BERT-jigsaw-severetoxic / trainer_state.json
Cameron Clarke
add model files
f0eaac5
{
"best_metric": 0.0,
"best_model_checkpoint": "/scratch/ccc779/jigsaw_severetoxic_model/checkpoint-59839",
"epoch": 2.0,
"global_step": 119678,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 1.2130648541044402e-05,
"loss": 0.0643,
"step": 500
},
{
"epoch": 0.02,
"learning_rate": 1.2079755555170145e-05,
"loss": 0.0878,
"step": 1000
},
{
"epoch": 0.03,
"learning_rate": 1.2028862569295889e-05,
"loss": 0.0959,
"step": 1500
},
{
"epoch": 0.03,
"learning_rate": 1.1977969583421632e-05,
"loss": 0.0328,
"step": 2000
},
{
"epoch": 0.04,
"learning_rate": 1.1927076597547376e-05,
"loss": 0.0508,
"step": 2500
},
{
"epoch": 0.05,
"learning_rate": 1.1876183611673118e-05,
"loss": 0.0522,
"step": 3000
},
{
"epoch": 0.06,
"learning_rate": 1.182529062579886e-05,
"loss": 0.0431,
"step": 3500
},
{
"epoch": 0.07,
"learning_rate": 1.1774397639924603e-05,
"loss": 0.0535,
"step": 4000
},
{
"epoch": 0.08,
"learning_rate": 1.1723504654050347e-05,
"loss": 0.0679,
"step": 4500
},
{
"epoch": 0.08,
"learning_rate": 1.167261166817609e-05,
"loss": 0.0714,
"step": 5000
},
{
"epoch": 0.09,
"learning_rate": 1.1621718682301832e-05,
"loss": 0.0482,
"step": 5500
},
{
"epoch": 0.1,
"learning_rate": 1.1570825696427575e-05,
"loss": 0.0648,
"step": 6000
},
{
"epoch": 0.11,
"learning_rate": 1.1519932710553319e-05,
"loss": 0.0569,
"step": 6500
},
{
"epoch": 0.12,
"learning_rate": 1.1469039724679063e-05,
"loss": 0.0393,
"step": 7000
},
{
"epoch": 0.13,
"learning_rate": 1.1418146738804806e-05,
"loss": 0.058,
"step": 7500
},
{
"epoch": 0.13,
"learning_rate": 1.1367253752930548e-05,
"loss": 0.0459,
"step": 8000
},
{
"epoch": 0.14,
"learning_rate": 1.131636076705629e-05,
"loss": 0.0565,
"step": 8500
},
{
"epoch": 0.15,
"learning_rate": 1.1265467781182033e-05,
"loss": 0.049,
"step": 9000
},
{
"epoch": 0.16,
"learning_rate": 1.1214574795307776e-05,
"loss": 0.0377,
"step": 9500
},
{
"epoch": 0.17,
"learning_rate": 1.116368180943352e-05,
"loss": 0.0548,
"step": 10000
},
{
"epoch": 0.18,
"learning_rate": 1.1112788823559262e-05,
"loss": 0.0944,
"step": 10500
},
{
"epoch": 0.18,
"learning_rate": 1.1061895837685007e-05,
"loss": 0.045,
"step": 11000
},
{
"epoch": 0.19,
"learning_rate": 1.1011002851810749e-05,
"loss": 0.0561,
"step": 11500
},
{
"epoch": 0.2,
"learning_rate": 1.0960109865936491e-05,
"loss": 0.0707,
"step": 12000
},
{
"epoch": 0.21,
"learning_rate": 1.0909216880062236e-05,
"loss": 0.0581,
"step": 12500
},
{
"epoch": 0.22,
"learning_rate": 1.0858323894187978e-05,
"loss": 0.0794,
"step": 13000
},
{
"epoch": 0.23,
"learning_rate": 1.080743090831372e-05,
"loss": 0.0863,
"step": 13500
},
{
"epoch": 0.23,
"learning_rate": 1.0756537922439463e-05,
"loss": 0.0733,
"step": 14000
},
{
"epoch": 0.24,
"learning_rate": 1.0705644936565207e-05,
"loss": 0.0425,
"step": 14500
},
{
"epoch": 0.25,
"learning_rate": 1.065475195069095e-05,
"loss": 0.065,
"step": 15000
},
{
"epoch": 0.26,
"learning_rate": 1.0603858964816694e-05,
"loss": 0.0392,
"step": 15500
},
{
"epoch": 0.27,
"learning_rate": 1.0552965978942436e-05,
"loss": 0.065,
"step": 16000
},
{
"epoch": 0.28,
"learning_rate": 1.0502072993068179e-05,
"loss": 0.0679,
"step": 16500
},
{
"epoch": 0.28,
"learning_rate": 1.0451180007193921e-05,
"loss": 0.0537,
"step": 17000
},
{
"epoch": 0.29,
"learning_rate": 1.0400287021319664e-05,
"loss": 0.0372,
"step": 17500
},
{
"epoch": 0.3,
"learning_rate": 1.0349394035445408e-05,
"loss": 0.0705,
"step": 18000
},
{
"epoch": 0.31,
"learning_rate": 1.029850104957115e-05,
"loss": 0.0771,
"step": 18500
},
{
"epoch": 0.32,
"learning_rate": 1.0247608063696895e-05,
"loss": 0.0689,
"step": 19000
},
{
"epoch": 0.33,
"learning_rate": 1.0196715077822637e-05,
"loss": 0.0722,
"step": 19500
},
{
"epoch": 0.33,
"learning_rate": 1.014582209194838e-05,
"loss": 0.0495,
"step": 20000
},
{
"epoch": 0.34,
"learning_rate": 1.0094929106074122e-05,
"loss": 0.0653,
"step": 20500
},
{
"epoch": 0.35,
"learning_rate": 1.0044036120199866e-05,
"loss": 0.0589,
"step": 21000
},
{
"epoch": 0.36,
"learning_rate": 9.993143134325609e-06,
"loss": 0.0655,
"step": 21500
},
{
"epoch": 0.37,
"learning_rate": 9.942250148451351e-06,
"loss": 0.0524,
"step": 22000
},
{
"epoch": 0.38,
"learning_rate": 9.891357162577094e-06,
"loss": 0.0801,
"step": 22500
},
{
"epoch": 0.38,
"learning_rate": 9.840464176702838e-06,
"loss": 0.061,
"step": 23000
},
{
"epoch": 0.39,
"learning_rate": 9.789571190828582e-06,
"loss": 0.0738,
"step": 23500
},
{
"epoch": 0.4,
"learning_rate": 9.738678204954325e-06,
"loss": 0.0558,
"step": 24000
},
{
"epoch": 0.41,
"learning_rate": 9.687785219080067e-06,
"loss": 0.0557,
"step": 24500
},
{
"epoch": 0.42,
"learning_rate": 9.63689223320581e-06,
"loss": 0.1104,
"step": 25000
},
{
"epoch": 0.43,
"learning_rate": 9.585999247331552e-06,
"loss": 0.0464,
"step": 25500
},
{
"epoch": 0.43,
"learning_rate": 9.535106261457295e-06,
"loss": 0.0692,
"step": 26000
},
{
"epoch": 0.44,
"learning_rate": 9.484213275583039e-06,
"loss": 0.063,
"step": 26500
},
{
"epoch": 0.45,
"learning_rate": 9.433320289708781e-06,
"loss": 0.0671,
"step": 27000
},
{
"epoch": 0.46,
"learning_rate": 9.382427303834525e-06,
"loss": 0.0622,
"step": 27500
},
{
"epoch": 0.47,
"learning_rate": 9.331534317960268e-06,
"loss": 0.0488,
"step": 28000
},
{
"epoch": 0.48,
"learning_rate": 9.28064133208601e-06,
"loss": 0.1113,
"step": 28500
},
{
"epoch": 0.48,
"learning_rate": 9.229748346211755e-06,
"loss": 0.0815,
"step": 29000
},
{
"epoch": 0.49,
"learning_rate": 9.178855360337497e-06,
"loss": 0.0589,
"step": 29500
},
{
"epoch": 0.5,
"learning_rate": 9.12796237446324e-06,
"loss": 0.0609,
"step": 30000
},
{
"epoch": 0.51,
"learning_rate": 9.077069388588982e-06,
"loss": 0.0898,
"step": 30500
},
{
"epoch": 0.52,
"learning_rate": 9.026176402714725e-06,
"loss": 0.0247,
"step": 31000
},
{
"epoch": 0.53,
"learning_rate": 8.975283416840469e-06,
"loss": 0.0581,
"step": 31500
},
{
"epoch": 0.53,
"learning_rate": 8.924390430966213e-06,
"loss": 0.0489,
"step": 32000
},
{
"epoch": 0.54,
"learning_rate": 8.873497445091955e-06,
"loss": 0.1036,
"step": 32500
},
{
"epoch": 0.55,
"learning_rate": 8.822604459217698e-06,
"loss": 0.0634,
"step": 33000
},
{
"epoch": 0.56,
"learning_rate": 8.77171147334344e-06,
"loss": 0.0456,
"step": 33500
},
{
"epoch": 0.57,
"learning_rate": 8.720818487469183e-06,
"loss": 0.0903,
"step": 34000
},
{
"epoch": 0.58,
"learning_rate": 8.669925501594925e-06,
"loss": 0.1017,
"step": 34500
},
{
"epoch": 0.58,
"learning_rate": 8.61903251572067e-06,
"loss": 0.0675,
"step": 35000
},
{
"epoch": 0.59,
"learning_rate": 8.568139529846412e-06,
"loss": 0.0473,
"step": 35500
},
{
"epoch": 0.6,
"learning_rate": 8.517246543972156e-06,
"loss": 0.0504,
"step": 36000
},
{
"epoch": 0.61,
"learning_rate": 8.466353558097899e-06,
"loss": 0.0489,
"step": 36500
},
{
"epoch": 0.62,
"learning_rate": 8.415460572223641e-06,
"loss": 0.0622,
"step": 37000
},
{
"epoch": 0.63,
"learning_rate": 8.364567586349385e-06,
"loss": 0.0454,
"step": 37500
},
{
"epoch": 0.64,
"learning_rate": 8.313674600475128e-06,
"loss": 0.0253,
"step": 38000
},
{
"epoch": 0.64,
"learning_rate": 8.26278161460087e-06,
"loss": 0.0592,
"step": 38500
},
{
"epoch": 0.65,
"learning_rate": 8.211888628726613e-06,
"loss": 0.0734,
"step": 39000
},
{
"epoch": 0.66,
"learning_rate": 8.160995642852357e-06,
"loss": 0.0595,
"step": 39500
},
{
"epoch": 0.67,
"learning_rate": 8.1101026569781e-06,
"loss": 0.0375,
"step": 40000
},
{
"epoch": 0.68,
"learning_rate": 8.059209671103844e-06,
"loss": 0.0542,
"step": 40500
},
{
"epoch": 0.69,
"learning_rate": 8.008316685229586e-06,
"loss": 0.0207,
"step": 41000
},
{
"epoch": 0.69,
"learning_rate": 7.957423699355329e-06,
"loss": 0.0493,
"step": 41500
},
{
"epoch": 0.7,
"learning_rate": 7.906530713481071e-06,
"loss": 0.0396,
"step": 42000
},
{
"epoch": 0.71,
"learning_rate": 7.855637727606814e-06,
"loss": 0.0734,
"step": 42500
},
{
"epoch": 0.72,
"learning_rate": 7.804744741732558e-06,
"loss": 0.0744,
"step": 43000
},
{
"epoch": 0.73,
"learning_rate": 7.7538517558583e-06,
"loss": 0.0579,
"step": 43500
},
{
"epoch": 0.74,
"learning_rate": 7.702958769984044e-06,
"loss": 0.0528,
"step": 44000
},
{
"epoch": 0.74,
"learning_rate": 7.652065784109787e-06,
"loss": 0.052,
"step": 44500
},
{
"epoch": 0.75,
"learning_rate": 7.601172798235529e-06,
"loss": 0.0654,
"step": 45000
},
{
"epoch": 0.76,
"learning_rate": 7.550279812361272e-06,
"loss": 0.0469,
"step": 45500
},
{
"epoch": 0.77,
"learning_rate": 7.499386826487016e-06,
"loss": 0.0457,
"step": 46000
},
{
"epoch": 0.78,
"learning_rate": 7.4484938406127585e-06,
"loss": 0.0665,
"step": 46500
},
{
"epoch": 0.79,
"learning_rate": 7.397600854738501e-06,
"loss": 0.0464,
"step": 47000
},
{
"epoch": 0.79,
"learning_rate": 7.346707868864244e-06,
"loss": 0.0581,
"step": 47500
},
{
"epoch": 0.8,
"learning_rate": 7.295814882989987e-06,
"loss": 0.0382,
"step": 48000
},
{
"epoch": 0.81,
"learning_rate": 7.244921897115731e-06,
"loss": 0.0675,
"step": 48500
},
{
"epoch": 0.82,
"learning_rate": 7.1940289112414734e-06,
"loss": 0.073,
"step": 49000
},
{
"epoch": 0.83,
"learning_rate": 7.143135925367217e-06,
"loss": 0.052,
"step": 49500
},
{
"epoch": 0.84,
"learning_rate": 7.092242939492959e-06,
"loss": 0.0752,
"step": 50000
},
{
"epoch": 0.84,
"learning_rate": 7.041349953618702e-06,
"loss": 0.0699,
"step": 50500
},
{
"epoch": 0.85,
"learning_rate": 6.990456967744445e-06,
"loss": 0.0898,
"step": 51000
},
{
"epoch": 0.86,
"learning_rate": 6.939563981870188e-06,
"loss": 0.067,
"step": 51500
},
{
"epoch": 0.87,
"learning_rate": 6.888670995995932e-06,
"loss": 0.0898,
"step": 52000
},
{
"epoch": 0.88,
"learning_rate": 6.837778010121674e-06,
"loss": 0.0536,
"step": 52500
},
{
"epoch": 0.89,
"learning_rate": 6.786885024247417e-06,
"loss": 0.0622,
"step": 53000
},
{
"epoch": 0.89,
"learning_rate": 6.73599203837316e-06,
"loss": 0.0901,
"step": 53500
},
{
"epoch": 0.9,
"learning_rate": 6.685099052498904e-06,
"loss": 0.0474,
"step": 54000
},
{
"epoch": 0.91,
"learning_rate": 6.634206066624647e-06,
"loss": 0.0783,
"step": 54500
},
{
"epoch": 0.92,
"learning_rate": 6.583313080750389e-06,
"loss": 0.1103,
"step": 55000
},
{
"epoch": 0.93,
"learning_rate": 6.5324200948761325e-06,
"loss": 0.0861,
"step": 55500
},
{
"epoch": 0.94,
"learning_rate": 6.481527109001875e-06,
"loss": 0.0615,
"step": 56000
},
{
"epoch": 0.94,
"learning_rate": 6.4306341231276175e-06,
"loss": 0.0398,
"step": 56500
},
{
"epoch": 0.95,
"learning_rate": 6.379741137253362e-06,
"loss": 0.0847,
"step": 57000
},
{
"epoch": 0.96,
"learning_rate": 6.328848151379104e-06,
"loss": 0.0403,
"step": 57500
},
{
"epoch": 0.97,
"learning_rate": 6.2779551655048475e-06,
"loss": 0.1037,
"step": 58000
},
{
"epoch": 0.98,
"learning_rate": 6.22706217963059e-06,
"loss": 0.1065,
"step": 58500
},
{
"epoch": 0.99,
"learning_rate": 6.1761691937563324e-06,
"loss": 0.1039,
"step": 59000
},
{
"epoch": 0.99,
"learning_rate": 6.125276207882077e-06,
"loss": 0.0597,
"step": 59500
},
{
"epoch": 1.0,
"eval_accuracy": 0.989822775925601,
"eval_f1": 0.0,
"eval_loss": 0.07756483554840088,
"eval_precision": 0.0,
"eval_recall": 0.0,
"eval_runtime": 233.8941,
"eval_samples_per_second": 170.56,
"step": 59839
},
{
"epoch": 1.0,
"learning_rate": 6.074383222007819e-06,
"loss": 0.0682,
"step": 60000
},
{
"epoch": 1.01,
"learning_rate": 6.0234902361335624e-06,
"loss": 0.0781,
"step": 60500
},
{
"epoch": 1.02,
"learning_rate": 5.972597250259305e-06,
"loss": 0.1027,
"step": 61000
},
{
"epoch": 1.03,
"learning_rate": 5.921704264385048e-06,
"loss": 0.0871,
"step": 61500
},
{
"epoch": 1.04,
"learning_rate": 5.870811278510792e-06,
"loss": 0.0525,
"step": 62000
},
{
"epoch": 1.04,
"learning_rate": 5.819918292636534e-06,
"loss": 0.0616,
"step": 62500
},
{
"epoch": 1.05,
"learning_rate": 5.769025306762277e-06,
"loss": 0.0599,
"step": 63000
},
{
"epoch": 1.06,
"learning_rate": 5.71813232088802e-06,
"loss": 0.0296,
"step": 63500
},
{
"epoch": 1.07,
"learning_rate": 5.667239335013763e-06,
"loss": 0.0515,
"step": 64000
},
{
"epoch": 1.08,
"learning_rate": 5.6163463491395066e-06,
"loss": 0.0436,
"step": 64500
},
{
"epoch": 1.09,
"learning_rate": 5.565453363265249e-06,
"loss": 0.0819,
"step": 65000
},
{
"epoch": 1.09,
"learning_rate": 5.5145603773909915e-06,
"loss": 0.0451,
"step": 65500
},
{
"epoch": 1.1,
"learning_rate": 5.463667391516736e-06,
"loss": 0.0485,
"step": 66000
},
{
"epoch": 1.11,
"learning_rate": 5.412774405642478e-06,
"loss": 0.0714,
"step": 66500
},
{
"epoch": 1.12,
"learning_rate": 5.3618814197682215e-06,
"loss": 0.0573,
"step": 67000
},
{
"epoch": 1.13,
"learning_rate": 5.310988433893964e-06,
"loss": 0.0972,
"step": 67500
},
{
"epoch": 1.14,
"learning_rate": 5.260095448019707e-06,
"loss": 0.0515,
"step": 68000
},
{
"epoch": 1.14,
"learning_rate": 5.209202462145451e-06,
"loss": 0.0446,
"step": 68500
},
{
"epoch": 1.15,
"learning_rate": 5.158309476271193e-06,
"loss": 0.0427,
"step": 69000
},
{
"epoch": 1.16,
"learning_rate": 5.107416490396936e-06,
"loss": 0.0566,
"step": 69500
},
{
"epoch": 1.17,
"learning_rate": 5.056523504522679e-06,
"loss": 0.0424,
"step": 70000
},
{
"epoch": 1.18,
"learning_rate": 5.005630518648422e-06,
"loss": 0.0746,
"step": 70500
},
{
"epoch": 1.19,
"learning_rate": 4.954737532774165e-06,
"loss": 0.0207,
"step": 71000
},
{
"epoch": 1.19,
"learning_rate": 4.903844546899908e-06,
"loss": 0.0351,
"step": 71500
},
{
"epoch": 1.2,
"learning_rate": 4.852951561025651e-06,
"loss": 0.0521,
"step": 72000
},
{
"epoch": 1.21,
"learning_rate": 4.802058575151395e-06,
"loss": 0.0747,
"step": 72500
},
{
"epoch": 1.22,
"learning_rate": 4.751165589277137e-06,
"loss": 0.0685,
"step": 73000
},
{
"epoch": 1.23,
"learning_rate": 4.70027260340288e-06,
"loss": 0.0461,
"step": 73500
},
{
"epoch": 1.24,
"learning_rate": 4.649379617528623e-06,
"loss": 0.0988,
"step": 74000
},
{
"epoch": 1.25,
"learning_rate": 4.598486631654366e-06,
"loss": 0.0502,
"step": 74500
},
{
"epoch": 1.25,
"learning_rate": 4.547593645780109e-06,
"loss": 0.0651,
"step": 75000
},
{
"epoch": 1.26,
"learning_rate": 4.496700659905852e-06,
"loss": 0.0731,
"step": 75500
},
{
"epoch": 1.27,
"learning_rate": 4.445807674031595e-06,
"loss": 0.0734,
"step": 76000
},
{
"epoch": 1.28,
"learning_rate": 4.394914688157338e-06,
"loss": 0.0377,
"step": 76500
},
{
"epoch": 1.29,
"learning_rate": 4.344021702283081e-06,
"loss": 0.0667,
"step": 77000
},
{
"epoch": 1.3,
"learning_rate": 4.293128716408824e-06,
"loss": 0.0657,
"step": 77500
},
{
"epoch": 1.3,
"learning_rate": 4.242235730534567e-06,
"loss": 0.0747,
"step": 78000
},
{
"epoch": 1.31,
"learning_rate": 4.1913427446603105e-06,
"loss": 0.0481,
"step": 78500
},
{
"epoch": 1.32,
"learning_rate": 4.140449758786053e-06,
"loss": 0.0489,
"step": 79000
},
{
"epoch": 1.33,
"learning_rate": 4.089556772911796e-06,
"loss": 0.029,
"step": 79500
},
{
"epoch": 1.34,
"learning_rate": 4.038663787037539e-06,
"loss": 0.0818,
"step": 80000
},
{
"epoch": 1.35,
"learning_rate": 3.987770801163282e-06,
"loss": 0.0396,
"step": 80500
},
{
"epoch": 1.35,
"learning_rate": 3.9368778152890255e-06,
"loss": 0.0758,
"step": 81000
},
{
"epoch": 1.36,
"learning_rate": 3.885984829414768e-06,
"loss": 0.0597,
"step": 81500
},
{
"epoch": 1.37,
"learning_rate": 3.8350918435405104e-06,
"loss": 0.0827,
"step": 82000
},
{
"epoch": 1.38,
"learning_rate": 3.784198857666254e-06,
"loss": 0.0465,
"step": 82500
},
{
"epoch": 1.39,
"learning_rate": 3.7333058717919967e-06,
"loss": 0.1027,
"step": 83000
},
{
"epoch": 1.4,
"learning_rate": 3.6824128859177396e-06,
"loss": 0.0399,
"step": 83500
},
{
"epoch": 1.4,
"learning_rate": 3.631519900043483e-06,
"loss": 0.0771,
"step": 84000
},
{
"epoch": 1.41,
"learning_rate": 3.580626914169226e-06,
"loss": 0.0592,
"step": 84500
},
{
"epoch": 1.42,
"learning_rate": 3.529733928294969e-06,
"loss": 0.0805,
"step": 85000
},
{
"epoch": 1.43,
"learning_rate": 3.478840942420712e-06,
"loss": 0.0478,
"step": 85500
},
{
"epoch": 1.44,
"learning_rate": 3.4279479565464546e-06,
"loss": 0.0633,
"step": 86000
},
{
"epoch": 1.45,
"learning_rate": 3.3770549706721983e-06,
"loss": 0.088,
"step": 86500
},
{
"epoch": 1.45,
"learning_rate": 3.326161984797941e-06,
"loss": 0.0506,
"step": 87000
},
{
"epoch": 1.46,
"learning_rate": 3.2752689989236837e-06,
"loss": 0.0905,
"step": 87500
},
{
"epoch": 1.47,
"learning_rate": 3.224376013049427e-06,
"loss": 0.0395,
"step": 88000
},
{
"epoch": 1.48,
"learning_rate": 3.17348302717517e-06,
"loss": 0.05,
"step": 88500
},
{
"epoch": 1.49,
"learning_rate": 3.1225900413009124e-06,
"loss": 0.0812,
"step": 89000
},
{
"epoch": 1.5,
"learning_rate": 3.071697055426656e-06,
"loss": 0.0501,
"step": 89500
},
{
"epoch": 1.5,
"learning_rate": 3.0208040695523987e-06,
"loss": 0.0635,
"step": 90000
},
{
"epoch": 1.51,
"learning_rate": 2.969911083678142e-06,
"loss": 0.1061,
"step": 90500
},
{
"epoch": 1.52,
"learning_rate": 2.919018097803885e-06,
"loss": 0.0743,
"step": 91000
},
{
"epoch": 1.53,
"learning_rate": 2.868125111929628e-06,
"loss": 0.0776,
"step": 91500
},
{
"epoch": 1.54,
"learning_rate": 2.8172321260553707e-06,
"loss": 0.0753,
"step": 92000
},
{
"epoch": 1.55,
"learning_rate": 2.7663391401811136e-06,
"loss": 0.0745,
"step": 92500
},
{
"epoch": 1.55,
"learning_rate": 2.715446154306857e-06,
"loss": 0.0534,
"step": 93000
},
{
"epoch": 1.56,
"learning_rate": 2.6645531684326e-06,
"loss": 0.0882,
"step": 93500
},
{
"epoch": 1.57,
"learning_rate": 2.6136601825583428e-06,
"loss": 0.0864,
"step": 94000
},
{
"epoch": 1.58,
"learning_rate": 2.5627671966840857e-06,
"loss": 0.0671,
"step": 94500
},
{
"epoch": 1.59,
"learning_rate": 2.511874210809829e-06,
"loss": 0.0446,
"step": 95000
},
{
"epoch": 1.6,
"learning_rate": 2.4609812249355715e-06,
"loss": 0.045,
"step": 95500
},
{
"epoch": 1.6,
"learning_rate": 2.410088239061315e-06,
"loss": 0.0581,
"step": 96000
},
{
"epoch": 1.61,
"learning_rate": 2.3591952531870577e-06,
"loss": 0.047,
"step": 96500
},
{
"epoch": 1.62,
"learning_rate": 2.308302267312801e-06,
"loss": 0.0367,
"step": 97000
},
{
"epoch": 1.63,
"learning_rate": 2.2574092814385436e-06,
"loss": 0.0394,
"step": 97500
},
{
"epoch": 1.64,
"learning_rate": 2.206516295564287e-06,
"loss": 0.0833,
"step": 98000
},
{
"epoch": 1.65,
"learning_rate": 2.15562330969003e-06,
"loss": 0.0566,
"step": 98500
},
{
"epoch": 1.65,
"learning_rate": 2.1047303238157727e-06,
"loss": 0.0211,
"step": 99000
},
{
"epoch": 1.66,
"learning_rate": 2.0538373379415156e-06,
"loss": 0.0526,
"step": 99500
},
{
"epoch": 1.67,
"learning_rate": 2.002944352067259e-06,
"loss": 0.0358,
"step": 100000
},
{
"epoch": 1.68,
"learning_rate": 1.952051366193002e-06,
"loss": 0.102,
"step": 100500
},
{
"epoch": 1.69,
"learning_rate": 1.9011583803187445e-06,
"loss": 0.0934,
"step": 101000
},
{
"epoch": 1.7,
"learning_rate": 1.8502653944444877e-06,
"loss": 0.0299,
"step": 101500
},
{
"epoch": 1.7,
"learning_rate": 1.7993724085702308e-06,
"loss": 0.0945,
"step": 102000
},
{
"epoch": 1.71,
"learning_rate": 1.748479422695974e-06,
"loss": 0.1056,
"step": 102500
},
{
"epoch": 1.72,
"learning_rate": 1.6975864368217166e-06,
"loss": 0.0283,
"step": 103000
},
{
"epoch": 1.73,
"learning_rate": 1.6466934509474597e-06,
"loss": 0.0572,
"step": 103500
},
{
"epoch": 1.74,
"learning_rate": 1.5958004650732028e-06,
"loss": 0.0695,
"step": 104000
},
{
"epoch": 1.75,
"learning_rate": 1.5449074791989455e-06,
"loss": 0.0397,
"step": 104500
},
{
"epoch": 1.75,
"learning_rate": 1.4940144933246887e-06,
"loss": 0.0654,
"step": 105000
},
{
"epoch": 1.76,
"learning_rate": 1.4431215074504318e-06,
"loss": 0.0487,
"step": 105500
},
{
"epoch": 1.77,
"learning_rate": 1.3922285215761747e-06,
"loss": 0.0425,
"step": 106000
},
{
"epoch": 1.78,
"learning_rate": 1.3413355357019178e-06,
"loss": 0.0469,
"step": 106500
},
{
"epoch": 1.79,
"learning_rate": 1.2904425498276607e-06,
"loss": 0.0713,
"step": 107000
},
{
"epoch": 1.8,
"learning_rate": 1.2395495639534036e-06,
"loss": 0.0615,
"step": 107500
},
{
"epoch": 1.8,
"learning_rate": 1.1886565780791467e-06,
"loss": 0.0501,
"step": 108000
},
{
"epoch": 1.81,
"learning_rate": 1.1377635922048896e-06,
"loss": 0.0548,
"step": 108500
},
{
"epoch": 1.82,
"learning_rate": 1.0868706063306326e-06,
"loss": 0.0402,
"step": 109000
},
{
"epoch": 1.83,
"learning_rate": 1.0359776204563757e-06,
"loss": 0.0453,
"step": 109500
},
{
"epoch": 1.84,
"learning_rate": 9.850846345821186e-07,
"loss": 0.0665,
"step": 110000
},
{
"epoch": 1.85,
"learning_rate": 9.341916487078616e-07,
"loss": 0.1019,
"step": 110500
},
{
"epoch": 1.85,
"learning_rate": 8.832986628336047e-07,
"loss": 0.0443,
"step": 111000
},
{
"epoch": 1.86,
"learning_rate": 8.324056769593476e-07,
"loss": 0.0584,
"step": 111500
},
{
"epoch": 1.87,
"learning_rate": 7.815126910850906e-07,
"loss": 0.0551,
"step": 112000
},
{
"epoch": 1.88,
"learning_rate": 7.306197052108335e-07,
"loss": 0.0371,
"step": 112500
},
{
"epoch": 1.89,
"learning_rate": 6.797267193365766e-07,
"loss": 0.0901,
"step": 113000
},
{
"epoch": 1.9,
"learning_rate": 6.288337334623196e-07,
"loss": 0.0482,
"step": 113500
},
{
"epoch": 1.91,
"learning_rate": 5.779407475880626e-07,
"loss": 0.0821,
"step": 114000
},
{
"epoch": 1.91,
"learning_rate": 5.270477617138056e-07,
"loss": 0.0482,
"step": 114500
},
{
"epoch": 1.92,
"learning_rate": 4.761547758395485e-07,
"loss": 0.0686,
"step": 115000
},
{
"epoch": 1.93,
"learning_rate": 4.252617899652915e-07,
"loss": 0.0765,
"step": 115500
},
{
"epoch": 1.94,
"learning_rate": 3.7436880409103453e-07,
"loss": 0.073,
"step": 116000
},
{
"epoch": 1.95,
"learning_rate": 3.234758182167775e-07,
"loss": 0.0592,
"step": 116500
},
{
"epoch": 1.96,
"learning_rate": 2.725828323425205e-07,
"loss": 0.076,
"step": 117000
},
{
"epoch": 1.96,
"learning_rate": 2.2168984646826352e-07,
"loss": 0.0532,
"step": 117500
},
{
"epoch": 1.97,
"learning_rate": 1.707968605940065e-07,
"loss": 0.053,
"step": 118000
},
{
"epoch": 1.98,
"learning_rate": 1.199038747197495e-07,
"loss": 0.0708,
"step": 118500
},
{
"epoch": 1.99,
"learning_rate": 6.90108888454925e-08,
"loss": 0.083,
"step": 119000
},
{
"epoch": 2.0,
"learning_rate": 1.8117902971235495e-08,
"loss": 0.0802,
"step": 119500
},
{
"epoch": 2.0,
"eval_accuracy": 0.989822775925601,
"eval_f1": 0.0,
"eval_loss": 0.061012040823698044,
"eval_precision": 0.0,
"eval_recall": 0.0,
"eval_runtime": 205.9172,
"eval_samples_per_second": 193.733,
"step": 119678
}
],
"max_steps": 119678,
"num_train_epochs": 2,
"total_flos": 2.160546753919596e+16,
"trial_name": null,
"trial_params": null
}