Upload logs/history.json with huggingface_hub
Browse files- logs/history.json +260 -0
logs/history.json
CHANGED
|
@@ -1038,5 +1038,265 @@
|
|
| 1038 |
],
|
| 1039 |
"grad_norm": 0.195,
|
| 1040 |
"sec": 76.3
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1041 |
}
|
| 1042 |
]
|
|
|
|
| 1038 |
],
|
| 1039 |
"grad_norm": 0.195,
|
| 1040 |
"sec": 76.3
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 41,
|
| 1044 |
+
"beta": 0.0003,
|
| 1045 |
+
"lr": 2.43e-05,
|
| 1046 |
+
"train_loss": 0.040033,
|
| 1047 |
+
"train_recon": 0.032655,
|
| 1048 |
+
"train_kl": 24.5927,
|
| 1049 |
+
"val_loss": 0.038497,
|
| 1050 |
+
"val_recon": 0.031199,
|
| 1051 |
+
"val_kl": 24.3254,
|
| 1052 |
+
"active_dims": 12,
|
| 1053 |
+
"active_per_slot": [
|
| 1054 |
+
4,
|
| 1055 |
+
4,
|
| 1056 |
+
4
|
| 1057 |
+
],
|
| 1058 |
+
"usage_max": 0.3343,
|
| 1059 |
+
"usage_min": 0.3318,
|
| 1060 |
+
"usage": [
|
| 1061 |
+
0.3318,
|
| 1062 |
+
0.3339,
|
| 1063 |
+
0.3343
|
| 1064 |
+
],
|
| 1065 |
+
"grad_norm": 0.164,
|
| 1066 |
+
"sec": 76.3
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 42,
|
| 1070 |
+
"beta": 0.0003,
|
| 1071 |
+
"lr": 1.95e-05,
|
| 1072 |
+
"train_loss": 0.039939,
|
| 1073 |
+
"train_recon": 0.032578,
|
| 1074 |
+
"train_kl": 24.5383,
|
| 1075 |
+
"val_loss": 0.038448,
|
| 1076 |
+
"val_recon": 0.031064,
|
| 1077 |
+
"val_kl": 24.6102,
|
| 1078 |
+
"active_dims": 12,
|
| 1079 |
+
"active_per_slot": [
|
| 1080 |
+
4,
|
| 1081 |
+
4,
|
| 1082 |
+
4
|
| 1083 |
+
],
|
| 1084 |
+
"usage_max": 0.3378,
|
| 1085 |
+
"usage_min": 0.3303,
|
| 1086 |
+
"usage": [
|
| 1087 |
+
0.3378,
|
| 1088 |
+
0.3303,
|
| 1089 |
+
0.3318
|
| 1090 |
+
],
|
| 1091 |
+
"grad_norm": 0.148,
|
| 1092 |
+
"sec": 76.2
|
| 1093 |
+
},
|
| 1094 |
+
{
|
| 1095 |
+
"epoch": 43,
|
| 1096 |
+
"beta": 0.0003,
|
| 1097 |
+
"lr": 1.52e-05,
|
| 1098 |
+
"train_loss": 0.039851,
|
| 1099 |
+
"train_recon": 0.032464,
|
| 1100 |
+
"train_kl": 24.6233,
|
| 1101 |
+
"val_loss": 0.038465,
|
| 1102 |
+
"val_recon": 0.031036,
|
| 1103 |
+
"val_kl": 24.7623,
|
| 1104 |
+
"active_dims": 12,
|
| 1105 |
+
"active_per_slot": [
|
| 1106 |
+
4,
|
| 1107 |
+
4,
|
| 1108 |
+
4
|
| 1109 |
+
],
|
| 1110 |
+
"usage_max": 0.3358,
|
| 1111 |
+
"usage_min": 0.33,
|
| 1112 |
+
"usage": [
|
| 1113 |
+
0.33,
|
| 1114 |
+
0.3342,
|
| 1115 |
+
0.3358
|
| 1116 |
+
],
|
| 1117 |
+
"grad_norm": 0.158,
|
| 1118 |
+
"sec": 76.4
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 44,
|
| 1122 |
+
"beta": 0.0003,
|
| 1123 |
+
"lr": 1.15e-05,
|
| 1124 |
+
"train_loss": 0.039883,
|
| 1125 |
+
"train_recon": 0.032494,
|
| 1126 |
+
"train_kl": 24.627,
|
| 1127 |
+
"val_loss": 0.038375,
|
| 1128 |
+
"val_recon": 0.030965,
|
| 1129 |
+
"val_kl": 24.7015,
|
| 1130 |
+
"active_dims": 12,
|
| 1131 |
+
"active_per_slot": [
|
| 1132 |
+
4,
|
| 1133 |
+
4,
|
| 1134 |
+
4
|
| 1135 |
+
],
|
| 1136 |
+
"usage_max": 0.3373,
|
| 1137 |
+
"usage_min": 0.3302,
|
| 1138 |
+
"usage": [
|
| 1139 |
+
0.3325,
|
| 1140 |
+
0.3302,
|
| 1141 |
+
0.3373
|
| 1142 |
+
],
|
| 1143 |
+
"grad_norm": 0.194,
|
| 1144 |
+
"sec": 76.3
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 45,
|
| 1148 |
+
"beta": 0.0003,
|
| 1149 |
+
"lr": 8.3e-06,
|
| 1150 |
+
"train_loss": 0.039779,
|
| 1151 |
+
"train_recon": 0.032389,
|
| 1152 |
+
"train_kl": 24.6331,
|
| 1153 |
+
"val_loss": 0.038241,
|
| 1154 |
+
"val_recon": 0.030881,
|
| 1155 |
+
"val_kl": 24.5332,
|
| 1156 |
+
"active_dims": 12,
|
| 1157 |
+
"active_per_slot": [
|
| 1158 |
+
4,
|
| 1159 |
+
4,
|
| 1160 |
+
4
|
| 1161 |
+
],
|
| 1162 |
+
"usage_max": 0.3363,
|
| 1163 |
+
"usage_min": 0.3313,
|
| 1164 |
+
"usage": [
|
| 1165 |
+
0.3363,
|
| 1166 |
+
0.3313,
|
| 1167 |
+
0.3324
|
| 1168 |
+
],
|
| 1169 |
+
"grad_norm": 0.129,
|
| 1170 |
+
"sec": 76.4
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"epoch": 46,
|
| 1174 |
+
"beta": 0.0003,
|
| 1175 |
+
"lr": 5.7e-06,
|
| 1176 |
+
"train_loss": 0.039808,
|
| 1177 |
+
"train_recon": 0.0324,
|
| 1178 |
+
"train_kl": 24.6945,
|
| 1179 |
+
"val_loss": 0.038271,
|
| 1180 |
+
"val_recon": 0.030882,
|
| 1181 |
+
"val_kl": 24.6279,
|
| 1182 |
+
"active_dims": 12,
|
| 1183 |
+
"active_per_slot": [
|
| 1184 |
+
4,
|
| 1185 |
+
4,
|
| 1186 |
+
4
|
| 1187 |
+
],
|
| 1188 |
+
"usage_max": 0.3353,
|
| 1189 |
+
"usage_min": 0.3317,
|
| 1190 |
+
"usage": [
|
| 1191 |
+
0.333,
|
| 1192 |
+
0.3317,
|
| 1193 |
+
0.3353
|
| 1194 |
+
],
|
| 1195 |
+
"grad_norm": 0.287,
|
| 1196 |
+
"sec": 76.2
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 47,
|
| 1200 |
+
"beta": 0.0003,
|
| 1201 |
+
"lr": 3.6e-06,
|
| 1202 |
+
"train_loss": 0.039715,
|
| 1203 |
+
"train_recon": 0.032329,
|
| 1204 |
+
"train_kl": 24.6202,
|
| 1205 |
+
"val_loss": 0.038265,
|
| 1206 |
+
"val_recon": 0.030881,
|
| 1207 |
+
"val_kl": 24.6125,
|
| 1208 |
+
"active_dims": 12,
|
| 1209 |
+
"active_per_slot": [
|
| 1210 |
+
4,
|
| 1211 |
+
4,
|
| 1212 |
+
4
|
| 1213 |
+
],
|
| 1214 |
+
"usage_max": 0.3355,
|
| 1215 |
+
"usage_min": 0.3317,
|
| 1216 |
+
"usage": [
|
| 1217 |
+
0.3328,
|
| 1218 |
+
0.3355,
|
| 1219 |
+
0.3317
|
| 1220 |
+
],
|
| 1221 |
+
"grad_norm": 0.125,
|
| 1222 |
+
"sec": 76.3
|
| 1223 |
+
},
|
| 1224 |
+
{
|
| 1225 |
+
"epoch": 48,
|
| 1226 |
+
"beta": 0.0003,
|
| 1227 |
+
"lr": 2.2e-06,
|
| 1228 |
+
"train_loss": 0.039696,
|
| 1229 |
+
"train_recon": 0.032302,
|
| 1230 |
+
"train_kl": 24.6479,
|
| 1231 |
+
"val_loss": 0.03823,
|
| 1232 |
+
"val_recon": 0.030834,
|
| 1233 |
+
"val_kl": 24.6528,
|
| 1234 |
+
"active_dims": 12,
|
| 1235 |
+
"active_per_slot": [
|
| 1236 |
+
4,
|
| 1237 |
+
4,
|
| 1238 |
+
4
|
| 1239 |
+
],
|
| 1240 |
+
"usage_max": 0.3369,
|
| 1241 |
+
"usage_min": 0.331,
|
| 1242 |
+
"usage": [
|
| 1243 |
+
0.3369,
|
| 1244 |
+
0.331,
|
| 1245 |
+
0.3322
|
| 1246 |
+
],
|
| 1247 |
+
"grad_norm": 0.126,
|
| 1248 |
+
"sec": 76.3
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 49,
|
| 1252 |
+
"beta": 0.0003,
|
| 1253 |
+
"lr": 1.3e-06,
|
| 1254 |
+
"train_loss": 0.039684,
|
| 1255 |
+
"train_recon": 0.032283,
|
| 1256 |
+
"train_kl": 24.6717,
|
| 1257 |
+
"val_loss": 0.038225,
|
| 1258 |
+
"val_recon": 0.030839,
|
| 1259 |
+
"val_kl": 24.6194,
|
| 1260 |
+
"active_dims": 12,
|
| 1261 |
+
"active_per_slot": [
|
| 1262 |
+
4,
|
| 1263 |
+
4,
|
| 1264 |
+
4
|
| 1265 |
+
],
|
| 1266 |
+
"usage_max": 0.3378,
|
| 1267 |
+
"usage_min": 0.3302,
|
| 1268 |
+
"usage": [
|
| 1269 |
+
0.3378,
|
| 1270 |
+
0.3302,
|
| 1271 |
+
0.332
|
| 1272 |
+
],
|
| 1273 |
+
"grad_norm": 0.114,
|
| 1274 |
+
"sec": 76.4
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 50,
|
| 1278 |
+
"beta": 0.0003,
|
| 1279 |
+
"lr": 1e-06,
|
| 1280 |
+
"train_loss": 0.039684,
|
| 1281 |
+
"train_recon": 0.032295,
|
| 1282 |
+
"train_kl": 24.6304,
|
| 1283 |
+
"val_loss": 0.038194,
|
| 1284 |
+
"val_recon": 0.030802,
|
| 1285 |
+
"val_kl": 24.6384,
|
| 1286 |
+
"active_dims": 12,
|
| 1287 |
+
"active_per_slot": [
|
| 1288 |
+
4,
|
| 1289 |
+
4,
|
| 1290 |
+
4
|
| 1291 |
+
],
|
| 1292 |
+
"usage_max": 0.3344,
|
| 1293 |
+
"usage_min": 0.3327,
|
| 1294 |
+
"usage": [
|
| 1295 |
+
0.3344,
|
| 1296 |
+
0.3328,
|
| 1297 |
+
0.3327
|
| 1298 |
+
],
|
| 1299 |
+
"grad_norm": 0.145,
|
| 1300 |
+
"sec": 76.3
|
| 1301 |
}
|
| 1302 |
]
|