dumb-dev's picture
at least i generates some kind of output now...
8bb1997 verified
raw
history blame contribute delete
No virus
47.4 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 5.0,
"eval_steps": 500,
"global_step": 3710,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 9.999875511956753e-05,
"loss": 1.7561,
"step": 10
},
{
"epoch": 0.03,
"learning_rate": 9.999502054025923e-05,
"loss": 1.7194,
"step": 20
},
{
"epoch": 0.04,
"learning_rate": 9.998879644803927e-05,
"loss": 1.7705,
"step": 30
},
{
"epoch": 0.05,
"learning_rate": 9.998008315283767e-05,
"loss": 1.5094,
"step": 40
},
{
"epoch": 0.07,
"learning_rate": 9.996888108853486e-05,
"loss": 1.5201,
"step": 50
},
{
"epoch": 0.08,
"learning_rate": 9.995519081294009e-05,
"loss": 1.637,
"step": 60
},
{
"epoch": 0.09,
"learning_rate": 9.993901300776359e-05,
"loss": 1.6338,
"step": 70
},
{
"epoch": 0.11,
"learning_rate": 9.992034847858267e-05,
"loss": 1.5496,
"step": 80
},
{
"epoch": 0.12,
"learning_rate": 9.989919815480165e-05,
"loss": 1.5669,
"step": 90
},
{
"epoch": 0.13,
"learning_rate": 9.987556308960549e-05,
"loss": 1.5622,
"step": 100
},
{
"epoch": 0.15,
"learning_rate": 9.984944445990735e-05,
"loss": 1.5836,
"step": 110
},
{
"epoch": 0.16,
"learning_rate": 9.982084356629015e-05,
"loss": 1.5877,
"step": 120
},
{
"epoch": 0.18,
"learning_rate": 9.978976183294153e-05,
"loss": 1.5465,
"step": 130
},
{
"epoch": 0.19,
"learning_rate": 9.97562008075832e-05,
"loss": 1.4834,
"step": 140
},
{
"epoch": 0.2,
"learning_rate": 9.97201621613937e-05,
"loss": 1.4956,
"step": 150
},
{
"epoch": 0.22,
"learning_rate": 9.968164768892526e-05,
"loss": 1.3877,
"step": 160
},
{
"epoch": 0.23,
"learning_rate": 9.964065930801438e-05,
"loss": 1.3466,
"step": 170
},
{
"epoch": 0.24,
"learning_rate": 9.95971990596864e-05,
"loss": 1.4023,
"step": 180
},
{
"epoch": 0.26,
"learning_rate": 9.955126910805386e-05,
"loss": 1.351,
"step": 190
},
{
"epoch": 0.27,
"learning_rate": 9.950287174020863e-05,
"loss": 1.424,
"step": 200
},
{
"epoch": 0.28,
"learning_rate": 9.94520093661082e-05,
"loss": 1.5038,
"step": 210
},
{
"epoch": 0.3,
"learning_rate": 9.939868451845554e-05,
"loss": 1.3436,
"step": 220
},
{
"epoch": 0.31,
"learning_rate": 9.934289985257299e-05,
"loss": 1.3446,
"step": 230
},
{
"epoch": 0.32,
"learning_rate": 9.928465814627014e-05,
"loss": 1.4063,
"step": 240
},
{
"epoch": 0.34,
"learning_rate": 9.922396229970541e-05,
"loss": 1.3803,
"step": 250
},
{
"epoch": 0.35,
"learning_rate": 9.916081533524167e-05,
"loss": 1.2975,
"step": 260
},
{
"epoch": 0.36,
"learning_rate": 9.909522039729571e-05,
"loss": 1.3689,
"step": 270
},
{
"epoch": 0.38,
"learning_rate": 9.902718075218176e-05,
"loss": 1.4446,
"step": 280
},
{
"epoch": 0.39,
"learning_rate": 9.895669978794869e-05,
"loss": 1.237,
"step": 290
},
{
"epoch": 0.4,
"learning_rate": 9.888378101421147e-05,
"loss": 1.333,
"step": 300
},
{
"epoch": 0.42,
"learning_rate": 9.880842806197625e-05,
"loss": 1.3933,
"step": 310
},
{
"epoch": 0.43,
"learning_rate": 9.873064468345969e-05,
"loss": 1.3735,
"step": 320
},
{
"epoch": 0.44,
"learning_rate": 9.865043475190201e-05,
"loss": 1.172,
"step": 330
},
{
"epoch": 0.46,
"learning_rate": 9.856780226137419e-05,
"loss": 1.2538,
"step": 340
},
{
"epoch": 0.47,
"learning_rate": 9.848275132657903e-05,
"loss": 1.2588,
"step": 350
},
{
"epoch": 0.49,
"learning_rate": 9.839528618264633e-05,
"loss": 1.3531,
"step": 360
},
{
"epoch": 0.5,
"learning_rate": 9.830541118492193e-05,
"loss": 1.4203,
"step": 370
},
{
"epoch": 0.51,
"learning_rate": 9.821313080875088e-05,
"loss": 1.4292,
"step": 380
},
{
"epoch": 0.53,
"learning_rate": 9.811844964925454e-05,
"loss": 1.3098,
"step": 390
},
{
"epoch": 0.54,
"learning_rate": 9.802137242110185e-05,
"loss": 1.2093,
"step": 400
},
{
"epoch": 0.55,
"learning_rate": 9.792190395827447e-05,
"loss": 1.2182,
"step": 410
},
{
"epoch": 0.57,
"learning_rate": 9.782004921382612e-05,
"loss": 1.3148,
"step": 420
},
{
"epoch": 0.58,
"learning_rate": 9.771581325963594e-05,
"loss": 1.4185,
"step": 430
},
{
"epoch": 0.59,
"learning_rate": 9.760920128615591e-05,
"loss": 1.409,
"step": 440
},
{
"epoch": 0.61,
"learning_rate": 9.750021860215241e-05,
"loss": 1.2089,
"step": 450
},
{
"epoch": 0.62,
"learning_rate": 9.738887063444188e-05,
"loss": 1.3114,
"step": 460
},
{
"epoch": 0.63,
"learning_rate": 9.727516292762058e-05,
"loss": 1.2528,
"step": 470
},
{
"epoch": 0.65,
"learning_rate": 9.715910114378845e-05,
"loss": 1.1718,
"step": 480
},
{
"epoch": 0.66,
"learning_rate": 9.704069106226727e-05,
"loss": 1.2825,
"step": 490
},
{
"epoch": 0.67,
"learning_rate": 9.691993857931277e-05,
"loss": 1.1629,
"step": 500
},
{
"epoch": 0.69,
"learning_rate": 9.679684970782106e-05,
"loss": 1.2463,
"step": 510
},
{
"epoch": 0.7,
"learning_rate": 9.667143057702926e-05,
"loss": 1.3297,
"step": 520
},
{
"epoch": 0.71,
"learning_rate": 9.654368743221022e-05,
"loss": 1.1995,
"step": 530
},
{
"epoch": 0.73,
"learning_rate": 9.641362663436161e-05,
"loss": 1.1135,
"step": 540
},
{
"epoch": 0.74,
"learning_rate": 9.628125465988913e-05,
"loss": 1.1311,
"step": 550
},
{
"epoch": 0.75,
"learning_rate": 9.614657810028403e-05,
"loss": 1.2915,
"step": 560
},
{
"epoch": 0.77,
"learning_rate": 9.60096036617948e-05,
"loss": 1.1898,
"step": 570
},
{
"epoch": 0.78,
"learning_rate": 9.587033816509341e-05,
"loss": 1.211,
"step": 580
},
{
"epoch": 0.8,
"learning_rate": 9.572878854493553e-05,
"loss": 1.1389,
"step": 590
},
{
"epoch": 0.81,
"learning_rate": 9.558496184981525e-05,
"loss": 1.0623,
"step": 600
},
{
"epoch": 0.82,
"learning_rate": 9.543886524161409e-05,
"loss": 1.1475,
"step": 610
},
{
"epoch": 0.84,
"learning_rate": 9.529050599524443e-05,
"loss": 1.0662,
"step": 620
},
{
"epoch": 0.85,
"learning_rate": 9.513989149828718e-05,
"loss": 1.2403,
"step": 630
},
{
"epoch": 0.86,
"learning_rate": 9.498702925062393e-05,
"loss": 1.3078,
"step": 640
},
{
"epoch": 0.88,
"learning_rate": 9.48319268640635e-05,
"loss": 1.2456,
"step": 650
},
{
"epoch": 0.89,
"learning_rate": 9.467459206196298e-05,
"loss": 1.1248,
"step": 660
},
{
"epoch": 0.9,
"learning_rate": 9.451503267884299e-05,
"loss": 1.0899,
"step": 670
},
{
"epoch": 0.92,
"learning_rate": 9.435325665999771e-05,
"loss": 1.3189,
"step": 680
},
{
"epoch": 0.93,
"learning_rate": 9.418927206109913e-05,
"loss": 0.9791,
"step": 690
},
{
"epoch": 0.94,
"learning_rate": 9.402308704779599e-05,
"loss": 1.112,
"step": 700
},
{
"epoch": 0.96,
"learning_rate": 9.385470989530716e-05,
"loss": 1.0378,
"step": 710
},
{
"epoch": 0.97,
"learning_rate": 9.368414898800952e-05,
"loss": 1.2503,
"step": 720
},
{
"epoch": 0.98,
"learning_rate": 9.35114128190205e-05,
"loss": 1.0507,
"step": 730
},
{
"epoch": 1.0,
"learning_rate": 9.333650998977518e-05,
"loss": 1.0279,
"step": 740
},
{
"epoch": 1.01,
"learning_rate": 9.315944920959798e-05,
"loss": 0.9083,
"step": 750
},
{
"epoch": 1.02,
"learning_rate": 9.298023929526888e-05,
"loss": 1.0812,
"step": 760
},
{
"epoch": 1.04,
"learning_rate": 9.279888917058452e-05,
"loss": 1.0576,
"step": 770
},
{
"epoch": 1.05,
"learning_rate": 9.261540786591377e-05,
"loss": 1.2718,
"step": 780
},
{
"epoch": 1.06,
"learning_rate": 9.242980451774807e-05,
"loss": 1.2115,
"step": 790
},
{
"epoch": 1.08,
"learning_rate": 9.224208836824647e-05,
"loss": 1.0058,
"step": 800
},
{
"epoch": 1.09,
"learning_rate": 9.20522687647754e-05,
"loss": 1.1806,
"step": 810
},
{
"epoch": 1.11,
"learning_rate": 9.18603551594433e-05,
"loss": 1.1535,
"step": 820
},
{
"epoch": 1.12,
"learning_rate": 9.166635710862983e-05,
"loss": 1.0037,
"step": 830
},
{
"epoch": 1.13,
"learning_rate": 9.14702842725101e-05,
"loss": 0.9461,
"step": 840
},
{
"epoch": 1.15,
"learning_rate": 9.127214641457358e-05,
"loss": 1.1394,
"step": 850
},
{
"epoch": 1.16,
"learning_rate": 9.107195340113794e-05,
"loss": 0.9338,
"step": 860
},
{
"epoch": 1.17,
"learning_rate": 9.086971520085783e-05,
"loss": 1.0181,
"step": 870
},
{
"epoch": 1.19,
"learning_rate": 9.066544188422836e-05,
"loss": 0.9725,
"step": 880
},
{
"epoch": 1.2,
"learning_rate": 9.04591436230837e-05,
"loss": 1.0929,
"step": 890
},
{
"epoch": 1.21,
"learning_rate": 9.025083069009063e-05,
"loss": 1.1141,
"step": 900
},
{
"epoch": 1.23,
"learning_rate": 9.004051345823689e-05,
"loss": 1.1072,
"step": 910
},
{
"epoch": 1.24,
"learning_rate": 8.982820240031474e-05,
"loss": 1.0488,
"step": 920
},
{
"epoch": 1.25,
"learning_rate": 8.961390808839945e-05,
"loss": 1.0832,
"step": 930
},
{
"epoch": 1.27,
"learning_rate": 8.939764119332284e-05,
"loss": 0.9583,
"step": 940
},
{
"epoch": 1.28,
"learning_rate": 8.917941248414196e-05,
"loss": 1.0901,
"step": 950
},
{
"epoch": 1.29,
"learning_rate": 8.895923282760279e-05,
"loss": 1.2054,
"step": 960
},
{
"epoch": 1.31,
"learning_rate": 8.873711318759919e-05,
"loss": 1.1045,
"step": 970
},
{
"epoch": 1.32,
"learning_rate": 8.851306462462688e-05,
"loss": 1.2223,
"step": 980
},
{
"epoch": 1.33,
"learning_rate": 8.828709829523276e-05,
"loss": 1.1103,
"step": 990
},
{
"epoch": 1.35,
"learning_rate": 8.805922545145928e-05,
"loss": 1.1602,
"step": 1000
},
{
"epoch": 1.36,
"learning_rate": 8.782945744028424e-05,
"loss": 1.0639,
"step": 1010
},
{
"epoch": 1.37,
"learning_rate": 8.759780570305565e-05,
"loss": 1.2104,
"step": 1020
},
{
"epoch": 1.39,
"learning_rate": 8.736428177492213e-05,
"loss": 0.85,
"step": 1030
},
{
"epoch": 1.4,
"learning_rate": 8.712889728425842e-05,
"loss": 0.9958,
"step": 1040
},
{
"epoch": 1.42,
"learning_rate": 8.689166395208636e-05,
"loss": 0.9216,
"step": 1050
},
{
"epoch": 1.43,
"learning_rate": 8.665259359149132e-05,
"loss": 0.9329,
"step": 1060
},
{
"epoch": 1.44,
"learning_rate": 8.641169810703382e-05,
"loss": 1.0304,
"step": 1070
},
{
"epoch": 1.46,
"learning_rate": 8.616898949415686e-05,
"loss": 1.0397,
"step": 1080
},
{
"epoch": 1.47,
"learning_rate": 8.592447983858858e-05,
"loss": 1.0056,
"step": 1090
},
{
"epoch": 1.48,
"learning_rate": 8.567818131574039e-05,
"loss": 0.9399,
"step": 1100
},
{
"epoch": 1.5,
"learning_rate": 8.543010619010077e-05,
"loss": 1.0151,
"step": 1110
},
{
"epoch": 1.51,
"learning_rate": 8.518026681462448e-05,
"loss": 0.8214,
"step": 1120
},
{
"epoch": 1.52,
"learning_rate": 8.492867563011752e-05,
"loss": 0.8622,
"step": 1130
},
{
"epoch": 1.54,
"learning_rate": 8.467534516461763e-05,
"loss": 0.962,
"step": 1140
},
{
"epoch": 1.55,
"learning_rate": 8.442028803277035e-05,
"loss": 1.2468,
"step": 1150
},
{
"epoch": 1.56,
"learning_rate": 8.4163516935201e-05,
"loss": 0.762,
"step": 1160
},
{
"epoch": 1.58,
"learning_rate": 8.390504465788216e-05,
"loss": 1.1067,
"step": 1170
},
{
"epoch": 1.59,
"learning_rate": 8.364488407149707e-05,
"loss": 1.0618,
"step": 1180
},
{
"epoch": 1.6,
"learning_rate": 8.338304813079865e-05,
"loss": 0.8968,
"step": 1190
},
{
"epoch": 1.62,
"learning_rate": 8.311954987396444e-05,
"loss": 0.9637,
"step": 1200
},
{
"epoch": 1.63,
"learning_rate": 8.285440242194744e-05,
"loss": 0.8944,
"step": 1210
},
{
"epoch": 1.64,
"learning_rate": 8.258761897782264e-05,
"loss": 0.9935,
"step": 1220
},
{
"epoch": 1.66,
"learning_rate": 8.231921282612957e-05,
"loss": 1.0193,
"step": 1230
},
{
"epoch": 1.67,
"learning_rate": 8.204919733221089e-05,
"loss": 0.9402,
"step": 1240
},
{
"epoch": 1.68,
"learning_rate": 8.177758594154681e-05,
"loss": 0.7653,
"step": 1250
},
{
"epoch": 1.7,
"learning_rate": 8.150439217908556e-05,
"loss": 0.9966,
"step": 1260
},
{
"epoch": 1.71,
"learning_rate": 8.122962964856989e-05,
"loss": 0.8538,
"step": 1270
},
{
"epoch": 1.73,
"learning_rate": 8.095331203185968e-05,
"loss": 1.078,
"step": 1280
},
{
"epoch": 1.74,
"learning_rate": 8.067545308825075e-05,
"loss": 0.8603,
"step": 1290
},
{
"epoch": 1.75,
"learning_rate": 8.039606665378956e-05,
"loss": 0.6651,
"step": 1300
},
{
"epoch": 1.77,
"learning_rate": 8.011516664058431e-05,
"loss": 0.7974,
"step": 1310
},
{
"epoch": 1.78,
"learning_rate": 7.983276703611222e-05,
"loss": 0.8603,
"step": 1320
},
{
"epoch": 1.79,
"learning_rate": 7.954888190252292e-05,
"loss": 0.8481,
"step": 1330
},
{
"epoch": 1.81,
"learning_rate": 7.926352537593837e-05,
"loss": 0.8794,
"step": 1340
},
{
"epoch": 1.82,
"learning_rate": 7.897671166574878e-05,
"loss": 0.6893,
"step": 1350
},
{
"epoch": 1.83,
"learning_rate": 7.86884550539052e-05,
"loss": 1.076,
"step": 1360
},
{
"epoch": 1.85,
"learning_rate": 7.839876989420827e-05,
"loss": 0.8875,
"step": 1370
},
{
"epoch": 1.86,
"learning_rate": 7.810767061159341e-05,
"loss": 0.7543,
"step": 1380
},
{
"epoch": 1.87,
"learning_rate": 7.78151717014127e-05,
"loss": 0.9026,
"step": 1390
},
{
"epoch": 1.89,
"learning_rate": 7.752128772871292e-05,
"loss": 0.7245,
"step": 1400
},
{
"epoch": 1.9,
"learning_rate": 7.722603332751034e-05,
"loss": 1.0003,
"step": 1410
},
{
"epoch": 1.91,
"learning_rate": 7.692942320006204e-05,
"loss": 1.08,
"step": 1420
},
{
"epoch": 1.93,
"learning_rate": 7.663147211613378e-05,
"loss": 0.9045,
"step": 1430
},
{
"epoch": 1.94,
"learning_rate": 7.633219491226448e-05,
"loss": 0.9683,
"step": 1440
},
{
"epoch": 1.95,
"learning_rate": 7.60316064910276e-05,
"loss": 1.0975,
"step": 1450
},
{
"epoch": 1.97,
"learning_rate": 7.572972182028885e-05,
"loss": 0.8337,
"step": 1460
},
{
"epoch": 1.98,
"learning_rate": 7.542655593246103e-05,
"loss": 0.944,
"step": 1470
},
{
"epoch": 1.99,
"learning_rate": 7.512212392375539e-05,
"loss": 0.9337,
"step": 1480
},
{
"epoch": 2.01,
"learning_rate": 7.481644095342996e-05,
"loss": 0.9375,
"step": 1490
},
{
"epoch": 2.02,
"learning_rate": 7.45095222430347e-05,
"loss": 0.8512,
"step": 1500
},
{
"epoch": 2.04,
"learning_rate": 7.420138307565344e-05,
"loss": 0.8152,
"step": 1510
},
{
"epoch": 2.05,
"learning_rate": 7.389203879514299e-05,
"loss": 1.1356,
"step": 1520
},
{
"epoch": 2.06,
"learning_rate": 7.358150480536904e-05,
"loss": 0.8811,
"step": 1530
},
{
"epoch": 2.08,
"learning_rate": 7.326979656943906e-05,
"loss": 0.7049,
"step": 1540
},
{
"epoch": 2.09,
"learning_rate": 7.295692960893242e-05,
"loss": 0.6571,
"step": 1550
},
{
"epoch": 2.1,
"learning_rate": 7.264291950312741e-05,
"loss": 0.8488,
"step": 1560
},
{
"epoch": 2.12,
"learning_rate": 7.232778188822543e-05,
"loss": 1.1176,
"step": 1570
},
{
"epoch": 2.13,
"learning_rate": 7.201153245657254e-05,
"loss": 0.8774,
"step": 1580
},
{
"epoch": 2.14,
"learning_rate": 7.169418695587791e-05,
"loss": 1.0086,
"step": 1590
},
{
"epoch": 2.16,
"learning_rate": 7.137576118842968e-05,
"loss": 0.8206,
"step": 1600
},
{
"epoch": 2.17,
"learning_rate": 7.105627101030817e-05,
"loss": 0.796,
"step": 1610
},
{
"epoch": 2.18,
"learning_rate": 7.073573233059617e-05,
"loss": 0.8781,
"step": 1620
},
{
"epoch": 2.2,
"learning_rate": 7.041416111058694e-05,
"loss": 0.7912,
"step": 1630
},
{
"epoch": 2.21,
"learning_rate": 7.009157336298924e-05,
"loss": 0.9046,
"step": 1640
},
{
"epoch": 2.22,
"learning_rate": 6.976798515113004e-05,
"loss": 0.8938,
"step": 1650
},
{
"epoch": 2.24,
"learning_rate": 6.94434125881547e-05,
"loss": 0.9112,
"step": 1660
},
{
"epoch": 2.25,
"learning_rate": 6.91178718362245e-05,
"loss": 0.988,
"step": 1670
},
{
"epoch": 2.26,
"learning_rate": 6.879137910571191e-05,
"loss": 0.8628,
"step": 1680
},
{
"epoch": 2.28,
"learning_rate": 6.846395065439342e-05,
"loss": 0.7263,
"step": 1690
},
{
"epoch": 2.29,
"learning_rate": 6.813560278663991e-05,
"loss": 0.7749,
"step": 1700
},
{
"epoch": 2.3,
"learning_rate": 6.780635185260479e-05,
"loss": 0.6977,
"step": 1710
},
{
"epoch": 2.32,
"learning_rate": 6.747621424740987e-05,
"loss": 0.8732,
"step": 1720
},
{
"epoch": 2.33,
"learning_rate": 6.714520641032894e-05,
"loss": 0.5676,
"step": 1730
},
{
"epoch": 2.35,
"learning_rate": 6.68133448239692e-05,
"loss": 0.7687,
"step": 1740
},
{
"epoch": 2.36,
"learning_rate": 6.64806460134504e-05,
"loss": 0.6224,
"step": 1750
},
{
"epoch": 2.37,
"learning_rate": 6.614712654558215e-05,
"loss": 0.7878,
"step": 1760
},
{
"epoch": 2.39,
"learning_rate": 6.581280302803881e-05,
"loss": 0.8812,
"step": 1770
},
{
"epoch": 2.4,
"learning_rate": 6.547769210853259e-05,
"loss": 0.972,
"step": 1780
},
{
"epoch": 2.41,
"learning_rate": 6.514181047398455e-05,
"loss": 0.7896,
"step": 1790
},
{
"epoch": 2.43,
"learning_rate": 6.480517484969366e-05,
"loss": 0.7557,
"step": 1800
},
{
"epoch": 2.44,
"learning_rate": 6.446780199850399e-05,
"loss": 0.933,
"step": 1810
},
{
"epoch": 2.45,
"learning_rate": 6.412970871996995e-05,
"loss": 0.6527,
"step": 1820
},
{
"epoch": 2.47,
"learning_rate": 6.379091184951984e-05,
"loss": 0.808,
"step": 1830
},
{
"epoch": 2.48,
"learning_rate": 6.345142825761744e-05,
"loss": 0.9055,
"step": 1840
},
{
"epoch": 2.49,
"learning_rate": 6.311127484892197e-05,
"loss": 0.7223,
"step": 1850
},
{
"epoch": 2.51,
"learning_rate": 6.277046856144633e-05,
"loss": 0.6202,
"step": 1860
},
{
"epoch": 2.52,
"learning_rate": 6.242902636571368e-05,
"loss": 0.7735,
"step": 1870
},
{
"epoch": 2.53,
"learning_rate": 6.208696526391234e-05,
"loss": 0.9887,
"step": 1880
},
{
"epoch": 2.55,
"learning_rate": 6.174430228904919e-05,
"loss": 0.7618,
"step": 1890
},
{
"epoch": 2.56,
"learning_rate": 6.140105450410154e-05,
"loss": 0.8471,
"step": 1900
},
{
"epoch": 2.57,
"learning_rate": 6.105723900116742e-05,
"loss": 0.8222,
"step": 1910
},
{
"epoch": 2.59,
"learning_rate": 6.071287290061451e-05,
"loss": 1.0693,
"step": 1920
},
{
"epoch": 2.6,
"learning_rate": 6.036797335022763e-05,
"loss": 0.5978,
"step": 1930
},
{
"epoch": 2.61,
"learning_rate": 6.002255752435483e-05,
"loss": 0.7915,
"step": 1940
},
{
"epoch": 2.63,
"learning_rate": 5.9676642623052213e-05,
"loss": 0.5858,
"step": 1950
},
{
"epoch": 2.64,
"learning_rate": 5.9330245871227454e-05,
"loss": 0.7407,
"step": 1960
},
{
"epoch": 2.65,
"learning_rate": 5.898338451778209e-05,
"loss": 0.7015,
"step": 1970
},
{
"epoch": 2.67,
"learning_rate": 5.863607583475259e-05,
"loss": 0.6859,
"step": 1980
},
{
"epoch": 2.68,
"learning_rate": 5.8288337116450276e-05,
"loss": 0.7478,
"step": 1990
},
{
"epoch": 2.7,
"learning_rate": 5.794018567860021e-05,
"loss": 0.5269,
"step": 2000
},
{
"epoch": 2.71,
"learning_rate": 5.75916388574789e-05,
"loss": 0.648,
"step": 2010
},
{
"epoch": 2.72,
"learning_rate": 5.724271400905101e-05,
"loss": 0.7577,
"step": 2020
},
{
"epoch": 2.74,
"learning_rate": 5.6893428508105225e-05,
"loss": 0.8501,
"step": 2030
},
{
"epoch": 2.75,
"learning_rate": 5.654379974738892e-05,
"loss": 0.8003,
"step": 2040
},
{
"epoch": 2.76,
"learning_rate": 5.619384513674226e-05,
"loss": 0.6311,
"step": 2050
},
{
"epoch": 2.78,
"learning_rate": 5.5843582102231096e-05,
"loss": 0.704,
"step": 2060
},
{
"epoch": 2.79,
"learning_rate": 5.5493028085279354e-05,
"loss": 0.6962,
"step": 2070
},
{
"epoch": 2.8,
"learning_rate": 5.514220054180046e-05,
"loss": 0.8652,
"step": 2080
},
{
"epoch": 2.82,
"learning_rate": 5.4791116941328194e-05,
"loss": 0.7469,
"step": 2090
},
{
"epoch": 2.83,
"learning_rate": 5.4439794766146746e-05,
"loss": 0.8403,
"step": 2100
},
{
"epoch": 2.84,
"learning_rate": 5.408825151042014e-05,
"loss": 0.7961,
"step": 2110
},
{
"epoch": 2.86,
"learning_rate": 5.373650467932122e-05,
"loss": 0.9164,
"step": 2120
},
{
"epoch": 2.87,
"learning_rate": 5.3384571788159845e-05,
"loss": 0.8157,
"step": 2130
},
{
"epoch": 2.88,
"learning_rate": 5.3032470361510825e-05,
"loss": 0.7424,
"step": 2140
},
{
"epoch": 2.9,
"learning_rate": 5.2680217932341183e-05,
"loss": 0.5375,
"step": 2150
},
{
"epoch": 2.91,
"learning_rate": 5.232783204113719e-05,
"loss": 0.4721,
"step": 2160
},
{
"epoch": 2.92,
"learning_rate": 5.197533023503089e-05,
"loss": 0.8567,
"step": 2170
},
{
"epoch": 2.94,
"learning_rate": 5.1622730066926294e-05,
"loss": 0.8027,
"step": 2180
},
{
"epoch": 2.95,
"learning_rate": 5.127004909462541e-05,
"loss": 0.7776,
"step": 2190
},
{
"epoch": 2.96,
"learning_rate": 5.0917304879953864e-05,
"loss": 0.8841,
"step": 2200
},
{
"epoch": 2.98,
"learning_rate": 5.05645149878865e-05,
"loss": 0.8787,
"step": 2210
},
{
"epoch": 2.99,
"learning_rate": 5.021169698567264e-05,
"loss": 0.6899,
"step": 2220
},
{
"epoch": 3.01,
"learning_rate": 4.985886844196138e-05,
"loss": 0.7251,
"step": 2230
},
{
"epoch": 3.02,
"learning_rate": 4.950604692592672e-05,
"loss": 0.793,
"step": 2240
},
{
"epoch": 3.03,
"learning_rate": 4.915325000639271e-05,
"loss": 0.6657,
"step": 2250
},
{
"epoch": 3.05,
"learning_rate": 4.880049525095865e-05,
"loss": 0.6907,
"step": 2260
},
{
"epoch": 3.06,
"learning_rate": 4.8447800225124214e-05,
"loss": 0.689,
"step": 2270
},
{
"epoch": 3.07,
"learning_rate": 4.8095182491414835e-05,
"loss": 0.5594,
"step": 2280
},
{
"epoch": 3.09,
"learning_rate": 4.774265960850724e-05,
"loss": 0.696,
"step": 2290
},
{
"epoch": 3.1,
"learning_rate": 4.7390249130354935e-05,
"loss": 0.8162,
"step": 2300
},
{
"epoch": 3.11,
"learning_rate": 4.703796860531429e-05,
"loss": 0.8038,
"step": 2310
},
{
"epoch": 3.13,
"learning_rate": 4.66858355752706e-05,
"loss": 0.6338,
"step": 2320
},
{
"epoch": 3.14,
"learning_rate": 4.6333867574764584e-05,
"loss": 0.8687,
"step": 2330
},
{
"epoch": 3.15,
"learning_rate": 4.5982082130119343e-05,
"loss": 0.7116,
"step": 2340
},
{
"epoch": 3.17,
"learning_rate": 4.563049675856749e-05,
"loss": 0.7132,
"step": 2350
},
{
"epoch": 3.18,
"learning_rate": 4.5279128967379065e-05,
"loss": 0.4794,
"step": 2360
},
{
"epoch": 3.19,
"learning_rate": 4.492799625298953e-05,
"loss": 0.8669,
"step": 2370
},
{
"epoch": 3.21,
"learning_rate": 4.4577116100128735e-05,
"loss": 0.8037,
"step": 2380
},
{
"epoch": 3.22,
"learning_rate": 4.422650598095011e-05,
"loss": 0.5912,
"step": 2390
},
{
"epoch": 3.23,
"learning_rate": 4.387618335416073e-05,
"loss": 0.7927,
"step": 2400
},
{
"epoch": 3.25,
"learning_rate": 4.352616566415192e-05,
"loss": 0.793,
"step": 2410
},
{
"epoch": 3.26,
"learning_rate": 4.317647034013064e-05,
"loss": 0.6888,
"step": 2420
},
{
"epoch": 3.27,
"learning_rate": 4.282711479525152e-05,
"loss": 0.4784,
"step": 2430
},
{
"epoch": 3.29,
"learning_rate": 4.247811642574981e-05,
"loss": 0.5894,
"step": 2440
},
{
"epoch": 3.3,
"learning_rate": 4.212949261007519e-05,
"loss": 0.7506,
"step": 2450
},
{
"epoch": 3.32,
"learning_rate": 4.1781260708026306e-05,
"loss": 0.9467,
"step": 2460
},
{
"epoch": 3.33,
"learning_rate": 4.143343805988637e-05,
"loss": 0.609,
"step": 2470
},
{
"epoch": 3.34,
"learning_rate": 4.1086041985559736e-05,
"loss": 0.6157,
"step": 2480
},
{
"epoch": 3.36,
"learning_rate": 4.073908978370944e-05,
"loss": 0.7386,
"step": 2490
},
{
"epoch": 3.37,
"learning_rate": 4.0392598730895756e-05,
"loss": 0.698,
"step": 2500
},
{
"epoch": 3.38,
"learning_rate": 4.004658608071593e-05,
"loss": 0.5692,
"step": 2510
},
{
"epoch": 3.4,
"learning_rate": 3.970106906294509e-05,
"loss": 0.7693,
"step": 2520
},
{
"epoch": 3.41,
"learning_rate": 3.9356064882678203e-05,
"loss": 0.8045,
"step": 2530
},
{
"epoch": 3.42,
"learning_rate": 3.901159071947339e-05,
"loss": 0.7021,
"step": 2540
},
{
"epoch": 3.44,
"learning_rate": 3.8667663726496454e-05,
"loss": 0.7647,
"step": 2550
},
{
"epoch": 3.45,
"learning_rate": 3.832430102966678e-05,
"loss": 0.6575,
"step": 2560
},
{
"epoch": 3.46,
"learning_rate": 3.7981519726804434e-05,
"loss": 0.6814,
"step": 2570
},
{
"epoch": 3.48,
"learning_rate": 3.76393368867789e-05,
"loss": 0.743,
"step": 2580
},
{
"epoch": 3.49,
"learning_rate": 3.729776954865905e-05,
"loss": 0.6576,
"step": 2590
},
{
"epoch": 3.5,
"learning_rate": 3.6956834720864704e-05,
"loss": 0.7939,
"step": 2600
},
{
"epoch": 3.52,
"learning_rate": 3.661654938031968e-05,
"loss": 0.7209,
"step": 2610
},
{
"epoch": 3.53,
"learning_rate": 3.6276930471606465e-05,
"loss": 0.5593,
"step": 2620
},
{
"epoch": 3.54,
"learning_rate": 3.5937994906122444e-05,
"loss": 0.6488,
"step": 2630
},
{
"epoch": 3.56,
"learning_rate": 3.559975956123772e-05,
"loss": 0.6735,
"step": 2640
},
{
"epoch": 3.57,
"learning_rate": 3.5262241279454785e-05,
"loss": 0.5263,
"step": 2650
},
{
"epoch": 3.58,
"learning_rate": 3.492545686756986e-05,
"loss": 0.6226,
"step": 2660
},
{
"epoch": 3.6,
"learning_rate": 3.4589423095835883e-05,
"loss": 0.5881,
"step": 2670
},
{
"epoch": 3.61,
"learning_rate": 3.425415669712754e-05,
"loss": 0.6564,
"step": 2680
},
{
"epoch": 3.63,
"learning_rate": 3.3919674366108e-05,
"loss": 0.439,
"step": 2690
},
{
"epoch": 3.64,
"learning_rate": 3.3585992758397656e-05,
"loss": 0.5981,
"step": 2700
},
{
"epoch": 3.65,
"learning_rate": 3.325312848974466e-05,
"loss": 0.8037,
"step": 2710
},
{
"epoch": 3.67,
"learning_rate": 3.2921098135197565e-05,
"loss": 0.5704,
"step": 2720
},
{
"epoch": 3.68,
"learning_rate": 3.258991822828007e-05,
"loss": 0.7255,
"step": 2730
},
{
"epoch": 3.69,
"learning_rate": 3.225960526016758e-05,
"loss": 0.597,
"step": 2740
},
{
"epoch": 3.71,
"learning_rate": 3.193017567886611e-05,
"loss": 0.7397,
"step": 2750
},
{
"epoch": 3.72,
"learning_rate": 3.160164588839328e-05,
"loss": 0.8098,
"step": 2760
},
{
"epoch": 3.73,
"learning_rate": 3.127403224796136e-05,
"loss": 0.7542,
"step": 2770
},
{
"epoch": 3.75,
"learning_rate": 3.09473510711628e-05,
"loss": 0.5898,
"step": 2780
},
{
"epoch": 3.76,
"learning_rate": 3.062161862515775e-05,
"loss": 0.7931,
"step": 2790
},
{
"epoch": 3.77,
"learning_rate": 3.0296851129864168e-05,
"loss": 0.7352,
"step": 2800
},
{
"epoch": 3.79,
"learning_rate": 2.997306475715005e-05,
"loss": 0.7473,
"step": 2810
},
{
"epoch": 3.8,
"learning_rate": 2.9650275630028146e-05,
"loss": 0.7002,
"step": 2820
},
{
"epoch": 3.81,
"learning_rate": 2.932849982185324e-05,
"loss": 0.8059,
"step": 2830
},
{
"epoch": 3.83,
"learning_rate": 2.9007753355521593e-05,
"loss": 0.8236,
"step": 2840
},
{
"epoch": 3.84,
"learning_rate": 2.8688052202673188e-05,
"loss": 0.4689,
"step": 2850
},
{
"epoch": 3.85,
"learning_rate": 2.8369412282896413e-05,
"loss": 0.5106,
"step": 2860
},
{
"epoch": 3.87,
"learning_rate": 2.805184946293532e-05,
"loss": 0.6069,
"step": 2870
},
{
"epoch": 3.88,
"learning_rate": 2.773537955589951e-05,
"loss": 0.6317,
"step": 2880
},
{
"epoch": 3.89,
"learning_rate": 2.742001832047677e-05,
"loss": 0.5961,
"step": 2890
},
{
"epoch": 3.91,
"learning_rate": 2.71057814601484e-05,
"loss": 0.777,
"step": 2900
},
{
"epoch": 3.92,
"learning_rate": 2.6792684622407082e-05,
"loss": 0.8788,
"step": 2910
},
{
"epoch": 3.94,
"learning_rate": 2.6480743397977915e-05,
"loss": 0.5528,
"step": 2920
},
{
"epoch": 3.95,
"learning_rate": 2.6169973320041968e-05,
"loss": 0.3569,
"step": 2930
},
{
"epoch": 3.96,
"learning_rate": 2.5860389863462765e-05,
"loss": 0.6363,
"step": 2940
},
{
"epoch": 3.98,
"learning_rate": 2.5552008444015817e-05,
"loss": 0.8244,
"step": 2950
},
{
"epoch": 3.99,
"learning_rate": 2.524484441762091e-05,
"loss": 0.8213,
"step": 2960
},
{
"epoch": 4.0,
"learning_rate": 2.4938913079577497e-05,
"loss": 0.7101,
"step": 2970
},
{
"epoch": 4.02,
"learning_rate": 2.4634229663803042e-05,
"loss": 0.6648,
"step": 2980
},
{
"epoch": 4.03,
"learning_rate": 2.433080934207441e-05,
"loss": 0.542,
"step": 2990
},
{
"epoch": 4.04,
"learning_rate": 2.4028667223272476e-05,
"loss": 0.7995,
"step": 3000
},
{
"epoch": 4.06,
"learning_rate": 2.3727818352629712e-05,
"loss": 0.7278,
"step": 3010
},
{
"epoch": 4.07,
"learning_rate": 2.3428277710980968e-05,
"loss": 0.5942,
"step": 3020
},
{
"epoch": 4.08,
"learning_rate": 2.3130060214017607e-05,
"loss": 0.8319,
"step": 3030
},
{
"epoch": 4.1,
"learning_rate": 2.2833180711544693e-05,
"loss": 0.7817,
"step": 3040
},
{
"epoch": 4.11,
"learning_rate": 2.2537653986741563e-05,
"loss": 0.7551,
"step": 3050
},
{
"epoch": 4.12,
"learning_rate": 2.2243494755425716e-05,
"loss": 0.7107,
"step": 3060
},
{
"epoch": 4.14,
"learning_rate": 2.1950717665319955e-05,
"loss": 0.8799,
"step": 3070
},
{
"epoch": 4.15,
"learning_rate": 2.1659337295323118e-05,
"loss": 0.6452,
"step": 3080
},
{
"epoch": 4.16,
"learning_rate": 2.1369368154784055e-05,
"loss": 0.7799,
"step": 3090
},
{
"epoch": 4.18,
"learning_rate": 2.1080824682779122e-05,
"loss": 0.5146,
"step": 3100
},
{
"epoch": 4.19,
"learning_rate": 2.0793721247393206e-05,
"loss": 0.5216,
"step": 3110
},
{
"epoch": 4.2,
"learning_rate": 2.0508072145004283e-05,
"loss": 0.6694,
"step": 3120
},
{
"epoch": 4.22,
"learning_rate": 2.0223891599571438e-05,
"loss": 0.7736,
"step": 3130
},
{
"epoch": 4.23,
"learning_rate": 1.99411937619267e-05,
"loss": 0.5249,
"step": 3140
},
{
"epoch": 4.25,
"learning_rate": 1.9659992709070345e-05,
"loss": 0.4881,
"step": 3150
},
{
"epoch": 4.26,
"learning_rate": 1.9380302443469856e-05,
"loss": 0.5865,
"step": 3160
},
{
"epoch": 4.27,
"learning_rate": 1.9102136892362844e-05,
"loss": 0.5865,
"step": 3170
},
{
"epoch": 4.29,
"learning_rate": 1.8825509907063327e-05,
"loss": 0.7465,
"step": 3180
},
{
"epoch": 4.3,
"learning_rate": 1.8550435262272165e-05,
"loss": 0.6462,
"step": 3190
},
{
"epoch": 4.31,
"learning_rate": 1.8276926655391085e-05,
"loss": 0.7105,
"step": 3200
},
{
"epoch": 4.33,
"learning_rate": 1.8004997705840564e-05,
"loss": 0.6498,
"step": 3210
},
{
"epoch": 4.34,
"learning_rate": 1.7734661954381754e-05,
"loss": 0.6075,
"step": 3220
},
{
"epoch": 4.35,
"learning_rate": 1.7465932862442136e-05,
"loss": 0.5235,
"step": 3230
},
{
"epoch": 4.37,
"learning_rate": 1.7198823811445253e-05,
"loss": 0.5452,
"step": 3240
},
{
"epoch": 4.38,
"learning_rate": 1.6933348102144324e-05,
"loss": 0.5263,
"step": 3250
},
{
"epoch": 4.39,
"learning_rate": 1.6669518953960008e-05,
"loss": 0.5772,
"step": 3260
},
{
"epoch": 4.41,
"learning_rate": 1.6407349504322033e-05,
"loss": 0.5608,
"step": 3270
},
{
"epoch": 4.42,
"learning_rate": 1.6146852808015135e-05,
"loss": 0.5716,
"step": 3280
},
{
"epoch": 4.43,
"learning_rate": 1.5888041836528915e-05,
"loss": 0.605,
"step": 3290
},
{
"epoch": 4.45,
"learning_rate": 1.5630929477411905e-05,
"loss": 0.5017,
"step": 3300
},
{
"epoch": 4.46,
"learning_rate": 1.5375528533629946e-05,
"loss": 0.5166,
"step": 3310
},
{
"epoch": 4.47,
"learning_rate": 1.5121851722928493e-05,
"loss": 0.4659,
"step": 3320
},
{
"epoch": 4.49,
"learning_rate": 1.4869911677199477e-05,
"loss": 0.6738,
"step": 3330
},
{
"epoch": 4.5,
"learning_rate": 1.4619720941852227e-05,
"loss": 0.5109,
"step": 3340
},
{
"epoch": 4.51,
"learning_rate": 1.4371291975188756e-05,
"loss": 0.58,
"step": 3350
},
{
"epoch": 4.53,
"learning_rate": 1.4124637147783437e-05,
"loss": 0.8329,
"step": 3360
},
{
"epoch": 4.54,
"learning_rate": 1.3879768741867039e-05,
"loss": 0.7531,
"step": 3370
},
{
"epoch": 4.56,
"learning_rate": 1.363669895071501e-05,
"loss": 0.6354,
"step": 3380
},
{
"epoch": 4.57,
"learning_rate": 1.3395439878040422e-05,
"loss": 0.5287,
"step": 3390
},
{
"epoch": 4.58,
"learning_rate": 1.3156003537391248e-05,
"loss": 0.6694,
"step": 3400
},
{
"epoch": 4.6,
"learning_rate": 1.291840185155207e-05,
"loss": 0.5892,
"step": 3410
},
{
"epoch": 4.61,
"learning_rate": 1.2682646651950475e-05,
"loss": 0.6901,
"step": 3420
},
{
"epoch": 4.62,
"learning_rate": 1.2448749678067856e-05,
"loss": 0.7459,
"step": 3430
},
{
"epoch": 4.64,
"learning_rate": 1.2216722576854861e-05,
"loss": 0.9782,
"step": 3440
},
{
"epoch": 4.65,
"learning_rate": 1.198657690215142e-05,
"loss": 0.6325,
"step": 3450
},
{
"epoch": 4.66,
"learning_rate": 1.175832411411139e-05,
"loss": 0.6926,
"step": 3460
},
{
"epoch": 4.68,
"learning_rate": 1.153197557863196e-05,
"loss": 0.5566,
"step": 3470
},
{
"epoch": 4.69,
"learning_rate": 1.1307542566787666e-05,
"loss": 0.6412,
"step": 3480
},
{
"epoch": 4.7,
"learning_rate": 1.108503625426906e-05,
"loss": 0.5914,
"step": 3490
},
{
"epoch": 4.72,
"learning_rate": 1.0864467720826343e-05,
"loss": 0.5399,
"step": 3500
},
{
"epoch": 4.73,
"learning_rate": 1.0645847949717568e-05,
"loss": 0.4704,
"step": 3510
},
{
"epoch": 4.74,
"learning_rate": 1.042918782716174e-05,
"loss": 0.4348,
"step": 3520
},
{
"epoch": 4.76,
"learning_rate": 1.0214498141796752e-05,
"loss": 0.6267,
"step": 3530
},
{
"epoch": 4.77,
"learning_rate": 1.000178958414214e-05,
"loss": 0.7144,
"step": 3540
},
{
"epoch": 4.78,
"learning_rate": 9.791072746066732e-06,
"loss": 0.8698,
"step": 3550
},
{
"epoch": 4.8,
"learning_rate": 9.58235812026128e-06,
"loss": 0.5476,
"step": 3560
},
{
"epoch": 4.81,
"learning_rate": 9.375656099715934e-06,
"loss": 0.7166,
"step": 3570
},
{
"epoch": 4.82,
"learning_rate": 9.170976977202716e-06,
"loss": 0.4887,
"step": 3580
},
{
"epoch": 4.84,
"learning_rate": 8.968330944763032e-06,
"loss": 0.5963,
"step": 3590
},
{
"epoch": 4.85,
"learning_rate": 8.76772809320006e-06,
"loss": 0.5586,
"step": 3600
},
{
"epoch": 4.87,
"learning_rate": 8.569178411576412e-06,
"loss": 0.6225,
"step": 3610
},
{
"epoch": 4.88,
"learning_rate": 8.37269178671663e-06,
"loss": 0.5414,
"step": 3620
},
{
"epoch": 4.89,
"learning_rate": 8.178278002714879e-06,
"loss": 0.4701,
"step": 3630
},
{
"epoch": 4.91,
"learning_rate": 7.985946740447791e-06,
"loss": 0.8415,
"step": 3640
},
{
"epoch": 4.92,
"learning_rate": 7.79570757709236e-06,
"loss": 0.8848,
"step": 3650
},
{
"epoch": 4.93,
"learning_rate": 7.607569985649066e-06,
"loss": 0.6962,
"step": 3660
},
{
"epoch": 4.95,
"learning_rate": 7.421543334470166e-06,
"loss": 0.5883,
"step": 3670
},
{
"epoch": 4.96,
"learning_rate": 7.237636886793158e-06,
"loss": 0.3532,
"step": 3680
},
{
"epoch": 4.97,
"learning_rate": 7.05585980027958e-06,
"loss": 0.6651,
"step": 3690
},
{
"epoch": 4.99,
"learning_rate": 6.876221126558951e-06,
"loss": 0.5999,
"step": 3700
},
{
"epoch": 5.0,
"learning_rate": 6.698729810778065e-06,
"loss": 0.5348,
"step": 3710
}
],
"logging_steps": 10,
"max_steps": 4452,
"num_train_epochs": 6,
"save_steps": 500,
"total_flos": 1.93884008030208e+16,
"trial_name": null,
"trial_params": null
}