Training in progress, step 1140, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 80792096
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:94341634f8a6afeb1417aef19e2dc8200584cdd3b7697243a52ef6a5a8499b76
|
| 3 |
size 80792096
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 41460084
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:40a3cb1571d7f507071d4bcdd907987f10dbfc1a03fcaba90d96eb78f0972ba5
|
| 3 |
size 41460084
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:71b5e22ff90c1903aaf724f48655dfadec626926b6c93dacb875ca649a4a2040
|
| 3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7580f3e3fea1333d62db7e78a941076f3d266b318c0579b5be4bec002280435a
|
| 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": 380,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -5351,6 +5351,2674 @@
|
|
| 5351 |
"eval_samples_per_second": 32.248,
|
| 5352 |
"eval_steps_per_second": 16.127,
|
| 5353 |
"step": 760
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5354 |
}
|
| 5355 |
],
|
| 5356 |
"logging_steps": 1,
|
|
@@ -5370,7 +8038,7 @@
|
|
| 5370 |
"attributes": {}
|
| 5371 |
}
|
| 5372 |
},
|
| 5373 |
-
"total_flos":
|
| 5374 |
"train_batch_size": 2,
|
| 5375 |
"trial_name": null,
|
| 5376 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.08278566500853274,
|
| 5 |
"eval_steps": 380,
|
| 6 |
+
"global_step": 1140,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 5351 |
"eval_samples_per_second": 32.248,
|
| 5352 |
"eval_steps_per_second": 16.127,
|
| 5353 |
"step": 760
|
| 5354 |
+
},
|
| 5355 |
+
{
|
| 5356 |
+
"epoch": 0.05526306234341527,
|
| 5357 |
+
"grad_norm": 1.1962695121765137,
|
| 5358 |
+
"learning_rate": 0.00010052116430522514,
|
| 5359 |
+
"loss": 0.7833,
|
| 5360 |
+
"step": 761
|
| 5361 |
+
},
|
| 5362 |
+
{
|
| 5363 |
+
"epoch": 0.055335681347808724,
|
| 5364 |
+
"grad_norm": 1.4395370483398438,
|
| 5365 |
+
"learning_rate": 0.00010031269948909516,
|
| 5366 |
+
"loss": 0.2363,
|
| 5367 |
+
"step": 762
|
| 5368 |
+
},
|
| 5369 |
+
{
|
| 5370 |
+
"epoch": 0.05540830035220217,
|
| 5371 |
+
"grad_norm": 1.3636575937271118,
|
| 5372 |
+
"learning_rate": 0.00010010423331402527,
|
| 5373 |
+
"loss": 0.343,
|
| 5374 |
+
"step": 763
|
| 5375 |
+
},
|
| 5376 |
+
{
|
| 5377 |
+
"epoch": 0.05548091935659562,
|
| 5378 |
+
"grad_norm": 2.326564073562622,
|
| 5379 |
+
"learning_rate": 9.989576668597476e-05,
|
| 5380 |
+
"loss": 0.2894,
|
| 5381 |
+
"step": 764
|
| 5382 |
+
},
|
| 5383 |
+
{
|
| 5384 |
+
"epoch": 0.05555353836098907,
|
| 5385 |
+
"grad_norm": 0.910338819026947,
|
| 5386 |
+
"learning_rate": 9.968730051090485e-05,
|
| 5387 |
+
"loss": 0.6565,
|
| 5388 |
+
"step": 765
|
| 5389 |
+
},
|
| 5390 |
+
{
|
| 5391 |
+
"epoch": 0.05562615736538252,
|
| 5392 |
+
"grad_norm": 1.8644438982009888,
|
| 5393 |
+
"learning_rate": 9.947883569477488e-05,
|
| 5394 |
+
"loss": 0.5779,
|
| 5395 |
+
"step": 766
|
| 5396 |
+
},
|
| 5397 |
+
{
|
| 5398 |
+
"epoch": 0.05569877636977597,
|
| 5399 |
+
"grad_norm": 1.2688733339309692,
|
| 5400 |
+
"learning_rate": 9.927037314353821e-05,
|
| 5401 |
+
"loss": 0.7624,
|
| 5402 |
+
"step": 767
|
| 5403 |
+
},
|
| 5404 |
+
{
|
| 5405 |
+
"epoch": 0.05577139537416942,
|
| 5406 |
+
"grad_norm": 1.6676098108291626,
|
| 5407 |
+
"learning_rate": 9.906191376313836e-05,
|
| 5408 |
+
"loss": 1.169,
|
| 5409 |
+
"step": 768
|
| 5410 |
+
},
|
| 5411 |
+
{
|
| 5412 |
+
"epoch": 0.05584401437856287,
|
| 5413 |
+
"grad_norm": 0.8322520852088928,
|
| 5414 |
+
"learning_rate": 9.88534584595051e-05,
|
| 5415 |
+
"loss": 0.9831,
|
| 5416 |
+
"step": 769
|
| 5417 |
+
},
|
| 5418 |
+
{
|
| 5419 |
+
"epoch": 0.05591663338295632,
|
| 5420 |
+
"grad_norm": 1.1143416166305542,
|
| 5421 |
+
"learning_rate": 9.86450081385505e-05,
|
| 5422 |
+
"loss": 0.447,
|
| 5423 |
+
"step": 770
|
| 5424 |
+
},
|
| 5425 |
+
{
|
| 5426 |
+
"epoch": 0.05598925238734977,
|
| 5427 |
+
"grad_norm": 2.037198543548584,
|
| 5428 |
+
"learning_rate": 9.84365637061649e-05,
|
| 5429 |
+
"loss": 1.061,
|
| 5430 |
+
"step": 771
|
| 5431 |
+
},
|
| 5432 |
+
{
|
| 5433 |
+
"epoch": 0.05606187139174322,
|
| 5434 |
+
"grad_norm": 0.8915815949440002,
|
| 5435 |
+
"learning_rate": 9.822812606821316e-05,
|
| 5436 |
+
"loss": 0.2336,
|
| 5437 |
+
"step": 772
|
| 5438 |
+
},
|
| 5439 |
+
{
|
| 5440 |
+
"epoch": 0.05613449039613667,
|
| 5441 |
+
"grad_norm": 0.7484176158905029,
|
| 5442 |
+
"learning_rate": 9.801969613053047e-05,
|
| 5443 |
+
"loss": 0.3113,
|
| 5444 |
+
"step": 773
|
| 5445 |
+
},
|
| 5446 |
+
{
|
| 5447 |
+
"epoch": 0.05620710940053012,
|
| 5448 |
+
"grad_norm": 0.8235967755317688,
|
| 5449 |
+
"learning_rate": 9.781127479891872e-05,
|
| 5450 |
+
"loss": 0.1354,
|
| 5451 |
+
"step": 774
|
| 5452 |
+
},
|
| 5453 |
+
{
|
| 5454 |
+
"epoch": 0.05627972840492357,
|
| 5455 |
+
"grad_norm": 2.569800853729248,
|
| 5456 |
+
"learning_rate": 9.760286297914223e-05,
|
| 5457 |
+
"loss": 0.9149,
|
| 5458 |
+
"step": 775
|
| 5459 |
+
},
|
| 5460 |
+
{
|
| 5461 |
+
"epoch": 0.056352347409317015,
|
| 5462 |
+
"grad_norm": 0.986344039440155,
|
| 5463 |
+
"learning_rate": 9.739446157692415e-05,
|
| 5464 |
+
"loss": 0.0597,
|
| 5465 |
+
"step": 776
|
| 5466 |
+
},
|
| 5467 |
+
{
|
| 5468 |
+
"epoch": 0.05642496641371047,
|
| 5469 |
+
"grad_norm": 1.5366631746292114,
|
| 5470 |
+
"learning_rate": 9.71860714979422e-05,
|
| 5471 |
+
"loss": 0.26,
|
| 5472 |
+
"step": 777
|
| 5473 |
+
},
|
| 5474 |
+
{
|
| 5475 |
+
"epoch": 0.056497585418103916,
|
| 5476 |
+
"grad_norm": 1.1993207931518555,
|
| 5477 |
+
"learning_rate": 9.697769364782501e-05,
|
| 5478 |
+
"loss": 0.9964,
|
| 5479 |
+
"step": 778
|
| 5480 |
+
},
|
| 5481 |
+
{
|
| 5482 |
+
"epoch": 0.05657020442249737,
|
| 5483 |
+
"grad_norm": 0.4609467685222626,
|
| 5484 |
+
"learning_rate": 9.676932893214805e-05,
|
| 5485 |
+
"loss": 0.1857,
|
| 5486 |
+
"step": 779
|
| 5487 |
+
},
|
| 5488 |
+
{
|
| 5489 |
+
"epoch": 0.056642823426890816,
|
| 5490 |
+
"grad_norm": 2.9757611751556396,
|
| 5491 |
+
"learning_rate": 9.656097825642961e-05,
|
| 5492 |
+
"loss": 0.5071,
|
| 5493 |
+
"step": 780
|
| 5494 |
+
},
|
| 5495 |
+
{
|
| 5496 |
+
"epoch": 0.05671544243128427,
|
| 5497 |
+
"grad_norm": 0.9907705187797546,
|
| 5498 |
+
"learning_rate": 9.635264252612711e-05,
|
| 5499 |
+
"loss": 1.1982,
|
| 5500 |
+
"step": 781
|
| 5501 |
+
},
|
| 5502 |
+
{
|
| 5503 |
+
"epoch": 0.05678806143567772,
|
| 5504 |
+
"grad_norm": 0.6433952450752258,
|
| 5505 |
+
"learning_rate": 9.614432264663287e-05,
|
| 5506 |
+
"loss": 0.6687,
|
| 5507 |
+
"step": 782
|
| 5508 |
+
},
|
| 5509 |
+
{
|
| 5510 |
+
"epoch": 0.056860680440071164,
|
| 5511 |
+
"grad_norm": 1.5051597356796265,
|
| 5512 |
+
"learning_rate": 9.593601952327048e-05,
|
| 5513 |
+
"loss": 0.7077,
|
| 5514 |
+
"step": 783
|
| 5515 |
+
},
|
| 5516 |
+
{
|
| 5517 |
+
"epoch": 0.05693329944446462,
|
| 5518 |
+
"grad_norm": 1.4717247486114502,
|
| 5519 |
+
"learning_rate": 9.572773406129057e-05,
|
| 5520 |
+
"loss": 0.7229,
|
| 5521 |
+
"step": 784
|
| 5522 |
+
},
|
| 5523 |
+
{
|
| 5524 |
+
"epoch": 0.057005918448858064,
|
| 5525 |
+
"grad_norm": 1.4327820539474487,
|
| 5526 |
+
"learning_rate": 9.551946716586713e-05,
|
| 5527 |
+
"loss": 1.2878,
|
| 5528 |
+
"step": 785
|
| 5529 |
+
},
|
| 5530 |
+
{
|
| 5531 |
+
"epoch": 0.05707853745325152,
|
| 5532 |
+
"grad_norm": 3.108001708984375,
|
| 5533 |
+
"learning_rate": 9.531121974209333e-05,
|
| 5534 |
+
"loss": 0.8526,
|
| 5535 |
+
"step": 786
|
| 5536 |
+
},
|
| 5537 |
+
{
|
| 5538 |
+
"epoch": 0.057151156457644965,
|
| 5539 |
+
"grad_norm": 2.0628154277801514,
|
| 5540 |
+
"learning_rate": 9.510299269497791e-05,
|
| 5541 |
+
"loss": 1.0106,
|
| 5542 |
+
"step": 787
|
| 5543 |
+
},
|
| 5544 |
+
{
|
| 5545 |
+
"epoch": 0.05722377546203841,
|
| 5546 |
+
"grad_norm": 0.8770999908447266,
|
| 5547 |
+
"learning_rate": 9.489478692944087e-05,
|
| 5548 |
+
"loss": 0.9365,
|
| 5549 |
+
"step": 788
|
| 5550 |
+
},
|
| 5551 |
+
{
|
| 5552 |
+
"epoch": 0.057296394466431866,
|
| 5553 |
+
"grad_norm": 1.6717029809951782,
|
| 5554 |
+
"learning_rate": 9.46866033503098e-05,
|
| 5555 |
+
"loss": 0.4152,
|
| 5556 |
+
"step": 789
|
| 5557 |
+
},
|
| 5558 |
+
{
|
| 5559 |
+
"epoch": 0.05736901347082531,
|
| 5560 |
+
"grad_norm": 0.6597108244895935,
|
| 5561 |
+
"learning_rate": 9.447844286231594e-05,
|
| 5562 |
+
"loss": 0.6531,
|
| 5563 |
+
"step": 790
|
| 5564 |
+
},
|
| 5565 |
+
{
|
| 5566 |
+
"epoch": 0.057441632475218767,
|
| 5567 |
+
"grad_norm": 1.3612674474716187,
|
| 5568 |
+
"learning_rate": 9.427030637009003e-05,
|
| 5569 |
+
"loss": 0.4148,
|
| 5570 |
+
"step": 791
|
| 5571 |
+
},
|
| 5572 |
+
{
|
| 5573 |
+
"epoch": 0.05751425147961221,
|
| 5574 |
+
"grad_norm": 0.5453882217407227,
|
| 5575 |
+
"learning_rate": 9.406219477815867e-05,
|
| 5576 |
+
"loss": 0.426,
|
| 5577 |
+
"step": 792
|
| 5578 |
+
},
|
| 5579 |
+
{
|
| 5580 |
+
"epoch": 0.05758687048400567,
|
| 5581 |
+
"grad_norm": 0.8962843418121338,
|
| 5582 |
+
"learning_rate": 9.385410899094013e-05,
|
| 5583 |
+
"loss": 0.341,
|
| 5584 |
+
"step": 793
|
| 5585 |
+
},
|
| 5586 |
+
{
|
| 5587 |
+
"epoch": 0.057659489488399114,
|
| 5588 |
+
"grad_norm": 0.7332125306129456,
|
| 5589 |
+
"learning_rate": 9.364604991274068e-05,
|
| 5590 |
+
"loss": 0.1671,
|
| 5591 |
+
"step": 794
|
| 5592 |
+
},
|
| 5593 |
+
{
|
| 5594 |
+
"epoch": 0.05773210849279256,
|
| 5595 |
+
"grad_norm": 1.698838233947754,
|
| 5596 |
+
"learning_rate": 9.343801844775034e-05,
|
| 5597 |
+
"loss": 0.2474,
|
| 5598 |
+
"step": 795
|
| 5599 |
+
},
|
| 5600 |
+
{
|
| 5601 |
+
"epoch": 0.057804727497186015,
|
| 5602 |
+
"grad_norm": 1.134534239768982,
|
| 5603 |
+
"learning_rate": 9.323001550003931e-05,
|
| 5604 |
+
"loss": 1.1692,
|
| 5605 |
+
"step": 796
|
| 5606 |
+
},
|
| 5607 |
+
{
|
| 5608 |
+
"epoch": 0.05787734650157946,
|
| 5609 |
+
"grad_norm": 0.7837601900100708,
|
| 5610 |
+
"learning_rate": 9.302204197355368e-05,
|
| 5611 |
+
"loss": 0.664,
|
| 5612 |
+
"step": 797
|
| 5613 |
+
},
|
| 5614 |
+
{
|
| 5615 |
+
"epoch": 0.057949965505972915,
|
| 5616 |
+
"grad_norm": 2.5851547718048096,
|
| 5617 |
+
"learning_rate": 9.28140987721118e-05,
|
| 5618 |
+
"loss": 0.7487,
|
| 5619 |
+
"step": 798
|
| 5620 |
+
},
|
| 5621 |
+
{
|
| 5622 |
+
"epoch": 0.05802258451036636,
|
| 5623 |
+
"grad_norm": 0.5738522410392761,
|
| 5624 |
+
"learning_rate": 9.260618679940025e-05,
|
| 5625 |
+
"loss": 0.4756,
|
| 5626 |
+
"step": 799
|
| 5627 |
+
},
|
| 5628 |
+
{
|
| 5629 |
+
"epoch": 0.05809520351475981,
|
| 5630 |
+
"grad_norm": 1.0114805698394775,
|
| 5631 |
+
"learning_rate": 9.239830695896978e-05,
|
| 5632 |
+
"loss": 0.3665,
|
| 5633 |
+
"step": 800
|
| 5634 |
+
},
|
| 5635 |
+
{
|
| 5636 |
+
"epoch": 0.05816782251915326,
|
| 5637 |
+
"grad_norm": 1.4201353788375854,
|
| 5638 |
+
"learning_rate": 9.219046015423162e-05,
|
| 5639 |
+
"loss": 1.0616,
|
| 5640 |
+
"step": 801
|
| 5641 |
+
},
|
| 5642 |
+
{
|
| 5643 |
+
"epoch": 0.05824044152354671,
|
| 5644 |
+
"grad_norm": 0.5711639523506165,
|
| 5645 |
+
"learning_rate": 9.198264728845332e-05,
|
| 5646 |
+
"loss": 0.2396,
|
| 5647 |
+
"step": 802
|
| 5648 |
+
},
|
| 5649 |
+
{
|
| 5650 |
+
"epoch": 0.058313060527940164,
|
| 5651 |
+
"grad_norm": 1.1684767007827759,
|
| 5652 |
+
"learning_rate": 9.177486926475507e-05,
|
| 5653 |
+
"loss": 0.3521,
|
| 5654 |
+
"step": 803
|
| 5655 |
+
},
|
| 5656 |
+
{
|
| 5657 |
+
"epoch": 0.05838567953233361,
|
| 5658 |
+
"grad_norm": 1.4156098365783691,
|
| 5659 |
+
"learning_rate": 9.156712698610547e-05,
|
| 5660 |
+
"loss": 0.9906,
|
| 5661 |
+
"step": 804
|
| 5662 |
+
},
|
| 5663 |
+
{
|
| 5664 |
+
"epoch": 0.058458298536727064,
|
| 5665 |
+
"grad_norm": 0.6224841475486755,
|
| 5666 |
+
"learning_rate": 9.135942135531796e-05,
|
| 5667 |
+
"loss": 0.4158,
|
| 5668 |
+
"step": 805
|
| 5669 |
+
},
|
| 5670 |
+
{
|
| 5671 |
+
"epoch": 0.05853091754112051,
|
| 5672 |
+
"grad_norm": 1.1172404289245605,
|
| 5673 |
+
"learning_rate": 9.115175327504655e-05,
|
| 5674 |
+
"loss": 0.4146,
|
| 5675 |
+
"step": 806
|
| 5676 |
+
},
|
| 5677 |
+
{
|
| 5678 |
+
"epoch": 0.05860353654551396,
|
| 5679 |
+
"grad_norm": 0.8851704001426697,
|
| 5680 |
+
"learning_rate": 9.094412364778221e-05,
|
| 5681 |
+
"loss": 0.5798,
|
| 5682 |
+
"step": 807
|
| 5683 |
+
},
|
| 5684 |
+
{
|
| 5685 |
+
"epoch": 0.05867615554990741,
|
| 5686 |
+
"grad_norm": 1.035550594329834,
|
| 5687 |
+
"learning_rate": 9.07365333758487e-05,
|
| 5688 |
+
"loss": 0.7614,
|
| 5689 |
+
"step": 808
|
| 5690 |
+
},
|
| 5691 |
+
{
|
| 5692 |
+
"epoch": 0.05874877455430086,
|
| 5693 |
+
"grad_norm": 0.9061303734779358,
|
| 5694 |
+
"learning_rate": 9.05289833613988e-05,
|
| 5695 |
+
"loss": 0.576,
|
| 5696 |
+
"step": 809
|
| 5697 |
+
},
|
| 5698 |
+
{
|
| 5699 |
+
"epoch": 0.05882139355869431,
|
| 5700 |
+
"grad_norm": 0.9312248229980469,
|
| 5701 |
+
"learning_rate": 9.032147450641028e-05,
|
| 5702 |
+
"loss": 0.2855,
|
| 5703 |
+
"step": 810
|
| 5704 |
+
},
|
| 5705 |
+
{
|
| 5706 |
+
"epoch": 0.05889401256308776,
|
| 5707 |
+
"grad_norm": 1.2476495504379272,
|
| 5708 |
+
"learning_rate": 9.01140077126821e-05,
|
| 5709 |
+
"loss": 0.7157,
|
| 5710 |
+
"step": 811
|
| 5711 |
+
},
|
| 5712 |
+
{
|
| 5713 |
+
"epoch": 0.058966631567481206,
|
| 5714 |
+
"grad_norm": 0.660798966884613,
|
| 5715 |
+
"learning_rate": 8.990658388183038e-05,
|
| 5716 |
+
"loss": 0.2476,
|
| 5717 |
+
"step": 812
|
| 5718 |
+
},
|
| 5719 |
+
{
|
| 5720 |
+
"epoch": 0.05903925057187466,
|
| 5721 |
+
"grad_norm": 1.4954866170883179,
|
| 5722 |
+
"learning_rate": 8.969920391528458e-05,
|
| 5723 |
+
"loss": 0.5552,
|
| 5724 |
+
"step": 813
|
| 5725 |
+
},
|
| 5726 |
+
{
|
| 5727 |
+
"epoch": 0.05911186957626811,
|
| 5728 |
+
"grad_norm": 0.8855932950973511,
|
| 5729 |
+
"learning_rate": 8.949186871428348e-05,
|
| 5730 |
+
"loss": 0.65,
|
| 5731 |
+
"step": 814
|
| 5732 |
+
},
|
| 5733 |
+
{
|
| 5734 |
+
"epoch": 0.05918448858066156,
|
| 5735 |
+
"grad_norm": 2.629485607147217,
|
| 5736 |
+
"learning_rate": 8.928457917987136e-05,
|
| 5737 |
+
"loss": 0.4325,
|
| 5738 |
+
"step": 815
|
| 5739 |
+
},
|
| 5740 |
+
{
|
| 5741 |
+
"epoch": 0.05925710758505501,
|
| 5742 |
+
"grad_norm": 0.5645459294319153,
|
| 5743 |
+
"learning_rate": 8.9077336212894e-05,
|
| 5744 |
+
"loss": 0.2746,
|
| 5745 |
+
"step": 816
|
| 5746 |
+
},
|
| 5747 |
+
{
|
| 5748 |
+
"epoch": 0.05932972658944846,
|
| 5749 |
+
"grad_norm": 1.0492244958877563,
|
| 5750 |
+
"learning_rate": 8.887014071399488e-05,
|
| 5751 |
+
"loss": 0.5085,
|
| 5752 |
+
"step": 817
|
| 5753 |
+
},
|
| 5754 |
+
{
|
| 5755 |
+
"epoch": 0.05940234559384191,
|
| 5756 |
+
"grad_norm": 0.8126876354217529,
|
| 5757 |
+
"learning_rate": 8.866299358361108e-05,
|
| 5758 |
+
"loss": 0.901,
|
| 5759 |
+
"step": 818
|
| 5760 |
+
},
|
| 5761 |
+
{
|
| 5762 |
+
"epoch": 0.059474964598235355,
|
| 5763 |
+
"grad_norm": 0.7482452392578125,
|
| 5764 |
+
"learning_rate": 8.845589572196961e-05,
|
| 5765 |
+
"loss": 0.3572,
|
| 5766 |
+
"step": 819
|
| 5767 |
+
},
|
| 5768 |
+
{
|
| 5769 |
+
"epoch": 0.05954758360262881,
|
| 5770 |
+
"grad_norm": 2.5570991039276123,
|
| 5771 |
+
"learning_rate": 8.82488480290832e-05,
|
| 5772 |
+
"loss": 1.3839,
|
| 5773 |
+
"step": 820
|
| 5774 |
+
},
|
| 5775 |
+
{
|
| 5776 |
+
"epoch": 0.059620202607022256,
|
| 5777 |
+
"grad_norm": 0.7802184224128723,
|
| 5778 |
+
"learning_rate": 8.804185140474675e-05,
|
| 5779 |
+
"loss": 0.4388,
|
| 5780 |
+
"step": 821
|
| 5781 |
+
},
|
| 5782 |
+
{
|
| 5783 |
+
"epoch": 0.05969282161141571,
|
| 5784 |
+
"grad_norm": 1.984308123588562,
|
| 5785 |
+
"learning_rate": 8.783490674853302e-05,
|
| 5786 |
+
"loss": 0.571,
|
| 5787 |
+
"step": 822
|
| 5788 |
+
},
|
| 5789 |
+
{
|
| 5790 |
+
"epoch": 0.05976544061580916,
|
| 5791 |
+
"grad_norm": 1.1061800718307495,
|
| 5792 |
+
"learning_rate": 8.762801495978914e-05,
|
| 5793 |
+
"loss": 0.3899,
|
| 5794 |
+
"step": 823
|
| 5795 |
+
},
|
| 5796 |
+
{
|
| 5797 |
+
"epoch": 0.05983805962020261,
|
| 5798 |
+
"grad_norm": 0.7607733011245728,
|
| 5799 |
+
"learning_rate": 8.742117693763227e-05,
|
| 5800 |
+
"loss": 0.6425,
|
| 5801 |
+
"step": 824
|
| 5802 |
+
},
|
| 5803 |
+
{
|
| 5804 |
+
"epoch": 0.05991067862459606,
|
| 5805 |
+
"grad_norm": 0.7867072224617004,
|
| 5806 |
+
"learning_rate": 8.721439358094612e-05,
|
| 5807 |
+
"loss": 0.3864,
|
| 5808 |
+
"step": 825
|
| 5809 |
+
},
|
| 5810 |
+
{
|
| 5811 |
+
"epoch": 0.059983297628989504,
|
| 5812 |
+
"grad_norm": 2.6383373737335205,
|
| 5813 |
+
"learning_rate": 8.700766578837664e-05,
|
| 5814 |
+
"loss": 0.6902,
|
| 5815 |
+
"step": 826
|
| 5816 |
+
},
|
| 5817 |
+
{
|
| 5818 |
+
"epoch": 0.06005591663338296,
|
| 5819 |
+
"grad_norm": 0.5204357504844666,
|
| 5820 |
+
"learning_rate": 8.680099445832842e-05,
|
| 5821 |
+
"loss": 0.1247,
|
| 5822 |
+
"step": 827
|
| 5823 |
+
},
|
| 5824 |
+
{
|
| 5825 |
+
"epoch": 0.060128535637776405,
|
| 5826 |
+
"grad_norm": 1.4301972389221191,
|
| 5827 |
+
"learning_rate": 8.659438048896071e-05,
|
| 5828 |
+
"loss": 1.4639,
|
| 5829 |
+
"step": 828
|
| 5830 |
+
},
|
| 5831 |
+
{
|
| 5832 |
+
"epoch": 0.06020115464216986,
|
| 5833 |
+
"grad_norm": 0.41357865929603577,
|
| 5834 |
+
"learning_rate": 8.638782477818334e-05,
|
| 5835 |
+
"loss": 0.1834,
|
| 5836 |
+
"step": 829
|
| 5837 |
+
},
|
| 5838 |
+
{
|
| 5839 |
+
"epoch": 0.060273773646563306,
|
| 5840 |
+
"grad_norm": 0.8870722651481628,
|
| 5841 |
+
"learning_rate": 8.618132822365311e-05,
|
| 5842 |
+
"loss": 0.4457,
|
| 5843 |
+
"step": 830
|
| 5844 |
+
},
|
| 5845 |
+
{
|
| 5846 |
+
"epoch": 0.06034639265095675,
|
| 5847 |
+
"grad_norm": 0.7810840606689453,
|
| 5848 |
+
"learning_rate": 8.597489172276957e-05,
|
| 5849 |
+
"loss": 0.3177,
|
| 5850 |
+
"step": 831
|
| 5851 |
+
},
|
| 5852 |
+
{
|
| 5853 |
+
"epoch": 0.060419011655350206,
|
| 5854 |
+
"grad_norm": 0.6351898312568665,
|
| 5855 |
+
"learning_rate": 8.57685161726715e-05,
|
| 5856 |
+
"loss": 0.3488,
|
| 5857 |
+
"step": 832
|
| 5858 |
+
},
|
| 5859 |
+
{
|
| 5860 |
+
"epoch": 0.06049163065974365,
|
| 5861 |
+
"grad_norm": 0.6372057795524597,
|
| 5862 |
+
"learning_rate": 8.556220247023256e-05,
|
| 5863 |
+
"loss": 0.767,
|
| 5864 |
+
"step": 833
|
| 5865 |
+
},
|
| 5866 |
+
{
|
| 5867 |
+
"epoch": 0.06056424966413711,
|
| 5868 |
+
"grad_norm": 1.307631015777588,
|
| 5869 |
+
"learning_rate": 8.535595151205786e-05,
|
| 5870 |
+
"loss": 0.8996,
|
| 5871 |
+
"step": 834
|
| 5872 |
+
},
|
| 5873 |
+
{
|
| 5874 |
+
"epoch": 0.060636868668530554,
|
| 5875 |
+
"grad_norm": 1.5109087228775024,
|
| 5876 |
+
"learning_rate": 8.514976419447964e-05,
|
| 5877 |
+
"loss": 0.7699,
|
| 5878 |
+
"step": 835
|
| 5879 |
+
},
|
| 5880 |
+
{
|
| 5881 |
+
"epoch": 0.06070948767292401,
|
| 5882 |
+
"grad_norm": 1.5807245969772339,
|
| 5883 |
+
"learning_rate": 8.494364141355368e-05,
|
| 5884 |
+
"loss": 0.9269,
|
| 5885 |
+
"step": 836
|
| 5886 |
+
},
|
| 5887 |
+
{
|
| 5888 |
+
"epoch": 0.060782106677317455,
|
| 5889 |
+
"grad_norm": 1.7503530979156494,
|
| 5890 |
+
"learning_rate": 8.47375840650553e-05,
|
| 5891 |
+
"loss": 0.825,
|
| 5892 |
+
"step": 837
|
| 5893 |
+
},
|
| 5894 |
+
{
|
| 5895 |
+
"epoch": 0.0608547256817109,
|
| 5896 |
+
"grad_norm": 0.9314043521881104,
|
| 5897 |
+
"learning_rate": 8.453159304447534e-05,
|
| 5898 |
+
"loss": 0.1621,
|
| 5899 |
+
"step": 838
|
| 5900 |
+
},
|
| 5901 |
+
{
|
| 5902 |
+
"epoch": 0.060927344686104355,
|
| 5903 |
+
"grad_norm": 1.5196501016616821,
|
| 5904 |
+
"learning_rate": 8.432566924701659e-05,
|
| 5905 |
+
"loss": 1.0021,
|
| 5906 |
+
"step": 839
|
| 5907 |
+
},
|
| 5908 |
+
{
|
| 5909 |
+
"epoch": 0.0609999636904978,
|
| 5910 |
+
"grad_norm": 0.8872591853141785,
|
| 5911 |
+
"learning_rate": 8.41198135675895e-05,
|
| 5912 |
+
"loss": 0.3018,
|
| 5913 |
+
"step": 840
|
| 5914 |
+
},
|
| 5915 |
+
{
|
| 5916 |
+
"epoch": 0.061072582694891256,
|
| 5917 |
+
"grad_norm": 0.8934780955314636,
|
| 5918 |
+
"learning_rate": 8.391402690080864e-05,
|
| 5919 |
+
"loss": 0.6862,
|
| 5920 |
+
"step": 841
|
| 5921 |
+
},
|
| 5922 |
+
{
|
| 5923 |
+
"epoch": 0.0611452016992847,
|
| 5924 |
+
"grad_norm": 0.7138051390647888,
|
| 5925 |
+
"learning_rate": 8.370831014098854e-05,
|
| 5926 |
+
"loss": 0.2865,
|
| 5927 |
+
"step": 842
|
| 5928 |
+
},
|
| 5929 |
+
{
|
| 5930 |
+
"epoch": 0.06121782070367815,
|
| 5931 |
+
"grad_norm": 0.8391185402870178,
|
| 5932 |
+
"learning_rate": 8.350266418214006e-05,
|
| 5933 |
+
"loss": 0.4947,
|
| 5934 |
+
"step": 843
|
| 5935 |
+
},
|
| 5936 |
+
{
|
| 5937 |
+
"epoch": 0.0612904397080716,
|
| 5938 |
+
"grad_norm": 0.9233042597770691,
|
| 5939 |
+
"learning_rate": 8.329708991796619e-05,
|
| 5940 |
+
"loss": 0.6456,
|
| 5941 |
+
"step": 844
|
| 5942 |
+
},
|
| 5943 |
+
{
|
| 5944 |
+
"epoch": 0.06136305871246505,
|
| 5945 |
+
"grad_norm": 1.451659917831421,
|
| 5946 |
+
"learning_rate": 8.309158824185854e-05,
|
| 5947 |
+
"loss": 1.4678,
|
| 5948 |
+
"step": 845
|
| 5949 |
+
},
|
| 5950 |
+
{
|
| 5951 |
+
"epoch": 0.061435677716858504,
|
| 5952 |
+
"grad_norm": 1.0687843561172485,
|
| 5953 |
+
"learning_rate": 8.28861600468932e-05,
|
| 5954 |
+
"loss": 1.3211,
|
| 5955 |
+
"step": 846
|
| 5956 |
+
},
|
| 5957 |
+
{
|
| 5958 |
+
"epoch": 0.06150829672125195,
|
| 5959 |
+
"grad_norm": 0.9472249150276184,
|
| 5960 |
+
"learning_rate": 8.268080622582682e-05,
|
| 5961 |
+
"loss": 1.1012,
|
| 5962 |
+
"step": 847
|
| 5963 |
+
},
|
| 5964 |
+
{
|
| 5965 |
+
"epoch": 0.061580915725645405,
|
| 5966 |
+
"grad_norm": 1.0834053754806519,
|
| 5967 |
+
"learning_rate": 8.2475527671093e-05,
|
| 5968 |
+
"loss": 0.3143,
|
| 5969 |
+
"step": 848
|
| 5970 |
+
},
|
| 5971 |
+
{
|
| 5972 |
+
"epoch": 0.06165353473003885,
|
| 5973 |
+
"grad_norm": 0.7939391136169434,
|
| 5974 |
+
"learning_rate": 8.227032527479806e-05,
|
| 5975 |
+
"loss": 0.3322,
|
| 5976 |
+
"step": 849
|
| 5977 |
+
},
|
| 5978 |
+
{
|
| 5979 |
+
"epoch": 0.0617261537344323,
|
| 5980 |
+
"grad_norm": 1.2913670539855957,
|
| 5981 |
+
"learning_rate": 8.206519992871756e-05,
|
| 5982 |
+
"loss": 1.1112,
|
| 5983 |
+
"step": 850
|
| 5984 |
+
},
|
| 5985 |
+
{
|
| 5986 |
+
"epoch": 0.06179877273882575,
|
| 5987 |
+
"grad_norm": 0.7568597197532654,
|
| 5988 |
+
"learning_rate": 8.1860152524292e-05,
|
| 5989 |
+
"loss": 0.5025,
|
| 5990 |
+
"step": 851
|
| 5991 |
+
},
|
| 5992 |
+
{
|
| 5993 |
+
"epoch": 0.0618713917432192,
|
| 5994 |
+
"grad_norm": 0.7415629029273987,
|
| 5995 |
+
"learning_rate": 8.165518395262335e-05,
|
| 5996 |
+
"loss": 0.7002,
|
| 5997 |
+
"step": 852
|
| 5998 |
+
},
|
| 5999 |
+
{
|
| 6000 |
+
"epoch": 0.06194401074761265,
|
| 6001 |
+
"grad_norm": 1.0701147317886353,
|
| 6002 |
+
"learning_rate": 8.14502951044708e-05,
|
| 6003 |
+
"loss": 0.3526,
|
| 6004 |
+
"step": 853
|
| 6005 |
+
},
|
| 6006 |
+
{
|
| 6007 |
+
"epoch": 0.0620166297520061,
|
| 6008 |
+
"grad_norm": 0.8759703636169434,
|
| 6009 |
+
"learning_rate": 8.124548687024723e-05,
|
| 6010 |
+
"loss": 1.0995,
|
| 6011 |
+
"step": 854
|
| 6012 |
+
},
|
| 6013 |
+
{
|
| 6014 |
+
"epoch": 0.06208924875639955,
|
| 6015 |
+
"grad_norm": 0.9400107860565186,
|
| 6016 |
+
"learning_rate": 8.104076014001514e-05,
|
| 6017 |
+
"loss": 0.3314,
|
| 6018 |
+
"step": 855
|
| 6019 |
+
},
|
| 6020 |
+
{
|
| 6021 |
+
"epoch": 0.062161867760793,
|
| 6022 |
+
"grad_norm": 1.0083138942718506,
|
| 6023 |
+
"learning_rate": 8.083611580348277e-05,
|
| 6024 |
+
"loss": 0.3453,
|
| 6025 |
+
"step": 856
|
| 6026 |
+
},
|
| 6027 |
+
{
|
| 6028 |
+
"epoch": 0.06223448676518645,
|
| 6029 |
+
"grad_norm": 0.9074184894561768,
|
| 6030 |
+
"learning_rate": 8.063155475000037e-05,
|
| 6031 |
+
"loss": 0.4892,
|
| 6032 |
+
"step": 857
|
| 6033 |
+
},
|
| 6034 |
+
{
|
| 6035 |
+
"epoch": 0.0623071057695799,
|
| 6036 |
+
"grad_norm": 1.6481777429580688,
|
| 6037 |
+
"learning_rate": 8.042707786855617e-05,
|
| 6038 |
+
"loss": 0.7171,
|
| 6039 |
+
"step": 858
|
| 6040 |
+
},
|
| 6041 |
+
{
|
| 6042 |
+
"epoch": 0.06237972477397335,
|
| 6043 |
+
"grad_norm": 1.227709174156189,
|
| 6044 |
+
"learning_rate": 8.022268604777271e-05,
|
| 6045 |
+
"loss": 0.7768,
|
| 6046 |
+
"step": 859
|
| 6047 |
+
},
|
| 6048 |
+
{
|
| 6049 |
+
"epoch": 0.0624523437783668,
|
| 6050 |
+
"grad_norm": 0.6494023203849792,
|
| 6051 |
+
"learning_rate": 8.001838017590277e-05,
|
| 6052 |
+
"loss": 0.2314,
|
| 6053 |
+
"step": 860
|
| 6054 |
+
},
|
| 6055 |
+
{
|
| 6056 |
+
"epoch": 0.06252496278276025,
|
| 6057 |
+
"grad_norm": 1.3288606405258179,
|
| 6058 |
+
"learning_rate": 7.981416114082568e-05,
|
| 6059 |
+
"loss": 0.4127,
|
| 6060 |
+
"step": 861
|
| 6061 |
+
},
|
| 6062 |
+
{
|
| 6063 |
+
"epoch": 0.0625975817871537,
|
| 6064 |
+
"grad_norm": 0.9424789547920227,
|
| 6065 |
+
"learning_rate": 7.961002983004335e-05,
|
| 6066 |
+
"loss": 0.7112,
|
| 6067 |
+
"step": 862
|
| 6068 |
+
},
|
| 6069 |
+
{
|
| 6070 |
+
"epoch": 0.06267020079154714,
|
| 6071 |
+
"grad_norm": 1.2619773149490356,
|
| 6072 |
+
"learning_rate": 7.940598713067646e-05,
|
| 6073 |
+
"loss": 1.2998,
|
| 6074 |
+
"step": 863
|
| 6075 |
+
},
|
| 6076 |
+
{
|
| 6077 |
+
"epoch": 0.0627428197959406,
|
| 6078 |
+
"grad_norm": 1.4239133596420288,
|
| 6079 |
+
"learning_rate": 7.920203392946065e-05,
|
| 6080 |
+
"loss": 1.1869,
|
| 6081 |
+
"step": 864
|
| 6082 |
+
},
|
| 6083 |
+
{
|
| 6084 |
+
"epoch": 0.06281543880033405,
|
| 6085 |
+
"grad_norm": 1.0152820348739624,
|
| 6086 |
+
"learning_rate": 7.899817111274248e-05,
|
| 6087 |
+
"loss": 0.3258,
|
| 6088 |
+
"step": 865
|
| 6089 |
+
},
|
| 6090 |
+
{
|
| 6091 |
+
"epoch": 0.0628880578047275,
|
| 6092 |
+
"grad_norm": 1.1045777797698975,
|
| 6093 |
+
"learning_rate": 7.879439956647594e-05,
|
| 6094 |
+
"loss": 0.244,
|
| 6095 |
+
"step": 866
|
| 6096 |
+
},
|
| 6097 |
+
{
|
| 6098 |
+
"epoch": 0.06296067680912094,
|
| 6099 |
+
"grad_norm": 1.7886762619018555,
|
| 6100 |
+
"learning_rate": 7.859072017621812e-05,
|
| 6101 |
+
"loss": 0.9382,
|
| 6102 |
+
"step": 867
|
| 6103 |
+
},
|
| 6104 |
+
{
|
| 6105 |
+
"epoch": 0.0630332958135144,
|
| 6106 |
+
"grad_norm": 1.1358466148376465,
|
| 6107 |
+
"learning_rate": 7.838713382712583e-05,
|
| 6108 |
+
"loss": 0.2925,
|
| 6109 |
+
"step": 868
|
| 6110 |
+
},
|
| 6111 |
+
{
|
| 6112 |
+
"epoch": 0.06310591481790785,
|
| 6113 |
+
"grad_norm": 1.4714925289154053,
|
| 6114 |
+
"learning_rate": 7.818364140395137e-05,
|
| 6115 |
+
"loss": 0.1758,
|
| 6116 |
+
"step": 869
|
| 6117 |
+
},
|
| 6118 |
+
{
|
| 6119 |
+
"epoch": 0.06317853382230129,
|
| 6120 |
+
"grad_norm": 1.0500401258468628,
|
| 6121 |
+
"learning_rate": 7.798024379103903e-05,
|
| 6122 |
+
"loss": 0.1407,
|
| 6123 |
+
"step": 870
|
| 6124 |
+
},
|
| 6125 |
+
{
|
| 6126 |
+
"epoch": 0.06325115282669475,
|
| 6127 |
+
"grad_norm": 0.9449998736381531,
|
| 6128 |
+
"learning_rate": 7.777694187232085e-05,
|
| 6129 |
+
"loss": 0.3281,
|
| 6130 |
+
"step": 871
|
| 6131 |
+
},
|
| 6132 |
+
{
|
| 6133 |
+
"epoch": 0.0633237718310882,
|
| 6134 |
+
"grad_norm": 1.001770257949829,
|
| 6135 |
+
"learning_rate": 7.757373653131322e-05,
|
| 6136 |
+
"loss": 0.6883,
|
| 6137 |
+
"step": 872
|
| 6138 |
+
},
|
| 6139 |
+
{
|
| 6140 |
+
"epoch": 0.06339639083548164,
|
| 6141 |
+
"grad_norm": 1.5930564403533936,
|
| 6142 |
+
"learning_rate": 7.737062865111264e-05,
|
| 6143 |
+
"loss": 1.0698,
|
| 6144 |
+
"step": 873
|
| 6145 |
+
},
|
| 6146 |
+
{
|
| 6147 |
+
"epoch": 0.06346900983987509,
|
| 6148 |
+
"grad_norm": 1.1424990892410278,
|
| 6149 |
+
"learning_rate": 7.716761911439216e-05,
|
| 6150 |
+
"loss": 0.5009,
|
| 6151 |
+
"step": 874
|
| 6152 |
+
},
|
| 6153 |
+
{
|
| 6154 |
+
"epoch": 0.06354162884426855,
|
| 6155 |
+
"grad_norm": 2.4575681686401367,
|
| 6156 |
+
"learning_rate": 7.69647088033975e-05,
|
| 6157 |
+
"loss": 1.098,
|
| 6158 |
+
"step": 875
|
| 6159 |
+
},
|
| 6160 |
+
{
|
| 6161 |
+
"epoch": 0.063614247848662,
|
| 6162 |
+
"grad_norm": 1.0910656452178955,
|
| 6163 |
+
"learning_rate": 7.676189859994298e-05,
|
| 6164 |
+
"loss": 0.7186,
|
| 6165 |
+
"step": 876
|
| 6166 |
+
},
|
| 6167 |
+
{
|
| 6168 |
+
"epoch": 0.06368686685305544,
|
| 6169 |
+
"grad_norm": 0.5516690611839294,
|
| 6170 |
+
"learning_rate": 7.655918938540808e-05,
|
| 6171 |
+
"loss": 0.3268,
|
| 6172 |
+
"step": 877
|
| 6173 |
+
},
|
| 6174 |
+
{
|
| 6175 |
+
"epoch": 0.0637594858574489,
|
| 6176 |
+
"grad_norm": 0.6600070595741272,
|
| 6177 |
+
"learning_rate": 7.63565820407332e-05,
|
| 6178 |
+
"loss": 0.2945,
|
| 6179 |
+
"step": 878
|
| 6180 |
+
},
|
| 6181 |
+
{
|
| 6182 |
+
"epoch": 0.06383210486184235,
|
| 6183 |
+
"grad_norm": 1.2723064422607422,
|
| 6184 |
+
"learning_rate": 7.615407744641619e-05,
|
| 6185 |
+
"loss": 0.9468,
|
| 6186 |
+
"step": 879
|
| 6187 |
+
},
|
| 6188 |
+
{
|
| 6189 |
+
"epoch": 0.06390472386623579,
|
| 6190 |
+
"grad_norm": 0.9849138259887695,
|
| 6191 |
+
"learning_rate": 7.595167648250825e-05,
|
| 6192 |
+
"loss": 0.8803,
|
| 6193 |
+
"step": 880
|
| 6194 |
+
},
|
| 6195 |
+
{
|
| 6196 |
+
"epoch": 0.06397734287062924,
|
| 6197 |
+
"grad_norm": 1.043287992477417,
|
| 6198 |
+
"learning_rate": 7.574938002861033e-05,
|
| 6199 |
+
"loss": 1.0998,
|
| 6200 |
+
"step": 881
|
| 6201 |
+
},
|
| 6202 |
+
{
|
| 6203 |
+
"epoch": 0.0640499618750227,
|
| 6204 |
+
"grad_norm": 1.1217182874679565,
|
| 6205 |
+
"learning_rate": 7.554718896386906e-05,
|
| 6206 |
+
"loss": 0.4497,
|
| 6207 |
+
"step": 882
|
| 6208 |
+
},
|
| 6209 |
+
{
|
| 6210 |
+
"epoch": 0.06412258087941615,
|
| 6211 |
+
"grad_norm": 1.0717918872833252,
|
| 6212 |
+
"learning_rate": 7.534510416697322e-05,
|
| 6213 |
+
"loss": 0.7797,
|
| 6214 |
+
"step": 883
|
| 6215 |
+
},
|
| 6216 |
+
{
|
| 6217 |
+
"epoch": 0.06419519988380959,
|
| 6218 |
+
"grad_norm": 2.1684038639068604,
|
| 6219 |
+
"learning_rate": 7.514312651614964e-05,
|
| 6220 |
+
"loss": 1.4668,
|
| 6221 |
+
"step": 884
|
| 6222 |
+
},
|
| 6223 |
+
{
|
| 6224 |
+
"epoch": 0.06426781888820304,
|
| 6225 |
+
"grad_norm": 0.8061167597770691,
|
| 6226 |
+
"learning_rate": 7.494125688915959e-05,
|
| 6227 |
+
"loss": 0.6487,
|
| 6228 |
+
"step": 885
|
| 6229 |
+
},
|
| 6230 |
+
{
|
| 6231 |
+
"epoch": 0.0643404378925965,
|
| 6232 |
+
"grad_norm": 1.2220314741134644,
|
| 6233 |
+
"learning_rate": 7.473949616329487e-05,
|
| 6234 |
+
"loss": 0.9514,
|
| 6235 |
+
"step": 886
|
| 6236 |
+
},
|
| 6237 |
+
{
|
| 6238 |
+
"epoch": 0.06441305689698994,
|
| 6239 |
+
"grad_norm": 1.135111927986145,
|
| 6240 |
+
"learning_rate": 7.453784521537392e-05,
|
| 6241 |
+
"loss": 1.068,
|
| 6242 |
+
"step": 887
|
| 6243 |
+
},
|
| 6244 |
+
{
|
| 6245 |
+
"epoch": 0.06448567590138339,
|
| 6246 |
+
"grad_norm": 0.7935314774513245,
|
| 6247 |
+
"learning_rate": 7.43363049217383e-05,
|
| 6248 |
+
"loss": 0.4185,
|
| 6249 |
+
"step": 888
|
| 6250 |
+
},
|
| 6251 |
+
{
|
| 6252 |
+
"epoch": 0.06455829490577684,
|
| 6253 |
+
"grad_norm": 1.032720923423767,
|
| 6254 |
+
"learning_rate": 7.413487615824847e-05,
|
| 6255 |
+
"loss": 0.7423,
|
| 6256 |
+
"step": 889
|
| 6257 |
+
},
|
| 6258 |
+
{
|
| 6259 |
+
"epoch": 0.0646309139101703,
|
| 6260 |
+
"grad_norm": 0.8119061589241028,
|
| 6261 |
+
"learning_rate": 7.393355980028039e-05,
|
| 6262 |
+
"loss": 0.5868,
|
| 6263 |
+
"step": 890
|
| 6264 |
+
},
|
| 6265 |
+
{
|
| 6266 |
+
"epoch": 0.06470353291456374,
|
| 6267 |
+
"grad_norm": 0.6888155937194824,
|
| 6268 |
+
"learning_rate": 7.373235672272131e-05,
|
| 6269 |
+
"loss": 0.8889,
|
| 6270 |
+
"step": 891
|
| 6271 |
+
},
|
| 6272 |
+
{
|
| 6273 |
+
"epoch": 0.06477615191895719,
|
| 6274 |
+
"grad_norm": 1.440739393234253,
|
| 6275 |
+
"learning_rate": 7.353126779996639e-05,
|
| 6276 |
+
"loss": 0.5643,
|
| 6277 |
+
"step": 892
|
| 6278 |
+
},
|
| 6279 |
+
{
|
| 6280 |
+
"epoch": 0.06484877092335065,
|
| 6281 |
+
"grad_norm": 0.3748563230037689,
|
| 6282 |
+
"learning_rate": 7.333029390591467e-05,
|
| 6283 |
+
"loss": 0.1296,
|
| 6284 |
+
"step": 893
|
| 6285 |
+
},
|
| 6286 |
+
{
|
| 6287 |
+
"epoch": 0.06492138992774409,
|
| 6288 |
+
"grad_norm": 0.951309323310852,
|
| 6289 |
+
"learning_rate": 7.312943591396508e-05,
|
| 6290 |
+
"loss": 0.3688,
|
| 6291 |
+
"step": 894
|
| 6292 |
+
},
|
| 6293 |
+
{
|
| 6294 |
+
"epoch": 0.06499400893213754,
|
| 6295 |
+
"grad_norm": 1.4929016828536987,
|
| 6296 |
+
"learning_rate": 7.292869469701315e-05,
|
| 6297 |
+
"loss": 0.1517,
|
| 6298 |
+
"step": 895
|
| 6299 |
+
},
|
| 6300 |
+
{
|
| 6301 |
+
"epoch": 0.065066627936531,
|
| 6302 |
+
"grad_norm": 0.5714091658592224,
|
| 6303 |
+
"learning_rate": 7.272807112744666e-05,
|
| 6304 |
+
"loss": 0.3717,
|
| 6305 |
+
"step": 896
|
| 6306 |
+
},
|
| 6307 |
+
{
|
| 6308 |
+
"epoch": 0.06513924694092443,
|
| 6309 |
+
"grad_norm": 1.1853396892547607,
|
| 6310 |
+
"learning_rate": 7.252756607714237e-05,
|
| 6311 |
+
"loss": 0.8886,
|
| 6312 |
+
"step": 897
|
| 6313 |
+
},
|
| 6314 |
+
{
|
| 6315 |
+
"epoch": 0.06521186594531789,
|
| 6316 |
+
"grad_norm": 0.8971914052963257,
|
| 6317 |
+
"learning_rate": 7.232718041746172e-05,
|
| 6318 |
+
"loss": 0.3482,
|
| 6319 |
+
"step": 898
|
| 6320 |
+
},
|
| 6321 |
+
{
|
| 6322 |
+
"epoch": 0.06528448494971134,
|
| 6323 |
+
"grad_norm": 0.9511501789093018,
|
| 6324 |
+
"learning_rate": 7.212691501924753e-05,
|
| 6325 |
+
"loss": 0.5241,
|
| 6326 |
+
"step": 899
|
| 6327 |
+
},
|
| 6328 |
+
{
|
| 6329 |
+
"epoch": 0.0653571039541048,
|
| 6330 |
+
"grad_norm": 1.5566978454589844,
|
| 6331 |
+
"learning_rate": 7.192677075281978e-05,
|
| 6332 |
+
"loss": 0.8287,
|
| 6333 |
+
"step": 900
|
| 6334 |
+
},
|
| 6335 |
+
{
|
| 6336 |
+
"epoch": 0.06542972295849823,
|
| 6337 |
+
"grad_norm": 3.092409133911133,
|
| 6338 |
+
"learning_rate": 7.172674848797219e-05,
|
| 6339 |
+
"loss": 0.7043,
|
| 6340 |
+
"step": 901
|
| 6341 |
+
},
|
| 6342 |
+
{
|
| 6343 |
+
"epoch": 0.06550234196289169,
|
| 6344 |
+
"grad_norm": 0.41825735569000244,
|
| 6345 |
+
"learning_rate": 7.152684909396822e-05,
|
| 6346 |
+
"loss": 0.1668,
|
| 6347 |
+
"step": 902
|
| 6348 |
+
},
|
| 6349 |
+
{
|
| 6350 |
+
"epoch": 0.06557496096728514,
|
| 6351 |
+
"grad_norm": 0.7303375005722046,
|
| 6352 |
+
"learning_rate": 7.132707343953732e-05,
|
| 6353 |
+
"loss": 0.3666,
|
| 6354 |
+
"step": 903
|
| 6355 |
+
},
|
| 6356 |
+
{
|
| 6357 |
+
"epoch": 0.06564757997167858,
|
| 6358 |
+
"grad_norm": 0.9642083048820496,
|
| 6359 |
+
"learning_rate": 7.112742239287126e-05,
|
| 6360 |
+
"loss": 0.8572,
|
| 6361 |
+
"step": 904
|
| 6362 |
+
},
|
| 6363 |
+
{
|
| 6364 |
+
"epoch": 0.06572019897607204,
|
| 6365 |
+
"grad_norm": 0.8542434573173523,
|
| 6366 |
+
"learning_rate": 7.092789682162024e-05,
|
| 6367 |
+
"loss": 0.399,
|
| 6368 |
+
"step": 905
|
| 6369 |
+
},
|
| 6370 |
+
{
|
| 6371 |
+
"epoch": 0.06579281798046549,
|
| 6372 |
+
"grad_norm": 0.8454334139823914,
|
| 6373 |
+
"learning_rate": 7.072849759288912e-05,
|
| 6374 |
+
"loss": 0.4181,
|
| 6375 |
+
"step": 906
|
| 6376 |
+
},
|
| 6377 |
+
{
|
| 6378 |
+
"epoch": 0.06586543698485894,
|
| 6379 |
+
"grad_norm": 0.9029088020324707,
|
| 6380 |
+
"learning_rate": 7.052922557323383e-05,
|
| 6381 |
+
"loss": 0.4566,
|
| 6382 |
+
"step": 907
|
| 6383 |
+
},
|
| 6384 |
+
{
|
| 6385 |
+
"epoch": 0.06593805598925238,
|
| 6386 |
+
"grad_norm": 0.9722921252250671,
|
| 6387 |
+
"learning_rate": 7.033008162865732e-05,
|
| 6388 |
+
"loss": 0.3817,
|
| 6389 |
+
"step": 908
|
| 6390 |
+
},
|
| 6391 |
+
{
|
| 6392 |
+
"epoch": 0.06601067499364584,
|
| 6393 |
+
"grad_norm": 0.9663965106010437,
|
| 6394 |
+
"learning_rate": 7.013106662460604e-05,
|
| 6395 |
+
"loss": 0.2543,
|
| 6396 |
+
"step": 909
|
| 6397 |
+
},
|
| 6398 |
+
{
|
| 6399 |
+
"epoch": 0.06608329399803929,
|
| 6400 |
+
"grad_norm": 0.8726044297218323,
|
| 6401 |
+
"learning_rate": 6.993218142596608e-05,
|
| 6402 |
+
"loss": 0.2746,
|
| 6403 |
+
"step": 910
|
| 6404 |
+
},
|
| 6405 |
+
{
|
| 6406 |
+
"epoch": 0.06615591300243273,
|
| 6407 |
+
"grad_norm": 1.117026925086975,
|
| 6408 |
+
"learning_rate": 6.973342689705937e-05,
|
| 6409 |
+
"loss": 0.814,
|
| 6410 |
+
"step": 911
|
| 6411 |
+
},
|
| 6412 |
+
{
|
| 6413 |
+
"epoch": 0.06622853200682619,
|
| 6414 |
+
"grad_norm": 0.27998337149620056,
|
| 6415 |
+
"learning_rate": 6.953480390164e-05,
|
| 6416 |
+
"loss": 0.0982,
|
| 6417 |
+
"step": 912
|
| 6418 |
+
},
|
| 6419 |
+
{
|
| 6420 |
+
"epoch": 0.06630115101121964,
|
| 6421 |
+
"grad_norm": 1.1643705368041992,
|
| 6422 |
+
"learning_rate": 6.93363133028905e-05,
|
| 6423 |
+
"loss": 0.874,
|
| 6424 |
+
"step": 913
|
| 6425 |
+
},
|
| 6426 |
+
{
|
| 6427 |
+
"epoch": 0.06637377001561309,
|
| 6428 |
+
"grad_norm": 1.0442330837249756,
|
| 6429 |
+
"learning_rate": 6.913795596341785e-05,
|
| 6430 |
+
"loss": 0.3492,
|
| 6431 |
+
"step": 914
|
| 6432 |
+
},
|
| 6433 |
+
{
|
| 6434 |
+
"epoch": 0.06644638902000653,
|
| 6435 |
+
"grad_norm": 0.45305201411247253,
|
| 6436 |
+
"learning_rate": 6.893973274525014e-05,
|
| 6437 |
+
"loss": 0.2096,
|
| 6438 |
+
"step": 915
|
| 6439 |
+
},
|
| 6440 |
+
{
|
| 6441 |
+
"epoch": 0.06651900802439999,
|
| 6442 |
+
"grad_norm": 1.278076171875,
|
| 6443 |
+
"learning_rate": 6.874164450983238e-05,
|
| 6444 |
+
"loss": 0.3226,
|
| 6445 |
+
"step": 916
|
| 6446 |
+
},
|
| 6447 |
+
{
|
| 6448 |
+
"epoch": 0.06659162702879344,
|
| 6449 |
+
"grad_norm": 1.0749460458755493,
|
| 6450 |
+
"learning_rate": 6.854369211802314e-05,
|
| 6451 |
+
"loss": 0.6293,
|
| 6452 |
+
"step": 917
|
| 6453 |
+
},
|
| 6454 |
+
{
|
| 6455 |
+
"epoch": 0.06666424603318688,
|
| 6456 |
+
"grad_norm": 1.6141176223754883,
|
| 6457 |
+
"learning_rate": 6.834587643009047e-05,
|
| 6458 |
+
"loss": 0.8815,
|
| 6459 |
+
"step": 918
|
| 6460 |
+
},
|
| 6461 |
+
{
|
| 6462 |
+
"epoch": 0.06673686503758033,
|
| 6463 |
+
"grad_norm": 0.4306516647338867,
|
| 6464 |
+
"learning_rate": 6.81481983057085e-05,
|
| 6465 |
+
"loss": 0.285,
|
| 6466 |
+
"step": 919
|
| 6467 |
+
},
|
| 6468 |
+
{
|
| 6469 |
+
"epoch": 0.06680948404197379,
|
| 6470 |
+
"grad_norm": 0.957610547542572,
|
| 6471 |
+
"learning_rate": 6.795065860395335e-05,
|
| 6472 |
+
"loss": 0.3719,
|
| 6473 |
+
"step": 920
|
| 6474 |
+
},
|
| 6475 |
+
{
|
| 6476 |
+
"epoch": 0.06688210304636723,
|
| 6477 |
+
"grad_norm": 0.8341255784034729,
|
| 6478 |
+
"learning_rate": 6.775325818329974e-05,
|
| 6479 |
+
"loss": 0.3517,
|
| 6480 |
+
"step": 921
|
| 6481 |
+
},
|
| 6482 |
+
{
|
| 6483 |
+
"epoch": 0.06695472205076068,
|
| 6484 |
+
"grad_norm": 0.8621962070465088,
|
| 6485 |
+
"learning_rate": 6.755599790161707e-05,
|
| 6486 |
+
"loss": 0.6972,
|
| 6487 |
+
"step": 922
|
| 6488 |
+
},
|
| 6489 |
+
{
|
| 6490 |
+
"epoch": 0.06702734105515414,
|
| 6491 |
+
"grad_norm": 0.983747124671936,
|
| 6492 |
+
"learning_rate": 6.735887861616556e-05,
|
| 6493 |
+
"loss": 0.5894,
|
| 6494 |
+
"step": 923
|
| 6495 |
+
},
|
| 6496 |
+
{
|
| 6497 |
+
"epoch": 0.06709996005954759,
|
| 6498 |
+
"grad_norm": 1.1814996004104614,
|
| 6499 |
+
"learning_rate": 6.716190118359292e-05,
|
| 6500 |
+
"loss": 0.501,
|
| 6501 |
+
"step": 924
|
| 6502 |
+
},
|
| 6503 |
+
{
|
| 6504 |
+
"epoch": 0.06717257906394103,
|
| 6505 |
+
"grad_norm": 1.239450454711914,
|
| 6506 |
+
"learning_rate": 6.696506645993019e-05,
|
| 6507 |
+
"loss": 1.1688,
|
| 6508 |
+
"step": 925
|
| 6509 |
+
},
|
| 6510 |
+
{
|
| 6511 |
+
"epoch": 0.06724519806833448,
|
| 6512 |
+
"grad_norm": 0.5906079411506653,
|
| 6513 |
+
"learning_rate": 6.67683753005884e-05,
|
| 6514 |
+
"loss": 0.3475,
|
| 6515 |
+
"step": 926
|
| 6516 |
+
},
|
| 6517 |
+
{
|
| 6518 |
+
"epoch": 0.06731781707272794,
|
| 6519 |
+
"grad_norm": 0.8814708590507507,
|
| 6520 |
+
"learning_rate": 6.657182856035447e-05,
|
| 6521 |
+
"loss": 1.0399,
|
| 6522 |
+
"step": 927
|
| 6523 |
+
},
|
| 6524 |
+
{
|
| 6525 |
+
"epoch": 0.06739043607712138,
|
| 6526 |
+
"grad_norm": 0.8364510536193848,
|
| 6527 |
+
"learning_rate": 6.637542709338789e-05,
|
| 6528 |
+
"loss": 0.3515,
|
| 6529 |
+
"step": 928
|
| 6530 |
+
},
|
| 6531 |
+
{
|
| 6532 |
+
"epoch": 0.06746305508151483,
|
| 6533 |
+
"grad_norm": 0.63116854429245,
|
| 6534 |
+
"learning_rate": 6.617917175321669e-05,
|
| 6535 |
+
"loss": 0.3783,
|
| 6536 |
+
"step": 929
|
| 6537 |
+
},
|
| 6538 |
+
{
|
| 6539 |
+
"epoch": 0.06753567408590828,
|
| 6540 |
+
"grad_norm": 0.7996227741241455,
|
| 6541 |
+
"learning_rate": 6.598306339273393e-05,
|
| 6542 |
+
"loss": 0.8184,
|
| 6543 |
+
"step": 930
|
| 6544 |
+
},
|
| 6545 |
+
{
|
| 6546 |
+
"epoch": 0.06760829309030174,
|
| 6547 |
+
"grad_norm": 0.5852695107460022,
|
| 6548 |
+
"learning_rate": 6.578710286419387e-05,
|
| 6549 |
+
"loss": 0.1764,
|
| 6550 |
+
"step": 931
|
| 6551 |
+
},
|
| 6552 |
+
{
|
| 6553 |
+
"epoch": 0.06768091209469518,
|
| 6554 |
+
"grad_norm": 1.080872893333435,
|
| 6555 |
+
"learning_rate": 6.559129101920835e-05,
|
| 6556 |
+
"loss": 1.0413,
|
| 6557 |
+
"step": 932
|
| 6558 |
+
},
|
| 6559 |
+
{
|
| 6560 |
+
"epoch": 0.06775353109908863,
|
| 6561 |
+
"grad_norm": 1.5430817604064941,
|
| 6562 |
+
"learning_rate": 6.53956287087431e-05,
|
| 6563 |
+
"loss": 1.04,
|
| 6564 |
+
"step": 933
|
| 6565 |
+
},
|
| 6566 |
+
{
|
| 6567 |
+
"epoch": 0.06782615010348209,
|
| 6568 |
+
"grad_norm": 0.8056606650352478,
|
| 6569 |
+
"learning_rate": 6.520011678311382e-05,
|
| 6570 |
+
"loss": 0.4264,
|
| 6571 |
+
"step": 934
|
| 6572 |
+
},
|
| 6573 |
+
{
|
| 6574 |
+
"epoch": 0.06789876910787553,
|
| 6575 |
+
"grad_norm": 0.8869556188583374,
|
| 6576 |
+
"learning_rate": 6.50047560919829e-05,
|
| 6577 |
+
"loss": 0.6154,
|
| 6578 |
+
"step": 935
|
| 6579 |
+
},
|
| 6580 |
+
{
|
| 6581 |
+
"epoch": 0.06797138811226898,
|
| 6582 |
+
"grad_norm": 1.0641229152679443,
|
| 6583 |
+
"learning_rate": 6.48095474843553e-05,
|
| 6584 |
+
"loss": 0.3356,
|
| 6585 |
+
"step": 936
|
| 6586 |
+
},
|
| 6587 |
+
{
|
| 6588 |
+
"epoch": 0.06804400711666243,
|
| 6589 |
+
"grad_norm": 0.6757221221923828,
|
| 6590 |
+
"learning_rate": 6.461449180857522e-05,
|
| 6591 |
+
"loss": 0.7945,
|
| 6592 |
+
"step": 937
|
| 6593 |
+
},
|
| 6594 |
+
{
|
| 6595 |
+
"epoch": 0.06811662612105589,
|
| 6596 |
+
"grad_norm": 1.6167094707489014,
|
| 6597 |
+
"learning_rate": 6.441958991232202e-05,
|
| 6598 |
+
"loss": 1.0524,
|
| 6599 |
+
"step": 938
|
| 6600 |
+
},
|
| 6601 |
+
{
|
| 6602 |
+
"epoch": 0.06818924512544933,
|
| 6603 |
+
"grad_norm": 0.6140128374099731,
|
| 6604 |
+
"learning_rate": 6.422484264260698e-05,
|
| 6605 |
+
"loss": 0.2873,
|
| 6606 |
+
"step": 939
|
| 6607 |
+
},
|
| 6608 |
+
{
|
| 6609 |
+
"epoch": 0.06826186412984278,
|
| 6610 |
+
"grad_norm": 1.1318567991256714,
|
| 6611 |
+
"learning_rate": 6.403025084576932e-05,
|
| 6612 |
+
"loss": 0.9774,
|
| 6613 |
+
"step": 940
|
| 6614 |
+
},
|
| 6615 |
+
{
|
| 6616 |
+
"epoch": 0.06833448313423623,
|
| 6617 |
+
"grad_norm": 1.5785590410232544,
|
| 6618 |
+
"learning_rate": 6.383581536747252e-05,
|
| 6619 |
+
"loss": 0.9335,
|
| 6620 |
+
"step": 941
|
| 6621 |
+
},
|
| 6622 |
+
{
|
| 6623 |
+
"epoch": 0.06840710213862967,
|
| 6624 |
+
"grad_norm": 1.1314034461975098,
|
| 6625 |
+
"learning_rate": 6.364153705270088e-05,
|
| 6626 |
+
"loss": 0.6233,
|
| 6627 |
+
"step": 942
|
| 6628 |
+
},
|
| 6629 |
+
{
|
| 6630 |
+
"epoch": 0.06847972114302313,
|
| 6631 |
+
"grad_norm": 1.5190768241882324,
|
| 6632 |
+
"learning_rate": 6.344741674575553e-05,
|
| 6633 |
+
"loss": 0.3765,
|
| 6634 |
+
"step": 943
|
| 6635 |
+
},
|
| 6636 |
+
{
|
| 6637 |
+
"epoch": 0.06855234014741658,
|
| 6638 |
+
"grad_norm": 1.107803225517273,
|
| 6639 |
+
"learning_rate": 6.325345529025106e-05,
|
| 6640 |
+
"loss": 0.548,
|
| 6641 |
+
"step": 944
|
| 6642 |
+
},
|
| 6643 |
+
{
|
| 6644 |
+
"epoch": 0.06862495915181002,
|
| 6645 |
+
"grad_norm": 0.8528901934623718,
|
| 6646 |
+
"learning_rate": 6.305965352911161e-05,
|
| 6647 |
+
"loss": 0.3526,
|
| 6648 |
+
"step": 945
|
| 6649 |
+
},
|
| 6650 |
+
{
|
| 6651 |
+
"epoch": 0.06869757815620348,
|
| 6652 |
+
"grad_norm": 1.8871947526931763,
|
| 6653 |
+
"learning_rate": 6.286601230456743e-05,
|
| 6654 |
+
"loss": 0.6336,
|
| 6655 |
+
"step": 946
|
| 6656 |
+
},
|
| 6657 |
+
{
|
| 6658 |
+
"epoch": 0.06877019716059693,
|
| 6659 |
+
"grad_norm": 0.7002166509628296,
|
| 6660 |
+
"learning_rate": 6.267253245815097e-05,
|
| 6661 |
+
"loss": 0.3596,
|
| 6662 |
+
"step": 947
|
| 6663 |
+
},
|
| 6664 |
+
{
|
| 6665 |
+
"epoch": 0.06884281616499038,
|
| 6666 |
+
"grad_norm": 2.7516982555389404,
|
| 6667 |
+
"learning_rate": 6.247921483069351e-05,
|
| 6668 |
+
"loss": 1.1462,
|
| 6669 |
+
"step": 948
|
| 6670 |
+
},
|
| 6671 |
+
{
|
| 6672 |
+
"epoch": 0.06891543516938382,
|
| 6673 |
+
"grad_norm": 1.5767769813537598,
|
| 6674 |
+
"learning_rate": 6.228606026232118e-05,
|
| 6675 |
+
"loss": 0.3461,
|
| 6676 |
+
"step": 949
|
| 6677 |
+
},
|
| 6678 |
+
{
|
| 6679 |
+
"epoch": 0.06898805417377728,
|
| 6680 |
+
"grad_norm": 2.7577474117279053,
|
| 6681 |
+
"learning_rate": 6.209306959245161e-05,
|
| 6682 |
+
"loss": 0.3272,
|
| 6683 |
+
"step": 950
|
| 6684 |
+
},
|
| 6685 |
+
{
|
| 6686 |
+
"epoch": 0.06906067317817073,
|
| 6687 |
+
"grad_norm": 1.220229983329773,
|
| 6688 |
+
"learning_rate": 6.190024365979017e-05,
|
| 6689 |
+
"loss": 0.8371,
|
| 6690 |
+
"step": 951
|
| 6691 |
+
},
|
| 6692 |
+
{
|
| 6693 |
+
"epoch": 0.06913329218256417,
|
| 6694 |
+
"grad_norm": 1.1744657754898071,
|
| 6695 |
+
"learning_rate": 6.170758330232621e-05,
|
| 6696 |
+
"loss": 0.6767,
|
| 6697 |
+
"step": 952
|
| 6698 |
+
},
|
| 6699 |
+
{
|
| 6700 |
+
"epoch": 0.06920591118695762,
|
| 6701 |
+
"grad_norm": 1.7130528688430786,
|
| 6702 |
+
"learning_rate": 6.151508935732956e-05,
|
| 6703 |
+
"loss": 0.4092,
|
| 6704 |
+
"step": 953
|
| 6705 |
+
},
|
| 6706 |
+
{
|
| 6707 |
+
"epoch": 0.06927853019135108,
|
| 6708 |
+
"grad_norm": 1.0144054889678955,
|
| 6709 |
+
"learning_rate": 6.132276266134687e-05,
|
| 6710 |
+
"loss": 0.939,
|
| 6711 |
+
"step": 954
|
| 6712 |
+
},
|
| 6713 |
+
{
|
| 6714 |
+
"epoch": 0.06935114919574453,
|
| 6715 |
+
"grad_norm": 0.8083431124687195,
|
| 6716 |
+
"learning_rate": 6.113060405019794e-05,
|
| 6717 |
+
"loss": 0.2313,
|
| 6718 |
+
"step": 955
|
| 6719 |
+
},
|
| 6720 |
+
{
|
| 6721 |
+
"epoch": 0.06942376820013797,
|
| 6722 |
+
"grad_norm": 1.4344497919082642,
|
| 6723 |
+
"learning_rate": 6.093861435897208e-05,
|
| 6724 |
+
"loss": 0.9125,
|
| 6725 |
+
"step": 956
|
| 6726 |
+
},
|
| 6727 |
+
{
|
| 6728 |
+
"epoch": 0.06949638720453143,
|
| 6729 |
+
"grad_norm": 0.576572835445404,
|
| 6730 |
+
"learning_rate": 6.074679442202455e-05,
|
| 6731 |
+
"loss": 0.4151,
|
| 6732 |
+
"step": 957
|
| 6733 |
+
},
|
| 6734 |
+
{
|
| 6735 |
+
"epoch": 0.06956900620892488,
|
| 6736 |
+
"grad_norm": 1.4292668104171753,
|
| 6737 |
+
"learning_rate": 6.055514507297284e-05,
|
| 6738 |
+
"loss": 0.4717,
|
| 6739 |
+
"step": 958
|
| 6740 |
+
},
|
| 6741 |
+
{
|
| 6742 |
+
"epoch": 0.06964162521331832,
|
| 6743 |
+
"grad_norm": 1.5132579803466797,
|
| 6744 |
+
"learning_rate": 6.0363667144693105e-05,
|
| 6745 |
+
"loss": 1.099,
|
| 6746 |
+
"step": 959
|
| 6747 |
+
},
|
| 6748 |
+
{
|
| 6749 |
+
"epoch": 0.06971424421771177,
|
| 6750 |
+
"grad_norm": 1.458929419517517,
|
| 6751 |
+
"learning_rate": 6.017236146931658e-05,
|
| 6752 |
+
"loss": 0.2586,
|
| 6753 |
+
"step": 960
|
| 6754 |
+
},
|
| 6755 |
+
{
|
| 6756 |
+
"epoch": 0.06978686322210523,
|
| 6757 |
+
"grad_norm": 0.528302013874054,
|
| 6758 |
+
"learning_rate": 5.998122887822583e-05,
|
| 6759 |
+
"loss": 0.1097,
|
| 6760 |
+
"step": 961
|
| 6761 |
+
},
|
| 6762 |
+
{
|
| 6763 |
+
"epoch": 0.06985948222649868,
|
| 6764 |
+
"grad_norm": 0.48788735270500183,
|
| 6765 |
+
"learning_rate": 5.9790270202051314e-05,
|
| 6766 |
+
"loss": 0.2783,
|
| 6767 |
+
"step": 962
|
| 6768 |
+
},
|
| 6769 |
+
{
|
| 6770 |
+
"epoch": 0.06993210123089212,
|
| 6771 |
+
"grad_norm": 1.215254783630371,
|
| 6772 |
+
"learning_rate": 5.9599486270667595e-05,
|
| 6773 |
+
"loss": 0.9113,
|
| 6774 |
+
"step": 963
|
| 6775 |
+
},
|
| 6776 |
+
{
|
| 6777 |
+
"epoch": 0.07000472023528558,
|
| 6778 |
+
"grad_norm": 0.9500570297241211,
|
| 6779 |
+
"learning_rate": 5.940887791318993e-05,
|
| 6780 |
+
"loss": 0.2876,
|
| 6781 |
+
"step": 964
|
| 6782 |
+
},
|
| 6783 |
+
{
|
| 6784 |
+
"epoch": 0.07007733923967903,
|
| 6785 |
+
"grad_norm": 1.8602180480957031,
|
| 6786 |
+
"learning_rate": 5.9218445957970436e-05,
|
| 6787 |
+
"loss": 0.8635,
|
| 6788 |
+
"step": 965
|
| 6789 |
+
},
|
| 6790 |
+
{
|
| 6791 |
+
"epoch": 0.07014995824407247,
|
| 6792 |
+
"grad_norm": 0.45447951555252075,
|
| 6793 |
+
"learning_rate": 5.9028191232594773e-05,
|
| 6794 |
+
"loss": 0.1845,
|
| 6795 |
+
"step": 966
|
| 6796 |
+
},
|
| 6797 |
+
{
|
| 6798 |
+
"epoch": 0.07022257724846592,
|
| 6799 |
+
"grad_norm": 1.5029722452163696,
|
| 6800 |
+
"learning_rate": 5.883811456387821e-05,
|
| 6801 |
+
"loss": 1.2559,
|
| 6802 |
+
"step": 967
|
| 6803 |
+
},
|
| 6804 |
+
{
|
| 6805 |
+
"epoch": 0.07029519625285938,
|
| 6806 |
+
"grad_norm": 1.3167767524719238,
|
| 6807 |
+
"learning_rate": 5.864821677786234e-05,
|
| 6808 |
+
"loss": 0.863,
|
| 6809 |
+
"step": 968
|
| 6810 |
+
},
|
| 6811 |
+
{
|
| 6812 |
+
"epoch": 0.07036781525725282,
|
| 6813 |
+
"grad_norm": 0.9414563775062561,
|
| 6814 |
+
"learning_rate": 5.845849869981137e-05,
|
| 6815 |
+
"loss": 0.3077,
|
| 6816 |
+
"step": 969
|
| 6817 |
+
},
|
| 6818 |
+
{
|
| 6819 |
+
"epoch": 0.07044043426164627,
|
| 6820 |
+
"grad_norm": 1.9273661375045776,
|
| 6821 |
+
"learning_rate": 5.826896115420839e-05,
|
| 6822 |
+
"loss": 0.4813,
|
| 6823 |
+
"step": 970
|
| 6824 |
+
},
|
| 6825 |
+
{
|
| 6826 |
+
"epoch": 0.07051305326603972,
|
| 6827 |
+
"grad_norm": 1.7446480989456177,
|
| 6828 |
+
"learning_rate": 5.807960496475212e-05,
|
| 6829 |
+
"loss": 0.9185,
|
| 6830 |
+
"step": 971
|
| 6831 |
+
},
|
| 6832 |
+
{
|
| 6833 |
+
"epoch": 0.07058567227043318,
|
| 6834 |
+
"grad_norm": 0.8334340453147888,
|
| 6835 |
+
"learning_rate": 5.7890430954352895e-05,
|
| 6836 |
+
"loss": 0.2089,
|
| 6837 |
+
"step": 972
|
| 6838 |
+
},
|
| 6839 |
+
{
|
| 6840 |
+
"epoch": 0.07065829127482662,
|
| 6841 |
+
"grad_norm": 0.8565029501914978,
|
| 6842 |
+
"learning_rate": 5.770143994512962e-05,
|
| 6843 |
+
"loss": 1.5406,
|
| 6844 |
+
"step": 973
|
| 6845 |
+
},
|
| 6846 |
+
{
|
| 6847 |
+
"epoch": 0.07073091027922007,
|
| 6848 |
+
"grad_norm": 0.8658803105354309,
|
| 6849 |
+
"learning_rate": 5.7512632758405616e-05,
|
| 6850 |
+
"loss": 0.8221,
|
| 6851 |
+
"step": 974
|
| 6852 |
+
},
|
| 6853 |
+
{
|
| 6854 |
+
"epoch": 0.07080352928361353,
|
| 6855 |
+
"grad_norm": 1.8876001834869385,
|
| 6856 |
+
"learning_rate": 5.73240102147056e-05,
|
| 6857 |
+
"loss": 0.8373,
|
| 6858 |
+
"step": 975
|
| 6859 |
+
},
|
| 6860 |
+
{
|
| 6861 |
+
"epoch": 0.07087614828800697,
|
| 6862 |
+
"grad_norm": 0.8740954399108887,
|
| 6863 |
+
"learning_rate": 5.713557313375162e-05,
|
| 6864 |
+
"loss": 0.3552,
|
| 6865 |
+
"step": 976
|
| 6866 |
+
},
|
| 6867 |
+
{
|
| 6868 |
+
"epoch": 0.07094876729240042,
|
| 6869 |
+
"grad_norm": 0.6312585473060608,
|
| 6870 |
+
"learning_rate": 5.694732233445996e-05,
|
| 6871 |
+
"loss": 0.1054,
|
| 6872 |
+
"step": 977
|
| 6873 |
+
},
|
| 6874 |
+
{
|
| 6875 |
+
"epoch": 0.07102138629679387,
|
| 6876 |
+
"grad_norm": 0.9607315063476562,
|
| 6877 |
+
"learning_rate": 5.675925863493721e-05,
|
| 6878 |
+
"loss": 0.7761,
|
| 6879 |
+
"step": 978
|
| 6880 |
+
},
|
| 6881 |
+
{
|
| 6882 |
+
"epoch": 0.07109400530118733,
|
| 6883 |
+
"grad_norm": 0.9000673294067383,
|
| 6884 |
+
"learning_rate": 5.657138285247687e-05,
|
| 6885 |
+
"loss": 0.4288,
|
| 6886 |
+
"step": 979
|
| 6887 |
+
},
|
| 6888 |
+
{
|
| 6889 |
+
"epoch": 0.07116662430558077,
|
| 6890 |
+
"grad_norm": 1.3778090476989746,
|
| 6891 |
+
"learning_rate": 5.638369580355587e-05,
|
| 6892 |
+
"loss": 0.9409,
|
| 6893 |
+
"step": 980
|
| 6894 |
+
},
|
| 6895 |
+
{
|
| 6896 |
+
"epoch": 0.07123924330997422,
|
| 6897 |
+
"grad_norm": 0.6998331546783447,
|
| 6898 |
+
"learning_rate": 5.619619830383083e-05,
|
| 6899 |
+
"loss": 0.8496,
|
| 6900 |
+
"step": 981
|
| 6901 |
+
},
|
| 6902 |
+
{
|
| 6903 |
+
"epoch": 0.07131186231436767,
|
| 6904 |
+
"grad_norm": 2.990966558456421,
|
| 6905 |
+
"learning_rate": 5.600889116813466e-05,
|
| 6906 |
+
"loss": 1.8518,
|
| 6907 |
+
"step": 982
|
| 6908 |
+
},
|
| 6909 |
+
{
|
| 6910 |
+
"epoch": 0.07138448131876111,
|
| 6911 |
+
"grad_norm": 1.4547016620635986,
|
| 6912 |
+
"learning_rate": 5.5821775210473016e-05,
|
| 6913 |
+
"loss": 0.621,
|
| 6914 |
+
"step": 983
|
| 6915 |
+
},
|
| 6916 |
+
{
|
| 6917 |
+
"epoch": 0.07145710032315457,
|
| 6918 |
+
"grad_norm": 1.0768303871154785,
|
| 6919 |
+
"learning_rate": 5.563485124402068e-05,
|
| 6920 |
+
"loss": 1.4578,
|
| 6921 |
+
"step": 984
|
| 6922 |
+
},
|
| 6923 |
+
{
|
| 6924 |
+
"epoch": 0.07152971932754802,
|
| 6925 |
+
"grad_norm": 1.0334311723709106,
|
| 6926 |
+
"learning_rate": 5.5448120081118085e-05,
|
| 6927 |
+
"loss": 0.1995,
|
| 6928 |
+
"step": 985
|
| 6929 |
+
},
|
| 6930 |
+
{
|
| 6931 |
+
"epoch": 0.07160233833194148,
|
| 6932 |
+
"grad_norm": 0.7292360663414001,
|
| 6933 |
+
"learning_rate": 5.526158253326776e-05,
|
| 6934 |
+
"loss": 0.1821,
|
| 6935 |
+
"step": 986
|
| 6936 |
+
},
|
| 6937 |
+
{
|
| 6938 |
+
"epoch": 0.07167495733633492,
|
| 6939 |
+
"grad_norm": 0.5135112404823303,
|
| 6940 |
+
"learning_rate": 5.5075239411130864e-05,
|
| 6941 |
+
"loss": 0.2844,
|
| 6942 |
+
"step": 987
|
| 6943 |
+
},
|
| 6944 |
+
{
|
| 6945 |
+
"epoch": 0.07174757634072837,
|
| 6946 |
+
"grad_norm": 1.301938772201538,
|
| 6947 |
+
"learning_rate": 5.488909152452357e-05,
|
| 6948 |
+
"loss": 0.3329,
|
| 6949 |
+
"step": 988
|
| 6950 |
+
},
|
| 6951 |
+
{
|
| 6952 |
+
"epoch": 0.07182019534512182,
|
| 6953 |
+
"grad_norm": 0.8821846842765808,
|
| 6954 |
+
"learning_rate": 5.4703139682413586e-05,
|
| 6955 |
+
"loss": 0.5103,
|
| 6956 |
+
"step": 989
|
| 6957 |
+
},
|
| 6958 |
+
{
|
| 6959 |
+
"epoch": 0.07189281434951526,
|
| 6960 |
+
"grad_norm": 1.6066642999649048,
|
| 6961 |
+
"learning_rate": 5.451738469291666e-05,
|
| 6962 |
+
"loss": 0.6446,
|
| 6963 |
+
"step": 990
|
| 6964 |
+
},
|
| 6965 |
+
{
|
| 6966 |
+
"epoch": 0.07196543335390872,
|
| 6967 |
+
"grad_norm": 0.9525912404060364,
|
| 6968 |
+
"learning_rate": 5.4331827363293055e-05,
|
| 6969 |
+
"loss": 0.175,
|
| 6970 |
+
"step": 991
|
| 6971 |
+
},
|
| 6972 |
+
{
|
| 6973 |
+
"epoch": 0.07203805235830217,
|
| 6974 |
+
"grad_norm": 0.9844843149185181,
|
| 6975 |
+
"learning_rate": 5.4146468499944026e-05,
|
| 6976 |
+
"loss": 0.9584,
|
| 6977 |
+
"step": 992
|
| 6978 |
+
},
|
| 6979 |
+
{
|
| 6980 |
+
"epoch": 0.07211067136269561,
|
| 6981 |
+
"grad_norm": 0.49875086545944214,
|
| 6982 |
+
"learning_rate": 5.3961308908408334e-05,
|
| 6983 |
+
"loss": 0.2495,
|
| 6984 |
+
"step": 993
|
| 6985 |
+
},
|
| 6986 |
+
{
|
| 6987 |
+
"epoch": 0.07218329036708906,
|
| 6988 |
+
"grad_norm": 1.9807820320129395,
|
| 6989 |
+
"learning_rate": 5.3776349393358736e-05,
|
| 6990 |
+
"loss": 0.5241,
|
| 6991 |
+
"step": 994
|
| 6992 |
+
},
|
| 6993 |
+
{
|
| 6994 |
+
"epoch": 0.07225590937148252,
|
| 6995 |
+
"grad_norm": 2.6492059230804443,
|
| 6996 |
+
"learning_rate": 5.3591590758598475e-05,
|
| 6997 |
+
"loss": 1.044,
|
| 6998 |
+
"step": 995
|
| 6999 |
+
},
|
| 7000 |
+
{
|
| 7001 |
+
"epoch": 0.07232852837587597,
|
| 7002 |
+
"grad_norm": 1.0260138511657715,
|
| 7003 |
+
"learning_rate": 5.3407033807057846e-05,
|
| 7004 |
+
"loss": 0.8203,
|
| 7005 |
+
"step": 996
|
| 7006 |
+
},
|
| 7007 |
+
{
|
| 7008 |
+
"epoch": 0.07240114738026941,
|
| 7009 |
+
"grad_norm": 3.9065654277801514,
|
| 7010 |
+
"learning_rate": 5.322267934079056e-05,
|
| 7011 |
+
"loss": 1.112,
|
| 7012 |
+
"step": 997
|
| 7013 |
+
},
|
| 7014 |
+
{
|
| 7015 |
+
"epoch": 0.07247376638466287,
|
| 7016 |
+
"grad_norm": 0.5905523300170898,
|
| 7017 |
+
"learning_rate": 5.303852816097057e-05,
|
| 7018 |
+
"loss": 0.269,
|
| 7019 |
+
"step": 998
|
| 7020 |
+
},
|
| 7021 |
+
{
|
| 7022 |
+
"epoch": 0.07254638538905632,
|
| 7023 |
+
"grad_norm": 1.3066952228546143,
|
| 7024 |
+
"learning_rate": 5.285458106788807e-05,
|
| 7025 |
+
"loss": 0.8614,
|
| 7026 |
+
"step": 999
|
| 7027 |
+
},
|
| 7028 |
+
{
|
| 7029 |
+
"epoch": 0.07261900439344976,
|
| 7030 |
+
"grad_norm": 1.337264895439148,
|
| 7031 |
+
"learning_rate": 5.267083886094668e-05,
|
| 7032 |
+
"loss": 0.6845,
|
| 7033 |
+
"step": 1000
|
| 7034 |
+
},
|
| 7035 |
+
{
|
| 7036 |
+
"epoch": 0.07269162339784321,
|
| 7037 |
+
"grad_norm": 1.4708069562911987,
|
| 7038 |
+
"learning_rate": 5.248730233865926e-05,
|
| 7039 |
+
"loss": 0.7446,
|
| 7040 |
+
"step": 1001
|
| 7041 |
+
},
|
| 7042 |
+
{
|
| 7043 |
+
"epoch": 0.07276424240223667,
|
| 7044 |
+
"grad_norm": 1.1049293279647827,
|
| 7045 |
+
"learning_rate": 5.2303972298645145e-05,
|
| 7046 |
+
"loss": 0.7114,
|
| 7047 |
+
"step": 1002
|
| 7048 |
+
},
|
| 7049 |
+
{
|
| 7050 |
+
"epoch": 0.07283686140663012,
|
| 7051 |
+
"grad_norm": 1.8978995084762573,
|
| 7052 |
+
"learning_rate": 5.212084953762599e-05,
|
| 7053 |
+
"loss": 0.9777,
|
| 7054 |
+
"step": 1003
|
| 7055 |
+
},
|
| 7056 |
+
{
|
| 7057 |
+
"epoch": 0.07290948041102356,
|
| 7058 |
+
"grad_norm": 1.2287487983703613,
|
| 7059 |
+
"learning_rate": 5.1937934851423006e-05,
|
| 7060 |
+
"loss": 0.7173,
|
| 7061 |
+
"step": 1004
|
| 7062 |
+
},
|
| 7063 |
+
{
|
| 7064 |
+
"epoch": 0.07298209941541701,
|
| 7065 |
+
"grad_norm": 0.7756360173225403,
|
| 7066 |
+
"learning_rate": 5.17552290349528e-05,
|
| 7067 |
+
"loss": 0.2439,
|
| 7068 |
+
"step": 1005
|
| 7069 |
+
},
|
| 7070 |
+
{
|
| 7071 |
+
"epoch": 0.07305471841981047,
|
| 7072 |
+
"grad_norm": 1.0052796602249146,
|
| 7073 |
+
"learning_rate": 5.157273288222455e-05,
|
| 7074 |
+
"loss": 0.9972,
|
| 7075 |
+
"step": 1006
|
| 7076 |
+
},
|
| 7077 |
+
{
|
| 7078 |
+
"epoch": 0.07312733742420391,
|
| 7079 |
+
"grad_norm": 0.9000952243804932,
|
| 7080 |
+
"learning_rate": 5.1390447186336085e-05,
|
| 7081 |
+
"loss": 0.1626,
|
| 7082 |
+
"step": 1007
|
| 7083 |
+
},
|
| 7084 |
+
{
|
| 7085 |
+
"epoch": 0.07319995642859736,
|
| 7086 |
+
"grad_norm": 1.6128787994384766,
|
| 7087 |
+
"learning_rate": 5.120837273947072e-05,
|
| 7088 |
+
"loss": 0.6554,
|
| 7089 |
+
"step": 1008
|
| 7090 |
+
},
|
| 7091 |
+
{
|
| 7092 |
+
"epoch": 0.07327257543299082,
|
| 7093 |
+
"grad_norm": 0.9931581616401672,
|
| 7094 |
+
"learning_rate": 5.10265103328937e-05,
|
| 7095 |
+
"loss": 0.4106,
|
| 7096 |
+
"step": 1009
|
| 7097 |
+
},
|
| 7098 |
+
{
|
| 7099 |
+
"epoch": 0.07334519443738427,
|
| 7100 |
+
"grad_norm": 2.278157949447632,
|
| 7101 |
+
"learning_rate": 5.084486075694871e-05,
|
| 7102 |
+
"loss": 0.5078,
|
| 7103 |
+
"step": 1010
|
| 7104 |
+
},
|
| 7105 |
+
{
|
| 7106 |
+
"epoch": 0.07341781344177771,
|
| 7107 |
+
"grad_norm": 0.5036839246749878,
|
| 7108 |
+
"learning_rate": 5.0663424801054595e-05,
|
| 7109 |
+
"loss": 0.3505,
|
| 7110 |
+
"step": 1011
|
| 7111 |
+
},
|
| 7112 |
+
{
|
| 7113 |
+
"epoch": 0.07349043244617116,
|
| 7114 |
+
"grad_norm": 0.7192487120628357,
|
| 7115 |
+
"learning_rate": 5.048220325370181e-05,
|
| 7116 |
+
"loss": 0.6513,
|
| 7117 |
+
"step": 1012
|
| 7118 |
+
},
|
| 7119 |
+
{
|
| 7120 |
+
"epoch": 0.07356305145056462,
|
| 7121 |
+
"grad_norm": 0.8298673629760742,
|
| 7122 |
+
"learning_rate": 5.0301196902449035e-05,
|
| 7123 |
+
"loss": 0.2502,
|
| 7124 |
+
"step": 1013
|
| 7125 |
+
},
|
| 7126 |
+
{
|
| 7127 |
+
"epoch": 0.07363567045495806,
|
| 7128 |
+
"grad_norm": 1.4358340501785278,
|
| 7129 |
+
"learning_rate": 5.0120406533919715e-05,
|
| 7130 |
+
"loss": 0.4269,
|
| 7131 |
+
"step": 1014
|
| 7132 |
+
},
|
| 7133 |
+
{
|
| 7134 |
+
"epoch": 0.07370828945935151,
|
| 7135 |
+
"grad_norm": 0.6739174127578735,
|
| 7136 |
+
"learning_rate": 4.99398329337987e-05,
|
| 7137 |
+
"loss": 0.464,
|
| 7138 |
+
"step": 1015
|
| 7139 |
+
},
|
| 7140 |
+
{
|
| 7141 |
+
"epoch": 0.07378090846374497,
|
| 7142 |
+
"grad_norm": 1.07339346408844,
|
| 7143 |
+
"learning_rate": 4.97594768868288e-05,
|
| 7144 |
+
"loss": 0.5514,
|
| 7145 |
+
"step": 1016
|
| 7146 |
+
},
|
| 7147 |
+
{
|
| 7148 |
+
"epoch": 0.0738535274681384,
|
| 7149 |
+
"grad_norm": 1.038041591644287,
|
| 7150 |
+
"learning_rate": 4.957933917680735e-05,
|
| 7151 |
+
"loss": 0.7342,
|
| 7152 |
+
"step": 1017
|
| 7153 |
+
},
|
| 7154 |
+
{
|
| 7155 |
+
"epoch": 0.07392614647253186,
|
| 7156 |
+
"grad_norm": 1.2801992893218994,
|
| 7157 |
+
"learning_rate": 4.9399420586582854e-05,
|
| 7158 |
+
"loss": 1.245,
|
| 7159 |
+
"step": 1018
|
| 7160 |
+
},
|
| 7161 |
+
{
|
| 7162 |
+
"epoch": 0.07399876547692531,
|
| 7163 |
+
"grad_norm": 1.4319050312042236,
|
| 7164 |
+
"learning_rate": 4.921972189805154e-05,
|
| 7165 |
+
"loss": 0.3972,
|
| 7166 |
+
"step": 1019
|
| 7167 |
+
},
|
| 7168 |
+
{
|
| 7169 |
+
"epoch": 0.07407138448131877,
|
| 7170 |
+
"grad_norm": 0.797933042049408,
|
| 7171 |
+
"learning_rate": 4.9040243892154e-05,
|
| 7172 |
+
"loss": 0.5278,
|
| 7173 |
+
"step": 1020
|
| 7174 |
+
},
|
| 7175 |
+
{
|
| 7176 |
+
"epoch": 0.0741440034857122,
|
| 7177 |
+
"grad_norm": 0.8828997015953064,
|
| 7178 |
+
"learning_rate": 4.8860987348871766e-05,
|
| 7179 |
+
"loss": 0.295,
|
| 7180 |
+
"step": 1021
|
| 7181 |
+
},
|
| 7182 |
+
{
|
| 7183 |
+
"epoch": 0.07421662249010566,
|
| 7184 |
+
"grad_norm": 0.9507952928543091,
|
| 7185 |
+
"learning_rate": 4.8681953047223914e-05,
|
| 7186 |
+
"loss": 0.9955,
|
| 7187 |
+
"step": 1022
|
| 7188 |
+
},
|
| 7189 |
+
{
|
| 7190 |
+
"epoch": 0.07428924149449911,
|
| 7191 |
+
"grad_norm": 1.0915498733520508,
|
| 7192 |
+
"learning_rate": 4.8503141765263726e-05,
|
| 7193 |
+
"loss": 0.257,
|
| 7194 |
+
"step": 1023
|
| 7195 |
+
},
|
| 7196 |
+
{
|
| 7197 |
+
"epoch": 0.07436186049889255,
|
| 7198 |
+
"grad_norm": 0.893052339553833,
|
| 7199 |
+
"learning_rate": 4.8324554280075255e-05,
|
| 7200 |
+
"loss": 0.8009,
|
| 7201 |
+
"step": 1024
|
| 7202 |
+
},
|
| 7203 |
+
{
|
| 7204 |
+
"epoch": 0.07443447950328601,
|
| 7205 |
+
"grad_norm": 0.890261173248291,
|
| 7206 |
+
"learning_rate": 4.8146191367769975e-05,
|
| 7207 |
+
"loss": 0.2356,
|
| 7208 |
+
"step": 1025
|
| 7209 |
+
},
|
| 7210 |
+
{
|
| 7211 |
+
"epoch": 0.07450709850767946,
|
| 7212 |
+
"grad_norm": 1.923423171043396,
|
| 7213 |
+
"learning_rate": 4.796805380348336e-05,
|
| 7214 |
+
"loss": 0.6498,
|
| 7215 |
+
"step": 1026
|
| 7216 |
+
},
|
| 7217 |
+
{
|
| 7218 |
+
"epoch": 0.07457971751207292,
|
| 7219 |
+
"grad_norm": 3.203328847885132,
|
| 7220 |
+
"learning_rate": 4.7790142361371736e-05,
|
| 7221 |
+
"loss": 0.7113,
|
| 7222 |
+
"step": 1027
|
| 7223 |
+
},
|
| 7224 |
+
{
|
| 7225 |
+
"epoch": 0.07465233651646636,
|
| 7226 |
+
"grad_norm": 0.9971923232078552,
|
| 7227 |
+
"learning_rate": 4.761245781460846e-05,
|
| 7228 |
+
"loss": 0.6342,
|
| 7229 |
+
"step": 1028
|
| 7230 |
+
},
|
| 7231 |
+
{
|
| 7232 |
+
"epoch": 0.07472495552085981,
|
| 7233 |
+
"grad_norm": 1.0747997760772705,
|
| 7234 |
+
"learning_rate": 4.7435000935381115e-05,
|
| 7235 |
+
"loss": 1.0257,
|
| 7236 |
+
"step": 1029
|
| 7237 |
+
},
|
| 7238 |
+
{
|
| 7239 |
+
"epoch": 0.07479757452525326,
|
| 7240 |
+
"grad_norm": 0.48684489727020264,
|
| 7241 |
+
"learning_rate": 4.725777249488761e-05,
|
| 7242 |
+
"loss": 0.1168,
|
| 7243 |
+
"step": 1030
|
| 7244 |
+
},
|
| 7245 |
+
{
|
| 7246 |
+
"epoch": 0.0748701935296467,
|
| 7247 |
+
"grad_norm": 3.6183338165283203,
|
| 7248 |
+
"learning_rate": 4.70807732633334e-05,
|
| 7249 |
+
"loss": 1.1131,
|
| 7250 |
+
"step": 1031
|
| 7251 |
+
},
|
| 7252 |
+
{
|
| 7253 |
+
"epoch": 0.07494281253404016,
|
| 7254 |
+
"grad_norm": 0.9645475149154663,
|
| 7255 |
+
"learning_rate": 4.690400400992752e-05,
|
| 7256 |
+
"loss": 0.7626,
|
| 7257 |
+
"step": 1032
|
| 7258 |
+
},
|
| 7259 |
+
{
|
| 7260 |
+
"epoch": 0.07501543153843361,
|
| 7261 |
+
"grad_norm": 0.8078812956809998,
|
| 7262 |
+
"learning_rate": 4.672746550287985e-05,
|
| 7263 |
+
"loss": 0.9775,
|
| 7264 |
+
"step": 1033
|
| 7265 |
+
},
|
| 7266 |
+
{
|
| 7267 |
+
"epoch": 0.07508805054282706,
|
| 7268 |
+
"grad_norm": 3.4486353397369385,
|
| 7269 |
+
"learning_rate": 4.655115850939722e-05,
|
| 7270 |
+
"loss": 0.7326,
|
| 7271 |
+
"step": 1034
|
| 7272 |
+
},
|
| 7273 |
+
{
|
| 7274 |
+
"epoch": 0.0751606695472205,
|
| 7275 |
+
"grad_norm": 0.9302204847335815,
|
| 7276 |
+
"learning_rate": 4.637508379568056e-05,
|
| 7277 |
+
"loss": 0.3621,
|
| 7278 |
+
"step": 1035
|
| 7279 |
+
},
|
| 7280 |
+
{
|
| 7281 |
+
"epoch": 0.07523328855161396,
|
| 7282 |
+
"grad_norm": 2.4916927814483643,
|
| 7283 |
+
"learning_rate": 4.619924212692125e-05,
|
| 7284 |
+
"loss": 0.5182,
|
| 7285 |
+
"step": 1036
|
| 7286 |
+
},
|
| 7287 |
+
{
|
| 7288 |
+
"epoch": 0.07530590755600741,
|
| 7289 |
+
"grad_norm": 0.9201470017433167,
|
| 7290 |
+
"learning_rate": 4.602363426729788e-05,
|
| 7291 |
+
"loss": 0.5608,
|
| 7292 |
+
"step": 1037
|
| 7293 |
+
},
|
| 7294 |
+
{
|
| 7295 |
+
"epoch": 0.07537852656040085,
|
| 7296 |
+
"grad_norm": 1.4069328308105469,
|
| 7297 |
+
"learning_rate": 4.584826097997297e-05,
|
| 7298 |
+
"loss": 0.3654,
|
| 7299 |
+
"step": 1038
|
| 7300 |
+
},
|
| 7301 |
+
{
|
| 7302 |
+
"epoch": 0.0754511455647943,
|
| 7303 |
+
"grad_norm": 1.4150211811065674,
|
| 7304 |
+
"learning_rate": 4.567312302708965e-05,
|
| 7305 |
+
"loss": 0.6301,
|
| 7306 |
+
"step": 1039
|
| 7307 |
+
},
|
| 7308 |
+
{
|
| 7309 |
+
"epoch": 0.07552376456918776,
|
| 7310 |
+
"grad_norm": 2.3295063972473145,
|
| 7311 |
+
"learning_rate": 4.5498221169768274e-05,
|
| 7312 |
+
"loss": 0.8831,
|
| 7313 |
+
"step": 1040
|
| 7314 |
+
},
|
| 7315 |
+
{
|
| 7316 |
+
"epoch": 0.0755963835735812,
|
| 7317 |
+
"grad_norm": 1.32217538356781,
|
| 7318 |
+
"learning_rate": 4.532355616810321e-05,
|
| 7319 |
+
"loss": 0.4149,
|
| 7320 |
+
"step": 1041
|
| 7321 |
+
},
|
| 7322 |
+
{
|
| 7323 |
+
"epoch": 0.07566900257797465,
|
| 7324 |
+
"grad_norm": 2.148969888687134,
|
| 7325 |
+
"learning_rate": 4.514912878115948e-05,
|
| 7326 |
+
"loss": 0.2224,
|
| 7327 |
+
"step": 1042
|
| 7328 |
+
},
|
| 7329 |
+
{
|
| 7330 |
+
"epoch": 0.07574162158236811,
|
| 7331 |
+
"grad_norm": 1.2118699550628662,
|
| 7332 |
+
"learning_rate": 4.497493976696945e-05,
|
| 7333 |
+
"loss": 0.9739,
|
| 7334 |
+
"step": 1043
|
| 7335 |
+
},
|
| 7336 |
+
{
|
| 7337 |
+
"epoch": 0.07581424058676156,
|
| 7338 |
+
"grad_norm": 7.353305339813232,
|
| 7339 |
+
"learning_rate": 4.4800989882529574e-05,
|
| 7340 |
+
"loss": 1.2647,
|
| 7341 |
+
"step": 1044
|
| 7342 |
+
},
|
| 7343 |
+
{
|
| 7344 |
+
"epoch": 0.075886859591155,
|
| 7345 |
+
"grad_norm": 1.8749953508377075,
|
| 7346 |
+
"learning_rate": 4.462727988379709e-05,
|
| 7347 |
+
"loss": 0.5645,
|
| 7348 |
+
"step": 1045
|
| 7349 |
+
},
|
| 7350 |
+
{
|
| 7351 |
+
"epoch": 0.07595947859554845,
|
| 7352 |
+
"grad_norm": 0.9001838564872742,
|
| 7353 |
+
"learning_rate": 4.4453810525686745e-05,
|
| 7354 |
+
"loss": 0.3672,
|
| 7355 |
+
"step": 1046
|
| 7356 |
+
},
|
| 7357 |
+
{
|
| 7358 |
+
"epoch": 0.07603209759994191,
|
| 7359 |
+
"grad_norm": 1.2034047842025757,
|
| 7360 |
+
"learning_rate": 4.4280582562067464e-05,
|
| 7361 |
+
"loss": 0.9263,
|
| 7362 |
+
"step": 1047
|
| 7363 |
+
},
|
| 7364 |
+
{
|
| 7365 |
+
"epoch": 0.07610471660433535,
|
| 7366 |
+
"grad_norm": 1.200669288635254,
|
| 7367 |
+
"learning_rate": 4.410759674575915e-05,
|
| 7368 |
+
"loss": 0.526,
|
| 7369 |
+
"step": 1048
|
| 7370 |
+
},
|
| 7371 |
+
{
|
| 7372 |
+
"epoch": 0.0761773356087288,
|
| 7373 |
+
"grad_norm": 1.4563775062561035,
|
| 7374 |
+
"learning_rate": 4.393485382852935e-05,
|
| 7375 |
+
"loss": 1.0811,
|
| 7376 |
+
"step": 1049
|
| 7377 |
+
},
|
| 7378 |
+
{
|
| 7379 |
+
"epoch": 0.07624995461312226,
|
| 7380 |
+
"grad_norm": 1.1171209812164307,
|
| 7381 |
+
"learning_rate": 4.3762354561090034e-05,
|
| 7382 |
+
"loss": 0.637,
|
| 7383 |
+
"step": 1050
|
| 7384 |
+
},
|
| 7385 |
+
{
|
| 7386 |
+
"epoch": 0.07632257361751571,
|
| 7387 |
+
"grad_norm": 1.1942858695983887,
|
| 7388 |
+
"learning_rate": 4.359009969309429e-05,
|
| 7389 |
+
"loss": 0.4594,
|
| 7390 |
+
"step": 1051
|
| 7391 |
+
},
|
| 7392 |
+
{
|
| 7393 |
+
"epoch": 0.07639519262190915,
|
| 7394 |
+
"grad_norm": 0.8663083910942078,
|
| 7395 |
+
"learning_rate": 4.341808997313308e-05,
|
| 7396 |
+
"loss": 0.288,
|
| 7397 |
+
"step": 1052
|
| 7398 |
+
},
|
| 7399 |
+
{
|
| 7400 |
+
"epoch": 0.0764678116263026,
|
| 7401 |
+
"grad_norm": 1.356735110282898,
|
| 7402 |
+
"learning_rate": 4.324632614873198e-05,
|
| 7403 |
+
"loss": 0.1273,
|
| 7404 |
+
"step": 1053
|
| 7405 |
+
},
|
| 7406 |
+
{
|
| 7407 |
+
"epoch": 0.07654043063069606,
|
| 7408 |
+
"grad_norm": 2.5246779918670654,
|
| 7409 |
+
"learning_rate": 4.30748089663481e-05,
|
| 7410 |
+
"loss": 0.777,
|
| 7411 |
+
"step": 1054
|
| 7412 |
+
},
|
| 7413 |
+
{
|
| 7414 |
+
"epoch": 0.0766130496350895,
|
| 7415 |
+
"grad_norm": 0.7004944086074829,
|
| 7416 |
+
"learning_rate": 4.2903539171366393e-05,
|
| 7417 |
+
"loss": 0.3555,
|
| 7418 |
+
"step": 1055
|
| 7419 |
+
},
|
| 7420 |
+
{
|
| 7421 |
+
"epoch": 0.07668566863948295,
|
| 7422 |
+
"grad_norm": 0.8813367486000061,
|
| 7423 |
+
"learning_rate": 4.2732517508097014e-05,
|
| 7424 |
+
"loss": 0.3604,
|
| 7425 |
+
"step": 1056
|
| 7426 |
+
},
|
| 7427 |
+
{
|
| 7428 |
+
"epoch": 0.0767582876438764,
|
| 7429 |
+
"grad_norm": 0.7513467073440552,
|
| 7430 |
+
"learning_rate": 4.25617447197715e-05,
|
| 7431 |
+
"loss": 0.4416,
|
| 7432 |
+
"step": 1057
|
| 7433 |
+
},
|
| 7434 |
+
{
|
| 7435 |
+
"epoch": 0.07683090664826986,
|
| 7436 |
+
"grad_norm": 0.5018694400787354,
|
| 7437 |
+
"learning_rate": 4.2391221548540094e-05,
|
| 7438 |
+
"loss": 0.1337,
|
| 7439 |
+
"step": 1058
|
| 7440 |
+
},
|
| 7441 |
+
{
|
| 7442 |
+
"epoch": 0.0769035256526633,
|
| 7443 |
+
"grad_norm": 0.914337158203125,
|
| 7444 |
+
"learning_rate": 4.2220948735467967e-05,
|
| 7445 |
+
"loss": 0.4022,
|
| 7446 |
+
"step": 1059
|
| 7447 |
+
},
|
| 7448 |
+
{
|
| 7449 |
+
"epoch": 0.07697614465705675,
|
| 7450 |
+
"grad_norm": 0.682499349117279,
|
| 7451 |
+
"learning_rate": 4.205092702053257e-05,
|
| 7452 |
+
"loss": 0.2247,
|
| 7453 |
+
"step": 1060
|
| 7454 |
+
},
|
| 7455 |
+
{
|
| 7456 |
+
"epoch": 0.0770487636614502,
|
| 7457 |
+
"grad_norm": 0.7339251041412354,
|
| 7458 |
+
"learning_rate": 4.18811571426198e-05,
|
| 7459 |
+
"loss": 0.1844,
|
| 7460 |
+
"step": 1061
|
| 7461 |
+
},
|
| 7462 |
+
{
|
| 7463 |
+
"epoch": 0.07712138266584365,
|
| 7464 |
+
"grad_norm": 1.9329663515090942,
|
| 7465 |
+
"learning_rate": 4.171163983952144e-05,
|
| 7466 |
+
"loss": 0.843,
|
| 7467 |
+
"step": 1062
|
| 7468 |
+
},
|
| 7469 |
+
{
|
| 7470 |
+
"epoch": 0.0771940016702371,
|
| 7471 |
+
"grad_norm": 1.3280580043792725,
|
| 7472 |
+
"learning_rate": 4.154237584793131e-05,
|
| 7473 |
+
"loss": 0.4244,
|
| 7474 |
+
"step": 1063
|
| 7475 |
+
},
|
| 7476 |
+
{
|
| 7477 |
+
"epoch": 0.07726662067463055,
|
| 7478 |
+
"grad_norm": 0.7914282083511353,
|
| 7479 |
+
"learning_rate": 4.137336590344266e-05,
|
| 7480 |
+
"loss": 0.7056,
|
| 7481 |
+
"step": 1064
|
| 7482 |
+
},
|
| 7483 |
+
{
|
| 7484 |
+
"epoch": 0.077339239679024,
|
| 7485 |
+
"grad_norm": 0.7150092720985413,
|
| 7486 |
+
"learning_rate": 4.1204610740544544e-05,
|
| 7487 |
+
"loss": 0.4983,
|
| 7488 |
+
"step": 1065
|
| 7489 |
+
},
|
| 7490 |
+
{
|
| 7491 |
+
"epoch": 0.07741185868341745,
|
| 7492 |
+
"grad_norm": 0.6459518671035767,
|
| 7493 |
+
"learning_rate": 4.1036111092618725e-05,
|
| 7494 |
+
"loss": 0.3573,
|
| 7495 |
+
"step": 1066
|
| 7496 |
+
},
|
| 7497 |
+
{
|
| 7498 |
+
"epoch": 0.0774844776878109,
|
| 7499 |
+
"grad_norm": 1.4271303415298462,
|
| 7500 |
+
"learning_rate": 4.0867867691936734e-05,
|
| 7501 |
+
"loss": 0.8561,
|
| 7502 |
+
"step": 1067
|
| 7503 |
+
},
|
| 7504 |
+
{
|
| 7505 |
+
"epoch": 0.07755709669220436,
|
| 7506 |
+
"grad_norm": 1.0434831380844116,
|
| 7507 |
+
"learning_rate": 4.069988126965625e-05,
|
| 7508 |
+
"loss": 0.2345,
|
| 7509 |
+
"step": 1068
|
| 7510 |
+
},
|
| 7511 |
+
{
|
| 7512 |
+
"epoch": 0.0776297156965978,
|
| 7513 |
+
"grad_norm": 1.7802876234054565,
|
| 7514 |
+
"learning_rate": 4.053215255581844e-05,
|
| 7515 |
+
"loss": 1.5387,
|
| 7516 |
+
"step": 1069
|
| 7517 |
+
},
|
| 7518 |
+
{
|
| 7519 |
+
"epoch": 0.07770233470099125,
|
| 7520 |
+
"grad_norm": 1.121539831161499,
|
| 7521 |
+
"learning_rate": 4.03646822793442e-05,
|
| 7522 |
+
"loss": 0.1665,
|
| 7523 |
+
"step": 1070
|
| 7524 |
+
},
|
| 7525 |
+
{
|
| 7526 |
+
"epoch": 0.0777749537053847,
|
| 7527 |
+
"grad_norm": 1.464437484741211,
|
| 7528 |
+
"learning_rate": 4.019747116803164e-05,
|
| 7529 |
+
"loss": 0.8184,
|
| 7530 |
+
"step": 1071
|
| 7531 |
+
},
|
| 7532 |
+
{
|
| 7533 |
+
"epoch": 0.07784757270977814,
|
| 7534 |
+
"grad_norm": 1.0533359050750732,
|
| 7535 |
+
"learning_rate": 4.003051994855226e-05,
|
| 7536 |
+
"loss": 0.6048,
|
| 7537 |
+
"step": 1072
|
| 7538 |
+
},
|
| 7539 |
+
{
|
| 7540 |
+
"epoch": 0.0779201917141716,
|
| 7541 |
+
"grad_norm": 0.6131143569946289,
|
| 7542 |
+
"learning_rate": 3.9863829346448356e-05,
|
| 7543 |
+
"loss": 0.2938,
|
| 7544 |
+
"step": 1073
|
| 7545 |
+
},
|
| 7546 |
+
{
|
| 7547 |
+
"epoch": 0.07799281071856505,
|
| 7548 |
+
"grad_norm": 0.9979797601699829,
|
| 7549 |
+
"learning_rate": 3.969740008612951e-05,
|
| 7550 |
+
"loss": 0.3266,
|
| 7551 |
+
"step": 1074
|
| 7552 |
+
},
|
| 7553 |
+
{
|
| 7554 |
+
"epoch": 0.0780654297229585,
|
| 7555 |
+
"grad_norm": 1.3746004104614258,
|
| 7556 |
+
"learning_rate": 3.953123289086957e-05,
|
| 7557 |
+
"loss": 1.1542,
|
| 7558 |
+
"step": 1075
|
| 7559 |
+
},
|
| 7560 |
+
{
|
| 7561 |
+
"epoch": 0.07813804872735194,
|
| 7562 |
+
"grad_norm": 0.7711180448532104,
|
| 7563 |
+
"learning_rate": 3.936532848280351e-05,
|
| 7564 |
+
"loss": 0.2947,
|
| 7565 |
+
"step": 1076
|
| 7566 |
+
},
|
| 7567 |
+
{
|
| 7568 |
+
"epoch": 0.0782106677317454,
|
| 7569 |
+
"grad_norm": 1.2094429731369019,
|
| 7570 |
+
"learning_rate": 3.919968758292425e-05,
|
| 7571 |
+
"loss": 0.2627,
|
| 7572 |
+
"step": 1077
|
| 7573 |
+
},
|
| 7574 |
+
{
|
| 7575 |
+
"epoch": 0.07828328673613885,
|
| 7576 |
+
"grad_norm": 1.353338360786438,
|
| 7577 |
+
"learning_rate": 3.903431091107956e-05,
|
| 7578 |
+
"loss": 1.3794,
|
| 7579 |
+
"step": 1078
|
| 7580 |
+
},
|
| 7581 |
+
{
|
| 7582 |
+
"epoch": 0.07835590574053229,
|
| 7583 |
+
"grad_norm": 1.0499879121780396,
|
| 7584 |
+
"learning_rate": 3.886919918596894e-05,
|
| 7585 |
+
"loss": 0.3597,
|
| 7586 |
+
"step": 1079
|
| 7587 |
+
},
|
| 7588 |
+
{
|
| 7589 |
+
"epoch": 0.07842852474492575,
|
| 7590 |
+
"grad_norm": 2.888306140899658,
|
| 7591 |
+
"learning_rate": 3.870435312514044e-05,
|
| 7592 |
+
"loss": 1.266,
|
| 7593 |
+
"step": 1080
|
| 7594 |
+
},
|
| 7595 |
+
{
|
| 7596 |
+
"epoch": 0.0785011437493192,
|
| 7597 |
+
"grad_norm": 0.808671772480011,
|
| 7598 |
+
"learning_rate": 3.8539773444987606e-05,
|
| 7599 |
+
"loss": 0.2154,
|
| 7600 |
+
"step": 1081
|
| 7601 |
+
},
|
| 7602 |
+
{
|
| 7603 |
+
"epoch": 0.07857376275371265,
|
| 7604 |
+
"grad_norm": 0.9751576781272888,
|
| 7605 |
+
"learning_rate": 3.8375460860746305e-05,
|
| 7606 |
+
"loss": 0.3425,
|
| 7607 |
+
"step": 1082
|
| 7608 |
+
},
|
| 7609 |
+
{
|
| 7610 |
+
"epoch": 0.0786463817581061,
|
| 7611 |
+
"grad_norm": 0.6964498162269592,
|
| 7612 |
+
"learning_rate": 3.821141608649169e-05,
|
| 7613 |
+
"loss": 0.5062,
|
| 7614 |
+
"step": 1083
|
| 7615 |
+
},
|
| 7616 |
+
{
|
| 7617 |
+
"epoch": 0.07871900076249955,
|
| 7618 |
+
"grad_norm": 1.0127582550048828,
|
| 7619 |
+
"learning_rate": 3.804763983513503e-05,
|
| 7620 |
+
"loss": 0.48,
|
| 7621 |
+
"step": 1084
|
| 7622 |
+
},
|
| 7623 |
+
{
|
| 7624 |
+
"epoch": 0.078791619766893,
|
| 7625 |
+
"grad_norm": 0.6307262182235718,
|
| 7626 |
+
"learning_rate": 3.788413281842065e-05,
|
| 7627 |
+
"loss": 0.6219,
|
| 7628 |
+
"step": 1085
|
| 7629 |
+
},
|
| 7630 |
+
{
|
| 7631 |
+
"epoch": 0.07886423877128644,
|
| 7632 |
+
"grad_norm": 1.0961401462554932,
|
| 7633 |
+
"learning_rate": 3.7720895746922814e-05,
|
| 7634 |
+
"loss": 0.6244,
|
| 7635 |
+
"step": 1086
|
| 7636 |
+
},
|
| 7637 |
+
{
|
| 7638 |
+
"epoch": 0.0789368577756799,
|
| 7639 |
+
"grad_norm": 1.224797248840332,
|
| 7640 |
+
"learning_rate": 3.755792933004265e-05,
|
| 7641 |
+
"loss": 0.7283,
|
| 7642 |
+
"step": 1087
|
| 7643 |
+
},
|
| 7644 |
+
{
|
| 7645 |
+
"epoch": 0.07900947678007335,
|
| 7646 |
+
"grad_norm": 0.8783885836601257,
|
| 7647 |
+
"learning_rate": 3.7395234276005087e-05,
|
| 7648 |
+
"loss": 0.367,
|
| 7649 |
+
"step": 1088
|
| 7650 |
+
},
|
| 7651 |
+
{
|
| 7652 |
+
"epoch": 0.07908209578446679,
|
| 7653 |
+
"grad_norm": 2.5618433952331543,
|
| 7654 |
+
"learning_rate": 3.723281129185574e-05,
|
| 7655 |
+
"loss": 0.5892,
|
| 7656 |
+
"step": 1089
|
| 7657 |
+
},
|
| 7658 |
+
{
|
| 7659 |
+
"epoch": 0.07915471478886024,
|
| 7660 |
+
"grad_norm": 1.455002784729004,
|
| 7661 |
+
"learning_rate": 3.7070661083457835e-05,
|
| 7662 |
+
"loss": 0.3932,
|
| 7663 |
+
"step": 1090
|
| 7664 |
+
},
|
| 7665 |
+
{
|
| 7666 |
+
"epoch": 0.0792273337932537,
|
| 7667 |
+
"grad_norm": 0.6232052445411682,
|
| 7668 |
+
"learning_rate": 3.690878435548917e-05,
|
| 7669 |
+
"loss": 0.7407,
|
| 7670 |
+
"step": 1091
|
| 7671 |
+
},
|
| 7672 |
+
{
|
| 7673 |
+
"epoch": 0.07929995279764715,
|
| 7674 |
+
"grad_norm": 0.6516185998916626,
|
| 7675 |
+
"learning_rate": 3.6747181811439146e-05,
|
| 7676 |
+
"loss": 0.2049,
|
| 7677 |
+
"step": 1092
|
| 7678 |
+
},
|
| 7679 |
+
{
|
| 7680 |
+
"epoch": 0.07937257180204059,
|
| 7681 |
+
"grad_norm": 0.46252021193504333,
|
| 7682 |
+
"learning_rate": 3.658585415360537e-05,
|
| 7683 |
+
"loss": 0.0476,
|
| 7684 |
+
"step": 1093
|
| 7685 |
+
},
|
| 7686 |
+
{
|
| 7687 |
+
"epoch": 0.07944519080643404,
|
| 7688 |
+
"grad_norm": 1.0539500713348389,
|
| 7689 |
+
"learning_rate": 3.642480208309112e-05,
|
| 7690 |
+
"loss": 0.7159,
|
| 7691 |
+
"step": 1094
|
| 7692 |
+
},
|
| 7693 |
+
{
|
| 7694 |
+
"epoch": 0.0795178098108275,
|
| 7695 |
+
"grad_norm": 0.7707275748252869,
|
| 7696 |
+
"learning_rate": 3.626402629980176e-05,
|
| 7697 |
+
"loss": 0.5077,
|
| 7698 |
+
"step": 1095
|
| 7699 |
+
},
|
| 7700 |
+
{
|
| 7701 |
+
"epoch": 0.07959042881522094,
|
| 7702 |
+
"grad_norm": 1.3060404062271118,
|
| 7703 |
+
"learning_rate": 3.610352750244218e-05,
|
| 7704 |
+
"loss": 0.6736,
|
| 7705 |
+
"step": 1096
|
| 7706 |
+
},
|
| 7707 |
+
{
|
| 7708 |
+
"epoch": 0.07966304781961439,
|
| 7709 |
+
"grad_norm": 1.0668853521347046,
|
| 7710 |
+
"learning_rate": 3.5943306388513333e-05,
|
| 7711 |
+
"loss": 0.2969,
|
| 7712 |
+
"step": 1097
|
| 7713 |
+
},
|
| 7714 |
+
{
|
| 7715 |
+
"epoch": 0.07973566682400784,
|
| 7716 |
+
"grad_norm": 1.3536769151687622,
|
| 7717 |
+
"learning_rate": 3.57833636543096e-05,
|
| 7718 |
+
"loss": 0.2062,
|
| 7719 |
+
"step": 1098
|
| 7720 |
+
},
|
| 7721 |
+
{
|
| 7722 |
+
"epoch": 0.0798082858284013,
|
| 7723 |
+
"grad_norm": 1.3943530321121216,
|
| 7724 |
+
"learning_rate": 3.562369999491536e-05,
|
| 7725 |
+
"loss": 0.6374,
|
| 7726 |
+
"step": 1099
|
| 7727 |
+
},
|
| 7728 |
+
{
|
| 7729 |
+
"epoch": 0.07988090483279474,
|
| 7730 |
+
"grad_norm": 0.9321604371070862,
|
| 7731 |
+
"learning_rate": 3.54643161042024e-05,
|
| 7732 |
+
"loss": 0.2545,
|
| 7733 |
+
"step": 1100
|
| 7734 |
+
},
|
| 7735 |
+
{
|
| 7736 |
+
"epoch": 0.07995352383718819,
|
| 7737 |
+
"grad_norm": 2.3198392391204834,
|
| 7738 |
+
"learning_rate": 3.530521267482653e-05,
|
| 7739 |
+
"loss": 1.1751,
|
| 7740 |
+
"step": 1101
|
| 7741 |
+
},
|
| 7742 |
+
{
|
| 7743 |
+
"epoch": 0.08002614284158165,
|
| 7744 |
+
"grad_norm": 1.0625810623168945,
|
| 7745 |
+
"learning_rate": 3.5146390398224746e-05,
|
| 7746 |
+
"loss": 0.3397,
|
| 7747 |
+
"step": 1102
|
| 7748 |
+
},
|
| 7749 |
+
{
|
| 7750 |
+
"epoch": 0.08009876184597509,
|
| 7751 |
+
"grad_norm": 1.262967824935913,
|
| 7752 |
+
"learning_rate": 3.498784996461225e-05,
|
| 7753 |
+
"loss": 0.387,
|
| 7754 |
+
"step": 1103
|
| 7755 |
+
},
|
| 7756 |
+
{
|
| 7757 |
+
"epoch": 0.08017138085036854,
|
| 7758 |
+
"grad_norm": 0.7193602323532104,
|
| 7759 |
+
"learning_rate": 3.4829592062979353e-05,
|
| 7760 |
+
"loss": 0.3205,
|
| 7761 |
+
"step": 1104
|
| 7762 |
+
},
|
| 7763 |
+
{
|
| 7764 |
+
"epoch": 0.080243999854762,
|
| 7765 |
+
"grad_norm": 1.541988730430603,
|
| 7766 |
+
"learning_rate": 3.467161738108855e-05,
|
| 7767 |
+
"loss": 0.8581,
|
| 7768 |
+
"step": 1105
|
| 7769 |
+
},
|
| 7770 |
+
{
|
| 7771 |
+
"epoch": 0.08031661885915545,
|
| 7772 |
+
"grad_norm": 2.6073052883148193,
|
| 7773 |
+
"learning_rate": 3.45139266054715e-05,
|
| 7774 |
+
"loss": 0.3846,
|
| 7775 |
+
"step": 1106
|
| 7776 |
+
},
|
| 7777 |
+
{
|
| 7778 |
+
"epoch": 0.08038923786354889,
|
| 7779 |
+
"grad_norm": 0.7341157793998718,
|
| 7780 |
+
"learning_rate": 3.435652042142606e-05,
|
| 7781 |
+
"loss": 0.4452,
|
| 7782 |
+
"step": 1107
|
| 7783 |
+
},
|
| 7784 |
+
{
|
| 7785 |
+
"epoch": 0.08046185686794234,
|
| 7786 |
+
"grad_norm": 0.8992398381233215,
|
| 7787 |
+
"learning_rate": 3.41993995130133e-05,
|
| 7788 |
+
"loss": 0.1214,
|
| 7789 |
+
"step": 1108
|
| 7790 |
+
},
|
| 7791 |
+
{
|
| 7792 |
+
"epoch": 0.0805344758723358,
|
| 7793 |
+
"grad_norm": 1.3411012887954712,
|
| 7794 |
+
"learning_rate": 3.4042564563054526e-05,
|
| 7795 |
+
"loss": 0.4329,
|
| 7796 |
+
"step": 1109
|
| 7797 |
+
},
|
| 7798 |
+
{
|
| 7799 |
+
"epoch": 0.08060709487672924,
|
| 7800 |
+
"grad_norm": 1.4646313190460205,
|
| 7801 |
+
"learning_rate": 3.3886016253128326e-05,
|
| 7802 |
+
"loss": 0.4565,
|
| 7803 |
+
"step": 1110
|
| 7804 |
+
},
|
| 7805 |
+
{
|
| 7806 |
+
"epoch": 0.08067971388112269,
|
| 7807 |
+
"grad_norm": 2.635862112045288,
|
| 7808 |
+
"learning_rate": 3.372975526356758e-05,
|
| 7809 |
+
"loss": 0.5001,
|
| 7810 |
+
"step": 1111
|
| 7811 |
+
},
|
| 7812 |
+
{
|
| 7813 |
+
"epoch": 0.08075233288551614,
|
| 7814 |
+
"grad_norm": 0.49482840299606323,
|
| 7815 |
+
"learning_rate": 3.357378227345652e-05,
|
| 7816 |
+
"loss": 0.4972,
|
| 7817 |
+
"step": 1112
|
| 7818 |
+
},
|
| 7819 |
+
{
|
| 7820 |
+
"epoch": 0.08082495188990958,
|
| 7821 |
+
"grad_norm": 1.237926959991455,
|
| 7822 |
+
"learning_rate": 3.341809796062783e-05,
|
| 7823 |
+
"loss": 0.6342,
|
| 7824 |
+
"step": 1113
|
| 7825 |
+
},
|
| 7826 |
+
{
|
| 7827 |
+
"epoch": 0.08089757089430304,
|
| 7828 |
+
"grad_norm": 0.6633093953132629,
|
| 7829 |
+
"learning_rate": 3.3262703001659565e-05,
|
| 7830 |
+
"loss": 0.2835,
|
| 7831 |
+
"step": 1114
|
| 7832 |
+
},
|
| 7833 |
+
{
|
| 7834 |
+
"epoch": 0.08097018989869649,
|
| 7835 |
+
"grad_norm": 0.7244913578033447,
|
| 7836 |
+
"learning_rate": 3.310759807187238e-05,
|
| 7837 |
+
"loss": 0.1479,
|
| 7838 |
+
"step": 1115
|
| 7839 |
+
},
|
| 7840 |
+
{
|
| 7841 |
+
"epoch": 0.08104280890308994,
|
| 7842 |
+
"grad_norm": 3.9820868968963623,
|
| 7843 |
+
"learning_rate": 3.295278384532646e-05,
|
| 7844 |
+
"loss": 0.9035,
|
| 7845 |
+
"step": 1116
|
| 7846 |
+
},
|
| 7847 |
+
{
|
| 7848 |
+
"epoch": 0.08111542790748338,
|
| 7849 |
+
"grad_norm": 1.3241500854492188,
|
| 7850 |
+
"learning_rate": 3.2798260994818666e-05,
|
| 7851 |
+
"loss": 0.7306,
|
| 7852 |
+
"step": 1117
|
| 7853 |
+
},
|
| 7854 |
+
{
|
| 7855 |
+
"epoch": 0.08118804691187684,
|
| 7856 |
+
"grad_norm": 1.0171231031417847,
|
| 7857 |
+
"learning_rate": 3.264403019187956e-05,
|
| 7858 |
+
"loss": 0.9498,
|
| 7859 |
+
"step": 1118
|
| 7860 |
+
},
|
| 7861 |
+
{
|
| 7862 |
+
"epoch": 0.08126066591627029,
|
| 7863 |
+
"grad_norm": 0.8324412107467651,
|
| 7864 |
+
"learning_rate": 3.249009210677054e-05,
|
| 7865 |
+
"loss": 0.2284,
|
| 7866 |
+
"step": 1119
|
| 7867 |
+
},
|
| 7868 |
+
{
|
| 7869 |
+
"epoch": 0.08133328492066373,
|
| 7870 |
+
"grad_norm": 1.500178575515747,
|
| 7871 |
+
"learning_rate": 3.233644740848087e-05,
|
| 7872 |
+
"loss": 1.1016,
|
| 7873 |
+
"step": 1120
|
| 7874 |
+
},
|
| 7875 |
+
{
|
| 7876 |
+
"epoch": 0.08140590392505719,
|
| 7877 |
+
"grad_norm": 1.238517165184021,
|
| 7878 |
+
"learning_rate": 3.2183096764724915e-05,
|
| 7879 |
+
"loss": 0.4627,
|
| 7880 |
+
"step": 1121
|
| 7881 |
+
},
|
| 7882 |
+
{
|
| 7883 |
+
"epoch": 0.08147852292945064,
|
| 7884 |
+
"grad_norm": 2.279787302017212,
|
| 7885 |
+
"learning_rate": 3.2030040841938924e-05,
|
| 7886 |
+
"loss": 0.7241,
|
| 7887 |
+
"step": 1122
|
| 7888 |
+
},
|
| 7889 |
+
{
|
| 7890 |
+
"epoch": 0.08155114193384409,
|
| 7891 |
+
"grad_norm": 0.5174742341041565,
|
| 7892 |
+
"learning_rate": 3.187728030527857e-05,
|
| 7893 |
+
"loss": 0.1982,
|
| 7894 |
+
"step": 1123
|
| 7895 |
+
},
|
| 7896 |
+
{
|
| 7897 |
+
"epoch": 0.08162376093823753,
|
| 7898 |
+
"grad_norm": 1.431541085243225,
|
| 7899 |
+
"learning_rate": 3.172481581861558e-05,
|
| 7900 |
+
"loss": 0.5524,
|
| 7901 |
+
"step": 1124
|
| 7902 |
+
},
|
| 7903 |
+
{
|
| 7904 |
+
"epoch": 0.08169637994263099,
|
| 7905 |
+
"grad_norm": 2.0064144134521484,
|
| 7906 |
+
"learning_rate": 3.157264804453537e-05,
|
| 7907 |
+
"loss": 1.0134,
|
| 7908 |
+
"step": 1125
|
| 7909 |
+
},
|
| 7910 |
+
{
|
| 7911 |
+
"epoch": 0.08176899894702444,
|
| 7912 |
+
"grad_norm": 0.7923216223716736,
|
| 7913 |
+
"learning_rate": 3.142077764433361e-05,
|
| 7914 |
+
"loss": 0.3506,
|
| 7915 |
+
"step": 1126
|
| 7916 |
+
},
|
| 7917 |
+
{
|
| 7918 |
+
"epoch": 0.08184161795141788,
|
| 7919 |
+
"grad_norm": 0.8654752373695374,
|
| 7920 |
+
"learning_rate": 3.12692052780139e-05,
|
| 7921 |
+
"loss": 0.3383,
|
| 7922 |
+
"step": 1127
|
| 7923 |
+
},
|
| 7924 |
+
{
|
| 7925 |
+
"epoch": 0.08191423695581133,
|
| 7926 |
+
"grad_norm": 3.1774401664733887,
|
| 7927 |
+
"learning_rate": 3.111793160428439e-05,
|
| 7928 |
+
"loss": 1.1845,
|
| 7929 |
+
"step": 1128
|
| 7930 |
+
},
|
| 7931 |
+
{
|
| 7932 |
+
"epoch": 0.08198685596020479,
|
| 7933 |
+
"grad_norm": 1.475877285003662,
|
| 7934 |
+
"learning_rate": 3.096695728055536e-05,
|
| 7935 |
+
"loss": 1.1822,
|
| 7936 |
+
"step": 1129
|
| 7937 |
+
},
|
| 7938 |
+
{
|
| 7939 |
+
"epoch": 0.08205947496459824,
|
| 7940 |
+
"grad_norm": 1.2860140800476074,
|
| 7941 |
+
"learning_rate": 3.081628296293608e-05,
|
| 7942 |
+
"loss": 0.4395,
|
| 7943 |
+
"step": 1130
|
| 7944 |
+
},
|
| 7945 |
+
{
|
| 7946 |
+
"epoch": 0.08213209396899168,
|
| 7947 |
+
"grad_norm": 1.0822505950927734,
|
| 7948 |
+
"learning_rate": 3.066590930623202e-05,
|
| 7949 |
+
"loss": 0.5476,
|
| 7950 |
+
"step": 1131
|
| 7951 |
+
},
|
| 7952 |
+
{
|
| 7953 |
+
"epoch": 0.08220471297338514,
|
| 7954 |
+
"grad_norm": 0.7372075319290161,
|
| 7955 |
+
"learning_rate": 3.0515836963942056e-05,
|
| 7956 |
+
"loss": 0.3629,
|
| 7957 |
+
"step": 1132
|
| 7958 |
+
},
|
| 7959 |
+
{
|
| 7960 |
+
"epoch": 0.08227733197777859,
|
| 7961 |
+
"grad_norm": 2.525468587875366,
|
| 7962 |
+
"learning_rate": 3.0366066588255614e-05,
|
| 7963 |
+
"loss": 2.0739,
|
| 7964 |
+
"step": 1133
|
| 7965 |
+
},
|
| 7966 |
+
{
|
| 7967 |
+
"epoch": 0.08234995098217203,
|
| 7968 |
+
"grad_norm": 1.0062713623046875,
|
| 7969 |
+
"learning_rate": 3.0216598830049804e-05,
|
| 7970 |
+
"loss": 0.3062,
|
| 7971 |
+
"step": 1134
|
| 7972 |
+
},
|
| 7973 |
+
{
|
| 7974 |
+
"epoch": 0.08242256998656548,
|
| 7975 |
+
"grad_norm": 0.4824628233909607,
|
| 7976 |
+
"learning_rate": 3.0067434338886623e-05,
|
| 7977 |
+
"loss": 0.3863,
|
| 7978 |
+
"step": 1135
|
| 7979 |
+
},
|
| 7980 |
+
{
|
| 7981 |
+
"epoch": 0.08249518899095894,
|
| 7982 |
+
"grad_norm": 0.7347169518470764,
|
| 7983 |
+
"learning_rate": 2.9918573763010106e-05,
|
| 7984 |
+
"loss": 0.6829,
|
| 7985 |
+
"step": 1136
|
| 7986 |
+
},
|
| 7987 |
+
{
|
| 7988 |
+
"epoch": 0.08256780799535238,
|
| 7989 |
+
"grad_norm": 0.7603349089622498,
|
| 7990 |
+
"learning_rate": 2.9770017749343527e-05,
|
| 7991 |
+
"loss": 0.6584,
|
| 7992 |
+
"step": 1137
|
| 7993 |
+
},
|
| 7994 |
+
{
|
| 7995 |
+
"epoch": 0.08264042699974583,
|
| 7996 |
+
"grad_norm": 1.750891923904419,
|
| 7997 |
+
"learning_rate": 2.9621766943486597e-05,
|
| 7998 |
+
"loss": 0.9981,
|
| 7999 |
+
"step": 1138
|
| 8000 |
+
},
|
| 8001 |
+
{
|
| 8002 |
+
"epoch": 0.08271304600413928,
|
| 8003 |
+
"grad_norm": 0.9961756467819214,
|
| 8004 |
+
"learning_rate": 2.9473821989712625e-05,
|
| 8005 |
+
"loss": 0.5043,
|
| 8006 |
+
"step": 1139
|
| 8007 |
+
},
|
| 8008 |
+
{
|
| 8009 |
+
"epoch": 0.08278566500853274,
|
| 8010 |
+
"grad_norm": 0.6989356875419617,
|
| 8011 |
+
"learning_rate": 2.9326183530965755e-05,
|
| 8012 |
+
"loss": 0.4037,
|
| 8013 |
+
"step": 1140
|
| 8014 |
+
},
|
| 8015 |
+
{
|
| 8016 |
+
"epoch": 0.08278566500853274,
|
| 8017 |
+
"eval_loss": 0.5331657528877258,
|
| 8018 |
+
"eval_runtime": 179.6619,
|
| 8019 |
+
"eval_samples_per_second": 32.277,
|
| 8020 |
+
"eval_steps_per_second": 16.141,
|
| 8021 |
+
"step": 1140
|
| 8022 |
}
|
| 8023 |
],
|
| 8024 |
"logging_steps": 1,
|
|
|
|
| 8038 |
"attributes": {}
|
| 8039 |
}
|
| 8040 |
},
|
| 8041 |
+
"total_flos": 2.1630469922095104e+17,
|
| 8042 |
"train_batch_size": 2,
|
| 8043 |
"trial_name": null,
|
| 8044 |
"trial_params": null
|