Training in progress, epoch 3, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 37789864
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c41d9243f7df55b50aa4775a15858d272a4ec8e8c563c0cd8ec6b9d0b3da9f8a
|
| 3 |
size 37789864
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 2622266
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:29c84bda34e9c4f410c99669a1f937ad7a00bee4dfd64bcf19e7795e6886813d
|
| 3 |
size 2622266
|
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": 3.
|
| 3 |
-
"best_model_checkpoint": "dq158/coqui/checkpoint-
|
| 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,
|
|
@@ -1011,13 +1011,518 @@
|
|
| 1011 |
"eval_steps_per_second": 0.543,
|
| 1012 |
"eval_translation_length": 4591104,
|
| 1013 |
"step": 80696
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1014 |
}
|
| 1015 |
],
|
| 1016 |
"logging_steps": 500,
|
| 1017 |
"max_steps": 1210440,
|
| 1018 |
"num_train_epochs": 30,
|
| 1019 |
"save_steps": 1000,
|
| 1020 |
-
"total_flos":
|
| 1021 |
"trial_name": null,
|
| 1022 |
"trial_params": null
|
| 1023 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 3.0360162258148193,
|
| 3 |
+
"best_model_checkpoint": "dq158/coqui/checkpoint-121044",
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 121044,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 1011 |
"eval_steps_per_second": 0.543,
|
| 1012 |
"eval_translation_length": 4591104,
|
| 1013 |
"step": 80696
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 2.01,
|
| 1017 |
+
"learning_rate": 7.912941309744704e-05,
|
| 1018 |
+
"loss": 3.1234,
|
| 1019 |
+
"step": 81000
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 2.02,
|
| 1023 |
+
"learning_rate": 7.91186047830294e-05,
|
| 1024 |
+
"loss": 3.1118,
|
| 1025 |
+
"step": 81500
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 2.03,
|
| 1029 |
+
"learning_rate": 7.910773053674371e-05,
|
| 1030 |
+
"loss": 3.1879,
|
| 1031 |
+
"step": 82000
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 2.04,
|
| 1035 |
+
"learning_rate": 7.909679037691783e-05,
|
| 1036 |
+
"loss": 3.2032,
|
| 1037 |
+
"step": 82500
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 2.06,
|
| 1041 |
+
"learning_rate": 7.908578432199069e-05,
|
| 1042 |
+
"loss": 3.1009,
|
| 1043 |
+
"step": 83000
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 2.07,
|
| 1047 |
+
"learning_rate": 7.907471239051224e-05,
|
| 1048 |
+
"loss": 3.0688,
|
| 1049 |
+
"step": 83500
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 2.08,
|
| 1053 |
+
"learning_rate": 7.906357460114355e-05,
|
| 1054 |
+
"loss": 3.0835,
|
| 1055 |
+
"step": 84000
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 2.09,
|
| 1059 |
+
"learning_rate": 7.90523709726566e-05,
|
| 1060 |
+
"loss": 3.1157,
|
| 1061 |
+
"step": 84500
|
| 1062 |
+
},
|
| 1063 |
+
{
|
| 1064 |
+
"epoch": 2.11,
|
| 1065 |
+
"learning_rate": 7.90411015239344e-05,
|
| 1066 |
+
"loss": 3.0581,
|
| 1067 |
+
"step": 85000
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 2.12,
|
| 1071 |
+
"learning_rate": 7.902976627397088e-05,
|
| 1072 |
+
"loss": 3.206,
|
| 1073 |
+
"step": 85500
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 2.13,
|
| 1077 |
+
"learning_rate": 7.901836524187085e-05,
|
| 1078 |
+
"loss": 3.2185,
|
| 1079 |
+
"step": 86000
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 2.14,
|
| 1083 |
+
"learning_rate": 7.900689844685002e-05,
|
| 1084 |
+
"loss": 3.1204,
|
| 1085 |
+
"step": 86500
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 2.16,
|
| 1089 |
+
"learning_rate": 7.899536590823493e-05,
|
| 1090 |
+
"loss": 3.2066,
|
| 1091 |
+
"step": 87000
|
| 1092 |
+
},
|
| 1093 |
+
{
|
| 1094 |
+
"epoch": 2.17,
|
| 1095 |
+
"learning_rate": 7.898376764546293e-05,
|
| 1096 |
+
"loss": 3.1684,
|
| 1097 |
+
"step": 87500
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 2.18,
|
| 1101 |
+
"learning_rate": 7.89721036780821e-05,
|
| 1102 |
+
"loss": 3.1742,
|
| 1103 |
+
"step": 88000
|
| 1104 |
+
},
|
| 1105 |
+
{
|
| 1106 |
+
"epoch": 2.19,
|
| 1107 |
+
"learning_rate": 7.896037402575136e-05,
|
| 1108 |
+
"loss": 3.1598,
|
| 1109 |
+
"step": 88500
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 2.21,
|
| 1113 |
+
"learning_rate": 7.894857870824023e-05,
|
| 1114 |
+
"loss": 3.0966,
|
| 1115 |
+
"step": 89000
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 2.22,
|
| 1119 |
+
"learning_rate": 7.893671774542899e-05,
|
| 1120 |
+
"loss": 3.2035,
|
| 1121 |
+
"step": 89500
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 2.23,
|
| 1125 |
+
"learning_rate": 7.892479115730848e-05,
|
| 1126 |
+
"loss": 3.215,
|
| 1127 |
+
"step": 90000
|
| 1128 |
+
},
|
| 1129 |
+
{
|
| 1130 |
+
"epoch": 2.24,
|
| 1131 |
+
"learning_rate": 7.891279896398023e-05,
|
| 1132 |
+
"loss": 3.153,
|
| 1133 |
+
"step": 90500
|
| 1134 |
+
},
|
| 1135 |
+
{
|
| 1136 |
+
"epoch": 2.26,
|
| 1137 |
+
"learning_rate": 7.890074118565626e-05,
|
| 1138 |
+
"loss": 3.1019,
|
| 1139 |
+
"step": 91000
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 2.27,
|
| 1143 |
+
"learning_rate": 7.88886178426592e-05,
|
| 1144 |
+
"loss": 3.1509,
|
| 1145 |
+
"step": 91500
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 2.28,
|
| 1149 |
+
"learning_rate": 7.887642895542218e-05,
|
| 1150 |
+
"loss": 3.1699,
|
| 1151 |
+
"step": 92000
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 2.29,
|
| 1155 |
+
"learning_rate": 7.886417454448872e-05,
|
| 1156 |
+
"loss": 3.1162,
|
| 1157 |
+
"step": 92500
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 2.3,
|
| 1161 |
+
"learning_rate": 7.885185463051289e-05,
|
| 1162 |
+
"loss": 3.1782,
|
| 1163 |
+
"step": 93000
|
| 1164 |
+
},
|
| 1165 |
+
{
|
| 1166 |
+
"epoch": 2.32,
|
| 1167 |
+
"learning_rate": 7.883946923425907e-05,
|
| 1168 |
+
"loss": 3.1546,
|
| 1169 |
+
"step": 93500
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 2.33,
|
| 1173 |
+
"learning_rate": 7.882701837660205e-05,
|
| 1174 |
+
"loss": 3.1551,
|
| 1175 |
+
"step": 94000
|
| 1176 |
+
},
|
| 1177 |
+
{
|
| 1178 |
+
"epoch": 2.34,
|
| 1179 |
+
"learning_rate": 7.881450207852696e-05,
|
| 1180 |
+
"loss": 3.1327,
|
| 1181 |
+
"step": 94500
|
| 1182 |
+
},
|
| 1183 |
+
{
|
| 1184 |
+
"epoch": 2.35,
|
| 1185 |
+
"learning_rate": 7.880192036112917e-05,
|
| 1186 |
+
"loss": 3.0316,
|
| 1187 |
+
"step": 95000
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 2.37,
|
| 1191 |
+
"learning_rate": 7.878927324561437e-05,
|
| 1192 |
+
"loss": 3.1897,
|
| 1193 |
+
"step": 95500
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 2.38,
|
| 1197 |
+
"learning_rate": 7.877656075329846e-05,
|
| 1198 |
+
"loss": 3.1447,
|
| 1199 |
+
"step": 96000
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 2.39,
|
| 1203 |
+
"learning_rate": 7.876378290560751e-05,
|
| 1204 |
+
"loss": 3.1764,
|
| 1205 |
+
"step": 96500
|
| 1206 |
+
},
|
| 1207 |
+
{
|
| 1208 |
+
"epoch": 2.4,
|
| 1209 |
+
"learning_rate": 7.875093972407774e-05,
|
| 1210 |
+
"loss": 3.1213,
|
| 1211 |
+
"step": 97000
|
| 1212 |
+
},
|
| 1213 |
+
{
|
| 1214 |
+
"epoch": 2.42,
|
| 1215 |
+
"learning_rate": 7.873803123035553e-05,
|
| 1216 |
+
"loss": 3.0774,
|
| 1217 |
+
"step": 97500
|
| 1218 |
+
},
|
| 1219 |
+
{
|
| 1220 |
+
"epoch": 2.43,
|
| 1221 |
+
"learning_rate": 7.872505744619728e-05,
|
| 1222 |
+
"loss": 3.1127,
|
| 1223 |
+
"step": 98000
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 2.44,
|
| 1227 |
+
"learning_rate": 7.871201839346947e-05,
|
| 1228 |
+
"loss": 3.2621,
|
| 1229 |
+
"step": 98500
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 2.45,
|
| 1233 |
+
"learning_rate": 7.869891409414858e-05,
|
| 1234 |
+
"loss": 3.2113,
|
| 1235 |
+
"step": 99000
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 2.47,
|
| 1239 |
+
"learning_rate": 7.868574457032106e-05,
|
| 1240 |
+
"loss": 3.1218,
|
| 1241 |
+
"step": 99500
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 2.48,
|
| 1245 |
+
"learning_rate": 7.867250984418328e-05,
|
| 1246 |
+
"loss": 3.1409,
|
| 1247 |
+
"step": 100000
|
| 1248 |
+
},
|
| 1249 |
+
{
|
| 1250 |
+
"epoch": 2.49,
|
| 1251 |
+
"learning_rate": 7.86592099380415e-05,
|
| 1252 |
+
"loss": 3.073,
|
| 1253 |
+
"step": 100500
|
| 1254 |
+
},
|
| 1255 |
+
{
|
| 1256 |
+
"epoch": 2.5,
|
| 1257 |
+
"learning_rate": 7.864584487431186e-05,
|
| 1258 |
+
"loss": 3.1624,
|
| 1259 |
+
"step": 101000
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 2.52,
|
| 1263 |
+
"learning_rate": 7.863241467552032e-05,
|
| 1264 |
+
"loss": 3.2052,
|
| 1265 |
+
"step": 101500
|
| 1266 |
+
},
|
| 1267 |
+
{
|
| 1268 |
+
"epoch": 2.53,
|
| 1269 |
+
"learning_rate": 7.861891936430258e-05,
|
| 1270 |
+
"loss": 3.1714,
|
| 1271 |
+
"step": 102000
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 2.54,
|
| 1275 |
+
"learning_rate": 7.860535896340414e-05,
|
| 1276 |
+
"loss": 3.1728,
|
| 1277 |
+
"step": 102500
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 2.55,
|
| 1281 |
+
"learning_rate": 7.859173349568015e-05,
|
| 1282 |
+
"loss": 3.1564,
|
| 1283 |
+
"step": 103000
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 2.57,
|
| 1287 |
+
"learning_rate": 7.857804298409547e-05,
|
| 1288 |
+
"loss": 3.0868,
|
| 1289 |
+
"step": 103500
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 2.58,
|
| 1293 |
+
"learning_rate": 7.856428745172455e-05,
|
| 1294 |
+
"loss": 3.1485,
|
| 1295 |
+
"step": 104000
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 2.59,
|
| 1299 |
+
"learning_rate": 7.855046692175145e-05,
|
| 1300 |
+
"loss": 3.2001,
|
| 1301 |
+
"step": 104500
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 2.6,
|
| 1305 |
+
"learning_rate": 7.853658141746979e-05,
|
| 1306 |
+
"loss": 3.0651,
|
| 1307 |
+
"step": 105000
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 2.61,
|
| 1311 |
+
"learning_rate": 7.852263096228267e-05,
|
| 1312 |
+
"loss": 3.1709,
|
| 1313 |
+
"step": 105500
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 2.63,
|
| 1317 |
+
"learning_rate": 7.850861557970269e-05,
|
| 1318 |
+
"loss": 3.1635,
|
| 1319 |
+
"step": 106000
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 2.64,
|
| 1323 |
+
"learning_rate": 7.849453529335188e-05,
|
| 1324 |
+
"loss": 3.1123,
|
| 1325 |
+
"step": 106500
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 2.65,
|
| 1329 |
+
"learning_rate": 7.848039012696161e-05,
|
| 1330 |
+
"loss": 3.1162,
|
| 1331 |
+
"step": 107000
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 2.66,
|
| 1335 |
+
"learning_rate": 7.846618010437265e-05,
|
| 1336 |
+
"loss": 3.1275,
|
| 1337 |
+
"step": 107500
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 2.68,
|
| 1341 |
+
"learning_rate": 7.84519052495351e-05,
|
| 1342 |
+
"loss": 3.1035,
|
| 1343 |
+
"step": 108000
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 2.69,
|
| 1347 |
+
"learning_rate": 7.843756558650827e-05,
|
| 1348 |
+
"loss": 3.0308,
|
| 1349 |
+
"step": 108500
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 2.7,
|
| 1353 |
+
"learning_rate": 7.842316113946073e-05,
|
| 1354 |
+
"loss": 3.1311,
|
| 1355 |
+
"step": 109000
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 2.71,
|
| 1359 |
+
"learning_rate": 7.840869193267027e-05,
|
| 1360 |
+
"loss": 3.1477,
|
| 1361 |
+
"step": 109500
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 2.73,
|
| 1365 |
+
"learning_rate": 7.839415799052378e-05,
|
| 1366 |
+
"loss": 3.1221,
|
| 1367 |
+
"step": 110000
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 2.74,
|
| 1371 |
+
"learning_rate": 7.837955933751725e-05,
|
| 1372 |
+
"loss": 3.1322,
|
| 1373 |
+
"step": 110500
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 2.75,
|
| 1377 |
+
"learning_rate": 7.83648959982558e-05,
|
| 1378 |
+
"loss": 3.1488,
|
| 1379 |
+
"step": 111000
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 2.76,
|
| 1383 |
+
"learning_rate": 7.835016799745353e-05,
|
| 1384 |
+
"loss": 3.0619,
|
| 1385 |
+
"step": 111500
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 2.78,
|
| 1389 |
+
"learning_rate": 7.833537535993351e-05,
|
| 1390 |
+
"loss": 3.237,
|
| 1391 |
+
"step": 112000
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 2.79,
|
| 1395 |
+
"learning_rate": 7.83205181106278e-05,
|
| 1396 |
+
"loss": 3.1022,
|
| 1397 |
+
"step": 112500
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 2.8,
|
| 1401 |
+
"learning_rate": 7.83055962745773e-05,
|
| 1402 |
+
"loss": 3.066,
|
| 1403 |
+
"step": 113000
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 2.81,
|
| 1407 |
+
"learning_rate": 7.829060987693179e-05,
|
| 1408 |
+
"loss": 3.1903,
|
| 1409 |
+
"step": 113500
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 2.83,
|
| 1413 |
+
"learning_rate": 7.827555894294991e-05,
|
| 1414 |
+
"loss": 3.2208,
|
| 1415 |
+
"step": 114000
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 2.84,
|
| 1419 |
+
"learning_rate": 7.8260443497999e-05,
|
| 1420 |
+
"loss": 3.2005,
|
| 1421 |
+
"step": 114500
|
| 1422 |
+
},
|
| 1423 |
+
{
|
| 1424 |
+
"epoch": 2.85,
|
| 1425 |
+
"learning_rate": 7.824526356755516e-05,
|
| 1426 |
+
"loss": 3.184,
|
| 1427 |
+
"step": 115000
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 2.86,
|
| 1431 |
+
"learning_rate": 7.82300191772032e-05,
|
| 1432 |
+
"loss": 3.1185,
|
| 1433 |
+
"step": 115500
|
| 1434 |
+
},
|
| 1435 |
+
{
|
| 1436 |
+
"epoch": 2.87,
|
| 1437 |
+
"learning_rate": 7.821471035263653e-05,
|
| 1438 |
+
"loss": 3.1077,
|
| 1439 |
+
"step": 116000
|
| 1440 |
+
},
|
| 1441 |
+
{
|
| 1442 |
+
"epoch": 2.89,
|
| 1443 |
+
"learning_rate": 7.819933711965718e-05,
|
| 1444 |
+
"loss": 3.1901,
|
| 1445 |
+
"step": 116500
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 2.9,
|
| 1449 |
+
"learning_rate": 7.818389950417574e-05,
|
| 1450 |
+
"loss": 3.1149,
|
| 1451 |
+
"step": 117000
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 2.91,
|
| 1455 |
+
"learning_rate": 7.816839753221132e-05,
|
| 1456 |
+
"loss": 3.1393,
|
| 1457 |
+
"step": 117500
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 2.92,
|
| 1461 |
+
"learning_rate": 7.815283122989147e-05,
|
| 1462 |
+
"loss": 3.0862,
|
| 1463 |
+
"step": 118000
|
| 1464 |
+
},
|
| 1465 |
+
{
|
| 1466 |
+
"epoch": 2.94,
|
| 1467 |
+
"learning_rate": 7.813720062345219e-05,
|
| 1468 |
+
"loss": 3.072,
|
| 1469 |
+
"step": 118500
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 2.95,
|
| 1473 |
+
"learning_rate": 7.812150573923785e-05,
|
| 1474 |
+
"loss": 3.1103,
|
| 1475 |
+
"step": 119000
|
| 1476 |
+
},
|
| 1477 |
+
{
|
| 1478 |
+
"epoch": 2.96,
|
| 1479 |
+
"learning_rate": 7.810574660370116e-05,
|
| 1480 |
+
"loss": 3.2114,
|
| 1481 |
+
"step": 119500
|
| 1482 |
+
},
|
| 1483 |
+
{
|
| 1484 |
+
"epoch": 2.97,
|
| 1485 |
+
"learning_rate": 7.808992324340312e-05,
|
| 1486 |
+
"loss": 3.1425,
|
| 1487 |
+
"step": 120000
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 2.99,
|
| 1491 |
+
"learning_rate": 7.807403568501297e-05,
|
| 1492 |
+
"loss": 3.1051,
|
| 1493 |
+
"step": 120500
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 3.0,
|
| 1497 |
+
"learning_rate": 7.805808395530816e-05,
|
| 1498 |
+
"loss": 3.1355,
|
| 1499 |
+
"step": 121000
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 3.0,
|
| 1503 |
+
"eval_bleu": 1.0,
|
| 1504 |
+
"eval_brevity_penalty": 1.0,
|
| 1505 |
+
"eval_length_ratio": 1.0,
|
| 1506 |
+
"eval_loss": 3.0360162258148193,
|
| 1507 |
+
"eval_precisions": [
|
| 1508 |
+
1.0,
|
| 1509 |
+
1.0,
|
| 1510 |
+
1.0,
|
| 1511 |
+
1.0
|
| 1512 |
+
],
|
| 1513 |
+
"eval_reference_length": 4591104,
|
| 1514 |
+
"eval_runtime": 8117.2024,
|
| 1515 |
+
"eval_samples_per_second": 1.105,
|
| 1516 |
+
"eval_steps_per_second": 0.552,
|
| 1517 |
+
"eval_translation_length": 4591104,
|
| 1518 |
+
"step": 121044
|
| 1519 |
}
|
| 1520 |
],
|
| 1521 |
"logging_steps": 500,
|
| 1522 |
"max_steps": 1210440,
|
| 1523 |
"num_train_epochs": 30,
|
| 1524 |
"save_steps": 1000,
|
| 1525 |
+
"total_flos": 2.077405682985861e+18,
|
| 1526 |
"trial_name": null,
|
| 1527 |
"trial_params": null
|
| 1528 |
}
|