Training in progress, step 1530, 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:fbf654fe0b4feb8ca20e63ac5ae6f5cf837d99aa2d5db0f429f7f6f79323982d
|
| 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 sha256:92715f67a640550d2d979f50ac3e29620507c040645efaac8e093ec98f77b00f
|
| 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 sha256:24678fed7bcc2f010b40a4ff7e6562e6ea1edb76805057b4f23f5ba162b6b70c
|
| 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:be4444afcc9f1dfd0b9fff2290771e76d51570802a901fda963f7c0bf832544e
|
| 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,
|
|
@@ -8082,6 +8082,2673 @@
|
|
| 8082 |
"eval_samples_per_second": 16.01,
|
| 8083 |
"eval_steps_per_second": 4.006,
|
| 8084 |
"step": 1149
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8085 |
}
|
| 8086 |
],
|
| 8087 |
"logging_steps": 1,
|
|
@@ -8096,12 +10763,12 @@
|
|
| 8096 |
"should_evaluate": false,
|
| 8097 |
"should_log": false,
|
| 8098 |
"should_save": true,
|
| 8099 |
-
"should_training_stop":
|
| 8100 |
},
|
| 8101 |
"attributes": {}
|
| 8102 |
}
|
| 8103 |
},
|
| 8104 |
-
"total_flos":
|
| 8105 |
"train_batch_size": 2,
|
| 8106 |
"trial_name": null,
|
| 8107 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9995100440960314,
|
| 5 |
"eval_steps": 383,
|
| 6 |
+
"global_step": 1530,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 8082 |
"eval_samples_per_second": 16.01,
|
| 8083 |
"eval_steps_per_second": 4.006,
|
| 8084 |
"step": 1149
|
| 8085 |
+
},
|
| 8086 |
+
{
|
| 8087 |
+
"epoch": 0.7512657194185857,
|
| 8088 |
+
"grad_norm": Infinity,
|
| 8089 |
+
"learning_rate": 3.080611310224539e-05,
|
| 8090 |
+
"loss": 4.3268,
|
| 8091 |
+
"step": 1150
|
| 8092 |
+
},
|
| 8093 |
+
{
|
| 8094 |
+
"epoch": 0.7519189939572105,
|
| 8095 |
+
"grad_norm": Infinity,
|
| 8096 |
+
"learning_rate": 3.0653019648938685e-05,
|
| 8097 |
+
"loss": 1.2606,
|
| 8098 |
+
"step": 1151
|
| 8099 |
+
},
|
| 8100 |
+
{
|
| 8101 |
+
"epoch": 0.7525722684958354,
|
| 8102 |
+
"grad_norm": Infinity,
|
| 8103 |
+
"learning_rate": 3.0500238662307212e-05,
|
| 8104 |
+
"loss": 1.3563,
|
| 8105 |
+
"step": 1152
|
| 8106 |
+
},
|
| 8107 |
+
{
|
| 8108 |
+
"epoch": 0.7532255430344602,
|
| 8109 |
+
"grad_norm": Infinity,
|
| 8110 |
+
"learning_rate": 3.0347770830758316e-05,
|
| 8111 |
+
"loss": 1.4777,
|
| 8112 |
+
"step": 1153
|
| 8113 |
+
},
|
| 8114 |
+
{
|
| 8115 |
+
"epoch": 0.7538788175730851,
|
| 8116 |
+
"grad_norm": Infinity,
|
| 8117 |
+
"learning_rate": 3.019561684128823e-05,
|
| 8118 |
+
"loss": 1.373,
|
| 8119 |
+
"step": 1154
|
| 8120 |
+
},
|
| 8121 |
+
{
|
| 8122 |
+
"epoch": 0.7545320921117099,
|
| 8123 |
+
"grad_norm": Infinity,
|
| 8124 |
+
"learning_rate": 3.0043777379479098e-05,
|
| 8125 |
+
"loss": 1.4508,
|
| 8126 |
+
"step": 1155
|
| 8127 |
+
},
|
| 8128 |
+
{
|
| 8129 |
+
"epoch": 0.7551853666503348,
|
| 8130 |
+
"grad_norm": Infinity,
|
| 8131 |
+
"learning_rate": 2.989225312949584e-05,
|
| 8132 |
+
"loss": 1.4653,
|
| 8133 |
+
"step": 1156
|
| 8134 |
+
},
|
| 8135 |
+
{
|
| 8136 |
+
"epoch": 0.7558386411889597,
|
| 8137 |
+
"grad_norm": Infinity,
|
| 8138 |
+
"learning_rate": 2.9741044774083094e-05,
|
| 8139 |
+
"loss": 1.5171,
|
| 8140 |
+
"step": 1157
|
| 8141 |
+
},
|
| 8142 |
+
{
|
| 8143 |
+
"epoch": 0.7564919157275846,
|
| 8144 |
+
"grad_norm": Infinity,
|
| 8145 |
+
"learning_rate": 2.9590152994562104e-05,
|
| 8146 |
+
"loss": 1.6195,
|
| 8147 |
+
"step": 1158
|
| 8148 |
+
},
|
| 8149 |
+
{
|
| 8150 |
+
"epoch": 0.7571451902662094,
|
| 8151 |
+
"grad_norm": Infinity,
|
| 8152 |
+
"learning_rate": 2.9439578470827755e-05,
|
| 8153 |
+
"loss": 1.5933,
|
| 8154 |
+
"step": 1159
|
| 8155 |
+
},
|
| 8156 |
+
{
|
| 8157 |
+
"epoch": 0.7577984648048343,
|
| 8158 |
+
"grad_norm": Infinity,
|
| 8159 |
+
"learning_rate": 2.9289321881345254e-05,
|
| 8160 |
+
"loss": 1.6688,
|
| 8161 |
+
"step": 1160
|
| 8162 |
+
},
|
| 8163 |
+
{
|
| 8164 |
+
"epoch": 0.7584517393434591,
|
| 8165 |
+
"grad_norm": Infinity,
|
| 8166 |
+
"learning_rate": 2.913938390314741e-05,
|
| 8167 |
+
"loss": 1.7732,
|
| 8168 |
+
"step": 1161
|
| 8169 |
+
},
|
| 8170 |
+
{
|
| 8171 |
+
"epoch": 0.759105013882084,
|
| 8172 |
+
"grad_norm": Infinity,
|
| 8173 |
+
"learning_rate": 2.8989765211831433e-05,
|
| 8174 |
+
"loss": 1.6398,
|
| 8175 |
+
"step": 1162
|
| 8176 |
+
},
|
| 8177 |
+
{
|
| 8178 |
+
"epoch": 0.7597582884207088,
|
| 8179 |
+
"grad_norm": Infinity,
|
| 8180 |
+
"learning_rate": 2.8840466481555672e-05,
|
| 8181 |
+
"loss": 1.8894,
|
| 8182 |
+
"step": 1163
|
| 8183 |
+
},
|
| 8184 |
+
{
|
| 8185 |
+
"epoch": 0.7604115629593337,
|
| 8186 |
+
"grad_norm": Infinity,
|
| 8187 |
+
"learning_rate": 2.8691488385037012e-05,
|
| 8188 |
+
"loss": 1.773,
|
| 8189 |
+
"step": 1164
|
| 8190 |
+
},
|
| 8191 |
+
{
|
| 8192 |
+
"epoch": 0.7610648374979585,
|
| 8193 |
+
"grad_norm": Infinity,
|
| 8194 |
+
"learning_rate": 2.854283159354748e-05,
|
| 8195 |
+
"loss": 1.9317,
|
| 8196 |
+
"step": 1165
|
| 8197 |
+
},
|
| 8198 |
+
{
|
| 8199 |
+
"epoch": 0.7617181120365834,
|
| 8200 |
+
"grad_norm": Infinity,
|
| 8201 |
+
"learning_rate": 2.8394496776911382e-05,
|
| 8202 |
+
"loss": 2.2086,
|
| 8203 |
+
"step": 1166
|
| 8204 |
+
},
|
| 8205 |
+
{
|
| 8206 |
+
"epoch": 0.7623713865752082,
|
| 8207 |
+
"grad_norm": Infinity,
|
| 8208 |
+
"learning_rate": 2.8246484603502275e-05,
|
| 8209 |
+
"loss": 2.3171,
|
| 8210 |
+
"step": 1167
|
| 8211 |
+
},
|
| 8212 |
+
{
|
| 8213 |
+
"epoch": 0.763024661113833,
|
| 8214 |
+
"grad_norm": Infinity,
|
| 8215 |
+
"learning_rate": 2.8098795740239923e-05,
|
| 8216 |
+
"loss": 2.2845,
|
| 8217 |
+
"step": 1168
|
| 8218 |
+
},
|
| 8219 |
+
{
|
| 8220 |
+
"epoch": 0.7636779356524579,
|
| 8221 |
+
"grad_norm": Infinity,
|
| 8222 |
+
"learning_rate": 2.7951430852587268e-05,
|
| 8223 |
+
"loss": 2.4874,
|
| 8224 |
+
"step": 1169
|
| 8225 |
+
},
|
| 8226 |
+
{
|
| 8227 |
+
"epoch": 0.7643312101910829,
|
| 8228 |
+
"grad_norm": Infinity,
|
| 8229 |
+
"learning_rate": 2.7804390604547557e-05,
|
| 8230 |
+
"loss": 2.7623,
|
| 8231 |
+
"step": 1170
|
| 8232 |
+
},
|
| 8233 |
+
{
|
| 8234 |
+
"epoch": 0.7649844847297077,
|
| 8235 |
+
"grad_norm": Infinity,
|
| 8236 |
+
"learning_rate": 2.7657675658661198e-05,
|
| 8237 |
+
"loss": 2.576,
|
| 8238 |
+
"step": 1171
|
| 8239 |
+
},
|
| 8240 |
+
{
|
| 8241 |
+
"epoch": 0.7656377592683326,
|
| 8242 |
+
"grad_norm": Infinity,
|
| 8243 |
+
"learning_rate": 2.7511286676002823e-05,
|
| 8244 |
+
"loss": 2.7432,
|
| 8245 |
+
"step": 1172
|
| 8246 |
+
},
|
| 8247 |
+
{
|
| 8248 |
+
"epoch": 0.7662910338069574,
|
| 8249 |
+
"grad_norm": Infinity,
|
| 8250 |
+
"learning_rate": 2.7365224316178384e-05,
|
| 8251 |
+
"loss": 2.7977,
|
| 8252 |
+
"step": 1173
|
| 8253 |
+
},
|
| 8254 |
+
{
|
| 8255 |
+
"epoch": 0.7669443083455822,
|
| 8256 |
+
"grad_norm": Infinity,
|
| 8257 |
+
"learning_rate": 2.7219489237322026e-05,
|
| 8258 |
+
"loss": 3.5219,
|
| 8259 |
+
"step": 1174
|
| 8260 |
+
},
|
| 8261 |
+
{
|
| 8262 |
+
"epoch": 0.7675975828842071,
|
| 8263 |
+
"grad_norm": Infinity,
|
| 8264 |
+
"learning_rate": 2.707408209609339e-05,
|
| 8265 |
+
"loss": 3.8416,
|
| 8266 |
+
"step": 1175
|
| 8267 |
+
},
|
| 8268 |
+
{
|
| 8269 |
+
"epoch": 0.768250857422832,
|
| 8270 |
+
"grad_norm": Infinity,
|
| 8271 |
+
"learning_rate": 2.692900354767425e-05,
|
| 8272 |
+
"loss": 1.4184,
|
| 8273 |
+
"step": 1176
|
| 8274 |
+
},
|
| 8275 |
+
{
|
| 8276 |
+
"epoch": 0.7689041319614568,
|
| 8277 |
+
"grad_norm": Infinity,
|
| 8278 |
+
"learning_rate": 2.678425424576596e-05,
|
| 8279 |
+
"loss": 1.3855,
|
| 8280 |
+
"step": 1177
|
| 8281 |
+
},
|
| 8282 |
+
{
|
| 8283 |
+
"epoch": 0.7695574065000816,
|
| 8284 |
+
"grad_norm": Infinity,
|
| 8285 |
+
"learning_rate": 2.6639834842586365e-05,
|
| 8286 |
+
"loss": 1.4003,
|
| 8287 |
+
"step": 1178
|
| 8288 |
+
},
|
| 8289 |
+
{
|
| 8290 |
+
"epoch": 0.7702106810387065,
|
| 8291 |
+
"grad_norm": Infinity,
|
| 8292 |
+
"learning_rate": 2.649574598886665e-05,
|
| 8293 |
+
"loss": 1.5033,
|
| 8294 |
+
"step": 1179
|
| 8295 |
+
},
|
| 8296 |
+
{
|
| 8297 |
+
"epoch": 0.7708639555773313,
|
| 8298 |
+
"grad_norm": Infinity,
|
| 8299 |
+
"learning_rate": 2.6351988333848788e-05,
|
| 8300 |
+
"loss": 1.4147,
|
| 8301 |
+
"step": 1180
|
| 8302 |
+
},
|
| 8303 |
+
{
|
| 8304 |
+
"epoch": 0.7715172301159562,
|
| 8305 |
+
"grad_norm": Infinity,
|
| 8306 |
+
"learning_rate": 2.620856252528232e-05,
|
| 8307 |
+
"loss": 1.4997,
|
| 8308 |
+
"step": 1181
|
| 8309 |
+
},
|
| 8310 |
+
{
|
| 8311 |
+
"epoch": 0.772170504654581,
|
| 8312 |
+
"grad_norm": Infinity,
|
| 8313 |
+
"learning_rate": 2.6065469209421566e-05,
|
| 8314 |
+
"loss": 1.5917,
|
| 8315 |
+
"step": 1182
|
| 8316 |
+
},
|
| 8317 |
+
{
|
| 8318 |
+
"epoch": 0.772823779193206,
|
| 8319 |
+
"grad_norm": Infinity,
|
| 8320 |
+
"learning_rate": 2.5922709031022686e-05,
|
| 8321 |
+
"loss": 1.5591,
|
| 8322 |
+
"step": 1183
|
| 8323 |
+
},
|
| 8324 |
+
{
|
| 8325 |
+
"epoch": 0.7734770537318308,
|
| 8326 |
+
"grad_norm": Infinity,
|
| 8327 |
+
"learning_rate": 2.578028263334078e-05,
|
| 8328 |
+
"loss": 1.5877,
|
| 8329 |
+
"step": 1184
|
| 8330 |
+
},
|
| 8331 |
+
{
|
| 8332 |
+
"epoch": 0.7741303282704557,
|
| 8333 |
+
"grad_norm": Infinity,
|
| 8334 |
+
"learning_rate": 2.5638190658126938e-05,
|
| 8335 |
+
"loss": 1.6238,
|
| 8336 |
+
"step": 1185
|
| 8337 |
+
},
|
| 8338 |
+
{
|
| 8339 |
+
"epoch": 0.7747836028090805,
|
| 8340 |
+
"grad_norm": Infinity,
|
| 8341 |
+
"learning_rate": 2.549643374562549e-05,
|
| 8342 |
+
"loss": 1.8098,
|
| 8343 |
+
"step": 1186
|
| 8344 |
+
},
|
| 8345 |
+
{
|
| 8346 |
+
"epoch": 0.7754368773477054,
|
| 8347 |
+
"grad_norm": Infinity,
|
| 8348 |
+
"learning_rate": 2.5355012534570953e-05,
|
| 8349 |
+
"loss": 1.8082,
|
| 8350 |
+
"step": 1187
|
| 8351 |
+
},
|
| 8352 |
+
{
|
| 8353 |
+
"epoch": 0.7760901518863302,
|
| 8354 |
+
"grad_norm": Infinity,
|
| 8355 |
+
"learning_rate": 2.52139276621852e-05,
|
| 8356 |
+
"loss": 1.8164,
|
| 8357 |
+
"step": 1188
|
| 8358 |
+
},
|
| 8359 |
+
{
|
| 8360 |
+
"epoch": 0.7767434264249551,
|
| 8361 |
+
"grad_norm": Infinity,
|
| 8362 |
+
"learning_rate": 2.507317976417475e-05,
|
| 8363 |
+
"loss": 2.0629,
|
| 8364 |
+
"step": 1189
|
| 8365 |
+
},
|
| 8366 |
+
{
|
| 8367 |
+
"epoch": 0.7773967009635799,
|
| 8368 |
+
"grad_norm": Infinity,
|
| 8369 |
+
"learning_rate": 2.493276947472756e-05,
|
| 8370 |
+
"loss": 2.1745,
|
| 8371 |
+
"step": 1190
|
| 8372 |
+
},
|
| 8373 |
+
{
|
| 8374 |
+
"epoch": 0.7780499755022048,
|
| 8375 |
+
"grad_norm": Infinity,
|
| 8376 |
+
"learning_rate": 2.4792697426510582e-05,
|
| 8377 |
+
"loss": 2.1568,
|
| 8378 |
+
"step": 1191
|
| 8379 |
+
},
|
| 8380 |
+
{
|
| 8381 |
+
"epoch": 0.7787032500408296,
|
| 8382 |
+
"grad_norm": Infinity,
|
| 8383 |
+
"learning_rate": 2.4652964250666567e-05,
|
| 8384 |
+
"loss": 2.3944,
|
| 8385 |
+
"step": 1192
|
| 8386 |
+
},
|
| 8387 |
+
{
|
| 8388 |
+
"epoch": 0.7793565245794545,
|
| 8389 |
+
"grad_norm": Infinity,
|
| 8390 |
+
"learning_rate": 2.4513570576811395e-05,
|
| 8391 |
+
"loss": 2.4762,
|
| 8392 |
+
"step": 1193
|
| 8393 |
+
},
|
| 8394 |
+
{
|
| 8395 |
+
"epoch": 0.7800097991180793,
|
| 8396 |
+
"grad_norm": Infinity,
|
| 8397 |
+
"learning_rate": 2.4374517033031285e-05,
|
| 8398 |
+
"loss": 2.3231,
|
| 8399 |
+
"step": 1194
|
| 8400 |
+
},
|
| 8401 |
+
{
|
| 8402 |
+
"epoch": 0.7806630736567042,
|
| 8403 |
+
"grad_norm": Infinity,
|
| 8404 |
+
"learning_rate": 2.4235804245879723e-05,
|
| 8405 |
+
"loss": 2.4375,
|
| 8406 |
+
"step": 1195
|
| 8407 |
+
},
|
| 8408 |
+
{
|
| 8409 |
+
"epoch": 0.7813163481953291,
|
| 8410 |
+
"grad_norm": Infinity,
|
| 8411 |
+
"learning_rate": 2.4097432840374945e-05,
|
| 8412 |
+
"loss": 2.8067,
|
| 8413 |
+
"step": 1196
|
| 8414 |
+
},
|
| 8415 |
+
{
|
| 8416 |
+
"epoch": 0.781969622733954,
|
| 8417 |
+
"grad_norm": Infinity,
|
| 8418 |
+
"learning_rate": 2.3959403439996907e-05,
|
| 8419 |
+
"loss": 2.957,
|
| 8420 |
+
"step": 1197
|
| 8421 |
+
},
|
| 8422 |
+
{
|
| 8423 |
+
"epoch": 0.7826228972725788,
|
| 8424 |
+
"grad_norm": Infinity,
|
| 8425 |
+
"learning_rate": 2.382171666668456e-05,
|
| 8426 |
+
"loss": 3.3291,
|
| 8427 |
+
"step": 1198
|
| 8428 |
+
},
|
| 8429 |
+
{
|
| 8430 |
+
"epoch": 0.7832761718112037,
|
| 8431 |
+
"grad_norm": Infinity,
|
| 8432 |
+
"learning_rate": 2.3684373140833016e-05,
|
| 8433 |
+
"loss": 3.5427,
|
| 8434 |
+
"step": 1199
|
| 8435 |
+
},
|
| 8436 |
+
{
|
| 8437 |
+
"epoch": 0.7839294463498285,
|
| 8438 |
+
"grad_norm": Infinity,
|
| 8439 |
+
"learning_rate": 2.354737348129077e-05,
|
| 8440 |
+
"loss": 3.9099,
|
| 8441 |
+
"step": 1200
|
| 8442 |
+
},
|
| 8443 |
+
{
|
| 8444 |
+
"epoch": 0.7845827208884534,
|
| 8445 |
+
"grad_norm": Infinity,
|
| 8446 |
+
"learning_rate": 2.3410718305356894e-05,
|
| 8447 |
+
"loss": 1.3826,
|
| 8448 |
+
"step": 1201
|
| 8449 |
+
},
|
| 8450 |
+
{
|
| 8451 |
+
"epoch": 0.7852359954270782,
|
| 8452 |
+
"grad_norm": Infinity,
|
| 8453 |
+
"learning_rate": 2.3274408228778355e-05,
|
| 8454 |
+
"loss": 1.4272,
|
| 8455 |
+
"step": 1202
|
| 8456 |
+
},
|
| 8457 |
+
{
|
| 8458 |
+
"epoch": 0.7858892699657031,
|
| 8459 |
+
"grad_norm": Infinity,
|
| 8460 |
+
"learning_rate": 2.3138443865747062e-05,
|
| 8461 |
+
"loss": 1.5302,
|
| 8462 |
+
"step": 1203
|
| 8463 |
+
},
|
| 8464 |
+
{
|
| 8465 |
+
"epoch": 0.7865425445043279,
|
| 8466 |
+
"grad_norm": Infinity,
|
| 8467 |
+
"learning_rate": 2.300282582889719e-05,
|
| 8468 |
+
"loss": 1.4341,
|
| 8469 |
+
"step": 1204
|
| 8470 |
+
},
|
| 8471 |
+
{
|
| 8472 |
+
"epoch": 0.7871958190429528,
|
| 8473 |
+
"grad_norm": Infinity,
|
| 8474 |
+
"learning_rate": 2.2867554729302542e-05,
|
| 8475 |
+
"loss": 1.5308,
|
| 8476 |
+
"step": 1205
|
| 8477 |
+
},
|
| 8478 |
+
{
|
| 8479 |
+
"epoch": 0.7878490935815776,
|
| 8480 |
+
"grad_norm": Infinity,
|
| 8481 |
+
"learning_rate": 2.2732631176473485e-05,
|
| 8482 |
+
"loss": 1.4427,
|
| 8483 |
+
"step": 1206
|
| 8484 |
+
},
|
| 8485 |
+
{
|
| 8486 |
+
"epoch": 0.7885023681202025,
|
| 8487 |
+
"grad_norm": Infinity,
|
| 8488 |
+
"learning_rate": 2.2598055778354587e-05,
|
| 8489 |
+
"loss": 1.4804,
|
| 8490 |
+
"step": 1207
|
| 8491 |
+
},
|
| 8492 |
+
{
|
| 8493 |
+
"epoch": 0.7891556426588274,
|
| 8494 |
+
"grad_norm": Infinity,
|
| 8495 |
+
"learning_rate": 2.2463829141321548e-05,
|
| 8496 |
+
"loss": 1.6638,
|
| 8497 |
+
"step": 1208
|
| 8498 |
+
},
|
| 8499 |
+
{
|
| 8500 |
+
"epoch": 0.7898089171974523,
|
| 8501 |
+
"grad_norm": Infinity,
|
| 8502 |
+
"learning_rate": 2.2329951870178655e-05,
|
| 8503 |
+
"loss": 1.5226,
|
| 8504 |
+
"step": 1209
|
| 8505 |
+
},
|
| 8506 |
+
{
|
| 8507 |
+
"epoch": 0.7904621917360771,
|
| 8508 |
+
"grad_norm": Infinity,
|
| 8509 |
+
"learning_rate": 2.2196424568156073e-05,
|
| 8510 |
+
"loss": 1.7148,
|
| 8511 |
+
"step": 1210
|
| 8512 |
+
},
|
| 8513 |
+
{
|
| 8514 |
+
"epoch": 0.791115466274702,
|
| 8515 |
+
"grad_norm": Infinity,
|
| 8516 |
+
"learning_rate": 2.2063247836906886e-05,
|
| 8517 |
+
"loss": 1.6323,
|
| 8518 |
+
"step": 1211
|
| 8519 |
+
},
|
| 8520 |
+
{
|
| 8521 |
+
"epoch": 0.7917687408133268,
|
| 8522 |
+
"grad_norm": Infinity,
|
| 8523 |
+
"learning_rate": 2.1930422276504747e-05,
|
| 8524 |
+
"loss": 1.7264,
|
| 8525 |
+
"step": 1212
|
| 8526 |
+
},
|
| 8527 |
+
{
|
| 8528 |
+
"epoch": 0.7924220153519517,
|
| 8529 |
+
"grad_norm": Infinity,
|
| 8530 |
+
"learning_rate": 2.17979484854409e-05,
|
| 8531 |
+
"loss": 1.8764,
|
| 8532 |
+
"step": 1213
|
| 8533 |
+
},
|
| 8534 |
+
{
|
| 8535 |
+
"epoch": 0.7930752898905765,
|
| 8536 |
+
"grad_norm": Infinity,
|
| 8537 |
+
"learning_rate": 2.1665827060621568e-05,
|
| 8538 |
+
"loss": 1.8152,
|
| 8539 |
+
"step": 1214
|
| 8540 |
+
},
|
| 8541 |
+
{
|
| 8542 |
+
"epoch": 0.7937285644292014,
|
| 8543 |
+
"grad_norm": Infinity,
|
| 8544 |
+
"learning_rate": 2.153405859736528e-05,
|
| 8545 |
+
"loss": 2.0223,
|
| 8546 |
+
"step": 1215
|
| 8547 |
+
},
|
| 8548 |
+
{
|
| 8549 |
+
"epoch": 0.7943818389678262,
|
| 8550 |
+
"grad_norm": Infinity,
|
| 8551 |
+
"learning_rate": 2.1402643689400192e-05,
|
| 8552 |
+
"loss": 1.9798,
|
| 8553 |
+
"step": 1216
|
| 8554 |
+
},
|
| 8555 |
+
{
|
| 8556 |
+
"epoch": 0.7950351135064511,
|
| 8557 |
+
"grad_norm": Infinity,
|
| 8558 |
+
"learning_rate": 2.1271582928861323e-05,
|
| 8559 |
+
"loss": 2.2652,
|
| 8560 |
+
"step": 1217
|
| 8561 |
+
},
|
| 8562 |
+
{
|
| 8563 |
+
"epoch": 0.7956883880450759,
|
| 8564 |
+
"grad_norm": Infinity,
|
| 8565 |
+
"learning_rate": 2.1140876906288086e-05,
|
| 8566 |
+
"loss": 2.5652,
|
| 8567 |
+
"step": 1218
|
| 8568 |
+
},
|
| 8569 |
+
{
|
| 8570 |
+
"epoch": 0.7963416625837008,
|
| 8571 |
+
"grad_norm": Infinity,
|
| 8572 |
+
"learning_rate": 2.1010526210621406e-05,
|
| 8573 |
+
"loss": 2.6046,
|
| 8574 |
+
"step": 1219
|
| 8575 |
+
},
|
| 8576 |
+
{
|
| 8577 |
+
"epoch": 0.7969949371223256,
|
| 8578 |
+
"grad_norm": Infinity,
|
| 8579 |
+
"learning_rate": 2.0880531429201145e-05,
|
| 8580 |
+
"loss": 2.4434,
|
| 8581 |
+
"step": 1220
|
| 8582 |
+
},
|
| 8583 |
+
{
|
| 8584 |
+
"epoch": 0.7976482116609506,
|
| 8585 |
+
"grad_norm": Infinity,
|
| 8586 |
+
"learning_rate": 2.0750893147763596e-05,
|
| 8587 |
+
"loss": 2.7545,
|
| 8588 |
+
"step": 1221
|
| 8589 |
+
},
|
| 8590 |
+
{
|
| 8591 |
+
"epoch": 0.7983014861995754,
|
| 8592 |
+
"grad_norm": Infinity,
|
| 8593 |
+
"learning_rate": 2.0621611950438512e-05,
|
| 8594 |
+
"loss": 3.1435,
|
| 8595 |
+
"step": 1222
|
| 8596 |
+
},
|
| 8597 |
+
{
|
| 8598 |
+
"epoch": 0.7989547607382003,
|
| 8599 |
+
"grad_norm": Infinity,
|
| 8600 |
+
"learning_rate": 2.0492688419746897e-05,
|
| 8601 |
+
"loss": 3.2472,
|
| 8602 |
+
"step": 1223
|
| 8603 |
+
},
|
| 8604 |
+
{
|
| 8605 |
+
"epoch": 0.7996080352768251,
|
| 8606 |
+
"grad_norm": Infinity,
|
| 8607 |
+
"learning_rate": 2.0364123136598035e-05,
|
| 8608 |
+
"loss": 3.4464,
|
| 8609 |
+
"step": 1224
|
| 8610 |
+
},
|
| 8611 |
+
{
|
| 8612 |
+
"epoch": 0.80026130981545,
|
| 8613 |
+
"grad_norm": Infinity,
|
| 8614 |
+
"learning_rate": 2.0235916680287015e-05,
|
| 8615 |
+
"loss": 3.9101,
|
| 8616 |
+
"step": 1225
|
| 8617 |
+
},
|
| 8618 |
+
{
|
| 8619 |
+
"epoch": 0.8009145843540748,
|
| 8620 |
+
"grad_norm": Infinity,
|
| 8621 |
+
"learning_rate": 2.010806962849219e-05,
|
| 8622 |
+
"loss": 1.3025,
|
| 8623 |
+
"step": 1226
|
| 8624 |
+
},
|
| 8625 |
+
{
|
| 8626 |
+
"epoch": 0.8015678588926997,
|
| 8627 |
+
"grad_norm": Infinity,
|
| 8628 |
+
"learning_rate": 1.998058255727234e-05,
|
| 8629 |
+
"loss": 1.2785,
|
| 8630 |
+
"step": 1227
|
| 8631 |
+
},
|
| 8632 |
+
{
|
| 8633 |
+
"epoch": 0.8022211334313245,
|
| 8634 |
+
"grad_norm": Infinity,
|
| 8635 |
+
"learning_rate": 1.985345604106439e-05,
|
| 8636 |
+
"loss": 1.397,
|
| 8637 |
+
"step": 1228
|
| 8638 |
+
},
|
| 8639 |
+
{
|
| 8640 |
+
"epoch": 0.8028744079699494,
|
| 8641 |
+
"grad_norm": Infinity,
|
| 8642 |
+
"learning_rate": 1.9726690652680578e-05,
|
| 8643 |
+
"loss": 1.4421,
|
| 8644 |
+
"step": 1229
|
| 8645 |
+
},
|
| 8646 |
+
{
|
| 8647 |
+
"epoch": 0.8035276825085742,
|
| 8648 |
+
"grad_norm": Infinity,
|
| 8649 |
+
"learning_rate": 1.9600286963305957e-05,
|
| 8650 |
+
"loss": 1.5002,
|
| 8651 |
+
"step": 1230
|
| 8652 |
+
},
|
| 8653 |
+
{
|
| 8654 |
+
"epoch": 0.804180957047199,
|
| 8655 |
+
"grad_norm": Infinity,
|
| 8656 |
+
"learning_rate": 1.9474245542495807e-05,
|
| 8657 |
+
"loss": 1.5233,
|
| 8658 |
+
"step": 1231
|
| 8659 |
+
},
|
| 8660 |
+
{
|
| 8661 |
+
"epoch": 0.8048342315858239,
|
| 8662 |
+
"grad_norm": Infinity,
|
| 8663 |
+
"learning_rate": 1.9348566958173197e-05,
|
| 8664 |
+
"loss": 1.5117,
|
| 8665 |
+
"step": 1232
|
| 8666 |
+
},
|
| 8667 |
+
{
|
| 8668 |
+
"epoch": 0.8054875061244487,
|
| 8669 |
+
"grad_norm": Infinity,
|
| 8670 |
+
"learning_rate": 1.9223251776626107e-05,
|
| 8671 |
+
"loss": 1.597,
|
| 8672 |
+
"step": 1233
|
| 8673 |
+
},
|
| 8674 |
+
{
|
| 8675 |
+
"epoch": 0.8061407806630737,
|
| 8676 |
+
"grad_norm": Infinity,
|
| 8677 |
+
"learning_rate": 1.9098300562505266e-05,
|
| 8678 |
+
"loss": 1.5572,
|
| 8679 |
+
"step": 1234
|
| 8680 |
+
},
|
| 8681 |
+
{
|
| 8682 |
+
"epoch": 0.8067940552016986,
|
| 8683 |
+
"grad_norm": Infinity,
|
| 8684 |
+
"learning_rate": 1.897371387882134e-05,
|
| 8685 |
+
"loss": 1.7387,
|
| 8686 |
+
"step": 1235
|
| 8687 |
+
},
|
| 8688 |
+
{
|
| 8689 |
+
"epoch": 0.8074473297403234,
|
| 8690 |
+
"grad_norm": Infinity,
|
| 8691 |
+
"learning_rate": 1.884949228694246e-05,
|
| 8692 |
+
"loss": 1.738,
|
| 8693 |
+
"step": 1236
|
| 8694 |
+
},
|
| 8695 |
+
{
|
| 8696 |
+
"epoch": 0.8081006042789483,
|
| 8697 |
+
"grad_norm": Infinity,
|
| 8698 |
+
"learning_rate": 1.8725636346591822e-05,
|
| 8699 |
+
"loss": 1.7022,
|
| 8700 |
+
"step": 1237
|
| 8701 |
+
},
|
| 8702 |
+
{
|
| 8703 |
+
"epoch": 0.8087538788175731,
|
| 8704 |
+
"grad_norm": Infinity,
|
| 8705 |
+
"learning_rate": 1.860214661584486e-05,
|
| 8706 |
+
"loss": 1.8925,
|
| 8707 |
+
"step": 1238
|
| 8708 |
+
},
|
| 8709 |
+
{
|
| 8710 |
+
"epoch": 0.809407153356198,
|
| 8711 |
+
"grad_norm": Infinity,
|
| 8712 |
+
"learning_rate": 1.8479023651127115e-05,
|
| 8713 |
+
"loss": 2.0958,
|
| 8714 |
+
"step": 1239
|
| 8715 |
+
},
|
| 8716 |
+
{
|
| 8717 |
+
"epoch": 0.8100604278948228,
|
| 8718 |
+
"grad_norm": Infinity,
|
| 8719 |
+
"learning_rate": 1.835626800721144e-05,
|
| 8720 |
+
"loss": 1.893,
|
| 8721 |
+
"step": 1240
|
| 8722 |
+
},
|
| 8723 |
+
{
|
| 8724 |
+
"epoch": 0.8107137024334476,
|
| 8725 |
+
"grad_norm": Infinity,
|
| 8726 |
+
"learning_rate": 1.8233880237215585e-05,
|
| 8727 |
+
"loss": 2.0566,
|
| 8728 |
+
"step": 1241
|
| 8729 |
+
},
|
| 8730 |
+
{
|
| 8731 |
+
"epoch": 0.8113669769720725,
|
| 8732 |
+
"grad_norm": Infinity,
|
| 8733 |
+
"learning_rate": 1.811186089259983e-05,
|
| 8734 |
+
"loss": 2.0772,
|
| 8735 |
+
"step": 1242
|
| 8736 |
+
},
|
| 8737 |
+
{
|
| 8738 |
+
"epoch": 0.8120202515106973,
|
| 8739 |
+
"grad_norm": Infinity,
|
| 8740 |
+
"learning_rate": 1.7990210523164198e-05,
|
| 8741 |
+
"loss": 2.3327,
|
| 8742 |
+
"step": 1243
|
| 8743 |
+
},
|
| 8744 |
+
{
|
| 8745 |
+
"epoch": 0.8126735260493222,
|
| 8746 |
+
"grad_norm": Infinity,
|
| 8747 |
+
"learning_rate": 1.7868929677046364e-05,
|
| 8748 |
+
"loss": 2.2319,
|
| 8749 |
+
"step": 1244
|
| 8750 |
+
},
|
| 8751 |
+
{
|
| 8752 |
+
"epoch": 0.813326800587947,
|
| 8753 |
+
"grad_norm": Infinity,
|
| 8754 |
+
"learning_rate": 1.7748018900718854e-05,
|
| 8755 |
+
"loss": 2.6381,
|
| 8756 |
+
"step": 1245
|
| 8757 |
+
},
|
| 8758 |
+
{
|
| 8759 |
+
"epoch": 0.8139800751265719,
|
| 8760 |
+
"grad_norm": Infinity,
|
| 8761 |
+
"learning_rate": 1.762747873898679e-05,
|
| 8762 |
+
"loss": 2.5717,
|
| 8763 |
+
"step": 1246
|
| 8764 |
+
},
|
| 8765 |
+
{
|
| 8766 |
+
"epoch": 0.8146333496651968,
|
| 8767 |
+
"grad_norm": Infinity,
|
| 8768 |
+
"learning_rate": 1.750730973498529e-05,
|
| 8769 |
+
"loss": 3.0298,
|
| 8770 |
+
"step": 1247
|
| 8771 |
+
},
|
| 8772 |
+
{
|
| 8773 |
+
"epoch": 0.8152866242038217,
|
| 8774 |
+
"grad_norm": Infinity,
|
| 8775 |
+
"learning_rate": 1.7387512430177234e-05,
|
| 8776 |
+
"loss": 3.1643,
|
| 8777 |
+
"step": 1248
|
| 8778 |
+
},
|
| 8779 |
+
{
|
| 8780 |
+
"epoch": 0.8159398987424465,
|
| 8781 |
+
"grad_norm": Infinity,
|
| 8782 |
+
"learning_rate": 1.726808736435046e-05,
|
| 8783 |
+
"loss": 3.1612,
|
| 8784 |
+
"step": 1249
|
| 8785 |
+
},
|
| 8786 |
+
{
|
| 8787 |
+
"epoch": 0.8165931732810714,
|
| 8788 |
+
"grad_norm": Infinity,
|
| 8789 |
+
"learning_rate": 1.7149035075615794e-05,
|
| 8790 |
+
"loss": 3.9105,
|
| 8791 |
+
"step": 1250
|
| 8792 |
+
},
|
| 8793 |
+
{
|
| 8794 |
+
"epoch": 0.8172464478196962,
|
| 8795 |
+
"grad_norm": Infinity,
|
| 8796 |
+
"learning_rate": 1.703035610040423e-05,
|
| 8797 |
+
"loss": 1.1581,
|
| 8798 |
+
"step": 1251
|
| 8799 |
+
},
|
| 8800 |
+
{
|
| 8801 |
+
"epoch": 0.8178997223583211,
|
| 8802 |
+
"grad_norm": Infinity,
|
| 8803 |
+
"learning_rate": 1.69120509734647e-05,
|
| 8804 |
+
"loss": 1.3175,
|
| 8805 |
+
"step": 1252
|
| 8806 |
+
},
|
| 8807 |
+
{
|
| 8808 |
+
"epoch": 0.8185529968969459,
|
| 8809 |
+
"grad_norm": Infinity,
|
| 8810 |
+
"learning_rate": 1.679412022786172e-05,
|
| 8811 |
+
"loss": 1.3558,
|
| 8812 |
+
"step": 1253
|
| 8813 |
+
},
|
| 8814 |
+
{
|
| 8815 |
+
"epoch": 0.8192062714355708,
|
| 8816 |
+
"grad_norm": Infinity,
|
| 8817 |
+
"learning_rate": 1.6676564394972727e-05,
|
| 8818 |
+
"loss": 1.48,
|
| 8819 |
+
"step": 1254
|
| 8820 |
+
},
|
| 8821 |
+
{
|
| 8822 |
+
"epoch": 0.8198595459741956,
|
| 8823 |
+
"grad_norm": Infinity,
|
| 8824 |
+
"learning_rate": 1.6559384004486055e-05,
|
| 8825 |
+
"loss": 1.5116,
|
| 8826 |
+
"step": 1255
|
| 8827 |
+
},
|
| 8828 |
+
{
|
| 8829 |
+
"epoch": 0.8205128205128205,
|
| 8830 |
+
"grad_norm": Infinity,
|
| 8831 |
+
"learning_rate": 1.6442579584398232e-05,
|
| 8832 |
+
"loss": 1.4282,
|
| 8833 |
+
"step": 1256
|
| 8834 |
+
},
|
| 8835 |
+
{
|
| 8836 |
+
"epoch": 0.8211660950514453,
|
| 8837 |
+
"grad_norm": Infinity,
|
| 8838 |
+
"learning_rate": 1.6326151661011724e-05,
|
| 8839 |
+
"loss": 1.4939,
|
| 8840 |
+
"step": 1257
|
| 8841 |
+
},
|
| 8842 |
+
{
|
| 8843 |
+
"epoch": 0.8218193695900702,
|
| 8844 |
+
"grad_norm": Infinity,
|
| 8845 |
+
"learning_rate": 1.6210100758932667e-05,
|
| 8846 |
+
"loss": 1.4974,
|
| 8847 |
+
"step": 1258
|
| 8848 |
+
},
|
| 8849 |
+
{
|
| 8850 |
+
"epoch": 0.822472644128695,
|
| 8851 |
+
"grad_norm": Infinity,
|
| 8852 |
+
"learning_rate": 1.6094427401068224e-05,
|
| 8853 |
+
"loss": 1.4659,
|
| 8854 |
+
"step": 1259
|
| 8855 |
+
},
|
| 8856 |
+
{
|
| 8857 |
+
"epoch": 0.82312591866732,
|
| 8858 |
+
"grad_norm": Infinity,
|
| 8859 |
+
"learning_rate": 1.5979132108624574e-05,
|
| 8860 |
+
"loss": 1.5381,
|
| 8861 |
+
"step": 1260
|
| 8862 |
+
},
|
| 8863 |
+
{
|
| 8864 |
+
"epoch": 0.8237791932059448,
|
| 8865 |
+
"grad_norm": Infinity,
|
| 8866 |
+
"learning_rate": 1.5864215401104287e-05,
|
| 8867 |
+
"loss": 1.7296,
|
| 8868 |
+
"step": 1261
|
| 8869 |
+
},
|
| 8870 |
+
{
|
| 8871 |
+
"epoch": 0.8244324677445697,
|
| 8872 |
+
"grad_norm": Infinity,
|
| 8873 |
+
"learning_rate": 1.574967779630414e-05,
|
| 8874 |
+
"loss": 1.7319,
|
| 8875 |
+
"step": 1262
|
| 8876 |
+
},
|
| 8877 |
+
{
|
| 8878 |
+
"epoch": 0.8250857422831945,
|
| 8879 |
+
"grad_norm": Infinity,
|
| 8880 |
+
"learning_rate": 1.563551981031267e-05,
|
| 8881 |
+
"loss": 1.7433,
|
| 8882 |
+
"step": 1263
|
| 8883 |
+
},
|
| 8884 |
+
{
|
| 8885 |
+
"epoch": 0.8257390168218194,
|
| 8886 |
+
"grad_norm": Infinity,
|
| 8887 |
+
"learning_rate": 1.552174195750803e-05,
|
| 8888 |
+
"loss": 1.8011,
|
| 8889 |
+
"step": 1264
|
| 8890 |
+
},
|
| 8891 |
+
{
|
| 8892 |
+
"epoch": 0.8263922913604442,
|
| 8893 |
+
"grad_norm": Infinity,
|
| 8894 |
+
"learning_rate": 1.5408344750555383e-05,
|
| 8895 |
+
"loss": 2.0143,
|
| 8896 |
+
"step": 1265
|
| 8897 |
+
},
|
| 8898 |
+
{
|
| 8899 |
+
"epoch": 0.8270455658990691,
|
| 8900 |
+
"grad_norm": Infinity,
|
| 8901 |
+
"learning_rate": 1.529532870040492e-05,
|
| 8902 |
+
"loss": 2.0743,
|
| 8903 |
+
"step": 1266
|
| 8904 |
+
},
|
| 8905 |
+
{
|
| 8906 |
+
"epoch": 0.8276988404376939,
|
| 8907 |
+
"grad_norm": Infinity,
|
| 8908 |
+
"learning_rate": 1.5182694316289314e-05,
|
| 8909 |
+
"loss": 2.1506,
|
| 8910 |
+
"step": 1267
|
| 8911 |
+
},
|
| 8912 |
+
{
|
| 8913 |
+
"epoch": 0.8283521149763188,
|
| 8914 |
+
"grad_norm": Infinity,
|
| 8915 |
+
"learning_rate": 1.5070442105721495e-05,
|
| 8916 |
+
"loss": 2.1783,
|
| 8917 |
+
"step": 1268
|
| 8918 |
+
},
|
| 8919 |
+
{
|
| 8920 |
+
"epoch": 0.8290053895149436,
|
| 8921 |
+
"grad_norm": Infinity,
|
| 8922 |
+
"learning_rate": 1.4958572574492501e-05,
|
| 8923 |
+
"loss": 2.5343,
|
| 8924 |
+
"step": 1269
|
| 8925 |
+
},
|
| 8926 |
+
{
|
| 8927 |
+
"epoch": 0.8296586640535685,
|
| 8928 |
+
"grad_norm": Infinity,
|
| 8929 |
+
"learning_rate": 1.4847086226668872e-05,
|
| 8930 |
+
"loss": 2.6854,
|
| 8931 |
+
"step": 1270
|
| 8932 |
+
},
|
| 8933 |
+
{
|
| 8934 |
+
"epoch": 0.8303119385921933,
|
| 8935 |
+
"grad_norm": Infinity,
|
| 8936 |
+
"learning_rate": 1.4735983564590783e-05,
|
| 8937 |
+
"loss": 2.8695,
|
| 8938 |
+
"step": 1271
|
| 8939 |
+
},
|
| 8940 |
+
{
|
| 8941 |
+
"epoch": 0.8309652131308183,
|
| 8942 |
+
"grad_norm": Infinity,
|
| 8943 |
+
"learning_rate": 1.4625265088869477e-05,
|
| 8944 |
+
"loss": 2.959,
|
| 8945 |
+
"step": 1272
|
| 8946 |
+
},
|
| 8947 |
+
{
|
| 8948 |
+
"epoch": 0.8316184876694431,
|
| 8949 |
+
"grad_norm": Infinity,
|
| 8950 |
+
"learning_rate": 1.4514931298385093e-05,
|
| 8951 |
+
"loss": 3.3266,
|
| 8952 |
+
"step": 1273
|
| 8953 |
+
},
|
| 8954 |
+
{
|
| 8955 |
+
"epoch": 0.832271762208068,
|
| 8956 |
+
"grad_norm": Infinity,
|
| 8957 |
+
"learning_rate": 1.4404982690284575e-05,
|
| 8958 |
+
"loss": 3.8966,
|
| 8959 |
+
"step": 1274
|
| 8960 |
+
},
|
| 8961 |
+
{
|
| 8962 |
+
"epoch": 0.8329250367466928,
|
| 8963 |
+
"grad_norm": Infinity,
|
| 8964 |
+
"learning_rate": 1.429541975997908e-05,
|
| 8965 |
+
"loss": 4.5467,
|
| 8966 |
+
"step": 1275
|
| 8967 |
+
},
|
| 8968 |
+
{
|
| 8969 |
+
"epoch": 0.8335783112853177,
|
| 8970 |
+
"grad_norm": Infinity,
|
| 8971 |
+
"learning_rate": 1.4186243001142164e-05,
|
| 8972 |
+
"loss": 1.2431,
|
| 8973 |
+
"step": 1276
|
| 8974 |
+
},
|
| 8975 |
+
{
|
| 8976 |
+
"epoch": 0.8342315858239425,
|
| 8977 |
+
"grad_norm": Infinity,
|
| 8978 |
+
"learning_rate": 1.4077452905707234e-05,
|
| 8979 |
+
"loss": 1.424,
|
| 8980 |
+
"step": 1277
|
| 8981 |
+
},
|
| 8982 |
+
{
|
| 8983 |
+
"epoch": 0.8348848603625674,
|
| 8984 |
+
"grad_norm": Infinity,
|
| 8985 |
+
"learning_rate": 1.396904996386551e-05,
|
| 8986 |
+
"loss": 1.4275,
|
| 8987 |
+
"step": 1278
|
| 8988 |
+
},
|
| 8989 |
+
{
|
| 8990 |
+
"epoch": 0.8355381349011922,
|
| 8991 |
+
"grad_norm": Infinity,
|
| 8992 |
+
"learning_rate": 1.38610346640637e-05,
|
| 8993 |
+
"loss": 1.3774,
|
| 8994 |
+
"step": 1279
|
| 8995 |
+
},
|
| 8996 |
+
{
|
| 8997 |
+
"epoch": 0.8361914094398171,
|
| 8998 |
+
"grad_norm": Infinity,
|
| 8999 |
+
"learning_rate": 1.3753407493001968e-05,
|
| 9000 |
+
"loss": 1.4958,
|
| 9001 |
+
"step": 1280
|
| 9002 |
+
},
|
| 9003 |
+
{
|
| 9004 |
+
"epoch": 0.8368446839784419,
|
| 9005 |
+
"grad_norm": Infinity,
|
| 9006 |
+
"learning_rate": 1.3646168935631464e-05,
|
| 9007 |
+
"loss": 1.4944,
|
| 9008 |
+
"step": 1281
|
| 9009 |
+
},
|
| 9010 |
+
{
|
| 9011 |
+
"epoch": 0.8374979585170668,
|
| 9012 |
+
"grad_norm": Infinity,
|
| 9013 |
+
"learning_rate": 1.3539319475152457e-05,
|
| 9014 |
+
"loss": 1.5639,
|
| 9015 |
+
"step": 1282
|
| 9016 |
+
},
|
| 9017 |
+
{
|
| 9018 |
+
"epoch": 0.8381512330556916,
|
| 9019 |
+
"grad_norm": Infinity,
|
| 9020 |
+
"learning_rate": 1.3432859593011948e-05,
|
| 9021 |
+
"loss": 1.6105,
|
| 9022 |
+
"step": 1283
|
| 9023 |
+
},
|
| 9024 |
+
{
|
| 9025 |
+
"epoch": 0.8388045075943165,
|
| 9026 |
+
"grad_norm": Infinity,
|
| 9027 |
+
"learning_rate": 1.3326789768901504e-05,
|
| 9028 |
+
"loss": 1.7179,
|
| 9029 |
+
"step": 1284
|
| 9030 |
+
},
|
| 9031 |
+
{
|
| 9032 |
+
"epoch": 0.8394577821329414,
|
| 9033 |
+
"grad_norm": Infinity,
|
| 9034 |
+
"learning_rate": 1.3221110480755305e-05,
|
| 9035 |
+
"loss": 1.5552,
|
| 9036 |
+
"step": 1285
|
| 9037 |
+
},
|
| 9038 |
+
{
|
| 9039 |
+
"epoch": 0.8401110566715663,
|
| 9040 |
+
"grad_norm": Infinity,
|
| 9041 |
+
"learning_rate": 1.3115822204747619e-05,
|
| 9042 |
+
"loss": 1.6532,
|
| 9043 |
+
"step": 1286
|
| 9044 |
+
},
|
| 9045 |
+
{
|
| 9046 |
+
"epoch": 0.8407643312101911,
|
| 9047 |
+
"grad_norm": Infinity,
|
| 9048 |
+
"learning_rate": 1.3010925415291075e-05,
|
| 9049 |
+
"loss": 1.7276,
|
| 9050 |
+
"step": 1287
|
| 9051 |
+
},
|
| 9052 |
+
{
|
| 9053 |
+
"epoch": 0.841417605748816,
|
| 9054 |
+
"grad_norm": Infinity,
|
| 9055 |
+
"learning_rate": 1.2906420585034229e-05,
|
| 9056 |
+
"loss": 1.731,
|
| 9057 |
+
"step": 1288
|
| 9058 |
+
},
|
| 9059 |
+
{
|
| 9060 |
+
"epoch": 0.8420708802874408,
|
| 9061 |
+
"grad_norm": Infinity,
|
| 9062 |
+
"learning_rate": 1.2802308184859502e-05,
|
| 9063 |
+
"loss": 1.8079,
|
| 9064 |
+
"step": 1289
|
| 9065 |
+
},
|
| 9066 |
+
{
|
| 9067 |
+
"epoch": 0.8427241548260657,
|
| 9068 |
+
"grad_norm": Infinity,
|
| 9069 |
+
"learning_rate": 1.2698588683881186e-05,
|
| 9070 |
+
"loss": 1.8644,
|
| 9071 |
+
"step": 1290
|
| 9072 |
+
},
|
| 9073 |
+
{
|
| 9074 |
+
"epoch": 0.8433774293646905,
|
| 9075 |
+
"grad_norm": Infinity,
|
| 9076 |
+
"learning_rate": 1.2595262549443133e-05,
|
| 9077 |
+
"loss": 2.2312,
|
| 9078 |
+
"step": 1291
|
| 9079 |
+
},
|
| 9080 |
+
{
|
| 9081 |
+
"epoch": 0.8440307039033154,
|
| 9082 |
+
"grad_norm": Infinity,
|
| 9083 |
+
"learning_rate": 1.2492330247116802e-05,
|
| 9084 |
+
"loss": 2.2468,
|
| 9085 |
+
"step": 1292
|
| 9086 |
+
},
|
| 9087 |
+
{
|
| 9088 |
+
"epoch": 0.8446839784419402,
|
| 9089 |
+
"grad_norm": Infinity,
|
| 9090 |
+
"learning_rate": 1.2389792240699084e-05,
|
| 9091 |
+
"loss": 2.4955,
|
| 9092 |
+
"step": 1293
|
| 9093 |
+
},
|
| 9094 |
+
{
|
| 9095 |
+
"epoch": 0.8453372529805651,
|
| 9096 |
+
"grad_norm": Infinity,
|
| 9097 |
+
"learning_rate": 1.2287648992210243e-05,
|
| 9098 |
+
"loss": 2.4229,
|
| 9099 |
+
"step": 1294
|
| 9100 |
+
},
|
| 9101 |
+
{
|
| 9102 |
+
"epoch": 0.8459905275191899,
|
| 9103 |
+
"grad_norm": Infinity,
|
| 9104 |
+
"learning_rate": 1.2185900961891794e-05,
|
| 9105 |
+
"loss": 2.4597,
|
| 9106 |
+
"step": 1295
|
| 9107 |
+
},
|
| 9108 |
+
{
|
| 9109 |
+
"epoch": 0.8466438020578148,
|
| 9110 |
+
"grad_norm": Infinity,
|
| 9111 |
+
"learning_rate": 1.208454860820456e-05,
|
| 9112 |
+
"loss": 3.0822,
|
| 9113 |
+
"step": 1296
|
| 9114 |
+
},
|
| 9115 |
+
{
|
| 9116 |
+
"epoch": 0.8472970765964396,
|
| 9117 |
+
"grad_norm": Infinity,
|
| 9118 |
+
"learning_rate": 1.1983592387826347e-05,
|
| 9119 |
+
"loss": 3.045,
|
| 9120 |
+
"step": 1297
|
| 9121 |
+
},
|
| 9122 |
+
{
|
| 9123 |
+
"epoch": 0.8479503511350646,
|
| 9124 |
+
"grad_norm": Infinity,
|
| 9125 |
+
"learning_rate": 1.1883032755650204e-05,
|
| 9126 |
+
"loss": 3.0758,
|
| 9127 |
+
"step": 1298
|
| 9128 |
+
},
|
| 9129 |
+
{
|
| 9130 |
+
"epoch": 0.8486036256736894,
|
| 9131 |
+
"grad_norm": Infinity,
|
| 9132 |
+
"learning_rate": 1.1782870164782111e-05,
|
| 9133 |
+
"loss": 3.4634,
|
| 9134 |
+
"step": 1299
|
| 9135 |
+
},
|
| 9136 |
+
{
|
| 9137 |
+
"epoch": 0.8492569002123143,
|
| 9138 |
+
"grad_norm": Infinity,
|
| 9139 |
+
"learning_rate": 1.1683105066539068e-05,
|
| 9140 |
+
"loss": 4.4563,
|
| 9141 |
+
"step": 1300
|
| 9142 |
+
},
|
| 9143 |
+
{
|
| 9144 |
+
"epoch": 0.8499101747509391,
|
| 9145 |
+
"grad_norm": Infinity,
|
| 9146 |
+
"learning_rate": 1.15837379104471e-05,
|
| 9147 |
+
"loss": 1.1769,
|
| 9148 |
+
"step": 1301
|
| 9149 |
+
},
|
| 9150 |
+
{
|
| 9151 |
+
"epoch": 0.850563449289564,
|
| 9152 |
+
"grad_norm": Infinity,
|
| 9153 |
+
"learning_rate": 1.1484769144239037e-05,
|
| 9154 |
+
"loss": 1.3331,
|
| 9155 |
+
"step": 1302
|
| 9156 |
+
},
|
| 9157 |
+
{
|
| 9158 |
+
"epoch": 0.8512167238281888,
|
| 9159 |
+
"grad_norm": Infinity,
|
| 9160 |
+
"learning_rate": 1.1386199213852755e-05,
|
| 9161 |
+
"loss": 1.361,
|
| 9162 |
+
"step": 1303
|
| 9163 |
+
},
|
| 9164 |
+
{
|
| 9165 |
+
"epoch": 0.8518699983668137,
|
| 9166 |
+
"grad_norm": Infinity,
|
| 9167 |
+
"learning_rate": 1.1288028563428965e-05,
|
| 9168 |
+
"loss": 1.4695,
|
| 9169 |
+
"step": 1304
|
| 9170 |
+
},
|
| 9171 |
+
{
|
| 9172 |
+
"epoch": 0.8525232729054385,
|
| 9173 |
+
"grad_norm": Infinity,
|
| 9174 |
+
"learning_rate": 1.1190257635309275e-05,
|
| 9175 |
+
"loss": 1.5785,
|
| 9176 |
+
"step": 1305
|
| 9177 |
+
},
|
| 9178 |
+
{
|
| 9179 |
+
"epoch": 0.8531765474440633,
|
| 9180 |
+
"grad_norm": Infinity,
|
| 9181 |
+
"learning_rate": 1.1092886870034291e-05,
|
| 9182 |
+
"loss": 1.5306,
|
| 9183 |
+
"step": 1306
|
| 9184 |
+
},
|
| 9185 |
+
{
|
| 9186 |
+
"epoch": 0.8538298219826882,
|
| 9187 |
+
"grad_norm": Infinity,
|
| 9188 |
+
"learning_rate": 1.0995916706341459e-05,
|
| 9189 |
+
"loss": 1.4277,
|
| 9190 |
+
"step": 1307
|
| 9191 |
+
},
|
| 9192 |
+
{
|
| 9193 |
+
"epoch": 0.854483096521313,
|
| 9194 |
+
"grad_norm": Infinity,
|
| 9195 |
+
"learning_rate": 1.0899347581163221e-05,
|
| 9196 |
+
"loss": 1.6138,
|
| 9197 |
+
"step": 1308
|
| 9198 |
+
},
|
| 9199 |
+
{
|
| 9200 |
+
"epoch": 0.8551363710599379,
|
| 9201 |
+
"grad_norm": Infinity,
|
| 9202 |
+
"learning_rate": 1.0803179929624973e-05,
|
| 9203 |
+
"loss": 1.6152,
|
| 9204 |
+
"step": 1309
|
| 9205 |
+
},
|
| 9206 |
+
{
|
| 9207 |
+
"epoch": 0.8557896455985627,
|
| 9208 |
+
"grad_norm": Infinity,
|
| 9209 |
+
"learning_rate": 1.0707414185043163e-05,
|
| 9210 |
+
"loss": 1.6077,
|
| 9211 |
+
"step": 1310
|
| 9212 |
+
},
|
| 9213 |
+
{
|
| 9214 |
+
"epoch": 0.8564429201371877,
|
| 9215 |
+
"grad_norm": Infinity,
|
| 9216 |
+
"learning_rate": 1.0612050778923276e-05,
|
| 9217 |
+
"loss": 1.6382,
|
| 9218 |
+
"step": 1311
|
| 9219 |
+
},
|
| 9220 |
+
{
|
| 9221 |
+
"epoch": 0.8570961946758126,
|
| 9222 |
+
"grad_norm": Infinity,
|
| 9223 |
+
"learning_rate": 1.0517090140957998e-05,
|
| 9224 |
+
"loss": 1.7534,
|
| 9225 |
+
"step": 1312
|
| 9226 |
+
},
|
| 9227 |
+
{
|
| 9228 |
+
"epoch": 0.8577494692144374,
|
| 9229 |
+
"grad_norm": Infinity,
|
| 9230 |
+
"learning_rate": 1.0422532699025068e-05,
|
| 9231 |
+
"loss": 1.8991,
|
| 9232 |
+
"step": 1313
|
| 9233 |
+
},
|
| 9234 |
+
{
|
| 9235 |
+
"epoch": 0.8584027437530622,
|
| 9236 |
+
"grad_norm": Infinity,
|
| 9237 |
+
"learning_rate": 1.0328378879185641e-05,
|
| 9238 |
+
"loss": 1.9086,
|
| 9239 |
+
"step": 1314
|
| 9240 |
+
},
|
| 9241 |
+
{
|
| 9242 |
+
"epoch": 0.8590560182916871,
|
| 9243 |
+
"grad_norm": Infinity,
|
| 9244 |
+
"learning_rate": 1.0234629105682103e-05,
|
| 9245 |
+
"loss": 2.1443,
|
| 9246 |
+
"step": 1315
|
| 9247 |
+
},
|
| 9248 |
+
{
|
| 9249 |
+
"epoch": 0.8597092928303119,
|
| 9250 |
+
"grad_norm": Infinity,
|
| 9251 |
+
"learning_rate": 1.0141283800936297e-05,
|
| 9252 |
+
"loss": 2.0489,
|
| 9253 |
+
"step": 1316
|
| 9254 |
+
},
|
| 9255 |
+
{
|
| 9256 |
+
"epoch": 0.8603625673689368,
|
| 9257 |
+
"grad_norm": Infinity,
|
| 9258 |
+
"learning_rate": 1.0048343385547676e-05,
|
| 9259 |
+
"loss": 2.0992,
|
| 9260 |
+
"step": 1317
|
| 9261 |
+
},
|
| 9262 |
+
{
|
| 9263 |
+
"epoch": 0.8610158419075616,
|
| 9264 |
+
"grad_norm": Infinity,
|
| 9265 |
+
"learning_rate": 9.955808278291156e-06,
|
| 9266 |
+
"loss": 2.5535,
|
| 9267 |
+
"step": 1318
|
| 9268 |
+
},
|
| 9269 |
+
{
|
| 9270 |
+
"epoch": 0.8616691164461865,
|
| 9271 |
+
"grad_norm": Infinity,
|
| 9272 |
+
"learning_rate": 9.863678896115559e-06,
|
| 9273 |
+
"loss": 2.5644,
|
| 9274 |
+
"step": 1319
|
| 9275 |
+
},
|
| 9276 |
+
{
|
| 9277 |
+
"epoch": 0.8623223909848113,
|
| 9278 |
+
"grad_norm": Infinity,
|
| 9279 |
+
"learning_rate": 9.771955654141496e-06,
|
| 9280 |
+
"loss": 2.6898,
|
| 9281 |
+
"step": 1320
|
| 9282 |
+
},
|
| 9283 |
+
{
|
| 9284 |
+
"epoch": 0.8629756655234362,
|
| 9285 |
+
"grad_norm": Infinity,
|
| 9286 |
+
"learning_rate": 9.68063896565955e-06,
|
| 9287 |
+
"loss": 2.8441,
|
| 9288 |
+
"step": 1321
|
| 9289 |
+
},
|
| 9290 |
+
{
|
| 9291 |
+
"epoch": 0.863628940062061,
|
| 9292 |
+
"grad_norm": Infinity,
|
| 9293 |
+
"learning_rate": 9.589729242128532e-06,
|
| 9294 |
+
"loss": 2.7607,
|
| 9295 |
+
"step": 1322
|
| 9296 |
+
},
|
| 9297 |
+
{
|
| 9298 |
+
"epoch": 0.8642822146006859,
|
| 9299 |
+
"grad_norm": Infinity,
|
| 9300 |
+
"learning_rate": 9.499226893173453e-06,
|
| 9301 |
+
"loss": 3.2205,
|
| 9302 |
+
"step": 1323
|
| 9303 |
+
},
|
| 9304 |
+
{
|
| 9305 |
+
"epoch": 0.8649354891393108,
|
| 9306 |
+
"grad_norm": Infinity,
|
| 9307 |
+
"learning_rate": 9.409132326583758e-06,
|
| 9308 |
+
"loss": 3.5218,
|
| 9309 |
+
"step": 1324
|
| 9310 |
+
},
|
| 9311 |
+
{
|
| 9312 |
+
"epoch": 0.8655887636779357,
|
| 9313 |
+
"grad_norm": Infinity,
|
| 9314 |
+
"learning_rate": 9.319445948311534e-06,
|
| 9315 |
+
"loss": 3.9074,
|
| 9316 |
+
"step": 1325
|
| 9317 |
+
},
|
| 9318 |
+
{
|
| 9319 |
+
"epoch": 0.8662420382165605,
|
| 9320 |
+
"grad_norm": Infinity,
|
| 9321 |
+
"learning_rate": 9.230168162469599e-06,
|
| 9322 |
+
"loss": 1.3831,
|
| 9323 |
+
"step": 1326
|
| 9324 |
+
},
|
| 9325 |
+
{
|
| 9326 |
+
"epoch": 0.8668953127551854,
|
| 9327 |
+
"grad_norm": Infinity,
|
| 9328 |
+
"learning_rate": 9.141299371329704e-06,
|
| 9329 |
+
"loss": 1.446,
|
| 9330 |
+
"step": 1327
|
| 9331 |
+
},
|
| 9332 |
+
{
|
| 9333 |
+
"epoch": 0.8675485872938102,
|
| 9334 |
+
"grad_norm": Infinity,
|
| 9335 |
+
"learning_rate": 9.052839975320836e-06,
|
| 9336 |
+
"loss": 1.3385,
|
| 9337 |
+
"step": 1328
|
| 9338 |
+
},
|
| 9339 |
+
{
|
| 9340 |
+
"epoch": 0.8682018618324351,
|
| 9341 |
+
"grad_norm": Infinity,
|
| 9342 |
+
"learning_rate": 8.964790373027132e-06,
|
| 9343 |
+
"loss": 1.4632,
|
| 9344 |
+
"step": 1329
|
| 9345 |
+
},
|
| 9346 |
+
{
|
| 9347 |
+
"epoch": 0.8688551363710599,
|
| 9348 |
+
"grad_norm": Infinity,
|
| 9349 |
+
"learning_rate": 8.87715096118642e-06,
|
| 9350 |
+
"loss": 1.5529,
|
| 9351 |
+
"step": 1330
|
| 9352 |
+
},
|
| 9353 |
+
{
|
| 9354 |
+
"epoch": 0.8695084109096848,
|
| 9355 |
+
"grad_norm": Infinity,
|
| 9356 |
+
"learning_rate": 8.789922134688244e-06,
|
| 9357 |
+
"loss": 1.572,
|
| 9358 |
+
"step": 1331
|
| 9359 |
+
},
|
| 9360 |
+
{
|
| 9361 |
+
"epoch": 0.8701616854483096,
|
| 9362 |
+
"grad_norm": Infinity,
|
| 9363 |
+
"learning_rate": 8.703104286572028e-06,
|
| 9364 |
+
"loss": 1.5595,
|
| 9365 |
+
"step": 1332
|
| 9366 |
+
},
|
| 9367 |
+
{
|
| 9368 |
+
"epoch": 0.8708149599869345,
|
| 9369 |
+
"grad_norm": Infinity,
|
| 9370 |
+
"learning_rate": 8.616697808025486e-06,
|
| 9371 |
+
"loss": 1.5692,
|
| 9372 |
+
"step": 1333
|
| 9373 |
+
},
|
| 9374 |
+
{
|
| 9375 |
+
"epoch": 0.8714682345255593,
|
| 9376 |
+
"grad_norm": Infinity,
|
| 9377 |
+
"learning_rate": 8.53070308838273e-06,
|
| 9378 |
+
"loss": 1.535,
|
| 9379 |
+
"step": 1334
|
| 9380 |
+
},
|
| 9381 |
+
{
|
| 9382 |
+
"epoch": 0.8721215090641842,
|
| 9383 |
+
"grad_norm": Infinity,
|
| 9384 |
+
"learning_rate": 8.445120515122551e-06,
|
| 9385 |
+
"loss": 1.5121,
|
| 9386 |
+
"step": 1335
|
| 9387 |
+
},
|
| 9388 |
+
{
|
| 9389 |
+
"epoch": 0.8727747836028091,
|
| 9390 |
+
"grad_norm": Infinity,
|
| 9391 |
+
"learning_rate": 8.359950473866663e-06,
|
| 9392 |
+
"loss": 1.6005,
|
| 9393 |
+
"step": 1336
|
| 9394 |
+
},
|
| 9395 |
+
{
|
| 9396 |
+
"epoch": 0.873428058141434,
|
| 9397 |
+
"grad_norm": Infinity,
|
| 9398 |
+
"learning_rate": 8.275193348377974e-06,
|
| 9399 |
+
"loss": 1.7782,
|
| 9400 |
+
"step": 1337
|
| 9401 |
+
},
|
| 9402 |
+
{
|
| 9403 |
+
"epoch": 0.8740813326800588,
|
| 9404 |
+
"grad_norm": Infinity,
|
| 9405 |
+
"learning_rate": 8.190849520558908e-06,
|
| 9406 |
+
"loss": 2.003,
|
| 9407 |
+
"step": 1338
|
| 9408 |
+
},
|
| 9409 |
+
{
|
| 9410 |
+
"epoch": 0.8747346072186837,
|
| 9411 |
+
"grad_norm": Infinity,
|
| 9412 |
+
"learning_rate": 8.106919370449572e-06,
|
| 9413 |
+
"loss": 1.9703,
|
| 9414 |
+
"step": 1339
|
| 9415 |
+
},
|
| 9416 |
+
{
|
| 9417 |
+
"epoch": 0.8753878817573085,
|
| 9418 |
+
"grad_norm": Infinity,
|
| 9419 |
+
"learning_rate": 8.023403276226126e-06,
|
| 9420 |
+
"loss": 2.0316,
|
| 9421 |
+
"step": 1340
|
| 9422 |
+
},
|
| 9423 |
+
{
|
| 9424 |
+
"epoch": 0.8760411562959334,
|
| 9425 |
+
"grad_norm": Infinity,
|
| 9426 |
+
"learning_rate": 7.940301614199074e-06,
|
| 9427 |
+
"loss": 1.9369,
|
| 9428 |
+
"step": 1341
|
| 9429 |
+
},
|
| 9430 |
+
{
|
| 9431 |
+
"epoch": 0.8766944308345582,
|
| 9432 |
+
"grad_norm": Infinity,
|
| 9433 |
+
"learning_rate": 7.857614758811527e-06,
|
| 9434 |
+
"loss": 2.2225,
|
| 9435 |
+
"step": 1342
|
| 9436 |
+
},
|
| 9437 |
+
{
|
| 9438 |
+
"epoch": 0.8773477053731831,
|
| 9439 |
+
"grad_norm": Infinity,
|
| 9440 |
+
"learning_rate": 7.775343082637553e-06,
|
| 9441 |
+
"loss": 2.26,
|
| 9442 |
+
"step": 1343
|
| 9443 |
+
},
|
| 9444 |
+
{
|
| 9445 |
+
"epoch": 0.8780009799118079,
|
| 9446 |
+
"grad_norm": Infinity,
|
| 9447 |
+
"learning_rate": 7.693486956380525e-06,
|
| 9448 |
+
"loss": 2.3674,
|
| 9449 |
+
"step": 1344
|
| 9450 |
+
},
|
| 9451 |
+
{
|
| 9452 |
+
"epoch": 0.8786542544504328,
|
| 9453 |
+
"grad_norm": Infinity,
|
| 9454 |
+
"learning_rate": 7.612046748871327e-06,
|
| 9455 |
+
"loss": 2.9854,
|
| 9456 |
+
"step": 1345
|
| 9457 |
+
},
|
| 9458 |
+
{
|
| 9459 |
+
"epoch": 0.8793075289890576,
|
| 9460 |
+
"grad_norm": Infinity,
|
| 9461 |
+
"learning_rate": 7.531022827066858e-06,
|
| 9462 |
+
"loss": 2.6,
|
| 9463 |
+
"step": 1346
|
| 9464 |
+
},
|
| 9465 |
+
{
|
| 9466 |
+
"epoch": 0.8799608035276825,
|
| 9467 |
+
"grad_norm": Infinity,
|
| 9468 |
+
"learning_rate": 7.450415556048296e-06,
|
| 9469 |
+
"loss": 3.0325,
|
| 9470 |
+
"step": 1347
|
| 9471 |
+
},
|
| 9472 |
+
{
|
| 9473 |
+
"epoch": 0.8806140780663073,
|
| 9474 |
+
"grad_norm": Infinity,
|
| 9475 |
+
"learning_rate": 7.370225299019362e-06,
|
| 9476 |
+
"loss": 3.0668,
|
| 9477 |
+
"step": 1348
|
| 9478 |
+
},
|
| 9479 |
+
{
|
| 9480 |
+
"epoch": 0.8812673526049323,
|
| 9481 |
+
"grad_norm": Infinity,
|
| 9482 |
+
"learning_rate": 7.290452417304916e-06,
|
| 9483 |
+
"loss": 3.2571,
|
| 9484 |
+
"step": 1349
|
| 9485 |
+
},
|
| 9486 |
+
{
|
| 9487 |
+
"epoch": 0.8819206271435571,
|
| 9488 |
+
"grad_norm": Infinity,
|
| 9489 |
+
"learning_rate": 7.211097270349066e-06,
|
| 9490 |
+
"loss": 3.7934,
|
| 9491 |
+
"step": 1350
|
| 9492 |
+
},
|
| 9493 |
+
{
|
| 9494 |
+
"epoch": 0.882573901682182,
|
| 9495 |
+
"grad_norm": Infinity,
|
| 9496 |
+
"learning_rate": 7.132160215713757e-06,
|
| 9497 |
+
"loss": 1.2629,
|
| 9498 |
+
"step": 1351
|
| 9499 |
+
},
|
| 9500 |
+
{
|
| 9501 |
+
"epoch": 0.8832271762208068,
|
| 9502 |
+
"grad_norm": Infinity,
|
| 9503 |
+
"learning_rate": 7.053641609077033e-06,
|
| 9504 |
+
"loss": 1.4117,
|
| 9505 |
+
"step": 1352
|
| 9506 |
+
},
|
| 9507 |
+
{
|
| 9508 |
+
"epoch": 0.8838804507594317,
|
| 9509 |
+
"grad_norm": Infinity,
|
| 9510 |
+
"learning_rate": 6.975541804231478e-06,
|
| 9511 |
+
"loss": 1.3355,
|
| 9512 |
+
"step": 1353
|
| 9513 |
+
},
|
| 9514 |
+
{
|
| 9515 |
+
"epoch": 0.8845337252980565,
|
| 9516 |
+
"grad_norm": Infinity,
|
| 9517 |
+
"learning_rate": 6.897861153082618e-06,
|
| 9518 |
+
"loss": 1.4311,
|
| 9519 |
+
"step": 1354
|
| 9520 |
+
},
|
| 9521 |
+
{
|
| 9522 |
+
"epoch": 0.8851869998366814,
|
| 9523 |
+
"grad_norm": Infinity,
|
| 9524 |
+
"learning_rate": 6.820600005647382e-06,
|
| 9525 |
+
"loss": 1.5243,
|
| 9526 |
+
"step": 1355
|
| 9527 |
+
},
|
| 9528 |
+
{
|
| 9529 |
+
"epoch": 0.8858402743753062,
|
| 9530 |
+
"grad_norm": Infinity,
|
| 9531 |
+
"learning_rate": 6.743758710052439e-06,
|
| 9532 |
+
"loss": 1.4378,
|
| 9533 |
+
"step": 1356
|
| 9534 |
+
},
|
| 9535 |
+
{
|
| 9536 |
+
"epoch": 0.8864935489139311,
|
| 9537 |
+
"grad_norm": Infinity,
|
| 9538 |
+
"learning_rate": 6.6673376125326645e-06,
|
| 9539 |
+
"loss": 1.4713,
|
| 9540 |
+
"step": 1357
|
| 9541 |
+
},
|
| 9542 |
+
{
|
| 9543 |
+
"epoch": 0.8871468234525559,
|
| 9544 |
+
"grad_norm": Infinity,
|
| 9545 |
+
"learning_rate": 6.591337057429603e-06,
|
| 9546 |
+
"loss": 1.5855,
|
| 9547 |
+
"step": 1358
|
| 9548 |
+
},
|
| 9549 |
+
{
|
| 9550 |
+
"epoch": 0.8878000979911808,
|
| 9551 |
+
"grad_norm": Infinity,
|
| 9552 |
+
"learning_rate": 6.515757387189902e-06,
|
| 9553 |
+
"loss": 1.5555,
|
| 9554 |
+
"step": 1359
|
| 9555 |
+
},
|
| 9556 |
+
{
|
| 9557 |
+
"epoch": 0.8884533725298056,
|
| 9558 |
+
"grad_norm": Infinity,
|
| 9559 |
+
"learning_rate": 6.440598942363796e-06,
|
| 9560 |
+
"loss": 1.573,
|
| 9561 |
+
"step": 1360
|
| 9562 |
+
},
|
| 9563 |
+
{
|
| 9564 |
+
"epoch": 0.8891066470684305,
|
| 9565 |
+
"grad_norm": Infinity,
|
| 9566 |
+
"learning_rate": 6.365862061603456e-06,
|
| 9567 |
+
"loss": 1.7148,
|
| 9568 |
+
"step": 1361
|
| 9569 |
+
},
|
| 9570 |
+
{
|
| 9571 |
+
"epoch": 0.8897599216070554,
|
| 9572 |
+
"grad_norm": Infinity,
|
| 9573 |
+
"learning_rate": 6.291547081661631e-06,
|
| 9574 |
+
"loss": 1.6704,
|
| 9575 |
+
"step": 1362
|
| 9576 |
+
},
|
| 9577 |
+
{
|
| 9578 |
+
"epoch": 0.8904131961456803,
|
| 9579 |
+
"grad_norm": Infinity,
|
| 9580 |
+
"learning_rate": 6.217654337390078e-06,
|
| 9581 |
+
"loss": 1.8345,
|
| 9582 |
+
"step": 1363
|
| 9583 |
+
},
|
| 9584 |
+
{
|
| 9585 |
+
"epoch": 0.8910664706843051,
|
| 9586 |
+
"grad_norm": Infinity,
|
| 9587 |
+
"learning_rate": 6.144184161737887e-06,
|
| 9588 |
+
"loss": 1.8895,
|
| 9589 |
+
"step": 1364
|
| 9590 |
+
},
|
| 9591 |
+
{
|
| 9592 |
+
"epoch": 0.89171974522293,
|
| 9593 |
+
"grad_norm": Infinity,
|
| 9594 |
+
"learning_rate": 6.071136885750272e-06,
|
| 9595 |
+
"loss": 1.9469,
|
| 9596 |
+
"step": 1365
|
| 9597 |
+
},
|
| 9598 |
+
{
|
| 9599 |
+
"epoch": 0.8923730197615548,
|
| 9600 |
+
"grad_norm": Infinity,
|
| 9601 |
+
"learning_rate": 5.99851283856685e-06,
|
| 9602 |
+
"loss": 2.2034,
|
| 9603 |
+
"step": 1366
|
| 9604 |
+
},
|
| 9605 |
+
{
|
| 9606 |
+
"epoch": 0.8930262943001797,
|
| 9607 |
+
"grad_norm": Infinity,
|
| 9608 |
+
"learning_rate": 5.926312347420238e-06,
|
| 9609 |
+
"loss": 2.2942,
|
| 9610 |
+
"step": 1367
|
| 9611 |
+
},
|
| 9612 |
+
{
|
| 9613 |
+
"epoch": 0.8936795688388045,
|
| 9614 |
+
"grad_norm": Infinity,
|
| 9615 |
+
"learning_rate": 5.854535737634581e-06,
|
| 9616 |
+
"loss": 2.514,
|
| 9617 |
+
"step": 1368
|
| 9618 |
+
},
|
| 9619 |
+
{
|
| 9620 |
+
"epoch": 0.8943328433774294,
|
| 9621 |
+
"grad_norm": Infinity,
|
| 9622 |
+
"learning_rate": 5.783183332624098e-06,
|
| 9623 |
+
"loss": 2.5924,
|
| 9624 |
+
"step": 1369
|
| 9625 |
+
},
|
| 9626 |
+
{
|
| 9627 |
+
"epoch": 0.8949861179160542,
|
| 9628 |
+
"grad_norm": Infinity,
|
| 9629 |
+
"learning_rate": 5.71225545389158e-06,
|
| 9630 |
+
"loss": 2.6971,
|
| 9631 |
+
"step": 1370
|
| 9632 |
+
},
|
| 9633 |
+
{
|
| 9634 |
+
"epoch": 0.895639392454679,
|
| 9635 |
+
"grad_norm": Infinity,
|
| 9636 |
+
"learning_rate": 5.641752421027014e-06,
|
| 9637 |
+
"loss": 2.4167,
|
| 9638 |
+
"step": 1371
|
| 9639 |
+
},
|
| 9640 |
+
{
|
| 9641 |
+
"epoch": 0.8962926669933039,
|
| 9642 |
+
"grad_norm": Infinity,
|
| 9643 |
+
"learning_rate": 5.571674551706041e-06,
|
| 9644 |
+
"loss": 3.0307,
|
| 9645 |
+
"step": 1372
|
| 9646 |
+
},
|
| 9647 |
+
{
|
| 9648 |
+
"epoch": 0.8969459415319287,
|
| 9649 |
+
"grad_norm": Infinity,
|
| 9650 |
+
"learning_rate": 5.5020221616886025e-06,
|
| 9651 |
+
"loss": 3.0822,
|
| 9652 |
+
"step": 1373
|
| 9653 |
+
},
|
| 9654 |
+
{
|
| 9655 |
+
"epoch": 0.8975992160705536,
|
| 9656 |
+
"grad_norm": Infinity,
|
| 9657 |
+
"learning_rate": 5.432795564817539e-06,
|
| 9658 |
+
"loss": 3.6224,
|
| 9659 |
+
"step": 1374
|
| 9660 |
+
},
|
| 9661 |
+
{
|
| 9662 |
+
"epoch": 0.8982524906091786,
|
| 9663 |
+
"grad_norm": Infinity,
|
| 9664 |
+
"learning_rate": 5.363995073017047e-06,
|
| 9665 |
+
"loss": 3.9485,
|
| 9666 |
+
"step": 1375
|
| 9667 |
+
},
|
| 9668 |
+
{
|
| 9669 |
+
"epoch": 0.8989057651478034,
|
| 9670 |
+
"grad_norm": Infinity,
|
| 9671 |
+
"learning_rate": 5.295620996291451e-06,
|
| 9672 |
+
"loss": 1.369,
|
| 9673 |
+
"step": 1376
|
| 9674 |
+
},
|
| 9675 |
+
{
|
| 9676 |
+
"epoch": 0.8995590396864283,
|
| 9677 |
+
"grad_norm": Infinity,
|
| 9678 |
+
"learning_rate": 5.227673642723651e-06,
|
| 9679 |
+
"loss": 1.3912,
|
| 9680 |
+
"step": 1377
|
| 9681 |
+
},
|
| 9682 |
+
{
|
| 9683 |
+
"epoch": 0.9002123142250531,
|
| 9684 |
+
"grad_norm": Infinity,
|
| 9685 |
+
"learning_rate": 5.160153318473815e-06,
|
| 9686 |
+
"loss": 1.4591,
|
| 9687 |
+
"step": 1378
|
| 9688 |
+
},
|
| 9689 |
+
{
|
| 9690 |
+
"epoch": 0.900865588763678,
|
| 9691 |
+
"grad_norm": Infinity,
|
| 9692 |
+
"learning_rate": 5.093060327778043e-06,
|
| 9693 |
+
"loss": 1.491,
|
| 9694 |
+
"step": 1379
|
| 9695 |
+
},
|
| 9696 |
+
{
|
| 9697 |
+
"epoch": 0.9015188633023028,
|
| 9698 |
+
"grad_norm": Infinity,
|
| 9699 |
+
"learning_rate": 5.026394972946813e-06,
|
| 9700 |
+
"loss": 1.472,
|
| 9701 |
+
"step": 1380
|
| 9702 |
+
},
|
| 9703 |
+
{
|
| 9704 |
+
"epoch": 0.9021721378409276,
|
| 9705 |
+
"grad_norm": Infinity,
|
| 9706 |
+
"learning_rate": 4.9601575543638535e-06,
|
| 9707 |
+
"loss": 1.4614,
|
| 9708 |
+
"step": 1381
|
| 9709 |
+
},
|
| 9710 |
+
{
|
| 9711 |
+
"epoch": 0.9028254123795525,
|
| 9712 |
+
"grad_norm": Infinity,
|
| 9713 |
+
"learning_rate": 4.8943483704846475e-06,
|
| 9714 |
+
"loss": 1.5487,
|
| 9715 |
+
"step": 1382
|
| 9716 |
+
},
|
| 9717 |
+
{
|
| 9718 |
+
"epoch": 0.9034786869181773,
|
| 9719 |
+
"grad_norm": Infinity,
|
| 9720 |
+
"learning_rate": 4.828967717835087e-06,
|
| 9721 |
+
"loss": 1.546,
|
| 9722 |
+
"step": 1383
|
| 9723 |
+
},
|
| 9724 |
+
{
|
| 9725 |
+
"epoch": 0.9041319614568022,
|
| 9726 |
+
"grad_norm": Infinity,
|
| 9727 |
+
"learning_rate": 4.76401589101021e-06,
|
| 9728 |
+
"loss": 1.4852,
|
| 9729 |
+
"step": 1384
|
| 9730 |
+
},
|
| 9731 |
+
{
|
| 9732 |
+
"epoch": 0.904785235995427,
|
| 9733 |
+
"grad_norm": Infinity,
|
| 9734 |
+
"learning_rate": 4.69949318267281e-06,
|
| 9735 |
+
"loss": 1.742,
|
| 9736 |
+
"step": 1385
|
| 9737 |
+
},
|
| 9738 |
+
{
|
| 9739 |
+
"epoch": 0.9054385105340519,
|
| 9740 |
+
"grad_norm": Infinity,
|
| 9741 |
+
"learning_rate": 4.635399883552128e-06,
|
| 9742 |
+
"loss": 1.6791,
|
| 9743 |
+
"step": 1386
|
| 9744 |
+
},
|
| 9745 |
+
{
|
| 9746 |
+
"epoch": 0.9060917850726768,
|
| 9747 |
+
"grad_norm": Infinity,
|
| 9748 |
+
"learning_rate": 4.571736282442607e-06,
|
| 9749 |
+
"loss": 1.7485,
|
| 9750 |
+
"step": 1387
|
| 9751 |
+
},
|
| 9752 |
+
{
|
| 9753 |
+
"epoch": 0.9067450596113017,
|
| 9754 |
+
"grad_norm": Infinity,
|
| 9755 |
+
"learning_rate": 4.508502666202474e-06,
|
| 9756 |
+
"loss": 1.9477,
|
| 9757 |
+
"step": 1388
|
| 9758 |
+
},
|
| 9759 |
+
{
|
| 9760 |
+
"epoch": 0.9073983341499265,
|
| 9761 |
+
"grad_norm": Infinity,
|
| 9762 |
+
"learning_rate": 4.445699319752539e-06,
|
| 9763 |
+
"loss": 1.8895,
|
| 9764 |
+
"step": 1389
|
| 9765 |
+
},
|
| 9766 |
+
{
|
| 9767 |
+
"epoch": 0.9080516086885514,
|
| 9768 |
+
"grad_norm": Infinity,
|
| 9769 |
+
"learning_rate": 4.383326526074916e-06,
|
| 9770 |
+
"loss": 1.9285,
|
| 9771 |
+
"step": 1390
|
| 9772 |
+
},
|
| 9773 |
+
{
|
| 9774 |
+
"epoch": 0.9087048832271762,
|
| 9775 |
+
"grad_norm": Infinity,
|
| 9776 |
+
"learning_rate": 4.32138456621164e-06,
|
| 9777 |
+
"loss": 2.2509,
|
| 9778 |
+
"step": 1391
|
| 9779 |
+
},
|
| 9780 |
+
{
|
| 9781 |
+
"epoch": 0.9093581577658011,
|
| 9782 |
+
"grad_norm": Infinity,
|
| 9783 |
+
"learning_rate": 4.2598737192635405e-06,
|
| 9784 |
+
"loss": 2.3064,
|
| 9785 |
+
"step": 1392
|
| 9786 |
+
},
|
| 9787 |
+
{
|
| 9788 |
+
"epoch": 0.9100114323044259,
|
| 9789 |
+
"grad_norm": Infinity,
|
| 9790 |
+
"learning_rate": 4.198794262388905e-06,
|
| 9791 |
+
"loss": 2.6755,
|
| 9792 |
+
"step": 1393
|
| 9793 |
+
},
|
| 9794 |
+
{
|
| 9795 |
+
"epoch": 0.9106647068430508,
|
| 9796 |
+
"grad_norm": Infinity,
|
| 9797 |
+
"learning_rate": 4.138146470802218e-06,
|
| 9798 |
+
"loss": 2.3408,
|
| 9799 |
+
"step": 1394
|
| 9800 |
+
},
|
| 9801 |
+
{
|
| 9802 |
+
"epoch": 0.9113179813816756,
|
| 9803 |
+
"grad_norm": Infinity,
|
| 9804 |
+
"learning_rate": 4.077930617773007e-06,
|
| 9805 |
+
"loss": 2.8286,
|
| 9806 |
+
"step": 1395
|
| 9807 |
+
},
|
| 9808 |
+
{
|
| 9809 |
+
"epoch": 0.9119712559203005,
|
| 9810 |
+
"grad_norm": Infinity,
|
| 9811 |
+
"learning_rate": 4.018146974624448e-06,
|
| 9812 |
+
"loss": 2.7035,
|
| 9813 |
+
"step": 1396
|
| 9814 |
+
},
|
| 9815 |
+
{
|
| 9816 |
+
"epoch": 0.9126245304589253,
|
| 9817 |
+
"grad_norm": Infinity,
|
| 9818 |
+
"learning_rate": 3.958795810732363e-06,
|
| 9819 |
+
"loss": 2.9971,
|
| 9820 |
+
"step": 1397
|
| 9821 |
+
},
|
| 9822 |
+
{
|
| 9823 |
+
"epoch": 0.9132778049975502,
|
| 9824 |
+
"grad_norm": Infinity,
|
| 9825 |
+
"learning_rate": 3.899877393523821e-06,
|
| 9826 |
+
"loss": 3.2009,
|
| 9827 |
+
"step": 1398
|
| 9828 |
+
},
|
| 9829 |
+
{
|
| 9830 |
+
"epoch": 0.913931079536175,
|
| 9831 |
+
"grad_norm": Infinity,
|
| 9832 |
+
"learning_rate": 3.841391988476018e-06,
|
| 9833 |
+
"loss": 3.1459,
|
| 9834 |
+
"step": 1399
|
| 9835 |
+
},
|
| 9836 |
+
{
|
| 9837 |
+
"epoch": 0.9145843540748,
|
| 9838 |
+
"grad_norm": Infinity,
|
| 9839 |
+
"learning_rate": 3.783339859115065e-06,
|
| 9840 |
+
"loss": 3.91,
|
| 9841 |
+
"step": 1400
|
| 9842 |
+
},
|
| 9843 |
+
{
|
| 9844 |
+
"epoch": 0.9152376286134248,
|
| 9845 |
+
"grad_norm": Infinity,
|
| 9846 |
+
"learning_rate": 3.7257212670148035e-06,
|
| 9847 |
+
"loss": 1.3089,
|
| 9848 |
+
"step": 1401
|
| 9849 |
+
},
|
| 9850 |
+
{
|
| 9851 |
+
"epoch": 0.9158909031520497,
|
| 9852 |
+
"grad_norm": Infinity,
|
| 9853 |
+
"learning_rate": 3.6685364717956138e-06,
|
| 9854 |
+
"loss": 1.3872,
|
| 9855 |
+
"step": 1402
|
| 9856 |
+
},
|
| 9857 |
+
{
|
| 9858 |
+
"epoch": 0.9165441776906745,
|
| 9859 |
+
"grad_norm": Infinity,
|
| 9860 |
+
"learning_rate": 3.611785731123274e-06,
|
| 9861 |
+
"loss": 1.3304,
|
| 9862 |
+
"step": 1403
|
| 9863 |
+
},
|
| 9864 |
+
{
|
| 9865 |
+
"epoch": 0.9171974522292994,
|
| 9866 |
+
"grad_norm": Infinity,
|
| 9867 |
+
"learning_rate": 3.555469300707759e-06,
|
| 9868 |
+
"loss": 1.3995,
|
| 9869 |
+
"step": 1404
|
| 9870 |
+
},
|
| 9871 |
+
{
|
| 9872 |
+
"epoch": 0.9178507267679242,
|
| 9873 |
+
"grad_norm": Infinity,
|
| 9874 |
+
"learning_rate": 3.4995874343021094e-06,
|
| 9875 |
+
"loss": 1.4048,
|
| 9876 |
+
"step": 1405
|
| 9877 |
+
},
|
| 9878 |
+
{
|
| 9879 |
+
"epoch": 0.9185040013065491,
|
| 9880 |
+
"grad_norm": Infinity,
|
| 9881 |
+
"learning_rate": 3.444140383701333e-06,
|
| 9882 |
+
"loss": 1.4064,
|
| 9883 |
+
"step": 1406
|
| 9884 |
+
},
|
| 9885 |
+
{
|
| 9886 |
+
"epoch": 0.9191572758451739,
|
| 9887 |
+
"grad_norm": Infinity,
|
| 9888 |
+
"learning_rate": 3.389128398741148e-06,
|
| 9889 |
+
"loss": 1.596,
|
| 9890 |
+
"step": 1407
|
| 9891 |
+
},
|
| 9892 |
+
{
|
| 9893 |
+
"epoch": 0.9198105503837988,
|
| 9894 |
+
"grad_norm": Infinity,
|
| 9895 |
+
"learning_rate": 3.3345517272970084e-06,
|
| 9896 |
+
"loss": 1.5013,
|
| 9897 |
+
"step": 1408
|
| 9898 |
+
},
|
| 9899 |
+
{
|
| 9900 |
+
"epoch": 0.9204638249224236,
|
| 9901 |
+
"grad_norm": Infinity,
|
| 9902 |
+
"learning_rate": 3.2804106152828582e-06,
|
| 9903 |
+
"loss": 1.7069,
|
| 9904 |
+
"step": 1409
|
| 9905 |
+
},
|
| 9906 |
+
{
|
| 9907 |
+
"epoch": 0.9211170994610485,
|
| 9908 |
+
"grad_norm": Infinity,
|
| 9909 |
+
"learning_rate": 3.226705306650113e-06,
|
| 9910 |
+
"loss": 1.6444,
|
| 9911 |
+
"step": 1410
|
| 9912 |
+
},
|
| 9913 |
+
{
|
| 9914 |
+
"epoch": 0.9217703739996733,
|
| 9915 |
+
"grad_norm": Infinity,
|
| 9916 |
+
"learning_rate": 3.173436043386535e-06,
|
| 9917 |
+
"loss": 1.5149,
|
| 9918 |
+
"step": 1411
|
| 9919 |
+
},
|
| 9920 |
+
{
|
| 9921 |
+
"epoch": 0.9224236485382982,
|
| 9922 |
+
"grad_norm": Infinity,
|
| 9923 |
+
"learning_rate": 3.120603065515071e-06,
|
| 9924 |
+
"loss": 1.7224,
|
| 9925 |
+
"step": 1412
|
| 9926 |
+
},
|
| 9927 |
+
{
|
| 9928 |
+
"epoch": 0.9230769230769231,
|
| 9929 |
+
"grad_norm": Infinity,
|
| 9930 |
+
"learning_rate": 3.068206611092905e-06,
|
| 9931 |
+
"loss": 1.7857,
|
| 9932 |
+
"step": 1413
|
| 9933 |
+
},
|
| 9934 |
+
{
|
| 9935 |
+
"epoch": 0.923730197615548,
|
| 9936 |
+
"grad_norm": Infinity,
|
| 9937 |
+
"learning_rate": 3.016246916210297e-06,
|
| 9938 |
+
"loss": 1.8834,
|
| 9939 |
+
"step": 1414
|
| 9940 |
+
},
|
| 9941 |
+
{
|
| 9942 |
+
"epoch": 0.9243834721541728,
|
| 9943 |
+
"grad_norm": Infinity,
|
| 9944 |
+
"learning_rate": 2.9647242149895006e-06,
|
| 9945 |
+
"loss": 1.9377,
|
| 9946 |
+
"step": 1415
|
| 9947 |
+
},
|
| 9948 |
+
{
|
| 9949 |
+
"epoch": 0.9250367466927977,
|
| 9950 |
+
"grad_norm": Infinity,
|
| 9951 |
+
"learning_rate": 2.9136387395837683e-06,
|
| 9952 |
+
"loss": 2.1515,
|
| 9953 |
+
"step": 1416
|
| 9954 |
+
},
|
| 9955 |
+
{
|
| 9956 |
+
"epoch": 0.9256900212314225,
|
| 9957 |
+
"grad_norm": Infinity,
|
| 9958 |
+
"learning_rate": 2.8629907201763283e-06,
|
| 9959 |
+
"loss": 2.3141,
|
| 9960 |
+
"step": 1417
|
| 9961 |
+
},
|
| 9962 |
+
{
|
| 9963 |
+
"epoch": 0.9263432957700474,
|
| 9964 |
+
"grad_norm": Infinity,
|
| 9965 |
+
"learning_rate": 2.8127803849791968e-06,
|
| 9966 |
+
"loss": 2.3314,
|
| 9967 |
+
"step": 1418
|
| 9968 |
+
},
|
| 9969 |
+
{
|
| 9970 |
+
"epoch": 0.9269965703086722,
|
| 9971 |
+
"grad_norm": Infinity,
|
| 9972 |
+
"learning_rate": 2.7630079602323442e-06,
|
| 9973 |
+
"loss": 2.3136,
|
| 9974 |
+
"step": 1419
|
| 9975 |
+
},
|
| 9976 |
+
{
|
| 9977 |
+
"epoch": 0.9276498448472971,
|
| 9978 |
+
"grad_norm": Infinity,
|
| 9979 |
+
"learning_rate": 2.7136736702025433e-06,
|
| 9980 |
+
"loss": 2.6563,
|
| 9981 |
+
"step": 1420
|
| 9982 |
+
},
|
| 9983 |
+
{
|
| 9984 |
+
"epoch": 0.9283031193859219,
|
| 9985 |
+
"grad_norm": Infinity,
|
| 9986 |
+
"learning_rate": 2.6647777371824e-06,
|
| 9987 |
+
"loss": 2.7181,
|
| 9988 |
+
"step": 1421
|
| 9989 |
+
},
|
| 9990 |
+
{
|
| 9991 |
+
"epoch": 0.9289563939245468,
|
| 9992 |
+
"grad_norm": Infinity,
|
| 9993 |
+
"learning_rate": 2.616320381489401e-06,
|
| 9994 |
+
"loss": 3.1091,
|
| 9995 |
+
"step": 1422
|
| 9996 |
+
},
|
| 9997 |
+
{
|
| 9998 |
+
"epoch": 0.9296096684631716,
|
| 9999 |
+
"grad_norm": Infinity,
|
| 10000 |
+
"learning_rate": 2.5683018214647693e-06,
|
| 10001 |
+
"loss": 3.1406,
|
| 10002 |
+
"step": 1423
|
| 10003 |
+
},
|
| 10004 |
+
{
|
| 10005 |
+
"epoch": 0.9302629430017965,
|
| 10006 |
+
"grad_norm": Infinity,
|
| 10007 |
+
"learning_rate": 2.520722273472698e-06,
|
| 10008 |
+
"loss": 3.4426,
|
| 10009 |
+
"step": 1424
|
| 10010 |
+
},
|
| 10011 |
+
{
|
| 10012 |
+
"epoch": 0.9309162175404213,
|
| 10013 |
+
"grad_norm": Infinity,
|
| 10014 |
+
"learning_rate": 2.473581951899184e-06,
|
| 10015 |
+
"loss": 4.6078,
|
| 10016 |
+
"step": 1425
|
| 10017 |
+
},
|
| 10018 |
+
{
|
| 10019 |
+
"epoch": 0.9315694920790463,
|
| 10020 |
+
"grad_norm": Infinity,
|
| 10021 |
+
"learning_rate": 2.4268810691511546e-06,
|
| 10022 |
+
"loss": 1.3089,
|
| 10023 |
+
"step": 1426
|
| 10024 |
+
},
|
| 10025 |
+
{
|
| 10026 |
+
"epoch": 0.9322227666176711,
|
| 10027 |
+
"grad_norm": Infinity,
|
| 10028 |
+
"learning_rate": 2.3806198356555287e-06,
|
| 10029 |
+
"loss": 1.4224,
|
| 10030 |
+
"step": 1427
|
| 10031 |
+
},
|
| 10032 |
+
{
|
| 10033 |
+
"epoch": 0.932876041156296,
|
| 10034 |
+
"grad_norm": Infinity,
|
| 10035 |
+
"learning_rate": 2.3347984598581783e-06,
|
| 10036 |
+
"loss": 1.4428,
|
| 10037 |
+
"step": 1428
|
| 10038 |
+
},
|
| 10039 |
+
{
|
| 10040 |
+
"epoch": 0.9335293156949208,
|
| 10041 |
+
"grad_norm": Infinity,
|
| 10042 |
+
"learning_rate": 2.289417148223094e-06,
|
| 10043 |
+
"loss": 1.4401,
|
| 10044 |
+
"step": 1429
|
| 10045 |
+
},
|
| 10046 |
+
{
|
| 10047 |
+
"epoch": 0.9341825902335457,
|
| 10048 |
+
"grad_norm": Infinity,
|
| 10049 |
+
"learning_rate": 2.2444761052313856e-06,
|
| 10050 |
+
"loss": 1.4837,
|
| 10051 |
+
"step": 1430
|
| 10052 |
+
},
|
| 10053 |
+
{
|
| 10054 |
+
"epoch": 0.9348358647721705,
|
| 10055 |
+
"grad_norm": Infinity,
|
| 10056 |
+
"learning_rate": 2.1999755333803718e-06,
|
| 10057 |
+
"loss": 1.5233,
|
| 10058 |
+
"step": 1431
|
| 10059 |
+
},
|
| 10060 |
+
{
|
| 10061 |
+
"epoch": 0.9354891393107954,
|
| 10062 |
+
"grad_norm": Infinity,
|
| 10063 |
+
"learning_rate": 2.1559156331826926e-06,
|
| 10064 |
+
"loss": 1.5427,
|
| 10065 |
+
"step": 1432
|
| 10066 |
+
},
|
| 10067 |
+
{
|
| 10068 |
+
"epoch": 0.9361424138494202,
|
| 10069 |
+
"grad_norm": Infinity,
|
| 10070 |
+
"learning_rate": 2.1122966031653977e-06,
|
| 10071 |
+
"loss": 1.5739,
|
| 10072 |
+
"step": 1433
|
| 10073 |
+
},
|
| 10074 |
+
{
|
| 10075 |
+
"epoch": 0.9367956883880451,
|
| 10076 |
+
"grad_norm": Infinity,
|
| 10077 |
+
"learning_rate": 2.069118639868994e-06,
|
| 10078 |
+
"loss": 1.525,
|
| 10079 |
+
"step": 1434
|
| 10080 |
+
},
|
| 10081 |
+
{
|
| 10082 |
+
"epoch": 0.9374489629266699,
|
| 10083 |
+
"grad_norm": Infinity,
|
| 10084 |
+
"learning_rate": 2.0263819378466884e-06,
|
| 10085 |
+
"loss": 1.7342,
|
| 10086 |
+
"step": 1435
|
| 10087 |
+
},
|
| 10088 |
+
{
|
| 10089 |
+
"epoch": 0.9381022374652948,
|
| 10090 |
+
"grad_norm": Infinity,
|
| 10091 |
+
"learning_rate": 1.984086689663378e-06,
|
| 10092 |
+
"loss": 1.9,
|
| 10093 |
+
"step": 1436
|
| 10094 |
+
},
|
| 10095 |
+
{
|
| 10096 |
+
"epoch": 0.9387555120039196,
|
| 10097 |
+
"grad_norm": Infinity,
|
| 10098 |
+
"learning_rate": 1.942233085894851e-06,
|
| 10099 |
+
"loss": 1.7909,
|
| 10100 |
+
"step": 1437
|
| 10101 |
+
},
|
| 10102 |
+
{
|
| 10103 |
+
"epoch": 0.9394087865425444,
|
| 10104 |
+
"grad_norm": Infinity,
|
| 10105 |
+
"learning_rate": 1.9008213151269328e-06,
|
| 10106 |
+
"loss": 1.8013,
|
| 10107 |
+
"step": 1438
|
| 10108 |
+
},
|
| 10109 |
+
{
|
| 10110 |
+
"epoch": 0.9400620610811694,
|
| 10111 |
+
"grad_norm": Infinity,
|
| 10112 |
+
"learning_rate": 1.8598515639545622e-06,
|
| 10113 |
+
"loss": 1.8483,
|
| 10114 |
+
"step": 1439
|
| 10115 |
+
},
|
| 10116 |
+
{
|
| 10117 |
+
"epoch": 0.9407153356197943,
|
| 10118 |
+
"grad_norm": Infinity,
|
| 10119 |
+
"learning_rate": 1.8193240169810943e-06,
|
| 10120 |
+
"loss": 2.0898,
|
| 10121 |
+
"step": 1440
|
| 10122 |
+
},
|
| 10123 |
+
{
|
| 10124 |
+
"epoch": 0.9413686101584191,
|
| 10125 |
+
"grad_norm": Infinity,
|
| 10126 |
+
"learning_rate": 1.7792388568173002e-06,
|
| 10127 |
+
"loss": 2.1185,
|
| 10128 |
+
"step": 1441
|
| 10129 |
+
},
|
| 10130 |
+
{
|
| 10131 |
+
"epoch": 0.942021884697044,
|
| 10132 |
+
"grad_norm": Infinity,
|
| 10133 |
+
"learning_rate": 1.7395962640806674e-06,
|
| 10134 |
+
"loss": 2.164,
|
| 10135 |
+
"step": 1442
|
| 10136 |
+
},
|
| 10137 |
+
{
|
| 10138 |
+
"epoch": 0.9426751592356688,
|
| 10139 |
+
"grad_norm": Infinity,
|
| 10140 |
+
"learning_rate": 1.7003964173945563e-06,
|
| 10141 |
+
"loss": 2.3025,
|
| 10142 |
+
"step": 1443
|
| 10143 |
+
},
|
| 10144 |
+
{
|
| 10145 |
+
"epoch": 0.9433284337742937,
|
| 10146 |
+
"grad_norm": Infinity,
|
| 10147 |
+
"learning_rate": 1.6616394933873235e-06,
|
| 10148 |
+
"loss": 2.4831,
|
| 10149 |
+
"step": 1444
|
| 10150 |
+
},
|
| 10151 |
+
{
|
| 10152 |
+
"epoch": 0.9439817083129185,
|
| 10153 |
+
"grad_norm": Infinity,
|
| 10154 |
+
"learning_rate": 1.623325666691644e-06,
|
| 10155 |
+
"loss": 2.6088,
|
| 10156 |
+
"step": 1445
|
| 10157 |
+
},
|
| 10158 |
+
{
|
| 10159 |
+
"epoch": 0.9446349828515433,
|
| 10160 |
+
"grad_norm": Infinity,
|
| 10161 |
+
"learning_rate": 1.5854551099436455e-06,
|
| 10162 |
+
"loss": 2.8619,
|
| 10163 |
+
"step": 1446
|
| 10164 |
+
},
|
| 10165 |
+
{
|
| 10166 |
+
"epoch": 0.9452882573901682,
|
| 10167 |
+
"grad_norm": Infinity,
|
| 10168 |
+
"learning_rate": 1.5480279937821418e-06,
|
| 10169 |
+
"loss": 2.9166,
|
| 10170 |
+
"step": 1447
|
| 10171 |
+
},
|
| 10172 |
+
{
|
| 10173 |
+
"epoch": 0.945941531928793,
|
| 10174 |
+
"grad_norm": Infinity,
|
| 10175 |
+
"learning_rate": 1.511044486847879e-06,
|
| 10176 |
+
"loss": 3.3526,
|
| 10177 |
+
"step": 1448
|
| 10178 |
+
},
|
| 10179 |
+
{
|
| 10180 |
+
"epoch": 0.9465948064674179,
|
| 10181 |
+
"grad_norm": Infinity,
|
| 10182 |
+
"learning_rate": 1.4745047557827796e-06,
|
| 10183 |
+
"loss": 3.6262,
|
| 10184 |
+
"step": 1449
|
| 10185 |
+
},
|
| 10186 |
+
{
|
| 10187 |
+
"epoch": 0.9472480810060427,
|
| 10188 |
+
"grad_norm": Infinity,
|
| 10189 |
+
"learning_rate": 1.4384089652291543e-06,
|
| 10190 |
+
"loss": 4.5926,
|
| 10191 |
+
"step": 1450
|
| 10192 |
+
},
|
| 10193 |
+
{
|
| 10194 |
+
"epoch": 0.9479013555446677,
|
| 10195 |
+
"grad_norm": Infinity,
|
| 10196 |
+
"learning_rate": 1.4027572778290255e-06,
|
| 10197 |
+
"loss": 1.2628,
|
| 10198 |
+
"step": 1451
|
| 10199 |
+
},
|
| 10200 |
+
{
|
| 10201 |
+
"epoch": 0.9485546300832925,
|
| 10202 |
+
"grad_norm": Infinity,
|
| 10203 |
+
"learning_rate": 1.3675498542233156e-06,
|
| 10204 |
+
"loss": 1.3955,
|
| 10205 |
+
"step": 1452
|
| 10206 |
+
},
|
| 10207 |
+
{
|
| 10208 |
+
"epoch": 0.9492079046219174,
|
| 10209 |
+
"grad_norm": Infinity,
|
| 10210 |
+
"learning_rate": 1.3327868530511934e-06,
|
| 10211 |
+
"loss": 1.3785,
|
| 10212 |
+
"step": 1453
|
| 10213 |
+
},
|
| 10214 |
+
{
|
| 10215 |
+
"epoch": 0.9498611791605422,
|
| 10216 |
+
"grad_norm": Infinity,
|
| 10217 |
+
"learning_rate": 1.2984684309493179e-06,
|
| 10218 |
+
"loss": 1.4708,
|
| 10219 |
+
"step": 1454
|
| 10220 |
+
},
|
| 10221 |
+
{
|
| 10222 |
+
"epoch": 0.9505144536991671,
|
| 10223 |
+
"grad_norm": Infinity,
|
| 10224 |
+
"learning_rate": 1.2645947425511395e-06,
|
| 10225 |
+
"loss": 1.4621,
|
| 10226 |
+
"step": 1455
|
| 10227 |
+
},
|
| 10228 |
+
{
|
| 10229 |
+
"epoch": 0.9511677282377919,
|
| 10230 |
+
"grad_norm": Infinity,
|
| 10231 |
+
"learning_rate": 1.231165940486234e-06,
|
| 10232 |
+
"loss": 1.567,
|
| 10233 |
+
"step": 1456
|
| 10234 |
+
},
|
| 10235 |
+
{
|
| 10236 |
+
"epoch": 0.9518210027764168,
|
| 10237 |
+
"grad_norm": Infinity,
|
| 10238 |
+
"learning_rate": 1.1981821753795365e-06,
|
| 10239 |
+
"loss": 1.6199,
|
| 10240 |
+
"step": 1457
|
| 10241 |
+
},
|
| 10242 |
+
{
|
| 10243 |
+
"epoch": 0.9524742773150416,
|
| 10244 |
+
"grad_norm": Infinity,
|
| 10245 |
+
"learning_rate": 1.1656435958507738e-06,
|
| 10246 |
+
"loss": 1.5783,
|
| 10247 |
+
"step": 1458
|
| 10248 |
+
},
|
| 10249 |
+
{
|
| 10250 |
+
"epoch": 0.9531275518536665,
|
| 10251 |
+
"grad_norm": Infinity,
|
| 10252 |
+
"learning_rate": 1.133550348513701e-06,
|
| 10253 |
+
"loss": 1.5525,
|
| 10254 |
+
"step": 1459
|
| 10255 |
+
},
|
| 10256 |
+
{
|
| 10257 |
+
"epoch": 0.9537808263922913,
|
| 10258 |
+
"grad_norm": Infinity,
|
| 10259 |
+
"learning_rate": 1.1019025779754666e-06,
|
| 10260 |
+
"loss": 1.6696,
|
| 10261 |
+
"step": 1460
|
| 10262 |
+
},
|
| 10263 |
+
{
|
| 10264 |
+
"epoch": 0.9544341009309162,
|
| 10265 |
+
"grad_norm": Infinity,
|
| 10266 |
+
"learning_rate": 1.0707004268360243e-06,
|
| 10267 |
+
"loss": 1.7598,
|
| 10268 |
+
"step": 1461
|
| 10269 |
+
},
|
| 10270 |
+
{
|
| 10271 |
+
"epoch": 0.955087375469541,
|
| 10272 |
+
"grad_norm": Infinity,
|
| 10273 |
+
"learning_rate": 1.039944035687368e-06,
|
| 10274 |
+
"loss": 1.8395,
|
| 10275 |
+
"step": 1462
|
| 10276 |
+
},
|
| 10277 |
+
{
|
| 10278 |
+
"epoch": 0.9557406500081659,
|
| 10279 |
+
"grad_norm": Infinity,
|
| 10280 |
+
"learning_rate": 1.0096335431130089e-06,
|
| 10281 |
+
"loss": 2.0216,
|
| 10282 |
+
"step": 1463
|
| 10283 |
+
},
|
| 10284 |
+
{
|
| 10285 |
+
"epoch": 0.9563939245467908,
|
| 10286 |
+
"grad_norm": Infinity,
|
| 10287 |
+
"learning_rate": 9.797690856872987e-07,
|
| 10288 |
+
"loss": 1.8501,
|
| 10289 |
+
"step": 1464
|
| 10290 |
+
},
|
| 10291 |
+
{
|
| 10292 |
+
"epoch": 0.9570471990854157,
|
| 10293 |
+
"grad_norm": Infinity,
|
| 10294 |
+
"learning_rate": 9.503507979748305e-07,
|
| 10295 |
+
"loss": 1.9238,
|
| 10296 |
+
"step": 1465
|
| 10297 |
+
},
|
| 10298 |
+
{
|
| 10299 |
+
"epoch": 0.9577004736240405,
|
| 10300 |
+
"grad_norm": Infinity,
|
| 10301 |
+
"learning_rate": 9.21378812529794e-07,
|
| 10302 |
+
"loss": 2.2574,
|
| 10303 |
+
"step": 1466
|
| 10304 |
+
},
|
| 10305 |
+
{
|
| 10306 |
+
"epoch": 0.9583537481626654,
|
| 10307 |
+
"grad_norm": Infinity,
|
| 10308 |
+
"learning_rate": 8.928532598954654e-07,
|
| 10309 |
+
"loss": 2.3334,
|
| 10310 |
+
"step": 1467
|
| 10311 |
+
},
|
| 10312 |
+
{
|
| 10313 |
+
"epoch": 0.9590070227012902,
|
| 10314 |
+
"grad_norm": Infinity,
|
| 10315 |
+
"learning_rate": 8.647742686035298e-07,
|
| 10316 |
+
"loss": 2.5487,
|
| 10317 |
+
"step": 1468
|
| 10318 |
+
},
|
| 10319 |
+
{
|
| 10320 |
+
"epoch": 0.9596602972399151,
|
| 10321 |
+
"grad_norm": Infinity,
|
| 10322 |
+
"learning_rate": 8.371419651735268e-07,
|
| 10323 |
+
"loss": 2.7241,
|
| 10324 |
+
"step": 1469
|
| 10325 |
+
},
|
| 10326 |
+
{
|
| 10327 |
+
"epoch": 0.9603135717785399,
|
| 10328 |
+
"grad_norm": Infinity,
|
| 10329 |
+
"learning_rate": 8.099564741123166e-07,
|
| 10330 |
+
"loss": 2.6671,
|
| 10331 |
+
"step": 1470
|
| 10332 |
+
},
|
| 10333 |
+
{
|
| 10334 |
+
"epoch": 0.9609668463171648,
|
| 10335 |
+
"grad_norm": Infinity,
|
| 10336 |
+
"learning_rate": 7.832179179134703e-07,
|
| 10337 |
+
"loss": 2.8966,
|
| 10338 |
+
"step": 1471
|
| 10339 |
+
},
|
| 10340 |
+
{
|
| 10341 |
+
"epoch": 0.9616201208557896,
|
| 10342 |
+
"grad_norm": Infinity,
|
| 10343 |
+
"learning_rate": 7.569264170567691e-07,
|
| 10344 |
+
"loss": 2.9527,
|
| 10345 |
+
"step": 1472
|
| 10346 |
+
},
|
| 10347 |
+
{
|
| 10348 |
+
"epoch": 0.9622733953944145,
|
| 10349 |
+
"grad_norm": Infinity,
|
| 10350 |
+
"learning_rate": 7.310820900075955e-07,
|
| 10351 |
+
"loss": 3.0179,
|
| 10352 |
+
"step": 1473
|
| 10353 |
+
},
|
| 10354 |
+
{
|
| 10355 |
+
"epoch": 0.9629266699330393,
|
| 10356 |
+
"grad_norm": Infinity,
|
| 10357 |
+
"learning_rate": 7.05685053216465e-07,
|
| 10358 |
+
"loss": 3.6077,
|
| 10359 |
+
"step": 1474
|
| 10360 |
+
},
|
| 10361 |
+
{
|
| 10362 |
+
"epoch": 0.9635799444716642,
|
| 10363 |
+
"grad_norm": Infinity,
|
| 10364 |
+
"learning_rate": 6.807354211184613e-07,
|
| 10365 |
+
"loss": 4.2041,
|
| 10366 |
+
"step": 1475
|
| 10367 |
+
},
|
| 10368 |
+
{
|
| 10369 |
+
"epoch": 0.964233219010289,
|
| 10370 |
+
"grad_norm": Infinity,
|
| 10371 |
+
"learning_rate": 6.562333061327364e-07,
|
| 10372 |
+
"loss": 1.3957,
|
| 10373 |
+
"step": 1476
|
| 10374 |
+
},
|
| 10375 |
+
{
|
| 10376 |
+
"epoch": 0.964886493548914,
|
| 10377 |
+
"grad_norm": Infinity,
|
| 10378 |
+
"learning_rate": 6.321788186619992e-07,
|
| 10379 |
+
"loss": 1.4075,
|
| 10380 |
+
"step": 1477
|
| 10381 |
+
},
|
| 10382 |
+
{
|
| 10383 |
+
"epoch": 0.9655397680875388,
|
| 10384 |
+
"grad_norm": Infinity,
|
| 10385 |
+
"learning_rate": 6.08572067092017e-07,
|
| 10386 |
+
"loss": 1.4448,
|
| 10387 |
+
"step": 1478
|
| 10388 |
+
},
|
| 10389 |
+
{
|
| 10390 |
+
"epoch": 0.9661930426261637,
|
| 10391 |
+
"grad_norm": Infinity,
|
| 10392 |
+
"learning_rate": 5.854131577911259e-07,
|
| 10393 |
+
"loss": 1.4893,
|
| 10394 |
+
"step": 1479
|
| 10395 |
+
},
|
| 10396 |
+
{
|
| 10397 |
+
"epoch": 0.9668463171647885,
|
| 10398 |
+
"grad_norm": Infinity,
|
| 10399 |
+
"learning_rate": 5.627021951097545e-07,
|
| 10400 |
+
"loss": 1.4216,
|
| 10401 |
+
"step": 1480
|
| 10402 |
+
},
|
| 10403 |
+
{
|
| 10404 |
+
"epoch": 0.9674995917034134,
|
| 10405 |
+
"grad_norm": Infinity,
|
| 10406 |
+
"learning_rate": 5.404392813799675e-07,
|
| 10407 |
+
"loss": 1.5241,
|
| 10408 |
+
"step": 1481
|
| 10409 |
+
},
|
| 10410 |
+
{
|
| 10411 |
+
"epoch": 0.9681528662420382,
|
| 10412 |
+
"grad_norm": Infinity,
|
| 10413 |
+
"learning_rate": 5.186245169149784e-07,
|
| 10414 |
+
"loss": 1.4522,
|
| 10415 |
+
"step": 1482
|
| 10416 |
+
},
|
| 10417 |
+
{
|
| 10418 |
+
"epoch": 0.9688061407806631,
|
| 10419 |
+
"grad_norm": Infinity,
|
| 10420 |
+
"learning_rate": 4.972580000087046e-07,
|
| 10421 |
+
"loss": 1.5146,
|
| 10422 |
+
"step": 1483
|
| 10423 |
+
},
|
| 10424 |
+
{
|
| 10425 |
+
"epoch": 0.9694594153192879,
|
| 10426 |
+
"grad_norm": Infinity,
|
| 10427 |
+
"learning_rate": 4.7633982693535693e-07,
|
| 10428 |
+
"loss": 1.6367,
|
| 10429 |
+
"step": 1484
|
| 10430 |
+
},
|
| 10431 |
+
{
|
| 10432 |
+
"epoch": 0.9701126898579128,
|
| 10433 |
+
"grad_norm": Infinity,
|
| 10434 |
+
"learning_rate": 4.5587009194894004e-07,
|
| 10435 |
+
"loss": 1.6609,
|
| 10436 |
+
"step": 1485
|
| 10437 |
+
},
|
| 10438 |
+
{
|
| 10439 |
+
"epoch": 0.9707659643965376,
|
| 10440 |
+
"grad_norm": Infinity,
|
| 10441 |
+
"learning_rate": 4.3584888728289695e-07,
|
| 10442 |
+
"loss": 1.6756,
|
| 10443 |
+
"step": 1486
|
| 10444 |
+
},
|
| 10445 |
+
{
|
| 10446 |
+
"epoch": 0.9714192389351625,
|
| 10447 |
+
"grad_norm": Infinity,
|
| 10448 |
+
"learning_rate": 4.1627630314965415e-07,
|
| 10449 |
+
"loss": 1.6661,
|
| 10450 |
+
"step": 1487
|
| 10451 |
+
},
|
| 10452 |
+
{
|
| 10453 |
+
"epoch": 0.9720725134737873,
|
| 10454 |
+
"grad_norm": Infinity,
|
| 10455 |
+
"learning_rate": 3.9715242774021055e-07,
|
| 10456 |
+
"loss": 1.8335,
|
| 10457 |
+
"step": 1488
|
| 10458 |
+
},
|
| 10459 |
+
{
|
| 10460 |
+
"epoch": 0.9727257880124122,
|
| 10461 |
+
"grad_norm": Infinity,
|
| 10462 |
+
"learning_rate": 3.7847734722378234e-07,
|
| 10463 |
+
"loss": 1.9859,
|
| 10464 |
+
"step": 1489
|
| 10465 |
+
},
|
| 10466 |
+
{
|
| 10467 |
+
"epoch": 0.9733790625510371,
|
| 10468 |
+
"grad_norm": Infinity,
|
| 10469 |
+
"learning_rate": 3.6025114574734785e-07,
|
| 10470 |
+
"loss": 2.034,
|
| 10471 |
+
"step": 1490
|
| 10472 |
+
},
|
| 10473 |
+
{
|
| 10474 |
+
"epoch": 0.974032337089662,
|
| 10475 |
+
"grad_norm": Infinity,
|
| 10476 |
+
"learning_rate": 3.424739054353476e-07,
|
| 10477 |
+
"loss": 1.9654,
|
| 10478 |
+
"step": 1491
|
| 10479 |
+
},
|
| 10480 |
+
{
|
| 10481 |
+
"epoch": 0.9746856116282868,
|
| 10482 |
+
"grad_norm": Infinity,
|
| 10483 |
+
"learning_rate": 3.2514570638925157e-07,
|
| 10484 |
+
"loss": 2.3266,
|
| 10485 |
+
"step": 1492
|
| 10486 |
+
},
|
| 10487 |
+
{
|
| 10488 |
+
"epoch": 0.9753388861669117,
|
| 10489 |
+
"grad_norm": Infinity,
|
| 10490 |
+
"learning_rate": 3.0826662668720364e-07,
|
| 10491 |
+
"loss": 2.358,
|
| 10492 |
+
"step": 1493
|
| 10493 |
+
},
|
| 10494 |
+
{
|
| 10495 |
+
"epoch": 0.9759921607055365,
|
| 10496 |
+
"grad_norm": Infinity,
|
| 10497 |
+
"learning_rate": 2.9183674238372206e-07,
|
| 10498 |
+
"loss": 2.384,
|
| 10499 |
+
"step": 1494
|
| 10500 |
+
},
|
| 10501 |
+
{
|
| 10502 |
+
"epoch": 0.9766454352441614,
|
| 10503 |
+
"grad_norm": Infinity,
|
| 10504 |
+
"learning_rate": 2.758561275092886e-07,
|
| 10505 |
+
"loss": 2.6566,
|
| 10506 |
+
"step": 1495
|
| 10507 |
+
},
|
| 10508 |
+
{
|
| 10509 |
+
"epoch": 0.9772987097827862,
|
| 10510 |
+
"grad_norm": Infinity,
|
| 10511 |
+
"learning_rate": 2.603248540700709e-07,
|
| 10512 |
+
"loss": 3.0257,
|
| 10513 |
+
"step": 1496
|
| 10514 |
+
},
|
| 10515 |
+
{
|
| 10516 |
+
"epoch": 0.9779519843214111,
|
| 10517 |
+
"grad_norm": Infinity,
|
| 10518 |
+
"learning_rate": 2.452429920475674e-07,
|
| 10519 |
+
"loss": 2.866,
|
| 10520 |
+
"step": 1497
|
| 10521 |
+
},
|
| 10522 |
+
{
|
| 10523 |
+
"epoch": 0.9786052588600359,
|
| 10524 |
+
"grad_norm": Infinity,
|
| 10525 |
+
"learning_rate": 2.3061060939828517e-07,
|
| 10526 |
+
"loss": 2.8559,
|
| 10527 |
+
"step": 1498
|
| 10528 |
+
},
|
| 10529 |
+
{
|
| 10530 |
+
"epoch": 0.9792585333986608,
|
| 10531 |
+
"grad_norm": Infinity,
|
| 10532 |
+
"learning_rate": 2.1642777205346242e-07,
|
| 10533 |
+
"loss": 3.8677,
|
| 10534 |
+
"step": 1499
|
| 10535 |
+
},
|
| 10536 |
+
{
|
| 10537 |
+
"epoch": 0.9799118079372856,
|
| 10538 |
+
"grad_norm": Infinity,
|
| 10539 |
+
"learning_rate": 2.0269454391874666e-07,
|
| 10540 |
+
"loss": 3.9609,
|
| 10541 |
+
"step": 1500
|
| 10542 |
+
},
|
| 10543 |
+
{
|
| 10544 |
+
"epoch": 0.9805650824759105,
|
| 10545 |
+
"grad_norm": Infinity,
|
| 10546 |
+
"learning_rate": 1.89410986873928e-07,
|
| 10547 |
+
"loss": 1.2609,
|
| 10548 |
+
"step": 1501
|
| 10549 |
+
},
|
| 10550 |
+
{
|
| 10551 |
+
"epoch": 0.9812183570145354,
|
| 10552 |
+
"grad_norm": Infinity,
|
| 10553 |
+
"learning_rate": 1.7657716077265073e-07,
|
| 10554 |
+
"loss": 1.307,
|
| 10555 |
+
"step": 1502
|
| 10556 |
+
},
|
| 10557 |
+
{
|
| 10558 |
+
"epoch": 0.9818716315531603,
|
| 10559 |
+
"grad_norm": Infinity,
|
| 10560 |
+
"learning_rate": 1.6419312344211347e-07,
|
| 10561 |
+
"loss": 1.3644,
|
| 10562 |
+
"step": 1503
|
| 10563 |
+
},
|
| 10564 |
+
{
|
| 10565 |
+
"epoch": 0.9825249060917851,
|
| 10566 |
+
"grad_norm": Infinity,
|
| 10567 |
+
"learning_rate": 1.5225893068286922e-07,
|
| 10568 |
+
"loss": 1.4293,
|
| 10569 |
+
"step": 1504
|
| 10570 |
+
},
|
| 10571 |
+
{
|
| 10572 |
+
"epoch": 0.98317818063041,
|
| 10573 |
+
"grad_norm": Infinity,
|
| 10574 |
+
"learning_rate": 1.4077463626852582e-07,
|
| 10575 |
+
"loss": 1.3799,
|
| 10576 |
+
"step": 1505
|
| 10577 |
+
},
|
| 10578 |
+
{
|
| 10579 |
+
"epoch": 0.9838314551690348,
|
| 10580 |
+
"grad_norm": Infinity,
|
| 10581 |
+
"learning_rate": 1.2974029194551262e-07,
|
| 10582 |
+
"loss": 1.4381,
|
| 10583 |
+
"step": 1506
|
| 10584 |
+
},
|
| 10585 |
+
{
|
| 10586 |
+
"epoch": 0.9844847297076597,
|
| 10587 |
+
"grad_norm": Infinity,
|
| 10588 |
+
"learning_rate": 1.1915594743288072e-07,
|
| 10589 |
+
"loss": 1.5463,
|
| 10590 |
+
"step": 1507
|
| 10591 |
+
},
|
| 10592 |
+
{
|
| 10593 |
+
"epoch": 0.9851380042462845,
|
| 10594 |
+
"grad_norm": Infinity,
|
| 10595 |
+
"learning_rate": 1.0902165042202539e-07,
|
| 10596 |
+
"loss": 1.5154,
|
| 10597 |
+
"step": 1508
|
| 10598 |
+
},
|
| 10599 |
+
{
|
| 10600 |
+
"epoch": 0.9857912787849094,
|
| 10601 |
+
"grad_norm": Infinity,
|
| 10602 |
+
"learning_rate": 9.933744657651956e-08,
|
| 10603 |
+
"loss": 1.6106,
|
| 10604 |
+
"step": 1509
|
| 10605 |
+
},
|
| 10606 |
+
{
|
| 10607 |
+
"epoch": 0.9864445533235342,
|
| 10608 |
+
"grad_norm": Infinity,
|
| 10609 |
+
"learning_rate": 9.010337953185843e-08,
|
| 10610 |
+
"loss": 1.6065,
|
| 10611 |
+
"step": 1510
|
| 10612 |
+
},
|
| 10613 |
+
{
|
| 10614 |
+
"epoch": 0.987097827862159,
|
| 10615 |
+
"grad_norm": Infinity,
|
| 10616 |
+
"learning_rate": 8.131949089531521e-08,
|
| 10617 |
+
"loss": 1.7721,
|
| 10618 |
+
"step": 1511
|
| 10619 |
+
},
|
| 10620 |
+
{
|
| 10621 |
+
"epoch": 0.9877511024007839,
|
| 10622 |
+
"grad_norm": Infinity,
|
| 10623 |
+
"learning_rate": 7.298582024571899e-08,
|
| 10624 |
+
"loss": 1.8281,
|
| 10625 |
+
"step": 1512
|
| 10626 |
+
},
|
| 10627 |
+
{
|
| 10628 |
+
"epoch": 0.9884043769394087,
|
| 10629 |
+
"grad_norm": Infinity,
|
| 10630 |
+
"learning_rate": 6.510240513328824e-08,
|
| 10631 |
+
"loss": 1.8281,
|
| 10632 |
+
"step": 1513
|
| 10633 |
+
},
|
| 10634 |
+
{
|
| 10635 |
+
"epoch": 0.9890576514780336,
|
| 10636 |
+
"grad_norm": Infinity,
|
| 10637 |
+
"learning_rate": 5.7669281079475446e-08,
|
| 10638 |
+
"loss": 1.8849,
|
| 10639 |
+
"step": 1514
|
| 10640 |
+
},
|
| 10641 |
+
{
|
| 10642 |
+
"epoch": 0.9897109260166586,
|
| 10643 |
+
"grad_norm": Infinity,
|
| 10644 |
+
"learning_rate": 5.068648157675604e-08,
|
| 10645 |
+
"loss": 1.9238,
|
| 10646 |
+
"step": 1515
|
| 10647 |
+
},
|
| 10648 |
+
{
|
| 10649 |
+
"epoch": 0.9903642005552834,
|
| 10650 |
+
"grad_norm": Infinity,
|
| 10651 |
+
"learning_rate": 4.4154038088561886e-08,
|
| 10652 |
+
"loss": 2.0281,
|
| 10653 |
+
"step": 1516
|
| 10654 |
+
},
|
| 10655 |
+
{
|
| 10656 |
+
"epoch": 0.9910174750939083,
|
| 10657 |
+
"grad_norm": Infinity,
|
| 10658 |
+
"learning_rate": 3.8071980049037006e-08,
|
| 10659 |
+
"loss": 2.3601,
|
| 10660 |
+
"step": 1517
|
| 10661 |
+
},
|
| 10662 |
+
{
|
| 10663 |
+
"epoch": 0.9916707496325331,
|
| 10664 |
+
"grad_norm": Infinity,
|
| 10665 |
+
"learning_rate": 3.244033486300424e-08,
|
| 10666 |
+
"loss": 2.4496,
|
| 10667 |
+
"step": 1518
|
| 10668 |
+
},
|
| 10669 |
+
{
|
| 10670 |
+
"epoch": 0.992324024171158,
|
| 10671 |
+
"grad_norm": Infinity,
|
| 10672 |
+
"learning_rate": 2.7259127905776562e-08,
|
| 10673 |
+
"loss": 2.399,
|
| 10674 |
+
"step": 1519
|
| 10675 |
+
},
|
| 10676 |
+
{
|
| 10677 |
+
"epoch": 0.9929772987097828,
|
| 10678 |
+
"grad_norm": Infinity,
|
| 10679 |
+
"learning_rate": 2.2528382523057113e-08,
|
| 10680 |
+
"loss": 2.5236,
|
| 10681 |
+
"step": 1520
|
| 10682 |
+
},
|
| 10683 |
+
{
|
| 10684 |
+
"epoch": 0.9936305732484076,
|
| 10685 |
+
"grad_norm": Infinity,
|
| 10686 |
+
"learning_rate": 1.8248120030850413e-08,
|
| 10687 |
+
"loss": 2.7336,
|
| 10688 |
+
"step": 1521
|
| 10689 |
+
},
|
| 10690 |
+
{
|
| 10691 |
+
"epoch": 0.9942838477870325,
|
| 10692 |
+
"grad_norm": Infinity,
|
| 10693 |
+
"learning_rate": 1.4418359715351327e-08,
|
| 10694 |
+
"loss": 2.898,
|
| 10695 |
+
"step": 1522
|
| 10696 |
+
},
|
| 10697 |
+
{
|
| 10698 |
+
"epoch": 0.9949371223256573,
|
| 10699 |
+
"grad_norm": Infinity,
|
| 10700 |
+
"learning_rate": 1.1039118832867345e-08,
|
| 10701 |
+
"loss": 3.4187,
|
| 10702 |
+
"step": 1523
|
| 10703 |
+
},
|
| 10704 |
+
{
|
| 10705 |
+
"epoch": 0.9955903968642822,
|
| 10706 |
+
"grad_norm": Infinity,
|
| 10707 |
+
"learning_rate": 8.110412609718676e-09,
|
| 10708 |
+
"loss": 3.3366,
|
| 10709 |
+
"step": 1524
|
| 10710 |
+
},
|
| 10711 |
+
{
|
| 10712 |
+
"epoch": 0.996243671402907,
|
| 10713 |
+
"grad_norm": Infinity,
|
| 10714 |
+
"learning_rate": 5.6322542422049266e-09,
|
| 10715 |
+
"loss": 4.1065,
|
| 10716 |
+
"step": 1525
|
| 10717 |
+
},
|
| 10718 |
+
{
|
| 10719 |
+
"epoch": 0.9968969459415319,
|
| 10720 |
+
"grad_norm": Infinity,
|
| 10721 |
+
"learning_rate": 3.604654896527393e-09,
|
| 10722 |
+
"loss": 1.3718,
|
| 10723 |
+
"step": 1526
|
| 10724 |
+
},
|
| 10725 |
+
{
|
| 10726 |
+
"epoch": 0.9975502204801567,
|
| 10727 |
+
"grad_norm": Infinity,
|
| 10728 |
+
"learning_rate": 2.027623708722448e-09,
|
| 10729 |
+
"loss": 1.5273,
|
| 10730 |
+
"step": 1527
|
| 10731 |
+
},
|
| 10732 |
+
{
|
| 10733 |
+
"epoch": 0.9982034950187817,
|
| 10734 |
+
"grad_norm": Infinity,
|
| 10735 |
+
"learning_rate": 9.011677846504363e-10,
|
| 10736 |
+
"loss": 1.8061,
|
| 10737 |
+
"step": 1528
|
| 10738 |
+
},
|
| 10739 |
+
{
|
| 10740 |
+
"epoch": 0.9988567695574065,
|
| 10741 |
+
"grad_norm": Infinity,
|
| 10742 |
+
"learning_rate": 2.252921999401636e-10,
|
| 10743 |
+
"loss": 2.0641,
|
| 10744 |
+
"step": 1529
|
| 10745 |
+
},
|
| 10746 |
+
{
|
| 10747 |
+
"epoch": 0.9995100440960314,
|
| 10748 |
+
"grad_norm": Infinity,
|
| 10749 |
+
"learning_rate": 0.0,
|
| 10750 |
+
"loss": 2.8842,
|
| 10751 |
+
"step": 1530
|
| 10752 |
}
|
| 10753 |
],
|
| 10754 |
"logging_steps": 1,
|
|
|
|
| 10763 |
"should_evaluate": false,
|
| 10764 |
"should_log": false,
|
| 10765 |
"should_save": true,
|
| 10766 |
+
"should_training_stop": true
|
| 10767 |
},
|
| 10768 |
"attributes": {}
|
| 10769 |
}
|
| 10770 |
},
|
| 10771 |
+
"total_flos": 6.299736884851507e+17,
|
| 10772 |
"train_batch_size": 2,
|
| 10773 |
"trial_name": null,
|
| 10774 |
"trial_params": null
|