566576808691025, "routing_entropy_normalized": 0.6858046742532881, "learning_rate": 0.0021529455436308975, "grad_norm": 0.7905598282814026, "tokens_per_second": 3658337.5993557135, "eta_seconds": 1582.1775226593018, "data_wait_seconds": 2.1189451217651367e-05} {"step": 25860, "total_steps": 28610, "tokens": 54232350720, "source": "cosmopedia_v2", "phase": 5, "loss": 2.8410918712615967, "ce_loss": 2.8273361921310425, "logit_z_loss": 28.94509243965149, "router_aux_loss": 1.2290131449699402, "router_z_loss": 1.4654806703329086, "routing_entropy_normalized": 0.693701591967548, "learning_rate": 0.0021430388421045813, "grad_norm": 1.0406668186187744, "tokens_per_second": 4199945.377512892, "eta_seconds": 1373.1530964374542, "data_wait_seconds": 2.205371856689453e-05} {"step": 25870, "total_steps": 28610, "tokens": 54253322240, "source": "finemath", "phase": 5, "loss": 2.223411485552788, "ce_loss": 2.209539532661438, "logit_z_loss": 38.78622579574585, "router_aux_loss": 1.3187766522169113, "router_z_loss": 0.6841153800487518, "routing_entropy_normalized": 0.6621431164945275, "learning_rate": 0.0021330976563566847, "grad_norm": 0.7831584215164185, "tokens_per_second": 4234349.930279344, "eta_seconds": 1357.0433654785159, "data_wait_seconds": 2.288818359375e-05} {"step": 25880, "total_steps": 28610, "tokens": 54274293760, "source": "cosmopedia_v2", "phase": 5, "loss": 2.81650710105896, "ce_loss": 2.802109658718109, "logit_z_loss": 29.209941625595093, "router_aux_loss": 1.3002807945013046, "router_z_loss": 1.3945920020341873, "routing_entropy_normalized": 0.6744864243838281, "learning_rate": 0.00212312251950283, "grad_norm": 1.969360113143921, "tokens_per_second": 3620085.592227352, "eta_seconds": 1581.5164625644684, "data_wait_seconds": 2.1964311599731445e-05} {"step": 25890, "total_steps": 28610, "tokens": 54295265280, "source": "fineweb_edu", "phase": 5, "loss": 3.430787742137909, "ce_loss": 3.4169659316539764, "logit_z_loss": 25.841191053390503, "router_aux_loss": 1.2736163586378098, "router_z_loss": 1.085673376917839, "routing_entropy_normalized": 0.688264056385114, "learning_rate": 0.002113113966479332, "grad_norm": 1.3308063745498657, "tokens_per_second": 4524369.160242552, "eta_seconds": 1260.7842636108398, "data_wait_seconds": 2.205371856689453e-05} {"step": 25900, "total_steps": 28610, "tokens": 54316236800, "source": "cosmopedia_v2", "phase": 5, "loss": 2.79316508769989, "ce_loss": 2.7787385880947113, "logit_z_loss": 28.538331985473633, "router_aux_loss": 1.3067277520895004, "router_z_loss": 1.3592262268066406, "routing_entropy_normalized": 0.6611341151259248, "learning_rate": 0.0021030725340145167, "grad_norm": 1.4404361248016357, "tokens_per_second": 3780365.747037241, "eta_seconds": 1503.3682718276978, "data_wait_seconds": 2.3663043975830078e-05} {"step": 25910, "total_steps": 28610, "tokens": 54337208320, "source": "npset2", "phase": 5, "loss": 1.6880659759044647, "ce_loss": 1.6740740239620209, "logit_z_loss": 40.23106527328491, "router_aux_loss": 1.3110373765230179, "router_z_loss": 0.8815775513648987, "routing_entropy_normalized": 0.7053053359413364, "learning_rate": 0.002092998760599932, "grad_norm": 0.9207278490066528, "tokens_per_second": 4007772.619269045, "eta_seconds": 1412.832248210907, "data_wait_seconds": 2.3424625396728516e-05} {"step": 25920, "total_steps": 28610, "tokens": 54358179840, "source": "fineweb_edu", "phase": 5, "loss": 3.390005439519882, "ce_loss": 3.375745564699173, "logit_z_loss": 26.137176513671875, "router_aux_loss": 1.321280151605606, "router_z_loss": 1.04704250395298, "routing_entropy_normalized": 0.6700883479738695, "learning_rate": 0.0020828931864614745, "grad_norm": 1.7576320171356201, "tokens_per_second": 4112025.9134402457, "eta_seconds": 1371.9122881889343, "data_wait_seconds": 2.1249055862426758e-05} {"step": 25930, "total_steps": 28610, "tokens": 54379151360, "source": "dclm", "phase": 5, "loss": 3.5004002451896667, "ce_loss": 3.4862572848796844, "logit_z_loss": 28.01809072494507, "router_aux_loss": 1.319250226020813, "router_z_loss": 0.950521282851696, "routing_entropy_normalized": 0.6682377198490823, "learning_rate": 0.002072756353530416, "grad_norm": 1.9349849224090576, "tokens_per_second": 4110479.6178060765, "eta_seconds": 1367.3264150619507, "data_wait_seconds": 2.053380012512207e-05} {"step": 25940, "total_steps": 28610, "tokens": 54400122880, "source": "finemath", "phase": 5, "loss": 2.2629907429218292, "ce_loss": 2.2494917511940002, "logit_z_loss": 37.40887403488159, "router_aux_loss": 1.2887169569730759, "router_z_loss": 0.6118207722902298, "routing_entropy_normalized": 0.722277415370357, "learning_rate": 0.002062588805414343, "grad_norm": 1.420856237411499, "tokens_per_second": 4159363.318769884, "eta_seconds": 1346.2146513462067, "data_wait_seconds": 1.9818544387817383e-05} {"step": 25950, "total_steps": 28610, "tokens": 54421094400, "source": "dclm", "phase": 5, "loss": 3.5104024410247803, "ce_loss": 3.496286630630493, "logit_z_loss": 26.958385944366455, "router_aux_loss": 1.3245084434747696, "router_z_loss": 0.8706828132271767, "routing_entropy_normalized": 0.6693902322065078, "learning_rate": 0.0020523910873680032, "grad_norm": 0.9793920516967773, "tokens_per_second": 3919889.7372026043, "eta_seconds": 1423.107458114624, "data_wait_seconds": 2.0444393157958984e-05} {"step": 25960, "total_steps": 28610, "tokens": 54442065920, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7751840949058533, "ce_loss": 2.7610473036766052, "logit_z_loss": 29.253642082214355, "router_aux_loss": 1.286216378211975, "router_z_loss": 1.274604007601738, "routing_entropy_normalized": 0.6708768749275054, "learning_rate": 0.002042163746264069, "grad_norm": 1.8901809453964233, "tokens_per_second": 4099006.3706385256, "eta_seconds": 1355.8048701286316, "data_wait_seconds": 2.0265579223632812e-05} {"step": 25970, "total_steps": 28610, "tokens": 54463037440, "source": "finemath", "phase": 5, "loss": 2.162233352661133, "ce_loss": 2.1489478945732117, "logit_z_loss": 37.67682933807373, "router_aux_loss": 1.2643873989582062, "router_z_loss": 0.6415451541543007, "routing_entropy_normalized": 0.7143326268191778, "learning_rate": 0.0020319073305638032, "grad_norm": 0.709261417388916, "tokens_per_second": 4073672.7797238464, "eta_seconds": 1359.0883655548096, "data_wait_seconds": 2.333521842956543e-05} {"step": 25980, "total_steps": 28610, "tokens": 54484008960, "source": "cosmopedia_v2", "phase": 5, "loss": 2.8107571601867676, "ce_loss": 2.7968991100788116, "logit_z_loss": 28.571171760559082, "router_aux_loss": 1.2547723948955536, "router_z_loss": 1.3103891909122467, "routing_entropy_normalized": 0.7070732677880366, "learning_rate": 0.002021622390287655, "grad_norm": 1.416067123413086, "tokens_per_second": 3603632.2648548298, "eta_seconds": 1530.5417852401733, "data_wait_seconds": 2.187490463256836e-05} {"step": 25990, "total_steps": 28610, "tokens": 54504980480, "source": "fineweb_edu", "phase": 5, "loss": 3.404345601797104, "ce_loss": 3.3909407258033752, "logit_z_loss": 25.008622407913208, "router_aux_loss": 1.243950977921486, "router_z_loss": 0.965381845831871, "routing_entropy_normalized": 0.744221272654482, "learning_rate": 0.002011309476985757, "grad_norm": 0.7725488543510437, "tokens_per_second": 4449618.254111895, "eta_seconds": 1234.8336253166199, "data_wait_seconds": 2.047419548034668e-05} {"step": 26000, "total_steps": 28610, "tokens": 54525952000, "source": "cosmopedia_v2", "phase": 5, "loss": 2.781717211008072, "ce_loss": 2.767430454492569, "logit_z_loss": 28.612836837768555, "router_aux_loss": 1.3013202995061874, "router_z_loss": 1.2735242247581482, "routing_entropy_normalized": 0.6572453196404654, "learning_rate": 0.002000969143708352, "grad_norm": 1.1834584474563599, "tokens_per_second": 4081789.084646668, "eta_seconds": 1340.9724526405334, "data_wait_seconds": 2.3066997528076172e-05} {"step": 26010, "total_steps": 28610, "tokens": 54546923520, "source": "npset2", "phase": 5, "loss": 1.6519658863544464, "ce_loss": 1.637620434165001, "logit_z_loss": 42.512879848480225, "router_aux_loss": 1.3507146537303925, "router_z_loss": 0.8383155986666679, "routing_entropy_normalized": 0.6347186974712778, "learning_rate": 0.0019906019449761328, "grad_norm": 1.428704857826233, "tokens_per_second": 3611768.0399799934, "eta_seconds": 1509.674801826477, "data_wait_seconds": 1.952052116394043e-05} {"step": 26020, "total_steps": 28610, "tokens": 54567895040, "source": "fineweb_edu", "phase": 5, "loss": 3.3892944753170013, "ce_loss": 3.375792473554611, "logit_z_loss": 25.411301374435425, "router_aux_loss": 1.2472090125083923, "router_z_loss": 1.0299023687839508, "routing_entropy_normalized": 0.7196089139069599, "learning_rate": 0.001980208436750503, "grad_norm": 1.2018969058990479, "tokens_per_second": 4363135.536578398, "eta_seconds": 1244.8899729251862, "data_wait_seconds": 2.0951032638549805e-05} {"step": 26030, "total_steps": 28610, "tokens": 54588866560, "source": "dclm", "phase": 5, "loss": 3.506024330854416, "ce_loss": 3.492424339056015, "logit_z_loss": 26.44134020805359, "router_aux_loss": 1.2736558616161346, "router_z_loss": 0.8634069859981537, "routing_entropy_normalized": 0.6848383937958716, "learning_rate": 0.0019697891764037688, "grad_norm": 1.3879752159118652, "tokens_per_second": 3732132.391799324, "eta_seconds": 1449.7481846809387, "data_wait_seconds": 2.0444393157958984e-05} {"step": 26040, "total_steps": 28610, "tokens": 54609838080, "source": "finemath", "phase": 5, "loss": 2.168168291449547, "ce_loss": 2.1545741856098175, "logit_z_loss": 37.61395740509033, "router_aux_loss": 1.304630532860756, "router_z_loss": 0.5478238798677921, "routing_entropy_normalized": 0.665708293026316, "learning_rate": 0.0019593447226892388, "grad_norm": 1.275975227355957, "tokens_per_second": 3837324.391569772, "eta_seconds": 1404.5413131713867, "data_wait_seconds": 1.8477439880371094e-05} {"step": 26050, "total_steps": 28610, "tokens": 54630809600, "source": "dclm", "phase": 5, "loss": 3.457745850086212, "ce_loss": 3.443114012479782, "logit_z_loss": 26.89117741584778, "router_aux_loss": 1.3758675456047058, "router_z_loss": 0.8731248378753662, "routing_entropy_normalized": 0.646580125131854, "learning_rate": 0.0019488756357112706, "grad_norm": 1.6617681980133057, "tokens_per_second": 4153141.1750923325, "eta_seconds": 1292.6864013671875, "data_wait_seconds": 2.0802021026611328e-05} {"step": 26060, "total_steps": 28610, "tokens": 54651781120, "source": "cosmopedia_v2", "phase": 5, "loss": 2.787946105003357, "ce_loss": 2.7737470269203186, "logit_z_loss": 29.607560396194458, "router_aux_loss": 1.2866849303245544, "router_z_loss": 1.3322384804487228, "routing_entropy_normalized": 0.6520809782356745, "learning_rate": 0.0019383824768952258, "grad_norm": 1.8996223211288452, "tokens_per_second": 4166413.107810057, "eta_seconds": 1283.535132408142, "data_wait_seconds": 2.2739171981811523e-05} {"step": 26070, "total_steps": 28610, "tokens": 54672752640, "source": "finemath", "phase": 5, "loss": 2.1098688542842865, "ce_loss": 2.0960132479667664, "logit_z_loss": 38.9520320892334, "router_aux_loss": 1.3161663264036179, "router_z_loss": 0.6939048245549202, "routing_entropy_normalized": 0.6873071000005557, "learning_rate": 0.001927865808957367, "grad_norm": 0.7987180352210999, "tokens_per_second": 3803706.85484121, "eta_seconds": 1400.4144597053528, "data_wait_seconds": 2.110004425048828e-05} {"step": 26080, "total_steps": 28610, "tokens": 54693724160, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7807356119155884, "ce_loss": 2.766353964805603, "logit_z_loss": 27.44449234008789, "router_aux_loss": 1.3086096048355103, "router_z_loss": 1.2955404669046402, "routing_entropy_normalized": 0.6590036879816016, "learning_rate": 0.0019173261958746791, "grad_norm": 2.0005147457122803, "tokens_per_second": 4401773.492720286, "eta_seconds": 1205.3765530586243, "data_wait_seconds": 2.4259090423583984e-05} {"step": 26090, "total_steps": 28610, "tokens": 54714695680, "source": "fineweb_edu", "phase": 5, "loss": 3.3678519427776337, "ce_loss": 3.3540929555892944, "logit_z_loss": 26.046154975891113, "router_aux_loss": 1.270874261856079, "router_z_loss": 1.0502297803759575, "routing_entropy_normalized": 0.7115094872587506, "learning_rate": 0.0019067642028546257, "grad_norm": 2.4822492599487305, "tokens_per_second": 3857671.4209206994, "eta_seconds": 1369.9515752792358, "data_wait_seconds": 2.0891427993774414e-05} {"step": 26100, "total_steps": 28610, "tokens": 54735667200, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7527546286582947, "ce_loss": 2.738636016845703, "logit_z_loss": 28.522263288497925, "router_aux_loss": 1.278403103351593, "router_z_loss": 1.3346151560544968, "routing_entropy_normalized": 0.6875299995875657, "learning_rate": 0.0018961803963048385, "grad_norm": 2.3101966381073, "tokens_per_second": 4074300.1746464856, "eta_seconds": 1291.964581489563, "data_wait_seconds": 2.256035804748535e-05} {"step": 26110, "total_steps": 28610, "tokens": 54756638720, "source": "npset2", "phase": 5, "loss": 1.6328068971633911, "ce_loss": 1.6188293099403381, "logit_z_loss": 39.44460153579712, "router_aux_loss": 1.30079884827137, "router_z_loss": 0.9696046561002731, "routing_entropy_normalized": 0.684760822055497, "learning_rate": 0.001885575343802742, "grad_norm": 1.0473271608352661, "tokens_per_second": 4059916.18836058, "eta_seconds": 1291.3764119148254, "data_wait_seconds": 2.0802021026611328e-05} {"step": 26120, "total_steps": 28610, "tokens": 54777610240, "source": "fineweb_edu", "phase": 5, "loss": 3.377428710460663, "ce_loss": 3.3630684912204742, "logit_z_loss": 24.94064426422119, "router_aux_loss": 1.335924968123436, "router_z_loss": 1.0009734630584717, "routing_entropy_normalized": 0.6849507517353013, "learning_rate": 0.0018749496140651187, "grad_norm": 1.4867390394210815, "tokens_per_second": 4378199.04030454, "eta_seconds": 1192.7069628238678, "data_wait_seconds": 2.1249055862426758e-05} {"step": 26130, "total_steps": 28610, "tokens": 54798581760, "source": "dclm", "phase": 5, "loss": 3.4885096848011017, "ce_loss": 3.474127620458603, "logit_z_loss": 27.371538400650024, "router_aux_loss": 1.337928131222725, "router_z_loss": 1.0027774274349213, "routing_entropy_normalized": 0.6798222125863413, "learning_rate": 0.0018643037769176057, "grad_norm": 0.8780933022499084, "tokens_per_second": 3974610.757827259, "eta_seconds": 1308.5399494171143, "data_wait_seconds": 1.8328428268432617e-05} {"step": 26140, "total_steps": 28610, "tokens": 54819553280, "source": "finemath", "phase": 5, "loss": 2.15644034743309, "ce_loss": 2.142666831612587, "logit_z_loss": 37.79618501663208, "router_aux_loss": 1.3017539978027344, "router_z_loss": 0.7559518814086914, "routing_entropy_normalized": 0.6994846508973913, "learning_rate": 0.0018536384032641412, "grad_norm": 1.2836058139801025, "tokens_per_second": 3824837.439478436, "eta_seconds": 1354.2968876361847, "data_wait_seconds": 1.9788742065429688e-05} {"step": 26150, "total_steps": 28610, "tokens": 54840524800, "source": "dclm", "phase": 5, "loss": 3.478436440229416, "ce_loss": 3.464638113975525, "logit_z_loss": 27.48058581352234, "router_aux_loss": 1.27597476541996, "router_z_loss": 1.0385882705450058, "routing_entropy_normalized": 0.7055452296114824, "learning_rate": 0.001842954065056346, "grad_norm": 0.7155717015266418, "tokens_per_second": 3531837.8493246213, "eta_seconds": 1460.710865020752, "data_wait_seconds": 2.142786979675293e-05} {"step": 26160, "total_steps": 28610, "tokens": 54861496320, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7506287693977356, "ce_loss": 2.7369565069675446, "logit_z_loss": 28.49308133125305, "router_aux_loss": 1.22204028069973, "router_z_loss": 1.451834723353386, "routing_entropy_normalized": 0.7053769285876061, "learning_rate": 0.001832251335262854, "grad_norm": 1.0610437393188477, "tokens_per_second": 4203582.480196395, "eta_seconds": 1222.29608297348, "data_wait_seconds": 2.1845102310180664e-05} {"step": 26170, "total_steps": 28610, "tokens": 54882467840, "source": "finemath", "phase": 5, "loss": 2.1848758459091187, "ce_loss": 2.171934798359871, "logit_z_loss": 39.344857692718506, "router_aux_loss": 1.2263746857643127, "router_z_loss": 0.6772928535938263, "routing_entropy_normalized": 0.7259154993515397, "learning_rate": 0.0018215307878385822, "grad_norm": 0.7551535367965698, "tokens_per_second": 4021866.092391474, "eta_seconds": 1272.3076210021973, "data_wait_seconds": 2.3692846298217773e-05} {"step": 26180, "total_steps": 28610, "tokens": 54903439360, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7926666736602783, "ce_loss": 2.7792481780052185, "logit_z_loss": 27.92425298690796, "router_aux_loss": 1.2099710702896118, "router_z_loss": 1.3188087791204453, "routing_entropy_normalized": 0.6737231583683816, "learning_rate": 0.0018107929976939538, "grad_norm": 1.581921100616455, "tokens_per_second": 3532512.7177578695, "eta_seconds": 1442.6216597557068, "data_wait_seconds": 2.0623207092285156e-05} {"step": 26190, "total_steps": 28610, "tokens": 54924410880, "source": "fineweb_edu", "phase": 5, "loss": 3.363036423921585, "ce_loss": 3.349186658859253, "logit_z_loss": 24.81829047203064, "router_aux_loss": 1.2894174456596375, "router_z_loss": 0.9556061252951622, "routing_entropy_normalized": 0.6929881830378326, "learning_rate": 0.0018000385406640668, "grad_norm": 1.812733769416809, "tokens_per_second": 3903853.5865379013, "eta_seconds": 1300.0251488685608, "data_wait_seconds": 2.1755695343017578e-05} {"step": 26200, "total_steps": 28610, "tokens": 54945382400, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7257577776908875, "ce_loss": 2.711452603340149, "logit_z_loss": 27.898787260055542, "router_aux_loss": 1.3075838088989258, "router_z_loss": 1.229298323392868, "routing_entropy_normalized": 0.6810070251284308, "learning_rate": 0.0017892679934778135, "grad_norm": 1.248279094696045, "tokens_per_second": 4162897.739095273, "eta_seconds": 1214.090913772583, "data_wait_seconds": 2.0503997802734375e-05} {"step": 26210, "total_steps": 28610, "tokens": 54966353920, "source": "npset2", "phase": 5, "loss": 1.6216282695531845, "ce_loss": 1.6077632308006287, "logit_z_loss": 39.528297901153564, "router_aux_loss": 1.2935359925031662, "router_z_loss": 0.9296544268727303, "routing_entropy_normalized": 0.658593791115024, "learning_rate": 0.0017784819337269531, "grad_norm": 0.6109414100646973, "tokens_per_second": 4004528.503035574, "eta_seconds": 1256.8682670593262, "data_wait_seconds": 2.1219253540039062e-05} {"step": 26220, "total_steps": 28610, "tokens": 54987325440, "source": "fineweb_edu", "phase": 5, "loss": 3.3573235273361206, "ce_loss": 3.3439034819602966, "logit_z_loss": 26.42137861251831, "router_aux_loss": 1.2421479374170303, "router_z_loss": 0.9985187500715256, "routing_entropy_normalized": 0.6948062296698289, "learning_rate": 0.0017676809398351357, "grad_norm": 1.7824323177337646, "tokens_per_second": 3892630.655526497, "eta_seconds": 1287.610801935196, "data_wait_seconds": 2.142786979675293e-05} {"step": 26230, "total_steps": 28610, "tokens": 55008296960, "source": "dclm", "phase": 5, "loss": 3.466436445713043, "ce_loss": 3.453154593706131, "logit_z_loss": 27.432345390319824, "router_aux_loss": 1.2406220138072968, "router_z_loss": 0.8756087943911552, "routing_entropy_normalized": 0.6913140241791635, "learning_rate": 0.0017568655910268842, "grad_norm": 0.9331544041633606, "tokens_per_second": 4251077.823482771, "eta_seconds": 1174.1073598861694, "data_wait_seconds": 2.250075340270996e-05} {"step": 26240, "total_steps": 28610, "tokens": 55029268480, "source": "finemath", "phase": 5, "loss": 2.2469785511493683, "ce_loss": 2.2335399091243744, "logit_z_loss": 37.261380672454834, "router_aux_loss": 1.2888060808181763, "router_z_loss": 0.5505532883107662, "routing_entropy_normalized": 0.6767800891259821, "learning_rate": 0.0017460364672965327, "grad_norm": 1.3549671173095703, "tokens_per_second": 3915788.2155288295, "eta_seconds": 1269.284743309021, "data_wait_seconds": 1.8537044525146484e-05} {"step": 26250, "total_steps": 28610, "tokens": 55050240000, "source": "dclm", "phase": 5, "loss": 3.478693872690201, "ce_loss": 3.4654670655727386, "logit_z_loss": 26.855010986328125, "router_aux_loss": 1.2348655313253403, "router_z_loss": 0.8781697973608971, "routing_entropy_normalized": 0.6829119870710819, "learning_rate": 0.0017351941493771227, "grad_norm": 0.8746938109397888, "tokens_per_second": 3942306.4266480603, "eta_seconds": 1255.4272003173828, "data_wait_seconds": 2.5779008865356445e-05} {"step": 26260, "total_steps": 28610, "tokens": 55071211520, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7460625767707825, "ce_loss": 2.7326146066188812, "logit_z_loss": 30.148548126220703, "router_aux_loss": 1.2181710451841354, "router_z_loss": 1.266198605298996, "routing_entropy_normalized": 0.665107372922723, "learning_rate": 0.0017243392187092574, "grad_norm": 1.8726959228515625, "tokens_per_second": 4024142.1371848304, "eta_seconds": 1224.6851706504822, "data_wait_seconds": 2.0951032638549805e-05} {"step": 26270, "total_steps": 28610, "tokens": 55092183040, "source": "finemath", "phase": 5, "loss": 2.1832275390625, "ce_loss": 2.1697809994220734, "logit_z_loss": 37.33620643615723, "router_aux_loss": 1.2894007414579391, "router_z_loss": 0.5525163859128952, "routing_entropy_normalized": 0.6606335980814779, "learning_rate": 0.001713472257409928, "grad_norm": 1.273890733718872, "tokens_per_second": 3853187.158880567, "eta_seconds": 1273.5783333778381, "data_wait_seconds": 1.7434358596801758e-05} {"step": 26280, "total_steps": 28610, "tokens": 55113154560, "source": "cosmopedia_v2", "phase": 5, "loss": 2.721426248550415, "ce_loss": 2.7073023915290833, "logit_z_loss": 28.94194507598877, "router_aux_loss": 1.2944137156009674, "router_z_loss": 1.1797640323638916, "routing_entropy_normalized": 0.6592279633717132, "learning_rate": 0.0017025938482412876, "grad_norm": 2.5991950035095215, "tokens_per_second": 4058807.5225058175, "eta_seconds": 1203.891569852829, "data_wait_seconds": 2.0116567611694336e-05} {"step": 26290, "total_steps": 28610, "tokens": 55134126080, "source": "fineweb_edu", "phase": 5, "loss": 3.3675479888916016, "ce_loss": 3.3538811206817627, "logit_z_loss": 26.195564031600952, "router_aux_loss": 1.2781090587377548, "router_z_loss": 0.8858034536242485, "routing_entropy_normalized": 0.6824100882662674, "learning_rate": 0.001691704574579407, "grad_norm": 1.0934199094772339, "tokens_per_second": 4154895.385322321, "eta_seconds": 1171.00244140625, "data_wait_seconds": 2.1070241928100586e-05} {"step": 26300, "total_steps": 28610, "tokens": 55155097600, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7334379851818085, "ce_loss": 2.7196489572525024, "logit_z_loss": 28.985244750976562, "router_aux_loss": 1.2549463510513306, "router_z_loss": 1.239524558186531, "routing_entropy_normalized": 0.6475379760921293, "learning_rate": 0.0016808050203829844, "grad_norm": 0.9249595403671265, "tokens_per_second": 4451757.184118246, "eta_seconds": 1088.2042572498322, "data_wait_seconds": 2.142786979675293e-05} {"step": 26310, "total_steps": 28610, "tokens": 55176069120, "source": "npset2", "phase": 5, "loss": 1.6440232545137405, "ce_loss": 1.6303927153348923, "logit_z_loss": 40.78071737289429, "router_aux_loss": 1.2793835252523422, "router_z_loss": 0.8367004543542862, "routing_entropy_normalized": 0.6605155702591615, "learning_rate": 0.0016698957701620338, "grad_norm": 1.1055511236190796, "tokens_per_second": 3862759.2413684316, "eta_seconds": 1248.7057304382324, "data_wait_seconds": 1.7851591110229492e-05} {"step": 26320, "total_steps": 28610, "tokens": 55197040640, "source": "fineweb_edu", "phase": 5, "loss": 3.3578014075756073, "ce_loss": 3.3436821699142456, "logit_z_loss": 25.724494457244873, "router_aux_loss": 1.3206904083490372, "router_z_loss": 0.9123143777251244, "routing_entropy_normalized": 0.64917282356277, "learning_rate": 0.0016589774089465366, "grad_norm": 1.2396092414855957, "tokens_per_second": 4402669.974978223, "eta_seconds": 1090.810373544693, "data_wait_seconds": 2.047419548034668e-05} {"step": 26330, "total_steps": 28610, "tokens": 55218012160, "source": "dclm", "phase": 5, "loss": 3.4442918598651886, "ce_loss": 3.4309054017066956, "logit_z_loss": 27.176717281341553, "router_aux_loss": 1.2568021565675735, "router_z_loss": 0.8184682130813599, "routing_entropy_normalized": 0.6548453507306833, "learning_rate": 0.0016480505222550683, "grad_norm": 1.4505504369735718, "tokens_per_second": 3360900.12413267, "eta_seconds": 1422.6862993240356, "data_wait_seconds": 2.0265579223632812e-05} {"step": 26340, "total_steps": 28610, "tokens": 55238983680, "source": "finemath", "phase": 5, "loss": 2.108493685722351, "ce_loss": 2.0950921922922134, "logit_z_loss": 38.696040630340576, "router_aux_loss": 1.279788613319397, "router_z_loss": 0.603610210120678, "routing_entropy_normalized": 0.6662653186438131, "learning_rate": 0.001637115696063402, "grad_norm": 0.8038490414619446, "tokens_per_second": 4149513.476203528, "eta_seconds": 1147.251374721527, "data_wait_seconds": 1.8417835235595703e-05} {"step": 26350, "total_steps": 28610, "tokens": 55259955200, "source": "dclm", "phase": 5, "loss": 3.4377028942108154, "ce_loss": 3.4240430295467377, "logit_z_loss": 26.814823627471924, "router_aux_loss": 1.2830093204975128, "router_z_loss": 0.8297396525740623, "routing_entropy_normalized": 0.6603606670144238, "learning_rate": 0.0016261735167730797, "grad_norm": 0.6339240670204163, "tokens_per_second": 4071646.6239750055, "eta_seconds": 1164.0409784317017, "data_wait_seconds": 2.3603439331054688e-05} {"step": 26360, "total_steps": 28610, "tokens": 55280926720, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7313534319400787, "ce_loss": 2.717469036579132, "logit_z_loss": 29.019818544387817, "router_aux_loss": 1.259063020348549, "router_z_loss": 1.2937944829463959, "routing_entropy_normalized": 0.6639655618114978, "learning_rate": 0.0016152245711799702, "grad_norm": 1.748723030090332, "tokens_per_second": 4237426.846487877, "eta_seconds": 1113.5512590408325, "data_wait_seconds": 2.4139881134033203e-05} {"step": 26370, "total_steps": 28610, "tokens": 55301898240, "source": "finemath", "phase": 5, "loss": 2.107446864247322, "ce_loss": 2.0946926921606064, "logit_z_loss": 37.93273067474365, "router_aux_loss": 1.2193006724119186, "router_z_loss": 0.5611627846956253, "routing_entropy_normalized": 0.7425565488580893, "learning_rate": 0.001604269446442797, "grad_norm": 0.7740422487258911, "tokens_per_second": 4112106.843994646, "eta_seconds": 1142.3877487182617, "data_wait_seconds": 2.1576881408691406e-05} {"step": 26380, "total_steps": 28610, "tokens": 55322869760, "source": "cosmopedia_v2", "phase": 5, "loss": 2.754505753517151, "ce_loss": 2.7405463457107544, "logit_z_loss": 28.26140332221985, "router_aux_loss": 1.2705006897449493, "router_z_loss": 1.2544752955436707, "routing_entropy_normalized": 0.6297631765697062, "learning_rate": 0.0015933087300516516, "grad_norm": 1.7061176300048828, "tokens_per_second": 4084901.2492569606, "eta_seconds": 1144.8621826171875, "data_wait_seconds": 2.0891427993774414e-05} {"step": 26390, "total_steps": 28610, "tokens": 55343841280, "source": "fineweb_edu", "phase": 5, "loss": 3.330838292837143, "ce_loss": 3.3177707493305206, "logit_z_loss": 25.544495344161987, "router_aux_loss": 1.2107177376747131, "router_z_loss": 0.9603558406233788, "routing_entropy_normalized": 0.7151360288581067, "learning_rate": 0.0015823430097964913, "grad_norm": 1.2212073802947998, "tokens_per_second": 4168918.379597048, "eta_seconds": 1116.7590765953064, "data_wait_seconds": 2.396106719970703e-05} {"step": 26400, "total_steps": 28610, "tokens": 55364812800, "source": "cosmopedia_v2", "phase": 5, "loss": 2.72541081905365, "ce_loss": 2.7117009460926056, "logit_z_loss": 27.37655258178711, "router_aux_loss": 1.252754956483841, "router_z_loss": 1.1823320239782333, "routing_entropy_normalized": 0.6840321833849592, "learning_rate": 0.0015713728737356137, "grad_norm": 1.42339026927948, "tokens_per_second": 3863818.199264521, "eta_seconds": 1199.5144908428192, "data_wait_seconds": 2.22623348236084e-05} {"step": 26410, "total_steps": 28610, "tokens": 55385784320, "source": "npset2", "phase": 5, "loss": 1.6003157049417496, "ce_loss": 1.5873009711503983, "logit_z_loss": 40.398706912994385, "router_aux_loss": 1.2269680052995682, "router_z_loss": 0.7450762465596199, "routing_entropy_normalized": 0.6995718119512073, "learning_rate": 0.0015603989101641228, "grad_norm": 0.8369653224945068, "tokens_per_second": 4156197.7780207503, "eta_seconds": 1110.0853824615479, "data_wait_seconds": 2.065300941467285e-05} {"step": 26420, "total_steps": 28610, "tokens": 55406755840, "source": "fineweb_edu", "phase": 5, "loss": 3.3638908863067627, "ce_loss": 3.350657105445862, "logit_z_loss": 26.36810278892517, "router_aux_loss": 1.2339442819356918, "router_z_loss": 0.894309401512146, "routing_entropy_normalized": 0.6729919887104259, "learning_rate": 0.0015494217075823812, "grad_norm": 0.8277478218078613, "tokens_per_second": 4099491.032881383, "eta_seconds": 1120.3251435756683, "data_wait_seconds": 2.047419548034668e-05} {"step": 26430, "total_steps": 28610, "tokens": 55427727360, "source": "dclm", "phase": 5, "loss": 3.49079030752182, "ce_loss": 3.477470487356186, "logit_z_loss": 25.80676555633545, "router_aux_loss": 1.2568572014570236, "router_z_loss": 0.7513342723250389, "routing_entropy_normalized": 0.647083324093781, "learning_rate": 0.001538441854664449, "grad_norm": 1.4281615018844604, "tokens_per_second": 3943557.962765872, "eta_seconds": 1159.3062415122986, "data_wait_seconds": 1.570582389831543e-05} {"step": 26440, "total_steps": 28610, "tokens": 55448698880, "source": "finemath", "phase": 5, "loss": 2.0860112011432648, "ce_loss": 2.073245346546173, "logit_z_loss": 38.14727163314819, "router_aux_loss": 1.221033737063408, "router_z_loss": 0.555486224591732, "routing_entropy_normalized": 0.7026053672418321, "learning_rate": 0.0015274599402265163, "grad_norm": 0.8255515694618225, "tokens_per_second": 3840367.371267082, "eta_seconds": 1184.9959652423859, "data_wait_seconds": 2.2113323211669922e-05} {"step": 26450, "total_steps": 28610, "tokens": 55469670400, "source": "dclm", "phase": 5, "loss": 3.4287246763706207, "ce_loss": 3.4155625700950623, "logit_z_loss": 27.18586540222168, "router_aux_loss": 1.2376267164945602, "router_z_loss": 0.7858434617519379, "routing_entropy_normalized": 0.6714082223266059, "learning_rate": 0.0015164765531953254, "grad_norm": 0.944913923740387, "tokens_per_second": 3981951.8547492037, "eta_seconds": 1137.59494972229, "data_wait_seconds": 1.990795135498047e-05} {"step": 26460, "total_steps": 28610, "tokens": 55490641920, "source": "cosmopedia_v2", "phase": 5, "loss": 2.719668447971344, "ce_loss": 2.7061092257499695, "logit_z_loss": 28.68957805633545, "router_aux_loss": 1.2434827983379364, "router_z_loss": 1.1243635639548302, "routing_entropy_normalized": 0.6802693563635495, "learning_rate": 0.0015054922825765908, "grad_norm": 2.6607542037963867, "tokens_per_second": 3813189.099929555, "eta_seconds": 1182.4424862861633, "data_wait_seconds": 1.862645149230957e-05} {"step": 26470, "total_steps": 28610, "tokens": 55511613440, "source": "finemath", "phase": 5, "loss": 2.1583929657936096, "ce_loss": 2.145143002271652, "logit_z_loss": 38.23094892501831, "router_aux_loss": 1.2763949930667877, "router_z_loss": 0.48604196310043335, "routing_entropy_normalized": 0.6719360059551769, "learning_rate": 0.0014945077174234097, "grad_norm": 0.8948416709899902, "tokens_per_second": 3711283.673375893, "eta_seconds": 1209.2595648765564, "data_wait_seconds": 1.9758939743041992e-05} {"step": 26480, "total_steps": 28610, "tokens": 55532584960, "source": "cosmopedia_v2", "phase": 5, "loss": 2.725513309240341, "ce_loss": 2.7123560905456543, "logit_z_loss": 29.653168201446533, "router_aux_loss": 1.2025906890630722, "router_z_loss": 1.131239652633667, "routing_entropy_normalized": 0.7047888831116288, "learning_rate": 0.0014835234468046745, "grad_norm": 1.266770362854004, "tokens_per_second": 4193180.5010244725, "eta_seconds": 1065.285350561142, "data_wait_seconds": 1.9341707229614258e-05} {"step": 26490, "total_steps": 28610, "tokens": 55553556480, "source": "fineweb_edu", "phase": 5, "loss": 3.345818907022476, "ce_loss": 3.33236888051033, "logit_z_loss": 25.958171844482422, "router_aux_loss": 1.2597253322601318, "router_z_loss": 0.8528016358613968, "routing_entropy_normalized": 0.6871290940896687, "learning_rate": 0.0014725400597734844, "grad_norm": 1.2805811166763306, "tokens_per_second": 4062345.0789906704, "eta_seconds": 1094.4324407577515, "data_wait_seconds": 1.96993350982666e-05} {"step": 26500, "total_steps": 28610, "tokens": 55574528000, "source": "cosmopedia_v2", "phase": 5, "loss": 2.69572913646698, "ce_loss": 2.6820767521858215, "logit_z_loss": 28.592874765396118, "router_aux_loss": 1.2423701137304306, "router_z_loss": 1.22872693836689, "routing_entropy_normalized": 0.6690208609734152, "learning_rate": 0.0014615581453355513, "grad_norm": 1.3565068244934082, "tokens_per_second": 4074203.3639650345, "eta_seconds": 1086.0996186733246, "data_wait_seconds": 2.0951032638549805e-05} {"step": 26510, "total_steps": 28610, "tokens": 55595499520, "source": "npset2", "phase": 5, "loss": 1.6138510555028915, "ce_loss": 1.6004473716020584, "logit_z_loss": 41.07929468154907, "router_aux_loss": 1.2567937076091766, "router_z_loss": 0.8357389494776726, "routing_entropy_normalized": 0.6693215766737655, "learning_rate": 0.001450578292417619, "grad_norm": 0.5583165287971497, "tokens_per_second": 4413152.290077707, "eta_seconds": 997.9304838180542, "data_wait_seconds": 2.2530555725097656e-05} {"step": 26520, "total_steps": 28610, "tokens": 55616471040, "source": "fineweb_edu", "phase": 5, "loss": 3.3158924281597137, "ce_loss": 3.3021465837955475, "logit_z_loss": 25.577357292175293, "router_aux_loss": 1.2832456231117249, "router_z_loss": 0.9133565947413445, "routing_entropy_normalized": 0.6473367725951029, "learning_rate": 0.0014396010898358777, "grad_norm": 0.769307017326355, "tokens_per_second": 4011916.223996652, "eta_seconds": 1092.5072796344757, "data_wait_seconds": 2.1636486053466797e-05} {"step": 26530, "total_steps": 28610, "tokens": 55637442560, "source": "dclm", "phase": 5, "loss": 3.3887805342674255, "ce_loss": 3.3757461607456207, "logit_z_loss": 27.58375120162964, "router_aux_loss": 1.2205445766448975, "router_z_loss": 0.8288892507553101, "routing_entropy_normalized": 0.7096485853074104, "learning_rate": 0.0014286271262643866, "grad_norm": 1.0057103633880615, "tokens_per_second": 3814425.819507668, "eta_seconds": 1143.5734672546387, "data_wait_seconds": 1.671910285949707e-05} {"step": 26540, "total_steps": 28610, "tokens": 55658414080, "source": "finemath", "phase": 5, "loss": 2.071742534637451, "ce_loss": 2.059154435992241, "logit_z_loss": 39.38363265991211, "router_aux_loss": 1.2026565968990326, "router_z_loss": 0.5614812523126602, "routing_entropy_normalized": 0.6705394682537945, "learning_rate": 0.0014176569902035088, "grad_norm": 0.9258538484573364, "tokens_per_second": 4110317.6958838194, "eta_seconds": 1056.1482009887695, "data_wait_seconds": 1.8864870071411133e-05} {"step": 26550, "total_steps": 28610, "tokens": 55679385600, "source": "dclm", "phase": 5, "loss": 3.461898624897003, "ce_loss": 3.4493768215179443, "logit_z_loss": 27.029990911483765, "router_aux_loss": 1.1768990010023117, "router_z_loss": 0.752819374203682, "routing_entropy_normalized": 0.6810650378239075, "learning_rate": 0.0014066912699483485, "grad_norm": 0.8725276589393616, "tokens_per_second": 4121094.1483930526, "eta_seconds": 1048.2976036071777, "data_wait_seconds": 2.5451183319091797e-05} {"step": 26560, "total_steps": 28610, "tokens": 55700357120, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7327428460121155, "ce_loss": 2.7200111746788025, "logit_z_loss": 29.411428689956665, "router_aux_loss": 1.1619348227977753, "router_z_loss": 1.1123671233654022, "routing_entropy_normalized": 0.6797715880379201, "learning_rate": 0.0013957305535572037, "grad_norm": 1.8395501375198364, "tokens_per_second": 4042848.511576898, "eta_seconds": 1063.399132490158, "data_wait_seconds": 1.9162893295288086e-05} {"step": 26570, "total_steps": 28610, "tokens": 55721328640, "source": "finemath", "phase": 5, "loss": 2.1204340159893036, "ce_loss": 2.107903480529785, "logit_z_loss": 39.24235439300537, "router_aux_loss": 1.2046575844287872, "router_z_loss": 0.4839385189116001, "routing_entropy_normalized": 0.6737986209026731, "learning_rate": 0.0013847754288200298, "grad_norm": 0.5850273966789246, "tokens_per_second": 4420825.699686776, "eta_seconds": 967.7355251312256, "data_wait_seconds": 2.1636486053466797e-05} {"step": 26580, "total_steps": 28610, "tokens": 55742300160, "source": "cosmopedia_v2", "phase": 5, "loss": 2.71760293841362, "ce_loss": 2.704732358455658, "logit_z_loss": 27.511191368103027, "router_aux_loss": 1.1820543706417084, "router_z_loss": 1.0500392764806747, "routing_entropy_normalized": 0.708944416085255, "learning_rate": 0.0013738264832269202, "grad_norm": 1.4178376197814941, "tokens_per_second": 3774483.0404126314, "eta_seconds": 1127.8944730758667, "data_wait_seconds": 2.086162567138672e-05} {"step": 26590, "total_steps": 28610, "tokens": 55763271680, "source": "fineweb_edu", "phase": 5, "loss": 3.3557665944099426, "ce_loss": 3.3430144786834717, "logit_z_loss": 26.120474815368652, "router_aux_loss": 1.201794683933258, "router_z_loss": 0.7341562435030937, "routing_entropy_normalized": 0.6907505484157362, "learning_rate": 0.0013628843039365985, "grad_norm": 1.176041603088379, "tokens_per_second": 4046236.001720782, "eta_seconds": 1046.9599494934082, "data_wait_seconds": 1.9818544387817383e-05} {"step": 26600, "total_steps": 28610, "tokens": 55784243200, "source": "cosmopedia_v2", "phase": 5, "loss": 2.6892812252044678, "ce_loss": 2.6759133636951447, "logit_z_loss": 28.492018461227417, "router_aux_loss": 1.2340872585773468, "router_z_loss": 1.0269992649555206, "routing_entropy_normalized": 0.6857035424247002, "learning_rate": 0.001351949477744932, "grad_norm": 0.9535841941833496, "tokens_per_second": 4142053.1304558706, "eta_seconds": 1017.677800655365, "data_wait_seconds": 2.0325183868408203e-05} {"step": 26610, "total_steps": 28610, "tokens": 55805214720, "source": "npset2", "phase": 5, "loss": 1.5816690474748611, "ce_loss": 1.5690102130174637, "logit_z_loss": 40.693235874176025, "router_aux_loss": 1.1968835443258286, "router_z_loss": 0.69002366065979, "routing_entropy_normalized": 0.673887201666591, "learning_rate": 0.0013410225910534636, "grad_norm": 0.4452550709247589, "tokens_per_second": 4169546.0083669713, "eta_seconds": 1005.9378147125244, "data_wait_seconds": 2.1964311599731445e-05} {"step": 26620, "total_steps": 28610, "tokens": 55826186240, "source": "fineweb_edu", "phase": 5, "loss": 3.3249068558216095, "ce_loss": 3.312472492456436, "logit_z_loss": 26.00437021255493, "router_aux_loss": 1.16546830534935, "router_z_loss": 0.7796386927366257, "routing_entropy_normalized": 0.711430632784104, "learning_rate": 0.001330104229837966, "grad_norm": 0.6452831625938416, "tokens_per_second": 4404266.656903544, "eta_seconds": 947.5658049583436, "data_wait_seconds": 2.014636993408203e-05} {"step": 26630, "total_steps": 28610, "tokens": 55847157760, "source": "dclm", "phase": 5, "loss": 3.4847119450569153, "ce_loss": 3.4723925292491913, "logit_z_loss": 26.597360372543335, "router_aux_loss": 1.1719747185707092, "router_z_loss": 0.5996079668402672, "routing_entropy_normalized": 0.6857895221799953, "learning_rate": 0.0013191949796170156, "grad_norm": 0.6408635973930359, "tokens_per_second": 3378664.559008659, "eta_seconds": 1228.9947366714478, "data_wait_seconds": 1.9669532775878906e-05} {"step": 26640, "total_steps": 28610, "tokens": 55868129280, "source": "finemath", "phase": 5, "loss": 2.1581434309482574, "ce_loss": 2.14564311504364, "logit_z_loss": 37.72417688369751, "router_aux_loss": 1.213011771440506, "router_z_loss": 0.3701830394566059, "routing_entropy_normalized": 0.6613989915420927, "learning_rate": 0.001308295425420593, "grad_norm": 1.814217209815979, "tokens_per_second": 3720902.5293781525, "eta_seconds": 1110.3191785812378, "data_wait_seconds": 1.6957521438598633e-05} {"step": 26650, "total_steps": 28610, "tokens": 55889100800, "source": "dclm", "phase": 5, "loss": 3.490166276693344, "ce_loss": 3.477458268404007, "logit_z_loss": 26.62082529067993, "router_aux_loss": 1.2177723050117493, "router_z_loss": 0.5302591025829315, "routing_entropy_normalized": 0.6771058171903787, "learning_rate": 0.0012974061517587122, "grad_norm": 1.4752663373947144, "tokens_per_second": 3857088.8373881085, "eta_seconds": 1065.6788301467896, "data_wait_seconds": 2.086162567138672e-05} {"step": 26660, "total_steps": 28610, "tokens": 55910072320, "source": "cosmopedia_v2", "phase": 5, "loss": 2.6869221329689026, "ce_loss": 2.673796832561493, "logit_z_loss": 29.791767835617065, "router_aux_loss": 1.2186841070652008, "router_z_loss": 0.938496045768261, "routing_entropy_normalized": 0.6703583793998382, "learning_rate": 0.0012865277425900725, "grad_norm": 1.9505070447921753, "tokens_per_second": 4056582.46512718, "eta_seconds": 1008.10138463974, "data_wait_seconds": 2.3692846298217773e-05} {"step": 26670, "total_steps": 28610, "tokens": 55931043840, "source": "finemath", "phase": 5, "loss": 2.0814130008220673, "ce_loss": 2.0690028369426727, "logit_z_loss": 38.29766511917114, "router_aux_loss": 1.203299656510353, "router_z_loss": 0.37713829427957535, "routing_entropy_normalized": 0.6894612184356115, "learning_rate": 0.0012756607812907427, "grad_norm": 0.6783751249313354, "tokens_per_second": 3897393.2236799137, "eta_seconds": 1043.8964319229126, "data_wait_seconds": 0.05696934461593628} {"step": 26680, "total_steps": 28610, "tokens": 55952015360, "source": "cosmopedia_v2", "phase": 5, "loss": 2.6932443380355835, "ce_loss": 2.6801024675369263, "logit_z_loss": 28.14855408668518, "router_aux_loss": 1.2224448472261429, "router_z_loss": 0.9174064695835114, "routing_entropy_normalized": 0.6669215292449919, "learning_rate": 0.0012648058506228778, "grad_norm": 3.2453434467315674, "tokens_per_second": 4183601.1807980062, "eta_seconds": 967.4687392711639, "data_wait_seconds": 2.238154411315918e-05} {"step": 26690, "total_steps": 28610, "tokens": 55972986880, "source": "fineweb_edu", "phase": 5, "loss": 3.3196659088134766, "ce_loss": 3.3064522743225098, "logit_z_loss": 25.76193356513977, "router_aux_loss": 1.2567742317914963, "router_z_loss": 0.6458994001150131, "routing_entropy_normalized": 0.6714140631237306, "learning_rate": 0.0012539635327034676, "grad_norm": 1.039702296257019, "tokens_per_second": 3991328.355573319, "eta_seconds": 1008.8199920654297, "data_wait_seconds": 2.5540590286254883e-05} {"step": 26700, "total_steps": 28610, "tokens": 55993958400, "source": "cosmopedia_v2", "phase": 5, "loss": 2.677870362997055, "ce_loss": 2.6649544835090637, "logit_z_loss": 28.27545142173767, "router_aux_loss": 1.1932083070278168, "router_z_loss": 0.9838202074170113, "routing_entropy_normalized": 0.6892674571132097, "learning_rate": 0.001243134408973116, "grad_norm": 0.9129019975662231, "tokens_per_second": 3883527.8140203673, "eta_seconds": 1031.423105955124, "data_wait_seconds": 2.4199485778808594e-05} {"step": 26710, "total_steps": 28610, "tokens": 56014929920, "source": "npset2", "phase": 5, "loss": 1.636982336640358, "ce_loss": 1.6243216693401337, "logit_z_loss": 40.092557430267334, "router_aux_loss": 1.2021003663539886, "router_z_loss": 0.6396542862057686, "routing_entropy_normalized": 0.6911723021934106, "learning_rate": 0.0012323190601648646, "grad_norm": 1.0221673250198364, "tokens_per_second": 3671087.7941661705, "eta_seconds": 1085.3973054885864, "data_wait_seconds": 1.8417835235595703e-05} {"step": 26720, "total_steps": 28610, "tokens": 56035901440, "source": "fineweb_edu", "phase": 5, "loss": 3.3013875782489777, "ce_loss": 3.2888972461223602, "logit_z_loss": 25.681470155715942, "router_aux_loss": 1.1761672794818878, "router_z_loss": 0.728621780872345, "routing_entropy_normalized": 0.6999486353568939, "learning_rate": 0.0012215180662730467, "grad_norm": 1.2504758834838867, "tokens_per_second": 4128465.166940111, "eta_seconds": 960.0704183578491, "data_wait_seconds": 2.0802021026611328e-05} {"step": 26730, "total_steps": 28610, "tokens": 56056872960, "source": "dclm", "phase": 5, "loss": 3.4584646224975586, "ce_loss": 3.4460761845111847, "logit_z_loss": 26.392282724380493, "router_aux_loss": 1.1824638098478317, "router_z_loss": 0.5637734979391098, "routing_entropy_normalized": 0.7099815616621364, "learning_rate": 0.0012107320065221866, "grad_norm": 1.449169635772705, "tokens_per_second": 3916802.500799405, "eta_seconds": 1006.5980501174927, "data_wait_seconds": 2.0772218704223633e-05} {"step": 26740, "total_steps": 28610, "tokens": 56077844480, "source": "finemath", "phase": 5, "loss": 2.1334515064954758, "ce_loss": 2.121169298887253, "logit_z_loss": 38.05854272842407, "router_aux_loss": 1.1912282407283783, "router_z_loss": 0.36993756890296936, "routing_entropy_normalized": 0.6567319736951651, "learning_rate": 0.0011999614593359337, "grad_norm": 0.3902529776096344, "tokens_per_second": 3583312.508519952, "eta_seconds": 1094.427078485489, "data_wait_seconds": 1.755356788635254e-05} {"step": 26750, "total_steps": 28610, "tokens": 56098816000, "source": "dclm", "phase": 5, "loss": 3.5189713537693024, "ce_loss": 3.5064379274845123, "logit_z_loss": 26.277918577194214, "router_aux_loss": 1.1922633647918701, "router_z_loss": 0.6108027324080467, "routing_entropy_normalized": 0.6738141201305857, "learning_rate": 0.0011892070023060467, "grad_norm": 1.0912598371505737, "tokens_per_second": 3907378.440307052, "eta_seconds": 998.2915091514587, "data_wait_seconds": 2.3692846298217773e-05} {"step": 26760, "total_steps": 28610, "tokens": 56119787520, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7025295794010162, "ce_loss": 2.6898871064186096, "logit_z_loss": 29.01583218574524, "router_aux_loss": 1.1751265078783035, "router_z_loss": 0.8911930024623871, "routing_entropy_normalized": 0.6860343997945464, "learning_rate": 0.0011784692121614186, "grad_norm": 1.7114814519882202, "tokens_per_second": 3417024.8323740955, "eta_seconds": 1135.4120588302612, "data_wait_seconds": 1.963973045349121e-05} {"step": 26770, "total_steps": 28610, "tokens": 56140759040, "source": "finemath", "phase": 5, "loss": 2.113733634352684, "ce_loss": 2.1017148792743683, "logit_z_loss": 38.758830547332764, "router_aux_loss": 1.1648533791303635, "router_z_loss": 0.37025464698672295, "routing_entropy_normalized": 0.7280109027683618, "learning_rate": 0.0011677486647371463, "grad_norm": 0.9790392518043518, "tokens_per_second": 3876351.6064954204, "eta_seconds": 995.4617309570312, "data_wait_seconds": 2.0056962966918945e-05} {"step": 26780, "total_steps": 28610, "tokens": 56161730560, "source": "cosmopedia_v2", "phase": 5, "loss": 2.6761710941791534, "ce_loss": 2.6632563173770905, "logit_z_loss": 27.766438245773315, "router_aux_loss": 1.2050562798976898, "router_z_loss": 0.86422910541296, "routing_entropy_normalized": 0.6420568492017625, "learning_rate": 0.0011570459349436542, "grad_norm": 1.5187362432479858, "tokens_per_second": 3598507.6015247265, "eta_seconds": 1066.4943873882294, "data_wait_seconds": 2.244114875793457e-05} {"step": 26790, "total_steps": 28610, "tokens": 56182702080, "source": "fineweb_edu", "phase": 5, "loss": 3.342780739068985, "ce_loss": 3.3303356766700745, "logit_z_loss": 26.164191484451294, "router_aux_loss": 1.1847972869873047, "router_z_loss": 0.5970424339175224, "routing_entropy_normalized": 0.6990804691621231, "learning_rate": 0.001146361596735859, "grad_norm": 1.0902901887893677, "tokens_per_second": 3803411.9574824898, "eta_seconds": 1003.5243835449219, "data_wait_seconds": 2.2411346435546875e-05} {"step": 26800, "total_steps": 28610, "tokens": 56203673600, "source": "cosmopedia_v2", "phase": 5, "loss": 2.683404356241226, "ce_loss": 2.6703766882419586, "logit_z_loss": 28.922803163528442, "router_aux_loss": 1.2161559760570526, "router_z_loss": 0.8661316186189651, "routing_entropy_normalized": 0.657490125507659, "learning_rate": 0.001135696223082395, "grad_norm": 1.4166899919509888, "tokens_per_second": 4035051.0952933114, "eta_seconds": 940.7179811000824, "data_wait_seconds": 2.345442771911621e-05} {"step": 26810, "total_steps": 28610, "tokens": 56224645120, "source": "npset2", "phase": 5, "loss": 1.5872646570205688, "ce_loss": 1.5746716856956482, "logit_z_loss": 42.061179637908936, "router_aux_loss": 1.2025295346975327, "router_z_loss": 0.5676737576723099, "routing_entropy_normalized": 0.6676891619032959, "learning_rate": 0.0011250503859348815, "grad_norm": 0.7660228610038757, "tokens_per_second": 3995144.3894717703, "eta_seconds": 944.8653745651245, "data_wait_seconds": 2.288818359375e-05} {"step": 26820, "total_steps": 28610, "tokens": 56245616640, "source": "fineweb_edu", "phase": 5, "loss": 3.34057879447937, "ce_loss": 3.3279893696308136, "logit_z_loss": 25.593172550201416, "router_aux_loss": 1.1983586996793747, "router_z_loss": 0.6058622598648071, "routing_entropy_normalized": 0.6897397578471065, "learning_rate": 0.001114424656197258, "grad_norm": 0.9086673855781555, "tokens_per_second": 3885775.58425857, "eta_seconds": 966.0625011920929, "data_wait_seconds": 2.059340476989746e-05} {"step": 26830, "total_steps": 28610, "tokens": 56266588160, "source": "dclm", "phase": 5, "loss": 3.4879902601242065, "ce_loss": 3.4754682183265686, "logit_z_loss": 26.955199718475342, "router_aux_loss": 1.2011335492134094, "router_z_loss": 0.510759137570858, "routing_entropy_normalized": 0.649463913423533, "learning_rate": 0.0011038196036951622, "grad_norm": 0.7980725765228271, "tokens_per_second": 4136339.2901324946, "eta_seconds": 902.4720406532288, "data_wait_seconds": 1.582503318786621e-05} {"step": 26840, "total_steps": 28610, "tokens": 56287559680, "source": "finemath", "phase": 5, "loss": 2.091466560959816, "ce_loss": 2.079535633325577, "logit_z_loss": 39.30641031265259, "router_aux_loss": 1.1581613719463348, "router_z_loss": 0.34936097264289856, "routing_entropy_normalized": 0.7114558206372813, "learning_rate": 0.0010932357971453744, "grad_norm": 0.9255495667457581, "tokens_per_second": 4219516.246984344, "eta_seconds": 879.7119913101196, "data_wait_seconds": 2.1219253540039062e-05} {"step": 26850, "total_steps": 28610, "tokens": 56308531200, "source": "dclm", "phase": 5, "loss": 3.497603714466095, "ce_loss": 3.4856384992599487, "logit_z_loss": 26.39293336868286, "router_aux_loss": 1.1479056030511856, "router_z_loss": 0.4862239845097065, "routing_entropy_normalized": 0.7171448210330112, "learning_rate": 0.0010826738041253212, "grad_norm": 0.9984221458435059, "tokens_per_second": 4043439.1238351734, "eta_seconds": 912.8337059020996, "data_wait_seconds": 2.0802021026611328e-05} {"step": 26860, "total_steps": 28610, "tokens": 56329502720, "source": "cosmopedia_v2", "phase": 5, "loss": 2.6861060559749603, "ce_loss": 2.6737519204616547, "logit_z_loss": 28.067408084869385, "router_aux_loss": 1.153065949678421, "router_z_loss": 0.8234642669558525, "routing_entropy_normalized": 0.6926280608388116, "learning_rate": 0.001072134191042633, "grad_norm": 1.4067950248718262, "tokens_per_second": 4149800.858941193, "eta_seconds": 884.3836426734924, "data_wait_seconds": 1.6063451766967773e-05} {"step": 26870, "total_steps": 28610, "tokens": 56350474240, "source": "finemath", "phase": 5, "loss": 2.0701867640018463, "ce_loss": 2.057514488697052, "logit_z_loss": 39.28429841995239, "router_aux_loss": 1.231732800602913, "router_z_loss": 0.35494449734687805, "routing_entropy_normalized": 0.6393875318475132, "learning_rate": 0.0010616175231047745, "grad_norm": 1.210857629776001, "tokens_per_second": 3711225.736718828, "eta_seconds": 983.2450890541076, "data_wait_seconds": 2.181529998779297e-05} {"step": 26880, "total_steps": 28610, "tokens": 56371445760, "source": "cosmopedia_v2", "phase": 5, "loss": 2.672047257423401, "ce_loss": 2.659090995788574, "logit_z_loss": 28.584959983825684, "router_aux_loss": 1.2137347608804703, "router_z_loss": 0.8188758492469788, "routing_entropy_normalized": 0.6753744106707195, "learning_rate": 0.0010511243642887295, "grad_norm": 1.4516658782958984, "tokens_per_second": 4207176.988366581, "eta_seconds": 862.3533000946045, "data_wait_seconds": 1.7851591110229492e-05} {"step": 26890, "total_steps": 28610, "tokens": 56392417280, "source": "fineweb_edu", "phase": 5, "loss": 3.3275181353092194, "ce_loss": 3.3151320815086365, "logit_z_loss": 25.73158884048462, "router_aux_loss": 1.1805078089237213, "router_z_loss": 0.5810066163539886, "routing_entropy_normalized": 0.681407693577358, "learning_rate": 0.001040655277310761, "grad_norm": 1.457423448562622, "tokens_per_second": 4246515.548864538, "eta_seconds": 849.426170349121, "data_wait_seconds": 2.0265579223632812e-05} {"step": 26900, "total_steps": 28610, "tokens": 56413388800, "source": "cosmopedia_v2", "phase": 5, "loss": 2.6830121278762817, "ce_loss": 2.670155018568039, "logit_z_loss": 28.878682851791382, "router_aux_loss": 1.2008183151483536, "router_z_loss": 0.8489385917782784, "routing_entropy_normalized": 0.6735486420677322, "learning_rate": 0.001030210823596232, "grad_norm": 1.0084223747253418, "tokens_per_second": 3671030.0332271447, "eta_seconds": 976.8729450702667, "data_wait_seconds": 2.2351741790771484e-05} {"step": 26910, "total_steps": 28610, "tokens": 56434360320, "source": "npset2", "phase": 5, "loss": 1.5682294070720673, "ce_loss": 1.5556501597166061, "logit_z_loss": 38.81826686859131, "router_aux_loss": 1.201519176363945, "router_z_loss": 0.5640315562486649, "routing_entropy_normalized": 0.6591180711311008, "learning_rate": 0.0010197915632494969, "grad_norm": 0.5162240862846375, "tokens_per_second": 3458382.806455045, "eta_seconds": 1030.8744287490845, "data_wait_seconds": 2.2530555725097656e-05} {"step": 26920, "total_steps": 28610, "tokens": 56455331840, "source": "fineweb_edu", "phase": 5, "loss": 3.309607148170471, "ce_loss": 3.2972211241722107, "logit_z_loss": 26.090455770492554, "router_aux_loss": 1.1754001528024673, "router_z_loss": 0.6320090964436531, "routing_entropy_normalized": 0.6674091951562922, "learning_rate": 0.0010093980550238675, "grad_norm": 0.9708734154701233, "tokens_per_second": 3957448.5300740185, "eta_seconds": 895.5737144947052, "data_wait_seconds": 1.96993350982666e-05} {"step": 26930, "total_steps": 28610, "tokens": 56476303360, "source": "dclm", "phase": 5, "loss": 3.509527087211609, "ce_loss": 3.497220367193222, "logit_z_loss": 25.976420640945435, "router_aux_loss": 1.177903562784195, "router_z_loss": 0.5277201980352402, "routing_entropy_normalized": 0.6802614646016311, "learning_rate": 0.000999030856291648, "grad_norm": 0.7383552193641663, "tokens_per_second": 4133890.300799375, "eta_seconds": 852.2759685516357, "data_wait_seconds": 1.901388168334961e-05} {"step": 26940, "total_steps": 28610, "tokens": 56497274880, "source": "finemath", "phase": 5, "loss": 2.136761888861656, "ce_loss": 2.124861091375351, "logit_z_loss": 36.74156188964844, "router_aux_loss": 1.1522388011217117, "router_z_loss": 0.37842628732323647, "routing_entropy_normalized": 0.6979221366451103, "learning_rate": 0.0009886905230142432, "grad_norm": 1.2455472946166992, "tokens_per_second": 3905725.860643967, "eta_seconds": 896.6947412490845, "data_wait_seconds": 2.0563602447509766e-05} {"step": 26950, "total_steps": 28610, "tokens": 56518246400, "source": "dclm", "phase": 5, "loss": 3.4925937950611115, "ce_loss": 3.4805973768234253, "logit_z_loss": 26.823571920394897, "router_aux_loss": 1.1475897282361984, "router_z_loss": 0.5205151326954365, "routing_entropy_normalized": 0.6918732768207556, "learning_rate": 0.0009783776097123454, "grad_norm": 0.5322127938270569, "tokens_per_second": 3765427.5685149888, "eta_seconds": 924.5357284545898, "data_wait_seconds": 2.09808349609375e-05} {"step": 26960, "total_steps": 28610, "tokens": 56539217920, "source": "cosmopedia_v2", "phase": 5, "loss": 2.674781858921051, "ce_loss": 2.662668377161026, "logit_z_loss": 28.156776189804077, "router_aux_loss": 1.1273177415132523, "router_z_loss": 0.8402318581938744, "routing_entropy_normalized": 0.7140384164029302, "learning_rate": 0.0009680926694361965, "grad_norm": 1.599409580230713, "tokens_per_second": 4223811.336793339, "eta_seconds": 819.2365908622742, "data_wait_seconds": 1.9669532775878906e-05} {"step": 26970, "total_steps": 28610, "tokens": 56560189440, "source": "finemath", "phase": 5, "loss": 2.0491650700569153, "ce_loss": 2.0373330116271973, "logit_z_loss": 37.24039125442505, "router_aux_loss": 1.1468837559223175, "router_z_loss": 0.3631895314902067, "routing_entropy_normalized": 0.7149362415310032, "learning_rate": 0.0009578362537359315, "grad_norm": 1.3410450220108032, "tokens_per_second": 4059970.5319266734, "eta_seconds": 847.1315870285034, "data_wait_seconds": 1.946091651916504e-05} {"step": 26980, "total_steps": 28610, "tokens": 56581160960, "source": "cosmopedia_v2", "phase": 5, "loss": 2.651888132095337, "ce_loss": 2.6397843658924103, "logit_z_loss": 28.300330638885498, "router_aux_loss": 1.1288045644760132, "router_z_loss": 0.8156850263476372, "routing_entropy_normalized": 0.7176337318994974, "learning_rate": 0.0009476089126319966, "grad_norm": 0.8704654574394226, "tokens_per_second": 3758632.9481236795, "eta_seconds": 909.4683644771576, "data_wait_seconds": 2.2858381271362305e-05} {"step": 26990, "total_steps": 28610, "tokens": 56602132480, "source": "fineweb_edu", "phase": 5, "loss": 3.297653079032898, "ce_loss": 3.285678446292877, "logit_z_loss": 25.941193342208862, "router_aux_loss": 1.1402114033699036, "router_z_loss": 0.572528749704361, "routing_entropy_normalized": 0.6974189064189417, "learning_rate": 0.0009374111945856571, "grad_norm": 0.7959719300270081, "tokens_per_second": 4284412.745617096, "eta_seconds": 792.9642734527588, "data_wait_seconds": 2.0623207092285156e-05} {"step": 27000, "total_steps": 28610, "tokens": 56623104000, "source": "cosmopedia_v2", "phase": 5, "loss": 2.657414436340332, "ce_loss": 2.6453470289707184, "logit_z_loss": 27.923677682876587, "router_aux_loss": 1.1301083862781525, "router_z_loss": 0.7663107439875603, "routing_entropy_normalized": 0.7053330909806086, "learning_rate": 0.000927243646469584, "grad_norm": 1.1093636751174927, "tokens_per_second": 4207573.447932212, "eta_seconds": 802.4612669944763, "data_wait_seconds": 2.0384788513183594e-05} {"step": 27010, "total_steps": 28610, "tokens": 56644075520, "source": "npset2", "phase": 5, "loss": 1.595350980758667, "ce_loss": 1.5836369693279266, "logit_z_loss": 39.23710632324219, "router_aux_loss": 1.1246405243873596, "router_z_loss": 0.46758976951241493, "routing_entropy_normalized": 0.718303438868533, "learning_rate": 0.0009171068135385256, "grad_norm": 0.5948740243911743, "tokens_per_second": 3962292.9479264896, "eta_seconds": 846.8437957763672, "data_wait_seconds": 2.130866050720215e-05} {"step": 27020, "total_steps": 28610, "tokens": 56665047040, "source": "fineweb_edu", "phase": 5, "loss": 3.283495932817459, "ce_loss": 3.271508276462555, "logit_z_loss": 25.5941960811615, "router_aux_loss": 1.1485477983951569, "router_z_loss": 0.5021529085934162, "routing_entropy_normalized": 0.6749819797640961, "learning_rate": 0.000907001239400068, "grad_norm": 1.1665195226669312, "tokens_per_second": 3927872.1612629, "eta_seconds": 848.9257142543793, "data_wait_seconds": 2.2411346435546875e-05} {"step": 27030, "total_steps": 28610, "tokens": 56686018560, "source": "dclm", "phase": 5, "loss": 3.4544862508773804, "ce_loss": 3.4428919851779938, "logit_z_loss": 27.02316188812256, "router_aux_loss": 1.1203729659318924, "router_z_loss": 0.3905331790447235, "routing_entropy_normalized": 0.7296527894946397, "learning_rate": 0.0008969274659854831, "grad_norm": 1.088804006576538, "tokens_per_second": 4205761.010305297, "eta_seconds": 787.8479428291321, "data_wait_seconds": 2.339482307434082e-05} {"step": 27040, "total_steps": 28610, "tokens": 56706990080, "source": "finemath", "phase": 5, "loss": 2.0888124108314514, "ce_loss": 2.0773110687732697, "logit_z_loss": 38.81069898605347, "router_aux_loss": 1.123294785618782, "router_z_loss": 0.2684058379381895, "routing_entropy_normalized": 0.6802595538509729, "learning_rate": 0.0008868860335206678, "grad_norm": 0.5251704454421997, "tokens_per_second": 3937082.639282001, "eta_seconds": 836.2863931655884, "data_wait_seconds": 1.7315149307250977e-05} {"step": 27050, "total_steps": 28610, "tokens": 56727961600, "source": "dclm", "phase": 5, "loss": 3.474793940782547, "ce_loss": 3.4632466435432434, "logit_z_loss": 26.37086534500122, "router_aux_loss": 1.1160024255514145, "router_z_loss": 0.38726136088371277, "routing_entropy_normalized": 0.6937214951870472, "learning_rate": 0.0008768774804971706, "grad_norm": 0.6344671845436096, "tokens_per_second": 3804078.2965002027, "eta_seconds": 860.013087272644, "data_wait_seconds": 2.3484230041503906e-05} {"step": 27060, "total_steps": 28610, "tokens": 56748933120, "source": "cosmopedia_v2", "phase": 5, "loss": 2.6539367139339447, "ce_loss": 2.6418879330158234, "logit_z_loss": 28.446779251098633, "router_aux_loss": 1.1359736174345016, "router_z_loss": 0.689084842801094, "routing_entropy_normalized": 0.6907683254700089, "learning_rate": 0.0008669023436433149, "grad_norm": 0.7828267216682434, "tokens_per_second": 3780783.1830204227, "eta_seconds": 859.7651445865631, "data_wait_seconds": 2.4199485778808594e-05} {"step": 27070, "total_steps": 28610, "tokens": 56769904640, "source": "finemath", "phase": 5, "loss": 2.130101576447487, "ce_loss": 2.1185649782419205, "logit_z_loss": 37.989882946014404, "router_aux_loss": 1.1257482320070267, "router_z_loss": 0.2791409492492676, "routing_entropy_normalized": 0.7173042995574955, "learning_rate": 0.0008569611578954185, "grad_norm": 0.6642813086509705, "tokens_per_second": 4313376.018210532, "eta_seconds": 748.7439227104186, "data_wait_seconds": 2.0891427993774414e-05} {"step": 27080, "total_steps": 28610, "tokens": 56790876160, "source": "cosmopedia_v2", "phase": 5, "loss": 2.64849516749382, "ce_loss": 2.6366628408432007, "logit_z_loss": 28.732802152633667, "router_aux_loss": 1.1197201609611511, "router_z_loss": 0.6351731196045876, "routing_entropy_normalized": 0.6595715444113781, "learning_rate": 0.0008470544563691032, "grad_norm": 1.1866569519042969, "tokens_per_second": 3917303.472868351, "eta_seconds": 819.0947120189667, "data_wait_seconds": 1.9669532775878906e-05} {"step": 27090, "total_steps": 28610, "tokens": 56811847680, "source": "fineweb_edu", "phase": 5, "loss": 3.3148223757743835, "ce_loss": 3.3030430674552917, "logit_z_loss": 24.514413833618164, "router_aux_loss": 1.1330312341451645, "router_z_loss": 0.448978528380394, "routing_entropy_normalized": 0.6967840093807388, "learning_rate": 0.0008371827703307048, "grad_norm": 2.369264841079712, "tokens_per_second": 3168277.4147677193, "eta_seconds": 1006.1211891174316, "data_wait_seconds": 2.2977590560913086e-05} {"step": 27100, "total_steps": 28610, "tokens": 56832819200, "source": "cosmopedia_v2", "phase": 5, "loss": 2.653329074382782, "ce_loss": 2.6411030292510986, "logit_z_loss": 28.855651378631592, "router_aux_loss": 1.1593342274427414, "router_z_loss": 0.632711261510849, "routing_entropy_normalized": 0.6589907281651515, "learning_rate": 0.0008273466291687882, "grad_norm": 0.8687689900398254, "tokens_per_second": 4562529.817294295, "eta_seconds": 694.0665917396545, "data_wait_seconds": 2.282857894897461e-05} {"step": 27110, "total_steps": 28610, "tokens": 56853790720, "source": "npset2", "phase": 5, "loss": 1.641847625374794, "ce_loss": 1.6290705800056458, "logit_z_loss": 38.75720167160034, "router_aux_loss": 1.2358923554420471, "router_z_loss": 0.4181269705295563, "routing_entropy_normalized": 0.6474294759472277, "learning_rate": 0.0008175465603657528, "grad_norm": 1.047889232635498, "tokens_per_second": 3471804.6235407256, "eta_seconds": 906.0786366462708, "data_wait_seconds": 3.1054019927978516e-05} {"step": 27120, "total_steps": 28610, "tokens": 56874762240, "source": "fineweb_edu", "phase": 5, "loss": 3.4966771006584167, "ce_loss": 3.4841882586479187, "logit_z_loss": 23.89732074737549, "router_aux_loss": 1.2032666355371475, "router_z_loss": 0.45619528368115425, "routing_entropy_normalized": 0.6822312679336763, "learning_rate": 0.0008077830894695457, "grad_norm": 0.5616018176078796, "tokens_per_second": 3768855.8186700786, "eta_seconds": 829.0995013713837, "data_wait_seconds": 2.753734588623047e-05} {"step": 27130, "total_steps": 28610, "tokens": 56895733760, "source": "dclm", "phase": 5, "loss": 3.622161418199539, "ce_loss": 3.610355854034424, "logit_z_loss": 24.716883182525635, "router_aux_loss": 1.142385944724083, "router_z_loss": 0.38176438957452774, "routing_entropy_normalized": 0.716047523735717, "learning_rate": 0.0007980567400654808, "grad_norm": 0.800957441329956, "tokens_per_second": 4297653.329330832, "eta_seconds": 722.20458984375, "data_wait_seconds": 2.7954578399658203e-05} {"step": 27140, "total_steps": 28610, "tokens": 56916705280, "source": "finemath", "phase": 5, "loss": 2.267115503549576, "ce_loss": 2.255267232656479, "logit_z_loss": 35.56045722961426, "router_aux_loss": 1.1585522294044495, "router_z_loss": 0.26272756047546864, "routing_entropy_normalized": 0.7056704430171978, "learning_rate": 0.0007883680337481599, "grad_norm": 0.6802553534507751, "tokens_per_second": 3957693.18519887, "eta_seconds": 778.9419987201691, "data_wait_seconds": 2.810359001159668e-05} {"step": 27150, "total_steps": 28610, "tokens": 56937676800, "source": "dclm", "phase": 5, "loss": 3.565791994333267, "ce_loss": 3.554145574569702, "logit_z_loss": 24.83358120918274, "router_aux_loss": 1.1286068260669708, "router_z_loss": 0.36037326976656914, "routing_entropy_normalized": 0.7379005432686291, "learning_rate": 0.0007787174900934976, "grad_norm": 0.771852970123291, "tokens_per_second": 1737382.6579812937, "eta_seconds": 1762.330195903778, "data_wait_seconds": 2.1964311599731445e-05} {"step": 27160, "total_steps": 28610, "tokens": 56958648320, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7658964097499847, "ce_loss": 2.753924936056137, "logit_z_loss": 27.595478057861328, "router_aux_loss": 1.1256768256425858, "router_z_loss": 0.7147043198347092, "routing_entropy_normalized": 0.7230427233392426, "learning_rate": 0.0007691056266308628, "grad_norm": 1.1873530149459839, "tokens_per_second": 3942826.1392121166, "eta_seconds": 771.2413108348846, "data_wait_seconds": 2.3990869522094727e-05} {"step": 27170, "total_steps": 28610, "tokens": 56979619840, "source": "finemath", "phase": 5, "loss": 2.1898454129695892, "ce_loss": 2.1781243085861206, "logit_z_loss": 35.876676082611084, "router_aux_loss": 1.1464245170354843, "router_z_loss": 0.25689063780009747, "routing_entropy_normalized": 0.6990924799985128, "learning_rate": 0.0007595329588153221, "grad_norm": 0.9159744381904602, "tokens_per_second": 4046207.15202699, "eta_seconds": 746.3530082702637, "data_wait_seconds": 2.0205974578857422e-05} {"step": 27180, "total_steps": 28610, "tokens": 57000591360, "source": "cosmopedia_v2", "phase": 5, "loss": 2.744966447353363, "ce_loss": 2.732774317264557, "logit_z_loss": 27.7472026348114, "router_aux_loss": 1.1514736115932465, "router_z_loss": 0.6773731112480164, "routing_entropy_normalized": 0.66993255428104, "learning_rate": 0.0007500000000000003, "grad_norm": 3.3361501693725586, "tokens_per_second": 4152975.482551904, "eta_seconds": 722.1153538227081, "data_wait_seconds": 2.0563602447509766e-05} {"step": 27190, "total_steps": 28610, "tokens": 57021562880, "source": "fineweb_edu", "phase": 5, "loss": 3.359998971223831, "ce_loss": 3.3480939865112305, "logit_z_loss": 24.516781330108643, "router_aux_loss": 1.1459475606679916, "router_z_loss": 0.4454890415072441, "routing_entropy_normalized": 0.7108187501580212, "learning_rate": 0.0007405072614085462, "grad_norm": 0.45561233162879944, "tokens_per_second": 4610843.146975151, "eta_seconds": 645.8592810630798, "data_wait_seconds": 1.8417835235595703e-05} {"step": 27200, "total_steps": 28610, "tokens": 57042534400, "source": "cosmopedia_v2", "phase": 5, "loss": 2.73199862241745, "ce_loss": 2.7198681831359863, "logit_z_loss": 27.525178909301758, "router_aux_loss": 1.144388496875763, "router_z_loss": 0.6865886449813843, "routing_entropy_normalized": 0.7208260226315351, "learning_rate": 0.0007310552521077218, "grad_norm": 0.8272308707237244, "tokens_per_second": 3772949.6790551734, "eta_seconds": 783.7327744960785, "data_wait_seconds": 2.008676528930664e-05} {"step": 27210, "total_steps": 28610, "tokens": 57063505920, "source": "npset2", "phase": 5, "loss": 1.6135769188404083, "ce_loss": 1.6016628444194794, "logit_z_loss": 40.03337001800537, "router_aux_loss": 1.143881395459175, "router_z_loss": 0.4752824157476425, "routing_entropy_normalized": 0.6523975004399571, "learning_rate": 0.0007216444789801009, "grad_norm": 0.7584613561630249, "tokens_per_second": 3970799.9516286203, "eta_seconds": 739.4008350372314, "data_wait_seconds": 1.817941665649414e-05} {"step": 27220, "total_steps": 28610, "tokens": 57084477440, "source": "fineweb_edu", "phase": 5, "loss": 3.380850166082382, "ce_loss": 3.3689653277397156, "logit_z_loss": 24.89562439918518, "router_aux_loss": 1.1439179182052612, "router_z_loss": 0.44564804807305336, "routing_entropy_normalized": 0.7003654366896488, "learning_rate": 0.0007122754466968858, "grad_norm": 0.47999805212020874, "tokens_per_second": 4251250.409344301, "eta_seconds": 685.6903262138367, "data_wait_seconds": 2.1249055862426758e-05} {"step": 27230, "total_steps": 28610, "tokens": 57105448960, "source": "dclm", "phase": 5, "loss": 3.563316822052002, "ce_loss": 3.551588863134384, "logit_z_loss": 25.18594217300415, "router_aux_loss": 1.1399169266223907, "router_z_loss": 0.3287147842347622, "routing_entropy_normalized": 0.7041120360176503, "learning_rate": 0.0007029486576908444, "grad_norm": 0.7160404324531555, "tokens_per_second": 3840182.775315622, "eta_seconds": 753.6281290054321, "data_wait_seconds": 1.677870750427246e-05} {"step": 27240, "total_steps": 28610, "tokens": 57126420480, "source": "finemath", "phase": 5, "loss": 2.2588150948286057, "ce_loss": 2.247038945555687, "logit_z_loss": 35.96598148345947, "router_aux_loss": 1.1464468389749527, "router_z_loss": 0.31171043775975704, "routing_entropy_normalized": 0.6681863720243265, "learning_rate": 0.0006936646121293654, "grad_norm": 0.7615252137184143, "tokens_per_second": 3818417.5935541485, "eta_seconds": 752.4316473007202, "data_wait_seconds": 1.9043684005737305e-05} {"step": 27250, "total_steps": 28610, "tokens": 57147392000, "source": "dclm", "phase": 5, "loss": 3.519827753305435, "ce_loss": 3.5080226957798004, "logit_z_loss": 25.702720642089844, "router_aux_loss": 1.1436737477779388, "router_z_loss": 0.36832766607403755, "routing_entropy_normalized": 0.6813647329517901, "learning_rate": 0.000684423807887638, "grad_norm": 1.0633331537246704, "tokens_per_second": 4141759.7990278928, "eta_seconds": 688.626781463623, "data_wait_seconds": 2.0205974578857422e-05} {"step": 27260, "total_steps": 28610, "tokens": 57168363520, "source": "cosmopedia_v2", "phase": 5, "loss": 2.710816890001297, "ce_loss": 2.69890233874321, "logit_z_loss": 29.35981559753418, "router_aux_loss": 1.1305030137300491, "router_z_loss": 0.6095302924513817, "routing_entropy_normalized": 0.7062633925879139, "learning_rate": 0.0006752267405219478, "grad_norm": 1.5235810279846191, "tokens_per_second": 4438035.205420683, "eta_seconds": 637.9298651218414, "data_wait_seconds": 2.1189451217651367e-05} {"step": 27270, "total_steps": 28610, "tokens": 57189335040, "source": "finemath", "phase": 5, "loss": 2.2848664224147797, "ce_loss": 2.2729388773441315, "logit_z_loss": 36.83523941040039, "router_aux_loss": 1.1691796481609344, "router_z_loss": 0.23578236997127533, "routing_entropy_normalized": 0.6808883757728483, "learning_rate": 0.0006660739032431065, "grad_norm": 1.8410810232162476, "tokens_per_second": 3651096.287434704, "eta_seconds": 769.6821608543396, "data_wait_seconds": 2.0563602447509766e-05} {"step": 27280, "total_steps": 28610, "tokens": 57210306560, "source": "cosmopedia_v2", "phase": 5, "loss": 2.697314113378525, "ce_loss": 2.684768259525299, "logit_z_loss": 28.13218903541565, "router_aux_loss": 1.1947495341300964, "router_z_loss": 0.598373532295227, "routing_entropy_normalized": 0.6577776415370906, "learning_rate": 0.00065696578689, "grad_norm": 1.1601126194000244, "tokens_per_second": 4193387.0005271714, "eta_seconds": 665.1454205513, "data_wait_seconds": 2.2739171981811523e-05} {"step": 27290, "total_steps": 28610, "tokens": 57231278080, "source": "fineweb_edu", "phase": 5, "loss": 3.3681235313415527, "ce_loss": 3.355970174074173, "logit_z_loss": 25.04009485244751, "router_aux_loss": 1.1735046356916428, "router_z_loss": 0.41834935918450356, "routing_entropy_normalized": 0.6899626958547367, "learning_rate": 0.0006479028799032665, "grad_norm": 1.1514455080032349, "tokens_per_second": 4563739.935056115, "eta_seconds": 606.5728282928467, "data_wait_seconds": 2.1725893020629883e-05} {"step": 27300, "total_steps": 28610, "tokens": 57252249600, "source": "cosmopedia_v2", "phase": 5, "loss": 2.697975993156433, "ce_loss": 2.685825824737549, "logit_z_loss": 28.32142186164856, "router_aux_loss": 1.152631938457489, "router_z_loss": 0.6238861307501793, "routing_entropy_normalized": 0.7051103679233057, "learning_rate": 0.0006388856682991034, "grad_norm": 1.2602440118789673, "tokens_per_second": 3905760.372682683, "eta_seconds": 703.3890607357025, "data_wait_seconds": 2.0176172256469727e-05} {"step": 27310, "total_steps": 28610, "tokens": 57273221120, "source": "npset2", "phase": 5, "loss": 1.5471699982881546, "ce_loss": 1.535287082195282, "logit_z_loss": 40.03988027572632, "router_aux_loss": 1.1413107365369797, "router_z_loss": 0.4698137529194355, "routing_entropy_normalized": 0.6981074530078091, "learning_rate": 0.000629914635643203, "grad_norm": 0.8176607489585876, "tokens_per_second": 3504955.8867586073, "eta_seconds": 777.8407740592957, "data_wait_seconds": 2.2232532501220703e-05} {"step": 27320, "total_steps": 28610, "tokens": 57294192640, "source": "fineweb_edu", "phase": 5, "loss": 3.3347651660442352, "ce_loss": 3.323015868663788, "logit_z_loss": 24.890592575073242, "router_aux_loss": 1.1315443217754364, "router_z_loss": 0.4339235797524452, "routing_entropy_normalized": 0.7069728879197142, "learning_rate": 0.000620990263024821, "grad_norm": 1.1690829992294312, "tokens_per_second": 4539463.978479164, "eta_seconds": 595.9571642875671, "data_wait_seconds": 2.014636993408203e-05} {"step": 27330, "total_steps": 28610, "tokens": 57315164160, "source": "dclm", "phase": 5, "loss": 3.4685001373291016, "ce_loss": 3.456776410341263, "logit_z_loss": 25.152488231658936, "router_aux_loss": 1.1367897987365723, "router_z_loss": 0.3557678982615471, "routing_entropy_normalized": 0.668205205625685, "learning_rate": 0.000612113029030977, "grad_norm": 1.41571843624115, "tokens_per_second": 4204147.246170723, "eta_seconds": 638.5015563964843, "data_wait_seconds": 1.9818544387817383e-05} {"step": 27340, "total_steps": 28610, "tokens": 57336135680, "source": "finemath", "phase": 5, "loss": 2.173094943165779, "ce_loss": 2.1611558198928833, "logit_z_loss": 36.4738974571228, "router_aux_loss": 1.166327714920044, "router_z_loss": 0.275822164490819, "routing_entropy_normalized": 0.6623565638146336, "learning_rate": 0.0006032834097207889, "grad_norm": 0.47295820713043213, "tokens_per_second": 3956775.795425879, "eta_seconds": 673.1195240020752, "data_wait_seconds": 2.136826515197754e-05} {"step": 27350, "total_steps": 28610, "tokens": 57357107200, "source": "dclm", "phase": 5, "loss": 3.512218713760376, "ce_loss": 3.5002668499946594, "logit_z_loss": 25.425736665725708, "router_aux_loss": 1.1613765507936478, "router_z_loss": 0.33813730627298355, "routing_entropy_normalized": 0.6550750432951279, "learning_rate": 0.0005945018785999439, "grad_norm": 0.3904390335083008, "tokens_per_second": 4222553.999950555, "eta_seconds": 625.7851338386536, "data_wait_seconds": 2.199411392211914e-05} {"step": 27360, "total_steps": 28610, "tokens": 57378078720, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7119091153144836, "ce_loss": 2.699716478586197, "logit_z_loss": 27.793972492218018, "router_aux_loss": 1.1521059572696686, "router_z_loss": 0.6715151593089104, "routing_entropy_normalized": 0.6942042894758622, "learning_rate": 0.0005857689065953056, "grad_norm": 2.734346866607666, "tokens_per_second": 4051173.6775461975, "eta_seconds": 647.0816135406494, "data_wait_seconds": 1.9729137420654297e-05} {"step": 27370, "total_steps": 28610, "tokens": 57399050240, "source": "finemath", "phase": 5, "loss": 2.120582640171051, "ce_loss": 2.1086553931236267, "logit_z_loss": 36.30908918380737, "router_aux_loss": 1.1621994525194168, "router_z_loss": 0.30525220930576324, "routing_entropy_normalized": 0.6844714568947154, "learning_rate": 0.0005770849620296591, "grad_norm": 0.7781955003738403, "tokens_per_second": 3567065.8765166714, "eta_seconds": 729.0217142105103, "data_wait_seconds": 2.276897430419922e-05} {"step": 27380, "total_steps": 28610, "tokens": 57420021760, "source": "cosmopedia_v2", "phase": 5, "loss": 2.701281398534775, "ce_loss": 2.6891345381736755, "logit_z_loss": 28.288653135299683, "router_aux_loss": 1.145484060049057, "router_z_loss": 0.6920041218400002, "routing_entropy_normalized": 0.6485221301872393, "learning_rate": 0.0005684505105965974, "grad_norm": 1.1340293884277344, "tokens_per_second": 4191959.111678167, "eta_seconds": 615.3440172672272, "data_wait_seconds": 2.4318695068359375e-05} {"step": 27390, "total_steps": 28610, "tokens": 57440993280, "source": "fineweb_edu", "phase": 5, "loss": 3.3724576234817505, "ce_loss": 3.3607043623924255, "logit_z_loss": 24.385291576385498, "router_aux_loss": 1.13286592066288, "router_z_loss": 0.42461635917425156, "routing_entropy_normalized": 0.7101786999377885, "learning_rate": 0.0005598660153355462, "grad_norm": 0.7322582602500916, "tokens_per_second": 4155169.577738428, "eta_seconds": 615.7451319694519, "data_wait_seconds": 2.1010637283325195e-05} {"step": 27400, "total_steps": 28610, "tokens": 57461964800, "source": "cosmopedia_v2", "phase": 5, "loss": 3.066123515367508, "ce_loss": 3.0538840293884277, "logit_z_loss": 26.53302240371704, "router_aux_loss": 1.1574629098176956, "router_z_loss": 0.6648423671722412, "routing_entropy_normalized": 0.6958644045276065, "learning_rate": 0.0005513319366069342, "grad_norm": 2.1996982097625732, "tokens_per_second": 3933601.4657241865, "eta_seconds": 645.096851348877, "data_wait_seconds": 2.2351741790771484e-05} {"step": 27410, "total_steps": 28610, "tokens": 57482936320, "source": "npset2", "phase": 5, "loss": 1.7342600226402283, "ce_loss": 1.7217050939798355, "logit_z_loss": 36.7751145362854, "router_aux_loss": 1.2107395231723785, "router_z_loss": 0.44751112535595894, "routing_entropy_normalized": 0.688461323638076, "learning_rate": 0.0005428487320675038, "grad_norm": 0.9691622257232666, "tokens_per_second": 3902544.8761108173, "eta_seconds": 644.8567485809326, "data_wait_seconds": 2.104043960571289e-05} {"step": 27420, "total_steps": 28610, "tokens": 57503907840, "source": "fineweb_edu", "phase": 5, "loss": 3.5529070794582367, "ce_loss": 3.540081173181534, "logit_z_loss": 23.973432302474976, "router_aux_loss": 1.2377067357301712, "router_z_loss": 0.4488038830459118, "routing_entropy_normalized": 0.663651910028144, "learning_rate": 0.0005344168566457695, "grad_norm": 0.4208899438381195, "tokens_per_second": 4351275.951286408, "eta_seconds": 573.5354199409485, "data_wait_seconds": 1.9997358322143555e-05} {"step": 27430, "total_steps": 28610, "tokens": 57524879360, "source": "dclm", "phase": 5, "loss": 3.6620669662952423, "ce_loss": 3.649735450744629, "logit_z_loss": 24.382521152496338, "router_aux_loss": 1.1981806010007858, "router_z_loss": 0.34979190677404404, "routing_entropy_normalized": 0.6724229517281588, "learning_rate": 0.0005260367625176202, "grad_norm": 0.61320561170578, "tokens_per_second": 3859270.369658558, "eta_seconds": 641.219485282898, "data_wait_seconds": 1.767277717590332e-05} {"step": 27440, "total_steps": 28610, "tokens": 57545850880, "source": "finemath", "phase": 5, "loss": 2.1695883721113205, "ce_loss": 2.1575977951288223, "logit_z_loss": 36.49573564529419, "router_aux_loss": 1.1722975671291351, "router_z_loss": 0.2676702532917261, "routing_entropy_normalized": 0.7403342219565815, "learning_rate": 0.0005177088990820725, "grad_norm": 0.5980548858642578, "tokens_per_second": 4184069.8326085336, "eta_seconds": 586.4309005737305, "data_wait_seconds": 1.946091651916504e-05} {"step": 27450, "total_steps": 28610, "tokens": 57566822400, "source": "dclm", "phase": 5, "loss": 3.6363372802734375, "ce_loss": 3.62434184551239, "logit_z_loss": 24.358548164367676, "router_aux_loss": 1.164070188999176, "router_z_loss": 0.35477619618177414, "routing_entropy_normalized": 0.7508924626149794, "learning_rate": 0.0005094337129371681, "grad_norm": 0.6511759161949158, "tokens_per_second": 4112515.9669025447, "eta_seconds": 591.5348024368286, "data_wait_seconds": 2.440810203552246e-05} {"step": 27460, "total_steps": 28610, "tokens": 57587793920, "source": "cosmopedia_v2", "phase": 5, "loss": 2.8173894584178925, "ce_loss": 2.805150091648102, "logit_z_loss": 27.46114730834961, "router_aux_loss": 1.1602921336889267, "router_z_loss": 0.6364258527755737, "routing_entropy_normalized": 0.7112096871838839, "learning_rate": 0.0005012116478560263, "grad_norm": 1.9540283679962158, "tokens_per_second": 3596758.201289896, "eta_seconds": 670.5273652076721, "data_wait_seconds": 2.3305416107177734e-05} {"step": 27470, "total_steps": 28610, "tokens": 57608765440, "source": "finemath", "phase": 5, "loss": 2.1544594168663025, "ce_loss": 2.1426647901535034, "logit_z_loss": 35.92954611778259, "router_aux_loss": 1.1465794295072556, "router_z_loss": 0.3288428373634815, "routing_entropy_normalized": 0.7780944910776741, "learning_rate": 0.0004930431447630445, "grad_norm": 0.5189215540885925, "tokens_per_second": 4303604.344132448, "eta_seconds": 555.5234842300415, "data_wait_seconds": 2.1666288375854492e-05} {"step": 27480, "total_steps": 28610, "tokens": 57629736960, "source": "cosmopedia_v2", "phase": 5, "loss": 2.8108617663383484, "ce_loss": 2.7987828254699707, "logit_z_loss": 27.06228256225586, "router_aux_loss": 1.1470867544412613, "router_z_loss": 0.6081330478191376, "routing_entropy_normalized": 0.7456178443489713, "learning_rate": 0.0004849286417102539, "grad_norm": 0.6440252661705017, "tokens_per_second": 3676986.514099815, "eta_seconds": 644.4901962280273, "data_wait_seconds": 1.817941665649414e-05} {"step": 27490, "total_steps": 28610, "tokens": 57650708480, "source": "fineweb_edu", "phase": 5, "loss": 3.445550709962845, "ce_loss": 3.4333740770816803, "logit_z_loss": 24.65852451324463, "router_aux_loss": 1.1776469945907593, "router_z_loss": 0.40017206221818924, "routing_entropy_normalized": 0.6845603437335309, "learning_rate": 0.00047686857385382674, "grad_norm": 1.757568597793579, "tokens_per_second": 3689267.872285071, "eta_seconds": 636.6602592468262, "data_wait_seconds": 1.862645149230957e-05} {"step": 27500, "total_steps": 28610, "tokens": 57671680000, "source": "cosmopedia_v2", "phase": 5, "loss": 2.778052270412445, "ce_loss": 2.7656134963035583, "logit_z_loss": 27.1903395652771, "router_aux_loss": 1.1812237650156021, "router_z_loss": 0.6265436410903931, "routing_entropy_normalized": 0.700057633968754, "learning_rate": 0.000468863373430741, "grad_norm": 0.6669081449508667, "tokens_per_second": 4192282.1750717796, "eta_seconds": 555.2676615715027, "data_wait_seconds": 2.205371856689453e-05} {"step": 27510, "total_steps": 28610, "tokens": 57692651520, "source": "npset2", "phase": 5, "loss": 1.5720415860414505, "ce_loss": 1.5598599314689636, "logit_z_loss": 38.37356901168823, "router_aux_loss": 1.1775330305099487, "router_z_loss": 0.4063064828515053, "routing_entropy_normalized": 0.6838369700740539, "learning_rate": 0.0004609134697356009, "grad_norm": 0.5728282332420349, "tokens_per_second": 4627358.5007601185, "eta_seconds": 498.5278749465942, "data_wait_seconds": 2.0891427993774414e-05} {"step": 27520, "total_steps": 28610, "tokens": 57713623040, "source": "fineweb_edu", "phase": 5, "loss": 3.38229900598526, "ce_loss": 3.370233654975891, "logit_z_loss": 24.675041913986206, "router_aux_loss": 1.1649102419614792, "router_z_loss": 0.41629115864634514, "routing_entropy_normalized": 0.724451127390397, "learning_rate": 0.0004530192890976151, "grad_norm": 1.1351481676101685, "tokens_per_second": 3849666.519909384, "eta_seconds": 593.7905707359314, "data_wait_seconds": 2.4646520614624023e-05} {"step": 27530, "total_steps": 28610, "tokens": 57734594560, "source": "dclm", "phase": 5, "loss": 3.557970643043518, "ce_loss": 3.5460150241851807, "logit_z_loss": 24.621973991394043, "router_aux_loss": 1.1624541282653809, "router_z_loss": 0.3310382701456547, "routing_entropy_normalized": 0.7178517426099329, "learning_rate": 0.00044518125485773334, "grad_norm": 0.7012308239936829, "tokens_per_second": 3791067.2617297834, "eta_seconds": 597.4370813369751, "data_wait_seconds": 2.199411392211914e-05} {"step": 27540, "total_steps": 28610, "tokens": 57755566080, "source": "finemath", "phase": 5, "loss": 2.0547412633895874, "ce_loss": 2.042922481894493, "logit_z_loss": 36.06462812423706, "router_aux_loss": 1.1562795490026474, "router_z_loss": 0.25598170794546604, "routing_entropy_normalized": 0.7469674111532102, "learning_rate": 0.00043739978734594494, "grad_norm": 0.4908483326435089, "tokens_per_second": 4073321.7115883483, "eta_seconds": 550.8901085853577, "data_wait_seconds": 1.9609928131103516e-05} {"step": 27550, "total_steps": 28610, "tokens": 57776537600, "source": "dclm", "phase": 5, "loss": 3.564127117395401, "ce_loss": 3.5523250102996826, "logit_z_loss": 24.759324550628662, "router_aux_loss": 1.1525534242391586, "router_z_loss": 0.27655231580138206, "routing_entropy_normalized": 0.7257696931146084, "learning_rate": 0.00042967530385873697, "grad_norm": 0.47635790705680847, "tokens_per_second": 3970219.794752279, "eta_seconds": 559.9138674736023, "data_wait_seconds": 2.276897430419922e-05} {"step": 27560, "total_steps": 28610, "tokens": 57797509120, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7337030470371246, "ce_loss": 2.721694737672806, "logit_z_loss": 27.768359899520874, "router_aux_loss": 1.142986610531807, "router_z_loss": 0.5784645825624466, "routing_entropy_normalized": 0.691453704430512, "learning_rate": 0.00042200821863671667, "grad_norm": 1.1140912771224976, "tokens_per_second": 3219577.010841031, "eta_seconds": 683.9437580108643, "data_wait_seconds": 2.187490463256836e-05} {"step": 27570, "total_steps": 28610, "tokens": 57818480640, "source": "finemath", "phase": 5, "loss": 2.1779327243566513, "ce_loss": 2.1662165820598602, "logit_z_loss": 36.34190368652344, "router_aux_loss": 1.1488913595676422, "router_z_loss": 0.22726093418896198, "routing_entropy_normalized": 0.7303074859835604, "learning_rate": 0.0004143989428423947, "grad_norm": 0.3000516891479492, "tokens_per_second": 4001469.1105673467, "eta_seconds": 545.0593318939209, "data_wait_seconds": 2.3305416107177734e-05} {"step": 27580, "total_steps": 28610, "tokens": 57839452160, "source": "cosmopedia_v2", "phase": 5, "loss": 2.767499625682831, "ce_loss": 2.755429983139038, "logit_z_loss": 27.709960460662842, "router_aux_loss": 1.14735846221447, "router_z_loss": 0.5960491895675659, "routing_entropy_normalized": 0.7326848285492075, "learning_rate": 0.0004068478845381405, "grad_norm": 0.584567666053772, "tokens_per_second": 4079096.452688106, "eta_seconds": 529.5453503131866, "data_wait_seconds": 1.9341707229614258e-05} {"step": 27590, "total_steps": 28610, "tokens": 57860423680, "source": "fineweb_edu", "phase": 5, "loss": 3.372015208005905, "ce_loss": 3.360063314437866, "logit_z_loss": 24.364545345306396, "router_aux_loss": 1.1558517664670944, "router_z_loss": 0.3934008814394474, "routing_entropy_normalized": 0.7002400413831071, "learning_rate": 0.0003993554486642944, "grad_norm": 1.0940029621124268, "tokens_per_second": 3615408.094000932, "eta_seconds": 591.6607432365417, "data_wait_seconds": 1.767277717590332e-05} {"step": 27600, "total_steps": 28610, "tokens": 57881395200, "source": "cosmopedia_v2", "phase": 5, "loss": 2.734100729227066, "ce_loss": 2.7219743728637695, "logit_z_loss": 28.118155002593994, "router_aux_loss": 1.1536212265491486, "router_z_loss": 0.590161457657814, "routing_entropy_normalized": 0.7194508000111869, "learning_rate": 0.0003919220370174532, "grad_norm": 1.1817651987075806, "tokens_per_second": 4327653.33472667, "eta_seconds": 489.43927717208857, "data_wait_seconds": 1.77919864654541e-05} {"step": 27610, "total_steps": 28610, "tokens": 57902366720, "source": "npset2", "phase": 5, "loss": 1.2614512294530869, "ce_loss": 1.2496893517673016, "logit_z_loss": 40.086236000061035, "router_aux_loss": 1.143567681312561, "router_z_loss": 0.3262030649930239, "routing_entropy_normalized": 0.7234530513302632, "learning_rate": 0.0003845480482289242, "grad_norm": 0.7926315069198608, "tokens_per_second": 4035048.133682905, "eta_seconds": 519.7340726852417, "data_wait_seconds": 1.9729137420654297e-05} {"step": 27620, "total_steps": 28610, "tokens": 57923338240, "source": "fineweb_edu", "phase": 5, "loss": 3.4163236916065216, "ce_loss": 3.4046172499656677, "logit_z_loss": 24.13103175163269, "router_aux_loss": 1.1362032890319824, "router_z_loss": 0.3443964868783951, "routing_entropy_normalized": 0.7433476527433595, "learning_rate": 0.0003772338777433482, "grad_norm": 0.7691269516944885, "tokens_per_second": 3953571.565207194, "eta_seconds": 525.1404826641083, "data_wait_seconds": 1.901388168334961e-05} {"step": 27630, "total_steps": 28610, "tokens": 57944309760, "source": "dclm", "phase": 5, "loss": 3.551723897457123, "ce_loss": 3.540062725543976, "logit_z_loss": 24.651623249053955, "router_aux_loss": 1.136534333229065, "router_z_loss": 0.2958680745214224, "routing_entropy_normalized": 0.7622628424578234, "learning_rate": 0.00036997991779749115, "grad_norm": 0.4423021376132965, "tokens_per_second": 4340584.122259875, "eta_seconds": 473.4867248535156, "data_wait_seconds": 2.199411392211914e-05} {"step": 27640, "total_steps": 28610, "tokens": 57965281280, "source": "finemath", "phase": 5, "loss": 2.150364711880684, "ce_loss": 2.1388886272907257, "logit_z_loss": 36.80145883560181, "router_aux_loss": 1.1270176768302917, "router_z_loss": 0.20588162541389465, "routing_entropy_normalized": 0.762633322775464, "learning_rate": 0.0003627865573992087, "grad_norm": 0.4371199905872345, "tokens_per_second": 3830269.065808735, "eta_seconds": 531.095180273056, "data_wait_seconds": 1.6570091247558594e-05} {"step": 27650, "total_steps": 28610, "tokens": 57986252800, "source": "dclm", "phase": 5, "loss": 3.4955612421035767, "ce_loss": 3.4838794469833374, "logit_z_loss": 25.015810251235962, "router_aux_loss": 1.1396625339984894, "router_z_loss": 0.2851655278354883, "routing_entropy_normalized": 0.7506955811748501, "learning_rate": 0.0003556541823065897, "grad_norm": 0.4225808084011078, "tokens_per_second": 3739549.1196477683, "eta_seconds": 538.3712997436523, "data_wait_seconds": 1.633167266845703e-05} {"step": 27660, "total_steps": 28610, "tokens": 58007224320, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7223654091358185, "ce_loss": 2.710407465696335, "logit_z_loss": 27.782416582107544, "router_aux_loss": 1.1393316090106964, "router_z_loss": 0.5645734593272209, "routing_entropy_normalized": 0.7182731258910863, "learning_rate": 0.0003485831750072649, "grad_norm": 0.3914169669151306, "tokens_per_second": 4182727.440512197, "eta_seconds": 476.31466031074524, "data_wait_seconds": 2.3573637008666992e-05} {"step": 27670, "total_steps": 28610, "tokens": 58028195840, "source": "finemath", "phase": 5, "loss": 2.1604300886392593, "ce_loss": 2.1488812416791916, "logit_z_loss": 37.02727127075195, "router_aux_loss": 1.1365265995264053, "router_z_loss": 0.18357402086257935, "routing_entropy_normalized": 0.7678460878578244, "learning_rate": 0.0003415739146978952, "grad_norm": 0.5129936337471008, "tokens_per_second": 3729616.446420361, "eta_seconds": 528.5591449737549, "data_wait_seconds": 1.8328428268432617e-05} {"step": 27680, "total_steps": 28610, "tokens": 58049167360, "source": "cosmopedia_v2", "phase": 5, "loss": 2.717371940612793, "ce_loss": 2.7055833637714386, "logit_z_loss": 27.751835107803345, "router_aux_loss": 1.1230333000421524, "router_z_loss": 0.5583273842930794, "routing_entropy_normalized": 0.7408505607367132, "learning_rate": 0.0003346267772638392, "grad_norm": 0.29500848054885864, "tokens_per_second": 4101095.80167952, "eta_seconds": 475.5683491230011, "data_wait_seconds": 2.372264862060547e-05} {"step": 27690, "total_steps": 28610, "tokens": 58070138880, "source": "fineweb_edu", "phase": 5, "loss": 3.338480919599533, "ce_loss": 3.326885849237442, "logit_z_loss": 24.668848514556885, "router_aux_loss": 1.1261379271745682, "router_z_loss": 0.333767581731081, "routing_entropy_normalized": 0.7506478305453763, "learning_rate": 0.00032774213525899493, "grad_norm": 1.0992536544799805, "tokens_per_second": 3938926.2519126274, "eta_seconds": 489.82380390167236, "data_wait_seconds": 2.3156404495239258e-05} {"step": 27700, "total_steps": 28610, "tokens": 58091110400, "source": "cosmopedia_v2", "phase": 5, "loss": 2.714723616838455, "ce_loss": 2.7029493153095245, "logit_z_loss": 27.913459539413452, "router_aux_loss": 1.1235522478818893, "router_z_loss": 0.5387637987732887, "routing_entropy_normalized": 0.7353492922296287, "learning_rate": 0.00032092035788581905, "grad_norm": 0.39336878061294556, "tokens_per_second": 4516121.640423276, "eta_seconds": 422.5768196582794, "data_wait_seconds": 2.1845102310180664e-05} {"step": 27710, "total_steps": 28610, "tokens": 58112081920, "source": "npset2", "phase": 5, "loss": 1.51713465154171, "ce_loss": 1.5055192708969116, "logit_z_loss": 39.36193656921387, "router_aux_loss": 1.126556396484375, "router_z_loss": 0.34981443360447884, "routing_entropy_normalized": 0.71747869100081, "learning_rate": 0.00031416181097552846, "grad_norm": 0.7251936197280884, "tokens_per_second": 3673098.7623040825, "eta_seconds": 513.8540840148926, "data_wait_seconds": 1.8090009689331055e-05} {"step": 27720, "total_steps": 28610, "tokens": 58133053440, "source": "fineweb_edu", "phase": 5, "loss": 3.383846491575241, "ce_loss": 3.371325343847275, "logit_z_loss": 22.65258002281189, "router_aux_loss": 1.2194788306951523, "router_z_loss": 0.3263409584760666, "routing_entropy_normalized": 0.6414140431461011, "learning_rate": 0.0003074668569684829, "grad_norm": 3.9371461868286133, "tokens_per_second": 3904019.5762621253, "eta_seconds": 478.08809447288513, "data_wait_seconds": 2.2917985916137695e-05} {"step": 27730, "total_steps": 28610, "tokens": 58154024960, "source": "dclm", "phase": 5, "loss": 3.4980911016464233, "ce_loss": 3.4848386645317078, "logit_z_loss": 24.763427257537842, "router_aux_loss": 1.3017573356628418, "router_z_loss": 0.23484240472316742, "routing_entropy_normalized": 0.6167399530235155, "learning_rate": 0.0003008358548947486, "grad_norm": 1.8664525747299194, "tokens_per_second": 4210482.766861056, "eta_seconds": 438.3093013763427, "data_wait_seconds": 2.053380012512207e-05} {"step": 27740, "total_steps": 28610, "tokens": 58174996480, "source": "finemath", "phase": 5, "loss": 2.1135820001363754, "ce_loss": 2.1014843434095383, "logit_z_loss": 37.0741286277771, "router_aux_loss": 1.1896660178899765, "router_z_loss": 0.2010026052594185, "routing_entropy_normalized": 0.6748067982159978, "learning_rate": 0.00029426916035484165, "grad_norm": 0.6280632019042969, "tokens_per_second": 3610586.0072669047, "eta_seconds": 505.32579374313354, "data_wait_seconds": 1.665949821472168e-05} {"step": 27750, "total_steps": 28610, "tokens": 58195968000, "source": "dclm", "phase": 5, "loss": 3.49196395277977, "ce_loss": 3.4802332520484924, "logit_z_loss": 24.838901042938232, "router_aux_loss": 1.1450781673192978, "router_z_loss": 0.27989829145371914, "routing_entropy_normalized": 0.7085990260586701, "learning_rate": 0.0002877671255006612, "grad_norm": 0.48371610045433044, "tokens_per_second": 4035806.2627049014, "eta_seconds": 446.88733863830566, "data_wait_seconds": 2.321600914001465e-05} {"step": 27760, "total_steps": 28610, "tokens": 58216939520, "source": "cosmopedia_v2", "phase": 5, "loss": 2.701373368501663, "ce_loss": 2.6894377768039703, "logit_z_loss": 27.893439531326294, "router_aux_loss": 1.1392162442207336, "router_z_loss": 0.5434527173638344, "routing_entropy_normalized": 0.7031680987134749, "learning_rate": 0.00028133009901660497, "grad_norm": 0.6565626263618469, "tokens_per_second": 3870674.6393131325, "eta_seconds": 460.53449749946594, "data_wait_seconds": 2.2798776626586914e-05} {"step": 27770, "total_steps": 28610, "tokens": 58237911040, "source": "finemath", "phase": 5, "loss": 2.2587705552577972, "ce_loss": 2.2472039461135864, "logit_z_loss": 35.48642158508301, "router_aux_loss": 1.1383630335330963, "router_z_loss": 0.18299024552106857, "routing_entropy_normalized": 0.7165015638687376, "learning_rate": 0.0002749584261008676, "grad_norm": 0.24051184952259064, "tokens_per_second": 3791855.4695031308, "eta_seconds": 464.5766944885254, "data_wait_seconds": 2.244114875793457e-05} {"step": 27780, "total_steps": 28610, "tokens": 58258882560, "source": "cosmopedia_v2", "phase": 5, "loss": 2.705506354570389, "ce_loss": 2.6936777532100677, "logit_z_loss": 27.664756298065186, "router_aux_loss": 1.1251948922872543, "router_z_loss": 0.5766317918896675, "routing_entropy_normalized": 0.75516974607052, "learning_rate": 0.00026865244844692996, "grad_norm": 0.4140530228614807, "tokens_per_second": 3642861.6248250026, "eta_seconds": 477.82110309600836, "data_wait_seconds": 1.9788742065429688e-05} {"step": 27790, "total_steps": 28610, "tokens": 58279854080, "source": "fineweb_edu", "phase": 5, "loss": 3.3298254907131195, "ce_loss": 3.3182373344898224, "logit_z_loss": 24.252888202667236, "router_aux_loss": 1.1213261038064957, "router_z_loss": 0.3748525194823742, "routing_entropy_normalized": 0.7373438245623659, "learning_rate": 0.00026241250422523585, "grad_norm": 0.4339379668235779, "tokens_per_second": 4240435.0527585335, "eta_seconds": 405.53967189788824, "data_wait_seconds": 2.0444393157958984e-05} {"step": 27800, "total_steps": 28610, "tokens": 58300825600, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7063998579978943, "ce_loss": 2.6945784986019135, "logit_z_loss": 27.460379362106323, "router_aux_loss": 1.1271987110376358, "router_z_loss": 0.5493593290448189, "routing_entropy_normalized": 0.7464171700833936, "learning_rate": 0.00025623892806505774, "grad_norm": 0.9743544459342957, "tokens_per_second": 3802096.249596766, "eta_seconds": 446.77804255485535, "data_wait_seconds": 2.5600194931030273e-05} {"step": 27810, "total_steps": 28610, "tokens": 58321797120, "source": "npset2", "phase": 5, "loss": 1.6617633402347565, "ce_loss": 1.6501868814229965, "logit_z_loss": 38.33955669403076, "router_aux_loss": 1.1166395992040634, "router_z_loss": 0.4100771136581898, "routing_entropy_normalized": 0.7689260357192811, "learning_rate": 0.0002501320510365484, "grad_norm": 0.5131340026855469, "tokens_per_second": 4103158.0792142767, "eta_seconds": 408.885440826416, "data_wait_seconds": 2.288818359375e-05} {"step": 27820, "total_steps": 28610, "tokens": 58342768640, "source": "fineweb_edu", "phase": 5, "loss": 3.369535267353058, "ce_loss": 3.357931226491928, "logit_z_loss": 24.21927046775818, "router_aux_loss": 1.1260649859905243, "router_z_loss": 0.3433970510959625, "routing_entropy_normalized": 0.7448027472894088, "learning_rate": 0.00024409220063298953, "grad_norm": 0.25202175974845886, "tokens_per_second": 3793870.377210933, "eta_seconds": 436.6912717819214, "data_wait_seconds": 2.1785497665405273e-05} {"step": 27830, "total_steps": 28610, "tokens": 58363740160, "source": "dclm", "phase": 5, "loss": 3.4526363015174866, "ce_loss": 3.441180497407913, "logit_z_loss": 24.910789728164673, "router_aux_loss": 1.114931970834732, "router_z_loss": 0.306452177464962, "routing_entropy_normalized": 0.783133231510063, "learning_rate": 0.00023811970075322803, "grad_norm": 0.3993011713027954, "tokens_per_second": 3991176.0469389283, "eta_seconds": 409.8487615585327, "data_wait_seconds": 2.047419548034668e-05} {"step": 27840, "total_steps": 28610, "tokens": 58384711680, "source": "finemath", "phase": 5, "loss": 2.1693107187747955, "ce_loss": 2.1578984558582306, "logit_z_loss": 36.06208324432373, "router_aux_loss": 1.120962604880333, "router_z_loss": 0.20255792699754238, "routing_entropy_normalized": 0.7675690151741489, "learning_rate": 0.0002322148716843081, "grad_norm": 0.5056644678115845, "tokens_per_second": 3507412.709255468, "eta_seconds": 460.3983545303345, "data_wait_seconds": 1.7464160919189453e-05} {"step": 27850, "total_steps": 28610, "tokens": 58405683200, "source": "dclm", "phase": 5, "loss": 3.442443937063217, "ce_loss": 3.430951237678528, "logit_z_loss": 24.61332082748413, "router_aux_loss": 1.117176428437233, "router_z_loss": 0.3208998218178749, "routing_entropy_normalized": 0.781347691420392, "learning_rate": 0.00022637803008429102, "grad_norm": 0.41088104248046875, "tokens_per_second": 3940437.760358782, "eta_seconds": 404.4818410873413, "data_wait_seconds": 2.0116567611694336e-05} {"step": 27860, "total_steps": 28610, "tokens": 58426654720, "source": "cosmopedia_v2", "phase": 5, "loss": 2.6912881433963776, "ce_loss": 2.679574340581894, "logit_z_loss": 27.730440616607666, "router_aux_loss": 1.117054060101509, "router_z_loss": 0.5432156659662724, "routing_entropy_normalized": 0.7575764777570495, "learning_rate": 0.00022060948896527872, "grad_norm": 0.777539849281311, "tokens_per_second": 3844382.9060194246, "eta_seconds": 409.13302302360535, "data_wait_seconds": 2.0265579223632812e-05} {"step": 27870, "total_steps": 28610, "tokens": 58447626240, "source": "finemath", "phase": 5, "loss": 2.1718658953905106, "ce_loss": 2.160458266735077, "logit_z_loss": 35.50574445724487, "router_aux_loss": 1.118214949965477, "router_z_loss": 0.2255100030452013, "routing_entropy_normalized": 0.7803306428710973, "learning_rate": 0.00021490955767662447, "grad_norm": 0.5993155241012573, "tokens_per_second": 3974727.4993909867, "eta_seconds": 390.439968585968, "data_wait_seconds": 1.6361474990844727e-05} {"step": 27880, "total_steps": 28610, "tokens": 58468597760, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7010094821453094, "ce_loss": 2.6892242431640625, "logit_z_loss": 27.78027868270874, "router_aux_loss": 1.122684195637703, "router_z_loss": 0.5583646595478058, "routing_entropy_normalized": 0.7400205049227344, "learning_rate": 0.00020927854188834507, "grad_norm": 1.039543628692627, "tokens_per_second": 3853110.0227408186, "eta_seconds": 397.32085275650024, "data_wait_seconds": 1.811981201171875e-05} {"step": 27890, "total_steps": 28610, "tokens": 58489569280, "source": "fineweb_edu", "phase": 5, "loss": 3.3493749499320984, "ce_loss": 3.3379043638706207, "logit_z_loss": 24.344539403915405, "router_aux_loss": 1.1134139150381088, "router_z_loss": 0.3364297114312649, "routing_entropy_normalized": 0.7723528111205986, "learning_rate": 0.0002037167435747279, "grad_norm": 0.2564642131328583, "tokens_per_second": 4069246.321748795, "eta_seconds": 371.0636615753174, "data_wait_seconds": 2.378225326538086e-05} {"step": 27900, "total_steps": 28610, "tokens": 58510540800, "source": "cosmopedia_v2", "phase": 5, "loss": 2.695162355899811, "ce_loss": 2.6831369400024414, "logit_z_loss": 27.76621913909912, "router_aux_loss": 1.14834263920784, "router_z_loss": 0.541979543864727, "routing_entropy_normalized": 0.6718089782434407, "learning_rate": 0.00019822446099813685, "grad_norm": 0.5913159847259521, "tokens_per_second": 4416124.579488824, "eta_seconds": 337.1684591770172, "data_wait_seconds": 1.9729137420654297e-05} {"step": 27910, "total_steps": 28610, "tokens": 58531512320, "source": "npset2", "phase": 5, "loss": 1.6481776684522629, "ce_loss": 1.6365756690502167, "logit_z_loss": 36.79128694534302, "router_aux_loss": 1.1228550970554352, "router_z_loss": 0.37343648076057434, "routing_entropy_normalized": 0.7884279792222472, "learning_rate": 0.0001928019886930197, "grad_norm": 0.670244038105011, "tokens_per_second": 3944093.5681051575, "eta_seconds": 372.20374584198, "data_wait_seconds": 1.671910285949707e-05} {"step": 27920, "total_steps": 28610, "tokens": 58552483840, "source": "fineweb_edu", "phase": 5, "loss": 3.3459804356098175, "ce_loss": 3.3339628875255585, "logit_z_loss": 24.279399394989014, "router_aux_loss": 1.1681936830282211, "router_z_loss": 0.3355780802667141, "routing_entropy_normalized": 0.6368045870925783, "learning_rate": 0.00018744961745010842, "grad_norm": 0.5130188465118408, "tokens_per_second": 4361349.895019809, "eta_seconds": 331.78601002693176, "data_wait_seconds": 2.345442771911621e-05} {"step": 27930, "total_steps": 28610, "tokens": 58573455360, "source": "dclm", "phase": 5, "loss": 3.5838514864444733, "ce_loss": 3.57213494181633, "logit_z_loss": 24.150202989578247, "router_aux_loss": 1.1447164416313171, "router_z_loss": 0.26937744580209255, "routing_entropy_normalized": 0.7283975053459159, "learning_rate": 0.00018216763430082983, "grad_norm": 1.0059657096862793, "tokens_per_second": 4125485.966980515, "eta_seconds": 345.6716060638428, "data_wait_seconds": 2.4378299713134766e-05} {"step": 27940, "total_steps": 28610, "tokens": 58594426880, "source": "finemath", "phase": 5, "loss": 2.2731394469738007, "ce_loss": 2.261336535215378, "logit_z_loss": 34.64069843292236, "router_aux_loss": 1.161469280719757, "router_z_loss": 0.18824726343154907, "routing_entropy_normalized": 0.742558525597635, "learning_rate": 0.00017695632250191003, "grad_norm": 0.8087918758392334, "tokens_per_second": 4058464.8166556233, "eta_seconds": 346.2126452922821, "data_wait_seconds": 1.7017126083374023e-05} {"step": 27950, "total_steps": 28610, "tokens": 58615398400, "source": "dclm", "phase": 5, "loss": 3.752820521593094, "ce_loss": 3.7409128546714783, "logit_z_loss": 21.282929182052612, "router_aux_loss": 1.1658814400434494, "router_z_loss": 0.24882402829825878, "routing_entropy_normalized": 0.7707282105474644, "learning_rate": 0.0001718159615201853, "grad_norm": 0.4362967014312744, "tokens_per_second": 3918430.081378731, "eta_seconds": 353.2333846092224, "data_wait_seconds": 2.0325183868408203e-05} {"step": 27960, "total_steps": 28610, "tokens": 58636369920, "source": "cosmopedia_v2", "phase": 5, "loss": 3.0612345337867737, "ce_loss": 3.0490361154079437, "logit_z_loss": 24.539011001586914, "router_aux_loss": 1.171160340309143, "router_z_loss": 0.4867748022079468, "routing_entropy_normalized": 0.7344704468204011, "learning_rate": 0.00016674682701761496, "grad_norm": 0.7010442018508911, "tokens_per_second": 3943809.3909322484, "eta_seconds": 345.6426680088043, "data_wait_seconds": 1.9431114196777344e-05} {"step": 27970, "total_steps": 28610, "tokens": 58657341440, "source": "finemath", "phase": 5, "loss": 2.265582114458084, "ce_loss": 2.2536762058734894, "logit_z_loss": 34.183842182159424, "router_aux_loss": 1.1686714142560959, "router_z_loss": 0.21920132264494896, "routing_entropy_normalized": 0.7689684164417554, "learning_rate": 0.0001617491908364983, "grad_norm": 0.3635607957839966, "tokens_per_second": 4006828.765778145, "eta_seconds": 334.9724578857422, "data_wait_seconds": 2.0772218704223633e-05} {"step": 27980, "total_steps": 28610, "tokens": 58678312960, "source": "cosmopedia_v2", "phase": 5, "loss": 2.9660346508026123, "ce_loss": 2.953992784023285, "logit_z_loss": 25.201317310333252, "router_aux_loss": 1.1566181927919388, "router_z_loss": 0.47574447840452194, "routing_entropy_normalized": 0.7758726536495058, "learning_rate": 0.0001568233209848966, "grad_norm": 0.3331252634525299, "tokens_per_second": 4150104.5334071624, "eta_seconds": 318.3548147678375, "data_wait_seconds": 2.047419548034668e-05} {"step": 27990, "total_steps": 28610, "tokens": 58699284480, "source": "fineweb_edu", "phase": 5, "loss": 3.5624370872974396, "ce_loss": 3.550575315952301, "logit_z_loss": 21.68261981010437, "router_aux_loss": 1.1555361300706863, "router_z_loss": 0.3063991069793701, "routing_entropy_normalized": 0.8089577601627074, "learning_rate": 0.00015196948162226115, "grad_norm": 0.3019684851169586, "tokens_per_second": 3850933.5929408353, "eta_seconds": 337.6413040161133, "data_wait_seconds": 2.0682811737060547e-05} {"step": 28000, "total_steps": 28610, "tokens": 58720256000, "source": "cosmopedia_v2", "phase": 5, "loss": 2.925056040287018, "ce_loss": 2.9130779206752777, "logit_z_loss": 25.42520260810852, "router_aux_loss": 1.1506331413984299, "router_z_loss": 0.4718184284865856, "routing_entropy_normalized": 0.8409485850295259, "learning_rate": 0.00014718793304526584, "grad_norm": 0.3107222616672516, "tokens_per_second": 3554947.515250123, "eta_seconds": 359.85417914390564, "data_wait_seconds": 2.1785497665405273e-05} {"step": 28010, "total_steps": 28610, "tokens": 58741227520, "source": "npset2", "phase": 5, "loss": 1.5751665085554123, "ce_loss": 1.5632983893156052, "logit_z_loss": 37.706584453582764, "router_aux_loss": 1.1509435176849365, "router_z_loss": 0.3586685322225094, "routing_entropy_normalized": 0.772169756708347, "learning_rate": 0.00014247893167384978, "grad_norm": 0.316896915435791, "tokens_per_second": 4309754.721011988, "eta_seconds": 291.963529586792, "data_wait_seconds": 1.9550323486328125e-05} {"step": 28020, "total_steps": 28610, "tokens": 58762199040, "source": "fineweb_edu", "phase": 5, "loss": 3.5036895275115967, "ce_loss": 3.4918562173843384, "logit_z_loss": 22.20421814918518, "router_aux_loss": 1.1515648663043976, "router_z_loss": 0.31770576909184456, "routing_entropy_normalized": 0.8071559160352636, "learning_rate": 0.00013784273003746567, "grad_norm": 0.44447430968284607, "tokens_per_second": 4509792.633359326, "eta_seconds": 274.36287665367126, "data_wait_seconds": 2.327561378479004e-05} {"step": 28030, "total_steps": 28610, "tokens": 58783170560, "source": "dclm", "phase": 5, "loss": 3.630942702293396, "ce_loss": 3.6191861033439636, "logit_z_loss": 22.50275158882141, "router_aux_loss": 1.1531541794538498, "router_z_loss": 0.2251303791999817, "routing_entropy_normalized": 0.7635803105731761, "learning_rate": 0.0001332795767615369, "grad_norm": 0.28792816400527954, "tokens_per_second": 3671684.0484204683, "eta_seconds": 331.2780032157898, "data_wait_seconds": 2.1904706954956055e-05} {"step": 28040, "total_steps": 28610, "tokens": 58804142080, "source": "finemath", "phase": 5, "loss": 2.242888778448105, "ce_loss": 2.2311703264713287, "logit_z_loss": 34.46228313446045, "router_aux_loss": 1.1514662355184555, "router_z_loss": 0.20383991114795208, "routing_entropy_normalized": 0.8179247004865856, "learning_rate": 0.00012878971655412515, "grad_norm": 0.3706262409687042, "tokens_per_second": 4233341.17597278, "eta_seconds": 282.3719115257263, "data_wait_seconds": 1.6510486602783203e-05} {"step": 28050, "total_steps": 28610, "tokens": 58825113600, "source": "dclm", "phase": 5, "loss": 3.598474830389023, "ce_loss": 3.5867534279823303, "logit_z_loss": 22.78657031059265, "router_aux_loss": 1.15047787129879, "router_z_loss": 0.21663139574229717, "routing_entropy_normalized": 0.8420933047150662, "learning_rate": 0.00012437339019280742, "grad_norm": 0.24644261598587036, "tokens_per_second": 4272765.397790203, "eta_seconds": 274.85832023620605, "data_wait_seconds": 1.7136335372924805e-05} {"step": 28060, "total_steps": 28610, "tokens": 58846085120, "source": "cosmopedia_v2", "phase": 5, "loss": 2.857569843530655, "ce_loss": 2.8455983996391296, "logit_z_loss": 25.779847383499146, "router_aux_loss": 1.1471699625253677, "router_z_loss": 0.49973373487591743, "routing_entropy_normalized": 0.8107803853481972, "learning_rate": 0.00012003083451176366, "grad_norm": 0.5798690319061279, "tokens_per_second": 3934000.1185232466, "eta_seconds": 293.19612741470337, "data_wait_seconds": 1.96993350982666e-05} {"step": 28070, "total_steps": 28610, "tokens": 58867056640, "source": "finemath", "phase": 5, "loss": 2.2018981873989105, "ce_loss": 2.1902053356170654, "logit_z_loss": 34.48916673660278, "router_aux_loss": 1.1503749340772629, "router_z_loss": 0.18907971866428852, "routing_entropy_normalized": 0.8204539886825669, "learning_rate": 0.00011576228238907605, "grad_norm": 0.2141193151473999, "tokens_per_second": 3933840.7180802342, "eta_seconds": 287.876953125, "data_wait_seconds": 1.8924474716186523e-05} {"step": 28080, "total_steps": 28610, "tokens": 58888028160, "source": "cosmopedia_v2", "phase": 5, "loss": 2.856248199939728, "ce_loss": 2.844309836626053, "logit_z_loss": 25.93678569793701, "router_aux_loss": 1.144879698753357, "router_z_loss": 0.4895361140370369, "routing_entropy_normalized": 0.8318042670321977, "learning_rate": 0.00011156796273424058, "grad_norm": 0.5318847298622131, "tokens_per_second": 4005079.522803191, "eta_seconds": 277.5202224254608, "data_wait_seconds": 1.5616416931152344e-05} {"step": 28090, "total_steps": 28610, "tokens": 58908999680, "source": "fineweb_edu", "phase": 5, "loss": 3.4692002832889557, "ce_loss": 3.457459509372711, "logit_z_loss": 22.52413058280945, "router_aux_loss": 1.1444062143564224, "router_z_loss": 0.29670859687030315, "routing_entropy_normalized": 0.826239041922377, "learning_rate": 0.00010744810047589115, "grad_norm": 0.32552576065063477, "tokens_per_second": 3668501.365980148, "eta_seconds": 297.26554012298584, "data_wait_seconds": 0.053449004888534546} {"step": 28100, "total_steps": 28610, "tokens": 58929971200, "source": "cosmopedia_v2", "phase": 5, "loss": 2.8385543525218964, "ce_loss": 2.826692223548889, "logit_z_loss": 26.14983296394348, "router_aux_loss": 1.1440565586090088, "router_z_loss": 0.4216091223061085, "routing_entropy_normalized": 0.8469309961037502, "learning_rate": 0.0001034029165497375, "grad_norm": 0.8730766177177429, "tokens_per_second": 4261380.212448583, "eta_seconds": 250.98617506027222, "data_wait_seconds": 1.9997358322143555e-05} {"step": 28110, "total_steps": 28610, "tokens": 58950942720, "source": "npset2", "phase": 5, "loss": 1.5759007185697556, "ce_loss": 1.5641705244779587, "logit_z_loss": 37.4142427444458, "router_aux_loss": 1.142189860343933, "router_z_loss": 0.30830928683280945, "routing_entropy_normalized": 0.8065408826439442, "learning_rate": 9.943262788671748e-05, "grad_norm": 0.3037864565849304, "tokens_per_second": 4231393.501198642, "eta_seconds": 247.8086709976196, "data_wait_seconds": 1.952052116394043e-05} {"step": 28120, "total_steps": 28610, "tokens": 58971914240, "source": "fineweb_edu", "phase": 5, "loss": 3.4753189086914062, "ce_loss": 3.463602304458618, "logit_z_loss": 22.915748834609985, "router_aux_loss": 1.1444860845804214, "router_z_loss": 0.2718108557164669, "routing_entropy_normalized": 0.8356689061221874, "learning_rate": 9.553744740136211e-05, "grad_norm": 0.4054260849952698, "tokens_per_second": 4047203.3560181684, "eta_seconds": 253.90482997894287, "data_wait_seconds": 2.142786979675293e-05} {"step": 28130, "total_steps": 28610, "tokens": 58992885760, "source": "dclm", "phase": 5, "loss": 3.5528814792633057, "ce_loss": 3.541232615709305, "logit_z_loss": 23.04280376434326, "router_aux_loss": 1.143473595380783, "router_z_loss": 0.21409964747726917, "routing_entropy_normalized": 0.8401052949936209, "learning_rate": 9.171758398038016e-05, "grad_norm": 0.2004852294921875, "tokens_per_second": 4335497.850071237, "eta_seconds": 232.1839370727539, "data_wait_seconds": 2.372264862060547e-05} {"step": 28140, "total_steps": 28610, "tokens": 59013857280, "source": "finemath", "phase": 5, "loss": 2.2527211606502533, "ce_loss": 2.2410978078842163, "logit_z_loss": 34.34602642059326, "router_aux_loss": 1.1417751461267471, "router_z_loss": 0.20564361847937107, "routing_entropy_normalized": 0.8369681773494698, "learning_rate": 8.797324247145411e-05, "grad_norm": 0.1915145367383957, "tokens_per_second": 4076619.1550633404, "eta_seconds": 241.78403782844543, "data_wait_seconds": 1.6421079635620117e-05} {"step": 28150, "total_steps": 28610, "tokens": 59034828800, "source": "dclm", "phase": 5, "loss": 3.565601944923401, "ce_loss": 3.5539133548736572, "logit_z_loss": 23.07322669029236, "router_aux_loss": 1.1441297382116318, "router_z_loss": 0.24734466150403023, "routing_entropy_normalized": 0.8211217550213764, "learning_rate": 8.430462367225562e-05, "grad_norm": 0.31337159872055054, "tokens_per_second": 3548874.05624914, "eta_seconds": 271.82985496520996, "data_wait_seconds": 1.8656253814697266e-05} {"step": 28160, "total_steps": 28610, "tokens": 59055800320, "source": "cosmopedia_v2", "phase": 5, "loss": 2.8249200880527496, "ce_loss": 2.8130362033843994, "logit_z_loss": 26.173656225204468, "router_aux_loss": 1.1436467468738556, "router_z_loss": 0.4474475719034672, "routing_entropy_normalized": 0.8230507695401179, "learning_rate": 8.07119243196771e-05, "grad_norm": 0.35555535554885864, "tokens_per_second": 3984821.6966118594, "eta_seconds": 236.82826280593872, "data_wait_seconds": 2.0742416381835938e-05} {"step": 28170, "total_steps": 28610, "tokens": 59076771840, "source": "finemath", "phase": 5, "loss": 2.2011674344539642, "ce_loss": 2.189516097307205, "logit_z_loss": 34.232972621917725, "router_aux_loss": 1.1433580070734024, "router_z_loss": 0.2177581014111638, "routing_entropy_normalized": 0.8152192616965523, "learning_rate": 7.719533707928179e-05, "grad_norm": 0.26549670100212097, "tokens_per_second": 3777935.108941766, "eta_seconds": 244.2463550567627, "data_wait_seconds": 0.0808371901512146} {"step": 28180, "total_steps": 28610, "tokens": 59097743360, "source": "cosmopedia_v2", "phase": 5, "loss": 2.825288772583008, "ce_loss": 2.813451737165451, "logit_z_loss": 26.08385133743286, "router_aux_loss": 1.1382818520069122, "router_z_loss": 0.4542282111942768, "routing_entropy_normalized": 0.8599172729154096, "learning_rate": 7.375505053497094e-05, "grad_norm": 0.2970309257507324, "tokens_per_second": 4345717.338486745, "eta_seconds": 207.5089771747589, "data_wait_seconds": 2.0235776901245117e-05} {"step": 28190, "total_steps": 28610, "tokens": 59118714880, "source": "fineweb_edu", "phase": 5, "loss": 3.46229550242424, "ce_loss": 3.450626254081726, "logit_z_loss": 22.68706464767456, "router_aux_loss": 1.1384910643100739, "router_z_loss": 0.28434498608112335, "routing_entropy_normalized": 0.8552725566802813, "learning_rate": 7.039124917887085e-05, "grad_norm": 0.21015027165412903, "tokens_per_second": 4285201.72301092, "eta_seconds": 205.54547882080075, "data_wait_seconds": 1.823902130126953e-05} {"step": 28200, "total_steps": 28610, "tokens": 59139686400, "source": "cosmopedia_v2", "phase": 5, "loss": 2.8146170675754547, "ce_loss": 2.8028166592121124, "logit_z_loss": 26.22274374961853, "router_aux_loss": 1.1355251520872116, "router_z_loss": 0.4451347477734089, "routing_entropy_normalized": 0.8575343142287454, "learning_rate": 6.710411340144007e-05, "grad_norm": 0.3364868760108948, "tokens_per_second": 4005561.196690192, "eta_seconds": 214.65963888168335, "data_wait_seconds": 2.1845102310180664e-05} {"step": 28210, "total_steps": 28610, "tokens": 59160657920, "source": "npset2", "phase": 5, "loss": 1.5706481337547302, "ce_loss": 1.5589661598205566, "logit_z_loss": 37.577370166778564, "router_aux_loss": 1.1358884572982788, "router_z_loss": 0.323131050914526, "routing_entropy_normalized": 0.8080924046325687, "learning_rate": 6.38938194817944e-05, "grad_norm": 0.30726292729377747, "tokens_per_second": 3628005.4402215914, "eta_seconds": 231.21817588806152, "data_wait_seconds": 1.895427703857422e-05} {"step": 28220, "total_steps": 28610, "tokens": 59181629440, "source": "fineweb_edu", "phase": 5, "loss": 3.3995717465877533, "ce_loss": 3.3878896236419678, "logit_z_loss": 23.089441299438477, "router_aux_loss": 1.1390522867441177, "router_z_loss": 0.2916587386280298, "routing_entropy_normalized": 0.8531743808163216, "learning_rate": 6.0760539578254115e-05, "grad_norm": 0.22132571041584015, "tokens_per_second": 4542284.952868354, "eta_seconds": 180.06120014190674, "data_wait_seconds": 1.940131187438965e-05} {"step": 28230, "total_steps": 28610, "tokens": 59202600960, "source": "dclm", "phase": 5, "loss": 3.5667548179626465, "ce_loss": 3.5551679134368896, "logit_z_loss": 23.128688097000122, "router_aux_loss": 1.1363331079483032, "router_z_loss": 0.22358847223222256, "routing_entropy_normalized": 0.8420471635179393, "learning_rate": 5.770444171911082e-05, "grad_norm": 0.15331889688968658, "tokens_per_second": 4196159.820767289, "eta_seconds": 189.91596937179565, "data_wait_seconds": 1.8537044525146484e-05} {"step": 28240, "total_steps": 28610, "tokens": 59223572480, "source": "finemath", "phase": 5, "loss": 2.2859044671058655, "ce_loss": 2.2743550539016724, "logit_z_loss": 34.651061058044434, "router_aux_loss": 1.1381640285253525, "router_z_loss": 0.1678282655775547, "routing_entropy_normalized": 0.835897037247811, "learning_rate": 5.472568979361853e-05, "grad_norm": 0.29108086228370667, "tokens_per_second": 3375927.8175498415, "eta_seconds": 229.84681010246277, "data_wait_seconds": 1.7344951629638672e-05} {"step": 28250, "total_steps": 28610, "tokens": 59244544000, "source": "dclm", "phase": 5, "loss": 3.498935490846634, "ce_loss": 3.487337440252304, "logit_z_loss": 23.117897748947144, "router_aux_loss": 1.1379653960466385, "router_z_loss": 0.21834231168031693, "routing_entropy_normalized": 0.8414088564111836, "learning_rate": 5.1824443543202984e-05, "grad_norm": 0.21605584025382996, "tokens_per_second": 3577338.3713546237, "eta_seconds": 211.04369831085205, "data_wait_seconds": 2.1010637283325195e-05} {"step": 28260, "total_steps": 28610, "tokens": 59265515520, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7975504100322723, "ce_loss": 2.785767138004303, "logit_z_loss": 26.407479524612427, "router_aux_loss": 1.1368837654590607, "router_z_loss": 0.41442912071943283, "routing_entropy_normalized": 0.8364829848851432, "learning_rate": 4.9000858552895375e-05, "grad_norm": 0.19018103182315826, "tokens_per_second": 3183927.079939361, "eta_seconds": 230.5339229106903, "data_wait_seconds": 1.817941665649414e-05} {"step": 28270, "total_steps": 28610, "tokens": 59286487040, "source": "finemath", "phase": 5, "loss": 2.1777670979499817, "ce_loss": 2.166230231523514, "logit_z_loss": 35.03705024719238, "router_aux_loss": 1.1353534907102585, "router_z_loss": 0.18324064277112484, "routing_entropy_normalized": 0.8486954127865337, "learning_rate": 4.625508624298952e-05, "grad_norm": 0.20284980535507202, "tokens_per_second": 4003842.0373605127, "eta_seconds": 178.08686590194702, "data_wait_seconds": 2.065300941467285e-05} {"step": 28280, "total_steps": 28610, "tokens": 59307458560, "source": "cosmopedia_v2", "phase": 5, "loss": 2.8057923018932343, "ce_loss": 2.7940456867218018, "logit_z_loss": 26.23194408416748, "router_aux_loss": 1.1350609511137009, "router_z_loss": 0.3959587588906288, "routing_entropy_normalized": 0.8548836148749686, "learning_rate": 4.3587273860921985e-05, "grad_norm": 0.7548964023590088, "tokens_per_second": 3869120.1782415668, "eta_seconds": 178.86757922172546, "data_wait_seconds": 2.047419548034668e-05} {"step": 28290, "total_steps": 28610, "tokens": 59328430080, "source": "fineweb_edu", "phase": 5, "loss": 3.4444962441921234, "ce_loss": 3.432883858680725, "logit_z_loss": 22.759189128875732, "router_aux_loss": 1.134047195315361, "router_z_loss": 0.2718684431165457, "routing_entropy_normalized": 0.8584434950743082, "learning_rate": 4.0997564473374104e-05, "grad_norm": 0.2661270499229431, "tokens_per_second": 3734109.833547744, "eta_seconds": 179.718505859375, "data_wait_seconds": 2.1785497665405273e-05} {"step": 28300, "total_steps": 28610, "tokens": 59349401600, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7988727390766144, "ce_loss": 2.787124067544937, "logit_z_loss": 26.359848260879517, "router_aux_loss": 1.13512821495533, "router_z_loss": 0.3973729945719242, "routing_entropy_normalized": 0.845530490635873, "learning_rate": 3.848609695860078e-05, "grad_norm": 0.1838477998971939, "tokens_per_second": 3868455.189134277, "eta_seconds": 168.05600380897522, "data_wait_seconds": 2.1070241928100586e-05} {"step": 28310, "total_steps": 28610, "tokens": 59370373120, "source": "npset2", "phase": 5, "loss": 1.5900703817605972, "ce_loss": 1.578428015112877, "logit_z_loss": 37.24407148361206, "router_aux_loss": 1.1328914910554886, "router_z_loss": 0.3134901858866215, "routing_entropy_normalized": 0.8001199363371495, "learning_rate": 3.6053005998983254e-05, "grad_norm": 0.30620670318603516, "tokens_per_second": 4159700.851915317, "eta_seconds": 151.2477993965149, "data_wait_seconds": 2.3871660232543945e-05} {"step": 28320, "total_steps": 28610, "tokens": 59391344640, "source": "fineweb_edu", "phase": 5, "loss": 3.4307896196842194, "ce_loss": 3.4192073047161102, "logit_z_loss": 22.98045325279236, "router_aux_loss": 1.1324403434991837, "router_z_loss": 0.2578986808657646, "routing_entropy_normalized": 0.8633059340499628, "learning_rate": 3.3698422073805744e-05, "grad_norm": 0.23244938254356384, "tokens_per_second": 4087862.565528613, "eta_seconds": 148.77556920051575, "data_wait_seconds": 1.940131187438965e-05} {"step": 28330, "total_steps": 28610, "tokens": 59412316160, "source": "dclm", "phase": 5, "loss": 3.5587405264377594, "ce_loss": 3.5471573173999786, "logit_z_loss": 23.03971004486084, "router_aux_loss": 1.1333892196416855, "router_z_loss": 0.24936968088150024, "routing_entropy_normalized": 0.877980665809458, "learning_rate": 3.1422471452258016e-05, "grad_norm": 0.2661083936691284, "tokens_per_second": 3548980.3011921076, "eta_seconds": 165.45669746398926, "data_wait_seconds": 1.773238182067871e-05} {"step": 28340, "total_steps": 28610, "tokens": 59433287680, "source": "finemath", "phase": 5, "loss": 2.161890536546707, "ce_loss": 2.1503761410713196, "logit_z_loss": 35.2562518119812, "router_aux_loss": 1.1331002712249756, "router_z_loss": 0.18342773988842964, "routing_entropy_normalized": 0.8298115933289071, "learning_rate": 2.9225276186664653e-05, "grad_norm": 0.241336390376091, "tokens_per_second": 4307812.275111535, "eta_seconds": 131.44283080101013, "data_wait_seconds": 1.996755599975586e-05} {"step": 28350, "total_steps": 28610, "tokens": 59454259200, "source": "dclm", "phase": 5, "loss": 3.603043019771576, "ce_loss": 3.5914676189422607, "logit_z_loss": 22.980189561843872, "router_aux_loss": 1.1349316537380219, "router_z_loss": 0.22611239552497864, "routing_entropy_normalized": 0.8539050499207519, "learning_rate": 2.710695410593994e-05, "grad_norm": 0.28729885816574097, "tokens_per_second": 3930034.079686987, "eta_seconds": 138.74167728424072, "data_wait_seconds": 1.9550323486328125e-05} {"step": 28360, "total_steps": 28610, "tokens": 59475230720, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7986084520816803, "ce_loss": 2.786814361810684, "logit_z_loss": 26.43400502204895, "router_aux_loss": 1.134001448750496, "router_z_loss": 0.4540867544710636, "routing_entropy_normalized": 0.863240817780386, "learning_rate": 2.5067618809268278e-05, "grad_norm": 0.3730792999267578, "tokens_per_second": 4191109.8343657176, "eta_seconds": 125.0952661037445, "data_wait_seconds": 2.092123031616211e-05} {"step": 28370, "total_steps": 28610, "tokens": 59496202240, "source": "finemath", "phase": 5, "loss": 2.338011145591736, "ce_loss": 2.3264894783496857, "logit_z_loss": 34.09506845474243, "router_aux_loss": 1.1331501007080078, "router_z_loss": 0.1901590581983328, "routing_entropy_normalized": 0.8632945612068849, "learning_rate": 2.3107379660012363e-05, "grad_norm": 0.2013065069913864, "tokens_per_second": 4073869.374125321, "eta_seconds": 123.54752540588379, "data_wait_seconds": 2.0265579223632812e-05} {"step": 28380, "total_steps": 28610, "tokens": 59517173760, "source": "cosmopedia_v2", "phase": 5, "loss": 2.8032900989055634, "ce_loss": 2.791520267724991, "logit_z_loss": 26.262795448303223, "router_aux_loss": 1.132930040359497, "router_z_loss": 0.4404784068465233, "routing_entropy_normalized": 0.8773594741207841, "learning_rate": 2.122634177984889e-05, "grad_norm": 0.30049648880958557, "tokens_per_second": 3905102.3210375705, "eta_seconds": 123.51659965515137, "data_wait_seconds": 2.053380012512207e-05} {"step": 28390, "total_steps": 28610, "tokens": 59538145280, "source": "fineweb_edu", "phase": 5, "loss": 3.4675089716911316, "ce_loss": 3.455911159515381, "logit_z_loss": 22.65322732925415, "router_aux_loss": 1.1331742405891418, "router_z_loss": 0.26605293340981007, "routing_entropy_normalized": 0.8719832066602838, "learning_rate": 1.942460604313073e-05, "grad_norm": 0.1557585448026657, "tokens_per_second": 4058769.8782141972, "eta_seconds": 113.67321968078613, "data_wait_seconds": 0.05573400855064392} {"step": 28400, "total_steps": 28610, "tokens": 59559116800, "source": "cosmopedia_v2", "phase": 5, "loss": 2.8138711154460907, "ce_loss": 2.802109569311142, "logit_z_loss": 26.221176624298096, "router_aux_loss": 1.1324796974658966, "router_z_loss": 0.4367651976644993, "routing_entropy_normalized": 0.8714061886179116, "learning_rate": 1.7702269071477428e-05, "grad_norm": 0.42910581827163696, "tokens_per_second": 4255137.711077464, "eta_seconds": 103.49886417388915, "data_wait_seconds": 2.002716064453125e-05} {"step": 28410, "total_steps": 28610, "tokens": 59580088320, "source": "npset2", "phase": 5, "loss": 1.5453701317310333, "ce_loss": 1.5337838977575302, "logit_z_loss": 37.7141489982605, "router_aux_loss": 1.1313526332378387, "router_z_loss": 0.27270414121448994, "routing_entropy_normalized": 0.8199449179025451, "learning_rate": 1.605942322859366e-05, "grad_norm": 0.18959227204322815, "tokens_per_second": 3947136.847038353, "eta_seconds": 106.26193523406982, "data_wait_seconds": 1.8864870071411133e-05} {"step": 28420, "total_steps": 28610, "tokens": 59601059840, "source": "fineweb_edu", "phase": 5, "loss": 3.3983368277549744, "ce_loss": 3.3867355585098267, "logit_z_loss": 23.005760669708252, "router_aux_loss": 1.1321180760860443, "router_z_loss": 0.2800303529947996, "routing_entropy_normalized": 0.8709546512488237, "learning_rate": 1.4496156615316391e-05, "grad_norm": 0.19778811931610107, "tokens_per_second": 4325836.830928206, "eta_seconds": 92.11139845848083, "data_wait_seconds": 2.181529998779297e-05} {"step": 28430, "total_steps": 28610, "tokens": 59622031360, "source": "dclm", "phase": 5, "loss": 3.557070940732956, "ce_loss": 3.5455488860607147, "logit_z_loss": 22.95970129966736, "router_aux_loss": 1.130581259727478, "router_z_loss": 0.21624510176479816, "routing_entropy_normalized": 0.8740810353545111, "learning_rate": 1.3012553064889632e-05, "grad_norm": 0.2222701758146286, "tokens_per_second": 3577949.2385683064, "eta_seconds": 105.5038332939148, "data_wait_seconds": 1.8805265426635742e-05} {"step": 28440, "total_steps": 28610, "tokens": 59643002880, "source": "finemath", "phase": 5, "loss": 2.1909379214048386, "ce_loss": 2.179443433880806, "logit_z_loss": 34.62312126159668, "router_aux_loss": 1.132166638970375, "router_z_loss": 0.17282235994935036, "routing_entropy_normalized": 0.8625835337495573, "learning_rate": 1.160869213846938e-05, "grad_norm": 0.3109655976295471, "tokens_per_second": 3886434.521763516, "eta_seconds": 91.73339676856995, "data_wait_seconds": 1.7642974853515625e-05} {"step": 28450, "total_steps": 28610, "tokens": 59663974400, "source": "dclm", "phase": 5, "loss": 3.7093681693077087, "ce_loss": 3.6978437900543213, "logit_z_loss": 22.514865159988403, "router_aux_loss": 1.1314284354448318, "router_z_loss": 0.2101354282349348, "routing_entropy_normalized": 0.8804778157302017, "learning_rate": 1.0284649120856527e-05, "grad_norm": 0.2852003574371338, "tokens_per_second": 4002724.9833053183, "eta_seconds": 83.82897186279297, "data_wait_seconds": 1.6927719116210938e-05} {"step": 28460, "total_steps": 28610, "tokens": 59684945920, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7980640828609467, "ce_loss": 2.7863463163375854, "logit_z_loss": 26.447378158569336, "router_aux_loss": 1.13259157538414, "router_z_loss": 0.3918267898261547, "routing_entropy_normalized": 0.8735579486976006, "learning_rate": 9.040495016460094e-06, "grad_norm": 0.2401314228773117, "tokens_per_second": 4194120.808001512, "eta_seconds": 75.00327587127686, "data_wait_seconds": 2.2292137145996094e-05} {"step": 28470, "total_steps": 28610, "tokens": 59705917440, "source": "finemath", "phase": 5, "loss": 2.147095575928688, "ce_loss": 2.1356002539396286, "logit_z_loss": 35.11658430099487, "router_aux_loss": 1.131084144115448, "router_z_loss": 0.18446779437363148, "routing_entropy_normalized": 0.849569109301432, "learning_rate": 7.876296545488603e-06, "grad_norm": 0.15857058763504028, "tokens_per_second": 4342030.1966386745, "eta_seconds": 67.61843347549438, "data_wait_seconds": 2.2113323211669922e-05} {"step": 28480, "total_steps": 28610, "tokens": 59726888960, "source": "cosmopedia_v2", "phase": 5, "loss": 2.79034361243248, "ce_loss": 2.778613954782486, "logit_z_loss": 26.55735969543457, "router_aux_loss": 1.1315055638551712, "router_z_loss": 0.41459862142801285, "routing_entropy_normalized": 0.8836999015030885, "learning_rate": 6.792116140373117e-06, "grad_norm": 0.8346433043479919, "tokens_per_second": 3836790.1106986864, "eta_seconds": 71.05673027038574, "data_wait_seconds": 2.3066997528076172e-05} {"step": 28490, "total_steps": 28610, "tokens": 59747860480, "source": "fineweb_edu", "phase": 5, "loss": 3.419430196285248, "ce_loss": 3.4078318774700165, "logit_z_loss": 22.95578694343567, "router_aux_loss": 1.1318596005439758, "router_z_loss": 0.2797381952404976, "routing_entropy_normalized": 0.8790217252021565, "learning_rate": 5.788011942418237e-06, "grad_norm": 0.19771647453308105, "tokens_per_second": 3427991.0365794655, "eta_seconds": 73.41274738311768, "data_wait_seconds": 0.1669522523880005} {"step": 28500, "total_steps": 28610, "tokens": 59768832000, "source": "cosmopedia_v2", "phase": 5, "loss": 2.808714598417282, "ce_loss": 2.7969537675380707, "logit_z_loss": 26.27467703819275, "router_aux_loss": 1.1325212270021439, "router_z_loss": 0.4355529546737671, "routing_entropy_normalized": 0.8767801980461712, "learning_rate": 4.864037798685106e-06, "grad_norm": 0.3214420676231384, "tokens_per_second": 4355744.177018424, "eta_seconds": 52.9614942073822, "data_wait_seconds": 2.1070241928100586e-05} {"step": 28510, "total_steps": 28610, "tokens": 59789803520, "source": "npset2", "phase": 5, "loss": 1.5507932305335999, "ce_loss": 1.5392421782016754, "logit_z_loss": 37.51888418197632, "router_aux_loss": 1.1288949251174927, "router_z_loss": 0.262073477730155, "routing_entropy_normalized": 0.8287835663251537, "learning_rate": 4.0202432591025515e-06, "grad_norm": 0.235471710562706, "tokens_per_second": 4154930.908569766, "eta_seconds": 50.473811626434326, "data_wait_seconds": 2.2590160369873047e-05} {"step": 28520, "total_steps": 28610, "tokens": 59810775040, "source": "fineweb_edu", "phase": 5, "loss": 3.4657942354679108, "ce_loss": 3.454213470220566, "logit_z_loss": 22.91712236404419, "router_aux_loss": 1.1308083683252335, "router_z_loss": 0.27263848669826984, "routing_entropy_normalized": 0.8812989727866994, "learning_rate": 3.2566735738112084e-06, "grad_norm": 0.2372068464756012, "tokens_per_second": 4119704.061784791, "eta_seconds": 45.81486368179321, "data_wait_seconds": 2.1070241928100586e-05} {"step": 28530, "total_steps": 28610, "tokens": 59831746560, "source": "dclm", "phase": 5, "loss": 3.616079032421112, "ce_loss": 3.6045749187469482, "logit_z_loss": 22.817455768585205, "router_aux_loss": 1.1303915530443192, "router_z_loss": 0.200216518715024, "routing_entropy_normalized": 0.876950894743613, "learning_rate": 2.5733696907359604e-06, "grad_norm": 0.16323983669281006, "tokens_per_second": 3744772.1354980124, "eta_seconds": 44.801700592041016, "data_wait_seconds": 1.8477439880371094e-05} {"step": 28540, "total_steps": 28610, "tokens": 59852718080, "source": "finemath", "phase": 5, "loss": 1.9722069054841995, "ce_loss": 1.9607226699590683, "logit_z_loss": 35.53999662399292, "router_aux_loss": 1.1305185109376907, "router_z_loss": 0.17905435152351856, "routing_entropy_normalized": 0.8603118755691536, "learning_rate": 1.9703682533901977e-06, "grad_norm": 0.14570476114749908, "tokens_per_second": 4053083.8809498916, "eta_seconds": 36.21949219703674, "data_wait_seconds": 1.6570091247558594e-05} {"step": 28550, "total_steps": 28610, "tokens": 59873689600, "source": "dclm", "phase": 5, "loss": 3.6269178688526154, "ce_loss": 3.6153849959373474, "logit_z_loss": 22.700026512145996, "router_aux_loss": 1.130830094218254, "router_z_loss": 0.2246313076466322, "routing_entropy_normalized": 0.883886407410607, "learning_rate": 1.4477015989110553e-06, "grad_norm": 0.17170587182044983, "tokens_per_second": 4308296.087019215, "eta_seconds": 29.206237792968754, "data_wait_seconds": 1.8924474716186523e-05} {"step": 28560, "total_steps": 28610, "tokens": 59894661120, "source": "cosmopedia_v2", "phase": 5, "loss": 2.790831744670868, "ce_loss": 2.7791003584861755, "logit_z_loss": 26.612406969070435, "router_aux_loss": 1.1315158903598785, "router_z_loss": 0.4162539392709732, "routing_entropy_normalized": 0.8776584520264326, "learning_rate": 1.0053977563251326e-06, "grad_norm": 0.1968858540058136, "tokens_per_second": 3863592.649082323, "eta_seconds": 27.139921188354492, "data_wait_seconds": 2.104043960571289e-05} {"step": 28570, "total_steps": 28610, "tokens": 59915632640, "source": "finemath", "phase": 5, "loss": 2.1863881051540375, "ce_loss": 2.1747566610574722, "logit_z_loss": 35.00465106964111, "router_aux_loss": 1.1303027719259262, "router_z_loss": 0.3284557834267616, "routing_entropy_normalized": 0.8679049635705, "learning_rate": 6.434804450450305e-07, "grad_norm": 0.20022663474082947, "tokens_per_second": 3654851.925509454, "eta_seconds": 22.951977729797363, "data_wait_seconds": 0.12939512729644775} {"step": 28580, "total_steps": 28610, "tokens": 59936604160, "source": "cosmopedia_v2", "phase": 5, "loss": 2.7991241812705994, "ce_loss": 2.787378579378128, "logit_z_loss": 26.49033236503601, "router_aux_loss": 1.1313853114843369, "router_z_loss": 0.43172862753272057, "routing_entropy_normalized": 0.8733043988490443, "learning_rate": 3.619690735978676e-07, "grad_norm": 0.24607665836811066, "tokens_per_second": 4320361.44392654, "eta_seconds": 14.562337160110474, "data_wait_seconds": 2.1576881408691406e-05} {"step": 28590, "total_steps": 28610, "tokens": 59957575680, "source": "fineweb_edu", "phase": 5, "loss": 3.399412751197815, "ce_loss": 3.387845039367676, "logit_z_loss": 23.18182921409607, "router_aux_loss": 1.1305192410945892, "router_z_loss": 0.26251762360334396, "routing_entropy_normalized": 0.8889308333070203, "learning_rate": 1.6087873858411373e-07, "grad_norm": 0.14011861383914948, "tokens_per_second": 4149526.2000738285, "eta_seconds": 10.107910633087158, "data_wait_seconds": 2.2113323211669922e-05} {"step": 28600, "total_steps": 28610, "tokens": 59978547200, "source": "cosmopedia_v2", "phase": 5, "loss": 2.774931013584137, "ce_loss": 2.7632090747356415, "logit_z_loss": 26.57383894920349, "router_aux_loss": 1.1307422965765, "router_z_loss": 0.41450344026088715, "routing_entropy_normalized": 0.8807967830612493, "learning_rate": 4.022022386823698e-08, "grad_norm": 0.23850508034229279, "tokens_per_second": 3769684.5747681344, "eta_seconds": 5.563202857971191, "data_wait_seconds": 2.0742416381835938e-05} {"step": 28610, "total_steps": 28610, "tokens": 59999518720, "source": "npset2", "phase": 5, "loss": 1.745762512087822, "ce_loss": 1.7341523617506027, "logit_z_loss": 36.487751960754395, "router_aux_loss": 1.1288997828960419, "router_z_loss": 0.3211454078555107, "routing_entropy_normalized": 0.8315624889632758, "learning_rate": 0.0, "grad_norm": 0.25187817215919495, "tokens_per_second": 4184896.9457226605, "eta_seconds": 0.0, "data_wait_seconds": 2.1636486053466797e-05}