Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +1023 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,1023 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 2.992969696969697,
|
| 6 |
+
"eval_steps": 50,
|
| 7 |
+
"global_step": 1158,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.02585858585858586,
|
| 14 |
+
"grad_norm": 129761.5390625,
|
| 15 |
+
"learning_rate": 7.758620689655173e-06,
|
| 16 |
+
"loss": 5.1637,
|
| 17 |
+
"step": 10
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.05171717171717172,
|
| 21 |
+
"grad_norm": 85007.0390625,
|
| 22 |
+
"learning_rate": 1.6379310344827585e-05,
|
| 23 |
+
"loss": 4.8462,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.07757575757575758,
|
| 28 |
+
"grad_norm": 77711.90625,
|
| 29 |
+
"learning_rate": 2.5e-05,
|
| 30 |
+
"loss": 4.6229,
|
| 31 |
+
"step": 30
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.10343434343434343,
|
| 35 |
+
"grad_norm": 75926.6015625,
|
| 36 |
+
"learning_rate": 3.3620689655172414e-05,
|
| 37 |
+
"loss": 4.4402,
|
| 38 |
+
"step": 40
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.1292929292929293,
|
| 42 |
+
"grad_norm": 81001.3046875,
|
| 43 |
+
"learning_rate": 4.224137931034483e-05,
|
| 44 |
+
"loss": 4.186,
|
| 45 |
+
"step": 50
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.1292929292929293,
|
| 49 |
+
"eval_loss": 8.234293937683105,
|
| 50 |
+
"eval_runtime": 21.2938,
|
| 51 |
+
"eval_samples_per_second": 46.962,
|
| 52 |
+
"eval_steps_per_second": 2.959,
|
| 53 |
+
"step": 50
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 0.15515151515151515,
|
| 57 |
+
"grad_norm": 77491.1875,
|
| 58 |
+
"learning_rate": 4.9999898041346155e-05,
|
| 59 |
+
"loss": 3.9984,
|
| 60 |
+
"step": 60
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 0.181010101010101,
|
| 64 |
+
"grad_norm": 70187.9375,
|
| 65 |
+
"learning_rate": 4.998766400914329e-05,
|
| 66 |
+
"loss": 3.7669,
|
| 67 |
+
"step": 70
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"epoch": 0.20686868686868687,
|
| 71 |
+
"grad_norm": 78757.3671875,
|
| 72 |
+
"learning_rate": 4.995504967976098e-05,
|
| 73 |
+
"loss": 3.4933,
|
| 74 |
+
"step": 80
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.23272727272727273,
|
| 78 |
+
"grad_norm": 93011.2109375,
|
| 79 |
+
"learning_rate": 4.9902081653914056e-05,
|
| 80 |
+
"loss": 3.3346,
|
| 81 |
+
"step": 90
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.2585858585858586,
|
| 85 |
+
"grad_norm": 82705.953125,
|
| 86 |
+
"learning_rate": 4.982880313308417e-05,
|
| 87 |
+
"loss": 3.2104,
|
| 88 |
+
"step": 100
|
| 89 |
+
},
|
| 90 |
+
{
|
| 91 |
+
"epoch": 0.2585858585858586,
|
| 92 |
+
"eval_loss": 6.281063556671143,
|
| 93 |
+
"eval_runtime": 21.0455,
|
| 94 |
+
"eval_samples_per_second": 47.516,
|
| 95 |
+
"eval_steps_per_second": 2.994,
|
| 96 |
+
"step": 100
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 0.28444444444444444,
|
| 100 |
+
"grad_norm": 98789.734375,
|
| 101 |
+
"learning_rate": 4.973527388428399e-05,
|
| 102 |
+
"loss": 3.0765,
|
| 103 |
+
"step": 110
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"epoch": 0.3103030303030303,
|
| 107 |
+
"grad_norm": 88273.296875,
|
| 108 |
+
"learning_rate": 4.962157019131037e-05,
|
| 109 |
+
"loss": 2.9833,
|
| 110 |
+
"step": 120
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.33616161616161616,
|
| 114 |
+
"grad_norm": 78808.640625,
|
| 115 |
+
"learning_rate": 4.948778479252625e-05,
|
| 116 |
+
"loss": 2.8664,
|
| 117 |
+
"step": 130
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.362020202020202,
|
| 121 |
+
"grad_norm": 68227.828125,
|
| 122 |
+
"learning_rate": 4.933402680522181e-05,
|
| 123 |
+
"loss": 2.7739,
|
| 124 |
+
"step": 140
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 0.3878787878787879,
|
| 128 |
+
"grad_norm": 70486.9453125,
|
| 129 |
+
"learning_rate": 4.916042163661697e-05,
|
| 130 |
+
"loss": 2.6577,
|
| 131 |
+
"step": 150
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 0.3878787878787879,
|
| 135 |
+
"eval_loss": 5.267953872680664,
|
| 136 |
+
"eval_runtime": 21.2432,
|
| 137 |
+
"eval_samples_per_second": 47.074,
|
| 138 |
+
"eval_steps_per_second": 2.966,
|
| 139 |
+
"step": 150
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"epoch": 0.41373737373737374,
|
| 143 |
+
"grad_norm": 129737.7421875,
|
| 144 |
+
"learning_rate": 4.896711088157736e-05,
|
| 145 |
+
"loss": 2.5488,
|
| 146 |
+
"step": 160
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.4395959595959596,
|
| 150 |
+
"grad_norm": 92119.6328125,
|
| 151 |
+
"learning_rate": 4.875425220712757e-05,
|
| 152 |
+
"loss": 2.5209,
|
| 153 |
+
"step": 170
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.46545454545454545,
|
| 157 |
+
"grad_norm": 76282.8046875,
|
| 158 |
+
"learning_rate": 4.852201922385564e-05,
|
| 159 |
+
"loss": 2.496,
|
| 160 |
+
"step": 180
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 0.4913131313131313,
|
| 164 |
+
"grad_norm": 64090.921875,
|
| 165 |
+
"learning_rate": 4.827060134431379e-05,
|
| 166 |
+
"loss": 2.4329,
|
| 167 |
+
"step": 190
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 0.5171717171717172,
|
| 171 |
+
"grad_norm": 62604.421875,
|
| 172 |
+
"learning_rate": 4.800020362853082e-05,
|
| 173 |
+
"loss": 2.3479,
|
| 174 |
+
"step": 200
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.5171717171717172,
|
| 178 |
+
"eval_loss": 4.751099109649658,
|
| 179 |
+
"eval_runtime": 21.4328,
|
| 180 |
+
"eval_samples_per_second": 46.658,
|
| 181 |
+
"eval_steps_per_second": 2.939,
|
| 182 |
+
"step": 200
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.5430303030303031,
|
| 186 |
+
"grad_norm": 89043.59375,
|
| 187 |
+
"learning_rate": 4.7711046616762206e-05,
|
| 188 |
+
"loss": 2.331,
|
| 189 |
+
"step": 210
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.5688888888888889,
|
| 193 |
+
"grad_norm": 70265.078125,
|
| 194 |
+
"learning_rate": 4.7403366149614304e-05,
|
| 195 |
+
"loss": 2.2867,
|
| 196 |
+
"step": 220
|
| 197 |
+
},
|
| 198 |
+
{
|
| 199 |
+
"epoch": 0.5947474747474748,
|
| 200 |
+
"grad_norm": 66474.015625,
|
| 201 |
+
"learning_rate": 4.7077413175689365e-05,
|
| 202 |
+
"loss": 2.2422,
|
| 203 |
+
"step": 230
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 0.6206060606060606,
|
| 207 |
+
"grad_norm": 72337.0703125,
|
| 208 |
+
"learning_rate": 4.6733453546908277e-05,
|
| 209 |
+
"loss": 2.2279,
|
| 210 |
+
"step": 240
|
| 211 |
+
},
|
| 212 |
+
{
|
| 213 |
+
"epoch": 0.6464646464646465,
|
| 214 |
+
"grad_norm": 77678.4921875,
|
| 215 |
+
"learning_rate": 4.63717678016779e-05,
|
| 216 |
+
"loss": 2.1829,
|
| 217 |
+
"step": 250
|
| 218 |
+
},
|
| 219 |
+
{
|
| 220 |
+
"epoch": 0.6464646464646465,
|
| 221 |
+
"eval_loss": 4.369379043579102,
|
| 222 |
+
"eval_runtime": 21.3413,
|
| 223 |
+
"eval_samples_per_second": 46.857,
|
| 224 |
+
"eval_steps_per_second": 2.952,
|
| 225 |
+
"step": 250
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.6723232323232323,
|
| 229 |
+
"grad_norm": 105560.984375,
|
| 230 |
+
"learning_rate": 4.599265093607993e-05,
|
| 231 |
+
"loss": 2.1284,
|
| 232 |
+
"step": 260
|
| 233 |
+
},
|
| 234 |
+
{
|
| 235 |
+
"epoch": 0.6981818181818182,
|
| 236 |
+
"grad_norm": 94847.96875,
|
| 237 |
+
"learning_rate": 4.5596412163267846e-05,
|
| 238 |
+
"loss": 2.1122,
|
| 239 |
+
"step": 270
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 0.724040404040404,
|
| 243 |
+
"grad_norm": 80977.328125,
|
| 244 |
+
"learning_rate": 4.518337466126826e-05,
|
| 245 |
+
"loss": 2.1004,
|
| 246 |
+
"step": 280
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 0.74989898989899,
|
| 250 |
+
"grad_norm": 97450.6796875,
|
| 251 |
+
"learning_rate": 4.4753875309392266e-05,
|
| 252 |
+
"loss": 2.0487,
|
| 253 |
+
"step": 290
|
| 254 |
+
},
|
| 255 |
+
{
|
| 256 |
+
"epoch": 0.7757575757575758,
|
| 257 |
+
"grad_norm": 69245.8125,
|
| 258 |
+
"learning_rate": 4.4308264413471814e-05,
|
| 259 |
+
"loss": 2.0288,
|
| 260 |
+
"step": 300
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 0.7757575757575758,
|
| 264 |
+
"eval_loss": 4.115316867828369,
|
| 265 |
+
"eval_runtime": 21.4236,
|
| 266 |
+
"eval_samples_per_second": 46.678,
|
| 267 |
+
"eval_steps_per_second": 2.941,
|
| 268 |
+
"step": 300
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.8016161616161617,
|
| 272 |
+
"grad_norm": 78553.4296875,
|
| 273 |
+
"learning_rate": 4.384690542014529e-05,
|
| 274 |
+
"loss": 2.0385,
|
| 275 |
+
"step": 310
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.8274747474747475,
|
| 279 |
+
"grad_norm": 68862.0859375,
|
| 280 |
+
"learning_rate": 4.3370174620425216e-05,
|
| 281 |
+
"loss": 2.0088,
|
| 282 |
+
"step": 320
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.8533333333333334,
|
| 286 |
+
"grad_norm": 96176.8515625,
|
| 287 |
+
"learning_rate": 4.2878460842789905e-05,
|
| 288 |
+
"loss": 2.0057,
|
| 289 |
+
"step": 330
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.8791919191919192,
|
| 293 |
+
"grad_norm": 109879.5859375,
|
| 294 |
+
"learning_rate": 4.237216513604933e-05,
|
| 295 |
+
"loss": 1.978,
|
| 296 |
+
"step": 340
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.9050505050505051,
|
| 300 |
+
"grad_norm": 111969.0234375,
|
| 301 |
+
"learning_rate": 4.185170044224404e-05,
|
| 302 |
+
"loss": 1.969,
|
| 303 |
+
"step": 350
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.9050505050505051,
|
| 307 |
+
"eval_loss": 3.927480936050415,
|
| 308 |
+
"eval_runtime": 21.4891,
|
| 309 |
+
"eval_samples_per_second": 46.535,
|
| 310 |
+
"eval_steps_per_second": 2.932,
|
| 311 |
+
"step": 350
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.9309090909090909,
|
| 315 |
+
"grad_norm": 94754.984375,
|
| 316 |
+
"learning_rate": 4.131749125984362e-05,
|
| 317 |
+
"loss": 1.9188,
|
| 318 |
+
"step": 360
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.9567676767676768,
|
| 322 |
+
"grad_norm": 96219.0234375,
|
| 323 |
+
"learning_rate": 4.076997329751977e-05,
|
| 324 |
+
"loss": 1.9077,
|
| 325 |
+
"step": 370
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.9826262626262626,
|
| 329 |
+
"grad_norm": 84781.5234375,
|
| 330 |
+
"learning_rate": 4.0209593118775937e-05,
|
| 331 |
+
"loss": 1.8951,
|
| 332 |
+
"step": 380
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 1.0077575757575759,
|
| 336 |
+
"grad_norm": 91395.078125,
|
| 337 |
+
"learning_rate": 3.963680777772377e-05,
|
| 338 |
+
"loss": 1.8716,
|
| 339 |
+
"step": 390
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 1.0336161616161617,
|
| 343 |
+
"grad_norm": 106231.5078125,
|
| 344 |
+
"learning_rate": 3.905208444630327e-05,
|
| 345 |
+
"loss": 1.8927,
|
| 346 |
+
"step": 400
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 1.0336161616161617,
|
| 350 |
+
"eval_loss": 3.7736313343048096,
|
| 351 |
+
"eval_runtime": 21.5204,
|
| 352 |
+
"eval_samples_per_second": 46.468,
|
| 353 |
+
"eval_steps_per_second": 2.927,
|
| 354 |
+
"step": 400
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 1.0594747474747475,
|
| 358 |
+
"grad_norm": 106825.3359375,
|
| 359 |
+
"learning_rate": 3.845590003325058e-05,
|
| 360 |
+
"loss": 1.8803,
|
| 361 |
+
"step": 410
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 1.0853333333333333,
|
| 365 |
+
"grad_norm": 80191.515625,
|
| 366 |
+
"learning_rate": 3.7848740795124436e-05,
|
| 367 |
+
"loss": 1.8041,
|
| 368 |
+
"step": 420
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 1.1111919191919193,
|
| 372 |
+
"grad_norm": 113625.0234375,
|
| 373 |
+
"learning_rate": 3.7231101939708305e-05,
|
| 374 |
+
"loss": 1.8161,
|
| 375 |
+
"step": 430
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 1.137050505050505,
|
| 379 |
+
"grad_norm": 91820.0546875,
|
| 380 |
+
"learning_rate": 3.660348722211186e-05,
|
| 381 |
+
"loss": 1.7924,
|
| 382 |
+
"step": 440
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 1.162909090909091,
|
| 386 |
+
"grad_norm": 108857.2578125,
|
| 387 |
+
"learning_rate": 3.596640853390103e-05,
|
| 388 |
+
"loss": 1.813,
|
| 389 |
+
"step": 450
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 1.162909090909091,
|
| 393 |
+
"eval_loss": 3.652933359146118,
|
| 394 |
+
"eval_runtime": 21.5614,
|
| 395 |
+
"eval_samples_per_second": 46.379,
|
| 396 |
+
"eval_steps_per_second": 2.922,
|
| 397 |
+
"step": 450
|
| 398 |
+
},
|
| 399 |
+
{
|
| 400 |
+
"epoch": 1.1887676767676767,
|
| 401 |
+
"grad_norm": 107753.078125,
|
| 402 |
+
"learning_rate": 3.532038548559193e-05,
|
| 403 |
+
"loss": 1.7731,
|
| 404 |
+
"step": 460
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 1.2146262626262627,
|
| 408 |
+
"grad_norm": 99259.5,
|
| 409 |
+
"learning_rate": 3.4665944982849086e-05,
|
| 410 |
+
"loss": 1.7662,
|
| 411 |
+
"step": 470
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 1.2404848484848485,
|
| 415 |
+
"grad_norm": 96071.25,
|
| 416 |
+
"learning_rate": 3.400362079673357e-05,
|
| 417 |
+
"loss": 1.8034,
|
| 418 |
+
"step": 480
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 1.2663434343434343,
|
| 422 |
+
"grad_norm": 100736.2890625,
|
| 423 |
+
"learning_rate": 3.33339531283517e-05,
|
| 424 |
+
"loss": 1.7841,
|
| 425 |
+
"step": 490
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 1.2922020202020201,
|
| 429 |
+
"grad_norm": 93535.171875,
|
| 430 |
+
"learning_rate": 3.26574881682593e-05,
|
| 431 |
+
"loss": 1.7428,
|
| 432 |
+
"step": 500
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 1.2922020202020201,
|
| 436 |
+
"eval_loss": 3.5527358055114746,
|
| 437 |
+
"eval_runtime": 21.3139,
|
| 438 |
+
"eval_samples_per_second": 46.918,
|
| 439 |
+
"eval_steps_per_second": 2.956,
|
| 440 |
+
"step": 500
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 1.3180606060606062,
|
| 444 |
+
"grad_norm": 78968.21875,
|
| 445 |
+
"learning_rate": 3.1974777650980735e-05,
|
| 446 |
+
"loss": 1.7202,
|
| 447 |
+
"step": 510
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 1.343919191919192,
|
| 451 |
+
"grad_norm": 115748.171875,
|
| 452 |
+
"learning_rate": 3.1286378405006465e-05,
|
| 453 |
+
"loss": 1.7063,
|
| 454 |
+
"step": 520
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 1.3697777777777778,
|
| 458 |
+
"grad_norm": 90799.15625,
|
| 459 |
+
"learning_rate": 3.059285189863564e-05,
|
| 460 |
+
"loss": 1.7286,
|
| 461 |
+
"step": 530
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 1.3956363636363636,
|
| 465 |
+
"grad_norm": 108626.28125,
|
| 466 |
+
"learning_rate": 2.9894763782034542e-05,
|
| 467 |
+
"loss": 1.709,
|
| 468 |
+
"step": 540
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 1.4214949494949494,
|
| 472 |
+
"grad_norm": 102933.5234375,
|
| 473 |
+
"learning_rate": 2.9192683425884164e-05,
|
| 474 |
+
"loss": 1.7114,
|
| 475 |
+
"step": 550
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 1.4214949494949494,
|
| 479 |
+
"eval_loss": 3.471935510635376,
|
| 480 |
+
"eval_runtime": 21.2904,
|
| 481 |
+
"eval_samples_per_second": 46.97,
|
| 482 |
+
"eval_steps_per_second": 2.959,
|
| 483 |
+
"step": 550
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 1.4473535353535354,
|
| 487 |
+
"grad_norm": 93085.421875,
|
| 488 |
+
"learning_rate": 2.8487183456993378e-05,
|
| 489 |
+
"loss": 1.6962,
|
| 490 |
+
"step": 560
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 1.4732121212121212,
|
| 494 |
+
"grad_norm": 79838.0859375,
|
| 495 |
+
"learning_rate": 2.7778839291256232e-05,
|
| 496 |
+
"loss": 1.6844,
|
| 497 |
+
"step": 570
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 1.499070707070707,
|
| 501 |
+
"grad_norm": 91634.3984375,
|
| 502 |
+
"learning_rate": 2.706822866433451e-05,
|
| 503 |
+
"loss": 1.6929,
|
| 504 |
+
"step": 580
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 1.524929292929293,
|
| 508 |
+
"grad_norm": 91946.4375,
|
| 509 |
+
"learning_rate": 2.6355931160448272e-05,
|
| 510 |
+
"loss": 1.6786,
|
| 511 |
+
"step": 590
|
| 512 |
+
},
|
| 513 |
+
{
|
| 514 |
+
"epoch": 1.5507878787878788,
|
| 515 |
+
"grad_norm": 69164.7265625,
|
| 516 |
+
"learning_rate": 2.564252773965861e-05,
|
| 517 |
+
"loss": 1.6562,
|
| 518 |
+
"step": 600
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 1.5507878787878788,
|
| 522 |
+
"eval_loss": 3.403024911880493,
|
| 523 |
+
"eval_runtime": 22.0604,
|
| 524 |
+
"eval_samples_per_second": 45.33,
|
| 525 |
+
"eval_steps_per_second": 2.856,
|
| 526 |
+
"step": 600
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 1.5766464646464646,
|
| 530 |
+
"grad_norm": 91655.7109375,
|
| 531 |
+
"learning_rate": 2.4928600264028312e-05,
|
| 532 |
+
"loss": 1.676,
|
| 533 |
+
"step": 610
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 1.6025050505050507,
|
| 537 |
+
"grad_norm": 88343.1875,
|
| 538 |
+
"learning_rate": 2.4214731023046793e-05,
|
| 539 |
+
"loss": 1.6749,
|
| 540 |
+
"step": 620
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 1.6283636363636362,
|
| 544 |
+
"grad_norm": 92792.515625,
|
| 545 |
+
"learning_rate": 2.3501502258706492e-05,
|
| 546 |
+
"loss": 1.6358,
|
| 547 |
+
"step": 630
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 1.6542222222222223,
|
| 551 |
+
"grad_norm": 89480.8671875,
|
| 552 |
+
"learning_rate": 2.2789495690617857e-05,
|
| 553 |
+
"loss": 1.6461,
|
| 554 |
+
"step": 640
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 1.680080808080808,
|
| 558 |
+
"grad_norm": 88156.109375,
|
| 559 |
+
"learning_rate": 2.2079292041550607e-05,
|
| 560 |
+
"loss": 1.6462,
|
| 561 |
+
"step": 650
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 1.680080808080808,
|
| 565 |
+
"eval_loss": 3.3437397480010986,
|
| 566 |
+
"eval_runtime": 21.5027,
|
| 567 |
+
"eval_samples_per_second": 46.506,
|
| 568 |
+
"eval_steps_per_second": 2.93,
|
| 569 |
+
"step": 650
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 1.7059393939393939,
|
| 573 |
+
"grad_norm": 113831.8671875,
|
| 574 |
+
"learning_rate": 2.1371470563787788e-05,
|
| 575 |
+
"loss": 1.6563,
|
| 576 |
+
"step": 660
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 1.73179797979798,
|
| 580 |
+
"grad_norm": 90395.5,
|
| 581 |
+
"learning_rate": 2.06666085666794e-05,
|
| 582 |
+
"loss": 1.6484,
|
| 583 |
+
"step": 670
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 1.7576565656565657,
|
| 587 |
+
"grad_norm": 102168.75,
|
| 588 |
+
"learning_rate": 1.9965280945780552e-05,
|
| 589 |
+
"loss": 1.6427,
|
| 590 |
+
"step": 680
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 1.7835151515151515,
|
| 594 |
+
"grad_norm": 107093.0625,
|
| 595 |
+
"learning_rate": 1.9268059713958437e-05,
|
| 596 |
+
"loss": 1.6518,
|
| 597 |
+
"step": 690
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 1.8093737373737375,
|
| 601 |
+
"grad_norm": 91756.90625,
|
| 602 |
+
"learning_rate": 1.857551353485039e-05,
|
| 603 |
+
"loss": 1.6305,
|
| 604 |
+
"step": 700
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 1.8093737373737375,
|
| 608 |
+
"eval_loss": 3.2984097003936768,
|
| 609 |
+
"eval_runtime": 21.4916,
|
| 610 |
+
"eval_samples_per_second": 46.53,
|
| 611 |
+
"eval_steps_per_second": 2.931,
|
| 612 |
+
"step": 700
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 1.835232323232323,
|
| 616 |
+
"grad_norm": 91322.09375,
|
| 617 |
+
"learning_rate": 1.788820725905373e-05,
|
| 618 |
+
"loss": 1.6342,
|
| 619 |
+
"step": 710
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 1.8610909090909091,
|
| 623 |
+
"grad_norm": 99315.203125,
|
| 624 |
+
"learning_rate": 1.7206701463425446e-05,
|
| 625 |
+
"loss": 1.6008,
|
| 626 |
+
"step": 720
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 1.886949494949495,
|
| 630 |
+
"grad_norm": 89938.4921875,
|
| 631 |
+
"learning_rate": 1.6531551993867717e-05,
|
| 632 |
+
"loss": 1.6294,
|
| 633 |
+
"step": 730
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 1.9128080808080807,
|
| 637 |
+
"grad_norm": 91629.78125,
|
| 638 |
+
"learning_rate": 1.5863309511971998e-05,
|
| 639 |
+
"loss": 1.5608,
|
| 640 |
+
"step": 740
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 1.9386666666666668,
|
| 644 |
+
"grad_norm": 91213.9296875,
|
| 645 |
+
"learning_rate": 1.5202519045891558e-05,
|
| 646 |
+
"loss": 1.627,
|
| 647 |
+
"step": 750
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 1.9386666666666668,
|
| 651 |
+
"eval_loss": 3.259477138519287,
|
| 652 |
+
"eval_runtime": 21.4965,
|
| 653 |
+
"eval_samples_per_second": 46.519,
|
| 654 |
+
"eval_steps_per_second": 2.931,
|
| 655 |
+
"step": 750
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 1.9645252525252526,
|
| 659 |
+
"grad_norm": 86313.796875,
|
| 660 |
+
"learning_rate": 1.4549719545808632e-05,
|
| 661 |
+
"loss": 1.5955,
|
| 662 |
+
"step": 760
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 1.9903838383838384,
|
| 666 |
+
"grad_norm": 80786.09375,
|
| 667 |
+
"learning_rate": 1.3905443444359025e-05,
|
| 668 |
+
"loss": 1.5658,
|
| 669 |
+
"step": 770
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 2.0155151515151517,
|
| 673 |
+
"grad_norm": 87623.234375,
|
| 674 |
+
"learning_rate": 1.3270216222372323e-05,
|
| 675 |
+
"loss": 1.5961,
|
| 676 |
+
"step": 780
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 2.0413737373737373,
|
| 680 |
+
"grad_norm": 81999.125,
|
| 681 |
+
"learning_rate": 1.2644555980282219e-05,
|
| 682 |
+
"loss": 1.5716,
|
| 683 |
+
"step": 790
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 2.0672323232323233,
|
| 687 |
+
"grad_norm": 78692.296875,
|
| 688 |
+
"learning_rate": 1.2028973015556413e-05,
|
| 689 |
+
"loss": 1.5748,
|
| 690 |
+
"step": 800
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 2.0672323232323233,
|
| 694 |
+
"eval_loss": 3.231915235519409,
|
| 695 |
+
"eval_runtime": 21.4941,
|
| 696 |
+
"eval_samples_per_second": 46.524,
|
| 697 |
+
"eval_steps_per_second": 2.931,
|
| 698 |
+
"step": 800
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 2.093090909090909,
|
| 702 |
+
"grad_norm": 85943.3359375,
|
| 703 |
+
"learning_rate": 1.142396940649062e-05,
|
| 704 |
+
"loss": 1.5775,
|
| 705 |
+
"step": 810
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 2.118949494949495,
|
| 709 |
+
"grad_norm": 89760.4765625,
|
| 710 |
+
"learning_rate": 1.0830038602706346e-05,
|
| 711 |
+
"loss": 1.5867,
|
| 712 |
+
"step": 820
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 2.144808080808081,
|
| 716 |
+
"grad_norm": 88100.2109375,
|
| 717 |
+
"learning_rate": 1.0247665022686262e-05,
|
| 718 |
+
"loss": 1.5792,
|
| 719 |
+
"step": 830
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 2.1706666666666665,
|
| 723 |
+
"grad_norm": 83300.453125,
|
| 724 |
+
"learning_rate": 9.677323658675594e-06,
|
| 725 |
+
"loss": 1.5703,
|
| 726 |
+
"step": 840
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 2.1965252525252525,
|
| 730 |
+
"grad_norm": 92031.609375,
|
| 731 |
+
"learning_rate": 9.11947968927157e-06,
|
| 732 |
+
"loss": 1.5711,
|
| 733 |
+
"step": 850
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 2.1965252525252525,
|
| 737 |
+
"eval_loss": 3.204101085662842,
|
| 738 |
+
"eval_runtime": 21.5224,
|
| 739 |
+
"eval_samples_per_second": 46.463,
|
| 740 |
+
"eval_steps_per_second": 2.927,
|
| 741 |
+
"step": 850
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 2.2223838383838386,
|
| 745 |
+
"grad_norm": 86591.40625,
|
| 746 |
+
"learning_rate": 8.574588100017159e-06,
|
| 747 |
+
"loss": 1.5752,
|
| 748 |
+
"step": 860
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 2.248242424242424,
|
| 752 |
+
"grad_norm": 101210.9609375,
|
| 753 |
+
"learning_rate": 8.043093312308248e-06,
|
| 754 |
+
"loss": 1.5828,
|
| 755 |
+
"step": 870
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 2.27410101010101,
|
| 759 |
+
"grad_norm": 88097.3984375,
|
| 760 |
+
"learning_rate": 7.525428820917288e-06,
|
| 761 |
+
"loss": 1.5628,
|
| 762 |
+
"step": 880
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 2.2999595959595958,
|
| 766 |
+
"grad_norm": 80520.515625,
|
| 767 |
+
"learning_rate": 7.022016840428614e-06,
|
| 768 |
+
"loss": 1.5892,
|
| 769 |
+
"step": 890
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 2.325818181818182,
|
| 773 |
+
"grad_norm": 71811.4921875,
|
| 774 |
+
"learning_rate": 6.533267960874282e-06,
|
| 775 |
+
"loss": 1.5501,
|
| 776 |
+
"step": 900
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 2.325818181818182,
|
| 780 |
+
"eval_loss": 3.186400890350342,
|
| 781 |
+
"eval_runtime": 21.3699,
|
| 782 |
+
"eval_samples_per_second": 46.795,
|
| 783 |
+
"eval_steps_per_second": 2.948,
|
| 784 |
+
"step": 900
|
| 785 |
+
},
|
| 786 |
+
{
|
| 787 |
+
"epoch": 2.351676767676768,
|
| 788 |
+
"grad_norm": 81277.7578125,
|
| 789 |
+
"learning_rate": 6.059580812850868e-06,
|
| 790 |
+
"loss": 1.5404,
|
| 791 |
+
"step": 910
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 2.3775353535353534,
|
| 795 |
+
"grad_norm": 72905.59375,
|
| 796 |
+
"learning_rate": 5.601341742390706e-06,
|
| 797 |
+
"loss": 1.5642,
|
| 798 |
+
"step": 920
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"epoch": 2.4033939393939394,
|
| 802 |
+
"grad_norm": 74492.1484375,
|
| 803 |
+
"learning_rate": 5.158924495852454e-06,
|
| 804 |
+
"loss": 1.5425,
|
| 805 |
+
"step": 930
|
| 806 |
+
},
|
| 807 |
+
{
|
| 808 |
+
"epoch": 2.4292525252525254,
|
| 809 |
+
"grad_norm": 88411.46875,
|
| 810 |
+
"learning_rate": 4.732689915088262e-06,
|
| 811 |
+
"loss": 1.5386,
|
| 812 |
+
"step": 940
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 2.455111111111111,
|
| 816 |
+
"grad_norm": 85327.9765625,
|
| 817 |
+
"learning_rate": 4.322985643135952e-06,
|
| 818 |
+
"loss": 1.5771,
|
| 819 |
+
"step": 950
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 2.455111111111111,
|
| 823 |
+
"eval_loss": 3.1719841957092285,
|
| 824 |
+
"eval_runtime": 21.5946,
|
| 825 |
+
"eval_samples_per_second": 46.308,
|
| 826 |
+
"eval_steps_per_second": 2.917,
|
| 827 |
+
"step": 950
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 2.480969696969697,
|
| 831 |
+
"grad_norm": 91103.5390625,
|
| 832 |
+
"learning_rate": 3.9301458406763746e-06,
|
| 833 |
+
"loss": 1.5707,
|
| 834 |
+
"step": 960
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 2.506828282828283,
|
| 838 |
+
"grad_norm": 81763.5546875,
|
| 839 |
+
"learning_rate": 3.5544909134871545e-06,
|
| 840 |
+
"loss": 1.5226,
|
| 841 |
+
"step": 970
|
| 842 |
+
},
|
| 843 |
+
{
|
| 844 |
+
"epoch": 2.5326868686868687,
|
| 845 |
+
"grad_norm": 72524.65625,
|
| 846 |
+
"learning_rate": 3.1963272511151555e-06,
|
| 847 |
+
"loss": 1.5312,
|
| 848 |
+
"step": 980
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 2.5585454545454547,
|
| 852 |
+
"grad_norm": 71866.671875,
|
| 853 |
+
"learning_rate": 2.855946976980775e-06,
|
| 854 |
+
"loss": 1.5427,
|
| 855 |
+
"step": 990
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 2.5844040404040403,
|
| 859 |
+
"grad_norm": 65724.171875,
|
| 860 |
+
"learning_rate": 2.533627710117889e-06,
|
| 861 |
+
"loss": 1.5305,
|
| 862 |
+
"step": 1000
|
| 863 |
+
},
|
| 864 |
+
{
|
| 865 |
+
"epoch": 2.5844040404040403,
|
| 866 |
+
"eval_loss": 3.1618518829345703,
|
| 867 |
+
"eval_runtime": 21.5264,
|
| 868 |
+
"eval_samples_per_second": 46.455,
|
| 869 |
+
"eval_steps_per_second": 2.927,
|
| 870 |
+
"step": 1000
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 2.6102626262626263,
|
| 874 |
+
"grad_norm": 72455.2890625,
|
| 875 |
+
"learning_rate": 2.229632338743773e-06,
|
| 876 |
+
"loss": 1.5533,
|
| 877 |
+
"step": 1010
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 2.6361212121212123,
|
| 881 |
+
"grad_norm": 74930.046875,
|
| 882 |
+
"learning_rate": 1.9442088058437113e-06,
|
| 883 |
+
"loss": 1.5636,
|
| 884 |
+
"step": 1020
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 2.661979797979798,
|
| 888 |
+
"grad_norm": 71263.109375,
|
| 889 |
+
"learning_rate": 1.6775899069451512e-06,
|
| 890 |
+
"loss": 1.5899,
|
| 891 |
+
"step": 1030
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 2.687838383838384,
|
| 895 |
+
"grad_norm": 64114.171875,
|
| 896 |
+
"learning_rate": 1.4299931002463047e-06,
|
| 897 |
+
"loss": 1.5348,
|
| 898 |
+
"step": 1040
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 2.7136969696969695,
|
| 902 |
+
"grad_norm": 67584.8671875,
|
| 903 |
+
"learning_rate": 1.2016203292541578e-06,
|
| 904 |
+
"loss": 1.5518,
|
| 905 |
+
"step": 1050
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 2.7136969696969695,
|
| 909 |
+
"eval_loss": 3.1565101146698,
|
| 910 |
+
"eval_runtime": 21.4913,
|
| 911 |
+
"eval_samples_per_second": 46.53,
|
| 912 |
+
"eval_steps_per_second": 2.931,
|
| 913 |
+
"step": 1050
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 2.7395555555555555,
|
| 917 |
+
"grad_norm": 65095.58984375,
|
| 918 |
+
"learning_rate": 9.926578580764234e-07,
|
| 919 |
+
"loss": 1.5439,
|
| 920 |
+
"step": 1060
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 2.7654141414141415,
|
| 924 |
+
"grad_norm": 55857.4296875,
|
| 925 |
+
"learning_rate": 8.032761195018723e-07,
|
| 926 |
+
"loss": 1.5378,
|
| 927 |
+
"step": 1070
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 2.791272727272727,
|
| 931 |
+
"grad_norm": 59888.99609375,
|
| 932 |
+
"learning_rate": 6.336295759929028e-07,
|
| 933 |
+
"loss": 1.5323,
|
| 934 |
+
"step": 1080
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 2.817131313131313,
|
| 938 |
+
"grad_norm": 58884.90234375,
|
| 939 |
+
"learning_rate": 4.83856593703752e-07,
|
| 940 |
+
"loss": 1.5375,
|
| 941 |
+
"step": 1090
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 2.8429898989898987,
|
| 945 |
+
"grad_norm": 54516.77734375,
|
| 946 |
+
"learning_rate": 3.540793296270578e-07,
|
| 947 |
+
"loss": 1.5388,
|
| 948 |
+
"step": 1100
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 2.8429898989898987,
|
| 952 |
+
"eval_loss": 3.153146505355835,
|
| 953 |
+
"eval_runtime": 21.5113,
|
| 954 |
+
"eval_samples_per_second": 46.487,
|
| 955 |
+
"eval_steps_per_second": 2.929,
|
| 956 |
+
"step": 1100
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 2.8688484848484848,
|
| 960 |
+
"grad_norm": 52652.86328125,
|
| 961 |
+
"learning_rate": 2.4440363196087945e-07,
|
| 962 |
+
"loss": 1.5775,
|
| 963 |
+
"step": 1110
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 2.894707070707071,
|
| 967 |
+
"grad_norm": 54156.91796875,
|
| 968 |
+
"learning_rate": 1.5491895377737764e-07,
|
| 969 |
+
"loss": 1.5358,
|
| 970 |
+
"step": 1120
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"epoch": 2.920565656565657,
|
| 974 |
+
"grad_norm": 46885.69921875,
|
| 975 |
+
"learning_rate": 8.569828006361469e-08,
|
| 976 |
+
"loss": 1.5392,
|
| 977 |
+
"step": 1130
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 2.9464242424242424,
|
| 981 |
+
"grad_norm": 50899.3359375,
|
| 982 |
+
"learning_rate": 3.6798068193946714e-08,
|
| 983 |
+
"loss": 1.5469,
|
| 984 |
+
"step": 1140
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 2.9722828282828284,
|
| 988 |
+
"grad_norm": 53669.515625,
|
| 989 |
+
"learning_rate": 8.25820188258275e-09,
|
| 990 |
+
"loss": 1.5226,
|
| 991 |
+
"step": 1150
|
| 992 |
+
},
|
| 993 |
+
{
|
| 994 |
+
"epoch": 2.9722828282828284,
|
| 995 |
+
"eval_loss": 3.152524948120117,
|
| 996 |
+
"eval_runtime": 21.5012,
|
| 997 |
+
"eval_samples_per_second": 46.509,
|
| 998 |
+
"eval_steps_per_second": 2.93,
|
| 999 |
+
"step": 1150
|
| 1000 |
+
}
|
| 1001 |
+
],
|
| 1002 |
+
"logging_steps": 10,
|
| 1003 |
+
"max_steps": 1158,
|
| 1004 |
+
"num_input_tokens_seen": 0,
|
| 1005 |
+
"num_train_epochs": 3,
|
| 1006 |
+
"save_steps": 50,
|
| 1007 |
+
"stateful_callbacks": {
|
| 1008 |
+
"TrainerControl": {
|
| 1009 |
+
"args": {
|
| 1010 |
+
"should_epoch_stop": false,
|
| 1011 |
+
"should_evaluate": false,
|
| 1012 |
+
"should_log": false,
|
| 1013 |
+
"should_save": true,
|
| 1014 |
+
"should_training_stop": true
|
| 1015 |
+
},
|
| 1016 |
+
"attributes": {}
|
| 1017 |
+
}
|
| 1018 |
+
},
|
| 1019 |
+
"total_flos": 4.35914399232e+16,
|
| 1020 |
+
"train_batch_size": 8,
|
| 1021 |
+
"trial_name": null,
|
| 1022 |
+
"trial_params": null
|
| 1023 |
+
}
|