Training in progress, step 1600
Browse files- last-checkpoint/optimizer.pt +1 -1
- last-checkpoint/pytorch_model.bin +1 -1
- last-checkpoint/rng_state.pth +1 -1
- last-checkpoint/scheduler.pt +1 -1
- last-checkpoint/trainer_state.json +1213 -5
- pytorch_model.bin +1 -1
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 4736616809
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e6a244c95c9746f9d57f6132321843371f3fa96ca9200f03525a3d8dbd62b8c3
|
3 |
size 4736616809
|
last-checkpoint/pytorch_model.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 2368281769
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b2bace5feec1de1af12f9432ac0a85b72ad18e1f09933a275a13f26e34955c1b
|
3 |
size 2368281769
|
last-checkpoint/rng_state.pth
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 14575
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3787c3971883551480b595d802b5952f989eaaf99e800545777f7aa67de4b26e
|
3 |
size 14575
|
last-checkpoint/scheduler.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 627
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:343a75a342817f18165e1e34ff2230eaacc1ae043cf4c83711f95c04ca831b8e
|
3 |
size 627
|
last-checkpoint/trainer_state.json
CHANGED
@@ -1,8 +1,8 @@
|
|
1 |
{
|
2 |
-
"best_metric": 2.
|
3 |
-
"best_model_checkpoint": "output/checkpoint-
|
4 |
-
"epoch":
|
5 |
-
"global_step":
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
@@ -8462,11 +8462,1219 @@
|
|
8462 |
"eval_samples_per_second": 6.072,
|
8463 |
"eval_steps_per_second": 6.072,
|
8464 |
"step": 1400
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
8465 |
}
|
8466 |
],
|
8467 |
"max_steps": 4386,
|
8468 |
"num_train_epochs": 3,
|
8469 |
-
"total_flos": 1.
|
8470 |
"trial_name": null,
|
8471 |
"trial_params": null
|
8472 |
}
|
|
|
1 |
{
|
2 |
+
"best_metric": 2.000626564025879,
|
3 |
+
"best_model_checkpoint": "output/checkpoint-1600",
|
4 |
+
"epoch": 1.0938943815063482,
|
5 |
+
"global_step": 1600,
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
|
|
8462 |
"eval_samples_per_second": 6.072,
|
8463 |
"eval_steps_per_second": 6.072,
|
8464 |
"step": 1400
|
8465 |
+
},
|
8466 |
+
{
|
8467 |
+
"epoch": 0.96,
|
8468 |
+
"learning_rate": 0.0004278547539417104,
|
8469 |
+
"loss": 2.0666,
|
8470 |
+
"step": 1401
|
8471 |
+
},
|
8472 |
+
{
|
8473 |
+
"epoch": 0.96,
|
8474 |
+
"learning_rate": 0.0004277114190157668,
|
8475 |
+
"loss": 2.1469,
|
8476 |
+
"step": 1402
|
8477 |
+
},
|
8478 |
+
{
|
8479 |
+
"epoch": 0.96,
|
8480 |
+
"learning_rate": 0.0004275680840898232,
|
8481 |
+
"loss": 2.0233,
|
8482 |
+
"step": 1403
|
8483 |
+
},
|
8484 |
+
{
|
8485 |
+
"epoch": 0.96,
|
8486 |
+
"learning_rate": 0.00042742474916387957,
|
8487 |
+
"loss": 2.0452,
|
8488 |
+
"step": 1404
|
8489 |
+
},
|
8490 |
+
{
|
8491 |
+
"epoch": 0.96,
|
8492 |
+
"learning_rate": 0.00042728141423793595,
|
8493 |
+
"loss": 2.0146,
|
8494 |
+
"step": 1405
|
8495 |
+
},
|
8496 |
+
{
|
8497 |
+
"epoch": 0.96,
|
8498 |
+
"learning_rate": 0.00042713807931199233,
|
8499 |
+
"loss": 2.1285,
|
8500 |
+
"step": 1406
|
8501 |
+
},
|
8502 |
+
{
|
8503 |
+
"epoch": 0.96,
|
8504 |
+
"learning_rate": 0.0004269947443860487,
|
8505 |
+
"loss": 2.0872,
|
8506 |
+
"step": 1407
|
8507 |
+
},
|
8508 |
+
{
|
8509 |
+
"epoch": 0.96,
|
8510 |
+
"learning_rate": 0.0004268514094601051,
|
8511 |
+
"loss": 2.0824,
|
8512 |
+
"step": 1408
|
8513 |
+
},
|
8514 |
+
{
|
8515 |
+
"epoch": 0.96,
|
8516 |
+
"learning_rate": 0.00042670807453416143,
|
8517 |
+
"loss": 2.1307,
|
8518 |
+
"step": 1409
|
8519 |
+
},
|
8520 |
+
{
|
8521 |
+
"epoch": 0.96,
|
8522 |
+
"learning_rate": 0.00042656473960821787,
|
8523 |
+
"loss": 2.0136,
|
8524 |
+
"step": 1410
|
8525 |
+
},
|
8526 |
+
{
|
8527 |
+
"epoch": 0.96,
|
8528 |
+
"learning_rate": 0.0004264214046822742,
|
8529 |
+
"loss": 2.0709,
|
8530 |
+
"step": 1411
|
8531 |
+
},
|
8532 |
+
{
|
8533 |
+
"epoch": 0.97,
|
8534 |
+
"learning_rate": 0.00042627806975633064,
|
8535 |
+
"loss": 2.0025,
|
8536 |
+
"step": 1412
|
8537 |
+
},
|
8538 |
+
{
|
8539 |
+
"epoch": 0.97,
|
8540 |
+
"learning_rate": 0.00042613473483038697,
|
8541 |
+
"loss": 2.1293,
|
8542 |
+
"step": 1413
|
8543 |
+
},
|
8544 |
+
{
|
8545 |
+
"epoch": 0.97,
|
8546 |
+
"learning_rate": 0.00042599139990444335,
|
8547 |
+
"loss": 2.1285,
|
8548 |
+
"step": 1414
|
8549 |
+
},
|
8550 |
+
{
|
8551 |
+
"epoch": 0.97,
|
8552 |
+
"learning_rate": 0.00042584806497849973,
|
8553 |
+
"loss": 2.1601,
|
8554 |
+
"step": 1415
|
8555 |
+
},
|
8556 |
+
{
|
8557 |
+
"epoch": 0.97,
|
8558 |
+
"learning_rate": 0.00042570473005255606,
|
8559 |
+
"loss": 2.0589,
|
8560 |
+
"step": 1416
|
8561 |
+
},
|
8562 |
+
{
|
8563 |
+
"epoch": 0.97,
|
8564 |
+
"learning_rate": 0.0004255613951266125,
|
8565 |
+
"loss": 2.087,
|
8566 |
+
"step": 1417
|
8567 |
+
},
|
8568 |
+
{
|
8569 |
+
"epoch": 0.97,
|
8570 |
+
"learning_rate": 0.00042541806020066883,
|
8571 |
+
"loss": 2.1115,
|
8572 |
+
"step": 1418
|
8573 |
+
},
|
8574 |
+
{
|
8575 |
+
"epoch": 0.97,
|
8576 |
+
"learning_rate": 0.00042527472527472527,
|
8577 |
+
"loss": 2.0327,
|
8578 |
+
"step": 1419
|
8579 |
+
},
|
8580 |
+
{
|
8581 |
+
"epoch": 0.97,
|
8582 |
+
"learning_rate": 0.0004251313903487816,
|
8583 |
+
"loss": 2.0255,
|
8584 |
+
"step": 1420
|
8585 |
+
},
|
8586 |
+
{
|
8587 |
+
"epoch": 0.97,
|
8588 |
+
"learning_rate": 0.00042498805542283804,
|
8589 |
+
"loss": 2.0882,
|
8590 |
+
"step": 1421
|
8591 |
+
},
|
8592 |
+
{
|
8593 |
+
"epoch": 0.97,
|
8594 |
+
"learning_rate": 0.00042484472049689436,
|
8595 |
+
"loss": 2.1193,
|
8596 |
+
"step": 1422
|
8597 |
+
},
|
8598 |
+
{
|
8599 |
+
"epoch": 0.97,
|
8600 |
+
"learning_rate": 0.00042470138557095075,
|
8601 |
+
"loss": 2.0616,
|
8602 |
+
"step": 1423
|
8603 |
+
},
|
8604 |
+
{
|
8605 |
+
"epoch": 0.97,
|
8606 |
+
"learning_rate": 0.00042455805064500713,
|
8607 |
+
"loss": 2.088,
|
8608 |
+
"step": 1424
|
8609 |
+
},
|
8610 |
+
{
|
8611 |
+
"epoch": 0.97,
|
8612 |
+
"learning_rate": 0.0004244147157190635,
|
8613 |
+
"loss": 2.1032,
|
8614 |
+
"step": 1425
|
8615 |
+
},
|
8616 |
+
{
|
8617 |
+
"epoch": 0.97,
|
8618 |
+
"learning_rate": 0.0004242713807931199,
|
8619 |
+
"loss": 2.1068,
|
8620 |
+
"step": 1426
|
8621 |
+
},
|
8622 |
+
{
|
8623 |
+
"epoch": 0.98,
|
8624 |
+
"learning_rate": 0.0004241280458671763,
|
8625 |
+
"loss": 2.0329,
|
8626 |
+
"step": 1427
|
8627 |
+
},
|
8628 |
+
{
|
8629 |
+
"epoch": 0.98,
|
8630 |
+
"learning_rate": 0.00042398471094123267,
|
8631 |
+
"loss": 2.2018,
|
8632 |
+
"step": 1428
|
8633 |
+
},
|
8634 |
+
{
|
8635 |
+
"epoch": 0.98,
|
8636 |
+
"learning_rate": 0.00042384137601528905,
|
8637 |
+
"loss": 2.107,
|
8638 |
+
"step": 1429
|
8639 |
+
},
|
8640 |
+
{
|
8641 |
+
"epoch": 0.98,
|
8642 |
+
"learning_rate": 0.0004236980410893454,
|
8643 |
+
"loss": 2.0242,
|
8644 |
+
"step": 1430
|
8645 |
+
},
|
8646 |
+
{
|
8647 |
+
"epoch": 0.98,
|
8648 |
+
"learning_rate": 0.00042355470616340176,
|
8649 |
+
"loss": 2.1247,
|
8650 |
+
"step": 1431
|
8651 |
+
},
|
8652 |
+
{
|
8653 |
+
"epoch": 0.98,
|
8654 |
+
"learning_rate": 0.00042341137123745815,
|
8655 |
+
"loss": 2.0634,
|
8656 |
+
"step": 1432
|
8657 |
+
},
|
8658 |
+
{
|
8659 |
+
"epoch": 0.98,
|
8660 |
+
"learning_rate": 0.00042326803631151453,
|
8661 |
+
"loss": 2.0362,
|
8662 |
+
"step": 1433
|
8663 |
+
},
|
8664 |
+
{
|
8665 |
+
"epoch": 0.98,
|
8666 |
+
"learning_rate": 0.0004231247013855709,
|
8667 |
+
"loss": 2.1368,
|
8668 |
+
"step": 1434
|
8669 |
+
},
|
8670 |
+
{
|
8671 |
+
"epoch": 0.98,
|
8672 |
+
"learning_rate": 0.0004229813664596273,
|
8673 |
+
"loss": 1.9856,
|
8674 |
+
"step": 1435
|
8675 |
+
},
|
8676 |
+
{
|
8677 |
+
"epoch": 0.98,
|
8678 |
+
"learning_rate": 0.0004228380315336837,
|
8679 |
+
"loss": 2.0913,
|
8680 |
+
"step": 1436
|
8681 |
+
},
|
8682 |
+
{
|
8683 |
+
"epoch": 0.98,
|
8684 |
+
"learning_rate": 0.00042269469660774,
|
8685 |
+
"loss": 2.1541,
|
8686 |
+
"step": 1437
|
8687 |
+
},
|
8688 |
+
{
|
8689 |
+
"epoch": 0.98,
|
8690 |
+
"learning_rate": 0.00042255136168179645,
|
8691 |
+
"loss": 2.1063,
|
8692 |
+
"step": 1438
|
8693 |
+
},
|
8694 |
+
{
|
8695 |
+
"epoch": 0.98,
|
8696 |
+
"learning_rate": 0.0004224080267558528,
|
8697 |
+
"loss": 2.0437,
|
8698 |
+
"step": 1439
|
8699 |
+
},
|
8700 |
+
{
|
8701 |
+
"epoch": 0.98,
|
8702 |
+
"learning_rate": 0.0004222646918299092,
|
8703 |
+
"loss": 2.1481,
|
8704 |
+
"step": 1440
|
8705 |
+
},
|
8706 |
+
{
|
8707 |
+
"epoch": 0.99,
|
8708 |
+
"learning_rate": 0.00042212135690396555,
|
8709 |
+
"loss": 2.2177,
|
8710 |
+
"step": 1441
|
8711 |
+
},
|
8712 |
+
{
|
8713 |
+
"epoch": 0.99,
|
8714 |
+
"learning_rate": 0.000421978021978022,
|
8715 |
+
"loss": 2.0707,
|
8716 |
+
"step": 1442
|
8717 |
+
},
|
8718 |
+
{
|
8719 |
+
"epoch": 0.99,
|
8720 |
+
"learning_rate": 0.0004218346870520783,
|
8721 |
+
"loss": 2.1417,
|
8722 |
+
"step": 1443
|
8723 |
+
},
|
8724 |
+
{
|
8725 |
+
"epoch": 0.99,
|
8726 |
+
"learning_rate": 0.0004216913521261347,
|
8727 |
+
"loss": 2.0127,
|
8728 |
+
"step": 1444
|
8729 |
+
},
|
8730 |
+
{
|
8731 |
+
"epoch": 0.99,
|
8732 |
+
"learning_rate": 0.0004215480172001911,
|
8733 |
+
"loss": 2.0886,
|
8734 |
+
"step": 1445
|
8735 |
+
},
|
8736 |
+
{
|
8737 |
+
"epoch": 0.99,
|
8738 |
+
"learning_rate": 0.00042140468227424747,
|
8739 |
+
"loss": 2.0717,
|
8740 |
+
"step": 1446
|
8741 |
+
},
|
8742 |
+
{
|
8743 |
+
"epoch": 0.99,
|
8744 |
+
"learning_rate": 0.00042126134734830385,
|
8745 |
+
"loss": 2.1458,
|
8746 |
+
"step": 1447
|
8747 |
+
},
|
8748 |
+
{
|
8749 |
+
"epoch": 0.99,
|
8750 |
+
"learning_rate": 0.0004211180124223602,
|
8751 |
+
"loss": 2.1917,
|
8752 |
+
"step": 1448
|
8753 |
+
},
|
8754 |
+
{
|
8755 |
+
"epoch": 0.99,
|
8756 |
+
"learning_rate": 0.00042097467749641656,
|
8757 |
+
"loss": 2.0842,
|
8758 |
+
"step": 1449
|
8759 |
+
},
|
8760 |
+
{
|
8761 |
+
"epoch": 0.99,
|
8762 |
+
"learning_rate": 0.00042083134257047295,
|
8763 |
+
"loss": 2.1187,
|
8764 |
+
"step": 1450
|
8765 |
+
},
|
8766 |
+
{
|
8767 |
+
"epoch": 0.99,
|
8768 |
+
"learning_rate": 0.00042068800764452933,
|
8769 |
+
"loss": 2.141,
|
8770 |
+
"step": 1451
|
8771 |
+
},
|
8772 |
+
{
|
8773 |
+
"epoch": 0.99,
|
8774 |
+
"learning_rate": 0.0004205446727185857,
|
8775 |
+
"loss": 2.079,
|
8776 |
+
"step": 1452
|
8777 |
+
},
|
8778 |
+
{
|
8779 |
+
"epoch": 0.99,
|
8780 |
+
"learning_rate": 0.0004204013377926421,
|
8781 |
+
"loss": 2.1057,
|
8782 |
+
"step": 1453
|
8783 |
+
},
|
8784 |
+
{
|
8785 |
+
"epoch": 0.99,
|
8786 |
+
"learning_rate": 0.0004202580028666985,
|
8787 |
+
"loss": 2.1162,
|
8788 |
+
"step": 1454
|
8789 |
+
},
|
8790 |
+
{
|
8791 |
+
"epoch": 0.99,
|
8792 |
+
"learning_rate": 0.00042011466794075487,
|
8793 |
+
"loss": 2.0668,
|
8794 |
+
"step": 1455
|
8795 |
+
},
|
8796 |
+
{
|
8797 |
+
"epoch": 1.0,
|
8798 |
+
"learning_rate": 0.0004199713330148112,
|
8799 |
+
"loss": 2.1383,
|
8800 |
+
"step": 1456
|
8801 |
+
},
|
8802 |
+
{
|
8803 |
+
"epoch": 1.0,
|
8804 |
+
"learning_rate": 0.00041982799808886763,
|
8805 |
+
"loss": 2.0792,
|
8806 |
+
"step": 1457
|
8807 |
+
},
|
8808 |
+
{
|
8809 |
+
"epoch": 1.0,
|
8810 |
+
"learning_rate": 0.00041968466316292396,
|
8811 |
+
"loss": 2.1121,
|
8812 |
+
"step": 1458
|
8813 |
+
},
|
8814 |
+
{
|
8815 |
+
"epoch": 1.0,
|
8816 |
+
"learning_rate": 0.0004195413282369804,
|
8817 |
+
"loss": 2.0756,
|
8818 |
+
"step": 1459
|
8819 |
+
},
|
8820 |
+
{
|
8821 |
+
"epoch": 1.0,
|
8822 |
+
"learning_rate": 0.00041939799331103673,
|
8823 |
+
"loss": 1.9645,
|
8824 |
+
"step": 1460
|
8825 |
+
},
|
8826 |
+
{
|
8827 |
+
"epoch": 1.0,
|
8828 |
+
"learning_rate": 0.00041925465838509317,
|
8829 |
+
"loss": 2.0766,
|
8830 |
+
"step": 1461
|
8831 |
+
},
|
8832 |
+
{
|
8833 |
+
"epoch": 1.0,
|
8834 |
+
"learning_rate": 0.0004191113234591495,
|
8835 |
+
"loss": 2.0359,
|
8836 |
+
"step": 1462
|
8837 |
+
},
|
8838 |
+
{
|
8839 |
+
"epoch": 1.0,
|
8840 |
+
"learning_rate": 0.0004189679885332059,
|
8841 |
+
"loss": 2.2025,
|
8842 |
+
"step": 1463
|
8843 |
+
},
|
8844 |
+
{
|
8845 |
+
"epoch": 1.0,
|
8846 |
+
"learning_rate": 0.00041882465360726227,
|
8847 |
+
"loss": 1.9701,
|
8848 |
+
"step": 1464
|
8849 |
+
},
|
8850 |
+
{
|
8851 |
+
"epoch": 1.0,
|
8852 |
+
"learning_rate": 0.0004186813186813186,
|
8853 |
+
"loss": 2.0596,
|
8854 |
+
"step": 1465
|
8855 |
+
},
|
8856 |
+
{
|
8857 |
+
"epoch": 1.0,
|
8858 |
+
"learning_rate": 0.00041853798375537503,
|
8859 |
+
"loss": 2.0145,
|
8860 |
+
"step": 1466
|
8861 |
+
},
|
8862 |
+
{
|
8863 |
+
"epoch": 1.0,
|
8864 |
+
"learning_rate": 0.00041839464882943136,
|
8865 |
+
"loss": 1.9526,
|
8866 |
+
"step": 1467
|
8867 |
+
},
|
8868 |
+
{
|
8869 |
+
"epoch": 1.0,
|
8870 |
+
"learning_rate": 0.0004182513139034878,
|
8871 |
+
"loss": 2.0803,
|
8872 |
+
"step": 1468
|
8873 |
+
},
|
8874 |
+
{
|
8875 |
+
"epoch": 1.0,
|
8876 |
+
"learning_rate": 0.00041810797897754413,
|
8877 |
+
"loss": 2.0068,
|
8878 |
+
"step": 1469
|
8879 |
+
},
|
8880 |
+
{
|
8881 |
+
"epoch": 1.01,
|
8882 |
+
"learning_rate": 0.0004179646440516005,
|
8883 |
+
"loss": 2.013,
|
8884 |
+
"step": 1470
|
8885 |
+
},
|
8886 |
+
{
|
8887 |
+
"epoch": 1.01,
|
8888 |
+
"learning_rate": 0.0004178213091256569,
|
8889 |
+
"loss": 2.0862,
|
8890 |
+
"step": 1471
|
8891 |
+
},
|
8892 |
+
{
|
8893 |
+
"epoch": 1.01,
|
8894 |
+
"learning_rate": 0.0004176779741997133,
|
8895 |
+
"loss": 1.8934,
|
8896 |
+
"step": 1472
|
8897 |
+
},
|
8898 |
+
{
|
8899 |
+
"epoch": 1.01,
|
8900 |
+
"learning_rate": 0.00041753463927376967,
|
8901 |
+
"loss": 2.0101,
|
8902 |
+
"step": 1473
|
8903 |
+
},
|
8904 |
+
{
|
8905 |
+
"epoch": 1.01,
|
8906 |
+
"learning_rate": 0.00041739130434782605,
|
8907 |
+
"loss": 2.0879,
|
8908 |
+
"step": 1474
|
8909 |
+
},
|
8910 |
+
{
|
8911 |
+
"epoch": 1.01,
|
8912 |
+
"learning_rate": 0.00041724796942188243,
|
8913 |
+
"loss": 2.0305,
|
8914 |
+
"step": 1475
|
8915 |
+
},
|
8916 |
+
{
|
8917 |
+
"epoch": 1.01,
|
8918 |
+
"learning_rate": 0.0004171046344959388,
|
8919 |
+
"loss": 1.9277,
|
8920 |
+
"step": 1476
|
8921 |
+
},
|
8922 |
+
{
|
8923 |
+
"epoch": 1.01,
|
8924 |
+
"learning_rate": 0.00041696129956999515,
|
8925 |
+
"loss": 2.0387,
|
8926 |
+
"step": 1477
|
8927 |
+
},
|
8928 |
+
{
|
8929 |
+
"epoch": 1.01,
|
8930 |
+
"learning_rate": 0.0004168179646440516,
|
8931 |
+
"loss": 2.0777,
|
8932 |
+
"step": 1478
|
8933 |
+
},
|
8934 |
+
{
|
8935 |
+
"epoch": 1.01,
|
8936 |
+
"learning_rate": 0.0004166746297181079,
|
8937 |
+
"loss": 2.0581,
|
8938 |
+
"step": 1479
|
8939 |
+
},
|
8940 |
+
{
|
8941 |
+
"epoch": 1.01,
|
8942 |
+
"learning_rate": 0.00041653129479216435,
|
8943 |
+
"loss": 1.9738,
|
8944 |
+
"step": 1480
|
8945 |
+
},
|
8946 |
+
{
|
8947 |
+
"epoch": 1.01,
|
8948 |
+
"learning_rate": 0.0004163879598662207,
|
8949 |
+
"loss": 2.0169,
|
8950 |
+
"step": 1481
|
8951 |
+
},
|
8952 |
+
{
|
8953 |
+
"epoch": 1.01,
|
8954 |
+
"learning_rate": 0.0004162446249402771,
|
8955 |
+
"loss": 1.9716,
|
8956 |
+
"step": 1482
|
8957 |
+
},
|
8958 |
+
{
|
8959 |
+
"epoch": 1.01,
|
8960 |
+
"learning_rate": 0.00041610129001433345,
|
8961 |
+
"loss": 2.0244,
|
8962 |
+
"step": 1483
|
8963 |
+
},
|
8964 |
+
{
|
8965 |
+
"epoch": 1.01,
|
8966 |
+
"learning_rate": 0.0004159579550883898,
|
8967 |
+
"loss": 2.0311,
|
8968 |
+
"step": 1484
|
8969 |
+
},
|
8970 |
+
{
|
8971 |
+
"epoch": 1.02,
|
8972 |
+
"learning_rate": 0.0004158146201624462,
|
8973 |
+
"loss": 2.0443,
|
8974 |
+
"step": 1485
|
8975 |
+
},
|
8976 |
+
{
|
8977 |
+
"epoch": 1.02,
|
8978 |
+
"learning_rate": 0.00041567128523650255,
|
8979 |
+
"loss": 1.9971,
|
8980 |
+
"step": 1486
|
8981 |
+
},
|
8982 |
+
{
|
8983 |
+
"epoch": 1.02,
|
8984 |
+
"learning_rate": 0.000415527950310559,
|
8985 |
+
"loss": 2.0571,
|
8986 |
+
"step": 1487
|
8987 |
+
},
|
8988 |
+
{
|
8989 |
+
"epoch": 1.02,
|
8990 |
+
"learning_rate": 0.0004153846153846153,
|
8991 |
+
"loss": 2.0578,
|
8992 |
+
"step": 1488
|
8993 |
+
},
|
8994 |
+
{
|
8995 |
+
"epoch": 1.02,
|
8996 |
+
"learning_rate": 0.00041524128045867175,
|
8997 |
+
"loss": 2.0293,
|
8998 |
+
"step": 1489
|
8999 |
+
},
|
9000 |
+
{
|
9001 |
+
"epoch": 1.02,
|
9002 |
+
"learning_rate": 0.0004150979455327281,
|
9003 |
+
"loss": 1.9722,
|
9004 |
+
"step": 1490
|
9005 |
+
},
|
9006 |
+
{
|
9007 |
+
"epoch": 1.02,
|
9008 |
+
"learning_rate": 0.00041495461060678447,
|
9009 |
+
"loss": 2.076,
|
9010 |
+
"step": 1491
|
9011 |
+
},
|
9012 |
+
{
|
9013 |
+
"epoch": 1.02,
|
9014 |
+
"learning_rate": 0.00041481127568084085,
|
9015 |
+
"loss": 2.0976,
|
9016 |
+
"step": 1492
|
9017 |
+
},
|
9018 |
+
{
|
9019 |
+
"epoch": 1.02,
|
9020 |
+
"learning_rate": 0.00041466794075489723,
|
9021 |
+
"loss": 2.0126,
|
9022 |
+
"step": 1493
|
9023 |
+
},
|
9024 |
+
{
|
9025 |
+
"epoch": 1.02,
|
9026 |
+
"learning_rate": 0.0004145246058289536,
|
9027 |
+
"loss": 1.9639,
|
9028 |
+
"step": 1494
|
9029 |
+
},
|
9030 |
+
{
|
9031 |
+
"epoch": 1.02,
|
9032 |
+
"learning_rate": 0.00041438127090301,
|
9033 |
+
"loss": 2.0662,
|
9034 |
+
"step": 1495
|
9035 |
+
},
|
9036 |
+
{
|
9037 |
+
"epoch": 1.02,
|
9038 |
+
"learning_rate": 0.0004142379359770664,
|
9039 |
+
"loss": 2.0507,
|
9040 |
+
"step": 1496
|
9041 |
+
},
|
9042 |
+
{
|
9043 |
+
"epoch": 1.02,
|
9044 |
+
"learning_rate": 0.00041409460105112277,
|
9045 |
+
"loss": 1.9897,
|
9046 |
+
"step": 1497
|
9047 |
+
},
|
9048 |
+
{
|
9049 |
+
"epoch": 1.02,
|
9050 |
+
"learning_rate": 0.0004139512661251791,
|
9051 |
+
"loss": 2.1243,
|
9052 |
+
"step": 1498
|
9053 |
+
},
|
9054 |
+
{
|
9055 |
+
"epoch": 1.02,
|
9056 |
+
"learning_rate": 0.00041380793119923554,
|
9057 |
+
"loss": 1.9646,
|
9058 |
+
"step": 1499
|
9059 |
+
},
|
9060 |
+
{
|
9061 |
+
"epoch": 1.03,
|
9062 |
+
"learning_rate": 0.00041366459627329187,
|
9063 |
+
"loss": 2.0035,
|
9064 |
+
"step": 1500
|
9065 |
+
},
|
9066 |
+
{
|
9067 |
+
"epoch": 1.03,
|
9068 |
+
"learning_rate": 0.0004135212613473483,
|
9069 |
+
"loss": 2.0319,
|
9070 |
+
"step": 1501
|
9071 |
+
},
|
9072 |
+
{
|
9073 |
+
"epoch": 1.03,
|
9074 |
+
"learning_rate": 0.00041337792642140463,
|
9075 |
+
"loss": 2.052,
|
9076 |
+
"step": 1502
|
9077 |
+
},
|
9078 |
+
{
|
9079 |
+
"epoch": 1.03,
|
9080 |
+
"learning_rate": 0.00041323459149546107,
|
9081 |
+
"loss": 2.0439,
|
9082 |
+
"step": 1503
|
9083 |
+
},
|
9084 |
+
{
|
9085 |
+
"epoch": 1.03,
|
9086 |
+
"learning_rate": 0.0004130912565695174,
|
9087 |
+
"loss": 2.0266,
|
9088 |
+
"step": 1504
|
9089 |
+
},
|
9090 |
+
{
|
9091 |
+
"epoch": 1.03,
|
9092 |
+
"learning_rate": 0.00041294792164357373,
|
9093 |
+
"loss": 2.0671,
|
9094 |
+
"step": 1505
|
9095 |
+
},
|
9096 |
+
{
|
9097 |
+
"epoch": 1.03,
|
9098 |
+
"learning_rate": 0.00041280458671763017,
|
9099 |
+
"loss": 1.9876,
|
9100 |
+
"step": 1506
|
9101 |
+
},
|
9102 |
+
{
|
9103 |
+
"epoch": 1.03,
|
9104 |
+
"learning_rate": 0.0004126612517916865,
|
9105 |
+
"loss": 2.0493,
|
9106 |
+
"step": 1507
|
9107 |
+
},
|
9108 |
+
{
|
9109 |
+
"epoch": 1.03,
|
9110 |
+
"learning_rate": 0.00041251791686574294,
|
9111 |
+
"loss": 2.1042,
|
9112 |
+
"step": 1508
|
9113 |
+
},
|
9114 |
+
{
|
9115 |
+
"epoch": 1.03,
|
9116 |
+
"learning_rate": 0.00041237458193979927,
|
9117 |
+
"loss": 1.9368,
|
9118 |
+
"step": 1509
|
9119 |
+
},
|
9120 |
+
{
|
9121 |
+
"epoch": 1.03,
|
9122 |
+
"learning_rate": 0.0004122312470138557,
|
9123 |
+
"loss": 2.0264,
|
9124 |
+
"step": 1510
|
9125 |
+
},
|
9126 |
+
{
|
9127 |
+
"epoch": 1.03,
|
9128 |
+
"learning_rate": 0.00041208791208791203,
|
9129 |
+
"loss": 2.0312,
|
9130 |
+
"step": 1511
|
9131 |
+
},
|
9132 |
+
{
|
9133 |
+
"epoch": 1.03,
|
9134 |
+
"learning_rate": 0.0004119445771619684,
|
9135 |
+
"loss": 2.0225,
|
9136 |
+
"step": 1512
|
9137 |
+
},
|
9138 |
+
{
|
9139 |
+
"epoch": 1.03,
|
9140 |
+
"learning_rate": 0.0004118012422360248,
|
9141 |
+
"loss": 2.0262,
|
9142 |
+
"step": 1513
|
9143 |
+
},
|
9144 |
+
{
|
9145 |
+
"epoch": 1.04,
|
9146 |
+
"learning_rate": 0.0004116579073100812,
|
9147 |
+
"loss": 2.0185,
|
9148 |
+
"step": 1514
|
9149 |
+
},
|
9150 |
+
{
|
9151 |
+
"epoch": 1.04,
|
9152 |
+
"learning_rate": 0.00041151457238413757,
|
9153 |
+
"loss": 1.9769,
|
9154 |
+
"step": 1515
|
9155 |
+
},
|
9156 |
+
{
|
9157 |
+
"epoch": 1.04,
|
9158 |
+
"learning_rate": 0.00041137123745819395,
|
9159 |
+
"loss": 2.0439,
|
9160 |
+
"step": 1516
|
9161 |
+
},
|
9162 |
+
{
|
9163 |
+
"epoch": 1.04,
|
9164 |
+
"learning_rate": 0.00041122790253225034,
|
9165 |
+
"loss": 2.0507,
|
9166 |
+
"step": 1517
|
9167 |
+
},
|
9168 |
+
{
|
9169 |
+
"epoch": 1.04,
|
9170 |
+
"learning_rate": 0.0004110845676063067,
|
9171 |
+
"loss": 2.0452,
|
9172 |
+
"step": 1518
|
9173 |
+
},
|
9174 |
+
{
|
9175 |
+
"epoch": 1.04,
|
9176 |
+
"learning_rate": 0.00041094123268036305,
|
9177 |
+
"loss": 2.1165,
|
9178 |
+
"step": 1519
|
9179 |
+
},
|
9180 |
+
{
|
9181 |
+
"epoch": 1.04,
|
9182 |
+
"learning_rate": 0.0004107978977544195,
|
9183 |
+
"loss": 1.9441,
|
9184 |
+
"step": 1520
|
9185 |
+
},
|
9186 |
+
{
|
9187 |
+
"epoch": 1.04,
|
9188 |
+
"learning_rate": 0.0004106545628284758,
|
9189 |
+
"loss": 1.9882,
|
9190 |
+
"step": 1521
|
9191 |
+
},
|
9192 |
+
{
|
9193 |
+
"epoch": 1.04,
|
9194 |
+
"learning_rate": 0.00041051122790253225,
|
9195 |
+
"loss": 2.0254,
|
9196 |
+
"step": 1522
|
9197 |
+
},
|
9198 |
+
{
|
9199 |
+
"epoch": 1.04,
|
9200 |
+
"learning_rate": 0.0004103678929765886,
|
9201 |
+
"loss": 1.978,
|
9202 |
+
"step": 1523
|
9203 |
+
},
|
9204 |
+
{
|
9205 |
+
"epoch": 1.04,
|
9206 |
+
"learning_rate": 0.000410224558050645,
|
9207 |
+
"loss": 1.9725,
|
9208 |
+
"step": 1524
|
9209 |
+
},
|
9210 |
+
{
|
9211 |
+
"epoch": 1.04,
|
9212 |
+
"learning_rate": 0.00041008122312470135,
|
9213 |
+
"loss": 1.9926,
|
9214 |
+
"step": 1525
|
9215 |
+
},
|
9216 |
+
{
|
9217 |
+
"epoch": 1.04,
|
9218 |
+
"learning_rate": 0.0004099378881987577,
|
9219 |
+
"loss": 1.9377,
|
9220 |
+
"step": 1526
|
9221 |
+
},
|
9222 |
+
{
|
9223 |
+
"epoch": 1.04,
|
9224 |
+
"learning_rate": 0.0004097945532728141,
|
9225 |
+
"loss": 1.9989,
|
9226 |
+
"step": 1527
|
9227 |
+
},
|
9228 |
+
{
|
9229 |
+
"epoch": 1.04,
|
9230 |
+
"learning_rate": 0.00040965121834687045,
|
9231 |
+
"loss": 1.999,
|
9232 |
+
"step": 1528
|
9233 |
+
},
|
9234 |
+
{
|
9235 |
+
"epoch": 1.05,
|
9236 |
+
"learning_rate": 0.0004095078834209269,
|
9237 |
+
"loss": 2.0974,
|
9238 |
+
"step": 1529
|
9239 |
+
},
|
9240 |
+
{
|
9241 |
+
"epoch": 1.05,
|
9242 |
+
"learning_rate": 0.0004093645484949832,
|
9243 |
+
"loss": 1.9753,
|
9244 |
+
"step": 1530
|
9245 |
+
},
|
9246 |
+
{
|
9247 |
+
"epoch": 1.05,
|
9248 |
+
"learning_rate": 0.00040922121356903965,
|
9249 |
+
"loss": 1.9776,
|
9250 |
+
"step": 1531
|
9251 |
+
},
|
9252 |
+
{
|
9253 |
+
"epoch": 1.05,
|
9254 |
+
"learning_rate": 0.000409077878643096,
|
9255 |
+
"loss": 2.004,
|
9256 |
+
"step": 1532
|
9257 |
+
},
|
9258 |
+
{
|
9259 |
+
"epoch": 1.05,
|
9260 |
+
"learning_rate": 0.00040893454371715237,
|
9261 |
+
"loss": 1.9964,
|
9262 |
+
"step": 1533
|
9263 |
+
},
|
9264 |
+
{
|
9265 |
+
"epoch": 1.05,
|
9266 |
+
"learning_rate": 0.00040879120879120875,
|
9267 |
+
"loss": 1.9928,
|
9268 |
+
"step": 1534
|
9269 |
+
},
|
9270 |
+
{
|
9271 |
+
"epoch": 1.05,
|
9272 |
+
"learning_rate": 0.00040864787386526514,
|
9273 |
+
"loss": 2.0106,
|
9274 |
+
"step": 1535
|
9275 |
+
},
|
9276 |
+
{
|
9277 |
+
"epoch": 1.05,
|
9278 |
+
"learning_rate": 0.0004085045389393215,
|
9279 |
+
"loss": 1.9374,
|
9280 |
+
"step": 1536
|
9281 |
+
},
|
9282 |
+
{
|
9283 |
+
"epoch": 1.05,
|
9284 |
+
"learning_rate": 0.0004083612040133779,
|
9285 |
+
"loss": 1.9205,
|
9286 |
+
"step": 1537
|
9287 |
+
},
|
9288 |
+
{
|
9289 |
+
"epoch": 1.05,
|
9290 |
+
"learning_rate": 0.0004082178690874343,
|
9291 |
+
"loss": 1.9928,
|
9292 |
+
"step": 1538
|
9293 |
+
},
|
9294 |
+
{
|
9295 |
+
"epoch": 1.05,
|
9296 |
+
"learning_rate": 0.00040807453416149067,
|
9297 |
+
"loss": 2.0717,
|
9298 |
+
"step": 1539
|
9299 |
+
},
|
9300 |
+
{
|
9301 |
+
"epoch": 1.05,
|
9302 |
+
"learning_rate": 0.000407931199235547,
|
9303 |
+
"loss": 2.0671,
|
9304 |
+
"step": 1540
|
9305 |
+
},
|
9306 |
+
{
|
9307 |
+
"epoch": 1.05,
|
9308 |
+
"learning_rate": 0.00040778786430960344,
|
9309 |
+
"loss": 1.955,
|
9310 |
+
"step": 1541
|
9311 |
+
},
|
9312 |
+
{
|
9313 |
+
"epoch": 1.05,
|
9314 |
+
"learning_rate": 0.00040764452938365977,
|
9315 |
+
"loss": 1.9376,
|
9316 |
+
"step": 1542
|
9317 |
+
},
|
9318 |
+
{
|
9319 |
+
"epoch": 1.05,
|
9320 |
+
"learning_rate": 0.0004075011944577162,
|
9321 |
+
"loss": 1.9598,
|
9322 |
+
"step": 1543
|
9323 |
+
},
|
9324 |
+
{
|
9325 |
+
"epoch": 1.06,
|
9326 |
+
"learning_rate": 0.00040735785953177254,
|
9327 |
+
"loss": 1.981,
|
9328 |
+
"step": 1544
|
9329 |
+
},
|
9330 |
+
{
|
9331 |
+
"epoch": 1.06,
|
9332 |
+
"learning_rate": 0.000407214524605829,
|
9333 |
+
"loss": 2.0193,
|
9334 |
+
"step": 1545
|
9335 |
+
},
|
9336 |
+
{
|
9337 |
+
"epoch": 1.06,
|
9338 |
+
"learning_rate": 0.0004070711896798853,
|
9339 |
+
"loss": 1.9919,
|
9340 |
+
"step": 1546
|
9341 |
+
},
|
9342 |
+
{
|
9343 |
+
"epoch": 1.06,
|
9344 |
+
"learning_rate": 0.00040692785475394163,
|
9345 |
+
"loss": 2.0287,
|
9346 |
+
"step": 1547
|
9347 |
+
},
|
9348 |
+
{
|
9349 |
+
"epoch": 1.06,
|
9350 |
+
"learning_rate": 0.00040678451982799807,
|
9351 |
+
"loss": 2.0366,
|
9352 |
+
"step": 1548
|
9353 |
+
},
|
9354 |
+
{
|
9355 |
+
"epoch": 1.06,
|
9356 |
+
"learning_rate": 0.0004066411849020544,
|
9357 |
+
"loss": 1.9564,
|
9358 |
+
"step": 1549
|
9359 |
+
},
|
9360 |
+
{
|
9361 |
+
"epoch": 1.06,
|
9362 |
+
"learning_rate": 0.00040649784997611084,
|
9363 |
+
"loss": 2.0538,
|
9364 |
+
"step": 1550
|
9365 |
+
},
|
9366 |
+
{
|
9367 |
+
"epoch": 1.06,
|
9368 |
+
"learning_rate": 0.00040635451505016717,
|
9369 |
+
"loss": 2.0049,
|
9370 |
+
"step": 1551
|
9371 |
+
},
|
9372 |
+
{
|
9373 |
+
"epoch": 1.06,
|
9374 |
+
"learning_rate": 0.00040621118012422355,
|
9375 |
+
"loss": 2.0463,
|
9376 |
+
"step": 1552
|
9377 |
+
},
|
9378 |
+
{
|
9379 |
+
"epoch": 1.06,
|
9380 |
+
"learning_rate": 0.00040606784519827994,
|
9381 |
+
"loss": 2.002,
|
9382 |
+
"step": 1553
|
9383 |
+
},
|
9384 |
+
{
|
9385 |
+
"epoch": 1.06,
|
9386 |
+
"learning_rate": 0.0004059245102723363,
|
9387 |
+
"loss": 2.0292,
|
9388 |
+
"step": 1554
|
9389 |
+
},
|
9390 |
+
{
|
9391 |
+
"epoch": 1.06,
|
9392 |
+
"learning_rate": 0.0004057811753463927,
|
9393 |
+
"loss": 2.0681,
|
9394 |
+
"step": 1555
|
9395 |
+
},
|
9396 |
+
{
|
9397 |
+
"epoch": 1.06,
|
9398 |
+
"learning_rate": 0.0004056378404204491,
|
9399 |
+
"loss": 2.0833,
|
9400 |
+
"step": 1556
|
9401 |
+
},
|
9402 |
+
{
|
9403 |
+
"epoch": 1.06,
|
9404 |
+
"learning_rate": 0.00040549450549450547,
|
9405 |
+
"loss": 2.0966,
|
9406 |
+
"step": 1557
|
9407 |
+
},
|
9408 |
+
{
|
9409 |
+
"epoch": 1.07,
|
9410 |
+
"learning_rate": 0.00040535117056856185,
|
9411 |
+
"loss": 1.9594,
|
9412 |
+
"step": 1558
|
9413 |
+
},
|
9414 |
+
{
|
9415 |
+
"epoch": 1.07,
|
9416 |
+
"learning_rate": 0.0004052078356426182,
|
9417 |
+
"loss": 1.9783,
|
9418 |
+
"step": 1559
|
9419 |
+
},
|
9420 |
+
{
|
9421 |
+
"epoch": 1.07,
|
9422 |
+
"learning_rate": 0.0004050645007166746,
|
9423 |
+
"loss": 2.0691,
|
9424 |
+
"step": 1560
|
9425 |
+
},
|
9426 |
+
{
|
9427 |
+
"epoch": 1.07,
|
9428 |
+
"learning_rate": 0.00040492116579073095,
|
9429 |
+
"loss": 2.0509,
|
9430 |
+
"step": 1561
|
9431 |
+
},
|
9432 |
+
{
|
9433 |
+
"epoch": 1.07,
|
9434 |
+
"learning_rate": 0.0004047778308647874,
|
9435 |
+
"loss": 2.0472,
|
9436 |
+
"step": 1562
|
9437 |
+
},
|
9438 |
+
{
|
9439 |
+
"epoch": 1.07,
|
9440 |
+
"learning_rate": 0.0004046344959388437,
|
9441 |
+
"loss": 1.9536,
|
9442 |
+
"step": 1563
|
9443 |
+
},
|
9444 |
+
{
|
9445 |
+
"epoch": 1.07,
|
9446 |
+
"learning_rate": 0.00040449116101290016,
|
9447 |
+
"loss": 2.0033,
|
9448 |
+
"step": 1564
|
9449 |
+
},
|
9450 |
+
{
|
9451 |
+
"epoch": 1.07,
|
9452 |
+
"learning_rate": 0.0004043478260869565,
|
9453 |
+
"loss": 2.0034,
|
9454 |
+
"step": 1565
|
9455 |
+
},
|
9456 |
+
{
|
9457 |
+
"epoch": 1.07,
|
9458 |
+
"learning_rate": 0.0004042044911610128,
|
9459 |
+
"loss": 1.9905,
|
9460 |
+
"step": 1566
|
9461 |
+
},
|
9462 |
+
{
|
9463 |
+
"epoch": 1.07,
|
9464 |
+
"learning_rate": 0.00040406115623506925,
|
9465 |
+
"loss": 2.0264,
|
9466 |
+
"step": 1567
|
9467 |
+
},
|
9468 |
+
{
|
9469 |
+
"epoch": 1.07,
|
9470 |
+
"learning_rate": 0.0004039178213091256,
|
9471 |
+
"loss": 1.982,
|
9472 |
+
"step": 1568
|
9473 |
+
},
|
9474 |
+
{
|
9475 |
+
"epoch": 1.07,
|
9476 |
+
"learning_rate": 0.000403774486383182,
|
9477 |
+
"loss": 2.0034,
|
9478 |
+
"step": 1569
|
9479 |
+
},
|
9480 |
+
{
|
9481 |
+
"epoch": 1.07,
|
9482 |
+
"learning_rate": 0.00040363115145723835,
|
9483 |
+
"loss": 1.994,
|
9484 |
+
"step": 1570
|
9485 |
+
},
|
9486 |
+
{
|
9487 |
+
"epoch": 1.07,
|
9488 |
+
"learning_rate": 0.0004034878165312948,
|
9489 |
+
"loss": 1.9319,
|
9490 |
+
"step": 1571
|
9491 |
+
},
|
9492 |
+
{
|
9493 |
+
"epoch": 1.07,
|
9494 |
+
"learning_rate": 0.0004033444816053511,
|
9495 |
+
"loss": 1.9808,
|
9496 |
+
"step": 1572
|
9497 |
+
},
|
9498 |
+
{
|
9499 |
+
"epoch": 1.08,
|
9500 |
+
"learning_rate": 0.0004032011466794075,
|
9501 |
+
"loss": 2.057,
|
9502 |
+
"step": 1573
|
9503 |
+
},
|
9504 |
+
{
|
9505 |
+
"epoch": 1.08,
|
9506 |
+
"learning_rate": 0.0004030578117534639,
|
9507 |
+
"loss": 1.9804,
|
9508 |
+
"step": 1574
|
9509 |
+
},
|
9510 |
+
{
|
9511 |
+
"epoch": 1.08,
|
9512 |
+
"learning_rate": 0.00040291447682752027,
|
9513 |
+
"loss": 1.9346,
|
9514 |
+
"step": 1575
|
9515 |
+
},
|
9516 |
+
{
|
9517 |
+
"epoch": 1.08,
|
9518 |
+
"learning_rate": 0.00040277114190157665,
|
9519 |
+
"loss": 1.9386,
|
9520 |
+
"step": 1576
|
9521 |
+
},
|
9522 |
+
{
|
9523 |
+
"epoch": 1.08,
|
9524 |
+
"learning_rate": 0.00040262780697563304,
|
9525 |
+
"loss": 2.0997,
|
9526 |
+
"step": 1577
|
9527 |
+
},
|
9528 |
+
{
|
9529 |
+
"epoch": 1.08,
|
9530 |
+
"learning_rate": 0.0004024844720496894,
|
9531 |
+
"loss": 1.9615,
|
9532 |
+
"step": 1578
|
9533 |
+
},
|
9534 |
+
{
|
9535 |
+
"epoch": 1.08,
|
9536 |
+
"learning_rate": 0.0004023411371237458,
|
9537 |
+
"loss": 2.0695,
|
9538 |
+
"step": 1579
|
9539 |
+
},
|
9540 |
+
{
|
9541 |
+
"epoch": 1.08,
|
9542 |
+
"learning_rate": 0.00040219780219780213,
|
9543 |
+
"loss": 2.0928,
|
9544 |
+
"step": 1580
|
9545 |
+
},
|
9546 |
+
{
|
9547 |
+
"epoch": 1.08,
|
9548 |
+
"learning_rate": 0.00040205446727185857,
|
9549 |
+
"loss": 1.9805,
|
9550 |
+
"step": 1581
|
9551 |
+
},
|
9552 |
+
{
|
9553 |
+
"epoch": 1.08,
|
9554 |
+
"learning_rate": 0.0004019111323459149,
|
9555 |
+
"loss": 1.9855,
|
9556 |
+
"step": 1582
|
9557 |
+
},
|
9558 |
+
{
|
9559 |
+
"epoch": 1.08,
|
9560 |
+
"learning_rate": 0.00040176779741997134,
|
9561 |
+
"loss": 2.0609,
|
9562 |
+
"step": 1583
|
9563 |
+
},
|
9564 |
+
{
|
9565 |
+
"epoch": 1.08,
|
9566 |
+
"learning_rate": 0.00040162446249402767,
|
9567 |
+
"loss": 2.0458,
|
9568 |
+
"step": 1584
|
9569 |
+
},
|
9570 |
+
{
|
9571 |
+
"epoch": 1.08,
|
9572 |
+
"learning_rate": 0.0004014811275680841,
|
9573 |
+
"loss": 1.9908,
|
9574 |
+
"step": 1585
|
9575 |
+
},
|
9576 |
+
{
|
9577 |
+
"epoch": 1.08,
|
9578 |
+
"learning_rate": 0.00040133779264214044,
|
9579 |
+
"loss": 2.1244,
|
9580 |
+
"step": 1586
|
9581 |
+
},
|
9582 |
+
{
|
9583 |
+
"epoch": 1.09,
|
9584 |
+
"learning_rate": 0.00040119445771619677,
|
9585 |
+
"loss": 2.0618,
|
9586 |
+
"step": 1587
|
9587 |
+
},
|
9588 |
+
{
|
9589 |
+
"epoch": 1.09,
|
9590 |
+
"learning_rate": 0.0004010511227902532,
|
9591 |
+
"loss": 1.9426,
|
9592 |
+
"step": 1588
|
9593 |
+
},
|
9594 |
+
{
|
9595 |
+
"epoch": 1.09,
|
9596 |
+
"learning_rate": 0.00040090778786430953,
|
9597 |
+
"loss": 2.0683,
|
9598 |
+
"step": 1589
|
9599 |
+
},
|
9600 |
+
{
|
9601 |
+
"epoch": 1.09,
|
9602 |
+
"learning_rate": 0.00040076445293836597,
|
9603 |
+
"loss": 1.9883,
|
9604 |
+
"step": 1590
|
9605 |
+
},
|
9606 |
+
{
|
9607 |
+
"epoch": 1.09,
|
9608 |
+
"learning_rate": 0.0004006211180124223,
|
9609 |
+
"loss": 2.081,
|
9610 |
+
"step": 1591
|
9611 |
+
},
|
9612 |
+
{
|
9613 |
+
"epoch": 1.09,
|
9614 |
+
"learning_rate": 0.00040047778308647874,
|
9615 |
+
"loss": 1.9766,
|
9616 |
+
"step": 1592
|
9617 |
+
},
|
9618 |
+
{
|
9619 |
+
"epoch": 1.09,
|
9620 |
+
"learning_rate": 0.00040033444816053507,
|
9621 |
+
"loss": 2.0358,
|
9622 |
+
"step": 1593
|
9623 |
+
},
|
9624 |
+
{
|
9625 |
+
"epoch": 1.09,
|
9626 |
+
"learning_rate": 0.00040019111323459145,
|
9627 |
+
"loss": 2.0776,
|
9628 |
+
"step": 1594
|
9629 |
+
},
|
9630 |
+
{
|
9631 |
+
"epoch": 1.09,
|
9632 |
+
"learning_rate": 0.00040004777830864784,
|
9633 |
+
"loss": 2.065,
|
9634 |
+
"step": 1595
|
9635 |
+
},
|
9636 |
+
{
|
9637 |
+
"epoch": 1.09,
|
9638 |
+
"learning_rate": 0.0003999044433827042,
|
9639 |
+
"loss": 2.0691,
|
9640 |
+
"step": 1596
|
9641 |
+
},
|
9642 |
+
{
|
9643 |
+
"epoch": 1.09,
|
9644 |
+
"learning_rate": 0.0003997611084567606,
|
9645 |
+
"loss": 2.1403,
|
9646 |
+
"step": 1597
|
9647 |
+
},
|
9648 |
+
{
|
9649 |
+
"epoch": 1.09,
|
9650 |
+
"learning_rate": 0.000399617773530817,
|
9651 |
+
"loss": 1.9704,
|
9652 |
+
"step": 1598
|
9653 |
+
},
|
9654 |
+
{
|
9655 |
+
"epoch": 1.09,
|
9656 |
+
"learning_rate": 0.00039947443860487337,
|
9657 |
+
"loss": 2.0766,
|
9658 |
+
"step": 1599
|
9659 |
+
},
|
9660 |
+
{
|
9661 |
+
"epoch": 1.09,
|
9662 |
+
"learning_rate": 0.00039933110367892976,
|
9663 |
+
"loss": 2.005,
|
9664 |
+
"step": 1600
|
9665 |
+
},
|
9666 |
+
{
|
9667 |
+
"epoch": 1.09,
|
9668 |
+
"eval_loss": 2.000626564025879,
|
9669 |
+
"eval_runtime": 1644.0479,
|
9670 |
+
"eval_samples_per_second": 6.083,
|
9671 |
+
"eval_steps_per_second": 6.083,
|
9672 |
+
"step": 1600
|
9673 |
}
|
9674 |
],
|
9675 |
"max_steps": 4386,
|
9676 |
"num_train_epochs": 3,
|
9677 |
+
"total_flos": 1.9592296256816364e+18,
|
9678 |
"trial_name": null,
|
9679 |
"trial_params": null
|
9680 |
}
|
pytorch_model.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 2368281769
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b2bace5feec1de1af12f9432ac0a85b72ad18e1f09933a275a13f26e34955c1b
|
3 |
size 2368281769
|