Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +553 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,553 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 10.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 730,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.136986301369863,
|
| 13 |
+
"grad_norm": 5.2195892333984375,
|
| 14 |
+
"learning_rate": 2.702702702702703e-05,
|
| 15 |
+
"loss": 0.9148,
|
| 16 |
+
"step": 10
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.273972602739726,
|
| 20 |
+
"grad_norm": 2.442189931869507,
|
| 21 |
+
"learning_rate": 5.405405405405406e-05,
|
| 22 |
+
"loss": 0.6052,
|
| 23 |
+
"step": 20
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.410958904109589,
|
| 27 |
+
"grad_norm": 2.444631576538086,
|
| 28 |
+
"learning_rate": 8.108108108108109e-05,
|
| 29 |
+
"loss": 0.4321,
|
| 30 |
+
"step": 30
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.547945205479452,
|
| 34 |
+
"grad_norm": 1.6064536571502686,
|
| 35 |
+
"learning_rate": 9.999537609289592e-05,
|
| 36 |
+
"loss": 0.328,
|
| 37 |
+
"step": 40
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.684931506849315,
|
| 41 |
+
"grad_norm": 0.9033578038215637,
|
| 42 |
+
"learning_rate": 9.991319708909113e-05,
|
| 43 |
+
"loss": 0.2676,
|
| 44 |
+
"step": 50
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.821917808219178,
|
| 48 |
+
"grad_norm": 0.881839394569397,
|
| 49 |
+
"learning_rate": 9.972845896501761e-05,
|
| 50 |
+
"loss": 0.241,
|
| 51 |
+
"step": 60
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.958904109589041,
|
| 55 |
+
"grad_norm": 0.7620624899864197,
|
| 56 |
+
"learning_rate": 9.944154131125642e-05,
|
| 57 |
+
"loss": 0.2151,
|
| 58 |
+
"step": 70
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 1.095890410958904,
|
| 62 |
+
"grad_norm": 0.8047911524772644,
|
| 63 |
+
"learning_rate": 9.905303367173336e-05,
|
| 64 |
+
"loss": 0.1981,
|
| 65 |
+
"step": 80
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 1.2328767123287672,
|
| 69 |
+
"grad_norm": 0.789812445640564,
|
| 70 |
+
"learning_rate": 9.856373433235373e-05,
|
| 71 |
+
"loss": 0.1722,
|
| 72 |
+
"step": 90
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 1.36986301369863,
|
| 76 |
+
"grad_norm": 0.8034406900405884,
|
| 77 |
+
"learning_rate": 9.797464868072488e-05,
|
| 78 |
+
"loss": 0.1565,
|
| 79 |
+
"step": 100
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 1.5068493150684932,
|
| 83 |
+
"grad_norm": 0.6497238278388977,
|
| 84 |
+
"learning_rate": 9.72869871403363e-05,
|
| 85 |
+
"loss": 0.1428,
|
| 86 |
+
"step": 110
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 1.643835616438356,
|
| 90 |
+
"grad_norm": 0.7455314993858337,
|
| 91 |
+
"learning_rate": 9.650216268344262e-05,
|
| 92 |
+
"loss": 0.136,
|
| 93 |
+
"step": 120
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 1.7808219178082192,
|
| 97 |
+
"grad_norm": 0.6564836502075195,
|
| 98 |
+
"learning_rate": 9.562178792775936e-05,
|
| 99 |
+
"loss": 0.1279,
|
| 100 |
+
"step": 130
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 1.9178082191780823,
|
| 104 |
+
"grad_norm": 0.7567775845527649,
|
| 105 |
+
"learning_rate": 9.464767182293739e-05,
|
| 106 |
+
"loss": 0.1189,
|
| 107 |
+
"step": 140
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 2.0547945205479454,
|
| 111 |
+
"grad_norm": 0.5241563320159912,
|
| 112 |
+
"learning_rate": 9.35818159336245e-05,
|
| 113 |
+
"loss": 0.1149,
|
| 114 |
+
"step": 150
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 2.191780821917808,
|
| 118 |
+
"grad_norm": 0.4330901801586151,
|
| 119 |
+
"learning_rate": 9.242641032675117e-05,
|
| 120 |
+
"loss": 0.1098,
|
| 121 |
+
"step": 160
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 2.328767123287671,
|
| 125 |
+
"grad_norm": 0.4459153413772583,
|
| 126 |
+
"learning_rate": 9.118382907149165e-05,
|
| 127 |
+
"loss": 0.1076,
|
| 128 |
+
"step": 170
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 2.4657534246575343,
|
| 132 |
+
"grad_norm": 0.467021644115448,
|
| 133 |
+
"learning_rate": 8.985662536114613e-05,
|
| 134 |
+
"loss": 0.1097,
|
| 135 |
+
"step": 180
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 2.602739726027397,
|
| 139 |
+
"grad_norm": 0.7806321382522583,
|
| 140 |
+
"learning_rate": 8.84475262669681e-05,
|
| 141 |
+
"loss": 0.0992,
|
| 142 |
+
"step": 190
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 2.73972602739726,
|
| 146 |
+
"grad_norm": 0.7798169851303101,
|
| 147 |
+
"learning_rate": 8.695942713471577e-05,
|
| 148 |
+
"loss": 0.1018,
|
| 149 |
+
"step": 200
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 2.8767123287671232,
|
| 153 |
+
"grad_norm": 0.40669694542884827,
|
| 154 |
+
"learning_rate": 8.539538563544163e-05,
|
| 155 |
+
"loss": 0.1017,
|
| 156 |
+
"step": 210
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 3.0136986301369864,
|
| 160 |
+
"grad_norm": 0.680648922920227,
|
| 161 |
+
"learning_rate": 8.375861548274418e-05,
|
| 162 |
+
"loss": 0.1015,
|
| 163 |
+
"step": 220
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 3.1506849315068495,
|
| 167 |
+
"grad_norm": 0.7035002112388611,
|
| 168 |
+
"learning_rate": 8.205247982939123e-05,
|
| 169 |
+
"loss": 0.0918,
|
| 170 |
+
"step": 230
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 3.287671232876712,
|
| 174 |
+
"grad_norm": 0.7396689653396606,
|
| 175 |
+
"learning_rate": 8.028048435688333e-05,
|
| 176 |
+
"loss": 0.0935,
|
| 177 |
+
"step": 240
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 3.4246575342465753,
|
| 181 |
+
"grad_norm": 0.746228039264679,
|
| 182 |
+
"learning_rate": 7.844627007215613e-05,
|
| 183 |
+
"loss": 0.0903,
|
| 184 |
+
"step": 250
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 3.5616438356164384,
|
| 188 |
+
"grad_norm": 0.4722535014152527,
|
| 189 |
+
"learning_rate": 7.655360582622286e-05,
|
| 190 |
+
"loss": 0.0891,
|
| 191 |
+
"step": 260
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 3.6986301369863015,
|
| 195 |
+
"grad_norm": 0.5862498879432678,
|
| 196 |
+
"learning_rate": 7.460638057012955e-05,
|
| 197 |
+
"loss": 0.0906,
|
| 198 |
+
"step": 270
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 3.8356164383561646,
|
| 202 |
+
"grad_norm": 0.5559934973716736,
|
| 203 |
+
"learning_rate": 7.260859536413429e-05,
|
| 204 |
+
"loss": 0.0839,
|
| 205 |
+
"step": 280
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 3.9726027397260273,
|
| 209 |
+
"grad_norm": 0.5900849103927612,
|
| 210 |
+
"learning_rate": 7.056435515653059e-05,
|
| 211 |
+
"loss": 0.0882,
|
| 212 |
+
"step": 290
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 4.109589041095891,
|
| 216 |
+
"grad_norm": 0.7324544787406921,
|
| 217 |
+
"learning_rate": 6.847786034900649e-05,
|
| 218 |
+
"loss": 0.0827,
|
| 219 |
+
"step": 300
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 4.2465753424657535,
|
| 223 |
+
"grad_norm": 0.5513842701911926,
|
| 224 |
+
"learning_rate": 6.635339816587109e-05,
|
| 225 |
+
"loss": 0.0874,
|
| 226 |
+
"step": 310
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 4.383561643835616,
|
| 230 |
+
"grad_norm": 0.44522860646247864,
|
| 231 |
+
"learning_rate": 6.41953338448822e-05,
|
| 232 |
+
"loss": 0.0824,
|
| 233 |
+
"step": 320
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 4.52054794520548,
|
| 237 |
+
"grad_norm": 0.64205002784729,
|
| 238 |
+
"learning_rate": 6.200810166777597e-05,
|
| 239 |
+
"loss": 0.0837,
|
| 240 |
+
"step": 330
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 4.657534246575342,
|
| 244 |
+
"grad_norm": 0.35937950015068054,
|
| 245 |
+
"learning_rate": 5.979619584892834e-05,
|
| 246 |
+
"loss": 0.0791,
|
| 247 |
+
"step": 340
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 4.794520547945205,
|
| 251 |
+
"grad_norm": 0.3836668133735657,
|
| 252 |
+
"learning_rate": 5.756416130087001e-05,
|
| 253 |
+
"loss": 0.0787,
|
| 254 |
+
"step": 350
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 4.931506849315069,
|
| 258 |
+
"grad_norm": 0.5425627827644348,
|
| 259 |
+
"learning_rate": 5.53165842956294e-05,
|
| 260 |
+
"loss": 0.077,
|
| 261 |
+
"step": 360
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 5.068493150684931,
|
| 265 |
+
"grad_norm": 0.5653792023658752,
|
| 266 |
+
"learning_rate": 5.305808304109214e-05,
|
| 267 |
+
"loss": 0.0809,
|
| 268 |
+
"step": 370
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 5.205479452054795,
|
| 272 |
+
"grad_norm": 0.6761691570281982,
|
| 273 |
+
"learning_rate": 5.0793298191740404e-05,
|
| 274 |
+
"loss": 0.0815,
|
| 275 |
+
"step": 380
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 5.342465753424658,
|
| 279 |
+
"grad_norm": 0.39536017179489136,
|
| 280 |
+
"learning_rate": 4.8526883313270106e-05,
|
| 281 |
+
"loss": 0.0853,
|
| 282 |
+
"step": 390
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 5.47945205479452,
|
| 286 |
+
"grad_norm": 0.47267112135887146,
|
| 287 |
+
"learning_rate": 4.626349532067879e-05,
|
| 288 |
+
"loss": 0.0756,
|
| 289 |
+
"step": 400
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 5.616438356164384,
|
| 293 |
+
"grad_norm": 0.42643123865127563,
|
| 294 |
+
"learning_rate": 4.400778490947157e-05,
|
| 295 |
+
"loss": 0.0743,
|
| 296 |
+
"step": 410
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 5.7534246575342465,
|
| 300 |
+
"grad_norm": 0.44257625937461853,
|
| 301 |
+
"learning_rate": 4.176438699964646e-05,
|
| 302 |
+
"loss": 0.0755,
|
| 303 |
+
"step": 420
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 5.890410958904109,
|
| 307 |
+
"grad_norm": 0.41078025102615356,
|
| 308 |
+
"learning_rate": 3.953791121209458e-05,
|
| 309 |
+
"loss": 0.0747,
|
| 310 |
+
"step": 430
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 6.027397260273973,
|
| 314 |
+
"grad_norm": 0.4885757863521576,
|
| 315 |
+
"learning_rate": 3.733293239698359e-05,
|
| 316 |
+
"loss": 0.0732,
|
| 317 |
+
"step": 440
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 6.164383561643835,
|
| 321 |
+
"grad_norm": 0.6123595237731934,
|
| 322 |
+
"learning_rate": 3.515398123358627e-05,
|
| 323 |
+
"loss": 0.0761,
|
| 324 |
+
"step": 450
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 6.301369863013699,
|
| 328 |
+
"grad_norm": 0.3499448001384735,
|
| 329 |
+
"learning_rate": 3.3005534920869176e-05,
|
| 330 |
+
"loss": 0.0767,
|
| 331 |
+
"step": 460
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 6.438356164383562,
|
| 335 |
+
"grad_norm": 0.43977606296539307,
|
| 336 |
+
"learning_rate": 3.0892007977970086e-05,
|
| 337 |
+
"loss": 0.0731,
|
| 338 |
+
"step": 470
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 6.575342465753424,
|
| 342 |
+
"grad_norm": 0.29874327778816223,
|
| 343 |
+
"learning_rate": 2.8817743173466636e-05,
|
| 344 |
+
"loss": 0.0735,
|
| 345 |
+
"step": 480
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 6.712328767123288,
|
| 349 |
+
"grad_norm": 0.3531714975833893,
|
| 350 |
+
"learning_rate": 2.678700260207449e-05,
|
| 351 |
+
"loss": 0.0721,
|
| 352 |
+
"step": 490
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 6.8493150684931505,
|
| 356 |
+
"grad_norm": 0.3463064730167389,
|
| 357 |
+
"learning_rate": 2.4803958927109965e-05,
|
| 358 |
+
"loss": 0.073,
|
| 359 |
+
"step": 500
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 6.986301369863014,
|
| 363 |
+
"grad_norm": 0.3994370698928833,
|
| 364 |
+
"learning_rate": 2.2872686806712035e-05,
|
| 365 |
+
"loss": 0.0713,
|
| 366 |
+
"step": 510
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 7.123287671232877,
|
| 370 |
+
"grad_norm": 0.3885277211666107,
|
| 371 |
+
"learning_rate": 2.09971545214401e-05,
|
| 372 |
+
"loss": 0.0768,
|
| 373 |
+
"step": 520
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 7.260273972602739,
|
| 377 |
+
"grad_norm": 0.5557729005813599,
|
| 378 |
+
"learning_rate": 1.918121582045132e-05,
|
| 379 |
+
"loss": 0.0681,
|
| 380 |
+
"step": 530
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 7.397260273972603,
|
| 384 |
+
"grad_norm": 0.34055230021476746,
|
| 385 |
+
"learning_rate": 1.7428602003011134e-05,
|
| 386 |
+
"loss": 0.0697,
|
| 387 |
+
"step": 540
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 7.534246575342466,
|
| 391 |
+
"grad_norm": 0.3750840723514557,
|
| 392 |
+
"learning_rate": 1.5742914251607793e-05,
|
| 393 |
+
"loss": 0.0679,
|
| 394 |
+
"step": 550
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 7.671232876712329,
|
| 398 |
+
"grad_norm": 0.48403847217559814,
|
| 399 |
+
"learning_rate": 1.412761623242404e-05,
|
| 400 |
+
"loss": 0.0669,
|
| 401 |
+
"step": 560
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 7.808219178082192,
|
| 405 |
+
"grad_norm": 0.37212374806404114,
|
| 406 |
+
"learning_rate": 1.2586026978370629e-05,
|
| 407 |
+
"loss": 0.0689,
|
| 408 |
+
"step": 570
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 7.945205479452055,
|
| 412 |
+
"grad_norm": 0.4119841158390045,
|
| 413 |
+
"learning_rate": 1.1121314069304811e-05,
|
| 414 |
+
"loss": 0.0699,
|
| 415 |
+
"step": 580
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 8.082191780821917,
|
| 419 |
+
"grad_norm": 0.529219388961792,
|
| 420 |
+
"learning_rate": 9.73648712344707e-06,
|
| 421 |
+
"loss": 0.0694,
|
| 422 |
+
"step": 590
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 8.219178082191782,
|
| 426 |
+
"grad_norm": 0.3745039701461792,
|
| 427 |
+
"learning_rate": 8.434391613369258e-06,
|
| 428 |
+
"loss": 0.0666,
|
| 429 |
+
"step": 600
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 8.356164383561644,
|
| 433 |
+
"grad_norm": 0.4455980956554413,
|
| 434 |
+
"learning_rate": 7.217703019261135e-06,
|
| 435 |
+
"loss": 0.0706,
|
| 436 |
+
"step": 610
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 8.493150684931507,
|
| 440 |
+
"grad_norm": 0.53988116979599,
|
| 441 |
+
"learning_rate": 6.088921331488568e-06,
|
| 442 |
+
"loss": 0.0707,
|
| 443 |
+
"step": 620
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 8.63013698630137,
|
| 447 |
+
"grad_norm": 0.33156082034111023,
|
| 448 |
+
"learning_rate": 5.050365913739441e-06,
|
| 449 |
+
"loss": 0.0645,
|
| 450 |
+
"step": 630
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 8.767123287671232,
|
| 454 |
+
"grad_norm": 0.3420356810092926,
|
| 455 |
+
"learning_rate": 4.1041707373120356e-06,
|
| 456 |
+
"loss": 0.0654,
|
| 457 |
+
"step": 640
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 8.904109589041095,
|
| 461 |
+
"grad_norm": 0.39363735914230347,
|
| 462 |
+
"learning_rate": 3.252279996338492e-06,
|
| 463 |
+
"loss": 0.0699,
|
| 464 |
+
"step": 650
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 9.04109589041096,
|
| 468 |
+
"grad_norm": 0.6195297241210938,
|
| 469 |
+
"learning_rate": 2.496444112952734e-06,
|
| 470 |
+
"loss": 0.0689,
|
| 471 |
+
"step": 660
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 9.178082191780822,
|
| 475 |
+
"grad_norm": 0.4905112087726593,
|
| 476 |
+
"learning_rate": 1.8382161406113208e-06,
|
| 477 |
+
"loss": 0.0654,
|
| 478 |
+
"step": 670
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 9.315068493150685,
|
| 482 |
+
"grad_norm": 0.43245476484298706,
|
| 483 |
+
"learning_rate": 1.2789485729575612e-06,
|
| 484 |
+
"loss": 0.0708,
|
| 485 |
+
"step": 680
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 9.452054794520548,
|
| 489 |
+
"grad_norm": 0.4771445393562317,
|
| 490 |
+
"learning_rate": 8.197905647857384e-07,
|
| 491 |
+
"loss": 0.0707,
|
| 492 |
+
"step": 690
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 9.58904109589041,
|
| 496 |
+
"grad_norm": 0.23996645212173462,
|
| 497 |
+
"learning_rate": 4.6168557081582854e-07,
|
| 498 |
+
"loss": 0.0634,
|
| 499 |
+
"step": 700
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 9.726027397260275,
|
| 503 |
+
"grad_norm": 0.5234819054603577,
|
| 504 |
+
"learning_rate": 2.0536940713028474e-07,
|
| 505 |
+
"loss": 0.0659,
|
| 506 |
+
"step": 710
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 9.863013698630137,
|
| 510 |
+
"grad_norm": 0.3188411295413971,
|
| 511 |
+
"learning_rate": 5.1368739256296706e-08,
|
| 512 |
+
"loss": 0.066,
|
| 513 |
+
"step": 720
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 10.0,
|
| 517 |
+
"grad_norm": 0.543819785118103,
|
| 518 |
+
"learning_rate": 0.0,
|
| 519 |
+
"loss": 0.0664,
|
| 520 |
+
"step": 730
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 10.0,
|
| 524 |
+
"step": 730,
|
| 525 |
+
"total_flos": 1.0568936642296032e+17,
|
| 526 |
+
"train_loss": 0.11973448420224124,
|
| 527 |
+
"train_runtime": 466.7787,
|
| 528 |
+
"train_samples_per_second": 198.659,
|
| 529 |
+
"train_steps_per_second": 1.564
|
| 530 |
+
}
|
| 531 |
+
],
|
| 532 |
+
"logging_steps": 10,
|
| 533 |
+
"max_steps": 730,
|
| 534 |
+
"num_input_tokens_seen": 0,
|
| 535 |
+
"num_train_epochs": 10,
|
| 536 |
+
"save_steps": 20000,
|
| 537 |
+
"stateful_callbacks": {
|
| 538 |
+
"TrainerControl": {
|
| 539 |
+
"args": {
|
| 540 |
+
"should_epoch_stop": false,
|
| 541 |
+
"should_evaluate": false,
|
| 542 |
+
"should_log": false,
|
| 543 |
+
"should_save": true,
|
| 544 |
+
"should_training_stop": true
|
| 545 |
+
},
|
| 546 |
+
"attributes": {}
|
| 547 |
+
}
|
| 548 |
+
},
|
| 549 |
+
"total_flos": 1.0568936642296032e+17,
|
| 550 |
+
"train_batch_size": 128,
|
| 551 |
+
"trial_name": null,
|
| 552 |
+
"trial_params": null
|
| 553 |
+
}
|