Training in progress, epoch 3, checkpoint
Browse files
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1256
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c62c435fa96b78ad6ecac93429cba4d88f3f2af5297b739e4a5280675f4784cf
|
| 3 |
size 1256
|
last-checkpoint/rng_state.pth
CHANGED
|
Binary files a/last-checkpoint/rng_state.pth and b/last-checkpoint/rng_state.pth differ
|
|
|
last-checkpoint/scheduler.pt
CHANGED
|
Binary files a/last-checkpoint/scheduler.pt and b/last-checkpoint/scheduler.pt differ
|
|
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": 2.873922109603882,
|
| 3 |
"best_model_checkpoint": "dq158/coqui/checkpoint-40162",
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -1005,13 +1005,512 @@
|
|
| 1005 |
"eval_steps_per_second": 0.568,
|
| 1006 |
"eval_translation_length": 4569600,
|
| 1007 |
"step": 80324
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1008 |
}
|
| 1009 |
],
|
| 1010 |
"logging_steps": 500,
|
| 1011 |
"max_steps": 1204860,
|
| 1012 |
"num_train_epochs": 30,
|
| 1013 |
"save_steps": 1000,
|
| 1014 |
-
"total_flos":
|
| 1015 |
"trial_name": null,
|
| 1016 |
"trial_params": null
|
| 1017 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": 2.873922109603882,
|
| 3 |
"best_model_checkpoint": "dq158/coqui/checkpoint-40162",
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 120486,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 1005 |
"eval_steps_per_second": 0.568,
|
| 1006 |
"eval_translation_length": 4569600,
|
| 1007 |
"step": 80324
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 2.0,
|
| 1011 |
+
"learning_rate": 7.913219840074152e-05,
|
| 1012 |
+
"loss": 3.0052,
|
| 1013 |
+
"step": 80500
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 2.02,
|
| 1017 |
+
"learning_rate": 7.912135699428588e-05,
|
| 1018 |
+
"loss": 2.9191,
|
| 1019 |
+
"step": 81000
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 2.03,
|
| 1023 |
+
"learning_rate": 7.91104490389186e-05,
|
| 1024 |
+
"loss": 3.0454,
|
| 1025 |
+
"step": 81500
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 2.04,
|
| 1029 |
+
"learning_rate": 7.909947455319513e-05,
|
| 1030 |
+
"loss": 3.0362,
|
| 1031 |
+
"step": 82000
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 2.05,
|
| 1035 |
+
"learning_rate": 7.908843355578402e-05,
|
| 1036 |
+
"loss": 3.016,
|
| 1037 |
+
"step": 82500
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 2.07,
|
| 1041 |
+
"learning_rate": 7.907732606546699e-05,
|
| 1042 |
+
"loss": 3.0735,
|
| 1043 |
+
"step": 83000
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 2.08,
|
| 1047 |
+
"learning_rate": 7.906615210113886e-05,
|
| 1048 |
+
"loss": 3.0626,
|
| 1049 |
+
"step": 83500
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 2.09,
|
| 1053 |
+
"learning_rate": 7.905491168180757e-05,
|
| 1054 |
+
"loss": 2.9955,
|
| 1055 |
+
"step": 84000
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 2.1,
|
| 1059 |
+
"learning_rate": 7.904360482659405e-05,
|
| 1060 |
+
"loss": 3.0104,
|
| 1061 |
+
"step": 84500
|
| 1062 |
+
},
|
| 1063 |
+
{
|
| 1064 |
+
"epoch": 2.12,
|
| 1065 |
+
"learning_rate": 7.903223155473227e-05,
|
| 1066 |
+
"loss": 3.0618,
|
| 1067 |
+
"step": 85000
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 2.13,
|
| 1071 |
+
"learning_rate": 7.902079188556919e-05,
|
| 1072 |
+
"loss": 3.0146,
|
| 1073 |
+
"step": 85500
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 2.14,
|
| 1077 |
+
"learning_rate": 7.900928583856469e-05,
|
| 1078 |
+
"loss": 3.0075,
|
| 1079 |
+
"step": 86000
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 2.15,
|
| 1083 |
+
"learning_rate": 7.899771343329159e-05,
|
| 1084 |
+
"loss": 2.9807,
|
| 1085 |
+
"step": 86500
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 2.17,
|
| 1089 |
+
"learning_rate": 7.898607468943558e-05,
|
| 1090 |
+
"loss": 3.0609,
|
| 1091 |
+
"step": 87000
|
| 1092 |
+
},
|
| 1093 |
+
{
|
| 1094 |
+
"epoch": 2.18,
|
| 1095 |
+
"learning_rate": 7.89743696267952e-05,
|
| 1096 |
+
"loss": 2.9379,
|
| 1097 |
+
"step": 87500
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 2.19,
|
| 1101 |
+
"learning_rate": 7.896259826528179e-05,
|
| 1102 |
+
"loss": 3.0411,
|
| 1103 |
+
"step": 88000
|
| 1104 |
+
},
|
| 1105 |
+
{
|
| 1106 |
+
"epoch": 2.2,
|
| 1107 |
+
"learning_rate": 7.89507606249195e-05,
|
| 1108 |
+
"loss": 3.0788,
|
| 1109 |
+
"step": 88500
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 2.22,
|
| 1113 |
+
"learning_rate": 7.893885672584521e-05,
|
| 1114 |
+
"loss": 3.115,
|
| 1115 |
+
"step": 89000
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 2.23,
|
| 1119 |
+
"learning_rate": 7.892688658830852e-05,
|
| 1120 |
+
"loss": 3.0927,
|
| 1121 |
+
"step": 89500
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 2.24,
|
| 1125 |
+
"learning_rate": 7.891485023267167e-05,
|
| 1126 |
+
"loss": 2.9474,
|
| 1127 |
+
"step": 90000
|
| 1128 |
+
},
|
| 1129 |
+
{
|
| 1130 |
+
"epoch": 2.25,
|
| 1131 |
+
"learning_rate": 7.890274767940961e-05,
|
| 1132 |
+
"loss": 3.0862,
|
| 1133 |
+
"step": 90500
|
| 1134 |
+
},
|
| 1135 |
+
{
|
| 1136 |
+
"epoch": 2.27,
|
| 1137 |
+
"learning_rate": 7.889057894910983e-05,
|
| 1138 |
+
"loss": 3.0357,
|
| 1139 |
+
"step": 91000
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 2.28,
|
| 1143 |
+
"learning_rate": 7.887834406247244e-05,
|
| 1144 |
+
"loss": 3.0165,
|
| 1145 |
+
"step": 91500
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 2.29,
|
| 1149 |
+
"learning_rate": 7.886604304031007e-05,
|
| 1150 |
+
"loss": 3.0627,
|
| 1151 |
+
"step": 92000
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 2.3,
|
| 1155 |
+
"learning_rate": 7.885367590354785e-05,
|
| 1156 |
+
"loss": 3.0763,
|
| 1157 |
+
"step": 92500
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 2.32,
|
| 1161 |
+
"learning_rate": 7.88412426732234e-05,
|
| 1162 |
+
"loss": 3.0735,
|
| 1163 |
+
"step": 93000
|
| 1164 |
+
},
|
| 1165 |
+
{
|
| 1166 |
+
"epoch": 2.33,
|
| 1167 |
+
"learning_rate": 7.882874337048669e-05,
|
| 1168 |
+
"loss": 2.9523,
|
| 1169 |
+
"step": 93500
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 2.34,
|
| 1173 |
+
"learning_rate": 7.881617801660023e-05,
|
| 1174 |
+
"loss": 2.9885,
|
| 1175 |
+
"step": 94000
|
| 1176 |
+
},
|
| 1177 |
+
{
|
| 1178 |
+
"epoch": 2.35,
|
| 1179 |
+
"learning_rate": 7.880354663293875e-05,
|
| 1180 |
+
"loss": 2.9825,
|
| 1181 |
+
"step": 94500
|
| 1182 |
+
},
|
| 1183 |
+
{
|
| 1184 |
+
"epoch": 2.37,
|
| 1185 |
+
"learning_rate": 7.879084924098938e-05,
|
| 1186 |
+
"loss": 2.9537,
|
| 1187 |
+
"step": 95000
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 2.38,
|
| 1191 |
+
"learning_rate": 7.877808586235151e-05,
|
| 1192 |
+
"loss": 3.0941,
|
| 1193 |
+
"step": 95500
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 2.39,
|
| 1197 |
+
"learning_rate": 7.876525651873679e-05,
|
| 1198 |
+
"loss": 3.0817,
|
| 1199 |
+
"step": 96000
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 2.4,
|
| 1203 |
+
"learning_rate": 7.875236123196905e-05,
|
| 1204 |
+
"loss": 3.0927,
|
| 1205 |
+
"step": 96500
|
| 1206 |
+
},
|
| 1207 |
+
{
|
| 1208 |
+
"epoch": 2.42,
|
| 1209 |
+
"learning_rate": 7.873940002398435e-05,
|
| 1210 |
+
"loss": 2.9777,
|
| 1211 |
+
"step": 97000
|
| 1212 |
+
},
|
| 1213 |
+
{
|
| 1214 |
+
"epoch": 2.43,
|
| 1215 |
+
"learning_rate": 7.872637291683084e-05,
|
| 1216 |
+
"loss": 3.0594,
|
| 1217 |
+
"step": 97500
|
| 1218 |
+
},
|
| 1219 |
+
{
|
| 1220 |
+
"epoch": 2.44,
|
| 1221 |
+
"learning_rate": 7.871327993266882e-05,
|
| 1222 |
+
"loss": 3.0813,
|
| 1223 |
+
"step": 98000
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 2.45,
|
| 1227 |
+
"learning_rate": 7.870012109377058e-05,
|
| 1228 |
+
"loss": 3.0248,
|
| 1229 |
+
"step": 98500
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 2.47,
|
| 1233 |
+
"learning_rate": 7.868689642252051e-05,
|
| 1234 |
+
"loss": 2.9885,
|
| 1235 |
+
"step": 99000
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 2.48,
|
| 1239 |
+
"learning_rate": 7.867360594141493e-05,
|
| 1240 |
+
"loss": 3.037,
|
| 1241 |
+
"step": 99500
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 2.49,
|
| 1245 |
+
"learning_rate": 7.866024967306214e-05,
|
| 1246 |
+
"loss": 3.1189,
|
| 1247 |
+
"step": 100000
|
| 1248 |
+
},
|
| 1249 |
+
{
|
| 1250 |
+
"epoch": 2.5,
|
| 1251 |
+
"learning_rate": 7.864682764018236e-05,
|
| 1252 |
+
"loss": 2.9825,
|
| 1253 |
+
"step": 100500
|
| 1254 |
+
},
|
| 1255 |
+
{
|
| 1256 |
+
"epoch": 2.51,
|
| 1257 |
+
"learning_rate": 7.863333986560763e-05,
|
| 1258 |
+
"loss": 3.0277,
|
| 1259 |
+
"step": 101000
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 2.53,
|
| 1263 |
+
"learning_rate": 7.861978637228187e-05,
|
| 1264 |
+
"loss": 3.1551,
|
| 1265 |
+
"step": 101500
|
| 1266 |
+
},
|
| 1267 |
+
{
|
| 1268 |
+
"epoch": 2.54,
|
| 1269 |
+
"learning_rate": 7.860616718326079e-05,
|
| 1270 |
+
"loss": 3.0137,
|
| 1271 |
+
"step": 102000
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 2.55,
|
| 1275 |
+
"learning_rate": 7.859248232171184e-05,
|
| 1276 |
+
"loss": 3.0373,
|
| 1277 |
+
"step": 102500
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 2.56,
|
| 1281 |
+
"learning_rate": 7.857873181091416e-05,
|
| 1282 |
+
"loss": 3.0931,
|
| 1283 |
+
"step": 103000
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 2.58,
|
| 1287 |
+
"learning_rate": 7.856491567425863e-05,
|
| 1288 |
+
"loss": 3.0635,
|
| 1289 |
+
"step": 103500
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 2.59,
|
| 1293 |
+
"learning_rate": 7.85510339352477e-05,
|
| 1294 |
+
"loss": 3.0677,
|
| 1295 |
+
"step": 104000
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 2.6,
|
| 1299 |
+
"learning_rate": 7.853708661749548e-05,
|
| 1300 |
+
"loss": 3.0483,
|
| 1301 |
+
"step": 104500
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 2.61,
|
| 1305 |
+
"learning_rate": 7.852307374472755e-05,
|
| 1306 |
+
"loss": 2.9974,
|
| 1307 |
+
"step": 105000
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 2.63,
|
| 1311 |
+
"learning_rate": 7.850899534078111e-05,
|
| 1312 |
+
"loss": 3.0567,
|
| 1313 |
+
"step": 105500
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 2.64,
|
| 1317 |
+
"learning_rate": 7.849485142960473e-05,
|
| 1318 |
+
"loss": 3.0159,
|
| 1319 |
+
"step": 106000
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 2.65,
|
| 1323 |
+
"learning_rate": 7.848064203525849e-05,
|
| 1324 |
+
"loss": 3.0221,
|
| 1325 |
+
"step": 106500
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 2.66,
|
| 1329 |
+
"learning_rate": 7.846636718191382e-05,
|
| 1330 |
+
"loss": 2.9791,
|
| 1331 |
+
"step": 107000
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 2.68,
|
| 1335 |
+
"learning_rate": 7.845202689385353e-05,
|
| 1336 |
+
"loss": 2.9952,
|
| 1337 |
+
"step": 107500
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 2.69,
|
| 1341 |
+
"learning_rate": 7.843762119547172e-05,
|
| 1342 |
+
"loss": 2.997,
|
| 1343 |
+
"step": 108000
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 2.7,
|
| 1347 |
+
"learning_rate": 7.842315011127377e-05,
|
| 1348 |
+
"loss": 2.9545,
|
| 1349 |
+
"step": 108500
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 2.71,
|
| 1353 |
+
"learning_rate": 7.840861366587628e-05,
|
| 1354 |
+
"loss": 3.0655,
|
| 1355 |
+
"step": 109000
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 2.73,
|
| 1359 |
+
"learning_rate": 7.839401188400704e-05,
|
| 1360 |
+
"loss": 3.0671,
|
| 1361 |
+
"step": 109500
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 2.74,
|
| 1365 |
+
"learning_rate": 7.837934479050496e-05,
|
| 1366 |
+
"loss": 3.0327,
|
| 1367 |
+
"step": 110000
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 2.75,
|
| 1371 |
+
"learning_rate": 7.836461241032011e-05,
|
| 1372 |
+
"loss": 3.0038,
|
| 1373 |
+
"step": 110500
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 2.76,
|
| 1377 |
+
"learning_rate": 7.834981476851354e-05,
|
| 1378 |
+
"loss": 3.0443,
|
| 1379 |
+
"step": 111000
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 2.78,
|
| 1383 |
+
"learning_rate": 7.83349518902574e-05,
|
| 1384 |
+
"loss": 2.9654,
|
| 1385 |
+
"step": 111500
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 2.79,
|
| 1389 |
+
"learning_rate": 7.832002380083474e-05,
|
| 1390 |
+
"loss": 3.0335,
|
| 1391 |
+
"step": 112000
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 2.8,
|
| 1395 |
+
"learning_rate": 7.830503052563956e-05,
|
| 1396 |
+
"loss": 3.0491,
|
| 1397 |
+
"step": 112500
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 2.81,
|
| 1401 |
+
"learning_rate": 7.828997209017678e-05,
|
| 1402 |
+
"loss": 3.0643,
|
| 1403 |
+
"step": 113000
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 2.83,
|
| 1407 |
+
"learning_rate": 7.827484852006215e-05,
|
| 1408 |
+
"loss": 3.1021,
|
| 1409 |
+
"step": 113500
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 2.84,
|
| 1413 |
+
"learning_rate": 7.825965984102216e-05,
|
| 1414 |
+
"loss": 3.0631,
|
| 1415 |
+
"step": 114000
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 2.85,
|
| 1419 |
+
"learning_rate": 7.824440607889415e-05,
|
| 1420 |
+
"loss": 2.9952,
|
| 1421 |
+
"step": 114500
|
| 1422 |
+
},
|
| 1423 |
+
{
|
| 1424 |
+
"epoch": 2.86,
|
| 1425 |
+
"learning_rate": 7.822908725962612e-05,
|
| 1426 |
+
"loss": 3.0093,
|
| 1427 |
+
"step": 115000
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 2.88,
|
| 1431 |
+
"learning_rate": 7.821370340927674e-05,
|
| 1432 |
+
"loss": 2.9657,
|
| 1433 |
+
"step": 115500
|
| 1434 |
+
},
|
| 1435 |
+
{
|
| 1436 |
+
"epoch": 2.89,
|
| 1437 |
+
"learning_rate": 7.819825455401529e-05,
|
| 1438 |
+
"loss": 3.0453,
|
| 1439 |
+
"step": 116000
|
| 1440 |
+
},
|
| 1441 |
+
{
|
| 1442 |
+
"epoch": 2.9,
|
| 1443 |
+
"learning_rate": 7.818274072012168e-05,
|
| 1444 |
+
"loss": 3.0879,
|
| 1445 |
+
"step": 116500
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 2.91,
|
| 1449 |
+
"learning_rate": 7.81671619339863e-05,
|
| 1450 |
+
"loss": 2.9641,
|
| 1451 |
+
"step": 117000
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 2.93,
|
| 1455 |
+
"learning_rate": 7.815151822211007e-05,
|
| 1456 |
+
"loss": 3.0363,
|
| 1457 |
+
"step": 117500
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 2.94,
|
| 1461 |
+
"learning_rate": 7.813580961110435e-05,
|
| 1462 |
+
"loss": 3.051,
|
| 1463 |
+
"step": 118000
|
| 1464 |
+
},
|
| 1465 |
+
{
|
| 1466 |
+
"epoch": 2.95,
|
| 1467 |
+
"learning_rate": 7.812003612769084e-05,
|
| 1468 |
+
"loss": 3.0135,
|
| 1469 |
+
"step": 118500
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 2.96,
|
| 1473 |
+
"learning_rate": 7.810419779870166e-05,
|
| 1474 |
+
"loss": 2.9968,
|
| 1475 |
+
"step": 119000
|
| 1476 |
+
},
|
| 1477 |
+
{
|
| 1478 |
+
"epoch": 2.98,
|
| 1479 |
+
"learning_rate": 7.808829465107923e-05,
|
| 1480 |
+
"loss": 3.0569,
|
| 1481 |
+
"step": 119500
|
| 1482 |
+
},
|
| 1483 |
+
{
|
| 1484 |
+
"epoch": 2.99,
|
| 1485 |
+
"learning_rate": 7.80723267118762e-05,
|
| 1486 |
+
"loss": 3.0546,
|
| 1487 |
+
"step": 120000
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 3.0,
|
| 1491 |
+
"eval_bleu": 1.0,
|
| 1492 |
+
"eval_brevity_penalty": 1.0,
|
| 1493 |
+
"eval_length_ratio": 1.0,
|
| 1494 |
+
"eval_loss": 2.873922109603882,
|
| 1495 |
+
"eval_precisions": [
|
| 1496 |
+
1.0,
|
| 1497 |
+
1.0,
|
| 1498 |
+
1.0,
|
| 1499 |
+
1.0
|
| 1500 |
+
],
|
| 1501 |
+
"eval_reference_length": 4569600,
|
| 1502 |
+
"eval_runtime": 7544.7431,
|
| 1503 |
+
"eval_samples_per_second": 1.183,
|
| 1504 |
+
"eval_steps_per_second": 0.592,
|
| 1505 |
+
"eval_translation_length": 4569600,
|
| 1506 |
+
"step": 120486
|
| 1507 |
}
|
| 1508 |
],
|
| 1509 |
"logging_steps": 500,
|
| 1510 |
"max_steps": 1204860,
|
| 1511 |
"num_train_epochs": 30,
|
| 1512 |
"save_steps": 1000,
|
| 1513 |
+
"total_flos": 2.0678546883965092e+18,
|
| 1514 |
"trial_name": null,
|
| 1515 |
"trial_params": null
|
| 1516 |
}
|