Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +547 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,547 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.8433710108449586,
|
| 3 |
+
"best_model_checkpoint": "result/my-unsup-simcse-roberta-base/SupCon-48-0.2-3e-6-0.07-0.00",
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
+
"global_step": 7813,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.02,
|
| 12 |
+
"eval_avg_sts": 0.7648697833708831,
|
| 13 |
+
"eval_sickr_spearman": 0.7122734723505795,
|
| 14 |
+
"eval_stsb_spearman": 0.8174660943911868,
|
| 15 |
+
"step": 125
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.03,
|
| 19 |
+
"eval_avg_sts": 0.7713268000072269,
|
| 20 |
+
"eval_sickr_spearman": 0.7159774387690809,
|
| 21 |
+
"eval_stsb_spearman": 0.8266761612453729,
|
| 22 |
+
"step": 250
|
| 23 |
+
},
|
| 24 |
+
{
|
| 25 |
+
"epoch": 0.05,
|
| 26 |
+
"eval_avg_sts": 0.7717212045194138,
|
| 27 |
+
"eval_sickr_spearman": 0.716408085624575,
|
| 28 |
+
"eval_stsb_spearman": 0.8270343234142526,
|
| 29 |
+
"step": 375
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"epoch": 0.06,
|
| 33 |
+
"learning_rate": 1.872008191475746e-05,
|
| 34 |
+
"loss": 0.1356,
|
| 35 |
+
"step": 500
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"epoch": 0.06,
|
| 39 |
+
"eval_avg_sts": 0.7714897999409676,
|
| 40 |
+
"eval_sickr_spearman": 0.714245581344527,
|
| 41 |
+
"eval_stsb_spearman": 0.8287340185374082,
|
| 42 |
+
"step": 500
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"epoch": 0.08,
|
| 46 |
+
"eval_avg_sts": 0.7683945074315737,
|
| 47 |
+
"eval_sickr_spearman": 0.707755714980412,
|
| 48 |
+
"eval_stsb_spearman": 0.8290332998827354,
|
| 49 |
+
"step": 625
|
| 50 |
+
},
|
| 51 |
+
{
|
| 52 |
+
"epoch": 0.1,
|
| 53 |
+
"eval_avg_sts": 0.7712141135572375,
|
| 54 |
+
"eval_sickr_spearman": 0.7136527334594763,
|
| 55 |
+
"eval_stsb_spearman": 0.8287754936549987,
|
| 56 |
+
"step": 750
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.11,
|
| 60 |
+
"eval_avg_sts": 0.7754955425035636,
|
| 61 |
+
"eval_sickr_spearman": 0.7167608740646381,
|
| 62 |
+
"eval_stsb_spearman": 0.8342302109424891,
|
| 63 |
+
"step": 875
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.13,
|
| 67 |
+
"learning_rate": 1.7440163829514915e-05,
|
| 68 |
+
"loss": 0.0005,
|
| 69 |
+
"step": 1000
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.13,
|
| 73 |
+
"eval_avg_sts": 0.7749643365975677,
|
| 74 |
+
"eval_sickr_spearman": 0.7165658197616978,
|
| 75 |
+
"eval_stsb_spearman": 0.8333628534334376,
|
| 76 |
+
"step": 1000
|
| 77 |
+
},
|
| 78 |
+
{
|
| 79 |
+
"epoch": 0.14,
|
| 80 |
+
"eval_avg_sts": 0.7747453579869918,
|
| 81 |
+
"eval_sickr_spearman": 0.707071944220339,
|
| 82 |
+
"eval_stsb_spearman": 0.8424187717536445,
|
| 83 |
+
"step": 1125
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 0.16,
|
| 87 |
+
"eval_avg_sts": 0.779538087156316,
|
| 88 |
+
"eval_sickr_spearman": 0.7169955059951632,
|
| 89 |
+
"eval_stsb_spearman": 0.8420806683174688,
|
| 90 |
+
"step": 1250
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.18,
|
| 94 |
+
"eval_avg_sts": 0.780701521846623,
|
| 95 |
+
"eval_sickr_spearman": 0.7190910068887052,
|
| 96 |
+
"eval_stsb_spearman": 0.8423120368045408,
|
| 97 |
+
"step": 1375
|
| 98 |
+
},
|
| 99 |
+
{
|
| 100 |
+
"epoch": 0.19,
|
| 101 |
+
"learning_rate": 1.616024574427237e-05,
|
| 102 |
+
"loss": 0.0003,
|
| 103 |
+
"step": 1500
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"epoch": 0.19,
|
| 107 |
+
"eval_avg_sts": 0.7822508597322496,
|
| 108 |
+
"eval_sickr_spearman": 0.725160745208541,
|
| 109 |
+
"eval_stsb_spearman": 0.8393409742559582,
|
| 110 |
+
"step": 1500
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.21,
|
| 114 |
+
"eval_avg_sts": 0.7806747469546087,
|
| 115 |
+
"eval_sickr_spearman": 0.7225573634483924,
|
| 116 |
+
"eval_stsb_spearman": 0.8387921304608251,
|
| 117 |
+
"step": 1625
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.22,
|
| 121 |
+
"eval_avg_sts": 0.783739242864979,
|
| 122 |
+
"eval_sickr_spearman": 0.7298345731433808,
|
| 123 |
+
"eval_stsb_spearman": 0.8376439125865772,
|
| 124 |
+
"step": 1750
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 0.24,
|
| 128 |
+
"eval_avg_sts": 0.7819519916053913,
|
| 129 |
+
"eval_sickr_spearman": 0.7310624707355461,
|
| 130 |
+
"eval_stsb_spearman": 0.8328415124752365,
|
| 131 |
+
"step": 1875
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 0.26,
|
| 135 |
+
"learning_rate": 1.4880327659029823e-05,
|
| 136 |
+
"loss": 0.0003,
|
| 137 |
+
"step": 2000
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 0.26,
|
| 141 |
+
"eval_avg_sts": 0.7824129920157619,
|
| 142 |
+
"eval_sickr_spearman": 0.731245325138721,
|
| 143 |
+
"eval_stsb_spearman": 0.8335806588928026,
|
| 144 |
+
"step": 2000
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.27,
|
| 148 |
+
"eval_avg_sts": 0.786146631865762,
|
| 149 |
+
"eval_sickr_spearman": 0.7289222528865652,
|
| 150 |
+
"eval_stsb_spearman": 0.8433710108449586,
|
| 151 |
+
"step": 2125
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 0.29,
|
| 155 |
+
"eval_avg_sts": 0.7797286001800978,
|
| 156 |
+
"eval_sickr_spearman": 0.7224563060109667,
|
| 157 |
+
"eval_stsb_spearman": 0.8370008943492291,
|
| 158 |
+
"step": 2250
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.3,
|
| 162 |
+
"eval_avg_sts": 0.780005893079553,
|
| 163 |
+
"eval_sickr_spearman": 0.7195845264559818,
|
| 164 |
+
"eval_stsb_spearman": 0.8404272597031243,
|
| 165 |
+
"step": 2375
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 0.32,
|
| 169 |
+
"learning_rate": 1.3600409573787278e-05,
|
| 170 |
+
"loss": 0.0003,
|
| 171 |
+
"step": 2500
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.32,
|
| 175 |
+
"eval_avg_sts": 0.7815935533979442,
|
| 176 |
+
"eval_sickr_spearman": 0.7247260157094245,
|
| 177 |
+
"eval_stsb_spearman": 0.8384610910864639,
|
| 178 |
+
"step": 2500
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.34,
|
| 182 |
+
"eval_avg_sts": 0.7790885863000977,
|
| 183 |
+
"eval_sickr_spearman": 0.719636736263245,
|
| 184 |
+
"eval_stsb_spearman": 0.8385404363369504,
|
| 185 |
+
"step": 2625
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.35,
|
| 189 |
+
"eval_avg_sts": 0.7786611915735007,
|
| 190 |
+
"eval_sickr_spearman": 0.720594044146008,
|
| 191 |
+
"eval_stsb_spearman": 0.8367283390009934,
|
| 192 |
+
"step": 2750
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.37,
|
| 196 |
+
"eval_avg_sts": 0.7740761960283342,
|
| 197 |
+
"eval_sickr_spearman": 0.7133921647341745,
|
| 198 |
+
"eval_stsb_spearman": 0.8347602273224938,
|
| 199 |
+
"step": 2875
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.38,
|
| 203 |
+
"learning_rate": 1.2320491488544735e-05,
|
| 204 |
+
"loss": 0.0003,
|
| 205 |
+
"step": 3000
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.38,
|
| 209 |
+
"eval_avg_sts": 0.7730594558706145,
|
| 210 |
+
"eval_sickr_spearman": 0.7107198103123609,
|
| 211 |
+
"eval_stsb_spearman": 0.8353991014288681,
|
| 212 |
+
"step": 3000
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.4,
|
| 216 |
+
"eval_avg_sts": 0.7735160682870633,
|
| 217 |
+
"eval_sickr_spearman": 0.7115497397141101,
|
| 218 |
+
"eval_stsb_spearman": 0.8354823968600166,
|
| 219 |
+
"step": 3125
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.42,
|
| 223 |
+
"eval_avg_sts": 0.7696881215537197,
|
| 224 |
+
"eval_sickr_spearman": 0.7034571235261068,
|
| 225 |
+
"eval_stsb_spearman": 0.8359191195813325,
|
| 226 |
+
"step": 3250
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.43,
|
| 230 |
+
"eval_avg_sts": 0.7731307195393922,
|
| 231 |
+
"eval_sickr_spearman": 0.7140373184886913,
|
| 232 |
+
"eval_stsb_spearman": 0.8322241205900931,
|
| 233 |
+
"step": 3375
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.45,
|
| 237 |
+
"learning_rate": 1.104057340330219e-05,
|
| 238 |
+
"loss": 0.0003,
|
| 239 |
+
"step": 3500
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 0.45,
|
| 243 |
+
"eval_avg_sts": 0.7744383399418961,
|
| 244 |
+
"eval_sickr_spearman": 0.7103547259103297,
|
| 245 |
+
"eval_stsb_spearman": 0.8385219539734627,
|
| 246 |
+
"step": 3500
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 0.46,
|
| 250 |
+
"eval_avg_sts": 0.7749032723185794,
|
| 251 |
+
"eval_sickr_spearman": 0.7106885420653246,
|
| 252 |
+
"eval_stsb_spearman": 0.839118002571834,
|
| 253 |
+
"step": 3625
|
| 254 |
+
},
|
| 255 |
+
{
|
| 256 |
+
"epoch": 0.48,
|
| 257 |
+
"eval_avg_sts": 0.7729790005154229,
|
| 258 |
+
"eval_sickr_spearman": 0.7092490979863258,
|
| 259 |
+
"eval_stsb_spearman": 0.8367089030445198,
|
| 260 |
+
"step": 3750
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 0.5,
|
| 264 |
+
"eval_avg_sts": 0.7675103682931668,
|
| 265 |
+
"eval_sickr_spearman": 0.7009157018868134,
|
| 266 |
+
"eval_stsb_spearman": 0.83410503469952,
|
| 267 |
+
"step": 3875
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 0.51,
|
| 271 |
+
"learning_rate": 9.760655318059645e-06,
|
| 272 |
+
"loss": 0.0003,
|
| 273 |
+
"step": 4000
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 0.51,
|
| 277 |
+
"eval_avg_sts": 0.7600618545185335,
|
| 278 |
+
"eval_sickr_spearman": 0.6939802029635556,
|
| 279 |
+
"eval_stsb_spearman": 0.8261435060735114,
|
| 280 |
+
"step": 4000
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.53,
|
| 284 |
+
"eval_avg_sts": 0.7611442447284995,
|
| 285 |
+
"eval_sickr_spearman": 0.6946870766834202,
|
| 286 |
+
"eval_stsb_spearman": 0.8276014127735787,
|
| 287 |
+
"step": 4125
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 0.54,
|
| 291 |
+
"eval_avg_sts": 0.765201470252675,
|
| 292 |
+
"eval_sickr_spearman": 0.6980777842583424,
|
| 293 |
+
"eval_stsb_spearman": 0.8323251562470075,
|
| 294 |
+
"step": 4250
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.56,
|
| 298 |
+
"eval_avg_sts": 0.7601537112935712,
|
| 299 |
+
"eval_sickr_spearman": 0.6892433277086275,
|
| 300 |
+
"eval_stsb_spearman": 0.831064094878515,
|
| 301 |
+
"step": 4375
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"epoch": 0.58,
|
| 305 |
+
"learning_rate": 8.4807372328171e-06,
|
| 306 |
+
"loss": 0.0002,
|
| 307 |
+
"step": 4500
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 0.58,
|
| 311 |
+
"eval_avg_sts": 0.7572602743940218,
|
| 312 |
+
"eval_sickr_spearman": 0.6881403895263041,
|
| 313 |
+
"eval_stsb_spearman": 0.8263801592617394,
|
| 314 |
+
"step": 4500
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.59,
|
| 318 |
+
"eval_avg_sts": 0.7620805832344215,
|
| 319 |
+
"eval_sickr_spearman": 0.6991238944111545,
|
| 320 |
+
"eval_stsb_spearman": 0.8250372720576886,
|
| 321 |
+
"step": 4625
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 0.61,
|
| 325 |
+
"eval_avg_sts": 0.7638147878715696,
|
| 326 |
+
"eval_sickr_spearman": 0.6998619475523625,
|
| 327 |
+
"eval_stsb_spearman": 0.8277676281907768,
|
| 328 |
+
"step": 4750
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 0.62,
|
| 332 |
+
"eval_avg_sts": 0.762036570031333,
|
| 333 |
+
"eval_sickr_spearman": 0.6968174001255049,
|
| 334 |
+
"eval_stsb_spearman": 0.8272557399371612,
|
| 335 |
+
"step": 4875
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 0.64,
|
| 339 |
+
"learning_rate": 7.200819147574556e-06,
|
| 340 |
+
"loss": 0.0002,
|
| 341 |
+
"step": 5000
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 0.64,
|
| 345 |
+
"eval_avg_sts": 0.7616123775261012,
|
| 346 |
+
"eval_sickr_spearman": 0.6975154841538064,
|
| 347 |
+
"eval_stsb_spearman": 0.8257092708983962,
|
| 348 |
+
"step": 5000
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 0.66,
|
| 352 |
+
"eval_avg_sts": 0.7572440478626198,
|
| 353 |
+
"eval_sickr_spearman": 0.6909725434226024,
|
| 354 |
+
"eval_stsb_spearman": 0.8235155523026372,
|
| 355 |
+
"step": 5125
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 0.67,
|
| 359 |
+
"eval_avg_sts": 0.7615420687641667,
|
| 360 |
+
"eval_sickr_spearman": 0.7007691109652242,
|
| 361 |
+
"eval_stsb_spearman": 0.8223150265631094,
|
| 362 |
+
"step": 5250
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.69,
|
| 366 |
+
"eval_avg_sts": 0.7622736227072948,
|
| 367 |
+
"eval_sickr_spearman": 0.70394151718411,
|
| 368 |
+
"eval_stsb_spearman": 0.8206057282304795,
|
| 369 |
+
"step": 5375
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.7,
|
| 373 |
+
"learning_rate": 5.920901062332012e-06,
|
| 374 |
+
"loss": 0.0002,
|
| 375 |
+
"step": 5500
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.7,
|
| 379 |
+
"eval_avg_sts": 0.7598959005042131,
|
| 380 |
+
"eval_sickr_spearman": 0.7014216615093624,
|
| 381 |
+
"eval_stsb_spearman": 0.8183701394990638,
|
| 382 |
+
"step": 5500
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 0.72,
|
| 386 |
+
"eval_avg_sts": 0.7617340962416614,
|
| 387 |
+
"eval_sickr_spearman": 0.705636198536151,
|
| 388 |
+
"eval_stsb_spearman": 0.8178319939471719,
|
| 389 |
+
"step": 5625
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 0.74,
|
| 393 |
+
"eval_avg_sts": 0.7602681846424262,
|
| 394 |
+
"eval_sickr_spearman": 0.6992537297148518,
|
| 395 |
+
"eval_stsb_spearman": 0.8212826395700007,
|
| 396 |
+
"step": 5750
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 0.75,
|
| 400 |
+
"eval_avg_sts": 0.7600256686195976,
|
| 401 |
+
"eval_sickr_spearman": 0.6983496883235836,
|
| 402 |
+
"eval_stsb_spearman": 0.8217016489156117,
|
| 403 |
+
"step": 5875
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 0.77,
|
| 407 |
+
"learning_rate": 4.640982977089467e-06,
|
| 408 |
+
"loss": 0.0003,
|
| 409 |
+
"step": 6000
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.77,
|
| 413 |
+
"eval_avg_sts": 0.7581145273334985,
|
| 414 |
+
"eval_sickr_spearman": 0.6957373727785674,
|
| 415 |
+
"eval_stsb_spearman": 0.8204916818884296,
|
| 416 |
+
"step": 6000
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.78,
|
| 420 |
+
"eval_avg_sts": 0.759467788986251,
|
| 421 |
+
"eval_sickr_spearman": 0.6969048167301223,
|
| 422 |
+
"eval_stsb_spearman": 0.8220307612423797,
|
| 423 |
+
"step": 6125
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.8,
|
| 427 |
+
"eval_avg_sts": 0.758567409617105,
|
| 428 |
+
"eval_sickr_spearman": 0.6932303894390052,
|
| 429 |
+
"eval_stsb_spearman": 0.8239044297952048,
|
| 430 |
+
"step": 6250
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.82,
|
| 434 |
+
"eval_avg_sts": 0.7593640477073755,
|
| 435 |
+
"eval_sickr_spearman": 0.6939898572149447,
|
| 436 |
+
"eval_stsb_spearman": 0.8247382381998064,
|
| 437 |
+
"step": 6375
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.83,
|
| 441 |
+
"learning_rate": 3.3610648918469217e-06,
|
| 442 |
+
"loss": 0.0002,
|
| 443 |
+
"step": 6500
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.83,
|
| 447 |
+
"eval_avg_sts": 0.7596970769949833,
|
| 448 |
+
"eval_sickr_spearman": 0.6966290701767661,
|
| 449 |
+
"eval_stsb_spearman": 0.8227650838132007,
|
| 450 |
+
"step": 6500
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.85,
|
| 454 |
+
"eval_avg_sts": 0.7608506682812288,
|
| 455 |
+
"eval_sickr_spearman": 0.6957939053849601,
|
| 456 |
+
"eval_stsb_spearman": 0.8259074311774975,
|
| 457 |
+
"step": 6625
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.86,
|
| 461 |
+
"eval_avg_sts": 0.7601011463122427,
|
| 462 |
+
"eval_sickr_spearman": 0.6948619579237564,
|
| 463 |
+
"eval_stsb_spearman": 0.8253403347007289,
|
| 464 |
+
"step": 6750
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.88,
|
| 468 |
+
"eval_avg_sts": 0.7613624100018648,
|
| 469 |
+
"eval_sickr_spearman": 0.6960120626476921,
|
| 470 |
+
"eval_stsb_spearman": 0.8267127573560374,
|
| 471 |
+
"step": 6875
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.9,
|
| 475 |
+
"learning_rate": 2.0811468066043775e-06,
|
| 476 |
+
"loss": 0.0002,
|
| 477 |
+
"step": 7000
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 0.9,
|
| 481 |
+
"eval_avg_sts": 0.7632853767372183,
|
| 482 |
+
"eval_sickr_spearman": 0.6985597763612739,
|
| 483 |
+
"eval_stsb_spearman": 0.8280109771131627,
|
| 484 |
+
"step": 7000
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 0.91,
|
| 488 |
+
"eval_avg_sts": 0.761317305690945,
|
| 489 |
+
"eval_sickr_spearman": 0.6965424220697717,
|
| 490 |
+
"eval_stsb_spearman": 0.8260921893121185,
|
| 491 |
+
"step": 7125
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 0.93,
|
| 495 |
+
"eval_avg_sts": 0.7612243104925751,
|
| 496 |
+
"eval_sickr_spearman": 0.6965610101060282,
|
| 497 |
+
"eval_stsb_spearman": 0.8258876108791219,
|
| 498 |
+
"step": 7250
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 0.94,
|
| 502 |
+
"eval_avg_sts": 0.7611869303629479,
|
| 503 |
+
"eval_sickr_spearman": 0.6964085113589622,
|
| 504 |
+
"eval_stsb_spearman": 0.8259653493669338,
|
| 505 |
+
"step": 7375
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 0.96,
|
| 509 |
+
"learning_rate": 8.012287213618329e-07,
|
| 510 |
+
"loss": 0.0002,
|
| 511 |
+
"step": 7500
|
| 512 |
+
},
|
| 513 |
+
{
|
| 514 |
+
"epoch": 0.96,
|
| 515 |
+
"eval_avg_sts": 0.761389925660918,
|
| 516 |
+
"eval_sickr_spearman": 0.6968965073495735,
|
| 517 |
+
"eval_stsb_spearman": 0.8258833439722626,
|
| 518 |
+
"step": 7500
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 0.98,
|
| 522 |
+
"eval_avg_sts": 0.7614202638478765,
|
| 523 |
+
"eval_sickr_spearman": 0.6971455005794286,
|
| 524 |
+
"eval_stsb_spearman": 0.8256950271163245,
|
| 525 |
+
"step": 7625
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 0.99,
|
| 529 |
+
"eval_avg_sts": 0.7612860433632234,
|
| 530 |
+
"eval_sickr_spearman": 0.6969965561338691,
|
| 531 |
+
"eval_stsb_spearman": 0.8255755305925775,
|
| 532 |
+
"step": 7750
|
| 533 |
+
},
|
| 534 |
+
{
|
| 535 |
+
"epoch": 1.0,
|
| 536 |
+
"step": 7813,
|
| 537 |
+
"total_flos": 48546397602349056,
|
| 538 |
+
"train_runtime": 6045.9604,
|
| 539 |
+
"train_samples_per_second": 1.292
|
| 540 |
+
}
|
| 541 |
+
],
|
| 542 |
+
"max_steps": 7813,
|
| 543 |
+
"num_train_epochs": 1,
|
| 544 |
+
"total_flos": 48546397602349056,
|
| 545 |
+
"trial_name": null,
|
| 546 |
+
"trial_params": null
|
| 547 |
+
}
|