jdannem6's picture
Uploaded checkpoint-30000
36707c4 verified
{
"best_metric": 0.6027244925498962,
"best_model_checkpoint": "runs/deepseek_CMU-AIR2/math-deepseek_FULL_HardArith_Interm_20240424-065814/checkpoint-5000",
"epoch": 0.25,
"eval_steps": 500,
"global_step": 10000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"grad_norm": 4.3125,
"learning_rate": 4.0000000000000003e-07,
"loss": 0.6839,
"step": 10
},
{
"epoch": 0.0,
"grad_norm": 6.53125,
"learning_rate": 8.000000000000001e-07,
"loss": 0.6587,
"step": 20
},
{
"epoch": 0.0,
"grad_norm": 6.34375,
"learning_rate": 1.2000000000000002e-06,
"loss": 0.6644,
"step": 30
},
{
"epoch": 0.0,
"grad_norm": 2.359375,
"learning_rate": 1.6000000000000001e-06,
"loss": 0.6101,
"step": 40
},
{
"epoch": 0.0,
"grad_norm": 4.90625,
"learning_rate": 2.0000000000000003e-06,
"loss": 0.6188,
"step": 50
},
{
"epoch": 0.0,
"grad_norm": 3.390625,
"learning_rate": 2.4000000000000003e-06,
"loss": 0.6524,
"step": 60
},
{
"epoch": 0.0,
"grad_norm": 3.75,
"learning_rate": 2.8000000000000003e-06,
"loss": 0.6322,
"step": 70
},
{
"epoch": 0.0,
"grad_norm": 3.484375,
"learning_rate": 3.2000000000000003e-06,
"loss": 0.676,
"step": 80
},
{
"epoch": 0.0,
"grad_norm": 4.40625,
"learning_rate": 3.6000000000000003e-06,
"loss": 0.6578,
"step": 90
},
{
"epoch": 0.0,
"grad_norm": 4.25,
"learning_rate": 4.000000000000001e-06,
"loss": 0.597,
"step": 100
},
{
"epoch": 0.0,
"grad_norm": 4.4375,
"learning_rate": 4.4e-06,
"loss": 0.6543,
"step": 110
},
{
"epoch": 0.0,
"grad_norm": 7.03125,
"learning_rate": 4.800000000000001e-06,
"loss": 0.6844,
"step": 120
},
{
"epoch": 0.0,
"grad_norm": 6.09375,
"learning_rate": 5.2e-06,
"loss": 0.6658,
"step": 130
},
{
"epoch": 0.0,
"grad_norm": 2.96875,
"learning_rate": 5.600000000000001e-06,
"loss": 0.7388,
"step": 140
},
{
"epoch": 0.0,
"grad_norm": 4.5625,
"learning_rate": 6e-06,
"loss": 0.6627,
"step": 150
},
{
"epoch": 0.0,
"grad_norm": 8.125,
"learning_rate": 6.4000000000000006e-06,
"loss": 0.7036,
"step": 160
},
{
"epoch": 0.0,
"grad_norm": 4.9375,
"learning_rate": 6.800000000000001e-06,
"loss": 0.6136,
"step": 170
},
{
"epoch": 0.0,
"grad_norm": 2.546875,
"learning_rate": 7.2000000000000005e-06,
"loss": 0.5902,
"step": 180
},
{
"epoch": 0.0,
"grad_norm": 2.265625,
"learning_rate": 7.600000000000001e-06,
"loss": 0.6254,
"step": 190
},
{
"epoch": 0.01,
"grad_norm": 8.9375,
"learning_rate": 8.000000000000001e-06,
"loss": 0.6322,
"step": 200
},
{
"epoch": 0.01,
"grad_norm": 2.546875,
"learning_rate": 8.400000000000001e-06,
"loss": 0.6152,
"step": 210
},
{
"epoch": 0.01,
"grad_norm": 5.15625,
"learning_rate": 8.8e-06,
"loss": 0.5843,
"step": 220
},
{
"epoch": 0.01,
"grad_norm": 5.96875,
"learning_rate": 9.200000000000002e-06,
"loss": 0.7049,
"step": 230
},
{
"epoch": 0.01,
"grad_norm": 4.34375,
"learning_rate": 9.600000000000001e-06,
"loss": 0.4978,
"step": 240
},
{
"epoch": 0.01,
"grad_norm": 3.6875,
"learning_rate": 1e-05,
"loss": 0.603,
"step": 250
},
{
"epoch": 0.01,
"grad_norm": 7.0625,
"learning_rate": 1.04e-05,
"loss": 0.6845,
"step": 260
},
{
"epoch": 0.01,
"grad_norm": 9.9375,
"learning_rate": 1.0800000000000002e-05,
"loss": 0.5937,
"step": 270
},
{
"epoch": 0.01,
"grad_norm": 5.6875,
"learning_rate": 1.1200000000000001e-05,
"loss": 0.5744,
"step": 280
},
{
"epoch": 0.01,
"grad_norm": 5.3125,
"learning_rate": 1.16e-05,
"loss": 0.7565,
"step": 290
},
{
"epoch": 0.01,
"grad_norm": 8.125,
"learning_rate": 1.2e-05,
"loss": 0.6332,
"step": 300
},
{
"epoch": 0.01,
"grad_norm": 5.84375,
"learning_rate": 1.2400000000000002e-05,
"loss": 0.5884,
"step": 310
},
{
"epoch": 0.01,
"grad_norm": 5.625,
"learning_rate": 1.2800000000000001e-05,
"loss": 0.6418,
"step": 320
},
{
"epoch": 0.01,
"grad_norm": 2.421875,
"learning_rate": 1.3200000000000002e-05,
"loss": 0.6532,
"step": 330
},
{
"epoch": 0.01,
"grad_norm": 4.125,
"learning_rate": 1.3600000000000002e-05,
"loss": 0.5984,
"step": 340
},
{
"epoch": 0.01,
"grad_norm": 5.4375,
"learning_rate": 1.4e-05,
"loss": 0.7039,
"step": 350
},
{
"epoch": 0.01,
"grad_norm": 4.15625,
"learning_rate": 1.4400000000000001e-05,
"loss": 0.5297,
"step": 360
},
{
"epoch": 0.01,
"grad_norm": 2.71875,
"learning_rate": 1.48e-05,
"loss": 0.6239,
"step": 370
},
{
"epoch": 0.01,
"grad_norm": 4.5,
"learning_rate": 1.5200000000000002e-05,
"loss": 0.6246,
"step": 380
},
{
"epoch": 0.01,
"grad_norm": 3.421875,
"learning_rate": 1.5600000000000003e-05,
"loss": 0.6023,
"step": 390
},
{
"epoch": 0.01,
"grad_norm": 2.671875,
"learning_rate": 1.6000000000000003e-05,
"loss": 0.6448,
"step": 400
},
{
"epoch": 0.01,
"grad_norm": 14.75,
"learning_rate": 1.64e-05,
"loss": 0.7238,
"step": 410
},
{
"epoch": 0.01,
"grad_norm": 4.1875,
"learning_rate": 1.6800000000000002e-05,
"loss": 0.6137,
"step": 420
},
{
"epoch": 0.01,
"grad_norm": 2.828125,
"learning_rate": 1.72e-05,
"loss": 0.6496,
"step": 430
},
{
"epoch": 0.01,
"grad_norm": 8.1875,
"learning_rate": 1.76e-05,
"loss": 0.5152,
"step": 440
},
{
"epoch": 0.01,
"grad_norm": 7.5625,
"learning_rate": 1.8e-05,
"loss": 0.5874,
"step": 450
},
{
"epoch": 0.01,
"grad_norm": 11.875,
"learning_rate": 1.8400000000000003e-05,
"loss": 0.6662,
"step": 460
},
{
"epoch": 0.01,
"grad_norm": 3.515625,
"learning_rate": 1.88e-05,
"loss": 0.6845,
"step": 470
},
{
"epoch": 0.01,
"grad_norm": 10.5625,
"learning_rate": 1.9200000000000003e-05,
"loss": 0.6917,
"step": 480
},
{
"epoch": 0.01,
"grad_norm": 3.0625,
"learning_rate": 1.9600000000000002e-05,
"loss": 0.5512,
"step": 490
},
{
"epoch": 0.01,
"grad_norm": 2.546875,
"learning_rate": 2e-05,
"loss": 0.6037,
"step": 500
},
{
"epoch": 0.01,
"eval_loss": 0.6478177309036255,
"eval_runtime": 38.0753,
"eval_samples_per_second": 26.264,
"eval_steps_per_second": 26.264,
"step": 500
},
{
"epoch": 0.01,
"grad_norm": 5.125,
"learning_rate": 1.9978947368421054e-05,
"loss": 0.6223,
"step": 510
},
{
"epoch": 0.01,
"grad_norm": 3.4375,
"learning_rate": 1.9957894736842107e-05,
"loss": 0.6301,
"step": 520
},
{
"epoch": 0.01,
"grad_norm": 7.65625,
"learning_rate": 1.993684210526316e-05,
"loss": 0.6224,
"step": 530
},
{
"epoch": 0.01,
"grad_norm": 9.125,
"learning_rate": 1.9915789473684212e-05,
"loss": 0.6255,
"step": 540
},
{
"epoch": 0.01,
"grad_norm": 2.96875,
"learning_rate": 1.9894736842105265e-05,
"loss": 0.6737,
"step": 550
},
{
"epoch": 0.01,
"grad_norm": 11.8125,
"learning_rate": 1.9873684210526318e-05,
"loss": 0.6218,
"step": 560
},
{
"epoch": 0.01,
"grad_norm": 5.15625,
"learning_rate": 1.985263157894737e-05,
"loss": 0.5187,
"step": 570
},
{
"epoch": 0.01,
"grad_norm": 6.0625,
"learning_rate": 1.9831578947368423e-05,
"loss": 0.6786,
"step": 580
},
{
"epoch": 0.01,
"grad_norm": 5.125,
"learning_rate": 1.9810526315789476e-05,
"loss": 0.6641,
"step": 590
},
{
"epoch": 0.01,
"grad_norm": 3.75,
"learning_rate": 1.9789473684210528e-05,
"loss": 0.5541,
"step": 600
},
{
"epoch": 0.02,
"grad_norm": 4.4375,
"learning_rate": 1.976842105263158e-05,
"loss": 0.565,
"step": 610
},
{
"epoch": 0.02,
"grad_norm": 10.375,
"learning_rate": 1.9747368421052633e-05,
"loss": 0.7099,
"step": 620
},
{
"epoch": 0.02,
"grad_norm": 5.09375,
"learning_rate": 1.9726315789473686e-05,
"loss": 0.5988,
"step": 630
},
{
"epoch": 0.02,
"grad_norm": 4.0625,
"learning_rate": 1.970526315789474e-05,
"loss": 0.7724,
"step": 640
},
{
"epoch": 0.02,
"grad_norm": 3.0625,
"learning_rate": 1.968421052631579e-05,
"loss": 0.582,
"step": 650
},
{
"epoch": 0.02,
"grad_norm": 3.65625,
"learning_rate": 1.9663157894736844e-05,
"loss": 0.6805,
"step": 660
},
{
"epoch": 0.02,
"grad_norm": 11.0,
"learning_rate": 1.9642105263157897e-05,
"loss": 0.6248,
"step": 670
},
{
"epoch": 0.02,
"grad_norm": 2.734375,
"learning_rate": 1.962105263157895e-05,
"loss": 0.5691,
"step": 680
},
{
"epoch": 0.02,
"grad_norm": 3.34375,
"learning_rate": 1.9600000000000002e-05,
"loss": 0.6657,
"step": 690
},
{
"epoch": 0.02,
"grad_norm": 2.078125,
"learning_rate": 1.9578947368421055e-05,
"loss": 0.4786,
"step": 700
},
{
"epoch": 0.02,
"grad_norm": 7.125,
"learning_rate": 1.9557894736842107e-05,
"loss": 0.6368,
"step": 710
},
{
"epoch": 0.02,
"grad_norm": 5.84375,
"learning_rate": 1.953684210526316e-05,
"loss": 0.6108,
"step": 720
},
{
"epoch": 0.02,
"grad_norm": 3.53125,
"learning_rate": 1.9515789473684213e-05,
"loss": 0.6331,
"step": 730
},
{
"epoch": 0.02,
"grad_norm": 3.96875,
"learning_rate": 1.9494736842105265e-05,
"loss": 0.5432,
"step": 740
},
{
"epoch": 0.02,
"grad_norm": 4.25,
"learning_rate": 1.9473684210526318e-05,
"loss": 0.7021,
"step": 750
},
{
"epoch": 0.02,
"grad_norm": 10.375,
"learning_rate": 1.945263157894737e-05,
"loss": 0.68,
"step": 760
},
{
"epoch": 0.02,
"grad_norm": 3.359375,
"learning_rate": 1.9431578947368423e-05,
"loss": 0.629,
"step": 770
},
{
"epoch": 0.02,
"grad_norm": 4.5625,
"learning_rate": 1.9410526315789476e-05,
"loss": 0.6685,
"step": 780
},
{
"epoch": 0.02,
"grad_norm": 4.21875,
"learning_rate": 1.9389473684210525e-05,
"loss": 0.5552,
"step": 790
},
{
"epoch": 0.02,
"grad_norm": 2.765625,
"learning_rate": 1.936842105263158e-05,
"loss": 0.657,
"step": 800
},
{
"epoch": 0.02,
"grad_norm": 7.71875,
"learning_rate": 1.9347368421052634e-05,
"loss": 0.5431,
"step": 810
},
{
"epoch": 0.02,
"grad_norm": 4.5,
"learning_rate": 1.9326315789473687e-05,
"loss": 0.619,
"step": 820
},
{
"epoch": 0.02,
"grad_norm": 7.53125,
"learning_rate": 1.930526315789474e-05,
"loss": 0.7058,
"step": 830
},
{
"epoch": 0.02,
"grad_norm": 8.5625,
"learning_rate": 1.9284210526315792e-05,
"loss": 0.631,
"step": 840
},
{
"epoch": 0.02,
"grad_norm": 6.1875,
"learning_rate": 1.9263157894736845e-05,
"loss": 0.5991,
"step": 850
},
{
"epoch": 0.02,
"grad_norm": 3.890625,
"learning_rate": 1.9242105263157894e-05,
"loss": 0.6852,
"step": 860
},
{
"epoch": 0.02,
"grad_norm": 7.46875,
"learning_rate": 1.922105263157895e-05,
"loss": 0.6474,
"step": 870
},
{
"epoch": 0.02,
"grad_norm": 3.46875,
"learning_rate": 1.9200000000000003e-05,
"loss": 0.5979,
"step": 880
},
{
"epoch": 0.02,
"grad_norm": 4.625,
"learning_rate": 1.9178947368421055e-05,
"loss": 0.7035,
"step": 890
},
{
"epoch": 0.02,
"grad_norm": 3.640625,
"learning_rate": 1.9157894736842108e-05,
"loss": 0.7001,
"step": 900
},
{
"epoch": 0.02,
"grad_norm": 2.578125,
"learning_rate": 1.913684210526316e-05,
"loss": 0.6453,
"step": 910
},
{
"epoch": 0.02,
"grad_norm": 4.4375,
"learning_rate": 1.9115789473684213e-05,
"loss": 0.6356,
"step": 920
},
{
"epoch": 0.02,
"grad_norm": 9.1875,
"learning_rate": 1.9094736842105262e-05,
"loss": 0.5679,
"step": 930
},
{
"epoch": 0.02,
"grad_norm": 3.46875,
"learning_rate": 1.907368421052632e-05,
"loss": 0.7062,
"step": 940
},
{
"epoch": 0.02,
"grad_norm": 5.5625,
"learning_rate": 1.9052631578947368e-05,
"loss": 0.665,
"step": 950
},
{
"epoch": 0.02,
"grad_norm": 7.96875,
"learning_rate": 1.9031578947368424e-05,
"loss": 0.6772,
"step": 960
},
{
"epoch": 0.02,
"grad_norm": 7.09375,
"learning_rate": 1.9010526315789476e-05,
"loss": 0.5347,
"step": 970
},
{
"epoch": 0.02,
"grad_norm": 4.65625,
"learning_rate": 1.898947368421053e-05,
"loss": 0.6161,
"step": 980
},
{
"epoch": 0.02,
"grad_norm": 3.796875,
"learning_rate": 1.8968421052631582e-05,
"loss": 0.595,
"step": 990
},
{
"epoch": 0.03,
"grad_norm": 3.03125,
"learning_rate": 1.894736842105263e-05,
"loss": 0.6752,
"step": 1000
},
{
"epoch": 0.03,
"eval_loss": 0.6129263639450073,
"eval_runtime": 38.0435,
"eval_samples_per_second": 26.286,
"eval_steps_per_second": 26.286,
"step": 1000
},
{
"epoch": 0.03,
"grad_norm": 5.5,
"learning_rate": 1.8926315789473687e-05,
"loss": 0.6876,
"step": 1010
},
{
"epoch": 0.03,
"grad_norm": 3.640625,
"learning_rate": 1.8905263157894736e-05,
"loss": 0.6364,
"step": 1020
},
{
"epoch": 0.03,
"grad_norm": 6.6875,
"learning_rate": 1.8884210526315792e-05,
"loss": 0.6598,
"step": 1030
},
{
"epoch": 0.03,
"grad_norm": 4.09375,
"learning_rate": 1.886315789473684e-05,
"loss": 0.6473,
"step": 1040
},
{
"epoch": 0.03,
"grad_norm": 5.125,
"learning_rate": 1.8842105263157898e-05,
"loss": 0.5972,
"step": 1050
},
{
"epoch": 0.03,
"grad_norm": 3.765625,
"learning_rate": 1.882105263157895e-05,
"loss": 0.7538,
"step": 1060
},
{
"epoch": 0.03,
"grad_norm": 4.625,
"learning_rate": 1.88e-05,
"loss": 0.6282,
"step": 1070
},
{
"epoch": 0.03,
"grad_norm": 3.9375,
"learning_rate": 1.8778947368421056e-05,
"loss": 0.7304,
"step": 1080
},
{
"epoch": 0.03,
"grad_norm": 12.625,
"learning_rate": 1.8757894736842105e-05,
"loss": 0.6744,
"step": 1090
},
{
"epoch": 0.03,
"grad_norm": 3.84375,
"learning_rate": 1.873684210526316e-05,
"loss": 0.7072,
"step": 1100
},
{
"epoch": 0.03,
"grad_norm": 3.65625,
"learning_rate": 1.871578947368421e-05,
"loss": 0.6935,
"step": 1110
},
{
"epoch": 0.03,
"grad_norm": 7.125,
"learning_rate": 1.8694736842105266e-05,
"loss": 0.6507,
"step": 1120
},
{
"epoch": 0.03,
"grad_norm": 0.326171875,
"learning_rate": 1.8673684210526316e-05,
"loss": 0.5451,
"step": 1130
},
{
"epoch": 0.03,
"grad_norm": 8.0,
"learning_rate": 1.8652631578947368e-05,
"loss": 0.6145,
"step": 1140
},
{
"epoch": 0.03,
"grad_norm": 4.09375,
"learning_rate": 1.8631578947368424e-05,
"loss": 0.6846,
"step": 1150
},
{
"epoch": 0.03,
"grad_norm": 3.734375,
"learning_rate": 1.8610526315789473e-05,
"loss": 0.6139,
"step": 1160
},
{
"epoch": 0.03,
"grad_norm": 6.71875,
"learning_rate": 1.858947368421053e-05,
"loss": 0.5147,
"step": 1170
},
{
"epoch": 0.03,
"grad_norm": 4.125,
"learning_rate": 1.856842105263158e-05,
"loss": 0.5526,
"step": 1180
},
{
"epoch": 0.03,
"grad_norm": 13.6875,
"learning_rate": 1.8547368421052635e-05,
"loss": 0.6973,
"step": 1190
},
{
"epoch": 0.03,
"grad_norm": 3.8125,
"learning_rate": 1.8526315789473684e-05,
"loss": 0.6338,
"step": 1200
},
{
"epoch": 0.03,
"grad_norm": 3.234375,
"learning_rate": 1.8505263157894737e-05,
"loss": 0.5985,
"step": 1210
},
{
"epoch": 0.03,
"grad_norm": 4.625,
"learning_rate": 1.8484210526315793e-05,
"loss": 0.6832,
"step": 1220
},
{
"epoch": 0.03,
"grad_norm": 4.59375,
"learning_rate": 1.8463157894736842e-05,
"loss": 0.6766,
"step": 1230
},
{
"epoch": 0.03,
"grad_norm": 5.09375,
"learning_rate": 1.8442105263157898e-05,
"loss": 0.6342,
"step": 1240
},
{
"epoch": 0.03,
"grad_norm": 5.0625,
"learning_rate": 1.8421052631578947e-05,
"loss": 0.6043,
"step": 1250
},
{
"epoch": 0.03,
"grad_norm": 3.34375,
"learning_rate": 1.8400000000000003e-05,
"loss": 0.5879,
"step": 1260
},
{
"epoch": 0.03,
"grad_norm": 4.84375,
"learning_rate": 1.8378947368421053e-05,
"loss": 0.6553,
"step": 1270
},
{
"epoch": 0.03,
"grad_norm": 2.984375,
"learning_rate": 1.8357894736842105e-05,
"loss": 0.6256,
"step": 1280
},
{
"epoch": 0.03,
"grad_norm": 5.75,
"learning_rate": 1.8336842105263158e-05,
"loss": 0.7604,
"step": 1290
},
{
"epoch": 0.03,
"grad_norm": 9.0,
"learning_rate": 1.831578947368421e-05,
"loss": 0.5902,
"step": 1300
},
{
"epoch": 0.03,
"grad_norm": 3.328125,
"learning_rate": 1.8294736842105267e-05,
"loss": 0.6895,
"step": 1310
},
{
"epoch": 0.03,
"grad_norm": 3.171875,
"learning_rate": 1.8273684210526316e-05,
"loss": 0.557,
"step": 1320
},
{
"epoch": 0.03,
"grad_norm": 2.59375,
"learning_rate": 1.8252631578947372e-05,
"loss": 0.5967,
"step": 1330
},
{
"epoch": 0.03,
"grad_norm": 2.171875,
"learning_rate": 1.823157894736842e-05,
"loss": 0.7342,
"step": 1340
},
{
"epoch": 0.03,
"grad_norm": 3.65625,
"learning_rate": 1.8210526315789477e-05,
"loss": 0.5225,
"step": 1350
},
{
"epoch": 0.03,
"grad_norm": 2.84375,
"learning_rate": 1.8189473684210527e-05,
"loss": 0.6079,
"step": 1360
},
{
"epoch": 0.03,
"grad_norm": 2.546875,
"learning_rate": 1.816842105263158e-05,
"loss": 0.5944,
"step": 1370
},
{
"epoch": 0.03,
"grad_norm": 3.0625,
"learning_rate": 1.8147368421052632e-05,
"loss": 0.5262,
"step": 1380
},
{
"epoch": 0.03,
"grad_norm": 1.7578125,
"learning_rate": 1.8126315789473685e-05,
"loss": 0.5117,
"step": 1390
},
{
"epoch": 0.04,
"grad_norm": 8.3125,
"learning_rate": 1.810526315789474e-05,
"loss": 0.7201,
"step": 1400
},
{
"epoch": 0.04,
"grad_norm": 2.171875,
"learning_rate": 1.808421052631579e-05,
"loss": 0.6592,
"step": 1410
},
{
"epoch": 0.04,
"grad_norm": 4.34375,
"learning_rate": 1.8063157894736846e-05,
"loss": 0.6966,
"step": 1420
},
{
"epoch": 0.04,
"grad_norm": 5.9375,
"learning_rate": 1.8042105263157895e-05,
"loss": 0.5328,
"step": 1430
},
{
"epoch": 0.04,
"grad_norm": 4.03125,
"learning_rate": 1.8021052631578948e-05,
"loss": 0.6245,
"step": 1440
},
{
"epoch": 0.04,
"grad_norm": 3.03125,
"learning_rate": 1.8e-05,
"loss": 0.5457,
"step": 1450
},
{
"epoch": 0.04,
"grad_norm": 2.9375,
"learning_rate": 1.7978947368421053e-05,
"loss": 0.6026,
"step": 1460
},
{
"epoch": 0.04,
"grad_norm": 3.546875,
"learning_rate": 1.795789473684211e-05,
"loss": 0.6519,
"step": 1470
},
{
"epoch": 0.04,
"grad_norm": 4.8125,
"learning_rate": 1.793684210526316e-05,
"loss": 0.5364,
"step": 1480
},
{
"epoch": 0.04,
"grad_norm": 4.9375,
"learning_rate": 1.7915789473684214e-05,
"loss": 0.6421,
"step": 1490
},
{
"epoch": 0.04,
"grad_norm": 5.03125,
"learning_rate": 1.7894736842105264e-05,
"loss": 0.5376,
"step": 1500
},
{
"epoch": 0.04,
"eval_loss": 0.6211587190628052,
"eval_runtime": 38.148,
"eval_samples_per_second": 26.214,
"eval_steps_per_second": 26.214,
"step": 1500
},
{
"epoch": 0.04,
"grad_norm": 5.125,
"learning_rate": 1.7873684210526316e-05,
"loss": 0.5633,
"step": 1510
},
{
"epoch": 0.04,
"grad_norm": 14.0,
"learning_rate": 1.785263157894737e-05,
"loss": 0.6917,
"step": 1520
},
{
"epoch": 0.04,
"grad_norm": 2.59375,
"learning_rate": 1.7831578947368422e-05,
"loss": 0.5882,
"step": 1530
},
{
"epoch": 0.04,
"grad_norm": 12.0625,
"learning_rate": 1.7810526315789474e-05,
"loss": 0.5752,
"step": 1540
},
{
"epoch": 0.04,
"grad_norm": 2.96875,
"learning_rate": 1.7789473684210527e-05,
"loss": 0.6439,
"step": 1550
},
{
"epoch": 0.04,
"grad_norm": 6.0625,
"learning_rate": 1.7768421052631583e-05,
"loss": 0.6806,
"step": 1560
},
{
"epoch": 0.04,
"grad_norm": 3.484375,
"learning_rate": 1.7747368421052632e-05,
"loss": 0.5903,
"step": 1570
},
{
"epoch": 0.04,
"grad_norm": 4.46875,
"learning_rate": 1.7726315789473685e-05,
"loss": 0.6394,
"step": 1580
},
{
"epoch": 0.04,
"grad_norm": 3.5,
"learning_rate": 1.7705263157894738e-05,
"loss": 0.53,
"step": 1590
},
{
"epoch": 0.04,
"grad_norm": 10.625,
"learning_rate": 1.768421052631579e-05,
"loss": 0.7113,
"step": 1600
},
{
"epoch": 0.04,
"grad_norm": 10.875,
"learning_rate": 1.7663157894736843e-05,
"loss": 0.6491,
"step": 1610
},
{
"epoch": 0.04,
"grad_norm": 4.96875,
"learning_rate": 1.7642105263157896e-05,
"loss": 0.6439,
"step": 1620
},
{
"epoch": 0.04,
"grad_norm": 9.5625,
"learning_rate": 1.7621052631578948e-05,
"loss": 0.7248,
"step": 1630
},
{
"epoch": 0.04,
"grad_norm": 6.96875,
"learning_rate": 1.76e-05,
"loss": 0.6754,
"step": 1640
},
{
"epoch": 0.04,
"grad_norm": 3.828125,
"learning_rate": 1.7578947368421054e-05,
"loss": 0.6029,
"step": 1650
},
{
"epoch": 0.04,
"grad_norm": 6.0,
"learning_rate": 1.7557894736842106e-05,
"loss": 0.6225,
"step": 1660
},
{
"epoch": 0.04,
"grad_norm": 3.265625,
"learning_rate": 1.753684210526316e-05,
"loss": 0.5892,
"step": 1670
},
{
"epoch": 0.04,
"grad_norm": 3.390625,
"learning_rate": 1.751578947368421e-05,
"loss": 0.7917,
"step": 1680
},
{
"epoch": 0.04,
"grad_norm": 3.015625,
"learning_rate": 1.7494736842105264e-05,
"loss": 0.5863,
"step": 1690
},
{
"epoch": 0.04,
"grad_norm": 3.734375,
"learning_rate": 1.7473684210526317e-05,
"loss": 0.5613,
"step": 1700
},
{
"epoch": 0.04,
"grad_norm": 11.6875,
"learning_rate": 1.745263157894737e-05,
"loss": 0.6582,
"step": 1710
},
{
"epoch": 0.04,
"grad_norm": 5.0,
"learning_rate": 1.7431578947368422e-05,
"loss": 0.6967,
"step": 1720
},
{
"epoch": 0.04,
"grad_norm": 4.21875,
"learning_rate": 1.7410526315789475e-05,
"loss": 0.6296,
"step": 1730
},
{
"epoch": 0.04,
"grad_norm": 26.0,
"learning_rate": 1.7389473684210527e-05,
"loss": 0.5804,
"step": 1740
},
{
"epoch": 0.04,
"grad_norm": 3.015625,
"learning_rate": 1.736842105263158e-05,
"loss": 0.7102,
"step": 1750
},
{
"epoch": 0.04,
"grad_norm": 2.640625,
"learning_rate": 1.7347368421052633e-05,
"loss": 0.6108,
"step": 1760
},
{
"epoch": 0.04,
"grad_norm": 3.9375,
"learning_rate": 1.7326315789473685e-05,
"loss": 0.6847,
"step": 1770
},
{
"epoch": 0.04,
"grad_norm": 3.890625,
"learning_rate": 1.7305263157894738e-05,
"loss": 0.7885,
"step": 1780
},
{
"epoch": 0.04,
"grad_norm": 5.53125,
"learning_rate": 1.728421052631579e-05,
"loss": 0.674,
"step": 1790
},
{
"epoch": 0.04,
"grad_norm": 5.25,
"learning_rate": 1.7263157894736843e-05,
"loss": 0.6753,
"step": 1800
},
{
"epoch": 0.05,
"grad_norm": 5.0,
"learning_rate": 1.7242105263157896e-05,
"loss": 0.5909,
"step": 1810
},
{
"epoch": 0.05,
"grad_norm": 1.8984375,
"learning_rate": 1.722105263157895e-05,
"loss": 0.6107,
"step": 1820
},
{
"epoch": 0.05,
"grad_norm": 2.875,
"learning_rate": 1.72e-05,
"loss": 0.623,
"step": 1830
},
{
"epoch": 0.05,
"grad_norm": 4.84375,
"learning_rate": 1.7178947368421054e-05,
"loss": 0.5906,
"step": 1840
},
{
"epoch": 0.05,
"grad_norm": 3.484375,
"learning_rate": 1.7157894736842107e-05,
"loss": 0.5833,
"step": 1850
},
{
"epoch": 0.05,
"grad_norm": 9.25,
"learning_rate": 1.713684210526316e-05,
"loss": 0.5935,
"step": 1860
},
{
"epoch": 0.05,
"grad_norm": 3.59375,
"learning_rate": 1.7115789473684212e-05,
"loss": 0.5972,
"step": 1870
},
{
"epoch": 0.05,
"grad_norm": 1.8046875,
"learning_rate": 1.7094736842105265e-05,
"loss": 0.5128,
"step": 1880
},
{
"epoch": 0.05,
"grad_norm": 4.9375,
"learning_rate": 1.7073684210526317e-05,
"loss": 0.6828,
"step": 1890
},
{
"epoch": 0.05,
"grad_norm": 3.671875,
"learning_rate": 1.705263157894737e-05,
"loss": 0.5969,
"step": 1900
},
{
"epoch": 0.05,
"grad_norm": 4.65625,
"learning_rate": 1.7031578947368423e-05,
"loss": 0.5967,
"step": 1910
},
{
"epoch": 0.05,
"grad_norm": 8.5625,
"learning_rate": 1.7010526315789475e-05,
"loss": 0.6486,
"step": 1920
},
{
"epoch": 0.05,
"grad_norm": 2.953125,
"learning_rate": 1.6989473684210528e-05,
"loss": 0.6343,
"step": 1930
},
{
"epoch": 0.05,
"grad_norm": 4.53125,
"learning_rate": 1.696842105263158e-05,
"loss": 0.6137,
"step": 1940
},
{
"epoch": 0.05,
"grad_norm": 3.1875,
"learning_rate": 1.6947368421052633e-05,
"loss": 0.5584,
"step": 1950
},
{
"epoch": 0.05,
"grad_norm": 4.6875,
"learning_rate": 1.6926315789473686e-05,
"loss": 0.5924,
"step": 1960
},
{
"epoch": 0.05,
"grad_norm": 3.359375,
"learning_rate": 1.690526315789474e-05,
"loss": 0.5658,
"step": 1970
},
{
"epoch": 0.05,
"grad_norm": 3.921875,
"learning_rate": 1.688421052631579e-05,
"loss": 0.6455,
"step": 1980
},
{
"epoch": 0.05,
"grad_norm": 4.0625,
"learning_rate": 1.6863157894736844e-05,
"loss": 0.7716,
"step": 1990
},
{
"epoch": 0.05,
"grad_norm": 5.34375,
"learning_rate": 1.6842105263157896e-05,
"loss": 0.6834,
"step": 2000
},
{
"epoch": 0.05,
"eval_loss": 0.6163671612739563,
"eval_runtime": 38.1399,
"eval_samples_per_second": 26.219,
"eval_steps_per_second": 26.219,
"step": 2000
},
{
"epoch": 0.05,
"grad_norm": 5.3125,
"learning_rate": 1.682105263157895e-05,
"loss": 0.655,
"step": 2010
},
{
"epoch": 0.05,
"grad_norm": 8.625,
"learning_rate": 1.6800000000000002e-05,
"loss": 0.6942,
"step": 2020
},
{
"epoch": 0.05,
"grad_norm": 4.8125,
"learning_rate": 1.6778947368421054e-05,
"loss": 0.6266,
"step": 2030
},
{
"epoch": 0.05,
"grad_norm": 6.46875,
"learning_rate": 1.6757894736842107e-05,
"loss": 0.6633,
"step": 2040
},
{
"epoch": 0.05,
"grad_norm": 6.625,
"learning_rate": 1.673684210526316e-05,
"loss": 0.6354,
"step": 2050
},
{
"epoch": 0.05,
"grad_norm": 3.203125,
"learning_rate": 1.6715789473684212e-05,
"loss": 0.6723,
"step": 2060
},
{
"epoch": 0.05,
"grad_norm": 3.734375,
"learning_rate": 1.6694736842105265e-05,
"loss": 0.677,
"step": 2070
},
{
"epoch": 0.05,
"grad_norm": 4.28125,
"learning_rate": 1.6673684210526318e-05,
"loss": 0.6329,
"step": 2080
},
{
"epoch": 0.05,
"grad_norm": 7.90625,
"learning_rate": 1.665263157894737e-05,
"loss": 0.6193,
"step": 2090
},
{
"epoch": 0.05,
"grad_norm": 6.46875,
"learning_rate": 1.6631578947368423e-05,
"loss": 0.6225,
"step": 2100
},
{
"epoch": 0.05,
"grad_norm": 2.75,
"learning_rate": 1.6610526315789476e-05,
"loss": 0.6449,
"step": 2110
},
{
"epoch": 0.05,
"grad_norm": 3.921875,
"learning_rate": 1.658947368421053e-05,
"loss": 0.5844,
"step": 2120
},
{
"epoch": 0.05,
"grad_norm": 6.59375,
"learning_rate": 1.656842105263158e-05,
"loss": 0.6374,
"step": 2130
},
{
"epoch": 0.05,
"grad_norm": 6.0,
"learning_rate": 1.6547368421052634e-05,
"loss": 0.6147,
"step": 2140
},
{
"epoch": 0.05,
"grad_norm": 5.6875,
"learning_rate": 1.6526315789473686e-05,
"loss": 0.6639,
"step": 2150
},
{
"epoch": 0.05,
"grad_norm": 3.109375,
"learning_rate": 1.650526315789474e-05,
"loss": 0.6715,
"step": 2160
},
{
"epoch": 0.05,
"grad_norm": 2.40625,
"learning_rate": 1.648421052631579e-05,
"loss": 0.718,
"step": 2170
},
{
"epoch": 0.05,
"grad_norm": 5.71875,
"learning_rate": 1.6463157894736844e-05,
"loss": 0.7205,
"step": 2180
},
{
"epoch": 0.05,
"grad_norm": 4.3125,
"learning_rate": 1.6442105263157897e-05,
"loss": 0.6613,
"step": 2190
},
{
"epoch": 0.06,
"grad_norm": 2.6875,
"learning_rate": 1.642105263157895e-05,
"loss": 0.7353,
"step": 2200
},
{
"epoch": 0.06,
"grad_norm": 2.078125,
"learning_rate": 1.64e-05,
"loss": 0.5747,
"step": 2210
},
{
"epoch": 0.06,
"grad_norm": 5.84375,
"learning_rate": 1.6378947368421055e-05,
"loss": 0.4669,
"step": 2220
},
{
"epoch": 0.06,
"grad_norm": 4.15625,
"learning_rate": 1.6357894736842108e-05,
"loss": 0.5741,
"step": 2230
},
{
"epoch": 0.06,
"grad_norm": 2.265625,
"learning_rate": 1.633684210526316e-05,
"loss": 0.6338,
"step": 2240
},
{
"epoch": 0.06,
"grad_norm": 2.96875,
"learning_rate": 1.6315789473684213e-05,
"loss": 0.5429,
"step": 2250
},
{
"epoch": 0.06,
"grad_norm": 2.34375,
"learning_rate": 1.6294736842105265e-05,
"loss": 0.6671,
"step": 2260
},
{
"epoch": 0.06,
"grad_norm": 2.59375,
"learning_rate": 1.6273684210526318e-05,
"loss": 0.681,
"step": 2270
},
{
"epoch": 0.06,
"grad_norm": 10.125,
"learning_rate": 1.6252631578947367e-05,
"loss": 0.5757,
"step": 2280
},
{
"epoch": 0.06,
"grad_norm": 4.75,
"learning_rate": 1.6231578947368423e-05,
"loss": 0.559,
"step": 2290
},
{
"epoch": 0.06,
"grad_norm": 4.09375,
"learning_rate": 1.6210526315789473e-05,
"loss": 0.6697,
"step": 2300
},
{
"epoch": 0.06,
"grad_norm": 5.15625,
"learning_rate": 1.618947368421053e-05,
"loss": 0.6065,
"step": 2310
},
{
"epoch": 0.06,
"grad_norm": 5.59375,
"learning_rate": 1.616842105263158e-05,
"loss": 0.6676,
"step": 2320
},
{
"epoch": 0.06,
"grad_norm": 5.1875,
"learning_rate": 1.6147368421052634e-05,
"loss": 0.5642,
"step": 2330
},
{
"epoch": 0.06,
"grad_norm": 4.1875,
"learning_rate": 1.6126315789473687e-05,
"loss": 0.6657,
"step": 2340
},
{
"epoch": 0.06,
"grad_norm": 2.921875,
"learning_rate": 1.6105263157894736e-05,
"loss": 0.6927,
"step": 2350
},
{
"epoch": 0.06,
"grad_norm": 5.46875,
"learning_rate": 1.6084210526315792e-05,
"loss": 0.6291,
"step": 2360
},
{
"epoch": 0.06,
"grad_norm": 2.640625,
"learning_rate": 1.606315789473684e-05,
"loss": 0.728,
"step": 2370
},
{
"epoch": 0.06,
"grad_norm": 6.4375,
"learning_rate": 1.6042105263157897e-05,
"loss": 0.621,
"step": 2380
},
{
"epoch": 0.06,
"grad_norm": 3.078125,
"learning_rate": 1.6021052631578947e-05,
"loss": 0.6176,
"step": 2390
},
{
"epoch": 0.06,
"grad_norm": 4.125,
"learning_rate": 1.6000000000000003e-05,
"loss": 0.6321,
"step": 2400
},
{
"epoch": 0.06,
"grad_norm": 3.25,
"learning_rate": 1.5978947368421055e-05,
"loss": 0.6174,
"step": 2410
},
{
"epoch": 0.06,
"grad_norm": 3.328125,
"learning_rate": 1.5957894736842105e-05,
"loss": 0.5024,
"step": 2420
},
{
"epoch": 0.06,
"grad_norm": 4.4375,
"learning_rate": 1.593684210526316e-05,
"loss": 0.5771,
"step": 2430
},
{
"epoch": 0.06,
"grad_norm": 4.09375,
"learning_rate": 1.591578947368421e-05,
"loss": 0.6005,
"step": 2440
},
{
"epoch": 0.06,
"grad_norm": 6.875,
"learning_rate": 1.5894736842105266e-05,
"loss": 0.6532,
"step": 2450
},
{
"epoch": 0.06,
"grad_norm": 4.21875,
"learning_rate": 1.5873684210526315e-05,
"loss": 0.7415,
"step": 2460
},
{
"epoch": 0.06,
"grad_norm": 3.84375,
"learning_rate": 1.585263157894737e-05,
"loss": 0.7431,
"step": 2470
},
{
"epoch": 0.06,
"grad_norm": 6.40625,
"learning_rate": 1.5831578947368424e-05,
"loss": 0.4914,
"step": 2480
},
{
"epoch": 0.06,
"grad_norm": 2.59375,
"learning_rate": 1.5810526315789473e-05,
"loss": 0.5853,
"step": 2490
},
{
"epoch": 0.06,
"grad_norm": 5.375,
"learning_rate": 1.578947368421053e-05,
"loss": 0.7048,
"step": 2500
},
{
"epoch": 0.06,
"eval_loss": 0.637090802192688,
"eval_runtime": 38.18,
"eval_samples_per_second": 26.192,
"eval_steps_per_second": 26.192,
"step": 2500
},
{
"epoch": 0.06,
"grad_norm": 5.25,
"learning_rate": 1.576842105263158e-05,
"loss": 0.6227,
"step": 2510
},
{
"epoch": 0.06,
"grad_norm": 3.796875,
"learning_rate": 1.5747368421052635e-05,
"loss": 0.5951,
"step": 2520
},
{
"epoch": 0.06,
"grad_norm": 4.96875,
"learning_rate": 1.5726315789473684e-05,
"loss": 0.6014,
"step": 2530
},
{
"epoch": 0.06,
"grad_norm": 5.90625,
"learning_rate": 1.570526315789474e-05,
"loss": 0.6057,
"step": 2540
},
{
"epoch": 0.06,
"grad_norm": 4.875,
"learning_rate": 1.568421052631579e-05,
"loss": 0.6255,
"step": 2550
},
{
"epoch": 0.06,
"grad_norm": 3.734375,
"learning_rate": 1.5663157894736842e-05,
"loss": 0.6768,
"step": 2560
},
{
"epoch": 0.06,
"grad_norm": 7.40625,
"learning_rate": 1.5642105263157898e-05,
"loss": 0.6953,
"step": 2570
},
{
"epoch": 0.06,
"grad_norm": 4.8125,
"learning_rate": 1.5621052631578947e-05,
"loss": 0.6471,
"step": 2580
},
{
"epoch": 0.06,
"grad_norm": 4.15625,
"learning_rate": 1.5600000000000003e-05,
"loss": 0.623,
"step": 2590
},
{
"epoch": 0.07,
"grad_norm": 3.34375,
"learning_rate": 1.5578947368421052e-05,
"loss": 0.6352,
"step": 2600
},
{
"epoch": 0.07,
"grad_norm": 3.46875,
"learning_rate": 1.555789473684211e-05,
"loss": 0.478,
"step": 2610
},
{
"epoch": 0.07,
"grad_norm": 2.140625,
"learning_rate": 1.5536842105263158e-05,
"loss": 0.553,
"step": 2620
},
{
"epoch": 0.07,
"grad_norm": 3.875,
"learning_rate": 1.551578947368421e-05,
"loss": 0.5354,
"step": 2630
},
{
"epoch": 0.07,
"grad_norm": 10.0,
"learning_rate": 1.5494736842105263e-05,
"loss": 0.6313,
"step": 2640
},
{
"epoch": 0.07,
"grad_norm": 4.09375,
"learning_rate": 1.5473684210526316e-05,
"loss": 0.6313,
"step": 2650
},
{
"epoch": 0.07,
"grad_norm": 4.84375,
"learning_rate": 1.545263157894737e-05,
"loss": 0.5761,
"step": 2660
},
{
"epoch": 0.07,
"grad_norm": 3.046875,
"learning_rate": 1.543157894736842e-05,
"loss": 0.6076,
"step": 2670
},
{
"epoch": 0.07,
"grad_norm": 4.875,
"learning_rate": 1.5410526315789477e-05,
"loss": 0.5827,
"step": 2680
},
{
"epoch": 0.07,
"grad_norm": 3.890625,
"learning_rate": 1.5389473684210526e-05,
"loss": 0.5284,
"step": 2690
},
{
"epoch": 0.07,
"grad_norm": 3.875,
"learning_rate": 1.536842105263158e-05,
"loss": 0.6868,
"step": 2700
},
{
"epoch": 0.07,
"grad_norm": 2.90625,
"learning_rate": 1.534736842105263e-05,
"loss": 0.6856,
"step": 2710
},
{
"epoch": 0.07,
"grad_norm": 2.8125,
"learning_rate": 1.5326315789473684e-05,
"loss": 0.6687,
"step": 2720
},
{
"epoch": 0.07,
"grad_norm": 4.34375,
"learning_rate": 1.530526315789474e-05,
"loss": 0.5599,
"step": 2730
},
{
"epoch": 0.07,
"grad_norm": 4.21875,
"learning_rate": 1.528421052631579e-05,
"loss": 0.5339,
"step": 2740
},
{
"epoch": 0.07,
"grad_norm": 8.1875,
"learning_rate": 1.5263157894736846e-05,
"loss": 0.5145,
"step": 2750
},
{
"epoch": 0.07,
"grad_norm": 2.75,
"learning_rate": 1.5242105263157897e-05,
"loss": 0.559,
"step": 2760
},
{
"epoch": 0.07,
"grad_norm": 3.796875,
"learning_rate": 1.5221052631578948e-05,
"loss": 0.6415,
"step": 2770
},
{
"epoch": 0.07,
"grad_norm": 4.8125,
"learning_rate": 1.5200000000000002e-05,
"loss": 0.6152,
"step": 2780
},
{
"epoch": 0.07,
"grad_norm": 2.71875,
"learning_rate": 1.5178947368421053e-05,
"loss": 0.5812,
"step": 2790
},
{
"epoch": 0.07,
"grad_norm": 5.0625,
"learning_rate": 1.5157894736842107e-05,
"loss": 0.5348,
"step": 2800
},
{
"epoch": 0.07,
"grad_norm": 3.546875,
"learning_rate": 1.5136842105263158e-05,
"loss": 0.5899,
"step": 2810
},
{
"epoch": 0.07,
"grad_norm": 3.578125,
"learning_rate": 1.5115789473684212e-05,
"loss": 0.5686,
"step": 2820
},
{
"epoch": 0.07,
"grad_norm": 3.59375,
"learning_rate": 1.5094736842105263e-05,
"loss": 0.4602,
"step": 2830
},
{
"epoch": 0.07,
"grad_norm": 5.4375,
"learning_rate": 1.5073684210526316e-05,
"loss": 0.6561,
"step": 2840
},
{
"epoch": 0.07,
"grad_norm": 8.75,
"learning_rate": 1.505263157894737e-05,
"loss": 0.5948,
"step": 2850
},
{
"epoch": 0.07,
"grad_norm": 4.1875,
"learning_rate": 1.5031578947368421e-05,
"loss": 0.6053,
"step": 2860
},
{
"epoch": 0.07,
"grad_norm": 4.0625,
"learning_rate": 1.5010526315789476e-05,
"loss": 0.5915,
"step": 2870
},
{
"epoch": 0.07,
"grad_norm": 3.671875,
"learning_rate": 1.4989473684210527e-05,
"loss": 0.6329,
"step": 2880
},
{
"epoch": 0.07,
"grad_norm": 3.421875,
"learning_rate": 1.4968421052631581e-05,
"loss": 0.6849,
"step": 2890
},
{
"epoch": 0.07,
"grad_norm": 11.0625,
"learning_rate": 1.4947368421052632e-05,
"loss": 0.6583,
"step": 2900
},
{
"epoch": 0.07,
"grad_norm": 3.25,
"learning_rate": 1.4926315789473686e-05,
"loss": 0.6508,
"step": 2910
},
{
"epoch": 0.07,
"grad_norm": 4.5,
"learning_rate": 1.4905263157894739e-05,
"loss": 0.5244,
"step": 2920
},
{
"epoch": 0.07,
"grad_norm": 4.03125,
"learning_rate": 1.488421052631579e-05,
"loss": 0.6146,
"step": 2930
},
{
"epoch": 0.07,
"grad_norm": 3.1875,
"learning_rate": 1.4863157894736844e-05,
"loss": 0.6245,
"step": 2940
},
{
"epoch": 0.07,
"grad_norm": 3.8125,
"learning_rate": 1.4842105263157895e-05,
"loss": 0.6862,
"step": 2950
},
{
"epoch": 0.07,
"grad_norm": 4.40625,
"learning_rate": 1.482105263157895e-05,
"loss": 0.5638,
"step": 2960
},
{
"epoch": 0.07,
"grad_norm": 3.0625,
"learning_rate": 1.48e-05,
"loss": 0.4932,
"step": 2970
},
{
"epoch": 0.07,
"grad_norm": 4.3125,
"learning_rate": 1.4778947368421055e-05,
"loss": 0.626,
"step": 2980
},
{
"epoch": 0.07,
"grad_norm": 5.125,
"learning_rate": 1.4757894736842106e-05,
"loss": 0.5843,
"step": 2990
},
{
"epoch": 0.07,
"grad_norm": 9.1875,
"learning_rate": 1.4736842105263159e-05,
"loss": 0.5773,
"step": 3000
},
{
"epoch": 0.07,
"eval_loss": 0.613276481628418,
"eval_runtime": 38.1471,
"eval_samples_per_second": 26.214,
"eval_steps_per_second": 26.214,
"step": 3000
},
{
"epoch": 0.08,
"grad_norm": 6.8125,
"learning_rate": 1.4715789473684213e-05,
"loss": 0.475,
"step": 3010
},
{
"epoch": 0.08,
"grad_norm": 3.21875,
"learning_rate": 1.4694736842105264e-05,
"loss": 0.592,
"step": 3020
},
{
"epoch": 0.08,
"grad_norm": 5.78125,
"learning_rate": 1.4673684210526318e-05,
"loss": 0.6149,
"step": 3030
},
{
"epoch": 0.08,
"grad_norm": 3.859375,
"learning_rate": 1.465263157894737e-05,
"loss": 0.5379,
"step": 3040
},
{
"epoch": 0.08,
"grad_norm": 3.109375,
"learning_rate": 1.4631578947368424e-05,
"loss": 0.5233,
"step": 3050
},
{
"epoch": 0.08,
"grad_norm": 4.375,
"learning_rate": 1.4610526315789474e-05,
"loss": 0.6428,
"step": 3060
},
{
"epoch": 0.08,
"grad_norm": 5.46875,
"learning_rate": 1.4589473684210527e-05,
"loss": 0.56,
"step": 3070
},
{
"epoch": 0.08,
"grad_norm": 13.0625,
"learning_rate": 1.456842105263158e-05,
"loss": 0.6643,
"step": 3080
},
{
"epoch": 0.08,
"grad_norm": 9.5625,
"learning_rate": 1.4547368421052632e-05,
"loss": 0.6606,
"step": 3090
},
{
"epoch": 0.08,
"grad_norm": 2.953125,
"learning_rate": 1.4526315789473687e-05,
"loss": 0.6734,
"step": 3100
},
{
"epoch": 0.08,
"grad_norm": 3.328125,
"learning_rate": 1.4505263157894738e-05,
"loss": 0.6948,
"step": 3110
},
{
"epoch": 0.08,
"grad_norm": 5.8125,
"learning_rate": 1.4484210526315792e-05,
"loss": 0.6735,
"step": 3120
},
{
"epoch": 0.08,
"grad_norm": 3.96875,
"learning_rate": 1.4463157894736843e-05,
"loss": 0.6107,
"step": 3130
},
{
"epoch": 0.08,
"grad_norm": 3.09375,
"learning_rate": 1.4442105263157896e-05,
"loss": 0.5645,
"step": 3140
},
{
"epoch": 0.08,
"grad_norm": 4.28125,
"learning_rate": 1.4421052631578948e-05,
"loss": 0.6415,
"step": 3150
},
{
"epoch": 0.08,
"grad_norm": 2.875,
"learning_rate": 1.4400000000000001e-05,
"loss": 0.4961,
"step": 3160
},
{
"epoch": 0.08,
"grad_norm": 3.484375,
"learning_rate": 1.4378947368421054e-05,
"loss": 0.609,
"step": 3170
},
{
"epoch": 0.08,
"grad_norm": 5.59375,
"learning_rate": 1.4357894736842106e-05,
"loss": 0.6139,
"step": 3180
},
{
"epoch": 0.08,
"grad_norm": 7.5,
"learning_rate": 1.433684210526316e-05,
"loss": 0.701,
"step": 3190
},
{
"epoch": 0.08,
"grad_norm": 4.4375,
"learning_rate": 1.4315789473684212e-05,
"loss": 0.5645,
"step": 3200
},
{
"epoch": 0.08,
"grad_norm": 3.0,
"learning_rate": 1.4294736842105263e-05,
"loss": 0.6149,
"step": 3210
},
{
"epoch": 0.08,
"grad_norm": 5.25,
"learning_rate": 1.4273684210526317e-05,
"loss": 0.6544,
"step": 3220
},
{
"epoch": 0.08,
"grad_norm": 3.21875,
"learning_rate": 1.425263157894737e-05,
"loss": 0.7131,
"step": 3230
},
{
"epoch": 0.08,
"grad_norm": 3.359375,
"learning_rate": 1.4231578947368422e-05,
"loss": 0.5974,
"step": 3240
},
{
"epoch": 0.08,
"grad_norm": 2.109375,
"learning_rate": 1.4210526315789475e-05,
"loss": 0.5418,
"step": 3250
},
{
"epoch": 0.08,
"grad_norm": 3.96875,
"learning_rate": 1.418947368421053e-05,
"loss": 0.5797,
"step": 3260
},
{
"epoch": 0.08,
"grad_norm": 2.5,
"learning_rate": 1.416842105263158e-05,
"loss": 0.5284,
"step": 3270
},
{
"epoch": 0.08,
"grad_norm": 4.21875,
"learning_rate": 1.4147368421052631e-05,
"loss": 0.6928,
"step": 3280
},
{
"epoch": 0.08,
"grad_norm": 5.90625,
"learning_rate": 1.4126315789473686e-05,
"loss": 0.5994,
"step": 3290
},
{
"epoch": 0.08,
"grad_norm": 7.09375,
"learning_rate": 1.4105263157894738e-05,
"loss": 0.4793,
"step": 3300
},
{
"epoch": 0.08,
"grad_norm": 3.984375,
"learning_rate": 1.4084210526315791e-05,
"loss": 0.6549,
"step": 3310
},
{
"epoch": 0.08,
"grad_norm": 2.078125,
"learning_rate": 1.4063157894736844e-05,
"loss": 0.6774,
"step": 3320
},
{
"epoch": 0.08,
"grad_norm": 18.5,
"learning_rate": 1.4042105263157896e-05,
"loss": 0.7274,
"step": 3330
},
{
"epoch": 0.08,
"grad_norm": 4.0625,
"learning_rate": 1.4021052631578949e-05,
"loss": 0.625,
"step": 3340
},
{
"epoch": 0.08,
"grad_norm": 4.5625,
"learning_rate": 1.4e-05,
"loss": 0.6704,
"step": 3350
},
{
"epoch": 0.08,
"grad_norm": 3.71875,
"learning_rate": 1.3978947368421054e-05,
"loss": 0.717,
"step": 3360
},
{
"epoch": 0.08,
"grad_norm": 5.53125,
"learning_rate": 1.3957894736842105e-05,
"loss": 0.5063,
"step": 3370
},
{
"epoch": 0.08,
"grad_norm": 4.875,
"learning_rate": 1.393684210526316e-05,
"loss": 0.5717,
"step": 3380
},
{
"epoch": 0.08,
"grad_norm": 4.03125,
"learning_rate": 1.3915789473684212e-05,
"loss": 0.5619,
"step": 3390
},
{
"epoch": 0.09,
"grad_norm": 5.75,
"learning_rate": 1.3894736842105265e-05,
"loss": 0.5526,
"step": 3400
},
{
"epoch": 0.09,
"grad_norm": 4.875,
"learning_rate": 1.3873684210526317e-05,
"loss": 0.5975,
"step": 3410
},
{
"epoch": 0.09,
"grad_norm": 3.890625,
"learning_rate": 1.3852631578947368e-05,
"loss": 0.6742,
"step": 3420
},
{
"epoch": 0.09,
"grad_norm": 4.4375,
"learning_rate": 1.3831578947368423e-05,
"loss": 0.5027,
"step": 3430
},
{
"epoch": 0.09,
"grad_norm": 6.8125,
"learning_rate": 1.3810526315789474e-05,
"loss": 0.6286,
"step": 3440
},
{
"epoch": 0.09,
"grad_norm": 5.1875,
"learning_rate": 1.3789473684210528e-05,
"loss": 0.5871,
"step": 3450
},
{
"epoch": 0.09,
"grad_norm": 8.0625,
"learning_rate": 1.3768421052631579e-05,
"loss": 0.5191,
"step": 3460
},
{
"epoch": 0.09,
"grad_norm": 8.4375,
"learning_rate": 1.3747368421052633e-05,
"loss": 0.6679,
"step": 3470
},
{
"epoch": 0.09,
"grad_norm": 42.5,
"learning_rate": 1.3726315789473686e-05,
"loss": 0.6627,
"step": 3480
},
{
"epoch": 0.09,
"grad_norm": 3.375,
"learning_rate": 1.3705263157894737e-05,
"loss": 0.5305,
"step": 3490
},
{
"epoch": 0.09,
"grad_norm": 3.6875,
"learning_rate": 1.3684210526315791e-05,
"loss": 0.6408,
"step": 3500
},
{
"epoch": 0.09,
"eval_loss": 0.6289076209068298,
"eval_runtime": 38.1752,
"eval_samples_per_second": 26.195,
"eval_steps_per_second": 26.195,
"step": 3500
},
{
"epoch": 0.09,
"grad_norm": 4.0625,
"learning_rate": 1.3663157894736842e-05,
"loss": 0.6025,
"step": 3510
},
{
"epoch": 0.09,
"grad_norm": 11.875,
"learning_rate": 1.3642105263157897e-05,
"loss": 0.6137,
"step": 3520
},
{
"epoch": 0.09,
"grad_norm": 22.25,
"learning_rate": 1.3621052631578948e-05,
"loss": 0.5813,
"step": 3530
},
{
"epoch": 0.09,
"grad_norm": 4.125,
"learning_rate": 1.3600000000000002e-05,
"loss": 0.4769,
"step": 3540
},
{
"epoch": 0.09,
"grad_norm": 7.03125,
"learning_rate": 1.3578947368421055e-05,
"loss": 0.7036,
"step": 3550
},
{
"epoch": 0.09,
"grad_norm": 4.5625,
"learning_rate": 1.3557894736842106e-05,
"loss": 0.6237,
"step": 3560
},
{
"epoch": 0.09,
"grad_norm": 2.796875,
"learning_rate": 1.353684210526316e-05,
"loss": 0.5743,
"step": 3570
},
{
"epoch": 0.09,
"grad_norm": 4.34375,
"learning_rate": 1.3515789473684211e-05,
"loss": 0.6427,
"step": 3580
},
{
"epoch": 0.09,
"grad_norm": 12.5,
"learning_rate": 1.3494736842105265e-05,
"loss": 0.5552,
"step": 3590
},
{
"epoch": 0.09,
"grad_norm": 4.0625,
"learning_rate": 1.3473684210526316e-05,
"loss": 0.6391,
"step": 3600
},
{
"epoch": 0.09,
"grad_norm": 6.125,
"learning_rate": 1.345263157894737e-05,
"loss": 0.5062,
"step": 3610
},
{
"epoch": 0.09,
"grad_norm": 4.375,
"learning_rate": 1.3431578947368421e-05,
"loss": 0.6708,
"step": 3620
},
{
"epoch": 0.09,
"grad_norm": 2.078125,
"learning_rate": 1.3410526315789474e-05,
"loss": 0.5664,
"step": 3630
},
{
"epoch": 0.09,
"grad_norm": 5.90625,
"learning_rate": 1.3389473684210528e-05,
"loss": 0.6353,
"step": 3640
},
{
"epoch": 0.09,
"grad_norm": 3.0,
"learning_rate": 1.336842105263158e-05,
"loss": 0.7542,
"step": 3650
},
{
"epoch": 0.09,
"grad_norm": 1.859375,
"learning_rate": 1.3347368421052634e-05,
"loss": 0.5818,
"step": 3660
},
{
"epoch": 0.09,
"grad_norm": 4.03125,
"learning_rate": 1.3326315789473685e-05,
"loss": 0.5839,
"step": 3670
},
{
"epoch": 0.09,
"grad_norm": 2.421875,
"learning_rate": 1.3305263157894739e-05,
"loss": 0.5676,
"step": 3680
},
{
"epoch": 0.09,
"grad_norm": 4.9375,
"learning_rate": 1.328421052631579e-05,
"loss": 0.6292,
"step": 3690
},
{
"epoch": 0.09,
"grad_norm": 5.34375,
"learning_rate": 1.3263157894736843e-05,
"loss": 0.5808,
"step": 3700
},
{
"epoch": 0.09,
"grad_norm": 4.125,
"learning_rate": 1.3242105263157895e-05,
"loss": 0.7451,
"step": 3710
},
{
"epoch": 0.09,
"grad_norm": 6.15625,
"learning_rate": 1.3221052631578948e-05,
"loss": 0.4822,
"step": 3720
},
{
"epoch": 0.09,
"grad_norm": 3.671875,
"learning_rate": 1.3200000000000002e-05,
"loss": 0.5966,
"step": 3730
},
{
"epoch": 0.09,
"grad_norm": 7.71875,
"learning_rate": 1.3178947368421053e-05,
"loss": 0.6074,
"step": 3740
},
{
"epoch": 0.09,
"grad_norm": 3.703125,
"learning_rate": 1.3157894736842108e-05,
"loss": 0.5465,
"step": 3750
},
{
"epoch": 0.09,
"grad_norm": 5.71875,
"learning_rate": 1.3136842105263159e-05,
"loss": 0.6226,
"step": 3760
},
{
"epoch": 0.09,
"grad_norm": 2.9375,
"learning_rate": 1.3115789473684211e-05,
"loss": 0.6031,
"step": 3770
},
{
"epoch": 0.09,
"grad_norm": 2.703125,
"learning_rate": 1.3094736842105264e-05,
"loss": 0.6693,
"step": 3780
},
{
"epoch": 0.09,
"grad_norm": 4.5,
"learning_rate": 1.3073684210526317e-05,
"loss": 0.6126,
"step": 3790
},
{
"epoch": 0.1,
"grad_norm": 3.96875,
"learning_rate": 1.305263157894737e-05,
"loss": 0.615,
"step": 3800
},
{
"epoch": 0.1,
"grad_norm": 4.78125,
"learning_rate": 1.3031578947368422e-05,
"loss": 0.6213,
"step": 3810
},
{
"epoch": 0.1,
"grad_norm": 7.9375,
"learning_rate": 1.3010526315789476e-05,
"loss": 0.5922,
"step": 3820
},
{
"epoch": 0.1,
"grad_norm": 3.453125,
"learning_rate": 1.2989473684210527e-05,
"loss": 0.5783,
"step": 3830
},
{
"epoch": 0.1,
"grad_norm": 5.03125,
"learning_rate": 1.2968421052631578e-05,
"loss": 0.5909,
"step": 3840
},
{
"epoch": 0.1,
"grad_norm": 3.1875,
"learning_rate": 1.2947368421052633e-05,
"loss": 0.6543,
"step": 3850
},
{
"epoch": 0.1,
"grad_norm": 3.625,
"learning_rate": 1.2926315789473685e-05,
"loss": 0.5652,
"step": 3860
},
{
"epoch": 0.1,
"grad_norm": 11.75,
"learning_rate": 1.2905263157894738e-05,
"loss": 0.6078,
"step": 3870
},
{
"epoch": 0.1,
"grad_norm": 3.828125,
"learning_rate": 1.288421052631579e-05,
"loss": 0.6246,
"step": 3880
},
{
"epoch": 0.1,
"grad_norm": 4.15625,
"learning_rate": 1.2863157894736845e-05,
"loss": 0.5988,
"step": 3890
},
{
"epoch": 0.1,
"grad_norm": 4.21875,
"learning_rate": 1.2842105263157896e-05,
"loss": 0.7113,
"step": 3900
},
{
"epoch": 0.1,
"grad_norm": 7.15625,
"learning_rate": 1.2821052631578947e-05,
"loss": 0.5497,
"step": 3910
},
{
"epoch": 0.1,
"grad_norm": 5.09375,
"learning_rate": 1.2800000000000001e-05,
"loss": 0.6188,
"step": 3920
},
{
"epoch": 0.1,
"grad_norm": 4.3125,
"learning_rate": 1.2778947368421054e-05,
"loss": 0.6139,
"step": 3930
},
{
"epoch": 0.1,
"grad_norm": 3.859375,
"learning_rate": 1.2757894736842106e-05,
"loss": 0.6298,
"step": 3940
},
{
"epoch": 0.1,
"grad_norm": 2.953125,
"learning_rate": 1.2736842105263159e-05,
"loss": 0.5542,
"step": 3950
},
{
"epoch": 0.1,
"grad_norm": 3.734375,
"learning_rate": 1.2715789473684212e-05,
"loss": 0.4915,
"step": 3960
},
{
"epoch": 0.1,
"grad_norm": 6.8125,
"learning_rate": 1.2694736842105264e-05,
"loss": 0.6733,
"step": 3970
},
{
"epoch": 0.1,
"grad_norm": 8.1875,
"learning_rate": 1.2673684210526315e-05,
"loss": 0.5441,
"step": 3980
},
{
"epoch": 0.1,
"grad_norm": 4.28125,
"learning_rate": 1.265263157894737e-05,
"loss": 0.6336,
"step": 3990
},
{
"epoch": 0.1,
"grad_norm": 4.09375,
"learning_rate": 1.263157894736842e-05,
"loss": 0.5396,
"step": 4000
},
{
"epoch": 0.1,
"eval_loss": 0.6335766315460205,
"eval_runtime": 38.1481,
"eval_samples_per_second": 26.214,
"eval_steps_per_second": 26.214,
"step": 4000
},
{
"epoch": 0.1,
"grad_norm": 3.015625,
"learning_rate": 1.2610526315789475e-05,
"loss": 0.5444,
"step": 4010
},
{
"epoch": 0.1,
"grad_norm": 4.8125,
"learning_rate": 1.2589473684210528e-05,
"loss": 0.5945,
"step": 4020
},
{
"epoch": 0.1,
"grad_norm": 3.90625,
"learning_rate": 1.256842105263158e-05,
"loss": 0.5489,
"step": 4030
},
{
"epoch": 0.1,
"grad_norm": 4.8125,
"learning_rate": 1.2547368421052633e-05,
"loss": 0.6619,
"step": 4040
},
{
"epoch": 0.1,
"grad_norm": 4.15625,
"learning_rate": 1.2526315789473684e-05,
"loss": 0.6392,
"step": 4050
},
{
"epoch": 0.1,
"grad_norm": 2.515625,
"learning_rate": 1.2505263157894738e-05,
"loss": 0.6275,
"step": 4060
},
{
"epoch": 0.1,
"grad_norm": 4.0,
"learning_rate": 1.248421052631579e-05,
"loss": 0.6209,
"step": 4070
},
{
"epoch": 0.1,
"grad_norm": 5.78125,
"learning_rate": 1.2463157894736844e-05,
"loss": 0.6417,
"step": 4080
},
{
"epoch": 0.1,
"grad_norm": 6.59375,
"learning_rate": 1.2442105263157895e-05,
"loss": 0.5294,
"step": 4090
},
{
"epoch": 0.1,
"grad_norm": 2.921875,
"learning_rate": 1.2421052631578949e-05,
"loss": 0.7045,
"step": 4100
},
{
"epoch": 0.1,
"grad_norm": 2.328125,
"learning_rate": 1.2400000000000002e-05,
"loss": 0.5133,
"step": 4110
},
{
"epoch": 0.1,
"grad_norm": 4.21875,
"learning_rate": 1.2378947368421053e-05,
"loss": 0.5812,
"step": 4120
},
{
"epoch": 0.1,
"grad_norm": 4.21875,
"learning_rate": 1.2357894736842107e-05,
"loss": 0.5628,
"step": 4130
},
{
"epoch": 0.1,
"grad_norm": 3.828125,
"learning_rate": 1.2336842105263158e-05,
"loss": 0.5895,
"step": 4140
},
{
"epoch": 0.1,
"grad_norm": 3.734375,
"learning_rate": 1.2315789473684212e-05,
"loss": 0.6542,
"step": 4150
},
{
"epoch": 0.1,
"grad_norm": 5.3125,
"learning_rate": 1.2294736842105263e-05,
"loss": 0.6245,
"step": 4160
},
{
"epoch": 0.1,
"grad_norm": 2.90625,
"learning_rate": 1.2273684210526317e-05,
"loss": 0.6157,
"step": 4170
},
{
"epoch": 0.1,
"grad_norm": 5.65625,
"learning_rate": 1.225263157894737e-05,
"loss": 0.602,
"step": 4180
},
{
"epoch": 0.1,
"grad_norm": 4.875,
"learning_rate": 1.2231578947368421e-05,
"loss": 0.6496,
"step": 4190
},
{
"epoch": 0.1,
"grad_norm": 2.34375,
"learning_rate": 1.2210526315789475e-05,
"loss": 0.6249,
"step": 4200
},
{
"epoch": 0.11,
"grad_norm": 4.3125,
"learning_rate": 1.2189473684210526e-05,
"loss": 0.6666,
"step": 4210
},
{
"epoch": 0.11,
"grad_norm": 4.6875,
"learning_rate": 1.216842105263158e-05,
"loss": 0.6132,
"step": 4220
},
{
"epoch": 0.11,
"grad_norm": 5.59375,
"learning_rate": 1.2147368421052632e-05,
"loss": 0.5799,
"step": 4230
},
{
"epoch": 0.11,
"grad_norm": 4.5625,
"learning_rate": 1.2126315789473686e-05,
"loss": 0.6745,
"step": 4240
},
{
"epoch": 0.11,
"grad_norm": 3.140625,
"learning_rate": 1.2105263157894737e-05,
"loss": 0.6618,
"step": 4250
},
{
"epoch": 0.11,
"grad_norm": 4.84375,
"learning_rate": 1.208421052631579e-05,
"loss": 0.5125,
"step": 4260
},
{
"epoch": 0.11,
"grad_norm": 2.75,
"learning_rate": 1.2063157894736844e-05,
"loss": 0.6907,
"step": 4270
},
{
"epoch": 0.11,
"grad_norm": 6.28125,
"learning_rate": 1.2042105263157895e-05,
"loss": 0.5876,
"step": 4280
},
{
"epoch": 0.11,
"grad_norm": 3.96875,
"learning_rate": 1.202105263157895e-05,
"loss": 0.5825,
"step": 4290
},
{
"epoch": 0.11,
"grad_norm": 2.421875,
"learning_rate": 1.2e-05,
"loss": 0.624,
"step": 4300
},
{
"epoch": 0.11,
"grad_norm": 3.0625,
"learning_rate": 1.1978947368421055e-05,
"loss": 0.6119,
"step": 4310
},
{
"epoch": 0.11,
"grad_norm": 5.8125,
"learning_rate": 1.1957894736842106e-05,
"loss": 0.6331,
"step": 4320
},
{
"epoch": 0.11,
"grad_norm": 4.59375,
"learning_rate": 1.1936842105263158e-05,
"loss": 0.5641,
"step": 4330
},
{
"epoch": 0.11,
"grad_norm": 3.578125,
"learning_rate": 1.1915789473684211e-05,
"loss": 0.4946,
"step": 4340
},
{
"epoch": 0.11,
"grad_norm": 4.84375,
"learning_rate": 1.1894736842105264e-05,
"loss": 0.5547,
"step": 4350
},
{
"epoch": 0.11,
"grad_norm": 18.5,
"learning_rate": 1.1873684210526318e-05,
"loss": 0.8141,
"step": 4360
},
{
"epoch": 0.11,
"grad_norm": 3.390625,
"learning_rate": 1.1852631578947369e-05,
"loss": 0.5756,
"step": 4370
},
{
"epoch": 0.11,
"grad_norm": 3.65625,
"learning_rate": 1.1831578947368423e-05,
"loss": 0.5353,
"step": 4380
},
{
"epoch": 0.11,
"grad_norm": 5.15625,
"learning_rate": 1.1810526315789474e-05,
"loss": 0.7419,
"step": 4390
},
{
"epoch": 0.11,
"grad_norm": 4.5,
"learning_rate": 1.1789473684210527e-05,
"loss": 0.72,
"step": 4400
},
{
"epoch": 0.11,
"grad_norm": 3.84375,
"learning_rate": 1.176842105263158e-05,
"loss": 0.6064,
"step": 4410
},
{
"epoch": 0.11,
"grad_norm": 3.46875,
"learning_rate": 1.1747368421052632e-05,
"loss": 0.5641,
"step": 4420
},
{
"epoch": 0.11,
"grad_norm": 9.125,
"learning_rate": 1.1726315789473685e-05,
"loss": 0.6017,
"step": 4430
},
{
"epoch": 0.11,
"grad_norm": 6.65625,
"learning_rate": 1.1705263157894737e-05,
"loss": 0.5863,
"step": 4440
},
{
"epoch": 0.11,
"grad_norm": 4.03125,
"learning_rate": 1.1684210526315792e-05,
"loss": 0.6267,
"step": 4450
},
{
"epoch": 0.11,
"grad_norm": 3.421875,
"learning_rate": 1.1663157894736843e-05,
"loss": 0.5873,
"step": 4460
},
{
"epoch": 0.11,
"grad_norm": 9.25,
"learning_rate": 1.1642105263157897e-05,
"loss": 0.5499,
"step": 4470
},
{
"epoch": 0.11,
"grad_norm": 4.09375,
"learning_rate": 1.1621052631578948e-05,
"loss": 0.621,
"step": 4480
},
{
"epoch": 0.11,
"grad_norm": 5.28125,
"learning_rate": 1.16e-05,
"loss": 0.5808,
"step": 4490
},
{
"epoch": 0.11,
"grad_norm": 3.46875,
"learning_rate": 1.1578947368421053e-05,
"loss": 0.7199,
"step": 4500
},
{
"epoch": 0.11,
"eval_loss": 0.6083159446716309,
"eval_runtime": 38.1826,
"eval_samples_per_second": 26.19,
"eval_steps_per_second": 26.19,
"step": 4500
},
{
"epoch": 0.11,
"grad_norm": 2.078125,
"learning_rate": 1.1557894736842106e-05,
"loss": 0.5804,
"step": 4510
},
{
"epoch": 0.11,
"grad_norm": 4.65625,
"learning_rate": 1.153684210526316e-05,
"loss": 0.7276,
"step": 4520
},
{
"epoch": 0.11,
"grad_norm": 2.90625,
"learning_rate": 1.1515789473684211e-05,
"loss": 0.6052,
"step": 4530
},
{
"epoch": 0.11,
"grad_norm": 14.6875,
"learning_rate": 1.1494736842105266e-05,
"loss": 0.6354,
"step": 4540
},
{
"epoch": 0.11,
"grad_norm": 5.03125,
"learning_rate": 1.1473684210526317e-05,
"loss": 0.7017,
"step": 4550
},
{
"epoch": 0.11,
"grad_norm": 23.875,
"learning_rate": 1.145263157894737e-05,
"loss": 0.6074,
"step": 4560
},
{
"epoch": 0.11,
"grad_norm": 5.5,
"learning_rate": 1.1431578947368422e-05,
"loss": 0.395,
"step": 4570
},
{
"epoch": 0.11,
"grad_norm": 7.0,
"learning_rate": 1.1410526315789475e-05,
"loss": 0.5674,
"step": 4580
},
{
"epoch": 0.11,
"grad_norm": 2.859375,
"learning_rate": 1.1389473684210527e-05,
"loss": 0.5567,
"step": 4590
},
{
"epoch": 0.12,
"grad_norm": 2.640625,
"learning_rate": 1.136842105263158e-05,
"loss": 0.607,
"step": 4600
},
{
"epoch": 0.12,
"grad_norm": 3.703125,
"learning_rate": 1.1347368421052634e-05,
"loss": 0.6487,
"step": 4610
},
{
"epoch": 0.12,
"grad_norm": 3.421875,
"learning_rate": 1.1326315789473685e-05,
"loss": 0.5341,
"step": 4620
},
{
"epoch": 0.12,
"grad_norm": 3.78125,
"learning_rate": 1.1305263157894736e-05,
"loss": 0.6164,
"step": 4630
},
{
"epoch": 0.12,
"grad_norm": 5.375,
"learning_rate": 1.128421052631579e-05,
"loss": 0.4827,
"step": 4640
},
{
"epoch": 0.12,
"grad_norm": 3.28125,
"learning_rate": 1.1263157894736843e-05,
"loss": 0.647,
"step": 4650
},
{
"epoch": 0.12,
"grad_norm": 7.9375,
"learning_rate": 1.1242105263157896e-05,
"loss": 0.5681,
"step": 4660
},
{
"epoch": 0.12,
"grad_norm": 3.890625,
"learning_rate": 1.1221052631578949e-05,
"loss": 0.5795,
"step": 4670
},
{
"epoch": 0.12,
"grad_norm": 4.0,
"learning_rate": 1.1200000000000001e-05,
"loss": 0.6383,
"step": 4680
},
{
"epoch": 0.12,
"grad_norm": 4.84375,
"learning_rate": 1.1178947368421054e-05,
"loss": 0.5925,
"step": 4690
},
{
"epoch": 0.12,
"grad_norm": 10.625,
"learning_rate": 1.1157894736842105e-05,
"loss": 0.5283,
"step": 4700
},
{
"epoch": 0.12,
"grad_norm": 2.671875,
"learning_rate": 1.1136842105263159e-05,
"loss": 0.7346,
"step": 4710
},
{
"epoch": 0.12,
"grad_norm": 2.734375,
"learning_rate": 1.111578947368421e-05,
"loss": 0.6452,
"step": 4720
},
{
"epoch": 0.12,
"grad_norm": 3.546875,
"learning_rate": 1.1094736842105264e-05,
"loss": 0.6553,
"step": 4730
},
{
"epoch": 0.12,
"grad_norm": 3.046875,
"learning_rate": 1.1073684210526317e-05,
"loss": 0.5828,
"step": 4740
},
{
"epoch": 0.12,
"grad_norm": 4.6875,
"learning_rate": 1.105263157894737e-05,
"loss": 0.5143,
"step": 4750
},
{
"epoch": 0.12,
"grad_norm": 4.21875,
"learning_rate": 1.1031578947368422e-05,
"loss": 0.6772,
"step": 4760
},
{
"epoch": 0.12,
"grad_norm": 3.421875,
"learning_rate": 1.1010526315789473e-05,
"loss": 0.6578,
"step": 4770
},
{
"epoch": 0.12,
"grad_norm": 4.1875,
"learning_rate": 1.0989473684210528e-05,
"loss": 0.5947,
"step": 4780
},
{
"epoch": 0.12,
"grad_norm": 4.9375,
"learning_rate": 1.0968421052631579e-05,
"loss": 0.6039,
"step": 4790
},
{
"epoch": 0.12,
"grad_norm": 2.34375,
"learning_rate": 1.0947368421052633e-05,
"loss": 0.6658,
"step": 4800
},
{
"epoch": 0.12,
"grad_norm": 8.0625,
"learning_rate": 1.0926315789473686e-05,
"loss": 0.4581,
"step": 4810
},
{
"epoch": 0.12,
"grad_norm": 4.84375,
"learning_rate": 1.0905263157894738e-05,
"loss": 0.5935,
"step": 4820
},
{
"epoch": 0.12,
"grad_norm": 4.25,
"learning_rate": 1.0884210526315791e-05,
"loss": 0.6028,
"step": 4830
},
{
"epoch": 0.12,
"grad_norm": 3.859375,
"learning_rate": 1.0863157894736842e-05,
"loss": 0.6113,
"step": 4840
},
{
"epoch": 0.12,
"grad_norm": 21.5,
"learning_rate": 1.0842105263157896e-05,
"loss": 0.575,
"step": 4850
},
{
"epoch": 0.12,
"grad_norm": 6.0,
"learning_rate": 1.0821052631578947e-05,
"loss": 0.5985,
"step": 4860
},
{
"epoch": 0.12,
"grad_norm": 3.140625,
"learning_rate": 1.0800000000000002e-05,
"loss": 0.643,
"step": 4870
},
{
"epoch": 0.12,
"grad_norm": 10.6875,
"learning_rate": 1.0778947368421053e-05,
"loss": 0.6964,
"step": 4880
},
{
"epoch": 0.12,
"grad_norm": 5.71875,
"learning_rate": 1.0757894736842107e-05,
"loss": 0.655,
"step": 4890
},
{
"epoch": 0.12,
"grad_norm": 5.90625,
"learning_rate": 1.073684210526316e-05,
"loss": 0.6014,
"step": 4900
},
{
"epoch": 0.12,
"grad_norm": 4.875,
"learning_rate": 1.071578947368421e-05,
"loss": 0.6139,
"step": 4910
},
{
"epoch": 0.12,
"grad_norm": 3.890625,
"learning_rate": 1.0694736842105265e-05,
"loss": 0.5815,
"step": 4920
},
{
"epoch": 0.12,
"grad_norm": 4.0,
"learning_rate": 1.0673684210526316e-05,
"loss": 0.6578,
"step": 4930
},
{
"epoch": 0.12,
"grad_norm": 3.59375,
"learning_rate": 1.065263157894737e-05,
"loss": 0.5326,
"step": 4940
},
{
"epoch": 0.12,
"grad_norm": 4.0,
"learning_rate": 1.0631578947368421e-05,
"loss": 0.5406,
"step": 4950
},
{
"epoch": 0.12,
"grad_norm": 4.84375,
"learning_rate": 1.0610526315789476e-05,
"loss": 0.6739,
"step": 4960
},
{
"epoch": 0.12,
"grad_norm": 3.8125,
"learning_rate": 1.0589473684210526e-05,
"loss": 0.5474,
"step": 4970
},
{
"epoch": 0.12,
"grad_norm": 3.578125,
"learning_rate": 1.0568421052631579e-05,
"loss": 0.5629,
"step": 4980
},
{
"epoch": 0.12,
"grad_norm": 3.1875,
"learning_rate": 1.0547368421052633e-05,
"loss": 0.7017,
"step": 4990
},
{
"epoch": 0.12,
"grad_norm": 3.984375,
"learning_rate": 1.0526315789473684e-05,
"loss": 0.4835,
"step": 5000
},
{
"epoch": 0.12,
"eval_loss": 0.6027244925498962,
"eval_runtime": 38.1439,
"eval_samples_per_second": 26.216,
"eval_steps_per_second": 26.216,
"step": 5000
},
{
"epoch": 0.13,
"grad_norm": 4.53125,
"learning_rate": 1.0505263157894739e-05,
"loss": 0.7573,
"step": 5010
},
{
"epoch": 0.13,
"grad_norm": 1.8515625,
"learning_rate": 1.048421052631579e-05,
"loss": 0.5092,
"step": 5020
},
{
"epoch": 0.13,
"grad_norm": 14.9375,
"learning_rate": 1.0463157894736844e-05,
"loss": 0.642,
"step": 5030
},
{
"epoch": 0.13,
"grad_norm": 4.90625,
"learning_rate": 1.0442105263157895e-05,
"loss": 0.6185,
"step": 5040
},
{
"epoch": 0.13,
"grad_norm": 4.5625,
"learning_rate": 1.0421052631578948e-05,
"loss": 0.6183,
"step": 5050
},
{
"epoch": 0.13,
"grad_norm": 6.125,
"learning_rate": 1.04e-05,
"loss": 0.5496,
"step": 5060
},
{
"epoch": 0.13,
"grad_norm": 4.5625,
"learning_rate": 1.0378947368421053e-05,
"loss": 0.5861,
"step": 5070
},
{
"epoch": 0.13,
"grad_norm": 3.515625,
"learning_rate": 1.0357894736842107e-05,
"loss": 0.6365,
"step": 5080
},
{
"epoch": 0.13,
"grad_norm": 4.75,
"learning_rate": 1.0336842105263158e-05,
"loss": 0.6688,
"step": 5090
},
{
"epoch": 0.13,
"grad_norm": 2.703125,
"learning_rate": 1.0315789473684213e-05,
"loss": 0.5361,
"step": 5100
},
{
"epoch": 0.13,
"grad_norm": 4.125,
"learning_rate": 1.0294736842105264e-05,
"loss": 0.6006,
"step": 5110
},
{
"epoch": 0.13,
"grad_norm": 3.890625,
"learning_rate": 1.0273684210526316e-05,
"loss": 0.6353,
"step": 5120
},
{
"epoch": 0.13,
"grad_norm": 6.25,
"learning_rate": 1.0252631578947369e-05,
"loss": 0.6475,
"step": 5130
},
{
"epoch": 0.13,
"grad_norm": 4.625,
"learning_rate": 1.0231578947368422e-05,
"loss": 0.498,
"step": 5140
},
{
"epoch": 0.13,
"grad_norm": 4.75,
"learning_rate": 1.0210526315789476e-05,
"loss": 0.6467,
"step": 5150
},
{
"epoch": 0.13,
"grad_norm": 4.1875,
"learning_rate": 1.0189473684210527e-05,
"loss": 0.611,
"step": 5160
},
{
"epoch": 0.13,
"grad_norm": 9.75,
"learning_rate": 1.0168421052631581e-05,
"loss": 0.5392,
"step": 5170
},
{
"epoch": 0.13,
"grad_norm": 13.25,
"learning_rate": 1.0147368421052632e-05,
"loss": 0.645,
"step": 5180
},
{
"epoch": 0.13,
"grad_norm": 4.125,
"learning_rate": 1.0126315789473685e-05,
"loss": 0.6287,
"step": 5190
},
{
"epoch": 0.13,
"grad_norm": 4.21875,
"learning_rate": 1.0105263157894738e-05,
"loss": 0.5719,
"step": 5200
},
{
"epoch": 0.13,
"grad_norm": 2.671875,
"learning_rate": 1.008421052631579e-05,
"loss": 0.621,
"step": 5210
},
{
"epoch": 0.13,
"grad_norm": 2.984375,
"learning_rate": 1.0063157894736843e-05,
"loss": 0.5169,
"step": 5220
},
{
"epoch": 0.13,
"grad_norm": 8.375,
"learning_rate": 1.0042105263157896e-05,
"loss": 0.6793,
"step": 5230
},
{
"epoch": 0.13,
"grad_norm": 117.0,
"learning_rate": 1.002105263157895e-05,
"loss": 0.6005,
"step": 5240
},
{
"epoch": 0.13,
"grad_norm": 3.5625,
"learning_rate": 1e-05,
"loss": 0.6674,
"step": 5250
},
{
"epoch": 0.13,
"grad_norm": 3.9375,
"learning_rate": 9.978947368421053e-06,
"loss": 0.6105,
"step": 5260
},
{
"epoch": 0.13,
"grad_norm": 4.75,
"learning_rate": 9.957894736842106e-06,
"loss": 0.5876,
"step": 5270
},
{
"epoch": 0.13,
"grad_norm": 3.203125,
"learning_rate": 9.936842105263159e-06,
"loss": 0.5551,
"step": 5280
},
{
"epoch": 0.13,
"grad_norm": 4.84375,
"learning_rate": 9.915789473684211e-06,
"loss": 0.5699,
"step": 5290
},
{
"epoch": 0.13,
"grad_norm": 3.5625,
"learning_rate": 9.894736842105264e-06,
"loss": 0.5652,
"step": 5300
},
{
"epoch": 0.13,
"grad_norm": 7.59375,
"learning_rate": 9.873684210526317e-06,
"loss": 0.6494,
"step": 5310
},
{
"epoch": 0.13,
"grad_norm": 6.03125,
"learning_rate": 9.85263157894737e-06,
"loss": 0.6,
"step": 5320
},
{
"epoch": 0.13,
"grad_norm": 3.3125,
"learning_rate": 9.831578947368422e-06,
"loss": 0.603,
"step": 5330
},
{
"epoch": 0.13,
"grad_norm": 3.546875,
"learning_rate": 9.810526315789475e-06,
"loss": 0.6117,
"step": 5340
},
{
"epoch": 0.13,
"grad_norm": 4.625,
"learning_rate": 9.789473684210527e-06,
"loss": 0.5883,
"step": 5350
},
{
"epoch": 0.13,
"grad_norm": 3.234375,
"learning_rate": 9.76842105263158e-06,
"loss": 0.5722,
"step": 5360
},
{
"epoch": 0.13,
"grad_norm": 3.171875,
"learning_rate": 9.747368421052633e-06,
"loss": 0.5692,
"step": 5370
},
{
"epoch": 0.13,
"grad_norm": 3.515625,
"learning_rate": 9.726315789473685e-06,
"loss": 0.7724,
"step": 5380
},
{
"epoch": 0.13,
"grad_norm": 4.25,
"learning_rate": 9.705263157894738e-06,
"loss": 0.6017,
"step": 5390
},
{
"epoch": 0.14,
"grad_norm": 6.53125,
"learning_rate": 9.68421052631579e-06,
"loss": 0.6049,
"step": 5400
},
{
"epoch": 0.14,
"grad_norm": 4.34375,
"learning_rate": 9.663157894736843e-06,
"loss": 0.5158,
"step": 5410
},
{
"epoch": 0.14,
"grad_norm": 5.34375,
"learning_rate": 9.642105263157896e-06,
"loss": 0.6487,
"step": 5420
},
{
"epoch": 0.14,
"grad_norm": 5.71875,
"learning_rate": 9.621052631578947e-06,
"loss": 0.5599,
"step": 5430
},
{
"epoch": 0.14,
"grad_norm": 5.34375,
"learning_rate": 9.600000000000001e-06,
"loss": 0.5035,
"step": 5440
},
{
"epoch": 0.14,
"grad_norm": 4.4375,
"learning_rate": 9.578947368421054e-06,
"loss": 0.5768,
"step": 5450
},
{
"epoch": 0.14,
"grad_norm": 48.25,
"learning_rate": 9.557894736842107e-06,
"loss": 0.5815,
"step": 5460
},
{
"epoch": 0.14,
"grad_norm": 3.40625,
"learning_rate": 9.53684210526316e-06,
"loss": 0.6133,
"step": 5470
},
{
"epoch": 0.14,
"grad_norm": 2.40625,
"learning_rate": 9.515789473684212e-06,
"loss": 0.5997,
"step": 5480
},
{
"epoch": 0.14,
"grad_norm": 3.578125,
"learning_rate": 9.494736842105265e-06,
"loss": 0.5689,
"step": 5490
},
{
"epoch": 0.14,
"grad_norm": 7.0625,
"learning_rate": 9.473684210526315e-06,
"loss": 0.6511,
"step": 5500
},
{
"epoch": 0.14,
"eval_loss": 0.6153059005737305,
"eval_runtime": 38.2125,
"eval_samples_per_second": 26.169,
"eval_steps_per_second": 26.169,
"step": 5500
},
{
"epoch": 0.14,
"grad_norm": 3.1875,
"learning_rate": 9.452631578947368e-06,
"loss": 0.5449,
"step": 5510
},
{
"epoch": 0.14,
"grad_norm": 4.40625,
"learning_rate": 9.43157894736842e-06,
"loss": 0.6634,
"step": 5520
},
{
"epoch": 0.14,
"grad_norm": 7.875,
"learning_rate": 9.410526315789475e-06,
"loss": 0.6222,
"step": 5530
},
{
"epoch": 0.14,
"grad_norm": 5.4375,
"learning_rate": 9.389473684210528e-06,
"loss": 0.5707,
"step": 5540
},
{
"epoch": 0.14,
"grad_norm": 3.828125,
"learning_rate": 9.36842105263158e-06,
"loss": 0.4959,
"step": 5550
},
{
"epoch": 0.14,
"grad_norm": 6.0625,
"learning_rate": 9.347368421052633e-06,
"loss": 0.5941,
"step": 5560
},
{
"epoch": 0.14,
"grad_norm": 1.6015625,
"learning_rate": 9.326315789473684e-06,
"loss": 0.6101,
"step": 5570
},
{
"epoch": 0.14,
"grad_norm": 4.4375,
"learning_rate": 9.305263157894737e-06,
"loss": 0.5916,
"step": 5580
},
{
"epoch": 0.14,
"grad_norm": 4.34375,
"learning_rate": 9.28421052631579e-06,
"loss": 0.5927,
"step": 5590
},
{
"epoch": 0.14,
"grad_norm": 5.90625,
"learning_rate": 9.263157894736842e-06,
"loss": 0.5575,
"step": 5600
},
{
"epoch": 0.14,
"grad_norm": 8.5,
"learning_rate": 9.242105263157896e-06,
"loss": 0.5709,
"step": 5610
},
{
"epoch": 0.14,
"grad_norm": 3.5,
"learning_rate": 9.221052631578949e-06,
"loss": 0.6237,
"step": 5620
},
{
"epoch": 0.14,
"grad_norm": 5.25,
"learning_rate": 9.200000000000002e-06,
"loss": 0.6183,
"step": 5630
},
{
"epoch": 0.14,
"grad_norm": 6.3125,
"learning_rate": 9.178947368421053e-06,
"loss": 0.579,
"step": 5640
},
{
"epoch": 0.14,
"grad_norm": 4.8125,
"learning_rate": 9.157894736842105e-06,
"loss": 0.5801,
"step": 5650
},
{
"epoch": 0.14,
"grad_norm": 3.71875,
"learning_rate": 9.136842105263158e-06,
"loss": 0.5568,
"step": 5660
},
{
"epoch": 0.14,
"grad_norm": 2.84375,
"learning_rate": 9.11578947368421e-06,
"loss": 0.6277,
"step": 5670
},
{
"epoch": 0.14,
"grad_norm": 3.125,
"learning_rate": 9.094736842105263e-06,
"loss": 0.5189,
"step": 5680
},
{
"epoch": 0.14,
"grad_norm": 2.3125,
"learning_rate": 9.073684210526316e-06,
"loss": 0.6304,
"step": 5690
},
{
"epoch": 0.14,
"grad_norm": 2.359375,
"learning_rate": 9.05263157894737e-06,
"loss": 0.5956,
"step": 5700
},
{
"epoch": 0.14,
"grad_norm": 11.9375,
"learning_rate": 9.031578947368423e-06,
"loss": 0.6483,
"step": 5710
},
{
"epoch": 0.14,
"grad_norm": 5.09375,
"learning_rate": 9.010526315789474e-06,
"loss": 0.601,
"step": 5720
},
{
"epoch": 0.14,
"grad_norm": 3.5,
"learning_rate": 8.989473684210527e-06,
"loss": 0.5887,
"step": 5730
},
{
"epoch": 0.14,
"grad_norm": 3.09375,
"learning_rate": 8.96842105263158e-06,
"loss": 0.63,
"step": 5740
},
{
"epoch": 0.14,
"grad_norm": 4.4375,
"learning_rate": 8.947368421052632e-06,
"loss": 0.5988,
"step": 5750
},
{
"epoch": 0.14,
"grad_norm": 3.265625,
"learning_rate": 8.926315789473685e-06,
"loss": 0.5692,
"step": 5760
},
{
"epoch": 0.14,
"grad_norm": 3.015625,
"learning_rate": 8.905263157894737e-06,
"loss": 0.5917,
"step": 5770
},
{
"epoch": 0.14,
"grad_norm": 4.40625,
"learning_rate": 8.884210526315792e-06,
"loss": 0.565,
"step": 5780
},
{
"epoch": 0.14,
"grad_norm": 4.15625,
"learning_rate": 8.863157894736842e-06,
"loss": 0.5388,
"step": 5790
},
{
"epoch": 0.14,
"grad_norm": 7.875,
"learning_rate": 8.842105263157895e-06,
"loss": 0.6622,
"step": 5800
},
{
"epoch": 0.15,
"grad_norm": 6.25,
"learning_rate": 8.821052631578948e-06,
"loss": 0.6043,
"step": 5810
},
{
"epoch": 0.15,
"grad_norm": 3.953125,
"learning_rate": 8.8e-06,
"loss": 0.6431,
"step": 5820
},
{
"epoch": 0.15,
"grad_norm": 2.421875,
"learning_rate": 8.778947368421053e-06,
"loss": 0.563,
"step": 5830
},
{
"epoch": 0.15,
"grad_norm": 7.4375,
"learning_rate": 8.757894736842106e-06,
"loss": 0.6342,
"step": 5840
},
{
"epoch": 0.15,
"grad_norm": 4.46875,
"learning_rate": 8.736842105263158e-06,
"loss": 0.6206,
"step": 5850
},
{
"epoch": 0.15,
"grad_norm": 2.828125,
"learning_rate": 8.715789473684211e-06,
"loss": 0.5046,
"step": 5860
},
{
"epoch": 0.15,
"grad_norm": 13.75,
"learning_rate": 8.694736842105264e-06,
"loss": 0.5405,
"step": 5870
},
{
"epoch": 0.15,
"grad_norm": 4.6875,
"learning_rate": 8.673684210526316e-06,
"loss": 0.6021,
"step": 5880
},
{
"epoch": 0.15,
"grad_norm": 4.0,
"learning_rate": 8.652631578947369e-06,
"loss": 0.7437,
"step": 5890
},
{
"epoch": 0.15,
"grad_norm": 3.4375,
"learning_rate": 8.631578947368422e-06,
"loss": 0.6618,
"step": 5900
},
{
"epoch": 0.15,
"grad_norm": 3.34375,
"learning_rate": 8.610526315789474e-06,
"loss": 0.6088,
"step": 5910
},
{
"epoch": 0.15,
"grad_norm": 4.90625,
"learning_rate": 8.589473684210527e-06,
"loss": 0.5808,
"step": 5920
},
{
"epoch": 0.15,
"grad_norm": 4.25,
"learning_rate": 8.56842105263158e-06,
"loss": 0.6744,
"step": 5930
},
{
"epoch": 0.15,
"grad_norm": 4.4375,
"learning_rate": 8.547368421052632e-06,
"loss": 0.5432,
"step": 5940
},
{
"epoch": 0.15,
"grad_norm": 6.65625,
"learning_rate": 8.526315789473685e-06,
"loss": 0.6601,
"step": 5950
},
{
"epoch": 0.15,
"grad_norm": 4.3125,
"learning_rate": 8.505263157894738e-06,
"loss": 0.4648,
"step": 5960
},
{
"epoch": 0.15,
"grad_norm": 2.953125,
"learning_rate": 8.48421052631579e-06,
"loss": 0.5452,
"step": 5970
},
{
"epoch": 0.15,
"grad_norm": 5.09375,
"learning_rate": 8.463157894736843e-06,
"loss": 0.5761,
"step": 5980
},
{
"epoch": 0.15,
"grad_norm": 2.828125,
"learning_rate": 8.442105263157896e-06,
"loss": 0.5394,
"step": 5990
},
{
"epoch": 0.15,
"grad_norm": 5.65625,
"learning_rate": 8.421052631578948e-06,
"loss": 0.6339,
"step": 6000
},
{
"epoch": 0.15,
"eval_loss": 0.6167545318603516,
"eval_runtime": 38.2022,
"eval_samples_per_second": 26.177,
"eval_steps_per_second": 26.177,
"step": 6000
},
{
"epoch": 0.15,
"grad_norm": 3.609375,
"learning_rate": 8.400000000000001e-06,
"loss": 0.5043,
"step": 6010
},
{
"epoch": 0.15,
"grad_norm": 3.171875,
"learning_rate": 8.378947368421054e-06,
"loss": 0.6012,
"step": 6020
},
{
"epoch": 0.15,
"grad_norm": 7.03125,
"learning_rate": 8.357894736842106e-06,
"loss": 0.5963,
"step": 6030
},
{
"epoch": 0.15,
"grad_norm": 5.9375,
"learning_rate": 8.336842105263159e-06,
"loss": 0.6281,
"step": 6040
},
{
"epoch": 0.15,
"grad_norm": 2.59375,
"learning_rate": 8.315789473684212e-06,
"loss": 0.4643,
"step": 6050
},
{
"epoch": 0.15,
"grad_norm": 7.4375,
"learning_rate": 8.294736842105264e-06,
"loss": 0.6371,
"step": 6060
},
{
"epoch": 0.15,
"grad_norm": 2.421875,
"learning_rate": 8.273684210526317e-06,
"loss": 0.5769,
"step": 6070
},
{
"epoch": 0.15,
"grad_norm": 3.703125,
"learning_rate": 8.25263157894737e-06,
"loss": 0.7226,
"step": 6080
},
{
"epoch": 0.15,
"grad_norm": 4.375,
"learning_rate": 8.231578947368422e-06,
"loss": 0.541,
"step": 6090
},
{
"epoch": 0.15,
"grad_norm": 4.46875,
"learning_rate": 8.210526315789475e-06,
"loss": 0.5965,
"step": 6100
},
{
"epoch": 0.15,
"grad_norm": 6.28125,
"learning_rate": 8.189473684210527e-06,
"loss": 0.5999,
"step": 6110
},
{
"epoch": 0.15,
"grad_norm": 6.34375,
"learning_rate": 8.16842105263158e-06,
"loss": 0.6388,
"step": 6120
},
{
"epoch": 0.15,
"grad_norm": 5.6875,
"learning_rate": 8.147368421052633e-06,
"loss": 0.5696,
"step": 6130
},
{
"epoch": 0.15,
"grad_norm": 2.4375,
"learning_rate": 8.126315789473684e-06,
"loss": 0.4886,
"step": 6140
},
{
"epoch": 0.15,
"grad_norm": 5.59375,
"learning_rate": 8.105263157894736e-06,
"loss": 0.5406,
"step": 6150
},
{
"epoch": 0.15,
"grad_norm": 6.9375,
"learning_rate": 8.08421052631579e-06,
"loss": 0.6134,
"step": 6160
},
{
"epoch": 0.15,
"grad_norm": 4.875,
"learning_rate": 8.063157894736843e-06,
"loss": 0.6944,
"step": 6170
},
{
"epoch": 0.15,
"grad_norm": 7.4375,
"learning_rate": 8.042105263157896e-06,
"loss": 0.5848,
"step": 6180
},
{
"epoch": 0.15,
"grad_norm": 3.40625,
"learning_rate": 8.021052631578949e-06,
"loss": 0.5441,
"step": 6190
},
{
"epoch": 0.15,
"grad_norm": 3.6875,
"learning_rate": 8.000000000000001e-06,
"loss": 0.678,
"step": 6200
},
{
"epoch": 0.16,
"grad_norm": 2.625,
"learning_rate": 7.978947368421052e-06,
"loss": 0.639,
"step": 6210
},
{
"epoch": 0.16,
"grad_norm": 6.09375,
"learning_rate": 7.957894736842105e-06,
"loss": 0.6824,
"step": 6220
},
{
"epoch": 0.16,
"grad_norm": 4.4375,
"learning_rate": 7.936842105263158e-06,
"loss": 0.6219,
"step": 6230
},
{
"epoch": 0.16,
"grad_norm": 3.796875,
"learning_rate": 7.915789473684212e-06,
"loss": 0.6267,
"step": 6240
},
{
"epoch": 0.16,
"grad_norm": 3.453125,
"learning_rate": 7.894736842105265e-06,
"loss": 0.5348,
"step": 6250
},
{
"epoch": 0.16,
"grad_norm": 5.90625,
"learning_rate": 7.873684210526317e-06,
"loss": 0.6523,
"step": 6260
},
{
"epoch": 0.16,
"grad_norm": 8.25,
"learning_rate": 7.85263157894737e-06,
"loss": 0.6024,
"step": 6270
},
{
"epoch": 0.16,
"grad_norm": 2.875,
"learning_rate": 7.831578947368421e-06,
"loss": 0.5896,
"step": 6280
},
{
"epoch": 0.16,
"grad_norm": 7.40625,
"learning_rate": 7.810526315789474e-06,
"loss": 0.7023,
"step": 6290
},
{
"epoch": 0.16,
"grad_norm": 4.09375,
"learning_rate": 7.789473684210526e-06,
"loss": 0.5793,
"step": 6300
},
{
"epoch": 0.16,
"grad_norm": 6.375,
"learning_rate": 7.768421052631579e-06,
"loss": 0.6332,
"step": 6310
},
{
"epoch": 0.16,
"grad_norm": 5.3125,
"learning_rate": 7.747368421052631e-06,
"loss": 0.5124,
"step": 6320
},
{
"epoch": 0.16,
"grad_norm": 2.203125,
"learning_rate": 7.726315789473686e-06,
"loss": 0.6329,
"step": 6330
},
{
"epoch": 0.16,
"grad_norm": 9.0,
"learning_rate": 7.705263157894738e-06,
"loss": 0.7025,
"step": 6340
},
{
"epoch": 0.16,
"grad_norm": 7.09375,
"learning_rate": 7.68421052631579e-06,
"loss": 0.5197,
"step": 6350
},
{
"epoch": 0.16,
"grad_norm": 3.78125,
"learning_rate": 7.663157894736842e-06,
"loss": 0.6173,
"step": 6360
},
{
"epoch": 0.16,
"grad_norm": 2.578125,
"learning_rate": 7.642105263157895e-06,
"loss": 0.5852,
"step": 6370
},
{
"epoch": 0.16,
"grad_norm": 3.109375,
"learning_rate": 7.621052631578948e-06,
"loss": 0.5688,
"step": 6380
},
{
"epoch": 0.16,
"grad_norm": 9.0625,
"learning_rate": 7.600000000000001e-06,
"loss": 0.5457,
"step": 6390
},
{
"epoch": 0.16,
"grad_norm": 1.9296875,
"learning_rate": 7.578947368421054e-06,
"loss": 0.6564,
"step": 6400
},
{
"epoch": 0.16,
"grad_norm": 2.15625,
"learning_rate": 7.557894736842106e-06,
"loss": 0.5177,
"step": 6410
},
{
"epoch": 0.16,
"grad_norm": 3.6875,
"learning_rate": 7.536842105263158e-06,
"loss": 0.6287,
"step": 6420
},
{
"epoch": 0.16,
"grad_norm": 3.578125,
"learning_rate": 7.515789473684211e-06,
"loss": 0.6026,
"step": 6430
},
{
"epoch": 0.16,
"grad_norm": 2.8125,
"learning_rate": 7.494736842105263e-06,
"loss": 0.5795,
"step": 6440
},
{
"epoch": 0.16,
"grad_norm": 3.09375,
"learning_rate": 7.473684210526316e-06,
"loss": 0.6848,
"step": 6450
},
{
"epoch": 0.16,
"grad_norm": 3.421875,
"learning_rate": 7.4526315789473695e-06,
"loss": 0.4139,
"step": 6460
},
{
"epoch": 0.16,
"grad_norm": 5.0,
"learning_rate": 7.431578947368422e-06,
"loss": 0.5696,
"step": 6470
},
{
"epoch": 0.16,
"grad_norm": 3.953125,
"learning_rate": 7.410526315789475e-06,
"loss": 0.6651,
"step": 6480
},
{
"epoch": 0.16,
"grad_norm": 2.40625,
"learning_rate": 7.3894736842105275e-06,
"loss": 0.6557,
"step": 6490
},
{
"epoch": 0.16,
"grad_norm": 8.5625,
"learning_rate": 7.368421052631579e-06,
"loss": 0.5214,
"step": 6500
},
{
"epoch": 0.16,
"eval_loss": 0.6133315563201904,
"eval_runtime": 38.1535,
"eval_samples_per_second": 26.21,
"eval_steps_per_second": 26.21,
"step": 6500
},
{
"epoch": 0.16,
"grad_norm": 2.3125,
"learning_rate": 7.347368421052632e-06,
"loss": 0.6287,
"step": 6510
},
{
"epoch": 0.16,
"grad_norm": 6.75,
"learning_rate": 7.326315789473685e-06,
"loss": 0.6126,
"step": 6520
},
{
"epoch": 0.16,
"grad_norm": 4.90625,
"learning_rate": 7.305263157894737e-06,
"loss": 0.68,
"step": 6530
},
{
"epoch": 0.16,
"grad_norm": 4.96875,
"learning_rate": 7.28421052631579e-06,
"loss": 0.4831,
"step": 6540
},
{
"epoch": 0.16,
"grad_norm": 3.328125,
"learning_rate": 7.263157894736843e-06,
"loss": 0.6983,
"step": 6550
},
{
"epoch": 0.16,
"grad_norm": 2.640625,
"learning_rate": 7.242105263157896e-06,
"loss": 0.5214,
"step": 6560
},
{
"epoch": 0.16,
"grad_norm": 4.5625,
"learning_rate": 7.221052631578948e-06,
"loss": 0.5789,
"step": 6570
},
{
"epoch": 0.16,
"grad_norm": 11.3125,
"learning_rate": 7.2000000000000005e-06,
"loss": 0.6369,
"step": 6580
},
{
"epoch": 0.16,
"grad_norm": 4.5625,
"learning_rate": 7.178947368421053e-06,
"loss": 0.608,
"step": 6590
},
{
"epoch": 0.17,
"grad_norm": 3.25,
"learning_rate": 7.157894736842106e-06,
"loss": 0.6514,
"step": 6600
},
{
"epoch": 0.17,
"grad_norm": 8.5625,
"learning_rate": 7.1368421052631585e-06,
"loss": 0.4964,
"step": 6610
},
{
"epoch": 0.17,
"grad_norm": 5.40625,
"learning_rate": 7.115789473684211e-06,
"loss": 0.6726,
"step": 6620
},
{
"epoch": 0.17,
"grad_norm": 2.65625,
"learning_rate": 7.094736842105265e-06,
"loss": 0.636,
"step": 6630
},
{
"epoch": 0.17,
"grad_norm": 5.5625,
"learning_rate": 7.073684210526316e-06,
"loss": 0.69,
"step": 6640
},
{
"epoch": 0.17,
"grad_norm": 15.375,
"learning_rate": 7.052631578947369e-06,
"loss": 0.5622,
"step": 6650
},
{
"epoch": 0.17,
"grad_norm": 6.5625,
"learning_rate": 7.031578947368422e-06,
"loss": 0.5597,
"step": 6660
},
{
"epoch": 0.17,
"grad_norm": 3.46875,
"learning_rate": 7.010526315789474e-06,
"loss": 0.6246,
"step": 6670
},
{
"epoch": 0.17,
"grad_norm": 3.515625,
"learning_rate": 6.989473684210527e-06,
"loss": 0.5107,
"step": 6680
},
{
"epoch": 0.17,
"grad_norm": 3.84375,
"learning_rate": 6.96842105263158e-06,
"loss": 0.646,
"step": 6690
},
{
"epoch": 0.17,
"grad_norm": 3.5625,
"learning_rate": 6.947368421052632e-06,
"loss": 0.659,
"step": 6700
},
{
"epoch": 0.17,
"grad_norm": 2.703125,
"learning_rate": 6.926315789473684e-06,
"loss": 0.4536,
"step": 6710
},
{
"epoch": 0.17,
"grad_norm": 2.703125,
"learning_rate": 6.905263157894737e-06,
"loss": 0.6416,
"step": 6720
},
{
"epoch": 0.17,
"grad_norm": 2.34375,
"learning_rate": 6.8842105263157895e-06,
"loss": 0.6546,
"step": 6730
},
{
"epoch": 0.17,
"grad_norm": 3.046875,
"learning_rate": 6.863157894736843e-06,
"loss": 0.6834,
"step": 6740
},
{
"epoch": 0.17,
"grad_norm": 5.53125,
"learning_rate": 6.842105263157896e-06,
"loss": 0.6329,
"step": 6750
},
{
"epoch": 0.17,
"grad_norm": 4.75,
"learning_rate": 6.821052631578948e-06,
"loss": 0.5559,
"step": 6760
},
{
"epoch": 0.17,
"grad_norm": 3.484375,
"learning_rate": 6.800000000000001e-06,
"loss": 0.5685,
"step": 6770
},
{
"epoch": 0.17,
"grad_norm": 5.03125,
"learning_rate": 6.778947368421053e-06,
"loss": 0.6151,
"step": 6780
},
{
"epoch": 0.17,
"grad_norm": 3.296875,
"learning_rate": 6.7578947368421054e-06,
"loss": 0.6197,
"step": 6790
},
{
"epoch": 0.17,
"grad_norm": 3.578125,
"learning_rate": 6.736842105263158e-06,
"loss": 0.565,
"step": 6800
},
{
"epoch": 0.17,
"grad_norm": 4.09375,
"learning_rate": 6.715789473684211e-06,
"loss": 0.5766,
"step": 6810
},
{
"epoch": 0.17,
"grad_norm": 2.5625,
"learning_rate": 6.694736842105264e-06,
"loss": 0.5622,
"step": 6820
},
{
"epoch": 0.17,
"grad_norm": 13.125,
"learning_rate": 6.673684210526317e-06,
"loss": 0.6691,
"step": 6830
},
{
"epoch": 0.17,
"grad_norm": 9.375,
"learning_rate": 6.6526315789473695e-06,
"loss": 0.5872,
"step": 6840
},
{
"epoch": 0.17,
"grad_norm": 3.296875,
"learning_rate": 6.631578947368421e-06,
"loss": 0.5883,
"step": 6850
},
{
"epoch": 0.17,
"grad_norm": 12.0625,
"learning_rate": 6.610526315789474e-06,
"loss": 0.6339,
"step": 6860
},
{
"epoch": 0.17,
"grad_norm": 2.96875,
"learning_rate": 6.589473684210527e-06,
"loss": 0.612,
"step": 6870
},
{
"epoch": 0.17,
"grad_norm": 3.34375,
"learning_rate": 6.568421052631579e-06,
"loss": 0.5187,
"step": 6880
},
{
"epoch": 0.17,
"grad_norm": 2.828125,
"learning_rate": 6.547368421052632e-06,
"loss": 0.5306,
"step": 6890
},
{
"epoch": 0.17,
"grad_norm": 3.671875,
"learning_rate": 6.526315789473685e-06,
"loss": 0.5485,
"step": 6900
},
{
"epoch": 0.17,
"grad_norm": 9.5625,
"learning_rate": 6.505263157894738e-06,
"loss": 0.6115,
"step": 6910
},
{
"epoch": 0.17,
"grad_norm": 3.9375,
"learning_rate": 6.484210526315789e-06,
"loss": 0.5446,
"step": 6920
},
{
"epoch": 0.17,
"grad_norm": 1.7109375,
"learning_rate": 6.463157894736843e-06,
"loss": 0.6284,
"step": 6930
},
{
"epoch": 0.17,
"grad_norm": 8.5625,
"learning_rate": 6.442105263157895e-06,
"loss": 0.526,
"step": 6940
},
{
"epoch": 0.17,
"grad_norm": 4.34375,
"learning_rate": 6.421052631578948e-06,
"loss": 0.6324,
"step": 6950
},
{
"epoch": 0.17,
"grad_norm": 5.28125,
"learning_rate": 6.4000000000000006e-06,
"loss": 0.575,
"step": 6960
},
{
"epoch": 0.17,
"grad_norm": 3.09375,
"learning_rate": 6.378947368421053e-06,
"loss": 0.5664,
"step": 6970
},
{
"epoch": 0.17,
"grad_norm": 3.75,
"learning_rate": 6.357894736842106e-06,
"loss": 0.5779,
"step": 6980
},
{
"epoch": 0.17,
"grad_norm": 5.15625,
"learning_rate": 6.336842105263158e-06,
"loss": 0.5258,
"step": 6990
},
{
"epoch": 0.17,
"grad_norm": 5.0625,
"learning_rate": 6.31578947368421e-06,
"loss": 0.5891,
"step": 7000
},
{
"epoch": 0.17,
"eval_loss": 0.6335042119026184,
"eval_runtime": 38.1328,
"eval_samples_per_second": 26.224,
"eval_steps_per_second": 26.224,
"step": 7000
},
{
"epoch": 0.18,
"grad_norm": 5.53125,
"learning_rate": 6.294736842105264e-06,
"loss": 0.524,
"step": 7010
},
{
"epoch": 0.18,
"grad_norm": 4.46875,
"learning_rate": 6.2736842105263165e-06,
"loss": 0.6672,
"step": 7020
},
{
"epoch": 0.18,
"grad_norm": 2.4375,
"learning_rate": 6.252631578947369e-06,
"loss": 0.5906,
"step": 7030
},
{
"epoch": 0.18,
"grad_norm": 3.703125,
"learning_rate": 6.231578947368422e-06,
"loss": 0.6795,
"step": 7040
},
{
"epoch": 0.18,
"grad_norm": 5.09375,
"learning_rate": 6.2105263157894745e-06,
"loss": 0.574,
"step": 7050
},
{
"epoch": 0.18,
"grad_norm": 3.921875,
"learning_rate": 6.189473684210526e-06,
"loss": 0.6008,
"step": 7060
},
{
"epoch": 0.18,
"grad_norm": 5.84375,
"learning_rate": 6.168421052631579e-06,
"loss": 0.5116,
"step": 7070
},
{
"epoch": 0.18,
"grad_norm": 3.203125,
"learning_rate": 6.1473684210526316e-06,
"loss": 0.6047,
"step": 7080
},
{
"epoch": 0.18,
"grad_norm": 2.15625,
"learning_rate": 6.126315789473685e-06,
"loss": 0.3864,
"step": 7090
},
{
"epoch": 0.18,
"grad_norm": 3.65625,
"learning_rate": 6.105263157894738e-06,
"loss": 0.5869,
"step": 7100
},
{
"epoch": 0.18,
"grad_norm": 9.875,
"learning_rate": 6.08421052631579e-06,
"loss": 0.5377,
"step": 7110
},
{
"epoch": 0.18,
"grad_norm": 3.078125,
"learning_rate": 6.063157894736843e-06,
"loss": 0.5649,
"step": 7120
},
{
"epoch": 0.18,
"grad_norm": 4.875,
"learning_rate": 6.042105263157895e-06,
"loss": 0.5955,
"step": 7130
},
{
"epoch": 0.18,
"grad_norm": 3.53125,
"learning_rate": 6.0210526315789475e-06,
"loss": 0.5352,
"step": 7140
},
{
"epoch": 0.18,
"grad_norm": 9.625,
"learning_rate": 6e-06,
"loss": 0.5801,
"step": 7150
},
{
"epoch": 0.18,
"grad_norm": 4.375,
"learning_rate": 5.978947368421053e-06,
"loss": 0.4599,
"step": 7160
},
{
"epoch": 0.18,
"grad_norm": 3.3125,
"learning_rate": 5.9578947368421055e-06,
"loss": 0.5337,
"step": 7170
},
{
"epoch": 0.18,
"grad_norm": 3.40625,
"learning_rate": 5.936842105263159e-06,
"loss": 0.5677,
"step": 7180
},
{
"epoch": 0.18,
"grad_norm": 2.90625,
"learning_rate": 5.915789473684212e-06,
"loss": 0.6926,
"step": 7190
},
{
"epoch": 0.18,
"grad_norm": 3.15625,
"learning_rate": 5.8947368421052634e-06,
"loss": 0.6243,
"step": 7200
},
{
"epoch": 0.18,
"grad_norm": 5.5,
"learning_rate": 5.873684210526316e-06,
"loss": 0.5837,
"step": 7210
},
{
"epoch": 0.18,
"grad_norm": 4.53125,
"learning_rate": 5.852631578947369e-06,
"loss": 0.6096,
"step": 7220
},
{
"epoch": 0.18,
"grad_norm": 3.984375,
"learning_rate": 5.831578947368421e-06,
"loss": 0.5598,
"step": 7230
},
{
"epoch": 0.18,
"grad_norm": 4.40625,
"learning_rate": 5.810526315789474e-06,
"loss": 0.5448,
"step": 7240
},
{
"epoch": 0.18,
"grad_norm": 3.671875,
"learning_rate": 5.789473684210527e-06,
"loss": 0.5487,
"step": 7250
},
{
"epoch": 0.18,
"grad_norm": 10.5625,
"learning_rate": 5.76842105263158e-06,
"loss": 0.5739,
"step": 7260
},
{
"epoch": 0.18,
"grad_norm": 9.4375,
"learning_rate": 5.747368421052633e-06,
"loss": 0.6278,
"step": 7270
},
{
"epoch": 0.18,
"grad_norm": 3.578125,
"learning_rate": 5.726315789473685e-06,
"loss": 0.5801,
"step": 7280
},
{
"epoch": 0.18,
"grad_norm": 3.796875,
"learning_rate": 5.705263157894737e-06,
"loss": 0.4992,
"step": 7290
},
{
"epoch": 0.18,
"grad_norm": 4.8125,
"learning_rate": 5.68421052631579e-06,
"loss": 0.607,
"step": 7300
},
{
"epoch": 0.18,
"grad_norm": 5.0,
"learning_rate": 5.663157894736843e-06,
"loss": 0.5189,
"step": 7310
},
{
"epoch": 0.18,
"grad_norm": 5.78125,
"learning_rate": 5.642105263157895e-06,
"loss": 0.6886,
"step": 7320
},
{
"epoch": 0.18,
"grad_norm": 3.46875,
"learning_rate": 5.621052631578948e-06,
"loss": 0.6127,
"step": 7330
},
{
"epoch": 0.18,
"grad_norm": 3.875,
"learning_rate": 5.600000000000001e-06,
"loss": 0.625,
"step": 7340
},
{
"epoch": 0.18,
"grad_norm": 16.25,
"learning_rate": 5.578947368421052e-06,
"loss": 0.537,
"step": 7350
},
{
"epoch": 0.18,
"grad_norm": 3.90625,
"learning_rate": 5.557894736842105e-06,
"loss": 0.5933,
"step": 7360
},
{
"epoch": 0.18,
"grad_norm": 13.375,
"learning_rate": 5.5368421052631586e-06,
"loss": 0.6001,
"step": 7370
},
{
"epoch": 0.18,
"grad_norm": 14.75,
"learning_rate": 5.515789473684211e-06,
"loss": 0.6809,
"step": 7380
},
{
"epoch": 0.18,
"grad_norm": 5.90625,
"learning_rate": 5.494736842105264e-06,
"loss": 0.5997,
"step": 7390
},
{
"epoch": 0.18,
"grad_norm": 4.03125,
"learning_rate": 5.4736842105263165e-06,
"loss": 0.6418,
"step": 7400
},
{
"epoch": 0.19,
"grad_norm": 3.65625,
"learning_rate": 5.452631578947369e-06,
"loss": 0.5285,
"step": 7410
},
{
"epoch": 0.19,
"grad_norm": 10.0625,
"learning_rate": 5.431578947368421e-06,
"loss": 0.6465,
"step": 7420
},
{
"epoch": 0.19,
"grad_norm": 3.359375,
"learning_rate": 5.410526315789474e-06,
"loss": 0.6697,
"step": 7430
},
{
"epoch": 0.19,
"grad_norm": 4.21875,
"learning_rate": 5.389473684210526e-06,
"loss": 0.5943,
"step": 7440
},
{
"epoch": 0.19,
"grad_norm": 5.875,
"learning_rate": 5.36842105263158e-06,
"loss": 0.6122,
"step": 7450
},
{
"epoch": 0.19,
"grad_norm": 2.4375,
"learning_rate": 5.3473684210526325e-06,
"loss": 0.5825,
"step": 7460
},
{
"epoch": 0.19,
"grad_norm": 27.75,
"learning_rate": 5.326315789473685e-06,
"loss": 0.6151,
"step": 7470
},
{
"epoch": 0.19,
"grad_norm": 3.28125,
"learning_rate": 5.305263157894738e-06,
"loss": 0.621,
"step": 7480
},
{
"epoch": 0.19,
"grad_norm": 3.265625,
"learning_rate": 5.2842105263157896e-06,
"loss": 0.5397,
"step": 7490
},
{
"epoch": 0.19,
"grad_norm": 4.53125,
"learning_rate": 5.263157894736842e-06,
"loss": 0.6428,
"step": 7500
},
{
"epoch": 0.19,
"eval_loss": 0.6119253039360046,
"eval_runtime": 38.1677,
"eval_samples_per_second": 26.2,
"eval_steps_per_second": 26.2,
"step": 7500
},
{
"epoch": 0.19,
"grad_norm": 4.6875,
"learning_rate": 5.242105263157895e-06,
"loss": 0.5938,
"step": 7510
},
{
"epoch": 0.19,
"grad_norm": 5.0625,
"learning_rate": 5.2210526315789475e-06,
"loss": 0.6286,
"step": 7520
},
{
"epoch": 0.19,
"grad_norm": 4.53125,
"learning_rate": 5.2e-06,
"loss": 0.7065,
"step": 7530
},
{
"epoch": 0.19,
"grad_norm": 3.28125,
"learning_rate": 5.178947368421054e-06,
"loss": 0.5862,
"step": 7540
},
{
"epoch": 0.19,
"grad_norm": 4.5,
"learning_rate": 5.157894736842106e-06,
"loss": 0.5015,
"step": 7550
},
{
"epoch": 0.19,
"grad_norm": 2.640625,
"learning_rate": 5.136842105263158e-06,
"loss": 0.6041,
"step": 7560
},
{
"epoch": 0.19,
"grad_norm": 5.59375,
"learning_rate": 5.115789473684211e-06,
"loss": 0.6845,
"step": 7570
},
{
"epoch": 0.19,
"grad_norm": 6.59375,
"learning_rate": 5.0947368421052635e-06,
"loss": 0.698,
"step": 7580
},
{
"epoch": 0.19,
"grad_norm": 9.8125,
"learning_rate": 5.073684210526316e-06,
"loss": 0.4452,
"step": 7590
},
{
"epoch": 0.19,
"grad_norm": 3.1875,
"learning_rate": 5.052631578947369e-06,
"loss": 0.6485,
"step": 7600
},
{
"epoch": 0.19,
"grad_norm": 5.8125,
"learning_rate": 5.0315789473684214e-06,
"loss": 0.5703,
"step": 7610
},
{
"epoch": 0.19,
"grad_norm": 3.953125,
"learning_rate": 5.010526315789475e-06,
"loss": 0.5871,
"step": 7620
},
{
"epoch": 0.19,
"grad_norm": 5.5625,
"learning_rate": 4.989473684210527e-06,
"loss": 0.6024,
"step": 7630
},
{
"epoch": 0.19,
"grad_norm": 2.9375,
"learning_rate": 4.968421052631579e-06,
"loss": 0.5723,
"step": 7640
},
{
"epoch": 0.19,
"grad_norm": 3.15625,
"learning_rate": 4.947368421052632e-06,
"loss": 0.584,
"step": 7650
},
{
"epoch": 0.19,
"grad_norm": 8.4375,
"learning_rate": 4.926315789473685e-06,
"loss": 0.6387,
"step": 7660
},
{
"epoch": 0.19,
"grad_norm": 4.34375,
"learning_rate": 4.905263157894737e-06,
"loss": 0.6574,
"step": 7670
},
{
"epoch": 0.19,
"grad_norm": 3.578125,
"learning_rate": 4.88421052631579e-06,
"loss": 0.5907,
"step": 7680
},
{
"epoch": 0.19,
"grad_norm": 4.5,
"learning_rate": 4.863157894736843e-06,
"loss": 0.6319,
"step": 7690
},
{
"epoch": 0.19,
"grad_norm": 3.234375,
"learning_rate": 4.842105263157895e-06,
"loss": 0.5053,
"step": 7700
},
{
"epoch": 0.19,
"grad_norm": 1.7421875,
"learning_rate": 4.821052631578948e-06,
"loss": 0.5135,
"step": 7710
},
{
"epoch": 0.19,
"grad_norm": 17.375,
"learning_rate": 4.800000000000001e-06,
"loss": 0.6893,
"step": 7720
},
{
"epoch": 0.19,
"grad_norm": 3.609375,
"learning_rate": 4.778947368421053e-06,
"loss": 0.6257,
"step": 7730
},
{
"epoch": 0.19,
"grad_norm": 4.46875,
"learning_rate": 4.757894736842106e-06,
"loss": 0.5808,
"step": 7740
},
{
"epoch": 0.19,
"grad_norm": 3.984375,
"learning_rate": 4.736842105263158e-06,
"loss": 0.518,
"step": 7750
},
{
"epoch": 0.19,
"grad_norm": 5.90625,
"learning_rate": 4.71578947368421e-06,
"loss": 0.5991,
"step": 7760
},
{
"epoch": 0.19,
"grad_norm": 4.71875,
"learning_rate": 4.694736842105264e-06,
"loss": 0.6514,
"step": 7770
},
{
"epoch": 0.19,
"grad_norm": 3.109375,
"learning_rate": 4.6736842105263166e-06,
"loss": 0.489,
"step": 7780
},
{
"epoch": 0.19,
"grad_norm": 3.609375,
"learning_rate": 4.652631578947368e-06,
"loss": 0.5181,
"step": 7790
},
{
"epoch": 0.2,
"grad_norm": 2.78125,
"learning_rate": 4.631578947368421e-06,
"loss": 0.5767,
"step": 7800
},
{
"epoch": 0.2,
"grad_norm": 3.171875,
"learning_rate": 4.6105263157894745e-06,
"loss": 0.4968,
"step": 7810
},
{
"epoch": 0.2,
"grad_norm": 4.8125,
"learning_rate": 4.589473684210526e-06,
"loss": 0.6562,
"step": 7820
},
{
"epoch": 0.2,
"grad_norm": 2.703125,
"learning_rate": 4.568421052631579e-06,
"loss": 0.6273,
"step": 7830
},
{
"epoch": 0.2,
"grad_norm": 4.75,
"learning_rate": 4.547368421052632e-06,
"loss": 0.6351,
"step": 7840
},
{
"epoch": 0.2,
"grad_norm": 2.9375,
"learning_rate": 4.526315789473685e-06,
"loss": 0.5495,
"step": 7850
},
{
"epoch": 0.2,
"grad_norm": 10.6875,
"learning_rate": 4.505263157894737e-06,
"loss": 0.6603,
"step": 7860
},
{
"epoch": 0.2,
"grad_norm": 4.25,
"learning_rate": 4.48421052631579e-06,
"loss": 0.5525,
"step": 7870
},
{
"epoch": 0.2,
"grad_norm": 3.171875,
"learning_rate": 4.463157894736842e-06,
"loss": 0.6219,
"step": 7880
},
{
"epoch": 0.2,
"grad_norm": 2.859375,
"learning_rate": 4.442105263157896e-06,
"loss": 0.5518,
"step": 7890
},
{
"epoch": 0.2,
"grad_norm": 3.109375,
"learning_rate": 4.4210526315789476e-06,
"loss": 0.5462,
"step": 7900
},
{
"epoch": 0.2,
"grad_norm": 3.359375,
"learning_rate": 4.4e-06,
"loss": 0.5776,
"step": 7910
},
{
"epoch": 0.2,
"grad_norm": 4.4375,
"learning_rate": 4.378947368421053e-06,
"loss": 0.4804,
"step": 7920
},
{
"epoch": 0.2,
"grad_norm": 3.578125,
"learning_rate": 4.3578947368421055e-06,
"loss": 0.5858,
"step": 7930
},
{
"epoch": 0.2,
"grad_norm": 3.71875,
"learning_rate": 4.336842105263158e-06,
"loss": 0.6847,
"step": 7940
},
{
"epoch": 0.2,
"grad_norm": 3.984375,
"learning_rate": 4.315789473684211e-06,
"loss": 0.6626,
"step": 7950
},
{
"epoch": 0.2,
"grad_norm": 9.4375,
"learning_rate": 4.2947368421052635e-06,
"loss": 0.5526,
"step": 7960
},
{
"epoch": 0.2,
"grad_norm": 4.625,
"learning_rate": 4.273684210526316e-06,
"loss": 0.5661,
"step": 7970
},
{
"epoch": 0.2,
"grad_norm": 4.28125,
"learning_rate": 4.252631578947369e-06,
"loss": 0.6182,
"step": 7980
},
{
"epoch": 0.2,
"grad_norm": 6.3125,
"learning_rate": 4.2315789473684215e-06,
"loss": 0.629,
"step": 7990
},
{
"epoch": 0.2,
"grad_norm": 3.59375,
"learning_rate": 4.210526315789474e-06,
"loss": 0.7233,
"step": 8000
},
{
"epoch": 0.2,
"eval_loss": 0.6029936671257019,
"eval_runtime": 38.1479,
"eval_samples_per_second": 26.214,
"eval_steps_per_second": 26.214,
"step": 8000
},
{
"epoch": 0.2,
"grad_norm": 2.828125,
"learning_rate": 4.189473684210527e-06,
"loss": 0.5451,
"step": 8010
},
{
"epoch": 0.2,
"grad_norm": 5.0,
"learning_rate": 4.1684210526315794e-06,
"loss": 0.5498,
"step": 8020
},
{
"epoch": 0.2,
"grad_norm": 4.875,
"learning_rate": 4.147368421052632e-06,
"loss": 0.5518,
"step": 8030
},
{
"epoch": 0.2,
"grad_norm": 3.609375,
"learning_rate": 4.126315789473685e-06,
"loss": 0.6088,
"step": 8040
},
{
"epoch": 0.2,
"grad_norm": 41.0,
"learning_rate": 4.105263157894737e-06,
"loss": 0.6399,
"step": 8050
},
{
"epoch": 0.2,
"grad_norm": 3.578125,
"learning_rate": 4.08421052631579e-06,
"loss": 0.5412,
"step": 8060
},
{
"epoch": 0.2,
"grad_norm": 4.21875,
"learning_rate": 4.063157894736842e-06,
"loss": 0.4933,
"step": 8070
},
{
"epoch": 0.2,
"grad_norm": 3.234375,
"learning_rate": 4.042105263157895e-06,
"loss": 0.6182,
"step": 8080
},
{
"epoch": 0.2,
"grad_norm": 2.953125,
"learning_rate": 4.021052631578948e-06,
"loss": 0.5342,
"step": 8090
},
{
"epoch": 0.2,
"grad_norm": 1.6328125,
"learning_rate": 4.000000000000001e-06,
"loss": 0.5838,
"step": 8100
},
{
"epoch": 0.2,
"grad_norm": 2.40625,
"learning_rate": 3.9789473684210525e-06,
"loss": 0.6717,
"step": 8110
},
{
"epoch": 0.2,
"grad_norm": 7.0,
"learning_rate": 3.957894736842106e-06,
"loss": 0.6484,
"step": 8120
},
{
"epoch": 0.2,
"grad_norm": 10.4375,
"learning_rate": 3.936842105263159e-06,
"loss": 0.6253,
"step": 8130
},
{
"epoch": 0.2,
"grad_norm": 3.078125,
"learning_rate": 3.9157894736842104e-06,
"loss": 0.5562,
"step": 8140
},
{
"epoch": 0.2,
"grad_norm": 5.4375,
"learning_rate": 3.894736842105263e-06,
"loss": 0.6162,
"step": 8150
},
{
"epoch": 0.2,
"grad_norm": 2.328125,
"learning_rate": 3.873684210526316e-06,
"loss": 0.5602,
"step": 8160
},
{
"epoch": 0.2,
"grad_norm": 6.78125,
"learning_rate": 3.852631578947369e-06,
"loss": 0.5414,
"step": 8170
},
{
"epoch": 0.2,
"grad_norm": 4.40625,
"learning_rate": 3.831578947368421e-06,
"loss": 0.5783,
"step": 8180
},
{
"epoch": 0.2,
"grad_norm": 3.46875,
"learning_rate": 3.810526315789474e-06,
"loss": 0.583,
"step": 8190
},
{
"epoch": 0.2,
"grad_norm": 6.5625,
"learning_rate": 3.789473684210527e-06,
"loss": 0.632,
"step": 8200
},
{
"epoch": 0.21,
"grad_norm": 3.625,
"learning_rate": 3.768421052631579e-06,
"loss": 0.5573,
"step": 8210
},
{
"epoch": 0.21,
"grad_norm": 3.578125,
"learning_rate": 3.7473684210526317e-06,
"loss": 0.5176,
"step": 8220
},
{
"epoch": 0.21,
"grad_norm": 2.796875,
"learning_rate": 3.7263157894736848e-06,
"loss": 0.4684,
"step": 8230
},
{
"epoch": 0.21,
"grad_norm": 6.71875,
"learning_rate": 3.7052631578947374e-06,
"loss": 0.6114,
"step": 8240
},
{
"epoch": 0.21,
"grad_norm": 5.5625,
"learning_rate": 3.6842105263157896e-06,
"loss": 0.5437,
"step": 8250
},
{
"epoch": 0.21,
"grad_norm": 4.53125,
"learning_rate": 3.6631578947368423e-06,
"loss": 0.5119,
"step": 8260
},
{
"epoch": 0.21,
"grad_norm": 4.21875,
"learning_rate": 3.642105263157895e-06,
"loss": 0.5742,
"step": 8270
},
{
"epoch": 0.21,
"grad_norm": 5.46875,
"learning_rate": 3.621052631578948e-06,
"loss": 0.6704,
"step": 8280
},
{
"epoch": 0.21,
"grad_norm": 3.25,
"learning_rate": 3.6000000000000003e-06,
"loss": 0.5655,
"step": 8290
},
{
"epoch": 0.21,
"grad_norm": 3.15625,
"learning_rate": 3.578947368421053e-06,
"loss": 0.5382,
"step": 8300
},
{
"epoch": 0.21,
"grad_norm": 6.875,
"learning_rate": 3.5578947368421056e-06,
"loss": 0.5485,
"step": 8310
},
{
"epoch": 0.21,
"grad_norm": 7.3125,
"learning_rate": 3.536842105263158e-06,
"loss": 0.5124,
"step": 8320
},
{
"epoch": 0.21,
"grad_norm": 4.625,
"learning_rate": 3.515789473684211e-06,
"loss": 0.4819,
"step": 8330
},
{
"epoch": 0.21,
"grad_norm": 6.53125,
"learning_rate": 3.4947368421052635e-06,
"loss": 0.5999,
"step": 8340
},
{
"epoch": 0.21,
"grad_norm": 4.3125,
"learning_rate": 3.473684210526316e-06,
"loss": 0.5241,
"step": 8350
},
{
"epoch": 0.21,
"grad_norm": 7.78125,
"learning_rate": 3.4526315789473684e-06,
"loss": 0.6193,
"step": 8360
},
{
"epoch": 0.21,
"grad_norm": 5.09375,
"learning_rate": 3.4315789473684215e-06,
"loss": 0.6701,
"step": 8370
},
{
"epoch": 0.21,
"grad_norm": 2.453125,
"learning_rate": 3.410526315789474e-06,
"loss": 0.4458,
"step": 8380
},
{
"epoch": 0.21,
"grad_norm": 2.53125,
"learning_rate": 3.3894736842105264e-06,
"loss": 0.5441,
"step": 8390
},
{
"epoch": 0.21,
"grad_norm": 3.71875,
"learning_rate": 3.368421052631579e-06,
"loss": 0.5787,
"step": 8400
},
{
"epoch": 0.21,
"grad_norm": 5.1875,
"learning_rate": 3.347368421052632e-06,
"loss": 0.5972,
"step": 8410
},
{
"epoch": 0.21,
"grad_norm": 3.828125,
"learning_rate": 3.3263157894736848e-06,
"loss": 0.5784,
"step": 8420
},
{
"epoch": 0.21,
"grad_norm": 2.40625,
"learning_rate": 3.305263157894737e-06,
"loss": 0.6005,
"step": 8430
},
{
"epoch": 0.21,
"grad_norm": 4.875,
"learning_rate": 3.2842105263157897e-06,
"loss": 0.5343,
"step": 8440
},
{
"epoch": 0.21,
"grad_norm": 3.0,
"learning_rate": 3.2631578947368423e-06,
"loss": 0.5407,
"step": 8450
},
{
"epoch": 0.21,
"grad_norm": 3.234375,
"learning_rate": 3.2421052631578945e-06,
"loss": 0.5047,
"step": 8460
},
{
"epoch": 0.21,
"grad_norm": 5.6875,
"learning_rate": 3.2210526315789476e-06,
"loss": 0.4705,
"step": 8470
},
{
"epoch": 0.21,
"grad_norm": 2.515625,
"learning_rate": 3.2000000000000003e-06,
"loss": 0.5838,
"step": 8480
},
{
"epoch": 0.21,
"grad_norm": 3.453125,
"learning_rate": 3.178947368421053e-06,
"loss": 0.547,
"step": 8490
},
{
"epoch": 0.21,
"grad_norm": 5.0625,
"learning_rate": 3.157894736842105e-06,
"loss": 0.6311,
"step": 8500
},
{
"epoch": 0.21,
"eval_loss": 0.6170388460159302,
"eval_runtime": 38.1693,
"eval_samples_per_second": 26.199,
"eval_steps_per_second": 26.199,
"step": 8500
},
{
"epoch": 0.21,
"grad_norm": 5.96875,
"learning_rate": 3.1368421052631582e-06,
"loss": 0.6415,
"step": 8510
},
{
"epoch": 0.21,
"grad_norm": 3.484375,
"learning_rate": 3.115789473684211e-06,
"loss": 0.5877,
"step": 8520
},
{
"epoch": 0.21,
"grad_norm": 3.8125,
"learning_rate": 3.094736842105263e-06,
"loss": 0.5951,
"step": 8530
},
{
"epoch": 0.21,
"grad_norm": 4.0625,
"learning_rate": 3.0736842105263158e-06,
"loss": 0.6047,
"step": 8540
},
{
"epoch": 0.21,
"grad_norm": 2.609375,
"learning_rate": 3.052631578947369e-06,
"loss": 0.5694,
"step": 8550
},
{
"epoch": 0.21,
"grad_norm": 4.0625,
"learning_rate": 3.0315789473684215e-06,
"loss": 0.6126,
"step": 8560
},
{
"epoch": 0.21,
"grad_norm": 2.75,
"learning_rate": 3.0105263157894737e-06,
"loss": 0.521,
"step": 8570
},
{
"epoch": 0.21,
"grad_norm": 4.46875,
"learning_rate": 2.9894736842105264e-06,
"loss": 0.6389,
"step": 8580
},
{
"epoch": 0.21,
"grad_norm": 4.75,
"learning_rate": 2.9684210526315795e-06,
"loss": 0.5909,
"step": 8590
},
{
"epoch": 0.21,
"grad_norm": 4.5625,
"learning_rate": 2.9473684210526317e-06,
"loss": 0.6578,
"step": 8600
},
{
"epoch": 0.22,
"grad_norm": 6.0,
"learning_rate": 2.9263157894736844e-06,
"loss": 0.5277,
"step": 8610
},
{
"epoch": 0.22,
"grad_norm": 2.71875,
"learning_rate": 2.905263157894737e-06,
"loss": 0.5313,
"step": 8620
},
{
"epoch": 0.22,
"grad_norm": 3.34375,
"learning_rate": 2.88421052631579e-06,
"loss": 0.5201,
"step": 8630
},
{
"epoch": 0.22,
"grad_norm": 4.0,
"learning_rate": 2.8631578947368423e-06,
"loss": 0.7751,
"step": 8640
},
{
"epoch": 0.22,
"grad_norm": 2.546875,
"learning_rate": 2.842105263157895e-06,
"loss": 0.6536,
"step": 8650
},
{
"epoch": 0.22,
"grad_norm": 6.34375,
"learning_rate": 2.8210526315789476e-06,
"loss": 0.6661,
"step": 8660
},
{
"epoch": 0.22,
"grad_norm": 4.65625,
"learning_rate": 2.8000000000000003e-06,
"loss": 0.6284,
"step": 8670
},
{
"epoch": 0.22,
"grad_norm": 5.5625,
"learning_rate": 2.7789473684210525e-06,
"loss": 0.673,
"step": 8680
},
{
"epoch": 0.22,
"grad_norm": 3.03125,
"learning_rate": 2.7578947368421056e-06,
"loss": 0.646,
"step": 8690
},
{
"epoch": 0.22,
"grad_norm": 2.765625,
"learning_rate": 2.7368421052631583e-06,
"loss": 0.5917,
"step": 8700
},
{
"epoch": 0.22,
"grad_norm": 6.03125,
"learning_rate": 2.7157894736842105e-06,
"loss": 0.5706,
"step": 8710
},
{
"epoch": 0.22,
"grad_norm": 3.171875,
"learning_rate": 2.694736842105263e-06,
"loss": 0.477,
"step": 8720
},
{
"epoch": 0.22,
"grad_norm": 5.875,
"learning_rate": 2.6736842105263162e-06,
"loss": 0.646,
"step": 8730
},
{
"epoch": 0.22,
"grad_norm": 4.75,
"learning_rate": 2.652631578947369e-06,
"loss": 0.638,
"step": 8740
},
{
"epoch": 0.22,
"grad_norm": 3.4375,
"learning_rate": 2.631578947368421e-06,
"loss": 0.6282,
"step": 8750
},
{
"epoch": 0.22,
"grad_norm": 3.6875,
"learning_rate": 2.6105263157894738e-06,
"loss": 0.6418,
"step": 8760
},
{
"epoch": 0.22,
"grad_norm": 2.828125,
"learning_rate": 2.589473684210527e-06,
"loss": 0.6228,
"step": 8770
},
{
"epoch": 0.22,
"grad_norm": 3.765625,
"learning_rate": 2.568421052631579e-06,
"loss": 0.5794,
"step": 8780
},
{
"epoch": 0.22,
"grad_norm": 4.9375,
"learning_rate": 2.5473684210526317e-06,
"loss": 0.5207,
"step": 8790
},
{
"epoch": 0.22,
"grad_norm": 4.75,
"learning_rate": 2.5263157894736844e-06,
"loss": 0.623,
"step": 8800
},
{
"epoch": 0.22,
"grad_norm": 4.40625,
"learning_rate": 2.5052631578947375e-06,
"loss": 0.6086,
"step": 8810
},
{
"epoch": 0.22,
"grad_norm": 3.109375,
"learning_rate": 2.4842105263157897e-06,
"loss": 0.579,
"step": 8820
},
{
"epoch": 0.22,
"grad_norm": 3.65625,
"learning_rate": 2.4631578947368424e-06,
"loss": 0.4531,
"step": 8830
},
{
"epoch": 0.22,
"grad_norm": 4.75,
"learning_rate": 2.442105263157895e-06,
"loss": 0.5167,
"step": 8840
},
{
"epoch": 0.22,
"grad_norm": 7.78125,
"learning_rate": 2.4210526315789477e-06,
"loss": 0.6362,
"step": 8850
},
{
"epoch": 0.22,
"grad_norm": 5.53125,
"learning_rate": 2.4000000000000003e-06,
"loss": 0.6628,
"step": 8860
},
{
"epoch": 0.22,
"grad_norm": 2.453125,
"learning_rate": 2.378947368421053e-06,
"loss": 0.722,
"step": 8870
},
{
"epoch": 0.22,
"grad_norm": 2.921875,
"learning_rate": 2.357894736842105e-06,
"loss": 0.5313,
"step": 8880
},
{
"epoch": 0.22,
"grad_norm": 5.25,
"learning_rate": 2.3368421052631583e-06,
"loss": 0.6277,
"step": 8890
},
{
"epoch": 0.22,
"grad_norm": 4.0,
"learning_rate": 2.3157894736842105e-06,
"loss": 0.5938,
"step": 8900
},
{
"epoch": 0.22,
"grad_norm": 3.546875,
"learning_rate": 2.294736842105263e-06,
"loss": 0.6851,
"step": 8910
},
{
"epoch": 0.22,
"grad_norm": 6.0,
"learning_rate": 2.273684210526316e-06,
"loss": 0.4746,
"step": 8920
},
{
"epoch": 0.22,
"grad_norm": 4.0,
"learning_rate": 2.2526315789473685e-06,
"loss": 0.5109,
"step": 8930
},
{
"epoch": 0.22,
"grad_norm": 2.421875,
"learning_rate": 2.231578947368421e-06,
"loss": 0.4732,
"step": 8940
},
{
"epoch": 0.22,
"grad_norm": 3.234375,
"learning_rate": 2.2105263157894738e-06,
"loss": 0.7736,
"step": 8950
},
{
"epoch": 0.22,
"grad_norm": 4.34375,
"learning_rate": 2.1894736842105264e-06,
"loss": 0.6251,
"step": 8960
},
{
"epoch": 0.22,
"grad_norm": 2.75,
"learning_rate": 2.168421052631579e-06,
"loss": 0.6428,
"step": 8970
},
{
"epoch": 0.22,
"grad_norm": 11.5625,
"learning_rate": 2.1473684210526317e-06,
"loss": 0.5886,
"step": 8980
},
{
"epoch": 0.22,
"grad_norm": 3.8125,
"learning_rate": 2.1263157894736844e-06,
"loss": 0.597,
"step": 8990
},
{
"epoch": 0.23,
"grad_norm": 1.8125,
"learning_rate": 2.105263157894737e-06,
"loss": 0.527,
"step": 9000
},
{
"epoch": 0.23,
"eval_loss": 0.6106051802635193,
"eval_runtime": 38.1285,
"eval_samples_per_second": 26.227,
"eval_steps_per_second": 26.227,
"step": 9000
},
{
"epoch": 0.23,
"grad_norm": 4.375,
"learning_rate": 2.0842105263157897e-06,
"loss": 0.5752,
"step": 9010
},
{
"epoch": 0.23,
"grad_norm": 4.9375,
"learning_rate": 2.0631578947368424e-06,
"loss": 0.5578,
"step": 9020
},
{
"epoch": 0.23,
"grad_norm": 3.9375,
"learning_rate": 2.042105263157895e-06,
"loss": 0.6715,
"step": 9030
},
{
"epoch": 0.23,
"grad_norm": 6.28125,
"learning_rate": 2.0210526315789477e-06,
"loss": 0.5559,
"step": 9040
},
{
"epoch": 0.23,
"grad_norm": 2.734375,
"learning_rate": 2.0000000000000003e-06,
"loss": 0.5839,
"step": 9050
},
{
"epoch": 0.23,
"grad_norm": 4.375,
"learning_rate": 1.978947368421053e-06,
"loss": 0.6546,
"step": 9060
},
{
"epoch": 0.23,
"grad_norm": 5.1875,
"learning_rate": 1.9578947368421052e-06,
"loss": 0.4581,
"step": 9070
},
{
"epoch": 0.23,
"grad_norm": 1.8046875,
"learning_rate": 1.936842105263158e-06,
"loss": 0.6111,
"step": 9080
},
{
"epoch": 0.23,
"grad_norm": 4.0625,
"learning_rate": 1.9157894736842105e-06,
"loss": 0.569,
"step": 9090
},
{
"epoch": 0.23,
"grad_norm": 3.25,
"learning_rate": 1.8947368421052634e-06,
"loss": 0.6089,
"step": 9100
},
{
"epoch": 0.23,
"grad_norm": 8.0625,
"learning_rate": 1.8736842105263158e-06,
"loss": 0.5164,
"step": 9110
},
{
"epoch": 0.23,
"grad_norm": 2.875,
"learning_rate": 1.8526315789473687e-06,
"loss": 0.5259,
"step": 9120
},
{
"epoch": 0.23,
"grad_norm": 4.0625,
"learning_rate": 1.8315789473684211e-06,
"loss": 0.6318,
"step": 9130
},
{
"epoch": 0.23,
"grad_norm": 3.625,
"learning_rate": 1.810526315789474e-06,
"loss": 0.5954,
"step": 9140
},
{
"epoch": 0.23,
"grad_norm": 4.25,
"learning_rate": 1.7894736842105265e-06,
"loss": 0.5146,
"step": 9150
},
{
"epoch": 0.23,
"grad_norm": 5.28125,
"learning_rate": 1.768421052631579e-06,
"loss": 0.6194,
"step": 9160
},
{
"epoch": 0.23,
"grad_norm": 3.3125,
"learning_rate": 1.7473684210526318e-06,
"loss": 0.5031,
"step": 9170
},
{
"epoch": 0.23,
"grad_norm": 3.59375,
"learning_rate": 1.7263157894736842e-06,
"loss": 0.5199,
"step": 9180
},
{
"epoch": 0.23,
"grad_norm": 12.75,
"learning_rate": 1.705263157894737e-06,
"loss": 0.6261,
"step": 9190
},
{
"epoch": 0.23,
"grad_norm": 3.640625,
"learning_rate": 1.6842105263157895e-06,
"loss": 0.6349,
"step": 9200
},
{
"epoch": 0.23,
"grad_norm": 3.59375,
"learning_rate": 1.6631578947368424e-06,
"loss": 0.5244,
"step": 9210
},
{
"epoch": 0.23,
"grad_norm": 5.15625,
"learning_rate": 1.6421052631578948e-06,
"loss": 0.5428,
"step": 9220
},
{
"epoch": 0.23,
"grad_norm": 2.625,
"learning_rate": 1.6210526315789473e-06,
"loss": 0.6296,
"step": 9230
},
{
"epoch": 0.23,
"grad_norm": 4.125,
"learning_rate": 1.6000000000000001e-06,
"loss": 0.5923,
"step": 9240
},
{
"epoch": 0.23,
"grad_norm": 3.875,
"learning_rate": 1.5789473684210526e-06,
"loss": 0.6146,
"step": 9250
},
{
"epoch": 0.23,
"grad_norm": 6.4375,
"learning_rate": 1.5578947368421054e-06,
"loss": 0.6558,
"step": 9260
},
{
"epoch": 0.23,
"grad_norm": 5.28125,
"learning_rate": 1.5368421052631579e-06,
"loss": 0.5426,
"step": 9270
},
{
"epoch": 0.23,
"grad_norm": 3.046875,
"learning_rate": 1.5157894736842108e-06,
"loss": 0.611,
"step": 9280
},
{
"epoch": 0.23,
"grad_norm": 4.5625,
"learning_rate": 1.4947368421052632e-06,
"loss": 0.5013,
"step": 9290
},
{
"epoch": 0.23,
"grad_norm": 3.34375,
"learning_rate": 1.4736842105263159e-06,
"loss": 0.599,
"step": 9300
},
{
"epoch": 0.23,
"grad_norm": 4.3125,
"learning_rate": 1.4526315789473685e-06,
"loss": 0.5549,
"step": 9310
},
{
"epoch": 0.23,
"grad_norm": 7.375,
"learning_rate": 1.4315789473684212e-06,
"loss": 0.5597,
"step": 9320
},
{
"epoch": 0.23,
"grad_norm": 4.9375,
"learning_rate": 1.4105263157894738e-06,
"loss": 0.5402,
"step": 9330
},
{
"epoch": 0.23,
"grad_norm": 3.34375,
"learning_rate": 1.3894736842105263e-06,
"loss": 0.56,
"step": 9340
},
{
"epoch": 0.23,
"grad_norm": 3.5,
"learning_rate": 1.3684210526315791e-06,
"loss": 0.5991,
"step": 9350
},
{
"epoch": 0.23,
"grad_norm": 3.625,
"learning_rate": 1.3473684210526316e-06,
"loss": 0.499,
"step": 9360
},
{
"epoch": 0.23,
"grad_norm": 4.125,
"learning_rate": 1.3263157894736844e-06,
"loss": 0.5946,
"step": 9370
},
{
"epoch": 0.23,
"grad_norm": 4.59375,
"learning_rate": 1.3052631578947369e-06,
"loss": 0.678,
"step": 9380
},
{
"epoch": 0.23,
"grad_norm": 2.9375,
"learning_rate": 1.2842105263157895e-06,
"loss": 0.7289,
"step": 9390
},
{
"epoch": 0.23,
"grad_norm": 3.953125,
"learning_rate": 1.2631578947368422e-06,
"loss": 0.6926,
"step": 9400
},
{
"epoch": 0.24,
"grad_norm": 9.1875,
"learning_rate": 1.2421052631578948e-06,
"loss": 0.6079,
"step": 9410
},
{
"epoch": 0.24,
"grad_norm": 9.4375,
"learning_rate": 1.2210526315789475e-06,
"loss": 0.6642,
"step": 9420
},
{
"epoch": 0.24,
"grad_norm": 4.28125,
"learning_rate": 1.2000000000000002e-06,
"loss": 0.5805,
"step": 9430
},
{
"epoch": 0.24,
"grad_norm": 4.6875,
"learning_rate": 1.1789473684210526e-06,
"loss": 0.564,
"step": 9440
},
{
"epoch": 0.24,
"grad_norm": 3.34375,
"learning_rate": 1.1578947368421053e-06,
"loss": 0.571,
"step": 9450
},
{
"epoch": 0.24,
"grad_norm": 4.625,
"learning_rate": 1.136842105263158e-06,
"loss": 0.538,
"step": 9460
},
{
"epoch": 0.24,
"grad_norm": 7.84375,
"learning_rate": 1.1157894736842106e-06,
"loss": 0.5777,
"step": 9470
},
{
"epoch": 0.24,
"grad_norm": 3.65625,
"learning_rate": 1.0947368421052632e-06,
"loss": 0.683,
"step": 9480
},
{
"epoch": 0.24,
"grad_norm": 3.3125,
"learning_rate": 1.0736842105263159e-06,
"loss": 0.5143,
"step": 9490
},
{
"epoch": 0.24,
"grad_norm": 3.109375,
"learning_rate": 1.0526315789473685e-06,
"loss": 0.4899,
"step": 9500
},
{
"epoch": 0.24,
"eval_loss": 0.6154074668884277,
"eval_runtime": 38.1761,
"eval_samples_per_second": 26.194,
"eval_steps_per_second": 26.194,
"step": 9500
},
{
"epoch": 0.24,
"grad_norm": 7.125,
"learning_rate": 1.0315789473684212e-06,
"loss": 0.7389,
"step": 9510
},
{
"epoch": 0.24,
"grad_norm": 3.734375,
"learning_rate": 1.0105263157894738e-06,
"loss": 0.5922,
"step": 9520
},
{
"epoch": 0.24,
"grad_norm": 4.46875,
"learning_rate": 9.894736842105265e-07,
"loss": 0.5658,
"step": 9530
},
{
"epoch": 0.24,
"grad_norm": 3.765625,
"learning_rate": 9.68421052631579e-07,
"loss": 0.608,
"step": 9540
},
{
"epoch": 0.24,
"grad_norm": 3.78125,
"learning_rate": 9.473684210526317e-07,
"loss": 0.5337,
"step": 9550
},
{
"epoch": 0.24,
"grad_norm": 3.515625,
"learning_rate": 9.263157894736844e-07,
"loss": 0.5736,
"step": 9560
},
{
"epoch": 0.24,
"grad_norm": 4.03125,
"learning_rate": 9.05263157894737e-07,
"loss": 0.5319,
"step": 9570
},
{
"epoch": 0.24,
"grad_norm": 3.859375,
"learning_rate": 8.842105263157895e-07,
"loss": 0.5011,
"step": 9580
},
{
"epoch": 0.24,
"grad_norm": 2.828125,
"learning_rate": 8.631578947368421e-07,
"loss": 0.6734,
"step": 9590
},
{
"epoch": 0.24,
"grad_norm": 2.625,
"learning_rate": 8.421052631578948e-07,
"loss": 0.5045,
"step": 9600
},
{
"epoch": 0.24,
"grad_norm": 4.78125,
"learning_rate": 8.210526315789474e-07,
"loss": 0.5957,
"step": 9610
},
{
"epoch": 0.24,
"grad_norm": 2.984375,
"learning_rate": 8.000000000000001e-07,
"loss": 0.6944,
"step": 9620
},
{
"epoch": 0.24,
"grad_norm": 2.546875,
"learning_rate": 7.789473684210527e-07,
"loss": 0.5249,
"step": 9630
},
{
"epoch": 0.24,
"grad_norm": 12.9375,
"learning_rate": 7.578947368421054e-07,
"loss": 0.5981,
"step": 9640
},
{
"epoch": 0.24,
"grad_norm": 7.25,
"learning_rate": 7.368421052631579e-07,
"loss": 0.633,
"step": 9650
},
{
"epoch": 0.24,
"grad_norm": 3.59375,
"learning_rate": 7.157894736842106e-07,
"loss": 0.5535,
"step": 9660
},
{
"epoch": 0.24,
"grad_norm": 5.8125,
"learning_rate": 6.947368421052631e-07,
"loss": 0.6233,
"step": 9670
},
{
"epoch": 0.24,
"grad_norm": 11.4375,
"learning_rate": 6.736842105263158e-07,
"loss": 0.5603,
"step": 9680
},
{
"epoch": 0.24,
"grad_norm": 3.890625,
"learning_rate": 6.526315789473684e-07,
"loss": 0.5977,
"step": 9690
},
{
"epoch": 0.24,
"grad_norm": 3.25,
"learning_rate": 6.315789473684211e-07,
"loss": 0.5737,
"step": 9700
},
{
"epoch": 0.24,
"grad_norm": 8.0625,
"learning_rate": 6.105263157894738e-07,
"loss": 0.5797,
"step": 9710
},
{
"epoch": 0.24,
"grad_norm": 6.4375,
"learning_rate": 5.894736842105263e-07,
"loss": 0.6256,
"step": 9720
},
{
"epoch": 0.24,
"grad_norm": 2.484375,
"learning_rate": 5.68421052631579e-07,
"loss": 0.4984,
"step": 9730
},
{
"epoch": 0.24,
"grad_norm": 4.28125,
"learning_rate": 5.473684210526316e-07,
"loss": 0.6489,
"step": 9740
},
{
"epoch": 0.24,
"grad_norm": 2.328125,
"learning_rate": 5.263157894736843e-07,
"loss": 0.5204,
"step": 9750
},
{
"epoch": 0.24,
"grad_norm": 3.09375,
"learning_rate": 5.052631578947369e-07,
"loss": 0.6044,
"step": 9760
},
{
"epoch": 0.24,
"grad_norm": 3.078125,
"learning_rate": 4.842105263157895e-07,
"loss": 0.6061,
"step": 9770
},
{
"epoch": 0.24,
"grad_norm": 5.59375,
"learning_rate": 4.631578947368422e-07,
"loss": 0.5491,
"step": 9780
},
{
"epoch": 0.24,
"grad_norm": 3.46875,
"learning_rate": 4.421052631578947e-07,
"loss": 0.5567,
"step": 9790
},
{
"epoch": 0.24,
"grad_norm": 3.671875,
"learning_rate": 4.210526315789474e-07,
"loss": 0.6661,
"step": 9800
},
{
"epoch": 0.25,
"grad_norm": 2.875,
"learning_rate": 4.0000000000000003e-07,
"loss": 0.672,
"step": 9810
},
{
"epoch": 0.25,
"grad_norm": 7.15625,
"learning_rate": 3.789473684210527e-07,
"loss": 0.5887,
"step": 9820
},
{
"epoch": 0.25,
"grad_norm": 2.453125,
"learning_rate": 3.578947368421053e-07,
"loss": 0.4978,
"step": 9830
},
{
"epoch": 0.25,
"grad_norm": 2.859375,
"learning_rate": 3.368421052631579e-07,
"loss": 0.5254,
"step": 9840
},
{
"epoch": 0.25,
"grad_norm": 3.515625,
"learning_rate": 3.1578947368421055e-07,
"loss": 0.5738,
"step": 9850
},
{
"epoch": 0.25,
"grad_norm": 3.828125,
"learning_rate": 2.9473684210526315e-07,
"loss": 0.5846,
"step": 9860
},
{
"epoch": 0.25,
"grad_norm": 6.0625,
"learning_rate": 2.736842105263158e-07,
"loss": 0.6591,
"step": 9870
},
{
"epoch": 0.25,
"grad_norm": 5.0,
"learning_rate": 2.5263157894736846e-07,
"loss": 0.6697,
"step": 9880
},
{
"epoch": 0.25,
"grad_norm": 4.25,
"learning_rate": 2.315789473684211e-07,
"loss": 0.4831,
"step": 9890
},
{
"epoch": 0.25,
"grad_norm": 2.65625,
"learning_rate": 2.105263157894737e-07,
"loss": 0.5504,
"step": 9900
},
{
"epoch": 0.25,
"grad_norm": 2.65625,
"learning_rate": 1.8947368421052634e-07,
"loss": 0.5576,
"step": 9910
},
{
"epoch": 0.25,
"grad_norm": 3.875,
"learning_rate": 1.6842105263157895e-07,
"loss": 0.4733,
"step": 9920
},
{
"epoch": 0.25,
"grad_norm": 3.03125,
"learning_rate": 1.4736842105263158e-07,
"loss": 0.5077,
"step": 9930
},
{
"epoch": 0.25,
"grad_norm": 4.5,
"learning_rate": 1.2631578947368423e-07,
"loss": 0.5068,
"step": 9940
},
{
"epoch": 0.25,
"grad_norm": 2.453125,
"learning_rate": 1.0526315789473685e-07,
"loss": 0.4754,
"step": 9950
},
{
"epoch": 0.25,
"grad_norm": 6.09375,
"learning_rate": 8.421052631578947e-08,
"loss": 0.5751,
"step": 9960
},
{
"epoch": 0.25,
"grad_norm": 3.65625,
"learning_rate": 6.315789473684211e-08,
"loss": 0.5133,
"step": 9970
},
{
"epoch": 0.25,
"grad_norm": 3.171875,
"learning_rate": 4.2105263157894737e-08,
"loss": 0.6084,
"step": 9980
},
{
"epoch": 0.25,
"grad_norm": 16.5,
"learning_rate": 2.1052631578947368e-08,
"loss": 0.6159,
"step": 9990
},
{
"epoch": 0.25,
"grad_norm": 4.46875,
"learning_rate": 0.0,
"loss": 0.6108,
"step": 10000
},
{
"epoch": 0.25,
"eval_loss": 0.6137469410896301,
"eval_runtime": 38.1278,
"eval_samples_per_second": 26.228,
"eval_steps_per_second": 26.228,
"step": 10000
}
],
"logging_steps": 10,
"max_steps": 10000,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 2500,
"total_flos": 1.5733698330624e+17,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}