Training in progress, step 1569, 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 5752
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:875978a9e15118a4a938537bce4d28a90550225e3c8c74ba119bd96b6f8c297d
|
3 |
size 5752
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 15814
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7421a53b982409b252d33acf57ef55db760c0c0cb94f3f7fc72365b294d98561
|
3 |
size 15814
|
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:b70ac0261b218442aed944e476fc98123a75eb5e472055f845bdf1266ed36d44
|
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:d47c943291779f5dce4e7a180d3b70d5d1ea5763673dabfc3f9b6bddb69d2b3c
|
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": 393,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -8292,6 +8292,2736 @@
|
|
8292 |
"eval_samples_per_second": 165.016,
|
8293 |
"eval_steps_per_second": 82.508,
|
8294 |
"step": 1179
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
8295 |
}
|
8296 |
],
|
8297 |
"logging_steps": 1,
|
@@ -8306,12 +11036,12 @@
|
|
8306 |
"should_evaluate": false,
|
8307 |
"should_log": false,
|
8308 |
"should_save": true,
|
8309 |
-
"should_training_stop":
|
8310 |
},
|
8311 |
"attributes": {}
|
8312 |
}
|
8313 |
},
|
8314 |
-
"total_flos":
|
8315 |
"train_batch_size": 2,
|
8316 |
"trial_name": null,
|
8317 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.031583079199251185,
|
5 |
"eval_steps": 393,
|
6 |
+
"global_step": 1569,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
8292 |
"eval_samples_per_second": 165.016,
|
8293 |
"eval_steps_per_second": 82.508,
|
8294 |
"step": 1179
|
8295 |
+
},
|
8296 |
+
{
|
8297 |
+
"epoch": 0.02375273005424882,
|
8298 |
+
"grad_norm": 0.00120261637493968,
|
8299 |
+
"learning_rate": 2.9182534103793825e-05,
|
8300 |
+
"loss": 46.0,
|
8301 |
+
"step": 1180
|
8302 |
+
},
|
8303 |
+
{
|
8304 |
+
"epoch": 0.023772859486498182,
|
8305 |
+
"grad_norm": 0.0013184483395889401,
|
8306 |
+
"learning_rate": 2.9040402068702977e-05,
|
8307 |
+
"loss": 46.0,
|
8308 |
+
"step": 1181
|
8309 |
+
},
|
8310 |
+
{
|
8311 |
+
"epoch": 0.023792988918747548,
|
8312 |
+
"grad_norm": 0.0016656328225508332,
|
8313 |
+
"learning_rate": 2.8898558183539283e-05,
|
8314 |
+
"loss": 46.0,
|
8315 |
+
"step": 1182
|
8316 |
+
},
|
8317 |
+
{
|
8318 |
+
"epoch": 0.02381311835099691,
|
8319 |
+
"grad_norm": 0.0012708749854937196,
|
8320 |
+
"learning_rate": 2.8757003024296768e-05,
|
8321 |
+
"loss": 46.0,
|
8322 |
+
"step": 1183
|
8323 |
+
},
|
8324 |
+
{
|
8325 |
+
"epoch": 0.023833247783246275,
|
8326 |
+
"grad_norm": 0.0014254259876906872,
|
8327 |
+
"learning_rate": 2.8615737165796974e-05,
|
8328 |
+
"loss": 46.0,
|
8329 |
+
"step": 1184
|
8330 |
+
},
|
8331 |
+
{
|
8332 |
+
"epoch": 0.023853377215495637,
|
8333 |
+
"grad_norm": 0.0012592601124197245,
|
8334 |
+
"learning_rate": 2.8474761181686772e-05,
|
8335 |
+
"loss": 46.0,
|
8336 |
+
"step": 1185
|
8337 |
+
},
|
8338 |
+
{
|
8339 |
+
"epoch": 0.023873506647745,
|
8340 |
+
"grad_norm": 0.001736863050609827,
|
8341 |
+
"learning_rate": 2.8334075644435807e-05,
|
8342 |
+
"loss": 46.0,
|
8343 |
+
"step": 1186
|
8344 |
+
},
|
8345 |
+
{
|
8346 |
+
"epoch": 0.023893636079994365,
|
8347 |
+
"grad_norm": 0.0012428145855665207,
|
8348 |
+
"learning_rate": 2.8193681125334393e-05,
|
8349 |
+
"loss": 46.0,
|
8350 |
+
"step": 1187
|
8351 |
+
},
|
8352 |
+
{
|
8353 |
+
"epoch": 0.023913765512243727,
|
8354 |
+
"grad_norm": 0.001219391357153654,
|
8355 |
+
"learning_rate": 2.805357819449098e-05,
|
8356 |
+
"loss": 46.0,
|
8357 |
+
"step": 1188
|
8358 |
+
},
|
8359 |
+
{
|
8360 |
+
"epoch": 0.023933894944493092,
|
8361 |
+
"grad_norm": 0.0018028883496299386,
|
8362 |
+
"learning_rate": 2.7913767420830105e-05,
|
8363 |
+
"loss": 46.0,
|
8364 |
+
"step": 1189
|
8365 |
+
},
|
8366 |
+
{
|
8367 |
+
"epoch": 0.023954024376742454,
|
8368 |
+
"grad_norm": 0.001996217295527458,
|
8369 |
+
"learning_rate": 2.7774249372089688e-05,
|
8370 |
+
"loss": 46.0,
|
8371 |
+
"step": 1190
|
8372 |
+
},
|
8373 |
+
{
|
8374 |
+
"epoch": 0.023974153808991816,
|
8375 |
+
"grad_norm": 0.0012688592541962862,
|
8376 |
+
"learning_rate": 2.7635024614819205e-05,
|
8377 |
+
"loss": 46.0,
|
8378 |
+
"step": 1191
|
8379 |
+
},
|
8380 |
+
{
|
8381 |
+
"epoch": 0.02399428324124118,
|
8382 |
+
"grad_norm": 0.0006761057302355766,
|
8383 |
+
"learning_rate": 2.749609371437697e-05,
|
8384 |
+
"loss": 46.0,
|
8385 |
+
"step": 1192
|
8386 |
+
},
|
8387 |
+
{
|
8388 |
+
"epoch": 0.024014412673490543,
|
8389 |
+
"grad_norm": 0.0008217204012908041,
|
8390 |
+
"learning_rate": 2.735745723492815e-05,
|
8391 |
+
"loss": 46.0,
|
8392 |
+
"step": 1193
|
8393 |
+
},
|
8394 |
+
{
|
8395 |
+
"epoch": 0.02403454210573991,
|
8396 |
+
"grad_norm": 0.0008044593851082027,
|
8397 |
+
"learning_rate": 2.7219115739442215e-05,
|
8398 |
+
"loss": 46.0,
|
8399 |
+
"step": 1194
|
8400 |
+
},
|
8401 |
+
{
|
8402 |
+
"epoch": 0.02405467153798927,
|
8403 |
+
"grad_norm": 0.0011257351143285632,
|
8404 |
+
"learning_rate": 2.7081069789690883e-05,
|
8405 |
+
"loss": 46.0,
|
8406 |
+
"step": 1195
|
8407 |
+
},
|
8408 |
+
{
|
8409 |
+
"epoch": 0.024074800970238633,
|
8410 |
+
"grad_norm": 0.001486622029915452,
|
8411 |
+
"learning_rate": 2.694331994624567e-05,
|
8412 |
+
"loss": 46.0,
|
8413 |
+
"step": 1196
|
8414 |
+
},
|
8415 |
+
{
|
8416 |
+
"epoch": 0.024094930402488,
|
8417 |
+
"grad_norm": 0.0016462607309222221,
|
8418 |
+
"learning_rate": 2.6805866768475663e-05,
|
8419 |
+
"loss": 46.0,
|
8420 |
+
"step": 1197
|
8421 |
+
},
|
8422 |
+
{
|
8423 |
+
"epoch": 0.02411505983473736,
|
8424 |
+
"grad_norm": 0.0006136257434263825,
|
8425 |
+
"learning_rate": 2.666871081454533e-05,
|
8426 |
+
"loss": 46.0,
|
8427 |
+
"step": 1198
|
8428 |
+
},
|
8429 |
+
{
|
8430 |
+
"epoch": 0.024135189266986726,
|
8431 |
+
"grad_norm": 0.0008609068463556468,
|
8432 |
+
"learning_rate": 2.6531852641412082e-05,
|
8433 |
+
"loss": 46.0,
|
8434 |
+
"step": 1199
|
8435 |
+
},
|
8436 |
+
{
|
8437 |
+
"epoch": 0.024155318699236088,
|
8438 |
+
"grad_norm": 0.0013050955021753907,
|
8439 |
+
"learning_rate": 2.6395292804824244e-05,
|
8440 |
+
"loss": 46.0,
|
8441 |
+
"step": 1200
|
8442 |
+
},
|
8443 |
+
{
|
8444 |
+
"epoch": 0.024175448131485453,
|
8445 |
+
"grad_norm": 0.00126195780467242,
|
8446 |
+
"learning_rate": 2.625903185931853e-05,
|
8447 |
+
"loss": 46.0,
|
8448 |
+
"step": 1201
|
8449 |
+
},
|
8450 |
+
{
|
8451 |
+
"epoch": 0.024195577563734815,
|
8452 |
+
"grad_norm": 0.001260231714695692,
|
8453 |
+
"learning_rate": 2.612307035821805e-05,
|
8454 |
+
"loss": 46.0,
|
8455 |
+
"step": 1202
|
8456 |
+
},
|
8457 |
+
{
|
8458 |
+
"epoch": 0.024215706995984177,
|
8459 |
+
"grad_norm": 0.0008337153121829033,
|
8460 |
+
"learning_rate": 2.5987408853629846e-05,
|
8461 |
+
"loss": 46.0,
|
8462 |
+
"step": 1203
|
8463 |
+
},
|
8464 |
+
{
|
8465 |
+
"epoch": 0.024235836428233543,
|
8466 |
+
"grad_norm": 0.0017989326734095812,
|
8467 |
+
"learning_rate": 2.5852047896442853e-05,
|
8468 |
+
"loss": 46.0,
|
8469 |
+
"step": 1204
|
8470 |
+
},
|
8471 |
+
{
|
8472 |
+
"epoch": 0.024255965860482905,
|
8473 |
+
"grad_norm": 0.0008393581956624985,
|
8474 |
+
"learning_rate": 2.5716988036325386e-05,
|
8475 |
+
"loss": 46.0,
|
8476 |
+
"step": 1205
|
8477 |
+
},
|
8478 |
+
{
|
8479 |
+
"epoch": 0.02427609529273227,
|
8480 |
+
"grad_norm": 0.001092239166609943,
|
8481 |
+
"learning_rate": 2.5582229821723257e-05,
|
8482 |
+
"loss": 46.0,
|
8483 |
+
"step": 1206
|
8484 |
+
},
|
8485 |
+
{
|
8486 |
+
"epoch": 0.024296224724981632,
|
8487 |
+
"grad_norm": 0.0011140386341139674,
|
8488 |
+
"learning_rate": 2.5447773799857244e-05,
|
8489 |
+
"loss": 46.0,
|
8490 |
+
"step": 1207
|
8491 |
+
},
|
8492 |
+
{
|
8493 |
+
"epoch": 0.024316354157230994,
|
8494 |
+
"grad_norm": 0.0014232645044103265,
|
8495 |
+
"learning_rate": 2.5313620516721105e-05,
|
8496 |
+
"loss": 46.0,
|
8497 |
+
"step": 1208
|
8498 |
+
},
|
8499 |
+
{
|
8500 |
+
"epoch": 0.02433648358948036,
|
8501 |
+
"grad_norm": 0.0015523895854130387,
|
8502 |
+
"learning_rate": 2.5179770517079093e-05,
|
8503 |
+
"loss": 46.0,
|
8504 |
+
"step": 1209
|
8505 |
+
},
|
8506 |
+
{
|
8507 |
+
"epoch": 0.02435661302172972,
|
8508 |
+
"grad_norm": 0.0020029693841934204,
|
8509 |
+
"learning_rate": 2.5046224344464074e-05,
|
8510 |
+
"loss": 46.0,
|
8511 |
+
"step": 1210
|
8512 |
+
},
|
8513 |
+
{
|
8514 |
+
"epoch": 0.024376742453979087,
|
8515 |
+
"grad_norm": 0.0009358267998322845,
|
8516 |
+
"learning_rate": 2.4912982541175033e-05,
|
8517 |
+
"loss": 46.0,
|
8518 |
+
"step": 1211
|
8519 |
+
},
|
8520 |
+
{
|
8521 |
+
"epoch": 0.02439687188622845,
|
8522 |
+
"grad_norm": 0.0009875150863081217,
|
8523 |
+
"learning_rate": 2.4780045648274975e-05,
|
8524 |
+
"loss": 46.0,
|
8525 |
+
"step": 1212
|
8526 |
+
},
|
8527 |
+
{
|
8528 |
+
"epoch": 0.02441700131847781,
|
8529 |
+
"grad_norm": 0.0015209922567009926,
|
8530 |
+
"learning_rate": 2.4647414205588827e-05,
|
8531 |
+
"loss": 46.0,
|
8532 |
+
"step": 1213
|
8533 |
+
},
|
8534 |
+
{
|
8535 |
+
"epoch": 0.024437130750727176,
|
8536 |
+
"grad_norm": 0.0008846839773468673,
|
8537 |
+
"learning_rate": 2.451508875170104e-05,
|
8538 |
+
"loss": 46.0,
|
8539 |
+
"step": 1214
|
8540 |
+
},
|
8541 |
+
{
|
8542 |
+
"epoch": 0.024457260182976538,
|
8543 |
+
"grad_norm": 0.0009960222523659468,
|
8544 |
+
"learning_rate": 2.4383069823953663e-05,
|
8545 |
+
"loss": 46.0,
|
8546 |
+
"step": 1215
|
8547 |
+
},
|
8548 |
+
{
|
8549 |
+
"epoch": 0.024477389615225904,
|
8550 |
+
"grad_norm": 0.001352517050690949,
|
8551 |
+
"learning_rate": 2.4251357958443867e-05,
|
8552 |
+
"loss": 46.0,
|
8553 |
+
"step": 1216
|
8554 |
+
},
|
8555 |
+
{
|
8556 |
+
"epoch": 0.024497519047475266,
|
8557 |
+
"grad_norm": 0.0013506961986422539,
|
8558 |
+
"learning_rate": 2.4119953690022025e-05,
|
8559 |
+
"loss": 46.0,
|
8560 |
+
"step": 1217
|
8561 |
+
},
|
8562 |
+
{
|
8563 |
+
"epoch": 0.02451764847972463,
|
8564 |
+
"grad_norm": 0.0009574260911904275,
|
8565 |
+
"learning_rate": 2.3988857552289333e-05,
|
8566 |
+
"loss": 46.0,
|
8567 |
+
"step": 1218
|
8568 |
+
},
|
8569 |
+
{
|
8570 |
+
"epoch": 0.024537777911973993,
|
8571 |
+
"grad_norm": 0.0007269601919688284,
|
8572 |
+
"learning_rate": 2.3858070077595908e-05,
|
8573 |
+
"loss": 46.0,
|
8574 |
+
"step": 1219
|
8575 |
+
},
|
8576 |
+
{
|
8577 |
+
"epoch": 0.024557907344223355,
|
8578 |
+
"grad_norm": 0.0006699137156829238,
|
8579 |
+
"learning_rate": 2.372759179703822e-05,
|
8580 |
+
"loss": 46.0,
|
8581 |
+
"step": 1220
|
8582 |
+
},
|
8583 |
+
{
|
8584 |
+
"epoch": 0.02457803677647272,
|
8585 |
+
"grad_norm": 0.0015291008166968822,
|
8586 |
+
"learning_rate": 2.3597423240457395e-05,
|
8587 |
+
"loss": 46.0,
|
8588 |
+
"step": 1221
|
8589 |
+
},
|
8590 |
+
{
|
8591 |
+
"epoch": 0.024598166208722082,
|
8592 |
+
"grad_norm": 0.002010623225942254,
|
8593 |
+
"learning_rate": 2.3467564936436703e-05,
|
8594 |
+
"loss": 46.0,
|
8595 |
+
"step": 1222
|
8596 |
+
},
|
8597 |
+
{
|
8598 |
+
"epoch": 0.024618295640971448,
|
8599 |
+
"grad_norm": 0.0008198167197406292,
|
8600 |
+
"learning_rate": 2.3338017412299663e-05,
|
8601 |
+
"loss": 46.0,
|
8602 |
+
"step": 1223
|
8603 |
+
},
|
8604 |
+
{
|
8605 |
+
"epoch": 0.02463842507322081,
|
8606 |
+
"grad_norm": 0.0006754023488610983,
|
8607 |
+
"learning_rate": 2.3208781194107664e-05,
|
8608 |
+
"loss": 46.0,
|
8609 |
+
"step": 1224
|
8610 |
+
},
|
8611 |
+
{
|
8612 |
+
"epoch": 0.024658554505470172,
|
8613 |
+
"grad_norm": 0.0009974197018891573,
|
8614 |
+
"learning_rate": 2.3079856806658107e-05,
|
8615 |
+
"loss": 46.0,
|
8616 |
+
"step": 1225
|
8617 |
+
},
|
8618 |
+
{
|
8619 |
+
"epoch": 0.024678683937719537,
|
8620 |
+
"grad_norm": 0.0012742577819153666,
|
8621 |
+
"learning_rate": 2.2951244773481995e-05,
|
8622 |
+
"loss": 46.0,
|
8623 |
+
"step": 1226
|
8624 |
+
},
|
8625 |
+
{
|
8626 |
+
"epoch": 0.0246988133699689,
|
8627 |
+
"grad_norm": 0.0013213262427598238,
|
8628 |
+
"learning_rate": 2.2822945616841963e-05,
|
8629 |
+
"loss": 46.0,
|
8630 |
+
"step": 1227
|
8631 |
+
},
|
8632 |
+
{
|
8633 |
+
"epoch": 0.024718942802218265,
|
8634 |
+
"grad_norm": 0.0006177601171657443,
|
8635 |
+
"learning_rate": 2.2694959857730204e-05,
|
8636 |
+
"loss": 46.0,
|
8637 |
+
"step": 1228
|
8638 |
+
},
|
8639 |
+
{
|
8640 |
+
"epoch": 0.024739072234467627,
|
8641 |
+
"grad_norm": 0.0009409788763150573,
|
8642 |
+
"learning_rate": 2.256728801586616e-05,
|
8643 |
+
"loss": 46.0,
|
8644 |
+
"step": 1229
|
8645 |
+
},
|
8646 |
+
{
|
8647 |
+
"epoch": 0.02475920166671699,
|
8648 |
+
"grad_norm": 0.001638015266507864,
|
8649 |
+
"learning_rate": 2.2439930609694658e-05,
|
8650 |
+
"loss": 46.0,
|
8651 |
+
"step": 1230
|
8652 |
+
},
|
8653 |
+
{
|
8654 |
+
"epoch": 0.024779331098966354,
|
8655 |
+
"grad_norm": 0.000653863069601357,
|
8656 |
+
"learning_rate": 2.231288815638356e-05,
|
8657 |
+
"loss": 46.0,
|
8658 |
+
"step": 1231
|
8659 |
+
},
|
8660 |
+
{
|
8661 |
+
"epoch": 0.024799460531215716,
|
8662 |
+
"grad_norm": 0.001285205944441259,
|
8663 |
+
"learning_rate": 2.2186161171821885e-05,
|
8664 |
+
"loss": 46.0,
|
8665 |
+
"step": 1232
|
8666 |
+
},
|
8667 |
+
{
|
8668 |
+
"epoch": 0.02481958996346508,
|
8669 |
+
"grad_norm": 0.001490448834374547,
|
8670 |
+
"learning_rate": 2.2059750170617523e-05,
|
8671 |
+
"loss": 46.0,
|
8672 |
+
"step": 1233
|
8673 |
+
},
|
8674 |
+
{
|
8675 |
+
"epoch": 0.024839719395714444,
|
8676 |
+
"grad_norm": 0.0018423368455842137,
|
8677 |
+
"learning_rate": 2.1933655666095275e-05,
|
8678 |
+
"loss": 46.0,
|
8679 |
+
"step": 1234
|
8680 |
+
},
|
8681 |
+
{
|
8682 |
+
"epoch": 0.02485984882796381,
|
8683 |
+
"grad_norm": 0.0011196645209565759,
|
8684 |
+
"learning_rate": 2.180787817029468e-05,
|
8685 |
+
"loss": 46.0,
|
8686 |
+
"step": 1235
|
8687 |
+
},
|
8688 |
+
{
|
8689 |
+
"epoch": 0.02487997826021317,
|
8690 |
+
"grad_norm": 0.000826796458568424,
|
8691 |
+
"learning_rate": 2.168241819396808e-05,
|
8692 |
+
"loss": 46.0,
|
8693 |
+
"step": 1236
|
8694 |
+
},
|
8695 |
+
{
|
8696 |
+
"epoch": 0.024900107692462533,
|
8697 |
+
"grad_norm": 0.0005586759652942419,
|
8698 |
+
"learning_rate": 2.1557276246578307e-05,
|
8699 |
+
"loss": 46.0,
|
8700 |
+
"step": 1237
|
8701 |
+
},
|
8702 |
+
{
|
8703 |
+
"epoch": 0.0249202371247119,
|
8704 |
+
"grad_norm": 0.0011236952850595117,
|
8705 |
+
"learning_rate": 2.1432452836296914e-05,
|
8706 |
+
"loss": 46.0,
|
8707 |
+
"step": 1238
|
8708 |
+
},
|
8709 |
+
{
|
8710 |
+
"epoch": 0.02494036655696126,
|
8711 |
+
"grad_norm": 0.001449284260161221,
|
8712 |
+
"learning_rate": 2.1307948470001782e-05,
|
8713 |
+
"loss": 46.0,
|
8714 |
+
"step": 1239
|
8715 |
+
},
|
8716 |
+
{
|
8717 |
+
"epoch": 0.024960495989210626,
|
8718 |
+
"grad_norm": 0.0007015886367298663,
|
8719 |
+
"learning_rate": 2.1183763653275378e-05,
|
8720 |
+
"loss": 46.0,
|
8721 |
+
"step": 1240
|
8722 |
+
},
|
8723 |
+
{
|
8724 |
+
"epoch": 0.024980625421459988,
|
8725 |
+
"grad_norm": 0.0011342605575919151,
|
8726 |
+
"learning_rate": 2.1059898890402473e-05,
|
8727 |
+
"loss": 46.0,
|
8728 |
+
"step": 1241
|
8729 |
+
},
|
8730 |
+
{
|
8731 |
+
"epoch": 0.02500075485370935,
|
8732 |
+
"grad_norm": 0.0012565052602440119,
|
8733 |
+
"learning_rate": 2.0936354684368163e-05,
|
8734 |
+
"loss": 46.0,
|
8735 |
+
"step": 1242
|
8736 |
+
},
|
8737 |
+
{
|
8738 |
+
"epoch": 0.025020884285958715,
|
8739 |
+
"grad_norm": 0.001035231165587902,
|
8740 |
+
"learning_rate": 2.0813131536855913e-05,
|
8741 |
+
"loss": 46.0,
|
8742 |
+
"step": 1243
|
8743 |
+
},
|
8744 |
+
{
|
8745 |
+
"epoch": 0.025041013718208077,
|
8746 |
+
"grad_norm": 0.0019417139701545238,
|
8747 |
+
"learning_rate": 2.0690229948245365e-05,
|
8748 |
+
"loss": 46.0,
|
8749 |
+
"step": 1244
|
8750 |
+
},
|
8751 |
+
{
|
8752 |
+
"epoch": 0.025061143150457443,
|
8753 |
+
"grad_norm": 0.0015144539065659046,
|
8754 |
+
"learning_rate": 2.0567650417610485e-05,
|
8755 |
+
"loss": 46.0,
|
8756 |
+
"step": 1245
|
8757 |
+
},
|
8758 |
+
{
|
8759 |
+
"epoch": 0.025081272582706805,
|
8760 |
+
"grad_norm": 0.0014989918563514948,
|
8761 |
+
"learning_rate": 2.0445393442717308e-05,
|
8762 |
+
"loss": 46.0,
|
8763 |
+
"step": 1246
|
8764 |
+
},
|
8765 |
+
{
|
8766 |
+
"epoch": 0.025101402014956167,
|
8767 |
+
"grad_norm": 0.0018054774263873696,
|
8768 |
+
"learning_rate": 2.032345952002218e-05,
|
8769 |
+
"loss": 46.0,
|
8770 |
+
"step": 1247
|
8771 |
+
},
|
8772 |
+
{
|
8773 |
+
"epoch": 0.025121531447205532,
|
8774 |
+
"grad_norm": 0.0009529809467494488,
|
8775 |
+
"learning_rate": 2.0201849144669525e-05,
|
8776 |
+
"loss": 46.0,
|
8777 |
+
"step": 1248
|
8778 |
+
},
|
8779 |
+
{
|
8780 |
+
"epoch": 0.025141660879454894,
|
8781 |
+
"grad_norm": 0.002208658494055271,
|
8782 |
+
"learning_rate": 2.0080562810489935e-05,
|
8783 |
+
"loss": 46.0,
|
8784 |
+
"step": 1249
|
8785 |
+
},
|
8786 |
+
{
|
8787 |
+
"epoch": 0.02516179031170426,
|
8788 |
+
"grad_norm": 0.0007078711641952395,
|
8789 |
+
"learning_rate": 1.995960100999814e-05,
|
8790 |
+
"loss": 46.0,
|
8791 |
+
"step": 1250
|
8792 |
+
},
|
8793 |
+
{
|
8794 |
+
"epoch": 0.02518191974395362,
|
8795 |
+
"grad_norm": 0.0010136510245501995,
|
8796 |
+
"learning_rate": 1.983896423439109e-05,
|
8797 |
+
"loss": 46.0,
|
8798 |
+
"step": 1251
|
8799 |
+
},
|
8800 |
+
{
|
8801 |
+
"epoch": 0.025202049176202987,
|
8802 |
+
"grad_norm": 0.0018538066651672125,
|
8803 |
+
"learning_rate": 1.9718652973545758e-05,
|
8804 |
+
"loss": 46.0,
|
8805 |
+
"step": 1252
|
8806 |
+
},
|
8807 |
+
{
|
8808 |
+
"epoch": 0.02522217860845235,
|
8809 |
+
"grad_norm": 0.0011417543282732368,
|
8810 |
+
"learning_rate": 1.9598667716017417e-05,
|
8811 |
+
"loss": 46.0,
|
8812 |
+
"step": 1253
|
8813 |
+
},
|
8814 |
+
{
|
8815 |
+
"epoch": 0.02524230804070171,
|
8816 |
+
"grad_norm": 0.000996634247712791,
|
8817 |
+
"learning_rate": 1.947900894903739e-05,
|
8818 |
+
"loss": 46.0,
|
8819 |
+
"step": 1254
|
8820 |
+
},
|
8821 |
+
{
|
8822 |
+
"epoch": 0.025262437472951076,
|
8823 |
+
"grad_norm": 0.001392628182657063,
|
8824 |
+
"learning_rate": 1.9359677158511334e-05,
|
8825 |
+
"loss": 46.0,
|
8826 |
+
"step": 1255
|
8827 |
+
},
|
8828 |
+
{
|
8829 |
+
"epoch": 0.02528256690520044,
|
8830 |
+
"grad_norm": 0.000720780692063272,
|
8831 |
+
"learning_rate": 1.9240672829017014e-05,
|
8832 |
+
"loss": 46.0,
|
8833 |
+
"step": 1256
|
8834 |
+
},
|
8835 |
+
{
|
8836 |
+
"epoch": 0.025302696337449804,
|
8837 |
+
"grad_norm": 0.001160190673545003,
|
8838 |
+
"learning_rate": 1.9121996443802482e-05,
|
8839 |
+
"loss": 46.0,
|
8840 |
+
"step": 1257
|
8841 |
+
},
|
8842 |
+
{
|
8843 |
+
"epoch": 0.025322825769699166,
|
8844 |
+
"grad_norm": 0.002251293743029237,
|
8845 |
+
"learning_rate": 1.9003648484784166e-05,
|
8846 |
+
"loss": 46.0,
|
8847 |
+
"step": 1258
|
8848 |
+
},
|
8849 |
+
{
|
8850 |
+
"epoch": 0.025342955201948528,
|
8851 |
+
"grad_norm": 0.00043235233169980347,
|
8852 |
+
"learning_rate": 1.8885629432544717e-05,
|
8853 |
+
"loss": 46.0,
|
8854 |
+
"step": 1259
|
8855 |
+
},
|
8856 |
+
{
|
8857 |
+
"epoch": 0.025363084634197893,
|
8858 |
+
"grad_norm": 0.0007145190611481667,
|
8859 |
+
"learning_rate": 1.8767939766331287e-05,
|
8860 |
+
"loss": 46.0,
|
8861 |
+
"step": 1260
|
8862 |
+
},
|
8863 |
+
{
|
8864 |
+
"epoch": 0.025383214066447255,
|
8865 |
+
"grad_norm": 0.0006261473754420877,
|
8866 |
+
"learning_rate": 1.865057996405336e-05,
|
8867 |
+
"loss": 46.0,
|
8868 |
+
"step": 1261
|
8869 |
+
},
|
8870 |
+
{
|
8871 |
+
"epoch": 0.02540334349869662,
|
8872 |
+
"grad_norm": 0.0010795745765790343,
|
8873 |
+
"learning_rate": 1.8533550502281015e-05,
|
8874 |
+
"loss": 46.0,
|
8875 |
+
"step": 1262
|
8876 |
+
},
|
8877 |
+
{
|
8878 |
+
"epoch": 0.025423472930945983,
|
8879 |
+
"grad_norm": 0.0009830056224018335,
|
8880 |
+
"learning_rate": 1.8416851856242868e-05,
|
8881 |
+
"loss": 46.0,
|
8882 |
+
"step": 1263
|
8883 |
+
},
|
8884 |
+
{
|
8885 |
+
"epoch": 0.025443602363195345,
|
8886 |
+
"grad_norm": 0.0014814147725701332,
|
8887 |
+
"learning_rate": 1.83004844998241e-05,
|
8888 |
+
"loss": 46.0,
|
8889 |
+
"step": 1264
|
8890 |
+
},
|
8891 |
+
{
|
8892 |
+
"epoch": 0.02546373179544471,
|
8893 |
+
"grad_norm": 0.0011380594223737717,
|
8894 |
+
"learning_rate": 1.8184448905564743e-05,
|
8895 |
+
"loss": 46.0,
|
8896 |
+
"step": 1265
|
8897 |
+
},
|
8898 |
+
{
|
8899 |
+
"epoch": 0.025483861227694072,
|
8900 |
+
"grad_norm": 0.0006726674037054181,
|
8901 |
+
"learning_rate": 1.8068745544657484e-05,
|
8902 |
+
"loss": 46.0,
|
8903 |
+
"step": 1266
|
8904 |
+
},
|
8905 |
+
{
|
8906 |
+
"epoch": 0.025503990659943437,
|
8907 |
+
"grad_norm": 0.0017750106053426862,
|
8908 |
+
"learning_rate": 1.7953374886946006e-05,
|
8909 |
+
"loss": 46.0,
|
8910 |
+
"step": 1267
|
8911 |
+
},
|
8912 |
+
{
|
8913 |
+
"epoch": 0.0255241200921928,
|
8914 |
+
"grad_norm": 0.0011626784689724445,
|
8915 |
+
"learning_rate": 1.7838337400922855e-05,
|
8916 |
+
"loss": 46.0,
|
8917 |
+
"step": 1268
|
8918 |
+
},
|
8919 |
+
{
|
8920 |
+
"epoch": 0.025544249524442165,
|
8921 |
+
"grad_norm": 0.0009647434344515204,
|
8922 |
+
"learning_rate": 1.772363355372776e-05,
|
8923 |
+
"loss": 46.0,
|
8924 |
+
"step": 1269
|
8925 |
+
},
|
8926 |
+
{
|
8927 |
+
"epoch": 0.025564378956691527,
|
8928 |
+
"grad_norm": 0.002181010087952018,
|
8929 |
+
"learning_rate": 1.7609263811145537e-05,
|
8930 |
+
"loss": 46.0,
|
8931 |
+
"step": 1270
|
8932 |
+
},
|
8933 |
+
{
|
8934 |
+
"epoch": 0.02558450838894089,
|
8935 |
+
"grad_norm": 0.0012630668934434652,
|
8936 |
+
"learning_rate": 1.74952286376043e-05,
|
8937 |
+
"loss": 46.0,
|
8938 |
+
"step": 1271
|
8939 |
+
},
|
8940 |
+
{
|
8941 |
+
"epoch": 0.025604637821190254,
|
8942 |
+
"grad_norm": 0.0009720325469970703,
|
8943 |
+
"learning_rate": 1.738152849617356e-05,
|
8944 |
+
"loss": 46.0,
|
8945 |
+
"step": 1272
|
8946 |
+
},
|
8947 |
+
{
|
8948 |
+
"epoch": 0.025624767253439616,
|
8949 |
+
"grad_norm": 0.0007137717329896986,
|
8950 |
+
"learning_rate": 1.72681638485624e-05,
|
8951 |
+
"loss": 46.0,
|
8952 |
+
"step": 1273
|
8953 |
+
},
|
8954 |
+
{
|
8955 |
+
"epoch": 0.02564489668568898,
|
8956 |
+
"grad_norm": 0.0014153624651953578,
|
8957 |
+
"learning_rate": 1.715513515511743e-05,
|
8958 |
+
"loss": 46.0,
|
8959 |
+
"step": 1274
|
8960 |
+
},
|
8961 |
+
{
|
8962 |
+
"epoch": 0.025665026117938344,
|
8963 |
+
"grad_norm": 0.001465531880967319,
|
8964 |
+
"learning_rate": 1.7042442874821164e-05,
|
8965 |
+
"loss": 46.0,
|
8966 |
+
"step": 1275
|
8967 |
+
},
|
8968 |
+
{
|
8969 |
+
"epoch": 0.025685155550187706,
|
8970 |
+
"grad_norm": 0.0011013116454705596,
|
8971 |
+
"learning_rate": 1.693008746528988e-05,
|
8972 |
+
"loss": 46.0,
|
8973 |
+
"step": 1276
|
8974 |
+
},
|
8975 |
+
{
|
8976 |
+
"epoch": 0.02570528498243707,
|
8977 |
+
"grad_norm": 0.0005420059314928949,
|
8978 |
+
"learning_rate": 1.681806938277205e-05,
|
8979 |
+
"loss": 46.0,
|
8980 |
+
"step": 1277
|
8981 |
+
},
|
8982 |
+
{
|
8983 |
+
"epoch": 0.025725414414686433,
|
8984 |
+
"grad_norm": 0.0007358815055340528,
|
8985 |
+
"learning_rate": 1.6706389082146244e-05,
|
8986 |
+
"loss": 46.0,
|
8987 |
+
"step": 1278
|
8988 |
+
},
|
8989 |
+
{
|
8990 |
+
"epoch": 0.0257455438469358,
|
8991 |
+
"grad_norm": 0.00043846582411788404,
|
8992 |
+
"learning_rate": 1.6595047016919373e-05,
|
8993 |
+
"loss": 46.0,
|
8994 |
+
"step": 1279
|
8995 |
+
},
|
8996 |
+
{
|
8997 |
+
"epoch": 0.02576567327918516,
|
8998 |
+
"grad_norm": 0.0007551101734861732,
|
8999 |
+
"learning_rate": 1.6484043639224955e-05,
|
9000 |
+
"loss": 46.0,
|
9001 |
+
"step": 1280
|
9002 |
+
},
|
9003 |
+
{
|
9004 |
+
"epoch": 0.025785802711434522,
|
9005 |
+
"grad_norm": 0.0011386704863980412,
|
9006 |
+
"learning_rate": 1.6373379399821033e-05,
|
9007 |
+
"loss": 46.0,
|
9008 |
+
"step": 1281
|
9009 |
+
},
|
9010 |
+
{
|
9011 |
+
"epoch": 0.025805932143683888,
|
9012 |
+
"grad_norm": 0.0014110167976468801,
|
9013 |
+
"learning_rate": 1.6263054748088658e-05,
|
9014 |
+
"loss": 46.0,
|
9015 |
+
"step": 1282
|
9016 |
+
},
|
9017 |
+
{
|
9018 |
+
"epoch": 0.02582606157593325,
|
9019 |
+
"grad_norm": 0.0017660473240539432,
|
9020 |
+
"learning_rate": 1.6153070132029723e-05,
|
9021 |
+
"loss": 46.0,
|
9022 |
+
"step": 1283
|
9023 |
+
},
|
9024 |
+
{
|
9025 |
+
"epoch": 0.025846191008182615,
|
9026 |
+
"grad_norm": 0.0021693487651646137,
|
9027 |
+
"learning_rate": 1.604342599826548e-05,
|
9028 |
+
"loss": 46.0,
|
9029 |
+
"step": 1284
|
9030 |
+
},
|
9031 |
+
{
|
9032 |
+
"epoch": 0.025866320440431977,
|
9033 |
+
"grad_norm": 0.001375765772536397,
|
9034 |
+
"learning_rate": 1.593412279203447e-05,
|
9035 |
+
"loss": 46.0,
|
9036 |
+
"step": 1285
|
9037 |
+
},
|
9038 |
+
{
|
9039 |
+
"epoch": 0.025886449872681343,
|
9040 |
+
"grad_norm": 0.001402435707859695,
|
9041 |
+
"learning_rate": 1.5825160957190798e-05,
|
9042 |
+
"loss": 46.0,
|
9043 |
+
"step": 1286
|
9044 |
+
},
|
9045 |
+
{
|
9046 |
+
"epoch": 0.025906579304930705,
|
9047 |
+
"grad_norm": 0.002132172929123044,
|
9048 |
+
"learning_rate": 1.5716540936202363e-05,
|
9049 |
+
"loss": 46.0,
|
9050 |
+
"step": 1287
|
9051 |
+
},
|
9052 |
+
{
|
9053 |
+
"epoch": 0.025926708737180067,
|
9054 |
+
"grad_norm": 0.00138960184995085,
|
9055 |
+
"learning_rate": 1.5608263170149095e-05,
|
9056 |
+
"loss": 46.0,
|
9057 |
+
"step": 1288
|
9058 |
+
},
|
9059 |
+
{
|
9060 |
+
"epoch": 0.025946838169429432,
|
9061 |
+
"grad_norm": 0.0014584781602025032,
|
9062 |
+
"learning_rate": 1.5500328098721017e-05,
|
9063 |
+
"loss": 46.0,
|
9064 |
+
"step": 1289
|
9065 |
+
},
|
9066 |
+
{
|
9067 |
+
"epoch": 0.025966967601678794,
|
9068 |
+
"grad_norm": 0.0017797322943806648,
|
9069 |
+
"learning_rate": 1.5392736160216635e-05,
|
9070 |
+
"loss": 46.0,
|
9071 |
+
"step": 1290
|
9072 |
+
},
|
9073 |
+
{
|
9074 |
+
"epoch": 0.02598709703392816,
|
9075 |
+
"grad_norm": 0.002001388929784298,
|
9076 |
+
"learning_rate": 1.5285487791541e-05,
|
9077 |
+
"loss": 46.0,
|
9078 |
+
"step": 1291
|
9079 |
+
},
|
9080 |
+
{
|
9081 |
+
"epoch": 0.02600722646617752,
|
9082 |
+
"grad_norm": 0.0019802197348326445,
|
9083 |
+
"learning_rate": 1.5178583428204085e-05,
|
9084 |
+
"loss": 46.0,
|
9085 |
+
"step": 1292
|
9086 |
+
},
|
9087 |
+
{
|
9088 |
+
"epoch": 0.026027355898426884,
|
9089 |
+
"grad_norm": 0.0018057803390547633,
|
9090 |
+
"learning_rate": 1.5072023504318867e-05,
|
9091 |
+
"loss": 46.0,
|
9092 |
+
"step": 1293
|
9093 |
+
},
|
9094 |
+
{
|
9095 |
+
"epoch": 0.02604748533067625,
|
9096 |
+
"grad_norm": 0.0009447969496250153,
|
9097 |
+
"learning_rate": 1.496580845259965e-05,
|
9098 |
+
"loss": 46.0,
|
9099 |
+
"step": 1294
|
9100 |
+
},
|
9101 |
+
{
|
9102 |
+
"epoch": 0.02606761476292561,
|
9103 |
+
"grad_norm": 0.0007863205391913652,
|
9104 |
+
"learning_rate": 1.4859938704360365e-05,
|
9105 |
+
"loss": 46.0,
|
9106 |
+
"step": 1295
|
9107 |
+
},
|
9108 |
+
{
|
9109 |
+
"epoch": 0.026087744195174976,
|
9110 |
+
"grad_norm": 0.0012318964581936598,
|
9111 |
+
"learning_rate": 1.475441468951263e-05,
|
9112 |
+
"loss": 46.0,
|
9113 |
+
"step": 1296
|
9114 |
+
},
|
9115 |
+
{
|
9116 |
+
"epoch": 0.02610787362742434,
|
9117 |
+
"grad_norm": 0.001262698438949883,
|
9118 |
+
"learning_rate": 1.4649236836564263e-05,
|
9119 |
+
"loss": 46.0,
|
9120 |
+
"step": 1297
|
9121 |
+
},
|
9122 |
+
{
|
9123 |
+
"epoch": 0.0261280030596737,
|
9124 |
+
"grad_norm": 0.0010360804153606296,
|
9125 |
+
"learning_rate": 1.4544405572617259e-05,
|
9126 |
+
"loss": 46.0,
|
9127 |
+
"step": 1298
|
9128 |
+
},
|
9129 |
+
{
|
9130 |
+
"epoch": 0.026148132491923066,
|
9131 |
+
"grad_norm": 0.001501108636148274,
|
9132 |
+
"learning_rate": 1.4439921323366323e-05,
|
9133 |
+
"loss": 46.0,
|
9134 |
+
"step": 1299
|
9135 |
+
},
|
9136 |
+
{
|
9137 |
+
"epoch": 0.026168261924172428,
|
9138 |
+
"grad_norm": 0.0006216936744749546,
|
9139 |
+
"learning_rate": 1.4335784513096929e-05,
|
9140 |
+
"loss": 46.0,
|
9141 |
+
"step": 1300
|
9142 |
+
},
|
9143 |
+
{
|
9144 |
+
"epoch": 0.026188391356421793,
|
9145 |
+
"grad_norm": 0.0012842519208788872,
|
9146 |
+
"learning_rate": 1.4231995564683732e-05,
|
9147 |
+
"loss": 46.0,
|
9148 |
+
"step": 1301
|
9149 |
+
},
|
9150 |
+
{
|
9151 |
+
"epoch": 0.026208520788671155,
|
9152 |
+
"grad_norm": 0.0013369874795898795,
|
9153 |
+
"learning_rate": 1.412855489958873e-05,
|
9154 |
+
"loss": 46.0,
|
9155 |
+
"step": 1302
|
9156 |
+
},
|
9157 |
+
{
|
9158 |
+
"epoch": 0.02622865022092052,
|
9159 |
+
"grad_norm": 0.0011299944017082453,
|
9160 |
+
"learning_rate": 1.4025462937859768e-05,
|
9161 |
+
"loss": 46.0,
|
9162 |
+
"step": 1303
|
9163 |
+
},
|
9164 |
+
{
|
9165 |
+
"epoch": 0.026248779653169883,
|
9166 |
+
"grad_norm": 0.0018812668276950717,
|
9167 |
+
"learning_rate": 1.3922720098128527e-05,
|
9168 |
+
"loss": 46.0,
|
9169 |
+
"step": 1304
|
9170 |
+
},
|
9171 |
+
{
|
9172 |
+
"epoch": 0.026268909085419245,
|
9173 |
+
"grad_norm": 0.001090607256628573,
|
9174 |
+
"learning_rate": 1.3820326797609129e-05,
|
9175 |
+
"loss": 46.0,
|
9176 |
+
"step": 1305
|
9177 |
+
},
|
9178 |
+
{
|
9179 |
+
"epoch": 0.02628903851766861,
|
9180 |
+
"grad_norm": 0.0012601654743775725,
|
9181 |
+
"learning_rate": 1.371828345209618e-05,
|
9182 |
+
"loss": 46.0,
|
9183 |
+
"step": 1306
|
9184 |
+
},
|
9185 |
+
{
|
9186 |
+
"epoch": 0.026309167949917972,
|
9187 |
+
"grad_norm": 0.0014517259551212192,
|
9188 |
+
"learning_rate": 1.361659047596332e-05,
|
9189 |
+
"loss": 46.0,
|
9190 |
+
"step": 1307
|
9191 |
+
},
|
9192 |
+
{
|
9193 |
+
"epoch": 0.026329297382167337,
|
9194 |
+
"grad_norm": 0.0019068201072514057,
|
9195 |
+
"learning_rate": 1.3515248282161319e-05,
|
9196 |
+
"loss": 46.0,
|
9197 |
+
"step": 1308
|
9198 |
+
},
|
9199 |
+
{
|
9200 |
+
"epoch": 0.0263494268144167,
|
9201 |
+
"grad_norm": 0.0016664571594446898,
|
9202 |
+
"learning_rate": 1.3414257282216535e-05,
|
9203 |
+
"loss": 46.0,
|
9204 |
+
"step": 1309
|
9205 |
+
},
|
9206 |
+
{
|
9207 |
+
"epoch": 0.02636955624666606,
|
9208 |
+
"grad_norm": 0.0009046939085237682,
|
9209 |
+
"learning_rate": 1.3313617886229269e-05,
|
9210 |
+
"loss": 46.0,
|
9211 |
+
"step": 1310
|
9212 |
+
},
|
9213 |
+
{
|
9214 |
+
"epoch": 0.026389685678915427,
|
9215 |
+
"grad_norm": 0.0005408989964053035,
|
9216 |
+
"learning_rate": 1.3213330502871956e-05,
|
9217 |
+
"loss": 46.0,
|
9218 |
+
"step": 1311
|
9219 |
+
},
|
9220 |
+
{
|
9221 |
+
"epoch": 0.02640981511116479,
|
9222 |
+
"grad_norm": 0.0008995769894681871,
|
9223 |
+
"learning_rate": 1.3113395539387674e-05,
|
9224 |
+
"loss": 46.0,
|
9225 |
+
"step": 1312
|
9226 |
+
},
|
9227 |
+
{
|
9228 |
+
"epoch": 0.026429944543414154,
|
9229 |
+
"grad_norm": 0.0012701263185590506,
|
9230 |
+
"learning_rate": 1.3013813401588315e-05,
|
9231 |
+
"loss": 46.0,
|
9232 |
+
"step": 1313
|
9233 |
+
},
|
9234 |
+
{
|
9235 |
+
"epoch": 0.026450073975663516,
|
9236 |
+
"grad_norm": 0.001226755790412426,
|
9237 |
+
"learning_rate": 1.2914584493853144e-05,
|
9238 |
+
"loss": 46.0,
|
9239 |
+
"step": 1314
|
9240 |
+
},
|
9241 |
+
{
|
9242 |
+
"epoch": 0.026470203407912878,
|
9243 |
+
"grad_norm": 0.0015894804382696748,
|
9244 |
+
"learning_rate": 1.2815709219126959e-05,
|
9245 |
+
"loss": 46.0,
|
9246 |
+
"step": 1315
|
9247 |
+
},
|
9248 |
+
{
|
9249 |
+
"epoch": 0.026490332840162244,
|
9250 |
+
"grad_norm": 0.0014253195840865374,
|
9251 |
+
"learning_rate": 1.2717187978918544e-05,
|
9252 |
+
"loss": 46.0,
|
9253 |
+
"step": 1316
|
9254 |
+
},
|
9255 |
+
{
|
9256 |
+
"epoch": 0.026510462272411606,
|
9257 |
+
"grad_norm": 0.0009140470647253096,
|
9258 |
+
"learning_rate": 1.2619021173299051e-05,
|
9259 |
+
"loss": 46.0,
|
9260 |
+
"step": 1317
|
9261 |
+
},
|
9262 |
+
{
|
9263 |
+
"epoch": 0.02653059170466097,
|
9264 |
+
"grad_norm": 0.0010910599958151579,
|
9265 |
+
"learning_rate": 1.2521209200900397e-05,
|
9266 |
+
"loss": 46.0,
|
9267 |
+
"step": 1318
|
9268 |
+
},
|
9269 |
+
{
|
9270 |
+
"epoch": 0.026550721136910333,
|
9271 |
+
"grad_norm": 0.000989207299426198,
|
9272 |
+
"learning_rate": 1.2423752458913518e-05,
|
9273 |
+
"loss": 46.0,
|
9274 |
+
"step": 1319
|
9275 |
+
},
|
9276 |
+
{
|
9277 |
+
"epoch": 0.0265708505691597,
|
9278 |
+
"grad_norm": 0.0009602408390492201,
|
9279 |
+
"learning_rate": 1.2326651343086937e-05,
|
9280 |
+
"loss": 46.0,
|
9281 |
+
"step": 1320
|
9282 |
+
},
|
9283 |
+
{
|
9284 |
+
"epoch": 0.02659098000140906,
|
9285 |
+
"grad_norm": 0.0004408732638694346,
|
9286 |
+
"learning_rate": 1.2229906247724998e-05,
|
9287 |
+
"loss": 46.0,
|
9288 |
+
"step": 1321
|
9289 |
+
},
|
9290 |
+
{
|
9291 |
+
"epoch": 0.026611109433658423,
|
9292 |
+
"grad_norm": 0.0010518889175727963,
|
9293 |
+
"learning_rate": 1.2133517565686381e-05,
|
9294 |
+
"loss": 46.0,
|
9295 |
+
"step": 1322
|
9296 |
+
},
|
9297 |
+
{
|
9298 |
+
"epoch": 0.026631238865907788,
|
9299 |
+
"grad_norm": 0.002272763755172491,
|
9300 |
+
"learning_rate": 1.2037485688382421e-05,
|
9301 |
+
"loss": 46.0,
|
9302 |
+
"step": 1323
|
9303 |
+
},
|
9304 |
+
{
|
9305 |
+
"epoch": 0.02665136829815715,
|
9306 |
+
"grad_norm": 0.0020480677485466003,
|
9307 |
+
"learning_rate": 1.1941811005775538e-05,
|
9308 |
+
"loss": 46.0,
|
9309 |
+
"step": 1324
|
9310 |
+
},
|
9311 |
+
{
|
9312 |
+
"epoch": 0.026671497730406515,
|
9313 |
+
"grad_norm": 0.002253229497000575,
|
9314 |
+
"learning_rate": 1.1846493906377743e-05,
|
9315 |
+
"loss": 46.0,
|
9316 |
+
"step": 1325
|
9317 |
+
},
|
9318 |
+
{
|
9319 |
+
"epoch": 0.026691627162655877,
|
9320 |
+
"grad_norm": 0.0012804159196093678,
|
9321 |
+
"learning_rate": 1.1751534777248885e-05,
|
9322 |
+
"loss": 46.0,
|
9323 |
+
"step": 1326
|
9324 |
+
},
|
9325 |
+
{
|
9326 |
+
"epoch": 0.02671175659490524,
|
9327 |
+
"grad_norm": 0.0019482570933178067,
|
9328 |
+
"learning_rate": 1.1656934003995302e-05,
|
9329 |
+
"loss": 46.0,
|
9330 |
+
"step": 1327
|
9331 |
+
},
|
9332 |
+
{
|
9333 |
+
"epoch": 0.026731886027154605,
|
9334 |
+
"grad_norm": 0.0008674189448356628,
|
9335 |
+
"learning_rate": 1.1562691970768014e-05,
|
9336 |
+
"loss": 46.0,
|
9337 |
+
"step": 1328
|
9338 |
+
},
|
9339 |
+
{
|
9340 |
+
"epoch": 0.026752015459403967,
|
9341 |
+
"grad_norm": 0.0017704269848763943,
|
9342 |
+
"learning_rate": 1.1468809060261399e-05,
|
9343 |
+
"loss": 46.0,
|
9344 |
+
"step": 1329
|
9345 |
+
},
|
9346 |
+
{
|
9347 |
+
"epoch": 0.026772144891653332,
|
9348 |
+
"grad_norm": 0.0014249221421778202,
|
9349 |
+
"learning_rate": 1.1375285653711399e-05,
|
9350 |
+
"loss": 46.0,
|
9351 |
+
"step": 1330
|
9352 |
+
},
|
9353 |
+
{
|
9354 |
+
"epoch": 0.026792274323902694,
|
9355 |
+
"grad_norm": 0.0009925938211381435,
|
9356 |
+
"learning_rate": 1.1282122130894202e-05,
|
9357 |
+
"loss": 46.0,
|
9358 |
+
"step": 1331
|
9359 |
+
},
|
9360 |
+
{
|
9361 |
+
"epoch": 0.026812403756152056,
|
9362 |
+
"grad_norm": 0.0008217705762945116,
|
9363 |
+
"learning_rate": 1.1189318870124531e-05,
|
9364 |
+
"loss": 46.0,
|
9365 |
+
"step": 1332
|
9366 |
+
},
|
9367 |
+
{
|
9368 |
+
"epoch": 0.02683253318840142,
|
9369 |
+
"grad_norm": 0.002136779949069023,
|
9370 |
+
"learning_rate": 1.1096876248254228e-05,
|
9371 |
+
"loss": 46.0,
|
9372 |
+
"step": 1333
|
9373 |
+
},
|
9374 |
+
{
|
9375 |
+
"epoch": 0.026852662620650784,
|
9376 |
+
"grad_norm": 0.0014067484298720956,
|
9377 |
+
"learning_rate": 1.1004794640670602e-05,
|
9378 |
+
"loss": 46.0,
|
9379 |
+
"step": 1334
|
9380 |
+
},
|
9381 |
+
{
|
9382 |
+
"epoch": 0.02687279205290015,
|
9383 |
+
"grad_norm": 0.0008450828609056771,
|
9384 |
+
"learning_rate": 1.0913074421295022e-05,
|
9385 |
+
"loss": 46.0,
|
9386 |
+
"step": 1335
|
9387 |
+
},
|
9388 |
+
{
|
9389 |
+
"epoch": 0.02689292148514951,
|
9390 |
+
"grad_norm": 0.0005786415422335267,
|
9391 |
+
"learning_rate": 1.0821715962581302e-05,
|
9392 |
+
"loss": 46.0,
|
9393 |
+
"step": 1336
|
9394 |
+
},
|
9395 |
+
{
|
9396 |
+
"epoch": 0.026913050917398876,
|
9397 |
+
"grad_norm": 0.0012445749016478658,
|
9398 |
+
"learning_rate": 1.0730719635514296e-05,
|
9399 |
+
"loss": 46.0,
|
9400 |
+
"step": 1337
|
9401 |
+
},
|
9402 |
+
{
|
9403 |
+
"epoch": 0.02693318034964824,
|
9404 |
+
"grad_norm": 0.0017225752817466855,
|
9405 |
+
"learning_rate": 1.0640085809608257e-05,
|
9406 |
+
"loss": 46.0,
|
9407 |
+
"step": 1338
|
9408 |
+
},
|
9409 |
+
{
|
9410 |
+
"epoch": 0.0269533097818976,
|
9411 |
+
"grad_norm": 0.0006431869696825743,
|
9412 |
+
"learning_rate": 1.0549814852905427e-05,
|
9413 |
+
"loss": 46.0,
|
9414 |
+
"step": 1339
|
9415 |
+
},
|
9416 |
+
{
|
9417 |
+
"epoch": 0.026973439214146966,
|
9418 |
+
"grad_norm": 0.0006889837677590549,
|
9419 |
+
"learning_rate": 1.0459907131974578e-05,
|
9420 |
+
"loss": 46.0,
|
9421 |
+
"step": 1340
|
9422 |
+
},
|
9423 |
+
{
|
9424 |
+
"epoch": 0.026993568646396328,
|
9425 |
+
"grad_norm": 0.0007062877994030714,
|
9426 |
+
"learning_rate": 1.0370363011909368e-05,
|
9427 |
+
"loss": 46.0,
|
9428 |
+
"step": 1341
|
9429 |
+
},
|
9430 |
+
{
|
9431 |
+
"epoch": 0.027013698078645693,
|
9432 |
+
"grad_norm": 0.0020865912083536386,
|
9433 |
+
"learning_rate": 1.0281182856327075e-05,
|
9434 |
+
"loss": 46.0,
|
9435 |
+
"step": 1342
|
9436 |
+
},
|
9437 |
+
{
|
9438 |
+
"epoch": 0.027033827510895055,
|
9439 |
+
"grad_norm": 0.000669913541059941,
|
9440 |
+
"learning_rate": 1.019236702736689e-05,
|
9441 |
+
"loss": 46.0,
|
9442 |
+
"step": 1343
|
9443 |
+
},
|
9444 |
+
{
|
9445 |
+
"epoch": 0.027053956943144417,
|
9446 |
+
"grad_norm": 0.001416582614183426,
|
9447 |
+
"learning_rate": 1.0103915885688686e-05,
|
9448 |
+
"loss": 46.0,
|
9449 |
+
"step": 1344
|
9450 |
+
},
|
9451 |
+
{
|
9452 |
+
"epoch": 0.027074086375393783,
|
9453 |
+
"grad_norm": 0.0017289246898144484,
|
9454 |
+
"learning_rate": 1.0015829790471288e-05,
|
9455 |
+
"loss": 46.0,
|
9456 |
+
"step": 1345
|
9457 |
+
},
|
9458 |
+
{
|
9459 |
+
"epoch": 0.027094215807643145,
|
9460 |
+
"grad_norm": 0.0016691044438630342,
|
9461 |
+
"learning_rate": 9.928109099411265e-06,
|
9462 |
+
"loss": 46.0,
|
9463 |
+
"step": 1346
|
9464 |
+
},
|
9465 |
+
{
|
9466 |
+
"epoch": 0.02711434523989251,
|
9467 |
+
"grad_norm": 0.0008188973879441619,
|
9468 |
+
"learning_rate": 9.840754168721289e-06,
|
9469 |
+
"loss": 46.0,
|
9470 |
+
"step": 1347
|
9471 |
+
},
|
9472 |
+
{
|
9473 |
+
"epoch": 0.027134474672141872,
|
9474 |
+
"grad_norm": 0.0006958742160350084,
|
9475 |
+
"learning_rate": 9.753765353128863e-06,
|
9476 |
+
"loss": 46.0,
|
9477 |
+
"step": 1348
|
9478 |
+
},
|
9479 |
+
{
|
9480 |
+
"epoch": 0.027154604104391234,
|
9481 |
+
"grad_norm": 0.0026792276185005903,
|
9482 |
+
"learning_rate": 9.667143005874679e-06,
|
9483 |
+
"loss": 46.0,
|
9484 |
+
"step": 1349
|
9485 |
+
},
|
9486 |
+
{
|
9487 |
+
"epoch": 0.0271747335366406,
|
9488 |
+
"grad_norm": 0.0012797150993719697,
|
9489 |
+
"learning_rate": 9.580887478711376e-06,
|
9490 |
+
"loss": 46.0,
|
9491 |
+
"step": 1350
|
9492 |
+
},
|
9493 |
+
{
|
9494 |
+
"epoch": 0.02719486296888996,
|
9495 |
+
"grad_norm": 0.0014746271772310138,
|
9496 |
+
"learning_rate": 9.494999121901948e-06,
|
9497 |
+
"loss": 46.0,
|
9498 |
+
"step": 1351
|
9499 |
+
},
|
9500 |
+
{
|
9501 |
+
"epoch": 0.027214992401139327,
|
9502 |
+
"grad_norm": 0.0014366628602147102,
|
9503 |
+
"learning_rate": 9.409478284218465e-06,
|
9504 |
+
"loss": 46.0,
|
9505 |
+
"step": 1352
|
9506 |
+
},
|
9507 |
+
{
|
9508 |
+
"epoch": 0.02723512183338869,
|
9509 |
+
"grad_norm": 0.0005030794418416917,
|
9510 |
+
"learning_rate": 9.32432531294054e-06,
|
9511 |
+
"loss": 46.0,
|
9512 |
+
"step": 1353
|
9513 |
+
},
|
9514 |
+
{
|
9515 |
+
"epoch": 0.027255251265638054,
|
9516 |
+
"grad_norm": 0.001364423893392086,
|
9517 |
+
"learning_rate": 9.239540553853987e-06,
|
9518 |
+
"loss": 46.0,
|
9519 |
+
"step": 1354
|
9520 |
+
},
|
9521 |
+
{
|
9522 |
+
"epoch": 0.027275380697887416,
|
9523 |
+
"grad_norm": 0.002308462280780077,
|
9524 |
+
"learning_rate": 9.155124351249434e-06,
|
9525 |
+
"loss": 46.0,
|
9526 |
+
"step": 1355
|
9527 |
+
},
|
9528 |
+
{
|
9529 |
+
"epoch": 0.02729551013013678,
|
9530 |
+
"grad_norm": 0.0006031619850546122,
|
9531 |
+
"learning_rate": 9.071077047920807e-06,
|
9532 |
+
"loss": 46.0,
|
9533 |
+
"step": 1356
|
9534 |
+
},
|
9535 |
+
{
|
9536 |
+
"epoch": 0.027315639562386144,
|
9537 |
+
"grad_norm": 0.0008249651291407645,
|
9538 |
+
"learning_rate": 8.987398985164108e-06,
|
9539 |
+
"loss": 46.0,
|
9540 |
+
"step": 1357
|
9541 |
+
},
|
9542 |
+
{
|
9543 |
+
"epoch": 0.027335768994635506,
|
9544 |
+
"grad_norm": 0.0013380757300183177,
|
9545 |
+
"learning_rate": 8.904090502775875e-06,
|
9546 |
+
"loss": 46.0,
|
9547 |
+
"step": 1358
|
9548 |
+
},
|
9549 |
+
{
|
9550 |
+
"epoch": 0.02735589842688487,
|
9551 |
+
"grad_norm": 0.0014677924336865544,
|
9552 |
+
"learning_rate": 8.821151939051953e-06,
|
9553 |
+
"loss": 46.0,
|
9554 |
+
"step": 1359
|
9555 |
+
},
|
9556 |
+
{
|
9557 |
+
"epoch": 0.027376027859134233,
|
9558 |
+
"grad_norm": 0.0010561698582023382,
|
9559 |
+
"learning_rate": 8.73858363078589e-06,
|
9560 |
+
"loss": 46.0,
|
9561 |
+
"step": 1360
|
9562 |
+
},
|
9563 |
+
{
|
9564 |
+
"epoch": 0.027396157291383595,
|
9565 |
+
"grad_norm": 0.0006301281973719597,
|
9566 |
+
"learning_rate": 8.656385913267872e-06,
|
9567 |
+
"loss": 46.0,
|
9568 |
+
"step": 1361
|
9569 |
+
},
|
9570 |
+
{
|
9571 |
+
"epoch": 0.02741628672363296,
|
9572 |
+
"grad_norm": 0.0013005957007408142,
|
9573 |
+
"learning_rate": 8.574559120283099e-06,
|
9574 |
+
"loss": 46.0,
|
9575 |
+
"step": 1362
|
9576 |
+
},
|
9577 |
+
{
|
9578 |
+
"epoch": 0.027436416155882323,
|
9579 |
+
"grad_norm": 0.0025409117806702852,
|
9580 |
+
"learning_rate": 8.493103584110595e-06,
|
9581 |
+
"loss": 46.0,
|
9582 |
+
"step": 1363
|
9583 |
+
},
|
9584 |
+
{
|
9585 |
+
"epoch": 0.027456545588131688,
|
9586 |
+
"grad_norm": 0.0010066272225230932,
|
9587 |
+
"learning_rate": 8.412019635521784e-06,
|
9588 |
+
"loss": 46.0,
|
9589 |
+
"step": 1364
|
9590 |
+
},
|
9591 |
+
{
|
9592 |
+
"epoch": 0.02747667502038105,
|
9593 |
+
"grad_norm": 0.0008767693652771413,
|
9594 |
+
"learning_rate": 8.331307603779137e-06,
|
9595 |
+
"loss": 46.0,
|
9596 |
+
"step": 1365
|
9597 |
+
},
|
9598 |
+
{
|
9599 |
+
"epoch": 0.027496804452630412,
|
9600 |
+
"grad_norm": 0.0005324966041371226,
|
9601 |
+
"learning_rate": 8.250967816634914e-06,
|
9602 |
+
"loss": 46.0,
|
9603 |
+
"step": 1366
|
9604 |
+
},
|
9605 |
+
{
|
9606 |
+
"epoch": 0.027516933884879777,
|
9607 |
+
"grad_norm": 0.0013536482583731413,
|
9608 |
+
"learning_rate": 8.171000600329682e-06,
|
9609 |
+
"loss": 46.0,
|
9610 |
+
"step": 1367
|
9611 |
+
},
|
9612 |
+
{
|
9613 |
+
"epoch": 0.02753706331712914,
|
9614 |
+
"grad_norm": 0.0011457927757874131,
|
9615 |
+
"learning_rate": 8.091406279591207e-06,
|
9616 |
+
"loss": 46.0,
|
9617 |
+
"step": 1368
|
9618 |
+
},
|
9619 |
+
{
|
9620 |
+
"epoch": 0.027557192749378505,
|
9621 |
+
"grad_norm": 0.0013843755004927516,
|
9622 |
+
"learning_rate": 8.012185177632914e-06,
|
9623 |
+
"loss": 46.0,
|
9624 |
+
"step": 1369
|
9625 |
+
},
|
9626 |
+
{
|
9627 |
+
"epoch": 0.027577322181627867,
|
9628 |
+
"grad_norm": 0.0013142600655555725,
|
9629 |
+
"learning_rate": 7.933337616152747e-06,
|
9630 |
+
"loss": 46.0,
|
9631 |
+
"step": 1370
|
9632 |
+
},
|
9633 |
+
{
|
9634 |
+
"epoch": 0.027597451613877232,
|
9635 |
+
"grad_norm": 0.0014753196155652404,
|
9636 |
+
"learning_rate": 7.854863915331745e-06,
|
9637 |
+
"loss": 46.0,
|
9638 |
+
"step": 1371
|
9639 |
+
},
|
9640 |
+
{
|
9641 |
+
"epoch": 0.027617581046126594,
|
9642 |
+
"grad_norm": 0.0005986293544992805,
|
9643 |
+
"learning_rate": 7.776764393832825e-06,
|
9644 |
+
"loss": 46.0,
|
9645 |
+
"step": 1372
|
9646 |
+
},
|
9647 |
+
{
|
9648 |
+
"epoch": 0.027637710478375956,
|
9649 |
+
"grad_norm": 0.001749454764649272,
|
9650 |
+
"learning_rate": 7.69903936879941e-06,
|
9651 |
+
"loss": 46.0,
|
9652 |
+
"step": 1373
|
9653 |
+
},
|
9654 |
+
{
|
9655 |
+
"epoch": 0.02765783991062532,
|
9656 |
+
"grad_norm": 0.0005876508657820523,
|
9657 |
+
"learning_rate": 7.6216891558542395e-06,
|
9658 |
+
"loss": 46.0,
|
9659 |
+
"step": 1374
|
9660 |
+
},
|
9661 |
+
{
|
9662 |
+
"epoch": 0.027677969342874684,
|
9663 |
+
"grad_norm": 0.0010496970498934388,
|
9664 |
+
"learning_rate": 7.54471406909798e-06,
|
9665 |
+
"loss": 46.0,
|
9666 |
+
"step": 1375
|
9667 |
+
},
|
9668 |
+
{
|
9669 |
+
"epoch": 0.02769809877512405,
|
9670 |
+
"grad_norm": 0.0011588014895096421,
|
9671 |
+
"learning_rate": 7.468114421107997e-06,
|
9672 |
+
"loss": 46.0,
|
9673 |
+
"step": 1376
|
9674 |
+
},
|
9675 |
+
{
|
9676 |
+
"epoch": 0.02771822820737341,
|
9677 |
+
"grad_norm": 0.002383692190051079,
|
9678 |
+
"learning_rate": 7.391890522937139e-06,
|
9679 |
+
"loss": 46.0,
|
9680 |
+
"step": 1377
|
9681 |
+
},
|
9682 |
+
{
|
9683 |
+
"epoch": 0.027738357639622773,
|
9684 |
+
"grad_norm": 0.0012233968591317534,
|
9685 |
+
"learning_rate": 7.3160426841123676e-06,
|
9686 |
+
"loss": 46.0,
|
9687 |
+
"step": 1378
|
9688 |
+
},
|
9689 |
+
{
|
9690 |
+
"epoch": 0.02775848707187214,
|
9691 |
+
"grad_norm": 0.0014729154063388705,
|
9692 |
+
"learning_rate": 7.240571212633618e-06,
|
9693 |
+
"loss": 46.0,
|
9694 |
+
"step": 1379
|
9695 |
+
},
|
9696 |
+
{
|
9697 |
+
"epoch": 0.0277786165041215,
|
9698 |
+
"grad_norm": 0.0009423243463970721,
|
9699 |
+
"learning_rate": 7.165476414972416e-06,
|
9700 |
+
"loss": 46.0,
|
9701 |
+
"step": 1380
|
9702 |
+
},
|
9703 |
+
{
|
9704 |
+
"epoch": 0.027798745936370866,
|
9705 |
+
"grad_norm": 0.0012143378844484687,
|
9706 |
+
"learning_rate": 7.090758596070801e-06,
|
9707 |
+
"loss": 46.0,
|
9708 |
+
"step": 1381
|
9709 |
+
},
|
9710 |
+
{
|
9711 |
+
"epoch": 0.027818875368620228,
|
9712 |
+
"grad_norm": 0.0015902521554380655,
|
9713 |
+
"learning_rate": 7.016418059339879e-06,
|
9714 |
+
"loss": 46.0,
|
9715 |
+
"step": 1382
|
9716 |
+
},
|
9717 |
+
{
|
9718 |
+
"epoch": 0.02783900480086959,
|
9719 |
+
"grad_norm": 0.001080973306670785,
|
9720 |
+
"learning_rate": 6.942455106658785e-06,
|
9721 |
+
"loss": 46.0,
|
9722 |
+
"step": 1383
|
9723 |
+
},
|
9724 |
+
{
|
9725 |
+
"epoch": 0.027859134233118955,
|
9726 |
+
"grad_norm": 0.0013885076623409986,
|
9727 |
+
"learning_rate": 6.868870038373332e-06,
|
9728 |
+
"loss": 46.0,
|
9729 |
+
"step": 1384
|
9730 |
+
},
|
9731 |
+
{
|
9732 |
+
"epoch": 0.027879263665368317,
|
9733 |
+
"grad_norm": 0.0010777899296954274,
|
9734 |
+
"learning_rate": 6.795663153294896e-06,
|
9735 |
+
"loss": 46.0,
|
9736 |
+
"step": 1385
|
9737 |
+
},
|
9738 |
+
{
|
9739 |
+
"epoch": 0.027899393097617683,
|
9740 |
+
"grad_norm": 0.0009947263170033693,
|
9741 |
+
"learning_rate": 6.7228347486990365e-06,
|
9742 |
+
"loss": 46.0,
|
9743 |
+
"step": 1386
|
9744 |
+
},
|
9745 |
+
{
|
9746 |
+
"epoch": 0.027919522529867045,
|
9747 |
+
"grad_norm": 0.0010311356745660305,
|
9748 |
+
"learning_rate": 6.6503851203245205e-06,
|
9749 |
+
"loss": 46.0,
|
9750 |
+
"step": 1387
|
9751 |
+
},
|
9752 |
+
{
|
9753 |
+
"epoch": 0.02793965196211641,
|
9754 |
+
"grad_norm": 0.0013068821281194687,
|
9755 |
+
"learning_rate": 6.57831456237189e-06,
|
9756 |
+
"loss": 46.0,
|
9757 |
+
"step": 1388
|
9758 |
+
},
|
9759 |
+
{
|
9760 |
+
"epoch": 0.027959781394365772,
|
9761 |
+
"grad_norm": 0.0010487777180969715,
|
9762 |
+
"learning_rate": 6.506623367502418e-06,
|
9763 |
+
"loss": 46.0,
|
9764 |
+
"step": 1389
|
9765 |
+
},
|
9766 |
+
{
|
9767 |
+
"epoch": 0.027979910826615134,
|
9768 |
+
"grad_norm": 0.0008867370779626071,
|
9769 |
+
"learning_rate": 6.4353118268368986e-06,
|
9770 |
+
"loss": 46.0,
|
9771 |
+
"step": 1390
|
9772 |
+
},
|
9773 |
+
{
|
9774 |
+
"epoch": 0.0280000402588645,
|
9775 |
+
"grad_norm": 0.0019279102561995387,
|
9776 |
+
"learning_rate": 6.3643802299543696e-06,
|
9777 |
+
"loss": 46.0,
|
9778 |
+
"step": 1391
|
9779 |
+
},
|
9780 |
+
{
|
9781 |
+
"epoch": 0.02802016969111386,
|
9782 |
+
"grad_norm": 0.001558710471726954,
|
9783 |
+
"learning_rate": 6.293828864891105e-06,
|
9784 |
+
"loss": 46.0,
|
9785 |
+
"step": 1392
|
9786 |
+
},
|
9787 |
+
{
|
9788 |
+
"epoch": 0.028040299123363227,
|
9789 |
+
"grad_norm": 0.0016826376086100936,
|
9790 |
+
"learning_rate": 6.223658018139245e-06,
|
9791 |
+
"loss": 46.0,
|
9792 |
+
"step": 1393
|
9793 |
+
},
|
9794 |
+
{
|
9795 |
+
"epoch": 0.02806042855561259,
|
9796 |
+
"grad_norm": 0.0012350878678262234,
|
9797 |
+
"learning_rate": 6.153867974645833e-06,
|
9798 |
+
"loss": 46.0,
|
9799 |
+
"step": 1394
|
9800 |
+
},
|
9801 |
+
{
|
9802 |
+
"epoch": 0.02808055798786195,
|
9803 |
+
"grad_norm": 0.0007092293817549944,
|
9804 |
+
"learning_rate": 6.084459017811473e-06,
|
9805 |
+
"loss": 46.0,
|
9806 |
+
"step": 1395
|
9807 |
+
},
|
9808 |
+
{
|
9809 |
+
"epoch": 0.028100687420111316,
|
9810 |
+
"grad_norm": 0.0008784400415606797,
|
9811 |
+
"learning_rate": 6.015431429489371e-06,
|
9812 |
+
"loss": 46.0,
|
9813 |
+
"step": 1396
|
9814 |
+
},
|
9815 |
+
{
|
9816 |
+
"epoch": 0.02812081685236068,
|
9817 |
+
"grad_norm": 0.0008223768672905862,
|
9818 |
+
"learning_rate": 5.946785489983941e-06,
|
9819 |
+
"loss": 46.0,
|
9820 |
+
"step": 1397
|
9821 |
+
},
|
9822 |
+
{
|
9823 |
+
"epoch": 0.028140946284610044,
|
9824 |
+
"grad_norm": 0.0015569966053590178,
|
9825 |
+
"learning_rate": 5.87852147804997e-06,
|
9826 |
+
"loss": 46.0,
|
9827 |
+
"step": 1398
|
9828 |
+
},
|
9829 |
+
{
|
9830 |
+
"epoch": 0.028161075716859406,
|
9831 |
+
"grad_norm": 0.002495410619303584,
|
9832 |
+
"learning_rate": 5.810639670891216e-06,
|
9833 |
+
"loss": 46.0,
|
9834 |
+
"step": 1399
|
9835 |
+
},
|
9836 |
+
{
|
9837 |
+
"epoch": 0.028181205149108768,
|
9838 |
+
"grad_norm": 0.0011495595099404454,
|
9839 |
+
"learning_rate": 5.743140344159459e-06,
|
9840 |
+
"loss": 46.0,
|
9841 |
+
"step": 1400
|
9842 |
+
},
|
9843 |
+
{
|
9844 |
+
"epoch": 0.028201334581358133,
|
9845 |
+
"grad_norm": 0.0013125301338732243,
|
9846 |
+
"learning_rate": 5.676023771953265e-06,
|
9847 |
+
"loss": 46.0,
|
9848 |
+
"step": 1401
|
9849 |
+
},
|
9850 |
+
{
|
9851 |
+
"epoch": 0.028221464013607495,
|
9852 |
+
"grad_norm": 0.0009252792806364596,
|
9853 |
+
"learning_rate": 5.6092902268169986e-06,
|
9854 |
+
"loss": 46.0,
|
9855 |
+
"step": 1402
|
9856 |
+
},
|
9857 |
+
{
|
9858 |
+
"epoch": 0.02824159344585686,
|
9859 |
+
"grad_norm": 0.0010239135008305311,
|
9860 |
+
"learning_rate": 5.542939979739559e-06,
|
9861 |
+
"loss": 46.0,
|
9862 |
+
"step": 1403
|
9863 |
+
},
|
9864 |
+
{
|
9865 |
+
"epoch": 0.028261722878106223,
|
9866 |
+
"grad_norm": 0.0005345203098841012,
|
9867 |
+
"learning_rate": 5.47697330015341e-06,
|
9868 |
+
"loss": 46.0,
|
9869 |
+
"step": 1404
|
9870 |
+
},
|
9871 |
+
{
|
9872 |
+
"epoch": 0.028281852310355588,
|
9873 |
+
"grad_norm": 0.0007038481417112052,
|
9874 |
+
"learning_rate": 5.411390455933463e-06,
|
9875 |
+
"loss": 46.0,
|
9876 |
+
"step": 1405
|
9877 |
+
},
|
9878 |
+
{
|
9879 |
+
"epoch": 0.02830198174260495,
|
9880 |
+
"grad_norm": 0.0007712701335549355,
|
9881 |
+
"learning_rate": 5.346191713395888e-06,
|
9882 |
+
"loss": 46.0,
|
9883 |
+
"step": 1406
|
9884 |
+
},
|
9885 |
+
{
|
9886 |
+
"epoch": 0.028322111174854312,
|
9887 |
+
"grad_norm": 0.0020393729209899902,
|
9888 |
+
"learning_rate": 5.2813773372971995e-06,
|
9889 |
+
"loss": 46.0,
|
9890 |
+
"step": 1407
|
9891 |
+
},
|
9892 |
+
{
|
9893 |
+
"epoch": 0.028342240607103678,
|
9894 |
+
"grad_norm": 0.0008605642360635102,
|
9895 |
+
"learning_rate": 5.216947590833032e-06,
|
9896 |
+
"loss": 46.0,
|
9897 |
+
"step": 1408
|
9898 |
+
},
|
9899 |
+
{
|
9900 |
+
"epoch": 0.02836237003935304,
|
9901 |
+
"grad_norm": 0.0014623169554397464,
|
9902 |
+
"learning_rate": 5.152902735637166e-06,
|
9903 |
+
"loss": 46.0,
|
9904 |
+
"step": 1409
|
9905 |
+
},
|
9906 |
+
{
|
9907 |
+
"epoch": 0.028382499471602405,
|
9908 |
+
"grad_norm": 0.0009752597543410957,
|
9909 |
+
"learning_rate": 5.089243031780389e-06,
|
9910 |
+
"loss": 46.0,
|
9911 |
+
"step": 1410
|
9912 |
+
},
|
9913 |
+
{
|
9914 |
+
"epoch": 0.028402628903851767,
|
9915 |
+
"grad_norm": 0.0009389633196406066,
|
9916 |
+
"learning_rate": 5.025968737769548e-06,
|
9917 |
+
"loss": 46.0,
|
9918 |
+
"step": 1411
|
9919 |
+
},
|
9920 |
+
{
|
9921 |
+
"epoch": 0.02842275833610113,
|
9922 |
+
"grad_norm": 0.0007833832642063498,
|
9923 |
+
"learning_rate": 4.963080110546336e-06,
|
9924 |
+
"loss": 46.0,
|
9925 |
+
"step": 1412
|
9926 |
+
},
|
9927 |
+
{
|
9928 |
+
"epoch": 0.028442887768350494,
|
9929 |
+
"grad_norm": 0.0010727515909820795,
|
9930 |
+
"learning_rate": 4.90057740548645e-06,
|
9931 |
+
"loss": 46.0,
|
9932 |
+
"step": 1413
|
9933 |
+
},
|
9934 |
+
{
|
9935 |
+
"epoch": 0.028463017200599856,
|
9936 |
+
"grad_norm": 0.0007540370570495725,
|
9937 |
+
"learning_rate": 4.838460876398365e-06,
|
9938 |
+
"loss": 46.0,
|
9939 |
+
"step": 1414
|
9940 |
+
},
|
9941 |
+
{
|
9942 |
+
"epoch": 0.028483146632849222,
|
9943 |
+
"grad_norm": 0.0013027754612267017,
|
9944 |
+
"learning_rate": 4.776730775522464e-06,
|
9945 |
+
"loss": 46.0,
|
9946 |
+
"step": 1415
|
9947 |
+
},
|
9948 |
+
{
|
9949 |
+
"epoch": 0.028503276065098584,
|
9950 |
+
"grad_norm": 0.0007448008400388062,
|
9951 |
+
"learning_rate": 4.715387353529855e-06,
|
9952 |
+
"loss": 46.0,
|
9953 |
+
"step": 1416
|
9954 |
+
},
|
9955 |
+
{
|
9956 |
+
"epoch": 0.028523405497347946,
|
9957 |
+
"grad_norm": 0.0008857371867634356,
|
9958 |
+
"learning_rate": 4.654430859521519e-06,
|
9959 |
+
"loss": 46.0,
|
9960 |
+
"step": 1417
|
9961 |
+
},
|
9962 |
+
{
|
9963 |
+
"epoch": 0.02854353492959731,
|
9964 |
+
"grad_norm": 0.002097605960443616,
|
9965 |
+
"learning_rate": 4.593861541027155e-06,
|
9966 |
+
"loss": 46.0,
|
9967 |
+
"step": 1418
|
9968 |
+
},
|
9969 |
+
{
|
9970 |
+
"epoch": 0.028563664361846673,
|
9971 |
+
"grad_norm": 0.0014931777259334922,
|
9972 |
+
"learning_rate": 4.53367964400423e-06,
|
9973 |
+
"loss": 46.0,
|
9974 |
+
"step": 1419
|
9975 |
+
},
|
9976 |
+
{
|
9977 |
+
"epoch": 0.02858379379409604,
|
9978 |
+
"grad_norm": 0.001164833316579461,
|
9979 |
+
"learning_rate": 4.473885412837065e-06,
|
9980 |
+
"loss": 46.0,
|
9981 |
+
"step": 1420
|
9982 |
+
},
|
9983 |
+
{
|
9984 |
+
"epoch": 0.0286039232263454,
|
9985 |
+
"grad_norm": 0.0007305769831873477,
|
9986 |
+
"learning_rate": 4.414479090335644e-06,
|
9987 |
+
"loss": 46.0,
|
9988 |
+
"step": 1421
|
9989 |
+
},
|
9990 |
+
{
|
9991 |
+
"epoch": 0.028624052658594766,
|
9992 |
+
"grad_norm": 0.0013536449987441301,
|
9993 |
+
"learning_rate": 4.355460917734866e-06,
|
9994 |
+
"loss": 46.0,
|
9995 |
+
"step": 1422
|
9996 |
+
},
|
9997 |
+
{
|
9998 |
+
"epoch": 0.028644182090844128,
|
9999 |
+
"grad_norm": 0.0007573326583951712,
|
10000 |
+
"learning_rate": 4.296831134693358e-06,
|
10001 |
+
"loss": 46.0,
|
10002 |
+
"step": 1423
|
10003 |
+
},
|
10004 |
+
{
|
10005 |
+
"epoch": 0.02866431152309349,
|
10006 |
+
"grad_norm": 0.0009904114995151758,
|
10007 |
+
"learning_rate": 4.238589979292651e-06,
|
10008 |
+
"loss": 46.0,
|
10009 |
+
"step": 1424
|
10010 |
+
},
|
10011 |
+
{
|
10012 |
+
"epoch": 0.028684440955342855,
|
10013 |
+
"grad_norm": 0.0010607121512293816,
|
10014 |
+
"learning_rate": 4.180737688036096e-06,
|
10015 |
+
"loss": 46.0,
|
10016 |
+
"step": 1425
|
10017 |
+
},
|
10018 |
+
{
|
10019 |
+
"epoch": 0.028704570387592217,
|
10020 |
+
"grad_norm": 0.001276315306313336,
|
10021 |
+
"learning_rate": 4.1232744958479955e-06,
|
10022 |
+
"loss": 46.0,
|
10023 |
+
"step": 1426
|
10024 |
+
},
|
10025 |
+
{
|
10026 |
+
"epoch": 0.028724699819841583,
|
10027 |
+
"grad_norm": 0.0012834541266784072,
|
10028 |
+
"learning_rate": 4.066200636072604e-06,
|
10029 |
+
"loss": 46.0,
|
10030 |
+
"step": 1427
|
10031 |
+
},
|
10032 |
+
{
|
10033 |
+
"epoch": 0.028744829252090945,
|
10034 |
+
"grad_norm": 0.001088009332306683,
|
10035 |
+
"learning_rate": 4.0095163404732075e-06,
|
10036 |
+
"loss": 46.0,
|
10037 |
+
"step": 1428
|
10038 |
+
},
|
10039 |
+
{
|
10040 |
+
"epoch": 0.028764958684340307,
|
10041 |
+
"grad_norm": 0.0008485732250846922,
|
10042 |
+
"learning_rate": 3.953221839231125e-06,
|
10043 |
+
"loss": 46.0,
|
10044 |
+
"step": 1429
|
10045 |
+
},
|
10046 |
+
{
|
10047 |
+
"epoch": 0.028785088116589672,
|
10048 |
+
"grad_norm": 0.00294103124178946,
|
10049 |
+
"learning_rate": 3.897317360944874e-06,
|
10050 |
+
"loss": 46.0,
|
10051 |
+
"step": 1430
|
10052 |
+
},
|
10053 |
+
{
|
10054 |
+
"epoch": 0.028805217548839034,
|
10055 |
+
"grad_norm": 0.0009056427516043186,
|
10056 |
+
"learning_rate": 3.841803132629107e-06,
|
10057 |
+
"loss": 46.0,
|
10058 |
+
"step": 1431
|
10059 |
+
},
|
10060 |
+
{
|
10061 |
+
"epoch": 0.0288253469810884,
|
10062 |
+
"grad_norm": 0.001108710654079914,
|
10063 |
+
"learning_rate": 3.786679379713842e-06,
|
10064 |
+
"loss": 46.0,
|
10065 |
+
"step": 1432
|
10066 |
+
},
|
10067 |
+
{
|
10068 |
+
"epoch": 0.02884547641333776,
|
10069 |
+
"grad_norm": 0.0013181203976273537,
|
10070 |
+
"learning_rate": 3.731946326043423e-06,
|
10071 |
+
"loss": 46.0,
|
10072 |
+
"step": 1433
|
10073 |
+
},
|
10074 |
+
{
|
10075 |
+
"epoch": 0.028865605845587124,
|
10076 |
+
"grad_norm": 0.0015622148057445884,
|
10077 |
+
"learning_rate": 3.677604193875639e-06,
|
10078 |
+
"loss": 46.0,
|
10079 |
+
"step": 1434
|
10080 |
+
},
|
10081 |
+
{
|
10082 |
+
"epoch": 0.02888573527783649,
|
10083 |
+
"grad_norm": 0.001430216245353222,
|
10084 |
+
"learning_rate": 3.6236532038809167e-06,
|
10085 |
+
"loss": 46.0,
|
10086 |
+
"step": 1435
|
10087 |
+
},
|
10088 |
+
{
|
10089 |
+
"epoch": 0.02890586471008585,
|
10090 |
+
"grad_norm": 0.0011702035553753376,
|
10091 |
+
"learning_rate": 3.5700935751412644e-06,
|
10092 |
+
"loss": 46.0,
|
10093 |
+
"step": 1436
|
10094 |
+
},
|
10095 |
+
{
|
10096 |
+
"epoch": 0.028925994142335217,
|
10097 |
+
"grad_norm": 0.0011749324621632695,
|
10098 |
+
"learning_rate": 3.5169255251495283e-06,
|
10099 |
+
"loss": 46.0,
|
10100 |
+
"step": 1437
|
10101 |
+
},
|
10102 |
+
{
|
10103 |
+
"epoch": 0.02894612357458458,
|
10104 |
+
"grad_norm": 0.0011062580160796642,
|
10105 |
+
"learning_rate": 3.464149269808392e-06,
|
10106 |
+
"loss": 46.0,
|
10107 |
+
"step": 1438
|
10108 |
+
},
|
10109 |
+
{
|
10110 |
+
"epoch": 0.028966253006833944,
|
10111 |
+
"grad_norm": 0.001277309376746416,
|
10112 |
+
"learning_rate": 3.411765023429625e-06,
|
10113 |
+
"loss": 46.0,
|
10114 |
+
"step": 1439
|
10115 |
+
},
|
10116 |
+
{
|
10117 |
+
"epoch": 0.028986382439083306,
|
10118 |
+
"grad_norm": 0.002218514448031783,
|
10119 |
+
"learning_rate": 3.3597729987330796e-06,
|
10120 |
+
"loss": 46.0,
|
10121 |
+
"step": 1440
|
10122 |
+
},
|
10123 |
+
{
|
10124 |
+
"epoch": 0.029006511871332668,
|
10125 |
+
"grad_norm": 0.0007218050304800272,
|
10126 |
+
"learning_rate": 3.3081734068459045e-06,
|
10127 |
+
"loss": 46.0,
|
10128 |
+
"step": 1441
|
10129 |
+
},
|
10130 |
+
{
|
10131 |
+
"epoch": 0.029026641303582033,
|
10132 |
+
"grad_norm": 0.0006786247249692678,
|
10133 |
+
"learning_rate": 3.25696645730168e-06,
|
10134 |
+
"loss": 46.0,
|
10135 |
+
"step": 1442
|
10136 |
+
},
|
10137 |
+
{
|
10138 |
+
"epoch": 0.029046770735831395,
|
10139 |
+
"grad_norm": 0.0008958657272160053,
|
10140 |
+
"learning_rate": 3.2061523580395824e-06,
|
10141 |
+
"loss": 46.0,
|
10142 |
+
"step": 1443
|
10143 |
+
},
|
10144 |
+
{
|
10145 |
+
"epoch": 0.02906690016808076,
|
10146 |
+
"grad_norm": 0.0011994513915851712,
|
10147 |
+
"learning_rate": 3.155731315403465e-06,
|
10148 |
+
"loss": 46.0,
|
10149 |
+
"step": 1444
|
10150 |
+
},
|
10151 |
+
{
|
10152 |
+
"epoch": 0.029087029600330123,
|
10153 |
+
"grad_norm": 0.001030144514515996,
|
10154 |
+
"learning_rate": 3.1057035341411357e-06,
|
10155 |
+
"loss": 46.0,
|
10156 |
+
"step": 1445
|
10157 |
+
},
|
10158 |
+
{
|
10159 |
+
"epoch": 0.029107159032579485,
|
10160 |
+
"grad_norm": 0.0007214623037725687,
|
10161 |
+
"learning_rate": 3.056069217403401e-06,
|
10162 |
+
"loss": 46.0,
|
10163 |
+
"step": 1446
|
10164 |
+
},
|
10165 |
+
{
|
10166 |
+
"epoch": 0.02912728846482885,
|
10167 |
+
"grad_norm": 0.001022842712700367,
|
10168 |
+
"learning_rate": 3.006828566743358e-06,
|
10169 |
+
"loss": 46.0,
|
10170 |
+
"step": 1447
|
10171 |
+
},
|
10172 |
+
{
|
10173 |
+
"epoch": 0.029147417897078212,
|
10174 |
+
"grad_norm": 0.0007378848385997117,
|
10175 |
+
"learning_rate": 2.957981782115471e-06,
|
10176 |
+
"loss": 46.0,
|
10177 |
+
"step": 1448
|
10178 |
+
},
|
10179 |
+
{
|
10180 |
+
"epoch": 0.029167547329327578,
|
10181 |
+
"grad_norm": 0.001484018168412149,
|
10182 |
+
"learning_rate": 2.909529061874816e-06,
|
10183 |
+
"loss": 46.0,
|
10184 |
+
"step": 1449
|
10185 |
+
},
|
10186 |
+
{
|
10187 |
+
"epoch": 0.02918767676157694,
|
10188 |
+
"grad_norm": 0.0008505574078299105,
|
10189 |
+
"learning_rate": 2.861470602776317e-06,
|
10190 |
+
"loss": 46.0,
|
10191 |
+
"step": 1450
|
10192 |
+
},
|
10193 |
+
{
|
10194 |
+
"epoch": 0.0292078061938263,
|
10195 |
+
"grad_norm": 0.0009992974810302258,
|
10196 |
+
"learning_rate": 2.8138065999738337e-06,
|
10197 |
+
"loss": 46.0,
|
10198 |
+
"step": 1451
|
10199 |
+
},
|
10200 |
+
{
|
10201 |
+
"epoch": 0.029227935626075667,
|
10202 |
+
"grad_norm": 0.000925807689782232,
|
10203 |
+
"learning_rate": 2.766537247019485e-06,
|
10204 |
+
"loss": 46.0,
|
10205 |
+
"step": 1452
|
10206 |
+
},
|
10207 |
+
{
|
10208 |
+
"epoch": 0.02924806505832503,
|
10209 |
+
"grad_norm": 0.001936123939231038,
|
10210 |
+
"learning_rate": 2.7196627358627713e-06,
|
10211 |
+
"loss": 46.0,
|
10212 |
+
"step": 1453
|
10213 |
+
},
|
10214 |
+
{
|
10215 |
+
"epoch": 0.029268194490574394,
|
10216 |
+
"grad_norm": 0.0008978885016404092,
|
10217 |
+
"learning_rate": 2.673183256849876e-06,
|
10218 |
+
"loss": 46.0,
|
10219 |
+
"step": 1454
|
10220 |
+
},
|
10221 |
+
{
|
10222 |
+
"epoch": 0.029288323922823756,
|
10223 |
+
"grad_norm": 0.001972075318917632,
|
10224 |
+
"learning_rate": 2.627098998722799e-06,
|
10225 |
+
"loss": 46.0,
|
10226 |
+
"step": 1455
|
10227 |
+
},
|
10228 |
+
{
|
10229 |
+
"epoch": 0.029308453355073122,
|
10230 |
+
"grad_norm": 0.001669483259320259,
|
10231 |
+
"learning_rate": 2.58141014861869e-06,
|
10232 |
+
"loss": 46.0,
|
10233 |
+
"step": 1456
|
10234 |
+
},
|
10235 |
+
{
|
10236 |
+
"epoch": 0.029328582787322484,
|
10237 |
+
"grad_norm": 0.0009523354819975793,
|
10238 |
+
"learning_rate": 2.536116892069007e-06,
|
10239 |
+
"loss": 46.0,
|
10240 |
+
"step": 1457
|
10241 |
+
},
|
10242 |
+
{
|
10243 |
+
"epoch": 0.029348712219571846,
|
10244 |
+
"grad_norm": 0.0025416603311896324,
|
10245 |
+
"learning_rate": 2.4912194129988353e-06,
|
10246 |
+
"loss": 46.0,
|
10247 |
+
"step": 1458
|
10248 |
+
},
|
10249 |
+
{
|
10250 |
+
"epoch": 0.02936884165182121,
|
10251 |
+
"grad_norm": 0.0017256122082471848,
|
10252 |
+
"learning_rate": 2.4467178937260692e-06,
|
10253 |
+
"loss": 46.0,
|
10254 |
+
"step": 1459
|
10255 |
+
},
|
10256 |
+
{
|
10257 |
+
"epoch": 0.029388971084070573,
|
10258 |
+
"grad_norm": 0.0008214665576815605,
|
10259 |
+
"learning_rate": 2.4026125149607225e-06,
|
10260 |
+
"loss": 46.0,
|
10261 |
+
"step": 1460
|
10262 |
+
},
|
10263 |
+
{
|
10264 |
+
"epoch": 0.02940910051631994,
|
10265 |
+
"grad_norm": 0.0010338842403143644,
|
10266 |
+
"learning_rate": 2.3589034558041624e-06,
|
10267 |
+
"loss": 46.0,
|
10268 |
+
"step": 1461
|
10269 |
+
},
|
10270 |
+
{
|
10271 |
+
"epoch": 0.0294292299485693,
|
10272 |
+
"grad_norm": 0.001692043850198388,
|
10273 |
+
"learning_rate": 2.3155908937484093e-06,
|
10274 |
+
"loss": 46.0,
|
10275 |
+
"step": 1462
|
10276 |
+
},
|
10277 |
+
{
|
10278 |
+
"epoch": 0.029449359380818663,
|
10279 |
+
"grad_norm": 0.0017285742796957493,
|
10280 |
+
"learning_rate": 2.2726750046754175e-06,
|
10281 |
+
"loss": 46.0,
|
10282 |
+
"step": 1463
|
10283 |
+
},
|
10284 |
+
{
|
10285 |
+
"epoch": 0.029469488813068028,
|
10286 |
+
"grad_norm": 0.0008371649892069399,
|
10287 |
+
"learning_rate": 2.2301559628563062e-06,
|
10288 |
+
"loss": 46.0,
|
10289 |
+
"step": 1464
|
10290 |
+
},
|
10291 |
+
{
|
10292 |
+
"epoch": 0.02948961824531739,
|
10293 |
+
"grad_norm": 0.0009613548754714429,
|
10294 |
+
"learning_rate": 2.1880339409507288e-06,
|
10295 |
+
"loss": 46.0,
|
10296 |
+
"step": 1465
|
10297 |
+
},
|
10298 |
+
{
|
10299 |
+
"epoch": 0.029509747677566756,
|
10300 |
+
"grad_norm": 0.0014171084621921182,
|
10301 |
+
"learning_rate": 2.146309110006128e-06,
|
10302 |
+
"loss": 46.0,
|
10303 |
+
"step": 1466
|
10304 |
+
},
|
10305 |
+
{
|
10306 |
+
"epoch": 0.029529877109816118,
|
10307 |
+
"grad_norm": 0.0011761389905586839,
|
10308 |
+
"learning_rate": 2.1049816394570486e-06,
|
10309 |
+
"loss": 46.0,
|
10310 |
+
"step": 1467
|
10311 |
+
},
|
10312 |
+
{
|
10313 |
+
"epoch": 0.02955000654206548,
|
10314 |
+
"grad_norm": 0.0011704186908900738,
|
10315 |
+
"learning_rate": 2.064051697124425e-06,
|
10316 |
+
"loss": 46.0,
|
10317 |
+
"step": 1468
|
10318 |
+
},
|
10319 |
+
{
|
10320 |
+
"epoch": 0.029570135974314845,
|
10321 |
+
"grad_norm": 0.0009983095806092024,
|
10322 |
+
"learning_rate": 2.0235194492149832e-06,
|
10323 |
+
"loss": 46.0,
|
10324 |
+
"step": 1469
|
10325 |
+
},
|
10326 |
+
{
|
10327 |
+
"epoch": 0.029590265406564207,
|
10328 |
+
"grad_norm": 0.0015483457827940583,
|
10329 |
+
"learning_rate": 1.983385060320453e-06,
|
10330 |
+
"loss": 46.0,
|
10331 |
+
"step": 1470
|
10332 |
+
},
|
10333 |
+
{
|
10334 |
+
"epoch": 0.029610394838813572,
|
10335 |
+
"grad_norm": 0.0015588031383231282,
|
10336 |
+
"learning_rate": 1.943648693416966e-06,
|
10337 |
+
"loss": 46.0,
|
10338 |
+
"step": 1471
|
10339 |
+
},
|
10340 |
+
{
|
10341 |
+
"epoch": 0.029630524271062934,
|
10342 |
+
"grad_norm": 0.00174389174208045,
|
10343 |
+
"learning_rate": 1.9043105098643931e-06,
|
10344 |
+
"loss": 46.0,
|
10345 |
+
"step": 1472
|
10346 |
+
},
|
10347 |
+
{
|
10348 |
+
"epoch": 0.0296506537033123,
|
10349 |
+
"grad_norm": 0.001446812180802226,
|
10350 |
+
"learning_rate": 1.865370669405675e-06,
|
10351 |
+
"loss": 46.0,
|
10352 |
+
"step": 1473
|
10353 |
+
},
|
10354 |
+
{
|
10355 |
+
"epoch": 0.029670783135561662,
|
10356 |
+
"grad_norm": 0.000889250251930207,
|
10357 |
+
"learning_rate": 1.82682933016618e-06,
|
10358 |
+
"loss": 46.0,
|
10359 |
+
"step": 1474
|
10360 |
+
},
|
10361 |
+
{
|
10362 |
+
"epoch": 0.029690912567811024,
|
10363 |
+
"grad_norm": 0.0005816498887725174,
|
10364 |
+
"learning_rate": 1.7886866486530374e-06,
|
10365 |
+
"loss": 46.0,
|
10366 |
+
"step": 1475
|
10367 |
+
},
|
10368 |
+
{
|
10369 |
+
"epoch": 0.02971104200006039,
|
10370 |
+
"grad_norm": 0.0008788988925516605,
|
10371 |
+
"learning_rate": 1.7509427797545718e-06,
|
10372 |
+
"loss": 46.0,
|
10373 |
+
"step": 1476
|
10374 |
+
},
|
10375 |
+
{
|
10376 |
+
"epoch": 0.02973117143230975,
|
10377 |
+
"grad_norm": 0.0010405541397631168,
|
10378 |
+
"learning_rate": 1.7135978767395588e-06,
|
10379 |
+
"loss": 46.0,
|
10380 |
+
"step": 1477
|
10381 |
+
},
|
10382 |
+
{
|
10383 |
+
"epoch": 0.029751300864559117,
|
10384 |
+
"grad_norm": 0.0011400578077882528,
|
10385 |
+
"learning_rate": 1.676652091256714e-06,
|
10386 |
+
"loss": 46.0,
|
10387 |
+
"step": 1478
|
10388 |
+
},
|
10389 |
+
{
|
10390 |
+
"epoch": 0.02977143029680848,
|
10391 |
+
"grad_norm": 0.0015322790713980794,
|
10392 |
+
"learning_rate": 1.6401055733340164e-06,
|
10393 |
+
"loss": 46.0,
|
10394 |
+
"step": 1479
|
10395 |
+
},
|
10396 |
+
{
|
10397 |
+
"epoch": 0.02979155972905784,
|
10398 |
+
"grad_norm": 0.0007389390957541764,
|
10399 |
+
"learning_rate": 1.6039584713781308e-06,
|
10400 |
+
"loss": 46.0,
|
10401 |
+
"step": 1480
|
10402 |
+
},
|
10403 |
+
{
|
10404 |
+
"epoch": 0.029811689161307206,
|
10405 |
+
"grad_norm": 0.0011101323179900646,
|
10406 |
+
"learning_rate": 1.5682109321737637e-06,
|
10407 |
+
"loss": 46.0,
|
10408 |
+
"step": 1481
|
10409 |
+
},
|
10410 |
+
{
|
10411 |
+
"epoch": 0.029831818593556568,
|
10412 |
+
"grad_norm": 0.0012494990369305015,
|
10413 |
+
"learning_rate": 1.5328631008831197e-06,
|
10414 |
+
"loss": 46.0,
|
10415 |
+
"step": 1482
|
10416 |
+
},
|
10417 |
+
{
|
10418 |
+
"epoch": 0.029851948025805933,
|
10419 |
+
"grad_norm": 0.0019527755212038755,
|
10420 |
+
"learning_rate": 1.497915121045268e-06,
|
10421 |
+
"loss": 46.0,
|
10422 |
+
"step": 1483
|
10423 |
+
},
|
10424 |
+
{
|
10425 |
+
"epoch": 0.029872077458055295,
|
10426 |
+
"grad_norm": 0.0015624084044247866,
|
10427 |
+
"learning_rate": 1.4633671345755884e-06,
|
10428 |
+
"loss": 46.0,
|
10429 |
+
"step": 1484
|
10430 |
+
},
|
10431 |
+
{
|
10432 |
+
"epoch": 0.029892206890304657,
|
10433 |
+
"grad_norm": 0.0013834653655067086,
|
10434 |
+
"learning_rate": 1.4292192817651706e-06,
|
10435 |
+
"loss": 46.0,
|
10436 |
+
"step": 1485
|
10437 |
+
},
|
10438 |
+
{
|
10439 |
+
"epoch": 0.029912336322554023,
|
10440 |
+
"grad_norm": 0.0015040615107864141,
|
10441 |
+
"learning_rate": 1.3954717012802599e-06,
|
10442 |
+
"loss": 46.0,
|
10443 |
+
"step": 1486
|
10444 |
+
},
|
10445 |
+
{
|
10446 |
+
"epoch": 0.029932465754803385,
|
10447 |
+
"grad_norm": 0.001884009805507958,
|
10448 |
+
"learning_rate": 1.3621245301617014e-06,
|
10449 |
+
"loss": 46.0,
|
10450 |
+
"step": 1487
|
10451 |
+
},
|
10452 |
+
{
|
10453 |
+
"epoch": 0.02995259518705275,
|
10454 |
+
"grad_norm": 0.0003439519787207246,
|
10455 |
+
"learning_rate": 1.3291779038243745e-06,
|
10456 |
+
"loss": 46.0,
|
10457 |
+
"step": 1488
|
10458 |
+
},
|
10459 |
+
{
|
10460 |
+
"epoch": 0.029972724619302112,
|
10461 |
+
"grad_norm": 0.0009911386296153069,
|
10462 |
+
"learning_rate": 1.2966319560566264e-06,
|
10463 |
+
"loss": 46.0,
|
10464 |
+
"step": 1489
|
10465 |
+
},
|
10466 |
+
{
|
10467 |
+
"epoch": 0.029992854051551478,
|
10468 |
+
"grad_norm": 0.001606732839718461,
|
10469 |
+
"learning_rate": 1.2644868190197501e-06,
|
10470 |
+
"loss": 46.0,
|
10471 |
+
"step": 1490
|
10472 |
+
},
|
10473 |
+
{
|
10474 |
+
"epoch": 0.03001298348380084,
|
10475 |
+
"grad_norm": 0.0009569000103510916,
|
10476 |
+
"learning_rate": 1.2327426232474626e-06,
|
10477 |
+
"loss": 46.0,
|
10478 |
+
"step": 1491
|
10479 |
+
},
|
10480 |
+
{
|
10481 |
+
"epoch": 0.0300331129160502,
|
10482 |
+
"grad_norm": 0.002008062554523349,
|
10483 |
+
"learning_rate": 1.201399497645328e-06,
|
10484 |
+
"loss": 46.0,
|
10485 |
+
"step": 1492
|
10486 |
+
},
|
10487 |
+
{
|
10488 |
+
"epoch": 0.030053242348299567,
|
10489 |
+
"grad_norm": 0.0015580368926748633,
|
10490 |
+
"learning_rate": 1.1704575694902686e-06,
|
10491 |
+
"loss": 46.0,
|
10492 |
+
"step": 1493
|
10493 |
+
},
|
10494 |
+
{
|
10495 |
+
"epoch": 0.03007337178054893,
|
10496 |
+
"grad_norm": 0.0017130931373685598,
|
10497 |
+
"learning_rate": 1.1399169644300323e-06,
|
10498 |
+
"loss": 46.0,
|
10499 |
+
"step": 1494
|
10500 |
+
},
|
10501 |
+
{
|
10502 |
+
"epoch": 0.030093501212798295,
|
10503 |
+
"grad_norm": 0.0015125928912311792,
|
10504 |
+
"learning_rate": 1.1097778064827257e-06,
|
10505 |
+
"loss": 46.0,
|
10506 |
+
"step": 1495
|
10507 |
+
},
|
10508 |
+
{
|
10509 |
+
"epoch": 0.030113630645047657,
|
10510 |
+
"grad_norm": 0.0013810923555865884,
|
10511 |
+
"learning_rate": 1.0800402180362158e-06,
|
10512 |
+
"loss": 46.0,
|
10513 |
+
"step": 1496
|
10514 |
+
},
|
10515 |
+
{
|
10516 |
+
"epoch": 0.03013376007729702,
|
10517 |
+
"grad_norm": 0.0014871679013594985,
|
10518 |
+
"learning_rate": 1.0507043198477617e-06,
|
10519 |
+
"loss": 46.0,
|
10520 |
+
"step": 1497
|
10521 |
+
},
|
10522 |
+
{
|
10523 |
+
"epoch": 0.030153889509546384,
|
10524 |
+
"grad_norm": 0.0008668963564559817,
|
10525 |
+
"learning_rate": 1.0217702310433842e-06,
|
10526 |
+
"loss": 46.0,
|
10527 |
+
"step": 1498
|
10528 |
+
},
|
10529 |
+
{
|
10530 |
+
"epoch": 0.030174018941795746,
|
10531 |
+
"grad_norm": 0.00135446572676301,
|
10532 |
+
"learning_rate": 9.93238069117508e-07,
|
10533 |
+
"loss": 46.0,
|
10534 |
+
"step": 1499
|
10535 |
+
},
|
10536 |
+
{
|
10537 |
+
"epoch": 0.03019414837404511,
|
10538 |
+
"grad_norm": 0.0012777193915098906,
|
10539 |
+
"learning_rate": 9.651079499323978e-07,
|
10540 |
+
"loss": 46.0,
|
10541 |
+
"step": 1500
|
10542 |
+
},
|
10543 |
+
{
|
10544 |
+
"epoch": 0.030214277806294473,
|
10545 |
+
"grad_norm": 0.0014395661419257522,
|
10546 |
+
"learning_rate": 9.373799877177236e-07,
|
10547 |
+
"loss": 46.0,
|
10548 |
+
"step": 1501
|
10549 |
+
},
|
10550 |
+
{
|
10551 |
+
"epoch": 0.030234407238543835,
|
10552 |
+
"grad_norm": 0.0019440649775788188,
|
10553 |
+
"learning_rate": 9.100542950701063e-07,
|
10554 |
+
"loss": 46.0,
|
10555 |
+
"step": 1502
|
10556 |
+
},
|
10557 |
+
{
|
10558 |
+
"epoch": 0.0302545366707932,
|
10559 |
+
"grad_norm": 0.0007341218297369778,
|
10560 |
+
"learning_rate": 8.831309829526291e-07,
|
10561 |
+
"loss": 46.0,
|
10562 |
+
"step": 1503
|
10563 |
+
},
|
10564 |
+
{
|
10565 |
+
"epoch": 0.030274666103042563,
|
10566 |
+
"grad_norm": 0.0010124749969691038,
|
10567 |
+
"learning_rate": 8.566101606944266e-07,
|
10568 |
+
"loss": 46.0,
|
10569 |
+
"step": 1504
|
10570 |
+
},
|
10571 |
+
{
|
10572 |
+
"epoch": 0.030294795535291928,
|
10573 |
+
"grad_norm": 0.0011337018804624677,
|
10574 |
+
"learning_rate": 8.304919359901963e-07,
|
10575 |
+
"loss": 46.0,
|
10576 |
+
"step": 1505
|
10577 |
+
},
|
10578 |
+
{
|
10579 |
+
"epoch": 0.03031492496754129,
|
10580 |
+
"grad_norm": 0.0006525327335111797,
|
10581 |
+
"learning_rate": 8.047764148997883e-07,
|
10582 |
+
"loss": 46.0,
|
10583 |
+
"step": 1506
|
10584 |
+
},
|
10585 |
+
{
|
10586 |
+
"epoch": 0.030335054399790656,
|
10587 |
+
"grad_norm": 0.0014475996140390635,
|
10588 |
+
"learning_rate": 7.794637018477824e-07,
|
10589 |
+
"loss": 46.0,
|
10590 |
+
"step": 1507
|
10591 |
+
},
|
10592 |
+
{
|
10593 |
+
"epoch": 0.030355183832040018,
|
10594 |
+
"grad_norm": 0.0006679428042843938,
|
10595 |
+
"learning_rate": 7.545538996230228e-07,
|
10596 |
+
"loss": 46.0,
|
10597 |
+
"step": 1508
|
10598 |
+
},
|
10599 |
+
{
|
10600 |
+
"epoch": 0.03037531326428938,
|
10601 |
+
"grad_norm": 0.001389230601489544,
|
10602 |
+
"learning_rate": 7.300471093782624e-07,
|
10603 |
+
"loss": 46.0,
|
10604 |
+
"step": 1509
|
10605 |
+
},
|
10606 |
+
{
|
10607 |
+
"epoch": 0.030395442696538745,
|
10608 |
+
"grad_norm": 0.0013642680132761598,
|
10609 |
+
"learning_rate": 7.059434306297075e-07,
|
10610 |
+
"loss": 46.0,
|
10611 |
+
"step": 1510
|
10612 |
+
},
|
10613 |
+
{
|
10614 |
+
"epoch": 0.030415572128788107,
|
10615 |
+
"grad_norm": 0.000978952506557107,
|
10616 |
+
"learning_rate": 6.822429612566184e-07,
|
10617 |
+
"loss": 46.0,
|
10618 |
+
"step": 1511
|
10619 |
+
},
|
10620 |
+
{
|
10621 |
+
"epoch": 0.030435701561037472,
|
10622 |
+
"grad_norm": 0.0009029002394527197,
|
10623 |
+
"learning_rate": 6.589457975009205e-07,
|
10624 |
+
"loss": 46.0,
|
10625 |
+
"step": 1512
|
10626 |
+
},
|
10627 |
+
{
|
10628 |
+
"epoch": 0.030455830993286834,
|
10629 |
+
"grad_norm": 0.002438169904053211,
|
10630 |
+
"learning_rate": 6.360520339668163e-07,
|
10631 |
+
"loss": 46.0,
|
10632 |
+
"step": 1513
|
10633 |
+
},
|
10634 |
+
{
|
10635 |
+
"epoch": 0.030475960425536196,
|
10636 |
+
"grad_norm": 0.0020010732114315033,
|
10637 |
+
"learning_rate": 6.135617636204072e-07,
|
10638 |
+
"loss": 46.0,
|
10639 |
+
"step": 1514
|
10640 |
+
},
|
10641 |
+
{
|
10642 |
+
"epoch": 0.030496089857785562,
|
10643 |
+
"grad_norm": 0.0014994231751188636,
|
10644 |
+
"learning_rate": 5.91475077789272e-07,
|
10645 |
+
"loss": 46.0,
|
10646 |
+
"step": 1515
|
10647 |
+
},
|
10648 |
+
{
|
10649 |
+
"epoch": 0.030516219290034924,
|
10650 |
+
"grad_norm": 0.0007911003194749355,
|
10651 |
+
"learning_rate": 5.697920661621558e-07,
|
10652 |
+
"loss": 46.0,
|
10653 |
+
"step": 1516
|
10654 |
+
},
|
10655 |
+
{
|
10656 |
+
"epoch": 0.03053634872228429,
|
10657 |
+
"grad_norm": 0.0011950345942750573,
|
10658 |
+
"learning_rate": 5.485128167885933e-07,
|
10659 |
+
"loss": 46.0,
|
10660 |
+
"step": 1517
|
10661 |
+
},
|
10662 |
+
{
|
10663 |
+
"epoch": 0.03055647815453365,
|
10664 |
+
"grad_norm": 0.0012883899034932256,
|
10665 |
+
"learning_rate": 5.276374160784858e-07,
|
10666 |
+
"loss": 46.0,
|
10667 |
+
"step": 1518
|
10668 |
+
},
|
10669 |
+
{
|
10670 |
+
"epoch": 0.030576607586783013,
|
10671 |
+
"grad_norm": 0.002167344558984041,
|
10672 |
+
"learning_rate": 5.071659488018688e-07,
|
10673 |
+
"loss": 46.0,
|
10674 |
+
"step": 1519
|
10675 |
+
},
|
10676 |
+
{
|
10677 |
+
"epoch": 0.03059673701903238,
|
10678 |
+
"grad_norm": 0.0008158805430866778,
|
10679 |
+
"learning_rate": 4.870984980884341e-07,
|
10680 |
+
"loss": 46.0,
|
10681 |
+
"step": 1520
|
10682 |
+
},
|
10683 |
+
{
|
10684 |
+
"epoch": 0.03061686645128174,
|
10685 |
+
"grad_norm": 0.00043665210250765085,
|
10686 |
+
"learning_rate": 4.674351454273307e-07,
|
10687 |
+
"loss": 46.0,
|
10688 |
+
"step": 1521
|
10689 |
+
},
|
10690 |
+
{
|
10691 |
+
"epoch": 0.030636995883531106,
|
10692 |
+
"grad_norm": 0.0012343135895207524,
|
10693 |
+
"learning_rate": 4.481759706666755e-07,
|
10694 |
+
"loss": 46.0,
|
10695 |
+
"step": 1522
|
10696 |
+
},
|
10697 |
+
{
|
10698 |
+
"epoch": 0.030657125315780468,
|
10699 |
+
"grad_norm": 0.0007814933778718114,
|
10700 |
+
"learning_rate": 4.2932105201339835e-07,
|
10701 |
+
"loss": 46.0,
|
10702 |
+
"step": 1523
|
10703 |
+
},
|
10704 |
+
{
|
10705 |
+
"epoch": 0.030677254748029834,
|
10706 |
+
"grad_norm": 0.0008475257782265544,
|
10707 |
+
"learning_rate": 4.1087046603279777e-07,
|
10708 |
+
"loss": 46.0,
|
10709 |
+
"step": 1524
|
10710 |
+
},
|
10711 |
+
{
|
10712 |
+
"epoch": 0.030697384180279196,
|
10713 |
+
"grad_norm": 0.0010306923650205135,
|
10714 |
+
"learning_rate": 3.9282428764827463e-07,
|
10715 |
+
"loss": 46.0,
|
10716 |
+
"step": 1525
|
10717 |
+
},
|
10718 |
+
{
|
10719 |
+
"epoch": 0.030717513612528557,
|
10720 |
+
"grad_norm": 0.0033900176640599966,
|
10721 |
+
"learning_rate": 3.751825901410433e-07,
|
10722 |
+
"loss": 46.0,
|
10723 |
+
"step": 1526
|
10724 |
+
},
|
10725 |
+
{
|
10726 |
+
"epoch": 0.030737643044777923,
|
10727 |
+
"grad_norm": 0.0020439354702830315,
|
10728 |
+
"learning_rate": 3.579454451498099e-07,
|
10729 |
+
"loss": 46.0,
|
10730 |
+
"step": 1527
|
10731 |
+
},
|
10732 |
+
{
|
10733 |
+
"epoch": 0.030757772477027285,
|
10734 |
+
"grad_norm": 0.0013795166742056608,
|
10735 |
+
"learning_rate": 3.411129226704945e-07,
|
10736 |
+
"loss": 46.0,
|
10737 |
+
"step": 1528
|
10738 |
+
},
|
10739 |
+
{
|
10740 |
+
"epoch": 0.03077790190927665,
|
10741 |
+
"grad_norm": 0.0019550782162696123,
|
10742 |
+
"learning_rate": 3.246850910559318e-07,
|
10743 |
+
"loss": 46.0,
|
10744 |
+
"step": 1529
|
10745 |
+
},
|
10746 |
+
{
|
10747 |
+
"epoch": 0.030798031341526012,
|
10748 |
+
"grad_norm": 0.001795038697309792,
|
10749 |
+
"learning_rate": 3.0866201701560406e-07,
|
10750 |
+
"loss": 46.0,
|
10751 |
+
"step": 1530
|
10752 |
+
},
|
10753 |
+
{
|
10754 |
+
"epoch": 0.030818160773775374,
|
10755 |
+
"grad_norm": 0.0019376088166609406,
|
10756 |
+
"learning_rate": 2.9304376561539726e-07,
|
10757 |
+
"loss": 46.0,
|
10758 |
+
"step": 1531
|
10759 |
+
},
|
10760 |
+
{
|
10761 |
+
"epoch": 0.03083829020602474,
|
10762 |
+
"grad_norm": 0.0016611182363703847,
|
10763 |
+
"learning_rate": 2.7783040027726804e-07,
|
10764 |
+
"loss": 46.0,
|
10765 |
+
"step": 1532
|
10766 |
+
},
|
10767 |
+
{
|
10768 |
+
"epoch": 0.030858419638274102,
|
10769 |
+
"grad_norm": 0.0016700802370905876,
|
10770 |
+
"learning_rate": 2.630219827790659e-07,
|
10771 |
+
"loss": 46.0,
|
10772 |
+
"step": 1533
|
10773 |
+
},
|
10774 |
+
{
|
10775 |
+
"epoch": 0.030878549070523467,
|
10776 |
+
"grad_norm": 0.0005473219207488,
|
10777 |
+
"learning_rate": 2.4861857325421123e-07,
|
10778 |
+
"loss": 46.0,
|
10779 |
+
"step": 1534
|
10780 |
+
},
|
10781 |
+
{
|
10782 |
+
"epoch": 0.03089867850277283,
|
10783 |
+
"grad_norm": 0.0012385062873363495,
|
10784 |
+
"learning_rate": 2.346202301915068e-07,
|
10785 |
+
"loss": 46.0,
|
10786 |
+
"step": 1535
|
10787 |
+
},
|
10788 |
+
{
|
10789 |
+
"epoch": 0.03091880793502219,
|
10790 |
+
"grad_norm": 0.0020921400282531977,
|
10791 |
+
"learning_rate": 2.2102701043487105e-07,
|
10792 |
+
"loss": 46.0,
|
10793 |
+
"step": 1536
|
10794 |
+
},
|
10795 |
+
{
|
10796 |
+
"epoch": 0.030938937367271557,
|
10797 |
+
"grad_norm": 0.0009741144021973014,
|
10798 |
+
"learning_rate": 2.0783896918310508e-07,
|
10799 |
+
"loss": 46.0,
|
10800 |
+
"step": 1537
|
10801 |
+
},
|
10802 |
+
{
|
10803 |
+
"epoch": 0.03095906679952092,
|
10804 |
+
"grad_norm": 0.0017884114058688283,
|
10805 |
+
"learning_rate": 1.9505615998969274e-07,
|
10806 |
+
"loss": 46.0,
|
10807 |
+
"step": 1538
|
10808 |
+
},
|
10809 |
+
{
|
10810 |
+
"epoch": 0.030979196231770284,
|
10811 |
+
"grad_norm": 0.0012170026311650872,
|
10812 |
+
"learning_rate": 1.8267863476255643e-07,
|
10813 |
+
"loss": 46.0,
|
10814 |
+
"step": 1539
|
10815 |
+
},
|
10816 |
+
{
|
10817 |
+
"epoch": 0.030999325664019646,
|
10818 |
+
"grad_norm": 0.0015093215042725205,
|
10819 |
+
"learning_rate": 1.7070644376386835e-07,
|
10820 |
+
"loss": 46.0,
|
10821 |
+
"step": 1540
|
10822 |
+
},
|
10823 |
+
{
|
10824 |
+
"epoch": 0.03101945509626901,
|
10825 |
+
"grad_norm": 0.0010898308828473091,
|
10826 |
+
"learning_rate": 1.5913963560981738e-07,
|
10827 |
+
"loss": 46.0,
|
10828 |
+
"step": 1541
|
10829 |
+
},
|
10830 |
+
{
|
10831 |
+
"epoch": 0.031039584528518373,
|
10832 |
+
"grad_norm": 0.000764137483201921,
|
10833 |
+
"learning_rate": 1.4797825727044246e-07,
|
10834 |
+
"loss": 46.0,
|
10835 |
+
"step": 1542
|
10836 |
+
},
|
10837 |
+
{
|
10838 |
+
"epoch": 0.031059713960767735,
|
10839 |
+
"grad_norm": 0.0012036709813401103,
|
10840 |
+
"learning_rate": 1.3722235406943285e-07,
|
10841 |
+
"loss": 46.0,
|
10842 |
+
"step": 1543
|
10843 |
+
},
|
10844 |
+
{
|
10845 |
+
"epoch": 0.0310798433930171,
|
10846 |
+
"grad_norm": 0.0012527679791674018,
|
10847 |
+
"learning_rate": 1.2687196968392822e-07,
|
10848 |
+
"loss": 46.0,
|
10849 |
+
"step": 1544
|
10850 |
+
},
|
10851 |
+
{
|
10852 |
+
"epoch": 0.031099972825266463,
|
10853 |
+
"grad_norm": 0.0017057860968634486,
|
10854 |
+
"learning_rate": 1.1692714614436329e-07,
|
10855 |
+
"loss": 46.0,
|
10856 |
+
"step": 1545
|
10857 |
+
},
|
10858 |
+
{
|
10859 |
+
"epoch": 0.03112010225751583,
|
10860 |
+
"grad_norm": 0.0012995371362194419,
|
10861 |
+
"learning_rate": 1.0738792383427898e-07,
|
10862 |
+
"loss": 46.0,
|
10863 |
+
"step": 1546
|
10864 |
+
},
|
10865 |
+
{
|
10866 |
+
"epoch": 0.03114023168976519,
|
10867 |
+
"grad_norm": 0.0005156396073289216,
|
10868 |
+
"learning_rate": 9.82543414901782e-08,
|
10869 |
+
"loss": 46.0,
|
10870 |
+
"step": 1547
|
10871 |
+
},
|
10872 |
+
{
|
10873 |
+
"epoch": 0.031160361122014552,
|
10874 |
+
"grad_norm": 0.0023463049437850714,
|
10875 |
+
"learning_rate": 8.952643620134815e-08,
|
10876 |
+
"loss": 46.0,
|
10877 |
+
"step": 1548
|
10878 |
+
},
|
10879 |
+
{
|
10880 |
+
"epoch": 0.031180490554263918,
|
10881 |
+
"grad_norm": 0.0005802233936265111,
|
10882 |
+
"learning_rate": 8.1204243409716e-08,
|
10883 |
+
"loss": 46.0,
|
10884 |
+
"step": 1549
|
10885 |
+
},
|
10886 |
+
{
|
10887 |
+
"epoch": 0.03120061998651328,
|
10888 |
+
"grad_norm": 0.0014365284005180001,
|
10889 |
+
"learning_rate": 7.328779690972675e-08,
|
10890 |
+
"loss": 46.0,
|
10891 |
+
"step": 1550
|
10892 |
+
},
|
10893 |
+
{
|
10894 |
+
"epoch": 0.031220749418762645,
|
10895 |
+
"grad_norm": 0.0009548702510073781,
|
10896 |
+
"learning_rate": 6.577712884816566e-08,
|
10897 |
+
"loss": 46.0,
|
10898 |
+
"step": 1551
|
10899 |
+
},
|
10900 |
+
{
|
10901 |
+
"epoch": 0.031240878851012007,
|
10902 |
+
"grad_norm": 0.000912398740183562,
|
10903 |
+
"learning_rate": 5.867226972404716e-08,
|
10904 |
+
"loss": 46.0,
|
10905 |
+
"step": 1552
|
10906 |
+
},
|
10907 |
+
{
|
10908 |
+
"epoch": 0.03126100828326137,
|
10909 |
+
"grad_norm": 0.0013671774649992585,
|
10910 |
+
"learning_rate": 5.197324838851492e-08,
|
10911 |
+
"loss": 46.0,
|
10912 |
+
"step": 1553
|
10913 |
+
},
|
10914 |
+
{
|
10915 |
+
"epoch": 0.031281137715510735,
|
10916 |
+
"grad_norm": 0.001228424021974206,
|
10917 |
+
"learning_rate": 4.5680092044686486e-08,
|
10918 |
+
"loss": 46.0,
|
10919 |
+
"step": 1554
|
10920 |
+
},
|
10921 |
+
{
|
10922 |
+
"epoch": 0.031301267147760096,
|
10923 |
+
"grad_norm": 0.0009602979407645762,
|
10924 |
+
"learning_rate": 3.9792826247553315e-08,
|
10925 |
+
"loss": 46.0,
|
10926 |
+
"step": 1555
|
10927 |
+
},
|
10928 |
+
{
|
10929 |
+
"epoch": 0.03132139658000946,
|
10930 |
+
"grad_norm": 0.0012437499826774001,
|
10931 |
+
"learning_rate": 3.431147490390307e-08,
|
10932 |
+
"loss": 46.0,
|
10933 |
+
"step": 1556
|
10934 |
+
},
|
10935 |
+
{
|
10936 |
+
"epoch": 0.03134152601225883,
|
10937 |
+
"grad_norm": 0.0018415412632748485,
|
10938 |
+
"learning_rate": 2.9236060272186395e-08,
|
10939 |
+
"loss": 46.0,
|
10940 |
+
"step": 1557
|
10941 |
+
},
|
10942 |
+
{
|
10943 |
+
"epoch": 0.03136165544450819,
|
10944 |
+
"grad_norm": 0.0008638726430945098,
|
10945 |
+
"learning_rate": 2.4566602962450282e-08,
|
10946 |
+
"loss": 46.0,
|
10947 |
+
"step": 1558
|
10948 |
+
},
|
10949 |
+
{
|
10950 |
+
"epoch": 0.03138178487675755,
|
10951 |
+
"grad_norm": 0.001755962148308754,
|
10952 |
+
"learning_rate": 2.0303121936227077e-08,
|
10953 |
+
"loss": 46.0,
|
10954 |
+
"step": 1559
|
10955 |
+
},
|
10956 |
+
{
|
10957 |
+
"epoch": 0.03140191430900691,
|
10958 |
+
"grad_norm": 0.000949330220464617,
|
10959 |
+
"learning_rate": 1.6445634506512265e-08,
|
10960 |
+
"loss": 46.0,
|
10961 |
+
"step": 1560
|
10962 |
+
},
|
10963 |
+
{
|
10964 |
+
"epoch": 0.031422043741256275,
|
10965 |
+
"grad_norm": 0.000980414217337966,
|
10966 |
+
"learning_rate": 1.2994156337620134e-08,
|
10967 |
+
"loss": 46.0,
|
10968 |
+
"step": 1561
|
10969 |
+
},
|
10970 |
+
{
|
10971 |
+
"epoch": 0.031442173173505644,
|
10972 |
+
"grad_norm": 0.001756619312800467,
|
10973 |
+
"learning_rate": 9.948701445194885e-09,
|
10974 |
+
"loss": 46.0,
|
10975 |
+
"step": 1562
|
10976 |
+
},
|
10977 |
+
{
|
10978 |
+
"epoch": 0.031462302605755006,
|
10979 |
+
"grad_norm": 0.0016652131453156471,
|
10980 |
+
"learning_rate": 7.3092821960774046e-09,
|
10981 |
+
"loss": 46.0,
|
10982 |
+
"step": 1563
|
10983 |
+
},
|
10984 |
+
{
|
10985 |
+
"epoch": 0.03148243203800437,
|
10986 |
+
"grad_norm": 0.0008927117451094091,
|
10987 |
+
"learning_rate": 5.0759093083385665e-09,
|
10988 |
+
"loss": 46.0,
|
10989 |
+
"step": 1564
|
10990 |
+
},
|
10991 |
+
{
|
10992 |
+
"epoch": 0.03150256147025373,
|
10993 |
+
"grad_norm": 0.0012494467664510012,
|
10994 |
+
"learning_rate": 3.24859185114601e-09,
|
10995 |
+
"loss": 46.0,
|
10996 |
+
"step": 1565
|
10997 |
+
},
|
10998 |
+
{
|
10999 |
+
"epoch": 0.03152269090250309,
|
11000 |
+
"grad_norm": 0.0012330285971984267,
|
11001 |
+
"learning_rate": 1.8273372448307512e-09,
|
11002 |
+
"loss": 46.0,
|
11003 |
+
"step": 1566
|
11004 |
+
},
|
11005 |
+
{
|
11006 |
+
"epoch": 0.03154282033475246,
|
11007 |
+
"grad_norm": 0.002677972661331296,
|
11008 |
+
"learning_rate": 8.121512607317528e-10,
|
11009 |
+
"loss": 46.0,
|
11010 |
+
"step": 1567
|
11011 |
+
},
|
11012 |
+
{
|
11013 |
+
"epoch": 0.03156294976700182,
|
11014 |
+
"grad_norm": 0.0020114348735660315,
|
11015 |
+
"learning_rate": 2.0303802130694493e-10,
|
11016 |
+
"loss": 46.0,
|
11017 |
+
"step": 1568
|
11018 |
+
},
|
11019 |
+
{
|
11020 |
+
"epoch": 0.031583079199251185,
|
11021 |
+
"grad_norm": 0.0010916402097791433,
|
11022 |
+
"learning_rate": 0.0,
|
11023 |
+
"loss": 46.0,
|
11024 |
+
"step": 1569
|
11025 |
}
|
11026 |
],
|
11027 |
"logging_steps": 1,
|
|
|
11036 |
"should_evaluate": false,
|
11037 |
"should_log": false,
|
11038 |
"should_save": true,
|
11039 |
+
"should_training_stop": true
|
11040 |
},
|
11041 |
"attributes": {}
|
11042 |
}
|
11043 |
},
|
11044 |
+
"total_flos": 36155043692544.0,
|
11045 |
"train_batch_size": 2,
|
11046 |
"trial_name": null,
|
11047 |
"trial_params": null
|