WSI-MLLM / s3 /trainer_state.json
yili7eli's picture
upload models (3 splits and 2 languages)
c00cf34 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 17.857142857142858,
"global_step": 8000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.02,
"learning_rate": 3.7174721189591085e-07,
"loss": 7.2406,
"step": 10
},
{
"epoch": 0.04,
"learning_rate": 7.434944237918217e-07,
"loss": 6.5969,
"step": 20
},
{
"epoch": 0.07,
"learning_rate": 1.1152416356877324e-06,
"loss": 5.4641,
"step": 30
},
{
"epoch": 0.09,
"learning_rate": 1.4869888475836434e-06,
"loss": 4.0734,
"step": 40
},
{
"epoch": 0.11,
"learning_rate": 1.858736059479554e-06,
"loss": 3.9453,
"step": 50
},
{
"epoch": 0.13,
"learning_rate": 2.2304832713754648e-06,
"loss": 3.4391,
"step": 60
},
{
"epoch": 0.16,
"learning_rate": 2.6022304832713758e-06,
"loss": 3.1609,
"step": 70
},
{
"epoch": 0.18,
"learning_rate": 2.973977695167287e-06,
"loss": 2.9453,
"step": 80
},
{
"epoch": 0.2,
"learning_rate": 3.3457249070631974e-06,
"loss": 2.8047,
"step": 90
},
{
"epoch": 0.22,
"learning_rate": 3.717472118959108e-06,
"loss": 2.2758,
"step": 100
},
{
"epoch": 0.25,
"learning_rate": 4.089219330855019e-06,
"loss": 2.0383,
"step": 110
},
{
"epoch": 0.27,
"learning_rate": 4.4609665427509296e-06,
"loss": 1.7414,
"step": 120
},
{
"epoch": 0.29,
"learning_rate": 4.83271375464684e-06,
"loss": 1.5773,
"step": 130
},
{
"epoch": 0.31,
"learning_rate": 5.2044609665427516e-06,
"loss": 1.3703,
"step": 140
},
{
"epoch": 0.33,
"learning_rate": 5.576208178438662e-06,
"loss": 1.2703,
"step": 150
},
{
"epoch": 0.36,
"learning_rate": 5.947955390334574e-06,
"loss": 1.1812,
"step": 160
},
{
"epoch": 0.38,
"learning_rate": 6.319702602230484e-06,
"loss": 1.0328,
"step": 170
},
{
"epoch": 0.4,
"learning_rate": 6.691449814126395e-06,
"loss": 0.943,
"step": 180
},
{
"epoch": 0.42,
"learning_rate": 7.063197026022306e-06,
"loss": 1.1418,
"step": 190
},
{
"epoch": 0.45,
"learning_rate": 7.434944237918216e-06,
"loss": 0.784,
"step": 200
},
{
"epoch": 0.47,
"learning_rate": 7.806691449814127e-06,
"loss": 0.6648,
"step": 210
},
{
"epoch": 0.49,
"learning_rate": 8.178438661710038e-06,
"loss": 0.7697,
"step": 220
},
{
"epoch": 0.51,
"learning_rate": 8.550185873605949e-06,
"loss": 0.6514,
"step": 230
},
{
"epoch": 0.54,
"learning_rate": 8.921933085501859e-06,
"loss": 0.6775,
"step": 240
},
{
"epoch": 0.56,
"learning_rate": 9.29368029739777e-06,
"loss": 0.5932,
"step": 250
},
{
"epoch": 0.58,
"learning_rate": 9.66542750929368e-06,
"loss": 0.6855,
"step": 260
},
{
"epoch": 0.6,
"learning_rate": 1.0037174721189591e-05,
"loss": 0.5135,
"step": 270
},
{
"epoch": 0.62,
"learning_rate": 1.0408921933085503e-05,
"loss": 0.5949,
"step": 280
},
{
"epoch": 0.65,
"learning_rate": 1.0780669144981412e-05,
"loss": 0.5139,
"step": 290
},
{
"epoch": 0.67,
"learning_rate": 1.1152416356877324e-05,
"loss": 0.4232,
"step": 300
},
{
"epoch": 0.69,
"learning_rate": 1.1524163568773235e-05,
"loss": 0.4325,
"step": 310
},
{
"epoch": 0.71,
"learning_rate": 1.1895910780669147e-05,
"loss": 0.4629,
"step": 320
},
{
"epoch": 0.74,
"learning_rate": 1.2267657992565056e-05,
"loss": 0.3893,
"step": 330
},
{
"epoch": 0.76,
"learning_rate": 1.2639405204460968e-05,
"loss": 0.5352,
"step": 340
},
{
"epoch": 0.78,
"learning_rate": 1.3011152416356879e-05,
"loss": 0.5012,
"step": 350
},
{
"epoch": 0.8,
"learning_rate": 1.338289962825279e-05,
"loss": 0.4872,
"step": 360
},
{
"epoch": 0.83,
"learning_rate": 1.37546468401487e-05,
"loss": 0.6109,
"step": 370
},
{
"epoch": 0.85,
"learning_rate": 1.4126394052044612e-05,
"loss": 0.4701,
"step": 380
},
{
"epoch": 0.87,
"learning_rate": 1.4498141263940521e-05,
"loss": 0.5756,
"step": 390
},
{
"epoch": 0.89,
"learning_rate": 1.4869888475836432e-05,
"loss": 0.4864,
"step": 400
},
{
"epoch": 0.92,
"learning_rate": 1.5241635687732344e-05,
"loss": 0.5752,
"step": 410
},
{
"epoch": 0.94,
"learning_rate": 1.5613382899628255e-05,
"loss": 0.5738,
"step": 420
},
{
"epoch": 0.96,
"learning_rate": 1.5985130111524165e-05,
"loss": 0.4296,
"step": 430
},
{
"epoch": 0.98,
"learning_rate": 1.6356877323420076e-05,
"loss": 0.4486,
"step": 440
},
{
"epoch": 1.0,
"learning_rate": 1.6728624535315986e-05,
"loss": 0.5186,
"step": 450
},
{
"epoch": 1.03,
"learning_rate": 1.7100371747211897e-05,
"loss": 0.3414,
"step": 460
},
{
"epoch": 1.05,
"learning_rate": 1.7472118959107808e-05,
"loss": 0.4656,
"step": 470
},
{
"epoch": 1.07,
"learning_rate": 1.7843866171003718e-05,
"loss": 0.361,
"step": 480
},
{
"epoch": 1.09,
"learning_rate": 1.8215613382899632e-05,
"loss": 0.4122,
"step": 490
},
{
"epoch": 1.12,
"learning_rate": 1.858736059479554e-05,
"loss": 0.331,
"step": 500
},
{
"epoch": 1.14,
"learning_rate": 1.8959107806691453e-05,
"loss": 0.4964,
"step": 510
},
{
"epoch": 1.16,
"learning_rate": 1.933085501858736e-05,
"loss": 0.3147,
"step": 520
},
{
"epoch": 1.18,
"learning_rate": 1.970260223048327e-05,
"loss": 0.3456,
"step": 530
},
{
"epoch": 1.21,
"learning_rate": 1.9999999346673545e-05,
"loss": 0.324,
"step": 540
},
{
"epoch": 1.23,
"learning_rate": 1.9999976480256544e-05,
"loss": 0.3336,
"step": 550
},
{
"epoch": 1.25,
"learning_rate": 1.9999920947602093e-05,
"loss": 0.3744,
"step": 560
},
{
"epoch": 1.27,
"learning_rate": 1.9999832748891602e-05,
"loss": 0.392,
"step": 570
},
{
"epoch": 1.29,
"learning_rate": 1.999971188441319e-05,
"loss": 0.4078,
"step": 580
},
{
"epoch": 1.32,
"learning_rate": 1.999955835456167e-05,
"loss": 0.4142,
"step": 590
},
{
"epoch": 1.34,
"learning_rate": 1.9999372159838563e-05,
"loss": 0.3998,
"step": 600
},
{
"epoch": 1.36,
"learning_rate": 1.9999153300852108e-05,
"loss": 0.3499,
"step": 610
},
{
"epoch": 1.38,
"learning_rate": 1.999890177831723e-05,
"loss": 0.3383,
"step": 620
},
{
"epoch": 1.41,
"learning_rate": 1.9998617593055565e-05,
"loss": 0.3296,
"step": 630
},
{
"epoch": 1.43,
"learning_rate": 1.9998300745995437e-05,
"loss": 0.3723,
"step": 640
},
{
"epoch": 1.45,
"learning_rate": 1.9997951238171875e-05,
"loss": 0.4653,
"step": 650
},
{
"epoch": 1.47,
"learning_rate": 1.9997569070726585e-05,
"loss": 0.3805,
"step": 660
},
{
"epoch": 1.5,
"learning_rate": 1.9997154244907972e-05,
"loss": 0.7628,
"step": 670
},
{
"epoch": 1.52,
"learning_rate": 1.999670676207112e-05,
"loss": 0.3567,
"step": 680
},
{
"epoch": 1.54,
"learning_rate": 1.999622662367778e-05,
"loss": 0.4469,
"step": 690
},
{
"epoch": 1.56,
"learning_rate": 1.99957138312964e-05,
"loss": 0.2795,
"step": 700
},
{
"epoch": 1.58,
"learning_rate": 1.999516838660208e-05,
"loss": 0.5107,
"step": 710
},
{
"epoch": 1.61,
"learning_rate": 1.9994590291376585e-05,
"loss": 0.4593,
"step": 720
},
{
"epoch": 1.63,
"learning_rate": 1.999397954750834e-05,
"loss": 0.4607,
"step": 730
},
{
"epoch": 1.65,
"learning_rate": 1.999333615699242e-05,
"loss": 0.4134,
"step": 740
},
{
"epoch": 1.67,
"learning_rate": 1.999266012193054e-05,
"loss": 0.4903,
"step": 750
},
{
"epoch": 1.7,
"learning_rate": 1.9991951444531067e-05,
"loss": 0.3674,
"step": 760
},
{
"epoch": 1.72,
"learning_rate": 1.999121012710898e-05,
"loss": 0.3483,
"step": 770
},
{
"epoch": 1.74,
"learning_rate": 1.9990436172085896e-05,
"loss": 0.3344,
"step": 780
},
{
"epoch": 1.76,
"learning_rate": 1.9989629581990038e-05,
"loss": 0.2979,
"step": 790
},
{
"epoch": 1.79,
"learning_rate": 1.9988790359456236e-05,
"loss": 0.3349,
"step": 800
},
{
"epoch": 1.81,
"learning_rate": 1.998791850722593e-05,
"loss": 0.3062,
"step": 810
},
{
"epoch": 1.83,
"learning_rate": 1.9987014028147124e-05,
"loss": 0.3093,
"step": 820
},
{
"epoch": 1.85,
"learning_rate": 1.9986076925174433e-05,
"loss": 0.3229,
"step": 830
},
{
"epoch": 1.88,
"learning_rate": 1.9985107201369024e-05,
"loss": 0.3677,
"step": 840
},
{
"epoch": 1.9,
"learning_rate": 1.998410485989862e-05,
"loss": 0.2956,
"step": 850
},
{
"epoch": 1.92,
"learning_rate": 1.9983069904037506e-05,
"loss": 0.4208,
"step": 860
},
{
"epoch": 1.94,
"learning_rate": 1.99820023371665e-05,
"loss": 0.2229,
"step": 870
},
{
"epoch": 1.96,
"learning_rate": 1.9980902162772952e-05,
"loss": 0.3343,
"step": 880
},
{
"epoch": 1.99,
"learning_rate": 1.9979769384450728e-05,
"loss": 0.4137,
"step": 890
},
{
"epoch": 2.01,
"learning_rate": 1.9978604005900187e-05,
"loss": 0.2521,
"step": 900
},
{
"epoch": 2.03,
"learning_rate": 1.9977406030928205e-05,
"loss": 0.3819,
"step": 910
},
{
"epoch": 2.05,
"learning_rate": 1.9976175463448113e-05,
"loss": 0.2854,
"step": 920
},
{
"epoch": 2.08,
"learning_rate": 1.997491230747973e-05,
"loss": 0.2011,
"step": 930
},
{
"epoch": 2.1,
"learning_rate": 1.9973616567149316e-05,
"loss": 0.2441,
"step": 940
},
{
"epoch": 2.12,
"learning_rate": 1.9972288246689576e-05,
"loss": 0.277,
"step": 950
},
{
"epoch": 2.14,
"learning_rate": 1.9970927350439647e-05,
"loss": 0.2837,
"step": 960
},
{
"epoch": 2.17,
"learning_rate": 1.9969533882845076e-05,
"loss": 0.1958,
"step": 970
},
{
"epoch": 2.19,
"learning_rate": 1.9968107848457804e-05,
"loss": 0.2656,
"step": 980
},
{
"epoch": 2.21,
"learning_rate": 1.9966649251936164e-05,
"loss": 0.227,
"step": 990
},
{
"epoch": 2.23,
"learning_rate": 1.996515809804485e-05,
"loss": 0.2097,
"step": 1000
},
{
"epoch": 2.25,
"learning_rate": 1.9963634391654912e-05,
"loss": 0.2186,
"step": 1010
},
{
"epoch": 2.28,
"learning_rate": 1.996207813774374e-05,
"loss": 0.2874,
"step": 1020
},
{
"epoch": 2.3,
"learning_rate": 1.9960489341395043e-05,
"loss": 0.1967,
"step": 1030
},
{
"epoch": 2.32,
"learning_rate": 1.9958868007798828e-05,
"loss": 0.263,
"step": 1040
},
{
"epoch": 2.34,
"learning_rate": 1.9957214142251392e-05,
"loss": 0.2814,
"step": 1050
},
{
"epoch": 2.37,
"learning_rate": 1.9955527750155315e-05,
"loss": 0.3031,
"step": 1060
},
{
"epoch": 2.39,
"learning_rate": 1.995380883701941e-05,
"loss": 0.273,
"step": 1070
},
{
"epoch": 2.41,
"learning_rate": 1.995205740845874e-05,
"loss": 0.3278,
"step": 1080
},
{
"epoch": 2.43,
"learning_rate": 1.9950273470194566e-05,
"loss": 0.2603,
"step": 1090
},
{
"epoch": 2.46,
"learning_rate": 1.9948457028054364e-05,
"loss": 0.2474,
"step": 1100
},
{
"epoch": 2.48,
"learning_rate": 1.994660808797178e-05,
"loss": 0.3549,
"step": 1110
},
{
"epoch": 2.5,
"learning_rate": 1.9944726655986618e-05,
"loss": 0.2238,
"step": 1120
},
{
"epoch": 2.52,
"learning_rate": 1.9942812738244827e-05,
"loss": 0.2438,
"step": 1130
},
{
"epoch": 2.54,
"learning_rate": 1.9940866340998464e-05,
"loss": 0.2068,
"step": 1140
},
{
"epoch": 2.57,
"learning_rate": 1.99388874706057e-05,
"loss": 0.2335,
"step": 1150
},
{
"epoch": 2.59,
"learning_rate": 1.9936876133530766e-05,
"loss": 0.2469,
"step": 1160
},
{
"epoch": 2.61,
"learning_rate": 1.9934832336343968e-05,
"loss": 0.3177,
"step": 1170
},
{
"epoch": 2.63,
"learning_rate": 1.993275608572163e-05,
"loss": 0.2442,
"step": 1180
},
{
"epoch": 2.66,
"learning_rate": 1.99306473884461e-05,
"loss": 0.2069,
"step": 1190
},
{
"epoch": 2.68,
"learning_rate": 1.992850625140572e-05,
"loss": 0.2246,
"step": 1200
},
{
"epoch": 2.7,
"learning_rate": 1.992633268159479e-05,
"loss": 0.2917,
"step": 1210
},
{
"epoch": 2.72,
"learning_rate": 1.992412668611356e-05,
"loss": 0.2328,
"step": 1220
},
{
"epoch": 2.75,
"learning_rate": 1.992188827216821e-05,
"loss": 0.2774,
"step": 1230
},
{
"epoch": 2.77,
"learning_rate": 1.9919617447070808e-05,
"loss": 0.3456,
"step": 1240
},
{
"epoch": 2.79,
"learning_rate": 1.991731421823931e-05,
"loss": 0.2944,
"step": 1250
},
{
"epoch": 2.81,
"learning_rate": 1.9914978593197507e-05,
"loss": 0.1965,
"step": 1260
},
{
"epoch": 2.83,
"learning_rate": 1.991261057957503e-05,
"loss": 0.2508,
"step": 1270
},
{
"epoch": 2.86,
"learning_rate": 1.9910210185107306e-05,
"loss": 0.2811,
"step": 1280
},
{
"epoch": 2.88,
"learning_rate": 1.9907777417635537e-05,
"loss": 0.3268,
"step": 1290
},
{
"epoch": 2.9,
"learning_rate": 1.9905312285106682e-05,
"loss": 0.2356,
"step": 1300
},
{
"epoch": 2.92,
"learning_rate": 1.9902814795573416e-05,
"loss": 0.2328,
"step": 1310
},
{
"epoch": 2.95,
"learning_rate": 1.9900284957194115e-05,
"loss": 0.2515,
"step": 1320
},
{
"epoch": 2.97,
"learning_rate": 1.9897722778232836e-05,
"loss": 0.4871,
"step": 1330
},
{
"epoch": 2.99,
"learning_rate": 1.9895128267059268e-05,
"loss": 0.2741,
"step": 1340
},
{
"epoch": 3.01,
"learning_rate": 1.9892501432148728e-05,
"loss": 0.3244,
"step": 1350
},
{
"epoch": 3.04,
"learning_rate": 1.988984228208211e-05,
"loss": 0.2452,
"step": 1360
},
{
"epoch": 3.06,
"learning_rate": 1.988715082554588e-05,
"loss": 0.1495,
"step": 1370
},
{
"epoch": 3.08,
"learning_rate": 1.988442707133204e-05,
"loss": 0.133,
"step": 1380
},
{
"epoch": 3.1,
"learning_rate": 1.988167102833808e-05,
"loss": 0.2249,
"step": 1390
},
{
"epoch": 3.12,
"learning_rate": 1.9878882705566988e-05,
"loss": 0.1724,
"step": 1400
},
{
"epoch": 3.15,
"learning_rate": 1.9876062112127176e-05,
"loss": 0.1692,
"step": 1410
},
{
"epoch": 3.17,
"learning_rate": 1.9873209257232494e-05,
"loss": 0.1605,
"step": 1420
},
{
"epoch": 3.19,
"learning_rate": 1.987032415020216e-05,
"loss": 0.2,
"step": 1430
},
{
"epoch": 3.21,
"learning_rate": 1.986740680046076e-05,
"loss": 0.2069,
"step": 1440
},
{
"epoch": 3.24,
"learning_rate": 1.9864457217538198e-05,
"loss": 0.194,
"step": 1450
},
{
"epoch": 3.26,
"learning_rate": 1.986147541106967e-05,
"loss": 0.2604,
"step": 1460
},
{
"epoch": 3.28,
"learning_rate": 1.9858461390795648e-05,
"loss": 0.1781,
"step": 1470
},
{
"epoch": 3.3,
"learning_rate": 1.985541516656182e-05,
"loss": 0.167,
"step": 1480
},
{
"epoch": 3.33,
"learning_rate": 1.985233674831908e-05,
"loss": 0.2232,
"step": 1490
},
{
"epoch": 3.35,
"learning_rate": 1.984922614612348e-05,
"loss": 0.1978,
"step": 1500
},
{
"epoch": 3.37,
"learning_rate": 1.9846083370136217e-05,
"loss": 0.2211,
"step": 1510
},
{
"epoch": 3.39,
"learning_rate": 1.984290843062358e-05,
"loss": 0.2264,
"step": 1520
},
{
"epoch": 3.42,
"learning_rate": 1.9839701337956922e-05,
"loss": 0.2231,
"step": 1530
},
{
"epoch": 3.44,
"learning_rate": 1.9836462102612643e-05,
"loss": 0.203,
"step": 1540
},
{
"epoch": 3.46,
"learning_rate": 1.9833190735172117e-05,
"loss": 0.1621,
"step": 1550
},
{
"epoch": 3.48,
"learning_rate": 1.982988724632171e-05,
"loss": 0.2412,
"step": 1560
},
{
"epoch": 3.5,
"learning_rate": 1.9826551646852696e-05,
"loss": 0.2393,
"step": 1570
},
{
"epoch": 3.53,
"learning_rate": 1.982318394766124e-05,
"loss": 0.2143,
"step": 1580
},
{
"epoch": 3.55,
"learning_rate": 1.9819784159748394e-05,
"loss": 0.197,
"step": 1590
},
{
"epoch": 3.57,
"learning_rate": 1.9816352294219995e-05,
"loss": 0.2317,
"step": 1600
},
{
"epoch": 3.59,
"learning_rate": 1.981288836228669e-05,
"loss": 0.1935,
"step": 1610
},
{
"epoch": 3.62,
"learning_rate": 1.9809392375263865e-05,
"loss": 0.2081,
"step": 1620
},
{
"epoch": 3.64,
"learning_rate": 1.9805864344571625e-05,
"loss": 0.1856,
"step": 1630
},
{
"epoch": 3.66,
"learning_rate": 1.980230428173474e-05,
"loss": 0.1731,
"step": 1640
},
{
"epoch": 3.68,
"learning_rate": 1.979871219838263e-05,
"loss": 0.201,
"step": 1650
},
{
"epoch": 3.71,
"learning_rate": 1.9795088106249306e-05,
"loss": 0.251,
"step": 1660
},
{
"epoch": 3.73,
"learning_rate": 1.9791432017173338e-05,
"loss": 0.2249,
"step": 1670
},
{
"epoch": 3.75,
"learning_rate": 1.978774394309782e-05,
"loss": 0.1694,
"step": 1680
},
{
"epoch": 3.77,
"learning_rate": 1.9784023896070336e-05,
"loss": 0.217,
"step": 1690
},
{
"epoch": 3.79,
"learning_rate": 1.9780271888242904e-05,
"loss": 0.1834,
"step": 1700
},
{
"epoch": 3.82,
"learning_rate": 1.9776487931871958e-05,
"loss": 0.1892,
"step": 1710
},
{
"epoch": 3.84,
"learning_rate": 1.9772672039318278e-05,
"loss": 0.1855,
"step": 1720
},
{
"epoch": 3.86,
"learning_rate": 1.976882422304699e-05,
"loss": 0.1846,
"step": 1730
},
{
"epoch": 3.88,
"learning_rate": 1.976494449562748e-05,
"loss": 0.1942,
"step": 1740
},
{
"epoch": 3.91,
"learning_rate": 1.9761032869733397e-05,
"loss": 0.1904,
"step": 1750
},
{
"epoch": 3.93,
"learning_rate": 1.9757089358142573e-05,
"loss": 0.2,
"step": 1760
},
{
"epoch": 3.95,
"learning_rate": 1.9753113973737016e-05,
"loss": 0.2034,
"step": 1770
},
{
"epoch": 3.97,
"learning_rate": 1.9749106729502833e-05,
"loss": 0.2563,
"step": 1780
},
{
"epoch": 4.0,
"learning_rate": 1.974506763853022e-05,
"loss": 0.2639,
"step": 1790
},
{
"epoch": 4.02,
"learning_rate": 1.9740996714013394e-05,
"loss": 0.17,
"step": 1800
},
{
"epoch": 4.04,
"learning_rate": 1.9736893969250567e-05,
"loss": 0.1453,
"step": 1810
},
{
"epoch": 4.06,
"learning_rate": 1.97327594176439e-05,
"loss": 0.1692,
"step": 1820
},
{
"epoch": 4.08,
"learning_rate": 1.972859307269944e-05,
"loss": 0.1397,
"step": 1830
},
{
"epoch": 4.11,
"learning_rate": 1.9724394948027102e-05,
"loss": 0.1819,
"step": 1840
},
{
"epoch": 4.13,
"learning_rate": 1.9720165057340616e-05,
"loss": 0.1902,
"step": 1850
},
{
"epoch": 4.15,
"learning_rate": 1.971590341445747e-05,
"loss": 0.1498,
"step": 1860
},
{
"epoch": 4.17,
"learning_rate": 1.9711610033298888e-05,
"loss": 0.1418,
"step": 1870
},
{
"epoch": 4.2,
"learning_rate": 1.9707284927889757e-05,
"loss": 0.1641,
"step": 1880
},
{
"epoch": 4.22,
"learning_rate": 1.970292811235861e-05,
"loss": 0.1604,
"step": 1890
},
{
"epoch": 4.24,
"learning_rate": 1.969853960093755e-05,
"loss": 0.1473,
"step": 1900
},
{
"epoch": 4.26,
"learning_rate": 1.969411940796223e-05,
"loss": 0.1341,
"step": 1910
},
{
"epoch": 4.29,
"learning_rate": 1.9689667547871788e-05,
"loss": 0.1327,
"step": 1920
},
{
"epoch": 4.31,
"learning_rate": 1.9685184035208814e-05,
"loss": 0.1508,
"step": 1930
},
{
"epoch": 4.33,
"learning_rate": 1.968066888461929e-05,
"loss": 0.175,
"step": 1940
},
{
"epoch": 4.35,
"learning_rate": 1.9676122110852547e-05,
"loss": 0.1844,
"step": 1950
},
{
"epoch": 4.38,
"learning_rate": 1.9671543728761226e-05,
"loss": 0.1875,
"step": 1960
},
{
"epoch": 4.4,
"learning_rate": 1.9666933753301203e-05,
"loss": 0.1373,
"step": 1970
},
{
"epoch": 4.42,
"learning_rate": 1.9662292199531575e-05,
"loss": 0.126,
"step": 1980
},
{
"epoch": 4.44,
"learning_rate": 1.9657619082614588e-05,
"loss": 0.1428,
"step": 1990
},
{
"epoch": 4.46,
"learning_rate": 1.965291441781559e-05,
"loss": 0.2099,
"step": 2000
},
{
"epoch": 4.49,
"learning_rate": 1.964817822050299e-05,
"loss": 0.1505,
"step": 2010
},
{
"epoch": 4.51,
"learning_rate": 1.9643410506148196e-05,
"loss": 0.1498,
"step": 2020
},
{
"epoch": 4.53,
"learning_rate": 1.9638611290325576e-05,
"loss": 0.158,
"step": 2030
},
{
"epoch": 4.55,
"learning_rate": 1.96337805887124e-05,
"loss": 0.13,
"step": 2040
},
{
"epoch": 4.58,
"learning_rate": 1.9628918417088785e-05,
"loss": 0.1162,
"step": 2050
},
{
"epoch": 4.6,
"learning_rate": 1.9624024791337657e-05,
"loss": 0.1587,
"step": 2060
},
{
"epoch": 4.62,
"learning_rate": 1.9619099727444686e-05,
"loss": 0.163,
"step": 2070
},
{
"epoch": 4.64,
"learning_rate": 1.961414324149824e-05,
"loss": 0.1551,
"step": 2080
},
{
"epoch": 4.67,
"learning_rate": 1.9609155349689338e-05,
"loss": 0.1859,
"step": 2090
},
{
"epoch": 4.69,
"learning_rate": 1.9604136068311577e-05,
"loss": 0.1807,
"step": 2100
},
{
"epoch": 4.71,
"learning_rate": 1.95990854137611e-05,
"loss": 0.148,
"step": 2110
},
{
"epoch": 4.73,
"learning_rate": 1.959400340253654e-05,
"loss": 0.1809,
"step": 2120
},
{
"epoch": 4.75,
"learning_rate": 1.9588890051238945e-05,
"loss": 0.1431,
"step": 2130
},
{
"epoch": 4.78,
"learning_rate": 1.9583745376571756e-05,
"loss": 0.1473,
"step": 2140
},
{
"epoch": 4.8,
"learning_rate": 1.9578569395340727e-05,
"loss": 0.1512,
"step": 2150
},
{
"epoch": 4.82,
"learning_rate": 1.9573362124453884e-05,
"loss": 0.1578,
"step": 2160
},
{
"epoch": 4.84,
"learning_rate": 1.9568123580921453e-05,
"loss": 0.1495,
"step": 2170
},
{
"epoch": 4.87,
"learning_rate": 1.9562853781855836e-05,
"loss": 0.167,
"step": 2180
},
{
"epoch": 4.89,
"learning_rate": 1.9557552744471517e-05,
"loss": 0.1947,
"step": 2190
},
{
"epoch": 4.91,
"learning_rate": 1.9552220486085033e-05,
"loss": 0.1611,
"step": 2200
},
{
"epoch": 4.93,
"learning_rate": 1.954685702411491e-05,
"loss": 0.1301,
"step": 2210
},
{
"epoch": 4.96,
"learning_rate": 1.9541462376081594e-05,
"loss": 0.1348,
"step": 2220
},
{
"epoch": 4.98,
"learning_rate": 1.953603655960742e-05,
"loss": 0.1641,
"step": 2230
},
{
"epoch": 5.0,
"learning_rate": 1.9530579592416522e-05,
"loss": 0.1603,
"step": 2240
},
{
"epoch": 5.02,
"learning_rate": 1.9525091492334813e-05,
"loss": 0.1247,
"step": 2250
},
{
"epoch": 5.04,
"learning_rate": 1.951957227728988e-05,
"loss": 0.1154,
"step": 2260
},
{
"epoch": 5.07,
"learning_rate": 1.9514021965310972e-05,
"loss": 0.1265,
"step": 2270
},
{
"epoch": 5.09,
"learning_rate": 1.950844057452891e-05,
"loss": 0.1025,
"step": 2280
},
{
"epoch": 5.11,
"learning_rate": 1.9502828123176042e-05,
"loss": 0.11,
"step": 2290
},
{
"epoch": 5.13,
"learning_rate": 1.9497184629586176e-05,
"loss": 0.1142,
"step": 2300
},
{
"epoch": 5.16,
"learning_rate": 1.949151011219453e-05,
"loss": 0.0908,
"step": 2310
},
{
"epoch": 5.18,
"learning_rate": 1.9485804589537655e-05,
"loss": 0.1159,
"step": 2320
},
{
"epoch": 5.2,
"learning_rate": 1.9480068080253393e-05,
"loss": 0.1335,
"step": 2330
},
{
"epoch": 5.22,
"learning_rate": 1.9474300603080805e-05,
"loss": 0.1093,
"step": 2340
},
{
"epoch": 5.25,
"learning_rate": 1.9468502176860117e-05,
"loss": 0.1255,
"step": 2350
},
{
"epoch": 5.27,
"learning_rate": 1.9462672820532643e-05,
"loss": 0.1081,
"step": 2360
},
{
"epoch": 5.29,
"learning_rate": 1.9456812553140744e-05,
"loss": 0.144,
"step": 2370
},
{
"epoch": 5.31,
"learning_rate": 1.945092139382776e-05,
"loss": 0.1115,
"step": 2380
},
{
"epoch": 5.33,
"learning_rate": 1.944499936183793e-05,
"loss": 0.1108,
"step": 2390
},
{
"epoch": 5.36,
"learning_rate": 1.9439046476516356e-05,
"loss": 0.1386,
"step": 2400
},
{
"epoch": 5.38,
"learning_rate": 1.9433062757308914e-05,
"loss": 0.1375,
"step": 2410
},
{
"epoch": 5.4,
"learning_rate": 1.9427048223762212e-05,
"loss": 0.1292,
"step": 2420
},
{
"epoch": 5.42,
"learning_rate": 1.9421002895523515e-05,
"loss": 0.1444,
"step": 2430
},
{
"epoch": 5.45,
"learning_rate": 1.941492679234068e-05,
"loss": 0.1225,
"step": 2440
},
{
"epoch": 5.47,
"learning_rate": 1.9408819934062098e-05,
"loss": 0.1526,
"step": 2450
},
{
"epoch": 5.49,
"learning_rate": 1.9402682340636625e-05,
"loss": 0.131,
"step": 2460
},
{
"epoch": 5.51,
"learning_rate": 1.9396514032113514e-05,
"loss": 0.114,
"step": 2470
},
{
"epoch": 5.54,
"learning_rate": 1.9390315028642355e-05,
"loss": 0.1298,
"step": 2480
},
{
"epoch": 5.56,
"learning_rate": 1.9384085350473016e-05,
"loss": 0.0971,
"step": 2490
},
{
"epoch": 5.58,
"learning_rate": 1.9377825017955548e-05,
"loss": 0.1352,
"step": 2500
},
{
"epoch": 5.6,
"learning_rate": 1.9371534051540158e-05,
"loss": 0.1494,
"step": 2510
},
{
"epoch": 5.62,
"learning_rate": 1.9365212471777113e-05,
"loss": 0.134,
"step": 2520
},
{
"epoch": 5.65,
"learning_rate": 1.935886029931668e-05,
"loss": 0.1332,
"step": 2530
},
{
"epoch": 5.67,
"learning_rate": 1.9352477554909067e-05,
"loss": 0.1328,
"step": 2540
},
{
"epoch": 5.69,
"learning_rate": 1.934606425940435e-05,
"loss": 0.1235,
"step": 2550
},
{
"epoch": 5.71,
"learning_rate": 1.93396204337524e-05,
"loss": 0.1609,
"step": 2560
},
{
"epoch": 5.74,
"learning_rate": 1.9333146099002826e-05,
"loss": 0.1408,
"step": 2570
},
{
"epoch": 5.76,
"learning_rate": 1.932664127630488e-05,
"loss": 0.1463,
"step": 2580
},
{
"epoch": 5.78,
"learning_rate": 1.9320105986907433e-05,
"loss": 0.1548,
"step": 2590
},
{
"epoch": 5.8,
"learning_rate": 1.931354025215886e-05,
"loss": 0.1282,
"step": 2600
},
{
"epoch": 5.83,
"learning_rate": 1.9306944093507e-05,
"loss": 0.1387,
"step": 2610
},
{
"epoch": 5.85,
"learning_rate": 1.930031753249907e-05,
"loss": 0.1259,
"step": 2620
},
{
"epoch": 5.87,
"learning_rate": 1.9293660590781603e-05,
"loss": 0.144,
"step": 2630
},
{
"epoch": 5.89,
"learning_rate": 1.928697329010037e-05,
"loss": 0.1063,
"step": 2640
},
{
"epoch": 5.92,
"learning_rate": 1.9280255652300326e-05,
"loss": 0.1188,
"step": 2650
},
{
"epoch": 5.94,
"learning_rate": 1.9273507699325513e-05,
"loss": 0.1581,
"step": 2660
},
{
"epoch": 5.96,
"learning_rate": 1.9266729453219008e-05,
"loss": 0.1357,
"step": 2670
},
{
"epoch": 5.98,
"learning_rate": 1.925992093612284e-05,
"loss": 0.1446,
"step": 2680
},
{
"epoch": 6.0,
"learning_rate": 1.925308217027792e-05,
"loss": 0.1305,
"step": 2690
},
{
"epoch": 6.03,
"learning_rate": 1.924621317802399e-05,
"loss": 0.1042,
"step": 2700
},
{
"epoch": 6.05,
"learning_rate": 1.9239313981799507e-05,
"loss": 0.0902,
"step": 2710
},
{
"epoch": 6.07,
"learning_rate": 1.92323846041416e-05,
"loss": 0.0978,
"step": 2720
},
{
"epoch": 6.09,
"learning_rate": 1.9225425067685995e-05,
"loss": 0.0895,
"step": 2730
},
{
"epoch": 6.12,
"learning_rate": 1.9218435395166933e-05,
"loss": 0.0901,
"step": 2740
},
{
"epoch": 6.14,
"learning_rate": 1.9211415609417097e-05,
"loss": 0.0958,
"step": 2750
},
{
"epoch": 6.16,
"learning_rate": 1.920436573336754e-05,
"loss": 0.0952,
"step": 2760
},
{
"epoch": 6.18,
"learning_rate": 1.919728579004761e-05,
"loss": 0.0866,
"step": 2770
},
{
"epoch": 6.21,
"learning_rate": 1.919017580258487e-05,
"loss": 0.1057,
"step": 2780
},
{
"epoch": 6.23,
"learning_rate": 1.918303579420503e-05,
"loss": 0.0941,
"step": 2790
},
{
"epoch": 6.25,
"learning_rate": 1.917586578823186e-05,
"loss": 0.1057,
"step": 2800
},
{
"epoch": 6.27,
"learning_rate": 1.916866580808714e-05,
"loss": 0.121,
"step": 2810
},
{
"epoch": 6.29,
"learning_rate": 1.9161435877290538e-05,
"loss": 0.1101,
"step": 2820
},
{
"epoch": 6.32,
"learning_rate": 1.915417601945958e-05,
"loss": 0.1027,
"step": 2830
},
{
"epoch": 6.34,
"learning_rate": 1.9146886258309548e-05,
"loss": 0.0803,
"step": 2840
},
{
"epoch": 6.36,
"learning_rate": 1.9139566617653395e-05,
"loss": 0.1232,
"step": 2850
},
{
"epoch": 6.38,
"learning_rate": 1.9132217121401698e-05,
"loss": 0.0986,
"step": 2860
},
{
"epoch": 6.41,
"learning_rate": 1.912483779356255e-05,
"loss": 0.1098,
"step": 2870
},
{
"epoch": 6.43,
"learning_rate": 1.9117428658241498e-05,
"loss": 0.1063,
"step": 2880
},
{
"epoch": 6.45,
"learning_rate": 1.9109989739641446e-05,
"loss": 0.0962,
"step": 2890
},
{
"epoch": 6.47,
"learning_rate": 1.9102521062062615e-05,
"loss": 0.0951,
"step": 2900
},
{
"epoch": 6.5,
"learning_rate": 1.909502264990241e-05,
"loss": 0.1091,
"step": 2910
},
{
"epoch": 6.52,
"learning_rate": 1.9087494527655383e-05,
"loss": 0.096,
"step": 2920
},
{
"epoch": 6.54,
"learning_rate": 1.9079936719913138e-05,
"loss": 0.1044,
"step": 2930
},
{
"epoch": 6.56,
"learning_rate": 1.9072349251364238e-05,
"loss": 0.0912,
"step": 2940
},
{
"epoch": 6.58,
"learning_rate": 1.906473214679416e-05,
"loss": 0.0998,
"step": 2950
},
{
"epoch": 6.61,
"learning_rate": 1.9057085431085163e-05,
"loss": 0.1122,
"step": 2960
},
{
"epoch": 6.63,
"learning_rate": 1.904940912921626e-05,
"loss": 0.098,
"step": 2970
},
{
"epoch": 6.65,
"learning_rate": 1.9041703266263095e-05,
"loss": 0.0913,
"step": 2980
},
{
"epoch": 6.67,
"learning_rate": 1.9033967867397883e-05,
"loss": 0.1545,
"step": 2990
},
{
"epoch": 6.7,
"learning_rate": 1.902620295788932e-05,
"loss": 0.0895,
"step": 3000
},
{
"epoch": 6.72,
"learning_rate": 1.9018408563102505e-05,
"loss": 0.1202,
"step": 3010
},
{
"epoch": 6.74,
"learning_rate": 1.901058470849885e-05,
"loss": 0.1023,
"step": 3020
},
{
"epoch": 6.76,
"learning_rate": 1.900273141963601e-05,
"loss": 0.0949,
"step": 3030
},
{
"epoch": 6.79,
"learning_rate": 1.899484872216778e-05,
"loss": 0.1018,
"step": 3040
},
{
"epoch": 6.81,
"learning_rate": 1.8986936641844025e-05,
"loss": 0.1039,
"step": 3050
},
{
"epoch": 6.83,
"learning_rate": 1.8978995204510605e-05,
"loss": 0.1055,
"step": 3060
},
{
"epoch": 6.85,
"learning_rate": 1.897102443610926e-05,
"loss": 0.1352,
"step": 3070
},
{
"epoch": 6.88,
"learning_rate": 1.8963024362677557e-05,
"loss": 0.1144,
"step": 3080
},
{
"epoch": 6.9,
"learning_rate": 1.895499501034878e-05,
"loss": 0.1253,
"step": 3090
},
{
"epoch": 6.92,
"learning_rate": 1.8946936405351877e-05,
"loss": 0.1012,
"step": 3100
},
{
"epoch": 6.94,
"learning_rate": 1.893884857401133e-05,
"loss": 0.125,
"step": 3110
},
{
"epoch": 6.96,
"learning_rate": 1.8930731542747108e-05,
"loss": 0.1364,
"step": 3120
},
{
"epoch": 6.99,
"learning_rate": 1.8922585338074556e-05,
"loss": 0.1229,
"step": 3130
},
{
"epoch": 7.01,
"learning_rate": 1.8914409986604327e-05,
"loss": 0.1061,
"step": 3140
},
{
"epoch": 7.03,
"learning_rate": 1.8906205515042272e-05,
"loss": 0.0676,
"step": 3150
},
{
"epoch": 7.05,
"learning_rate": 1.8897971950189385e-05,
"loss": 0.0866,
"step": 3160
},
{
"epoch": 7.08,
"learning_rate": 1.888970931894169e-05,
"loss": 0.073,
"step": 3170
},
{
"epoch": 7.1,
"learning_rate": 1.888141764829015e-05,
"loss": 0.0978,
"step": 3180
},
{
"epoch": 7.12,
"learning_rate": 1.8873096965320597e-05,
"loss": 0.0781,
"step": 3190
},
{
"epoch": 7.14,
"learning_rate": 1.886474729721364e-05,
"loss": 0.0948,
"step": 3200
},
{
"epoch": 7.17,
"learning_rate": 1.8856368671244565e-05,
"loss": 0.0655,
"step": 3210
},
{
"epoch": 7.19,
"learning_rate": 1.8847961114783254e-05,
"loss": 0.0934,
"step": 3220
},
{
"epoch": 7.21,
"learning_rate": 1.88395246552941e-05,
"loss": 0.1012,
"step": 3230
},
{
"epoch": 7.23,
"learning_rate": 1.8831059320335902e-05,
"loss": 0.0872,
"step": 3240
},
{
"epoch": 7.25,
"learning_rate": 1.882256513756179e-05,
"loss": 0.0904,
"step": 3250
},
{
"epoch": 7.28,
"learning_rate": 1.881404213471913e-05,
"loss": 0.093,
"step": 3260
},
{
"epoch": 7.3,
"learning_rate": 1.8805490339649428e-05,
"loss": 0.0921,
"step": 3270
},
{
"epoch": 7.32,
"learning_rate": 1.879690978028825e-05,
"loss": 0.0907,
"step": 3280
},
{
"epoch": 7.34,
"learning_rate": 1.8788300484665118e-05,
"loss": 0.0783,
"step": 3290
},
{
"epoch": 7.37,
"learning_rate": 1.877966248090343e-05,
"loss": 0.0941,
"step": 3300
},
{
"epoch": 7.39,
"learning_rate": 1.8770995797220356e-05,
"loss": 0.1035,
"step": 3310
},
{
"epoch": 7.41,
"learning_rate": 1.8762300461926766e-05,
"loss": 0.0814,
"step": 3320
},
{
"epoch": 7.43,
"learning_rate": 1.8753576503427107e-05,
"loss": 0.0937,
"step": 3330
},
{
"epoch": 7.46,
"learning_rate": 1.874482395021934e-05,
"loss": 0.0916,
"step": 3340
},
{
"epoch": 7.48,
"learning_rate": 1.8736042830894828e-05,
"loss": 0.0981,
"step": 3350
},
{
"epoch": 7.5,
"learning_rate": 1.8727233174138254e-05,
"loss": 0.0897,
"step": 3360
},
{
"epoch": 7.52,
"learning_rate": 1.871839500872752e-05,
"loss": 0.0911,
"step": 3370
},
{
"epoch": 7.54,
"learning_rate": 1.8709528363533653e-05,
"loss": 0.1,
"step": 3380
},
{
"epoch": 7.57,
"learning_rate": 1.8700633267520715e-05,
"loss": 0.1036,
"step": 3390
},
{
"epoch": 7.59,
"learning_rate": 1.8691709749745705e-05,
"loss": 0.1104,
"step": 3400
},
{
"epoch": 7.61,
"learning_rate": 1.8682757839358472e-05,
"loss": 0.107,
"step": 3410
},
{
"epoch": 7.63,
"learning_rate": 1.86737775656016e-05,
"loss": 0.1009,
"step": 3420
},
{
"epoch": 7.66,
"learning_rate": 1.866476895781034e-05,
"loss": 0.0983,
"step": 3430
},
{
"epoch": 7.68,
"learning_rate": 1.8655732045412488e-05,
"loss": 0.0831,
"step": 3440
},
{
"epoch": 7.7,
"learning_rate": 1.8646666857928314e-05,
"loss": 0.0984,
"step": 3450
},
{
"epoch": 7.72,
"learning_rate": 1.8637573424970435e-05,
"loss": 0.0921,
"step": 3460
},
{
"epoch": 7.75,
"learning_rate": 1.862845177624375e-05,
"loss": 0.1029,
"step": 3470
},
{
"epoch": 7.77,
"learning_rate": 1.8619301941545323e-05,
"loss": 0.1106,
"step": 3480
},
{
"epoch": 7.79,
"learning_rate": 1.8610123950764288e-05,
"loss": 0.13,
"step": 3490
},
{
"epoch": 7.81,
"learning_rate": 1.8600917833881765e-05,
"loss": 0.1031,
"step": 3500
},
{
"epoch": 7.83,
"learning_rate": 1.8591683620970737e-05,
"loss": 0.0851,
"step": 3510
},
{
"epoch": 7.86,
"learning_rate": 1.858242134219598e-05,
"loss": 0.0879,
"step": 3520
},
{
"epoch": 7.88,
"learning_rate": 1.8573131027813945e-05,
"loss": 0.0817,
"step": 3530
},
{
"epoch": 7.9,
"learning_rate": 1.856381270817266e-05,
"loss": 0.1026,
"step": 3540
},
{
"epoch": 7.92,
"learning_rate": 1.8554466413711644e-05,
"loss": 0.0979,
"step": 3550
},
{
"epoch": 7.95,
"learning_rate": 1.8545092174961795e-05,
"loss": 0.0904,
"step": 3560
},
{
"epoch": 7.97,
"learning_rate": 1.85356900225453e-05,
"loss": 0.0887,
"step": 3570
},
{
"epoch": 7.99,
"learning_rate": 1.852625998717552e-05,
"loss": 0.0993,
"step": 3580
},
{
"epoch": 8.01,
"learning_rate": 1.8516802099656907e-05,
"loss": 0.0935,
"step": 3590
},
{
"epoch": 8.04,
"learning_rate": 1.8507316390884894e-05,
"loss": 0.081,
"step": 3600
},
{
"epoch": 8.06,
"learning_rate": 1.84978028918458e-05,
"loss": 0.0646,
"step": 3610
},
{
"epoch": 8.08,
"learning_rate": 1.848826163361671e-05,
"loss": 0.075,
"step": 3620
},
{
"epoch": 8.1,
"learning_rate": 1.8478692647365402e-05,
"loss": 0.0654,
"step": 3630
},
{
"epoch": 8.12,
"learning_rate": 1.846909596435023e-05,
"loss": 0.0885,
"step": 3640
},
{
"epoch": 8.15,
"learning_rate": 1.845947161592002e-05,
"loss": 0.0861,
"step": 3650
},
{
"epoch": 8.17,
"learning_rate": 1.844981963351397e-05,
"loss": 0.0781,
"step": 3660
},
{
"epoch": 8.19,
"learning_rate": 1.8440140048661547e-05,
"loss": 0.0802,
"step": 3670
},
{
"epoch": 8.21,
"learning_rate": 1.843043289298239e-05,
"loss": 0.0633,
"step": 3680
},
{
"epoch": 8.24,
"learning_rate": 1.8420698198186197e-05,
"loss": 0.0585,
"step": 3690
},
{
"epoch": 8.26,
"learning_rate": 1.841093599607263e-05,
"loss": 0.0751,
"step": 3700
},
{
"epoch": 8.28,
"learning_rate": 1.8401146318531204e-05,
"loss": 0.0834,
"step": 3710
},
{
"epoch": 8.3,
"learning_rate": 1.8391329197541186e-05,
"loss": 0.0924,
"step": 3720
},
{
"epoch": 8.33,
"learning_rate": 1.838148466517149e-05,
"loss": 0.0761,
"step": 3730
},
{
"epoch": 8.35,
"learning_rate": 1.8371612753580583e-05,
"loss": 0.0866,
"step": 3740
},
{
"epoch": 8.37,
"learning_rate": 1.8361713495016354e-05,
"loss": 0.0847,
"step": 3750
},
{
"epoch": 8.39,
"learning_rate": 1.8351786921816037e-05,
"loss": 0.0706,
"step": 3760
},
{
"epoch": 8.42,
"learning_rate": 1.8341833066406083e-05,
"loss": 0.0917,
"step": 3770
},
{
"epoch": 8.44,
"learning_rate": 1.8331851961302075e-05,
"loss": 0.0748,
"step": 3780
},
{
"epoch": 8.46,
"learning_rate": 1.83218436391086e-05,
"loss": 0.0767,
"step": 3790
},
{
"epoch": 8.48,
"learning_rate": 1.8311808132519157e-05,
"loss": 0.0702,
"step": 3800
},
{
"epoch": 8.5,
"learning_rate": 1.830174547431605e-05,
"loss": 0.0851,
"step": 3810
},
{
"epoch": 8.53,
"learning_rate": 1.8291655697370276e-05,
"loss": 0.0713,
"step": 3820
},
{
"epoch": 8.55,
"learning_rate": 1.8281538834641416e-05,
"loss": 0.0869,
"step": 3830
},
{
"epoch": 8.57,
"learning_rate": 1.8271394919177528e-05,
"loss": 0.0823,
"step": 3840
},
{
"epoch": 8.59,
"learning_rate": 1.8261223984115052e-05,
"loss": 0.0872,
"step": 3850
},
{
"epoch": 8.62,
"learning_rate": 1.8251026062678673e-05,
"loss": 0.0775,
"step": 3860
},
{
"epoch": 8.64,
"learning_rate": 1.8240801188181257e-05,
"loss": 0.0862,
"step": 3870
},
{
"epoch": 8.66,
"learning_rate": 1.823054939402369e-05,
"loss": 0.1073,
"step": 3880
},
{
"epoch": 8.68,
"learning_rate": 1.8220270713694803e-05,
"loss": 0.0979,
"step": 3890
},
{
"epoch": 8.71,
"learning_rate": 1.8209965180771262e-05,
"loss": 0.0851,
"step": 3900
},
{
"epoch": 8.73,
"learning_rate": 1.8199632828917445e-05,
"loss": 0.0912,
"step": 3910
},
{
"epoch": 8.75,
"learning_rate": 1.8189273691885336e-05,
"loss": 0.0803,
"step": 3920
},
{
"epoch": 8.77,
"learning_rate": 1.8178887803514415e-05,
"loss": 0.0788,
"step": 3930
},
{
"epoch": 8.79,
"learning_rate": 1.8168475197731553e-05,
"loss": 0.0764,
"step": 3940
},
{
"epoch": 8.82,
"learning_rate": 1.81580359085509e-05,
"loss": 0.1058,
"step": 3950
},
{
"epoch": 8.84,
"learning_rate": 1.814756997007376e-05,
"loss": 0.1,
"step": 3960
},
{
"epoch": 8.86,
"learning_rate": 1.8137077416488496e-05,
"loss": 0.09,
"step": 3970
},
{
"epoch": 8.88,
"learning_rate": 1.8126558282070417e-05,
"loss": 0.0848,
"step": 3980
},
{
"epoch": 8.91,
"learning_rate": 1.8116012601181655e-05,
"loss": 0.0852,
"step": 3990
},
{
"epoch": 8.93,
"learning_rate": 1.810544040827107e-05,
"loss": 0.1047,
"step": 4000
},
{
"epoch": 8.95,
"learning_rate": 1.8094841737874108e-05,
"loss": 0.096,
"step": 4010
},
{
"epoch": 8.97,
"learning_rate": 1.8084216624612726e-05,
"loss": 0.0942,
"step": 4020
},
{
"epoch": 9.0,
"learning_rate": 1.8073565103195254e-05,
"loss": 0.0985,
"step": 4030
},
{
"epoch": 9.02,
"learning_rate": 1.8062887208416282e-05,
"loss": 0.0743,
"step": 4040
},
{
"epoch": 9.04,
"learning_rate": 1.8052182975156557e-05,
"loss": 0.0912,
"step": 4050
},
{
"epoch": 9.06,
"learning_rate": 1.8041452438382873e-05,
"loss": 0.0667,
"step": 4060
},
{
"epoch": 9.08,
"learning_rate": 1.8030695633147926e-05,
"loss": 0.0882,
"step": 4070
},
{
"epoch": 9.11,
"learning_rate": 1.801991259459024e-05,
"loss": 0.0683,
"step": 4080
},
{
"epoch": 9.13,
"learning_rate": 1.8009103357934024e-05,
"loss": 0.0732,
"step": 4090
},
{
"epoch": 9.15,
"learning_rate": 1.7998267958489076e-05,
"loss": 0.0658,
"step": 4100
},
{
"epoch": 9.17,
"learning_rate": 1.7987406431650653e-05,
"loss": 0.0709,
"step": 4110
},
{
"epoch": 9.2,
"learning_rate": 1.797651881289935e-05,
"loss": 0.0759,
"step": 4120
},
{
"epoch": 9.22,
"learning_rate": 1.7965605137801015e-05,
"loss": 0.0793,
"step": 4130
},
{
"epoch": 9.24,
"learning_rate": 1.79546654420066e-05,
"loss": 0.0936,
"step": 4140
},
{
"epoch": 9.26,
"learning_rate": 1.7943699761252057e-05,
"loss": 0.0862,
"step": 4150
},
{
"epoch": 9.29,
"learning_rate": 1.7932708131358222e-05,
"loss": 0.0682,
"step": 4160
},
{
"epoch": 9.31,
"learning_rate": 1.7921690588230698e-05,
"loss": 0.0783,
"step": 4170
},
{
"epoch": 9.33,
"learning_rate": 1.7910647167859744e-05,
"loss": 0.0736,
"step": 4180
},
{
"epoch": 9.35,
"learning_rate": 1.7899577906320135e-05,
"loss": 0.063,
"step": 4190
},
{
"epoch": 9.38,
"learning_rate": 1.7888482839771074e-05,
"loss": 0.0746,
"step": 4200
},
{
"epoch": 9.4,
"learning_rate": 1.787736200445606e-05,
"loss": 0.0806,
"step": 4210
},
{
"epoch": 9.42,
"learning_rate": 1.786621543670275e-05,
"loss": 0.081,
"step": 4220
},
{
"epoch": 9.44,
"learning_rate": 1.7855043172922883e-05,
"loss": 0.0712,
"step": 4230
},
{
"epoch": 9.46,
"learning_rate": 1.7843845249612122e-05,
"loss": 0.0719,
"step": 4240
},
{
"epoch": 9.49,
"learning_rate": 1.7832621703349956e-05,
"loss": 0.0617,
"step": 4250
},
{
"epoch": 9.51,
"learning_rate": 1.7821372570799574e-05,
"loss": 0.1026,
"step": 4260
},
{
"epoch": 9.53,
"learning_rate": 1.781009788870775e-05,
"loss": 0.0749,
"step": 4270
},
{
"epoch": 9.55,
"learning_rate": 1.779879769390471e-05,
"loss": 0.0857,
"step": 4280
},
{
"epoch": 9.58,
"learning_rate": 1.7787472023304023e-05,
"loss": 0.0699,
"step": 4290
},
{
"epoch": 9.6,
"learning_rate": 1.7776120913902487e-05,
"loss": 0.0845,
"step": 4300
},
{
"epoch": 9.62,
"learning_rate": 1.7764744402779992e-05,
"loss": 0.0678,
"step": 4310
},
{
"epoch": 9.64,
"learning_rate": 1.77533425270994e-05,
"loss": 0.0859,
"step": 4320
},
{
"epoch": 9.67,
"learning_rate": 1.7741915324106445e-05,
"loss": 0.0802,
"step": 4330
},
{
"epoch": 9.69,
"learning_rate": 1.7730462831129584e-05,
"loss": 0.0708,
"step": 4340
},
{
"epoch": 9.71,
"learning_rate": 1.771898508557989e-05,
"loss": 0.0808,
"step": 4350
},
{
"epoch": 9.73,
"learning_rate": 1.7707482124950923e-05,
"loss": 0.0799,
"step": 4360
},
{
"epoch": 9.75,
"learning_rate": 1.7695953986818625e-05,
"loss": 0.0734,
"step": 4370
},
{
"epoch": 9.78,
"learning_rate": 1.7684400708841165e-05,
"loss": 0.0818,
"step": 4380
},
{
"epoch": 9.8,
"learning_rate": 1.7672822328758852e-05,
"loss": 0.0706,
"step": 4390
},
{
"epoch": 9.82,
"learning_rate": 1.7661218884393977e-05,
"loss": 0.0775,
"step": 4400
},
{
"epoch": 9.84,
"learning_rate": 1.764959041365073e-05,
"loss": 0.0839,
"step": 4410
},
{
"epoch": 9.87,
"learning_rate": 1.7637936954515026e-05,
"loss": 0.0785,
"step": 4420
},
{
"epoch": 9.89,
"learning_rate": 1.7626258545054425e-05,
"loss": 0.0739,
"step": 4430
},
{
"epoch": 9.91,
"learning_rate": 1.7614555223417987e-05,
"loss": 0.0892,
"step": 4440
},
{
"epoch": 9.93,
"learning_rate": 1.7602827027836153e-05,
"loss": 0.0835,
"step": 4450
},
{
"epoch": 9.96,
"learning_rate": 1.7591073996620607e-05,
"loss": 0.0785,
"step": 4460
},
{
"epoch": 9.98,
"learning_rate": 1.757929616816418e-05,
"loss": 0.0927,
"step": 4470
},
{
"epoch": 10.0,
"learning_rate": 1.756749358094069e-05,
"loss": 0.1003,
"step": 4480
},
{
"epoch": 10.02,
"learning_rate": 1.755566627350484e-05,
"loss": 0.0677,
"step": 4490
},
{
"epoch": 10.04,
"learning_rate": 1.754381428449209e-05,
"loss": 0.076,
"step": 4500
},
{
"epoch": 10.07,
"learning_rate": 1.7531937652618515e-05,
"loss": 0.0721,
"step": 4510
},
{
"epoch": 10.09,
"learning_rate": 1.7520036416680687e-05,
"loss": 0.0616,
"step": 4520
},
{
"epoch": 10.11,
"learning_rate": 1.7508110615555573e-05,
"loss": 0.0682,
"step": 4530
},
{
"epoch": 10.13,
"learning_rate": 1.749616028820036e-05,
"loss": 0.0621,
"step": 4540
},
{
"epoch": 10.16,
"learning_rate": 1.748418547365236e-05,
"loss": 0.07,
"step": 4550
},
{
"epoch": 10.18,
"learning_rate": 1.7472186211028884e-05,
"loss": 0.0971,
"step": 4560
},
{
"epoch": 10.2,
"learning_rate": 1.7460162539527104e-05,
"loss": 0.0721,
"step": 4570
},
{
"epoch": 10.22,
"learning_rate": 1.7448114498423915e-05,
"loss": 0.0736,
"step": 4580
},
{
"epoch": 10.25,
"learning_rate": 1.743604212707583e-05,
"loss": 0.0674,
"step": 4590
},
{
"epoch": 10.27,
"learning_rate": 1.7423945464918835e-05,
"loss": 0.0675,
"step": 4600
},
{
"epoch": 10.29,
"learning_rate": 1.741182455146827e-05,
"loss": 0.0677,
"step": 4610
},
{
"epoch": 10.31,
"learning_rate": 1.739967942631869e-05,
"loss": 0.0677,
"step": 4620
},
{
"epoch": 10.33,
"learning_rate": 1.738751012914375e-05,
"loss": 0.0595,
"step": 4630
},
{
"epoch": 10.36,
"learning_rate": 1.7375316699696042e-05,
"loss": 0.0675,
"step": 4640
},
{
"epoch": 10.38,
"learning_rate": 1.736309917780702e-05,
"loss": 0.0688,
"step": 4650
},
{
"epoch": 10.4,
"learning_rate": 1.7350857603386816e-05,
"loss": 0.0588,
"step": 4660
},
{
"epoch": 10.42,
"learning_rate": 1.733859201642415e-05,
"loss": 0.0733,
"step": 4670
},
{
"epoch": 10.45,
"learning_rate": 1.732630245698617e-05,
"loss": 0.0745,
"step": 4680
},
{
"epoch": 10.47,
"learning_rate": 1.7313988965218337e-05,
"loss": 0.0731,
"step": 4690
},
{
"epoch": 10.49,
"learning_rate": 1.730165158134429e-05,
"loss": 0.0991,
"step": 4700
},
{
"epoch": 10.51,
"learning_rate": 1.7289290345665713e-05,
"loss": 0.0693,
"step": 4710
},
{
"epoch": 10.54,
"learning_rate": 1.7276905298562208e-05,
"loss": 0.0782,
"step": 4720
},
{
"epoch": 10.56,
"learning_rate": 1.7264496480491165e-05,
"loss": 0.0831,
"step": 4730
},
{
"epoch": 10.58,
"learning_rate": 1.7252063931987607e-05,
"loss": 0.0764,
"step": 4740
},
{
"epoch": 10.6,
"learning_rate": 1.7239607693664103e-05,
"loss": 0.0632,
"step": 4750
},
{
"epoch": 10.62,
"learning_rate": 1.7227127806210578e-05,
"loss": 0.06,
"step": 4760
},
{
"epoch": 10.65,
"learning_rate": 1.7214624310394236e-05,
"loss": 0.0835,
"step": 4770
},
{
"epoch": 10.67,
"learning_rate": 1.7202097247059383e-05,
"loss": 0.0731,
"step": 4780
},
{
"epoch": 10.69,
"learning_rate": 1.7189546657127315e-05,
"loss": 0.0789,
"step": 4790
},
{
"epoch": 10.71,
"learning_rate": 1.717697258159619e-05,
"loss": 0.0813,
"step": 4800
},
{
"epoch": 10.74,
"learning_rate": 1.7164375061540877e-05,
"loss": 0.0744,
"step": 4810
},
{
"epoch": 10.76,
"learning_rate": 1.715175413811283e-05,
"loss": 0.0639,
"step": 4820
},
{
"epoch": 10.78,
"learning_rate": 1.7139109852539954e-05,
"loss": 0.076,
"step": 4830
},
{
"epoch": 10.8,
"learning_rate": 1.712644224612647e-05,
"loss": 0.0798,
"step": 4840
},
{
"epoch": 10.83,
"learning_rate": 1.7113751360252777e-05,
"loss": 0.0736,
"step": 4850
},
{
"epoch": 10.85,
"learning_rate": 1.7101037236375324e-05,
"loss": 0.0759,
"step": 4860
},
{
"epoch": 10.87,
"learning_rate": 1.708829991602647e-05,
"loss": 0.0781,
"step": 4870
},
{
"epoch": 10.89,
"learning_rate": 1.707553944081434e-05,
"loss": 0.0787,
"step": 4880
},
{
"epoch": 10.92,
"learning_rate": 1.7062755852422705e-05,
"loss": 0.0717,
"step": 4890
},
{
"epoch": 10.94,
"learning_rate": 1.7049949192610845e-05,
"loss": 0.0961,
"step": 4900
},
{
"epoch": 10.96,
"learning_rate": 1.7037119503213385e-05,
"loss": 0.0717,
"step": 4910
},
{
"epoch": 10.98,
"learning_rate": 1.7024266826140194e-05,
"loss": 0.0863,
"step": 4920
},
{
"epoch": 11.0,
"learning_rate": 1.701139120337624e-05,
"loss": 0.0834,
"step": 4930
},
{
"epoch": 11.03,
"learning_rate": 1.699849267698143e-05,
"loss": 0.0523,
"step": 4940
},
{
"epoch": 11.05,
"learning_rate": 1.698557128909049e-05,
"loss": 0.0642,
"step": 4950
},
{
"epoch": 11.07,
"learning_rate": 1.6972627081912848e-05,
"loss": 0.0741,
"step": 4960
},
{
"epoch": 11.09,
"learning_rate": 1.695966009773244e-05,
"loss": 0.0673,
"step": 4970
},
{
"epoch": 11.12,
"learning_rate": 1.6946670378907635e-05,
"loss": 0.0599,
"step": 4980
},
{
"epoch": 11.14,
"learning_rate": 1.6933657967871056e-05,
"loss": 0.067,
"step": 4990
},
{
"epoch": 11.16,
"learning_rate": 1.6920622907129452e-05,
"loss": 0.0747,
"step": 5000
},
{
"epoch": 11.18,
"learning_rate": 1.690756523926356e-05,
"loss": 0.0662,
"step": 5010
},
{
"epoch": 11.21,
"learning_rate": 1.6894485006927972e-05,
"loss": 0.0572,
"step": 5020
},
{
"epoch": 11.23,
"learning_rate": 1.688138225285098e-05,
"loss": 0.0635,
"step": 5030
},
{
"epoch": 11.25,
"learning_rate": 1.6868257019834464e-05,
"loss": 0.0675,
"step": 5040
},
{
"epoch": 11.27,
"learning_rate": 1.685510935075371e-05,
"loss": 0.0637,
"step": 5050
},
{
"epoch": 11.29,
"learning_rate": 1.684193928855731e-05,
"loss": 0.0649,
"step": 5060
},
{
"epoch": 11.32,
"learning_rate": 1.682874687626701e-05,
"loss": 0.0656,
"step": 5070
},
{
"epoch": 11.34,
"learning_rate": 1.6815532156977553e-05,
"loss": 0.0773,
"step": 5080
},
{
"epoch": 11.36,
"learning_rate": 1.6802295173856558e-05,
"loss": 0.0664,
"step": 5090
},
{
"epoch": 11.38,
"learning_rate": 1.678903597014437e-05,
"loss": 0.0864,
"step": 5100
},
{
"epoch": 11.41,
"learning_rate": 1.6775754589153913e-05,
"loss": 0.0873,
"step": 5110
},
{
"epoch": 11.43,
"learning_rate": 1.676245107427058e-05,
"loss": 0.0736,
"step": 5120
},
{
"epoch": 11.45,
"learning_rate": 1.6749125468952033e-05,
"loss": 0.0716,
"step": 5130
},
{
"epoch": 11.47,
"learning_rate": 1.673577781672812e-05,
"loss": 0.0653,
"step": 5140
},
{
"epoch": 11.5,
"learning_rate": 1.672240816120071e-05,
"loss": 0.0621,
"step": 5150
},
{
"epoch": 11.52,
"learning_rate": 1.670901654604353e-05,
"loss": 0.067,
"step": 5160
},
{
"epoch": 11.54,
"learning_rate": 1.669560301500205e-05,
"loss": 0.0693,
"step": 5170
},
{
"epoch": 11.56,
"learning_rate": 1.668216761189334e-05,
"loss": 0.0696,
"step": 5180
},
{
"epoch": 11.58,
"learning_rate": 1.6668710380605902e-05,
"loss": 0.079,
"step": 5190
},
{
"epoch": 11.61,
"learning_rate": 1.6655231365099556e-05,
"loss": 0.0593,
"step": 5200
},
{
"epoch": 11.63,
"learning_rate": 1.6641730609405276e-05,
"loss": 0.0737,
"step": 5210
},
{
"epoch": 11.65,
"learning_rate": 1.6628208157625055e-05,
"loss": 0.0909,
"step": 5220
},
{
"epoch": 11.67,
"learning_rate": 1.6614664053931757e-05,
"loss": 0.0885,
"step": 5230
},
{
"epoch": 11.7,
"learning_rate": 1.6601098342568978e-05,
"loss": 0.0679,
"step": 5240
},
{
"epoch": 11.72,
"learning_rate": 1.65875110678509e-05,
"loss": 0.073,
"step": 5250
},
{
"epoch": 11.74,
"learning_rate": 1.6573902274162135e-05,
"loss": 0.0751,
"step": 5260
},
{
"epoch": 11.76,
"learning_rate": 1.6560272005957604e-05,
"loss": 0.0575,
"step": 5270
},
{
"epoch": 11.79,
"learning_rate": 1.6546620307762364e-05,
"loss": 0.064,
"step": 5280
},
{
"epoch": 11.81,
"learning_rate": 1.6532947224171482e-05,
"loss": 0.0773,
"step": 5290
},
{
"epoch": 11.83,
"learning_rate": 1.6519252799849887e-05,
"loss": 0.059,
"step": 5300
},
{
"epoch": 11.85,
"learning_rate": 1.650553707953221e-05,
"loss": 0.0674,
"step": 5310
},
{
"epoch": 11.88,
"learning_rate": 1.6491800108022657e-05,
"loss": 0.0746,
"step": 5320
},
{
"epoch": 11.9,
"learning_rate": 1.6478041930194848e-05,
"loss": 0.0781,
"step": 5330
},
{
"epoch": 11.92,
"learning_rate": 1.6464262590991683e-05,
"loss": 0.0619,
"step": 5340
},
{
"epoch": 11.94,
"learning_rate": 1.6450462135425187e-05,
"loss": 0.0694,
"step": 5350
},
{
"epoch": 11.96,
"learning_rate": 1.6436640608576354e-05,
"loss": 0.074,
"step": 5360
},
{
"epoch": 11.99,
"learning_rate": 1.642279805559502e-05,
"loss": 0.0714,
"step": 5370
},
{
"epoch": 12.01,
"learning_rate": 1.6408934521699706e-05,
"loss": 0.0652,
"step": 5380
},
{
"epoch": 12.03,
"learning_rate": 1.639505005217747e-05,
"loss": 0.0722,
"step": 5390
},
{
"epoch": 12.05,
"learning_rate": 1.6381144692383754e-05,
"loss": 0.0635,
"step": 5400
},
{
"epoch": 12.08,
"learning_rate": 1.636721848774224e-05,
"loss": 0.0502,
"step": 5410
},
{
"epoch": 12.1,
"learning_rate": 1.635327148374471e-05,
"loss": 0.0481,
"step": 5420
},
{
"epoch": 12.12,
"learning_rate": 1.633930372595088e-05,
"loss": 0.0622,
"step": 5430
},
{
"epoch": 12.14,
"learning_rate": 1.6325315259988275e-05,
"loss": 0.0765,
"step": 5440
},
{
"epoch": 12.17,
"learning_rate": 1.631130613155205e-05,
"loss": 0.0598,
"step": 5450
},
{
"epoch": 12.19,
"learning_rate": 1.6297276386404872e-05,
"loss": 0.0578,
"step": 5460
},
{
"epoch": 12.21,
"learning_rate": 1.628322607037674e-05,
"loss": 0.0714,
"step": 5470
},
{
"epoch": 12.23,
"learning_rate": 1.626915522936486e-05,
"loss": 0.0503,
"step": 5480
},
{
"epoch": 12.25,
"learning_rate": 1.6255063909333486e-05,
"loss": 0.0532,
"step": 5490
},
{
"epoch": 12.28,
"learning_rate": 1.6240952156313762e-05,
"loss": 0.0534,
"step": 5500
},
{
"epoch": 12.3,
"learning_rate": 1.622682001640359e-05,
"loss": 0.0687,
"step": 5510
},
{
"epoch": 12.32,
"learning_rate": 1.6212667535767456e-05,
"loss": 0.0721,
"step": 5520
},
{
"epoch": 12.34,
"learning_rate": 1.6198494760636303e-05,
"loss": 0.0604,
"step": 5530
},
{
"epoch": 12.37,
"learning_rate": 1.618430173730736e-05,
"loss": 0.0712,
"step": 5540
},
{
"epoch": 12.39,
"learning_rate": 1.617008851214401e-05,
"loss": 0.0704,
"step": 5550
},
{
"epoch": 12.41,
"learning_rate": 1.6155855131575614e-05,
"loss": 0.0701,
"step": 5560
},
{
"epoch": 12.43,
"learning_rate": 1.6141601642097382e-05,
"loss": 0.0629,
"step": 5570
},
{
"epoch": 12.46,
"learning_rate": 1.6127328090270213e-05,
"loss": 0.0712,
"step": 5580
},
{
"epoch": 12.48,
"learning_rate": 1.611303452272053e-05,
"loss": 0.0794,
"step": 5590
},
{
"epoch": 12.5,
"learning_rate": 1.609872098614017e-05,
"loss": 0.0689,
"step": 5600
},
{
"epoch": 12.52,
"learning_rate": 1.608438752728616e-05,
"loss": 0.067,
"step": 5610
},
{
"epoch": 12.54,
"learning_rate": 1.6070034192980638e-05,
"loss": 0.0595,
"step": 5620
},
{
"epoch": 12.57,
"learning_rate": 1.6055661030110655e-05,
"loss": 0.0716,
"step": 5630
},
{
"epoch": 12.59,
"learning_rate": 1.6041268085628042e-05,
"loss": 0.074,
"step": 5640
},
{
"epoch": 12.61,
"learning_rate": 1.602685540654924e-05,
"loss": 0.0628,
"step": 5650
},
{
"epoch": 12.63,
"learning_rate": 1.6012423039955153e-05,
"loss": 0.0742,
"step": 5660
},
{
"epoch": 12.66,
"learning_rate": 1.5997971032991007e-05,
"loss": 0.0745,
"step": 5670
},
{
"epoch": 12.68,
"learning_rate": 1.5983499432866187e-05,
"loss": 0.0544,
"step": 5680
},
{
"epoch": 12.7,
"learning_rate": 1.596900828685407e-05,
"loss": 0.0643,
"step": 5690
},
{
"epoch": 12.72,
"learning_rate": 1.5954497642291897e-05,
"loss": 0.061,
"step": 5700
},
{
"epoch": 12.75,
"learning_rate": 1.593996754658059e-05,
"loss": 0.0602,
"step": 5710
},
{
"epoch": 12.77,
"learning_rate": 1.5925418047184615e-05,
"loss": 0.069,
"step": 5720
},
{
"epoch": 12.79,
"learning_rate": 1.591084919163183e-05,
"loss": 0.0646,
"step": 5730
},
{
"epoch": 12.81,
"learning_rate": 1.589626102751331e-05,
"loss": 0.0649,
"step": 5740
},
{
"epoch": 12.83,
"learning_rate": 1.588165360248321e-05,
"loss": 0.0705,
"step": 5750
},
{
"epoch": 12.86,
"learning_rate": 1.5867026964258614e-05,
"loss": 0.0566,
"step": 5760
},
{
"epoch": 12.88,
"learning_rate": 1.5852381160619343e-05,
"loss": 0.0815,
"step": 5770
},
{
"epoch": 12.9,
"learning_rate": 1.5837716239407855e-05,
"loss": 0.0844,
"step": 5780
},
{
"epoch": 12.92,
"learning_rate": 1.582303224852903e-05,
"loss": 0.0728,
"step": 5790
},
{
"epoch": 12.95,
"learning_rate": 1.580832923595006e-05,
"loss": 0.0701,
"step": 5800
},
{
"epoch": 12.97,
"learning_rate": 1.5793607249700268e-05,
"loss": 0.0592,
"step": 5810
},
{
"epoch": 12.99,
"learning_rate": 1.5778866337870952e-05,
"loss": 0.0654,
"step": 5820
},
{
"epoch": 13.01,
"learning_rate": 1.5764106548615244e-05,
"loss": 0.0604,
"step": 5830
},
{
"epoch": 13.04,
"learning_rate": 1.5749327930147932e-05,
"loss": 0.0587,
"step": 5840
},
{
"epoch": 13.06,
"learning_rate": 1.573453053074532e-05,
"loss": 0.0545,
"step": 5850
},
{
"epoch": 13.08,
"learning_rate": 1.571971439874505e-05,
"loss": 0.0531,
"step": 5860
},
{
"epoch": 13.1,
"learning_rate": 1.570487958254597e-05,
"loss": 0.0651,
"step": 5870
},
{
"epoch": 13.12,
"learning_rate": 1.569002613060796e-05,
"loss": 0.0521,
"step": 5880
},
{
"epoch": 13.15,
"learning_rate": 1.5675154091451765e-05,
"loss": 0.0496,
"step": 5890
},
{
"epoch": 13.17,
"learning_rate": 1.566026351365886e-05,
"loss": 0.0479,
"step": 5900
},
{
"epoch": 13.19,
"learning_rate": 1.5645354445871274e-05,
"loss": 0.0563,
"step": 5910
},
{
"epoch": 13.21,
"learning_rate": 1.5630426936791433e-05,
"loss": 0.061,
"step": 5920
},
{
"epoch": 13.24,
"learning_rate": 1.5615481035182013e-05,
"loss": 0.0489,
"step": 5930
},
{
"epoch": 13.26,
"learning_rate": 1.5600516789865767e-05,
"loss": 0.0558,
"step": 5940
},
{
"epoch": 13.28,
"learning_rate": 1.5585534249725362e-05,
"loss": 0.0708,
"step": 5950
},
{
"epoch": 13.3,
"learning_rate": 1.5570533463703233e-05,
"loss": 0.0541,
"step": 5960
},
{
"epoch": 13.33,
"learning_rate": 1.555551448080143e-05,
"loss": 0.074,
"step": 5970
},
{
"epoch": 13.35,
"learning_rate": 1.5540477350081423e-05,
"loss": 0.075,
"step": 5980
},
{
"epoch": 13.37,
"learning_rate": 1.5525422120663986e-05,
"loss": 0.0741,
"step": 5990
},
{
"epoch": 13.39,
"learning_rate": 1.5510348841728997e-05,
"loss": 0.0765,
"step": 6000
},
{
"epoch": 13.42,
"learning_rate": 1.5495257562515308e-05,
"loss": 0.0761,
"step": 6010
},
{
"epoch": 13.44,
"learning_rate": 1.5480148332320562e-05,
"loss": 0.0858,
"step": 6020
},
{
"epoch": 13.46,
"learning_rate": 1.5465021200501046e-05,
"loss": 0.0736,
"step": 6030
},
{
"epoch": 13.48,
"learning_rate": 1.5449876216471525e-05,
"loss": 0.0637,
"step": 6040
},
{
"epoch": 13.5,
"learning_rate": 1.5434713429705078e-05,
"loss": 0.082,
"step": 6050
},
{
"epoch": 13.53,
"learning_rate": 1.5419532889732943e-05,
"loss": 0.0609,
"step": 6060
},
{
"epoch": 13.55,
"learning_rate": 1.540433464614435e-05,
"loss": 0.0865,
"step": 6070
},
{
"epoch": 13.57,
"learning_rate": 1.5389118748586357e-05,
"loss": 0.0696,
"step": 6080
},
{
"epoch": 13.59,
"learning_rate": 1.537388524676369e-05,
"loss": 0.0639,
"step": 6090
},
{
"epoch": 13.62,
"learning_rate": 1.5358634190438592e-05,
"loss": 0.0698,
"step": 6100
},
{
"epoch": 13.64,
"learning_rate": 1.5343365629430638e-05,
"loss": 0.0603,
"step": 6110
},
{
"epoch": 13.66,
"learning_rate": 1.5328079613616592e-05,
"loss": 0.0865,
"step": 6120
},
{
"epoch": 13.68,
"learning_rate": 1.531277619293023e-05,
"loss": 0.0667,
"step": 6130
},
{
"epoch": 13.71,
"learning_rate": 1.5297455417362194e-05,
"loss": 0.0736,
"step": 6140
},
{
"epoch": 13.73,
"learning_rate": 1.52821173369598e-05,
"loss": 0.0717,
"step": 6150
},
{
"epoch": 13.75,
"learning_rate": 1.526676200182691e-05,
"loss": 0.0644,
"step": 6160
},
{
"epoch": 13.77,
"learning_rate": 1.5251389462123748e-05,
"loss": 0.06,
"step": 6170
},
{
"epoch": 13.79,
"learning_rate": 1.5235999768066729e-05,
"loss": 0.0661,
"step": 6180
},
{
"epoch": 13.82,
"learning_rate": 1.5220592969928313e-05,
"loss": 0.0659,
"step": 6190
},
{
"epoch": 13.84,
"learning_rate": 1.5205169118036831e-05,
"loss": 0.0663,
"step": 6200
},
{
"epoch": 13.86,
"learning_rate": 1.5189728262776325e-05,
"loss": 0.074,
"step": 6210
},
{
"epoch": 13.88,
"learning_rate": 1.5174270454586375e-05,
"loss": 0.0685,
"step": 6220
},
{
"epoch": 13.91,
"learning_rate": 1.5158795743961942e-05,
"loss": 0.0735,
"step": 6230
},
{
"epoch": 13.93,
"learning_rate": 1.5143304181453204e-05,
"loss": 0.0651,
"step": 6240
},
{
"epoch": 13.95,
"learning_rate": 1.5127795817665389e-05,
"loss": 0.0663,
"step": 6250
},
{
"epoch": 13.97,
"learning_rate": 1.5112270703258602e-05,
"loss": 0.0624,
"step": 6260
},
{
"epoch": 14.0,
"learning_rate": 1.5096728888947669e-05,
"loss": 0.0694,
"step": 6270
},
{
"epoch": 14.02,
"learning_rate": 1.508117042550197e-05,
"loss": 0.0576,
"step": 6280
},
{
"epoch": 14.04,
"learning_rate": 1.5065595363745272e-05,
"loss": 0.0617,
"step": 6290
},
{
"epoch": 14.06,
"learning_rate": 1.505000375455556e-05,
"loss": 0.07,
"step": 6300
},
{
"epoch": 14.08,
"learning_rate": 1.503439564886487e-05,
"loss": 0.062,
"step": 6310
},
{
"epoch": 14.11,
"learning_rate": 1.501877109765914e-05,
"loss": 0.0588,
"step": 6320
},
{
"epoch": 14.13,
"learning_rate": 1.5003130151978012e-05,
"loss": 0.0494,
"step": 6330
},
{
"epoch": 14.15,
"learning_rate": 1.4987472862914697e-05,
"loss": 0.0531,
"step": 6340
},
{
"epoch": 14.17,
"learning_rate": 1.4971799281615782e-05,
"loss": 0.0543,
"step": 6350
},
{
"epoch": 14.2,
"learning_rate": 1.4956109459281083e-05,
"loss": 0.0561,
"step": 6360
},
{
"epoch": 14.22,
"learning_rate": 1.4940403447163467e-05,
"loss": 0.0581,
"step": 6370
},
{
"epoch": 14.24,
"learning_rate": 1.4924681296568689e-05,
"loss": 0.0728,
"step": 6380
},
{
"epoch": 14.26,
"learning_rate": 1.4908943058855213e-05,
"loss": 0.0556,
"step": 6390
},
{
"epoch": 14.29,
"learning_rate": 1.4893188785434067e-05,
"loss": 0.0574,
"step": 6400
},
{
"epoch": 14.31,
"learning_rate": 1.4877418527768654e-05,
"loss": 0.0706,
"step": 6410
},
{
"epoch": 14.33,
"learning_rate": 1.4861632337374596e-05,
"loss": 0.0667,
"step": 6420
},
{
"epoch": 14.35,
"learning_rate": 1.4845830265819552e-05,
"loss": 0.0584,
"step": 6430
},
{
"epoch": 14.38,
"learning_rate": 1.483001236472307e-05,
"loss": 0.054,
"step": 6440
},
{
"epoch": 14.4,
"learning_rate": 1.4814178685756405e-05,
"loss": 0.0609,
"step": 6450
},
{
"epoch": 14.42,
"learning_rate": 1.4798329280642345e-05,
"loss": 0.0667,
"step": 6460
},
{
"epoch": 14.44,
"learning_rate": 1.4782464201155057e-05,
"loss": 0.0561,
"step": 6470
},
{
"epoch": 14.46,
"learning_rate": 1.476658349911991e-05,
"loss": 0.0611,
"step": 6480
},
{
"epoch": 14.49,
"learning_rate": 1.4750687226413305e-05,
"loss": 0.0637,
"step": 6490
},
{
"epoch": 14.51,
"learning_rate": 1.4734775434962504e-05,
"loss": 0.0629,
"step": 6500
},
{
"epoch": 14.53,
"learning_rate": 1.471884817674546e-05,
"loss": 0.0655,
"step": 6510
},
{
"epoch": 14.55,
"learning_rate": 1.4702905503790668e-05,
"loss": 0.0625,
"step": 6520
},
{
"epoch": 14.58,
"learning_rate": 1.4686947468176955e-05,
"loss": 0.0559,
"step": 6530
},
{
"epoch": 14.6,
"learning_rate": 1.467097412203334e-05,
"loss": 0.069,
"step": 6540
},
{
"epoch": 14.62,
"learning_rate": 1.4654985517538864e-05,
"loss": 0.0687,
"step": 6550
},
{
"epoch": 14.64,
"learning_rate": 1.4638981706922401e-05,
"loss": 0.0493,
"step": 6560
},
{
"epoch": 14.67,
"learning_rate": 1.4622962742462503e-05,
"loss": 0.0691,
"step": 6570
},
{
"epoch": 14.69,
"learning_rate": 1.4606928676487223e-05,
"loss": 0.0516,
"step": 6580
},
{
"epoch": 14.71,
"learning_rate": 1.459087956137394e-05,
"loss": 0.0676,
"step": 6590
},
{
"epoch": 14.73,
"learning_rate": 1.4574815449549209e-05,
"loss": 0.0629,
"step": 6600
},
{
"epoch": 14.75,
"learning_rate": 1.4558736393488553e-05,
"loss": 0.0646,
"step": 6610
},
{
"epoch": 14.78,
"learning_rate": 1.4542642445716326e-05,
"loss": 0.0594,
"step": 6620
},
{
"epoch": 14.8,
"learning_rate": 1.4526533658805517e-05,
"loss": 0.0718,
"step": 6630
},
{
"epoch": 14.82,
"learning_rate": 1.4510410085377606e-05,
"loss": 0.055,
"step": 6640
},
{
"epoch": 14.84,
"learning_rate": 1.4494271778102358e-05,
"loss": 0.0549,
"step": 6650
},
{
"epoch": 14.87,
"learning_rate": 1.4478118789697675e-05,
"loss": 0.0587,
"step": 6660
},
{
"epoch": 14.89,
"learning_rate": 1.4461951172929419e-05,
"loss": 0.0508,
"step": 6670
},
{
"epoch": 14.91,
"learning_rate": 1.4445768980611233e-05,
"loss": 0.0717,
"step": 6680
},
{
"epoch": 14.93,
"learning_rate": 1.4429572265604375e-05,
"loss": 0.0635,
"step": 6690
},
{
"epoch": 14.96,
"learning_rate": 1.4413361080817545e-05,
"loss": 0.0743,
"step": 6700
},
{
"epoch": 14.98,
"learning_rate": 1.4397135479206705e-05,
"loss": 0.0793,
"step": 6710
},
{
"epoch": 15.0,
"learning_rate": 1.4380895513774922e-05,
"loss": 0.0636,
"step": 6720
},
{
"epoch": 15.02,
"learning_rate": 1.436464123757217e-05,
"loss": 0.0548,
"step": 6730
},
{
"epoch": 15.04,
"learning_rate": 1.4348372703695184e-05,
"loss": 0.0564,
"step": 6740
},
{
"epoch": 15.07,
"learning_rate": 1.4332089965287266e-05,
"loss": 0.052,
"step": 6750
},
{
"epoch": 15.09,
"learning_rate": 1.431579307553812e-05,
"loss": 0.0621,
"step": 6760
},
{
"epoch": 15.11,
"learning_rate": 1.429948208768368e-05,
"loss": 0.0538,
"step": 6770
},
{
"epoch": 15.13,
"learning_rate": 1.4283157055005928e-05,
"loss": 0.0598,
"step": 6780
},
{
"epoch": 15.16,
"learning_rate": 1.4266818030832732e-05,
"loss": 0.0495,
"step": 6790
},
{
"epoch": 15.18,
"learning_rate": 1.4250465068537664e-05,
"loss": 0.0549,
"step": 6800
},
{
"epoch": 15.2,
"learning_rate": 1.4234098221539818e-05,
"loss": 0.0596,
"step": 6810
},
{
"epoch": 15.22,
"learning_rate": 1.4217717543303657e-05,
"loss": 0.0586,
"step": 6820
},
{
"epoch": 15.25,
"learning_rate": 1.4201323087338816e-05,
"loss": 0.0634,
"step": 6830
},
{
"epoch": 15.27,
"learning_rate": 1.4184914907199942e-05,
"loss": 0.0582,
"step": 6840
},
{
"epoch": 15.29,
"learning_rate": 1.4168493056486512e-05,
"loss": 0.0531,
"step": 6850
},
{
"epoch": 15.31,
"learning_rate": 1.4152057588842657e-05,
"loss": 0.0523,
"step": 6860
},
{
"epoch": 15.33,
"learning_rate": 1.4135608557956992e-05,
"loss": 0.0532,
"step": 6870
},
{
"epoch": 15.36,
"learning_rate": 1.4119146017562441e-05,
"loss": 0.0528,
"step": 6880
},
{
"epoch": 15.38,
"learning_rate": 1.4102670021436059e-05,
"loss": 0.0503,
"step": 6890
},
{
"epoch": 15.4,
"learning_rate": 1.4086180623398842e-05,
"loss": 0.0463,
"step": 6900
},
{
"epoch": 15.42,
"learning_rate": 1.4069677877315587e-05,
"loss": 0.0672,
"step": 6910
},
{
"epoch": 15.45,
"learning_rate": 1.4053161837094675e-05,
"loss": 0.0583,
"step": 6920
},
{
"epoch": 15.47,
"learning_rate": 1.4036632556687927e-05,
"loss": 0.0539,
"step": 6930
},
{
"epoch": 15.49,
"learning_rate": 1.4020090090090408e-05,
"loss": 0.0426,
"step": 6940
},
{
"epoch": 15.51,
"learning_rate": 1.4003534491340259e-05,
"loss": 0.0569,
"step": 6950
},
{
"epoch": 15.54,
"learning_rate": 1.3986965814518521e-05,
"loss": 0.0527,
"step": 6960
},
{
"epoch": 15.56,
"learning_rate": 1.3970384113748951e-05,
"loss": 0.0593,
"step": 6970
},
{
"epoch": 15.58,
"learning_rate": 1.3953789443197857e-05,
"loss": 0.0618,
"step": 6980
},
{
"epoch": 15.6,
"learning_rate": 1.3937181857073912e-05,
"loss": 0.0595,
"step": 6990
},
{
"epoch": 15.62,
"learning_rate": 1.3920561409627974e-05,
"loss": 0.0626,
"step": 7000
},
{
"epoch": 15.65,
"learning_rate": 1.3903928155152926e-05,
"loss": 0.0622,
"step": 7010
},
{
"epoch": 15.67,
"learning_rate": 1.3887282147983472e-05,
"loss": 0.0583,
"step": 7020
},
{
"epoch": 15.69,
"learning_rate": 1.3870623442495987e-05,
"loss": 0.0582,
"step": 7030
},
{
"epoch": 15.71,
"learning_rate": 1.3853952093108323e-05,
"loss": 0.0948,
"step": 7040
},
{
"epoch": 15.74,
"learning_rate": 1.3837268154279628e-05,
"loss": 0.101,
"step": 7050
},
{
"epoch": 15.76,
"learning_rate": 1.3820571680510187e-05,
"loss": 0.0675,
"step": 7060
},
{
"epoch": 15.78,
"learning_rate": 1.3803862726341224e-05,
"loss": 0.0683,
"step": 7070
},
{
"epoch": 15.8,
"learning_rate": 1.3787141346354733e-05,
"loss": 0.0725,
"step": 7080
},
{
"epoch": 15.83,
"learning_rate": 1.3770407595173301e-05,
"loss": 0.0792,
"step": 7090
},
{
"epoch": 15.85,
"learning_rate": 1.375366152745992e-05,
"loss": 0.0737,
"step": 7100
},
{
"epoch": 15.87,
"learning_rate": 1.373690319791783e-05,
"loss": 0.067,
"step": 7110
},
{
"epoch": 15.89,
"learning_rate": 1.3720132661290311e-05,
"loss": 0.0694,
"step": 7120
},
{
"epoch": 15.92,
"learning_rate": 1.3703349972360527e-05,
"loss": 0.0655,
"step": 7130
},
{
"epoch": 15.94,
"learning_rate": 1.3686555185951334e-05,
"loss": 0.0683,
"step": 7140
},
{
"epoch": 15.96,
"learning_rate": 1.3669748356925112e-05,
"loss": 0.0771,
"step": 7150
},
{
"epoch": 15.98,
"learning_rate": 1.3652929540183578e-05,
"loss": 0.0552,
"step": 7160
},
{
"epoch": 16.0,
"learning_rate": 1.3636098790667605e-05,
"loss": 0.0605,
"step": 7170
},
{
"epoch": 16.03,
"learning_rate": 1.3619256163357046e-05,
"loss": 0.0511,
"step": 7180
},
{
"epoch": 16.05,
"learning_rate": 1.3602401713270566e-05,
"loss": 0.0545,
"step": 7190
},
{
"epoch": 16.07,
"learning_rate": 1.3585535495465432e-05,
"loss": 0.0435,
"step": 7200
},
{
"epoch": 16.09,
"learning_rate": 1.3568657565037365e-05,
"loss": 0.0541,
"step": 7210
},
{
"epoch": 16.12,
"learning_rate": 1.3551767977120341e-05,
"loss": 0.0466,
"step": 7220
},
{
"epoch": 16.14,
"learning_rate": 1.353486678688642e-05,
"loss": 0.0594,
"step": 7230
},
{
"epoch": 16.16,
"learning_rate": 1.351795404954556e-05,
"loss": 0.0581,
"step": 7240
},
{
"epoch": 16.18,
"learning_rate": 1.3501029820345446e-05,
"loss": 0.0542,
"step": 7250
},
{
"epoch": 16.21,
"learning_rate": 1.3484094154571286e-05,
"loss": 0.0551,
"step": 7260
},
{
"epoch": 16.23,
"learning_rate": 1.3467147107545668e-05,
"loss": 0.0636,
"step": 7270
},
{
"epoch": 16.25,
"learning_rate": 1.3450188734628344e-05,
"loss": 0.0518,
"step": 7280
},
{
"epoch": 16.27,
"learning_rate": 1.3433219091216069e-05,
"loss": 0.059,
"step": 7290
},
{
"epoch": 16.29,
"learning_rate": 1.3416238232742414e-05,
"loss": 0.0601,
"step": 7300
},
{
"epoch": 16.32,
"learning_rate": 1.3399246214677583e-05,
"loss": 0.0518,
"step": 7310
},
{
"epoch": 16.34,
"learning_rate": 1.338224309252824e-05,
"loss": 0.0554,
"step": 7320
},
{
"epoch": 16.36,
"learning_rate": 1.3365228921837314e-05,
"loss": 0.0554,
"step": 7330
},
{
"epoch": 16.38,
"learning_rate": 1.3348203758183831e-05,
"loss": 0.0533,
"step": 7340
},
{
"epoch": 16.41,
"learning_rate": 1.3331167657182726e-05,
"loss": 0.048,
"step": 7350
},
{
"epoch": 16.43,
"learning_rate": 1.3314120674484663e-05,
"loss": 0.0533,
"step": 7360
},
{
"epoch": 16.45,
"learning_rate": 1.3297062865775851e-05,
"loss": 0.0553,
"step": 7370
},
{
"epoch": 16.47,
"learning_rate": 1.327999428677786e-05,
"loss": 0.0521,
"step": 7380
},
{
"epoch": 16.5,
"learning_rate": 1.3262914993247454e-05,
"loss": 0.0573,
"step": 7390
},
{
"epoch": 16.52,
"learning_rate": 1.324582504097638e-05,
"loss": 0.0546,
"step": 7400
},
{
"epoch": 16.54,
"learning_rate": 1.3228724485791225e-05,
"loss": 0.0527,
"step": 7410
},
{
"epoch": 16.56,
"learning_rate": 1.321161338355319e-05,
"loss": 0.0555,
"step": 7420
},
{
"epoch": 16.58,
"learning_rate": 1.3194491790157947e-05,
"loss": 0.0528,
"step": 7430
},
{
"epoch": 16.61,
"learning_rate": 1.3177359761535427e-05,
"loss": 0.0493,
"step": 7440
},
{
"epoch": 16.63,
"learning_rate": 1.3160217353649652e-05,
"loss": 0.0492,
"step": 7450
},
{
"epoch": 16.65,
"learning_rate": 1.3143064622498551e-05,
"loss": 0.0389,
"step": 7460
},
{
"epoch": 16.67,
"learning_rate": 1.312590162411378e-05,
"loss": 0.064,
"step": 7470
},
{
"epoch": 16.7,
"learning_rate": 1.310872841456052e-05,
"loss": 0.0567,
"step": 7480
},
{
"epoch": 16.72,
"learning_rate": 1.3091545049937322e-05,
"loss": 0.0606,
"step": 7490
},
{
"epoch": 16.74,
"learning_rate": 1.3074351586375906e-05,
"loss": 0.0553,
"step": 7500
},
{
"epoch": 16.76,
"learning_rate": 1.305714808004098e-05,
"loss": 0.0594,
"step": 7510
},
{
"epoch": 16.79,
"learning_rate": 1.3039934587130056e-05,
"loss": 0.0504,
"step": 7520
},
{
"epoch": 16.81,
"learning_rate": 1.3022711163873272e-05,
"loss": 0.0567,
"step": 7530
},
{
"epoch": 16.83,
"learning_rate": 1.3005477866533202e-05,
"loss": 0.0516,
"step": 7540
},
{
"epoch": 16.85,
"learning_rate": 1.2988234751404683e-05,
"loss": 0.0469,
"step": 7550
},
{
"epoch": 16.88,
"learning_rate": 1.2970981874814613e-05,
"loss": 0.0529,
"step": 7560
},
{
"epoch": 16.9,
"learning_rate": 1.2953719293121775e-05,
"loss": 0.0524,
"step": 7570
},
{
"epoch": 16.92,
"learning_rate": 1.2936447062716668e-05,
"loss": 0.0564,
"step": 7580
},
{
"epoch": 16.94,
"learning_rate": 1.2919165240021303e-05,
"loss": 0.0623,
"step": 7590
},
{
"epoch": 16.96,
"learning_rate": 1.2901873881489021e-05,
"loss": 0.0711,
"step": 7600
},
{
"epoch": 16.99,
"learning_rate": 1.288457304360432e-05,
"loss": 0.0605,
"step": 7610
},
{
"epoch": 17.01,
"learning_rate": 1.2867262782882662e-05,
"loss": 0.0533,
"step": 7620
},
{
"epoch": 17.03,
"learning_rate": 1.2849943155870284e-05,
"loss": 0.0508,
"step": 7630
},
{
"epoch": 17.05,
"learning_rate": 1.2832614219144027e-05,
"loss": 0.0406,
"step": 7640
},
{
"epoch": 17.08,
"learning_rate": 1.2815276029311138e-05,
"loss": 0.0545,
"step": 7650
},
{
"epoch": 17.1,
"learning_rate": 1.2797928643009097e-05,
"loss": 0.0526,
"step": 7660
},
{
"epoch": 17.12,
"learning_rate": 1.2780572116905418e-05,
"loss": 0.0453,
"step": 7670
},
{
"epoch": 17.14,
"learning_rate": 1.276320650769748e-05,
"loss": 0.0373,
"step": 7680
},
{
"epoch": 17.17,
"learning_rate": 1.2745831872112318e-05,
"loss": 0.0534,
"step": 7690
},
{
"epoch": 17.19,
"learning_rate": 1.2728448266906468e-05,
"loss": 0.0424,
"step": 7700
},
{
"epoch": 17.21,
"learning_rate": 1.2711055748865765e-05,
"loss": 0.0443,
"step": 7710
},
{
"epoch": 17.23,
"learning_rate": 1.2693654374805148e-05,
"loss": 0.0491,
"step": 7720
},
{
"epoch": 17.25,
"learning_rate": 1.2676244201568498e-05,
"loss": 0.0369,
"step": 7730
},
{
"epoch": 17.28,
"learning_rate": 1.2658825286028428e-05,
"loss": 0.0428,
"step": 7740
},
{
"epoch": 17.3,
"learning_rate": 1.2641397685086124e-05,
"loss": 0.0455,
"step": 7750
},
{
"epoch": 17.32,
"learning_rate": 1.2623961455671125e-05,
"loss": 0.0429,
"step": 7760
},
{
"epoch": 17.34,
"learning_rate": 1.2606516654741172e-05,
"loss": 0.0456,
"step": 7770
},
{
"epoch": 17.37,
"learning_rate": 1.2589063339281995e-05,
"loss": 0.0584,
"step": 7780
},
{
"epoch": 17.39,
"learning_rate": 1.257160156630715e-05,
"loss": 0.0545,
"step": 7790
},
{
"epoch": 17.41,
"learning_rate": 1.2554131392857812e-05,
"loss": 0.073,
"step": 7800
},
{
"epoch": 17.43,
"learning_rate": 1.253665287600259e-05,
"loss": 0.0552,
"step": 7810
},
{
"epoch": 17.46,
"learning_rate": 1.2519166072837368e-05,
"loss": 0.0486,
"step": 7820
},
{
"epoch": 17.48,
"learning_rate": 1.250167104048508e-05,
"loss": 0.0489,
"step": 7830
},
{
"epoch": 17.5,
"learning_rate": 1.248416783609555e-05,
"loss": 0.0629,
"step": 7840
},
{
"epoch": 17.52,
"learning_rate": 1.2466656516845293e-05,
"loss": 0.0472,
"step": 7850
},
{
"epoch": 17.54,
"learning_rate": 1.244913713993734e-05,
"loss": 0.0588,
"step": 7860
},
{
"epoch": 17.57,
"learning_rate": 1.2431609762601036e-05,
"loss": 0.0581,
"step": 7870
},
{
"epoch": 17.59,
"learning_rate": 1.241407444209186e-05,
"loss": 0.0473,
"step": 7880
},
{
"epoch": 17.61,
"learning_rate": 1.2396531235691245e-05,
"loss": 0.0497,
"step": 7890
},
{
"epoch": 17.63,
"learning_rate": 1.2378980200706376e-05,
"loss": 0.0538,
"step": 7900
},
{
"epoch": 17.66,
"learning_rate": 1.236142139447002e-05,
"loss": 0.0606,
"step": 7910
},
{
"epoch": 17.68,
"learning_rate": 1.2343854874340324e-05,
"loss": 0.0633,
"step": 7920
},
{
"epoch": 17.7,
"learning_rate": 1.2326280697700632e-05,
"loss": 0.0562,
"step": 7930
},
{
"epoch": 17.72,
"learning_rate": 1.2308698921959306e-05,
"loss": 0.0545,
"step": 7940
},
{
"epoch": 17.75,
"learning_rate": 1.2291109604549525e-05,
"loss": 0.0502,
"step": 7950
},
{
"epoch": 17.77,
"learning_rate": 1.2273512802929107e-05,
"loss": 0.0442,
"step": 7960
},
{
"epoch": 17.79,
"learning_rate": 1.2255908574580311e-05,
"loss": 0.0564,
"step": 7970
},
{
"epoch": 17.81,
"learning_rate": 1.2238296977009672e-05,
"loss": 0.0526,
"step": 7980
},
{
"epoch": 17.83,
"learning_rate": 1.2220678067747785e-05,
"loss": 0.0612,
"step": 7990
},
{
"epoch": 17.86,
"learning_rate": 1.2203051904349128e-05,
"loss": 0.0516,
"step": 8000
}
],
"max_steps": 17920,
"num_train_epochs": 40,
"total_flos": 6.984706930455347e+16,
"trial_name": null,
"trial_params": null
}