{ "best_global_step": 4470, "best_metric": 1.0481648445129395, "best_model_checkpoint": "./MediBerta/checkpoint-4470", "epoch": 10.0, "eval_steps": 500, "global_step": 4470, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0022371364653243847, "grad_norm": 32.67197036743164, "learning_rate": 0.0, "loss": 31.76995086669922, "num_input_tokens_seen": 327680, "step": 1, "train_runtime": 7.8573, "train_tokens_per_second": 41704.008 }, { "epoch": 0.11185682326621924, "grad_norm": 31.965707778930664, "learning_rate": 1.821561338289963e-05, "loss": 32.44386509486607, "num_input_tokens_seen": 16384000, "step": 50, "train_runtime": 380.1137, "train_tokens_per_second": 43102.889 }, { "epoch": 0.22371364653243847, "grad_norm": 34.0458984375, "learning_rate": 3.680297397769517e-05, "loss": 31.56456787109375, "num_input_tokens_seen": 32768000, "step": 100, "train_runtime": 759.9744, "train_tokens_per_second": 43117.243 }, { "epoch": 0.33557046979865773, "grad_norm": 36.13215255737305, "learning_rate": 5.539033457249071e-05, "loss": 30.703720703125, "num_input_tokens_seen": 49152000, "step": 150, "train_runtime": 1140.0254, "train_tokens_per_second": 43114.831 }, { "epoch": 0.44742729306487694, "grad_norm": 75.77301788330078, "learning_rate": 7.397769516728626e-05, "loss": 30.0342919921875, "num_input_tokens_seen": 65536000, "step": 200, "train_runtime": 1519.7868, "train_tokens_per_second": 43121.837 }, { "epoch": 0.5592841163310962, "grad_norm": 31.4502010345459, "learning_rate": 9.256505576208179e-05, "loss": 29.2935986328125, "num_input_tokens_seen": 81920000, "step": 250, "train_runtime": 1899.4384, "train_tokens_per_second": 43128.538 }, { "epoch": 0.6711409395973155, "grad_norm": 35.78462600708008, "learning_rate": 9.928588431325875e-05, "loss": 28.63412353515625, "num_input_tokens_seen": 98304000, "step": 300, "train_runtime": 2279.106, "train_tokens_per_second": 43132.703 }, { "epoch": 0.7829977628635347, "grad_norm": 32.92461395263672, "learning_rate": 9.809569150202333e-05, "loss": 27.89616455078125, "num_input_tokens_seen": 114688000, "step": 350, "train_runtime": 2659.6638, "train_tokens_per_second": 43121.24 }, { "epoch": 0.8948545861297539, "grad_norm": 31.881704330444336, "learning_rate": 9.69054986907879e-05, "loss": 27.38276123046875, "num_input_tokens_seen": 131072000, "step": 400, "train_runtime": 3039.907, "train_tokens_per_second": 43117.108 }, { "epoch": 1.0067114093959733, "grad_norm": 28.32501220703125, "learning_rate": 9.57153058795525e-05, "loss": 26.72684814453125, "num_input_tokens_seen": 147446784, "step": 450, "train_runtime": 3419.3158, "train_tokens_per_second": 43121.721 }, { "epoch": 1.1185682326621924, "grad_norm": 27.903852462768555, "learning_rate": 9.452511306831707e-05, "loss": 26.2695166015625, "num_input_tokens_seen": 163830784, "step": 500, "train_runtime": 3799.2571, "train_tokens_per_second": 43121.794 }, { "epoch": 1.1185682326621924, "eval_loss": 1.3941264152526855, "eval_masked_accuracy": 0.7140736610865701, "eval_runtime": 25.352, "eval_samples_per_second": 219.627, "eval_steps_per_second": 3.432, "num_input_tokens_seen": 163830784, "step": 500 }, { "epoch": 1.2304250559284116, "grad_norm": 25.51409339904785, "learning_rate": 9.333492025708165e-05, "loss": 25.86959228515625, "num_input_tokens_seen": 180214784, "step": 550, "train_runtime": 4207.1627, "train_tokens_per_second": 42835.231 }, { "epoch": 1.342281879194631, "grad_norm": 31.309749603271484, "learning_rate": 9.214472744584624e-05, "loss": 25.3655517578125, "num_input_tokens_seen": 196598784, "step": 600, "train_runtime": 4586.3773, "train_tokens_per_second": 42865.812 }, { "epoch": 1.45413870246085, "grad_norm": 28.270841598510742, "learning_rate": 9.095453463461082e-05, "loss": 25.1278271484375, "num_input_tokens_seen": 212982784, "step": 650, "train_runtime": 4966.2335, "train_tokens_per_second": 42886.18 }, { "epoch": 1.5659955257270695, "grad_norm": 27.522308349609375, "learning_rate": 8.976434182337538e-05, "loss": 24.80788818359375, "num_input_tokens_seen": 229366784, "step": 700, "train_runtime": 5346.6697, "train_tokens_per_second": 42899.0 }, { "epoch": 1.6778523489932886, "grad_norm": 27.093711853027344, "learning_rate": 8.857414901213997e-05, "loss": 24.6148046875, "num_input_tokens_seen": 245750784, "step": 750, "train_runtime": 5726.4531, "train_tokens_per_second": 42915.008 }, { "epoch": 1.7897091722595078, "grad_norm": 25.49102210998535, "learning_rate": 8.738395620090455e-05, "loss": 24.40305908203125, "num_input_tokens_seen": 262134784, "step": 800, "train_runtime": 6106.8482, "train_tokens_per_second": 42924.726 }, { "epoch": 1.901565995525727, "grad_norm": 25.048629760742188, "learning_rate": 8.619376338966914e-05, "loss": 24.0977783203125, "num_input_tokens_seen": 278518784, "step": 850, "train_runtime": 6486.6563, "train_tokens_per_second": 42937.189 }, { "epoch": 2.0134228187919465, "grad_norm": 25.157657623291016, "learning_rate": 8.500357057843372e-05, "loss": 23.89361572265625, "num_input_tokens_seen": 294893568, "step": 900, "train_runtime": 6867.068, "train_tokens_per_second": 42943.155 }, { "epoch": 2.1252796420581657, "grad_norm": 25.902904510498047, "learning_rate": 8.381337776719829e-05, "loss": 23.57096435546875, "num_input_tokens_seen": 311277568, "step": 950, "train_runtime": 7246.8643, "train_tokens_per_second": 42953.415 }, { "epoch": 2.237136465324385, "grad_norm": 24.553022384643555, "learning_rate": 8.262318495596287e-05, "loss": 23.42177490234375, "num_input_tokens_seen": 327661568, "step": 1000, "train_runtime": 7626.2318, "train_tokens_per_second": 42965.068 }, { "epoch": 2.237136465324385, "eval_loss": 1.2587531805038452, "eval_masked_accuracy": 0.7357594334234474, "eval_runtime": 25.3576, "eval_samples_per_second": 219.579, "eval_steps_per_second": 3.431, "num_input_tokens_seen": 327661568, "step": 1000 }, { "epoch": 2.348993288590604, "grad_norm": 24.940990447998047, "learning_rate": 8.143299214472745e-05, "loss": 23.259130859375, "num_input_tokens_seen": 344045568, "step": 1050, "train_runtime": 8033.7212, "train_tokens_per_second": 42825.182 }, { "epoch": 2.460850111856823, "grad_norm": 24.946720123291016, "learning_rate": 8.024279933349202e-05, "loss": 23.12722900390625, "num_input_tokens_seen": 360429568, "step": 1100, "train_runtime": 8413.8773, "train_tokens_per_second": 42837.512 }, { "epoch": 2.5727069351230423, "grad_norm": 24.135683059692383, "learning_rate": 7.905260652225661e-05, "loss": 22.9796826171875, "num_input_tokens_seen": 376813568, "step": 1150, "train_runtime": 8793.1262, "train_tokens_per_second": 42853.197 }, { "epoch": 2.684563758389262, "grad_norm": 25.850025177001953, "learning_rate": 7.786241371102119e-05, "loss": 22.77516845703125, "num_input_tokens_seen": 393197568, "step": 1200, "train_runtime": 9172.793, "train_tokens_per_second": 42865.632 }, { "epoch": 2.796420581655481, "grad_norm": 22.62376594543457, "learning_rate": 7.667222089978577e-05, "loss": 22.6738720703125, "num_input_tokens_seen": 409581568, "step": 1250, "train_runtime": 9553.7727, "train_tokens_per_second": 42871.186 }, { "epoch": 2.9082774049217, "grad_norm": 22.361886978149414, "learning_rate": 7.548202808855035e-05, "loss": 22.53384033203125, "num_input_tokens_seen": 425965568, "step": 1300, "train_runtime": 9932.9496, "train_tokens_per_second": 42884.096 }, { "epoch": 3.0201342281879193, "grad_norm": 24.093069076538086, "learning_rate": 7.429183527731492e-05, "loss": 22.38953125, "num_input_tokens_seen": 442340352, "step": 1350, "train_runtime": 10313.033, "train_tokens_per_second": 42891.393 }, { "epoch": 3.131991051454139, "grad_norm": 22.419540405273438, "learning_rate": 7.310164246607951e-05, "loss": 22.18701416015625, "num_input_tokens_seen": 458724352, "step": 1400, "train_runtime": 10693.1337, "train_tokens_per_second": 42898.964 }, { "epoch": 3.243847874720358, "grad_norm": 24.061424255371094, "learning_rate": 7.191144965484409e-05, "loss": 22.1206396484375, "num_input_tokens_seen": 475108352, "step": 1450, "train_runtime": 11073.944, "train_tokens_per_second": 42903.265 }, { "epoch": 3.3557046979865772, "grad_norm": 21.660381317138672, "learning_rate": 7.072125684360867e-05, "loss": 21.99459228515625, "num_input_tokens_seen": 491492352, "step": 1500, "train_runtime": 11453.1596, "train_tokens_per_second": 42913.254 }, { "epoch": 3.3557046979865772, "eval_loss": 1.194376826286316, "eval_masked_accuracy": 0.7462490242091618, "eval_runtime": 25.3329, "eval_samples_per_second": 219.794, "eval_steps_per_second": 3.434, "num_input_tokens_seen": 491492352, "step": 1500 }, { "epoch": 3.4675615212527964, "grad_norm": 23.65436553955078, "learning_rate": 6.953106403237324e-05, "loss": 21.8618212890625, "num_input_tokens_seen": 507876352, "step": 1550, "train_runtime": 11860.4601, "train_tokens_per_second": 42820.965 }, { "epoch": 3.5794183445190155, "grad_norm": 23.492694854736328, "learning_rate": 6.834087122113782e-05, "loss": 21.84644287109375, "num_input_tokens_seen": 524260352, "step": 1600, "train_runtime": 12240.4952, "train_tokens_per_second": 42829.995 }, { "epoch": 3.6912751677852347, "grad_norm": 22.143146514892578, "learning_rate": 6.715067840990241e-05, "loss": 21.695458984375, "num_input_tokens_seen": 540644352, "step": 1650, "train_runtime": 12620.3553, "train_tokens_per_second": 42839.075 }, { "epoch": 3.8031319910514543, "grad_norm": 22.582658767700195, "learning_rate": 6.596048559866699e-05, "loss": 21.49891357421875, "num_input_tokens_seen": 557028352, "step": 1700, "train_runtime": 13000.3344, "train_tokens_per_second": 42847.233 }, { "epoch": 3.9149888143176734, "grad_norm": 22.46770477294922, "learning_rate": 6.477029278743157e-05, "loss": 21.4741015625, "num_input_tokens_seen": 573412352, "step": 1750, "train_runtime": 13380.7098, "train_tokens_per_second": 42853.657 }, { "epoch": 4.026845637583893, "grad_norm": 22.605365753173828, "learning_rate": 6.358009997619616e-05, "loss": 21.39329833984375, "num_input_tokens_seen": 589787136, "step": 1800, "train_runtime": 13759.8334, "train_tokens_per_second": 42862.956 }, { "epoch": 4.138702460850112, "grad_norm": 21.95088005065918, "learning_rate": 6.238990716496072e-05, "loss": 21.18496826171875, "num_input_tokens_seen": 606171136, "step": 1850, "train_runtime": 14139.2097, "train_tokens_per_second": 42871.642 }, { "epoch": 4.250559284116331, "grad_norm": 22.34777069091797, "learning_rate": 6.11997143537253e-05, "loss": 21.2357177734375, "num_input_tokens_seen": 622555136, "step": 1900, "train_runtime": 14518.7366, "train_tokens_per_second": 42879.429 }, { "epoch": 4.3624161073825505, "grad_norm": 22.553071975708008, "learning_rate": 6.000952154248989e-05, "loss": 21.16220458984375, "num_input_tokens_seen": 638939136, "step": 1950, "train_runtime": 14898.759, "train_tokens_per_second": 42885.393 }, { "epoch": 4.47427293064877, "grad_norm": 22.87005615234375, "learning_rate": 5.8819328731254464e-05, "loss": 21.01521240234375, "num_input_tokens_seen": 655323136, "step": 2000, "train_runtime": 15278.457, "train_tokens_per_second": 42891.971 }, { "epoch": 4.47427293064877, "eval_loss": 1.14796781539917, "eval_masked_accuracy": 0.7536936821619982, "eval_runtime": 25.3387, "eval_samples_per_second": 219.743, "eval_steps_per_second": 3.433, "num_input_tokens_seen": 655323136, "step": 2000 }, { "epoch": 4.586129753914989, "grad_norm": 22.233734130859375, "learning_rate": 5.762913592001905e-05, "loss": 20.9041943359375, "num_input_tokens_seen": 671707136, "step": 2050, "train_runtime": 15686.3972, "train_tokens_per_second": 42820.995 }, { "epoch": 4.697986577181208, "grad_norm": 21.8343448638916, "learning_rate": 5.6438943108783625e-05, "loss": 20.91035888671875, "num_input_tokens_seen": 688091136, "step": 2100, "train_runtime": 16066.5233, "train_tokens_per_second": 42827.631 }, { "epoch": 4.809843400447427, "grad_norm": 21.778446197509766, "learning_rate": 5.52487502975482e-05, "loss": 20.8124365234375, "num_input_tokens_seen": 704475136, "step": 2150, "train_runtime": 16446.1615, "train_tokens_per_second": 42835.231 }, { "epoch": 4.921700223713646, "grad_norm": 21.686033248901367, "learning_rate": 5.4058557486312786e-05, "loss": 20.7349560546875, "num_input_tokens_seen": 720859136, "step": 2200, "train_runtime": 16825.7207, "train_tokens_per_second": 42842.69 }, { "epoch": 5.033557046979865, "grad_norm": 21.80782127380371, "learning_rate": 5.286836467507736e-05, "loss": 20.7165283203125, "num_input_tokens_seen": 737233920, "step": 2250, "train_runtime": 17206.3274, "train_tokens_per_second": 42846.675 }, { "epoch": 5.145413870246085, "grad_norm": 23.681589126586914, "learning_rate": 5.167817186384194e-05, "loss": 20.5538916015625, "num_input_tokens_seen": 753617920, "step": 2300, "train_runtime": 17586.976, "train_tokens_per_second": 42850.91 }, { "epoch": 5.257270693512305, "grad_norm": 23.062753677368164, "learning_rate": 5.0487979052606524e-05, "loss": 20.505771484375, "num_input_tokens_seen": 770001920, "step": 2350, "train_runtime": 17966.3576, "train_tokens_per_second": 42857.987 }, { "epoch": 5.369127516778524, "grad_norm": 22.336442947387695, "learning_rate": 4.92977862413711e-05, "loss": 20.479666748046874, "num_input_tokens_seen": 786385920, "step": 2400, "train_runtime": 18346.1922, "train_tokens_per_second": 42863.713 }, { "epoch": 5.480984340044743, "grad_norm": 22.454973220825195, "learning_rate": 4.8107593430135685e-05, "loss": 20.462833251953125, "num_input_tokens_seen": 802769920, "step": 2450, "train_runtime": 18726.6308, "train_tokens_per_second": 42867.824 }, { "epoch": 5.592841163310962, "grad_norm": 21.835525512695312, "learning_rate": 4.691740061890026e-05, "loss": 20.33416748046875, "num_input_tokens_seen": 819153920, "step": 2500, "train_runtime": 19106.2918, "train_tokens_per_second": 42873.517 }, { "epoch": 5.592841163310962, "eval_loss": 1.1132277250289917, "eval_masked_accuracy": 0.7584756667111459, "eval_runtime": 25.3527, "eval_samples_per_second": 219.621, "eval_steps_per_second": 3.432, "num_input_tokens_seen": 819153920, "step": 2500 }, { "epoch": 5.704697986577181, "grad_norm": 23.234500885009766, "learning_rate": 4.5727207807664845e-05, "loss": 20.25551513671875, "num_input_tokens_seen": 835537920, "step": 2550, "train_runtime": 19513.5356, "train_tokens_per_second": 42818.377 }, { "epoch": 5.8165548098434, "grad_norm": 21.292207717895508, "learning_rate": 4.453701499642942e-05, "loss": 20.156134033203124, "num_input_tokens_seen": 851921920, "step": 2600, "train_runtime": 19893.1597, "train_tokens_per_second": 42824.867 }, { "epoch": 5.9284116331096195, "grad_norm": 20.622861862182617, "learning_rate": 4.3346822185194e-05, "loss": 20.159989013671876, "num_input_tokens_seen": 868305920, "step": 2650, "train_runtime": 20273.9368, "train_tokens_per_second": 42828.679 }, { "epoch": 6.040268456375839, "grad_norm": 23.077646255493164, "learning_rate": 4.215662937395858e-05, "loss": 20.13439697265625, "num_input_tokens_seen": 884680704, "step": 2700, "train_runtime": 20654.2855, "train_tokens_per_second": 42832.791 }, { "epoch": 6.152125279642058, "grad_norm": 21.280527114868164, "learning_rate": 4.096643656272317e-05, "loss": 19.994425048828123, "num_input_tokens_seen": 901064704, "step": 2750, "train_runtime": 21033.1909, "train_tokens_per_second": 42840.133 }, { "epoch": 6.263982102908278, "grad_norm": 23.213178634643555, "learning_rate": 3.977624375148774e-05, "loss": 19.980313720703126, "num_input_tokens_seen": 917448704, "step": 2800, "train_runtime": 21412.6494, "train_tokens_per_second": 42846.109 }, { "epoch": 6.375838926174497, "grad_norm": 22.994335174560547, "learning_rate": 3.858605094025232e-05, "loss": 19.9128857421875, "num_input_tokens_seen": 933832704, "step": 2850, "train_runtime": 21792.7904, "train_tokens_per_second": 42850.534 }, { "epoch": 6.487695749440716, "grad_norm": 25.134178161621094, "learning_rate": 3.7395858129016905e-05, "loss": 19.874437255859377, "num_input_tokens_seen": 950216704, "step": 2900, "train_runtime": 22173.1567, "train_tokens_per_second": 42854.372 }, { "epoch": 6.599552572706935, "grad_norm": 20.69732093811035, "learning_rate": 3.620566531778148e-05, "loss": 19.910538330078126, "num_input_tokens_seen": 966600704, "step": 2950, "train_runtime": 22553.2542, "train_tokens_per_second": 42858.591 }, { "epoch": 6.7114093959731544, "grad_norm": 21.227550506591797, "learning_rate": 3.501547250654606e-05, "loss": 19.87051025390625, "num_input_tokens_seen": 982984704, "step": 3000, "train_runtime": 22933.4022, "train_tokens_per_second": 42862.576 }, { "epoch": 6.7114093959731544, "eval_loss": 1.0944103002548218, "eval_masked_accuracy": 0.7619272139314145, "eval_runtime": 25.3399, "eval_samples_per_second": 219.732, "eval_steps_per_second": 3.433, "num_input_tokens_seen": 982984704, "step": 3000 }, { "epoch": 6.823266219239374, "grad_norm": 21.8638858795166, "learning_rate": 3.382527969531064e-05, "loss": 19.65084228515625, "num_input_tokens_seen": 999368704, "step": 3050, "train_runtime": 23340.7393, "train_tokens_per_second": 42816.497 }, { "epoch": 6.935123042505593, "grad_norm": 25.967254638671875, "learning_rate": 3.263508688407522e-05, "loss": 19.7602978515625, "num_input_tokens_seen": 1015752704, "step": 3100, "train_runtime": 23721.0833, "train_tokens_per_second": 42820.671 }, { "epoch": 7.046979865771812, "grad_norm": 29.288837432861328, "learning_rate": 3.1444894072839803e-05, "loss": 19.64591064453125, "num_input_tokens_seen": 1032127488, "step": 3150, "train_runtime": 24100.8116, "train_tokens_per_second": 42825.425 }, { "epoch": 7.158836689038031, "grad_norm": 20.559537887573242, "learning_rate": 3.025470126160438e-05, "loss": 19.643092041015624, "num_input_tokens_seen": 1048511488, "step": 3200, "train_runtime": 24479.6774, "train_tokens_per_second": 42831.916 }, { "epoch": 7.27069351230425, "grad_norm": 21.216949462890625, "learning_rate": 2.906450845036896e-05, "loss": 19.5406787109375, "num_input_tokens_seen": 1064895488, "step": 3250, "train_runtime": 24859.9356, "train_tokens_per_second": 42835.81 }, { "epoch": 7.382550335570469, "grad_norm": 19.948974609375, "learning_rate": 2.787431563913354e-05, "loss": 19.52265625, "num_input_tokens_seen": 1081279488, "step": 3300, "train_runtime": 25239.4624, "train_tokens_per_second": 42840.829 }, { "epoch": 7.494407158836689, "grad_norm": 24.806529998779297, "learning_rate": 2.668412282789812e-05, "loss": 19.483994140625, "num_input_tokens_seen": 1097663488, "step": 3350, "train_runtime": 25619.1507, "train_tokens_per_second": 42845.428 }, { "epoch": 7.6062639821029085, "grad_norm": 22.306764602661133, "learning_rate": 2.54939300166627e-05, "loss": 19.500614013671875, "num_input_tokens_seen": 1114047488, "step": 3400, "train_runtime": 25998.4643, "train_tokens_per_second": 42850.511 }, { "epoch": 7.718120805369128, "grad_norm": 20.259811401367188, "learning_rate": 2.430373720542728e-05, "loss": 19.44253173828125, "num_input_tokens_seen": 1130431488, "step": 3450, "train_runtime": 26378.2281, "train_tokens_per_second": 42854.716 }, { "epoch": 7.829977628635347, "grad_norm": 31.171615600585938, "learning_rate": 2.311354439419186e-05, "loss": 19.37197021484375, "num_input_tokens_seen": 1146815488, "step": 3500, "train_runtime": 26757.8966, "train_tokens_per_second": 42858.955 }, { "epoch": 7.829977628635347, "eval_loss": 1.0732760429382324, "eval_masked_accuracy": 0.7661423014370904, "eval_runtime": 25.3349, "eval_samples_per_second": 219.775, "eval_steps_per_second": 3.434, "num_input_tokens_seen": 1146815488, "step": 3500 }, { "epoch": 7.941834451901566, "grad_norm": 19.443883895874023, "learning_rate": 2.192335158295644e-05, "loss": 19.302330322265625, "num_input_tokens_seen": 1163199488, "step": 3550, "train_runtime": 27165.3477, "train_tokens_per_second": 42819.238 }, { "epoch": 8.053691275167786, "grad_norm": 19.65326690673828, "learning_rate": 2.073315877172102e-05, "loss": 19.334464111328124, "num_input_tokens_seen": 1179574272, "step": 3600, "train_runtime": 27545.6424, "train_tokens_per_second": 42822.536 }, { "epoch": 8.165548098434005, "grad_norm": 20.760732650756836, "learning_rate": 1.9542965960485597e-05, "loss": 19.248583984375, "num_input_tokens_seen": 1195958272, "step": 3650, "train_runtime": 27925.9284, "train_tokens_per_second": 42826.088 }, { "epoch": 8.277404921700224, "grad_norm": 19.925457000732422, "learning_rate": 1.835277314925018e-05, "loss": 19.22830322265625, "num_input_tokens_seen": 1212342272, "step": 3700, "train_runtime": 28305.1518, "train_tokens_per_second": 42831.152 }, { "epoch": 8.389261744966444, "grad_norm": 20.903831481933594, "learning_rate": 1.7162580338014758e-05, "loss": 19.20845458984375, "num_input_tokens_seen": 1228726272, "step": 3750, "train_runtime": 28684.7612, "train_tokens_per_second": 42835.506 }, { "epoch": 8.501118568232663, "grad_norm": 25.746238708496094, "learning_rate": 1.597238752677934e-05, "loss": 19.181019287109375, "num_input_tokens_seen": 1245110272, "step": 3800, "train_runtime": 29065.5625, "train_tokens_per_second": 42837.99 }, { "epoch": 8.612975391498882, "grad_norm": 19.735681533813477, "learning_rate": 1.4782194715543917e-05, "loss": 19.200714111328125, "num_input_tokens_seen": 1261494272, "step": 3850, "train_runtime": 29444.9518, "train_tokens_per_second": 42842.463 }, { "epoch": 8.724832214765101, "grad_norm": 20.348718643188477, "learning_rate": 1.35920019043085e-05, "loss": 19.14758056640625, "num_input_tokens_seen": 1277878272, "step": 3900, "train_runtime": 29825.2862, "train_tokens_per_second": 42845.466 }, { "epoch": 8.83668903803132, "grad_norm": 19.208051681518555, "learning_rate": 1.2401809093073078e-05, "loss": 19.07127685546875, "num_input_tokens_seen": 1294262272, "step": 3950, "train_runtime": 30205.5851, "train_tokens_per_second": 42848.442 }, { "epoch": 8.94854586129754, "grad_norm": 20.35933494567871, "learning_rate": 1.1211616281837658e-05, "loss": 19.101888427734377, "num_input_tokens_seen": 1310646272, "step": 4000, "train_runtime": 30585.6001, "train_tokens_per_second": 42851.743 }, { "epoch": 8.94854586129754, "eval_loss": 1.0551483631134033, "eval_masked_accuracy": 0.7687605175839743, "eval_runtime": 25.3745, "eval_samples_per_second": 219.433, "eval_steps_per_second": 3.429, "num_input_tokens_seen": 1310646272, "step": 4000 }, { "epoch": 9.060402684563758, "grad_norm": 19.61882972717285, "learning_rate": 1.0021423470602239e-05, "loss": 19.01905517578125, "num_input_tokens_seen": 1327021056, "step": 4050, "train_runtime": 30993.7338, "train_tokens_per_second": 42815.785 }, { "epoch": 9.172259507829978, "grad_norm": 20.218429565429688, "learning_rate": 8.831230659366818e-06, "loss": 19.05180419921875, "num_input_tokens_seen": 1343405056, "step": 4100, "train_runtime": 31373.034, "train_tokens_per_second": 42820.374 }, { "epoch": 9.284116331096197, "grad_norm": 19.39629554748535, "learning_rate": 7.641037848131398e-06, "loss": 19.03988525390625, "num_input_tokens_seen": 1359789056, "step": 4150, "train_runtime": 31753.5802, "train_tokens_per_second": 42823.173 }, { "epoch": 9.395973154362416, "grad_norm": 20.272720336914062, "learning_rate": 6.4508450368959775e-06, "loss": 18.968330078125, "num_input_tokens_seen": 1376173056, "step": 4200, "train_runtime": 32133.9665, "train_tokens_per_second": 42826.119 }, { "epoch": 9.507829977628635, "grad_norm": 19.632165908813477, "learning_rate": 5.260652225660558e-06, "loss": 18.9827783203125, "num_input_tokens_seen": 1392557056, "step": 4250, "train_runtime": 32514.0065, "train_tokens_per_second": 42829.451 }, { "epoch": 9.619686800894854, "grad_norm": 19.80462646484375, "learning_rate": 4.0704594144251375e-06, "loss": 18.956285400390627, "num_input_tokens_seen": 1408941056, "step": 4300, "train_runtime": 32894.5758, "train_tokens_per_second": 42832.018 }, { "epoch": 9.731543624161073, "grad_norm": 20.74568748474121, "learning_rate": 2.8802666031897166e-06, "loss": 18.866341552734376, "num_input_tokens_seen": 1425325056, "step": 4350, "train_runtime": 33275.028, "train_tokens_per_second": 42834.676 }, { "epoch": 9.843400447427292, "grad_norm": 19.217836380004883, "learning_rate": 1.6900737919542966e-06, "loss": 18.920970458984375, "num_input_tokens_seen": 1441709056, "step": 4400, "train_runtime": 33654.5878, "train_tokens_per_second": 42838.411 }, { "epoch": 9.955257270693512, "grad_norm": 19.824554443359375, "learning_rate": 4.998809807188764e-07, "loss": 18.913228759765627, "num_input_tokens_seen": 1458093056, "step": 4450, "train_runtime": 34033.2944, "train_tokens_per_second": 42843.136 }, { "epoch": 10.0, "eval_loss": 1.0481648445129395, "eval_masked_accuracy": 0.7704025727385737, "eval_runtime": 25.3462, "eval_samples_per_second": 219.678, "eval_steps_per_second": 3.432, "num_input_tokens_seen": 1464637440, "step": 4470 }, { "epoch": 10.0, "num_input_tokens_seen": 1464637440, "step": 4470, "total_flos": 4.917799558028206e+18, "train_loss": 21.80173199032777, "train_runtime": 34212.9191, "train_samples_per_second": 83.612, "train_steps_per_second": 0.131 } ], "logging_steps": 50, "max_steps": 4470, "num_input_tokens_seen": 1464637440, "num_train_epochs": 10, "save_steps": 500, "stateful_callbacks": { "EarlyStoppingCallback": { "args": { "early_stopping_patience": 5, "early_stopping_threshold": 0.0001 }, "attributes": { "early_stopping_patience_counter": 0 } }, "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 4.917799558028206e+18, "train_batch_size": 40, "trial_name": null, "trial_params": null }