{"step": 0, "eval": {"dense": 0.29589015431702137, "uniform2": 0.18351956829428673, "mosaic": 0.24955991158882776, "gap_uniform2": 0.6732382861276468, "gap_mosaic": 0.6934149401883284}, "eval_extent": {"1x1x1": 0.4007962744130407, "1x1x2": 0.31223080786820945, "1x1x4": 0.2370571091101227, "1x2x1": 0.29517988055875455, "1x2x2": 0.19886096763228508, "1x2x4": 0.13515716742974146, "1x4x1": 0.2090887156431299, "1x4x2": 0.11668259091330348, "1x4x4": 0.0879126813342951}} {"step": 1, "loss": 0.33032920956611633, "data": 0.33032920956611633, "gap": null, "sigma": 0.7373471856117249, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.1198805719614029, "seconds": 10.4} {"step": 2, "loss": 0.7165993452072144, "data": 0.20391950011253357, "gap": 0.5126798152923584, "sigma": 0.8721965551376343, "kind": "mosaic", "tokens": 163, "dense": 448, "grad_norm": 0.34913313388824463, "seconds": 20.6} {"step": 3, "loss": 2.1590161323547363, "data": 0.17840678989887238, "gap": 1.9806092977523804, "sigma": 0.9452754855155945, "kind": "mosaic", "tokens": 197, "dense": 448, "grad_norm": 0.6287012696266174, "seconds": 25.2} {"step": 4, "loss": 0.22655856609344482, "data": 0.1100403293967247, "gap": 0.11651823669672012, "sigma": 0.5696934461593628, "kind": "mosaic", "tokens": 268, "dense": 448, "grad_norm": 0.23569534718990326, "seconds": 29.8} {"step": 5, "loss": 0.7774600386619568, "data": 0.31573188304901123, "gap": 0.46172815561294556, "sigma": 0.36320504546165466, "kind": "mosaic", "tokens": 265, "dense": 448, "grad_norm": 0.18135619163513184, "seconds": 34.4} {"step": 6, "loss": 0.8143993020057678, "data": 0.09257905185222626, "gap": 0.7218202352523804, "sigma": 0.842214047908783, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.2544681429862976, "seconds": 44.1} {"step": 7, "loss": 0.5624452233314514, "data": 0.2109021693468094, "gap": 0.3515430688858032, "sigma": 0.3472445607185364, "kind": "mosaic", "tokens": 215, "dense": 448, "grad_norm": 0.24435855448246002, "seconds": 48.7} {"step": 8, "loss": 0.41258567571640015, "data": 0.41258567571640015, "gap": null, "sigma": 0.9829752445220947, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.6541537642478943, "seconds": 51.4} {"step": 9, "loss": 0.3437822461128235, "data": 0.18008363246917725, "gap": 0.16369861364364624, "sigma": 0.9218659400939941, "kind": "mosaic", "tokens": 166, "dense": 448, "grad_norm": 0.2190304696559906, "seconds": 55.8} {"step": 10, "loss": 0.9349219799041748, "data": 0.2638360857963562, "gap": 0.6710858941078186, "sigma": 0.259257048368454, "kind": "mosaic", "tokens": 244, "dense": 448, "grad_norm": 0.39679864048957825, "seconds": 60.4} {"step": 11, "loss": 0.28165414929389954, "data": 0.28165414929389954, "gap": null, "sigma": 0.6205832958221436, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.06823430955410004, "seconds": 63.2} {"step": 12, "loss": 0.1852559745311737, "data": 0.09324035048484802, "gap": 0.09201562404632568, "sigma": 0.4896657466888428, "kind": "mosaic", "tokens": 262, "dense": 448, "grad_norm": 0.10733352601528168, "seconds": 67.7} {"step": 13, "loss": 0.3819490671157837, "data": 0.15440909564495087, "gap": 0.22753997147083282, "sigma": 0.6818842887878418, "kind": "mosaic", "tokens": 208, "dense": 448, "grad_norm": 0.12982311844825745, "seconds": 72.3} {"step": 14, "loss": 0.55352383852005, "data": 0.22644896805286407, "gap": 0.3270748555660248, "sigma": 0.5749927163124084, "kind": "mosaic", "tokens": 252, "dense": 448, "grad_norm": 0.22618603706359863, "seconds": 77.0} {"step": 15, "loss": 0.5084904432296753, "data": 0.16274391114711761, "gap": 0.3457465171813965, "sigma": 0.9040144085884094, "kind": "mosaic", "tokens": 268, "dense": 448, "grad_norm": 0.39170992374420166, "seconds": 81.4} {"step": 16, "loss": 0.3901768624782562, "data": 0.3901768624782562, "gap": null, "sigma": 0.9801896810531616, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3412907123565674, "seconds": 84.2} {"step": 17, "loss": 1.917094111442566, "data": 0.21885259449481964, "gap": 1.6982414722442627, "sigma": 0.9983839392662048, "kind": "mosaic", "tokens": 194, "dense": 448, "grad_norm": 0.6679621934890747, "seconds": 88.7} {"step": 18, "loss": 0.4554670453071594, "data": 0.4554670453071594, "gap": null, "sigma": 0.9970348477363586, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3849268853664398, "seconds": 91.5} {"step": 19, "loss": 2.160745859146118, "data": 0.20186756551265717, "gap": 1.9588782787322998, "sigma": 0.9919070601463318, "kind": "mosaic", "tokens": 165, "dense": 448, "grad_norm": 1.39857017993927, "seconds": 95.9} {"step": 20, "loss": 0.24279159307479858, "data": 0.24279159307479858, "gap": null, "sigma": 0.9079496264457703, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2695155441761017, "seconds": 98.7} {"step": 21, "loss": 0.438040167093277, "data": 0.19111160933971405, "gap": 0.24692855775356293, "sigma": 0.7235287427902222, "kind": "mosaic", "tokens": 238, "dense": 448, "grad_norm": 0.20259042084217072, "seconds": 103.1} {"step": 22, "loss": 0.4724307358264923, "data": 0.4724307358264923, "gap": null, "sigma": 0.9975983500480652, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.576013445854187, "seconds": 105.9} {"step": 23, "loss": 0.21345342695713043, "data": 0.21345342695713043, "gap": null, "sigma": 0.38285207748413086, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.0903380811214447, "seconds": 108.6} {"step": 24, "loss": 5.938042640686035, "data": 1.3378125429153442, "gap": 4.6002302169799805, "sigma": 0.0562245137989521, "kind": "mosaic", "tokens": 229, "dense": 448, "grad_norm": 8.81228256225586, "seconds": 113.0} {"step": 25, "loss": 0.28568753600120544, "data": 0.05090280994772911, "gap": 0.23478472232818604, "sigma": 0.6575742959976196, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.3276703357696533, "seconds": 117.0} {"step": 26, "loss": 0.48871758580207825, "data": 0.48871758580207825, "gap": null, "sigma": 0.9531753063201904, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.32613858580589294, "seconds": 119.8} {"step": 27, "loss": 1.3503762483596802, "data": 0.24707671999931335, "gap": 1.1032994985580444, "sigma": 0.9611159563064575, "kind": "mosaic", "tokens": 218, "dense": 448, "grad_norm": 2.959946632385254, "seconds": 124.4} {"step": 28, "loss": 0.3705834746360779, "data": 0.2133364975452423, "gap": 0.15724696218967438, "sigma": 0.883303165435791, "kind": "mosaic", "tokens": 220, "dense": 448, "grad_norm": 0.1958794742822647, "seconds": 129.1} {"step": 29, "loss": 0.43168312311172485, "data": 0.17168089747428894, "gap": 0.2600022256374359, "sigma": 0.5874873399734497, "kind": "mosaic", "tokens": 194, "dense": 448, "grad_norm": 0.16283239424228668, "seconds": 133.5} {"step": 30, "loss": 1.3499568700790405, "data": 0.1961619108915329, "gap": 1.1537950038909912, "sigma": 0.9807025194168091, "kind": "mosaic", "tokens": 239, "dense": 448, "grad_norm": 0.8665856122970581, "seconds": 138.1} {"step": 31, "loss": 0.5049365162849426, "data": 0.07469840347766876, "gap": 0.43023809790611267, "sigma": 0.7971756458282471, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.24856269359588623, "seconds": 142.4} {"step": 32, "loss": 1.4926280975341797, "data": 0.16532272100448608, "gap": 1.3273054361343384, "sigma": 0.8798080086708069, "kind": "mosaic", "tokens": 193, "dense": 448, "grad_norm": 0.7262457609176636, "seconds": 146.9} {"step": 33, "loss": 0.3120899796485901, "data": 0.3120899796485901, "gap": null, "sigma": 0.7214223146438599, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.15642787516117096, "seconds": 149.6} {"step": 34, "loss": 0.5732499957084656, "data": 0.18486008048057556, "gap": 0.38838991522789, "sigma": 0.9430251121520996, "kind": "mosaic", "tokens": 222, "dense": 448, "grad_norm": 0.32356399297714233, "seconds": 153.9} {"step": 35, "loss": 0.553977906703949, "data": 0.2172396183013916, "gap": 0.3367382884025574, "sigma": 0.8159809112548828, "kind": "mosaic", "tokens": 235, "dense": 448, "grad_norm": 0.23115405440330505, "seconds": 158.5} {"step": 36, "loss": 0.462267130613327, "data": 0.1954767107963562, "gap": 0.2667904198169708, "sigma": 0.7095539569854736, "kind": "mosaic", "tokens": 187, "dense": 448, "grad_norm": 0.16404451429843903, "seconds": 163.0} {"step": 37, "loss": 3.6629912853240967, "data": 0.687340497970581, "gap": 2.9756507873535156, "sigma": 0.14861637353897095, "kind": "mosaic", "tokens": 238, "dense": 448, "grad_norm": 1.8617035150527954, "seconds": 167.4} {"step": 38, "loss": 2.6010611057281494, "data": 0.6354860067367554, "gap": 1.965575098991394, "sigma": 0.16700637340545654, "kind": "mosaic", "tokens": 221, "dense": 448, "grad_norm": 1.391778588294983, "seconds": 171.8} {"step": 39, "loss": 15.467447280883789, "data": 3.201385259628296, "gap": 12.266061782836914, "sigma": 0.02287657931447029, "kind": "mosaic", "tokens": 246, "dense": 448, "grad_norm": 14.160988807678223, "seconds": 176.3} {"step": 40, "loss": 0.42948007583618164, "data": 0.11711739748716354, "gap": 0.3123626708984375, "sigma": 0.9097589254379272, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.3333139717578888, "seconds": 180.6} {"step": 41, "loss": 0.44425851106643677, "data": 0.21918997168540955, "gap": 0.22506852447986603, "sigma": 0.8902490735054016, "kind": "mosaic", "tokens": 258, "dense": 448, "grad_norm": 0.297703355550766, "seconds": 185.0} {"step": 42, "loss": 0.1740870326757431, "data": 0.1740870326757431, "gap": null, "sigma": 0.5688877105712891, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.18639230728149414, "seconds": 187.8} {"step": 43, "loss": 0.6531744003295898, "data": 0.21444007754325867, "gap": 0.4387343227863312, "sigma": 0.8499942421913147, "kind": "mosaic", "tokens": 235, "dense": 448, "grad_norm": 0.337968111038208, "seconds": 192.2} {"step": 44, "loss": 0.9108379483222961, "data": 0.3891944885253906, "gap": 0.5216434597969055, "sigma": 0.3240469694137573, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.5456856489181519, "seconds": 196.6} {"step": 45, "loss": 0.2811942398548126, "data": 0.10635096579790115, "gap": 0.17484326660633087, "sigma": 0.8034327030181885, "kind": "mosaic", "tokens": 199, "dense": 448, "grad_norm": 0.5673373341560364, "seconds": 201.2} {"step": 46, "loss": 0.5196132063865662, "data": 0.22333763539791107, "gap": 0.2962755560874939, "sigma": 0.6973215341567993, "kind": "mosaic", "tokens": 271, "dense": 576, "grad_norm": 0.21249732375144958, "seconds": 205.5} {"step": 47, "loss": 0.2956501543521881, "data": 0.2956501543521881, "gap": null, "sigma": 0.40601053833961487, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.10780730843544006, "seconds": 208.3} {"step": 48, "loss": 0.27250251173973083, "data": 0.27250251173973083, "gap": null, "sigma": 0.8695079684257507, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.34843164682388306, "seconds": 211.0} {"step": 49, "loss": 0.48162856698036194, "data": 0.3044094145298004, "gap": 0.17721915245056152, "sigma": 0.9447206258773804, "kind": "mosaic", "tokens": 243, "dense": 448, "grad_norm": 0.26419129967689514, "seconds": 215.6} {"step": 50, "loss": 0.8979701995849609, "data": 0.28426405787467957, "gap": 0.6137061715126038, "sigma": 0.3810195326805115, "kind": "mosaic", "tokens": 247, "dense": 448, "grad_norm": 0.23261266946792603, "seconds": 220.1} {"step": 51, "loss": 0.4901348948478699, "data": 0.11722519248723984, "gap": 0.37290969491004944, "sigma": 0.7262241840362549, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 1.5662167072296143, "seconds": 224.5} {"step": 52, "loss": 0.2989625930786133, "data": 0.1266181766986847, "gap": 0.17234443128108978, "sigma": 0.6749216318130493, "kind": "mosaic", "tokens": 229, "dense": 448, "grad_norm": 0.19652052223682404, "seconds": 229.1} {"step": 53, "loss": 0.8429771065711975, "data": 0.28972065448760986, "gap": 0.5532564520835876, "sigma": 0.2918144166469574, "kind": "mosaic", "tokens": 207, "dense": 448, "grad_norm": 0.3518405556678772, "seconds": 233.6} {"step": 54, "loss": 0.3066231608390808, "data": 0.3066231608390808, "gap": null, "sigma": 0.9475508332252502, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.29021909832954407, "seconds": 236.4} {"step": 55, "loss": 0.4153955578804016, "data": 0.4153955578804016, "gap": null, "sigma": 0.904132068157196, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.23008836805820465, "seconds": 239.3} {"step": 56, "loss": 0.5897945165634155, "data": 0.5897945165634155, "gap": null, "sigma": 0.9900473356246948, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.39974820613861084, "seconds": 242.2} {"step": 57, "loss": 0.5703562498092651, "data": 0.22518737614154816, "gap": 0.3451688885688782, "sigma": 0.8753262162208557, "kind": "mosaic", "tokens": 195, "dense": 448, "grad_norm": 0.2218632698059082, "seconds": 246.6} {"step": 58, "loss": 0.4363149106502533, "data": 0.18604639172554016, "gap": 0.25026851892471313, "sigma": 0.6107502579689026, "kind": "mosaic", "tokens": 241, "dense": 448, "grad_norm": 0.1488625407218933, "seconds": 251.1} {"step": 59, "loss": 0.5137021541595459, "data": 0.20825771987438202, "gap": 0.3054444491863251, "sigma": 0.6731486320495605, "kind": "mosaic", "tokens": 253, "dense": 448, "grad_norm": 0.2727620303630829, "seconds": 255.6} {"step": 60, "loss": 1.123119831085205, "data": 0.3337200880050659, "gap": 0.7893996834754944, "sigma": 0.964516282081604, "kind": "mosaic", "tokens": 235, "dense": 448, "grad_norm": 0.6805928945541382, "seconds": 260.1} {"step": 61, "loss": 0.7810416221618652, "data": 0.2598695158958435, "gap": 0.5211721062660217, "sigma": 0.536760151386261, "kind": "mosaic", "tokens": 238, "dense": 448, "grad_norm": 0.21997669339179993, "seconds": 264.6} {"step": 62, "loss": 0.5073575377464294, "data": 0.20749010145664215, "gap": 0.2998674511909485, "sigma": 0.7090136408805847, "kind": "mosaic", "tokens": 218, "dense": 448, "grad_norm": 0.2000909000635147, "seconds": 269.0} {"step": 63, "loss": 0.5771428346633911, "data": 0.18074512481689453, "gap": 0.3963977098464966, "sigma": 0.7883142232894897, "kind": "mosaic", "tokens": 197, "dense": 448, "grad_norm": 0.3043515682220459, "seconds": 273.4} {"step": 64, "loss": 1.0055410861968994, "data": 0.33956918120384216, "gap": 0.6659719347953796, "sigma": 0.30127590894699097, "kind": "mosaic", "tokens": 294, "dense": 512, "grad_norm": 0.3158489167690277, "seconds": 277.8} {"step": 65, "loss": 0.5077523589134216, "data": 0.2756282687187195, "gap": 0.23212409019470215, "sigma": 0.883789598941803, "kind": "mosaic", "tokens": 264, "dense": 448, "grad_norm": 0.21064306795597076, "seconds": 282.4} {"step": 66, "loss": 0.6175780296325684, "data": 0.12127716094255447, "gap": 0.4963008463382721, "sigma": 0.866838276386261, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.3942413926124573, "seconds": 287.0} {"step": 67, "loss": 0.4011308550834656, "data": 0.13823631405830383, "gap": 0.26289454102516174, "sigma": 0.8361613750457764, "kind": "mosaic", "tokens": 170, "dense": 448, "grad_norm": 0.21885697543621063, "seconds": 291.6} {"step": 68, "loss": 0.35830366611480713, "data": 0.35830366611480713, "gap": null, "sigma": 0.9574754238128662, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.22729060053825378, "seconds": 294.4} {"step": 69, "loss": 0.360796719789505, "data": 0.360796719789505, "gap": null, "sigma": 0.7496079206466675, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.13947390019893646, "seconds": 297.1} {"step": 70, "loss": 0.532070517539978, "data": 0.21615977585315704, "gap": 0.3159107267856598, "sigma": 0.6859972476959229, "kind": "mosaic", "tokens": 261, "dense": 448, "grad_norm": 0.4236219525337219, "seconds": 301.3} {"step": 71, "loss": 1.1800878047943115, "data": 0.18694494664669037, "gap": 0.9931429028511047, "sigma": 0.4421517848968506, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.38669106364250183, "seconds": 305.4} {"step": 72, "loss": 2.179370403289795, "data": 0.6090444922447205, "gap": 1.5703258514404297, "sigma": 0.14330963790416718, "kind": "mosaic", "tokens": 247, "dense": 448, "grad_norm": 1.1646651029586792, "seconds": 309.9} {"step": 73, "loss": 0.5558246374130249, "data": 0.04466618224978447, "gap": 0.5111584663391113, "sigma": 0.6991196870803833, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 0.595018208026886, "seconds": 319.6} {"step": 74, "loss": 0.48831242322921753, "data": 0.1813194900751114, "gap": 0.30699291825294495, "sigma": 0.8070201873779297, "kind": "mosaic", "tokens": 229, "dense": 448, "grad_norm": 0.387435644865036, "seconds": 324.3} {"step": 75, "loss": 0.38943564891815186, "data": 0.38943564891815186, "gap": null, "sigma": 0.9162572622299194, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.30116283893585205, "seconds": 327.2} {"step": 76, "loss": 0.8311829566955566, "data": 0.265895813703537, "gap": 0.565287172794342, "sigma": 0.3446841239929199, "kind": "mosaic", "tokens": 169, "dense": 448, "grad_norm": 2.3919601440429688, "seconds": 331.6} {"step": 77, "loss": 0.6968094110488892, "data": 0.25202783942222595, "gap": 0.4447816014289856, "sigma": 0.9097294211387634, "kind": "mosaic", "tokens": 235, "dense": 448, "grad_norm": 0.3046470880508423, "seconds": 335.9} {"step": 78, "loss": 2.051173448562622, "data": 0.5047298073768616, "gap": 1.5464435815811157, "sigma": 0.21765707433223724, "kind": "mosaic", "tokens": 161, "dense": 448, "grad_norm": 1.0632028579711914, "seconds": 340.2} {"step": 79, "loss": 0.894727349281311, "data": 0.15496523678302765, "gap": 0.7397621273994446, "sigma": 0.9381996393203735, "kind": "mosaic", "tokens": 193, "dense": 448, "grad_norm": 1.297400951385498, "seconds": 344.7} {"step": 80, "loss": 0.6631811857223511, "data": 0.15060998499393463, "gap": 0.5125712156295776, "sigma": 0.8576589226722717, "kind": "mosaic", "tokens": 194, "dense": 448, "grad_norm": 0.46838510036468506, "seconds": 349.1} {"step": 81, "loss": 1.7826879024505615, "data": 0.7417845129966736, "gap": 1.0409033298492432, "sigma": 0.1732017993927002, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.8609848618507385, "seconds": 353.5} {"step": 82, "loss": 1.8845279216766357, "data": 0.4545080065727234, "gap": 1.4300198554992676, "sigma": 0.22213801741600037, "kind": "mosaic", "tokens": 187, "dense": 448, "grad_norm": 0.6489255428314209, "seconds": 357.9} {"step": 83, "loss": 0.4628092050552368, "data": 0.1603657305240631, "gap": 0.3024434745311737, "sigma": 0.5492868423461914, "kind": "mosaic", "tokens": 219, "dense": 448, "grad_norm": 0.16750507056713104, "seconds": 362.5} {"step": 84, "loss": 0.41581207513809204, "data": 0.22167329490184784, "gap": 0.194138765335083, "sigma": 0.7166568040847778, "kind": "mosaic", "tokens": 227, "dense": 448, "grad_norm": 0.2338017225265503, "seconds": 367.0} {"step": 85, "loss": 5.289532661437988, "data": 1.1781748533248901, "gap": 4.111357688903809, "sigma": 0.12481698393821716, "kind": "mosaic", "tokens": 194, "dense": 448, "grad_norm": 3.177321195602417, "seconds": 371.3} {"step": 86, "loss": 0.3419943153858185, "data": 0.3419943153858185, "gap": null, "sigma": 0.3913078308105469, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.08062022179365158, "seconds": 374.2} {"step": 87, "loss": 0.699615478515625, "data": 0.3219970464706421, "gap": 0.3776184618473053, "sigma": 0.40536177158355713, "kind": "mosaic", "tokens": 256, "dense": 448, "grad_norm": 0.3360126316547394, "seconds": 378.6} {"step": 88, "loss": 0.21823547780513763, "data": 0.21823547780513763, "gap": null, "sigma": 0.8996612429618835, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.1498260796070099, "seconds": 381.4} {"step": 89, "loss": 0.45752033591270447, "data": 0.19951078295707703, "gap": 0.25800955295562744, "sigma": 0.46569836139678955, "kind": "mosaic", "tokens": 222, "dense": 448, "grad_norm": 0.15549620985984802, "seconds": 386.0} {"step": 90, "loss": 2.4420433044433594, "data": 0.694107711315155, "gap": 1.7479356527328491, "sigma": 0.17671434581279755, "kind": "mosaic", "tokens": 224, "dense": 448, "grad_norm": 0.9405331015586853, "seconds": 390.5} {"step": 91, "loss": 3.2730636596679688, "data": 0.5211013555526733, "gap": 2.751962184906006, "sigma": 0.1892843246459961, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 1.946187138557434, "seconds": 394.8} {"step": 92, "loss": 0.7482444643974304, "data": 0.22532857954502106, "gap": 0.5229158997535706, "sigma": 0.585851788520813, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.22697001695632935, "seconds": 399.1} {"step": 93, "loss": 0.4436571002006531, "data": 0.17463427782058716, "gap": 0.2690228223800659, "sigma": 0.8260942697525024, "kind": "mosaic", "tokens": 271, "dense": 448, "grad_norm": 0.3401586711406708, "seconds": 403.7} {"step": 94, "loss": 1.497236967086792, "data": 0.44470298290252686, "gap": 1.0525339841842651, "sigma": 0.17647522687911987, "kind": "mosaic", "tokens": 243, "dense": 448, "grad_norm": 0.8475401997566223, "seconds": 408.1} {"step": 95, "loss": 0.5426442623138428, "data": 0.5426442623138428, "gap": null, "sigma": 0.9994184374809265, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.7475112676620483, "seconds": 410.8} {"step": 96, "loss": 0.6527393460273743, "data": 0.2251865565776825, "gap": 0.4275527894496918, "sigma": 0.7679210305213928, "kind": "mosaic", "tokens": 214, "dense": 448, "grad_norm": 0.5796886086463928, "seconds": 415.2} {"step": 97, "loss": 0.38527634739875793, "data": 0.38527634739875793, "gap": null, "sigma": 0.9387632608413696, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.20573538541793823, "seconds": 418.0} {"step": 98, "loss": 0.3967325687408447, "data": 0.17814745008945465, "gap": 0.21858510375022888, "sigma": 0.6352647542953491, "kind": "mosaic", "tokens": 274, "dense": 448, "grad_norm": 0.151432603597641, "seconds": 422.5} {"step": 99, "loss": 0.9146636128425598, "data": 0.2315940260887146, "gap": 0.6830695867538452, "sigma": 0.28034543991088867, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.49060726165771484, "seconds": 426.7} {"step": 100, "loss": 19.877897262573242, "data": 3.827111005783081, "gap": 16.0507869720459, "sigma": 0.05663831904530525, "kind": "mosaic", "tokens": 233, "dense": 448, "grad_norm": 13.811223030090332, "seconds": 431.2} {"step": 101, "loss": 0.4436681866645813, "data": 0.2720414996147156, "gap": 0.17162668704986572, "sigma": 0.866452693939209, "kind": "mosaic", "tokens": 251, "dense": 448, "grad_norm": 0.20566529035568237, "seconds": 435.6} {"step": 102, "loss": 0.5257474780082703, "data": 0.21020330488681793, "gap": 0.3155441880226135, "sigma": 0.7871720790863037, "kind": "mosaic", "tokens": 202, "dense": 448, "grad_norm": 0.30040648579597473, "seconds": 440.1} {"step": 103, "loss": 0.8200852870941162, "data": 0.8200852870941162, "gap": null, "sigma": 0.9899352788925171, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.8315299153327942, "seconds": 442.9} {"step": 104, "loss": 0.4728187322616577, "data": 0.19603502750396729, "gap": 0.27678370475769043, "sigma": 0.8577414155006409, "kind": "mosaic", "tokens": 173, "dense": 448, "grad_norm": 0.15592586994171143, "seconds": 447.4} {"step": 105, "loss": 0.6385059356689453, "data": 0.18680644035339355, "gap": 0.45169946551322937, "sigma": 0.9232167601585388, "kind": "mosaic", "tokens": 212, "dense": 448, "grad_norm": 0.3065749406814575, "seconds": 452.0} {"step": 106, "loss": 0.3866788446903229, "data": 0.3866788446903229, "gap": null, "sigma": 0.970257043838501, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3367624282836914, "seconds": 454.8} {"step": 107, "loss": 0.37098661065101624, "data": 0.19308581948280334, "gap": 0.1779007911682129, "sigma": 0.7906244397163391, "kind": "mosaic", "tokens": 217, "dense": 480, "grad_norm": 0.19312770664691925, "seconds": 459.4} {"step": 108, "loss": 0.5915383696556091, "data": 0.2354704737663269, "gap": 0.3560678958892822, "sigma": 0.588651180267334, "kind": "mosaic", "tokens": 267, "dense": 448, "grad_norm": 0.23423981666564941, "seconds": 463.9} {"step": 109, "loss": 0.9405341744422913, "data": 0.1735764890909195, "gap": 0.766957700252533, "sigma": 0.9998534917831421, "kind": "mosaic", "tokens": 211, "dense": 448, "grad_norm": 0.5798701643943787, "seconds": 468.4} {"step": 110, "loss": 1.3436394929885864, "data": 0.11892064660787582, "gap": 1.2247188091278076, "sigma": 0.867237389087677, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.8033882975578308, "seconds": 472.7} {"step": 111, "loss": 0.6354517340660095, "data": 0.6354517340660095, "gap": null, "sigma": 0.08297558128833771, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3928394317626953, "seconds": 475.5} {"step": 112, "loss": 0.45041805505752563, "data": 0.21576330065727234, "gap": 0.2346547394990921, "sigma": 0.7401117086410522, "kind": "mosaic", "tokens": 263, "dense": 448, "grad_norm": 0.198158860206604, "seconds": 480.0} {"step": 113, "loss": 0.5872549414634705, "data": 0.5872549414634705, "gap": null, "sigma": 0.06824713200330734, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.11150394380092621, "seconds": 482.8} {"step": 114, "loss": 0.3429108262062073, "data": 0.3429108262062073, "gap": null, "sigma": 0.9817643761634827, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.24944987893104553, "seconds": 485.6} {"step": 115, "loss": 0.43585851788520813, "data": 0.18213766813278198, "gap": 0.25372084975242615, "sigma": 0.7147499918937683, "kind": "mosaic", "tokens": 232, "dense": 448, "grad_norm": 0.3066031336784363, "seconds": 489.9} {"step": 116, "loss": 0.25881227850914, "data": 0.25881227850914, "gap": null, "sigma": 0.9638495445251465, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.20400993525981903, "seconds": 492.6} {"step": 117, "loss": 0.9494789242744446, "data": 0.1723574995994568, "gap": 0.7771214246749878, "sigma": 0.9320509433746338, "kind": "mosaic", "tokens": 237, "dense": 448, "grad_norm": 0.30234190821647644, "seconds": 497.2} {"step": 118, "loss": 1.201762318611145, "data": 0.3946881592273712, "gap": 0.8070741295814514, "sigma": 0.32777169346809387, "kind": "mosaic", "tokens": 244, "dense": 448, "grad_norm": 0.34696629643440247, "seconds": 501.7} {"step": 119, "loss": 0.4400336444377899, "data": 0.1978328675031662, "gap": 0.24220077693462372, "sigma": 0.7973437905311584, "kind": "mosaic", "tokens": 238, "dense": 448, "grad_norm": 0.22253437340259552, "seconds": 506.2} {"step": 120, "loss": 1.3926905393600464, "data": 0.30989864468574524, "gap": 1.0827919244766235, "sigma": 0.9652847051620483, "kind": "mosaic", "tokens": 261, "dense": 448, "grad_norm": 0.7026775479316711, "seconds": 510.8} {"step": 121, "loss": 0.5730089545249939, "data": 0.19563870131969452, "gap": 0.37737026810646057, "sigma": 0.9785884618759155, "kind": "mosaic", "tokens": 242, "dense": 448, "grad_norm": 0.5736625790596008, "seconds": 515.4} {"step": 122, "loss": 1.1115844249725342, "data": 0.3594154119491577, "gap": 0.7521690130233765, "sigma": 0.30183207988739014, "kind": "mosaic", "tokens": 248, "dense": 448, "grad_norm": 0.30032220482826233, "seconds": 519.9} {"step": 123, "loss": 0.6718800067901611, "data": 0.2359296679496765, "gap": 0.43595030903816223, "sigma": 0.34453147649765015, "kind": "mosaic", "tokens": 248, "dense": 448, "grad_norm": 0.2175651490688324, "seconds": 524.4} {"step": 124, "loss": 1.5287425518035889, "data": 0.04162333160638809, "gap": 1.487119197845459, "sigma": 0.9988503456115723, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 0.9031606316566467, "seconds": 528.3} {"step": 125, "loss": 0.2827400863170624, "data": 0.2827400863170624, "gap": null, "sigma": 0.9082300066947937, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3440075218677521, "seconds": 531.1} {"step": 126, "loss": 0.3292282223701477, "data": 0.3292282223701477, "gap": null, "sigma": 0.9674188494682312, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3075787425041199, "seconds": 534.0} {"step": 127, "loss": 1.6965832710266113, "data": 0.5881088972091675, "gap": 1.1084743738174438, "sigma": 0.15018552541732788, "kind": "mosaic", "tokens": 248, "dense": 448, "grad_norm": 0.9683441519737244, "seconds": 538.5} {"step": 128, "loss": 0.6677253246307373, "data": 0.27842727303504944, "gap": 0.3892980217933655, "sigma": 0.9221664667129517, "kind": "mosaic", "tokens": 251, "dense": 448, "grad_norm": 0.2624225914478302, "seconds": 543.0} {"step": 129, "loss": 0.43095266819000244, "data": 0.43095266819000244, "gap": null, "sigma": 0.9129348397254944, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3679400384426117, "seconds": 545.8} {"step": 130, "loss": 0.5519667863845825, "data": 0.5519667863845825, "gap": null, "sigma": 0.9765364527702332, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.6332886219024658, "seconds": 548.6} {"step": 131, "loss": 1.2147352695465088, "data": 0.42183560132980347, "gap": 0.7928996682167053, "sigma": 0.2829054892063141, "kind": "mosaic", "tokens": 260, "dense": 448, "grad_norm": 0.2904525399208069, "seconds": 553.1} {"step": 132, "loss": 0.7184230089187622, "data": 0.19871585071086884, "gap": 0.5197071433067322, "sigma": 0.8067536354064941, "kind": "mosaic", "tokens": 158, "dense": 448, "grad_norm": 0.33210819959640503, "seconds": 557.5} {"step": 133, "loss": 0.375456303358078, "data": 0.375456303358078, "gap": null, "sigma": 0.9753865003585815, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2995269298553467, "seconds": 560.3} {"step": 134, "loss": 0.2136842906475067, "data": 0.1117003858089447, "gap": 0.10198389738798141, "sigma": 0.8265514373779297, "kind": "mosaic", "tokens": 260, "dense": 448, "grad_norm": 0.5311508178710938, "seconds": 564.9} {"step": 135, "loss": 0.5463481545448303, "data": 0.2439507395029068, "gap": 0.30239740014076233, "sigma": 0.6929938793182373, "kind": "mosaic", "tokens": 273, "dense": 448, "grad_norm": 0.22034381330013275, "seconds": 569.4} {"step": 136, "loss": 0.5948335528373718, "data": 0.11417745798826218, "gap": 0.48065611720085144, "sigma": 0.7112604379653931, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.271480530500412, "seconds": 573.5} {"step": 137, "loss": 0.5550205707550049, "data": 0.28355345129966736, "gap": 0.2714671492576599, "sigma": 0.7593069672584534, "kind": "mosaic", "tokens": 270, "dense": 448, "grad_norm": 0.45529207587242126, "seconds": 578.0} {"step": 138, "loss": 0.5071971416473389, "data": 0.21723955869674683, "gap": 0.28995761275291443, "sigma": 0.8108517527580261, "kind": "mosaic", "tokens": 224, "dense": 448, "grad_norm": 0.2926119565963745, "seconds": 582.4} {"step": 139, "loss": 0.9544516801834106, "data": 0.3125646710395813, "gap": 0.6418870091438293, "sigma": 0.4958465099334717, "kind": "mosaic", "tokens": 181, "dense": 448, "grad_norm": 0.27805328369140625, "seconds": 586.9} {"step": 140, "loss": 0.6950629353523254, "data": 0.1985962837934494, "gap": 0.49646666646003723, "sigma": 0.7816876769065857, "kind": "mosaic", "tokens": 167, "dense": 448, "grad_norm": 0.28701862692832947, "seconds": 591.5} {"step": 141, "loss": 0.9080395698547363, "data": 0.25541961193084717, "gap": 0.6526199579238892, "sigma": 0.34991028904914856, "kind": "mosaic", "tokens": 246, "dense": 448, "grad_norm": 0.3116987347602844, "seconds": 596.1} {"step": 142, "loss": 0.47855210304260254, "data": 0.20566114783287048, "gap": 0.27289095520973206, "sigma": 0.7214951515197754, "kind": "mosaic", "tokens": 223, "dense": 448, "grad_norm": 0.548799455165863, "seconds": 600.6} {"step": 143, "loss": 0.39908111095428467, "data": 0.39908111095428467, "gap": null, "sigma": 0.9488863945007324, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.30682608485221863, "seconds": 603.5} {"step": 144, "loss": 0.3559800386428833, "data": 0.153113454580307, "gap": 0.2028665989637375, "sigma": 0.729826807975769, "kind": "mosaic", "tokens": 214, "dense": 448, "grad_norm": 0.1299360990524292, "seconds": 607.9} {"step": 145, "loss": 0.48205167055130005, "data": 0.48205167055130005, "gap": null, "sigma": 0.9929391145706177, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.7736615538597107, "seconds": 610.7} {"step": 146, "loss": 0.5600395798683167, "data": 0.19761288166046143, "gap": 0.3624266982078552, "sigma": 0.5923941731452942, "kind": "mosaic", "tokens": 191, "dense": 448, "grad_norm": 0.14027105271816254, "seconds": 615.1} {"step": 147, "loss": 0.3289729654788971, "data": 0.3289729654788971, "gap": null, "sigma": 0.9829492568969727, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.489402711391449, "seconds": 617.9} {"step": 148, "loss": 13.453330039978027, "data": 3.0071332454681396, "gap": 10.446196556091309, "sigma": 0.06013205647468567, "kind": "mosaic", "tokens": 222, "dense": 448, "grad_norm": 8.57374095916748, "seconds": 622.5} {"step": 149, "loss": 0.7440004944801331, "data": 0.26368528604507446, "gap": 0.4803152084350586, "sigma": 0.5592179298400879, "kind": "mosaic", "tokens": 182, "dense": 448, "grad_norm": 0.19767750799655914, "seconds": 626.9} {"step": 150, "loss": 0.560051441192627, "data": 0.22391855716705322, "gap": 0.3361329138278961, "sigma": 0.9097912907600403, "kind": "mosaic", "tokens": 247, "dense": 448, "grad_norm": 0.16848132014274597, "seconds": 631.4} {"step": 151, "loss": 1.7678587436676025, "data": 0.5029610395431519, "gap": 1.2648977041244507, "sigma": 0.19795361161231995, "kind": "mosaic", "tokens": 249, "dense": 448, "grad_norm": 0.618722140789032, "seconds": 635.9} {"step": 152, "loss": 0.4172258973121643, "data": 0.19948133826255798, "gap": 0.21774454414844513, "sigma": 0.5005940198898315, "kind": "mosaic", "tokens": 245, "dense": 448, "grad_norm": 0.13756531476974487, "seconds": 640.3} {"step": 153, "loss": 0.6376960873603821, "data": 0.2497517317533493, "gap": 0.3879443407058716, "sigma": 0.38135889172554016, "kind": "mosaic", "tokens": 258, "dense": 448, "grad_norm": 0.21962520480155945, "seconds": 644.9} {"step": 154, "loss": 0.5266963243484497, "data": 0.21321114897727966, "gap": 0.31348514556884766, "sigma": 0.766990602016449, "kind": "mosaic", "tokens": 231, "dense": 448, "grad_norm": 0.23680774867534637, "seconds": 649.3} {"step": 155, "loss": 1.4236818552017212, "data": 0.3997478485107422, "gap": 1.023934006690979, "sigma": 0.30959010124206543, "kind": "mosaic", "tokens": 183, "dense": 448, "grad_norm": 0.36494433879852295, "seconds": 653.7} {"step": 156, "loss": 0.25909990072250366, "data": 0.25909990072250366, "gap": null, "sigma": 0.9943181872367859, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.27169540524482727, "seconds": 656.4} {"step": 157, "loss": 0.3148914575576782, "data": 0.3148914575576782, "gap": null, "sigma": 0.29632967710494995, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.20233331620693207, "seconds": 659.3} {"step": 158, "loss": 0.32121896743774414, "data": 0.10556155443191528, "gap": 0.21565741300582886, "sigma": 0.8222589492797852, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.3895075023174286, "seconds": 663.5} {"step": 159, "loss": 0.9583067893981934, "data": 0.2773948311805725, "gap": 0.6809119582176208, "sigma": 0.9182301163673401, "kind": "mosaic", "tokens": 194, "dense": 448, "grad_norm": 1.7011202573776245, "seconds": 668.0} {"step": 160, "loss": 2.860076904296875, "data": 0.24819351732730865, "gap": 2.6118834018707275, "sigma": 0.9959679841995239, "kind": "mosaic", "tokens": 189, "dense": 448, "grad_norm": 0.9667125344276428, "seconds": 672.5} {"step": 161, "loss": 0.517736554145813, "data": 0.33004727959632874, "gap": 0.18768924474716187, "sigma": 0.9697849750518799, "kind": "mosaic", "tokens": 264, "dense": 448, "grad_norm": 0.28621476888656616, "seconds": 677.1} {"step": 162, "loss": 0.4934651255607605, "data": 0.21996180713176727, "gap": 0.27350330352783203, "sigma": 0.658065915107727, "kind": "mosaic", "tokens": 238, "dense": 448, "grad_norm": 0.22903022170066833, "seconds": 681.6} {"step": 163, "loss": 1.4290534257888794, "data": 0.23891186714172363, "gap": 1.1901415586471558, "sigma": 0.9900049567222595, "kind": "mosaic", "tokens": 191, "dense": 448, "grad_norm": 0.9212079644203186, "seconds": 686.1} {"step": 164, "loss": 0.6077125072479248, "data": 0.27770712971687317, "gap": 0.33000537753105164, "sigma": 0.9154115915298462, "kind": "mosaic", "tokens": 188, "dense": 448, "grad_norm": 0.3394985795021057, "seconds": 690.6} {"step": 165, "loss": 0.28178757429122925, "data": 0.28178757429122925, "gap": null, "sigma": 0.6754851937294006, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.11036789417266846, "seconds": 693.3} {"step": 166, "loss": 0.2963383197784424, "data": 0.2963383197784424, "gap": null, "sigma": 0.7650582790374756, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.21723487973213196, "seconds": 696.1} {"step": 167, "loss": 2.2623918056488037, "data": 0.5777410864830017, "gap": 1.6846507787704468, "sigma": 0.11992492526769638, "kind": "mosaic", "tokens": 256, "dense": 448, "grad_norm": 1.4750880002975464, "seconds": 700.6} {"step": 168, "loss": 1.155909538269043, "data": 0.31783217191696167, "gap": 0.8380773067474365, "sigma": 0.35013553500175476, "kind": "mosaic", "tokens": 186, "dense": 448, "grad_norm": 0.293810099363327, "seconds": 705.1} {"step": 169, "loss": 0.5775285363197327, "data": 0.23521673679351807, "gap": 0.3423117995262146, "sigma": 0.8990097045898438, "kind": "mosaic", "tokens": 216, "dense": 448, "grad_norm": 0.6904009580612183, "seconds": 709.6} {"step": 170, "loss": 0.5462774634361267, "data": 0.1699894666671753, "gap": 0.3762879967689514, "sigma": 0.6237866878509521, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.20682094991207123, "seconds": 714.1} {"step": 171, "loss": 0.6026087403297424, "data": 0.10517193377017975, "gap": 0.4974367916584015, "sigma": 0.5618345141410828, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.3202956020832062, "seconds": 718.2} {"step": 172, "loss": 0.8532137274742126, "data": 0.24421370029449463, "gap": 0.609000027179718, "sigma": 0.9310535788536072, "kind": "mosaic", "tokens": 228, "dense": 448, "grad_norm": 0.31372544169425964, "seconds": 722.7} {"step": 173, "loss": 0.39852088689804077, "data": 0.19968092441558838, "gap": 0.1988399624824524, "sigma": 0.7928066849708557, "kind": "mosaic", "tokens": 209, "dense": 448, "grad_norm": 0.12792541086673737, "seconds": 727.3} {"step": 174, "loss": 7.733833312988281, "data": 1.7210664749145508, "gap": 6.0127668380737305, "sigma": 0.07108180969953537, "kind": "mosaic", "tokens": 187, "dense": 448, "grad_norm": 5.395622730255127, "seconds": 731.6} {"step": 175, "loss": 1.2434282302856445, "data": 0.2763561010360718, "gap": 0.967072069644928, "sigma": 0.997687578201294, "kind": "mosaic", "tokens": 253, "dense": 448, "grad_norm": 0.8231188654899597, "seconds": 736.0} {"step": 176, "loss": 0.31154686212539673, "data": 0.31154686212539673, "gap": null, "sigma": 0.8773794174194336, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.35981568694114685, "seconds": 738.7} {"step": 177, "loss": 0.18726149201393127, "data": 0.18726149201393127, "gap": null, "sigma": 0.8168197870254517, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2571219503879547, "seconds": 741.4} {"step": 178, "loss": 4.140461444854736, "data": 0.6812040209770203, "gap": 3.4592576026916504, "sigma": 0.18993183970451355, "kind": "uniform", "tokens": 64, "dense": 512, "grad_norm": 3.110297203063965, "seconds": 745.5} {"step": 179, "loss": 3.168130397796631, "data": 0.7192150354385376, "gap": 2.448915481567383, "sigma": 0.173668771982193, "kind": "mosaic", "tokens": 209, "dense": 448, "grad_norm": 1.1865487098693848, "seconds": 750.1} {"step": 180, "loss": 1.2361689805984497, "data": 0.31024834513664246, "gap": 0.9259206056594849, "sigma": 0.9706968665122986, "kind": "mosaic", "tokens": 218, "dense": 448, "grad_norm": 0.7666411399841309, "seconds": 754.5} {"step": 181, "loss": 1.1965672969818115, "data": 0.39471137523651123, "gap": 0.8018559217453003, "sigma": 0.9948006868362427, "kind": "mosaic", "tokens": 229, "dense": 448, "grad_norm": 1.3146076202392578, "seconds": 758.9} {"step": 182, "loss": 0.3604251444339752, "data": 0.04347032681107521, "gap": 0.3169548213481903, "sigma": 0.8551115393638611, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 0.497768372297287, "seconds": 762.9} {"step": 183, "loss": 0.7593560814857483, "data": 0.22325466573238373, "gap": 0.5361014008522034, "sigma": 0.28986653685569763, "kind": "mosaic", "tokens": 160, "dense": 448, "grad_norm": 0.28929629921913147, "seconds": 767.5} {"step": 184, "loss": 0.5635526776313782, "data": 0.25076475739479065, "gap": 0.3127879202365875, "sigma": 0.823826014995575, "kind": "mosaic", "tokens": 219, "dense": 448, "grad_norm": 0.3022436797618866, "seconds": 771.7} {"step": 185, "loss": 0.438984751701355, "data": 0.2059921771287918, "gap": 0.23299255967140198, "sigma": 0.7827209234237671, "kind": "mosaic", "tokens": 257, "dense": 448, "grad_norm": 0.26490095257759094, "seconds": 776.2} {"step": 186, "loss": 0.44357967376708984, "data": 0.21065807342529297, "gap": 0.23292160034179688, "sigma": 0.7722525596618652, "kind": "mosaic", "tokens": 233, "dense": 448, "grad_norm": 0.23225444555282593, "seconds": 780.8} {"step": 187, "loss": 0.5484648942947388, "data": 0.24622631072998047, "gap": 0.3022385537624359, "sigma": 0.7650861740112305, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.39676061272621155, "seconds": 785.2} {"step": 188, "loss": 0.5273564457893372, "data": 0.5273564457893372, "gap": null, "sigma": 0.9320231676101685, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.34656813740730286, "seconds": 788.1} {"step": 189, "loss": 21.067031860351562, "data": 1.712969183921814, "gap": 19.354063034057617, "sigma": 0.07016436010599136, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 16.82317352294922, "seconds": 792.2} {"step": 190, "loss": 0.6341750621795654, "data": 0.18644769489765167, "gap": 0.44772735238075256, "sigma": 0.6623656153678894, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.37600916624069214, "seconds": 796.7} {"step": 191, "loss": 0.731170117855072, "data": 0.2800367474555969, "gap": 0.4511333703994751, "sigma": 0.26420673727989197, "kind": "mosaic", "tokens": 261, "dense": 448, "grad_norm": 0.3303695619106293, "seconds": 801.1} {"step": 192, "loss": 0.5642489194869995, "data": 0.2553991377353668, "gap": 0.3088498115539551, "sigma": 0.8700852394104004, "kind": "mosaic", "tokens": 266, "dense": 448, "grad_norm": 0.3893255293369293, "seconds": 805.7} {"step": 193, "loss": 2.566958427429199, "data": 0.3197973966598511, "gap": 2.2471609115600586, "sigma": 0.999584972858429, "kind": "mosaic", "tokens": 233, "dense": 448, "grad_norm": 1.9555236101150513, "seconds": 810.3} {"step": 194, "loss": 0.41955000162124634, "data": 0.1702376753091812, "gap": 0.24931234121322632, "sigma": 0.635114848613739, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.22625477612018585, "seconds": 814.7} {"step": 195, "loss": 0.47498705983161926, "data": 0.18937942385673523, "gap": 0.28560763597488403, "sigma": 0.6206375956535339, "kind": "mosaic", "tokens": 202, "dense": 448, "grad_norm": 0.16002324223518372, "seconds": 819.1} {"step": 196, "loss": 0.5365963578224182, "data": 0.21605876088142395, "gap": 0.32053759694099426, "sigma": 0.8106085658073425, "kind": "mosaic", "tokens": 193, "dense": 448, "grad_norm": 0.25237753987312317, "seconds": 823.6} {"step": 197, "loss": 0.5570013523101807, "data": 0.19131529331207275, "gap": 0.3656860291957855, "sigma": 0.518254816532135, "kind": "mosaic", "tokens": 199, "dense": 448, "grad_norm": 0.167988121509552, "seconds": 828.1} {"step": 198, "loss": 0.8427335619926453, "data": 0.30743950605392456, "gap": 0.5352940559387207, "sigma": 0.9809545278549194, "kind": "mosaic", "tokens": 292, "dense": 448, "grad_norm": 0.734469473361969, "seconds": 832.8} {"step": 199, "loss": 1.535677194595337, "data": 0.546074628829956, "gap": 0.9896025657653809, "sigma": 0.18608358502388, "kind": "mosaic", "tokens": 238, "dense": 448, "grad_norm": 0.5397521257400513, "seconds": 837.3} {"step": 200, "loss": 1.1364877223968506, "data": 0.2841629683971405, "gap": 0.8523247241973877, "sigma": 0.9085556864738464, "kind": "mosaic", "tokens": 235, "dense": 448, "grad_norm": 0.6272070407867432, "seconds": 841.8} {"step": 201, "loss": 0.45366623997688293, "data": 0.45366623997688293, "gap": null, "sigma": 0.9947360157966614, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.6347963809967041, "seconds": 844.6} {"step": 202, "loss": 0.5992074608802795, "data": 0.20897401869297028, "gap": 0.39023345708847046, "sigma": 0.5631557106971741, "kind": "mosaic", "tokens": 212, "dense": 448, "grad_norm": 0.17740222811698914, "seconds": 849.1} {"step": 203, "loss": 0.5890113115310669, "data": 0.19139249622821808, "gap": 0.39761883020401, "sigma": 0.8877082467079163, "kind": "mosaic", "tokens": 255, "dense": 448, "grad_norm": 1.0488841533660889, "seconds": 853.6} {"step": 204, "loss": 0.6623252630233765, "data": 0.23882848024368286, "gap": 0.4234967827796936, "sigma": 0.5396007299423218, "kind": "mosaic", "tokens": 217, "dense": 448, "grad_norm": 0.3350867033004761, "seconds": 858.2} {"step": 205, "loss": 8.021191596984863, "data": 1.607499599456787, "gap": 6.413691997528076, "sigma": 0.08530023694038391, "kind": "mosaic", "tokens": 247, "dense": 448, "grad_norm": 4.694334506988525, "seconds": 862.7} {"step": 206, "loss": 0.7982033491134644, "data": 0.23437993228435516, "gap": 0.563823401927948, "sigma": 0.7898280620574951, "kind": "mosaic", "tokens": 166, "dense": 448, "grad_norm": 0.4922213852405548, "seconds": 866.9} {"step": 207, "loss": 0.883753776550293, "data": 0.40448683500289917, "gap": 0.4792669117450714, "sigma": 0.9719775319099426, "kind": "mosaic", "tokens": 269, "dense": 448, "grad_norm": 0.8202059268951416, "seconds": 871.4} {"step": 208, "loss": 0.35554876923561096, "data": 0.35554876923561096, "gap": null, "sigma": 0.9958375096321106, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3821500837802887, "seconds": 874.2} {"step": 209, "loss": 0.5422831773757935, "data": 0.21877574920654297, "gap": 0.3235074579715729, "sigma": 0.8072879314422607, "kind": "mosaic", "tokens": 206, "dense": 448, "grad_norm": 0.3853963315486908, "seconds": 878.6} {"step": 210, "loss": 0.405597984790802, "data": 0.2255362570285797, "gap": 0.18006174266338348, "sigma": 0.8324878811836243, "kind": "mosaic", "tokens": 214, "dense": 448, "grad_norm": 0.19357749819755554, "seconds": 883.0} {"step": 211, "loss": 0.5867710113525391, "data": 0.17791998386383057, "gap": 0.4088510572910309, "sigma": 0.8389511108398438, "kind": "mosaic", "tokens": 210, "dense": 448, "grad_norm": 0.2857275605201721, "seconds": 887.3} {"step": 212, "loss": 1.2272435426712036, "data": 0.5474704504013062, "gap": 0.6797730922698975, "sigma": 0.2209552526473999, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.43391337990760803, "seconds": 891.9} {"step": 213, "loss": 0.6440621018409729, "data": 0.24111434817314148, "gap": 0.4029477536678314, "sigma": 0.6118550300598145, "kind": "mosaic", "tokens": 219, "dense": 448, "grad_norm": 0.4053405821323395, "seconds": 896.2} {"step": 214, "loss": 0.3007428050041199, "data": 0.3007428050041199, "gap": null, "sigma": 0.8422703742980957, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.19835273921489716, "seconds": 899.0} {"step": 215, "loss": 0.8486059904098511, "data": 0.2725585401058197, "gap": 0.576047420501709, "sigma": 0.4569619297981262, "kind": "mosaic", "tokens": 202, "dense": 448, "grad_norm": 0.192000150680542, "seconds": 903.6} {"step": 216, "loss": 0.14553941786289215, "data": 0.14553941786289215, "gap": null, "sigma": 0.40634578466415405, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.07238782942295074, "seconds": 906.4} {"step": 217, "loss": 0.6749057769775391, "data": 0.2905949056148529, "gap": 0.38431084156036377, "sigma": 0.50923752784729, "kind": "mosaic", "tokens": 248, "dense": 448, "grad_norm": 0.22035697102546692, "seconds": 911.0} {"step": 218, "loss": 1.073736548423767, "data": 0.3374469578266144, "gap": 0.7362895607948303, "sigma": 0.31084761023521423, "kind": "mosaic", "tokens": 219, "dense": 448, "grad_norm": 0.3174094259738922, "seconds": 915.5} {"step": 219, "loss": 0.8527858853340149, "data": 0.12147413194179535, "gap": 0.7313117384910583, "sigma": 0.8961567282676697, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.8283525109291077, "seconds": 919.8} {"step": 220, "loss": 0.6437791585922241, "data": 0.29499202966690063, "gap": 0.3487870991230011, "sigma": 0.3902435004711151, "kind": "mosaic", "tokens": 244, "dense": 448, "grad_norm": 0.18219275772571564, "seconds": 924.2} {"step": 221, "loss": 0.8522221446037292, "data": 0.2797362208366394, "gap": 0.5724859237670898, "sigma": 0.3623722493648529, "kind": "mosaic", "tokens": 228, "dense": 448, "grad_norm": 0.2211207151412964, "seconds": 928.8} {"step": 222, "loss": 0.3182138204574585, "data": 0.3182138204574585, "gap": null, "sigma": 0.9666451811790466, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.43092313408851624, "seconds": 931.7} {"step": 223, "loss": 0.3888959586620331, "data": 0.3888959586620331, "gap": null, "sigma": 0.9978122115135193, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2627260982990265, "seconds": 934.5} {"step": 224, "loss": 0.6792328357696533, "data": 0.23010095953941345, "gap": 0.44913187623023987, "sigma": 0.8089739680290222, "kind": "mosaic", "tokens": 272, "dense": 448, "grad_norm": 0.5183472037315369, "seconds": 938.9} {"step": 225, "loss": 0.7495555281639099, "data": 0.19001835584640503, "gap": 0.5595371723175049, "sigma": 0.9871394634246826, "kind": "mosaic", "tokens": 242, "dense": 448, "grad_norm": 0.7756423354148865, "seconds": 943.4} {"step": 226, "loss": 0.6912479400634766, "data": 0.07426445186138153, "gap": 0.6169834733009338, "sigma": 0.951616644859314, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.4770926833152771, "seconds": 947.6} {"step": 227, "loss": 1.2155967950820923, "data": 0.1746760606765747, "gap": 1.0409207344055176, "sigma": 0.8999932408332825, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.572147786617279, "seconds": 951.8} {"step": 228, "loss": 0.559857964515686, "data": 0.19497744739055634, "gap": 0.3648805022239685, "sigma": 0.5992234945297241, "kind": "mosaic", "tokens": 157, "dense": 448, "grad_norm": 0.3463054299354553, "seconds": 956.0} {"step": 229, "loss": 0.7871907949447632, "data": 0.23687569797039032, "gap": 0.5503150820732117, "sigma": 0.8348854184150696, "kind": "mosaic", "tokens": 242, "dense": 448, "grad_norm": 0.7639517784118652, "seconds": 960.6} {"step": 230, "loss": 0.5576143860816956, "data": 0.5576143860816956, "gap": null, "sigma": 0.9954399466514587, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.43624022603034973, "seconds": 963.4} {"step": 231, "loss": 0.5559945702552795, "data": 0.5559945702552795, "gap": null, "sigma": 0.9949360489845276, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.5551934838294983, "seconds": 966.2} {"step": 232, "loss": 0.6823974847793579, "data": 0.1874149739742279, "gap": 0.49498251080513, "sigma": 0.5477771759033203, "kind": "mosaic", "tokens": 142, "dense": 448, "grad_norm": 0.1564575582742691, "seconds": 970.4} {"step": 233, "loss": 0.7205120325088501, "data": 0.22419725358486176, "gap": 0.49631479382514954, "sigma": 0.9008174538612366, "kind": "mosaic", "tokens": 292, "dense": 576, "grad_norm": 0.3865128457546234, "seconds": 974.9} {"step": 234, "loss": 3.31353497505188, "data": 0.9342889785766602, "gap": 2.3792459964752197, "sigma": 0.11550905555486679, "kind": "mosaic", "tokens": 194, "dense": 448, "grad_norm": 1.6889599561691284, "seconds": 979.6} {"step": 235, "loss": 0.8802636861801147, "data": 0.22523748874664307, "gap": 0.6550261974334717, "sigma": 0.990145742893219, "kind": "mosaic", "tokens": 193, "dense": 448, "grad_norm": 0.6195921301841736, "seconds": 984.1} {"step": 236, "loss": 0.8828775882720947, "data": 0.32057836651802063, "gap": 0.5622992515563965, "sigma": 0.20232939720153809, "kind": "mosaic", "tokens": 246, "dense": 448, "grad_norm": 0.3861270248889923, "seconds": 988.6} {"step": 237, "loss": 0.8290023803710938, "data": 0.2914925515651703, "gap": 0.5375097990036011, "sigma": 0.8971456289291382, "kind": "mosaic", "tokens": 254, "dense": 448, "grad_norm": 0.6721734404563904, "seconds": 993.0} {"step": 238, "loss": 0.6073449850082397, "data": 0.23468375205993652, "gap": 0.3726612329483032, "sigma": 0.7649084329605103, "kind": "mosaic", "tokens": 239, "dense": 448, "grad_norm": 0.45506444573402405, "seconds": 997.6} {"step": 239, "loss": 0.4652020037174225, "data": 0.2527809143066406, "gap": 0.21242108941078186, "sigma": 0.6171761155128479, "kind": "mosaic", "tokens": 285, "dense": 448, "grad_norm": 0.1703997105360031, "seconds": 1002.2} {"step": 240, "loss": 0.734667956829071, "data": 0.24468272924423218, "gap": 0.48998522758483887, "sigma": 0.9747119545936584, "kind": "mosaic", "tokens": 208, "dense": 448, "grad_norm": 0.3828068673610687, "seconds": 1006.8} {"step": 241, "loss": 0.6142264604568481, "data": 0.25942203402519226, "gap": 0.3548044264316559, "sigma": 0.9242699146270752, "kind": "mosaic", "tokens": 257, "dense": 448, "grad_norm": 0.287966251373291, "seconds": 1011.2} {"step": 242, "loss": 12.536796569824219, "data": 2.6321768760681152, "gap": 9.904620170593262, "sigma": 0.0634167343378067, "kind": "mosaic", "tokens": 207, "dense": 448, "grad_norm": 8.71889591217041, "seconds": 1015.7} {"step": 243, "loss": 0.346904456615448, "data": 0.346904456615448, "gap": null, "sigma": 0.937456488609314, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2252207249403, "seconds": 1018.5} {"step": 244, "loss": 2.5894970893859863, "data": 0.38385847210884094, "gap": 2.2056386470794678, "sigma": 0.24807842075824738, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 97.8486557006836, "seconds": 1022.7} {"step": 245, "loss": 0.631010115146637, "data": 0.1849251240491867, "gap": 0.44608500599861145, "sigma": 0.8771204948425293, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.49887815117836, "seconds": 1027.0} {"step": 246, "loss": 1.4075629711151123, "data": 0.3338907063007355, "gap": 1.0736722946166992, "sigma": 0.9424181580543518, "kind": "mosaic", "tokens": 285, "dense": 576, "grad_norm": 0.44841721653938293, "seconds": 1031.5} {"step": 247, "loss": 0.43190330266952515, "data": 0.2277270257472992, "gap": 0.20417627692222595, "sigma": 0.5705147981643677, "kind": "mosaic", "tokens": 248, "dense": 448, "grad_norm": 0.19411548972129822, "seconds": 1036.0} {"step": 248, "loss": 2.146162986755371, "data": 0.19317497313022614, "gap": 1.9529880285263062, "sigma": 0.32773125171661377, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 0.9036575555801392, "seconds": 1040.2} {"step": 249, "loss": 0.34110477566719055, "data": 0.34110477566719055, "gap": null, "sigma": 0.9647676348686218, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.21120378375053406, "seconds": 1043.1} {"step": 250, "loss": 0.6107117533683777, "data": 0.27678757905960083, "gap": 0.33392417430877686, "sigma": 0.8571346998214722, "kind": "mosaic", "tokens": 236, "dense": 448, "grad_norm": 0.32230836153030396, "seconds": 1047.7} {"step": 250, "eval": {"dense": 0.30757818433145684, "uniform2": 0.18783317475269237, "mosaic": 0.25507365725934505, "gap_uniform2": 0.6375641288856665, "gap_mosaic": 0.6474883891642094}, "eval_extent": {"1x1x1": 0.4128509304450659, "1x1x2": 0.3276627306139008, "1x1x4": 0.23578619867105757, "1x2x1": 0.3035550272281937, "1x2x2": 0.2021679424460193, "1x2x4": 0.13328863113586392, "1x4x1": 0.21017847633888934, "1x4x2": 0.11779781580316248, "1x4x4": 0.0874572291286016}} {"step": 251, "loss": 0.4372028112411499, "data": 0.13830628991127014, "gap": 0.29889652132987976, "sigma": 0.7883396148681641, "kind": "mosaic", "tokens": 173, "dense": 448, "grad_norm": 0.43833446502685547, "seconds": 1293.8} {"step": 252, "loss": 0.905138373374939, "data": 0.054018259048461914, "gap": 0.851120114326477, "sigma": 0.9362566471099854, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 2.279933214187622, "seconds": 1297.9} {"step": 253, "loss": 0.5321667194366455, "data": 0.1777879148721695, "gap": 0.3543788194656372, "sigma": 0.7739601135253906, "kind": "mosaic", "tokens": 181, "dense": 448, "grad_norm": 0.5706455111503601, "seconds": 1302.2} {"step": 254, "loss": 0.6403937935829163, "data": 0.2199188619852066, "gap": 0.42047491669654846, "sigma": 0.553886353969574, "kind": "mosaic", "tokens": 237, "dense": 448, "grad_norm": 0.18563182651996613, "seconds": 1306.8} {"step": 255, "loss": 0.6090810298919678, "data": 0.18153037130832672, "gap": 0.42755067348480225, "sigma": 0.5762924551963806, "kind": "mosaic", "tokens": 191, "dense": 448, "grad_norm": 0.28041645884513855, "seconds": 1311.3} {"step": 256, "loss": 1.0044100284576416, "data": 0.17099808156490326, "gap": 0.8334119319915771, "sigma": 0.4423125088214874, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.34318962693214417, "seconds": 1315.5} {"step": 257, "loss": 2.9128336906433105, "data": 0.7522268891334534, "gap": 2.160606861114502, "sigma": 0.17139457166194916, "kind": "mosaic", "tokens": 243, "dense": 448, "grad_norm": 1.0842069387435913, "seconds": 1320.1} {"step": 258, "loss": 0.6076375246047974, "data": 0.1766541302204132, "gap": 0.43098339438438416, "sigma": 0.7803082466125488, "kind": "mosaic", "tokens": 203, "dense": 448, "grad_norm": 0.7164493799209595, "seconds": 1324.7} {"step": 259, "loss": 0.4440474808216095, "data": 0.4440474808216095, "gap": null, "sigma": 0.9170628786087036, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.21291102468967438, "seconds": 1327.6} {"step": 260, "loss": 0.5439385175704956, "data": 0.12382662296295166, "gap": 0.42011186480522156, "sigma": 0.732705295085907, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 1.7035675048828125, "seconds": 1331.9} {"step": 261, "loss": 1.4733760356903076, "data": 0.2084919810295105, "gap": 1.2648839950561523, "sigma": 0.9820805788040161, "kind": "mosaic", "tokens": 203, "dense": 448, "grad_norm": 0.9798614382743835, "seconds": 1336.5} {"step": 262, "loss": 1.273881196975708, "data": 0.40547361969947815, "gap": 0.8684075474739075, "sigma": 0.2609492838382721, "kind": "mosaic", "tokens": 226, "dense": 448, "grad_norm": 0.3614118695259094, "seconds": 1341.0} {"step": 263, "loss": 0.31690725684165955, "data": 0.11565103381872177, "gap": 0.20125623047351837, "sigma": 0.6664053797721863, "kind": "mosaic", "tokens": 173, "dense": 448, "grad_norm": 0.12405400723218918, "seconds": 1345.3} {"step": 264, "loss": 0.3573039174079895, "data": 0.1616552472114563, "gap": 0.195648655295372, "sigma": 0.38644716143608093, "kind": "mosaic", "tokens": 272, "dense": 448, "grad_norm": 0.2590979337692261, "seconds": 1349.9} {"step": 265, "loss": 0.5249887704849243, "data": 0.18709182739257812, "gap": 0.3378969430923462, "sigma": 0.6598610877990723, "kind": "mosaic", "tokens": 198, "dense": 448, "grad_norm": 0.17447444796562195, "seconds": 1354.4} {"step": 266, "loss": 0.4875723123550415, "data": 0.4875723123550415, "gap": null, "sigma": 0.9794538617134094, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.5830115675926208, "seconds": 1357.2} {"step": 267, "loss": 0.48375850915908813, "data": 0.21532772481441498, "gap": 0.26843076944351196, "sigma": 0.7329633831977844, "kind": "mosaic", "tokens": 210, "dense": 448, "grad_norm": 0.2785542607307434, "seconds": 1361.9} {"step": 268, "loss": 1.1334538459777832, "data": 0.3578414022922516, "gap": 0.775612473487854, "sigma": 0.3179420828819275, "kind": "mosaic", "tokens": 216, "dense": 448, "grad_norm": 0.34439513087272644, "seconds": 1366.4} {"step": 269, "loss": 0.281065970659256, "data": 0.281065970659256, "gap": null, "sigma": 0.7728172540664673, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.5923858284950256, "seconds": 1369.1} {"step": 270, "loss": 0.4830992519855499, "data": 0.4830992519855499, "gap": null, "sigma": 0.9835397005081177, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.47270452976226807, "seconds": 1372.0} {"step": 271, "loss": 0.4130561351776123, "data": 0.21339021623134613, "gap": 0.19966590404510498, "sigma": 0.8106812238693237, "kind": "mosaic", "tokens": 198, "dense": 448, "grad_norm": 0.18891498446464539, "seconds": 1376.5} {"step": 272, "loss": 0.5676478743553162, "data": 0.25828617811203003, "gap": 0.30936169624328613, "sigma": 0.5831453204154968, "kind": "mosaic", "tokens": 250, "dense": 448, "grad_norm": 0.14474479854106903, "seconds": 1381.1} {"step": 273, "loss": 0.33985811471939087, "data": 0.12074713408946991, "gap": 0.21911098062992096, "sigma": 0.6414647102355957, "kind": "mosaic", "tokens": 248, "dense": 448, "grad_norm": 0.18822439014911652, "seconds": 1385.8} {"step": 274, "loss": 0.5361970067024231, "data": 0.5361970067024231, "gap": null, "sigma": 0.9899405837059021, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.5343494415283203, "seconds": 1388.7} {"step": 275, "loss": 0.8053241968154907, "data": 0.29485347867012024, "gap": 0.5104707479476929, "sigma": 0.4273379147052765, "kind": "mosaic", "tokens": 222, "dense": 448, "grad_norm": 0.2748173773288727, "seconds": 1393.2} {"step": 276, "loss": 0.40500977635383606, "data": 0.40500977635383606, "gap": null, "sigma": 0.9611828923225403, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2616569697856903, "seconds": 1395.9} {"step": 277, "loss": 0.6023797392845154, "data": 0.24601870775222778, "gap": 0.3563610315322876, "sigma": 0.6622490286827087, "kind": "mosaic", "tokens": 224, "dense": 448, "grad_norm": 0.21260109543800354, "seconds": 1400.6} {"step": 278, "loss": 0.708319365978241, "data": 0.24748484790325165, "gap": 0.4608345329761505, "sigma": 0.7982119917869568, "kind": "mosaic", "tokens": 274, "dense": 448, "grad_norm": 0.4167706370353699, "seconds": 1405.2} {"step": 279, "loss": 0.29480791091918945, "data": 0.12536612153053284, "gap": 0.16944178938865662, "sigma": 0.5577189326286316, "kind": "mosaic", "tokens": 239, "dense": 448, "grad_norm": 2.0296547412872314, "seconds": 1409.7} {"step": 280, "loss": 0.42509812116622925, "data": 0.2507224380970001, "gap": 0.17437569797039032, "sigma": 0.9045059084892273, "kind": "mosaic", "tokens": 344, "dense": 576, "grad_norm": 0.31011417508125305, "seconds": 1414.4} {"step": 281, "loss": 0.29913052916526794, "data": 0.29913052916526794, "gap": null, "sigma": 0.9409881830215454, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.16950474679470062, "seconds": 1417.3} {"step": 282, "loss": 0.5759353637695312, "data": 0.25569775700569153, "gap": 0.32023757696151733, "sigma": 0.4764254093170166, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.22712457180023193, "seconds": 1421.8} {"step": 283, "loss": 0.7067773938179016, "data": 0.12353949248790741, "gap": 0.583237886428833, "sigma": 0.6010485291481018, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.35070157051086426, "seconds": 1426.1} {"step": 284, "loss": 6.895668983459473, "data": 1.5535292625427246, "gap": 5.342139720916748, "sigma": 0.005293257534503937, "kind": "mosaic", "tokens": 279, "dense": 448, "grad_norm": 6.9344892501831055, "seconds": 1430.7} {"step": 285, "loss": 0.5278677940368652, "data": 0.2661702632904053, "gap": 0.26169756054878235, "sigma": 0.8982024788856506, "kind": "mosaic", "tokens": 233, "dense": 448, "grad_norm": 0.2267053723335266, "seconds": 1435.3} {"step": 286, "loss": 0.6315478086471558, "data": 0.23653048276901245, "gap": 0.3950172960758209, "sigma": 0.6375497579574585, "kind": "mosaic", "tokens": 240, "dense": 448, "grad_norm": 0.17144803702831268, "seconds": 1439.8} {"step": 287, "loss": 0.962918758392334, "data": 0.30350372195243835, "gap": 0.659415066242218, "sigma": 0.39004430174827576, "kind": "mosaic", "tokens": 191, "dense": 448, "grad_norm": 0.2389390766620636, "seconds": 1444.3} {"step": 288, "loss": 0.45056718587875366, "data": 0.1584319919347763, "gap": 0.29213517904281616, "sigma": 0.8518177270889282, "kind": "mosaic", "tokens": 171, "dense": 448, "grad_norm": 0.7682028412818909, "seconds": 1448.8} {"step": 289, "loss": 0.6879714131355286, "data": 0.16603298485279083, "gap": 0.5219384431838989, "sigma": 0.9108604192733765, "kind": "mosaic", "tokens": 231, "dense": 448, "grad_norm": 1.1996911764144897, "seconds": 1453.2} {"step": 290, "loss": 0.6144726276397705, "data": 0.22730904817581177, "gap": 0.38716360926628113, "sigma": 0.9415664672851562, "kind": "mosaic", "tokens": 241, "dense": 448, "grad_norm": 0.5323389768600464, "seconds": 1457.7} {"step": 291, "loss": 0.3835620880126953, "data": 0.3835620880126953, "gap": null, "sigma": 0.8634136319160461, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2847796380519867, "seconds": 1460.5} {"step": 292, "loss": 0.4712546169757843, "data": 0.18700537085533142, "gap": 0.2842492461204529, "sigma": 0.9357562065124512, "kind": "mosaic", "tokens": 225, "dense": 448, "grad_norm": 0.28587138652801514, "seconds": 1464.9} {"step": 293, "loss": 0.5392004251480103, "data": 0.24591493606567383, "gap": 0.29328545928001404, "sigma": 0.7433866858482361, "kind": "mosaic", "tokens": 252, "dense": 448, "grad_norm": 0.21908524632453918, "seconds": 1469.4} {"step": 294, "loss": 0.5312210321426392, "data": 0.5312210321426392, "gap": null, "sigma": 0.9821851253509521, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.7950584292411804, "seconds": 1472.2} {"step": 295, "loss": 3.1244029998779297, "data": 0.7125664949417114, "gap": 2.4118363857269287, "sigma": 0.14683914184570312, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 2.539074182510376, "seconds": 1476.5} {"step": 296, "loss": 0.25785672664642334, "data": 0.25785672664642334, "gap": null, "sigma": 0.8311596512794495, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.14512762427330017, "seconds": 1479.5} {"step": 297, "loss": 0.7463303804397583, "data": 0.13684459030628204, "gap": 0.6094858050346375, "sigma": 0.5881388783454895, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.28587213158607483, "seconds": 1483.7} {"step": 298, "loss": 0.7401875257492065, "data": 0.217086523771286, "gap": 0.5231009721755981, "sigma": 0.5242453813552856, "kind": "mosaic", "tokens": 172, "dense": 448, "grad_norm": 0.24010604619979858, "seconds": 1488.1} {"step": 299, "loss": 0.4557393789291382, "data": 0.4557393789291382, "gap": null, "sigma": 0.9086105823516846, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3198341131210327, "seconds": 1491.0} {"step": 300, "loss": 0.39674079418182373, "data": 0.2354382574558258, "gap": 0.16130255162715912, "sigma": 0.7345548868179321, "kind": "mosaic", "tokens": 260, "dense": 448, "grad_norm": 0.24555093050003052, "seconds": 1495.5} {"step": 301, "loss": 0.9006186723709106, "data": 0.24221870303153992, "gap": 0.6583999991416931, "sigma": 0.4945265054702759, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.2428838163614273, "seconds": 1499.7} {"step": 302, "loss": 0.43032756447792053, "data": 0.43032756447792053, "gap": null, "sigma": 0.9795109033584595, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3436446487903595, "seconds": 1502.5} {"step": 303, "loss": 0.5251606702804565, "data": 0.2563377916812897, "gap": 0.26882287859916687, "sigma": 0.845496654510498, "kind": "mosaic", "tokens": 223, "dense": 448, "grad_norm": 0.24554666876792908, "seconds": 1507.1} {"step": 304, "loss": 0.6635675430297852, "data": 0.27430570125579834, "gap": 0.3892618417739868, "sigma": 0.5134761333465576, "kind": "mosaic", "tokens": 217, "dense": 448, "grad_norm": 0.1938241720199585, "seconds": 1511.7} {"step": 305, "loss": 0.40690523386001587, "data": 0.17316660284996033, "gap": 0.23373861610889435, "sigma": 0.403530091047287, "kind": "mosaic", "tokens": 255, "dense": 448, "grad_norm": 0.15994156897068024, "seconds": 1516.2} {"step": 306, "loss": 0.6547458171844482, "data": 0.06476639956235886, "gap": 0.5899794101715088, "sigma": 0.8537775278091431, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.38285398483276367, "seconds": 1520.5} {"step": 307, "loss": 1.223478078842163, "data": 0.37528854608535767, "gap": 0.8481895923614502, "sigma": 0.9959495663642883, "kind": "mosaic", "tokens": 271, "dense": 448, "grad_norm": 0.5066401362419128, "seconds": 1525.0} {"step": 308, "loss": 0.4937276244163513, "data": 0.20433975756168365, "gap": 0.28938785195350647, "sigma": 0.7760845422744751, "kind": "mosaic", "tokens": 255, "dense": 448, "grad_norm": 0.3363538086414337, "seconds": 1529.3} {"step": 309, "loss": 0.7350465059280396, "data": 0.091682069003582, "gap": 0.643364429473877, "sigma": 0.7700542211532593, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.5297286510467529, "seconds": 1533.5} {"step": 310, "loss": 0.6385846138000488, "data": 0.3081343173980713, "gap": 0.3304503262042999, "sigma": 0.95588618516922, "kind": "mosaic", "tokens": 249, "dense": 448, "grad_norm": 0.45417892932891846, "seconds": 1538.2} {"step": 311, "loss": 0.7161381244659424, "data": 0.2558789849281311, "gap": 0.4602591395378113, "sigma": 0.46200740337371826, "kind": "mosaic", "tokens": 259, "dense": 576, "grad_norm": 0.13006427884101868, "seconds": 1542.8} {"step": 312, "loss": 0.36733677983283997, "data": 0.36733677983283997, "gap": null, "sigma": 0.9647648930549622, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.19516079127788544, "seconds": 1545.7} {"step": 313, "loss": 0.4874264895915985, "data": 0.4874264895915985, "gap": null, "sigma": 0.9371665716171265, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2023429423570633, "seconds": 1548.6} {"step": 314, "loss": 0.7477830648422241, "data": 0.21483898162841797, "gap": 0.5329440832138062, "sigma": 0.9501174688339233, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.4737938940525055, "seconds": 1552.9} {"step": 315, "loss": 0.19459784030914307, "data": 0.19459784030914307, "gap": null, "sigma": 0.5617126822471619, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.10642693936824799, "seconds": 1555.7} {"step": 316, "loss": 0.7304381132125854, "data": 0.26645487546920776, "gap": 0.4639832377433777, "sigma": 0.49355149269104004, "kind": "mosaic", "tokens": 225, "dense": 448, "grad_norm": 0.16012455523014069, "seconds": 1560.3} {"step": 317, "loss": 1.7750639915466309, "data": 0.49814489483833313, "gap": 1.2769191265106201, "sigma": 0.22518815100193024, "kind": "mosaic", "tokens": 225, "dense": 448, "grad_norm": 0.5972217917442322, "seconds": 1564.9} {"step": 318, "loss": 0.5809921026229858, "data": 0.24587313830852509, "gap": 0.33511897921562195, "sigma": 0.5985762476921082, "kind": "mosaic", "tokens": 220, "dense": 448, "grad_norm": 0.18419867753982544, "seconds": 1569.4} {"step": 319, "loss": 0.5586219429969788, "data": 0.0479971244931221, "gap": 0.5106248259544373, "sigma": 0.7729633450508118, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 0.49441099166870117, "seconds": 1573.4} {"step": 320, "loss": 1.2143923044204712, "data": 0.16194990277290344, "gap": 1.0524424314498901, "sigma": 0.907882034778595, "kind": "mosaic", "tokens": 169, "dense": 448, "grad_norm": 0.5977810621261597, "seconds": 1577.7} {"step": 321, "loss": 1.2573097944259644, "data": 0.2830694317817688, "gap": 0.9742403626441956, "sigma": 0.9522266387939453, "kind": "mosaic", "tokens": 241, "dense": 448, "grad_norm": 0.4250909984111786, "seconds": 1582.1} {"step": 322, "loss": 0.7091445326805115, "data": 0.21069113910198212, "gap": 0.49845340847969055, "sigma": 0.9534180164337158, "kind": "mosaic", "tokens": 191, "dense": 448, "grad_norm": 0.32404547929763794, "seconds": 1586.6} {"step": 323, "loss": 0.27496537566185, "data": 0.27496537566185, "gap": null, "sigma": 0.9674642086029053, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.21847650408744812, "seconds": 1589.4} {"step": 324, "loss": 0.8202602863311768, "data": 0.29321834444999695, "gap": 0.5270419120788574, "sigma": 0.9513542056083679, "kind": "mosaic", "tokens": 267, "dense": 448, "grad_norm": 0.4019835889339447, "seconds": 1594.0} {"step": 325, "loss": 3.048384428024292, "data": 0.48790034651756287, "gap": 2.5604841709136963, "sigma": 0.23645898699760437, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 1.0677248239517212, "seconds": 1598.0} {"step": 326, "loss": 0.34791189432144165, "data": 0.1791282743215561, "gap": 0.16878363490104675, "sigma": 0.7357252240180969, "kind": "mosaic", "tokens": 247, "dense": 448, "grad_norm": 0.20896805822849274, "seconds": 1602.5} {"step": 327, "loss": 0.8058457970619202, "data": 0.20399384200572968, "gap": 0.6018519401550293, "sigma": 0.7571561336517334, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.9850484132766724, "seconds": 1606.6} {"step": 328, "loss": 0.38086265325546265, "data": 0.14915713667869568, "gap": 0.23170553147792816, "sigma": 0.6431710720062256, "kind": "mosaic", "tokens": 254, "dense": 448, "grad_norm": 0.11619003117084503, "seconds": 1611.1} {"step": 329, "loss": 1.1912992000579834, "data": 0.31486260890960693, "gap": 0.8764366507530212, "sigma": 0.9713262319564819, "kind": "mosaic", "tokens": 221, "dense": 448, "grad_norm": 0.9715947508811951, "seconds": 1615.6} {"step": 330, "loss": 0.3381865918636322, "data": 0.12513844668865204, "gap": 0.21304814517498016, "sigma": 0.7748801708221436, "kind": "mosaic", "tokens": 191, "dense": 448, "grad_norm": 0.260044127702713, "seconds": 1620.3} {"step": 331, "loss": 0.6529961824417114, "data": 0.2754485607147217, "gap": 0.37754759192466736, "sigma": 0.9263343811035156, "kind": "mosaic", "tokens": 255, "dense": 448, "grad_norm": 0.7032442092895508, "seconds": 1624.9} {"step": 332, "loss": 0.5086123943328857, "data": 0.24859823286533356, "gap": 0.260014146566391, "sigma": 0.9476487636566162, "kind": "mosaic", "tokens": 208, "dense": 448, "grad_norm": 0.29790782928466797, "seconds": 1629.4} {"step": 333, "loss": 0.8151348233222961, "data": 0.38083964586257935, "gap": 0.4342951774597168, "sigma": 0.3161226511001587, "kind": "mosaic", "tokens": 292, "dense": 480, "grad_norm": 0.17338688671588898, "seconds": 1634.0} {"step": 334, "loss": 1.6334186792373657, "data": 0.4701533615589142, "gap": 1.163265347480774, "sigma": 0.2905557155609131, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.7570621967315674, "seconds": 1638.2} {"step": 335, "loss": 1.2146364450454712, "data": 0.23476023972034454, "gap": 0.9798761606216431, "sigma": 0.9344934225082397, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.6294605135917664, "seconds": 1642.7} {"step": 336, "loss": 0.4247748851776123, "data": 0.19324974715709686, "gap": 0.23152515292167664, "sigma": 0.7009024024009705, "kind": "mosaic", "tokens": 237, "dense": 448, "grad_norm": 0.30030038952827454, "seconds": 1647.3} {"step": 337, "loss": 0.5482196807861328, "data": 0.1523682326078415, "gap": 0.3958514630794525, "sigma": 0.7597953081130981, "kind": "mosaic", "tokens": 180, "dense": 448, "grad_norm": 0.3359658420085907, "seconds": 1651.7} {"step": 338, "loss": 0.3926600217819214, "data": 0.20489883422851562, "gap": 0.18776118755340576, "sigma": 0.7092301845550537, "kind": "mosaic", "tokens": 268, "dense": 448, "grad_norm": 0.19282080233097076, "seconds": 1656.3} {"step": 339, "loss": 25.15540313720703, "data": 3.732485055923462, "gap": 21.42291831970215, "sigma": 0.05289802327752113, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 20.039106369018555, "seconds": 1660.6} {"step": 340, "loss": 1.0131604671478271, "data": 0.239837646484375, "gap": 0.7733227610588074, "sigma": 0.9952735304832458, "kind": "mosaic", "tokens": 264, "dense": 448, "grad_norm": 1.6455154418945312, "seconds": 1665.2} {"step": 341, "loss": 0.6310325860977173, "data": 0.2346152365207672, "gap": 0.3964173495769501, "sigma": 0.5689442753791809, "kind": "mosaic", "tokens": 169, "dense": 448, "grad_norm": 0.16031228005886078, "seconds": 1669.8} {"step": 342, "loss": 0.5722355246543884, "data": 0.5722355246543884, "gap": null, "sigma": 0.043690044432878494, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.13918590545654297, "seconds": 1672.5} {"step": 343, "loss": 0.395765095949173, "data": 0.395765095949173, "gap": null, "sigma": 0.9804736375808716, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.28801971673965454, "seconds": 1675.4} {"step": 344, "loss": 0.39880889654159546, "data": 0.39880889654159546, "gap": null, "sigma": 0.9759591221809387, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2773137092590332, "seconds": 1678.3} {"step": 345, "loss": 0.798069953918457, "data": 0.11610617488622665, "gap": 0.6819638013839722, "sigma": 0.8865379691123962, "kind": "mosaic", "tokens": 176, "dense": 448, "grad_norm": 0.8122803568840027, "seconds": 1682.7} {"step": 346, "loss": 0.3333757221698761, "data": 0.3333757221698761, "gap": null, "sigma": 0.7427453994750977, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.09206278622150421, "seconds": 1685.6} {"step": 347, "loss": 3.2942428588867188, "data": 0.45408380031585693, "gap": 2.8401591777801514, "sigma": 0.987452507019043, "kind": "mosaic", "tokens": 236, "dense": 448, "grad_norm": 1.244240641593933, "seconds": 1690.2} {"step": 348, "loss": 0.8484736680984497, "data": 0.20880945026874542, "gap": 0.6396642327308655, "sigma": 0.8272638320922852, "kind": "mosaic", "tokens": 220, "dense": 448, "grad_norm": 0.50182044506073, "seconds": 1694.8} {"step": 349, "loss": 1.5371595621109009, "data": 0.3791566789150238, "gap": 1.1580028533935547, "sigma": 0.9846264719963074, "kind": "mosaic", "tokens": 254, "dense": 448, "grad_norm": 1.0034940242767334, "seconds": 1699.4} {"step": 350, "loss": 1.1967411041259766, "data": 0.18967880308628082, "gap": 1.007062315940857, "sigma": 0.9757298231124878, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.9663933515548706, "seconds": 1704.0} {"step": 351, "loss": 0.43533971905708313, "data": 0.1754048764705658, "gap": 0.25993484258651733, "sigma": 0.7951448559761047, "kind": "mosaic", "tokens": 217, "dense": 448, "grad_norm": 0.2692643105983734, "seconds": 1708.5} {"step": 352, "loss": 0.5269989967346191, "data": 0.1584351360797882, "gap": 0.36856383085250854, "sigma": 0.854241132736206, "kind": "mosaic", "tokens": 177, "dense": 448, "grad_norm": 0.31177106499671936, "seconds": 1713.2} {"step": 353, "loss": 0.33698245882987976, "data": 0.33698245882987976, "gap": null, "sigma": 0.8627971410751343, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.43504419922828674, "seconds": 1716.0} {"step": 354, "loss": 0.748405396938324, "data": 0.19436614215373993, "gap": 0.5540392398834229, "sigma": 0.7582835555076599, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.6610565781593323, "seconds": 1720.2} {"step": 355, "loss": 0.49298128485679626, "data": 0.49298128485679626, "gap": null, "sigma": 0.987938642501831, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3894905149936676, "seconds": 1723.1} {"step": 356, "loss": 0.7593151330947876, "data": 0.21380801498889923, "gap": 0.5455071330070496, "sigma": 0.8731184005737305, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.654954731464386, "seconds": 1727.7} {"step": 357, "loss": 10.129827499389648, "data": 1.758139729499817, "gap": 8.371687889099121, "sigma": 0.07278608530759811, "kind": "mosaic", "tokens": 208, "dense": 448, "grad_norm": 6.757364749908447, "seconds": 1732.3} {"step": 358, "loss": 0.6649446487426758, "data": 0.14449571073055267, "gap": 0.5204489231109619, "sigma": 0.8455610275268555, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.4814606308937073, "seconds": 1736.6} {"step": 359, "loss": 0.4552561938762665, "data": 0.15472504496574402, "gap": 0.30053114891052246, "sigma": 0.8130870461463928, "kind": "mosaic", "tokens": 245, "dense": 448, "grad_norm": 0.3058438301086426, "seconds": 1741.1} {"step": 360, "loss": 0.44212663173675537, "data": 0.07625972479581833, "gap": 0.36586689949035645, "sigma": 0.8953219652175903, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.4352197051048279, "seconds": 1745.3} {"step": 361, "loss": 0.41244375705718994, "data": 0.1819961816072464, "gap": 0.23044759035110474, "sigma": 0.8242061138153076, "kind": "mosaic", "tokens": 150, "dense": 448, "grad_norm": 0.28877946734428406, "seconds": 1749.4} {"step": 362, "loss": 1.4495817422866821, "data": 0.28601059317588806, "gap": 1.1635711193084717, "sigma": 0.9931473135948181, "kind": "mosaic", "tokens": 254, "dense": 448, "grad_norm": 1.811444640159607, "seconds": 1754.2} {"step": 363, "loss": 0.6902762651443481, "data": 0.2436809539794922, "gap": 0.44659531116485596, "sigma": 0.9095615148544312, "kind": "mosaic", "tokens": 227, "dense": 448, "grad_norm": 0.44408702850341797, "seconds": 1758.7} {"step": 364, "loss": 1.0515000820159912, "data": 0.26858752965927124, "gap": 0.7829126119613647, "sigma": 0.9705040454864502, "kind": "mosaic", "tokens": 208, "dense": 448, "grad_norm": 1.0892678499221802, "seconds": 1763.2} {"step": 365, "loss": 0.4523703455924988, "data": 0.4523703455924988, "gap": null, "sigma": 0.9896461963653564, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.4011242091655731, "seconds": 1766.0} {"step": 366, "loss": 0.8418399095535278, "data": 0.27956777811050415, "gap": 0.5622721314430237, "sigma": 0.4849911332130432, "kind": "mosaic", "tokens": 187, "dense": 448, "grad_norm": 0.19253025949001312, "seconds": 1770.5} {"step": 367, "loss": 13.492177963256836, "data": 4.601916313171387, "gap": 8.89026165008545, "sigma": 0.030234476551413536, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 15.657668113708496, "seconds": 1774.8} {"step": 368, "loss": 0.45722663402557373, "data": 0.24420715868473053, "gap": 0.2130194902420044, "sigma": 0.8323892951011658, "kind": "mosaic", "tokens": 267, "dense": 448, "grad_norm": 0.19282649457454681, "seconds": 1779.4} {"step": 369, "loss": 0.5978840589523315, "data": 0.2580778896808624, "gap": 0.3398061692714691, "sigma": 0.6903886198997498, "kind": "mosaic", "tokens": 227, "dense": 448, "grad_norm": 0.19645021855831146, "seconds": 1784.0} {"step": 370, "loss": 0.4555921256542206, "data": 0.4555921256542206, "gap": null, "sigma": 0.9156671762466431, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.19014836847782135, "seconds": 1786.7} {"step": 371, "loss": 1.1502443552017212, "data": 0.3542303442955017, "gap": 0.7960140109062195, "sigma": 0.2860487997531891, "kind": "mosaic", "tokens": 243, "dense": 448, "grad_norm": 0.4686652421951294, "seconds": 1791.2} {"step": 372, "loss": 0.8059353232383728, "data": 0.28032994270324707, "gap": 0.5256053805351257, "sigma": 0.25746455788612366, "kind": "mosaic", "tokens": 229, "dense": 448, "grad_norm": 0.34843531250953674, "seconds": 1795.8} {"step": 373, "loss": 1.1468218564987183, "data": 0.10366084426641464, "gap": 1.0431610345840454, "sigma": 0.34592077136039734, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 0.611047625541687, "seconds": 1799.9} {"step": 374, "loss": 0.46767640113830566, "data": 0.46767640113830566, "gap": null, "sigma": 0.9608465433120728, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2869615852832794, "seconds": 1802.7} {"step": 375, "loss": 1.2018051147460938, "data": 0.34563249349594116, "gap": 0.8561726212501526, "sigma": 0.33055606484413147, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.5495043396949768, "seconds": 1807.1} {"step": 376, "loss": 0.8107073307037354, "data": 0.2476406991481781, "gap": 0.5630666017532349, "sigma": 0.6017033457756042, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.3135278820991516, "seconds": 1811.5} {"step": 377, "loss": 0.5488544702529907, "data": 0.24659736454486847, "gap": 0.30225712060928345, "sigma": 0.9469717741012573, "kind": "mosaic", "tokens": 260, "dense": 448, "grad_norm": 0.33347266912460327, "seconds": 1816.1} {"step": 378, "loss": 0.5348591208457947, "data": 0.21630863845348358, "gap": 0.3185504972934723, "sigma": 0.34987160563468933, "kind": "mosaic", "tokens": 245, "dense": 448, "grad_norm": 0.17553476989269257, "seconds": 1820.5} {"step": 379, "loss": 7.221506595611572, "data": 1.6060256958007812, "gap": 5.615480899810791, "sigma": 0.07748499512672424, "kind": "mosaic", "tokens": 176, "dense": 448, "grad_norm": 4.04480504989624, "seconds": 1825.2} {"step": 380, "loss": 0.49808353185653687, "data": 0.24780943989753723, "gap": 0.25027409195899963, "sigma": 0.6043687462806702, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.19077685475349426, "seconds": 1829.7} {"step": 381, "loss": 0.7303680181503296, "data": 0.12948158383369446, "gap": 0.6008864641189575, "sigma": 0.6349045038223267, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.24791480600833893, "seconds": 1834.0} {"step": 382, "loss": 2.3588109016418457, "data": 0.7028156518936157, "gap": 1.65599524974823, "sigma": 0.17963731288909912, "kind": "mosaic", "tokens": 213, "dense": 448, "grad_norm": 0.9340973496437073, "seconds": 1838.4} {"step": 383, "loss": 0.7181691527366638, "data": 0.23773683607578278, "gap": 0.48043230175971985, "sigma": 0.4716298580169678, "kind": "mosaic", "tokens": 233, "dense": 448, "grad_norm": 0.19792024791240692, "seconds": 1842.9} {"step": 384, "loss": 0.3983939290046692, "data": 0.12758877873420715, "gap": 0.27080515027046204, "sigma": 0.9547759294509888, "kind": "mosaic", "tokens": 178, "dense": 448, "grad_norm": 0.6526463627815247, "seconds": 1847.3} {"step": 385, "loss": 1.0979405641555786, "data": 0.2872186303138733, "gap": 0.8107219338417053, "sigma": 0.97437584400177, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.6765439510345459, "seconds": 1851.7} {"step": 386, "loss": 0.7506192922592163, "data": 0.3603571653366089, "gap": 0.39026209712028503, "sigma": 0.9697842597961426, "kind": "mosaic", "tokens": 236, "dense": 448, "grad_norm": 0.6287931203842163, "seconds": 1856.1} {"step": 387, "loss": 0.5353881120681763, "data": 0.18238505721092224, "gap": 0.35300305485725403, "sigma": 0.7825959920883179, "kind": "mosaic", "tokens": 191, "dense": 448, "grad_norm": 0.9200378656387329, "seconds": 1860.8} {"step": 388, "loss": 0.35550594329833984, "data": 0.35550594329833984, "gap": null, "sigma": 0.896433413028717, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.21747025847434998, "seconds": 1863.5} {"step": 389, "loss": 0.6374740600585938, "data": 0.19474661350250244, "gap": 0.4427274167537689, "sigma": 0.7605071663856506, "kind": "mosaic", "tokens": 161, "dense": 448, "grad_norm": 0.2523884177207947, "seconds": 1867.9} {"step": 390, "loss": 0.4768678545951843, "data": 0.1432790607213974, "gap": 0.3335888087749481, "sigma": 0.7570829391479492, "kind": "mosaic", "tokens": 154, "dense": 448, "grad_norm": 0.39649441838264465, "seconds": 1872.2} {"step": 391, "loss": 1.038119912147522, "data": 0.027959972620010376, "gap": 1.010159969329834, "sigma": 0.9850722551345825, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 2.207273006439209, "seconds": 1876.3} {"step": 392, "loss": 0.26090317964553833, "data": 0.26090317964553833, "gap": null, "sigma": 0.7762604355812073, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.4830061197280884, "seconds": 1879.1} {"step": 393, "loss": 2.1589443683624268, "data": 0.3635757267475128, "gap": 1.7953685522079468, "sigma": 0.27861154079437256, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.7041717767715454, "seconds": 1883.4} {"step": 394, "loss": 1.5669245719909668, "data": 0.35458996891975403, "gap": 1.2123346328735352, "sigma": 0.2579236924648285, "kind": "mosaic", "tokens": 165, "dense": 448, "grad_norm": 0.5782503485679626, "seconds": 1887.5} {"step": 395, "loss": 0.4760229289531708, "data": 0.14773878455162048, "gap": 0.3282841444015503, "sigma": 0.6942564249038696, "kind": "mosaic", "tokens": 169, "dense": 448, "grad_norm": 0.2516288459300995, "seconds": 1891.8} {"step": 396, "loss": 1.0717092752456665, "data": 0.2557303011417389, "gap": 0.8159789443016052, "sigma": 0.999140202999115, "kind": "mosaic", "tokens": 197, "dense": 448, "grad_norm": 1.4198875427246094, "seconds": 1896.3} {"step": 397, "loss": 0.3892132043838501, "data": 0.1584966480731964, "gap": 0.2307165563106537, "sigma": 0.6975432634353638, "kind": "mosaic", "tokens": 201, "dense": 448, "grad_norm": 0.18732543289661407, "seconds": 1900.8} {"step": 398, "loss": 0.1954939365386963, "data": 0.09468121826648712, "gap": 0.10081271827220917, "sigma": 0.6197863221168518, "kind": "mosaic", "tokens": 251, "dense": 448, "grad_norm": 0.18334127962589264, "seconds": 1905.2} {"step": 399, "loss": 0.4559931755065918, "data": 0.22878269851207733, "gap": 0.22721047699451447, "sigma": 0.6756412982940674, "kind": "mosaic", "tokens": 250, "dense": 448, "grad_norm": 0.13824835419654846, "seconds": 1909.8} {"step": 400, "loss": 0.5684159398078918, "data": 0.10161026567220688, "gap": 0.46680566668510437, "sigma": 0.6709287166595459, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.24475954473018646, "seconds": 1913.9} {"step": 401, "loss": 0.4698932468891144, "data": 0.4698932468891144, "gap": null, "sigma": 0.9516916275024414, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 1.1506520509719849, "seconds": 1916.8} {"step": 402, "loss": 0.683815598487854, "data": 0.2530786097049713, "gap": 0.4307369589805603, "sigma": 0.7917740941047668, "kind": "mosaic", "tokens": 211, "dense": 448, "grad_norm": 0.2717921733856201, "seconds": 1921.2} {"step": 403, "loss": 1.1206833124160767, "data": 0.28573697805404663, "gap": 0.83494633436203, "sigma": 0.9463574886322021, "kind": "mosaic", "tokens": 253, "dense": 448, "grad_norm": 0.9063707590103149, "seconds": 1925.9} {"step": 404, "loss": 0.5529130101203918, "data": 0.17954973876476288, "gap": 0.37336328625679016, "sigma": 0.916120171546936, "kind": "mosaic", "tokens": 239, "dense": 448, "grad_norm": 0.22091078758239746, "seconds": 1930.2} {"step": 405, "loss": 0.6239154934883118, "data": 0.21296261250972748, "gap": 0.4109528660774231, "sigma": 0.6501662731170654, "kind": "mosaic", "tokens": 177, "dense": 448, "grad_norm": 0.21162542700767517, "seconds": 1934.7} {"step": 406, "loss": 0.8370000123977661, "data": 0.28517642617225647, "gap": 0.551823616027832, "sigma": 0.961632251739502, "kind": "mosaic", "tokens": 232, "dense": 448, "grad_norm": 0.9145351648330688, "seconds": 1939.3} {"step": 407, "loss": 1.4305355548858643, "data": 0.2709117531776428, "gap": 1.1596237421035767, "sigma": 0.8984763622283936, "kind": "mosaic", "tokens": 234, "dense": 448, "grad_norm": 0.3253244161605835, "seconds": 1943.8} {"step": 408, "loss": 0.419537752866745, "data": 0.21601863205432892, "gap": 0.20351912081241608, "sigma": 0.7061851024627686, "kind": "mosaic", "tokens": 276, "dense": 448, "grad_norm": 0.25841566920280457, "seconds": 1948.2} {"step": 409, "loss": 0.6332957744598389, "data": 0.268278568983078, "gap": 0.3650171756744385, "sigma": 0.9371362328529358, "kind": "mosaic", "tokens": 211, "dense": 448, "grad_norm": 0.5488366484642029, "seconds": 1952.7} {"step": 410, "loss": 0.4149998426437378, "data": 0.16059009730815887, "gap": 0.2544097602367401, "sigma": 0.6486701965332031, "kind": "mosaic", "tokens": 195, "dense": 448, "grad_norm": 0.17736244201660156, "seconds": 1957.2} {"step": 411, "loss": 0.40099990367889404, "data": 0.14090918004512787, "gap": 0.260090708732605, "sigma": 0.5436296463012695, "kind": "mosaic", "tokens": 243, "dense": 448, "grad_norm": 0.27970078587532043, "seconds": 1961.7} {"step": 412, "loss": 0.985715389251709, "data": 0.31506672501564026, "gap": 0.6706486940383911, "sigma": 0.9527208805084229, "kind": "mosaic", "tokens": 223, "dense": 448, "grad_norm": 0.46087220311164856, "seconds": 1966.0} {"step": 413, "loss": 0.7166297435760498, "data": 0.34068354964256287, "gap": 0.3759462237358093, "sigma": 0.9242942929267883, "kind": "mosaic", "tokens": 260, "dense": 448, "grad_norm": 0.3951287865638733, "seconds": 1970.4} {"step": 414, "loss": 0.8755292296409607, "data": 0.13162051141262054, "gap": 0.743908703327179, "sigma": 0.6721863746643066, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.9753072261810303, "seconds": 1974.5} {"step": 415, "loss": 0.43891316652297974, "data": 0.1880326122045517, "gap": 0.25088056921958923, "sigma": 0.8335139751434326, "kind": "mosaic", "tokens": 239, "dense": 448, "grad_norm": 0.21817509829998016, "seconds": 1979.0} {"step": 416, "loss": 1.0183571577072144, "data": 0.3656502366065979, "gap": 0.6527069211006165, "sigma": 0.3624064028263092, "kind": "mosaic", "tokens": 260, "dense": 448, "grad_norm": 0.2716214060783386, "seconds": 1983.5} {"step": 417, "loss": 0.45607808232307434, "data": 0.45607808232307434, "gap": null, "sigma": 0.9843724966049194, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.29933905601501465, "seconds": 1986.3} {"step": 418, "loss": 0.28402918577194214, "data": 0.28402918577194214, "gap": null, "sigma": 0.9177502393722534, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3493815064430237, "seconds": 1989.2} {"step": 419, "loss": 0.5308923125267029, "data": 0.2117038518190384, "gap": 0.3191884458065033, "sigma": 0.681638240814209, "kind": "mosaic", "tokens": 270, "dense": 448, "grad_norm": 0.31936419010162354, "seconds": 1993.7} {"step": 420, "loss": 0.7195339202880859, "data": 0.33110949397087097, "gap": 0.38842442631721497, "sigma": 0.4103640615940094, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.4057977795600891, "seconds": 1998.2} {"step": 421, "loss": 1.7916820049285889, "data": 0.5926019549369812, "gap": 1.199079990386963, "sigma": 0.1723010241985321, "kind": "mosaic", "tokens": 269, "dense": 448, "grad_norm": 0.7723122835159302, "seconds": 2002.6} {"step": 422, "loss": 0.4675273299217224, "data": 0.1211860328912735, "gap": 0.3463412821292877, "sigma": 0.7573082447052002, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.38219499588012695, "seconds": 2006.8} {"step": 423, "loss": 0.5904764533042908, "data": 0.5904764533042908, "gap": null, "sigma": 0.9973974823951721, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.8287491798400879, "seconds": 2009.6} {"step": 424, "loss": 0.6332943439483643, "data": 0.2800010144710541, "gap": 0.35329335927963257, "sigma": 0.8194261789321899, "kind": "mosaic", "tokens": 271, "dense": 448, "grad_norm": 0.32043981552124023, "seconds": 2014.2} {"step": 425, "loss": 0.5423440337181091, "data": 0.154855415225029, "gap": 0.38748860359191895, "sigma": 0.7062969207763672, "kind": "mosaic", "tokens": 164, "dense": 448, "grad_norm": 0.24586161971092224, "seconds": 2018.7} {"step": 426, "loss": 0.7406890392303467, "data": 0.18275251984596252, "gap": 0.5579364895820618, "sigma": 0.79355388879776, "kind": "mosaic", "tokens": 229, "dense": 448, "grad_norm": 0.32008078694343567, "seconds": 2023.2} {"step": 427, "loss": 0.49585267901420593, "data": 0.09946614503860474, "gap": 0.3963865339756012, "sigma": 0.9745064973831177, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.6231560707092285, "seconds": 2027.3} {"step": 428, "loss": 0.5740694999694824, "data": 0.5740694999694824, "gap": null, "sigma": 0.9878292083740234, "kind": "dense", "tokens": 576, "dense": 576, "grad_norm": 0.6021348834037781, "seconds": 2036.8} {"step": 429, "loss": 2.071870803833008, "data": 0.6753448247909546, "gap": 1.3965258598327637, "sigma": 0.1872224658727646, "kind": "mosaic", "tokens": 206, "dense": 448, "grad_norm": 0.6151880621910095, "seconds": 2041.2} {"step": 430, "loss": 0.5091110467910767, "data": 0.2104039490222931, "gap": 0.29870712757110596, "sigma": 0.645693838596344, "kind": "mosaic", "tokens": 259, "dense": 448, "grad_norm": 0.30188697576522827, "seconds": 2045.8} {"step": 431, "loss": 0.5386544466018677, "data": 0.08923304826021194, "gap": 0.44942137598991394, "sigma": 0.5807091593742371, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.304201602935791, "seconds": 2049.9} {"step": 432, "loss": 1.0909984111785889, "data": 0.27493423223495483, "gap": 0.8160641193389893, "sigma": 0.96468585729599, "kind": "mosaic", "tokens": 238, "dense": 448, "grad_norm": 0.6831342577934265, "seconds": 2054.5} {"step": 433, "loss": 0.3778247833251953, "data": 0.1644592136144638, "gap": 0.2133655846118927, "sigma": 0.6641697287559509, "kind": "mosaic", "tokens": 251, "dense": 448, "grad_norm": 0.13616600632667542, "seconds": 2059.0} {"step": 434, "loss": 0.5909565687179565, "data": 0.24544139206409454, "gap": 0.3455151915550232, "sigma": 0.9135600328445435, "kind": "mosaic", "tokens": 263, "dense": 448, "grad_norm": 0.3454219400882721, "seconds": 2063.4} {"step": 435, "loss": 0.47497549653053284, "data": 0.0973258689045906, "gap": 0.3776496350765228, "sigma": 0.8902079463005066, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.3661458194255829, "seconds": 2067.8} {"step": 436, "loss": 0.49702292680740356, "data": 0.18687520921230316, "gap": 0.3101477324962616, "sigma": 0.6832191944122314, "kind": "mosaic", "tokens": 187, "dense": 448, "grad_norm": 0.197907492518425, "seconds": 2072.3} {"step": 437, "loss": 0.9177178144454956, "data": 0.28590983152389526, "gap": 0.6318079829216003, "sigma": 0.5217355489730835, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.31489071249961853, "seconds": 2076.7} {"step": 438, "loss": 0.3721289038658142, "data": 0.14679144322872162, "gap": 0.22533747553825378, "sigma": 0.8125389814376831, "kind": "mosaic", "tokens": 167, "dense": 448, "grad_norm": 0.14178451895713806, "seconds": 2080.9} {"step": 439, "loss": 0.8448892831802368, "data": 0.2856302559375763, "gap": 0.5592590570449829, "sigma": 0.36542046070098877, "kind": "mosaic", "tokens": 245, "dense": 448, "grad_norm": 0.23627278208732605, "seconds": 2085.1} {"step": 440, "loss": 0.7011206150054932, "data": 0.0826922133564949, "gap": 0.6184284090995789, "sigma": 0.9495969414710999, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.7504321336746216, "seconds": 2089.3} {"step": 441, "loss": 0.5930875539779663, "data": 0.19922268390655518, "gap": 0.39386484026908875, "sigma": 0.561107337474823, "kind": "mosaic", "tokens": 187, "dense": 448, "grad_norm": 0.1574895828962326, "seconds": 2093.6} {"step": 442, "loss": 0.4161832332611084, "data": 0.162639781832695, "gap": 0.2535434365272522, "sigma": 0.7435521483421326, "kind": "mosaic", "tokens": 195, "dense": 448, "grad_norm": 0.2052135169506073, "seconds": 2097.9} {"step": 443, "loss": 0.3908698856830597, "data": 0.3908698856830597, "gap": null, "sigma": 0.995671808719635, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.37842118740081787, "seconds": 2100.8} {"step": 444, "loss": 0.4707486927509308, "data": 0.21474099159240723, "gap": 0.25600770115852356, "sigma": 0.8071426153182983, "kind": "mosaic", "tokens": 287, "dense": 448, "grad_norm": 0.24181179702281952, "seconds": 2105.2} {"step": 445, "loss": 0.7878513336181641, "data": 0.20939327776432037, "gap": 0.5784580707550049, "sigma": 0.7873393297195435, "kind": "mosaic", "tokens": 227, "dense": 448, "grad_norm": 0.6579792499542236, "seconds": 2109.8} {"step": 446, "loss": 0.5728569030761719, "data": 0.189169779419899, "gap": 0.3836871385574341, "sigma": 0.7705565690994263, "kind": "mosaic", "tokens": 176, "dense": 448, "grad_norm": 0.31961163878440857, "seconds": 2114.2} {"step": 447, "loss": 0.49647802114486694, "data": 0.18441109359264374, "gap": 0.312066912651062, "sigma": 0.6607406735420227, "kind": "mosaic", "tokens": 233, "dense": 512, "grad_norm": 0.1554870307445526, "seconds": 2118.7} {"step": 448, "loss": 0.5177197456359863, "data": 0.17532561719417572, "gap": 0.3423941135406494, "sigma": 0.5334318280220032, "kind": "mosaic", "tokens": 227, "dense": 448, "grad_norm": 0.33577173948287964, "seconds": 2123.2} {"step": 449, "loss": 0.5260446071624756, "data": 0.2378542572259903, "gap": 0.2881903350353241, "sigma": 0.6831845641136169, "kind": "mosaic", "tokens": 214, "dense": 448, "grad_norm": 0.200101837515831, "seconds": 2127.6} {"step": 450, "loss": 0.578449010848999, "data": 0.19882452487945557, "gap": 0.37962445616722107, "sigma": 0.555865466594696, "kind": "mosaic", "tokens": 243, "dense": 448, "grad_norm": 0.1659374237060547, "seconds": 2132.0} {"step": 451, "loss": 0.7904136180877686, "data": 0.2976987361907959, "gap": 0.49271485209465027, "sigma": 0.47993287444114685, "kind": "mosaic", "tokens": 223, "dense": 448, "grad_norm": 0.21704241633415222, "seconds": 2136.4} {"step": 452, "loss": 0.17883750796318054, "data": 0.17883750796318054, "gap": null, "sigma": 0.7097806930541992, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.1996365487575531, "seconds": 2139.2} {"step": 453, "loss": 0.2522302269935608, "data": 0.10745178163051605, "gap": 0.14477844536304474, "sigma": 0.644693911075592, "kind": "mosaic", "tokens": 229, "dense": 448, "grad_norm": 0.20126047730445862, "seconds": 2143.8} {"step": 454, "loss": 0.8473032712936401, "data": 0.2865183651447296, "gap": 0.5607849359512329, "sigma": 0.2876988351345062, "kind": "mosaic", "tokens": 208, "dense": 448, "grad_norm": 0.2884256839752197, "seconds": 2148.2} {"step": 455, "loss": 0.13756659626960754, "data": 0.13756659626960754, "gap": null, "sigma": 0.5945030450820923, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.09656592458486557, "seconds": 2151.0} {"step": 456, "loss": 0.6289593577384949, "data": 0.2232983112335205, "gap": 0.40566104650497437, "sigma": 0.7389830946922302, "kind": "mosaic", "tokens": 196, "dense": 448, "grad_norm": 0.31616881489753723, "seconds": 2155.4} {"step": 457, "loss": 1.9220998287200928, "data": 0.1754876971244812, "gap": 1.7466121912002563, "sigma": 0.9652824997901917, "kind": "mosaic", "tokens": 160, "dense": 448, "grad_norm": 2.0866284370422363, "seconds": 2159.7} {"step": 458, "loss": 0.6738990545272827, "data": 0.29854175448417664, "gap": 0.3753572702407837, "sigma": 0.9216415882110596, "kind": "mosaic", "tokens": 273, "dense": 448, "grad_norm": 0.39584222435951233, "seconds": 2164.3} {"step": 459, "loss": 0.2617093324661255, "data": 0.2617093324661255, "gap": null, "sigma": 0.7097460627555847, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.1826772540807724, "seconds": 2167.0} {"step": 460, "loss": 0.9194301962852478, "data": 0.15163849294185638, "gap": 0.7677916884422302, "sigma": 0.9836145639419556, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 1.1494654417037964, "seconds": 2171.2} {"step": 461, "loss": 0.6436156034469604, "data": 0.13646194338798523, "gap": 0.5071536302566528, "sigma": 0.8841478228569031, "kind": "mosaic", "tokens": 212, "dense": 448, "grad_norm": 0.575471818447113, "seconds": 2175.7} {"step": 462, "loss": 2.9035568237304688, "data": 0.7040643095970154, "gap": 2.1994924545288086, "sigma": 0.16693997383117676, "kind": "mosaic", "tokens": 174, "dense": 448, "grad_norm": 1.1955984830856323, "seconds": 2180.1} {"step": 463, "loss": 0.8609639406204224, "data": 0.15905450284481049, "gap": 0.7019094228744507, "sigma": 0.8461818099021912, "kind": "mosaic", "tokens": 158, "dense": 448, "grad_norm": 0.6752170324325562, "seconds": 2184.5} {"step": 464, "loss": 0.32086294889450073, "data": 0.17889566719532013, "gap": 0.1419672816991806, "sigma": 0.49029937386512756, "kind": "mosaic", "tokens": 266, "dense": 448, "grad_norm": 0.18728749454021454, "seconds": 2189.1} {"step": 465, "loss": 0.6541079878807068, "data": 0.2147427350282669, "gap": 0.4393652379512787, "sigma": 0.9201403856277466, "kind": "mosaic", "tokens": 190, "dense": 448, "grad_norm": 0.5618835687637329, "seconds": 2193.4} {"step": 466, "loss": 0.9571107625961304, "data": 0.35743120312690735, "gap": 0.5996795296669006, "sigma": 0.27728593349456787, "kind": "mosaic", "tokens": 266, "dense": 448, "grad_norm": 0.2771761119365692, "seconds": 2198.0} {"step": 467, "loss": 0.6827589273452759, "data": 0.353054940700531, "gap": 0.32970401644706726, "sigma": 0.93071049451828, "kind": "mosaic", "tokens": 254, "dense": 448, "grad_norm": 0.6358369588851929, "seconds": 2202.5} {"step": 468, "loss": 0.29316821694374084, "data": 0.11616411805152893, "gap": 0.17700409889221191, "sigma": 0.6089036464691162, "kind": "mosaic", "tokens": 200, "dense": 448, "grad_norm": 0.1910792738199234, "seconds": 2207.0} {"step": 469, "loss": 0.7151952981948853, "data": 0.3228975534439087, "gap": 0.3922977149486542, "sigma": 0.4576173722743988, "kind": "mosaic", "tokens": 257, "dense": 448, "grad_norm": 0.18690602481365204, "seconds": 2211.4} {"step": 470, "loss": 5.758792877197266, "data": 1.5148284435272217, "gap": 4.243964195251465, "sigma": 0.09901382029056549, "kind": "mosaic", "tokens": 220, "dense": 448, "grad_norm": 2.6202385425567627, "seconds": 2215.8} {"step": 471, "loss": 0.6474127173423767, "data": 0.26193875074386597, "gap": 0.38547396659851074, "sigma": 0.5515977144241333, "kind": "mosaic", "tokens": 201, "dense": 448, "grad_norm": 0.14425338804721832, "seconds": 2220.2} {"step": 472, "loss": 0.47625482082366943, "data": 0.18648764491081238, "gap": 0.28976717591285706, "sigma": 0.7023944854736328, "kind": "mosaic", "tokens": 213, "dense": 448, "grad_norm": 0.30416181683540344, "seconds": 2224.7} {"step": 473, "loss": 0.44082337617874146, "data": 0.22098959982395172, "gap": 0.21983377635478973, "sigma": 0.9136592745780945, "kind": "mosaic", "tokens": 206, "dense": 448, "grad_norm": 0.2373274266719818, "seconds": 2229.2} {"step": 474, "loss": 0.42822420597076416, "data": 0.42822420597076416, "gap": null, "sigma": 0.9612993001937866, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.36949270963668823, "seconds": 2232.1} {"step": 475, "loss": 0.3478180468082428, "data": 0.1718215048313141, "gap": 0.1759965419769287, "sigma": 0.8562758564949036, "kind": "mosaic", "tokens": 206, "dense": 448, "grad_norm": 0.1640133112668991, "seconds": 2236.5} {"step": 476, "loss": 1.965775966644287, "data": 0.055280182510614395, "gap": 1.9104957580566406, "sigma": 0.9931210279464722, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 1.8718091249465942, "seconds": 2240.6} {"step": 477, "loss": 0.56341552734375, "data": 0.2421397715806961, "gap": 0.3212757706642151, "sigma": 0.24716095626354218, "kind": "mosaic", "tokens": 241, "dense": 448, "grad_norm": 0.28991392254829407, "seconds": 2245.0} {"step": 478, "loss": 0.41960960626602173, "data": 0.19727720320224762, "gap": 0.22233238816261292, "sigma": 0.6472739577293396, "kind": "mosaic", "tokens": 241, "dense": 448, "grad_norm": 0.15292289853096008, "seconds": 2249.4} {"step": 479, "loss": 0.4806552231311798, "data": 0.4806552231311798, "gap": null, "sigma": 0.10670732706785202, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.23285745084285736, "seconds": 2252.2} {"step": 480, "loss": 0.36749163269996643, "data": 0.36749163269996643, "gap": null, "sigma": 0.8090542554855347, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.23056069016456604, "seconds": 2255.1} {"step": 481, "loss": 0.46602925658226013, "data": 0.25245213508605957, "gap": 0.21357712149620056, "sigma": 0.9050066471099854, "kind": "mosaic", "tokens": 247, "dense": 448, "grad_norm": 0.5501452088356018, "seconds": 2259.7} {"step": 482, "loss": 0.42954742908477783, "data": 0.2538677752017975, "gap": 0.17567965388298035, "sigma": 0.8587731719017029, "kind": "mosaic", "tokens": 225, "dense": 448, "grad_norm": 0.19693763554096222, "seconds": 2264.1} {"step": 483, "loss": 0.3685532808303833, "data": 0.2087550014257431, "gap": 0.159798264503479, "sigma": 0.6020909547805786, "kind": "mosaic", "tokens": 218, "dense": 448, "grad_norm": 0.1116890087723732, "seconds": 2268.6} {"step": 484, "loss": 0.625128984451294, "data": 0.1922522336244583, "gap": 0.43287673592567444, "sigma": 0.7533942461013794, "kind": "mosaic", "tokens": 222, "dense": 448, "grad_norm": 0.3524937033653259, "seconds": 2273.0} {"step": 485, "loss": 3.0699477195739746, "data": 0.828798770904541, "gap": 2.2411489486694336, "sigma": 0.15600910782814026, "kind": "mosaic", "tokens": 236, "dense": 448, "grad_norm": 1.1888213157653809, "seconds": 2277.6} {"step": 486, "loss": 0.6001922488212585, "data": 0.1892540007829666, "gap": 0.4109382629394531, "sigma": 0.6235172152519226, "kind": "uniform", "tokens": 144, "dense": 576, "grad_norm": 0.19561772048473358, "seconds": 2282.0} {"step": 487, "loss": 0.7624719142913818, "data": 0.28765869140625, "gap": 0.47481319308280945, "sigma": 0.44984084367752075, "kind": "mosaic", "tokens": 231, "dense": 448, "grad_norm": 0.17641477286815643, "seconds": 2286.3} {"step": 488, "loss": 0.36227333545684814, "data": 0.17093288898468018, "gap": 0.19134046137332916, "sigma": 0.5315719842910767, "kind": "mosaic", "tokens": 225, "dense": 448, "grad_norm": 0.1778021603822708, "seconds": 2290.8} {"step": 489, "loss": 11.204666137695312, "data": 2.8513405323028564, "gap": 8.353325843811035, "sigma": 0.009373909793794155, "kind": "mosaic", "tokens": 225, "dense": 448, "grad_norm": 8.947733879089355, "seconds": 2295.3} {"step": 490, "loss": 0.49312061071395874, "data": 0.23643697798252106, "gap": 0.2566836178302765, "sigma": 0.8114317655563354, "kind": "mosaic", "tokens": 236, "dense": 448, "grad_norm": 0.3584156632423401, "seconds": 2299.9} {"step": 491, "loss": 0.5375998616218567, "data": 0.2619209289550781, "gap": 0.27567893266677856, "sigma": 0.9463034272193909, "kind": "mosaic", "tokens": 195, "dense": 448, "grad_norm": 0.2920377254486084, "seconds": 2304.3} {"step": 492, "loss": 0.9552438259124756, "data": 0.28276583552360535, "gap": 0.6724779605865479, "sigma": 0.9523870944976807, "kind": "mosaic", "tokens": 181, "dense": 448, "grad_norm": 0.6488070487976074, "seconds": 2308.8} {"step": 493, "loss": 1.1264857053756714, "data": 0.0940379872918129, "gap": 1.0324476957321167, "sigma": 0.927281379699707, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.8036583662033081, "seconds": 2313.1} {"step": 494, "loss": 0.7462653517723083, "data": 0.2711988091468811, "gap": 0.47506654262542725, "sigma": 0.41473516821861267, "kind": "mosaic", "tokens": 216, "dense": 448, "grad_norm": 0.2013409435749054, "seconds": 2317.5} {"step": 495, "loss": 0.5553525686264038, "data": 0.20125934481620789, "gap": 0.3540932536125183, "sigma": 0.5759297609329224, "kind": "mosaic", "tokens": 243, "dense": 448, "grad_norm": 0.2461618036031723, "seconds": 2322.1} {"step": 496, "loss": 0.5124784708023071, "data": 0.18181172013282776, "gap": 0.33066675066947937, "sigma": 0.6038357019424438, "kind": "mosaic", "tokens": 196, "dense": 448, "grad_norm": 0.4526306688785553, "seconds": 2326.6} {"step": 497, "loss": 0.46828019618988037, "data": 0.23632769286632538, "gap": 0.2319524884223938, "sigma": 0.7509141564369202, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.2210393100976944, "seconds": 2331.1} {"step": 498, "loss": 1.1276739835739136, "data": 0.23594768345355988, "gap": 0.8917263150215149, "sigma": 0.9524529576301575, "kind": "mosaic", "tokens": 229, "dense": 448, "grad_norm": 0.6597152948379517, "seconds": 2335.5} {"step": 499, "loss": 0.4749147891998291, "data": 0.19109386205673218, "gap": 0.2838209271430969, "sigma": 0.7695508003234863, "kind": "mosaic", "tokens": 243, "dense": 448, "grad_norm": 0.2827998995780945, "seconds": 2339.9} {"step": 500, "loss": 0.6258120536804199, "data": 0.21104617416858673, "gap": 0.414765864610672, "sigma": 0.7327790260314941, "kind": "mosaic", "tokens": 184, "dense": 448, "grad_norm": 0.17692753672599792, "seconds": 2344.5} {"step": 500, "eval": {"dense": 0.3094475908825795, "uniform2": 0.18935116628805795, "mosaic": 0.25361086552341777, "gap_uniform2": 0.6171817394594351, "gap_mosaic": 0.6118613642950853}, "eval_extent": {"1x1x1": 0.41361204463948564, "1x1x2": 0.3231521863479256, "1x1x4": 0.2324761874450926, "1x2x1": 0.30401915144175296, "1x2x2": 0.20143478800248552, "1x2x4": 0.131201674548901, "1x4x1": 0.2064344999215686, "1x4x2": 0.11641151202083738, "1x4x4": 0.08584292435558204}} {"step": 501, "loss": 0.7303758859634399, "data": 0.28911954164505005, "gap": 0.4412563741207123, "sigma": 0.43320587277412415, "kind": "mosaic", "tokens": 244, "dense": 448, "grad_norm": 0.18869540095329285, "seconds": 2589.6} {"step": 502, "loss": 0.2992444634437561, "data": 0.2992444634437561, "gap": null, "sigma": 0.7619395852088928, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.17257912456989288, "seconds": 2592.5} {"step": 503, "loss": 6.671875953674316, "data": 1.4701383113861084, "gap": 5.201737403869629, "sigma": 0.06664541363716125, "kind": "mosaic", "tokens": 259, "dense": 448, "grad_norm": 5.283036708831787, "seconds": 2597.0} {"step": 504, "loss": 0.3458186089992523, "data": 0.3458186089992523, "gap": null, "sigma": 0.8299679756164551, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.22927092015743256, "seconds": 2599.8} {"step": 505, "loss": 0.895984947681427, "data": 0.1524476408958435, "gap": 0.7435373067855835, "sigma": 0.5261205434799194, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.3258568346500397, "seconds": 2604.0} {"step": 506, "loss": 0.4635730981826782, "data": 0.18340973556041718, "gap": 0.28016334772109985, "sigma": 0.5336090326309204, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.2306731641292572, "seconds": 2608.6} {"step": 507, "loss": 0.5755161046981812, "data": 0.23787054419517517, "gap": 0.33764559030532837, "sigma": 0.7883734107017517, "kind": "mosaic", "tokens": 210, "dense": 448, "grad_norm": 0.3093458116054535, "seconds": 2612.9} {"step": 508, "loss": 0.5802709460258484, "data": 0.21535219252109528, "gap": 0.3649187386035919, "sigma": 0.43845757842063904, "kind": "mosaic", "tokens": 216, "dense": 448, "grad_norm": 0.17341248691082, "seconds": 2617.5} {"step": 509, "loss": 0.7938857674598694, "data": 0.2099582552909851, "gap": 0.5839275121688843, "sigma": 0.851204514503479, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.48044484853744507, "seconds": 2622.1} {"step": 510, "loss": 0.31494471430778503, "data": 0.31494471430778503, "gap": null, "sigma": 0.9846660494804382, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.20261885225772858, "seconds": 2624.9} {"step": 511, "loss": 0.24756154417991638, "data": 0.24756154417991638, "gap": null, "sigma": 0.8457221984863281, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.17474351823329926, "seconds": 2627.8} {"step": 512, "loss": 0.23665615916252136, "data": 0.1043514534831047, "gap": 0.13230469822883606, "sigma": 0.7548661828041077, "kind": "mosaic", "tokens": 176, "dense": 448, "grad_norm": 0.2251376509666443, "seconds": 2632.4} {"step": 513, "loss": 0.5215970277786255, "data": 0.22463642060756683, "gap": 0.29696062207221985, "sigma": 0.9287819862365723, "kind": "mosaic", "tokens": 203, "dense": 448, "grad_norm": 0.4224298596382141, "seconds": 2637.0} {"step": 514, "loss": 0.3483107089996338, "data": 0.171517014503479, "gap": 0.1767936795949936, "sigma": 0.48731669783592224, "kind": "mosaic", "tokens": 275, "dense": 448, "grad_norm": 0.12170592695474625, "seconds": 2641.5} {"step": 515, "loss": 0.7871044278144836, "data": 0.201385498046875, "gap": 0.5857189297676086, "sigma": 0.8605022430419922, "kind": "mosaic", "tokens": 232, "dense": 448, "grad_norm": 0.4219207167625427, "seconds": 2645.9} {"step": 516, "loss": 1.0779962539672852, "data": 0.342020183801651, "gap": 0.7359760999679565, "sigma": 0.9928251504898071, "kind": "mosaic", "tokens": 209, "dense": 448, "grad_norm": 0.7389780879020691, "seconds": 2650.4} {"step": 517, "loss": 0.659812331199646, "data": 0.25431182980537415, "gap": 0.40550053119659424, "sigma": 0.42617762088775635, "kind": "mosaic", "tokens": 233, "dense": 448, "grad_norm": 0.17031346261501312, "seconds": 2654.9} {"step": 518, "loss": 0.5285844802856445, "data": 0.22490990161895752, "gap": 0.3036746084690094, "sigma": 0.8137616515159607, "kind": "mosaic", "tokens": 261, "dense": 448, "grad_norm": 0.30518823862075806, "seconds": 2659.5} {"step": 519, "loss": 0.6341268420219421, "data": 0.17627829313278198, "gap": 0.45784854888916016, "sigma": 0.6409002542495728, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.2611408233642578, "seconds": 2663.7} {"step": 520, "loss": 0.5010050535202026, "data": 0.2479952871799469, "gap": 0.2530097961425781, "sigma": 0.6033961176872253, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.19907011091709137, "seconds": 2668.1} {"step": 521, "loss": 0.8356750011444092, "data": 0.2833038568496704, "gap": 0.5523711442947388, "sigma": 0.9673087000846863, "kind": "mosaic", "tokens": 240, "dense": 448, "grad_norm": 0.49759072065353394, "seconds": 2672.7} {"step": 522, "loss": 0.6431851387023926, "data": 0.26768267154693604, "gap": 0.37550243735313416, "sigma": 0.9087241888046265, "kind": "mosaic", "tokens": 253, "dense": 448, "grad_norm": 0.3489854335784912, "seconds": 2677.2} {"step": 523, "loss": 0.35219961404800415, "data": 0.1996416449546814, "gap": 0.15255795419216156, "sigma": 0.8759015202522278, "kind": "mosaic", "tokens": 210, "dense": 448, "grad_norm": 0.21328817307949066, "seconds": 2681.6} {"step": 524, "loss": 0.5337058901786804, "data": 0.2376471906900406, "gap": 0.296058714389801, "sigma": 0.9111482501029968, "kind": "mosaic", "tokens": 196, "dense": 448, "grad_norm": 0.6698153614997864, "seconds": 2686.1} {"step": 525, "loss": 0.48339372873306274, "data": 0.19058726727962494, "gap": 0.292806476354599, "sigma": 0.7110147476196289, "kind": "mosaic", "tokens": 174, "dense": 448, "grad_norm": 0.3756173849105835, "seconds": 2690.6} {"step": 526, "loss": 0.2643872797489166, "data": 0.2643872797489166, "gap": null, "sigma": 0.7072087526321411, "kind": "dense", "tokens": 480, "dense": 480, "grad_norm": 0.09311629086732864, "seconds": 2693.5} {"step": 527, "loss": 0.2524343729019165, "data": 0.11844715476036072, "gap": 0.1339872032403946, "sigma": 0.5783913135528564, "kind": "mosaic", "tokens": 226, "dense": 448, "grad_norm": 0.34478381276130676, "seconds": 2697.9} {"step": 528, "loss": 0.39411476254463196, "data": 0.06658890843391418, "gap": 0.3275258541107178, "sigma": 0.6954914331436157, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.8187848925590515, "seconds": 2702.2} {"step": 529, "loss": 0.8726797103881836, "data": 0.2578037977218628, "gap": 0.6148759126663208, "sigma": 0.362684428691864, "kind": "mosaic", "tokens": 183, "dense": 448, "grad_norm": 0.3079374134540558, "seconds": 2706.7} {"step": 530, "loss": 0.5447330474853516, "data": 0.25694596767425537, "gap": 0.2877870798110962, "sigma": 0.8678755760192871, "kind": "mosaic", "tokens": 210, "dense": 448, "grad_norm": 0.67527174949646, "seconds": 2711.1} {"step": 531, "loss": 0.32716065645217896, "data": 0.1520361453294754, "gap": 0.17512452602386475, "sigma": 0.6074947714805603, "kind": "mosaic", "tokens": 194, "dense": 448, "grad_norm": 0.1377895176410675, "seconds": 2715.7} {"step": 532, "loss": 0.23137333989143372, "data": 0.23137333989143372, "gap": null, "sigma": 0.7296580076217651, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2312547117471695, "seconds": 2718.5} {"step": 533, "loss": 1.0967386960983276, "data": 0.5063854455947876, "gap": 0.59035325050354, "sigma": 0.3040080964565277, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.33210262656211853, "seconds": 2722.9} {"step": 534, "loss": 0.31705838441848755, "data": 0.31705838441848755, "gap": null, "sigma": 0.9160110950469971, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.379415899515152, "seconds": 2725.8} {"step": 535, "loss": 0.4777032136917114, "data": 0.19591401517391205, "gap": 0.2817891836166382, "sigma": 0.7071049809455872, "kind": "mosaic", "tokens": 177, "dense": 448, "grad_norm": 0.5424913167953491, "seconds": 2730.2} {"step": 536, "loss": 0.44559741020202637, "data": 0.1951015144586563, "gap": 0.25049591064453125, "sigma": 0.6768110990524292, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.22516211867332458, "seconds": 2734.7} {"step": 537, "loss": 1.1832492351531982, "data": 0.31274470686912537, "gap": 0.8705045580863953, "sigma": 0.3325622081756592, "kind": "mosaic", "tokens": 170, "dense": 448, "grad_norm": 0.8379989862442017, "seconds": 2739.1} {"step": 538, "loss": 1.6195547580718994, "data": 0.5289751887321472, "gap": 1.090579628944397, "sigma": 0.2195238322019577, "kind": "mosaic", "tokens": 233, "dense": 448, "grad_norm": 0.47130468487739563, "seconds": 2743.5} {"step": 539, "loss": 0.26766863465309143, "data": 0.13310636579990387, "gap": 0.13456226885318756, "sigma": 0.6849071383476257, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.4602532982826233, "seconds": 2748.0} {"step": 540, "loss": 0.3479422926902771, "data": 0.3479422926902771, "gap": null, "sigma": 0.3920571804046631, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.1372244656085968, "seconds": 2750.9} {"step": 541, "loss": 0.24328286945819855, "data": 0.24328286945819855, "gap": null, "sigma": 0.9640375971794128, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.24815966188907623, "seconds": 2753.8} {"step": 542, "loss": 0.621511697769165, "data": 0.2507683038711548, "gap": 0.37074336409568787, "sigma": 0.6019627451896667, "kind": "mosaic", "tokens": 238, "dense": 448, "grad_norm": 0.178416445851326, "seconds": 2758.3} {"step": 543, "loss": 0.6596875190734863, "data": 0.1629093885421753, "gap": 0.49677810072898865, "sigma": 0.8047723770141602, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.3484151363372803, "seconds": 2762.7} {"step": 544, "loss": 0.6386972665786743, "data": 0.10946191102266312, "gap": 0.5292353630065918, "sigma": 0.8046832084655762, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.3363024592399597, "seconds": 2766.9} {"step": 545, "loss": 0.18610823154449463, "data": 0.18610823154449463, "gap": null, "sigma": 0.6857646703720093, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.20550744235515594, "seconds": 2769.6} {"step": 546, "loss": 0.7777907252311707, "data": 0.1618875116109848, "gap": 0.6159031987190247, "sigma": 0.879374623298645, "kind": "mosaic", "tokens": 146, "dense": 448, "grad_norm": 0.35910138487815857, "seconds": 2774.0} {"step": 547, "loss": 0.34667378664016724, "data": 0.1252405047416687, "gap": 0.22143328189849854, "sigma": 0.7108222246170044, "kind": "mosaic", "tokens": 176, "dense": 448, "grad_norm": 0.20150314271450043, "seconds": 2778.6} {"step": 548, "loss": 0.5001686811447144, "data": 0.21928292512893677, "gap": 0.2808857858181, "sigma": 0.6638980507850647, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.23218689858913422, "seconds": 2783.1} {"step": 549, "loss": 0.48350411653518677, "data": 0.20918859541416168, "gap": 0.2743155062198639, "sigma": 0.59815514087677, "kind": "mosaic", "tokens": 264, "dense": 448, "grad_norm": 0.21046793460845947, "seconds": 2787.7} {"step": 550, "loss": 19.891464233398438, "data": 3.5062336921691895, "gap": 16.385231018066406, "sigma": 0.012561117298901081, "kind": "mosaic", "tokens": 254, "dense": 448, "grad_norm": 23.837080001831055, "seconds": 2792.1} {"step": 551, "loss": 0.4013737738132477, "data": 0.19343213737010956, "gap": 0.20794163644313812, "sigma": 0.7933424711227417, "kind": "mosaic", "tokens": 196, "dense": 448, "grad_norm": 0.1586441993713379, "seconds": 2796.6} {"step": 552, "loss": 1.4482333660125732, "data": 0.3999469578266144, "gap": 1.0482864379882812, "sigma": 0.1958608478307724, "kind": "mosaic", "tokens": 167, "dense": 448, "grad_norm": 0.5610320568084717, "seconds": 2801.3} {"step": 553, "loss": 0.6605817079544067, "data": 0.2542572617530823, "gap": 0.40632444620132446, "sigma": 0.9044156074523926, "kind": "mosaic", "tokens": 250, "dense": 448, "grad_norm": 0.6607577800750732, "seconds": 2805.8} {"step": 554, "loss": 0.4670455753803253, "data": 0.4670455753803253, "gap": null, "sigma": 0.9061928987503052, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.19561481475830078, "seconds": 2808.6} {"step": 555, "loss": 0.3775748908519745, "data": 0.3775748908519745, "gap": null, "sigma": 0.9138321280479431, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.1901237964630127, "seconds": 2811.4} {"step": 556, "loss": 0.3196237087249756, "data": 0.14197631180286407, "gap": 0.17764738202095032, "sigma": 0.6842758059501648, "kind": "mosaic", "tokens": 244, "dense": 448, "grad_norm": 0.15081565082073212, "seconds": 2815.9} {"step": 557, "loss": 0.43687188625335693, "data": 0.17911715805530548, "gap": 0.25775474309921265, "sigma": 0.8571357131004333, "kind": "mosaic", "tokens": 265, "dense": 448, "grad_norm": 0.2883536219596863, "seconds": 2820.3} {"step": 558, "loss": 0.14846490323543549, "data": 0.14846490323543549, "gap": null, "sigma": 0.48527538776397705, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.11113908141851425, "seconds": 2823.1} {"step": 559, "loss": 0.22442838549613953, "data": 0.22442838549613953, "gap": null, "sigma": 0.7414125204086304, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.12996584177017212, "seconds": 2826.0} {"step": 560, "loss": 0.49680179357528687, "data": 0.27881714701652527, "gap": 0.2179846465587616, "sigma": 0.914211630821228, "kind": "mosaic", "tokens": 232, "dense": 448, "grad_norm": 0.29308995604515076, "seconds": 2830.5} {"step": 561, "loss": 0.6729484796524048, "data": 0.04883886128664017, "gap": 0.6241096258163452, "sigma": 0.8586887121200562, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 0.5075064897537231, "seconds": 2834.7} {"step": 562, "loss": 0.2918180227279663, "data": 0.16776899993419647, "gap": 0.12404901534318924, "sigma": 0.6498490571975708, "kind": "mosaic", "tokens": 272, "dense": 448, "grad_norm": 0.2366592288017273, "seconds": 2839.1} {"step": 563, "loss": 0.3882882297039032, "data": 0.3882882297039032, "gap": null, "sigma": 0.8676369190216064, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.14055968821048737, "seconds": 2841.9} {"step": 564, "loss": 0.41822177171707153, "data": 0.1690608710050583, "gap": 0.24916090071201324, "sigma": 0.7618739604949951, "kind": "mosaic", "tokens": 180, "dense": 448, "grad_norm": 0.1371668577194214, "seconds": 2846.4} {"step": 565, "loss": 0.5310354232788086, "data": 0.21156834065914154, "gap": 0.31946706771850586, "sigma": 0.6601697206497192, "kind": "mosaic", "tokens": 178, "dense": 448, "grad_norm": 0.21871554851531982, "seconds": 2851.1} {"step": 566, "loss": 0.4638785421848297, "data": 0.23383526504039764, "gap": 0.23004327714443207, "sigma": 0.6641474962234497, "kind": "mosaic", "tokens": 278, "dense": 448, "grad_norm": 0.33266061544418335, "seconds": 2855.7} {"step": 567, "loss": 1.6212550401687622, "data": 0.29639604687690735, "gap": 1.3248590230941772, "sigma": 0.9982728362083435, "kind": "mosaic", "tokens": 192, "dense": 448, "grad_norm": 0.6887263655662537, "seconds": 2860.1} {"step": 568, "loss": 0.4841572046279907, "data": 0.2651166617870331, "gap": 0.21904055774211884, "sigma": 0.6677011847496033, "kind": "mosaic", "tokens": 277, "dense": 448, "grad_norm": 0.3902854323387146, "seconds": 2864.6} {"step": 569, "loss": 0.7677408456802368, "data": 0.18043413758277893, "gap": 0.5873066782951355, "sigma": 0.9411762356758118, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.6652207970619202, "seconds": 2868.7} {"step": 570, "loss": 0.7807719707489014, "data": 0.23634234070777893, "gap": 0.5444296598434448, "sigma": 0.8655397891998291, "kind": "mosaic", "tokens": 247, "dense": 448, "grad_norm": 3.08911395072937, "seconds": 2873.2} {"step": 571, "loss": 0.3011496961116791, "data": 0.3011496961116791, "gap": null, "sigma": 0.9346188306808472, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.29462018609046936, "seconds": 2876.0} {"step": 572, "loss": 0.31225332617759705, "data": 0.17734411358833313, "gap": 0.13490921258926392, "sigma": 0.5859031081199646, "kind": "mosaic", "tokens": 267, "dense": 448, "grad_norm": 0.1368826925754547, "seconds": 2880.6} {"step": 573, "loss": 0.522284209728241, "data": 0.2062598615884781, "gap": 0.3160243630409241, "sigma": 0.7436637282371521, "kind": "mosaic", "tokens": 213, "dense": 448, "grad_norm": 0.17335203289985657, "seconds": 2885.1} {"step": 574, "loss": 0.6090855598449707, "data": 0.12400771677494049, "gap": 0.4850778579711914, "sigma": 0.8259785771369934, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.4328000843524933, "seconds": 2889.2} {"step": 575, "loss": 0.4265294075012207, "data": 0.1778913289308548, "gap": 0.2486380934715271, "sigma": 0.6940250992774963, "kind": "mosaic", "tokens": 230, "dense": 448, "grad_norm": 0.18393729627132416, "seconds": 2893.6} {"step": 576, "loss": 0.7186475992202759, "data": 0.25436902046203613, "gap": 0.46427857875823975, "sigma": 0.9716201424598694, "kind": "mosaic", "tokens": 251, "dense": 448, "grad_norm": 0.4304986000061035, "seconds": 2898.1} {"step": 577, "loss": 0.8038208484649658, "data": 0.2744750678539276, "gap": 0.5293458104133606, "sigma": 0.9804878830909729, "kind": "mosaic", "tokens": 259, "dense": 448, "grad_norm": 0.44240063428878784, "seconds": 2902.7} {"step": 578, "loss": 0.6484039425849915, "data": 0.24749520421028137, "gap": 0.4009087383747101, "sigma": 0.4466565251350403, "kind": "mosaic", "tokens": 263, "dense": 448, "grad_norm": 0.15289893746376038, "seconds": 2907.3} {"step": 579, "loss": 1.219536542892456, "data": 0.35498055815696716, "gap": 0.8645559549331665, "sigma": 0.9422459006309509, "kind": "mosaic", "tokens": 268, "dense": 448, "grad_norm": 0.6542341113090515, "seconds": 2911.7} {"step": 580, "loss": 17.676422119140625, "data": 3.9747519493103027, "gap": 13.701669692993164, "sigma": 0.05530267581343651, "kind": "mosaic", "tokens": 194, "dense": 448, "grad_norm": 10.389241218566895, "seconds": 2916.1} {"step": 581, "loss": 0.3677763044834137, "data": 0.3677763044834137, "gap": null, "sigma": 0.8927273750305176, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.20035356283187866, "seconds": 2919.0} {"step": 582, "loss": 0.5900076031684875, "data": 0.10373617708683014, "gap": 0.4862714409828186, "sigma": 0.9372949004173279, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.5635491013526917, "seconds": 2923.2} {"step": 583, "loss": 0.19847257435321808, "data": 0.19847257435321808, "gap": null, "sigma": 0.8427485823631287, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.18142814934253693, "seconds": 2926.0} {"step": 584, "loss": 0.5806030035018921, "data": 0.10634192824363708, "gap": 0.4742611050605774, "sigma": 0.7505175471305847, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.3386111855506897, "seconds": 2930.1} {"step": 585, "loss": 0.2734239101409912, "data": 0.11826902627944946, "gap": 0.15515488386154175, "sigma": 0.6193878054618835, "kind": "mosaic", "tokens": 206, "dense": 448, "grad_norm": 0.16268566250801086, "seconds": 2934.6} {"step": 586, "loss": 0.36044055223464966, "data": 0.36044055223464966, "gap": null, "sigma": 0.8664708733558655, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.20371833443641663, "seconds": 2937.4} {"step": 587, "loss": 0.5310864448547363, "data": 0.22650441527366638, "gap": 0.30458199977874756, "sigma": 0.9513582587242126, "kind": "mosaic", "tokens": 196, "dense": 448, "grad_norm": 0.45917952060699463, "seconds": 2941.9} {"step": 588, "loss": 0.3371756672859192, "data": 0.1749984622001648, "gap": 0.1621771901845932, "sigma": 0.8314533233642578, "kind": "mosaic", "tokens": 264, "dense": 448, "grad_norm": 0.2324070930480957, "seconds": 2946.4} {"step": 589, "loss": 1.128939151763916, "data": 0.3055304288864136, "gap": 0.8234087228775024, "sigma": 0.9768217206001282, "kind": "mosaic", "tokens": 227, "dense": 448, "grad_norm": 0.37436017394065857, "seconds": 2950.9} {"step": 590, "loss": 0.6472638845443726, "data": 0.6472638845443726, "gap": null, "sigma": 0.9997388124465942, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.5840306282043457, "seconds": 2953.8} {"step": 591, "loss": 0.36290037631988525, "data": 0.36290037631988525, "gap": null, "sigma": 0.22831711173057556, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.17109155654907227, "seconds": 2956.6} {"step": 592, "loss": 0.5309725999832153, "data": 0.20980465412139893, "gap": 0.3211679756641388, "sigma": 0.7092174291610718, "kind": "mosaic", "tokens": 178, "dense": 448, "grad_norm": 0.14907246828079224, "seconds": 2960.9} {"step": 593, "loss": 0.6216516494750977, "data": 0.17277538776397705, "gap": 0.448876291513443, "sigma": 0.9410053491592407, "kind": "mosaic", "tokens": 188, "dense": 448, "grad_norm": 0.6363325119018555, "seconds": 2965.5} {"step": 594, "loss": 1.1730715036392212, "data": 0.4089486300945282, "gap": 0.7641228437423706, "sigma": 0.995600163936615, "kind": "mosaic", "tokens": 271, "dense": 448, "grad_norm": 1.3241682052612305, "seconds": 2970.0} {"step": 595, "loss": 0.667158842086792, "data": 0.25453418493270874, "gap": 0.41262462735176086, "sigma": 0.7403305172920227, "kind": "mosaic", "tokens": 186, "dense": 448, "grad_norm": 0.2579876184463501, "seconds": 2974.5} {"step": 596, "loss": 0.3999210596084595, "data": 0.15588080883026123, "gap": 0.24404026567935944, "sigma": 0.6306561231613159, "kind": "mosaic", "tokens": 176, "dense": 448, "grad_norm": 0.16203300654888153, "seconds": 2978.9} {"step": 597, "loss": 0.3349168002605438, "data": 0.17454056441783905, "gap": 0.16037623584270477, "sigma": 0.7088419795036316, "kind": "mosaic", "tokens": 252, "dense": 448, "grad_norm": 0.17043568193912506, "seconds": 2983.4} {"step": 598, "loss": 1.472124457359314, "data": 0.2370651215314865, "gap": 1.235059380531311, "sigma": 0.9984227418899536, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 1.3852919340133667, "seconds": 2987.5} {"step": 599, "loss": 0.5595969557762146, "data": 0.23479698598384857, "gap": 0.32479995489120483, "sigma": 0.4508785009384155, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.30556267499923706, "seconds": 2992.0} {"step": 600, "loss": 0.4997134208679199, "data": 0.2210136204957962, "gap": 0.2786997854709625, "sigma": 0.7233200073242188, "kind": "mosaic", "tokens": 228, "dense": 448, "grad_norm": 0.30149081349372864, "seconds": 2996.5} {"step": 601, "loss": 0.632541298866272, "data": 0.21214301884174347, "gap": 0.4203982651233673, "sigma": 0.5753442645072937, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.26307693123817444, "seconds": 3000.8} {"step": 602, "loss": 0.7090511918067932, "data": 0.08956367522478104, "gap": 0.6194875240325928, "sigma": 0.9578245282173157, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.7727471590042114, "seconds": 3004.9} {"step": 603, "loss": 0.6281289458274841, "data": 0.6281289458274841, "gap": null, "sigma": 0.9863549470901489, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.8784211874008179, "seconds": 3007.7} {"step": 604, "loss": 0.4843340814113617, "data": 0.17973071336746216, "gap": 0.30460336804389954, "sigma": 0.673646092414856, "kind": "mosaic", "tokens": 166, "dense": 448, "grad_norm": 0.1679663360118866, "seconds": 3012.3} {"step": 605, "loss": 0.3599297106266022, "data": 0.1664302945137024, "gap": 0.19349941611289978, "sigma": 0.714084267616272, "kind": "mosaic", "tokens": 277, "dense": 448, "grad_norm": 0.18196743726730347, "seconds": 3016.6} {"step": 606, "loss": 0.3922044634819031, "data": 0.3922044634819031, "gap": null, "sigma": 0.8984763026237488, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.1642322838306427, "seconds": 3019.4} {"step": 607, "loss": 0.6155181527137756, "data": 0.1938837617635727, "gap": 0.42163437604904175, "sigma": 0.9484402537345886, "kind": "mosaic", "tokens": 227, "dense": 448, "grad_norm": 0.3720586895942688, "seconds": 3024.0} {"step": 608, "loss": 0.64290851354599, "data": 0.22676384449005127, "gap": 0.4161446690559387, "sigma": 0.6516799330711365, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.18578492105007172, "seconds": 3028.3} {"step": 609, "loss": 0.8074110746383667, "data": 0.2616219222545624, "gap": 0.5457891821861267, "sigma": 0.4366152286529541, "kind": "mosaic", "tokens": 220, "dense": 448, "grad_norm": 0.1762215793132782, "seconds": 3032.6} {"step": 610, "loss": 1.1132328510284424, "data": 0.24964028596878052, "gap": 0.8635925054550171, "sigma": 0.9963024258613586, "kind": "mosaic", "tokens": 169, "dense": 448, "grad_norm": 0.8666104674339294, "seconds": 3037.0} {"step": 611, "loss": 0.6774488687515259, "data": 0.21400411427021027, "gap": 0.4634447395801544, "sigma": 0.9171174764633179, "kind": "mosaic", "tokens": 191, "dense": 448, "grad_norm": 0.6089087724685669, "seconds": 3041.3} {"step": 612, "loss": 0.6413711905479431, "data": 0.24086162447929382, "gap": 0.4005095660686493, "sigma": 0.5927878618240356, "kind": "mosaic", "tokens": 210, "dense": 448, "grad_norm": 0.1608181893825531, "seconds": 3045.8} {"step": 613, "loss": 0.33741143345832825, "data": 0.33741143345832825, "gap": null, "sigma": 0.8253292441368103, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.10495122522115707, "seconds": 3048.6} {"step": 614, "loss": 10.780787467956543, "data": 2.5224266052246094, "gap": 8.258360862731934, "sigma": 0.07485844939947128, "kind": "mosaic", "tokens": 207, "dense": 448, "grad_norm": 5.96135139465332, "seconds": 3053.1} {"step": 615, "loss": 2.1761112213134766, "data": 0.4527842402458191, "gap": 1.7233270406723022, "sigma": 0.9835661053657532, "kind": "mosaic", "tokens": 224, "dense": 512, "grad_norm": 3.140077829360962, "seconds": 3057.6} {"step": 616, "loss": 0.6984573006629944, "data": 0.6984573006629944, "gap": null, "sigma": 0.9959392547607422, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.6287544965744019, "seconds": 3060.5} {"step": 617, "loss": 1.3382065296173096, "data": 0.3125850558280945, "gap": 1.0256215333938599, "sigma": 0.9803633689880371, "kind": "mosaic", "tokens": 249, "dense": 448, "grad_norm": 2.31349515914917, "seconds": 3065.0} {"step": 618, "loss": 0.4826255440711975, "data": 0.21855531632900238, "gap": 0.26407021284103394, "sigma": 0.7860410809516907, "kind": "mosaic", "tokens": 245, "dense": 448, "grad_norm": 0.3635345697402954, "seconds": 3069.6} {"step": 619, "loss": 0.4622167944908142, "data": 0.4622167944908142, "gap": null, "sigma": 0.9978255033493042, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.37107178568840027, "seconds": 3072.4} {"step": 620, "loss": 0.6302393674850464, "data": 0.2532418966293335, "gap": 0.3769974410533905, "sigma": 0.37070223689079285, "kind": "mosaic", "tokens": 194, "dense": 448, "grad_norm": 0.2006395161151886, "seconds": 3076.9} {"step": 621, "loss": 0.35407784581184387, "data": 0.35407784581184387, "gap": null, "sigma": 0.9454600214958191, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2831818163394928, "seconds": 3079.8} {"step": 622, "loss": 2.166779041290283, "data": 0.6078952550888062, "gap": 1.5588839054107666, "sigma": 0.1517450362443924, "kind": "mosaic", "tokens": 257, "dense": 448, "grad_norm": 1.088019847869873, "seconds": 3084.4} {"step": 623, "loss": 0.45150431990623474, "data": 0.14626187086105347, "gap": 0.3052424490451813, "sigma": 0.5872308611869812, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.31616753339767456, "seconds": 3088.7} {"step": 624, "loss": 0.5589070320129395, "data": 0.5589070320129395, "gap": null, "sigma": 0.9810838103294373, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.35431715846061707, "seconds": 3091.4} {"step": 625, "loss": 0.6353654265403748, "data": 0.6353654265403748, "gap": null, "sigma": 0.9765263199806213, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.5140491127967834, "seconds": 3094.3} {"step": 626, "loss": 0.5415413975715637, "data": 0.2687472999095917, "gap": 0.27279409766197205, "sigma": 0.5672892928123474, "kind": "mosaic", "tokens": 270, "dense": 448, "grad_norm": 0.36527153849601746, "seconds": 3098.9} {"step": 627, "loss": 0.6890230178833008, "data": 0.21344031393527985, "gap": 0.47558271884918213, "sigma": 0.5640961527824402, "kind": "mosaic", "tokens": 169, "dense": 448, "grad_norm": 0.2606440782546997, "seconds": 3103.4} {"step": 628, "loss": 0.687424898147583, "data": 0.08298167586326599, "gap": 0.6044432520866394, "sigma": 0.9165192246437073, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.688159167766571, "seconds": 3107.7} {"step": 629, "loss": 0.44103559851646423, "data": 0.44103559851646423, "gap": null, "sigma": 0.9476974606513977, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.167276531457901, "seconds": 3110.5} {"step": 630, "loss": 1.0810108184814453, "data": 0.2879849672317505, "gap": 0.7930258512496948, "sigma": 0.4222119152545929, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.5540751218795776, "seconds": 3114.8} {"step": 631, "loss": 0.9751229882240295, "data": 0.16330741345882416, "gap": 0.8118155598640442, "sigma": 0.4683610200881958, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.7680302262306213, "seconds": 3118.9} {"step": 632, "loss": 2.2413642406463623, "data": 0.27409836649894714, "gap": 1.9672658443450928, "sigma": 0.9864919781684875, "kind": "mosaic", "tokens": 219, "dense": 448, "grad_norm": 1.5453888177871704, "seconds": 3123.4} {"step": 633, "loss": 0.3197292983531952, "data": 0.3197292983531952, "gap": null, "sigma": 0.8851401805877686, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.21329320967197418, "seconds": 3126.2} {"step": 634, "loss": 0.6887068748474121, "data": 0.23444747924804688, "gap": 0.45425936579704285, "sigma": 0.8495097160339355, "kind": "mosaic", "tokens": 196, "dense": 448, "grad_norm": 0.3182891011238098, "seconds": 3130.7} {"step": 635, "loss": 0.8716298937797546, "data": 0.18307869136333466, "gap": 0.6885511875152588, "sigma": 0.9739781618118286, "kind": "mosaic", "tokens": 248, "dense": 448, "grad_norm": 0.5717733502388, "seconds": 3135.4} {"step": 636, "loss": 0.6518639326095581, "data": 0.2366582602262497, "gap": 0.4152056872844696, "sigma": 0.8209174275398254, "kind": "mosaic", "tokens": 212, "dense": 448, "grad_norm": 0.3431320786476135, "seconds": 3139.9} {"step": 637, "loss": 0.4298558533191681, "data": 0.4298558533191681, "gap": null, "sigma": 0.9465991854667664, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.20877599716186523, "seconds": 3142.7} {"step": 638, "loss": 0.5418793559074402, "data": 0.5418793559074402, "gap": null, "sigma": 0.9614440202713013, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.33042460680007935, "seconds": 3145.6} {"step": 639, "loss": 0.4846666157245636, "data": 0.4846666157245636, "gap": null, "sigma": 0.9854356646537781, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.32951295375823975, "seconds": 3148.4} {"step": 640, "loss": 0.7106929421424866, "data": 0.24471020698547363, "gap": 0.46598273515701294, "sigma": 0.37830665707588196, "kind": "mosaic", "tokens": 226, "dense": 448, "grad_norm": 0.1757616251707077, "seconds": 3153.0} {"step": 641, "loss": 1.9197500944137573, "data": 0.27152153849601746, "gap": 1.6482285261154175, "sigma": 0.9774367809295654, "kind": "mosaic", "tokens": 274, "dense": 448, "grad_norm": 0.8656370043754578, "seconds": 3157.6} {"step": 642, "loss": 0.6449145078659058, "data": 0.35367605090141296, "gap": 0.2912384271621704, "sigma": 0.9789921045303345, "kind": "mosaic", "tokens": 177, "dense": 480, "grad_norm": 0.3605271279811859, "seconds": 3161.9} {"step": 643, "loss": 0.6662139892578125, "data": 0.21347405016422272, "gap": 0.452739953994751, "sigma": 0.5090160965919495, "kind": "mosaic", "tokens": 210, "dense": 448, "grad_norm": 0.22472307085990906, "seconds": 3166.4} {"step": 644, "loss": 10.918396949768066, "data": 2.637814521789551, "gap": 8.280582427978516, "sigma": 0.04834757372736931, "kind": "mosaic", "tokens": 236, "dense": 448, "grad_norm": 8.248929977416992, "seconds": 3171.0} {"step": 645, "loss": 0.7658491730690002, "data": 0.7658491730690002, "gap": null, "sigma": 0.9967558979988098, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 1.07303786277771, "seconds": 3173.8} {"step": 646, "loss": 0.6252927780151367, "data": 0.3163737952709198, "gap": 0.3089189827442169, "sigma": 0.9433229565620422, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.9141508340835571, "seconds": 3178.4} {"step": 647, "loss": 0.3837168216705322, "data": 0.2011144608259201, "gap": 0.18260234594345093, "sigma": 0.5827264785766602, "kind": "mosaic", "tokens": 248, "dense": 448, "grad_norm": 0.14716307818889618, "seconds": 3182.8} {"step": 648, "loss": 0.7922645807266235, "data": 0.3573393225669861, "gap": 0.43492528796195984, "sigma": 0.4307067096233368, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.225803405046463, "seconds": 3187.2} {"step": 649, "loss": 0.5063779354095459, "data": 0.21695050597190857, "gap": 0.28942739963531494, "sigma": 0.4661094546318054, "kind": "mosaic", "tokens": 228, "dense": 448, "grad_norm": 0.14085835218429565, "seconds": 3191.6} {"step": 650, "loss": 5.526341438293457, "data": 1.3590469360351562, "gap": 4.167294502258301, "sigma": 0.0793784111738205, "kind": "mosaic", "tokens": 272, "dense": 448, "grad_norm": 3.6214492321014404, "seconds": 3196.1} {"step": 651, "loss": 1.598552942276001, "data": 0.4281005859375, "gap": 1.170452356338501, "sigma": 0.23713117837905884, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.576165497303009, "seconds": 3200.2} {"step": 652, "loss": 0.7019327282905579, "data": 0.2551519274711609, "gap": 0.446780800819397, "sigma": 0.4662719666957855, "kind": "mosaic", "tokens": 246, "dense": 448, "grad_norm": 0.1828506588935852, "seconds": 3204.7} {"step": 653, "loss": 1.0450506210327148, "data": 0.4631370007991791, "gap": 0.5819136500358582, "sigma": 0.1980801373720169, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.623894214630127, "seconds": 3209.2} {"step": 654, "loss": 0.6196338534355164, "data": 0.2359962910413742, "gap": 0.38363757729530334, "sigma": 0.7650236487388611, "kind": "mosaic", "tokens": 243, "dense": 448, "grad_norm": 0.22381111979484558, "seconds": 3213.7} {"step": 655, "loss": 0.639747142791748, "data": 0.05663367733359337, "gap": 0.5831134915351868, "sigma": 0.6934536695480347, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 0.4438813328742981, "seconds": 3217.7} {"step": 656, "loss": 0.7514646053314209, "data": 0.28462696075439453, "gap": 0.466837614774704, "sigma": 0.9810476899147034, "kind": "mosaic", "tokens": 214, "dense": 448, "grad_norm": 0.30805084109306335, "seconds": 3222.2} {"step": 657, "loss": 0.5230062007904053, "data": 0.1940709501504898, "gap": 0.3289352357387543, "sigma": 0.7352482676506042, "kind": "mosaic", "tokens": 178, "dense": 448, "grad_norm": 0.1859387755393982, "seconds": 3226.8} {"step": 658, "loss": 0.5063279867172241, "data": 0.20646093785762787, "gap": 0.29986706376075745, "sigma": 0.7248632907867432, "kind": "mosaic", "tokens": 268, "dense": 448, "grad_norm": 0.24272115528583527, "seconds": 3231.5} {"step": 659, "loss": 21.07189178466797, "data": 3.899838924407959, "gap": 17.17205238342285, "sigma": 0.01658381149172783, "kind": "mosaic", "tokens": 203, "dense": 448, "grad_norm": 17.793668746948242, "seconds": 3236.0} {"step": 660, "loss": 0.7212751507759094, "data": 0.22379006445407867, "gap": 0.49748510122299194, "sigma": 0.948836624622345, "kind": "mosaic", "tokens": 281, "dense": 448, "grad_norm": 0.48602914810180664, "seconds": 3240.4} {"step": 661, "loss": 5.408147811889648, "data": 1.1615149974822998, "gap": 4.2466325759887695, "sigma": 0.0921802744269371, "kind": "mosaic", "tokens": 274, "dense": 448, "grad_norm": 3.276650905609131, "seconds": 3245.1} {"step": 662, "loss": 0.2520478069782257, "data": 0.2520478069782257, "gap": null, "sigma": 0.8902727365493774, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.44396722316741943, "seconds": 3247.9} {"step": 663, "loss": 0.949101448059082, "data": 0.2582993805408478, "gap": 0.6908020973205566, "sigma": 0.49807074666023254, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.42998701333999634, "seconds": 3252.2} {"step": 664, "loss": 0.6459502577781677, "data": 0.24978719651699066, "gap": 0.39616307616233826, "sigma": 0.8658459782600403, "kind": "mosaic", "tokens": 253, "dense": 448, "grad_norm": 0.3306601047515869, "seconds": 3256.7} {"step": 665, "loss": 0.47505366802215576, "data": 0.09702317416667938, "gap": 0.3780305087566376, "sigma": 0.7713029980659485, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.18843816220760345, "seconds": 3260.8} {"step": 666, "loss": 0.25886014103889465, "data": 0.25886014103889465, "gap": null, "sigma": 0.8137025237083435, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.14431869983673096, "seconds": 3263.6} {"step": 667, "loss": 0.7791592478752136, "data": 0.28501784801483154, "gap": 0.4941413998603821, "sigma": 0.5261076092720032, "kind": "mosaic", "tokens": 256, "dense": 448, "grad_norm": 0.23108813166618347, "seconds": 3268.1} {"step": 668, "loss": 0.3881083130836487, "data": 0.3881083130836487, "gap": null, "sigma": 0.8343539834022522, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.16777634620666504, "seconds": 3271.0} {"step": 669, "loss": 0.4445474445819855, "data": 0.16633838415145874, "gap": 0.27820906043052673, "sigma": 0.7417117357254028, "kind": "mosaic", "tokens": 246, "dense": 448, "grad_norm": 0.20094716548919678, "seconds": 3275.7} {"step": 670, "loss": 0.530053973197937, "data": 0.23397332429885864, "gap": 0.29608067870140076, "sigma": 0.6345142126083374, "kind": "mosaic", "tokens": 264, "dense": 448, "grad_norm": 0.1194424256682396, "seconds": 3280.4} {"step": 671, "loss": 1.5564216375350952, "data": 0.2959761321544647, "gap": 1.260445475578308, "sigma": 0.9578574299812317, "kind": "mosaic", "tokens": 236, "dense": 448, "grad_norm": 0.8197267651557922, "seconds": 3285.0} {"step": 672, "loss": 0.5821194648742676, "data": 0.5821194648742676, "gap": null, "sigma": 0.9722999930381775, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.6621083617210388, "seconds": 3287.8} {"step": 673, "loss": 0.8811781406402588, "data": 0.1337674856185913, "gap": 0.7474106550216675, "sigma": 0.6276999115943909, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.5304073095321655, "seconds": 3292.0} {"step": 674, "loss": 0.8771281242370605, "data": 0.273703932762146, "gap": 0.6034241914749146, "sigma": 0.7828751802444458, "kind": "mosaic", "tokens": 261, "dense": 448, "grad_norm": 0.3268881142139435, "seconds": 3296.6} {"step": 675, "loss": 0.5317139625549316, "data": 0.3092404901981354, "gap": 0.22247345745563507, "sigma": 0.9320824146270752, "kind": "mosaic", "tokens": 241, "dense": 448, "grad_norm": 0.15887685120105743, "seconds": 3301.1} {"step": 676, "loss": 0.3477385640144348, "data": 0.141903817653656, "gap": 0.20583473145961761, "sigma": 0.6696447730064392, "kind": "mosaic", "tokens": 198, "dense": 448, "grad_norm": 0.20167851448059082, "seconds": 3305.6} {"step": 677, "loss": 0.5867166519165039, "data": 0.2013402283191681, "gap": 0.3853764235973358, "sigma": 0.8091822862625122, "kind": "mosaic", "tokens": 203, "dense": 448, "grad_norm": 0.23144885897636414, "seconds": 3310.0} {"step": 678, "loss": 0.5907676219940186, "data": 0.5907676219940186, "gap": null, "sigma": 0.9788689017295837, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.9251371026039124, "seconds": 3312.7} {"step": 679, "loss": 0.25615012645721436, "data": 0.25615012645721436, "gap": null, "sigma": 0.878984272480011, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3852507770061493, "seconds": 3315.6} {"step": 680, "loss": 11.874727249145508, "data": 2.6056151390075684, "gap": 9.269111633300781, "sigma": 0.0019371593371033669, "kind": "mosaic", "tokens": 245, "dense": 448, "grad_norm": 9.70048713684082, "seconds": 3320.3} {"step": 681, "loss": 0.9343290328979492, "data": 0.2951987087726593, "gap": 0.6391303539276123, "sigma": 0.9242391586303711, "kind": "mosaic", "tokens": 240, "dense": 448, "grad_norm": 1.1428871154785156, "seconds": 3324.8} {"step": 682, "loss": 0.4316154718399048, "data": 0.4316154718399048, "gap": null, "sigma": 0.9273623824119568, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.28979256749153137, "seconds": 3327.6} {"step": 683, "loss": 0.15695905685424805, "data": 0.15695905685424805, "gap": null, "sigma": 0.7294620871543884, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.08788876235485077, "seconds": 3330.6} {"step": 684, "loss": 0.3434312343597412, "data": 0.23204205930233002, "gap": 0.11138918995857239, "sigma": 0.8159549236297607, "kind": "mosaic", "tokens": 281, "dense": 480, "grad_norm": 0.16454951465129852, "seconds": 3335.2} {"step": 685, "loss": 0.4504837989807129, "data": 0.18988527357578278, "gap": 0.2605985105037689, "sigma": 0.4332335591316223, "kind": "mosaic", "tokens": 261, "dense": 448, "grad_norm": 0.14415381848812103, "seconds": 3339.7} {"step": 686, "loss": 0.4152678847312927, "data": 0.17936797440052032, "gap": 0.2358999103307724, "sigma": 0.8424046635627747, "kind": "mosaic", "tokens": 238, "dense": 448, "grad_norm": 0.2309446483850479, "seconds": 3344.3} {"step": 687, "loss": 0.5922013521194458, "data": 0.2236071079969406, "gap": 0.368594229221344, "sigma": 0.8968657851219177, "kind": "mosaic", "tokens": 237, "dense": 448, "grad_norm": 0.3175244629383087, "seconds": 3348.9} {"step": 688, "loss": 0.7447052001953125, "data": 0.32143256068229675, "gap": 0.42327260971069336, "sigma": 0.42261046171188354, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.2668343186378479, "seconds": 3353.4} {"step": 689, "loss": 1.122119426727295, "data": 0.29387885332107544, "gap": 0.8282405138015747, "sigma": 0.9971565008163452, "kind": "mosaic", "tokens": 237, "dense": 448, "grad_norm": 0.5296809673309326, "seconds": 3357.9} {"step": 690, "loss": 0.38608282804489136, "data": 0.1608627885580063, "gap": 0.22522002458572388, "sigma": 0.533557653427124, "kind": "mosaic", "tokens": 220, "dense": 448, "grad_norm": 0.18517668545246124, "seconds": 3362.4} {"step": 691, "loss": 1.200059175491333, "data": 0.39412087202072144, "gap": 0.8059383630752563, "sigma": 0.19917765259742737, "kind": "mosaic", "tokens": 250, "dense": 448, "grad_norm": 0.5153498649597168, "seconds": 3367.1} {"step": 692, "loss": 0.6014264822006226, "data": 0.19672246277332306, "gap": 0.4047040343284607, "sigma": 0.5623985528945923, "kind": "mosaic", "tokens": 177, "dense": 448, "grad_norm": 0.2320159673690796, "seconds": 3371.5} {"step": 693, "loss": 0.4499407410621643, "data": 0.28441351652145386, "gap": 0.16552723944187164, "sigma": 0.9346681833267212, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.30066797137260437, "seconds": 3376.0} {"step": 694, "loss": 0.7648155689239502, "data": 0.2676493525505066, "gap": 0.4971662163734436, "sigma": 0.27139517664909363, "kind": "mosaic", "tokens": 192, "dense": 448, "grad_norm": 0.4503791928291321, "seconds": 3380.6} {"step": 695, "loss": 35.82472229003906, "data": 5.326043128967285, "gap": 30.49867820739746, "sigma": 0.032799720764160156, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 22.175127029418945, "seconds": 3384.8} {"step": 696, "loss": 0.3449355363845825, "data": 0.16354495286941528, "gap": 0.18139058351516724, "sigma": 0.6458543539047241, "kind": "mosaic", "tokens": 234, "dense": 448, "grad_norm": 0.20695237815380096, "seconds": 3389.3} {"step": 697, "loss": 0.68879234790802, "data": 0.17165502905845642, "gap": 0.5171372890472412, "sigma": 0.8395284414291382, "kind": "mosaic", "tokens": 225, "dense": 448, "grad_norm": 0.8280972242355347, "seconds": 3393.8} {"step": 698, "loss": 0.893413782119751, "data": 0.25948506593704224, "gap": 0.6339287161827087, "sigma": 0.9400593638420105, "kind": "mosaic", "tokens": 257, "dense": 448, "grad_norm": 0.6264004111289978, "seconds": 3398.3} {"step": 699, "loss": 0.7489341497421265, "data": 0.1933615654706955, "gap": 0.5555725693702698, "sigma": 0.34572166204452515, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.4029827117919922, "seconds": 3402.5} {"step": 700, "loss": 1.303199291229248, "data": 0.4152586758136749, "gap": 0.8879406452178955, "sigma": 0.18939395248889923, "kind": "mosaic", "tokens": 210, "dense": 448, "grad_norm": 0.49612390995025635, "seconds": 3406.9} {"step": 701, "loss": 0.8327755928039551, "data": 0.2615521550178528, "gap": 0.5712234377861023, "sigma": 0.9631719589233398, "kind": "mosaic", "tokens": 229, "dense": 448, "grad_norm": 0.7426238656044006, "seconds": 3411.4} {"step": 702, "loss": 0.7513352036476135, "data": 0.18620632588863373, "gap": 0.5651288628578186, "sigma": 0.8635388016700745, "kind": "mosaic", "tokens": 186, "dense": 448, "grad_norm": 0.4836704730987549, "seconds": 3416.1} {"step": 703, "loss": 0.6553263664245605, "data": 0.2039862871170044, "gap": 0.45134007930755615, "sigma": 0.8054440021514893, "kind": "mosaic", "tokens": 201, "dense": 448, "grad_norm": 0.6876180171966553, "seconds": 3420.6} {"step": 704, "loss": 0.34616392850875854, "data": 0.34616392850875854, "gap": null, "sigma": 0.8626114726066589, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.20328740775585175, "seconds": 3423.4} {"step": 705, "loss": 0.5398792028427124, "data": 0.12115433067083359, "gap": 0.4187248945236206, "sigma": 0.8575121164321899, "kind": "mosaic", "tokens": 215, "dense": 448, "grad_norm": 0.38532549142837524, "seconds": 3428.0} {"step": 706, "loss": 0.34727761149406433, "data": 0.34727761149406433, "gap": null, "sigma": 0.8484840393066406, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.12062614411115646, "seconds": 3430.7} {"step": 707, "loss": 0.6079947352409363, "data": 0.1700207144021988, "gap": 0.4379740357398987, "sigma": 0.525104284286499, "kind": "mosaic", "tokens": 163, "dense": 448, "grad_norm": 0.20151807367801666, "seconds": 3435.1} {"step": 708, "loss": 0.4741035997867584, "data": 0.16503360867500305, "gap": 0.30906999111175537, "sigma": 0.7839000225067139, "kind": "mosaic", "tokens": 212, "dense": 448, "grad_norm": 0.1979176104068756, "seconds": 3439.7} {"step": 709, "loss": 0.8530269861221313, "data": 0.4007440507411957, "gap": 0.45228293538093567, "sigma": 0.365640252828598, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.20270518958568573, "seconds": 3444.4} {"step": 710, "loss": 0.71275794506073, "data": 0.23222097754478455, "gap": 0.48053693771362305, "sigma": 0.4543392062187195, "kind": "mosaic", "tokens": 220, "dense": 448, "grad_norm": 0.20184555649757385, "seconds": 3448.7} {"step": 711, "loss": 0.4223926067352295, "data": 0.19300763309001923, "gap": 0.22938497364521027, "sigma": 0.9429751634597778, "kind": "mosaic", "tokens": 223, "dense": 448, "grad_norm": 0.401650607585907, "seconds": 3453.2} {"step": 712, "loss": 0.28060558438301086, "data": 0.28060558438301086, "gap": null, "sigma": 0.907170832157135, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.19317808747291565, "seconds": 3456.0} {"step": 713, "loss": 0.4764442443847656, "data": 0.27959850430488586, "gap": 0.19684572517871857, "sigma": 0.911615252494812, "kind": "mosaic", "tokens": 257, "dense": 448, "grad_norm": 0.22556036710739136, "seconds": 3460.7} {"step": 714, "loss": 0.17518217861652374, "data": 0.17518217861652374, "gap": null, "sigma": 0.7083002328872681, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.14429114758968353, "seconds": 3463.5} {"step": 715, "loss": 0.5807385444641113, "data": 0.21413204073905945, "gap": 0.36660653352737427, "sigma": 0.5056746602058411, "kind": "mosaic", "tokens": 255, "dense": 448, "grad_norm": 0.34701988101005554, "seconds": 3467.9} {"step": 716, "loss": 0.221553236246109, "data": 0.09187160432338715, "gap": 0.12968163192272186, "sigma": 0.6976786255836487, "kind": "mosaic", "tokens": 248, "dense": 448, "grad_norm": 0.12280245870351791, "seconds": 3472.3} {"step": 717, "loss": 0.3517340421676636, "data": 0.16888055205345154, "gap": 0.18285349011421204, "sigma": 0.6732515692710876, "kind": "mosaic", "tokens": 218, "dense": 448, "grad_norm": 0.14497020840644836, "seconds": 3476.9} {"step": 718, "loss": 0.3547697067260742, "data": 0.1082749143242836, "gap": 0.24649478495121002, "sigma": 0.4725411832332611, "kind": "mosaic", "tokens": 198, "dense": 448, "grad_norm": 0.1522953063249588, "seconds": 3481.2} {"step": 719, "loss": 0.6324082612991333, "data": 0.2953089773654938, "gap": 0.3370993137359619, "sigma": 0.47421857714653015, "kind": "mosaic", "tokens": 282, "dense": 448, "grad_norm": 0.2284211963415146, "seconds": 3485.6} {"step": 720, "loss": 1.0186305046081543, "data": 0.24572795629501343, "gap": 0.7729025483131409, "sigma": 0.9424241781234741, "kind": "mosaic", "tokens": 237, "dense": 448, "grad_norm": 0.6781524419784546, "seconds": 3490.1} {"step": 721, "loss": 0.9552250504493713, "data": 0.23988784849643707, "gap": 0.7153372168540955, "sigma": 0.42352378368377686, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.22374027967453003, "seconds": 3494.4} {"step": 722, "loss": 0.6324424743652344, "data": 0.2588430941104889, "gap": 0.37359941005706787, "sigma": 0.8417008519172668, "kind": "mosaic", "tokens": 262, "dense": 448, "grad_norm": 0.21090270578861237, "seconds": 3499.0} {"step": 723, "loss": 0.21204203367233276, "data": 0.21204203367233276, "gap": null, "sigma": 0.7617525458335876, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.1295800805091858, "seconds": 3501.8} {"step": 724, "loss": 0.45824694633483887, "data": 0.19074171781539917, "gap": 0.2675052285194397, "sigma": 0.9593435525894165, "kind": "mosaic", "tokens": 177, "dense": 448, "grad_norm": 0.28332605957984924, "seconds": 3506.2} {"step": 725, "loss": 0.4325408935546875, "data": 0.4325408935546875, "gap": null, "sigma": 0.9996548295021057, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.7158041000366211, "seconds": 3509.1} {"step": 726, "loss": 0.6586617231369019, "data": 0.22398129105567932, "gap": 0.43468040227890015, "sigma": 0.9396181702613831, "kind": "mosaic", "tokens": 210, "dense": 448, "grad_norm": 0.9497061967849731, "seconds": 3513.6} {"step": 727, "loss": 0.3669689893722534, "data": 0.3669689893722534, "gap": null, "sigma": 0.9139070510864258, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.42512214183807373, "seconds": 3516.4} {"step": 728, "loss": 0.5175096392631531, "data": 0.17777971923351288, "gap": 0.339729905128479, "sigma": 0.7832565307617188, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.22590386867523193, "seconds": 3520.8} {"step": 729, "loss": 0.5724551677703857, "data": 0.23025953769683838, "gap": 0.34219563007354736, "sigma": 0.6245192289352417, "kind": "mosaic", "tokens": 216, "dense": 448, "grad_norm": 0.14688342809677124, "seconds": 3525.2} {"step": 730, "loss": 12.23194694519043, "data": 2.517249822616577, "gap": 9.714696884155273, "sigma": 0.018742535263299942, "kind": "mosaic", "tokens": 171, "dense": 448, "grad_norm": 10.290884017944336, "seconds": 3529.5} {"step": 731, "loss": 0.7150282859802246, "data": 0.2514228820800781, "gap": 0.4636053740978241, "sigma": 0.8460284471511841, "kind": "mosaic", "tokens": 212, "dense": 448, "grad_norm": 0.5646591782569885, "seconds": 3534.2} {"step": 732, "loss": 1.452558994293213, "data": 0.5533562898635864, "gap": 0.8992026448249817, "sigma": 0.1718212068080902, "kind": "mosaic", "tokens": 276, "dense": 448, "grad_norm": 1.999131441116333, "seconds": 3538.7} {"step": 733, "loss": 0.5210540294647217, "data": 0.5210540294647217, "gap": null, "sigma": 0.9430974125862122, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2542833387851715, "seconds": 3541.5} {"step": 734, "loss": 0.15059654414653778, "data": 0.15059654414653778, "gap": null, "sigma": 0.5632902979850769, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.06352725625038147, "seconds": 3544.3} {"step": 735, "loss": 1.2822093963623047, "data": 0.514902651309967, "gap": 0.7673068046569824, "sigma": 0.23635978996753693, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.6041049957275391, "seconds": 3548.7} {"step": 736, "loss": 0.6200051307678223, "data": 0.1088196262717247, "gap": 0.5111855268478394, "sigma": 0.5840829610824585, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.2693057656288147, "seconds": 3552.8} {"step": 737, "loss": 0.5830777883529663, "data": 0.2369278073310852, "gap": 0.3461500108242035, "sigma": 0.5374797582626343, "kind": "mosaic", "tokens": 224, "dense": 448, "grad_norm": 0.13046970963478088, "seconds": 3557.4} {"step": 738, "loss": 2.3475406169891357, "data": 0.6825255751609802, "gap": 1.6650149822235107, "sigma": 0.20348230004310608, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 1.1449813842773438, "seconds": 3561.6} {"step": 739, "loss": 0.5803711414337158, "data": 0.22724688053131104, "gap": 0.3531242311000824, "sigma": 0.546532154083252, "kind": "mosaic", "tokens": 246, "dense": 448, "grad_norm": 0.1587912142276764, "seconds": 3566.2} {"step": 740, "loss": 0.6122793555259705, "data": 0.19636322557926178, "gap": 0.4159161448478699, "sigma": 0.42388835549354553, "kind": "mosaic", "tokens": 227, "dense": 448, "grad_norm": 0.18217845261096954, "seconds": 3570.7} {"step": 741, "loss": 0.2584170997142792, "data": 0.2584170997142792, "gap": null, "sigma": 0.7930293679237366, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.19235539436340332, "seconds": 3573.6} {"step": 742, "loss": 0.460939884185791, "data": 0.2531794607639313, "gap": 0.20776042342185974, "sigma": 0.9644140005111694, "kind": "mosaic", "tokens": 217, "dense": 448, "grad_norm": 0.35087019205093384, "seconds": 3578.3} {"step": 743, "loss": 0.7484767436981201, "data": 0.18676158785820007, "gap": 0.5617151260375977, "sigma": 0.8180500864982605, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.33317580819129944, "seconds": 3582.6} {"step": 744, "loss": 1.2402784824371338, "data": 0.37196046113967896, "gap": 0.8683180212974548, "sigma": 0.3508053719997406, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.4929460287094116, "seconds": 3587.0} {"step": 745, "loss": 0.3656574785709381, "data": 0.3656574785709381, "gap": null, "sigma": 0.9060502648353577, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.10902601480484009, "seconds": 3589.9} {"step": 746, "loss": 0.7829079627990723, "data": 0.23694384098052979, "gap": 0.5459641218185425, "sigma": 0.9148995876312256, "kind": "mosaic", "tokens": 224, "dense": 448, "grad_norm": 0.4636574387550354, "seconds": 3594.4} {"step": 747, "loss": 0.47281163930892944, "data": 0.1663920134305954, "gap": 0.30641964077949524, "sigma": 0.8481619358062744, "kind": "mosaic", "tokens": 187, "dense": 448, "grad_norm": 0.2054213583469391, "seconds": 3599.0} {"step": 748, "loss": 5.434081554412842, "data": 1.103040099143982, "gap": 4.33104133605957, "sigma": 0.0880354642868042, "kind": "mosaic", "tokens": 217, "dense": 448, "grad_norm": 3.150491237640381, "seconds": 3603.7} {"step": 749, "loss": 0.6595754623413086, "data": 0.15486900508403778, "gap": 0.5047064423561096, "sigma": 0.8503766059875488, "kind": "mosaic", "tokens": 222, "dense": 448, "grad_norm": 0.31972965598106384, "seconds": 3608.3} {"step": 750, "loss": 0.3745877146720886, "data": 0.15583397448062897, "gap": 0.21875375509262085, "sigma": 0.7665547728538513, "kind": "mosaic", "tokens": 218, "dense": 448, "grad_norm": 0.20192772150039673, "seconds": 3613.0} {"step": 750, "eval": {"dense": 0.3093707642207543, "uniform2": 0.18456867119918266, "mosaic": 0.2516392469406128, "gap_uniform2": 0.6092246932288011, "gap_mosaic": 0.6071007971962293}, "eval_extent": {"1x1x1": 0.414478103033482, "1x1x2": 0.32072878537001087, "1x1x4": 0.23084642235679845, "1x2x1": 0.2985250933337267, "1x2x2": 0.19827988705793886, "1x2x4": 0.13018494442395576, "1x4x1": 0.2033258731736719, "1x4x2": 0.11258935491657919, "1x4x4": 0.0846940486170267}} {"step": 751, "loss": 0.3970564603805542, "data": 0.15951204299926758, "gap": 0.23754441738128662, "sigma": 0.5674445033073425, "kind": "mosaic", "tokens": 226, "dense": 448, "grad_norm": 0.13537746667861938, "seconds": 3858.7} {"step": 752, "loss": 0.2064877301454544, "data": 0.2064877301454544, "gap": null, "sigma": 0.6535435914993286, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.1700417548418045, "seconds": 3861.5} {"step": 753, "loss": 1.656245231628418, "data": 0.4849478602409363, "gap": 1.171297311782837, "sigma": 0.2340957224369049, "kind": "mosaic", "tokens": 220, "dense": 448, "grad_norm": 0.47960326075553894, "seconds": 3866.0} {"step": 754, "loss": 0.49185246229171753, "data": 0.21509504318237305, "gap": 0.2767574191093445, "sigma": 0.7782038450241089, "kind": "mosaic", "tokens": 265, "dense": 448, "grad_norm": 0.23092731833457947, "seconds": 3870.5} {"step": 755, "loss": 0.5267477035522461, "data": 0.19626013934612274, "gap": 0.33048754930496216, "sigma": 0.9197144508361816, "kind": "mosaic", "tokens": 181, "dense": 448, "grad_norm": 0.3150429129600525, "seconds": 3874.9} {"step": 756, "loss": 0.22601772844791412, "data": 0.22601772844791412, "gap": null, "sigma": 0.7703086137771606, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.09659882634878159, "seconds": 3877.9} {"step": 757, "loss": 0.7035772204399109, "data": 0.2948472797870636, "gap": 0.4087299406528473, "sigma": 0.38697996735572815, "kind": "mosaic", "tokens": 280, "dense": 448, "grad_norm": 0.20954373478889465, "seconds": 3882.5} {"step": 758, "loss": 0.2998746633529663, "data": 0.2998746633529663, "gap": null, "sigma": 0.8883606791496277, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.1324847787618637, "seconds": 3885.2} {"step": 759, "loss": 0.41389453411102295, "data": 0.41389453411102295, "gap": null, "sigma": 0.9553840160369873, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.17714080214500427, "seconds": 3888.2} {"step": 760, "loss": 0.4973906874656677, "data": 0.11910147964954376, "gap": 0.37828919291496277, "sigma": 0.6476667523384094, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.26068201661109924, "seconds": 3892.5} {"step": 761, "loss": 0.511201024055481, "data": 0.2151799201965332, "gap": 0.29602107405662537, "sigma": 0.600942850112915, "kind": "mosaic", "tokens": 223, "dense": 448, "grad_norm": 0.12559184432029724, "seconds": 3896.9} {"step": 762, "loss": 0.34681418538093567, "data": 0.34681418538093567, "gap": null, "sigma": 0.8606471419334412, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.20361356437206268, "seconds": 3899.8} {"step": 763, "loss": 0.6629012227058411, "data": 0.25433388352394104, "gap": 0.4085673391819, "sigma": 0.9045979976654053, "kind": "mosaic", "tokens": 239, "dense": 448, "grad_norm": 0.1699177473783493, "seconds": 3904.2} {"step": 764, "loss": 0.47531652450561523, "data": 0.2957429587841034, "gap": 0.17957355082035065, "sigma": 0.9075822830200195, "kind": "mosaic", "tokens": 249, "dense": 448, "grad_norm": 0.40210291743278503, "seconds": 3908.7} {"step": 765, "loss": 0.7798808813095093, "data": 0.24741792678833008, "gap": 0.5324629545211792, "sigma": 0.9520424604415894, "kind": "mosaic", "tokens": 161, "dense": 448, "grad_norm": 0.481220006942749, "seconds": 3913.2} {"step": 766, "loss": 11.283700942993164, "data": 2.6058571338653564, "gap": 8.677844047546387, "sigma": 0.06688656657934189, "kind": "mosaic", "tokens": 225, "dense": 448, "grad_norm": 6.677892684936523, "seconds": 3917.6} {"step": 767, "loss": 0.4233129322528839, "data": 0.4233129322528839, "gap": null, "sigma": 0.8963865637779236, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.12290921062231064, "seconds": 3920.4} {"step": 768, "loss": 0.42617592215538025, "data": 0.14097577333450317, "gap": 0.2852001488208771, "sigma": 0.8961041569709778, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.357430636882782, "seconds": 3924.6} {"step": 769, "loss": 0.9104592800140381, "data": 0.31669825315475464, "gap": 0.5937610268592834, "sigma": 0.3313000202178955, "kind": "mosaic", "tokens": 208, "dense": 448, "grad_norm": 0.26135435700416565, "seconds": 3929.2} {"step": 770, "loss": 20.27513313293457, "data": 3.3247299194335938, "gap": 16.950403213500977, "sigma": 0.04732297360897064, "kind": "mosaic", "tokens": 151, "dense": 448, "grad_norm": 12.801359176635742, "seconds": 3933.6} {"step": 771, "loss": 0.2205205261707306, "data": 0.12178389728069305, "gap": 0.09873663634061813, "sigma": 0.6084046959877014, "kind": "mosaic", "tokens": 259, "dense": 448, "grad_norm": 0.08936277776956558, "seconds": 3938.1} {"step": 772, "loss": 0.9082176685333252, "data": 0.20179317891597748, "gap": 0.7064244747161865, "sigma": 0.8333340883255005, "kind": "mosaic", "tokens": 270, "dense": 448, "grad_norm": 0.521181583404541, "seconds": 3942.6} {"step": 773, "loss": 0.24990777671337128, "data": 0.09485134482383728, "gap": 0.155056431889534, "sigma": 0.7289704084396362, "kind": "mosaic", "tokens": 253, "dense": 448, "grad_norm": 0.12980684638023376, "seconds": 3947.1} {"step": 774, "loss": 0.6205723285675049, "data": 0.214820995926857, "gap": 0.4057513475418091, "sigma": 0.8406376242637634, "kind": "mosaic", "tokens": 247, "dense": 448, "grad_norm": 0.2623092234134674, "seconds": 3951.5} {"step": 775, "loss": 0.5860112905502319, "data": 0.21067661046981812, "gap": 0.3753347098827362, "sigma": 0.5196271538734436, "kind": "mosaic", "tokens": 217, "dense": 448, "grad_norm": 0.1581587940454483, "seconds": 3956.0} {"step": 776, "loss": 0.4407847225666046, "data": 0.19241860508918762, "gap": 0.248366117477417, "sigma": 0.475276917219162, "kind": "mosaic", "tokens": 246, "dense": 448, "grad_norm": 0.14605805277824402, "seconds": 3960.6} {"step": 777, "loss": 0.48180994391441345, "data": 0.21103575825691223, "gap": 0.2707741856575012, "sigma": 0.7374532222747803, "kind": "mosaic", "tokens": 273, "dense": 448, "grad_norm": 0.21219462156295776, "seconds": 3965.1} {"step": 778, "loss": 0.5796850323677063, "data": 0.22452908754348755, "gap": 0.35515594482421875, "sigma": 0.43916037678718567, "kind": "mosaic", "tokens": 246, "dense": 448, "grad_norm": 0.1673080027103424, "seconds": 3969.6} {"step": 779, "loss": 0.5758578181266785, "data": 0.23458613455295563, "gap": 0.34127166867256165, "sigma": 0.7529283165931702, "kind": "mosaic", "tokens": 257, "dense": 448, "grad_norm": 0.26670408248901367, "seconds": 3974.3} {"step": 780, "loss": 0.9812554717063904, "data": 0.08224757760763168, "gap": 0.8990079164505005, "sigma": 0.975770115852356, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.7315782308578491, "seconds": 3978.4} {"step": 781, "loss": 0.8330259919166565, "data": 0.19845432043075562, "gap": 0.6345716714859009, "sigma": 0.8129281401634216, "kind": "mosaic", "tokens": 168, "dense": 448, "grad_norm": 0.5347458720207214, "seconds": 3982.9} {"step": 782, "loss": 0.35170137882232666, "data": 0.1770046353340149, "gap": 0.17469674348831177, "sigma": 0.6482487916946411, "kind": "mosaic", "tokens": 222, "dense": 448, "grad_norm": 0.1430014967918396, "seconds": 3987.4} {"step": 783, "loss": 0.3709954619407654, "data": 0.3709954619407654, "gap": null, "sigma": 0.9705536961555481, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.18086281418800354, "seconds": 3990.1} {"step": 784, "loss": 0.39656493067741394, "data": 0.39656493067741394, "gap": null, "sigma": 0.9697387218475342, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.29597562551498413, "seconds": 3992.9} {"step": 785, "loss": 0.47435033321380615, "data": 0.20658543705940247, "gap": 0.2677648961544037, "sigma": 0.7193962931632996, "kind": "mosaic", "tokens": 264, "dense": 448, "grad_norm": 0.19632507860660553, "seconds": 3997.4} {"step": 786, "loss": 0.2942967116832733, "data": 0.2942967116832733, "gap": null, "sigma": 0.9267498850822449, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.18875105679035187, "seconds": 4000.3} {"step": 787, "loss": 0.9243342876434326, "data": 0.2397843301296234, "gap": 0.6845499873161316, "sigma": 0.8341066241264343, "kind": "mosaic", "tokens": 200, "dense": 448, "grad_norm": 0.4325745403766632, "seconds": 4004.8} {"step": 788, "loss": 0.36346903443336487, "data": 0.20849494636058807, "gap": 0.1549740880727768, "sigma": 0.9504423141479492, "kind": "mosaic", "tokens": 253, "dense": 448, "grad_norm": 0.1667724996805191, "seconds": 4009.3} {"step": 789, "loss": 0.5395041704177856, "data": 0.20585207641124725, "gap": 0.3336521089076996, "sigma": 0.6406974196434021, "kind": "mosaic", "tokens": 184, "dense": 448, "grad_norm": 0.6703975200653076, "seconds": 4013.7} {"step": 790, "loss": 0.3750778138637543, "data": 0.3750778138637543, "gap": null, "sigma": 0.9196171760559082, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2611802816390991, "seconds": 4016.5} {"step": 791, "loss": 0.3940132260322571, "data": 0.18749332427978516, "gap": 0.20651988685131073, "sigma": 0.9432879686355591, "kind": "mosaic", "tokens": 249, "dense": 448, "grad_norm": 0.24498894810676575, "seconds": 4021.1} {"step": 792, "loss": 0.7372498512268066, "data": 0.22452640533447266, "gap": 0.512723445892334, "sigma": 0.9679096937179565, "kind": "mosaic", "tokens": 240, "dense": 448, "grad_norm": 0.5793561935424805, "seconds": 4025.7} {"step": 793, "loss": 0.7399216890335083, "data": 0.36906546354293823, "gap": 0.37085622549057007, "sigma": 0.9536997079849243, "kind": "mosaic", "tokens": 214, "dense": 448, "grad_norm": 0.46734708547592163, "seconds": 4030.2} {"step": 794, "loss": 0.5067400932312012, "data": 0.5067400932312012, "gap": null, "sigma": 0.9794433116912842, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.29219362139701843, "seconds": 4033.1} {"step": 795, "loss": 0.4739565849304199, "data": 0.1738358587026596, "gap": 0.3001207411289215, "sigma": 0.8705429434776306, "kind": "mosaic", "tokens": 185, "dense": 448, "grad_norm": 0.3203490078449249, "seconds": 4037.5} {"step": 796, "loss": 0.9268105030059814, "data": 0.23041298985481262, "gap": 0.6963975429534912, "sigma": 0.9348006248474121, "kind": "mosaic", "tokens": 167, "dense": 448, "grad_norm": 0.4429875314235687, "seconds": 4041.9} {"step": 797, "loss": 0.46093499660491943, "data": 0.2088678479194641, "gap": 0.2520671486854553, "sigma": 0.6805504560470581, "kind": "mosaic", "tokens": 236, "dense": 448, "grad_norm": 0.15592031180858612, "seconds": 4046.4} {"step": 798, "loss": 0.6086567640304565, "data": 0.11093809455633163, "gap": 0.4977186918258667, "sigma": 0.6387742161750793, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.28103265166282654, "seconds": 4050.4} {"step": 799, "loss": 0.16912154853343964, "data": 0.16912154853343964, "gap": null, "sigma": 0.7220737338066101, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.09112287312746048, "seconds": 4053.2} {"step": 800, "loss": 0.6469006538391113, "data": 0.6469006538391113, "gap": null, "sigma": 0.967792272567749, "kind": "dense", "tokens": 512, "dense": 512, "grad_norm": 0.48123109340667725, "seconds": 4056.2} {"step": 801, "loss": 0.6018365621566772, "data": 0.24333728849887848, "gap": 0.3584992587566376, "sigma": 0.9314521551132202, "kind": "mosaic", "tokens": 214, "dense": 448, "grad_norm": 0.25332850217819214, "seconds": 4060.7} {"step": 802, "loss": 0.5370608568191528, "data": 0.23061755299568176, "gap": 0.30644330382347107, "sigma": 0.7366891503334045, "kind": "mosaic", "tokens": 213, "dense": 448, "grad_norm": 0.22797082364559174, "seconds": 4065.1} {"step": 803, "loss": 0.4651367664337158, "data": 0.25008097290992737, "gap": 0.21505579352378845, "sigma": 0.8273669481277466, "kind": "mosaic", "tokens": 244, "dense": 448, "grad_norm": 0.3089946508407593, "seconds": 4069.7} {"step": 804, "loss": 0.4159529209136963, "data": 0.4159529209136963, "gap": null, "sigma": 0.9957619905471802, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.5882324576377869, "seconds": 4072.6} {"step": 805, "loss": 18.748640060424805, "data": 1.5086103677749634, "gap": 17.24003028869629, "sigma": 0.08817686140537262, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 13.292985916137695, "seconds": 4076.6} {"step": 806, "loss": 1.1022213697433472, "data": 0.15114225447177887, "gap": 0.9510790705680847, "sigma": 0.9831520318984985, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.926260232925415, "seconds": 4080.8} {"step": 807, "loss": 0.32255858182907104, "data": 0.22271570563316345, "gap": 0.0998428612947464, "sigma": 0.9268670082092285, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.24364972114562988, "seconds": 4085.4} {"step": 808, "loss": 0.49195340275764465, "data": 0.21769464015960693, "gap": 0.2742587625980377, "sigma": 0.5619732141494751, "kind": "uniform", "tokens": 256, "dense": 512, "grad_norm": 0.2882483899593353, "seconds": 4089.8} {"step": 809, "loss": 0.4997834265232086, "data": 0.22917765378952026, "gap": 0.27060577273368835, "sigma": 0.9390066862106323, "kind": "mosaic", "tokens": 186, "dense": 448, "grad_norm": 0.35549992322921753, "seconds": 4094.4} {"step": 810, "loss": 0.2033139169216156, "data": 0.2033139169216156, "gap": null, "sigma": 0.6420786380767822, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.10612118989229202, "seconds": 4097.2} {"step": 811, "loss": 0.5887365937232971, "data": 0.05575883015990257, "gap": 0.5329777598381042, "sigma": 0.6941087245941162, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 0.3177620768547058, "seconds": 4101.4} {"step": 812, "loss": 0.5696932077407837, "data": 0.21180352568626404, "gap": 0.35788968205451965, "sigma": 0.7735101580619812, "kind": "mosaic", "tokens": 180, "dense": 448, "grad_norm": 0.491940438747406, "seconds": 4105.8} {"step": 813, "loss": 0.6287624835968018, "data": 0.23492318391799927, "gap": 0.3938392996788025, "sigma": 0.9725367426872253, "kind": "mosaic", "tokens": 206, "dense": 448, "grad_norm": 0.5302606821060181, "seconds": 4110.2} {"step": 814, "loss": 0.4075065851211548, "data": 0.16603755950927734, "gap": 0.24146901071071625, "sigma": 0.5563505291938782, "kind": "mosaic", "tokens": 260, "dense": 448, "grad_norm": 0.1706247478723526, "seconds": 4114.9} {"step": 815, "loss": 0.6701952219009399, "data": 0.21201738715171814, "gap": 0.4581778049468994, "sigma": 0.7225020527839661, "kind": "mosaic", "tokens": 179, "dense": 448, "grad_norm": 0.3026641309261322, "seconds": 4119.3} {"step": 816, "loss": 0.3454676568508148, "data": 0.3454676568508148, "gap": null, "sigma": 0.8662192225456238, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.11382506042718887, "seconds": 4122.2} {"step": 817, "loss": 0.4715646207332611, "data": 0.4715646207332611, "gap": null, "sigma": 0.937330424785614, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.22090058028697968, "seconds": 4125.0} {"step": 818, "loss": 0.5629855990409851, "data": 0.1999187022447586, "gap": 0.3630669116973877, "sigma": 0.7177385091781616, "kind": "mosaic", "tokens": 186, "dense": 448, "grad_norm": 0.2212521731853485, "seconds": 4129.5} {"step": 819, "loss": 0.27925604581832886, "data": 0.27925604581832886, "gap": null, "sigma": 0.7593469023704529, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.15719451010227203, "seconds": 4132.2} {"step": 820, "loss": 0.6942301988601685, "data": 0.2615242600440979, "gap": 0.43270596861839294, "sigma": 0.9774985909461975, "kind": "mosaic", "tokens": 177, "dense": 448, "grad_norm": 0.5284779071807861, "seconds": 4136.8} {"step": 821, "loss": 0.6017171144485474, "data": 0.22859272360801697, "gap": 0.373124361038208, "sigma": 0.5241003036499023, "kind": "mosaic", "tokens": 278, "dense": 448, "grad_norm": 0.1976230889558792, "seconds": 4141.3} {"step": 822, "loss": 0.924832820892334, "data": 0.4021691083908081, "gap": 0.5226637125015259, "sigma": 0.3301829397678375, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.42361658811569214, "seconds": 4145.8} {"step": 823, "loss": 0.40910804271698, "data": 0.20383413136005402, "gap": 0.20527391135692596, "sigma": 0.8777901530265808, "kind": "mosaic", "tokens": 221, "dense": 448, "grad_norm": 0.2511211335659027, "seconds": 4150.5} {"step": 824, "loss": 0.5828503370285034, "data": 0.2547001540660858, "gap": 0.3281501829624176, "sigma": 0.6354565620422363, "kind": "mosaic", "tokens": 233, "dense": 448, "grad_norm": 0.20190949738025665, "seconds": 4154.9} {"step": 825, "loss": 11.675667762756348, "data": 2.2668378353118896, "gap": 9.408829689025879, "sigma": 0.0041807121597230434, "kind": "mosaic", "tokens": 238, "dense": 448, "grad_norm": 11.277350425720215, "seconds": 4159.6} {"step": 826, "loss": 0.4819974899291992, "data": 0.09296410530805588, "gap": 0.38903337717056274, "sigma": 0.7661120891571045, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.26128679513931274, "seconds": 4163.7} {"step": 827, "loss": 0.4446813464164734, "data": 0.20294098556041718, "gap": 0.2417403757572174, "sigma": 0.6036339402198792, "kind": "mosaic", "tokens": 225, "dense": 448, "grad_norm": 0.11859236657619476, "seconds": 4168.0} {"step": 828, "loss": 0.21185927093029022, "data": 0.21185927093029022, "gap": null, "sigma": 0.46568408608436584, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.05098184943199158, "seconds": 4170.8} {"step": 829, "loss": 0.6749594211578369, "data": 0.2808263897895813, "gap": 0.3941330015659332, "sigma": 0.7174521088600159, "kind": "mosaic", "tokens": 234, "dense": 448, "grad_norm": 0.3392501771450043, "seconds": 4175.5} {"step": 830, "loss": 0.536220371723175, "data": 0.19072823226451874, "gap": 0.3454921245574951, "sigma": 0.8913125991821289, "kind": "mosaic", "tokens": 193, "dense": 448, "grad_norm": 0.4216088652610779, "seconds": 4180.1} {"step": 831, "loss": 0.7465187311172485, "data": 0.26137223839759827, "gap": 0.48514649271965027, "sigma": 0.48620983958244324, "kind": "mosaic", "tokens": 245, "dense": 448, "grad_norm": 0.14772437512874603, "seconds": 4184.3} {"step": 832, "loss": 0.8024895191192627, "data": 0.2413993775844574, "gap": 0.5610901117324829, "sigma": 0.32310211658477783, "kind": "mosaic", "tokens": 163, "dense": 448, "grad_norm": 0.27200400829315186, "seconds": 4188.7} {"step": 833, "loss": 0.5357288122177124, "data": 0.23371519148349762, "gap": 0.3020136058330536, "sigma": 0.8436406254768372, "kind": "mosaic", "tokens": 250, "dense": 448, "grad_norm": 0.3905397057533264, "seconds": 4193.1} {"step": 834, "loss": 0.44599032402038574, "data": 0.22818025946617126, "gap": 0.21781006455421448, "sigma": 0.7847570776939392, "kind": "mosaic", "tokens": 238, "dense": 448, "grad_norm": 0.2217138707637787, "seconds": 4197.6} {"step": 835, "loss": 0.7240354418754578, "data": 0.14236658811569214, "gap": 0.5816688537597656, "sigma": 0.8359743356704712, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.41309839487075806, "seconds": 4201.8} {"step": 836, "loss": 0.33104461431503296, "data": 0.33104461431503296, "gap": null, "sigma": 0.3818751275539398, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.041846178472042084, "seconds": 4204.6} {"step": 837, "loss": 0.384621262550354, "data": 0.384621262550354, "gap": null, "sigma": 0.9552162289619446, "kind": "dense", "tokens": 480, "dense": 480, "grad_norm": 0.2013457715511322, "seconds": 4207.3} {"step": 838, "loss": 0.4203576445579529, "data": 0.19891811907291412, "gap": 0.22143951058387756, "sigma": 0.52955162525177, "kind": "mosaic", "tokens": 250, "dense": 448, "grad_norm": 0.14743472635746002, "seconds": 4211.9} {"step": 839, "loss": 0.6932604312896729, "data": 0.21597345173358917, "gap": 0.4772869944572449, "sigma": 0.6393191814422607, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.19537825882434845, "seconds": 4216.1} {"step": 840, "loss": 0.2560671865940094, "data": 0.2560671865940094, "gap": null, "sigma": 0.27019956707954407, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.09497058391571045, "seconds": 4218.9} {"step": 841, "loss": 2.1524155139923096, "data": 0.5314754843711853, "gap": 1.620940089225769, "sigma": 0.18476878106594086, "kind": "mosaic", "tokens": 206, "dense": 448, "grad_norm": 0.877163827419281, "seconds": 4223.5} {"step": 842, "loss": 0.4139013886451721, "data": 0.4139013886451721, "gap": null, "sigma": 0.8919349312782288, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.321499764919281, "seconds": 4226.3} {"step": 843, "loss": 0.3150131106376648, "data": 0.3150131106376648, "gap": null, "sigma": 0.9789379835128784, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3205969035625458, "seconds": 4229.1} {"step": 844, "loss": 0.6879628896713257, "data": 0.21240562200546265, "gap": 0.47555726766586304, "sigma": 0.809234619140625, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.3187706470489502, "seconds": 4233.5} {"step": 845, "loss": 0.5403379797935486, "data": 0.29313334822654724, "gap": 0.24720463156700134, "sigma": 0.7644014954566956, "kind": "mosaic", "tokens": 273, "dense": 448, "grad_norm": 0.20612114667892456, "seconds": 4238.1} {"step": 846, "loss": 0.24888770282268524, "data": 0.24888770282268524, "gap": null, "sigma": 0.4547978639602661, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.4633076786994934, "seconds": 4240.9} {"step": 847, "loss": 0.4713108539581299, "data": 0.16059093177318573, "gap": 0.31071993708610535, "sigma": 0.897446870803833, "kind": "mosaic", "tokens": 210, "dense": 448, "grad_norm": 0.2411968857049942, "seconds": 4245.4} {"step": 848, "loss": 0.7707529067993164, "data": 0.36061665415763855, "gap": 0.41013628244400024, "sigma": 0.9866827130317688, "kind": "mosaic", "tokens": 274, "dense": 448, "grad_norm": 0.31832030415534973, "seconds": 4249.9} {"step": 849, "loss": 0.5306548476219177, "data": 0.20837025344371796, "gap": 0.3222845792770386, "sigma": 0.7946681380271912, "kind": "mosaic", "tokens": 186, "dense": 448, "grad_norm": 0.2724551260471344, "seconds": 4254.2} {"step": 850, "loss": 0.5975220203399658, "data": 0.25136467814445496, "gap": 0.34615737199783325, "sigma": 0.6846866011619568, "kind": "mosaic", "tokens": 209, "dense": 448, "grad_norm": 0.20270924270153046, "seconds": 4258.7} {"step": 851, "loss": 0.7377545833587646, "data": 0.7377545833587646, "gap": null, "sigma": 0.05108585208654404, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.17536693811416626, "seconds": 4261.5} {"step": 852, "loss": 5.690752983093262, "data": 1.281458854675293, "gap": 4.409294128417969, "sigma": 0.09241458028554916, "kind": "mosaic", "tokens": 234, "dense": 448, "grad_norm": 4.052290916442871, "seconds": 4265.9} {"step": 853, "loss": 1.2653586864471436, "data": 0.3978751599788666, "gap": 0.8674835562705994, "sigma": 0.22424420714378357, "kind": "mosaic", "tokens": 170, "dense": 448, "grad_norm": 0.45906317234039307, "seconds": 4270.4} {"step": 854, "loss": 0.644829273223877, "data": 0.2481548935174942, "gap": 0.39667436480522156, "sigma": 0.6311622262001038, "kind": "mosaic", "tokens": 188, "dense": 448, "grad_norm": 0.20083735883235931, "seconds": 4274.7} {"step": 855, "loss": 0.6052733063697815, "data": 0.2310466766357422, "gap": 0.3742266297340393, "sigma": 0.937882661819458, "kind": "mosaic", "tokens": 207, "dense": 448, "grad_norm": 0.44451192021369934, "seconds": 4279.2} {"step": 856, "loss": 0.5096300840377808, "data": 0.2139144241809845, "gap": 0.29571568965911865, "sigma": 0.9644410014152527, "kind": "mosaic", "tokens": 208, "dense": 448, "grad_norm": 0.4254402816295624, "seconds": 4283.7} {"step": 857, "loss": 0.783126711845398, "data": 0.20385225117206573, "gap": 0.5792744755744934, "sigma": 0.39846259355545044, "kind": "mosaic", "tokens": 169, "dense": 448, "grad_norm": 0.23435893654823303, "seconds": 4288.1} {"step": 858, "loss": 1.3452284336090088, "data": 0.3957778811454773, "gap": 0.9494506120681763, "sigma": 0.267468124628067, "kind": "mosaic", "tokens": 208, "dense": 512, "grad_norm": 0.4067380130290985, "seconds": 4292.6} {"step": 859, "loss": 0.41081666946411133, "data": 0.14491237699985504, "gap": 0.2659042775630951, "sigma": 0.8771469593048096, "kind": "mosaic", "tokens": 198, "dense": 448, "grad_norm": 0.30374908447265625, "seconds": 4297.2} {"step": 860, "loss": 0.8218495845794678, "data": 0.1933879405260086, "gap": 0.6284616589546204, "sigma": 0.8172085881233215, "kind": "mosaic", "tokens": 220, "dense": 448, "grad_norm": 0.7929482460021973, "seconds": 4301.7} {"step": 861, "loss": 0.5852071046829224, "data": 0.17464229464530945, "gap": 0.4105647802352905, "sigma": 0.5419301986694336, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.22791291773319244, "seconds": 4305.9} {"step": 862, "loss": 0.38324692845344543, "data": 0.38324692845344543, "gap": null, "sigma": 0.8743548393249512, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.11268367618322372, "seconds": 4308.7} {"step": 863, "loss": 0.6879309415817261, "data": 0.24298030138015747, "gap": 0.444950670003891, "sigma": 0.30513760447502136, "kind": "mosaic", "tokens": 219, "dense": 448, "grad_norm": 0.24950067698955536, "seconds": 4313.0} {"step": 864, "loss": 0.5661649107933044, "data": 0.24176813662052155, "gap": 0.3243967592716217, "sigma": 0.8944974541664124, "kind": "mosaic", "tokens": 205, "dense": 448, "grad_norm": 0.24863585829734802, "seconds": 4317.3} {"step": 865, "loss": 0.5035446286201477, "data": 0.11743895709514618, "gap": 0.3861056864261627, "sigma": 0.699860155582428, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.23157647252082825, "seconds": 4321.5} {"step": 866, "loss": 0.43851059675216675, "data": 0.43851059675216675, "gap": null, "sigma": 0.8831008076667786, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3444897532463074, "seconds": 4324.3} {"step": 867, "loss": 0.4283161759376526, "data": 0.4283161759376526, "gap": null, "sigma": 0.9379317760467529, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.18443402647972107, "seconds": 4327.1} {"step": 868, "loss": 0.31746527552604675, "data": 0.1280519813299179, "gap": 0.18941329419612885, "sigma": 0.6747951507568359, "kind": "mosaic", "tokens": 206, "dense": 448, "grad_norm": 0.14221639931201935, "seconds": 4331.6} {"step": 869, "loss": 0.21952545642852783, "data": 0.09539175033569336, "gap": 0.12413370609283447, "sigma": 0.7036177515983582, "kind": "mosaic", "tokens": 186, "dense": 448, "grad_norm": 0.2659405469894409, "seconds": 4335.8} {"step": 870, "loss": 16.760177612304688, "data": 2.337632179260254, "gap": 14.422544479370117, "sigma": 0.06292714178562164, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 11.371613502502441, "seconds": 4340.0} {"step": 871, "loss": 0.48319536447525024, "data": 0.20343150198459625, "gap": 0.2797638773918152, "sigma": 0.684837281703949, "kind": "mosaic", "tokens": 169, "dense": 448, "grad_norm": 0.13512757420539856, "seconds": 4344.4} {"step": 872, "loss": 0.7422710657119751, "data": 0.2769504189491272, "gap": 0.4653206765651703, "sigma": 0.4958283305168152, "kind": "mosaic", "tokens": 226, "dense": 448, "grad_norm": 0.21769511699676514, "seconds": 4349.0} {"step": 873, "loss": 0.5387780070304871, "data": 0.20437444746494293, "gap": 0.3344035744667053, "sigma": 0.7822176814079285, "kind": "mosaic", "tokens": 211, "dense": 448, "grad_norm": 0.9617516398429871, "seconds": 4353.3} {"step": 874, "loss": 0.2767862379550934, "data": 0.16112104058265686, "gap": 0.11566518992185593, "sigma": 0.7070191502571106, "kind": "mosaic", "tokens": 243, "dense": 448, "grad_norm": 0.13885027170181274, "seconds": 4357.8} {"step": 875, "loss": 0.9021367430686951, "data": 0.23673196136951447, "gap": 0.6654047966003418, "sigma": 0.9879230856895447, "kind": "mosaic", "tokens": 223, "dense": 448, "grad_norm": 0.49167612195014954, "seconds": 4362.2} {"step": 876, "loss": 0.85472571849823, "data": 0.1610512137413025, "gap": 0.6936745047569275, "sigma": 0.9108087420463562, "kind": "mosaic", "tokens": 177, "dense": 448, "grad_norm": 1.8441498279571533, "seconds": 4366.7} {"step": 877, "loss": 0.6384915113449097, "data": 0.19567212462425232, "gap": 0.44281935691833496, "sigma": 0.38078242540359497, "kind": "mosaic", "tokens": 153, "dense": 448, "grad_norm": 0.21761056780815125, "seconds": 4371.0} {"step": 878, "loss": 0.5132217407226562, "data": 0.20012567937374115, "gap": 0.3130960464477539, "sigma": 0.8812643885612488, "kind": "mosaic", "tokens": 244, "dense": 448, "grad_norm": 0.30762946605682373, "seconds": 4375.5} {"step": 879, "loss": 0.4357546865940094, "data": 0.4357546865940094, "gap": null, "sigma": 0.9952211976051331, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.33655568957328796, "seconds": 4378.2} {"step": 880, "loss": 0.5698719024658203, "data": 0.22456026077270508, "gap": 0.34531161189079285, "sigma": 0.8412088751792908, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 1.2831041812896729, "seconds": 4382.7} {"step": 881, "loss": 0.1370449960231781, "data": 0.1370449960231781, "gap": null, "sigma": 0.8570893406867981, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.19948695600032806, "seconds": 4385.5} {"step": 882, "loss": 0.2468668818473816, "data": 0.2468668818473816, "gap": null, "sigma": 0.6783685088157654, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.12724827229976654, "seconds": 4388.3} {"step": 883, "loss": 0.645697832107544, "data": 0.19627751410007477, "gap": 0.44942033290863037, "sigma": 0.8757330179214478, "kind": "mosaic", "tokens": 209, "dense": 448, "grad_norm": 0.597321093082428, "seconds": 4392.8} {"step": 884, "loss": 0.7061998844146729, "data": 0.2670883536338806, "gap": 0.43911150097846985, "sigma": 0.9581974148750305, "kind": "mosaic", "tokens": 209, "dense": 448, "grad_norm": 0.3665255904197693, "seconds": 4397.4} {"step": 885, "loss": 0.5795241594314575, "data": 0.25531038641929626, "gap": 0.32421380281448364, "sigma": 0.625719428062439, "kind": "mosaic", "tokens": 247, "dense": 448, "grad_norm": 0.15491606295108795, "seconds": 4402.0} {"step": 886, "loss": 0.46501845121383667, "data": 0.14254046976566315, "gap": 0.3224779963493347, "sigma": 0.9418220520019531, "kind": "mosaic", "tokens": 227, "dense": 448, "grad_norm": 0.26580825448036194, "seconds": 4406.4} {"step": 887, "loss": 0.6541826725006104, "data": 0.28354182839393616, "gap": 0.3706408143043518, "sigma": 0.8062604069709778, "kind": "mosaic", "tokens": 250, "dense": 448, "grad_norm": 2.5305254459381104, "seconds": 4411.0} {"step": 888, "loss": 0.7172226905822754, "data": 0.1984274834394455, "gap": 0.5187951922416687, "sigma": 0.998376190662384, "kind": "mosaic", "tokens": 214, "dense": 448, "grad_norm": 0.39112356305122375, "seconds": 4415.4} {"step": 889, "loss": 0.5058653950691223, "data": 0.2058887928724289, "gap": 0.2999765872955322, "sigma": 0.36887142062187195, "kind": "mosaic", "tokens": 265, "dense": 448, "grad_norm": 0.1471603810787201, "seconds": 4419.9} {"step": 890, "loss": 1.433901071548462, "data": 0.36831265687942505, "gap": 1.0655884742736816, "sigma": 0.9894871115684509, "kind": "mosaic", "tokens": 161, "dense": 448, "grad_norm": 0.8974464535713196, "seconds": 4424.0} {"step": 891, "loss": 0.43201500177383423, "data": 0.04052826389670372, "gap": 0.3914867341518402, "sigma": 0.6435359120368958, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 0.28703317046165466, "seconds": 4427.9} {"step": 892, "loss": 0.662798285484314, "data": 0.2362857609987259, "gap": 0.4265125095844269, "sigma": 0.4377850592136383, "kind": "mosaic", "tokens": 226, "dense": 448, "grad_norm": 0.21468068659305573, "seconds": 4432.2} {"step": 893, "loss": 0.6901369690895081, "data": 0.2186281383037567, "gap": 0.47150883078575134, "sigma": 0.43793460726737976, "kind": "mosaic", "tokens": 242, "dense": 448, "grad_norm": 0.16775821149349213, "seconds": 4436.6} {"step": 894, "loss": 0.4167778789997101, "data": 0.4167778789997101, "gap": null, "sigma": 0.9793217778205872, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.26596128940582275, "seconds": 4439.5} {"step": 895, "loss": 0.3899003863334656, "data": 0.3899003863334656, "gap": null, "sigma": 0.9555850625038147, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.23727428913116455, "seconds": 4442.3} {"step": 896, "loss": 1.6280627250671387, "data": 0.5073208808898926, "gap": 1.120741844177246, "sigma": 0.2921959161758423, "kind": "mosaic", "tokens": 181, "dense": 448, "grad_norm": 0.4312632083892822, "seconds": 4446.7} {"step": 897, "loss": 0.5593488216400146, "data": 0.2242072969675064, "gap": 0.33514153957366943, "sigma": 0.4674382209777832, "kind": "mosaic", "tokens": 270, "dense": 448, "grad_norm": 0.1722329556941986, "seconds": 4451.3} {"step": 898, "loss": 0.6158201098442078, "data": 0.19175304472446442, "gap": 0.42406708002090454, "sigma": 0.32919254899024963, "kind": "mosaic", "tokens": 186, "dense": 448, "grad_norm": 0.265837162733078, "seconds": 4455.8} {"step": 899, "loss": 0.30270785093307495, "data": 0.30270785093307495, "gap": null, "sigma": 0.690287709236145, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.07728047668933868, "seconds": 4458.6} {"step": 900, "loss": 0.6395718455314636, "data": 0.17147302627563477, "gap": 0.46809881925582886, "sigma": 0.820640504360199, "kind": "mosaic", "tokens": 180, "dense": 448, "grad_norm": 0.21441510319709778, "seconds": 4463.1} {"step": 901, "loss": 0.43941667675971985, "data": 0.1546555459499359, "gap": 0.28476113080978394, "sigma": 0.692371666431427, "kind": "mosaic", "tokens": 221, "dense": 448, "grad_norm": 0.18130835890769958, "seconds": 4467.5} {"step": 902, "loss": 0.5307878255844116, "data": 0.1909867525100708, "gap": 0.33980104327201843, "sigma": 0.8753413558006287, "kind": "mosaic", "tokens": 209, "dense": 448, "grad_norm": 0.7827073335647583, "seconds": 4472.1} {"step": 903, "loss": 0.8141303062438965, "data": 0.206970676779747, "gap": 0.6071596145629883, "sigma": 0.9148131608963013, "kind": "mosaic", "tokens": 214, "dense": 448, "grad_norm": 0.7552091479301453, "seconds": 4476.7} {"step": 904, "loss": 0.5703315138816833, "data": 0.23454700410366058, "gap": 0.3357844948768616, "sigma": 0.8891435265541077, "kind": "mosaic", "tokens": 278, "dense": 448, "grad_norm": 0.24941104650497437, "seconds": 4481.3} {"step": 905, "loss": 0.5631601810455322, "data": 0.2488320767879486, "gap": 0.31432807445526123, "sigma": 0.6263977885246277, "kind": "mosaic", "tokens": 230, "dense": 448, "grad_norm": 0.26100772619247437, "seconds": 4485.9} {"step": 906, "loss": 0.7994927167892456, "data": 0.28907325863838196, "gap": 0.5104194283485413, "sigma": 0.4182201027870178, "kind": "mosaic", "tokens": 205, "dense": 448, "grad_norm": 0.24447546899318695, "seconds": 4490.5} {"step": 907, "loss": 0.5612282156944275, "data": 0.23701320588588715, "gap": 0.32421499490737915, "sigma": 0.6728585362434387, "kind": "mosaic", "tokens": 207, "dense": 448, "grad_norm": 0.18740113079547882, "seconds": 4495.0} {"step": 908, "loss": 0.3984089493751526, "data": 0.1883065402507782, "gap": 0.2101023942232132, "sigma": 0.7701070308685303, "kind": "mosaic", "tokens": 241, "dense": 448, "grad_norm": 0.36138325929641724, "seconds": 4499.6} {"step": 909, "loss": 1.609906554222107, "data": 0.4364798963069916, "gap": 1.173426628112793, "sigma": 0.36228039860725403, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.8244976997375488, "seconds": 4503.8} {"step": 910, "loss": 0.8526519536972046, "data": 0.39719483256340027, "gap": 0.4554571509361267, "sigma": 0.3666902482509613, "kind": "mosaic", "tokens": 238, "dense": 448, "grad_norm": 0.19555562734603882, "seconds": 4508.3} {"step": 911, "loss": 1.950699806213379, "data": 0.3020196259021759, "gap": 1.6486802101135254, "sigma": 0.9732034802436829, "kind": "mosaic", "tokens": 165, "dense": 448, "grad_norm": 1.0743300914764404, "seconds": 4512.6} {"step": 912, "loss": 0.43209266662597656, "data": 0.19627493619918823, "gap": 0.23581774532794952, "sigma": 0.597602128982544, "kind": "mosaic", "tokens": 221, "dense": 480, "grad_norm": 0.10856279730796814, "seconds": 4517.1} {"step": 913, "loss": 0.5826010704040527, "data": 0.21089601516723633, "gap": 0.3717050850391388, "sigma": 0.7203344106674194, "kind": "mosaic", "tokens": 185, "dense": 448, "grad_norm": 0.20069319009780884, "seconds": 4521.5} {"step": 914, "loss": 0.9473181366920471, "data": 0.18228565156459808, "gap": 0.7650324702262878, "sigma": 0.9049611687660217, "kind": "mosaic", "tokens": 214, "dense": 448, "grad_norm": 0.8056186437606812, "seconds": 4526.0} {"step": 915, "loss": 1.0314539670944214, "data": 0.22468163073062897, "gap": 0.8067722916603088, "sigma": 0.9243664741516113, "kind": "mosaic", "tokens": 173, "dense": 448, "grad_norm": 0.6116030812263489, "seconds": 4530.3} {"step": 916, "loss": 0.24366211891174316, "data": 0.24366211891174316, "gap": null, "sigma": 0.7871280908584595, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.40943825244903564, "seconds": 4533.1} {"step": 917, "loss": 0.4189526438713074, "data": 0.1713370382785797, "gap": 0.24761560559272766, "sigma": 0.6673973798751831, "kind": "mosaic", "tokens": 277, "dense": 448, "grad_norm": 1.4157894849777222, "seconds": 4537.5} {"step": 918, "loss": 1.3099415302276611, "data": 0.3483274579048157, "gap": 0.9616140127182007, "sigma": 0.9938333630561829, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.9987826943397522, "seconds": 4541.8} {"step": 919, "loss": 0.8125879764556885, "data": 0.13172194361686707, "gap": 0.6808660626411438, "sigma": 0.42910462617874146, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.5875247716903687, "seconds": 4546.0} {"step": 920, "loss": 0.4227215349674225, "data": 0.23967847228050232, "gap": 0.18304306268692017, "sigma": 0.9314998388290405, "kind": "mosaic", "tokens": 255, "dense": 448, "grad_norm": 0.39391109347343445, "seconds": 4550.4} {"step": 921, "loss": 0.5523589849472046, "data": 0.2928445637226105, "gap": 0.2595144212245941, "sigma": 0.8774778842926025, "kind": "mosaic", "tokens": 208, "dense": 448, "grad_norm": 0.24092993140220642, "seconds": 4554.9} {"step": 922, "loss": 0.5135234594345093, "data": 0.5135234594345093, "gap": null, "sigma": 0.963417112827301, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.26267576217651367, "seconds": 4557.7} {"step": 923, "loss": 1.11664617061615, "data": 0.3526313602924347, "gap": 0.7640147805213928, "sigma": 0.9563412666320801, "kind": "mosaic", "tokens": 221, "dense": 448, "grad_norm": 0.5024541616439819, "seconds": 4562.3} {"step": 924, "loss": 0.9118128418922424, "data": 0.26531994342803955, "gap": 0.6464928984642029, "sigma": 0.8843939304351807, "kind": "mosaic", "tokens": 251, "dense": 448, "grad_norm": 0.5648916363716125, "seconds": 4566.8} {"step": 925, "loss": 0.6736491918563843, "data": 0.1795172095298767, "gap": 0.4941319525241852, "sigma": 0.9529484510421753, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.37135180830955505, "seconds": 4571.3} {"step": 926, "loss": 0.9581338167190552, "data": 0.27221187949180603, "gap": 0.6859219074249268, "sigma": 0.39264532923698425, "kind": "mosaic", "tokens": 167, "dense": 448, "grad_norm": 4.612842082977295, "seconds": 4575.6} {"step": 927, "loss": 0.6343080997467041, "data": 0.25620025396347046, "gap": 0.37810781598091125, "sigma": 0.8282588124275208, "kind": "mosaic", "tokens": 226, "dense": 448, "grad_norm": 0.4088355004787445, "seconds": 4580.1} {"step": 928, "loss": 0.35924017429351807, "data": 0.35924017429351807, "gap": null, "sigma": 0.7877016663551331, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.1499939113855362, "seconds": 4582.9} {"step": 929, "loss": 0.5491909980773926, "data": 0.2422756552696228, "gap": 0.30691537261009216, "sigma": 0.9366806149482727, "kind": "mosaic", "tokens": 276, "dense": 448, "grad_norm": 0.42356932163238525, "seconds": 4587.4} {"step": 930, "loss": 0.6298969388008118, "data": 0.20178602635860443, "gap": 0.42811089754104614, "sigma": 0.6701515316963196, "kind": "mosaic", "tokens": 156, "dense": 448, "grad_norm": 0.1803509145975113, "seconds": 4591.7} {"step": 931, "loss": 0.21895559132099152, "data": 0.21895559132099152, "gap": null, "sigma": 0.8289626836776733, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.21133878827095032, "seconds": 4594.5} {"step": 932, "loss": 0.4645935297012329, "data": 0.4645935297012329, "gap": null, "sigma": 0.9968076944351196, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.41300642490386963, "seconds": 4597.2} {"step": 933, "loss": 0.9740005731582642, "data": 0.27231112122535706, "gap": 0.7016894817352295, "sigma": 0.9973908066749573, "kind": "mosaic", "tokens": 228, "dense": 448, "grad_norm": 0.7542763352394104, "seconds": 4601.7} {"step": 934, "loss": 1.067112922668457, "data": 0.33132103085517883, "gap": 0.7357918620109558, "sigma": 0.3531397879123688, "kind": "mosaic", "tokens": 228, "dense": 448, "grad_norm": 0.2575750946998596, "seconds": 4606.4} {"step": 935, "loss": 0.5077934265136719, "data": 0.19663043320178986, "gap": 0.3111629784107208, "sigma": 0.8538703322410583, "kind": "mosaic", "tokens": 249, "dense": 448, "grad_norm": 0.5982997417449951, "seconds": 4610.9} {"step": 936, "loss": 12.368341445922852, "data": 2.316833734512329, "gap": 10.051507949829102, "sigma": 0.014877088367938995, "kind": "mosaic", "tokens": 233, "dense": 448, "grad_norm": 9.617398262023926, "seconds": 4615.3} {"step": 937, "loss": 0.6098065376281738, "data": 0.24378368258476257, "gap": 0.36602288484573364, "sigma": 0.9788767099380493, "kind": "mosaic", "tokens": 267, "dense": 448, "grad_norm": 0.4898965358734131, "seconds": 4619.8} {"step": 938, "loss": 0.5218601822853088, "data": 0.5218601822853088, "gap": null, "sigma": 0.989542543888092, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.5254406929016113, "seconds": 4622.7} {"step": 939, "loss": 0.39430272579193115, "data": 0.39430272579193115, "gap": null, "sigma": 0.8593692779541016, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3071850836277008, "seconds": 4625.4} {"step": 940, "loss": 6.5290141105651855, "data": 1.5528775453567505, "gap": 4.976136684417725, "sigma": 0.07329759001731873, "kind": "mosaic", "tokens": 206, "dense": 448, "grad_norm": 4.964498519897461, "seconds": 4629.8} {"step": 941, "loss": 0.37513643503189087, "data": 0.21550840139389038, "gap": 0.15962804853916168, "sigma": 0.8946330547332764, "kind": "mosaic", "tokens": 224, "dense": 448, "grad_norm": 0.18711644411087036, "seconds": 4634.4} {"step": 942, "loss": 1.4517120122909546, "data": 0.3819155991077423, "gap": 1.0697964429855347, "sigma": 0.9935258626937866, "kind": "mosaic", "tokens": 173, "dense": 448, "grad_norm": 0.5925320386886597, "seconds": 4638.9} {"step": 943, "loss": 0.8127514123916626, "data": 0.37212252616882324, "gap": 0.44062885642051697, "sigma": 0.9697485566139221, "kind": "mosaic", "tokens": 261, "dense": 448, "grad_norm": 0.4882544279098511, "seconds": 4643.3} {"step": 944, "loss": 0.5142344236373901, "data": 0.2522129714488983, "gap": 0.26202142238616943, "sigma": 0.9042695164680481, "kind": "mosaic", "tokens": 240, "dense": 448, "grad_norm": 0.25285622477531433, "seconds": 4647.9} {"step": 945, "loss": 0.5200191736221313, "data": 0.1694725602865219, "gap": 0.35054662823677063, "sigma": 0.843517541885376, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 1.0095739364624023, "seconds": 4652.3} {"step": 946, "loss": 0.4461120367050171, "data": 0.4461120367050171, "gap": null, "sigma": 0.9990170001983643, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.4607139825820923, "seconds": 4655.1} {"step": 947, "loss": 0.46636152267456055, "data": 0.184409037232399, "gap": 0.28195250034332275, "sigma": 0.4443865418434143, "kind": "mosaic", "tokens": 191, "dense": 448, "grad_norm": 0.3209933042526245, "seconds": 4659.8} {"step": 948, "loss": 5.669368267059326, "data": 1.181899905204773, "gap": 4.487468242645264, "sigma": 0.08173393458127975, "kind": "mosaic", "tokens": 238, "dense": 448, "grad_norm": 3.927670955657959, "seconds": 4664.4} {"step": 949, "loss": 0.9678671360015869, "data": 0.3419308066368103, "gap": 0.6259363293647766, "sigma": 0.2457047551870346, "kind": "mosaic", "tokens": 209, "dense": 448, "grad_norm": 0.2998318672180176, "seconds": 4668.8} {"step": 950, "loss": 0.4929983615875244, "data": 0.1596653312444687, "gap": 0.33333301544189453, "sigma": 0.7072388529777527, "kind": "mosaic", "tokens": 236, "dense": 448, "grad_norm": 0.1597563773393631, "seconds": 4673.4} {"step": 951, "loss": 0.6653465032577515, "data": 0.23561927676200867, "gap": 0.4297271966934204, "sigma": 0.5467924475669861, "kind": "mosaic", "tokens": 222, "dense": 448, "grad_norm": 0.1827097088098526, "seconds": 4677.8} {"step": 952, "loss": 0.5298440456390381, "data": 0.09470430016517639, "gap": 0.4351397454738617, "sigma": 0.9434079527854919, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.41629672050476074, "seconds": 4682.1} {"step": 953, "loss": 0.5001550912857056, "data": 0.20990332961082458, "gap": 0.2902517318725586, "sigma": 0.5881150960922241, "kind": "mosaic", "tokens": 259, "dense": 448, "grad_norm": 0.1417178511619568, "seconds": 4686.6} {"step": 954, "loss": 2.1188607215881348, "data": 0.19671428203582764, "gap": 1.9221465587615967, "sigma": 0.9926055073738098, "kind": "uniform", "tokens": 64, "dense": 512, "grad_norm": 2.373570442199707, "seconds": 4690.7} {"step": 955, "loss": 0.871590256690979, "data": 0.2928062081336975, "gap": 0.5787840485572815, "sigma": 0.9618791341781616, "kind": "mosaic", "tokens": 232, "dense": 448, "grad_norm": 0.4809127748012543, "seconds": 4695.4} {"step": 956, "loss": 0.6544598340988159, "data": 0.6544598340988159, "gap": null, "sigma": 0.9978505969047546, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.8250143527984619, "seconds": 4698.3} {"step": 957, "loss": 1.0911850929260254, "data": 0.18399016559123993, "gap": 0.9071949124336243, "sigma": 0.39914947748184204, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.38645312190055847, "seconds": 4702.4} {"step": 958, "loss": 0.5770105123519897, "data": 0.16594627499580383, "gap": 0.4110642373561859, "sigma": 0.8892453908920288, "kind": "mosaic", "tokens": 225, "dense": 448, "grad_norm": 0.47609156370162964, "seconds": 4707.0} {"step": 959, "loss": 0.3472342789173126, "data": 0.3472342789173126, "gap": null, "sigma": 0.9012266397476196, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.18770669400691986, "seconds": 4709.9} {"step": 960, "loss": 1.5878039598464966, "data": 0.13447138667106628, "gap": 1.453332543373108, "sigma": 0.9555015563964844, "kind": "mosaic", "tokens": 194, "dense": 448, "grad_norm": 0.930060863494873, "seconds": 4714.4} {"step": 961, "loss": 0.6135008335113525, "data": 0.22501905262470245, "gap": 0.3884817957878113, "sigma": 0.9575514793395996, "kind": "mosaic", "tokens": 194, "dense": 448, "grad_norm": 0.3450048565864563, "seconds": 4719.0} {"step": 962, "loss": 0.4947543442249298, "data": 0.4947543442249298, "gap": null, "sigma": 0.9714997410774231, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.4214710593223572, "seconds": 4721.8} {"step": 963, "loss": 0.647269606590271, "data": 0.3292847275733948, "gap": 0.3179849088191986, "sigma": 0.938025712966919, "kind": "mosaic", "tokens": 274, "dense": 448, "grad_norm": 0.2280578762292862, "seconds": 4726.3} {"step": 964, "loss": 0.359392911195755, "data": 0.359392911195755, "gap": null, "sigma": 0.35897374153137207, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.056212496012449265, "seconds": 4729.1} {"step": 965, "loss": 0.6229067444801331, "data": 0.1780952364206314, "gap": 0.44481152296066284, "sigma": 0.8433358073234558, "kind": "mosaic", "tokens": 166, "dense": 448, "grad_norm": 0.2548031210899353, "seconds": 4733.6} {"step": 966, "loss": 0.7292206287384033, "data": 0.3870922923088074, "gap": 0.34212830662727356, "sigma": 0.31986165046691895, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.3444022536277771, "seconds": 4738.1} {"step": 967, "loss": 0.8102014064788818, "data": 0.24831745028495789, "gap": 0.5618839263916016, "sigma": 0.8564670085906982, "kind": "mosaic", "tokens": 179, "dense": 448, "grad_norm": 0.38077038526535034, "seconds": 4742.5} {"step": 968, "loss": 0.6528950929641724, "data": 0.17850007116794586, "gap": 0.4743950068950653, "sigma": 0.9230995178222656, "kind": "mosaic", "tokens": 192, "dense": 448, "grad_norm": 0.4860987663269043, "seconds": 4747.0} {"step": 969, "loss": 0.46332237124443054, "data": 0.46332237124443054, "gap": null, "sigma": 0.9908865094184875, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3575896918773651, "seconds": 4749.8} {"step": 970, "loss": 0.4918352961540222, "data": 0.16274039447307587, "gap": 0.32909491658210754, "sigma": 0.9148757457733154, "kind": "mosaic", "tokens": 239, "dense": 448, "grad_norm": 0.23607514798641205, "seconds": 4754.4} {"step": 971, "loss": 0.34467294812202454, "data": 0.10050906985998154, "gap": 0.2441638708114624, "sigma": 0.6075635552406311, "kind": "mosaic", "tokens": 149, "dense": 448, "grad_norm": 0.17876510322093964, "seconds": 4758.7} {"step": 972, "loss": 0.32340705394744873, "data": 0.32340705394744873, "gap": null, "sigma": 0.8424490690231323, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.14285686612129211, "seconds": 4761.5} {"step": 973, "loss": 0.48867326974868774, "data": 0.237550288438797, "gap": 0.25112298130989075, "sigma": 0.4401220679283142, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.25444725155830383, "seconds": 4766.1} {"step": 974, "loss": 1.0908015966415405, "data": 0.37641847133636475, "gap": 0.7143831253051758, "sigma": 0.31361570954322815, "kind": "mosaic", "tokens": 223, "dense": 448, "grad_norm": 0.29235750436782837, "seconds": 4770.8} {"step": 975, "loss": 0.44927918910980225, "data": 0.20095764100551605, "gap": 0.2483215481042862, "sigma": 0.6129260063171387, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.23634089529514313, "seconds": 4775.2} {"step": 976, "loss": 0.749432384967804, "data": 0.2531672418117523, "gap": 0.49626514315605164, "sigma": 0.42559701204299927, "kind": "mosaic", "tokens": 222, "dense": 480, "grad_norm": 0.2493315488100052, "seconds": 4779.9} {"step": 977, "loss": 0.3524496257305145, "data": 0.049963947385549545, "gap": 0.3024856746196747, "sigma": 0.856918215751648, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 0.43503832817077637, "seconds": 4783.8} {"step": 978, "loss": 0.19819055497646332, "data": 0.19819055497646332, "gap": null, "sigma": 0.6538556814193726, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.07941420376300812, "seconds": 4786.6} {"step": 979, "loss": 0.5594848394393921, "data": 0.13793005049228668, "gap": 0.4215548038482666, "sigma": 0.5348535776138306, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.2519742548465729, "seconds": 4790.6} {"step": 980, "loss": 0.8039634227752686, "data": 0.30211785435676575, "gap": 0.5018455386161804, "sigma": 0.4472363293170929, "kind": "mosaic", "tokens": 244, "dense": 448, "grad_norm": 0.16938768327236176, "seconds": 4795.0} {"step": 981, "loss": 0.32701215147972107, "data": 0.32701215147972107, "gap": null, "sigma": 0.9648091197013855, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2693025469779968, "seconds": 4797.8} {"step": 982, "loss": 0.37081480026245117, "data": 0.13234193623065948, "gap": 0.2384728640317917, "sigma": 0.48584961891174316, "kind": "mosaic", "tokens": 252, "dense": 448, "grad_norm": 0.19918012619018555, "seconds": 4802.3} {"step": 983, "loss": 3.601055860519409, "data": 0.807525098323822, "gap": 2.7935307025909424, "sigma": 0.1417187601327896, "kind": "mosaic", "tokens": 259, "dense": 448, "grad_norm": 1.7141846418380737, "seconds": 4806.9} {"step": 984, "loss": 0.23627915978431702, "data": 0.11065113544464111, "gap": 0.1256280243396759, "sigma": 0.8246293663978577, "kind": "mosaic", "tokens": 269, "dense": 448, "grad_norm": 0.32091155648231506, "seconds": 4811.4} {"step": 985, "loss": 0.3667675852775574, "data": 0.22785992920398712, "gap": 0.13890767097473145, "sigma": 0.8410481214523315, "kind": "mosaic", "tokens": 245, "dense": 480, "grad_norm": 0.2663307785987854, "seconds": 4815.9} {"step": 986, "loss": 0.5493003129959106, "data": 0.5493003129959106, "gap": null, "sigma": 0.999401330947876, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.4508645236492157, "seconds": 4818.8} {"step": 987, "loss": 0.5196805596351624, "data": 0.19740256667137146, "gap": 0.3222779929637909, "sigma": 0.690977931022644, "kind": "mosaic", "tokens": 266, "dense": 448, "grad_norm": 0.2533419728279114, "seconds": 4823.4} {"step": 988, "loss": 0.5387366414070129, "data": 0.11342750489711761, "gap": 0.42530912160873413, "sigma": 0.8498438596725464, "kind": "mosaic", "tokens": 170, "dense": 448, "grad_norm": 0.4689837098121643, "seconds": 4827.9} {"step": 989, "loss": 0.2617708444595337, "data": 0.09584604948759079, "gap": 0.1659247875213623, "sigma": 0.6140139102935791, "kind": "mosaic", "tokens": 226, "dense": 448, "grad_norm": 0.14495748281478882, "seconds": 4832.6} {"step": 990, "loss": 0.48337486386299133, "data": 0.20564070343971252, "gap": 0.2777341604232788, "sigma": 0.7638294100761414, "kind": "mosaic", "tokens": 196, "dense": 448, "grad_norm": 0.21544091403484344, "seconds": 4837.2} {"step": 991, "loss": 0.7220826148986816, "data": 0.30061739683151245, "gap": 0.4214651882648468, "sigma": 0.4396510422229767, "kind": "mosaic", "tokens": 254, "dense": 448, "grad_norm": 0.20866818726062775, "seconds": 4841.6} {"step": 992, "loss": 0.4505801796913147, "data": 0.22253164649009705, "gap": 0.22804853320121765, "sigma": 0.8074902296066284, "kind": "mosaic", "tokens": 247, "dense": 448, "grad_norm": 0.18021297454833984, "seconds": 4846.0} {"step": 993, "loss": 0.7082308530807495, "data": 0.262818306684494, "gap": 0.4454125761985779, "sigma": 0.8918776512145996, "kind": "mosaic", "tokens": 243, "dense": 448, "grad_norm": 0.7782512903213501, "seconds": 4850.6} {"step": 994, "loss": 0.9093747735023499, "data": 0.29266470670700073, "gap": 0.6167100667953491, "sigma": 0.471550852060318, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.4074123799800873, "seconds": 4854.8} {"step": 995, "loss": 0.3465879261493683, "data": 0.3465879261493683, "gap": null, "sigma": 0.9229808449745178, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.1691761314868927, "seconds": 4857.6} {"step": 996, "loss": 0.3502306342124939, "data": 0.20896664261817932, "gap": 0.14126400649547577, "sigma": 0.8787692189216614, "kind": "mosaic", "tokens": 176, "dense": 448, "grad_norm": 0.18582390248775482, "seconds": 4862.0} {"step": 997, "loss": 0.5155526995658875, "data": 0.23347048461437225, "gap": 0.282082200050354, "sigma": 0.4440465271472931, "kind": "mosaic", "tokens": 215, "dense": 448, "grad_norm": 0.18478058278560638, "seconds": 4866.5} {"step": 998, "loss": 0.25540006160736084, "data": 0.25540006160736084, "gap": null, "sigma": 0.694516658782959, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.11173064261674881, "seconds": 4869.3} {"step": 999, "loss": 0.4276329278945923, "data": 0.17340825498104095, "gap": 0.25422465801239014, "sigma": 0.5592190027236938, "kind": "mosaic", "tokens": 251, "dense": 448, "grad_norm": 0.12767742574214935, "seconds": 4873.9} {"step": 1000, "loss": 0.15178312361240387, "data": 0.15178312361240387, "gap": null, "sigma": 0.7374703288078308, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.6785165667533875, "seconds": 4876.7} {"step": 1000, "eval": {"dense": 0.30914735856155556, "uniform2": 0.18364609157045683, "mosaic": 0.2513229406128327, "gap_uniform2": 0.5969258627543846, "gap_mosaic": 0.6043623816221952}, "eval_extent": {"1x1x1": 0.4156660583814035, "1x1x2": 0.3183877985708282, "1x1x4": 0.23165325165252412, "1x2x1": 0.29897064914219773, "1x2x2": 0.19718564191441001, "1x2x4": 0.1305640237313167, "1x4x1": 0.2000811755752229, "1x4x2": 0.11348232635538334, "1x4x4": 0.08357604812459003}} {"step": 1001, "loss": 0.42317381501197815, "data": 0.22007474303245544, "gap": 0.2030990719795227, "sigma": 0.7076337337493896, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.16426114737987518, "seconds": 5124.4} {"step": 1002, "loss": 0.46126097440719604, "data": 0.17710937559604645, "gap": 0.2841516137123108, "sigma": 0.8317615389823914, "kind": "mosaic", "tokens": 182, "dense": 448, "grad_norm": 0.235671266913414, "seconds": 5128.7} {"step": 1003, "loss": 0.2887260317802429, "data": 0.1928386688232422, "gap": 0.09588735550642014, "sigma": 0.8704923987388611, "kind": "mosaic", "tokens": 176, "dense": 448, "grad_norm": 0.16929852962493896, "seconds": 5133.2} {"step": 1004, "loss": 0.9109222292900085, "data": 0.31275832653045654, "gap": 0.598163902759552, "sigma": 0.9383535385131836, "kind": "mosaic", "tokens": 276, "dense": 448, "grad_norm": 0.4278707206249237, "seconds": 5137.6} {"step": 1005, "loss": 0.5969045758247375, "data": 0.23307569324970245, "gap": 0.3638288974761963, "sigma": 0.9227415323257446, "kind": "mosaic", "tokens": 172, "dense": 448, "grad_norm": 1.7350465059280396, "seconds": 5142.0} {"step": 1006, "loss": 0.5095736980438232, "data": 0.2486260086297989, "gap": 0.26094770431518555, "sigma": 0.5734415054321289, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.20066645741462708, "seconds": 5146.6} {"step": 1007, "loss": 25.17913818359375, "data": 4.579980850219727, "gap": 20.599157333374023, "sigma": 0.02855224348604679, "kind": "mosaic", "tokens": 202, "dense": 448, "grad_norm": 17.723453521728516, "seconds": 5151.2} {"step": 1008, "loss": 0.6451873779296875, "data": 0.2730760872364044, "gap": 0.3721112906932831, "sigma": 0.8246299624443054, "kind": "mosaic", "tokens": 251, "dense": 448, "grad_norm": 0.5619493126869202, "seconds": 5155.8} {"step": 1009, "loss": 0.5231263637542725, "data": 0.185856431722641, "gap": 0.3372699022293091, "sigma": 0.5682690143585205, "kind": "mosaic", "tokens": 165, "dense": 448, "grad_norm": 0.194000244140625, "seconds": 5160.4} {"step": 1010, "loss": 0.76918625831604, "data": 0.20624519884586334, "gap": 0.5629410743713379, "sigma": 0.8309997320175171, "kind": "mosaic", "tokens": 198, "dense": 448, "grad_norm": 0.40973055362701416, "seconds": 5165.1} {"step": 1011, "loss": 0.5802898406982422, "data": 0.0645979642868042, "gap": 0.515691876411438, "sigma": 0.7093879580497742, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 0.2786600887775421, "seconds": 5169.2} {"step": 1012, "loss": 0.7627004384994507, "data": 0.3077065050601959, "gap": 0.45499393343925476, "sigma": 0.952200710773468, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.467195600271225, "seconds": 5173.8} {"step": 1013, "loss": 0.39122796058654785, "data": 0.13891306519508362, "gap": 0.25231489539146423, "sigma": 0.6584538817405701, "kind": "mosaic", "tokens": 197, "dense": 448, "grad_norm": 0.2684306502342224, "seconds": 5178.6} {"step": 1014, "loss": 0.39219480752944946, "data": 0.19567035138607025, "gap": 0.1965244561433792, "sigma": 0.7640566825866699, "kind": "mosaic", "tokens": 262, "dense": 448, "grad_norm": 0.16747207939624786, "seconds": 5183.0} {"step": 1015, "loss": 0.4811283349990845, "data": 0.2342124581336975, "gap": 0.24691586196422577, "sigma": 0.6736651659011841, "kind": "mosaic", "tokens": 330, "dense": 576, "grad_norm": 0.13823302090168, "seconds": 5187.8} {"step": 1016, "loss": 0.7633862495422363, "data": 0.263152539730072, "gap": 0.5002337098121643, "sigma": 0.43019524216651917, "kind": "mosaic", "tokens": 236, "dense": 448, "grad_norm": 0.19617469608783722, "seconds": 5192.5} {"step": 1017, "loss": 0.23025113344192505, "data": 0.23025113344192505, "gap": null, "sigma": 0.5378860831260681, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.05421293154358864, "seconds": 5195.4} {"step": 1018, "loss": 0.6816202402114868, "data": 0.2711215615272522, "gap": 0.4104986786842346, "sigma": 0.9054797291755676, "kind": "mosaic", "tokens": 210, "dense": 448, "grad_norm": 0.38674384355545044, "seconds": 5199.9} {"step": 1019, "loss": 0.7956869602203369, "data": 0.28919586539268494, "gap": 0.5064910650253296, "sigma": 0.9524341821670532, "kind": "mosaic", "tokens": 223, "dense": 448, "grad_norm": 0.3905261158943176, "seconds": 5204.5} {"step": 1020, "loss": 0.3216121196746826, "data": 0.16451439261436462, "gap": 0.157097727060318, "sigma": 0.47344061732292175, "kind": "mosaic", "tokens": 243, "dense": 448, "grad_norm": 0.1933985948562622, "seconds": 5209.0} {"step": 1021, "loss": 0.5136346220970154, "data": 0.18890586495399475, "gap": 0.32472875714302063, "sigma": 0.8073803782463074, "kind": "mosaic", "tokens": 268, "dense": 448, "grad_norm": 0.2623669505119324, "seconds": 5213.6} {"step": 1022, "loss": 0.811898946762085, "data": 0.22783447802066803, "gap": 0.5840644836425781, "sigma": 0.46945133805274963, "kind": "mosaic", "tokens": 158, "dense": 448, "grad_norm": 0.24211427569389343, "seconds": 5218.2} {"step": 1023, "loss": 0.3542839586734772, "data": 0.3542839586734772, "gap": null, "sigma": 0.8748735189437866, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.20500892400741577, "seconds": 5221.1} {"step": 1024, "loss": 0.9679667353630066, "data": 0.9679667353630066, "gap": null, "sigma": 0.9966344833374023, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 1.5189019441604614, "seconds": 5224.0} {"step": 1025, "loss": 0.26064199209213257, "data": 0.11192503571510315, "gap": 0.1487169712781906, "sigma": 0.5567399859428406, "kind": "mosaic", "tokens": 263, "dense": 448, "grad_norm": 0.12283901125192642, "seconds": 5228.7} {"step": 1026, "loss": 0.46928170323371887, "data": 0.46928170323371887, "gap": null, "sigma": 0.9293007850646973, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.17709344625473022, "seconds": 5231.6} {"step": 1027, "loss": 0.3437578082084656, "data": 0.1271398514509201, "gap": 0.21661797165870667, "sigma": 0.6392093300819397, "kind": "mosaic", "tokens": 171, "dense": 448, "grad_norm": 0.17460252344608307, "seconds": 5236.0} {"step": 1028, "loss": 0.5412437319755554, "data": 0.1789558082818985, "gap": 0.3622879385948181, "sigma": 0.5081088542938232, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.25038084387779236, "seconds": 5240.4} {"step": 1029, "loss": 0.513910174369812, "data": 0.18952535092830658, "gap": 0.3243848383426666, "sigma": 0.5988051891326904, "kind": "mosaic", "tokens": 218, "dense": 448, "grad_norm": 0.17665112018585205, "seconds": 5244.8} {"step": 1030, "loss": 0.44962403178215027, "data": 0.08351068943738937, "gap": 0.3661133348941803, "sigma": 0.9208945631980896, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.2819100022315979, "seconds": 5249.0} {"step": 1031, "loss": 0.48847055435180664, "data": 0.266996830701828, "gap": 0.22147372364997864, "sigma": 0.7855920791625977, "kind": "mosaic", "tokens": 283, "dense": 448, "grad_norm": 0.22610978782176971, "seconds": 5253.5} {"step": 1032, "loss": 0.24624021351337433, "data": 0.24624021351337433, "gap": null, "sigma": 0.7624310851097107, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.19219428300857544, "seconds": 5256.4} {"step": 1033, "loss": 0.46027812361717224, "data": 0.46027812361717224, "gap": null, "sigma": 0.9788763523101807, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3669865131378174, "seconds": 5259.2} {"step": 1034, "loss": 0.32814836502075195, "data": 0.32814836502075195, "gap": null, "sigma": 0.9849645495414734, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.365865558385849, "seconds": 5262.2} {"step": 1035, "loss": 0.4176579713821411, "data": 0.2043430209159851, "gap": 0.2133149355649948, "sigma": 0.47599250078201294, "kind": "mosaic", "tokens": 279, "dense": 448, "grad_norm": 0.16100184619426727, "seconds": 5266.9} {"step": 1036, "loss": 0.7472516298294067, "data": 0.21070127189159393, "gap": 0.5365503430366516, "sigma": 0.8863189220428467, "kind": "mosaic", "tokens": 189, "dense": 448, "grad_norm": 0.2807862162590027, "seconds": 5271.5} {"step": 1037, "loss": 0.9868320822715759, "data": 0.3210234045982361, "gap": 0.6658086776733398, "sigma": 0.371773898601532, "kind": "mosaic", "tokens": 230, "dense": 448, "grad_norm": 0.22404684126377106, "seconds": 5276.1} {"step": 1038, "loss": 0.7633349895477295, "data": 0.24749179184436798, "gap": 0.5158432126045227, "sigma": 0.4033597707748413, "kind": "mosaic", "tokens": 225, "dense": 448, "grad_norm": 0.20380431413650513, "seconds": 5280.7} {"step": 1039, "loss": 0.4321824014186859, "data": 0.16247805953025818, "gap": 0.26970434188842773, "sigma": 0.8042165637016296, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.4924965798854828, "seconds": 5285.2} {"step": 1040, "loss": 0.5569412708282471, "data": 0.09873943030834198, "gap": 0.4582018256187439, "sigma": 0.8705103397369385, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.7197544574737549, "seconds": 5289.6} {"step": 1041, "loss": 1.5179500579833984, "data": 0.4767736792564392, "gap": 1.0411763191223145, "sigma": 0.18743811547756195, "kind": "mosaic", "tokens": 233, "dense": 448, "grad_norm": 0.6340556144714355, "seconds": 5294.2} {"step": 1042, "loss": 2.9713058471679688, "data": 0.7376881837844849, "gap": 2.2336175441741943, "sigma": 0.1137637123465538, "kind": "mosaic", "tokens": 240, "dense": 448, "grad_norm": 1.821761965751648, "seconds": 5298.9} {"step": 1043, "loss": 1.4613512754440308, "data": 0.32163915038108826, "gap": 1.1397120952606201, "sigma": 0.997545063495636, "kind": "mosaic", "tokens": 240, "dense": 448, "grad_norm": 0.8495290875434875, "seconds": 5303.5} {"step": 1044, "loss": 0.34877923130989075, "data": 0.34877923130989075, "gap": null, "sigma": 0.9499409794807434, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.26726964116096497, "seconds": 5306.3} {"step": 1045, "loss": 0.2461685985326767, "data": 0.2461685985326767, "gap": null, "sigma": 0.8377148509025574, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.1827106475830078, "seconds": 5309.1} {"step": 1046, "loss": 0.8490095734596252, "data": 0.3118436336517334, "gap": 0.5371659398078918, "sigma": 0.2911547124385834, "kind": "mosaic", "tokens": 204, "dense": 448, "grad_norm": 0.25065112113952637, "seconds": 5313.6} {"step": 1047, "loss": 0.6732103228569031, "data": 0.23489730060100555, "gap": 0.43831300735473633, "sigma": 0.5444642901420593, "kind": "mosaic", "tokens": 234, "dense": 448, "grad_norm": 0.2231307029724121, "seconds": 5318.2} {"step": 1048, "loss": 0.5806450843811035, "data": 0.2741895318031311, "gap": 0.3064555525779724, "sigma": 0.773483395576477, "kind": "mosaic", "tokens": 231, "dense": 448, "grad_norm": 0.20258736610412598, "seconds": 5322.8} {"step": 1049, "loss": 0.4773963689804077, "data": 0.17549575865268707, "gap": 0.30190059542655945, "sigma": 0.5317038893699646, "kind": "mosaic", "tokens": 202, "dense": 448, "grad_norm": 0.23432274162769318, "seconds": 5327.2} {"step": 1050, "loss": 0.6659759283065796, "data": 0.22294454276561737, "gap": 0.443031370639801, "sigma": 0.5307636857032776, "kind": "mosaic", "tokens": 200, "dense": 448, "grad_norm": 0.1829400509595871, "seconds": 5331.8} {"step": 1051, "loss": 0.33818262815475464, "data": 0.15387240052223206, "gap": 0.18431024253368378, "sigma": 0.7164784669876099, "kind": "mosaic", "tokens": 183, "dense": 448, "grad_norm": 0.1361282467842102, "seconds": 5336.1} {"step": 1052, "loss": 0.6646921634674072, "data": 0.26026245951652527, "gap": 0.40442973375320435, "sigma": 0.7144175171852112, "kind": "mosaic", "tokens": 198, "dense": 448, "grad_norm": 0.1980087161064148, "seconds": 5340.5} {"step": 1053, "loss": 2.068913459777832, "data": 0.30591505765914917, "gap": 1.762998342514038, "sigma": 0.9953092932701111, "kind": "mosaic", "tokens": 243, "dense": 448, "grad_norm": 1.3672767877578735, "seconds": 5345.1} {"step": 1054, "loss": 0.5160597562789917, "data": 0.2183280736207962, "gap": 0.2977316975593567, "sigma": 0.929814875125885, "kind": "mosaic", "tokens": 252, "dense": 448, "grad_norm": 0.31560519337654114, "seconds": 5349.8} {"step": 1055, "loss": 0.5315836071968079, "data": 0.146707221865654, "gap": 0.3848763704299927, "sigma": 0.811864972114563, "kind": "mosaic", "tokens": 180, "dense": 448, "grad_norm": 0.2918856739997864, "seconds": 5354.3} {"step": 1056, "loss": 0.6296399235725403, "data": 0.6296399235725403, "gap": null, "sigma": 0.9797152876853943, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.5650292634963989, "seconds": 5357.1} {"step": 1057, "loss": 0.5423616170883179, "data": 0.1973448246717453, "gap": 0.34501680731773376, "sigma": 0.3968855142593384, "kind": "mosaic", "tokens": 188, "dense": 448, "grad_norm": 0.20221729576587677, "seconds": 5361.6} {"step": 1058, "loss": 0.5494867563247681, "data": 0.24367618560791016, "gap": 0.3058105409145355, "sigma": 0.9619984030723572, "kind": "mosaic", "tokens": 191, "dense": 448, "grad_norm": 0.34438464045524597, "seconds": 5366.2} {"step": 1059, "loss": 8.808418273925781, "data": 0.7303032875061035, "gap": 8.07811450958252, "sigma": 0.1081450879573822, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 7.526813507080078, "seconds": 5370.3} {"step": 1060, "loss": 0.39167988300323486, "data": 0.1661175638437271, "gap": 0.22556230425834656, "sigma": 0.5545751452445984, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.22087925672531128, "seconds": 5375.0} {"step": 1061, "loss": 0.9488617181777954, "data": 0.2839175760746002, "gap": 0.6649441719055176, "sigma": 0.4140867292881012, "kind": "mosaic", "tokens": 165, "dense": 448, "grad_norm": 0.1920468807220459, "seconds": 5379.3} {"step": 1062, "loss": 0.5558767318725586, "data": 0.23351992666721344, "gap": 0.32235679030418396, "sigma": 0.6965533494949341, "kind": "mosaic", "tokens": 250, "dense": 448, "grad_norm": 0.22214621305465698, "seconds": 5383.8} {"step": 1063, "loss": 0.25611525774002075, "data": 0.25611525774002075, "gap": null, "sigma": 0.8462982773780823, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.11792074888944626, "seconds": 5386.7} {"step": 1064, "loss": 0.41572901606559753, "data": 0.16550800204277039, "gap": 0.25022101402282715, "sigma": 0.710229754447937, "kind": "mosaic", "tokens": 184, "dense": 448, "grad_norm": 0.14588230848312378, "seconds": 5391.2} {"step": 1065, "loss": 0.7596769332885742, "data": 0.197792187333107, "gap": 0.5618847608566284, "sigma": 0.4376428723335266, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.3257407248020172, "seconds": 5395.4} {"step": 1066, "loss": 0.5868393778800964, "data": 0.18948067724704742, "gap": 0.3973587155342102, "sigma": 0.5838697552680969, "kind": "mosaic", "tokens": 218, "dense": 448, "grad_norm": 0.17808613181114197, "seconds": 5400.0} {"step": 1067, "loss": 0.34283825755119324, "data": 0.34283825755119324, "gap": null, "sigma": 0.8000940680503845, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.16092371940612793, "seconds": 5402.8} {"step": 1068, "loss": 0.5722324848175049, "data": 0.25407305359840393, "gap": 0.31815946102142334, "sigma": 0.5879275798797607, "kind": "mosaic", "tokens": 214, "dense": 448, "grad_norm": 0.13171331584453583, "seconds": 5407.4} {"step": 1069, "loss": 0.5914723873138428, "data": 0.24761654436588287, "gap": 0.3438558578491211, "sigma": 0.776183009147644, "kind": "mosaic", "tokens": 237, "dense": 448, "grad_norm": 0.2995753586292267, "seconds": 5411.9} {"step": 1070, "loss": 0.7621771097183228, "data": 0.34465599060058594, "gap": 0.41752108931541443, "sigma": 0.9403496384620667, "kind": "mosaic", "tokens": 260, "dense": 448, "grad_norm": 0.35887372493743896, "seconds": 5416.5} {"step": 1071, "loss": 1.5589189529418945, "data": 0.4367356300354004, "gap": 1.1221833229064941, "sigma": 0.9932959675788879, "kind": "mosaic", "tokens": 234, "dense": 448, "grad_norm": 0.6282365322113037, "seconds": 5421.1} {"step": 1072, "loss": 0.4807141423225403, "data": 0.19468970596790314, "gap": 0.28602442145347595, "sigma": 0.6974320411682129, "kind": "mosaic", "tokens": 218, "dense": 448, "grad_norm": 0.1402588039636612, "seconds": 5425.6} {"step": 1073, "loss": 0.3763623535633087, "data": 0.3763623535633087, "gap": null, "sigma": 0.9327767491340637, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.23000767827033997, "seconds": 5428.5} {"step": 1074, "loss": 0.5740652680397034, "data": 0.1707366406917572, "gap": 0.40332862734794617, "sigma": 0.8877229690551758, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.3541799485683441, "seconds": 5432.7} {"step": 1075, "loss": 0.35667550563812256, "data": 0.35667550563812256, "gap": null, "sigma": 0.9402605891227722, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.1485312581062317, "seconds": 5435.4} {"step": 1076, "loss": 0.2195223867893219, "data": 0.2195223867893219, "gap": null, "sigma": 0.357572466135025, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.05481978505849838, "seconds": 5438.3} {"step": 1077, "loss": 0.4590340852737427, "data": 0.07657715678215027, "gap": 0.3824569284915924, "sigma": 0.9273834824562073, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.3582788407802582, "seconds": 5442.5} {"step": 1078, "loss": 2.9466733932495117, "data": 0.721542239189148, "gap": 2.225131034851074, "sigma": 0.15393684804439545, "kind": "mosaic", "tokens": 211, "dense": 448, "grad_norm": 1.090230941772461, "seconds": 5446.9} {"step": 1079, "loss": 0.2976628541946411, "data": 0.12308889627456665, "gap": 0.17457397282123566, "sigma": 0.7489888072013855, "kind": "mosaic", "tokens": 215, "dense": 448, "grad_norm": 0.16753114759922028, "seconds": 5451.5} {"step": 1080, "loss": 30.125778198242188, "data": 4.122977256774902, "gap": 26.0028018951416, "sigma": 0.06634006649255753, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 16.656564712524414, "seconds": 5455.6} {"step": 1081, "loss": 0.6668461561203003, "data": 0.1846190243959427, "gap": 0.4822271168231964, "sigma": 0.9425415396690369, "kind": "mosaic", "tokens": 241, "dense": 448, "grad_norm": 0.4850384294986725, "seconds": 5460.3} {"step": 1082, "loss": 0.8296034932136536, "data": 0.27045226097106934, "gap": 0.5591512322425842, "sigma": 0.9733427166938782, "kind": "mosaic", "tokens": 196, "dense": 448, "grad_norm": 1.333114743232727, "seconds": 5464.7} {"step": 1083, "loss": 0.3532910645008087, "data": 0.3532910645008087, "gap": null, "sigma": 0.9104374647140503, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.11285459995269775, "seconds": 5467.5} {"step": 1084, "loss": 10.038267135620117, "data": 1.9416660070419312, "gap": 8.096601486206055, "sigma": 0.07520657032728195, "kind": "mosaic", "tokens": 235, "dense": 448, "grad_norm": 5.939075946807861, "seconds": 5472.1} {"step": 1085, "loss": 0.48448410630226135, "data": 0.219343364238739, "gap": 0.26514074206352234, "sigma": 0.63996422290802, "kind": "mosaic", "tokens": 286, "dense": 448, "grad_norm": 0.18520094454288483, "seconds": 5476.8} {"step": 1086, "loss": 0.7045303583145142, "data": 0.7045303583145142, "gap": null, "sigma": 0.9947553873062134, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.7092891335487366, "seconds": 5479.6} {"step": 1087, "loss": 1.051194429397583, "data": 0.24087919294834137, "gap": 0.8103152513504028, "sigma": 0.9634746313095093, "kind": "mosaic", "tokens": 218, "dense": 448, "grad_norm": 0.761205792427063, "seconds": 5484.3} {"step": 1088, "loss": 0.5316907167434692, "data": 0.1748420000076294, "gap": 0.35684868693351746, "sigma": 0.40124407410621643, "kind": "mosaic", "tokens": 172, "dense": 448, "grad_norm": 0.2508532404899597, "seconds": 5488.8} {"step": 1089, "loss": 0.4725102484226227, "data": 0.08582236617803574, "gap": 0.38668787479400635, "sigma": 0.8165298700332642, "kind": "uniform", "tokens": 64, "dense": 512, "grad_norm": 0.38536566495895386, "seconds": 5493.1} {"step": 1090, "loss": 0.9086751937866211, "data": 0.4011988937854767, "gap": 0.5074763298034668, "sigma": 0.3647995889186859, "kind": "mosaic", "tokens": 258, "dense": 448, "grad_norm": 0.1733141392469406, "seconds": 5497.4} {"step": 1091, "loss": 1.2790864706039429, "data": 0.33713796734809875, "gap": 0.9419485330581665, "sigma": 0.3708074390888214, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.4006980359554291, "seconds": 5501.5} {"step": 1092, "loss": 0.44487398862838745, "data": 0.18937437236309052, "gap": 0.25549960136413574, "sigma": 0.49332505464553833, "kind": "mosaic", "tokens": 228, "dense": 448, "grad_norm": 0.16389024257659912, "seconds": 5505.9} {"step": 1093, "loss": 0.6527213454246521, "data": 0.2344895303249359, "gap": 0.4182318150997162, "sigma": 0.5558990836143494, "kind": "mosaic", "tokens": 238, "dense": 448, "grad_norm": 0.1867440938949585, "seconds": 5510.4} {"step": 1094, "loss": 0.23487776517868042, "data": 0.08283346146345139, "gap": 0.15204429626464844, "sigma": 0.68096923828125, "kind": "mosaic", "tokens": 165, "dense": 448, "grad_norm": 0.12412945181131363, "seconds": 5514.8} {"step": 1095, "loss": 1.1890543699264526, "data": 0.4018424451351166, "gap": 0.7872118949890137, "sigma": 0.9848859310150146, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.8794618844985962, "seconds": 5519.2} {"step": 1096, "loss": 0.2875632047653198, "data": 0.09766291826963425, "gap": 0.18990029394626617, "sigma": 0.7208936214447021, "kind": "mosaic", "tokens": 159, "dense": 448, "grad_norm": 1.069785475730896, "seconds": 5523.5} {"step": 1097, "loss": 2.543593168258667, "data": 0.5716834664344788, "gap": 1.9719096422195435, "sigma": 0.1532510668039322, "kind": "mosaic", "tokens": 184, "dense": 448, "grad_norm": 1.1550196409225464, "seconds": 5527.9} {"step": 1098, "loss": 0.5301804542541504, "data": 0.24183060228824615, "gap": 0.28834986686706543, "sigma": 0.7988494634628296, "kind": "mosaic", "tokens": 260, "dense": 448, "grad_norm": 0.34595969319343567, "seconds": 5532.3} {"step": 1099, "loss": 0.7145719528198242, "data": 0.23268413543701172, "gap": 0.4818878471851349, "sigma": 0.46486812829971313, "kind": "mosaic", "tokens": 198, "dense": 448, "grad_norm": 0.17780110239982605, "seconds": 5536.8} {"step": 1100, "loss": 0.6212562322616577, "data": 0.6212562322616577, "gap": null, "sigma": 0.9976843595504761, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.5311785340309143, "seconds": 5539.6} {"step": 1101, "loss": 0.4331949055194855, "data": 0.183376744389534, "gap": 0.24981816112995148, "sigma": 0.48620009422302246, "kind": "mosaic", "tokens": 207, "dense": 448, "grad_norm": 0.18599919974803925, "seconds": 5544.1} {"step": 1102, "loss": 0.17835716903209686, "data": 0.17835716903209686, "gap": null, "sigma": 0.764269232749939, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.10045916587114334, "seconds": 5546.9} {"step": 1103, "loss": 1.070594310760498, "data": 0.09532175958156586, "gap": 0.9752725958824158, "sigma": 0.5380146503448486, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 0.43176889419555664, "seconds": 5550.8} {"step": 1104, "loss": 0.6376440525054932, "data": 0.16254806518554688, "gap": 0.4750960171222687, "sigma": 0.9270014762878418, "kind": "mosaic", "tokens": 163, "dense": 448, "grad_norm": 0.33127617835998535, "seconds": 5554.9} {"step": 1105, "loss": 1.253437876701355, "data": 0.33184441924095154, "gap": 0.9215934872627258, "sigma": 0.22447019815444946, "kind": "mosaic", "tokens": 203, "dense": 448, "grad_norm": 0.5293046236038208, "seconds": 5559.3} {"step": 1106, "loss": 0.25319477915763855, "data": 0.25319477915763855, "gap": null, "sigma": 0.3980754017829895, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.04731527715921402, "seconds": 5562.1} {"step": 1107, "loss": 0.722582221031189, "data": 0.18624085187911987, "gap": 0.5363413691520691, "sigma": 0.8812757134437561, "kind": "mosaic", "tokens": 256, "dense": 448, "grad_norm": 0.28073519468307495, "seconds": 5566.6} {"step": 1108, "loss": 0.2908565402030945, "data": 0.11029636859893799, "gap": 0.1805601716041565, "sigma": 0.6922088265419006, "kind": "mosaic", "tokens": 162, "dense": 448, "grad_norm": 0.4214276075363159, "seconds": 5571.1} {"step": 1109, "loss": 0.3320874869823456, "data": 0.3320874869823456, "gap": null, "sigma": 0.6757888197898865, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.06401187926530838, "seconds": 5573.9} {"step": 1110, "loss": 0.8115679621696472, "data": 0.3561456501483917, "gap": 0.4554223120212555, "sigma": 0.9538038969039917, "kind": "mosaic", "tokens": 271, "dense": 448, "grad_norm": 0.2617117762565613, "seconds": 5578.4} {"step": 1111, "loss": 0.758773148059845, "data": 0.24944140017032623, "gap": 0.5093317627906799, "sigma": 0.5989706516265869, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.3614620864391327, "seconds": 5582.6} {"step": 1112, "loss": 0.4153836965560913, "data": 0.19486477971076965, "gap": 0.22051891684532166, "sigma": 0.6684442758560181, "kind": "mosaic", "tokens": 254, "dense": 448, "grad_norm": 0.14816942811012268, "seconds": 5587.2} {"step": 1113, "loss": 0.4061901867389679, "data": 0.21918326616287231, "gap": 0.18700692057609558, "sigma": 0.9077011942863464, "kind": "mosaic", "tokens": 261, "dense": 448, "grad_norm": 0.36534130573272705, "seconds": 5591.7} {"step": 1114, "loss": 1.1596304178237915, "data": 0.2399803102016449, "gap": 0.9196500778198242, "sigma": 0.9595515727996826, "kind": "mosaic", "tokens": 245, "dense": 448, "grad_norm": 1.6391184329986572, "seconds": 5596.0} {"step": 1115, "loss": 0.676224946975708, "data": 0.676224946975708, "gap": null, "sigma": 0.9914494752883911, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.5860730409622192, "seconds": 5598.8} {"step": 1116, "loss": 1.061888575553894, "data": 0.33594372868537903, "gap": 0.7259448170661926, "sigma": 0.37206920981407166, "kind": "mosaic", "tokens": 220, "dense": 448, "grad_norm": 0.2470504641532898, "seconds": 5603.2} {"step": 1117, "loss": 0.1902187466621399, "data": 0.08784224838018417, "gap": 0.10237650573253632, "sigma": 0.6642188429832458, "kind": "mosaic", "tokens": 248, "dense": 448, "grad_norm": 0.11904767155647278, "seconds": 5607.7} {"step": 1118, "loss": 0.7633500099182129, "data": 0.23473379015922546, "gap": 0.528616189956665, "sigma": 0.5506759285926819, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.26045656204223633, "seconds": 5612.0} {"step": 1119, "loss": 3.46065616607666, "data": 1.32425856590271, "gap": 2.13639760017395, "sigma": 0.11660044640302658, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 1.9334826469421387, "seconds": 5616.5} {"step": 1120, "loss": 0.6924300789833069, "data": 0.2546023428440094, "gap": 0.4378277361392975, "sigma": 0.9382022619247437, "kind": "mosaic", "tokens": 210, "dense": 448, "grad_norm": 0.3334818184375763, "seconds": 5620.9} {"step": 1121, "loss": 0.520317554473877, "data": 0.170669823884964, "gap": 0.3496477007865906, "sigma": 0.7871043682098389, "kind": "mosaic", "tokens": 176, "dense": 448, "grad_norm": 0.22970227897167206, "seconds": 5625.3} {"step": 1122, "loss": 0.509617269039154, "data": 0.509617269039154, "gap": null, "sigma": 0.9764912128448486, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.29420188069343567, "seconds": 5628.0} {"step": 1123, "loss": 0.3944372832775116, "data": 0.1674283742904663, "gap": 0.2270089089870453, "sigma": 0.5466588139533997, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.18482035398483276, "seconds": 5632.5} {"step": 1124, "loss": 0.361293226480484, "data": 0.361293226480484, "gap": null, "sigma": 0.9079922437667847, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.15407328307628632, "seconds": 5635.3} {"step": 1125, "loss": 0.3929159343242645, "data": 0.17152172327041626, "gap": 0.22139421105384827, "sigma": 0.5407975316047668, "kind": "mosaic", "tokens": 236, "dense": 448, "grad_norm": 0.16587898135185242, "seconds": 5639.8} {"step": 1126, "loss": 1.1798393726348877, "data": 0.3752896189689636, "gap": 0.8045498132705688, "sigma": 0.3230438232421875, "kind": "mosaic", "tokens": 247, "dense": 448, "grad_norm": 0.26566460728645325, "seconds": 5644.3} {"step": 1127, "loss": 0.6235711574554443, "data": 0.23492349684238434, "gap": 0.3886476755142212, "sigma": 0.5601056218147278, "kind": "mosaic", "tokens": 225, "dense": 448, "grad_norm": 0.1793409138917923, "seconds": 5648.8} {"step": 1128, "loss": 1.0500576496124268, "data": 0.23441708087921143, "gap": 0.8156405091285706, "sigma": 0.9852910041809082, "kind": "mosaic", "tokens": 246, "dense": 448, "grad_norm": 0.6132308840751648, "seconds": 5653.3} {"step": 1129, "loss": 0.5342364311218262, "data": 0.1740698218345642, "gap": 0.36016663908958435, "sigma": 0.9785823225975037, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 1.035783052444458, "seconds": 5657.5} {"step": 1130, "loss": 0.40407997369766235, "data": 0.1845339685678482, "gap": 0.21954602003097534, "sigma": 0.7703566551208496, "kind": "mosaic", "tokens": 270, "dense": 448, "grad_norm": 0.20386740565299988, "seconds": 5662.1} {"step": 1131, "loss": 0.6887365579605103, "data": 0.16803911328315735, "gap": 0.5206974148750305, "sigma": 0.8497990965843201, "kind": "mosaic", "tokens": 217, "dense": 448, "grad_norm": 0.36487284302711487, "seconds": 5666.6} {"step": 1132, "loss": 0.6878799200057983, "data": 0.3567565381526947, "gap": 0.33112335205078125, "sigma": 0.9341872930526733, "kind": "mosaic", "tokens": 221, "dense": 448, "grad_norm": 0.22055508196353912, "seconds": 5671.1} {"step": 1133, "loss": 0.1905568689107895, "data": 0.1905568689107895, "gap": null, "sigma": 0.8941738605499268, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.17647358775138855, "seconds": 5673.9} {"step": 1134, "loss": 0.6148480772972107, "data": 0.23210649192333221, "gap": 0.3827416002750397, "sigma": 0.9620356559753418, "kind": "mosaic", "tokens": 225, "dense": 448, "grad_norm": 0.28729766607284546, "seconds": 5678.3} {"step": 1135, "loss": 0.30459660291671753, "data": 0.30459660291671753, "gap": null, "sigma": 0.8684489130973816, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.24040207266807556, "seconds": 5681.1} {"step": 1136, "loss": 0.5645150542259216, "data": 0.13960117101669312, "gap": 0.4249138832092285, "sigma": 0.5754328966140747, "kind": "mosaic", "tokens": 148, "dense": 448, "grad_norm": 0.2275608628988266, "seconds": 5685.4} {"step": 1137, "loss": 1.4641605615615845, "data": 0.413545697927475, "gap": 1.050614833831787, "sigma": 0.2649020254611969, "kind": "mosaic", "tokens": 241, "dense": 448, "grad_norm": 0.4706929326057434, "seconds": 5689.8} {"step": 1138, "loss": 0.43709641695022583, "data": 0.21504366397857666, "gap": 0.22205276787281036, "sigma": 0.7510265111923218, "kind": "mosaic", "tokens": 213, "dense": 448, "grad_norm": 0.17552641034126282, "seconds": 5694.3} {"step": 1139, "loss": 0.5417038798332214, "data": 0.5417038798332214, "gap": null, "sigma": 0.992391049861908, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.36326733231544495, "seconds": 5697.1} {"step": 1140, "loss": 0.6780068278312683, "data": 0.23497150838375092, "gap": 0.4430353343486786, "sigma": 0.967195987701416, "kind": "mosaic", "tokens": 204, "dense": 448, "grad_norm": 0.5299682021141052, "seconds": 5701.5} {"step": 1141, "loss": 0.5968104600906372, "data": 0.2344159185886383, "gap": 0.3623945713043213, "sigma": 0.9251276254653931, "kind": "mosaic", "tokens": 174, "dense": 448, "grad_norm": 0.19623711705207825, "seconds": 5706.0} {"step": 1142, "loss": 0.3838403522968292, "data": 0.3838403522968292, "gap": null, "sigma": 0.973733127117157, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.29662948846817017, "seconds": 5708.7} {"step": 1143, "loss": 0.6690487265586853, "data": 0.10957632958889008, "gap": 0.559472382068634, "sigma": 0.7088268399238586, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.3138124942779541, "seconds": 5712.9} {"step": 1144, "loss": 0.4018961191177368, "data": 0.12346011400222778, "gap": 0.27843600511550903, "sigma": 0.5584775805473328, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.2769882380962372, "seconds": 5717.1} {"step": 1145, "loss": 0.5208287239074707, "data": 0.5208287239074707, "gap": null, "sigma": 0.9997884035110474, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.6160470843315125, "seconds": 5719.8} {"step": 1146, "loss": 1.7204276323318481, "data": 0.3146289587020874, "gap": 1.4057986736297607, "sigma": 0.9543823599815369, "kind": "mosaic", "tokens": 282, "dense": 448, "grad_norm": 1.2566522359848022, "seconds": 5724.3} {"step": 1147, "loss": 0.4117056131362915, "data": 0.18724291026592255, "gap": 0.22446270287036896, "sigma": 0.9819751977920532, "kind": "mosaic", "tokens": 202, "dense": 448, "grad_norm": 0.28696510195732117, "seconds": 5728.6} {"step": 1148, "loss": 1.5013079643249512, "data": 0.315388023853302, "gap": 1.1859198808670044, "sigma": 0.9969931840896606, "kind": "mosaic", "tokens": 280, "dense": 448, "grad_norm": 0.9600368142127991, "seconds": 5733.2} {"step": 1149, "loss": 1.4229037761688232, "data": 0.28043419122695923, "gap": 1.1424696445465088, "sigma": 0.9410423636436462, "kind": "mosaic", "tokens": 195, "dense": 448, "grad_norm": 1.2325125932693481, "seconds": 5737.6} {"step": 1150, "loss": 0.6477733850479126, "data": 0.29710596799850464, "gap": 0.35066741704940796, "sigma": 0.9484527707099915, "kind": "mosaic", "tokens": 201, "dense": 448, "grad_norm": 0.399944007396698, "seconds": 5742.0} {"step": 1151, "loss": 0.9270390272140503, "data": 0.27184292674064636, "gap": 0.6551961302757263, "sigma": 0.8522072434425354, "kind": "mosaic", "tokens": 218, "dense": 448, "grad_norm": 0.32491129636764526, "seconds": 5746.5} {"step": 1152, "loss": 0.26070302724838257, "data": 0.1425912082195282, "gap": 0.11811182647943497, "sigma": 0.7787394523620605, "kind": "mosaic", "tokens": 232, "dense": 448, "grad_norm": 0.17568440735340118, "seconds": 5751.1} {"step": 1153, "loss": 0.5594401359558105, "data": 0.22226645052433014, "gap": 0.3371736705303192, "sigma": 0.3674337565898895, "kind": "mosaic", "tokens": 223, "dense": 448, "grad_norm": 0.1880679875612259, "seconds": 5755.5} {"step": 1154, "loss": 0.27358120679855347, "data": 0.11581455171108246, "gap": 0.1577666699886322, "sigma": 0.6519041657447815, "kind": "mosaic", "tokens": 202, "dense": 448, "grad_norm": 0.13458126783370972, "seconds": 5760.0} {"step": 1155, "loss": 0.299772173166275, "data": 0.299772173166275, "gap": null, "sigma": 0.6639820337295532, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.11490924656391144, "seconds": 5762.9} {"step": 1156, "loss": 0.43857330083847046, "data": 0.43857330083847046, "gap": null, "sigma": 0.8446211218833923, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2274121791124344, "seconds": 5765.8} {"step": 1157, "loss": 0.37797871232032776, "data": 0.20106098055839539, "gap": 0.17691773176193237, "sigma": 0.4427777826786041, "kind": "mosaic", "tokens": 259, "dense": 448, "grad_norm": 0.17230160534381866, "seconds": 5770.4} {"step": 1158, "loss": 0.4956088066101074, "data": 0.2667941749095917, "gap": 0.22881464660167694, "sigma": 0.960012674331665, "kind": "mosaic", "tokens": 169, "dense": 448, "grad_norm": 0.41051238775253296, "seconds": 5774.9} {"step": 1159, "loss": 0.5138607621192932, "data": 0.1728602796792984, "gap": 0.341000497341156, "sigma": 0.919021725654602, "kind": "mosaic", "tokens": 145, "dense": 448, "grad_norm": 0.3315127193927765, "seconds": 5779.1} {"step": 1160, "loss": 1.1627037525177002, "data": 0.18577779829502106, "gap": 0.9769259095191956, "sigma": 0.9726966023445129, "kind": "mosaic", "tokens": 205, "dense": 448, "grad_norm": 0.7801617383956909, "seconds": 5783.6} {"step": 1161, "loss": 0.46351760625839233, "data": 0.46351760625839233, "gap": null, "sigma": 0.9776397347450256, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3752403259277344, "seconds": 5786.5} {"step": 1162, "loss": 0.6185255646705627, "data": 0.23776109516620636, "gap": 0.3807644844055176, "sigma": 0.5716795325279236, "kind": "mosaic", "tokens": 264, "dense": 448, "grad_norm": 0.20435155928134918, "seconds": 5791.1} {"step": 1163, "loss": 1.5935072898864746, "data": 0.5058942437171936, "gap": 1.0876131057739258, "sigma": 0.20365266501903534, "kind": "mosaic", "tokens": 229, "dense": 448, "grad_norm": 0.49796339869499207, "seconds": 5795.3} {"step": 1164, "loss": 0.3556889295578003, "data": 0.3556889295578003, "gap": null, "sigma": 0.27626505494117737, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.06753575056791306, "seconds": 5798.2} {"step": 1165, "loss": 0.46095961332321167, "data": 0.46095961332321167, "gap": null, "sigma": 0.9886307716369629, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3733719289302826, "seconds": 5801.0} {"step": 1166, "loss": 0.30258771777153015, "data": 0.11810337007045746, "gap": 0.1844843477010727, "sigma": 0.7034448981285095, "kind": "mosaic", "tokens": 213, "dense": 448, "grad_norm": 0.2271318882703781, "seconds": 5805.4} {"step": 1167, "loss": 0.24017637968063354, "data": 0.127549409866333, "gap": 0.11262697726488113, "sigma": 0.4983237385749817, "kind": "mosaic", "tokens": 276, "dense": 448, "grad_norm": 0.16664843261241913, "seconds": 5810.1} {"step": 1168, "loss": 1.0152910947799683, "data": 0.30346792936325073, "gap": 0.7118231654167175, "sigma": 0.9236355423927307, "kind": "mosaic", "tokens": 268, "dense": 448, "grad_norm": 0.35348325967788696, "seconds": 5814.6} {"step": 1169, "loss": 0.5114656686782837, "data": 0.21924282610416412, "gap": 0.2922228276729584, "sigma": 0.573508083820343, "kind": "mosaic", "tokens": 259, "dense": 448, "grad_norm": 0.1855572611093521, "seconds": 5819.0} {"step": 1170, "loss": 0.4435794949531555, "data": 0.4435794949531555, "gap": null, "sigma": 0.9967042803764343, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.44744589924812317, "seconds": 5821.8} {"step": 1171, "loss": 0.4576120376586914, "data": 0.2176201045513153, "gap": 0.2399919331073761, "sigma": 0.5981383323669434, "kind": "mosaic", "tokens": 250, "dense": 448, "grad_norm": 0.1227702647447586, "seconds": 5826.2} {"step": 1172, "loss": 1.0484079122543335, "data": 0.36422795057296753, "gap": 0.684179961681366, "sigma": 0.9952342510223389, "kind": "mosaic", "tokens": 208, "dense": 448, "grad_norm": 0.5672023296356201, "seconds": 5830.7} {"step": 1173, "loss": 0.4750683903694153, "data": 0.23098868131637573, "gap": 0.24407969415187836, "sigma": 0.7160871624946594, "kind": "mosaic", "tokens": 221, "dense": 448, "grad_norm": 0.22617536783218384, "seconds": 5835.1} {"step": 1174, "loss": 0.3219832479953766, "data": 0.3219832479953766, "gap": null, "sigma": 0.22987347841262817, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.07507602870464325, "seconds": 5838.0} {"step": 1175, "loss": 0.7093007564544678, "data": 0.231230691075325, "gap": 0.47807005047798157, "sigma": 0.5397286415100098, "kind": "mosaic", "tokens": 171, "dense": 448, "grad_norm": 0.3972438871860504, "seconds": 5842.3} {"step": 1176, "loss": 0.4315584897994995, "data": 0.4315584897994995, "gap": null, "sigma": 0.8802674412727356, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.24490666389465332, "seconds": 5845.1} {"step": 1177, "loss": 0.9415217638015747, "data": 0.32756510376930237, "gap": 0.61395663022995, "sigma": 0.9928057789802551, "kind": "mosaic", "tokens": 178, "dense": 448, "grad_norm": 0.733924925327301, "seconds": 5849.7} {"step": 1178, "loss": 1.7931865453720093, "data": 0.4584563970565796, "gap": 1.3347301483154297, "sigma": 0.24180485308170319, "kind": "mosaic", "tokens": 173, "dense": 448, "grad_norm": 0.5254939198493958, "seconds": 5854.3} {"step": 1179, "loss": 0.5434774160385132, "data": 0.22606195509433746, "gap": 0.3174154460430145, "sigma": 0.7234876751899719, "kind": "mosaic", "tokens": 221, "dense": 448, "grad_norm": 0.22462132573127747, "seconds": 5858.7} {"step": 1180, "loss": 0.5169233083724976, "data": 0.2342739850282669, "gap": 0.28264933824539185, "sigma": 0.5327479243278503, "kind": "mosaic", "tokens": 266, "dense": 448, "grad_norm": 0.15813039243221283, "seconds": 5863.2} {"step": 1181, "loss": 0.3620091378688812, "data": 0.3620091378688812, "gap": null, "sigma": 0.9467616677284241, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2018948197364807, "seconds": 5866.0} {"step": 1182, "loss": 30.490861892700195, "data": 5.601978302001953, "gap": 24.888883590698242, "sigma": 0.04592099413275719, "kind": "mosaic", "tokens": 190, "dense": 448, "grad_norm": 16.84977912902832, "seconds": 5870.5} {"step": 1183, "loss": 1.5475939512252808, "data": 0.14233674108982086, "gap": 1.405257225036621, "sigma": 0.9777448773384094, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 2.2149503231048584, "seconds": 5874.6} {"step": 1184, "loss": 0.9184987545013428, "data": 0.2585442364215851, "gap": 0.6599544882774353, "sigma": 0.38265296816825867, "kind": "mosaic", "tokens": 205, "dense": 448, "grad_norm": 0.2550915777683258, "seconds": 5879.3} {"step": 1185, "loss": 0.5444692969322205, "data": 0.05806489288806915, "gap": 0.4864043891429901, "sigma": 0.9512425065040588, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 1.8882287740707397, "seconds": 5883.3} {"step": 1186, "loss": 0.6003530025482178, "data": 0.2414599359035492, "gap": 0.35889309644699097, "sigma": 0.429592102766037, "kind": "mosaic", "tokens": 199, "dense": 448, "grad_norm": 0.20700526237487793, "seconds": 5887.8} {"step": 1187, "loss": 25.0596923828125, "data": 3.886491298675537, "gap": 21.173200607299805, "sigma": 0.01982817053794861, "kind": "mosaic", "tokens": 141, "dense": 448, "grad_norm": 14.873600959777832, "seconds": 5892.0} {"step": 1188, "loss": 0.38105735182762146, "data": 0.38105735182762146, "gap": null, "sigma": 0.9306297302246094, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.4731261730194092, "seconds": 5894.8} {"step": 1189, "loss": 0.4727679193019867, "data": 0.4727679193019867, "gap": null, "sigma": 0.9645923972129822, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.22287535667419434, "seconds": 5897.6} {"step": 1190, "loss": 1.0174700021743774, "data": 0.2920369803905487, "gap": 0.7254329919815063, "sigma": 0.9679778218269348, "kind": "mosaic", "tokens": 234, "dense": 448, "grad_norm": 0.6699033975601196, "seconds": 5902.1} {"step": 1191, "loss": 12.120241165161133, "data": 4.22196102142334, "gap": 7.898280143737793, "sigma": 0.045917339622974396, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 16.844388961791992, "seconds": 5906.5} {"step": 1192, "loss": 0.5281602144241333, "data": 0.5281602144241333, "gap": null, "sigma": 0.9789389967918396, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.35137632489204407, "seconds": 5909.3} {"step": 1193, "loss": 0.48556506633758545, "data": 0.1583787351846695, "gap": 0.32718634605407715, "sigma": 0.7235040664672852, "kind": "mosaic", "tokens": 185, "dense": 448, "grad_norm": 0.429992139339447, "seconds": 5913.9} {"step": 1194, "loss": 0.38914355635643005, "data": 0.05983507260680199, "gap": 0.32930848002433777, "sigma": 0.7650658488273621, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.35954636335372925, "seconds": 5918.0} {"step": 1195, "loss": 1.07766592502594, "data": 0.28376495838165283, "gap": 0.7939009666442871, "sigma": 0.9054263830184937, "kind": "mosaic", "tokens": 236, "dense": 448, "grad_norm": 0.46740928292274475, "seconds": 5922.4} {"step": 1196, "loss": 0.8285946846008301, "data": 0.2832973301410675, "gap": 0.5452973246574402, "sigma": 0.4855693578720093, "kind": "mosaic", "tokens": 191, "dense": 448, "grad_norm": 0.3204786777496338, "seconds": 5926.9} {"step": 1197, "loss": 0.2747468948364258, "data": 0.1050284281373024, "gap": 0.16971847414970398, "sigma": 0.7709569334983826, "kind": "mosaic", "tokens": 247, "dense": 448, "grad_norm": 0.21229103207588196, "seconds": 5931.5} {"step": 1198, "loss": 0.43188875913619995, "data": 0.43188875913619995, "gap": null, "sigma": 0.9485855102539062, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.16165465116500854, "seconds": 5934.2} {"step": 1199, "loss": 0.2970306873321533, "data": 0.2970306873321533, "gap": null, "sigma": 0.8690032362937927, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2792714238166809, "seconds": 5937.0} {"step": 1200, "loss": 0.4695337414741516, "data": 0.1877894252538681, "gap": 0.2817443311214447, "sigma": 0.8085001707077026, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.35450899600982666, "seconds": 5941.6} {"step": 1201, "loss": 0.472847580909729, "data": 0.27130669355392456, "gap": 0.20154087245464325, "sigma": 0.8797296285629272, "kind": "mosaic", "tokens": 254, "dense": 448, "grad_norm": 0.1872125118970871, "seconds": 5946.1} {"step": 1202, "loss": 12.26253890991211, "data": 1.9945635795593262, "gap": 10.267974853515625, "sigma": 0.013523386791348457, "kind": "mosaic", "tokens": 214, "dense": 448, "grad_norm": 9.534131050109863, "seconds": 5950.6} {"step": 1203, "loss": 1.202975869178772, "data": 0.33815035223960876, "gap": 0.8648255467414856, "sigma": 0.303699254989624, "kind": "mosaic", "tokens": 204, "dense": 448, "grad_norm": 0.4569113552570343, "seconds": 5955.1} {"step": 1204, "loss": 0.6300504803657532, "data": 0.2712266743183136, "gap": 0.3588238060474396, "sigma": 0.8326717615127563, "kind": "mosaic", "tokens": 257, "dense": 448, "grad_norm": 0.44411081075668335, "seconds": 5959.6} {"step": 1205, "loss": 0.4270155727863312, "data": 0.16888055205345154, "gap": 0.25813502073287964, "sigma": 0.8912881016731262, "kind": "mosaic", "tokens": 171, "dense": 448, "grad_norm": 0.25263887643814087, "seconds": 5964.0} {"step": 1206, "loss": 1.640791893005371, "data": 0.4428757429122925, "gap": 1.1979161500930786, "sigma": 0.9953187704086304, "kind": "mosaic", "tokens": 216, "dense": 448, "grad_norm": 2.0872442722320557, "seconds": 5968.6} {"step": 1207, "loss": 2.5787081718444824, "data": 0.5306673049926758, "gap": 2.0480408668518066, "sigma": 0.18244747817516327, "kind": "mosaic", "tokens": 181, "dense": 448, "grad_norm": 1.164306640625, "seconds": 5973.0} {"step": 1208, "loss": 0.2560201585292816, "data": 0.10661514848470688, "gap": 0.14940500259399414, "sigma": 0.7640351057052612, "kind": "mosaic", "tokens": 194, "dense": 448, "grad_norm": 0.11451786756515503, "seconds": 5977.4} {"step": 1209, "loss": 0.6322422027587891, "data": 0.20728234946727753, "gap": 0.42495986819267273, "sigma": 0.911945104598999, "kind": "mosaic", "tokens": 199, "dense": 448, "grad_norm": 0.4067329168319702, "seconds": 5981.8} {"step": 1210, "loss": 0.2822720408439636, "data": 0.13437700271606445, "gap": 0.14789505302906036, "sigma": 0.7864701747894287, "kind": "mosaic", "tokens": 171, "dense": 448, "grad_norm": 0.17077498137950897, "seconds": 5986.1} {"step": 1211, "loss": 0.74405837059021, "data": 0.2884061634540558, "gap": 0.45565223693847656, "sigma": 0.2889105975627899, "kind": "mosaic", "tokens": 223, "dense": 448, "grad_norm": 0.312376469373703, "seconds": 5990.6} {"step": 1212, "loss": 7.528404235839844, "data": 1.7256624698638916, "gap": 5.802741527557373, "sigma": 0.005660830996930599, "kind": "mosaic", "tokens": 242, "dense": 448, "grad_norm": 6.594488620758057, "seconds": 5995.0} {"step": 1213, "loss": 17.84762954711914, "data": 3.762418508529663, "gap": 14.085210800170898, "sigma": 0.0376635417342186, "kind": "mosaic", "tokens": 243, "dense": 448, "grad_norm": 10.43154239654541, "seconds": 5999.6} {"step": 1214, "loss": 1.1241153478622437, "data": 0.31389978528022766, "gap": 0.8102155923843384, "sigma": 0.23924116790294647, "kind": "mosaic", "tokens": 190, "dense": 448, "grad_norm": 0.39487186074256897, "seconds": 6004.1} {"step": 1215, "loss": 0.31855839490890503, "data": 0.09776362776756287, "gap": 0.22079478204250336, "sigma": 0.854988694190979, "kind": "mosaic", "tokens": 219, "dense": 448, "grad_norm": 0.26039522886276245, "seconds": 6008.6} {"step": 1216, "loss": 0.4347403049468994, "data": 0.21508745849132538, "gap": 0.21965286135673523, "sigma": 0.664432168006897, "kind": "mosaic", "tokens": 244, "dense": 512, "grad_norm": 0.1280934065580368, "seconds": 6013.2} {"step": 1217, "loss": 0.6671208143234253, "data": 0.07413139194250107, "gap": 0.592989444732666, "sigma": 0.9170433282852173, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.4475281536579132, "seconds": 6017.4} {"step": 1218, "loss": 1.0722390413284302, "data": 0.2542558014392853, "gap": 0.8179832100868225, "sigma": 0.9959109425544739, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 1.1774975061416626, "seconds": 6021.5} {"step": 1219, "loss": 0.9652119874954224, "data": 0.2649802267551422, "gap": 0.7002317309379578, "sigma": 0.3456399142742157, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.5472096800804138, "seconds": 6025.8} {"step": 1220, "loss": 0.4400928020477295, "data": 0.15582822263240814, "gap": 0.28426459431648254, "sigma": 0.7451555728912354, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.19607888162136078, "seconds": 6030.0} {"step": 1221, "loss": 0.48139500617980957, "data": 0.2093213051557541, "gap": 0.2720736861228943, "sigma": 0.777519941329956, "kind": "mosaic", "tokens": 268, "dense": 448, "grad_norm": 0.4257563352584839, "seconds": 6034.6} {"step": 1222, "loss": 0.3630562722682953, "data": 0.3630562722682953, "gap": null, "sigma": 0.9453775882720947, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.12962602078914642, "seconds": 6037.4} {"step": 1223, "loss": 0.6107989549636841, "data": 0.10327843576669693, "gap": 0.5075204968452454, "sigma": 0.6393275260925293, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.2691779136657715, "seconds": 6041.6} {"step": 1224, "loss": 0.8833929300308228, "data": 0.32331275939941406, "gap": 0.5600801706314087, "sigma": 0.27208518981933594, "kind": "mosaic", "tokens": 193, "dense": 448, "grad_norm": 0.30327239632606506, "seconds": 6046.2} {"step": 1225, "loss": 1.4883222579956055, "data": 0.25668638944625854, "gap": 1.2316359281539917, "sigma": 0.9140727519989014, "kind": "mosaic", "tokens": 200, "dense": 448, "grad_norm": 0.5164814591407776, "seconds": 6050.7} {"step": 1226, "loss": 0.44967421889305115, "data": 0.44967421889305115, "gap": null, "sigma": 0.9736781716346741, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3028452396392822, "seconds": 6053.5} {"step": 1227, "loss": 0.6491806507110596, "data": 0.20231519639492035, "gap": 0.446865439414978, "sigma": 0.8440452814102173, "kind": "mosaic", "tokens": 204, "dense": 448, "grad_norm": 0.4189724624156952, "seconds": 6058.0} {"step": 1228, "loss": 0.49774065613746643, "data": 0.03743300214409828, "gap": 0.46030765771865845, "sigma": 0.9102482795715332, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 1.1122568845748901, "seconds": 6062.2} {"step": 1229, "loss": 0.785446047782898, "data": 0.21160180866718292, "gap": 0.5738442540168762, "sigma": 0.9405650496482849, "kind": "mosaic", "tokens": 188, "dense": 448, "grad_norm": 0.7664144039154053, "seconds": 6066.6} {"step": 1230, "loss": 0.5876821279525757, "data": 0.2461465299129486, "gap": 0.3415355682373047, "sigma": 0.7780280709266663, "kind": "mosaic", "tokens": 206, "dense": 448, "grad_norm": 0.16200454533100128, "seconds": 6071.0} {"step": 1231, "loss": 0.3940000534057617, "data": 0.3940000534057617, "gap": null, "sigma": 0.9878641366958618, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.34680265188217163, "seconds": 6073.8} {"step": 1232, "loss": 0.33370041847229004, "data": 0.13757814466953278, "gap": 0.19612228870391846, "sigma": 0.7393891215324402, "kind": "mosaic", "tokens": 198, "dense": 448, "grad_norm": 0.18840302526950836, "seconds": 6078.2} {"step": 1233, "loss": 0.3439218997955322, "data": 0.3439218997955322, "gap": null, "sigma": 0.6412743330001831, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3368954658508301, "seconds": 6081.0} {"step": 1234, "loss": 1.311041235923767, "data": 0.2601769268512726, "gap": 1.050864338874817, "sigma": 0.8933722972869873, "kind": "mosaic", "tokens": 222, "dense": 448, "grad_norm": 1.3748635053634644, "seconds": 6085.7} {"step": 1235, "loss": 0.9080632328987122, "data": 0.32958298921585083, "gap": 0.5784802436828613, "sigma": 0.3464507758617401, "kind": "mosaic", "tokens": 300, "dense": 576, "grad_norm": 0.28052276372909546, "seconds": 6090.3} {"step": 1236, "loss": 3.4219655990600586, "data": 0.6932103633880615, "gap": 2.728755235671997, "sigma": 0.15808779001235962, "kind": "mosaic", "tokens": 187, "dense": 448, "grad_norm": 1.2891496419906616, "seconds": 6094.9} {"step": 1237, "loss": 0.3931904733181, "data": 0.3931904733181, "gap": null, "sigma": 0.9117760062217712, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.22181499004364014, "seconds": 6097.7} {"step": 1238, "loss": 1.2755391597747803, "data": 0.3566509485244751, "gap": 0.9188882112503052, "sigma": 0.967670738697052, "kind": "mosaic", "tokens": 278, "dense": 448, "grad_norm": 1.6226859092712402, "seconds": 6102.4} {"step": 1239, "loss": 0.5393623113632202, "data": 0.14230996370315552, "gap": 0.3970523178577423, "sigma": 0.8127281665802002, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.30198055505752563, "seconds": 6106.6} {"step": 1240, "loss": 1.3542723655700684, "data": 0.19890935719013214, "gap": 1.1553629636764526, "sigma": 0.8742634057998657, "kind": "mosaic", "tokens": 241, "dense": 448, "grad_norm": 0.9429448843002319, "seconds": 6111.1} {"step": 1241, "loss": 0.5950388312339783, "data": 0.2205740213394165, "gap": 0.37446480989456177, "sigma": 0.6897470355033875, "kind": "mosaic", "tokens": 224, "dense": 448, "grad_norm": 0.21030648052692413, "seconds": 6115.6} {"step": 1242, "loss": 0.7014414072036743, "data": 0.09073686599731445, "gap": 0.6107045412063599, "sigma": 0.8399406671524048, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.44460031390190125, "seconds": 6119.9} {"step": 1243, "loss": 0.4421716630458832, "data": 0.1832658350467682, "gap": 0.258905827999115, "sigma": 0.928512692451477, "kind": "mosaic", "tokens": 239, "dense": 448, "grad_norm": 0.24304470419883728, "seconds": 6124.3} {"step": 1244, "loss": 0.6914880275726318, "data": 0.2514975368976593, "gap": 0.43999046087265015, "sigma": 0.7944323420524597, "kind": "mosaic", "tokens": 189, "dense": 448, "grad_norm": 0.22786206007003784, "seconds": 6129.0} {"step": 1245, "loss": 0.7622337341308594, "data": 0.2817728519439697, "gap": 0.48046085238456726, "sigma": 0.47749415040016174, "kind": "mosaic", "tokens": 240, "dense": 448, "grad_norm": 0.17427588999271393, "seconds": 6133.5} {"step": 1246, "loss": 0.35544854402542114, "data": 0.14549589157104492, "gap": 0.20995266735553741, "sigma": 0.8579429984092712, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.2325521856546402, "seconds": 6137.8} {"step": 1247, "loss": 0.5015722513198853, "data": 0.2551080584526062, "gap": 0.24646417796611786, "sigma": 0.710247278213501, "kind": "mosaic", "tokens": 252, "dense": 448, "grad_norm": 0.1959594339132309, "seconds": 6142.4} {"step": 1248, "loss": 18.28800392150879, "data": 3.593991279602051, "gap": 14.694012641906738, "sigma": 0.04035298153758049, "kind": "mosaic", "tokens": 204, "dense": 448, "grad_norm": 11.58227252960205, "seconds": 6146.9} {"step": 1249, "loss": 0.4787887930870056, "data": 0.1807546615600586, "gap": 0.298034131526947, "sigma": 0.5723892450332642, "kind": "mosaic", "tokens": 248, "dense": 448, "grad_norm": 0.15256236493587494, "seconds": 6151.6} {"step": 1250, "loss": 0.30611950159072876, "data": 0.17269261181354523, "gap": 0.13342690467834473, "sigma": 0.6176990270614624, "kind": "mosaic", "tokens": 271, "dense": 448, "grad_norm": 0.12208734452724457, "seconds": 6156.2} {"step": 1250, "eval": {"dense": 0.317887743934989, "uniform2": 0.18828316343327364, "mosaic": 0.2566743604838848, "gap_uniform2": 0.5832831803709269, "gap_mosaic": 0.5825953626384338}, "eval_extent": {"1x1x1": 0.4261193123709529, "1x1x2": 0.32911722833523527, "1x1x4": 0.23377780759249603, "1x2x1": 0.3101658871794051, "1x2x2": 0.19879875507352776, "1x2x4": 0.13109872068290584, "1x4x1": 0.20103961090692438, "1x4x2": 0.1170429069203911, "1x4x4": 0.08371990975879487}} {"step": 1251, "loss": 0.6948075890541077, "data": 0.19549478590488434, "gap": 0.49931278824806213, "sigma": 0.4552381634712219, "kind": "mosaic", "tokens": 154, "dense": 448, "grad_norm": 0.18718256056308746, "seconds": 6403.6} {"step": 1252, "loss": 0.44078654050827026, "data": 0.2272522896528244, "gap": 0.21353425085544586, "sigma": 0.9018217325210571, "kind": "mosaic", "tokens": 257, "dense": 448, "grad_norm": 0.5533727407455444, "seconds": 6408.3} {"step": 1253, "loss": 0.5236278772354126, "data": 0.5236278772354126, "gap": null, "sigma": 0.9649940133094788, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.32579562067985535, "seconds": 6411.2} {"step": 1254, "loss": 1.006080150604248, "data": 0.3410835266113281, "gap": 0.6649965643882751, "sigma": 0.3282422125339508, "kind": "mosaic", "tokens": 179, "dense": 448, "grad_norm": 0.23354746401309967, "seconds": 6415.7} {"step": 1255, "loss": 0.44054532051086426, "data": 0.44054532051086426, "gap": null, "sigma": 0.9981684684753418, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.5774902105331421, "seconds": 6418.7} {"step": 1256, "loss": 0.45923566818237305, "data": 0.19714964926242828, "gap": 0.26208603382110596, "sigma": 0.6614478826522827, "kind": "mosaic", "tokens": 269, "dense": 448, "grad_norm": 0.18460294604301453, "seconds": 6423.4} {"step": 1257, "loss": 26.292827606201172, "data": 4.850458145141602, "gap": 21.44236946105957, "sigma": 0.03799961134791374, "kind": "mosaic", "tokens": 207, "dense": 448, "grad_norm": 16.4770565032959, "seconds": 6427.9} {"step": 1258, "loss": 0.5079457759857178, "data": 0.24791479110717773, "gap": 0.26003095507621765, "sigma": 0.746726393699646, "kind": "mosaic", "tokens": 233, "dense": 448, "grad_norm": 0.18724019825458527, "seconds": 6432.4} {"step": 1259, "loss": 0.636534571647644, "data": 0.21444593369960785, "gap": 0.422088623046875, "sigma": 0.4098154306411743, "kind": "mosaic", "tokens": 157, "dense": 448, "grad_norm": 0.17958053946495056, "seconds": 6436.7} {"step": 1260, "loss": 0.6492305994033813, "data": 0.1790204644203186, "gap": 0.47021013498306274, "sigma": 0.789652943611145, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.29932984709739685, "seconds": 6441.0} {"step": 1261, "loss": 0.36334580183029175, "data": 0.15588320791721344, "gap": 0.2074626088142395, "sigma": 0.5460543632507324, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.5705715417861938, "seconds": 6445.5} {"step": 1262, "loss": 0.50497967004776, "data": 0.50497967004776, "gap": null, "sigma": 0.9770143628120422, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3003907799720764, "seconds": 6448.3} {"step": 1263, "loss": 1.1255805492401123, "data": 0.24988079071044922, "gap": 0.8756998181343079, "sigma": 0.9138520956039429, "kind": "mosaic", "tokens": 187, "dense": 448, "grad_norm": 0.8471600413322449, "seconds": 6452.9} {"step": 1264, "loss": 0.5107576847076416, "data": 0.2717616856098175, "gap": 0.23899602890014648, "sigma": 0.948460042476654, "kind": "mosaic", "tokens": 250, "dense": 448, "grad_norm": 0.38474947214126587, "seconds": 6457.4} {"step": 1265, "loss": 0.499345064163208, "data": 0.16183827817440033, "gap": 0.33750680088996887, "sigma": 0.7704138159751892, "kind": "mosaic", "tokens": 191, "dense": 448, "grad_norm": 0.27163171768188477, "seconds": 6461.8} {"step": 1266, "loss": 0.5163990259170532, "data": 0.19167396426200867, "gap": 0.32472503185272217, "sigma": 0.5216159820556641, "kind": "mosaic", "tokens": 193, "dense": 448, "grad_norm": 0.16761094331741333, "seconds": 6466.1} {"step": 1267, "loss": 0.3463433086872101, "data": 0.22510266304016113, "gap": 0.12124063819646835, "sigma": 0.7404614090919495, "kind": "mosaic", "tokens": 269, "dense": 448, "grad_norm": 0.2056792825460434, "seconds": 6470.7} {"step": 1268, "loss": 0.9694433808326721, "data": 0.4140739440917969, "gap": 0.5553694367408752, "sigma": 0.9937536120414734, "kind": "mosaic", "tokens": 192, "dense": 448, "grad_norm": 0.7378539443016052, "seconds": 6475.1} {"step": 1269, "loss": 0.20561712980270386, "data": 0.20561712980270386, "gap": null, "sigma": 0.9539848566055298, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.20806987583637238, "seconds": 6478.0} {"step": 1270, "loss": 0.4132592976093292, "data": 0.15190324187278748, "gap": 0.26135605573654175, "sigma": 0.6317962408065796, "kind": "mosaic", "tokens": 208, "dense": 448, "grad_norm": 0.21544408798217773, "seconds": 6482.6} {"step": 1271, "loss": 0.46384134888648987, "data": 0.18663200736045837, "gap": 0.2772093415260315, "sigma": 0.7132688164710999, "kind": "mosaic", "tokens": 228, "dense": 448, "grad_norm": 0.34610745310783386, "seconds": 6487.1} {"step": 1272, "loss": 0.5186699628829956, "data": 0.5186699628829956, "gap": null, "sigma": 0.9613469839096069, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.4127104580402374, "seconds": 6489.9} {"step": 1273, "loss": 0.17782722413539886, "data": 0.09564156830310822, "gap": 0.08218565583229065, "sigma": 0.6758443713188171, "kind": "mosaic", "tokens": 243, "dense": 448, "grad_norm": 0.5909442901611328, "seconds": 6494.5} {"step": 1274, "loss": 0.4178018271923065, "data": 0.4178018271923065, "gap": null, "sigma": 0.8716690540313721, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.1833704710006714, "seconds": 6497.4} {"step": 1275, "loss": 0.4214552044868469, "data": 0.15005123615264893, "gap": 0.271403968334198, "sigma": 0.6608511805534363, "kind": "mosaic", "tokens": 205, "dense": 448, "grad_norm": 0.1845911145210266, "seconds": 6501.9} {"step": 1276, "loss": 26.847259521484375, "data": 5.483437538146973, "gap": 21.363821029663086, "sigma": 0.047425899654626846, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 15.599447250366211, "seconds": 6506.2} {"step": 1277, "loss": 0.6870579719543457, "data": 0.2911421060562134, "gap": 0.3959158658981323, "sigma": 0.29159975051879883, "kind": "mosaic", "tokens": 234, "dense": 448, "grad_norm": 0.26195597648620605, "seconds": 6510.8} {"step": 1278, "loss": 0.5432636141777039, "data": 0.22732758522033691, "gap": 0.31593602895736694, "sigma": 0.3771766424179077, "kind": "mosaic", "tokens": 225, "dense": 448, "grad_norm": 0.17623387277126312, "seconds": 6515.4} {"step": 1279, "loss": 0.5979784727096558, "data": 0.22665898501873016, "gap": 0.3713195025920868, "sigma": 0.7043400406837463, "kind": "mosaic", "tokens": 185, "dense": 448, "grad_norm": 0.21211472153663635, "seconds": 6519.9} {"step": 1280, "loss": 0.649121880531311, "data": 0.2883909046649933, "gap": 0.36073094606399536, "sigma": 0.9318294525146484, "kind": "mosaic", "tokens": 167, "dense": 448, "grad_norm": 0.60417640209198, "seconds": 6524.4} {"step": 1281, "loss": 5.227969169616699, "data": 0.74046391248703, "gap": 4.4875054359436035, "sigma": 0.164638489484787, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 2.008213996887207, "seconds": 6528.4} {"step": 1282, "loss": 0.2941339612007141, "data": 0.1538342982530594, "gap": 0.14029966294765472, "sigma": 0.5266886353492737, "kind": "mosaic", "tokens": 250, "dense": 448, "grad_norm": 0.15183953940868378, "seconds": 6533.0} {"step": 1283, "loss": 0.44915032386779785, "data": 0.20646511018276215, "gap": 0.2426852136850357, "sigma": 0.627489447593689, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.1610833704471588, "seconds": 6537.4} {"step": 1284, "loss": 0.3122981786727905, "data": 0.3122981786727905, "gap": null, "sigma": 0.6573240160942078, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.10742305964231491, "seconds": 6540.2} {"step": 1285, "loss": 0.3151891231536865, "data": 0.14388197660446167, "gap": 0.17130714654922485, "sigma": 0.542468249797821, "kind": "mosaic", "tokens": 198, "dense": 448, "grad_norm": 0.1420496106147766, "seconds": 6544.8} {"step": 1286, "loss": 0.2532482147216797, "data": 0.2532482147216797, "gap": null, "sigma": 0.9222472310066223, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.24749509990215302, "seconds": 6547.5} {"step": 1287, "loss": 0.5842273235321045, "data": 0.17359404265880585, "gap": 0.41063326597213745, "sigma": 0.5159069299697876, "kind": "mosaic", "tokens": 201, "dense": 448, "grad_norm": 0.24514789879322052, "seconds": 6552.0} {"step": 1288, "loss": 1.0259613990783691, "data": 0.19913047552108765, "gap": 0.8268309235572815, "sigma": 0.9514344930648804, "kind": "mosaic", "tokens": 190, "dense": 448, "grad_norm": 1.146267056465149, "seconds": 6556.3} {"step": 1289, "loss": 0.4346361756324768, "data": 0.2260805368423462, "gap": 0.20855562388896942, "sigma": 0.9518154263496399, "kind": "mosaic", "tokens": 222, "dense": 448, "grad_norm": 0.27045559883117676, "seconds": 6561.0} {"step": 1290, "loss": 0.45710521936416626, "data": 0.06357155740261078, "gap": 0.3935336470603943, "sigma": 0.9899908900260925, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 0.9200772643089294, "seconds": 6565.1} {"step": 1291, "loss": 0.4202639162540436, "data": 0.16152077913284302, "gap": 0.25874313712120056, "sigma": 0.56560218334198, "kind": "mosaic", "tokens": 232, "dense": 448, "grad_norm": 0.17224127054214478, "seconds": 6569.6} {"step": 1292, "loss": 0.8482611775398254, "data": 0.31088292598724365, "gap": 0.5373782515525818, "sigma": 0.30924689769744873, "kind": "mosaic", "tokens": 264, "dense": 448, "grad_norm": 0.2434358447790146, "seconds": 6574.3} {"step": 1293, "loss": 0.4834918677806854, "data": 0.28688156604766846, "gap": 0.19661030173301697, "sigma": 0.9180749654769897, "kind": "mosaic", "tokens": 201, "dense": 448, "grad_norm": 0.3442705273628235, "seconds": 6578.7} {"step": 1294, "loss": 0.39655032753944397, "data": 0.39655032753944397, "gap": null, "sigma": 0.2915310263633728, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.5301505327224731, "seconds": 6581.5} {"step": 1295, "loss": 0.3225627541542053, "data": 0.3225627541542053, "gap": null, "sigma": 0.9394508600234985, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.25075745582580566, "seconds": 6584.4} {"step": 1296, "loss": 0.6234104633331299, "data": 0.2665548622608185, "gap": 0.356855571269989, "sigma": 0.9360342025756836, "kind": "mosaic", "tokens": 219, "dense": 448, "grad_norm": 0.26108118891716003, "seconds": 6589.0} {"step": 1297, "loss": 0.41709619760513306, "data": 0.18746666610240936, "gap": 0.2296295464038849, "sigma": 0.8885848522186279, "kind": "mosaic", "tokens": 177, "dense": 448, "grad_norm": 0.25917142629623413, "seconds": 6593.5} {"step": 1298, "loss": 1.0494564771652222, "data": 0.3149811923503876, "gap": 0.734475314617157, "sigma": 0.2304140031337738, "kind": "mosaic", "tokens": 204, "dense": 448, "grad_norm": 0.5013070106506348, "seconds": 6597.9} {"step": 1299, "loss": 0.5745129585266113, "data": 0.21361206471920013, "gap": 0.3609009087085724, "sigma": 0.5092173218727112, "kind": "mosaic", "tokens": 207, "dense": 448, "grad_norm": 0.1878659576177597, "seconds": 6602.4} {"step": 1300, "loss": 0.3099791407585144, "data": 0.16640332341194153, "gap": 0.14357583224773407, "sigma": 0.6420443654060364, "kind": "mosaic", "tokens": 256, "dense": 448, "grad_norm": 0.198793426156044, "seconds": 6606.8} {"step": 1301, "loss": 0.2994893491268158, "data": 0.2994893491268158, "gap": null, "sigma": 0.914350152015686, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.19981515407562256, "seconds": 6609.8} {"step": 1302, "loss": 0.6637924909591675, "data": 0.22469162940979004, "gap": 0.43910089135169983, "sigma": 0.9564015865325928, "kind": "mosaic", "tokens": 169, "dense": 448, "grad_norm": 0.3959435522556305, "seconds": 6614.3} {"step": 1303, "loss": 0.5308856964111328, "data": 0.16688945889472961, "gap": 0.3639962673187256, "sigma": 0.7851558327674866, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.21332715451717377, "seconds": 6618.4} {"step": 1304, "loss": 0.7519024610519409, "data": 0.23966893553733826, "gap": 0.512233555316925, "sigma": 0.4039786159992218, "kind": "mosaic", "tokens": 241, "dense": 448, "grad_norm": 0.30447256565093994, "seconds": 6622.9} {"step": 1305, "loss": 0.6771277785301208, "data": 0.23329167068004608, "gap": 0.44383612275123596, "sigma": 0.89528888463974, "kind": "mosaic", "tokens": 276, "dense": 448, "grad_norm": 0.2868490219116211, "seconds": 6627.4} {"step": 1306, "loss": 0.5551354885101318, "data": 0.5551354885101318, "gap": null, "sigma": 0.984750509262085, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.4009217321872711, "seconds": 6630.3} {"step": 1307, "loss": 0.3191230893135071, "data": 0.14672166109085083, "gap": 0.17240141332149506, "sigma": 0.7491763830184937, "kind": "mosaic", "tokens": 221, "dense": 448, "grad_norm": 0.2721499502658844, "seconds": 6634.9} {"step": 1308, "loss": 0.4828453063964844, "data": 0.1485241800546646, "gap": 0.33432114124298096, "sigma": 0.3805551528930664, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.34868255257606506, "seconds": 6639.3} {"step": 1309, "loss": 0.48303860425949097, "data": 0.051772721111774445, "gap": 0.4312658905982971, "sigma": 0.9435943961143494, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 1.1202176809310913, "seconds": 6643.1} {"step": 1310, "loss": 1.9452106952667236, "data": 0.5023368000984192, "gap": 1.4428738355636597, "sigma": 0.17556315660476685, "kind": "mosaic", "tokens": 196, "dense": 448, "grad_norm": 1.0015034675598145, "seconds": 6647.6} {"step": 1311, "loss": 0.9511620998382568, "data": 0.2936520278453827, "gap": 0.6575100421905518, "sigma": 0.979694664478302, "kind": "mosaic", "tokens": 224, "dense": 448, "grad_norm": 0.5398383736610413, "seconds": 6652.1} {"step": 1312, "loss": 0.5595380067825317, "data": 0.21901294589042664, "gap": 0.3405250608921051, "sigma": 0.4331512749195099, "kind": "mosaic", "tokens": 198, "dense": 448, "grad_norm": 0.1664285510778427, "seconds": 6656.7} {"step": 1313, "loss": 1.6449166536331177, "data": 0.39505088329315186, "gap": 1.2498657703399658, "sigma": 0.2807530462741852, "kind": "mosaic", "tokens": 200, "dense": 448, "grad_norm": 0.4550210237503052, "seconds": 6661.3} {"step": 1314, "loss": 0.616104245185852, "data": 0.1125062108039856, "gap": 0.5035980343818665, "sigma": 0.9454535245895386, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.473024845123291, "seconds": 6665.6} {"step": 1315, "loss": 0.4847322702407837, "data": 0.1868617981672287, "gap": 0.2978704571723938, "sigma": 0.4862816333770752, "kind": "mosaic", "tokens": 174, "dense": 448, "grad_norm": 0.24828514456748962, "seconds": 6670.1} {"step": 1316, "loss": 0.3327028453350067, "data": 0.3327028453350067, "gap": null, "sigma": 0.6359722018241882, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.07145437598228455, "seconds": 6672.9} {"step": 1317, "loss": 0.22042538225650787, "data": 0.22042538225650787, "gap": null, "sigma": 0.8100459575653076, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.29199594259262085, "seconds": 6675.7} {"step": 1318, "loss": 2.6746609210968018, "data": 0.7612511515617371, "gap": 1.9134098291397095, "sigma": 0.161203071475029, "kind": "mosaic", "tokens": 232, "dense": 448, "grad_norm": 0.9457828402519226, "seconds": 6680.2} {"step": 1319, "loss": 0.36920690536499023, "data": 0.21977034211158752, "gap": 0.14943654835224152, "sigma": 0.815479576587677, "kind": "mosaic", "tokens": 241, "dense": 448, "grad_norm": 0.18485073745250702, "seconds": 6684.7} {"step": 1320, "loss": 0.489896297454834, "data": 0.2262297421693802, "gap": 0.263666570186615, "sigma": 0.8891128301620483, "kind": "mosaic", "tokens": 223, "dense": 448, "grad_norm": 0.31042203307151794, "seconds": 6689.3} {"step": 1321, "loss": 0.3793552815914154, "data": 0.3793552815914154, "gap": null, "sigma": 0.8451932668685913, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.09677451848983765, "seconds": 6692.1} {"step": 1322, "loss": 0.3809649348258972, "data": 0.19123874604701996, "gap": 0.18972620368003845, "sigma": 0.5741051435470581, "kind": "mosaic", "tokens": 287, "dense": 448, "grad_norm": 0.14056354761123657, "seconds": 6696.8} {"step": 1323, "loss": 0.6195443868637085, "data": 0.21422545611858368, "gap": 0.405318945646286, "sigma": 0.48178285360336304, "kind": "mosaic", "tokens": 218, "dense": 448, "grad_norm": 0.2781301438808441, "seconds": 6701.3} {"step": 1324, "loss": 0.49571794271469116, "data": 0.1826888918876648, "gap": 0.31302905082702637, "sigma": 0.7739997506141663, "kind": "mosaic", "tokens": 163, "dense": 448, "grad_norm": 0.19846364855766296, "seconds": 6705.9} {"step": 1325, "loss": 0.52834153175354, "data": 0.22336697578430176, "gap": 0.30497458577156067, "sigma": 0.7615875005722046, "kind": "mosaic", "tokens": 187, "dense": 448, "grad_norm": 0.18484145402908325, "seconds": 6710.4} {"step": 1326, "loss": 0.41722291707992554, "data": 0.16921775043010712, "gap": 0.24800515174865723, "sigma": 0.651889979839325, "kind": "mosaic", "tokens": 242, "dense": 448, "grad_norm": 0.1512114256620407, "seconds": 6715.0} {"step": 1327, "loss": 1.2667516469955444, "data": 0.5760520100593567, "gap": 0.6906996369361877, "sigma": 0.9845273494720459, "kind": "mosaic", "tokens": 199, "dense": 448, "grad_norm": 1.1847230195999146, "seconds": 6719.6} {"step": 1328, "loss": 2.337001323699951, "data": 0.5678730010986328, "gap": 1.769128441810608, "sigma": 0.19802382588386536, "kind": "mosaic", "tokens": 195, "dense": 448, "grad_norm": 0.7544717788696289, "seconds": 6724.1} {"step": 1329, "loss": 0.35216954350471497, "data": 0.0649276152253151, "gap": 0.28724193572998047, "sigma": 0.6470629572868347, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.20844408869743347, "seconds": 6728.3} {"step": 1330, "loss": 0.4399424195289612, "data": 0.4399424195289612, "gap": null, "sigma": 0.9345276355743408, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.16900886595249176, "seconds": 6731.1} {"step": 1331, "loss": 0.7713003158569336, "data": 0.204959437251091, "gap": 0.5663408637046814, "sigma": 0.928803563117981, "kind": "mosaic", "tokens": 235, "dense": 448, "grad_norm": 0.3406233787536621, "seconds": 6735.8} {"step": 1332, "loss": 0.4250951409339905, "data": 0.21721479296684265, "gap": 0.20788034796714783, "sigma": 0.9150250554084778, "kind": "mosaic", "tokens": 180, "dense": 448, "grad_norm": 0.18778178095817566, "seconds": 6740.3} {"step": 1333, "loss": 2.270880699157715, "data": 0.5195262432098389, "gap": 1.751354455947876, "sigma": 0.2008810192346573, "kind": "mosaic", "tokens": 213, "dense": 448, "grad_norm": 0.8067936897277832, "seconds": 6744.8} {"step": 1334, "loss": 0.7139082551002502, "data": 0.30777251720428467, "gap": 0.4061357378959656, "sigma": 0.9913612008094788, "kind": "mosaic", "tokens": 196, "dense": 448, "grad_norm": 0.7461036443710327, "seconds": 6749.2} {"step": 1335, "loss": 0.5135942697525024, "data": 0.1413363516330719, "gap": 0.37225788831710815, "sigma": 0.40805721282958984, "kind": "mosaic", "tokens": 173, "dense": 448, "grad_norm": 0.2616389989852905, "seconds": 6753.9} {"step": 1336, "loss": 0.7330893874168396, "data": 0.23841159045696259, "gap": 0.4946778118610382, "sigma": 0.5028476715087891, "kind": "mosaic", "tokens": 181, "dense": 448, "grad_norm": 0.17539817094802856, "seconds": 6758.4} {"step": 1337, "loss": 0.370927631855011, "data": 0.370927631855011, "gap": null, "sigma": 0.9715027213096619, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.22294770181179047, "seconds": 6761.2} {"step": 1338, "loss": 1.0033199787139893, "data": 0.3692147135734558, "gap": 0.6341052651405334, "sigma": 0.3099265992641449, "kind": "mosaic", "tokens": 260, "dense": 448, "grad_norm": 0.36117440462112427, "seconds": 6765.9} {"step": 1339, "loss": 0.6434587240219116, "data": 0.2726311981678009, "gap": 0.3708275258541107, "sigma": 0.6576107144355774, "kind": "mosaic", "tokens": 265, "dense": 448, "grad_norm": 0.33408066630363464, "seconds": 6770.6} {"step": 1340, "loss": 0.7291574478149414, "data": 0.28078919649124146, "gap": 0.44836825132369995, "sigma": 0.45101410150527954, "kind": "mosaic", "tokens": 251, "dense": 448, "grad_norm": 0.16640765964984894, "seconds": 6775.0} {"step": 1341, "loss": 0.5610836148262024, "data": 0.213157519698143, "gap": 0.3479260802268982, "sigma": 0.6565534472465515, "kind": "mosaic", "tokens": 197, "dense": 448, "grad_norm": 0.15024027228355408, "seconds": 6779.6} {"step": 1342, "loss": 0.5984879732131958, "data": 0.5984879732131958, "gap": null, "sigma": 0.9885547161102295, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.4850974678993225, "seconds": 6782.4} {"step": 1343, "loss": 0.6060903072357178, "data": 0.2474404275417328, "gap": 0.358649879693985, "sigma": 0.6424817442893982, "kind": "mosaic", "tokens": 209, "dense": 448, "grad_norm": 0.15867573022842407, "seconds": 6787.0} {"step": 1344, "loss": 3.1659440994262695, "data": 0.8794477581977844, "gap": 2.28649640083313, "sigma": 0.16077381372451782, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 1.8097813129425049, "seconds": 6791.1} {"step": 1345, "loss": 0.45339435338974, "data": 0.45339435338974, "gap": null, "sigma": 0.9694780707359314, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.1793207973241806, "seconds": 6794.0} {"step": 1346, "loss": 0.41397029161453247, "data": 0.17642882466316223, "gap": 0.23754145205020905, "sigma": 0.658919095993042, "kind": "mosaic", "tokens": 213, "dense": 448, "grad_norm": 0.17854520678520203, "seconds": 6798.6} {"step": 1347, "loss": 0.5579885244369507, "data": 0.1251646727323532, "gap": 0.4328238368034363, "sigma": 0.8867175579071045, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.3741181194782257, "seconds": 6802.8} {"step": 1348, "loss": 0.3976721167564392, "data": 0.21810507774353027, "gap": 0.17956703901290894, "sigma": 0.703881561756134, "kind": "mosaic", "tokens": 276, "dense": 448, "grad_norm": 0.1790294498205185, "seconds": 6807.3} {"step": 1349, "loss": 0.5951454043388367, "data": 0.3180483877658844, "gap": 0.27709701657295227, "sigma": 0.8844652771949768, "kind": "mosaic", "tokens": 244, "dense": 448, "grad_norm": 0.28969839215278625, "seconds": 6811.9} {"step": 1350, "loss": 4.1382036209106445, "data": 0.6023309230804443, "gap": 3.535872459411621, "sigma": 0.18623068928718567, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 4.400602340698242, "seconds": 6816.0} {"step": 1351, "loss": 1.4216554164886475, "data": 0.41519683599472046, "gap": 1.0064585208892822, "sigma": 0.21498149633407593, "kind": "mosaic", "tokens": 196, "dense": 448, "grad_norm": 0.5485637784004211, "seconds": 6820.6} {"step": 1352, "loss": 0.3023001551628113, "data": 0.3023001551628113, "gap": null, "sigma": 0.948209285736084, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.22915859520435333, "seconds": 6823.4} {"step": 1353, "loss": 0.5248367786407471, "data": 0.24753829836845398, "gap": 0.2772984802722931, "sigma": 0.8754170536994934, "kind": "mosaic", "tokens": 188, "dense": 448, "grad_norm": 0.21974919736385345, "seconds": 6828.0} {"step": 1354, "loss": 0.5803518295288086, "data": 0.32859280705451965, "gap": 0.25175905227661133, "sigma": 0.4697096347808838, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.1945771723985672, "seconds": 6832.5} {"step": 1355, "loss": 0.3062819540500641, "data": 0.3062819540500641, "gap": null, "sigma": 0.8878649473190308, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.14114071428775787, "seconds": 6835.3} {"step": 1356, "loss": 0.7205300331115723, "data": 0.13111349940299988, "gap": 0.58941650390625, "sigma": 0.5098729133605957, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.2709334194660187, "seconds": 6839.3} {"step": 1357, "loss": 1.4149153232574463, "data": 0.24730786681175232, "gap": 1.1676074266433716, "sigma": 0.9848139882087708, "kind": "mosaic", "tokens": 199, "dense": 448, "grad_norm": 0.708096981048584, "seconds": 6843.8} {"step": 1358, "loss": 0.6883646249771118, "data": 0.17773286998271942, "gap": 0.5106317400932312, "sigma": 0.8977243304252625, "kind": "mosaic", "tokens": 220, "dense": 448, "grad_norm": 0.3584267497062683, "seconds": 6848.2} {"step": 1359, "loss": 0.41722023487091064, "data": 0.246552512049675, "gap": 0.17066772282123566, "sigma": 0.969619870185852, "kind": "mosaic", "tokens": 241, "dense": 480, "grad_norm": 0.3775538206100464, "seconds": 6853.0} {"step": 1360, "loss": 0.3837625980377197, "data": 0.20918582379817963, "gap": 0.1745767593383789, "sigma": 0.9083773493766785, "kind": "mosaic", "tokens": 184, "dense": 448, "grad_norm": 0.15636935830116272, "seconds": 6857.5} {"step": 1361, "loss": 0.4572713077068329, "data": 0.4572713077068329, "gap": null, "sigma": 0.9290695190429688, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.15673376619815826, "seconds": 6860.4} {"step": 1362, "loss": 0.650105893611908, "data": 0.14584358036518097, "gap": 0.5042623281478882, "sigma": 0.9692832231521606, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.6380126476287842, "seconds": 6864.7} {"step": 1363, "loss": 0.5664235353469849, "data": 0.1854013204574585, "gap": 0.381022185087204, "sigma": 0.6861804723739624, "kind": "mosaic", "tokens": 158, "dense": 448, "grad_norm": 0.22220489382743835, "seconds": 6868.9} {"step": 1364, "loss": 0.4892685115337372, "data": 0.20787754654884338, "gap": 0.2813909649848938, "sigma": 0.7778042554855347, "kind": "mosaic", "tokens": 237, "dense": 448, "grad_norm": 0.3119089901447296, "seconds": 6873.4} {"step": 1365, "loss": 0.39773237705230713, "data": 0.20068666338920593, "gap": 0.1970457285642624, "sigma": 0.7231472730636597, "kind": "mosaic", "tokens": 268, "dense": 512, "grad_norm": 0.2549866735935211, "seconds": 6878.1} {"step": 1366, "loss": 0.44086959958076477, "data": 0.2617577910423279, "gap": 0.1791118085384369, "sigma": 0.8250665664672852, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.21006648242473602, "seconds": 6882.7} {"step": 1367, "loss": 0.4591726064682007, "data": 0.4591726064682007, "gap": null, "sigma": 0.9514068961143494, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2393326610326767, "seconds": 6885.6} {"step": 1368, "loss": 0.6428494453430176, "data": 0.2412101924419403, "gap": 0.4016392230987549, "sigma": 0.7239550948143005, "kind": "mosaic", "tokens": 202, "dense": 448, "grad_norm": 0.17063003778457642, "seconds": 6890.1} {"step": 1369, "loss": 0.7157201170921326, "data": 0.21145790815353394, "gap": 0.5042622089385986, "sigma": 0.9584047198295593, "kind": "mosaic", "tokens": 183, "dense": 448, "grad_norm": 0.2949156165122986, "seconds": 6894.5} {"step": 1370, "loss": 11.877554893493652, "data": 2.6866705417633057, "gap": 9.190884590148926, "sigma": 0.06805871427059174, "kind": "mosaic", "tokens": 230, "dense": 448, "grad_norm": 6.041816711425781, "seconds": 6899.0} {"step": 1371, "loss": 0.4232149124145508, "data": 0.16526435315608978, "gap": 0.2579505443572998, "sigma": 0.5851765871047974, "kind": "mosaic", "tokens": 183, "dense": 448, "grad_norm": 0.18371690809726715, "seconds": 6903.3} {"step": 1372, "loss": 0.19808699190616608, "data": 0.19808699190616608, "gap": null, "sigma": 0.8887573480606079, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.28974273800849915, "seconds": 6906.1} {"step": 1373, "loss": 0.7258933782577515, "data": 0.3483249545097351, "gap": 0.37756842374801636, "sigma": 0.9721778035163879, "kind": "mosaic", "tokens": 269, "dense": 448, "grad_norm": 0.43904685974121094, "seconds": 6910.8} {"step": 1374, "loss": 3.5438809394836426, "data": 0.7431834936141968, "gap": 2.8006975650787354, "sigma": 0.12954720854759216, "kind": "mosaic", "tokens": 219, "dense": 448, "grad_norm": 1.678282380104065, "seconds": 6915.3} {"step": 1375, "loss": 0.29972121119499207, "data": 0.29972121119499207, "gap": null, "sigma": 0.9690725803375244, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2456941306591034, "seconds": 6918.2} {"step": 1376, "loss": 0.4409101605415344, "data": 0.25958025455474854, "gap": 0.18132992088794708, "sigma": 0.91443932056427, "kind": "mosaic", "tokens": 223, "dense": 448, "grad_norm": 0.19733190536499023, "seconds": 6922.6} {"step": 1377, "loss": 2.0529725551605225, "data": 0.6820146441459656, "gap": 1.370957851409912, "sigma": 0.1505359411239624, "kind": "mosaic", "tokens": 236, "dense": 448, "grad_norm": 0.7555981278419495, "seconds": 6927.2} {"step": 1378, "loss": 0.29352521896362305, "data": 0.12733173370361328, "gap": 0.16619350016117096, "sigma": 0.9699312448501587, "kind": "mosaic", "tokens": 179, "dense": 448, "grad_norm": 0.155923992395401, "seconds": 6931.6} {"step": 1379, "loss": 0.728835940361023, "data": 0.19461500644683838, "gap": 0.5342209339141846, "sigma": 0.9136301279067993, "kind": "mosaic", "tokens": 217, "dense": 448, "grad_norm": 0.7646372318267822, "seconds": 6936.1} {"step": 1380, "loss": 0.7330697774887085, "data": 0.27224135398864746, "gap": 0.46082839369773865, "sigma": 0.7104095816612244, "kind": "mosaic", "tokens": 235, "dense": 448, "grad_norm": 0.3681747615337372, "seconds": 6940.7} {"step": 1381, "loss": 0.7356852293014526, "data": 0.31894659996032715, "gap": 0.4167386293411255, "sigma": 0.9265767931938171, "kind": "mosaic", "tokens": 273, "dense": 448, "grad_norm": 0.22787907719612122, "seconds": 6945.2} {"step": 1382, "loss": 0.6144567728042603, "data": 0.24716591835021973, "gap": 0.3672908544540405, "sigma": 0.574693500995636, "kind": "mosaic", "tokens": 231, "dense": 448, "grad_norm": 0.16533158719539642, "seconds": 6949.9} {"step": 1383, "loss": 0.6074392795562744, "data": 0.29799458384513855, "gap": 0.30944469571113586, "sigma": 0.9564588069915771, "kind": "mosaic", "tokens": 242, "dense": 448, "grad_norm": 0.46432673931121826, "seconds": 6954.3} {"step": 1384, "loss": 0.4849432110786438, "data": 0.23303495347499847, "gap": 0.2519082725048065, "sigma": 0.6760470867156982, "kind": "mosaic", "tokens": 227, "dense": 448, "grad_norm": 0.15980498492717743, "seconds": 6959.0} {"step": 1385, "loss": 0.8308184146881104, "data": 0.35546281933784485, "gap": 0.4753556251525879, "sigma": 0.9563965201377869, "kind": "mosaic", "tokens": 313, "dense": 576, "grad_norm": 0.6780422329902649, "seconds": 6963.7} {"step": 1386, "loss": 0.5019461512565613, "data": 0.2652541995048523, "gap": 0.23669195175170898, "sigma": 0.7818140983581543, "kind": "mosaic", "tokens": 242, "dense": 448, "grad_norm": 0.29281631112098694, "seconds": 6968.2} {"step": 1387, "loss": 0.5124505758285522, "data": 0.5124505758285522, "gap": null, "sigma": 0.9985653162002563, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.7251112461090088, "seconds": 6971.1} {"step": 1388, "loss": 1.1387618780136108, "data": 0.29647740721702576, "gap": 0.8422845005989075, "sigma": 0.8917315006256104, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.7045133709907532, "seconds": 6975.6} {"step": 1389, "loss": 0.7272714376449585, "data": 0.2391202598810196, "gap": 0.4881511926651001, "sigma": 0.9188022017478943, "kind": "mosaic", "tokens": 177, "dense": 448, "grad_norm": 0.7310009598731995, "seconds": 6980.3} {"step": 1390, "loss": 1.4145296812057495, "data": 0.12963245809078217, "gap": 1.2848972082138062, "sigma": 0.9411247968673706, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.8862014412879944, "seconds": 6984.2} {"step": 1391, "loss": 0.3358156979084015, "data": 0.3358156979084015, "gap": null, "sigma": 0.9559448957443237, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.20929424464702606, "seconds": 6987.0} {"step": 1392, "loss": 0.6085718274116516, "data": 0.22303709387779236, "gap": 0.38553473353385925, "sigma": 0.7842364311218262, "kind": "mosaic", "tokens": 253, "dense": 448, "grad_norm": 0.6018134355545044, "seconds": 6991.6} {"step": 1393, "loss": 0.6674807071685791, "data": 0.22549910843372345, "gap": 0.44198158383369446, "sigma": 0.941228449344635, "kind": "mosaic", "tokens": 217, "dense": 448, "grad_norm": 0.30260899662971497, "seconds": 6996.0} {"step": 1394, "loss": 0.5720090866088867, "data": 0.2245132327079773, "gap": 0.34749582409858704, "sigma": 0.8761517405509949, "kind": "mosaic", "tokens": 223, "dense": 448, "grad_norm": 0.2830106317996979, "seconds": 7000.6} {"step": 1395, "loss": 0.44842448830604553, "data": 0.18220627307891846, "gap": 0.2662182152271271, "sigma": 0.3845944404602051, "kind": "mosaic", "tokens": 211, "dense": 448, "grad_norm": 0.1310836672782898, "seconds": 7005.0} {"step": 1396, "loss": 0.37387213110923767, "data": 0.37387213110923767, "gap": null, "sigma": 0.3837389051914215, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.055184196680784225, "seconds": 7007.8} {"step": 1397, "loss": 1.0240362882614136, "data": 0.08730337023735046, "gap": 0.9367328882217407, "sigma": 0.8321618437767029, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.589857280254364, "seconds": 7012.0} {"step": 1398, "loss": 0.3414623737335205, "data": 0.13135051727294922, "gap": 0.2101118564605713, "sigma": 0.9536046385765076, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.2750065326690674, "seconds": 7016.4} {"step": 1399, "loss": 0.4594399929046631, "data": 0.23221175372600555, "gap": 0.22722823917865753, "sigma": 0.8762326836585999, "kind": "mosaic", "tokens": 210, "dense": 448, "grad_norm": 0.22311903536319733, "seconds": 7020.9} {"step": 1400, "loss": 0.41630372405052185, "data": 0.41630372405052185, "gap": null, "sigma": 0.9224656224250793, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.18576768040657043, "seconds": 7023.7} {"step": 1401, "loss": 0.5639905333518982, "data": 0.5639905333518982, "gap": null, "sigma": 0.9941412210464478, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.482944130897522, "seconds": 7026.6} {"step": 1402, "loss": 0.3475746810436249, "data": 0.3475746810436249, "gap": null, "sigma": 0.8029953241348267, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.09610430151224136, "seconds": 7029.5} {"step": 1403, "loss": 0.4760434925556183, "data": 0.2739332318305969, "gap": 0.20211026072502136, "sigma": 0.8734526038169861, "kind": "mosaic", "tokens": 308, "dense": 576, "grad_norm": 0.21165987849235535, "seconds": 7034.1} {"step": 1404, "loss": 0.3991348445415497, "data": 0.21810245513916016, "gap": 0.18103238940238953, "sigma": 0.8244709968566895, "kind": "mosaic", "tokens": 252, "dense": 576, "grad_norm": 0.18624120950698853, "seconds": 7038.7} {"step": 1405, "loss": 0.5422800779342651, "data": 0.14851917326450348, "gap": 0.39376088976860046, "sigma": 0.8211078643798828, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.22501084208488464, "seconds": 7042.9} {"step": 1406, "loss": 0.3096601068973541, "data": 0.3096601068973541, "gap": null, "sigma": 0.8300448060035706, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.39769601821899414, "seconds": 7045.7} {"step": 1407, "loss": 0.8111215829849243, "data": 0.2902128994464874, "gap": 0.5209087133407593, "sigma": 0.35771238803863525, "kind": "mosaic", "tokens": 217, "dense": 448, "grad_norm": 0.2574816644191742, "seconds": 7050.3} {"step": 1408, "loss": 16.16707992553711, "data": 3.2333364486694336, "gap": 12.93374252319336, "sigma": 0.007067585363984108, "kind": "mosaic", "tokens": 293, "dense": 448, "grad_norm": 12.922807693481445, "seconds": 7055.1} {"step": 1409, "loss": 3.60945987701416, "data": 0.885873019695282, "gap": 2.7235867977142334, "sigma": 0.11054890602827072, "kind": "mosaic", "tokens": 232, "dense": 448, "grad_norm": 1.9034944772720337, "seconds": 7059.6} {"step": 1410, "loss": 0.942893385887146, "data": 0.3058381676673889, "gap": 0.6370552182197571, "sigma": 0.981106698513031, "kind": "mosaic", "tokens": 198, "dense": 448, "grad_norm": 1.096866250038147, "seconds": 7064.1} {"step": 1411, "loss": 0.5569319725036621, "data": 0.22853446006774902, "gap": 0.3283975422382355, "sigma": 0.8387524485588074, "kind": "mosaic", "tokens": 235, "dense": 448, "grad_norm": 0.35716769099235535, "seconds": 7068.7} {"step": 1412, "loss": 0.44740527868270874, "data": 0.20095889270305634, "gap": 0.2464463710784912, "sigma": 0.6951658129692078, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.2055385708808899, "seconds": 7073.5} {"step": 1413, "loss": 0.48743265867233276, "data": 0.18287602066993713, "gap": 0.30455663800239563, "sigma": 0.7210680246353149, "kind": "mosaic", "tokens": 169, "dense": 448, "grad_norm": 0.1785326451063156, "seconds": 7077.9} {"step": 1414, "loss": 0.604417085647583, "data": 0.604417085647583, "gap": null, "sigma": 0.06300510466098785, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2173214554786682, "seconds": 7080.7} {"step": 1415, "loss": 0.6674522161483765, "data": 0.22896939516067505, "gap": 0.43848279118537903, "sigma": 0.4808443486690521, "kind": "mosaic", "tokens": 187, "dense": 448, "grad_norm": 0.16200833022594452, "seconds": 7085.3} {"step": 1416, "loss": 0.597815990447998, "data": 0.2649049460887909, "gap": 0.33291104435920715, "sigma": 0.6679958701133728, "kind": "mosaic", "tokens": 235, "dense": 448, "grad_norm": 0.18935315310955048, "seconds": 7090.0} {"step": 1417, "loss": 1.5070171356201172, "data": 0.13709761202335358, "gap": 1.3699195384979248, "sigma": 0.47111067175865173, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 0.6147500276565552, "seconds": 7094.2} {"step": 1418, "loss": 0.3352077901363373, "data": 0.18102282285690308, "gap": 0.1541849672794342, "sigma": 0.6205079555511475, "kind": "mosaic", "tokens": 245, "dense": 448, "grad_norm": 0.13332046568393707, "seconds": 7098.6} {"step": 1419, "loss": 1.1664090156555176, "data": 0.3618786334991455, "gap": 0.8045303821563721, "sigma": 0.9892550110816956, "kind": "mosaic", "tokens": 268, "dense": 448, "grad_norm": 0.7624567747116089, "seconds": 7103.1} {"step": 1420, "loss": 1.989133596420288, "data": 0.3330586850643158, "gap": 1.65607488155365, "sigma": 0.2929283082485199, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.49254733324050903, "seconds": 7107.0} {"step": 1421, "loss": 0.41758573055267334, "data": 0.20348899066448212, "gap": 0.21409675478935242, "sigma": 0.884518563747406, "kind": "mosaic", "tokens": 194, "dense": 448, "grad_norm": 0.21299304068088531, "seconds": 7111.5} {"step": 1422, "loss": 9.66800308227539, "data": 1.469996690750122, "gap": 8.198006629943848, "sigma": 0.11248151957988739, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 5.1856369972229, "seconds": 7115.6} {"step": 1423, "loss": 0.5320342779159546, "data": 0.23435896635055542, "gap": 0.29767534136772156, "sigma": 0.7849530577659607, "kind": "mosaic", "tokens": 275, "dense": 448, "grad_norm": 0.17418797314167023, "seconds": 7120.2} {"step": 1424, "loss": 0.46224498748779297, "data": 0.20118564367294312, "gap": 0.26105934381484985, "sigma": 0.7761704921722412, "kind": "mosaic", "tokens": 193, "dense": 448, "grad_norm": 0.3168668746948242, "seconds": 7124.7} {"step": 1425, "loss": 0.8959571123123169, "data": 0.31668463349342346, "gap": 0.579272449016571, "sigma": 0.41787388920783997, "kind": "mosaic", "tokens": 241, "dense": 448, "grad_norm": 0.18081584572792053, "seconds": 7129.4} {"step": 1426, "loss": 1.3335661888122559, "data": 0.3267408013343811, "gap": 1.0068254470825195, "sigma": 0.9980595707893372, "kind": "mosaic", "tokens": 177, "dense": 448, "grad_norm": 0.8678915500640869, "seconds": 7134.0} {"step": 1427, "loss": 0.49027425050735474, "data": 0.1926022469997406, "gap": 0.29767200350761414, "sigma": 0.6074439287185669, "kind": "mosaic", "tokens": 196, "dense": 448, "grad_norm": 0.3996369540691376, "seconds": 7138.3} {"step": 1428, "loss": 0.6679857969284058, "data": 0.19964468479156494, "gap": 0.4683411419391632, "sigma": 0.4920847415924072, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.26034244894981384, "seconds": 7142.6} {"step": 1429, "loss": 0.5988292098045349, "data": 0.5988292098045349, "gap": null, "sigma": 0.9947904944419861, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.532961368560791, "seconds": 7145.4} {"step": 1430, "loss": 0.29904723167419434, "data": 0.07108774781227112, "gap": 0.22795948386192322, "sigma": 0.8268660306930542, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.25074318051338196, "seconds": 7149.8} {"step": 1431, "loss": 0.31494781374931335, "data": 0.31494781374931335, "gap": null, "sigma": 0.9238685369491577, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2003905326128006, "seconds": 7152.6} {"step": 1432, "loss": 0.39111918210983276, "data": 0.39111918210983276, "gap": null, "sigma": 0.9812575578689575, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.24611859023571014, "seconds": 7155.4} {"step": 1433, "loss": 1.285394310951233, "data": 0.31297042965888977, "gap": 0.9724238514900208, "sigma": 0.2879796624183655, "kind": "mosaic", "tokens": 219, "dense": 448, "grad_norm": 0.41711002588272095, "seconds": 7160.0} {"step": 1434, "loss": 0.4701237380504608, "data": 0.13132959604263306, "gap": 0.33879414200782776, "sigma": 0.9270645380020142, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.3491703271865845, "seconds": 7164.4} {"step": 1435, "loss": 1.7315462827682495, "data": 0.5552664995193481, "gap": 1.1762797832489014, "sigma": 0.13451074063777924, "kind": "mosaic", "tokens": 231, "dense": 448, "grad_norm": 1.014819622039795, "seconds": 7169.0} {"step": 1436, "loss": 0.5568435788154602, "data": 0.1850087195634842, "gap": 0.3718348741531372, "sigma": 0.9281647801399231, "kind": "mosaic", "tokens": 205, "dense": 448, "grad_norm": 0.31087803840637207, "seconds": 7173.6} {"step": 1437, "loss": 0.4419329762458801, "data": 0.20793497562408447, "gap": 0.23399800062179565, "sigma": 0.7198114395141602, "kind": "mosaic", "tokens": 268, "dense": 448, "grad_norm": 0.16861499845981598, "seconds": 7178.2} {"step": 1438, "loss": 0.6980526447296143, "data": 0.253900945186615, "gap": 0.4441516697406769, "sigma": 0.5208756327629089, "kind": "mosaic", "tokens": 208, "dense": 448, "grad_norm": 0.14811579883098602, "seconds": 7182.9} {"step": 1439, "loss": 1.0804882049560547, "data": 0.3158940076828003, "gap": 0.7645941376686096, "sigma": 0.967907726764679, "kind": "mosaic", "tokens": 239, "dense": 448, "grad_norm": 0.5250407457351685, "seconds": 7187.5} {"step": 1440, "loss": 0.47730982303619385, "data": 0.04383439943194389, "gap": 0.43347543478012085, "sigma": 0.8413079977035522, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 0.5817248225212097, "seconds": 7191.6} {"step": 1441, "loss": 0.42761924862861633, "data": 0.42761924862861633, "gap": null, "sigma": 0.8904969096183777, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.15968002378940582, "seconds": 7194.5} {"step": 1442, "loss": 0.23115849494934082, "data": 0.23115849494934082, "gap": null, "sigma": 0.39926645159721375, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.06555525958538055, "seconds": 7197.4} {"step": 1443, "loss": 0.5282911062240601, "data": 0.1623258888721466, "gap": 0.36596518754959106, "sigma": 0.6148238778114319, "kind": "mosaic", "tokens": 161, "dense": 448, "grad_norm": 0.23955121636390686, "seconds": 7201.9} {"step": 1444, "loss": 0.5172969102859497, "data": 0.19067010283470154, "gap": 0.3266267776489258, "sigma": 0.6161246299743652, "kind": "mosaic", "tokens": 146, "dense": 448, "grad_norm": 0.15579208731651306, "seconds": 7206.0} {"step": 1445, "loss": 1.1972272396087646, "data": 0.30157625675201416, "gap": 0.8956509828567505, "sigma": 0.9702240228652954, "kind": "mosaic", "tokens": 179, "dense": 448, "grad_norm": 1.7639319896697998, "seconds": 7210.5} {"step": 1446, "loss": 0.4812111258506775, "data": 0.2378145456314087, "gap": 0.2433965802192688, "sigma": 0.7647007703781128, "kind": "mosaic", "tokens": 243, "dense": 448, "grad_norm": 0.260422945022583, "seconds": 7215.0} {"step": 1447, "loss": 0.6462776064872742, "data": 0.057683151215314865, "gap": 0.5885944366455078, "sigma": 0.7611172795295715, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 0.33833763003349304, "seconds": 7219.1} {"step": 1448, "loss": 1.2682607173919678, "data": 0.4210496246814728, "gap": 0.8472111225128174, "sigma": 0.288015753030777, "kind": "mosaic", "tokens": 222, "dense": 448, "grad_norm": 0.3251086175441742, "seconds": 7223.7} {"step": 1449, "loss": 0.34906983375549316, "data": 0.15035046637058258, "gap": 0.19871936738491058, "sigma": 0.5346701145172119, "kind": "mosaic", "tokens": 228, "dense": 448, "grad_norm": 0.12914544343948364, "seconds": 7228.3} {"step": 1450, "loss": 0.6140000820159912, "data": 0.07268554717302322, "gap": 0.5413145422935486, "sigma": 0.951514482498169, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 0.5720393061637878, "seconds": 7232.5} {"step": 1451, "loss": 0.5683006048202515, "data": 0.2915555536746979, "gap": 0.2767450213432312, "sigma": 0.9399465322494507, "kind": "mosaic", "tokens": 211, "dense": 448, "grad_norm": 0.3341832458972931, "seconds": 7237.0} {"step": 1452, "loss": 0.3713381290435791, "data": 0.14712858200073242, "gap": 0.22420956194400787, "sigma": 0.7805057764053345, "kind": "mosaic", "tokens": 197, "dense": 448, "grad_norm": 0.23925583064556122, "seconds": 7241.4} {"step": 1453, "loss": 0.5404295921325684, "data": 0.23087665438652039, "gap": 0.309552937746048, "sigma": 0.6471999883651733, "kind": "mosaic", "tokens": 218, "dense": 448, "grad_norm": 0.1959504783153534, "seconds": 7246.1} {"step": 1454, "loss": 0.9300282597541809, "data": 0.29601848125457764, "gap": 0.6340097784996033, "sigma": 0.9245558977127075, "kind": "mosaic", "tokens": 262, "dense": 448, "grad_norm": 0.5177614688873291, "seconds": 7250.6} {"step": 1455, "loss": 0.5303148031234741, "data": 0.1036253422498703, "gap": 0.4266894459724426, "sigma": 0.7794855833053589, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.22043722867965698, "seconds": 7254.7} {"step": 1456, "loss": 0.5796514749526978, "data": 0.23731790482997894, "gap": 0.3423335552215576, "sigma": 0.5950989723205566, "kind": "mosaic", "tokens": 212, "dense": 448, "grad_norm": 0.18549445271492004, "seconds": 7259.2} {"step": 1457, "loss": 0.38438835740089417, "data": 0.1958605945110321, "gap": 0.18852776288986206, "sigma": 0.5573468804359436, "kind": "mosaic", "tokens": 231, "dense": 480, "grad_norm": 0.08532954007387161, "seconds": 7263.6} {"step": 1458, "loss": 0.49264463782310486, "data": 0.23223373293876648, "gap": 0.2604109048843384, "sigma": 0.6641812324523926, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.18762463331222534, "seconds": 7268.1} {"step": 1459, "loss": 2.7792224884033203, "data": 0.7703859210014343, "gap": 2.008836507797241, "sigma": 0.22659462690353394, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 1.092300534248352, "seconds": 7272.4} {"step": 1460, "loss": 0.5998682379722595, "data": 0.2176099568605423, "gap": 0.3822582960128784, "sigma": 0.4064277410507202, "kind": "mosaic", "tokens": 211, "dense": 448, "grad_norm": 0.15679214894771576, "seconds": 7276.9} {"step": 1461, "loss": 0.609290599822998, "data": 0.2655642032623291, "gap": 0.34372636675834656, "sigma": 0.9135783910751343, "kind": "mosaic", "tokens": 273, "dense": 448, "grad_norm": 0.28403738141059875, "seconds": 7281.3} {"step": 1462, "loss": 0.5297508239746094, "data": 0.16316017508506775, "gap": 0.3665906488895416, "sigma": 0.7127271890640259, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.3038853704929352, "seconds": 7285.6} {"step": 1463, "loss": 0.29480457305908203, "data": 0.29480457305908203, "gap": null, "sigma": 0.9948443174362183, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.19101712107658386, "seconds": 7288.4} {"step": 1464, "loss": 0.42718052864074707, "data": 0.15697920322418213, "gap": 0.27020132541656494, "sigma": 0.7697693705558777, "kind": "mosaic", "tokens": 237, "dense": 448, "grad_norm": 0.18614408373832703, "seconds": 7293.0} {"step": 1465, "loss": 0.34596461057662964, "data": 0.12754938006401062, "gap": 0.21841521561145782, "sigma": 0.9218516945838928, "kind": "mosaic", "tokens": 211, "dense": 448, "grad_norm": 0.34844839572906494, "seconds": 7297.6} {"step": 1466, "loss": 0.2882182002067566, "data": 0.1483761966228485, "gap": 0.13984201848506927, "sigma": 0.8202810883522034, "kind": "mosaic", "tokens": 240, "dense": 448, "grad_norm": 0.159956157207489, "seconds": 7302.2} {"step": 1467, "loss": 0.6214948296546936, "data": 0.09078850597143173, "gap": 0.5307063460350037, "sigma": 0.9203404784202576, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.8594367504119873, "seconds": 7306.3} {"step": 1468, "loss": 2.4335756301879883, "data": 0.29167234897613525, "gap": 2.1419034004211426, "sigma": 0.9937447905540466, "kind": "mosaic", "tokens": 254, "dense": 448, "grad_norm": 1.4437285661697388, "seconds": 7310.8} {"step": 1469, "loss": 0.6535840630531311, "data": 0.1187669038772583, "gap": 0.5348171591758728, "sigma": 0.867919921875, "kind": "mosaic", "tokens": 167, "dense": 448, "grad_norm": 0.3175254464149475, "seconds": 7315.2} {"step": 1470, "loss": 0.39026010036468506, "data": 0.1764688789844513, "gap": 0.21379123628139496, "sigma": 0.6488395929336548, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.26023852825164795, "seconds": 7319.7} {"step": 1471, "loss": 3.3633975982666016, "data": 0.7256022691726685, "gap": 2.6377952098846436, "sigma": 0.12007235735654831, "kind": "mosaic", "tokens": 241, "dense": 448, "grad_norm": 1.6807620525360107, "seconds": 7324.1} {"step": 1472, "loss": 13.903316497802734, "data": 2.732811689376831, "gap": 11.170504570007324, "sigma": 0.07161525636911392, "kind": "mosaic", "tokens": 176, "dense": 448, "grad_norm": 8.57572078704834, "seconds": 7328.6} {"step": 1473, "loss": 0.47731226682662964, "data": 0.26608186960220337, "gap": 0.21123039722442627, "sigma": 0.8206531405448914, "kind": "mosaic", "tokens": 271, "dense": 448, "grad_norm": 0.2922333776950836, "seconds": 7333.0} {"step": 1474, "loss": 0.5604218244552612, "data": 0.2980491518974304, "gap": 0.2623726725578308, "sigma": 0.9214096665382385, "kind": "mosaic", "tokens": 245, "dense": 448, "grad_norm": 0.3339020013809204, "seconds": 7337.5} {"step": 1475, "loss": 0.2834397256374359, "data": 0.2834397256374359, "gap": null, "sigma": 0.9222444891929626, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.1628524363040924, "seconds": 7340.4} {"step": 1476, "loss": 0.6072125434875488, "data": 0.6072125434875488, "gap": null, "sigma": 0.9849733114242554, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.33215829730033875, "seconds": 7343.1} {"step": 1477, "loss": 4.961613655090332, "data": 0.8942397236824036, "gap": 4.067373752593994, "sigma": 0.11929246038198471, "kind": "mosaic", "tokens": 169, "dense": 448, "grad_norm": 2.0332818031311035, "seconds": 7347.7} {"step": 1478, "loss": 0.49298739433288574, "data": 0.22608257830142975, "gap": 0.2669048309326172, "sigma": 0.41207900643348694, "kind": "mosaic", "tokens": 248, "dense": 448, "grad_norm": 0.130674347281456, "seconds": 7352.2} {"step": 1479, "loss": 0.32789838314056396, "data": 0.21309657394886017, "gap": 0.1148018166422844, "sigma": 0.5832644701004028, "kind": "mosaic", "tokens": 274, "dense": 448, "grad_norm": 0.09058970957994461, "seconds": 7356.9} {"step": 1480, "loss": 0.2075449824333191, "data": 0.11704706400632858, "gap": 0.09049791097640991, "sigma": 0.5049463510513306, "kind": "mosaic", "tokens": 264, "dense": 448, "grad_norm": 0.2578018009662628, "seconds": 7361.5} {"step": 1481, "loss": 0.5326172709465027, "data": 0.19724763929843903, "gap": 0.33536964654922485, "sigma": 0.7275194525718689, "kind": "mosaic", "tokens": 198, "dense": 448, "grad_norm": 0.16180191934108734, "seconds": 7365.8} {"step": 1482, "loss": 0.5615016222000122, "data": 0.24270962178707123, "gap": 0.3187919855117798, "sigma": 0.8216086626052856, "kind": "mosaic", "tokens": 233, "dense": 448, "grad_norm": 0.2827746868133545, "seconds": 7370.1} {"step": 1483, "loss": 1.7752671241760254, "data": 0.47021567821502686, "gap": 1.3050514459609985, "sigma": 0.23950323462486267, "kind": "mosaic", "tokens": 167, "dense": 448, "grad_norm": 0.6411735415458679, "seconds": 7374.3} {"step": 1484, "loss": 0.36276376247406006, "data": 0.2152796983718872, "gap": 0.14748404920101166, "sigma": 0.7363161444664001, "kind": "mosaic", "tokens": 250, "dense": 448, "grad_norm": 0.15770873427391052, "seconds": 7378.8} {"step": 1485, "loss": 0.42557939887046814, "data": 0.16887569427490234, "gap": 0.2567037045955658, "sigma": 0.6724399328231812, "kind": "mosaic", "tokens": 185, "dense": 480, "grad_norm": 0.15681949257850647, "seconds": 7383.2} {"step": 1486, "loss": 45.278839111328125, "data": 6.158079147338867, "gap": 39.12076187133789, "sigma": 0.04020042344927788, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 24.25543975830078, "seconds": 7387.2} {"step": 1487, "loss": 0.6434590816497803, "data": 0.2353844940662384, "gap": 0.40807458758354187, "sigma": 0.49755188822746277, "kind": "mosaic", "tokens": 183, "dense": 448, "grad_norm": 0.16660267114639282, "seconds": 7391.7} {"step": 1488, "loss": 0.6632486581802368, "data": 0.23704665899276733, "gap": 0.42620202898979187, "sigma": 0.8099837899208069, "kind": "mosaic", "tokens": 198, "dense": 448, "grad_norm": 0.30214932560920715, "seconds": 7396.1} {"step": 1489, "loss": 0.38789182901382446, "data": 0.1622764766216278, "gap": 0.22561533749103546, "sigma": 0.6629610657691956, "kind": "mosaic", "tokens": 194, "dense": 448, "grad_norm": 0.2086697667837143, "seconds": 7400.7} {"step": 1490, "loss": 0.2214268445968628, "data": 0.1281246691942215, "gap": 0.0933021679520607, "sigma": 0.7358480095863342, "kind": "mosaic", "tokens": 289, "dense": 448, "grad_norm": 0.22690726816654205, "seconds": 7405.2} {"step": 1491, "loss": 0.8551100492477417, "data": 0.3354041278362274, "gap": 0.5197058916091919, "sigma": 0.9901406168937683, "kind": "mosaic", "tokens": 248, "dense": 448, "grad_norm": 0.5325312614440918, "seconds": 7409.7} {"step": 1492, "loss": 1.0552715063095093, "data": 0.22771118581295013, "gap": 0.8275603652000427, "sigma": 0.9970018267631531, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 2.385162591934204, "seconds": 7413.9} {"step": 1493, "loss": 0.6049383878707886, "data": 0.22905397415161133, "gap": 0.37588438391685486, "sigma": 0.8553662300109863, "kind": "mosaic", "tokens": 192, "dense": 448, "grad_norm": 0.25654348731040955, "seconds": 7418.4} {"step": 1494, "loss": 0.6389641165733337, "data": 0.6389641165733337, "gap": null, "sigma": 0.9472702741622925, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.683286190032959, "seconds": 7421.2} {"step": 1495, "loss": 10.1241455078125, "data": 1.9055732488632202, "gap": 8.218572616577148, "sigma": 0.039892882108688354, "kind": "mosaic", "tokens": 231, "dense": 448, "grad_norm": 9.513251304626465, "seconds": 7425.7} {"step": 1496, "loss": 0.4403505027294159, "data": 0.4403505027294159, "gap": null, "sigma": 0.9367082118988037, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.21411244571208954, "seconds": 7428.6} {"step": 1497, "loss": 1.2958515882492065, "data": 0.32110220193862915, "gap": 0.9747493863105774, "sigma": 0.9829187393188477, "kind": "mosaic", "tokens": 185, "dense": 448, "grad_norm": 0.6704176068305969, "seconds": 7432.9} {"step": 1498, "loss": 0.13628391921520233, "data": 0.13628391921520233, "gap": null, "sigma": 0.8324170112609863, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.10566200315952301, "seconds": 7435.7} {"step": 1499, "loss": 0.5158421993255615, "data": 0.5158421993255615, "gap": null, "sigma": 0.9675086736679077, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2593404948711395, "seconds": 7438.4} {"step": 1500, "loss": 0.46643680334091187, "data": 0.1816651076078415, "gap": 0.2847716808319092, "sigma": 0.49239489436149597, "kind": "mosaic", "tokens": 227, "dense": 448, "grad_norm": 0.3110707700252533, "seconds": 7442.9} {"step": 1500, "eval": {"dense": 0.30779860603312653, "uniform2": 0.1822254341095686, "mosaic": 0.25121789425611496, "gap_uniform2": 0.5794254907717308, "gap_mosaic": 0.5832862493892511}, "eval_extent": {"1x1x1": 0.41669221716112953, "1x1x2": 0.3175062806162584, "1x1x4": 0.22900635954972945, "1x2x1": 0.3020929985127791, "1x2x2": 0.19555534662090635, "1x2x4": 0.12854241228650334, "1x4x1": 0.1986555529019189, "1x4x2": 0.11520703461504095, "1x4x4": 0.08158179056744119}} {"step": 1501, "loss": 0.22147299349308014, "data": 0.22147299349308014, "gap": null, "sigma": 0.8351448178291321, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.12318243831396103, "seconds": 7686.8} {"step": 1502, "loss": 0.9418059587478638, "data": 0.19665595889091492, "gap": 0.7451499700546265, "sigma": 0.9400414228439331, "kind": "mosaic", "tokens": 179, "dense": 448, "grad_norm": 0.8442859053611755, "seconds": 7691.2} {"step": 1503, "loss": 0.48146578669548035, "data": 0.20463240146636963, "gap": 0.2768333852291107, "sigma": 0.7262700200080872, "kind": "mosaic", "tokens": 172, "dense": 448, "grad_norm": 0.14951714873313904, "seconds": 7695.6} {"step": 1504, "loss": 0.4092634916305542, "data": 0.08635666221380234, "gap": 0.32290682196617126, "sigma": 0.8554280996322632, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.33657780289649963, "seconds": 7699.7} {"step": 1505, "loss": 0.4766451120376587, "data": 0.4766451120376587, "gap": null, "sigma": 0.9655206799507141, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.22433984279632568, "seconds": 7702.5} {"step": 1506, "loss": 0.8339093923568726, "data": 0.28029391169548035, "gap": 0.5536154508590698, "sigma": 0.40804323554039, "kind": "mosaic", "tokens": 204, "dense": 448, "grad_norm": 0.3654005825519562, "seconds": 7707.1} {"step": 1507, "loss": 0.7411749362945557, "data": 0.1281319558620453, "gap": 0.613042950630188, "sigma": 0.668359637260437, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.3425532877445221, "seconds": 7711.2} {"step": 1508, "loss": 1.3623590469360352, "data": 0.45229947566986084, "gap": 0.9100595712661743, "sigma": 0.2632439136505127, "kind": "mosaic", "tokens": 205, "dense": 448, "grad_norm": 0.4860222041606903, "seconds": 7715.5} {"step": 1509, "loss": 0.39048635959625244, "data": 0.19025224447250366, "gap": 0.20023411512374878, "sigma": 0.8077954053878784, "kind": "mosaic", "tokens": 222, "dense": 448, "grad_norm": 0.3385635018348694, "seconds": 7720.1} {"step": 1510, "loss": 0.6878389120101929, "data": 0.300672322511673, "gap": 0.3871665596961975, "sigma": 0.9445793032646179, "kind": "mosaic", "tokens": 248, "dense": 448, "grad_norm": 0.3191573917865753, "seconds": 7724.6} {"step": 1511, "loss": 0.4532995820045471, "data": 0.1824229508638382, "gap": 0.27087661623954773, "sigma": 0.5191866755485535, "kind": "mosaic", "tokens": 211, "dense": 448, "grad_norm": 0.18733544647693634, "seconds": 7728.8} {"step": 1512, "loss": 0.22477728128433228, "data": 0.11594665795564651, "gap": 0.10883061587810516, "sigma": 0.683230459690094, "kind": "mosaic", "tokens": 271, "dense": 448, "grad_norm": 0.2532709836959839, "seconds": 7733.4} {"step": 1513, "loss": 0.3409378230571747, "data": 0.14347133040428162, "gap": 0.19746649265289307, "sigma": 0.8049307465553284, "kind": "uniform", "tokens": 144, "dense": 576, "grad_norm": 0.15549254417419434, "seconds": 7737.7} {"step": 1514, "loss": 0.5485414862632751, "data": 0.17373719811439514, "gap": 0.37480428814888, "sigma": 0.978912889957428, "kind": "mosaic", "tokens": 252, "dense": 448, "grad_norm": 0.3625815808773041, "seconds": 7742.3} {"step": 1515, "loss": 0.347724974155426, "data": 0.347724974155426, "gap": null, "sigma": 0.8847159147262573, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.12498831748962402, "seconds": 7745.1} {"step": 1516, "loss": 0.5714150667190552, "data": 0.21511757373809814, "gap": 0.35629749298095703, "sigma": 0.8557885885238647, "kind": "mosaic", "tokens": 187, "dense": 448, "grad_norm": 0.48135361075401306, "seconds": 7749.7} {"step": 1517, "loss": 0.21169540286064148, "data": 0.21169540286064148, "gap": null, "sigma": 0.8196719288825989, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.1353766769170761, "seconds": 7752.5} {"step": 1518, "loss": 1.0169944763183594, "data": 0.28433331847190857, "gap": 0.7326611876487732, "sigma": 0.9312057495117188, "kind": "mosaic", "tokens": 221, "dense": 448, "grad_norm": 0.40082091093063354, "seconds": 7757.1} {"step": 1519, "loss": 0.5442178249359131, "data": 0.5442178249359131, "gap": null, "sigma": 0.9918376803398132, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.565360963344574, "seconds": 7759.9} {"step": 1520, "loss": 1.1200193166732788, "data": 0.35920464992523193, "gap": 0.7608146667480469, "sigma": 0.4012872874736786, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.33118337392807007, "seconds": 7764.0} {"step": 1521, "loss": 0.932431697845459, "data": 0.3134939968585968, "gap": 0.6189377307891846, "sigma": 0.3224819600582123, "kind": "mosaic", "tokens": 243, "dense": 448, "grad_norm": 0.28012847900390625, "seconds": 7768.6} {"step": 1522, "loss": 0.8251136541366577, "data": 0.2082350105047226, "gap": 0.6168786287307739, "sigma": 0.33691224455833435, "kind": "mosaic", "tokens": 155, "dense": 448, "grad_norm": 0.5775439143180847, "seconds": 7772.8} {"step": 1523, "loss": 0.29989346861839294, "data": 0.29989346861839294, "gap": null, "sigma": 0.8215184211730957, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.15460403263568878, "seconds": 7775.7} {"step": 1524, "loss": 0.36015552282333374, "data": 0.36015552282333374, "gap": null, "sigma": 0.7861268520355225, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.10146621614694595, "seconds": 7778.6} {"step": 1525, "loss": 0.4777596890926361, "data": 0.17725548148155212, "gap": 0.300504207611084, "sigma": 0.6171398162841797, "kind": "mosaic", "tokens": 215, "dense": 448, "grad_norm": 0.17921362817287445, "seconds": 7783.1} {"step": 1526, "loss": 0.55703204870224, "data": 0.2302795648574829, "gap": 0.3267524838447571, "sigma": 0.4989279806613922, "kind": "mosaic", "tokens": 267, "dense": 448, "grad_norm": 0.29899483919143677, "seconds": 7787.5} {"step": 1527, "loss": 0.4081399440765381, "data": 0.14088331162929535, "gap": 0.26725664734840393, "sigma": 0.8079867362976074, "kind": "mosaic", "tokens": 154, "dense": 448, "grad_norm": 0.20230303704738617, "seconds": 7791.7} {"step": 1528, "loss": 0.8589821457862854, "data": 0.21911238133907318, "gap": 0.639869749546051, "sigma": 0.8512510657310486, "kind": "mosaic", "tokens": 185, "dense": 448, "grad_norm": 0.45079973340034485, "seconds": 7796.3} {"step": 1529, "loss": 0.6965327262878418, "data": 0.299707293510437, "gap": 0.3968254625797272, "sigma": 0.9521087408065796, "kind": "mosaic", "tokens": 278, "dense": 448, "grad_norm": 0.3900119662284851, "seconds": 7800.7} {"step": 1530, "loss": 0.3937046527862549, "data": 0.18040721118450165, "gap": 0.21329742670059204, "sigma": 0.6789407134056091, "kind": "mosaic", "tokens": 237, "dense": 448, "grad_norm": 0.1400144398212433, "seconds": 7805.2} {"step": 1531, "loss": 1.8896677494049072, "data": 0.46142637729644775, "gap": 1.4282413721084595, "sigma": 0.9853607416152954, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 1.1544193029403687, "seconds": 7809.6} {"step": 1532, "loss": 0.880010187625885, "data": 0.17414061725139618, "gap": 0.7058695554733276, "sigma": 0.8737109303474426, "kind": "mosaic", "tokens": 234, "dense": 448, "grad_norm": 0.44992324709892273, "seconds": 7814.2} {"step": 1533, "loss": 0.34642505645751953, "data": 0.15674176812171936, "gap": 0.18968330323696136, "sigma": 0.87864089012146, "kind": "mosaic", "tokens": 173, "dense": 448, "grad_norm": 0.17335452139377594, "seconds": 7818.9} {"step": 1534, "loss": 0.263857901096344, "data": 0.13181893527507782, "gap": 0.13203896582126617, "sigma": 0.7805534601211548, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.2853250503540039, "seconds": 7823.5} {"step": 1535, "loss": 1.0950052738189697, "data": 0.3302870988845825, "gap": 0.7647181153297424, "sigma": 0.34490054845809937, "kind": "mosaic", "tokens": 176, "dense": 448, "grad_norm": 0.2543371617794037, "seconds": 7828.1} {"step": 1536, "loss": 0.8350332379341125, "data": 0.2583329677581787, "gap": 0.5767002701759338, "sigma": 0.9233949184417725, "kind": "mosaic", "tokens": 239, "dense": 448, "grad_norm": 0.6384419202804565, "seconds": 7832.5} {"step": 1537, "loss": 0.3530083894729614, "data": 0.17739790678024292, "gap": 0.1756104677915573, "sigma": 0.5123065710067749, "kind": "mosaic", "tokens": 235, "dense": 448, "grad_norm": 0.15737135708332062, "seconds": 7836.9} {"step": 1538, "loss": 0.4487609267234802, "data": 0.09872150421142578, "gap": 0.35003942251205444, "sigma": 0.7495488524436951, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.19924741983413696, "seconds": 7841.1} {"step": 1539, "loss": 1.3222026824951172, "data": 0.3315218687057495, "gap": 0.9906808733940125, "sigma": 0.9925317168235779, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 1.0800789594650269, "seconds": 7845.4} {"step": 1540, "loss": 0.6527599096298218, "data": 0.11550094932317734, "gap": 0.5372589826583862, "sigma": 0.7018038034439087, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.3086315095424652, "seconds": 7849.5} {"step": 1541, "loss": 0.4271344840526581, "data": 0.4271344840526581, "gap": null, "sigma": 0.9481004476547241, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.23858100175857544, "seconds": 7852.2} {"step": 1542, "loss": 0.38945910334587097, "data": 0.1073032021522522, "gap": 0.2821559011936188, "sigma": 0.8000980019569397, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.24755632877349854, "seconds": 7856.3} {"step": 1543, "loss": 0.44455498456954956, "data": 0.1905282884836197, "gap": 0.25402671098709106, "sigma": 0.8327643871307373, "kind": "mosaic", "tokens": 164, "dense": 448, "grad_norm": 0.1869007647037506, "seconds": 7860.8} {"step": 1544, "loss": 0.5910052061080933, "data": 0.29598814249038696, "gap": 0.2950170338153839, "sigma": 0.9596852660179138, "kind": "mosaic", "tokens": 232, "dense": 448, "grad_norm": 0.23840400576591492, "seconds": 7865.4} {"step": 1545, "loss": 0.28990840911865234, "data": 0.28990840911865234, "gap": null, "sigma": 0.36506277322769165, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.06096356362104416, "seconds": 7868.2} {"step": 1546, "loss": 0.6252846121788025, "data": 0.6252846121788025, "gap": null, "sigma": 0.9970841407775879, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.7162275910377502, "seconds": 7871.0} {"step": 1547, "loss": 0.5269337892532349, "data": 0.27566462755203247, "gap": 0.2512691915035248, "sigma": 0.9666575193405151, "kind": "mosaic", "tokens": 171, "dense": 448, "grad_norm": 0.39779913425445557, "seconds": 7875.4} {"step": 1548, "loss": 0.4707501530647278, "data": 0.19856344163417816, "gap": 0.2721867263317108, "sigma": 0.8300489783287048, "kind": "mosaic", "tokens": 238, "dense": 448, "grad_norm": 0.32175320386886597, "seconds": 7879.9} {"step": 1549, "loss": 8.358476638793945, "data": 1.7173335552215576, "gap": 6.641143321990967, "sigma": 0.08665693551301956, "kind": "mosaic", "tokens": 184, "dense": 448, "grad_norm": 4.213086128234863, "seconds": 7884.4} {"step": 1550, "loss": 0.8255200386047363, "data": 0.27136027812957764, "gap": 0.5541597604751587, "sigma": 0.36274632811546326, "kind": "mosaic", "tokens": 230, "dense": 448, "grad_norm": 0.26600784063339233, "seconds": 7889.0} {"step": 1551, "loss": 12.80055046081543, "data": 2.5216803550720215, "gap": 10.278870582580566, "sigma": 0.04701371490955353, "kind": "mosaic", "tokens": 264, "dense": 448, "grad_norm": 9.040083885192871, "seconds": 7893.6} {"step": 1552, "loss": 5.543376922607422, "data": 1.2099639177322388, "gap": 4.333413124084473, "sigma": 0.09393343329429626, "kind": "mosaic", "tokens": 242, "dense": 448, "grad_norm": 3.1453709602355957, "seconds": 7898.0} {"step": 1553, "loss": 0.5874965190887451, "data": 0.2470054179430008, "gap": 0.3404911160469055, "sigma": 0.3370923101902008, "kind": "mosaic", "tokens": 255, "dense": 448, "grad_norm": 0.17143842577934265, "seconds": 7902.5} {"step": 1554, "loss": 0.4081990420818329, "data": 0.4081990420818329, "gap": null, "sigma": 0.993853747844696, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.4451645314693451, "seconds": 7905.3} {"step": 1555, "loss": 1.2253880500793457, "data": 0.3294285535812378, "gap": 0.8959594964981079, "sigma": 0.23282873630523682, "kind": "mosaic", "tokens": 231, "dense": 448, "grad_norm": 0.5239813923835754, "seconds": 7909.9} {"step": 1556, "loss": 0.3805372714996338, "data": 0.18505005538463593, "gap": 0.19548720121383667, "sigma": 0.7195606827735901, "kind": "mosaic", "tokens": 232, "dense": 448, "grad_norm": 0.14316721260547638, "seconds": 7914.6} {"step": 1557, "loss": 0.5568892955780029, "data": 0.22158926725387573, "gap": 0.3352999985218048, "sigma": 0.8928053379058838, "kind": "mosaic", "tokens": 250, "dense": 448, "grad_norm": 0.21288642287254333, "seconds": 7919.2} {"step": 1558, "loss": 0.6422296166419983, "data": 0.042205389589071274, "gap": 0.6000242233276367, "sigma": 0.7689943909645081, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 0.3509048521518707, "seconds": 7923.3} {"step": 1559, "loss": 0.526191234588623, "data": 0.2677822709083557, "gap": 0.2584089934825897, "sigma": 0.6620844602584839, "kind": "mosaic", "tokens": 285, "dense": 448, "grad_norm": 0.1813514232635498, "seconds": 7927.8} {"step": 1560, "loss": 1.4291455745697021, "data": 0.5156757235527039, "gap": 0.9134697914123535, "sigma": 0.9835338592529297, "kind": "mosaic", "tokens": 224, "dense": 448, "grad_norm": 1.189340591430664, "seconds": 7932.3} {"step": 1561, "loss": 0.5081158876419067, "data": 0.24385806918144226, "gap": 0.2642578184604645, "sigma": 0.5347477793693542, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.20553235709667206, "seconds": 7936.8} {"step": 1562, "loss": 0.5479768514633179, "data": 0.18184424936771393, "gap": 0.36613261699676514, "sigma": 0.7350665926933289, "kind": "mosaic", "tokens": 165, "dense": 448, "grad_norm": 0.1916249394416809, "seconds": 7941.3} {"step": 1563, "loss": 0.3519108295440674, "data": 0.17560893297195435, "gap": 0.17630188167095184, "sigma": 0.7217486500740051, "kind": "mosaic", "tokens": 231, "dense": 448, "grad_norm": 0.21304264664649963, "seconds": 7945.9} {"step": 1564, "loss": 3.4251410961151123, "data": 0.9489942193031311, "gap": 2.476146936416626, "sigma": 0.1317964643239975, "kind": "mosaic", "tokens": 249, "dense": 448, "grad_norm": 1.5282421112060547, "seconds": 7950.5} {"step": 1565, "loss": 0.4502798318862915, "data": 0.25179582834243774, "gap": 0.19848401844501495, "sigma": 0.7195114493370056, "kind": "mosaic", "tokens": 268, "dense": 448, "grad_norm": 0.10781295597553253, "seconds": 7955.1} {"step": 1566, "loss": 0.48340168595314026, "data": 0.06365174055099487, "gap": 0.4197499454021454, "sigma": 0.9283283352851868, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 0.848031759262085, "seconds": 7959.2} {"step": 1567, "loss": 0.7774266600608826, "data": 0.24890422821044922, "gap": 0.5285224318504333, "sigma": 0.93828284740448, "kind": "mosaic", "tokens": 238, "dense": 448, "grad_norm": 0.37232059240341187, "seconds": 7963.8} {"step": 1568, "loss": 0.8320417404174805, "data": 0.2877412736415863, "gap": 0.5443004369735718, "sigma": 0.94963139295578, "kind": "mosaic", "tokens": 262, "dense": 448, "grad_norm": 0.9686019420623779, "seconds": 7968.4} {"step": 1569, "loss": 0.2697578966617584, "data": 0.13916794955730438, "gap": 0.13058994710445404, "sigma": 0.792912483215332, "kind": "mosaic", "tokens": 257, "dense": 448, "grad_norm": 0.16781680285930634, "seconds": 7972.9} {"step": 1570, "loss": 1.132190465927124, "data": 0.17273478209972382, "gap": 0.9594557285308838, "sigma": 0.8684549927711487, "kind": "mosaic", "tokens": 133, "dense": 448, "grad_norm": 0.5624797344207764, "seconds": 7977.3} {"step": 1571, "loss": 0.20141252875328064, "data": 0.20141252875328064, "gap": null, "sigma": 0.6915074586868286, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.11521267145872116, "seconds": 7980.2} {"step": 1572, "loss": 0.9977130889892578, "data": 0.3373005986213684, "gap": 0.6604124903678894, "sigma": 0.9717550873756409, "kind": "mosaic", "tokens": 236, "dense": 448, "grad_norm": 0.6973474025726318, "seconds": 7984.6} {"step": 1573, "loss": 0.5827810168266296, "data": 0.18898499011993408, "gap": 0.39379602670669556, "sigma": 0.8407337665557861, "kind": "mosaic", "tokens": 198, "dense": 448, "grad_norm": 0.18715159595012665, "seconds": 7989.2} {"step": 1574, "loss": 0.7650186419487, "data": 0.2598950266838074, "gap": 0.5051236152648926, "sigma": 0.4888937473297119, "kind": "mosaic", "tokens": 238, "dense": 448, "grad_norm": 0.19662386178970337, "seconds": 7993.6} {"step": 1575, "loss": 0.7953474521636963, "data": 0.39325642585754395, "gap": 0.40209099650382996, "sigma": 0.9412887692451477, "kind": "mosaic", "tokens": 276, "dense": 448, "grad_norm": 0.4604335427284241, "seconds": 7998.3} {"step": 1576, "loss": 0.5945767164230347, "data": 0.5945767164230347, "gap": null, "sigma": 0.9932649731636047, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.7779814600944519, "seconds": 8001.1} {"step": 1577, "loss": 0.8708304166793823, "data": 0.2991122305393219, "gap": 0.5717182159423828, "sigma": 0.9731492400169373, "kind": "mosaic", "tokens": 217, "dense": 448, "grad_norm": 0.6282585263252258, "seconds": 8005.6} {"step": 1578, "loss": 0.26464080810546875, "data": 0.1283927857875824, "gap": 0.13624803721904755, "sigma": 0.8136669397354126, "kind": "mosaic", "tokens": 253, "dense": 448, "grad_norm": 0.14752478897571564, "seconds": 8010.2} {"step": 1579, "loss": 0.30738234519958496, "data": 0.21648967266082764, "gap": 0.09089267253875732, "sigma": 0.8878491520881653, "kind": "mosaic", "tokens": 226, "dense": 512, "grad_norm": 0.2280563861131668, "seconds": 8014.9} {"step": 1580, "loss": 0.47065502405166626, "data": 0.23080000281333923, "gap": 0.23985503613948822, "sigma": 0.7936095595359802, "kind": "mosaic", "tokens": 258, "dense": 448, "grad_norm": 0.2897648811340332, "seconds": 8019.5} {"step": 1581, "loss": 1.216360330581665, "data": 0.44640660285949707, "gap": 0.7699536681175232, "sigma": 0.25463584065437317, "kind": "mosaic", "tokens": 282, "dense": 448, "grad_norm": 0.9001820087432861, "seconds": 8024.0} {"step": 1582, "loss": 0.3912981152534485, "data": 0.13287217915058136, "gap": 0.2584259510040283, "sigma": 0.7165802121162415, "kind": "mosaic", "tokens": 166, "dense": 448, "grad_norm": 0.1683388501405716, "seconds": 8028.5} {"step": 1583, "loss": 0.2154570370912552, "data": 0.2154570370912552, "gap": null, "sigma": 0.7642790079116821, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.11311089247465134, "seconds": 8031.4} {"step": 1584, "loss": 0.3852121829986572, "data": 0.13233917951583862, "gap": 0.2528730034828186, "sigma": 0.7383633852005005, "kind": "mosaic", "tokens": 196, "dense": 448, "grad_norm": 0.31467095017433167, "seconds": 8036.0} {"step": 1585, "loss": 0.6321872472763062, "data": 0.12613698840141296, "gap": 0.5060502886772156, "sigma": 0.9615856409072876, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.4906949996948242, "seconds": 8040.3} {"step": 1586, "loss": 0.35331517457962036, "data": 0.35331517457962036, "gap": null, "sigma": 0.9639140963554382, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.26640793681144714, "seconds": 8043.1} {"step": 1587, "loss": 0.3543321490287781, "data": 0.3543321490287781, "gap": null, "sigma": 0.8179088234901428, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2028566151857376, "seconds": 8046.0} {"step": 1588, "loss": 0.7627674341201782, "data": 0.227525994181633, "gap": 0.535241425037384, "sigma": 0.8079352974891663, "kind": "mosaic", "tokens": 177, "dense": 448, "grad_norm": 0.2340361475944519, "seconds": 8050.5} {"step": 1589, "loss": 0.6981188654899597, "data": 0.2899462878704071, "gap": 0.4081725776195526, "sigma": 0.943119466304779, "kind": "mosaic", "tokens": 253, "dense": 448, "grad_norm": 0.5820873379707336, "seconds": 8054.9} {"step": 1590, "loss": 0.48492884635925293, "data": 0.19349683821201324, "gap": 0.2914320230484009, "sigma": 0.9609889388084412, "kind": "mosaic", "tokens": 220, "dense": 448, "grad_norm": 0.3980358839035034, "seconds": 8059.5} {"step": 1591, "loss": 0.4315492510795593, "data": 0.1668349653482437, "gap": 0.2647143006324768, "sigma": 0.6841920614242554, "kind": "mosaic", "tokens": 225, "dense": 448, "grad_norm": 0.2597391605377197, "seconds": 8063.9} {"step": 1592, "loss": 0.2870674133300781, "data": 0.1471593677997589, "gap": 0.1399080604314804, "sigma": 0.688482940196991, "kind": "mosaic", "tokens": 243, "dense": 448, "grad_norm": 0.10719295591115952, "seconds": 8068.4} {"step": 1593, "loss": 30.591503143310547, "data": 5.680951118469238, "gap": 24.910552978515625, "sigma": 0.0342019759118557, "kind": "mosaic", "tokens": 181, "dense": 448, "grad_norm": 16.552288055419922, "seconds": 8072.9} {"step": 1594, "loss": 0.31601300835609436, "data": 0.2531026303768158, "gap": 0.06291037052869797, "sigma": 0.8439772725105286, "kind": "mosaic", "tokens": 256, "dense": 448, "grad_norm": 0.11878158897161484, "seconds": 8077.4} {"step": 1595, "loss": 0.6429871320724487, "data": 0.17949995398521423, "gap": 0.4634872078895569, "sigma": 0.8059854507446289, "kind": "mosaic", "tokens": 205, "dense": 448, "grad_norm": 0.3523135483264923, "seconds": 8081.8} {"step": 1596, "loss": 0.5896758437156677, "data": 0.2725856602191925, "gap": 0.3170901834964752, "sigma": 0.8141917586326599, "kind": "mosaic", "tokens": 228, "dense": 448, "grad_norm": 0.2887744903564453, "seconds": 8086.2} {"step": 1597, "loss": 1.0181511640548706, "data": 0.3172806203365326, "gap": 0.7008705735206604, "sigma": 0.9905053377151489, "kind": "mosaic", "tokens": 178, "dense": 448, "grad_norm": 0.5644872188568115, "seconds": 8090.6} {"step": 1598, "loss": 0.630178689956665, "data": 0.2698930501937866, "gap": 0.3602856695652008, "sigma": 0.8168280124664307, "kind": "mosaic", "tokens": 246, "dense": 448, "grad_norm": 0.20181292295455933, "seconds": 8095.1} {"step": 1599, "loss": 0.46078765392303467, "data": 0.28542089462280273, "gap": 0.17536674439907074, "sigma": 0.8820586800575256, "kind": "mosaic", "tokens": 228, "dense": 448, "grad_norm": 0.2219296395778656, "seconds": 8099.6} {"step": 1600, "loss": 0.4197598099708557, "data": 0.21500235795974731, "gap": 0.2047574669122696, "sigma": 0.9404025077819824, "kind": "mosaic", "tokens": 202, "dense": 448, "grad_norm": 0.4342648684978485, "seconds": 8104.1} {"step": 1601, "loss": 0.5656286478042603, "data": 0.5656286478042603, "gap": null, "sigma": 0.9813887476921082, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.282623291015625, "seconds": 8106.9} {"step": 1602, "loss": 0.5351091027259827, "data": 0.19213351607322693, "gap": 0.34297558665275574, "sigma": 0.8333748579025269, "kind": "mosaic", "tokens": 161, "dense": 448, "grad_norm": 0.4029352068901062, "seconds": 8111.2} {"step": 1603, "loss": 0.48588526248931885, "data": 0.15302686393260956, "gap": 0.3328584134578705, "sigma": 0.7882262468338013, "kind": "mosaic", "tokens": 181, "dense": 448, "grad_norm": 0.5335471034049988, "seconds": 8115.5} {"step": 1604, "loss": 0.5556089282035828, "data": 0.5556089282035828, "gap": null, "sigma": 0.10498855262994766, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.10069143772125244, "seconds": 8118.4} {"step": 1605, "loss": 0.39394819736480713, "data": 0.05169433727860451, "gap": 0.3422538638114929, "sigma": 0.8042766451835632, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.2104026824235916, "seconds": 8122.6} {"step": 1606, "loss": 0.6206785440444946, "data": 0.22986571490764618, "gap": 0.39081281423568726, "sigma": 0.5322911739349365, "kind": "mosaic", "tokens": 223, "dense": 448, "grad_norm": 0.15882296860218048, "seconds": 8127.1} {"step": 1607, "loss": 0.7252863645553589, "data": 0.22981686890125275, "gap": 0.49546951055526733, "sigma": 0.5567057132720947, "kind": "mosaic", "tokens": 176, "dense": 448, "grad_norm": 0.3428666889667511, "seconds": 8131.6} {"step": 1608, "loss": 1.213742733001709, "data": 0.22013919055461884, "gap": 0.9936035871505737, "sigma": 0.5513324737548828, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.4947963356971741, "seconds": 8135.7} {"step": 1609, "loss": 0.3856632709503174, "data": 0.3856632709503174, "gap": null, "sigma": 0.9531391263008118, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2103019654750824, "seconds": 8138.6} {"step": 1610, "loss": 21.23310661315918, "data": 4.060643672943115, "gap": 17.172462463378906, "sigma": 0.020749518647789955, "kind": "mosaic", "tokens": 223, "dense": 448, "grad_norm": 12.700820922851562, "seconds": 8143.2} {"step": 1611, "loss": 0.5328546166419983, "data": 0.15895986557006836, "gap": 0.37389475107192993, "sigma": 0.8236212134361267, "kind": "mosaic", "tokens": 155, "dense": 448, "grad_norm": 0.3847852051258087, "seconds": 8147.5} {"step": 1612, "loss": 0.5087197422981262, "data": 0.1600807011127472, "gap": 0.34863904118537903, "sigma": 0.966793417930603, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.7613604068756104, "seconds": 8151.9} {"step": 1613, "loss": 1.0722193717956543, "data": 0.30815285444259644, "gap": 0.7640664577484131, "sigma": 0.34552040696144104, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.439284086227417, "seconds": 8156.2} {"step": 1614, "loss": 1.571068525314331, "data": 0.3269934356212616, "gap": 1.244075059890747, "sigma": 0.967911422252655, "kind": "mosaic", "tokens": 252, "dense": 448, "grad_norm": 0.9861205220222473, "seconds": 8160.9} {"step": 1615, "loss": 0.42479148507118225, "data": 0.18903464078903198, "gap": 0.23575684428215027, "sigma": 0.8691299557685852, "kind": "mosaic", "tokens": 277, "dense": 448, "grad_norm": 0.38459697365760803, "seconds": 8165.6} {"step": 1616, "loss": 0.45205679535865784, "data": 0.45205679535865784, "gap": null, "sigma": 0.9902719259262085, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.4121178090572357, "seconds": 8168.4} {"step": 1617, "loss": 0.5982878804206848, "data": 0.1760711371898651, "gap": 0.4222167432308197, "sigma": 0.9766368865966797, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.46881604194641113, "seconds": 8172.8} {"step": 1618, "loss": 0.4312058985233307, "data": 0.1816602647304535, "gap": 0.2495456337928772, "sigma": 0.9336195588111877, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.8225336670875549, "seconds": 8177.0} {"step": 1619, "loss": 0.47095781564712524, "data": 0.47095781564712524, "gap": null, "sigma": 0.9810846447944641, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.30845147371292114, "seconds": 8179.9} {"step": 1620, "loss": 12.990300178527832, "data": 2.6558446884155273, "gap": 10.334455490112305, "sigma": 0.012465600855648518, "kind": "mosaic", "tokens": 274, "dense": 448, "grad_norm": 9.960613250732422, "seconds": 8184.5} {"step": 1621, "loss": 0.4915577471256256, "data": 0.24601741135120392, "gap": 0.2455403357744217, "sigma": 0.8239129185676575, "kind": "mosaic", "tokens": 216, "dense": 448, "grad_norm": 0.5096129775047302, "seconds": 8189.0} {"step": 1622, "loss": 0.5100130438804626, "data": 0.14703291654586792, "gap": 0.3629801273345947, "sigma": 0.8863934874534607, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.35722580552101135, "seconds": 8193.4} {"step": 1623, "loss": 0.3864057660102844, "data": 0.3864057660102844, "gap": null, "sigma": 0.9607797861099243, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.18336914479732513, "seconds": 8196.3} {"step": 1624, "loss": 0.2949258089065552, "data": 0.17396441102027893, "gap": 0.12096139043569565, "sigma": 0.5161855816841125, "kind": "mosaic", "tokens": 261, "dense": 448, "grad_norm": 0.13576306402683258, "seconds": 8200.7} {"step": 1625, "loss": 0.5999426245689392, "data": 0.24510560929775238, "gap": 0.354837030172348, "sigma": 0.8185607194900513, "kind": "mosaic", "tokens": 212, "dense": 448, "grad_norm": 0.3365190923213959, "seconds": 8205.3} {"step": 1626, "loss": 0.5312177538871765, "data": 0.04457489028573036, "gap": 0.48664286732673645, "sigma": 0.9503052234649658, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 0.6640366911888123, "seconds": 8209.4} {"step": 1627, "loss": 0.817671537399292, "data": 0.2801196873188019, "gap": 0.5375518202781677, "sigma": 0.42269250750541687, "kind": "mosaic", "tokens": 207, "dense": 448, "grad_norm": 0.26904913783073425, "seconds": 8214.0} {"step": 1628, "loss": 0.4597303867340088, "data": 0.17432260513305664, "gap": 0.28540778160095215, "sigma": 0.49420833587646484, "kind": "mosaic", "tokens": 194, "dense": 448, "grad_norm": 0.18212467432022095, "seconds": 8218.6} {"step": 1629, "loss": 0.6948856711387634, "data": 0.18526078760623932, "gap": 0.5096248984336853, "sigma": 0.9521225690841675, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 1.2316787242889404, "seconds": 8222.8} {"step": 1630, "loss": 0.27707087993621826, "data": 0.27707087993621826, "gap": null, "sigma": 0.9477213621139526, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.16731084883213043, "seconds": 8225.6} {"step": 1631, "loss": 0.31172892451286316, "data": 0.14419658482074738, "gap": 0.16753233969211578, "sigma": 0.6836837530136108, "kind": "mosaic", "tokens": 249, "dense": 448, "grad_norm": 0.1195819079875946, "seconds": 8230.3} {"step": 1632, "loss": 0.7039411664009094, "data": 0.24307119846343994, "gap": 0.4608699679374695, "sigma": 0.4095860421657562, "kind": "mosaic", "tokens": 258, "dense": 448, "grad_norm": 0.19254592061042786, "seconds": 8234.7} {"step": 1633, "loss": 0.21762621402740479, "data": 0.21762621402740479, "gap": null, "sigma": 0.49444955587387085, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3899778723716736, "seconds": 8237.5} {"step": 1634, "loss": 0.6287393569946289, "data": 0.15900519490242004, "gap": 0.46973416209220886, "sigma": 0.4753146767616272, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.23286184668540955, "seconds": 8241.8} {"step": 1635, "loss": 0.355564683675766, "data": 0.2532109320163727, "gap": 0.10235374420881271, "sigma": 0.871271550655365, "kind": "mosaic", "tokens": 251, "dense": 448, "grad_norm": 0.13643115758895874, "seconds": 8246.3} {"step": 1636, "loss": 0.568617045879364, "data": 0.20524190366268158, "gap": 0.36337512731552124, "sigma": 0.8503953218460083, "kind": "mosaic", "tokens": 285, "dense": 448, "grad_norm": 0.2809276580810547, "seconds": 8251.0} {"step": 1637, "loss": 0.34750229120254517, "data": 0.34750229120254517, "gap": null, "sigma": 0.16875173151493073, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.06692758202552795, "seconds": 8253.8} {"step": 1638, "loss": 0.4466387629508972, "data": 0.16890452802181244, "gap": 0.27773424983024597, "sigma": 0.8014353513717651, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.33579057455062866, "seconds": 8258.0} {"step": 1639, "loss": 0.4577081799507141, "data": 0.258283793926239, "gap": 0.1994244009256363, "sigma": 0.8005805611610413, "kind": "mosaic", "tokens": 260, "dense": 448, "grad_norm": 1.1618269681930542, "seconds": 8262.5} {"step": 1640, "loss": 0.7237353920936584, "data": 0.1525612622499466, "gap": 0.571174144744873, "sigma": 0.8050190806388855, "kind": "mosaic", "tokens": 178, "dense": 448, "grad_norm": 0.42663294076919556, "seconds": 8266.8} {"step": 1641, "loss": 0.7374570965766907, "data": 0.2384011447429657, "gap": 0.499055951833725, "sigma": 0.8589612245559692, "kind": "mosaic", "tokens": 226, "dense": 448, "grad_norm": 0.90545254945755, "seconds": 8271.3} {"step": 1642, "loss": 9.382854461669922, "data": 3.498342752456665, "gap": 5.884511470794678, "sigma": 0.053194399923086166, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 11.222241401672363, "seconds": 8275.9} {"step": 1643, "loss": 0.6709558963775635, "data": 0.3535163104534149, "gap": 0.31743961572647095, "sigma": 0.9624764323234558, "kind": "mosaic", "tokens": 235, "dense": 448, "grad_norm": 0.3424440324306488, "seconds": 8280.2} {"step": 1644, "loss": 0.6532996892929077, "data": 0.06149561330676079, "gap": 0.5918040871620178, "sigma": 0.6574679613113403, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 0.3863585293292999, "seconds": 8284.4} {"step": 1645, "loss": 0.46101298928260803, "data": 0.10692919045686722, "gap": 0.3540838062763214, "sigma": 0.9576483368873596, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.6441848874092102, "seconds": 8288.4} {"step": 1646, "loss": 0.5906720161437988, "data": 0.1986430138349533, "gap": 0.3920289874076843, "sigma": 0.5431181788444519, "kind": "mosaic", "tokens": 207, "dense": 448, "grad_norm": 0.16443198919296265, "seconds": 8292.9} {"step": 1647, "loss": 1.2363061904907227, "data": 0.38413339853286743, "gap": 0.8521727323532104, "sigma": 0.9776546955108643, "kind": "mosaic", "tokens": 253, "dense": 448, "grad_norm": 0.5613142848014832, "seconds": 8297.3} {"step": 1648, "loss": 0.8739140629768372, "data": 0.1579521894454956, "gap": 0.7159618735313416, "sigma": 0.944829523563385, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.9277197122573853, "seconds": 8301.5} {"step": 1649, "loss": 0.9453339576721191, "data": 0.25802186131477356, "gap": 0.6873120665550232, "sigma": 0.383095383644104, "kind": "mosaic", "tokens": 164, "dense": 448, "grad_norm": 0.3122669756412506, "seconds": 8305.8} {"step": 1650, "loss": 0.2574528455734253, "data": 0.1308005005121231, "gap": 0.126652330160141, "sigma": 0.44050833582878113, "kind": "mosaic", "tokens": 238, "dense": 448, "grad_norm": 0.09485596418380737, "seconds": 8310.2} {"step": 1651, "loss": 1.035864233970642, "data": 0.1666092723608017, "gap": 0.869255006313324, "sigma": 0.9306882619857788, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.4990893602371216, "seconds": 8314.5} {"step": 1652, "loss": 0.4046265780925751, "data": 0.4046265780925751, "gap": null, "sigma": 0.9392194747924805, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.1742335855960846, "seconds": 8317.3} {"step": 1653, "loss": 0.43406230211257935, "data": 0.21287430822849274, "gap": 0.2211880087852478, "sigma": 0.5068150162696838, "kind": "mosaic", "tokens": 265, "dense": 480, "grad_norm": 0.2628073990345001, "seconds": 8322.0} {"step": 1654, "loss": 1.5006721019744873, "data": 0.33555757999420166, "gap": 1.1651145219802856, "sigma": 0.9616013169288635, "kind": "mosaic", "tokens": 226, "dense": 448, "grad_norm": 0.7007074952125549, "seconds": 8326.5} {"step": 1655, "loss": 0.4302511215209961, "data": 0.17424757778644562, "gap": 0.2560035288333893, "sigma": 0.6948705315589905, "kind": "mosaic", "tokens": 187, "dense": 448, "grad_norm": 0.2850112318992615, "seconds": 8330.9} {"step": 1656, "loss": 0.37633195519447327, "data": 0.18072079122066498, "gap": 0.1956111639738083, "sigma": 0.5975814461708069, "kind": "mosaic", "tokens": 254, "dense": 448, "grad_norm": 0.16686813533306122, "seconds": 8335.5} {"step": 1657, "loss": 1.4009919166564941, "data": 0.23013582825660706, "gap": 1.1708561182022095, "sigma": 0.35122376680374146, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.42958635091781616, "seconds": 8339.6} {"step": 1658, "loss": 0.5158624649047852, "data": 0.23802462220191956, "gap": 0.277837872505188, "sigma": 0.5604804754257202, "kind": "mosaic", "tokens": 247, "dense": 448, "grad_norm": 0.1591702252626419, "seconds": 8344.1} {"step": 1659, "loss": 0.7006934881210327, "data": 0.324858158826828, "gap": 0.3758353292942047, "sigma": 0.8073840737342834, "kind": "mosaic", "tokens": 282, "dense": 448, "grad_norm": 0.1652640998363495, "seconds": 8348.7} {"step": 1660, "loss": 0.5082634091377258, "data": 0.1891898661851883, "gap": 0.31907355785369873, "sigma": 0.6229444742202759, "kind": "mosaic", "tokens": 217, "dense": 448, "grad_norm": 0.16954436898231506, "seconds": 8353.2} {"step": 1661, "loss": 0.895728588104248, "data": 0.18881700932979584, "gap": 0.706911563873291, "sigma": 0.979132354259491, "kind": "mosaic", "tokens": 224, "dense": 448, "grad_norm": 0.6344785690307617, "seconds": 8357.7} {"step": 1662, "loss": 0.5815675854682922, "data": 0.21493496000766754, "gap": 0.3666326403617859, "sigma": 0.45783668756484985, "kind": "mosaic", "tokens": 179, "dense": 448, "grad_norm": 0.29444193840026855, "seconds": 8362.2} {"step": 1663, "loss": 0.40247172117233276, "data": 0.40247172117233276, "gap": null, "sigma": 0.9836114048957825, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.43781667947769165, "seconds": 8365.1} {"step": 1664, "loss": 0.8405012488365173, "data": 0.28212565183639526, "gap": 0.5583755970001221, "sigma": 0.91651850938797, "kind": "mosaic", "tokens": 242, "dense": 448, "grad_norm": 0.4168262779712677, "seconds": 8369.6} {"step": 1665, "loss": 0.37380942702293396, "data": 0.37380942702293396, "gap": null, "sigma": 0.9080700278282166, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.11573927104473114, "seconds": 8372.4} {"step": 1666, "loss": 0.3880687952041626, "data": 0.3880687952041626, "gap": null, "sigma": 0.9986225366592407, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.4339071214199066, "seconds": 8375.2} {"step": 1667, "loss": 0.22244031727313995, "data": 0.22244031727313995, "gap": null, "sigma": 0.37008994817733765, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.15017150342464447, "seconds": 8378.1} {"step": 1668, "loss": 0.5720159411430359, "data": 0.23714597523212433, "gap": 0.33486998081207275, "sigma": 0.9610300064086914, "kind": "mosaic", "tokens": 172, "dense": 448, "grad_norm": 0.31672385334968567, "seconds": 8382.6} {"step": 1669, "loss": 0.49238038063049316, "data": 0.22833819687366486, "gap": 0.2640421688556671, "sigma": 0.777835488319397, "kind": "mosaic", "tokens": 205, "dense": 448, "grad_norm": 0.19988787174224854, "seconds": 8387.1} {"step": 1670, "loss": 15.847990036010742, "data": 2.4958395957946777, "gap": 13.352149963378906, "sigma": 0.008885146118700504, "kind": "mosaic", "tokens": 239, "dense": 448, "grad_norm": 12.597951889038086, "seconds": 8391.6} {"step": 1671, "loss": 0.31000760197639465, "data": 0.131772980093956, "gap": 0.17823462188243866, "sigma": 0.4361647963523865, "kind": "mosaic", "tokens": 173, "dense": 448, "grad_norm": 0.18462318181991577, "seconds": 8396.1} {"step": 1672, "loss": 0.3971003293991089, "data": 0.17210303246974945, "gap": 0.22499728202819824, "sigma": 0.7075631022453308, "kind": "mosaic", "tokens": 202, "dense": 448, "grad_norm": 0.23846124112606049, "seconds": 8400.3} {"step": 1673, "loss": 0.5179673433303833, "data": 0.20413458347320557, "gap": 0.3138327896595001, "sigma": 0.5996171832084656, "kind": "mosaic", "tokens": 223, "dense": 448, "grad_norm": 0.1894093006849289, "seconds": 8404.9} {"step": 1674, "loss": 0.54248046875, "data": 0.1125333234667778, "gap": 0.429947167634964, "sigma": 0.6767571568489075, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.23237544298171997, "seconds": 8409.0} {"step": 1675, "loss": 0.7545969486236572, "data": 0.2865016460418701, "gap": 0.4680952727794647, "sigma": 0.26822397112846375, "kind": "mosaic", "tokens": 208, "dense": 448, "grad_norm": 0.2725636661052704, "seconds": 8413.7} {"step": 1676, "loss": 13.3843412399292, "data": 4.809527397155762, "gap": 8.574813842773438, "sigma": 0.028580551967024803, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 14.333508491516113, "seconds": 8418.1} {"step": 1677, "loss": 0.6816824674606323, "data": 0.14095231890678406, "gap": 0.5407301783561707, "sigma": 0.9785293936729431, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.8984217047691345, "seconds": 8422.3} {"step": 1678, "loss": 32.05704879760742, "data": 3.8120038509368896, "gap": 28.245046615600586, "sigma": 0.010495592840015888, "kind": "mosaic", "tokens": 219, "dense": 448, "grad_norm": 21.58635711669922, "seconds": 8427.0} {"step": 1679, "loss": 0.9793719053268433, "data": 0.34934136271476746, "gap": 0.6300305128097534, "sigma": 0.16992530226707458, "kind": "mosaic", "tokens": 265, "dense": 448, "grad_norm": 0.5863739252090454, "seconds": 8431.7} {"step": 1680, "loss": 0.45706290006637573, "data": 0.17462487518787384, "gap": 0.2824380397796631, "sigma": 0.6969552636146545, "kind": "mosaic", "tokens": 203, "dense": 448, "grad_norm": 0.17368583381175995, "seconds": 8436.2} {"step": 1681, "loss": 0.49041640758514404, "data": 0.26353490352630615, "gap": 0.22688151895999908, "sigma": 0.8315995335578918, "kind": "mosaic", "tokens": 222, "dense": 448, "grad_norm": 0.2278827577829361, "seconds": 8440.8} {"step": 1682, "loss": 0.33087870478630066, "data": 0.33087870478630066, "gap": null, "sigma": 0.979454517364502, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.27061668038368225, "seconds": 8443.6} {"step": 1683, "loss": 0.49454790353775024, "data": 0.49454790353775024, "gap": null, "sigma": 0.992022693157196, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.40276965498924255, "seconds": 8446.4} {"step": 1684, "loss": 0.7031530737876892, "data": 0.18962128460407257, "gap": 0.5135318040847778, "sigma": 0.8116737008094788, "kind": "mosaic", "tokens": 226, "dense": 448, "grad_norm": 0.3194410800933838, "seconds": 8451.0} {"step": 1685, "loss": 0.5348843932151794, "data": 0.21784056723117828, "gap": 0.31704384088516235, "sigma": 0.6534914374351501, "kind": "mosaic", "tokens": 246, "dense": 448, "grad_norm": 0.17964619398117065, "seconds": 8455.5} {"step": 1686, "loss": 0.4640064835548401, "data": 0.322755366563797, "gap": 0.1412511169910431, "sigma": 0.9904740452766418, "kind": "mosaic", "tokens": 237, "dense": 448, "grad_norm": 0.3194381892681122, "seconds": 8460.1} {"step": 1687, "loss": 0.48906442523002625, "data": 0.185316264629364, "gap": 0.30374816060066223, "sigma": 0.7505090236663818, "kind": "mosaic", "tokens": 217, "dense": 576, "grad_norm": 0.18346388638019562, "seconds": 8464.6} {"step": 1688, "loss": 0.6163613796234131, "data": 0.2549881637096405, "gap": 0.3613731861114502, "sigma": 0.8067760467529297, "kind": "mosaic", "tokens": 241, "dense": 448, "grad_norm": 0.3608759641647339, "seconds": 8469.2} {"step": 1689, "loss": 0.333263099193573, "data": 0.10677790641784668, "gap": 0.2264852076768875, "sigma": 0.628372848033905, "kind": "mosaic", "tokens": 189, "dense": 448, "grad_norm": 0.12261666357517242, "seconds": 8473.7} {"step": 1690, "loss": 0.4154701828956604, "data": 0.17339099943637848, "gap": 0.24207918345928192, "sigma": 0.915192723274231, "kind": "mosaic", "tokens": 198, "dense": 448, "grad_norm": 0.46372681856155396, "seconds": 8478.3} {"step": 1691, "loss": 0.6656615734100342, "data": 0.2669885456562042, "gap": 0.39867299795150757, "sigma": 0.9379357695579529, "kind": "mosaic", "tokens": 213, "dense": 448, "grad_norm": 0.5457644462585449, "seconds": 8482.8} {"step": 1692, "loss": 1.4667744636535645, "data": 0.4110559821128845, "gap": 1.0557184219360352, "sigma": 0.2870543599128723, "kind": "mosaic", "tokens": 176, "dense": 448, "grad_norm": 0.3999749422073364, "seconds": 8487.5} {"step": 1693, "loss": 0.7262560725212097, "data": 0.296083003282547, "gap": 0.4301730692386627, "sigma": 0.4506506621837616, "kind": "mosaic", "tokens": 254, "dense": 448, "grad_norm": 0.1748872548341751, "seconds": 8492.0} {"step": 1694, "loss": 0.309922456741333, "data": 0.309922456741333, "gap": null, "sigma": 0.6829853057861328, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.1426319181919098, "seconds": 8494.8} {"step": 1695, "loss": 0.5533945560455322, "data": 0.15165530145168304, "gap": 0.4017392694950104, "sigma": 0.6968883275985718, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.24384695291519165, "seconds": 8499.0} {"step": 1696, "loss": 0.5569049119949341, "data": 0.20750197768211365, "gap": 0.34940290451049805, "sigma": 0.7847759127616882, "kind": "mosaic", "tokens": 206, "dense": 448, "grad_norm": 0.24942195415496826, "seconds": 8503.4} {"step": 1697, "loss": 0.551138699054718, "data": 0.18628337979316711, "gap": 0.3648553192615509, "sigma": 0.8558889627456665, "kind": "mosaic", "tokens": 233, "dense": 448, "grad_norm": 1.4552116394042969, "seconds": 8508.0} {"step": 1698, "loss": 0.6076482534408569, "data": 0.17458686232566833, "gap": 0.4330613613128662, "sigma": 0.913856565952301, "kind": "mosaic", "tokens": 201, "dense": 448, "grad_norm": 1.1035300493240356, "seconds": 8512.2} {"step": 1699, "loss": 0.09086768329143524, "data": 0.09086768329143524, "gap": null, "sigma": 0.7014030814170837, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.09753841906785965, "seconds": 8515.1} {"step": 1700, "loss": 1.2098369598388672, "data": 0.27671077847480774, "gap": 0.9331262111663818, "sigma": 0.9827108979225159, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 1.2205650806427002, "seconds": 8519.6} {"step": 1701, "loss": 12.452228546142578, "data": 2.587681293487549, "gap": 9.864546775817871, "sigma": 0.03361928462982178, "kind": "mosaic", "tokens": 213, "dense": 448, "grad_norm": 9.827425956726074, "seconds": 8524.0} {"step": 1702, "loss": 0.2937347888946533, "data": 0.16401590406894684, "gap": 0.12971889972686768, "sigma": 0.7065370678901672, "kind": "mosaic", "tokens": 250, "dense": 448, "grad_norm": 0.10351086407899857, "seconds": 8528.5} {"step": 1703, "loss": 0.5142069458961487, "data": 0.14913706481456757, "gap": 0.3650698661804199, "sigma": 0.6112670302391052, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.16183403134346008, "seconds": 8532.5} {"step": 1704, "loss": 0.33841031789779663, "data": 0.33841031789779663, "gap": null, "sigma": 0.9795947074890137, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2682681977748871, "seconds": 8535.3} {"step": 1705, "loss": 0.30486923456192017, "data": 0.13507995009422302, "gap": 0.16978926956653595, "sigma": 0.3876408636569977, "kind": "mosaic", "tokens": 225, "dense": 448, "grad_norm": 0.12578755617141724, "seconds": 8540.0} {"step": 1706, "loss": 0.8491102457046509, "data": 0.32138529419898987, "gap": 0.5277249813079834, "sigma": 0.969257116317749, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.4648822247982025, "seconds": 8544.6} {"step": 1707, "loss": 0.2002582550048828, "data": 0.2002582550048828, "gap": null, "sigma": 0.7550355195999146, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.12742824852466583, "seconds": 8547.4} {"step": 1708, "loss": 0.42353302240371704, "data": 0.23972970247268677, "gap": 0.18380333483219147, "sigma": 0.9628182649612427, "kind": "mosaic", "tokens": 263, "dense": 448, "grad_norm": 0.317339152097702, "seconds": 8551.9} {"step": 1709, "loss": 0.9412072896957397, "data": 0.43576040863990784, "gap": 0.5054469108581543, "sigma": 0.32800692319869995, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.3876255452632904, "seconds": 8556.5} {"step": 1710, "loss": 0.7044885754585266, "data": 0.11184117943048477, "gap": 0.5926473736763, "sigma": 0.5568628907203674, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.4387650191783905, "seconds": 8560.8} {"step": 1711, "loss": 0.7879486083984375, "data": 0.13638776540756226, "gap": 0.6515608429908752, "sigma": 0.4110057055950165, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.4905140697956085, "seconds": 8565.0} {"step": 1712, "loss": 0.8349670171737671, "data": 0.2517872750759125, "gap": 0.583179771900177, "sigma": 0.9530853033065796, "kind": "mosaic", "tokens": 193, "dense": 448, "grad_norm": 0.6526331305503845, "seconds": 8569.5} {"step": 1713, "loss": 1.3539884090423584, "data": 0.3934059143066406, "gap": 0.960582435131073, "sigma": 0.23994815349578857, "kind": "mosaic", "tokens": 266, "dense": 448, "grad_norm": 1.1313295364379883, "seconds": 8574.1} {"step": 1714, "loss": 0.18639300763607025, "data": 0.18639300763607025, "gap": null, "sigma": 0.6894939541816711, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.1352231800556183, "seconds": 8576.9} {"step": 1715, "loss": 0.7612740993499756, "data": 0.1914779543876648, "gap": 0.5697961449623108, "sigma": 0.9900455474853516, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.9463137984275818, "seconds": 8581.3} {"step": 1716, "loss": 0.35838067531585693, "data": 0.16178230941295624, "gap": 0.1965983510017395, "sigma": 0.5590078830718994, "kind": "mosaic", "tokens": 205, "dense": 448, "grad_norm": 0.1409565955400467, "seconds": 8585.8} {"step": 1717, "loss": 0.23273682594299316, "data": 0.23273682594299316, "gap": null, "sigma": 0.7960026860237122, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.13255563378334045, "seconds": 8588.6} {"step": 1718, "loss": 0.3988789916038513, "data": 0.1879173368215561, "gap": 0.21096165478229523, "sigma": 0.7949718832969666, "kind": "mosaic", "tokens": 207, "dense": 448, "grad_norm": 0.47184622287750244, "seconds": 8593.3} {"step": 1719, "loss": 0.6998664736747742, "data": 0.2716416120529175, "gap": 0.4282248616218567, "sigma": 0.9517519474029541, "kind": "mosaic", "tokens": 197, "dense": 448, "grad_norm": 0.9068513512611389, "seconds": 8597.9} {"step": 1720, "loss": 5.015310764312744, "data": 1.0043407678604126, "gap": 4.010970115661621, "sigma": 0.1068900004029274, "kind": "mosaic", "tokens": 228, "dense": 448, "grad_norm": 2.7907183170318604, "seconds": 8602.4} {"step": 1721, "loss": 1.010434865951538, "data": 0.18430276215076447, "gap": 0.8261321187019348, "sigma": 0.43663784861564636, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.46852582693099976, "seconds": 8606.8} {"step": 1722, "loss": 0.4200071692466736, "data": 0.4200071692466736, "gap": null, "sigma": 0.9974561333656311, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.5331478714942932, "seconds": 8609.7} {"step": 1723, "loss": 0.35297155380249023, "data": 0.052748262882232666, "gap": 0.30022329092025757, "sigma": 0.7615898251533508, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.27073580026626587, "seconds": 8613.9} {"step": 1724, "loss": 0.830582320690155, "data": 0.1750478595495224, "gap": 0.6555344462394714, "sigma": 0.9176673293113708, "kind": "mosaic", "tokens": 157, "dense": 448, "grad_norm": 0.37361225485801697, "seconds": 8618.1} {"step": 1725, "loss": 0.492007851600647, "data": 0.17294463515281677, "gap": 0.3190632164478302, "sigma": 0.8068520426750183, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.4431726038455963, "seconds": 8622.2} {"step": 1726, "loss": 0.3211041986942291, "data": 0.3211041986942291, "gap": null, "sigma": 0.8728373050689697, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.15093840658664703, "seconds": 8625.1} {"step": 1727, "loss": 0.4629461467266083, "data": 0.1571798026561737, "gap": 0.30576634407043457, "sigma": 0.746263325214386, "kind": "mosaic", "tokens": 169, "dense": 448, "grad_norm": 0.20709781348705292, "seconds": 8629.4} {"step": 1728, "loss": 0.3843121826648712, "data": 0.22967571020126343, "gap": 0.1546364724636078, "sigma": 0.8764760494232178, "kind": "mosaic", "tokens": 277, "dense": 448, "grad_norm": 0.22596164047718048, "seconds": 8633.9} {"step": 1729, "loss": 0.6177667379379272, "data": 0.2232172191143036, "gap": 0.39454951882362366, "sigma": 0.8832027912139893, "kind": "mosaic", "tokens": 204, "dense": 448, "grad_norm": 0.3931237459182739, "seconds": 8638.4} {"step": 1730, "loss": 0.5676979422569275, "data": 0.1968940794467926, "gap": 0.3708038628101349, "sigma": 0.4466145634651184, "kind": "mosaic", "tokens": 191, "dense": 448, "grad_norm": 0.20536845922470093, "seconds": 8643.0} {"step": 1731, "loss": 0.5659658312797546, "data": 0.24661454558372498, "gap": 0.31935128569602966, "sigma": 0.8263524770736694, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.2837275564670563, "seconds": 8647.7} {"step": 1732, "loss": 0.40057191252708435, "data": 0.40057191252708435, "gap": null, "sigma": 0.8884390592575073, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.1522771418094635, "seconds": 8650.5} {"step": 1733, "loss": 0.33936184644699097, "data": 0.33936184644699097, "gap": null, "sigma": 0.9954643249511719, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3382023572921753, "seconds": 8653.3} {"step": 1734, "loss": 2.645937442779541, "data": 1.0833771228790283, "gap": 1.5625603199005127, "sigma": 0.10556480288505554, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 2.208294630050659, "seconds": 8657.8} {"step": 1735, "loss": 0.8384928703308105, "data": 0.28341561555862427, "gap": 0.5550772547721863, "sigma": 0.3607812523841858, "kind": "mosaic", "tokens": 201, "dense": 448, "grad_norm": 2.478513717651367, "seconds": 8662.2} {"step": 1736, "loss": 0.6079694032669067, "data": 0.2511869966983795, "gap": 0.3567824065685272, "sigma": 0.6226018071174622, "kind": "mosaic", "tokens": 238, "dense": 448, "grad_norm": 0.19684338569641113, "seconds": 8666.7} {"step": 1737, "loss": 0.8462505340576172, "data": 0.22459688782691956, "gap": 0.62165367603302, "sigma": 0.988797664642334, "kind": "mosaic", "tokens": 185, "dense": 448, "grad_norm": 1.765156626701355, "seconds": 8671.2} {"step": 1738, "loss": 0.386868417263031, "data": 0.1820869892835617, "gap": 0.2047814279794693, "sigma": 0.8667861223220825, "kind": "mosaic", "tokens": 216, "dense": 512, "grad_norm": 0.1482391506433487, "seconds": 8675.6} {"step": 1739, "loss": 0.5980804562568665, "data": 0.2562790811061859, "gap": 0.34180137515068054, "sigma": 0.7369697690010071, "kind": "mosaic", "tokens": 239, "dense": 448, "grad_norm": 0.19008876383304596, "seconds": 8680.1} {"step": 1740, "loss": 0.31650346517562866, "data": 0.31650346517562866, "gap": null, "sigma": 0.9927324056625366, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.34566617012023926, "seconds": 8683.0} {"step": 1741, "loss": 0.48349374532699585, "data": 0.18231503665447235, "gap": 0.3011786937713623, "sigma": 0.6452013254165649, "kind": "mosaic", "tokens": 213, "dense": 448, "grad_norm": 0.4426615536212921, "seconds": 8687.4} {"step": 1742, "loss": 0.7575092911720276, "data": 0.2242448925971985, "gap": 0.5332643985748291, "sigma": 0.96553635597229, "kind": "mosaic", "tokens": 147, "dense": 448, "grad_norm": 0.4371545910835266, "seconds": 8691.7} {"step": 1743, "loss": 0.3413030207157135, "data": 0.3413030207157135, "gap": null, "sigma": 0.8474776148796082, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.11839363723993301, "seconds": 8694.5} {"step": 1744, "loss": 1.0180588960647583, "data": 0.08248953521251678, "gap": 0.9355694055557251, "sigma": 0.8233808279037476, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.4003293216228485, "seconds": 8698.8} {"step": 1745, "loss": 0.47137197852134705, "data": 0.47137197852134705, "gap": null, "sigma": 0.10516412556171417, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.13004839420318604, "seconds": 8701.7} {"step": 1746, "loss": 0.6498785018920898, "data": 0.09273331612348557, "gap": 0.5571451783180237, "sigma": 0.8647935390472412, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.33670392632484436, "seconds": 8705.9} {"step": 1747, "loss": 0.7839305400848389, "data": 0.31532877683639526, "gap": 0.4686017334461212, "sigma": 0.31438323855400085, "kind": "mosaic", "tokens": 269, "dense": 448, "grad_norm": 0.22427718341350555, "seconds": 8710.4} {"step": 1748, "loss": 0.3902641534805298, "data": 0.3902641534805298, "gap": null, "sigma": 0.9493298530578613, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.22089868783950806, "seconds": 8713.2} {"step": 1749, "loss": 3.6593880653381348, "data": 0.8307843804359436, "gap": 2.828603744506836, "sigma": 0.10572757571935654, "kind": "mosaic", "tokens": 230, "dense": 448, "grad_norm": 2.2389142513275146, "seconds": 8717.7} {"step": 1750, "loss": 0.49832284450531006, "data": 0.12340673804283142, "gap": 0.37491610646247864, "sigma": 0.7893756031990051, "kind": "mosaic", "tokens": 186, "dense": 448, "grad_norm": 0.5165205001831055, "seconds": 8722.1} {"step": 1750, "eval": {"dense": 0.3090555363645156, "uniform2": 0.1850779348363479, "mosaic": 0.2537253890186548, "gap_uniform2": 0.5609348621219397, "gap_mosaic": 0.5726460969696442}, "eval_extent": {"1x1x1": 0.42550859946140906, "1x1x2": 0.32227283858006733, "1x1x4": 0.22816494014114141, "1x2x1": 0.30252244278133195, "1x2x2": 0.19612615572435682, "1x2x4": 0.12796776715093555, "1x4x1": 0.19630965101437603, "1x4x2": 0.1131066456678565, "1x4x4": 0.08203251483066688}} {"step": 1751, "loss": 0.4922263026237488, "data": 0.24146728217601776, "gap": 0.2507590353488922, "sigma": 0.40543442964553833, "kind": "mosaic", "tokens": 278, "dense": 448, "grad_norm": 0.18982331454753876, "seconds": 8970.9} {"step": 1752, "loss": 1.2080665826797485, "data": 0.3775319755077362, "gap": 0.8305346369743347, "sigma": 0.30485451221466064, "kind": "mosaic", "tokens": 212, "dense": 448, "grad_norm": 0.34533947706222534, "seconds": 8975.4} {"step": 1753, "loss": 0.5802593231201172, "data": 0.21258288621902466, "gap": 0.36767643690109253, "sigma": 0.6940491795539856, "kind": "mosaic", "tokens": 197, "dense": 448, "grad_norm": 0.20453031361103058, "seconds": 8980.0} {"step": 1754, "loss": 0.7377596497535706, "data": 0.23230703175067902, "gap": 0.5054526329040527, "sigma": 0.27630358934402466, "kind": "mosaic", "tokens": 216, "dense": 448, "grad_norm": 0.2960119843482971, "seconds": 8984.5} {"step": 1755, "loss": 0.48814743757247925, "data": 0.20893992483615875, "gap": 0.2792074978351593, "sigma": 0.6665709018707275, "kind": "mosaic", "tokens": 185, "dense": 448, "grad_norm": 0.2611277401447296, "seconds": 8989.0} {"step": 1756, "loss": 0.20202487707138062, "data": 0.20202487707138062, "gap": null, "sigma": 0.6867275834083557, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.10855942219495773, "seconds": 8991.8} {"step": 1757, "loss": 0.4819530248641968, "data": 0.22172905504703522, "gap": 0.26022398471832275, "sigma": 0.806391716003418, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.4446260929107666, "seconds": 8996.2} {"step": 1758, "loss": 0.5089944005012512, "data": 0.2655229866504669, "gap": 0.2434714287519455, "sigma": 0.6941328644752502, "kind": "mosaic", "tokens": 264, "dense": 448, "grad_norm": 0.2028035968542099, "seconds": 9000.8} {"step": 1759, "loss": 1.264040231704712, "data": 0.40440428256988525, "gap": 0.8596358895301819, "sigma": 0.21734677255153656, "kind": "mosaic", "tokens": 216, "dense": 448, "grad_norm": 0.8755666613578796, "seconds": 9005.3} {"step": 1760, "loss": 0.8745378255844116, "data": 0.3390520215034485, "gap": 0.5354858040809631, "sigma": 0.34623610973358154, "kind": "mosaic", "tokens": 264, "dense": 448, "grad_norm": 0.20855207741260529, "seconds": 9009.8} {"step": 1761, "loss": 0.3010953664779663, "data": 0.3010953664779663, "gap": null, "sigma": 0.9642571806907654, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3242103159427643, "seconds": 9012.7} {"step": 1762, "loss": 0.32470616698265076, "data": 0.32470616698265076, "gap": null, "sigma": 0.9854804873466492, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.26089468598365784, "seconds": 9015.7} {"step": 1763, "loss": 0.6338148713111877, "data": 0.22032655775547028, "gap": 0.4134882986545563, "sigma": 0.9255484938621521, "kind": "mosaic", "tokens": 194, "dense": 448, "grad_norm": 0.28783342242240906, "seconds": 9020.3} {"step": 1764, "loss": 0.343828022480011, "data": 0.343828022480011, "gap": null, "sigma": 0.8447589874267578, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.15492470562458038, "seconds": 9023.3} {"step": 1765, "loss": 0.3545030355453491, "data": 0.3545030355453491, "gap": null, "sigma": 0.9321630001068115, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.24997703731060028, "seconds": 9026.2} {"step": 1766, "loss": 0.47331511974334717, "data": 0.47331511974334717, "gap": null, "sigma": 0.9221360087394714, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.654153048992157, "seconds": 9029.1} {"step": 1767, "loss": 3.5928187370300293, "data": 0.8725759387016296, "gap": 2.720242738723755, "sigma": 0.12385988980531693, "kind": "mosaic", "tokens": 181, "dense": 448, "grad_norm": 2.5159380435943604, "seconds": 9033.6} {"step": 1768, "loss": 0.3003511130809784, "data": 0.133214071393013, "gap": 0.1671370416879654, "sigma": 0.6516607403755188, "kind": "mosaic", "tokens": 185, "dense": 448, "grad_norm": 0.15292175114154816, "seconds": 9038.1} {"step": 1769, "loss": 0.49509862065315247, "data": 0.23320025205612183, "gap": 0.26189836859703064, "sigma": 0.5879515409469604, "kind": "mosaic", "tokens": 248, "dense": 448, "grad_norm": 0.1493227332830429, "seconds": 9042.8} {"step": 1770, "loss": 1.0023386478424072, "data": 0.26024529337882996, "gap": 0.7420933842658997, "sigma": 0.9377836585044861, "kind": "mosaic", "tokens": 248, "dense": 448, "grad_norm": 0.5050286650657654, "seconds": 9047.2} {"step": 1771, "loss": 0.3087472915649414, "data": 0.0839703232049942, "gap": 0.2247769683599472, "sigma": 0.5286058783531189, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.21422183513641357, "seconds": 9051.4} {"step": 1772, "loss": 0.5333324074745178, "data": 0.19184991717338562, "gap": 0.3414824903011322, "sigma": 0.8675159215927124, "kind": "mosaic", "tokens": 194, "dense": 448, "grad_norm": 0.4174427092075348, "seconds": 9055.8} {"step": 1773, "loss": 0.3434242904186249, "data": 0.3434242904186249, "gap": null, "sigma": 0.8741481304168701, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.13692708313465118, "seconds": 9058.7} {"step": 1774, "loss": 0.23743265867233276, "data": 0.23743265867233276, "gap": null, "sigma": 0.6102765202522278, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.0699511468410492, "seconds": 9061.5} {"step": 1775, "loss": 0.6460991501808167, "data": 0.31844452023506165, "gap": 0.327654629945755, "sigma": 0.9658215045928955, "kind": "mosaic", "tokens": 181, "dense": 448, "grad_norm": 0.7671970129013062, "seconds": 9066.0} {"step": 1776, "loss": 0.44426584243774414, "data": 0.44426584243774414, "gap": null, "sigma": 0.9991317391395569, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.37881967425346375, "seconds": 9068.7} {"step": 1777, "loss": 0.5733949542045593, "data": 0.1728963702917099, "gap": 0.4004985988140106, "sigma": 0.8270677924156189, "kind": "mosaic", "tokens": 219, "dense": 448, "grad_norm": 0.2581977844238281, "seconds": 9073.1} {"step": 1778, "loss": 0.49909693002700806, "data": 0.49909693002700806, "gap": null, "sigma": 0.985234260559082, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.39548808336257935, "seconds": 9076.0} {"step": 1779, "loss": 0.42802315950393677, "data": 0.21229004859924316, "gap": 0.2157330960035324, "sigma": 0.8762543797492981, "kind": "mosaic", "tokens": 185, "dense": 448, "grad_norm": 0.7553852200508118, "seconds": 9080.4} {"step": 1780, "loss": 0.2916453778743744, "data": 0.2916453778743744, "gap": null, "sigma": 0.9438785910606384, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.127312570810318, "seconds": 9083.3} {"step": 1781, "loss": 0.18117551505565643, "data": 0.18117551505565643, "gap": null, "sigma": 0.6273133754730225, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.07243163138628006, "seconds": 9086.0} {"step": 1782, "loss": 0.42830654978752136, "data": 0.42830654978752136, "gap": null, "sigma": 0.9506284594535828, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2274957001209259, "seconds": 9088.9} {"step": 1783, "loss": 0.6390833258628845, "data": 0.20424985885620117, "gap": 0.43483346700668335, "sigma": 0.7589311599731445, "kind": "mosaic", "tokens": 165, "dense": 448, "grad_norm": 0.32185041904449463, "seconds": 9093.4} {"step": 1784, "loss": 22.863988876342773, "data": 3.083721876144409, "gap": 19.7802677154541, "sigma": 0.024975541979074478, "kind": "mosaic", "tokens": 187, "dense": 448, "grad_norm": 17.254615783691406, "seconds": 9097.9} {"step": 1785, "loss": 1.7799302339553833, "data": 0.08846618235111237, "gap": 1.6914640665054321, "sigma": 0.983870804309845, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 3.2112112045288086, "seconds": 9102.1} {"step": 1786, "loss": 0.23595952987670898, "data": 0.11568528413772583, "gap": 0.12027423828840256, "sigma": 0.4747004508972168, "kind": "mosaic", "tokens": 252, "dense": 448, "grad_norm": 0.11045002192258835, "seconds": 9106.7} {"step": 1787, "loss": 6.643525123596191, "data": 1.4816415309906006, "gap": 5.161883354187012, "sigma": 0.06724988669157028, "kind": "mosaic", "tokens": 242, "dense": 448, "grad_norm": 5.280479907989502, "seconds": 9111.2} {"step": 1788, "loss": 0.5703645944595337, "data": 0.25499284267425537, "gap": 0.31537172198295593, "sigma": 0.4407155215740204, "kind": "mosaic", "tokens": 265, "dense": 448, "grad_norm": 0.14349880814552307, "seconds": 9115.7} {"step": 1789, "loss": 0.44243115186691284, "data": 0.44243115186691284, "gap": null, "sigma": 0.9477269649505615, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.18454550206661224, "seconds": 9118.5} {"step": 1790, "loss": 0.34861651062965393, "data": 0.34861651062965393, "gap": null, "sigma": 0.9311885237693787, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.14959783852100372, "seconds": 9121.2} {"step": 1791, "loss": 0.37808042764663696, "data": 0.16844728589057922, "gap": 0.20963314175605774, "sigma": 0.7605984210968018, "kind": "mosaic", "tokens": 239, "dense": 448, "grad_norm": 0.20846131443977356, "seconds": 9125.8} {"step": 1792, "loss": 0.3035885691642761, "data": 0.1333925575017929, "gap": 0.1701960265636444, "sigma": 0.7932084202766418, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.22948803007602692, "seconds": 9130.3} {"step": 1793, "loss": 0.5598907470703125, "data": 0.2680419683456421, "gap": 0.2918487787246704, "sigma": 0.5048010945320129, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.1879730224609375, "seconds": 9134.8} {"step": 1794, "loss": 0.6717102527618408, "data": 0.11512431502342224, "gap": 0.5565859079360962, "sigma": 0.48537930846214294, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.240975022315979, "seconds": 9139.0} {"step": 1795, "loss": 0.478715717792511, "data": 0.26022621989250183, "gap": 0.21848948299884796, "sigma": 0.9306328296661377, "kind": "mosaic", "tokens": 266, "dense": 448, "grad_norm": 0.203182190656662, "seconds": 9143.4} {"step": 1796, "loss": 1.599799394607544, "data": 0.4680529236793518, "gap": 1.131746530532837, "sigma": 0.2600129246711731, "kind": "mosaic", "tokens": 174, "dense": 448, "grad_norm": 0.36986789107322693, "seconds": 9147.8} {"step": 1797, "loss": 0.7784003019332886, "data": 0.311858594417572, "gap": 0.46654173731803894, "sigma": 0.38241633772850037, "kind": "mosaic", "tokens": 259, "dense": 448, "grad_norm": 0.27106696367263794, "seconds": 9152.3} {"step": 1798, "loss": 0.21907007694244385, "data": 0.21907007694244385, "gap": null, "sigma": 0.38102036714553833, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.06065767630934715, "seconds": 9155.0} {"step": 1799, "loss": 0.6959280371665955, "data": 0.27039599418640137, "gap": 0.4255320429801941, "sigma": 0.8343157172203064, "kind": "mosaic", "tokens": 236, "dense": 448, "grad_norm": 0.6953896284103394, "seconds": 9159.5} {"step": 1800, "loss": 0.20538070797920227, "data": 0.13038380444049835, "gap": 0.07499689608812332, "sigma": 0.7415564656257629, "kind": "mosaic", "tokens": 230, "dense": 448, "grad_norm": 0.10734472423791885, "seconds": 9164.0} {"step": 1801, "loss": 0.4382043778896332, "data": 0.4382043778896332, "gap": null, "sigma": 0.9915019273757935, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2697437107563019, "seconds": 9166.8} {"step": 1802, "loss": 22.95489501953125, "data": 5.1476850509643555, "gap": 17.807209014892578, "sigma": 0.04434267431497574, "kind": "mosaic", "tokens": 183, "dense": 448, "grad_norm": 11.564712524414062, "seconds": 9171.3} {"step": 1803, "loss": 0.3953845798969269, "data": 0.19619455933570862, "gap": 0.19919002056121826, "sigma": 0.3786449730396271, "kind": "mosaic", "tokens": 267, "dense": 448, "grad_norm": 0.1560639590024948, "seconds": 9175.8} {"step": 1804, "loss": 0.4943264424800873, "data": 0.23196545243263245, "gap": 0.26236099004745483, "sigma": 0.5861932039260864, "kind": "mosaic", "tokens": 297, "dense": 576, "grad_norm": 0.16148583590984344, "seconds": 9180.3} {"step": 1805, "loss": 1.0155298709869385, "data": 0.3272641599178314, "gap": 0.6882656812667847, "sigma": 0.9582517147064209, "kind": "mosaic", "tokens": 263, "dense": 448, "grad_norm": 0.39371252059936523, "seconds": 9184.7} {"step": 1806, "loss": 0.25850602984428406, "data": 0.25850602984428406, "gap": null, "sigma": 0.7633128762245178, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.10230947285890579, "seconds": 9187.4} {"step": 1807, "loss": 0.4210078716278076, "data": 0.20088885724544525, "gap": 0.22011901438236237, "sigma": 0.6236246824264526, "kind": "mosaic", "tokens": 196, "dense": 448, "grad_norm": 0.16280153393745422, "seconds": 9191.9} {"step": 1808, "loss": 0.6463806629180908, "data": 0.2219526171684265, "gap": 0.4244280457496643, "sigma": 0.7657499313354492, "kind": "mosaic", "tokens": 200, "dense": 448, "grad_norm": 0.2682598829269409, "seconds": 9196.4} {"step": 1809, "loss": 0.7242786288261414, "data": 0.21460455656051636, "gap": 0.509674072265625, "sigma": 0.8600147366523743, "kind": "mosaic", "tokens": 252, "dense": 448, "grad_norm": 0.40860310196876526, "seconds": 9200.8} {"step": 1810, "loss": 1.6199254989624023, "data": 0.44403332471847534, "gap": 1.1758922338485718, "sigma": 0.29452505707740784, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.6662678122520447, "seconds": 9205.0} {"step": 1811, "loss": 0.8596001267433167, "data": 0.2854387164115906, "gap": 0.5741614103317261, "sigma": 0.3094981610774994, "kind": "mosaic", "tokens": 172, "dense": 448, "grad_norm": 0.2479272484779358, "seconds": 9209.4} {"step": 1812, "loss": 0.5298253893852234, "data": 0.225102961063385, "gap": 0.3047224283218384, "sigma": 0.7743606567382812, "kind": "mosaic", "tokens": 294, "dense": 576, "grad_norm": 0.19730496406555176, "seconds": 9213.9} {"step": 1813, "loss": 0.3797646164894104, "data": 0.16132502257823944, "gap": 0.21843960881233215, "sigma": 0.6601525545120239, "kind": "mosaic", "tokens": 257, "dense": 448, "grad_norm": 0.1619727611541748, "seconds": 9218.4} {"step": 1814, "loss": 0.6127519607543945, "data": 0.12536028027534485, "gap": 0.4873916506767273, "sigma": 0.8700461387634277, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.39693987369537354, "seconds": 9222.6} {"step": 1815, "loss": 0.4070962071418762, "data": 0.16889388859272003, "gap": 0.238202303647995, "sigma": 0.7630405426025391, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.3007378876209259, "seconds": 9227.1} {"step": 1816, "loss": 0.33596041798591614, "data": 0.16294339299201965, "gap": 0.17301702499389648, "sigma": 0.8653186559677124, "kind": "mosaic", "tokens": 291, "dense": 448, "grad_norm": 0.17437662184238434, "seconds": 9231.7} {"step": 1817, "loss": 0.7238740921020508, "data": 0.28506380319595337, "gap": 0.4388102889060974, "sigma": 0.8735060691833496, "kind": "mosaic", "tokens": 192, "dense": 448, "grad_norm": 0.40646225214004517, "seconds": 9236.1} {"step": 1818, "loss": 0.243477001786232, "data": 0.243477001786232, "gap": null, "sigma": 0.7436856627464294, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.05824030190706253, "seconds": 9238.9} {"step": 1819, "loss": 0.38058459758758545, "data": 0.38058459758758545, "gap": null, "sigma": 0.4270435571670532, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.062193166464567184, "seconds": 9241.6} {"step": 1820, "loss": 0.5253310203552246, "data": 0.06884073466062546, "gap": 0.45649027824401855, "sigma": 0.8801900148391724, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.6630149483680725, "seconds": 9245.7} {"step": 1821, "loss": 0.3626585602760315, "data": 0.3626585602760315, "gap": null, "sigma": 0.9694437384605408, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3266032636165619, "seconds": 9248.5} {"step": 1822, "loss": 0.6063501238822937, "data": 0.13106541335582733, "gap": 0.47528472542762756, "sigma": 0.6960331201553345, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.2903394103050232, "seconds": 9252.6} {"step": 1823, "loss": 0.7040694952011108, "data": 0.2150098979473114, "gap": 0.48905956745147705, "sigma": 0.9141229391098022, "kind": "mosaic", "tokens": 217, "dense": 448, "grad_norm": 0.4493906795978546, "seconds": 9257.0} {"step": 1824, "loss": 0.8616975545883179, "data": 0.3304786682128906, "gap": 0.5312188863754272, "sigma": 0.8996359705924988, "kind": "mosaic", "tokens": 271, "dense": 448, "grad_norm": 0.4406268298625946, "seconds": 9261.5} {"step": 1825, "loss": 0.4443119168281555, "data": 0.225986048579216, "gap": 0.21832585334777832, "sigma": 0.7922226190567017, "kind": "mosaic", "tokens": 197, "dense": 448, "grad_norm": 0.3152100741863251, "seconds": 9266.1} {"step": 1826, "loss": 0.3867514133453369, "data": 0.3867514133453369, "gap": null, "sigma": 0.9492192268371582, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.12792693078517914, "seconds": 9268.9} {"step": 1827, "loss": 0.6393895149230957, "data": 0.22995327413082123, "gap": 0.4094362258911133, "sigma": 0.9482722878456116, "kind": "mosaic", "tokens": 210, "dense": 448, "grad_norm": 0.3344247043132782, "seconds": 9273.3} {"step": 1828, "loss": 0.17917901277542114, "data": 0.17917901277542114, "gap": null, "sigma": 0.021497776731848717, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.04899419844150543, "seconds": 9276.2} {"step": 1829, "loss": 0.1951371133327484, "data": 0.09094898402690887, "gap": 0.10418812185525894, "sigma": 0.5585637092590332, "kind": "mosaic", "tokens": 231, "dense": 448, "grad_norm": 0.1066640317440033, "seconds": 9280.5} {"step": 1830, "loss": 0.2234029769897461, "data": 0.2234029769897461, "gap": null, "sigma": 0.8693828582763672, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.1568954885005951, "seconds": 9283.4} {"step": 1831, "loss": 0.4593939185142517, "data": 0.24688725173473358, "gap": 0.21250666677951813, "sigma": 0.8589140176773071, "kind": "mosaic", "tokens": 216, "dense": 448, "grad_norm": 0.2529510259628296, "seconds": 9288.0} {"step": 1832, "loss": 0.44079965353012085, "data": 0.18752364814281464, "gap": 0.253275990486145, "sigma": 0.6074203252792358, "kind": "mosaic", "tokens": 267, "dense": 448, "grad_norm": 1.6048630475997925, "seconds": 9292.6} {"step": 1833, "loss": 0.421373188495636, "data": 0.19968661665916443, "gap": 0.22168658673763275, "sigma": 0.7364602088928223, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.3436439633369446, "seconds": 9297.0} {"step": 1834, "loss": 0.46488499641418457, "data": 0.46488499641418457, "gap": null, "sigma": 0.9605180621147156, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2426627278327942, "seconds": 9299.8} {"step": 1835, "loss": 1.0601282119750977, "data": 0.6753143668174744, "gap": 0.3848137855529785, "sigma": 0.9998030662536621, "kind": "mosaic", "tokens": 180, "dense": 448, "grad_norm": 2.0217342376708984, "seconds": 9304.4} {"step": 1836, "loss": 0.5465027093887329, "data": 0.2372574806213379, "gap": 0.3092452585697174, "sigma": 0.8618253469467163, "kind": "mosaic", "tokens": 212, "dense": 448, "grad_norm": 0.33359378576278687, "seconds": 9308.8} {"step": 1837, "loss": 0.622683048248291, "data": 0.277963250875473, "gap": 0.344719797372818, "sigma": 0.6170687675476074, "kind": "mosaic", "tokens": 203, "dense": 448, "grad_norm": 0.13062208890914917, "seconds": 9313.4} {"step": 1838, "loss": 0.6896097660064697, "data": 0.22965508699417114, "gap": 0.4599546492099762, "sigma": 0.45873910188674927, "kind": "mosaic", "tokens": 211, "dense": 448, "grad_norm": 0.1819348931312561, "seconds": 9317.9} {"step": 1839, "loss": 0.45760613679885864, "data": 0.45760613679885864, "gap": null, "sigma": 0.9980952739715576, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.25542858242988586, "seconds": 9320.7} {"step": 1840, "loss": 1.0902928113937378, "data": 0.37517669796943665, "gap": 0.7151160836219788, "sigma": 0.2945709526538849, "kind": "mosaic", "tokens": 220, "dense": 448, "grad_norm": 0.32627546787261963, "seconds": 9325.2} {"step": 1841, "loss": 0.33100515604019165, "data": 0.1586572229862213, "gap": 0.17234793305397034, "sigma": 0.723516047000885, "kind": "mosaic", "tokens": 208, "dense": 448, "grad_norm": 0.36142316460609436, "seconds": 9329.6} {"step": 1842, "loss": 2.891839027404785, "data": 0.4289451837539673, "gap": 2.4628939628601074, "sigma": 0.20330901443958282, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.9381049275398254, "seconds": 9333.7} {"step": 1843, "loss": 0.46419334411621094, "data": 0.2095710188150406, "gap": 0.25462231040000916, "sigma": 0.6570915579795837, "kind": "mosaic", "tokens": 236, "dense": 448, "grad_norm": 0.15774209797382355, "seconds": 9338.3} {"step": 1844, "loss": 0.3563181757926941, "data": 0.3563181757926941, "gap": null, "sigma": 0.9884665012359619, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.28287431597709656, "seconds": 9341.1} {"step": 1845, "loss": 0.4856538474559784, "data": 0.4856538474559784, "gap": null, "sigma": 0.9874264001846313, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3600825369358063, "seconds": 9344.0} {"step": 1846, "loss": 17.618101119995117, "data": 4.193418502807617, "gap": 13.4246826171875, "sigma": 0.003509050700813532, "kind": "mosaic", "tokens": 218, "dense": 448, "grad_norm": 11.55666446685791, "seconds": 9348.4} {"step": 1847, "loss": 0.39458975195884705, "data": 0.39458975195884705, "gap": null, "sigma": 0.9610535502433777, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.1952173262834549, "seconds": 9351.2} {"step": 1848, "loss": 0.517085611820221, "data": 0.23131152987480164, "gap": 0.2857740819454193, "sigma": 0.5437006950378418, "kind": "mosaic", "tokens": 263, "dense": 448, "grad_norm": 0.16943463683128357, "seconds": 9355.8} {"step": 1849, "loss": 0.6260804533958435, "data": 0.23846615850925446, "gap": 0.38761430978775024, "sigma": 0.42885667085647583, "kind": "mosaic", "tokens": 236, "dense": 448, "grad_norm": 0.33624792098999023, "seconds": 9360.2} {"step": 1850, "loss": 0.1879933923482895, "data": 0.1879933923482895, "gap": null, "sigma": 0.7332957983016968, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.09198377281427383, "seconds": 9363.0} {"step": 1851, "loss": 0.8591098785400391, "data": 0.18841131031513214, "gap": 0.6706985831260681, "sigma": 0.948627769947052, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.9260655045509338, "seconds": 9367.3} {"step": 1852, "loss": 0.3375488817691803, "data": 0.1986865997314453, "gap": 0.13886228203773499, "sigma": 0.8840422034263611, "kind": "mosaic", "tokens": 209, "dense": 448, "grad_norm": 0.21482771635055542, "seconds": 9371.7} {"step": 1853, "loss": 0.19123263657093048, "data": 0.19123263657093048, "gap": null, "sigma": 0.6914812326431274, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.24971288442611694, "seconds": 9374.5} {"step": 1854, "loss": 0.5075455904006958, "data": 0.2226690649986267, "gap": 0.2848765552043915, "sigma": 0.36003541946411133, "kind": "mosaic", "tokens": 249, "dense": 448, "grad_norm": 0.15976017713546753, "seconds": 9379.1} {"step": 1855, "loss": 0.7802276015281677, "data": 0.22323393821716309, "gap": 0.5569936633110046, "sigma": 0.8186874985694885, "kind": "mosaic", "tokens": 217, "dense": 448, "grad_norm": 0.38276970386505127, "seconds": 9383.6} {"step": 1856, "loss": 3.1457653045654297, "data": 0.3029724359512329, "gap": 2.8427929878234863, "sigma": 0.9838695526123047, "kind": "mosaic", "tokens": 246, "dense": 448, "grad_norm": 1.8076711893081665, "seconds": 9388.1} {"step": 1857, "loss": 0.7345371246337891, "data": 0.2140599638223648, "gap": 0.5204771757125854, "sigma": 0.8430945873260498, "kind": "mosaic", "tokens": 269, "dense": 448, "grad_norm": 0.31737953424453735, "seconds": 9392.8} {"step": 1858, "loss": 0.447844535112381, "data": 0.24016115069389343, "gap": 0.20768338441848755, "sigma": 0.7441024780273438, "kind": "mosaic", "tokens": 277, "dense": 448, "grad_norm": 0.23500044643878937, "seconds": 9397.3} {"step": 1859, "loss": 0.6489553451538086, "data": 0.2113323211669922, "gap": 0.4376230537891388, "sigma": 0.42542436718940735, "kind": "mosaic", "tokens": 187, "dense": 448, "grad_norm": 1.9642175436019897, "seconds": 9401.8} {"step": 1860, "loss": 0.453355073928833, "data": 0.18809150159358978, "gap": 0.2652635872364044, "sigma": 0.635231614112854, "kind": "mosaic", "tokens": 230, "dense": 448, "grad_norm": 0.29587826132774353, "seconds": 9406.3} {"step": 1861, "loss": 0.6486620306968689, "data": 0.1930224746465683, "gap": 0.4556395411491394, "sigma": 0.7021804451942444, "kind": "mosaic", "tokens": 151, "dense": 448, "grad_norm": 0.2506428360939026, "seconds": 9410.7} {"step": 1862, "loss": 0.40258410573005676, "data": 0.19578532874584198, "gap": 0.20679877698421478, "sigma": 0.74516361951828, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.7610877156257629, "seconds": 9415.2} {"step": 1863, "loss": 0.36288565397262573, "data": 0.18037812411785126, "gap": 0.18250754475593567, "sigma": 0.8042358160018921, "kind": "mosaic", "tokens": 231, "dense": 448, "grad_norm": 0.22078412771224976, "seconds": 9419.6} {"step": 1864, "loss": 0.8232202529907227, "data": 0.21561257541179657, "gap": 0.6076076626777649, "sigma": 0.8935835957527161, "kind": "mosaic", "tokens": 207, "dense": 448, "grad_norm": 0.4381158649921417, "seconds": 9424.0} {"step": 1865, "loss": 0.8247151374816895, "data": 0.30866095423698425, "gap": 0.5160542130470276, "sigma": 0.9904812574386597, "kind": "mosaic", "tokens": 210, "dense": 448, "grad_norm": 0.695939302444458, "seconds": 9428.5} {"step": 1866, "loss": 0.5194749236106873, "data": 0.19700250029563904, "gap": 0.3224724233150482, "sigma": 0.8698437809944153, "kind": "mosaic", "tokens": 186, "dense": 448, "grad_norm": 0.2549305260181427, "seconds": 9433.0} {"step": 1867, "loss": 0.5671470761299133, "data": 0.24048279225826263, "gap": 0.3266642689704895, "sigma": 0.8111867904663086, "kind": "mosaic", "tokens": 217, "dense": 448, "grad_norm": 0.30811774730682373, "seconds": 9437.5} {"step": 1868, "loss": 0.5338906049728394, "data": 0.2275545299053192, "gap": 0.30633610486984253, "sigma": 0.8633973598480225, "kind": "mosaic", "tokens": 276, "dense": 448, "grad_norm": 0.3604991137981415, "seconds": 9442.2} {"step": 1869, "loss": 0.6035542488098145, "data": 0.09379646182060242, "gap": 0.5097578167915344, "sigma": 0.8634383082389832, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.7611007690429688, "seconds": 9446.5} {"step": 1870, "loss": 0.6342345476150513, "data": 0.1505391001701355, "gap": 0.48369544744491577, "sigma": 0.9106093645095825, "kind": "mosaic", "tokens": 215, "dense": 448, "grad_norm": 0.2430712729692459, "seconds": 9451.0} {"step": 1871, "loss": 0.46507754921913147, "data": 0.24355801939964294, "gap": 0.22151952981948853, "sigma": 0.7549047470092773, "kind": "mosaic", "tokens": 220, "dense": 448, "grad_norm": 0.2461412250995636, "seconds": 9455.5} {"step": 1872, "loss": 0.23625916242599487, "data": 0.23625916242599487, "gap": null, "sigma": 0.8545709848403931, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2404809594154358, "seconds": 9458.3} {"step": 1873, "loss": 0.2488132268190384, "data": 0.2488132268190384, "gap": null, "sigma": 0.6341221928596497, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.11307187378406525, "seconds": 9461.1} {"step": 1874, "loss": 0.37097227573394775, "data": 0.10713748633861542, "gap": 0.26383477449417114, "sigma": 0.8729689717292786, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.2860071063041687, "seconds": 9465.2} {"step": 1875, "loss": 0.689430832862854, "data": 0.2915078103542328, "gap": 0.3979230523109436, "sigma": 0.3972129225730896, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.3785829544067383, "seconds": 9469.6} {"step": 1876, "loss": 0.9456132054328918, "data": 0.45881813764572144, "gap": 0.4867950677871704, "sigma": 0.9901885986328125, "kind": "mosaic", "tokens": 238, "dense": 448, "grad_norm": 0.6315404772758484, "seconds": 9474.0} {"step": 1877, "loss": 1.1660146713256836, "data": 0.4290401041507721, "gap": 0.7369745373725891, "sigma": 0.9537861347198486, "kind": "mosaic", "tokens": 279, "dense": 448, "grad_norm": 0.837799608707428, "seconds": 9478.6} {"step": 1878, "loss": 0.8255940079689026, "data": 0.23273272812366486, "gap": 0.5928612947463989, "sigma": 0.6526350975036621, "kind": "mosaic", "tokens": 192, "dense": 448, "grad_norm": 0.3254748284816742, "seconds": 9482.9} {"step": 1879, "loss": 1.0494825839996338, "data": 0.2146589159965515, "gap": 0.834823727607727, "sigma": 0.9019454717636108, "kind": "mosaic", "tokens": 194, "dense": 448, "grad_norm": 0.32236090302467346, "seconds": 9487.5} {"step": 1880, "loss": 0.2763134837150574, "data": 0.2763134837150574, "gap": null, "sigma": 0.9571199417114258, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.21788789331912994, "seconds": 9490.3} {"step": 1881, "loss": 0.6251684427261353, "data": 0.24674926698207855, "gap": 0.3784191906452179, "sigma": 0.7578032612800598, "kind": "mosaic", "tokens": 231, "dense": 448, "grad_norm": 0.48643338680267334, "seconds": 9494.9} {"step": 1882, "loss": 0.3346209228038788, "data": 0.22732199728488922, "gap": 0.10729892551898956, "sigma": 0.9360959529876709, "kind": "mosaic", "tokens": 199, "dense": 448, "grad_norm": 0.226141557097435, "seconds": 9499.3} {"step": 1883, "loss": 0.6048402786254883, "data": 0.24713961780071259, "gap": 0.3577006757259369, "sigma": 0.8756887316703796, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.7095493674278259, "seconds": 9503.8} {"step": 1884, "loss": 0.3665076494216919, "data": 0.06925806403160095, "gap": 0.29724958539009094, "sigma": 0.4913748502731323, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.21050888299942017, "seconds": 9508.1} {"step": 1885, "loss": 0.6276627779006958, "data": 0.17577937245368958, "gap": 0.4518834352493286, "sigma": 0.7792412042617798, "kind": "mosaic", "tokens": 199, "dense": 448, "grad_norm": 0.5556933283805847, "seconds": 9512.4} {"step": 1886, "loss": 0.7342725396156311, "data": 0.7342725396156311, "gap": null, "sigma": 0.056586429476737976, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.30250272154808044, "seconds": 9515.3} {"step": 1887, "loss": 0.8087910413742065, "data": 0.23047325015068054, "gap": 0.5783178210258484, "sigma": 0.8126043081283569, "kind": "mosaic", "tokens": 240, "dense": 448, "grad_norm": 0.5397114753723145, "seconds": 9519.8} {"step": 1888, "loss": 0.25463563203811646, "data": 0.25463563203811646, "gap": null, "sigma": 0.8900747299194336, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.16088593006134033, "seconds": 9522.6} {"step": 1889, "loss": 1.5657835006713867, "data": 0.4538816809654236, "gap": 1.111901879310608, "sigma": 0.15848933160305023, "kind": "mosaic", "tokens": 226, "dense": 448, "grad_norm": 0.6485915780067444, "seconds": 9527.1} {"step": 1890, "loss": 0.403978168964386, "data": 0.403978168964386, "gap": null, "sigma": 0.9934735298156738, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3837203085422516, "seconds": 9529.8} {"step": 1891, "loss": 0.790664553642273, "data": 0.2506163418292999, "gap": 0.5400482416152954, "sigma": 0.5680558681488037, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.2101828157901764, "seconds": 9534.2} {"step": 1892, "loss": 0.34602320194244385, "data": 0.026093600317835808, "gap": 0.3199295997619629, "sigma": 0.862929105758667, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 0.19502004981040955, "seconds": 9538.1} {"step": 1893, "loss": 0.28260713815689087, "data": 0.28260713815689087, "gap": null, "sigma": 0.896030604839325, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.16197288036346436, "seconds": 9541.0} {"step": 1894, "loss": 0.31871044635772705, "data": 0.31871044635772705, "gap": null, "sigma": 0.7304611206054688, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.14841417968273163, "seconds": 9543.9} {"step": 1895, "loss": 2.003450870513916, "data": 0.5605863332748413, "gap": 1.4428646564483643, "sigma": 0.17567934095859528, "kind": "mosaic", "tokens": 199, "dense": 448, "grad_norm": 0.7975695729255676, "seconds": 9548.3} {"step": 1896, "loss": 0.6533153653144836, "data": 0.06615352630615234, "gap": 0.5871618390083313, "sigma": 0.866733729839325, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.45120275020599365, "seconds": 9552.6} {"step": 1897, "loss": 0.6260743737220764, "data": 0.2920742630958557, "gap": 0.3340001106262207, "sigma": 0.8840846419334412, "kind": "mosaic", "tokens": 211, "dense": 448, "grad_norm": 0.3786766231060028, "seconds": 9557.1} {"step": 1898, "loss": 0.22323767840862274, "data": 0.22323767840862274, "gap": null, "sigma": 0.565712034702301, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.08401668071746826, "seconds": 9559.9} {"step": 1899, "loss": 0.4456254243850708, "data": 0.17392560839653015, "gap": 0.27169981598854065, "sigma": 0.5433105826377869, "kind": "mosaic", "tokens": 172, "dense": 448, "grad_norm": 0.20083193480968475, "seconds": 9564.3} {"step": 1900, "loss": 0.43298712372779846, "data": 0.43298712372779846, "gap": null, "sigma": 0.9176280498504639, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.14994415640830994, "seconds": 9567.1} {"step": 1901, "loss": 0.47898608446121216, "data": 0.17823322117328644, "gap": 0.3007528781890869, "sigma": 0.5983017086982727, "kind": "mosaic", "tokens": 233, "dense": 448, "grad_norm": 0.19130124151706696, "seconds": 9571.8} {"step": 1902, "loss": 0.4472602307796478, "data": 0.0934162437915802, "gap": 0.3538439869880676, "sigma": 0.7970174551010132, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 0.24668294191360474, "seconds": 9575.9} {"step": 1903, "loss": 2.9556612968444824, "data": 0.773840069770813, "gap": 2.181821346282959, "sigma": 0.13898873329162598, "kind": "mosaic", "tokens": 205, "dense": 448, "grad_norm": 1.3456138372421265, "seconds": 9580.3} {"step": 1904, "loss": 1.0851900577545166, "data": 0.3152673542499542, "gap": 0.7699227333068848, "sigma": 0.1794966757297516, "kind": "mosaic", "tokens": 197, "dense": 448, "grad_norm": 0.7697901129722595, "seconds": 9585.0} {"step": 1905, "loss": 0.3980124294757843, "data": 0.16832588613033295, "gap": 0.22968654334545135, "sigma": 0.7429895401000977, "kind": "mosaic", "tokens": 249, "dense": 448, "grad_norm": 0.35885992646217346, "seconds": 9589.5} {"step": 1906, "loss": 0.8395388126373291, "data": 0.2620410621166229, "gap": 0.5774977207183838, "sigma": 0.9136096835136414, "kind": "mosaic", "tokens": 213, "dense": 448, "grad_norm": 1.5930041074752808, "seconds": 9594.1} {"step": 1907, "loss": 0.48646074533462524, "data": 0.22718589007854462, "gap": 0.25927484035491943, "sigma": 0.8195105791091919, "kind": "mosaic", "tokens": 244, "dense": 448, "grad_norm": 0.20262369513511658, "seconds": 9598.6} {"step": 1908, "loss": 1.2825976610183716, "data": 0.2525792419910431, "gap": 1.0300184488296509, "sigma": 0.8866308331489563, "kind": "mosaic", "tokens": 180, "dense": 448, "grad_norm": 0.4577101171016693, "seconds": 9603.1} {"step": 1909, "loss": 0.6016460061073303, "data": 0.22719816863536835, "gap": 0.37444785237312317, "sigma": 0.859393298625946, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.5508343577384949, "seconds": 9607.4} {"step": 1910, "loss": 0.49666357040405273, "data": 0.20105108618736267, "gap": 0.29561248421669006, "sigma": 0.8641841411590576, "kind": "mosaic", "tokens": 206, "dense": 448, "grad_norm": 0.20052184164524078, "seconds": 9612.0} {"step": 1911, "loss": 0.3006277084350586, "data": 0.15428949892520905, "gap": 0.14633820950984955, "sigma": 0.6517595648765564, "kind": "mosaic", "tokens": 249, "dense": 448, "grad_norm": 0.5961738228797913, "seconds": 9616.5} {"step": 1912, "loss": 0.5161480903625488, "data": 0.27752962708473206, "gap": 0.23861847817897797, "sigma": 0.9279775619506836, "kind": "mosaic", "tokens": 248, "dense": 448, "grad_norm": 0.3153591454029083, "seconds": 9621.0} {"step": 1913, "loss": 0.41294950246810913, "data": 0.15212616324424744, "gap": 0.2608233392238617, "sigma": 0.820780336856842, "kind": "mosaic", "tokens": 179, "dense": 448, "grad_norm": 0.15339526534080505, "seconds": 9625.3} {"step": 1914, "loss": 0.30206865072250366, "data": 0.1347048580646515, "gap": 0.16736377775669098, "sigma": 0.34556981921195984, "kind": "mosaic", "tokens": 224, "dense": 448, "grad_norm": 0.1997552216053009, "seconds": 9629.8} {"step": 1915, "loss": 0.7028588056564331, "data": 0.2646220028400421, "gap": 0.438236802816391, "sigma": 0.861661434173584, "kind": "mosaic", "tokens": 249, "dense": 448, "grad_norm": 0.3324490189552307, "seconds": 9634.3} {"step": 1916, "loss": 0.44822901487350464, "data": 0.44822901487350464, "gap": null, "sigma": 0.950702428817749, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2535193860530853, "seconds": 9637.1} {"step": 1917, "loss": 1.5965471267700195, "data": 0.4736346900463104, "gap": 1.1229124069213867, "sigma": 0.24826167523860931, "kind": "mosaic", "tokens": 181, "dense": 448, "grad_norm": 0.5522674918174744, "seconds": 9641.6} {"step": 1918, "loss": 0.39529746770858765, "data": 0.1718614250421524, "gap": 0.22343605756759644, "sigma": 0.38408157229423523, "kind": "mosaic", "tokens": 200, "dense": 448, "grad_norm": 0.27072635293006897, "seconds": 9646.1} {"step": 1919, "loss": 0.8921281099319458, "data": 0.26506564021110535, "gap": 0.6270624399185181, "sigma": 0.8606554269790649, "kind": "mosaic", "tokens": 265, "dense": 448, "grad_norm": 1.3109347820281982, "seconds": 9650.7} {"step": 1920, "loss": 0.9702404141426086, "data": 0.37725579738616943, "gap": 0.5929846167564392, "sigma": 0.1875341832637787, "kind": "mosaic", "tokens": 276, "dense": 448, "grad_norm": 0.5422198176383972, "seconds": 9655.2} {"step": 1921, "loss": 1.139386773109436, "data": 0.35991808772087097, "gap": 0.7794686555862427, "sigma": 0.2648078203201294, "kind": "mosaic", "tokens": 223, "dense": 448, "grad_norm": 0.4180464744567871, "seconds": 9659.6} {"step": 1922, "loss": 0.29228705167770386, "data": 0.12535011768341064, "gap": 0.1669369488954544, "sigma": 0.7282798886299133, "kind": "mosaic", "tokens": 182, "dense": 448, "grad_norm": 0.17955569922924042, "seconds": 9664.2} {"step": 1923, "loss": 0.6795258522033691, "data": 0.288603276014328, "gap": 0.3909226059913635, "sigma": 0.43386000394821167, "kind": "mosaic", "tokens": 277, "dense": 448, "grad_norm": 0.1931370347738266, "seconds": 9668.8} {"step": 1924, "loss": 8.09012222290039, "data": 2.2195258140563965, "gap": 5.870596885681152, "sigma": 0.05863872915506363, "kind": "mosaic", "tokens": 238, "dense": 448, "grad_norm": 5.98729133605957, "seconds": 9673.3} {"step": 1925, "loss": 0.45032966136932373, "data": 0.45032966136932373, "gap": null, "sigma": 0.932781994342804, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2520724833011627, "seconds": 9676.1} {"step": 1926, "loss": 0.6598349809646606, "data": 0.32169511914253235, "gap": 0.3381398320198059, "sigma": 0.9419991970062256, "kind": "mosaic", "tokens": 170, "dense": 448, "grad_norm": 0.3125684857368469, "seconds": 9680.6} {"step": 1927, "loss": 0.5021758079528809, "data": 0.19622516632080078, "gap": 0.3059506118297577, "sigma": 0.7919990420341492, "kind": "mosaic", "tokens": 250, "dense": 448, "grad_norm": 0.2667343318462372, "seconds": 9685.2} {"step": 1928, "loss": 0.35829466581344604, "data": 0.15906982123851776, "gap": 0.1992248296737671, "sigma": 0.6894808411598206, "kind": "mosaic", "tokens": 237, "dense": 448, "grad_norm": 0.13098721206188202, "seconds": 9689.7} {"step": 1929, "loss": 0.4171498715877533, "data": 0.4171498715877533, "gap": null, "sigma": 0.9668997526168823, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2617061734199524, "seconds": 9692.4} {"step": 1930, "loss": 0.24641327559947968, "data": 0.24641327559947968, "gap": null, "sigma": 0.8590896129608154, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.15795478224754333, "seconds": 9695.3} {"step": 1931, "loss": 0.8454611897468567, "data": 0.30720841884613037, "gap": 0.5382527709007263, "sigma": 0.4081938862800598, "kind": "mosaic", "tokens": 228, "dense": 448, "grad_norm": 0.19188134372234344, "seconds": 9699.8} {"step": 1932, "loss": 0.4555387496948242, "data": 0.20601072907447815, "gap": 0.24952800571918488, "sigma": 0.66887366771698, "kind": "mosaic", "tokens": 241, "dense": 448, "grad_norm": 0.2193724364042282, "seconds": 9704.4} {"step": 1933, "loss": 0.47318610548973083, "data": 0.47318610548973083, "gap": null, "sigma": 0.9330216646194458, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.2997860014438629, "seconds": 9707.2} {"step": 1934, "loss": 0.40504488348960876, "data": 0.24978184700012207, "gap": 0.1552630364894867, "sigma": 0.8680221438407898, "kind": "mosaic", "tokens": 193, "dense": 448, "grad_norm": 0.2679409086704254, "seconds": 9711.7} {"step": 1935, "loss": 0.4435756206512451, "data": 0.17665909230709076, "gap": 0.26691654324531555, "sigma": 0.6419216394424438, "kind": "mosaic", "tokens": 237, "dense": 448, "grad_norm": 0.22790174186229706, "seconds": 9716.2} {"step": 1936, "loss": 0.6896287202835083, "data": 0.267948716878891, "gap": 0.4216800332069397, "sigma": 0.47287502884864807, "kind": "mosaic", "tokens": 244, "dense": 448, "grad_norm": 0.22672143578529358, "seconds": 9720.7} {"step": 1937, "loss": 0.24210761487483978, "data": 0.04837295040488243, "gap": 0.19373466074466705, "sigma": 0.8745854496955872, "kind": "uniform", "tokens": 28, "dense": 448, "grad_norm": 0.2811664342880249, "seconds": 9725.0} {"step": 1938, "loss": 0.3277875781059265, "data": 0.12176825106143951, "gap": 0.2060193121433258, "sigma": 0.9297114610671997, "kind": "mosaic", "tokens": 211, "dense": 448, "grad_norm": 0.5460731387138367, "seconds": 9729.5} {"step": 1939, "loss": 0.31744346022605896, "data": 0.31744346022605896, "gap": null, "sigma": 0.9743668437004089, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.40679794549942017, "seconds": 9732.4} {"step": 1940, "loss": 0.2756849527359009, "data": 0.2756849527359009, "gap": null, "sigma": 0.885698676109314, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.09044106304645538, "seconds": 9735.3} {"step": 1941, "loss": 0.9876707792282104, "data": 0.2960178256034851, "gap": 0.6916529536247253, "sigma": 0.4069709777832031, "kind": "mosaic", "tokens": 169, "dense": 448, "grad_norm": 0.22927486896514893, "seconds": 9739.8} {"step": 1942, "loss": 0.35350704193115234, "data": 0.35350704193115234, "gap": null, "sigma": 0.956369936466217, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.21264362335205078, "seconds": 9742.7} {"step": 1943, "loss": 3.2816295623779297, "data": 0.9298925995826721, "gap": 2.3517370223999023, "sigma": 0.1408500075340271, "kind": "mosaic", "tokens": 193, "dense": 448, "grad_norm": 1.7151094675064087, "seconds": 9747.4} {"step": 1944, "loss": 0.6266343593597412, "data": 0.21125714480876923, "gap": 0.4153772294521332, "sigma": 0.8644717931747437, "kind": "mosaic", "tokens": 238, "dense": 448, "grad_norm": 0.4683779180049896, "seconds": 9752.0} {"step": 1945, "loss": 0.9576468467712402, "data": 0.4478652775287628, "gap": 0.509781539440155, "sigma": 0.2067091166973114, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.39912617206573486, "seconds": 9756.5} {"step": 1946, "loss": 0.5991236567497253, "data": 0.2573601305484772, "gap": 0.34176352620124817, "sigma": 0.4819517731666565, "kind": "mosaic", "tokens": 260, "dense": 448, "grad_norm": 0.1465994119644165, "seconds": 9760.9} {"step": 1947, "loss": 0.7877618074417114, "data": 0.14725083112716675, "gap": 0.6405109763145447, "sigma": 0.9426004886627197, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.6124942898750305, "seconds": 9765.1} {"step": 1948, "loss": 1.4611692428588867, "data": 0.4283083975315094, "gap": 1.0328608751296997, "sigma": 0.23292168974876404, "kind": "mosaic", "tokens": 187, "dense": 448, "grad_norm": 0.6615998148918152, "seconds": 9769.5} {"step": 1949, "loss": 0.7452423572540283, "data": 0.2476556897163391, "gap": 0.4975866675376892, "sigma": 0.5153200626373291, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.2734513282775879, "seconds": 9773.5} {"step": 1950, "loss": 0.1637597680091858, "data": 0.1637597680091858, "gap": null, "sigma": 0.8652151226997375, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.15657079219818115, "seconds": 9776.3} {"step": 1951, "loss": 0.5402190089225769, "data": 0.141841322183609, "gap": 0.3983776867389679, "sigma": 0.7435728311538696, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.4202820062637329, "seconds": 9780.6} {"step": 1952, "loss": 1.144334077835083, "data": 0.47782468795776367, "gap": 0.6665093898773193, "sigma": 0.9960194826126099, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.7145214080810547, "seconds": 9785.1} {"step": 1953, "loss": 0.31167012453079224, "data": 0.12983228266239166, "gap": 0.18183785676956177, "sigma": 0.6884264945983887, "kind": "mosaic", "tokens": 229, "dense": 448, "grad_norm": 0.18165534734725952, "seconds": 9789.7} {"step": 1954, "loss": 0.6981062293052673, "data": 0.24784360826015472, "gap": 0.4502626359462738, "sigma": 0.9679232835769653, "kind": "mosaic", "tokens": 231, "dense": 448, "grad_norm": 0.48701032996177673, "seconds": 9794.3} {"step": 1955, "loss": 0.3011314868927002, "data": 0.1778644174337387, "gap": 0.12326706200838089, "sigma": 0.8507726788520813, "kind": "mosaic", "tokens": 187, "dense": 448, "grad_norm": 0.1596371829509735, "seconds": 9798.9} {"step": 1956, "loss": 1.1758912801742554, "data": 0.22445325553417206, "gap": 0.9514380693435669, "sigma": 0.9438490867614746, "kind": "mosaic", "tokens": 257, "dense": 448, "grad_norm": 0.7417433857917786, "seconds": 9803.5} {"step": 1957, "loss": 0.5076464414596558, "data": 0.2117641121149063, "gap": 0.29588234424591064, "sigma": 0.9286202192306519, "kind": "mosaic", "tokens": 158, "dense": 448, "grad_norm": 0.2849789261817932, "seconds": 9807.8} {"step": 1958, "loss": 0.33912399411201477, "data": 0.33912399411201477, "gap": null, "sigma": 0.78435879945755, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.17319093644618988, "seconds": 9810.5} {"step": 1959, "loss": 0.6757417321205139, "data": 0.2249072939157486, "gap": 0.4508344233036041, "sigma": 0.8167232275009155, "kind": "mosaic", "tokens": 179, "dense": 448, "grad_norm": 0.382127970457077, "seconds": 9815.0} {"step": 1960, "loss": 0.6525791883468628, "data": 0.23230962455272675, "gap": 0.42026954889297485, "sigma": 0.4662105441093445, "kind": "mosaic", "tokens": 220, "dense": 448, "grad_norm": 0.2096283733844757, "seconds": 9819.6} {"step": 1961, "loss": 0.37128177285194397, "data": 0.37128177285194397, "gap": null, "sigma": 0.8661512136459351, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.21894614398479462, "seconds": 9822.3} {"step": 1962, "loss": 0.5317282676696777, "data": 0.15269845724105835, "gap": 0.3790298104286194, "sigma": 0.5596303939819336, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.19053272902965546, "seconds": 9826.5} {"step": 1963, "loss": 0.35706937313079834, "data": 0.35706937313079834, "gap": null, "sigma": 0.7960952520370483, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.19134193658828735, "seconds": 9829.3} {"step": 1964, "loss": 0.45285820960998535, "data": 0.2037036120891571, "gap": 0.24915459752082825, "sigma": 0.8931668400764465, "kind": "mosaic", "tokens": 228, "dense": 448, "grad_norm": 0.28182151913642883, "seconds": 9833.8} {"step": 1965, "loss": 0.6186186671257019, "data": 0.6186186671257019, "gap": null, "sigma": 0.9660919904708862, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.9690698385238647, "seconds": 9836.6} {"step": 1966, "loss": 0.3674125075340271, "data": 0.1918545663356781, "gap": 0.1755579560995102, "sigma": 0.5758040547370911, "kind": "mosaic", "tokens": 252, "dense": 448, "grad_norm": 0.13727545738220215, "seconds": 9841.3} {"step": 1967, "loss": 0.4595712423324585, "data": 0.19517703354358673, "gap": 0.26439419388771057, "sigma": 0.8182379603385925, "kind": "mosaic", "tokens": 245, "dense": 448, "grad_norm": 0.5318899154663086, "seconds": 9845.7} {"step": 1968, "loss": 1.0513544082641602, "data": 0.3018818199634552, "gap": 0.7494726181030273, "sigma": 0.9781084656715393, "kind": "mosaic", "tokens": 156, "dense": 448, "grad_norm": 1.2363240718841553, "seconds": 9849.9} {"step": 1969, "loss": 0.34327220916748047, "data": 0.1461087465286255, "gap": 0.19716347754001617, "sigma": 0.822695791721344, "kind": "mosaic", "tokens": 199, "dense": 448, "grad_norm": 0.26920416951179504, "seconds": 9854.3} {"step": 1970, "loss": 0.4716431200504303, "data": 0.20566076040267944, "gap": 0.26598235964775085, "sigma": 0.5572866797447205, "kind": "uniform", "tokens": 224, "dense": 448, "grad_norm": 0.20721298456192017, "seconds": 9858.9} {"step": 1971, "loss": 1.0652639865875244, "data": 0.1390756070613861, "gap": 0.9261883497238159, "sigma": 0.9864957332611084, "kind": "uniform", "tokens": 56, "dense": 448, "grad_norm": 1.3558324575424194, "seconds": 9863.2} {"step": 1972, "loss": 0.714440107345581, "data": 0.2734259366989136, "gap": 0.4410141408443451, "sigma": 0.536274790763855, "kind": "mosaic", "tokens": 212, "dense": 448, "grad_norm": 0.19761116802692413, "seconds": 9867.9} {"step": 1973, "loss": 0.7029111385345459, "data": 0.21105735003948212, "gap": 0.491853803396225, "sigma": 0.9516200423240662, "kind": "mosaic", "tokens": 214, "dense": 448, "grad_norm": 0.8584670424461365, "seconds": 9872.5} {"step": 1974, "loss": 0.6159664392471313, "data": 0.21399053931236267, "gap": 0.4019758999347687, "sigma": 0.5147782564163208, "kind": "mosaic", "tokens": 170, "dense": 448, "grad_norm": 0.20585007965564728, "seconds": 9877.0} {"step": 1975, "loss": 0.6281715631484985, "data": 0.2622607946395874, "gap": 0.36591076850891113, "sigma": 0.5501153469085693, "kind": "mosaic", "tokens": 227, "dense": 448, "grad_norm": 0.13882087171077728, "seconds": 9881.4} {"step": 1976, "loss": 0.6282368302345276, "data": 0.22681936621665955, "gap": 0.40141746401786804, "sigma": 0.4685368239879608, "kind": "mosaic", "tokens": 215, "dense": 448, "grad_norm": 0.22364985942840576, "seconds": 9885.7} {"step": 1977, "loss": 0.28426873683929443, "data": 0.28426873683929443, "gap": null, "sigma": 0.6949211955070496, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.12569156289100647, "seconds": 9888.4} {"step": 1978, "loss": 0.5220824480056763, "data": 0.3056332767009735, "gap": 0.21644917130470276, "sigma": 0.9347962141036987, "kind": "mosaic", "tokens": 275, "dense": 448, "grad_norm": 0.35725316405296326, "seconds": 9892.9} {"step": 1979, "loss": 0.5309968590736389, "data": 0.28054389357566833, "gap": 0.2504529654979706, "sigma": 0.5892459750175476, "kind": "mosaic", "tokens": 288, "dense": 448, "grad_norm": 0.14103098213672638, "seconds": 9897.3} {"step": 1980, "loss": 0.48237401247024536, "data": 0.48237401247024536, "gap": null, "sigma": 0.9759893417358398, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3702309727668762, "seconds": 9900.2} {"step": 1981, "loss": 0.985645055770874, "data": 0.289465069770813, "gap": 0.696179986000061, "sigma": 0.3842757046222687, "kind": "mosaic", "tokens": 228, "dense": 448, "grad_norm": 0.3689408302307129, "seconds": 9904.7} {"step": 1982, "loss": 1.0194259881973267, "data": 0.35975882411003113, "gap": 0.6596671342849731, "sigma": 0.3370242416858673, "kind": "mosaic", "tokens": 203, "dense": 448, "grad_norm": 0.2308553159236908, "seconds": 9909.1} {"step": 1983, "loss": 0.9418983459472656, "data": 0.1863347440958023, "gap": 0.7555636167526245, "sigma": 0.7812823057174683, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.48069673776626587, "seconds": 9913.4} {"step": 1984, "loss": 2.345465898513794, "data": 0.6063842177391052, "gap": 1.7390817403793335, "sigma": 0.1837388426065445, "kind": "mosaic", "tokens": 201, "dense": 448, "grad_norm": 0.8750503659248352, "seconds": 9918.0} {"step": 1985, "loss": 0.5392331480979919, "data": 0.19142545759677887, "gap": 0.34780770540237427, "sigma": 0.566608726978302, "kind": "mosaic", "tokens": 225, "dense": 448, "grad_norm": 0.20301508903503418, "seconds": 9922.5} {"step": 1986, "loss": 0.44954627752304077, "data": 0.44954627752304077, "gap": null, "sigma": 0.9629234671592712, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.18306387960910797, "seconds": 9925.5} {"step": 1987, "loss": 0.5236812829971313, "data": 0.26944777369499207, "gap": 0.2542334794998169, "sigma": 0.49704834818840027, "kind": "mosaic", "tokens": 301, "dense": 448, "grad_norm": 0.13391146063804626, "seconds": 9930.1} {"step": 1988, "loss": 0.6512086391448975, "data": 0.255746990442276, "gap": 0.39546164870262146, "sigma": 0.42362430691719055, "kind": "mosaic", "tokens": 272, "dense": 448, "grad_norm": 0.1438077837228775, "seconds": 9934.7} {"step": 1989, "loss": 0.7242143154144287, "data": 0.23985843360424042, "gap": 0.4843558669090271, "sigma": 0.8031491637229919, "kind": "mosaic", "tokens": 230, "dense": 448, "grad_norm": 0.23447811603546143, "seconds": 9939.2} {"step": 1990, "loss": 0.4828375279903412, "data": 0.18597492575645447, "gap": 0.2968626022338867, "sigma": 0.4538833796977997, "kind": "mosaic", "tokens": 213, "dense": 448, "grad_norm": 0.14313451945781708, "seconds": 9943.6} {"step": 1991, "loss": 1.0436418056488037, "data": 0.3297715485095978, "gap": 0.7138702273368835, "sigma": 0.43190428614616394, "kind": "uniform", "tokens": 112, "dense": 448, "grad_norm": 0.39192959666252136, "seconds": 9947.7} {"step": 1992, "loss": 1.4175562858581543, "data": 0.4095439910888672, "gap": 1.008012294769287, "sigma": 0.9851781129837036, "kind": "mosaic", "tokens": 219, "dense": 448, "grad_norm": 1.2789032459259033, "seconds": 9952.2} {"step": 1993, "loss": 0.8791083097457886, "data": 0.3528580963611603, "gap": 0.5262502431869507, "sigma": 0.35901808738708496, "kind": "mosaic", "tokens": 221, "dense": 448, "grad_norm": 0.18786229193210602, "seconds": 9956.7} {"step": 1994, "loss": 0.7295423150062561, "data": 0.27649056911468506, "gap": 0.45305174589157104, "sigma": 0.4429360330104828, "kind": "mosaic", "tokens": 212, "dense": 448, "grad_norm": 0.2693067193031311, "seconds": 9961.2} {"step": 1995, "loss": 0.382304847240448, "data": 0.21287396550178528, "gap": 0.1694308966398239, "sigma": 0.6063232421875, "kind": "mosaic", "tokens": 244, "dense": 448, "grad_norm": 0.5237278342247009, "seconds": 9965.7} {"step": 1996, "loss": 0.40654829144477844, "data": 0.40654829144477844, "gap": null, "sigma": 0.9624500274658203, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.20620521903038025, "seconds": 9968.5} {"step": 1997, "loss": 0.15612301230430603, "data": 0.15612301230430603, "gap": null, "sigma": 0.485383540391922, "kind": "dense", "tokens": 448, "dense": 448, "grad_norm": 0.3323831260204315, "seconds": 9971.3} {"step": 1998, "loss": 0.8651532530784607, "data": 0.2626063823699951, "gap": 0.6025468707084656, "sigma": 0.4470447301864624, "kind": "mosaic", "tokens": 163, "dense": 448, "grad_norm": 0.25905996561050415, "seconds": 9975.7} {"step": 1999, "loss": 0.8115382194519043, "data": 0.2675084173679352, "gap": 0.5440298318862915, "sigma": 0.9251821637153625, "kind": "mosaic", "tokens": 197, "dense": 448, "grad_norm": 0.3610991835594177, "seconds": 9980.3} {"step": 2000, "loss": 0.791424036026001, "data": 0.2437272071838379, "gap": 0.5476968288421631, "sigma": 0.860593855381012, "kind": "mosaic", "tokens": 212, "dense": 448, "grad_norm": 0.48871931433677673, "seconds": 9984.7} {"step": 2000, "eval": {"dense": 0.3055709395557642, "uniform2": 0.18303990767647824, "mosaic": 0.2523188032209873, "gap_uniform2": 0.5688122659921646, "gap_mosaic": 0.5670113184799751}, "eval_extent": {"1x1x1": 0.42630242245308997, "1x1x2": 0.31803739938020176, "1x1x4": 0.22709114320184054, "1x2x1": 0.3001476560469234, "1x2x2": 0.19257446268426354, "1x2x4": 0.12678435496760138, "1x4x1": 0.19429905736056535, "1x4x2": 0.11132421701353182, "1x4x4": 0.08084111534135074}}