english-to-latin-v2 / trainer_state.json
BryanFalkowski's picture
Upload trainer_state.json with huggingface_hub
5945a3e
raw
history blame
No virus
100 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.4577316256652815,
"global_step": 8500,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.05,
"learning_rate": 1.978544061302682e-05,
"loss": 9.6964,
"step": 20
},
{
"epoch": 0.09,
"learning_rate": 1.9478927203065136e-05,
"loss": 7.2368,
"step": 40
},
{
"epoch": 0.14,
"learning_rate": 1.917241379310345e-05,
"loss": 6.2605,
"step": 60
},
{
"epoch": 0.18,
"learning_rate": 1.8865900383141766e-05,
"loss": 4.7284,
"step": 80
},
{
"epoch": 0.23,
"learning_rate": 1.855938697318008e-05,
"loss": 4.0496,
"step": 100
},
{
"epoch": 0.28,
"learning_rate": 1.8252873563218392e-05,
"loss": 3.7911,
"step": 120
},
{
"epoch": 0.32,
"learning_rate": 1.7946360153256706e-05,
"loss": 3.6108,
"step": 140
},
{
"epoch": 0.37,
"learning_rate": 1.7639846743295022e-05,
"loss": 3.5038,
"step": 160
},
{
"epoch": 0.41,
"learning_rate": 1.7333333333333336e-05,
"loss": 3.409,
"step": 180
},
{
"epoch": 0.46,
"learning_rate": 1.702681992337165e-05,
"loss": 3.3203,
"step": 200
},
{
"epoch": 0.51,
"learning_rate": 1.6720306513409962e-05,
"loss": 3.2725,
"step": 220
},
{
"epoch": 0.55,
"learning_rate": 1.6413793103448276e-05,
"loss": 3.2151,
"step": 240
},
{
"epoch": 0.6,
"learning_rate": 1.610727969348659e-05,
"loss": 3.1602,
"step": 260
},
{
"epoch": 0.64,
"learning_rate": 1.5800766283524906e-05,
"loss": 3.1164,
"step": 280
},
{
"epoch": 0.69,
"learning_rate": 1.549425287356322e-05,
"loss": 3.0772,
"step": 300
},
{
"epoch": 0.74,
"learning_rate": 1.5187739463601534e-05,
"loss": 3.0443,
"step": 320
},
{
"epoch": 0.78,
"learning_rate": 1.4881226053639847e-05,
"loss": 3.022,
"step": 340
},
{
"epoch": 0.83,
"learning_rate": 1.4574712643678162e-05,
"loss": 3.0014,
"step": 360
},
{
"epoch": 0.87,
"learning_rate": 1.4268199233716476e-05,
"loss": 2.9582,
"step": 380
},
{
"epoch": 0.92,
"learning_rate": 1.396168582375479e-05,
"loss": 2.9272,
"step": 400
},
{
"epoch": 0.97,
"learning_rate": 1.3655172413793106e-05,
"loss": 2.8951,
"step": 420
},
{
"epoch": 1.01,
"learning_rate": 1.3348659003831419e-05,
"loss": 2.9101,
"step": 440
},
{
"epoch": 1.06,
"learning_rate": 1.3042145593869732e-05,
"loss": 2.8377,
"step": 460
},
{
"epoch": 1.1,
"learning_rate": 1.2735632183908047e-05,
"loss": 2.8211,
"step": 480
},
{
"epoch": 1.15,
"learning_rate": 1.242911877394636e-05,
"loss": 2.7957,
"step": 500
},
{
"epoch": 0.03,
"learning_rate": 1.981905326081103e-05,
"loss": 2.8029,
"step": 510
},
{
"epoch": 0.03,
"learning_rate": 1.9815463047731884e-05,
"loss": 2.8451,
"step": 520
},
{
"epoch": 0.03,
"learning_rate": 1.981187283465274e-05,
"loss": 2.9085,
"step": 530
},
{
"epoch": 0.03,
"learning_rate": 1.9808282621573592e-05,
"loss": 2.9984,
"step": 540
},
{
"epoch": 0.03,
"learning_rate": 1.9804692408494444e-05,
"loss": 2.8696,
"step": 550
},
{
"epoch": 0.03,
"learning_rate": 1.98011021954153e-05,
"loss": 2.8919,
"step": 560
},
{
"epoch": 0.03,
"learning_rate": 1.9797511982336152e-05,
"loss": 2.8115,
"step": 570
},
{
"epoch": 0.03,
"learning_rate": 1.9793921769257008e-05,
"loss": 2.7233,
"step": 580
},
{
"epoch": 0.03,
"learning_rate": 1.979033155617786e-05,
"loss": 2.9758,
"step": 590
},
{
"epoch": 0.03,
"learning_rate": 1.9786741343098715e-05,
"loss": 2.8503,
"step": 600
},
{
"epoch": 0.03,
"learning_rate": 1.9783151130019567e-05,
"loss": 2.7586,
"step": 610
},
{
"epoch": 0.03,
"learning_rate": 1.977956091694042e-05,
"loss": 2.898,
"step": 620
},
{
"epoch": 0.03,
"learning_rate": 1.9775970703861275e-05,
"loss": 2.7126,
"step": 630
},
{
"epoch": 0.03,
"learning_rate": 1.977238049078213e-05,
"loss": 2.8214,
"step": 640
},
{
"epoch": 0.04,
"learning_rate": 1.9768790277702983e-05,
"loss": 2.8384,
"step": 650
},
{
"epoch": 0.04,
"learning_rate": 1.976520006462384e-05,
"loss": 2.8592,
"step": 660
},
{
"epoch": 0.04,
"learning_rate": 1.976160985154469e-05,
"loss": 2.7752,
"step": 670
},
{
"epoch": 0.04,
"learning_rate": 1.9758019638465543e-05,
"loss": 2.9652,
"step": 680
},
{
"epoch": 0.04,
"learning_rate": 1.9754429425386398e-05,
"loss": 2.8639,
"step": 690
},
{
"epoch": 0.04,
"learning_rate": 1.975083921230725e-05,
"loss": 3.5679,
"step": 700
},
{
"epoch": 0.04,
"learning_rate": 1.9747248999228106e-05,
"loss": 3.842,
"step": 710
},
{
"epoch": 0.04,
"learning_rate": 1.974365878614896e-05,
"loss": 2.9028,
"step": 720
},
{
"epoch": 0.04,
"learning_rate": 1.9740068573069814e-05,
"loss": 2.9596,
"step": 730
},
{
"epoch": 0.04,
"learning_rate": 1.9736478359990666e-05,
"loss": 3.5545,
"step": 740
},
{
"epoch": 0.04,
"learning_rate": 1.9733247168219437e-05,
"loss": 9.8081,
"step": 750
},
{
"epoch": 0.04,
"learning_rate": 1.972965695514029e-05,
"loss": 4.6766,
"step": 760
},
{
"epoch": 0.04,
"learning_rate": 1.9726784784676972e-05,
"loss": 6.1851,
"step": 770
},
{
"epoch": 0.04,
"learning_rate": 1.9723194571597824e-05,
"loss": 15.1653,
"step": 780
},
{
"epoch": 0.04,
"learning_rate": 1.971960435851868e-05,
"loss": 16.7837,
"step": 790
},
{
"epoch": 0.04,
"learning_rate": 1.9716014145439532e-05,
"loss": 16.0732,
"step": 800
},
{
"epoch": 0.04,
"learning_rate": 1.9712423932360387e-05,
"loss": 15.0973,
"step": 810
},
{
"epoch": 0.04,
"learning_rate": 1.9708833719281243e-05,
"loss": 14.4623,
"step": 820
},
{
"epoch": 0.04,
"learning_rate": 1.9705243506202095e-05,
"loss": 13.6565,
"step": 830
},
{
"epoch": 0.05,
"learning_rate": 1.9701653293122947e-05,
"loss": 13.3604,
"step": 840
},
{
"epoch": 0.05,
"learning_rate": 1.9698063080043803e-05,
"loss": 12.3429,
"step": 850
},
{
"epoch": 0.05,
"learning_rate": 1.9694472866964655e-05,
"loss": 11.7149,
"step": 860
},
{
"epoch": 0.05,
"learning_rate": 1.969088265388551e-05,
"loss": 8.7238,
"step": 870
},
{
"epoch": 0.05,
"learning_rate": 1.9687292440806363e-05,
"loss": 7.7324,
"step": 880
},
{
"epoch": 0.05,
"learning_rate": 1.9683702227727218e-05,
"loss": 9.1205,
"step": 890
},
{
"epoch": 0.05,
"learning_rate": 1.968011201464807e-05,
"loss": 5.5195,
"step": 900
},
{
"epoch": 0.05,
"learning_rate": 1.9676521801568923e-05,
"loss": 4.6745,
"step": 910
},
{
"epoch": 0.05,
"learning_rate": 1.9672931588489778e-05,
"loss": 3.2255,
"step": 920
},
{
"epoch": 0.05,
"learning_rate": 1.9669341375410634e-05,
"loss": 3.071,
"step": 930
},
{
"epoch": 0.05,
"learning_rate": 1.9665751162331486e-05,
"loss": 7.5739,
"step": 940
},
{
"epoch": 0.05,
"learning_rate": 1.966216094925234e-05,
"loss": 10.6607,
"step": 950
},
{
"epoch": 0.05,
"learning_rate": 1.9658570736173194e-05,
"loss": 9.1666,
"step": 960
},
{
"epoch": 0.05,
"learning_rate": 1.9654980523094046e-05,
"loss": 3.3352,
"step": 970
},
{
"epoch": 0.05,
"learning_rate": 1.96513903100149e-05,
"loss": 3.0204,
"step": 980
},
{
"epoch": 0.05,
"learning_rate": 1.9647800096935753e-05,
"loss": 3.0261,
"step": 990
},
{
"epoch": 0.05,
"learning_rate": 1.964420988385661e-05,
"loss": 2.9761,
"step": 1000
},
{
"epoch": 0.05,
"learning_rate": 1.9640619670777464e-05,
"loss": 2.9513,
"step": 1010
},
{
"epoch": 0.05,
"learning_rate": 1.9637029457698317e-05,
"loss": 2.9555,
"step": 1020
},
{
"epoch": 0.06,
"learning_rate": 1.963343924461917e-05,
"loss": 2.8127,
"step": 1030
},
{
"epoch": 0.06,
"learning_rate": 1.962984903154002e-05,
"loss": 2.8066,
"step": 1040
},
{
"epoch": 0.06,
"learning_rate": 1.9626258818460876e-05,
"loss": 2.8596,
"step": 1050
},
{
"epoch": 0.06,
"learning_rate": 1.9622668605381732e-05,
"loss": 2.8612,
"step": 1060
},
{
"epoch": 0.06,
"learning_rate": 1.9619078392302584e-05,
"loss": 2.8639,
"step": 1070
},
{
"epoch": 0.06,
"learning_rate": 1.961548817922344e-05,
"loss": 2.8358,
"step": 1080
},
{
"epoch": 0.06,
"learning_rate": 1.9611897966144292e-05,
"loss": 2.951,
"step": 1090
},
{
"epoch": 0.06,
"learning_rate": 1.9608307753065144e-05,
"loss": 2.961,
"step": 1100
},
{
"epoch": 0.06,
"learning_rate": 1.9604717539986e-05,
"loss": 2.7049,
"step": 1110
},
{
"epoch": 0.06,
"learning_rate": 1.9601127326906852e-05,
"loss": 2.9308,
"step": 1120
},
{
"epoch": 0.06,
"learning_rate": 1.9597537113827707e-05,
"loss": 2.9189,
"step": 1130
},
{
"epoch": 0.06,
"learning_rate": 1.9593946900748563e-05,
"loss": 2.8701,
"step": 1140
},
{
"epoch": 0.06,
"learning_rate": 1.9590356687669415e-05,
"loss": 3.0191,
"step": 1150
},
{
"epoch": 0.06,
"learning_rate": 1.9586766474590267e-05,
"loss": 2.851,
"step": 1160
},
{
"epoch": 0.06,
"learning_rate": 1.9583176261511123e-05,
"loss": 2.9326,
"step": 1170
},
{
"epoch": 0.06,
"learning_rate": 1.9579586048431975e-05,
"loss": 2.916,
"step": 1180
},
{
"epoch": 0.06,
"learning_rate": 1.957599583535283e-05,
"loss": 2.8135,
"step": 1190
},
{
"epoch": 0.06,
"learning_rate": 1.9572405622273683e-05,
"loss": 2.8072,
"step": 1200
},
{
"epoch": 0.07,
"learning_rate": 1.9568815409194538e-05,
"loss": 2.7695,
"step": 1210
},
{
"epoch": 0.07,
"learning_rate": 1.956522519611539e-05,
"loss": 2.7315,
"step": 1220
},
{
"epoch": 0.07,
"learning_rate": 1.9561634983036242e-05,
"loss": 2.7673,
"step": 1230
},
{
"epoch": 0.07,
"learning_rate": 1.9558044769957098e-05,
"loss": 2.8883,
"step": 1240
},
{
"epoch": 0.07,
"learning_rate": 1.9554454556877954e-05,
"loss": 2.6834,
"step": 1250
},
{
"epoch": 0.07,
"learning_rate": 1.9550864343798806e-05,
"loss": 2.8404,
"step": 1260
},
{
"epoch": 0.07,
"learning_rate": 1.954727413071966e-05,
"loss": 2.7066,
"step": 1270
},
{
"epoch": 0.07,
"learning_rate": 1.9543683917640513e-05,
"loss": 2.8951,
"step": 1280
},
{
"epoch": 0.07,
"learning_rate": 1.9540093704561366e-05,
"loss": 2.7959,
"step": 1290
},
{
"epoch": 0.07,
"learning_rate": 1.953650349148222e-05,
"loss": 2.7034,
"step": 1300
},
{
"epoch": 0.07,
"learning_rate": 1.9532913278403073e-05,
"loss": 2.838,
"step": 1310
},
{
"epoch": 0.07,
"learning_rate": 1.952932306532393e-05,
"loss": 2.8125,
"step": 1320
},
{
"epoch": 0.07,
"learning_rate": 1.9525732852244784e-05,
"loss": 2.7806,
"step": 1330
},
{
"epoch": 0.07,
"learning_rate": 1.9522142639165636e-05,
"loss": 2.8089,
"step": 1340
},
{
"epoch": 0.07,
"learning_rate": 1.951855242608649e-05,
"loss": 2.8305,
"step": 1350
},
{
"epoch": 0.07,
"learning_rate": 1.9514962213007344e-05,
"loss": 2.7721,
"step": 1360
},
{
"epoch": 0.07,
"learning_rate": 1.9511371999928196e-05,
"loss": 2.8546,
"step": 1370
},
{
"epoch": 0.07,
"learning_rate": 1.9507781786849052e-05,
"loss": 2.6916,
"step": 1380
},
{
"epoch": 0.07,
"learning_rate": 1.9504191573769904e-05,
"loss": 2.8091,
"step": 1390
},
{
"epoch": 0.08,
"learning_rate": 1.950060136069076e-05,
"loss": 2.7681,
"step": 1400
},
{
"epoch": 0.08,
"learning_rate": 1.9497011147611612e-05,
"loss": 2.8239,
"step": 1410
},
{
"epoch": 0.08,
"learning_rate": 1.9493420934532464e-05,
"loss": 2.8446,
"step": 1420
},
{
"epoch": 0.08,
"learning_rate": 1.948983072145332e-05,
"loss": 2.8963,
"step": 1430
},
{
"epoch": 0.08,
"learning_rate": 1.9486240508374175e-05,
"loss": 2.8163,
"step": 1440
},
{
"epoch": 0.08,
"learning_rate": 1.9482650295295027e-05,
"loss": 2.7866,
"step": 1450
},
{
"epoch": 0.08,
"learning_rate": 1.9479060082215883e-05,
"loss": 2.78,
"step": 1460
},
{
"epoch": 0.08,
"learning_rate": 1.9475469869136735e-05,
"loss": 2.7873,
"step": 1470
},
{
"epoch": 0.08,
"learning_rate": 1.9471879656057587e-05,
"loss": 2.7717,
"step": 1480
},
{
"epoch": 0.08,
"learning_rate": 1.9468289442978443e-05,
"loss": 3.0517,
"step": 1490
},
{
"epoch": 0.08,
"learning_rate": 1.9464699229899295e-05,
"loss": 2.8272,
"step": 1500
},
{
"epoch": 0.08,
"learning_rate": 1.946110901682015e-05,
"loss": 2.8358,
"step": 1510
},
{
"epoch": 0.08,
"learning_rate": 1.9457518803741006e-05,
"loss": 2.6929,
"step": 1520
},
{
"epoch": 0.08,
"learning_rate": 1.9453928590661858e-05,
"loss": 2.7598,
"step": 1530
},
{
"epoch": 0.08,
"learning_rate": 1.945033837758271e-05,
"loss": 2.7529,
"step": 1540
},
{
"epoch": 0.08,
"learning_rate": 1.9446748164503562e-05,
"loss": 2.6802,
"step": 1550
},
{
"epoch": 0.08,
"learning_rate": 1.9443157951424418e-05,
"loss": 2.8318,
"step": 1560
},
{
"epoch": 0.08,
"learning_rate": 1.9439567738345273e-05,
"loss": 2.8196,
"step": 1570
},
{
"epoch": 0.09,
"learning_rate": 1.9435977525266125e-05,
"loss": 2.7937,
"step": 1580
},
{
"epoch": 0.09,
"learning_rate": 1.943238731218698e-05,
"loss": 2.7443,
"step": 1590
},
{
"epoch": 0.09,
"learning_rate": 1.9428797099107833e-05,
"loss": 2.7506,
"step": 1600
},
{
"epoch": 0.09,
"learning_rate": 1.9425206886028685e-05,
"loss": 2.8196,
"step": 1610
},
{
"epoch": 0.09,
"learning_rate": 1.942161667294954e-05,
"loss": 2.8199,
"step": 1620
},
{
"epoch": 0.09,
"learning_rate": 1.9418026459870393e-05,
"loss": 2.7978,
"step": 1630
},
{
"epoch": 0.09,
"learning_rate": 1.941443624679125e-05,
"loss": 2.835,
"step": 1640
},
{
"epoch": 0.09,
"learning_rate": 1.9410846033712104e-05,
"loss": 2.6939,
"step": 1650
},
{
"epoch": 0.09,
"learning_rate": 1.9407255820632956e-05,
"loss": 2.6551,
"step": 1660
},
{
"epoch": 0.09,
"learning_rate": 1.940366560755381e-05,
"loss": 2.9222,
"step": 1670
},
{
"epoch": 0.09,
"learning_rate": 1.9400075394474664e-05,
"loss": 2.5975,
"step": 1680
},
{
"epoch": 0.09,
"learning_rate": 1.9396485181395516e-05,
"loss": 2.7544,
"step": 1690
},
{
"epoch": 0.09,
"learning_rate": 1.9392894968316372e-05,
"loss": 2.6074,
"step": 1700
},
{
"epoch": 0.09,
"learning_rate": 1.9389304755237227e-05,
"loss": 2.6687,
"step": 1710
},
{
"epoch": 0.09,
"learning_rate": 1.938571454215808e-05,
"loss": 2.7668,
"step": 1720
},
{
"epoch": 0.09,
"learning_rate": 1.938212432907893e-05,
"loss": 2.8017,
"step": 1730
},
{
"epoch": 0.09,
"learning_rate": 1.9378534115999784e-05,
"loss": 2.8552,
"step": 1740
},
{
"epoch": 0.09,
"learning_rate": 1.937494390292064e-05,
"loss": 2.6845,
"step": 1750
},
{
"epoch": 0.09,
"learning_rate": 1.9371353689841495e-05,
"loss": 2.7429,
"step": 1760
},
{
"epoch": 0.1,
"learning_rate": 1.9367763476762347e-05,
"loss": 2.8029,
"step": 1770
},
{
"epoch": 0.1,
"learning_rate": 1.9364173263683203e-05,
"loss": 2.7683,
"step": 1780
},
{
"epoch": 0.1,
"learning_rate": 1.9360583050604055e-05,
"loss": 2.8638,
"step": 1790
},
{
"epoch": 0.1,
"learning_rate": 1.9356992837524907e-05,
"loss": 2.75,
"step": 1800
},
{
"epoch": 0.1,
"learning_rate": 1.9353402624445762e-05,
"loss": 2.6314,
"step": 1810
},
{
"epoch": 0.1,
"learning_rate": 1.9349812411366615e-05,
"loss": 2.766,
"step": 1820
},
{
"epoch": 0.1,
"learning_rate": 1.934622219828747e-05,
"loss": 2.6745,
"step": 1830
},
{
"epoch": 0.1,
"learning_rate": 1.9342631985208326e-05,
"loss": 2.7482,
"step": 1840
},
{
"epoch": 0.1,
"learning_rate": 1.9339041772129178e-05,
"loss": 2.723,
"step": 1850
},
{
"epoch": 0.1,
"learning_rate": 1.933545155905003e-05,
"loss": 2.7507,
"step": 1860
},
{
"epoch": 0.1,
"learning_rate": 1.9331861345970885e-05,
"loss": 2.7127,
"step": 1870
},
{
"epoch": 0.1,
"learning_rate": 1.9328271132891738e-05,
"loss": 2.6458,
"step": 1880
},
{
"epoch": 0.1,
"learning_rate": 1.9324680919812593e-05,
"loss": 2.7481,
"step": 1890
},
{
"epoch": 0.1,
"learning_rate": 1.9321090706733445e-05,
"loss": 2.6548,
"step": 1900
},
{
"epoch": 0.1,
"learning_rate": 1.93175004936543e-05,
"loss": 2.6856,
"step": 1910
},
{
"epoch": 0.1,
"learning_rate": 1.9313910280575153e-05,
"loss": 2.7039,
"step": 1920
},
{
"epoch": 0.1,
"learning_rate": 1.9310320067496005e-05,
"loss": 2.7392,
"step": 1930
},
{
"epoch": 0.1,
"learning_rate": 1.930672985441686e-05,
"loss": 2.6267,
"step": 1940
},
{
"epoch": 0.11,
"learning_rate": 1.9303139641337716e-05,
"loss": 2.7609,
"step": 1950
},
{
"epoch": 0.11,
"learning_rate": 1.929954942825857e-05,
"loss": 2.7663,
"step": 1960
},
{
"epoch": 0.11,
"learning_rate": 1.9295959215179424e-05,
"loss": 2.7452,
"step": 1970
},
{
"epoch": 0.11,
"learning_rate": 1.9292369002100276e-05,
"loss": 2.791,
"step": 1980
},
{
"epoch": 0.11,
"learning_rate": 1.9288778789021128e-05,
"loss": 2.8045,
"step": 1990
},
{
"epoch": 0.11,
"learning_rate": 1.9285188575941984e-05,
"loss": 2.6488,
"step": 2000
},
{
"epoch": 0.11,
"learning_rate": 1.9281598362862836e-05,
"loss": 2.6689,
"step": 2010
},
{
"epoch": 0.11,
"learning_rate": 1.927800814978369e-05,
"loss": 2.5595,
"step": 2020
},
{
"epoch": 0.11,
"learning_rate": 1.9274417936704547e-05,
"loss": 2.6463,
"step": 2030
},
{
"epoch": 0.11,
"learning_rate": 1.92708277236254e-05,
"loss": 2.7077,
"step": 2040
},
{
"epoch": 0.11,
"learning_rate": 1.926723751054625e-05,
"loss": 2.726,
"step": 2050
},
{
"epoch": 0.11,
"learning_rate": 1.9263647297467104e-05,
"loss": 2.6916,
"step": 2060
},
{
"epoch": 0.11,
"learning_rate": 1.926005708438796e-05,
"loss": 2.7012,
"step": 2070
},
{
"epoch": 0.11,
"learning_rate": 1.9256466871308815e-05,
"loss": 2.601,
"step": 2080
},
{
"epoch": 0.11,
"learning_rate": 1.9252876658229667e-05,
"loss": 2.751,
"step": 2090
},
{
"epoch": 0.11,
"learning_rate": 1.9249286445150522e-05,
"loss": 2.5344,
"step": 2100
},
{
"epoch": 0.11,
"learning_rate": 1.9245696232071375e-05,
"loss": 2.589,
"step": 2110
},
{
"epoch": 0.11,
"learning_rate": 1.9242106018992227e-05,
"loss": 2.6888,
"step": 2120
},
{
"epoch": 0.11,
"learning_rate": 1.9238515805913082e-05,
"loss": 2.6296,
"step": 2130
},
{
"epoch": 0.12,
"learning_rate": 1.9234925592833938e-05,
"loss": 2.5918,
"step": 2140
},
{
"epoch": 0.12,
"learning_rate": 1.923133537975479e-05,
"loss": 2.5673,
"step": 2150
},
{
"epoch": 0.12,
"learning_rate": 1.9227745166675645e-05,
"loss": 2.5967,
"step": 2160
},
{
"epoch": 0.12,
"learning_rate": 1.9224154953596498e-05,
"loss": 2.6612,
"step": 2170
},
{
"epoch": 0.12,
"learning_rate": 1.922056474051735e-05,
"loss": 2.7359,
"step": 2180
},
{
"epoch": 0.12,
"learning_rate": 1.9216974527438205e-05,
"loss": 2.593,
"step": 2190
},
{
"epoch": 0.12,
"learning_rate": 1.9213384314359057e-05,
"loss": 2.63,
"step": 2200
},
{
"epoch": 0.12,
"learning_rate": 1.9209794101279913e-05,
"loss": 2.6424,
"step": 2210
},
{
"epoch": 0.12,
"learning_rate": 1.920620388820077e-05,
"loss": 2.7708,
"step": 2220
},
{
"epoch": 0.12,
"learning_rate": 1.920261367512162e-05,
"loss": 2.6115,
"step": 2230
},
{
"epoch": 0.12,
"learning_rate": 1.9199023462042473e-05,
"loss": 2.7522,
"step": 2240
},
{
"epoch": 0.12,
"learning_rate": 1.9195433248963325e-05,
"loss": 2.7271,
"step": 2250
},
{
"epoch": 0.12,
"learning_rate": 1.919184303588418e-05,
"loss": 2.5119,
"step": 2260
},
{
"epoch": 0.12,
"learning_rate": 1.9188252822805036e-05,
"loss": 2.7152,
"step": 2270
},
{
"epoch": 0.12,
"learning_rate": 1.9184662609725888e-05,
"loss": 2.6712,
"step": 2280
},
{
"epoch": 0.12,
"learning_rate": 1.9181072396646744e-05,
"loss": 2.6579,
"step": 2290
},
{
"epoch": 0.12,
"learning_rate": 1.9177482183567596e-05,
"loss": 2.556,
"step": 2300
},
{
"epoch": 0.12,
"learning_rate": 1.9173891970488448e-05,
"loss": 2.755,
"step": 2310
},
{
"epoch": 0.12,
"learning_rate": 1.9170301757409304e-05,
"loss": 2.5799,
"step": 2320
},
{
"epoch": 0.13,
"learning_rate": 1.9166711544330156e-05,
"loss": 2.8061,
"step": 2330
},
{
"epoch": 0.13,
"learning_rate": 1.916312133125101e-05,
"loss": 2.5838,
"step": 2340
},
{
"epoch": 0.13,
"learning_rate": 1.9159531118171867e-05,
"loss": 2.5489,
"step": 2350
},
{
"epoch": 0.13,
"learning_rate": 1.915594090509272e-05,
"loss": 2.6664,
"step": 2360
},
{
"epoch": 0.13,
"learning_rate": 1.915235069201357e-05,
"loss": 2.638,
"step": 2370
},
{
"epoch": 0.13,
"learning_rate": 1.9148760478934427e-05,
"loss": 2.6145,
"step": 2380
},
{
"epoch": 0.13,
"learning_rate": 1.914517026585528e-05,
"loss": 2.6475,
"step": 2390
},
{
"epoch": 0.13,
"learning_rate": 1.9141580052776134e-05,
"loss": 2.731,
"step": 2400
},
{
"epoch": 0.13,
"learning_rate": 1.9137989839696987e-05,
"loss": 2.624,
"step": 2410
},
{
"epoch": 0.13,
"learning_rate": 1.9134399626617842e-05,
"loss": 2.6253,
"step": 2420
},
{
"epoch": 0.13,
"learning_rate": 1.9130809413538694e-05,
"loss": 2.6922,
"step": 2430
},
{
"epoch": 0.13,
"learning_rate": 1.9127219200459546e-05,
"loss": 2.7505,
"step": 2440
},
{
"epoch": 0.13,
"learning_rate": 1.9123628987380402e-05,
"loss": 2.7062,
"step": 2450
},
{
"epoch": 0.13,
"learning_rate": 1.9120038774301258e-05,
"loss": 2.6434,
"step": 2460
},
{
"epoch": 0.13,
"learning_rate": 1.911644856122211e-05,
"loss": 2.7078,
"step": 2470
},
{
"epoch": 0.13,
"learning_rate": 1.9112858348142965e-05,
"loss": 2.66,
"step": 2480
},
{
"epoch": 0.13,
"learning_rate": 1.9109268135063817e-05,
"loss": 2.7149,
"step": 2490
},
{
"epoch": 0.13,
"learning_rate": 1.910567792198467e-05,
"loss": 2.6047,
"step": 2500
},
{
"epoch": 0.14,
"learning_rate": 1.9102087708905525e-05,
"loss": 2.6814,
"step": 2510
},
{
"epoch": 0.14,
"learning_rate": 1.9098497495826377e-05,
"loss": 2.5845,
"step": 2520
},
{
"epoch": 0.14,
"learning_rate": 1.9094907282747233e-05,
"loss": 2.5178,
"step": 2530
},
{
"epoch": 0.14,
"learning_rate": 1.909131706966809e-05,
"loss": 2.5956,
"step": 2540
},
{
"epoch": 0.14,
"learning_rate": 1.908772685658894e-05,
"loss": 2.5979,
"step": 2550
},
{
"epoch": 0.14,
"learning_rate": 1.9084136643509793e-05,
"loss": 2.7305,
"step": 2560
},
{
"epoch": 0.14,
"learning_rate": 1.9080546430430648e-05,
"loss": 2.7569,
"step": 2570
},
{
"epoch": 0.14,
"learning_rate": 1.90769562173515e-05,
"loss": 2.5463,
"step": 2580
},
{
"epoch": 0.14,
"learning_rate": 1.9073366004272356e-05,
"loss": 2.6905,
"step": 2590
},
{
"epoch": 0.14,
"learning_rate": 1.9069775791193208e-05,
"loss": 2.6339,
"step": 2600
},
{
"epoch": 0.14,
"learning_rate": 1.9066185578114064e-05,
"loss": 2.6937,
"step": 2610
},
{
"epoch": 0.14,
"learning_rate": 1.9062595365034916e-05,
"loss": 2.6612,
"step": 2620
},
{
"epoch": 0.14,
"learning_rate": 1.9059005151955768e-05,
"loss": 2.6495,
"step": 2630
},
{
"epoch": 0.14,
"learning_rate": 1.9055414938876624e-05,
"loss": 2.6662,
"step": 2640
},
{
"epoch": 0.14,
"learning_rate": 1.905182472579748e-05,
"loss": 2.6109,
"step": 2650
},
{
"epoch": 0.14,
"learning_rate": 1.904823451271833e-05,
"loss": 2.6967,
"step": 2660
},
{
"epoch": 0.14,
"learning_rate": 1.9044644299639187e-05,
"loss": 2.6573,
"step": 2670
},
{
"epoch": 0.14,
"learning_rate": 1.904105408656004e-05,
"loss": 2.6437,
"step": 2680
},
{
"epoch": 0.14,
"learning_rate": 1.903746387348089e-05,
"loss": 2.6396,
"step": 2690
},
{
"epoch": 0.15,
"learning_rate": 1.9033873660401747e-05,
"loss": 2.5431,
"step": 2700
},
{
"epoch": 0.15,
"learning_rate": 1.90302834473226e-05,
"loss": 2.6333,
"step": 2710
},
{
"epoch": 0.15,
"learning_rate": 1.9026693234243454e-05,
"loss": 2.562,
"step": 2720
},
{
"epoch": 0.15,
"learning_rate": 1.902310302116431e-05,
"loss": 2.6076,
"step": 2730
},
{
"epoch": 0.15,
"learning_rate": 1.9019512808085162e-05,
"loss": 2.7656,
"step": 2740
},
{
"epoch": 0.15,
"learning_rate": 1.9015922595006014e-05,
"loss": 2.6859,
"step": 2750
},
{
"epoch": 0.15,
"learning_rate": 1.9012332381926866e-05,
"loss": 2.5723,
"step": 2760
},
{
"epoch": 0.15,
"learning_rate": 1.9008742168847722e-05,
"loss": 2.5823,
"step": 2770
},
{
"epoch": 0.15,
"learning_rate": 1.9005151955768577e-05,
"loss": 2.6527,
"step": 2780
},
{
"epoch": 0.15,
"learning_rate": 1.900156174268943e-05,
"loss": 2.6323,
"step": 2790
},
{
"epoch": 0.15,
"learning_rate": 1.8997971529610285e-05,
"loss": 2.5642,
"step": 2800
},
{
"epoch": 0.15,
"learning_rate": 1.8994381316531137e-05,
"loss": 2.706,
"step": 2810
},
{
"epoch": 0.15,
"learning_rate": 1.899079110345199e-05,
"loss": 2.4924,
"step": 2820
},
{
"epoch": 0.15,
"learning_rate": 1.8987200890372845e-05,
"loss": 2.6242,
"step": 2830
},
{
"epoch": 0.15,
"learning_rate": 1.8983610677293697e-05,
"loss": 2.503,
"step": 2840
},
{
"epoch": 0.15,
"learning_rate": 1.8980020464214553e-05,
"loss": 2.6463,
"step": 2850
},
{
"epoch": 0.15,
"learning_rate": 1.8976430251135408e-05,
"loss": 2.6109,
"step": 2860
},
{
"epoch": 0.15,
"learning_rate": 1.897284003805626e-05,
"loss": 2.6042,
"step": 2870
},
{
"epoch": 0.16,
"learning_rate": 1.8969249824977113e-05,
"loss": 2.6578,
"step": 2880
},
{
"epoch": 0.16,
"learning_rate": 1.8965659611897968e-05,
"loss": 2.6372,
"step": 2890
},
{
"epoch": 0.16,
"learning_rate": 1.896206939881882e-05,
"loss": 2.5956,
"step": 2900
},
{
"epoch": 0.16,
"learning_rate": 1.8958479185739676e-05,
"loss": 2.5386,
"step": 2910
},
{
"epoch": 0.16,
"learning_rate": 1.8954888972660528e-05,
"loss": 2.6261,
"step": 2920
},
{
"epoch": 0.16,
"learning_rate": 1.8951298759581383e-05,
"loss": 2.5414,
"step": 2930
},
{
"epoch": 0.16,
"learning_rate": 1.8947708546502236e-05,
"loss": 2.6658,
"step": 2940
},
{
"epoch": 0.16,
"learning_rate": 1.8944118333423088e-05,
"loss": 2.5249,
"step": 2950
},
{
"epoch": 0.16,
"learning_rate": 1.8940528120343943e-05,
"loss": 2.5465,
"step": 2960
},
{
"epoch": 0.16,
"learning_rate": 1.89369379072648e-05,
"loss": 2.5472,
"step": 2970
},
{
"epoch": 0.16,
"learning_rate": 1.893334769418565e-05,
"loss": 2.659,
"step": 2980
},
{
"epoch": 0.16,
"learning_rate": 1.8929757481106507e-05,
"loss": 2.6079,
"step": 2990
},
{
"epoch": 0.16,
"learning_rate": 1.892616726802736e-05,
"loss": 2.508,
"step": 3000
},
{
"epoch": 0.16,
"learning_rate": 1.892257705494821e-05,
"loss": 2.496,
"step": 3010
},
{
"epoch": 0.16,
"learning_rate": 1.8918986841869066e-05,
"loss": 2.6098,
"step": 3020
},
{
"epoch": 0.16,
"learning_rate": 1.891539662878992e-05,
"loss": 2.5876,
"step": 3030
},
{
"epoch": 0.16,
"learning_rate": 1.8911806415710774e-05,
"loss": 2.6408,
"step": 3040
},
{
"epoch": 0.16,
"learning_rate": 1.890821620263163e-05,
"loss": 2.5723,
"step": 3050
},
{
"epoch": 0.16,
"learning_rate": 1.8904625989552482e-05,
"loss": 2.5794,
"step": 3060
},
{
"epoch": 0.17,
"learning_rate": 1.8901035776473334e-05,
"loss": 2.6547,
"step": 3070
},
{
"epoch": 0.17,
"learning_rate": 1.889744556339419e-05,
"loss": 2.729,
"step": 3080
},
{
"epoch": 0.17,
"learning_rate": 1.8893855350315042e-05,
"loss": 2.6229,
"step": 3090
},
{
"epoch": 0.17,
"learning_rate": 1.8890265137235897e-05,
"loss": 2.5611,
"step": 3100
},
{
"epoch": 0.17,
"learning_rate": 1.888667492415675e-05,
"loss": 2.6038,
"step": 3110
},
{
"epoch": 0.17,
"learning_rate": 1.8883084711077605e-05,
"loss": 2.5261,
"step": 3120
},
{
"epoch": 0.17,
"learning_rate": 1.8879494497998457e-05,
"loss": 2.5488,
"step": 3130
},
{
"epoch": 0.17,
"learning_rate": 1.887590428491931e-05,
"loss": 2.4828,
"step": 3140
},
{
"epoch": 0.17,
"learning_rate": 1.8872314071840165e-05,
"loss": 2.6341,
"step": 3150
},
{
"epoch": 0.17,
"learning_rate": 1.886872385876102e-05,
"loss": 2.5842,
"step": 3160
},
{
"epoch": 0.17,
"learning_rate": 1.8865133645681873e-05,
"loss": 2.4895,
"step": 3170
},
{
"epoch": 0.17,
"learning_rate": 1.8861543432602728e-05,
"loss": 2.6954,
"step": 3180
},
{
"epoch": 0.17,
"learning_rate": 1.885795321952358e-05,
"loss": 2.3721,
"step": 3190
},
{
"epoch": 0.17,
"learning_rate": 1.8854363006444432e-05,
"loss": 2.5129,
"step": 3200
},
{
"epoch": 0.17,
"learning_rate": 1.8850772793365288e-05,
"loss": 2.5214,
"step": 3210
},
{
"epoch": 0.17,
"learning_rate": 1.884718258028614e-05,
"loss": 2.4711,
"step": 3220
},
{
"epoch": 0.17,
"learning_rate": 1.8843592367206996e-05,
"loss": 2.6295,
"step": 3230
},
{
"epoch": 0.17,
"learning_rate": 1.884000215412785e-05,
"loss": 2.7043,
"step": 3240
},
{
"epoch": 0.18,
"learning_rate": 1.8836411941048703e-05,
"loss": 2.5009,
"step": 3250
},
{
"epoch": 0.18,
"learning_rate": 1.8832821727969555e-05,
"loss": 2.5467,
"step": 3260
},
{
"epoch": 0.18,
"learning_rate": 1.8829231514890408e-05,
"loss": 2.5931,
"step": 3270
},
{
"epoch": 0.18,
"learning_rate": 1.8825641301811263e-05,
"loss": 2.4969,
"step": 3280
},
{
"epoch": 0.18,
"learning_rate": 1.882205108873212e-05,
"loss": 2.4967,
"step": 3290
},
{
"epoch": 0.18,
"learning_rate": 1.881846087565297e-05,
"loss": 2.6285,
"step": 3300
},
{
"epoch": 0.18,
"learning_rate": 1.8814870662573826e-05,
"loss": 2.5414,
"step": 3310
},
{
"epoch": 0.18,
"learning_rate": 1.881128044949468e-05,
"loss": 2.6132,
"step": 3320
},
{
"epoch": 0.18,
"learning_rate": 1.880769023641553e-05,
"loss": 2.5526,
"step": 3330
},
{
"epoch": 0.18,
"learning_rate": 1.8804100023336386e-05,
"loss": 2.424,
"step": 3340
},
{
"epoch": 0.18,
"learning_rate": 1.880050981025724e-05,
"loss": 2.4913,
"step": 3350
},
{
"epoch": 0.18,
"learning_rate": 1.8796919597178094e-05,
"loss": 2.5996,
"step": 3360
},
{
"epoch": 0.18,
"learning_rate": 1.879332938409895e-05,
"loss": 2.5916,
"step": 3370
},
{
"epoch": 0.18,
"learning_rate": 1.87897391710198e-05,
"loss": 2.6065,
"step": 3380
},
{
"epoch": 0.18,
"learning_rate": 1.8786148957940654e-05,
"loss": 2.5286,
"step": 3390
},
{
"epoch": 0.18,
"learning_rate": 1.878255874486151e-05,
"loss": 2.3606,
"step": 3400
},
{
"epoch": 0.18,
"learning_rate": 1.877896853178236e-05,
"loss": 2.5906,
"step": 3410
},
{
"epoch": 0.18,
"learning_rate": 1.8775378318703217e-05,
"loss": 2.5598,
"step": 3420
},
{
"epoch": 0.18,
"learning_rate": 1.8771788105624073e-05,
"loss": 2.567,
"step": 3430
},
{
"epoch": 0.19,
"learning_rate": 1.8768197892544925e-05,
"loss": 2.4522,
"step": 3440
},
{
"epoch": 0.19,
"learning_rate": 1.8764607679465777e-05,
"loss": 2.5184,
"step": 3450
},
{
"epoch": 0.19,
"learning_rate": 1.876101746638663e-05,
"loss": 2.5049,
"step": 3460
},
{
"epoch": 0.19,
"learning_rate": 1.8757427253307485e-05,
"loss": 2.5587,
"step": 3470
},
{
"epoch": 0.19,
"learning_rate": 1.875383704022834e-05,
"loss": 2.5336,
"step": 3480
},
{
"epoch": 0.19,
"learning_rate": 1.8750246827149192e-05,
"loss": 2.5556,
"step": 3490
},
{
"epoch": 0.19,
"learning_rate": 1.8746656614070048e-05,
"loss": 2.5977,
"step": 3500
},
{
"epoch": 0.19,
"learning_rate": 1.87430664009909e-05,
"loss": 2.635,
"step": 3510
},
{
"epoch": 0.19,
"learning_rate": 1.8739476187911752e-05,
"loss": 2.4791,
"step": 3520
},
{
"epoch": 0.19,
"learning_rate": 1.8735885974832608e-05,
"loss": 2.4704,
"step": 3530
},
{
"epoch": 0.19,
"learning_rate": 1.873229576175346e-05,
"loss": 2.5109,
"step": 3540
},
{
"epoch": 0.19,
"learning_rate": 1.8728705548674315e-05,
"loss": 2.5131,
"step": 3550
},
{
"epoch": 0.19,
"learning_rate": 1.872511533559517e-05,
"loss": 2.6331,
"step": 3560
},
{
"epoch": 0.19,
"learning_rate": 1.8721525122516023e-05,
"loss": 2.4951,
"step": 3570
},
{
"epoch": 0.19,
"learning_rate": 1.8717934909436875e-05,
"loss": 2.3377,
"step": 3580
},
{
"epoch": 0.19,
"learning_rate": 1.871434469635773e-05,
"loss": 2.5551,
"step": 3590
},
{
"epoch": 0.19,
"learning_rate": 1.8710754483278583e-05,
"loss": 2.5181,
"step": 3600
},
{
"epoch": 0.19,
"learning_rate": 1.870716427019944e-05,
"loss": 2.3867,
"step": 3610
},
{
"epoch": 0.19,
"learning_rate": 1.870357405712029e-05,
"loss": 2.4953,
"step": 3620
},
{
"epoch": 0.2,
"learning_rate": 1.8699983844041146e-05,
"loss": 2.4266,
"step": 3630
},
{
"epoch": 0.2,
"learning_rate": 1.8696393630962e-05,
"loss": 2.5761,
"step": 3640
},
{
"epoch": 0.2,
"learning_rate": 1.869280341788285e-05,
"loss": 2.5007,
"step": 3650
},
{
"epoch": 0.2,
"learning_rate": 1.8689213204803706e-05,
"loss": 2.641,
"step": 3660
},
{
"epoch": 0.2,
"learning_rate": 1.868562299172456e-05,
"loss": 2.509,
"step": 3670
},
{
"epoch": 0.2,
"learning_rate": 1.8682032778645414e-05,
"loss": 2.4467,
"step": 3680
},
{
"epoch": 0.2,
"learning_rate": 1.867844256556627e-05,
"loss": 2.5679,
"step": 3690
},
{
"epoch": 0.2,
"learning_rate": 1.867485235248712e-05,
"loss": 2.4515,
"step": 3700
},
{
"epoch": 0.2,
"learning_rate": 1.8671262139407974e-05,
"loss": 2.5014,
"step": 3710
},
{
"epoch": 0.2,
"learning_rate": 1.866767192632883e-05,
"loss": 2.5634,
"step": 3720
},
{
"epoch": 0.2,
"learning_rate": 1.866408171324968e-05,
"loss": 2.5036,
"step": 3730
},
{
"epoch": 0.2,
"learning_rate": 1.8660491500170537e-05,
"loss": 2.4928,
"step": 3740
},
{
"epoch": 0.2,
"learning_rate": 1.8656901287091392e-05,
"loss": 2.4937,
"step": 3750
},
{
"epoch": 0.2,
"learning_rate": 1.8653311074012245e-05,
"loss": 2.5571,
"step": 3760
},
{
"epoch": 0.2,
"learning_rate": 1.8649720860933097e-05,
"loss": 2.5574,
"step": 3770
},
{
"epoch": 0.2,
"learning_rate": 1.864613064785395e-05,
"loss": 2.4821,
"step": 3780
},
{
"epoch": 0.2,
"learning_rate": 1.8642540434774804e-05,
"loss": 2.4531,
"step": 3790
},
{
"epoch": 0.2,
"learning_rate": 1.863895022169566e-05,
"loss": 2.5505,
"step": 3800
},
{
"epoch": 0.21,
"learning_rate": 1.8635360008616512e-05,
"loss": 2.4979,
"step": 3810
},
{
"epoch": 0.21,
"learning_rate": 1.8631769795537368e-05,
"loss": 2.5249,
"step": 3820
},
{
"epoch": 0.21,
"learning_rate": 1.862817958245822e-05,
"loss": 2.4761,
"step": 3830
},
{
"epoch": 0.21,
"learning_rate": 1.8624589369379072e-05,
"loss": 2.4575,
"step": 3840
},
{
"epoch": 0.21,
"learning_rate": 1.8620999156299928e-05,
"loss": 2.5435,
"step": 3850
},
{
"epoch": 0.21,
"learning_rate": 1.8617408943220783e-05,
"loss": 2.5849,
"step": 3860
},
{
"epoch": 0.21,
"learning_rate": 1.8613818730141635e-05,
"loss": 2.5028,
"step": 3870
},
{
"epoch": 0.21,
"learning_rate": 1.861022851706249e-05,
"loss": 2.4835,
"step": 3880
},
{
"epoch": 0.21,
"learning_rate": 1.8606638303983343e-05,
"loss": 2.6491,
"step": 3890
},
{
"epoch": 0.21,
"learning_rate": 1.8603048090904195e-05,
"loss": 2.4101,
"step": 3900
},
{
"epoch": 0.21,
"learning_rate": 1.859945787782505e-05,
"loss": 2.5407,
"step": 3910
},
{
"epoch": 0.21,
"learning_rate": 1.859622668605382e-05,
"loss": 6.4295,
"step": 3920
},
{
"epoch": 0.21,
"learning_rate": 1.8592636472974674e-05,
"loss": 11.3457,
"step": 3930
},
{
"epoch": 0.21,
"learning_rate": 1.8589046259895526e-05,
"loss": 15.0764,
"step": 3940
},
{
"epoch": 0.21,
"learning_rate": 1.8585456046816378e-05,
"loss": 14.1912,
"step": 3950
},
{
"epoch": 0.21,
"learning_rate": 1.8581865833737234e-05,
"loss": 12.5133,
"step": 3960
},
{
"epoch": 0.21,
"learning_rate": 1.857827562065809e-05,
"loss": 11.572,
"step": 3970
},
{
"epoch": 0.21,
"learning_rate": 1.857468540757894e-05,
"loss": 11.2024,
"step": 3980
},
{
"epoch": 0.21,
"learning_rate": 1.8571095194499794e-05,
"loss": 10.5553,
"step": 3990
},
{
"epoch": 0.22,
"learning_rate": 1.856750498142065e-05,
"loss": 9.9466,
"step": 4000
},
{
"epoch": 0.22,
"learning_rate": 1.85639147683415e-05,
"loss": 5.7554,
"step": 4010
},
{
"epoch": 0.22,
"learning_rate": 1.8560324555262357e-05,
"loss": 2.8853,
"step": 4020
},
{
"epoch": 0.22,
"learning_rate": 1.855673434218321e-05,
"loss": 2.7084,
"step": 4030
},
{
"epoch": 0.22,
"learning_rate": 1.8553144129104065e-05,
"loss": 3.5929,
"step": 4040
},
{
"epoch": 0.22,
"learning_rate": 1.8549553916024917e-05,
"loss": 5.2618,
"step": 4050
},
{
"epoch": 0.22,
"learning_rate": 1.8545963702945772e-05,
"loss": 4.1107,
"step": 4060
},
{
"epoch": 0.22,
"learning_rate": 1.8542373489866624e-05,
"loss": 3.6029,
"step": 4070
},
{
"epoch": 0.22,
"learning_rate": 1.853878327678748e-05,
"loss": 3.5567,
"step": 4080
},
{
"epoch": 0.22,
"learning_rate": 1.8535193063708332e-05,
"loss": 3.7352,
"step": 4090
},
{
"epoch": 0.22,
"learning_rate": 1.8531602850629188e-05,
"loss": 3.6084,
"step": 4100
},
{
"epoch": 0.22,
"learning_rate": 1.852801263755004e-05,
"loss": 3.0987,
"step": 4110
},
{
"epoch": 0.22,
"learning_rate": 1.8524422424470895e-05,
"loss": 3.2507,
"step": 4120
},
{
"epoch": 0.22,
"learning_rate": 1.8520832211391747e-05,
"loss": 3.3224,
"step": 4130
},
{
"epoch": 0.22,
"learning_rate": 1.85172419983126e-05,
"loss": 3.1285,
"step": 4140
},
{
"epoch": 0.22,
"learning_rate": 1.8513651785233455e-05,
"loss": 2.9518,
"step": 4150
},
{
"epoch": 0.22,
"learning_rate": 1.851006157215431e-05,
"loss": 3.2708,
"step": 4160
},
{
"epoch": 0.22,
"learning_rate": 1.8506471359075163e-05,
"loss": 3.1881,
"step": 4170
},
{
"epoch": 0.23,
"learning_rate": 1.8502881145996015e-05,
"loss": 2.9374,
"step": 4180
},
{
"epoch": 0.23,
"learning_rate": 1.849929093291687e-05,
"loss": 2.8588,
"step": 4190
},
{
"epoch": 0.23,
"learning_rate": 1.8495700719837723e-05,
"loss": 2.8595,
"step": 4200
},
{
"epoch": 0.23,
"learning_rate": 1.8492110506758578e-05,
"loss": 2.8636,
"step": 4210
},
{
"epoch": 0.23,
"learning_rate": 1.848852029367943e-05,
"loss": 2.839,
"step": 4220
},
{
"epoch": 0.23,
"learning_rate": 1.8484930080600286e-05,
"loss": 2.8617,
"step": 4230
},
{
"epoch": 0.23,
"learning_rate": 1.8481339867521138e-05,
"loss": 2.8889,
"step": 4240
},
{
"epoch": 0.23,
"learning_rate": 1.8477749654441994e-05,
"loss": 2.8111,
"step": 4250
},
{
"epoch": 0.23,
"learning_rate": 1.8474159441362846e-05,
"loss": 2.805,
"step": 4260
},
{
"epoch": 0.23,
"learning_rate": 1.84705692282837e-05,
"loss": 2.7259,
"step": 4270
},
{
"epoch": 0.23,
"learning_rate": 1.8466979015204554e-05,
"loss": 2.8082,
"step": 4280
},
{
"epoch": 0.23,
"learning_rate": 1.846338880212541e-05,
"loss": 2.822,
"step": 4290
},
{
"epoch": 0.23,
"learning_rate": 1.845979858904626e-05,
"loss": 2.7717,
"step": 4300
},
{
"epoch": 0.23,
"learning_rate": 1.8456208375967117e-05,
"loss": 2.8857,
"step": 4310
},
{
"epoch": 0.23,
"learning_rate": 1.845261816288797e-05,
"loss": 2.8493,
"step": 4320
},
{
"epoch": 0.23,
"learning_rate": 1.844902794980882e-05,
"loss": 2.9915,
"step": 4330
},
{
"epoch": 0.23,
"learning_rate": 1.8445437736729677e-05,
"loss": 3.125,
"step": 4340
},
{
"epoch": 0.23,
"learning_rate": 1.8441847523650532e-05,
"loss": 2.8721,
"step": 4350
},
{
"epoch": 0.23,
"learning_rate": 1.8438257310571384e-05,
"loss": 2.8248,
"step": 4360
},
{
"epoch": 0.24,
"learning_rate": 1.8434667097492237e-05,
"loss": 2.6402,
"step": 4370
},
{
"epoch": 0.24,
"learning_rate": 1.8431076884413092e-05,
"loss": 2.7797,
"step": 4380
},
{
"epoch": 0.24,
"learning_rate": 1.8427486671333944e-05,
"loss": 2.913,
"step": 4390
},
{
"epoch": 0.24,
"learning_rate": 1.84238964582548e-05,
"loss": 2.7321,
"step": 4400
},
{
"epoch": 0.24,
"learning_rate": 1.8420306245175652e-05,
"loss": 2.7276,
"step": 4410
},
{
"epoch": 0.24,
"learning_rate": 1.8416716032096507e-05,
"loss": 2.8455,
"step": 4420
},
{
"epoch": 0.24,
"learning_rate": 1.841312581901736e-05,
"loss": 2.5952,
"step": 4430
},
{
"epoch": 0.24,
"learning_rate": 1.8409535605938215e-05,
"loss": 2.6574,
"step": 4440
},
{
"epoch": 0.24,
"learning_rate": 1.8405945392859067e-05,
"loss": 2.7382,
"step": 4450
},
{
"epoch": 0.24,
"learning_rate": 1.840235517977992e-05,
"loss": 2.7432,
"step": 4460
},
{
"epoch": 0.24,
"learning_rate": 1.8398764966700775e-05,
"loss": 2.8787,
"step": 4470
},
{
"epoch": 0.24,
"learning_rate": 1.839517475362163e-05,
"loss": 2.7913,
"step": 4480
},
{
"epoch": 0.24,
"learning_rate": 1.8391584540542483e-05,
"loss": 2.6114,
"step": 4490
},
{
"epoch": 0.24,
"learning_rate": 1.8387994327463335e-05,
"loss": 2.7506,
"step": 4500
},
{
"epoch": 0.24,
"learning_rate": 1.838440411438419e-05,
"loss": 2.6786,
"step": 4510
},
{
"epoch": 0.24,
"learning_rate": 1.8380813901305043e-05,
"loss": 2.7556,
"step": 4520
},
{
"epoch": 0.24,
"learning_rate": 1.8377223688225898e-05,
"loss": 2.5142,
"step": 4530
},
{
"epoch": 0.24,
"learning_rate": 1.837363347514675e-05,
"loss": 2.5796,
"step": 4540
},
{
"epoch": 0.25,
"learning_rate": 1.8370043262067606e-05,
"loss": 2.5861,
"step": 4550
},
{
"epoch": 0.25,
"learning_rate": 1.8366453048988458e-05,
"loss": 2.4686,
"step": 4560
},
{
"epoch": 0.25,
"learning_rate": 1.8362862835909314e-05,
"loss": 2.5343,
"step": 4570
},
{
"epoch": 0.25,
"learning_rate": 1.8359272622830166e-05,
"loss": 2.4751,
"step": 4580
},
{
"epoch": 0.25,
"learning_rate": 1.835568240975102e-05,
"loss": 2.5097,
"step": 4590
},
{
"epoch": 0.25,
"learning_rate": 1.8352092196671873e-05,
"loss": 2.4695,
"step": 4600
},
{
"epoch": 0.25,
"learning_rate": 1.834850198359273e-05,
"loss": 2.3584,
"step": 4610
},
{
"epoch": 0.25,
"learning_rate": 1.834491177051358e-05,
"loss": 2.5559,
"step": 4620
},
{
"epoch": 0.25,
"learning_rate": 1.8341321557434437e-05,
"loss": 2.501,
"step": 4630
},
{
"epoch": 0.25,
"learning_rate": 1.833773134435529e-05,
"loss": 2.3888,
"step": 4640
},
{
"epoch": 0.25,
"learning_rate": 1.833414113127614e-05,
"loss": 2.4763,
"step": 4650
},
{
"epoch": 0.25,
"learning_rate": 1.8330550918196996e-05,
"loss": 2.4321,
"step": 4660
},
{
"epoch": 0.25,
"learning_rate": 1.8326960705117852e-05,
"loss": 2.5815,
"step": 4670
},
{
"epoch": 0.25,
"learning_rate": 1.8323370492038704e-05,
"loss": 2.5292,
"step": 4680
},
{
"epoch": 0.25,
"learning_rate": 1.8319780278959556e-05,
"loss": 2.4254,
"step": 4690
},
{
"epoch": 0.25,
"learning_rate": 1.8316190065880412e-05,
"loss": 2.4508,
"step": 4700
},
{
"epoch": 0.25,
"learning_rate": 1.8312599852801264e-05,
"loss": 2.5653,
"step": 4710
},
{
"epoch": 0.25,
"learning_rate": 1.830900963972212e-05,
"loss": 2.8117,
"step": 4720
},
{
"epoch": 0.25,
"learning_rate": 1.8305419426642972e-05,
"loss": 3.3514,
"step": 4730
},
{
"epoch": 0.26,
"learning_rate": 1.8301829213563827e-05,
"loss": 3.1663,
"step": 4740
},
{
"epoch": 0.26,
"learning_rate": 1.829823900048468e-05,
"loss": 2.6344,
"step": 4750
},
{
"epoch": 0.26,
"learning_rate": 1.8294648787405535e-05,
"loss": 2.5842,
"step": 4760
},
{
"epoch": 0.26,
"learning_rate": 1.8291058574326387e-05,
"loss": 2.5508,
"step": 4770
},
{
"epoch": 0.26,
"learning_rate": 1.8287468361247243e-05,
"loss": 2.5438,
"step": 4780
},
{
"epoch": 0.26,
"learning_rate": 1.8283878148168095e-05,
"loss": 2.5024,
"step": 4790
},
{
"epoch": 0.26,
"learning_rate": 1.828028793508895e-05,
"loss": 2.524,
"step": 4800
},
{
"epoch": 0.26,
"learning_rate": 1.8276697722009803e-05,
"loss": 2.5516,
"step": 4810
},
{
"epoch": 0.26,
"learning_rate": 1.8273107508930658e-05,
"loss": 2.5282,
"step": 4820
},
{
"epoch": 0.26,
"learning_rate": 1.826951729585151e-05,
"loss": 2.5368,
"step": 4830
},
{
"epoch": 0.26,
"learning_rate": 1.8265927082772362e-05,
"loss": 2.6432,
"step": 4840
},
{
"epoch": 0.26,
"learning_rate": 1.8262336869693218e-05,
"loss": 2.4558,
"step": 4850
},
{
"epoch": 0.26,
"learning_rate": 1.8258746656614074e-05,
"loss": 2.5538,
"step": 4860
},
{
"epoch": 0.26,
"learning_rate": 1.8255156443534926e-05,
"loss": 2.5044,
"step": 4870
},
{
"epoch": 0.26,
"learning_rate": 1.8251566230455778e-05,
"loss": 2.4873,
"step": 4880
},
{
"epoch": 0.26,
"learning_rate": 1.8247976017376633e-05,
"loss": 2.5056,
"step": 4890
},
{
"epoch": 0.26,
"learning_rate": 1.8244385804297486e-05,
"loss": 2.506,
"step": 4900
},
{
"epoch": 0.26,
"learning_rate": 1.824079559121834e-05,
"loss": 2.4669,
"step": 4910
},
{
"epoch": 0.26,
"learning_rate": 1.8237205378139193e-05,
"loss": 2.416,
"step": 4920
},
{
"epoch": 0.27,
"learning_rate": 1.823361516506005e-05,
"loss": 2.4406,
"step": 4930
},
{
"epoch": 0.27,
"learning_rate": 1.82300249519809e-05,
"loss": 2.5012,
"step": 4940
},
{
"epoch": 0.27,
"learning_rate": 1.8226434738901756e-05,
"loss": 2.5044,
"step": 4950
},
{
"epoch": 0.27,
"learning_rate": 1.822284452582261e-05,
"loss": 2.6877,
"step": 4960
},
{
"epoch": 0.27,
"learning_rate": 1.821925431274346e-05,
"loss": 2.4853,
"step": 4970
},
{
"epoch": 0.27,
"learning_rate": 1.8215664099664316e-05,
"loss": 2.4769,
"step": 4980
},
{
"epoch": 0.27,
"learning_rate": 1.8212073886585172e-05,
"loss": 2.6282,
"step": 4990
},
{
"epoch": 0.27,
"learning_rate": 1.8208483673506024e-05,
"loss": 2.612,
"step": 5000
},
{
"epoch": 0.27,
"learning_rate": 1.8204893460426876e-05,
"loss": 2.7836,
"step": 5010
},
{
"epoch": 0.27,
"learning_rate": 1.8201303247347732e-05,
"loss": 2.8112,
"step": 5020
},
{
"epoch": 0.27,
"learning_rate": 1.8197713034268584e-05,
"loss": 2.3967,
"step": 5030
},
{
"epoch": 0.27,
"learning_rate": 1.819412282118944e-05,
"loss": 2.5992,
"step": 5040
},
{
"epoch": 0.27,
"learning_rate": 1.819053260811029e-05,
"loss": 2.5028,
"step": 5050
},
{
"epoch": 0.27,
"learning_rate": 1.8186942395031147e-05,
"loss": 2.3993,
"step": 5060
},
{
"epoch": 0.27,
"learning_rate": 1.8183352181952e-05,
"loss": 2.4996,
"step": 5070
},
{
"epoch": 0.27,
"learning_rate": 1.8179761968872855e-05,
"loss": 2.5207,
"step": 5080
},
{
"epoch": 0.27,
"learning_rate": 1.8176171755793707e-05,
"loss": 2.5934,
"step": 5090
},
{
"epoch": 0.27,
"learning_rate": 1.8172581542714563e-05,
"loss": 2.4164,
"step": 5100
},
{
"epoch": 0.28,
"learning_rate": 1.8168991329635415e-05,
"loss": 2.4869,
"step": 5110
},
{
"epoch": 0.28,
"learning_rate": 1.816540111655627e-05,
"loss": 2.562,
"step": 5120
},
{
"epoch": 0.28,
"learning_rate": 1.8161810903477122e-05,
"loss": 2.4455,
"step": 5130
},
{
"epoch": 0.28,
"learning_rate": 1.8158220690397978e-05,
"loss": 2.5116,
"step": 5140
},
{
"epoch": 0.28,
"learning_rate": 1.815463047731883e-05,
"loss": 2.5017,
"step": 5150
},
{
"epoch": 0.28,
"learning_rate": 1.8151040264239682e-05,
"loss": 2.411,
"step": 5160
},
{
"epoch": 0.28,
"learning_rate": 1.8147450051160538e-05,
"loss": 2.4743,
"step": 5170
},
{
"epoch": 0.28,
"learning_rate": 1.8143859838081393e-05,
"loss": 2.2396,
"step": 5180
},
{
"epoch": 0.28,
"learning_rate": 1.8140269625002246e-05,
"loss": 2.532,
"step": 5190
},
{
"epoch": 0.28,
"learning_rate": 1.8136679411923098e-05,
"loss": 2.4305,
"step": 5200
},
{
"epoch": 0.28,
"learning_rate": 1.8133089198843953e-05,
"loss": 2.4409,
"step": 5210
},
{
"epoch": 0.28,
"learning_rate": 1.8129498985764805e-05,
"loss": 2.5456,
"step": 5220
},
{
"epoch": 0.28,
"learning_rate": 1.812590877268566e-05,
"loss": 2.4283,
"step": 5230
},
{
"epoch": 0.28,
"learning_rate": 1.8122318559606513e-05,
"loss": 2.3817,
"step": 5240
},
{
"epoch": 0.28,
"learning_rate": 1.811872834652737e-05,
"loss": 2.4492,
"step": 5250
},
{
"epoch": 0.28,
"learning_rate": 1.811513813344822e-05,
"loss": 2.4183,
"step": 5260
},
{
"epoch": 0.28,
"learning_rate": 1.8111547920369076e-05,
"loss": 2.3994,
"step": 5270
},
{
"epoch": 0.28,
"learning_rate": 1.810795770728993e-05,
"loss": 2.5341,
"step": 5280
},
{
"epoch": 0.28,
"learning_rate": 1.8104367494210784e-05,
"loss": 2.5021,
"step": 5290
},
{
"epoch": 0.29,
"learning_rate": 1.8100777281131636e-05,
"loss": 2.3663,
"step": 5300
},
{
"epoch": 0.29,
"learning_rate": 1.8097187068052492e-05,
"loss": 2.3879,
"step": 5310
},
{
"epoch": 0.29,
"learning_rate": 1.8093596854973344e-05,
"loss": 2.4557,
"step": 5320
},
{
"epoch": 0.29,
"learning_rate": 1.80900066418942e-05,
"loss": 2.5436,
"step": 5330
},
{
"epoch": 0.29,
"learning_rate": 1.808641642881505e-05,
"loss": 2.4606,
"step": 5340
},
{
"epoch": 0.29,
"learning_rate": 1.8082826215735904e-05,
"loss": 2.3418,
"step": 5350
},
{
"epoch": 0.29,
"learning_rate": 1.807923600265676e-05,
"loss": 2.4659,
"step": 5360
},
{
"epoch": 0.29,
"learning_rate": 1.8075645789577615e-05,
"loss": 2.3854,
"step": 5370
},
{
"epoch": 0.29,
"learning_rate": 1.8072055576498467e-05,
"loss": 2.3809,
"step": 5380
},
{
"epoch": 0.29,
"learning_rate": 1.806846536341932e-05,
"loss": 2.4259,
"step": 5390
},
{
"epoch": 0.29,
"learning_rate": 1.8064875150340175e-05,
"loss": 2.3579,
"step": 5400
},
{
"epoch": 0.29,
"learning_rate": 1.8061284937261027e-05,
"loss": 2.4097,
"step": 5410
},
{
"epoch": 0.29,
"learning_rate": 1.8057694724181882e-05,
"loss": 2.5435,
"step": 5420
},
{
"epoch": 0.29,
"learning_rate": 1.8054104511102735e-05,
"loss": 2.4496,
"step": 5430
},
{
"epoch": 0.29,
"learning_rate": 1.805051429802359e-05,
"loss": 2.4193,
"step": 5440
},
{
"epoch": 0.29,
"learning_rate": 1.8046924084944442e-05,
"loss": 2.4353,
"step": 5450
},
{
"epoch": 0.29,
"learning_rate": 1.8043333871865298e-05,
"loss": 2.3613,
"step": 5460
},
{
"epoch": 0.29,
"learning_rate": 1.803974365878615e-05,
"loss": 2.4537,
"step": 5470
},
{
"epoch": 0.3,
"learning_rate": 1.8036153445707002e-05,
"loss": 2.3768,
"step": 5480
},
{
"epoch": 0.3,
"learning_rate": 1.8032563232627858e-05,
"loss": 2.3984,
"step": 5490
},
{
"epoch": 0.3,
"learning_rate": 1.8028973019548713e-05,
"loss": 2.5764,
"step": 5500
},
{
"epoch": 0.3,
"learning_rate": 1.8025382806469565e-05,
"loss": 2.5871,
"step": 5510
},
{
"epoch": 0.3,
"learning_rate": 1.802179259339042e-05,
"loss": 2.4675,
"step": 5520
},
{
"epoch": 0.3,
"learning_rate": 1.8018202380311273e-05,
"loss": 2.4377,
"step": 5530
},
{
"epoch": 0.3,
"learning_rate": 1.8014612167232125e-05,
"loss": 2.3109,
"step": 5540
},
{
"epoch": 0.3,
"learning_rate": 1.801102195415298e-05,
"loss": 2.2933,
"step": 5550
},
{
"epoch": 0.3,
"learning_rate": 1.8007431741073836e-05,
"loss": 2.3383,
"step": 5560
},
{
"epoch": 0.3,
"learning_rate": 1.800384152799469e-05,
"loss": 2.4377,
"step": 5570
},
{
"epoch": 0.3,
"learning_rate": 1.800025131491554e-05,
"loss": 2.4252,
"step": 5580
},
{
"epoch": 0.3,
"learning_rate": 1.7996661101836396e-05,
"loss": 2.5476,
"step": 5590
},
{
"epoch": 0.3,
"learning_rate": 1.7993070888757248e-05,
"loss": 2.251,
"step": 5600
},
{
"epoch": 0.3,
"learning_rate": 1.7989480675678104e-05,
"loss": 2.2933,
"step": 5610
},
{
"epoch": 0.3,
"learning_rate": 1.7985890462598956e-05,
"loss": 2.5557,
"step": 5620
},
{
"epoch": 0.3,
"learning_rate": 1.798230024951981e-05,
"loss": 2.5928,
"step": 5630
},
{
"epoch": 0.3,
"learning_rate": 1.7978710036440664e-05,
"loss": 2.4235,
"step": 5640
},
{
"epoch": 0.3,
"learning_rate": 1.797511982336152e-05,
"loss": 2.3717,
"step": 5650
},
{
"epoch": 0.3,
"learning_rate": 1.797152961028237e-05,
"loss": 2.4914,
"step": 5660
},
{
"epoch": 0.31,
"learning_rate": 1.7967939397203224e-05,
"loss": 2.779,
"step": 5670
},
{
"epoch": 0.31,
"learning_rate": 1.796434918412408e-05,
"loss": 2.8167,
"step": 5680
},
{
"epoch": 0.31,
"learning_rate": 1.7960758971044935e-05,
"loss": 2.5284,
"step": 5690
},
{
"epoch": 0.31,
"learning_rate": 1.7957168757965787e-05,
"loss": 2.3563,
"step": 5700
},
{
"epoch": 0.31,
"learning_rate": 1.795357854488664e-05,
"loss": 2.4062,
"step": 5710
},
{
"epoch": 0.31,
"learning_rate": 1.7949988331807495e-05,
"loss": 2.2945,
"step": 5720
},
{
"epoch": 0.31,
"learning_rate": 1.7946398118728347e-05,
"loss": 2.4604,
"step": 5730
},
{
"epoch": 0.31,
"learning_rate": 1.7942807905649202e-05,
"loss": 2.4568,
"step": 5740
},
{
"epoch": 0.31,
"learning_rate": 1.7939217692570054e-05,
"loss": 2.4982,
"step": 5750
},
{
"epoch": 0.31,
"learning_rate": 1.793562747949091e-05,
"loss": 2.525,
"step": 5760
},
{
"epoch": 0.31,
"learning_rate": 1.7932037266411762e-05,
"loss": 2.2969,
"step": 5770
},
{
"epoch": 0.31,
"learning_rate": 1.7928447053332618e-05,
"loss": 2.3002,
"step": 5780
},
{
"epoch": 0.31,
"learning_rate": 1.792485684025347e-05,
"loss": 2.3354,
"step": 5790
},
{
"epoch": 0.31,
"learning_rate": 1.7921266627174325e-05,
"loss": 2.46,
"step": 5800
},
{
"epoch": 0.31,
"learning_rate": 1.7917676414095177e-05,
"loss": 2.2436,
"step": 5810
},
{
"epoch": 0.31,
"learning_rate": 1.7914086201016033e-05,
"loss": 2.3628,
"step": 5820
},
{
"epoch": 0.31,
"learning_rate": 1.7910495987936885e-05,
"loss": 2.4019,
"step": 5830
},
{
"epoch": 0.31,
"learning_rate": 1.790690577485774e-05,
"loss": 2.3565,
"step": 5840
},
{
"epoch": 0.32,
"learning_rate": 1.7903315561778593e-05,
"loss": 2.3806,
"step": 5850
},
{
"epoch": 0.32,
"learning_rate": 1.7899725348699445e-05,
"loss": 2.4735,
"step": 5860
},
{
"epoch": 0.32,
"learning_rate": 1.78961351356203e-05,
"loss": 2.2858,
"step": 5870
},
{
"epoch": 0.32,
"learning_rate": 1.7892544922541156e-05,
"loss": 2.4405,
"step": 5880
},
{
"epoch": 0.32,
"learning_rate": 1.7888954709462008e-05,
"loss": 2.51,
"step": 5890
},
{
"epoch": 0.32,
"learning_rate": 1.788536449638286e-05,
"loss": 2.3536,
"step": 5900
},
{
"epoch": 0.32,
"learning_rate": 1.7881774283303716e-05,
"loss": 2.3228,
"step": 5910
},
{
"epoch": 0.32,
"learning_rate": 1.7878184070224568e-05,
"loss": 2.3953,
"step": 5920
},
{
"epoch": 0.32,
"learning_rate": 1.7874593857145424e-05,
"loss": 2.3964,
"step": 5930
},
{
"epoch": 0.32,
"learning_rate": 1.7871003644066276e-05,
"loss": 2.2246,
"step": 5940
},
{
"epoch": 0.32,
"learning_rate": 1.786741343098713e-05,
"loss": 2.4914,
"step": 5950
},
{
"epoch": 0.32,
"learning_rate": 1.7863823217907984e-05,
"loss": 2.4168,
"step": 5960
},
{
"epoch": 0.32,
"learning_rate": 1.786023300482884e-05,
"loss": 2.386,
"step": 5970
},
{
"epoch": 0.32,
"learning_rate": 1.785664279174969e-05,
"loss": 2.4283,
"step": 5980
},
{
"epoch": 0.32,
"learning_rate": 1.7853052578670547e-05,
"loss": 2.4177,
"step": 5990
},
{
"epoch": 0.32,
"learning_rate": 1.78494623655914e-05,
"loss": 2.5387,
"step": 6000
},
{
"epoch": 0.32,
"learning_rate": 1.7845872152512254e-05,
"loss": 2.4892,
"step": 6010
},
{
"epoch": 0.32,
"learning_rate": 1.7842281939433107e-05,
"loss": 2.3955,
"step": 6020
},
{
"epoch": 0.32,
"learning_rate": 1.7838691726353962e-05,
"loss": 2.2883,
"step": 6030
},
{
"epoch": 0.33,
"learning_rate": 1.7835101513274814e-05,
"loss": 2.3865,
"step": 6040
},
{
"epoch": 0.33,
"learning_rate": 1.7831511300195667e-05,
"loss": 2.4296,
"step": 6050
},
{
"epoch": 0.33,
"learning_rate": 1.7827921087116522e-05,
"loss": 2.1582,
"step": 6060
},
{
"epoch": 0.33,
"learning_rate": 1.7824330874037378e-05,
"loss": 2.3292,
"step": 6070
},
{
"epoch": 0.33,
"learning_rate": 1.782074066095823e-05,
"loss": 2.2857,
"step": 6080
},
{
"epoch": 0.33,
"learning_rate": 1.7817150447879082e-05,
"loss": 2.3803,
"step": 6090
},
{
"epoch": 0.33,
"learning_rate": 1.7813560234799937e-05,
"loss": 2.2857,
"step": 6100
},
{
"epoch": 0.33,
"learning_rate": 1.780997002172079e-05,
"loss": 2.4408,
"step": 6110
},
{
"epoch": 0.33,
"learning_rate": 1.7806379808641645e-05,
"loss": 2.3569,
"step": 6120
},
{
"epoch": 0.33,
"learning_rate": 1.7802789595562497e-05,
"loss": 2.3872,
"step": 6130
},
{
"epoch": 0.33,
"learning_rate": 1.7799199382483353e-05,
"loss": 2.476,
"step": 6140
},
{
"epoch": 0.33,
"learning_rate": 1.7795609169404205e-05,
"loss": 2.3956,
"step": 6150
},
{
"epoch": 0.33,
"learning_rate": 1.779201895632506e-05,
"loss": 2.3636,
"step": 6160
},
{
"epoch": 0.33,
"learning_rate": 1.7788428743245913e-05,
"loss": 2.4033,
"step": 6170
},
{
"epoch": 0.33,
"learning_rate": 1.7784838530166765e-05,
"loss": 2.4184,
"step": 6180
},
{
"epoch": 0.33,
"learning_rate": 1.778124831708762e-05,
"loss": 2.3507,
"step": 6190
},
{
"epoch": 0.33,
"learning_rate": 1.7777658104008476e-05,
"loss": 2.2903,
"step": 6200
},
{
"epoch": 0.33,
"learning_rate": 1.7774067890929328e-05,
"loss": 2.3092,
"step": 6210
},
{
"epoch": 0.33,
"learning_rate": 1.777047767785018e-05,
"loss": 2.4227,
"step": 6220
},
{
"epoch": 0.34,
"learning_rate": 1.7766887464771036e-05,
"loss": 2.2057,
"step": 6230
},
{
"epoch": 0.34,
"learning_rate": 1.7763297251691888e-05,
"loss": 2.3168,
"step": 6240
},
{
"epoch": 0.34,
"learning_rate": 1.7759707038612744e-05,
"loss": 2.2859,
"step": 6250
},
{
"epoch": 0.34,
"learning_rate": 1.7756116825533596e-05,
"loss": 2.3328,
"step": 6260
},
{
"epoch": 0.34,
"learning_rate": 1.775252661245445e-05,
"loss": 2.3735,
"step": 6270
},
{
"epoch": 0.34,
"learning_rate": 1.7748936399375303e-05,
"loss": 2.3695,
"step": 6280
},
{
"epoch": 0.34,
"learning_rate": 1.774534618629616e-05,
"loss": 2.2232,
"step": 6290
},
{
"epoch": 0.34,
"learning_rate": 1.774175597321701e-05,
"loss": 2.2689,
"step": 6300
},
{
"epoch": 0.34,
"learning_rate": 1.7738165760137867e-05,
"loss": 2.3546,
"step": 6310
},
{
"epoch": 0.34,
"learning_rate": 1.773457554705872e-05,
"loss": 2.2545,
"step": 6320
},
{
"epoch": 0.34,
"learning_rate": 1.7730985333979574e-05,
"loss": 2.4258,
"step": 6330
},
{
"epoch": 0.34,
"learning_rate": 1.7727395120900426e-05,
"loss": 2.3306,
"step": 6340
},
{
"epoch": 0.34,
"learning_rate": 1.7723804907821282e-05,
"loss": 2.4172,
"step": 6350
},
{
"epoch": 0.34,
"learning_rate": 1.7720214694742134e-05,
"loss": 2.3494,
"step": 6360
},
{
"epoch": 0.34,
"learning_rate": 1.7716624481662986e-05,
"loss": 2.511,
"step": 6370
},
{
"epoch": 0.34,
"learning_rate": 1.7713034268583842e-05,
"loss": 2.2993,
"step": 6380
},
{
"epoch": 0.34,
"learning_rate": 1.7709444055504697e-05,
"loss": 2.4266,
"step": 6390
},
{
"epoch": 0.34,
"learning_rate": 1.770585384242555e-05,
"loss": 2.3397,
"step": 6400
},
{
"epoch": 0.35,
"learning_rate": 1.7702263629346402e-05,
"loss": 2.2552,
"step": 6410
},
{
"epoch": 0.35,
"learning_rate": 1.7698673416267257e-05,
"loss": 2.287,
"step": 6420
},
{
"epoch": 0.35,
"learning_rate": 1.769508320318811e-05,
"loss": 2.2259,
"step": 6430
},
{
"epoch": 0.35,
"learning_rate": 1.7691492990108965e-05,
"loss": 2.3773,
"step": 6440
},
{
"epoch": 0.35,
"learning_rate": 1.7687902777029817e-05,
"loss": 2.273,
"step": 6450
},
{
"epoch": 0.35,
"learning_rate": 1.7684312563950673e-05,
"loss": 2.2841,
"step": 6460
},
{
"epoch": 0.35,
"learning_rate": 1.7680722350871525e-05,
"loss": 2.3063,
"step": 6470
},
{
"epoch": 0.35,
"learning_rate": 1.767713213779238e-05,
"loss": 2.2447,
"step": 6480
},
{
"epoch": 0.35,
"learning_rate": 1.7673541924713233e-05,
"loss": 2.3697,
"step": 6490
},
{
"epoch": 0.35,
"learning_rate": 1.7669951711634088e-05,
"loss": 2.19,
"step": 6500
},
{
"epoch": 0.35,
"learning_rate": 1.766636149855494e-05,
"loss": 2.2496,
"step": 6510
},
{
"epoch": 0.35,
"learning_rate": 1.7662771285475796e-05,
"loss": 2.3629,
"step": 6520
},
{
"epoch": 0.35,
"learning_rate": 1.7659181072396648e-05,
"loss": 2.3822,
"step": 6530
},
{
"epoch": 0.35,
"learning_rate": 1.7655590859317503e-05,
"loss": 2.3546,
"step": 6540
},
{
"epoch": 0.35,
"learning_rate": 1.7652000646238356e-05,
"loss": 2.3238,
"step": 6550
},
{
"epoch": 0.35,
"learning_rate": 1.7648410433159208e-05,
"loss": 2.4195,
"step": 6560
},
{
"epoch": 0.35,
"learning_rate": 1.7644820220080063e-05,
"loss": 2.3532,
"step": 6570
},
{
"epoch": 0.35,
"learning_rate": 1.764123000700092e-05,
"loss": 2.407,
"step": 6580
},
{
"epoch": 0.35,
"learning_rate": 1.763763979392177e-05,
"loss": 2.2077,
"step": 6590
},
{
"epoch": 0.36,
"learning_rate": 1.7634049580842623e-05,
"loss": 2.3895,
"step": 6600
},
{
"epoch": 0.36,
"learning_rate": 1.763045936776348e-05,
"loss": 2.2603,
"step": 6610
},
{
"epoch": 0.36,
"learning_rate": 1.762686915468433e-05,
"loss": 2.2514,
"step": 6620
},
{
"epoch": 0.36,
"learning_rate": 1.7623278941605186e-05,
"loss": 2.4831,
"step": 6630
},
{
"epoch": 0.36,
"learning_rate": 1.761968872852604e-05,
"loss": 2.3825,
"step": 6640
},
{
"epoch": 0.36,
"learning_rate": 1.7616098515446894e-05,
"loss": 2.2755,
"step": 6650
},
{
"epoch": 0.36,
"learning_rate": 1.7612508302367746e-05,
"loss": 2.2863,
"step": 6660
},
{
"epoch": 0.36,
"learning_rate": 1.7608918089288602e-05,
"loss": 2.3306,
"step": 6670
},
{
"epoch": 0.36,
"learning_rate": 1.7605327876209454e-05,
"loss": 2.5017,
"step": 6680
},
{
"epoch": 0.36,
"learning_rate": 1.7601737663130306e-05,
"loss": 2.4517,
"step": 6690
},
{
"epoch": 0.36,
"learning_rate": 1.7598147450051162e-05,
"loss": 2.3501,
"step": 6700
},
{
"epoch": 0.36,
"learning_rate": 1.7594557236972017e-05,
"loss": 2.753,
"step": 6710
},
{
"epoch": 0.36,
"learning_rate": 1.759096702389287e-05,
"loss": 5.5076,
"step": 6720
},
{
"epoch": 0.36,
"learning_rate": 1.7587376810813725e-05,
"loss": 3.9846,
"step": 6730
},
{
"epoch": 0.36,
"learning_rate": 1.7583786597734577e-05,
"loss": 2.4476,
"step": 6740
},
{
"epoch": 0.36,
"learning_rate": 1.758019638465543e-05,
"loss": 2.3937,
"step": 6750
},
{
"epoch": 0.36,
"learning_rate": 1.7576606171576285e-05,
"loss": 2.3642,
"step": 6760
},
{
"epoch": 0.36,
"learning_rate": 1.7573015958497137e-05,
"loss": 2.5941,
"step": 6770
},
{
"epoch": 0.37,
"learning_rate": 1.7569425745417993e-05,
"loss": 2.4031,
"step": 6780
},
{
"epoch": 0.37,
"learning_rate": 1.7565835532338845e-05,
"loss": 2.3498,
"step": 6790
},
{
"epoch": 0.37,
"learning_rate": 1.75622453192597e-05,
"loss": 2.4207,
"step": 6800
},
{
"epoch": 0.37,
"learning_rate": 1.7558655106180552e-05,
"loss": 2.3942,
"step": 6810
},
{
"epoch": 0.37,
"learning_rate": 1.7555064893101408e-05,
"loss": 2.5219,
"step": 6820
},
{
"epoch": 0.37,
"learning_rate": 1.755147468002226e-05,
"loss": 2.5068,
"step": 6830
},
{
"epoch": 0.37,
"learning_rate": 1.7547884466943116e-05,
"loss": 2.6013,
"step": 6840
},
{
"epoch": 0.37,
"learning_rate": 1.7544294253863968e-05,
"loss": 2.4904,
"step": 6850
},
{
"epoch": 0.37,
"learning_rate": 1.7540704040784823e-05,
"loss": 2.575,
"step": 6860
},
{
"epoch": 0.37,
"learning_rate": 1.7537113827705675e-05,
"loss": 2.3668,
"step": 6870
},
{
"epoch": 0.37,
"learning_rate": 1.7533523614626528e-05,
"loss": 2.5214,
"step": 6880
},
{
"epoch": 0.37,
"learning_rate": 1.7529933401547383e-05,
"loss": 2.3257,
"step": 6890
},
{
"epoch": 0.37,
"learning_rate": 1.752634318846824e-05,
"loss": 2.2955,
"step": 6900
},
{
"epoch": 0.37,
"learning_rate": 1.752275297538909e-05,
"loss": 2.1785,
"step": 6910
},
{
"epoch": 0.37,
"learning_rate": 1.7519162762309943e-05,
"loss": 2.3821,
"step": 6920
},
{
"epoch": 0.37,
"learning_rate": 1.75155725492308e-05,
"loss": 2.4626,
"step": 6930
},
{
"epoch": 0.37,
"learning_rate": 1.751198233615165e-05,
"loss": 2.2706,
"step": 6940
},
{
"epoch": 0.37,
"learning_rate": 1.7508392123072506e-05,
"loss": 2.2919,
"step": 6950
},
{
"epoch": 0.37,
"learning_rate": 1.750480190999336e-05,
"loss": 2.2875,
"step": 6960
},
{
"epoch": 0.38,
"learning_rate": 1.7501211696914214e-05,
"loss": 2.4828,
"step": 6970
},
{
"epoch": 0.38,
"learning_rate": 1.7497621483835066e-05,
"loss": 2.3753,
"step": 6980
},
{
"epoch": 0.38,
"learning_rate": 1.7494031270755922e-05,
"loss": 2.2839,
"step": 6990
},
{
"epoch": 0.38,
"learning_rate": 1.7490441057676774e-05,
"loss": 2.2531,
"step": 7000
},
{
"epoch": 0.38,
"learning_rate": 1.748685084459763e-05,
"loss": 2.3371,
"step": 7010
},
{
"epoch": 0.38,
"learning_rate": 1.748326063151848e-05,
"loss": 2.3223,
"step": 7020
},
{
"epoch": 0.38,
"learning_rate": 1.7479670418439337e-05,
"loss": 2.3399,
"step": 7030
},
{
"epoch": 0.38,
"learning_rate": 1.747608020536019e-05,
"loss": 2.3653,
"step": 7040
},
{
"epoch": 0.38,
"learning_rate": 1.7472489992281045e-05,
"loss": 2.3339,
"step": 7050
},
{
"epoch": 0.38,
"learning_rate": 1.7468899779201897e-05,
"loss": 2.2744,
"step": 7060
},
{
"epoch": 0.38,
"learning_rate": 1.746530956612275e-05,
"loss": 2.2977,
"step": 7070
},
{
"epoch": 0.38,
"learning_rate": 1.7461719353043605e-05,
"loss": 2.2017,
"step": 7080
},
{
"epoch": 0.38,
"learning_rate": 1.745812913996446e-05,
"loss": 2.3658,
"step": 7090
},
{
"epoch": 0.38,
"learning_rate": 1.7454538926885312e-05,
"loss": 2.3233,
"step": 7100
},
{
"epoch": 0.38,
"learning_rate": 1.7450948713806165e-05,
"loss": 2.3938,
"step": 7110
},
{
"epoch": 0.38,
"learning_rate": 1.744735850072702e-05,
"loss": 2.3039,
"step": 7120
},
{
"epoch": 0.38,
"learning_rate": 1.7443768287647872e-05,
"loss": 2.4032,
"step": 7130
},
{
"epoch": 0.38,
"learning_rate": 1.7440178074568728e-05,
"loss": 2.4282,
"step": 7140
},
{
"epoch": 0.39,
"learning_rate": 1.743658786148958e-05,
"loss": 2.3364,
"step": 7150
},
{
"epoch": 0.39,
"learning_rate": 1.7432997648410435e-05,
"loss": 2.3028,
"step": 7160
},
{
"epoch": 0.39,
"learning_rate": 1.7429407435331288e-05,
"loss": 2.3196,
"step": 7170
},
{
"epoch": 0.39,
"learning_rate": 1.7425817222252143e-05,
"loss": 2.4483,
"step": 7180
},
{
"epoch": 0.39,
"learning_rate": 1.7422227009172995e-05,
"loss": 2.3709,
"step": 7190
},
{
"epoch": 0.39,
"learning_rate": 1.7418636796093847e-05,
"loss": 2.2443,
"step": 7200
},
{
"epoch": 0.39,
"learning_rate": 1.7415046583014703e-05,
"loss": 2.501,
"step": 7210
},
{
"epoch": 0.39,
"learning_rate": 1.741145636993556e-05,
"loss": 2.3832,
"step": 7220
},
{
"epoch": 0.39,
"learning_rate": 1.740786615685641e-05,
"loss": 2.3266,
"step": 7230
},
{
"epoch": 0.39,
"learning_rate": 1.7404275943777266e-05,
"loss": 2.2613,
"step": 7240
},
{
"epoch": 0.39,
"learning_rate": 1.740068573069812e-05,
"loss": 2.362,
"step": 7250
},
{
"epoch": 0.39,
"learning_rate": 1.739709551761897e-05,
"loss": 2.2679,
"step": 7260
},
{
"epoch": 0.39,
"learning_rate": 1.7393505304539826e-05,
"loss": 2.2904,
"step": 7270
},
{
"epoch": 0.39,
"learning_rate": 1.738991509146068e-05,
"loss": 2.108,
"step": 7280
},
{
"epoch": 0.39,
"learning_rate": 1.7386324878381534e-05,
"loss": 2.2689,
"step": 7290
},
{
"epoch": 0.39,
"learning_rate": 1.7382734665302386e-05,
"loss": 2.317,
"step": 7300
},
{
"epoch": 0.39,
"learning_rate": 1.737914445222324e-05,
"loss": 2.3217,
"step": 7310
},
{
"epoch": 0.39,
"learning_rate": 1.7375554239144094e-05,
"loss": 2.2032,
"step": 7320
},
{
"epoch": 0.39,
"learning_rate": 1.737196402606495e-05,
"loss": 2.3136,
"step": 7330
},
{
"epoch": 0.4,
"learning_rate": 1.73683738129858e-05,
"loss": 2.381,
"step": 7340
},
{
"epoch": 0.4,
"learning_rate": 1.7364783599906657e-05,
"loss": 2.1501,
"step": 7350
},
{
"epoch": 0.4,
"learning_rate": 1.736119338682751e-05,
"loss": 2.215,
"step": 7360
},
{
"epoch": 0.4,
"learning_rate": 1.7357603173748365e-05,
"loss": 2.2151,
"step": 7370
},
{
"epoch": 0.4,
"learning_rate": 1.7354012960669217e-05,
"loss": 2.2561,
"step": 7380
},
{
"epoch": 0.4,
"learning_rate": 1.735042274759007e-05,
"loss": 2.2621,
"step": 7390
},
{
"epoch": 0.4,
"learning_rate": 1.7346832534510924e-05,
"loss": 2.2839,
"step": 7400
},
{
"epoch": 0.4,
"learning_rate": 1.734324232143178e-05,
"loss": 2.2633,
"step": 7410
},
{
"epoch": 0.4,
"learning_rate": 1.7339652108352632e-05,
"loss": 2.2914,
"step": 7420
},
{
"epoch": 0.4,
"learning_rate": 1.7336061895273484e-05,
"loss": 2.2161,
"step": 7430
},
{
"epoch": 0.4,
"learning_rate": 1.733247168219434e-05,
"loss": 2.1786,
"step": 7440
},
{
"epoch": 0.4,
"learning_rate": 1.7328881469115192e-05,
"loss": 2.2575,
"step": 7450
},
{
"epoch": 0.4,
"learning_rate": 1.7325291256036048e-05,
"loss": 2.2606,
"step": 7460
},
{
"epoch": 0.4,
"learning_rate": 1.73217010429569e-05,
"loss": 2.2184,
"step": 7470
},
{
"epoch": 0.4,
"learning_rate": 1.7318110829877755e-05,
"loss": 2.2805,
"step": 7480
},
{
"epoch": 0.4,
"learning_rate": 1.7314520616798607e-05,
"loss": 2.3843,
"step": 7490
},
{
"epoch": 0.4,
"learning_rate": 1.7310930403719463e-05,
"loss": 2.3597,
"step": 7500
},
{
"epoch": 0.4,
"learning_rate": 1.7307340190640315e-05,
"loss": 2.2498,
"step": 7510
},
{
"epoch": 0.4,
"learning_rate": 1.730374997756117e-05,
"loss": 2.3065,
"step": 7520
},
{
"epoch": 0.41,
"learning_rate": 1.7300159764482023e-05,
"loss": 2.2381,
"step": 7530
},
{
"epoch": 0.41,
"learning_rate": 1.729656955140288e-05,
"loss": 2.1535,
"step": 7540
},
{
"epoch": 0.41,
"learning_rate": 1.729297933832373e-05,
"loss": 2.3371,
"step": 7550
},
{
"epoch": 0.41,
"learning_rate": 1.7289389125244586e-05,
"loss": 2.143,
"step": 7560
},
{
"epoch": 0.41,
"learning_rate": 1.7285798912165438e-05,
"loss": 2.4662,
"step": 7570
},
{
"epoch": 0.41,
"learning_rate": 1.728220869908629e-05,
"loss": 2.2632,
"step": 7580
},
{
"epoch": 0.41,
"learning_rate": 1.7278618486007146e-05,
"loss": 2.2501,
"step": 7590
},
{
"epoch": 0.41,
"learning_rate": 1.7275028272928e-05,
"loss": 2.2385,
"step": 7600
},
{
"epoch": 0.41,
"learning_rate": 1.7271438059848854e-05,
"loss": 2.1904,
"step": 7610
},
{
"epoch": 0.41,
"learning_rate": 1.7267847846769706e-05,
"loss": 2.2508,
"step": 7620
},
{
"epoch": 0.41,
"learning_rate": 1.726425763369056e-05,
"loss": 2.2618,
"step": 7630
},
{
"epoch": 0.41,
"learning_rate": 1.7260667420611414e-05,
"loss": 2.3615,
"step": 7640
},
{
"epoch": 0.41,
"learning_rate": 1.725707720753227e-05,
"loss": 2.273,
"step": 7650
},
{
"epoch": 0.41,
"learning_rate": 1.725348699445312e-05,
"loss": 2.2675,
"step": 7660
},
{
"epoch": 0.41,
"learning_rate": 1.7249896781373977e-05,
"loss": 2.1635,
"step": 7670
},
{
"epoch": 0.41,
"learning_rate": 1.724630656829483e-05,
"loss": 2.2832,
"step": 7680
},
{
"epoch": 0.41,
"learning_rate": 1.7242716355215684e-05,
"loss": 2.2099,
"step": 7690
},
{
"epoch": 0.41,
"learning_rate": 1.7239126142136537e-05,
"loss": 2.3396,
"step": 7700
},
{
"epoch": 0.42,
"learning_rate": 1.7235535929057392e-05,
"loss": 2.2692,
"step": 7710
},
{
"epoch": 0.42,
"learning_rate": 1.7231945715978244e-05,
"loss": 2.3635,
"step": 7720
},
{
"epoch": 0.42,
"learning_rate": 1.72283555028991e-05,
"loss": 2.2206,
"step": 7730
},
{
"epoch": 0.42,
"learning_rate": 1.7224765289819952e-05,
"loss": 2.2761,
"step": 7740
},
{
"epoch": 0.42,
"learning_rate": 1.7221175076740808e-05,
"loss": 2.3127,
"step": 7750
},
{
"epoch": 0.42,
"learning_rate": 1.721758486366166e-05,
"loss": 2.2085,
"step": 7760
},
{
"epoch": 0.42,
"learning_rate": 1.7213994650582512e-05,
"loss": 2.2609,
"step": 7770
},
{
"epoch": 0.42,
"learning_rate": 1.7210404437503367e-05,
"loss": 2.1903,
"step": 7780
},
{
"epoch": 0.42,
"learning_rate": 1.7206814224424223e-05,
"loss": 2.2663,
"step": 7790
},
{
"epoch": 0.42,
"learning_rate": 1.7203224011345075e-05,
"loss": 2.2721,
"step": 7800
},
{
"epoch": 0.42,
"learning_rate": 1.7199633798265927e-05,
"loss": 2.3648,
"step": 7810
},
{
"epoch": 0.42,
"learning_rate": 1.7196043585186783e-05,
"loss": 2.1694,
"step": 7820
},
{
"epoch": 0.42,
"learning_rate": 1.7192453372107635e-05,
"loss": 2.2729,
"step": 7830
},
{
"epoch": 0.42,
"learning_rate": 1.718886315902849e-05,
"loss": 2.2186,
"step": 7840
},
{
"epoch": 0.42,
"learning_rate": 1.7185272945949343e-05,
"loss": 2.4567,
"step": 7850
},
{
"epoch": 0.42,
"learning_rate": 1.7181682732870198e-05,
"loss": 2.4213,
"step": 7860
},
{
"epoch": 0.42,
"learning_rate": 1.717809251979105e-05,
"loss": 2.2289,
"step": 7870
},
{
"epoch": 0.42,
"learning_rate": 1.7174502306711906e-05,
"loss": 2.2743,
"step": 7880
},
{
"epoch": 0.42,
"learning_rate": 1.7170912093632758e-05,
"loss": 2.4067,
"step": 7890
},
{
"epoch": 0.43,
"learning_rate": 1.716732188055361e-05,
"loss": 2.2565,
"step": 7900
},
{
"epoch": 0.43,
"learning_rate": 1.7163731667474466e-05,
"loss": 2.1664,
"step": 7910
},
{
"epoch": 0.43,
"learning_rate": 1.716014145439532e-05,
"loss": 2.3281,
"step": 7920
},
{
"epoch": 0.43,
"learning_rate": 1.7156551241316174e-05,
"loss": 2.2045,
"step": 7930
},
{
"epoch": 0.43,
"learning_rate": 1.7152961028237026e-05,
"loss": 2.2015,
"step": 7940
},
{
"epoch": 0.43,
"learning_rate": 1.714937081515788e-05,
"loss": 2.2742,
"step": 7950
},
{
"epoch": 0.43,
"learning_rate": 1.7145780602078733e-05,
"loss": 2.3435,
"step": 7960
},
{
"epoch": 0.43,
"learning_rate": 1.714219038899959e-05,
"loss": 2.2946,
"step": 7970
},
{
"epoch": 0.43,
"learning_rate": 1.713860017592044e-05,
"loss": 2.3054,
"step": 7980
},
{
"epoch": 0.43,
"learning_rate": 1.7135009962841297e-05,
"loss": 2.2602,
"step": 7990
},
{
"epoch": 0.43,
"learning_rate": 1.713141974976215e-05,
"loss": 2.1613,
"step": 8000
},
{
"epoch": 0.43,
"learning_rate": 1.7127829536683004e-05,
"loss": 2.1763,
"step": 8010
},
{
"epoch": 0.43,
"learning_rate": 1.7124239323603856e-05,
"loss": 2.195,
"step": 8020
},
{
"epoch": 0.43,
"learning_rate": 1.7120649110524712e-05,
"loss": 2.1862,
"step": 8030
},
{
"epoch": 0.43,
"learning_rate": 1.7117058897445564e-05,
"loss": 2.1873,
"step": 8040
},
{
"epoch": 0.43,
"learning_rate": 1.711346868436642e-05,
"loss": 2.3154,
"step": 8050
},
{
"epoch": 0.43,
"learning_rate": 1.7109878471287272e-05,
"loss": 2.3125,
"step": 8060
},
{
"epoch": 0.43,
"learning_rate": 1.7106288258208127e-05,
"loss": 2.3045,
"step": 8070
},
{
"epoch": 0.44,
"learning_rate": 1.710269804512898e-05,
"loss": 2.2409,
"step": 8080
},
{
"epoch": 0.44,
"learning_rate": 1.7099107832049832e-05,
"loss": 2.1399,
"step": 8090
},
{
"epoch": 0.44,
"learning_rate": 1.7095517618970687e-05,
"loss": 2.2779,
"step": 8100
},
{
"epoch": 0.44,
"learning_rate": 1.7091927405891543e-05,
"loss": 2.2896,
"step": 8110
},
{
"epoch": 0.44,
"learning_rate": 1.7088337192812395e-05,
"loss": 2.1175,
"step": 8120
},
{
"epoch": 0.44,
"learning_rate": 1.7084746979733247e-05,
"loss": 2.3036,
"step": 8130
},
{
"epoch": 0.44,
"learning_rate": 1.7081156766654103e-05,
"loss": 2.1518,
"step": 8140
},
{
"epoch": 0.44,
"learning_rate": 1.7077566553574955e-05,
"loss": 2.265,
"step": 8150
},
{
"epoch": 0.44,
"learning_rate": 1.707397634049581e-05,
"loss": 2.2341,
"step": 8160
},
{
"epoch": 0.44,
"learning_rate": 1.7070386127416663e-05,
"loss": 2.1126,
"step": 8170
},
{
"epoch": 0.44,
"learning_rate": 1.7066795914337518e-05,
"loss": 2.2285,
"step": 8180
},
{
"epoch": 0.44,
"learning_rate": 1.706320570125837e-05,
"loss": 2.298,
"step": 8190
},
{
"epoch": 0.44,
"learning_rate": 1.7059615488179226e-05,
"loss": 2.343,
"step": 8200
},
{
"epoch": 0.44,
"learning_rate": 1.7056025275100078e-05,
"loss": 2.2423,
"step": 8210
},
{
"epoch": 0.44,
"learning_rate": 1.7052435062020933e-05,
"loss": 2.2843,
"step": 8220
},
{
"epoch": 0.44,
"learning_rate": 1.7048844848941786e-05,
"loss": 2.2622,
"step": 8230
},
{
"epoch": 0.44,
"learning_rate": 1.704525463586264e-05,
"loss": 2.2375,
"step": 8240
},
{
"epoch": 0.44,
"learning_rate": 1.7041664422783493e-05,
"loss": 2.2831,
"step": 8250
},
{
"epoch": 0.44,
"learning_rate": 1.703807420970435e-05,
"loss": 2.1778,
"step": 8260
},
{
"epoch": 0.45,
"learning_rate": 1.70344839966252e-05,
"loss": 2.1791,
"step": 8270
},
{
"epoch": 0.45,
"learning_rate": 1.7030893783546053e-05,
"loss": 2.306,
"step": 8280
},
{
"epoch": 0.45,
"learning_rate": 1.702730357046691e-05,
"loss": 2.3793,
"step": 8290
},
{
"epoch": 0.45,
"learning_rate": 1.7023713357387764e-05,
"loss": 2.2369,
"step": 8300
},
{
"epoch": 0.45,
"learning_rate": 1.7020123144308616e-05,
"loss": 2.2126,
"step": 8310
},
{
"epoch": 0.45,
"learning_rate": 1.701653293122947e-05,
"loss": 2.226,
"step": 8320
},
{
"epoch": 0.45,
"learning_rate": 1.7012942718150324e-05,
"loss": 2.2276,
"step": 8330
},
{
"epoch": 0.45,
"learning_rate": 1.7009352505071176e-05,
"loss": 2.2843,
"step": 8340
},
{
"epoch": 0.45,
"learning_rate": 1.7005762291992032e-05,
"loss": 2.36,
"step": 8350
},
{
"epoch": 0.45,
"learning_rate": 1.7002172078912884e-05,
"loss": 2.2753,
"step": 8360
},
{
"epoch": 0.45,
"learning_rate": 1.699858186583374e-05,
"loss": 2.3235,
"step": 8370
},
{
"epoch": 0.45,
"learning_rate": 1.6994991652754592e-05,
"loss": 2.2682,
"step": 8380
},
{
"epoch": 0.45,
"learning_rate": 1.6991401439675447e-05,
"loss": 2.2015,
"step": 8390
},
{
"epoch": 0.45,
"learning_rate": 1.69878112265963e-05,
"loss": 2.1672,
"step": 8400
},
{
"epoch": 0.45,
"learning_rate": 1.698422101351715e-05,
"loss": 2.2932,
"step": 8410
},
{
"epoch": 0.45,
"learning_rate": 1.6980630800438007e-05,
"loss": 2.1829,
"step": 8420
},
{
"epoch": 0.45,
"learning_rate": 1.6977040587358863e-05,
"loss": 2.1634,
"step": 8430
},
{
"epoch": 0.45,
"learning_rate": 1.6973450374279715e-05,
"loss": 2.0578,
"step": 8440
},
{
"epoch": 0.46,
"learning_rate": 1.696986016120057e-05,
"loss": 2.2226,
"step": 8450
},
{
"epoch": 0.46,
"learning_rate": 1.6966269948121423e-05,
"loss": 2.3056,
"step": 8460
},
{
"epoch": 0.46,
"learning_rate": 1.6962679735042275e-05,
"loss": 2.2172,
"step": 8470
},
{
"epoch": 0.46,
"learning_rate": 1.695908952196313e-05,
"loss": 2.2584,
"step": 8480
},
{
"epoch": 0.46,
"learning_rate": 1.6955499308883982e-05,
"loss": 2.0855,
"step": 8490
},
{
"epoch": 0.46,
"learning_rate": 1.6951909095804838e-05,
"loss": 2.1562,
"step": 8500
}
],
"max_steps": 55707,
"num_train_epochs": 3,
"total_flos": 1.9077026634163814e+17,
"trial_name": null,
"trial_params": null
}