Training in progress, step 1303, 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 sha256:7d1517bbb68b73118ab298b46d78154dff83c4336e98408645d842041e1e7e41
|
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 sha256:988c28e29b391965d592d06f8e5c0720e58d3b0a4325154940bcc92abedb4644
|
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:f79d0ab40c06f3104eb07d7ccc26e9ca2de4bde54f83f4e379cd88a2f7887023
|
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 sha256:9eabf0c977468cdbb0fb34466e43f850400f356cb7be4266d2d1ac09878380a5
|
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": 326,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -6885,6 +6885,2281 @@
|
|
6885 |
"eval_samples_per_second": 165.165,
|
6886 |
"eval_steps_per_second": 82.582,
|
6887 |
"step": 978
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
6888 |
}
|
6889 |
],
|
6890 |
"logging_steps": 1,
|
@@ -6899,12 +9174,12 @@
|
|
6899 |
"should_evaluate": false,
|
6900 |
"should_log": false,
|
6901 |
"should_save": true,
|
6902 |
-
"should_training_stop":
|
6903 |
},
|
6904 |
"attributes": {}
|
6905 |
}
|
6906 |
},
|
6907 |
-
"total_flos":
|
6908 |
"train_batch_size": 2,
|
6909 |
"trial_name": null,
|
6910 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.02622865022092052,
|
5 |
"eval_steps": 326,
|
6 |
+
"global_step": 1303,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
6885 |
"eval_samples_per_second": 165.165,
|
6886 |
"eval_steps_per_second": 82.582,
|
6887 |
"step": 978
|
6888 |
+
},
|
6889 |
+
{
|
6890 |
+
"epoch": 0.019706714172126775,
|
6891 |
+
"grad_norm": 0.00031089509138837457,
|
6892 |
+
"learning_rate": 2.9418293126798603e-05,
|
6893 |
+
"loss": 46.0,
|
6894 |
+
"step": 979
|
6895 |
+
},
|
6896 |
+
{
|
6897 |
+
"epoch": 0.01972684360437614,
|
6898 |
+
"grad_norm": 0.0008214189438149333,
|
6899 |
+
"learning_rate": 2.9246383624288387e-05,
|
6900 |
+
"loss": 46.0,
|
6901 |
+
"step": 980
|
6902 |
+
},
|
6903 |
+
{
|
6904 |
+
"epoch": 0.019746973036625502,
|
6905 |
+
"grad_norm": 0.0012812769273295999,
|
6906 |
+
"learning_rate": 2.9074891808937753e-05,
|
6907 |
+
"loss": 46.0,
|
6908 |
+
"step": 981
|
6909 |
+
},
|
6910 |
+
{
|
6911 |
+
"epoch": 0.019767102468874864,
|
6912 |
+
"grad_norm": 0.0005773354787379503,
|
6913 |
+
"learning_rate": 2.8903818693132077e-05,
|
6914 |
+
"loss": 46.0,
|
6915 |
+
"step": 982
|
6916 |
+
},
|
6917 |
+
{
|
6918 |
+
"epoch": 0.01978723190112423,
|
6919 |
+
"grad_norm": 0.000667093729134649,
|
6920 |
+
"learning_rate": 2.873316528678507e-05,
|
6921 |
+
"loss": 46.0,
|
6922 |
+
"step": 983
|
6923 |
+
},
|
6924 |
+
{
|
6925 |
+
"epoch": 0.01980736133337359,
|
6926 |
+
"grad_norm": 0.0005955504602752626,
|
6927 |
+
"learning_rate": 2.856293259733266e-05,
|
6928 |
+
"loss": 46.0,
|
6929 |
+
"step": 984
|
6930 |
+
},
|
6931 |
+
{
|
6932 |
+
"epoch": 0.019827490765622957,
|
6933 |
+
"grad_norm": 0.0009190890123136342,
|
6934 |
+
"learning_rate": 2.8393121629727138e-05,
|
6935 |
+
"loss": 46.0,
|
6936 |
+
"step": 985
|
6937 |
+
},
|
6938 |
+
{
|
6939 |
+
"epoch": 0.01984762019787232,
|
6940 |
+
"grad_norm": 0.000597997335717082,
|
6941 |
+
"learning_rate": 2.8223733386431185e-05,
|
6942 |
+
"loss": 46.0,
|
6943 |
+
"step": 986
|
6944 |
+
},
|
6945 |
+
{
|
6946 |
+
"epoch": 0.01986774963012168,
|
6947 |
+
"grad_norm": 0.001080994145013392,
|
6948 |
+
"learning_rate": 2.8054768867411974e-05,
|
6949 |
+
"loss": 46.0,
|
6950 |
+
"step": 987
|
6951 |
+
},
|
6952 |
+
{
|
6953 |
+
"epoch": 0.019887879062371046,
|
6954 |
+
"grad_norm": 0.0009957716101780534,
|
6955 |
+
"learning_rate": 2.788622907013526e-05,
|
6956 |
+
"loss": 46.0,
|
6957 |
+
"step": 988
|
6958 |
+
},
|
6959 |
+
{
|
6960 |
+
"epoch": 0.01990800849462041,
|
6961 |
+
"grad_norm": 0.0005448372685350478,
|
6962 |
+
"learning_rate": 2.7718114989559552e-05,
|
6963 |
+
"loss": 46.0,
|
6964 |
+
"step": 989
|
6965 |
+
},
|
6966 |
+
{
|
6967 |
+
"epoch": 0.019928137926869774,
|
6968 |
+
"grad_norm": 0.0005995734827592969,
|
6969 |
+
"learning_rate": 2.7550427618130127e-05,
|
6970 |
+
"loss": 46.0,
|
6971 |
+
"step": 990
|
6972 |
+
},
|
6973 |
+
{
|
6974 |
+
"epoch": 0.019948267359119136,
|
6975 |
+
"grad_norm": 0.0009233965538442135,
|
6976 |
+
"learning_rate": 2.738316794577315e-05,
|
6977 |
+
"loss": 46.0,
|
6978 |
+
"step": 991
|
6979 |
+
},
|
6980 |
+
{
|
6981 |
+
"epoch": 0.019968396791368498,
|
6982 |
+
"grad_norm": 0.000919658865313977,
|
6983 |
+
"learning_rate": 2.7216336959890076e-05,
|
6984 |
+
"loss": 46.0,
|
6985 |
+
"step": 992
|
6986 |
+
},
|
6987 |
+
{
|
6988 |
+
"epoch": 0.019988526223617863,
|
6989 |
+
"grad_norm": 0.0008890883764252067,
|
6990 |
+
"learning_rate": 2.704993564535152e-05,
|
6991 |
+
"loss": 46.0,
|
6992 |
+
"step": 993
|
6993 |
+
},
|
6994 |
+
{
|
6995 |
+
"epoch": 0.020008655655867225,
|
6996 |
+
"grad_norm": 0.0005916538066230714,
|
6997 |
+
"learning_rate": 2.688396498449164e-05,
|
6998 |
+
"loss": 46.0,
|
6999 |
+
"step": 994
|
7000 |
+
},
|
7001 |
+
{
|
7002 |
+
"epoch": 0.02002878508811659,
|
7003 |
+
"grad_norm": 0.0006696307682432234,
|
7004 |
+
"learning_rate": 2.671842595710219e-05,
|
7005 |
+
"loss": 46.0,
|
7006 |
+
"step": 995
|
7007 |
+
},
|
7008 |
+
{
|
7009 |
+
"epoch": 0.020048914520365953,
|
7010 |
+
"grad_norm": 0.0006121333572082222,
|
7011 |
+
"learning_rate": 2.655331954042699e-05,
|
7012 |
+
"loss": 46.0,
|
7013 |
+
"step": 996
|
7014 |
+
},
|
7015 |
+
{
|
7016 |
+
"epoch": 0.020069043952615318,
|
7017 |
+
"grad_norm": 0.0007363299373537302,
|
7018 |
+
"learning_rate": 2.638864670915572e-05,
|
7019 |
+
"loss": 46.0,
|
7020 |
+
"step": 997
|
7021 |
+
},
|
7022 |
+
{
|
7023 |
+
"epoch": 0.02008917338486468,
|
7024 |
+
"grad_norm": 0.0006978671881370246,
|
7025 |
+
"learning_rate": 2.622440843541869e-05,
|
7026 |
+
"loss": 46.0,
|
7027 |
+
"step": 998
|
7028 |
+
},
|
7029 |
+
{
|
7030 |
+
"epoch": 0.020109302817114042,
|
7031 |
+
"grad_norm": 0.0010457762982696295,
|
7032 |
+
"learning_rate": 2.6060605688780694e-05,
|
7033 |
+
"loss": 46.0,
|
7034 |
+
"step": 999
|
7035 |
+
},
|
7036 |
+
{
|
7037 |
+
"epoch": 0.020129432249363408,
|
7038 |
+
"grad_norm": 0.0013878996251150966,
|
7039 |
+
"learning_rate": 2.5897239436235466e-05,
|
7040 |
+
"loss": 46.0,
|
7041 |
+
"step": 1000
|
7042 |
+
},
|
7043 |
+
{
|
7044 |
+
"epoch": 0.02014956168161277,
|
7045 |
+
"grad_norm": 0.0007541460217908025,
|
7046 |
+
"learning_rate": 2.5734310642199943e-05,
|
7047 |
+
"loss": 46.0,
|
7048 |
+
"step": 1001
|
7049 |
+
},
|
7050 |
+
{
|
7051 |
+
"epoch": 0.020169691113862135,
|
7052 |
+
"grad_norm": 0.0015688574640080333,
|
7053 |
+
"learning_rate": 2.557182026850855e-05,
|
7054 |
+
"loss": 46.0,
|
7055 |
+
"step": 1002
|
7056 |
+
},
|
7057 |
+
{
|
7058 |
+
"epoch": 0.020189820546111497,
|
7059 |
+
"grad_norm": 0.0007213862845674157,
|
7060 |
+
"learning_rate": 2.5409769274407637e-05,
|
7061 |
+
"loss": 46.0,
|
7062 |
+
"step": 1003
|
7063 |
+
},
|
7064 |
+
{
|
7065 |
+
"epoch": 0.02020994997836086,
|
7066 |
+
"grad_norm": 0.0005947791505604982,
|
7067 |
+
"learning_rate": 2.524815861654952e-05,
|
7068 |
+
"loss": 46.0,
|
7069 |
+
"step": 1004
|
7070 |
+
},
|
7071 |
+
{
|
7072 |
+
"epoch": 0.020230079410610224,
|
7073 |
+
"grad_norm": 0.0006019077845849097,
|
7074 |
+
"learning_rate": 2.5086989248987248e-05,
|
7075 |
+
"loss": 46.0,
|
7076 |
+
"step": 1005
|
7077 |
+
},
|
7078 |
+
{
|
7079 |
+
"epoch": 0.020250208842859586,
|
7080 |
+
"grad_norm": 0.0012076911516487598,
|
7081 |
+
"learning_rate": 2.492626212316862e-05,
|
7082 |
+
"loss": 46.0,
|
7083 |
+
"step": 1006
|
7084 |
+
},
|
7085 |
+
{
|
7086 |
+
"epoch": 0.020270338275108952,
|
7087 |
+
"grad_norm": 0.0010813012486323714,
|
7088 |
+
"learning_rate": 2.476597818793075e-05,
|
7089 |
+
"loss": 46.0,
|
7090 |
+
"step": 1007
|
7091 |
+
},
|
7092 |
+
{
|
7093 |
+
"epoch": 0.020290467707358314,
|
7094 |
+
"grad_norm": 0.000990718137472868,
|
7095 |
+
"learning_rate": 2.460613838949437e-05,
|
7096 |
+
"loss": 46.0,
|
7097 |
+
"step": 1008
|
7098 |
+
},
|
7099 |
+
{
|
7100 |
+
"epoch": 0.020310597139607676,
|
7101 |
+
"grad_norm": 0.001392314094118774,
|
7102 |
+
"learning_rate": 2.444674367145845e-05,
|
7103 |
+
"loss": 46.0,
|
7104 |
+
"step": 1009
|
7105 |
+
},
|
7106 |
+
{
|
7107 |
+
"epoch": 0.02033072657185704,
|
7108 |
+
"grad_norm": 0.0014354052254930139,
|
7109 |
+
"learning_rate": 2.4287794974794208e-05,
|
7110 |
+
"loss": 46.0,
|
7111 |
+
"step": 1010
|
7112 |
+
},
|
7113 |
+
{
|
7114 |
+
"epoch": 0.020350856004106403,
|
7115 |
+
"grad_norm": 0.00045173740363679826,
|
7116 |
+
"learning_rate": 2.4129293237840066e-05,
|
7117 |
+
"loss": 46.0,
|
7118 |
+
"step": 1011
|
7119 |
+
},
|
7120 |
+
{
|
7121 |
+
"epoch": 0.02037098543635577,
|
7122 |
+
"grad_norm": 0.0011678735027089715,
|
7123 |
+
"learning_rate": 2.397123939629574e-05,
|
7124 |
+
"loss": 46.0,
|
7125 |
+
"step": 1012
|
7126 |
+
},
|
7127 |
+
{
|
7128 |
+
"epoch": 0.02039111486860513,
|
7129 |
+
"grad_norm": 0.0005970303900539875,
|
7130 |
+
"learning_rate": 2.3813634383216853e-05,
|
7131 |
+
"loss": 46.0,
|
7132 |
+
"step": 1013
|
7133 |
+
},
|
7134 |
+
{
|
7135 |
+
"epoch": 0.020411244300854496,
|
7136 |
+
"grad_norm": 0.0007846534135751426,
|
7137 |
+
"learning_rate": 2.3656479129009422e-05,
|
7138 |
+
"loss": 46.0,
|
7139 |
+
"step": 1014
|
7140 |
+
},
|
7141 |
+
{
|
7142 |
+
"epoch": 0.020431373733103858,
|
7143 |
+
"grad_norm": 0.00044201669516041875,
|
7144 |
+
"learning_rate": 2.3499774561424327e-05,
|
7145 |
+
"loss": 46.0,
|
7146 |
+
"step": 1015
|
7147 |
+
},
|
7148 |
+
{
|
7149 |
+
"epoch": 0.02045150316535322,
|
7150 |
+
"grad_norm": 0.001156677259132266,
|
7151 |
+
"learning_rate": 2.3343521605551967e-05,
|
7152 |
+
"loss": 46.0,
|
7153 |
+
"step": 1016
|
7154 |
+
},
|
7155 |
+
{
|
7156 |
+
"epoch": 0.020471632597602585,
|
7157 |
+
"grad_norm": 0.0006299121305346489,
|
7158 |
+
"learning_rate": 2.3187721183816503e-05,
|
7159 |
+
"loss": 46.0,
|
7160 |
+
"step": 1017
|
7161 |
+
},
|
7162 |
+
{
|
7163 |
+
"epoch": 0.020491762029851947,
|
7164 |
+
"grad_norm": 0.0007507737609557807,
|
7165 |
+
"learning_rate": 2.303237421597082e-05,
|
7166 |
+
"loss": 46.0,
|
7167 |
+
"step": 1018
|
7168 |
+
},
|
7169 |
+
{
|
7170 |
+
"epoch": 0.020511891462101313,
|
7171 |
+
"grad_norm": 0.0007438535685651004,
|
7172 |
+
"learning_rate": 2.2877481619090734e-05,
|
7173 |
+
"loss": 46.0,
|
7174 |
+
"step": 1019
|
7175 |
+
},
|
7176 |
+
{
|
7177 |
+
"epoch": 0.020532020894350675,
|
7178 |
+
"grad_norm": 0.0016225421568378806,
|
7179 |
+
"learning_rate": 2.2723044307569775e-05,
|
7180 |
+
"loss": 46.0,
|
7181 |
+
"step": 1020
|
7182 |
+
},
|
7183 |
+
{
|
7184 |
+
"epoch": 0.020552150326600037,
|
7185 |
+
"grad_norm": 0.00039501202991232276,
|
7186 |
+
"learning_rate": 2.2569063193113716e-05,
|
7187 |
+
"loss": 46.0,
|
7188 |
+
"step": 1021
|
7189 |
+
},
|
7190 |
+
{
|
7191 |
+
"epoch": 0.020572279758849402,
|
7192 |
+
"grad_norm": 0.0004267230106052011,
|
7193 |
+
"learning_rate": 2.2415539184735323e-05,
|
7194 |
+
"loss": 46.0,
|
7195 |
+
"step": 1022
|
7196 |
+
},
|
7197 |
+
{
|
7198 |
+
"epoch": 0.020592409191098764,
|
7199 |
+
"grad_norm": 0.0008113220101222396,
|
7200 |
+
"learning_rate": 2.2262473188748667e-05,
|
7201 |
+
"loss": 46.0,
|
7202 |
+
"step": 1023
|
7203 |
+
},
|
7204 |
+
{
|
7205 |
+
"epoch": 0.02061253862334813,
|
7206 |
+
"grad_norm": 0.0018733137985691428,
|
7207 |
+
"learning_rate": 2.210986610876421e-05,
|
7208 |
+
"loss": 46.0,
|
7209 |
+
"step": 1024
|
7210 |
+
},
|
7211 |
+
{
|
7212 |
+
"epoch": 0.02063266805559749,
|
7213 |
+
"grad_norm": 0.0011937421513721347,
|
7214 |
+
"learning_rate": 2.1957718845683106e-05,
|
7215 |
+
"loss": 46.0,
|
7216 |
+
"step": 1025
|
7217 |
+
},
|
7218 |
+
{
|
7219 |
+
"epoch": 0.020652797487846854,
|
7220 |
+
"grad_norm": 0.0011691252002492547,
|
7221 |
+
"learning_rate": 2.180603229769208e-05,
|
7222 |
+
"loss": 46.0,
|
7223 |
+
"step": 1026
|
7224 |
+
},
|
7225 |
+
{
|
7226 |
+
"epoch": 0.02067292692009622,
|
7227 |
+
"grad_norm": 0.0005129252676852047,
|
7228 |
+
"learning_rate": 2.1654807360258068e-05,
|
7229 |
+
"loss": 46.0,
|
7230 |
+
"step": 1027
|
7231 |
+
},
|
7232 |
+
{
|
7233 |
+
"epoch": 0.02069305635234558,
|
7234 |
+
"grad_norm": 0.001547716441564262,
|
7235 |
+
"learning_rate": 2.15040449261229e-05,
|
7236 |
+
"loss": 46.0,
|
7237 |
+
"step": 1028
|
7238 |
+
},
|
7239 |
+
{
|
7240 |
+
"epoch": 0.020713185784594947,
|
7241 |
+
"grad_norm": 0.0005555424140766263,
|
7242 |
+
"learning_rate": 2.1353745885298103e-05,
|
7243 |
+
"loss": 46.0,
|
7244 |
+
"step": 1029
|
7245 |
+
},
|
7246 |
+
{
|
7247 |
+
"epoch": 0.02073331521684431,
|
7248 |
+
"grad_norm": 0.0006457091076299548,
|
7249 |
+
"learning_rate": 2.120391112505955e-05,
|
7250 |
+
"loss": 46.0,
|
7251 |
+
"step": 1030
|
7252 |
+
},
|
7253 |
+
{
|
7254 |
+
"epoch": 0.020753444649093674,
|
7255 |
+
"grad_norm": 0.0005488627939485013,
|
7256 |
+
"learning_rate": 2.1054541529942374e-05,
|
7257 |
+
"loss": 46.0,
|
7258 |
+
"step": 1031
|
7259 |
+
},
|
7260 |
+
{
|
7261 |
+
"epoch": 0.020773574081343036,
|
7262 |
+
"grad_norm": 0.001043295138515532,
|
7263 |
+
"learning_rate": 2.090563798173557e-05,
|
7264 |
+
"loss": 46.0,
|
7265 |
+
"step": 1032
|
7266 |
+
},
|
7267 |
+
{
|
7268 |
+
"epoch": 0.020793703513592398,
|
7269 |
+
"grad_norm": 0.0010731170186772943,
|
7270 |
+
"learning_rate": 2.0757201359476884e-05,
|
7271 |
+
"loss": 46.0,
|
7272 |
+
"step": 1033
|
7273 |
+
},
|
7274 |
+
{
|
7275 |
+
"epoch": 0.020813832945841763,
|
7276 |
+
"grad_norm": 0.0010378467850387096,
|
7277 |
+
"learning_rate": 2.060923253944761e-05,
|
7278 |
+
"loss": 46.0,
|
7279 |
+
"step": 1034
|
7280 |
+
},
|
7281 |
+
{
|
7282 |
+
"epoch": 0.020833962378091125,
|
7283 |
+
"grad_norm": 0.0009367475286126137,
|
7284 |
+
"learning_rate": 2.0461732395167475e-05,
|
7285 |
+
"loss": 46.0,
|
7286 |
+
"step": 1035
|
7287 |
+
},
|
7288 |
+
{
|
7289 |
+
"epoch": 0.02085409181034049,
|
7290 |
+
"grad_norm": 0.0012599321780726314,
|
7291 |
+
"learning_rate": 2.03147017973893e-05,
|
7292 |
+
"loss": 46.0,
|
7293 |
+
"step": 1036
|
7294 |
+
},
|
7295 |
+
{
|
7296 |
+
"epoch": 0.020874221242589853,
|
7297 |
+
"grad_norm": 0.0007634823559783399,
|
7298 |
+
"learning_rate": 2.0168141614094126e-05,
|
7299 |
+
"loss": 46.0,
|
7300 |
+
"step": 1037
|
7301 |
+
},
|
7302 |
+
{
|
7303 |
+
"epoch": 0.020894350674839215,
|
7304 |
+
"grad_norm": 0.0008573421509936452,
|
7305 |
+
"learning_rate": 2.0022052710485874e-05,
|
7306 |
+
"loss": 46.0,
|
7307 |
+
"step": 1038
|
7308 |
+
},
|
7309 |
+
{
|
7310 |
+
"epoch": 0.02091448010708858,
|
7311 |
+
"grad_norm": 0.0009650752181187272,
|
7312 |
+
"learning_rate": 1.9876435948986228e-05,
|
7313 |
+
"loss": 46.0,
|
7314 |
+
"step": 1039
|
7315 |
+
},
|
7316 |
+
{
|
7317 |
+
"epoch": 0.020934609539337942,
|
7318 |
+
"grad_norm": 0.0008246242650784552,
|
7319 |
+
"learning_rate": 1.973129218922981e-05,
|
7320 |
+
"loss": 46.0,
|
7321 |
+
"step": 1040
|
7322 |
+
},
|
7323 |
+
{
|
7324 |
+
"epoch": 0.020954738971587308,
|
7325 |
+
"grad_norm": 0.0008325594244524837,
|
7326 |
+
"learning_rate": 1.95866222880588e-05,
|
7327 |
+
"loss": 46.0,
|
7328 |
+
"step": 1041
|
7329 |
+
},
|
7330 |
+
{
|
7331 |
+
"epoch": 0.02097486840383667,
|
7332 |
+
"grad_norm": 0.0007449231925420463,
|
7333 |
+
"learning_rate": 1.9442427099518022e-05,
|
7334 |
+
"loss": 46.0,
|
7335 |
+
"step": 1042
|
7336 |
+
},
|
7337 |
+
{
|
7338 |
+
"epoch": 0.02099499783608603,
|
7339 |
+
"grad_norm": 0.0016277192626148462,
|
7340 |
+
"learning_rate": 1.9298707474849843e-05,
|
7341 |
+
"loss": 46.0,
|
7342 |
+
"step": 1043
|
7343 |
+
},
|
7344 |
+
{
|
7345 |
+
"epoch": 0.021015127268335397,
|
7346 |
+
"grad_norm": 0.0005063241114839911,
|
7347 |
+
"learning_rate": 1.9155464262489298e-05,
|
7348 |
+
"loss": 46.0,
|
7349 |
+
"step": 1044
|
7350 |
+
},
|
7351 |
+
{
|
7352 |
+
"epoch": 0.02103525670058476,
|
7353 |
+
"grad_norm": 0.0008978778496384621,
|
7354 |
+
"learning_rate": 1.9012698308058852e-05,
|
7355 |
+
"loss": 46.0,
|
7356 |
+
"step": 1045
|
7357 |
+
},
|
7358 |
+
{
|
7359 |
+
"epoch": 0.021055386132834124,
|
7360 |
+
"grad_norm": 0.0007100282236933708,
|
7361 |
+
"learning_rate": 1.8870410454363573e-05,
|
7362 |
+
"loss": 46.0,
|
7363 |
+
"step": 1046
|
7364 |
+
},
|
7365 |
+
{
|
7366 |
+
"epoch": 0.021075515565083486,
|
7367 |
+
"grad_norm": 0.0006800219998694956,
|
7368 |
+
"learning_rate": 1.872860154138608e-05,
|
7369 |
+
"loss": 46.0,
|
7370 |
+
"step": 1047
|
7371 |
+
},
|
7372 |
+
{
|
7373 |
+
"epoch": 0.021095644997332852,
|
7374 |
+
"grad_norm": 0.0006419627461582422,
|
7375 |
+
"learning_rate": 1.858727240628171e-05,
|
7376 |
+
"loss": 46.0,
|
7377 |
+
"step": 1048
|
7378 |
+
},
|
7379 |
+
{
|
7380 |
+
"epoch": 0.021115774429582214,
|
7381 |
+
"grad_norm": 0.0004995979252271354,
|
7382 |
+
"learning_rate": 1.8446423883373286e-05,
|
7383 |
+
"loss": 46.0,
|
7384 |
+
"step": 1049
|
7385 |
+
},
|
7386 |
+
{
|
7387 |
+
"epoch": 0.021135903861831576,
|
7388 |
+
"grad_norm": 0.0007924246601760387,
|
7389 |
+
"learning_rate": 1.8306056804146575e-05,
|
7390 |
+
"loss": 46.0,
|
7391 |
+
"step": 1050
|
7392 |
+
},
|
7393 |
+
{
|
7394 |
+
"epoch": 0.02115603329408094,
|
7395 |
+
"grad_norm": 0.0006777087110094726,
|
7396 |
+
"learning_rate": 1.816617199724512e-05,
|
7397 |
+
"loss": 46.0,
|
7398 |
+
"step": 1051
|
7399 |
+
},
|
7400 |
+
{
|
7401 |
+
"epoch": 0.021176162726330303,
|
7402 |
+
"grad_norm": 0.0013369604712352157,
|
7403 |
+
"learning_rate": 1.8026770288465323e-05,
|
7404 |
+
"loss": 46.0,
|
7405 |
+
"step": 1052
|
7406 |
+
},
|
7407 |
+
{
|
7408 |
+
"epoch": 0.02119629215857967,
|
7409 |
+
"grad_norm": 0.0005232554394751787,
|
7410 |
+
"learning_rate": 1.7887852500751822e-05,
|
7411 |
+
"loss": 46.0,
|
7412 |
+
"step": 1053
|
7413 |
+
},
|
7414 |
+
{
|
7415 |
+
"epoch": 0.02121642159082903,
|
7416 |
+
"grad_norm": 0.0013328958302736282,
|
7417 |
+
"learning_rate": 1.7749419454192373e-05,
|
7418 |
+
"loss": 46.0,
|
7419 |
+
"step": 1054
|
7420 |
+
},
|
7421 |
+
{
|
7422 |
+
"epoch": 0.021236551023078393,
|
7423 |
+
"grad_norm": 0.0014558390248566866,
|
7424 |
+
"learning_rate": 1.7611471966013127e-05,
|
7425 |
+
"loss": 46.0,
|
7426 |
+
"step": 1055
|
7427 |
+
},
|
7428 |
+
{
|
7429 |
+
"epoch": 0.021256680455327758,
|
7430 |
+
"grad_norm": 0.0010977151105180383,
|
7431 |
+
"learning_rate": 1.7474010850573775e-05,
|
7432 |
+
"loss": 46.0,
|
7433 |
+
"step": 1056
|
7434 |
+
},
|
7435 |
+
{
|
7436 |
+
"epoch": 0.02127680988757712,
|
7437 |
+
"grad_norm": 0.000998729607090354,
|
7438 |
+
"learning_rate": 1.7337036919362827e-05,
|
7439 |
+
"loss": 46.0,
|
7440 |
+
"step": 1057
|
7441 |
+
},
|
7442 |
+
{
|
7443 |
+
"epoch": 0.021296939319826486,
|
7444 |
+
"grad_norm": 0.00059292814694345,
|
7445 |
+
"learning_rate": 1.7200550980992647e-05,
|
7446 |
+
"loss": 46.0,
|
7447 |
+
"step": 1058
|
7448 |
+
},
|
7449 |
+
{
|
7450 |
+
"epoch": 0.021317068752075848,
|
7451 |
+
"grad_norm": 0.0010517132468521595,
|
7452 |
+
"learning_rate": 1.706455384119485e-05,
|
7453 |
+
"loss": 46.0,
|
7454 |
+
"step": 1059
|
7455 |
+
},
|
7456 |
+
{
|
7457 |
+
"epoch": 0.02133719818432521,
|
7458 |
+
"grad_norm": 0.0010053809965029359,
|
7459 |
+
"learning_rate": 1.6929046302815443e-05,
|
7460 |
+
"loss": 46.0,
|
7461 |
+
"step": 1060
|
7462 |
+
},
|
7463 |
+
{
|
7464 |
+
"epoch": 0.021357327616574575,
|
7465 |
+
"grad_norm": 0.0006442563608288765,
|
7466 |
+
"learning_rate": 1.6794029165810133e-05,
|
7467 |
+
"loss": 46.0,
|
7468 |
+
"step": 1061
|
7469 |
+
},
|
7470 |
+
{
|
7471 |
+
"epoch": 0.021377457048823937,
|
7472 |
+
"grad_norm": 0.0004650287446565926,
|
7473 |
+
"learning_rate": 1.665950322723957e-05,
|
7474 |
+
"loss": 46.0,
|
7475 |
+
"step": 1062
|
7476 |
+
},
|
7477 |
+
{
|
7478 |
+
"epoch": 0.021397586481073302,
|
7479 |
+
"grad_norm": 0.0011159973219037056,
|
7480 |
+
"learning_rate": 1.652546928126466e-05,
|
7481 |
+
"loss": 46.0,
|
7482 |
+
"step": 1063
|
7483 |
+
},
|
7484 |
+
{
|
7485 |
+
"epoch": 0.021417715913322664,
|
7486 |
+
"grad_norm": 0.0005395881598815322,
|
7487 |
+
"learning_rate": 1.6391928119141965e-05,
|
7488 |
+
"loss": 46.0,
|
7489 |
+
"step": 1064
|
7490 |
+
},
|
7491 |
+
{
|
7492 |
+
"epoch": 0.02143784534557203,
|
7493 |
+
"grad_norm": 0.0005475838552229106,
|
7494 |
+
"learning_rate": 1.625888052921878e-05,
|
7495 |
+
"loss": 46.0,
|
7496 |
+
"step": 1065
|
7497 |
+
},
|
7498 |
+
{
|
7499 |
+
"epoch": 0.021457974777821392,
|
7500 |
+
"grad_norm": 0.0009001667494885623,
|
7501 |
+
"learning_rate": 1.612632729692881e-05,
|
7502 |
+
"loss": 46.0,
|
7503 |
+
"step": 1066
|
7504 |
+
},
|
7505 |
+
{
|
7506 |
+
"epoch": 0.021478104210070754,
|
7507 |
+
"grad_norm": 0.0011171525111421943,
|
7508 |
+
"learning_rate": 1.599426920478726e-05,
|
7509 |
+
"loss": 46.0,
|
7510 |
+
"step": 1067
|
7511 |
+
},
|
7512 |
+
{
|
7513 |
+
"epoch": 0.02149823364232012,
|
7514 |
+
"grad_norm": 0.0007830538670532405,
|
7515 |
+
"learning_rate": 1.586270703238637e-05,
|
7516 |
+
"loss": 46.0,
|
7517 |
+
"step": 1068
|
7518 |
+
},
|
7519 |
+
{
|
7520 |
+
"epoch": 0.02151836307456948,
|
7521 |
+
"grad_norm": 0.0005228519439697266,
|
7522 |
+
"learning_rate": 1.573164155639073e-05,
|
7523 |
+
"loss": 46.0,
|
7524 |
+
"step": 1069
|
7525 |
+
},
|
7526 |
+
{
|
7527 |
+
"epoch": 0.021538492506818847,
|
7528 |
+
"grad_norm": 0.0009138689492829144,
|
7529 |
+
"learning_rate": 1.560107355053282e-05,
|
7530 |
+
"loss": 46.0,
|
7531 |
+
"step": 1070
|
7532 |
+
},
|
7533 |
+
{
|
7534 |
+
"epoch": 0.02155862193906821,
|
7535 |
+
"grad_norm": 0.0011538650142028928,
|
7536 |
+
"learning_rate": 1.5471003785608184e-05,
|
7537 |
+
"loss": 46.0,
|
7538 |
+
"step": 1071
|
7539 |
+
},
|
7540 |
+
{
|
7541 |
+
"epoch": 0.02157875137131757,
|
7542 |
+
"grad_norm": 0.0009152950951829553,
|
7543 |
+
"learning_rate": 1.534143302947123e-05,
|
7544 |
+
"loss": 46.0,
|
7545 |
+
"step": 1072
|
7546 |
+
},
|
7547 |
+
{
|
7548 |
+
"epoch": 0.021598880803566936,
|
7549 |
+
"grad_norm": 0.0008602161542512476,
|
7550 |
+
"learning_rate": 1.5212362047030427e-05,
|
7551 |
+
"loss": 46.0,
|
7552 |
+
"step": 1073
|
7553 |
+
},
|
7554 |
+
{
|
7555 |
+
"epoch": 0.021619010235816298,
|
7556 |
+
"grad_norm": 0.000630200607702136,
|
7557 |
+
"learning_rate": 1.5083791600243857e-05,
|
7558 |
+
"loss": 46.0,
|
7559 |
+
"step": 1074
|
7560 |
+
},
|
7561 |
+
{
|
7562 |
+
"epoch": 0.021639139668065663,
|
7563 |
+
"grad_norm": 0.001679250504821539,
|
7564 |
+
"learning_rate": 1.4955722448114807e-05,
|
7565 |
+
"loss": 46.0,
|
7566 |
+
"step": 1075
|
7567 |
+
},
|
7568 |
+
{
|
7569 |
+
"epoch": 0.021659269100315025,
|
7570 |
+
"grad_norm": 0.0008510241750627756,
|
7571 |
+
"learning_rate": 1.4828155346687123e-05,
|
7572 |
+
"loss": 46.0,
|
7573 |
+
"step": 1076
|
7574 |
+
},
|
7575 |
+
{
|
7576 |
+
"epoch": 0.021679398532564387,
|
7577 |
+
"grad_norm": 0.0007962991949170828,
|
7578 |
+
"learning_rate": 1.4701091049040994e-05,
|
7579 |
+
"loss": 46.0,
|
7580 |
+
"step": 1077
|
7581 |
+
},
|
7582 |
+
{
|
7583 |
+
"epoch": 0.021699527964813753,
|
7584 |
+
"grad_norm": 0.00033583008917048573,
|
7585 |
+
"learning_rate": 1.4574530305288158e-05,
|
7586 |
+
"loss": 46.0,
|
7587 |
+
"step": 1078
|
7588 |
+
},
|
7589 |
+
{
|
7590 |
+
"epoch": 0.021719657397063115,
|
7591 |
+
"grad_norm": 0.0015587556408718228,
|
7592 |
+
"learning_rate": 1.4448473862567857e-05,
|
7593 |
+
"loss": 46.0,
|
7594 |
+
"step": 1079
|
7595 |
+
},
|
7596 |
+
{
|
7597 |
+
"epoch": 0.02173978682931248,
|
7598 |
+
"grad_norm": 0.0016931117279455066,
|
7599 |
+
"learning_rate": 1.4322922465042132e-05,
|
7600 |
+
"loss": 46.0,
|
7601 |
+
"step": 1080
|
7602 |
+
},
|
7603 |
+
{
|
7604 |
+
"epoch": 0.021759916261561842,
|
7605 |
+
"grad_norm": 0.0011070282198488712,
|
7606 |
+
"learning_rate": 1.4197876853891557e-05,
|
7607 |
+
"loss": 46.0,
|
7608 |
+
"step": 1081
|
7609 |
+
},
|
7610 |
+
{
|
7611 |
+
"epoch": 0.021780045693811208,
|
7612 |
+
"grad_norm": 0.0012120172614231706,
|
7613 |
+
"learning_rate": 1.4073337767310834e-05,
|
7614 |
+
"loss": 46.0,
|
7615 |
+
"step": 1082
|
7616 |
+
},
|
7617 |
+
{
|
7618 |
+
"epoch": 0.02180017512606057,
|
7619 |
+
"grad_norm": 0.00137874367646873,
|
7620 |
+
"learning_rate": 1.3949305940504541e-05,
|
7621 |
+
"loss": 46.0,
|
7622 |
+
"step": 1083
|
7623 |
+
},
|
7624 |
+
{
|
7625 |
+
"epoch": 0.02182030455830993,
|
7626 |
+
"grad_norm": 0.0011860841186717153,
|
7627 |
+
"learning_rate": 1.3825782105682527e-05,
|
7628 |
+
"loss": 46.0,
|
7629 |
+
"step": 1084
|
7630 |
+
},
|
7631 |
+
{
|
7632 |
+
"epoch": 0.021840433990559297,
|
7633 |
+
"grad_norm": 0.0007199230021797121,
|
7634 |
+
"learning_rate": 1.3702766992055927e-05,
|
7635 |
+
"loss": 46.0,
|
7636 |
+
"step": 1085
|
7637 |
+
},
|
7638 |
+
{
|
7639 |
+
"epoch": 0.02186056342280866,
|
7640 |
+
"grad_norm": 0.0006517477449961007,
|
7641 |
+
"learning_rate": 1.3580261325832578e-05,
|
7642 |
+
"loss": 46.0,
|
7643 |
+
"step": 1086
|
7644 |
+
},
|
7645 |
+
{
|
7646 |
+
"epoch": 0.021880692855058025,
|
7647 |
+
"grad_norm": 0.0005604965263046324,
|
7648 |
+
"learning_rate": 1.3458265830212891e-05,
|
7649 |
+
"loss": 46.0,
|
7650 |
+
"step": 1087
|
7651 |
+
},
|
7652 |
+
{
|
7653 |
+
"epoch": 0.021900822287307387,
|
7654 |
+
"grad_norm": 0.0008536526001989841,
|
7655 |
+
"learning_rate": 1.333678122538553e-05,
|
7656 |
+
"loss": 46.0,
|
7657 |
+
"step": 1088
|
7658 |
+
},
|
7659 |
+
{
|
7660 |
+
"epoch": 0.02192095171955675,
|
7661 |
+
"grad_norm": 0.00035072650643996894,
|
7662 |
+
"learning_rate": 1.321580822852313e-05,
|
7663 |
+
"loss": 46.0,
|
7664 |
+
"step": 1089
|
7665 |
+
},
|
7666 |
+
{
|
7667 |
+
"epoch": 0.021941081151806114,
|
7668 |
+
"grad_norm": 0.0014688886003568769,
|
7669 |
+
"learning_rate": 1.3095347553778193e-05,
|
7670 |
+
"loss": 46.0,
|
7671 |
+
"step": 1090
|
7672 |
+
},
|
7673 |
+
{
|
7674 |
+
"epoch": 0.021961210584055476,
|
7675 |
+
"grad_norm": 0.0017621091101318598,
|
7676 |
+
"learning_rate": 1.2975399912278608e-05,
|
7677 |
+
"loss": 46.0,
|
7678 |
+
"step": 1091
|
7679 |
+
},
|
7680 |
+
{
|
7681 |
+
"epoch": 0.02198134001630484,
|
7682 |
+
"grad_norm": 0.0005523571744561195,
|
7683 |
+
"learning_rate": 1.2855966012123822e-05,
|
7684 |
+
"loss": 46.0,
|
7685 |
+
"step": 1092
|
7686 |
+
},
|
7687 |
+
{
|
7688 |
+
"epoch": 0.022001469448554203,
|
7689 |
+
"grad_norm": 0.0010343191679567099,
|
7690 |
+
"learning_rate": 1.2737046558380305e-05,
|
7691 |
+
"loss": 46.0,
|
7692 |
+
"step": 1093
|
7693 |
+
},
|
7694 |
+
{
|
7695 |
+
"epoch": 0.022021598880803565,
|
7696 |
+
"grad_norm": 0.0010563414543867111,
|
7697 |
+
"learning_rate": 1.2618642253077628e-05,
|
7698 |
+
"loss": 46.0,
|
7699 |
+
"step": 1094
|
7700 |
+
},
|
7701 |
+
{
|
7702 |
+
"epoch": 0.02204172831305293,
|
7703 |
+
"grad_norm": 0.0008712798589840531,
|
7704 |
+
"learning_rate": 1.2500753795204157e-05,
|
7705 |
+
"loss": 46.0,
|
7706 |
+
"step": 1095
|
7707 |
+
},
|
7708 |
+
{
|
7709 |
+
"epoch": 0.022061857745302293,
|
7710 |
+
"grad_norm": 0.0011983743170276284,
|
7711 |
+
"learning_rate": 1.2383381880703138e-05,
|
7712 |
+
"loss": 46.0,
|
7713 |
+
"step": 1096
|
7714 |
+
},
|
7715 |
+
{
|
7716 |
+
"epoch": 0.022081987177551658,
|
7717 |
+
"grad_norm": 0.001204630360007286,
|
7718 |
+
"learning_rate": 1.2266527202468248e-05,
|
7719 |
+
"loss": 46.0,
|
7720 |
+
"step": 1097
|
7721 |
+
},
|
7722 |
+
{
|
7723 |
+
"epoch": 0.02210211660980102,
|
7724 |
+
"grad_norm": 0.0008704798528924584,
|
7725 |
+
"learning_rate": 1.2150190450339915e-05,
|
7726 |
+
"loss": 46.0,
|
7727 |
+
"step": 1098
|
7728 |
+
},
|
7729 |
+
{
|
7730 |
+
"epoch": 0.022122246042050386,
|
7731 |
+
"grad_norm": 0.0009507142240181565,
|
7732 |
+
"learning_rate": 1.2034372311100905e-05,
|
7733 |
+
"loss": 46.0,
|
7734 |
+
"step": 1099
|
7735 |
+
},
|
7736 |
+
{
|
7737 |
+
"epoch": 0.022142375474299748,
|
7738 |
+
"grad_norm": 0.0018596797017380595,
|
7739 |
+
"learning_rate": 1.1919073468472475e-05,
|
7740 |
+
"loss": 46.0,
|
7741 |
+
"step": 1100
|
7742 |
+
},
|
7743 |
+
{
|
7744 |
+
"epoch": 0.02216250490654911,
|
7745 |
+
"grad_norm": 0.0010968039277940989,
|
7746 |
+
"learning_rate": 1.1804294603110222e-05,
|
7747 |
+
"loss": 46.0,
|
7748 |
+
"step": 1101
|
7749 |
+
},
|
7750 |
+
{
|
7751 |
+
"epoch": 0.022182634338798475,
|
7752 |
+
"grad_norm": 0.0012583578936755657,
|
7753 |
+
"learning_rate": 1.1690036392600112e-05,
|
7754 |
+
"loss": 46.0,
|
7755 |
+
"step": 1102
|
7756 |
+
},
|
7757 |
+
{
|
7758 |
+
"epoch": 0.022202763771047837,
|
7759 |
+
"grad_norm": 0.0011010583257302642,
|
7760 |
+
"learning_rate": 1.1576299511454513e-05,
|
7761 |
+
"loss": 46.0,
|
7762 |
+
"step": 1103
|
7763 |
+
},
|
7764 |
+
{
|
7765 |
+
"epoch": 0.022222893203297202,
|
7766 |
+
"grad_norm": 0.000499493908137083,
|
7767 |
+
"learning_rate": 1.1463084631108101e-05,
|
7768 |
+
"loss": 46.0,
|
7769 |
+
"step": 1104
|
7770 |
+
},
|
7771 |
+
{
|
7772 |
+
"epoch": 0.022243022635546564,
|
7773 |
+
"grad_norm": 0.0009412445360794663,
|
7774 |
+
"learning_rate": 1.135039241991408e-05,
|
7775 |
+
"loss": 46.0,
|
7776 |
+
"step": 1105
|
7777 |
+
},
|
7778 |
+
{
|
7779 |
+
"epoch": 0.022263152067795926,
|
7780 |
+
"grad_norm": 0.0010741801233962178,
|
7781 |
+
"learning_rate": 1.1238223543140024e-05,
|
7782 |
+
"loss": 46.0,
|
7783 |
+
"step": 1106
|
7784 |
+
},
|
7785 |
+
{
|
7786 |
+
"epoch": 0.022283281500045292,
|
7787 |
+
"grad_norm": 0.0016603496624156833,
|
7788 |
+
"learning_rate": 1.1126578662964115e-05,
|
7789 |
+
"loss": 46.0,
|
7790 |
+
"step": 1107
|
7791 |
+
},
|
7792 |
+
{
|
7793 |
+
"epoch": 0.022303410932294654,
|
7794 |
+
"grad_norm": 0.0007736904663033783,
|
7795 |
+
"learning_rate": 1.1015458438471116e-05,
|
7796 |
+
"loss": 46.0,
|
7797 |
+
"step": 1108
|
7798 |
+
},
|
7799 |
+
{
|
7800 |
+
"epoch": 0.02232354036454402,
|
7801 |
+
"grad_norm": 0.0007681693532504141,
|
7802 |
+
"learning_rate": 1.0904863525648633e-05,
|
7803 |
+
"loss": 46.0,
|
7804 |
+
"step": 1109
|
7805 |
+
},
|
7806 |
+
{
|
7807 |
+
"epoch": 0.02234366979679338,
|
7808 |
+
"grad_norm": 0.0009165824740193784,
|
7809 |
+
"learning_rate": 1.0794794577383016e-05,
|
7810 |
+
"loss": 46.0,
|
7811 |
+
"step": 1110
|
7812 |
+
},
|
7813 |
+
{
|
7814 |
+
"epoch": 0.022363799229042743,
|
7815 |
+
"grad_norm": 0.0012590873520821333,
|
7816 |
+
"learning_rate": 1.0685252243455712e-05,
|
7817 |
+
"loss": 46.0,
|
7818 |
+
"step": 1111
|
7819 |
+
},
|
7820 |
+
{
|
7821 |
+
"epoch": 0.02238392866129211,
|
7822 |
+
"grad_norm": 0.0006520768511109054,
|
7823 |
+
"learning_rate": 1.0576237170539383e-05,
|
7824 |
+
"loss": 46.0,
|
7825 |
+
"step": 1112
|
7826 |
+
},
|
7827 |
+
{
|
7828 |
+
"epoch": 0.02240405809354147,
|
7829 |
+
"grad_norm": 0.0008068094030022621,
|
7830 |
+
"learning_rate": 1.0467750002193944e-05,
|
7831 |
+
"loss": 46.0,
|
7832 |
+
"step": 1113
|
7833 |
+
},
|
7834 |
+
{
|
7835 |
+
"epoch": 0.022424187525790836,
|
7836 |
+
"grad_norm": 0.0007286273175850511,
|
7837 |
+
"learning_rate": 1.0359791378863005e-05,
|
7838 |
+
"loss": 46.0,
|
7839 |
+
"step": 1114
|
7840 |
+
},
|
7841 |
+
{
|
7842 |
+
"epoch": 0.022444316958040198,
|
7843 |
+
"grad_norm": 0.00037491964758373797,
|
7844 |
+
"learning_rate": 1.025236193786987e-05,
|
7845 |
+
"loss": 46.0,
|
7846 |
+
"step": 1115
|
7847 |
+
},
|
7848 |
+
{
|
7849 |
+
"epoch": 0.022464446390289564,
|
7850 |
+
"grad_norm": 0.0006626130198128521,
|
7851 |
+
"learning_rate": 1.014546231341391e-05,
|
7852 |
+
"loss": 46.0,
|
7853 |
+
"step": 1116
|
7854 |
+
},
|
7855 |
+
{
|
7856 |
+
"epoch": 0.022484575822538926,
|
7857 |
+
"grad_norm": 0.0007106042467057705,
|
7858 |
+
"learning_rate": 1.0039093136566735e-05,
|
7859 |
+
"loss": 46.0,
|
7860 |
+
"step": 1117
|
7861 |
+
},
|
7862 |
+
{
|
7863 |
+
"epoch": 0.022504705254788288,
|
7864 |
+
"grad_norm": 0.0007863907376304269,
|
7865 |
+
"learning_rate": 9.933255035268574e-06,
|
7866 |
+
"loss": 46.0,
|
7867 |
+
"step": 1118
|
7868 |
+
},
|
7869 |
+
{
|
7870 |
+
"epoch": 0.022524834687037653,
|
7871 |
+
"grad_norm": 0.0013863188214600086,
|
7872 |
+
"learning_rate": 9.827948634324447e-06,
|
7873 |
+
"loss": 46.0,
|
7874 |
+
"step": 1119
|
7875 |
+
},
|
7876 |
+
{
|
7877 |
+
"epoch": 0.022544964119287015,
|
7878 |
+
"grad_norm": 0.0006982397171668708,
|
7879 |
+
"learning_rate": 9.72317455540055e-06,
|
7880 |
+
"loss": 46.0,
|
7881 |
+
"step": 1120
|
7882 |
+
},
|
7883 |
+
{
|
7884 |
+
"epoch": 0.02256509355153638,
|
7885 |
+
"grad_norm": 0.0007036144379526377,
|
7886 |
+
"learning_rate": 9.61893341702056e-06,
|
7887 |
+
"loss": 46.0,
|
7888 |
+
"step": 1121
|
7889 |
+
},
|
7890 |
+
{
|
7891 |
+
"epoch": 0.022585222983785742,
|
7892 |
+
"grad_norm": 0.0005324577214196324,
|
7893 |
+
"learning_rate": 9.515225834562003e-06,
|
7894 |
+
"loss": 46.0,
|
7895 |
+
"step": 1122
|
7896 |
+
},
|
7897 |
+
{
|
7898 |
+
"epoch": 0.022605352416035104,
|
7899 |
+
"grad_norm": 0.00036683998769149184,
|
7900 |
+
"learning_rate": 9.412052420252605e-06,
|
7901 |
+
"loss": 46.0,
|
7902 |
+
"step": 1123
|
7903 |
+
},
|
7904 |
+
{
|
7905 |
+
"epoch": 0.02262548184828447,
|
7906 |
+
"grad_norm": 0.0008661012398079038,
|
7907 |
+
"learning_rate": 9.309413783166654e-06,
|
7908 |
+
"loss": 46.0,
|
7909 |
+
"step": 1124
|
7910 |
+
},
|
7911 |
+
{
|
7912 |
+
"epoch": 0.022645611280533832,
|
7913 |
+
"grad_norm": 0.0007851457339711487,
|
7914 |
+
"learning_rate": 9.207310529221525e-06,
|
7915 |
+
"loss": 46.0,
|
7916 |
+
"step": 1125
|
7917 |
+
},
|
7918 |
+
{
|
7919 |
+
"epoch": 0.022665740712783197,
|
7920 |
+
"grad_norm": 0.0005590534419752657,
|
7921 |
+
"learning_rate": 9.10574326117386e-06,
|
7922 |
+
"loss": 46.0,
|
7923 |
+
"step": 1126
|
7924 |
+
},
|
7925 |
+
{
|
7926 |
+
"epoch": 0.02268587014503256,
|
7927 |
+
"grad_norm": 0.001089409808628261,
|
7928 |
+
"learning_rate": 9.004712578616304e-06,
|
7929 |
+
"loss": 46.0,
|
7930 |
+
"step": 1127
|
7931 |
+
},
|
7932 |
+
{
|
7933 |
+
"epoch": 0.02270599957728192,
|
7934 |
+
"grad_norm": 0.0006295640487223864,
|
7935 |
+
"learning_rate": 8.90421907797374e-06,
|
7936 |
+
"loss": 46.0,
|
7937 |
+
"step": 1128
|
7938 |
+
},
|
7939 |
+
{
|
7940 |
+
"epoch": 0.022726129009531287,
|
7941 |
+
"grad_norm": 0.0009275046759285033,
|
7942 |
+
"learning_rate": 8.804263352499864e-06,
|
7943 |
+
"loss": 46.0,
|
7944 |
+
"step": 1129
|
7945 |
+
},
|
7946 |
+
{
|
7947 |
+
"epoch": 0.02274625844178065,
|
7948 |
+
"grad_norm": 0.0012473628157749772,
|
7949 |
+
"learning_rate": 8.70484599227367e-06,
|
7950 |
+
"loss": 46.0,
|
7951 |
+
"step": 1130
|
7952 |
+
},
|
7953 |
+
{
|
7954 |
+
"epoch": 0.022766387874030014,
|
7955 |
+
"grad_norm": 0.0009947452927008271,
|
7956 |
+
"learning_rate": 8.605967584195995e-06,
|
7957 |
+
"loss": 46.0,
|
7958 |
+
"step": 1131
|
7959 |
+
},
|
7960 |
+
{
|
7961 |
+
"epoch": 0.022786517306279376,
|
7962 |
+
"grad_norm": 0.0007231653435155749,
|
7963 |
+
"learning_rate": 8.507628711985983e-06,
|
7964 |
+
"loss": 46.0,
|
7965 |
+
"step": 1132
|
7966 |
+
},
|
7967 |
+
{
|
7968 |
+
"epoch": 0.02280664673852874,
|
7969 |
+
"grad_norm": 0.0004182531265541911,
|
7970 |
+
"learning_rate": 8.409829956177684e-06,
|
7971 |
+
"loss": 46.0,
|
7972 |
+
"step": 1133
|
7973 |
+
},
|
7974 |
+
{
|
7975 |
+
"epoch": 0.022826776170778103,
|
7976 |
+
"grad_norm": 0.0012393246870487928,
|
7977 |
+
"learning_rate": 8.312571894116649e-06,
|
7978 |
+
"loss": 46.0,
|
7979 |
+
"step": 1134
|
7980 |
+
},
|
7981 |
+
{
|
7982 |
+
"epoch": 0.022846905603027465,
|
7983 |
+
"grad_norm": 0.0009794539073482156,
|
7984 |
+
"learning_rate": 8.215855099956472e-06,
|
7985 |
+
"loss": 46.0,
|
7986 |
+
"step": 1135
|
7987 |
+
},
|
7988 |
+
{
|
7989 |
+
"epoch": 0.02286703503527683,
|
7990 |
+
"grad_norm": 0.000817556690890342,
|
7991 |
+
"learning_rate": 8.119680144655428e-06,
|
7992 |
+
"loss": 46.0,
|
7993 |
+
"step": 1136
|
7994 |
+
},
|
7995 |
+
{
|
7996 |
+
"epoch": 0.022887164467526193,
|
7997 |
+
"grad_norm": 0.001215306343510747,
|
7998 |
+
"learning_rate": 8.024047595973095e-06,
|
7999 |
+
"loss": 46.0,
|
8000 |
+
"step": 1137
|
8001 |
+
},
|
8002 |
+
{
|
8003 |
+
"epoch": 0.02290729389977556,
|
8004 |
+
"grad_norm": 0.0013395050773397088,
|
8005 |
+
"learning_rate": 7.92895801846707e-06,
|
8006 |
+
"loss": 46.0,
|
8007 |
+
"step": 1138
|
8008 |
+
},
|
8009 |
+
{
|
8010 |
+
"epoch": 0.02292742333202492,
|
8011 |
+
"grad_norm": 0.0012201687786728144,
|
8012 |
+
"learning_rate": 7.834411973489419e-06,
|
8013 |
+
"loss": 46.0,
|
8014 |
+
"step": 1139
|
8015 |
+
},
|
8016 |
+
{
|
8017 |
+
"epoch": 0.022947552764274282,
|
8018 |
+
"grad_norm": 0.0008355136960744858,
|
8019 |
+
"learning_rate": 7.740410019183697e-06,
|
8020 |
+
"loss": 46.0,
|
8021 |
+
"step": 1140
|
8022 |
+
},
|
8023 |
+
{
|
8024 |
+
"epoch": 0.022967682196523648,
|
8025 |
+
"grad_norm": 0.0009616176830604672,
|
8026 |
+
"learning_rate": 7.646952710481336e-06,
|
8027 |
+
"loss": 46.0,
|
8028 |
+
"step": 1141
|
8029 |
+
},
|
8030 |
+
{
|
8031 |
+
"epoch": 0.02298781162877301,
|
8032 |
+
"grad_norm": 0.000676050316542387,
|
8033 |
+
"learning_rate": 7.554040599098588e-06,
|
8034 |
+
"loss": 46.0,
|
8035 |
+
"step": 1142
|
8036 |
+
},
|
8037 |
+
{
|
8038 |
+
"epoch": 0.023007941061022375,
|
8039 |
+
"grad_norm": 0.0018640294438228011,
|
8040 |
+
"learning_rate": 7.461674233533123e-06,
|
8041 |
+
"loss": 46.0,
|
8042 |
+
"step": 1143
|
8043 |
+
},
|
8044 |
+
{
|
8045 |
+
"epoch": 0.023028070493271737,
|
8046 |
+
"grad_norm": 0.0013213737402111292,
|
8047 |
+
"learning_rate": 7.369854159060929e-06,
|
8048 |
+
"loss": 46.0,
|
8049 |
+
"step": 1144
|
8050 |
+
},
|
8051 |
+
{
|
8052 |
+
"epoch": 0.0230481999255211,
|
8053 |
+
"grad_norm": 0.001049902755767107,
|
8054 |
+
"learning_rate": 7.278580917732913e-06,
|
8055 |
+
"loss": 46.0,
|
8056 |
+
"step": 1145
|
8057 |
+
},
|
8058 |
+
{
|
8059 |
+
"epoch": 0.023068329357770465,
|
8060 |
+
"grad_norm": 0.0006768841994926333,
|
8061 |
+
"learning_rate": 7.187855048371917e-06,
|
8062 |
+
"loss": 46.0,
|
8063 |
+
"step": 1146
|
8064 |
+
},
|
8065 |
+
{
|
8066 |
+
"epoch": 0.023088458790019827,
|
8067 |
+
"grad_norm": 0.0010121595114469528,
|
8068 |
+
"learning_rate": 7.097677086569343e-06,
|
8069 |
+
"loss": 46.0,
|
8070 |
+
"step": 1147
|
8071 |
+
},
|
8072 |
+
{
|
8073 |
+
"epoch": 0.023108588222269192,
|
8074 |
+
"grad_norm": 0.001295650377869606,
|
8075 |
+
"learning_rate": 7.008047564682119e-06,
|
8076 |
+
"loss": 46.0,
|
8077 |
+
"step": 1148
|
8078 |
+
},
|
8079 |
+
{
|
8080 |
+
"epoch": 0.023128717654518554,
|
8081 |
+
"grad_norm": 0.0004917937330901623,
|
8082 |
+
"learning_rate": 6.91896701182948e-06,
|
8083 |
+
"loss": 46.0,
|
8084 |
+
"step": 1149
|
8085 |
+
},
|
8086 |
+
{
|
8087 |
+
"epoch": 0.02314884708676792,
|
8088 |
+
"grad_norm": 0.0013024702202528715,
|
8089 |
+
"learning_rate": 6.83043595388988e-06,
|
8090 |
+
"loss": 46.0,
|
8091 |
+
"step": 1150
|
8092 |
+
},
|
8093 |
+
{
|
8094 |
+
"epoch": 0.02316897651901728,
|
8095 |
+
"grad_norm": 0.0009011050569824874,
|
8096 |
+
"learning_rate": 6.742454913497942e-06,
|
8097 |
+
"loss": 46.0,
|
8098 |
+
"step": 1151
|
8099 |
+
},
|
8100 |
+
{
|
8101 |
+
"epoch": 0.023189105951266643,
|
8102 |
+
"grad_norm": 0.0012644693488255143,
|
8103 |
+
"learning_rate": 6.6550244100412e-06,
|
8104 |
+
"loss": 46.0,
|
8105 |
+
"step": 1152
|
8106 |
+
},
|
8107 |
+
{
|
8108 |
+
"epoch": 0.02320923538351601,
|
8109 |
+
"grad_norm": 0.0013667414896190166,
|
8110 |
+
"learning_rate": 6.568144959657263e-06,
|
8111 |
+
"loss": 46.0,
|
8112 |
+
"step": 1153
|
8113 |
+
},
|
8114 |
+
{
|
8115 |
+
"epoch": 0.02322936481576537,
|
8116 |
+
"grad_norm": 0.0011150363134220243,
|
8117 |
+
"learning_rate": 6.481817075230567e-06,
|
8118 |
+
"loss": 46.0,
|
8119 |
+
"step": 1154
|
8120 |
+
},
|
8121 |
+
{
|
8122 |
+
"epoch": 0.023249494248014736,
|
8123 |
+
"grad_norm": 0.0010298212291672826,
|
8124 |
+
"learning_rate": 6.396041266389474e-06,
|
8125 |
+
"loss": 46.0,
|
8126 |
+
"step": 1155
|
8127 |
+
},
|
8128 |
+
{
|
8129 |
+
"epoch": 0.023269623680264098,
|
8130 |
+
"grad_norm": 0.001109607401303947,
|
8131 |
+
"learning_rate": 6.3108180395031965e-06,
|
8132 |
+
"loss": 46.0,
|
8133 |
+
"step": 1156
|
8134 |
+
},
|
8135 |
+
{
|
8136 |
+
"epoch": 0.02328975311251346,
|
8137 |
+
"grad_norm": 0.0005489352042786777,
|
8138 |
+
"learning_rate": 6.22614789767888e-06,
|
8139 |
+
"loss": 46.0,
|
8140 |
+
"step": 1157
|
8141 |
+
},
|
8142 |
+
{
|
8143 |
+
"epoch": 0.023309882544762826,
|
8144 |
+
"grad_norm": 0.0007645548903383315,
|
8145 |
+
"learning_rate": 6.142031340758525e-06,
|
8146 |
+
"loss": 46.0,
|
8147 |
+
"step": 1158
|
8148 |
+
},
|
8149 |
+
{
|
8150 |
+
"epoch": 0.023330011977012188,
|
8151 |
+
"grad_norm": 0.0007147680153138936,
|
8152 |
+
"learning_rate": 6.058468865316102e-06,
|
8153 |
+
"loss": 46.0,
|
8154 |
+
"step": 1159
|
8155 |
+
},
|
8156 |
+
{
|
8157 |
+
"epoch": 0.023350141409261553,
|
8158 |
+
"grad_norm": 0.0006330362521111965,
|
8159 |
+
"learning_rate": 5.975460964654689e-06,
|
8160 |
+
"loss": 46.0,
|
8161 |
+
"step": 1160
|
8162 |
+
},
|
8163 |
+
{
|
8164 |
+
"epoch": 0.023370270841510915,
|
8165 |
+
"grad_norm": 0.0008396542398259044,
|
8166 |
+
"learning_rate": 5.8930081288034014e-06,
|
8167 |
+
"loss": 46.0,
|
8168 |
+
"step": 1161
|
8169 |
+
},
|
8170 |
+
{
|
8171 |
+
"epoch": 0.023390400273760277,
|
8172 |
+
"grad_norm": 0.001648742356337607,
|
8173 |
+
"learning_rate": 5.8111108445146116e-06,
|
8174 |
+
"loss": 46.0,
|
8175 |
+
"step": 1162
|
8176 |
+
},
|
8177 |
+
{
|
8178 |
+
"epoch": 0.023410529706009642,
|
8179 |
+
"grad_norm": 0.0008351008291356266,
|
8180 |
+
"learning_rate": 5.72976959526107e-06,
|
8181 |
+
"loss": 46.0,
|
8182 |
+
"step": 1163
|
8183 |
+
},
|
8184 |
+
{
|
8185 |
+
"epoch": 0.023430659138259004,
|
8186 |
+
"grad_norm": 0.0012098524020984769,
|
8187 |
+
"learning_rate": 5.648984861232986e-06,
|
8188 |
+
"loss": 46.0,
|
8189 |
+
"step": 1164
|
8190 |
+
},
|
8191 |
+
{
|
8192 |
+
"epoch": 0.02345078857050837,
|
8193 |
+
"grad_norm": 0.0010954445460811257,
|
8194 |
+
"learning_rate": 5.568757119335244e-06,
|
8195 |
+
"loss": 46.0,
|
8196 |
+
"step": 1165
|
8197 |
+
},
|
8198 |
+
{
|
8199 |
+
"epoch": 0.023470918002757732,
|
8200 |
+
"grad_norm": 0.0008584211464039981,
|
8201 |
+
"learning_rate": 5.489086843184632e-06,
|
8202 |
+
"loss": 46.0,
|
8203 |
+
"step": 1166
|
8204 |
+
},
|
8205 |
+
{
|
8206 |
+
"epoch": 0.023491047435007097,
|
8207 |
+
"grad_norm": 0.000767943391110748,
|
8208 |
+
"learning_rate": 5.40997450310693e-06,
|
8209 |
+
"loss": 46.0,
|
8210 |
+
"step": 1167
|
8211 |
+
},
|
8212 |
+
{
|
8213 |
+
"epoch": 0.02351117686725646,
|
8214 |
+
"grad_norm": 0.001566538936458528,
|
8215 |
+
"learning_rate": 5.3314205661342155e-06,
|
8216 |
+
"loss": 46.0,
|
8217 |
+
"step": 1168
|
8218 |
+
},
|
8219 |
+
{
|
8220 |
+
"epoch": 0.02353130629950582,
|
8221 |
+
"grad_norm": 0.0008029242744669318,
|
8222 |
+
"learning_rate": 5.253425496002084e-06,
|
8223 |
+
"loss": 46.0,
|
8224 |
+
"step": 1169
|
8225 |
+
},
|
8226 |
+
{
|
8227 |
+
"epoch": 0.023551435731755187,
|
8228 |
+
"grad_norm": 0.0011748820543289185,
|
8229 |
+
"learning_rate": 5.175989753146948e-06,
|
8230 |
+
"loss": 46.0,
|
8231 |
+
"step": 1170
|
8232 |
+
},
|
8233 |
+
{
|
8234 |
+
"epoch": 0.02357156516400455,
|
8235 |
+
"grad_norm": 0.000719308911357075,
|
8236 |
+
"learning_rate": 5.099113794703225e-06,
|
8237 |
+
"loss": 46.0,
|
8238 |
+
"step": 1171
|
8239 |
+
},
|
8240 |
+
{
|
8241 |
+
"epoch": 0.023591694596253914,
|
8242 |
+
"grad_norm": 0.0005550024216063321,
|
8243 |
+
"learning_rate": 5.0227980745007345e-06,
|
8244 |
+
"loss": 46.0,
|
8245 |
+
"step": 1172
|
8246 |
+
},
|
8247 |
+
{
|
8248 |
+
"epoch": 0.023611824028503276,
|
8249 |
+
"grad_norm": 0.0007492146105505526,
|
8250 |
+
"learning_rate": 4.947043043062016e-06,
|
8251 |
+
"loss": 46.0,
|
8252 |
+
"step": 1173
|
8253 |
+
},
|
8254 |
+
{
|
8255 |
+
"epoch": 0.023631953460752638,
|
8256 |
+
"grad_norm": 0.0011210053926333785,
|
8257 |
+
"learning_rate": 4.87184914759955e-06,
|
8258 |
+
"loss": 46.0,
|
8259 |
+
"step": 1174
|
8260 |
+
},
|
8261 |
+
{
|
8262 |
+
"epoch": 0.023652082893002004,
|
8263 |
+
"grad_norm": 0.000586840498726815,
|
8264 |
+
"learning_rate": 4.7972168320132845e-06,
|
8265 |
+
"loss": 46.0,
|
8266 |
+
"step": 1175
|
8267 |
+
},
|
8268 |
+
{
|
8269 |
+
"epoch": 0.023672212325251366,
|
8270 |
+
"grad_norm": 0.0013588924193754792,
|
8271 |
+
"learning_rate": 4.7231465368879124e-06,
|
8272 |
+
"loss": 46.0,
|
8273 |
+
"step": 1176
|
8274 |
+
},
|
8275 |
+
{
|
8276 |
+
"epoch": 0.02369234175750073,
|
8277 |
+
"grad_norm": 0.00061926303897053,
|
8278 |
+
"learning_rate": 4.649638699490266e-06,
|
8279 |
+
"loss": 46.0,
|
8280 |
+
"step": 1177
|
8281 |
+
},
|
8282 |
+
{
|
8283 |
+
"epoch": 0.023712471189750093,
|
8284 |
+
"grad_norm": 0.0006894692778587341,
|
8285 |
+
"learning_rate": 4.576693753766792e-06,
|
8286 |
+
"loss": 46.0,
|
8287 |
+
"step": 1178
|
8288 |
+
},
|
8289 |
+
{
|
8290 |
+
"epoch": 0.023732600621999455,
|
8291 |
+
"grad_norm": 0.0008896641083993018,
|
8292 |
+
"learning_rate": 4.5043121303409595e-06,
|
8293 |
+
"loss": 46.0,
|
8294 |
+
"step": 1179
|
8295 |
+
},
|
8296 |
+
{
|
8297 |
+
"epoch": 0.02375273005424882,
|
8298 |
+
"grad_norm": 0.0010777831776067615,
|
8299 |
+
"learning_rate": 4.432494256510711e-06,
|
8300 |
+
"loss": 46.0,
|
8301 |
+
"step": 1180
|
8302 |
+
},
|
8303 |
+
{
|
8304 |
+
"epoch": 0.023772859486498182,
|
8305 |
+
"grad_norm": 0.0012661231448873878,
|
8306 |
+
"learning_rate": 4.361240556245938e-06,
|
8307 |
+
"loss": 46.0,
|
8308 |
+
"step": 1181
|
8309 |
+
},
|
8310 |
+
{
|
8311 |
+
"epoch": 0.023792988918747548,
|
8312 |
+
"grad_norm": 0.0015156505396589637,
|
8313 |
+
"learning_rate": 4.290551450185986e-06,
|
8314 |
+
"loss": 46.0,
|
8315 |
+
"step": 1182
|
8316 |
+
},
|
8317 |
+
{
|
8318 |
+
"epoch": 0.02381311835099691,
|
8319 |
+
"grad_norm": 0.000532010046299547,
|
8320 |
+
"learning_rate": 4.220427355637224e-06,
|
8321 |
+
"loss": 46.0,
|
8322 |
+
"step": 1183
|
8323 |
+
},
|
8324 |
+
{
|
8325 |
+
"epoch": 0.023833247783246275,
|
8326 |
+
"grad_norm": 0.0007818634621798992,
|
8327 |
+
"learning_rate": 4.150868686570464e-06,
|
8328 |
+
"loss": 46.0,
|
8329 |
+
"step": 1184
|
8330 |
+
},
|
8331 |
+
{
|
8332 |
+
"epoch": 0.023853377215495637,
|
8333 |
+
"grad_norm": 0.0010528319980949163,
|
8334 |
+
"learning_rate": 4.081875853618588e-06,
|
8335 |
+
"loss": 46.0,
|
8336 |
+
"step": 1185
|
8337 |
+
},
|
8338 |
+
{
|
8339 |
+
"epoch": 0.023873506647745,
|
8340 |
+
"grad_norm": 0.001450626994483173,
|
8341 |
+
"learning_rate": 4.013449264074187e-06,
|
8342 |
+
"loss": 46.0,
|
8343 |
+
"step": 1186
|
8344 |
+
},
|
8345 |
+
{
|
8346 |
+
"epoch": 0.023893636079994365,
|
8347 |
+
"grad_norm": 0.0018947275821119547,
|
8348 |
+
"learning_rate": 3.945589321886989e-06,
|
8349 |
+
"loss": 46.0,
|
8350 |
+
"step": 1187
|
8351 |
+
},
|
8352 |
+
{
|
8353 |
+
"epoch": 0.023913765512243727,
|
8354 |
+
"grad_norm": 0.0009489529766142368,
|
8355 |
+
"learning_rate": 3.878296427661676e-06,
|
8356 |
+
"loss": 46.0,
|
8357 |
+
"step": 1188
|
8358 |
+
},
|
8359 |
+
{
|
8360 |
+
"epoch": 0.023933894944493092,
|
8361 |
+
"grad_norm": 0.0008835981134325266,
|
8362 |
+
"learning_rate": 3.8115709786553435e-06,
|
8363 |
+
"loss": 46.0,
|
8364 |
+
"step": 1189
|
8365 |
+
},
|
8366 |
+
{
|
8367 |
+
"epoch": 0.023954024376742454,
|
8368 |
+
"grad_norm": 0.0008584621245972812,
|
8369 |
+
"learning_rate": 3.7454133687752524e-06,
|
8370 |
+
"loss": 46.0,
|
8371 |
+
"step": 1190
|
8372 |
+
},
|
8373 |
+
{
|
8374 |
+
"epoch": 0.023974153808991816,
|
8375 |
+
"grad_norm": 0.001248899381607771,
|
8376 |
+
"learning_rate": 3.6798239885764806e-06,
|
8377 |
+
"loss": 46.0,
|
8378 |
+
"step": 1191
|
8379 |
+
},
|
8380 |
+
{
|
8381 |
+
"epoch": 0.02399428324124118,
|
8382 |
+
"grad_norm": 0.0006818815018050373,
|
8383 |
+
"learning_rate": 3.614803225259622e-06,
|
8384 |
+
"loss": 46.0,
|
8385 |
+
"step": 1192
|
8386 |
+
},
|
8387 |
+
{
|
8388 |
+
"epoch": 0.024014412673490543,
|
8389 |
+
"grad_norm": 0.0011521173873916268,
|
8390 |
+
"learning_rate": 3.550351462668489e-06,
|
8391 |
+
"loss": 46.0,
|
8392 |
+
"step": 1193
|
8393 |
+
},
|
8394 |
+
{
|
8395 |
+
"epoch": 0.02403454210573991,
|
8396 |
+
"grad_norm": 0.0009178062318824232,
|
8397 |
+
"learning_rate": 3.4864690812878688e-06,
|
8398 |
+
"loss": 46.0,
|
8399 |
+
"step": 1194
|
8400 |
+
},
|
8401 |
+
{
|
8402 |
+
"epoch": 0.02405467153798927,
|
8403 |
+
"grad_norm": 0.0005374921602196991,
|
8404 |
+
"learning_rate": 3.4231564582412167e-06,
|
8405 |
+
"loss": 46.0,
|
8406 |
+
"step": 1195
|
8407 |
+
},
|
8408 |
+
{
|
8409 |
+
"epoch": 0.024074800970238633,
|
8410 |
+
"grad_norm": 0.0010567499557510018,
|
8411 |
+
"learning_rate": 3.3604139672885227e-06,
|
8412 |
+
"loss": 46.0,
|
8413 |
+
"step": 1196
|
8414 |
+
},
|
8415 |
+
{
|
8416 |
+
"epoch": 0.024094930402488,
|
8417 |
+
"grad_norm": 0.00042650941759347916,
|
8418 |
+
"learning_rate": 3.298241978824046e-06,
|
8419 |
+
"loss": 46.0,
|
8420 |
+
"step": 1197
|
8421 |
+
},
|
8422 |
+
{
|
8423 |
+
"epoch": 0.02411505983473736,
|
8424 |
+
"grad_norm": 0.001136740087531507,
|
8425 |
+
"learning_rate": 3.2366408598741072e-06,
|
8426 |
+
"loss": 46.0,
|
8427 |
+
"step": 1198
|
8428 |
+
},
|
8429 |
+
{
|
8430 |
+
"epoch": 0.024135189266986726,
|
8431 |
+
"grad_norm": 0.0011373634915798903,
|
8432 |
+
"learning_rate": 3.175610974095011e-06,
|
8433 |
+
"loss": 46.0,
|
8434 |
+
"step": 1199
|
8435 |
+
},
|
8436 |
+
{
|
8437 |
+
"epoch": 0.024155318699236088,
|
8438 |
+
"grad_norm": 0.00101361027918756,
|
8439 |
+
"learning_rate": 3.115152681770783e-06,
|
8440 |
+
"loss": 46.0,
|
8441 |
+
"step": 1200
|
8442 |
+
},
|
8443 |
+
{
|
8444 |
+
"epoch": 0.024175448131485453,
|
8445 |
+
"grad_norm": 0.000945060164667666,
|
8446 |
+
"learning_rate": 3.055266339811147e-06,
|
8447 |
+
"loss": 46.0,
|
8448 |
+
"step": 1201
|
8449 |
+
},
|
8450 |
+
{
|
8451 |
+
"epoch": 0.024195577563734815,
|
8452 |
+
"grad_norm": 0.0009245016844943166,
|
8453 |
+
"learning_rate": 2.9959523017493386e-06,
|
8454 |
+
"loss": 46.0,
|
8455 |
+
"step": 1202
|
8456 |
+
},
|
8457 |
+
{
|
8458 |
+
"epoch": 0.024215706995984177,
|
8459 |
+
"grad_norm": 0.00034907733788713813,
|
8460 |
+
"learning_rate": 2.9372109177400854e-06,
|
8461 |
+
"loss": 46.0,
|
8462 |
+
"step": 1203
|
8463 |
+
},
|
8464 |
+
{
|
8465 |
+
"epoch": 0.024235836428233543,
|
8466 |
+
"grad_norm": 0.0015949602238833904,
|
8467 |
+
"learning_rate": 2.8790425345574745e-06,
|
8468 |
+
"loss": 46.0,
|
8469 |
+
"step": 1204
|
8470 |
+
},
|
8471 |
+
{
|
8472 |
+
"epoch": 0.024255965860482905,
|
8473 |
+
"grad_norm": 0.0008049356401897967,
|
8474 |
+
"learning_rate": 2.821447495592977e-06,
|
8475 |
+
"loss": 46.0,
|
8476 |
+
"step": 1205
|
8477 |
+
},
|
8478 |
+
{
|
8479 |
+
"epoch": 0.02427609529273227,
|
8480 |
+
"grad_norm": 0.0010959411738440394,
|
8481 |
+
"learning_rate": 2.7644261408533155e-06,
|
8482 |
+
"loss": 46.0,
|
8483 |
+
"step": 1206
|
8484 |
+
},
|
8485 |
+
{
|
8486 |
+
"epoch": 0.024296224724981632,
|
8487 |
+
"grad_norm": 0.001113040023483336,
|
8488 |
+
"learning_rate": 2.707978806958611e-06,
|
8489 |
+
"loss": 46.0,
|
8490 |
+
"step": 1207
|
8491 |
+
},
|
8492 |
+
{
|
8493 |
+
"epoch": 0.024316354157230994,
|
8494 |
+
"grad_norm": 0.0006179322372190654,
|
8495 |
+
"learning_rate": 2.6521058271402386e-06,
|
8496 |
+
"loss": 46.0,
|
8497 |
+
"step": 1208
|
8498 |
+
},
|
8499 |
+
{
|
8500 |
+
"epoch": 0.02433648358948036,
|
8501 |
+
"grad_norm": 0.0004471320426091552,
|
8502 |
+
"learning_rate": 2.596807531238965e-06,
|
8503 |
+
"loss": 46.0,
|
8504 |
+
"step": 1209
|
8505 |
+
},
|
8506 |
+
{
|
8507 |
+
"epoch": 0.02435661302172972,
|
8508 |
+
"grad_norm": 0.002076149685308337,
|
8509 |
+
"learning_rate": 2.542084245702947e-06,
|
8510 |
+
"loss": 46.0,
|
8511 |
+
"step": 1210
|
8512 |
+
},
|
8513 |
+
{
|
8514 |
+
"epoch": 0.024376742453979087,
|
8515 |
+
"grad_norm": 0.0005335964378900826,
|
8516 |
+
"learning_rate": 2.487936293585813e-06,
|
8517 |
+
"loss": 46.0,
|
8518 |
+
"step": 1211
|
8519 |
+
},
|
8520 |
+
{
|
8521 |
+
"epoch": 0.02439687188622845,
|
8522 |
+
"grad_norm": 0.001850920612923801,
|
8523 |
+
"learning_rate": 2.4343639945448306e-06,
|
8524 |
+
"loss": 46.0,
|
8525 |
+
"step": 1212
|
8526 |
+
},
|
8527 |
+
{
|
8528 |
+
"epoch": 0.02441700131847781,
|
8529 |
+
"grad_norm": 0.0012666831025853753,
|
8530 |
+
"learning_rate": 2.3813676648388517e-06,
|
8531 |
+
"loss": 46.0,
|
8532 |
+
"step": 1213
|
8533 |
+
},
|
8534 |
+
{
|
8535 |
+
"epoch": 0.024437130750727176,
|
8536 |
+
"grad_norm": 0.00046365856542252004,
|
8537 |
+
"learning_rate": 2.3289476173266376e-06,
|
8538 |
+
"loss": 46.0,
|
8539 |
+
"step": 1214
|
8540 |
+
},
|
8541 |
+
{
|
8542 |
+
"epoch": 0.024457260182976538,
|
8543 |
+
"grad_norm": 0.0010178579250350595,
|
8544 |
+
"learning_rate": 2.2771041614648825e-06,
|
8545 |
+
"loss": 46.0,
|
8546 |
+
"step": 1215
|
8547 |
+
},
|
8548 |
+
{
|
8549 |
+
"epoch": 0.024477389615225904,
|
8550 |
+
"grad_norm": 0.001140277599915862,
|
8551 |
+
"learning_rate": 2.2258376033064354e-06,
|
8552 |
+
"loss": 46.0,
|
8553 |
+
"step": 1216
|
8554 |
+
},
|
8555 |
+
{
|
8556 |
+
"epoch": 0.024497519047475266,
|
8557 |
+
"grad_norm": 0.0006400442798621953,
|
8558 |
+
"learning_rate": 2.1751482454984706e-06,
|
8559 |
+
"loss": 46.0,
|
8560 |
+
"step": 1217
|
8561 |
+
},
|
8562 |
+
{
|
8563 |
+
"epoch": 0.02451764847972463,
|
8564 |
+
"grad_norm": 0.0014165055472403765,
|
8565 |
+
"learning_rate": 2.1250363872807655e-06,
|
8566 |
+
"loss": 46.0,
|
8567 |
+
"step": 1218
|
8568 |
+
},
|
8569 |
+
{
|
8570 |
+
"epoch": 0.024537777911973993,
|
8571 |
+
"grad_norm": 0.0007371717365458608,
|
8572 |
+
"learning_rate": 2.0755023244838136e-06,
|
8573 |
+
"loss": 46.0,
|
8574 |
+
"step": 1219
|
8575 |
+
},
|
8576 |
+
{
|
8577 |
+
"epoch": 0.024557907344223355,
|
8578 |
+
"grad_norm": 0.0013694085646420717,
|
8579 |
+
"learning_rate": 2.026546349527181e-06,
|
8580 |
+
"loss": 46.0,
|
8581 |
+
"step": 1220
|
8582 |
+
},
|
8583 |
+
{
|
8584 |
+
"epoch": 0.02457803677647272,
|
8585 |
+
"grad_norm": 0.0012485695770010352,
|
8586 |
+
"learning_rate": 1.978168751417786e-06,
|
8587 |
+
"loss": 46.0,
|
8588 |
+
"step": 1221
|
8589 |
+
},
|
8590 |
+
{
|
8591 |
+
"epoch": 0.024598166208722082,
|
8592 |
+
"grad_norm": 0.001493003685027361,
|
8593 |
+
"learning_rate": 1.93036981574809e-06,
|
8594 |
+
"loss": 46.0,
|
8595 |
+
"step": 1222
|
8596 |
+
},
|
8597 |
+
{
|
8598 |
+
"epoch": 0.024618295640971448,
|
8599 |
+
"grad_norm": 0.0006664522225037217,
|
8600 |
+
"learning_rate": 1.8831498246945189e-06,
|
8601 |
+
"loss": 46.0,
|
8602 |
+
"step": 1223
|
8603 |
+
},
|
8604 |
+
{
|
8605 |
+
"epoch": 0.02463842507322081,
|
8606 |
+
"grad_norm": 0.0011530322954058647,
|
8607 |
+
"learning_rate": 1.836509057015734e-06,
|
8608 |
+
"loss": 46.0,
|
8609 |
+
"step": 1224
|
8610 |
+
},
|
8611 |
+
{
|
8612 |
+
"epoch": 0.024658554505470172,
|
8613 |
+
"grad_norm": 0.0006030354998074472,
|
8614 |
+
"learning_rate": 1.7904477880510307e-06,
|
8615 |
+
"loss": 46.0,
|
8616 |
+
"step": 1225
|
8617 |
+
},
|
8618 |
+
{
|
8619 |
+
"epoch": 0.024678683937719537,
|
8620 |
+
"grad_norm": 0.001384797622449696,
|
8621 |
+
"learning_rate": 1.7449662897186414e-06,
|
8622 |
+
"loss": 46.0,
|
8623 |
+
"step": 1226
|
8624 |
+
},
|
8625 |
+
{
|
8626 |
+
"epoch": 0.0246988133699689,
|
8627 |
+
"grad_norm": 0.0006306437426246703,
|
8628 |
+
"learning_rate": 1.7000648305142364e-06,
|
8629 |
+
"loss": 46.0,
|
8630 |
+
"step": 1227
|
8631 |
+
},
|
8632 |
+
{
|
8633 |
+
"epoch": 0.024718942802218265,
|
8634 |
+
"grad_norm": 0.0005167327472008765,
|
8635 |
+
"learning_rate": 1.655743675509258e-06,
|
8636 |
+
"loss": 46.0,
|
8637 |
+
"step": 1228
|
8638 |
+
},
|
8639 |
+
{
|
8640 |
+
"epoch": 0.024739072234467627,
|
8641 |
+
"grad_norm": 0.00042707190732471645,
|
8642 |
+
"learning_rate": 1.6120030863493674e-06,
|
8643 |
+
"loss": 46.0,
|
8644 |
+
"step": 1229
|
8645 |
+
},
|
8646 |
+
{
|
8647 |
+
"epoch": 0.02475920166671699,
|
8648 |
+
"grad_norm": 0.0003029134531971067,
|
8649 |
+
"learning_rate": 1.5688433212529107e-06,
|
8650 |
+
"loss": 46.0,
|
8651 |
+
"step": 1230
|
8652 |
+
},
|
8653 |
+
{
|
8654 |
+
"epoch": 0.024779331098966354,
|
8655 |
+
"grad_norm": 0.0007286612526513636,
|
8656 |
+
"learning_rate": 1.5262646350094334e-06,
|
8657 |
+
"loss": 46.0,
|
8658 |
+
"step": 1231
|
8659 |
+
},
|
8660 |
+
{
|
8661 |
+
"epoch": 0.024799460531215716,
|
8662 |
+
"grad_norm": 0.0004967825370840728,
|
8663 |
+
"learning_rate": 1.4842672789780798e-06,
|
8664 |
+
"loss": 46.0,
|
8665 |
+
"step": 1232
|
8666 |
+
},
|
8667 |
+
{
|
8668 |
+
"epoch": 0.02481958996346508,
|
8669 |
+
"grad_norm": 0.0014918498927727342,
|
8670 |
+
"learning_rate": 1.4428515010861955e-06,
|
8671 |
+
"loss": 46.0,
|
8672 |
+
"step": 1233
|
8673 |
+
},
|
8674 |
+
{
|
8675 |
+
"epoch": 0.024839719395714444,
|
8676 |
+
"grad_norm": 0.0008161486475728452,
|
8677 |
+
"learning_rate": 1.4020175458278607e-06,
|
8678 |
+
"loss": 46.0,
|
8679 |
+
"step": 1234
|
8680 |
+
},
|
8681 |
+
{
|
8682 |
+
"epoch": 0.02485984882796381,
|
8683 |
+
"grad_norm": 0.0008656067657284439,
|
8684 |
+
"learning_rate": 1.3617656542623813e-06,
|
8685 |
+
"loss": 46.0,
|
8686 |
+
"step": 1235
|
8687 |
+
},
|
8688 |
+
{
|
8689 |
+
"epoch": 0.02487997826021317,
|
8690 |
+
"grad_norm": 0.0005301354103721678,
|
8691 |
+
"learning_rate": 1.322096064012912e-06,
|
8692 |
+
"loss": 46.0,
|
8693 |
+
"step": 1236
|
8694 |
+
},
|
8695 |
+
{
|
8696 |
+
"epoch": 0.024900107692462533,
|
8697 |
+
"grad_norm": 0.0011264794738963246,
|
8698 |
+
"learning_rate": 1.2830090092650904e-06,
|
8699 |
+
"loss": 46.0,
|
8700 |
+
"step": 1237
|
8701 |
+
},
|
8702 |
+
{
|
8703 |
+
"epoch": 0.0249202371247119,
|
8704 |
+
"grad_norm": 0.000983362435363233,
|
8705 |
+
"learning_rate": 1.244504720765549e-06,
|
8706 |
+
"loss": 46.0,
|
8707 |
+
"step": 1238
|
8708 |
+
},
|
8709 |
+
{
|
8710 |
+
"epoch": 0.02494036655696126,
|
8711 |
+
"grad_norm": 0.0011257297592237592,
|
8712 |
+
"learning_rate": 1.20658342582064e-06,
|
8713 |
+
"loss": 46.0,
|
8714 |
+
"step": 1239
|
8715 |
+
},
|
8716 |
+
{
|
8717 |
+
"epoch": 0.024960495989210626,
|
8718 |
+
"grad_norm": 0.0009586882661096752,
|
8719 |
+
"learning_rate": 1.1692453482951115e-06,
|
8720 |
+
"loss": 46.0,
|
8721 |
+
"step": 1240
|
8722 |
+
},
|
8723 |
+
{
|
8724 |
+
"epoch": 0.024980625421459988,
|
8725 |
+
"grad_norm": 0.0005041824770160019,
|
8726 |
+
"learning_rate": 1.1324907086106895e-06,
|
8727 |
+
"loss": 46.0,
|
8728 |
+
"step": 1241
|
8729 |
+
},
|
8730 |
+
{
|
8731 |
+
"epoch": 0.02500075485370935,
|
8732 |
+
"grad_norm": 0.0012379593681544065,
|
8733 |
+
"learning_rate": 1.096319723744843e-06,
|
8734 |
+
"loss": 46.0,
|
8735 |
+
"step": 1242
|
8736 |
+
},
|
8737 |
+
{
|
8738 |
+
"epoch": 0.025020884285958715,
|
8739 |
+
"grad_norm": 0.0007608251180499792,
|
8740 |
+
"learning_rate": 1.0607326072295087e-06,
|
8741 |
+
"loss": 46.0,
|
8742 |
+
"step": 1243
|
8743 |
+
},
|
8744 |
+
{
|
8745 |
+
"epoch": 0.025041013718208077,
|
8746 |
+
"grad_norm": 0.001463228720240295,
|
8747 |
+
"learning_rate": 1.0257295691497914e-06,
|
8748 |
+
"loss": 46.0,
|
8749 |
+
"step": 1244
|
8750 |
+
},
|
8751 |
+
{
|
8752 |
+
"epoch": 0.025061143150457443,
|
8753 |
+
"grad_norm": 0.0005632633110508323,
|
8754 |
+
"learning_rate": 9.913108161427543e-07,
|
8755 |
+
"loss": 46.0,
|
8756 |
+
"step": 1245
|
8757 |
+
},
|
8758 |
+
{
|
8759 |
+
"epoch": 0.025081272582706805,
|
8760 |
+
"grad_norm": 0.0013915124582126737,
|
8761 |
+
"learning_rate": 9.57476551396197e-07,
|
8762 |
+
"loss": 46.0,
|
8763 |
+
"step": 1246
|
8764 |
+
},
|
8765 |
+
{
|
8766 |
+
"epoch": 0.025101402014956167,
|
8767 |
+
"grad_norm": 0.0010674886871129274,
|
8768 |
+
"learning_rate": 9.242269746474575e-07,
|
8769 |
+
"loss": 46.0,
|
8770 |
+
"step": 1247
|
8771 |
+
},
|
8772 |
+
{
|
8773 |
+
"epoch": 0.025121531447205532,
|
8774 |
+
"grad_norm": 0.0008928573224693537,
|
8775 |
+
"learning_rate": 8.915622821821789e-07,
|
8776 |
+
"loss": 46.0,
|
8777 |
+
"step": 1248
|
8778 |
+
},
|
8779 |
+
{
|
8780 |
+
"epoch": 0.025141660879454894,
|
8781 |
+
"grad_norm": 0.0014497325755655766,
|
8782 |
+
"learning_rate": 8.594826668332445e-07,
|
8783 |
+
"loss": 46.0,
|
8784 |
+
"step": 1249
|
8785 |
+
},
|
8786 |
+
{
|
8787 |
+
"epoch": 0.02516179031170426,
|
8788 |
+
"grad_norm": 0.0011807921109721065,
|
8789 |
+
"learning_rate": 8.279883179795666e-07,
|
8790 |
+
"loss": 46.0,
|
8791 |
+
"step": 1250
|
8792 |
+
},
|
8793 |
+
{
|
8794 |
+
"epoch": 0.02518191974395362,
|
8795 |
+
"grad_norm": 0.001125653157941997,
|
8796 |
+
"learning_rate": 7.970794215450106e-07,
|
8797 |
+
"loss": 46.0,
|
8798 |
+
"step": 1251
|
8799 |
+
},
|
8800 |
+
{
|
8801 |
+
"epoch": 0.025202049176202987,
|
8802 |
+
"grad_norm": 0.0013557058991864324,
|
8803 |
+
"learning_rate": 7.667561599972505e-07,
|
8804 |
+
"loss": 46.0,
|
8805 |
+
"step": 1252
|
8806 |
+
},
|
8807 |
+
{
|
8808 |
+
"epoch": 0.02522217860845235,
|
8809 |
+
"grad_norm": 0.0003978684253524989,
|
8810 |
+
"learning_rate": 7.370187123467708e-07,
|
8811 |
+
"loss": 46.0,
|
8812 |
+
"step": 1253
|
8813 |
+
},
|
8814 |
+
{
|
8815 |
+
"epoch": 0.02524230804070171,
|
8816 |
+
"grad_norm": 0.0008096559904515743,
|
8817 |
+
"learning_rate": 7.078672541456999e-07,
|
8818 |
+
"loss": 46.0,
|
8819 |
+
"step": 1254
|
8820 |
+
},
|
8821 |
+
{
|
8822 |
+
"epoch": 0.025262437472951076,
|
8823 |
+
"grad_norm": 0.0015750976745039225,
|
8824 |
+
"learning_rate": 6.793019574868775e-07,
|
8825 |
+
"loss": 46.0,
|
8826 |
+
"step": 1255
|
8827 |
+
},
|
8828 |
+
{
|
8829 |
+
"epoch": 0.02528256690520044,
|
8830 |
+
"grad_norm": 0.000689225154928863,
|
8831 |
+
"learning_rate": 6.513229910027896e-07,
|
8832 |
+
"loss": 46.0,
|
8833 |
+
"step": 1256
|
8834 |
+
},
|
8835 |
+
{
|
8836 |
+
"epoch": 0.025302696337449804,
|
8837 |
+
"grad_norm": 0.0008678320446051657,
|
8838 |
+
"learning_rate": 6.239305198645462e-07,
|
8839 |
+
"loss": 46.0,
|
8840 |
+
"step": 1257
|
8841 |
+
},
|
8842 |
+
{
|
8843 |
+
"epoch": 0.025322825769699166,
|
8844 |
+
"grad_norm": 0.001171753858216107,
|
8845 |
+
"learning_rate": 5.971247057809826e-07,
|
8846 |
+
"loss": 46.0,
|
8847 |
+
"step": 1258
|
8848 |
+
},
|
8849 |
+
{
|
8850 |
+
"epoch": 0.025342955201948528,
|
8851 |
+
"grad_norm": 0.000669551664032042,
|
8852 |
+
"learning_rate": 5.709057069976265e-07,
|
8853 |
+
"loss": 46.0,
|
8854 |
+
"step": 1259
|
8855 |
+
},
|
8856 |
+
{
|
8857 |
+
"epoch": 0.025363084634197893,
|
8858 |
+
"grad_norm": 0.00028675812063738704,
|
8859 |
+
"learning_rate": 5.452736782958323e-07,
|
8860 |
+
"loss": 46.0,
|
8861 |
+
"step": 1260
|
8862 |
+
},
|
8863 |
+
{
|
8864 |
+
"epoch": 0.025383214066447255,
|
8865 |
+
"grad_norm": 0.0010722498409450054,
|
8866 |
+
"learning_rate": 5.20228770991793e-07,
|
8867 |
+
"loss": 46.0,
|
8868 |
+
"step": 1261
|
8869 |
+
},
|
8870 |
+
{
|
8871 |
+
"epoch": 0.02540334349869662,
|
8872 |
+
"grad_norm": 0.0011695049470290542,
|
8873 |
+
"learning_rate": 4.957711329357073e-07,
|
8874 |
+
"loss": 46.0,
|
8875 |
+
"step": 1262
|
8876 |
+
},
|
8877 |
+
{
|
8878 |
+
"epoch": 0.025423472930945983,
|
8879 |
+
"grad_norm": 0.0005382261006161571,
|
8880 |
+
"learning_rate": 4.7190090851090274e-07,
|
8881 |
+
"loss": 46.0,
|
8882 |
+
"step": 1263
|
8883 |
+
},
|
8884 |
+
{
|
8885 |
+
"epoch": 0.025443602363195345,
|
8886 |
+
"grad_norm": 0.0011571204522624612,
|
8887 |
+
"learning_rate": 4.4861823863292516e-07,
|
8888 |
+
"loss": 46.0,
|
8889 |
+
"step": 1264
|
8890 |
+
},
|
8891 |
+
{
|
8892 |
+
"epoch": 0.02546373179544471,
|
8893 |
+
"grad_norm": 0.0008085128501988947,
|
8894 |
+
"learning_rate": 4.259232607487951e-07,
|
8895 |
+
"loss": 46.0,
|
8896 |
+
"step": 1265
|
8897 |
+
},
|
8898 |
+
{
|
8899 |
+
"epoch": 0.025483861227694072,
|
8900 |
+
"grad_norm": 0.0012183074140921235,
|
8901 |
+
"learning_rate": 4.038161088361192e-07,
|
8902 |
+
"loss": 46.0,
|
8903 |
+
"step": 1266
|
8904 |
+
},
|
8905 |
+
{
|
8906 |
+
"epoch": 0.025503990659943437,
|
8907 |
+
"grad_norm": 0.0012540913885459304,
|
8908 |
+
"learning_rate": 3.8229691340234684e-07,
|
8909 |
+
"loss": 46.0,
|
8910 |
+
"step": 1267
|
8911 |
+
},
|
8912 |
+
{
|
8913 |
+
"epoch": 0.0255241200921928,
|
8914 |
+
"grad_norm": 0.0005392631865106523,
|
8915 |
+
"learning_rate": 3.613658014839594e-07,
|
8916 |
+
"loss": 46.0,
|
8917 |
+
"step": 1268
|
8918 |
+
},
|
8919 |
+
{
|
8920 |
+
"epoch": 0.025544249524442165,
|
8921 |
+
"grad_norm": 0.0005177839775569737,
|
8922 |
+
"learning_rate": 3.4102289664578177e-07,
|
8923 |
+
"loss": 46.0,
|
8924 |
+
"step": 1269
|
8925 |
+
},
|
8926 |
+
{
|
8927 |
+
"epoch": 0.025564378956691527,
|
8928 |
+
"grad_norm": 0.0007628992316313088,
|
8929 |
+
"learning_rate": 3.212683189801724e-07,
|
8930 |
+
"loss": 46.0,
|
8931 |
+
"step": 1270
|
8932 |
+
},
|
8933 |
+
{
|
8934 |
+
"epoch": 0.02558450838894089,
|
8935 |
+
"grad_norm": 0.0019209292950108647,
|
8936 |
+
"learning_rate": 3.021021851063899e-07,
|
8937 |
+
"loss": 46.0,
|
8938 |
+
"step": 1271
|
8939 |
+
},
|
8940 |
+
{
|
8941 |
+
"epoch": 0.025604637821190254,
|
8942 |
+
"grad_norm": 0.0006337051163427532,
|
8943 |
+
"learning_rate": 2.8352460816986057e-07,
|
8944 |
+
"loss": 46.0,
|
8945 |
+
"step": 1272
|
8946 |
+
},
|
8947 |
+
{
|
8948 |
+
"epoch": 0.025624767253439616,
|
8949 |
+
"grad_norm": 0.0004375329881440848,
|
8950 |
+
"learning_rate": 2.6553569784152357e-07,
|
8951 |
+
"loss": 46.0,
|
8952 |
+
"step": 1273
|
8953 |
+
},
|
8954 |
+
{
|
8955 |
+
"epoch": 0.02564489668568898,
|
8956 |
+
"grad_norm": 0.0008538602269254625,
|
8957 |
+
"learning_rate": 2.481355603171531e-07,
|
8958 |
+
"loss": 46.0,
|
8959 |
+
"step": 1274
|
8960 |
+
},
|
8961 |
+
{
|
8962 |
+
"epoch": 0.025665026117938344,
|
8963 |
+
"grad_norm": 0.0008380180224776268,
|
8964 |
+
"learning_rate": 2.3132429831682622e-07,
|
8965 |
+
"loss": 46.0,
|
8966 |
+
"step": 1275
|
8967 |
+
},
|
8968 |
+
{
|
8969 |
+
"epoch": 0.025685155550187706,
|
8970 |
+
"grad_norm": 0.0007942443480715156,
|
8971 |
+
"learning_rate": 2.1510201108416728e-07,
|
8972 |
+
"loss": 46.0,
|
8973 |
+
"step": 1276
|
8974 |
+
},
|
8975 |
+
{
|
8976 |
+
"epoch": 0.02570528498243707,
|
8977 |
+
"grad_norm": 0.0008215562556870282,
|
8978 |
+
"learning_rate": 1.9946879438592636e-07,
|
8979 |
+
"loss": 46.0,
|
8980 |
+
"step": 1277
|
8981 |
+
},
|
8982 |
+
{
|
8983 |
+
"epoch": 0.025725414414686433,
|
8984 |
+
"grad_norm": 0.0011535886442288756,
|
8985 |
+
"learning_rate": 1.8442474051125757e-07,
|
8986 |
+
"loss": 46.0,
|
8987 |
+
"step": 1278
|
8988 |
+
},
|
8989 |
+
{
|
8990 |
+
"epoch": 0.0257455438469358,
|
8991 |
+
"grad_norm": 0.00102553132455796,
|
8992 |
+
"learning_rate": 1.6996993827129715e-07,
|
8993 |
+
"loss": 46.0,
|
8994 |
+
"step": 1279
|
8995 |
+
},
|
8996 |
+
{
|
8997 |
+
"epoch": 0.02576567327918516,
|
8998 |
+
"grad_norm": 0.0006207975093275309,
|
8999 |
+
"learning_rate": 1.561044729985861e-07,
|
9000 |
+
"loss": 46.0,
|
9001 |
+
"step": 1280
|
9002 |
+
},
|
9003 |
+
{
|
9004 |
+
"epoch": 0.025785802711434522,
|
9005 |
+
"grad_norm": 0.0006774533540010452,
|
9006 |
+
"learning_rate": 1.428284265465596e-07,
|
9007 |
+
"loss": 46.0,
|
9008 |
+
"step": 1281
|
9009 |
+
},
|
9010 |
+
{
|
9011 |
+
"epoch": 0.025805932143683888,
|
9012 |
+
"grad_norm": 0.0017024496337398887,
|
9013 |
+
"learning_rate": 1.3014187728906945e-07,
|
9014 |
+
"loss": 46.0,
|
9015 |
+
"step": 1282
|
9016 |
+
},
|
9017 |
+
{
|
9018 |
+
"epoch": 0.02582606157593325,
|
9019 |
+
"grad_norm": 0.0017212866805493832,
|
9020 |
+
"learning_rate": 1.1804490011995129e-07,
|
9021 |
+
"loss": 46.0,
|
9022 |
+
"step": 1283
|
9023 |
+
},
|
9024 |
+
{
|
9025 |
+
"epoch": 0.025846191008182615,
|
9026 |
+
"grad_norm": 0.0024159506428986788,
|
9027 |
+
"learning_rate": 1.0653756645252477e-07,
|
9028 |
+
"loss": 46.0,
|
9029 |
+
"step": 1284
|
9030 |
+
},
|
9031 |
+
{
|
9032 |
+
"epoch": 0.025866320440431977,
|
9033 |
+
"grad_norm": 0.0017721692565828562,
|
9034 |
+
"learning_rate": 9.561994421924958e-08,
|
9035 |
+
"loss": 46.0,
|
9036 |
+
"step": 1285
|
9037 |
+
},
|
9038 |
+
{
|
9039 |
+
"epoch": 0.025886449872681343,
|
9040 |
+
"grad_norm": 0.0007376950816251338,
|
9041 |
+
"learning_rate": 8.529209787123682e-08,
|
9042 |
+
"loss": 46.0,
|
9043 |
+
"step": 1286
|
9044 |
+
},
|
9045 |
+
{
|
9046 |
+
"epoch": 0.025906579304930705,
|
9047 |
+
"grad_norm": 0.0007463957881554961,
|
9048 |
+
"learning_rate": 7.555408837794931e-08,
|
9049 |
+
"loss": 46.0,
|
9050 |
+
"step": 1287
|
9051 |
+
},
|
9052 |
+
{
|
9053 |
+
"epoch": 0.025926708737180067,
|
9054 |
+
"grad_norm": 0.0006077784928493202,
|
9055 |
+
"learning_rate": 6.640597322677967e-08,
|
9056 |
+
"loss": 46.0,
|
9057 |
+
"step": 1288
|
9058 |
+
},
|
9059 |
+
{
|
9060 |
+
"epoch": 0.025946838169429432,
|
9061 |
+
"grad_norm": 0.002363705076277256,
|
9062 |
+
"learning_rate": 5.784780642275056e-08,
|
9063 |
+
"loss": 46.0,
|
9064 |
+
"step": 1289
|
9065 |
+
},
|
9066 |
+
{
|
9067 |
+
"epoch": 0.025966967601678794,
|
9068 |
+
"grad_norm": 0.0013910114066675305,
|
9069 |
+
"learning_rate": 4.9879638488159465e-08,
|
9070 |
+
"loss": 46.0,
|
9071 |
+
"step": 1290
|
9072 |
+
},
|
9073 |
+
{
|
9074 |
+
"epoch": 0.02598709703392816,
|
9075 |
+
"grad_norm": 0.001159190433099866,
|
9076 |
+
"learning_rate": 4.2501516462334356e-08,
|
9077 |
+
"loss": 46.0,
|
9078 |
+
"step": 1291
|
9079 |
+
},
|
9080 |
+
{
|
9081 |
+
"epoch": 0.02600722646617752,
|
9082 |
+
"grad_norm": 0.0009958260925486684,
|
9083 |
+
"learning_rate": 3.5713483901300696e-08,
|
9084 |
+
"loss": 46.0,
|
9085 |
+
"step": 1292
|
9086 |
+
},
|
9087 |
+
{
|
9088 |
+
"epoch": 0.026027355898426884,
|
9089 |
+
"grad_norm": 0.0004471206630114466,
|
9090 |
+
"learning_rate": 2.9515580877559346e-08,
|
9091 |
+
"loss": 46.0,
|
9092 |
+
"step": 1293
|
9093 |
+
},
|
9094 |
+
{
|
9095 |
+
"epoch": 0.02604748533067625,
|
9096 |
+
"grad_norm": 0.0009266930283047259,
|
9097 |
+
"learning_rate": 2.3907843979831257e-08,
|
9098 |
+
"loss": 46.0,
|
9099 |
+
"step": 1294
|
9100 |
+
},
|
9101 |
+
{
|
9102 |
+
"epoch": 0.02606761476292561,
|
9103 |
+
"grad_norm": 0.0013211799087002873,
|
9104 |
+
"learning_rate": 1.8890306312846495e-08,
|
9105 |
+
"loss": 46.0,
|
9106 |
+
"step": 1295
|
9107 |
+
},
|
9108 |
+
{
|
9109 |
+
"epoch": 0.026087744195174976,
|
9110 |
+
"grad_norm": 0.001065763528458774,
|
9111 |
+
"learning_rate": 1.446299749716662e-08,
|
9112 |
+
"loss": 46.0,
|
9113 |
+
"step": 1296
|
9114 |
+
},
|
9115 |
+
{
|
9116 |
+
"epoch": 0.02610787362742434,
|
9117 |
+
"grad_norm": 0.001209865789860487,
|
9118 |
+
"learning_rate": 1.0625943668973736e-08,
|
9119 |
+
"loss": 46.0,
|
9120 |
+
"step": 1297
|
9121 |
+
},
|
9122 |
+
{
|
9123 |
+
"epoch": 0.0261280030596737,
|
9124 |
+
"grad_norm": 0.0011283751809969544,
|
9125 |
+
"learning_rate": 7.379167479948379e-09,
|
9126 |
+
"loss": 46.0,
|
9127 |
+
"step": 1298
|
9128 |
+
},
|
9129 |
+
{
|
9130 |
+
"epoch": 0.026148132491923066,
|
9131 |
+
"grad_norm": 0.0004522628150880337,
|
9132 |
+
"learning_rate": 4.722688097125172e-09,
|
9133 |
+
"loss": 46.0,
|
9134 |
+
"step": 1299
|
9135 |
+
},
|
9136 |
+
{
|
9137 |
+
"epoch": 0.026168261924172428,
|
9138 |
+
"grad_norm": 0.0005747976247221231,
|
9139 |
+
"learning_rate": 2.656521202770712e-09,
|
9140 |
+
"loss": 46.0,
|
9141 |
+
"step": 1300
|
9142 |
+
},
|
9143 |
+
{
|
9144 |
+
"epoch": 0.026188391356421793,
|
9145 |
+
"grad_norm": 0.0009918762370944023,
|
9146 |
+
"learning_rate": 1.1806789942947484e-09,
|
9147 |
+
"loss": 46.0,
|
9148 |
+
"step": 1301
|
9149 |
+
},
|
9150 |
+
{
|
9151 |
+
"epoch": 0.026208520788671155,
|
9152 |
+
"grad_norm": 0.0008219339651986957,
|
9153 |
+
"learning_rate": 2.9517018420577305e-10,
|
9154 |
+
"loss": 46.0,
|
9155 |
+
"step": 1302
|
9156 |
+
},
|
9157 |
+
{
|
9158 |
+
"epoch": 0.02622865022092052,
|
9159 |
+
"grad_norm": 0.0007112828898243606,
|
9160 |
+
"learning_rate": 0.0,
|
9161 |
+
"loss": 46.0,
|
9162 |
+
"step": 1303
|
9163 |
}
|
9164 |
],
|
9165 |
"logging_steps": 1,
|
|
|
9174 |
"should_evaluate": false,
|
9175 |
"should_log": false,
|
9176 |
"should_save": true,
|
9177 |
+
"should_training_stop": true
|
9178 |
},
|
9179 |
"attributes": {}
|
9180 |
}
|
9181 |
},
|
9182 |
+
"total_flos": 30018797125632.0,
|
9183 |
"train_batch_size": 2,
|
9184 |
"trial_name": null,
|
9185 |
"trial_params": null
|