Upload bayes_trial_metrics.txt with huggingface_hub
Browse files- bayes_trial_metrics.txt +562 -0
bayes_trial_metrics.txt
ADDED
|
@@ -0,0 +1,562 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[
|
| 2 |
+
{
|
| 3 |
+
"trial_number": 0,
|
| 4 |
+
"params": {
|
| 5 |
+
"num_heads": 6,
|
| 6 |
+
"num_layers": 1
|
| 7 |
+
},
|
| 8 |
+
"epoch_metrics": [
|
| 9 |
+
{
|
| 10 |
+
"epoch": 1,
|
| 11 |
+
"train_loss": 1.1832761391537991,
|
| 12 |
+
"val_loss": 0.7498329176104127,
|
| 13 |
+
"macro_f1": 0.7103340867858469
|
| 14 |
+
},
|
| 15 |
+
{
|
| 16 |
+
"epoch": 2,
|
| 17 |
+
"train_loss": 0.6320729479836771,
|
| 18 |
+
"val_loss": 0.586862810580014,
|
| 19 |
+
"macro_f1": 0.7913692983317114
|
| 20 |
+
},
|
| 21 |
+
{
|
| 22 |
+
"epoch": 3,
|
| 23 |
+
"train_loss": 0.5160054902002932,
|
| 24 |
+
"val_loss": 0.5864161346933524,
|
| 25 |
+
"macro_f1": 0.7848025166573419
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"epoch": 4,
|
| 29 |
+
"train_loss": 0.411905202302627,
|
| 30 |
+
"val_loss": 0.511089034415231,
|
| 31 |
+
"macro_f1": 0.8072666472893073
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 5,
|
| 35 |
+
"train_loss": 0.3297956578902046,
|
| 36 |
+
"val_loss": 0.464968596516278,
|
| 37 |
+
"macro_f1": 0.834067964116802
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 6,
|
| 41 |
+
"train_loss": 0.2837876489732768,
|
| 42 |
+
"val_loss": 0.38831818485494907,
|
| 43 |
+
"macro_f1": 0.8705889067805529
|
| 44 |
+
},
|
| 45 |
+
{
|
| 46 |
+
"epoch": 7,
|
| 47 |
+
"train_loss": 0.22080586585141582,
|
| 48 |
+
"val_loss": 0.351857351157539,
|
| 49 |
+
"macro_f1": 0.8878972081948868
|
| 50 |
+
},
|
| 51 |
+
{
|
| 52 |
+
"epoch": 8,
|
| 53 |
+
"train_loss": 0.18440273702695836,
|
| 54 |
+
"val_loss": 0.36548399620602284,
|
| 55 |
+
"macro_f1": 0.8804424351677086
|
| 56 |
+
},
|
| 57 |
+
{
|
| 58 |
+
"epoch": 9,
|
| 59 |
+
"train_loss": 0.14913206719753186,
|
| 60 |
+
"val_loss": 0.38098949995442416,
|
| 61 |
+
"macro_f1": 0.8723550071229064
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 10,
|
| 65 |
+
"train_loss": 0.12198060535528363,
|
| 66 |
+
"val_loss": 0.4393947260646984,
|
| 67 |
+
"macro_f1": 0.8655638957350722
|
| 68 |
+
}
|
| 69 |
+
],
|
| 70 |
+
"final_macro_f1": 0.8655638957350722
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"trial_number": 1,
|
| 74 |
+
"params": {
|
| 75 |
+
"num_heads": 4,
|
| 76 |
+
"num_layers": 2
|
| 77 |
+
},
|
| 78 |
+
"epoch_metrics": [
|
| 79 |
+
{
|
| 80 |
+
"epoch": 1,
|
| 81 |
+
"train_loss": 1.1230820458561512,
|
| 82 |
+
"val_loss": 0.7704554694920338,
|
| 83 |
+
"macro_f1": 0.7036616295856207
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 2,
|
| 87 |
+
"train_loss": 0.6215897324493869,
|
| 88 |
+
"val_loss": 0.6329705550752837,
|
| 89 |
+
"macro_f1": 0.747321270691553
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 3,
|
| 93 |
+
"train_loss": 0.5201769666298397,
|
| 94 |
+
"val_loss": 0.5488664073309875,
|
| 95 |
+
"macro_f1": 0.8103063427800269
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 4,
|
| 99 |
+
"train_loss": 0.3938881587673498,
|
| 100 |
+
"val_loss": 0.506728891118113,
|
| 101 |
+
"macro_f1": 0.8111035470961853
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 5,
|
| 105 |
+
"train_loss": 0.3405428771871233,
|
| 106 |
+
"val_loss": 0.4811670455824712,
|
| 107 |
+
"macro_f1": 0.8425270143911638
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 6,
|
| 111 |
+
"train_loss": 0.29019304942776036,
|
| 112 |
+
"val_loss": 0.41828343503434084,
|
| 113 |
+
"macro_f1": 0.8464476111160173
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 7,
|
| 117 |
+
"train_loss": 0.23187170347196548,
|
| 118 |
+
"val_loss": 0.42985851424080984,
|
| 119 |
+
"macro_f1": 0.8487845567154556
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 8,
|
| 123 |
+
"train_loss": 0.19625346978113478,
|
| 124 |
+
"val_loss": 0.42691103950730097,
|
| 125 |
+
"macro_f1": 0.8671223625188696
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 9,
|
| 129 |
+
"train_loss": 0.14775698915877267,
|
| 130 |
+
"val_loss": 0.44993914651175215,
|
| 131 |
+
"macro_f1": 0.8621576111982389
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 10,
|
| 135 |
+
"train_loss": 0.1359217370229827,
|
| 136 |
+
"val_loss": 0.40257360588924346,
|
| 137 |
+
"macro_f1": 0.8679347055481997
|
| 138 |
+
}
|
| 139 |
+
],
|
| 140 |
+
"final_macro_f1": 0.8679347055481997
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"trial_number": 4,
|
| 144 |
+
"params": {
|
| 145 |
+
"num_heads": 6,
|
| 146 |
+
"num_layers": 2
|
| 147 |
+
},
|
| 148 |
+
"epoch_metrics": [
|
| 149 |
+
{
|
| 150 |
+
"epoch": 1,
|
| 151 |
+
"train_loss": 1.3173663898402166,
|
| 152 |
+
"val_loss": 0.759101252397293,
|
| 153 |
+
"macro_f1": 0.6990005637236743
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 2,
|
| 157 |
+
"train_loss": 0.6467822624046617,
|
| 158 |
+
"val_loss": 0.5983039810152476,
|
| 159 |
+
"macro_f1": 0.78449167857198
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 3,
|
| 163 |
+
"train_loss": 0.5089347752778068,
|
| 164 |
+
"val_loss": 0.5390065236044634,
|
| 165 |
+
"macro_f1": 0.8095188828227796
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 4,
|
| 169 |
+
"train_loss": 0.40332894629894145,
|
| 170 |
+
"val_loss": 0.4772564168340467,
|
| 171 |
+
"macro_f1": 0.8295270726746916
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 5,
|
| 175 |
+
"train_loss": 0.33736765722752204,
|
| 176 |
+
"val_loss": 0.46593107330828465,
|
| 177 |
+
"macro_f1": 0.8366828266689496
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 6,
|
| 181 |
+
"train_loss": 0.26871140042156827,
|
| 182 |
+
"val_loss": 0.46045689791294153,
|
| 183 |
+
"macro_f1": 0.8419257227035141
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 7,
|
| 187 |
+
"train_loss": 0.21393143714539956,
|
| 188 |
+
"val_loss": 0.39201908384270856,
|
| 189 |
+
"macro_f1": 0.8753177666423365
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 8,
|
| 193 |
+
"train_loss": 0.1685226441265696,
|
| 194 |
+
"val_loss": 0.4256001700615061,
|
| 195 |
+
"macro_f1": 0.8579319045518754
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 9,
|
| 199 |
+
"train_loss": 0.11896324822100388,
|
| 200 |
+
"val_loss": 0.43303222099108063,
|
| 201 |
+
"macro_f1": 0.8677495291666183
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 10,
|
| 205 |
+
"train_loss": 0.1051060429541107,
|
| 206 |
+
"val_loss": 0.49004084796741093,
|
| 207 |
+
"macro_f1": 0.8619553747771226
|
| 208 |
+
}
|
| 209 |
+
],
|
| 210 |
+
"final_macro_f1": 0.8619553747771226
|
| 211 |
+
},
|
| 212 |
+
{
|
| 213 |
+
"trial_number": 5,
|
| 214 |
+
"params": {
|
| 215 |
+
"num_heads": 6,
|
| 216 |
+
"num_layers": 2
|
| 217 |
+
},
|
| 218 |
+
"epoch_metrics": [
|
| 219 |
+
{
|
| 220 |
+
"epoch": 1,
|
| 221 |
+
"train_loss": 1.1954575238701153,
|
| 222 |
+
"val_loss": 0.7813598942286862,
|
| 223 |
+
"macro_f1": 0.7221217333984522
|
| 224 |
+
},
|
| 225 |
+
{
|
| 226 |
+
"epoch": 2,
|
| 227 |
+
"train_loss": 0.6292221567486718,
|
| 228 |
+
"val_loss": 0.6820206618661364,
|
| 229 |
+
"macro_f1": 0.730189742107776
|
| 230 |
+
},
|
| 231 |
+
{
|
| 232 |
+
"epoch": 3,
|
| 233 |
+
"train_loss": 0.5232538815344306,
|
| 234 |
+
"val_loss": 0.5693920787625713,
|
| 235 |
+
"macro_f1": 0.7875359818027032
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 4,
|
| 239 |
+
"train_loss": 0.4423905862759427,
|
| 240 |
+
"val_loss": 0.5050926569647389,
|
| 241 |
+
"macro_f1": 0.8160427418514019
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 5,
|
| 245 |
+
"train_loss": 0.35481051066212826,
|
| 246 |
+
"val_loss": 0.45971177048577466,
|
| 247 |
+
"macro_f1": 0.8303637056003775
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 6,
|
| 251 |
+
"train_loss": 0.29134497639944756,
|
| 252 |
+
"val_loss": 0.4631793054160226,
|
| 253 |
+
"macro_f1": 0.8430694304621142
|
| 254 |
+
},
|
| 255 |
+
{
|
| 256 |
+
"epoch": 7,
|
| 257 |
+
"train_loss": 0.2574463309639773,
|
| 258 |
+
"val_loss": 0.4040321405481529,
|
| 259 |
+
"macro_f1": 0.8648266468143123
|
| 260 |
+
},
|
| 261 |
+
{
|
| 262 |
+
"epoch": 8,
|
| 263 |
+
"train_loss": 0.17930574953997033,
|
| 264 |
+
"val_loss": 0.41403597299688555,
|
| 265 |
+
"macro_f1": 0.8802919260037383
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 9,
|
| 269 |
+
"train_loss": 0.15790989192182017,
|
| 270 |
+
"val_loss": 0.44370942524281043,
|
| 271 |
+
"macro_f1": 0.8783024228283974
|
| 272 |
+
},
|
| 273 |
+
{
|
| 274 |
+
"epoch": 10,
|
| 275 |
+
"train_loss": 0.13971920217285175,
|
| 276 |
+
"val_loss": 0.4138593869329673,
|
| 277 |
+
"macro_f1": 0.8782744869992875
|
| 278 |
+
}
|
| 279 |
+
],
|
| 280 |
+
"final_macro_f1": 0.8782744869992875
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"trial_number": 7,
|
| 284 |
+
"params": {
|
| 285 |
+
"num_heads": 4,
|
| 286 |
+
"num_layers": 2
|
| 287 |
+
},
|
| 288 |
+
"epoch_metrics": [
|
| 289 |
+
{
|
| 290 |
+
"epoch": 1,
|
| 291 |
+
"train_loss": 1.1595548766902872,
|
| 292 |
+
"val_loss": 0.7383359403445803,
|
| 293 |
+
"macro_f1": 0.712648753378513
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 2,
|
| 297 |
+
"train_loss": 0.6291670470364144,
|
| 298 |
+
"val_loss": 0.6041549987393647,
|
| 299 |
+
"macro_f1": 0.7628297499692137
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 3,
|
| 303 |
+
"train_loss": 0.5113105823535072,
|
| 304 |
+
"val_loss": 0.5913394728464446,
|
| 305 |
+
"macro_f1": 0.762223321208065
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 4,
|
| 309 |
+
"train_loss": 0.429566226302475,
|
| 310 |
+
"val_loss": 0.5100458779285106,
|
| 311 |
+
"macro_f1": 0.8092755734750158
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 5,
|
| 315 |
+
"train_loss": 0.34133534183043585,
|
| 316 |
+
"val_loss": 0.4490610392663279,
|
| 317 |
+
"macro_f1": 0.8514615846241297
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 6,
|
| 321 |
+
"train_loss": 0.28462050195539923,
|
| 322 |
+
"val_loss": 0.4344335053795077,
|
| 323 |
+
"macro_f1": 0.8501548126586678
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 7,
|
| 327 |
+
"train_loss": 0.2287863478266649,
|
| 328 |
+
"val_loss": 0.43782980841001856,
|
| 329 |
+
"macro_f1": 0.8551092733111149
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 8,
|
| 333 |
+
"train_loss": 0.17019800981449287,
|
| 334 |
+
"val_loss": 0.4136471379493258,
|
| 335 |
+
"macro_f1": 0.8771639159798625
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 9,
|
| 339 |
+
"train_loss": 0.13897078628760892,
|
| 340 |
+
"val_loss": 0.3903981795387682,
|
| 341 |
+
"macro_f1": 0.8690399950721789
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 10,
|
| 345 |
+
"train_loss": 0.1324176407532833,
|
| 346 |
+
"val_loss": 0.39504132581315016,
|
| 347 |
+
"macro_f1": 0.8887459647104133
|
| 348 |
+
}
|
| 349 |
+
],
|
| 350 |
+
"final_macro_f1": 0.8887459647104133
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"trial_number": 8,
|
| 354 |
+
"params": {
|
| 355 |
+
"num_heads": 8,
|
| 356 |
+
"num_layers": 3
|
| 357 |
+
},
|
| 358 |
+
"epoch_metrics": [
|
| 359 |
+
{
|
| 360 |
+
"epoch": 1,
|
| 361 |
+
"train_loss": 1.2111919544334036,
|
| 362 |
+
"val_loss": 0.7954880866804733,
|
| 363 |
+
"macro_f1": 0.692173249027094
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 2,
|
| 367 |
+
"train_loss": 0.6736151260012616,
|
| 368 |
+
"val_loss": 0.6812303161679818,
|
| 369 |
+
"macro_f1": 0.7403447069975888
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 3,
|
| 373 |
+
"train_loss": 0.5472460934655734,
|
| 374 |
+
"val_loss": 0.5812115360950602,
|
| 375 |
+
"macro_f1": 0.7688338387674025
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 4,
|
| 379 |
+
"train_loss": 0.46018385821918967,
|
| 380 |
+
"val_loss": 0.5203743778132452,
|
| 381 |
+
"macro_f1": 0.8032102618967811
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 5,
|
| 385 |
+
"train_loss": 0.37938465003461636,
|
| 386 |
+
"val_loss": 0.5028770503409097,
|
| 387 |
+
"macro_f1": 0.8164099342185992
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 6,
|
| 391 |
+
"train_loss": 0.316088581005279,
|
| 392 |
+
"val_loss": 0.41271519390269745,
|
| 393 |
+
"macro_f1": 0.8469800383056026
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 7,
|
| 397 |
+
"train_loss": 0.25914997950141144,
|
| 398 |
+
"val_loss": 0.44095426879529587,
|
| 399 |
+
"macro_f1": 0.8534820881525831
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 8,
|
| 403 |
+
"train_loss": 0.20093080279694092,
|
| 404 |
+
"val_loss": 0.4070406252233853,
|
| 405 |
+
"macro_f1": 0.8711430333008884
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 9,
|
| 409 |
+
"train_loss": 0.18884633775942863,
|
| 410 |
+
"val_loss": 0.4072782529949261,
|
| 411 |
+
"macro_f1": 0.8589692051642381
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 10,
|
| 415 |
+
"train_loss": 0.13745949061360077,
|
| 416 |
+
"val_loss": 0.38686535329434085,
|
| 417 |
+
"macro_f1": 0.8790647966307543
|
| 418 |
+
}
|
| 419 |
+
],
|
| 420 |
+
"final_macro_f1": 0.8790647966307543
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"trial_number": 9,
|
| 424 |
+
"params": {
|
| 425 |
+
"num_heads": 6,
|
| 426 |
+
"num_layers": 1
|
| 427 |
+
},
|
| 428 |
+
"epoch_metrics": [
|
| 429 |
+
{
|
| 430 |
+
"epoch": 1,
|
| 431 |
+
"train_loss": 1.1922890041230199,
|
| 432 |
+
"val_loss": 0.717364141946943,
|
| 433 |
+
"macro_f1": 0.7111832146234033
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 2,
|
| 437 |
+
"train_loss": 0.614923968415078,
|
| 438 |
+
"val_loss": 0.6068004227330532,
|
| 439 |
+
"macro_f1": 0.7572401128403796
|
| 440 |
+
},
|
| 441 |
+
{
|
| 442 |
+
"epoch": 3,
|
| 443 |
+
"train_loss": 0.5134530554166117,
|
| 444 |
+
"val_loss": 0.5359148223118242,
|
| 445 |
+
"macro_f1": 0.8108260298969783
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 4,
|
| 449 |
+
"train_loss": 0.4162116786775107,
|
| 450 |
+
"val_loss": 0.5204935171113813,
|
| 451 |
+
"macro_f1": 0.8078057743572108
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 5,
|
| 455 |
+
"train_loss": 0.33961279247456794,
|
| 456 |
+
"val_loss": 0.5185712169984291,
|
| 457 |
+
"macro_f1": 0.8156870159488682
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 6,
|
| 461 |
+
"train_loss": 0.2688734014731006,
|
| 462 |
+
"val_loss": 0.5058216502422036,
|
| 463 |
+
"macro_f1": 0.8127487330103519
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 7,
|
| 467 |
+
"train_loss": 0.24128333852154407,
|
| 468 |
+
"val_loss": 0.44323977637173506,
|
| 469 |
+
"macro_f1": 0.8459015831040886
|
| 470 |
+
},
|
| 471 |
+
{
|
| 472 |
+
"epoch": 8,
|
| 473 |
+
"train_loss": 0.19852170058779886,
|
| 474 |
+
"val_loss": 0.3472442598325278,
|
| 475 |
+
"macro_f1": 0.8831003044387479
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 9,
|
| 479 |
+
"train_loss": 0.12439552349590785,
|
| 480 |
+
"val_loss": 0.3821052959179643,
|
| 481 |
+
"macro_f1": 0.8758038683098553
|
| 482 |
+
},
|
| 483 |
+
{
|
| 484 |
+
"epoch": 10,
|
| 485 |
+
"train_loss": 0.14608042874935045,
|
| 486 |
+
"val_loss": 0.3752268017476122,
|
| 487 |
+
"macro_f1": 0.8712872641014485
|
| 488 |
+
}
|
| 489 |
+
],
|
| 490 |
+
"final_macro_f1": 0.8712872641014485
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"trial_number": 11,
|
| 494 |
+
"params": {
|
| 495 |
+
"num_heads": 8,
|
| 496 |
+
"num_layers": 3
|
| 497 |
+
},
|
| 498 |
+
"epoch_metrics": [
|
| 499 |
+
{
|
| 500 |
+
"epoch": 1,
|
| 501 |
+
"train_loss": 1.3417097340382718,
|
| 502 |
+
"val_loss": 0.7926707279505988,
|
| 503 |
+
"macro_f1": 0.7129917353338494
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 2,
|
| 507 |
+
"train_loss": 0.6865363907431851,
|
| 508 |
+
"val_loss": 0.6855129791891633,
|
| 509 |
+
"macro_f1": 0.7429053146243679
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 3,
|
| 513 |
+
"train_loss": 0.5470005607854864,
|
| 514 |
+
"val_loss": 0.6210718623229435,
|
| 515 |
+
"macro_f1": 0.763000265572399
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 4,
|
| 519 |
+
"train_loss": 0.4693222763946994,
|
| 520 |
+
"val_loss": 0.5969800439300795,
|
| 521 |
+
"macro_f1": 0.7693955538580185
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 5,
|
| 525 |
+
"train_loss": 0.36977021348388217,
|
| 526 |
+
"val_loss": 0.5299910488974284,
|
| 527 |
+
"macro_f1": 0.8096957303892384
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 6,
|
| 531 |
+
"train_loss": 0.31427258129992997,
|
| 532 |
+
"val_loss": 0.525438177046077,
|
| 533 |
+
"macro_f1": 0.830292606084595
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 7,
|
| 537 |
+
"train_loss": 0.278600519309429,
|
| 538 |
+
"val_loss": 0.4611006113402362,
|
| 539 |
+
"macro_f1": 0.8491567596313283
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 8,
|
| 543 |
+
"train_loss": 0.21794261691492378,
|
| 544 |
+
"val_loss": 0.5124526821774215,
|
| 545 |
+
"macro_f1": 0.8308818032791065
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 9,
|
| 549 |
+
"train_loss": 0.2131353972018059,
|
| 550 |
+
"val_loss": 0.4707893036269202,
|
| 551 |
+
"macro_f1": 0.8478198878118651
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 10,
|
| 555 |
+
"train_loss": 0.15050037305815886,
|
| 556 |
+
"val_loss": 0.4796101963101643,
|
| 557 |
+
"macro_f1": 0.8644901534658801
|
| 558 |
+
}
|
| 559 |
+
],
|
| 560 |
+
"final_macro_f1": 0.8644901534658801
|
| 561 |
+
}
|
| 562 |
+
]
|