coding_llamaduo_60k / trainer_state.json
chansung's picture
Model save
d51d9ca verified
raw
history blame contribute delete
No virus
85.7 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 10.0,
"eval_steps": 500,
"global_step": 2520,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.003968253968253968,
"grad_norm": 458.0,
"learning_rate": 7.936507936507937e-07,
"loss": 46.2081,
"step": 1
},
{
"epoch": 0.01984126984126984,
"grad_norm": 342.0,
"learning_rate": 3.968253968253968e-06,
"loss": 43.6637,
"step": 5
},
{
"epoch": 0.03968253968253968,
"grad_norm": 304.0,
"learning_rate": 7.936507936507936e-06,
"loss": 42.8752,
"step": 10
},
{
"epoch": 0.05952380952380952,
"grad_norm": 205.0,
"learning_rate": 1.1904761904761905e-05,
"loss": 35.6419,
"step": 15
},
{
"epoch": 0.07936507936507936,
"grad_norm": 139.0,
"learning_rate": 1.5873015873015872e-05,
"loss": 26.112,
"step": 20
},
{
"epoch": 0.0992063492063492,
"grad_norm": 24.75,
"learning_rate": 1.984126984126984e-05,
"loss": 21.3785,
"step": 25
},
{
"epoch": 0.11904761904761904,
"grad_norm": 23.75,
"learning_rate": 2.380952380952381e-05,
"loss": 18.754,
"step": 30
},
{
"epoch": 0.1388888888888889,
"grad_norm": 11.5625,
"learning_rate": 2.777777777777778e-05,
"loss": 17.301,
"step": 35
},
{
"epoch": 0.15873015873015872,
"grad_norm": 5.46875,
"learning_rate": 3.1746031746031745e-05,
"loss": 16.1126,
"step": 40
},
{
"epoch": 0.17857142857142858,
"grad_norm": 4.0,
"learning_rate": 3.571428571428572e-05,
"loss": 15.745,
"step": 45
},
{
"epoch": 0.1984126984126984,
"grad_norm": 4.84375,
"learning_rate": 3.968253968253968e-05,
"loss": 15.3598,
"step": 50
},
{
"epoch": 0.21825396825396826,
"grad_norm": 7.0,
"learning_rate": 4.3650793650793655e-05,
"loss": 14.6923,
"step": 55
},
{
"epoch": 0.23809523809523808,
"grad_norm": 11.5625,
"learning_rate": 4.761904761904762e-05,
"loss": 13.8362,
"step": 60
},
{
"epoch": 0.25793650793650796,
"grad_norm": 23.625,
"learning_rate": 5.158730158730159e-05,
"loss": 12.2662,
"step": 65
},
{
"epoch": 0.2777777777777778,
"grad_norm": 40.0,
"learning_rate": 5.555555555555556e-05,
"loss": 8.9932,
"step": 70
},
{
"epoch": 0.2976190476190476,
"grad_norm": 12.0,
"learning_rate": 5.9523809523809524e-05,
"loss": 3.4302,
"step": 75
},
{
"epoch": 0.31746031746031744,
"grad_norm": 2.65625,
"learning_rate": 6.349206349206349e-05,
"loss": 1.7075,
"step": 80
},
{
"epoch": 0.3373015873015873,
"grad_norm": 1.84375,
"learning_rate": 6.746031746031747e-05,
"loss": 1.4439,
"step": 85
},
{
"epoch": 0.35714285714285715,
"grad_norm": 1.578125,
"learning_rate": 7.142857142857143e-05,
"loss": 1.2581,
"step": 90
},
{
"epoch": 0.376984126984127,
"grad_norm": 1.7265625,
"learning_rate": 7.53968253968254e-05,
"loss": 1.1475,
"step": 95
},
{
"epoch": 0.3968253968253968,
"grad_norm": 1.046875,
"learning_rate": 7.936507936507937e-05,
"loss": 1.0353,
"step": 100
},
{
"epoch": 0.4166666666666667,
"grad_norm": 1.0859375,
"learning_rate": 8.333333333333334e-05,
"loss": 0.9727,
"step": 105
},
{
"epoch": 0.4365079365079365,
"grad_norm": 1.2578125,
"learning_rate": 8.730158730158731e-05,
"loss": 0.9454,
"step": 110
},
{
"epoch": 0.45634920634920634,
"grad_norm": 2.90625,
"learning_rate": 9.126984126984128e-05,
"loss": 0.8831,
"step": 115
},
{
"epoch": 0.47619047619047616,
"grad_norm": 4.40625,
"learning_rate": 9.523809523809524e-05,
"loss": 0.8701,
"step": 120
},
{
"epoch": 0.49603174603174605,
"grad_norm": 3.21875,
"learning_rate": 9.920634920634922e-05,
"loss": 0.8496,
"step": 125
},
{
"epoch": 0.5158730158730159,
"grad_norm": 31.75,
"learning_rate": 0.00010317460317460319,
"loss": 0.8262,
"step": 130
},
{
"epoch": 0.5357142857142857,
"grad_norm": 0.9765625,
"learning_rate": 0.00010714285714285715,
"loss": 0.8224,
"step": 135
},
{
"epoch": 0.5555555555555556,
"grad_norm": 3.25,
"learning_rate": 0.00011111111111111112,
"loss": 0.8101,
"step": 140
},
{
"epoch": 0.5753968253968254,
"grad_norm": 1.1171875,
"learning_rate": 0.00011507936507936508,
"loss": 0.797,
"step": 145
},
{
"epoch": 0.5952380952380952,
"grad_norm": 1.0625,
"learning_rate": 0.00011904761904761905,
"loss": 0.7696,
"step": 150
},
{
"epoch": 0.6150793650793651,
"grad_norm": 5.9375,
"learning_rate": 0.00012301587301587303,
"loss": 0.7579,
"step": 155
},
{
"epoch": 0.6349206349206349,
"grad_norm": 3.390625,
"learning_rate": 0.00012698412698412698,
"loss": 0.7647,
"step": 160
},
{
"epoch": 0.6547619047619048,
"grad_norm": 1.59375,
"learning_rate": 0.00013095238095238096,
"loss": 0.7303,
"step": 165
},
{
"epoch": 0.6746031746031746,
"grad_norm": 2.953125,
"learning_rate": 0.00013492063492063494,
"loss": 0.7226,
"step": 170
},
{
"epoch": 0.6944444444444444,
"grad_norm": 1.265625,
"learning_rate": 0.0001388888888888889,
"loss": 0.711,
"step": 175
},
{
"epoch": 0.7142857142857143,
"grad_norm": 17.5,
"learning_rate": 0.00014285714285714287,
"loss": 0.7247,
"step": 180
},
{
"epoch": 0.7341269841269841,
"grad_norm": 3.25,
"learning_rate": 0.00014682539682539682,
"loss": 0.7103,
"step": 185
},
{
"epoch": 0.753968253968254,
"grad_norm": 2.265625,
"learning_rate": 0.0001507936507936508,
"loss": 0.726,
"step": 190
},
{
"epoch": 0.7738095238095238,
"grad_norm": 1.3671875,
"learning_rate": 0.00015476190476190478,
"loss": 0.6925,
"step": 195
},
{
"epoch": 0.7936507936507936,
"grad_norm": 1.4375,
"learning_rate": 0.00015873015873015873,
"loss": 0.7095,
"step": 200
},
{
"epoch": 0.8134920634920635,
"grad_norm": 0.5703125,
"learning_rate": 0.0001626984126984127,
"loss": 0.671,
"step": 205
},
{
"epoch": 0.8333333333333334,
"grad_norm": 0.6015625,
"learning_rate": 0.0001666666666666667,
"loss": 0.6731,
"step": 210
},
{
"epoch": 0.8531746031746031,
"grad_norm": 0.54296875,
"learning_rate": 0.00017063492063492064,
"loss": 0.6706,
"step": 215
},
{
"epoch": 0.873015873015873,
"grad_norm": 3.078125,
"learning_rate": 0.00017460317460317462,
"loss": 0.6931,
"step": 220
},
{
"epoch": 0.8928571428571429,
"grad_norm": 0.8671875,
"learning_rate": 0.0001785714285714286,
"loss": 0.6979,
"step": 225
},
{
"epoch": 0.9126984126984127,
"grad_norm": 0.80078125,
"learning_rate": 0.00018253968253968255,
"loss": 0.6797,
"step": 230
},
{
"epoch": 0.9325396825396826,
"grad_norm": 0.828125,
"learning_rate": 0.00018650793650793653,
"loss": 0.6679,
"step": 235
},
{
"epoch": 0.9523809523809523,
"grad_norm": 2.8125,
"learning_rate": 0.00019047619047619048,
"loss": 0.6605,
"step": 240
},
{
"epoch": 0.9722222222222222,
"grad_norm": 1.5703125,
"learning_rate": 0.00019444444444444446,
"loss": 0.6615,
"step": 245
},
{
"epoch": 0.9920634920634921,
"grad_norm": 0.98046875,
"learning_rate": 0.00019841269841269844,
"loss": 0.6618,
"step": 250
},
{
"epoch": 1.0,
"eval_loss": 1.2071309089660645,
"eval_runtime": 0.5456,
"eval_samples_per_second": 3.666,
"eval_steps_per_second": 1.833,
"step": 252
},
{
"epoch": 1.0119047619047619,
"grad_norm": 1.578125,
"learning_rate": 0.0001999991365731819,
"loss": 0.6218,
"step": 255
},
{
"epoch": 1.0317460317460316,
"grad_norm": 3.71875,
"learning_rate": 0.00019999386012995552,
"loss": 0.6584,
"step": 260
},
{
"epoch": 1.0515873015873016,
"grad_norm": 1.828125,
"learning_rate": 0.0001999837871778622,
"loss": 0.635,
"step": 265
},
{
"epoch": 1.0714285714285714,
"grad_norm": 1.4375,
"learning_rate": 0.00019996891820008164,
"loss": 0.6303,
"step": 270
},
{
"epoch": 1.0912698412698412,
"grad_norm": 1.03125,
"learning_rate": 0.00019994925390984946,
"loss": 0.6337,
"step": 275
},
{
"epoch": 1.1111111111111112,
"grad_norm": 1.265625,
"learning_rate": 0.00019992479525042303,
"loss": 0.6218,
"step": 280
},
{
"epoch": 1.130952380952381,
"grad_norm": 0.97265625,
"learning_rate": 0.00019989554339503612,
"loss": 0.6194,
"step": 285
},
{
"epoch": 1.1507936507936507,
"grad_norm": 0.369140625,
"learning_rate": 0.0001998614997468427,
"loss": 0.613,
"step": 290
},
{
"epoch": 1.1706349206349207,
"grad_norm": 1.5078125,
"learning_rate": 0.00019982266593884972,
"loss": 0.6137,
"step": 295
},
{
"epoch": 1.1904761904761905,
"grad_norm": 1.6875,
"learning_rate": 0.0001997790438338385,
"loss": 0.6055,
"step": 300
},
{
"epoch": 1.2103174603174602,
"grad_norm": 1.4921875,
"learning_rate": 0.00019973063552427572,
"loss": 0.6144,
"step": 305
},
{
"epoch": 1.2301587301587302,
"grad_norm": 0.78515625,
"learning_rate": 0.00019967744333221278,
"loss": 0.6101,
"step": 310
},
{
"epoch": 1.25,
"grad_norm": 0.384765625,
"learning_rate": 0.00019961946980917456,
"loss": 0.6202,
"step": 315
},
{
"epoch": 1.2698412698412698,
"grad_norm": 0.451171875,
"learning_rate": 0.00019955671773603696,
"loss": 0.6088,
"step": 320
},
{
"epoch": 1.2896825396825398,
"grad_norm": 0.76171875,
"learning_rate": 0.00019948919012289357,
"loss": 0.5923,
"step": 325
},
{
"epoch": 1.3095238095238095,
"grad_norm": 0.5859375,
"learning_rate": 0.0001994168902089112,
"loss": 0.6051,
"step": 330
},
{
"epoch": 1.3293650793650793,
"grad_norm": 2.015625,
"learning_rate": 0.00019933982146217463,
"loss": 0.5887,
"step": 335
},
{
"epoch": 1.3492063492063493,
"grad_norm": 1.4765625,
"learning_rate": 0.00019925798757952,
"loss": 0.5903,
"step": 340
},
{
"epoch": 1.369047619047619,
"grad_norm": 0.921875,
"learning_rate": 0.00019917139248635786,
"loss": 0.5895,
"step": 345
},
{
"epoch": 1.3888888888888888,
"grad_norm": 0.61328125,
"learning_rate": 0.00019908004033648453,
"loss": 0.616,
"step": 350
},
{
"epoch": 1.4087301587301586,
"grad_norm": 0.416015625,
"learning_rate": 0.00019898393551188304,
"loss": 0.5878,
"step": 355
},
{
"epoch": 1.4285714285714286,
"grad_norm": 0.55859375,
"learning_rate": 0.00019888308262251285,
"loss": 0.589,
"step": 360
},
{
"epoch": 1.4484126984126984,
"grad_norm": 1.1015625,
"learning_rate": 0.0001987774865060888,
"loss": 0.5866,
"step": 365
},
{
"epoch": 1.4682539682539684,
"grad_norm": 0.431640625,
"learning_rate": 0.00019866715222784895,
"loss": 0.5823,
"step": 370
},
{
"epoch": 1.4880952380952381,
"grad_norm": 0.57421875,
"learning_rate": 0.0001985520850803117,
"loss": 0.586,
"step": 375
},
{
"epoch": 1.507936507936508,
"grad_norm": 0.78125,
"learning_rate": 0.0001984322905830219,
"loss": 0.5844,
"step": 380
},
{
"epoch": 1.5277777777777777,
"grad_norm": 1.1953125,
"learning_rate": 0.00019830777448228603,
"loss": 0.5813,
"step": 385
},
{
"epoch": 1.5476190476190477,
"grad_norm": 1.21875,
"learning_rate": 0.0001981785427508966,
"loss": 0.5916,
"step": 390
},
{
"epoch": 1.5674603174603174,
"grad_norm": 0.81640625,
"learning_rate": 0.0001980446015878457,
"loss": 0.5879,
"step": 395
},
{
"epoch": 1.5873015873015874,
"grad_norm": 5.0625,
"learning_rate": 0.00019790595741802757,
"loss": 0.5762,
"step": 400
},
{
"epoch": 1.6071428571428572,
"grad_norm": 0.88671875,
"learning_rate": 0.00019776261689193048,
"loss": 0.5798,
"step": 405
},
{
"epoch": 1.626984126984127,
"grad_norm": 1.015625,
"learning_rate": 0.00019761458688531756,
"loss": 0.5942,
"step": 410
},
{
"epoch": 1.6468253968253967,
"grad_norm": 0.46875,
"learning_rate": 0.00019746187449889725,
"loss": 0.5794,
"step": 415
},
{
"epoch": 1.6666666666666665,
"grad_norm": 0.6171875,
"learning_rate": 0.00019730448705798239,
"loss": 0.5885,
"step": 420
},
{
"epoch": 1.6865079365079365,
"grad_norm": 0.8359375,
"learning_rate": 0.00019714243211213915,
"loss": 0.572,
"step": 425
},
{
"epoch": 1.7063492063492065,
"grad_norm": 0.5625,
"learning_rate": 0.0001969757174348246,
"loss": 0.5707,
"step": 430
},
{
"epoch": 1.7261904761904763,
"grad_norm": 0.80078125,
"learning_rate": 0.00019680435102301412,
"loss": 0.5813,
"step": 435
},
{
"epoch": 1.746031746031746,
"grad_norm": 0.42578125,
"learning_rate": 0.0001966283410968174,
"loss": 0.584,
"step": 440
},
{
"epoch": 1.7658730158730158,
"grad_norm": 0.65625,
"learning_rate": 0.00019644769609908466,
"loss": 0.5634,
"step": 445
},
{
"epoch": 1.7857142857142856,
"grad_norm": 0.48828125,
"learning_rate": 0.0001962624246950012,
"loss": 0.5764,
"step": 450
},
{
"epoch": 1.8055555555555556,
"grad_norm": 0.365234375,
"learning_rate": 0.00019607253577167205,
"loss": 0.5763,
"step": 455
},
{
"epoch": 1.8253968253968254,
"grad_norm": 0.435546875,
"learning_rate": 0.0001958780384376955,
"loss": 0.5704,
"step": 460
},
{
"epoch": 1.8452380952380953,
"grad_norm": 0.765625,
"learning_rate": 0.0001956789420227262,
"loss": 0.5748,
"step": 465
},
{
"epoch": 1.8650793650793651,
"grad_norm": 0.58984375,
"learning_rate": 0.00019547525607702774,
"loss": 0.5647,
"step": 470
},
{
"epoch": 1.8849206349206349,
"grad_norm": 0.90234375,
"learning_rate": 0.00019526699037101454,
"loss": 0.5781,
"step": 475
},
{
"epoch": 1.9047619047619047,
"grad_norm": 0.65625,
"learning_rate": 0.0001950541548947829,
"loss": 0.5906,
"step": 480
},
{
"epoch": 1.9246031746031746,
"grad_norm": 0.390625,
"learning_rate": 0.00019483675985763226,
"loss": 0.574,
"step": 485
},
{
"epoch": 1.9444444444444444,
"grad_norm": 0.42578125,
"learning_rate": 0.00019461481568757506,
"loss": 0.5732,
"step": 490
},
{
"epoch": 1.9642857142857144,
"grad_norm": 0.8515625,
"learning_rate": 0.00019438833303083678,
"loss": 0.595,
"step": 495
},
{
"epoch": 1.9841269841269842,
"grad_norm": 0.81640625,
"learning_rate": 0.00019415732275134513,
"loss": 0.5731,
"step": 500
},
{
"epoch": 2.0,
"eval_loss": 1.1435573101043701,
"eval_runtime": 0.545,
"eval_samples_per_second": 3.67,
"eval_steps_per_second": 1.835,
"step": 504
},
{
"epoch": 2.003968253968254,
"grad_norm": 0.404296875,
"learning_rate": 0.00019392179593020906,
"loss": 0.5499,
"step": 505
},
{
"epoch": 2.0238095238095237,
"grad_norm": 0.369140625,
"learning_rate": 0.0001936817638651871,
"loss": 0.5257,
"step": 510
},
{
"epoch": 2.0436507936507935,
"grad_norm": 0.78515625,
"learning_rate": 0.00019343723807014544,
"loss": 0.5286,
"step": 515
},
{
"epoch": 2.0634920634920633,
"grad_norm": 1.921875,
"learning_rate": 0.0001931882302745057,
"loss": 0.5322,
"step": 520
},
{
"epoch": 2.0833333333333335,
"grad_norm": 0.8046875,
"learning_rate": 0.00019293475242268223,
"loss": 0.5154,
"step": 525
},
{
"epoch": 2.1031746031746033,
"grad_norm": 1.7578125,
"learning_rate": 0.00019267681667350928,
"loss": 0.5199,
"step": 530
},
{
"epoch": 2.123015873015873,
"grad_norm": 2.21875,
"learning_rate": 0.0001924144353996575,
"loss": 0.5328,
"step": 535
},
{
"epoch": 2.142857142857143,
"grad_norm": 2.03125,
"learning_rate": 0.00019214762118704076,
"loss": 0.529,
"step": 540
},
{
"epoch": 2.1626984126984126,
"grad_norm": 1.078125,
"learning_rate": 0.00019187638683421226,
"loss": 0.5325,
"step": 545
},
{
"epoch": 2.1825396825396823,
"grad_norm": 1.71875,
"learning_rate": 0.00019160074535175058,
"loss": 0.5314,
"step": 550
},
{
"epoch": 2.2023809523809526,
"grad_norm": 1.40625,
"learning_rate": 0.00019132070996163568,
"loss": 0.5414,
"step": 555
},
{
"epoch": 2.2222222222222223,
"grad_norm": 4.0,
"learning_rate": 0.0001910362940966147,
"loss": 0.5265,
"step": 560
},
{
"epoch": 2.242063492063492,
"grad_norm": 1.1953125,
"learning_rate": 0.00019074751139955737,
"loss": 0.5357,
"step": 565
},
{
"epoch": 2.261904761904762,
"grad_norm": 0.921875,
"learning_rate": 0.00019045437572280194,
"loss": 0.5291,
"step": 570
},
{
"epoch": 2.2817460317460316,
"grad_norm": 0.5546875,
"learning_rate": 0.00019015690112749044,
"loss": 0.5209,
"step": 575
},
{
"epoch": 2.3015873015873014,
"grad_norm": 0.8125,
"learning_rate": 0.0001898551018828944,
"loss": 0.5152,
"step": 580
},
{
"epoch": 2.3214285714285716,
"grad_norm": 0.51953125,
"learning_rate": 0.0001895489924657301,
"loss": 0.525,
"step": 585
},
{
"epoch": 2.3412698412698414,
"grad_norm": 0.4140625,
"learning_rate": 0.0001892385875594645,
"loss": 0.5308,
"step": 590
},
{
"epoch": 2.361111111111111,
"grad_norm": 0.431640625,
"learning_rate": 0.00018892390205361062,
"loss": 0.5136,
"step": 595
},
{
"epoch": 2.380952380952381,
"grad_norm": 0.75390625,
"learning_rate": 0.00018860495104301345,
"loss": 0.5228,
"step": 600
},
{
"epoch": 2.4007936507936507,
"grad_norm": 0.37109375,
"learning_rate": 0.00018828174982712587,
"loss": 0.5294,
"step": 605
},
{
"epoch": 2.4206349206349205,
"grad_norm": 0.7265625,
"learning_rate": 0.0001879543139092747,
"loss": 0.5259,
"step": 610
},
{
"epoch": 2.4404761904761907,
"grad_norm": 0.482421875,
"learning_rate": 0.00018762265899591722,
"loss": 0.5203,
"step": 615
},
{
"epoch": 2.4603174603174605,
"grad_norm": 0.349609375,
"learning_rate": 0.00018728680099588748,
"loss": 0.5241,
"step": 620
},
{
"epoch": 2.4801587301587302,
"grad_norm": 0.51171875,
"learning_rate": 0.00018694675601963338,
"loss": 0.5314,
"step": 625
},
{
"epoch": 2.5,
"grad_norm": 0.458984375,
"learning_rate": 0.00018660254037844388,
"loss": 0.5272,
"step": 630
},
{
"epoch": 2.5198412698412698,
"grad_norm": 0.50390625,
"learning_rate": 0.0001862541705836665,
"loss": 0.5253,
"step": 635
},
{
"epoch": 2.5396825396825395,
"grad_norm": 0.34375,
"learning_rate": 0.00018590166334591531,
"loss": 0.5283,
"step": 640
},
{
"epoch": 2.5595238095238093,
"grad_norm": 0.734375,
"learning_rate": 0.00018554503557426948,
"loss": 0.5236,
"step": 645
},
{
"epoch": 2.5793650793650795,
"grad_norm": 0.68359375,
"learning_rate": 0.000185184304375462,
"loss": 0.5308,
"step": 650
},
{
"epoch": 2.5992063492063493,
"grad_norm": 0.3046875,
"learning_rate": 0.0001848194870530592,
"loss": 0.5281,
"step": 655
},
{
"epoch": 2.619047619047619,
"grad_norm": 0.31640625,
"learning_rate": 0.0001844506011066308,
"loss": 0.5229,
"step": 660
},
{
"epoch": 2.638888888888889,
"grad_norm": 0.396484375,
"learning_rate": 0.00018407766423091034,
"loss": 0.5169,
"step": 665
},
{
"epoch": 2.6587301587301586,
"grad_norm": 0.38671875,
"learning_rate": 0.00018370069431494646,
"loss": 0.5277,
"step": 670
},
{
"epoch": 2.678571428571429,
"grad_norm": 0.345703125,
"learning_rate": 0.0001833197094412449,
"loss": 0.5152,
"step": 675
},
{
"epoch": 2.6984126984126986,
"grad_norm": 0.380859375,
"learning_rate": 0.00018293472788490095,
"loss": 0.5286,
"step": 680
},
{
"epoch": 2.7182539682539684,
"grad_norm": 0.306640625,
"learning_rate": 0.00018254576811272294,
"loss": 0.5177,
"step": 685
},
{
"epoch": 2.738095238095238,
"grad_norm": 0.59765625,
"learning_rate": 0.00018215284878234642,
"loss": 0.5308,
"step": 690
},
{
"epoch": 2.757936507936508,
"grad_norm": 0.45703125,
"learning_rate": 0.00018175598874133904,
"loss": 0.5173,
"step": 695
},
{
"epoch": 2.7777777777777777,
"grad_norm": 0.95703125,
"learning_rate": 0.00018135520702629675,
"loss": 0.5236,
"step": 700
},
{
"epoch": 2.7976190476190474,
"grad_norm": 0.7578125,
"learning_rate": 0.0001809505228619304,
"loss": 0.524,
"step": 705
},
{
"epoch": 2.817460317460317,
"grad_norm": 0.49609375,
"learning_rate": 0.0001805419556601437,
"loss": 0.5396,
"step": 710
},
{
"epoch": 2.8373015873015874,
"grad_norm": 0.376953125,
"learning_rate": 0.00018012952501910201,
"loss": 0.5259,
"step": 715
},
{
"epoch": 2.857142857142857,
"grad_norm": 0.35546875,
"learning_rate": 0.00017971325072229226,
"loss": 0.534,
"step": 720
},
{
"epoch": 2.876984126984127,
"grad_norm": 0.470703125,
"learning_rate": 0.00017929315273757402,
"loss": 0.523,
"step": 725
},
{
"epoch": 2.8968253968253967,
"grad_norm": 0.396484375,
"learning_rate": 0.0001788692512162216,
"loss": 0.5138,
"step": 730
},
{
"epoch": 2.9166666666666665,
"grad_norm": 0.466796875,
"learning_rate": 0.00017844156649195759,
"loss": 0.533,
"step": 735
},
{
"epoch": 2.9365079365079367,
"grad_norm": 0.423828125,
"learning_rate": 0.00017801011907997725,
"loss": 0.53,
"step": 740
},
{
"epoch": 2.9563492063492065,
"grad_norm": 0.462890625,
"learning_rate": 0.0001775749296759647,
"loss": 0.5302,
"step": 745
},
{
"epoch": 2.9761904761904763,
"grad_norm": 0.640625,
"learning_rate": 0.0001771360191551,
"loss": 0.5436,
"step": 750
},
{
"epoch": 2.996031746031746,
"grad_norm": 1.0703125,
"learning_rate": 0.00017669340857105777,
"loss": 0.5198,
"step": 755
},
{
"epoch": 3.0,
"eval_loss": 1.1346014738082886,
"eval_runtime": 0.5471,
"eval_samples_per_second": 3.655,
"eval_steps_per_second": 1.828,
"step": 756
},
{
"epoch": 3.015873015873016,
"grad_norm": 0.7734375,
"learning_rate": 0.00017624711915499764,
"loss": 0.481,
"step": 760
},
{
"epoch": 3.0357142857142856,
"grad_norm": 0.5390625,
"learning_rate": 0.0001757971723145453,
"loss": 0.4658,
"step": 765
},
{
"epoch": 3.0555555555555554,
"grad_norm": 0.58203125,
"learning_rate": 0.00017534358963276607,
"loss": 0.4725,
"step": 770
},
{
"epoch": 3.0753968253968256,
"grad_norm": 0.59765625,
"learning_rate": 0.0001748863928671294,
"loss": 0.4651,
"step": 775
},
{
"epoch": 3.0952380952380953,
"grad_norm": 0.78515625,
"learning_rate": 0.00017442560394846516,
"loss": 0.4651,
"step": 780
},
{
"epoch": 3.115079365079365,
"grad_norm": 0.71484375,
"learning_rate": 0.00017396124497991186,
"loss": 0.4732,
"step": 785
},
{
"epoch": 3.134920634920635,
"grad_norm": 0.671875,
"learning_rate": 0.00017349333823585617,
"loss": 0.4705,
"step": 790
},
{
"epoch": 3.1547619047619047,
"grad_norm": 0.46484375,
"learning_rate": 0.00017302190616086464,
"loss": 0.4634,
"step": 795
},
{
"epoch": 3.1746031746031744,
"grad_norm": 0.421875,
"learning_rate": 0.00017254697136860703,
"loss": 0.4825,
"step": 800
},
{
"epoch": 3.1944444444444446,
"grad_norm": 1.15625,
"learning_rate": 0.00017206855664077147,
"loss": 0.4796,
"step": 805
},
{
"epoch": 3.2142857142857144,
"grad_norm": 0.8828125,
"learning_rate": 0.00017158668492597186,
"loss": 0.4892,
"step": 810
},
{
"epoch": 3.234126984126984,
"grad_norm": 1.1015625,
"learning_rate": 0.0001711013793386469,
"loss": 0.4782,
"step": 815
},
{
"epoch": 3.253968253968254,
"grad_norm": 0.64453125,
"learning_rate": 0.00017061266315795146,
"loss": 0.4714,
"step": 820
},
{
"epoch": 3.2738095238095237,
"grad_norm": 0.67578125,
"learning_rate": 0.0001701205598266398,
"loss": 0.4696,
"step": 825
},
{
"epoch": 3.2936507936507935,
"grad_norm": 0.419921875,
"learning_rate": 0.0001696250929499412,
"loss": 0.4746,
"step": 830
},
{
"epoch": 3.3134920634920633,
"grad_norm": 0.5078125,
"learning_rate": 0.00016912628629442753,
"loss": 0.4703,
"step": 835
},
{
"epoch": 3.3333333333333335,
"grad_norm": 0.59375,
"learning_rate": 0.0001686241637868734,
"loss": 0.4729,
"step": 840
},
{
"epoch": 3.3531746031746033,
"grad_norm": 0.96484375,
"learning_rate": 0.00016811874951310813,
"loss": 0.4783,
"step": 845
},
{
"epoch": 3.373015873015873,
"grad_norm": 0.494140625,
"learning_rate": 0.0001676100677168608,
"loss": 0.4578,
"step": 850
},
{
"epoch": 3.392857142857143,
"grad_norm": 0.53515625,
"learning_rate": 0.00016709814279859702,
"loss": 0.4771,
"step": 855
},
{
"epoch": 3.4126984126984126,
"grad_norm": 0.50390625,
"learning_rate": 0.00016658299931434858,
"loss": 0.4889,
"step": 860
},
{
"epoch": 3.432539682539683,
"grad_norm": 0.326171875,
"learning_rate": 0.00016606466197453555,
"loss": 0.4818,
"step": 865
},
{
"epoch": 3.4523809523809526,
"grad_norm": 0.37109375,
"learning_rate": 0.000165543155642781,
"loss": 0.4736,
"step": 870
},
{
"epoch": 3.4722222222222223,
"grad_norm": 0.314453125,
"learning_rate": 0.00016501850533471836,
"loss": 0.4694,
"step": 875
},
{
"epoch": 3.492063492063492,
"grad_norm": 0.408203125,
"learning_rate": 0.00016449073621679127,
"loss": 0.4748,
"step": 880
},
{
"epoch": 3.511904761904762,
"grad_norm": 0.373046875,
"learning_rate": 0.00016395987360504668,
"loss": 0.4778,
"step": 885
},
{
"epoch": 3.5317460317460316,
"grad_norm": 0.5234375,
"learning_rate": 0.0001634259429639203,
"loss": 0.4849,
"step": 890
},
{
"epoch": 3.5515873015873014,
"grad_norm": 0.6796875,
"learning_rate": 0.00016288896990501525,
"loss": 0.4903,
"step": 895
},
{
"epoch": 3.571428571428571,
"grad_norm": 0.4296875,
"learning_rate": 0.00016234898018587337,
"loss": 0.4843,
"step": 900
},
{
"epoch": 3.5912698412698414,
"grad_norm": 0.357421875,
"learning_rate": 0.00016180599970873973,
"loss": 0.4781,
"step": 905
},
{
"epoch": 3.611111111111111,
"grad_norm": 0.546875,
"learning_rate": 0.0001612600545193203,
"loss": 0.4921,
"step": 910
},
{
"epoch": 3.630952380952381,
"grad_norm": 0.83203125,
"learning_rate": 0.00016071117080553236,
"loss": 0.4722,
"step": 915
},
{
"epoch": 3.6507936507936507,
"grad_norm": 0.8984375,
"learning_rate": 0.00016015937489624848,
"loss": 0.4836,
"step": 920
},
{
"epoch": 3.6706349206349205,
"grad_norm": 0.5625,
"learning_rate": 0.00015960469326003354,
"loss": 0.4793,
"step": 925
},
{
"epoch": 3.6904761904761907,
"grad_norm": 0.69140625,
"learning_rate": 0.00015904715250387498,
"loss": 0.484,
"step": 930
},
{
"epoch": 3.7103174603174605,
"grad_norm": 0.390625,
"learning_rate": 0.00015848677937190675,
"loss": 0.4839,
"step": 935
},
{
"epoch": 3.7301587301587302,
"grad_norm": 0.326171875,
"learning_rate": 0.00015792360074412613,
"loss": 0.4695,
"step": 940
},
{
"epoch": 3.75,
"grad_norm": 0.326171875,
"learning_rate": 0.0001573576436351046,
"loss": 0.484,
"step": 945
},
{
"epoch": 3.7698412698412698,
"grad_norm": 0.3515625,
"learning_rate": 0.00015678893519269197,
"loss": 0.4743,
"step": 950
},
{
"epoch": 3.7896825396825395,
"grad_norm": 0.31640625,
"learning_rate": 0.00015621750269671403,
"loss": 0.481,
"step": 955
},
{
"epoch": 3.8095238095238093,
"grad_norm": 0.341796875,
"learning_rate": 0.00015564337355766412,
"loss": 0.4834,
"step": 960
},
{
"epoch": 3.8293650793650795,
"grad_norm": 0.466796875,
"learning_rate": 0.00015506657531538814,
"loss": 0.4691,
"step": 965
},
{
"epoch": 3.8492063492063493,
"grad_norm": 0.361328125,
"learning_rate": 0.00015448713563776374,
"loss": 0.4764,
"step": 970
},
{
"epoch": 3.869047619047619,
"grad_norm": 0.5,
"learning_rate": 0.00015390508231937297,
"loss": 0.4725,
"step": 975
},
{
"epoch": 3.888888888888889,
"grad_norm": 0.453125,
"learning_rate": 0.00015332044328016914,
"loss": 0.4815,
"step": 980
},
{
"epoch": 3.9087301587301586,
"grad_norm": 0.427734375,
"learning_rate": 0.00015273324656413744,
"loss": 0.4893,
"step": 985
},
{
"epoch": 3.928571428571429,
"grad_norm": 0.59375,
"learning_rate": 0.0001521435203379498,
"loss": 0.4839,
"step": 990
},
{
"epoch": 3.9484126984126986,
"grad_norm": 0.54296875,
"learning_rate": 0.00015155129288961386,
"loss": 0.4741,
"step": 995
},
{
"epoch": 3.9682539682539684,
"grad_norm": 0.62890625,
"learning_rate": 0.0001509565926271159,
"loss": 0.5001,
"step": 1000
},
{
"epoch": 3.988095238095238,
"grad_norm": 0.703125,
"learning_rate": 0.0001503594480770581,
"loss": 0.4783,
"step": 1005
},
{
"epoch": 4.0,
"eval_loss": 1.1535594463348389,
"eval_runtime": 0.5468,
"eval_samples_per_second": 3.658,
"eval_steps_per_second": 1.829,
"step": 1008
},
{
"epoch": 4.007936507936508,
"grad_norm": 0.80859375,
"learning_rate": 0.00014975988788329064,
"loss": 0.4632,
"step": 1010
},
{
"epoch": 4.027777777777778,
"grad_norm": 0.7265625,
"learning_rate": 0.00014915794080553707,
"loss": 0.4185,
"step": 1015
},
{
"epoch": 4.0476190476190474,
"grad_norm": 0.474609375,
"learning_rate": 0.00014855363571801523,
"loss": 0.4187,
"step": 1020
},
{
"epoch": 4.067460317460317,
"grad_norm": 0.400390625,
"learning_rate": 0.00014794700160805196,
"loss": 0.4173,
"step": 1025
},
{
"epoch": 4.087301587301587,
"grad_norm": 0.376953125,
"learning_rate": 0.00014733806757469286,
"loss": 0.4114,
"step": 1030
},
{
"epoch": 4.107142857142857,
"grad_norm": 0.361328125,
"learning_rate": 0.0001467268628273062,
"loss": 0.4308,
"step": 1035
},
{
"epoch": 4.1269841269841265,
"grad_norm": 0.390625,
"learning_rate": 0.000146113416684182,
"loss": 0.4169,
"step": 1040
},
{
"epoch": 4.146825396825397,
"grad_norm": 0.3515625,
"learning_rate": 0.00014549775857112558,
"loss": 0.4183,
"step": 1045
},
{
"epoch": 4.166666666666667,
"grad_norm": 0.34765625,
"learning_rate": 0.00014487991802004623,
"loss": 0.4229,
"step": 1050
},
{
"epoch": 4.186507936507937,
"grad_norm": 0.68359375,
"learning_rate": 0.00014425992466754033,
"loss": 0.4206,
"step": 1055
},
{
"epoch": 4.2063492063492065,
"grad_norm": 0.41796875,
"learning_rate": 0.00014363780825347005,
"loss": 0.4142,
"step": 1060
},
{
"epoch": 4.226190476190476,
"grad_norm": 0.4453125,
"learning_rate": 0.0001430135986195365,
"loss": 0.4298,
"step": 1065
},
{
"epoch": 4.246031746031746,
"grad_norm": 0.5234375,
"learning_rate": 0.00014238732570784866,
"loss": 0.4189,
"step": 1070
},
{
"epoch": 4.265873015873016,
"grad_norm": 0.439453125,
"learning_rate": 0.0001417590195594867,
"loss": 0.4259,
"step": 1075
},
{
"epoch": 4.285714285714286,
"grad_norm": 0.369140625,
"learning_rate": 0.00014112871031306119,
"loss": 0.4286,
"step": 1080
},
{
"epoch": 4.305555555555555,
"grad_norm": 0.375,
"learning_rate": 0.00014049642820326735,
"loss": 0.4289,
"step": 1085
},
{
"epoch": 4.325396825396825,
"grad_norm": 0.42578125,
"learning_rate": 0.00013986220355943494,
"loss": 0.4187,
"step": 1090
},
{
"epoch": 4.345238095238095,
"grad_norm": 0.5390625,
"learning_rate": 0.00013922606680407307,
"loss": 0.423,
"step": 1095
},
{
"epoch": 4.365079365079365,
"grad_norm": 0.515625,
"learning_rate": 0.00013858804845141116,
"loss": 0.4297,
"step": 1100
},
{
"epoch": 4.384920634920635,
"grad_norm": 0.5625,
"learning_rate": 0.00013794817910593512,
"loss": 0.4292,
"step": 1105
},
{
"epoch": 4.404761904761905,
"grad_norm": 0.421875,
"learning_rate": 0.0001373064894609194,
"loss": 0.4332,
"step": 1110
},
{
"epoch": 4.424603174603175,
"grad_norm": 0.5234375,
"learning_rate": 0.00013666301029695462,
"loss": 0.4299,
"step": 1115
},
{
"epoch": 4.444444444444445,
"grad_norm": 0.37890625,
"learning_rate": 0.00013601777248047105,
"loss": 0.4239,
"step": 1120
},
{
"epoch": 4.464285714285714,
"grad_norm": 0.34765625,
"learning_rate": 0.00013537080696225814,
"loss": 0.4253,
"step": 1125
},
{
"epoch": 4.484126984126984,
"grad_norm": 0.423828125,
"learning_rate": 0.00013472214477597977,
"loss": 0.4296,
"step": 1130
},
{
"epoch": 4.503968253968254,
"grad_norm": 0.361328125,
"learning_rate": 0.00013407181703668566,
"loss": 0.4283,
"step": 1135
},
{
"epoch": 4.523809523809524,
"grad_norm": 0.40234375,
"learning_rate": 0.00013341985493931877,
"loss": 0.4284,
"step": 1140
},
{
"epoch": 4.5436507936507935,
"grad_norm": 0.33984375,
"learning_rate": 0.00013276628975721904,
"loss": 0.4313,
"step": 1145
},
{
"epoch": 4.563492063492063,
"grad_norm": 0.376953125,
"learning_rate": 0.00013211115284062335,
"loss": 0.4241,
"step": 1150
},
{
"epoch": 4.583333333333333,
"grad_norm": 0.51171875,
"learning_rate": 0.00013145447561516138,
"loss": 0.4367,
"step": 1155
},
{
"epoch": 4.603174603174603,
"grad_norm": 0.54296875,
"learning_rate": 0.00013079628958034855,
"loss": 0.4291,
"step": 1160
},
{
"epoch": 4.6230158730158735,
"grad_norm": 2.125,
"learning_rate": 0.00013013662630807484,
"loss": 0.435,
"step": 1165
},
{
"epoch": 4.642857142857143,
"grad_norm": 0.72265625,
"learning_rate": 0.00012947551744109043,
"loss": 0.4542,
"step": 1170
},
{
"epoch": 4.662698412698413,
"grad_norm": 0.7421875,
"learning_rate": 0.00012881299469148785,
"loss": 0.4381,
"step": 1175
},
{
"epoch": 4.682539682539683,
"grad_norm": 0.6015625,
"learning_rate": 0.00012814908983918073,
"loss": 0.4425,
"step": 1180
},
{
"epoch": 4.7023809523809526,
"grad_norm": 0.44140625,
"learning_rate": 0.00012748383473037948,
"loss": 0.4291,
"step": 1185
},
{
"epoch": 4.722222222222222,
"grad_norm": 0.76953125,
"learning_rate": 0.00012681726127606376,
"loss": 0.4327,
"step": 1190
},
{
"epoch": 4.742063492063492,
"grad_norm": 0.5625,
"learning_rate": 0.0001261494014504515,
"loss": 0.4381,
"step": 1195
},
{
"epoch": 4.761904761904762,
"grad_norm": 0.470703125,
"learning_rate": 0.0001254802872894655,
"loss": 0.437,
"step": 1200
},
{
"epoch": 4.781746031746032,
"grad_norm": 0.62890625,
"learning_rate": 0.00012480995088919635,
"loss": 0.435,
"step": 1205
},
{
"epoch": 4.801587301587301,
"grad_norm": 0.57421875,
"learning_rate": 0.00012413842440436333,
"loss": 0.4329,
"step": 1210
},
{
"epoch": 4.821428571428571,
"grad_norm": 0.41796875,
"learning_rate": 0.00012346574004677154,
"loss": 0.4308,
"step": 1215
},
{
"epoch": 4.841269841269841,
"grad_norm": 0.326171875,
"learning_rate": 0.000122791930083767,
"loss": 0.4272,
"step": 1220
},
{
"epoch": 4.861111111111111,
"grad_norm": 0.333984375,
"learning_rate": 0.00012211702683668878,
"loss": 0.4319,
"step": 1225
},
{
"epoch": 4.880952380952381,
"grad_norm": 0.345703125,
"learning_rate": 0.00012144106267931876,
"loss": 0.4402,
"step": 1230
},
{
"epoch": 4.900793650793651,
"grad_norm": 0.498046875,
"learning_rate": 0.00012076407003632846,
"loss": 0.432,
"step": 1235
},
{
"epoch": 4.920634920634921,
"grad_norm": 0.357421875,
"learning_rate": 0.00012008608138172393,
"loss": 0.4354,
"step": 1240
},
{
"epoch": 4.940476190476191,
"grad_norm": 0.419921875,
"learning_rate": 0.00011940712923728783,
"loss": 0.4272,
"step": 1245
},
{
"epoch": 4.9603174603174605,
"grad_norm": 0.32421875,
"learning_rate": 0.00011872724617101969,
"loss": 0.4286,
"step": 1250
},
{
"epoch": 4.98015873015873,
"grad_norm": 0.35546875,
"learning_rate": 0.00011804646479557342,
"loss": 0.4335,
"step": 1255
},
{
"epoch": 5.0,
"grad_norm": 0.43359375,
"learning_rate": 0.00011736481776669306,
"loss": 0.4378,
"step": 1260
},
{
"epoch": 5.0,
"eval_loss": 1.2224926948547363,
"eval_runtime": 0.5464,
"eval_samples_per_second": 3.661,
"eval_steps_per_second": 1.83,
"step": 1260
},
{
"epoch": 5.01984126984127,
"grad_norm": 0.64453125,
"learning_rate": 0.00011668233778164633,
"loss": 0.3758,
"step": 1265
},
{
"epoch": 5.0396825396825395,
"grad_norm": 0.75390625,
"learning_rate": 0.0001159990575776563,
"loss": 0.3679,
"step": 1270
},
{
"epoch": 5.059523809523809,
"grad_norm": 0.44140625,
"learning_rate": 0.00011531500993033093,
"loss": 0.3671,
"step": 1275
},
{
"epoch": 5.079365079365079,
"grad_norm": 0.5390625,
"learning_rate": 0.00011463022765209088,
"loss": 0.3633,
"step": 1280
},
{
"epoch": 5.099206349206349,
"grad_norm": 0.421875,
"learning_rate": 0.00011394474359059567,
"loss": 0.3777,
"step": 1285
},
{
"epoch": 5.119047619047619,
"grad_norm": 0.53125,
"learning_rate": 0.00011325859062716795,
"loss": 0.3714,
"step": 1290
},
{
"epoch": 5.138888888888889,
"grad_norm": 0.431640625,
"learning_rate": 0.00011257180167521629,
"loss": 0.3801,
"step": 1295
},
{
"epoch": 5.158730158730159,
"grad_norm": 0.400390625,
"learning_rate": 0.00011188440967865641,
"loss": 0.3817,
"step": 1300
},
{
"epoch": 5.178571428571429,
"grad_norm": 0.380859375,
"learning_rate": 0.00011119644761033078,
"loss": 0.3724,
"step": 1305
},
{
"epoch": 5.198412698412699,
"grad_norm": 0.48046875,
"learning_rate": 0.00011050794847042731,
"loss": 0.3799,
"step": 1310
},
{
"epoch": 5.218253968253968,
"grad_norm": 0.5078125,
"learning_rate": 0.00010981894528489596,
"loss": 0.3735,
"step": 1315
},
{
"epoch": 5.238095238095238,
"grad_norm": 0.462890625,
"learning_rate": 0.00010912947110386484,
"loss": 0.3828,
"step": 1320
},
{
"epoch": 5.257936507936508,
"grad_norm": 0.5,
"learning_rate": 0.00010843955900005485,
"loss": 0.3704,
"step": 1325
},
{
"epoch": 5.277777777777778,
"grad_norm": 0.44140625,
"learning_rate": 0.0001077492420671931,
"loss": 0.386,
"step": 1330
},
{
"epoch": 5.2976190476190474,
"grad_norm": 0.44140625,
"learning_rate": 0.00010705855341842563,
"loss": 0.3726,
"step": 1335
},
{
"epoch": 5.317460317460317,
"grad_norm": 0.388671875,
"learning_rate": 0.00010636752618472887,
"loss": 0.381,
"step": 1340
},
{
"epoch": 5.337301587301587,
"grad_norm": 0.4609375,
"learning_rate": 0.00010567619351332057,
"loss": 0.3725,
"step": 1345
},
{
"epoch": 5.357142857142857,
"grad_norm": 0.423828125,
"learning_rate": 0.00010498458856606972,
"loss": 0.377,
"step": 1350
},
{
"epoch": 5.3769841269841265,
"grad_norm": 0.37109375,
"learning_rate": 0.00010429274451790585,
"loss": 0.376,
"step": 1355
},
{
"epoch": 5.396825396825397,
"grad_norm": 0.443359375,
"learning_rate": 0.00010360069455522765,
"loss": 0.3845,
"step": 1360
},
{
"epoch": 5.416666666666667,
"grad_norm": 0.4609375,
"learning_rate": 0.00010290847187431113,
"loss": 0.3818,
"step": 1365
},
{
"epoch": 5.436507936507937,
"grad_norm": 0.3984375,
"learning_rate": 0.00010221610967971735,
"loss": 0.3774,
"step": 1370
},
{
"epoch": 5.4563492063492065,
"grad_norm": 0.58203125,
"learning_rate": 0.00010152364118269943,
"loss": 0.3757,
"step": 1375
},
{
"epoch": 5.476190476190476,
"grad_norm": 0.58203125,
"learning_rate": 0.00010083109959960973,
"loss": 0.383,
"step": 1380
},
{
"epoch": 5.496031746031746,
"grad_norm": 0.392578125,
"learning_rate": 0.00010013851815030624,
"loss": 0.3814,
"step": 1385
},
{
"epoch": 5.515873015873016,
"grad_norm": 0.421875,
"learning_rate": 9.944593005655947e-05,
"loss": 0.3829,
"step": 1390
},
{
"epoch": 5.535714285714286,
"grad_norm": 0.48828125,
"learning_rate": 9.875336854045851e-05,
"loss": 0.3951,
"step": 1395
},
{
"epoch": 5.555555555555555,
"grad_norm": 0.443359375,
"learning_rate": 9.806086682281758e-05,
"loss": 0.3923,
"step": 1400
},
{
"epoch": 5.575396825396825,
"grad_norm": 0.44921875,
"learning_rate": 9.736845812158252e-05,
"loss": 0.3828,
"step": 1405
},
{
"epoch": 5.595238095238095,
"grad_norm": 0.6875,
"learning_rate": 9.667617565023735e-05,
"loss": 0.3908,
"step": 1410
},
{
"epoch": 5.615079365079366,
"grad_norm": 0.5234375,
"learning_rate": 9.598405261621105e-05,
"loss": 0.3918,
"step": 1415
},
{
"epoch": 5.634920634920634,
"grad_norm": 0.484375,
"learning_rate": 9.529212221928483e-05,
"loss": 0.3781,
"step": 1420
},
{
"epoch": 5.654761904761905,
"grad_norm": 0.4609375,
"learning_rate": 9.460041764999928e-05,
"loss": 0.3803,
"step": 1425
},
{
"epoch": 5.674603174603175,
"grad_norm": 0.515625,
"learning_rate": 9.390897208806266e-05,
"loss": 0.3745,
"step": 1430
},
{
"epoch": 5.694444444444445,
"grad_norm": 0.6640625,
"learning_rate": 9.321781870075908e-05,
"loss": 0.3807,
"step": 1435
},
{
"epoch": 5.714285714285714,
"grad_norm": 0.484375,
"learning_rate": 9.252699064135758e-05,
"loss": 0.3797,
"step": 1440
},
{
"epoch": 5.734126984126984,
"grad_norm": 0.546875,
"learning_rate": 9.183652104752195e-05,
"loss": 0.3848,
"step": 1445
},
{
"epoch": 5.753968253968254,
"grad_norm": 0.37890625,
"learning_rate": 9.114644303972096e-05,
"loss": 0.3788,
"step": 1450
},
{
"epoch": 5.773809523809524,
"grad_norm": 0.40625,
"learning_rate": 9.045678971963988e-05,
"loss": 0.3824,
"step": 1455
},
{
"epoch": 5.7936507936507935,
"grad_norm": 0.353515625,
"learning_rate": 8.976759416859256e-05,
"loss": 0.3895,
"step": 1460
},
{
"epoch": 5.813492063492063,
"grad_norm": 0.41015625,
"learning_rate": 8.907888944593442e-05,
"loss": 0.3773,
"step": 1465
},
{
"epoch": 5.833333333333333,
"grad_norm": 0.44921875,
"learning_rate": 8.839070858747697e-05,
"loss": 0.3789,
"step": 1470
},
{
"epoch": 5.853174603174603,
"grad_norm": 0.388671875,
"learning_rate": 8.770308460390293e-05,
"loss": 0.3892,
"step": 1475
},
{
"epoch": 5.8730158730158735,
"grad_norm": 0.419921875,
"learning_rate": 8.701605047918276e-05,
"loss": 0.3795,
"step": 1480
},
{
"epoch": 5.892857142857143,
"grad_norm": 0.57421875,
"learning_rate": 8.632963916899268e-05,
"loss": 0.383,
"step": 1485
},
{
"epoch": 5.912698412698413,
"grad_norm": 0.5546875,
"learning_rate": 8.564388359913356e-05,
"loss": 0.3871,
"step": 1490
},
{
"epoch": 5.932539682539683,
"grad_norm": 0.43359375,
"learning_rate": 8.495881666395185e-05,
"loss": 0.3799,
"step": 1495
},
{
"epoch": 5.9523809523809526,
"grad_norm": 0.7265625,
"learning_rate": 8.427447122476148e-05,
"loss": 0.3831,
"step": 1500
},
{
"epoch": 5.972222222222222,
"grad_norm": 0.609375,
"learning_rate": 8.35908801082676e-05,
"loss": 0.3798,
"step": 1505
},
{
"epoch": 5.992063492063492,
"grad_norm": 0.58203125,
"learning_rate": 8.290807610499206e-05,
"loss": 0.3836,
"step": 1510
},
{
"epoch": 6.0,
"eval_loss": 1.2893120050430298,
"eval_runtime": 0.547,
"eval_samples_per_second": 3.656,
"eval_steps_per_second": 1.828,
"step": 1512
},
{
"epoch": 6.011904761904762,
"grad_norm": 0.54296875,
"learning_rate": 8.222609196770036e-05,
"loss": 0.3411,
"step": 1515
},
{
"epoch": 6.031746031746032,
"grad_norm": 0.53125,
"learning_rate": 8.154496040983073e-05,
"loss": 0.3316,
"step": 1520
},
{
"epoch": 6.051587301587301,
"grad_norm": 0.59375,
"learning_rate": 8.086471410392476e-05,
"loss": 0.3286,
"step": 1525
},
{
"epoch": 6.071428571428571,
"grad_norm": 0.46875,
"learning_rate": 8.018538568006027e-05,
"loss": 0.3326,
"step": 1530
},
{
"epoch": 6.091269841269841,
"grad_norm": 0.41796875,
"learning_rate": 7.950700772428605e-05,
"loss": 0.3298,
"step": 1535
},
{
"epoch": 6.111111111111111,
"grad_norm": 0.458984375,
"learning_rate": 7.882961277705895e-05,
"loss": 0.3318,
"step": 1540
},
{
"epoch": 6.130952380952381,
"grad_norm": 0.4453125,
"learning_rate": 7.815323333168262e-05,
"loss": 0.3276,
"step": 1545
},
{
"epoch": 6.150793650793651,
"grad_norm": 0.42578125,
"learning_rate": 7.747790183274922e-05,
"loss": 0.3387,
"step": 1550
},
{
"epoch": 6.170634920634921,
"grad_norm": 0.41796875,
"learning_rate": 7.6803650674583e-05,
"loss": 0.332,
"step": 1555
},
{
"epoch": 6.190476190476191,
"grad_norm": 0.4140625,
"learning_rate": 7.613051219968623e-05,
"loss": 0.3401,
"step": 1560
},
{
"epoch": 6.2103174603174605,
"grad_norm": 0.392578125,
"learning_rate": 7.545851869718817e-05,
"loss": 0.3277,
"step": 1565
},
{
"epoch": 6.23015873015873,
"grad_norm": 0.392578125,
"learning_rate": 7.478770240129579e-05,
"loss": 0.3328,
"step": 1570
},
{
"epoch": 6.25,
"grad_norm": 0.39453125,
"learning_rate": 7.411809548974792e-05,
"loss": 0.3312,
"step": 1575
},
{
"epoch": 6.26984126984127,
"grad_norm": 0.408203125,
"learning_rate": 7.344973008227161e-05,
"loss": 0.3317,
"step": 1580
},
{
"epoch": 6.2896825396825395,
"grad_norm": 0.490234375,
"learning_rate": 7.278263823904123e-05,
"loss": 0.3322,
"step": 1585
},
{
"epoch": 6.309523809523809,
"grad_norm": 0.3984375,
"learning_rate": 7.211685195914097e-05,
"loss": 0.333,
"step": 1590
},
{
"epoch": 6.329365079365079,
"grad_norm": 0.52734375,
"learning_rate": 7.145240317902965e-05,
"loss": 0.3403,
"step": 1595
},
{
"epoch": 6.349206349206349,
"grad_norm": 0.41015625,
"learning_rate": 7.078932377100877e-05,
"loss": 0.3369,
"step": 1600
},
{
"epoch": 6.369047619047619,
"grad_norm": 0.4921875,
"learning_rate": 7.012764554169393e-05,
"loss": 0.3317,
"step": 1605
},
{
"epoch": 6.388888888888889,
"grad_norm": 0.515625,
"learning_rate": 6.94674002304887e-05,
"loss": 0.3318,
"step": 1610
},
{
"epoch": 6.408730158730159,
"grad_norm": 0.494140625,
"learning_rate": 6.880861950806256e-05,
"loss": 0.3456,
"step": 1615
},
{
"epoch": 6.428571428571429,
"grad_norm": 0.427734375,
"learning_rate": 6.815133497483157e-05,
"loss": 0.3345,
"step": 1620
},
{
"epoch": 6.448412698412699,
"grad_norm": 0.388671875,
"learning_rate": 6.749557815944236e-05,
"loss": 0.3332,
"step": 1625
},
{
"epoch": 6.468253968253968,
"grad_norm": 0.55859375,
"learning_rate": 6.684138051726012e-05,
"loss": 0.3331,
"step": 1630
},
{
"epoch": 6.488095238095238,
"grad_norm": 0.5390625,
"learning_rate": 6.618877342885945e-05,
"loss": 0.3321,
"step": 1635
},
{
"epoch": 6.507936507936508,
"grad_norm": 0.447265625,
"learning_rate": 6.553778819851926e-05,
"loss": 0.3317,
"step": 1640
},
{
"epoch": 6.527777777777778,
"grad_norm": 0.453125,
"learning_rate": 6.488845605272113e-05,
"loss": 0.3379,
"step": 1645
},
{
"epoch": 6.5476190476190474,
"grad_norm": 0.423828125,
"learning_rate": 6.424080813865138e-05,
"loss": 0.3457,
"step": 1650
},
{
"epoch": 6.567460317460317,
"grad_norm": 0.408203125,
"learning_rate": 6.359487552270708e-05,
"loss": 0.342,
"step": 1655
},
{
"epoch": 6.587301587301587,
"grad_norm": 0.46875,
"learning_rate": 6.295068918900586e-05,
"loss": 0.3371,
"step": 1660
},
{
"epoch": 6.607142857142857,
"grad_norm": 0.703125,
"learning_rate": 6.230828003789949e-05,
"loss": 0.3347,
"step": 1665
},
{
"epoch": 6.6269841269841265,
"grad_norm": 0.5234375,
"learning_rate": 6.16676788844919e-05,
"loss": 0.3391,
"step": 1670
},
{
"epoch": 6.646825396825397,
"grad_norm": 0.5234375,
"learning_rate": 6.102891645716096e-05,
"loss": 0.3319,
"step": 1675
},
{
"epoch": 6.666666666666667,
"grad_norm": 0.5078125,
"learning_rate": 6.039202339608432e-05,
"loss": 0.3471,
"step": 1680
},
{
"epoch": 6.686507936507937,
"grad_norm": 0.4140625,
"learning_rate": 5.975703025176997e-05,
"loss": 0.3311,
"step": 1685
},
{
"epoch": 6.7063492063492065,
"grad_norm": 0.45703125,
"learning_rate": 5.912396748359046e-05,
"loss": 0.339,
"step": 1690
},
{
"epoch": 6.726190476190476,
"grad_norm": 0.376953125,
"learning_rate": 5.849286545832211e-05,
"loss": 0.3409,
"step": 1695
},
{
"epoch": 6.746031746031746,
"grad_norm": 0.44921875,
"learning_rate": 5.786375444868828e-05,
"loss": 0.3389,
"step": 1700
},
{
"epoch": 6.765873015873016,
"grad_norm": 0.47265625,
"learning_rate": 5.723666463190713e-05,
"loss": 0.3379,
"step": 1705
},
{
"epoch": 6.785714285714286,
"grad_norm": 0.4296875,
"learning_rate": 5.6611626088244194e-05,
"loss": 0.3418,
"step": 1710
},
{
"epoch": 6.805555555555555,
"grad_norm": 0.46484375,
"learning_rate": 5.5988668799569545e-05,
"loss": 0.3379,
"step": 1715
},
{
"epoch": 6.825396825396825,
"grad_norm": 0.4140625,
"learning_rate": 5.5367822647919424e-05,
"loss": 0.3369,
"step": 1720
},
{
"epoch": 6.845238095238095,
"grad_norm": 0.40625,
"learning_rate": 5.47491174140631e-05,
"loss": 0.3391,
"step": 1725
},
{
"epoch": 6.865079365079366,
"grad_norm": 0.431640625,
"learning_rate": 5.4132582776074126e-05,
"loss": 0.3449,
"step": 1730
},
{
"epoch": 6.884920634920634,
"grad_norm": 0.390625,
"learning_rate": 5.35182483079069e-05,
"loss": 0.3303,
"step": 1735
},
{
"epoch": 6.904761904761905,
"grad_norm": 0.400390625,
"learning_rate": 5.290614347797802e-05,
"loss": 0.3348,
"step": 1740
},
{
"epoch": 6.924603174603175,
"grad_norm": 0.4375,
"learning_rate": 5.229629764775261e-05,
"loss": 0.331,
"step": 1745
},
{
"epoch": 6.944444444444445,
"grad_norm": 0.4140625,
"learning_rate": 5.168874007033615e-05,
"loss": 0.3425,
"step": 1750
},
{
"epoch": 6.964285714285714,
"grad_norm": 0.431640625,
"learning_rate": 5.108349988907111e-05,
"loss": 0.3392,
"step": 1755
},
{
"epoch": 6.984126984126984,
"grad_norm": 0.44140625,
"learning_rate": 5.048060613613888e-05,
"loss": 0.3381,
"step": 1760
},
{
"epoch": 7.0,
"eval_loss": 1.4050453901290894,
"eval_runtime": 0.5475,
"eval_samples_per_second": 3.653,
"eval_steps_per_second": 1.826,
"step": 1764
},
{
"epoch": 7.003968253968254,
"grad_norm": 0.369140625,
"learning_rate": 4.9880087731167515e-05,
"loss": 0.3209,
"step": 1765
},
{
"epoch": 7.023809523809524,
"grad_norm": 0.44140625,
"learning_rate": 4.92819734798441e-05,
"loss": 0.3078,
"step": 1770
},
{
"epoch": 7.0436507936507935,
"grad_norm": 0.392578125,
"learning_rate": 4.868629207253332e-05,
"loss": 0.2901,
"step": 1775
},
{
"epoch": 7.063492063492063,
"grad_norm": 0.404296875,
"learning_rate": 4.809307208290114e-05,
"loss": 0.296,
"step": 1780
},
{
"epoch": 7.083333333333333,
"grad_norm": 0.416015625,
"learning_rate": 4.7502341966544e-05,
"loss": 0.2953,
"step": 1785
},
{
"epoch": 7.103174603174603,
"grad_norm": 0.447265625,
"learning_rate": 4.691413005962415e-05,
"loss": 0.2979,
"step": 1790
},
{
"epoch": 7.123015873015873,
"grad_norm": 0.40625,
"learning_rate": 4.632846457751031e-05,
"loss": 0.3008,
"step": 1795
},
{
"epoch": 7.142857142857143,
"grad_norm": 0.451171875,
"learning_rate": 4.574537361342407e-05,
"loss": 0.2998,
"step": 1800
},
{
"epoch": 7.162698412698413,
"grad_norm": 0.4375,
"learning_rate": 4.5164885137092585e-05,
"loss": 0.3035,
"step": 1805
},
{
"epoch": 7.182539682539683,
"grad_norm": 0.40234375,
"learning_rate": 4.458702699340667e-05,
"loss": 0.3013,
"step": 1810
},
{
"epoch": 7.2023809523809526,
"grad_norm": 0.41796875,
"learning_rate": 4.401182690108534e-05,
"loss": 0.2971,
"step": 1815
},
{
"epoch": 7.222222222222222,
"grad_norm": 0.40625,
"learning_rate": 4.343931245134616e-05,
"loss": 0.3012,
"step": 1820
},
{
"epoch": 7.242063492063492,
"grad_norm": 0.46484375,
"learning_rate": 4.2869511106581536e-05,
"loss": 0.2973,
"step": 1825
},
{
"epoch": 7.261904761904762,
"grad_norm": 0.4375,
"learning_rate": 4.23024501990417e-05,
"loss": 0.3019,
"step": 1830
},
{
"epoch": 7.281746031746032,
"grad_norm": 0.41796875,
"learning_rate": 4.1738156929523486e-05,
"loss": 0.2937,
"step": 1835
},
{
"epoch": 7.301587301587301,
"grad_norm": 0.390625,
"learning_rate": 4.117665836606549e-05,
"loss": 0.3012,
"step": 1840
},
{
"epoch": 7.321428571428571,
"grad_norm": 0.408203125,
"learning_rate": 4.0617981442649855e-05,
"loss": 0.3027,
"step": 1845
},
{
"epoch": 7.341269841269841,
"grad_norm": 0.421875,
"learning_rate": 4.00621529579101e-05,
"loss": 0.2984,
"step": 1850
},
{
"epoch": 7.361111111111111,
"grad_norm": 0.39453125,
"learning_rate": 3.950919957384582e-05,
"loss": 0.3087,
"step": 1855
},
{
"epoch": 7.380952380952381,
"grad_norm": 0.4375,
"learning_rate": 3.89591478145437e-05,
"loss": 0.3008,
"step": 1860
},
{
"epoch": 7.400793650793651,
"grad_norm": 0.427734375,
"learning_rate": 3.841202406490506e-05,
"loss": 0.3046,
"step": 1865
},
{
"epoch": 7.420634920634921,
"grad_norm": 0.419921875,
"learning_rate": 3.786785456938049e-05,
"loss": 0.3092,
"step": 1870
},
{
"epoch": 7.440476190476191,
"grad_norm": 0.466796875,
"learning_rate": 3.732666543071079e-05,
"loss": 0.2992,
"step": 1875
},
{
"epoch": 7.4603174603174605,
"grad_norm": 0.57421875,
"learning_rate": 3.6788482608674826e-05,
"loss": 0.2964,
"step": 1880
},
{
"epoch": 7.48015873015873,
"grad_norm": 0.423828125,
"learning_rate": 3.625333191884447e-05,
"loss": 0.2964,
"step": 1885
},
{
"epoch": 7.5,
"grad_norm": 0.40625,
"learning_rate": 3.5721239031346066e-05,
"loss": 0.2938,
"step": 1890
},
{
"epoch": 7.51984126984127,
"grad_norm": 0.40625,
"learning_rate": 3.519222946962929e-05,
"loss": 0.2882,
"step": 1895
},
{
"epoch": 7.5396825396825395,
"grad_norm": 0.453125,
"learning_rate": 3.4666328609242725e-05,
"loss": 0.3071,
"step": 1900
},
{
"epoch": 7.559523809523809,
"grad_norm": 0.4140625,
"learning_rate": 3.4143561676616575e-05,
"loss": 0.3,
"step": 1905
},
{
"epoch": 7.579365079365079,
"grad_norm": 0.376953125,
"learning_rate": 3.362395374785283e-05,
"loss": 0.303,
"step": 1910
},
{
"epoch": 7.599206349206349,
"grad_norm": 0.435546875,
"learning_rate": 3.3107529747522135e-05,
"loss": 0.2965,
"step": 1915
},
{
"epoch": 7.619047619047619,
"grad_norm": 0.5078125,
"learning_rate": 3.259431444746846e-05,
"loss": 0.3073,
"step": 1920
},
{
"epoch": 7.638888888888889,
"grad_norm": 0.37890625,
"learning_rate": 3.2084332465620694e-05,
"loss": 0.3038,
"step": 1925
},
{
"epoch": 7.658730158730159,
"grad_norm": 0.431640625,
"learning_rate": 3.157760826481174e-05,
"loss": 0.2962,
"step": 1930
},
{
"epoch": 7.678571428571429,
"grad_norm": 0.4140625,
"learning_rate": 3.10741661516053e-05,
"loss": 0.299,
"step": 1935
},
{
"epoch": 7.698412698412699,
"grad_norm": 0.41015625,
"learning_rate": 3.057403027512963e-05,
"loss": 0.2947,
"step": 1940
},
{
"epoch": 7.718253968253968,
"grad_norm": 0.4140625,
"learning_rate": 3.007722462591942e-05,
"loss": 0.2982,
"step": 1945
},
{
"epoch": 7.738095238095238,
"grad_norm": 0.388671875,
"learning_rate": 2.9583773034764826e-05,
"loss": 0.2921,
"step": 1950
},
{
"epoch": 7.757936507936508,
"grad_norm": 0.43359375,
"learning_rate": 2.9093699171568522e-05,
"loss": 0.2991,
"step": 1955
},
{
"epoch": 7.777777777777778,
"grad_norm": 0.4375,
"learning_rate": 2.8607026544210114e-05,
"loss": 0.3019,
"step": 1960
},
{
"epoch": 7.7976190476190474,
"grad_norm": 0.4140625,
"learning_rate": 2.8123778497418685e-05,
"loss": 0.3034,
"step": 1965
},
{
"epoch": 7.817460317460317,
"grad_norm": 0.41796875,
"learning_rate": 2.764397821165292e-05,
"loss": 0.2991,
"step": 1970
},
{
"epoch": 7.837301587301587,
"grad_norm": 0.421875,
"learning_rate": 2.7167648701989135e-05,
"loss": 0.3041,
"step": 1975
},
{
"epoch": 7.857142857142857,
"grad_norm": 0.447265625,
"learning_rate": 2.669481281701739e-05,
"loss": 0.3065,
"step": 1980
},
{
"epoch": 7.8769841269841265,
"grad_norm": 0.408203125,
"learning_rate": 2.622549323774539e-05,
"loss": 0.3037,
"step": 1985
},
{
"epoch": 7.896825396825397,
"grad_norm": 0.404296875,
"learning_rate": 2.5759712476510622e-05,
"loss": 0.3086,
"step": 1990
},
{
"epoch": 7.916666666666667,
"grad_norm": 0.41015625,
"learning_rate": 2.529749287590042e-05,
"loss": 0.2866,
"step": 1995
},
{
"epoch": 7.936507936507937,
"grad_norm": 0.390625,
"learning_rate": 2.4838856607680183e-05,
"loss": 0.3094,
"step": 2000
},
{
"epoch": 7.9563492063492065,
"grad_norm": 0.423828125,
"learning_rate": 2.4383825671729975e-05,
"loss": 0.2962,
"step": 2005
},
{
"epoch": 7.976190476190476,
"grad_norm": 0.404296875,
"learning_rate": 2.3932421894989167e-05,
"loss": 0.3078,
"step": 2010
},
{
"epoch": 7.996031746031746,
"grad_norm": 0.419921875,
"learning_rate": 2.3484666930409327e-05,
"loss": 0.3043,
"step": 2015
},
{
"epoch": 8.0,
"eval_loss": 1.5184850692749023,
"eval_runtime": 0.5467,
"eval_samples_per_second": 3.658,
"eval_steps_per_second": 1.829,
"step": 2016
},
{
"epoch": 8.015873015873016,
"grad_norm": 0.37890625,
"learning_rate": 2.304058225591581e-05,
"loss": 0.2823,
"step": 2020
},
{
"epoch": 8.035714285714286,
"grad_norm": 0.4140625,
"learning_rate": 2.260018917337726e-05,
"loss": 0.2887,
"step": 2025
},
{
"epoch": 8.055555555555555,
"grad_norm": 0.44140625,
"learning_rate": 2.2163508807583998e-05,
"loss": 0.2806,
"step": 2030
},
{
"epoch": 8.075396825396826,
"grad_norm": 0.42578125,
"learning_rate": 2.173056210523461e-05,
"loss": 0.276,
"step": 2035
},
{
"epoch": 8.095238095238095,
"grad_norm": 0.384765625,
"learning_rate": 2.1301369833931117e-05,
"loss": 0.288,
"step": 2040
},
{
"epoch": 8.115079365079366,
"grad_norm": 0.3828125,
"learning_rate": 2.0875952581182932e-05,
"loss": 0.2754,
"step": 2045
},
{
"epoch": 8.134920634920634,
"grad_norm": 0.37109375,
"learning_rate": 2.045433075341927e-05,
"loss": 0.2838,
"step": 2050
},
{
"epoch": 8.154761904761905,
"grad_norm": 0.4296875,
"learning_rate": 2.0036524575010172e-05,
"loss": 0.2782,
"step": 2055
},
{
"epoch": 8.174603174603174,
"grad_norm": 0.41015625,
"learning_rate": 1.962255408729662e-05,
"loss": 0.2802,
"step": 2060
},
{
"epoch": 8.194444444444445,
"grad_norm": 0.392578125,
"learning_rate": 1.921243914762889e-05,
"loss": 0.2802,
"step": 2065
},
{
"epoch": 8.214285714285714,
"grad_norm": 0.373046875,
"learning_rate": 1.880619942841435e-05,
"loss": 0.2819,
"step": 2070
},
{
"epoch": 8.234126984126984,
"grad_norm": 0.396484375,
"learning_rate": 1.8403854416173595e-05,
"loss": 0.2791,
"step": 2075
},
{
"epoch": 8.253968253968253,
"grad_norm": 0.412109375,
"learning_rate": 1.8005423410605772e-05,
"loss": 0.2781,
"step": 2080
},
{
"epoch": 8.273809523809524,
"grad_norm": 0.388671875,
"learning_rate": 1.7610925523662835e-05,
"loss": 0.2863,
"step": 2085
},
{
"epoch": 8.293650793650794,
"grad_norm": 0.42578125,
"learning_rate": 1.7220379678632814e-05,
"loss": 0.2842,
"step": 2090
},
{
"epoch": 8.313492063492063,
"grad_norm": 0.404296875,
"learning_rate": 1.6833804609231973e-05,
"loss": 0.2819,
"step": 2095
},
{
"epoch": 8.333333333333334,
"grad_norm": 0.3984375,
"learning_rate": 1.6451218858706374e-05,
"loss": 0.2937,
"step": 2100
},
{
"epoch": 8.353174603174603,
"grad_norm": 0.408203125,
"learning_rate": 1.6072640778942183e-05,
"loss": 0.2846,
"step": 2105
},
{
"epoch": 8.373015873015873,
"grad_norm": 0.3984375,
"learning_rate": 1.5698088529585597e-05,
"loss": 0.2834,
"step": 2110
},
{
"epoch": 8.392857142857142,
"grad_norm": 0.4375,
"learning_rate": 1.5327580077171587e-05,
"loss": 0.2801,
"step": 2115
},
{
"epoch": 8.412698412698413,
"grad_norm": 0.40234375,
"learning_rate": 1.49611331942621e-05,
"loss": 0.278,
"step": 2120
},
{
"epoch": 8.432539682539682,
"grad_norm": 0.404296875,
"learning_rate": 1.4598765458593655e-05,
"loss": 0.2777,
"step": 2125
},
{
"epoch": 8.452380952380953,
"grad_norm": 0.376953125,
"learning_rate": 1.4240494252234049e-05,
"loss": 0.2713,
"step": 2130
},
{
"epoch": 8.472222222222221,
"grad_norm": 0.4140625,
"learning_rate": 1.388633676074862e-05,
"loss": 0.2839,
"step": 2135
},
{
"epoch": 8.492063492063492,
"grad_norm": 0.392578125,
"learning_rate": 1.3536309972375948e-05,
"loss": 0.2835,
"step": 2140
},
{
"epoch": 8.511904761904763,
"grad_norm": 0.408203125,
"learning_rate": 1.3190430677212794e-05,
"loss": 0.2783,
"step": 2145
},
{
"epoch": 8.531746031746032,
"grad_norm": 0.412109375,
"learning_rate": 1.2848715466408967e-05,
"loss": 0.2837,
"step": 2150
},
{
"epoch": 8.551587301587302,
"grad_norm": 0.390625,
"learning_rate": 1.2511180731371208e-05,
"loss": 0.2779,
"step": 2155
},
{
"epoch": 8.571428571428571,
"grad_norm": 0.421875,
"learning_rate": 1.2177842662977135e-05,
"loss": 0.277,
"step": 2160
},
{
"epoch": 8.591269841269842,
"grad_norm": 0.408203125,
"learning_rate": 1.1848717250798524e-05,
"loss": 0.285,
"step": 2165
},
{
"epoch": 8.61111111111111,
"grad_norm": 0.384765625,
"learning_rate": 1.1523820282334219e-05,
"loss": 0.2848,
"step": 2170
},
{
"epoch": 8.630952380952381,
"grad_norm": 0.427734375,
"learning_rate": 1.1203167342253062e-05,
"loss": 0.2809,
"step": 2175
},
{
"epoch": 8.65079365079365,
"grad_norm": 0.384765625,
"learning_rate": 1.088677381164609e-05,
"loss": 0.278,
"step": 2180
},
{
"epoch": 8.670634920634921,
"grad_norm": 0.412109375,
"learning_rate": 1.0574654867288925e-05,
"loss": 0.2786,
"step": 2185
},
{
"epoch": 8.69047619047619,
"grad_norm": 0.39453125,
"learning_rate": 1.0266825480913611e-05,
"loss": 0.2826,
"step": 2190
},
{
"epoch": 8.71031746031746,
"grad_norm": 0.44140625,
"learning_rate": 9.963300418490617e-06,
"loss": 0.2849,
"step": 2195
},
{
"epoch": 8.73015873015873,
"grad_norm": 0.44140625,
"learning_rate": 9.664094239520372e-06,
"loss": 0.2899,
"step": 2200
},
{
"epoch": 8.75,
"grad_norm": 0.3828125,
"learning_rate": 9.369221296335006e-06,
"loss": 0.2776,
"step": 2205
},
{
"epoch": 8.76984126984127,
"grad_norm": 0.43359375,
"learning_rate": 9.07869573340987e-06,
"loss": 0.2866,
"step": 2210
},
{
"epoch": 8.78968253968254,
"grad_norm": 0.408203125,
"learning_rate": 8.79253148668494e-06,
"loss": 0.2815,
"step": 2215
},
{
"epoch": 8.80952380952381,
"grad_norm": 0.380859375,
"learning_rate": 8.510742282896544e-06,
"loss": 0.2828,
"step": 2220
},
{
"epoch": 8.829365079365079,
"grad_norm": 0.419921875,
"learning_rate": 8.233341638918712e-06,
"loss": 0.2809,
"step": 2225
},
{
"epoch": 8.84920634920635,
"grad_norm": 0.416015625,
"learning_rate": 7.960342861114921e-06,
"loss": 0.2798,
"step": 2230
},
{
"epoch": 8.869047619047619,
"grad_norm": 0.400390625,
"learning_rate": 7.69175904469982e-06,
"loss": 0.2828,
"step": 2235
},
{
"epoch": 8.88888888888889,
"grad_norm": 0.39453125,
"learning_rate": 7.427603073110967e-06,
"loss": 0.2843,
"step": 2240
},
{
"epoch": 8.908730158730158,
"grad_norm": 0.3828125,
"learning_rate": 7.1678876173909556e-06,
"loss": 0.2779,
"step": 2245
},
{
"epoch": 8.928571428571429,
"grad_norm": 0.3984375,
"learning_rate": 6.9126251355795864e-06,
"loss": 0.2816,
"step": 2250
},
{
"epoch": 8.948412698412698,
"grad_norm": 0.404296875,
"learning_rate": 6.661827872116177e-06,
"loss": 0.2846,
"step": 2255
},
{
"epoch": 8.968253968253968,
"grad_norm": 0.431640625,
"learning_rate": 6.415507857252389e-06,
"loss": 0.2883,
"step": 2260
},
{
"epoch": 8.988095238095237,
"grad_norm": 0.419921875,
"learning_rate": 6.173676906475012e-06,
"loss": 0.2778,
"step": 2265
},
{
"epoch": 9.0,
"eval_loss": 1.6142504215240479,
"eval_runtime": 0.5468,
"eval_samples_per_second": 3.657,
"eval_steps_per_second": 1.829,
"step": 2268
},
{
"epoch": 9.007936507936508,
"grad_norm": 0.376953125,
"learning_rate": 5.936346619939271e-06,
"loss": 0.2831,
"step": 2270
},
{
"epoch": 9.027777777777779,
"grad_norm": 0.392578125,
"learning_rate": 5.7035283819124155e-06,
"loss": 0.2754,
"step": 2275
},
{
"epoch": 9.047619047619047,
"grad_norm": 0.408203125,
"learning_rate": 5.475233360227516e-06,
"loss": 0.2794,
"step": 2280
},
{
"epoch": 9.067460317460318,
"grad_norm": 0.375,
"learning_rate": 5.2514725057479455e-06,
"loss": 0.2828,
"step": 2285
},
{
"epoch": 9.087301587301587,
"grad_norm": 0.384765625,
"learning_rate": 5.03225655184194e-06,
"loss": 0.269,
"step": 2290
},
{
"epoch": 9.107142857142858,
"grad_norm": 0.3984375,
"learning_rate": 4.817596013867764e-06,
"loss": 0.2773,
"step": 2295
},
{
"epoch": 9.126984126984127,
"grad_norm": 0.388671875,
"learning_rate": 4.607501188669394e-06,
"loss": 0.277,
"step": 2300
},
{
"epoch": 9.146825396825397,
"grad_norm": 0.392578125,
"learning_rate": 4.401982154082507e-06,
"loss": 0.2775,
"step": 2305
},
{
"epoch": 9.166666666666666,
"grad_norm": 0.37890625,
"learning_rate": 4.20104876845111e-06,
"loss": 0.272,
"step": 2310
},
{
"epoch": 9.186507936507937,
"grad_norm": 0.400390625,
"learning_rate": 4.004710670154655e-06,
"loss": 0.2817,
"step": 2315
},
{
"epoch": 9.206349206349206,
"grad_norm": 0.396484375,
"learning_rate": 3.8129772771456797e-06,
"loss": 0.2805,
"step": 2320
},
{
"epoch": 9.226190476190476,
"grad_norm": 0.423828125,
"learning_rate": 3.625857786498055e-06,
"loss": 0.2795,
"step": 2325
},
{
"epoch": 9.246031746031745,
"grad_norm": 0.396484375,
"learning_rate": 3.4433611739658645e-06,
"loss": 0.2846,
"step": 2330
},
{
"epoch": 9.265873015873016,
"grad_norm": 0.390625,
"learning_rate": 3.265496193552775e-06,
"loss": 0.2818,
"step": 2335
},
{
"epoch": 9.285714285714286,
"grad_norm": 0.39453125,
"learning_rate": 3.092271377092215e-06,
"loss": 0.2776,
"step": 2340
},
{
"epoch": 9.305555555555555,
"grad_norm": 0.390625,
"learning_rate": 2.9236950338380033e-06,
"loss": 0.2744,
"step": 2345
},
{
"epoch": 9.325396825396826,
"grad_norm": 0.3984375,
"learning_rate": 2.759775250065899e-06,
"loss": 0.2834,
"step": 2350
},
{
"epoch": 9.345238095238095,
"grad_norm": 0.421875,
"learning_rate": 2.6005198886856487e-06,
"loss": 0.2738,
"step": 2355
},
{
"epoch": 9.365079365079366,
"grad_norm": 0.388671875,
"learning_rate": 2.4459365888638062e-06,
"loss": 0.2766,
"step": 2360
},
{
"epoch": 9.384920634920634,
"grad_norm": 0.3984375,
"learning_rate": 2.296032765657319e-06,
"loss": 0.2786,
"step": 2365
},
{
"epoch": 9.404761904761905,
"grad_norm": 0.388671875,
"learning_rate": 2.150815609657875e-06,
"loss": 0.2723,
"step": 2370
},
{
"epoch": 9.424603174603174,
"grad_norm": 0.412109375,
"learning_rate": 2.010292086646903e-06,
"loss": 0.2782,
"step": 2375
},
{
"epoch": 9.444444444444445,
"grad_norm": 0.396484375,
"learning_rate": 1.874468937261531e-06,
"loss": 0.2764,
"step": 2380
},
{
"epoch": 9.464285714285714,
"grad_norm": 0.388671875,
"learning_rate": 1.7433526766711728e-06,
"loss": 0.279,
"step": 2385
},
{
"epoch": 9.484126984126984,
"grad_norm": 0.3984375,
"learning_rate": 1.6169495942650714e-06,
"loss": 0.2832,
"step": 2390
},
{
"epoch": 9.503968253968253,
"grad_norm": 0.384765625,
"learning_rate": 1.4952657533505277e-06,
"loss": 0.2782,
"step": 2395
},
{
"epoch": 9.523809523809524,
"grad_norm": 0.392578125,
"learning_rate": 1.378306990862177e-06,
"loss": 0.2861,
"step": 2400
},
{
"epoch": 9.543650793650794,
"grad_norm": 0.41015625,
"learning_rate": 1.2660789170818809e-06,
"loss": 0.2742,
"step": 2405
},
{
"epoch": 9.563492063492063,
"grad_norm": 0.384765625,
"learning_rate": 1.158586915369675e-06,
"loss": 0.2712,
"step": 2410
},
{
"epoch": 9.583333333333334,
"grad_norm": 0.416015625,
"learning_rate": 1.055836141905553e-06,
"loss": 0.2838,
"step": 2415
},
{
"epoch": 9.603174603174603,
"grad_norm": 0.396484375,
"learning_rate": 9.578315254420767e-07,
"loss": 0.2779,
"step": 2420
},
{
"epoch": 9.623015873015873,
"grad_norm": 0.388671875,
"learning_rate": 8.645777670680089e-07,
"loss": 0.2688,
"step": 2425
},
{
"epoch": 9.642857142857142,
"grad_norm": 0.384765625,
"learning_rate": 7.760793399827937e-07,
"loss": 0.2776,
"step": 2430
},
{
"epoch": 9.662698412698413,
"grad_norm": 0.392578125,
"learning_rate": 6.92340489281973e-07,
"loss": 0.2791,
"step": 2435
},
{
"epoch": 9.682539682539682,
"grad_norm": 0.400390625,
"learning_rate": 6.13365231753571e-07,
"loss": 0.2775,
"step": 2440
},
{
"epoch": 9.702380952380953,
"grad_norm": 0.41015625,
"learning_rate": 5.391573556854157e-07,
"loss": 0.2791,
"step": 2445
},
{
"epoch": 9.722222222222221,
"grad_norm": 0.3984375,
"learning_rate": 4.6972042068341714e-07,
"loss": 0.2818,
"step": 2450
},
{
"epoch": 9.742063492063492,
"grad_norm": 0.3671875,
"learning_rate": 4.050577575008263e-07,
"loss": 0.2768,
"step": 2455
},
{
"epoch": 9.761904761904763,
"grad_norm": 0.3828125,
"learning_rate": 3.451724678784518e-07,
"loss": 0.2791,
"step": 2460
},
{
"epoch": 9.781746031746032,
"grad_norm": 0.392578125,
"learning_rate": 2.9006742439590116e-07,
"loss": 0.274,
"step": 2465
},
{
"epoch": 9.801587301587302,
"grad_norm": 0.41015625,
"learning_rate": 2.397452703337577e-07,
"loss": 0.2767,
"step": 2470
},
{
"epoch": 9.821428571428571,
"grad_norm": 0.380859375,
"learning_rate": 1.9420841954681525e-07,
"loss": 0.2742,
"step": 2475
},
{
"epoch": 9.841269841269842,
"grad_norm": 0.388671875,
"learning_rate": 1.5345905634827074e-07,
"loss": 0.2723,
"step": 2480
},
{
"epoch": 9.86111111111111,
"grad_norm": 0.373046875,
"learning_rate": 1.1749913540496371e-07,
"loss": 0.2813,
"step": 2485
},
{
"epoch": 9.880952380952381,
"grad_norm": 0.37890625,
"learning_rate": 8.633038164358454e-08,
"loss": 0.2789,
"step": 2490
},
{
"epoch": 9.90079365079365,
"grad_norm": 0.41015625,
"learning_rate": 5.995429016798505e-08,
"loss": 0.2819,
"step": 2495
},
{
"epoch": 9.920634920634921,
"grad_norm": 0.400390625,
"learning_rate": 3.8372126187413704e-08,
"loss": 0.2792,
"step": 2500
},
{
"epoch": 9.94047619047619,
"grad_norm": 0.390625,
"learning_rate": 2.1584924955819764e-08,
"loss": 0.2759,
"step": 2505
},
{
"epoch": 9.96031746031746,
"grad_norm": 0.38671875,
"learning_rate": 9.593491722270642e-09,
"loss": 0.2839,
"step": 2510
},
{
"epoch": 9.98015873015873,
"grad_norm": 0.388671875,
"learning_rate": 2.398401692216279e-09,
"loss": 0.2708,
"step": 2515
},
{
"epoch": 10.0,
"grad_norm": 0.390625,
"learning_rate": 0.0,
"loss": 0.2748,
"step": 2520
},
{
"epoch": 10.0,
"eval_loss": 1.6317710876464844,
"eval_runtime": 0.5467,
"eval_samples_per_second": 3.658,
"eval_steps_per_second": 1.829,
"step": 2520
},
{
"epoch": 10.0,
"step": 2520,
"total_flos": 3.8651476831067177e+18,
"train_loss": 1.0268459252421818,
"train_runtime": 8757.4109,
"train_samples_per_second": 4.596,
"train_steps_per_second": 0.288
}
],
"logging_steps": 5,
"max_steps": 2520,
"num_input_tokens_seen": 0,
"num_train_epochs": 10,
"save_steps": 100,
"total_flos": 3.8651476831067177e+18,
"train_batch_size": 2,
"trial_name": null,
"trial_params": null
}