Sunsimiao-Qwen1.5-1.8B-full-2.0 / trainer_state.json
Tsumugii's picture
initial commit
d87bb35
raw
history blame contribute delete
No virus
83.8 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.9984,
"eval_steps": 500,
"global_step": 2811,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"grad_norm": 8.1875,
"learning_rate": 1.25e-05,
"loss": 1.5548,
"step": 5
},
{
"epoch": 0.01,
"grad_norm": 7.59375,
"learning_rate": 2.5e-05,
"loss": 1.5637,
"step": 10
},
{
"epoch": 0.02,
"grad_norm": 7.90625,
"learning_rate": 3.7500000000000003e-05,
"loss": 1.5441,
"step": 15
},
{
"epoch": 0.02,
"grad_norm": 7.5,
"learning_rate": 5e-05,
"loss": 1.5212,
"step": 20
},
{
"epoch": 0.03,
"grad_norm": 6.90625,
"learning_rate": 4.999960406039296e-05,
"loss": 1.3853,
"step": 25
},
{
"epoch": 0.03,
"grad_norm": 6.59375,
"learning_rate": 4.9998416254113275e-05,
"loss": 1.3669,
"step": 30
},
{
"epoch": 0.04,
"grad_norm": 4.6875,
"learning_rate": 4.9996436618784923e-05,
"loss": 1.339,
"step": 35
},
{
"epoch": 0.04,
"grad_norm": 5.125,
"learning_rate": 4.9993665217113186e-05,
"loss": 1.4381,
"step": 40
},
{
"epoch": 0.05,
"grad_norm": 6.4375,
"learning_rate": 4.9990102136882665e-05,
"loss": 1.3781,
"step": 45
},
{
"epoch": 0.05,
"grad_norm": 4.875,
"learning_rate": 4.998574749095454e-05,
"loss": 1.3353,
"step": 50
},
{
"epoch": 0.06,
"grad_norm": 5.90625,
"learning_rate": 4.998060141726295e-05,
"loss": 1.326,
"step": 55
},
{
"epoch": 0.06,
"grad_norm": 5.125,
"learning_rate": 4.997466407881065e-05,
"loss": 1.314,
"step": 60
},
{
"epoch": 0.07,
"grad_norm": 5.21875,
"learning_rate": 4.996793566366384e-05,
"loss": 1.2821,
"step": 65
},
{
"epoch": 0.07,
"grad_norm": 5.46875,
"learning_rate": 4.99604163849462e-05,
"loss": 1.3627,
"step": 70
},
{
"epoch": 0.08,
"grad_norm": 6.25,
"learning_rate": 4.995210648083214e-05,
"loss": 1.3492,
"step": 75
},
{
"epoch": 0.09,
"grad_norm": 4.59375,
"learning_rate": 4.994300621453928e-05,
"loss": 1.3351,
"step": 80
},
{
"epoch": 0.09,
"grad_norm": 5.53125,
"learning_rate": 4.99331158743201e-05,
"loss": 1.3716,
"step": 85
},
{
"epoch": 0.1,
"grad_norm": 4.78125,
"learning_rate": 4.992243577345278e-05,
"loss": 1.3064,
"step": 90
},
{
"epoch": 0.1,
"grad_norm": 6.0625,
"learning_rate": 4.9910966250231324e-05,
"loss": 1.3729,
"step": 95
},
{
"epoch": 0.11,
"grad_norm": 5.59375,
"learning_rate": 4.989870766795481e-05,
"loss": 1.3587,
"step": 100
},
{
"epoch": 0.11,
"grad_norm": 6.09375,
"learning_rate": 4.988566041491589e-05,
"loss": 1.3231,
"step": 105
},
{
"epoch": 0.12,
"grad_norm": 6.09375,
"learning_rate": 4.987182490438852e-05,
"loss": 1.2753,
"step": 110
},
{
"epoch": 0.12,
"grad_norm": 6.09375,
"learning_rate": 4.985720157461482e-05,
"loss": 1.2592,
"step": 115
},
{
"epoch": 0.13,
"grad_norm": 5.1875,
"learning_rate": 4.984179088879122e-05,
"loss": 1.3175,
"step": 120
},
{
"epoch": 0.13,
"grad_norm": 6.59375,
"learning_rate": 4.9825593335053796e-05,
"loss": 1.4169,
"step": 125
},
{
"epoch": 0.14,
"grad_norm": 5.46875,
"learning_rate": 4.9808609426462794e-05,
"loss": 1.3497,
"step": 130
},
{
"epoch": 0.14,
"grad_norm": 6.6875,
"learning_rate": 4.979083970098638e-05,
"loss": 1.4249,
"step": 135
},
{
"epoch": 0.15,
"grad_norm": 5.40625,
"learning_rate": 4.977228472148361e-05,
"loss": 1.3558,
"step": 140
},
{
"epoch": 0.15,
"grad_norm": 6.53125,
"learning_rate": 4.9752945075686574e-05,
"loss": 1.371,
"step": 145
},
{
"epoch": 0.16,
"grad_norm": 5.90625,
"learning_rate": 4.973282137618182e-05,
"loss": 1.2899,
"step": 150
},
{
"epoch": 0.17,
"grad_norm": 7.96875,
"learning_rate": 4.9711914260390924e-05,
"loss": 1.3947,
"step": 155
},
{
"epoch": 0.17,
"grad_norm": 5.625,
"learning_rate": 4.96902243905503e-05,
"loss": 1.2762,
"step": 160
},
{
"epoch": 0.18,
"grad_norm": 9.125,
"learning_rate": 4.966775245369024e-05,
"loss": 1.392,
"step": 165
},
{
"epoch": 0.18,
"grad_norm": 6.1875,
"learning_rate": 4.964449916161312e-05,
"loss": 1.3061,
"step": 170
},
{
"epoch": 0.19,
"grad_norm": 5.53125,
"learning_rate": 4.9620465250870886e-05,
"loss": 1.2861,
"step": 175
},
{
"epoch": 0.19,
"grad_norm": 5.71875,
"learning_rate": 4.959565148274172e-05,
"loss": 1.3093,
"step": 180
},
{
"epoch": 0.2,
"grad_norm": 6.5,
"learning_rate": 4.95700586432059e-05,
"loss": 1.2894,
"step": 185
},
{
"epoch": 0.2,
"grad_norm": 6.3125,
"learning_rate": 4.954368754292095e-05,
"loss": 1.2801,
"step": 190
},
{
"epoch": 0.21,
"grad_norm": 4.75,
"learning_rate": 4.951653901719589e-05,
"loss": 1.3643,
"step": 195
},
{
"epoch": 0.21,
"grad_norm": 5.125,
"learning_rate": 4.9488613925964886e-05,
"loss": 1.2965,
"step": 200
},
{
"epoch": 0.22,
"grad_norm": 5.125,
"learning_rate": 4.945991315375987e-05,
"loss": 1.3055,
"step": 205
},
{
"epoch": 0.22,
"grad_norm": 4.53125,
"learning_rate": 4.943043760968267e-05,
"loss": 1.2909,
"step": 210
},
{
"epoch": 0.23,
"grad_norm": 6.8125,
"learning_rate": 4.940018822737609e-05,
"loss": 1.3486,
"step": 215
},
{
"epoch": 0.23,
"grad_norm": 6.625,
"learning_rate": 4.9369165964994434e-05,
"loss": 1.3847,
"step": 220
},
{
"epoch": 0.24,
"grad_norm": 5.125,
"learning_rate": 4.9337371805173074e-05,
"loss": 1.3065,
"step": 225
},
{
"epoch": 0.25,
"grad_norm": 5.75,
"learning_rate": 4.93048067549974e-05,
"loss": 1.309,
"step": 230
},
{
"epoch": 0.25,
"grad_norm": 5.21875,
"learning_rate": 4.9271471845970854e-05,
"loss": 1.2953,
"step": 235
},
{
"epoch": 0.26,
"grad_norm": 5.59375,
"learning_rate": 4.9237368133982286e-05,
"loss": 1.3291,
"step": 240
},
{
"epoch": 0.26,
"grad_norm": 6.25,
"learning_rate": 4.920249669927255e-05,
"loss": 1.2906,
"step": 245
},
{
"epoch": 0.27,
"grad_norm": 5.46875,
"learning_rate": 4.916685864640021e-05,
"loss": 1.3583,
"step": 250
},
{
"epoch": 0.27,
"grad_norm": 5.5625,
"learning_rate": 4.913045510420658e-05,
"loss": 1.2911,
"step": 255
},
{
"epoch": 0.28,
"grad_norm": 4.46875,
"learning_rate": 4.9093287225780004e-05,
"loss": 1.2865,
"step": 260
},
{
"epoch": 0.28,
"grad_norm": 4.9375,
"learning_rate": 4.905535618841931e-05,
"loss": 1.3438,
"step": 265
},
{
"epoch": 0.29,
"grad_norm": 6.53125,
"learning_rate": 4.901666319359649e-05,
"loss": 1.3012,
"step": 270
},
{
"epoch": 0.29,
"grad_norm": 5.84375,
"learning_rate": 4.897720946691867e-05,
"loss": 1.313,
"step": 275
},
{
"epoch": 0.3,
"grad_norm": 4.71875,
"learning_rate": 4.893699625808931e-05,
"loss": 1.3882,
"step": 280
},
{
"epoch": 0.3,
"grad_norm": 6.40625,
"learning_rate": 4.889602484086857e-05,
"loss": 1.2409,
"step": 285
},
{
"epoch": 0.31,
"grad_norm": 5.46875,
"learning_rate": 4.8854296513032996e-05,
"loss": 1.3014,
"step": 290
},
{
"epoch": 0.31,
"grad_norm": 5.125,
"learning_rate": 4.8811812596334395e-05,
"loss": 1.3349,
"step": 295
},
{
"epoch": 0.32,
"grad_norm": 6.34375,
"learning_rate": 4.8768574436458e-05,
"loss": 1.2216,
"step": 300
},
{
"epoch": 0.33,
"grad_norm": 5.5,
"learning_rate": 4.8724583402979826e-05,
"loss": 1.27,
"step": 305
},
{
"epoch": 0.33,
"grad_norm": 5.40625,
"learning_rate": 4.867984088932326e-05,
"loss": 1.2596,
"step": 310
},
{
"epoch": 0.34,
"grad_norm": 6.21875,
"learning_rate": 4.8634348312714956e-05,
"loss": 1.3084,
"step": 315
},
{
"epoch": 0.34,
"grad_norm": 5.625,
"learning_rate": 4.8588107114139955e-05,
"loss": 1.3364,
"step": 320
},
{
"epoch": 0.35,
"grad_norm": 5.40625,
"learning_rate": 4.8541118758296025e-05,
"loss": 1.1912,
"step": 325
},
{
"epoch": 0.35,
"grad_norm": 5.03125,
"learning_rate": 4.849338473354725e-05,
"loss": 1.3021,
"step": 330
},
{
"epoch": 0.36,
"grad_norm": 6.0,
"learning_rate": 4.8444906551876914e-05,
"loss": 1.2595,
"step": 335
},
{
"epoch": 0.36,
"grad_norm": 5.25,
"learning_rate": 4.839568574883958e-05,
"loss": 1.2704,
"step": 340
},
{
"epoch": 0.37,
"grad_norm": 5.25,
"learning_rate": 4.83457238835125e-05,
"loss": 1.4043,
"step": 345
},
{
"epoch": 0.37,
"grad_norm": 6.5625,
"learning_rate": 4.8295022538446164e-05,
"loss": 1.3419,
"step": 350
},
{
"epoch": 0.38,
"grad_norm": 5.75,
"learning_rate": 4.824358331961423e-05,
"loss": 1.2798,
"step": 355
},
{
"epoch": 0.38,
"grad_norm": 5.40625,
"learning_rate": 4.819140785636262e-05,
"loss": 1.3005,
"step": 360
},
{
"epoch": 0.39,
"grad_norm": 4.9375,
"learning_rate": 4.813849780135795e-05,
"loss": 1.3454,
"step": 365
},
{
"epoch": 0.39,
"grad_norm": 5.40625,
"learning_rate": 4.8084854830535095e-05,
"loss": 1.3375,
"step": 370
},
{
"epoch": 0.4,
"grad_norm": 6.09375,
"learning_rate": 4.803048064304423e-05,
"loss": 1.3316,
"step": 375
},
{
"epoch": 0.41,
"grad_norm": 4.875,
"learning_rate": 4.797537696119689e-05,
"loss": 1.2503,
"step": 380
},
{
"epoch": 0.41,
"grad_norm": 6.375,
"learning_rate": 4.7919545530411494e-05,
"loss": 1.2164,
"step": 385
},
{
"epoch": 0.42,
"grad_norm": 7.28125,
"learning_rate": 4.786298811915801e-05,
"loss": 1.2928,
"step": 390
},
{
"epoch": 0.42,
"grad_norm": 5.8125,
"learning_rate": 4.7805706518902e-05,
"loss": 1.2657,
"step": 395
},
{
"epoch": 0.43,
"grad_norm": 5.375,
"learning_rate": 4.774770254404779e-05,
"loss": 1.2557,
"step": 400
},
{
"epoch": 0.43,
"grad_norm": 7.90625,
"learning_rate": 4.768897803188105e-05,
"loss": 1.3147,
"step": 405
},
{
"epoch": 0.44,
"grad_norm": 5.8125,
"learning_rate": 4.7629534842510626e-05,
"loss": 1.289,
"step": 410
},
{
"epoch": 0.44,
"grad_norm": 5.96875,
"learning_rate": 4.7569374858809546e-05,
"loss": 1.2597,
"step": 415
},
{
"epoch": 0.45,
"grad_norm": 6.53125,
"learning_rate": 4.750849998635544e-05,
"loss": 1.247,
"step": 420
},
{
"epoch": 0.45,
"grad_norm": 8.0625,
"learning_rate": 4.744691215337015e-05,
"loss": 1.3946,
"step": 425
},
{
"epoch": 0.46,
"grad_norm": 8.8125,
"learning_rate": 4.738461331065867e-05,
"loss": 1.3241,
"step": 430
},
{
"epoch": 0.46,
"grad_norm": 5.9375,
"learning_rate": 4.732160543154735e-05,
"loss": 1.2764,
"step": 435
},
{
"epoch": 0.47,
"grad_norm": 10.125,
"learning_rate": 4.725789051182136e-05,
"loss": 1.2941,
"step": 440
},
{
"epoch": 0.47,
"grad_norm": 6.03125,
"learning_rate": 4.719347056966155e-05,
"loss": 1.3391,
"step": 445
},
{
"epoch": 0.48,
"grad_norm": 7.40625,
"learning_rate": 4.7128347645580434e-05,
"loss": 1.3739,
"step": 450
},
{
"epoch": 0.49,
"grad_norm": 6.34375,
"learning_rate": 4.7062523802357606e-05,
"loss": 1.2787,
"step": 455
},
{
"epoch": 0.49,
"grad_norm": 6.4375,
"learning_rate": 4.699600112497441e-05,
"loss": 1.2488,
"step": 460
},
{
"epoch": 0.5,
"grad_norm": 5.3125,
"learning_rate": 4.692878172054784e-05,
"loss": 1.217,
"step": 465
},
{
"epoch": 0.5,
"grad_norm": 5.65625,
"learning_rate": 4.686086771826389e-05,
"loss": 1.2612,
"step": 470
},
{
"epoch": 0.51,
"grad_norm": 5.59375,
"learning_rate": 4.679226126931001e-05,
"loss": 1.2482,
"step": 475
},
{
"epoch": 0.51,
"grad_norm": 4.625,
"learning_rate": 4.672296454680705e-05,
"loss": 1.2179,
"step": 480
},
{
"epoch": 0.52,
"grad_norm": 7.28125,
"learning_rate": 4.665297974574037e-05,
"loss": 1.2427,
"step": 485
},
{
"epoch": 0.52,
"grad_norm": 5.34375,
"learning_rate": 4.6582309082890327e-05,
"loss": 1.3199,
"step": 490
},
{
"epoch": 0.53,
"grad_norm": 5.46875,
"learning_rate": 4.6510954796762094e-05,
"loss": 1.2884,
"step": 495
},
{
"epoch": 0.53,
"grad_norm": 5.8125,
"learning_rate": 4.643891914751472e-05,
"loss": 1.3176,
"step": 500
},
{
"epoch": 0.54,
"grad_norm": 5.40625,
"learning_rate": 4.6366204416889526e-05,
"loss": 1.361,
"step": 505
},
{
"epoch": 0.54,
"grad_norm": 6.6875,
"learning_rate": 4.629281290813785e-05,
"loss": 1.3099,
"step": 510
},
{
"epoch": 0.55,
"grad_norm": 4.96875,
"learning_rate": 4.6218746945948124e-05,
"loss": 1.2626,
"step": 515
},
{
"epoch": 0.55,
"grad_norm": 5.71875,
"learning_rate": 4.614400887637218e-05,
"loss": 1.2902,
"step": 520
},
{
"epoch": 0.56,
"grad_norm": 4.6875,
"learning_rate": 4.6068601066750964e-05,
"loss": 1.242,
"step": 525
},
{
"epoch": 0.57,
"grad_norm": 5.53125,
"learning_rate": 4.599252590563956e-05,
"loss": 1.2424,
"step": 530
},
{
"epoch": 0.57,
"grad_norm": 4.9375,
"learning_rate": 4.5915785802731515e-05,
"loss": 1.263,
"step": 535
},
{
"epoch": 0.58,
"grad_norm": 5.625,
"learning_rate": 4.583838318878253e-05,
"loss": 1.237,
"step": 540
},
{
"epoch": 0.58,
"grad_norm": 5.8125,
"learning_rate": 4.576032051553345e-05,
"loss": 1.3625,
"step": 545
},
{
"epoch": 0.59,
"grad_norm": 6.34375,
"learning_rate": 4.5681600255632605e-05,
"loss": 1.2604,
"step": 550
},
{
"epoch": 0.59,
"grad_norm": 4.875,
"learning_rate": 4.56022249025575e-05,
"loss": 1.4195,
"step": 555
},
{
"epoch": 0.6,
"grad_norm": 6.03125,
"learning_rate": 4.5522196970535825e-05,
"loss": 1.2599,
"step": 560
},
{
"epoch": 0.6,
"grad_norm": 5.21875,
"learning_rate": 4.5441518994465806e-05,
"loss": 1.3454,
"step": 565
},
{
"epoch": 0.61,
"grad_norm": 6.46875,
"learning_rate": 4.5360193529835935e-05,
"loss": 1.3825,
"step": 570
},
{
"epoch": 0.61,
"grad_norm": 5.8125,
"learning_rate": 4.527822315264403e-05,
"loss": 1.2646,
"step": 575
},
{
"epoch": 0.62,
"grad_norm": 5.78125,
"learning_rate": 4.519561045931559e-05,
"loss": 1.262,
"step": 580
},
{
"epoch": 0.62,
"grad_norm": 5.34375,
"learning_rate": 4.5112358066621605e-05,
"loss": 1.2725,
"step": 585
},
{
"epoch": 0.63,
"grad_norm": 8.1875,
"learning_rate": 4.502846861159565e-05,
"loss": 1.282,
"step": 590
},
{
"epoch": 0.63,
"grad_norm": 5.09375,
"learning_rate": 4.494394475145036e-05,
"loss": 1.292,
"step": 595
},
{
"epoch": 0.64,
"grad_norm": 5.84375,
"learning_rate": 4.485878916349324e-05,
"loss": 1.2546,
"step": 600
},
{
"epoch": 0.65,
"grad_norm": 6.0,
"learning_rate": 4.4773004545041894e-05,
"loss": 1.2673,
"step": 605
},
{
"epoch": 0.65,
"grad_norm": 5.09375,
"learning_rate": 4.468659361333857e-05,
"loss": 1.2391,
"step": 610
},
{
"epoch": 0.66,
"grad_norm": 8.4375,
"learning_rate": 4.4599559105464106e-05,
"loss": 1.3886,
"step": 615
},
{
"epoch": 0.66,
"grad_norm": 5.1875,
"learning_rate": 4.451190377825121e-05,
"loss": 1.1826,
"step": 620
},
{
"epoch": 0.67,
"grad_norm": 5.375,
"learning_rate": 4.442363040819714e-05,
"loss": 1.215,
"step": 625
},
{
"epoch": 0.67,
"grad_norm": 5.53125,
"learning_rate": 4.433474179137577e-05,
"loss": 1.18,
"step": 630
},
{
"epoch": 0.68,
"grad_norm": 5.875,
"learning_rate": 4.424524074334902e-05,
"loss": 1.2138,
"step": 635
},
{
"epoch": 0.68,
"grad_norm": 5.40625,
"learning_rate": 4.4155130099077676e-05,
"loss": 1.2433,
"step": 640
},
{
"epoch": 0.69,
"grad_norm": 5.0,
"learning_rate": 4.406441271283159e-05,
"loss": 1.2532,
"step": 645
},
{
"epoch": 0.69,
"grad_norm": 4.40625,
"learning_rate": 4.3973091458099266e-05,
"loss": 1.263,
"step": 650
},
{
"epoch": 0.7,
"grad_norm": 5.9375,
"learning_rate": 4.388116922749683e-05,
"loss": 1.2687,
"step": 655
},
{
"epoch": 0.7,
"grad_norm": 6.46875,
"learning_rate": 4.3788648932676436e-05,
"loss": 1.2414,
"step": 660
},
{
"epoch": 0.71,
"grad_norm": 5.5,
"learning_rate": 4.369553350423401e-05,
"loss": 1.2109,
"step": 665
},
{
"epoch": 0.71,
"grad_norm": 5.34375,
"learning_rate": 4.3601825891616453e-05,
"loss": 1.3227,
"step": 670
},
{
"epoch": 0.72,
"grad_norm": 5.0625,
"learning_rate": 4.35075290630282e-05,
"loss": 1.2769,
"step": 675
},
{
"epoch": 0.73,
"grad_norm": 5.59375,
"learning_rate": 4.3412646005337166e-05,
"loss": 1.2736,
"step": 680
},
{
"epoch": 0.73,
"grad_norm": 5.875,
"learning_rate": 4.331717972398022e-05,
"loss": 1.1944,
"step": 685
},
{
"epoch": 0.74,
"grad_norm": 5.875,
"learning_rate": 4.32211332428679e-05,
"loss": 1.2209,
"step": 690
},
{
"epoch": 0.74,
"grad_norm": 8.0625,
"learning_rate": 4.31245096042887e-05,
"loss": 1.2117,
"step": 695
},
{
"epoch": 0.75,
"grad_norm": 6.03125,
"learning_rate": 4.302731186881265e-05,
"loss": 1.2603,
"step": 700
},
{
"epoch": 0.75,
"grad_norm": 5.5625,
"learning_rate": 4.29295431151944e-05,
"loss": 1.3085,
"step": 705
},
{
"epoch": 0.76,
"grad_norm": 5.5,
"learning_rate": 4.283120644027571e-05,
"loss": 1.1815,
"step": 710
},
{
"epoch": 0.76,
"grad_norm": 5.1875,
"learning_rate": 4.2732304958887336e-05,
"loss": 1.2594,
"step": 715
},
{
"epoch": 0.77,
"grad_norm": 5.65625,
"learning_rate": 4.2632841803750365e-05,
"loss": 1.2502,
"step": 720
},
{
"epoch": 0.77,
"grad_norm": 7.09375,
"learning_rate": 4.253282012537701e-05,
"loss": 1.3242,
"step": 725
},
{
"epoch": 0.78,
"grad_norm": 6.65625,
"learning_rate": 4.2432243091970784e-05,
"loss": 1.286,
"step": 730
},
{
"epoch": 0.78,
"grad_norm": 5.625,
"learning_rate": 4.233111388932618e-05,
"loss": 1.2395,
"step": 735
},
{
"epoch": 0.79,
"grad_norm": 5.1875,
"learning_rate": 4.222943572072773e-05,
"loss": 1.2041,
"step": 740
},
{
"epoch": 0.79,
"grad_norm": 4.375,
"learning_rate": 4.2127211806848575e-05,
"loss": 1.2588,
"step": 745
},
{
"epoch": 0.8,
"grad_norm": 4.8125,
"learning_rate": 4.202444538564842e-05,
"loss": 1.2544,
"step": 750
},
{
"epoch": 0.81,
"grad_norm": 5.3125,
"learning_rate": 4.192113971227097e-05,
"loss": 1.2406,
"step": 755
},
{
"epoch": 0.81,
"grad_norm": 5.6875,
"learning_rate": 4.181729805894083e-05,
"loss": 1.3559,
"step": 760
},
{
"epoch": 0.82,
"grad_norm": 6.0625,
"learning_rate": 4.171292371485991e-05,
"loss": 1.2854,
"step": 765
},
{
"epoch": 0.82,
"grad_norm": 5.78125,
"learning_rate": 4.160801998610312e-05,
"loss": 1.2643,
"step": 770
},
{
"epoch": 0.83,
"grad_norm": 4.625,
"learning_rate": 4.150259019551377e-05,
"loss": 1.1756,
"step": 775
},
{
"epoch": 0.83,
"grad_norm": 5.78125,
"learning_rate": 4.1396637682598235e-05,
"loss": 1.2207,
"step": 780
},
{
"epoch": 0.84,
"grad_norm": 7.96875,
"learning_rate": 4.1290165803420235e-05,
"loss": 1.2095,
"step": 785
},
{
"epoch": 0.84,
"grad_norm": 5.9375,
"learning_rate": 4.118317793049448e-05,
"loss": 1.2021,
"step": 790
},
{
"epoch": 0.85,
"grad_norm": 5.25,
"learning_rate": 4.107567745267987e-05,
"loss": 1.2172,
"step": 795
},
{
"epoch": 0.85,
"grad_norm": 7.28125,
"learning_rate": 4.096766777507218e-05,
"loss": 1.1723,
"step": 800
},
{
"epoch": 0.86,
"grad_norm": 5.3125,
"learning_rate": 4.085915231889615e-05,
"loss": 1.2254,
"step": 805
},
{
"epoch": 0.86,
"grad_norm": 6.6875,
"learning_rate": 4.075013452139714e-05,
"loss": 1.2538,
"step": 810
},
{
"epoch": 0.87,
"grad_norm": 5.09375,
"learning_rate": 4.064061783573227e-05,
"loss": 1.1936,
"step": 815
},
{
"epoch": 0.87,
"grad_norm": 5.1875,
"learning_rate": 4.053060573086101e-05,
"loss": 1.19,
"step": 820
},
{
"epoch": 0.88,
"grad_norm": 6.125,
"learning_rate": 4.042010169143533e-05,
"loss": 1.2572,
"step": 825
},
{
"epoch": 0.89,
"grad_norm": 6.6875,
"learning_rate": 4.030910921768931e-05,
"loss": 1.2529,
"step": 830
},
{
"epoch": 0.89,
"grad_norm": 7.34375,
"learning_rate": 4.0197631825328264e-05,
"loss": 1.2959,
"step": 835
},
{
"epoch": 0.9,
"grad_norm": 5.84375,
"learning_rate": 4.008567304541737e-05,
"loss": 1.3033,
"step": 840
},
{
"epoch": 0.9,
"grad_norm": 5.4375,
"learning_rate": 3.9973236424269876e-05,
"loss": 1.3391,
"step": 845
},
{
"epoch": 0.91,
"grad_norm": 4.9375,
"learning_rate": 3.9860325523334695e-05,
"loss": 1.2128,
"step": 850
},
{
"epoch": 0.91,
"grad_norm": 5.15625,
"learning_rate": 3.9746943919083665e-05,
"loss": 1.2255,
"step": 855
},
{
"epoch": 0.92,
"grad_norm": 7.46875,
"learning_rate": 3.963309520289819e-05,
"loss": 1.2787,
"step": 860
},
{
"epoch": 0.92,
"grad_norm": 7.625,
"learning_rate": 3.951878298095556e-05,
"loss": 1.4147,
"step": 865
},
{
"epoch": 0.93,
"grad_norm": 5.65625,
"learning_rate": 3.940401087411467e-05,
"loss": 1.2919,
"step": 870
},
{
"epoch": 0.93,
"grad_norm": 5.96875,
"learning_rate": 3.928878251780135e-05,
"loss": 1.291,
"step": 875
},
{
"epoch": 0.94,
"grad_norm": 5.78125,
"learning_rate": 3.91731015618932e-05,
"loss": 1.2669,
"step": 880
},
{
"epoch": 0.94,
"grad_norm": 6.28125,
"learning_rate": 3.905697167060402e-05,
"loss": 1.2935,
"step": 885
},
{
"epoch": 0.95,
"grad_norm": 5.03125,
"learning_rate": 3.894039652236767e-05,
"loss": 1.3249,
"step": 890
},
{
"epoch": 0.95,
"grad_norm": 6.59375,
"learning_rate": 3.882337980972164e-05,
"loss": 1.2558,
"step": 895
},
{
"epoch": 0.96,
"grad_norm": 4.84375,
"learning_rate": 3.870592523919002e-05,
"loss": 1.1875,
"step": 900
},
{
"epoch": 0.97,
"grad_norm": 6.375,
"learning_rate": 3.8588036531166124e-05,
"loss": 1.2176,
"step": 905
},
{
"epoch": 0.97,
"grad_norm": 7.21875,
"learning_rate": 3.846971741979465e-05,
"loss": 1.2816,
"step": 910
},
{
"epoch": 0.98,
"grad_norm": 7.34375,
"learning_rate": 3.835097165285341e-05,
"loss": 1.2478,
"step": 915
},
{
"epoch": 0.98,
"grad_norm": 5.75,
"learning_rate": 3.823180299163458e-05,
"loss": 1.2156,
"step": 920
},
{
"epoch": 0.99,
"grad_norm": 5.9375,
"learning_rate": 3.811221521082558e-05,
"loss": 1.3427,
"step": 925
},
{
"epoch": 0.99,
"grad_norm": 5.3125,
"learning_rate": 3.7992212098389536e-05,
"loss": 1.2286,
"step": 930
},
{
"epoch": 1.0,
"grad_norm": 5.4375,
"learning_rate": 3.7871797455445256e-05,
"loss": 1.2868,
"step": 935
},
{
"epoch": 1.0,
"grad_norm": 5.0625,
"learning_rate": 3.775097509614687e-05,
"loss": 1.2123,
"step": 940
},
{
"epoch": 1.01,
"grad_norm": 5.5625,
"learning_rate": 3.762974884756296e-05,
"loss": 1.114,
"step": 945
},
{
"epoch": 1.01,
"grad_norm": 4.5,
"learning_rate": 3.7508122549555384e-05,
"loss": 1.1734,
"step": 950
},
{
"epoch": 1.02,
"grad_norm": 5.21875,
"learning_rate": 3.738610005465764e-05,
"loss": 1.1267,
"step": 955
},
{
"epoch": 1.02,
"grad_norm": 4.9375,
"learning_rate": 3.726368522795282e-05,
"loss": 1.2459,
"step": 960
},
{
"epoch": 1.03,
"grad_norm": 5.28125,
"learning_rate": 3.714088194695119e-05,
"loss": 1.1431,
"step": 965
},
{
"epoch": 1.03,
"grad_norm": 9.1875,
"learning_rate": 3.7017694101467384e-05,
"loss": 1.2636,
"step": 970
},
{
"epoch": 1.04,
"grad_norm": 5.96875,
"learning_rate": 3.6894125593497167e-05,
"loss": 1.1851,
"step": 975
},
{
"epoch": 1.05,
"grad_norm": 5.75,
"learning_rate": 3.677018033709386e-05,
"loss": 1.134,
"step": 980
},
{
"epoch": 1.05,
"grad_norm": 5.1875,
"learning_rate": 3.6645862258244344e-05,
"loss": 1.1106,
"step": 985
},
{
"epoch": 1.06,
"grad_norm": 5.71875,
"learning_rate": 3.6521175294744735e-05,
"loss": 1.1718,
"step": 990
},
{
"epoch": 1.06,
"grad_norm": 4.8125,
"learning_rate": 3.639612339607561e-05,
"loss": 1.1642,
"step": 995
},
{
"epoch": 1.07,
"grad_norm": 5.6875,
"learning_rate": 3.627071052327695e-05,
"loss": 1.1399,
"step": 1000
},
{
"epoch": 1.07,
"grad_norm": 6.4375,
"learning_rate": 3.614494064882263e-05,
"loss": 1.1741,
"step": 1005
},
{
"epoch": 1.08,
"grad_norm": 5.03125,
"learning_rate": 3.6018817756494634e-05,
"loss": 1.1551,
"step": 1010
},
{
"epoch": 1.08,
"grad_norm": 5.5,
"learning_rate": 3.5892345841256825e-05,
"loss": 1.1106,
"step": 1015
},
{
"epoch": 1.09,
"grad_norm": 7.75,
"learning_rate": 3.576552890912844e-05,
"loss": 1.1462,
"step": 1020
},
{
"epoch": 1.09,
"grad_norm": 4.90625,
"learning_rate": 3.5638370977057186e-05,
"loss": 1.1489,
"step": 1025
},
{
"epoch": 1.1,
"grad_norm": 5.125,
"learning_rate": 3.5510876072791986e-05,
"loss": 1.1427,
"step": 1030
},
{
"epoch": 1.1,
"grad_norm": 5.125,
"learning_rate": 3.538304823475543e-05,
"loss": 1.1172,
"step": 1035
},
{
"epoch": 1.11,
"grad_norm": 6.28125,
"learning_rate": 3.5254891511915836e-05,
"loss": 1.2579,
"step": 1040
},
{
"epoch": 1.11,
"grad_norm": 4.84375,
"learning_rate": 3.5126409963659004e-05,
"loss": 1.1606,
"step": 1045
},
{
"epoch": 1.12,
"grad_norm": 4.6875,
"learning_rate": 3.4997607659659614e-05,
"loss": 1.2115,
"step": 1050
},
{
"epoch": 1.13,
"grad_norm": 5.15625,
"learning_rate": 3.4868488679752384e-05,
"loss": 1.1618,
"step": 1055
},
{
"epoch": 1.13,
"grad_norm": 4.875,
"learning_rate": 3.4739057113802756e-05,
"loss": 1.1754,
"step": 1060
},
{
"epoch": 1.14,
"grad_norm": 6.125,
"learning_rate": 3.46093170615774e-05,
"loss": 1.2046,
"step": 1065
},
{
"epoch": 1.14,
"grad_norm": 6.03125,
"learning_rate": 3.4479272632614326e-05,
"loss": 1.2066,
"step": 1070
},
{
"epoch": 1.15,
"grad_norm": 6.90625,
"learning_rate": 3.4348927946092766e-05,
"loss": 1.1538,
"step": 1075
},
{
"epoch": 1.15,
"grad_norm": 5.5625,
"learning_rate": 3.421828713070261e-05,
"loss": 1.1924,
"step": 1080
},
{
"epoch": 1.16,
"grad_norm": 5.78125,
"learning_rate": 3.4087354324513734e-05,
"loss": 1.1463,
"step": 1085
},
{
"epoch": 1.16,
"grad_norm": 8.75,
"learning_rate": 3.3956133674844826e-05,
"loss": 1.2498,
"step": 1090
},
{
"epoch": 1.17,
"grad_norm": 5.90625,
"learning_rate": 3.38246293381321e-05,
"loss": 1.1543,
"step": 1095
},
{
"epoch": 1.17,
"grad_norm": 5.3125,
"learning_rate": 3.3692845479797564e-05,
"loss": 1.1401,
"step": 1100
},
{
"epoch": 1.18,
"grad_norm": 5.53125,
"learning_rate": 3.356078627411716e-05,
"loss": 1.1909,
"step": 1105
},
{
"epoch": 1.18,
"grad_norm": 5.9375,
"learning_rate": 3.342845590408849e-05,
"loss": 1.1075,
"step": 1110
},
{
"epoch": 1.19,
"grad_norm": 5.15625,
"learning_rate": 3.329585856129833e-05,
"loss": 1.2045,
"step": 1115
},
{
"epoch": 1.19,
"grad_norm": 5.21875,
"learning_rate": 3.316299844578985e-05,
"loss": 1.1752,
"step": 1120
},
{
"epoch": 1.2,
"grad_norm": 4.5625,
"learning_rate": 3.302987976592963e-05,
"loss": 1.1077,
"step": 1125
},
{
"epoch": 1.21,
"grad_norm": 5.15625,
"learning_rate": 3.289650673827427e-05,
"loss": 1.1641,
"step": 1130
},
{
"epoch": 1.21,
"grad_norm": 4.96875,
"learning_rate": 3.276288358743691e-05,
"loss": 1.1683,
"step": 1135
},
{
"epoch": 1.22,
"grad_norm": 6.6875,
"learning_rate": 3.262901454595338e-05,
"loss": 1.1507,
"step": 1140
},
{
"epoch": 1.22,
"grad_norm": 6.1875,
"learning_rate": 3.2494903854148134e-05,
"loss": 1.1706,
"step": 1145
},
{
"epoch": 1.23,
"grad_norm": 5.5,
"learning_rate": 3.236055575999994e-05,
"loss": 1.1484,
"step": 1150
},
{
"epoch": 1.23,
"grad_norm": 5.03125,
"learning_rate": 3.222597451900733e-05,
"loss": 1.2247,
"step": 1155
},
{
"epoch": 1.24,
"grad_norm": 5.84375,
"learning_rate": 3.2091164394053785e-05,
"loss": 1.1492,
"step": 1160
},
{
"epoch": 1.24,
"grad_norm": 5.375,
"learning_rate": 3.1956129655272755e-05,
"loss": 1.1822,
"step": 1165
},
{
"epoch": 1.25,
"grad_norm": 6.59375,
"learning_rate": 3.182087457991233e-05,
"loss": 1.2857,
"step": 1170
},
{
"epoch": 1.25,
"grad_norm": 7.75,
"learning_rate": 3.168540345219985e-05,
"loss": 1.1828,
"step": 1175
},
{
"epoch": 1.26,
"grad_norm": 7.75,
"learning_rate": 3.15497205632061e-05,
"loss": 1.2212,
"step": 1180
},
{
"epoch": 1.26,
"grad_norm": 6.03125,
"learning_rate": 3.141383021070947e-05,
"loss": 1.1151,
"step": 1185
},
{
"epoch": 1.27,
"grad_norm": 4.71875,
"learning_rate": 3.127773669905978e-05,
"loss": 1.226,
"step": 1190
},
{
"epoch": 1.27,
"grad_norm": 6.09375,
"learning_rate": 3.114144433904194e-05,
"loss": 1.2121,
"step": 1195
},
{
"epoch": 1.28,
"grad_norm": 5.8125,
"learning_rate": 3.100495744773944e-05,
"loss": 1.1956,
"step": 1200
},
{
"epoch": 1.29,
"grad_norm": 6.90625,
"learning_rate": 3.086828034839756e-05,
"loss": 1.1705,
"step": 1205
},
{
"epoch": 1.29,
"grad_norm": 5.8125,
"learning_rate": 3.073141737028648e-05,
"loss": 1.1738,
"step": 1210
},
{
"epoch": 1.3,
"grad_norm": 6.28125,
"learning_rate": 3.059437284856409e-05,
"loss": 1.2063,
"step": 1215
},
{
"epoch": 1.3,
"grad_norm": 5.4375,
"learning_rate": 3.04571511241387e-05,
"loss": 1.1492,
"step": 1220
},
{
"epoch": 1.31,
"grad_norm": 5.78125,
"learning_rate": 3.031975654353158e-05,
"loss": 1.2052,
"step": 1225
},
{
"epoch": 1.31,
"grad_norm": 5.0,
"learning_rate": 3.018219345873923e-05,
"loss": 1.1688,
"step": 1230
},
{
"epoch": 1.32,
"grad_norm": 6.21875,
"learning_rate": 3.004446622709554e-05,
"loss": 1.0857,
"step": 1235
},
{
"epoch": 1.32,
"grad_norm": 6.40625,
"learning_rate": 2.99065792111338e-05,
"loss": 1.2369,
"step": 1240
},
{
"epoch": 1.33,
"grad_norm": 6.46875,
"learning_rate": 2.976853677844847e-05,
"loss": 1.1783,
"step": 1245
},
{
"epoch": 1.33,
"grad_norm": 5.65625,
"learning_rate": 2.9630343301556883e-05,
"loss": 1.1596,
"step": 1250
},
{
"epoch": 1.34,
"grad_norm": 5.90625,
"learning_rate": 2.9492003157760713e-05,
"loss": 1.2145,
"step": 1255
},
{
"epoch": 1.34,
"grad_norm": 5.625,
"learning_rate": 2.9353520729007326e-05,
"loss": 1.157,
"step": 1260
},
{
"epoch": 1.35,
"grad_norm": 5.28125,
"learning_rate": 2.921490040175101e-05,
"loss": 1.2069,
"step": 1265
},
{
"epoch": 1.35,
"grad_norm": 5.90625,
"learning_rate": 2.9076146566813994e-05,
"loss": 1.1791,
"step": 1270
},
{
"epoch": 1.36,
"grad_norm": 5.21875,
"learning_rate": 2.893726361924738e-05,
"loss": 1.1243,
"step": 1275
},
{
"epoch": 1.37,
"grad_norm": 6.53125,
"learning_rate": 2.8798255958191946e-05,
"loss": 1.191,
"step": 1280
},
{
"epoch": 1.37,
"grad_norm": 5.625,
"learning_rate": 2.8659127986738782e-05,
"loss": 1.1712,
"step": 1285
},
{
"epoch": 1.38,
"grad_norm": 4.75,
"learning_rate": 2.851988411178985e-05,
"loss": 1.1619,
"step": 1290
},
{
"epoch": 1.38,
"grad_norm": 5.71875,
"learning_rate": 2.8380528743918348e-05,
"loss": 1.1699,
"step": 1295
},
{
"epoch": 1.39,
"grad_norm": 6.375,
"learning_rate": 2.824106629722906e-05,
"loss": 1.2141,
"step": 1300
},
{
"epoch": 1.39,
"grad_norm": 6.1875,
"learning_rate": 2.810150118921847e-05,
"loss": 1.1444,
"step": 1305
},
{
"epoch": 1.4,
"grad_norm": 5.875,
"learning_rate": 2.7961837840634913e-05,
"loss": 1.1494,
"step": 1310
},
{
"epoch": 1.4,
"grad_norm": 6.3125,
"learning_rate": 2.7822080675338492e-05,
"loss": 1.1783,
"step": 1315
},
{
"epoch": 1.41,
"grad_norm": 5.25,
"learning_rate": 2.768223412016099e-05,
"loss": 1.1855,
"step": 1320
},
{
"epoch": 1.41,
"grad_norm": 5.96875,
"learning_rate": 2.7542302604765598e-05,
"loss": 1.1509,
"step": 1325
},
{
"epoch": 1.42,
"grad_norm": 5.03125,
"learning_rate": 2.740229056150666e-05,
"loss": 1.1274,
"step": 1330
},
{
"epoch": 1.42,
"grad_norm": 4.53125,
"learning_rate": 2.7262202425289253e-05,
"loss": 1.1581,
"step": 1335
},
{
"epoch": 1.43,
"grad_norm": 5.28125,
"learning_rate": 2.7122042633428697e-05,
"loss": 1.0967,
"step": 1340
},
{
"epoch": 1.43,
"grad_norm": 5.71875,
"learning_rate": 2.698181562551002e-05,
"loss": 1.181,
"step": 1345
},
{
"epoch": 1.44,
"grad_norm": 5.625,
"learning_rate": 2.684152584324735e-05,
"loss": 1.151,
"step": 1350
},
{
"epoch": 1.45,
"grad_norm": 7.09375,
"learning_rate": 2.6701177730343186e-05,
"loss": 1.1978,
"step": 1355
},
{
"epoch": 1.45,
"grad_norm": 5.53125,
"learning_rate": 2.6560775732347654e-05,
"loss": 1.1315,
"step": 1360
},
{
"epoch": 1.46,
"grad_norm": 5.25,
"learning_rate": 2.6420324296517708e-05,
"loss": 1.1845,
"step": 1365
},
{
"epoch": 1.46,
"grad_norm": 5.53125,
"learning_rate": 2.627982787167625e-05,
"loss": 1.1386,
"step": 1370
},
{
"epoch": 1.47,
"grad_norm": 4.96875,
"learning_rate": 2.613929090807123e-05,
"loss": 1.159,
"step": 1375
},
{
"epoch": 1.47,
"grad_norm": 5.3125,
"learning_rate": 2.599871785723465e-05,
"loss": 1.1653,
"step": 1380
},
{
"epoch": 1.48,
"grad_norm": 5.53125,
"learning_rate": 2.5858113171841604e-05,
"loss": 1.2584,
"step": 1385
},
{
"epoch": 1.48,
"grad_norm": 5.75,
"learning_rate": 2.5717481305569186e-05,
"loss": 1.2103,
"step": 1390
},
{
"epoch": 1.49,
"grad_norm": 5.875,
"learning_rate": 2.5576826712955472e-05,
"loss": 1.2361,
"step": 1395
},
{
"epoch": 1.49,
"grad_norm": 5.15625,
"learning_rate": 2.5436153849258398e-05,
"loss": 1.1491,
"step": 1400
},
{
"epoch": 1.5,
"grad_norm": 5.5625,
"learning_rate": 2.5295467170314624e-05,
"loss": 1.1003,
"step": 1405
},
{
"epoch": 1.5,
"grad_norm": 5.6875,
"learning_rate": 2.515477113239843e-05,
"loss": 1.2162,
"step": 1410
},
{
"epoch": 1.51,
"grad_norm": 6.1875,
"learning_rate": 2.5014070192080526e-05,
"loss": 1.1498,
"step": 1415
},
{
"epoch": 1.51,
"grad_norm": 6.28125,
"learning_rate": 2.4873368806086918e-05,
"loss": 1.2132,
"step": 1420
},
{
"epoch": 1.52,
"grad_norm": 9.4375,
"learning_rate": 2.4732671431157716e-05,
"loss": 1.1903,
"step": 1425
},
{
"epoch": 1.53,
"grad_norm": 7.625,
"learning_rate": 2.4591982523906e-05,
"loss": 1.1472,
"step": 1430
},
{
"epoch": 1.53,
"grad_norm": 5.25,
"learning_rate": 2.4451306540676615e-05,
"loss": 1.1085,
"step": 1435
},
{
"epoch": 1.54,
"grad_norm": 4.84375,
"learning_rate": 2.431064793740505e-05,
"loss": 1.0636,
"step": 1440
},
{
"epoch": 1.54,
"grad_norm": 6.1875,
"learning_rate": 2.4170011169476254e-05,
"loss": 1.1439,
"step": 1445
},
{
"epoch": 1.55,
"grad_norm": 4.96875,
"learning_rate": 2.4029400691583574e-05,
"loss": 1.2272,
"step": 1450
},
{
"epoch": 1.55,
"grad_norm": 5.71875,
"learning_rate": 2.3888820957587605e-05,
"loss": 1.158,
"step": 1455
},
{
"epoch": 1.56,
"grad_norm": 5.90625,
"learning_rate": 2.3748276420375095e-05,
"loss": 1.1893,
"step": 1460
},
{
"epoch": 1.56,
"grad_norm": 5.6875,
"learning_rate": 2.3607771531717982e-05,
"loss": 1.1827,
"step": 1465
},
{
"epoch": 1.57,
"grad_norm": 5.09375,
"learning_rate": 2.3467310742132266e-05,
"loss": 1.2413,
"step": 1470
},
{
"epoch": 1.57,
"grad_norm": 5.125,
"learning_rate": 2.332689850073715e-05,
"loss": 1.1588,
"step": 1475
},
{
"epoch": 1.58,
"grad_norm": 7.46875,
"learning_rate": 2.3186539255114058e-05,
"loss": 1.2112,
"step": 1480
},
{
"epoch": 1.58,
"grad_norm": 5.53125,
"learning_rate": 2.304623745116573e-05,
"loss": 1.1712,
"step": 1485
},
{
"epoch": 1.59,
"grad_norm": 5.625,
"learning_rate": 2.2905997532975484e-05,
"loss": 1.1619,
"step": 1490
},
{
"epoch": 1.59,
"grad_norm": 5.15625,
"learning_rate": 2.2765823942666343e-05,
"loss": 1.1737,
"step": 1495
},
{
"epoch": 1.6,
"grad_norm": 6.03125,
"learning_rate": 2.2625721120260416e-05,
"loss": 1.1432,
"step": 1500
},
{
"epoch": 1.61,
"grad_norm": 6.0,
"learning_rate": 2.2485693503538232e-05,
"loss": 1.1612,
"step": 1505
},
{
"epoch": 1.61,
"grad_norm": 4.78125,
"learning_rate": 2.2345745527898138e-05,
"loss": 1.1659,
"step": 1510
},
{
"epoch": 1.62,
"grad_norm": 5.78125,
"learning_rate": 2.2205881626215868e-05,
"loss": 1.1377,
"step": 1515
},
{
"epoch": 1.62,
"grad_norm": 5.28125,
"learning_rate": 2.2066106228704058e-05,
"loss": 1.1139,
"step": 1520
},
{
"epoch": 1.63,
"grad_norm": 6.03125,
"learning_rate": 2.1926423762772017e-05,
"loss": 1.1575,
"step": 1525
},
{
"epoch": 1.63,
"grad_norm": 6.1875,
"learning_rate": 2.1786838652885373e-05,
"loss": 1.1132,
"step": 1530
},
{
"epoch": 1.64,
"grad_norm": 4.84375,
"learning_rate": 2.1647355320426017e-05,
"loss": 1.1038,
"step": 1535
},
{
"epoch": 1.64,
"grad_norm": 5.625,
"learning_rate": 2.1507978183552034e-05,
"loss": 1.064,
"step": 1540
},
{
"epoch": 1.65,
"grad_norm": 5.1875,
"learning_rate": 2.1368711657057705e-05,
"loss": 1.1183,
"step": 1545
},
{
"epoch": 1.65,
"grad_norm": 4.71875,
"learning_rate": 2.122956015223375e-05,
"loss": 1.1619,
"step": 1550
},
{
"epoch": 1.66,
"grad_norm": 5.0,
"learning_rate": 2.109052807672752e-05,
"loss": 1.2064,
"step": 1555
},
{
"epoch": 1.66,
"grad_norm": 5.9375,
"learning_rate": 2.0951619834403462e-05,
"loss": 1.1686,
"step": 1560
},
{
"epoch": 1.67,
"grad_norm": 5.375,
"learning_rate": 2.081283982520356e-05,
"loss": 1.1372,
"step": 1565
},
{
"epoch": 1.67,
"grad_norm": 5.375,
"learning_rate": 2.0674192445007997e-05,
"loss": 1.1645,
"step": 1570
},
{
"epoch": 1.68,
"grad_norm": 5.75,
"learning_rate": 2.0535682085495912e-05,
"loss": 1.1098,
"step": 1575
},
{
"epoch": 1.69,
"grad_norm": 6.40625,
"learning_rate": 2.039731313400629e-05,
"loss": 1.1762,
"step": 1580
},
{
"epoch": 1.69,
"grad_norm": 6.1875,
"learning_rate": 2.025908997339899e-05,
"loss": 1.1623,
"step": 1585
},
{
"epoch": 1.7,
"grad_norm": 4.65625,
"learning_rate": 2.0121016981915934e-05,
"loss": 1.169,
"step": 1590
},
{
"epoch": 1.7,
"grad_norm": 5.6875,
"learning_rate": 1.9983098533042384e-05,
"loss": 1.1352,
"step": 1595
},
{
"epoch": 1.71,
"grad_norm": 5.125,
"learning_rate": 1.9845338995368483e-05,
"loss": 1.1707,
"step": 1600
},
{
"epoch": 1.71,
"grad_norm": 5.78125,
"learning_rate": 1.9707742732450775e-05,
"loss": 1.1958,
"step": 1605
},
{
"epoch": 1.72,
"grad_norm": 5.28125,
"learning_rate": 1.9570314102674108e-05,
"loss": 1.1699,
"step": 1610
},
{
"epoch": 1.72,
"grad_norm": 5.8125,
"learning_rate": 1.9433057459113485e-05,
"loss": 1.1558,
"step": 1615
},
{
"epoch": 1.73,
"grad_norm": 5.78125,
"learning_rate": 1.9295977149396225e-05,
"loss": 1.1885,
"step": 1620
},
{
"epoch": 1.73,
"grad_norm": 5.875,
"learning_rate": 1.9159077515564257e-05,
"loss": 1.1141,
"step": 1625
},
{
"epoch": 1.74,
"grad_norm": 4.375,
"learning_rate": 1.902236289393654e-05,
"loss": 1.1866,
"step": 1630
},
{
"epoch": 1.74,
"grad_norm": 4.90625,
"learning_rate": 1.888583761497178e-05,
"loss": 1.1655,
"step": 1635
},
{
"epoch": 1.75,
"grad_norm": 5.25,
"learning_rate": 1.874950600313118e-05,
"loss": 1.167,
"step": 1640
},
{
"epoch": 1.75,
"grad_norm": 6.625,
"learning_rate": 1.861337237674154e-05,
"loss": 1.1927,
"step": 1645
},
{
"epoch": 1.76,
"grad_norm": 5.375,
"learning_rate": 1.8477441047858425e-05,
"loss": 1.1902,
"step": 1650
},
{
"epoch": 1.77,
"grad_norm": 5.65625,
"learning_rate": 1.8341716322129575e-05,
"loss": 1.1921,
"step": 1655
},
{
"epoch": 1.77,
"grad_norm": 5.25,
"learning_rate": 1.820620249865858e-05,
"loss": 1.1864,
"step": 1660
},
{
"epoch": 1.78,
"grad_norm": 6.875,
"learning_rate": 1.8070903869868615e-05,
"loss": 1.1866,
"step": 1665
},
{
"epoch": 1.78,
"grad_norm": 5.65625,
"learning_rate": 1.7935824721366574e-05,
"loss": 1.1595,
"step": 1670
},
{
"epoch": 1.79,
"grad_norm": 7.1875,
"learning_rate": 1.7800969331807237e-05,
"loss": 1.1422,
"step": 1675
},
{
"epoch": 1.79,
"grad_norm": 5.8125,
"learning_rate": 1.766634197275781e-05,
"loss": 1.1428,
"step": 1680
},
{
"epoch": 1.8,
"grad_norm": 5.9375,
"learning_rate": 1.7531946908562596e-05,
"loss": 1.1056,
"step": 1685
},
{
"epoch": 1.8,
"grad_norm": 7.1875,
"learning_rate": 1.7397788396207883e-05,
"loss": 1.2066,
"step": 1690
},
{
"epoch": 1.81,
"grad_norm": 5.1875,
"learning_rate": 1.7263870685187188e-05,
"loss": 1.1753,
"step": 1695
},
{
"epoch": 1.81,
"grad_norm": 6.15625,
"learning_rate": 1.7130198017366556e-05,
"loss": 1.1243,
"step": 1700
},
{
"epoch": 1.82,
"grad_norm": 5.09375,
"learning_rate": 1.699677462685029e-05,
"loss": 1.1959,
"step": 1705
},
{
"epoch": 1.82,
"grad_norm": 5.25,
"learning_rate": 1.686360473984678e-05,
"loss": 1.1313,
"step": 1710
},
{
"epoch": 1.83,
"grad_norm": 5.21875,
"learning_rate": 1.6730692574534622e-05,
"loss": 1.1578,
"step": 1715
},
{
"epoch": 1.83,
"grad_norm": 6.5625,
"learning_rate": 1.659804234092908e-05,
"loss": 1.1415,
"step": 1720
},
{
"epoch": 1.84,
"grad_norm": 6.1875,
"learning_rate": 1.646565824074865e-05,
"loss": 1.1565,
"step": 1725
},
{
"epoch": 1.85,
"grad_norm": 5.3125,
"learning_rate": 1.6333544467282028e-05,
"loss": 1.1677,
"step": 1730
},
{
"epoch": 1.85,
"grad_norm": 9.375,
"learning_rate": 1.6201705205255244e-05,
"loss": 1.1842,
"step": 1735
},
{
"epoch": 1.86,
"grad_norm": 7.8125,
"learning_rate": 1.6070144630699165e-05,
"loss": 1.2085,
"step": 1740
},
{
"epoch": 1.86,
"grad_norm": 6.0625,
"learning_rate": 1.5938866910817152e-05,
"loss": 1.156,
"step": 1745
},
{
"epoch": 1.87,
"grad_norm": 5.28125,
"learning_rate": 1.5807876203853118e-05,
"loss": 1.1278,
"step": 1750
},
{
"epoch": 1.87,
"grad_norm": 5.59375,
"learning_rate": 1.5677176658959782e-05,
"loss": 1.1558,
"step": 1755
},
{
"epoch": 1.88,
"grad_norm": 5.75,
"learning_rate": 1.554677241606726e-05,
"loss": 1.1674,
"step": 1760
},
{
"epoch": 1.88,
"grad_norm": 5.0625,
"learning_rate": 1.5416667605751933e-05,
"loss": 1.1816,
"step": 1765
},
{
"epoch": 1.89,
"grad_norm": 5.65625,
"learning_rate": 1.5286866349105593e-05,
"loss": 1.1665,
"step": 1770
},
{
"epoch": 1.89,
"grad_norm": 5.96875,
"learning_rate": 1.5157372757604932e-05,
"loss": 1.1067,
"step": 1775
},
{
"epoch": 1.9,
"grad_norm": 6.75,
"learning_rate": 1.502819093298128e-05,
"loss": 1.2465,
"step": 1780
},
{
"epoch": 1.9,
"grad_norm": 5.65625,
"learning_rate": 1.4899324967090705e-05,
"loss": 1.1388,
"step": 1785
},
{
"epoch": 1.91,
"grad_norm": 6.125,
"learning_rate": 1.4770778941784401e-05,
"loss": 1.1855,
"step": 1790
},
{
"epoch": 1.91,
"grad_norm": 5.03125,
"learning_rate": 1.4642556928779403e-05,
"loss": 1.1639,
"step": 1795
},
{
"epoch": 1.92,
"grad_norm": 5.625,
"learning_rate": 1.4514662989529562e-05,
"loss": 1.1454,
"step": 1800
},
{
"epoch": 1.93,
"grad_norm": 5.46875,
"learning_rate": 1.4387101175096985e-05,
"loss": 1.1658,
"step": 1805
},
{
"epoch": 1.93,
"grad_norm": 5.3125,
"learning_rate": 1.4259875526023622e-05,
"loss": 1.1015,
"step": 1810
},
{
"epoch": 1.94,
"grad_norm": 4.84375,
"learning_rate": 1.4132990072203378e-05,
"loss": 1.1869,
"step": 1815
},
{
"epoch": 1.94,
"grad_norm": 5.90625,
"learning_rate": 1.400644883275437e-05,
"loss": 1.1581,
"step": 1820
},
{
"epoch": 1.95,
"grad_norm": 5.4375,
"learning_rate": 1.3880255815891696e-05,
"loss": 1.2049,
"step": 1825
},
{
"epoch": 1.95,
"grad_norm": 5.3125,
"learning_rate": 1.3754415018800448e-05,
"loss": 1.1779,
"step": 1830
},
{
"epoch": 1.96,
"grad_norm": 5.03125,
"learning_rate": 1.3628930427509068e-05,
"loss": 1.1909,
"step": 1835
},
{
"epoch": 1.96,
"grad_norm": 4.625,
"learning_rate": 1.3503806016763152e-05,
"loss": 1.1462,
"step": 1840
},
{
"epoch": 1.97,
"grad_norm": 6.03125,
"learning_rate": 1.3379045749899485e-05,
"loss": 1.1609,
"step": 1845
},
{
"epoch": 1.97,
"grad_norm": 6.0625,
"learning_rate": 1.325465357872056e-05,
"loss": 1.1463,
"step": 1850
},
{
"epoch": 1.98,
"grad_norm": 5.6875,
"learning_rate": 1.3130633443369377e-05,
"loss": 1.1447,
"step": 1855
},
{
"epoch": 1.98,
"grad_norm": 5.78125,
"learning_rate": 1.3006989272204611e-05,
"loss": 1.1105,
"step": 1860
},
{
"epoch": 1.99,
"grad_norm": 8.3125,
"learning_rate": 1.2883724981676243e-05,
"loss": 1.2189,
"step": 1865
},
{
"epoch": 1.99,
"grad_norm": 5.25,
"learning_rate": 1.2760844476201429e-05,
"loss": 1.1292,
"step": 1870
},
{
"epoch": 2.0,
"grad_norm": 5.21875,
"learning_rate": 1.2638351648040918e-05,
"loss": 1.1375,
"step": 1875
},
{
"epoch": 2.01,
"grad_norm": 4.65625,
"learning_rate": 1.2516250377175673e-05,
"loss": 1.1576,
"step": 1880
},
{
"epoch": 2.01,
"grad_norm": 4.84375,
"learning_rate": 1.2394544531184036e-05,
"loss": 1.168,
"step": 1885
},
{
"epoch": 2.02,
"grad_norm": 6.3125,
"learning_rate": 1.2273237965119202e-05,
"loss": 1.1362,
"step": 1890
},
{
"epoch": 2.02,
"grad_norm": 6.0,
"learning_rate": 1.2152334521387082e-05,
"loss": 1.147,
"step": 1895
},
{
"epoch": 2.03,
"grad_norm": 7.28125,
"learning_rate": 1.2031838029624657e-05,
"loss": 1.0798,
"step": 1900
},
{
"epoch": 2.03,
"grad_norm": 4.375,
"learning_rate": 1.1911752306578593e-05,
"loss": 1.0446,
"step": 1905
},
{
"epoch": 2.04,
"grad_norm": 5.21875,
"learning_rate": 1.1792081155984422e-05,
"loss": 1.0947,
"step": 1910
},
{
"epoch": 2.04,
"grad_norm": 5.25,
"learning_rate": 1.1672828368446018e-05,
"loss": 1.1792,
"step": 1915
},
{
"epoch": 2.05,
"grad_norm": 5.96875,
"learning_rate": 1.1553997721315509e-05,
"loss": 1.1014,
"step": 1920
},
{
"epoch": 2.05,
"grad_norm": 5.53125,
"learning_rate": 1.1435592978573695e-05,
"loss": 1.1151,
"step": 1925
},
{
"epoch": 2.06,
"grad_norm": 5.71875,
"learning_rate": 1.1317617890710738e-05,
"loss": 1.1495,
"step": 1930
},
{
"epoch": 2.06,
"grad_norm": 5.125,
"learning_rate": 1.1200076194607448e-05,
"loss": 1.203,
"step": 1935
},
{
"epoch": 2.07,
"grad_norm": 6.84375,
"learning_rate": 1.1082971613416865e-05,
"loss": 1.1496,
"step": 1940
},
{
"epoch": 2.07,
"grad_norm": 5.5625,
"learning_rate": 1.0966307856446322e-05,
"loss": 1.098,
"step": 1945
},
{
"epoch": 2.08,
"grad_norm": 5.3125,
"learning_rate": 1.0850088619040005e-05,
"loss": 1.1397,
"step": 1950
},
{
"epoch": 2.09,
"grad_norm": 5.0,
"learning_rate": 1.0734317582461833e-05,
"loss": 1.2423,
"step": 1955
},
{
"epoch": 2.09,
"grad_norm": 4.875,
"learning_rate": 1.061899841377892e-05,
"loss": 1.1154,
"step": 1960
},
{
"epoch": 2.1,
"grad_norm": 4.78125,
"learning_rate": 1.0504134765745351e-05,
"loss": 1.2164,
"step": 1965
},
{
"epoch": 2.1,
"grad_norm": 5.0,
"learning_rate": 1.0389730276686557e-05,
"loss": 1.0513,
"step": 1970
},
{
"epoch": 2.11,
"grad_norm": 4.9375,
"learning_rate": 1.0275788570384016e-05,
"loss": 1.0955,
"step": 1975
},
{
"epoch": 2.11,
"grad_norm": 5.59375,
"learning_rate": 1.0162313255960465e-05,
"loss": 1.075,
"step": 1980
},
{
"epoch": 2.12,
"grad_norm": 5.03125,
"learning_rate": 1.0049307927765634e-05,
"loss": 1.1331,
"step": 1985
},
{
"epoch": 2.12,
"grad_norm": 6.0,
"learning_rate": 9.936776165262324e-06,
"loss": 1.1564,
"step": 1990
},
{
"epoch": 2.13,
"grad_norm": 4.75,
"learning_rate": 9.824721532913091e-06,
"loss": 1.0781,
"step": 1995
},
{
"epoch": 2.13,
"grad_norm": 5.46875,
"learning_rate": 9.71314758006731e-06,
"loss": 1.1172,
"step": 2000
},
{
"epoch": 2.14,
"grad_norm": 6.09375,
"learning_rate": 9.602057840848732e-06,
"loss": 1.1993,
"step": 2005
},
{
"epoch": 2.14,
"grad_norm": 4.75,
"learning_rate": 9.491455834043583e-06,
"loss": 1.0923,
"step": 2010
},
{
"epoch": 2.15,
"grad_norm": 5.75,
"learning_rate": 9.38134506298906e-06,
"loss": 1.1308,
"step": 2015
},
{
"epoch": 2.15,
"grad_norm": 4.84375,
"learning_rate": 9.271729015462411e-06,
"loss": 1.1036,
"step": 2020
},
{
"epoch": 2.16,
"grad_norm": 4.65625,
"learning_rate": 9.162611163570403e-06,
"loss": 1.1123,
"step": 2025
},
{
"epoch": 2.17,
"grad_norm": 6.21875,
"learning_rate": 9.053994963639406e-06,
"loss": 1.1922,
"step": 2030
},
{
"epoch": 2.17,
"grad_norm": 6.3125,
"learning_rate": 8.945883856105841e-06,
"loss": 1.2439,
"step": 2035
},
{
"epoch": 2.18,
"grad_norm": 5.9375,
"learning_rate": 8.838281265407284e-06,
"loss": 1.1455,
"step": 2040
},
{
"epoch": 2.18,
"grad_norm": 5.09375,
"learning_rate": 8.731190599873912e-06,
"loss": 1.1547,
"step": 2045
},
{
"epoch": 2.19,
"grad_norm": 5.40625,
"learning_rate": 8.624615251620627e-06,
"loss": 1.1167,
"step": 2050
},
{
"epoch": 2.19,
"grad_norm": 6.53125,
"learning_rate": 8.518558596439532e-06,
"loss": 1.1272,
"step": 2055
},
{
"epoch": 2.2,
"grad_norm": 6.4375,
"learning_rate": 8.413023993693075e-06,
"loss": 1.1031,
"step": 2060
},
{
"epoch": 2.2,
"grad_norm": 5.71875,
"learning_rate": 8.308014786207571e-06,
"loss": 1.1688,
"step": 2065
},
{
"epoch": 2.21,
"grad_norm": 4.84375,
"learning_rate": 8.203534300167382e-06,
"loss": 1.1888,
"step": 2070
},
{
"epoch": 2.21,
"grad_norm": 5.1875,
"learning_rate": 8.0995858450095e-06,
"loss": 1.1494,
"step": 2075
},
{
"epoch": 2.22,
"grad_norm": 4.8125,
"learning_rate": 7.99617271331877e-06,
"loss": 1.1709,
"step": 2080
},
{
"epoch": 2.22,
"grad_norm": 5.46875,
"learning_rate": 7.893298180723582e-06,
"loss": 1.1323,
"step": 2085
},
{
"epoch": 2.23,
"grad_norm": 5.96875,
"learning_rate": 7.790965505792078e-06,
"loss": 1.1689,
"step": 2090
},
{
"epoch": 2.23,
"grad_norm": 6.15625,
"learning_rate": 7.689177929929004e-06,
"loss": 1.1852,
"step": 2095
},
{
"epoch": 2.24,
"grad_norm": 5.34375,
"learning_rate": 7.5879386772729625e-06,
"loss": 1.1371,
"step": 2100
},
{
"epoch": 2.25,
"grad_norm": 4.5625,
"learning_rate": 7.4872509545943715e-06,
"loss": 1.0184,
"step": 2105
},
{
"epoch": 2.25,
"grad_norm": 4.78125,
"learning_rate": 7.387117951193792e-06,
"loss": 1.1552,
"step": 2110
},
{
"epoch": 2.26,
"grad_norm": 6.125,
"learning_rate": 7.287542838801003e-06,
"loss": 1.1553,
"step": 2115
},
{
"epoch": 2.26,
"grad_norm": 5.28125,
"learning_rate": 7.188528771474478e-06,
"loss": 1.1062,
"step": 2120
},
{
"epoch": 2.27,
"grad_norm": 5.125,
"learning_rate": 7.090078885501475e-06,
"loss": 1.0785,
"step": 2125
},
{
"epoch": 2.27,
"grad_norm": 4.90625,
"learning_rate": 6.992196299298739e-06,
"loss": 1.0447,
"step": 2130
},
{
"epoch": 2.28,
"grad_norm": 6.375,
"learning_rate": 6.894884113313679e-06,
"loss": 1.1757,
"step": 2135
},
{
"epoch": 2.28,
"grad_norm": 5.6875,
"learning_rate": 6.798145409926191e-06,
"loss": 1.1345,
"step": 2140
},
{
"epoch": 2.29,
"grad_norm": 5.46875,
"learning_rate": 6.701983253351021e-06,
"loss": 1.1477,
"step": 2145
},
{
"epoch": 2.29,
"grad_norm": 5.78125,
"learning_rate": 6.606400689540673e-06,
"loss": 1.1445,
"step": 2150
},
{
"epoch": 2.3,
"grad_norm": 5.0625,
"learning_rate": 6.511400746088986e-06,
"loss": 1.1672,
"step": 2155
},
{
"epoch": 2.3,
"grad_norm": 5.75,
"learning_rate": 6.416986432135161e-06,
"loss": 1.0833,
"step": 2160
},
{
"epoch": 2.31,
"grad_norm": 5.03125,
"learning_rate": 6.323160738268524e-06,
"loss": 1.0998,
"step": 2165
},
{
"epoch": 2.31,
"grad_norm": 4.53125,
"learning_rate": 6.229926636433733e-06,
"loss": 1.0866,
"step": 2170
},
{
"epoch": 2.32,
"grad_norm": 5.96875,
"learning_rate": 6.137287079836681e-06,
"loss": 1.142,
"step": 2175
},
{
"epoch": 2.33,
"grad_norm": 5.15625,
"learning_rate": 6.045245002850952e-06,
"loss": 1.0761,
"step": 2180
},
{
"epoch": 2.33,
"grad_norm": 6.15625,
"learning_rate": 5.95380332092483e-06,
"loss": 1.1538,
"step": 2185
},
{
"epoch": 2.34,
"grad_norm": 4.9375,
"learning_rate": 5.8629649304890235e-06,
"loss": 1.1283,
"step": 2190
},
{
"epoch": 2.34,
"grad_norm": 6.3125,
"learning_rate": 5.772732708864842e-06,
"loss": 1.2129,
"step": 2195
},
{
"epoch": 2.35,
"grad_norm": 5.625,
"learning_rate": 5.683109514173124e-06,
"loss": 1.1755,
"step": 2200
},
{
"epoch": 2.35,
"grad_norm": 5.125,
"learning_rate": 5.594098185243676e-06,
"loss": 1.072,
"step": 2205
},
{
"epoch": 2.36,
"grad_norm": 5.5625,
"learning_rate": 5.505701541525335e-06,
"loss": 1.1341,
"step": 2210
},
{
"epoch": 2.36,
"grad_norm": 5.09375,
"learning_rate": 5.417922382996699e-06,
"loss": 1.1623,
"step": 2215
},
{
"epoch": 2.37,
"grad_norm": 5.875,
"learning_rate": 5.330763490077404e-06,
"loss": 1.09,
"step": 2220
},
{
"epoch": 2.37,
"grad_norm": 7.875,
"learning_rate": 5.244227623540074e-06,
"loss": 1.1628,
"step": 2225
},
{
"epoch": 2.38,
"grad_norm": 5.59375,
"learning_rate": 5.1583175244228785e-06,
"loss": 1.1654,
"step": 2230
},
{
"epoch": 2.38,
"grad_norm": 5.5,
"learning_rate": 5.073035913942678e-06,
"loss": 1.117,
"step": 2235
},
{
"epoch": 2.39,
"grad_norm": 6.28125,
"learning_rate": 4.988385493408871e-06,
"loss": 1.1308,
"step": 2240
},
{
"epoch": 2.39,
"grad_norm": 5.0,
"learning_rate": 4.904368944137783e-06,
"loss": 1.1502,
"step": 2245
},
{
"epoch": 2.4,
"grad_norm": 5.84375,
"learning_rate": 4.82098892736779e-06,
"loss": 1.137,
"step": 2250
},
{
"epoch": 2.41,
"grad_norm": 4.96875,
"learning_rate": 4.738248084174959e-06,
"loss": 1.1274,
"step": 2255
},
{
"epoch": 2.41,
"grad_norm": 6.125,
"learning_rate": 4.656149035389457e-06,
"loss": 1.0886,
"step": 2260
},
{
"epoch": 2.42,
"grad_norm": 4.90625,
"learning_rate": 4.574694381512498e-06,
"loss": 1.2045,
"step": 2265
},
{
"epoch": 2.42,
"grad_norm": 5.875,
"learning_rate": 4.493886702633962e-06,
"loss": 1.1118,
"step": 2270
},
{
"epoch": 2.43,
"grad_norm": 5.21875,
"learning_rate": 4.41372855835071e-06,
"loss": 1.1354,
"step": 2275
},
{
"epoch": 2.43,
"grad_norm": 5.5,
"learning_rate": 4.3342224876854604e-06,
"loss": 1.1385,
"step": 2280
},
{
"epoch": 2.44,
"grad_norm": 5.3125,
"learning_rate": 4.255371009006415e-06,
"loss": 1.2304,
"step": 2285
},
{
"epoch": 2.44,
"grad_norm": 5.8125,
"learning_rate": 4.177176619947451e-06,
"loss": 1.1728,
"step": 2290
},
{
"epoch": 2.45,
"grad_norm": 5.625,
"learning_rate": 4.099641797329015e-06,
"loss": 1.14,
"step": 2295
},
{
"epoch": 2.45,
"grad_norm": 6.15625,
"learning_rate": 4.0227689970796956e-06,
"loss": 1.1576,
"step": 2300
},
{
"epoch": 2.46,
"grad_norm": 5.90625,
"learning_rate": 3.946560654158385e-06,
"loss": 1.1663,
"step": 2305
},
{
"epoch": 2.46,
"grad_norm": 7.5,
"learning_rate": 3.871019182477203e-06,
"loss": 1.2139,
"step": 2310
},
{
"epoch": 2.47,
"grad_norm": 5.09375,
"learning_rate": 3.796146974824988e-06,
"loss": 1.142,
"step": 2315
},
{
"epoch": 2.47,
"grad_norm": 4.8125,
"learning_rate": 3.7219464027915403e-06,
"loss": 1.1319,
"step": 2320
},
{
"epoch": 2.48,
"grad_norm": 5.03125,
"learning_rate": 3.6484198166924973e-06,
"loss": 1.1338,
"step": 2325
},
{
"epoch": 2.49,
"grad_norm": 5.375,
"learning_rate": 3.5755695454948558e-06,
"loss": 1.1259,
"step": 2330
},
{
"epoch": 2.49,
"grad_norm": 6.40625,
"learning_rate": 3.5033978967432437e-06,
"loss": 1.226,
"step": 2335
},
{
"epoch": 2.5,
"grad_norm": 4.96875,
"learning_rate": 3.431907156486791e-06,
"loss": 1.1347,
"step": 2340
},
{
"epoch": 2.5,
"grad_norm": 5.25,
"learning_rate": 3.361099589206751e-06,
"loss": 1.133,
"step": 2345
},
{
"epoch": 2.51,
"grad_norm": 4.5,
"learning_rate": 3.2909774377447606e-06,
"loss": 1.0546,
"step": 2350
},
{
"epoch": 2.51,
"grad_norm": 6.125,
"learning_rate": 3.221542923231774e-06,
"loss": 1.1597,
"step": 2355
},
{
"epoch": 2.52,
"grad_norm": 5.75,
"learning_rate": 3.1527982450177547e-06,
"loss": 1.1044,
"step": 2360
},
{
"epoch": 2.52,
"grad_norm": 6.1875,
"learning_rate": 3.084745580601964e-06,
"loss": 1.1435,
"step": 2365
},
{
"epoch": 2.53,
"grad_norm": 5.5,
"learning_rate": 3.0173870855640246e-06,
"loss": 1.2131,
"step": 2370
},
{
"epoch": 2.53,
"grad_norm": 6.25,
"learning_rate": 2.9507248934956155e-06,
"loss": 1.1718,
"step": 2375
},
{
"epoch": 2.54,
"grad_norm": 4.40625,
"learning_rate": 2.8847611159329072e-06,
"loss": 1.1415,
"step": 2380
},
{
"epoch": 2.54,
"grad_norm": 5.5625,
"learning_rate": 2.819497842289684e-06,
"loss": 1.1762,
"step": 2385
},
{
"epoch": 2.55,
"grad_norm": 5.375,
"learning_rate": 2.754937139791125e-06,
"loss": 1.1139,
"step": 2390
},
{
"epoch": 2.55,
"grad_norm": 5.78125,
"learning_rate": 2.691081053408376e-06,
"loss": 1.1348,
"step": 2395
},
{
"epoch": 2.56,
"grad_norm": 5.59375,
"learning_rate": 2.6279316057937263e-06,
"loss": 1.2093,
"step": 2400
},
{
"epoch": 2.57,
"grad_norm": 6.9375,
"learning_rate": 2.5654907972165853e-06,
"loss": 1.1454,
"step": 2405
},
{
"epoch": 2.57,
"grad_norm": 5.46875,
"learning_rate": 2.50376060550008e-06,
"loss": 1.1167,
"step": 2410
},
{
"epoch": 2.58,
"grad_norm": 5.59375,
"learning_rate": 2.442742985958446e-06,
"loss": 1.1659,
"step": 2415
},
{
"epoch": 2.58,
"grad_norm": 5.125,
"learning_rate": 2.3824398713350597e-06,
"loss": 1.1642,
"step": 2420
},
{
"epoch": 2.59,
"grad_norm": 6.0625,
"learning_rate": 2.3228531717412515e-06,
"loss": 1.1461,
"step": 2425
},
{
"epoch": 2.59,
"grad_norm": 5.15625,
"learning_rate": 2.2639847745957647e-06,
"loss": 1.1391,
"step": 2430
},
{
"epoch": 2.6,
"grad_norm": 5.6875,
"learning_rate": 2.2058365445650126e-06,
"loss": 1.1331,
"step": 2435
},
{
"epoch": 2.6,
"grad_norm": 5.8125,
"learning_rate": 2.1484103235039714e-06,
"loss": 1.108,
"step": 2440
},
{
"epoch": 2.61,
"grad_norm": 5.65625,
"learning_rate": 2.091707930397885e-06,
"loss": 1.142,
"step": 2445
},
{
"epoch": 2.61,
"grad_norm": 6.1875,
"learning_rate": 2.035731161304602e-06,
"loss": 1.0998,
"step": 2450
},
{
"epoch": 2.62,
"grad_norm": 4.40625,
"learning_rate": 1.9804817892977277e-06,
"loss": 1.1734,
"step": 2455
},
{
"epoch": 2.62,
"grad_norm": 6.90625,
"learning_rate": 1.925961564410428e-06,
"loss": 1.2022,
"step": 2460
},
{
"epoch": 2.63,
"grad_norm": 5.46875,
"learning_rate": 1.8721722135800174e-06,
"loss": 1.1813,
"step": 2465
},
{
"epoch": 2.63,
"grad_norm": 5.46875,
"learning_rate": 1.819115440593258e-06,
"loss": 1.1642,
"step": 2470
},
{
"epoch": 2.64,
"grad_norm": 5.71875,
"learning_rate": 1.766792926032365e-06,
"loss": 1.1271,
"step": 2475
},
{
"epoch": 2.65,
"grad_norm": 7.1875,
"learning_rate": 1.715206327221816e-06,
"loss": 1.1829,
"step": 2480
},
{
"epoch": 2.65,
"grad_norm": 5.53125,
"learning_rate": 1.6643572781758181e-06,
"loss": 1.1764,
"step": 2485
},
{
"epoch": 2.66,
"grad_norm": 7.03125,
"learning_rate": 1.614247389546575e-06,
"loss": 1.0975,
"step": 2490
},
{
"epoch": 2.66,
"grad_norm": 5.09375,
"learning_rate": 1.5648782485732566e-06,
"loss": 1.1742,
"step": 2495
},
{
"epoch": 2.67,
"grad_norm": 6.46875,
"learning_rate": 1.5162514190317195e-06,
"loss": 1.1686,
"step": 2500
},
{
"epoch": 2.67,
"grad_norm": 6.0,
"learning_rate": 1.4683684411849947e-06,
"loss": 1.1135,
"step": 2505
},
{
"epoch": 2.68,
"grad_norm": 4.96875,
"learning_rate": 1.4212308317344686e-06,
"loss": 1.1648,
"step": 2510
},
{
"epoch": 2.68,
"grad_norm": 5.96875,
"learning_rate": 1.3748400837718696e-06,
"loss": 1.1331,
"step": 2515
},
{
"epoch": 2.69,
"grad_norm": 5.0,
"learning_rate": 1.32919766673196e-06,
"loss": 1.1018,
"step": 2520
},
{
"epoch": 2.69,
"grad_norm": 6.125,
"learning_rate": 1.2843050263459888e-06,
"loss": 1.1353,
"step": 2525
},
{
"epoch": 2.7,
"grad_norm": 5.8125,
"learning_rate": 1.2401635845959159e-06,
"loss": 1.1662,
"step": 2530
},
{
"epoch": 2.7,
"grad_norm": 6.4375,
"learning_rate": 1.1967747396693401e-06,
"loss": 1.2042,
"step": 2535
},
{
"epoch": 2.71,
"grad_norm": 5.59375,
"learning_rate": 1.1541398659152464e-06,
"loss": 1.132,
"step": 2540
},
{
"epoch": 2.71,
"grad_norm": 6.5625,
"learning_rate": 1.1122603138004422e-06,
"loss": 1.1555,
"step": 2545
},
{
"epoch": 2.72,
"grad_norm": 5.8125,
"learning_rate": 1.0711374098667987e-06,
"loss": 1.1753,
"step": 2550
},
{
"epoch": 2.73,
"grad_norm": 5.46875,
"learning_rate": 1.0307724566892385e-06,
"loss": 1.1657,
"step": 2555
},
{
"epoch": 2.73,
"grad_norm": 5.0,
"learning_rate": 9.911667328344477e-07,
"loss": 1.1786,
"step": 2560
},
{
"epoch": 2.74,
"grad_norm": 5.03125,
"learning_rate": 9.523214928204127e-07,
"loss": 1.1759,
"step": 2565
},
{
"epoch": 2.74,
"grad_norm": 6.25,
"learning_rate": 9.142379670766532e-07,
"loss": 1.19,
"step": 2570
},
{
"epoch": 2.75,
"grad_norm": 6.0,
"learning_rate": 8.76917361905269e-07,
"loss": 1.1569,
"step": 2575
},
{
"epoch": 2.75,
"grad_norm": 7.28125,
"learning_rate": 8.403608594427237e-07,
"loss": 1.1427,
"step": 2580
},
{
"epoch": 2.76,
"grad_norm": 5.1875,
"learning_rate": 8.045696176223888e-07,
"loss": 1.1103,
"step": 2585
},
{
"epoch": 2.76,
"grad_norm": 5.75,
"learning_rate": 7.695447701378866e-07,
"loss": 1.1479,
"step": 2590
},
{
"epoch": 2.77,
"grad_norm": 5.0,
"learning_rate": 7.352874264071579e-07,
"loss": 1.1034,
"step": 2595
},
{
"epoch": 2.77,
"grad_norm": 5.84375,
"learning_rate": 7.017986715373515e-07,
"loss": 1.0887,
"step": 2600
},
{
"epoch": 2.78,
"grad_norm": 6.21875,
"learning_rate": 6.69079566290412e-07,
"loss": 1.2236,
"step": 2605
},
{
"epoch": 2.78,
"grad_norm": 6.46875,
"learning_rate": 6.371311470495162e-07,
"loss": 1.1793,
"step": 2610
},
{
"epoch": 2.79,
"grad_norm": 5.15625,
"learning_rate": 6.059544257862343e-07,
"loss": 1.0842,
"step": 2615
},
{
"epoch": 2.79,
"grad_norm": 4.71875,
"learning_rate": 5.75550390028462e-07,
"loss": 1.077,
"step": 2620
},
{
"epoch": 2.8,
"grad_norm": 6.03125,
"learning_rate": 5.459200028291617e-07,
"loss": 1.1295,
"step": 2625
},
{
"epoch": 2.81,
"grad_norm": 5.90625,
"learning_rate": 5.170642027358341e-07,
"loss": 1.1019,
"step": 2630
},
{
"epoch": 2.81,
"grad_norm": 5.96875,
"learning_rate": 4.889839037608202e-07,
"loss": 1.1931,
"step": 2635
},
{
"epoch": 2.82,
"grad_norm": 5.75,
"learning_rate": 4.616799953523182e-07,
"loss": 1.1147,
"step": 2640
},
{
"epoch": 2.82,
"grad_norm": 5.8125,
"learning_rate": 4.351533423662318e-07,
"loss": 1.1505,
"step": 2645
},
{
"epoch": 2.83,
"grad_norm": 5.375,
"learning_rate": 4.094047850387667e-07,
"loss": 1.1301,
"step": 2650
},
{
"epoch": 2.83,
"grad_norm": 5.6875,
"learning_rate": 3.8443513895981e-07,
"loss": 1.1204,
"step": 2655
},
{
"epoch": 2.84,
"grad_norm": 5.6875,
"learning_rate": 3.6024519504711806e-07,
"loss": 1.0925,
"step": 2660
},
{
"epoch": 2.84,
"grad_norm": 5.84375,
"learning_rate": 3.3683571952123906e-07,
"loss": 1.1177,
"step": 2665
},
{
"epoch": 2.85,
"grad_norm": 6.09375,
"learning_rate": 3.142074538812545e-07,
"loss": 1.1465,
"step": 2670
},
{
"epoch": 2.85,
"grad_norm": 5.25,
"learning_rate": 2.9236111488129545e-07,
"loss": 1.1202,
"step": 2675
},
{
"epoch": 2.86,
"grad_norm": 6.84375,
"learning_rate": 2.712973945078301e-07,
"loss": 1.1668,
"step": 2680
},
{
"epoch": 2.86,
"grad_norm": 5.09375,
"learning_rate": 2.510169599577533e-07,
"loss": 1.1126,
"step": 2685
},
{
"epoch": 2.87,
"grad_norm": 6.28125,
"learning_rate": 2.315204536172455e-07,
"loss": 1.1607,
"step": 2690
},
{
"epoch": 2.87,
"grad_norm": 5.84375,
"learning_rate": 2.128084930414359e-07,
"loss": 1.1588,
"step": 2695
},
{
"epoch": 2.88,
"grad_norm": 6.90625,
"learning_rate": 1.9488167093482667e-07,
"loss": 1.1142,
"step": 2700
},
{
"epoch": 2.89,
"grad_norm": 5.90625,
"learning_rate": 1.7774055513253006e-07,
"loss": 1.1313,
"step": 2705
},
{
"epoch": 2.89,
"grad_norm": 5.46875,
"learning_rate": 1.613856885822801e-07,
"loss": 1.0984,
"step": 2710
},
{
"epoch": 2.9,
"grad_norm": 5.34375,
"learning_rate": 1.4581758932722956e-07,
"loss": 1.0768,
"step": 2715
},
{
"epoch": 2.9,
"grad_norm": 5.375,
"learning_rate": 1.3103675048954934e-07,
"loss": 1.1069,
"step": 2720
},
{
"epoch": 2.91,
"grad_norm": 5.25,
"learning_rate": 1.1704364025479642e-07,
"loss": 1.1131,
"step": 2725
},
{
"epoch": 2.91,
"grad_norm": 6.46875,
"learning_rate": 1.0383870185710076e-07,
"loss": 1.1476,
"step": 2730
},
{
"epoch": 2.92,
"grad_norm": 6.71875,
"learning_rate": 9.142235356510987e-08,
"loss": 1.1705,
"step": 2735
},
{
"epoch": 2.92,
"grad_norm": 5.0,
"learning_rate": 7.979498866874935e-08,
"loss": 1.069,
"step": 2740
},
{
"epoch": 2.93,
"grad_norm": 5.4375,
"learning_rate": 6.895697546676349e-08,
"loss": 1.1401,
"step": 2745
},
{
"epoch": 2.93,
"grad_norm": 6.65625,
"learning_rate": 5.890865725504402e-08,
"loss": 1.119,
"step": 2750
},
{
"epoch": 2.94,
"grad_norm": 6.09375,
"learning_rate": 4.9650352315766535e-08,
"loss": 1.1177,
"step": 2755
},
{
"epoch": 2.94,
"grad_norm": 5.6875,
"learning_rate": 4.118235390730141e-08,
"loss": 1.0892,
"step": 2760
},
{
"epoch": 2.95,
"grad_norm": 8.9375,
"learning_rate": 3.350493025492396e-08,
"loss": 1.1853,
"step": 2765
},
{
"epoch": 2.95,
"grad_norm": 5.21875,
"learning_rate": 2.661832454232127e-08,
"loss": 1.1004,
"step": 2770
},
{
"epoch": 2.96,
"grad_norm": 5.71875,
"learning_rate": 2.0522754903892772e-08,
"loss": 1.113,
"step": 2775
},
{
"epoch": 2.97,
"grad_norm": 5.5,
"learning_rate": 1.5218414417833558e-08,
"loss": 1.1525,
"step": 2780
},
{
"epoch": 2.97,
"grad_norm": 5.34375,
"learning_rate": 1.0705471100022624e-08,
"loss": 1.1145,
"step": 2785
},
{
"epoch": 2.98,
"grad_norm": 4.46875,
"learning_rate": 6.9840678986993205e-09,
"loss": 1.0893,
"step": 2790
},
{
"epoch": 2.98,
"grad_norm": 6.3125,
"learning_rate": 4.054322689936441e-09,
"loss": 1.1528,
"step": 2795
},
{
"epoch": 2.99,
"grad_norm": 5.6875,
"learning_rate": 1.9163282739126376e-09,
"loss": 1.1335,
"step": 2800
},
{
"epoch": 2.99,
"grad_norm": 7.53125,
"learning_rate": 5.701523719536805e-10,
"loss": 1.1066,
"step": 2805
},
{
"epoch": 3.0,
"grad_norm": 5.875,
"learning_rate": 1.5837624414705154e-11,
"loss": 1.2276,
"step": 2810
},
{
"epoch": 3.0,
"step": 2811,
"total_flos": 3.2918248209933926e+17,
"train_loss": 1.199945468658317,
"train_runtime": 5291.6066,
"train_samples_per_second": 17.008,
"train_steps_per_second": 0.531
}
],
"logging_steps": 5,
"max_steps": 2811,
"num_input_tokens_seen": 0,
"num_train_epochs": 3,
"save_steps": 30000,
"total_flos": 3.2918248209933926e+17,
"train_batch_size": 8,
"trial_name": null,
"trial_params": null
}