Commit
•
d66c37a
1
Parent(s):
67fa0a8
End of training
Browse files- all_results.json +14 -0
- eval_results.json +9 -0
- runs/Jun14_15-53-02_labgpu/events.out.tfevents.1655236207.labgpu.32510.2 +3 -0
- train_results.json +8 -0
- trainer_state.json +415 -0
all_results.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 10.0,
|
| 3 |
+
"eval_accuracy": 0.9068277503203368,
|
| 4 |
+
"eval_loss": 0.3730911314487457,
|
| 5 |
+
"eval_runtime": 29.6788,
|
| 6 |
+
"eval_samples": 5463,
|
| 7 |
+
"eval_samples_per_second": 184.071,
|
| 8 |
+
"eval_steps_per_second": 23.013,
|
| 9 |
+
"train_loss": 3521.3033845717396,
|
| 10 |
+
"train_runtime": 13419.3371,
|
| 11 |
+
"train_samples": 104743,
|
| 12 |
+
"train_samples_per_second": 78.054,
|
| 13 |
+
"train_steps_per_second": 2.44
|
| 14 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 10.0,
|
| 3 |
+
"eval_accuracy": 0.9068277503203368,
|
| 4 |
+
"eval_loss": 0.3730911314487457,
|
| 5 |
+
"eval_runtime": 29.6788,
|
| 6 |
+
"eval_samples": 5463,
|
| 7 |
+
"eval_samples_per_second": 184.071,
|
| 8 |
+
"eval_steps_per_second": 23.013
|
| 9 |
+
}
|
runs/Jun14_15-53-02_labgpu/events.out.tfevents.1655236207.labgpu.32510.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f0d33d55a47a5f0735b7809bca8630fbf0b989e3bedc196fc957b18d91422ca7
|
| 3 |
+
size 369
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 10.0,
|
| 3 |
+
"train_loss": 3521.3033845717396,
|
| 4 |
+
"train_runtime": 13419.3371,
|
| 5 |
+
"train_samples": 104743,
|
| 6 |
+
"train_samples_per_second": 78.054,
|
| 7 |
+
"train_steps_per_second": 2.44
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,415 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 10.0,
|
| 5 |
+
"global_step": 32740,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.15,
|
| 12 |
+
"learning_rate": 9.847281612706171e-06,
|
| 13 |
+
"loss": 230560.032,
|
| 14 |
+
"step": 500
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.31,
|
| 18 |
+
"learning_rate": 9.69456322541234e-06,
|
| 19 |
+
"loss": 0.4568,
|
| 20 |
+
"step": 1000
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.46,
|
| 24 |
+
"learning_rate": 9.541844838118511e-06,
|
| 25 |
+
"loss": 0.4604,
|
| 26 |
+
"step": 1500
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.61,
|
| 30 |
+
"learning_rate": 9.389126450824681e-06,
|
| 31 |
+
"loss": 0.4138,
|
| 32 |
+
"step": 2000
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.76,
|
| 36 |
+
"learning_rate": 9.23640806353085e-06,
|
| 37 |
+
"loss": 0.3819,
|
| 38 |
+
"step": 2500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.92,
|
| 42 |
+
"learning_rate": 9.08368967623702e-06,
|
| 43 |
+
"loss": 0.3615,
|
| 44 |
+
"step": 3000
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 1.07,
|
| 48 |
+
"learning_rate": 8.93097128894319e-06,
|
| 49 |
+
"loss": 0.348,
|
| 50 |
+
"step": 3500
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 1.22,
|
| 54 |
+
"learning_rate": 8.77825290164936e-06,
|
| 55 |
+
"loss": 0.4017,
|
| 56 |
+
"step": 4000
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 1.37,
|
| 60 |
+
"learning_rate": 8.62553451435553e-06,
|
| 61 |
+
"loss": 0.3136,
|
| 62 |
+
"step": 4500
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 1.53,
|
| 66 |
+
"learning_rate": 8.4728161270617e-06,
|
| 67 |
+
"loss": 0.3419,
|
| 68 |
+
"step": 5000
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 1.68,
|
| 72 |
+
"learning_rate": 8.320097739767868e-06,
|
| 73 |
+
"loss": 0.3969,
|
| 74 |
+
"step": 5500
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 1.83,
|
| 78 |
+
"learning_rate": 8.167379352474038e-06,
|
| 79 |
+
"loss": 0.304,
|
| 80 |
+
"step": 6000
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 1.99,
|
| 84 |
+
"learning_rate": 8.014660965180208e-06,
|
| 85 |
+
"loss": 0.2897,
|
| 86 |
+
"step": 6500
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 2.14,
|
| 90 |
+
"learning_rate": 7.861942577886378e-06,
|
| 91 |
+
"loss": 0.2875,
|
| 92 |
+
"step": 7000
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 2.29,
|
| 96 |
+
"learning_rate": 7.709224190592548e-06,
|
| 97 |
+
"loss": 0.2658,
|
| 98 |
+
"step": 7500
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 2.44,
|
| 102 |
+
"learning_rate": 7.556505803298718e-06,
|
| 103 |
+
"loss": 0.2658,
|
| 104 |
+
"step": 8000
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 2.6,
|
| 108 |
+
"learning_rate": 7.403787416004888e-06,
|
| 109 |
+
"loss": 0.2861,
|
| 110 |
+
"step": 8500
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 2.75,
|
| 114 |
+
"learning_rate": 7.251069028711058e-06,
|
| 115 |
+
"loss": 0.2862,
|
| 116 |
+
"step": 9000
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 2.9,
|
| 120 |
+
"learning_rate": 7.098350641417228e-06,
|
| 121 |
+
"loss": 0.2594,
|
| 122 |
+
"step": 9500
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 3.05,
|
| 126 |
+
"learning_rate": 6.945632254123397e-06,
|
| 127 |
+
"loss": 0.2509,
|
| 128 |
+
"step": 10000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 3.21,
|
| 132 |
+
"learning_rate": 6.792913866829567e-06,
|
| 133 |
+
"loss": 0.2535,
|
| 134 |
+
"step": 10500
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 3.36,
|
| 138 |
+
"learning_rate": 6.640195479535737e-06,
|
| 139 |
+
"loss": 0.2358,
|
| 140 |
+
"step": 11000
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 3.51,
|
| 144 |
+
"learning_rate": 6.487477092241906e-06,
|
| 145 |
+
"loss": 0.2357,
|
| 146 |
+
"step": 11500
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 3.67,
|
| 150 |
+
"learning_rate": 6.334758704948076e-06,
|
| 151 |
+
"loss": 0.2331,
|
| 152 |
+
"step": 12000
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 3.82,
|
| 156 |
+
"learning_rate": 6.1820403176542464e-06,
|
| 157 |
+
"loss": 0.2806,
|
| 158 |
+
"step": 12500
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 3.97,
|
| 162 |
+
"learning_rate": 6.029321930360416e-06,
|
| 163 |
+
"loss": 0.2245,
|
| 164 |
+
"step": 13000
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 4.12,
|
| 168 |
+
"learning_rate": 5.876603543066586e-06,
|
| 169 |
+
"loss": 0.2074,
|
| 170 |
+
"step": 13500
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 4.28,
|
| 174 |
+
"learning_rate": 5.723885155772756e-06,
|
| 175 |
+
"loss": 0.2115,
|
| 176 |
+
"step": 14000
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 4.43,
|
| 180 |
+
"learning_rate": 5.571166768478925e-06,
|
| 181 |
+
"loss": 0.2124,
|
| 182 |
+
"step": 14500
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 4.58,
|
| 186 |
+
"learning_rate": 5.418448381185095e-06,
|
| 187 |
+
"loss": 0.2149,
|
| 188 |
+
"step": 15000
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 4.73,
|
| 192 |
+
"learning_rate": 5.265729993891265e-06,
|
| 193 |
+
"loss": 0.2133,
|
| 194 |
+
"step": 15500
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 4.89,
|
| 198 |
+
"learning_rate": 5.113011606597434e-06,
|
| 199 |
+
"loss": 0.2119,
|
| 200 |
+
"step": 16000
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 5.04,
|
| 204 |
+
"learning_rate": 4.960293219303604e-06,
|
| 205 |
+
"loss": 0.2128,
|
| 206 |
+
"step": 16500
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 5.19,
|
| 210 |
+
"learning_rate": 4.8075748320097744e-06,
|
| 211 |
+
"loss": 0.1881,
|
| 212 |
+
"step": 17000
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 5.35,
|
| 216 |
+
"learning_rate": 4.6548564447159445e-06,
|
| 217 |
+
"loss": 0.1974,
|
| 218 |
+
"step": 17500
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 5.5,
|
| 222 |
+
"learning_rate": 4.502138057422114e-06,
|
| 223 |
+
"loss": 0.1985,
|
| 224 |
+
"step": 18000
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 5.65,
|
| 228 |
+
"learning_rate": 4.349419670128284e-06,
|
| 229 |
+
"loss": 0.1979,
|
| 230 |
+
"step": 18500
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 5.8,
|
| 234 |
+
"learning_rate": 4.196701282834454e-06,
|
| 235 |
+
"loss": 0.1986,
|
| 236 |
+
"step": 19000
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 5.96,
|
| 240 |
+
"learning_rate": 4.043982895540624e-06,
|
| 241 |
+
"loss": 0.1906,
|
| 242 |
+
"step": 19500
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 6.11,
|
| 246 |
+
"learning_rate": 3.891264508246793e-06,
|
| 247 |
+
"loss": 0.1873,
|
| 248 |
+
"step": 20000
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 6.26,
|
| 252 |
+
"learning_rate": 3.7385461209529628e-06,
|
| 253 |
+
"loss": 0.1832,
|
| 254 |
+
"step": 20500
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 6.41,
|
| 258 |
+
"learning_rate": 3.585827733659133e-06,
|
| 259 |
+
"loss": 0.1786,
|
| 260 |
+
"step": 21000
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 6.57,
|
| 264 |
+
"learning_rate": 3.4331093463653025e-06,
|
| 265 |
+
"loss": 0.1811,
|
| 266 |
+
"step": 21500
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 6.72,
|
| 270 |
+
"learning_rate": 3.280390959071472e-06,
|
| 271 |
+
"loss": 0.1856,
|
| 272 |
+
"step": 22000
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 6.87,
|
| 276 |
+
"learning_rate": 3.1276725717776426e-06,
|
| 277 |
+
"loss": 0.1767,
|
| 278 |
+
"step": 22500
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 7.03,
|
| 282 |
+
"learning_rate": 2.974954184483812e-06,
|
| 283 |
+
"loss": 0.1744,
|
| 284 |
+
"step": 23000
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 7.18,
|
| 288 |
+
"learning_rate": 2.8222357971899814e-06,
|
| 289 |
+
"loss": 0.17,
|
| 290 |
+
"step": 23500
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 7.33,
|
| 294 |
+
"learning_rate": 2.669517409896152e-06,
|
| 295 |
+
"loss": 0.178,
|
| 296 |
+
"step": 24000
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 7.48,
|
| 300 |
+
"learning_rate": 2.5167990226023216e-06,
|
| 301 |
+
"loss": 0.1742,
|
| 302 |
+
"step": 24500
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 7.64,
|
| 306 |
+
"learning_rate": 2.364080635308491e-06,
|
| 307 |
+
"loss": 0.1673,
|
| 308 |
+
"step": 25000
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 7.79,
|
| 312 |
+
"learning_rate": 2.2113622480146613e-06,
|
| 313 |
+
"loss": 0.1634,
|
| 314 |
+
"step": 25500
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 7.94,
|
| 318 |
+
"learning_rate": 2.058643860720831e-06,
|
| 319 |
+
"loss": 0.1735,
|
| 320 |
+
"step": 26000
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 8.09,
|
| 324 |
+
"learning_rate": 1.9059254734270008e-06,
|
| 325 |
+
"loss": 0.1691,
|
| 326 |
+
"step": 26500
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 8.25,
|
| 330 |
+
"learning_rate": 1.7532070861331706e-06,
|
| 331 |
+
"loss": 0.1697,
|
| 332 |
+
"step": 27000
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 8.4,
|
| 336 |
+
"learning_rate": 1.6004886988393402e-06,
|
| 337 |
+
"loss": 0.1585,
|
| 338 |
+
"step": 27500
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 8.55,
|
| 342 |
+
"learning_rate": 1.4477703115455103e-06,
|
| 343 |
+
"loss": 0.1591,
|
| 344 |
+
"step": 28000
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 8.7,
|
| 348 |
+
"learning_rate": 1.2950519242516802e-06,
|
| 349 |
+
"loss": 0.1622,
|
| 350 |
+
"step": 28500
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 8.86,
|
| 354 |
+
"learning_rate": 1.1423335369578498e-06,
|
| 355 |
+
"loss": 0.1558,
|
| 356 |
+
"step": 29000
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 9.01,
|
| 360 |
+
"learning_rate": 9.896151496640196e-07,
|
| 361 |
+
"loss": 0.1598,
|
| 362 |
+
"step": 29500
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 9.16,
|
| 366 |
+
"learning_rate": 8.368967623701894e-07,
|
| 367 |
+
"loss": 0.1515,
|
| 368 |
+
"step": 30000
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 9.32,
|
| 372 |
+
"learning_rate": 6.841783750763592e-07,
|
| 373 |
+
"loss": 0.1556,
|
| 374 |
+
"step": 30500
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 9.47,
|
| 378 |
+
"learning_rate": 5.314599877825291e-07,
|
| 379 |
+
"loss": 0.1526,
|
| 380 |
+
"step": 31000
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 9.62,
|
| 384 |
+
"learning_rate": 3.787416004886989e-07,
|
| 385 |
+
"loss": 0.1586,
|
| 386 |
+
"step": 31500
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 9.77,
|
| 390 |
+
"learning_rate": 2.2602321319486868e-07,
|
| 391 |
+
"loss": 0.1489,
|
| 392 |
+
"step": 32000
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 9.93,
|
| 396 |
+
"learning_rate": 7.330482590103849e-08,
|
| 397 |
+
"loss": 0.1567,
|
| 398 |
+
"step": 32500
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 10.0,
|
| 402 |
+
"step": 32740,
|
| 403 |
+
"total_flos": 1.64206965272832e+16,
|
| 404 |
+
"train_loss": 3521.3033845717396,
|
| 405 |
+
"train_runtime": 13419.3371,
|
| 406 |
+
"train_samples_per_second": 78.054,
|
| 407 |
+
"train_steps_per_second": 2.44
|
| 408 |
+
}
|
| 409 |
+
],
|
| 410 |
+
"max_steps": 32740,
|
| 411 |
+
"num_train_epochs": 10,
|
| 412 |
+
"total_flos": 1.64206965272832e+16,
|
| 413 |
+
"trial_name": null,
|
| 414 |
+
"trial_params": null
|
| 415 |
+
}
|