| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0003621220351258, | |
| "eval_steps": 346, | |
| "global_step": 1381, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0007242440702516748, | |
| "grad_norm": 0.22644081711769104, | |
| "learning_rate": 2e-05, | |
| "loss": 10.3782, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0014484881405033496, | |
| "grad_norm": 0.21413658559322357, | |
| "learning_rate": 4e-05, | |
| "loss": 10.378, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0021727322107550242, | |
| "grad_norm": 0.20930856466293335, | |
| "learning_rate": 6e-05, | |
| "loss": 10.3801, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.002896976281006699, | |
| "grad_norm": 0.21580834686756134, | |
| "learning_rate": 8e-05, | |
| "loss": 10.3772, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.003621220351258374, | |
| "grad_norm": 0.22320739924907684, | |
| "learning_rate": 0.0001, | |
| "loss": 10.3771, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0043454644215100485, | |
| "grad_norm": 0.20463065803050995, | |
| "learning_rate": 0.00012, | |
| "loss": 10.3765, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.005069708491761723, | |
| "grad_norm": 0.2269536256790161, | |
| "learning_rate": 0.00014, | |
| "loss": 10.374, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.005793952562013398, | |
| "grad_norm": 0.23299536108970642, | |
| "learning_rate": 0.00016, | |
| "loss": 10.3733, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.006518196632265073, | |
| "grad_norm": 0.2280343770980835, | |
| "learning_rate": 0.00018, | |
| "loss": 10.3716, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.007242440702516748, | |
| "grad_norm": 0.23425506055355072, | |
| "learning_rate": 0.0002, | |
| "loss": 10.371, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.007966684772768424, | |
| "grad_norm": 0.26600638031959534, | |
| "learning_rate": 0.00019999973746050225, | |
| "loss": 10.3668, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.008690928843020097, | |
| "grad_norm": 0.2833026051521301, | |
| "learning_rate": 0.00019999894984338746, | |
| "loss": 10.3668, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.009415172913271772, | |
| "grad_norm": 0.31404921412467957, | |
| "learning_rate": 0.00019999763715279132, | |
| "loss": 10.3583, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.010139416983523447, | |
| "grad_norm": 0.3399479389190674, | |
| "learning_rate": 0.00019999579939560644, | |
| "loss": 10.3628, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.010863661053775122, | |
| "grad_norm": 0.3927531838417053, | |
| "learning_rate": 0.00019999343658148253, | |
| "loss": 10.3534, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.011587905124026797, | |
| "grad_norm": 0.43685317039489746, | |
| "learning_rate": 0.00019999054872282622, | |
| "loss": 10.3498, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.012312149194278471, | |
| "grad_norm": 0.5002313852310181, | |
| "learning_rate": 0.00019998713583480103, | |
| "loss": 10.3404, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.013036393264530146, | |
| "grad_norm": 0.5821593999862671, | |
| "learning_rate": 0.00019998319793532735, | |
| "loss": 10.3355, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.013760637334781821, | |
| "grad_norm": 0.6062823534011841, | |
| "learning_rate": 0.00019997873504508222, | |
| "loss": 10.329, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.014484881405033496, | |
| "grad_norm": 0.6338623762130737, | |
| "learning_rate": 0.0001999737471874994, | |
| "loss": 10.3277, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.015209125475285171, | |
| "grad_norm": 0.7111278176307678, | |
| "learning_rate": 0.00019996823438876902, | |
| "loss": 10.3189, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.015933369545536848, | |
| "grad_norm": 0.7820473313331604, | |
| "learning_rate": 0.00019996219667783765, | |
| "loss": 10.3128, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.01665761361578852, | |
| "grad_norm": 0.8882144093513489, | |
| "learning_rate": 0.00019995563408640806, | |
| "loss": 10.3005, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.017381857686040194, | |
| "grad_norm": 0.9162967205047607, | |
| "learning_rate": 0.00019994854664893906, | |
| "loss": 10.2918, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.01810610175629187, | |
| "grad_norm": 0.9656468033790588, | |
| "learning_rate": 0.00019994093440264522, | |
| "loss": 10.277, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.018830345826543544, | |
| "grad_norm": 0.9979098439216614, | |
| "learning_rate": 0.00019993279738749687, | |
| "loss": 10.2707, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.01955458989679522, | |
| "grad_norm": 0.9871697425842285, | |
| "learning_rate": 0.00019992413564621985, | |
| "loss": 10.2556, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.020278833967046894, | |
| "grad_norm": 1.0156805515289307, | |
| "learning_rate": 0.00019991494922429504, | |
| "loss": 10.2469, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.02100307803729857, | |
| "grad_norm": 0.9702807068824768, | |
| "learning_rate": 0.00019990523816995848, | |
| "loss": 10.2315, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.021727322107550243, | |
| "grad_norm": 0.9247342348098755, | |
| "learning_rate": 0.0001998950025342008, | |
| "loss": 10.2134, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.022451566177801918, | |
| "grad_norm": 0.8712878227233887, | |
| "learning_rate": 0.00019988424237076728, | |
| "loss": 10.1968, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.023175810248053593, | |
| "grad_norm": 0.8503078818321228, | |
| "learning_rate": 0.0001998729577361572, | |
| "loss": 10.1812, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.023900054318305268, | |
| "grad_norm": 0.8318528532981873, | |
| "learning_rate": 0.0001998611486896238, | |
| "loss": 10.1752, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.024624298388556943, | |
| "grad_norm": 0.7345334887504578, | |
| "learning_rate": 0.00019984881529317393, | |
| "loss": 10.1601, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.025348542458808618, | |
| "grad_norm": 0.7379999160766602, | |
| "learning_rate": 0.0001998359576115677, | |
| "loss": 10.1499, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.026072786529060293, | |
| "grad_norm": 0.674487829208374, | |
| "learning_rate": 0.00019982257571231804, | |
| "loss": 10.1309, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.026797030599311968, | |
| "grad_norm": 0.6472688317298889, | |
| "learning_rate": 0.00019980866966569054, | |
| "loss": 10.1182, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.027521274669563642, | |
| "grad_norm": 0.6024581789970398, | |
| "learning_rate": 0.00019979423954470286, | |
| "loss": 10.1074, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.028245518739815317, | |
| "grad_norm": 0.610586404800415, | |
| "learning_rate": 0.0001997792854251246, | |
| "loss": 10.0958, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.028969762810066992, | |
| "grad_norm": 0.5543527007102966, | |
| "learning_rate": 0.00019976380738547666, | |
| "loss": 10.0964, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.029694006880318667, | |
| "grad_norm": 0.5993051528930664, | |
| "learning_rate": 0.000199747805507031, | |
| "loss": 10.0835, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.030418250950570342, | |
| "grad_norm": 0.5518122911453247, | |
| "learning_rate": 0.0001997312798738101, | |
| "loss": 10.0662, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.031142495020822017, | |
| "grad_norm": 0.5521525144577026, | |
| "learning_rate": 0.00019971423057258664, | |
| "loss": 10.0451, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.031866739091073695, | |
| "grad_norm": 0.5692468285560608, | |
| "learning_rate": 0.00019969665769288284, | |
| "loss": 10.0466, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.03259098316132537, | |
| "grad_norm": 0.5686005353927612, | |
| "learning_rate": 0.00019967856132697027, | |
| "loss": 10.0256, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.03331522723157704, | |
| "grad_norm": 0.5743457078933716, | |
| "learning_rate": 0.00019965994156986912, | |
| "loss": 10.0049, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.034039471301828716, | |
| "grad_norm": 0.5410902500152588, | |
| "learning_rate": 0.0001996407985193478, | |
| "loss": 10.0077, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.03476371537208039, | |
| "grad_norm": 0.5433526635169983, | |
| "learning_rate": 0.0001996211322759225, | |
| "loss": 9.9934, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.035487959442332066, | |
| "grad_norm": 0.5521161556243896, | |
| "learning_rate": 0.00019960094294285647, | |
| "loss": 9.9753, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.03621220351258374, | |
| "grad_norm": 0.6015803217887878, | |
| "learning_rate": 0.00019958023062615973, | |
| "loss": 9.976, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.036936447582835416, | |
| "grad_norm": 0.5433859825134277, | |
| "learning_rate": 0.00019955899543458824, | |
| "loss": 9.9617, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.03766069165308709, | |
| "grad_norm": 0.5231461524963379, | |
| "learning_rate": 0.00019953723747964355, | |
| "loss": 9.9525, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.038384935723338766, | |
| "grad_norm": 0.5396632552146912, | |
| "learning_rate": 0.00019951495687557213, | |
| "loss": 9.9383, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.03910917979359044, | |
| "grad_norm": 0.5196948051452637, | |
| "learning_rate": 0.00019949215373936475, | |
| "loss": 9.928, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.039833423863842116, | |
| "grad_norm": 0.4962172210216522, | |
| "learning_rate": 0.00019946882819075587, | |
| "loss": 9.9235, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.04055766793409379, | |
| "grad_norm": 0.5424382090568542, | |
| "learning_rate": 0.00019944498035222305, | |
| "loss": 9.9033, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.041281912004345465, | |
| "grad_norm": 0.5209097266197205, | |
| "learning_rate": 0.00019942061034898626, | |
| "loss": 9.902, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.04200615607459714, | |
| "grad_norm": 0.4981401562690735, | |
| "learning_rate": 0.00019939571830900735, | |
| "loss": 9.8884, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.042730400144848815, | |
| "grad_norm": 0.4946765899658203, | |
| "learning_rate": 0.0001993703043629891, | |
| "loss": 9.8841, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.04345464421510049, | |
| "grad_norm": 0.526029646396637, | |
| "learning_rate": 0.00019934436864437485, | |
| "loss": 9.8607, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.044178888285352165, | |
| "grad_norm": 0.528313159942627, | |
| "learning_rate": 0.0001993179112893476, | |
| "loss": 9.8549, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.044903132355603836, | |
| "grad_norm": 0.5022369623184204, | |
| "learning_rate": 0.00019929093243682938, | |
| "loss": 9.83, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.045627376425855515, | |
| "grad_norm": 0.5330261588096619, | |
| "learning_rate": 0.00019926343222848042, | |
| "loss": 9.8244, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.046351620496107186, | |
| "grad_norm": 0.5101606249809265, | |
| "learning_rate": 0.0001992354108086986, | |
| "loss": 9.8145, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.047075864566358865, | |
| "grad_norm": 0.5162172317504883, | |
| "learning_rate": 0.0001992068683246185, | |
| "loss": 9.8123, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.047800108636610536, | |
| "grad_norm": 0.5125104784965515, | |
| "learning_rate": 0.0001991778049261107, | |
| "loss": 9.7947, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.048524352706862214, | |
| "grad_norm": 0.5619991421699524, | |
| "learning_rate": 0.00019914822076578097, | |
| "loss": 9.773, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.049248596777113886, | |
| "grad_norm": 0.5185785889625549, | |
| "learning_rate": 0.0001991181159989696, | |
| "loss": 9.7618, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.049972840847365564, | |
| "grad_norm": 0.4937543272972107, | |
| "learning_rate": 0.0001990874907837503, | |
| "loss": 9.7543, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.050697084917617236, | |
| "grad_norm": 0.5016326904296875, | |
| "learning_rate": 0.00019905634528092972, | |
| "loss": 9.7509, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.051421328987868914, | |
| "grad_norm": 0.5426475405693054, | |
| "learning_rate": 0.0001990246796540463, | |
| "loss": 9.7287, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.052145573058120585, | |
| "grad_norm": 0.5001341700553894, | |
| "learning_rate": 0.00019899249406936964, | |
| "loss": 9.7159, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.052869817128372264, | |
| "grad_norm": 0.5133237242698669, | |
| "learning_rate": 0.00019895978869589946, | |
| "loss": 9.7129, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.053594061198623935, | |
| "grad_norm": 0.5022423267364502, | |
| "learning_rate": 0.00019892656370536482, | |
| "loss": 9.6986, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.054318305268875614, | |
| "grad_norm": 0.5012578368186951, | |
| "learning_rate": 0.0001988928192722231, | |
| "loss": 9.6968, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.055042549339127285, | |
| "grad_norm": 0.5184592008590698, | |
| "learning_rate": 0.00019885855557365937, | |
| "loss": 9.6754, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.05576679340937896, | |
| "grad_norm": 0.49060916900634766, | |
| "learning_rate": 0.000198823772789585, | |
| "loss": 9.6662, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.056491037479630635, | |
| "grad_norm": 0.5203105211257935, | |
| "learning_rate": 0.0001987884711026371, | |
| "loss": 9.6577, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.05721528154988231, | |
| "grad_norm": 0.5055034160614014, | |
| "learning_rate": 0.00019875265069817743, | |
| "loss": 9.6392, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.057939525620133984, | |
| "grad_norm": 0.5237613320350647, | |
| "learning_rate": 0.00019871631176429145, | |
| "loss": 9.6415, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.05866376969038566, | |
| "grad_norm": 0.5136992931365967, | |
| "learning_rate": 0.0001986794544917872, | |
| "loss": 9.6265, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.059388013760637334, | |
| "grad_norm": 0.5193312168121338, | |
| "learning_rate": 0.00019864207907419447, | |
| "loss": 9.6138, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.06011225783088901, | |
| "grad_norm": 0.5376698970794678, | |
| "learning_rate": 0.0001986041857077638, | |
| "loss": 9.6057, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.060836501901140684, | |
| "grad_norm": 0.52781081199646, | |
| "learning_rate": 0.00019856577459146526, | |
| "loss": 9.5839, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.06156074597139236, | |
| "grad_norm": 0.5407407879829407, | |
| "learning_rate": 0.00019852684592698756, | |
| "loss": 9.5891, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.062284990041644034, | |
| "grad_norm": 0.5036855936050415, | |
| "learning_rate": 0.0001984873999187369, | |
| "loss": 9.5722, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.06300923411189571, | |
| "grad_norm": 0.5385954976081848, | |
| "learning_rate": 0.00019844743677383604, | |
| "loss": 9.5572, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.06373347818214739, | |
| "grad_norm": 0.5041705965995789, | |
| "learning_rate": 0.00019840695670212302, | |
| "loss": 9.5528, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.06445772225239906, | |
| "grad_norm": 0.5461394190788269, | |
| "learning_rate": 0.00019836595991615022, | |
| "loss": 9.5181, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.06518196632265073, | |
| "grad_norm": 0.5438934564590454, | |
| "learning_rate": 0.00019832444663118315, | |
| "loss": 9.5303, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.06590621039290241, | |
| "grad_norm": 0.5632613301277161, | |
| "learning_rate": 0.00019828241706519934, | |
| "loss": 9.5119, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.06663045446315408, | |
| "grad_norm": 0.4781850278377533, | |
| "learning_rate": 0.0001982398714388872, | |
| "loss": 9.5208, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.06735469853340575, | |
| "grad_norm": 0.5311410427093506, | |
| "learning_rate": 0.00019819680997564492, | |
| "loss": 9.5111, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.06807894260365743, | |
| "grad_norm": 0.5310368537902832, | |
| "learning_rate": 0.00019815323290157916, | |
| "loss": 9.4963, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.06880318667390911, | |
| "grad_norm": 0.5059214234352112, | |
| "learning_rate": 0.000198109140445504, | |
| "loss": 9.4878, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.06952743074416078, | |
| "grad_norm": 0.517311155796051, | |
| "learning_rate": 0.00019806453283893963, | |
| "loss": 9.478, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.07025167481441245, | |
| "grad_norm": 0.5341803431510925, | |
| "learning_rate": 0.00019801941031611126, | |
| "loss": 9.4307, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.07097591888466413, | |
| "grad_norm": 0.5468002557754517, | |
| "learning_rate": 0.0001979737731139478, | |
| "loss": 9.462, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.07170016295491581, | |
| "grad_norm": 0.5668586492538452, | |
| "learning_rate": 0.00019792762147208056, | |
| "loss": 9.4286, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.07242440702516748, | |
| "grad_norm": 0.6626901626586914, | |
| "learning_rate": 0.00019788095563284217, | |
| "loss": 9.4239, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.07314865109541915, | |
| "grad_norm": 0.5568472743034363, | |
| "learning_rate": 0.00019783377584126508, | |
| "loss": 9.4093, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.07387289516567083, | |
| "grad_norm": 0.5283701419830322, | |
| "learning_rate": 0.00019778608234508055, | |
| "loss": 9.4273, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.07459713923592251, | |
| "grad_norm": 0.5079399943351746, | |
| "learning_rate": 0.00019773787539471705, | |
| "loss": 9.3889, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.07532138330617417, | |
| "grad_norm": 0.5189304351806641, | |
| "learning_rate": 0.00019768915524329917, | |
| "loss": 9.3884, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.07604562737642585, | |
| "grad_norm": 0.504858136177063, | |
| "learning_rate": 0.00019763992214664615, | |
| "loss": 9.4027, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.07676987144667753, | |
| "grad_norm": 0.5144493579864502, | |
| "learning_rate": 0.00019759017636327073, | |
| "loss": 9.3547, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.07749411551692921, | |
| "grad_norm": 0.49943238496780396, | |
| "learning_rate": 0.0001975399181543775, | |
| "loss": 9.3584, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.07821835958718087, | |
| "grad_norm": 0.5128799080848694, | |
| "learning_rate": 0.0001974891477838618, | |
| "loss": 9.3521, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.07894260365743255, | |
| "grad_norm": 0.5011847019195557, | |
| "learning_rate": 0.00019743786551830813, | |
| "loss": 9.3468, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.07966684772768423, | |
| "grad_norm": 0.5017261505126953, | |
| "learning_rate": 0.00019738607162698895, | |
| "loss": 9.3335, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.08039109179793591, | |
| "grad_norm": 0.5072489976882935, | |
| "learning_rate": 0.00019733376638186308, | |
| "loss": 9.3045, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.08111533586818757, | |
| "grad_norm": 0.5118771195411682, | |
| "learning_rate": 0.00019728095005757434, | |
| "loss": 9.3186, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.08183957993843925, | |
| "grad_norm": 0.5291309356689453, | |
| "learning_rate": 0.0001972276229314502, | |
| "loss": 9.3077, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.08256382400869093, | |
| "grad_norm": 0.49927669763565063, | |
| "learning_rate": 0.00019717378528350023, | |
| "loss": 9.2945, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.08328806807894261, | |
| "grad_norm": 0.5379682183265686, | |
| "learning_rate": 0.00019711943739641452, | |
| "loss": 9.2844, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.08401231214919427, | |
| "grad_norm": 0.5146782398223877, | |
| "learning_rate": 0.00019706457955556247, | |
| "loss": 9.2589, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.08473655621944595, | |
| "grad_norm": 0.5160855054855347, | |
| "learning_rate": 0.0001970092120489911, | |
| "loss": 9.2522, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.08546080028969763, | |
| "grad_norm": 0.5257864594459534, | |
| "learning_rate": 0.0001969533351674235, | |
| "loss": 9.2622, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.08618504435994931, | |
| "grad_norm": 0.4924103617668152, | |
| "learning_rate": 0.00019689694920425746, | |
| "loss": 9.2382, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.08690928843020097, | |
| "grad_norm": 0.5249960422515869, | |
| "learning_rate": 0.00019684005445556383, | |
| "loss": 9.2173, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.08763353250045265, | |
| "grad_norm": 0.5177789330482483, | |
| "learning_rate": 0.000196782651220085, | |
| "loss": 9.2171, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.08835777657070433, | |
| "grad_norm": 0.5228692889213562, | |
| "learning_rate": 0.0001967247397992333, | |
| "loss": 9.2168, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.08908202064095601, | |
| "grad_norm": 0.5426861047744751, | |
| "learning_rate": 0.00019666632049708942, | |
| "loss": 9.1786, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.08980626471120767, | |
| "grad_norm": 0.5005182027816772, | |
| "learning_rate": 0.00019660739362040082, | |
| "loss": 9.1994, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.09053050878145935, | |
| "grad_norm": 0.516734778881073, | |
| "learning_rate": 0.00019654795947858023, | |
| "loss": 9.172, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.09125475285171103, | |
| "grad_norm": 0.5049043297767639, | |
| "learning_rate": 0.00019648801838370377, | |
| "loss": 9.1714, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.09197899692196271, | |
| "grad_norm": 0.5162491202354431, | |
| "learning_rate": 0.00019642757065050956, | |
| "loss": 9.149, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.09270324099221437, | |
| "grad_norm": 0.49220365285873413, | |
| "learning_rate": 0.000196366616596396, | |
| "loss": 9.164, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.09342748506246605, | |
| "grad_norm": 0.508711576461792, | |
| "learning_rate": 0.00019630515654141996, | |
| "loss": 9.1311, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.09415172913271773, | |
| "grad_norm": 0.5290361642837524, | |
| "learning_rate": 0.0001962431908082953, | |
| "loss": 9.1079, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.09487597320296941, | |
| "grad_norm": 0.5396744608879089, | |
| "learning_rate": 0.00019618071972239107, | |
| "loss": 9.115, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.09560021727322107, | |
| "grad_norm": 0.49481114745140076, | |
| "learning_rate": 0.0001961177436117298, | |
| "loss": 9.1326, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.09632446134347275, | |
| "grad_norm": 0.4906870126724243, | |
| "learning_rate": 0.0001960542628069859, | |
| "loss": 9.1192, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.09704870541372443, | |
| "grad_norm": 0.5055575370788574, | |
| "learning_rate": 0.00019599027764148367, | |
| "loss": 9.1099, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.0977729494839761, | |
| "grad_norm": 0.5195087790489197, | |
| "learning_rate": 0.00019592578845119575, | |
| "loss": 9.0824, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.09849719355422777, | |
| "grad_norm": 0.48597005009651184, | |
| "learning_rate": 0.0001958607955747414, | |
| "loss": 9.0973, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.09922143762447945, | |
| "grad_norm": 0.5294395089149475, | |
| "learning_rate": 0.0001957952993533845, | |
| "loss": 9.0725, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.09994568169473113, | |
| "grad_norm": 0.5333034992218018, | |
| "learning_rate": 0.00019572930013103202, | |
| "loss": 9.0177, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.10066992576498279, | |
| "grad_norm": 0.5079728960990906, | |
| "learning_rate": 0.00019566279825423196, | |
| "loss": 9.0231, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.10139416983523447, | |
| "grad_norm": 0.5101923942565918, | |
| "learning_rate": 0.00019559579407217172, | |
| "loss": 9.0478, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.10211841390548615, | |
| "grad_norm": 0.5175127387046814, | |
| "learning_rate": 0.0001955282879366762, | |
| "loss": 9.0333, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.10284265797573783, | |
| "grad_norm": 0.5011202692985535, | |
| "learning_rate": 0.00019546028020220595, | |
| "loss": 9.0051, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.10356690204598949, | |
| "grad_norm": 0.5366819500923157, | |
| "learning_rate": 0.00019539177122585523, | |
| "loss": 9.014, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.10429114611624117, | |
| "grad_norm": 0.5233102440834045, | |
| "learning_rate": 0.00019532276136735038, | |
| "loss": 8.9821, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.10501539018649285, | |
| "grad_norm": 0.5242369771003723, | |
| "learning_rate": 0.00019525325098904757, | |
| "loss": 9.0231, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.10573963425674453, | |
| "grad_norm": 0.5241793394088745, | |
| "learning_rate": 0.00019518324045593132, | |
| "loss": 8.9769, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.10646387832699619, | |
| "grad_norm": 0.5393111109733582, | |
| "learning_rate": 0.0001951127301356121, | |
| "loss": 9.0189, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.10718812239724787, | |
| "grad_norm": 0.5141487717628479, | |
| "learning_rate": 0.00019504172039832492, | |
| "loss": 8.9846, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.10791236646749955, | |
| "grad_norm": 0.5347380042076111, | |
| "learning_rate": 0.00019497021161692687, | |
| "loss": 8.9623, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.10863661053775123, | |
| "grad_norm": 0.5859756469726562, | |
| "learning_rate": 0.00019489820416689565, | |
| "loss": 8.9612, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.10936085460800289, | |
| "grad_norm": 0.5630178451538086, | |
| "learning_rate": 0.0001948256984263272, | |
| "loss": 8.9316, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.11008509867825457, | |
| "grad_norm": 0.5126703381538391, | |
| "learning_rate": 0.00019475269477593394, | |
| "loss": 8.9421, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.11080934274850625, | |
| "grad_norm": 0.5410084128379822, | |
| "learning_rate": 0.0001946791935990427, | |
| "loss": 8.9093, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.11153358681875793, | |
| "grad_norm": 0.500529408454895, | |
| "learning_rate": 0.00019460519528159275, | |
| "loss": 8.9258, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.11225783088900959, | |
| "grad_norm": 0.505597710609436, | |
| "learning_rate": 0.00019453070021213366, | |
| "loss": 8.9067, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.11298207495926127, | |
| "grad_norm": 0.5078559517860413, | |
| "learning_rate": 0.00019445570878182342, | |
| "loss": 8.8985, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.11370631902951295, | |
| "grad_norm": 0.48962968587875366, | |
| "learning_rate": 0.0001943802213844263, | |
| "loss": 8.8889, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.11443056309976463, | |
| "grad_norm": 0.521984338760376, | |
| "learning_rate": 0.00019430423841631074, | |
| "loss": 8.8726, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.11515480717001629, | |
| "grad_norm": 0.510870635509491, | |
| "learning_rate": 0.00019422776027644737, | |
| "loss": 8.8657, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.11587905124026797, | |
| "grad_norm": 0.48674073815345764, | |
| "learning_rate": 0.0001941507873664068, | |
| "loss": 8.8745, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.11660329531051965, | |
| "grad_norm": 0.5165760517120361, | |
| "learning_rate": 0.00019407332009035769, | |
| "loss": 8.8614, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.11732753938077133, | |
| "grad_norm": 0.5049470067024231, | |
| "learning_rate": 0.00019399535885506432, | |
| "loss": 8.8781, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.11805178345102299, | |
| "grad_norm": 0.49627047777175903, | |
| "learning_rate": 0.00019391690406988485, | |
| "loss": 8.832, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.11877602752127467, | |
| "grad_norm": 0.5152491331100464, | |
| "learning_rate": 0.00019383795614676886, | |
| "loss": 8.852, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.11950027159152635, | |
| "grad_norm": 0.49169403314590454, | |
| "learning_rate": 0.00019375851550025529, | |
| "loss": 8.8621, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.12022451566177803, | |
| "grad_norm": 0.5024704337120056, | |
| "learning_rate": 0.00019367858254747028, | |
| "loss": 8.8057, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.12094875973202969, | |
| "grad_norm": 0.5297830104827881, | |
| "learning_rate": 0.00019359815770812503, | |
| "loss": 8.7918, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.12167300380228137, | |
| "grad_norm": 0.5009456276893616, | |
| "learning_rate": 0.00019351724140451344, | |
| "loss": 8.795, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.12239724787253305, | |
| "grad_norm": 0.49791526794433594, | |
| "learning_rate": 0.00019343583406151004, | |
| "loss": 8.7908, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.12312149194278472, | |
| "grad_norm": 0.4869917035102844, | |
| "learning_rate": 0.00019335393610656767, | |
| "loss": 8.8117, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.12384573601303639, | |
| "grad_norm": 0.4747772216796875, | |
| "learning_rate": 0.00019327154796971527, | |
| "loss": 8.7738, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.12456998008328807, | |
| "grad_norm": 0.49815577268600464, | |
| "learning_rate": 0.0001931886700835557, | |
| "loss": 8.7769, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.12529422415353975, | |
| "grad_norm": 0.4809548556804657, | |
| "learning_rate": 0.00019310530288326329, | |
| "loss": 8.7769, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.12601846822379142, | |
| "grad_norm": 0.519578218460083, | |
| "learning_rate": 0.00019302144680658173, | |
| "loss": 8.7652, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.1267427122940431, | |
| "grad_norm": 0.49192696809768677, | |
| "learning_rate": 0.0001929371022938216, | |
| "loss": 8.7399, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.12746695636429478, | |
| "grad_norm": 0.5109142661094666, | |
| "learning_rate": 0.00019285226978785832, | |
| "loss": 8.7374, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.12819120043454643, | |
| "grad_norm": 0.48110830783843994, | |
| "learning_rate": 0.00019276694973412948, | |
| "loss": 8.7524, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.1289154445047981, | |
| "grad_norm": 0.511137843132019, | |
| "learning_rate": 0.0001926811425806328, | |
| "loss": 8.7089, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.1296396885750498, | |
| "grad_norm": 0.49463972449302673, | |
| "learning_rate": 0.00019259484877792358, | |
| "loss": 8.7089, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.13036393264530147, | |
| "grad_norm": 0.4988570213317871, | |
| "learning_rate": 0.00019250806877911249, | |
| "loss": 8.7283, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.13108817671555315, | |
| "grad_norm": 0.48762744665145874, | |
| "learning_rate": 0.00019242080303986305, | |
| "loss": 8.6905, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.13181242078580482, | |
| "grad_norm": 0.5100088119506836, | |
| "learning_rate": 0.00019233305201838937, | |
| "loss": 8.6756, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.1325366648560565, | |
| "grad_norm": 0.502967119216919, | |
| "learning_rate": 0.00019224481617545358, | |
| "loss": 8.6831, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.13326090892630815, | |
| "grad_norm": 0.4762354791164398, | |
| "learning_rate": 0.00019215609597436362, | |
| "loss": 8.6856, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.13398515299655983, | |
| "grad_norm": 0.5262447595596313, | |
| "learning_rate": 0.00019206689188097054, | |
| "loss": 8.6393, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.1347093970668115, | |
| "grad_norm": 0.5070264339447021, | |
| "learning_rate": 0.00019197720436366637, | |
| "loss": 8.628, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.1354336411370632, | |
| "grad_norm": 0.4977290630340576, | |
| "learning_rate": 0.00019188703389338142, | |
| "loss": 8.6412, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.13615788520731487, | |
| "grad_norm": 0.5031901001930237, | |
| "learning_rate": 0.00019179638094358187, | |
| "loss": 8.6122, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.13688212927756654, | |
| "grad_norm": 0.5076611638069153, | |
| "learning_rate": 0.00019170524599026732, | |
| "loss": 8.6475, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.13760637334781822, | |
| "grad_norm": 0.5410529971122742, | |
| "learning_rate": 0.00019161362951196825, | |
| "loss": 8.6091, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.1383306174180699, | |
| "grad_norm": 0.5248331427574158, | |
| "learning_rate": 0.0001915215319897436, | |
| "loss": 8.6022, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.13905486148832155, | |
| "grad_norm": 0.5243874788284302, | |
| "learning_rate": 0.00019142895390717804, | |
| "loss": 8.6125, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.13977910555857323, | |
| "grad_norm": 0.597634494304657, | |
| "learning_rate": 0.0001913358957503797, | |
| "loss": 8.6093, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.1405033496288249, | |
| "grad_norm": 0.528044581413269, | |
| "learning_rate": 0.0001912423580079774, | |
| "loss": 8.6003, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.1412275936990766, | |
| "grad_norm": 0.5066649913787842, | |
| "learning_rate": 0.00019114834117111814, | |
| "loss": 8.5842, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.14195183776932827, | |
| "grad_norm": 0.49779024720191956, | |
| "learning_rate": 0.00019105384573346463, | |
| "loss": 8.6286, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.14267608183957994, | |
| "grad_norm": 0.5297601222991943, | |
| "learning_rate": 0.00019095887219119256, | |
| "loss": 8.5814, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.14340032590983162, | |
| "grad_norm": 0.5315724015235901, | |
| "learning_rate": 0.000190863421042988, | |
| "loss": 8.5826, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.1441245699800833, | |
| "grad_norm": 0.5855052471160889, | |
| "learning_rate": 0.00019076749279004496, | |
| "loss": 8.5369, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.14484881405033495, | |
| "grad_norm": 0.5880944728851318, | |
| "learning_rate": 0.0001906710879360625, | |
| "loss": 8.5681, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.14557305812058663, | |
| "grad_norm": 0.5085429549217224, | |
| "learning_rate": 0.00019057420698724223, | |
| "loss": 8.5739, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.1462973021908383, | |
| "grad_norm": 0.541529655456543, | |
| "learning_rate": 0.00019047685045228569, | |
| "loss": 8.5405, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.14702154626108999, | |
| "grad_norm": 0.5351887941360474, | |
| "learning_rate": 0.0001903790188423916, | |
| "loss": 8.5307, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.14774579033134166, | |
| "grad_norm": 0.4913345277309418, | |
| "learning_rate": 0.00019028071267125323, | |
| "loss": 8.5303, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.14847003440159334, | |
| "grad_norm": 0.476534903049469, | |
| "learning_rate": 0.0001901819324550556, | |
| "loss": 8.546, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.14919427847184502, | |
| "grad_norm": 0.511883020401001, | |
| "learning_rate": 0.00019008267871247286, | |
| "loss": 8.523, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.1499185225420967, | |
| "grad_norm": 0.5081936717033386, | |
| "learning_rate": 0.0001899829519646656, | |
| "loss": 8.5286, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.15064276661234835, | |
| "grad_norm": 0.4755452275276184, | |
| "learning_rate": 0.000189882752735278, | |
| "loss": 8.5464, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.15136701068260003, | |
| "grad_norm": 0.5027110576629639, | |
| "learning_rate": 0.0001897820815504352, | |
| "loss": 8.5272, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.1520912547528517, | |
| "grad_norm": 0.5083355903625488, | |
| "learning_rate": 0.0001896809389387404, | |
| "loss": 8.4798, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.15281549882310339, | |
| "grad_norm": 0.5499453544616699, | |
| "learning_rate": 0.00018957932543127226, | |
| "loss": 8.5331, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.15353974289335506, | |
| "grad_norm": 0.51893150806427, | |
| "learning_rate": 0.00018947724156158192, | |
| "loss": 8.4911, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.15426398696360674, | |
| "grad_norm": 0.5075845718383789, | |
| "learning_rate": 0.00018937468786569034, | |
| "loss": 8.4956, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.15498823103385842, | |
| "grad_norm": 0.5007838606834412, | |
| "learning_rate": 0.00018927166488208548, | |
| "loss": 8.4777, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.1557124751041101, | |
| "grad_norm": 0.5123993158340454, | |
| "learning_rate": 0.00018916817315171934, | |
| "loss": 8.4535, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.15643671917436175, | |
| "grad_norm": 0.5103585124015808, | |
| "learning_rate": 0.00018906421321800528, | |
| "loss": 8.4328, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.15716096324461343, | |
| "grad_norm": 0.49165982007980347, | |
| "learning_rate": 0.00018895978562681506, | |
| "loss": 8.4659, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.1578852073148651, | |
| "grad_norm": 0.5067943930625916, | |
| "learning_rate": 0.00018885489092647606, | |
| "loss": 8.4503, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.15860945138511678, | |
| "grad_norm": 0.5133797526359558, | |
| "learning_rate": 0.0001887495296677683, | |
| "loss": 8.4473, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.15933369545536846, | |
| "grad_norm": 0.4692786633968353, | |
| "learning_rate": 0.0001886437024039216, | |
| "loss": 8.4552, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.16005793952562014, | |
| "grad_norm": 0.4994044005870819, | |
| "learning_rate": 0.00018853740969061272, | |
| "loss": 8.4293, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.16078218359587182, | |
| "grad_norm": 0.5162525773048401, | |
| "learning_rate": 0.00018843065208596236, | |
| "loss": 8.3714, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.16150642766612347, | |
| "grad_norm": 0.48502400517463684, | |
| "learning_rate": 0.00018832343015053228, | |
| "loss": 8.4212, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.16223067173637515, | |
| "grad_norm": 0.48316872119903564, | |
| "learning_rate": 0.00018821574444732235, | |
| "loss": 8.3908, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.16295491580662683, | |
| "grad_norm": 0.5301130414009094, | |
| "learning_rate": 0.0001881075955417676, | |
| "loss": 8.4019, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.1636791598768785, | |
| "grad_norm": 0.5196858048439026, | |
| "learning_rate": 0.0001879989840017351, | |
| "loss": 8.3669, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.16440340394713018, | |
| "grad_norm": 0.4746645390987396, | |
| "learning_rate": 0.0001878899103975214, | |
| "loss": 8.3958, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.16512764801738186, | |
| "grad_norm": 0.4931572675704956, | |
| "learning_rate": 0.000187780375301849, | |
| "loss": 8.3942, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.16585189208763354, | |
| "grad_norm": 0.5066322684288025, | |
| "learning_rate": 0.00018767037928986367, | |
| "loss": 8.3708, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.16657613615788522, | |
| "grad_norm": 0.46926993131637573, | |
| "learning_rate": 0.00018755992293913135, | |
| "loss": 8.3315, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.16730038022813687, | |
| "grad_norm": 0.4935818314552307, | |
| "learning_rate": 0.00018744900682963523, | |
| "loss": 8.3768, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.16802462429838855, | |
| "grad_norm": 0.5058098435401917, | |
| "learning_rate": 0.0001873376315437724, | |
| "loss": 8.332, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.16874886836864023, | |
| "grad_norm": 0.4851613938808441, | |
| "learning_rate": 0.00018722579766635117, | |
| "loss": 8.3578, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.1694731124388919, | |
| "grad_norm": 0.4681190550327301, | |
| "learning_rate": 0.00018711350578458767, | |
| "loss": 8.3757, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.17019735650914358, | |
| "grad_norm": 0.48669615387916565, | |
| "learning_rate": 0.00018700075648810303, | |
| "loss": 8.3425, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.17092160057939526, | |
| "grad_norm": 0.4987237751483917, | |
| "learning_rate": 0.00018688755036892012, | |
| "loss": 8.3175, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.17164584464964694, | |
| "grad_norm": 0.4954805076122284, | |
| "learning_rate": 0.0001867738880214605, | |
| "loss": 8.3055, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.17237008871989862, | |
| "grad_norm": 0.5057269930839539, | |
| "learning_rate": 0.00018665977004254125, | |
| "loss": 8.3266, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.17309433279015027, | |
| "grad_norm": 0.5142120122909546, | |
| "learning_rate": 0.00018654519703137191, | |
| "loss": 8.2739, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.17381857686040195, | |
| "grad_norm": 0.49817487597465515, | |
| "learning_rate": 0.00018643016958955135, | |
| "loss": 8.292, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.17454282093065362, | |
| "grad_norm": 0.49923181533813477, | |
| "learning_rate": 0.00018631468832106446, | |
| "loss": 8.2962, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.1752670650009053, | |
| "grad_norm": 0.5132392048835754, | |
| "learning_rate": 0.00018619875383227912, | |
| "loss": 8.3461, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.17599130907115698, | |
| "grad_norm": 0.5324509739875793, | |
| "learning_rate": 0.000186082366731943, | |
| "loss": 8.2242, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.17671555314140866, | |
| "grad_norm": 0.5011392831802368, | |
| "learning_rate": 0.0001859655276311803, | |
| "loss": 8.2918, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.17743979721166034, | |
| "grad_norm": 0.5092015862464905, | |
| "learning_rate": 0.0001858482371434886, | |
| "loss": 8.2892, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.17816404128191202, | |
| "grad_norm": 0.531007707118988, | |
| "learning_rate": 0.00018573049588473564, | |
| "loss": 8.2796, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.17888828535216367, | |
| "grad_norm": 0.5116980075836182, | |
| "learning_rate": 0.00018561230447315604, | |
| "loss": 8.2792, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.17961252942241535, | |
| "grad_norm": 0.5366231203079224, | |
| "learning_rate": 0.000185493663529348, | |
| "loss": 8.2506, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.18033677349266702, | |
| "grad_norm": 0.595176637172699, | |
| "learning_rate": 0.0001853745736762703, | |
| "loss": 8.2615, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.1810610175629187, | |
| "grad_norm": 0.5762497186660767, | |
| "learning_rate": 0.0001852550355392387, | |
| "loss": 8.2754, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.18178526163317038, | |
| "grad_norm": 0.5076980590820312, | |
| "learning_rate": 0.00018513504974592283, | |
| "loss": 8.2569, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.18250950570342206, | |
| "grad_norm": 0.5133912563323975, | |
| "learning_rate": 0.0001850146169263429, | |
| "loss": 8.2293, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.18323374977367374, | |
| "grad_norm": 0.5173690915107727, | |
| "learning_rate": 0.00018489373771286637, | |
| "loss": 8.2451, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.18395799384392542, | |
| "grad_norm": 0.5147438645362854, | |
| "learning_rate": 0.00018477241274020458, | |
| "loss": 8.1972, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.18468223791417707, | |
| "grad_norm": 0.48920467495918274, | |
| "learning_rate": 0.00018465064264540945, | |
| "loss": 8.2376, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.18540648198442874, | |
| "grad_norm": 0.4959597587585449, | |
| "learning_rate": 0.00018452842806787026, | |
| "loss": 8.2552, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.18613072605468042, | |
| "grad_norm": 0.48315390944480896, | |
| "learning_rate": 0.00018440576964930998, | |
| "loss": 8.2317, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.1868549701249321, | |
| "grad_norm": 0.5170494914054871, | |
| "learning_rate": 0.00018428266803378226, | |
| "loss": 8.2036, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.18757921419518378, | |
| "grad_norm": 0.4896693229675293, | |
| "learning_rate": 0.00018415912386766781, | |
| "loss": 8.2285, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.18830345826543546, | |
| "grad_norm": 0.5025380849838257, | |
| "learning_rate": 0.00018403513779967115, | |
| "loss": 8.2247, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.18902770233568714, | |
| "grad_norm": 0.4876105785369873, | |
| "learning_rate": 0.000183910710480817, | |
| "loss": 8.2347, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.18975194640593882, | |
| "grad_norm": 0.4978480041027069, | |
| "learning_rate": 0.00018378584256444712, | |
| "loss": 8.1992, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.19047619047619047, | |
| "grad_norm": 0.47433796525001526, | |
| "learning_rate": 0.00018366053470621668, | |
| "loss": 8.2212, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.19120043454644214, | |
| "grad_norm": 0.4850807785987854, | |
| "learning_rate": 0.00018353478756409096, | |
| "loss": 8.2226, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.19192467861669382, | |
| "grad_norm": 0.5015024542808533, | |
| "learning_rate": 0.00018340860179834177, | |
| "loss": 8.1905, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.1926489226869455, | |
| "grad_norm": 0.5145589113235474, | |
| "learning_rate": 0.00018328197807154407, | |
| "loss": 8.1793, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.19337316675719718, | |
| "grad_norm": 0.48222506046295166, | |
| "learning_rate": 0.00018315491704857246, | |
| "loss": 8.1716, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.19409741082744886, | |
| "grad_norm": 0.5012249946594238, | |
| "learning_rate": 0.00018302741939659763, | |
| "loss": 8.1869, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.19482165489770054, | |
| "grad_norm": 0.494550496339798, | |
| "learning_rate": 0.00018289948578508307, | |
| "loss": 8.1733, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.1955458989679522, | |
| "grad_norm": 0.51914381980896, | |
| "learning_rate": 0.00018277111688578122, | |
| "loss": 8.1836, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.19627014303820386, | |
| "grad_norm": 0.5137649774551392, | |
| "learning_rate": 0.00018264231337273022, | |
| "loss": 8.1376, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.19699438710845554, | |
| "grad_norm": 0.4887925982475281, | |
| "learning_rate": 0.0001825130759222503, | |
| "loss": 8.1645, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.19771863117870722, | |
| "grad_norm": 0.48226258158683777, | |
| "learning_rate": 0.0001823834052129401, | |
| "loss": 8.1425, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.1984428752489589, | |
| "grad_norm": 0.4919883906841278, | |
| "learning_rate": 0.00018225330192567335, | |
| "loss": 8.1622, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.19916711931921058, | |
| "grad_norm": 0.49928978085517883, | |
| "learning_rate": 0.00018212276674359508, | |
| "loss": 8.1414, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.19989136338946226, | |
| "grad_norm": 0.5267078280448914, | |
| "learning_rate": 0.00018199180035211805, | |
| "loss": 8.0985, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.20061560745971393, | |
| "grad_norm": 0.5004377365112305, | |
| "learning_rate": 0.0001818604034389193, | |
| "loss": 8.1365, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.20133985152996559, | |
| "grad_norm": 0.5070396661758423, | |
| "learning_rate": 0.00018172857669393645, | |
| "loss": 8.1672, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.20206409560021726, | |
| "grad_norm": 0.49480506777763367, | |
| "learning_rate": 0.000181596320809364, | |
| "loss": 8.1374, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.20278833967046894, | |
| "grad_norm": 0.4668210446834564, | |
| "learning_rate": 0.0001814636364796499, | |
| "loss": 8.1812, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.20351258374072062, | |
| "grad_norm": 0.49613460898399353, | |
| "learning_rate": 0.00018133052440149163, | |
| "loss": 8.1357, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.2042368278109723, | |
| "grad_norm": 0.5054727792739868, | |
| "learning_rate": 0.00018119698527383274, | |
| "loss": 8.0969, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.20496107188122398, | |
| "grad_norm": 0.49716633558273315, | |
| "learning_rate": 0.0001810630197978592, | |
| "loss": 8.1176, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.20568531595147566, | |
| "grad_norm": 0.491693377494812, | |
| "learning_rate": 0.00018092862867699557, | |
| "loss": 8.118, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.20640956002172733, | |
| "grad_norm": 0.498813658952713, | |
| "learning_rate": 0.00018079381261690134, | |
| "loss": 8.0726, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.20713380409197898, | |
| "grad_norm": 0.5227881669998169, | |
| "learning_rate": 0.00018065857232546736, | |
| "loss": 8.0672, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.20785804816223066, | |
| "grad_norm": 0.4753732979297638, | |
| "learning_rate": 0.00018052290851281204, | |
| "loss": 8.06, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.20858229223248234, | |
| "grad_norm": 0.510330080986023, | |
| "learning_rate": 0.0001803868218912775, | |
| "loss": 8.0716, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.20930653630273402, | |
| "grad_norm": 0.5118574500083923, | |
| "learning_rate": 0.000180250313175426, | |
| "loss": 8.1125, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.2100307803729857, | |
| "grad_norm": 0.5137168169021606, | |
| "learning_rate": 0.00018011338308203623, | |
| "loss": 8.0358, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.21075502444323738, | |
| "grad_norm": 0.48737525939941406, | |
| "learning_rate": 0.00017997603233009922, | |
| "loss": 8.1156, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.21147926851348905, | |
| "grad_norm": 0.5204469561576843, | |
| "learning_rate": 0.00017983826164081503, | |
| "loss": 8.0484, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.21220351258374073, | |
| "grad_norm": 0.4880678951740265, | |
| "learning_rate": 0.00017970007173758856, | |
| "loss": 8.0651, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.21292775665399238, | |
| "grad_norm": 0.47253793478012085, | |
| "learning_rate": 0.00017956146334602595, | |
| "loss": 8.0648, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.21365200072424406, | |
| "grad_norm": 0.5147983431816101, | |
| "learning_rate": 0.00017942243719393076, | |
| "loss": 8.0625, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.21437624479449574, | |
| "grad_norm": 0.5061482191085815, | |
| "learning_rate": 0.00017928299401130012, | |
| "loss": 8.059, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.21510048886474742, | |
| "grad_norm": 0.5227460265159607, | |
| "learning_rate": 0.00017914313453032093, | |
| "loss": 8.0338, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.2158247329349991, | |
| "grad_norm": 0.5411704182624817, | |
| "learning_rate": 0.00017900285948536587, | |
| "loss": 8.0671, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.21654897700525078, | |
| "grad_norm": 0.5224537253379822, | |
| "learning_rate": 0.00017886216961298981, | |
| "loss": 8.0801, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.21727322107550245, | |
| "grad_norm": 0.5319260954856873, | |
| "learning_rate": 0.00017872106565192567, | |
| "loss": 8.1099, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.21799746514575413, | |
| "grad_norm": 0.5229703783988953, | |
| "learning_rate": 0.00017857954834308074, | |
| "loss": 8.0345, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.21872170921600578, | |
| "grad_norm": 0.4948280453681946, | |
| "learning_rate": 0.0001784376184295327, | |
| "loss": 8.0936, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.21944595328625746, | |
| "grad_norm": 0.4888227581977844, | |
| "learning_rate": 0.00017829527665652562, | |
| "loss": 8.062, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.22017019735650914, | |
| "grad_norm": 0.49134767055511475, | |
| "learning_rate": 0.00017815252377146638, | |
| "loss": 8.0392, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.22089444142676082, | |
| "grad_norm": 0.483167827129364, | |
| "learning_rate": 0.0001780093605239203, | |
| "loss": 8.07, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.2216186854970125, | |
| "grad_norm": 0.4626474380493164, | |
| "learning_rate": 0.00017786578766560758, | |
| "loss": 8.0982, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.22234292956726417, | |
| "grad_norm": 0.4981623888015747, | |
| "learning_rate": 0.0001777218059503991, | |
| "loss": 8.0233, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.22306717363751585, | |
| "grad_norm": 0.5132870674133301, | |
| "learning_rate": 0.00017757741613431263, | |
| "loss": 8.0564, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.22379141770776753, | |
| "grad_norm": 0.4659659266471863, | |
| "learning_rate": 0.00017743261897550875, | |
| "loss": 8.0599, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.22451566177801918, | |
| "grad_norm": 0.48992919921875, | |
| "learning_rate": 0.00017728741523428696, | |
| "loss": 8.0318, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.22523990584827086, | |
| "grad_norm": 0.465867817401886, | |
| "learning_rate": 0.00017714180567308157, | |
| "loss": 8.042, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.22596414991852254, | |
| "grad_norm": 0.48665499687194824, | |
| "learning_rate": 0.0001769957910564578, | |
| "loss": 8.0519, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.22668839398877422, | |
| "grad_norm": 0.47514137625694275, | |
| "learning_rate": 0.00017684937215110778, | |
| "loss": 8.0537, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.2274126380590259, | |
| "grad_norm": 0.5170972943305969, | |
| "learning_rate": 0.00017670254972584638, | |
| "loss": 8.0094, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.22813688212927757, | |
| "grad_norm": 0.5160325765609741, | |
| "learning_rate": 0.0001765553245516073, | |
| "loss": 8.021, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.22886112619952925, | |
| "grad_norm": 0.48066890239715576, | |
| "learning_rate": 0.00017640769740143904, | |
| "loss": 8.0412, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.2295853702697809, | |
| "grad_norm": 0.4954223334789276, | |
| "learning_rate": 0.00017625966905050077, | |
| "loss": 8.0338, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.23030961434003258, | |
| "grad_norm": 0.5118348002433777, | |
| "learning_rate": 0.00017611124027605825, | |
| "loss": 8.0008, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.23103385841028426, | |
| "grad_norm": 0.5256980657577515, | |
| "learning_rate": 0.00017596241185747978, | |
| "loss": 7.9931, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.23175810248053594, | |
| "grad_norm": 0.48749062418937683, | |
| "learning_rate": 0.00017581318457623218, | |
| "loss": 7.9715, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.23248234655078762, | |
| "grad_norm": 0.5072448253631592, | |
| "learning_rate": 0.0001756635592158765, | |
| "loss": 8.011, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.2332065906210393, | |
| "grad_norm": 0.49068352580070496, | |
| "learning_rate": 0.00017551353656206412, | |
| "loss": 8.0219, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.23393083469129097, | |
| "grad_norm": 0.47556236386299133, | |
| "learning_rate": 0.00017536311740253243, | |
| "loss": 8.0213, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.23465507876154265, | |
| "grad_norm": 0.5052993297576904, | |
| "learning_rate": 0.0001752123025271009, | |
| "loss": 7.9925, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.2353793228317943, | |
| "grad_norm": 0.49751394987106323, | |
| "learning_rate": 0.00017506109272766673, | |
| "loss": 7.9627, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.23610356690204598, | |
| "grad_norm": 0.5014733672142029, | |
| "learning_rate": 0.00017490948879820084, | |
| "loss": 7.9567, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.23682781097229766, | |
| "grad_norm": 0.49920758605003357, | |
| "learning_rate": 0.0001747574915347436, | |
| "loss": 8.0418, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.23755205504254934, | |
| "grad_norm": 0.5061244368553162, | |
| "learning_rate": 0.00017460510173540072, | |
| "loss": 7.9813, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.23827629911280102, | |
| "grad_norm": 0.5036886930465698, | |
| "learning_rate": 0.00017445232020033902, | |
| "loss": 7.9537, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.2390005431830527, | |
| "grad_norm": 0.48648810386657715, | |
| "learning_rate": 0.00017429914773178228, | |
| "loss": 7.9614, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.23972478725330437, | |
| "grad_norm": 0.47009527683258057, | |
| "learning_rate": 0.00017414558513400693, | |
| "loss": 8.0103, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.24044903132355605, | |
| "grad_norm": 0.4699738919734955, | |
| "learning_rate": 0.00017399163321333793, | |
| "loss": 7.9483, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.2411732753938077, | |
| "grad_norm": 0.4880366623401642, | |
| "learning_rate": 0.00017383729277814446, | |
| "loss": 8.0285, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.24189751946405938, | |
| "grad_norm": 0.4812169373035431, | |
| "learning_rate": 0.00017368256463883578, | |
| "loss": 8.0094, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.24262176353431106, | |
| "grad_norm": 0.5153531432151794, | |
| "learning_rate": 0.00017352744960785676, | |
| "loss": 7.9488, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.24334600760456274, | |
| "grad_norm": 0.4880661368370056, | |
| "learning_rate": 0.0001733719484996839, | |
| "loss": 7.9457, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.24407025167481441, | |
| "grad_norm": 0.5033547282218933, | |
| "learning_rate": 0.00017321606213082088, | |
| "loss": 7.9726, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.2447944957450661, | |
| "grad_norm": 0.4773882329463959, | |
| "learning_rate": 0.0001730597913197942, | |
| "loss": 8.0068, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.24551873981531777, | |
| "grad_norm": 0.48063740134239197, | |
| "learning_rate": 0.00017290313688714915, | |
| "loss": 7.9865, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.24624298388556945, | |
| "grad_norm": 0.5227491855621338, | |
| "learning_rate": 0.00017274609965544523, | |
| "loss": 7.8864, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.2469672279558211, | |
| "grad_norm": 0.490004301071167, | |
| "learning_rate": 0.00017258868044925195, | |
| "loss": 7.9176, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.24769147202607278, | |
| "grad_norm": 0.49324241280555725, | |
| "learning_rate": 0.0001724308800951445, | |
| "loss": 7.9429, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.24841571609632446, | |
| "grad_norm": 0.5013875961303711, | |
| "learning_rate": 0.00017227269942169936, | |
| "loss": 8.0167, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.24913996016657614, | |
| "grad_norm": 0.49397408962249756, | |
| "learning_rate": 0.00017211413925949005, | |
| "loss": 7.9341, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.2498642042368278, | |
| "grad_norm": 0.48179903626441956, | |
| "learning_rate": 0.00017195520044108268, | |
| "loss": 7.9138, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.2505884483070795, | |
| "grad_norm": 0.5134261846542358, | |
| "learning_rate": 0.00017179588380103163, | |
| "loss": 7.9132, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.2505884483070795, | |
| "eval_loss": 7.9483160972595215, | |
| "eval_runtime": 5.1008, | |
| "eval_samples_per_second": 228.001, | |
| "eval_steps_per_second": 114.099, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.25131269237733117, | |
| "grad_norm": 0.5437692403793335, | |
| "learning_rate": 0.00017163619017587504, | |
| "loss": 7.9127, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.25203693644758285, | |
| "grad_norm": 0.5261752605438232, | |
| "learning_rate": 0.00017147612040413065, | |
| "loss": 7.9209, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.2527611805178345, | |
| "grad_norm": 0.5796931385993958, | |
| "learning_rate": 0.0001713156753262912, | |
| "loss": 7.8785, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.2534854245880862, | |
| "grad_norm": 0.6015821099281311, | |
| "learning_rate": 0.00017115485578482006, | |
| "loss": 8.0264, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.2542096686583379, | |
| "grad_norm": 0.5010411739349365, | |
| "learning_rate": 0.00017099366262414694, | |
| "loss": 7.9666, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.25493391272858956, | |
| "grad_norm": 0.4945250451564789, | |
| "learning_rate": 0.00017083209669066317, | |
| "loss": 7.9885, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.2556581567988412, | |
| "grad_norm": 0.5084425210952759, | |
| "learning_rate": 0.0001706701588327176, | |
| "loss": 7.9407, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.25638240086909286, | |
| "grad_norm": 0.47225329279899597, | |
| "learning_rate": 0.0001705078499006119, | |
| "loss": 7.9456, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.25710664493934454, | |
| "grad_norm": 0.5019710659980774, | |
| "learning_rate": 0.00017034517074659617, | |
| "loss": 7.9546, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.2578308890095962, | |
| "grad_norm": 0.49253225326538086, | |
| "learning_rate": 0.00017018212222486446, | |
| "loss": 7.9919, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.2585551330798479, | |
| "grad_norm": 0.4921097457408905, | |
| "learning_rate": 0.0001700187051915503, | |
| "loss": 7.9367, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.2592793771500996, | |
| "grad_norm": 0.4596784710884094, | |
| "learning_rate": 0.00016985492050472227, | |
| "loss": 7.9454, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.26000362122035126, | |
| "grad_norm": 0.48740360140800476, | |
| "learning_rate": 0.00016969076902437932, | |
| "loss": 7.9385, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.26072786529060293, | |
| "grad_norm": 0.4537501931190491, | |
| "learning_rate": 0.00016952625161244638, | |
| "loss": 7.9553, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.2614521093608546, | |
| "grad_norm": 0.5013776421546936, | |
| "learning_rate": 0.00016936136913276982, | |
| "loss": 7.9313, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.2621763534311063, | |
| "grad_norm": 0.49748286604881287, | |
| "learning_rate": 0.00016919612245111295, | |
| "loss": 7.9337, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.26290059750135797, | |
| "grad_norm": 0.4852049648761749, | |
| "learning_rate": 0.0001690305124351514, | |
| "loss": 7.9322, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.26362484157160965, | |
| "grad_norm": 0.4738268554210663, | |
| "learning_rate": 0.0001688645399544685, | |
| "loss": 7.9062, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.2643490856418613, | |
| "grad_norm": 0.5000738501548767, | |
| "learning_rate": 0.00016869820588055095, | |
| "loss": 7.9441, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.265073329712113, | |
| "grad_norm": 0.4679807424545288, | |
| "learning_rate": 0.00016853151108678398, | |
| "loss": 7.9791, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.2657975737823647, | |
| "grad_norm": 0.4647904336452484, | |
| "learning_rate": 0.00016836445644844697, | |
| "loss": 7.9007, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.2665218178526163, | |
| "grad_norm": 0.491827130317688, | |
| "learning_rate": 0.00016819704284270874, | |
| "loss": 7.9301, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.267246061922868, | |
| "grad_norm": 0.4953727126121521, | |
| "learning_rate": 0.000168029271148623, | |
| "loss": 7.8832, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.26797030599311966, | |
| "grad_norm": 0.5065143704414368, | |
| "learning_rate": 0.0001678611422471236, | |
| "loss": 7.8904, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.26869455006337134, | |
| "grad_norm": 0.49114301800727844, | |
| "learning_rate": 0.00016769265702102018, | |
| "loss": 7.8696, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.269418794133623, | |
| "grad_norm": 0.4655497670173645, | |
| "learning_rate": 0.00016752381635499317, | |
| "loss": 7.9315, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.2701430382038747, | |
| "grad_norm": 0.4854629933834076, | |
| "learning_rate": 0.0001673546211355895, | |
| "loss": 7.8832, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.2708672822741264, | |
| "grad_norm": 0.4579346776008606, | |
| "learning_rate": 0.0001671850722512178, | |
| "loss": 7.8855, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.27159152634437805, | |
| "grad_norm": 0.4732455909252167, | |
| "learning_rate": 0.00016701517059214348, | |
| "loss": 7.9345, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.27231577041462973, | |
| "grad_norm": 0.4654392600059509, | |
| "learning_rate": 0.00016684491705048457, | |
| "loss": 7.9374, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.2730400144848814, | |
| "grad_norm": 0.4884020686149597, | |
| "learning_rate": 0.0001666743125202067, | |
| "loss": 7.9046, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.2737642585551331, | |
| "grad_norm": 0.4872707724571228, | |
| "learning_rate": 0.00016650335789711833, | |
| "loss": 7.9233, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.27448850262538477, | |
| "grad_norm": 0.47144293785095215, | |
| "learning_rate": 0.0001663320540788663, | |
| "loss": 7.9096, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.27521274669563645, | |
| "grad_norm": 0.45968198776245117, | |
| "learning_rate": 0.00016616040196493103, | |
| "loss": 7.8761, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.2759369907658881, | |
| "grad_norm": 0.47557854652404785, | |
| "learning_rate": 0.00016598840245662166, | |
| "loss": 7.8756, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.2766612348361398, | |
| "grad_norm": 0.45728030800819397, | |
| "learning_rate": 0.0001658160564570715, | |
| "loss": 7.9193, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.2773854789063915, | |
| "grad_norm": 0.5167350172996521, | |
| "learning_rate": 0.0001656433648712332, | |
| "loss": 7.8849, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.2781097229766431, | |
| "grad_norm": 0.4883110523223877, | |
| "learning_rate": 0.00016547032860587398, | |
| "loss": 7.8263, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.2788339670468948, | |
| "grad_norm": 0.47416940331459045, | |
| "learning_rate": 0.00016529694856957098, | |
| "loss": 7.9048, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.27955821111714646, | |
| "grad_norm": 0.4862803816795349, | |
| "learning_rate": 0.0001651232256727063, | |
| "loss": 7.8855, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.28028245518739814, | |
| "grad_norm": 0.4723484516143799, | |
| "learning_rate": 0.0001649491608274624, | |
| "loss": 7.9657, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.2810066992576498, | |
| "grad_norm": 0.4739847779273987, | |
| "learning_rate": 0.00016477475494781717, | |
| "loss": 7.8765, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.2817309433279015, | |
| "grad_norm": 0.4680808484554291, | |
| "learning_rate": 0.00016460000894953934, | |
| "loss": 7.9019, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.2824551873981532, | |
| "grad_norm": 0.5003647804260254, | |
| "learning_rate": 0.00016442492375018343, | |
| "loss": 7.839, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.28317943146840485, | |
| "grad_norm": 0.5076538324356079, | |
| "learning_rate": 0.00016424950026908497, | |
| "loss": 7.8935, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.28390367553865653, | |
| "grad_norm": 0.47889307141304016, | |
| "learning_rate": 0.0001640737394273559, | |
| "loss": 7.9099, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.2846279196089082, | |
| "grad_norm": 0.48014241456985474, | |
| "learning_rate": 0.0001638976421478794, | |
| "loss": 7.9593, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.2853521636791599, | |
| "grad_norm": 0.48349729180336, | |
| "learning_rate": 0.00016372120935530536, | |
| "loss": 7.8625, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.28607640774941157, | |
| "grad_norm": 0.4960671067237854, | |
| "learning_rate": 0.00016354444197604529, | |
| "loss": 7.8534, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.28680065181966324, | |
| "grad_norm": 0.515612006187439, | |
| "learning_rate": 0.00016336734093826756, | |
| "loss": 7.872, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.2875248958899149, | |
| "grad_norm": 0.49004271626472473, | |
| "learning_rate": 0.00016318990717189256, | |
| "loss": 7.9025, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.2882491399601666, | |
| "grad_norm": 0.5171812176704407, | |
| "learning_rate": 0.00016301214160858768, | |
| "loss": 7.8026, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.2889733840304182, | |
| "grad_norm": 0.5030158162117004, | |
| "learning_rate": 0.00016283404518176257, | |
| "loss": 7.9481, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.2896976281006699, | |
| "grad_norm": 0.5945943593978882, | |
| "learning_rate": 0.0001626556188265642, | |
| "loss": 7.9135, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.2904218721709216, | |
| "grad_norm": 0.5005214810371399, | |
| "learning_rate": 0.00016247686347987183, | |
| "loss": 7.9069, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.29114611624117326, | |
| "grad_norm": 0.4819411635398865, | |
| "learning_rate": 0.00016229778008029224, | |
| "loss": 7.8972, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.29187036031142494, | |
| "grad_norm": 0.48914363980293274, | |
| "learning_rate": 0.00016211836956815477, | |
| "loss": 7.9025, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.2925946043816766, | |
| "grad_norm": 0.4711272418498993, | |
| "learning_rate": 0.00016193863288550638, | |
| "loss": 7.923, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.2933188484519283, | |
| "grad_norm": 0.4771362841129303, | |
| "learning_rate": 0.00016175857097610653, | |
| "loss": 7.8727, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.29404309252217997, | |
| "grad_norm": 0.4811650216579437, | |
| "learning_rate": 0.00016157818478542254, | |
| "loss": 7.8895, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.29476733659243165, | |
| "grad_norm": 0.49894389510154724, | |
| "learning_rate": 0.00016139747526062442, | |
| "loss": 7.8914, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.29549158066268333, | |
| "grad_norm": 0.48443737626075745, | |
| "learning_rate": 0.00016121644335057993, | |
| "loss": 7.8962, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.296215824732935, | |
| "grad_norm": 0.4693881869316101, | |
| "learning_rate": 0.00016103509000584958, | |
| "loss": 7.8654, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.2969400688031867, | |
| "grad_norm": 0.4889278709888458, | |
| "learning_rate": 0.0001608534161786817, | |
| "loss": 7.8896, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.29766431287343836, | |
| "grad_norm": 0.48877543210983276, | |
| "learning_rate": 0.0001606714228230074, | |
| "loss": 7.9108, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.29838855694369004, | |
| "grad_norm": 0.46180957555770874, | |
| "learning_rate": 0.00016048911089443558, | |
| "loss": 7.9667, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.2991128010139417, | |
| "grad_norm": 0.466511994600296, | |
| "learning_rate": 0.00016030648135024786, | |
| "loss": 7.9088, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.2998370450841934, | |
| "grad_norm": 0.4758537709712982, | |
| "learning_rate": 0.00016012353514939363, | |
| "loss": 7.8952, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.300561289154445, | |
| "grad_norm": 0.5132808089256287, | |
| "learning_rate": 0.00015994027325248492, | |
| "loss": 7.8671, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.3012855332246967, | |
| "grad_norm": 0.45275020599365234, | |
| "learning_rate": 0.00015975669662179152, | |
| "loss": 7.8962, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.3020097772949484, | |
| "grad_norm": 0.5065222382545471, | |
| "learning_rate": 0.00015957280622123574, | |
| "loss": 7.8559, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.30273402136520006, | |
| "grad_norm": 0.4880678057670593, | |
| "learning_rate": 0.00015938860301638742, | |
| "loss": 7.8497, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.30345826543545174, | |
| "grad_norm": 0.4863564968109131, | |
| "learning_rate": 0.0001592040879744589, | |
| "loss": 7.8586, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.3041825095057034, | |
| "grad_norm": 0.49322959780693054, | |
| "learning_rate": 0.0001590192620643, | |
| "loss": 7.8449, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.3049067535759551, | |
| "grad_norm": 0.4878312349319458, | |
| "learning_rate": 0.00015883412625639263, | |
| "loss": 7.8653, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.30563099764620677, | |
| "grad_norm": 0.5146703720092773, | |
| "learning_rate": 0.00015864868152284608, | |
| "loss": 7.9, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.30635524171645845, | |
| "grad_norm": 0.46097350120544434, | |
| "learning_rate": 0.00015846292883739171, | |
| "loss": 7.8997, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.3070794857867101, | |
| "grad_norm": 0.48234981298446655, | |
| "learning_rate": 0.00015827686917537783, | |
| "loss": 7.9008, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.3078037298569618, | |
| "grad_norm": 0.49214646220207214, | |
| "learning_rate": 0.00015809050351376467, | |
| "loss": 7.9094, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.3085279739272135, | |
| "grad_norm": 0.4872777462005615, | |
| "learning_rate": 0.00015790383283111913, | |
| "loss": 7.8567, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.30925221799746516, | |
| "grad_norm": 0.4729660451412201, | |
| "learning_rate": 0.00015771685810760978, | |
| "loss": 7.8577, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.30997646206771684, | |
| "grad_norm": 0.4986751973628998, | |
| "learning_rate": 0.00015752958032500165, | |
| "loss": 7.8491, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.3107007061379685, | |
| "grad_norm": 0.4702273905277252, | |
| "learning_rate": 0.000157342000466651, | |
| "loss": 7.8647, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.3114249502082202, | |
| "grad_norm": 0.48451560735702515, | |
| "learning_rate": 0.0001571541195175003, | |
| "loss": 7.8365, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.3121491942784718, | |
| "grad_norm": 0.4823257029056549, | |
| "learning_rate": 0.0001569659384640729, | |
| "loss": 7.8654, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.3128734383487235, | |
| "grad_norm": 0.4614291191101074, | |
| "learning_rate": 0.00015677745829446803, | |
| "loss": 7.8945, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.3135976824189752, | |
| "grad_norm": 0.47423475980758667, | |
| "learning_rate": 0.00015658867999835546, | |
| "loss": 7.8436, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.31432192648922685, | |
| "grad_norm": 0.486380398273468, | |
| "learning_rate": 0.00015639960456697037, | |
| "loss": 7.836, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.31504617055947853, | |
| "grad_norm": 0.4982565641403198, | |
| "learning_rate": 0.00015621023299310812, | |
| "loss": 7.8526, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.3157704146297302, | |
| "grad_norm": 0.4835852086544037, | |
| "learning_rate": 0.00015602056627111907, | |
| "loss": 7.831, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.3164946586999819, | |
| "grad_norm": 0.508026659488678, | |
| "learning_rate": 0.0001558306053969034, | |
| "loss": 7.837, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.31721890277023357, | |
| "grad_norm": 0.5414742231369019, | |
| "learning_rate": 0.00015564035136790566, | |
| "loss": 7.8466, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.31794314684048525, | |
| "grad_norm": 0.5005372762680054, | |
| "learning_rate": 0.00015544980518310988, | |
| "loss": 7.8585, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.3186673909107369, | |
| "grad_norm": 0.485451877117157, | |
| "learning_rate": 0.00015525896784303398, | |
| "loss": 7.8713, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.3193916349809886, | |
| "grad_norm": 0.4814086854457855, | |
| "learning_rate": 0.0001550678403497248, | |
| "loss": 7.8185, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.3201158790512403, | |
| "grad_norm": 0.4870661497116089, | |
| "learning_rate": 0.00015487642370675265, | |
| "loss": 7.8919, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.32084012312149196, | |
| "grad_norm": 0.530167281627655, | |
| "learning_rate": 0.00015468471891920613, | |
| "loss": 7.8137, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.32156436719174364, | |
| "grad_norm": 0.4842655658721924, | |
| "learning_rate": 0.0001544927269936868, | |
| "loss": 7.7832, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.3222886112619953, | |
| "grad_norm": 0.47289198637008667, | |
| "learning_rate": 0.0001543004489383039, | |
| "loss": 7.8456, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.32301285533224694, | |
| "grad_norm": 0.5019034147262573, | |
| "learning_rate": 0.00015410788576266916, | |
| "loss": 7.8209, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.3237370994024986, | |
| "grad_norm": 0.5903533697128296, | |
| "learning_rate": 0.00015391503847789136, | |
| "loss": 7.9144, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.3244613434727503, | |
| "grad_norm": 0.5589230060577393, | |
| "learning_rate": 0.00015372190809657106, | |
| "loss": 7.7157, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.325185587543002, | |
| "grad_norm": 0.5412530899047852, | |
| "learning_rate": 0.00015352849563279536, | |
| "loss": 7.7944, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.32590983161325365, | |
| "grad_norm": 0.6175811290740967, | |
| "learning_rate": 0.00015333480210213244, | |
| "loss": 7.8087, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.32663407568350533, | |
| "grad_norm": 0.504089891910553, | |
| "learning_rate": 0.0001531408285216264, | |
| "loss": 7.8893, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.327358319753757, | |
| "grad_norm": 0.5161882638931274, | |
| "learning_rate": 0.00015294657590979172, | |
| "loss": 7.8682, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.3280825638240087, | |
| "grad_norm": 0.4699746072292328, | |
| "learning_rate": 0.0001527520452866081, | |
| "loss": 7.8638, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.32880680789426037, | |
| "grad_norm": 0.48906633257865906, | |
| "learning_rate": 0.00015255723767351495, | |
| "loss": 7.8385, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.32953105196451205, | |
| "grad_norm": 0.48731744289398193, | |
| "learning_rate": 0.00015236215409340616, | |
| "loss": 7.8139, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.3302552960347637, | |
| "grad_norm": 0.4950321614742279, | |
| "learning_rate": 0.0001521667955706246, | |
| "loss": 7.8719, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.3309795401050154, | |
| "grad_norm": 0.4786604046821594, | |
| "learning_rate": 0.00015197116313095683, | |
| "loss": 7.8451, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.3317037841752671, | |
| "grad_norm": 0.46545907855033875, | |
| "learning_rate": 0.00015177525780162775, | |
| "loss": 7.8802, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.33242802824551876, | |
| "grad_norm": 0.4908168315887451, | |
| "learning_rate": 0.00015157908061129508, | |
| "loss": 7.8576, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.33315227231577044, | |
| "grad_norm": 0.5020480751991272, | |
| "learning_rate": 0.00015138263259004402, | |
| "loss": 7.8606, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.3338765163860221, | |
| "grad_norm": 0.48662593960762024, | |
| "learning_rate": 0.00015118591476938188, | |
| "loss": 7.8735, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.33460076045627374, | |
| "grad_norm": 0.46343091130256653, | |
| "learning_rate": 0.00015098892818223258, | |
| "loss": 7.9559, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.3353250045265254, | |
| "grad_norm": 0.467752069234848, | |
| "learning_rate": 0.0001507916738629314, | |
| "loss": 7.8742, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.3360492485967771, | |
| "grad_norm": 0.4827868342399597, | |
| "learning_rate": 0.00015059415284721924, | |
| "loss": 7.8779, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.3367734926670288, | |
| "grad_norm": 0.4806783199310303, | |
| "learning_rate": 0.00015039636617223754, | |
| "loss": 7.8949, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.33749773673728045, | |
| "grad_norm": 0.4673584997653961, | |
| "learning_rate": 0.00015019831487652255, | |
| "loss": 7.8997, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.33822198080753213, | |
| "grad_norm": 0.5170037746429443, | |
| "learning_rate": 0.00015000000000000001, | |
| "loss": 7.8233, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.3389462248777838, | |
| "grad_norm": 0.5105298757553101, | |
| "learning_rate": 0.00014980142258397972, | |
| "loss": 7.8252, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.3396704689480355, | |
| "grad_norm": 0.476662814617157, | |
| "learning_rate": 0.00014960258367114997, | |
| "loss": 7.843, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.34039471301828716, | |
| "grad_norm": 0.49604159593582153, | |
| "learning_rate": 0.0001494034843055721, | |
| "loss": 7.8556, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.34111895708853884, | |
| "grad_norm": 0.48032739758491516, | |
| "learning_rate": 0.00014920412553267508, | |
| "loss": 7.8577, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.3418432011587905, | |
| "grad_norm": 0.49723532795906067, | |
| "learning_rate": 0.00014900450839924994, | |
| "loss": 7.864, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.3425674452290422, | |
| "grad_norm": 0.47292062640190125, | |
| "learning_rate": 0.00014880463395344434, | |
| "loss": 7.8405, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 0.3432916892992939, | |
| "grad_norm": 0.49656134843826294, | |
| "learning_rate": 0.00014860450324475703, | |
| "loss": 7.832, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.34401593336954556, | |
| "grad_norm": 0.4985675811767578, | |
| "learning_rate": 0.0001484041173240323, | |
| "loss": 7.8131, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.34474017743979724, | |
| "grad_norm": 0.4732464551925659, | |
| "learning_rate": 0.0001482034772434545, | |
| "loss": 7.8819, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.3454644215100489, | |
| "grad_norm": 0.4878876507282257, | |
| "learning_rate": 0.00014800258405654257, | |
| "loss": 7.8256, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 0.34618866558030054, | |
| "grad_norm": 0.49212250113487244, | |
| "learning_rate": 0.00014780143881814442, | |
| "loss": 7.816, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.3469129096505522, | |
| "grad_norm": 0.4787651598453522, | |
| "learning_rate": 0.00014760004258443151, | |
| "loss": 7.856, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 0.3476371537208039, | |
| "grad_norm": 0.48412755131721497, | |
| "learning_rate": 0.00014739839641289313, | |
| "loss": 7.8497, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.34836139779105557, | |
| "grad_norm": 0.4971597492694855, | |
| "learning_rate": 0.00014719650136233096, | |
| "loss": 7.8446, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 0.34908564186130725, | |
| "grad_norm": 0.4982038736343384, | |
| "learning_rate": 0.00014699435849285352, | |
| "loss": 7.7839, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.34980988593155893, | |
| "grad_norm": 0.5108397603034973, | |
| "learning_rate": 0.00014679196886587052, | |
| "loss": 7.7931, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.3505341300018106, | |
| "grad_norm": 0.4721708297729492, | |
| "learning_rate": 0.00014658933354408743, | |
| "loss": 7.8607, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.3512583740720623, | |
| "grad_norm": 0.4693983495235443, | |
| "learning_rate": 0.0001463864535914997, | |
| "loss": 7.8773, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.35198261814231396, | |
| "grad_norm": 0.5026690363883972, | |
| "learning_rate": 0.00014618333007338744, | |
| "loss": 7.8628, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.35270686221256564, | |
| "grad_norm": 0.48804745078086853, | |
| "learning_rate": 0.00014597996405630947, | |
| "loss": 7.84, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 0.3534311062828173, | |
| "grad_norm": 0.4530211389064789, | |
| "learning_rate": 0.0001457763566080981, | |
| "loss": 7.859, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.354155350353069, | |
| "grad_norm": 0.48494988679885864, | |
| "learning_rate": 0.0001455725087978533, | |
| "loss": 7.7981, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 0.3548795944233207, | |
| "grad_norm": 0.4591243267059326, | |
| "learning_rate": 0.00014536842169593703, | |
| "loss": 7.8371, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.35560383849357236, | |
| "grad_norm": 0.49914512038230896, | |
| "learning_rate": 0.00014516409637396787, | |
| "loss": 7.829, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 0.35632808256382403, | |
| "grad_norm": 0.48346269130706787, | |
| "learning_rate": 0.00014495953390481506, | |
| "loss": 7.7837, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.35705232663407566, | |
| "grad_norm": 0.4839797019958496, | |
| "learning_rate": 0.00014475473536259325, | |
| "loss": 7.8779, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 0.35777657070432733, | |
| "grad_norm": 0.4998239576816559, | |
| "learning_rate": 0.00014454970182265655, | |
| "loss": 7.8213, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.358500814774579, | |
| "grad_norm": 0.5087825059890747, | |
| "learning_rate": 0.000144344434361593, | |
| "loss": 7.8142, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.3592250588448307, | |
| "grad_norm": 0.47627493739128113, | |
| "learning_rate": 0.00014413893405721895, | |
| "loss": 7.8951, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.35994930291508237, | |
| "grad_norm": 0.5072402954101562, | |
| "learning_rate": 0.0001439332019885733, | |
| "loss": 7.8125, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 0.36067354698533405, | |
| "grad_norm": 0.5031032562255859, | |
| "learning_rate": 0.0001437272392359119, | |
| "loss": 7.8159, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.3613977910555857, | |
| "grad_norm": 0.5236200094223022, | |
| "learning_rate": 0.000143521046880702, | |
| "loss": 7.8519, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 0.3621220351258374, | |
| "grad_norm": 0.5609799027442932, | |
| "learning_rate": 0.00014331462600561626, | |
| "loss": 7.8537, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.3628462791960891, | |
| "grad_norm": 0.4694886803627014, | |
| "learning_rate": 0.0001431079776945274, | |
| "loss": 7.9079, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 0.36357052326634076, | |
| "grad_norm": 0.5094586610794067, | |
| "learning_rate": 0.00014290110303250225, | |
| "loss": 7.8228, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.36429476733659244, | |
| "grad_norm": 0.4992700219154358, | |
| "learning_rate": 0.00014269400310579623, | |
| "loss": 7.8663, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 0.3650190114068441, | |
| "grad_norm": 0.4908367395401001, | |
| "learning_rate": 0.00014248667900184752, | |
| "loss": 7.8586, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.3657432554770958, | |
| "grad_norm": 0.46805062890052795, | |
| "learning_rate": 0.00014227913180927152, | |
| "loss": 7.8644, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.3664674995473475, | |
| "grad_norm": 0.49678292870521545, | |
| "learning_rate": 0.00014207136261785484, | |
| "loss": 7.8759, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.36719174361759915, | |
| "grad_norm": 0.46577930450439453, | |
| "learning_rate": 0.00014186337251854994, | |
| "loss": 7.8668, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 0.36791598768785083, | |
| "grad_norm": 0.4829874038696289, | |
| "learning_rate": 0.00014165516260346913, | |
| "loss": 7.8516, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.36864023175810245, | |
| "grad_norm": 0.4831247925758362, | |
| "learning_rate": 0.00014144673396587892, | |
| "loss": 7.8148, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 0.36936447582835413, | |
| "grad_norm": 0.49740880727767944, | |
| "learning_rate": 0.00014123808770019432, | |
| "loss": 7.864, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.3700887198986058, | |
| "grad_norm": 0.4826888144016266, | |
| "learning_rate": 0.00014102922490197308, | |
| "loss": 7.887, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 0.3708129639688575, | |
| "grad_norm": 0.4531131684780121, | |
| "learning_rate": 0.0001408201466679098, | |
| "loss": 7.901, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.37153720803910917, | |
| "grad_norm": 0.4634702801704407, | |
| "learning_rate": 0.00014061085409583043, | |
| "loss": 7.9295, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 0.37226145210936085, | |
| "grad_norm": 0.48286962509155273, | |
| "learning_rate": 0.0001404013482846863, | |
| "loss": 7.8609, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.3729856961796125, | |
| "grad_norm": 0.49540573358535767, | |
| "learning_rate": 0.00014019163033454843, | |
| "loss": 7.8966, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.3737099402498642, | |
| "grad_norm": 0.471544474363327, | |
| "learning_rate": 0.00013998170134660169, | |
| "loss": 7.9148, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.3744341843201159, | |
| "grad_norm": 0.4708135426044464, | |
| "learning_rate": 0.0001397715624231391, | |
| "loss": 7.882, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 0.37515842839036756, | |
| "grad_norm": 0.5138322710990906, | |
| "learning_rate": 0.0001395612146675561, | |
| "loss": 7.8323, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.37588267246061924, | |
| "grad_norm": 0.5133697390556335, | |
| "learning_rate": 0.00013935065918434445, | |
| "loss": 7.867, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 0.3766069165308709, | |
| "grad_norm": 0.44749224185943604, | |
| "learning_rate": 0.00013913989707908683, | |
| "loss": 7.8333, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.3773311606011226, | |
| "grad_norm": 0.47725266218185425, | |
| "learning_rate": 0.00013892892945845077, | |
| "loss": 7.8102, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 0.3780554046713743, | |
| "grad_norm": 0.4651695787906647, | |
| "learning_rate": 0.00013871775743018293, | |
| "loss": 7.9104, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.37877964874162595, | |
| "grad_norm": 0.4996732771396637, | |
| "learning_rate": 0.0001385063821031033, | |
| "loss": 7.8928, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 0.37950389281187763, | |
| "grad_norm": 0.465753436088562, | |
| "learning_rate": 0.00013829480458709927, | |
| "loss": 7.8479, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.38022813688212925, | |
| "grad_norm": 0.4831571877002716, | |
| "learning_rate": 0.00013808302599312, | |
| "loss": 7.8302, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.38095238095238093, | |
| "grad_norm": 0.513978123664856, | |
| "learning_rate": 0.0001378710474331704, | |
| "loss": 7.8111, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 0.3816766250226326, | |
| "grad_norm": 0.44832077622413635, | |
| "learning_rate": 0.00013765887002030529, | |
| "loss": 7.9085, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 0.3824008690928843, | |
| "grad_norm": 0.46730518341064453, | |
| "learning_rate": 0.00013744649486862378, | |
| "loss": 7.8403, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.38312511316313597, | |
| "grad_norm": 0.47098594903945923, | |
| "learning_rate": 0.00013723392309326316, | |
| "loss": 7.8301, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 0.38384935723338764, | |
| "grad_norm": 0.46994248032569885, | |
| "learning_rate": 0.00013702115581039313, | |
| "loss": 7.8654, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.3845736013036393, | |
| "grad_norm": 0.46916016936302185, | |
| "learning_rate": 0.00013680819413721003, | |
| "loss": 7.8193, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 0.385297845373891, | |
| "grad_norm": 0.48833534121513367, | |
| "learning_rate": 0.00013659503919193089, | |
| "loss": 7.7667, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 0.3860220894441427, | |
| "grad_norm": 0.444770485162735, | |
| "learning_rate": 0.00013638169209378756, | |
| "loss": 7.8987, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 0.38674633351439436, | |
| "grad_norm": 0.46796807646751404, | |
| "learning_rate": 0.00013616815396302081, | |
| "loss": 7.8328, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 0.38747057758464604, | |
| "grad_norm": 0.49306872487068176, | |
| "learning_rate": 0.00013595442592087453, | |
| "loss": 7.8562, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.3881948216548977, | |
| "grad_norm": 0.45947471261024475, | |
| "learning_rate": 0.00013574050908958976, | |
| "loss": 7.8148, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 0.3889190657251494, | |
| "grad_norm": 0.48289933800697327, | |
| "learning_rate": 0.00013552640459239888, | |
| "loss": 7.8374, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.38964330979540107, | |
| "grad_norm": 0.49065014719963074, | |
| "learning_rate": 0.00013531211355351962, | |
| "loss": 7.782, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 0.39036755386565275, | |
| "grad_norm": 0.48593437671661377, | |
| "learning_rate": 0.00013509763709814923, | |
| "loss": 7.8095, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 0.3910917979359044, | |
| "grad_norm": 0.4995581805706024, | |
| "learning_rate": 0.00013488297635245848, | |
| "loss": 7.8081, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.39181604200615605, | |
| "grad_norm": 0.4646143317222595, | |
| "learning_rate": 0.0001346681324435859, | |
| "loss": 7.9134, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 0.39254028607640773, | |
| "grad_norm": 0.49974748492240906, | |
| "learning_rate": 0.00013445310649963169, | |
| "loss": 7.8236, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 0.3932645301466594, | |
| "grad_norm": 0.4993489384651184, | |
| "learning_rate": 0.00013423789964965194, | |
| "loss": 7.8141, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 0.3939887742169111, | |
| "grad_norm": 0.4687751829624176, | |
| "learning_rate": 0.00013402251302365264, | |
| "loss": 7.8152, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.39471301828716276, | |
| "grad_norm": 0.47237464785575867, | |
| "learning_rate": 0.00013380694775258367, | |
| "loss": 7.8618, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.39543726235741444, | |
| "grad_norm": 0.4689088463783264, | |
| "learning_rate": 0.00013359120496833304, | |
| "loss": 7.897, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 0.3961615064276661, | |
| "grad_norm": 0.4774571657180786, | |
| "learning_rate": 0.00013337528580372078, | |
| "loss": 7.8143, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 0.3968857504979178, | |
| "grad_norm": 0.49935322999954224, | |
| "learning_rate": 0.00013315919139249307, | |
| "loss": 7.8004, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 0.3976099945681695, | |
| "grad_norm": 0.5764620304107666, | |
| "learning_rate": 0.00013294292286931627, | |
| "loss": 7.8003, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 0.39833423863842116, | |
| "grad_norm": 0.5829343795776367, | |
| "learning_rate": 0.00013272648136977092, | |
| "loss": 7.771, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.39905848270867283, | |
| "grad_norm": 0.49663594365119934, | |
| "learning_rate": 0.00013250986803034598, | |
| "loss": 7.8275, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 0.3997827267789245, | |
| "grad_norm": 0.4658512473106384, | |
| "learning_rate": 0.0001322930839884325, | |
| "loss": 7.8529, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 0.4005069708491762, | |
| "grad_norm": 0.47856324911117554, | |
| "learning_rate": 0.000132076130382318, | |
| "loss": 7.8692, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 0.40123121491942787, | |
| "grad_norm": 0.5036735534667969, | |
| "learning_rate": 0.00013185900835118025, | |
| "loss": 7.8315, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 0.40195545898967955, | |
| "grad_norm": 0.4810890853404999, | |
| "learning_rate": 0.00013164171903508153, | |
| "loss": 7.8672, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.40267970305993117, | |
| "grad_norm": 0.5102052092552185, | |
| "learning_rate": 0.00013142426357496225, | |
| "loss": 7.8202, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 0.40340394713018285, | |
| "grad_norm": 0.4854986071586609, | |
| "learning_rate": 0.0001312066431126355, | |
| "loss": 7.8389, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 0.40412819120043453, | |
| "grad_norm": 0.4611469805240631, | |
| "learning_rate": 0.0001309888587907805, | |
| "loss": 7.8226, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 0.4048524352706862, | |
| "grad_norm": 0.49772679805755615, | |
| "learning_rate": 0.00013077091175293706, | |
| "loss": 7.8254, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 0.4055766793409379, | |
| "grad_norm": 0.47026780247688293, | |
| "learning_rate": 0.00013055280314349928, | |
| "loss": 7.8621, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.40630092341118956, | |
| "grad_norm": 0.4732389748096466, | |
| "learning_rate": 0.00013033453410770963, | |
| "loss": 7.877, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 0.40702516748144124, | |
| "grad_norm": 0.47972285747528076, | |
| "learning_rate": 0.000130116105791653, | |
| "loss": 7.8747, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 0.4077494115516929, | |
| "grad_norm": 0.48406916856765747, | |
| "learning_rate": 0.0001298975193422506, | |
| "loss": 7.8502, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 0.4084736556219446, | |
| "grad_norm": 0.47579431533813477, | |
| "learning_rate": 0.0001296787759072539, | |
| "loss": 7.8294, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 0.4091978996921963, | |
| "grad_norm": 0.48034512996673584, | |
| "learning_rate": 0.0001294598766352388, | |
| "loss": 7.8261, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.40992214376244795, | |
| "grad_norm": 0.44753938913345337, | |
| "learning_rate": 0.00012924082267559939, | |
| "loss": 7.8905, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 0.41064638783269963, | |
| "grad_norm": 0.4587608575820923, | |
| "learning_rate": 0.00012902161517854197, | |
| "loss": 7.8683, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 0.4113706319029513, | |
| "grad_norm": 0.47434911131858826, | |
| "learning_rate": 0.00012880225529507912, | |
| "loss": 7.8635, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 0.412094875973203, | |
| "grad_norm": 0.4594435691833496, | |
| "learning_rate": 0.00012858274417702344, | |
| "loss": 7.8942, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 0.41281912004345467, | |
| "grad_norm": 0.4879729747772217, | |
| "learning_rate": 0.00012836308297698175, | |
| "loss": 7.8623, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.41354336411370635, | |
| "grad_norm": 0.5010982155799866, | |
| "learning_rate": 0.00012814327284834886, | |
| "loss": 7.8099, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 0.41426760818395797, | |
| "grad_norm": 0.47567158937454224, | |
| "learning_rate": 0.00012792331494530158, | |
| "loss": 7.8321, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 0.41499185225420965, | |
| "grad_norm": 0.47433096170425415, | |
| "learning_rate": 0.00012770321042279264, | |
| "loss": 7.8615, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 0.4157160963244613, | |
| "grad_norm": 0.47150593996047974, | |
| "learning_rate": 0.00012748296043654472, | |
| "loss": 7.8348, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 0.416440340394713, | |
| "grad_norm": 0.43587684631347656, | |
| "learning_rate": 0.0001272625661430442, | |
| "loss": 7.8704, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.4171645844649647, | |
| "grad_norm": 0.4814154803752899, | |
| "learning_rate": 0.00012704202869953521, | |
| "loss": 7.8232, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.41788882853521636, | |
| "grad_norm": 0.4678634703159332, | |
| "learning_rate": 0.00012682134926401354, | |
| "loss": 7.8351, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 0.41861307260546804, | |
| "grad_norm": 0.48871251940727234, | |
| "learning_rate": 0.00012660052899522058, | |
| "loss": 7.7851, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 0.4193373166757197, | |
| "grad_norm": 0.48269665241241455, | |
| "learning_rate": 0.00012637956905263718, | |
| "loss": 7.8597, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 0.4200615607459714, | |
| "grad_norm": 0.46834123134613037, | |
| "learning_rate": 0.00012615847059647752, | |
| "loss": 7.8509, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.4207858048162231, | |
| "grad_norm": 0.47934436798095703, | |
| "learning_rate": 0.00012593723478768323, | |
| "loss": 7.8651, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 0.42151004888647475, | |
| "grad_norm": 0.45488452911376953, | |
| "learning_rate": 0.00012571586278791705, | |
| "loss": 7.8629, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.42223429295672643, | |
| "grad_norm": 0.4674306809902191, | |
| "learning_rate": 0.00012549435575955683, | |
| "loss": 7.8591, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 0.4229585370269781, | |
| "grad_norm": 0.47342684864997864, | |
| "learning_rate": 0.0001252727148656895, | |
| "loss": 7.7871, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 0.4236827810972298, | |
| "grad_norm": 0.49287477135658264, | |
| "learning_rate": 0.0001250509412701048, | |
| "loss": 7.8044, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.42440702516748147, | |
| "grad_norm": 0.47948122024536133, | |
| "learning_rate": 0.00012482903613728928, | |
| "loss": 7.872, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 0.4251312692377331, | |
| "grad_norm": 0.46859344840049744, | |
| "learning_rate": 0.00012460700063242027, | |
| "loss": 7.8114, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 0.42585551330798477, | |
| "grad_norm": 0.49053072929382324, | |
| "learning_rate": 0.00012438483592135948, | |
| "loss": 7.7944, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.42657975737823645, | |
| "grad_norm": 0.5255662202835083, | |
| "learning_rate": 0.00012416254317064714, | |
| "loss": 7.7864, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 0.4273040014484881, | |
| "grad_norm": 0.5337214469909668, | |
| "learning_rate": 0.00012394012354749584, | |
| "loss": 7.8267, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.4280282455187398, | |
| "grad_norm": 0.48263901472091675, | |
| "learning_rate": 0.0001237175782197843, | |
| "loss": 7.7757, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 0.4287524895889915, | |
| "grad_norm": 0.468227356672287, | |
| "learning_rate": 0.00012349490835605127, | |
| "loss": 7.8892, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 0.42947673365924316, | |
| "grad_norm": 0.4746133089065552, | |
| "learning_rate": 0.00012327211512548945, | |
| "loss": 7.8251, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 0.43020097772949484, | |
| "grad_norm": 0.4931454062461853, | |
| "learning_rate": 0.00012304919969793928, | |
| "loss": 7.7892, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 0.4309252217997465, | |
| "grad_norm": 0.44330355525016785, | |
| "learning_rate": 0.00012282616324388283, | |
| "loss": 7.8673, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.4316494658699982, | |
| "grad_norm": 0.5240752100944519, | |
| "learning_rate": 0.00012260300693443777, | |
| "loss": 7.8191, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 0.4323737099402499, | |
| "grad_norm": 0.5226469039916992, | |
| "learning_rate": 0.00012237973194135086, | |
| "loss": 7.799, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 0.43309795401050155, | |
| "grad_norm": 0.5419146418571472, | |
| "learning_rate": 0.00012215633943699232, | |
| "loss": 7.753, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 0.43382219808075323, | |
| "grad_norm": 0.5095767974853516, | |
| "learning_rate": 0.00012193283059434918, | |
| "loss": 7.899, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 0.4345464421510049, | |
| "grad_norm": 0.5662295818328857, | |
| "learning_rate": 0.0001217092065870195, | |
| "loss": 7.8183, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.4352706862212566, | |
| "grad_norm": 0.46471965312957764, | |
| "learning_rate": 0.00012148546858920591, | |
| "loss": 7.8282, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 0.43599493029150826, | |
| "grad_norm": 0.5106796622276306, | |
| "learning_rate": 0.00012126161777570967, | |
| "loss": 7.8318, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 0.4367191743617599, | |
| "grad_norm": 0.48581328988075256, | |
| "learning_rate": 0.00012103765532192437, | |
| "loss": 7.9178, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 0.43744341843201157, | |
| "grad_norm": 0.4774625301361084, | |
| "learning_rate": 0.00012081358240382983, | |
| "loss": 7.861, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 0.43816766250226324, | |
| "grad_norm": 0.4969928562641144, | |
| "learning_rate": 0.00012058940019798588, | |
| "loss": 7.8667, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.4388919065725149, | |
| "grad_norm": 0.4804958403110504, | |
| "learning_rate": 0.00012036510988152618, | |
| "loss": 7.8884, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 0.4396161506427666, | |
| "grad_norm": 0.4955686330795288, | |
| "learning_rate": 0.0001201407126321521, | |
| "loss": 7.8276, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 0.4403403947130183, | |
| "grad_norm": 0.49440160393714905, | |
| "learning_rate": 0.00011991620962812638, | |
| "loss": 7.8653, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 0.44106463878326996, | |
| "grad_norm": 0.49569594860076904, | |
| "learning_rate": 0.00011969160204826721, | |
| "loss": 7.8461, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 0.44178888285352164, | |
| "grad_norm": 0.4727310240268707, | |
| "learning_rate": 0.00011946689107194182, | |
| "loss": 7.8543, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.4425131269237733, | |
| "grad_norm": 0.4789380729198456, | |
| "learning_rate": 0.00011924207787906032, | |
| "loss": 7.8264, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 0.443237370994025, | |
| "grad_norm": 0.4611952006816864, | |
| "learning_rate": 0.00011901716365006956, | |
| "loss": 7.868, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 0.44396161506427667, | |
| "grad_norm": 0.4959096312522888, | |
| "learning_rate": 0.00011879214956594693, | |
| "loss": 7.8141, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 0.44468585913452835, | |
| "grad_norm": 0.4807751178741455, | |
| "learning_rate": 0.00011856703680819414, | |
| "loss": 7.8085, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 0.44541010320478003, | |
| "grad_norm": 0.4968527555465698, | |
| "learning_rate": 0.00011834182655883097, | |
| "loss": 7.8014, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.4461343472750317, | |
| "grad_norm": 0.4878495931625366, | |
| "learning_rate": 0.00011811652000038915, | |
| "loss": 7.7984, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 0.4468585913452834, | |
| "grad_norm": 0.4903671443462372, | |
| "learning_rate": 0.00011789111831590606, | |
| "loss": 7.862, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 0.44758283541553506, | |
| "grad_norm": 0.4804092347621918, | |
| "learning_rate": 0.00011766562268891864, | |
| "loss": 7.798, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 0.4483070794857867, | |
| "grad_norm": 0.5215314626693726, | |
| "learning_rate": 0.00011744003430345705, | |
| "loss": 7.7989, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 0.44903132355603836, | |
| "grad_norm": 0.49652373790740967, | |
| "learning_rate": 0.00011721435434403849, | |
| "loss": 7.8358, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.44975556762629004, | |
| "grad_norm": 0.4879864454269409, | |
| "learning_rate": 0.00011698858399566109, | |
| "loss": 7.8349, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 0.4504798116965417, | |
| "grad_norm": 0.4709030091762543, | |
| "learning_rate": 0.00011676272444379748, | |
| "loss": 7.7974, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 0.4512040557667934, | |
| "grad_norm": 0.47246500849723816, | |
| "learning_rate": 0.00011653677687438874, | |
| "loss": 7.8776, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 0.4519282998370451, | |
| "grad_norm": 0.46343374252319336, | |
| "learning_rate": 0.00011631074247383808, | |
| "loss": 7.8622, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 0.45265254390729676, | |
| "grad_norm": 0.4682675898075104, | |
| "learning_rate": 0.00011608462242900471, | |
| "loss": 7.8559, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.45337678797754843, | |
| "grad_norm": 0.4654393494129181, | |
| "learning_rate": 0.00011585841792719741, | |
| "loss": 7.8423, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 0.4541010320478001, | |
| "grad_norm": 0.44032153487205505, | |
| "learning_rate": 0.00011563213015616856, | |
| "loss": 7.8894, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 0.4548252761180518, | |
| "grad_norm": 0.4499657452106476, | |
| "learning_rate": 0.0001154057603041077, | |
| "loss": 7.8771, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.45554952018830347, | |
| "grad_norm": 0.4855179190635681, | |
| "learning_rate": 0.0001151793095596354, | |
| "loss": 7.8613, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 0.45627376425855515, | |
| "grad_norm": 0.4786494970321655, | |
| "learning_rate": 0.00011495277911179694, | |
| "loss": 7.8229, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.4569980083288068, | |
| "grad_norm": 0.48533812165260315, | |
| "learning_rate": 0.00011472617015005609, | |
| "loss": 7.8304, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 0.4577222523990585, | |
| "grad_norm": 0.4566459059715271, | |
| "learning_rate": 0.00011449948386428894, | |
| "loss": 7.8429, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 0.4584464964693102, | |
| "grad_norm": 0.4726155400276184, | |
| "learning_rate": 0.00011427272144477757, | |
| "loss": 7.7996, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 0.4591707405395618, | |
| "grad_norm": 0.44978711009025574, | |
| "learning_rate": 0.0001140458840822038, | |
| "loss": 7.8547, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 0.4598949846098135, | |
| "grad_norm": 0.503386378288269, | |
| "learning_rate": 0.00011381897296764296, | |
| "loss": 7.8874, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.46061922868006516, | |
| "grad_norm": 0.4718452990055084, | |
| "learning_rate": 0.0001135919892925577, | |
| "loss": 7.7894, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 0.46134347275031684, | |
| "grad_norm": 0.47960880398750305, | |
| "learning_rate": 0.00011336493424879158, | |
| "loss": 7.8652, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 0.4620677168205685, | |
| "grad_norm": 0.48030349612236023, | |
| "learning_rate": 0.00011313780902856294, | |
| "loss": 7.8594, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 0.4627919608908202, | |
| "grad_norm": 0.448631227016449, | |
| "learning_rate": 0.00011291061482445861, | |
| "loss": 7.9241, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 0.4635162049610719, | |
| "grad_norm": 0.5188454985618591, | |
| "learning_rate": 0.00011268335282942765, | |
| "loss": 7.7786, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.46424044903132355, | |
| "grad_norm": 0.47939813137054443, | |
| "learning_rate": 0.00011245602423677506, | |
| "loss": 7.8136, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 0.46496469310157523, | |
| "grad_norm": 0.5121312737464905, | |
| "learning_rate": 0.00011222863024015551, | |
| "loss": 7.91, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 0.4656889371718269, | |
| "grad_norm": 0.49285969138145447, | |
| "learning_rate": 0.00011200117203356715, | |
| "loss": 7.8379, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 0.4664131812420786, | |
| "grad_norm": 0.4866684079170227, | |
| "learning_rate": 0.0001117736508113452, | |
| "loss": 7.8206, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 0.46713742531233027, | |
| "grad_norm": 0.46585652232170105, | |
| "learning_rate": 0.00011154606776815587, | |
| "loss": 7.8355, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.46786166938258195, | |
| "grad_norm": 0.49136561155319214, | |
| "learning_rate": 0.00011131842409898982, | |
| "loss": 7.8392, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 0.4685859134528336, | |
| "grad_norm": 0.4672096073627472, | |
| "learning_rate": 0.00011109072099915625, | |
| "loss": 7.8878, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 0.4693101575230853, | |
| "grad_norm": 0.5394557118415833, | |
| "learning_rate": 0.00011086295966427622, | |
| "loss": 7.819, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 0.470034401593337, | |
| "grad_norm": 0.4872957766056061, | |
| "learning_rate": 0.00011063514129027672, | |
| "loss": 7.8399, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 0.4707586456635886, | |
| "grad_norm": 0.5272302627563477, | |
| "learning_rate": 0.00011040726707338416, | |
| "loss": 7.9755, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.4714828897338403, | |
| "grad_norm": 0.4649381935596466, | |
| "learning_rate": 0.00011017933821011819, | |
| "loss": 7.8441, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 0.47220713380409196, | |
| "grad_norm": 0.4975007474422455, | |
| "learning_rate": 0.0001099513558972854, | |
| "loss": 7.8123, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 0.47293137787434364, | |
| "grad_norm": 0.4799993336200714, | |
| "learning_rate": 0.000109723321331973, | |
| "loss": 7.8462, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 0.4736556219445953, | |
| "grad_norm": 0.46020984649658203, | |
| "learning_rate": 0.00010949523571154266, | |
| "loss": 7.8492, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 0.474379866014847, | |
| "grad_norm": 0.48827457427978516, | |
| "learning_rate": 0.00010926710023362398, | |
| "loss": 7.8295, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.4751041100850987, | |
| "grad_norm": 0.509655773639679, | |
| "learning_rate": 0.0001090389160961085, | |
| "loss": 7.8165, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 0.47582835415535035, | |
| "grad_norm": 0.48283475637435913, | |
| "learning_rate": 0.00010881068449714315, | |
| "loss": 7.8341, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 0.47655259822560203, | |
| "grad_norm": 0.4844430088996887, | |
| "learning_rate": 0.00010858240663512416, | |
| "loss": 7.8634, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 0.4772768422958537, | |
| "grad_norm": 0.4722267985343933, | |
| "learning_rate": 0.00010835408370869063, | |
| "loss": 7.8826, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 0.4780010863661054, | |
| "grad_norm": 0.4804382026195526, | |
| "learning_rate": 0.00010812571691671826, | |
| "loss": 7.8242, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.47872533043635707, | |
| "grad_norm": 0.46896272897720337, | |
| "learning_rate": 0.00010789730745831312, | |
| "loss": 7.9247, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 0.47944957450660874, | |
| "grad_norm": 0.5058935284614563, | |
| "learning_rate": 0.00010766885653280532, | |
| "loss": 7.8299, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 0.4801738185768604, | |
| "grad_norm": 0.49389636516571045, | |
| "learning_rate": 0.00010744036533974267, | |
| "loss": 7.8462, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 0.4808980626471121, | |
| "grad_norm": 0.47579723596572876, | |
| "learning_rate": 0.00010721183507888442, | |
| "loss": 7.8666, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 0.4816223067173637, | |
| "grad_norm": 0.47958090901374817, | |
| "learning_rate": 0.00010698326695019496, | |
| "loss": 7.8241, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.4823465507876154, | |
| "grad_norm": 0.5065767765045166, | |
| "learning_rate": 0.00010675466215383758, | |
| "loss": 7.8085, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 0.4830707948578671, | |
| "grad_norm": 0.4563339352607727, | |
| "learning_rate": 0.00010652602189016799, | |
| "loss": 7.8053, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 0.48379503892811876, | |
| "grad_norm": 0.4857904314994812, | |
| "learning_rate": 0.00010629734735972818, | |
| "loss": 7.8281, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 0.48451928299837044, | |
| "grad_norm": 0.4706745743751526, | |
| "learning_rate": 0.00010606863976324015, | |
| "loss": 7.8073, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 0.4852435270686221, | |
| "grad_norm": 0.49210935831069946, | |
| "learning_rate": 0.00010583990030159939, | |
| "loss": 7.8297, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.4859677711388738, | |
| "grad_norm": 0.4665246307849884, | |
| "learning_rate": 0.00010561113017586878, | |
| "loss": 7.8966, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 0.4866920152091255, | |
| "grad_norm": 0.47893619537353516, | |
| "learning_rate": 0.00010538233058727225, | |
| "loss": 7.8491, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 0.48741625927937715, | |
| "grad_norm": 0.4599825441837311, | |
| "learning_rate": 0.00010515350273718829, | |
| "loss": 7.8459, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 0.48814050334962883, | |
| "grad_norm": 0.47248682379722595, | |
| "learning_rate": 0.00010492464782714395, | |
| "loss": 7.8785, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 0.4888647474198805, | |
| "grad_norm": 0.46690088510513306, | |
| "learning_rate": 0.00010469576705880826, | |
| "loss": 7.8327, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.4895889914901322, | |
| "grad_norm": 0.48305100202560425, | |
| "learning_rate": 0.00010446686163398604, | |
| "loss": 7.8324, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 0.49031323556038386, | |
| "grad_norm": 0.47751349210739136, | |
| "learning_rate": 0.00010423793275461162, | |
| "loss": 7.8515, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 0.49103747963063554, | |
| "grad_norm": 0.46247628331184387, | |
| "learning_rate": 0.00010400898162274248, | |
| "loss": 7.8304, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 0.4917617237008872, | |
| "grad_norm": 0.4933788478374481, | |
| "learning_rate": 0.00010378000944055291, | |
| "loss": 7.7668, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.4924859677711389, | |
| "grad_norm": 0.4568544626235962, | |
| "learning_rate": 0.00010355101741032771, | |
| "loss": 7.8139, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.4932102118413905, | |
| "grad_norm": 0.5135279297828674, | |
| "learning_rate": 0.000103322006734456, | |
| "loss": 7.8382, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 0.4939344559116422, | |
| "grad_norm": 0.4543817639350891, | |
| "learning_rate": 0.00010309297861542468, | |
| "loss": 7.8125, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 0.4946586999818939, | |
| "grad_norm": 0.48243218660354614, | |
| "learning_rate": 0.00010286393425581231, | |
| "loss": 7.8269, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 0.49538294405214556, | |
| "grad_norm": 0.4711065888404846, | |
| "learning_rate": 0.00010263487485828271, | |
| "loss": 7.881, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 0.49610718812239724, | |
| "grad_norm": 0.48097798228263855, | |
| "learning_rate": 0.0001024058016255787, | |
| "loss": 7.8061, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.4968314321926489, | |
| "grad_norm": 0.4807778000831604, | |
| "learning_rate": 0.00010217671576051564, | |
| "loss": 7.8052, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 0.4975556762629006, | |
| "grad_norm": 0.48832613229751587, | |
| "learning_rate": 0.00010194761846597534, | |
| "loss": 7.814, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 0.49827992033315227, | |
| "grad_norm": 0.4787043035030365, | |
| "learning_rate": 0.00010171851094489957, | |
| "loss": 7.829, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 0.49900416440340395, | |
| "grad_norm": 0.4968035817146301, | |
| "learning_rate": 0.0001014893944002838, | |
| "loss": 7.7458, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 0.4997284084736556, | |
| "grad_norm": 0.458019495010376, | |
| "learning_rate": 0.00010126027003517086, | |
| "loss": 7.8039, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.5004526525439073, | |
| "grad_norm": 0.4970747232437134, | |
| "learning_rate": 0.0001010311390526447, | |
| "loss": 7.7751, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 0.501176896614159, | |
| "grad_norm": 0.45407259464263916, | |
| "learning_rate": 0.00010080200265582394, | |
| "loss": 7.836, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 0.501176896614159, | |
| "eval_loss": 7.830474376678467, | |
| "eval_runtime": 4.779, | |
| "eval_samples_per_second": 243.355, | |
| "eval_steps_per_second": 121.782, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 0.5019011406844106, | |
| "grad_norm": 0.5260797142982483, | |
| "learning_rate": 0.0001005728620478557, | |
| "loss": 7.7982, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 0.5026253847546623, | |
| "grad_norm": 0.4505165219306946, | |
| "learning_rate": 0.00010034371843190915, | |
| "loss": 7.8403, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 0.503349628824914, | |
| "grad_norm": 0.5103182196617126, | |
| "learning_rate": 0.00010011457301116933, | |
| "loss": 7.7234, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.5040738728951657, | |
| "grad_norm": 0.5652449727058411, | |
| "learning_rate": 9.988542698883068e-05, | |
| "loss": 7.8412, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 0.5047981169654173, | |
| "grad_norm": 0.5191313624382019, | |
| "learning_rate": 9.965628156809087e-05, | |
| "loss": 7.7998, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 0.505522361035669, | |
| "grad_norm": 0.4937920570373535, | |
| "learning_rate": 9.94271379521443e-05, | |
| "loss": 7.8072, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 0.5062466051059207, | |
| "grad_norm": 0.547224760055542, | |
| "learning_rate": 9.919799734417608e-05, | |
| "loss": 7.8387, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 0.5069708491761724, | |
| "grad_norm": 0.5999129414558411, | |
| "learning_rate": 9.896886094735535e-05, | |
| "loss": 7.8445, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.507695093246424, | |
| "grad_norm": 0.5010003447532654, | |
| "learning_rate": 9.873972996482916e-05, | |
| "loss": 7.8172, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 0.5084193373166758, | |
| "grad_norm": 0.48822104930877686, | |
| "learning_rate": 9.851060559971624e-05, | |
| "loss": 7.8536, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 0.5091435813869274, | |
| "grad_norm": 0.5012331604957581, | |
| "learning_rate": 9.828148905510044e-05, | |
| "loss": 7.8445, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 0.5098678254571791, | |
| "grad_norm": 0.4834780693054199, | |
| "learning_rate": 9.805238153402469e-05, | |
| "loss": 7.822, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 0.5105920695274307, | |
| "grad_norm": 0.4714999496936798, | |
| "learning_rate": 9.782328423948435e-05, | |
| "loss": 7.877, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.5113163135976824, | |
| "grad_norm": 0.46891549229621887, | |
| "learning_rate": 9.759419837442134e-05, | |
| "loss": 7.8555, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 0.5120405576679341, | |
| "grad_norm": 0.4694564938545227, | |
| "learning_rate": 9.736512514171732e-05, | |
| "loss": 7.8477, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 0.5127648017381857, | |
| "grad_norm": 0.48359888792037964, | |
| "learning_rate": 9.71360657441877e-05, | |
| "loss": 7.8765, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 0.5134890458084375, | |
| "grad_norm": 0.47419044375419617, | |
| "learning_rate": 9.690702138457534e-05, | |
| "loss": 7.8814, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 0.5142132898786891, | |
| "grad_norm": 0.47822949290275574, | |
| "learning_rate": 9.667799326554401e-05, | |
| "loss": 7.8609, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.5149375339489408, | |
| "grad_norm": 0.4817030429840088, | |
| "learning_rate": 9.644898258967232e-05, | |
| "loss": 7.8694, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 0.5156617780191924, | |
| "grad_norm": 0.4727485775947571, | |
| "learning_rate": 9.62199905594471e-05, | |
| "loss": 7.903, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 0.5163860220894442, | |
| "grad_norm": 0.5098878741264343, | |
| "learning_rate": 9.599101837725753e-05, | |
| "loss": 7.8711, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 0.5171102661596958, | |
| "grad_norm": 0.449669748544693, | |
| "learning_rate": 9.576206724538837e-05, | |
| "loss": 7.868, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 0.5178345102299475, | |
| "grad_norm": 0.4713840186595917, | |
| "learning_rate": 9.553313836601398e-05, | |
| "loss": 7.8244, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.5185587543001992, | |
| "grad_norm": 0.4921199381351471, | |
| "learning_rate": 9.53042329411918e-05, | |
| "loss": 7.8841, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.5192829983704509, | |
| "grad_norm": 0.47159916162490845, | |
| "learning_rate": 9.507535217285607e-05, | |
| "loss": 7.8828, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 0.5200072424407025, | |
| "grad_norm": 0.48348769545555115, | |
| "learning_rate": 9.484649726281173e-05, | |
| "loss": 7.8346, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 0.5207314865109542, | |
| "grad_norm": 0.48519793152809143, | |
| "learning_rate": 9.461766941272778e-05, | |
| "loss": 7.8751, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 0.5214557305812059, | |
| "grad_norm": 0.4665411412715912, | |
| "learning_rate": 9.438886982413124e-05, | |
| "loss": 7.9164, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.5221799746514575, | |
| "grad_norm": 0.4838716387748718, | |
| "learning_rate": 9.416009969840061e-05, | |
| "loss": 7.8461, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 0.5229042187217092, | |
| "grad_norm": 0.4719375669956207, | |
| "learning_rate": 9.393136023675988e-05, | |
| "loss": 7.8157, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 0.5236284627919608, | |
| "grad_norm": 0.5168206691741943, | |
| "learning_rate": 9.370265264027185e-05, | |
| "loss": 7.8118, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 0.5243527068622126, | |
| "grad_norm": 0.466214656829834, | |
| "learning_rate": 9.347397810983204e-05, | |
| "loss": 7.7755, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 0.5250769509324642, | |
| "grad_norm": 0.5822585821151733, | |
| "learning_rate": 9.324533784616247e-05, | |
| "loss": 7.85, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.5258011950027159, | |
| "grad_norm": 0.48217180371284485, | |
| "learning_rate": 9.301673304980504e-05, | |
| "loss": 7.8665, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 0.5265254390729676, | |
| "grad_norm": 0.5124320983886719, | |
| "learning_rate": 9.278816492111562e-05, | |
| "loss": 7.8112, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 0.5272496831432193, | |
| "grad_norm": 0.4656013548374176, | |
| "learning_rate": 9.255963466025735e-05, | |
| "loss": 7.8952, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 0.5279739272134709, | |
| "grad_norm": 0.4809459447860718, | |
| "learning_rate": 9.233114346719472e-05, | |
| "loss": 7.8659, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 0.5286981712837227, | |
| "grad_norm": 0.46756666898727417, | |
| "learning_rate": 9.21026925416869e-05, | |
| "loss": 7.8359, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.5294224153539743, | |
| "grad_norm": 0.503579318523407, | |
| "learning_rate": 9.187428308328176e-05, | |
| "loss": 7.8012, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 0.530146659424226, | |
| "grad_norm": 0.46779781579971313, | |
| "learning_rate": 9.164591629130941e-05, | |
| "loss": 7.8359, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 0.5308709034944776, | |
| "grad_norm": 0.4926343560218811, | |
| "learning_rate": 9.141759336487584e-05, | |
| "loss": 7.8221, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 0.5315951475647294, | |
| "grad_norm": 0.464432954788208, | |
| "learning_rate": 9.118931550285687e-05, | |
| "loss": 7.8713, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 0.532319391634981, | |
| "grad_norm": 0.49543848633766174, | |
| "learning_rate": 9.096108390389151e-05, | |
| "loss": 7.8084, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.5330436357052326, | |
| "grad_norm": 0.45875051617622375, | |
| "learning_rate": 9.073289976637603e-05, | |
| "loss": 7.8152, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 0.5337678797754843, | |
| "grad_norm": 0.4713042974472046, | |
| "learning_rate": 9.050476428845739e-05, | |
| "loss": 7.7912, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 0.534492123845736, | |
| "grad_norm": 0.4827256500720978, | |
| "learning_rate": 9.027667866802701e-05, | |
| "loss": 7.8026, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 0.5352163679159877, | |
| "grad_norm": 0.45872411131858826, | |
| "learning_rate": 9.004864410271462e-05, | |
| "loss": 7.8695, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 0.5359406119862393, | |
| "grad_norm": 0.4700596034526825, | |
| "learning_rate": 8.982066178988182e-05, | |
| "loss": 7.8606, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.5366648560564911, | |
| "grad_norm": 0.45632404088974, | |
| "learning_rate": 8.959273292661586e-05, | |
| "loss": 7.7807, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 0.5373891001267427, | |
| "grad_norm": 0.4901478886604309, | |
| "learning_rate": 8.936485870972328e-05, | |
| "loss": 7.8481, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 0.5381133441969944, | |
| "grad_norm": 0.4826202392578125, | |
| "learning_rate": 8.913704033572379e-05, | |
| "loss": 7.7886, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 0.538837588267246, | |
| "grad_norm": 0.5499749779701233, | |
| "learning_rate": 8.89092790008438e-05, | |
| "loss": 7.8111, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 0.5395618323374978, | |
| "grad_norm": 0.4294477105140686, | |
| "learning_rate": 8.868157590101019e-05, | |
| "loss": 7.8344, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.5402860764077494, | |
| "grad_norm": 0.4754398763179779, | |
| "learning_rate": 8.845393223184418e-05, | |
| "loss": 7.9187, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 0.5410103204780011, | |
| "grad_norm": 0.526314377784729, | |
| "learning_rate": 8.822634918865482e-05, | |
| "loss": 7.7336, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 0.5417345645482528, | |
| "grad_norm": 0.5151805281639099, | |
| "learning_rate": 8.799882796643288e-05, | |
| "loss": 7.7485, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 0.5424588086185045, | |
| "grad_norm": 0.56130450963974, | |
| "learning_rate": 8.777136975984448e-05, | |
| "loss": 7.8075, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 0.5431830526887561, | |
| "grad_norm": 0.6159570813179016, | |
| "learning_rate": 8.754397576322498e-05, | |
| "loss": 7.7995, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.5439072967590078, | |
| "grad_norm": 0.49635598063468933, | |
| "learning_rate": 8.731664717057236e-05, | |
| "loss": 7.8466, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 0.5446315408292595, | |
| "grad_norm": 0.49500927329063416, | |
| "learning_rate": 8.708938517554141e-05, | |
| "loss": 7.8124, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 0.5453557848995111, | |
| "grad_norm": 0.48374736309051514, | |
| "learning_rate": 8.68621909714371e-05, | |
| "loss": 7.8808, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 0.5460800289697628, | |
| "grad_norm": 0.4718135893344879, | |
| "learning_rate": 8.663506575120845e-05, | |
| "loss": 7.8721, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 0.5468042730400144, | |
| "grad_norm": 0.49202045798301697, | |
| "learning_rate": 8.640801070744233e-05, | |
| "loss": 7.8248, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.5475285171102662, | |
| "grad_norm": 0.4670674502849579, | |
| "learning_rate": 8.618102703235702e-05, | |
| "loss": 7.8563, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 0.5482527611805178, | |
| "grad_norm": 0.5087535381317139, | |
| "learning_rate": 8.59541159177962e-05, | |
| "loss": 7.8356, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 0.5489770052507695, | |
| "grad_norm": 0.4710357189178467, | |
| "learning_rate": 8.572727855522243e-05, | |
| "loss": 7.8835, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 0.5497012493210212, | |
| "grad_norm": 0.48206937313079834, | |
| "learning_rate": 8.550051613571108e-05, | |
| "loss": 7.8386, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 0.5504254933912729, | |
| "grad_norm": 0.4598138630390167, | |
| "learning_rate": 8.527382984994394e-05, | |
| "loss": 7.9155, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.5511497374615245, | |
| "grad_norm": 0.46072736382484436, | |
| "learning_rate": 8.504722088820309e-05, | |
| "loss": 7.9004, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 0.5518739815317762, | |
| "grad_norm": 0.47069671750068665, | |
| "learning_rate": 8.482069044036462e-05, | |
| "loss": 7.8558, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 0.5525982256020279, | |
| "grad_norm": 0.48600372672080994, | |
| "learning_rate": 8.459423969589228e-05, | |
| "loss": 7.8204, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 0.5533224696722796, | |
| "grad_norm": 0.456483393907547, | |
| "learning_rate": 8.436786984383146e-05, | |
| "loss": 7.8135, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 0.5540467137425312, | |
| "grad_norm": 0.4625873565673828, | |
| "learning_rate": 8.414158207280259e-05, | |
| "loss": 7.8563, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.554770957812783, | |
| "grad_norm": 0.47437793016433716, | |
| "learning_rate": 8.391537757099533e-05, | |
| "loss": 7.8649, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 0.5554952018830346, | |
| "grad_norm": 0.48973995447158813, | |
| "learning_rate": 8.368925752616194e-05, | |
| "loss": 7.8504, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 0.5562194459532862, | |
| "grad_norm": 0.49295151233673096, | |
| "learning_rate": 8.346322312561127e-05, | |
| "loss": 7.8525, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 0.5569436900235379, | |
| "grad_norm": 0.4774750769138336, | |
| "learning_rate": 8.323727555620256e-05, | |
| "loss": 7.8677, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 0.5576679340937896, | |
| "grad_norm": 0.4751247465610504, | |
| "learning_rate": 8.301141600433891e-05, | |
| "loss": 7.8658, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.5583921781640413, | |
| "grad_norm": 0.48602089285850525, | |
| "learning_rate": 8.278564565596152e-05, | |
| "loss": 7.8284, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 0.5591164222342929, | |
| "grad_norm": 0.4605376124382019, | |
| "learning_rate": 8.255996569654296e-05, | |
| "loss": 7.8761, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 0.5598406663045447, | |
| "grad_norm": 0.46493852138519287, | |
| "learning_rate": 8.233437731108139e-05, | |
| "loss": 7.8206, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 0.5605649103747963, | |
| "grad_norm": 0.4919988512992859, | |
| "learning_rate": 8.210888168409399e-05, | |
| "loss": 7.7783, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 0.561289154445048, | |
| "grad_norm": 0.4530481994152069, | |
| "learning_rate": 8.188347999961087e-05, | |
| "loss": 7.8679, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.5620133985152996, | |
| "grad_norm": 0.4982988238334656, | |
| "learning_rate": 8.165817344116906e-05, | |
| "loss": 7.7636, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 0.5627376425855514, | |
| "grad_norm": 0.49008551239967346, | |
| "learning_rate": 8.143296319180588e-05, | |
| "loss": 7.864, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 0.563461886655803, | |
| "grad_norm": 0.4614430367946625, | |
| "learning_rate": 8.120785043405309e-05, | |
| "loss": 7.8705, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 0.5641861307260547, | |
| "grad_norm": 0.46965253353118896, | |
| "learning_rate": 8.098283634993045e-05, | |
| "loss": 7.8487, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 0.5649103747963063, | |
| "grad_norm": 0.4675351083278656, | |
| "learning_rate": 8.07579221209397e-05, | |
| "loss": 7.8998, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.5656346188665581, | |
| "grad_norm": 0.5022285580635071, | |
| "learning_rate": 8.053310892805823e-05, | |
| "loss": 7.7928, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 0.5663588629368097, | |
| "grad_norm": 0.46838775277137756, | |
| "learning_rate": 8.03083979517328e-05, | |
| "loss": 7.8427, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 0.5670831070070613, | |
| "grad_norm": 0.501288652420044, | |
| "learning_rate": 8.008379037187366e-05, | |
| "loss": 7.7874, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 0.5678073510773131, | |
| "grad_norm": 0.49566730856895447, | |
| "learning_rate": 7.985928736784794e-05, | |
| "loss": 7.7951, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 0.5685315951475647, | |
| "grad_norm": 0.4716828167438507, | |
| "learning_rate": 7.963489011847385e-05, | |
| "loss": 7.7954, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.5692558392178164, | |
| "grad_norm": 0.47677019238471985, | |
| "learning_rate": 7.941059980201413e-05, | |
| "loss": 7.8447, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 0.569980083288068, | |
| "grad_norm": 0.4770268499851227, | |
| "learning_rate": 7.918641759617018e-05, | |
| "loss": 7.804, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 0.5707043273583198, | |
| "grad_norm": 0.48790669441223145, | |
| "learning_rate": 7.896234467807561e-05, | |
| "loss": 7.7361, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 0.5714285714285714, | |
| "grad_norm": 0.49754399061203003, | |
| "learning_rate": 7.873838222429035e-05, | |
| "loss": 7.7453, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 0.5721528154988231, | |
| "grad_norm": 0.4493372142314911, | |
| "learning_rate": 7.851453141079413e-05, | |
| "loss": 7.8176, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.5728770595690748, | |
| "grad_norm": 0.4799216091632843, | |
| "learning_rate": 7.829079341298051e-05, | |
| "loss": 7.814, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 0.5736013036393265, | |
| "grad_norm": 0.4880140721797943, | |
| "learning_rate": 7.806716940565084e-05, | |
| "loss": 7.7657, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 0.5743255477095781, | |
| "grad_norm": 0.46878695487976074, | |
| "learning_rate": 7.784366056300769e-05, | |
| "loss": 7.8544, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 0.5750497917798298, | |
| "grad_norm": 0.5069106817245483, | |
| "learning_rate": 7.762026805864915e-05, | |
| "loss": 7.8384, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 0.5757740358500815, | |
| "grad_norm": 0.521747350692749, | |
| "learning_rate": 7.739699306556228e-05, | |
| "loss": 7.7701, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.5764982799203332, | |
| "grad_norm": 0.5047403573989868, | |
| "learning_rate": 7.717383675611718e-05, | |
| "loss": 7.8321, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 0.5772225239905848, | |
| "grad_norm": 0.48073387145996094, | |
| "learning_rate": 7.695080030206076e-05, | |
| "loss": 7.7588, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 0.5779467680608364, | |
| "grad_norm": 0.4945087730884552, | |
| "learning_rate": 7.672788487451058e-05, | |
| "loss": 7.8004, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 0.5786710121310882, | |
| "grad_norm": 0.537676215171814, | |
| "learning_rate": 7.650509164394876e-05, | |
| "loss": 7.7014, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 0.5793952562013398, | |
| "grad_norm": 0.5863990783691406, | |
| "learning_rate": 7.628242178021572e-05, | |
| "loss": 7.9051, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.5801195002715915, | |
| "grad_norm": 0.5090913772583008, | |
| "learning_rate": 7.605987645250419e-05, | |
| "loss": 7.8515, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 0.5808437443418432, | |
| "grad_norm": 0.507625162601471, | |
| "learning_rate": 7.583745682935286e-05, | |
| "loss": 7.8064, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 0.5815679884120949, | |
| "grad_norm": 0.4949605166912079, | |
| "learning_rate": 7.561516407864055e-05, | |
| "loss": 7.7985, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 0.5822922324823465, | |
| "grad_norm": 0.48158496618270874, | |
| "learning_rate": 7.539299936757978e-05, | |
| "loss": 7.8514, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 0.5830164765525983, | |
| "grad_norm": 0.48431625962257385, | |
| "learning_rate": 7.517096386271072e-05, | |
| "loss": 7.8289, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.5837407206228499, | |
| "grad_norm": 0.5053116679191589, | |
| "learning_rate": 7.494905872989522e-05, | |
| "loss": 7.8756, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 0.5844649646931016, | |
| "grad_norm": 0.49544015526771545, | |
| "learning_rate": 7.472728513431051e-05, | |
| "loss": 7.8154, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 0.5851892087633532, | |
| "grad_norm": 0.4702587425708771, | |
| "learning_rate": 7.450564424044318e-05, | |
| "loss": 7.8741, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 0.585913452833605, | |
| "grad_norm": 0.47806039452552795, | |
| "learning_rate": 7.428413721208296e-05, | |
| "loss": 7.8546, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 0.5866376969038566, | |
| "grad_norm": 0.49512267112731934, | |
| "learning_rate": 7.40627652123168e-05, | |
| "loss": 7.8856, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.5873619409741083, | |
| "grad_norm": 0.5060679316520691, | |
| "learning_rate": 7.384152940352253e-05, | |
| "loss": 7.7994, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 0.5880861850443599, | |
| "grad_norm": 0.5014981627464294, | |
| "learning_rate": 7.362043094736287e-05, | |
| "loss": 7.8225, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 0.5888104291146117, | |
| "grad_norm": 0.5063635110855103, | |
| "learning_rate": 7.339947100477947e-05, | |
| "loss": 7.8719, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 0.5895346731848633, | |
| "grad_norm": 0.4955052137374878, | |
| "learning_rate": 7.317865073598648e-05, | |
| "loss": 7.9004, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 0.5902589172551149, | |
| "grad_norm": 0.4668295085430145, | |
| "learning_rate": 7.295797130046482e-05, | |
| "loss": 7.8558, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.5909831613253667, | |
| "grad_norm": 0.47807928919792175, | |
| "learning_rate": 7.273743385695582e-05, | |
| "loss": 7.8619, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 0.5917074053956183, | |
| "grad_norm": 0.48628920316696167, | |
| "learning_rate": 7.25170395634553e-05, | |
| "loss": 7.8668, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 0.59243164946587, | |
| "grad_norm": 0.4596603214740753, | |
| "learning_rate": 7.229678957720738e-05, | |
| "loss": 7.8634, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 0.5931558935361216, | |
| "grad_norm": 0.4702138900756836, | |
| "learning_rate": 7.207668505469844e-05, | |
| "loss": 7.8422, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 0.5938801376063734, | |
| "grad_norm": 0.4690391421318054, | |
| "learning_rate": 7.185672715165119e-05, | |
| "loss": 7.8206, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.594604381676625, | |
| "grad_norm": 0.5008434057235718, | |
| "learning_rate": 7.163691702301827e-05, | |
| "loss": 7.8105, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 0.5953286257468767, | |
| "grad_norm": 0.5092424750328064, | |
| "learning_rate": 7.14172558229766e-05, | |
| "loss": 7.8332, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 0.5960528698171284, | |
| "grad_norm": 0.4641984701156616, | |
| "learning_rate": 7.119774470492092e-05, | |
| "loss": 7.8529, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 0.5967771138873801, | |
| "grad_norm": 0.4499223828315735, | |
| "learning_rate": 7.097838482145802e-05, | |
| "loss": 7.8605, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 0.5975013579576317, | |
| "grad_norm": 0.47683218121528625, | |
| "learning_rate": 7.075917732440061e-05, | |
| "loss": 7.8361, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.5982256020278834, | |
| "grad_norm": 0.48220375180244446, | |
| "learning_rate": 7.054012336476121e-05, | |
| "loss": 7.8493, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 0.5989498460981351, | |
| "grad_norm": 0.5097104907035828, | |
| "learning_rate": 7.032122409274613e-05, | |
| "loss": 7.894, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 0.5996740901683868, | |
| "grad_norm": 0.44973528385162354, | |
| "learning_rate": 7.010248065774943e-05, | |
| "loss": 7.8853, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 0.6003983342386384, | |
| "grad_norm": 0.48968738317489624, | |
| "learning_rate": 6.988389420834703e-05, | |
| "loss": 7.8434, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 0.60112257830889, | |
| "grad_norm": 0.4859766960144043, | |
| "learning_rate": 6.966546589229037e-05, | |
| "loss": 7.8775, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.6018468223791418, | |
| "grad_norm": 0.5055286884307861, | |
| "learning_rate": 6.944719685650075e-05, | |
| "loss": 7.793, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 0.6025710664493934, | |
| "grad_norm": 0.4732415974140167, | |
| "learning_rate": 6.922908824706295e-05, | |
| "loss": 7.8474, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 0.6032953105196451, | |
| "grad_norm": 0.5133833885192871, | |
| "learning_rate": 6.90111412092195e-05, | |
| "loss": 7.7488, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 0.6040195545898968, | |
| "grad_norm": 0.47122323513031006, | |
| "learning_rate": 6.879335688736454e-05, | |
| "loss": 7.8389, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 0.6047437986601485, | |
| "grad_norm": 0.5096733570098877, | |
| "learning_rate": 6.857573642503776e-05, | |
| "loss": 7.791, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.6054680427304001, | |
| "grad_norm": 0.4919492304325104, | |
| "learning_rate": 6.835828096491854e-05, | |
| "loss": 7.7837, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 0.6061922868006518, | |
| "grad_norm": 0.48628994822502136, | |
| "learning_rate": 6.814099164881975e-05, | |
| "loss": 7.8275, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 0.6069165308709035, | |
| "grad_norm": 0.5138300061225891, | |
| "learning_rate": 6.792386961768204e-05, | |
| "loss": 7.8181, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 0.6076407749411552, | |
| "grad_norm": 0.4759021997451782, | |
| "learning_rate": 6.77069160115675e-05, | |
| "loss": 7.7909, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 0.6083650190114068, | |
| "grad_norm": 0.5031505227088928, | |
| "learning_rate": 6.749013196965406e-05, | |
| "loss": 7.8561, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.6090892630816586, | |
| "grad_norm": 0.45728060603141785, | |
| "learning_rate": 6.72735186302291e-05, | |
| "loss": 7.8218, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 0.6098135071519102, | |
| "grad_norm": 0.4515739381313324, | |
| "learning_rate": 6.705707713068376e-05, | |
| "loss": 7.8055, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 0.6105377512221619, | |
| "grad_norm": 0.507045567035675, | |
| "learning_rate": 6.684080860750697e-05, | |
| "loss": 7.8195, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 0.6112619952924135, | |
| "grad_norm": 0.5178161263465881, | |
| "learning_rate": 6.662471419627924e-05, | |
| "loss": 7.8287, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 0.6119862393626652, | |
| "grad_norm": 0.5024030804634094, | |
| "learning_rate": 6.640879503166698e-05, | |
| "loss": 7.7915, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.6127104834329169, | |
| "grad_norm": 0.5156453847885132, | |
| "learning_rate": 6.619305224741632e-05, | |
| "loss": 7.8114, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 0.6134347275031685, | |
| "grad_norm": 0.50360107421875, | |
| "learning_rate": 6.597748697634739e-05, | |
| "loss": 7.8968, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 0.6141589715734203, | |
| "grad_norm": 0.5216660499572754, | |
| "learning_rate": 6.57621003503481e-05, | |
| "loss": 7.8246, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 0.6148832156436719, | |
| "grad_norm": 0.48182037472724915, | |
| "learning_rate": 6.554689350036834e-05, | |
| "loss": 7.8687, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 0.6156074597139236, | |
| "grad_norm": 0.5355033874511719, | |
| "learning_rate": 6.533186755641416e-05, | |
| "loss": 7.8434, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.6163317037841752, | |
| "grad_norm": 0.49577125906944275, | |
| "learning_rate": 6.511702364754153e-05, | |
| "loss": 7.8567, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 0.617055947854427, | |
| "grad_norm": 0.4947705566883087, | |
| "learning_rate": 6.49023629018508e-05, | |
| "loss": 7.9015, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 0.6177801919246786, | |
| "grad_norm": 0.4933142364025116, | |
| "learning_rate": 6.468788644648039e-05, | |
| "loss": 7.8388, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 0.6185044359949303, | |
| "grad_norm": 0.48275384306907654, | |
| "learning_rate": 6.447359540760114e-05, | |
| "loss": 7.8168, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 0.619228680065182, | |
| "grad_norm": 0.4693118929862976, | |
| "learning_rate": 6.425949091041027e-05, | |
| "loss": 7.8784, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.6199529241354337, | |
| "grad_norm": 0.49441099166870117, | |
| "learning_rate": 6.404557407912549e-05, | |
| "loss": 7.8285, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 0.6206771682056853, | |
| "grad_norm": 0.4684365689754486, | |
| "learning_rate": 6.383184603697922e-05, | |
| "loss": 7.8689, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 0.621401412275937, | |
| "grad_norm": 0.4639764130115509, | |
| "learning_rate": 6.361830790621246e-05, | |
| "loss": 7.8302, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 0.6221256563461887, | |
| "grad_norm": 0.5014294385910034, | |
| "learning_rate": 6.340496080806914e-05, | |
| "loss": 7.8206, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 0.6228499004164404, | |
| "grad_norm": 0.4852553904056549, | |
| "learning_rate": 6.319180586278996e-05, | |
| "loss": 7.8383, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.623574144486692, | |
| "grad_norm": 0.4782876670360565, | |
| "learning_rate": 6.297884418960691e-05, | |
| "loss": 7.8477, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 0.6242983885569436, | |
| "grad_norm": 0.49755236506462097, | |
| "learning_rate": 6.276607690673688e-05, | |
| "loss": 7.8563, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 0.6250226326271954, | |
| "grad_norm": 0.485841304063797, | |
| "learning_rate": 6.255350513137626e-05, | |
| "loss": 7.8672, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 0.625746876697447, | |
| "grad_norm": 0.4920748770236969, | |
| "learning_rate": 6.234112997969475e-05, | |
| "loss": 7.8371, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 0.6264711207676987, | |
| "grad_norm": 0.4922964572906494, | |
| "learning_rate": 6.212895256682964e-05, | |
| "loss": 7.7465, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.6271953648379504, | |
| "grad_norm": 0.46496015787124634, | |
| "learning_rate": 6.191697400688001e-05, | |
| "loss": 7.8745, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 0.6279196089082021, | |
| "grad_norm": 0.4966541826725006, | |
| "learning_rate": 6.170519541290072e-05, | |
| "loss": 7.8253, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 0.6286438529784537, | |
| "grad_norm": 0.47178417444229126, | |
| "learning_rate": 6.149361789689674e-05, | |
| "loss": 7.8381, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 0.6293680970487054, | |
| "grad_norm": 0.4999862611293793, | |
| "learning_rate": 6.128224256981706e-05, | |
| "loss": 7.796, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 0.6300923411189571, | |
| "grad_norm": 0.5017781853675842, | |
| "learning_rate": 6.107107054154924e-05, | |
| "loss": 7.8559, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.6308165851892088, | |
| "grad_norm": 0.46709203720092773, | |
| "learning_rate": 6.0860102920913196e-05, | |
| "loss": 7.8881, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 0.6315408292594604, | |
| "grad_norm": 0.49039193987846375, | |
| "learning_rate": 6.064934081565557e-05, | |
| "loss": 7.7982, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 0.6322650733297122, | |
| "grad_norm": 0.4402787685394287, | |
| "learning_rate": 6.0438785332443946e-05, | |
| "loss": 7.8388, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 0.6329893173999638, | |
| "grad_norm": 0.48174357414245605, | |
| "learning_rate": 6.0228437576860874e-05, | |
| "loss": 7.8222, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 0.6337135614702155, | |
| "grad_norm": 0.4780054986476898, | |
| "learning_rate": 6.0018298653398345e-05, | |
| "loss": 7.8427, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.6344378055404671, | |
| "grad_norm": 0.48912590742111206, | |
| "learning_rate": 5.980836966545158e-05, | |
| "loss": 7.8059, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 0.6351620496107188, | |
| "grad_norm": 0.5173190832138062, | |
| "learning_rate": 5.9598651715313715e-05, | |
| "loss": 7.8253, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 0.6358862936809705, | |
| "grad_norm": 0.4782562255859375, | |
| "learning_rate": 5.9389145904169595e-05, | |
| "loss": 7.8532, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 0.6366105377512221, | |
| "grad_norm": 0.47767338156700134, | |
| "learning_rate": 5.917985333209022e-05, | |
| "loss": 7.8198, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 0.6373347818214739, | |
| "grad_norm": 0.4958306849002838, | |
| "learning_rate": 5.8970775098026973e-05, | |
| "loss": 7.8114, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.6380590258917255, | |
| "grad_norm": 0.49203237891197205, | |
| "learning_rate": 5.87619122998057e-05, | |
| "loss": 7.8273, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 0.6387832699619772, | |
| "grad_norm": 0.5006452798843384, | |
| "learning_rate": 5.8553266034121124e-05, | |
| "loss": 7.7925, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 0.6395075140322288, | |
| "grad_norm": 0.4803057312965393, | |
| "learning_rate": 5.834483739653089e-05, | |
| "loss": 7.8057, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 0.6402317581024806, | |
| "grad_norm": 0.4590706527233124, | |
| "learning_rate": 5.813662748145008e-05, | |
| "loss": 7.8248, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 0.6409560021727322, | |
| "grad_norm": 0.4630698263645172, | |
| "learning_rate": 5.7928637382145203e-05, | |
| "loss": 7.8183, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.6416802462429839, | |
| "grad_norm": 0.5254594683647156, | |
| "learning_rate": 5.772086819072853e-05, | |
| "loss": 7.7863, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 0.6424044903132355, | |
| "grad_norm": 0.49607181549072266, | |
| "learning_rate": 5.75133209981525e-05, | |
| "loss": 7.8268, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 0.6431287343834873, | |
| "grad_norm": 0.4501619338989258, | |
| "learning_rate": 5.73059968942038e-05, | |
| "loss": 7.8368, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 0.6438529784537389, | |
| "grad_norm": 0.49222031235694885, | |
| "learning_rate": 5.7098896967497775e-05, | |
| "loss": 7.7554, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 0.6445772225239906, | |
| "grad_norm": 0.45921745896339417, | |
| "learning_rate": 5.689202230547259e-05, | |
| "loss": 7.8255, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.6453014665942423, | |
| "grad_norm": 0.4936200976371765, | |
| "learning_rate": 5.668537399438374e-05, | |
| "loss": 7.8921, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 0.6460257106644939, | |
| "grad_norm": 0.4984248876571655, | |
| "learning_rate": 5.647895311929803e-05, | |
| "loss": 7.8222, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 0.6467499547347456, | |
| "grad_norm": 0.47879624366760254, | |
| "learning_rate": 5.627276076408807e-05, | |
| "loss": 7.9071, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 0.6474741988049972, | |
| "grad_norm": 0.49307674169540405, | |
| "learning_rate": 5.6066798011426737e-05, | |
| "loss": 7.7861, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 0.648198442875249, | |
| "grad_norm": 0.5343567728996277, | |
| "learning_rate": 5.586106594278109e-05, | |
| "loss": 7.8635, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.6489226869455006, | |
| "grad_norm": 0.5100454092025757, | |
| "learning_rate": 5.565556563840699e-05, | |
| "loss": 7.7338, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 0.6496469310157523, | |
| "grad_norm": 0.4881967306137085, | |
| "learning_rate": 5.5450298177343466e-05, | |
| "loss": 7.8221, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 0.650371175086004, | |
| "grad_norm": 0.5178599953651428, | |
| "learning_rate": 5.524526463740678e-05, | |
| "loss": 7.7642, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 0.6510954191562557, | |
| "grad_norm": 0.5256778001785278, | |
| "learning_rate": 5.5040466095184965e-05, | |
| "loss": 7.778, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 0.6518196632265073, | |
| "grad_norm": 0.5190709829330444, | |
| "learning_rate": 5.4835903626032195e-05, | |
| "loss": 7.8178, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.652543907296759, | |
| "grad_norm": 0.49377232789993286, | |
| "learning_rate": 5.4631578304063e-05, | |
| "loss": 7.8374, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 0.6532681513670107, | |
| "grad_norm": 0.4822869598865509, | |
| "learning_rate": 5.442749120214672e-05, | |
| "loss": 7.8543, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 0.6539923954372624, | |
| "grad_norm": 0.5066609978675842, | |
| "learning_rate": 5.4223643391901916e-05, | |
| "loss": 7.8336, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 0.654716639507514, | |
| "grad_norm": 0.49571654200553894, | |
| "learning_rate": 5.402003594369053e-05, | |
| "loss": 7.7989, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 0.6554408835777658, | |
| "grad_norm": 0.47712838649749756, | |
| "learning_rate": 5.3816669926612605e-05, | |
| "loss": 7.8491, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.6561651276480174, | |
| "grad_norm": 0.5059483647346497, | |
| "learning_rate": 5.361354640850029e-05, | |
| "loss": 7.8269, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 0.6568893717182691, | |
| "grad_norm": 0.4979402422904968, | |
| "learning_rate": 5.3410666455912604e-05, | |
| "loss": 7.7683, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 0.6576136157885207, | |
| "grad_norm": 0.4917674660682678, | |
| "learning_rate": 5.320803113412951e-05, | |
| "loss": 7.8254, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 0.6583378598587724, | |
| "grad_norm": 0.4933214783668518, | |
| "learning_rate": 5.3005641507146495e-05, | |
| "loss": 7.8163, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 0.6590621039290241, | |
| "grad_norm": 0.4836273193359375, | |
| "learning_rate": 5.280349863766906e-05, | |
| "loss": 7.8663, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.6597863479992757, | |
| "grad_norm": 0.45807477831840515, | |
| "learning_rate": 5.2601603587106863e-05, | |
| "loss": 7.8553, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 0.6605105920695274, | |
| "grad_norm": 0.4703587293624878, | |
| "learning_rate": 5.239995741556848e-05, | |
| "loss": 7.8577, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 0.6612348361397791, | |
| "grad_norm": 0.48380714654922485, | |
| "learning_rate": 5.2198561181855574e-05, | |
| "loss": 7.8219, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 0.6619590802100308, | |
| "grad_norm": 0.4754285216331482, | |
| "learning_rate": 5.199741594345744e-05, | |
| "loss": 7.8467, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 0.6626833242802824, | |
| "grad_norm": 0.4489299952983856, | |
| "learning_rate": 5.179652275654554e-05, | |
| "loss": 7.8807, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.6634075683505342, | |
| "grad_norm": 0.4758758842945099, | |
| "learning_rate": 5.1595882675967755e-05, | |
| "loss": 7.7741, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 0.6641318124207858, | |
| "grad_norm": 0.49364110827445984, | |
| "learning_rate": 5.139549675524301e-05, | |
| "loss": 7.8486, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 0.6648560564910375, | |
| "grad_norm": 0.48365652561187744, | |
| "learning_rate": 5.1195366046555656e-05, | |
| "loss": 7.8576, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 0.6655803005612891, | |
| "grad_norm": 0.4603043794631958, | |
| "learning_rate": 5.099549160075008e-05, | |
| "loss": 7.8672, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 0.6663045446315409, | |
| "grad_norm": 0.46025213599205017, | |
| "learning_rate": 5.079587446732493e-05, | |
| "loss": 7.8447, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.6670287887017925, | |
| "grad_norm": 0.47760942578315735, | |
| "learning_rate": 5.059651569442794e-05, | |
| "loss": 7.8622, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 0.6677530327720442, | |
| "grad_norm": 0.4995022416114807, | |
| "learning_rate": 5.039741632885009e-05, | |
| "loss": 7.7676, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 0.6684772768422959, | |
| "grad_norm": 0.48363152146339417, | |
| "learning_rate": 5.01985774160203e-05, | |
| "loss": 7.8692, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 0.6692015209125475, | |
| "grad_norm": 0.4707562029361725, | |
| "learning_rate": 5.000000000000002e-05, | |
| "loss": 7.8341, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 0.6699257649827992, | |
| "grad_norm": 0.46814507246017456, | |
| "learning_rate": 4.980168512347747e-05, | |
| "loss": 7.864, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.6706500090530508, | |
| "grad_norm": 0.4722250699996948, | |
| "learning_rate": 4.9603633827762484e-05, | |
| "loss": 7.8269, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 0.6713742531233026, | |
| "grad_norm": 0.45258694887161255, | |
| "learning_rate": 4.940584715278075e-05, | |
| "loss": 7.8147, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 0.6720984971935542, | |
| "grad_norm": 0.4681549370288849, | |
| "learning_rate": 4.9208326137068625e-05, | |
| "loss": 7.8421, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 0.6728227412638059, | |
| "grad_norm": 0.49207526445388794, | |
| "learning_rate": 4.901107181776743e-05, | |
| "loss": 7.7759, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 0.6735469853340575, | |
| "grad_norm": 0.4637949764728546, | |
| "learning_rate": 4.881408523061813e-05, | |
| "loss": 7.8506, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.6742712294043093, | |
| "grad_norm": 0.5171796679496765, | |
| "learning_rate": 4.861736740995601e-05, | |
| "loss": 7.8593, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 0.6749954734745609, | |
| "grad_norm": 0.47917360067367554, | |
| "learning_rate": 4.8420919388704925e-05, | |
| "loss": 7.8142, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 0.6757197175448126, | |
| "grad_norm": 0.4940485656261444, | |
| "learning_rate": 4.822474219837225e-05, | |
| "loss": 7.7892, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 0.6764439616150643, | |
| "grad_norm": 0.5015715956687927, | |
| "learning_rate": 4.802883686904318e-05, | |
| "loss": 7.9136, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 0.677168205685316, | |
| "grad_norm": 0.4832720160484314, | |
| "learning_rate": 4.7833204429375454e-05, | |
| "loss": 7.7575, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.6778924497555676, | |
| "grad_norm": 0.45720604062080383, | |
| "learning_rate": 4.763784590659387e-05, | |
| "loss": 7.8713, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 0.6786166938258194, | |
| "grad_norm": 0.47305944561958313, | |
| "learning_rate": 4.744276232648508e-05, | |
| "loss": 7.7939, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 0.679340937896071, | |
| "grad_norm": 0.4715365171432495, | |
| "learning_rate": 4.724795471339195e-05, | |
| "loss": 7.806, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 0.6800651819663226, | |
| "grad_norm": 0.468777060508728, | |
| "learning_rate": 4.7053424090208295e-05, | |
| "loss": 7.7986, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 0.6807894260365743, | |
| "grad_norm": 0.47867533564567566, | |
| "learning_rate": 4.685917147837364e-05, | |
| "loss": 7.8439, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.681513670106826, | |
| "grad_norm": 0.4867823123931885, | |
| "learning_rate": 4.666519789786756e-05, | |
| "loss": 7.8547, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 0.6822379141770777, | |
| "grad_norm": 0.48606693744659424, | |
| "learning_rate": 4.6471504367204674e-05, | |
| "loss": 7.836, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 0.6829621582473293, | |
| "grad_norm": 0.4838757812976837, | |
| "learning_rate": 4.6278091903428945e-05, | |
| "loss": 7.8131, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 0.683686402317581, | |
| "grad_norm": 0.46455007791519165, | |
| "learning_rate": 4.608496152210867e-05, | |
| "loss": 7.8649, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 0.6844106463878327, | |
| "grad_norm": 0.5098406076431274, | |
| "learning_rate": 4.589211423733087e-05, | |
| "loss": 7.8212, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.6851348904580844, | |
| "grad_norm": 0.48916125297546387, | |
| "learning_rate": 4.569955106169611e-05, | |
| "loss": 7.7915, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 0.685859134528336, | |
| "grad_norm": 0.525736927986145, | |
| "learning_rate": 4.5507273006313245e-05, | |
| "loss": 7.7649, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 0.6865833785985878, | |
| "grad_norm": 0.5023128986358643, | |
| "learning_rate": 4.531528108079387e-05, | |
| "loss": 7.7986, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 0.6873076226688394, | |
| "grad_norm": 0.5856012105941772, | |
| "learning_rate": 4.5123576293247364e-05, | |
| "loss": 7.7212, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 0.6880318667390911, | |
| "grad_norm": 0.5098285675048828, | |
| "learning_rate": 4.493215965027519e-05, | |
| "loss": 7.9312, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.6887561108093427, | |
| "grad_norm": 0.5016021728515625, | |
| "learning_rate": 4.4741032156966025e-05, | |
| "loss": 7.8514, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 0.6894803548795945, | |
| "grad_norm": 0.4811919033527374, | |
| "learning_rate": 4.455019481689016e-05, | |
| "loss": 7.8562, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 0.6902045989498461, | |
| "grad_norm": 0.45924901962280273, | |
| "learning_rate": 4.435964863209437e-05, | |
| "loss": 7.8562, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 0.6909288430200978, | |
| "grad_norm": 0.5059201717376709, | |
| "learning_rate": 4.416939460309667e-05, | |
| "loss": 7.8266, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 0.6916530870903494, | |
| "grad_norm": 0.47285687923431396, | |
| "learning_rate": 4.3979433728880936e-05, | |
| "loss": 7.8448, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.6923773311606011, | |
| "grad_norm": 0.47951385378837585, | |
| "learning_rate": 4.378976700689192e-05, | |
| "loss": 7.8623, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 0.6931015752308528, | |
| "grad_norm": 0.45039263367652893, | |
| "learning_rate": 4.360039543302965e-05, | |
| "loss": 7.876, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 0.6938258193011044, | |
| "grad_norm": 0.47055962681770325, | |
| "learning_rate": 4.3411320001644576e-05, | |
| "loss": 7.8539, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 0.6945500633713562, | |
| "grad_norm": 0.4704470634460449, | |
| "learning_rate": 4.322254170553201e-05, | |
| "loss": 7.8378, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 0.6952743074416078, | |
| "grad_norm": 0.47437623143196106, | |
| "learning_rate": 4.303406153592712e-05, | |
| "loss": 7.8915, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.6959985515118595, | |
| "grad_norm": 0.450575590133667, | |
| "learning_rate": 4.284588048249974e-05, | |
| "loss": 7.8715, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 0.6967227955821111, | |
| "grad_norm": 0.49726375937461853, | |
| "learning_rate": 4.2657999533349e-05, | |
| "loss": 7.8248, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 0.6974470396523629, | |
| "grad_norm": 0.5131885409355164, | |
| "learning_rate": 4.247041967499837e-05, | |
| "loss": 7.8141, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 0.6981712837226145, | |
| "grad_norm": 0.5051475167274475, | |
| "learning_rate": 4.228314189239021e-05, | |
| "loss": 7.7858, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 0.6988955277928662, | |
| "grad_norm": 0.49921953678131104, | |
| "learning_rate": 4.209616716888088e-05, | |
| "loss": 7.806, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.6996197718631179, | |
| "grad_norm": 0.4952443838119507, | |
| "learning_rate": 4.190949648623538e-05, | |
| "loss": 7.8111, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 0.7003440159333696, | |
| "grad_norm": 0.48354852199554443, | |
| "learning_rate": 4.172313082462218e-05, | |
| "loss": 7.8454, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 0.7010682600036212, | |
| "grad_norm": 0.4903525412082672, | |
| "learning_rate": 4.153707116260831e-05, | |
| "loss": 7.796, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 0.701792504073873, | |
| "grad_norm": 0.4461687505245209, | |
| "learning_rate": 4.135131847715391e-05, | |
| "loss": 7.8763, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 0.7025167481441246, | |
| "grad_norm": 0.4997590482234955, | |
| "learning_rate": 4.116587374360738e-05, | |
| "loss": 7.7374, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.7032409922143762, | |
| "grad_norm": 0.4803599715232849, | |
| "learning_rate": 4.0980737935700045e-05, | |
| "loss": 7.8267, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 0.7039652362846279, | |
| "grad_norm": 0.4723648130893707, | |
| "learning_rate": 4.0795912025541106e-05, | |
| "loss": 7.8528, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 0.7046894803548795, | |
| "grad_norm": 0.4950774312019348, | |
| "learning_rate": 4.061139698361259e-05, | |
| "loss": 7.846, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 0.7054137244251313, | |
| "grad_norm": 0.4853728115558624, | |
| "learning_rate": 4.0427193778764307e-05, | |
| "loss": 7.8703, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 0.7061379684953829, | |
| "grad_norm": 0.4416038393974304, | |
| "learning_rate": 4.024330337820853e-05, | |
| "loss": 7.8747, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.7068622125656346, | |
| "grad_norm": 0.4781404435634613, | |
| "learning_rate": 4.0059726747515104e-05, | |
| "loss": 7.8101, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 0.7075864566358863, | |
| "grad_norm": 0.4613541066646576, | |
| "learning_rate": 3.9876464850606435e-05, | |
| "loss": 7.8109, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 0.708310700706138, | |
| "grad_norm": 0.4910728633403778, | |
| "learning_rate": 3.969351864975216e-05, | |
| "loss": 7.8568, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 0.7090349447763896, | |
| "grad_norm": 0.45888233184814453, | |
| "learning_rate": 3.9510889105564454e-05, | |
| "loss": 7.7947, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 0.7097591888466414, | |
| "grad_norm": 0.4573422968387604, | |
| "learning_rate": 3.9328577176992595e-05, | |
| "loss": 7.8149, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.710483432916893, | |
| "grad_norm": 0.4781007170677185, | |
| "learning_rate": 3.914658382131832e-05, | |
| "loss": 7.8372, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 0.7112076769871447, | |
| "grad_norm": 0.4993189871311188, | |
| "learning_rate": 3.8964909994150456e-05, | |
| "loss": 7.8337, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 0.7119319210573963, | |
| "grad_norm": 0.46670612692832947, | |
| "learning_rate": 3.8783556649420085e-05, | |
| "loss": 7.7961, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 0.7126561651276481, | |
| "grad_norm": 0.48190969228744507, | |
| "learning_rate": 3.860252473937559e-05, | |
| "loss": 7.8113, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 0.7133804091978997, | |
| "grad_norm": 0.4879286587238312, | |
| "learning_rate": 3.8421815214577454e-05, | |
| "loss": 7.8245, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.7141046532681513, | |
| "grad_norm": 0.4496977627277374, | |
| "learning_rate": 3.8241429023893494e-05, | |
| "loss": 7.8067, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 0.714828897338403, | |
| "grad_norm": 0.47890475392341614, | |
| "learning_rate": 3.806136711449363e-05, | |
| "loss": 7.8695, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 0.7155531414086547, | |
| "grad_norm": 0.46888604760169983, | |
| "learning_rate": 3.7881630431845215e-05, | |
| "loss": 7.775, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 0.7162773854789064, | |
| "grad_norm": 0.458762526512146, | |
| "learning_rate": 3.770221991970777e-05, | |
| "loss": 7.8523, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 0.717001629549158, | |
| "grad_norm": 0.4727189242839813, | |
| "learning_rate": 3.752313652012817e-05, | |
| "loss": 7.7896, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.7177258736194098, | |
| "grad_norm": 0.4806780219078064, | |
| "learning_rate": 3.734438117343582e-05, | |
| "loss": 7.7772, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 0.7184501176896614, | |
| "grad_norm": 0.4579210877418518, | |
| "learning_rate": 3.7165954818237436e-05, | |
| "loss": 7.8326, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 0.7191743617599131, | |
| "grad_norm": 0.4929826259613037, | |
| "learning_rate": 3.698785839141236e-05, | |
| "loss": 7.8425, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 0.7198986058301647, | |
| "grad_norm": 0.5137585997581482, | |
| "learning_rate": 3.681009282810746e-05, | |
| "loss": 7.765, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 0.7206228499004165, | |
| "grad_norm": 0.4695113003253937, | |
| "learning_rate": 3.663265906173245e-05, | |
| "loss": 7.8759, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.7213470939706681, | |
| "grad_norm": 0.5344631671905518, | |
| "learning_rate": 3.645555802395476e-05, | |
| "loss": 7.7362, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 0.7220713380409198, | |
| "grad_norm": 0.5240874886512756, | |
| "learning_rate": 3.6278790644694674e-05, | |
| "loss": 7.8663, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 0.7227955821111715, | |
| "grad_norm": 0.5019177198410034, | |
| "learning_rate": 3.610235785212064e-05, | |
| "loss": 7.6963, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 0.7235198261814232, | |
| "grad_norm": 0.5077367424964905, | |
| "learning_rate": 3.592626057264413e-05, | |
| "loss": 7.8491, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 0.7242440702516748, | |
| "grad_norm": 0.5551170706748962, | |
| "learning_rate": 3.575049973091506e-05, | |
| "loss": 7.8329, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.7249683143219265, | |
| "grad_norm": 0.49724796414375305, | |
| "learning_rate": 3.5575076249816584e-05, | |
| "loss": 7.8, | |
| "step": 1001 | |
| }, | |
| { | |
| "epoch": 0.7256925583921782, | |
| "grad_norm": 0.4674682319164276, | |
| "learning_rate": 3.5399991050460655e-05, | |
| "loss": 7.8876, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 0.7264168024624298, | |
| "grad_norm": 0.49887341260910034, | |
| "learning_rate": 3.522524505218281e-05, | |
| "loss": 7.8503, | |
| "step": 1003 | |
| }, | |
| { | |
| "epoch": 0.7271410465326815, | |
| "grad_norm": 0.47441011667251587, | |
| "learning_rate": 3.505083917253763e-05, | |
| "loss": 7.8196, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 0.7278652906029331, | |
| "grad_norm": 0.4713391065597534, | |
| "learning_rate": 3.4876774327293734e-05, | |
| "loss": 7.8007, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 0.7285895346731849, | |
| "grad_norm": 0.4885476231575012, | |
| "learning_rate": 3.4703051430429024e-05, | |
| "loss": 7.8527, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 0.7293137787434365, | |
| "grad_norm": 0.4708382785320282, | |
| "learning_rate": 3.452967139412602e-05, | |
| "loss": 7.8204, | |
| "step": 1007 | |
| }, | |
| { | |
| "epoch": 0.7300380228136882, | |
| "grad_norm": 0.45463642477989197, | |
| "learning_rate": 3.435663512876679e-05, | |
| "loss": 7.8397, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 0.7307622668839399, | |
| "grad_norm": 0.46222880482673645, | |
| "learning_rate": 3.4183943542928496e-05, | |
| "loss": 7.8621, | |
| "step": 1009 | |
| }, | |
| { | |
| "epoch": 0.7314865109541916, | |
| "grad_norm": 0.4764266610145569, | |
| "learning_rate": 3.401159754337836e-05, | |
| "loss": 7.8469, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.7322107550244432, | |
| "grad_norm": 0.4880512058734894, | |
| "learning_rate": 3.383959803506901e-05, | |
| "loss": 7.8698, | |
| "step": 1011 | |
| }, | |
| { | |
| "epoch": 0.732934999094695, | |
| "grad_norm": 0.4619225859642029, | |
| "learning_rate": 3.3667945921133734e-05, | |
| "loss": 7.8439, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 0.7336592431649466, | |
| "grad_norm": 0.4897652566432953, | |
| "learning_rate": 3.34966421028817e-05, | |
| "loss": 7.7954, | |
| "step": 1013 | |
| }, | |
| { | |
| "epoch": 0.7343834872351983, | |
| "grad_norm": 0.47381868958473206, | |
| "learning_rate": 3.332568747979335e-05, | |
| "loss": 7.8305, | |
| "step": 1014 | |
| }, | |
| { | |
| "epoch": 0.7351077313054499, | |
| "grad_norm": 0.4567718207836151, | |
| "learning_rate": 3.3155082949515424e-05, | |
| "loss": 7.8574, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 0.7358319753757017, | |
| "grad_norm": 0.4839675724506378, | |
| "learning_rate": 3.298482940785655e-05, | |
| "loss": 7.8414, | |
| "step": 1016 | |
| }, | |
| { | |
| "epoch": 0.7365562194459533, | |
| "grad_norm": 0.4948679804801941, | |
| "learning_rate": 3.2814927748782245e-05, | |
| "loss": 7.8058, | |
| "step": 1017 | |
| }, | |
| { | |
| "epoch": 0.7372804635162049, | |
| "grad_norm": 0.47992077469825745, | |
| "learning_rate": 3.26453788644105e-05, | |
| "loss": 7.7794, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 0.7380047075864566, | |
| "grad_norm": 0.4672059714794159, | |
| "learning_rate": 3.2476183645006854e-05, | |
| "loss": 7.8264, | |
| "step": 1019 | |
| }, | |
| { | |
| "epoch": 0.7387289516567083, | |
| "grad_norm": 0.4785839319229126, | |
| "learning_rate": 3.2307342978979847e-05, | |
| "loss": 7.8405, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.73945319572696, | |
| "grad_norm": 0.48339641094207764, | |
| "learning_rate": 3.2138857752876406e-05, | |
| "loss": 7.8538, | |
| "step": 1021 | |
| }, | |
| { | |
| "epoch": 0.7401774397972116, | |
| "grad_norm": 0.4682196378707886, | |
| "learning_rate": 3.1970728851377005e-05, | |
| "loss": 7.8156, | |
| "step": 1022 | |
| }, | |
| { | |
| "epoch": 0.7409016838674634, | |
| "grad_norm": 0.4627414643764496, | |
| "learning_rate": 3.1802957157291256e-05, | |
| "loss": 7.8184, | |
| "step": 1023 | |
| }, | |
| { | |
| "epoch": 0.741625927937715, | |
| "grad_norm": 0.4607759714126587, | |
| "learning_rate": 3.1635543551553015e-05, | |
| "loss": 7.8599, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 0.7423501720079667, | |
| "grad_norm": 0.4652021825313568, | |
| "learning_rate": 3.146848891321604e-05, | |
| "loss": 7.7955, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.7430744160782183, | |
| "grad_norm": 0.4668468236923218, | |
| "learning_rate": 3.130179411944909e-05, | |
| "loss": 7.8534, | |
| "step": 1026 | |
| }, | |
| { | |
| "epoch": 0.7437986601484701, | |
| "grad_norm": 0.46713390946388245, | |
| "learning_rate": 3.113546004553151e-05, | |
| "loss": 7.8634, | |
| "step": 1027 | |
| }, | |
| { | |
| "epoch": 0.7445229042187217, | |
| "grad_norm": 0.4680757224559784, | |
| "learning_rate": 3.096948756484863e-05, | |
| "loss": 7.8308, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 0.7452471482889734, | |
| "grad_norm": 0.48597460985183716, | |
| "learning_rate": 3.080387754888706e-05, | |
| "loss": 7.7773, | |
| "step": 1029 | |
| }, | |
| { | |
| "epoch": 0.745971392359225, | |
| "grad_norm": 0.4744811952114105, | |
| "learning_rate": 3.06386308672302e-05, | |
| "loss": 7.7997, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.7466956364294768, | |
| "grad_norm": 0.4475364685058594, | |
| "learning_rate": 3.0473748387553647e-05, | |
| "loss": 7.8358, | |
| "step": 1031 | |
| }, | |
| { | |
| "epoch": 0.7474198804997284, | |
| "grad_norm": 0.4579427242279053, | |
| "learning_rate": 3.0309230975620717e-05, | |
| "loss": 7.8671, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 0.74814412456998, | |
| "grad_norm": 0.4434613883495331, | |
| "learning_rate": 3.0145079495277772e-05, | |
| "loss": 7.8496, | |
| "step": 1033 | |
| }, | |
| { | |
| "epoch": 0.7488683686402318, | |
| "grad_norm": 0.5059404373168945, | |
| "learning_rate": 2.9981294808449713e-05, | |
| "loss": 7.7336, | |
| "step": 1034 | |
| }, | |
| { | |
| "epoch": 0.7495926127104834, | |
| "grad_norm": 0.4605514407157898, | |
| "learning_rate": 2.9817877775135593e-05, | |
| "loss": 7.8329, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 0.7503168567807351, | |
| "grad_norm": 0.48954278230667114, | |
| "learning_rate": 2.965482925340386e-05, | |
| "loss": 7.8067, | |
| "step": 1036 | |
| }, | |
| { | |
| "epoch": 0.7510411008509867, | |
| "grad_norm": 0.4908781945705414, | |
| "learning_rate": 2.9492150099388127e-05, | |
| "loss": 7.9183, | |
| "step": 1037 | |
| }, | |
| { | |
| "epoch": 0.7517653449212385, | |
| "grad_norm": 0.46575653553009033, | |
| "learning_rate": 2.932984116728239e-05, | |
| "loss": 7.7655, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 0.7517653449212385, | |
| "eval_loss": 7.828281402587891, | |
| "eval_runtime": 4.8059, | |
| "eval_samples_per_second": 241.993, | |
| "eval_steps_per_second": 121.101, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 0.7524895889914901, | |
| "grad_norm": 0.46100151538848877, | |
| "learning_rate": 2.916790330933683e-05, | |
| "loss": 7.8197, | |
| "step": 1039 | |
| }, | |
| { | |
| "epoch": 0.7532138330617418, | |
| "grad_norm": 0.47532182931900024, | |
| "learning_rate": 2.9006337375853064e-05, | |
| "loss": 7.7012, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.7539380771319935, | |
| "grad_norm": 0.486884742975235, | |
| "learning_rate": 2.884514421517993e-05, | |
| "loss": 7.8531, | |
| "step": 1041 | |
| }, | |
| { | |
| "epoch": 0.7546623212022452, | |
| "grad_norm": 0.4965648055076599, | |
| "learning_rate": 2.868432467370882e-05, | |
| "loss": 7.8061, | |
| "step": 1042 | |
| }, | |
| { | |
| "epoch": 0.7553865652724968, | |
| "grad_norm": 0.49399492144584656, | |
| "learning_rate": 2.852387959586934e-05, | |
| "loss": 7.7895, | |
| "step": 1043 | |
| }, | |
| { | |
| "epoch": 0.7561108093427485, | |
| "grad_norm": 0.48604780435562134, | |
| "learning_rate": 2.8363809824124964e-05, | |
| "loss": 7.8522, | |
| "step": 1044 | |
| }, | |
| { | |
| "epoch": 0.7568350534130002, | |
| "grad_norm": 0.46038612723350525, | |
| "learning_rate": 2.820411619896838e-05, | |
| "loss": 7.9004, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 0.7575592974832519, | |
| "grad_norm": 0.5330672860145569, | |
| "learning_rate": 2.8044799558917313e-05, | |
| "loss": 7.85, | |
| "step": 1046 | |
| }, | |
| { | |
| "epoch": 0.7582835415535035, | |
| "grad_norm": 0.5077647566795349, | |
| "learning_rate": 2.7885860740509963e-05, | |
| "loss": 7.845, | |
| "step": 1047 | |
| }, | |
| { | |
| "epoch": 0.7590077856237553, | |
| "grad_norm": 0.480844646692276, | |
| "learning_rate": 2.7727300578300674e-05, | |
| "loss": 7.7689, | |
| "step": 1048 | |
| }, | |
| { | |
| "epoch": 0.7597320296940069, | |
| "grad_norm": 0.49301832914352417, | |
| "learning_rate": 2.756911990485552e-05, | |
| "loss": 7.8533, | |
| "step": 1049 | |
| }, | |
| { | |
| "epoch": 0.7604562737642585, | |
| "grad_norm": 0.598520040512085, | |
| "learning_rate": 2.741131955074807e-05, | |
| "loss": 7.7991, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.7611805178345102, | |
| "grad_norm": 0.4859587252140045, | |
| "learning_rate": 2.7253900344554795e-05, | |
| "loss": 7.8235, | |
| "step": 1051 | |
| }, | |
| { | |
| "epoch": 0.7619047619047619, | |
| "grad_norm": 0.49456787109375, | |
| "learning_rate": 2.7096863112850847e-05, | |
| "loss": 7.84, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 0.7626290059750136, | |
| "grad_norm": 0.4734579622745514, | |
| "learning_rate": 2.6940208680205802e-05, | |
| "loss": 7.8954, | |
| "step": 1053 | |
| }, | |
| { | |
| "epoch": 0.7633532500452652, | |
| "grad_norm": 0.4647412896156311, | |
| "learning_rate": 2.6783937869179143e-05, | |
| "loss": 7.8155, | |
| "step": 1054 | |
| }, | |
| { | |
| "epoch": 0.764077494115517, | |
| "grad_norm": 0.4898563623428345, | |
| "learning_rate": 2.662805150031612e-05, | |
| "loss": 7.8175, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 0.7648017381857686, | |
| "grad_norm": 0.5130847692489624, | |
| "learning_rate": 2.647255039214328e-05, | |
| "loss": 7.8289, | |
| "step": 1056 | |
| }, | |
| { | |
| "epoch": 0.7655259822560203, | |
| "grad_norm": 0.4688289761543274, | |
| "learning_rate": 2.6317435361164256e-05, | |
| "loss": 7.837, | |
| "step": 1057 | |
| }, | |
| { | |
| "epoch": 0.7662502263262719, | |
| "grad_norm": 0.4437507688999176, | |
| "learning_rate": 2.6162707221855552e-05, | |
| "loss": 7.8955, | |
| "step": 1058 | |
| }, | |
| { | |
| "epoch": 0.7669744703965237, | |
| "grad_norm": 0.44463828206062317, | |
| "learning_rate": 2.6008366786662073e-05, | |
| "loss": 7.8781, | |
| "step": 1059 | |
| }, | |
| { | |
| "epoch": 0.7676987144667753, | |
| "grad_norm": 0.44665560126304626, | |
| "learning_rate": 2.585441486599308e-05, | |
| "loss": 7.8517, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.768422958537027, | |
| "grad_norm": 0.45422571897506714, | |
| "learning_rate": 2.570085226821771e-05, | |
| "loss": 7.8811, | |
| "step": 1061 | |
| }, | |
| { | |
| "epoch": 0.7691472026072786, | |
| "grad_norm": 0.46825090050697327, | |
| "learning_rate": 2.554767979966097e-05, | |
| "loss": 7.8527, | |
| "step": 1062 | |
| }, | |
| { | |
| "epoch": 0.7698714466775304, | |
| "grad_norm": 0.4797806441783905, | |
| "learning_rate": 2.5394898264599297e-05, | |
| "loss": 7.8624, | |
| "step": 1063 | |
| }, | |
| { | |
| "epoch": 0.770595690747782, | |
| "grad_norm": 0.48127833008766174, | |
| "learning_rate": 2.5242508465256397e-05, | |
| "loss": 7.8368, | |
| "step": 1064 | |
| }, | |
| { | |
| "epoch": 0.7713199348180336, | |
| "grad_norm": 0.480258584022522, | |
| "learning_rate": 2.5090511201799172e-05, | |
| "loss": 7.8536, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 0.7720441788882854, | |
| "grad_norm": 0.495200514793396, | |
| "learning_rate": 2.493890727233329e-05, | |
| "loss": 7.8352, | |
| "step": 1066 | |
| }, | |
| { | |
| "epoch": 0.772768422958537, | |
| "grad_norm": 0.4558762013912201, | |
| "learning_rate": 2.478769747289912e-05, | |
| "loss": 7.8246, | |
| "step": 1067 | |
| }, | |
| { | |
| "epoch": 0.7734926670287887, | |
| "grad_norm": 0.46966373920440674, | |
| "learning_rate": 2.4636882597467593e-05, | |
| "loss": 7.8572, | |
| "step": 1068 | |
| }, | |
| { | |
| "epoch": 0.7742169110990403, | |
| "grad_norm": 0.471635103225708, | |
| "learning_rate": 2.4486463437935934e-05, | |
| "loss": 7.8459, | |
| "step": 1069 | |
| }, | |
| { | |
| "epoch": 0.7749411551692921, | |
| "grad_norm": 0.47857850790023804, | |
| "learning_rate": 2.433644078412355e-05, | |
| "loss": 7.8279, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.7756653992395437, | |
| "grad_norm": 0.48017922043800354, | |
| "learning_rate": 2.418681542376785e-05, | |
| "loss": 7.8697, | |
| "step": 1071 | |
| }, | |
| { | |
| "epoch": 0.7763896433097954, | |
| "grad_norm": 0.4738040864467621, | |
| "learning_rate": 2.403758814252024e-05, | |
| "loss": 7.8073, | |
| "step": 1072 | |
| }, | |
| { | |
| "epoch": 0.777113887380047, | |
| "grad_norm": 0.46145448088645935, | |
| "learning_rate": 2.3888759723941766e-05, | |
| "loss": 7.8748, | |
| "step": 1073 | |
| }, | |
| { | |
| "epoch": 0.7778381314502988, | |
| "grad_norm": 0.4822898805141449, | |
| "learning_rate": 2.3740330949499257e-05, | |
| "loss": 7.8332, | |
| "step": 1074 | |
| }, | |
| { | |
| "epoch": 0.7785623755205504, | |
| "grad_norm": 0.47280868887901306, | |
| "learning_rate": 2.359230259856097e-05, | |
| "loss": 7.8453, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.7792866195908021, | |
| "grad_norm": 0.47516578435897827, | |
| "learning_rate": 2.3444675448392728e-05, | |
| "loss": 7.8826, | |
| "step": 1076 | |
| }, | |
| { | |
| "epoch": 0.7800108636610538, | |
| "grad_norm": 0.4623286724090576, | |
| "learning_rate": 2.3297450274153643e-05, | |
| "loss": 7.872, | |
| "step": 1077 | |
| }, | |
| { | |
| "epoch": 0.7807351077313055, | |
| "grad_norm": 0.5090032815933228, | |
| "learning_rate": 2.3150627848892248e-05, | |
| "loss": 7.8377, | |
| "step": 1078 | |
| }, | |
| { | |
| "epoch": 0.7814593518015571, | |
| "grad_norm": 0.4724178612232208, | |
| "learning_rate": 2.3004208943542215e-05, | |
| "loss": 7.8654, | |
| "step": 1079 | |
| }, | |
| { | |
| "epoch": 0.7821835958718087, | |
| "grad_norm": 0.48950740694999695, | |
| "learning_rate": 2.2858194326918435e-05, | |
| "loss": 7.882, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.7829078399420605, | |
| "grad_norm": 0.4657925069332123, | |
| "learning_rate": 2.2712584765713064e-05, | |
| "loss": 7.8123, | |
| "step": 1081 | |
| }, | |
| { | |
| "epoch": 0.7836320840123121, | |
| "grad_norm": 0.5018364191055298, | |
| "learning_rate": 2.256738102449124e-05, | |
| "loss": 7.7877, | |
| "step": 1082 | |
| }, | |
| { | |
| "epoch": 0.7843563280825638, | |
| "grad_norm": 0.4764616787433624, | |
| "learning_rate": 2.2422583865687375e-05, | |
| "loss": 7.8885, | |
| "step": 1083 | |
| }, | |
| { | |
| "epoch": 0.7850805721528155, | |
| "grad_norm": 0.496661514043808, | |
| "learning_rate": 2.227819404960092e-05, | |
| "loss": 7.8584, | |
| "step": 1084 | |
| }, | |
| { | |
| "epoch": 0.7858048162230672, | |
| "grad_norm": 0.48577845096588135, | |
| "learning_rate": 2.2134212334392434e-05, | |
| "loss": 7.7898, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 0.7865290602933188, | |
| "grad_norm": 0.4589325189590454, | |
| "learning_rate": 2.1990639476079712e-05, | |
| "loss": 7.8669, | |
| "step": 1086 | |
| }, | |
| { | |
| "epoch": 0.7872533043635705, | |
| "grad_norm": 0.5179975032806396, | |
| "learning_rate": 2.1847476228533648e-05, | |
| "loss": 7.7761, | |
| "step": 1087 | |
| }, | |
| { | |
| "epoch": 0.7879775484338222, | |
| "grad_norm": 0.45577552914619446, | |
| "learning_rate": 2.1704723343474396e-05, | |
| "loss": 7.8018, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 0.7887017925040739, | |
| "grad_norm": 0.4571017622947693, | |
| "learning_rate": 2.156238157046734e-05, | |
| "loss": 7.8382, | |
| "step": 1089 | |
| }, | |
| { | |
| "epoch": 0.7894260365743255, | |
| "grad_norm": 0.49493101239204407, | |
| "learning_rate": 2.1420451656919284e-05, | |
| "loss": 7.7954, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.7901502806445773, | |
| "grad_norm": 0.4955536723136902, | |
| "learning_rate": 2.1278934348074332e-05, | |
| "loss": 7.9138, | |
| "step": 1091 | |
| }, | |
| { | |
| "epoch": 0.7908745247148289, | |
| "grad_norm": 0.5027803182601929, | |
| "learning_rate": 2.1137830387010206e-05, | |
| "loss": 7.7502, | |
| "step": 1092 | |
| }, | |
| { | |
| "epoch": 0.7915987687850806, | |
| "grad_norm": 0.47927114367485046, | |
| "learning_rate": 2.099714051463415e-05, | |
| "loss": 7.8775, | |
| "step": 1093 | |
| }, | |
| { | |
| "epoch": 0.7923230128553322, | |
| "grad_norm": 0.47723886370658875, | |
| "learning_rate": 2.085686546967909e-05, | |
| "loss": 7.7893, | |
| "step": 1094 | |
| }, | |
| { | |
| "epoch": 0.793047256925584, | |
| "grad_norm": 0.5063787698745728, | |
| "learning_rate": 2.0717005988699887e-05, | |
| "loss": 7.8202, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 0.7937715009958356, | |
| "grad_norm": 0.4967922270298004, | |
| "learning_rate": 2.0577562806069238e-05, | |
| "loss": 7.8676, | |
| "step": 1096 | |
| }, | |
| { | |
| "epoch": 0.7944957450660872, | |
| "grad_norm": 0.4945312738418579, | |
| "learning_rate": 2.0438536653974073e-05, | |
| "loss": 7.7993, | |
| "step": 1097 | |
| }, | |
| { | |
| "epoch": 0.795219989136339, | |
| "grad_norm": 0.47871074080467224, | |
| "learning_rate": 2.029992826241145e-05, | |
| "loss": 7.8309, | |
| "step": 1098 | |
| }, | |
| { | |
| "epoch": 0.7959442332065906, | |
| "grad_norm": 0.5307055115699768, | |
| "learning_rate": 2.016173835918498e-05, | |
| "loss": 7.7908, | |
| "step": 1099 | |
| }, | |
| { | |
| "epoch": 0.7966684772768423, | |
| "grad_norm": 0.5632224678993225, | |
| "learning_rate": 2.0023967669900777e-05, | |
| "loss": 7.7806, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.7973927213470939, | |
| "grad_norm": 0.4810802638530731, | |
| "learning_rate": 1.9886616917963785e-05, | |
| "loss": 7.8422, | |
| "step": 1101 | |
| }, | |
| { | |
| "epoch": 0.7981169654173457, | |
| "grad_norm": 0.49098509550094604, | |
| "learning_rate": 1.9749686824573987e-05, | |
| "loss": 7.8427, | |
| "step": 1102 | |
| }, | |
| { | |
| "epoch": 0.7988412094875973, | |
| "grad_norm": 0.4845903217792511, | |
| "learning_rate": 1.9613178108722507e-05, | |
| "loss": 7.8179, | |
| "step": 1103 | |
| }, | |
| { | |
| "epoch": 0.799565453557849, | |
| "grad_norm": 0.4725109040737152, | |
| "learning_rate": 1.9477091487187983e-05, | |
| "loss": 7.8455, | |
| "step": 1104 | |
| }, | |
| { | |
| "epoch": 0.8002896976281006, | |
| "grad_norm": 0.49514099955558777, | |
| "learning_rate": 1.9341427674532643e-05, | |
| "loss": 7.8411, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 0.8010139416983524, | |
| "grad_norm": 0.4852292239665985, | |
| "learning_rate": 1.9206187383098694e-05, | |
| "loss": 7.85, | |
| "step": 1106 | |
| }, | |
| { | |
| "epoch": 0.801738185768604, | |
| "grad_norm": 0.505436897277832, | |
| "learning_rate": 1.9071371323004494e-05, | |
| "loss": 7.8205, | |
| "step": 1107 | |
| }, | |
| { | |
| "epoch": 0.8024624298388557, | |
| "grad_norm": 0.4951551556587219, | |
| "learning_rate": 1.893698020214082e-05, | |
| "loss": 7.8409, | |
| "step": 1108 | |
| }, | |
| { | |
| "epoch": 0.8031866739091074, | |
| "grad_norm": 0.49560704827308655, | |
| "learning_rate": 1.8803014726167266e-05, | |
| "loss": 7.8177, | |
| "step": 1109 | |
| }, | |
| { | |
| "epoch": 0.8039109179793591, | |
| "grad_norm": 0.4628080725669861, | |
| "learning_rate": 1.866947559850839e-05, | |
| "loss": 7.8229, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.8046351620496107, | |
| "grad_norm": 0.47996366024017334, | |
| "learning_rate": 1.853636352035012e-05, | |
| "loss": 7.8377, | |
| "step": 1111 | |
| }, | |
| { | |
| "epoch": 0.8053594061198623, | |
| "grad_norm": 0.48397380113601685, | |
| "learning_rate": 1.840367919063598e-05, | |
| "loss": 7.8254, | |
| "step": 1112 | |
| }, | |
| { | |
| "epoch": 0.8060836501901141, | |
| "grad_norm": 0.47569262981414795, | |
| "learning_rate": 1.8271423306063564e-05, | |
| "loss": 7.8227, | |
| "step": 1113 | |
| }, | |
| { | |
| "epoch": 0.8068078942603657, | |
| "grad_norm": 0.47170665860176086, | |
| "learning_rate": 1.8139596561080696e-05, | |
| "loss": 7.8185, | |
| "step": 1114 | |
| }, | |
| { | |
| "epoch": 0.8075321383306174, | |
| "grad_norm": 0.48356300592422485, | |
| "learning_rate": 1.800819964788196e-05, | |
| "loss": 7.847, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 0.8082563824008691, | |
| "grad_norm": 0.4541751444339752, | |
| "learning_rate": 1.7877233256404956e-05, | |
| "loss": 7.861, | |
| "step": 1116 | |
| }, | |
| { | |
| "epoch": 0.8089806264711208, | |
| "grad_norm": 0.43556642532348633, | |
| "learning_rate": 1.7746698074326638e-05, | |
| "loss": 7.8939, | |
| "step": 1117 | |
| }, | |
| { | |
| "epoch": 0.8097048705413724, | |
| "grad_norm": 0.47422105073928833, | |
| "learning_rate": 1.761659478705989e-05, | |
| "loss": 7.8543, | |
| "step": 1118 | |
| }, | |
| { | |
| "epoch": 0.8104291146116241, | |
| "grad_norm": 0.501738965511322, | |
| "learning_rate": 1.7486924077749712e-05, | |
| "loss": 7.8054, | |
| "step": 1119 | |
| }, | |
| { | |
| "epoch": 0.8111533586818758, | |
| "grad_norm": 0.48210737109184265, | |
| "learning_rate": 1.7357686627269788e-05, | |
| "loss": 7.889, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.8118776027521275, | |
| "grad_norm": 0.5010181069374084, | |
| "learning_rate": 1.7228883114218775e-05, | |
| "loss": 7.8183, | |
| "step": 1121 | |
| }, | |
| { | |
| "epoch": 0.8126018468223791, | |
| "grad_norm": 0.5056195855140686, | |
| "learning_rate": 1.710051421491694e-05, | |
| "loss": 7.8112, | |
| "step": 1122 | |
| }, | |
| { | |
| "epoch": 0.8133260908926309, | |
| "grad_norm": 0.49855145812034607, | |
| "learning_rate": 1.6972580603402364e-05, | |
| "loss": 7.788, | |
| "step": 1123 | |
| }, | |
| { | |
| "epoch": 0.8140503349628825, | |
| "grad_norm": 0.479667603969574, | |
| "learning_rate": 1.6845082951427572e-05, | |
| "loss": 7.8281, | |
| "step": 1124 | |
| }, | |
| { | |
| "epoch": 0.8147745790331342, | |
| "grad_norm": 0.4779108464717865, | |
| "learning_rate": 1.671802192845594e-05, | |
| "loss": 7.8274, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.8154988231033858, | |
| "grad_norm": 0.4873594641685486, | |
| "learning_rate": 1.659139820165825e-05, | |
| "loss": 7.8795, | |
| "step": 1126 | |
| }, | |
| { | |
| "epoch": 0.8162230671736375, | |
| "grad_norm": 0.47458213567733765, | |
| "learning_rate": 1.6465212435909073e-05, | |
| "loss": 7.8462, | |
| "step": 1127 | |
| }, | |
| { | |
| "epoch": 0.8169473112438892, | |
| "grad_norm": 0.4490852355957031, | |
| "learning_rate": 1.6339465293783328e-05, | |
| "loss": 7.8258, | |
| "step": 1128 | |
| }, | |
| { | |
| "epoch": 0.8176715553141408, | |
| "grad_norm": 0.4825502038002014, | |
| "learning_rate": 1.6214157435552914e-05, | |
| "loss": 7.8331, | |
| "step": 1129 | |
| }, | |
| { | |
| "epoch": 0.8183957993843926, | |
| "grad_norm": 0.4452913701534271, | |
| "learning_rate": 1.6089289519183036e-05, | |
| "loss": 7.8999, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.8191200434546442, | |
| "grad_norm": 0.4826939105987549, | |
| "learning_rate": 1.596486220032888e-05, | |
| "loss": 7.8164, | |
| "step": 1131 | |
| }, | |
| { | |
| "epoch": 0.8198442875248959, | |
| "grad_norm": 0.46260425448417664, | |
| "learning_rate": 1.5840876132332195e-05, | |
| "loss": 7.8404, | |
| "step": 1132 | |
| }, | |
| { | |
| "epoch": 0.8205685315951475, | |
| "grad_norm": 0.476392537355423, | |
| "learning_rate": 1.571733196621774e-05, | |
| "loss": 7.8025, | |
| "step": 1133 | |
| }, | |
| { | |
| "epoch": 0.8212927756653993, | |
| "grad_norm": 0.43944284319877625, | |
| "learning_rate": 1.5594230350690044e-05, | |
| "loss": 7.8283, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 0.8220170197356509, | |
| "grad_norm": 0.48247215151786804, | |
| "learning_rate": 1.547157193212977e-05, | |
| "loss": 7.7797, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 0.8227412638059026, | |
| "grad_norm": 0.4725157618522644, | |
| "learning_rate": 1.5349357354590555e-05, | |
| "loss": 7.844, | |
| "step": 1136 | |
| }, | |
| { | |
| "epoch": 0.8234655078761542, | |
| "grad_norm": 0.46587786078453064, | |
| "learning_rate": 1.5227587259795462e-05, | |
| "loss": 7.8471, | |
| "step": 1137 | |
| }, | |
| { | |
| "epoch": 0.824189751946406, | |
| "grad_norm": 0.4861607253551483, | |
| "learning_rate": 1.5106262287133643e-05, | |
| "loss": 7.745, | |
| "step": 1138 | |
| }, | |
| { | |
| "epoch": 0.8249139960166576, | |
| "grad_norm": 0.4683452248573303, | |
| "learning_rate": 1.4985383073657112e-05, | |
| "loss": 7.7614, | |
| "step": 1139 | |
| }, | |
| { | |
| "epoch": 0.8256382400869093, | |
| "grad_norm": 0.47723594307899475, | |
| "learning_rate": 1.4864950254077181e-05, | |
| "loss": 7.8326, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.826362484157161, | |
| "grad_norm": 0.49085745215415955, | |
| "learning_rate": 1.4744964460761312e-05, | |
| "loss": 7.7305, | |
| "step": 1141 | |
| }, | |
| { | |
| "epoch": 0.8270867282274127, | |
| "grad_norm": 0.5058842301368713, | |
| "learning_rate": 1.4625426323729708e-05, | |
| "loss": 7.7709, | |
| "step": 1142 | |
| }, | |
| { | |
| "epoch": 0.8278109722976643, | |
| "grad_norm": 0.47824400663375854, | |
| "learning_rate": 1.4506336470651982e-05, | |
| "loss": 7.7977, | |
| "step": 1143 | |
| }, | |
| { | |
| "epoch": 0.8285352163679159, | |
| "grad_norm": 0.5179119110107422, | |
| "learning_rate": 1.438769552684398e-05, | |
| "loss": 7.7603, | |
| "step": 1144 | |
| }, | |
| { | |
| "epoch": 0.8292594604381677, | |
| "grad_norm": 0.4722190201282501, | |
| "learning_rate": 1.4269504115264376e-05, | |
| "loss": 7.7913, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 0.8299837045084193, | |
| "grad_norm": 0.47829577326774597, | |
| "learning_rate": 1.4151762856511419e-05, | |
| "loss": 7.8386, | |
| "step": 1146 | |
| }, | |
| { | |
| "epoch": 0.830707948578671, | |
| "grad_norm": 0.47289183735847473, | |
| "learning_rate": 1.4034472368819718e-05, | |
| "loss": 7.8422, | |
| "step": 1147 | |
| }, | |
| { | |
| "epoch": 0.8314321926489227, | |
| "grad_norm": 0.502065122127533, | |
| "learning_rate": 1.391763326805704e-05, | |
| "loss": 7.8705, | |
| "step": 1148 | |
| }, | |
| { | |
| "epoch": 0.8321564367191744, | |
| "grad_norm": 0.529735267162323, | |
| "learning_rate": 1.3801246167720904e-05, | |
| "loss": 7.7626, | |
| "step": 1149 | |
| }, | |
| { | |
| "epoch": 0.832880680789426, | |
| "grad_norm": 0.5681172609329224, | |
| "learning_rate": 1.3685311678935575e-05, | |
| "loss": 7.8454, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.8336049248596777, | |
| "grad_norm": 0.4957883954048157, | |
| "learning_rate": 1.3569830410448658e-05, | |
| "loss": 7.8634, | |
| "step": 1151 | |
| }, | |
| { | |
| "epoch": 0.8343291689299294, | |
| "grad_norm": 0.4802315831184387, | |
| "learning_rate": 1.34548029686281e-05, | |
| "loss": 7.8578, | |
| "step": 1152 | |
| }, | |
| { | |
| "epoch": 0.8350534130001811, | |
| "grad_norm": 0.49144551157951355, | |
| "learning_rate": 1.3340229957458783e-05, | |
| "loss": 7.802, | |
| "step": 1153 | |
| }, | |
| { | |
| "epoch": 0.8357776570704327, | |
| "grad_norm": 0.457791805267334, | |
| "learning_rate": 1.3226111978539524e-05, | |
| "loss": 7.8177, | |
| "step": 1154 | |
| }, | |
| { | |
| "epoch": 0.8365019011406845, | |
| "grad_norm": 0.4937816858291626, | |
| "learning_rate": 1.3112449631079892e-05, | |
| "loss": 7.8694, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 0.8372261452109361, | |
| "grad_norm": 0.4611557126045227, | |
| "learning_rate": 1.2999243511896974e-05, | |
| "loss": 7.8976, | |
| "step": 1156 | |
| }, | |
| { | |
| "epoch": 0.8379503892811878, | |
| "grad_norm": 0.4825960099697113, | |
| "learning_rate": 1.2886494215412348e-05, | |
| "loss": 7.8651, | |
| "step": 1157 | |
| }, | |
| { | |
| "epoch": 0.8386746333514394, | |
| "grad_norm": 0.4975496530532837, | |
| "learning_rate": 1.2774202333648844e-05, | |
| "loss": 7.8353, | |
| "step": 1158 | |
| }, | |
| { | |
| "epoch": 0.8393988774216911, | |
| "grad_norm": 0.46706926822662354, | |
| "learning_rate": 1.2662368456227602e-05, | |
| "loss": 7.8567, | |
| "step": 1159 | |
| }, | |
| { | |
| "epoch": 0.8401231214919428, | |
| "grad_norm": 0.46040868759155273, | |
| "learning_rate": 1.25509931703648e-05, | |
| "loss": 7.8983, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.8408473655621944, | |
| "grad_norm": 0.4722301661968231, | |
| "learning_rate": 1.2440077060868638e-05, | |
| "loss": 7.8863, | |
| "step": 1161 | |
| }, | |
| { | |
| "epoch": 0.8415716096324461, | |
| "grad_norm": 0.47437435388565063, | |
| "learning_rate": 1.2329620710136358e-05, | |
| "loss": 7.8298, | |
| "step": 1162 | |
| }, | |
| { | |
| "epoch": 0.8422958537026978, | |
| "grad_norm": 0.4701695144176483, | |
| "learning_rate": 1.2219624698151033e-05, | |
| "loss": 7.8605, | |
| "step": 1163 | |
| }, | |
| { | |
| "epoch": 0.8430200977729495, | |
| "grad_norm": 0.4844394326210022, | |
| "learning_rate": 1.2110089602478624e-05, | |
| "loss": 7.8583, | |
| "step": 1164 | |
| }, | |
| { | |
| "epoch": 0.8437443418432011, | |
| "grad_norm": 0.4796792268753052, | |
| "learning_rate": 1.2001015998264886e-05, | |
| "loss": 7.8471, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 0.8444685859134529, | |
| "grad_norm": 0.4662293493747711, | |
| "learning_rate": 1.1892404458232454e-05, | |
| "loss": 7.9045, | |
| "step": 1166 | |
| }, | |
| { | |
| "epoch": 0.8451928299837045, | |
| "grad_norm": 0.49967148900032043, | |
| "learning_rate": 1.178425555267768e-05, | |
| "loss": 7.847, | |
| "step": 1167 | |
| }, | |
| { | |
| "epoch": 0.8459170740539562, | |
| "grad_norm": 0.45847171545028687, | |
| "learning_rate": 1.1676569849467733e-05, | |
| "loss": 7.8271, | |
| "step": 1168 | |
| }, | |
| { | |
| "epoch": 0.8466413181242078, | |
| "grad_norm": 0.4743190407752991, | |
| "learning_rate": 1.1569347914037664e-05, | |
| "loss": 7.8559, | |
| "step": 1169 | |
| }, | |
| { | |
| "epoch": 0.8473655621944596, | |
| "grad_norm": 0.48577260971069336, | |
| "learning_rate": 1.1462590309387289e-05, | |
| "loss": 7.8242, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.8480898062647112, | |
| "grad_norm": 0.4353601932525635, | |
| "learning_rate": 1.1356297596078425e-05, | |
| "loss": 7.8549, | |
| "step": 1171 | |
| }, | |
| { | |
| "epoch": 0.8488140503349629, | |
| "grad_norm": 0.46052125096321106, | |
| "learning_rate": 1.125047033223171e-05, | |
| "loss": 7.8489, | |
| "step": 1172 | |
| }, | |
| { | |
| "epoch": 0.8495382944052146, | |
| "grad_norm": 0.4780481159687042, | |
| "learning_rate": 1.1145109073523951e-05, | |
| "loss": 7.8461, | |
| "step": 1173 | |
| }, | |
| { | |
| "epoch": 0.8502625384754662, | |
| "grad_norm": 0.48932063579559326, | |
| "learning_rate": 1.1040214373184954e-05, | |
| "loss": 7.8926, | |
| "step": 1174 | |
| }, | |
| { | |
| "epoch": 0.8509867825457179, | |
| "grad_norm": 0.5182289481163025, | |
| "learning_rate": 1.0935786781994738e-05, | |
| "loss": 7.8103, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.8517110266159695, | |
| "grad_norm": 0.4565328359603882, | |
| "learning_rate": 1.0831826848280679e-05, | |
| "loss": 7.853, | |
| "step": 1176 | |
| }, | |
| { | |
| "epoch": 0.8524352706862213, | |
| "grad_norm": 0.4468229115009308, | |
| "learning_rate": 1.0728335117914534e-05, | |
| "loss": 7.8157, | |
| "step": 1177 | |
| }, | |
| { | |
| "epoch": 0.8531595147564729, | |
| "grad_norm": 0.4953078329563141, | |
| "learning_rate": 1.0625312134309662e-05, | |
| "loss": 7.8499, | |
| "step": 1178 | |
| }, | |
| { | |
| "epoch": 0.8538837588267246, | |
| "grad_norm": 0.4719950258731842, | |
| "learning_rate": 1.052275843841809e-05, | |
| "loss": 7.8703, | |
| "step": 1179 | |
| }, | |
| { | |
| "epoch": 0.8546080028969762, | |
| "grad_norm": 0.506100594997406, | |
| "learning_rate": 1.042067456872775e-05, | |
| "loss": 7.7924, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.855332246967228, | |
| "grad_norm": 0.46659165620803833, | |
| "learning_rate": 1.0319061061259606e-05, | |
| "loss": 7.8599, | |
| "step": 1181 | |
| }, | |
| { | |
| "epoch": 0.8560564910374796, | |
| "grad_norm": 0.4609448313713074, | |
| "learning_rate": 1.0217918449564812e-05, | |
| "loss": 7.8043, | |
| "step": 1182 | |
| }, | |
| { | |
| "epoch": 0.8567807351077313, | |
| "grad_norm": 0.4527445435523987, | |
| "learning_rate": 1.0117247264722008e-05, | |
| "loss": 7.8937, | |
| "step": 1183 | |
| }, | |
| { | |
| "epoch": 0.857504979177983, | |
| "grad_norm": 0.478190153837204, | |
| "learning_rate": 1.0017048035334408e-05, | |
| "loss": 7.8079, | |
| "step": 1184 | |
| }, | |
| { | |
| "epoch": 0.8582292232482347, | |
| "grad_norm": 0.46092936396598816, | |
| "learning_rate": 9.917321287527148e-06, | |
| "loss": 7.8575, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 0.8589534673184863, | |
| "grad_norm": 0.4953816533088684, | |
| "learning_rate": 9.81806754494441e-06, | |
| "loss": 7.8281, | |
| "step": 1186 | |
| }, | |
| { | |
| "epoch": 0.859677711388738, | |
| "grad_norm": 0.4743238687515259, | |
| "learning_rate": 9.719287328746773e-06, | |
| "loss": 7.8017, | |
| "step": 1187 | |
| }, | |
| { | |
| "epoch": 0.8604019554589897, | |
| "grad_norm": 0.48036321997642517, | |
| "learning_rate": 9.620981157608389e-06, | |
| "loss": 7.8096, | |
| "step": 1188 | |
| }, | |
| { | |
| "epoch": 0.8611261995292414, | |
| "grad_norm": 0.4978514611721039, | |
| "learning_rate": 9.523149547714327e-06, | |
| "loss": 7.7873, | |
| "step": 1189 | |
| }, | |
| { | |
| "epoch": 0.861850443599493, | |
| "grad_norm": 0.5149461627006531, | |
| "learning_rate": 9.425793012757812e-06, | |
| "loss": 7.8773, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.8625746876697447, | |
| "grad_norm": 0.4775876998901367, | |
| "learning_rate": 9.328912063937544e-06, | |
| "loss": 7.8557, | |
| "step": 1191 | |
| }, | |
| { | |
| "epoch": 0.8632989317399964, | |
| "grad_norm": 0.4656752943992615, | |
| "learning_rate": 9.232507209955077e-06, | |
| "loss": 7.8102, | |
| "step": 1192 | |
| }, | |
| { | |
| "epoch": 0.864023175810248, | |
| "grad_norm": 0.49488770961761475, | |
| "learning_rate": 9.136578957011998e-06, | |
| "loss": 7.8236, | |
| "step": 1193 | |
| }, | |
| { | |
| "epoch": 0.8647474198804997, | |
| "grad_norm": 0.4837650656700134, | |
| "learning_rate": 9.04112780880747e-06, | |
| "loss": 7.7681, | |
| "step": 1194 | |
| }, | |
| { | |
| "epoch": 0.8654716639507514, | |
| "grad_norm": 0.4603443443775177, | |
| "learning_rate": 8.946154266535366e-06, | |
| "loss": 7.8176, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 0.8661959080210031, | |
| "grad_norm": 0.4513426423072815, | |
| "learning_rate": 8.851658828881858e-06, | |
| "loss": 7.8717, | |
| "step": 1196 | |
| }, | |
| { | |
| "epoch": 0.8669201520912547, | |
| "grad_norm": 0.48408597707748413, | |
| "learning_rate": 8.757641992022614e-06, | |
| "loss": 7.758, | |
| "step": 1197 | |
| }, | |
| { | |
| "epoch": 0.8676443961615065, | |
| "grad_norm": 0.5038301348686218, | |
| "learning_rate": 8.664104249620298e-06, | |
| "loss": 7.8486, | |
| "step": 1198 | |
| }, | |
| { | |
| "epoch": 0.8683686402317581, | |
| "grad_norm": 0.5601458549499512, | |
| "learning_rate": 8.571046092821955e-06, | |
| "loss": 7.713, | |
| "step": 1199 | |
| }, | |
| { | |
| "epoch": 0.8690928843020098, | |
| "grad_norm": 0.5590780377388, | |
| "learning_rate": 8.478468010256425e-06, | |
| "loss": 7.7364, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.8698171283722614, | |
| "grad_norm": 0.4802273213863373, | |
| "learning_rate": 8.386370488031759e-06, | |
| "loss": 7.8386, | |
| "step": 1201 | |
| }, | |
| { | |
| "epoch": 0.8705413724425132, | |
| "grad_norm": 0.5128775238990784, | |
| "learning_rate": 8.294754009732696e-06, | |
| "loss": 7.7979, | |
| "step": 1202 | |
| }, | |
| { | |
| "epoch": 0.8712656165127648, | |
| "grad_norm": 0.49584537744522095, | |
| "learning_rate": 8.203619056418155e-06, | |
| "loss": 7.7934, | |
| "step": 1203 | |
| }, | |
| { | |
| "epoch": 0.8719898605830165, | |
| "grad_norm": 0.47992539405822754, | |
| "learning_rate": 8.112966106618602e-06, | |
| "loss": 7.8282, | |
| "step": 1204 | |
| }, | |
| { | |
| "epoch": 0.8727141046532682, | |
| "grad_norm": 0.4876195788383484, | |
| "learning_rate": 8.022795636333636e-06, | |
| "loss": 7.8194, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 0.8734383487235198, | |
| "grad_norm": 0.49370482563972473, | |
| "learning_rate": 7.933108119029475e-06, | |
| "loss": 7.8294, | |
| "step": 1206 | |
| }, | |
| { | |
| "epoch": 0.8741625927937715, | |
| "grad_norm": 0.47466379404067993, | |
| "learning_rate": 7.843904025636417e-06, | |
| "loss": 7.8737, | |
| "step": 1207 | |
| }, | |
| { | |
| "epoch": 0.8748868368640231, | |
| "grad_norm": 0.5090891122817993, | |
| "learning_rate": 7.75518382454643e-06, | |
| "loss": 7.8391, | |
| "step": 1208 | |
| }, | |
| { | |
| "epoch": 0.8756110809342749, | |
| "grad_norm": 0.5014375448226929, | |
| "learning_rate": 7.66694798161064e-06, | |
| "loss": 7.8252, | |
| "step": 1209 | |
| }, | |
| { | |
| "epoch": 0.8763353250045265, | |
| "grad_norm": 0.46806249022483826, | |
| "learning_rate": 7.579196960136959e-06, | |
| "loss": 7.8172, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.8770595690747782, | |
| "grad_norm": 0.4725843071937561, | |
| "learning_rate": 7.4919312208875385e-06, | |
| "loss": 7.8868, | |
| "step": 1211 | |
| }, | |
| { | |
| "epoch": 0.8777838131450298, | |
| "grad_norm": 0.5052827000617981, | |
| "learning_rate": 7.405151222076434e-06, | |
| "loss": 7.8121, | |
| "step": 1212 | |
| }, | |
| { | |
| "epoch": 0.8785080572152816, | |
| "grad_norm": 0.4352813959121704, | |
| "learning_rate": 7.318857419367242e-06, | |
| "loss": 7.8594, | |
| "step": 1213 | |
| }, | |
| { | |
| "epoch": 0.8792323012855332, | |
| "grad_norm": 0.4714949429035187, | |
| "learning_rate": 7.233050265870534e-06, | |
| "loss": 7.8642, | |
| "step": 1214 | |
| }, | |
| { | |
| "epoch": 0.8799565453557849, | |
| "grad_norm": 0.48388224840164185, | |
| "learning_rate": 7.147730212141701e-06, | |
| "loss": 7.8853, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 0.8806807894260366, | |
| "grad_norm": 0.48352184891700745, | |
| "learning_rate": 7.062897706178384e-06, | |
| "loss": 7.8965, | |
| "step": 1216 | |
| }, | |
| { | |
| "epoch": 0.8814050334962883, | |
| "grad_norm": 0.46383845806121826, | |
| "learning_rate": 6.9785531934182915e-06, | |
| "loss": 7.8627, | |
| "step": 1217 | |
| }, | |
| { | |
| "epoch": 0.8821292775665399, | |
| "grad_norm": 0.48393815755844116, | |
| "learning_rate": 6.894697116736715e-06, | |
| "loss": 7.874, | |
| "step": 1218 | |
| }, | |
| { | |
| "epoch": 0.8828535216367916, | |
| "grad_norm": 0.4774603843688965, | |
| "learning_rate": 6.8113299164443205e-06, | |
| "loss": 7.8833, | |
| "step": 1219 | |
| }, | |
| { | |
| "epoch": 0.8835777657070433, | |
| "grad_norm": 0.47935643792152405, | |
| "learning_rate": 6.728452030284738e-06, | |
| "loss": 7.8931, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.8843020097772949, | |
| "grad_norm": 0.45838284492492676, | |
| "learning_rate": 6.646063893432364e-06, | |
| "loss": 7.8718, | |
| "step": 1221 | |
| }, | |
| { | |
| "epoch": 0.8850262538475466, | |
| "grad_norm": 0.4871087074279785, | |
| "learning_rate": 6.564165938489996e-06, | |
| "loss": 7.8648, | |
| "step": 1222 | |
| }, | |
| { | |
| "epoch": 0.8857504979177983, | |
| "grad_norm": 0.4909784495830536, | |
| "learning_rate": 6.482758595486571e-06, | |
| "loss": 7.793, | |
| "step": 1223 | |
| }, | |
| { | |
| "epoch": 0.88647474198805, | |
| "grad_norm": 0.4485389292240143, | |
| "learning_rate": 6.401842291874982e-06, | |
| "loss": 7.8831, | |
| "step": 1224 | |
| }, | |
| { | |
| "epoch": 0.8871989860583016, | |
| "grad_norm": 0.479890376329422, | |
| "learning_rate": 6.32141745252971e-06, | |
| "loss": 7.8632, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.8879232301285533, | |
| "grad_norm": 0.456013023853302, | |
| "learning_rate": 6.241484499744732e-06, | |
| "loss": 7.8373, | |
| "step": 1226 | |
| }, | |
| { | |
| "epoch": 0.888647474198805, | |
| "grad_norm": 0.4660279452800751, | |
| "learning_rate": 6.16204385323117e-06, | |
| "loss": 7.8827, | |
| "step": 1227 | |
| }, | |
| { | |
| "epoch": 0.8893717182690567, | |
| "grad_norm": 0.4719277620315552, | |
| "learning_rate": 6.083095930115157e-06, | |
| "loss": 7.7943, | |
| "step": 1228 | |
| }, | |
| { | |
| "epoch": 0.8900959623393083, | |
| "grad_norm": 0.5206422209739685, | |
| "learning_rate": 6.004641144935696e-06, | |
| "loss": 7.7849, | |
| "step": 1229 | |
| }, | |
| { | |
| "epoch": 0.8908202064095601, | |
| "grad_norm": 0.4772622585296631, | |
| "learning_rate": 5.926679909642341e-06, | |
| "loss": 7.8206, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.8915444504798117, | |
| "grad_norm": 0.4761473536491394, | |
| "learning_rate": 5.849212633593193e-06, | |
| "loss": 7.83, | |
| "step": 1231 | |
| }, | |
| { | |
| "epoch": 0.8922686945500634, | |
| "grad_norm": 0.46055835485458374, | |
| "learning_rate": 5.77223972355263e-06, | |
| "loss": 7.8043, | |
| "step": 1232 | |
| }, | |
| { | |
| "epoch": 0.892992938620315, | |
| "grad_norm": 0.5095073580741882, | |
| "learning_rate": 5.695761583689263e-06, | |
| "loss": 7.7667, | |
| "step": 1233 | |
| }, | |
| { | |
| "epoch": 0.8937171826905668, | |
| "grad_norm": 0.466761976480484, | |
| "learning_rate": 5.619778615573712e-06, | |
| "loss": 7.8028, | |
| "step": 1234 | |
| }, | |
| { | |
| "epoch": 0.8944414267608184, | |
| "grad_norm": 0.5017898082733154, | |
| "learning_rate": 5.544291218176578e-06, | |
| "loss": 7.79, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 0.8951656708310701, | |
| "grad_norm": 0.464933305978775, | |
| "learning_rate": 5.469299787866355e-06, | |
| "loss": 7.8344, | |
| "step": 1236 | |
| }, | |
| { | |
| "epoch": 0.8958899149013217, | |
| "grad_norm": 0.47450631856918335, | |
| "learning_rate": 5.394804718407276e-06, | |
| "loss": 7.8237, | |
| "step": 1237 | |
| }, | |
| { | |
| "epoch": 0.8966141589715734, | |
| "grad_norm": 0.4888070821762085, | |
| "learning_rate": 5.320806400957312e-06, | |
| "loss": 7.7909, | |
| "step": 1238 | |
| }, | |
| { | |
| "epoch": 0.8973384030418251, | |
| "grad_norm": 0.4835636019706726, | |
| "learning_rate": 5.247305224066079e-06, | |
| "loss": 7.8204, | |
| "step": 1239 | |
| }, | |
| { | |
| "epoch": 0.8980626471120767, | |
| "grad_norm": 0.5021434426307678, | |
| "learning_rate": 5.174301573672813e-06, | |
| "loss": 7.824, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.8987868911823285, | |
| "grad_norm": 0.47264546155929565, | |
| "learning_rate": 5.101795833104362e-06, | |
| "loss": 7.8474, | |
| "step": 1241 | |
| }, | |
| { | |
| "epoch": 0.8995111352525801, | |
| "grad_norm": 0.514746367931366, | |
| "learning_rate": 5.02978838307312e-06, | |
| "loss": 7.818, | |
| "step": 1242 | |
| }, | |
| { | |
| "epoch": 0.9002353793228318, | |
| "grad_norm": 0.45614975690841675, | |
| "learning_rate": 4.958279601675109e-06, | |
| "loss": 7.7936, | |
| "step": 1243 | |
| }, | |
| { | |
| "epoch": 0.9009596233930834, | |
| "grad_norm": 0.477163165807724, | |
| "learning_rate": 4.887269864387889e-06, | |
| "loss": 7.8263, | |
| "step": 1244 | |
| }, | |
| { | |
| "epoch": 0.9016838674633352, | |
| "grad_norm": 0.4801469147205353, | |
| "learning_rate": 4.816759544068705e-06, | |
| "loss": 7.7873, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 0.9024081115335868, | |
| "grad_norm": 0.5064570307731628, | |
| "learning_rate": 4.746749010952412e-06, | |
| "loss": 7.7897, | |
| "step": 1246 | |
| }, | |
| { | |
| "epoch": 0.9031323556038385, | |
| "grad_norm": 0.4912499785423279, | |
| "learning_rate": 4.677238632649639e-06, | |
| "loss": 7.7679, | |
| "step": 1247 | |
| }, | |
| { | |
| "epoch": 0.9038565996740902, | |
| "grad_norm": 0.5194550156593323, | |
| "learning_rate": 4.608228774144785e-06, | |
| "loss": 7.7607, | |
| "step": 1248 | |
| }, | |
| { | |
| "epoch": 0.9045808437443419, | |
| "grad_norm": 0.5179385542869568, | |
| "learning_rate": 4.5397197977940845e-06, | |
| "loss": 7.7615, | |
| "step": 1249 | |
| }, | |
| { | |
| "epoch": 0.9053050878145935, | |
| "grad_norm": 0.5159029364585876, | |
| "learning_rate": 4.471712063323818e-06, | |
| "loss": 7.7898, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.9060293318848452, | |
| "grad_norm": 0.4930214285850525, | |
| "learning_rate": 4.4042059278282865e-06, | |
| "loss": 7.8184, | |
| "step": 1251 | |
| }, | |
| { | |
| "epoch": 0.9067535759550969, | |
| "grad_norm": 0.5010021328926086, | |
| "learning_rate": 4.33720174576806e-06, | |
| "loss": 7.8191, | |
| "step": 1252 | |
| }, | |
| { | |
| "epoch": 0.9074778200253485, | |
| "grad_norm": 0.47560325264930725, | |
| "learning_rate": 4.270699868967998e-06, | |
| "loss": 7.8237, | |
| "step": 1253 | |
| }, | |
| { | |
| "epoch": 0.9082020640956002, | |
| "grad_norm": 0.4990387260913849, | |
| "learning_rate": 4.2047006466155115e-06, | |
| "loss": 7.841, | |
| "step": 1254 | |
| }, | |
| { | |
| "epoch": 0.9089263081658518, | |
| "grad_norm": 0.5125808715820312, | |
| "learning_rate": 4.139204425258625e-06, | |
| "loss": 7.8264, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 0.9096505522361036, | |
| "grad_norm": 0.47210493683815, | |
| "learning_rate": 4.0742115488042636e-06, | |
| "loss": 7.8051, | |
| "step": 1256 | |
| }, | |
| { | |
| "epoch": 0.9103747963063552, | |
| "grad_norm": 0.4914381206035614, | |
| "learning_rate": 4.009722358516366e-06, | |
| "loss": 7.8372, | |
| "step": 1257 | |
| }, | |
| { | |
| "epoch": 0.9110990403766069, | |
| "grad_norm": 0.48577770590782166, | |
| "learning_rate": 3.945737193014121e-06, | |
| "loss": 7.848, | |
| "step": 1258 | |
| }, | |
| { | |
| "epoch": 0.9118232844468586, | |
| "grad_norm": 0.49301815032958984, | |
| "learning_rate": 3.8822563882702e-06, | |
| "loss": 7.8581, | |
| "step": 1259 | |
| }, | |
| { | |
| "epoch": 0.9125475285171103, | |
| "grad_norm": 0.4755497872829437, | |
| "learning_rate": 3.819280277608939e-06, | |
| "loss": 7.8607, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.9132717725873619, | |
| "grad_norm": 0.483230322599411, | |
| "learning_rate": 3.7568091917047244e-06, | |
| "loss": 7.8094, | |
| "step": 1261 | |
| }, | |
| { | |
| "epoch": 0.9139960166576137, | |
| "grad_norm": 0.4857102632522583, | |
| "learning_rate": 3.694843458580055e-06, | |
| "loss": 7.8308, | |
| "step": 1262 | |
| }, | |
| { | |
| "epoch": 0.9147202607278653, | |
| "grad_norm": 0.4947606921195984, | |
| "learning_rate": 3.633383403604018e-06, | |
| "loss": 7.8509, | |
| "step": 1263 | |
| }, | |
| { | |
| "epoch": 0.915444504798117, | |
| "grad_norm": 0.489041268825531, | |
| "learning_rate": 3.572429349490436e-06, | |
| "loss": 7.8645, | |
| "step": 1264 | |
| }, | |
| { | |
| "epoch": 0.9161687488683686, | |
| "grad_norm": 0.5049325227737427, | |
| "learning_rate": 3.511981616296245e-06, | |
| "loss": 7.8248, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 0.9168929929386204, | |
| "grad_norm": 0.4949936866760254, | |
| "learning_rate": 3.4520405214197972e-06, | |
| "loss": 7.8719, | |
| "step": 1266 | |
| }, | |
| { | |
| "epoch": 0.917617237008872, | |
| "grad_norm": 0.49333250522613525, | |
| "learning_rate": 3.3926063795991723e-06, | |
| "loss": 7.8477, | |
| "step": 1267 | |
| }, | |
| { | |
| "epoch": 0.9183414810791236, | |
| "grad_norm": 0.4706898331642151, | |
| "learning_rate": 3.3336795029106027e-06, | |
| "loss": 7.831, | |
| "step": 1268 | |
| }, | |
| { | |
| "epoch": 0.9190657251493753, | |
| "grad_norm": 0.5014546513557434, | |
| "learning_rate": 3.2752602007667167e-06, | |
| "loss": 7.8772, | |
| "step": 1269 | |
| }, | |
| { | |
| "epoch": 0.919789969219627, | |
| "grad_norm": 0.50562584400177, | |
| "learning_rate": 3.2173487799150083e-06, | |
| "loss": 7.8077, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.9205142132898787, | |
| "grad_norm": 0.4861434996128082, | |
| "learning_rate": 3.159945544436171e-06, | |
| "loss": 7.8048, | |
| "step": 1271 | |
| }, | |
| { | |
| "epoch": 0.9212384573601303, | |
| "grad_norm": 0.4890470802783966, | |
| "learning_rate": 3.103050795742546e-06, | |
| "loss": 7.8218, | |
| "step": 1272 | |
| }, | |
| { | |
| "epoch": 0.9219627014303821, | |
| "grad_norm": 0.46198517084121704, | |
| "learning_rate": 3.046664832576518e-06, | |
| "loss": 7.8283, | |
| "step": 1273 | |
| }, | |
| { | |
| "epoch": 0.9226869455006337, | |
| "grad_norm": 0.4833485186100006, | |
| "learning_rate": 2.990787951008911e-06, | |
| "loss": 7.8617, | |
| "step": 1274 | |
| }, | |
| { | |
| "epoch": 0.9234111895708854, | |
| "grad_norm": 0.47587546706199646, | |
| "learning_rate": 2.93542044443752e-06, | |
| "loss": 7.8225, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.924135433641137, | |
| "grad_norm": 0.4739225506782532, | |
| "learning_rate": 2.8805626035854793e-06, | |
| "loss": 7.8603, | |
| "step": 1276 | |
| }, | |
| { | |
| "epoch": 0.9248596777113888, | |
| "grad_norm": 0.48468372225761414, | |
| "learning_rate": 2.8262147164997975e-06, | |
| "loss": 7.8818, | |
| "step": 1277 | |
| }, | |
| { | |
| "epoch": 0.9255839217816404, | |
| "grad_norm": 0.47420841455459595, | |
| "learning_rate": 2.772377068549792e-06, | |
| "loss": 7.8859, | |
| "step": 1278 | |
| }, | |
| { | |
| "epoch": 0.9263081658518921, | |
| "grad_norm": 0.4387821555137634, | |
| "learning_rate": 2.7190499424256665e-06, | |
| "loss": 7.8423, | |
| "step": 1279 | |
| }, | |
| { | |
| "epoch": 0.9270324099221438, | |
| "grad_norm": 0.4843416213989258, | |
| "learning_rate": 2.6662336181369485e-06, | |
| "loss": 7.8529, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.9277566539923955, | |
| "grad_norm": 0.45708930492401123, | |
| "learning_rate": 2.613928373011065e-06, | |
| "loss": 7.7918, | |
| "step": 1281 | |
| }, | |
| { | |
| "epoch": 0.9284808980626471, | |
| "grad_norm": 0.44832369685173035, | |
| "learning_rate": 2.5621344816918803e-06, | |
| "loss": 7.9067, | |
| "step": 1282 | |
| }, | |
| { | |
| "epoch": 0.9292051421328987, | |
| "grad_norm": 0.485809862613678, | |
| "learning_rate": 2.5108522161382153e-06, | |
| "loss": 7.8254, | |
| "step": 1283 | |
| }, | |
| { | |
| "epoch": 0.9299293862031505, | |
| "grad_norm": 0.4524907171726227, | |
| "learning_rate": 2.4600818456225083e-06, | |
| "loss": 7.8376, | |
| "step": 1284 | |
| }, | |
| { | |
| "epoch": 0.9306536302734021, | |
| "grad_norm": 0.4835200309753418, | |
| "learning_rate": 2.4098236367292805e-06, | |
| "loss": 7.7911, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 0.9313778743436538, | |
| "grad_norm": 0.45698094367980957, | |
| "learning_rate": 2.360077853353848e-06, | |
| "loss": 7.9004, | |
| "step": 1286 | |
| }, | |
| { | |
| "epoch": 0.9321021184139054, | |
| "grad_norm": 0.47620901465415955, | |
| "learning_rate": 2.3108447567008695e-06, | |
| "loss": 7.7822, | |
| "step": 1287 | |
| }, | |
| { | |
| "epoch": 0.9328263624841572, | |
| "grad_norm": 0.502024233341217, | |
| "learning_rate": 2.262124605282978e-06, | |
| "loss": 7.8447, | |
| "step": 1288 | |
| }, | |
| { | |
| "epoch": 0.9335506065544088, | |
| "grad_norm": 0.5169337391853333, | |
| "learning_rate": 2.213917654919473e-06, | |
| "loss": 7.8207, | |
| "step": 1289 | |
| }, | |
| { | |
| "epoch": 0.9342748506246605, | |
| "grad_norm": 0.5109481811523438, | |
| "learning_rate": 2.1662241587349195e-06, | |
| "loss": 7.8592, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.9349990946949122, | |
| "grad_norm": 0.47905823588371277, | |
| "learning_rate": 2.119044367157852e-06, | |
| "loss": 7.9079, | |
| "step": 1291 | |
| }, | |
| { | |
| "epoch": 0.9357233387651639, | |
| "grad_norm": 0.48844465613365173, | |
| "learning_rate": 2.0723785279194386e-06, | |
| "loss": 7.7795, | |
| "step": 1292 | |
| }, | |
| { | |
| "epoch": 0.9364475828354155, | |
| "grad_norm": 0.49026617407798767, | |
| "learning_rate": 2.026226886052207e-06, | |
| "loss": 7.8043, | |
| "step": 1293 | |
| }, | |
| { | |
| "epoch": 0.9371718269056672, | |
| "grad_norm": 0.505928635597229, | |
| "learning_rate": 1.9805896838887337e-06, | |
| "loss": 7.7842, | |
| "step": 1294 | |
| }, | |
| { | |
| "epoch": 0.9378960709759189, | |
| "grad_norm": 0.4998481869697571, | |
| "learning_rate": 1.935467161060378e-06, | |
| "loss": 7.903, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 0.9386203150461706, | |
| "grad_norm": 0.5379026532173157, | |
| "learning_rate": 1.8908595544960272e-06, | |
| "loss": 7.8672, | |
| "step": 1296 | |
| }, | |
| { | |
| "epoch": 0.9393445591164222, | |
| "grad_norm": 0.4858308732509613, | |
| "learning_rate": 1.8467670984208652e-06, | |
| "loss": 7.7751, | |
| "step": 1297 | |
| }, | |
| { | |
| "epoch": 0.940068803186674, | |
| "grad_norm": 0.51478111743927, | |
| "learning_rate": 1.8031900243550948e-06, | |
| "loss": 7.8418, | |
| "step": 1298 | |
| }, | |
| { | |
| "epoch": 0.9407930472569256, | |
| "grad_norm": 0.4762341380119324, | |
| "learning_rate": 1.760128561112795e-06, | |
| "loss": 7.8597, | |
| "step": 1299 | |
| }, | |
| { | |
| "epoch": 0.9415172913271772, | |
| "grad_norm": 0.5281409025192261, | |
| "learning_rate": 1.7175829348006767e-06, | |
| "loss": 7.7901, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.9422415353974289, | |
| "grad_norm": 0.4912181496620178, | |
| "learning_rate": 1.6755533688168624e-06, | |
| "loss": 7.8392, | |
| "step": 1301 | |
| }, | |
| { | |
| "epoch": 0.9429657794676806, | |
| "grad_norm": 0.47923725843429565, | |
| "learning_rate": 1.634040083849786e-06, | |
| "loss": 7.8797, | |
| "step": 1302 | |
| }, | |
| { | |
| "epoch": 0.9436900235379323, | |
| "grad_norm": 0.48932698369026184, | |
| "learning_rate": 1.593043297876984e-06, | |
| "loss": 7.8072, | |
| "step": 1303 | |
| }, | |
| { | |
| "epoch": 0.9444142676081839, | |
| "grad_norm": 0.4930644631385803, | |
| "learning_rate": 1.5525632261639722e-06, | |
| "loss": 7.8609, | |
| "step": 1304 | |
| }, | |
| { | |
| "epoch": 0.9451385116784357, | |
| "grad_norm": 0.46192994713783264, | |
| "learning_rate": 1.5126000812631159e-06, | |
| "loss": 7.8686, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 0.9458627557486873, | |
| "grad_norm": 0.4716580808162689, | |
| "learning_rate": 1.4731540730124616e-06, | |
| "loss": 7.8314, | |
| "step": 1306 | |
| }, | |
| { | |
| "epoch": 0.946586999818939, | |
| "grad_norm": 0.45859482884407043, | |
| "learning_rate": 1.4342254085347506e-06, | |
| "loss": 7.8843, | |
| "step": 1307 | |
| }, | |
| { | |
| "epoch": 0.9473112438891906, | |
| "grad_norm": 0.4815140664577484, | |
| "learning_rate": 1.3958142922362083e-06, | |
| "loss": 7.8706, | |
| "step": 1308 | |
| }, | |
| { | |
| "epoch": 0.9480354879594424, | |
| "grad_norm": 0.46593400835990906, | |
| "learning_rate": 1.3579209258055226e-06, | |
| "loss": 7.8941, | |
| "step": 1309 | |
| }, | |
| { | |
| "epoch": 0.948759732029694, | |
| "grad_norm": 0.47369858622550964, | |
| "learning_rate": 1.3205455082128225e-06, | |
| "loss": 7.8729, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.9494839760999457, | |
| "grad_norm": 0.46756669878959656, | |
| "learning_rate": 1.283688235708569e-06, | |
| "loss": 7.8424, | |
| "step": 1311 | |
| }, | |
| { | |
| "epoch": 0.9502082201701973, | |
| "grad_norm": 0.5048179626464844, | |
| "learning_rate": 1.2473493018225646e-06, | |
| "loss": 7.854, | |
| "step": 1312 | |
| }, | |
| { | |
| "epoch": 0.9509324642404491, | |
| "grad_norm": 0.45838841795921326, | |
| "learning_rate": 1.2115288973629014e-06, | |
| "loss": 7.8695, | |
| "step": 1313 | |
| }, | |
| { | |
| "epoch": 0.9516567083107007, | |
| "grad_norm": 0.5052929520606995, | |
| "learning_rate": 1.176227210415015e-06, | |
| "loss": 7.788, | |
| "step": 1314 | |
| }, | |
| { | |
| "epoch": 0.9523809523809523, | |
| "grad_norm": 0.4793596863746643, | |
| "learning_rate": 1.1414444263406432e-06, | |
| "loss": 7.8184, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 0.9531051964512041, | |
| "grad_norm": 0.47426638007164, | |
| "learning_rate": 1.1071807277768798e-06, | |
| "loss": 7.8186, | |
| "step": 1316 | |
| }, | |
| { | |
| "epoch": 0.9538294405214557, | |
| "grad_norm": 0.48723068833351135, | |
| "learning_rate": 1.0734362946352107e-06, | |
| "loss": 7.8259, | |
| "step": 1317 | |
| }, | |
| { | |
| "epoch": 0.9545536845917074, | |
| "grad_norm": 0.4758625626564026, | |
| "learning_rate": 1.0402113041005468e-06, | |
| "loss": 7.8321, | |
| "step": 1318 | |
| }, | |
| { | |
| "epoch": 0.955277928661959, | |
| "grad_norm": 0.4779294431209564, | |
| "learning_rate": 1.0075059306303702e-06, | |
| "loss": 7.8489, | |
| "step": 1319 | |
| }, | |
| { | |
| "epoch": 0.9560021727322108, | |
| "grad_norm": 0.48141980171203613, | |
| "learning_rate": 9.753203459537009e-07, | |
| "loss": 7.8922, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.9567264168024624, | |
| "grad_norm": 0.4511309862136841, | |
| "learning_rate": 9.436547190702971e-07, | |
| "loss": 7.8431, | |
| "step": 1321 | |
| }, | |
| { | |
| "epoch": 0.9574506608727141, | |
| "grad_norm": 0.4854510724544525, | |
| "learning_rate": 9.125092162497129e-07, | |
| "loss": 7.8434, | |
| "step": 1322 | |
| }, | |
| { | |
| "epoch": 0.9581749049429658, | |
| "grad_norm": 0.48352810740470886, | |
| "learning_rate": 8.818840010304308e-07, | |
| "loss": 7.864, | |
| "step": 1323 | |
| }, | |
| { | |
| "epoch": 0.9588991490132175, | |
| "grad_norm": 0.4762645661830902, | |
| "learning_rate": 8.5177923421903e-07, | |
| "loss": 7.8196, | |
| "step": 1324 | |
| }, | |
| { | |
| "epoch": 0.9596233930834691, | |
| "grad_norm": 0.4877321720123291, | |
| "learning_rate": 8.221950738893203e-07, | |
| "loss": 7.7954, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.9603476371537208, | |
| "grad_norm": 0.4776611924171448, | |
| "learning_rate": 7.931316753815088e-07, | |
| "loss": 7.8505, | |
| "step": 1326 | |
| }, | |
| { | |
| "epoch": 0.9610718812239725, | |
| "grad_norm": 0.479509562253952, | |
| "learning_rate": 7.645891913014013e-07, | |
| "loss": 7.8779, | |
| "step": 1327 | |
| }, | |
| { | |
| "epoch": 0.9617961252942242, | |
| "grad_norm": 0.46871259808540344, | |
| "learning_rate": 7.365677715195918e-07, | |
| "loss": 7.8008, | |
| "step": 1328 | |
| }, | |
| { | |
| "epoch": 0.9625203693644758, | |
| "grad_norm": 0.4688069820404053, | |
| "learning_rate": 7.090675631706512e-07, | |
| "loss": 7.8098, | |
| "step": 1329 | |
| }, | |
| { | |
| "epoch": 0.9632446134347274, | |
| "grad_norm": 0.4722239077091217, | |
| "learning_rate": 6.820887106524065e-07, | |
| "loss": 7.8363, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.9639688575049792, | |
| "grad_norm": 0.4427089989185333, | |
| "learning_rate": 6.556313556251636e-07, | |
| "loss": 7.8216, | |
| "step": 1331 | |
| }, | |
| { | |
| "epoch": 0.9646931015752308, | |
| "grad_norm": 0.475266695022583, | |
| "learning_rate": 6.296956370109075e-07, | |
| "loss": 7.8203, | |
| "step": 1332 | |
| }, | |
| { | |
| "epoch": 0.9654173456454825, | |
| "grad_norm": 0.47641658782958984, | |
| "learning_rate": 6.042816909926585e-07, | |
| "loss": 7.85, | |
| "step": 1333 | |
| }, | |
| { | |
| "epoch": 0.9661415897157342, | |
| "grad_norm": 0.47987788915634155, | |
| "learning_rate": 5.793896510137287e-07, | |
| "loss": 7.7702, | |
| "step": 1334 | |
| }, | |
| { | |
| "epoch": 0.9668658337859859, | |
| "grad_norm": 0.5124553442001343, | |
| "learning_rate": 5.550196477769665e-07, | |
| "loss": 7.8409, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 0.9675900778562375, | |
| "grad_norm": 0.4751659333705902, | |
| "learning_rate": 5.311718092441465e-07, | |
| "loss": 7.8114, | |
| "step": 1336 | |
| }, | |
| { | |
| "epoch": 0.9683143219264893, | |
| "grad_norm": 0.45295917987823486, | |
| "learning_rate": 5.078462606352585e-07, | |
| "loss": 7.9019, | |
| "step": 1337 | |
| }, | |
| { | |
| "epoch": 0.9690385659967409, | |
| "grad_norm": 0.49618884921073914, | |
| "learning_rate": 4.850431244278753e-07, | |
| "loss": 7.7931, | |
| "step": 1338 | |
| }, | |
| { | |
| "epoch": 0.9697628100669926, | |
| "grad_norm": 0.48623406887054443, | |
| "learning_rate": 4.627625203564523e-07, | |
| "loss": 7.8004, | |
| "step": 1339 | |
| }, | |
| { | |
| "epoch": 0.9704870541372442, | |
| "grad_norm": 0.43766334652900696, | |
| "learning_rate": 4.4100456541177335e-07, | |
| "loss": 7.8297, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.971211298207496, | |
| "grad_norm": 0.46884864568710327, | |
| "learning_rate": 4.1976937384028417e-07, | |
| "loss": 7.8893, | |
| "step": 1341 | |
| }, | |
| { | |
| "epoch": 0.9719355422777476, | |
| "grad_norm": 0.48704731464385986, | |
| "learning_rate": 3.990570571435259e-07, | |
| "loss": 7.7991, | |
| "step": 1342 | |
| }, | |
| { | |
| "epoch": 0.9726597863479993, | |
| "grad_norm": 0.4837970733642578, | |
| "learning_rate": 3.7886772407751406e-07, | |
| "loss": 7.8021, | |
| "step": 1343 | |
| }, | |
| { | |
| "epoch": 0.973384030418251, | |
| "grad_norm": 0.5404148697853088, | |
| "learning_rate": 3.5920148065220484e-07, | |
| "loss": 7.7664, | |
| "step": 1344 | |
| }, | |
| { | |
| "epoch": 0.9741082744885027, | |
| "grad_norm": 0.544996976852417, | |
| "learning_rate": 3.4005843013089625e-07, | |
| "loss": 7.8745, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 0.9748325185587543, | |
| "grad_norm": 0.5073729753494263, | |
| "learning_rate": 3.2143867302973917e-07, | |
| "loss": 7.774, | |
| "step": 1346 | |
| }, | |
| { | |
| "epoch": 0.9755567626290059, | |
| "grad_norm": 0.5275909304618835, | |
| "learning_rate": 3.033423071171604e-07, | |
| "loss": 7.7614, | |
| "step": 1347 | |
| }, | |
| { | |
| "epoch": 0.9762810066992577, | |
| "grad_norm": 0.5046758651733398, | |
| "learning_rate": 2.857694274133849e-07, | |
| "loss": 7.7907, | |
| "step": 1348 | |
| }, | |
| { | |
| "epoch": 0.9770052507695093, | |
| "grad_norm": 0.5203927159309387, | |
| "learning_rate": 2.6872012618990306e-07, | |
| "loss": 7.852, | |
| "step": 1349 | |
| }, | |
| { | |
| "epoch": 0.977729494839761, | |
| "grad_norm": 0.6230973601341248, | |
| "learning_rate": 2.5219449296900455e-07, | |
| "loss": 7.8475, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.9784537389100126, | |
| "grad_norm": 0.4698044955730438, | |
| "learning_rate": 2.3619261452335617e-07, | |
| "loss": 7.83, | |
| "step": 1351 | |
| }, | |
| { | |
| "epoch": 0.9791779829802644, | |
| "grad_norm": 0.48747655749320984, | |
| "learning_rate": 2.207145748754247e-07, | |
| "loss": 7.8553, | |
| "step": 1352 | |
| }, | |
| { | |
| "epoch": 0.979902227050516, | |
| "grad_norm": 0.5164722800254822, | |
| "learning_rate": 2.0576045529715482e-07, | |
| "loss": 7.8262, | |
| "step": 1353 | |
| }, | |
| { | |
| "epoch": 0.9806264711207677, | |
| "grad_norm": 0.4546986222267151, | |
| "learning_rate": 1.9133033430949186e-07, | |
| "loss": 7.9124, | |
| "step": 1354 | |
| }, | |
| { | |
| "epoch": 0.9813507151910194, | |
| "grad_norm": 0.4600694179534912, | |
| "learning_rate": 1.7742428768195985e-07, | |
| "loss": 7.8223, | |
| "step": 1355 | |
| }, | |
| { | |
| "epoch": 0.9820749592612711, | |
| "grad_norm": 0.4773062765598297, | |
| "learning_rate": 1.6404238843230612e-07, | |
| "loss": 7.8256, | |
| "step": 1356 | |
| }, | |
| { | |
| "epoch": 0.9827992033315227, | |
| "grad_norm": 0.46029308438301086, | |
| "learning_rate": 1.5118470682605745e-07, | |
| "loss": 7.8241, | |
| "step": 1357 | |
| }, | |
| { | |
| "epoch": 0.9835234474017744, | |
| "grad_norm": 0.45993316173553467, | |
| "learning_rate": 1.38851310376209e-07, | |
| "loss": 7.8153, | |
| "step": 1358 | |
| }, | |
| { | |
| "epoch": 0.9842476914720261, | |
| "grad_norm": 0.45841482281684875, | |
| "learning_rate": 1.2704226384282482e-07, | |
| "loss": 7.8534, | |
| "step": 1359 | |
| }, | |
| { | |
| "epoch": 0.9849719355422778, | |
| "grad_norm": 0.4997026026248932, | |
| "learning_rate": 1.157576292327378e-07, | |
| "loss": 7.8132, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.9856961796125294, | |
| "grad_norm": 0.4676493704319, | |
| "learning_rate": 1.0499746579919478e-07, | |
| "loss": 7.8852, | |
| "step": 1361 | |
| }, | |
| { | |
| "epoch": 0.986420423682781, | |
| "grad_norm": 0.497711718082428, | |
| "learning_rate": 9.476183004154537e-08, | |
| "loss": 7.8372, | |
| "step": 1362 | |
| }, | |
| { | |
| "epoch": 0.9871446677530328, | |
| "grad_norm": 0.46188995242118835, | |
| "learning_rate": 8.505077570496456e-08, | |
| "loss": 7.8758, | |
| "step": 1363 | |
| }, | |
| { | |
| "epoch": 0.9878689118232844, | |
| "grad_norm": 0.46226826310157776, | |
| "learning_rate": 7.586435378016399e-08, | |
| "loss": 7.8781, | |
| "step": 1364 | |
| }, | |
| { | |
| "epoch": 0.9885931558935361, | |
| "grad_norm": 0.47206389904022217, | |
| "learning_rate": 6.720261250311444e-08, | |
| "loss": 7.8517, | |
| "step": 1365 | |
| }, | |
| { | |
| "epoch": 0.9893173999637878, | |
| "grad_norm": 0.4836471378803253, | |
| "learning_rate": 5.9065597354790445e-08, | |
| "loss": 7.8837, | |
| "step": 1366 | |
| }, | |
| { | |
| "epoch": 0.9900416440340395, | |
| "grad_norm": 0.4508313834667206, | |
| "learning_rate": 5.1453351060959387e-08, | |
| "loss": 7.8247, | |
| "step": 1367 | |
| }, | |
| { | |
| "epoch": 0.9907658881042911, | |
| "grad_norm": 0.46068304777145386, | |
| "learning_rate": 4.43659135919372e-08, | |
| "loss": 7.8923, | |
| "step": 1368 | |
| }, | |
| { | |
| "epoch": 0.9914901321745428, | |
| "grad_norm": 0.4875946342945099, | |
| "learning_rate": 3.780332216234417e-08, | |
| "loss": 7.866, | |
| "step": 1369 | |
| }, | |
| { | |
| "epoch": 0.9922143762447945, | |
| "grad_norm": 0.48047712445259094, | |
| "learning_rate": 3.1765611230993865e-08, | |
| "loss": 7.7948, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.9929386203150462, | |
| "grad_norm": 0.4504997134208679, | |
| "learning_rate": 2.625281250061562e-08, | |
| "loss": 7.808, | |
| "step": 1371 | |
| }, | |
| { | |
| "epoch": 0.9936628643852978, | |
| "grad_norm": 0.44184571504592896, | |
| "learning_rate": 2.1264954917776802e-08, | |
| "loss": 7.8365, | |
| "step": 1372 | |
| }, | |
| { | |
| "epoch": 0.9943871084555496, | |
| "grad_norm": 0.46836909651756287, | |
| "learning_rate": 1.6802064672660767e-08, | |
| "loss": 7.8799, | |
| "step": 1373 | |
| }, | |
| { | |
| "epoch": 0.9951113525258012, | |
| "grad_norm": 0.5343976616859436, | |
| "learning_rate": 1.286416519897804e-08, | |
| "loss": 7.7852, | |
| "step": 1374 | |
| }, | |
| { | |
| "epoch": 0.9958355965960529, | |
| "grad_norm": 0.5096919536590576, | |
| "learning_rate": 9.451277173788687e-09, | |
| "loss": 7.7644, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.9965598406663045, | |
| "grad_norm": 0.538051426410675, | |
| "learning_rate": 6.563418517469e-09, | |
| "loss": 7.8173, | |
| "step": 1376 | |
| }, | |
| { | |
| "epoch": 0.9972840847365562, | |
| "grad_norm": 0.4815247356891632, | |
| "learning_rate": 4.200604393556073e-09, | |
| "loss": 7.8366, | |
| "step": 1377 | |
| }, | |
| { | |
| "epoch": 0.9980083288068079, | |
| "grad_norm": 0.49304264783859253, | |
| "learning_rate": 2.3628472086811847e-09, | |
| "loss": 7.8003, | |
| "step": 1378 | |
| }, | |
| { | |
| "epoch": 0.9987325728770595, | |
| "grad_norm": 0.5107442736625671, | |
| "learning_rate": 1.0501566125364904e-09, | |
| "loss": 7.7771, | |
| "step": 1379 | |
| }, | |
| { | |
| "epoch": 0.9994568169473113, | |
| "grad_norm": 0.5584591031074524, | |
| "learning_rate": 2.6253949776400275e-10, | |
| "loss": 7.8327, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 1.0003621220351258, | |
| "grad_norm": 0.8555117845535278, | |
| "learning_rate": 0.0, | |
| "loss": 11.8993, | |
| "step": 1381 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 1381, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 346, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 564726482337792.0, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |