migration: trainer_state.json
Browse files
runs/exp_t1_fusion_sft_20260427_h100/checkpoint-1000/trainer_state.json
ADDED
|
@@ -0,0 +1,750 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.22857142857142856,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 1000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.002285714285714286,
|
| 14 |
+
"grad_norm": 4.173460960388184,
|
| 15 |
+
"learning_rate": 1.3636363636363636e-06,
|
| 16 |
+
"loss": 2.5613529205322267,
|
| 17 |
+
"step": 10
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.004571428571428572,
|
| 21 |
+
"grad_norm": 2.588500499725342,
|
| 22 |
+
"learning_rate": 2.8787878787878793e-06,
|
| 23 |
+
"loss": 2.5523008346557616,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.006857142857142857,
|
| 28 |
+
"grad_norm": 2.0487074851989746,
|
| 29 |
+
"learning_rate": 4.393939393939394e-06,
|
| 30 |
+
"loss": 2.5377822875976563,
|
| 31 |
+
"step": 30
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.009142857142857144,
|
| 35 |
+
"grad_norm": 3.0702173709869385,
|
| 36 |
+
"learning_rate": 5.90909090909091e-06,
|
| 37 |
+
"loss": 2.5055368423461912,
|
| 38 |
+
"step": 40
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.011428571428571429,
|
| 42 |
+
"grad_norm": 1.7121827602386475,
|
| 43 |
+
"learning_rate": 7.424242424242425e-06,
|
| 44 |
+
"loss": 2.4827037811279298,
|
| 45 |
+
"step": 50
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.013714285714285714,
|
| 49 |
+
"grad_norm": 3.549866199493408,
|
| 50 |
+
"learning_rate": 8.93939393939394e-06,
|
| 51 |
+
"loss": 2.414329719543457,
|
| 52 |
+
"step": 60
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.016,
|
| 56 |
+
"grad_norm": 3.476240873336792,
|
| 57 |
+
"learning_rate": 1.0454545454545455e-05,
|
| 58 |
+
"loss": 2.349599266052246,
|
| 59 |
+
"step": 70
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.018285714285714287,
|
| 63 |
+
"grad_norm": 6.9574198722839355,
|
| 64 |
+
"learning_rate": 1.1969696969696971e-05,
|
| 65 |
+
"loss": 2.25291748046875,
|
| 66 |
+
"step": 80
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.02057142857142857,
|
| 70 |
+
"grad_norm": 1.808807134628296,
|
| 71 |
+
"learning_rate": 1.3484848484848486e-05,
|
| 72 |
+
"loss": 2.228924560546875,
|
| 73 |
+
"step": 90
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.022857142857142857,
|
| 77 |
+
"grad_norm": 0.5981293320655823,
|
| 78 |
+
"learning_rate": 1.5000000000000002e-05,
|
| 79 |
+
"loss": 2.197663688659668,
|
| 80 |
+
"step": 100
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.025142857142857144,
|
| 84 |
+
"grad_norm": 0.7301533222198486,
|
| 85 |
+
"learning_rate": 1.6515151515151517e-05,
|
| 86 |
+
"loss": 2.2144283294677733,
|
| 87 |
+
"step": 110
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.027428571428571427,
|
| 91 |
+
"grad_norm": 0.7283045053482056,
|
| 92 |
+
"learning_rate": 1.803030303030303e-05,
|
| 93 |
+
"loss": 2.1870325088500975,
|
| 94 |
+
"step": 120
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.029714285714285714,
|
| 98 |
+
"grad_norm": 0.7445758581161499,
|
| 99 |
+
"learning_rate": 1.9545454545454546e-05,
|
| 100 |
+
"loss": 2.20316104888916,
|
| 101 |
+
"step": 130
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.032,
|
| 105 |
+
"grad_norm": 0.5618444681167603,
|
| 106 |
+
"learning_rate": 1.996700447796371e-05,
|
| 107 |
+
"loss": 2.1872159957885744,
|
| 108 |
+
"step": 140
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.03428571428571429,
|
| 112 |
+
"grad_norm": 0.40402787923812866,
|
| 113 |
+
"learning_rate": 1.9919868017911858e-05,
|
| 114 |
+
"loss": 2.170643615722656,
|
| 115 |
+
"step": 150
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.036571428571428574,
|
| 119 |
+
"grad_norm": 0.45000511407852173,
|
| 120 |
+
"learning_rate": 1.9872731557860004e-05,
|
| 121 |
+
"loss": 2.1895257949829103,
|
| 122 |
+
"step": 160
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.038857142857142854,
|
| 126 |
+
"grad_norm": 0.5675016045570374,
|
| 127 |
+
"learning_rate": 1.9825595097808157e-05,
|
| 128 |
+
"loss": 2.208065414428711,
|
| 129 |
+
"step": 170
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.04114285714285714,
|
| 133 |
+
"grad_norm": 0.5914866924285889,
|
| 134 |
+
"learning_rate": 1.9778458637756306e-05,
|
| 135 |
+
"loss": 2.191765022277832,
|
| 136 |
+
"step": 180
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.04342857142857143,
|
| 140 |
+
"grad_norm": 0.5335065722465515,
|
| 141 |
+
"learning_rate": 1.9731322177704455e-05,
|
| 142 |
+
"loss": 2.1627159118652344,
|
| 143 |
+
"step": 190
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.045714285714285714,
|
| 147 |
+
"grad_norm": 0.9274171590805054,
|
| 148 |
+
"learning_rate": 1.9684185717652608e-05,
|
| 149 |
+
"loss": 2.1878808975219726,
|
| 150 |
+
"step": 200
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.048,
|
| 154 |
+
"grad_norm": 0.7307443618774414,
|
| 155 |
+
"learning_rate": 1.9637049257600757e-05,
|
| 156 |
+
"loss": 2.192724609375,
|
| 157 |
+
"step": 210
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.05028571428571429,
|
| 161 |
+
"grad_norm": 0.6710821390151978,
|
| 162 |
+
"learning_rate": 1.9589912797548906e-05,
|
| 163 |
+
"loss": 2.1558170318603516,
|
| 164 |
+
"step": 220
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.052571428571428575,
|
| 168 |
+
"grad_norm": 0.8309326171875,
|
| 169 |
+
"learning_rate": 1.9542776337497055e-05,
|
| 170 |
+
"loss": 2.1750204086303713,
|
| 171 |
+
"step": 230
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.054857142857142854,
|
| 175 |
+
"grad_norm": 0.9823929071426392,
|
| 176 |
+
"learning_rate": 1.9495639877445205e-05,
|
| 177 |
+
"loss": 2.158173942565918,
|
| 178 |
+
"step": 240
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.05714285714285714,
|
| 182 |
+
"grad_norm": 0.8871349692344666,
|
| 183 |
+
"learning_rate": 1.9448503417393357e-05,
|
| 184 |
+
"loss": 2.1888866424560547,
|
| 185 |
+
"step": 250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.05942857142857143,
|
| 189 |
+
"grad_norm": 0.6405887007713318,
|
| 190 |
+
"learning_rate": 1.9401366957341506e-05,
|
| 191 |
+
"loss": 2.169106674194336,
|
| 192 |
+
"step": 260
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.061714285714285715,
|
| 196 |
+
"grad_norm": 0.7272316217422485,
|
| 197 |
+
"learning_rate": 1.9354230497289656e-05,
|
| 198 |
+
"loss": 2.1696027755737304,
|
| 199 |
+
"step": 270
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.064,
|
| 203 |
+
"grad_norm": 0.6185906529426575,
|
| 204 |
+
"learning_rate": 1.9307094037237805e-05,
|
| 205 |
+
"loss": 2.1781539916992188,
|
| 206 |
+
"step": 280
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.06628571428571428,
|
| 210 |
+
"grad_norm": 0.7201874852180481,
|
| 211 |
+
"learning_rate": 1.9259957577185954e-05,
|
| 212 |
+
"loss": 2.15869197845459,
|
| 213 |
+
"step": 290
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.06857142857142857,
|
| 217 |
+
"grad_norm": 1.0094012022018433,
|
| 218 |
+
"learning_rate": 1.9212821117134103e-05,
|
| 219 |
+
"loss": 2.162426567077637,
|
| 220 |
+
"step": 300
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.07085714285714285,
|
| 224 |
+
"grad_norm": 0.6247529983520508,
|
| 225 |
+
"learning_rate": 1.9165684657082256e-05,
|
| 226 |
+
"loss": 2.149643325805664,
|
| 227 |
+
"step": 310
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.07314285714285715,
|
| 231 |
+
"grad_norm": 0.7491098642349243,
|
| 232 |
+
"learning_rate": 1.9118548197030405e-05,
|
| 233 |
+
"loss": 2.179397201538086,
|
| 234 |
+
"step": 320
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.07542857142857143,
|
| 238 |
+
"grad_norm": 0.7959390878677368,
|
| 239 |
+
"learning_rate": 1.9071411736978555e-05,
|
| 240 |
+
"loss": 2.1876123428344725,
|
| 241 |
+
"step": 330
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.07771428571428571,
|
| 245 |
+
"grad_norm": 0.7109071612358093,
|
| 246 |
+
"learning_rate": 1.9024275276926704e-05,
|
| 247 |
+
"loss": 2.1702051162719727,
|
| 248 |
+
"step": 340
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.08,
|
| 252 |
+
"grad_norm": 0.6978886127471924,
|
| 253 |
+
"learning_rate": 1.8977138816874853e-05,
|
| 254 |
+
"loss": 2.180721092224121,
|
| 255 |
+
"step": 350
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.08228571428571428,
|
| 259 |
+
"grad_norm": 0.761053740978241,
|
| 260 |
+
"learning_rate": 1.8930002356823006e-05,
|
| 261 |
+
"loss": 2.177662658691406,
|
| 262 |
+
"step": 360
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.08457142857142858,
|
| 266 |
+
"grad_norm": 0.6913453936576843,
|
| 267 |
+
"learning_rate": 1.8882865896771155e-05,
|
| 268 |
+
"loss": 2.1632457733154298,
|
| 269 |
+
"step": 370
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.08685714285714285,
|
| 273 |
+
"grad_norm": 0.5731646418571472,
|
| 274 |
+
"learning_rate": 1.8835729436719304e-05,
|
| 275 |
+
"loss": 2.1325332641601564,
|
| 276 |
+
"step": 380
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.08914285714285715,
|
| 280 |
+
"grad_norm": 0.946429967880249,
|
| 281 |
+
"learning_rate": 1.8788592976667453e-05,
|
| 282 |
+
"loss": 2.1733421325683593,
|
| 283 |
+
"step": 390
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.09142857142857143,
|
| 287 |
+
"grad_norm": 0.9916898012161255,
|
| 288 |
+
"learning_rate": 1.8741456516615603e-05,
|
| 289 |
+
"loss": 2.166789436340332,
|
| 290 |
+
"step": 400
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.09371428571428571,
|
| 294 |
+
"grad_norm": 0.9256937503814697,
|
| 295 |
+
"learning_rate": 1.8694320056563752e-05,
|
| 296 |
+
"loss": 2.177092361450195,
|
| 297 |
+
"step": 410
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.096,
|
| 301 |
+
"grad_norm": 0.7875392436981201,
|
| 302 |
+
"learning_rate": 1.8647183596511905e-05,
|
| 303 |
+
"loss": 2.1483591079711912,
|
| 304 |
+
"step": 420
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.09828571428571428,
|
| 308 |
+
"grad_norm": 0.7831864356994629,
|
| 309 |
+
"learning_rate": 1.8600047136460054e-05,
|
| 310 |
+
"loss": 2.1530668258666994,
|
| 311 |
+
"step": 430
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.10057142857142858,
|
| 315 |
+
"grad_norm": 0.7145215272903442,
|
| 316 |
+
"learning_rate": 1.8552910676408203e-05,
|
| 317 |
+
"loss": 2.1778982162475584,
|
| 318 |
+
"step": 440
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.10285714285714286,
|
| 322 |
+
"grad_norm": 0.7052364349365234,
|
| 323 |
+
"learning_rate": 1.8505774216356352e-05,
|
| 324 |
+
"loss": 2.166706085205078,
|
| 325 |
+
"step": 450
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.10514285714285715,
|
| 329 |
+
"grad_norm": 0.7628682255744934,
|
| 330 |
+
"learning_rate": 1.84586377563045e-05,
|
| 331 |
+
"loss": 2.1655509948730467,
|
| 332 |
+
"step": 460
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.10742857142857143,
|
| 336 |
+
"grad_norm": 0.6400886178016663,
|
| 337 |
+
"learning_rate": 1.8411501296252654e-05,
|
| 338 |
+
"loss": 2.1709674835205077,
|
| 339 |
+
"step": 470
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.10971428571428571,
|
| 343 |
+
"grad_norm": 1.3466713428497314,
|
| 344 |
+
"learning_rate": 1.8364364836200803e-05,
|
| 345 |
+
"loss": 2.167324256896973,
|
| 346 |
+
"step": 480
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.112,
|
| 350 |
+
"grad_norm": 0.9079979658126831,
|
| 351 |
+
"learning_rate": 1.8317228376148953e-05,
|
| 352 |
+
"loss": 2.1815256118774413,
|
| 353 |
+
"step": 490
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.11428571428571428,
|
| 357 |
+
"grad_norm": 0.837339460849762,
|
| 358 |
+
"learning_rate": 1.8270091916097102e-05,
|
| 359 |
+
"loss": 2.169051742553711,
|
| 360 |
+
"step": 500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.11428571428571428,
|
| 364 |
+
"eval_loss": 2.171773910522461,
|
| 365 |
+
"eval_runtime": 1237.5541,
|
| 366 |
+
"eval_samples_per_second": 5.656,
|
| 367 |
+
"eval_steps_per_second": 1.414,
|
| 368 |
+
"step": 500
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 0.11657142857142858,
|
| 372 |
+
"grad_norm": 1.8823212385177612,
|
| 373 |
+
"learning_rate": 1.822295545604525e-05,
|
| 374 |
+
"loss": 2.1702648162841798,
|
| 375 |
+
"step": 510
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.11885714285714286,
|
| 379 |
+
"grad_norm": 0.9993029236793518,
|
| 380 |
+
"learning_rate": 1.81758189959934e-05,
|
| 381 |
+
"loss": 2.185104751586914,
|
| 382 |
+
"step": 520
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 0.12114285714285715,
|
| 386 |
+
"grad_norm": 0.9917933344841003,
|
| 387 |
+
"learning_rate": 1.8128682535941553e-05,
|
| 388 |
+
"loss": 2.1710590362548827,
|
| 389 |
+
"step": 530
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 0.12342857142857143,
|
| 393 |
+
"grad_norm": 1.0610461235046387,
|
| 394 |
+
"learning_rate": 1.8081546075889702e-05,
|
| 395 |
+
"loss": 2.15472297668457,
|
| 396 |
+
"step": 540
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 0.12571428571428572,
|
| 400 |
+
"grad_norm": 0.8838741183280945,
|
| 401 |
+
"learning_rate": 1.803440961583785e-05,
|
| 402 |
+
"loss": 2.1636852264404296,
|
| 403 |
+
"step": 550
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 0.128,
|
| 407 |
+
"grad_norm": 0.5796688795089722,
|
| 408 |
+
"learning_rate": 1.7987273155786004e-05,
|
| 409 |
+
"loss": 2.1625608444213866,
|
| 410 |
+
"step": 560
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 0.13028571428571428,
|
| 414 |
+
"grad_norm": 0.7740033864974976,
|
| 415 |
+
"learning_rate": 1.7940136695734153e-05,
|
| 416 |
+
"loss": 2.1263044357299803,
|
| 417 |
+
"step": 570
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.13257142857142856,
|
| 421 |
+
"grad_norm": 0.7221623659133911,
|
| 422 |
+
"learning_rate": 1.78930002356823e-05,
|
| 423 |
+
"loss": 2.1715375900268556,
|
| 424 |
+
"step": 580
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 0.13485714285714287,
|
| 428 |
+
"grad_norm": 0.7279105186462402,
|
| 429 |
+
"learning_rate": 1.7845863775630452e-05,
|
| 430 |
+
"loss": 2.1706336975097655,
|
| 431 |
+
"step": 590
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.13714285714285715,
|
| 435 |
+
"grad_norm": 0.957526683807373,
|
| 436 |
+
"learning_rate": 1.77987273155786e-05,
|
| 437 |
+
"loss": 2.1569234848022463,
|
| 438 |
+
"step": 600
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 0.13942857142857143,
|
| 442 |
+
"grad_norm": 0.7820221781730652,
|
| 443 |
+
"learning_rate": 1.7751590855526754e-05,
|
| 444 |
+
"loss": 2.1689067840576173,
|
| 445 |
+
"step": 610
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 0.1417142857142857,
|
| 449 |
+
"grad_norm": 2.0388848781585693,
|
| 450 |
+
"learning_rate": 1.7704454395474903e-05,
|
| 451 |
+
"loss": 2.1743133544921873,
|
| 452 |
+
"step": 620
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 0.144,
|
| 456 |
+
"grad_norm": 1.6090975999832153,
|
| 457 |
+
"learning_rate": 1.7657317935423052e-05,
|
| 458 |
+
"loss": 2.1662643432617186,
|
| 459 |
+
"step": 630
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.1462857142857143,
|
| 463 |
+
"grad_norm": 2.1735026836395264,
|
| 464 |
+
"learning_rate": 1.76101814753712e-05,
|
| 465 |
+
"loss": 2.1489837646484373,
|
| 466 |
+
"step": 640
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 0.14857142857142858,
|
| 470 |
+
"grad_norm": 0.9704180955886841,
|
| 471 |
+
"learning_rate": 1.756304501531935e-05,
|
| 472 |
+
"loss": 2.1514299392700194,
|
| 473 |
+
"step": 650
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 0.15085714285714286,
|
| 477 |
+
"grad_norm": 1.312035083770752,
|
| 478 |
+
"learning_rate": 1.75159085552675e-05,
|
| 479 |
+
"loss": 2.154615020751953,
|
| 480 |
+
"step": 660
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 0.15314285714285714,
|
| 484 |
+
"grad_norm": 1.2137097120285034,
|
| 485 |
+
"learning_rate": 1.7468772095215653e-05,
|
| 486 |
+
"loss": 2.157608222961426,
|
| 487 |
+
"step": 670
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 0.15542857142857142,
|
| 491 |
+
"grad_norm": 0.7296802401542664,
|
| 492 |
+
"learning_rate": 1.7421635635163802e-05,
|
| 493 |
+
"loss": 2.1513166427612305,
|
| 494 |
+
"step": 680
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 0.15771428571428572,
|
| 498 |
+
"grad_norm": 1.8267555236816406,
|
| 499 |
+
"learning_rate": 1.737449917511195e-05,
|
| 500 |
+
"loss": 2.175881195068359,
|
| 501 |
+
"step": 690
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.16,
|
| 505 |
+
"grad_norm": 0.9844738841056824,
|
| 506 |
+
"learning_rate": 1.73273627150601e-05,
|
| 507 |
+
"loss": 2.1672784805297853,
|
| 508 |
+
"step": 700
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 0.16228571428571428,
|
| 512 |
+
"grad_norm": 0.8195302486419678,
|
| 513 |
+
"learning_rate": 1.728022625500825e-05,
|
| 514 |
+
"loss": 2.154075050354004,
|
| 515 |
+
"step": 710
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 0.16457142857142856,
|
| 519 |
+
"grad_norm": 1.0060882568359375,
|
| 520 |
+
"learning_rate": 1.72330897949564e-05,
|
| 521 |
+
"loss": 2.1396883010864256,
|
| 522 |
+
"step": 720
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 0.16685714285714287,
|
| 526 |
+
"grad_norm": 1.0614882707595825,
|
| 527 |
+
"learning_rate": 1.718595333490455e-05,
|
| 528 |
+
"loss": 2.167150688171387,
|
| 529 |
+
"step": 730
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 0.16914285714285715,
|
| 533 |
+
"grad_norm": 0.9078527092933655,
|
| 534 |
+
"learning_rate": 1.71388168748527e-05,
|
| 535 |
+
"loss": 2.157125473022461,
|
| 536 |
+
"step": 740
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 0.17142857142857143,
|
| 540 |
+
"grad_norm": 0.7285545468330383,
|
| 541 |
+
"learning_rate": 1.709168041480085e-05,
|
| 542 |
+
"loss": 2.1346229553222655,
|
| 543 |
+
"step": 750
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 0.1737142857142857,
|
| 547 |
+
"grad_norm": 0.7541137337684631,
|
| 548 |
+
"learning_rate": 1.7044543954749e-05,
|
| 549 |
+
"loss": 2.1761722564697266,
|
| 550 |
+
"step": 760
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 0.176,
|
| 554 |
+
"grad_norm": 1.071502923965454,
|
| 555 |
+
"learning_rate": 1.699740749469715e-05,
|
| 556 |
+
"loss": 2.162173843383789,
|
| 557 |
+
"step": 770
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 0.1782857142857143,
|
| 561 |
+
"grad_norm": 0.724347710609436,
|
| 562 |
+
"learning_rate": 1.69502710346453e-05,
|
| 563 |
+
"loss": 2.1582818984985352,
|
| 564 |
+
"step": 780
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 0.18057142857142858,
|
| 568 |
+
"grad_norm": 1.0418791770935059,
|
| 569 |
+
"learning_rate": 1.690313457459345e-05,
|
| 570 |
+
"loss": 2.157000923156738,
|
| 571 |
+
"step": 790
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 0.18285714285714286,
|
| 575 |
+
"grad_norm": 1.191541075706482,
|
| 576 |
+
"learning_rate": 1.68559981145416e-05,
|
| 577 |
+
"loss": 2.1548776626586914,
|
| 578 |
+
"step": 800
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 0.18514285714285714,
|
| 582 |
+
"grad_norm": 2.068469762802124,
|
| 583 |
+
"learning_rate": 1.680886165448975e-05,
|
| 584 |
+
"loss": 2.1657318115234374,
|
| 585 |
+
"step": 810
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.18742857142857142,
|
| 589 |
+
"grad_norm": 0.9738569259643555,
|
| 590 |
+
"learning_rate": 1.6761725194437898e-05,
|
| 591 |
+
"loss": 2.1749725341796875,
|
| 592 |
+
"step": 820
|
| 593 |
+
},
|
| 594 |
+
{
|
| 595 |
+
"epoch": 0.18971428571428572,
|
| 596 |
+
"grad_norm": 1.656999945640564,
|
| 597 |
+
"learning_rate": 1.6714588734386047e-05,
|
| 598 |
+
"loss": 2.160741996765137,
|
| 599 |
+
"step": 830
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 0.192,
|
| 603 |
+
"grad_norm": 1.535428762435913,
|
| 604 |
+
"learning_rate": 1.66674522743342e-05,
|
| 605 |
+
"loss": 2.1628345489501952,
|
| 606 |
+
"step": 840
|
| 607 |
+
},
|
| 608 |
+
{
|
| 609 |
+
"epoch": 0.19428571428571428,
|
| 610 |
+
"grad_norm": 1.3933148384094238,
|
| 611 |
+
"learning_rate": 1.662031581428235e-05,
|
| 612 |
+
"loss": 2.1727434158325196,
|
| 613 |
+
"step": 850
|
| 614 |
+
},
|
| 615 |
+
{
|
| 616 |
+
"epoch": 0.19657142857142856,
|
| 617 |
+
"grad_norm": 1.040420413017273,
|
| 618 |
+
"learning_rate": 1.6573179354230498e-05,
|
| 619 |
+
"loss": 2.1670541763305664,
|
| 620 |
+
"step": 860
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 0.19885714285714284,
|
| 624 |
+
"grad_norm": 0.9024999141693115,
|
| 625 |
+
"learning_rate": 1.6526042894178648e-05,
|
| 626 |
+
"loss": 2.172655487060547,
|
| 627 |
+
"step": 870
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 0.20114285714285715,
|
| 631 |
+
"grad_norm": 1.2898517847061157,
|
| 632 |
+
"learning_rate": 1.6478906434126797e-05,
|
| 633 |
+
"loss": 2.1662887573242187,
|
| 634 |
+
"step": 880
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 0.20342857142857143,
|
| 638 |
+
"grad_norm": 1.0849480628967285,
|
| 639 |
+
"learning_rate": 1.643176997407495e-05,
|
| 640 |
+
"loss": 2.1510431289672853,
|
| 641 |
+
"step": 890
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 0.2057142857142857,
|
| 645 |
+
"grad_norm": 0.8024755716323853,
|
| 646 |
+
"learning_rate": 1.63846335140231e-05,
|
| 647 |
+
"loss": 2.1584331512451174,
|
| 648 |
+
"step": 900
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 0.208,
|
| 652 |
+
"grad_norm": 1.9280140399932861,
|
| 653 |
+
"learning_rate": 1.6337497053971248e-05,
|
| 654 |
+
"loss": 2.1537303924560547,
|
| 655 |
+
"step": 910
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 0.2102857142857143,
|
| 659 |
+
"grad_norm": 1.0073834657669067,
|
| 660 |
+
"learning_rate": 1.6290360593919397e-05,
|
| 661 |
+
"loss": 2.1719333648681642,
|
| 662 |
+
"step": 920
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 0.21257142857142858,
|
| 666 |
+
"grad_norm": 1.2503358125686646,
|
| 667 |
+
"learning_rate": 1.6243224133867546e-05,
|
| 668 |
+
"loss": 2.1825456619262695,
|
| 669 |
+
"step": 930
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 0.21485714285714286,
|
| 673 |
+
"grad_norm": 1.8010166883468628,
|
| 674 |
+
"learning_rate": 1.6196087673815696e-05,
|
| 675 |
+
"loss": 2.1685020446777346,
|
| 676 |
+
"step": 940
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 0.21714285714285714,
|
| 680 |
+
"grad_norm": 1.0193595886230469,
|
| 681 |
+
"learning_rate": 1.6148951213763848e-05,
|
| 682 |
+
"loss": 2.158196449279785,
|
| 683 |
+
"step": 950
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 0.21942857142857142,
|
| 687 |
+
"grad_norm": 1.8305720090866089,
|
| 688 |
+
"learning_rate": 1.6101814753711997e-05,
|
| 689 |
+
"loss": 2.141801643371582,
|
| 690 |
+
"step": 960
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 0.22171428571428572,
|
| 694 |
+
"grad_norm": 1.456999659538269,
|
| 695 |
+
"learning_rate": 1.6054678293660147e-05,
|
| 696 |
+
"loss": 2.1524885177612303,
|
| 697 |
+
"step": 970
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 0.224,
|
| 701 |
+
"grad_norm": 1.3395929336547852,
|
| 702 |
+
"learning_rate": 1.60075418336083e-05,
|
| 703 |
+
"loss": 2.158906173706055,
|
| 704 |
+
"step": 980
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 0.22628571428571428,
|
| 708 |
+
"grad_norm": 1.1533492803573608,
|
| 709 |
+
"learning_rate": 1.5960405373556445e-05,
|
| 710 |
+
"loss": 2.145818901062012,
|
| 711 |
+
"step": 990
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 0.22857142857142856,
|
| 715 |
+
"grad_norm": 1.014697551727295,
|
| 716 |
+
"learning_rate": 1.5913268913504598e-05,
|
| 717 |
+
"loss": 2.165022277832031,
|
| 718 |
+
"step": 1000
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 0.22857142857142856,
|
| 722 |
+
"eval_loss": 2.1635704040527344,
|
| 723 |
+
"eval_runtime": 1167.8266,
|
| 724 |
+
"eval_samples_per_second": 5.994,
|
| 725 |
+
"eval_steps_per_second": 1.499,
|
| 726 |
+
"step": 1000
|
| 727 |
+
}
|
| 728 |
+
],
|
| 729 |
+
"logging_steps": 10,
|
| 730 |
+
"max_steps": 4375,
|
| 731 |
+
"num_input_tokens_seen": 0,
|
| 732 |
+
"num_train_epochs": 1,
|
| 733 |
+
"save_steps": 1000,
|
| 734 |
+
"stateful_callbacks": {
|
| 735 |
+
"TrainerControl": {
|
| 736 |
+
"args": {
|
| 737 |
+
"should_epoch_stop": false,
|
| 738 |
+
"should_evaluate": false,
|
| 739 |
+
"should_log": false,
|
| 740 |
+
"should_save": true,
|
| 741 |
+
"should_training_stop": false
|
| 742 |
+
},
|
| 743 |
+
"attributes": {}
|
| 744 |
+
}
|
| 745 |
+
},
|
| 746 |
+
"total_flos": 0.0,
|
| 747 |
+
"train_batch_size": 4,
|
| 748 |
+
"trial_name": null,
|
| 749 |
+
"trial_params": null
|
| 750 |
+
}
|