Training in progress, step 1000, 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 100966336
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8a34e1c3eefe67e3c3b7ca69b7bf2b61344d84bf7fcbffe4179a85113175b184
|
3 |
size 100966336
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 51613668
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9e32ec59747e1e82a1285705a0ea2dc4241c13f194910cbdadecdd9efc35bb19
|
3 |
size 51613668
|
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 sha256:5ba63c0f98e0aaab10c7eb5cb1d2fe94362ee6fea2eb67e46f979e752587f0af
|
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 sha256:15f8669faa402ee0cdd93395e0f0ae74d45b29ab3377a6deed5d79a02e86432c
|
3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"best_metric": 0.0015288225840777159,
|
3 |
"best_model_checkpoint": "miner_id_24/checkpoint-400",
|
4 |
-
"epoch": 0.
|
5 |
"eval_steps": 200,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -5647,6 +5647,1414 @@
|
|
5647 |
"eval_samples_per_second": 21.883,
|
5648 |
"eval_steps_per_second": 10.941,
|
5649 |
"step": 800
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5650 |
}
|
5651 |
],
|
5652 |
"logging_steps": 1,
|
@@ -5661,7 +7069,7 @@
|
|
5661 |
"early_stopping_threshold": 0.0
|
5662 |
},
|
5663 |
"attributes": {
|
5664 |
-
"early_stopping_patience_counter":
|
5665 |
}
|
5666 |
},
|
5667 |
"TrainerControl": {
|
@@ -5670,12 +7078,12 @@
|
|
5670 |
"should_evaluate": false,
|
5671 |
"should_log": false,
|
5672 |
"should_save": true,
|
5673 |
-
"should_training_stop":
|
5674 |
},
|
5675 |
"attributes": {}
|
5676 |
}
|
5677 |
},
|
5678 |
-
"total_flos":
|
5679 |
"train_batch_size": 2,
|
5680 |
"trial_name": null,
|
5681 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": 0.0015288225840777159,
|
3 |
"best_model_checkpoint": "miner_id_24/checkpoint-400",
|
4 |
+
"epoch": 0.18506523549551218,
|
5 |
"eval_steps": 200,
|
6 |
+
"global_step": 1000,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
5647 |
"eval_samples_per_second": 21.883,
|
5648 |
"eval_steps_per_second": 10.941,
|
5649 |
"step": 800
|
5650 |
+
},
|
5651 |
+
{
|
5652 |
+
"epoch": 0.14823725363190524,
|
5653 |
+
"grad_norm": 0.0027204654179513454,
|
5654 |
+
"learning_rate": 0.00019888142826480451,
|
5655 |
+
"loss": 0.0002,
|
5656 |
+
"step": 801
|
5657 |
+
},
|
5658 |
+
{
|
5659 |
+
"epoch": 0.14842231886740076,
|
5660 |
+
"grad_norm": 0.003083963645622134,
|
5661 |
+
"learning_rate": 0.00019887853020944678,
|
5662 |
+
"loss": 0.0002,
|
5663 |
+
"step": 802
|
5664 |
+
},
|
5665 |
+
{
|
5666 |
+
"epoch": 0.14860738410289628,
|
5667 |
+
"grad_norm": 0.00834161601960659,
|
5668 |
+
"learning_rate": 0.00019887562842589593,
|
5669 |
+
"loss": 0.0006,
|
5670 |
+
"step": 803
|
5671 |
+
},
|
5672 |
+
{
|
5673 |
+
"epoch": 0.14879244933839178,
|
5674 |
+
"grad_norm": 0.0013405329082161188,
|
5675 |
+
"learning_rate": 0.00019887272291426141,
|
5676 |
+
"loss": 0.0002,
|
5677 |
+
"step": 804
|
5678 |
+
},
|
5679 |
+
{
|
5680 |
+
"epoch": 0.1489775145738873,
|
5681 |
+
"grad_norm": 0.09376765787601471,
|
5682 |
+
"learning_rate": 0.0001988698136746528,
|
5683 |
+
"loss": 0.0045,
|
5684 |
+
"step": 805
|
5685 |
+
},
|
5686 |
+
{
|
5687 |
+
"epoch": 0.1491625798093828,
|
5688 |
+
"grad_norm": 0.01702548936009407,
|
5689 |
+
"learning_rate": 0.00019886690070717978,
|
5690 |
+
"loss": 0.0007,
|
5691 |
+
"step": 806
|
5692 |
+
},
|
5693 |
+
{
|
5694 |
+
"epoch": 0.14934764504487832,
|
5695 |
+
"grad_norm": 0.0030693926382809877,
|
5696 |
+
"learning_rate": 0.00019886398401195213,
|
5697 |
+
"loss": 0.0003,
|
5698 |
+
"step": 807
|
5699 |
+
},
|
5700 |
+
{
|
5701 |
+
"epoch": 0.14953271028037382,
|
5702 |
+
"grad_norm": 0.09932032227516174,
|
5703 |
+
"learning_rate": 0.00019886106358907982,
|
5704 |
+
"loss": 0.0028,
|
5705 |
+
"step": 808
|
5706 |
+
},
|
5707 |
+
{
|
5708 |
+
"epoch": 0.14971777551586934,
|
5709 |
+
"grad_norm": 0.10470867156982422,
|
5710 |
+
"learning_rate": 0.0001988581394386731,
|
5711 |
+
"loss": 0.0013,
|
5712 |
+
"step": 809
|
5713 |
+
},
|
5714 |
+
{
|
5715 |
+
"epoch": 0.14990284075136487,
|
5716 |
+
"grad_norm": 0.013230692595243454,
|
5717 |
+
"learning_rate": 0.00019885521156084207,
|
5718 |
+
"loss": 0.0005,
|
5719 |
+
"step": 810
|
5720 |
+
},
|
5721 |
+
{
|
5722 |
+
"epoch": 0.15008790598686036,
|
5723 |
+
"grad_norm": 0.0016286876052618027,
|
5724 |
+
"learning_rate": 0.00019885227995569715,
|
5725 |
+
"loss": 0.0002,
|
5726 |
+
"step": 811
|
5727 |
+
},
|
5728 |
+
{
|
5729 |
+
"epoch": 0.1502729712223559,
|
5730 |
+
"grad_norm": 0.03855406120419502,
|
5731 |
+
"learning_rate": 0.00019884934462334898,
|
5732 |
+
"loss": 0.0008,
|
5733 |
+
"step": 812
|
5734 |
+
},
|
5735 |
+
{
|
5736 |
+
"epoch": 0.15045803645785139,
|
5737 |
+
"grad_norm": 0.005248669069260359,
|
5738 |
+
"learning_rate": 0.0001988464055639081,
|
5739 |
+
"loss": 0.0003,
|
5740 |
+
"step": 813
|
5741 |
+
},
|
5742 |
+
{
|
5743 |
+
"epoch": 0.1506431016933469,
|
5744 |
+
"grad_norm": 0.0074467347003519535,
|
5745 |
+
"learning_rate": 0.00019884346277748545,
|
5746 |
+
"loss": 0.0004,
|
5747 |
+
"step": 814
|
5748 |
+
},
|
5749 |
+
{
|
5750 |
+
"epoch": 0.1508281669288424,
|
5751 |
+
"grad_norm": 0.4559827446937561,
|
5752 |
+
"learning_rate": 0.0001988405162641919,
|
5753 |
+
"loss": 0.0047,
|
5754 |
+
"step": 815
|
5755 |
+
},
|
5756 |
+
{
|
5757 |
+
"epoch": 0.15101323216433793,
|
5758 |
+
"grad_norm": 0.0009584883810020983,
|
5759 |
+
"learning_rate": 0.00019883756602413858,
|
5760 |
+
"loss": 0.0001,
|
5761 |
+
"step": 816
|
5762 |
+
},
|
5763 |
+
{
|
5764 |
+
"epoch": 0.15119829739983345,
|
5765 |
+
"grad_norm": 0.0027457585092633963,
|
5766 |
+
"learning_rate": 0.0001988346120574367,
|
5767 |
+
"loss": 0.0002,
|
5768 |
+
"step": 817
|
5769 |
+
},
|
5770 |
+
{
|
5771 |
+
"epoch": 0.15138336263532895,
|
5772 |
+
"grad_norm": 0.06770993024110794,
|
5773 |
+
"learning_rate": 0.0001988316543641977,
|
5774 |
+
"loss": 0.0042,
|
5775 |
+
"step": 818
|
5776 |
+
},
|
5777 |
+
{
|
5778 |
+
"epoch": 0.15156842787082447,
|
5779 |
+
"grad_norm": 0.08119744807481766,
|
5780 |
+
"learning_rate": 0.00019882869294453307,
|
5781 |
+
"loss": 0.0015,
|
5782 |
+
"step": 819
|
5783 |
+
},
|
5784 |
+
{
|
5785 |
+
"epoch": 0.15175349310631997,
|
5786 |
+
"grad_norm": 0.00217386893928051,
|
5787 |
+
"learning_rate": 0.00019882572779855438,
|
5788 |
+
"loss": 0.0002,
|
5789 |
+
"step": 820
|
5790 |
+
},
|
5791 |
+
{
|
5792 |
+
"epoch": 0.1519385583418155,
|
5793 |
+
"grad_norm": 0.00926368311047554,
|
5794 |
+
"learning_rate": 0.0001988227589263736,
|
5795 |
+
"loss": 0.0002,
|
5796 |
+
"step": 821
|
5797 |
+
},
|
5798 |
+
{
|
5799 |
+
"epoch": 0.152123623577311,
|
5800 |
+
"grad_norm": 0.0006100459722802043,
|
5801 |
+
"learning_rate": 0.00019881978632810255,
|
5802 |
+
"loss": 0.0001,
|
5803 |
+
"step": 822
|
5804 |
+
},
|
5805 |
+
{
|
5806 |
+
"epoch": 0.15230868881280651,
|
5807 |
+
"grad_norm": 0.003751414595171809,
|
5808 |
+
"learning_rate": 0.00019881681000385334,
|
5809 |
+
"loss": 0.0002,
|
5810 |
+
"step": 823
|
5811 |
+
},
|
5812 |
+
{
|
5813 |
+
"epoch": 0.15249375404830204,
|
5814 |
+
"grad_norm": 0.00555981183424592,
|
5815 |
+
"learning_rate": 0.00019881382995373822,
|
5816 |
+
"loss": 0.0004,
|
5817 |
+
"step": 824
|
5818 |
+
},
|
5819 |
+
{
|
5820 |
+
"epoch": 0.15267881928379753,
|
5821 |
+
"grad_norm": 0.0011659067822620273,
|
5822 |
+
"learning_rate": 0.0001988108461778695,
|
5823 |
+
"loss": 0.0001,
|
5824 |
+
"step": 825
|
5825 |
+
},
|
5826 |
+
{
|
5827 |
+
"epoch": 0.15286388451929306,
|
5828 |
+
"grad_norm": 0.0023037681821733713,
|
5829 |
+
"learning_rate": 0.0001988078586763597,
|
5830 |
+
"loss": 0.0003,
|
5831 |
+
"step": 826
|
5832 |
+
},
|
5833 |
+
{
|
5834 |
+
"epoch": 0.15304894975478855,
|
5835 |
+
"grad_norm": 0.001487642410211265,
|
5836 |
+
"learning_rate": 0.0001988048674493215,
|
5837 |
+
"loss": 0.0002,
|
5838 |
+
"step": 827
|
5839 |
+
},
|
5840 |
+
{
|
5841 |
+
"epoch": 0.15323401499028408,
|
5842 |
+
"grad_norm": 0.0758136436343193,
|
5843 |
+
"learning_rate": 0.00019880187249686766,
|
5844 |
+
"loss": 0.0055,
|
5845 |
+
"step": 828
|
5846 |
+
},
|
5847 |
+
{
|
5848 |
+
"epoch": 0.15341908022577957,
|
5849 |
+
"grad_norm": 0.0020462728571146727,
|
5850 |
+
"learning_rate": 0.0001987988738191111,
|
5851 |
+
"loss": 0.0002,
|
5852 |
+
"step": 829
|
5853 |
+
},
|
5854 |
+
{
|
5855 |
+
"epoch": 0.1536041454612751,
|
5856 |
+
"grad_norm": 0.0192215908318758,
|
5857 |
+
"learning_rate": 0.00019879587141616485,
|
5858 |
+
"loss": 0.0006,
|
5859 |
+
"step": 830
|
5860 |
+
},
|
5861 |
+
{
|
5862 |
+
"epoch": 0.15378921069677062,
|
5863 |
+
"grad_norm": 0.0020349507685750723,
|
5864 |
+
"learning_rate": 0.00019879286528814218,
|
5865 |
+
"loss": 0.0002,
|
5866 |
+
"step": 831
|
5867 |
+
},
|
5868 |
+
{
|
5869 |
+
"epoch": 0.15397427593226612,
|
5870 |
+
"grad_norm": 0.0010202473495155573,
|
5871 |
+
"learning_rate": 0.0001987898554351564,
|
5872 |
+
"loss": 0.0001,
|
5873 |
+
"step": 832
|
5874 |
+
},
|
5875 |
+
{
|
5876 |
+
"epoch": 0.15415934116776164,
|
5877 |
+
"grad_norm": 0.25521135330200195,
|
5878 |
+
"learning_rate": 0.00019878684185732103,
|
5879 |
+
"loss": 0.0027,
|
5880 |
+
"step": 833
|
5881 |
+
},
|
5882 |
+
{
|
5883 |
+
"epoch": 0.15434440640325714,
|
5884 |
+
"grad_norm": 0.02357971854507923,
|
5885 |
+
"learning_rate": 0.00019878382455474963,
|
5886 |
+
"loss": 0.0008,
|
5887 |
+
"step": 834
|
5888 |
+
},
|
5889 |
+
{
|
5890 |
+
"epoch": 0.15452947163875266,
|
5891 |
+
"grad_norm": 0.0028601621743291616,
|
5892 |
+
"learning_rate": 0.00019878080352755598,
|
5893 |
+
"loss": 0.0002,
|
5894 |
+
"step": 835
|
5895 |
+
},
|
5896 |
+
{
|
5897 |
+
"epoch": 0.15471453687424816,
|
5898 |
+
"grad_norm": 0.0013797554420307279,
|
5899 |
+
"learning_rate": 0.00019877777877585405,
|
5900 |
+
"loss": 0.0001,
|
5901 |
+
"step": 836
|
5902 |
+
},
|
5903 |
+
{
|
5904 |
+
"epoch": 0.15489960210974368,
|
5905 |
+
"grad_norm": 0.011994694359600544,
|
5906 |
+
"learning_rate": 0.00019877475029975785,
|
5907 |
+
"loss": 0.0006,
|
5908 |
+
"step": 837
|
5909 |
+
},
|
5910 |
+
{
|
5911 |
+
"epoch": 0.1550846673452392,
|
5912 |
+
"grad_norm": 0.2087889313697815,
|
5913 |
+
"learning_rate": 0.00019877171809938156,
|
5914 |
+
"loss": 0.0034,
|
5915 |
+
"step": 838
|
5916 |
+
},
|
5917 |
+
{
|
5918 |
+
"epoch": 0.1552697325807347,
|
5919 |
+
"grad_norm": 0.01124395802617073,
|
5920 |
+
"learning_rate": 0.0001987686821748395,
|
5921 |
+
"loss": 0.001,
|
5922 |
+
"step": 839
|
5923 |
+
},
|
5924 |
+
{
|
5925 |
+
"epoch": 0.15545479781623023,
|
5926 |
+
"grad_norm": 0.04885182902216911,
|
5927 |
+
"learning_rate": 0.00019876564252624616,
|
5928 |
+
"loss": 0.0004,
|
5929 |
+
"step": 840
|
5930 |
+
},
|
5931 |
+
{
|
5932 |
+
"epoch": 0.15563986305172572,
|
5933 |
+
"grad_norm": 0.0007935105240903795,
|
5934 |
+
"learning_rate": 0.00019876259915371618,
|
5935 |
+
"loss": 0.0001,
|
5936 |
+
"step": 841
|
5937 |
+
},
|
5938 |
+
{
|
5939 |
+
"epoch": 0.15582492828722125,
|
5940 |
+
"grad_norm": 0.04080585017800331,
|
5941 |
+
"learning_rate": 0.00019875955205736422,
|
5942 |
+
"loss": 0.0008,
|
5943 |
+
"step": 842
|
5944 |
+
},
|
5945 |
+
{
|
5946 |
+
"epoch": 0.15600999352271674,
|
5947 |
+
"grad_norm": 0.016090860590338707,
|
5948 |
+
"learning_rate": 0.00019875650123730526,
|
5949 |
+
"loss": 0.0003,
|
5950 |
+
"step": 843
|
5951 |
+
},
|
5952 |
+
{
|
5953 |
+
"epoch": 0.15619505875821227,
|
5954 |
+
"grad_norm": 0.001955505693331361,
|
5955 |
+
"learning_rate": 0.0001987534466936543,
|
5956 |
+
"loss": 0.0002,
|
5957 |
+
"step": 844
|
5958 |
+
},
|
5959 |
+
{
|
5960 |
+
"epoch": 0.1563801239937078,
|
5961 |
+
"grad_norm": 0.044548399746418,
|
5962 |
+
"learning_rate": 0.0001987503884265265,
|
5963 |
+
"loss": 0.0025,
|
5964 |
+
"step": 845
|
5965 |
+
},
|
5966 |
+
{
|
5967 |
+
"epoch": 0.1565651892292033,
|
5968 |
+
"grad_norm": 0.00806332565844059,
|
5969 |
+
"learning_rate": 0.00019874732643603715,
|
5970 |
+
"loss": 0.0005,
|
5971 |
+
"step": 846
|
5972 |
+
},
|
5973 |
+
{
|
5974 |
+
"epoch": 0.1567502544646988,
|
5975 |
+
"grad_norm": 0.1479962319135666,
|
5976 |
+
"learning_rate": 0.00019874426072230173,
|
5977 |
+
"loss": 0.0011,
|
5978 |
+
"step": 847
|
5979 |
+
},
|
5980 |
+
{
|
5981 |
+
"epoch": 0.1569353197001943,
|
5982 |
+
"grad_norm": 0.003856030525639653,
|
5983 |
+
"learning_rate": 0.00019874119128543584,
|
5984 |
+
"loss": 0.0002,
|
5985 |
+
"step": 848
|
5986 |
+
},
|
5987 |
+
{
|
5988 |
+
"epoch": 0.15712038493568983,
|
5989 |
+
"grad_norm": 0.0037897711154073477,
|
5990 |
+
"learning_rate": 0.00019873811812555522,
|
5991 |
+
"loss": 0.0003,
|
5992 |
+
"step": 849
|
5993 |
+
},
|
5994 |
+
{
|
5995 |
+
"epoch": 0.15730545017118533,
|
5996 |
+
"grad_norm": 0.0020272263791412115,
|
5997 |
+
"learning_rate": 0.0001987350412427757,
|
5998 |
+
"loss": 0.0002,
|
5999 |
+
"step": 850
|
6000 |
+
},
|
6001 |
+
{
|
6002 |
+
"epoch": 0.15749051540668085,
|
6003 |
+
"grad_norm": 0.001225305488333106,
|
6004 |
+
"learning_rate": 0.00019873196063721334,
|
6005 |
+
"loss": 0.0001,
|
6006 |
+
"step": 851
|
6007 |
+
},
|
6008 |
+
{
|
6009 |
+
"epoch": 0.15767558064217638,
|
6010 |
+
"grad_norm": 0.0012447525514289737,
|
6011 |
+
"learning_rate": 0.00019872887630898424,
|
6012 |
+
"loss": 0.0001,
|
6013 |
+
"step": 852
|
6014 |
+
},
|
6015 |
+
{
|
6016 |
+
"epoch": 0.15786064587767187,
|
6017 |
+
"grad_norm": 0.018615124747157097,
|
6018 |
+
"learning_rate": 0.00019872578825820477,
|
6019 |
+
"loss": 0.0004,
|
6020 |
+
"step": 853
|
6021 |
+
},
|
6022 |
+
{
|
6023 |
+
"epoch": 0.1580457111131674,
|
6024 |
+
"grad_norm": 0.01611863076686859,
|
6025 |
+
"learning_rate": 0.00019872269648499126,
|
6026 |
+
"loss": 0.0004,
|
6027 |
+
"step": 854
|
6028 |
+
},
|
6029 |
+
{
|
6030 |
+
"epoch": 0.1582307763486629,
|
6031 |
+
"grad_norm": 0.3293641209602356,
|
6032 |
+
"learning_rate": 0.00019871960098946038,
|
6033 |
+
"loss": 0.0155,
|
6034 |
+
"step": 855
|
6035 |
+
},
|
6036 |
+
{
|
6037 |
+
"epoch": 0.15841584158415842,
|
6038 |
+
"grad_norm": 0.0038168844766914845,
|
6039 |
+
"learning_rate": 0.00019871650177172883,
|
6040 |
+
"loss": 0.0002,
|
6041 |
+
"step": 856
|
6042 |
+
},
|
6043 |
+
{
|
6044 |
+
"epoch": 0.15860090681965394,
|
6045 |
+
"grad_norm": 0.04572943598031998,
|
6046 |
+
"learning_rate": 0.00019871339883191342,
|
6047 |
+
"loss": 0.0019,
|
6048 |
+
"step": 857
|
6049 |
+
},
|
6050 |
+
{
|
6051 |
+
"epoch": 0.15878597205514944,
|
6052 |
+
"grad_norm": 0.0027950333897024393,
|
6053 |
+
"learning_rate": 0.00019871029217013116,
|
6054 |
+
"loss": 0.0002,
|
6055 |
+
"step": 858
|
6056 |
+
},
|
6057 |
+
{
|
6058 |
+
"epoch": 0.15897103729064496,
|
6059 |
+
"grad_norm": 0.024270273745059967,
|
6060 |
+
"learning_rate": 0.00019870718178649922,
|
6061 |
+
"loss": 0.0008,
|
6062 |
+
"step": 859
|
6063 |
+
},
|
6064 |
+
{
|
6065 |
+
"epoch": 0.15915610252614046,
|
6066 |
+
"grad_norm": 0.0011921890545636415,
|
6067 |
+
"learning_rate": 0.00019870406768113485,
|
6068 |
+
"loss": 0.0001,
|
6069 |
+
"step": 860
|
6070 |
+
},
|
6071 |
+
{
|
6072 |
+
"epoch": 0.15934116776163598,
|
6073 |
+
"grad_norm": 0.1097538024187088,
|
6074 |
+
"learning_rate": 0.00019870094985415546,
|
6075 |
+
"loss": 0.0007,
|
6076 |
+
"step": 861
|
6077 |
+
},
|
6078 |
+
{
|
6079 |
+
"epoch": 0.15952623299713148,
|
6080 |
+
"grad_norm": 0.20224367082118988,
|
6081 |
+
"learning_rate": 0.00019869782830567862,
|
6082 |
+
"loss": 0.0064,
|
6083 |
+
"step": 862
|
6084 |
+
},
|
6085 |
+
{
|
6086 |
+
"epoch": 0.159711298232627,
|
6087 |
+
"grad_norm": 0.007074209861457348,
|
6088 |
+
"learning_rate": 0.00019869470303582201,
|
6089 |
+
"loss": 0.0002,
|
6090 |
+
"step": 863
|
6091 |
+
},
|
6092 |
+
{
|
6093 |
+
"epoch": 0.15989636346812253,
|
6094 |
+
"grad_norm": 0.004815272521227598,
|
6095 |
+
"learning_rate": 0.0001986915740447035,
|
6096 |
+
"loss": 0.0003,
|
6097 |
+
"step": 864
|
6098 |
+
},
|
6099 |
+
{
|
6100 |
+
"epoch": 0.16008142870361802,
|
6101 |
+
"grad_norm": 0.0019826674833893776,
|
6102 |
+
"learning_rate": 0.00019868844133244106,
|
6103 |
+
"loss": 0.0001,
|
6104 |
+
"step": 865
|
6105 |
+
},
|
6106 |
+
{
|
6107 |
+
"epoch": 0.16026649393911355,
|
6108 |
+
"grad_norm": 0.014026563614606857,
|
6109 |
+
"learning_rate": 0.00019868530489915279,
|
6110 |
+
"loss": 0.0004,
|
6111 |
+
"step": 866
|
6112 |
+
},
|
6113 |
+
{
|
6114 |
+
"epoch": 0.16045155917460904,
|
6115 |
+
"grad_norm": 0.006988345645368099,
|
6116 |
+
"learning_rate": 0.00019868216474495693,
|
6117 |
+
"loss": 0.0003,
|
6118 |
+
"step": 867
|
6119 |
+
},
|
6120 |
+
{
|
6121 |
+
"epoch": 0.16063662441010457,
|
6122 |
+
"grad_norm": 0.010466384701430798,
|
6123 |
+
"learning_rate": 0.00019867902086997196,
|
6124 |
+
"loss": 0.0003,
|
6125 |
+
"step": 868
|
6126 |
+
},
|
6127 |
+
{
|
6128 |
+
"epoch": 0.16082168964560006,
|
6129 |
+
"grad_norm": 0.0015272882301360369,
|
6130 |
+
"learning_rate": 0.0001986758732743163,
|
6131 |
+
"loss": 0.0002,
|
6132 |
+
"step": 869
|
6133 |
+
},
|
6134 |
+
{
|
6135 |
+
"epoch": 0.1610067548810956,
|
6136 |
+
"grad_norm": 0.030166354030370712,
|
6137 |
+
"learning_rate": 0.00019867272195810875,
|
6138 |
+
"loss": 0.0008,
|
6139 |
+
"step": 870
|
6140 |
+
},
|
6141 |
+
{
|
6142 |
+
"epoch": 0.1611918201165911,
|
6143 |
+
"grad_norm": 0.004422664642333984,
|
6144 |
+
"learning_rate": 0.000198669566921468,
|
6145 |
+
"loss": 0.0004,
|
6146 |
+
"step": 871
|
6147 |
+
},
|
6148 |
+
{
|
6149 |
+
"epoch": 0.1613768853520866,
|
6150 |
+
"grad_norm": 0.0035160244442522526,
|
6151 |
+
"learning_rate": 0.00019866640816451313,
|
6152 |
+
"loss": 0.0003,
|
6153 |
+
"step": 872
|
6154 |
+
},
|
6155 |
+
{
|
6156 |
+
"epoch": 0.16156195058758213,
|
6157 |
+
"grad_norm": 0.09532395750284195,
|
6158 |
+
"learning_rate": 0.00019866324568736321,
|
6159 |
+
"loss": 0.0018,
|
6160 |
+
"step": 873
|
6161 |
+
},
|
6162 |
+
{
|
6163 |
+
"epoch": 0.16174701582307763,
|
6164 |
+
"grad_norm": 0.0015503017930313945,
|
6165 |
+
"learning_rate": 0.00019866007949013743,
|
6166 |
+
"loss": 0.0002,
|
6167 |
+
"step": 874
|
6168 |
+
},
|
6169 |
+
{
|
6170 |
+
"epoch": 0.16193208105857315,
|
6171 |
+
"grad_norm": 0.03559022769331932,
|
6172 |
+
"learning_rate": 0.00019865690957295522,
|
6173 |
+
"loss": 0.0006,
|
6174 |
+
"step": 875
|
6175 |
+
},
|
6176 |
+
{
|
6177 |
+
"epoch": 0.16211714629406865,
|
6178 |
+
"grad_norm": 0.011263100430369377,
|
6179 |
+
"learning_rate": 0.00019865373593593608,
|
6180 |
+
"loss": 0.0003,
|
6181 |
+
"step": 876
|
6182 |
+
},
|
6183 |
+
{
|
6184 |
+
"epoch": 0.16230221152956417,
|
6185 |
+
"grad_norm": 0.0601922944188118,
|
6186 |
+
"learning_rate": 0.00019865055857919966,
|
6187 |
+
"loss": 0.0016,
|
6188 |
+
"step": 877
|
6189 |
+
},
|
6190 |
+
{
|
6191 |
+
"epoch": 0.1624872767650597,
|
6192 |
+
"grad_norm": 0.04134982451796532,
|
6193 |
+
"learning_rate": 0.0001986473775028658,
|
6194 |
+
"loss": 0.0003,
|
6195 |
+
"step": 878
|
6196 |
+
},
|
6197 |
+
{
|
6198 |
+
"epoch": 0.1626723420005552,
|
6199 |
+
"grad_norm": 0.26190969347953796,
|
6200 |
+
"learning_rate": 0.00019864419270705443,
|
6201 |
+
"loss": 0.0017,
|
6202 |
+
"step": 879
|
6203 |
+
},
|
6204 |
+
{
|
6205 |
+
"epoch": 0.16285740723605072,
|
6206 |
+
"grad_norm": 0.1426112949848175,
|
6207 |
+
"learning_rate": 0.00019864100419188563,
|
6208 |
+
"loss": 0.0007,
|
6209 |
+
"step": 880
|
6210 |
+
},
|
6211 |
+
{
|
6212 |
+
"epoch": 0.1630424724715462,
|
6213 |
+
"grad_norm": 0.002557241590693593,
|
6214 |
+
"learning_rate": 0.00019863781195747957,
|
6215 |
+
"loss": 0.0002,
|
6216 |
+
"step": 881
|
6217 |
+
},
|
6218 |
+
{
|
6219 |
+
"epoch": 0.16322753770704174,
|
6220 |
+
"grad_norm": 0.020958442240953445,
|
6221 |
+
"learning_rate": 0.0001986346160039567,
|
6222 |
+
"loss": 0.0006,
|
6223 |
+
"step": 882
|
6224 |
+
},
|
6225 |
+
{
|
6226 |
+
"epoch": 0.16341260294253723,
|
6227 |
+
"grad_norm": 0.08520295470952988,
|
6228 |
+
"learning_rate": 0.00019863141633143748,
|
6229 |
+
"loss": 0.0021,
|
6230 |
+
"step": 883
|
6231 |
+
},
|
6232 |
+
{
|
6233 |
+
"epoch": 0.16359766817803276,
|
6234 |
+
"grad_norm": 0.026120366528630257,
|
6235 |
+
"learning_rate": 0.00019862821294004253,
|
6236 |
+
"loss": 0.0003,
|
6237 |
+
"step": 884
|
6238 |
+
},
|
6239 |
+
{
|
6240 |
+
"epoch": 0.16378273341352828,
|
6241 |
+
"grad_norm": 0.08190575987100601,
|
6242 |
+
"learning_rate": 0.00019862500582989263,
|
6243 |
+
"loss": 0.0014,
|
6244 |
+
"step": 885
|
6245 |
+
},
|
6246 |
+
{
|
6247 |
+
"epoch": 0.16396779864902378,
|
6248 |
+
"grad_norm": 0.027544621378183365,
|
6249 |
+
"learning_rate": 0.00019862179500110876,
|
6250 |
+
"loss": 0.0008,
|
6251 |
+
"step": 886
|
6252 |
+
},
|
6253 |
+
{
|
6254 |
+
"epoch": 0.1641528638845193,
|
6255 |
+
"grad_norm": 0.004529684316366911,
|
6256 |
+
"learning_rate": 0.00019861858045381196,
|
6257 |
+
"loss": 0.0003,
|
6258 |
+
"step": 887
|
6259 |
+
},
|
6260 |
+
{
|
6261 |
+
"epoch": 0.1643379291200148,
|
6262 |
+
"grad_norm": 0.03812885656952858,
|
6263 |
+
"learning_rate": 0.00019861536218812343,
|
6264 |
+
"loss": 0.0017,
|
6265 |
+
"step": 888
|
6266 |
+
},
|
6267 |
+
{
|
6268 |
+
"epoch": 0.16452299435551032,
|
6269 |
+
"grad_norm": 0.005853751208633184,
|
6270 |
+
"learning_rate": 0.00019861214020416447,
|
6271 |
+
"loss": 0.0002,
|
6272 |
+
"step": 889
|
6273 |
+
},
|
6274 |
+
{
|
6275 |
+
"epoch": 0.16470805959100582,
|
6276 |
+
"grad_norm": 0.004154008813202381,
|
6277 |
+
"learning_rate": 0.00019860891450205665,
|
6278 |
+
"loss": 0.0003,
|
6279 |
+
"step": 890
|
6280 |
+
},
|
6281 |
+
{
|
6282 |
+
"epoch": 0.16489312482650134,
|
6283 |
+
"grad_norm": 0.0007245794404298067,
|
6284 |
+
"learning_rate": 0.00019860568508192153,
|
6285 |
+
"loss": 0.0001,
|
6286 |
+
"step": 891
|
6287 |
+
},
|
6288 |
+
{
|
6289 |
+
"epoch": 0.16507819006199687,
|
6290 |
+
"grad_norm": 0.00185197067912668,
|
6291 |
+
"learning_rate": 0.0001986024519438809,
|
6292 |
+
"loss": 0.0002,
|
6293 |
+
"step": 892
|
6294 |
+
},
|
6295 |
+
{
|
6296 |
+
"epoch": 0.16526325529749236,
|
6297 |
+
"grad_norm": 0.01505447831004858,
|
6298 |
+
"learning_rate": 0.00019859921508805665,
|
6299 |
+
"loss": 0.0004,
|
6300 |
+
"step": 893
|
6301 |
+
},
|
6302 |
+
{
|
6303 |
+
"epoch": 0.16544832053298789,
|
6304 |
+
"grad_norm": 0.016156723722815514,
|
6305 |
+
"learning_rate": 0.00019859597451457085,
|
6306 |
+
"loss": 0.0004,
|
6307 |
+
"step": 894
|
6308 |
+
},
|
6309 |
+
{
|
6310 |
+
"epoch": 0.16563338576848338,
|
6311 |
+
"grad_norm": 0.004411355126649141,
|
6312 |
+
"learning_rate": 0.00019859273022354565,
|
6313 |
+
"loss": 0.0003,
|
6314 |
+
"step": 895
|
6315 |
+
},
|
6316 |
+
{
|
6317 |
+
"epoch": 0.1658184510039789,
|
6318 |
+
"grad_norm": 0.0008460478275083005,
|
6319 |
+
"learning_rate": 0.0001985894822151034,
|
6320 |
+
"loss": 0.0001,
|
6321 |
+
"step": 896
|
6322 |
+
},
|
6323 |
+
{
|
6324 |
+
"epoch": 0.1660035162394744,
|
6325 |
+
"grad_norm": 0.0012289606966078281,
|
6326 |
+
"learning_rate": 0.00019858623048936655,
|
6327 |
+
"loss": 0.0001,
|
6328 |
+
"step": 897
|
6329 |
+
},
|
6330 |
+
{
|
6331 |
+
"epoch": 0.16618858147496993,
|
6332 |
+
"grad_norm": 0.004417366348206997,
|
6333 |
+
"learning_rate": 0.00019858297504645774,
|
6334 |
+
"loss": 0.0005,
|
6335 |
+
"step": 898
|
6336 |
+
},
|
6337 |
+
{
|
6338 |
+
"epoch": 0.16637364671046545,
|
6339 |
+
"grad_norm": 0.00527498684823513,
|
6340 |
+
"learning_rate": 0.00019857971588649965,
|
6341 |
+
"loss": 0.0002,
|
6342 |
+
"step": 899
|
6343 |
+
},
|
6344 |
+
{
|
6345 |
+
"epoch": 0.16655871194596095,
|
6346 |
+
"grad_norm": 1.9224200248718262,
|
6347 |
+
"learning_rate": 0.00019857645300961526,
|
6348 |
+
"loss": 0.0117,
|
6349 |
+
"step": 900
|
6350 |
+
},
|
6351 |
+
{
|
6352 |
+
"epoch": 0.16674377718145647,
|
6353 |
+
"grad_norm": 0.0036594411358237267,
|
6354 |
+
"learning_rate": 0.0001985731864159275,
|
6355 |
+
"loss": 0.0002,
|
6356 |
+
"step": 901
|
6357 |
+
},
|
6358 |
+
{
|
6359 |
+
"epoch": 0.16692884241695197,
|
6360 |
+
"grad_norm": 0.003251464106142521,
|
6361 |
+
"learning_rate": 0.0001985699161055596,
|
6362 |
+
"loss": 0.0001,
|
6363 |
+
"step": 902
|
6364 |
+
},
|
6365 |
+
{
|
6366 |
+
"epoch": 0.1671139076524475,
|
6367 |
+
"grad_norm": 0.13795721530914307,
|
6368 |
+
"learning_rate": 0.00019856664207863482,
|
6369 |
+
"loss": 0.0092,
|
6370 |
+
"step": 903
|
6371 |
+
},
|
6372 |
+
{
|
6373 |
+
"epoch": 0.167298972887943,
|
6374 |
+
"grad_norm": 0.3705841898918152,
|
6375 |
+
"learning_rate": 0.00019856336433527667,
|
6376 |
+
"loss": 0.0091,
|
6377 |
+
"step": 904
|
6378 |
+
},
|
6379 |
+
{
|
6380 |
+
"epoch": 0.1674840381234385,
|
6381 |
+
"grad_norm": 0.0026172541547566652,
|
6382 |
+
"learning_rate": 0.0001985600828756087,
|
6383 |
+
"loss": 0.0002,
|
6384 |
+
"step": 905
|
6385 |
+
},
|
6386 |
+
{
|
6387 |
+
"epoch": 0.16766910335893404,
|
6388 |
+
"grad_norm": 0.003219661535695195,
|
6389 |
+
"learning_rate": 0.00019855679769975457,
|
6390 |
+
"loss": 0.0001,
|
6391 |
+
"step": 906
|
6392 |
+
},
|
6393 |
+
{
|
6394 |
+
"epoch": 0.16785416859442953,
|
6395 |
+
"grad_norm": 0.001061516348272562,
|
6396 |
+
"learning_rate": 0.00019855350880783825,
|
6397 |
+
"loss": 0.0001,
|
6398 |
+
"step": 907
|
6399 |
+
},
|
6400 |
+
{
|
6401 |
+
"epoch": 0.16803923382992506,
|
6402 |
+
"grad_norm": 0.0020027856808155775,
|
6403 |
+
"learning_rate": 0.0001985502161999837,
|
6404 |
+
"loss": 0.0002,
|
6405 |
+
"step": 908
|
6406 |
+
},
|
6407 |
+
{
|
6408 |
+
"epoch": 0.16822429906542055,
|
6409 |
+
"grad_norm": 0.004068360198289156,
|
6410 |
+
"learning_rate": 0.0001985469198763151,
|
6411 |
+
"loss": 0.0002,
|
6412 |
+
"step": 909
|
6413 |
+
},
|
6414 |
+
{
|
6415 |
+
"epoch": 0.16840936430091608,
|
6416 |
+
"grad_norm": 0.013775911182165146,
|
6417 |
+
"learning_rate": 0.00019854361983695672,
|
6418 |
+
"loss": 0.0013,
|
6419 |
+
"step": 910
|
6420 |
+
},
|
6421 |
+
{
|
6422 |
+
"epoch": 0.16859442953641157,
|
6423 |
+
"grad_norm": 0.05193403735756874,
|
6424 |
+
"learning_rate": 0.00019854031608203295,
|
6425 |
+
"loss": 0.0007,
|
6426 |
+
"step": 911
|
6427 |
+
},
|
6428 |
+
{
|
6429 |
+
"epoch": 0.1687794947719071,
|
6430 |
+
"grad_norm": 0.07376479357481003,
|
6431 |
+
"learning_rate": 0.00019853700861166839,
|
6432 |
+
"loss": 0.0027,
|
6433 |
+
"step": 912
|
6434 |
+
},
|
6435 |
+
{
|
6436 |
+
"epoch": 0.16896456000740262,
|
6437 |
+
"grad_norm": 0.14925335347652435,
|
6438 |
+
"learning_rate": 0.00019853369742598776,
|
6439 |
+
"loss": 0.004,
|
6440 |
+
"step": 913
|
6441 |
+
},
|
6442 |
+
{
|
6443 |
+
"epoch": 0.16914962524289812,
|
6444 |
+
"grad_norm": 0.0016460709739476442,
|
6445 |
+
"learning_rate": 0.00019853038252511587,
|
6446 |
+
"loss": 0.0001,
|
6447 |
+
"step": 914
|
6448 |
+
},
|
6449 |
+
{
|
6450 |
+
"epoch": 0.16933469047839364,
|
6451 |
+
"grad_norm": 0.17655791342258453,
|
6452 |
+
"learning_rate": 0.00019852706390917775,
|
6453 |
+
"loss": 0.0047,
|
6454 |
+
"step": 915
|
6455 |
+
},
|
6456 |
+
{
|
6457 |
+
"epoch": 0.16951975571388914,
|
6458 |
+
"grad_norm": 0.07149423658847809,
|
6459 |
+
"learning_rate": 0.0001985237415782985,
|
6460 |
+
"loss": 0.0055,
|
6461 |
+
"step": 916
|
6462 |
+
},
|
6463 |
+
{
|
6464 |
+
"epoch": 0.16970482094938466,
|
6465 |
+
"grad_norm": 0.015162936411798,
|
6466 |
+
"learning_rate": 0.00019852041553260338,
|
6467 |
+
"loss": 0.0008,
|
6468 |
+
"step": 917
|
6469 |
+
},
|
6470 |
+
{
|
6471 |
+
"epoch": 0.16988988618488016,
|
6472 |
+
"grad_norm": 0.003818758763372898,
|
6473 |
+
"learning_rate": 0.00019851708577221783,
|
6474 |
+
"loss": 0.0002,
|
6475 |
+
"step": 918
|
6476 |
+
},
|
6477 |
+
{
|
6478 |
+
"epoch": 0.17007495142037568,
|
6479 |
+
"grad_norm": 0.04823250323534012,
|
6480 |
+
"learning_rate": 0.0001985137522972674,
|
6481 |
+
"loss": 0.0013,
|
6482 |
+
"step": 919
|
6483 |
+
},
|
6484 |
+
{
|
6485 |
+
"epoch": 0.1702600166558712,
|
6486 |
+
"grad_norm": 0.019358035176992416,
|
6487 |
+
"learning_rate": 0.00019851041510787773,
|
6488 |
+
"loss": 0.0009,
|
6489 |
+
"step": 920
|
6490 |
+
},
|
6491 |
+
{
|
6492 |
+
"epoch": 0.1704450818913667,
|
6493 |
+
"grad_norm": 0.006376199424266815,
|
6494 |
+
"learning_rate": 0.00019850707420417468,
|
6495 |
+
"loss": 0.0003,
|
6496 |
+
"step": 921
|
6497 |
+
},
|
6498 |
+
{
|
6499 |
+
"epoch": 0.17063014712686222,
|
6500 |
+
"grad_norm": 0.1438623070716858,
|
6501 |
+
"learning_rate": 0.00019850372958628423,
|
6502 |
+
"loss": 0.0008,
|
6503 |
+
"step": 922
|
6504 |
+
},
|
6505 |
+
{
|
6506 |
+
"epoch": 0.17081521236235772,
|
6507 |
+
"grad_norm": 0.005809268448501825,
|
6508 |
+
"learning_rate": 0.00019850038125433247,
|
6509 |
+
"loss": 0.0004,
|
6510 |
+
"step": 923
|
6511 |
+
},
|
6512 |
+
{
|
6513 |
+
"epoch": 0.17100027759785325,
|
6514 |
+
"grad_norm": 0.0019713384099304676,
|
6515 |
+
"learning_rate": 0.00019849702920844565,
|
6516 |
+
"loss": 0.0002,
|
6517 |
+
"step": 924
|
6518 |
+
},
|
6519 |
+
{
|
6520 |
+
"epoch": 0.17118534283334877,
|
6521 |
+
"grad_norm": 0.11557551473379135,
|
6522 |
+
"learning_rate": 0.00019849367344875013,
|
6523 |
+
"loss": 0.003,
|
6524 |
+
"step": 925
|
6525 |
+
},
|
6526 |
+
{
|
6527 |
+
"epoch": 0.17137040806884427,
|
6528 |
+
"grad_norm": 0.050921566784381866,
|
6529 |
+
"learning_rate": 0.0001984903139753725,
|
6530 |
+
"loss": 0.0039,
|
6531 |
+
"step": 926
|
6532 |
+
},
|
6533 |
+
{
|
6534 |
+
"epoch": 0.1715554733043398,
|
6535 |
+
"grad_norm": 0.031566072255373,
|
6536 |
+
"learning_rate": 0.0001984869507884394,
|
6537 |
+
"loss": 0.0011,
|
6538 |
+
"step": 927
|
6539 |
+
},
|
6540 |
+
{
|
6541 |
+
"epoch": 0.17174053853983529,
|
6542 |
+
"grad_norm": 0.19348770380020142,
|
6543 |
+
"learning_rate": 0.00019848358388807762,
|
6544 |
+
"loss": 0.0085,
|
6545 |
+
"step": 928
|
6546 |
+
},
|
6547 |
+
{
|
6548 |
+
"epoch": 0.1719256037753308,
|
6549 |
+
"grad_norm": 0.06429725885391235,
|
6550 |
+
"learning_rate": 0.00019848021327441412,
|
6551 |
+
"loss": 0.0028,
|
6552 |
+
"step": 929
|
6553 |
+
},
|
6554 |
+
{
|
6555 |
+
"epoch": 0.1721106690108263,
|
6556 |
+
"grad_norm": 0.06797130405902863,
|
6557 |
+
"learning_rate": 0.00019847683894757602,
|
6558 |
+
"loss": 0.0034,
|
6559 |
+
"step": 930
|
6560 |
+
},
|
6561 |
+
{
|
6562 |
+
"epoch": 0.17229573424632183,
|
6563 |
+
"grad_norm": 0.07783009111881256,
|
6564 |
+
"learning_rate": 0.00019847346090769047,
|
6565 |
+
"loss": 0.0113,
|
6566 |
+
"step": 931
|
6567 |
+
},
|
6568 |
+
{
|
6569 |
+
"epoch": 0.17248079948181735,
|
6570 |
+
"grad_norm": 0.01035092119127512,
|
6571 |
+
"learning_rate": 0.0001984700791548849,
|
6572 |
+
"loss": 0.0003,
|
6573 |
+
"step": 932
|
6574 |
+
},
|
6575 |
+
{
|
6576 |
+
"epoch": 0.17266586471731285,
|
6577 |
+
"grad_norm": 0.021911177784204483,
|
6578 |
+
"learning_rate": 0.00019846669368928682,
|
6579 |
+
"loss": 0.0004,
|
6580 |
+
"step": 933
|
6581 |
+
},
|
6582 |
+
{
|
6583 |
+
"epoch": 0.17285092995280837,
|
6584 |
+
"grad_norm": 0.0650099590420723,
|
6585 |
+
"learning_rate": 0.00019846330451102383,
|
6586 |
+
"loss": 0.0014,
|
6587 |
+
"step": 934
|
6588 |
+
},
|
6589 |
+
{
|
6590 |
+
"epoch": 0.17303599518830387,
|
6591 |
+
"grad_norm": 0.010482696816325188,
|
6592 |
+
"learning_rate": 0.00019845991162022379,
|
6593 |
+
"loss": 0.0002,
|
6594 |
+
"step": 935
|
6595 |
+
},
|
6596 |
+
{
|
6597 |
+
"epoch": 0.1732210604237994,
|
6598 |
+
"grad_norm": 0.0009689448052085936,
|
6599 |
+
"learning_rate": 0.0001984565150170146,
|
6600 |
+
"loss": 0.0001,
|
6601 |
+
"step": 936
|
6602 |
+
},
|
6603 |
+
{
|
6604 |
+
"epoch": 0.1734061256592949,
|
6605 |
+
"grad_norm": 0.0025378544814884663,
|
6606 |
+
"learning_rate": 0.00019845311470152429,
|
6607 |
+
"loss": 0.0002,
|
6608 |
+
"step": 937
|
6609 |
+
},
|
6610 |
+
{
|
6611 |
+
"epoch": 0.17359119089479041,
|
6612 |
+
"grad_norm": 0.029655590653419495,
|
6613 |
+
"learning_rate": 0.00019844971067388108,
|
6614 |
+
"loss": 0.0011,
|
6615 |
+
"step": 938
|
6616 |
+
},
|
6617 |
+
{
|
6618 |
+
"epoch": 0.17377625613028594,
|
6619 |
+
"grad_norm": 0.005666040349751711,
|
6620 |
+
"learning_rate": 0.00019844630293421338,
|
6621 |
+
"loss": 0.0002,
|
6622 |
+
"step": 939
|
6623 |
+
},
|
6624 |
+
{
|
6625 |
+
"epoch": 0.17396132136578144,
|
6626 |
+
"grad_norm": 0.0009374783257953823,
|
6627 |
+
"learning_rate": 0.00019844289148264962,
|
6628 |
+
"loss": 0.0001,
|
6629 |
+
"step": 940
|
6630 |
+
},
|
6631 |
+
{
|
6632 |
+
"epoch": 0.17414638660127696,
|
6633 |
+
"grad_norm": 0.005437355954200029,
|
6634 |
+
"learning_rate": 0.00019843947631931842,
|
6635 |
+
"loss": 0.0003,
|
6636 |
+
"step": 941
|
6637 |
+
},
|
6638 |
+
{
|
6639 |
+
"epoch": 0.17433145183677246,
|
6640 |
+
"grad_norm": 0.14348870515823364,
|
6641 |
+
"learning_rate": 0.0001984360574443486,
|
6642 |
+
"loss": 0.0043,
|
6643 |
+
"step": 942
|
6644 |
+
},
|
6645 |
+
{
|
6646 |
+
"epoch": 0.17451651707226798,
|
6647 |
+
"grad_norm": 0.19243773818016052,
|
6648 |
+
"learning_rate": 0.00019843263485786898,
|
6649 |
+
"loss": 0.0012,
|
6650 |
+
"step": 943
|
6651 |
+
},
|
6652 |
+
{
|
6653 |
+
"epoch": 0.17470158230776348,
|
6654 |
+
"grad_norm": 0.008438481949269772,
|
6655 |
+
"learning_rate": 0.0001984292085600087,
|
6656 |
+
"loss": 0.0003,
|
6657 |
+
"step": 944
|
6658 |
+
},
|
6659 |
+
{
|
6660 |
+
"epoch": 0.174886647543259,
|
6661 |
+
"grad_norm": 0.0023594368249177933,
|
6662 |
+
"learning_rate": 0.0001984257785508969,
|
6663 |
+
"loss": 0.0002,
|
6664 |
+
"step": 945
|
6665 |
+
},
|
6666 |
+
{
|
6667 |
+
"epoch": 0.17507171277875452,
|
6668 |
+
"grad_norm": 0.005180665757507086,
|
6669 |
+
"learning_rate": 0.00019842234483066293,
|
6670 |
+
"loss": 0.0003,
|
6671 |
+
"step": 946
|
6672 |
+
},
|
6673 |
+
{
|
6674 |
+
"epoch": 0.17525677801425002,
|
6675 |
+
"grad_norm": 0.0011170993093401194,
|
6676 |
+
"learning_rate": 0.00019841890739943625,
|
6677 |
+
"loss": 0.0001,
|
6678 |
+
"step": 947
|
6679 |
+
},
|
6680 |
+
{
|
6681 |
+
"epoch": 0.17544184324974554,
|
6682 |
+
"grad_norm": 0.0031029570382088423,
|
6683 |
+
"learning_rate": 0.00019841546625734647,
|
6684 |
+
"loss": 0.0002,
|
6685 |
+
"step": 948
|
6686 |
+
},
|
6687 |
+
{
|
6688 |
+
"epoch": 0.17562690848524104,
|
6689 |
+
"grad_norm": 0.001714157173410058,
|
6690 |
+
"learning_rate": 0.0001984120214045233,
|
6691 |
+
"loss": 0.0002,
|
6692 |
+
"step": 949
|
6693 |
+
},
|
6694 |
+
{
|
6695 |
+
"epoch": 0.17581197372073656,
|
6696 |
+
"grad_norm": 0.0033893748186528683,
|
6697 |
+
"learning_rate": 0.0001984085728410967,
|
6698 |
+
"loss": 0.0002,
|
6699 |
+
"step": 950
|
6700 |
+
},
|
6701 |
+
{
|
6702 |
+
"epoch": 0.17599703895623206,
|
6703 |
+
"grad_norm": 0.004998108372092247,
|
6704 |
+
"learning_rate": 0.00019840512056719666,
|
6705 |
+
"loss": 0.0003,
|
6706 |
+
"step": 951
|
6707 |
+
},
|
6708 |
+
{
|
6709 |
+
"epoch": 0.17618210419172758,
|
6710 |
+
"grad_norm": 0.0266425758600235,
|
6711 |
+
"learning_rate": 0.0001984016645829533,
|
6712 |
+
"loss": 0.0006,
|
6713 |
+
"step": 952
|
6714 |
+
},
|
6715 |
+
{
|
6716 |
+
"epoch": 0.1763671694272231,
|
6717 |
+
"grad_norm": 0.0037740166299045086,
|
6718 |
+
"learning_rate": 0.000198398204888497,
|
6719 |
+
"loss": 0.0002,
|
6720 |
+
"step": 953
|
6721 |
+
},
|
6722 |
+
{
|
6723 |
+
"epoch": 0.1765522346627186,
|
6724 |
+
"grad_norm": 0.0013826234498992562,
|
6725 |
+
"learning_rate": 0.00019839474148395818,
|
6726 |
+
"loss": 0.0002,
|
6727 |
+
"step": 954
|
6728 |
+
},
|
6729 |
+
{
|
6730 |
+
"epoch": 0.17673729989821413,
|
6731 |
+
"grad_norm": 0.0025008702650666237,
|
6732 |
+
"learning_rate": 0.0001983912743694674,
|
6733 |
+
"loss": 0.0002,
|
6734 |
+
"step": 955
|
6735 |
+
},
|
6736 |
+
{
|
6737 |
+
"epoch": 0.17692236513370962,
|
6738 |
+
"grad_norm": 0.006042696535587311,
|
6739 |
+
"learning_rate": 0.00019838780354515545,
|
6740 |
+
"loss": 0.0004,
|
6741 |
+
"step": 956
|
6742 |
+
},
|
6743 |
+
{
|
6744 |
+
"epoch": 0.17710743036920515,
|
6745 |
+
"grad_norm": 0.385301411151886,
|
6746 |
+
"learning_rate": 0.00019838432901115314,
|
6747 |
+
"loss": 0.0049,
|
6748 |
+
"step": 957
|
6749 |
+
},
|
6750 |
+
{
|
6751 |
+
"epoch": 0.17729249560470065,
|
6752 |
+
"grad_norm": 0.0020321677438914776,
|
6753 |
+
"learning_rate": 0.0001983808507675915,
|
6754 |
+
"loss": 0.0001,
|
6755 |
+
"step": 958
|
6756 |
+
},
|
6757 |
+
{
|
6758 |
+
"epoch": 0.17747756084019617,
|
6759 |
+
"grad_norm": 0.008854767307639122,
|
6760 |
+
"learning_rate": 0.00019837736881460165,
|
6761 |
+
"loss": 0.0002,
|
6762 |
+
"step": 959
|
6763 |
+
},
|
6764 |
+
{
|
6765 |
+
"epoch": 0.1776626260756917,
|
6766 |
+
"grad_norm": 0.0018682387890294194,
|
6767 |
+
"learning_rate": 0.00019837388315231493,
|
6768 |
+
"loss": 0.0002,
|
6769 |
+
"step": 960
|
6770 |
+
},
|
6771 |
+
{
|
6772 |
+
"epoch": 0.1778476913111872,
|
6773 |
+
"grad_norm": 0.001613518688827753,
|
6774 |
+
"learning_rate": 0.0001983703937808627,
|
6775 |
+
"loss": 0.0001,
|
6776 |
+
"step": 961
|
6777 |
+
},
|
6778 |
+
{
|
6779 |
+
"epoch": 0.1780327565466827,
|
6780 |
+
"grad_norm": 0.002380665624514222,
|
6781 |
+
"learning_rate": 0.0001983669007003766,
|
6782 |
+
"loss": 0.0002,
|
6783 |
+
"step": 962
|
6784 |
+
},
|
6785 |
+
{
|
6786 |
+
"epoch": 0.1782178217821782,
|
6787 |
+
"grad_norm": 0.004421447869390249,
|
6788 |
+
"learning_rate": 0.00019836340391098827,
|
6789 |
+
"loss": 0.0002,
|
6790 |
+
"step": 963
|
6791 |
+
},
|
6792 |
+
{
|
6793 |
+
"epoch": 0.17840288701767373,
|
6794 |
+
"grad_norm": 0.004585796967148781,
|
6795 |
+
"learning_rate": 0.00019835990341282957,
|
6796 |
+
"loss": 0.0002,
|
6797 |
+
"step": 964
|
6798 |
+
},
|
6799 |
+
{
|
6800 |
+
"epoch": 0.17858795225316923,
|
6801 |
+
"grad_norm": 0.016321241855621338,
|
6802 |
+
"learning_rate": 0.00019835639920603253,
|
6803 |
+
"loss": 0.0006,
|
6804 |
+
"step": 965
|
6805 |
+
},
|
6806 |
+
{
|
6807 |
+
"epoch": 0.17877301748866475,
|
6808 |
+
"grad_norm": 0.3031005859375,
|
6809 |
+
"learning_rate": 0.00019835289129072924,
|
6810 |
+
"loss": 0.0029,
|
6811 |
+
"step": 966
|
6812 |
+
},
|
6813 |
+
{
|
6814 |
+
"epoch": 0.17895808272416028,
|
6815 |
+
"grad_norm": 0.0025489667896181345,
|
6816 |
+
"learning_rate": 0.00019834937966705193,
|
6817 |
+
"loss": 0.0002,
|
6818 |
+
"step": 967
|
6819 |
+
},
|
6820 |
+
{
|
6821 |
+
"epoch": 0.17914314795965577,
|
6822 |
+
"grad_norm": 0.0011104342993348837,
|
6823 |
+
"learning_rate": 0.00019834586433513306,
|
6824 |
+
"loss": 0.0001,
|
6825 |
+
"step": 968
|
6826 |
+
},
|
6827 |
+
{
|
6828 |
+
"epoch": 0.1793282131951513,
|
6829 |
+
"grad_norm": 0.012867497280240059,
|
6830 |
+
"learning_rate": 0.00019834234529510518,
|
6831 |
+
"loss": 0.0002,
|
6832 |
+
"step": 969
|
6833 |
+
},
|
6834 |
+
{
|
6835 |
+
"epoch": 0.1795132784306468,
|
6836 |
+
"grad_norm": 0.09900082647800446,
|
6837 |
+
"learning_rate": 0.00019833882254710093,
|
6838 |
+
"loss": 0.0044,
|
6839 |
+
"step": 970
|
6840 |
+
},
|
6841 |
+
{
|
6842 |
+
"epoch": 0.17969834366614232,
|
6843 |
+
"grad_norm": 0.1001380980014801,
|
6844 |
+
"learning_rate": 0.00019833529609125316,
|
6845 |
+
"loss": 0.0024,
|
6846 |
+
"step": 971
|
6847 |
+
},
|
6848 |
+
{
|
6849 |
+
"epoch": 0.17988340890163781,
|
6850 |
+
"grad_norm": 0.005729401484131813,
|
6851 |
+
"learning_rate": 0.00019833176592769483,
|
6852 |
+
"loss": 0.0005,
|
6853 |
+
"step": 972
|
6854 |
+
},
|
6855 |
+
{
|
6856 |
+
"epoch": 0.18006847413713334,
|
6857 |
+
"grad_norm": 0.007758236024528742,
|
6858 |
+
"learning_rate": 0.00019832823205655906,
|
6859 |
+
"loss": 0.0004,
|
6860 |
+
"step": 973
|
6861 |
+
},
|
6862 |
+
{
|
6863 |
+
"epoch": 0.18025353937262886,
|
6864 |
+
"grad_norm": 0.0032970982138067484,
|
6865 |
+
"learning_rate": 0.00019832469447797905,
|
6866 |
+
"loss": 0.0002,
|
6867 |
+
"step": 974
|
6868 |
+
},
|
6869 |
+
{
|
6870 |
+
"epoch": 0.18043860460812436,
|
6871 |
+
"grad_norm": 0.0036083597224205732,
|
6872 |
+
"learning_rate": 0.00019832115319208825,
|
6873 |
+
"loss": 0.0003,
|
6874 |
+
"step": 975
|
6875 |
+
},
|
6876 |
+
{
|
6877 |
+
"epoch": 0.18062366984361988,
|
6878 |
+
"grad_norm": 0.13452349603176117,
|
6879 |
+
"learning_rate": 0.0001983176081990201,
|
6880 |
+
"loss": 0.0035,
|
6881 |
+
"step": 976
|
6882 |
+
},
|
6883 |
+
{
|
6884 |
+
"epoch": 0.18080873507911538,
|
6885 |
+
"grad_norm": 0.008456684648990631,
|
6886 |
+
"learning_rate": 0.00019831405949890832,
|
6887 |
+
"loss": 0.0004,
|
6888 |
+
"step": 977
|
6889 |
+
},
|
6890 |
+
{
|
6891 |
+
"epoch": 0.1809938003146109,
|
6892 |
+
"grad_norm": 0.0014329359401017427,
|
6893 |
+
"learning_rate": 0.00019831050709188675,
|
6894 |
+
"loss": 0.0002,
|
6895 |
+
"step": 978
|
6896 |
+
},
|
6897 |
+
{
|
6898 |
+
"epoch": 0.1811788655501064,
|
6899 |
+
"grad_norm": 0.008513165637850761,
|
6900 |
+
"learning_rate": 0.00019830695097808923,
|
6901 |
+
"loss": 0.0005,
|
6902 |
+
"step": 979
|
6903 |
+
},
|
6904 |
+
{
|
6905 |
+
"epoch": 0.18136393078560192,
|
6906 |
+
"grad_norm": 0.0037564258091151714,
|
6907 |
+
"learning_rate": 0.0001983033911576499,
|
6908 |
+
"loss": 0.0002,
|
6909 |
+
"step": 980
|
6910 |
+
},
|
6911 |
+
{
|
6912 |
+
"epoch": 0.18154899602109745,
|
6913 |
+
"grad_norm": 0.20619681477546692,
|
6914 |
+
"learning_rate": 0.00019829982763070302,
|
6915 |
+
"loss": 0.0119,
|
6916 |
+
"step": 981
|
6917 |
+
},
|
6918 |
+
{
|
6919 |
+
"epoch": 0.18173406125659294,
|
6920 |
+
"grad_norm": 0.09338407963514328,
|
6921 |
+
"learning_rate": 0.00019829626039738287,
|
6922 |
+
"loss": 0.0052,
|
6923 |
+
"step": 982
|
6924 |
+
},
|
6925 |
+
{
|
6926 |
+
"epoch": 0.18191912649208847,
|
6927 |
+
"grad_norm": 0.004537677392363548,
|
6928 |
+
"learning_rate": 0.000198292689457824,
|
6929 |
+
"loss": 0.0002,
|
6930 |
+
"step": 983
|
6931 |
+
},
|
6932 |
+
{
|
6933 |
+
"epoch": 0.18210419172758396,
|
6934 |
+
"grad_norm": 0.02287222072482109,
|
6935 |
+
"learning_rate": 0.00019828911481216107,
|
6936 |
+
"loss": 0.0006,
|
6937 |
+
"step": 984
|
6938 |
+
},
|
6939 |
+
{
|
6940 |
+
"epoch": 0.1822892569630795,
|
6941 |
+
"grad_norm": 0.10385712236166,
|
6942 |
+
"learning_rate": 0.0001982855364605288,
|
6943 |
+
"loss": 0.0028,
|
6944 |
+
"step": 985
|
6945 |
+
},
|
6946 |
+
{
|
6947 |
+
"epoch": 0.18247432219857498,
|
6948 |
+
"grad_norm": 0.0036787688732147217,
|
6949 |
+
"learning_rate": 0.00019828195440306216,
|
6950 |
+
"loss": 0.0002,
|
6951 |
+
"step": 986
|
6952 |
+
},
|
6953 |
+
{
|
6954 |
+
"epoch": 0.1826593874340705,
|
6955 |
+
"grad_norm": 0.1489401012659073,
|
6956 |
+
"learning_rate": 0.00019827836863989623,
|
6957 |
+
"loss": 0.006,
|
6958 |
+
"step": 987
|
6959 |
+
},
|
6960 |
+
{
|
6961 |
+
"epoch": 0.18284445266956603,
|
6962 |
+
"grad_norm": 0.005185346119105816,
|
6963 |
+
"learning_rate": 0.00019827477917116614,
|
6964 |
+
"loss": 0.0002,
|
6965 |
+
"step": 988
|
6966 |
+
},
|
6967 |
+
{
|
6968 |
+
"epoch": 0.18302951790506153,
|
6969 |
+
"grad_norm": 0.009372856467962265,
|
6970 |
+
"learning_rate": 0.00019827118599700726,
|
6971 |
+
"loss": 0.0003,
|
6972 |
+
"step": 989
|
6973 |
+
},
|
6974 |
+
{
|
6975 |
+
"epoch": 0.18321458314055705,
|
6976 |
+
"grad_norm": 0.006225073244422674,
|
6977 |
+
"learning_rate": 0.0001982675891175551,
|
6978 |
+
"loss": 0.0005,
|
6979 |
+
"step": 990
|
6980 |
+
},
|
6981 |
+
{
|
6982 |
+
"epoch": 0.18339964837605255,
|
6983 |
+
"grad_norm": 0.12533916532993317,
|
6984 |
+
"learning_rate": 0.00019826398853294526,
|
6985 |
+
"loss": 0.0037,
|
6986 |
+
"step": 991
|
6987 |
+
},
|
6988 |
+
{
|
6989 |
+
"epoch": 0.18358471361154807,
|
6990 |
+
"grad_norm": 0.002677407581359148,
|
6991 |
+
"learning_rate": 0.00019826038424331348,
|
6992 |
+
"loss": 0.0002,
|
6993 |
+
"step": 992
|
6994 |
+
},
|
6995 |
+
{
|
6996 |
+
"epoch": 0.1837697788470436,
|
6997 |
+
"grad_norm": 0.002158583840355277,
|
6998 |
+
"learning_rate": 0.0001982567762487957,
|
6999 |
+
"loss": 0.0002,
|
7000 |
+
"step": 993
|
7001 |
+
},
|
7002 |
+
{
|
7003 |
+
"epoch": 0.1839548440825391,
|
7004 |
+
"grad_norm": 0.0065282066352665424,
|
7005 |
+
"learning_rate": 0.00019825316454952788,
|
7006 |
+
"loss": 0.0005,
|
7007 |
+
"step": 994
|
7008 |
+
},
|
7009 |
+
{
|
7010 |
+
"epoch": 0.18413990931803462,
|
7011 |
+
"grad_norm": 0.012533469125628471,
|
7012 |
+
"learning_rate": 0.0001982495491456463,
|
7013 |
+
"loss": 0.0005,
|
7014 |
+
"step": 995
|
7015 |
+
},
|
7016 |
+
{
|
7017 |
+
"epoch": 0.1843249745535301,
|
7018 |
+
"grad_norm": 0.06735439598560333,
|
7019 |
+
"learning_rate": 0.00019824593003728722,
|
7020 |
+
"loss": 0.0007,
|
7021 |
+
"step": 996
|
7022 |
+
},
|
7023 |
+
{
|
7024 |
+
"epoch": 0.18451003978902564,
|
7025 |
+
"grad_norm": 1.4106947183609009,
|
7026 |
+
"learning_rate": 0.0001982423072245871,
|
7027 |
+
"loss": 0.0625,
|
7028 |
+
"step": 997
|
7029 |
+
},
|
7030 |
+
{
|
7031 |
+
"epoch": 0.18469510502452113,
|
7032 |
+
"grad_norm": 0.001669297693297267,
|
7033 |
+
"learning_rate": 0.00019823868070768257,
|
7034 |
+
"loss": 0.0002,
|
7035 |
+
"step": 998
|
7036 |
+
},
|
7037 |
+
{
|
7038 |
+
"epoch": 0.18488017026001666,
|
7039 |
+
"grad_norm": 0.009132337756454945,
|
7040 |
+
"learning_rate": 0.00019823505048671033,
|
7041 |
+
"loss": 0.0003,
|
7042 |
+
"step": 999
|
7043 |
+
},
|
7044 |
+
{
|
7045 |
+
"epoch": 0.18506523549551218,
|
7046 |
+
"grad_norm": 0.1724046766757965,
|
7047 |
+
"learning_rate": 0.00019823141656180723,
|
7048 |
+
"loss": 0.0089,
|
7049 |
+
"step": 1000
|
7050 |
+
},
|
7051 |
+
{
|
7052 |
+
"epoch": 0.18506523549551218,
|
7053 |
+
"eval_loss": 0.004897969774901867,
|
7054 |
+
"eval_runtime": 9.9751,
|
7055 |
+
"eval_samples_per_second": 21.854,
|
7056 |
+
"eval_steps_per_second": 10.927,
|
7057 |
+
"step": 1000
|
7058 |
}
|
7059 |
],
|
7060 |
"logging_steps": 1,
|
|
|
7069 |
"early_stopping_threshold": 0.0
|
7070 |
},
|
7071 |
"attributes": {
|
7072 |
+
"early_stopping_patience_counter": 3
|
7073 |
}
|
7074 |
},
|
7075 |
"TrainerControl": {
|
|
|
7078 |
"should_evaluate": false,
|
7079 |
"should_log": false,
|
7080 |
"should_save": true,
|
7081 |
+
"should_training_stop": true
|
7082 |
},
|
7083 |
"attributes": {}
|
7084 |
}
|
7085 |
},
|
7086 |
+
"total_flos": 3.10903380639744e+16,
|
7087 |
"train_batch_size": 2,
|
7088 |
"trial_name": null,
|
7089 |
"trial_params": null
|