Uploaded checkpoint-3000
Browse files- model.safetensors +1 -1
- optimizer.pt +1 -1
- rng_state.pth +1 -1
- scheduler.pt +1 -1
- trainer_state.json +711 -3
model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 2836579040
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:324380f8549051ad119f7a5f1ecf2e45c0459be0dc3ff417db289f9e126509ba
|
3 |
size 2836579040
|
optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 5673376169
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:577bb6ff5b16221ddbe7a96caee67d6e6e0b7837ed6fef1eb4c5f9ac3ea49052
|
3 |
size 5673376169
|
rng_state.pth
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 14244
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5e5fc0f601c838403e67acb49c7b54f73c6cd7b1fce2cc26fd3c09d0ed6bb5bf
|
3 |
size 14244
|
scheduler.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1064
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a89895d690ce4c7033aa40dfb76059ed31cde2d2d86428cdbd6159c4cf947465
|
3 |
size 1064
|
trainer_state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch": 0.
|
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,
|
@@ -1423,6 +1423,714 @@
|
|
1423 |
"eval_samples_per_second": 15.462,
|
1424 |
"eval_steps_per_second": 15.462,
|
1425 |
"step": 2000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1426 |
}
|
1427 |
],
|
1428 |
"logging_steps": 10,
|
@@ -1430,7 +2138,7 @@
|
|
1430 |
"num_input_tokens_seen": 0,
|
1431 |
"num_train_epochs": 1,
|
1432 |
"save_steps": 1000,
|
1433 |
-
"total_flos":
|
1434 |
"train_batch_size": 1,
|
1435 |
"trial_name": null,
|
1436 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.15,
|
5 |
"eval_steps": 1000,
|
6 |
+
"global_step": 3000,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
1423 |
"eval_samples_per_second": 15.462,
|
1424 |
"eval_steps_per_second": 15.462,
|
1425 |
"step": 2000
|
1426 |
+
},
|
1427 |
+
{
|
1428 |
+
"epoch": 0.1,
|
1429 |
+
"grad_norm": 8.0,
|
1430 |
+
"learning_rate": 7.308888888888889e-07,
|
1431 |
+
"loss": 0.0596,
|
1432 |
+
"step": 2010
|
1433 |
+
},
|
1434 |
+
{
|
1435 |
+
"epoch": 0.1,
|
1436 |
+
"grad_norm": 5.84375,
|
1437 |
+
"learning_rate": 7.284444444444445e-07,
|
1438 |
+
"loss": 0.0587,
|
1439 |
+
"step": 2020
|
1440 |
+
},
|
1441 |
+
{
|
1442 |
+
"epoch": 0.1,
|
1443 |
+
"grad_norm": 5.78125,
|
1444 |
+
"learning_rate": 7.260000000000001e-07,
|
1445 |
+
"loss": 0.0605,
|
1446 |
+
"step": 2030
|
1447 |
+
},
|
1448 |
+
{
|
1449 |
+
"epoch": 0.1,
|
1450 |
+
"grad_norm": 6.65625,
|
1451 |
+
"learning_rate": 7.235555555555556e-07,
|
1452 |
+
"loss": 0.0557,
|
1453 |
+
"step": 2040
|
1454 |
+
},
|
1455 |
+
{
|
1456 |
+
"epoch": 0.1,
|
1457 |
+
"grad_norm": 5.65625,
|
1458 |
+
"learning_rate": 7.211111111111112e-07,
|
1459 |
+
"loss": 0.0572,
|
1460 |
+
"step": 2050
|
1461 |
+
},
|
1462 |
+
{
|
1463 |
+
"epoch": 0.1,
|
1464 |
+
"grad_norm": 5.84375,
|
1465 |
+
"learning_rate": 7.186666666666667e-07,
|
1466 |
+
"loss": 0.053,
|
1467 |
+
"step": 2060
|
1468 |
+
},
|
1469 |
+
{
|
1470 |
+
"epoch": 0.1,
|
1471 |
+
"grad_norm": 5.78125,
|
1472 |
+
"learning_rate": 7.162222222222222e-07,
|
1473 |
+
"loss": 0.0582,
|
1474 |
+
"step": 2070
|
1475 |
+
},
|
1476 |
+
{
|
1477 |
+
"epoch": 0.1,
|
1478 |
+
"grad_norm": 6.90625,
|
1479 |
+
"learning_rate": 7.137777777777777e-07,
|
1480 |
+
"loss": 0.0644,
|
1481 |
+
"step": 2080
|
1482 |
+
},
|
1483 |
+
{
|
1484 |
+
"epoch": 0.1,
|
1485 |
+
"grad_norm": 7.96875,
|
1486 |
+
"learning_rate": 7.113333333333333e-07,
|
1487 |
+
"loss": 0.0588,
|
1488 |
+
"step": 2090
|
1489 |
+
},
|
1490 |
+
{
|
1491 |
+
"epoch": 0.1,
|
1492 |
+
"grad_norm": 5.15625,
|
1493 |
+
"learning_rate": 7.08888888888889e-07,
|
1494 |
+
"loss": 0.0535,
|
1495 |
+
"step": 2100
|
1496 |
+
},
|
1497 |
+
{
|
1498 |
+
"epoch": 0.11,
|
1499 |
+
"grad_norm": 7.46875,
|
1500 |
+
"learning_rate": 7.064444444444445e-07,
|
1501 |
+
"loss": 0.0573,
|
1502 |
+
"step": 2110
|
1503 |
+
},
|
1504 |
+
{
|
1505 |
+
"epoch": 0.11,
|
1506 |
+
"grad_norm": 6.28125,
|
1507 |
+
"learning_rate": 7.040000000000001e-07,
|
1508 |
+
"loss": 0.0584,
|
1509 |
+
"step": 2120
|
1510 |
+
},
|
1511 |
+
{
|
1512 |
+
"epoch": 0.11,
|
1513 |
+
"grad_norm": 5.75,
|
1514 |
+
"learning_rate": 7.015555555555556e-07,
|
1515 |
+
"loss": 0.0532,
|
1516 |
+
"step": 2130
|
1517 |
+
},
|
1518 |
+
{
|
1519 |
+
"epoch": 0.11,
|
1520 |
+
"grad_norm": 7.15625,
|
1521 |
+
"learning_rate": 6.991111111111111e-07,
|
1522 |
+
"loss": 0.0526,
|
1523 |
+
"step": 2140
|
1524 |
+
},
|
1525 |
+
{
|
1526 |
+
"epoch": 0.11,
|
1527 |
+
"grad_norm": 7.40625,
|
1528 |
+
"learning_rate": 6.966666666666666e-07,
|
1529 |
+
"loss": 0.0621,
|
1530 |
+
"step": 2150
|
1531 |
+
},
|
1532 |
+
{
|
1533 |
+
"epoch": 0.11,
|
1534 |
+
"grad_norm": 9.125,
|
1535 |
+
"learning_rate": 6.942222222222222e-07,
|
1536 |
+
"loss": 0.055,
|
1537 |
+
"step": 2160
|
1538 |
+
},
|
1539 |
+
{
|
1540 |
+
"epoch": 0.11,
|
1541 |
+
"grad_norm": 8.625,
|
1542 |
+
"learning_rate": 6.917777777777779e-07,
|
1543 |
+
"loss": 0.0614,
|
1544 |
+
"step": 2170
|
1545 |
+
},
|
1546 |
+
{
|
1547 |
+
"epoch": 0.11,
|
1548 |
+
"grad_norm": 6.375,
|
1549 |
+
"learning_rate": 6.893333333333334e-07,
|
1550 |
+
"loss": 0.0527,
|
1551 |
+
"step": 2180
|
1552 |
+
},
|
1553 |
+
{
|
1554 |
+
"epoch": 0.11,
|
1555 |
+
"grad_norm": 5.21875,
|
1556 |
+
"learning_rate": 6.868888888888889e-07,
|
1557 |
+
"loss": 0.069,
|
1558 |
+
"step": 2190
|
1559 |
+
},
|
1560 |
+
{
|
1561 |
+
"epoch": 0.11,
|
1562 |
+
"grad_norm": 5.65625,
|
1563 |
+
"learning_rate": 6.844444444444445e-07,
|
1564 |
+
"loss": 0.0524,
|
1565 |
+
"step": 2200
|
1566 |
+
},
|
1567 |
+
{
|
1568 |
+
"epoch": 0.11,
|
1569 |
+
"grad_norm": 5.84375,
|
1570 |
+
"learning_rate": 6.82e-07,
|
1571 |
+
"loss": 0.0596,
|
1572 |
+
"step": 2210
|
1573 |
+
},
|
1574 |
+
{
|
1575 |
+
"epoch": 0.11,
|
1576 |
+
"grad_norm": 5.15625,
|
1577 |
+
"learning_rate": 6.795555555555555e-07,
|
1578 |
+
"loss": 0.0587,
|
1579 |
+
"step": 2220
|
1580 |
+
},
|
1581 |
+
{
|
1582 |
+
"epoch": 0.11,
|
1583 |
+
"grad_norm": 5.375,
|
1584 |
+
"learning_rate": 6.771111111111111e-07,
|
1585 |
+
"loss": 0.0563,
|
1586 |
+
"step": 2230
|
1587 |
+
},
|
1588 |
+
{
|
1589 |
+
"epoch": 0.11,
|
1590 |
+
"grad_norm": 5.71875,
|
1591 |
+
"learning_rate": 6.746666666666667e-07,
|
1592 |
+
"loss": 0.0553,
|
1593 |
+
"step": 2240
|
1594 |
+
},
|
1595 |
+
{
|
1596 |
+
"epoch": 0.11,
|
1597 |
+
"grad_norm": 5.5625,
|
1598 |
+
"learning_rate": 6.722222222222223e-07,
|
1599 |
+
"loss": 0.0557,
|
1600 |
+
"step": 2250
|
1601 |
+
},
|
1602 |
+
{
|
1603 |
+
"epoch": 0.11,
|
1604 |
+
"grad_norm": 6.8125,
|
1605 |
+
"learning_rate": 6.697777777777778e-07,
|
1606 |
+
"loss": 0.0557,
|
1607 |
+
"step": 2260
|
1608 |
+
},
|
1609 |
+
{
|
1610 |
+
"epoch": 0.11,
|
1611 |
+
"grad_norm": 5.8125,
|
1612 |
+
"learning_rate": 6.673333333333334e-07,
|
1613 |
+
"loss": 0.0551,
|
1614 |
+
"step": 2270
|
1615 |
+
},
|
1616 |
+
{
|
1617 |
+
"epoch": 0.11,
|
1618 |
+
"grad_norm": 5.78125,
|
1619 |
+
"learning_rate": 6.648888888888889e-07,
|
1620 |
+
"loss": 0.0535,
|
1621 |
+
"step": 2280
|
1622 |
+
},
|
1623 |
+
{
|
1624 |
+
"epoch": 0.11,
|
1625 |
+
"grad_norm": 5.375,
|
1626 |
+
"learning_rate": 6.624444444444445e-07,
|
1627 |
+
"loss": 0.0501,
|
1628 |
+
"step": 2290
|
1629 |
+
},
|
1630 |
+
{
|
1631 |
+
"epoch": 0.12,
|
1632 |
+
"grad_norm": 6.40625,
|
1633 |
+
"learning_rate": 6.6e-07,
|
1634 |
+
"loss": 0.0586,
|
1635 |
+
"step": 2300
|
1636 |
+
},
|
1637 |
+
{
|
1638 |
+
"epoch": 0.12,
|
1639 |
+
"grad_norm": 7.3125,
|
1640 |
+
"learning_rate": 6.575555555555556e-07,
|
1641 |
+
"loss": 0.0548,
|
1642 |
+
"step": 2310
|
1643 |
+
},
|
1644 |
+
{
|
1645 |
+
"epoch": 0.12,
|
1646 |
+
"grad_norm": 8.5,
|
1647 |
+
"learning_rate": 6.551111111111111e-07,
|
1648 |
+
"loss": 0.0566,
|
1649 |
+
"step": 2320
|
1650 |
+
},
|
1651 |
+
{
|
1652 |
+
"epoch": 0.12,
|
1653 |
+
"grad_norm": 5.5,
|
1654 |
+
"learning_rate": 6.526666666666667e-07,
|
1655 |
+
"loss": 0.0656,
|
1656 |
+
"step": 2330
|
1657 |
+
},
|
1658 |
+
{
|
1659 |
+
"epoch": 0.12,
|
1660 |
+
"grad_norm": 5.625,
|
1661 |
+
"learning_rate": 6.502222222222222e-07,
|
1662 |
+
"loss": 0.052,
|
1663 |
+
"step": 2340
|
1664 |
+
},
|
1665 |
+
{
|
1666 |
+
"epoch": 0.12,
|
1667 |
+
"grad_norm": 6.1875,
|
1668 |
+
"learning_rate": 6.477777777777779e-07,
|
1669 |
+
"loss": 0.0574,
|
1670 |
+
"step": 2350
|
1671 |
+
},
|
1672 |
+
{
|
1673 |
+
"epoch": 0.12,
|
1674 |
+
"grad_norm": 7.53125,
|
1675 |
+
"learning_rate": 6.453333333333334e-07,
|
1676 |
+
"loss": 0.0564,
|
1677 |
+
"step": 2360
|
1678 |
+
},
|
1679 |
+
{
|
1680 |
+
"epoch": 0.12,
|
1681 |
+
"grad_norm": 5.0,
|
1682 |
+
"learning_rate": 6.428888888888889e-07,
|
1683 |
+
"loss": 0.0554,
|
1684 |
+
"step": 2370
|
1685 |
+
},
|
1686 |
+
{
|
1687 |
+
"epoch": 0.12,
|
1688 |
+
"grad_norm": 6.1875,
|
1689 |
+
"learning_rate": 6.404444444444444e-07,
|
1690 |
+
"loss": 0.0513,
|
1691 |
+
"step": 2380
|
1692 |
+
},
|
1693 |
+
{
|
1694 |
+
"epoch": 0.12,
|
1695 |
+
"grad_norm": 7.03125,
|
1696 |
+
"learning_rate": 6.38e-07,
|
1697 |
+
"loss": 0.0525,
|
1698 |
+
"step": 2390
|
1699 |
+
},
|
1700 |
+
{
|
1701 |
+
"epoch": 0.12,
|
1702 |
+
"grad_norm": 5.25,
|
1703 |
+
"learning_rate": 6.355555555555555e-07,
|
1704 |
+
"loss": 0.0522,
|
1705 |
+
"step": 2400
|
1706 |
+
},
|
1707 |
+
{
|
1708 |
+
"epoch": 0.12,
|
1709 |
+
"grad_norm": 6.125,
|
1710 |
+
"learning_rate": 6.331111111111112e-07,
|
1711 |
+
"loss": 0.06,
|
1712 |
+
"step": 2410
|
1713 |
+
},
|
1714 |
+
{
|
1715 |
+
"epoch": 0.12,
|
1716 |
+
"grad_norm": 5.0625,
|
1717 |
+
"learning_rate": 6.306666666666668e-07,
|
1718 |
+
"loss": 0.0542,
|
1719 |
+
"step": 2420
|
1720 |
+
},
|
1721 |
+
{
|
1722 |
+
"epoch": 0.12,
|
1723 |
+
"grad_norm": 5.71875,
|
1724 |
+
"learning_rate": 6.282222222222223e-07,
|
1725 |
+
"loss": 0.0614,
|
1726 |
+
"step": 2430
|
1727 |
+
},
|
1728 |
+
{
|
1729 |
+
"epoch": 0.12,
|
1730 |
+
"grad_norm": 6.8125,
|
1731 |
+
"learning_rate": 6.257777777777778e-07,
|
1732 |
+
"loss": 0.055,
|
1733 |
+
"step": 2440
|
1734 |
+
},
|
1735 |
+
{
|
1736 |
+
"epoch": 0.12,
|
1737 |
+
"grad_norm": 4.78125,
|
1738 |
+
"learning_rate": 6.233333333333333e-07,
|
1739 |
+
"loss": 0.0533,
|
1740 |
+
"step": 2450
|
1741 |
+
},
|
1742 |
+
{
|
1743 |
+
"epoch": 0.12,
|
1744 |
+
"grad_norm": 5.6875,
|
1745 |
+
"learning_rate": 6.208888888888889e-07,
|
1746 |
+
"loss": 0.0595,
|
1747 |
+
"step": 2460
|
1748 |
+
},
|
1749 |
+
{
|
1750 |
+
"epoch": 0.12,
|
1751 |
+
"grad_norm": 7.875,
|
1752 |
+
"learning_rate": 6.184444444444444e-07,
|
1753 |
+
"loss": 0.0589,
|
1754 |
+
"step": 2470
|
1755 |
+
},
|
1756 |
+
{
|
1757 |
+
"epoch": 0.12,
|
1758 |
+
"grad_norm": 6.71875,
|
1759 |
+
"learning_rate": 6.160000000000001e-07,
|
1760 |
+
"loss": 0.0513,
|
1761 |
+
"step": 2480
|
1762 |
+
},
|
1763 |
+
{
|
1764 |
+
"epoch": 0.12,
|
1765 |
+
"grad_norm": 7.125,
|
1766 |
+
"learning_rate": 6.135555555555556e-07,
|
1767 |
+
"loss": 0.0566,
|
1768 |
+
"step": 2490
|
1769 |
+
},
|
1770 |
+
{
|
1771 |
+
"epoch": 0.12,
|
1772 |
+
"grad_norm": 5.875,
|
1773 |
+
"learning_rate": 6.111111111111112e-07,
|
1774 |
+
"loss": 0.0549,
|
1775 |
+
"step": 2500
|
1776 |
+
},
|
1777 |
+
{
|
1778 |
+
"epoch": 0.13,
|
1779 |
+
"grad_norm": 7.125,
|
1780 |
+
"learning_rate": 6.086666666666667e-07,
|
1781 |
+
"loss": 0.0583,
|
1782 |
+
"step": 2510
|
1783 |
+
},
|
1784 |
+
{
|
1785 |
+
"epoch": 0.13,
|
1786 |
+
"grad_norm": 5.875,
|
1787 |
+
"learning_rate": 6.062222222222222e-07,
|
1788 |
+
"loss": 0.0632,
|
1789 |
+
"step": 2520
|
1790 |
+
},
|
1791 |
+
{
|
1792 |
+
"epoch": 0.13,
|
1793 |
+
"grad_norm": 4.875,
|
1794 |
+
"learning_rate": 6.037777777777777e-07,
|
1795 |
+
"loss": 0.058,
|
1796 |
+
"step": 2530
|
1797 |
+
},
|
1798 |
+
{
|
1799 |
+
"epoch": 0.13,
|
1800 |
+
"grad_norm": 5.03125,
|
1801 |
+
"learning_rate": 6.013333333333334e-07,
|
1802 |
+
"loss": 0.0566,
|
1803 |
+
"step": 2540
|
1804 |
+
},
|
1805 |
+
{
|
1806 |
+
"epoch": 0.13,
|
1807 |
+
"grad_norm": 5.8125,
|
1808 |
+
"learning_rate": 5.988888888888889e-07,
|
1809 |
+
"loss": 0.0588,
|
1810 |
+
"step": 2550
|
1811 |
+
},
|
1812 |
+
{
|
1813 |
+
"epoch": 0.13,
|
1814 |
+
"grad_norm": 6.21875,
|
1815 |
+
"learning_rate": 5.964444444444445e-07,
|
1816 |
+
"loss": 0.059,
|
1817 |
+
"step": 2560
|
1818 |
+
},
|
1819 |
+
{
|
1820 |
+
"epoch": 0.13,
|
1821 |
+
"grad_norm": 5.09375,
|
1822 |
+
"learning_rate": 5.94e-07,
|
1823 |
+
"loss": 0.0592,
|
1824 |
+
"step": 2570
|
1825 |
+
},
|
1826 |
+
{
|
1827 |
+
"epoch": 0.13,
|
1828 |
+
"grad_norm": 7.59375,
|
1829 |
+
"learning_rate": 5.915555555555556e-07,
|
1830 |
+
"loss": 0.0599,
|
1831 |
+
"step": 2580
|
1832 |
+
},
|
1833 |
+
{
|
1834 |
+
"epoch": 0.13,
|
1835 |
+
"grad_norm": 5.8125,
|
1836 |
+
"learning_rate": 5.891111111111111e-07,
|
1837 |
+
"loss": 0.0518,
|
1838 |
+
"step": 2590
|
1839 |
+
},
|
1840 |
+
{
|
1841 |
+
"epoch": 0.13,
|
1842 |
+
"grad_norm": 8.0,
|
1843 |
+
"learning_rate": 5.866666666666667e-07,
|
1844 |
+
"loss": 0.0578,
|
1845 |
+
"step": 2600
|
1846 |
+
},
|
1847 |
+
{
|
1848 |
+
"epoch": 0.13,
|
1849 |
+
"grad_norm": 7.125,
|
1850 |
+
"learning_rate": 5.842222222222223e-07,
|
1851 |
+
"loss": 0.0509,
|
1852 |
+
"step": 2610
|
1853 |
+
},
|
1854 |
+
{
|
1855 |
+
"epoch": 0.13,
|
1856 |
+
"grad_norm": 4.96875,
|
1857 |
+
"learning_rate": 5.817777777777778e-07,
|
1858 |
+
"loss": 0.0536,
|
1859 |
+
"step": 2620
|
1860 |
+
},
|
1861 |
+
{
|
1862 |
+
"epoch": 0.13,
|
1863 |
+
"grad_norm": 7.25,
|
1864 |
+
"learning_rate": 5.793333333333333e-07,
|
1865 |
+
"loss": 0.052,
|
1866 |
+
"step": 2630
|
1867 |
+
},
|
1868 |
+
{
|
1869 |
+
"epoch": 0.13,
|
1870 |
+
"grad_norm": 8.8125,
|
1871 |
+
"learning_rate": 5.768888888888889e-07,
|
1872 |
+
"loss": 0.0555,
|
1873 |
+
"step": 2640
|
1874 |
+
},
|
1875 |
+
{
|
1876 |
+
"epoch": 0.13,
|
1877 |
+
"grad_norm": 10.375,
|
1878 |
+
"learning_rate": 5.744444444444445e-07,
|
1879 |
+
"loss": 0.0574,
|
1880 |
+
"step": 2650
|
1881 |
+
},
|
1882 |
+
{
|
1883 |
+
"epoch": 0.13,
|
1884 |
+
"grad_norm": 7.625,
|
1885 |
+
"learning_rate": 5.720000000000001e-07,
|
1886 |
+
"loss": 0.0569,
|
1887 |
+
"step": 2660
|
1888 |
+
},
|
1889 |
+
{
|
1890 |
+
"epoch": 0.13,
|
1891 |
+
"grad_norm": 7.1875,
|
1892 |
+
"learning_rate": 5.695555555555556e-07,
|
1893 |
+
"loss": 0.0565,
|
1894 |
+
"step": 2670
|
1895 |
+
},
|
1896 |
+
{
|
1897 |
+
"epoch": 0.13,
|
1898 |
+
"grad_norm": 6.71875,
|
1899 |
+
"learning_rate": 5.671111111111111e-07,
|
1900 |
+
"loss": 0.0518,
|
1901 |
+
"step": 2680
|
1902 |
+
},
|
1903 |
+
{
|
1904 |
+
"epoch": 0.13,
|
1905 |
+
"grad_norm": 6.71875,
|
1906 |
+
"learning_rate": 5.646666666666667e-07,
|
1907 |
+
"loss": 0.0614,
|
1908 |
+
"step": 2690
|
1909 |
+
},
|
1910 |
+
{
|
1911 |
+
"epoch": 0.14,
|
1912 |
+
"grad_norm": 5.59375,
|
1913 |
+
"learning_rate": 5.622222222222222e-07,
|
1914 |
+
"loss": 0.054,
|
1915 |
+
"step": 2700
|
1916 |
+
},
|
1917 |
+
{
|
1918 |
+
"epoch": 0.14,
|
1919 |
+
"grad_norm": 8.0625,
|
1920 |
+
"learning_rate": 5.597777777777777e-07,
|
1921 |
+
"loss": 0.0573,
|
1922 |
+
"step": 2710
|
1923 |
+
},
|
1924 |
+
{
|
1925 |
+
"epoch": 0.14,
|
1926 |
+
"grad_norm": 8.25,
|
1927 |
+
"learning_rate": 5.573333333333335e-07,
|
1928 |
+
"loss": 0.0602,
|
1929 |
+
"step": 2720
|
1930 |
+
},
|
1931 |
+
{
|
1932 |
+
"epoch": 0.14,
|
1933 |
+
"grad_norm": 8.625,
|
1934 |
+
"learning_rate": 5.54888888888889e-07,
|
1935 |
+
"loss": 0.0643,
|
1936 |
+
"step": 2730
|
1937 |
+
},
|
1938 |
+
{
|
1939 |
+
"epoch": 0.14,
|
1940 |
+
"grad_norm": 7.25,
|
1941 |
+
"learning_rate": 5.524444444444445e-07,
|
1942 |
+
"loss": 0.0589,
|
1943 |
+
"step": 2740
|
1944 |
+
},
|
1945 |
+
{
|
1946 |
+
"epoch": 0.14,
|
1947 |
+
"grad_norm": 5.78125,
|
1948 |
+
"learning_rate": 5.5e-07,
|
1949 |
+
"loss": 0.0508,
|
1950 |
+
"step": 2750
|
1951 |
+
},
|
1952 |
+
{
|
1953 |
+
"epoch": 0.14,
|
1954 |
+
"grad_norm": 5.625,
|
1955 |
+
"learning_rate": 5.475555555555556e-07,
|
1956 |
+
"loss": 0.0644,
|
1957 |
+
"step": 2760
|
1958 |
+
},
|
1959 |
+
{
|
1960 |
+
"epoch": 0.14,
|
1961 |
+
"grad_norm": 6.15625,
|
1962 |
+
"learning_rate": 5.451111111111112e-07,
|
1963 |
+
"loss": 0.0525,
|
1964 |
+
"step": 2770
|
1965 |
+
},
|
1966 |
+
{
|
1967 |
+
"epoch": 0.14,
|
1968 |
+
"grad_norm": 7.5625,
|
1969 |
+
"learning_rate": 5.426666666666667e-07,
|
1970 |
+
"loss": 0.0626,
|
1971 |
+
"step": 2780
|
1972 |
+
},
|
1973 |
+
{
|
1974 |
+
"epoch": 0.14,
|
1975 |
+
"grad_norm": 7.875,
|
1976 |
+
"learning_rate": 5.402222222222222e-07,
|
1977 |
+
"loss": 0.0568,
|
1978 |
+
"step": 2790
|
1979 |
+
},
|
1980 |
+
{
|
1981 |
+
"epoch": 0.14,
|
1982 |
+
"grad_norm": 5.46875,
|
1983 |
+
"learning_rate": 5.377777777777778e-07,
|
1984 |
+
"loss": 0.0535,
|
1985 |
+
"step": 2800
|
1986 |
+
},
|
1987 |
+
{
|
1988 |
+
"epoch": 0.14,
|
1989 |
+
"grad_norm": 5.46875,
|
1990 |
+
"learning_rate": 5.353333333333334e-07,
|
1991 |
+
"loss": 0.0539,
|
1992 |
+
"step": 2810
|
1993 |
+
},
|
1994 |
+
{
|
1995 |
+
"epoch": 0.14,
|
1996 |
+
"grad_norm": 5.15625,
|
1997 |
+
"learning_rate": 5.328888888888889e-07,
|
1998 |
+
"loss": 0.0467,
|
1999 |
+
"step": 2820
|
2000 |
+
},
|
2001 |
+
{
|
2002 |
+
"epoch": 0.14,
|
2003 |
+
"grad_norm": 5.90625,
|
2004 |
+
"learning_rate": 5.304444444444444e-07,
|
2005 |
+
"loss": 0.0613,
|
2006 |
+
"step": 2830
|
2007 |
+
},
|
2008 |
+
{
|
2009 |
+
"epoch": 0.14,
|
2010 |
+
"grad_norm": 6.125,
|
2011 |
+
"learning_rate": 5.28e-07,
|
2012 |
+
"loss": 0.0516,
|
2013 |
+
"step": 2840
|
2014 |
+
},
|
2015 |
+
{
|
2016 |
+
"epoch": 0.14,
|
2017 |
+
"grad_norm": 5.8125,
|
2018 |
+
"learning_rate": 5.255555555555556e-07,
|
2019 |
+
"loss": 0.0555,
|
2020 |
+
"step": 2850
|
2021 |
+
},
|
2022 |
+
{
|
2023 |
+
"epoch": 0.14,
|
2024 |
+
"grad_norm": 8.125,
|
2025 |
+
"learning_rate": 5.231111111111111e-07,
|
2026 |
+
"loss": 0.0608,
|
2027 |
+
"step": 2860
|
2028 |
+
},
|
2029 |
+
{
|
2030 |
+
"epoch": 0.14,
|
2031 |
+
"grad_norm": 7.78125,
|
2032 |
+
"learning_rate": 5.206666666666666e-07,
|
2033 |
+
"loss": 0.0559,
|
2034 |
+
"step": 2870
|
2035 |
+
},
|
2036 |
+
{
|
2037 |
+
"epoch": 0.14,
|
2038 |
+
"grad_norm": 6.28125,
|
2039 |
+
"learning_rate": 5.182222222222223e-07,
|
2040 |
+
"loss": 0.0566,
|
2041 |
+
"step": 2880
|
2042 |
+
},
|
2043 |
+
{
|
2044 |
+
"epoch": 0.14,
|
2045 |
+
"grad_norm": 6.53125,
|
2046 |
+
"learning_rate": 5.157777777777778e-07,
|
2047 |
+
"loss": 0.0475,
|
2048 |
+
"step": 2890
|
2049 |
+
},
|
2050 |
+
{
|
2051 |
+
"epoch": 0.14,
|
2052 |
+
"grad_norm": 5.59375,
|
2053 |
+
"learning_rate": 5.133333333333333e-07,
|
2054 |
+
"loss": 0.0537,
|
2055 |
+
"step": 2900
|
2056 |
+
},
|
2057 |
+
{
|
2058 |
+
"epoch": 0.15,
|
2059 |
+
"grad_norm": 7.75,
|
2060 |
+
"learning_rate": 5.10888888888889e-07,
|
2061 |
+
"loss": 0.0558,
|
2062 |
+
"step": 2910
|
2063 |
+
},
|
2064 |
+
{
|
2065 |
+
"epoch": 0.15,
|
2066 |
+
"grad_norm": 5.9375,
|
2067 |
+
"learning_rate": 5.084444444444445e-07,
|
2068 |
+
"loss": 0.0518,
|
2069 |
+
"step": 2920
|
2070 |
+
},
|
2071 |
+
{
|
2072 |
+
"epoch": 0.15,
|
2073 |
+
"grad_norm": 7.0625,
|
2074 |
+
"learning_rate": 5.06e-07,
|
2075 |
+
"loss": 0.0578,
|
2076 |
+
"step": 2930
|
2077 |
+
},
|
2078 |
+
{
|
2079 |
+
"epoch": 0.15,
|
2080 |
+
"grad_norm": 6.0625,
|
2081 |
+
"learning_rate": 5.035555555555556e-07,
|
2082 |
+
"loss": 0.052,
|
2083 |
+
"step": 2940
|
2084 |
+
},
|
2085 |
+
{
|
2086 |
+
"epoch": 0.15,
|
2087 |
+
"grad_norm": 5.53125,
|
2088 |
+
"learning_rate": 5.011111111111112e-07,
|
2089 |
+
"loss": 0.048,
|
2090 |
+
"step": 2950
|
2091 |
+
},
|
2092 |
+
{
|
2093 |
+
"epoch": 0.15,
|
2094 |
+
"grad_norm": 6.59375,
|
2095 |
+
"learning_rate": 4.986666666666667e-07,
|
2096 |
+
"loss": 0.0566,
|
2097 |
+
"step": 2960
|
2098 |
+
},
|
2099 |
+
{
|
2100 |
+
"epoch": 0.15,
|
2101 |
+
"grad_norm": 5.1875,
|
2102 |
+
"learning_rate": 4.962222222222223e-07,
|
2103 |
+
"loss": 0.0501,
|
2104 |
+
"step": 2970
|
2105 |
+
},
|
2106 |
+
{
|
2107 |
+
"epoch": 0.15,
|
2108 |
+
"grad_norm": 6.09375,
|
2109 |
+
"learning_rate": 4.937777777777778e-07,
|
2110 |
+
"loss": 0.0554,
|
2111 |
+
"step": 2980
|
2112 |
+
},
|
2113 |
+
{
|
2114 |
+
"epoch": 0.15,
|
2115 |
+
"grad_norm": 7.34375,
|
2116 |
+
"learning_rate": 4.913333333333334e-07,
|
2117 |
+
"loss": 0.0582,
|
2118 |
+
"step": 2990
|
2119 |
+
},
|
2120 |
+
{
|
2121 |
+
"epoch": 0.15,
|
2122 |
+
"grad_norm": 5.09375,
|
2123 |
+
"learning_rate": 4.888888888888889e-07,
|
2124 |
+
"loss": 0.0568,
|
2125 |
+
"step": 3000
|
2126 |
+
},
|
2127 |
+
{
|
2128 |
+
"epoch": 0.15,
|
2129 |
+
"eval_loss": 0.05641532689332962,
|
2130 |
+
"eval_runtime": 64.515,
|
2131 |
+
"eval_samples_per_second": 15.5,
|
2132 |
+
"eval_steps_per_second": 15.5,
|
2133 |
+
"step": 3000
|
2134 |
}
|
2135 |
],
|
2136 |
"logging_steps": 10,
|
|
|
2138 |
"num_input_tokens_seen": 0,
|
2139 |
"num_train_epochs": 1,
|
2140 |
"save_steps": 1000,
|
2141 |
+
"total_flos": 4.841766125568e+16,
|
2142 |
"train_batch_size": 1,
|
2143 |
"trial_name": null,
|
2144 |
"trial_params": null
|