Training in progress, step 881, 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 100059752
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:963d789019eec055b55d8d0df77a361b9d00860e685194befed461d6ac6a82ce
|
3 |
size 100059752
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 51244404
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:37bd0c6ada81cd6cdd2769c64bfc8a35423409aa0a478994e558c9d683ca82f2
|
3 |
size 51244404
|
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 sha256:0e3e9e31194f80109fdbafd3c1e2d72c76110c2f31392f20db890a3fe3baca3d
|
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:aff26be62fe19bc1ef53bf2f0f6ceb3378a857f001e55c5a2722905806c357ff
|
3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch":
|
5 |
"eval_steps": 221,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -4672,6 +4672,1532 @@
|
|
4672 |
"eval_samples_per_second": 15.589,
|
4673 |
"eval_steps_per_second": 7.794,
|
4674 |
"step": 663
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
4675 |
}
|
4676 |
],
|
4677 |
"logging_steps": 1,
|
@@ -4686,12 +6212,12 @@
|
|
4686 |
"should_evaluate": false,
|
4687 |
"should_log": false,
|
4688 |
"should_save": true,
|
4689 |
-
"should_training_stop":
|
4690 |
},
|
4691 |
"attributes": {}
|
4692 |
}
|
4693 |
},
|
4694 |
-
"total_flos": 1.
|
4695 |
"train_batch_size": 2,
|
4696 |
"trial_name": null,
|
4697 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 1.0009226401703335,
|
5 |
"eval_steps": 221,
|
6 |
+
"global_step": 881,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
4672 |
"eval_samples_per_second": 15.589,
|
4673 |
"eval_steps_per_second": 7.794,
|
4674 |
"step": 663
|
4675 |
+
},
|
4676 |
+
{
|
4677 |
+
"epoch": 0.7540099361249113,
|
4678 |
+
"grad_norm": 0.7721088528633118,
|
4679 |
+
"learning_rate": 2.9098297132430265e-05,
|
4680 |
+
"loss": 1.6602,
|
4681 |
+
"step": 664
|
4682 |
+
},
|
4683 |
+
{
|
4684 |
+
"epoch": 0.7551454932576295,
|
4685 |
+
"grad_norm": 0.7854897379875183,
|
4686 |
+
"learning_rate": 2.8844404809282978e-05,
|
4687 |
+
"loss": 1.6649,
|
4688 |
+
"step": 665
|
4689 |
+
},
|
4690 |
+
{
|
4691 |
+
"epoch": 0.7562810503903478,
|
4692 |
+
"grad_norm": 0.8402066230773926,
|
4693 |
+
"learning_rate": 2.8591438189879748e-05,
|
4694 |
+
"loss": 1.6524,
|
4695 |
+
"step": 666
|
4696 |
+
},
|
4697 |
+
{
|
4698 |
+
"epoch": 0.757416607523066,
|
4699 |
+
"grad_norm": 0.7977060079574585,
|
4700 |
+
"learning_rate": 2.833940056520772e-05,
|
4701 |
+
"loss": 1.6691,
|
4702 |
+
"step": 667
|
4703 |
+
},
|
4704 |
+
{
|
4705 |
+
"epoch": 0.7585521646557842,
|
4706 |
+
"grad_norm": 0.822481095790863,
|
4707 |
+
"learning_rate": 2.8088295214168147e-05,
|
4708 |
+
"loss": 1.6679,
|
4709 |
+
"step": 668
|
4710 |
+
},
|
4711 |
+
{
|
4712 |
+
"epoch": 0.7596877217885025,
|
4713 |
+
"grad_norm": 0.8438048958778381,
|
4714 |
+
"learning_rate": 2.7838125403533854e-05,
|
4715 |
+
"loss": 1.5296,
|
4716 |
+
"step": 669
|
4717 |
+
},
|
4718 |
+
{
|
4719 |
+
"epoch": 0.7608232789212207,
|
4720 |
+
"grad_norm": 0.8359370231628418,
|
4721 |
+
"learning_rate": 2.7588894387906585e-05,
|
4722 |
+
"loss": 1.588,
|
4723 |
+
"step": 670
|
4724 |
+
},
|
4725 |
+
{
|
4726 |
+
"epoch": 0.761958836053939,
|
4727 |
+
"grad_norm": 0.8355525135993958,
|
4728 |
+
"learning_rate": 2.734060540967499e-05,
|
4729 |
+
"loss": 1.6429,
|
4730 |
+
"step": 671
|
4731 |
+
},
|
4732 |
+
{
|
4733 |
+
"epoch": 0.7630943931866572,
|
4734 |
+
"grad_norm": 0.8638067245483398,
|
4735 |
+
"learning_rate": 2.7093261698972005e-05,
|
4736 |
+
"loss": 1.5093,
|
4737 |
+
"step": 672
|
4738 |
+
},
|
4739 |
+
{
|
4740 |
+
"epoch": 0.7642299503193755,
|
4741 |
+
"grad_norm": 0.8546352982521057,
|
4742 |
+
"learning_rate": 2.6846866473633125e-05,
|
4743 |
+
"loss": 1.5595,
|
4744 |
+
"step": 673
|
4745 |
+
},
|
4746 |
+
{
|
4747 |
+
"epoch": 0.7653655074520936,
|
4748 |
+
"grad_norm": 0.9073834419250488,
|
4749 |
+
"learning_rate": 2.6601422939154407e-05,
|
4750 |
+
"loss": 1.6569,
|
4751 |
+
"step": 674
|
4752 |
+
},
|
4753 |
+
{
|
4754 |
+
"epoch": 0.7665010645848119,
|
4755 |
+
"grad_norm": 0.8955768346786499,
|
4756 |
+
"learning_rate": 2.6356934288650903e-05,
|
4757 |
+
"loss": 1.6477,
|
4758 |
+
"step": 675
|
4759 |
+
},
|
4760 |
+
{
|
4761 |
+
"epoch": 0.7676366217175301,
|
4762 |
+
"grad_norm": 0.8899676203727722,
|
4763 |
+
"learning_rate": 2.6113403702814998e-05,
|
4764 |
+
"loss": 1.5093,
|
4765 |
+
"step": 676
|
4766 |
+
},
|
4767 |
+
{
|
4768 |
+
"epoch": 0.7687721788502484,
|
4769 |
+
"grad_norm": 0.872996985912323,
|
4770 |
+
"learning_rate": 2.587083434987505e-05,
|
4771 |
+
"loss": 1.4281,
|
4772 |
+
"step": 677
|
4773 |
+
},
|
4774 |
+
{
|
4775 |
+
"epoch": 0.7699077359829667,
|
4776 |
+
"grad_norm": 0.9623343348503113,
|
4777 |
+
"learning_rate": 2.5629229385554142e-05,
|
4778 |
+
"loss": 1.5363,
|
4779 |
+
"step": 678
|
4780 |
+
},
|
4781 |
+
{
|
4782 |
+
"epoch": 0.7710432931156849,
|
4783 |
+
"grad_norm": 0.9317876696586609,
|
4784 |
+
"learning_rate": 2.538859195302922e-05,
|
4785 |
+
"loss": 1.451,
|
4786 |
+
"step": 679
|
4787 |
+
},
|
4788 |
+
{
|
4789 |
+
"epoch": 0.7721788502484032,
|
4790 |
+
"grad_norm": 1.059510350227356,
|
4791 |
+
"learning_rate": 2.514892518288988e-05,
|
4792 |
+
"loss": 1.5472,
|
4793 |
+
"step": 680
|
4794 |
+
},
|
4795 |
+
{
|
4796 |
+
"epoch": 0.7733144073811213,
|
4797 |
+
"grad_norm": 0.9734624624252319,
|
4798 |
+
"learning_rate": 2.4910232193097994e-05,
|
4799 |
+
"loss": 1.4922,
|
4800 |
+
"step": 681
|
4801 |
+
},
|
4802 |
+
{
|
4803 |
+
"epoch": 0.7744499645138396,
|
4804 |
+
"grad_norm": 0.951866090297699,
|
4805 |
+
"learning_rate": 2.467251608894683e-05,
|
4806 |
+
"loss": 1.3553,
|
4807 |
+
"step": 682
|
4808 |
+
},
|
4809 |
+
{
|
4810 |
+
"epoch": 0.7755855216465578,
|
4811 |
+
"grad_norm": 1.0854949951171875,
|
4812 |
+
"learning_rate": 2.443577996302081e-05,
|
4813 |
+
"loss": 1.5345,
|
4814 |
+
"step": 683
|
4815 |
+
},
|
4816 |
+
{
|
4817 |
+
"epoch": 0.7767210787792761,
|
4818 |
+
"grad_norm": 1.0024741888046265,
|
4819 |
+
"learning_rate": 2.420002689515537e-05,
|
4820 |
+
"loss": 1.3282,
|
4821 |
+
"step": 684
|
4822 |
+
},
|
4823 |
+
{
|
4824 |
+
"epoch": 0.7778566359119943,
|
4825 |
+
"grad_norm": 1.0837198495864868,
|
4826 |
+
"learning_rate": 2.3965259952396646e-05,
|
4827 |
+
"loss": 1.5353,
|
4828 |
+
"step": 685
|
4829 |
+
},
|
4830 |
+
{
|
4831 |
+
"epoch": 0.7789921930447126,
|
4832 |
+
"grad_norm": 1.062203049659729,
|
4833 |
+
"learning_rate": 2.3731482188961818e-05,
|
4834 |
+
"loss": 1.4269,
|
4835 |
+
"step": 686
|
4836 |
+
},
|
4837 |
+
{
|
4838 |
+
"epoch": 0.7801277501774309,
|
4839 |
+
"grad_norm": 1.1522090435028076,
|
4840 |
+
"learning_rate": 2.349869664619917e-05,
|
4841 |
+
"loss": 1.4395,
|
4842 |
+
"step": 687
|
4843 |
+
},
|
4844 |
+
{
|
4845 |
+
"epoch": 0.781263307310149,
|
4846 |
+
"grad_norm": 1.122259259223938,
|
4847 |
+
"learning_rate": 2.326690635254872e-05,
|
4848 |
+
"loss": 1.4416,
|
4849 |
+
"step": 688
|
4850 |
+
},
|
4851 |
+
{
|
4852 |
+
"epoch": 0.7823988644428673,
|
4853 |
+
"grad_norm": 1.2164281606674194,
|
4854 |
+
"learning_rate": 2.3036114323502655e-05,
|
4855 |
+
"loss": 1.4305,
|
4856 |
+
"step": 689
|
4857 |
+
},
|
4858 |
+
{
|
4859 |
+
"epoch": 0.7835344215755855,
|
4860 |
+
"grad_norm": 1.2222996950149536,
|
4861 |
+
"learning_rate": 2.2806323561566146e-05,
|
4862 |
+
"loss": 1.403,
|
4863 |
+
"step": 690
|
4864 |
+
},
|
4865 |
+
{
|
4866 |
+
"epoch": 0.7846699787083038,
|
4867 |
+
"grad_norm": 1.3191629648208618,
|
4868 |
+
"learning_rate": 2.257753705621839e-05,
|
4869 |
+
"loss": 1.3559,
|
4870 |
+
"step": 691
|
4871 |
+
},
|
4872 |
+
{
|
4873 |
+
"epoch": 0.785805535841022,
|
4874 |
+
"grad_norm": 1.3164441585540771,
|
4875 |
+
"learning_rate": 2.2349757783873627e-05,
|
4876 |
+
"loss": 1.2384,
|
4877 |
+
"step": 692
|
4878 |
+
},
|
4879 |
+
{
|
4880 |
+
"epoch": 0.7869410929737403,
|
4881 |
+
"grad_norm": 1.3931750059127808,
|
4882 |
+
"learning_rate": 2.2122988707842353e-05,
|
4883 |
+
"loss": 1.4181,
|
4884 |
+
"step": 693
|
4885 |
+
},
|
4886 |
+
{
|
4887 |
+
"epoch": 0.7880766501064584,
|
4888 |
+
"grad_norm": 1.4053387641906738,
|
4889 |
+
"learning_rate": 2.18972327782929e-05,
|
4890 |
+
"loss": 1.2748,
|
4891 |
+
"step": 694
|
4892 |
+
},
|
4893 |
+
{
|
4894 |
+
"epoch": 0.7892122072391767,
|
4895 |
+
"grad_norm": 1.4738166332244873,
|
4896 |
+
"learning_rate": 2.167249293221293e-05,
|
4897 |
+
"loss": 1.2564,
|
4898 |
+
"step": 695
|
4899 |
+
},
|
4900 |
+
{
|
4901 |
+
"epoch": 0.7903477643718949,
|
4902 |
+
"grad_norm": 1.6781202554702759,
|
4903 |
+
"learning_rate": 2.144877209337145e-05,
|
4904 |
+
"loss": 1.3486,
|
4905 |
+
"step": 696
|
4906 |
+
},
|
4907 |
+
{
|
4908 |
+
"epoch": 0.7914833215046132,
|
4909 |
+
"grad_norm": 1.7127809524536133,
|
4910 |
+
"learning_rate": 2.122607317228049e-05,
|
4911 |
+
"loss": 1.1686,
|
4912 |
+
"step": 697
|
4913 |
+
},
|
4914 |
+
{
|
4915 |
+
"epoch": 0.7926188786373315,
|
4916 |
+
"grad_norm": 1.9314239025115967,
|
4917 |
+
"learning_rate": 2.100439906615739e-05,
|
4918 |
+
"loss": 1.192,
|
4919 |
+
"step": 698
|
4920 |
+
},
|
4921 |
+
{
|
4922 |
+
"epoch": 0.7937544357700497,
|
4923 |
+
"grad_norm": 2.266965389251709,
|
4924 |
+
"learning_rate": 2.0783752658887066e-05,
|
4925 |
+
"loss": 0.9822,
|
4926 |
+
"step": 699
|
4927 |
+
},
|
4928 |
+
{
|
4929 |
+
"epoch": 0.794889992902768,
|
4930 |
+
"grad_norm": 3.223053216934204,
|
4931 |
+
"learning_rate": 2.056413682098459e-05,
|
4932 |
+
"loss": 1.1714,
|
4933 |
+
"step": 700
|
4934 |
+
},
|
4935 |
+
{
|
4936 |
+
"epoch": 0.7960255500354861,
|
4937 |
+
"grad_norm": 0.675081193447113,
|
4938 |
+
"learning_rate": 2.034555440955773e-05,
|
4939 |
+
"loss": 1.9009,
|
4940 |
+
"step": 701
|
4941 |
+
},
|
4942 |
+
{
|
4943 |
+
"epoch": 0.7971611071682044,
|
4944 |
+
"grad_norm": 0.6774374842643738,
|
4945 |
+
"learning_rate": 2.0128008268269815e-05,
|
4946 |
+
"loss": 1.8244,
|
4947 |
+
"step": 702
|
4948 |
+
},
|
4949 |
+
{
|
4950 |
+
"epoch": 0.7982966643009226,
|
4951 |
+
"grad_norm": 0.686852216720581,
|
4952 |
+
"learning_rate": 1.9911501227302687e-05,
|
4953 |
+
"loss": 1.7523,
|
4954 |
+
"step": 703
|
4955 |
+
},
|
4956 |
+
{
|
4957 |
+
"epoch": 0.7994322214336409,
|
4958 |
+
"grad_norm": 0.7346818447113037,
|
4959 |
+
"learning_rate": 1.969603610332007e-05,
|
4960 |
+
"loss": 1.7847,
|
4961 |
+
"step": 704
|
4962 |
+
},
|
4963 |
+
{
|
4964 |
+
"epoch": 0.8005677785663591,
|
4965 |
+
"grad_norm": 0.7139765620231628,
|
4966 |
+
"learning_rate": 1.9481615699430654e-05,
|
4967 |
+
"loss": 1.6362,
|
4968 |
+
"step": 705
|
4969 |
+
},
|
4970 |
+
{
|
4971 |
+
"epoch": 0.8017033356990774,
|
4972 |
+
"grad_norm": 0.7906776070594788,
|
4973 |
+
"learning_rate": 1.9268242805151902e-05,
|
4974 |
+
"loss": 1.8393,
|
4975 |
+
"step": 706
|
4976 |
+
},
|
4977 |
+
{
|
4978 |
+
"epoch": 0.8028388928317955,
|
4979 |
+
"grad_norm": 0.7764397859573364,
|
4980 |
+
"learning_rate": 1.9055920196373523e-05,
|
4981 |
+
"loss": 1.8045,
|
4982 |
+
"step": 707
|
4983 |
+
},
|
4984 |
+
{
|
4985 |
+
"epoch": 0.8039744499645138,
|
4986 |
+
"grad_norm": 0.7411718368530273,
|
4987 |
+
"learning_rate": 1.8844650635321483e-05,
|
4988 |
+
"loss": 1.6908,
|
4989 |
+
"step": 708
|
4990 |
+
},
|
4991 |
+
{
|
4992 |
+
"epoch": 0.8051100070972321,
|
4993 |
+
"grad_norm": 0.7711935639381409,
|
4994 |
+
"learning_rate": 1.863443687052211e-05,
|
4995 |
+
"loss": 1.7031,
|
4996 |
+
"step": 709
|
4997 |
+
},
|
4998 |
+
{
|
4999 |
+
"epoch": 0.8062455642299503,
|
5000 |
+
"grad_norm": 0.7640191912651062,
|
5001 |
+
"learning_rate": 1.842528163676619e-05,
|
5002 |
+
"loss": 1.7085,
|
5003 |
+
"step": 710
|
5004 |
+
},
|
5005 |
+
{
|
5006 |
+
"epoch": 0.8073811213626686,
|
5007 |
+
"grad_norm": 0.8145517110824585,
|
5008 |
+
"learning_rate": 1.8217187655073564e-05,
|
5009 |
+
"loss": 1.7607,
|
5010 |
+
"step": 711
|
5011 |
+
},
|
5012 |
+
{
|
5013 |
+
"epoch": 0.8085166784953868,
|
5014 |
+
"grad_norm": 0.7967603206634521,
|
5015 |
+
"learning_rate": 1.8010157632657543e-05,
|
5016 |
+
"loss": 1.6411,
|
5017 |
+
"step": 712
|
5018 |
+
},
|
5019 |
+
{
|
5020 |
+
"epoch": 0.8096522356281051,
|
5021 |
+
"grad_norm": 0.8301655054092407,
|
5022 |
+
"learning_rate": 1.7804194262889874e-05,
|
5023 |
+
"loss": 1.6753,
|
5024 |
+
"step": 713
|
5025 |
+
},
|
5026 |
+
{
|
5027 |
+
"epoch": 0.8107877927608232,
|
5028 |
+
"grad_norm": 0.7865291237831116,
|
5029 |
+
"learning_rate": 1.759930022526556e-05,
|
5030 |
+
"loss": 1.641,
|
5031 |
+
"step": 714
|
5032 |
+
},
|
5033 |
+
{
|
5034 |
+
"epoch": 0.8119233498935415,
|
5035 |
+
"grad_norm": 0.835369884967804,
|
5036 |
+
"learning_rate": 1.739547818536804e-05,
|
5037 |
+
"loss": 1.6512,
|
5038 |
+
"step": 715
|
5039 |
+
},
|
5040 |
+
{
|
5041 |
+
"epoch": 0.8130589070262597,
|
5042 |
+
"grad_norm": 0.8319673538208008,
|
5043 |
+
"learning_rate": 1.7192730794834556e-05,
|
5044 |
+
"loss": 1.651,
|
5045 |
+
"step": 716
|
5046 |
+
},
|
5047 |
+
{
|
5048 |
+
"epoch": 0.814194464158978,
|
5049 |
+
"grad_norm": 0.8641321659088135,
|
5050 |
+
"learning_rate": 1.699106069132165e-05,
|
5051 |
+
"loss": 1.674,
|
5052 |
+
"step": 717
|
5053 |
+
},
|
5054 |
+
{
|
5055 |
+
"epoch": 0.8153300212916962,
|
5056 |
+
"grad_norm": 0.8129961490631104,
|
5057 |
+
"learning_rate": 1.6790470498470744e-05,
|
5058 |
+
"loss": 1.5624,
|
5059 |
+
"step": 718
|
5060 |
+
},
|
5061 |
+
{
|
5062 |
+
"epoch": 0.8164655784244145,
|
5063 |
+
"grad_norm": 0.8343265056610107,
|
5064 |
+
"learning_rate": 1.6590962825874146e-05,
|
5065 |
+
"loss": 1.6264,
|
5066 |
+
"step": 719
|
5067 |
+
},
|
5068 |
+
{
|
5069 |
+
"epoch": 0.8176011355571328,
|
5070 |
+
"grad_norm": 0.8597946763038635,
|
5071 |
+
"learning_rate": 1.639254026904099e-05,
|
5072 |
+
"loss": 1.5274,
|
5073 |
+
"step": 720
|
5074 |
+
},
|
5075 |
+
{
|
5076 |
+
"epoch": 0.8187366926898509,
|
5077 |
+
"grad_norm": 0.8967453837394714,
|
5078 |
+
"learning_rate": 1.6195205409363577e-05,
|
5079 |
+
"loss": 1.5423,
|
5080 |
+
"step": 721
|
5081 |
+
},
|
5082 |
+
{
|
5083 |
+
"epoch": 0.8198722498225692,
|
5084 |
+
"grad_norm": 0.8765518069267273,
|
5085 |
+
"learning_rate": 1.599896081408373e-05,
|
5086 |
+
"loss": 1.6939,
|
5087 |
+
"step": 722
|
5088 |
+
},
|
5089 |
+
{
|
5090 |
+
"epoch": 0.8210078069552874,
|
5091 |
+
"grad_norm": 0.886623740196228,
|
5092 |
+
"learning_rate": 1.5803809036259364e-05,
|
5093 |
+
"loss": 1.558,
|
5094 |
+
"step": 723
|
5095 |
+
},
|
5096 |
+
{
|
5097 |
+
"epoch": 0.8221433640880057,
|
5098 |
+
"grad_norm": 0.9663245677947998,
|
5099 |
+
"learning_rate": 1.5609752614731288e-05,
|
5100 |
+
"loss": 1.6809,
|
5101 |
+
"step": 724
|
5102 |
+
},
|
5103 |
+
{
|
5104 |
+
"epoch": 0.8232789212207239,
|
5105 |
+
"grad_norm": 0.940026044845581,
|
5106 |
+
"learning_rate": 1.5416794074090258e-05,
|
5107 |
+
"loss": 1.5393,
|
5108 |
+
"step": 725
|
5109 |
+
},
|
5110 |
+
{
|
5111 |
+
"epoch": 0.8244144783534422,
|
5112 |
+
"grad_norm": 0.9838845133781433,
|
5113 |
+
"learning_rate": 1.5224935924644069e-05,
|
5114 |
+
"loss": 1.5611,
|
5115 |
+
"step": 726
|
5116 |
+
},
|
5117 |
+
{
|
5118 |
+
"epoch": 0.8255500354861603,
|
5119 |
+
"grad_norm": 1.0032014846801758,
|
5120 |
+
"learning_rate": 1.5034180662384857e-05,
|
5121 |
+
"loss": 1.5944,
|
5122 |
+
"step": 727
|
5123 |
+
},
|
5124 |
+
{
|
5125 |
+
"epoch": 0.8266855926188786,
|
5126 |
+
"grad_norm": 0.991060197353363,
|
5127 |
+
"learning_rate": 1.4844530768956656e-05,
|
5128 |
+
"loss": 1.5018,
|
5129 |
+
"step": 728
|
5130 |
+
},
|
5131 |
+
{
|
5132 |
+
"epoch": 0.8278211497515969,
|
5133 |
+
"grad_norm": 0.9556703567504883,
|
5134 |
+
"learning_rate": 1.4655988711623203e-05,
|
5135 |
+
"loss": 1.5761,
|
5136 |
+
"step": 729
|
5137 |
+
},
|
5138 |
+
{
|
5139 |
+
"epoch": 0.8289567068843151,
|
5140 |
+
"grad_norm": 1.0259032249450684,
|
5141 |
+
"learning_rate": 1.4468556943235678e-05,
|
5142 |
+
"loss": 1.5824,
|
5143 |
+
"step": 730
|
5144 |
+
},
|
5145 |
+
{
|
5146 |
+
"epoch": 0.8300922640170334,
|
5147 |
+
"grad_norm": 1.0417308807373047,
|
5148 |
+
"learning_rate": 1.4282237902200957e-05,
|
5149 |
+
"loss": 1.4678,
|
5150 |
+
"step": 731
|
5151 |
+
},
|
5152 |
+
{
|
5153 |
+
"epoch": 0.8312278211497516,
|
5154 |
+
"grad_norm": 1.0187368392944336,
|
5155 |
+
"learning_rate": 1.409703401244975e-05,
|
5156 |
+
"loss": 1.5475,
|
5157 |
+
"step": 732
|
5158 |
+
},
|
5159 |
+
{
|
5160 |
+
"epoch": 0.8323633782824699,
|
5161 |
+
"grad_norm": 1.1188650131225586,
|
5162 |
+
"learning_rate": 1.391294768340513e-05,
|
5163 |
+
"loss": 1.6506,
|
5164 |
+
"step": 733
|
5165 |
+
},
|
5166 |
+
{
|
5167 |
+
"epoch": 0.833498935415188,
|
5168 |
+
"grad_norm": 1.173802375793457,
|
5169 |
+
"learning_rate": 1.3729981309951245e-05,
|
5170 |
+
"loss": 1.5189,
|
5171 |
+
"step": 734
|
5172 |
+
},
|
5173 |
+
{
|
5174 |
+
"epoch": 0.8346344925479063,
|
5175 |
+
"grad_norm": 1.0794363021850586,
|
5176 |
+
"learning_rate": 1.3548137272402006e-05,
|
5177 |
+
"loss": 1.4119,
|
5178 |
+
"step": 735
|
5179 |
+
},
|
5180 |
+
{
|
5181 |
+
"epoch": 0.8357700496806245,
|
5182 |
+
"grad_norm": 1.0807902812957764,
|
5183 |
+
"learning_rate": 1.3367417936470328e-05,
|
5184 |
+
"loss": 1.3185,
|
5185 |
+
"step": 736
|
5186 |
+
},
|
5187 |
+
{
|
5188 |
+
"epoch": 0.8369056068133428,
|
5189 |
+
"grad_norm": 1.0555070638656616,
|
5190 |
+
"learning_rate": 1.318782565323714e-05,
|
5191 |
+
"loss": 1.2576,
|
5192 |
+
"step": 737
|
5193 |
+
},
|
5194 |
+
{
|
5195 |
+
"epoch": 0.838041163946061,
|
5196 |
+
"grad_norm": 1.1013679504394531,
|
5197 |
+
"learning_rate": 1.300936275912098e-05,
|
5198 |
+
"loss": 1.2131,
|
5199 |
+
"step": 738
|
5200 |
+
},
|
5201 |
+
{
|
5202 |
+
"epoch": 0.8391767210787793,
|
5203 |
+
"grad_norm": 1.1725547313690186,
|
5204 |
+
"learning_rate": 1.2832031575847448e-05,
|
5205 |
+
"loss": 1.3464,
|
5206 |
+
"step": 739
|
5207 |
+
},
|
5208 |
+
{
|
5209 |
+
"epoch": 0.8403122782114976,
|
5210 |
+
"grad_norm": 1.2003037929534912,
|
5211 |
+
"learning_rate": 1.265583441041911e-05,
|
5212 |
+
"loss": 1.3536,
|
5213 |
+
"step": 740
|
5214 |
+
},
|
5215 |
+
{
|
5216 |
+
"epoch": 0.8414478353442157,
|
5217 |
+
"grad_norm": 1.1799437999725342,
|
5218 |
+
"learning_rate": 1.2480773555085434e-05,
|
5219 |
+
"loss": 1.2299,
|
5220 |
+
"step": 741
|
5221 |
+
},
|
5222 |
+
{
|
5223 |
+
"epoch": 0.842583392476934,
|
5224 |
+
"grad_norm": 1.3554902076721191,
|
5225 |
+
"learning_rate": 1.2306851287313025e-05,
|
5226 |
+
"loss": 1.4439,
|
5227 |
+
"step": 742
|
5228 |
+
},
|
5229 |
+
{
|
5230 |
+
"epoch": 0.8437189496096522,
|
5231 |
+
"grad_norm": 1.3761941194534302,
|
5232 |
+
"learning_rate": 1.2134069869755893e-05,
|
5233 |
+
"loss": 1.3434,
|
5234 |
+
"step": 743
|
5235 |
+
},
|
5236 |
+
{
|
5237 |
+
"epoch": 0.8448545067423705,
|
5238 |
+
"grad_norm": 1.4420794248580933,
|
5239 |
+
"learning_rate": 1.1962431550226105e-05,
|
5240 |
+
"loss": 1.315,
|
5241 |
+
"step": 744
|
5242 |
+
},
|
5243 |
+
{
|
5244 |
+
"epoch": 0.8459900638750887,
|
5245 |
+
"grad_norm": 1.3914544582366943,
|
5246 |
+
"learning_rate": 1.1791938561664485e-05,
|
5247 |
+
"loss": 1.1473,
|
5248 |
+
"step": 745
|
5249 |
+
},
|
5250 |
+
{
|
5251 |
+
"epoch": 0.847125621007807,
|
5252 |
+
"grad_norm": 1.5766267776489258,
|
5253 |
+
"learning_rate": 1.1622593122111624e-05,
|
5254 |
+
"loss": 1.3514,
|
5255 |
+
"step": 746
|
5256 |
+
},
|
5257 |
+
{
|
5258 |
+
"epoch": 0.8482611781405252,
|
5259 |
+
"grad_norm": 1.5784944295883179,
|
5260 |
+
"learning_rate": 1.1454397434679021e-05,
|
5261 |
+
"loss": 1.2336,
|
5262 |
+
"step": 747
|
5263 |
+
},
|
5264 |
+
{
|
5265 |
+
"epoch": 0.8493967352732434,
|
5266 |
+
"grad_norm": 1.8584637641906738,
|
5267 |
+
"learning_rate": 1.128735368752033e-05,
|
5268 |
+
"loss": 1.1117,
|
5269 |
+
"step": 748
|
5270 |
+
},
|
5271 |
+
{
|
5272 |
+
"epoch": 0.8505322924059616,
|
5273 |
+
"grad_norm": 2.064455270767212,
|
5274 |
+
"learning_rate": 1.1121464053802965e-05,
|
5275 |
+
"loss": 1.0516,
|
5276 |
+
"step": 749
|
5277 |
+
},
|
5278 |
+
{
|
5279 |
+
"epoch": 0.8516678495386799,
|
5280 |
+
"grad_norm": 2.9727470874786377,
|
5281 |
+
"learning_rate": 1.0956730691679861e-05,
|
5282 |
+
"loss": 1.0157,
|
5283 |
+
"step": 750
|
5284 |
+
},
|
5285 |
+
{
|
5286 |
+
"epoch": 0.8528034066713982,
|
5287 |
+
"grad_norm": 0.5627864599227905,
|
5288 |
+
"learning_rate": 1.0793155744261351e-05,
|
5289 |
+
"loss": 1.816,
|
5290 |
+
"step": 751
|
5291 |
+
},
|
5292 |
+
{
|
5293 |
+
"epoch": 0.8539389638041164,
|
5294 |
+
"grad_norm": 0.6203926205635071,
|
5295 |
+
"learning_rate": 1.0630741339587257e-05,
|
5296 |
+
"loss": 1.8486,
|
5297 |
+
"step": 752
|
5298 |
+
},
|
5299 |
+
{
|
5300 |
+
"epoch": 0.8550745209368347,
|
5301 |
+
"grad_norm": 0.6811462044715881,
|
5302 |
+
"learning_rate": 1.0469489590599257e-05,
|
5303 |
+
"loss": 1.764,
|
5304 |
+
"step": 753
|
5305 |
+
},
|
5306 |
+
{
|
5307 |
+
"epoch": 0.8562100780695528,
|
5308 |
+
"grad_norm": 0.6945324540138245,
|
5309 |
+
"learning_rate": 1.0309402595113338e-05,
|
5310 |
+
"loss": 1.7796,
|
5311 |
+
"step": 754
|
5312 |
+
},
|
5313 |
+
{
|
5314 |
+
"epoch": 0.8573456352022711,
|
5315 |
+
"grad_norm": 0.6891714334487915,
|
5316 |
+
"learning_rate": 1.0150482435792618e-05,
|
5317 |
+
"loss": 1.7623,
|
5318 |
+
"step": 755
|
5319 |
+
},
|
5320 |
+
{
|
5321 |
+
"epoch": 0.8584811923349893,
|
5322 |
+
"grad_norm": 0.7250452637672424,
|
5323 |
+
"learning_rate": 9.992731180120164e-06,
|
5324 |
+
"loss": 1.7345,
|
5325 |
+
"step": 756
|
5326 |
+
},
|
5327 |
+
{
|
5328 |
+
"epoch": 0.8596167494677076,
|
5329 |
+
"grad_norm": 0.7625033259391785,
|
5330 |
+
"learning_rate": 9.836150880372041e-06,
|
5331 |
+
"loss": 1.7901,
|
5332 |
+
"step": 757
|
5333 |
+
},
|
5334 |
+
{
|
5335 |
+
"epoch": 0.8607523066004258,
|
5336 |
+
"grad_norm": 0.746115505695343,
|
5337 |
+
"learning_rate": 9.680743573590733e-06,
|
5338 |
+
"loss": 1.7118,
|
5339 |
+
"step": 758
|
5340 |
+
},
|
5341 |
+
{
|
5342 |
+
"epoch": 0.8618878637331441,
|
5343 |
+
"grad_norm": 0.7322573661804199,
|
5344 |
+
"learning_rate": 9.526511281558593e-06,
|
5345 |
+
"loss": 1.6324,
|
5346 |
+
"step": 759
|
5347 |
+
},
|
5348 |
+
{
|
5349 |
+
"epoch": 0.8630234208658624,
|
5350 |
+
"grad_norm": 0.7643139362335205,
|
5351 |
+
"learning_rate": 9.373456010771509e-06,
|
5352 |
+
"loss": 1.6827,
|
5353 |
+
"step": 760
|
5354 |
+
},
|
5355 |
+
{
|
5356 |
+
"epoch": 0.8641589779985805,
|
5357 |
+
"grad_norm": 0.7623146772384644,
|
5358 |
+
"learning_rate": 9.221579752412856e-06,
|
5359 |
+
"loss": 1.5998,
|
5360 |
+
"step": 761
|
5361 |
+
},
|
5362 |
+
{
|
5363 |
+
"epoch": 0.8652945351312988,
|
5364 |
+
"grad_norm": 0.7586025595664978,
|
5365 |
+
"learning_rate": 9.070884482327524e-06,
|
5366 |
+
"loss": 1.6652,
|
5367 |
+
"step": 762
|
5368 |
+
},
|
5369 |
+
{
|
5370 |
+
"epoch": 0.866430092264017,
|
5371 |
+
"grad_norm": 0.7911468744277954,
|
5372 |
+
"learning_rate": 8.921372160996322e-06,
|
5373 |
+
"loss": 1.6911,
|
5374 |
+
"step": 763
|
5375 |
+
},
|
5376 |
+
{
|
5377 |
+
"epoch": 0.8675656493967353,
|
5378 |
+
"grad_norm": 0.7929603457450867,
|
5379 |
+
"learning_rate": 8.773044733510338e-06,
|
5380 |
+
"loss": 1.6144,
|
5381 |
+
"step": 764
|
5382 |
+
},
|
5383 |
+
{
|
5384 |
+
"epoch": 0.8687012065294535,
|
5385 |
+
"grad_norm": 0.8179545402526855,
|
5386 |
+
"learning_rate": 8.625904129545692e-06,
|
5387 |
+
"loss": 1.6333,
|
5388 |
+
"step": 765
|
5389 |
+
},
|
5390 |
+
{
|
5391 |
+
"epoch": 0.8698367636621718,
|
5392 |
+
"grad_norm": 0.8298175930976868,
|
5393 |
+
"learning_rate": 8.479952263338509e-06,
|
5394 |
+
"loss": 1.6375,
|
5395 |
+
"step": 766
|
5396 |
+
},
|
5397 |
+
{
|
5398 |
+
"epoch": 0.87097232079489,
|
5399 |
+
"grad_norm": 0.8845016956329346,
|
5400 |
+
"learning_rate": 8.335191033659907e-06,
|
5401 |
+
"loss": 1.6268,
|
5402 |
+
"step": 767
|
5403 |
+
},
|
5404 |
+
{
|
5405 |
+
"epoch": 0.8721078779276082,
|
5406 |
+
"grad_norm": 0.8354196548461914,
|
5407 |
+
"learning_rate": 8.191622323791315e-06,
|
5408 |
+
"loss": 1.6815,
|
5409 |
+
"step": 768
|
5410 |
+
},
|
5411 |
+
{
|
5412 |
+
"epoch": 0.8732434350603264,
|
5413 |
+
"grad_norm": 0.8513210415840149,
|
5414 |
+
"learning_rate": 8.049248001500021e-06,
|
5415 |
+
"loss": 1.6325,
|
5416 |
+
"step": 769
|
5417 |
+
},
|
5418 |
+
{
|
5419 |
+
"epoch": 0.8743789921930447,
|
5420 |
+
"grad_norm": 0.8357150554656982,
|
5421 |
+
"learning_rate": 7.908069919014815e-06,
|
5422 |
+
"loss": 1.5492,
|
5423 |
+
"step": 770
|
5424 |
+
},
|
5425 |
+
{
|
5426 |
+
"epoch": 0.875514549325763,
|
5427 |
+
"grad_norm": 0.923727810382843,
|
5428 |
+
"learning_rate": 7.768089913001941e-06,
|
5429 |
+
"loss": 1.6026,
|
5430 |
+
"step": 771
|
5431 |
+
},
|
5432 |
+
{
|
5433 |
+
"epoch": 0.8766501064584812,
|
5434 |
+
"grad_norm": 0.9044084548950195,
|
5435 |
+
"learning_rate": 7.629309804541207e-06,
|
5436 |
+
"loss": 1.577,
|
5437 |
+
"step": 772
|
5438 |
+
},
|
5439 |
+
{
|
5440 |
+
"epoch": 0.8777856635911995,
|
5441 |
+
"grad_norm": 0.8825026750564575,
|
5442 |
+
"learning_rate": 7.491731399102231e-06,
|
5443 |
+
"loss": 1.5215,
|
5444 |
+
"step": 773
|
5445 |
+
},
|
5446 |
+
{
|
5447 |
+
"epoch": 0.8789212207239177,
|
5448 |
+
"grad_norm": 0.9344937205314636,
|
5449 |
+
"learning_rate": 7.355356486520959e-06,
|
5450 |
+
"loss": 1.515,
|
5451 |
+
"step": 774
|
5452 |
+
},
|
5453 |
+
{
|
5454 |
+
"epoch": 0.8800567778566359,
|
5455 |
+
"grad_norm": 0.9843137264251709,
|
5456 |
+
"learning_rate": 7.220186840976495e-06,
|
5457 |
+
"loss": 1.5535,
|
5458 |
+
"step": 775
|
5459 |
+
},
|
5460 |
+
{
|
5461 |
+
"epoch": 0.8811923349893541,
|
5462 |
+
"grad_norm": 0.9081253409385681,
|
5463 |
+
"learning_rate": 7.086224220967907e-06,
|
5464 |
+
"loss": 1.5453,
|
5465 |
+
"step": 776
|
5466 |
+
},
|
5467 |
+
{
|
5468 |
+
"epoch": 0.8823278921220724,
|
5469 |
+
"grad_norm": 0.9501760005950928,
|
5470 |
+
"learning_rate": 6.953470369291348e-06,
|
5471 |
+
"loss": 1.6231,
|
5472 |
+
"step": 777
|
5473 |
+
},
|
5474 |
+
{
|
5475 |
+
"epoch": 0.8834634492547906,
|
5476 |
+
"grad_norm": 0.9582067131996155,
|
5477 |
+
"learning_rate": 6.821927013017426e-06,
|
5478 |
+
"loss": 1.503,
|
5479 |
+
"step": 778
|
5480 |
+
},
|
5481 |
+
{
|
5482 |
+
"epoch": 0.8845990063875089,
|
5483 |
+
"grad_norm": 1.0171207189559937,
|
5484 |
+
"learning_rate": 6.691595863468703e-06,
|
5485 |
+
"loss": 1.6333,
|
5486 |
+
"step": 779
|
5487 |
+
},
|
5488 |
+
{
|
5489 |
+
"epoch": 0.8857345635202271,
|
5490 |
+
"grad_norm": 0.9824504256248474,
|
5491 |
+
"learning_rate": 6.562478616197554e-06,
|
5492 |
+
"loss": 1.4158,
|
5493 |
+
"step": 780
|
5494 |
+
},
|
5495 |
+
{
|
5496 |
+
"epoch": 0.8868701206529453,
|
5497 |
+
"grad_norm": 1.0205291509628296,
|
5498 |
+
"learning_rate": 6.4345769509638776e-06,
|
5499 |
+
"loss": 1.5389,
|
5500 |
+
"step": 781
|
5501 |
+
},
|
5502 |
+
{
|
5503 |
+
"epoch": 0.8880056777856636,
|
5504 |
+
"grad_norm": 0.9647020101547241,
|
5505 |
+
"learning_rate": 6.307892531713444e-06,
|
5506 |
+
"loss": 1.4057,
|
5507 |
+
"step": 782
|
5508 |
+
},
|
5509 |
+
{
|
5510 |
+
"epoch": 0.8891412349183818,
|
5511 |
+
"grad_norm": 1.0226200819015503,
|
5512 |
+
"learning_rate": 6.182427006556135e-06,
|
5513 |
+
"loss": 1.4316,
|
5514 |
+
"step": 783
|
5515 |
+
},
|
5516 |
+
{
|
5517 |
+
"epoch": 0.8902767920511001,
|
5518 |
+
"grad_norm": 1.0292648077011108,
|
5519 |
+
"learning_rate": 6.058182007744584e-06,
|
5520 |
+
"loss": 1.3882,
|
5521 |
+
"step": 784
|
5522 |
+
},
|
5523 |
+
{
|
5524 |
+
"epoch": 0.8914123491838183,
|
5525 |
+
"grad_norm": 1.1634976863861084,
|
5526 |
+
"learning_rate": 5.935159151652902e-06,
|
5527 |
+
"loss": 1.4706,
|
5528 |
+
"step": 785
|
5529 |
+
},
|
5530 |
+
{
|
5531 |
+
"epoch": 0.8925479063165366,
|
5532 |
+
"grad_norm": 0.993560791015625,
|
5533 |
+
"learning_rate": 5.813360038755611e-06,
|
5534 |
+
"loss": 1.2897,
|
5535 |
+
"step": 786
|
5536 |
+
},
|
5537 |
+
{
|
5538 |
+
"epoch": 0.8936834634492548,
|
5539 |
+
"grad_norm": 1.1387815475463867,
|
5540 |
+
"learning_rate": 5.6927862536068635e-06,
|
5541 |
+
"loss": 1.5279,
|
5542 |
+
"step": 787
|
5543 |
+
},
|
5544 |
+
{
|
5545 |
+
"epoch": 0.894819020581973,
|
5546 |
+
"grad_norm": 1.2450308799743652,
|
5547 |
+
"learning_rate": 5.573439364819855e-06,
|
5548 |
+
"loss": 1.4883,
|
5549 |
+
"step": 788
|
5550 |
+
},
|
5551 |
+
{
|
5552 |
+
"epoch": 0.8959545777146912,
|
5553 |
+
"grad_norm": 1.23203444480896,
|
5554 |
+
"learning_rate": 5.455320925046359e-06,
|
5555 |
+
"loss": 1.4462,
|
5556 |
+
"step": 789
|
5557 |
+
},
|
5558 |
+
{
|
5559 |
+
"epoch": 0.8970901348474095,
|
5560 |
+
"grad_norm": 1.1987688541412354,
|
5561 |
+
"learning_rate": 5.338432470956589e-06,
|
5562 |
+
"loss": 1.3306,
|
5563 |
+
"step": 790
|
5564 |
+
},
|
5565 |
+
{
|
5566 |
+
"epoch": 0.8982256919801278,
|
5567 |
+
"grad_norm": 1.291505217552185,
|
5568 |
+
"learning_rate": 5.222775523219125e-06,
|
5569 |
+
"loss": 1.2691,
|
5570 |
+
"step": 791
|
5571 |
+
},
|
5572 |
+
{
|
5573 |
+
"epoch": 0.899361249112846,
|
5574 |
+
"grad_norm": 1.3048465251922607,
|
5575 |
+
"learning_rate": 5.108351586481197e-06,
|
5576 |
+
"loss": 1.4038,
|
5577 |
+
"step": 792
|
5578 |
+
},
|
5579 |
+
{
|
5580 |
+
"epoch": 0.9004968062455643,
|
5581 |
+
"grad_norm": 1.4124550819396973,
|
5582 |
+
"learning_rate": 4.99516214934912e-06,
|
5583 |
+
"loss": 1.2919,
|
5584 |
+
"step": 793
|
5585 |
+
},
|
5586 |
+
{
|
5587 |
+
"epoch": 0.9016323633782825,
|
5588 |
+
"grad_norm": 1.405280590057373,
|
5589 |
+
"learning_rate": 4.8832086843688564e-06,
|
5590 |
+
"loss": 1.3023,
|
5591 |
+
"step": 794
|
5592 |
+
},
|
5593 |
+
{
|
5594 |
+
"epoch": 0.9027679205110007,
|
5595 |
+
"grad_norm": 1.6001452207565308,
|
5596 |
+
"learning_rate": 4.772492648006932e-06,
|
5597 |
+
"loss": 1.3003,
|
5598 |
+
"step": 795
|
5599 |
+
},
|
5600 |
+
{
|
5601 |
+
"epoch": 0.9039034776437189,
|
5602 |
+
"grad_norm": 1.7145951986312866,
|
5603 |
+
"learning_rate": 4.663015480631428e-06,
|
5604 |
+
"loss": 1.2998,
|
5605 |
+
"step": 796
|
5606 |
+
},
|
5607 |
+
{
|
5608 |
+
"epoch": 0.9050390347764372,
|
5609 |
+
"grad_norm": 1.6823632717132568,
|
5610 |
+
"learning_rate": 4.554778606493315e-06,
|
5611 |
+
"loss": 1.2216,
|
5612 |
+
"step": 797
|
5613 |
+
},
|
5614 |
+
{
|
5615 |
+
"epoch": 0.9061745919091554,
|
5616 |
+
"grad_norm": 1.7903900146484375,
|
5617 |
+
"learning_rate": 4.447783433707842e-06,
|
5618 |
+
"loss": 0.927,
|
5619 |
+
"step": 798
|
5620 |
+
},
|
5621 |
+
{
|
5622 |
+
"epoch": 0.9073101490418737,
|
5623 |
+
"grad_norm": 2.1676876544952393,
|
5624 |
+
"learning_rate": 4.342031354236265e-06,
|
5625 |
+
"loss": 1.1205,
|
5626 |
+
"step": 799
|
5627 |
+
},
|
5628 |
+
{
|
5629 |
+
"epoch": 0.9084457061745919,
|
5630 |
+
"grad_norm": 3.2513821125030518,
|
5631 |
+
"learning_rate": 4.237523743867744e-06,
|
5632 |
+
"loss": 1.2247,
|
5633 |
+
"step": 800
|
5634 |
+
},
|
5635 |
+
{
|
5636 |
+
"epoch": 0.9095812633073102,
|
5637 |
+
"grad_norm": 0.5881676077842712,
|
5638 |
+
"learning_rate": 4.134261962201425e-06,
|
5639 |
+
"loss": 1.8548,
|
5640 |
+
"step": 801
|
5641 |
+
},
|
5642 |
+
{
|
5643 |
+
"epoch": 0.9107168204400284,
|
5644 |
+
"grad_norm": 0.6239389777183533,
|
5645 |
+
"learning_rate": 4.032247352628748e-06,
|
5646 |
+
"loss": 1.8171,
|
5647 |
+
"step": 802
|
5648 |
+
},
|
5649 |
+
{
|
5650 |
+
"epoch": 0.9118523775727466,
|
5651 |
+
"grad_norm": 0.667490541934967,
|
5652 |
+
"learning_rate": 3.931481242315993e-06,
|
5653 |
+
"loss": 1.8324,
|
5654 |
+
"step": 803
|
5655 |
+
},
|
5656 |
+
{
|
5657 |
+
"epoch": 0.9129879347054649,
|
5658 |
+
"grad_norm": 0.7072851061820984,
|
5659 |
+
"learning_rate": 3.8319649421869495e-06,
|
5660 |
+
"loss": 1.8454,
|
5661 |
+
"step": 804
|
5662 |
+
},
|
5663 |
+
{
|
5664 |
+
"epoch": 0.9141234918381831,
|
5665 |
+
"grad_norm": 0.695285975933075,
|
5666 |
+
"learning_rate": 3.7336997469060276e-06,
|
5667 |
+
"loss": 1.6505,
|
5668 |
+
"step": 805
|
5669 |
+
},
|
5670 |
+
{
|
5671 |
+
"epoch": 0.9152590489709014,
|
5672 |
+
"grad_norm": 0.6985710263252258,
|
5673 |
+
"learning_rate": 3.6366869348611887e-06,
|
5674 |
+
"loss": 1.7544,
|
5675 |
+
"step": 806
|
5676 |
+
},
|
5677 |
+
{
|
5678 |
+
"epoch": 0.9163946061036196,
|
5679 |
+
"grad_norm": 0.7601833939552307,
|
5680 |
+
"learning_rate": 3.540927768147484e-06,
|
5681 |
+
"loss": 1.794,
|
5682 |
+
"step": 807
|
5683 |
+
},
|
5684 |
+
{
|
5685 |
+
"epoch": 0.9175301632363378,
|
5686 |
+
"grad_norm": 0.7707709670066833,
|
5687 |
+
"learning_rate": 3.4464234925505213e-06,
|
5688 |
+
"loss": 1.6881,
|
5689 |
+
"step": 808
|
5690 |
+
},
|
5691 |
+
{
|
5692 |
+
"epoch": 0.918665720369056,
|
5693 |
+
"grad_norm": 0.7650094032287598,
|
5694 |
+
"learning_rate": 3.3531753375303897e-06,
|
5695 |
+
"loss": 1.6556,
|
5696 |
+
"step": 809
|
5697 |
+
},
|
5698 |
+
{
|
5699 |
+
"epoch": 0.9198012775017743,
|
5700 |
+
"grad_norm": 0.7413115501403809,
|
5701 |
+
"learning_rate": 3.261184516205551e-06,
|
5702 |
+
"loss": 1.6912,
|
5703 |
+
"step": 810
|
5704 |
+
},
|
5705 |
+
{
|
5706 |
+
"epoch": 0.9209368346344925,
|
5707 |
+
"grad_norm": 0.8219679594039917,
|
5708 |
+
"learning_rate": 3.1704522253370947e-06,
|
5709 |
+
"loss": 1.6756,
|
5710 |
+
"step": 811
|
5711 |
+
},
|
5712 |
+
{
|
5713 |
+
"epoch": 0.9220723917672108,
|
5714 |
+
"grad_norm": 0.7987946271896362,
|
5715 |
+
"learning_rate": 3.080979645313142e-06,
|
5716 |
+
"loss": 1.7316,
|
5717 |
+
"step": 812
|
5718 |
+
},
|
5719 |
+
{
|
5720 |
+
"epoch": 0.9232079488999291,
|
5721 |
+
"grad_norm": 0.7636008858680725,
|
5722 |
+
"learning_rate": 2.9927679401335785e-06,
|
5723 |
+
"loss": 1.6475,
|
5724 |
+
"step": 813
|
5725 |
+
},
|
5726 |
+
{
|
5727 |
+
"epoch": 0.9243435060326473,
|
5728 |
+
"grad_norm": 0.8155189752578735,
|
5729 |
+
"learning_rate": 2.905818257394799e-06,
|
5730 |
+
"loss": 1.7,
|
5731 |
+
"step": 814
|
5732 |
+
},
|
5733 |
+
{
|
5734 |
+
"epoch": 0.9254790631653655,
|
5735 |
+
"grad_norm": 0.7985911965370178,
|
5736 |
+
"learning_rate": 2.8201317282748552e-06,
|
5737 |
+
"loss": 1.5697,
|
5738 |
+
"step": 815
|
5739 |
+
},
|
5740 |
+
{
|
5741 |
+
"epoch": 0.9266146202980837,
|
5742 |
+
"grad_norm": 0.8054267764091492,
|
5743 |
+
"learning_rate": 2.735709467518699e-06,
|
5744 |
+
"loss": 1.5872,
|
5745 |
+
"step": 816
|
5746 |
+
},
|
5747 |
+
{
|
5748 |
+
"epoch": 0.927750177430802,
|
5749 |
+
"grad_norm": 0.8176944851875305,
|
5750 |
+
"learning_rate": 2.6525525734236944e-06,
|
5751 |
+
"loss": 1.5912,
|
5752 |
+
"step": 817
|
5753 |
+
},
|
5754 |
+
{
|
5755 |
+
"epoch": 0.9288857345635202,
|
5756 |
+
"grad_norm": 0.8318896293640137,
|
5757 |
+
"learning_rate": 2.5706621278253406e-06,
|
5758 |
+
"loss": 1.5092,
|
5759 |
+
"step": 818
|
5760 |
+
},
|
5761 |
+
{
|
5762 |
+
"epoch": 0.9300212916962385,
|
5763 |
+
"grad_norm": 0.8194354772567749,
|
5764 |
+
"learning_rate": 2.49003919608316e-06,
|
5765 |
+
"loss": 1.5934,
|
5766 |
+
"step": 819
|
5767 |
+
},
|
5768 |
+
{
|
5769 |
+
"epoch": 0.9311568488289567,
|
5770 |
+
"grad_norm": 0.9440639615058899,
|
5771 |
+
"learning_rate": 2.4106848270669e-06,
|
5772 |
+
"loss": 1.6014,
|
5773 |
+
"step": 820
|
5774 |
+
},
|
5775 |
+
{
|
5776 |
+
"epoch": 0.932292405961675,
|
5777 |
+
"grad_norm": 0.9147624969482422,
|
5778 |
+
"learning_rate": 2.3326000531428195e-06,
|
5779 |
+
"loss": 1.6544,
|
5780 |
+
"step": 821
|
5781 |
+
},
|
5782 |
+
{
|
5783 |
+
"epoch": 0.9334279630943931,
|
5784 |
+
"grad_norm": 0.8831505179405212,
|
5785 |
+
"learning_rate": 2.255785890160311e-06,
|
5786 |
+
"loss": 1.5767,
|
5787 |
+
"step": 822
|
5788 |
+
},
|
5789 |
+
{
|
5790 |
+
"epoch": 0.9345635202271114,
|
5791 |
+
"grad_norm": 0.8842580318450928,
|
5792 |
+
"learning_rate": 2.1802433374386588e-06,
|
5793 |
+
"loss": 1.57,
|
5794 |
+
"step": 823
|
5795 |
+
},
|
5796 |
+
{
|
5797 |
+
"epoch": 0.9356990773598297,
|
5798 |
+
"grad_norm": 0.9388715028762817,
|
5799 |
+
"learning_rate": 2.1059733777540225e-06,
|
5800 |
+
"loss": 1.5907,
|
5801 |
+
"step": 824
|
5802 |
+
},
|
5803 |
+
{
|
5804 |
+
"epoch": 0.9368346344925479,
|
5805 |
+
"grad_norm": 1.0418598651885986,
|
5806 |
+
"learning_rate": 2.032976977326706e-06,
|
5807 |
+
"loss": 1.6648,
|
5808 |
+
"step": 825
|
5809 |
+
},
|
5810 |
+
{
|
5811 |
+
"epoch": 0.9379701916252662,
|
5812 |
+
"grad_norm": 0.9417559504508972,
|
5813 |
+
"learning_rate": 1.9612550858085334e-06,
|
5814 |
+
"loss": 1.5464,
|
5815 |
+
"step": 826
|
5816 |
+
},
|
5817 |
+
{
|
5818 |
+
"epoch": 0.9391057487579844,
|
5819 |
+
"grad_norm": 0.9899365305900574,
|
5820 |
+
"learning_rate": 1.8908086362705357e-06,
|
5821 |
+
"loss": 1.6173,
|
5822 |
+
"step": 827
|
5823 |
+
},
|
5824 |
+
{
|
5825 |
+
"epoch": 0.9402413058907026,
|
5826 |
+
"grad_norm": 0.9272261261940002,
|
5827 |
+
"learning_rate": 1.8216385451907624e-06,
|
5828 |
+
"loss": 1.3998,
|
5829 |
+
"step": 828
|
5830 |
+
},
|
5831 |
+
{
|
5832 |
+
"epoch": 0.9413768630234208,
|
5833 |
+
"grad_norm": 1.0163697004318237,
|
5834 |
+
"learning_rate": 1.7537457124423895e-06,
|
5835 |
+
"loss": 1.4026,
|
5836 |
+
"step": 829
|
5837 |
+
},
|
5838 |
+
{
|
5839 |
+
"epoch": 0.9425124201561391,
|
5840 |
+
"grad_norm": 0.9311161041259766,
|
5841 |
+
"learning_rate": 1.68713102128204e-06,
|
5842 |
+
"loss": 1.3825,
|
5843 |
+
"step": 830
|
5844 |
+
},
|
5845 |
+
{
|
5846 |
+
"epoch": 0.9436479772888573,
|
5847 |
+
"grad_norm": 1.0091522932052612,
|
5848 |
+
"learning_rate": 1.62179533833825e-06,
|
5849 |
+
"loss": 1.4272,
|
5850 |
+
"step": 831
|
5851 |
+
},
|
5852 |
+
{
|
5853 |
+
"epoch": 0.9447835344215756,
|
5854 |
+
"grad_norm": 1.104512095451355,
|
5855 |
+
"learning_rate": 1.5577395136001982e-06,
|
5856 |
+
"loss": 1.6002,
|
5857 |
+
"step": 832
|
5858 |
+
},
|
5859 |
+
{
|
5860 |
+
"epoch": 0.9459190915542939,
|
5861 |
+
"grad_norm": 1.098473310470581,
|
5862 |
+
"learning_rate": 1.4949643804066493e-06,
|
5863 |
+
"loss": 1.4385,
|
5864 |
+
"step": 833
|
5865 |
+
},
|
5866 |
+
{
|
5867 |
+
"epoch": 0.9470546486870121,
|
5868 |
+
"grad_norm": 1.0314809083938599,
|
5869 |
+
"learning_rate": 1.4334707554351511e-06,
|
5870 |
+
"loss": 1.3949,
|
5871 |
+
"step": 834
|
5872 |
+
},
|
5873 |
+
{
|
5874 |
+
"epoch": 0.9481902058197303,
|
5875 |
+
"grad_norm": 1.039817452430725,
|
5876 |
+
"learning_rate": 1.3732594386913655e-06,
|
5877 |
+
"loss": 1.4388,
|
5878 |
+
"step": 835
|
5879 |
+
},
|
5880 |
+
{
|
5881 |
+
"epoch": 0.9493257629524485,
|
5882 |
+
"grad_norm": 1.1446057558059692,
|
5883 |
+
"learning_rate": 1.3143312134986651e-06,
|
5884 |
+
"loss": 1.5356,
|
5885 |
+
"step": 836
|
5886 |
+
},
|
5887 |
+
{
|
5888 |
+
"epoch": 0.9504613200851668,
|
5889 |
+
"grad_norm": 1.1136023998260498,
|
5890 |
+
"learning_rate": 1.2566868464879533e-06,
|
5891 |
+
"loss": 1.4069,
|
5892 |
+
"step": 837
|
5893 |
+
},
|
5894 |
+
{
|
5895 |
+
"epoch": 0.951596877217885,
|
5896 |
+
"grad_norm": 1.1392496824264526,
|
5897 |
+
"learning_rate": 1.200327087587716e-06,
|
5898 |
+
"loss": 1.311,
|
5899 |
+
"step": 838
|
5900 |
+
},
|
5901 |
+
{
|
5902 |
+
"epoch": 0.9527324343506033,
|
5903 |
+
"grad_norm": 1.1572725772857666,
|
5904 |
+
"learning_rate": 1.1452526700141964e-06,
|
5905 |
+
"loss": 1.2519,
|
5906 |
+
"step": 839
|
5907 |
+
},
|
5908 |
+
{
|
5909 |
+
"epoch": 0.9538679914833215,
|
5910 |
+
"grad_norm": 1.2496389150619507,
|
5911 |
+
"learning_rate": 1.091464310261947e-06,
|
5912 |
+
"loss": 1.3081,
|
5913 |
+
"step": 840
|
5914 |
+
},
|
5915 |
+
{
|
5916 |
+
"epoch": 0.9550035486160398,
|
5917 |
+
"grad_norm": 1.2605552673339844,
|
5918 |
+
"learning_rate": 1.0389627080944153e-06,
|
5919 |
+
"loss": 1.3012,
|
5920 |
+
"step": 841
|
5921 |
+
},
|
5922 |
+
{
|
5923 |
+
"epoch": 0.9561391057487579,
|
5924 |
+
"grad_norm": 1.4142736196517944,
|
5925 |
+
"learning_rate": 9.877485465349058e-07,
|
5926 |
+
"loss": 1.4079,
|
5927 |
+
"step": 842
|
5928 |
+
},
|
5929 |
+
{
|
5930 |
+
"epoch": 0.9572746628814762,
|
5931 |
+
"grad_norm": 1.5376014709472656,
|
5932 |
+
"learning_rate": 9.378224918576872e-07,
|
5933 |
+
"loss": 1.5231,
|
5934 |
+
"step": 843
|
5935 |
+
},
|
5936 |
+
{
|
5937 |
+
"epoch": 0.9584102200141945,
|
5938 |
+
"grad_norm": 1.5845942497253418,
|
5939 |
+
"learning_rate": 8.891851935792673e-07,
|
5940 |
+
"loss": 1.2707,
|
5941 |
+
"step": 844
|
5942 |
+
},
|
5943 |
+
{
|
5944 |
+
"epoch": 0.9595457771469127,
|
5945 |
+
"grad_norm": 1.518162488937378,
|
5946 |
+
"learning_rate": 8.418372844500532e-07,
|
5947 |
+
"loss": 1.2705,
|
5948 |
+
"step": 845
|
5949 |
+
},
|
5950 |
+
{
|
5951 |
+
"epoch": 0.960681334279631,
|
5952 |
+
"grad_norm": 1.7950546741485596,
|
5953 |
+
"learning_rate": 7.957793804459824e-07,
|
5954 |
+
"loss": 1.2607,
|
5955 |
+
"step": 846
|
5956 |
+
},
|
5957 |
+
{
|
5958 |
+
"epoch": 0.9618168914123492,
|
5959 |
+
"grad_norm": 2.0692031383514404,
|
5960 |
+
"learning_rate": 7.51012080760638e-07,
|
5961 |
+
"loss": 1.4285,
|
5962 |
+
"step": 847
|
5963 |
+
},
|
5964 |
+
{
|
5965 |
+
"epoch": 0.9629524485450675,
|
5966 |
+
"grad_norm": 2.0200307369232178,
|
5967 |
+
"learning_rate": 7.075359677973569e-07,
|
5968 |
+
"loss": 1.1366,
|
5969 |
+
"step": 848
|
5970 |
+
},
|
5971 |
+
{
|
5972 |
+
"epoch": 0.9640880056777856,
|
5973 |
+
"grad_norm": 2.89170503616333,
|
5974 |
+
"learning_rate": 6.653516071616906e-07,
|
5975 |
+
"loss": 1.2507,
|
5976 |
+
"step": 849
|
5977 |
+
},
|
5978 |
+
{
|
5979 |
+
"epoch": 0.9652235628105039,
|
5980 |
+
"grad_norm": 4.06318998336792,
|
5981 |
+
"learning_rate": 6.24459547654066e-07,
|
5982 |
+
"loss": 1.4431,
|
5983 |
+
"step": 850
|
5984 |
+
},
|
5985 |
+
{
|
5986 |
+
"epoch": 0.9663591199432221,
|
5987 |
+
"grad_norm": 0.6128149628639221,
|
5988 |
+
"learning_rate": 5.84860321262648e-07,
|
5989 |
+
"loss": 1.8312,
|
5990 |
+
"step": 851
|
5991 |
+
},
|
5992 |
+
{
|
5993 |
+
"epoch": 0.9674946770759404,
|
5994 |
+
"grad_norm": 0.7000106573104858,
|
5995 |
+
"learning_rate": 5.46554443156333e-07,
|
5996 |
+
"loss": 1.8172,
|
5997 |
+
"step": 852
|
5998 |
+
},
|
5999 |
+
{
|
6000 |
+
"epoch": 0.9686302342086586,
|
6001 |
+
"grad_norm": 0.7351566553115845,
|
6002 |
+
"learning_rate": 5.095424116781767e-07,
|
6003 |
+
"loss": 1.6679,
|
6004 |
+
"step": 853
|
6005 |
+
},
|
6006 |
+
{
|
6007 |
+
"epoch": 0.9697657913413769,
|
6008 |
+
"grad_norm": 0.7055298089981079,
|
6009 |
+
"learning_rate": 4.738247083387992e-07,
|
6010 |
+
"loss": 1.71,
|
6011 |
+
"step": 854
|
6012 |
+
},
|
6013 |
+
{
|
6014 |
+
"epoch": 0.9709013484740951,
|
6015 |
+
"grad_norm": 0.7372353076934814,
|
6016 |
+
"learning_rate": 4.3940179781019055e-07,
|
6017 |
+
"loss": 1.7349,
|
6018 |
+
"step": 855
|
6019 |
+
},
|
6020 |
+
{
|
6021 |
+
"epoch": 0.9720369056068133,
|
6022 |
+
"grad_norm": 0.7767588496208191,
|
6023 |
+
"learning_rate": 4.06274127919648e-07,
|
6024 |
+
"loss": 1.7337,
|
6025 |
+
"step": 856
|
6026 |
+
},
|
6027 |
+
{
|
6028 |
+
"epoch": 0.9731724627395316,
|
6029 |
+
"grad_norm": 0.765540599822998,
|
6030 |
+
"learning_rate": 3.74442129643926e-07,
|
6031 |
+
"loss": 1.646,
|
6032 |
+
"step": 857
|
6033 |
+
},
|
6034 |
+
{
|
6035 |
+
"epoch": 0.9743080198722498,
|
6036 |
+
"grad_norm": 0.7871925234794617,
|
6037 |
+
"learning_rate": 3.439062171036511e-07,
|
6038 |
+
"loss": 1.5931,
|
6039 |
+
"step": 858
|
6040 |
+
},
|
6041 |
+
{
|
6042 |
+
"epoch": 0.9754435770049681,
|
6043 |
+
"grad_norm": 0.8065288066864014,
|
6044 |
+
"learning_rate": 3.14666787557949e-07,
|
6045 |
+
"loss": 1.5187,
|
6046 |
+
"step": 859
|
6047 |
+
},
|
6048 |
+
{
|
6049 |
+
"epoch": 0.9765791341376863,
|
6050 |
+
"grad_norm": 0.8028612732887268,
|
6051 |
+
"learning_rate": 2.8672422139923715e-07,
|
6052 |
+
"loss": 1.5718,
|
6053 |
+
"step": 860
|
6054 |
+
},
|
6055 |
+
{
|
6056 |
+
"epoch": 0.9777146912704046,
|
6057 |
+
"grad_norm": 0.8129795789718628,
|
6058 |
+
"learning_rate": 2.600788821483069e-07,
|
6059 |
+
"loss": 1.6316,
|
6060 |
+
"step": 861
|
6061 |
+
},
|
6062 |
+
{
|
6063 |
+
"epoch": 0.9788502484031227,
|
6064 |
+
"grad_norm": 0.8282390832901001,
|
6065 |
+
"learning_rate": 2.3473111644957135e-07,
|
6066 |
+
"loss": 1.5771,
|
6067 |
+
"step": 862
|
6068 |
+
},
|
6069 |
+
{
|
6070 |
+
"epoch": 0.979985805535841,
|
6071 |
+
"grad_norm": 0.8801934719085693,
|
6072 |
+
"learning_rate": 2.1068125406659145e-07,
|
6073 |
+
"loss": 1.6199,
|
6074 |
+
"step": 863
|
6075 |
+
},
|
6076 |
+
{
|
6077 |
+
"epoch": 0.9811213626685593,
|
6078 |
+
"grad_norm": 0.9070190191268921,
|
6079 |
+
"learning_rate": 1.8792960787774593e-07,
|
6080 |
+
"loss": 1.6148,
|
6081 |
+
"step": 864
|
6082 |
+
},
|
6083 |
+
{
|
6084 |
+
"epoch": 0.9822569198012775,
|
6085 |
+
"grad_norm": 0.9105101823806763,
|
6086 |
+
"learning_rate": 1.6647647387219023e-07,
|
6087 |
+
"loss": 1.5872,
|
6088 |
+
"step": 865
|
6089 |
+
},
|
6090 |
+
{
|
6091 |
+
"epoch": 0.9833924769339958,
|
6092 |
+
"grad_norm": 0.97794109582901,
|
6093 |
+
"learning_rate": 1.463221311459817e-07,
|
6094 |
+
"loss": 1.5238,
|
6095 |
+
"step": 866
|
6096 |
+
},
|
6097 |
+
{
|
6098 |
+
"epoch": 0.984528034066714,
|
6099 |
+
"grad_norm": 1.0621849298477173,
|
6100 |
+
"learning_rate": 1.2746684189846036e-07,
|
6101 |
+
"loss": 1.6259,
|
6102 |
+
"step": 867
|
6103 |
+
},
|
6104 |
+
{
|
6105 |
+
"epoch": 0.9856635911994323,
|
6106 |
+
"grad_norm": 1.1340172290802002,
|
6107 |
+
"learning_rate": 1.0991085142886271e-07,
|
6108 |
+
"loss": 1.7028,
|
6109 |
+
"step": 868
|
6110 |
+
},
|
6111 |
+
{
|
6112 |
+
"epoch": 0.9867991483321504,
|
6113 |
+
"grad_norm": 1.1114485263824463,
|
6114 |
+
"learning_rate": 9.365438813306871e-08,
|
6115 |
+
"loss": 1.5325,
|
6116 |
+
"step": 869
|
6117 |
+
},
|
6118 |
+
{
|
6119 |
+
"epoch": 0.9879347054648687,
|
6120 |
+
"grad_norm": 1.0111680030822754,
|
6121 |
+
"learning_rate": 7.869766350069308e-08,
|
6122 |
+
"loss": 1.4184,
|
6123 |
+
"step": 870
|
6124 |
+
},
|
6125 |
+
{
|
6126 |
+
"epoch": 0.9890702625975869,
|
6127 |
+
"grad_norm": 1.1053588390350342,
|
6128 |
+
"learning_rate": 6.504087211229859e-08,
|
6129 |
+
"loss": 1.5139,
|
6130 |
+
"step": 871
|
6131 |
+
},
|
6132 |
+
{
|
6133 |
+
"epoch": 0.9902058197303052,
|
6134 |
+
"grad_norm": 1.0900163650512695,
|
6135 |
+
"learning_rate": 5.268419163688698e-08,
|
6136 |
+
"loss": 1.4397,
|
6137 |
+
"step": 872
|
6138 |
+
},
|
6139 |
+
{
|
6140 |
+
"epoch": 0.9913413768630234,
|
6141 |
+
"grad_norm": 1.1527384519577026,
|
6142 |
+
"learning_rate": 4.1627782829567476e-08,
|
6143 |
+
"loss": 1.2827,
|
6144 |
+
"step": 873
|
6145 |
+
},
|
6146 |
+
{
|
6147 |
+
"epoch": 0.9924769339957417,
|
6148 |
+
"grad_norm": 1.1613572835922241,
|
6149 |
+
"learning_rate": 3.187178952945846e-08,
|
6150 |
+
"loss": 1.3983,
|
6151 |
+
"step": 874
|
6152 |
+
},
|
6153 |
+
{
|
6154 |
+
"epoch": 0.99361249112846,
|
6155 |
+
"grad_norm": 1.2781202793121338,
|
6156 |
+
"learning_rate": 2.341633865784454e-08,
|
6157 |
+
"loss": 1.3543,
|
6158 |
+
"step": 875
|
6159 |
+
},
|
6160 |
+
{
|
6161 |
+
"epoch": 0.9947480482611781,
|
6162 |
+
"grad_norm": 1.6025346517562866,
|
6163 |
+
"learning_rate": 1.6261540216522264e-08,
|
6164 |
+
"loss": 1.4821,
|
6165 |
+
"step": 876
|
6166 |
+
},
|
6167 |
+
{
|
6168 |
+
"epoch": 0.9958836053938964,
|
6169 |
+
"grad_norm": 1.4461325407028198,
|
6170 |
+
"learning_rate": 1.0407487286345774e-08,
|
6171 |
+
"loss": 1.2549,
|
6172 |
+
"step": 877
|
6173 |
+
},
|
6174 |
+
{
|
6175 |
+
"epoch": 0.9970191625266146,
|
6176 |
+
"grad_norm": 1.5728317499160767,
|
6177 |
+
"learning_rate": 5.854256026027738e-09,
|
6178 |
+
"loss": 1.1668,
|
6179 |
+
"step": 878
|
6180 |
+
},
|
6181 |
+
{
|
6182 |
+
"epoch": 0.9981547196593329,
|
6183 |
+
"grad_norm": 2.2646923065185547,
|
6184 |
+
"learning_rate": 2.6019056711512614e-09,
|
6185 |
+
"loss": 1.2263,
|
6186 |
+
"step": 879
|
6187 |
+
},
|
6188 |
+
{
|
6189 |
+
"epoch": 0.9992902767920511,
|
6190 |
+
"grad_norm": 2.1974527835845947,
|
6191 |
+
"learning_rate": 6.50478533403831e-10,
|
6192 |
+
"loss": 0.9378,
|
6193 |
+
"step": 880
|
6194 |
+
},
|
6195 |
+
{
|
6196 |
+
"epoch": 1.0009226401703335,
|
6197 |
+
"grad_norm": 5.330093860626221,
|
6198 |
+
"learning_rate": 0.0,
|
6199 |
+
"loss": 2.3759,
|
6200 |
+
"step": 881
|
6201 |
}
|
6202 |
],
|
6203 |
"logging_steps": 1,
|
|
|
6212 |
"should_evaluate": false,
|
6213 |
"should_log": false,
|
6214 |
"should_save": true,
|
6215 |
+
"should_training_stop": true
|
6216 |
},
|
6217 |
"attributes": {}
|
6218 |
}
|
6219 |
},
|
6220 |
+
"total_flos": 1.3469372301862502e+18,
|
6221 |
"train_batch_size": 2,
|
6222 |
"trial_name": null,
|
6223 |
"trial_params": null
|