Training in progress, step 1600, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 73911112
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dd8edd4b23894593994dd1597dbc694ccc81001cb7faf035867a3339f3d6ca28
|
| 3 |
size 73911112
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 37965684
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e3fa9ed80396e8f083b3d374dcff89bc7c9037cfc4e5585d635d883ef34faa3f
|
| 3 |
size 37965684
|
last-checkpoint/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:e9ec4f7c75fc8dc73feec982dc49c1db4c92d75f32040d95e28ccfad97fcf3c4
|
| 3 |
size 14244
|
last-checkpoint/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 sha256:0a342d756ed2deadb7604676fd81f55043100e7cbb57f36d28b57fa6ef3c9fda
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": 0.04516534134745598,
|
| 3 |
"best_model_checkpoint": "miner_id_24/checkpoint-1400",
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 100,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -10635,6 +10635,714 @@
|
|
| 10635 |
"eval_samples_per_second": 29.681,
|
| 10636 |
"eval_steps_per_second": 7.42,
|
| 10637 |
"step": 1500
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 10638 |
}
|
| 10639 |
],
|
| 10640 |
"logging_steps": 1,
|
|
@@ -10649,7 +11357,7 @@
|
|
| 10649 |
"early_stopping_threshold": 0.0
|
| 10650 |
},
|
| 10651 |
"attributes": {
|
| 10652 |
-
"early_stopping_patience_counter":
|
| 10653 |
}
|
| 10654 |
},
|
| 10655 |
"TrainerControl": {
|
|
@@ -10658,12 +11366,12 @@
|
|
| 10658 |
"should_evaluate": false,
|
| 10659 |
"should_log": false,
|
| 10660 |
"should_save": true,
|
| 10661 |
-
"should_training_stop":
|
| 10662 |
},
|
| 10663 |
"attributes": {}
|
| 10664 |
}
|
| 10665 |
},
|
| 10666 |
-
"total_flos": 4.
|
| 10667 |
"train_batch_size": 4,
|
| 10668 |
"trial_name": null,
|
| 10669 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": 0.04516534134745598,
|
| 3 |
"best_model_checkpoint": "miner_id_24/checkpoint-1400",
|
| 4 |
+
"epoch": 0.13984944333366112,
|
| 5 |
"eval_steps": 100,
|
| 6 |
+
"global_step": 1600,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 10635 |
"eval_samples_per_second": 29.681,
|
| 10636 |
"eval_steps_per_second": 7.42,
|
| 10637 |
"step": 1500
|
| 10638 |
+
},
|
| 10639 |
+
{
|
| 10640 |
+
"epoch": 0.13119625902739082,
|
| 10641 |
+
"grad_norm": 0.08697141706943512,
|
| 10642 |
+
"learning_rate": 0.00010350138745749725,
|
| 10643 |
+
"loss": 0.0364,
|
| 10644 |
+
"step": 1501
|
| 10645 |
+
},
|
| 10646 |
+
{
|
| 10647 |
+
"epoch": 0.13128366492947435,
|
| 10648 |
+
"grad_norm": 0.11794944107532501,
|
| 10649 |
+
"learning_rate": 0.00010339844573921038,
|
| 10650 |
+
"loss": 0.0406,
|
| 10651 |
+
"step": 1502
|
| 10652 |
+
},
|
| 10653 |
+
{
|
| 10654 |
+
"epoch": 0.1313710708315579,
|
| 10655 |
+
"grad_norm": 0.20966710150241852,
|
| 10656 |
+
"learning_rate": 0.00010329550041530005,
|
| 10657 |
+
"loss": 0.0717,
|
| 10658 |
+
"step": 1503
|
| 10659 |
+
},
|
| 10660 |
+
{
|
| 10661 |
+
"epoch": 0.13145847673364144,
|
| 10662 |
+
"grad_norm": 0.15292756259441376,
|
| 10663 |
+
"learning_rate": 0.0001031925515949874,
|
| 10664 |
+
"loss": 0.0329,
|
| 10665 |
+
"step": 1504
|
| 10666 |
+
},
|
| 10667 |
+
{
|
| 10668 |
+
"epoch": 0.13154588263572498,
|
| 10669 |
+
"grad_norm": 0.06401754915714264,
|
| 10670 |
+
"learning_rate": 0.00010308959938749729,
|
| 10671 |
+
"loss": 0.024,
|
| 10672 |
+
"step": 1505
|
| 10673 |
+
},
|
| 10674 |
+
{
|
| 10675 |
+
"epoch": 0.1316332885378085,
|
| 10676 |
+
"grad_norm": 0.0975053682923317,
|
| 10677 |
+
"learning_rate": 0.00010298664390205812,
|
| 10678 |
+
"loss": 0.0453,
|
| 10679 |
+
"step": 1506
|
| 10680 |
+
},
|
| 10681 |
+
{
|
| 10682 |
+
"epoch": 0.13172069443989207,
|
| 10683 |
+
"grad_norm": 0.07814626395702362,
|
| 10684 |
+
"learning_rate": 0.00010288368524790182,
|
| 10685 |
+
"loss": 0.029,
|
| 10686 |
+
"step": 1507
|
| 10687 |
+
},
|
| 10688 |
+
{
|
| 10689 |
+
"epoch": 0.1318081003419756,
|
| 10690 |
+
"grad_norm": 0.08279038220643997,
|
| 10691 |
+
"learning_rate": 0.00010278072353426365,
|
| 10692 |
+
"loss": 0.0267,
|
| 10693 |
+
"step": 1508
|
| 10694 |
+
},
|
| 10695 |
+
{
|
| 10696 |
+
"epoch": 0.13189550624405913,
|
| 10697 |
+
"grad_norm": 0.10478601604700089,
|
| 10698 |
+
"learning_rate": 0.0001026777588703821,
|
| 10699 |
+
"loss": 0.0389,
|
| 10700 |
+
"step": 1509
|
| 10701 |
+
},
|
| 10702 |
+
{
|
| 10703 |
+
"epoch": 0.13198291214614266,
|
| 10704 |
+
"grad_norm": 0.0825965404510498,
|
| 10705 |
+
"learning_rate": 0.00010257479136549889,
|
| 10706 |
+
"loss": 0.0278,
|
| 10707 |
+
"step": 1510
|
| 10708 |
+
},
|
| 10709 |
+
{
|
| 10710 |
+
"epoch": 0.1320703180482262,
|
| 10711 |
+
"grad_norm": 0.09572532027959824,
|
| 10712 |
+
"learning_rate": 0.00010247182112885861,
|
| 10713 |
+
"loss": 0.0352,
|
| 10714 |
+
"step": 1511
|
| 10715 |
+
},
|
| 10716 |
+
{
|
| 10717 |
+
"epoch": 0.13215772395030975,
|
| 10718 |
+
"grad_norm": 0.06361362338066101,
|
| 10719 |
+
"learning_rate": 0.00010236884826970878,
|
| 10720 |
+
"loss": 0.0197,
|
| 10721 |
+
"step": 1512
|
| 10722 |
+
},
|
| 10723 |
+
{
|
| 10724 |
+
"epoch": 0.13224512985239328,
|
| 10725 |
+
"grad_norm": 0.08282926678657532,
|
| 10726 |
+
"learning_rate": 0.00010226587289729988,
|
| 10727 |
+
"loss": 0.0324,
|
| 10728 |
+
"step": 1513
|
| 10729 |
+
},
|
| 10730 |
+
{
|
| 10731 |
+
"epoch": 0.13233253575447682,
|
| 10732 |
+
"grad_norm": 0.13065838813781738,
|
| 10733 |
+
"learning_rate": 0.00010216289512088479,
|
| 10734 |
+
"loss": 0.0547,
|
| 10735 |
+
"step": 1514
|
| 10736 |
+
},
|
| 10737 |
+
{
|
| 10738 |
+
"epoch": 0.13241994165656035,
|
| 10739 |
+
"grad_norm": 0.060922764241695404,
|
| 10740 |
+
"learning_rate": 0.00010205991504971912,
|
| 10741 |
+
"loss": 0.0205,
|
| 10742 |
+
"step": 1515
|
| 10743 |
+
},
|
| 10744 |
+
{
|
| 10745 |
+
"epoch": 0.1325073475586439,
|
| 10746 |
+
"grad_norm": 0.07022725045681,
|
| 10747 |
+
"learning_rate": 0.00010195693279306089,
|
| 10748 |
+
"loss": 0.0221,
|
| 10749 |
+
"step": 1516
|
| 10750 |
+
},
|
| 10751 |
+
{
|
| 10752 |
+
"epoch": 0.13259475346072744,
|
| 10753 |
+
"grad_norm": 0.10593367367982864,
|
| 10754 |
+
"learning_rate": 0.00010185394846017036,
|
| 10755 |
+
"loss": 0.036,
|
| 10756 |
+
"step": 1517
|
| 10757 |
+
},
|
| 10758 |
+
{
|
| 10759 |
+
"epoch": 0.13268215936281097,
|
| 10760 |
+
"grad_norm": 0.07627231627702713,
|
| 10761 |
+
"learning_rate": 0.00010175096216031006,
|
| 10762 |
+
"loss": 0.0322,
|
| 10763 |
+
"step": 1518
|
| 10764 |
+
},
|
| 10765 |
+
{
|
| 10766 |
+
"epoch": 0.1327695652648945,
|
| 10767 |
+
"grad_norm": 0.05789932608604431,
|
| 10768 |
+
"learning_rate": 0.0001016479740027446,
|
| 10769 |
+
"loss": 0.0289,
|
| 10770 |
+
"step": 1519
|
| 10771 |
+
},
|
| 10772 |
+
{
|
| 10773 |
+
"epoch": 0.13285697116697806,
|
| 10774 |
+
"grad_norm": 0.09948685020208359,
|
| 10775 |
+
"learning_rate": 0.00010154498409674051,
|
| 10776 |
+
"loss": 0.0331,
|
| 10777 |
+
"step": 1520
|
| 10778 |
+
},
|
| 10779 |
+
{
|
| 10780 |
+
"epoch": 0.1329443770690616,
|
| 10781 |
+
"grad_norm": 0.07831980288028717,
|
| 10782 |
+
"learning_rate": 0.00010144199255156631,
|
| 10783 |
+
"loss": 0.0306,
|
| 10784 |
+
"step": 1521
|
| 10785 |
+
},
|
| 10786 |
+
{
|
| 10787 |
+
"epoch": 0.13303178297114512,
|
| 10788 |
+
"grad_norm": 0.10856369137763977,
|
| 10789 |
+
"learning_rate": 0.00010133899947649209,
|
| 10790 |
+
"loss": 0.0355,
|
| 10791 |
+
"step": 1522
|
| 10792 |
+
},
|
| 10793 |
+
{
|
| 10794 |
+
"epoch": 0.13311918887322866,
|
| 10795 |
+
"grad_norm": 0.16798873245716095,
|
| 10796 |
+
"learning_rate": 0.0001012360049807896,
|
| 10797 |
+
"loss": 0.0508,
|
| 10798 |
+
"step": 1523
|
| 10799 |
+
},
|
| 10800 |
+
{
|
| 10801 |
+
"epoch": 0.13320659477531221,
|
| 10802 |
+
"grad_norm": 0.11936990171670914,
|
| 10803 |
+
"learning_rate": 0.00010113300917373226,
|
| 10804 |
+
"loss": 0.0575,
|
| 10805 |
+
"step": 1524
|
| 10806 |
+
},
|
| 10807 |
+
{
|
| 10808 |
+
"epoch": 0.13329400067739575,
|
| 10809 |
+
"grad_norm": 0.09081139415502548,
|
| 10810 |
+
"learning_rate": 0.00010103001216459469,
|
| 10811 |
+
"loss": 0.0288,
|
| 10812 |
+
"step": 1525
|
| 10813 |
+
},
|
| 10814 |
+
{
|
| 10815 |
+
"epoch": 0.13338140657947928,
|
| 10816 |
+
"grad_norm": 0.08312739431858063,
|
| 10817 |
+
"learning_rate": 0.0001009270140626528,
|
| 10818 |
+
"loss": 0.0196,
|
| 10819 |
+
"step": 1526
|
| 10820 |
+
},
|
| 10821 |
+
{
|
| 10822 |
+
"epoch": 0.1334688124815628,
|
| 10823 |
+
"grad_norm": 0.08712684363126755,
|
| 10824 |
+
"learning_rate": 0.0001008240149771838,
|
| 10825 |
+
"loss": 0.0302,
|
| 10826 |
+
"step": 1527
|
| 10827 |
+
},
|
| 10828 |
+
{
|
| 10829 |
+
"epoch": 0.13355621838364637,
|
| 10830 |
+
"grad_norm": 0.17785540223121643,
|
| 10831 |
+
"learning_rate": 0.0001007210150174658,
|
| 10832 |
+
"loss": 0.0321,
|
| 10833 |
+
"step": 1528
|
| 10834 |
+
},
|
| 10835 |
+
{
|
| 10836 |
+
"epoch": 0.1336436242857299,
|
| 10837 |
+
"grad_norm": 0.119221031665802,
|
| 10838 |
+
"learning_rate": 0.00010061801429277796,
|
| 10839 |
+
"loss": 0.0481,
|
| 10840 |
+
"step": 1529
|
| 10841 |
+
},
|
| 10842 |
+
{
|
| 10843 |
+
"epoch": 0.13373103018781343,
|
| 10844 |
+
"grad_norm": 0.04911545664072037,
|
| 10845 |
+
"learning_rate": 0.00010051501291240008,
|
| 10846 |
+
"loss": 0.0181,
|
| 10847 |
+
"step": 1530
|
| 10848 |
+
},
|
| 10849 |
+
{
|
| 10850 |
+
"epoch": 0.13381843608989696,
|
| 10851 |
+
"grad_norm": 0.05027123540639877,
|
| 10852 |
+
"learning_rate": 0.00010041201098561286,
|
| 10853 |
+
"loss": 0.0183,
|
| 10854 |
+
"step": 1531
|
| 10855 |
+
},
|
| 10856 |
+
{
|
| 10857 |
+
"epoch": 0.1339058419919805,
|
| 10858 |
+
"grad_norm": 0.13207297027111053,
|
| 10859 |
+
"learning_rate": 0.00010030900862169744,
|
| 10860 |
+
"loss": 0.0185,
|
| 10861 |
+
"step": 1532
|
| 10862 |
+
},
|
| 10863 |
+
{
|
| 10864 |
+
"epoch": 0.13399324789406405,
|
| 10865 |
+
"grad_norm": 0.05222494527697563,
|
| 10866 |
+
"learning_rate": 0.00010020600592993548,
|
| 10867 |
+
"loss": 0.0165,
|
| 10868 |
+
"step": 1533
|
| 10869 |
+
},
|
| 10870 |
+
{
|
| 10871 |
+
"epoch": 0.1340806537961476,
|
| 10872 |
+
"grad_norm": 0.07455245405435562,
|
| 10873 |
+
"learning_rate": 0.00010010300301960888,
|
| 10874 |
+
"loss": 0.0299,
|
| 10875 |
+
"step": 1534
|
| 10876 |
+
},
|
| 10877 |
+
{
|
| 10878 |
+
"epoch": 0.13416805969823112,
|
| 10879 |
+
"grad_norm": 0.07873023301362991,
|
| 10880 |
+
"learning_rate": 0.0001,
|
| 10881 |
+
"loss": 0.0283,
|
| 10882 |
+
"step": 1535
|
| 10883 |
+
},
|
| 10884 |
+
{
|
| 10885 |
+
"epoch": 0.13425546560031465,
|
| 10886 |
+
"grad_norm": 0.07464205473661423,
|
| 10887 |
+
"learning_rate": 9.989699698039111e-05,
|
| 10888 |
+
"loss": 0.0312,
|
| 10889 |
+
"step": 1536
|
| 10890 |
+
},
|
| 10891 |
+
{
|
| 10892 |
+
"epoch": 0.1343428715023982,
|
| 10893 |
+
"grad_norm": 0.08801527321338654,
|
| 10894 |
+
"learning_rate": 9.979399407006458e-05,
|
| 10895 |
+
"loss": 0.0233,
|
| 10896 |
+
"step": 1537
|
| 10897 |
+
},
|
| 10898 |
+
{
|
| 10899 |
+
"epoch": 0.13443027740448174,
|
| 10900 |
+
"grad_norm": 0.104781374335289,
|
| 10901 |
+
"learning_rate": 9.969099137830259e-05,
|
| 10902 |
+
"loss": 0.0353,
|
| 10903 |
+
"step": 1538
|
| 10904 |
+
},
|
| 10905 |
+
{
|
| 10906 |
+
"epoch": 0.13451768330656527,
|
| 10907 |
+
"grad_norm": 0.0791974663734436,
|
| 10908 |
+
"learning_rate": 9.958798901438715e-05,
|
| 10909 |
+
"loss": 0.0283,
|
| 10910 |
+
"step": 1539
|
| 10911 |
+
},
|
| 10912 |
+
{
|
| 10913 |
+
"epoch": 0.1346050892086488,
|
| 10914 |
+
"grad_norm": 0.060504913330078125,
|
| 10915 |
+
"learning_rate": 9.948498708759993e-05,
|
| 10916 |
+
"loss": 0.0192,
|
| 10917 |
+
"step": 1540
|
| 10918 |
+
},
|
| 10919 |
+
{
|
| 10920 |
+
"epoch": 0.13469249511073236,
|
| 10921 |
+
"grad_norm": 0.04931584745645523,
|
| 10922 |
+
"learning_rate": 9.938198570722207e-05,
|
| 10923 |
+
"loss": 0.0225,
|
| 10924 |
+
"step": 1541
|
| 10925 |
+
},
|
| 10926 |
+
{
|
| 10927 |
+
"epoch": 0.1347799010128159,
|
| 10928 |
+
"grad_norm": 0.10144636034965515,
|
| 10929 |
+
"learning_rate": 9.927898498253422e-05,
|
| 10930 |
+
"loss": 0.0259,
|
| 10931 |
+
"step": 1542
|
| 10932 |
+
},
|
| 10933 |
+
{
|
| 10934 |
+
"epoch": 0.13486730691489943,
|
| 10935 |
+
"grad_norm": 0.06569879502058029,
|
| 10936 |
+
"learning_rate": 9.917598502281621e-05,
|
| 10937 |
+
"loss": 0.0258,
|
| 10938 |
+
"step": 1543
|
| 10939 |
+
},
|
| 10940 |
+
{
|
| 10941 |
+
"epoch": 0.13495471281698296,
|
| 10942 |
+
"grad_norm": 0.06734506040811539,
|
| 10943 |
+
"learning_rate": 9.90729859373472e-05,
|
| 10944 |
+
"loss": 0.0259,
|
| 10945 |
+
"step": 1544
|
| 10946 |
+
},
|
| 10947 |
+
{
|
| 10948 |
+
"epoch": 0.13504211871906652,
|
| 10949 |
+
"grad_norm": 0.07036790996789932,
|
| 10950 |
+
"learning_rate": 9.896998783540536e-05,
|
| 10951 |
+
"loss": 0.0289,
|
| 10952 |
+
"step": 1545
|
| 10953 |
+
},
|
| 10954 |
+
{
|
| 10955 |
+
"epoch": 0.13512952462115005,
|
| 10956 |
+
"grad_norm": 0.05877295881509781,
|
| 10957 |
+
"learning_rate": 9.886699082626775e-05,
|
| 10958 |
+
"loss": 0.0193,
|
| 10959 |
+
"step": 1546
|
| 10960 |
+
},
|
| 10961 |
+
{
|
| 10962 |
+
"epoch": 0.13521693052323358,
|
| 10963 |
+
"grad_norm": 0.0712662935256958,
|
| 10964 |
+
"learning_rate": 9.87639950192104e-05,
|
| 10965 |
+
"loss": 0.0256,
|
| 10966 |
+
"step": 1547
|
| 10967 |
+
},
|
| 10968 |
+
{
|
| 10969 |
+
"epoch": 0.1353043364253171,
|
| 10970 |
+
"grad_norm": 0.10854820907115936,
|
| 10971 |
+
"learning_rate": 9.866100052350796e-05,
|
| 10972 |
+
"loss": 0.0423,
|
| 10973 |
+
"step": 1548
|
| 10974 |
+
},
|
| 10975 |
+
{
|
| 10976 |
+
"epoch": 0.13539174232740067,
|
| 10977 |
+
"grad_norm": 0.10757939517498016,
|
| 10978 |
+
"learning_rate": 9.855800744843372e-05,
|
| 10979 |
+
"loss": 0.0378,
|
| 10980 |
+
"step": 1549
|
| 10981 |
+
},
|
| 10982 |
+
{
|
| 10983 |
+
"epoch": 0.1354791482294842,
|
| 10984 |
+
"grad_norm": 0.07478786259889603,
|
| 10985 |
+
"learning_rate": 9.845501590325948e-05,
|
| 10986 |
+
"loss": 0.0251,
|
| 10987 |
+
"step": 1550
|
| 10988 |
+
},
|
| 10989 |
+
{
|
| 10990 |
+
"epoch": 0.13556655413156773,
|
| 10991 |
+
"grad_norm": 0.08897072076797485,
|
| 10992 |
+
"learning_rate": 9.835202599725544e-05,
|
| 10993 |
+
"loss": 0.027,
|
| 10994 |
+
"step": 1551
|
| 10995 |
+
},
|
| 10996 |
+
{
|
| 10997 |
+
"epoch": 0.13565396003365127,
|
| 10998 |
+
"grad_norm": 0.07537046819925308,
|
| 10999 |
+
"learning_rate": 9.824903783968996e-05,
|
| 11000 |
+
"loss": 0.0223,
|
| 11001 |
+
"step": 1552
|
| 11002 |
+
},
|
| 11003 |
+
{
|
| 11004 |
+
"epoch": 0.1357413659357348,
|
| 11005 |
+
"grad_norm": 0.0578019842505455,
|
| 11006 |
+
"learning_rate": 9.814605153982967e-05,
|
| 11007 |
+
"loss": 0.0197,
|
| 11008 |
+
"step": 1553
|
| 11009 |
+
},
|
| 11010 |
+
{
|
| 11011 |
+
"epoch": 0.13582877183781836,
|
| 11012 |
+
"grad_norm": 0.0902201235294342,
|
| 11013 |
+
"learning_rate": 9.804306720693913e-05,
|
| 11014 |
+
"loss": 0.0264,
|
| 11015 |
+
"step": 1554
|
| 11016 |
+
},
|
| 11017 |
+
{
|
| 11018 |
+
"epoch": 0.1359161777399019,
|
| 11019 |
+
"grad_norm": 0.1797989010810852,
|
| 11020 |
+
"learning_rate": 9.794008495028087e-05,
|
| 11021 |
+
"loss": 0.0301,
|
| 11022 |
+
"step": 1555
|
| 11023 |
+
},
|
| 11024 |
+
{
|
| 11025 |
+
"epoch": 0.13600358364198542,
|
| 11026 |
+
"grad_norm": 0.1257629692554474,
|
| 11027 |
+
"learning_rate": 9.783710487911523e-05,
|
| 11028 |
+
"loss": 0.0292,
|
| 11029 |
+
"step": 1556
|
| 11030 |
+
},
|
| 11031 |
+
{
|
| 11032 |
+
"epoch": 0.13609098954406895,
|
| 11033 |
+
"grad_norm": 0.0894462838768959,
|
| 11034 |
+
"learning_rate": 9.773412710270016e-05,
|
| 11035 |
+
"loss": 0.0302,
|
| 11036 |
+
"step": 1557
|
| 11037 |
+
},
|
| 11038 |
+
{
|
| 11039 |
+
"epoch": 0.1361783954461525,
|
| 11040 |
+
"grad_norm": 0.1309974491596222,
|
| 11041 |
+
"learning_rate": 9.763115173029121e-05,
|
| 11042 |
+
"loss": 0.038,
|
| 11043 |
+
"step": 1558
|
| 11044 |
+
},
|
| 11045 |
+
{
|
| 11046 |
+
"epoch": 0.13626580134823604,
|
| 11047 |
+
"grad_norm": 0.15583781898021698,
|
| 11048 |
+
"learning_rate": 9.752817887114146e-05,
|
| 11049 |
+
"loss": 0.0389,
|
| 11050 |
+
"step": 1559
|
| 11051 |
+
},
|
| 11052 |
+
{
|
| 11053 |
+
"epoch": 0.13635320725031957,
|
| 11054 |
+
"grad_norm": 0.055186927318573,
|
| 11055 |
+
"learning_rate": 9.742520863450115e-05,
|
| 11056 |
+
"loss": 0.0215,
|
| 11057 |
+
"step": 1560
|
| 11058 |
+
},
|
| 11059 |
+
{
|
| 11060 |
+
"epoch": 0.1364406131524031,
|
| 11061 |
+
"grad_norm": 0.06592228263616562,
|
| 11062 |
+
"learning_rate": 9.73222411296179e-05,
|
| 11063 |
+
"loss": 0.0205,
|
| 11064 |
+
"step": 1561
|
| 11065 |
+
},
|
| 11066 |
+
{
|
| 11067 |
+
"epoch": 0.13652801905448667,
|
| 11068 |
+
"grad_norm": 0.16731403768062592,
|
| 11069 |
+
"learning_rate": 9.721927646573639e-05,
|
| 11070 |
+
"loss": 0.0476,
|
| 11071 |
+
"step": 1562
|
| 11072 |
+
},
|
| 11073 |
+
{
|
| 11074 |
+
"epoch": 0.1366154249565702,
|
| 11075 |
+
"grad_norm": 0.07706723362207413,
|
| 11076 |
+
"learning_rate": 9.71163147520982e-05,
|
| 11077 |
+
"loss": 0.0264,
|
| 11078 |
+
"step": 1563
|
| 11079 |
+
},
|
| 11080 |
+
{
|
| 11081 |
+
"epoch": 0.13670283085865373,
|
| 11082 |
+
"grad_norm": 0.0804000049829483,
|
| 11083 |
+
"learning_rate": 9.70133560979419e-05,
|
| 11084 |
+
"loss": 0.0284,
|
| 11085 |
+
"step": 1564
|
| 11086 |
+
},
|
| 11087 |
+
{
|
| 11088 |
+
"epoch": 0.13679023676073726,
|
| 11089 |
+
"grad_norm": 0.1061941385269165,
|
| 11090 |
+
"learning_rate": 9.691040061250273e-05,
|
| 11091 |
+
"loss": 0.0333,
|
| 11092 |
+
"step": 1565
|
| 11093 |
+
},
|
| 11094 |
+
{
|
| 11095 |
+
"epoch": 0.13687764266282082,
|
| 11096 |
+
"grad_norm": 0.06735522300004959,
|
| 11097 |
+
"learning_rate": 9.680744840501261e-05,
|
| 11098 |
+
"loss": 0.0242,
|
| 11099 |
+
"step": 1566
|
| 11100 |
+
},
|
| 11101 |
+
{
|
| 11102 |
+
"epoch": 0.13696504856490435,
|
| 11103 |
+
"grad_norm": 0.16169671714305878,
|
| 11104 |
+
"learning_rate": 9.670449958469999e-05,
|
| 11105 |
+
"loss": 0.0599,
|
| 11106 |
+
"step": 1567
|
| 11107 |
+
},
|
| 11108 |
+
{
|
| 11109 |
+
"epoch": 0.13705245446698788,
|
| 11110 |
+
"grad_norm": 0.054925642907619476,
|
| 11111 |
+
"learning_rate": 9.660155426078964e-05,
|
| 11112 |
+
"loss": 0.0177,
|
| 11113 |
+
"step": 1568
|
| 11114 |
+
},
|
| 11115 |
+
{
|
| 11116 |
+
"epoch": 0.13713986036907141,
|
| 11117 |
+
"grad_norm": 0.10755883902311325,
|
| 11118 |
+
"learning_rate": 9.649861254250275e-05,
|
| 11119 |
+
"loss": 0.0349,
|
| 11120 |
+
"step": 1569
|
| 11121 |
+
},
|
| 11122 |
+
{
|
| 11123 |
+
"epoch": 0.13722726627115495,
|
| 11124 |
+
"grad_norm": 0.06160590425133705,
|
| 11125 |
+
"learning_rate": 9.639567453905661e-05,
|
| 11126 |
+
"loss": 0.024,
|
| 11127 |
+
"step": 1570
|
| 11128 |
+
},
|
| 11129 |
+
{
|
| 11130 |
+
"epoch": 0.1373146721732385,
|
| 11131 |
+
"grad_norm": 0.20590682327747345,
|
| 11132 |
+
"learning_rate": 9.629274035966457e-05,
|
| 11133 |
+
"loss": 0.0551,
|
| 11134 |
+
"step": 1571
|
| 11135 |
+
},
|
| 11136 |
+
{
|
| 11137 |
+
"epoch": 0.13740207807532204,
|
| 11138 |
+
"grad_norm": 0.1019708439707756,
|
| 11139 |
+
"learning_rate": 9.618981011353592e-05,
|
| 11140 |
+
"loss": 0.0335,
|
| 11141 |
+
"step": 1572
|
| 11142 |
+
},
|
| 11143 |
+
{
|
| 11144 |
+
"epoch": 0.13748948397740557,
|
| 11145 |
+
"grad_norm": 0.06255567073822021,
|
| 11146 |
+
"learning_rate": 9.608688390987584e-05,
|
| 11147 |
+
"loss": 0.0234,
|
| 11148 |
+
"step": 1573
|
| 11149 |
+
},
|
| 11150 |
+
{
|
| 11151 |
+
"epoch": 0.1375768898794891,
|
| 11152 |
+
"grad_norm": 0.10318184643983841,
|
| 11153 |
+
"learning_rate": 9.59839618578851e-05,
|
| 11154 |
+
"loss": 0.039,
|
| 11155 |
+
"step": 1574
|
| 11156 |
+
},
|
| 11157 |
+
{
|
| 11158 |
+
"epoch": 0.13766429578157266,
|
| 11159 |
+
"grad_norm": 0.1166694313287735,
|
| 11160 |
+
"learning_rate": 9.58810440667602e-05,
|
| 11161 |
+
"loss": 0.0457,
|
| 11162 |
+
"step": 1575
|
| 11163 |
+
},
|
| 11164 |
+
{
|
| 11165 |
+
"epoch": 0.1377517016836562,
|
| 11166 |
+
"grad_norm": 0.1220555454492569,
|
| 11167 |
+
"learning_rate": 9.577813064569301e-05,
|
| 11168 |
+
"loss": 0.0469,
|
| 11169 |
+
"step": 1576
|
| 11170 |
+
},
|
| 11171 |
+
{
|
| 11172 |
+
"epoch": 0.13783910758573972,
|
| 11173 |
+
"grad_norm": 0.06090294197201729,
|
| 11174 |
+
"learning_rate": 9.567522170387082e-05,
|
| 11175 |
+
"loss": 0.0246,
|
| 11176 |
+
"step": 1577
|
| 11177 |
+
},
|
| 11178 |
+
{
|
| 11179 |
+
"epoch": 0.13792651348782325,
|
| 11180 |
+
"grad_norm": 0.14577405154705048,
|
| 11181 |
+
"learning_rate": 9.557231735047619e-05,
|
| 11182 |
+
"loss": 0.0258,
|
| 11183 |
+
"step": 1578
|
| 11184 |
+
},
|
| 11185 |
+
{
|
| 11186 |
+
"epoch": 0.13801391938990681,
|
| 11187 |
+
"grad_norm": 0.10201065242290497,
|
| 11188 |
+
"learning_rate": 9.546941769468673e-05,
|
| 11189 |
+
"loss": 0.0313,
|
| 11190 |
+
"step": 1579
|
| 11191 |
+
},
|
| 11192 |
+
{
|
| 11193 |
+
"epoch": 0.13810132529199035,
|
| 11194 |
+
"grad_norm": 0.10618524253368378,
|
| 11195 |
+
"learning_rate": 9.536652284567513e-05,
|
| 11196 |
+
"loss": 0.0301,
|
| 11197 |
+
"step": 1580
|
| 11198 |
+
},
|
| 11199 |
+
{
|
| 11200 |
+
"epoch": 0.13818873119407388,
|
| 11201 |
+
"grad_norm": 0.1036718562245369,
|
| 11202 |
+
"learning_rate": 9.5263632912609e-05,
|
| 11203 |
+
"loss": 0.0289,
|
| 11204 |
+
"step": 1581
|
| 11205 |
+
},
|
| 11206 |
+
{
|
| 11207 |
+
"epoch": 0.1382761370961574,
|
| 11208 |
+
"grad_norm": 0.10260613262653351,
|
| 11209 |
+
"learning_rate": 9.516074800465064e-05,
|
| 11210 |
+
"loss": 0.0358,
|
| 11211 |
+
"step": 1582
|
| 11212 |
+
},
|
| 11213 |
+
{
|
| 11214 |
+
"epoch": 0.13836354299824097,
|
| 11215 |
+
"grad_norm": 0.09655741602182388,
|
| 11216 |
+
"learning_rate": 9.505786823095712e-05,
|
| 11217 |
+
"loss": 0.0363,
|
| 11218 |
+
"step": 1583
|
| 11219 |
+
},
|
| 11220 |
+
{
|
| 11221 |
+
"epoch": 0.1384509489003245,
|
| 11222 |
+
"grad_norm": 0.17335118353366852,
|
| 11223 |
+
"learning_rate": 9.495499370068004e-05,
|
| 11224 |
+
"loss": 0.029,
|
| 11225 |
+
"step": 1584
|
| 11226 |
+
},
|
| 11227 |
+
{
|
| 11228 |
+
"epoch": 0.13853835480240803,
|
| 11229 |
+
"grad_norm": 0.07364212721586227,
|
| 11230 |
+
"learning_rate": 9.485212452296535e-05,
|
| 11231 |
+
"loss": 0.0333,
|
| 11232 |
+
"step": 1585
|
| 11233 |
+
},
|
| 11234 |
+
{
|
| 11235 |
+
"epoch": 0.13862576070449156,
|
| 11236 |
+
"grad_norm": 0.1601741462945938,
|
| 11237 |
+
"learning_rate": 9.474926080695345e-05,
|
| 11238 |
+
"loss": 0.0691,
|
| 11239 |
+
"step": 1586
|
| 11240 |
+
},
|
| 11241 |
+
{
|
| 11242 |
+
"epoch": 0.13871316660657512,
|
| 11243 |
+
"grad_norm": 0.06292049586772919,
|
| 11244 |
+
"learning_rate": 9.464640266177883e-05,
|
| 11245 |
+
"loss": 0.0215,
|
| 11246 |
+
"step": 1587
|
| 11247 |
+
},
|
| 11248 |
+
{
|
| 11249 |
+
"epoch": 0.13880057250865865,
|
| 11250 |
+
"grad_norm": 0.09334374964237213,
|
| 11251 |
+
"learning_rate": 9.454355019657017e-05,
|
| 11252 |
+
"loss": 0.0269,
|
| 11253 |
+
"step": 1588
|
| 11254 |
+
},
|
| 11255 |
+
{
|
| 11256 |
+
"epoch": 0.13888797841074219,
|
| 11257 |
+
"grad_norm": 0.07139234989881516,
|
| 11258 |
+
"learning_rate": 9.444070352045006e-05,
|
| 11259 |
+
"loss": 0.025,
|
| 11260 |
+
"step": 1589
|
| 11261 |
+
},
|
| 11262 |
+
{
|
| 11263 |
+
"epoch": 0.13897538431282572,
|
| 11264 |
+
"grad_norm": 0.07587097585201263,
|
| 11265 |
+
"learning_rate": 9.433786274253495e-05,
|
| 11266 |
+
"loss": 0.0256,
|
| 11267 |
+
"step": 1590
|
| 11268 |
+
},
|
| 11269 |
+
{
|
| 11270 |
+
"epoch": 0.13906279021490925,
|
| 11271 |
+
"grad_norm": 0.1509503573179245,
|
| 11272 |
+
"learning_rate": 9.423502797193508e-05,
|
| 11273 |
+
"loss": 0.0459,
|
| 11274 |
+
"step": 1591
|
| 11275 |
+
},
|
| 11276 |
+
{
|
| 11277 |
+
"epoch": 0.1391501961169928,
|
| 11278 |
+
"grad_norm": 0.15848566591739655,
|
| 11279 |
+
"learning_rate": 9.413219931775426e-05,
|
| 11280 |
+
"loss": 0.0469,
|
| 11281 |
+
"step": 1592
|
| 11282 |
+
},
|
| 11283 |
+
{
|
| 11284 |
+
"epoch": 0.13923760201907634,
|
| 11285 |
+
"grad_norm": 0.126972034573555,
|
| 11286 |
+
"learning_rate": 9.402937688908983e-05,
|
| 11287 |
+
"loss": 0.0427,
|
| 11288 |
+
"step": 1593
|
| 11289 |
+
},
|
| 11290 |
+
{
|
| 11291 |
+
"epoch": 0.13932500792115987,
|
| 11292 |
+
"grad_norm": 0.10220225900411606,
|
| 11293 |
+
"learning_rate": 9.392656079503256e-05,
|
| 11294 |
+
"loss": 0.0349,
|
| 11295 |
+
"step": 1594
|
| 11296 |
+
},
|
| 11297 |
+
{
|
| 11298 |
+
"epoch": 0.1394124138232434,
|
| 11299 |
+
"grad_norm": 0.07176347076892853,
|
| 11300 |
+
"learning_rate": 9.382375114466644e-05,
|
| 11301 |
+
"loss": 0.0295,
|
| 11302 |
+
"step": 1595
|
| 11303 |
+
},
|
| 11304 |
+
{
|
| 11305 |
+
"epoch": 0.13949981972532696,
|
| 11306 |
+
"grad_norm": 0.13212533295154572,
|
| 11307 |
+
"learning_rate": 9.372094804706867e-05,
|
| 11308 |
+
"loss": 0.0551,
|
| 11309 |
+
"step": 1596
|
| 11310 |
+
},
|
| 11311 |
+
{
|
| 11312 |
+
"epoch": 0.1395872256274105,
|
| 11313 |
+
"grad_norm": 0.10252740979194641,
|
| 11314 |
+
"learning_rate": 9.36181516113095e-05,
|
| 11315 |
+
"loss": 0.0403,
|
| 11316 |
+
"step": 1597
|
| 11317 |
+
},
|
| 11318 |
+
{
|
| 11319 |
+
"epoch": 0.13967463152949403,
|
| 11320 |
+
"grad_norm": 0.06956984102725983,
|
| 11321 |
+
"learning_rate": 9.351536194645198e-05,
|
| 11322 |
+
"loss": 0.0216,
|
| 11323 |
+
"step": 1598
|
| 11324 |
+
},
|
| 11325 |
+
{
|
| 11326 |
+
"epoch": 0.13976203743157756,
|
| 11327 |
+
"grad_norm": 0.11074226349592209,
|
| 11328 |
+
"learning_rate": 9.341257916155222e-05,
|
| 11329 |
+
"loss": 0.0274,
|
| 11330 |
+
"step": 1599
|
| 11331 |
+
},
|
| 11332 |
+
{
|
| 11333 |
+
"epoch": 0.13984944333366112,
|
| 11334 |
+
"grad_norm": 0.14012524485588074,
|
| 11335 |
+
"learning_rate": 9.330980336565887e-05,
|
| 11336 |
+
"loss": 0.0365,
|
| 11337 |
+
"step": 1600
|
| 11338 |
+
},
|
| 11339 |
+
{
|
| 11340 |
+
"epoch": 0.13984944333366112,
|
| 11341 |
+
"eval_loss": 0.04779767617583275,
|
| 11342 |
+
"eval_runtime": 168.5729,
|
| 11343 |
+
"eval_samples_per_second": 29.661,
|
| 11344 |
+
"eval_steps_per_second": 7.415,
|
| 11345 |
+
"step": 1600
|
| 11346 |
}
|
| 11347 |
],
|
| 11348 |
"logging_steps": 1,
|
|
|
|
| 11357 |
"early_stopping_threshold": 0.0
|
| 11358 |
},
|
| 11359 |
"attributes": {
|
| 11360 |
+
"early_stopping_patience_counter": 2
|
| 11361 |
}
|
| 11362 |
},
|
| 11363 |
"TrainerControl": {
|
|
|
|
| 11366 |
"should_evaluate": false,
|
| 11367 |
"should_log": false,
|
| 11368 |
"should_save": true,
|
| 11369 |
+
"should_training_stop": true
|
| 11370 |
},
|
| 11371 |
"attributes": {}
|
| 11372 |
}
|
| 11373 |
},
|
| 11374 |
+
"total_flos": 4.835480666709688e+17,
|
| 11375 |
"train_batch_size": 4,
|
| 11376 |
"trial_name": null,
|
| 11377 |
"trial_params": null
|