| [ | |
| { | |
| "loss": 0.2659, | |
| "grad_norm": 3.4286012649536133, | |
| "learning_rate": 1.9715261958997724e-05, | |
| "epoch": 0.2847380410022779, | |
| "step": 500 | |
| }, | |
| { | |
| "loss": 0.1081, | |
| "grad_norm": 0.7319045066833496, | |
| "learning_rate": 1.9430523917995446e-05, | |
| "epoch": 0.5694760820045558, | |
| "step": 1000 | |
| }, | |
| { | |
| "loss": 0.0777, | |
| "grad_norm": 3.7819042205810547, | |
| "learning_rate": 1.9145785876993168e-05, | |
| "epoch": 0.8542141230068337, | |
| "step": 1500 | |
| }, | |
| { | |
| "eval_loss": 0.07203580439090729, | |
| "eval_overall": { | |
| "precision": 0.8912693298969072, | |
| "recall": 0.931167956916863, | |
| "f1": 0.9107818930041152, | |
| "accuracy": 0.9814269735680226 | |
| }, | |
| "eval_per_label": { | |
| "LOC": { | |
| "precision": 0.9239187076602398, | |
| "recall": 0.965160587915079, | |
| "f1": 0.9440894568690097, | |
| "number": 1837 | |
| }, | |
| "MISC": { | |
| "precision": 0.7727272727272727, | |
| "recall": 0.8297180043383948, | |
| "f1": 0.8002092050209204, | |
| "number": 922 | |
| }, | |
| "ORG": { | |
| "precision": 0.8624823695345557, | |
| "recall": 0.9120059656972409, | |
| "f1": 0.8865530989488946, | |
| "number": 1341 | |
| }, | |
| "PER": { | |
| "precision": 0.9420520999468368, | |
| "recall": 0.9619978284473398, | |
| "f1": 0.9519204942250873, | |
| "number": 1842 | |
| } | |
| }, | |
| "eval_runtime": 7.4533, | |
| "eval_samples_per_second": 436.049, | |
| "eval_steps_per_second": 54.607, | |
| "epoch": 1.0, | |
| "step": 1756 | |
| }, | |
| { | |
| "loss": 0.0631, | |
| "grad_norm": 0.11871737241744995, | |
| "learning_rate": 1.886104783599089e-05, | |
| "epoch": 1.1389521640091116, | |
| "step": 2000 | |
| }, | |
| { | |
| "loss": 0.0464, | |
| "grad_norm": 0.2318667322397232, | |
| "learning_rate": 1.8576309794988612e-05, | |
| "epoch": 1.4236902050113895, | |
| "step": 2500 | |
| }, | |
| { | |
| "loss": 0.0442, | |
| "grad_norm": 0.35079020261764526, | |
| "learning_rate": 1.8291571753986334e-05, | |
| "epoch": 1.7084282460136673, | |
| "step": 3000 | |
| }, | |
| { | |
| "loss": 0.0411, | |
| "grad_norm": 0.06829982250928879, | |
| "learning_rate": 1.8006833712984056e-05, | |
| "epoch": 1.9931662870159452, | |
| "step": 3500 | |
| }, | |
| { | |
| "eval_loss": 0.07787470519542694, | |
| "eval_overall": { | |
| "precision": 0.9304723885562209, | |
| "recall": 0.9414338606529788, | |
| "f1": 0.9359210306173666, | |
| "accuracy": 0.9842968152116324 | |
| }, | |
| "eval_per_label": { | |
| "LOC": { | |
| "precision": 0.9511015583019882, | |
| "recall": 0.9635274904735982, | |
| "f1": 0.9572742022714981, | |
| "number": 1837 | |
| }, | |
| "MISC": { | |
| "precision": 0.8725910064239829, | |
| "recall": 0.8839479392624728, | |
| "f1": 0.8782327586206896, | |
| "number": 922 | |
| }, | |
| "ORG": { | |
| "precision": 0.9050822122571002, | |
| "recall": 0.9030574198359433, | |
| "f1": 0.9040686823441583, | |
| "number": 1341 | |
| }, | |
| "PER": { | |
| "precision": 0.9568919638105375, | |
| "recall": 0.9761129207383279, | |
| "f1": 0.9664068798710024, | |
| "number": 1842 | |
| } | |
| }, | |
| "eval_runtime": 7.5011, | |
| "eval_samples_per_second": 433.267, | |
| "eval_steps_per_second": 54.258, | |
| "epoch": 2.0, | |
| "step": 3512 | |
| }, | |
| { | |
| "loss": 0.0258, | |
| "grad_norm": 4.330985069274902, | |
| "learning_rate": 1.7722095671981778e-05, | |
| "epoch": 2.277904328018223, | |
| "step": 4000 | |
| }, | |
| { | |
| "loss": 0.0284, | |
| "grad_norm": 0.06637139618396759, | |
| "learning_rate": 1.74373576309795e-05, | |
| "epoch": 2.562642369020501, | |
| "step": 4500 | |
| }, | |
| { | |
| "loss": 0.0269, | |
| "grad_norm": 11.263030052185059, | |
| "learning_rate": 1.7152619589977222e-05, | |
| "epoch": 2.847380410022779, | |
| "step": 5000 | |
| }, | |
| { | |
| "eval_loss": 0.07202505320310593, | |
| "eval_overall": { | |
| "precision": 0.9318106587222774, | |
| "recall": 0.9474924267923258, | |
| "f1": 0.9395861148197597, | |
| "accuracy": 0.9857685288750221 | |
| }, | |
| "eval_per_label": { | |
| "LOC": { | |
| "precision": 0.9626218851570965, | |
| "recall": 0.9673380511703865, | |
| "f1": 0.9649742058104807, | |
| "number": 1837 | |
| }, | |
| "MISC": { | |
| "precision": 0.8943355119825708, | |
| "recall": 0.8904555314533622, | |
| "f1": 0.8923913043478261, | |
| "number": 922 | |
| }, | |
| "ORG": { | |
| "precision": 0.8772791023842917, | |
| "recall": 0.9328859060402684, | |
| "f1": 0.9042284062161186, | |
| "number": 1341 | |
| }, | |
| "PER": { | |
| "precision": 0.9616630669546437, | |
| "recall": 0.9668838219326819, | |
| "f1": 0.9642663779101246, | |
| "number": 1842 | |
| } | |
| }, | |
| "eval_runtime": 6.5043, | |
| "eval_samples_per_second": 499.669, | |
| "eval_steps_per_second": 62.574, | |
| "epoch": 3.0, | |
| "step": 5268 | |
| }, | |
| { | |
| "loss": 0.0226, | |
| "grad_norm": 0.7242124676704407, | |
| "learning_rate": 1.6867881548974945e-05, | |
| "epoch": 3.132118451025057, | |
| "step": 5500 | |
| }, | |
| { | |
| "loss": 0.0165, | |
| "grad_norm": 0.006902824155986309, | |
| "learning_rate": 1.6583143507972667e-05, | |
| "epoch": 3.416856492027335, | |
| "step": 6000 | |
| }, | |
| { | |
| "loss": 0.0176, | |
| "grad_norm": 0.031127002090215683, | |
| "learning_rate": 1.629840546697039e-05, | |
| "epoch": 3.7015945330296125, | |
| "step": 6500 | |
| }, | |
| { | |
| "loss": 0.0196, | |
| "grad_norm": 0.0038718737196177244, | |
| "learning_rate": 1.601366742596811e-05, | |
| "epoch": 3.9863325740318905, | |
| "step": 7000 | |
| }, | |
| { | |
| "eval_loss": 0.08569859713315964, | |
| "eval_overall": { | |
| "precision": 0.930635838150289, | |
| "recall": 0.9483338943116796, | |
| "f1": 0.9394015170459282, | |
| "accuracy": 0.9850621063165951 | |
| }, | |
| "eval_per_label": { | |
| "LOC": { | |
| "precision": 0.9533011272141707, | |
| "recall": 0.9667936853565596, | |
| "f1": 0.96, | |
| "number": 1837 | |
| }, | |
| "MISC": { | |
| "precision": 0.8623188405797102, | |
| "recall": 0.903470715835141, | |
| "f1": 0.8824152542372882, | |
| "number": 922 | |
| }, | |
| "ORG": { | |
| "precision": 0.8995664739884393, | |
| "recall": 0.9284116331096197, | |
| "f1": 0.9137614678899083, | |
| "number": 1341 | |
| }, | |
| "PER": { | |
| "precision": 0.9668838219326819, | |
| "recall": 0.9668838219326819, | |
| "f1": 0.9668838219326819, | |
| "number": 1842 | |
| } | |
| }, | |
| "eval_runtime": 7.5249, | |
| "eval_samples_per_second": 431.899, | |
| "eval_steps_per_second": 54.087, | |
| "epoch": 4.0, | |
| "step": 7024 | |
| }, | |
| { | |
| "loss": 0.0138, | |
| "grad_norm": 0.17987537384033203, | |
| "learning_rate": 1.5728929384965833e-05, | |
| "epoch": 4.271070615034168, | |
| "step": 7500 | |
| }, | |
| { | |
| "loss": 0.0134, | |
| "grad_norm": 1.4385559558868408, | |
| "learning_rate": 1.5444191343963555e-05, | |
| "epoch": 4.555808656036446, | |
| "step": 8000 | |
| }, | |
| { | |
| "loss": 0.0139, | |
| "grad_norm": 0.02356315404176712, | |
| "learning_rate": 1.5159453302961277e-05, | |
| "epoch": 4.840546697038724, | |
| "step": 8500 | |
| }, | |
| { | |
| "eval_loss": 0.08417785912752151, | |
| "eval_overall": { | |
| "precision": 0.9310686015831134, | |
| "recall": 0.9501851228542578, | |
| "f1": 0.9405297351324339, | |
| "accuracy": 0.9851062577264967 | |
| }, | |
| "eval_per_label": { | |
| "LOC": { | |
| "precision": 0.9627228525121556, | |
| "recall": 0.9700598802395209, | |
| "f1": 0.9663774403470715, | |
| "number": 1837 | |
| }, | |
| "MISC": { | |
| "precision": 0.8386454183266933, | |
| "recall": 0.913232104121475, | |
| "f1": 0.8743509865005192, | |
| "number": 922 | |
| }, | |
| "ORG": { | |
| "precision": 0.9070110701107011, | |
| "recall": 0.9164802386278896, | |
| "f1": 0.9117210682492581, | |
| "number": 1341 | |
| }, | |
| "PER": { | |
| "precision": 0.9670981661272924, | |
| "recall": 0.9733984799131379, | |
| "f1": 0.9702380952380952, | |
| "number": 1842 | |
| } | |
| }, | |
| "eval_runtime": 7.5038, | |
| "eval_samples_per_second": 433.114, | |
| "eval_steps_per_second": 54.239, | |
| "epoch": 5.0, | |
| "step": 8780 | |
| }, | |
| { | |
| "loss": 0.0092, | |
| "grad_norm": 0.7477717399597168, | |
| "learning_rate": 1.4874715261958999e-05, | |
| "epoch": 5.125284738041002, | |
| "step": 9000 | |
| }, | |
| { | |
| "loss": 0.006, | |
| "grad_norm": 0.04322722181677818, | |
| "learning_rate": 1.4589977220956721e-05, | |
| "epoch": 5.41002277904328, | |
| "step": 9500 | |
| }, | |
| { | |
| "loss": 0.0073, | |
| "grad_norm": 0.051729559898376465, | |
| "learning_rate": 1.4305239179954442e-05, | |
| "epoch": 5.694760820045558, | |
| "step": 10000 | |
| }, | |
| { | |
| "loss": 0.0079, | |
| "grad_norm": 0.012189100496470928, | |
| "learning_rate": 1.4020501138952165e-05, | |
| "epoch": 5.979498861047836, | |
| "step": 10500 | |
| }, | |
| { | |
| "eval_loss": 0.09725591540336609, | |
| "eval_overall": { | |
| "precision": 0.9291845493562232, | |
| "recall": 0.9473241332884551, | |
| "f1": 0.9381666666666667, | |
| "accuracy": 0.9844734208512392 | |
| }, | |
| "eval_per_label": { | |
| "LOC": { | |
| "precision": 0.9546424759871932, | |
| "recall": 0.9738704409363091, | |
| "f1": 0.9641606036108865, | |
| "number": 1837 | |
| }, | |
| "MISC": { | |
| "precision": 0.858739837398374, | |
| "recall": 0.9164859002169198, | |
| "f1": 0.8866736621196223, | |
| "number": 922 | |
| }, | |
| "ORG": { | |
| "precision": 0.8921852387843705, | |
| "recall": 0.9194630872483222, | |
| "f1": 0.9056188027910393, | |
| "number": 1341 | |
| }, | |
| "PER": { | |
| "precision": 0.9691969196919692, | |
| "recall": 0.9565689467969598, | |
| "f1": 0.9628415300546448, | |
| "number": 1842 | |
| } | |
| }, | |
| "eval_runtime": 7.4822, | |
| "eval_samples_per_second": 434.362, | |
| "eval_steps_per_second": 54.395, | |
| "epoch": 6.0, | |
| "step": 10536 | |
| }, | |
| { | |
| "loss": 0.008, | |
| "grad_norm": 0.0029869996942579746, | |
| "learning_rate": 1.3735763097949887e-05, | |
| "epoch": 6.264236902050114, | |
| "step": 11000 | |
| }, | |
| { | |
| "loss": 0.0066, | |
| "grad_norm": 0.018263721838593483, | |
| "learning_rate": 1.3451025056947608e-05, | |
| "epoch": 6.548974943052392, | |
| "step": 11500 | |
| }, | |
| { | |
| "loss": 0.0057, | |
| "grad_norm": 0.020874306559562683, | |
| "learning_rate": 1.3166287015945332e-05, | |
| "epoch": 6.83371298405467, | |
| "step": 12000 | |
| }, | |
| { | |
| "eval_loss": 0.097112737596035, | |
| "eval_overall": { | |
| "precision": 0.9343584656084656, | |
| "recall": 0.9510265903736116, | |
| "f1": 0.9426188490408675, | |
| "accuracy": 0.9859745687878966 | |
| }, | |
| "eval_per_label": { | |
| "LOC": { | |
| "precision": 0.9689373297002725, | |
| "recall": 0.9678824169842134, | |
| "f1": 0.968409586056645, | |
| "number": 1837 | |
| }, | |
| "MISC": { | |
| "precision": 0.8805031446540881, | |
| "recall": 0.911062906724512, | |
| "f1": 0.8955223880597015, | |
| "number": 922 | |
| }, | |
| "ORG": { | |
| "precision": 0.8851063829787233, | |
| "recall": 0.930648769574944, | |
| "f1": 0.9073064340239914, | |
| "number": 1341 | |
| }, | |
| "PER": { | |
| "precision": 0.965386695511087, | |
| "recall": 0.9690553745928339, | |
| "f1": 0.967217556217827, | |
| "number": 1842 | |
| } | |
| }, | |
| "eval_runtime": 7.4789, | |
| "eval_samples_per_second": 434.555, | |
| "eval_steps_per_second": 54.42, | |
| "epoch": 7.0, | |
| "step": 12292 | |
| }, | |
| { | |
| "loss": 0.0031, | |
| "grad_norm": 0.0007487820694223046, | |
| "learning_rate": 1.2881548974943054e-05, | |
| "epoch": 7.118451025056948, | |
| "step": 12500 | |
| }, | |
| { | |
| "loss": 0.0037, | |
| "grad_norm": 15.724783897399902, | |
| "learning_rate": 1.2596810933940776e-05, | |
| "epoch": 7.403189066059226, | |
| "step": 13000 | |
| }, | |
| { | |
| "loss": 0.0061, | |
| "grad_norm": 0.614613950252533, | |
| "learning_rate": 1.2312072892938498e-05, | |
| "epoch": 7.687927107061503, | |
| "step": 13500 | |
| }, | |
| { | |
| "loss": 0.0036, | |
| "grad_norm": 0.0012391641503199935, | |
| "learning_rate": 1.2027334851936218e-05, | |
| "epoch": 7.972665148063781, | |
| "step": 14000 | |
| }, | |
| { | |
| "eval_loss": 0.10294844955205917, | |
| "eval_overall": { | |
| "precision": 0.9285831285831286, | |
| "recall": 0.9540558734432851, | |
| "f1": 0.9411471735701834, | |
| "accuracy": 0.9852387119562018 | |
| }, | |
| "eval_per_label": { | |
| "LOC": { | |
| "precision": 0.9664864864864865, | |
| "recall": 0.9733260751224823, | |
| "f1": 0.9698942229454841, | |
| "number": 1837 | |
| }, | |
| "MISC": { | |
| "precision": 0.8497983870967742, | |
| "recall": 0.9143167028199566, | |
| "f1": 0.8808777429467085, | |
| "number": 922 | |
| }, | |
| "ORG": { | |
| "precision": 0.9019607843137255, | |
| "recall": 0.9261744966442953, | |
| "f1": 0.9139072847682119, | |
| "number": 1341 | |
| }, | |
| "PER": { | |
| "precision": 0.9522799575821845, | |
| "recall": 0.9750271444082519, | |
| "f1": 0.9635193133047211, | |
| "number": 1842 | |
| } | |
| }, | |
| "eval_runtime": 7.6218, | |
| "eval_samples_per_second": 426.41, | |
| "eval_steps_per_second": 53.4, | |
| "epoch": 8.0, | |
| "step": 14048 | |
| }, | |
| { | |
| "loss": 0.0042, | |
| "grad_norm": 0.15165293216705322, | |
| "learning_rate": 1.1742596810933942e-05, | |
| "epoch": 8.257403189066059, | |
| "step": 14500 | |
| }, | |
| { | |
| "loss": 0.0038, | |
| "grad_norm": 0.01020512543618679, | |
| "learning_rate": 1.1457858769931664e-05, | |
| "epoch": 8.542141230068337, | |
| "step": 15000 | |
| }, | |
| { | |
| "loss": 0.0028, | |
| "grad_norm": 0.00046127362293191254, | |
| "learning_rate": 1.1173120728929384e-05, | |
| "epoch": 8.826879271070615, | |
| "step": 15500 | |
| }, | |
| { | |
| "eval_loss": 0.11473368108272552, | |
| "eval_overall": { | |
| "precision": 0.9368072787427626, | |
| "recall": 0.9530461124200605, | |
| "f1": 0.944856928339034, | |
| "accuracy": 0.9860187201977983 | |
| }, | |
| "eval_per_label": { | |
| "LOC": { | |
| "precision": 0.9632034632034632, | |
| "recall": 0.9689711486118672, | |
| "f1": 0.966078697421981, | |
| "number": 1837 | |
| }, | |
| "MISC": { | |
| "precision": 0.8518145161290323, | |
| "recall": 0.9164859002169198, | |
| "f1": 0.8829676071055382, | |
| "number": 922 | |
| }, | |
| "ORG": { | |
| "precision": 0.9313207547169812, | |
| "recall": 0.9202087994034303, | |
| "f1": 0.9257314328582145, | |
| "number": 1341 | |
| }, | |
| "PER": { | |
| "precision": 0.9595744680851064, | |
| "recall": 0.9793702497285559, | |
| "f1": 0.969371305749597, | |
| "number": 1842 | |
| } | |
| }, | |
| "eval_runtime": 7.5249, | |
| "eval_samples_per_second": 431.9, | |
| "eval_steps_per_second": 54.087, | |
| "epoch": 9.0, | |
| "step": 15804 | |
| }, | |
| { | |
| "loss": 0.0049, | |
| "grad_norm": 0.17741906642913818, | |
| "learning_rate": 1.0888382687927108e-05, | |
| "epoch": 9.111617312072893, | |
| "step": 16000 | |
| }, | |
| { | |
| "loss": 0.003, | |
| "grad_norm": 0.0018998866435140371, | |
| "learning_rate": 1.060364464692483e-05, | |
| "epoch": 9.39635535307517, | |
| "step": 16500 | |
| }, | |
| { | |
| "loss": 0.0042, | |
| "grad_norm": 0.0021700740326195955, | |
| "learning_rate": 1.0318906605922552e-05, | |
| "epoch": 9.681093394077449, | |
| "step": 17000 | |
| }, | |
| { | |
| "loss": 0.0026, | |
| "grad_norm": 0.014039441011846066, | |
| "learning_rate": 1.0034168564920275e-05, | |
| "epoch": 9.965831435079727, | |
| "step": 17500 | |
| }, | |
| { | |
| "eval_loss": 0.1184961125254631, | |
| "eval_overall": { | |
| "precision": 0.9358889623265036, | |
| "recall": 0.9532144059239314, | |
| "f1": 0.9444722361180591, | |
| "accuracy": 0.9852681462294696 | |
| }, | |
| "eval_per_label": { | |
| "LOC": { | |
| "precision": 0.9669197396963124, | |
| "recall": 0.9706042460533478, | |
| "f1": 0.9687584895408857, | |
| "number": 1837 | |
| }, | |
| "MISC": { | |
| "precision": 0.8826638477801269, | |
| "recall": 0.9056399132321041, | |
| "f1": 0.8940042826552462, | |
| "number": 922 | |
| }, | |
| "ORG": { | |
| "precision": 0.8961318051575932, | |
| "recall": 0.9328859060402684, | |
| "f1": 0.9141395688710267, | |
| "number": 1341 | |
| }, | |
| "PER": { | |
| "precision": 0.9619506966773848, | |
| "recall": 0.9744842562432139, | |
| "f1": 0.9681769147788565, | |
| "number": 1842 | |
| } | |
| }, | |
| "eval_runtime": 7.5803, | |
| "eval_samples_per_second": 428.744, | |
| "eval_steps_per_second": 53.692, | |
| "epoch": 10.0, | |
| "step": 17560 | |
| }, | |
| { | |
| "loss": 0.0011, | |
| "grad_norm": 0.00040622701635584235, | |
| "learning_rate": 9.749430523917997e-06, | |
| "epoch": 10.250569476082005, | |
| "step": 18000 | |
| }, | |
| { | |
| "loss": 0.001, | |
| "grad_norm": 0.0004215097869746387, | |
| "learning_rate": 9.464692482915719e-06, | |
| "epoch": 10.535307517084282, | |
| "step": 18500 | |
| }, | |
| { | |
| "loss": 0.002, | |
| "grad_norm": 0.0020023963879793882, | |
| "learning_rate": 9.17995444191344e-06, | |
| "epoch": 10.82004555808656, | |
| "step": 19000 | |
| }, | |
| { | |
| "eval_loss": 0.11947210878133774, | |
| "eval_overall": { | |
| "precision": 0.9384717168375786, | |
| "recall": 0.9548973409626389, | |
| "f1": 0.9466132799466134, | |
| "accuracy": 0.9857685288750221 | |
| }, | |
| "eval_per_label": { | |
| "LOC": { | |
| "precision": 0.9658906334596643, | |
| "recall": 0.9711486118671747, | |
| "f1": 0.9685124864277959, | |
| "number": 1837 | |
| }, | |
| "MISC": { | |
| "precision": 0.8954108858057631, | |
| "recall": 0.9099783080260304, | |
| "f1": 0.9026358257127487, | |
| "number": 922 | |
| }, | |
| "ORG": { | |
| "precision": 0.9075812274368231, | |
| "recall": 0.9373601789709173, | |
| "f1": 0.922230374174615, | |
| "number": 1341 | |
| }, | |
| "PER": { | |
| "precision": 0.9557805007991476, | |
| "recall": 0.9739413680781759, | |
| "f1": 0.9647754772788384, | |
| "number": 1842 | |
| } | |
| }, | |
| "eval_runtime": 7.5556, | |
| "eval_samples_per_second": 430.142, | |
| "eval_steps_per_second": 53.867, | |
| "epoch": 11.0, | |
| "step": 19316 | |
| }, | |
| { | |
| "loss": 0.0028, | |
| "grad_norm": 0.030722877010703087, | |
| "learning_rate": 8.895216400911163e-06, | |
| "epoch": 11.104783599088838, | |
| "step": 19500 | |
| }, | |
| { | |
| "loss": 0.0033, | |
| "grad_norm": 0.004747629631310701, | |
| "learning_rate": 8.610478359908885e-06, | |
| "epoch": 11.389521640091116, | |
| "step": 20000 | |
| }, | |
| { | |
| "loss": 0.0018, | |
| "grad_norm": 0.0014466517604887486, | |
| "learning_rate": 8.325740318906607e-06, | |
| "epoch": 11.674259681093394, | |
| "step": 20500 | |
| }, | |
| { | |
| "loss": 0.0004, | |
| "grad_norm": 0.005812987219542265, | |
| "learning_rate": 8.041002277904329e-06, | |
| "epoch": 11.958997722095672, | |
| "step": 21000 | |
| }, | |
| { | |
| "eval_loss": 0.12146918475627899, | |
| "eval_overall": { | |
| "precision": 0.9431799302209669, | |
| "recall": 0.9554022214742511, | |
| "f1": 0.9492517348047823, | |
| "accuracy": 0.9866074056631542 | |
| }, | |
| "eval_per_label": { | |
| "LOC": { | |
| "precision": 0.971195652173913, | |
| "recall": 0.9727817093086554, | |
| "f1": 0.9719880337231438, | |
| "number": 1837 | |
| }, | |
| "MISC": { | |
| "precision": 0.9096844396082698, | |
| "recall": 0.9067245119305857, | |
| "f1": 0.9082020640956002, | |
| "number": 922 | |
| }, | |
| "ORG": { | |
| "precision": 0.9121522693997072, | |
| "recall": 0.9291573452647278, | |
| "f1": 0.9205762837089029, | |
| "number": 1341 | |
| }, | |
| "PER": { | |
| "precision": 0.9545934530095037, | |
| "recall": 0.9815418023887079, | |
| "f1": 0.9678800856531049, | |
| "number": 1842 | |
| } | |
| }, | |
| "eval_runtime": 7.7732, | |
| "eval_samples_per_second": 418.106, | |
| "eval_steps_per_second": 52.36, | |
| "epoch": 12.0, | |
| "step": 21072 | |
| }, | |
| { | |
| "loss": 0.0006, | |
| "grad_norm": 0.0022160038352012634, | |
| "learning_rate": 7.75626423690205e-06, | |
| "epoch": 12.24373576309795, | |
| "step": 21500 | |
| }, | |
| { | |
| "loss": 0.0008, | |
| "grad_norm": 0.00019432637782301754, | |
| "learning_rate": 7.471526195899773e-06, | |
| "epoch": 12.528473804100228, | |
| "step": 22000 | |
| }, | |
| { | |
| "loss": 0.0011, | |
| "grad_norm": 1.6397913694381714, | |
| "learning_rate": 7.186788154897495e-06, | |
| "epoch": 12.813211845102506, | |
| "step": 22500 | |
| }, | |
| { | |
| "eval_loss": 0.12505799531936646, | |
| "eval_overall": { | |
| "precision": 0.9386808087504143, | |
| "recall": 0.9532144059239314, | |
| "f1": 0.9458917835671342, | |
| "accuracy": 0.9860923058809677 | |
| }, | |
| "eval_per_label": { | |
| "LOC": { | |
| "precision": 0.9675148890092041, | |
| "recall": 0.9727817093086554, | |
| "f1": 0.9701411509229099, | |
| "number": 1837 | |
| }, | |
| "MISC": { | |
| "precision": 0.8782791185729276, | |
| "recall": 0.9078091106290672, | |
| "f1": 0.8927999999999999, | |
| "number": 922 | |
| }, | |
| "ORG": { | |
| "precision": 0.9241741741741741, | |
| "recall": 0.9179716629381058, | |
| "f1": 0.9210624766180322, | |
| "number": 1341 | |
| }, | |
| "PER": { | |
| "precision": 0.9511041009463722, | |
| "recall": 0.9820846905537459, | |
| "f1": 0.9663461538461537, | |
| "number": 1842 | |
| } | |
| }, | |
| "eval_runtime": 7.5826, | |
| "eval_samples_per_second": 428.613, | |
| "eval_steps_per_second": 53.676, | |
| "epoch": 13.0, | |
| "step": 22828 | |
| }, | |
| { | |
| "loss": 0.0015, | |
| "grad_norm": 0.00018138765881303698, | |
| "learning_rate": 6.9020501138952166e-06, | |
| "epoch": 13.097949886104784, | |
| "step": 23000 | |
| }, | |
| { | |
| "loss": 0.0007, | |
| "grad_norm": 0.00034025911008939147, | |
| "learning_rate": 6.617312072892939e-06, | |
| "epoch": 13.382687927107062, | |
| "step": 23500 | |
| }, | |
| { | |
| "loss": 0.0006, | |
| "grad_norm": 0.00041584973223507404, | |
| "learning_rate": 6.3325740318906616e-06, | |
| "epoch": 13.66742596810934, | |
| "step": 24000 | |
| }, | |
| { | |
| "loss": 0.0008, | |
| "grad_norm": 0.0006968477973714471, | |
| "learning_rate": 6.047835990888384e-06, | |
| "epoch": 13.952164009111618, | |
| "step": 24500 | |
| }, | |
| { | |
| "eval_loss": 0.12853111326694489, | |
| "eval_overall": { | |
| "precision": 0.9430043203722167, | |
| "recall": 0.9550656344665096, | |
| "f1": 0.9489966555183946, | |
| "accuracy": 0.986342497203744 | |
| }, | |
| "eval_per_label": { | |
| "LOC": { | |
| "precision": 0.9736986301369863, | |
| "recall": 0.9673380511703865, | |
| "f1": 0.9705079191698526, | |
| "number": 1837 | |
| }, | |
| "MISC": { | |
| "precision": 0.8892438764643238, | |
| "recall": 0.9056399132321041, | |
| "f1": 0.8973670069854918, | |
| "number": 922 | |
| }, | |
| "ORG": { | |
| "precision": 0.9200293470286134, | |
| "recall": 0.9351230425055929, | |
| "f1": 0.9275147928994084, | |
| "number": 1341 | |
| }, | |
| "PER": { | |
| "precision": 0.9566367001586462, | |
| "recall": 0.9820846905537459, | |
| "f1": 0.9691936780069649, | |
| "number": 1842 | |
| } | |
| }, | |
| "eval_runtime": 7.5688, | |
| "eval_samples_per_second": 429.396, | |
| "eval_steps_per_second": 53.774, | |
| "epoch": 14.0, | |
| "step": 24584 | |
| }, | |
| { | |
| "loss": 0.0006, | |
| "grad_norm": 0.00020477738871704787, | |
| "learning_rate": 5.763097949886105e-06, | |
| "epoch": 14.236902050113896, | |
| "step": 25000 | |
| }, | |
| { | |
| "loss": 0.0007, | |
| "grad_norm": 0.00032146400189958513, | |
| "learning_rate": 5.478359908883827e-06, | |
| "epoch": 14.521640091116174, | |
| "step": 25500 | |
| }, | |
| { | |
| "loss": 0.0014, | |
| "grad_norm": 0.000293695367872715, | |
| "learning_rate": 5.19362186788155e-06, | |
| "epoch": 14.806378132118452, | |
| "step": 26000 | |
| }, | |
| { | |
| "eval_loss": 0.1253676861524582, | |
| "eval_overall": { | |
| "precision": 0.938937613767996, | |
| "recall": 0.9548973409626389, | |
| "f1": 0.9468502294534834, | |
| "accuracy": 0.9861364572908695 | |
| }, | |
| "eval_per_label": { | |
| "LOC": { | |
| "precision": 0.9607948442534908, | |
| "recall": 0.9738704409363091, | |
| "f1": 0.9672884563395511, | |
| "number": 1837 | |
| }, | |
| "MISC": { | |
| "precision": 0.8742203742203742, | |
| "recall": 0.9121475054229935, | |
| "f1": 0.8927813163481952, | |
| "number": 922 | |
| }, | |
| "ORG": { | |
| "precision": 0.9322289156626506, | |
| "recall": 0.9231916480238628, | |
| "f1": 0.9276882727613338, | |
| "number": 1341 | |
| }, | |
| "PER": { | |
| "precision": 0.9550502379693284, | |
| "recall": 0.9804560260586319, | |
| "f1": 0.9675863916421109, | |
| "number": 1842 | |
| } | |
| }, | |
| "eval_runtime": 8.6163, | |
| "eval_samples_per_second": 377.191, | |
| "eval_steps_per_second": 47.236, | |
| "epoch": 15.0, | |
| "step": 26340 | |
| }, | |
| { | |
| "loss": 0.0008, | |
| "grad_norm": 0.00019819244334939867, | |
| "learning_rate": 4.908883826879272e-06, | |
| "epoch": 15.09111617312073, | |
| "step": 26500 | |
| }, | |
| { | |
| "loss": 0.0008, | |
| "grad_norm": 0.00021079520229250193, | |
| "learning_rate": 4.624145785876993e-06, | |
| "epoch": 15.375854214123008, | |
| "step": 27000 | |
| }, | |
| { | |
| "loss": 0.0003, | |
| "grad_norm": 0.00012124140630476177, | |
| "learning_rate": 4.339407744874715e-06, | |
| "epoch": 15.660592255125284, | |
| "step": 27500 | |
| }, | |
| { | |
| "loss": 0.0008, | |
| "grad_norm": 0.000182148942258209, | |
| "learning_rate": 4.054669703872437e-06, | |
| "epoch": 15.945330296127562, | |
| "step": 28000 | |
| }, | |
| { | |
| "eval_loss": 0.12416187673807144, | |
| "eval_overall": { | |
| "precision": 0.9447862963578912, | |
| "recall": 0.9560753954897341, | |
| "f1": 0.9503973232956922, | |
| "accuracy": 0.9866957084829575 | |
| }, | |
| "eval_per_label": { | |
| "LOC": { | |
| "precision": 0.9680043383947939, | |
| "recall": 0.9716929776810016, | |
| "f1": 0.969845150774246, | |
| "number": 1837 | |
| }, | |
| "MISC": { | |
| "precision": 0.8954108858057631, | |
| "recall": 0.9099783080260304, | |
| "f1": 0.9026358257127487, | |
| "number": 922 | |
| }, | |
| "ORG": { | |
| "precision": 0.9301634472511144, | |
| "recall": 0.9336316181953765, | |
| "f1": 0.9318943059173799, | |
| "number": 1341 | |
| }, | |
| "PER": { | |
| "precision": 0.957051961823966, | |
| "recall": 0.9799131378935939, | |
| "f1": 0.9683476394849786, | |
| "number": 1842 | |
| } | |
| }, | |
| "eval_runtime": 8.6628, | |
| "eval_samples_per_second": 375.168, | |
| "eval_steps_per_second": 46.983, | |
| "epoch": 16.0, | |
| "step": 28096 | |
| }, | |
| { | |
| "loss": 0.001, | |
| "grad_norm": 0.019741835072636604, | |
| "learning_rate": 3.76993166287016e-06, | |
| "epoch": 16.23006833712984, | |
| "step": 28500 | |
| }, | |
| { | |
| "loss": 0.0003, | |
| "grad_norm": 0.00012148160021752119, | |
| "learning_rate": 3.4851936218678815e-06, | |
| "epoch": 16.514806378132118, | |
| "step": 29000 | |
| }, | |
| { | |
| "loss": 0.0005, | |
| "grad_norm": 0.00011341737263137475, | |
| "learning_rate": 3.200455580865604e-06, | |
| "epoch": 16.799544419134396, | |
| "step": 29500 | |
| }, | |
| { | |
| "eval_loss": 0.12274094671010971, | |
| "eval_overall": { | |
| "precision": 0.9438631456568676, | |
| "recall": 0.9564119824974756, | |
| "f1": 0.9500961297333445, | |
| "accuracy": 0.9868428798492965 | |
| }, | |
| "eval_per_label": { | |
| "LOC": { | |
| "precision": 0.971195652173913, | |
| "recall": 0.9727817093086554, | |
| "f1": 0.9719880337231438, | |
| "number": 1837 | |
| }, | |
| "MISC": { | |
| "precision": 0.8761609907120743, | |
| "recall": 0.920824295010846, | |
| "f1": 0.8979375991538868, | |
| "number": 922 | |
| }, | |
| "ORG": { | |
| "precision": 0.9307004470938898, | |
| "recall": 0.9313944817300522, | |
| "f1": 0.9310473350726798, | |
| "number": 1341 | |
| }, | |
| "PER": { | |
| "precision": 0.9614973262032086, | |
| "recall": 0.9761129207383279, | |
| "f1": 0.9687500000000001, | |
| "number": 1842 | |
| } | |
| }, | |
| "eval_runtime": 8.9038, | |
| "eval_samples_per_second": 365.012, | |
| "eval_steps_per_second": 45.711, | |
| "epoch": 17.0, | |
| "step": 29852 | |
| }, | |
| { | |
| "loss": 0.0007, | |
| "grad_norm": 0.00015434053784701973, | |
| "learning_rate": 2.9157175398633257e-06, | |
| "epoch": 17.084282460136674, | |
| "step": 30000 | |
| }, | |
| { | |
| "loss": 0.0, | |
| "grad_norm": 0.00021848917822353542, | |
| "learning_rate": 2.6309794988610482e-06, | |
| "epoch": 17.36902050113895, | |
| "step": 30500 | |
| }, | |
| { | |
| "loss": 0.0002, | |
| "grad_norm": 0.0001007779865176417, | |
| "learning_rate": 2.34624145785877e-06, | |
| "epoch": 17.65375854214123, | |
| "step": 31000 | |
| }, | |
| { | |
| "loss": 0.0002, | |
| "grad_norm": 0.00010259783448418602, | |
| "learning_rate": 2.061503416856492e-06, | |
| "epoch": 17.938496583143507, | |
| "step": 31500 | |
| }, | |
| { | |
| "eval_loss": 0.12293359637260437, | |
| "eval_overall": { | |
| "precision": 0.9440199335548173, | |
| "recall": 0.9564119824974756, | |
| "f1": 0.9501755559271025, | |
| "accuracy": 0.9870636368988049 | |
| }, | |
| "eval_per_label": { | |
| "LOC": { | |
| "precision": 0.9701249321021184, | |
| "recall": 0.9722373434948285, | |
| "f1": 0.9711799891245242, | |
| "number": 1837 | |
| }, | |
| "MISC": { | |
| "precision": 0.8810020876826722, | |
| "recall": 0.9154013015184381, | |
| "f1": 0.8978723404255319, | |
| "number": 922 | |
| }, | |
| "ORG": { | |
| "precision": 0.926829268292683, | |
| "recall": 0.9351230425055929, | |
| "f1": 0.9309576837416481, | |
| "number": 1341 | |
| }, | |
| "PER": { | |
| "precision": 0.9630620985010707, | |
| "recall": 0.9766558089033659, | |
| "f1": 0.969811320754717, | |
| "number": 1842 | |
| } | |
| }, | |
| "eval_runtime": 8.7313, | |
| "eval_samples_per_second": 372.226, | |
| "eval_steps_per_second": 46.614, | |
| "epoch": 18.0, | |
| "step": 31608 | |
| }, | |
| { | |
| "loss": 0.0001, | |
| "grad_norm": 0.0010326108895242214, | |
| "learning_rate": 1.7767653758542143e-06, | |
| "epoch": 18.223234624145785, | |
| "step": 32000 | |
| }, | |
| { | |
| "loss": 0.0007, | |
| "grad_norm": 0.00016551795124541968, | |
| "learning_rate": 1.4920273348519363e-06, | |
| "epoch": 18.507972665148063, | |
| "step": 32500 | |
| }, | |
| { | |
| "loss": 0.0004, | |
| "grad_norm": 0.00024008983746170998, | |
| "learning_rate": 1.2072892938496584e-06, | |
| "epoch": 18.79271070615034, | |
| "step": 33000 | |
| }, | |
| { | |
| "eval_loss": 0.12236841022968292, | |
| "eval_overall": { | |
| "precision": 0.944813829787234, | |
| "recall": 0.9565802760013463, | |
| "f1": 0.9506606455929084, | |
| "accuracy": 0.9872255254017779 | |
| }, | |
| "eval_per_label": { | |
| "LOC": { | |
| "precision": 0.971677559912854, | |
| "recall": 0.9711486118671747, | |
| "f1": 0.9714130138851075, | |
| "number": 1837 | |
| }, | |
| "MISC": { | |
| "precision": 0.8712667353244078, | |
| "recall": 0.9175704989154013, | |
| "f1": 0.8938193343898574, | |
| "number": 922 | |
| }, | |
| "ORG": { | |
| "precision": 0.9383458646616541, | |
| "recall": 0.930648769574944, | |
| "f1": 0.9344814676151254, | |
| "number": 1341 | |
| }, | |
| "PER": { | |
| "precision": 0.961149547631719, | |
| "recall": 0.9804560260586319, | |
| "f1": 0.9707067992475142, | |
| "number": 1842 | |
| } | |
| }, | |
| "eval_runtime": 6.5081, | |
| "eval_samples_per_second": 499.381, | |
| "eval_steps_per_second": 62.538, | |
| "epoch": 19.0, | |
| "step": 33364 | |
| }, | |
| { | |
| "loss": 0.0001, | |
| "grad_norm": 0.00016071839490905404, | |
| "learning_rate": 9.225512528473805e-07, | |
| "epoch": 19.07744874715262, | |
| "step": 33500 | |
| }, | |
| { | |
| "loss": 0.0001, | |
| "grad_norm": 9.665234392741695e-05, | |
| "learning_rate": 6.378132118451026e-07, | |
| "epoch": 19.362186788154897, | |
| "step": 34000 | |
| }, | |
| { | |
| "loss": 0.0001, | |
| "grad_norm": 0.004949676804244518, | |
| "learning_rate": 3.530751708428246e-07, | |
| "epoch": 19.646924829157175, | |
| "step": 34500 | |
| }, | |
| { | |
| "loss": 0.0005, | |
| "grad_norm": 0.00014002641546539962, | |
| "learning_rate": 6.83371298405467e-08, | |
| "epoch": 19.931662870159453, | |
| "step": 35000 | |
| }, | |
| { | |
| "eval_loss": 0.12258101999759674, | |
| "eval_overall": { | |
| "precision": 0.9443891102257637, | |
| "recall": 0.9574217435207001, | |
| "f1": 0.9508607721878657, | |
| "accuracy": 0.9872549596750456 | |
| }, | |
| "eval_per_label": { | |
| "LOC": { | |
| "precision": 0.9727371864776445, | |
| "recall": 0.9711486118671747, | |
| "f1": 0.9719422500681013, | |
| "number": 1837 | |
| }, | |
| "MISC": { | |
| "precision": 0.8746113989637305, | |
| "recall": 0.9154013015184381, | |
| "f1": 0.8945416004239534, | |
| "number": 922 | |
| }, | |
| "ORG": { | |
| "precision": 0.9336810730253353, | |
| "recall": 0.9343773303504848, | |
| "f1": 0.9340290719344019, | |
| "number": 1341 | |
| }, | |
| "PER": { | |
| "precision": 0.960169941582581, | |
| "recall": 0.9815418023887079, | |
| "f1": 0.970738255033557, | |
| "number": 1842 | |
| } | |
| }, | |
| "eval_runtime": 6.7405, | |
| "eval_samples_per_second": 482.16, | |
| "eval_steps_per_second": 60.381, | |
| "epoch": 20.0, | |
| "step": 35120 | |
| }, | |
| { | |
| "train_runtime": 3604.2086, | |
| "train_samples_per_second": 77.914, | |
| "train_steps_per_second": 9.744, | |
| "total_flos": 6145780901050062.0, | |
| "train_loss": 0.013757273165407275, | |
| "epoch": 20.0, | |
| "step": 35120 | |
| } | |
| ] |