YeRyeongLee's picture
End of training
4ceacd6
raw
history blame
47.9 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 9.999680409076383,
"global_step": 3910,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.03,
"learning_rate": 5.115089514066497e-07,
"loss": 2.3876,
"step": 10
},
{
"epoch": 0.05,
"learning_rate": 1.0230179028132994e-06,
"loss": 2.382,
"step": 20
},
{
"epoch": 0.08,
"learning_rate": 1.534526854219949e-06,
"loss": 2.3645,
"step": 30
},
{
"epoch": 0.1,
"learning_rate": 2.0460358056265987e-06,
"loss": 2.3333,
"step": 40
},
{
"epoch": 0.13,
"learning_rate": 2.5575447570332483e-06,
"loss": 2.2959,
"step": 50
},
{
"epoch": 0.15,
"learning_rate": 3.069053708439898e-06,
"loss": 2.2527,
"step": 60
},
{
"epoch": 0.18,
"learning_rate": 3.5805626598465474e-06,
"loss": 2.2021,
"step": 70
},
{
"epoch": 0.2,
"learning_rate": 4.092071611253197e-06,
"loss": 2.1613,
"step": 80
},
{
"epoch": 0.23,
"learning_rate": 4.603580562659847e-06,
"loss": 2.0852,
"step": 90
},
{
"epoch": 0.26,
"learning_rate": 5.1150895140664966e-06,
"loss": 2.0602,
"step": 100
},
{
"epoch": 0.28,
"learning_rate": 5.626598465473146e-06,
"loss": 2.0174,
"step": 110
},
{
"epoch": 0.31,
"learning_rate": 6.138107416879796e-06,
"loss": 1.9743,
"step": 120
},
{
"epoch": 0.33,
"learning_rate": 6.649616368286445e-06,
"loss": 1.8913,
"step": 130
},
{
"epoch": 0.36,
"learning_rate": 7.161125319693095e-06,
"loss": 1.8595,
"step": 140
},
{
"epoch": 0.38,
"learning_rate": 7.672634271099745e-06,
"loss": 1.7506,
"step": 150
},
{
"epoch": 0.41,
"learning_rate": 8.184143222506395e-06,
"loss": 1.6762,
"step": 160
},
{
"epoch": 0.43,
"learning_rate": 8.695652173913044e-06,
"loss": 1.6315,
"step": 170
},
{
"epoch": 0.46,
"learning_rate": 9.207161125319694e-06,
"loss": 1.5075,
"step": 180
},
{
"epoch": 0.49,
"learning_rate": 9.718670076726344e-06,
"loss": 1.522,
"step": 190
},
{
"epoch": 0.51,
"learning_rate": 1.0230179028132993e-05,
"loss": 1.4587,
"step": 200
},
{
"epoch": 0.54,
"learning_rate": 1.0741687979539643e-05,
"loss": 1.4112,
"step": 210
},
{
"epoch": 0.56,
"learning_rate": 1.1253196930946292e-05,
"loss": 1.4088,
"step": 220
},
{
"epoch": 0.59,
"learning_rate": 1.1764705882352942e-05,
"loss": 1.3279,
"step": 230
},
{
"epoch": 0.61,
"learning_rate": 1.2276214833759591e-05,
"loss": 1.3437,
"step": 240
},
{
"epoch": 0.64,
"learning_rate": 1.2787723785166241e-05,
"loss": 1.312,
"step": 250
},
{
"epoch": 0.66,
"learning_rate": 1.329923273657289e-05,
"loss": 1.2475,
"step": 260
},
{
"epoch": 0.69,
"learning_rate": 1.381074168797954e-05,
"loss": 1.2143,
"step": 270
},
{
"epoch": 0.72,
"learning_rate": 1.432225063938619e-05,
"loss": 1.2112,
"step": 280
},
{
"epoch": 0.74,
"learning_rate": 1.483375959079284e-05,
"loss": 1.1612,
"step": 290
},
{
"epoch": 0.77,
"learning_rate": 1.534526854219949e-05,
"loss": 1.1555,
"step": 300
},
{
"epoch": 0.79,
"learning_rate": 1.585677749360614e-05,
"loss": 1.1404,
"step": 310
},
{
"epoch": 0.82,
"learning_rate": 1.636828644501279e-05,
"loss": 1.1411,
"step": 320
},
{
"epoch": 0.84,
"learning_rate": 1.687979539641944e-05,
"loss": 1.1204,
"step": 330
},
{
"epoch": 0.87,
"learning_rate": 1.739130434782609e-05,
"loss": 1.0784,
"step": 340
},
{
"epoch": 0.89,
"learning_rate": 1.790281329923274e-05,
"loss": 1.0616,
"step": 350
},
{
"epoch": 0.92,
"learning_rate": 1.8414322250639388e-05,
"loss": 1.0479,
"step": 360
},
{
"epoch": 0.95,
"learning_rate": 1.8925831202046038e-05,
"loss": 1.087,
"step": 370
},
{
"epoch": 0.97,
"learning_rate": 1.9437340153452687e-05,
"loss": 1.0054,
"step": 380
},
{
"epoch": 1.0,
"learning_rate": 1.9948849104859337e-05,
"loss": 1.0543,
"step": 390
},
{
"epoch": 1.02,
"learning_rate": 1.9948849104859337e-05,
"loss": 0.9918,
"step": 400
},
{
"epoch": 1.05,
"learning_rate": 1.9892014776925264e-05,
"loss": 0.9886,
"step": 410
},
{
"epoch": 1.07,
"learning_rate": 1.983518044899119e-05,
"loss": 0.9545,
"step": 420
},
{
"epoch": 1.1,
"learning_rate": 1.9778346121057118e-05,
"loss": 0.9254,
"step": 430
},
{
"epoch": 1.13,
"learning_rate": 1.972151179312305e-05,
"loss": 0.9518,
"step": 440
},
{
"epoch": 1.15,
"learning_rate": 1.9664677465188975e-05,
"loss": 0.938,
"step": 450
},
{
"epoch": 1.18,
"learning_rate": 1.9607843137254903e-05,
"loss": 0.9437,
"step": 460
},
{
"epoch": 1.2,
"learning_rate": 1.955100880932083e-05,
"loss": 0.9041,
"step": 470
},
{
"epoch": 1.23,
"learning_rate": 1.949417448138676e-05,
"loss": 0.9089,
"step": 480
},
{
"epoch": 1.25,
"learning_rate": 1.9437340153452687e-05,
"loss": 0.8944,
"step": 490
},
{
"epoch": 1.28,
"learning_rate": 1.9380505825518614e-05,
"loss": 0.9,
"step": 500
},
{
"epoch": 1.3,
"learning_rate": 1.932367149758454e-05,
"loss": 0.9285,
"step": 510
},
{
"epoch": 1.33,
"learning_rate": 1.926683716965047e-05,
"loss": 0.9065,
"step": 520
},
{
"epoch": 1.36,
"learning_rate": 1.92100028417164e-05,
"loss": 0.8607,
"step": 530
},
{
"epoch": 1.38,
"learning_rate": 1.9153168513782326e-05,
"loss": 0.8576,
"step": 540
},
{
"epoch": 1.41,
"learning_rate": 1.9096334185848253e-05,
"loss": 0.8888,
"step": 550
},
{
"epoch": 1.43,
"learning_rate": 1.903949985791418e-05,
"loss": 0.8894,
"step": 560
},
{
"epoch": 1.46,
"learning_rate": 1.898266552998011e-05,
"loss": 0.8637,
"step": 570
},
{
"epoch": 1.48,
"learning_rate": 1.8925831202046038e-05,
"loss": 0.8725,
"step": 580
},
{
"epoch": 1.51,
"learning_rate": 1.8868996874111965e-05,
"loss": 0.852,
"step": 590
},
{
"epoch": 1.53,
"learning_rate": 1.881216254617789e-05,
"loss": 0.8559,
"step": 600
},
{
"epoch": 1.56,
"learning_rate": 1.875532821824382e-05,
"loss": 0.8671,
"step": 610
},
{
"epoch": 1.59,
"learning_rate": 1.869849389030975e-05,
"loss": 0.8656,
"step": 620
},
{
"epoch": 1.61,
"learning_rate": 1.8641659562375676e-05,
"loss": 0.8336,
"step": 630
},
{
"epoch": 1.64,
"learning_rate": 1.8584825234441603e-05,
"loss": 0.8274,
"step": 640
},
{
"epoch": 1.66,
"learning_rate": 1.852799090650753e-05,
"loss": 0.8555,
"step": 650
},
{
"epoch": 1.69,
"learning_rate": 1.847115657857346e-05,
"loss": 0.8158,
"step": 660
},
{
"epoch": 1.71,
"learning_rate": 1.8414322250639388e-05,
"loss": 0.8182,
"step": 670
},
{
"epoch": 1.74,
"learning_rate": 1.8357487922705315e-05,
"loss": 0.8143,
"step": 680
},
{
"epoch": 1.76,
"learning_rate": 1.8300653594771242e-05,
"loss": 0.8632,
"step": 690
},
{
"epoch": 1.79,
"learning_rate": 1.824381926683717e-05,
"loss": 0.8461,
"step": 700
},
{
"epoch": 1.82,
"learning_rate": 1.81869849389031e-05,
"loss": 0.8144,
"step": 710
},
{
"epoch": 1.84,
"learning_rate": 1.8130150610969027e-05,
"loss": 0.834,
"step": 720
},
{
"epoch": 1.87,
"learning_rate": 1.8073316283034954e-05,
"loss": 0.8729,
"step": 730
},
{
"epoch": 1.89,
"learning_rate": 1.801648195510088e-05,
"loss": 0.8187,
"step": 740
},
{
"epoch": 1.92,
"learning_rate": 1.795964762716681e-05,
"loss": 0.827,
"step": 750
},
{
"epoch": 1.94,
"learning_rate": 1.790281329923274e-05,
"loss": 0.8571,
"step": 760
},
{
"epoch": 1.97,
"learning_rate": 1.7845978971298665e-05,
"loss": 0.8204,
"step": 770
},
{
"epoch": 1.99,
"learning_rate": 1.7789144643364593e-05,
"loss": 0.8573,
"step": 780
},
{
"epoch": 2.02,
"learning_rate": 1.773231031543052e-05,
"loss": 0.7526,
"step": 790
},
{
"epoch": 2.05,
"learning_rate": 1.767547598749645e-05,
"loss": 0.7175,
"step": 800
},
{
"epoch": 2.07,
"learning_rate": 1.7618641659562377e-05,
"loss": 0.7448,
"step": 810
},
{
"epoch": 2.1,
"learning_rate": 1.7561807331628304e-05,
"loss": 0.7375,
"step": 820
},
{
"epoch": 2.12,
"learning_rate": 1.750497300369423e-05,
"loss": 0.7215,
"step": 830
},
{
"epoch": 2.15,
"learning_rate": 1.7448138675760162e-05,
"loss": 0.7259,
"step": 840
},
{
"epoch": 2.17,
"learning_rate": 1.739130434782609e-05,
"loss": 0.7564,
"step": 850
},
{
"epoch": 2.2,
"learning_rate": 1.7334470019892016e-05,
"loss": 0.7001,
"step": 860
},
{
"epoch": 2.22,
"learning_rate": 1.7277635691957943e-05,
"loss": 0.7136,
"step": 870
},
{
"epoch": 2.25,
"learning_rate": 1.722080136402387e-05,
"loss": 0.7273,
"step": 880
},
{
"epoch": 2.28,
"learning_rate": 1.71639670360898e-05,
"loss": 0.7035,
"step": 890
},
{
"epoch": 2.3,
"learning_rate": 1.7107132708155728e-05,
"loss": 0.7125,
"step": 900
},
{
"epoch": 2.33,
"learning_rate": 1.7050298380221655e-05,
"loss": 0.73,
"step": 910
},
{
"epoch": 2.35,
"learning_rate": 1.6993464052287582e-05,
"loss": 0.7052,
"step": 920
},
{
"epoch": 2.38,
"learning_rate": 1.6936629724353512e-05,
"loss": 0.7362,
"step": 930
},
{
"epoch": 2.4,
"learning_rate": 1.687979539641944e-05,
"loss": 0.7649,
"step": 940
},
{
"epoch": 2.43,
"learning_rate": 1.6822961068485366e-05,
"loss": 0.7001,
"step": 950
},
{
"epoch": 2.46,
"learning_rate": 1.6766126740551293e-05,
"loss": 0.7006,
"step": 960
},
{
"epoch": 2.48,
"learning_rate": 1.670929241261722e-05,
"loss": 0.7253,
"step": 970
},
{
"epoch": 2.51,
"learning_rate": 1.665245808468315e-05,
"loss": 0.7042,
"step": 980
},
{
"epoch": 2.53,
"learning_rate": 1.6595623756749078e-05,
"loss": 0.7072,
"step": 990
},
{
"epoch": 2.56,
"learning_rate": 1.6538789428815005e-05,
"loss": 0.7279,
"step": 1000
},
{
"epoch": 2.58,
"learning_rate": 1.6481955100880932e-05,
"loss": 0.7273,
"step": 1010
},
{
"epoch": 2.61,
"learning_rate": 1.6425120772946863e-05,
"loss": 0.7148,
"step": 1020
},
{
"epoch": 2.63,
"learning_rate": 1.636828644501279e-05,
"loss": 0.7136,
"step": 1030
},
{
"epoch": 2.66,
"learning_rate": 1.6311452117078717e-05,
"loss": 0.6997,
"step": 1040
},
{
"epoch": 2.69,
"learning_rate": 1.6254617789144644e-05,
"loss": 0.7115,
"step": 1050
},
{
"epoch": 2.71,
"learning_rate": 1.619778346121057e-05,
"loss": 0.6905,
"step": 1060
},
{
"epoch": 2.74,
"learning_rate": 1.61409491332765e-05,
"loss": 0.725,
"step": 1070
},
{
"epoch": 2.76,
"learning_rate": 1.608411480534243e-05,
"loss": 0.6954,
"step": 1080
},
{
"epoch": 2.79,
"learning_rate": 1.6027280477408356e-05,
"loss": 0.7409,
"step": 1090
},
{
"epoch": 2.81,
"learning_rate": 1.5970446149474283e-05,
"loss": 0.7275,
"step": 1100
},
{
"epoch": 2.84,
"learning_rate": 1.591361182154021e-05,
"loss": 0.7164,
"step": 1110
},
{
"epoch": 2.86,
"learning_rate": 1.585677749360614e-05,
"loss": 0.7317,
"step": 1120
},
{
"epoch": 2.89,
"learning_rate": 1.5799943165672067e-05,
"loss": 0.7129,
"step": 1130
},
{
"epoch": 2.92,
"learning_rate": 1.5743108837737994e-05,
"loss": 0.6981,
"step": 1140
},
{
"epoch": 2.94,
"learning_rate": 1.568627450980392e-05,
"loss": 0.6901,
"step": 1150
},
{
"epoch": 2.97,
"learning_rate": 1.5629440181869852e-05,
"loss": 0.7029,
"step": 1160
},
{
"epoch": 2.99,
"learning_rate": 1.557260585393578e-05,
"loss": 0.7217,
"step": 1170
},
{
"epoch": 3.02,
"learning_rate": 1.5515771526001706e-05,
"loss": 0.6391,
"step": 1180
},
{
"epoch": 3.04,
"learning_rate": 1.5458937198067633e-05,
"loss": 0.6264,
"step": 1190
},
{
"epoch": 3.07,
"learning_rate": 1.540210287013356e-05,
"loss": 0.6258,
"step": 1200
},
{
"epoch": 3.09,
"learning_rate": 1.534526854219949e-05,
"loss": 0.6432,
"step": 1210
},
{
"epoch": 3.12,
"learning_rate": 1.5288434214265418e-05,
"loss": 0.6276,
"step": 1220
},
{
"epoch": 3.15,
"learning_rate": 1.5231599886331345e-05,
"loss": 0.6057,
"step": 1230
},
{
"epoch": 3.17,
"learning_rate": 1.5174765558397273e-05,
"loss": 0.6249,
"step": 1240
},
{
"epoch": 3.2,
"learning_rate": 1.51179312304632e-05,
"loss": 0.6112,
"step": 1250
},
{
"epoch": 3.22,
"learning_rate": 1.506109690252913e-05,
"loss": 0.6232,
"step": 1260
},
{
"epoch": 3.25,
"learning_rate": 1.5004262574595056e-05,
"loss": 0.6441,
"step": 1270
},
{
"epoch": 3.27,
"learning_rate": 1.4947428246660985e-05,
"loss": 0.6036,
"step": 1280
},
{
"epoch": 3.3,
"learning_rate": 1.4890593918726912e-05,
"loss": 0.6509,
"step": 1290
},
{
"epoch": 3.32,
"learning_rate": 1.483375959079284e-05,
"loss": 0.611,
"step": 1300
},
{
"epoch": 3.35,
"learning_rate": 1.4776925262858768e-05,
"loss": 0.6206,
"step": 1310
},
{
"epoch": 3.38,
"learning_rate": 1.4720090934924695e-05,
"loss": 0.6064,
"step": 1320
},
{
"epoch": 3.4,
"learning_rate": 1.4663256606990624e-05,
"loss": 0.6353,
"step": 1330
},
{
"epoch": 3.43,
"learning_rate": 1.4606422279056551e-05,
"loss": 0.6394,
"step": 1340
},
{
"epoch": 3.45,
"learning_rate": 1.454958795112248e-05,
"loss": 0.6163,
"step": 1350
},
{
"epoch": 3.48,
"learning_rate": 1.4492753623188407e-05,
"loss": 0.614,
"step": 1360
},
{
"epoch": 3.5,
"learning_rate": 1.4435919295254336e-05,
"loss": 0.6185,
"step": 1370
},
{
"epoch": 3.53,
"learning_rate": 1.4379084967320263e-05,
"loss": 0.6439,
"step": 1380
},
{
"epoch": 3.55,
"learning_rate": 1.432225063938619e-05,
"loss": 0.6156,
"step": 1390
},
{
"epoch": 3.58,
"learning_rate": 1.4265416311452118e-05,
"loss": 0.6665,
"step": 1400
},
{
"epoch": 3.61,
"learning_rate": 1.4208581983518046e-05,
"loss": 0.6015,
"step": 1410
},
{
"epoch": 3.63,
"learning_rate": 1.4151747655583974e-05,
"loss": 0.5993,
"step": 1420
},
{
"epoch": 3.66,
"learning_rate": 1.4094913327649901e-05,
"loss": 0.6466,
"step": 1430
},
{
"epoch": 3.68,
"learning_rate": 1.403807899971583e-05,
"loss": 0.6505,
"step": 1440
},
{
"epoch": 3.71,
"learning_rate": 1.3981244671781757e-05,
"loss": 0.6328,
"step": 1450
},
{
"epoch": 3.73,
"learning_rate": 1.3924410343847686e-05,
"loss": 0.6456,
"step": 1460
},
{
"epoch": 3.76,
"learning_rate": 1.3867576015913613e-05,
"loss": 0.6334,
"step": 1470
},
{
"epoch": 3.78,
"learning_rate": 1.381074168797954e-05,
"loss": 0.5947,
"step": 1480
},
{
"epoch": 3.81,
"learning_rate": 1.3753907360045469e-05,
"loss": 0.6485,
"step": 1490
},
{
"epoch": 3.84,
"learning_rate": 1.3697073032111396e-05,
"loss": 0.6559,
"step": 1500
},
{
"epoch": 3.86,
"learning_rate": 1.3640238704177325e-05,
"loss": 0.6182,
"step": 1510
},
{
"epoch": 3.89,
"learning_rate": 1.3583404376243252e-05,
"loss": 0.6184,
"step": 1520
},
{
"epoch": 3.91,
"learning_rate": 1.352657004830918e-05,
"loss": 0.6376,
"step": 1530
},
{
"epoch": 3.94,
"learning_rate": 1.3469735720375108e-05,
"loss": 0.6126,
"step": 1540
},
{
"epoch": 3.96,
"learning_rate": 1.3412901392441036e-05,
"loss": 0.6122,
"step": 1550
},
{
"epoch": 3.99,
"learning_rate": 1.3356067064506963e-05,
"loss": 0.6305,
"step": 1560
},
{
"epoch": 4.02,
"learning_rate": 1.329923273657289e-05,
"loss": 0.5854,
"step": 1570
},
{
"epoch": 4.04,
"learning_rate": 1.324239840863882e-05,
"loss": 0.5463,
"step": 1580
},
{
"epoch": 4.07,
"learning_rate": 1.3185564080704746e-05,
"loss": 0.5565,
"step": 1590
},
{
"epoch": 4.09,
"learning_rate": 1.3128729752770675e-05,
"loss": 0.5883,
"step": 1600
},
{
"epoch": 4.12,
"learning_rate": 1.3071895424836602e-05,
"loss": 0.5709,
"step": 1610
},
{
"epoch": 4.14,
"learning_rate": 1.3015061096902531e-05,
"loss": 0.5554,
"step": 1620
},
{
"epoch": 4.17,
"learning_rate": 1.2958226768968458e-05,
"loss": 0.541,
"step": 1630
},
{
"epoch": 4.19,
"learning_rate": 1.2901392441034387e-05,
"loss": 0.5633,
"step": 1640
},
{
"epoch": 4.22,
"learning_rate": 1.2844558113100314e-05,
"loss": 0.53,
"step": 1650
},
{
"epoch": 4.25,
"learning_rate": 1.2787723785166241e-05,
"loss": 0.5695,
"step": 1660
},
{
"epoch": 4.27,
"learning_rate": 1.273088945723217e-05,
"loss": 0.5689,
"step": 1670
},
{
"epoch": 4.3,
"learning_rate": 1.2674055129298097e-05,
"loss": 0.561,
"step": 1680
},
{
"epoch": 4.32,
"learning_rate": 1.2617220801364026e-05,
"loss": 0.5441,
"step": 1690
},
{
"epoch": 4.35,
"learning_rate": 1.2560386473429953e-05,
"loss": 0.5052,
"step": 1700
},
{
"epoch": 4.37,
"learning_rate": 1.2503552145495881e-05,
"loss": 0.5565,
"step": 1710
},
{
"epoch": 4.4,
"learning_rate": 1.2446717817561809e-05,
"loss": 0.569,
"step": 1720
},
{
"epoch": 4.42,
"learning_rate": 1.2389883489627737e-05,
"loss": 0.5383,
"step": 1730
},
{
"epoch": 4.45,
"learning_rate": 1.2333049161693664e-05,
"loss": 0.5603,
"step": 1740
},
{
"epoch": 4.48,
"learning_rate": 1.2276214833759591e-05,
"loss": 0.5318,
"step": 1750
},
{
"epoch": 4.5,
"learning_rate": 1.221938050582552e-05,
"loss": 0.5463,
"step": 1760
},
{
"epoch": 4.53,
"learning_rate": 1.2162546177891447e-05,
"loss": 0.5893,
"step": 1770
},
{
"epoch": 4.55,
"learning_rate": 1.2105711849957376e-05,
"loss": 0.584,
"step": 1780
},
{
"epoch": 4.58,
"learning_rate": 1.2048877522023303e-05,
"loss": 0.5615,
"step": 1790
},
{
"epoch": 4.6,
"learning_rate": 1.1992043194089232e-05,
"loss": 0.5359,
"step": 1800
},
{
"epoch": 4.63,
"learning_rate": 1.1935208866155159e-05,
"loss": 0.5391,
"step": 1810
},
{
"epoch": 4.65,
"learning_rate": 1.1878374538221088e-05,
"loss": 0.588,
"step": 1820
},
{
"epoch": 4.68,
"learning_rate": 1.1821540210287015e-05,
"loss": 0.5374,
"step": 1830
},
{
"epoch": 4.71,
"learning_rate": 1.1764705882352942e-05,
"loss": 0.5946,
"step": 1840
},
{
"epoch": 4.73,
"learning_rate": 1.170787155441887e-05,
"loss": 0.5641,
"step": 1850
},
{
"epoch": 4.76,
"learning_rate": 1.1651037226484798e-05,
"loss": 0.5724,
"step": 1860
},
{
"epoch": 4.78,
"learning_rate": 1.1594202898550726e-05,
"loss": 0.5327,
"step": 1870
},
{
"epoch": 4.81,
"learning_rate": 1.1537368570616654e-05,
"loss": 0.5543,
"step": 1880
},
{
"epoch": 4.83,
"learning_rate": 1.1480534242682582e-05,
"loss": 0.5694,
"step": 1890
},
{
"epoch": 4.86,
"learning_rate": 1.142369991474851e-05,
"loss": 0.5487,
"step": 1900
},
{
"epoch": 4.88,
"learning_rate": 1.1366865586814438e-05,
"loss": 0.5794,
"step": 1910
},
{
"epoch": 4.91,
"learning_rate": 1.1310031258880365e-05,
"loss": 0.5981,
"step": 1920
},
{
"epoch": 4.94,
"learning_rate": 1.1253196930946292e-05,
"loss": 0.5559,
"step": 1930
},
{
"epoch": 4.96,
"learning_rate": 1.1196362603012221e-05,
"loss": 0.5132,
"step": 1940
},
{
"epoch": 4.99,
"learning_rate": 1.1139528275078148e-05,
"loss": 0.5294,
"step": 1950
},
{
"epoch": 5.01,
"learning_rate": 1.1082693947144077e-05,
"loss": 0.5457,
"step": 1960
},
{
"epoch": 5.04,
"learning_rate": 1.1025859619210004e-05,
"loss": 0.5082,
"step": 1970
},
{
"epoch": 5.06,
"learning_rate": 1.0969025291275933e-05,
"loss": 0.4839,
"step": 1980
},
{
"epoch": 5.09,
"learning_rate": 1.091219096334186e-05,
"loss": 0.5042,
"step": 1990
},
{
"epoch": 5.12,
"learning_rate": 1.0855356635407789e-05,
"loss": 0.4862,
"step": 2000
},
{
"epoch": 5.14,
"learning_rate": 1.0798522307473716e-05,
"loss": 0.5559,
"step": 2010
},
{
"epoch": 5.17,
"learning_rate": 1.0741687979539643e-05,
"loss": 0.5107,
"step": 2020
},
{
"epoch": 5.19,
"learning_rate": 1.0684853651605571e-05,
"loss": 0.4919,
"step": 2030
},
{
"epoch": 5.22,
"learning_rate": 1.0628019323671499e-05,
"loss": 0.5198,
"step": 2040
},
{
"epoch": 5.24,
"learning_rate": 1.0571184995737427e-05,
"loss": 0.4967,
"step": 2050
},
{
"epoch": 5.27,
"learning_rate": 1.0514350667803354e-05,
"loss": 0.4895,
"step": 2060
},
{
"epoch": 5.29,
"learning_rate": 1.0457516339869283e-05,
"loss": 0.492,
"step": 2070
},
{
"epoch": 5.32,
"learning_rate": 1.040068201193521e-05,
"loss": 0.494,
"step": 2080
},
{
"epoch": 5.35,
"learning_rate": 1.0343847684001139e-05,
"loss": 0.4864,
"step": 2090
},
{
"epoch": 5.37,
"learning_rate": 1.0287013356067066e-05,
"loss": 0.5043,
"step": 2100
},
{
"epoch": 5.4,
"learning_rate": 1.0230179028132993e-05,
"loss": 0.5088,
"step": 2110
},
{
"epoch": 5.42,
"learning_rate": 1.0173344700198922e-05,
"loss": 0.5035,
"step": 2120
},
{
"epoch": 5.45,
"learning_rate": 1.0116510372264849e-05,
"loss": 0.5052,
"step": 2130
},
{
"epoch": 5.47,
"learning_rate": 1.0059676044330778e-05,
"loss": 0.5096,
"step": 2140
},
{
"epoch": 5.5,
"learning_rate": 1.0002841716396705e-05,
"loss": 0.4916,
"step": 2150
},
{
"epoch": 5.52,
"learning_rate": 9.946007388462632e-06,
"loss": 0.4942,
"step": 2160
},
{
"epoch": 5.55,
"learning_rate": 9.889173060528559e-06,
"loss": 0.4957,
"step": 2170
},
{
"epoch": 5.58,
"learning_rate": 9.832338732594488e-06,
"loss": 0.5078,
"step": 2180
},
{
"epoch": 5.6,
"learning_rate": 9.775504404660415e-06,
"loss": 0.4841,
"step": 2190
},
{
"epoch": 5.63,
"learning_rate": 9.718670076726344e-06,
"loss": 0.4971,
"step": 2200
},
{
"epoch": 5.65,
"learning_rate": 9.66183574879227e-06,
"loss": 0.485,
"step": 2210
},
{
"epoch": 5.68,
"learning_rate": 9.6050014208582e-06,
"loss": 0.4804,
"step": 2220
},
{
"epoch": 5.7,
"learning_rate": 9.548167092924126e-06,
"loss": 0.4811,
"step": 2230
},
{
"epoch": 5.73,
"learning_rate": 9.491332764990055e-06,
"loss": 0.5065,
"step": 2240
},
{
"epoch": 5.75,
"learning_rate": 9.434498437055982e-06,
"loss": 0.4964,
"step": 2250
},
{
"epoch": 5.78,
"learning_rate": 9.37766410912191e-06,
"loss": 0.5003,
"step": 2260
},
{
"epoch": 5.81,
"learning_rate": 9.320829781187838e-06,
"loss": 0.4964,
"step": 2270
},
{
"epoch": 5.83,
"learning_rate": 9.263995453253765e-06,
"loss": 0.5077,
"step": 2280
},
{
"epoch": 5.86,
"learning_rate": 9.207161125319694e-06,
"loss": 0.5197,
"step": 2290
},
{
"epoch": 5.88,
"learning_rate": 9.150326797385621e-06,
"loss": 0.5071,
"step": 2300
},
{
"epoch": 5.91,
"learning_rate": 9.09349246945155e-06,
"loss": 0.4853,
"step": 2310
},
{
"epoch": 5.93,
"learning_rate": 9.036658141517477e-06,
"loss": 0.4799,
"step": 2320
},
{
"epoch": 5.96,
"learning_rate": 8.979823813583406e-06,
"loss": 0.4857,
"step": 2330
},
{
"epoch": 5.98,
"learning_rate": 8.922989485649333e-06,
"loss": 0.4689,
"step": 2340
},
{
"epoch": 6.01,
"learning_rate": 8.86615515771526e-06,
"loss": 0.5046,
"step": 2350
},
{
"epoch": 6.04,
"learning_rate": 8.809320829781189e-06,
"loss": 0.4501,
"step": 2360
},
{
"epoch": 6.06,
"learning_rate": 8.752486501847116e-06,
"loss": 0.4295,
"step": 2370
},
{
"epoch": 6.09,
"learning_rate": 8.695652173913044e-06,
"loss": 0.4513,
"step": 2380
},
{
"epoch": 6.11,
"learning_rate": 8.638817845978971e-06,
"loss": 0.4742,
"step": 2390
},
{
"epoch": 6.14,
"learning_rate": 8.5819835180449e-06,
"loss": 0.4412,
"step": 2400
},
{
"epoch": 6.16,
"learning_rate": 8.525149190110827e-06,
"loss": 0.4546,
"step": 2410
},
{
"epoch": 6.19,
"learning_rate": 8.468314862176756e-06,
"loss": 0.4547,
"step": 2420
},
{
"epoch": 6.21,
"learning_rate": 8.411480534242683e-06,
"loss": 0.4286,
"step": 2430
},
{
"epoch": 6.24,
"learning_rate": 8.35464620630861e-06,
"loss": 0.4475,
"step": 2440
},
{
"epoch": 6.27,
"learning_rate": 8.297811878374539e-06,
"loss": 0.4666,
"step": 2450
},
{
"epoch": 6.29,
"learning_rate": 8.240977550440466e-06,
"loss": 0.4564,
"step": 2460
},
{
"epoch": 6.32,
"learning_rate": 8.184143222506395e-06,
"loss": 0.4504,
"step": 2470
},
{
"epoch": 6.34,
"learning_rate": 8.127308894572322e-06,
"loss": 0.4422,
"step": 2480
},
{
"epoch": 6.37,
"learning_rate": 8.07047456663825e-06,
"loss": 0.4548,
"step": 2490
},
{
"epoch": 6.39,
"learning_rate": 8.013640238704178e-06,
"loss": 0.457,
"step": 2500
},
{
"epoch": 6.42,
"learning_rate": 7.956805910770105e-06,
"loss": 0.4474,
"step": 2510
},
{
"epoch": 6.44,
"learning_rate": 7.899971582836034e-06,
"loss": 0.4462,
"step": 2520
},
{
"epoch": 6.47,
"learning_rate": 7.84313725490196e-06,
"loss": 0.4246,
"step": 2530
},
{
"epoch": 6.5,
"learning_rate": 7.78630292696789e-06,
"loss": 0.4488,
"step": 2540
},
{
"epoch": 6.52,
"learning_rate": 7.729468599033817e-06,
"loss": 0.4616,
"step": 2550
},
{
"epoch": 6.55,
"learning_rate": 7.672634271099745e-06,
"loss": 0.4273,
"step": 2560
},
{
"epoch": 6.57,
"learning_rate": 7.615799943165672e-06,
"loss": 0.4671,
"step": 2570
},
{
"epoch": 6.6,
"learning_rate": 7.5589656152316e-06,
"loss": 0.4738,
"step": 2580
},
{
"epoch": 6.62,
"learning_rate": 7.502131287297528e-06,
"loss": 0.4563,
"step": 2590
},
{
"epoch": 6.65,
"learning_rate": 7.445296959363456e-06,
"loss": 0.4773,
"step": 2600
},
{
"epoch": 6.67,
"learning_rate": 7.388462631429384e-06,
"loss": 0.4678,
"step": 2610
},
{
"epoch": 6.7,
"learning_rate": 7.331628303495312e-06,
"loss": 0.4758,
"step": 2620
},
{
"epoch": 6.73,
"learning_rate": 7.27479397556124e-06,
"loss": 0.4357,
"step": 2630
},
{
"epoch": 6.75,
"learning_rate": 7.217959647627168e-06,
"loss": 0.4732,
"step": 2640
},
{
"epoch": 6.78,
"learning_rate": 7.161125319693095e-06,
"loss": 0.4376,
"step": 2650
},
{
"epoch": 6.8,
"learning_rate": 7.104290991759023e-06,
"loss": 0.4503,
"step": 2660
},
{
"epoch": 6.83,
"learning_rate": 7.047456663824951e-06,
"loss": 0.452,
"step": 2670
},
{
"epoch": 6.85,
"learning_rate": 6.990622335890879e-06,
"loss": 0.4345,
"step": 2680
},
{
"epoch": 6.88,
"learning_rate": 6.9337880079568065e-06,
"loss": 0.4928,
"step": 2690
},
{
"epoch": 6.91,
"learning_rate": 6.8769536800227344e-06,
"loss": 0.4375,
"step": 2700
},
{
"epoch": 6.93,
"learning_rate": 6.820119352088662e-06,
"loss": 0.4412,
"step": 2710
},
{
"epoch": 6.96,
"learning_rate": 6.76328502415459e-06,
"loss": 0.4186,
"step": 2720
},
{
"epoch": 6.98,
"learning_rate": 6.706450696220518e-06,
"loss": 0.433,
"step": 2730
},
{
"epoch": 7.01,
"learning_rate": 6.649616368286445e-06,
"loss": 0.4575,
"step": 2740
},
{
"epoch": 7.03,
"learning_rate": 6.592782040352373e-06,
"loss": 0.4553,
"step": 2750
},
{
"epoch": 7.06,
"learning_rate": 6.535947712418301e-06,
"loss": 0.4123,
"step": 2760
},
{
"epoch": 7.08,
"learning_rate": 6.479113384484229e-06,
"loss": 0.4148,
"step": 2770
},
{
"epoch": 7.11,
"learning_rate": 6.422279056550157e-06,
"loss": 0.4094,
"step": 2780
},
{
"epoch": 7.14,
"learning_rate": 6.365444728616085e-06,
"loss": 0.402,
"step": 2790
},
{
"epoch": 7.16,
"learning_rate": 6.308610400682013e-06,
"loss": 0.4316,
"step": 2800
},
{
"epoch": 7.19,
"learning_rate": 6.251776072747941e-06,
"loss": 0.4182,
"step": 2810
},
{
"epoch": 7.21,
"learning_rate": 6.194941744813869e-06,
"loss": 0.4226,
"step": 2820
},
{
"epoch": 7.24,
"learning_rate": 6.138107416879796e-06,
"loss": 0.4167,
"step": 2830
},
{
"epoch": 7.26,
"learning_rate": 6.081273088945724e-06,
"loss": 0.3813,
"step": 2840
},
{
"epoch": 7.29,
"learning_rate": 6.0244387610116515e-06,
"loss": 0.4131,
"step": 2850
},
{
"epoch": 7.31,
"learning_rate": 5.9676044330775795e-06,
"loss": 0.4336,
"step": 2860
},
{
"epoch": 7.34,
"learning_rate": 5.910770105143507e-06,
"loss": 0.401,
"step": 2870
},
{
"epoch": 7.37,
"learning_rate": 5.853935777209435e-06,
"loss": 0.4051,
"step": 2880
},
{
"epoch": 7.39,
"learning_rate": 5.797101449275363e-06,
"loss": 0.4045,
"step": 2890
},
{
"epoch": 7.42,
"learning_rate": 5.740267121341291e-06,
"loss": 0.4039,
"step": 2900
},
{
"epoch": 7.44,
"learning_rate": 5.683432793407219e-06,
"loss": 0.4158,
"step": 2910
},
{
"epoch": 7.47,
"learning_rate": 5.626598465473146e-06,
"loss": 0.3805,
"step": 2920
},
{
"epoch": 7.49,
"learning_rate": 5.569764137539074e-06,
"loss": 0.4288,
"step": 2930
},
{
"epoch": 7.52,
"learning_rate": 5.512929809605002e-06,
"loss": 0.3905,
"step": 2940
},
{
"epoch": 7.54,
"learning_rate": 5.45609548167093e-06,
"loss": 0.4227,
"step": 2950
},
{
"epoch": 7.57,
"learning_rate": 5.399261153736858e-06,
"loss": 0.437,
"step": 2960
},
{
"epoch": 7.6,
"learning_rate": 5.342426825802786e-06,
"loss": 0.4289,
"step": 2970
},
{
"epoch": 7.62,
"learning_rate": 5.285592497868714e-06,
"loss": 0.4034,
"step": 2980
},
{
"epoch": 7.65,
"learning_rate": 5.2287581699346416e-06,
"loss": 0.4394,
"step": 2990
},
{
"epoch": 7.67,
"learning_rate": 5.1719238420005695e-06,
"loss": 0.4292,
"step": 3000
},
{
"epoch": 7.7,
"learning_rate": 5.1150895140664966e-06,
"loss": 0.4219,
"step": 3010
},
{
"epoch": 7.72,
"learning_rate": 5.0582551861324245e-06,
"loss": 0.4181,
"step": 3020
},
{
"epoch": 7.75,
"learning_rate": 5.001420858198352e-06,
"loss": 0.3983,
"step": 3030
},
{
"epoch": 7.77,
"learning_rate": 4.9445865302642795e-06,
"loss": 0.4167,
"step": 3040
},
{
"epoch": 7.8,
"learning_rate": 4.887752202330207e-06,
"loss": 0.3953,
"step": 3050
},
{
"epoch": 7.83,
"learning_rate": 4.830917874396135e-06,
"loss": 0.4081,
"step": 3060
},
{
"epoch": 7.85,
"learning_rate": 4.774083546462063e-06,
"loss": 0.3817,
"step": 3070
},
{
"epoch": 7.88,
"learning_rate": 4.717249218527991e-06,
"loss": 0.4229,
"step": 3080
},
{
"epoch": 7.9,
"learning_rate": 4.660414890593919e-06,
"loss": 0.41,
"step": 3090
},
{
"epoch": 7.93,
"learning_rate": 4.603580562659847e-06,
"loss": 0.3895,
"step": 3100
},
{
"epoch": 7.95,
"learning_rate": 4.546746234725775e-06,
"loss": 0.4021,
"step": 3110
},
{
"epoch": 7.98,
"learning_rate": 4.489911906791703e-06,
"loss": 0.419,
"step": 3120
},
{
"epoch": 8.01,
"learning_rate": 4.43307757885763e-06,
"loss": 0.4308,
"step": 3130
},
{
"epoch": 8.03,
"learning_rate": 4.376243250923558e-06,
"loss": 0.3788,
"step": 3140
},
{
"epoch": 8.06,
"learning_rate": 4.319408922989486e-06,
"loss": 0.3693,
"step": 3150
},
{
"epoch": 8.08,
"learning_rate": 4.262574595055414e-06,
"loss": 0.3889,
"step": 3160
},
{
"epoch": 8.11,
"learning_rate": 4.205740267121342e-06,
"loss": 0.4061,
"step": 3170
},
{
"epoch": 8.13,
"learning_rate": 4.1489059391872695e-06,
"loss": 0.3893,
"step": 3180
},
{
"epoch": 8.16,
"learning_rate": 4.092071611253197e-06,
"loss": 0.3829,
"step": 3190
},
{
"epoch": 8.18,
"learning_rate": 4.035237283319125e-06,
"loss": 0.3624,
"step": 3200
},
{
"epoch": 8.21,
"learning_rate": 3.978402955385052e-06,
"loss": 0.3947,
"step": 3210
},
{
"epoch": 8.24,
"learning_rate": 3.92156862745098e-06,
"loss": 0.4006,
"step": 3220
},
{
"epoch": 8.26,
"learning_rate": 3.864734299516908e-06,
"loss": 0.3909,
"step": 3230
},
{
"epoch": 8.29,
"learning_rate": 3.807899971582836e-06,
"loss": 0.3744,
"step": 3240
},
{
"epoch": 8.31,
"learning_rate": 3.751065643648764e-06,
"loss": 0.4052,
"step": 3250
},
{
"epoch": 8.34,
"learning_rate": 3.694231315714692e-06,
"loss": 0.3801,
"step": 3260
},
{
"epoch": 8.36,
"learning_rate": 3.63739698778062e-06,
"loss": 0.3822,
"step": 3270
},
{
"epoch": 8.39,
"learning_rate": 3.5805626598465474e-06,
"loss": 0.3856,
"step": 3280
},
{
"epoch": 8.41,
"learning_rate": 3.5237283319124753e-06,
"loss": 0.3644,
"step": 3290
},
{
"epoch": 8.44,
"learning_rate": 3.4668940039784033e-06,
"loss": 0.3844,
"step": 3300
},
{
"epoch": 8.47,
"learning_rate": 3.410059676044331e-06,
"loss": 0.4009,
"step": 3310
},
{
"epoch": 8.49,
"learning_rate": 3.353225348110259e-06,
"loss": 0.3886,
"step": 3320
},
{
"epoch": 8.52,
"learning_rate": 3.2963910201761866e-06,
"loss": 0.3711,
"step": 3330
},
{
"epoch": 8.54,
"learning_rate": 3.2395566922421145e-06,
"loss": 0.3844,
"step": 3340
},
{
"epoch": 8.57,
"learning_rate": 3.1827223643080424e-06,
"loss": 0.3861,
"step": 3350
},
{
"epoch": 8.59,
"learning_rate": 3.1258880363739704e-06,
"loss": 0.3874,
"step": 3360
},
{
"epoch": 8.62,
"learning_rate": 3.069053708439898e-06,
"loss": 0.3617,
"step": 3370
},
{
"epoch": 8.64,
"learning_rate": 3.0122193805058258e-06,
"loss": 0.37,
"step": 3380
},
{
"epoch": 8.67,
"learning_rate": 2.9553850525717537e-06,
"loss": 0.3907,
"step": 3390
},
{
"epoch": 8.7,
"learning_rate": 2.8985507246376816e-06,
"loss": 0.3958,
"step": 3400
},
{
"epoch": 8.72,
"learning_rate": 2.8417163967036095e-06,
"loss": 0.3868,
"step": 3410
},
{
"epoch": 8.75,
"learning_rate": 2.784882068769537e-06,
"loss": 0.3755,
"step": 3420
},
{
"epoch": 8.77,
"learning_rate": 2.728047740835465e-06,
"loss": 0.371,
"step": 3430
},
{
"epoch": 8.8,
"learning_rate": 2.671213412901393e-06,
"loss": 0.3793,
"step": 3440
},
{
"epoch": 8.82,
"learning_rate": 2.6143790849673208e-06,
"loss": 0.3588,
"step": 3450
},
{
"epoch": 8.85,
"learning_rate": 2.5575447570332483e-06,
"loss": 0.3868,
"step": 3460
},
{
"epoch": 8.87,
"learning_rate": 2.500710429099176e-06,
"loss": 0.3826,
"step": 3470
},
{
"epoch": 8.9,
"learning_rate": 2.4438761011651037e-06,
"loss": 0.3868,
"step": 3480
},
{
"epoch": 8.93,
"learning_rate": 2.3870417732310316e-06,
"loss": 0.4038,
"step": 3490
},
{
"epoch": 8.95,
"learning_rate": 2.3302074452969595e-06,
"loss": 0.3527,
"step": 3500
},
{
"epoch": 8.98,
"learning_rate": 2.2733731173628875e-06,
"loss": 0.3766,
"step": 3510
},
{
"epoch": 9.0,
"learning_rate": 2.216538789428815e-06,
"loss": 0.3853,
"step": 3520
},
{
"epoch": 9.03,
"learning_rate": 2.159704461494743e-06,
"loss": 0.36,
"step": 3530
},
{
"epoch": 9.05,
"learning_rate": 2.102870133560671e-06,
"loss": 0.3456,
"step": 3540
},
{
"epoch": 9.08,
"learning_rate": 2.0460358056265987e-06,
"loss": 0.3509,
"step": 3550
},
{
"epoch": 9.1,
"learning_rate": 1.989201477692526e-06,
"loss": 0.3565,
"step": 3560
},
{
"epoch": 9.13,
"learning_rate": 1.932367149758454e-06,
"loss": 0.3911,
"step": 3570
},
{
"epoch": 9.16,
"learning_rate": 1.875532821824382e-06,
"loss": 0.3853,
"step": 3580
},
{
"epoch": 9.18,
"learning_rate": 1.81869849389031e-06,
"loss": 0.3816,
"step": 3590
},
{
"epoch": 9.21,
"learning_rate": 1.7618641659562377e-06,
"loss": 0.3679,
"step": 3600
},
{
"epoch": 9.23,
"learning_rate": 1.7050298380221656e-06,
"loss": 0.365,
"step": 3610
},
{
"epoch": 9.26,
"learning_rate": 1.6481955100880933e-06,
"loss": 0.346,
"step": 3620
},
{
"epoch": 9.28,
"learning_rate": 1.5913611821540212e-06,
"loss": 0.3598,
"step": 3630
},
{
"epoch": 9.31,
"learning_rate": 1.534526854219949e-06,
"loss": 0.3428,
"step": 3640
},
{
"epoch": 9.33,
"learning_rate": 1.4776925262858768e-06,
"loss": 0.3959,
"step": 3650
},
{
"epoch": 9.36,
"learning_rate": 1.4208581983518048e-06,
"loss": 0.3707,
"step": 3660
},
{
"epoch": 9.39,
"learning_rate": 1.3640238704177325e-06,
"loss": 0.3541,
"step": 3670
},
{
"epoch": 9.41,
"learning_rate": 1.3071895424836604e-06,
"loss": 0.3691,
"step": 3680
},
{
"epoch": 9.44,
"learning_rate": 1.250355214549588e-06,
"loss": 0.3736,
"step": 3690
},
{
"epoch": 9.46,
"learning_rate": 1.1935208866155158e-06,
"loss": 0.3496,
"step": 3700
},
{
"epoch": 9.49,
"learning_rate": 1.1366865586814437e-06,
"loss": 0.3369,
"step": 3710
},
{
"epoch": 9.51,
"learning_rate": 1.0798522307473714e-06,
"loss": 0.3959,
"step": 3720
},
{
"epoch": 9.54,
"learning_rate": 1.0230179028132994e-06,
"loss": 0.3604,
"step": 3730
},
{
"epoch": 9.57,
"learning_rate": 9.66183574879227e-07,
"loss": 0.3569,
"step": 3740
},
{
"epoch": 9.59,
"learning_rate": 9.09349246945155e-07,
"loss": 0.3648,
"step": 3750
},
{
"epoch": 9.62,
"learning_rate": 8.525149190110828e-07,
"loss": 0.354,
"step": 3760
},
{
"epoch": 9.64,
"learning_rate": 7.956805910770106e-07,
"loss": 0.3331,
"step": 3770
},
{
"epoch": 9.67,
"learning_rate": 7.388462631429384e-07,
"loss": 0.384,
"step": 3780
},
{
"epoch": 9.69,
"learning_rate": 6.820119352088662e-07,
"loss": 0.3496,
"step": 3790
},
{
"epoch": 9.72,
"learning_rate": 6.25177607274794e-07,
"loss": 0.3779,
"step": 3800
},
{
"epoch": 9.74,
"learning_rate": 5.683432793407219e-07,
"loss": 0.3726,
"step": 3810
},
{
"epoch": 9.77,
"learning_rate": 5.115089514066497e-07,
"loss": 0.3931,
"step": 3820
},
{
"epoch": 9.8,
"learning_rate": 4.546746234725775e-07,
"loss": 0.3714,
"step": 3830
},
{
"epoch": 9.82,
"learning_rate": 3.978402955385053e-07,
"loss": 0.3703,
"step": 3840
},
{
"epoch": 9.85,
"learning_rate": 3.410059676044331e-07,
"loss": 0.3559,
"step": 3850
},
{
"epoch": 9.87,
"learning_rate": 2.8417163967036093e-07,
"loss": 0.3656,
"step": 3860
},
{
"epoch": 9.9,
"learning_rate": 2.2733731173628875e-07,
"loss": 0.3534,
"step": 3870
},
{
"epoch": 9.92,
"learning_rate": 1.7050298380221656e-07,
"loss": 0.3769,
"step": 3880
},
{
"epoch": 9.95,
"learning_rate": 1.1366865586814437e-07,
"loss": 0.3442,
"step": 3890
},
{
"epoch": 9.97,
"learning_rate": 5.6834327934072186e-08,
"loss": 0.3558,
"step": 3900
},
{
"epoch": 10.0,
"learning_rate": 0.0,
"loss": 0.3395,
"step": 3910
},
{
"epoch": 10.0,
"step": 3910,
"total_flos": 1.0186138340339622e+17,
"train_loss": 0.6484309974533823,
"train_runtime": 10912.4775,
"train_samples_per_second": 91.732,
"train_steps_per_second": 0.358
}
],
"max_steps": 3910,
"num_train_epochs": 10,
"total_flos": 1.0186138340339622e+17,
"trial_name": null,
"trial_params": null
}