Training in progress, step 5000, checkpoint
Browse files- last-checkpoint/global_step5000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt +2 -2
- last-checkpoint/global_step5000/mp_rank_00_model_states.pt +2 -2
- last-checkpoint/latest +1 -1
- last-checkpoint/model.safetensors +1 -1
- last-checkpoint/rng_state.pth +1 -1
- last-checkpoint/scheduler.pt +1 -1
- last-checkpoint/trainer_state.json +295 -5
last-checkpoint/global_step5000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dadec7b95e9c63b7f4e5b0497d74ff5f09c5a2d9cf367942e470953f3f7cc7f1
|
| 3 |
+
size 5117197489
|
last-checkpoint/global_step5000/mp_rank_00_model_states.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:757e473d63fc9cb2a005c0657327d432b5292ea2e001b1ac5f2a00023815e9a9
|
| 3 |
+
size 859127933
|
last-checkpoint/latest
CHANGED
|
@@ -1 +1 @@
|
|
| 1 |
-
|
|
|
|
| 1 |
+
global_step5000
|
last-checkpoint/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 962205216
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e917f6578a37f477ce51d824ef2c22355d57ba680883e19ec30d3b97940c7e3b
|
| 3 |
size 962205216
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14709
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fe58c8283b537c6ee9a4dd56ebbea21d90b446075eea802c036a3707078dd25c
|
| 3 |
size 14709
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1465
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4fdaaa58d591c2d03b0ec95bb2576cb7c7885945b5e85c3aae63ede0ea16cfc5
|
| 3 |
size 1465
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 86.
|
| 3 |
-
"best_model_checkpoint": "./iteboshi_student_model_temp/checkpoint-
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 1000,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -1167,6 +1167,296 @@
|
|
| 1167 |
"eval_steps_per_second": 2.028,
|
| 1168 |
"eval_wer": 86.4875058934465,
|
| 1169 |
"step": 4000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1170 |
}
|
| 1171 |
],
|
| 1172 |
"logging_steps": 25,
|
|
@@ -1186,7 +1476,7 @@
|
|
| 1186 |
"attributes": {}
|
| 1187 |
}
|
| 1188 |
},
|
| 1189 |
-
"total_flos":
|
| 1190 |
"train_batch_size": 4,
|
| 1191 |
"trial_name": null,
|
| 1192 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 86.11975483262611,
|
| 3 |
+
"best_model_checkpoint": "./iteboshi_student_model_temp/checkpoint-5000",
|
| 4 |
+
"epoch": 5.506607929515418,
|
| 5 |
"eval_steps": 1000,
|
| 6 |
+
"global_step": 5000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 1167 |
"eval_steps_per_second": 2.028,
|
| 1168 |
"eval_wer": 86.4875058934465,
|
| 1169 |
"step": 4000
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 4.432819383259912,
|
| 1173 |
+
"grad_norm": 0.6854081153869629,
|
| 1174 |
+
"learning_rate": 1.6384615384615384e-05,
|
| 1175 |
+
"loss": 0.2609,
|
| 1176 |
+
"step": 4025
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"epoch": 4.460352422907489,
|
| 1180 |
+
"grad_norm": 0.8021137714385986,
|
| 1181 |
+
"learning_rate": 1.635897435897436e-05,
|
| 1182 |
+
"loss": 0.2604,
|
| 1183 |
+
"step": 4050
|
| 1184 |
+
},
|
| 1185 |
+
{
|
| 1186 |
+
"epoch": 4.487885462555066,
|
| 1187 |
+
"grad_norm": 0.9230350255966187,
|
| 1188 |
+
"learning_rate": 1.6333333333333335e-05,
|
| 1189 |
+
"loss": 0.2594,
|
| 1190 |
+
"step": 4075
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 4.515418502202643,
|
| 1194 |
+
"grad_norm": 0.8251164555549622,
|
| 1195 |
+
"learning_rate": 1.630769230769231e-05,
|
| 1196 |
+
"loss": 0.2191,
|
| 1197 |
+
"step": 4100
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 4.54295154185022,
|
| 1201 |
+
"grad_norm": 0.7396982312202454,
|
| 1202 |
+
"learning_rate": 1.6282051282051282e-05,
|
| 1203 |
+
"loss": 0.2508,
|
| 1204 |
+
"step": 4125
|
| 1205 |
+
},
|
| 1206 |
+
{
|
| 1207 |
+
"epoch": 4.570484581497797,
|
| 1208 |
+
"grad_norm": 0.7925761938095093,
|
| 1209 |
+
"learning_rate": 1.625641025641026e-05,
|
| 1210 |
+
"loss": 0.2603,
|
| 1211 |
+
"step": 4150
|
| 1212 |
+
},
|
| 1213 |
+
{
|
| 1214 |
+
"epoch": 4.598017621145375,
|
| 1215 |
+
"grad_norm": 0.6864134669303894,
|
| 1216 |
+
"learning_rate": 1.6230769230769233e-05,
|
| 1217 |
+
"loss": 0.249,
|
| 1218 |
+
"step": 4175
|
| 1219 |
+
},
|
| 1220 |
+
{
|
| 1221 |
+
"epoch": 4.6255506607929515,
|
| 1222 |
+
"grad_norm": 0.5276267528533936,
|
| 1223 |
+
"learning_rate": 1.6205128205128207e-05,
|
| 1224 |
+
"loss": 0.2445,
|
| 1225 |
+
"step": 4200
|
| 1226 |
+
},
|
| 1227 |
+
{
|
| 1228 |
+
"epoch": 4.653083700440528,
|
| 1229 |
+
"grad_norm": 1.1504285335540771,
|
| 1230 |
+
"learning_rate": 1.617948717948718e-05,
|
| 1231 |
+
"loss": 0.2383,
|
| 1232 |
+
"step": 4225
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 4.680616740088106,
|
| 1236 |
+
"grad_norm": 0.7452952861785889,
|
| 1237 |
+
"learning_rate": 1.6153846153846154e-05,
|
| 1238 |
+
"loss": 0.2319,
|
| 1239 |
+
"step": 4250
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 4.708149779735683,
|
| 1243 |
+
"grad_norm": 0.5664868950843811,
|
| 1244 |
+
"learning_rate": 1.612820512820513e-05,
|
| 1245 |
+
"loss": 0.2386,
|
| 1246 |
+
"step": 4275
|
| 1247 |
+
},
|
| 1248 |
+
{
|
| 1249 |
+
"epoch": 4.73568281938326,
|
| 1250 |
+
"grad_norm": 0.7903388738632202,
|
| 1251 |
+
"learning_rate": 1.6102564102564105e-05,
|
| 1252 |
+
"loss": 0.2476,
|
| 1253 |
+
"step": 4300
|
| 1254 |
+
},
|
| 1255 |
+
{
|
| 1256 |
+
"epoch": 4.763215859030837,
|
| 1257 |
+
"grad_norm": 0.6549268364906311,
|
| 1258 |
+
"learning_rate": 1.607692307692308e-05,
|
| 1259 |
+
"loss": 0.2392,
|
| 1260 |
+
"step": 4325
|
| 1261 |
+
},
|
| 1262 |
+
{
|
| 1263 |
+
"epoch": 4.790748898678414,
|
| 1264 |
+
"grad_norm": 0.6780884265899658,
|
| 1265 |
+
"learning_rate": 1.6051282051282052e-05,
|
| 1266 |
+
"loss": 0.2354,
|
| 1267 |
+
"step": 4350
|
| 1268 |
+
},
|
| 1269 |
+
{
|
| 1270 |
+
"epoch": 4.818281938325991,
|
| 1271 |
+
"grad_norm": 0.9416743516921997,
|
| 1272 |
+
"learning_rate": 1.602564102564103e-05,
|
| 1273 |
+
"loss": 0.2514,
|
| 1274 |
+
"step": 4375
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 4.845814977973569,
|
| 1278 |
+
"grad_norm": 0.5058385133743286,
|
| 1279 |
+
"learning_rate": 1.6000000000000003e-05,
|
| 1280 |
+
"loss": 0.2173,
|
| 1281 |
+
"step": 4400
|
| 1282 |
+
},
|
| 1283 |
+
{
|
| 1284 |
+
"epoch": 4.8733480176211454,
|
| 1285 |
+
"grad_norm": 0.6523875594139099,
|
| 1286 |
+
"learning_rate": 1.5974358974358976e-05,
|
| 1287 |
+
"loss": 0.2323,
|
| 1288 |
+
"step": 4425
|
| 1289 |
+
},
|
| 1290 |
+
{
|
| 1291 |
+
"epoch": 4.900881057268722,
|
| 1292 |
+
"grad_norm": 0.6458995342254639,
|
| 1293 |
+
"learning_rate": 1.594871794871795e-05,
|
| 1294 |
+
"loss": 0.242,
|
| 1295 |
+
"step": 4450
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 4.9284140969163,
|
| 1299 |
+
"grad_norm": 0.5865331292152405,
|
| 1300 |
+
"learning_rate": 1.5923076923076924e-05,
|
| 1301 |
+
"loss": 0.2181,
|
| 1302 |
+
"step": 4475
|
| 1303 |
+
},
|
| 1304 |
+
{
|
| 1305 |
+
"epoch": 4.955947136563877,
|
| 1306 |
+
"grad_norm": 0.5348775386810303,
|
| 1307 |
+
"learning_rate": 1.5897435897435897e-05,
|
| 1308 |
+
"loss": 0.2379,
|
| 1309 |
+
"step": 4500
|
| 1310 |
+
},
|
| 1311 |
+
{
|
| 1312 |
+
"epoch": 4.983480176211454,
|
| 1313 |
+
"grad_norm": 0.6599372029304504,
|
| 1314 |
+
"learning_rate": 1.587179487179487e-05,
|
| 1315 |
+
"loss": 0.2482,
|
| 1316 |
+
"step": 4525
|
| 1317 |
+
},
|
| 1318 |
+
{
|
| 1319 |
+
"epoch": 5.011013215859031,
|
| 1320 |
+
"grad_norm": 0.625035285949707,
|
| 1321 |
+
"learning_rate": 1.5846153846153848e-05,
|
| 1322 |
+
"loss": 0.1922,
|
| 1323 |
+
"step": 4550
|
| 1324 |
+
},
|
| 1325 |
+
{
|
| 1326 |
+
"epoch": 5.038546255506608,
|
| 1327 |
+
"grad_norm": 0.6018031239509583,
|
| 1328 |
+
"learning_rate": 1.582051282051282e-05,
|
| 1329 |
+
"loss": 0.1687,
|
| 1330 |
+
"step": 4575
|
| 1331 |
+
},
|
| 1332 |
+
{
|
| 1333 |
+
"epoch": 5.066079295154185,
|
| 1334 |
+
"grad_norm": 0.6142588257789612,
|
| 1335 |
+
"learning_rate": 1.5794871794871795e-05,
|
| 1336 |
+
"loss": 0.1399,
|
| 1337 |
+
"step": 4600
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 5.093612334801762,
|
| 1341 |
+
"grad_norm": 0.445803701877594,
|
| 1342 |
+
"learning_rate": 1.576923076923077e-05,
|
| 1343 |
+
"loss": 0.1796,
|
| 1344 |
+
"step": 4625
|
| 1345 |
+
},
|
| 1346 |
+
{
|
| 1347 |
+
"epoch": 5.121145374449339,
|
| 1348 |
+
"grad_norm": 0.4978330135345459,
|
| 1349 |
+
"learning_rate": 1.5743589743589746e-05,
|
| 1350 |
+
"loss": 0.1716,
|
| 1351 |
+
"step": 4650
|
| 1352 |
+
},
|
| 1353 |
+
{
|
| 1354 |
+
"epoch": 5.148678414096916,
|
| 1355 |
+
"grad_norm": 0.43810775876045227,
|
| 1356 |
+
"learning_rate": 1.571794871794872e-05,
|
| 1357 |
+
"loss": 0.176,
|
| 1358 |
+
"step": 4675
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 5.176211453744493,
|
| 1362 |
+
"grad_norm": 0.5028232336044312,
|
| 1363 |
+
"learning_rate": 1.5692307692307693e-05,
|
| 1364 |
+
"loss": 0.1435,
|
| 1365 |
+
"step": 4700
|
| 1366 |
+
},
|
| 1367 |
+
{
|
| 1368 |
+
"epoch": 5.203744493392071,
|
| 1369 |
+
"grad_norm": 0.3771626949310303,
|
| 1370 |
+
"learning_rate": 1.5666666666666667e-05,
|
| 1371 |
+
"loss": 0.1577,
|
| 1372 |
+
"step": 4725
|
| 1373 |
+
},
|
| 1374 |
+
{
|
| 1375 |
+
"epoch": 5.2312775330396475,
|
| 1376 |
+
"grad_norm": 0.823710024356842,
|
| 1377 |
+
"learning_rate": 1.5641025641025644e-05,
|
| 1378 |
+
"loss": 0.1346,
|
| 1379 |
+
"step": 4750
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 5.258810572687224,
|
| 1383 |
+
"grad_norm": 0.4752519428730011,
|
| 1384 |
+
"learning_rate": 1.5615384615384618e-05,
|
| 1385 |
+
"loss": 0.1996,
|
| 1386 |
+
"step": 4775
|
| 1387 |
+
},
|
| 1388 |
+
{
|
| 1389 |
+
"epoch": 5.286343612334802,
|
| 1390 |
+
"grad_norm": 0.3718922436237335,
|
| 1391 |
+
"learning_rate": 1.558974358974359e-05,
|
| 1392 |
+
"loss": 0.1477,
|
| 1393 |
+
"step": 4800
|
| 1394 |
+
},
|
| 1395 |
+
{
|
| 1396 |
+
"epoch": 5.313876651982379,
|
| 1397 |
+
"grad_norm": 0.37068554759025574,
|
| 1398 |
+
"learning_rate": 1.5564102564102565e-05,
|
| 1399 |
+
"loss": 0.1384,
|
| 1400 |
+
"step": 4825
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 5.341409691629956,
|
| 1404 |
+
"grad_norm": 0.4219229221343994,
|
| 1405 |
+
"learning_rate": 1.553846153846154e-05,
|
| 1406 |
+
"loss": 0.1534,
|
| 1407 |
+
"step": 4850
|
| 1408 |
+
},
|
| 1409 |
+
{
|
| 1410 |
+
"epoch": 5.368942731277533,
|
| 1411 |
+
"grad_norm": 0.6927037835121155,
|
| 1412 |
+
"learning_rate": 1.5512820512820516e-05,
|
| 1413 |
+
"loss": 0.1623,
|
| 1414 |
+
"step": 4875
|
| 1415 |
+
},
|
| 1416 |
+
{
|
| 1417 |
+
"epoch": 5.39647577092511,
|
| 1418 |
+
"grad_norm": 0.6902387142181396,
|
| 1419 |
+
"learning_rate": 1.548717948717949e-05,
|
| 1420 |
+
"loss": 0.1451,
|
| 1421 |
+
"step": 4900
|
| 1422 |
+
},
|
| 1423 |
+
{
|
| 1424 |
+
"epoch": 5.424008810572687,
|
| 1425 |
+
"grad_norm": 0.5900410413742065,
|
| 1426 |
+
"learning_rate": 1.5461538461538463e-05,
|
| 1427 |
+
"loss": 0.1971,
|
| 1428 |
+
"step": 4925
|
| 1429 |
+
},
|
| 1430 |
+
{
|
| 1431 |
+
"epoch": 5.451541850220265,
|
| 1432 |
+
"grad_norm": 0.5337275266647339,
|
| 1433 |
+
"learning_rate": 1.5435897435897436e-05,
|
| 1434 |
+
"loss": 0.1688,
|
| 1435 |
+
"step": 4950
|
| 1436 |
+
},
|
| 1437 |
+
{
|
| 1438 |
+
"epoch": 5.479074889867841,
|
| 1439 |
+
"grad_norm": 0.5641874074935913,
|
| 1440 |
+
"learning_rate": 1.5410256410256414e-05,
|
| 1441 |
+
"loss": 0.1567,
|
| 1442 |
+
"step": 4975
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 5.506607929515418,
|
| 1446 |
+
"grad_norm": 0.5372440218925476,
|
| 1447 |
+
"learning_rate": 1.5384615384615387e-05,
|
| 1448 |
+
"loss": 0.1609,
|
| 1449 |
+
"step": 5000
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 5.506607929515418,
|
| 1453 |
+
"eval_cer": 28.45941329542719,
|
| 1454 |
+
"eval_loss": 0.7670999765396118,
|
| 1455 |
+
"eval_runtime": 1323.4879,
|
| 1456 |
+
"eval_samples_per_second": 7.995,
|
| 1457 |
+
"eval_steps_per_second": 1.999,
|
| 1458 |
+
"eval_wer": 86.11975483262611,
|
| 1459 |
+
"step": 5000
|
| 1460 |
}
|
| 1461 |
],
|
| 1462 |
"logging_steps": 25,
|
|
|
|
| 1476 |
"attributes": {}
|
| 1477 |
}
|
| 1478 |
},
|
| 1479 |
+
"total_flos": 8.590715932450488e+19,
|
| 1480 |
"train_batch_size": 4,
|
| 1481 |
"trial_name": null,
|
| 1482 |
"trial_params": null
|