Training in progress, step 1149, checkpoint
Browse files
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 101184122
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:153eed1481bc61fe36b1f1cf1654f85452147fe230553dc8731acf81acb3a244
|
| 3 |
size 101184122
|
last-checkpoint/rng_state_0.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14512
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:37c3a76300b214839815f53708f1b6bb815b6feaa2f5036f6a9d3156acf37e12
|
| 3 |
size 14512
|
last-checkpoint/rng_state_1.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14512
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d77c99f309bad08baed1afae6c22e6ad4dbb4c1a734ee71c48d6072810519747
|
| 3 |
size 14512
|
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:b1f750763acba28ae9170e730eb75e38578658a5d1599aebf4d0966d360ccf4a
|
| 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": 0.
|
| 5 |
"eval_steps": 383,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -5393,6 +5393,2695 @@
|
|
| 5393 |
"eval_samples_per_second": 15.995,
|
| 5394 |
"eval_steps_per_second": 4.002,
|
| 5395 |
"step": 766
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5396 |
}
|
| 5397 |
],
|
| 5398 |
"logging_steps": 1,
|
|
@@ -5412,7 +8101,7 @@
|
|
| 5412 |
"attributes": {}
|
| 5413 |
}
|
| 5414 |
},
|
| 5415 |
-
"total_flos":
|
| 5416 |
"train_batch_size": 2,
|
| 5417 |
"trial_name": null,
|
| 5418 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.7506124448799608,
|
| 5 |
"eval_steps": 383,
|
| 6 |
+
"global_step": 1149,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 5393 |
"eval_samples_per_second": 15.995,
|
| 5394 |
"eval_steps_per_second": 4.002,
|
| 5395 |
"step": 766
|
| 5396 |
+
},
|
| 5397 |
+
{
|
| 5398 |
+
"epoch": 0.5010615711252654,
|
| 5399 |
+
"grad_norm": Infinity,
|
| 5400 |
+
"learning_rate": 0.00010488026549961863,
|
| 5401 |
+
"loss": 2.3278,
|
| 5402 |
+
"step": 767
|
| 5403 |
+
},
|
| 5404 |
+
{
|
| 5405 |
+
"epoch": 0.5017148456638902,
|
| 5406 |
+
"grad_norm": Infinity,
|
| 5407 |
+
"learning_rate": 0.00010466823782214927,
|
| 5408 |
+
"loss": 2.3015,
|
| 5409 |
+
"step": 768
|
| 5410 |
+
},
|
| 5411 |
+
{
|
| 5412 |
+
"epoch": 0.5023681202025151,
|
| 5413 |
+
"grad_norm": Infinity,
|
| 5414 |
+
"learning_rate": 0.00010445618911032853,
|
| 5415 |
+
"loss": 2.6376,
|
| 5416 |
+
"step": 769
|
| 5417 |
+
},
|
| 5418 |
+
{
|
| 5419 |
+
"epoch": 0.5030213947411399,
|
| 5420 |
+
"grad_norm": Infinity,
|
| 5421 |
+
"learning_rate": 0.00010424412031961484,
|
| 5422 |
+
"loss": 2.5291,
|
| 5423 |
+
"step": 770
|
| 5424 |
+
},
|
| 5425 |
+
{
|
| 5426 |
+
"epoch": 0.5036746692797648,
|
| 5427 |
+
"grad_norm": Infinity,
|
| 5428 |
+
"learning_rate": 0.00010403203240555707,
|
| 5429 |
+
"loss": 2.7269,
|
| 5430 |
+
"step": 771
|
| 5431 |
+
},
|
| 5432 |
+
{
|
| 5433 |
+
"epoch": 0.5043279438183896,
|
| 5434 |
+
"grad_norm": Infinity,
|
| 5435 |
+
"learning_rate": 0.0001038199263237903,
|
| 5436 |
+
"loss": 2.5242,
|
| 5437 |
+
"step": 772
|
| 5438 |
+
},
|
| 5439 |
+
{
|
| 5440 |
+
"epoch": 0.5049812183570145,
|
| 5441 |
+
"grad_norm": Infinity,
|
| 5442 |
+
"learning_rate": 0.00010360780303003139,
|
| 5443 |
+
"loss": 3.0819,
|
| 5444 |
+
"step": 773
|
| 5445 |
+
},
|
| 5446 |
+
{
|
| 5447 |
+
"epoch": 0.5056344928956394,
|
| 5448 |
+
"grad_norm": Infinity,
|
| 5449 |
+
"learning_rate": 0.00010339566348007487,
|
| 5450 |
+
"loss": 3.4961,
|
| 5451 |
+
"step": 774
|
| 5452 |
+
},
|
| 5453 |
+
{
|
| 5454 |
+
"epoch": 0.5062877674342643,
|
| 5455 |
+
"grad_norm": Infinity,
|
| 5456 |
+
"learning_rate": 0.00010318350862978848,
|
| 5457 |
+
"loss": 4.0663,
|
| 5458 |
+
"step": 775
|
| 5459 |
+
},
|
| 5460 |
+
{
|
| 5461 |
+
"epoch": 0.5069410419728891,
|
| 5462 |
+
"grad_norm": Infinity,
|
| 5463 |
+
"learning_rate": 0.00010297133943510879,
|
| 5464 |
+
"loss": 1.2678,
|
| 5465 |
+
"step": 776
|
| 5466 |
+
},
|
| 5467 |
+
{
|
| 5468 |
+
"epoch": 0.507594316511514,
|
| 5469 |
+
"grad_norm": Infinity,
|
| 5470 |
+
"learning_rate": 0.00010275915685203712,
|
| 5471 |
+
"loss": 1.3923,
|
| 5472 |
+
"step": 777
|
| 5473 |
+
},
|
| 5474 |
+
{
|
| 5475 |
+
"epoch": 0.5082475910501388,
|
| 5476 |
+
"grad_norm": Infinity,
|
| 5477 |
+
"learning_rate": 0.00010254696183663511,
|
| 5478 |
+
"loss": 1.403,
|
| 5479 |
+
"step": 778
|
| 5480 |
+
},
|
| 5481 |
+
{
|
| 5482 |
+
"epoch": 0.5089008655887637,
|
| 5483 |
+
"grad_norm": Infinity,
|
| 5484 |
+
"learning_rate": 0.00010233475534502042,
|
| 5485 |
+
"loss": 1.4813,
|
| 5486 |
+
"step": 779
|
| 5487 |
+
},
|
| 5488 |
+
{
|
| 5489 |
+
"epoch": 0.5095541401273885,
|
| 5490 |
+
"grad_norm": Infinity,
|
| 5491 |
+
"learning_rate": 0.00010212253833336237,
|
| 5492 |
+
"loss": 1.454,
|
| 5493 |
+
"step": 780
|
| 5494 |
+
},
|
| 5495 |
+
{
|
| 5496 |
+
"epoch": 0.5102074146660134,
|
| 5497 |
+
"grad_norm": Infinity,
|
| 5498 |
+
"learning_rate": 0.00010191031175787768,
|
| 5499 |
+
"loss": 1.4327,
|
| 5500 |
+
"step": 781
|
| 5501 |
+
},
|
| 5502 |
+
{
|
| 5503 |
+
"epoch": 0.5108606892046382,
|
| 5504 |
+
"grad_norm": Infinity,
|
| 5505 |
+
"learning_rate": 0.00010169807657482623,
|
| 5506 |
+
"loss": 1.4784,
|
| 5507 |
+
"step": 782
|
| 5508 |
+
},
|
| 5509 |
+
{
|
| 5510 |
+
"epoch": 0.5115139637432631,
|
| 5511 |
+
"grad_norm": Infinity,
|
| 5512 |
+
"learning_rate": 0.00010148583374050667,
|
| 5513 |
+
"loss": 1.6763,
|
| 5514 |
+
"step": 783
|
| 5515 |
+
},
|
| 5516 |
+
{
|
| 5517 |
+
"epoch": 0.5121672382818879,
|
| 5518 |
+
"grad_norm": Infinity,
|
| 5519 |
+
"learning_rate": 0.00010127358421125204,
|
| 5520 |
+
"loss": 1.6503,
|
| 5521 |
+
"step": 784
|
| 5522 |
+
},
|
| 5523 |
+
{
|
| 5524 |
+
"epoch": 0.5128205128205128,
|
| 5525 |
+
"grad_norm": Infinity,
|
| 5526 |
+
"learning_rate": 0.00010106132894342564,
|
| 5527 |
+
"loss": 1.8225,
|
| 5528 |
+
"step": 785
|
| 5529 |
+
},
|
| 5530 |
+
{
|
| 5531 |
+
"epoch": 0.5134737873591376,
|
| 5532 |
+
"grad_norm": Infinity,
|
| 5533 |
+
"learning_rate": 0.00010084906889341656,
|
| 5534 |
+
"loss": 1.5835,
|
| 5535 |
+
"step": 786
|
| 5536 |
+
},
|
| 5537 |
+
{
|
| 5538 |
+
"epoch": 0.5141270618977626,
|
| 5539 |
+
"grad_norm": Infinity,
|
| 5540 |
+
"learning_rate": 0.00010063680501763552,
|
| 5541 |
+
"loss": 1.8056,
|
| 5542 |
+
"step": 787
|
| 5543 |
+
},
|
| 5544 |
+
{
|
| 5545 |
+
"epoch": 0.5147803364363874,
|
| 5546 |
+
"grad_norm": Infinity,
|
| 5547 |
+
"learning_rate": 0.00010042453827251044,
|
| 5548 |
+
"loss": 1.9522,
|
| 5549 |
+
"step": 788
|
| 5550 |
+
},
|
| 5551 |
+
{
|
| 5552 |
+
"epoch": 0.5154336109750123,
|
| 5553 |
+
"grad_norm": Infinity,
|
| 5554 |
+
"learning_rate": 0.00010021226961448209,
|
| 5555 |
+
"loss": 2.0843,
|
| 5556 |
+
"step": 789
|
| 5557 |
+
},
|
| 5558 |
+
{
|
| 5559 |
+
"epoch": 0.5160868855136371,
|
| 5560 |
+
"grad_norm": Infinity,
|
| 5561 |
+
"learning_rate": 0.0001,
|
| 5562 |
+
"loss": 2.0254,
|
| 5563 |
+
"step": 790
|
| 5564 |
+
},
|
| 5565 |
+
{
|
| 5566 |
+
"epoch": 0.516740160052262,
|
| 5567 |
+
"grad_norm": Infinity,
|
| 5568 |
+
"learning_rate": 9.97877303855179e-05,
|
| 5569 |
+
"loss": 2.1822,
|
| 5570 |
+
"step": 791
|
| 5571 |
+
},
|
| 5572 |
+
{
|
| 5573 |
+
"epoch": 0.5173934345908868,
|
| 5574 |
+
"grad_norm": Infinity,
|
| 5575 |
+
"learning_rate": 9.957546172748958e-05,
|
| 5576 |
+
"loss": 2.2943,
|
| 5577 |
+
"step": 792
|
| 5578 |
+
},
|
| 5579 |
+
{
|
| 5580 |
+
"epoch": 0.5180467091295117,
|
| 5581 |
+
"grad_norm": Infinity,
|
| 5582 |
+
"learning_rate": 9.936319498236446e-05,
|
| 5583 |
+
"loss": 2.3479,
|
| 5584 |
+
"step": 793
|
| 5585 |
+
},
|
| 5586 |
+
{
|
| 5587 |
+
"epoch": 0.5186999836681365,
|
| 5588 |
+
"grad_norm": Infinity,
|
| 5589 |
+
"learning_rate": 9.915093110658346e-05,
|
| 5590 |
+
"loss": 2.7068,
|
| 5591 |
+
"step": 794
|
| 5592 |
+
},
|
| 5593 |
+
{
|
| 5594 |
+
"epoch": 0.5193532582067614,
|
| 5595 |
+
"grad_norm": Infinity,
|
| 5596 |
+
"learning_rate": 9.89386710565744e-05,
|
| 5597 |
+
"loss": 2.4189,
|
| 5598 |
+
"step": 795
|
| 5599 |
+
},
|
| 5600 |
+
{
|
| 5601 |
+
"epoch": 0.5200065327453862,
|
| 5602 |
+
"grad_norm": Infinity,
|
| 5603 |
+
"learning_rate": 9.8726415788748e-05,
|
| 5604 |
+
"loss": 3.0649,
|
| 5605 |
+
"step": 796
|
| 5606 |
+
},
|
| 5607 |
+
{
|
| 5608 |
+
"epoch": 0.5206598072840111,
|
| 5609 |
+
"grad_norm": Infinity,
|
| 5610 |
+
"learning_rate": 9.851416625949334e-05,
|
| 5611 |
+
"loss": 3.1619,
|
| 5612 |
+
"step": 797
|
| 5613 |
+
},
|
| 5614 |
+
{
|
| 5615 |
+
"epoch": 0.5213130818226359,
|
| 5616 |
+
"grad_norm": Infinity,
|
| 5617 |
+
"learning_rate": 9.830192342517379e-05,
|
| 5618 |
+
"loss": 3.4088,
|
| 5619 |
+
"step": 798
|
| 5620 |
+
},
|
| 5621 |
+
{
|
| 5622 |
+
"epoch": 0.5219663563612609,
|
| 5623 |
+
"grad_norm": Infinity,
|
| 5624 |
+
"learning_rate": 9.808968824212234e-05,
|
| 5625 |
+
"loss": 3.5337,
|
| 5626 |
+
"step": 799
|
| 5627 |
+
},
|
| 5628 |
+
{
|
| 5629 |
+
"epoch": 0.5226196308998857,
|
| 5630 |
+
"grad_norm": Infinity,
|
| 5631 |
+
"learning_rate": 9.787746166663764e-05,
|
| 5632 |
+
"loss": 4.3343,
|
| 5633 |
+
"step": 800
|
| 5634 |
+
},
|
| 5635 |
+
{
|
| 5636 |
+
"epoch": 0.5232729054385106,
|
| 5637 |
+
"grad_norm": Infinity,
|
| 5638 |
+
"learning_rate": 9.76652446549796e-05,
|
| 5639 |
+
"loss": 1.4279,
|
| 5640 |
+
"step": 801
|
| 5641 |
+
},
|
| 5642 |
+
{
|
| 5643 |
+
"epoch": 0.5239261799771354,
|
| 5644 |
+
"grad_norm": Infinity,
|
| 5645 |
+
"learning_rate": 9.745303816336489e-05,
|
| 5646 |
+
"loss": 1.3824,
|
| 5647 |
+
"step": 802
|
| 5648 |
+
},
|
| 5649 |
+
{
|
| 5650 |
+
"epoch": 0.5245794545157603,
|
| 5651 |
+
"grad_norm": Infinity,
|
| 5652 |
+
"learning_rate": 9.724084314796292e-05,
|
| 5653 |
+
"loss": 1.3597,
|
| 5654 |
+
"step": 803
|
| 5655 |
+
},
|
| 5656 |
+
{
|
| 5657 |
+
"epoch": 0.5252327290543851,
|
| 5658 |
+
"grad_norm": Infinity,
|
| 5659 |
+
"learning_rate": 9.702866056489125e-05,
|
| 5660 |
+
"loss": 1.3498,
|
| 5661 |
+
"step": 804
|
| 5662 |
+
},
|
| 5663 |
+
{
|
| 5664 |
+
"epoch": 0.52588600359301,
|
| 5665 |
+
"grad_norm": Infinity,
|
| 5666 |
+
"learning_rate": 9.681649137021158e-05,
|
| 5667 |
+
"loss": 1.3747,
|
| 5668 |
+
"step": 805
|
| 5669 |
+
},
|
| 5670 |
+
{
|
| 5671 |
+
"epoch": 0.5265392781316348,
|
| 5672 |
+
"grad_norm": Infinity,
|
| 5673 |
+
"learning_rate": 9.660433651992514e-05,
|
| 5674 |
+
"loss": 1.3791,
|
| 5675 |
+
"step": 806
|
| 5676 |
+
},
|
| 5677 |
+
{
|
| 5678 |
+
"epoch": 0.5271925526702597,
|
| 5679 |
+
"grad_norm": Infinity,
|
| 5680 |
+
"learning_rate": 9.639219696996861e-05,
|
| 5681 |
+
"loss": 1.5428,
|
| 5682 |
+
"step": 807
|
| 5683 |
+
},
|
| 5684 |
+
{
|
| 5685 |
+
"epoch": 0.5278458272088845,
|
| 5686 |
+
"grad_norm": Infinity,
|
| 5687 |
+
"learning_rate": 9.618007367620972e-05,
|
| 5688 |
+
"loss": 1.586,
|
| 5689 |
+
"step": 808
|
| 5690 |
+
},
|
| 5691 |
+
{
|
| 5692 |
+
"epoch": 0.5284991017475094,
|
| 5693 |
+
"grad_norm": Infinity,
|
| 5694 |
+
"learning_rate": 9.596796759444293e-05,
|
| 5695 |
+
"loss": 1.5683,
|
| 5696 |
+
"step": 809
|
| 5697 |
+
},
|
| 5698 |
+
{
|
| 5699 |
+
"epoch": 0.5291523762861342,
|
| 5700 |
+
"grad_norm": Infinity,
|
| 5701 |
+
"learning_rate": 9.57558796803852e-05,
|
| 5702 |
+
"loss": 1.6566,
|
| 5703 |
+
"step": 810
|
| 5704 |
+
},
|
| 5705 |
+
{
|
| 5706 |
+
"epoch": 0.5298056508247591,
|
| 5707 |
+
"grad_norm": Infinity,
|
| 5708 |
+
"learning_rate": 9.554381088967148e-05,
|
| 5709 |
+
"loss": 1.7084,
|
| 5710 |
+
"step": 811
|
| 5711 |
+
},
|
| 5712 |
+
{
|
| 5713 |
+
"epoch": 0.530458925363384,
|
| 5714 |
+
"grad_norm": Infinity,
|
| 5715 |
+
"learning_rate": 9.533176217785075e-05,
|
| 5716 |
+
"loss": 1.9305,
|
| 5717 |
+
"step": 812
|
| 5718 |
+
},
|
| 5719 |
+
{
|
| 5720 |
+
"epoch": 0.5311121999020089,
|
| 5721 |
+
"grad_norm": Infinity,
|
| 5722 |
+
"learning_rate": 9.51197345003814e-05,
|
| 5723 |
+
"loss": 1.8239,
|
| 5724 |
+
"step": 813
|
| 5725 |
+
},
|
| 5726 |
+
{
|
| 5727 |
+
"epoch": 0.5317654744406337,
|
| 5728 |
+
"grad_norm": Infinity,
|
| 5729 |
+
"learning_rate": 9.490772881262709e-05,
|
| 5730 |
+
"loss": 1.7874,
|
| 5731 |
+
"step": 814
|
| 5732 |
+
},
|
| 5733 |
+
{
|
| 5734 |
+
"epoch": 0.5324187489792586,
|
| 5735 |
+
"grad_norm": Infinity,
|
| 5736 |
+
"learning_rate": 9.46957460698523e-05,
|
| 5737 |
+
"loss": 1.8129,
|
| 5738 |
+
"step": 815
|
| 5739 |
+
},
|
| 5740 |
+
{
|
| 5741 |
+
"epoch": 0.5330720235178834,
|
| 5742 |
+
"grad_norm": Infinity,
|
| 5743 |
+
"learning_rate": 9.448378722721825e-05,
|
| 5744 |
+
"loss": 2.0876,
|
| 5745 |
+
"step": 816
|
| 5746 |
+
},
|
| 5747 |
+
{
|
| 5748 |
+
"epoch": 0.5337252980565083,
|
| 5749 |
+
"grad_norm": Infinity,
|
| 5750 |
+
"learning_rate": 9.427185323977845e-05,
|
| 5751 |
+
"loss": 2.2503,
|
| 5752 |
+
"step": 817
|
| 5753 |
+
},
|
| 5754 |
+
{
|
| 5755 |
+
"epoch": 0.5343785725951331,
|
| 5756 |
+
"grad_norm": Infinity,
|
| 5757 |
+
"learning_rate": 9.405994506247432e-05,
|
| 5758 |
+
"loss": 2.2582,
|
| 5759 |
+
"step": 818
|
| 5760 |
+
},
|
| 5761 |
+
{
|
| 5762 |
+
"epoch": 0.535031847133758,
|
| 5763 |
+
"grad_norm": Infinity,
|
| 5764 |
+
"learning_rate": 9.384806365013113e-05,
|
| 5765 |
+
"loss": 2.4585,
|
| 5766 |
+
"step": 819
|
| 5767 |
+
},
|
| 5768 |
+
{
|
| 5769 |
+
"epoch": 0.5356851216723828,
|
| 5770 |
+
"grad_norm": Infinity,
|
| 5771 |
+
"learning_rate": 9.363620995745337e-05,
|
| 5772 |
+
"loss": 2.5637,
|
| 5773 |
+
"step": 820
|
| 5774 |
+
},
|
| 5775 |
+
{
|
| 5776 |
+
"epoch": 0.5363383962110077,
|
| 5777 |
+
"grad_norm": Infinity,
|
| 5778 |
+
"learning_rate": 9.342438493902085e-05,
|
| 5779 |
+
"loss": 2.8806,
|
| 5780 |
+
"step": 821
|
| 5781 |
+
},
|
| 5782 |
+
{
|
| 5783 |
+
"epoch": 0.5369916707496325,
|
| 5784 |
+
"grad_norm": Infinity,
|
| 5785 |
+
"learning_rate": 9.321258954928393e-05,
|
| 5786 |
+
"loss": 2.8588,
|
| 5787 |
+
"step": 822
|
| 5788 |
+
},
|
| 5789 |
+
{
|
| 5790 |
+
"epoch": 0.5376449452882573,
|
| 5791 |
+
"grad_norm": Infinity,
|
| 5792 |
+
"learning_rate": 9.300082474255967e-05,
|
| 5793 |
+
"loss": 3.0883,
|
| 5794 |
+
"step": 823
|
| 5795 |
+
},
|
| 5796 |
+
{
|
| 5797 |
+
"epoch": 0.5382982198268822,
|
| 5798 |
+
"grad_norm": Infinity,
|
| 5799 |
+
"learning_rate": 9.278909147302727e-05,
|
| 5800 |
+
"loss": 3.3387,
|
| 5801 |
+
"step": 824
|
| 5802 |
+
},
|
| 5803 |
+
{
|
| 5804 |
+
"epoch": 0.5389514943655072,
|
| 5805 |
+
"grad_norm": Infinity,
|
| 5806 |
+
"learning_rate": 9.257739069472374e-05,
|
| 5807 |
+
"loss": 4.0844,
|
| 5808 |
+
"step": 825
|
| 5809 |
+
},
|
| 5810 |
+
{
|
| 5811 |
+
"epoch": 0.539604768904132,
|
| 5812 |
+
"grad_norm": Infinity,
|
| 5813 |
+
"learning_rate": 9.236572336153986e-05,
|
| 5814 |
+
"loss": 1.3864,
|
| 5815 |
+
"step": 826
|
| 5816 |
+
},
|
| 5817 |
+
{
|
| 5818 |
+
"epoch": 0.5402580434427569,
|
| 5819 |
+
"grad_norm": Infinity,
|
| 5820 |
+
"learning_rate": 9.215409042721552e-05,
|
| 5821 |
+
"loss": 1.3945,
|
| 5822 |
+
"step": 827
|
| 5823 |
+
},
|
| 5824 |
+
{
|
| 5825 |
+
"epoch": 0.5409113179813817,
|
| 5826 |
+
"grad_norm": Infinity,
|
| 5827 |
+
"learning_rate": 9.194249284533576e-05,
|
| 5828 |
+
"loss": 1.4255,
|
| 5829 |
+
"step": 828
|
| 5830 |
+
},
|
| 5831 |
+
{
|
| 5832 |
+
"epoch": 0.5415645925200065,
|
| 5833 |
+
"grad_norm": Infinity,
|
| 5834 |
+
"learning_rate": 9.173093156932623e-05,
|
| 5835 |
+
"loss": 1.3899,
|
| 5836 |
+
"step": 829
|
| 5837 |
+
},
|
| 5838 |
+
{
|
| 5839 |
+
"epoch": 0.5422178670586314,
|
| 5840 |
+
"grad_norm": Infinity,
|
| 5841 |
+
"learning_rate": 9.151940755244912e-05,
|
| 5842 |
+
"loss": 1.4801,
|
| 5843 |
+
"step": 830
|
| 5844 |
+
},
|
| 5845 |
+
{
|
| 5846 |
+
"epoch": 0.5428711415972562,
|
| 5847 |
+
"grad_norm": Infinity,
|
| 5848 |
+
"learning_rate": 9.130792174779854e-05,
|
| 5849 |
+
"loss": 1.5015,
|
| 5850 |
+
"step": 831
|
| 5851 |
+
},
|
| 5852 |
+
{
|
| 5853 |
+
"epoch": 0.5435244161358811,
|
| 5854 |
+
"grad_norm": Infinity,
|
| 5855 |
+
"learning_rate": 9.109647510829657e-05,
|
| 5856 |
+
"loss": 1.5265,
|
| 5857 |
+
"step": 832
|
| 5858 |
+
},
|
| 5859 |
+
{
|
| 5860 |
+
"epoch": 0.5441776906745059,
|
| 5861 |
+
"grad_norm": Infinity,
|
| 5862 |
+
"learning_rate": 9.088506858668888e-05,
|
| 5863 |
+
"loss": 1.5976,
|
| 5864 |
+
"step": 833
|
| 5865 |
+
},
|
| 5866 |
+
{
|
| 5867 |
+
"epoch": 0.5448309652131308,
|
| 5868 |
+
"grad_norm": Infinity,
|
| 5869 |
+
"learning_rate": 9.067370313554015e-05,
|
| 5870 |
+
"loss": 1.5211,
|
| 5871 |
+
"step": 834
|
| 5872 |
+
},
|
| 5873 |
+
{
|
| 5874 |
+
"epoch": 0.5454842397517556,
|
| 5875 |
+
"grad_norm": Infinity,
|
| 5876 |
+
"learning_rate": 9.04623797072302e-05,
|
| 5877 |
+
"loss": 1.7078,
|
| 5878 |
+
"step": 835
|
| 5879 |
+
},
|
| 5880 |
+
{
|
| 5881 |
+
"epoch": 0.5461375142903805,
|
| 5882 |
+
"grad_norm": Infinity,
|
| 5883 |
+
"learning_rate": 9.025109925394939e-05,
|
| 5884 |
+
"loss": 1.6801,
|
| 5885 |
+
"step": 836
|
| 5886 |
+
},
|
| 5887 |
+
{
|
| 5888 |
+
"epoch": 0.5467907888290053,
|
| 5889 |
+
"grad_norm": Infinity,
|
| 5890 |
+
"learning_rate": 9.003986272769455e-05,
|
| 5891 |
+
"loss": 1.755,
|
| 5892 |
+
"step": 837
|
| 5893 |
+
},
|
| 5894 |
+
{
|
| 5895 |
+
"epoch": 0.5474440633676303,
|
| 5896 |
+
"grad_norm": Infinity,
|
| 5897 |
+
"learning_rate": 8.982867108026442e-05,
|
| 5898 |
+
"loss": 1.796,
|
| 5899 |
+
"step": 838
|
| 5900 |
+
},
|
| 5901 |
+
{
|
| 5902 |
+
"epoch": 0.5480973379062551,
|
| 5903 |
+
"grad_norm": Infinity,
|
| 5904 |
+
"learning_rate": 8.961752526325565e-05,
|
| 5905 |
+
"loss": 1.9277,
|
| 5906 |
+
"step": 839
|
| 5907 |
+
},
|
| 5908 |
+
{
|
| 5909 |
+
"epoch": 0.54875061244488,
|
| 5910 |
+
"grad_norm": Infinity,
|
| 5911 |
+
"learning_rate": 8.94064262280584e-05,
|
| 5912 |
+
"loss": 2.0079,
|
| 5913 |
+
"step": 840
|
| 5914 |
+
},
|
| 5915 |
+
{
|
| 5916 |
+
"epoch": 0.5494038869835048,
|
| 5917 |
+
"grad_norm": Infinity,
|
| 5918 |
+
"learning_rate": 8.919537492585194e-05,
|
| 5919 |
+
"loss": 2.1747,
|
| 5920 |
+
"step": 841
|
| 5921 |
+
},
|
| 5922 |
+
{
|
| 5923 |
+
"epoch": 0.5500571615221297,
|
| 5924 |
+
"grad_norm": Infinity,
|
| 5925 |
+
"learning_rate": 8.898437230760058e-05,
|
| 5926 |
+
"loss": 2.2313,
|
| 5927 |
+
"step": 842
|
| 5928 |
+
},
|
| 5929 |
+
{
|
| 5930 |
+
"epoch": 0.5507104360607545,
|
| 5931 |
+
"grad_norm": Infinity,
|
| 5932 |
+
"learning_rate": 8.877341932404909e-05,
|
| 5933 |
+
"loss": 2.6037,
|
| 5934 |
+
"step": 843
|
| 5935 |
+
},
|
| 5936 |
+
{
|
| 5937 |
+
"epoch": 0.5513637105993794,
|
| 5938 |
+
"grad_norm": Infinity,
|
| 5939 |
+
"learning_rate": 8.856251692571879e-05,
|
| 5940 |
+
"loss": 2.5185,
|
| 5941 |
+
"step": 844
|
| 5942 |
+
},
|
| 5943 |
+
{
|
| 5944 |
+
"epoch": 0.5520169851380042,
|
| 5945 |
+
"grad_norm": Infinity,
|
| 5946 |
+
"learning_rate": 8.835166606290295e-05,
|
| 5947 |
+
"loss": 2.5605,
|
| 5948 |
+
"step": 845
|
| 5949 |
+
},
|
| 5950 |
+
{
|
| 5951 |
+
"epoch": 0.5526702596766291,
|
| 5952 |
+
"grad_norm": Infinity,
|
| 5953 |
+
"learning_rate": 8.814086768566272e-05,
|
| 5954 |
+
"loss": 3.0754,
|
| 5955 |
+
"step": 846
|
| 5956 |
+
},
|
| 5957 |
+
{
|
| 5958 |
+
"epoch": 0.5533235342152539,
|
| 5959 |
+
"grad_norm": Infinity,
|
| 5960 |
+
"learning_rate": 8.793012274382261e-05,
|
| 5961 |
+
"loss": 3.1223,
|
| 5962 |
+
"step": 847
|
| 5963 |
+
},
|
| 5964 |
+
{
|
| 5965 |
+
"epoch": 0.5539768087538788,
|
| 5966 |
+
"grad_norm": Infinity,
|
| 5967 |
+
"learning_rate": 8.771943218696649e-05,
|
| 5968 |
+
"loss": 3.3635,
|
| 5969 |
+
"step": 848
|
| 5970 |
+
},
|
| 5971 |
+
{
|
| 5972 |
+
"epoch": 0.5546300832925036,
|
| 5973 |
+
"grad_norm": Infinity,
|
| 5974 |
+
"learning_rate": 8.750879696443321e-05,
|
| 5975 |
+
"loss": 3.5695,
|
| 5976 |
+
"step": 849
|
| 5977 |
+
},
|
| 5978 |
+
{
|
| 5979 |
+
"epoch": 0.5552833578311285,
|
| 5980 |
+
"grad_norm": Infinity,
|
| 5981 |
+
"learning_rate": 8.729821802531212e-05,
|
| 5982 |
+
"loss": 4.3607,
|
| 5983 |
+
"step": 850
|
| 5984 |
+
},
|
| 5985 |
+
{
|
| 5986 |
+
"epoch": 0.5559366323697534,
|
| 5987 |
+
"grad_norm": Infinity,
|
| 5988 |
+
"learning_rate": 8.708769631843916e-05,
|
| 5989 |
+
"loss": 1.1649,
|
| 5990 |
+
"step": 851
|
| 5991 |
+
},
|
| 5992 |
+
{
|
| 5993 |
+
"epoch": 0.5565899069083783,
|
| 5994 |
+
"grad_norm": Infinity,
|
| 5995 |
+
"learning_rate": 8.687723279239222e-05,
|
| 5996 |
+
"loss": 1.412,
|
| 5997 |
+
"step": 852
|
| 5998 |
+
},
|
| 5999 |
+
{
|
| 6000 |
+
"epoch": 0.5572431814470031,
|
| 6001 |
+
"grad_norm": Infinity,
|
| 6002 |
+
"learning_rate": 8.666682839548719e-05,
|
| 6003 |
+
"loss": 1.341,
|
| 6004 |
+
"step": 853
|
| 6005 |
+
},
|
| 6006 |
+
{
|
| 6007 |
+
"epoch": 0.557896455985628,
|
| 6008 |
+
"grad_norm": Infinity,
|
| 6009 |
+
"learning_rate": 8.645648407577338e-05,
|
| 6010 |
+
"loss": 1.3507,
|
| 6011 |
+
"step": 854
|
| 6012 |
+
},
|
| 6013 |
+
{
|
| 6014 |
+
"epoch": 0.5585497305242528,
|
| 6015 |
+
"grad_norm": Infinity,
|
| 6016 |
+
"learning_rate": 8.624620078102951e-05,
|
| 6017 |
+
"loss": 1.4897,
|
| 6018 |
+
"step": 855
|
| 6019 |
+
},
|
| 6020 |
+
{
|
| 6021 |
+
"epoch": 0.5592030050628777,
|
| 6022 |
+
"grad_norm": Infinity,
|
| 6023 |
+
"learning_rate": 8.603597945875935e-05,
|
| 6024 |
+
"loss": 1.4492,
|
| 6025 |
+
"step": 856
|
| 6026 |
+
},
|
| 6027 |
+
{
|
| 6028 |
+
"epoch": 0.5598562796015025,
|
| 6029 |
+
"grad_norm": Infinity,
|
| 6030 |
+
"learning_rate": 8.582582105618733e-05,
|
| 6031 |
+
"loss": 1.4236,
|
| 6032 |
+
"step": 857
|
| 6033 |
+
},
|
| 6034 |
+
{
|
| 6035 |
+
"epoch": 0.5605095541401274,
|
| 6036 |
+
"grad_norm": Infinity,
|
| 6037 |
+
"learning_rate": 8.561572652025447e-05,
|
| 6038 |
+
"loss": 1.5401,
|
| 6039 |
+
"step": 858
|
| 6040 |
+
},
|
| 6041 |
+
{
|
| 6042 |
+
"epoch": 0.5611628286787522,
|
| 6043 |
+
"grad_norm": Infinity,
|
| 6044 |
+
"learning_rate": 8.540569679761391e-05,
|
| 6045 |
+
"loss": 1.6108,
|
| 6046 |
+
"step": 859
|
| 6047 |
+
},
|
| 6048 |
+
{
|
| 6049 |
+
"epoch": 0.5618161032173771,
|
| 6050 |
+
"grad_norm": Infinity,
|
| 6051 |
+
"learning_rate": 8.519573283462687e-05,
|
| 6052 |
+
"loss": 1.582,
|
| 6053 |
+
"step": 860
|
| 6054 |
+
},
|
| 6055 |
+
{
|
| 6056 |
+
"epoch": 0.5624693777560019,
|
| 6057 |
+
"grad_norm": Infinity,
|
| 6058 |
+
"learning_rate": 8.498583557735819e-05,
|
| 6059 |
+
"loss": 1.6815,
|
| 6060 |
+
"step": 861
|
| 6061 |
+
},
|
| 6062 |
+
{
|
| 6063 |
+
"epoch": 0.5631226522946268,
|
| 6064 |
+
"grad_norm": Infinity,
|
| 6065 |
+
"learning_rate": 8.47760059715722e-05,
|
| 6066 |
+
"loss": 1.5396,
|
| 6067 |
+
"step": 862
|
| 6068 |
+
},
|
| 6069 |
+
{
|
| 6070 |
+
"epoch": 0.5637759268332517,
|
| 6071 |
+
"grad_norm": Infinity,
|
| 6072 |
+
"learning_rate": 8.456624496272829e-05,
|
| 6073 |
+
"loss": 1.7314,
|
| 6074 |
+
"step": 863
|
| 6075 |
+
},
|
| 6076 |
+
{
|
| 6077 |
+
"epoch": 0.5644292013718766,
|
| 6078 |
+
"grad_norm": Infinity,
|
| 6079 |
+
"learning_rate": 8.435655349597689e-05,
|
| 6080 |
+
"loss": 1.9425,
|
| 6081 |
+
"step": 864
|
| 6082 |
+
},
|
| 6083 |
+
{
|
| 6084 |
+
"epoch": 0.5650824759105014,
|
| 6085 |
+
"grad_norm": Infinity,
|
| 6086 |
+
"learning_rate": 8.414693251615512e-05,
|
| 6087 |
+
"loss": 1.8167,
|
| 6088 |
+
"step": 865
|
| 6089 |
+
},
|
| 6090 |
+
{
|
| 6091 |
+
"epoch": 0.5657357504491263,
|
| 6092 |
+
"grad_norm": Infinity,
|
| 6093 |
+
"learning_rate": 8.393738296778228e-05,
|
| 6094 |
+
"loss": 2.0253,
|
| 6095 |
+
"step": 866
|
| 6096 |
+
},
|
| 6097 |
+
{
|
| 6098 |
+
"epoch": 0.5663890249877511,
|
| 6099 |
+
"grad_norm": Infinity,
|
| 6100 |
+
"learning_rate": 8.3727905795056e-05,
|
| 6101 |
+
"loss": 2.0881,
|
| 6102 |
+
"step": 867
|
| 6103 |
+
},
|
| 6104 |
+
{
|
| 6105 |
+
"epoch": 0.567042299526376,
|
| 6106 |
+
"grad_norm": Infinity,
|
| 6107 |
+
"learning_rate": 8.351850194184775e-05,
|
| 6108 |
+
"loss": 2.3032,
|
| 6109 |
+
"step": 868
|
| 6110 |
+
},
|
| 6111 |
+
{
|
| 6112 |
+
"epoch": 0.5676955740650008,
|
| 6113 |
+
"grad_norm": Infinity,
|
| 6114 |
+
"learning_rate": 8.330917235169867e-05,
|
| 6115 |
+
"loss": 2.4988,
|
| 6116 |
+
"step": 869
|
| 6117 |
+
},
|
| 6118 |
+
{
|
| 6119 |
+
"epoch": 0.5683488486036257,
|
| 6120 |
+
"grad_norm": Infinity,
|
| 6121 |
+
"learning_rate": 8.309991796781511e-05,
|
| 6122 |
+
"loss": 2.4876,
|
| 6123 |
+
"step": 870
|
| 6124 |
+
},
|
| 6125 |
+
{
|
| 6126 |
+
"epoch": 0.5690021231422505,
|
| 6127 |
+
"grad_norm": Infinity,
|
| 6128 |
+
"learning_rate": 8.289073973306478e-05,
|
| 6129 |
+
"loss": 2.951,
|
| 6130 |
+
"step": 871
|
| 6131 |
+
},
|
| 6132 |
+
{
|
| 6133 |
+
"epoch": 0.5696553976808754,
|
| 6134 |
+
"grad_norm": Infinity,
|
| 6135 |
+
"learning_rate": 8.268163858997215e-05,
|
| 6136 |
+
"loss": 2.8837,
|
| 6137 |
+
"step": 872
|
| 6138 |
+
},
|
| 6139 |
+
{
|
| 6140 |
+
"epoch": 0.5703086722195002,
|
| 6141 |
+
"grad_norm": Infinity,
|
| 6142 |
+
"learning_rate": 8.247261548071435e-05,
|
| 6143 |
+
"loss": 3.1574,
|
| 6144 |
+
"step": 873
|
| 6145 |
+
},
|
| 6146 |
+
{
|
| 6147 |
+
"epoch": 0.5709619467581251,
|
| 6148 |
+
"grad_norm": Infinity,
|
| 6149 |
+
"learning_rate": 8.226367134711692e-05,
|
| 6150 |
+
"loss": 3.2576,
|
| 6151 |
+
"step": 874
|
| 6152 |
+
},
|
| 6153 |
+
{
|
| 6154 |
+
"epoch": 0.5716152212967499,
|
| 6155 |
+
"grad_norm": Infinity,
|
| 6156 |
+
"learning_rate": 8.205480713064946e-05,
|
| 6157 |
+
"loss": 4.0353,
|
| 6158 |
+
"step": 875
|
| 6159 |
+
},
|
| 6160 |
+
{
|
| 6161 |
+
"epoch": 0.5722684958353749,
|
| 6162 |
+
"grad_norm": Infinity,
|
| 6163 |
+
"learning_rate": 8.184602377242168e-05,
|
| 6164 |
+
"loss": 1.2324,
|
| 6165 |
+
"step": 876
|
| 6166 |
+
},
|
| 6167 |
+
{
|
| 6168 |
+
"epoch": 0.5729217703739997,
|
| 6169 |
+
"grad_norm": Infinity,
|
| 6170 |
+
"learning_rate": 8.163732221317868e-05,
|
| 6171 |
+
"loss": 1.3859,
|
| 6172 |
+
"step": 877
|
| 6173 |
+
},
|
| 6174 |
+
{
|
| 6175 |
+
"epoch": 0.5735750449126246,
|
| 6176 |
+
"grad_norm": Infinity,
|
| 6177 |
+
"learning_rate": 8.142870339329723e-05,
|
| 6178 |
+
"loss": 1.4323,
|
| 6179 |
+
"step": 878
|
| 6180 |
+
},
|
| 6181 |
+
{
|
| 6182 |
+
"epoch": 0.5742283194512494,
|
| 6183 |
+
"grad_norm": Infinity,
|
| 6184 |
+
"learning_rate": 8.12201682527811e-05,
|
| 6185 |
+
"loss": 1.4522,
|
| 6186 |
+
"step": 879
|
| 6187 |
+
},
|
| 6188 |
+
{
|
| 6189 |
+
"epoch": 0.5748815939898743,
|
| 6190 |
+
"grad_norm": Infinity,
|
| 6191 |
+
"learning_rate": 8.101171773125716e-05,
|
| 6192 |
+
"loss": 1.5134,
|
| 6193 |
+
"step": 880
|
| 6194 |
+
},
|
| 6195 |
+
{
|
| 6196 |
+
"epoch": 0.5755348685284991,
|
| 6197 |
+
"grad_norm": Infinity,
|
| 6198 |
+
"learning_rate": 8.080335276797097e-05,
|
| 6199 |
+
"loss": 1.5799,
|
| 6200 |
+
"step": 881
|
| 6201 |
+
},
|
| 6202 |
+
{
|
| 6203 |
+
"epoch": 0.576188143067124,
|
| 6204 |
+
"grad_norm": Infinity,
|
| 6205 |
+
"learning_rate": 8.059507430178247e-05,
|
| 6206 |
+
"loss": 1.4346,
|
| 6207 |
+
"step": 882
|
| 6208 |
+
},
|
| 6209 |
+
{
|
| 6210 |
+
"epoch": 0.5768414176057488,
|
| 6211 |
+
"grad_norm": Infinity,
|
| 6212 |
+
"learning_rate": 8.0386883271162e-05,
|
| 6213 |
+
"loss": 1.5796,
|
| 6214 |
+
"step": 883
|
| 6215 |
+
},
|
| 6216 |
+
{
|
| 6217 |
+
"epoch": 0.5774946921443737,
|
| 6218 |
+
"grad_norm": Infinity,
|
| 6219 |
+
"learning_rate": 8.017878061418582e-05,
|
| 6220 |
+
"loss": 1.6774,
|
| 6221 |
+
"step": 884
|
| 6222 |
+
},
|
| 6223 |
+
{
|
| 6224 |
+
"epoch": 0.5781479666829985,
|
| 6225 |
+
"grad_norm": Infinity,
|
| 6226 |
+
"learning_rate": 7.99707672685321e-05,
|
| 6227 |
+
"loss": 1.6591,
|
| 6228 |
+
"step": 885
|
| 6229 |
+
},
|
| 6230 |
+
{
|
| 6231 |
+
"epoch": 0.5788012412216234,
|
| 6232 |
+
"grad_norm": Infinity,
|
| 6233 |
+
"learning_rate": 7.976284417147644e-05,
|
| 6234 |
+
"loss": 1.7876,
|
| 6235 |
+
"step": 886
|
| 6236 |
+
},
|
| 6237 |
+
{
|
| 6238 |
+
"epoch": 0.5794545157602482,
|
| 6239 |
+
"grad_norm": Infinity,
|
| 6240 |
+
"learning_rate": 7.955501225988792e-05,
|
| 6241 |
+
"loss": 1.7762,
|
| 6242 |
+
"step": 887
|
| 6243 |
+
},
|
| 6244 |
+
{
|
| 6245 |
+
"epoch": 0.580107790298873,
|
| 6246 |
+
"grad_norm": Infinity,
|
| 6247 |
+
"learning_rate": 7.934727247022475e-05,
|
| 6248 |
+
"loss": 1.8609,
|
| 6249 |
+
"step": 888
|
| 6250 |
+
},
|
| 6251 |
+
{
|
| 6252 |
+
"epoch": 0.580761064837498,
|
| 6253 |
+
"grad_norm": Infinity,
|
| 6254 |
+
"learning_rate": 7.913962573852996e-05,
|
| 6255 |
+
"loss": 1.9347,
|
| 6256 |
+
"step": 889
|
| 6257 |
+
},
|
| 6258 |
+
{
|
| 6259 |
+
"epoch": 0.5814143393761229,
|
| 6260 |
+
"grad_norm": Infinity,
|
| 6261 |
+
"learning_rate": 7.89320730004274e-05,
|
| 6262 |
+
"loss": 2.2418,
|
| 6263 |
+
"step": 890
|
| 6264 |
+
},
|
| 6265 |
+
{
|
| 6266 |
+
"epoch": 0.5820676139147477,
|
| 6267 |
+
"grad_norm": Infinity,
|
| 6268 |
+
"learning_rate": 7.87246151911172e-05,
|
| 6269 |
+
"loss": 2.1152,
|
| 6270 |
+
"step": 891
|
| 6271 |
+
},
|
| 6272 |
+
{
|
| 6273 |
+
"epoch": 0.5827208884533726,
|
| 6274 |
+
"grad_norm": Infinity,
|
| 6275 |
+
"learning_rate": 7.851725324537204e-05,
|
| 6276 |
+
"loss": 2.2936,
|
| 6277 |
+
"step": 892
|
| 6278 |
+
},
|
| 6279 |
+
{
|
| 6280 |
+
"epoch": 0.5833741629919974,
|
| 6281 |
+
"grad_norm": Infinity,
|
| 6282 |
+
"learning_rate": 7.830998809753237e-05,
|
| 6283 |
+
"loss": 2.5237,
|
| 6284 |
+
"step": 893
|
| 6285 |
+
},
|
| 6286 |
+
{
|
| 6287 |
+
"epoch": 0.5840274375306223,
|
| 6288 |
+
"grad_norm": Infinity,
|
| 6289 |
+
"learning_rate": 7.81028206815027e-05,
|
| 6290 |
+
"loss": 2.33,
|
| 6291 |
+
"step": 894
|
| 6292 |
+
},
|
| 6293 |
+
{
|
| 6294 |
+
"epoch": 0.5846807120692471,
|
| 6295 |
+
"grad_norm": Infinity,
|
| 6296 |
+
"learning_rate": 7.789575193074704e-05,
|
| 6297 |
+
"loss": 2.8523,
|
| 6298 |
+
"step": 895
|
| 6299 |
+
},
|
| 6300 |
+
{
|
| 6301 |
+
"epoch": 0.585333986607872,
|
| 6302 |
+
"grad_norm": Infinity,
|
| 6303 |
+
"learning_rate": 7.768878277828486e-05,
|
| 6304 |
+
"loss": 2.9459,
|
| 6305 |
+
"step": 896
|
| 6306 |
+
},
|
| 6307 |
+
{
|
| 6308 |
+
"epoch": 0.5859872611464968,
|
| 6309 |
+
"grad_norm": Infinity,
|
| 6310 |
+
"learning_rate": 7.748191415668698e-05,
|
| 6311 |
+
"loss": 3.0787,
|
| 6312 |
+
"step": 897
|
| 6313 |
+
},
|
| 6314 |
+
{
|
| 6315 |
+
"epoch": 0.5866405356851216,
|
| 6316 |
+
"grad_norm": Infinity,
|
| 6317 |
+
"learning_rate": 7.727514699807102e-05,
|
| 6318 |
+
"loss": 2.9497,
|
| 6319 |
+
"step": 898
|
| 6320 |
+
},
|
| 6321 |
+
{
|
| 6322 |
+
"epoch": 0.5872938102237465,
|
| 6323 |
+
"grad_norm": Infinity,
|
| 6324 |
+
"learning_rate": 7.706848223409759e-05,
|
| 6325 |
+
"loss": 3.8379,
|
| 6326 |
+
"step": 899
|
| 6327 |
+
},
|
| 6328 |
+
{
|
| 6329 |
+
"epoch": 0.5879470847623713,
|
| 6330 |
+
"grad_norm": Infinity,
|
| 6331 |
+
"learning_rate": 7.686192079596586e-05,
|
| 6332 |
+
"loss": 4.5886,
|
| 6333 |
+
"step": 900
|
| 6334 |
+
},
|
| 6335 |
+
{
|
| 6336 |
+
"epoch": 0.5886003593009962,
|
| 6337 |
+
"grad_norm": Infinity,
|
| 6338 |
+
"learning_rate": 7.66554636144095e-05,
|
| 6339 |
+
"loss": 1.2221,
|
| 6340 |
+
"step": 901
|
| 6341 |
+
},
|
| 6342 |
+
{
|
| 6343 |
+
"epoch": 0.5892536338396211,
|
| 6344 |
+
"grad_norm": Infinity,
|
| 6345 |
+
"learning_rate": 7.644911161969226e-05,
|
| 6346 |
+
"loss": 1.3855,
|
| 6347 |
+
"step": 902
|
| 6348 |
+
},
|
| 6349 |
+
{
|
| 6350 |
+
"epoch": 0.589906908378246,
|
| 6351 |
+
"grad_norm": Infinity,
|
| 6352 |
+
"learning_rate": 7.624286574160409e-05,
|
| 6353 |
+
"loss": 1.3057,
|
| 6354 |
+
"step": 903
|
| 6355 |
+
},
|
| 6356 |
+
{
|
| 6357 |
+
"epoch": 0.5905601829168708,
|
| 6358 |
+
"grad_norm": Infinity,
|
| 6359 |
+
"learning_rate": 7.603672690945682e-05,
|
| 6360 |
+
"loss": 1.4719,
|
| 6361 |
+
"step": 904
|
| 6362 |
+
},
|
| 6363 |
+
{
|
| 6364 |
+
"epoch": 0.5912134574554957,
|
| 6365 |
+
"grad_norm": Infinity,
|
| 6366 |
+
"learning_rate": 7.583069605207975e-05,
|
| 6367 |
+
"loss": 1.3908,
|
| 6368 |
+
"step": 905
|
| 6369 |
+
},
|
| 6370 |
+
{
|
| 6371 |
+
"epoch": 0.5918667319941205,
|
| 6372 |
+
"grad_norm": Infinity,
|
| 6373 |
+
"learning_rate": 7.562477409781586e-05,
|
| 6374 |
+
"loss": 1.5683,
|
| 6375 |
+
"step": 906
|
| 6376 |
+
},
|
| 6377 |
+
{
|
| 6378 |
+
"epoch": 0.5925200065327454,
|
| 6379 |
+
"grad_norm": Infinity,
|
| 6380 |
+
"learning_rate": 7.541896197451727e-05,
|
| 6381 |
+
"loss": 1.4981,
|
| 6382 |
+
"step": 907
|
| 6383 |
+
},
|
| 6384 |
+
{
|
| 6385 |
+
"epoch": 0.5931732810713702,
|
| 6386 |
+
"grad_norm": Infinity,
|
| 6387 |
+
"learning_rate": 7.521326060954143e-05,
|
| 6388 |
+
"loss": 1.5204,
|
| 6389 |
+
"step": 908
|
| 6390 |
+
},
|
| 6391 |
+
{
|
| 6392 |
+
"epoch": 0.5938265556099951,
|
| 6393 |
+
"grad_norm": Infinity,
|
| 6394 |
+
"learning_rate": 7.500767092974647e-05,
|
| 6395 |
+
"loss": 1.6046,
|
| 6396 |
+
"step": 909
|
| 6397 |
+
},
|
| 6398 |
+
{
|
| 6399 |
+
"epoch": 0.5944798301486199,
|
| 6400 |
+
"grad_norm": Infinity,
|
| 6401 |
+
"learning_rate": 7.48021938614875e-05,
|
| 6402 |
+
"loss": 1.4955,
|
| 6403 |
+
"step": 910
|
| 6404 |
+
},
|
| 6405 |
+
{
|
| 6406 |
+
"epoch": 0.5951331046872448,
|
| 6407 |
+
"grad_norm": Infinity,
|
| 6408 |
+
"learning_rate": 7.459683033061208e-05,
|
| 6409 |
+
"loss": 1.6148,
|
| 6410 |
+
"step": 911
|
| 6411 |
+
},
|
| 6412 |
+
{
|
| 6413 |
+
"epoch": 0.5957863792258696,
|
| 6414 |
+
"grad_norm": Infinity,
|
| 6415 |
+
"learning_rate": 7.439158126245627e-05,
|
| 6416 |
+
"loss": 1.6611,
|
| 6417 |
+
"step": 912
|
| 6418 |
+
},
|
| 6419 |
+
{
|
| 6420 |
+
"epoch": 0.5964396537644945,
|
| 6421 |
+
"grad_norm": Infinity,
|
| 6422 |
+
"learning_rate": 7.418644758184038e-05,
|
| 6423 |
+
"loss": 1.9497,
|
| 6424 |
+
"step": 913
|
| 6425 |
+
},
|
| 6426 |
+
{
|
| 6427 |
+
"epoch": 0.5970929283031194,
|
| 6428 |
+
"grad_norm": Infinity,
|
| 6429 |
+
"learning_rate": 7.398143021306471e-05,
|
| 6430 |
+
"loss": 1.964,
|
| 6431 |
+
"step": 914
|
| 6432 |
+
},
|
| 6433 |
+
{
|
| 6434 |
+
"epoch": 0.5977462028417443,
|
| 6435 |
+
"grad_norm": Infinity,
|
| 6436 |
+
"learning_rate": 7.37765300799056e-05,
|
| 6437 |
+
"loss": 2.007,
|
| 6438 |
+
"step": 915
|
| 6439 |
+
},
|
| 6440 |
+
{
|
| 6441 |
+
"epoch": 0.5983994773803691,
|
| 6442 |
+
"grad_norm": Infinity,
|
| 6443 |
+
"learning_rate": 7.357174810561103e-05,
|
| 6444 |
+
"loss": 2.3996,
|
| 6445 |
+
"step": 916
|
| 6446 |
+
},
|
| 6447 |
+
{
|
| 6448 |
+
"epoch": 0.599052751918994,
|
| 6449 |
+
"grad_norm": Infinity,
|
| 6450 |
+
"learning_rate": 7.336708521289669e-05,
|
| 6451 |
+
"loss": 2.3556,
|
| 6452 |
+
"step": 917
|
| 6453 |
+
},
|
| 6454 |
+
{
|
| 6455 |
+
"epoch": 0.5997060264576188,
|
| 6456 |
+
"grad_norm": Infinity,
|
| 6457 |
+
"learning_rate": 7.316254232394158e-05,
|
| 6458 |
+
"loss": 2.3301,
|
| 6459 |
+
"step": 918
|
| 6460 |
+
},
|
| 6461 |
+
{
|
| 6462 |
+
"epoch": 0.6003593009962437,
|
| 6463 |
+
"grad_norm": Infinity,
|
| 6464 |
+
"learning_rate": 7.295812036038407e-05,
|
| 6465 |
+
"loss": 2.5678,
|
| 6466 |
+
"step": 919
|
| 6467 |
+
},
|
| 6468 |
+
{
|
| 6469 |
+
"epoch": 0.6010125755348685,
|
| 6470 |
+
"grad_norm": Infinity,
|
| 6471 |
+
"learning_rate": 7.275382024331772e-05,
|
| 6472 |
+
"loss": 2.7143,
|
| 6473 |
+
"step": 920
|
| 6474 |
+
},
|
| 6475 |
+
{
|
| 6476 |
+
"epoch": 0.6016658500734934,
|
| 6477 |
+
"grad_norm": Infinity,
|
| 6478 |
+
"learning_rate": 7.254964289328688e-05,
|
| 6479 |
+
"loss": 2.7722,
|
| 6480 |
+
"step": 921
|
| 6481 |
+
},
|
| 6482 |
+
{
|
| 6483 |
+
"epoch": 0.6023191246121182,
|
| 6484 |
+
"grad_norm": Infinity,
|
| 6485 |
+
"learning_rate": 7.234558923028289e-05,
|
| 6486 |
+
"loss": 3.4389,
|
| 6487 |
+
"step": 922
|
| 6488 |
+
},
|
| 6489 |
+
{
|
| 6490 |
+
"epoch": 0.6029723991507431,
|
| 6491 |
+
"grad_norm": Infinity,
|
| 6492 |
+
"learning_rate": 7.21416601737397e-05,
|
| 6493 |
+
"loss": 3.2806,
|
| 6494 |
+
"step": 923
|
| 6495 |
+
},
|
| 6496 |
+
{
|
| 6497 |
+
"epoch": 0.6036256736893679,
|
| 6498 |
+
"grad_norm": Infinity,
|
| 6499 |
+
"learning_rate": 7.193785664252987e-05,
|
| 6500 |
+
"loss": 3.4732,
|
| 6501 |
+
"step": 924
|
| 6502 |
+
},
|
| 6503 |
+
{
|
| 6504 |
+
"epoch": 0.6042789482279928,
|
| 6505 |
+
"grad_norm": Infinity,
|
| 6506 |
+
"learning_rate": 7.173417955496024e-05,
|
| 6507 |
+
"loss": 4.3247,
|
| 6508 |
+
"step": 925
|
| 6509 |
+
},
|
| 6510 |
+
{
|
| 6511 |
+
"epoch": 0.6049322227666176,
|
| 6512 |
+
"grad_norm": Infinity,
|
| 6513 |
+
"learning_rate": 7.153062982876805e-05,
|
| 6514 |
+
"loss": 1.3828,
|
| 6515 |
+
"step": 926
|
| 6516 |
+
},
|
| 6517 |
+
{
|
| 6518 |
+
"epoch": 0.6055854973052426,
|
| 6519 |
+
"grad_norm": Infinity,
|
| 6520 |
+
"learning_rate": 7.132720838111659e-05,
|
| 6521 |
+
"loss": 1.3161,
|
| 6522 |
+
"step": 927
|
| 6523 |
+
},
|
| 6524 |
+
{
|
| 6525 |
+
"epoch": 0.6062387718438674,
|
| 6526 |
+
"grad_norm": Infinity,
|
| 6527 |
+
"learning_rate": 7.112391612859118e-05,
|
| 6528 |
+
"loss": 1.4399,
|
| 6529 |
+
"step": 928
|
| 6530 |
+
},
|
| 6531 |
+
{
|
| 6532 |
+
"epoch": 0.6068920463824923,
|
| 6533 |
+
"grad_norm": Infinity,
|
| 6534 |
+
"learning_rate": 7.092075398719502e-05,
|
| 6535 |
+
"loss": 1.3693,
|
| 6536 |
+
"step": 929
|
| 6537 |
+
},
|
| 6538 |
+
{
|
| 6539 |
+
"epoch": 0.6075453209211171,
|
| 6540 |
+
"grad_norm": Infinity,
|
| 6541 |
+
"learning_rate": 7.071772287234497e-05,
|
| 6542 |
+
"loss": 1.453,
|
| 6543 |
+
"step": 930
|
| 6544 |
+
},
|
| 6545 |
+
{
|
| 6546 |
+
"epoch": 0.608198595459742,
|
| 6547 |
+
"grad_norm": Infinity,
|
| 6548 |
+
"learning_rate": 7.051482369886761e-05,
|
| 6549 |
+
"loss": 1.4425,
|
| 6550 |
+
"step": 931
|
| 6551 |
+
},
|
| 6552 |
+
{
|
| 6553 |
+
"epoch": 0.6088518699983668,
|
| 6554 |
+
"grad_norm": Infinity,
|
| 6555 |
+
"learning_rate": 7.031205738099491e-05,
|
| 6556 |
+
"loss": 1.5857,
|
| 6557 |
+
"step": 932
|
| 6558 |
+
},
|
| 6559 |
+
{
|
| 6560 |
+
"epoch": 0.6095051445369917,
|
| 6561 |
+
"grad_norm": Infinity,
|
| 6562 |
+
"learning_rate": 7.010942483236036e-05,
|
| 6563 |
+
"loss": 1.6388,
|
| 6564 |
+
"step": 933
|
| 6565 |
+
},
|
| 6566 |
+
{
|
| 6567 |
+
"epoch": 0.6101584190756165,
|
| 6568 |
+
"grad_norm": Infinity,
|
| 6569 |
+
"learning_rate": 6.990692696599447e-05,
|
| 6570 |
+
"loss": 1.6966,
|
| 6571 |
+
"step": 934
|
| 6572 |
+
},
|
| 6573 |
+
{
|
| 6574 |
+
"epoch": 0.6108116936142414,
|
| 6575 |
+
"grad_norm": Infinity,
|
| 6576 |
+
"learning_rate": 6.970456469432117e-05,
|
| 6577 |
+
"loss": 1.7996,
|
| 6578 |
+
"step": 935
|
| 6579 |
+
},
|
| 6580 |
+
{
|
| 6581 |
+
"epoch": 0.6114649681528662,
|
| 6582 |
+
"grad_norm": Infinity,
|
| 6583 |
+
"learning_rate": 6.950233892915323e-05,
|
| 6584 |
+
"loss": 1.6195,
|
| 6585 |
+
"step": 936
|
| 6586 |
+
},
|
| 6587 |
+
{
|
| 6588 |
+
"epoch": 0.6121182426914911,
|
| 6589 |
+
"grad_norm": Infinity,
|
| 6590 |
+
"learning_rate": 6.93002505816884e-05,
|
| 6591 |
+
"loss": 1.6846,
|
| 6592 |
+
"step": 937
|
| 6593 |
+
},
|
| 6594 |
+
{
|
| 6595 |
+
"epoch": 0.6127715172301159,
|
| 6596 |
+
"grad_norm": Infinity,
|
| 6597 |
+
"learning_rate": 6.909830056250527e-05,
|
| 6598 |
+
"loss": 1.9402,
|
| 6599 |
+
"step": 938
|
| 6600 |
+
},
|
| 6601 |
+
{
|
| 6602 |
+
"epoch": 0.6134247917687408,
|
| 6603 |
+
"grad_norm": Infinity,
|
| 6604 |
+
"learning_rate": 6.889648978155909e-05,
|
| 6605 |
+
"loss": 1.8907,
|
| 6606 |
+
"step": 939
|
| 6607 |
+
},
|
| 6608 |
+
{
|
| 6609 |
+
"epoch": 0.6140780663073657,
|
| 6610 |
+
"grad_norm": Infinity,
|
| 6611 |
+
"learning_rate": 6.869481914817779e-05,
|
| 6612 |
+
"loss": 1.855,
|
| 6613 |
+
"step": 940
|
| 6614 |
+
},
|
| 6615 |
+
{
|
| 6616 |
+
"epoch": 0.6147313408459906,
|
| 6617 |
+
"grad_norm": Infinity,
|
| 6618 |
+
"learning_rate": 6.849328957105773e-05,
|
| 6619 |
+
"loss": 2.1814,
|
| 6620 |
+
"step": 941
|
| 6621 |
+
},
|
| 6622 |
+
{
|
| 6623 |
+
"epoch": 0.6153846153846154,
|
| 6624 |
+
"grad_norm": Infinity,
|
| 6625 |
+
"learning_rate": 6.829190195825983e-05,
|
| 6626 |
+
"loss": 2.4298,
|
| 6627 |
+
"step": 942
|
| 6628 |
+
},
|
| 6629 |
+
{
|
| 6630 |
+
"epoch": 0.6160378899232403,
|
| 6631 |
+
"grad_norm": Infinity,
|
| 6632 |
+
"learning_rate": 6.809065721720516e-05,
|
| 6633 |
+
"loss": 2.5652,
|
| 6634 |
+
"step": 943
|
| 6635 |
+
},
|
| 6636 |
+
{
|
| 6637 |
+
"epoch": 0.6166911644618651,
|
| 6638 |
+
"grad_norm": Infinity,
|
| 6639 |
+
"learning_rate": 6.788955625467119e-05,
|
| 6640 |
+
"loss": 2.3691,
|
| 6641 |
+
"step": 944
|
| 6642 |
+
},
|
| 6643 |
+
{
|
| 6644 |
+
"epoch": 0.61734443900049,
|
| 6645 |
+
"grad_norm": Infinity,
|
| 6646 |
+
"learning_rate": 6.768859997678751e-05,
|
| 6647 |
+
"loss": 2.7052,
|
| 6648 |
+
"step": 945
|
| 6649 |
+
},
|
| 6650 |
+
{
|
| 6651 |
+
"epoch": 0.6179977135391148,
|
| 6652 |
+
"grad_norm": Infinity,
|
| 6653 |
+
"learning_rate": 6.748778928903166e-05,
|
| 6654 |
+
"loss": 3.2163,
|
| 6655 |
+
"step": 946
|
| 6656 |
+
},
|
| 6657 |
+
{
|
| 6658 |
+
"epoch": 0.6186509880777397,
|
| 6659 |
+
"grad_norm": Infinity,
|
| 6660 |
+
"learning_rate": 6.728712509622539e-05,
|
| 6661 |
+
"loss": 2.8301,
|
| 6662 |
+
"step": 947
|
| 6663 |
+
},
|
| 6664 |
+
{
|
| 6665 |
+
"epoch": 0.6193042626163645,
|
| 6666 |
+
"grad_norm": Infinity,
|
| 6667 |
+
"learning_rate": 6.708660830253015e-05,
|
| 6668 |
+
"loss": 3.2148,
|
| 6669 |
+
"step": 948
|
| 6670 |
+
},
|
| 6671 |
+
{
|
| 6672 |
+
"epoch": 0.6199575371549894,
|
| 6673 |
+
"grad_norm": Infinity,
|
| 6674 |
+
"learning_rate": 6.688623981144339e-05,
|
| 6675 |
+
"loss": 3.767,
|
| 6676 |
+
"step": 949
|
| 6677 |
+
},
|
| 6678 |
+
{
|
| 6679 |
+
"epoch": 0.6206108116936142,
|
| 6680 |
+
"grad_norm": Infinity,
|
| 6681 |
+
"learning_rate": 6.668602052579424e-05,
|
| 6682 |
+
"loss": 4.4411,
|
| 6683 |
+
"step": 950
|
| 6684 |
+
},
|
| 6685 |
+
{
|
| 6686 |
+
"epoch": 0.621264086232239,
|
| 6687 |
+
"grad_norm": Infinity,
|
| 6688 |
+
"learning_rate": 6.648595134773958e-05,
|
| 6689 |
+
"loss": 1.2227,
|
| 6690 |
+
"step": 951
|
| 6691 |
+
},
|
| 6692 |
+
{
|
| 6693 |
+
"epoch": 0.6219173607708639,
|
| 6694 |
+
"grad_norm": Infinity,
|
| 6695 |
+
"learning_rate": 6.628603317875996e-05,
|
| 6696 |
+
"loss": 1.466,
|
| 6697 |
+
"step": 952
|
| 6698 |
+
},
|
| 6699 |
+
{
|
| 6700 |
+
"epoch": 0.6225706353094889,
|
| 6701 |
+
"grad_norm": Infinity,
|
| 6702 |
+
"learning_rate": 6.608626691965541e-05,
|
| 6703 |
+
"loss": 1.3718,
|
| 6704 |
+
"step": 953
|
| 6705 |
+
},
|
| 6706 |
+
{
|
| 6707 |
+
"epoch": 0.6232239098481137,
|
| 6708 |
+
"grad_norm": Infinity,
|
| 6709 |
+
"learning_rate": 6.588665347054153e-05,
|
| 6710 |
+
"loss": 1.4628,
|
| 6711 |
+
"step": 954
|
| 6712 |
+
},
|
| 6713 |
+
{
|
| 6714 |
+
"epoch": 0.6238771843867386,
|
| 6715 |
+
"grad_norm": Infinity,
|
| 6716 |
+
"learning_rate": 6.568719373084538e-05,
|
| 6717 |
+
"loss": 1.4955,
|
| 6718 |
+
"step": 955
|
| 6719 |
+
},
|
| 6720 |
+
{
|
| 6721 |
+
"epoch": 0.6245304589253634,
|
| 6722 |
+
"grad_norm": Infinity,
|
| 6723 |
+
"learning_rate": 6.548788859930146e-05,
|
| 6724 |
+
"loss": 1.5999,
|
| 6725 |
+
"step": 956
|
| 6726 |
+
},
|
| 6727 |
+
{
|
| 6728 |
+
"epoch": 0.6251837334639883,
|
| 6729 |
+
"grad_norm": Infinity,
|
| 6730 |
+
"learning_rate": 6.528873897394757e-05,
|
| 6731 |
+
"loss": 1.5084,
|
| 6732 |
+
"step": 957
|
| 6733 |
+
},
|
| 6734 |
+
{
|
| 6735 |
+
"epoch": 0.6258370080026131,
|
| 6736 |
+
"grad_norm": Infinity,
|
| 6737 |
+
"learning_rate": 6.508974575212088e-05,
|
| 6738 |
+
"loss": 1.4152,
|
| 6739 |
+
"step": 958
|
| 6740 |
+
},
|
| 6741 |
+
{
|
| 6742 |
+
"epoch": 0.626490282541238,
|
| 6743 |
+
"grad_norm": Infinity,
|
| 6744 |
+
"learning_rate": 6.489090983045379e-05,
|
| 6745 |
+
"loss": 1.6017,
|
| 6746 |
+
"step": 959
|
| 6747 |
+
},
|
| 6748 |
+
{
|
| 6749 |
+
"epoch": 0.6271435570798628,
|
| 6750 |
+
"grad_norm": Infinity,
|
| 6751 |
+
"learning_rate": 6.469223210486992e-05,
|
| 6752 |
+
"loss": 1.642,
|
| 6753 |
+
"step": 960
|
| 6754 |
+
},
|
| 6755 |
+
{
|
| 6756 |
+
"epoch": 0.6277968316184876,
|
| 6757 |
+
"grad_norm": Infinity,
|
| 6758 |
+
"learning_rate": 6.449371347058019e-05,
|
| 6759 |
+
"loss": 1.6396,
|
| 6760 |
+
"step": 961
|
| 6761 |
+
},
|
| 6762 |
+
{
|
| 6763 |
+
"epoch": 0.6284501061571125,
|
| 6764 |
+
"grad_norm": Infinity,
|
| 6765 |
+
"learning_rate": 6.429535482207846e-05,
|
| 6766 |
+
"loss": 1.8072,
|
| 6767 |
+
"step": 962
|
| 6768 |
+
},
|
| 6769 |
+
{
|
| 6770 |
+
"epoch": 0.6291033806957373,
|
| 6771 |
+
"grad_norm": Infinity,
|
| 6772 |
+
"learning_rate": 6.409715705313804e-05,
|
| 6773 |
+
"loss": 1.7597,
|
| 6774 |
+
"step": 963
|
| 6775 |
+
},
|
| 6776 |
+
{
|
| 6777 |
+
"epoch": 0.6297566552343622,
|
| 6778 |
+
"grad_norm": Infinity,
|
| 6779 |
+
"learning_rate": 6.389912105680697e-05,
|
| 6780 |
+
"loss": 1.8747,
|
| 6781 |
+
"step": 964
|
| 6782 |
+
},
|
| 6783 |
+
{
|
| 6784 |
+
"epoch": 0.630409929772987,
|
| 6785 |
+
"grad_norm": Infinity,
|
| 6786 |
+
"learning_rate": 6.370124772540469e-05,
|
| 6787 |
+
"loss": 2.0892,
|
| 6788 |
+
"step": 965
|
| 6789 |
+
},
|
| 6790 |
+
{
|
| 6791 |
+
"epoch": 0.631063204311612,
|
| 6792 |
+
"grad_norm": Infinity,
|
| 6793 |
+
"learning_rate": 6.350353795051748e-05,
|
| 6794 |
+
"loss": 2.1664,
|
| 6795 |
+
"step": 966
|
| 6796 |
+
},
|
| 6797 |
+
{
|
| 6798 |
+
"epoch": 0.6317164788502369,
|
| 6799 |
+
"grad_norm": Infinity,
|
| 6800 |
+
"learning_rate": 6.330599262299479e-05,
|
| 6801 |
+
"loss": 2.094,
|
| 6802 |
+
"step": 967
|
| 6803 |
+
},
|
| 6804 |
+
{
|
| 6805 |
+
"epoch": 0.6323697533888617,
|
| 6806 |
+
"grad_norm": Infinity,
|
| 6807 |
+
"learning_rate": 6.310861263294507e-05,
|
| 6808 |
+
"loss": 2.3747,
|
| 6809 |
+
"step": 968
|
| 6810 |
+
},
|
| 6811 |
+
{
|
| 6812 |
+
"epoch": 0.6330230279274865,
|
| 6813 |
+
"grad_norm": Infinity,
|
| 6814 |
+
"learning_rate": 6.291139886973169e-05,
|
| 6815 |
+
"loss": 2.2346,
|
| 6816 |
+
"step": 969
|
| 6817 |
+
},
|
| 6818 |
+
{
|
| 6819 |
+
"epoch": 0.6336763024661114,
|
| 6820 |
+
"grad_norm": Infinity,
|
| 6821 |
+
"learning_rate": 6.271435222196916e-05,
|
| 6822 |
+
"loss": 2.6554,
|
| 6823 |
+
"step": 970
|
| 6824 |
+
},
|
| 6825 |
+
{
|
| 6826 |
+
"epoch": 0.6343295770047362,
|
| 6827 |
+
"grad_norm": Infinity,
|
| 6828 |
+
"learning_rate": 6.251747357751889e-05,
|
| 6829 |
+
"loss": 2.7962,
|
| 6830 |
+
"step": 971
|
| 6831 |
+
},
|
| 6832 |
+
{
|
| 6833 |
+
"epoch": 0.6349828515433611,
|
| 6834 |
+
"grad_norm": Infinity,
|
| 6835 |
+
"learning_rate": 6.232076382348539e-05,
|
| 6836 |
+
"loss": 3.1112,
|
| 6837 |
+
"step": 972
|
| 6838 |
+
},
|
| 6839 |
+
{
|
| 6840 |
+
"epoch": 0.6356361260819859,
|
| 6841 |
+
"grad_norm": Infinity,
|
| 6842 |
+
"learning_rate": 6.212422384621208e-05,
|
| 6843 |
+
"loss": 3.0868,
|
| 6844 |
+
"step": 973
|
| 6845 |
+
},
|
| 6846 |
+
{
|
| 6847 |
+
"epoch": 0.6362894006206108,
|
| 6848 |
+
"grad_norm": Infinity,
|
| 6849 |
+
"learning_rate": 6.192785453127748e-05,
|
| 6850 |
+
"loss": 3.2682,
|
| 6851 |
+
"step": 974
|
| 6852 |
+
},
|
| 6853 |
+
{
|
| 6854 |
+
"epoch": 0.6369426751592356,
|
| 6855 |
+
"grad_norm": Infinity,
|
| 6856 |
+
"learning_rate": 6.173165676349103e-05,
|
| 6857 |
+
"loss": 4.1133,
|
| 6858 |
+
"step": 975
|
| 6859 |
+
},
|
| 6860 |
+
{
|
| 6861 |
+
"epoch": 0.6375959496978605,
|
| 6862 |
+
"grad_norm": Infinity,
|
| 6863 |
+
"learning_rate": 6.15356314268893e-05,
|
| 6864 |
+
"loss": 1.2235,
|
| 6865 |
+
"step": 976
|
| 6866 |
+
},
|
| 6867 |
+
{
|
| 6868 |
+
"epoch": 0.6382492242364853,
|
| 6869 |
+
"grad_norm": Infinity,
|
| 6870 |
+
"learning_rate": 6.13397794047319e-05,
|
| 6871 |
+
"loss": 1.3234,
|
| 6872 |
+
"step": 977
|
| 6873 |
+
},
|
| 6874 |
+
{
|
| 6875 |
+
"epoch": 0.6389024987751103,
|
| 6876 |
+
"grad_norm": Infinity,
|
| 6877 |
+
"learning_rate": 6.114410157949745e-05,
|
| 6878 |
+
"loss": 1.4325,
|
| 6879 |
+
"step": 978
|
| 6880 |
+
},
|
| 6881 |
+
{
|
| 6882 |
+
"epoch": 0.6395557733137351,
|
| 6883 |
+
"grad_norm": Infinity,
|
| 6884 |
+
"learning_rate": 6.094859883287977e-05,
|
| 6885 |
+
"loss": 1.3914,
|
| 6886 |
+
"step": 979
|
| 6887 |
+
},
|
| 6888 |
+
{
|
| 6889 |
+
"epoch": 0.64020904785236,
|
| 6890 |
+
"grad_norm": Infinity,
|
| 6891 |
+
"learning_rate": 6.0753272045783625e-05,
|
| 6892 |
+
"loss": 1.4688,
|
| 6893 |
+
"step": 980
|
| 6894 |
+
},
|
| 6895 |
+
{
|
| 6896 |
+
"epoch": 0.6408623223909848,
|
| 6897 |
+
"grad_norm": Infinity,
|
| 6898 |
+
"learning_rate": 6.0558122098321144e-05,
|
| 6899 |
+
"loss": 1.4596,
|
| 6900 |
+
"step": 981
|
| 6901 |
+
},
|
| 6902 |
+
{
|
| 6903 |
+
"epoch": 0.6415155969296097,
|
| 6904 |
+
"grad_norm": Infinity,
|
| 6905 |
+
"learning_rate": 6.036314986980749e-05,
|
| 6906 |
+
"loss": 1.5476,
|
| 6907 |
+
"step": 982
|
| 6908 |
+
},
|
| 6909 |
+
{
|
| 6910 |
+
"epoch": 0.6421688714682345,
|
| 6911 |
+
"grad_norm": Infinity,
|
| 6912 |
+
"learning_rate": 6.016835623875716e-05,
|
| 6913 |
+
"loss": 1.5854,
|
| 6914 |
+
"step": 983
|
| 6915 |
+
},
|
| 6916 |
+
{
|
| 6917 |
+
"epoch": 0.6428221460068594,
|
| 6918 |
+
"grad_norm": Infinity,
|
| 6919 |
+
"learning_rate": 5.9973742082879815e-05,
|
| 6920 |
+
"loss": 1.7013,
|
| 6921 |
+
"step": 984
|
| 6922 |
+
},
|
| 6923 |
+
{
|
| 6924 |
+
"epoch": 0.6434754205454842,
|
| 6925 |
+
"grad_norm": Infinity,
|
| 6926 |
+
"learning_rate": 5.977930827907649e-05,
|
| 6927 |
+
"loss": 1.7004,
|
| 6928 |
+
"step": 985
|
| 6929 |
+
},
|
| 6930 |
+
{
|
| 6931 |
+
"epoch": 0.6441286950841091,
|
| 6932 |
+
"grad_norm": Infinity,
|
| 6933 |
+
"learning_rate": 5.958505570343561e-05,
|
| 6934 |
+
"loss": 1.7823,
|
| 6935 |
+
"step": 986
|
| 6936 |
+
},
|
| 6937 |
+
{
|
| 6938 |
+
"epoch": 0.6447819696227339,
|
| 6939 |
+
"grad_norm": Infinity,
|
| 6940 |
+
"learning_rate": 5.9390985231228936e-05,
|
| 6941 |
+
"loss": 1.7044,
|
| 6942 |
+
"step": 987
|
| 6943 |
+
},
|
| 6944 |
+
{
|
| 6945 |
+
"epoch": 0.6454352441613588,
|
| 6946 |
+
"grad_norm": Infinity,
|
| 6947 |
+
"learning_rate": 5.91970977369078e-05,
|
| 6948 |
+
"loss": 1.9262,
|
| 6949 |
+
"step": 988
|
| 6950 |
+
},
|
| 6951 |
+
{
|
| 6952 |
+
"epoch": 0.6460885186999836,
|
| 6953 |
+
"grad_norm": Infinity,
|
| 6954 |
+
"learning_rate": 5.90033940940989e-05,
|
| 6955 |
+
"loss": 1.9572,
|
| 6956 |
+
"step": 989
|
| 6957 |
+
},
|
| 6958 |
+
{
|
| 6959 |
+
"epoch": 0.6467417932386085,
|
| 6960 |
+
"grad_norm": Infinity,
|
| 6961 |
+
"learning_rate": 5.880987517560075e-05,
|
| 6962 |
+
"loss": 1.9857,
|
| 6963 |
+
"step": 990
|
| 6964 |
+
},
|
| 6965 |
+
{
|
| 6966 |
+
"epoch": 0.6473950677772334,
|
| 6967 |
+
"grad_norm": Infinity,
|
| 6968 |
+
"learning_rate": 5.861654185337933e-05,
|
| 6969 |
+
"loss": 2.3491,
|
| 6970 |
+
"step": 991
|
| 6971 |
+
},
|
| 6972 |
+
{
|
| 6973 |
+
"epoch": 0.6480483423158583,
|
| 6974 |
+
"grad_norm": Infinity,
|
| 6975 |
+
"learning_rate": 5.842339499856444e-05,
|
| 6976 |
+
"loss": 2.3709,
|
| 6977 |
+
"step": 992
|
| 6978 |
+
},
|
| 6979 |
+
{
|
| 6980 |
+
"epoch": 0.6487016168544831,
|
| 6981 |
+
"grad_norm": Infinity,
|
| 6982 |
+
"learning_rate": 5.8230435481445734e-05,
|
| 6983 |
+
"loss": 2.3352,
|
| 6984 |
+
"step": 993
|
| 6985 |
+
},
|
| 6986 |
+
{
|
| 6987 |
+
"epoch": 0.649354891393108,
|
| 6988 |
+
"grad_norm": Infinity,
|
| 6989 |
+
"learning_rate": 5.8037664171468596e-05,
|
| 6990 |
+
"loss": 2.7182,
|
| 6991 |
+
"step": 994
|
| 6992 |
+
},
|
| 6993 |
+
{
|
| 6994 |
+
"epoch": 0.6500081659317328,
|
| 6995 |
+
"grad_norm": Infinity,
|
| 6996 |
+
"learning_rate": 5.784508193723057e-05,
|
| 6997 |
+
"loss": 2.823,
|
| 6998 |
+
"step": 995
|
| 6999 |
+
},
|
| 7000 |
+
{
|
| 7001 |
+
"epoch": 0.6506614404703577,
|
| 7002 |
+
"grad_norm": Infinity,
|
| 7003 |
+
"learning_rate": 5.765268964647712e-05,
|
| 7004 |
+
"loss": 2.8539,
|
| 7005 |
+
"step": 996
|
| 7006 |
+
},
|
| 7007 |
+
{
|
| 7008 |
+
"epoch": 0.6513147150089825,
|
| 7009 |
+
"grad_norm": Infinity,
|
| 7010 |
+
"learning_rate": 5.746048816609788e-05,
|
| 7011 |
+
"loss": 2.871,
|
| 7012 |
+
"step": 997
|
| 7013 |
+
},
|
| 7014 |
+
{
|
| 7015 |
+
"epoch": 0.6519679895476074,
|
| 7016 |
+
"grad_norm": Infinity,
|
| 7017 |
+
"learning_rate": 5.726847836212277e-05,
|
| 7018 |
+
"loss": 3.035,
|
| 7019 |
+
"step": 998
|
| 7020 |
+
},
|
| 7021 |
+
{
|
| 7022 |
+
"epoch": 0.6526212640862322,
|
| 7023 |
+
"grad_norm": Infinity,
|
| 7024 |
+
"learning_rate": 5.7076661099717986e-05,
|
| 7025 |
+
"loss": 4.0633,
|
| 7026 |
+
"step": 999
|
| 7027 |
+
},
|
| 7028 |
+
{
|
| 7029 |
+
"epoch": 0.6532745386248571,
|
| 7030 |
+
"grad_norm": Infinity,
|
| 7031 |
+
"learning_rate": 5.688503724318217e-05,
|
| 7032 |
+
"loss": 3.8549,
|
| 7033 |
+
"step": 1000
|
| 7034 |
+
},
|
| 7035 |
+
{
|
| 7036 |
+
"epoch": 0.6539278131634819,
|
| 7037 |
+
"grad_norm": Infinity,
|
| 7038 |
+
"learning_rate": 5.669360765594258e-05,
|
| 7039 |
+
"loss": 1.2725,
|
| 7040 |
+
"step": 1001
|
| 7041 |
+
},
|
| 7042 |
+
{
|
| 7043 |
+
"epoch": 0.6545810877021068,
|
| 7044 |
+
"grad_norm": Infinity,
|
| 7045 |
+
"learning_rate": 5.6502373200551065e-05,
|
| 7046 |
+
"loss": 1.3742,
|
| 7047 |
+
"step": 1002
|
| 7048 |
+
},
|
| 7049 |
+
{
|
| 7050 |
+
"epoch": 0.6552343622407316,
|
| 7051 |
+
"grad_norm": Infinity,
|
| 7052 |
+
"learning_rate": 5.631133473868018e-05,
|
| 7053 |
+
"loss": 1.3706,
|
| 7054 |
+
"step": 1003
|
| 7055 |
+
},
|
| 7056 |
+
{
|
| 7057 |
+
"epoch": 0.6558876367793566,
|
| 7058 |
+
"grad_norm": Infinity,
|
| 7059 |
+
"learning_rate": 5.6120493131119555e-05,
|
| 7060 |
+
"loss": 1.4821,
|
| 7061 |
+
"step": 1004
|
| 7062 |
+
},
|
| 7063 |
+
{
|
| 7064 |
+
"epoch": 0.6565409113179814,
|
| 7065 |
+
"grad_norm": Infinity,
|
| 7066 |
+
"learning_rate": 5.5929849237771556e-05,
|
| 7067 |
+
"loss": 1.4066,
|
| 7068 |
+
"step": 1005
|
| 7069 |
+
},
|
| 7070 |
+
{
|
| 7071 |
+
"epoch": 0.6571941858566063,
|
| 7072 |
+
"grad_norm": Infinity,
|
| 7073 |
+
"learning_rate": 5.573940391764796e-05,
|
| 7074 |
+
"loss": 1.4638,
|
| 7075 |
+
"step": 1006
|
| 7076 |
+
},
|
| 7077 |
+
{
|
| 7078 |
+
"epoch": 0.6578474603952311,
|
| 7079 |
+
"grad_norm": Infinity,
|
| 7080 |
+
"learning_rate": 5.554915802886558e-05,
|
| 7081 |
+
"loss": 1.4518,
|
| 7082 |
+
"step": 1007
|
| 7083 |
+
},
|
| 7084 |
+
{
|
| 7085 |
+
"epoch": 0.658500734933856,
|
| 7086 |
+
"grad_norm": Infinity,
|
| 7087 |
+
"learning_rate": 5.535911242864271e-05,
|
| 7088 |
+
"loss": 1.515,
|
| 7089 |
+
"step": 1008
|
| 7090 |
+
},
|
| 7091 |
+
{
|
| 7092 |
+
"epoch": 0.6591540094724808,
|
| 7093 |
+
"grad_norm": Infinity,
|
| 7094 |
+
"learning_rate": 5.5169267973295294e-05,
|
| 7095 |
+
"loss": 1.5506,
|
| 7096 |
+
"step": 1009
|
| 7097 |
+
},
|
| 7098 |
+
{
|
| 7099 |
+
"epoch": 0.6598072840111057,
|
| 7100 |
+
"grad_norm": Infinity,
|
| 7101 |
+
"learning_rate": 5.497962551823266e-05,
|
| 7102 |
+
"loss": 1.6251,
|
| 7103 |
+
"step": 1010
|
| 7104 |
+
},
|
| 7105 |
+
{
|
| 7106 |
+
"epoch": 0.6604605585497305,
|
| 7107 |
+
"grad_norm": Infinity,
|
| 7108 |
+
"learning_rate": 5.479018591795426e-05,
|
| 7109 |
+
"loss": 1.5996,
|
| 7110 |
+
"step": 1011
|
| 7111 |
+
},
|
| 7112 |
+
{
|
| 7113 |
+
"epoch": 0.6611138330883554,
|
| 7114 |
+
"grad_norm": Infinity,
|
| 7115 |
+
"learning_rate": 5.4600950026045326e-05,
|
| 7116 |
+
"loss": 1.6873,
|
| 7117 |
+
"step": 1012
|
| 7118 |
+
},
|
| 7119 |
+
{
|
| 7120 |
+
"epoch": 0.6617671076269802,
|
| 7121 |
+
"grad_norm": Infinity,
|
| 7122 |
+
"learning_rate": 5.441191869517328e-05,
|
| 7123 |
+
"loss": 1.7156,
|
| 7124 |
+
"step": 1013
|
| 7125 |
+
},
|
| 7126 |
+
{
|
| 7127 |
+
"epoch": 0.6624203821656051,
|
| 7128 |
+
"grad_norm": Infinity,
|
| 7129 |
+
"learning_rate": 5.422309277708379e-05,
|
| 7130 |
+
"loss": 1.8523,
|
| 7131 |
+
"step": 1014
|
| 7132 |
+
},
|
| 7133 |
+
{
|
| 7134 |
+
"epoch": 0.6630736567042299,
|
| 7135 |
+
"grad_norm": Infinity,
|
| 7136 |
+
"learning_rate": 5.403447312259702e-05,
|
| 7137 |
+
"loss": 2.0574,
|
| 7138 |
+
"step": 1015
|
| 7139 |
+
},
|
| 7140 |
+
{
|
| 7141 |
+
"epoch": 0.6637269312428548,
|
| 7142 |
+
"grad_norm": Infinity,
|
| 7143 |
+
"learning_rate": 5.384606058160363e-05,
|
| 7144 |
+
"loss": 1.8924,
|
| 7145 |
+
"step": 1016
|
| 7146 |
+
},
|
| 7147 |
+
{
|
| 7148 |
+
"epoch": 0.6643802057814797,
|
| 7149 |
+
"grad_norm": Infinity,
|
| 7150 |
+
"learning_rate": 5.365785600306124e-05,
|
| 7151 |
+
"loss": 2.3307,
|
| 7152 |
+
"step": 1017
|
| 7153 |
+
},
|
| 7154 |
+
{
|
| 7155 |
+
"epoch": 0.6650334803201046,
|
| 7156 |
+
"grad_norm": Infinity,
|
| 7157 |
+
"learning_rate": 5.346986023499027e-05,
|
| 7158 |
+
"loss": 2.5785,
|
| 7159 |
+
"step": 1018
|
| 7160 |
+
},
|
| 7161 |
+
{
|
| 7162 |
+
"epoch": 0.6656867548587294,
|
| 7163 |
+
"grad_norm": Infinity,
|
| 7164 |
+
"learning_rate": 5.3282074124470284e-05,
|
| 7165 |
+
"loss": 2.4408,
|
| 7166 |
+
"step": 1019
|
| 7167 |
+
},
|
| 7168 |
+
{
|
| 7169 |
+
"epoch": 0.6663400293973543,
|
| 7170 |
+
"grad_norm": Infinity,
|
| 7171 |
+
"learning_rate": 5.309449851763633e-05,
|
| 7172 |
+
"loss": 2.4936,
|
| 7173 |
+
"step": 1020
|
| 7174 |
+
},
|
| 7175 |
+
{
|
| 7176 |
+
"epoch": 0.6669933039359791,
|
| 7177 |
+
"grad_norm": Infinity,
|
| 7178 |
+
"learning_rate": 5.290713425967466e-05,
|
| 7179 |
+
"loss": 2.7817,
|
| 7180 |
+
"step": 1021
|
| 7181 |
+
},
|
| 7182 |
+
{
|
| 7183 |
+
"epoch": 0.667646578474604,
|
| 7184 |
+
"grad_norm": Infinity,
|
| 7185 |
+
"learning_rate": 5.271998219481953e-05,
|
| 7186 |
+
"loss": 3.0024,
|
| 7187 |
+
"step": 1022
|
| 7188 |
+
},
|
| 7189 |
+
{
|
| 7190 |
+
"epoch": 0.6682998530132288,
|
| 7191 |
+
"grad_norm": Infinity,
|
| 7192 |
+
"learning_rate": 5.2533043166348886e-05,
|
| 7193 |
+
"loss": 2.867,
|
| 7194 |
+
"step": 1023
|
| 7195 |
+
},
|
| 7196 |
+
{
|
| 7197 |
+
"epoch": 0.6689531275518537,
|
| 7198 |
+
"grad_norm": Infinity,
|
| 7199 |
+
"learning_rate": 5.234631801658081e-05,
|
| 7200 |
+
"loss": 3.1437,
|
| 7201 |
+
"step": 1024
|
| 7202 |
+
},
|
| 7203 |
+
{
|
| 7204 |
+
"epoch": 0.6696064020904785,
|
| 7205 |
+
"grad_norm": Infinity,
|
| 7206 |
+
"learning_rate": 5.215980758686978e-05,
|
| 7207 |
+
"loss": 3.8861,
|
| 7208 |
+
"step": 1025
|
| 7209 |
+
},
|
| 7210 |
+
{
|
| 7211 |
+
"epoch": 0.6702596766291034,
|
| 7212 |
+
"grad_norm": Infinity,
|
| 7213 |
+
"learning_rate": 5.197351271760258e-05,
|
| 7214 |
+
"loss": 1.2455,
|
| 7215 |
+
"step": 1026
|
| 7216 |
+
},
|
| 7217 |
+
{
|
| 7218 |
+
"epoch": 0.6709129511677282,
|
| 7219 |
+
"grad_norm": Infinity,
|
| 7220 |
+
"learning_rate": 5.178743424819492e-05,
|
| 7221 |
+
"loss": 1.4044,
|
| 7222 |
+
"step": 1027
|
| 7223 |
+
},
|
| 7224 |
+
{
|
| 7225 |
+
"epoch": 0.671566225706353,
|
| 7226 |
+
"grad_norm": Infinity,
|
| 7227 |
+
"learning_rate": 5.160157301708732e-05,
|
| 7228 |
+
"loss": 1.4414,
|
| 7229 |
+
"step": 1028
|
| 7230 |
+
},
|
| 7231 |
+
{
|
| 7232 |
+
"epoch": 0.6722195002449779,
|
| 7233 |
+
"grad_norm": Infinity,
|
| 7234 |
+
"learning_rate": 5.141592986174151e-05,
|
| 7235 |
+
"loss": 1.4843,
|
| 7236 |
+
"step": 1029
|
| 7237 |
+
},
|
| 7238 |
+
{
|
| 7239 |
+
"epoch": 0.6728727747836029,
|
| 7240 |
+
"grad_norm": Infinity,
|
| 7241 |
+
"learning_rate": 5.123050561863657e-05,
|
| 7242 |
+
"loss": 1.473,
|
| 7243 |
+
"step": 1030
|
| 7244 |
+
},
|
| 7245 |
+
{
|
| 7246 |
+
"epoch": 0.6735260493222277,
|
| 7247 |
+
"grad_norm": Infinity,
|
| 7248 |
+
"learning_rate": 5.104530112326522e-05,
|
| 7249 |
+
"loss": 1.5881,
|
| 7250 |
+
"step": 1031
|
| 7251 |
+
},
|
| 7252 |
+
{
|
| 7253 |
+
"epoch": 0.6741793238608526,
|
| 7254 |
+
"grad_norm": Infinity,
|
| 7255 |
+
"learning_rate": 5.086031721012998e-05,
|
| 7256 |
+
"loss": 1.5053,
|
| 7257 |
+
"step": 1032
|
| 7258 |
+
},
|
| 7259 |
+
{
|
| 7260 |
+
"epoch": 0.6748325983994774,
|
| 7261 |
+
"grad_norm": Infinity,
|
| 7262 |
+
"learning_rate": 5.067555471273957e-05,
|
| 7263 |
+
"loss": 1.5942,
|
| 7264 |
+
"step": 1033
|
| 7265 |
+
},
|
| 7266 |
+
{
|
| 7267 |
+
"epoch": 0.6754858729381022,
|
| 7268 |
+
"grad_norm": Infinity,
|
| 7269 |
+
"learning_rate": 5.049101446360498e-05,
|
| 7270 |
+
"loss": 1.5184,
|
| 7271 |
+
"step": 1034
|
| 7272 |
+
},
|
| 7273 |
+
{
|
| 7274 |
+
"epoch": 0.6761391474767271,
|
| 7275 |
+
"grad_norm": Infinity,
|
| 7276 |
+
"learning_rate": 5.0306697294235714e-05,
|
| 7277 |
+
"loss": 1.6699,
|
| 7278 |
+
"step": 1035
|
| 7279 |
+
},
|
| 7280 |
+
{
|
| 7281 |
+
"epoch": 0.676792422015352,
|
| 7282 |
+
"grad_norm": Infinity,
|
| 7283 |
+
"learning_rate": 5.01226040351363e-05,
|
| 7284 |
+
"loss": 1.7125,
|
| 7285 |
+
"step": 1036
|
| 7286 |
+
},
|
| 7287 |
+
{
|
| 7288 |
+
"epoch": 0.6774456965539768,
|
| 7289 |
+
"grad_norm": Infinity,
|
| 7290 |
+
"learning_rate": 4.99387355158021e-05,
|
| 7291 |
+
"loss": 1.8265,
|
| 7292 |
+
"step": 1037
|
| 7293 |
+
},
|
| 7294 |
+
{
|
| 7295 |
+
"epoch": 0.6780989710926016,
|
| 7296 |
+
"grad_norm": Infinity,
|
| 7297 |
+
"learning_rate": 4.97550925647161e-05,
|
| 7298 |
+
"loss": 1.8673,
|
| 7299 |
+
"step": 1038
|
| 7300 |
+
},
|
| 7301 |
+
{
|
| 7302 |
+
"epoch": 0.6787522456312265,
|
| 7303 |
+
"grad_norm": Infinity,
|
| 7304 |
+
"learning_rate": 4.957167600934474e-05,
|
| 7305 |
+
"loss": 1.935,
|
| 7306 |
+
"step": 1039
|
| 7307 |
+
},
|
| 7308 |
+
{
|
| 7309 |
+
"epoch": 0.6794055201698513,
|
| 7310 |
+
"grad_norm": Infinity,
|
| 7311 |
+
"learning_rate": 4.938848667613436e-05,
|
| 7312 |
+
"loss": 2.0304,
|
| 7313 |
+
"step": 1040
|
| 7314 |
+
},
|
| 7315 |
+
{
|
| 7316 |
+
"epoch": 0.6800587947084762,
|
| 7317 |
+
"grad_norm": Infinity,
|
| 7318 |
+
"learning_rate": 4.9205525390507644e-05,
|
| 7319 |
+
"loss": 1.9897,
|
| 7320 |
+
"step": 1041
|
| 7321 |
+
},
|
| 7322 |
+
{
|
| 7323 |
+
"epoch": 0.6807120692471011,
|
| 7324 |
+
"grad_norm": Infinity,
|
| 7325 |
+
"learning_rate": 4.9022792976859455e-05,
|
| 7326 |
+
"loss": 2.2204,
|
| 7327 |
+
"step": 1042
|
| 7328 |
+
},
|
| 7329 |
+
{
|
| 7330 |
+
"epoch": 0.681365343785726,
|
| 7331 |
+
"grad_norm": Infinity,
|
| 7332 |
+
"learning_rate": 4.884029025855364e-05,
|
| 7333 |
+
"loss": 2.268,
|
| 7334 |
+
"step": 1043
|
| 7335 |
+
},
|
| 7336 |
+
{
|
| 7337 |
+
"epoch": 0.6820186183243508,
|
| 7338 |
+
"grad_norm": Infinity,
|
| 7339 |
+
"learning_rate": 4.8658018057918964e-05,
|
| 7340 |
+
"loss": 2.5273,
|
| 7341 |
+
"step": 1044
|
| 7342 |
+
},
|
| 7343 |
+
{
|
| 7344 |
+
"epoch": 0.6826718928629757,
|
| 7345 |
+
"grad_norm": Infinity,
|
| 7346 |
+
"learning_rate": 4.8475977196245504e-05,
|
| 7347 |
+
"loss": 2.3542,
|
| 7348 |
+
"step": 1045
|
| 7349 |
+
},
|
| 7350 |
+
{
|
| 7351 |
+
"epoch": 0.6833251674016005,
|
| 7352 |
+
"grad_norm": Infinity,
|
| 7353 |
+
"learning_rate": 4.8294168493780955e-05,
|
| 7354 |
+
"loss": 2.7903,
|
| 7355 |
+
"step": 1046
|
| 7356 |
+
},
|
| 7357 |
+
{
|
| 7358 |
+
"epoch": 0.6839784419402254,
|
| 7359 |
+
"grad_norm": Infinity,
|
| 7360 |
+
"learning_rate": 4.8112592769727085e-05,
|
| 7361 |
+
"loss": 2.9665,
|
| 7362 |
+
"step": 1047
|
| 7363 |
+
},
|
| 7364 |
+
{
|
| 7365 |
+
"epoch": 0.6846317164788502,
|
| 7366 |
+
"grad_norm": Infinity,
|
| 7367 |
+
"learning_rate": 4.793125084223562e-05,
|
| 7368 |
+
"loss": 3.3895,
|
| 7369 |
+
"step": 1048
|
| 7370 |
+
},
|
| 7371 |
+
{
|
| 7372 |
+
"epoch": 0.6852849910174751,
|
| 7373 |
+
"grad_norm": Infinity,
|
| 7374 |
+
"learning_rate": 4.7750143528405126e-05,
|
| 7375 |
+
"loss": 3.2519,
|
| 7376 |
+
"step": 1049
|
| 7377 |
+
},
|
| 7378 |
+
{
|
| 7379 |
+
"epoch": 0.6859382655560999,
|
| 7380 |
+
"grad_norm": Infinity,
|
| 7381 |
+
"learning_rate": 4.756927164427685e-05,
|
| 7382 |
+
"loss": 4.7098,
|
| 7383 |
+
"step": 1050
|
| 7384 |
+
},
|
| 7385 |
+
{
|
| 7386 |
+
"epoch": 0.6865915400947248,
|
| 7387 |
+
"grad_norm": Infinity,
|
| 7388 |
+
"learning_rate": 4.738863600483125e-05,
|
| 7389 |
+
"loss": 1.2615,
|
| 7390 |
+
"step": 1051
|
| 7391 |
+
},
|
| 7392 |
+
{
|
| 7393 |
+
"epoch": 0.6872448146333496,
|
| 7394 |
+
"grad_norm": Infinity,
|
| 7395 |
+
"learning_rate": 4.720823742398447e-05,
|
| 7396 |
+
"loss": 1.3003,
|
| 7397 |
+
"step": 1052
|
| 7398 |
+
},
|
| 7399 |
+
{
|
| 7400 |
+
"epoch": 0.6878980891719745,
|
| 7401 |
+
"grad_norm": Infinity,
|
| 7402 |
+
"learning_rate": 4.70280767145842e-05,
|
| 7403 |
+
"loss": 1.46,
|
| 7404 |
+
"step": 1053
|
| 7405 |
+
},
|
| 7406 |
+
{
|
| 7407 |
+
"epoch": 0.6885513637105993,
|
| 7408 |
+
"grad_norm": Infinity,
|
| 7409 |
+
"learning_rate": 4.684815468840662e-05,
|
| 7410 |
+
"loss": 1.4526,
|
| 7411 |
+
"step": 1054
|
| 7412 |
+
},
|
| 7413 |
+
{
|
| 7414 |
+
"epoch": 0.6892046382492243,
|
| 7415 |
+
"grad_norm": Infinity,
|
| 7416 |
+
"learning_rate": 4.666847215615226e-05,
|
| 7417 |
+
"loss": 1.4448,
|
| 7418 |
+
"step": 1055
|
| 7419 |
+
},
|
| 7420 |
+
{
|
| 7421 |
+
"epoch": 0.6898579127878491,
|
| 7422 |
+
"grad_norm": Infinity,
|
| 7423 |
+
"learning_rate": 4.648902992744255e-05,
|
| 7424 |
+
"loss": 1.4982,
|
| 7425 |
+
"step": 1056
|
| 7426 |
+
},
|
| 7427 |
+
{
|
| 7428 |
+
"epoch": 0.690511187326474,
|
| 7429 |
+
"grad_norm": Infinity,
|
| 7430 |
+
"learning_rate": 4.630982881081629e-05,
|
| 7431 |
+
"loss": 1.5638,
|
| 7432 |
+
"step": 1057
|
| 7433 |
+
},
|
| 7434 |
+
{
|
| 7435 |
+
"epoch": 0.6911644618650988,
|
| 7436 |
+
"grad_norm": Infinity,
|
| 7437 |
+
"learning_rate": 4.6130869613725614e-05,
|
| 7438 |
+
"loss": 1.5881,
|
| 7439 |
+
"step": 1058
|
| 7440 |
+
},
|
| 7441 |
+
{
|
| 7442 |
+
"epoch": 0.6918177364037237,
|
| 7443 |
+
"grad_norm": Infinity,
|
| 7444 |
+
"learning_rate": 4.595215314253285e-05,
|
| 7445 |
+
"loss": 1.6855,
|
| 7446 |
+
"step": 1059
|
| 7447 |
+
},
|
| 7448 |
+
{
|
| 7449 |
+
"epoch": 0.6924710109423485,
|
| 7450 |
+
"grad_norm": Infinity,
|
| 7451 |
+
"learning_rate": 4.57736802025065e-05,
|
| 7452 |
+
"loss": 1.7135,
|
| 7453 |
+
"step": 1060
|
| 7454 |
+
},
|
| 7455 |
+
{
|
| 7456 |
+
"epoch": 0.6931242854809734,
|
| 7457 |
+
"grad_norm": Infinity,
|
| 7458 |
+
"learning_rate": 4.5595451597817795e-05,
|
| 7459 |
+
"loss": 1.6494,
|
| 7460 |
+
"step": 1061
|
| 7461 |
+
},
|
| 7462 |
+
{
|
| 7463 |
+
"epoch": 0.6937775600195982,
|
| 7464 |
+
"grad_norm": Infinity,
|
| 7465 |
+
"learning_rate": 4.5417468131536975e-05,
|
| 7466 |
+
"loss": 1.7395,
|
| 7467 |
+
"step": 1062
|
| 7468 |
+
},
|
| 7469 |
+
{
|
| 7470 |
+
"epoch": 0.6944308345582231,
|
| 7471 |
+
"grad_norm": Infinity,
|
| 7472 |
+
"learning_rate": 4.52397306056299e-05,
|
| 7473 |
+
"loss": 1.8408,
|
| 7474 |
+
"step": 1063
|
| 7475 |
+
},
|
| 7476 |
+
{
|
| 7477 |
+
"epoch": 0.6950841090968479,
|
| 7478 |
+
"grad_norm": Infinity,
|
| 7479 |
+
"learning_rate": 4.5062239820953986e-05,
|
| 7480 |
+
"loss": 1.9947,
|
| 7481 |
+
"step": 1064
|
| 7482 |
+
},
|
| 7483 |
+
{
|
| 7484 |
+
"epoch": 0.6957373836354728,
|
| 7485 |
+
"grad_norm": Infinity,
|
| 7486 |
+
"learning_rate": 4.488499657725511e-05,
|
| 7487 |
+
"loss": 2.0566,
|
| 7488 |
+
"step": 1065
|
| 7489 |
+
},
|
| 7490 |
+
{
|
| 7491 |
+
"epoch": 0.6963906581740976,
|
| 7492 |
+
"grad_norm": Infinity,
|
| 7493 |
+
"learning_rate": 4.470800167316367e-05,
|
| 7494 |
+
"loss": 2.0069,
|
| 7495 |
+
"step": 1066
|
| 7496 |
+
},
|
| 7497 |
+
{
|
| 7498 |
+
"epoch": 0.6970439327127225,
|
| 7499 |
+
"grad_norm": Infinity,
|
| 7500 |
+
"learning_rate": 4.453125590619104e-05,
|
| 7501 |
+
"loss": 2.2135,
|
| 7502 |
+
"step": 1067
|
| 7503 |
+
},
|
| 7504 |
+
{
|
| 7505 |
+
"epoch": 0.6976972072513474,
|
| 7506 |
+
"grad_norm": Infinity,
|
| 7507 |
+
"learning_rate": 4.4354760072726176e-05,
|
| 7508 |
+
"loss": 2.2858,
|
| 7509 |
+
"step": 1068
|
| 7510 |
+
},
|
| 7511 |
+
{
|
| 7512 |
+
"epoch": 0.6983504817899723,
|
| 7513 |
+
"grad_norm": Infinity,
|
| 7514 |
+
"learning_rate": 4.417851496803164e-05,
|
| 7515 |
+
"loss": 2.7049,
|
| 7516 |
+
"step": 1069
|
| 7517 |
+
},
|
| 7518 |
+
{
|
| 7519 |
+
"epoch": 0.6990037563285971,
|
| 7520 |
+
"grad_norm": Infinity,
|
| 7521 |
+
"learning_rate": 4.4002521386240466e-05,
|
| 7522 |
+
"loss": 2.6026,
|
| 7523 |
+
"step": 1070
|
| 7524 |
+
},
|
| 7525 |
+
{
|
| 7526 |
+
"epoch": 0.699657030867222,
|
| 7527 |
+
"grad_norm": Infinity,
|
| 7528 |
+
"learning_rate": 4.382678012035227e-05,
|
| 7529 |
+
"loss": 2.8106,
|
| 7530 |
+
"step": 1071
|
| 7531 |
+
},
|
| 7532 |
+
{
|
| 7533 |
+
"epoch": 0.7003103054058468,
|
| 7534 |
+
"grad_norm": Infinity,
|
| 7535 |
+
"learning_rate": 4.365129196222972e-05,
|
| 7536 |
+
"loss": 2.8605,
|
| 7537 |
+
"step": 1072
|
| 7538 |
+
},
|
| 7539 |
+
{
|
| 7540 |
+
"epoch": 0.7009635799444717,
|
| 7541 |
+
"grad_norm": Infinity,
|
| 7542 |
+
"learning_rate": 4.347605770259522e-05,
|
| 7543 |
+
"loss": 3.2817,
|
| 7544 |
+
"step": 1073
|
| 7545 |
+
},
|
| 7546 |
+
{
|
| 7547 |
+
"epoch": 0.7016168544830965,
|
| 7548 |
+
"grad_norm": Infinity,
|
| 7549 |
+
"learning_rate": 4.3301078131026826e-05,
|
| 7550 |
+
"loss": 3.3869,
|
| 7551 |
+
"step": 1074
|
| 7552 |
+
},
|
| 7553 |
+
{
|
| 7554 |
+
"epoch": 0.7022701290217214,
|
| 7555 |
+
"grad_norm": Infinity,
|
| 7556 |
+
"learning_rate": 4.312635403595532e-05,
|
| 7557 |
+
"loss": 4.4504,
|
| 7558 |
+
"step": 1075
|
| 7559 |
+
},
|
| 7560 |
+
{
|
| 7561 |
+
"epoch": 0.7029234035603462,
|
| 7562 |
+
"grad_norm": Infinity,
|
| 7563 |
+
"learning_rate": 4.2951886204660175e-05,
|
| 7564 |
+
"loss": 1.2931,
|
| 7565 |
+
"step": 1076
|
| 7566 |
+
},
|
| 7567 |
+
{
|
| 7568 |
+
"epoch": 0.7035766780989711,
|
| 7569 |
+
"grad_norm": Infinity,
|
| 7570 |
+
"learning_rate": 4.2777675423266227e-05,
|
| 7571 |
+
"loss": 1.3579,
|
| 7572 |
+
"step": 1077
|
| 7573 |
+
},
|
| 7574 |
+
{
|
| 7575 |
+
"epoch": 0.7042299526375959,
|
| 7576 |
+
"grad_norm": Infinity,
|
| 7577 |
+
"learning_rate": 4.260372247674004e-05,
|
| 7578 |
+
"loss": 1.4018,
|
| 7579 |
+
"step": 1078
|
| 7580 |
+
},
|
| 7581 |
+
{
|
| 7582 |
+
"epoch": 0.7048832271762208,
|
| 7583 |
+
"grad_norm": Infinity,
|
| 7584 |
+
"learning_rate": 4.243002814888656e-05,
|
| 7585 |
+
"loss": 1.415,
|
| 7586 |
+
"step": 1079
|
| 7587 |
+
},
|
| 7588 |
+
{
|
| 7589 |
+
"epoch": 0.7055365017148456,
|
| 7590 |
+
"grad_norm": Infinity,
|
| 7591 |
+
"learning_rate": 4.2256593222345185e-05,
|
| 7592 |
+
"loss": 1.4629,
|
| 7593 |
+
"step": 1080
|
| 7594 |
+
},
|
| 7595 |
+
{
|
| 7596 |
+
"epoch": 0.7061897762534706,
|
| 7597 |
+
"grad_norm": Infinity,
|
| 7598 |
+
"learning_rate": 4.2083418478586755e-05,
|
| 7599 |
+
"loss": 1.478,
|
| 7600 |
+
"step": 1081
|
| 7601 |
+
},
|
| 7602 |
+
{
|
| 7603 |
+
"epoch": 0.7068430507920954,
|
| 7604 |
+
"grad_norm": Infinity,
|
| 7605 |
+
"learning_rate": 4.1910504697909614e-05,
|
| 7606 |
+
"loss": 1.5336,
|
| 7607 |
+
"step": 1082
|
| 7608 |
+
},
|
| 7609 |
+
{
|
| 7610 |
+
"epoch": 0.7074963253307203,
|
| 7611 |
+
"grad_norm": Infinity,
|
| 7612 |
+
"learning_rate": 4.173785265943624e-05,
|
| 7613 |
+
"loss": 1.5161,
|
| 7614 |
+
"step": 1083
|
| 7615 |
+
},
|
| 7616 |
+
{
|
| 7617 |
+
"epoch": 0.7081495998693451,
|
| 7618 |
+
"grad_norm": Infinity,
|
| 7619 |
+
"learning_rate": 4.1565463141109894e-05,
|
| 7620 |
+
"loss": 1.5435,
|
| 7621 |
+
"step": 1084
|
| 7622 |
+
},
|
| 7623 |
+
{
|
| 7624 |
+
"epoch": 0.70880287440797,
|
| 7625 |
+
"grad_norm": Infinity,
|
| 7626 |
+
"learning_rate": 4.139333691969071e-05,
|
| 7627 |
+
"loss": 1.7314,
|
| 7628 |
+
"step": 1085
|
| 7629 |
+
},
|
| 7630 |
+
{
|
| 7631 |
+
"epoch": 0.7094561489465948,
|
| 7632 |
+
"grad_norm": Infinity,
|
| 7633 |
+
"learning_rate": 4.12214747707527e-05,
|
| 7634 |
+
"loss": 1.6485,
|
| 7635 |
+
"step": 1086
|
| 7636 |
+
},
|
| 7637 |
+
{
|
| 7638 |
+
"epoch": 0.7101094234852197,
|
| 7639 |
+
"grad_norm": Infinity,
|
| 7640 |
+
"learning_rate": 4.1049877468679856e-05,
|
| 7641 |
+
"loss": 1.8288,
|
| 7642 |
+
"step": 1087
|
| 7643 |
+
},
|
| 7644 |
+
{
|
| 7645 |
+
"epoch": 0.7107626980238445,
|
| 7646 |
+
"grad_norm": Infinity,
|
| 7647 |
+
"learning_rate": 4.087854578666282e-05,
|
| 7648 |
+
"loss": 1.945,
|
| 7649 |
+
"step": 1088
|
| 7650 |
+
},
|
| 7651 |
+
{
|
| 7652 |
+
"epoch": 0.7114159725624694,
|
| 7653 |
+
"grad_norm": Infinity,
|
| 7654 |
+
"learning_rate": 4.0707480496695514e-05,
|
| 7655 |
+
"loss": 2.0181,
|
| 7656 |
+
"step": 1089
|
| 7657 |
+
},
|
| 7658 |
+
{
|
| 7659 |
+
"epoch": 0.7120692471010942,
|
| 7660 |
+
"grad_norm": Infinity,
|
| 7661 |
+
"learning_rate": 4.053668236957134e-05,
|
| 7662 |
+
"loss": 2.1645,
|
| 7663 |
+
"step": 1090
|
| 7664 |
+
},
|
| 7665 |
+
{
|
| 7666 |
+
"epoch": 0.712722521639719,
|
| 7667 |
+
"grad_norm": Infinity,
|
| 7668 |
+
"learning_rate": 4.036615217488009e-05,
|
| 7669 |
+
"loss": 2.2705,
|
| 7670 |
+
"step": 1091
|
| 7671 |
+
},
|
| 7672 |
+
{
|
| 7673 |
+
"epoch": 0.7133757961783439,
|
| 7674 |
+
"grad_norm": Infinity,
|
| 7675 |
+
"learning_rate": 4.01958906810042e-05,
|
| 7676 |
+
"loss": 2.2283,
|
| 7677 |
+
"step": 1092
|
| 7678 |
+
},
|
| 7679 |
+
{
|
| 7680 |
+
"epoch": 0.7140290707169689,
|
| 7681 |
+
"grad_norm": Infinity,
|
| 7682 |
+
"learning_rate": 4.0025898655115394e-05,
|
| 7683 |
+
"loss": 2.4277,
|
| 7684 |
+
"step": 1093
|
| 7685 |
+
},
|
| 7686 |
+
{
|
| 7687 |
+
"epoch": 0.7146823452555937,
|
| 7688 |
+
"grad_norm": Infinity,
|
| 7689 |
+
"learning_rate": 3.985617686317118e-05,
|
| 7690 |
+
"loss": 2.6549,
|
| 7691 |
+
"step": 1094
|
| 7692 |
+
},
|
| 7693 |
+
{
|
| 7694 |
+
"epoch": 0.7153356197942186,
|
| 7695 |
+
"grad_norm": Infinity,
|
| 7696 |
+
"learning_rate": 3.96867260699116e-05,
|
| 7697 |
+
"loss": 2.3398,
|
| 7698 |
+
"step": 1095
|
| 7699 |
+
},
|
| 7700 |
+
{
|
| 7701 |
+
"epoch": 0.7159888943328434,
|
| 7702 |
+
"grad_norm": Infinity,
|
| 7703 |
+
"learning_rate": 3.951754703885533e-05,
|
| 7704 |
+
"loss": 2.6731,
|
| 7705 |
+
"step": 1096
|
| 7706 |
+
},
|
| 7707 |
+
{
|
| 7708 |
+
"epoch": 0.7166421688714683,
|
| 7709 |
+
"grad_norm": Infinity,
|
| 7710 |
+
"learning_rate": 3.934864053229681e-05,
|
| 7711 |
+
"loss": 2.6785,
|
| 7712 |
+
"step": 1097
|
| 7713 |
+
},
|
| 7714 |
+
{
|
| 7715 |
+
"epoch": 0.7172954434100931,
|
| 7716 |
+
"grad_norm": Infinity,
|
| 7717 |
+
"learning_rate": 3.918000731130238e-05,
|
| 7718 |
+
"loss": 3.2614,
|
| 7719 |
+
"step": 1098
|
| 7720 |
+
},
|
| 7721 |
+
{
|
| 7722 |
+
"epoch": 0.717948717948718,
|
| 7723 |
+
"grad_norm": Infinity,
|
| 7724 |
+
"learning_rate": 3.9011648135706966e-05,
|
| 7725 |
+
"loss": 3.9284,
|
| 7726 |
+
"step": 1099
|
| 7727 |
+
},
|
| 7728 |
+
{
|
| 7729 |
+
"epoch": 0.7186019924873428,
|
| 7730 |
+
"grad_norm": Infinity,
|
| 7731 |
+
"learning_rate": 3.884356376411089e-05,
|
| 7732 |
+
"loss": 4.2049,
|
| 7733 |
+
"step": 1100
|
| 7734 |
+
},
|
| 7735 |
+
{
|
| 7736 |
+
"epoch": 0.7192552670259676,
|
| 7737 |
+
"grad_norm": Infinity,
|
| 7738 |
+
"learning_rate": 3.8675754953875956e-05,
|
| 7739 |
+
"loss": 1.3125,
|
| 7740 |
+
"step": 1101
|
| 7741 |
+
},
|
| 7742 |
+
{
|
| 7743 |
+
"epoch": 0.7199085415645925,
|
| 7744 |
+
"grad_norm": Infinity,
|
| 7745 |
+
"learning_rate": 3.8508222461122567e-05,
|
| 7746 |
+
"loss": 1.3905,
|
| 7747 |
+
"step": 1102
|
| 7748 |
+
},
|
| 7749 |
+
{
|
| 7750 |
+
"epoch": 0.7205618161032173,
|
| 7751 |
+
"grad_norm": Infinity,
|
| 7752 |
+
"learning_rate": 3.8340967040725995e-05,
|
| 7753 |
+
"loss": 1.4399,
|
| 7754 |
+
"step": 1103
|
| 7755 |
+
},
|
| 7756 |
+
{
|
| 7757 |
+
"epoch": 0.7212150906418422,
|
| 7758 |
+
"grad_norm": Infinity,
|
| 7759 |
+
"learning_rate": 3.817398944631302e-05,
|
| 7760 |
+
"loss": 1.4085,
|
| 7761 |
+
"step": 1104
|
| 7762 |
+
},
|
| 7763 |
+
{
|
| 7764 |
+
"epoch": 0.721868365180467,
|
| 7765 |
+
"grad_norm": Infinity,
|
| 7766 |
+
"learning_rate": 3.800729043025871e-05,
|
| 7767 |
+
"loss": 1.463,
|
| 7768 |
+
"step": 1105
|
| 7769 |
+
},
|
| 7770 |
+
{
|
| 7771 |
+
"epoch": 0.722521639719092,
|
| 7772 |
+
"grad_norm": Infinity,
|
| 7773 |
+
"learning_rate": 3.784087074368279e-05,
|
| 7774 |
+
"loss": 1.4755,
|
| 7775 |
+
"step": 1106
|
| 7776 |
+
},
|
| 7777 |
+
{
|
| 7778 |
+
"epoch": 0.7231749142577168,
|
| 7779 |
+
"grad_norm": Infinity,
|
| 7780 |
+
"learning_rate": 3.767473113644641e-05,
|
| 7781 |
+
"loss": 1.5932,
|
| 7782 |
+
"step": 1107
|
| 7783 |
+
},
|
| 7784 |
+
{
|
| 7785 |
+
"epoch": 0.7238281887963417,
|
| 7786 |
+
"grad_norm": Infinity,
|
| 7787 |
+
"learning_rate": 3.750887235714872e-05,
|
| 7788 |
+
"loss": 1.6179,
|
| 7789 |
+
"step": 1108
|
| 7790 |
+
},
|
| 7791 |
+
{
|
| 7792 |
+
"epoch": 0.7244814633349665,
|
| 7793 |
+
"grad_norm": Infinity,
|
| 7794 |
+
"learning_rate": 3.734329515312349e-05,
|
| 7795 |
+
"loss": 1.5271,
|
| 7796 |
+
"step": 1109
|
| 7797 |
+
},
|
| 7798 |
+
{
|
| 7799 |
+
"epoch": 0.7251347378735914,
|
| 7800 |
+
"grad_norm": Infinity,
|
| 7801 |
+
"learning_rate": 3.717800027043576e-05,
|
| 7802 |
+
"loss": 1.5595,
|
| 7803 |
+
"step": 1110
|
| 7804 |
+
},
|
| 7805 |
+
{
|
| 7806 |
+
"epoch": 0.7257880124122162,
|
| 7807 |
+
"grad_norm": Infinity,
|
| 7808 |
+
"learning_rate": 3.7012988453878586e-05,
|
| 7809 |
+
"loss": 1.6898,
|
| 7810 |
+
"step": 1111
|
| 7811 |
+
},
|
| 7812 |
+
{
|
| 7813 |
+
"epoch": 0.7264412869508411,
|
| 7814 |
+
"grad_norm": Infinity,
|
| 7815 |
+
"learning_rate": 3.6848260446969306e-05,
|
| 7816 |
+
"loss": 1.7962,
|
| 7817 |
+
"step": 1112
|
| 7818 |
+
},
|
| 7819 |
+
{
|
| 7820 |
+
"epoch": 0.7270945614894659,
|
| 7821 |
+
"grad_norm": Infinity,
|
| 7822 |
+
"learning_rate": 3.668381699194676e-05,
|
| 7823 |
+
"loss": 1.7829,
|
| 7824 |
+
"step": 1113
|
| 7825 |
+
},
|
| 7826 |
+
{
|
| 7827 |
+
"epoch": 0.7277478360280908,
|
| 7828 |
+
"grad_norm": Infinity,
|
| 7829 |
+
"learning_rate": 3.6519658829767455e-05,
|
| 7830 |
+
"loss": 1.874,
|
| 7831 |
+
"step": 1114
|
| 7832 |
+
},
|
| 7833 |
+
{
|
| 7834 |
+
"epoch": 0.7284011105667156,
|
| 7835 |
+
"grad_norm": Infinity,
|
| 7836 |
+
"learning_rate": 3.635578670010242e-05,
|
| 7837 |
+
"loss": 1.9904,
|
| 7838 |
+
"step": 1115
|
| 7839 |
+
},
|
| 7840 |
+
{
|
| 7841 |
+
"epoch": 0.7290543851053405,
|
| 7842 |
+
"grad_norm": Infinity,
|
| 7843 |
+
"learning_rate": 3.619220134133402e-05,
|
| 7844 |
+
"loss": 1.9767,
|
| 7845 |
+
"step": 1116
|
| 7846 |
+
},
|
| 7847 |
+
{
|
| 7848 |
+
"epoch": 0.7297076596439653,
|
| 7849 |
+
"grad_norm": Infinity,
|
| 7850 |
+
"learning_rate": 3.602890349055221e-05,
|
| 7851 |
+
"loss": 2.2286,
|
| 7852 |
+
"step": 1117
|
| 7853 |
+
},
|
| 7854 |
+
{
|
| 7855 |
+
"epoch": 0.7303609341825902,
|
| 7856 |
+
"grad_norm": Infinity,
|
| 7857 |
+
"learning_rate": 3.586589388355176e-05,
|
| 7858 |
+
"loss": 2.4116,
|
| 7859 |
+
"step": 1118
|
| 7860 |
+
},
|
| 7861 |
+
{
|
| 7862 |
+
"epoch": 0.7310142087212151,
|
| 7863 |
+
"grad_norm": Infinity,
|
| 7864 |
+
"learning_rate": 3.570317325482847e-05,
|
| 7865 |
+
"loss": 2.5585,
|
| 7866 |
+
"step": 1119
|
| 7867 |
+
},
|
| 7868 |
+
{
|
| 7869 |
+
"epoch": 0.73166748325984,
|
| 7870 |
+
"grad_norm": Infinity,
|
| 7871 |
+
"learning_rate": 3.554074233757608e-05,
|
| 7872 |
+
"loss": 2.3875,
|
| 7873 |
+
"step": 1120
|
| 7874 |
+
},
|
| 7875 |
+
{
|
| 7876 |
+
"epoch": 0.7323207577984648,
|
| 7877 |
+
"grad_norm": Infinity,
|
| 7878 |
+
"learning_rate": 3.537860186368305e-05,
|
| 7879 |
+
"loss": 2.9333,
|
| 7880 |
+
"step": 1121
|
| 7881 |
+
},
|
| 7882 |
+
{
|
| 7883 |
+
"epoch": 0.7329740323370897,
|
| 7884 |
+
"grad_norm": Infinity,
|
| 7885 |
+
"learning_rate": 3.5216752563729e-05,
|
| 7886 |
+
"loss": 3.1705,
|
| 7887 |
+
"step": 1122
|
| 7888 |
+
},
|
| 7889 |
+
{
|
| 7890 |
+
"epoch": 0.7336273068757145,
|
| 7891 |
+
"grad_norm": Infinity,
|
| 7892 |
+
"learning_rate": 3.5055195166981645e-05,
|
| 7893 |
+
"loss": 3.3577,
|
| 7894 |
+
"step": 1123
|
| 7895 |
+
},
|
| 7896 |
+
{
|
| 7897 |
+
"epoch": 0.7342805814143394,
|
| 7898 |
+
"grad_norm": Infinity,
|
| 7899 |
+
"learning_rate": 3.4893930401393406e-05,
|
| 7900 |
+
"loss": 3.4905,
|
| 7901 |
+
"step": 1124
|
| 7902 |
+
},
|
| 7903 |
+
{
|
| 7904 |
+
"epoch": 0.7349338559529642,
|
| 7905 |
+
"grad_norm": Infinity,
|
| 7906 |
+
"learning_rate": 3.4732958993598154e-05,
|
| 7907 |
+
"loss": 4.1618,
|
| 7908 |
+
"step": 1125
|
| 7909 |
+
},
|
| 7910 |
+
{
|
| 7911 |
+
"epoch": 0.7355871304915891,
|
| 7912 |
+
"grad_norm": Infinity,
|
| 7913 |
+
"learning_rate": 3.457228166890791e-05,
|
| 7914 |
+
"loss": 1.2767,
|
| 7915 |
+
"step": 1126
|
| 7916 |
+
},
|
| 7917 |
+
{
|
| 7918 |
+
"epoch": 0.7362404050302139,
|
| 7919 |
+
"grad_norm": Infinity,
|
| 7920 |
+
"learning_rate": 3.441189915130974e-05,
|
| 7921 |
+
"loss": 1.3949,
|
| 7922 |
+
"step": 1127
|
| 7923 |
+
},
|
| 7924 |
+
{
|
| 7925 |
+
"epoch": 0.7368936795688388,
|
| 7926 |
+
"grad_norm": Infinity,
|
| 7927 |
+
"learning_rate": 3.425181216346213e-05,
|
| 7928 |
+
"loss": 1.4078,
|
| 7929 |
+
"step": 1128
|
| 7930 |
+
},
|
| 7931 |
+
{
|
| 7932 |
+
"epoch": 0.7375469541074636,
|
| 7933 |
+
"grad_norm": Infinity,
|
| 7934 |
+
"learning_rate": 3.409202142669213e-05,
|
| 7935 |
+
"loss": 1.4977,
|
| 7936 |
+
"step": 1129
|
| 7937 |
+
},
|
| 7938 |
+
{
|
| 7939 |
+
"epoch": 0.7382002286460885,
|
| 7940 |
+
"grad_norm": Infinity,
|
| 7941 |
+
"learning_rate": 3.393252766099187e-05,
|
| 7942 |
+
"loss": 1.4244,
|
| 7943 |
+
"step": 1130
|
| 7944 |
+
},
|
| 7945 |
+
{
|
| 7946 |
+
"epoch": 0.7388535031847133,
|
| 7947 |
+
"grad_norm": Infinity,
|
| 7948 |
+
"learning_rate": 3.377333158501534e-05,
|
| 7949 |
+
"loss": 1.6322,
|
| 7950 |
+
"step": 1131
|
| 7951 |
+
},
|
| 7952 |
+
{
|
| 7953 |
+
"epoch": 0.7395067777233383,
|
| 7954 |
+
"grad_norm": Infinity,
|
| 7955 |
+
"learning_rate": 3.3614433916075326e-05,
|
| 7956 |
+
"loss": 1.5867,
|
| 7957 |
+
"step": 1132
|
| 7958 |
+
},
|
| 7959 |
+
{
|
| 7960 |
+
"epoch": 0.7401600522619631,
|
| 7961 |
+
"grad_norm": Infinity,
|
| 7962 |
+
"learning_rate": 3.34558353701398e-05,
|
| 7963 |
+
"loss": 1.539,
|
| 7964 |
+
"step": 1133
|
| 7965 |
+
},
|
| 7966 |
+
{
|
| 7967 |
+
"epoch": 0.740813326800588,
|
| 7968 |
+
"grad_norm": Infinity,
|
| 7969 |
+
"learning_rate": 3.3297536661829155e-05,
|
| 7970 |
+
"loss": 1.5412,
|
| 7971 |
+
"step": 1134
|
| 7972 |
+
},
|
| 7973 |
+
{
|
| 7974 |
+
"epoch": 0.7414666013392128,
|
| 7975 |
+
"grad_norm": Infinity,
|
| 7976 |
+
"learning_rate": 3.313953850441266e-05,
|
| 7977 |
+
"loss": 1.6732,
|
| 7978 |
+
"step": 1135
|
| 7979 |
+
},
|
| 7980 |
+
{
|
| 7981 |
+
"epoch": 0.7421198758778377,
|
| 7982 |
+
"grad_norm": Infinity,
|
| 7983 |
+
"learning_rate": 3.298184160980532e-05,
|
| 7984 |
+
"loss": 1.6536,
|
| 7985 |
+
"step": 1136
|
| 7986 |
+
},
|
| 7987 |
+
{
|
| 7988 |
+
"epoch": 0.7427731504164625,
|
| 7989 |
+
"grad_norm": Infinity,
|
| 7990 |
+
"learning_rate": 3.2824446688564815e-05,
|
| 7991 |
+
"loss": 1.7995,
|
| 7992 |
+
"step": 1137
|
| 7993 |
+
},
|
| 7994 |
+
{
|
| 7995 |
+
"epoch": 0.7434264249550874,
|
| 7996 |
+
"grad_norm": Infinity,
|
| 7997 |
+
"learning_rate": 3.266735444988808e-05,
|
| 7998 |
+
"loss": 1.7265,
|
| 7999 |
+
"step": 1138
|
| 8000 |
+
},
|
| 8001 |
+
{
|
| 8002 |
+
"epoch": 0.7440796994937122,
|
| 8003 |
+
"grad_norm": Infinity,
|
| 8004 |
+
"learning_rate": 3.251056560160821e-05,
|
| 8005 |
+
"loss": 1.9421,
|
| 8006 |
+
"step": 1139
|
| 8007 |
+
},
|
| 8008 |
+
{
|
| 8009 |
+
"epoch": 0.7447329740323371,
|
| 8010 |
+
"grad_norm": Infinity,
|
| 8011 |
+
"learning_rate": 3.2354080850191324e-05,
|
| 8012 |
+
"loss": 2.1221,
|
| 8013 |
+
"step": 1140
|
| 8014 |
+
},
|
| 8015 |
+
{
|
| 8016 |
+
"epoch": 0.7453862485709619,
|
| 8017 |
+
"grad_norm": Infinity,
|
| 8018 |
+
"learning_rate": 3.219790090073329e-05,
|
| 8019 |
+
"loss": 2.1755,
|
| 8020 |
+
"step": 1141
|
| 8021 |
+
},
|
| 8022 |
+
{
|
| 8023 |
+
"epoch": 0.7460395231095868,
|
| 8024 |
+
"grad_norm": Infinity,
|
| 8025 |
+
"learning_rate": 3.2042026456956555e-05,
|
| 8026 |
+
"loss": 2.3494,
|
| 8027 |
+
"step": 1142
|
| 8028 |
+
},
|
| 8029 |
+
{
|
| 8030 |
+
"epoch": 0.7466927976482116,
|
| 8031 |
+
"grad_norm": Infinity,
|
| 8032 |
+
"learning_rate": 3.188645822120716e-05,
|
| 8033 |
+
"loss": 2.5545,
|
| 8034 |
+
"step": 1143
|
| 8035 |
+
},
|
| 8036 |
+
{
|
| 8037 |
+
"epoch": 0.7473460721868365,
|
| 8038 |
+
"grad_norm": Infinity,
|
| 8039 |
+
"learning_rate": 3.1731196894451154e-05,
|
| 8040 |
+
"loss": 2.7081,
|
| 8041 |
+
"step": 1144
|
| 8042 |
+
},
|
| 8043 |
+
{
|
| 8044 |
+
"epoch": 0.7479993467254614,
|
| 8045 |
+
"grad_norm": Infinity,
|
| 8046 |
+
"learning_rate": 3.157624317627195e-05,
|
| 8047 |
+
"loss": 2.7034,
|
| 8048 |
+
"step": 1145
|
| 8049 |
+
},
|
| 8050 |
+
{
|
| 8051 |
+
"epoch": 0.7486526212640863,
|
| 8052 |
+
"grad_norm": Infinity,
|
| 8053 |
+
"learning_rate": 3.1421597764866864e-05,
|
| 8054 |
+
"loss": 2.9047,
|
| 8055 |
+
"step": 1146
|
| 8056 |
+
},
|
| 8057 |
+
{
|
| 8058 |
+
"epoch": 0.7493058958027111,
|
| 8059 |
+
"grad_norm": Infinity,
|
| 8060 |
+
"learning_rate": 3.126726135704389e-05,
|
| 8061 |
+
"loss": 2.915,
|
| 8062 |
+
"step": 1147
|
| 8063 |
+
},
|
| 8064 |
+
{
|
| 8065 |
+
"epoch": 0.749959170341336,
|
| 8066 |
+
"grad_norm": Infinity,
|
| 8067 |
+
"learning_rate": 3.1113234648218883e-05,
|
| 8068 |
+
"loss": 3.1268,
|
| 8069 |
+
"step": 1148
|
| 8070 |
+
},
|
| 8071 |
+
{
|
| 8072 |
+
"epoch": 0.7506124448799608,
|
| 8073 |
+
"grad_norm": Infinity,
|
| 8074 |
+
"learning_rate": 3.095951833241213e-05,
|
| 8075 |
+
"loss": 3.4791,
|
| 8076 |
+
"step": 1149
|
| 8077 |
+
},
|
| 8078 |
+
{
|
| 8079 |
+
"epoch": 0.7506124448799608,
|
| 8080 |
+
"eval_loss": 2.0990819931030273,
|
| 8081 |
+
"eval_runtime": 161.0223,
|
| 8082 |
+
"eval_samples_per_second": 16.01,
|
| 8083 |
+
"eval_steps_per_second": 4.006,
|
| 8084 |
+
"step": 1149
|
| 8085 |
}
|
| 8086 |
],
|
| 8087 |
"logging_steps": 1,
|
|
|
|
| 8101 |
"attributes": {}
|
| 8102 |
}
|
| 8103 |
},
|
| 8104 |
+
"total_flos": 4.730978876270838e+17,
|
| 8105 |
"train_batch_size": 2,
|
| 8106 |
"trial_name": null,
|
| 8107 |
"trial_params": null
|