Upload logs/history.json with huggingface_hub
Browse files- logs/history.json +330 -0
logs/history.json
CHANGED
|
@@ -1318,5 +1318,335 @@
|
|
| 1318 |
"active_slots": 6,
|
| 1319 |
"grad_norm": 0.07,
|
| 1320 |
"sec": 138.1
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1321 |
}
|
| 1322 |
]
|
|
|
|
| 1318 |
"active_slots": 6,
|
| 1319 |
"grad_norm": 0.07,
|
| 1320 |
"sec": 138.1
|
| 1321 |
+
},
|
| 1322 |
+
{
|
| 1323 |
+
"epoch": 41,
|
| 1324 |
+
"beta": 0.0003,
|
| 1325 |
+
"lr": 2.43e-05,
|
| 1326 |
+
"train_loss": 0.045967,
|
| 1327 |
+
"train_recon": 0.036589,
|
| 1328 |
+
"train_kl": 31.2603,
|
| 1329 |
+
"val_loss": 0.043983,
|
| 1330 |
+
"val_recon": 0.034637,
|
| 1331 |
+
"val_kl": 31.1529,
|
| 1332 |
+
"active_dims": 60,
|
| 1333 |
+
"active_per_slot": [
|
| 1334 |
+
10,
|
| 1335 |
+
10,
|
| 1336 |
+
10,
|
| 1337 |
+
10,
|
| 1338 |
+
10,
|
| 1339 |
+
10
|
| 1340 |
+
],
|
| 1341 |
+
"usage_max": 0.1667,
|
| 1342 |
+
"usage_min": 0.1667,
|
| 1343 |
+
"usage": [
|
| 1344 |
+
0.1667,
|
| 1345 |
+
0.1667,
|
| 1346 |
+
0.1667,
|
| 1347 |
+
0.1667,
|
| 1348 |
+
0.1667,
|
| 1349 |
+
0.1667
|
| 1350 |
+
],
|
| 1351 |
+
"active_slots": 6,
|
| 1352 |
+
"grad_norm": 0.066,
|
| 1353 |
+
"sec": 137.4
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 42,
|
| 1357 |
+
"beta": 0.0003,
|
| 1358 |
+
"lr": 1.95e-05,
|
| 1359 |
+
"train_loss": 0.045923,
|
| 1360 |
+
"train_recon": 0.036516,
|
| 1361 |
+
"train_kl": 31.3569,
|
| 1362 |
+
"val_loss": 0.043882,
|
| 1363 |
+
"val_recon": 0.034463,
|
| 1364 |
+
"val_kl": 31.3971,
|
| 1365 |
+
"active_dims": 60,
|
| 1366 |
+
"active_per_slot": [
|
| 1367 |
+
10,
|
| 1368 |
+
10,
|
| 1369 |
+
10,
|
| 1370 |
+
10,
|
| 1371 |
+
10,
|
| 1372 |
+
10
|
| 1373 |
+
],
|
| 1374 |
+
"usage_max": 0.1667,
|
| 1375 |
+
"usage_min": 0.1667,
|
| 1376 |
+
"usage": [
|
| 1377 |
+
0.1667,
|
| 1378 |
+
0.1667,
|
| 1379 |
+
0.1667,
|
| 1380 |
+
0.1667,
|
| 1381 |
+
0.1667,
|
| 1382 |
+
0.1667
|
| 1383 |
+
],
|
| 1384 |
+
"active_slots": 6,
|
| 1385 |
+
"grad_norm": 0.064,
|
| 1386 |
+
"sec": 137.4
|
| 1387 |
+
},
|
| 1388 |
+
{
|
| 1389 |
+
"epoch": 43,
|
| 1390 |
+
"beta": 0.0003,
|
| 1391 |
+
"lr": 1.52e-05,
|
| 1392 |
+
"train_loss": 0.045843,
|
| 1393 |
+
"train_recon": 0.036448,
|
| 1394 |
+
"train_kl": 31.319,
|
| 1395 |
+
"val_loss": 0.043677,
|
| 1396 |
+
"val_recon": 0.03434,
|
| 1397 |
+
"val_kl": 31.1244,
|
| 1398 |
+
"active_dims": 60,
|
| 1399 |
+
"active_per_slot": [
|
| 1400 |
+
10,
|
| 1401 |
+
10,
|
| 1402 |
+
10,
|
| 1403 |
+
10,
|
| 1404 |
+
10,
|
| 1405 |
+
10
|
| 1406 |
+
],
|
| 1407 |
+
"usage_max": 0.1667,
|
| 1408 |
+
"usage_min": 0.1667,
|
| 1409 |
+
"usage": [
|
| 1410 |
+
0.1667,
|
| 1411 |
+
0.1667,
|
| 1412 |
+
0.1667,
|
| 1413 |
+
0.1667,
|
| 1414 |
+
0.1667,
|
| 1415 |
+
0.1667
|
| 1416 |
+
],
|
| 1417 |
+
"active_slots": 6,
|
| 1418 |
+
"grad_norm": 0.065,
|
| 1419 |
+
"sec": 137.2
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 44,
|
| 1423 |
+
"beta": 0.0003,
|
| 1424 |
+
"lr": 1.15e-05,
|
| 1425 |
+
"train_loss": 0.045838,
|
| 1426 |
+
"train_recon": 0.036422,
|
| 1427 |
+
"train_kl": 31.388,
|
| 1428 |
+
"val_loss": 0.043778,
|
| 1429 |
+
"val_recon": 0.034326,
|
| 1430 |
+
"val_kl": 31.5083,
|
| 1431 |
+
"active_dims": 60,
|
| 1432 |
+
"active_per_slot": [
|
| 1433 |
+
10,
|
| 1434 |
+
10,
|
| 1435 |
+
10,
|
| 1436 |
+
10,
|
| 1437 |
+
10,
|
| 1438 |
+
10
|
| 1439 |
+
],
|
| 1440 |
+
"usage_max": 0.1667,
|
| 1441 |
+
"usage_min": 0.1667,
|
| 1442 |
+
"usage": [
|
| 1443 |
+
0.1667,
|
| 1444 |
+
0.1667,
|
| 1445 |
+
0.1667,
|
| 1446 |
+
0.1667,
|
| 1447 |
+
0.1667,
|
| 1448 |
+
0.1667
|
| 1449 |
+
],
|
| 1450 |
+
"active_slots": 6,
|
| 1451 |
+
"grad_norm": 0.063,
|
| 1452 |
+
"sec": 137.2
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 45,
|
| 1456 |
+
"beta": 0.0003,
|
| 1457 |
+
"lr": 8.3e-06,
|
| 1458 |
+
"train_loss": 0.045801,
|
| 1459 |
+
"train_recon": 0.036387,
|
| 1460 |
+
"train_kl": 31.3817,
|
| 1461 |
+
"val_loss": 0.043698,
|
| 1462 |
+
"val_recon": 0.034245,
|
| 1463 |
+
"val_kl": 31.5125,
|
| 1464 |
+
"active_dims": 60,
|
| 1465 |
+
"active_per_slot": [
|
| 1466 |
+
10,
|
| 1467 |
+
10,
|
| 1468 |
+
10,
|
| 1469 |
+
10,
|
| 1470 |
+
10,
|
| 1471 |
+
10
|
| 1472 |
+
],
|
| 1473 |
+
"usage_max": 0.1667,
|
| 1474 |
+
"usage_min": 0.1667,
|
| 1475 |
+
"usage": [
|
| 1476 |
+
0.1667,
|
| 1477 |
+
0.1667,
|
| 1478 |
+
0.1667,
|
| 1479 |
+
0.1667,
|
| 1480 |
+
0.1667,
|
| 1481 |
+
0.1667
|
| 1482 |
+
],
|
| 1483 |
+
"active_slots": 6,
|
| 1484 |
+
"grad_norm": 0.061,
|
| 1485 |
+
"sec": 138.0
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 46,
|
| 1489 |
+
"beta": 0.0003,
|
| 1490 |
+
"lr": 5.7e-06,
|
| 1491 |
+
"train_loss": 0.045769,
|
| 1492 |
+
"train_recon": 0.036352,
|
| 1493 |
+
"train_kl": 31.3907,
|
| 1494 |
+
"val_loss": 0.043514,
|
| 1495 |
+
"val_recon": 0.034157,
|
| 1496 |
+
"val_kl": 31.1888,
|
| 1497 |
+
"active_dims": 60,
|
| 1498 |
+
"active_per_slot": [
|
| 1499 |
+
10,
|
| 1500 |
+
10,
|
| 1501 |
+
10,
|
| 1502 |
+
10,
|
| 1503 |
+
10,
|
| 1504 |
+
10
|
| 1505 |
+
],
|
| 1506 |
+
"usage_max": 0.1667,
|
| 1507 |
+
"usage_min": 0.1667,
|
| 1508 |
+
"usage": [
|
| 1509 |
+
0.1667,
|
| 1510 |
+
0.1667,
|
| 1511 |
+
0.1667,
|
| 1512 |
+
0.1667,
|
| 1513 |
+
0.1667,
|
| 1514 |
+
0.1667
|
| 1515 |
+
],
|
| 1516 |
+
"active_slots": 6,
|
| 1517 |
+
"grad_norm": 0.06,
|
| 1518 |
+
"sec": 137.9
|
| 1519 |
+
},
|
| 1520 |
+
{
|
| 1521 |
+
"epoch": 47,
|
| 1522 |
+
"beta": 0.0003,
|
| 1523 |
+
"lr": 3.6e-06,
|
| 1524 |
+
"train_loss": 0.045749,
|
| 1525 |
+
"train_recon": 0.036326,
|
| 1526 |
+
"train_kl": 31.4093,
|
| 1527 |
+
"val_loss": 0.043671,
|
| 1528 |
+
"val_recon": 0.034251,
|
| 1529 |
+
"val_kl": 31.4004,
|
| 1530 |
+
"active_dims": 60,
|
| 1531 |
+
"active_per_slot": [
|
| 1532 |
+
10,
|
| 1533 |
+
10,
|
| 1534 |
+
10,
|
| 1535 |
+
10,
|
| 1536 |
+
10,
|
| 1537 |
+
10
|
| 1538 |
+
],
|
| 1539 |
+
"usage_max": 0.1667,
|
| 1540 |
+
"usage_min": 0.1667,
|
| 1541 |
+
"usage": [
|
| 1542 |
+
0.1667,
|
| 1543 |
+
0.1667,
|
| 1544 |
+
0.1667,
|
| 1545 |
+
0.1667,
|
| 1546 |
+
0.1667,
|
| 1547 |
+
0.1667
|
| 1548 |
+
],
|
| 1549 |
+
"active_slots": 6,
|
| 1550 |
+
"grad_norm": 0.06,
|
| 1551 |
+
"sec": 138.1
|
| 1552 |
+
},
|
| 1553 |
+
{
|
| 1554 |
+
"epoch": 48,
|
| 1555 |
+
"beta": 0.0003,
|
| 1556 |
+
"lr": 2.2e-06,
|
| 1557 |
+
"train_loss": 0.045732,
|
| 1558 |
+
"train_recon": 0.036288,
|
| 1559 |
+
"train_kl": 31.4821,
|
| 1560 |
+
"val_loss": 0.043779,
|
| 1561 |
+
"val_recon": 0.034306,
|
| 1562 |
+
"val_kl": 31.5788,
|
| 1563 |
+
"active_dims": 60,
|
| 1564 |
+
"active_per_slot": [
|
| 1565 |
+
10,
|
| 1566 |
+
10,
|
| 1567 |
+
10,
|
| 1568 |
+
10,
|
| 1569 |
+
10,
|
| 1570 |
+
10
|
| 1571 |
+
],
|
| 1572 |
+
"usage_max": 0.1667,
|
| 1573 |
+
"usage_min": 0.1667,
|
| 1574 |
+
"usage": [
|
| 1575 |
+
0.1667,
|
| 1576 |
+
0.1667,
|
| 1577 |
+
0.1667,
|
| 1578 |
+
0.1667,
|
| 1579 |
+
0.1667,
|
| 1580 |
+
0.1667
|
| 1581 |
+
],
|
| 1582 |
+
"active_slots": 6,
|
| 1583 |
+
"grad_norm": 0.059,
|
| 1584 |
+
"sec": 138.6
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 49,
|
| 1588 |
+
"beta": 0.0003,
|
| 1589 |
+
"lr": 1.3e-06,
|
| 1590 |
+
"train_loss": 0.045732,
|
| 1591 |
+
"train_recon": 0.036296,
|
| 1592 |
+
"train_kl": 31.4532,
|
| 1593 |
+
"val_loss": 0.043684,
|
| 1594 |
+
"val_recon": 0.03423,
|
| 1595 |
+
"val_kl": 31.5142,
|
| 1596 |
+
"active_dims": 60,
|
| 1597 |
+
"active_per_slot": [
|
| 1598 |
+
10,
|
| 1599 |
+
10,
|
| 1600 |
+
10,
|
| 1601 |
+
10,
|
| 1602 |
+
10,
|
| 1603 |
+
10
|
| 1604 |
+
],
|
| 1605 |
+
"usage_max": 0.1667,
|
| 1606 |
+
"usage_min": 0.1667,
|
| 1607 |
+
"usage": [
|
| 1608 |
+
0.1667,
|
| 1609 |
+
0.1667,
|
| 1610 |
+
0.1667,
|
| 1611 |
+
0.1667,
|
| 1612 |
+
0.1667,
|
| 1613 |
+
0.1667
|
| 1614 |
+
],
|
| 1615 |
+
"active_slots": 6,
|
| 1616 |
+
"grad_norm": 0.058,
|
| 1617 |
+
"sec": 138.1
|
| 1618 |
+
},
|
| 1619 |
+
{
|
| 1620 |
+
"epoch": 50,
|
| 1621 |
+
"beta": 0.0003,
|
| 1622 |
+
"lr": 1e-06,
|
| 1623 |
+
"train_loss": 0.045697,
|
| 1624 |
+
"train_recon": 0.036254,
|
| 1625 |
+
"train_kl": 31.4775,
|
| 1626 |
+
"val_loss": 0.043661,
|
| 1627 |
+
"val_recon": 0.034226,
|
| 1628 |
+
"val_kl": 31.4496,
|
| 1629 |
+
"active_dims": 60,
|
| 1630 |
+
"active_per_slot": [
|
| 1631 |
+
10,
|
| 1632 |
+
10,
|
| 1633 |
+
10,
|
| 1634 |
+
10,
|
| 1635 |
+
10,
|
| 1636 |
+
10
|
| 1637 |
+
],
|
| 1638 |
+
"usage_max": 0.1667,
|
| 1639 |
+
"usage_min": 0.1667,
|
| 1640 |
+
"usage": [
|
| 1641 |
+
0.1667,
|
| 1642 |
+
0.1667,
|
| 1643 |
+
0.1667,
|
| 1644 |
+
0.1667,
|
| 1645 |
+
0.1667,
|
| 1646 |
+
0.1667
|
| 1647 |
+
],
|
| 1648 |
+
"active_slots": 6,
|
| 1649 |
+
"grad_norm": 0.058,
|
| 1650 |
+
"sec": 138.5
|
| 1651 |
}
|
| 1652 |
]
|