5174 lines
108 KiB
JSON
5174 lines
108 KiB
JSON
{
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 1.9992927864214993,
|
|
"eval_steps": 35,
|
|
"global_step": 712,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"epoch": 0.0,
|
|
"grad_norm": 51.75,
|
|
"learning_rate": 8.571428571428572e-08,
|
|
"loss": 1.9845,
|
|
"step": 1
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"eval_loss": 2.040191650390625,
|
|
"eval_runtime": 20.1394,
|
|
"eval_samples_per_second": 64.997,
|
|
"eval_steps_per_second": 64.997,
|
|
"step": 1
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"grad_norm": 48.75,
|
|
"learning_rate": 1.7142857142857143e-07,
|
|
"loss": 1.9423,
|
|
"step": 2
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"grad_norm": 45.5,
|
|
"learning_rate": 2.5714285714285716e-07,
|
|
"loss": 1.9446,
|
|
"step": 3
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"grad_norm": 56.25,
|
|
"learning_rate": 3.4285714285714286e-07,
|
|
"loss": 2.1833,
|
|
"step": 4
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"grad_norm": 55.25,
|
|
"learning_rate": 4.2857142857142857e-07,
|
|
"loss": 2.0796,
|
|
"step": 5
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"grad_norm": 49.75,
|
|
"learning_rate": 5.142857142857143e-07,
|
|
"loss": 1.9704,
|
|
"step": 6
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"grad_norm": 53.0,
|
|
"learning_rate": 6.000000000000001e-07,
|
|
"loss": 1.9781,
|
|
"step": 7
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"grad_norm": 49.0,
|
|
"learning_rate": 6.857142857142857e-07,
|
|
"loss": 1.9854,
|
|
"step": 8
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"grad_norm": 51.5,
|
|
"learning_rate": 7.714285714285714e-07,
|
|
"loss": 1.927,
|
|
"step": 9
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"grad_norm": 51.5,
|
|
"learning_rate": 8.571428571428571e-07,
|
|
"loss": 2.0879,
|
|
"step": 10
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"grad_norm": 48.5,
|
|
"learning_rate": 9.428571428571429e-07,
|
|
"loss": 1.8861,
|
|
"step": 11
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"grad_norm": 49.5,
|
|
"learning_rate": 1.0285714285714286e-06,
|
|
"loss": 1.9962,
|
|
"step": 12
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"grad_norm": 42.5,
|
|
"learning_rate": 1.1142857142857143e-06,
|
|
"loss": 1.904,
|
|
"step": 13
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"grad_norm": 38.0,
|
|
"learning_rate": 1.2000000000000002e-06,
|
|
"loss": 1.8107,
|
|
"step": 14
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"grad_norm": 37.25,
|
|
"learning_rate": 1.2857142857142856e-06,
|
|
"loss": 1.8908,
|
|
"step": 15
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"grad_norm": 35.75,
|
|
"learning_rate": 1.3714285714285715e-06,
|
|
"loss": 1.9207,
|
|
"step": 16
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"grad_norm": 32.25,
|
|
"learning_rate": 1.4571428571428571e-06,
|
|
"loss": 1.8448,
|
|
"step": 17
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"grad_norm": 28.625,
|
|
"learning_rate": 1.5428571428571428e-06,
|
|
"loss": 1.7089,
|
|
"step": 18
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"grad_norm": 25.5,
|
|
"learning_rate": 1.6285714285714284e-06,
|
|
"loss": 1.6403,
|
|
"step": 19
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"grad_norm": 23.75,
|
|
"learning_rate": 1.7142857142857143e-06,
|
|
"loss": 1.867,
|
|
"step": 20
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"grad_norm": 21.875,
|
|
"learning_rate": 1.8e-06,
|
|
"loss": 1.6772,
|
|
"step": 21
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"grad_norm": 20.0,
|
|
"learning_rate": 1.8857142857142858e-06,
|
|
"loss": 1.4738,
|
|
"step": 22
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"grad_norm": 19.375,
|
|
"learning_rate": 1.9714285714285714e-06,
|
|
"loss": 1.584,
|
|
"step": 23
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"grad_norm": 18.25,
|
|
"learning_rate": 2.0571428571428573e-06,
|
|
"loss": 1.5077,
|
|
"step": 24
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"grad_norm": 15.625,
|
|
"learning_rate": 2.142857142857143e-06,
|
|
"loss": 1.3541,
|
|
"step": 25
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"grad_norm": 17.375,
|
|
"learning_rate": 2.2285714285714286e-06,
|
|
"loss": 1.4149,
|
|
"step": 26
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"grad_norm": 14.25,
|
|
"learning_rate": 2.3142857142857145e-06,
|
|
"loss": 1.2507,
|
|
"step": 27
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"grad_norm": 13.125,
|
|
"learning_rate": 2.4000000000000003e-06,
|
|
"loss": 1.2992,
|
|
"step": 28
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"grad_norm": 10.75,
|
|
"learning_rate": 2.4857142857142858e-06,
|
|
"loss": 1.2994,
|
|
"step": 29
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"grad_norm": 9.1875,
|
|
"learning_rate": 2.571428571428571e-06,
|
|
"loss": 1.2327,
|
|
"step": 30
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"grad_norm": 9.3125,
|
|
"learning_rate": 2.657142857142857e-06,
|
|
"loss": 1.3238,
|
|
"step": 31
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"grad_norm": 5.71875,
|
|
"learning_rate": 2.742857142857143e-06,
|
|
"loss": 1.2781,
|
|
"step": 32
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"grad_norm": 6.5,
|
|
"learning_rate": 2.8285714285714288e-06,
|
|
"loss": 1.2028,
|
|
"step": 33
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"grad_norm": 5.96875,
|
|
"learning_rate": 2.9142857142857142e-06,
|
|
"loss": 1.3073,
|
|
"step": 34
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"grad_norm": 4.875,
|
|
"learning_rate": 3e-06,
|
|
"loss": 1.1666,
|
|
"step": 35
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"eval_loss": 1.2127766609191895,
|
|
"eval_runtime": 20.1589,
|
|
"eval_samples_per_second": 64.934,
|
|
"eval_steps_per_second": 64.934,
|
|
"step": 35
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"grad_norm": 4.71875,
|
|
"learning_rate": 3.0857142857142855e-06,
|
|
"loss": 1.1807,
|
|
"step": 36
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"grad_norm": 4.65625,
|
|
"learning_rate": 3.1714285714285714e-06,
|
|
"loss": 1.178,
|
|
"step": 37
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 3.257142857142857e-06,
|
|
"loss": 1.08,
|
|
"step": 38
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 3.342857142857143e-06,
|
|
"loss": 1.2125,
|
|
"step": 39
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 3.4285714285714285e-06,
|
|
"loss": 1.2026,
|
|
"step": 40
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"grad_norm": 3.875,
|
|
"learning_rate": 3.5142857142857144e-06,
|
|
"loss": 1.1884,
|
|
"step": 41
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"grad_norm": 3.875,
|
|
"learning_rate": 3.6e-06,
|
|
"loss": 1.3009,
|
|
"step": 42
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 3.685714285714286e-06,
|
|
"loss": 1.0866,
|
|
"step": 43
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 3.7714285714285716e-06,
|
|
"loss": 1.1396,
|
|
"step": 44
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 3.857142857142858e-06,
|
|
"loss": 1.0933,
|
|
"step": 45
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 3.942857142857143e-06,
|
|
"loss": 1.0827,
|
|
"step": 46
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 4.028571428571429e-06,
|
|
"loss": 1.1054,
|
|
"step": 47
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"grad_norm": 3.453125,
|
|
"learning_rate": 4.114285714285715e-06,
|
|
"loss": 1.1369,
|
|
"step": 48
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"grad_norm": 3.265625,
|
|
"learning_rate": 4.2e-06,
|
|
"loss": 1.086,
|
|
"step": 49
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"grad_norm": 3.46875,
|
|
"learning_rate": 4.285714285714286e-06,
|
|
"loss": 1.0688,
|
|
"step": 50
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"grad_norm": 3.421875,
|
|
"learning_rate": 4.371428571428571e-06,
|
|
"loss": 1.0986,
|
|
"step": 51
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"grad_norm": 2.84375,
|
|
"learning_rate": 4.457142857142857e-06,
|
|
"loss": 1.0268,
|
|
"step": 52
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"grad_norm": 2.796875,
|
|
"learning_rate": 4.542857142857143e-06,
|
|
"loss": 1.0134,
|
|
"step": 53
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"grad_norm": 2.625,
|
|
"learning_rate": 4.628571428571429e-06,
|
|
"loss": 1.0973,
|
|
"step": 54
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"grad_norm": 2.59375,
|
|
"learning_rate": 4.714285714285714e-06,
|
|
"loss": 1.1536,
|
|
"step": 55
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"grad_norm": 2.203125,
|
|
"learning_rate": 4.800000000000001e-06,
|
|
"loss": 1.034,
|
|
"step": 56
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"grad_norm": 2.265625,
|
|
"learning_rate": 4.885714285714286e-06,
|
|
"loss": 1.1263,
|
|
"step": 57
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"grad_norm": 2.0,
|
|
"learning_rate": 4.9714285714285715e-06,
|
|
"loss": 1.1458,
|
|
"step": 58
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"grad_norm": 2.125,
|
|
"learning_rate": 5.057142857142857e-06,
|
|
"loss": 1.0678,
|
|
"step": 59
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"grad_norm": 2.046875,
|
|
"learning_rate": 5.142857142857142e-06,
|
|
"loss": 1.0277,
|
|
"step": 60
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"grad_norm": 1.84375,
|
|
"learning_rate": 5.228571428571429e-06,
|
|
"loss": 1.0104,
|
|
"step": 61
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"grad_norm": 1.9609375,
|
|
"learning_rate": 5.314285714285714e-06,
|
|
"loss": 1.2226,
|
|
"step": 62
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"grad_norm": 1.9296875,
|
|
"learning_rate": 5.4e-06,
|
|
"loss": 1.1612,
|
|
"step": 63
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"grad_norm": 1.875,
|
|
"learning_rate": 5.485714285714286e-06,
|
|
"loss": 1.0489,
|
|
"step": 64
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"grad_norm": 1.7734375,
|
|
"learning_rate": 5.571428571428572e-06,
|
|
"loss": 1.0256,
|
|
"step": 65
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"grad_norm": 1.7890625,
|
|
"learning_rate": 5.6571428571428576e-06,
|
|
"loss": 1.0704,
|
|
"step": 66
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"grad_norm": 1.9140625,
|
|
"learning_rate": 5.742857142857143e-06,
|
|
"loss": 1.0114,
|
|
"step": 67
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"grad_norm": 1.90625,
|
|
"learning_rate": 5.8285714285714284e-06,
|
|
"loss": 0.999,
|
|
"step": 68
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"grad_norm": 1.8359375,
|
|
"learning_rate": 5.914285714285715e-06,
|
|
"loss": 1.0967,
|
|
"step": 69
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"grad_norm": 1.8203125,
|
|
"learning_rate": 6e-06,
|
|
"loss": 1.0044,
|
|
"step": 70
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"eval_loss": 1.0669639110565186,
|
|
"eval_runtime": 20.1538,
|
|
"eval_samples_per_second": 64.951,
|
|
"eval_steps_per_second": 64.951,
|
|
"step": 70
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"grad_norm": 1.921875,
|
|
"learning_rate": 5.999986378009738e-06,
|
|
"loss": 0.9961,
|
|
"step": 71
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"grad_norm": 1.9375,
|
|
"learning_rate": 5.9999455121763995e-06,
|
|
"loss": 1.141,
|
|
"step": 72
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"grad_norm": 1.90625,
|
|
"learning_rate": 5.999877402912338e-06,
|
|
"loss": 0.976,
|
|
"step": 73
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"grad_norm": 1.828125,
|
|
"learning_rate": 5.999782050904801e-06,
|
|
"loss": 0.998,
|
|
"step": 74
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"grad_norm": 1.890625,
|
|
"learning_rate": 5.9996594571159225e-06,
|
|
"loss": 1.0277,
|
|
"step": 75
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"grad_norm": 2.03125,
|
|
"learning_rate": 5.9995096227827206e-06,
|
|
"loss": 1.0644,
|
|
"step": 76
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"grad_norm": 1.9921875,
|
|
"learning_rate": 5.999332549417079e-06,
|
|
"loss": 1.0319,
|
|
"step": 77
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"grad_norm": 1.875,
|
|
"learning_rate": 5.999128238805728e-06,
|
|
"loss": 1.1474,
|
|
"step": 78
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"grad_norm": 1.9296875,
|
|
"learning_rate": 5.99889669301024e-06,
|
|
"loss": 1.0158,
|
|
"step": 79
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"grad_norm": 2.234375,
|
|
"learning_rate": 5.998637914366994e-06,
|
|
"loss": 1.0571,
|
|
"step": 80
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"grad_norm": 2.09375,
|
|
"learning_rate": 5.99835190548716e-06,
|
|
"loss": 1.1034,
|
|
"step": 81
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"grad_norm": 2.234375,
|
|
"learning_rate": 5.998038669256674e-06,
|
|
"loss": 1.0618,
|
|
"step": 82
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"grad_norm": 2.03125,
|
|
"learning_rate": 5.997698208836202e-06,
|
|
"loss": 0.9915,
|
|
"step": 83
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"grad_norm": 2.515625,
|
|
"learning_rate": 5.997330527661113e-06,
|
|
"loss": 1.0523,
|
|
"step": 84
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"grad_norm": 2.21875,
|
|
"learning_rate": 5.996935629441444e-06,
|
|
"loss": 0.9552,
|
|
"step": 85
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"grad_norm": 2.765625,
|
|
"learning_rate": 5.996513518161861e-06,
|
|
"loss": 1.0611,
|
|
"step": 86
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"grad_norm": 2.765625,
|
|
"learning_rate": 5.996064198081619e-06,
|
|
"loss": 0.9936,
|
|
"step": 87
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"grad_norm": 3.046875,
|
|
"learning_rate": 5.9955876737345236e-06,
|
|
"loss": 1.0302,
|
|
"step": 88
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"grad_norm": 3.0625,
|
|
"learning_rate": 5.995083949928876e-06,
|
|
"loss": 1.043,
|
|
"step": 89
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"grad_norm": 3.484375,
|
|
"learning_rate": 5.994553031747435e-06,
|
|
"loss": 1.0136,
|
|
"step": 90
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"grad_norm": 3.40625,
|
|
"learning_rate": 5.993994924547356e-06,
|
|
"loss": 1.1077,
|
|
"step": 91
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"grad_norm": 3.875,
|
|
"learning_rate": 5.993409633960142e-06,
|
|
"loss": 0.9621,
|
|
"step": 92
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"grad_norm": 3.375,
|
|
"learning_rate": 5.99279716589159e-06,
|
|
"loss": 1.0648,
|
|
"step": 93
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"grad_norm": 3.171875,
|
|
"learning_rate": 5.992157526521724e-06,
|
|
"loss": 0.9725,
|
|
"step": 94
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"grad_norm": 3.109375,
|
|
"learning_rate": 5.991490722304738e-06,
|
|
"loss": 1.0358,
|
|
"step": 95
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"grad_norm": 3.03125,
|
|
"learning_rate": 5.990796759968929e-06,
|
|
"loss": 1.1569,
|
|
"step": 96
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"grad_norm": 3.03125,
|
|
"learning_rate": 5.990075646516628e-06,
|
|
"loss": 1.0382,
|
|
"step": 97
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"grad_norm": 4.34375,
|
|
"learning_rate": 5.989327389224132e-06,
|
|
"loss": 0.9393,
|
|
"step": 98
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"grad_norm": 4.6875,
|
|
"learning_rate": 5.988551995641629e-06,
|
|
"loss": 1.0306,
|
|
"step": 99
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 5.987749473593123e-06,
|
|
"loss": 1.092,
|
|
"step": 100
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"grad_norm": 3.625,
|
|
"learning_rate": 5.98691983117635e-06,
|
|
"loss": 0.962,
|
|
"step": 101
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"grad_norm": 2.984375,
|
|
"learning_rate": 5.986063076762707e-06,
|
|
"loss": 0.941,
|
|
"step": 102
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"grad_norm": 2.125,
|
|
"learning_rate": 5.985179218997154e-06,
|
|
"loss": 0.9805,
|
|
"step": 103
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"grad_norm": 1.59375,
|
|
"learning_rate": 5.9842682667981374e-06,
|
|
"loss": 0.938,
|
|
"step": 104
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 5.983330229357497e-06,
|
|
"loss": 1.0686,
|
|
"step": 105
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"eval_loss": 1.0165659189224243,
|
|
"eval_runtime": 20.3105,
|
|
"eval_samples_per_second": 64.449,
|
|
"eval_steps_per_second": 64.449,
|
|
"step": 105
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 5.982365116140371e-06,
|
|
"loss": 1.1855,
|
|
"step": 106
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"grad_norm": 1.4921875,
|
|
"learning_rate": 5.981372936885102e-06,
|
|
"loss": 0.9464,
|
|
"step": 107
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"grad_norm": 1.609375,
|
|
"learning_rate": 5.980353701603138e-06,
|
|
"loss": 0.9249,
|
|
"step": 108
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 5.979307420578934e-06,
|
|
"loss": 0.9662,
|
|
"step": 109
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 5.978234104369846e-06,
|
|
"loss": 0.9772,
|
|
"step": 110
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 5.9771337638060235e-06,
|
|
"loss": 1.0412,
|
|
"step": 111
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 5.9760064099903025e-06,
|
|
"loss": 1.0262,
|
|
"step": 112
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 5.974852054298093e-06,
|
|
"loss": 0.9824,
|
|
"step": 113
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 5.973670708377263e-06,
|
|
"loss": 0.9425,
|
|
"step": 114
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 5.972462384148021e-06,
|
|
"loss": 1.0243,
|
|
"step": 115
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"grad_norm": 2.796875,
|
|
"learning_rate": 5.971227093802798e-06,
|
|
"loss": 0.9326,
|
|
"step": 116
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"grad_norm": 1.7109375,
|
|
"learning_rate": 5.9699648498061216e-06,
|
|
"loss": 1.0175,
|
|
"step": 117
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 5.968675664894493e-06,
|
|
"loss": 1.013,
|
|
"step": 118
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"grad_norm": 1.5390625,
|
|
"learning_rate": 5.967359552076256e-06,
|
|
"loss": 1.0109,
|
|
"step": 119
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"grad_norm": 1.5078125,
|
|
"learning_rate": 5.966016524631466e-06,
|
|
"loss": 0.9164,
|
|
"step": 120
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"grad_norm": 1.4921875,
|
|
"learning_rate": 5.964646596111758e-06,
|
|
"loss": 1.0451,
|
|
"step": 121
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"grad_norm": 1.546875,
|
|
"learning_rate": 5.9632497803402104e-06,
|
|
"loss": 0.9873,
|
|
"step": 122
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"grad_norm": 1.484375,
|
|
"learning_rate": 5.961826091411198e-06,
|
|
"loss": 0.9989,
|
|
"step": 123
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"grad_norm": 1.5390625,
|
|
"learning_rate": 5.960375543690261e-06,
|
|
"loss": 1.1006,
|
|
"step": 124
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"grad_norm": 1.546875,
|
|
"learning_rate": 5.9588981518139525e-06,
|
|
"loss": 1.0773,
|
|
"step": 125
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 5.957393930689694e-06,
|
|
"loss": 0.905,
|
|
"step": 126
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 5.955862895495621e-06,
|
|
"loss": 0.9727,
|
|
"step": 127
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 5.9543050616804355e-06,
|
|
"loss": 1.0183,
|
|
"step": 128
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 5.952720444963247e-06,
|
|
"loss": 1.008,
|
|
"step": 129
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"grad_norm": 1.6171875,
|
|
"learning_rate": 5.951109061333413e-06,
|
|
"loss": 0.9817,
|
|
"step": 130
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"grad_norm": 1.6171875,
|
|
"learning_rate": 5.94947092705038e-06,
|
|
"loss": 1.0044,
|
|
"step": 131
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"grad_norm": 1.5546875,
|
|
"learning_rate": 5.947806058643517e-06,
|
|
"loss": 1.0978,
|
|
"step": 132
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 5.946114472911952e-06,
|
|
"loss": 0.9711,
|
|
"step": 133
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"grad_norm": 1.5390625,
|
|
"learning_rate": 5.944396186924398e-06,
|
|
"loss": 1.0369,
|
|
"step": 134
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 5.942651218018985e-06,
|
|
"loss": 1.0105,
|
|
"step": 135
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"grad_norm": 1.5078125,
|
|
"learning_rate": 5.940879583803084e-06,
|
|
"loss": 1.0037,
|
|
"step": 136
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"grad_norm": 1.609375,
|
|
"learning_rate": 5.939081302153127e-06,
|
|
"loss": 1.0426,
|
|
"step": 137
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 5.937256391214429e-06,
|
|
"loss": 1.0015,
|
|
"step": 138
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"grad_norm": 1.5625,
|
|
"learning_rate": 5.935404869401004e-06,
|
|
"loss": 0.9964,
|
|
"step": 139
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 5.933526755395378e-06,
|
|
"loss": 1.0053,
|
|
"step": 140
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"eval_loss": 1.0025994777679443,
|
|
"eval_runtime": 20.1779,
|
|
"eval_samples_per_second": 64.873,
|
|
"eval_steps_per_second": 64.873,
|
|
"step": 140
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"grad_norm": 1.3515625,
|
|
"learning_rate": 5.931622068148405e-06,
|
|
"loss": 1.0178,
|
|
"step": 141
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 5.929690826879072e-06,
|
|
"loss": 0.9558,
|
|
"step": 142
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 5.927733051074303e-06,
|
|
"loss": 0.9541,
|
|
"step": 143
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 5.925748760488767e-06,
|
|
"loss": 0.9655,
|
|
"step": 144
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"grad_norm": 1.71875,
|
|
"learning_rate": 5.923737975144677e-06,
|
|
"loss": 0.9171,
|
|
"step": 145
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 5.921700715331588e-06,
|
|
"loss": 1.0064,
|
|
"step": 146
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"grad_norm": 1.78125,
|
|
"learning_rate": 5.919637001606191e-06,
|
|
"loss": 0.9423,
|
|
"step": 147
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"grad_norm": 1.515625,
|
|
"learning_rate": 5.917546854792106e-06,
|
|
"loss": 1.0746,
|
|
"step": 148
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 5.915430295979674e-06,
|
|
"loss": 0.9868,
|
|
"step": 149
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 5.913287346525743e-06,
|
|
"loss": 0.9514,
|
|
"step": 150
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 5.911118028053448e-06,
|
|
"loss": 1.0203,
|
|
"step": 151
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"grad_norm": 1.53125,
|
|
"learning_rate": 5.908922362452005e-06,
|
|
"loss": 0.9735,
|
|
"step": 152
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 5.906700371876474e-06,
|
|
"loss": 1.036,
|
|
"step": 153
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 5.904452078747547e-06,
|
|
"loss": 0.979,
|
|
"step": 154
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 5.90217750575132e-06,
|
|
"loss": 0.9801,
|
|
"step": 155
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 5.899876675839058e-06,
|
|
"loss": 0.9683,
|
|
"step": 156
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 5.897549612226973e-06,
|
|
"loss": 0.9777,
|
|
"step": 157
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 5.895196338395979e-06,
|
|
"loss": 1.0319,
|
|
"step": 158
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 5.892816878091468e-06,
|
|
"loss": 0.9227,
|
|
"step": 159
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"grad_norm": 1.4921875,
|
|
"learning_rate": 5.8904112553230535e-06,
|
|
"loss": 0.9632,
|
|
"step": 160
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 5.887979494364347e-06,
|
|
"loss": 0.9928,
|
|
"step": 161
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 5.885521619752696e-06,
|
|
"loss": 1.0652,
|
|
"step": 162
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 5.883037656288951e-06,
|
|
"loss": 0.9787,
|
|
"step": 163
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 5.880527629037203e-06,
|
|
"loss": 0.9502,
|
|
"step": 164
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 5.877991563324541e-06,
|
|
"loss": 0.9333,
|
|
"step": 165
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 5.875429484740787e-06,
|
|
"loss": 0.9805,
|
|
"step": 166
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 5.872841419138246e-06,
|
|
"loss": 1.0546,
|
|
"step": 167
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 5.870227392631439e-06,
|
|
"loss": 0.9738,
|
|
"step": 168
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 5.8675874315968425e-06,
|
|
"loss": 0.945,
|
|
"step": 169
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"grad_norm": 1.5546875,
|
|
"learning_rate": 5.8649215626726225e-06,
|
|
"loss": 0.9291,
|
|
"step": 170
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 5.8622298127583635e-06,
|
|
"loss": 0.9397,
|
|
"step": 171
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 5.859512209014801e-06,
|
|
"loss": 0.8954,
|
|
"step": 172
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"grad_norm": 1.515625,
|
|
"learning_rate": 5.856768778863542e-06,
|
|
"loss": 0.9867,
|
|
"step": 173
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 5.853999549986795e-06,
|
|
"loss": 0.9465,
|
|
"step": 174
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 5.851204550327085e-06,
|
|
"loss": 0.9735,
|
|
"step": 175
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"eval_loss": 0.9932471513748169,
|
|
"eval_runtime": 20.1373,
|
|
"eval_samples_per_second": 65.004,
|
|
"eval_steps_per_second": 65.004,
|
|
"step": 175
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 5.848383808086973e-06,
|
|
"loss": 1.112,
|
|
"step": 176
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 5.845537351728772e-06,
|
|
"loss": 0.9184,
|
|
"step": 177
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 5.842665209974259e-06,
|
|
"loss": 1.0099,
|
|
"step": 178
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 5.839767411804391e-06,
|
|
"loss": 1.0246,
|
|
"step": 179
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 5.836843986459001e-06,
|
|
"loss": 0.9687,
|
|
"step": 180
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 5.833894963436514e-06,
|
|
"loss": 0.9731,
|
|
"step": 181
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 5.8309203724936425e-06,
|
|
"loss": 1.0483,
|
|
"step": 182
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 5.82792024364509e-06,
|
|
"loss": 0.9702,
|
|
"step": 183
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 5.824894607163246e-06,
|
|
"loss": 0.9713,
|
|
"step": 184
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"grad_norm": 1.53125,
|
|
"learning_rate": 5.82184349357788e-06,
|
|
"loss": 1.0378,
|
|
"step": 185
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 5.818766933675838e-06,
|
|
"loss": 1.0155,
|
|
"step": 186
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 5.815664958500725e-06,
|
|
"loss": 1.0087,
|
|
"step": 187
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 5.812537599352597e-06,
|
|
"loss": 0.9576,
|
|
"step": 188
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"grad_norm": 1.4921875,
|
|
"learning_rate": 5.8093848877876455e-06,
|
|
"loss": 0.9713,
|
|
"step": 189
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 5.8062068556178745e-06,
|
|
"loss": 0.9434,
|
|
"step": 190
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"grad_norm": 1.5546875,
|
|
"learning_rate": 5.803003534910781e-06,
|
|
"loss": 1.0159,
|
|
"step": 191
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 5.799774957989036e-06,
|
|
"loss": 0.9089,
|
|
"step": 192
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"grad_norm": 1.53125,
|
|
"learning_rate": 5.796521157430152e-06,
|
|
"loss": 1.073,
|
|
"step": 193
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 5.79324216606616e-06,
|
|
"loss": 1.0105,
|
|
"step": 194
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 5.7899380169832715e-06,
|
|
"loss": 1.2134,
|
|
"step": 195
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 5.786608743521553e-06,
|
|
"loss": 0.932,
|
|
"step": 196
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"grad_norm": 1.4921875,
|
|
"learning_rate": 5.783254379274581e-06,
|
|
"loss": 0.9971,
|
|
"step": 197
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"grad_norm": 1.484375,
|
|
"learning_rate": 5.779874958089108e-06,
|
|
"loss": 1.02,
|
|
"step": 198
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 5.7764705140647235e-06,
|
|
"loss": 0.9966,
|
|
"step": 199
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"grad_norm": 1.7578125,
|
|
"learning_rate": 5.773041081553503e-06,
|
|
"loss": 0.9188,
|
|
"step": 200
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 5.769586695159664e-06,
|
|
"loss": 0.8953,
|
|
"step": 201
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"grad_norm": 1.59375,
|
|
"learning_rate": 5.7661073897392245e-06,
|
|
"loss": 0.9489,
|
|
"step": 202
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 5.7626032003996355e-06,
|
|
"loss": 0.909,
|
|
"step": 203
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"grad_norm": 2.078125,
|
|
"learning_rate": 5.759074162499441e-06,
|
|
"loss": 0.9914,
|
|
"step": 204
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"grad_norm": 1.5859375,
|
|
"learning_rate": 5.7555203116479166e-06,
|
|
"loss": 0.9092,
|
|
"step": 205
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 5.7519416837047065e-06,
|
|
"loss": 0.9878,
|
|
"step": 206
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"grad_norm": 1.4921875,
|
|
"learning_rate": 5.748338314779465e-06,
|
|
"loss": 0.9933,
|
|
"step": 207
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"grad_norm": 1.671875,
|
|
"learning_rate": 5.744710241231496e-06,
|
|
"loss": 1.0544,
|
|
"step": 208
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"grad_norm": 1.2890625,
|
|
"learning_rate": 5.741057499669377e-06,
|
|
"loss": 1.0482,
|
|
"step": 209
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 5.737380126950598e-06,
|
|
"loss": 0.9966,
|
|
"step": 210
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"eval_loss": 0.9859724044799805,
|
|
"eval_runtime": 20.1941,
|
|
"eval_samples_per_second": 64.821,
|
|
"eval_steps_per_second": 64.821,
|
|
"step": 210
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"grad_norm": 1.84375,
|
|
"learning_rate": 5.733678160181186e-06,
|
|
"loss": 1.0631,
|
|
"step": 211
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 5.729951636715328e-06,
|
|
"loss": 0.8644,
|
|
"step": 212
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 5.726200594155e-06,
|
|
"loss": 0.9526,
|
|
"step": 213
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"grad_norm": 1.4921875,
|
|
"learning_rate": 5.722425070349584e-06,
|
|
"loss": 0.9792,
|
|
"step": 214
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 5.718625103395486e-06,
|
|
"loss": 0.9734,
|
|
"step": 215
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 5.7148007316357526e-06,
|
|
"loss": 1.07,
|
|
"step": 216
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 5.7109519936596835e-06,
|
|
"loss": 1.0494,
|
|
"step": 217
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 5.707078928302444e-06,
|
|
"loss": 0.9834,
|
|
"step": 218
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 5.703181574644667e-06,
|
|
"loss": 0.9465,
|
|
"step": 219
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"grad_norm": 1.640625,
|
|
"learning_rate": 5.699259972012069e-06,
|
|
"loss": 1.0075,
|
|
"step": 220
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"grad_norm": 1.5703125,
|
|
"learning_rate": 5.695314159975044e-06,
|
|
"loss": 1.0926,
|
|
"step": 221
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"grad_norm": 1.484375,
|
|
"learning_rate": 5.691344178348268e-06,
|
|
"loss": 0.9195,
|
|
"step": 222
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"grad_norm": 1.546875,
|
|
"learning_rate": 5.687350067190298e-06,
|
|
"loss": 0.9602,
|
|
"step": 223
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"grad_norm": 1.5625,
|
|
"learning_rate": 5.683331866803166e-06,
|
|
"loss": 0.9602,
|
|
"step": 224
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 5.679289617731972e-06,
|
|
"loss": 1.0241,
|
|
"step": 225
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 5.675223360764478e-06,
|
|
"loss": 0.9545,
|
|
"step": 226
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"grad_norm": 1.5625,
|
|
"learning_rate": 5.671133136930693e-06,
|
|
"loss": 1.0943,
|
|
"step": 227
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"grad_norm": 1.75,
|
|
"learning_rate": 5.667018987502462e-06,
|
|
"loss": 0.9773,
|
|
"step": 228
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"grad_norm": 1.609375,
|
|
"learning_rate": 5.6628809539930465e-06,
|
|
"loss": 0.8895,
|
|
"step": 229
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"grad_norm": 1.7734375,
|
|
"learning_rate": 5.658719078156706e-06,
|
|
"loss": 0.9635,
|
|
"step": 230
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 5.654533401988281e-06,
|
|
"loss": 0.9368,
|
|
"step": 231
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 5.650323967722762e-06,
|
|
"loss": 0.9458,
|
|
"step": 232
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"grad_norm": 1.5390625,
|
|
"learning_rate": 5.646090817834871e-06,
|
|
"loss": 0.8872,
|
|
"step": 233
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 5.641833995038629e-06,
|
|
"loss": 0.9918,
|
|
"step": 234
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 5.63755354228692e-06,
|
|
"loss": 0.9063,
|
|
"step": 235
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 5.633249502771071e-06,
|
|
"loss": 1.0154,
|
|
"step": 236
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 5.6289219199203996e-06,
|
|
"loss": 1.0322,
|
|
"step": 237
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 5.624570837401791e-06,
|
|
"loss": 0.9797,
|
|
"step": 238
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 5.620196299119248e-06,
|
|
"loss": 1.0433,
|
|
"step": 239
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 5.61579834921345e-06,
|
|
"loss": 0.8911,
|
|
"step": 240
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 5.611377032061308e-06,
|
|
"loss": 1.0065,
|
|
"step": 241
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 5.60693239227552e-06,
|
|
"loss": 0.9498,
|
|
"step": 242
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 5.602464474704111e-06,
|
|
"loss": 0.8877,
|
|
"step": 243
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"grad_norm": 1.6015625,
|
|
"learning_rate": 5.597973324429997e-06,
|
|
"loss": 1.0399,
|
|
"step": 244
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 5.593458986770512e-06,
|
|
"loss": 1.0804,
|
|
"step": 245
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"eval_loss": 0.9796280264854431,
|
|
"eval_runtime": 20.1507,
|
|
"eval_samples_per_second": 64.96,
|
|
"eval_steps_per_second": 64.96,
|
|
"step": 245
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 5.588921507276964e-06,
|
|
"loss": 0.9844,
|
|
"step": 246
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 5.584360931734168e-06,
|
|
"loss": 0.9852,
|
|
"step": 247
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 5.5797773061599876e-06,
|
|
"loss": 0.9387,
|
|
"step": 248
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"grad_norm": 1.5703125,
|
|
"learning_rate": 5.57517067680487e-06,
|
|
"loss": 0.9369,
|
|
"step": 249
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 5.570541090151377e-06,
|
|
"loss": 0.9211,
|
|
"step": 250
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"grad_norm": 1.609375,
|
|
"learning_rate": 5.565888592913721e-06,
|
|
"loss": 1.0429,
|
|
"step": 251
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 5.561213232037286e-06,
|
|
"loss": 0.939,
|
|
"step": 252
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"grad_norm": 1.5078125,
|
|
"learning_rate": 5.556515054698165e-06,
|
|
"loss": 1.0304,
|
|
"step": 253
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 5.5517941083026704e-06,
|
|
"loss": 0.8778,
|
|
"step": 254
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 5.547050440486868e-06,
|
|
"loss": 0.9531,
|
|
"step": 255
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"grad_norm": 1.65625,
|
|
"learning_rate": 5.542284099116085e-06,
|
|
"loss": 1.0964,
|
|
"step": 256
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"grad_norm": 1.5390625,
|
|
"learning_rate": 5.537495132284442e-06,
|
|
"loss": 1.0354,
|
|
"step": 257
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"grad_norm": 1.578125,
|
|
"learning_rate": 5.53268358831435e-06,
|
|
"loss": 0.9608,
|
|
"step": 258
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 5.527849515756036e-06,
|
|
"loss": 0.9937,
|
|
"step": 259
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 5.522992963387047e-06,
|
|
"loss": 0.9523,
|
|
"step": 260
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"grad_norm": 1.515625,
|
|
"learning_rate": 5.518113980211762e-06,
|
|
"loss": 0.929,
|
|
"step": 261
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 5.51321261546089e-06,
|
|
"loss": 0.9928,
|
|
"step": 262
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 5.5082889185909834e-06,
|
|
"loss": 0.8508,
|
|
"step": 263
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"grad_norm": 1.609375,
|
|
"learning_rate": 5.503342939283931e-06,
|
|
"loss": 0.9067,
|
|
"step": 264
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"grad_norm": 1.5078125,
|
|
"learning_rate": 5.498374727446461e-06,
|
|
"loss": 0.9878,
|
|
"step": 265
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"grad_norm": 1.578125,
|
|
"learning_rate": 5.493384333209634e-06,
|
|
"loss": 0.8992,
|
|
"step": 266
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 5.488371806928342e-06,
|
|
"loss": 1.0696,
|
|
"step": 267
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"grad_norm": 1.5078125,
|
|
"learning_rate": 5.483337199180793e-06,
|
|
"loss": 0.9069,
|
|
"step": 268
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"grad_norm": 1.5703125,
|
|
"learning_rate": 5.4782805607680085e-06,
|
|
"loss": 0.9608,
|
|
"step": 269
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 5.473201942713307e-06,
|
|
"loss": 1.0443,
|
|
"step": 270
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"grad_norm": 1.4921875,
|
|
"learning_rate": 5.46810139626179e-06,
|
|
"loss": 0.888,
|
|
"step": 271
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 5.46297897287982e-06,
|
|
"loss": 0.9668,
|
|
"step": 272
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"grad_norm": 1.53125,
|
|
"learning_rate": 5.457834724254512e-06,
|
|
"loss": 0.9828,
|
|
"step": 273
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 5.452668702293203e-06,
|
|
"loss": 0.9154,
|
|
"step": 274
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"grad_norm": 1.5625,
|
|
"learning_rate": 5.447480959122929e-06,
|
|
"loss": 0.9498,
|
|
"step": 275
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 5.442271547089904e-06,
|
|
"loss": 0.9782,
|
|
"step": 276
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 5.437040518758986e-06,
|
|
"loss": 0.9729,
|
|
"step": 277
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 5.431787926913152e-06,
|
|
"loss": 0.9455,
|
|
"step": 278
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"grad_norm": 1.515625,
|
|
"learning_rate": 5.42651382455296e-06,
|
|
"loss": 0.9834,
|
|
"step": 279
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 5.421218264896019e-06,
|
|
"loss": 0.8952,
|
|
"step": 280
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"eval_loss": 0.9743253588676453,
|
|
"eval_runtime": 20.2702,
|
|
"eval_samples_per_second": 64.578,
|
|
"eval_steps_per_second": 64.578,
|
|
"step": 280
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 5.415901301376447e-06,
|
|
"loss": 1.1117,
|
|
"step": 281
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 5.4105629876443405e-06,
|
|
"loss": 0.9905,
|
|
"step": 282
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"grad_norm": 1.5546875,
|
|
"learning_rate": 5.40520337756522e-06,
|
|
"loss": 1.0321,
|
|
"step": 283
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"grad_norm": 1.484375,
|
|
"learning_rate": 5.3998225252195e-06,
|
|
"loss": 0.9664,
|
|
"step": 284
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"grad_norm": 1.5390625,
|
|
"learning_rate": 5.394420484901934e-06,
|
|
"loss": 1.0156,
|
|
"step": 285
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 5.38899731112107e-06,
|
|
"loss": 0.8818,
|
|
"step": 286
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"grad_norm": 1.59375,
|
|
"learning_rate": 5.383553058598702e-06,
|
|
"loss": 0.9483,
|
|
"step": 287
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 5.378087782269314e-06,
|
|
"loss": 0.9671,
|
|
"step": 288
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 5.37260153727953e-06,
|
|
"loss": 0.915,
|
|
"step": 289
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 5.367094378987554e-06,
|
|
"loss": 0.9215,
|
|
"step": 290
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"grad_norm": 1.7421875,
|
|
"learning_rate": 5.361566362962613e-06,
|
|
"loss": 0.9655,
|
|
"step": 291
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 5.356017544984396e-06,
|
|
"loss": 0.9641,
|
|
"step": 292
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 5.350447981042491e-06,
|
|
"loss": 0.9303,
|
|
"step": 293
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 5.344857727335822e-06,
|
|
"loss": 0.9223,
|
|
"step": 294
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"grad_norm": 1.59375,
|
|
"learning_rate": 5.339246840272079e-06,
|
|
"loss": 1.0255,
|
|
"step": 295
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 5.333615376467149e-06,
|
|
"loss": 0.9899,
|
|
"step": 296
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 5.3279633927445486e-06,
|
|
"loss": 0.9336,
|
|
"step": 297
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 5.322290946134844e-06,
|
|
"loss": 0.9654,
|
|
"step": 298
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 5.316598093875082e-06,
|
|
"loss": 1.0265,
|
|
"step": 299
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"grad_norm": 1.4921875,
|
|
"learning_rate": 5.310884893408211e-06,
|
|
"loss": 0.9132,
|
|
"step": 300
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"grad_norm": 1.6796875,
|
|
"learning_rate": 5.305151402382497e-06,
|
|
"loss": 0.9248,
|
|
"step": 301
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 5.299397678650947e-06,
|
|
"loss": 0.9737,
|
|
"step": 302
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 5.293623780270723e-06,
|
|
"loss": 0.9366,
|
|
"step": 303
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 5.287829765502556e-06,
|
|
"loss": 0.9379,
|
|
"step": 304
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"grad_norm": 1.5625,
|
|
"learning_rate": 5.282015692810159e-06,
|
|
"loss": 1.0213,
|
|
"step": 305
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 5.276181620859639e-06,
|
|
"loss": 1.013,
|
|
"step": 306
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 5.270327608518898e-06,
|
|
"loss": 1.0134,
|
|
"step": 307
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"grad_norm": 1.5625,
|
|
"learning_rate": 5.264453714857047e-06,
|
|
"loss": 0.9575,
|
|
"step": 308
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 5.258559999143809e-06,
|
|
"loss": 0.8686,
|
|
"step": 309
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 5.252646520848912e-06,
|
|
"loss": 1.0265,
|
|
"step": 310
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 5.2467133396415016e-06,
|
|
"loss": 0.88,
|
|
"step": 311
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 5.24076051538953e-06,
|
|
"loss": 0.9549,
|
|
"step": 312
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 5.2347881081591565e-06,
|
|
"loss": 0.9798,
|
|
"step": 313
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"grad_norm": 1.6015625,
|
|
"learning_rate": 5.228796178214138e-06,
|
|
"loss": 0.8834,
|
|
"step": 314
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 5.222784786015225e-06,
|
|
"loss": 0.9394,
|
|
"step": 315
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"eval_loss": 0.9702575206756592,
|
|
"eval_runtime": 20.1318,
|
|
"eval_samples_per_second": 65.022,
|
|
"eval_steps_per_second": 65.022,
|
|
"step": 315
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 5.216753992219547e-06,
|
|
"loss": 0.9804,
|
|
"step": 316
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 5.210703857680003e-06,
|
|
"loss": 0.9201,
|
|
"step": 317
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 5.204634443444651e-06,
|
|
"loss": 0.9839,
|
|
"step": 318
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 5.198545810756079e-06,
|
|
"loss": 1.024,
|
|
"step": 319
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 5.192438021050806e-06,
|
|
"loss": 0.9319,
|
|
"step": 320
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"grad_norm": 1.515625,
|
|
"learning_rate": 5.186311135958647e-06,
|
|
"loss": 0.8863,
|
|
"step": 321
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"grad_norm": 1.484375,
|
|
"learning_rate": 5.180165217302099e-06,
|
|
"loss": 0.9621,
|
|
"step": 322
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 5.174000327095711e-06,
|
|
"loss": 0.8818,
|
|
"step": 323
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"grad_norm": 2.953125,
|
|
"learning_rate": 5.167816527545466e-06,
|
|
"loss": 0.807,
|
|
"step": 324
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 5.161613881048147e-06,
|
|
"loss": 0.9364,
|
|
"step": 325
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 5.155392450190707e-06,
|
|
"loss": 0.9787,
|
|
"step": 326
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 5.149152297749646e-06,
|
|
"loss": 0.9272,
|
|
"step": 327
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 5.142893486690366e-06,
|
|
"loss": 0.8953,
|
|
"step": 328
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 5.136616080166545e-06,
|
|
"loss": 0.952,
|
|
"step": 329
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"grad_norm": 1.7265625,
|
|
"learning_rate": 5.1303201415194946e-06,
|
|
"loss": 1.046,
|
|
"step": 330
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"grad_norm": 1.5078125,
|
|
"learning_rate": 5.1240057342775204e-06,
|
|
"loss": 0.9028,
|
|
"step": 331
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"grad_norm": 1.53125,
|
|
"learning_rate": 5.1176729221552865e-06,
|
|
"loss": 0.9819,
|
|
"step": 332
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 5.111321769053166e-06,
|
|
"loss": 0.9189,
|
|
"step": 333
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"grad_norm": 1.515625,
|
|
"learning_rate": 5.104952339056601e-06,
|
|
"loss": 0.9145,
|
|
"step": 334
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 5.098564696435452e-06,
|
|
"loss": 0.8783,
|
|
"step": 335
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 5.092158905643353e-06,
|
|
"loss": 1.0541,
|
|
"step": 336
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"grad_norm": 1.5078125,
|
|
"learning_rate": 5.085735031317061e-06,
|
|
"loss": 0.8956,
|
|
"step": 337
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"grad_norm": 1.5859375,
|
|
"learning_rate": 5.0792931382758e-06,
|
|
"loss": 1.0593,
|
|
"step": 338
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 5.072833291520609e-06,
|
|
"loss": 0.9345,
|
|
"step": 339
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"grad_norm": 1.5625,
|
|
"learning_rate": 5.066355556233688e-06,
|
|
"loss": 0.9642,
|
|
"step": 340
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 5.059859997777741e-06,
|
|
"loss": 0.8651,
|
|
"step": 341
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 5.053346681695309e-06,
|
|
"loss": 0.9767,
|
|
"step": 342
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 5.046815673708117e-06,
|
|
"loss": 0.8192,
|
|
"step": 343
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"grad_norm": 1.484375,
|
|
"learning_rate": 5.040267039716409e-06,
|
|
"loss": 0.856,
|
|
"step": 344
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 5.033700845798278e-06,
|
|
"loss": 0.9764,
|
|
"step": 345
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 5.0271171582090065e-06,
|
|
"loss": 0.9693,
|
|
"step": 346
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"grad_norm": 1.59375,
|
|
"learning_rate": 5.020516043380393e-06,
|
|
"loss": 0.8832,
|
|
"step": 347
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 5.013897567920083e-06,
|
|
"loss": 0.9449,
|
|
"step": 348
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 5.007261798610898e-06,
|
|
"loss": 0.9854,
|
|
"step": 349
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"grad_norm": 1.5546875,
|
|
"learning_rate": 5.00060880241016e-06,
|
|
"loss": 0.8984,
|
|
"step": 350
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"eval_loss": 0.9668439626693726,
|
|
"eval_runtime": 20.1513,
|
|
"eval_samples_per_second": 64.958,
|
|
"eval_steps_per_second": 64.958,
|
|
"step": 350
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 4.993938646449017e-06,
|
|
"loss": 0.9833,
|
|
"step": 351
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"grad_norm": 1.7109375,
|
|
"learning_rate": 4.9872513980317635e-06,
|
|
"loss": 0.8848,
|
|
"step": 352
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 4.9805471246351666e-06,
|
|
"loss": 0.9356,
|
|
"step": 353
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"grad_norm": 1.5625,
|
|
"learning_rate": 4.973825893907778e-06,
|
|
"loss": 1.0285,
|
|
"step": 354
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 4.9670877736692575e-06,
|
|
"loss": 1.0421,
|
|
"step": 355
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 4.960332831909685e-06,
|
|
"loss": 0.9617,
|
|
"step": 356
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"grad_norm": 1.53125,
|
|
"learning_rate": 4.953561136788877e-06,
|
|
"loss": 0.9683,
|
|
"step": 357
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 4.946772756635694e-06,
|
|
"loss": 0.9737,
|
|
"step": 358
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 4.939967759947358e-06,
|
|
"loss": 1.0571,
|
|
"step": 359
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"grad_norm": 1.5625,
|
|
"learning_rate": 4.933146215388759e-06,
|
|
"loss": 0.8903,
|
|
"step": 360
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 4.926308191791757e-06,
|
|
"loss": 0.893,
|
|
"step": 361
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 4.919453758154493e-06,
|
|
"loss": 0.9087,
|
|
"step": 362
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 4.912582983640692e-06,
|
|
"loss": 0.8752,
|
|
"step": 363
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 4.905695937578965e-06,
|
|
"loss": 0.9943,
|
|
"step": 364
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 4.898792689462106e-06,
|
|
"loss": 0.9087,
|
|
"step": 365
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 4.891873308946399e-06,
|
|
"loss": 0.9659,
|
|
"step": 366
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"grad_norm": 1.4921875,
|
|
"learning_rate": 4.884937865850902e-06,
|
|
"loss": 0.924,
|
|
"step": 367
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 4.877986430156756e-06,
|
|
"loss": 0.87,
|
|
"step": 368
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 4.871019072006473e-06,
|
|
"loss": 1.0459,
|
|
"step": 369
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"grad_norm": 1.515625,
|
|
"learning_rate": 4.864035861703225e-06,
|
|
"loss": 0.8948,
|
|
"step": 370
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 4.8570368697101415e-06,
|
|
"loss": 0.8847,
|
|
"step": 371
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 4.850022166649592e-06,
|
|
"loss": 0.9415,
|
|
"step": 372
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 4.84299182330248e-06,
|
|
"loss": 0.9833,
|
|
"step": 373
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 4.8359459106075204e-06,
|
|
"loss": 0.9669,
|
|
"step": 374
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 4.828884499660533e-06,
|
|
"loss": 0.8532,
|
|
"step": 375
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 4.8218076617137175e-06,
|
|
"loss": 0.9317,
|
|
"step": 376
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"grad_norm": 1.515625,
|
|
"learning_rate": 4.814715468174939e-06,
|
|
"loss": 0.8942,
|
|
"step": 377
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 4.807607990607007e-06,
|
|
"loss": 0.8847,
|
|
"step": 378
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 4.8004853007269504e-06,
|
|
"loss": 0.9326,
|
|
"step": 379
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 4.793347470405296e-06,
|
|
"loss": 0.8939,
|
|
"step": 380
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 4.786194571665346e-06,
|
|
"loss": 0.8817,
|
|
"step": 381
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"grad_norm": 1.609375,
|
|
"learning_rate": 4.7790266766824435e-06,
|
|
"loss": 0.9413,
|
|
"step": 382
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"grad_norm": 1.3515625,
|
|
"learning_rate": 4.771843857783254e-06,
|
|
"loss": 0.8537,
|
|
"step": 383
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 4.7646461874450294e-06,
|
|
"loss": 0.9405,
|
|
"step": 384
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 4.757433738294874e-06,
|
|
"loss": 0.9424,
|
|
"step": 385
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"eval_loss": 0.9648303389549255,
|
|
"eval_runtime": 20.2432,
|
|
"eval_samples_per_second": 64.664,
|
|
"eval_steps_per_second": 64.664,
|
|
"step": 385
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 4.750206583109022e-06,
|
|
"loss": 0.9327,
|
|
"step": 386
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 4.742964794812093e-06,
|
|
"loss": 0.8816,
|
|
"step": 387
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 4.735708446476361e-06,
|
|
"loss": 0.8918,
|
|
"step": 388
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 4.728437611321015e-06,
|
|
"loss": 0.958,
|
|
"step": 389
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 4.721152362711423e-06,
|
|
"loss": 0.8336,
|
|
"step": 390
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 4.713852774158389e-06,
|
|
"loss": 1.0331,
|
|
"step": 391
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 4.706538919317411e-06,
|
|
"loss": 0.8288,
|
|
"step": 392
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"grad_norm": 1.515625,
|
|
"learning_rate": 4.6992108719879445e-06,
|
|
"loss": 0.8429,
|
|
"step": 393
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"grad_norm": 1.4921875,
|
|
"learning_rate": 4.691868706112643e-06,
|
|
"loss": 0.8227,
|
|
"step": 394
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"grad_norm": 1.53125,
|
|
"learning_rate": 4.684512495776631e-06,
|
|
"loss": 1.0182,
|
|
"step": 395
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 4.6771423152067395e-06,
|
|
"loss": 0.915,
|
|
"step": 396
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 4.669758238770767e-06,
|
|
"loss": 0.9564,
|
|
"step": 397
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 4.662360340976728e-06,
|
|
"loss": 0.9006,
|
|
"step": 398
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 4.654948696472097e-06,
|
|
"loss": 0.9841,
|
|
"step": 399
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 4.6475233800430586e-06,
|
|
"loss": 0.8723,
|
|
"step": 400
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 4.6400844666137525e-06,
|
|
"loss": 1.0067,
|
|
"step": 401
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 4.632632031245516e-06,
|
|
"loss": 0.9236,
|
|
"step": 402
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"grad_norm": 1.625,
|
|
"learning_rate": 4.6251661491361285e-06,
|
|
"loss": 0.9849,
|
|
"step": 403
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 4.617686895619052e-06,
|
|
"loss": 0.923,
|
|
"step": 404
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"grad_norm": 1.2421875,
|
|
"learning_rate": 4.610194346162673e-06,
|
|
"loss": 0.9378,
|
|
"step": 405
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"grad_norm": 1.515625,
|
|
"learning_rate": 4.602688576369531e-06,
|
|
"loss": 1.0061,
|
|
"step": 406
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 4.595169661975573e-06,
|
|
"loss": 0.9105,
|
|
"step": 407
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 4.5876376788493754e-06,
|
|
"loss": 0.9587,
|
|
"step": 408
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 4.580092702991383e-06,
|
|
"loss": 0.877,
|
|
"step": 409
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 4.572534810533142e-06,
|
|
"loss": 0.923,
|
|
"step": 410
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 4.564964077736531e-06,
|
|
"loss": 0.9448,
|
|
"step": 411
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 4.5573805809929946e-06,
|
|
"loss": 0.9546,
|
|
"step": 412
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 4.549784396822769e-06,
|
|
"loss": 0.9835,
|
|
"step": 413
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 4.542175601874111e-06,
|
|
"loss": 0.8655,
|
|
"step": 414
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 4.534554272922525e-06,
|
|
"loss": 0.8878,
|
|
"step": 415
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 4.526920486869986e-06,
|
|
"loss": 0.9834,
|
|
"step": 416
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 4.5192743207441685e-06,
|
|
"loss": 0.8446,
|
|
"step": 417
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 4.511615851697668e-06,
|
|
"loss": 0.9525,
|
|
"step": 418
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 4.503945157007216e-06,
|
|
"loss": 1.0056,
|
|
"step": 419
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 4.4962623140729074e-06,
|
|
"loss": 0.9117,
|
|
"step": 420
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"eval_loss": 0.9622918367385864,
|
|
"eval_runtime": 20.1675,
|
|
"eval_samples_per_second": 64.906,
|
|
"eval_steps_per_second": 64.906,
|
|
"step": 420
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 4.488567400417417e-06,
|
|
"loss": 0.9416,
|
|
"step": 421
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 4.48086049368522e-06,
|
|
"loss": 0.9492,
|
|
"step": 422
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 4.473141671641803e-06,
|
|
"loss": 0.8472,
|
|
"step": 423
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"grad_norm": 1.4921875,
|
|
"learning_rate": 4.4654110121728855e-06,
|
|
"loss": 0.9374,
|
|
"step": 424
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 4.457668593283625e-06,
|
|
"loss": 0.8029,
|
|
"step": 425
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 4.449914493097844e-06,
|
|
"loss": 0.8902,
|
|
"step": 426
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 4.442148789857225e-06,
|
|
"loss": 0.9008,
|
|
"step": 427
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 4.434371561920539e-06,
|
|
"loss": 0.87,
|
|
"step": 428
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 4.426582887762836e-06,
|
|
"loss": 0.9065,
|
|
"step": 429
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 4.418782845974669e-06,
|
|
"loss": 0.9354,
|
|
"step": 430
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 4.410971515261291e-06,
|
|
"loss": 0.8697,
|
|
"step": 431
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"grad_norm": 1.6015625,
|
|
"learning_rate": 4.403148974441867e-06,
|
|
"loss": 0.8788,
|
|
"step": 432
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 4.3953153024486716e-06,
|
|
"loss": 0.9644,
|
|
"step": 433
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 4.387470578326305e-06,
|
|
"loss": 1.0109,
|
|
"step": 434
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 4.379614881230879e-06,
|
|
"loss": 0.9348,
|
|
"step": 435
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"grad_norm": 1.53125,
|
|
"learning_rate": 4.371748290429231e-06,
|
|
"loss": 0.9769,
|
|
"step": 436
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 4.363870885298119e-06,
|
|
"loss": 0.9734,
|
|
"step": 437
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"grad_norm": 1.5546875,
|
|
"learning_rate": 4.355982745323421e-06,
|
|
"loss": 0.8801,
|
|
"step": 438
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 4.348083950099335e-06,
|
|
"loss": 0.9323,
|
|
"step": 439
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 4.340174579327574e-06,
|
|
"loss": 0.9672,
|
|
"step": 440
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 4.332254712816559e-06,
|
|
"loss": 0.9541,
|
|
"step": 441
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 4.32432443048062e-06,
|
|
"loss": 0.8939,
|
|
"step": 442
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 4.316383812339188e-06,
|
|
"loss": 0.9381,
|
|
"step": 443
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 4.30843293851598e-06,
|
|
"loss": 0.8943,
|
|
"step": 444
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 4.300471889238203e-06,
|
|
"loss": 1.0116,
|
|
"step": 445
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 4.292500744835735e-06,
|
|
"loss": 0.9023,
|
|
"step": 446
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 4.284519585740317e-06,
|
|
"loss": 0.8961,
|
|
"step": 447
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 4.276528492484743e-06,
|
|
"loss": 0.9492,
|
|
"step": 448
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 4.2685275457020465e-06,
|
|
"loss": 0.8869,
|
|
"step": 449
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"grad_norm": 1.3515625,
|
|
"learning_rate": 4.260516826124686e-06,
|
|
"loss": 0.8692,
|
|
"step": 450
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"grad_norm": 1.484375,
|
|
"learning_rate": 4.2524964145837305e-06,
|
|
"loss": 0.857,
|
|
"step": 451
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"grad_norm": 1.2890625,
|
|
"learning_rate": 4.244466392008045e-06,
|
|
"loss": 0.853,
|
|
"step": 452
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 4.236426839423474e-06,
|
|
"loss": 0.8563,
|
|
"step": 453
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"grad_norm": 1.5078125,
|
|
"learning_rate": 4.2283778379520235e-06,
|
|
"loss": 0.9452,
|
|
"step": 454
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 4.22031946881104e-06,
|
|
"loss": 0.9345,
|
|
"step": 455
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"eval_loss": 0.9603658318519592,
|
|
"eval_runtime": 20.3177,
|
|
"eval_samples_per_second": 64.427,
|
|
"eval_steps_per_second": 64.427,
|
|
"step": 455
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 4.212251813312395e-06,
|
|
"loss": 0.8672,
|
|
"step": 456
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 4.204174952861663e-06,
|
|
"loss": 0.9066,
|
|
"step": 457
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 4.196088968957298e-06,
|
|
"loss": 0.9464,
|
|
"step": 458
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 4.187993943189815e-06,
|
|
"loss": 0.921,
|
|
"step": 459
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 4.179889957240963e-06,
|
|
"loss": 0.8512,
|
|
"step": 460
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 4.171777092882905e-06,
|
|
"loss": 0.8376,
|
|
"step": 461
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 4.1636554319773865e-06,
|
|
"loss": 0.961,
|
|
"step": 462
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"grad_norm": 1.4921875,
|
|
"learning_rate": 4.155525056474917e-06,
|
|
"loss": 0.9431,
|
|
"step": 463
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 4.147386048413936e-06,
|
|
"loss": 0.9156,
|
|
"step": 464
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 4.1392384899199916e-06,
|
|
"loss": 1.0053,
|
|
"step": 465
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 4.13108246320491e-06,
|
|
"loss": 0.9785,
|
|
"step": 466
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 4.122918050565959e-06,
|
|
"loss": 1.0079,
|
|
"step": 467
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 4.11474533438503e-06,
|
|
"loss": 0.8462,
|
|
"step": 468
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 4.1065643971277945e-06,
|
|
"loss": 0.9099,
|
|
"step": 469
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 4.098375321342883e-06,
|
|
"loss": 0.9068,
|
|
"step": 470
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"grad_norm": 1.671875,
|
|
"learning_rate": 4.090178189661043e-06,
|
|
"loss": 0.8288,
|
|
"step": 471
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"grad_norm": 1.5078125,
|
|
"learning_rate": 4.08197308479431e-06,
|
|
"loss": 0.9653,
|
|
"step": 472
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"grad_norm": 1.5390625,
|
|
"learning_rate": 4.073760089535173e-06,
|
|
"loss": 0.9364,
|
|
"step": 473
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"grad_norm": 1.4921875,
|
|
"learning_rate": 4.065539286755734e-06,
|
|
"loss": 0.9165,
|
|
"step": 474
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 4.057310759406881e-06,
|
|
"loss": 0.9906,
|
|
"step": 475
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 4.0490745905174425e-06,
|
|
"loss": 0.8814,
|
|
"step": 476
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 4.040830863193354e-06,
|
|
"loss": 0.9109,
|
|
"step": 477
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 4.032579660616816e-06,
|
|
"loss": 0.8054,
|
|
"step": 478
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 4.024321066045462e-06,
|
|
"loss": 0.95,
|
|
"step": 479
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 4.016055162811508e-06,
|
|
"loss": 0.9861,
|
|
"step": 480
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 4.00778203432092e-06,
|
|
"loss": 0.8921,
|
|
"step": 481
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 3.999501764052571e-06,
|
|
"loss": 0.9639,
|
|
"step": 482
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"grad_norm": 1.296875,
|
|
"learning_rate": 3.991214435557392e-06,
|
|
"loss": 0.8409,
|
|
"step": 483
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 3.982920132457538e-06,
|
|
"loss": 0.9423,
|
|
"step": 484
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 3.97461893844554e-06,
|
|
"loss": 0.9888,
|
|
"step": 485
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"grad_norm": 1.484375,
|
|
"learning_rate": 3.966310937283461e-06,
|
|
"loss": 0.9327,
|
|
"step": 486
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 3.957996212802049e-06,
|
|
"loss": 0.9888,
|
|
"step": 487
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 3.949674848899893e-06,
|
|
"loss": 0.9688,
|
|
"step": 488
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 3.9413469295425764e-06,
|
|
"loss": 0.853,
|
|
"step": 489
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"grad_norm": 1.53125,
|
|
"learning_rate": 3.933012538761832e-06,
|
|
"loss": 0.951,
|
|
"step": 490
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"eval_loss": 0.9591352343559265,
|
|
"eval_runtime": 20.1897,
|
|
"eval_samples_per_second": 64.835,
|
|
"eval_steps_per_second": 64.835,
|
|
"step": 490
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 3.924671760654688e-06,
|
|
"loss": 0.9298,
|
|
"step": 491
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"grad_norm": 1.59375,
|
|
"learning_rate": 3.916324679382624e-06,
|
|
"loss": 0.8488,
|
|
"step": 492
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 3.907971379170722e-06,
|
|
"loss": 0.8833,
|
|
"step": 493
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 3.899611944306816e-06,
|
|
"loss": 0.8187,
|
|
"step": 494
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 3.891246459140637e-06,
|
|
"loss": 0.9175,
|
|
"step": 495
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 3.88287500808297e-06,
|
|
"loss": 0.9665,
|
|
"step": 496
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 3.874497675604795e-06,
|
|
"loss": 0.9449,
|
|
"step": 497
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 3.866114546236441e-06,
|
|
"loss": 1.0137,
|
|
"step": 498
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 3.857725704566727e-06,
|
|
"loss": 0.8451,
|
|
"step": 499
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 3.849331235242112e-06,
|
|
"loss": 0.8812,
|
|
"step": 500
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 3.8409312229658385e-06,
|
|
"loss": 0.9713,
|
|
"step": 501
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 3.8325257524970825e-06,
|
|
"loss": 0.9424,
|
|
"step": 502
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 3.8241149086500925e-06,
|
|
"loss": 0.9523,
|
|
"step": 503
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"grad_norm": 1.5078125,
|
|
"learning_rate": 3.815698776293337e-06,
|
|
"loss": 0.8275,
|
|
"step": 504
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 3.807277440348651e-06,
|
|
"loss": 0.8957,
|
|
"step": 505
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 3.7988509857903683e-06,
|
|
"loss": 0.9782,
|
|
"step": 506
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 3.7904194976444786e-06,
|
|
"loss": 0.8956,
|
|
"step": 507
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 3.7819830609877576e-06,
|
|
"loss": 0.8747,
|
|
"step": 508
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 3.773541760946916e-06,
|
|
"loss": 0.9444,
|
|
"step": 509
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 3.7650956826977353e-06,
|
|
"loss": 0.873,
|
|
"step": 510
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 3.7566449114642136e-06,
|
|
"loss": 0.9871,
|
|
"step": 511
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 3.7481895325177007e-06,
|
|
"loss": 0.8154,
|
|
"step": 512
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 3.7397296311760422e-06,
|
|
"loss": 0.9796,
|
|
"step": 513
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 3.731265292802713e-06,
|
|
"loss": 0.8966,
|
|
"step": 514
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 3.722796602805963e-06,
|
|
"loss": 0.9269,
|
|
"step": 515
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 3.7143236466379477e-06,
|
|
"loss": 0.8601,
|
|
"step": 516
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"grad_norm": 1.4921875,
|
|
"learning_rate": 3.705846509793875e-06,
|
|
"loss": 0.939,
|
|
"step": 517
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 3.69736527781113e-06,
|
|
"loss": 0.9095,
|
|
"step": 518
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"grad_norm": 1.53125,
|
|
"learning_rate": 3.6888800362684264e-06,
|
|
"loss": 1.0163,
|
|
"step": 519
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 3.6803908707849324e-06,
|
|
"loss": 0.8745,
|
|
"step": 520
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"grad_norm": 1.71875,
|
|
"learning_rate": 3.6718978670194105e-06,
|
|
"loss": 1.003,
|
|
"step": 521
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 3.663401110669353e-06,
|
|
"loss": 0.9032,
|
|
"step": 522
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"grad_norm": 1.6484375,
|
|
"learning_rate": 3.654900687470116e-06,
|
|
"loss": 0.8658,
|
|
"step": 523
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 3.646396683194058e-06,
|
|
"loss": 0.9152,
|
|
"step": 524
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 3.6378891836496702e-06,
|
|
"loss": 0.9348,
|
|
"step": 525
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"eval_loss": 0.9573694467544556,
|
|
"eval_runtime": 20.2011,
|
|
"eval_samples_per_second": 64.798,
|
|
"eval_steps_per_second": 64.798,
|
|
"step": 525
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 3.629378274680713e-06,
|
|
"loss": 0.9352,
|
|
"step": 526
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"grad_norm": 1.4921875,
|
|
"learning_rate": 3.6208640421653465e-06,
|
|
"loss": 0.9221,
|
|
"step": 527
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"grad_norm": 1.546875,
|
|
"learning_rate": 3.612346572015271e-06,
|
|
"loss": 0.8442,
|
|
"step": 528
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 3.603825950174853e-06,
|
|
"loss": 0.9139,
|
|
"step": 529
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 3.5953022626202615e-06,
|
|
"loss": 0.9462,
|
|
"step": 530
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 3.5867755953585975e-06,
|
|
"loss": 0.9609,
|
|
"step": 531
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 3.578246034427033e-06,
|
|
"loss": 0.9211,
|
|
"step": 532
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 3.5697136658919335e-06,
|
|
"loss": 0.959,
|
|
"step": 533
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 3.5611785758479966e-06,
|
|
"loss": 0.8621,
|
|
"step": 534
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 3.552640850417381e-06,
|
|
"loss": 0.9998,
|
|
"step": 535
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 3.544100575748835e-06,
|
|
"loss": 0.8606,
|
|
"step": 536
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"grad_norm": 1.484375,
|
|
"learning_rate": 3.535557838016833e-06,
|
|
"loss": 1.0495,
|
|
"step": 537
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 3.527012723420701e-06,
|
|
"loss": 0.9242,
|
|
"step": 538
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 3.518465318183748e-06,
|
|
"loss": 0.9776,
|
|
"step": 539
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 3.5099157085523975e-06,
|
|
"loss": 0.9708,
|
|
"step": 540
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 3.501363980795315e-06,
|
|
"loss": 0.9329,
|
|
"step": 541
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 3.4928102212025395e-06,
|
|
"loss": 1.007,
|
|
"step": 542
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"grad_norm": 1.5078125,
|
|
"learning_rate": 3.484254516084612e-06,
|
|
"loss": 0.9819,
|
|
"step": 543
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 3.475696951771704e-06,
|
|
"loss": 0.8111,
|
|
"step": 544
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 3.467137614612746e-06,
|
|
"loss": 0.9027,
|
|
"step": 545
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 3.4585765909745595e-06,
|
|
"loss": 0.9196,
|
|
"step": 546
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 3.450013967240982e-06,
|
|
"loss": 0.9815,
|
|
"step": 547
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"grad_norm": 1.296875,
|
|
"learning_rate": 3.441449829811996e-06,
|
|
"loss": 0.8236,
|
|
"step": 548
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 3.4328842651028592e-06,
|
|
"loss": 0.87,
|
|
"step": 549
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 3.4243173595432297e-06,
|
|
"loss": 0.9304,
|
|
"step": 550
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 3.4157491995762946e-06,
|
|
"loss": 0.957,
|
|
"step": 551
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 3.4071798716579008e-06,
|
|
"loss": 0.9195,
|
|
"step": 552
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"grad_norm": 1.7421875,
|
|
"learning_rate": 3.3986094622556796e-06,
|
|
"loss": 0.8709,
|
|
"step": 553
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 3.3900380578481727e-06,
|
|
"loss": 0.9077,
|
|
"step": 554
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 3.3814657449239643e-06,
|
|
"loss": 1.0009,
|
|
"step": 555
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 3.372892609980805e-06,
|
|
"loss": 0.9224,
|
|
"step": 556
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 3.3643187395247386e-06,
|
|
"loss": 0.894,
|
|
"step": 557
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 3.3557442200692337e-06,
|
|
"loss": 0.9728,
|
|
"step": 558
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"grad_norm": 1.3515625,
|
|
"learning_rate": 3.3471691381343042e-06,
|
|
"loss": 0.9809,
|
|
"step": 559
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 3.3385935802456405e-06,
|
|
"loss": 0.9539,
|
|
"step": 560
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"eval_loss": 0.9559576511383057,
|
|
"eval_runtime": 20.2178,
|
|
"eval_samples_per_second": 64.745,
|
|
"eval_steps_per_second": 64.745,
|
|
"step": 560
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 3.330017632933736e-06,
|
|
"loss": 0.9418,
|
|
"step": 561
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 3.3214413827330156e-06,
|
|
"loss": 0.9327,
|
|
"step": 562
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"grad_norm": 1.4921875,
|
|
"learning_rate": 3.3128649161809558e-06,
|
|
"loss": 0.8283,
|
|
"step": 563
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 3.3042883198172203e-06,
|
|
"loss": 0.8813,
|
|
"step": 564
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 3.2957116801827805e-06,
|
|
"loss": 0.983,
|
|
"step": 565
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 3.287135083819045e-06,
|
|
"loss": 0.91,
|
|
"step": 566
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 3.278558617266985e-06,
|
|
"loss": 0.8266,
|
|
"step": 567
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 3.2699823670662644e-06,
|
|
"loss": 0.9438,
|
|
"step": 568
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 3.2614064197543607e-06,
|
|
"loss": 0.8858,
|
|
"step": 569
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 3.2528308618656974e-06,
|
|
"loss": 0.8525,
|
|
"step": 570
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 3.2442557799307675e-06,
|
|
"loss": 0.9663,
|
|
"step": 571
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 3.235681260475262e-06,
|
|
"loss": 1.0201,
|
|
"step": 572
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 3.2271073900191958e-06,
|
|
"loss": 0.9675,
|
|
"step": 573
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 3.2185342550760365e-06,
|
|
"loss": 0.9893,
|
|
"step": 574
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"grad_norm": 1.7109375,
|
|
"learning_rate": 3.2099619421518285e-06,
|
|
"loss": 0.9938,
|
|
"step": 575
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"grad_norm": 1.296875,
|
|
"learning_rate": 3.2013905377443207e-06,
|
|
"loss": 0.8972,
|
|
"step": 576
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 3.1928201283421e-06,
|
|
"loss": 0.8968,
|
|
"step": 577
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 3.1842508004237066e-06,
|
|
"loss": 0.9195,
|
|
"step": 578
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 3.175682640456772e-06,
|
|
"loss": 0.9322,
|
|
"step": 579
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 3.1671157348971415e-06,
|
|
"loss": 0.8878,
|
|
"step": 580
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 3.1585501701880053e-06,
|
|
"loss": 0.8621,
|
|
"step": 581
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 3.1499860327590182e-06,
|
|
"loss": 0.966,
|
|
"step": 582
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 3.1414234090254413e-06,
|
|
"loss": 0.8862,
|
|
"step": 583
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"grad_norm": 1.3515625,
|
|
"learning_rate": 3.1328623853872558e-06,
|
|
"loss": 0.9531,
|
|
"step": 584
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"grad_norm": 1.484375,
|
|
"learning_rate": 3.1243030482282973e-06,
|
|
"loss": 0.9865,
|
|
"step": 585
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 3.1157454839153885e-06,
|
|
"loss": 0.8452,
|
|
"step": 586
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 3.1071897787974613e-06,
|
|
"loss": 0.8975,
|
|
"step": 587
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 3.0986360192046853e-06,
|
|
"loss": 0.8988,
|
|
"step": 588
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 3.0900842914476024e-06,
|
|
"loss": 0.855,
|
|
"step": 589
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"grad_norm": 1.53125,
|
|
"learning_rate": 3.0815346818162534e-06,
|
|
"loss": 0.9749,
|
|
"step": 590
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 3.0729872765793e-06,
|
|
"loss": 0.9262,
|
|
"step": 591
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 3.0644421619831683e-06,
|
|
"loss": 0.8623,
|
|
"step": 592
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 3.0558994242511667e-06,
|
|
"loss": 0.9703,
|
|
"step": 593
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 3.047359149582621e-06,
|
|
"loss": 0.936,
|
|
"step": 594
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"grad_norm": 1.484375,
|
|
"learning_rate": 3.038821424152004e-06,
|
|
"loss": 1.2291,
|
|
"step": 595
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"eval_loss": 0.954813539981842,
|
|
"eval_runtime": 20.1791,
|
|
"eval_samples_per_second": 64.869,
|
|
"eval_steps_per_second": 64.869,
|
|
"step": 595
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 3.030286334108068e-06,
|
|
"loss": 0.8685,
|
|
"step": 596
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 3.021753965572967e-06,
|
|
"loss": 0.8731,
|
|
"step": 597
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 3.0132244046414032e-06,
|
|
"loss": 0.9185,
|
|
"step": 598
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 3.0046977373797396e-06,
|
|
"loss": 0.893,
|
|
"step": 599
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 2.9961740498251477e-06,
|
|
"loss": 0.8581,
|
|
"step": 600
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 2.987653427984729e-06,
|
|
"loss": 0.8867,
|
|
"step": 601
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 2.9791359578346543e-06,
|
|
"loss": 0.8389,
|
|
"step": 602
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 2.9706217253192877e-06,
|
|
"loss": 0.8954,
|
|
"step": 603
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"grad_norm": 1.71875,
|
|
"learning_rate": 2.962110816350331e-06,
|
|
"loss": 0.922,
|
|
"step": 604
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 2.953603316805943e-06,
|
|
"loss": 0.8986,
|
|
"step": 605
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 2.9450993125298848e-06,
|
|
"loss": 0.8654,
|
|
"step": 606
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"grad_norm": 1.53125,
|
|
"learning_rate": 2.936598889330648e-06,
|
|
"loss": 0.88,
|
|
"step": 607
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 2.9281021329805902e-06,
|
|
"loss": 0.8357,
|
|
"step": 608
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 2.9196091292150675e-06,
|
|
"loss": 0.8443,
|
|
"step": 609
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"grad_norm": 1.484375,
|
|
"learning_rate": 2.911119963731574e-06,
|
|
"loss": 0.9116,
|
|
"step": 610
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 2.902634722188871e-06,
|
|
"loss": 0.8414,
|
|
"step": 611
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 2.894153490206126e-06,
|
|
"loss": 1.032,
|
|
"step": 612
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 2.885676353362053e-06,
|
|
"loss": 1.0082,
|
|
"step": 613
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 2.8772033971940384e-06,
|
|
"loss": 0.8801,
|
|
"step": 614
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 2.8687347071972875e-06,
|
|
"loss": 0.9127,
|
|
"step": 615
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 2.8602703688239585e-06,
|
|
"loss": 0.8387,
|
|
"step": 616
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 2.8518104674823e-06,
|
|
"loss": 0.9205,
|
|
"step": 617
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"grad_norm": 1.6015625,
|
|
"learning_rate": 2.8433550885357863e-06,
|
|
"loss": 0.8927,
|
|
"step": 618
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 2.8349043173022655e-06,
|
|
"loss": 0.8627,
|
|
"step": 619
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 2.8264582390530854e-06,
|
|
"loss": 0.8549,
|
|
"step": 620
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 2.818016939012243e-06,
|
|
"loss": 0.8739,
|
|
"step": 621
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"grad_norm": 1.6328125,
|
|
"learning_rate": 2.809580502355522e-06,
|
|
"loss": 1.0039,
|
|
"step": 622
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"grad_norm": 1.5390625,
|
|
"learning_rate": 2.8011490142096316e-06,
|
|
"loss": 0.9061,
|
|
"step": 623
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 2.79272255965135e-06,
|
|
"loss": 0.9938,
|
|
"step": 624
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"grad_norm": 1.546875,
|
|
"learning_rate": 2.7843012237066627e-06,
|
|
"loss": 1.0029,
|
|
"step": 625
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 2.775885091349909e-06,
|
|
"loss": 0.9585,
|
|
"step": 626
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 2.767474247502919e-06,
|
|
"loss": 0.8371,
|
|
"step": 627
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 2.7590687770341626e-06,
|
|
"loss": 0.8878,
|
|
"step": 628
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"grad_norm": 1.484375,
|
|
"learning_rate": 2.7506687647578895e-06,
|
|
"loss": 0.9514,
|
|
"step": 629
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 2.7422742954332735e-06,
|
|
"loss": 0.9397,
|
|
"step": 630
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"eval_loss": 0.9540159106254578,
|
|
"eval_runtime": 20.3197,
|
|
"eval_samples_per_second": 64.42,
|
|
"eval_steps_per_second": 64.42,
|
|
"step": 630
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 2.7338854537635585e-06,
|
|
"loss": 0.9597,
|
|
"step": 631
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 2.725502324395205e-06,
|
|
"loss": 0.9125,
|
|
"step": 632
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 2.7171249919170304e-06,
|
|
"loss": 0.9403,
|
|
"step": 633
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 2.7087535408593643e-06,
|
|
"loss": 0.9128,
|
|
"step": 634
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 2.7003880556931855e-06,
|
|
"loss": 0.8996,
|
|
"step": 635
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 2.6920286208292777e-06,
|
|
"loss": 0.9428,
|
|
"step": 636
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"grad_norm": 1.484375,
|
|
"learning_rate": 2.683675320617376e-06,
|
|
"loss": 0.8807,
|
|
"step": 637
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 2.6753282393453134e-06,
|
|
"loss": 1.0051,
|
|
"step": 638
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 2.6669874612381685e-06,
|
|
"loss": 0.926,
|
|
"step": 639
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 2.6586530704574235e-06,
|
|
"loss": 0.9606,
|
|
"step": 640
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 2.650325151100108e-06,
|
|
"loss": 0.8918,
|
|
"step": 641
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"grad_norm": 1.3515625,
|
|
"learning_rate": 2.642003787197952e-06,
|
|
"loss": 0.9572,
|
|
"step": 642
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 2.6336890627165397e-06,
|
|
"loss": 0.8793,
|
|
"step": 643
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 2.6253810615544603e-06,
|
|
"loss": 0.8986,
|
|
"step": 644
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 2.6170798675424617e-06,
|
|
"loss": 0.904,
|
|
"step": 645
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 2.608785564442609e-06,
|
|
"loss": 0.9042,
|
|
"step": 646
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 2.600498235947431e-06,
|
|
"loss": 0.9596,
|
|
"step": 647
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"grad_norm": 1.4921875,
|
|
"learning_rate": 2.5922179656790796e-06,
|
|
"loss": 0.8224,
|
|
"step": 648
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"grad_norm": 1.5390625,
|
|
"learning_rate": 2.5839448371884926e-06,
|
|
"loss": 0.9048,
|
|
"step": 649
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 2.5756789339545387e-06,
|
|
"loss": 0.9585,
|
|
"step": 650
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"grad_norm": 1.5546875,
|
|
"learning_rate": 2.567420339383184e-06,
|
|
"loss": 0.9117,
|
|
"step": 651
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 2.5591691368066475e-06,
|
|
"loss": 0.9752,
|
|
"step": 652
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 2.550925409482559e-06,
|
|
"loss": 0.9207,
|
|
"step": 653
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 2.542689240593119e-06,
|
|
"loss": 0.8864,
|
|
"step": 654
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 2.5344607132442667e-06,
|
|
"loss": 0.8982,
|
|
"step": 655
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"grad_norm": 1.5625,
|
|
"learning_rate": 2.5262399104648286e-06,
|
|
"loss": 0.9327,
|
|
"step": 656
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 2.51802691520569e-06,
|
|
"loss": 0.8947,
|
|
"step": 657
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 2.509821810338957e-06,
|
|
"loss": 0.9742,
|
|
"step": 658
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 2.5016246786571174e-06,
|
|
"loss": 0.8728,
|
|
"step": 659
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 2.493435602872206e-06,
|
|
"loss": 0.9128,
|
|
"step": 660
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"grad_norm": 1.5390625,
|
|
"learning_rate": 2.485254665614971e-06,
|
|
"loss": 0.9039,
|
|
"step": 661
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 2.4770819494340417e-06,
|
|
"loss": 0.8575,
|
|
"step": 662
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 2.4689175367950906e-06,
|
|
"loss": 1.0352,
|
|
"step": 663
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 2.4607615100800083e-06,
|
|
"loss": 0.9921,
|
|
"step": 664
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"grad_norm": 1.3515625,
|
|
"learning_rate": 2.452613951586065e-06,
|
|
"loss": 0.985,
|
|
"step": 665
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"eval_loss": 0.9532026052474976,
|
|
"eval_runtime": 20.1343,
|
|
"eval_samples_per_second": 65.014,
|
|
"eval_steps_per_second": 65.014,
|
|
"step": 665
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 2.4444749435250837e-06,
|
|
"loss": 0.8722,
|
|
"step": 666
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 2.436344568022614e-06,
|
|
"loss": 0.8208,
|
|
"step": 667
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 2.4282229071170965e-06,
|
|
"loss": 0.9407,
|
|
"step": 668
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"grad_norm": 1.578125,
|
|
"learning_rate": 2.4201100427590365e-06,
|
|
"loss": 0.8911,
|
|
"step": 669
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 2.4120060568101856e-06,
|
|
"loss": 0.8809,
|
|
"step": 670
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 2.403911031042702e-06,
|
|
"loss": 0.8505,
|
|
"step": 671
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 2.3958250471383373e-06,
|
|
"loss": 0.9088,
|
|
"step": 672
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"grad_norm": 1.53125,
|
|
"learning_rate": 2.387748186687606e-06,
|
|
"loss": 0.826,
|
|
"step": 673
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 2.3796805311889613e-06,
|
|
"loss": 0.8774,
|
|
"step": 674
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 2.3716221620479772e-06,
|
|
"loss": 0.8676,
|
|
"step": 675
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 2.3635731605765264e-06,
|
|
"loss": 0.9997,
|
|
"step": 676
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"grad_norm": 1.5859375,
|
|
"learning_rate": 2.3555336079919567e-06,
|
|
"loss": 0.8903,
|
|
"step": 677
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 2.3475035854162707e-06,
|
|
"loss": 0.8055,
|
|
"step": 678
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"grad_norm": 1.5390625,
|
|
"learning_rate": 2.3394831738753154e-06,
|
|
"loss": 0.8353,
|
|
"step": 679
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 2.3314724542979543e-06,
|
|
"loss": 0.9792,
|
|
"step": 680
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 2.323471507515257e-06,
|
|
"loss": 0.8619,
|
|
"step": 681
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 2.3154804142596834e-06,
|
|
"loss": 0.8911,
|
|
"step": 682
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"grad_norm": 1.53125,
|
|
"learning_rate": 2.307499255164266e-06,
|
|
"loss": 0.9566,
|
|
"step": 683
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"grad_norm": 1.3515625,
|
|
"learning_rate": 2.2995281107617967e-06,
|
|
"loss": 0.8491,
|
|
"step": 684
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"grad_norm": 1.59375,
|
|
"learning_rate": 2.291567061484021e-06,
|
|
"loss": 0.9986,
|
|
"step": 685
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 2.2836161876608136e-06,
|
|
"loss": 0.9711,
|
|
"step": 686
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 2.27567556951938e-06,
|
|
"loss": 0.9344,
|
|
"step": 687
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 2.267745287183442e-06,
|
|
"loss": 0.9524,
|
|
"step": 688
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 2.2598254206724275e-06,
|
|
"loss": 0.8871,
|
|
"step": 689
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 2.251916049900664e-06,
|
|
"loss": 1.095,
|
|
"step": 690
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"grad_norm": 1.5390625,
|
|
"learning_rate": 2.2440172546765796e-06,
|
|
"loss": 0.8317,
|
|
"step": 691
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 2.2361291147018816e-06,
|
|
"loss": 0.8479,
|
|
"step": 692
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 2.2282517095707697e-06,
|
|
"loss": 0.9132,
|
|
"step": 693
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"grad_norm": 1.4921875,
|
|
"learning_rate": 2.2203851187691216e-06,
|
|
"loss": 0.8728,
|
|
"step": 694
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 2.2125294216736956e-06,
|
|
"loss": 0.927,
|
|
"step": 695
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"grad_norm": 1.5859375,
|
|
"learning_rate": 2.2046846975513275e-06,
|
|
"loss": 0.8764,
|
|
"step": 696
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 2.1968510255581348e-06,
|
|
"loss": 0.9311,
|
|
"step": 697
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 2.1890284847387107e-06,
|
|
"loss": 0.9595,
|
|
"step": 698
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 2.1812171540253324e-06,
|
|
"loss": 0.9611,
|
|
"step": 699
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 2.1734171122371646e-06,
|
|
"loss": 0.9543,
|
|
"step": 700
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"eval_loss": 0.9524132609367371,
|
|
"eval_runtime": 20.1667,
|
|
"eval_samples_per_second": 64.909,
|
|
"eval_steps_per_second": 64.909,
|
|
"step": 700
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 2.165628438079463e-06,
|
|
"loss": 0.9835,
|
|
"step": 701
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"grad_norm": 1.4921875,
|
|
"learning_rate": 2.1578512101427746e-06,
|
|
"loss": 0.9356,
|
|
"step": 702
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 2.1500855069021567e-06,
|
|
"loss": 0.906,
|
|
"step": 703
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"grad_norm": 1.578125,
|
|
"learning_rate": 2.1423314067163747e-06,
|
|
"loss": 0.9937,
|
|
"step": 704
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 2.1345889878271152e-06,
|
|
"loss": 0.9486,
|
|
"step": 705
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 2.126858328358197e-06,
|
|
"loss": 0.9641,
|
|
"step": 706
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 2.11913950631478e-06,
|
|
"loss": 0.866,
|
|
"step": 707
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"grad_norm": 1.5078125,
|
|
"learning_rate": 2.111432599582582e-06,
|
|
"loss": 0.9687,
|
|
"step": 708
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"grad_norm": 1.5546875,
|
|
"learning_rate": 2.1037376859270937e-06,
|
|
"loss": 0.9431,
|
|
"step": 709
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 2.0960548429927853e-06,
|
|
"loss": 0.8683,
|
|
"step": 710
|
|
},
|
|
{
|
|
"epoch": 2.0,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 2.088384148302332e-06,
|
|
"loss": 0.895,
|
|
"step": 711
|
|
},
|
|
{
|
|
"epoch": 2.0,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 2.0807256792558305e-06,
|
|
"loss": 0.919,
|
|
"step": 712
|
|
}
|
|
],
|
|
"logging_steps": 1,
|
|
"max_steps": 1059,
|
|
"num_input_tokens_seen": 0,
|
|
"num_train_epochs": 3,
|
|
"save_steps": 89,
|
|
"total_flos": 2.1011521993955082e+18,
|
|
"train_batch_size": 1,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|