Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +938 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,938 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.5025785565376282,
|
| 3 |
+
"best_model_checkpoint": "models/E-Coli-FFT/KCYHSM/checkpoint-21500",
|
| 4 |
+
"epoch": 8.96,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 28000,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.16,
|
| 13 |
+
"grad_norm": 0.560947835445404,
|
| 14 |
+
"learning_rate": 4.996016e-05,
|
| 15 |
+
"loss": 2.0872,
|
| 16 |
+
"step": 500
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.16,
|
| 20 |
+
"eval_accuracy_per_token": 0.30261626839637756,
|
| 21 |
+
"eval_loss": 2.0467002391815186,
|
| 22 |
+
"eval_runtime": 226.6763,
|
| 23 |
+
"eval_samples_per_second": 110.289,
|
| 24 |
+
"eval_steps_per_second": 6.895,
|
| 25 |
+
"step": 500
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"epoch": 0.32,
|
| 29 |
+
"grad_norm": 0.5276467204093933,
|
| 30 |
+
"learning_rate": 4.992016000000001e-05,
|
| 31 |
+
"loss": 2.033,
|
| 32 |
+
"step": 1000
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.32,
|
| 36 |
+
"eval_accuracy_per_token": 0.31899595260620117,
|
| 37 |
+
"eval_loss": 2.0086700916290283,
|
| 38 |
+
"eval_runtime": 215.1833,
|
| 39 |
+
"eval_samples_per_second": 116.18,
|
| 40 |
+
"eval_steps_per_second": 7.264,
|
| 41 |
+
"step": 1000
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"epoch": 0.48,
|
| 45 |
+
"grad_norm": 1.3343302011489868,
|
| 46 |
+
"learning_rate": 4.988016e-05,
|
| 47 |
+
"loss": 1.9439,
|
| 48 |
+
"step": 1500
|
| 49 |
+
},
|
| 50 |
+
{
|
| 51 |
+
"epoch": 0.48,
|
| 52 |
+
"eval_accuracy_per_token": 0.3831162750720978,
|
| 53 |
+
"eval_loss": 1.8512518405914307,
|
| 54 |
+
"eval_runtime": 215.5299,
|
| 55 |
+
"eval_samples_per_second": 115.993,
|
| 56 |
+
"eval_steps_per_second": 7.252,
|
| 57 |
+
"step": 1500
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.64,
|
| 61 |
+
"grad_norm": 1.4562740325927734,
|
| 62 |
+
"learning_rate": 4.984016e-05,
|
| 63 |
+
"loss": 1.7198,
|
| 64 |
+
"step": 2000
|
| 65 |
+
},
|
| 66 |
+
{
|
| 67 |
+
"epoch": 0.64,
|
| 68 |
+
"eval_accuracy_per_token": 0.47227564454078674,
|
| 69 |
+
"eval_loss": 1.6024138927459717,
|
| 70 |
+
"eval_runtime": 215.7037,
|
| 71 |
+
"eval_samples_per_second": 115.9,
|
| 72 |
+
"eval_steps_per_second": 7.246,
|
| 73 |
+
"step": 2000
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.8,
|
| 77 |
+
"grad_norm": 2.136140823364258,
|
| 78 |
+
"learning_rate": 4.980016e-05,
|
| 79 |
+
"loss": 1.5124,
|
| 80 |
+
"step": 2500
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.8,
|
| 84 |
+
"eval_accuracy_per_token": 0.5323951840400696,
|
| 85 |
+
"eval_loss": 1.4281255006790161,
|
| 86 |
+
"eval_runtime": 215.8041,
|
| 87 |
+
"eval_samples_per_second": 115.846,
|
| 88 |
+
"eval_steps_per_second": 7.243,
|
| 89 |
+
"step": 2500
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 0.96,
|
| 93 |
+
"grad_norm": 1.6090725660324097,
|
| 94 |
+
"learning_rate": 4.976016e-05,
|
| 95 |
+
"loss": 1.3463,
|
| 96 |
+
"step": 3000
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 0.96,
|
| 100 |
+
"eval_accuracy_per_token": 0.5815901160240173,
|
| 101 |
+
"eval_loss": 1.2842084169387817,
|
| 102 |
+
"eval_runtime": 215.7888,
|
| 103 |
+
"eval_samples_per_second": 115.854,
|
| 104 |
+
"eval_steps_per_second": 7.243,
|
| 105 |
+
"step": 3000
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 1.12,
|
| 109 |
+
"grad_norm": 2.4967901706695557,
|
| 110 |
+
"learning_rate": 4.972016e-05,
|
| 111 |
+
"loss": 1.1657,
|
| 112 |
+
"step": 3500
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 1.12,
|
| 116 |
+
"eval_accuracy_per_token": 0.6198335886001587,
|
| 117 |
+
"eval_loss": 1.1727232933044434,
|
| 118 |
+
"eval_runtime": 215.8392,
|
| 119 |
+
"eval_samples_per_second": 115.827,
|
| 120 |
+
"eval_steps_per_second": 7.242,
|
| 121 |
+
"step": 3500
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 1.28,
|
| 125 |
+
"grad_norm": 2.750229597091675,
|
| 126 |
+
"learning_rate": 4.968016e-05,
|
| 127 |
+
"loss": 1.0624,
|
| 128 |
+
"step": 4000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 1.28,
|
| 132 |
+
"eval_accuracy_per_token": 0.6496000289916992,
|
| 133 |
+
"eval_loss": 1.0832889080047607,
|
| 134 |
+
"eval_runtime": 216.0152,
|
| 135 |
+
"eval_samples_per_second": 115.733,
|
| 136 |
+
"eval_steps_per_second": 7.236,
|
| 137 |
+
"step": 4000
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 1.44,
|
| 141 |
+
"grad_norm": 2.4956400394439697,
|
| 142 |
+
"learning_rate": 4.9640160000000003e-05,
|
| 143 |
+
"loss": 0.9957,
|
| 144 |
+
"step": 4500
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 1.44,
|
| 148 |
+
"eval_accuracy_per_token": 0.6765108108520508,
|
| 149 |
+
"eval_loss": 1.0036195516586304,
|
| 150 |
+
"eval_runtime": 215.7199,
|
| 151 |
+
"eval_samples_per_second": 115.891,
|
| 152 |
+
"eval_steps_per_second": 7.246,
|
| 153 |
+
"step": 4500
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 1.6,
|
| 157 |
+
"grad_norm": 2.816002368927002,
|
| 158 |
+
"learning_rate": 4.9600160000000004e-05,
|
| 159 |
+
"loss": 0.913,
|
| 160 |
+
"step": 5000
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 1.6,
|
| 164 |
+
"eval_accuracy_per_token": 0.6984472274780273,
|
| 165 |
+
"eval_loss": 0.9418182969093323,
|
| 166 |
+
"eval_runtime": 215.8333,
|
| 167 |
+
"eval_samples_per_second": 115.83,
|
| 168 |
+
"eval_steps_per_second": 7.242,
|
| 169 |
+
"step": 5000
|
| 170 |
+
},
|
| 171 |
+
{
|
| 172 |
+
"epoch": 1.76,
|
| 173 |
+
"grad_norm": 2.293124198913574,
|
| 174 |
+
"learning_rate": 4.956016e-05,
|
| 175 |
+
"loss": 0.8747,
|
| 176 |
+
"step": 5500
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 1.76,
|
| 180 |
+
"eval_accuracy_per_token": 0.7186679244041443,
|
| 181 |
+
"eval_loss": 0.8788002133369446,
|
| 182 |
+
"eval_runtime": 215.9244,
|
| 183 |
+
"eval_samples_per_second": 115.781,
|
| 184 |
+
"eval_steps_per_second": 7.239,
|
| 185 |
+
"step": 5500
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 1.92,
|
| 189 |
+
"grad_norm": 2.6259233951568604,
|
| 190 |
+
"learning_rate": 4.9520160000000005e-05,
|
| 191 |
+
"loss": 0.8064,
|
| 192 |
+
"step": 6000
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 1.92,
|
| 196 |
+
"eval_accuracy_per_token": 0.7395341992378235,
|
| 197 |
+
"eval_loss": 0.8194286823272705,
|
| 198 |
+
"eval_runtime": 215.8802,
|
| 199 |
+
"eval_samples_per_second": 115.805,
|
| 200 |
+
"eval_steps_per_second": 7.24,
|
| 201 |
+
"step": 6000
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 2.08,
|
| 205 |
+
"grad_norm": 2.7774460315704346,
|
| 206 |
+
"learning_rate": 4.9480160000000005e-05,
|
| 207 |
+
"loss": 0.7329,
|
| 208 |
+
"step": 6500
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 2.08,
|
| 212 |
+
"eval_accuracy_per_token": 0.7511833310127258,
|
| 213 |
+
"eval_loss": 0.7874646782875061,
|
| 214 |
+
"eval_runtime": 215.9936,
|
| 215 |
+
"eval_samples_per_second": 115.744,
|
| 216 |
+
"eval_steps_per_second": 7.236,
|
| 217 |
+
"step": 6500
|
| 218 |
+
},
|
| 219 |
+
{
|
| 220 |
+
"epoch": 2.24,
|
| 221 |
+
"grad_norm": 2.8111910820007324,
|
| 222 |
+
"learning_rate": 4.9440160000000005e-05,
|
| 223 |
+
"loss": 0.6541,
|
| 224 |
+
"step": 7000
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 2.24,
|
| 228 |
+
"eval_accuracy_per_token": 0.7654264569282532,
|
| 229 |
+
"eval_loss": 0.7462261915206909,
|
| 230 |
+
"eval_runtime": 215.7774,
|
| 231 |
+
"eval_samples_per_second": 115.86,
|
| 232 |
+
"eval_steps_per_second": 7.244,
|
| 233 |
+
"step": 7000
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 2.4,
|
| 237 |
+
"grad_norm": 2.734023332595825,
|
| 238 |
+
"learning_rate": 4.940016e-05,
|
| 239 |
+
"loss": 0.6368,
|
| 240 |
+
"step": 7500
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 2.4,
|
| 244 |
+
"eval_accuracy_per_token": 0.7762272953987122,
|
| 245 |
+
"eval_loss": 0.7165001630783081,
|
| 246 |
+
"eval_runtime": 215.5714,
|
| 247 |
+
"eval_samples_per_second": 115.971,
|
| 248 |
+
"eval_steps_per_second": 7.25,
|
| 249 |
+
"step": 7500
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 2.56,
|
| 253 |
+
"grad_norm": 3.085864782333374,
|
| 254 |
+
"learning_rate": 4.936016e-05,
|
| 255 |
+
"loss": 0.6019,
|
| 256 |
+
"step": 8000
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 2.56,
|
| 260 |
+
"eval_accuracy_per_token": 0.7840808033943176,
|
| 261 |
+
"eval_loss": 0.6941312551498413,
|
| 262 |
+
"eval_runtime": 215.5763,
|
| 263 |
+
"eval_samples_per_second": 115.968,
|
| 264 |
+
"eval_steps_per_second": 7.25,
|
| 265 |
+
"step": 8000
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 2.7199999999999998,
|
| 269 |
+
"grad_norm": 2.5468504428863525,
|
| 270 |
+
"learning_rate": 4.9320320000000004e-05,
|
| 271 |
+
"loss": 0.5855,
|
| 272 |
+
"step": 8500
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 2.7199999999999998,
|
| 276 |
+
"eval_accuracy_per_token": 0.7905924916267395,
|
| 277 |
+
"eval_loss": 0.6726610064506531,
|
| 278 |
+
"eval_runtime": 215.5941,
|
| 279 |
+
"eval_samples_per_second": 115.959,
|
| 280 |
+
"eval_steps_per_second": 7.25,
|
| 281 |
+
"step": 8500
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 2.88,
|
| 285 |
+
"grad_norm": 2.7991392612457275,
|
| 286 |
+
"learning_rate": 4.9280320000000005e-05,
|
| 287 |
+
"loss": 0.565,
|
| 288 |
+
"step": 9000
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 2.88,
|
| 292 |
+
"eval_accuracy_per_token": 0.7970213890075684,
|
| 293 |
+
"eval_loss": 0.6574403047561646,
|
| 294 |
+
"eval_runtime": 215.6086,
|
| 295 |
+
"eval_samples_per_second": 115.951,
|
| 296 |
+
"eval_steps_per_second": 7.249,
|
| 297 |
+
"step": 9000
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 3.04,
|
| 301 |
+
"grad_norm": 2.0875983238220215,
|
| 302 |
+
"learning_rate": 4.924032e-05,
|
| 303 |
+
"loss": 0.5437,
|
| 304 |
+
"step": 9500
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 3.04,
|
| 308 |
+
"eval_accuracy_per_token": 0.8036227226257324,
|
| 309 |
+
"eval_loss": 0.6450381278991699,
|
| 310 |
+
"eval_runtime": 215.7698,
|
| 311 |
+
"eval_samples_per_second": 115.864,
|
| 312 |
+
"eval_steps_per_second": 7.244,
|
| 313 |
+
"step": 9500
|
| 314 |
+
},
|
| 315 |
+
{
|
| 316 |
+
"epoch": 3.2,
|
| 317 |
+
"grad_norm": 2.820004940032959,
|
| 318 |
+
"learning_rate": 4.920032e-05,
|
| 319 |
+
"loss": 0.4618,
|
| 320 |
+
"step": 10000
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 3.2,
|
| 324 |
+
"eval_accuracy_per_token": 0.8065482378005981,
|
| 325 |
+
"eval_loss": 0.6290712952613831,
|
| 326 |
+
"eval_runtime": 215.5858,
|
| 327 |
+
"eval_samples_per_second": 115.963,
|
| 328 |
+
"eval_steps_per_second": 7.25,
|
| 329 |
+
"step": 10000
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 3.36,
|
| 333 |
+
"grad_norm": 2.534975051879883,
|
| 334 |
+
"learning_rate": 4.916048e-05,
|
| 335 |
+
"loss": 0.4836,
|
| 336 |
+
"step": 10500
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 3.36,
|
| 340 |
+
"eval_accuracy_per_token": 0.8109478950500488,
|
| 341 |
+
"eval_loss": 0.621654748916626,
|
| 342 |
+
"eval_runtime": 215.6832,
|
| 343 |
+
"eval_samples_per_second": 115.911,
|
| 344 |
+
"eval_steps_per_second": 7.247,
|
| 345 |
+
"step": 10500
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 3.52,
|
| 349 |
+
"grad_norm": 2.2555694580078125,
|
| 350 |
+
"learning_rate": 4.9120480000000004e-05,
|
| 351 |
+
"loss": 0.4793,
|
| 352 |
+
"step": 11000
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 3.52,
|
| 356 |
+
"eval_accuracy_per_token": 0.8153020739555359,
|
| 357 |
+
"eval_loss": 0.6032074093818665,
|
| 358 |
+
"eval_runtime": 215.7864,
|
| 359 |
+
"eval_samples_per_second": 115.855,
|
| 360 |
+
"eval_steps_per_second": 7.243,
|
| 361 |
+
"step": 11000
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 3.68,
|
| 365 |
+
"grad_norm": 2.6011011600494385,
|
| 366 |
+
"learning_rate": 4.9080480000000004e-05,
|
| 367 |
+
"loss": 0.4639,
|
| 368 |
+
"step": 11500
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 3.68,
|
| 372 |
+
"eval_accuracy_per_token": 0.8189014792442322,
|
| 373 |
+
"eval_loss": 0.5884661078453064,
|
| 374 |
+
"eval_runtime": 215.9181,
|
| 375 |
+
"eval_samples_per_second": 115.785,
|
| 376 |
+
"eval_steps_per_second": 7.239,
|
| 377 |
+
"step": 11500
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 3.84,
|
| 381 |
+
"grad_norm": 2.201638698577881,
|
| 382 |
+
"learning_rate": 4.904048e-05,
|
| 383 |
+
"loss": 0.4734,
|
| 384 |
+
"step": 12000
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 3.84,
|
| 388 |
+
"eval_accuracy_per_token": 0.8225154280662537,
|
| 389 |
+
"eval_loss": 0.5901506543159485,
|
| 390 |
+
"eval_runtime": 215.7481,
|
| 391 |
+
"eval_samples_per_second": 115.876,
|
| 392 |
+
"eval_steps_per_second": 7.245,
|
| 393 |
+
"step": 12000
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 4.0,
|
| 397 |
+
"grad_norm": 2.115771770477295,
|
| 398 |
+
"learning_rate": 4.9000480000000005e-05,
|
| 399 |
+
"loss": 0.4589,
|
| 400 |
+
"step": 12500
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"epoch": 4.0,
|
| 404 |
+
"eval_accuracy_per_token": 0.824912428855896,
|
| 405 |
+
"eval_loss": 0.57036292552948,
|
| 406 |
+
"eval_runtime": 215.9409,
|
| 407 |
+
"eval_samples_per_second": 115.772,
|
| 408 |
+
"eval_steps_per_second": 7.238,
|
| 409 |
+
"step": 12500
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 4.16,
|
| 413 |
+
"grad_norm": 2.13425350189209,
|
| 414 |
+
"learning_rate": 4.8960480000000005e-05,
|
| 415 |
+
"loss": 0.3846,
|
| 416 |
+
"step": 13000
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 4.16,
|
| 420 |
+
"eval_accuracy_per_token": 0.8280689716339111,
|
| 421 |
+
"eval_loss": 0.5732296705245972,
|
| 422 |
+
"eval_runtime": 215.8629,
|
| 423 |
+
"eval_samples_per_second": 115.814,
|
| 424 |
+
"eval_steps_per_second": 7.241,
|
| 425 |
+
"step": 13000
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 4.32,
|
| 429 |
+
"grad_norm": 2.227823257446289,
|
| 430 |
+
"learning_rate": 4.892048e-05,
|
| 431 |
+
"loss": 0.3907,
|
| 432 |
+
"step": 13500
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 4.32,
|
| 436 |
+
"eval_accuracy_per_token": 0.8309552073478699,
|
| 437 |
+
"eval_loss": 0.5606500506401062,
|
| 438 |
+
"eval_runtime": 215.8455,
|
| 439 |
+
"eval_samples_per_second": 115.824,
|
| 440 |
+
"eval_steps_per_second": 7.241,
|
| 441 |
+
"step": 13500
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 4.48,
|
| 445 |
+
"grad_norm": 1.9877572059631348,
|
| 446 |
+
"learning_rate": 4.888048e-05,
|
| 447 |
+
"loss": 0.3781,
|
| 448 |
+
"step": 14000
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"epoch": 4.48,
|
| 452 |
+
"eval_accuracy_per_token": 0.833513617515564,
|
| 453 |
+
"eval_loss": 0.5528165102005005,
|
| 454 |
+
"eval_runtime": 216.0002,
|
| 455 |
+
"eval_samples_per_second": 115.741,
|
| 456 |
+
"eval_steps_per_second": 7.236,
|
| 457 |
+
"step": 14000
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 4.64,
|
| 461 |
+
"grad_norm": 2.0553159713745117,
|
| 462 |
+
"learning_rate": 4.884048e-05,
|
| 463 |
+
"loss": 0.3914,
|
| 464 |
+
"step": 14500
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 4.64,
|
| 468 |
+
"eval_accuracy_per_token": 0.8335487246513367,
|
| 469 |
+
"eval_loss": 0.549156904220581,
|
| 470 |
+
"eval_runtime": 215.7201,
|
| 471 |
+
"eval_samples_per_second": 115.891,
|
| 472 |
+
"eval_steps_per_second": 7.246,
|
| 473 |
+
"step": 14500
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 4.8,
|
| 477 |
+
"grad_norm": 2.2741925716400146,
|
| 478 |
+
"learning_rate": 4.880048000000001e-05,
|
| 479 |
+
"loss": 0.3894,
|
| 480 |
+
"step": 15000
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 4.8,
|
| 484 |
+
"eval_accuracy_per_token": 0.8362560272216797,
|
| 485 |
+
"eval_loss": 0.5504783391952515,
|
| 486 |
+
"eval_runtime": 215.7979,
|
| 487 |
+
"eval_samples_per_second": 115.849,
|
| 488 |
+
"eval_steps_per_second": 7.243,
|
| 489 |
+
"step": 15000
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 4.96,
|
| 493 |
+
"grad_norm": 2.08858585357666,
|
| 494 |
+
"learning_rate": 4.876048e-05,
|
| 495 |
+
"loss": 0.3944,
|
| 496 |
+
"step": 15500
|
| 497 |
+
},
|
| 498 |
+
{
|
| 499 |
+
"epoch": 4.96,
|
| 500 |
+
"eval_accuracy_per_token": 0.8395103812217712,
|
| 501 |
+
"eval_loss": 0.5324631929397583,
|
| 502 |
+
"eval_runtime": 215.9188,
|
| 503 |
+
"eval_samples_per_second": 115.784,
|
| 504 |
+
"eval_steps_per_second": 7.239,
|
| 505 |
+
"step": 15500
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 5.12,
|
| 509 |
+
"grad_norm": 2.1368110179901123,
|
| 510 |
+
"learning_rate": 4.872048e-05,
|
| 511 |
+
"loss": 0.3307,
|
| 512 |
+
"step": 16000
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 5.12,
|
| 516 |
+
"eval_accuracy_per_token": 0.839469850063324,
|
| 517 |
+
"eval_loss": 0.5465147495269775,
|
| 518 |
+
"eval_runtime": 216.142,
|
| 519 |
+
"eval_samples_per_second": 115.665,
|
| 520 |
+
"eval_steps_per_second": 7.231,
|
| 521 |
+
"step": 16000
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 5.28,
|
| 525 |
+
"grad_norm": 2.2641913890838623,
|
| 526 |
+
"learning_rate": 4.868048e-05,
|
| 527 |
+
"loss": 0.3125,
|
| 528 |
+
"step": 16500
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 5.28,
|
| 532 |
+
"eval_accuracy_per_token": 0.8418903946876526,
|
| 533 |
+
"eval_loss": 0.5389049053192139,
|
| 534 |
+
"eval_runtime": 216.0271,
|
| 535 |
+
"eval_samples_per_second": 115.726,
|
| 536 |
+
"eval_steps_per_second": 7.235,
|
| 537 |
+
"step": 16500
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 5.44,
|
| 541 |
+
"grad_norm": 1.6883608102798462,
|
| 542 |
+
"learning_rate": 4.864048e-05,
|
| 543 |
+
"loss": 0.3213,
|
| 544 |
+
"step": 17000
|
| 545 |
+
},
|
| 546 |
+
{
|
| 547 |
+
"epoch": 5.44,
|
| 548 |
+
"eval_accuracy_per_token": 0.8448163270950317,
|
| 549 |
+
"eval_loss": 0.5376200675964355,
|
| 550 |
+
"eval_runtime": 216.0239,
|
| 551 |
+
"eval_samples_per_second": 115.728,
|
| 552 |
+
"eval_steps_per_second": 7.235,
|
| 553 |
+
"step": 17000
|
| 554 |
+
},
|
| 555 |
+
{
|
| 556 |
+
"epoch": 5.6,
|
| 557 |
+
"grad_norm": 2.2113685607910156,
|
| 558 |
+
"learning_rate": 4.860048e-05,
|
| 559 |
+
"loss": 0.3269,
|
| 560 |
+
"step": 17500
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 5.6,
|
| 564 |
+
"eval_accuracy_per_token": 0.8447655439376831,
|
| 565 |
+
"eval_loss": 0.5422470569610596,
|
| 566 |
+
"eval_runtime": 215.062,
|
| 567 |
+
"eval_samples_per_second": 116.246,
|
| 568 |
+
"eval_steps_per_second": 7.268,
|
| 569 |
+
"step": 17500
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 5.76,
|
| 573 |
+
"grad_norm": 1.9853556156158447,
|
| 574 |
+
"learning_rate": 4.856048e-05,
|
| 575 |
+
"loss": 0.3374,
|
| 576 |
+
"step": 18000
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 5.76,
|
| 580 |
+
"eval_accuracy_per_token": 0.847917914390564,
|
| 581 |
+
"eval_loss": 0.5255292057991028,
|
| 582 |
+
"eval_runtime": 213.7043,
|
| 583 |
+
"eval_samples_per_second": 116.984,
|
| 584 |
+
"eval_steps_per_second": 7.314,
|
| 585 |
+
"step": 18000
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 5.92,
|
| 589 |
+
"grad_norm": 2.584245204925537,
|
| 590 |
+
"learning_rate": 4.852048e-05,
|
| 591 |
+
"loss": 0.3229,
|
| 592 |
+
"step": 18500
|
| 593 |
+
},
|
| 594 |
+
{
|
| 595 |
+
"epoch": 5.92,
|
| 596 |
+
"eval_accuracy_per_token": 0.8491668701171875,
|
| 597 |
+
"eval_loss": 0.5113908052444458,
|
| 598 |
+
"eval_runtime": 214.1955,
|
| 599 |
+
"eval_samples_per_second": 116.716,
|
| 600 |
+
"eval_steps_per_second": 7.297,
|
| 601 |
+
"step": 18500
|
| 602 |
+
},
|
| 603 |
+
{
|
| 604 |
+
"epoch": 6.08,
|
| 605 |
+
"grad_norm": 2.0866963863372803,
|
| 606 |
+
"learning_rate": 4.848056e-05,
|
| 607 |
+
"loss": 0.3061,
|
| 608 |
+
"step": 19000
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 6.08,
|
| 612 |
+
"eval_accuracy_per_token": 0.8502768874168396,
|
| 613 |
+
"eval_loss": 0.5418649911880493,
|
| 614 |
+
"eval_runtime": 214.0933,
|
| 615 |
+
"eval_samples_per_second": 116.771,
|
| 616 |
+
"eval_steps_per_second": 7.301,
|
| 617 |
+
"step": 19000
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 6.24,
|
| 621 |
+
"grad_norm": 2.0000994205474854,
|
| 622 |
+
"learning_rate": 4.844056e-05,
|
| 623 |
+
"loss": 0.2556,
|
| 624 |
+
"step": 19500
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 6.24,
|
| 628 |
+
"eval_accuracy_per_token": 0.85043865442276,
|
| 629 |
+
"eval_loss": 0.542536199092865,
|
| 630 |
+
"eval_runtime": 213.1241,
|
| 631 |
+
"eval_samples_per_second": 117.303,
|
| 632 |
+
"eval_steps_per_second": 7.334,
|
| 633 |
+
"step": 19500
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 6.4,
|
| 637 |
+
"grad_norm": 2.849112033843994,
|
| 638 |
+
"learning_rate": 4.840064e-05,
|
| 639 |
+
"loss": 0.2672,
|
| 640 |
+
"step": 20000
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 6.4,
|
| 644 |
+
"eval_accuracy_per_token": 0.8512768149375916,
|
| 645 |
+
"eval_loss": 0.5348747372627258,
|
| 646 |
+
"eval_runtime": 215.5865,
|
| 647 |
+
"eval_samples_per_second": 115.963,
|
| 648 |
+
"eval_steps_per_second": 7.25,
|
| 649 |
+
"step": 20000
|
| 650 |
+
},
|
| 651 |
+
{
|
| 652 |
+
"epoch": 6.5600000000000005,
|
| 653 |
+
"grad_norm": 2.216937303543091,
|
| 654 |
+
"learning_rate": 4.836064e-05,
|
| 655 |
+
"loss": 0.2791,
|
| 656 |
+
"step": 20500
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 6.5600000000000005,
|
| 660 |
+
"eval_accuracy_per_token": 0.8536346554756165,
|
| 661 |
+
"eval_loss": 0.5185777544975281,
|
| 662 |
+
"eval_runtime": 215.5365,
|
| 663 |
+
"eval_samples_per_second": 115.99,
|
| 664 |
+
"eval_steps_per_second": 7.252,
|
| 665 |
+
"step": 20500
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 6.72,
|
| 669 |
+
"grad_norm": 2.2237415313720703,
|
| 670 |
+
"learning_rate": 4.832064e-05,
|
| 671 |
+
"loss": 0.2792,
|
| 672 |
+
"step": 21000
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 6.72,
|
| 676 |
+
"eval_accuracy_per_token": 0.8542323112487793,
|
| 677 |
+
"eval_loss": 0.5137789249420166,
|
| 678 |
+
"eval_runtime": 215.6897,
|
| 679 |
+
"eval_samples_per_second": 115.907,
|
| 680 |
+
"eval_steps_per_second": 7.247,
|
| 681 |
+
"step": 21000
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 6.88,
|
| 685 |
+
"grad_norm": 2.6220571994781494,
|
| 686 |
+
"learning_rate": 4.828064e-05,
|
| 687 |
+
"loss": 0.2867,
|
| 688 |
+
"step": 21500
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 6.88,
|
| 692 |
+
"eval_accuracy_per_token": 0.8552057147026062,
|
| 693 |
+
"eval_loss": 0.5025785565376282,
|
| 694 |
+
"eval_runtime": 215.6135,
|
| 695 |
+
"eval_samples_per_second": 115.948,
|
| 696 |
+
"eval_steps_per_second": 7.249,
|
| 697 |
+
"step": 21500
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 7.04,
|
| 701 |
+
"grad_norm": 2.4813005924224854,
|
| 702 |
+
"learning_rate": 4.82408e-05,
|
| 703 |
+
"loss": 0.2628,
|
| 704 |
+
"step": 22000
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 7.04,
|
| 708 |
+
"eval_accuracy_per_token": 0.8553439974784851,
|
| 709 |
+
"eval_loss": 0.5604137778282166,
|
| 710 |
+
"eval_runtime": 215.6781,
|
| 711 |
+
"eval_samples_per_second": 115.913,
|
| 712 |
+
"eval_steps_per_second": 7.247,
|
| 713 |
+
"step": 22000
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 7.2,
|
| 717 |
+
"grad_norm": 2.316831588745117,
|
| 718 |
+
"learning_rate": 4.82008e-05,
|
| 719 |
+
"loss": 0.2124,
|
| 720 |
+
"step": 22500
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 7.2,
|
| 724 |
+
"eval_accuracy_per_token": 0.8566614389419556,
|
| 725 |
+
"eval_loss": 0.5436919927597046,
|
| 726 |
+
"eval_runtime": 215.5335,
|
| 727 |
+
"eval_samples_per_second": 115.991,
|
| 728 |
+
"eval_steps_per_second": 7.252,
|
| 729 |
+
"step": 22500
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 7.36,
|
| 733 |
+
"grad_norm": 2.0445761680603027,
|
| 734 |
+
"learning_rate": 4.81608e-05,
|
| 735 |
+
"loss": 0.2246,
|
| 736 |
+
"step": 23000
|
| 737 |
+
},
|
| 738 |
+
{
|
| 739 |
+
"epoch": 7.36,
|
| 740 |
+
"eval_accuracy_per_token": 0.8579392433166504,
|
| 741 |
+
"eval_loss": 0.5327755212783813,
|
| 742 |
+
"eval_runtime": 215.6442,
|
| 743 |
+
"eval_samples_per_second": 115.932,
|
| 744 |
+
"eval_steps_per_second": 7.248,
|
| 745 |
+
"step": 23000
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 7.52,
|
| 749 |
+
"grad_norm": 1.8908203840255737,
|
| 750 |
+
"learning_rate": 4.81208e-05,
|
| 751 |
+
"loss": 0.2266,
|
| 752 |
+
"step": 23500
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 7.52,
|
| 756 |
+
"eval_accuracy_per_token": 0.8593510389328003,
|
| 757 |
+
"eval_loss": 0.5242588520050049,
|
| 758 |
+
"eval_runtime": 215.5467,
|
| 759 |
+
"eval_samples_per_second": 115.984,
|
| 760 |
+
"eval_steps_per_second": 7.251,
|
| 761 |
+
"step": 23500
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 7.68,
|
| 765 |
+
"grad_norm": 1.698264241218567,
|
| 766 |
+
"learning_rate": 4.80808e-05,
|
| 767 |
+
"loss": 0.2335,
|
| 768 |
+
"step": 24000
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 7.68,
|
| 772 |
+
"eval_accuracy_per_token": 0.8600181341171265,
|
| 773 |
+
"eval_loss": 0.5188168883323669,
|
| 774 |
+
"eval_runtime": 215.5491,
|
| 775 |
+
"eval_samples_per_second": 115.983,
|
| 776 |
+
"eval_steps_per_second": 7.251,
|
| 777 |
+
"step": 24000
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 7.84,
|
| 781 |
+
"grad_norm": 2.196866035461426,
|
| 782 |
+
"learning_rate": 4.80408e-05,
|
| 783 |
+
"loss": 0.2375,
|
| 784 |
+
"step": 24500
|
| 785 |
+
},
|
| 786 |
+
{
|
| 787 |
+
"epoch": 7.84,
|
| 788 |
+
"eval_accuracy_per_token": 0.8607679605484009,
|
| 789 |
+
"eval_loss": 0.5174950957298279,
|
| 790 |
+
"eval_runtime": 215.872,
|
| 791 |
+
"eval_samples_per_second": 115.809,
|
| 792 |
+
"eval_steps_per_second": 7.24,
|
| 793 |
+
"step": 24500
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 8.0,
|
| 797 |
+
"grad_norm": 2.599553346633911,
|
| 798 |
+
"learning_rate": 4.800088e-05,
|
| 799 |
+
"loss": 0.2426,
|
| 800 |
+
"step": 25000
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 8.0,
|
| 804 |
+
"eval_accuracy_per_token": 0.8615555763244629,
|
| 805 |
+
"eval_loss": 0.5121429562568665,
|
| 806 |
+
"eval_runtime": 215.6323,
|
| 807 |
+
"eval_samples_per_second": 115.938,
|
| 808 |
+
"eval_steps_per_second": 7.248,
|
| 809 |
+
"step": 25000
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 8.16,
|
| 813 |
+
"grad_norm": 2.2151308059692383,
|
| 814 |
+
"learning_rate": 4.796088e-05,
|
| 815 |
+
"loss": 0.1708,
|
| 816 |
+
"step": 25500
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 8.16,
|
| 820 |
+
"eval_accuracy_per_token": 0.8606916666030884,
|
| 821 |
+
"eval_loss": 0.5612675547599792,
|
| 822 |
+
"eval_runtime": 215.6202,
|
| 823 |
+
"eval_samples_per_second": 115.945,
|
| 824 |
+
"eval_steps_per_second": 7.249,
|
| 825 |
+
"step": 25500
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 8.32,
|
| 829 |
+
"grad_norm": 2.0478286743164062,
|
| 830 |
+
"learning_rate": 4.792088e-05,
|
| 831 |
+
"loss": 0.1739,
|
| 832 |
+
"step": 26000
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 8.32,
|
| 836 |
+
"eval_accuracy_per_token": 0.8610015511512756,
|
| 837 |
+
"eval_loss": 0.5620591640472412,
|
| 838 |
+
"eval_runtime": 215.5503,
|
| 839 |
+
"eval_samples_per_second": 115.982,
|
| 840 |
+
"eval_steps_per_second": 7.251,
|
| 841 |
+
"step": 26000
|
| 842 |
+
},
|
| 843 |
+
{
|
| 844 |
+
"epoch": 8.48,
|
| 845 |
+
"grad_norm": 2.2109131813049316,
|
| 846 |
+
"learning_rate": 4.788088e-05,
|
| 847 |
+
"loss": 0.1893,
|
| 848 |
+
"step": 26500
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 8.48,
|
| 852 |
+
"eval_accuracy_per_token": 0.861874520778656,
|
| 853 |
+
"eval_loss": 0.5581731200218201,
|
| 854 |
+
"eval_runtime": 215.6555,
|
| 855 |
+
"eval_samples_per_second": 115.926,
|
| 856 |
+
"eval_steps_per_second": 7.248,
|
| 857 |
+
"step": 26500
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 8.64,
|
| 861 |
+
"grad_norm": 2.688985586166382,
|
| 862 |
+
"learning_rate": 4.784096e-05,
|
| 863 |
+
"loss": 0.1948,
|
| 864 |
+
"step": 27000
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 8.64,
|
| 868 |
+
"eval_accuracy_per_token": 0.864142119884491,
|
| 869 |
+
"eval_loss": 0.5396531820297241,
|
| 870 |
+
"eval_runtime": 215.5944,
|
| 871 |
+
"eval_samples_per_second": 115.959,
|
| 872 |
+
"eval_steps_per_second": 7.25,
|
| 873 |
+
"step": 27000
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 8.8,
|
| 877 |
+
"grad_norm": 2.5816333293914795,
|
| 878 |
+
"learning_rate": 4.780096e-05,
|
| 879 |
+
"loss": 0.2022,
|
| 880 |
+
"step": 27500
|
| 881 |
+
},
|
| 882 |
+
{
|
| 883 |
+
"epoch": 8.8,
|
| 884 |
+
"eval_accuracy_per_token": 0.8647276759147644,
|
| 885 |
+
"eval_loss": 0.537101149559021,
|
| 886 |
+
"eval_runtime": 215.522,
|
| 887 |
+
"eval_samples_per_second": 115.997,
|
| 888 |
+
"eval_steps_per_second": 7.252,
|
| 889 |
+
"step": 27500
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 8.96,
|
| 893 |
+
"grad_norm": 2.262164354324341,
|
| 894 |
+
"learning_rate": 4.776096e-05,
|
| 895 |
+
"loss": 0.2027,
|
| 896 |
+
"step": 28000
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 8.96,
|
| 900 |
+
"eval_accuracy_per_token": 0.8652188777923584,
|
| 901 |
+
"eval_loss": 0.5241075158119202,
|
| 902 |
+
"eval_runtime": 215.8588,
|
| 903 |
+
"eval_samples_per_second": 115.816,
|
| 904 |
+
"eval_steps_per_second": 7.241,
|
| 905 |
+
"step": 28000
|
| 906 |
+
}
|
| 907 |
+
],
|
| 908 |
+
"logging_steps": 500,
|
| 909 |
+
"max_steps": 625000,
|
| 910 |
+
"num_input_tokens_seen": 0,
|
| 911 |
+
"num_train_epochs": 200,
|
| 912 |
+
"save_steps": 500,
|
| 913 |
+
"stateful_callbacks": {
|
| 914 |
+
"EarlyStoppingCallback": {
|
| 915 |
+
"args": {
|
| 916 |
+
"early_stopping_patience": 19,
|
| 917 |
+
"early_stopping_threshold": 0.01
|
| 918 |
+
},
|
| 919 |
+
"attributes": {
|
| 920 |
+
"early_stopping_patience_counter": 0
|
| 921 |
+
}
|
| 922 |
+
},
|
| 923 |
+
"TrainerControl": {
|
| 924 |
+
"args": {
|
| 925 |
+
"should_epoch_stop": false,
|
| 926 |
+
"should_evaluate": false,
|
| 927 |
+
"should_log": false,
|
| 928 |
+
"should_save": true,
|
| 929 |
+
"should_training_stop": true
|
| 930 |
+
},
|
| 931 |
+
"attributes": {}
|
| 932 |
+
}
|
| 933 |
+
},
|
| 934 |
+
"total_flos": 1.043038124900352e+18,
|
| 935 |
+
"train_batch_size": 16,
|
| 936 |
+
"trial_name": null,
|
| 937 |
+
"trial_params": null
|
| 938 |
+
}
|