MeedoSam's picture
Uploaded checkpoint-2000
898ed2e verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.1,
"eval_steps": 2000,
"global_step": 2000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"grad_norm": 6.03125,
"learning_rate": 3.0000000000000004e-08,
"loss": 0.5688,
"step": 10
},
{
"epoch": 0.0,
"grad_norm": 5.5,
"learning_rate": 6.000000000000001e-08,
"loss": 0.4811,
"step": 20
},
{
"epoch": 0.0,
"grad_norm": 4.71875,
"learning_rate": 9e-08,
"loss": 0.6172,
"step": 30
},
{
"epoch": 0.0,
"grad_norm": 6.59375,
"learning_rate": 1.2000000000000002e-07,
"loss": 0.5229,
"step": 40
},
{
"epoch": 0.0,
"grad_norm": 7.125,
"learning_rate": 1.5000000000000002e-07,
"loss": 0.5498,
"step": 50
},
{
"epoch": 0.0,
"grad_norm": 4.875,
"learning_rate": 1.8e-07,
"loss": 0.6114,
"step": 60
},
{
"epoch": 0.0,
"grad_norm": 5.40625,
"learning_rate": 2.1000000000000003e-07,
"loss": 0.5094,
"step": 70
},
{
"epoch": 0.0,
"grad_norm": 6.46875,
"learning_rate": 2.4000000000000003e-07,
"loss": 0.5955,
"step": 80
},
{
"epoch": 0.0,
"grad_norm": 5.0625,
"learning_rate": 2.7e-07,
"loss": 0.5804,
"step": 90
},
{
"epoch": 0.01,
"grad_norm": 6.0625,
"learning_rate": 3.0000000000000004e-07,
"loss": 0.5598,
"step": 100
},
{
"epoch": 0.01,
"grad_norm": 5.40625,
"learning_rate": 3.3e-07,
"loss": 0.5773,
"step": 110
},
{
"epoch": 0.01,
"grad_norm": 6.6875,
"learning_rate": 3.6e-07,
"loss": 0.6036,
"step": 120
},
{
"epoch": 0.01,
"grad_norm": 8.25,
"learning_rate": 3.9e-07,
"loss": 0.5187,
"step": 130
},
{
"epoch": 0.01,
"grad_norm": 6.09375,
"learning_rate": 4.2000000000000006e-07,
"loss": 0.6931,
"step": 140
},
{
"epoch": 0.01,
"grad_norm": 4.40625,
"learning_rate": 4.5e-07,
"loss": 0.5209,
"step": 150
},
{
"epoch": 0.01,
"grad_norm": 4.375,
"learning_rate": 4.800000000000001e-07,
"loss": 0.6007,
"step": 160
},
{
"epoch": 0.01,
"grad_norm": 7.375,
"learning_rate": 5.100000000000001e-07,
"loss": 0.5626,
"step": 170
},
{
"epoch": 0.01,
"grad_norm": 5.9375,
"learning_rate": 5.4e-07,
"loss": 0.5869,
"step": 180
},
{
"epoch": 0.01,
"grad_norm": 5.53125,
"learning_rate": 5.7e-07,
"loss": 0.5639,
"step": 190
},
{
"epoch": 0.01,
"grad_norm": 4.25,
"learning_rate": 6.000000000000001e-07,
"loss": 0.5354,
"step": 200
},
{
"epoch": 0.01,
"grad_norm": 7.09375,
"learning_rate": 6.3e-07,
"loss": 0.492,
"step": 210
},
{
"epoch": 0.01,
"grad_norm": 4.9375,
"learning_rate": 6.6e-07,
"loss": 0.5569,
"step": 220
},
{
"epoch": 0.01,
"grad_norm": 6.71875,
"learning_rate": 6.900000000000001e-07,
"loss": 0.5664,
"step": 230
},
{
"epoch": 0.01,
"grad_norm": 6.375,
"learning_rate": 7.2e-07,
"loss": 0.4673,
"step": 240
},
{
"epoch": 0.01,
"grad_norm": 5.09375,
"learning_rate": 7.5e-07,
"loss": 0.6147,
"step": 250
},
{
"epoch": 0.01,
"grad_norm": 5.65625,
"learning_rate": 7.8e-07,
"loss": 0.6587,
"step": 260
},
{
"epoch": 0.01,
"grad_norm": 5.53125,
"learning_rate": 8.100000000000001e-07,
"loss": 0.5524,
"step": 270
},
{
"epoch": 0.01,
"grad_norm": 5.6875,
"learning_rate": 8.400000000000001e-07,
"loss": 0.5877,
"step": 280
},
{
"epoch": 0.01,
"grad_norm": 6.125,
"learning_rate": 8.699999999999999e-07,
"loss": 0.5573,
"step": 290
},
{
"epoch": 0.01,
"grad_norm": 5.71875,
"learning_rate": 9e-07,
"loss": 0.5665,
"step": 300
},
{
"epoch": 0.02,
"grad_norm": 4.65625,
"learning_rate": 9.3e-07,
"loss": 0.5659,
"step": 310
},
{
"epoch": 0.02,
"grad_norm": 5.5625,
"learning_rate": 9.600000000000001e-07,
"loss": 0.4814,
"step": 320
},
{
"epoch": 0.02,
"grad_norm": 4.875,
"learning_rate": 9.9e-07,
"loss": 0.5048,
"step": 330
},
{
"epoch": 0.02,
"grad_norm": 8.5625,
"learning_rate": 1.0200000000000002e-06,
"loss": 0.5021,
"step": 340
},
{
"epoch": 0.02,
"grad_norm": 6.5,
"learning_rate": 1.05e-06,
"loss": 0.5175,
"step": 350
},
{
"epoch": 0.02,
"grad_norm": 5.0625,
"learning_rate": 1.08e-06,
"loss": 0.55,
"step": 360
},
{
"epoch": 0.02,
"grad_norm": 5.6875,
"learning_rate": 1.11e-06,
"loss": 0.5565,
"step": 370
},
{
"epoch": 0.02,
"grad_norm": 6.90625,
"learning_rate": 1.14e-06,
"loss": 0.4719,
"step": 380
},
{
"epoch": 0.02,
"grad_norm": 5.59375,
"learning_rate": 1.17e-06,
"loss": 0.5117,
"step": 390
},
{
"epoch": 0.02,
"grad_norm": 5.09375,
"learning_rate": 1.2000000000000002e-06,
"loss": 0.4718,
"step": 400
},
{
"epoch": 0.02,
"grad_norm": 3.921875,
"learning_rate": 1.2299999999999999e-06,
"loss": 0.5335,
"step": 410
},
{
"epoch": 0.02,
"grad_norm": 6.21875,
"learning_rate": 1.26e-06,
"loss": 0.539,
"step": 420
},
{
"epoch": 0.02,
"grad_norm": 5.96875,
"learning_rate": 1.29e-06,
"loss": 0.4928,
"step": 430
},
{
"epoch": 0.02,
"grad_norm": 7.0625,
"learning_rate": 1.32e-06,
"loss": 0.6026,
"step": 440
},
{
"epoch": 0.02,
"grad_norm": 5.3125,
"learning_rate": 1.35e-06,
"loss": 0.5829,
"step": 450
},
{
"epoch": 0.02,
"grad_norm": 5.21875,
"learning_rate": 1.3800000000000001e-06,
"loss": 0.5167,
"step": 460
},
{
"epoch": 0.02,
"grad_norm": 4.8125,
"learning_rate": 1.41e-06,
"loss": 0.475,
"step": 470
},
{
"epoch": 0.02,
"grad_norm": 5.34375,
"learning_rate": 1.44e-06,
"loss": 0.5965,
"step": 480
},
{
"epoch": 0.02,
"grad_norm": 7.6875,
"learning_rate": 1.4700000000000001e-06,
"loss": 0.5396,
"step": 490
},
{
"epoch": 0.03,
"grad_norm": 5.53125,
"learning_rate": 1.5e-06,
"loss": 0.4778,
"step": 500
},
{
"epoch": 0.03,
"grad_norm": 4.78125,
"learning_rate": 1.498421052631579e-06,
"loss": 0.4872,
"step": 510
},
{
"epoch": 0.03,
"grad_norm": 6.375,
"learning_rate": 1.4968421052631579e-06,
"loss": 0.5117,
"step": 520
},
{
"epoch": 0.03,
"grad_norm": 5.375,
"learning_rate": 1.4952631578947368e-06,
"loss": 0.5354,
"step": 530
},
{
"epoch": 0.03,
"grad_norm": 4.875,
"learning_rate": 1.4936842105263159e-06,
"loss": 0.5136,
"step": 540
},
{
"epoch": 0.03,
"grad_norm": 6.0625,
"learning_rate": 1.4921052631578948e-06,
"loss": 0.6154,
"step": 550
},
{
"epoch": 0.03,
"grad_norm": 5.09375,
"learning_rate": 1.4905263157894737e-06,
"loss": 0.5474,
"step": 560
},
{
"epoch": 0.03,
"grad_norm": 5.625,
"learning_rate": 1.4889473684210526e-06,
"loss": 0.4899,
"step": 570
},
{
"epoch": 0.03,
"grad_norm": 5.125,
"learning_rate": 1.4873684210526315e-06,
"loss": 0.5327,
"step": 580
},
{
"epoch": 0.03,
"grad_norm": 5.625,
"learning_rate": 1.4857894736842106e-06,
"loss": 0.5615,
"step": 590
},
{
"epoch": 0.03,
"grad_norm": 4.28125,
"learning_rate": 1.4842105263157895e-06,
"loss": 0.5382,
"step": 600
},
{
"epoch": 0.03,
"grad_norm": 6.125,
"learning_rate": 1.4826315789473684e-06,
"loss": 0.4741,
"step": 610
},
{
"epoch": 0.03,
"grad_norm": 5.15625,
"learning_rate": 1.4810526315789473e-06,
"loss": 0.4843,
"step": 620
},
{
"epoch": 0.03,
"grad_norm": 5.09375,
"learning_rate": 1.4794736842105265e-06,
"loss": 0.5417,
"step": 630
},
{
"epoch": 0.03,
"grad_norm": 4.8125,
"learning_rate": 1.4778947368421054e-06,
"loss": 0.5187,
"step": 640
},
{
"epoch": 0.03,
"grad_norm": 6.21875,
"learning_rate": 1.4763157894736843e-06,
"loss": 0.5706,
"step": 650
},
{
"epoch": 0.03,
"grad_norm": 6.21875,
"learning_rate": 1.4747368421052632e-06,
"loss": 0.4604,
"step": 660
},
{
"epoch": 0.03,
"grad_norm": 6.25,
"learning_rate": 1.473157894736842e-06,
"loss": 0.5064,
"step": 670
},
{
"epoch": 0.03,
"grad_norm": 4.09375,
"learning_rate": 1.4715789473684212e-06,
"loss": 0.54,
"step": 680
},
{
"epoch": 0.03,
"grad_norm": 5.3125,
"learning_rate": 1.4700000000000001e-06,
"loss": 0.5333,
"step": 690
},
{
"epoch": 0.04,
"grad_norm": 6.09375,
"learning_rate": 1.468421052631579e-06,
"loss": 0.5975,
"step": 700
},
{
"epoch": 0.04,
"grad_norm": 6.1875,
"learning_rate": 1.466842105263158e-06,
"loss": 0.5563,
"step": 710
},
{
"epoch": 0.04,
"grad_norm": 5.1875,
"learning_rate": 1.4652631578947368e-06,
"loss": 0.4694,
"step": 720
},
{
"epoch": 0.04,
"grad_norm": 5.09375,
"learning_rate": 1.463684210526316e-06,
"loss": 0.6063,
"step": 730
},
{
"epoch": 0.04,
"grad_norm": 6.125,
"learning_rate": 1.4621052631578946e-06,
"loss": 0.4839,
"step": 740
},
{
"epoch": 0.04,
"grad_norm": 4.625,
"learning_rate": 1.4605263157894738e-06,
"loss": 0.4172,
"step": 750
},
{
"epoch": 0.04,
"grad_norm": 6.34375,
"learning_rate": 1.4589473684210527e-06,
"loss": 0.4914,
"step": 760
},
{
"epoch": 0.04,
"grad_norm": 7.4375,
"learning_rate": 1.4573684210526316e-06,
"loss": 0.4758,
"step": 770
},
{
"epoch": 0.04,
"grad_norm": 4.625,
"learning_rate": 1.4557894736842105e-06,
"loss": 0.4667,
"step": 780
},
{
"epoch": 0.04,
"grad_norm": 5.21875,
"learning_rate": 1.4542105263157894e-06,
"loss": 0.5244,
"step": 790
},
{
"epoch": 0.04,
"grad_norm": 5.625,
"learning_rate": 1.4526315789473685e-06,
"loss": 0.4719,
"step": 800
},
{
"epoch": 0.04,
"grad_norm": 6.21875,
"learning_rate": 1.4510526315789474e-06,
"loss": 0.4166,
"step": 810
},
{
"epoch": 0.04,
"grad_norm": 3.78125,
"learning_rate": 1.4494736842105263e-06,
"loss": 0.4209,
"step": 820
},
{
"epoch": 0.04,
"grad_norm": 6.46875,
"learning_rate": 1.4478947368421052e-06,
"loss": 0.4946,
"step": 830
},
{
"epoch": 0.04,
"grad_norm": 5.0625,
"learning_rate": 1.4463157894736844e-06,
"loss": 0.4399,
"step": 840
},
{
"epoch": 0.04,
"grad_norm": 4.65625,
"learning_rate": 1.4447368421052633e-06,
"loss": 0.4578,
"step": 850
},
{
"epoch": 0.04,
"grad_norm": 5.25,
"learning_rate": 1.4431578947368422e-06,
"loss": 0.5727,
"step": 860
},
{
"epoch": 0.04,
"grad_norm": 5.75,
"learning_rate": 1.441578947368421e-06,
"loss": 0.4572,
"step": 870
},
{
"epoch": 0.04,
"grad_norm": 5.03125,
"learning_rate": 1.44e-06,
"loss": 0.5346,
"step": 880
},
{
"epoch": 0.04,
"grad_norm": 5.6875,
"learning_rate": 1.438421052631579e-06,
"loss": 0.4093,
"step": 890
},
{
"epoch": 0.04,
"grad_norm": 4.875,
"learning_rate": 1.436842105263158e-06,
"loss": 0.5276,
"step": 900
},
{
"epoch": 0.05,
"grad_norm": 6.21875,
"learning_rate": 1.435263157894737e-06,
"loss": 0.4809,
"step": 910
},
{
"epoch": 0.05,
"grad_norm": 5.3125,
"learning_rate": 1.4336842105263158e-06,
"loss": 0.5037,
"step": 920
},
{
"epoch": 0.05,
"grad_norm": 3.984375,
"learning_rate": 1.4321052631578947e-06,
"loss": 0.5046,
"step": 930
},
{
"epoch": 0.05,
"grad_norm": 3.890625,
"learning_rate": 1.4305263157894738e-06,
"loss": 0.4633,
"step": 940
},
{
"epoch": 0.05,
"grad_norm": 5.0625,
"learning_rate": 1.4289473684210525e-06,
"loss": 0.3924,
"step": 950
},
{
"epoch": 0.05,
"grad_norm": 5.0,
"learning_rate": 1.4273684210526317e-06,
"loss": 0.5488,
"step": 960
},
{
"epoch": 0.05,
"grad_norm": 6.0,
"learning_rate": 1.4257894736842106e-06,
"loss": 0.5566,
"step": 970
},
{
"epoch": 0.05,
"grad_norm": 4.71875,
"learning_rate": 1.4242105263157895e-06,
"loss": 0.4756,
"step": 980
},
{
"epoch": 0.05,
"grad_norm": 4.1875,
"learning_rate": 1.4226315789473684e-06,
"loss": 0.3859,
"step": 990
},
{
"epoch": 0.05,
"grad_norm": 4.8125,
"learning_rate": 1.4210526315789473e-06,
"loss": 0.4438,
"step": 1000
},
{
"epoch": 0.05,
"grad_norm": 5.875,
"learning_rate": 1.4194736842105264e-06,
"loss": 0.5533,
"step": 1010
},
{
"epoch": 0.05,
"grad_norm": 6.09375,
"learning_rate": 1.4178947368421053e-06,
"loss": 0.5438,
"step": 1020
},
{
"epoch": 0.05,
"grad_norm": 4.59375,
"learning_rate": 1.4163157894736842e-06,
"loss": 0.574,
"step": 1030
},
{
"epoch": 0.05,
"grad_norm": 4.375,
"learning_rate": 1.4147368421052631e-06,
"loss": 0.4077,
"step": 1040
},
{
"epoch": 0.05,
"grad_norm": 5.21875,
"learning_rate": 1.4131578947368422e-06,
"loss": 0.5732,
"step": 1050
},
{
"epoch": 0.05,
"grad_norm": 5.40625,
"learning_rate": 1.4115789473684211e-06,
"loss": 0.4751,
"step": 1060
},
{
"epoch": 0.05,
"grad_norm": 4.78125,
"learning_rate": 1.41e-06,
"loss": 0.4842,
"step": 1070
},
{
"epoch": 0.05,
"grad_norm": 5.125,
"learning_rate": 1.408421052631579e-06,
"loss": 0.5282,
"step": 1080
},
{
"epoch": 0.05,
"grad_norm": 4.6875,
"learning_rate": 1.4068421052631579e-06,
"loss": 0.4528,
"step": 1090
},
{
"epoch": 0.06,
"grad_norm": 4.1875,
"learning_rate": 1.405263157894737e-06,
"loss": 0.5392,
"step": 1100
},
{
"epoch": 0.06,
"grad_norm": 3.5,
"learning_rate": 1.4036842105263159e-06,
"loss": 0.4842,
"step": 1110
},
{
"epoch": 0.06,
"grad_norm": 5.3125,
"learning_rate": 1.4021052631578948e-06,
"loss": 0.4341,
"step": 1120
},
{
"epoch": 0.06,
"grad_norm": 4.84375,
"learning_rate": 1.4005263157894737e-06,
"loss": 0.4241,
"step": 1130
},
{
"epoch": 0.06,
"grad_norm": 4.96875,
"learning_rate": 1.3989473684210526e-06,
"loss": 0.4254,
"step": 1140
},
{
"epoch": 0.06,
"grad_norm": 5.8125,
"learning_rate": 1.3973684210526317e-06,
"loss": 0.5204,
"step": 1150
},
{
"epoch": 0.06,
"grad_norm": 5.15625,
"learning_rate": 1.3957894736842104e-06,
"loss": 0.5232,
"step": 1160
},
{
"epoch": 0.06,
"grad_norm": 5.5625,
"learning_rate": 1.3942105263157895e-06,
"loss": 0.4818,
"step": 1170
},
{
"epoch": 0.06,
"grad_norm": 4.15625,
"learning_rate": 1.3926315789473685e-06,
"loss": 0.5224,
"step": 1180
},
{
"epoch": 0.06,
"grad_norm": 5.53125,
"learning_rate": 1.3910526315789476e-06,
"loss": 0.5449,
"step": 1190
},
{
"epoch": 0.06,
"grad_norm": 5.53125,
"learning_rate": 1.3894736842105263e-06,
"loss": 0.5367,
"step": 1200
},
{
"epoch": 0.06,
"grad_norm": 5.71875,
"learning_rate": 1.3878947368421052e-06,
"loss": 0.5962,
"step": 1210
},
{
"epoch": 0.06,
"grad_norm": 4.125,
"learning_rate": 1.3863157894736843e-06,
"loss": 0.3955,
"step": 1220
},
{
"epoch": 0.06,
"grad_norm": 4.96875,
"learning_rate": 1.3847368421052632e-06,
"loss": 0.5646,
"step": 1230
},
{
"epoch": 0.06,
"grad_norm": 4.84375,
"learning_rate": 1.383157894736842e-06,
"loss": 0.515,
"step": 1240
},
{
"epoch": 0.06,
"grad_norm": 4.65625,
"learning_rate": 1.381578947368421e-06,
"loss": 0.4916,
"step": 1250
},
{
"epoch": 0.06,
"grad_norm": 4.34375,
"learning_rate": 1.3800000000000001e-06,
"loss": 0.4971,
"step": 1260
},
{
"epoch": 0.06,
"grad_norm": 5.5625,
"learning_rate": 1.378421052631579e-06,
"loss": 0.5121,
"step": 1270
},
{
"epoch": 0.06,
"grad_norm": 6.84375,
"learning_rate": 1.376842105263158e-06,
"loss": 0.4628,
"step": 1280
},
{
"epoch": 0.06,
"grad_norm": 5.6875,
"learning_rate": 1.3752631578947368e-06,
"loss": 0.4309,
"step": 1290
},
{
"epoch": 0.07,
"grad_norm": 4.09375,
"learning_rate": 1.3736842105263158e-06,
"loss": 0.4871,
"step": 1300
},
{
"epoch": 0.07,
"grad_norm": 4.5,
"learning_rate": 1.3721052631578949e-06,
"loss": 0.5438,
"step": 1310
},
{
"epoch": 0.07,
"grad_norm": 4.03125,
"learning_rate": 1.3705263157894738e-06,
"loss": 0.4459,
"step": 1320
},
{
"epoch": 0.07,
"grad_norm": 4.78125,
"learning_rate": 1.3689473684210527e-06,
"loss": 0.4734,
"step": 1330
},
{
"epoch": 0.07,
"grad_norm": 6.125,
"learning_rate": 1.3673684210526316e-06,
"loss": 0.5321,
"step": 1340
},
{
"epoch": 0.07,
"grad_norm": 5.59375,
"learning_rate": 1.3657894736842107e-06,
"loss": 0.4859,
"step": 1350
},
{
"epoch": 0.07,
"grad_norm": 5.59375,
"learning_rate": 1.3642105263157896e-06,
"loss": 0.4804,
"step": 1360
},
{
"epoch": 0.07,
"grad_norm": 5.21875,
"learning_rate": 1.3626315789473683e-06,
"loss": 0.5041,
"step": 1370
},
{
"epoch": 0.07,
"grad_norm": 6.71875,
"learning_rate": 1.3610526315789474e-06,
"loss": 0.5103,
"step": 1380
},
{
"epoch": 0.07,
"grad_norm": 5.84375,
"learning_rate": 1.3594736842105263e-06,
"loss": 0.4737,
"step": 1390
},
{
"epoch": 0.07,
"grad_norm": 6.125,
"learning_rate": 1.3578947368421055e-06,
"loss": 0.4954,
"step": 1400
},
{
"epoch": 0.07,
"grad_norm": 5.78125,
"learning_rate": 1.3563157894736842e-06,
"loss": 0.5358,
"step": 1410
},
{
"epoch": 0.07,
"grad_norm": 5.625,
"learning_rate": 1.3547368421052633e-06,
"loss": 0.4667,
"step": 1420
},
{
"epoch": 0.07,
"grad_norm": 5.84375,
"learning_rate": 1.3531578947368422e-06,
"loss": 0.4511,
"step": 1430
},
{
"epoch": 0.07,
"grad_norm": 4.875,
"learning_rate": 1.351578947368421e-06,
"loss": 0.4853,
"step": 1440
},
{
"epoch": 0.07,
"grad_norm": 7.34375,
"learning_rate": 1.35e-06,
"loss": 0.4906,
"step": 1450
},
{
"epoch": 0.07,
"grad_norm": 3.515625,
"learning_rate": 1.348421052631579e-06,
"loss": 0.4553,
"step": 1460
},
{
"epoch": 0.07,
"grad_norm": 5.1875,
"learning_rate": 1.346842105263158e-06,
"loss": 0.4767,
"step": 1470
},
{
"epoch": 0.07,
"grad_norm": 4.84375,
"learning_rate": 1.345263157894737e-06,
"loss": 0.5855,
"step": 1480
},
{
"epoch": 0.07,
"grad_norm": 5.90625,
"learning_rate": 1.3436842105263158e-06,
"loss": 0.548,
"step": 1490
},
{
"epoch": 0.07,
"grad_norm": 5.0625,
"learning_rate": 1.3421052631578947e-06,
"loss": 0.4498,
"step": 1500
},
{
"epoch": 0.08,
"grad_norm": 5.09375,
"learning_rate": 1.3405263157894736e-06,
"loss": 0.453,
"step": 1510
},
{
"epoch": 0.08,
"grad_norm": 5.3125,
"learning_rate": 1.3389473684210528e-06,
"loss": 0.4706,
"step": 1520
},
{
"epoch": 0.08,
"grad_norm": 3.984375,
"learning_rate": 1.3373684210526317e-06,
"loss": 0.492,
"step": 1530
},
{
"epoch": 0.08,
"grad_norm": 4.8125,
"learning_rate": 1.3357894736842106e-06,
"loss": 0.5406,
"step": 1540
},
{
"epoch": 0.08,
"grad_norm": 5.3125,
"learning_rate": 1.3342105263157895e-06,
"loss": 0.4736,
"step": 1550
},
{
"epoch": 0.08,
"grad_norm": 5.5625,
"learning_rate": 1.3326315789473686e-06,
"loss": 0.6124,
"step": 1560
},
{
"epoch": 0.08,
"grad_norm": 4.96875,
"learning_rate": 1.3310526315789475e-06,
"loss": 0.4797,
"step": 1570
},
{
"epoch": 0.08,
"grad_norm": 7.65625,
"learning_rate": 1.3294736842105262e-06,
"loss": 0.5659,
"step": 1580
},
{
"epoch": 0.08,
"grad_norm": 4.53125,
"learning_rate": 1.3278947368421053e-06,
"loss": 0.4673,
"step": 1590
},
{
"epoch": 0.08,
"grad_norm": 5.625,
"learning_rate": 1.3263157894736842e-06,
"loss": 0.5162,
"step": 1600
},
{
"epoch": 0.08,
"grad_norm": 4.96875,
"learning_rate": 1.3247368421052633e-06,
"loss": 0.4925,
"step": 1610
},
{
"epoch": 0.08,
"grad_norm": 4.0625,
"learning_rate": 1.323157894736842e-06,
"loss": 0.4602,
"step": 1620
},
{
"epoch": 0.08,
"grad_norm": 4.03125,
"learning_rate": 1.3215789473684212e-06,
"loss": 0.4743,
"step": 1630
},
{
"epoch": 0.08,
"grad_norm": 5.5625,
"learning_rate": 1.32e-06,
"loss": 0.4944,
"step": 1640
},
{
"epoch": 0.08,
"grad_norm": 5.625,
"learning_rate": 1.318421052631579e-06,
"loss": 0.5525,
"step": 1650
},
{
"epoch": 0.08,
"grad_norm": 4.90625,
"learning_rate": 1.3168421052631579e-06,
"loss": 0.4512,
"step": 1660
},
{
"epoch": 0.08,
"grad_norm": 4.90625,
"learning_rate": 1.3152631578947368e-06,
"loss": 0.4107,
"step": 1670
},
{
"epoch": 0.08,
"grad_norm": 3.984375,
"learning_rate": 1.313684210526316e-06,
"loss": 0.3901,
"step": 1680
},
{
"epoch": 0.08,
"grad_norm": 5.59375,
"learning_rate": 1.3121052631578948e-06,
"loss": 0.4924,
"step": 1690
},
{
"epoch": 0.09,
"grad_norm": 5.84375,
"learning_rate": 1.3105263157894737e-06,
"loss": 0.6272,
"step": 1700
},
{
"epoch": 0.09,
"grad_norm": 6.4375,
"learning_rate": 1.3089473684210526e-06,
"loss": 0.4816,
"step": 1710
},
{
"epoch": 0.09,
"grad_norm": 5.09375,
"learning_rate": 1.3073684210526315e-06,
"loss": 0.4654,
"step": 1720
},
{
"epoch": 0.09,
"grad_norm": 4.96875,
"learning_rate": 1.3057894736842107e-06,
"loss": 0.3862,
"step": 1730
},
{
"epoch": 0.09,
"grad_norm": 4.65625,
"learning_rate": 1.3042105263157896e-06,
"loss": 0.491,
"step": 1740
},
{
"epoch": 0.09,
"grad_norm": 5.84375,
"learning_rate": 1.3026315789473685e-06,
"loss": 0.4092,
"step": 1750
},
{
"epoch": 0.09,
"grad_norm": 4.5625,
"learning_rate": 1.3010526315789474e-06,
"loss": 0.43,
"step": 1760
},
{
"epoch": 0.09,
"grad_norm": 5.28125,
"learning_rate": 1.2994736842105265e-06,
"loss": 0.4754,
"step": 1770
},
{
"epoch": 0.09,
"grad_norm": 4.71875,
"learning_rate": 1.2978947368421054e-06,
"loss": 0.5338,
"step": 1780
},
{
"epoch": 0.09,
"grad_norm": 6.3125,
"learning_rate": 1.296315789473684e-06,
"loss": 0.4727,
"step": 1790
},
{
"epoch": 0.09,
"grad_norm": 4.03125,
"learning_rate": 1.2947368421052632e-06,
"loss": 0.485,
"step": 1800
},
{
"epoch": 0.09,
"grad_norm": 4.3125,
"learning_rate": 1.2931578947368421e-06,
"loss": 0.4643,
"step": 1810
},
{
"epoch": 0.09,
"grad_norm": 4.28125,
"learning_rate": 1.2915789473684212e-06,
"loss": 0.5617,
"step": 1820
},
{
"epoch": 0.09,
"grad_norm": 4.875,
"learning_rate": 1.29e-06,
"loss": 0.4752,
"step": 1830
},
{
"epoch": 0.09,
"grad_norm": 4.6875,
"learning_rate": 1.288421052631579e-06,
"loss": 0.4927,
"step": 1840
},
{
"epoch": 0.09,
"grad_norm": 7.9375,
"learning_rate": 1.286842105263158e-06,
"loss": 0.5522,
"step": 1850
},
{
"epoch": 0.09,
"grad_norm": 4.53125,
"learning_rate": 1.2852631578947369e-06,
"loss": 0.5096,
"step": 1860
},
{
"epoch": 0.09,
"grad_norm": 3.390625,
"learning_rate": 1.2836842105263158e-06,
"loss": 0.5059,
"step": 1870
},
{
"epoch": 0.09,
"grad_norm": 5.0,
"learning_rate": 1.2821052631578947e-06,
"loss": 0.5205,
"step": 1880
},
{
"epoch": 0.09,
"grad_norm": 5.0625,
"learning_rate": 1.2805263157894738e-06,
"loss": 0.5575,
"step": 1890
},
{
"epoch": 0.1,
"grad_norm": 5.59375,
"learning_rate": 1.2789473684210527e-06,
"loss": 0.4836,
"step": 1900
},
{
"epoch": 0.1,
"grad_norm": 6.53125,
"learning_rate": 1.2773684210526316e-06,
"loss": 0.4563,
"step": 1910
},
{
"epoch": 0.1,
"grad_norm": 5.8125,
"learning_rate": 1.2757894736842105e-06,
"loss": 0.4431,
"step": 1920
},
{
"epoch": 0.1,
"grad_norm": 5.21875,
"learning_rate": 1.2742105263157894e-06,
"loss": 0.4686,
"step": 1930
},
{
"epoch": 0.1,
"grad_norm": 4.96875,
"learning_rate": 1.2726315789473685e-06,
"loss": 0.4486,
"step": 1940
},
{
"epoch": 0.1,
"grad_norm": 4.875,
"learning_rate": 1.2710526315789474e-06,
"loss": 0.4044,
"step": 1950
},
{
"epoch": 0.1,
"grad_norm": 5.03125,
"learning_rate": 1.2694736842105264e-06,
"loss": 0.5884,
"step": 1960
},
{
"epoch": 0.1,
"grad_norm": 4.5,
"learning_rate": 1.2678947368421053e-06,
"loss": 0.4116,
"step": 1970
},
{
"epoch": 0.1,
"grad_norm": 5.0625,
"learning_rate": 1.2663157894736844e-06,
"loss": 0.404,
"step": 1980
},
{
"epoch": 0.1,
"grad_norm": 4.03125,
"learning_rate": 1.2647368421052633e-06,
"loss": 0.4443,
"step": 1990
},
{
"epoch": 0.1,
"grad_norm": 5.0625,
"learning_rate": 1.263157894736842e-06,
"loss": 0.5572,
"step": 2000
},
{
"epoch": 0.1,
"eval_loss": 0.4979419410228729,
"eval_runtime": 64.9548,
"eval_samples_per_second": 15.395,
"eval_steps_per_second": 15.395,
"step": 2000
}
],
"logging_steps": 10,
"max_steps": 10000,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 2000,
"total_flos": 3.227844083712e+16,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}