Abhi4's picture
End of training
2d243fe
{
"best_metric": 0.9615316328342309,
"best_model_checkpoint": "swinv2-tiny-patch4-window8-256-finetuned-eurosat/checkpoint-3564",
"epoch": 2.9993688196928256,
"eval_steps": 500,
"global_step": 3564,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 1.4005602240896359e-06,
"loss": 1.3358,
"step": 10
},
{
"epoch": 0.02,
"learning_rate": 2.8011204481792718e-06,
"loss": 1.2686,
"step": 20
},
{
"epoch": 0.03,
"learning_rate": 4.2016806722689085e-06,
"loss": 1.1951,
"step": 30
},
{
"epoch": 0.03,
"learning_rate": 5.6022408963585436e-06,
"loss": 1.0686,
"step": 40
},
{
"epoch": 0.04,
"learning_rate": 7.0028011204481795e-06,
"loss": 1.0255,
"step": 50
},
{
"epoch": 0.05,
"learning_rate": 8.403361344537817e-06,
"loss": 0.9536,
"step": 60
},
{
"epoch": 0.06,
"learning_rate": 9.803921568627451e-06,
"loss": 0.8115,
"step": 70
},
{
"epoch": 0.07,
"learning_rate": 1.1204481792717087e-05,
"loss": 0.735,
"step": 80
},
{
"epoch": 0.08,
"learning_rate": 1.2605042016806723e-05,
"loss": 0.666,
"step": 90
},
{
"epoch": 0.08,
"learning_rate": 1.4005602240896359e-05,
"loss": 0.6396,
"step": 100
},
{
"epoch": 0.09,
"learning_rate": 1.5406162464985995e-05,
"loss": 0.5814,
"step": 110
},
{
"epoch": 0.1,
"learning_rate": 1.6806722689075634e-05,
"loss": 0.5461,
"step": 120
},
{
"epoch": 0.11,
"learning_rate": 1.8207282913165267e-05,
"loss": 0.5389,
"step": 130
},
{
"epoch": 0.12,
"learning_rate": 1.9607843137254903e-05,
"loss": 0.4734,
"step": 140
},
{
"epoch": 0.13,
"learning_rate": 2.100840336134454e-05,
"loss": 0.4689,
"step": 150
},
{
"epoch": 0.13,
"learning_rate": 2.2408963585434174e-05,
"loss": 0.4593,
"step": 160
},
{
"epoch": 0.14,
"learning_rate": 2.380952380952381e-05,
"loss": 0.4205,
"step": 170
},
{
"epoch": 0.15,
"learning_rate": 2.5210084033613446e-05,
"loss": 0.4288,
"step": 180
},
{
"epoch": 0.16,
"learning_rate": 2.6610644257703082e-05,
"loss": 0.4006,
"step": 190
},
{
"epoch": 0.17,
"learning_rate": 2.8011204481792718e-05,
"loss": 0.4362,
"step": 200
},
{
"epoch": 0.18,
"learning_rate": 2.9411764705882354e-05,
"loss": 0.3418,
"step": 210
},
{
"epoch": 0.19,
"learning_rate": 3.081232492997199e-05,
"loss": 0.3842,
"step": 220
},
{
"epoch": 0.19,
"learning_rate": 3.221288515406163e-05,
"loss": 0.3478,
"step": 230
},
{
"epoch": 0.2,
"learning_rate": 3.361344537815127e-05,
"loss": 0.3627,
"step": 240
},
{
"epoch": 0.21,
"learning_rate": 3.5014005602240894e-05,
"loss": 0.3705,
"step": 250
},
{
"epoch": 0.22,
"learning_rate": 3.641456582633053e-05,
"loss": 0.3953,
"step": 260
},
{
"epoch": 0.23,
"learning_rate": 3.7815126050420166e-05,
"loss": 0.38,
"step": 270
},
{
"epoch": 0.24,
"learning_rate": 3.9215686274509805e-05,
"loss": 0.3714,
"step": 280
},
{
"epoch": 0.24,
"learning_rate": 4.0616246498599444e-05,
"loss": 0.3445,
"step": 290
},
{
"epoch": 0.25,
"learning_rate": 4.201680672268908e-05,
"loss": 0.3509,
"step": 300
},
{
"epoch": 0.26,
"learning_rate": 4.3417366946778716e-05,
"loss": 0.3525,
"step": 310
},
{
"epoch": 0.27,
"learning_rate": 4.481792717086835e-05,
"loss": 0.3626,
"step": 320
},
{
"epoch": 0.28,
"learning_rate": 4.621848739495799e-05,
"loss": 0.3336,
"step": 330
},
{
"epoch": 0.29,
"learning_rate": 4.761904761904762e-05,
"loss": 0.3407,
"step": 340
},
{
"epoch": 0.29,
"learning_rate": 4.901960784313725e-05,
"loss": 0.3259,
"step": 350
},
{
"epoch": 0.3,
"learning_rate": 4.99532273152479e-05,
"loss": 0.3566,
"step": 360
},
{
"epoch": 0.31,
"learning_rate": 4.979731836607422e-05,
"loss": 0.3209,
"step": 370
},
{
"epoch": 0.32,
"learning_rate": 4.964140941690053e-05,
"loss": 0.3593,
"step": 380
},
{
"epoch": 0.33,
"learning_rate": 4.948550046772685e-05,
"loss": 0.3349,
"step": 390
},
{
"epoch": 0.34,
"learning_rate": 4.932959151855317e-05,
"loss": 0.3043,
"step": 400
},
{
"epoch": 0.35,
"learning_rate": 4.917368256937949e-05,
"loss": 0.3387,
"step": 410
},
{
"epoch": 0.35,
"learning_rate": 4.90177736202058e-05,
"loss": 0.3448,
"step": 420
},
{
"epoch": 0.36,
"learning_rate": 4.886186467103212e-05,
"loss": 0.3246,
"step": 430
},
{
"epoch": 0.37,
"learning_rate": 4.870595572185844e-05,
"loss": 0.3293,
"step": 440
},
{
"epoch": 0.38,
"learning_rate": 4.855004677268475e-05,
"loss": 0.3269,
"step": 450
},
{
"epoch": 0.39,
"learning_rate": 4.839413782351107e-05,
"loss": 0.2847,
"step": 460
},
{
"epoch": 0.4,
"learning_rate": 4.8238228874337385e-05,
"loss": 0.2986,
"step": 470
},
{
"epoch": 0.4,
"learning_rate": 4.8082319925163704e-05,
"loss": 0.3196,
"step": 480
},
{
"epoch": 0.41,
"learning_rate": 4.7926410975990024e-05,
"loss": 0.3191,
"step": 490
},
{
"epoch": 0.42,
"learning_rate": 4.777050202681634e-05,
"loss": 0.2793,
"step": 500
},
{
"epoch": 0.43,
"learning_rate": 4.761459307764266e-05,
"loss": 0.2967,
"step": 510
},
{
"epoch": 0.44,
"learning_rate": 4.7458684128468975e-05,
"loss": 0.3299,
"step": 520
},
{
"epoch": 0.45,
"learning_rate": 4.7302775179295295e-05,
"loss": 0.2825,
"step": 530
},
{
"epoch": 0.45,
"learning_rate": 4.7146866230121614e-05,
"loss": 0.3363,
"step": 540
},
{
"epoch": 0.46,
"learning_rate": 4.699095728094793e-05,
"loss": 0.2974,
"step": 550
},
{
"epoch": 0.47,
"learning_rate": 4.6835048331774246e-05,
"loss": 0.2847,
"step": 560
},
{
"epoch": 0.48,
"learning_rate": 4.667913938260056e-05,
"loss": 0.2943,
"step": 570
},
{
"epoch": 0.49,
"learning_rate": 4.652323043342688e-05,
"loss": 0.2425,
"step": 580
},
{
"epoch": 0.5,
"learning_rate": 4.63673214842532e-05,
"loss": 0.3261,
"step": 590
},
{
"epoch": 0.5,
"learning_rate": 4.621141253507952e-05,
"loss": 0.2602,
"step": 600
},
{
"epoch": 0.51,
"learning_rate": 4.605550358590584e-05,
"loss": 0.2779,
"step": 610
},
{
"epoch": 0.52,
"learning_rate": 4.589959463673215e-05,
"loss": 0.2807,
"step": 620
},
{
"epoch": 0.53,
"learning_rate": 4.574368568755847e-05,
"loss": 0.2705,
"step": 630
},
{
"epoch": 0.54,
"learning_rate": 4.558777673838479e-05,
"loss": 0.2932,
"step": 640
},
{
"epoch": 0.55,
"learning_rate": 4.54318677892111e-05,
"loss": 0.2662,
"step": 650
},
{
"epoch": 0.56,
"learning_rate": 4.527595884003742e-05,
"loss": 0.2452,
"step": 660
},
{
"epoch": 0.56,
"learning_rate": 4.512004989086373e-05,
"loss": 0.2549,
"step": 670
},
{
"epoch": 0.57,
"learning_rate": 4.496414094169005e-05,
"loss": 0.2792,
"step": 680
},
{
"epoch": 0.58,
"learning_rate": 4.480823199251637e-05,
"loss": 0.2771,
"step": 690
},
{
"epoch": 0.59,
"learning_rate": 4.465232304334269e-05,
"loss": 0.296,
"step": 700
},
{
"epoch": 0.6,
"learning_rate": 4.449641409416901e-05,
"loss": 0.2566,
"step": 710
},
{
"epoch": 0.61,
"learning_rate": 4.4340505144995324e-05,
"loss": 0.293,
"step": 720
},
{
"epoch": 0.61,
"learning_rate": 4.4184596195821643e-05,
"loss": 0.2751,
"step": 730
},
{
"epoch": 0.62,
"learning_rate": 4.4028687246647956e-05,
"loss": 0.2603,
"step": 740
},
{
"epoch": 0.63,
"learning_rate": 4.3872778297474276e-05,
"loss": 0.2984,
"step": 750
},
{
"epoch": 0.64,
"learning_rate": 4.3716869348300595e-05,
"loss": 0.2358,
"step": 760
},
{
"epoch": 0.65,
"learning_rate": 4.356096039912691e-05,
"loss": 0.2454,
"step": 770
},
{
"epoch": 0.66,
"learning_rate": 4.340505144995323e-05,
"loss": 0.2366,
"step": 780
},
{
"epoch": 0.66,
"learning_rate": 4.324914250077955e-05,
"loss": 0.2831,
"step": 790
},
{
"epoch": 0.67,
"learning_rate": 4.3093233551605866e-05,
"loss": 0.3191,
"step": 800
},
{
"epoch": 0.68,
"learning_rate": 4.2937324602432186e-05,
"loss": 0.2633,
"step": 810
},
{
"epoch": 0.69,
"learning_rate": 4.27814156532585e-05,
"loss": 0.2857,
"step": 820
},
{
"epoch": 0.7,
"learning_rate": 4.262550670408482e-05,
"loss": 0.2503,
"step": 830
},
{
"epoch": 0.71,
"learning_rate": 4.246959775491113e-05,
"loss": 0.2761,
"step": 840
},
{
"epoch": 0.72,
"learning_rate": 4.231368880573745e-05,
"loss": 0.2761,
"step": 850
},
{
"epoch": 0.72,
"learning_rate": 4.215777985656377e-05,
"loss": 0.2628,
"step": 860
},
{
"epoch": 0.73,
"learning_rate": 4.200187090739008e-05,
"loss": 0.2573,
"step": 870
},
{
"epoch": 0.74,
"learning_rate": 4.18459619582164e-05,
"loss": 0.2736,
"step": 880
},
{
"epoch": 0.75,
"learning_rate": 4.169005300904272e-05,
"loss": 0.2847,
"step": 890
},
{
"epoch": 0.76,
"learning_rate": 4.153414405986904e-05,
"loss": 0.2567,
"step": 900
},
{
"epoch": 0.77,
"learning_rate": 4.137823511069536e-05,
"loss": 0.256,
"step": 910
},
{
"epoch": 0.77,
"learning_rate": 4.122232616152167e-05,
"loss": 0.2705,
"step": 920
},
{
"epoch": 0.78,
"learning_rate": 4.106641721234799e-05,
"loss": 0.2551,
"step": 930
},
{
"epoch": 0.79,
"learning_rate": 4.0910508263174305e-05,
"loss": 0.2679,
"step": 940
},
{
"epoch": 0.8,
"learning_rate": 4.0754599314000624e-05,
"loss": 0.2729,
"step": 950
},
{
"epoch": 0.81,
"learning_rate": 4.0598690364826944e-05,
"loss": 0.2574,
"step": 960
},
{
"epoch": 0.82,
"learning_rate": 4.0442781415653257e-05,
"loss": 0.2625,
"step": 970
},
{
"epoch": 0.82,
"learning_rate": 4.028687246647958e-05,
"loss": 0.2673,
"step": 980
},
{
"epoch": 0.83,
"learning_rate": 4.0130963517305895e-05,
"loss": 0.2618,
"step": 990
},
{
"epoch": 0.84,
"learning_rate": 3.9975054568132215e-05,
"loss": 0.2675,
"step": 1000
},
{
"epoch": 0.85,
"learning_rate": 3.9819145618958534e-05,
"loss": 0.2439,
"step": 1010
},
{
"epoch": 0.86,
"learning_rate": 3.966323666978485e-05,
"loss": 0.2537,
"step": 1020
},
{
"epoch": 0.87,
"learning_rate": 3.9507327720611167e-05,
"loss": 0.2383,
"step": 1030
},
{
"epoch": 0.88,
"learning_rate": 3.935141877143748e-05,
"loss": 0.2545,
"step": 1040
},
{
"epoch": 0.88,
"learning_rate": 3.91955098222638e-05,
"loss": 0.2303,
"step": 1050
},
{
"epoch": 0.89,
"learning_rate": 3.903960087309012e-05,
"loss": 0.2312,
"step": 1060
},
{
"epoch": 0.9,
"learning_rate": 3.888369192391643e-05,
"loss": 0.2846,
"step": 1070
},
{
"epoch": 0.91,
"learning_rate": 3.872778297474276e-05,
"loss": 0.2839,
"step": 1080
},
{
"epoch": 0.92,
"learning_rate": 3.857187402556907e-05,
"loss": 0.2489,
"step": 1090
},
{
"epoch": 0.93,
"learning_rate": 3.841596507639539e-05,
"loss": 0.2635,
"step": 1100
},
{
"epoch": 0.93,
"learning_rate": 3.82600561272217e-05,
"loss": 0.2591,
"step": 1110
},
{
"epoch": 0.94,
"learning_rate": 3.810414717804802e-05,
"loss": 0.2256,
"step": 1120
},
{
"epoch": 0.95,
"learning_rate": 3.794823822887434e-05,
"loss": 0.2216,
"step": 1130
},
{
"epoch": 0.96,
"learning_rate": 3.7792329279700654e-05,
"loss": 0.2441,
"step": 1140
},
{
"epoch": 0.97,
"learning_rate": 3.763642033052697e-05,
"loss": 0.2335,
"step": 1150
},
{
"epoch": 0.98,
"learning_rate": 3.748051138135329e-05,
"loss": 0.2729,
"step": 1160
},
{
"epoch": 0.98,
"learning_rate": 3.7324602432179605e-05,
"loss": 0.2336,
"step": 1170
},
{
"epoch": 0.99,
"learning_rate": 3.716869348300593e-05,
"loss": 0.2183,
"step": 1180
},
{
"epoch": 1.0,
"eval_accuracy": 0.9541930520210689,
"eval_loss": 0.12830650806427002,
"eval_runtime": 310.446,
"eval_samples_per_second": 54.428,
"eval_steps_per_second": 1.704,
"step": 1188
},
{
"epoch": 1.0,
"learning_rate": 3.7012784533832244e-05,
"loss": 0.2144,
"step": 1190
},
{
"epoch": 1.01,
"learning_rate": 3.6856875584658564e-05,
"loss": 0.2249,
"step": 1200
},
{
"epoch": 1.02,
"learning_rate": 3.6700966635484876e-05,
"loss": 0.2721,
"step": 1210
},
{
"epoch": 1.03,
"learning_rate": 3.6545057686311196e-05,
"loss": 0.2508,
"step": 1220
},
{
"epoch": 1.04,
"learning_rate": 3.6389148737137515e-05,
"loss": 0.2781,
"step": 1230
},
{
"epoch": 1.04,
"learning_rate": 3.623323978796383e-05,
"loss": 0.233,
"step": 1240
},
{
"epoch": 1.05,
"learning_rate": 3.607733083879015e-05,
"loss": 0.2861,
"step": 1250
},
{
"epoch": 1.06,
"learning_rate": 3.592142188961646e-05,
"loss": 0.2512,
"step": 1260
},
{
"epoch": 1.07,
"learning_rate": 3.5765512940442786e-05,
"loss": 0.2532,
"step": 1270
},
{
"epoch": 1.08,
"learning_rate": 3.5609603991269106e-05,
"loss": 0.2321,
"step": 1280
},
{
"epoch": 1.09,
"learning_rate": 3.545369504209542e-05,
"loss": 0.2722,
"step": 1290
},
{
"epoch": 1.09,
"learning_rate": 3.529778609292174e-05,
"loss": 0.2788,
"step": 1300
},
{
"epoch": 1.1,
"learning_rate": 3.514187714374805e-05,
"loss": 0.2427,
"step": 1310
},
{
"epoch": 1.11,
"learning_rate": 3.498596819457437e-05,
"loss": 0.2556,
"step": 1320
},
{
"epoch": 1.12,
"learning_rate": 3.483005924540069e-05,
"loss": 0.2325,
"step": 1330
},
{
"epoch": 1.13,
"learning_rate": 3.4674150296227e-05,
"loss": 0.2431,
"step": 1340
},
{
"epoch": 1.14,
"learning_rate": 3.451824134705332e-05,
"loss": 0.2179,
"step": 1350
},
{
"epoch": 1.14,
"learning_rate": 3.4362332397879635e-05,
"loss": 0.2631,
"step": 1360
},
{
"epoch": 1.15,
"learning_rate": 3.420642344870596e-05,
"loss": 0.2296,
"step": 1370
},
{
"epoch": 1.16,
"learning_rate": 3.4050514499532273e-05,
"loss": 0.2425,
"step": 1380
},
{
"epoch": 1.17,
"learning_rate": 3.389460555035859e-05,
"loss": 0.2098,
"step": 1390
},
{
"epoch": 1.18,
"learning_rate": 3.373869660118491e-05,
"loss": 0.2488,
"step": 1400
},
{
"epoch": 1.19,
"learning_rate": 3.3582787652011225e-05,
"loss": 0.2532,
"step": 1410
},
{
"epoch": 1.2,
"learning_rate": 3.3426878702837545e-05,
"loss": 0.2135,
"step": 1420
},
{
"epoch": 1.2,
"learning_rate": 3.3270969753663864e-05,
"loss": 0.2263,
"step": 1430
},
{
"epoch": 1.21,
"learning_rate": 3.311506080449018e-05,
"loss": 0.2859,
"step": 1440
},
{
"epoch": 1.22,
"learning_rate": 3.2959151855316496e-05,
"loss": 0.2486,
"step": 1450
},
{
"epoch": 1.23,
"learning_rate": 3.280324290614281e-05,
"loss": 0.264,
"step": 1460
},
{
"epoch": 1.24,
"learning_rate": 3.2647333956969135e-05,
"loss": 0.2348,
"step": 1470
},
{
"epoch": 1.25,
"learning_rate": 3.249142500779545e-05,
"loss": 0.2529,
"step": 1480
},
{
"epoch": 1.25,
"learning_rate": 3.233551605862177e-05,
"loss": 0.2424,
"step": 1490
},
{
"epoch": 1.26,
"learning_rate": 3.217960710944809e-05,
"loss": 0.215,
"step": 1500
},
{
"epoch": 1.27,
"learning_rate": 3.20236981602744e-05,
"loss": 0.242,
"step": 1510
},
{
"epoch": 1.28,
"learning_rate": 3.186778921110072e-05,
"loss": 0.2328,
"step": 1520
},
{
"epoch": 1.29,
"learning_rate": 3.171188026192703e-05,
"loss": 0.2489,
"step": 1530
},
{
"epoch": 1.3,
"learning_rate": 3.155597131275335e-05,
"loss": 0.213,
"step": 1540
},
{
"epoch": 1.3,
"learning_rate": 3.140006236357967e-05,
"loss": 0.2347,
"step": 1550
},
{
"epoch": 1.31,
"learning_rate": 3.124415341440599e-05,
"loss": 0.2158,
"step": 1560
},
{
"epoch": 1.32,
"learning_rate": 3.108824446523231e-05,
"loss": 0.2464,
"step": 1570
},
{
"epoch": 1.33,
"learning_rate": 3.093233551605862e-05,
"loss": 0.2185,
"step": 1580
},
{
"epoch": 1.34,
"learning_rate": 3.077642656688494e-05,
"loss": 0.2299,
"step": 1590
},
{
"epoch": 1.35,
"learning_rate": 3.062051761771126e-05,
"loss": 0.2343,
"step": 1600
},
{
"epoch": 1.35,
"learning_rate": 3.0464608668537574e-05,
"loss": 0.2384,
"step": 1610
},
{
"epoch": 1.36,
"learning_rate": 3.0308699719363893e-05,
"loss": 0.2382,
"step": 1620
},
{
"epoch": 1.37,
"learning_rate": 3.015279077019021e-05,
"loss": 0.21,
"step": 1630
},
{
"epoch": 1.38,
"learning_rate": 2.9996881821016526e-05,
"loss": 0.2803,
"step": 1640
},
{
"epoch": 1.39,
"learning_rate": 2.984097287184284e-05,
"loss": 0.2284,
"step": 1650
},
{
"epoch": 1.4,
"learning_rate": 2.9685063922669164e-05,
"loss": 0.223,
"step": 1660
},
{
"epoch": 1.41,
"learning_rate": 2.952915497349548e-05,
"loss": 0.2582,
"step": 1670
},
{
"epoch": 1.41,
"learning_rate": 2.93732460243218e-05,
"loss": 0.2435,
"step": 1680
},
{
"epoch": 1.42,
"learning_rate": 2.9217337075148116e-05,
"loss": 0.2211,
"step": 1690
},
{
"epoch": 1.43,
"learning_rate": 2.9061428125974432e-05,
"loss": 0.2541,
"step": 1700
},
{
"epoch": 1.44,
"learning_rate": 2.8905519176800748e-05,
"loss": 0.2283,
"step": 1710
},
{
"epoch": 1.45,
"learning_rate": 2.8749610227627068e-05,
"loss": 0.2357,
"step": 1720
},
{
"epoch": 1.46,
"learning_rate": 2.8593701278453384e-05,
"loss": 0.2264,
"step": 1730
},
{
"epoch": 1.46,
"learning_rate": 2.84377923292797e-05,
"loss": 0.2504,
"step": 1740
},
{
"epoch": 1.47,
"learning_rate": 2.8281883380106016e-05,
"loss": 0.2338,
"step": 1750
},
{
"epoch": 1.48,
"learning_rate": 2.812597443093234e-05,
"loss": 0.2298,
"step": 1760
},
{
"epoch": 1.49,
"learning_rate": 2.7970065481758655e-05,
"loss": 0.2282,
"step": 1770
},
{
"epoch": 1.5,
"learning_rate": 2.7814156532584974e-05,
"loss": 0.2402,
"step": 1780
},
{
"epoch": 1.51,
"learning_rate": 2.765824758341129e-05,
"loss": 0.2186,
"step": 1790
},
{
"epoch": 1.51,
"learning_rate": 2.7502338634237607e-05,
"loss": 0.2131,
"step": 1800
},
{
"epoch": 1.52,
"learning_rate": 2.7346429685063923e-05,
"loss": 0.2254,
"step": 1810
},
{
"epoch": 1.53,
"learning_rate": 2.719052073589024e-05,
"loss": 0.2042,
"step": 1820
},
{
"epoch": 1.54,
"learning_rate": 2.7034611786716558e-05,
"loss": 0.2537,
"step": 1830
},
{
"epoch": 1.55,
"learning_rate": 2.6878702837542874e-05,
"loss": 0.2386,
"step": 1840
},
{
"epoch": 1.56,
"learning_rate": 2.6722793888369197e-05,
"loss": 0.2017,
"step": 1850
},
{
"epoch": 1.57,
"learning_rate": 2.6566884939195513e-05,
"loss": 0.1893,
"step": 1860
},
{
"epoch": 1.57,
"learning_rate": 2.641097599002183e-05,
"loss": 0.2256,
"step": 1870
},
{
"epoch": 1.58,
"learning_rate": 2.6255067040848145e-05,
"loss": 0.2499,
"step": 1880
},
{
"epoch": 1.59,
"learning_rate": 2.6099158091674465e-05,
"loss": 0.2405,
"step": 1890
},
{
"epoch": 1.6,
"learning_rate": 2.594324914250078e-05,
"loss": 0.2505,
"step": 1900
},
{
"epoch": 1.61,
"learning_rate": 2.5787340193327097e-05,
"loss": 0.2102,
"step": 1910
},
{
"epoch": 1.62,
"learning_rate": 2.5631431244153413e-05,
"loss": 0.2263,
"step": 1920
},
{
"epoch": 1.62,
"learning_rate": 2.5475522294979733e-05,
"loss": 0.2314,
"step": 1930
},
{
"epoch": 1.63,
"learning_rate": 2.531961334580605e-05,
"loss": 0.2166,
"step": 1940
},
{
"epoch": 1.64,
"learning_rate": 2.516370439663237e-05,
"loss": 0.2281,
"step": 1950
},
{
"epoch": 1.65,
"learning_rate": 2.5007795447458688e-05,
"loss": 0.2122,
"step": 1960
},
{
"epoch": 1.66,
"learning_rate": 2.4851886498285e-05,
"loss": 0.2381,
"step": 1970
},
{
"epoch": 1.67,
"learning_rate": 2.469597754911132e-05,
"loss": 0.2209,
"step": 1980
},
{
"epoch": 1.67,
"learning_rate": 2.454006859993764e-05,
"loss": 0.2274,
"step": 1990
},
{
"epoch": 1.68,
"learning_rate": 2.4384159650763955e-05,
"loss": 0.2267,
"step": 2000
},
{
"epoch": 1.69,
"learning_rate": 2.422825070159027e-05,
"loss": 0.2145,
"step": 2010
},
{
"epoch": 1.7,
"learning_rate": 2.407234175241659e-05,
"loss": 0.2255,
"step": 2020
},
{
"epoch": 1.71,
"learning_rate": 2.3916432803242907e-05,
"loss": 0.2436,
"step": 2030
},
{
"epoch": 1.72,
"learning_rate": 2.3760523854069226e-05,
"loss": 0.2258,
"step": 2040
},
{
"epoch": 1.73,
"learning_rate": 2.3604614904895543e-05,
"loss": 0.2201,
"step": 2050
},
{
"epoch": 1.73,
"learning_rate": 2.344870595572186e-05,
"loss": 0.211,
"step": 2060
},
{
"epoch": 1.74,
"learning_rate": 2.3292797006548178e-05,
"loss": 0.2884,
"step": 2070
},
{
"epoch": 1.75,
"learning_rate": 2.3136888057374494e-05,
"loss": 0.2147,
"step": 2080
},
{
"epoch": 1.76,
"learning_rate": 2.298097910820081e-05,
"loss": 0.2356,
"step": 2090
},
{
"epoch": 1.77,
"learning_rate": 2.282507015902713e-05,
"loss": 0.214,
"step": 2100
},
{
"epoch": 1.78,
"learning_rate": 2.2669161209853446e-05,
"loss": 0.221,
"step": 2110
},
{
"epoch": 1.78,
"learning_rate": 2.2513252260679765e-05,
"loss": 0.2366,
"step": 2120
},
{
"epoch": 1.79,
"learning_rate": 2.235734331150608e-05,
"loss": 0.2046,
"step": 2130
},
{
"epoch": 1.8,
"learning_rate": 2.2201434362332397e-05,
"loss": 0.2113,
"step": 2140
},
{
"epoch": 1.81,
"learning_rate": 2.2045525413158717e-05,
"loss": 0.217,
"step": 2150
},
{
"epoch": 1.82,
"learning_rate": 2.1889616463985033e-05,
"loss": 0.1887,
"step": 2160
},
{
"epoch": 1.83,
"learning_rate": 2.1733707514811352e-05,
"loss": 0.1934,
"step": 2170
},
{
"epoch": 1.83,
"learning_rate": 2.157779856563767e-05,
"loss": 0.2295,
"step": 2180
},
{
"epoch": 1.84,
"learning_rate": 2.1421889616463985e-05,
"loss": 0.2161,
"step": 2190
},
{
"epoch": 1.85,
"learning_rate": 2.1265980667290304e-05,
"loss": 0.2304,
"step": 2200
},
{
"epoch": 1.86,
"learning_rate": 2.111007171811662e-05,
"loss": 0.2535,
"step": 2210
},
{
"epoch": 1.87,
"learning_rate": 2.095416276894294e-05,
"loss": 0.2198,
"step": 2220
},
{
"epoch": 1.88,
"learning_rate": 2.0798253819769256e-05,
"loss": 0.2422,
"step": 2230
},
{
"epoch": 1.89,
"learning_rate": 2.0642344870595572e-05,
"loss": 0.2231,
"step": 2240
},
{
"epoch": 1.89,
"learning_rate": 2.048643592142189e-05,
"loss": 0.2184,
"step": 2250
},
{
"epoch": 1.9,
"learning_rate": 2.0330526972248207e-05,
"loss": 0.2294,
"step": 2260
},
{
"epoch": 1.91,
"learning_rate": 2.0174618023074527e-05,
"loss": 0.1991,
"step": 2270
},
{
"epoch": 1.92,
"learning_rate": 2.0018709073900843e-05,
"loss": 0.2187,
"step": 2280
},
{
"epoch": 1.93,
"learning_rate": 1.986280012472716e-05,
"loss": 0.1991,
"step": 2290
},
{
"epoch": 1.94,
"learning_rate": 1.970689117555348e-05,
"loss": 0.1823,
"step": 2300
},
{
"epoch": 1.94,
"learning_rate": 1.9550982226379798e-05,
"loss": 0.237,
"step": 2310
},
{
"epoch": 1.95,
"learning_rate": 1.9395073277206114e-05,
"loss": 0.2189,
"step": 2320
},
{
"epoch": 1.96,
"learning_rate": 1.923916432803243e-05,
"loss": 0.2173,
"step": 2330
},
{
"epoch": 1.97,
"learning_rate": 1.9083255378858746e-05,
"loss": 0.2319,
"step": 2340
},
{
"epoch": 1.98,
"learning_rate": 1.8927346429685062e-05,
"loss": 0.237,
"step": 2350
},
{
"epoch": 1.99,
"learning_rate": 1.8771437480511385e-05,
"loss": 0.2368,
"step": 2360
},
{
"epoch": 1.99,
"learning_rate": 1.86155285313377e-05,
"loss": 0.2099,
"step": 2370
},
{
"epoch": 2.0,
"eval_accuracy": 0.9570337929810026,
"eval_loss": 0.11882077902555466,
"eval_runtime": 304.3918,
"eval_samples_per_second": 55.511,
"eval_steps_per_second": 1.738,
"step": 2376
},
{
"epoch": 2.0,
"learning_rate": 1.8459619582164017e-05,
"loss": 0.2455,
"step": 2380
},
{
"epoch": 2.01,
"learning_rate": 1.8303710632990333e-05,
"loss": 0.2108,
"step": 2390
},
{
"epoch": 2.02,
"learning_rate": 1.814780168381665e-05,
"loss": 0.1959,
"step": 2400
},
{
"epoch": 2.03,
"learning_rate": 1.799189273464297e-05,
"loss": 0.2015,
"step": 2410
},
{
"epoch": 2.04,
"learning_rate": 1.783598378546929e-05,
"loss": 0.1989,
"step": 2420
},
{
"epoch": 2.05,
"learning_rate": 1.7680074836295604e-05,
"loss": 0.1953,
"step": 2430
},
{
"epoch": 2.05,
"learning_rate": 1.752416588712192e-05,
"loss": 0.2103,
"step": 2440
},
{
"epoch": 2.06,
"learning_rate": 1.7368256937948237e-05,
"loss": 0.2186,
"step": 2450
},
{
"epoch": 2.07,
"learning_rate": 1.7212347988774556e-05,
"loss": 0.2499,
"step": 2460
},
{
"epoch": 2.08,
"learning_rate": 1.7056439039600876e-05,
"loss": 0.234,
"step": 2470
},
{
"epoch": 2.09,
"learning_rate": 1.690053009042719e-05,
"loss": 0.2212,
"step": 2480
},
{
"epoch": 2.1,
"learning_rate": 1.6744621141253508e-05,
"loss": 0.2011,
"step": 2490
},
{
"epoch": 2.1,
"learning_rate": 1.6588712192079824e-05,
"loss": 0.2273,
"step": 2500
},
{
"epoch": 2.11,
"learning_rate": 1.6432803242906143e-05,
"loss": 0.2124,
"step": 2510
},
{
"epoch": 2.12,
"learning_rate": 1.6276894293732463e-05,
"loss": 0.1946,
"step": 2520
},
{
"epoch": 2.13,
"learning_rate": 1.612098534455878e-05,
"loss": 0.2414,
"step": 2530
},
{
"epoch": 2.14,
"learning_rate": 1.5965076395385095e-05,
"loss": 0.2319,
"step": 2540
},
{
"epoch": 2.15,
"learning_rate": 1.580916744621141e-05,
"loss": 0.1864,
"step": 2550
},
{
"epoch": 2.15,
"learning_rate": 1.565325849703773e-05,
"loss": 0.2163,
"step": 2560
},
{
"epoch": 2.16,
"learning_rate": 1.549734954786405e-05,
"loss": 0.223,
"step": 2570
},
{
"epoch": 2.17,
"learning_rate": 1.5341440598690366e-05,
"loss": 0.202,
"step": 2580
},
{
"epoch": 2.18,
"learning_rate": 1.5185531649516682e-05,
"loss": 0.2173,
"step": 2590
},
{
"epoch": 2.19,
"learning_rate": 1.5029622700343002e-05,
"loss": 0.2205,
"step": 2600
},
{
"epoch": 2.2,
"learning_rate": 1.487371375116932e-05,
"loss": 0.2409,
"step": 2610
},
{
"epoch": 2.2,
"learning_rate": 1.4717804801995635e-05,
"loss": 0.2188,
"step": 2620
},
{
"epoch": 2.21,
"learning_rate": 1.4561895852821952e-05,
"loss": 0.2067,
"step": 2630
},
{
"epoch": 2.22,
"learning_rate": 1.440598690364827e-05,
"loss": 0.2117,
"step": 2640
},
{
"epoch": 2.23,
"learning_rate": 1.4250077954474589e-05,
"loss": 0.2101,
"step": 2650
},
{
"epoch": 2.24,
"learning_rate": 1.4094169005300905e-05,
"loss": 0.222,
"step": 2660
},
{
"epoch": 2.25,
"learning_rate": 1.3938260056127223e-05,
"loss": 0.2403,
"step": 2670
},
{
"epoch": 2.26,
"learning_rate": 1.3782351106953539e-05,
"loss": 0.2037,
"step": 2680
},
{
"epoch": 2.26,
"learning_rate": 1.3626442157779856e-05,
"loss": 0.2064,
"step": 2690
},
{
"epoch": 2.27,
"learning_rate": 1.3470533208606176e-05,
"loss": 0.2301,
"step": 2700
},
{
"epoch": 2.28,
"learning_rate": 1.3314624259432492e-05,
"loss": 0.1992,
"step": 2710
},
{
"epoch": 2.29,
"learning_rate": 1.315871531025881e-05,
"loss": 0.1956,
"step": 2720
},
{
"epoch": 2.3,
"learning_rate": 1.3002806361085126e-05,
"loss": 0.1989,
"step": 2730
},
{
"epoch": 2.31,
"learning_rate": 1.2846897411911444e-05,
"loss": 0.2183,
"step": 2740
},
{
"epoch": 2.31,
"learning_rate": 1.2690988462737763e-05,
"loss": 0.2311,
"step": 2750
},
{
"epoch": 2.32,
"learning_rate": 1.253507951356408e-05,
"loss": 0.2116,
"step": 2760
},
{
"epoch": 2.33,
"learning_rate": 1.2379170564390397e-05,
"loss": 0.1907,
"step": 2770
},
{
"epoch": 2.34,
"learning_rate": 1.2223261615216713e-05,
"loss": 0.2253,
"step": 2780
},
{
"epoch": 2.35,
"learning_rate": 1.2067352666043031e-05,
"loss": 0.2089,
"step": 2790
},
{
"epoch": 2.36,
"learning_rate": 1.1911443716869349e-05,
"loss": 0.2271,
"step": 2800
},
{
"epoch": 2.36,
"learning_rate": 1.1755534767695666e-05,
"loss": 0.2093,
"step": 2810
},
{
"epoch": 2.37,
"learning_rate": 1.1599625818521984e-05,
"loss": 0.2115,
"step": 2820
},
{
"epoch": 2.38,
"learning_rate": 1.14437168693483e-05,
"loss": 0.2228,
"step": 2830
},
{
"epoch": 2.39,
"learning_rate": 1.1287807920174618e-05,
"loss": 0.2002,
"step": 2840
},
{
"epoch": 2.4,
"learning_rate": 1.1131898971000936e-05,
"loss": 0.212,
"step": 2850
},
{
"epoch": 2.41,
"learning_rate": 1.0975990021827254e-05,
"loss": 0.2017,
"step": 2860
},
{
"epoch": 2.42,
"learning_rate": 1.0820081072653571e-05,
"loss": 0.2283,
"step": 2870
},
{
"epoch": 2.42,
"learning_rate": 1.066417212347989e-05,
"loss": 0.215,
"step": 2880
},
{
"epoch": 2.43,
"learning_rate": 1.0508263174306205e-05,
"loss": 0.2146,
"step": 2890
},
{
"epoch": 2.44,
"learning_rate": 1.0352354225132523e-05,
"loss": 0.2017,
"step": 2900
},
{
"epoch": 2.45,
"learning_rate": 1.019644527595884e-05,
"loss": 0.2078,
"step": 2910
},
{
"epoch": 2.46,
"learning_rate": 1.0040536326785157e-05,
"loss": 0.2092,
"step": 2920
},
{
"epoch": 2.47,
"learning_rate": 9.884627377611476e-06,
"loss": 0.1804,
"step": 2930
},
{
"epoch": 2.47,
"learning_rate": 9.728718428437792e-06,
"loss": 0.1877,
"step": 2940
},
{
"epoch": 2.48,
"learning_rate": 9.57280947926411e-06,
"loss": 0.216,
"step": 2950
},
{
"epoch": 2.49,
"learning_rate": 9.416900530090428e-06,
"loss": 0.2339,
"step": 2960
},
{
"epoch": 2.5,
"learning_rate": 9.260991580916744e-06,
"loss": 0.2199,
"step": 2970
},
{
"epoch": 2.51,
"learning_rate": 9.105082631743064e-06,
"loss": 0.2094,
"step": 2980
},
{
"epoch": 2.52,
"learning_rate": 8.94917368256938e-06,
"loss": 0.2139,
"step": 2990
},
{
"epoch": 2.52,
"learning_rate": 8.793264733395697e-06,
"loss": 0.2141,
"step": 3000
},
{
"epoch": 2.53,
"learning_rate": 8.637355784222015e-06,
"loss": 0.1935,
"step": 3010
},
{
"epoch": 2.54,
"learning_rate": 8.481446835048331e-06,
"loss": 0.1691,
"step": 3020
},
{
"epoch": 2.55,
"learning_rate": 8.32553788587465e-06,
"loss": 0.2109,
"step": 3030
},
{
"epoch": 2.56,
"learning_rate": 8.169628936700967e-06,
"loss": 0.2142,
"step": 3040
},
{
"epoch": 2.57,
"learning_rate": 8.013719987527285e-06,
"loss": 0.221,
"step": 3050
},
{
"epoch": 2.58,
"learning_rate": 7.857811038353602e-06,
"loss": 0.1718,
"step": 3060
},
{
"epoch": 2.58,
"learning_rate": 7.701902089179918e-06,
"loss": 0.2,
"step": 3070
},
{
"epoch": 2.59,
"learning_rate": 7.545993140006237e-06,
"loss": 0.1856,
"step": 3080
},
{
"epoch": 2.6,
"learning_rate": 7.390084190832554e-06,
"loss": 0.1998,
"step": 3090
},
{
"epoch": 2.61,
"learning_rate": 7.234175241658872e-06,
"loss": 0.1969,
"step": 3100
},
{
"epoch": 2.62,
"learning_rate": 7.078266292485189e-06,
"loss": 0.1886,
"step": 3110
},
{
"epoch": 2.63,
"learning_rate": 6.922357343311506e-06,
"loss": 0.1732,
"step": 3120
},
{
"epoch": 2.63,
"learning_rate": 6.766448394137824e-06,
"loss": 0.2157,
"step": 3130
},
{
"epoch": 2.64,
"learning_rate": 6.610539444964141e-06,
"loss": 0.214,
"step": 3140
},
{
"epoch": 2.65,
"learning_rate": 6.454630495790459e-06,
"loss": 0.1919,
"step": 3150
},
{
"epoch": 2.66,
"learning_rate": 6.298721546616776e-06,
"loss": 0.2152,
"step": 3160
},
{
"epoch": 2.67,
"learning_rate": 6.142812597443094e-06,
"loss": 0.1989,
"step": 3170
},
{
"epoch": 2.68,
"learning_rate": 5.9869036482694114e-06,
"loss": 0.1901,
"step": 3180
},
{
"epoch": 2.68,
"learning_rate": 5.830994699095728e-06,
"loss": 0.2068,
"step": 3190
},
{
"epoch": 2.69,
"learning_rate": 5.675085749922046e-06,
"loss": 0.1901,
"step": 3200
},
{
"epoch": 2.7,
"learning_rate": 5.519176800748363e-06,
"loss": 0.2091,
"step": 3210
},
{
"epoch": 2.71,
"learning_rate": 5.363267851574681e-06,
"loss": 0.206,
"step": 3220
},
{
"epoch": 2.72,
"learning_rate": 5.207358902400998e-06,
"loss": 0.2135,
"step": 3230
},
{
"epoch": 2.73,
"learning_rate": 5.0514499532273156e-06,
"loss": 0.2252,
"step": 3240
},
{
"epoch": 2.74,
"learning_rate": 4.895541004053633e-06,
"loss": 0.1835,
"step": 3250
},
{
"epoch": 2.74,
"learning_rate": 4.73963205487995e-06,
"loss": 0.1613,
"step": 3260
},
{
"epoch": 2.75,
"learning_rate": 4.583723105706267e-06,
"loss": 0.2149,
"step": 3270
},
{
"epoch": 2.76,
"learning_rate": 4.427814156532585e-06,
"loss": 0.1654,
"step": 3280
},
{
"epoch": 2.77,
"learning_rate": 4.271905207358903e-06,
"loss": 0.2135,
"step": 3290
},
{
"epoch": 2.78,
"learning_rate": 4.1159962581852205e-06,
"loss": 0.1936,
"step": 3300
},
{
"epoch": 2.79,
"learning_rate": 3.9600873090115375e-06,
"loss": 0.2129,
"step": 3310
},
{
"epoch": 2.79,
"learning_rate": 3.8041783598378544e-06,
"loss": 0.2319,
"step": 3320
},
{
"epoch": 2.8,
"learning_rate": 3.648269410664172e-06,
"loss": 0.237,
"step": 3330
},
{
"epoch": 2.81,
"learning_rate": 3.4923604614904895e-06,
"loss": 0.2078,
"step": 3340
},
{
"epoch": 2.82,
"learning_rate": 3.3364515123168073e-06,
"loss": 0.1914,
"step": 3350
},
{
"epoch": 2.83,
"learning_rate": 3.1805425631431246e-06,
"loss": 0.1764,
"step": 3360
},
{
"epoch": 2.84,
"learning_rate": 3.024633613969442e-06,
"loss": 0.2198,
"step": 3370
},
{
"epoch": 2.84,
"learning_rate": 2.8687246647957593e-06,
"loss": 0.2305,
"step": 3380
},
{
"epoch": 2.85,
"learning_rate": 2.7128157156220767e-06,
"loss": 0.1813,
"step": 3390
},
{
"epoch": 2.86,
"learning_rate": 2.556906766448394e-06,
"loss": 0.174,
"step": 3400
},
{
"epoch": 2.87,
"learning_rate": 2.400997817274712e-06,
"loss": 0.1667,
"step": 3410
},
{
"epoch": 2.88,
"learning_rate": 2.2450888681010288e-06,
"loss": 0.2069,
"step": 3420
},
{
"epoch": 2.89,
"learning_rate": 2.0891799189273465e-06,
"loss": 0.2115,
"step": 3430
},
{
"epoch": 2.9,
"learning_rate": 1.933270969753664e-06,
"loss": 0.2187,
"step": 3440
},
{
"epoch": 2.9,
"learning_rate": 1.7773620205799812e-06,
"loss": 0.2239,
"step": 3450
},
{
"epoch": 2.91,
"learning_rate": 1.6214530714062988e-06,
"loss": 0.2191,
"step": 3460
},
{
"epoch": 2.92,
"learning_rate": 1.4655441222326162e-06,
"loss": 0.194,
"step": 3470
},
{
"epoch": 2.93,
"learning_rate": 1.3096351730589337e-06,
"loss": 0.1813,
"step": 3480
},
{
"epoch": 2.94,
"learning_rate": 1.153726223885251e-06,
"loss": 0.2153,
"step": 3490
},
{
"epoch": 2.95,
"learning_rate": 9.978172747115684e-07,
"loss": 0.1929,
"step": 3500
},
{
"epoch": 2.95,
"learning_rate": 8.41908325537886e-07,
"loss": 0.2387,
"step": 3510
},
{
"epoch": 2.96,
"learning_rate": 6.859993763642034e-07,
"loss": 0.1916,
"step": 3520
},
{
"epoch": 2.97,
"learning_rate": 5.300904271905207e-07,
"loss": 0.2146,
"step": 3530
},
{
"epoch": 2.98,
"learning_rate": 3.741814780168382e-07,
"loss": 0.1924,
"step": 3540
},
{
"epoch": 2.99,
"learning_rate": 2.182725288431556e-07,
"loss": 0.1937,
"step": 3550
},
{
"epoch": 3.0,
"learning_rate": 6.236357966947303e-08,
"loss": 0.1668,
"step": 3560
},
{
"epoch": 3.0,
"eval_accuracy": 0.9615316328342309,
"eval_loss": 0.1066935583949089,
"eval_runtime": 302.9553,
"eval_samples_per_second": 55.774,
"eval_steps_per_second": 1.746,
"step": 3564
},
{
"epoch": 3.0,
"step": 3564,
"total_flos": 1.484110125228884e+19,
"train_loss": 0.2671308586192051,
"train_runtime": 11690.3858,
"train_samples_per_second": 39.025,
"train_steps_per_second": 0.305
}
],
"logging_steps": 10,
"max_steps": 3564,
"num_train_epochs": 3,
"save_steps": 500,
"total_flos": 1.484110125228884e+19,
"trial_name": null,
"trial_params": null
}