Training in progress, step 1377, 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 125040
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:60d91bc2e6e8c113e8d3110776d8c738660c1130d8ae5d4b86eeb3773e80b249
|
3 |
size 125040
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 162868
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:06957f85a83fa2687b3a6fc8225cedd80d935e20ba3aa3cecfab3cf531216193
|
3 |
size 162868
|
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:e13cc6fcf98ce463649d00e6c1e62e50e1d02c22175c9f77243626e309aaaceb
|
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 sha256:1e0e85e0e32fab9354024de5a685b6fd556a114cb069dd71bf06e582a1b393f5
|
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": 345,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -7284,6 +7284,2400 @@
|
|
7284 |
"eval_samples_per_second": 377.57,
|
7285 |
"eval_steps_per_second": 188.829,
|
7286 |
"step": 1035
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
7287 |
}
|
7288 |
],
|
7289 |
"logging_steps": 1,
|
@@ -7298,12 +9692,12 @@
|
|
7298 |
"should_evaluate": false,
|
7299 |
"should_log": false,
|
7300 |
"should_save": true,
|
7301 |
-
"should_training_stop":
|
7302 |
},
|
7303 |
"attributes": {}
|
7304 |
}
|
7305 |
},
|
7306 |
-
"total_flos":
|
7307 |
"train_batch_size": 2,
|
7308 |
"trial_name": null,
|
7309 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.1365800436421345,
|
5 |
"eval_steps": 345,
|
6 |
+
"global_step": 1377,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
7284 |
"eval_samples_per_second": 377.57,
|
7285 |
"eval_steps_per_second": 188.829,
|
7286 |
"step": 1035
|
7287 |
+
},
|
7288 |
+
{
|
7289 |
+
"epoch": 0.10275738940686371,
|
7290 |
+
"grad_norm": 1.6864278316497803,
|
7291 |
+
"learning_rate": 2.9167548331365014e-05,
|
7292 |
+
"loss": 43.3224,
|
7293 |
+
"step": 1036
|
7294 |
+
},
|
7295 |
+
{
|
7296 |
+
"epoch": 0.10285657607617536,
|
7297 |
+
"grad_norm": 2.01654314994812,
|
7298 |
+
"learning_rate": 2.900551100340455e-05,
|
7299 |
+
"loss": 43.3088,
|
7300 |
+
"step": 1037
|
7301 |
+
},
|
7302 |
+
{
|
7303 |
+
"epoch": 0.10295576274548701,
|
7304 |
+
"grad_norm": 1.9112672805786133,
|
7305 |
+
"learning_rate": 2.8843848637362712e-05,
|
7306 |
+
"loss": 43.2829,
|
7307 |
+
"step": 1038
|
7308 |
+
},
|
7309 |
+
{
|
7310 |
+
"epoch": 0.10305494941479865,
|
7311 |
+
"grad_norm": 1.7659448385238647,
|
7312 |
+
"learning_rate": 2.8682562087069486e-05,
|
7313 |
+
"loss": 43.0748,
|
7314 |
+
"step": 1039
|
7315 |
+
},
|
7316 |
+
{
|
7317 |
+
"epoch": 0.1031541360841103,
|
7318 |
+
"grad_norm": 2.215243339538574,
|
7319 |
+
"learning_rate": 2.852165220437015e-05,
|
7320 |
+
"loss": 43.0117,
|
7321 |
+
"step": 1040
|
7322 |
+
},
|
7323 |
+
{
|
7324 |
+
"epoch": 0.10325332275342194,
|
7325 |
+
"grad_norm": 2.1228878498077393,
|
7326 |
+
"learning_rate": 2.8361119839120488e-05,
|
7327 |
+
"loss": 43.401,
|
7328 |
+
"step": 1041
|
7329 |
+
},
|
7330 |
+
{
|
7331 |
+
"epoch": 0.10335250942273358,
|
7332 |
+
"grad_norm": 2.2890191078186035,
|
7333 |
+
"learning_rate": 2.820096583918236e-05,
|
7334 |
+
"loss": 43.1889,
|
7335 |
+
"step": 1042
|
7336 |
+
},
|
7337 |
+
{
|
7338 |
+
"epoch": 0.10345169609204523,
|
7339 |
+
"grad_norm": 1.8007656335830688,
|
7340 |
+
"learning_rate": 2.8041191050419403e-05,
|
7341 |
+
"loss": 43.191,
|
7342 |
+
"step": 1043
|
7343 |
+
},
|
7344 |
+
{
|
7345 |
+
"epoch": 0.10355088276135688,
|
7346 |
+
"grad_norm": 1.9802706241607666,
|
7347 |
+
"learning_rate": 2.788179631669232e-05,
|
7348 |
+
"loss": 43.2676,
|
7349 |
+
"step": 1044
|
7350 |
+
},
|
7351 |
+
{
|
7352 |
+
"epoch": 0.10365006943066851,
|
7353 |
+
"grad_norm": 1.9914954900741577,
|
7354 |
+
"learning_rate": 2.7722782479854514e-05,
|
7355 |
+
"loss": 43.2411,
|
7356 |
+
"step": 1045
|
7357 |
+
},
|
7358 |
+
{
|
7359 |
+
"epoch": 0.10374925609998016,
|
7360 |
+
"grad_norm": 2.4769716262817383,
|
7361 |
+
"learning_rate": 2.7564150379747777e-05,
|
7362 |
+
"loss": 43.3384,
|
7363 |
+
"step": 1046
|
7364 |
+
},
|
7365 |
+
{
|
7366 |
+
"epoch": 0.10384844276929181,
|
7367 |
+
"grad_norm": 2.1755826473236084,
|
7368 |
+
"learning_rate": 2.7405900854197607e-05,
|
7369 |
+
"loss": 43.0251,
|
7370 |
+
"step": 1047
|
7371 |
+
},
|
7372 |
+
{
|
7373 |
+
"epoch": 0.10394762943860344,
|
7374 |
+
"grad_norm": 2.272775650024414,
|
7375 |
+
"learning_rate": 2.7248034739008942e-05,
|
7376 |
+
"loss": 42.9522,
|
7377 |
+
"step": 1048
|
7378 |
+
},
|
7379 |
+
{
|
7380 |
+
"epoch": 0.1040468161079151,
|
7381 |
+
"grad_norm": 2.054165840148926,
|
7382 |
+
"learning_rate": 2.7090552867961817e-05,
|
7383 |
+
"loss": 43.3634,
|
7384 |
+
"step": 1049
|
7385 |
+
},
|
7386 |
+
{
|
7387 |
+
"epoch": 0.10414600277722674,
|
7388 |
+
"grad_norm": 1.8856379985809326,
|
7389 |
+
"learning_rate": 2.693345607280665e-05,
|
7390 |
+
"loss": 43.4114,
|
7391 |
+
"step": 1050
|
7392 |
+
},
|
7393 |
+
{
|
7394 |
+
"epoch": 0.10424518944653839,
|
7395 |
+
"grad_norm": 2.4174342155456543,
|
7396 |
+
"learning_rate": 2.6776745183260278e-05,
|
7397 |
+
"loss": 42.8189,
|
7398 |
+
"step": 1051
|
7399 |
+
},
|
7400 |
+
{
|
7401 |
+
"epoch": 0.10434437611585003,
|
7402 |
+
"grad_norm": 2.0024263858795166,
|
7403 |
+
"learning_rate": 2.662042102700121e-05,
|
7404 |
+
"loss": 43.2766,
|
7405 |
+
"step": 1052
|
7406 |
+
},
|
7407 |
+
{
|
7408 |
+
"epoch": 0.10444356278516168,
|
7409 |
+
"grad_norm": 2.0610110759735107,
|
7410 |
+
"learning_rate": 2.6464484429665404e-05,
|
7411 |
+
"loss": 43.7177,
|
7412 |
+
"step": 1053
|
7413 |
+
},
|
7414 |
+
{
|
7415 |
+
"epoch": 0.10454274945447332,
|
7416 |
+
"grad_norm": 1.9586541652679443,
|
7417 |
+
"learning_rate": 2.6308936214842016e-05,
|
7418 |
+
"loss": 43.3399,
|
7419 |
+
"step": 1054
|
7420 |
+
},
|
7421 |
+
{
|
7422 |
+
"epoch": 0.10464193612378496,
|
7423 |
+
"grad_norm": 2.1504197120666504,
|
7424 |
+
"learning_rate": 2.6153777204068818e-05,
|
7425 |
+
"loss": 43.2787,
|
7426 |
+
"step": 1055
|
7427 |
+
},
|
7428 |
+
{
|
7429 |
+
"epoch": 0.10474112279309661,
|
7430 |
+
"grad_norm": 2.3864498138427734,
|
7431 |
+
"learning_rate": 2.599900821682799e-05,
|
7432 |
+
"loss": 43.372,
|
7433 |
+
"step": 1056
|
7434 |
+
},
|
7435 |
+
{
|
7436 |
+
"epoch": 0.10484030946240826,
|
7437 |
+
"grad_norm": 2.1138644218444824,
|
7438 |
+
"learning_rate": 2.5844630070541865e-05,
|
7439 |
+
"loss": 43.1379,
|
7440 |
+
"step": 1057
|
7441 |
+
},
|
7442 |
+
{
|
7443 |
+
"epoch": 0.10493949613171989,
|
7444 |
+
"grad_norm": 1.9664846658706665,
|
7445 |
+
"learning_rate": 2.569064358056841e-05,
|
7446 |
+
"loss": 43.295,
|
7447 |
+
"step": 1058
|
7448 |
+
},
|
7449 |
+
{
|
7450 |
+
"epoch": 0.10503868280103154,
|
7451 |
+
"grad_norm": 2.0142152309417725,
|
7452 |
+
"learning_rate": 2.5537049560197112e-05,
|
7453 |
+
"loss": 43.0202,
|
7454 |
+
"step": 1059
|
7455 |
+
},
|
7456 |
+
{
|
7457 |
+
"epoch": 0.10513786947034319,
|
7458 |
+
"grad_norm": 1.9622864723205566,
|
7459 |
+
"learning_rate": 2.538384882064453e-05,
|
7460 |
+
"loss": 43.2141,
|
7461 |
+
"step": 1060
|
7462 |
+
},
|
7463 |
+
{
|
7464 |
+
"epoch": 0.10523705613965482,
|
7465 |
+
"grad_norm": 2.4546916484832764,
|
7466 |
+
"learning_rate": 2.523104217105021e-05,
|
7467 |
+
"loss": 43.4144,
|
7468 |
+
"step": 1061
|
7469 |
+
},
|
7470 |
+
{
|
7471 |
+
"epoch": 0.10533624280896647,
|
7472 |
+
"grad_norm": 2.5703399181365967,
|
7473 |
+
"learning_rate": 2.507863041847217e-05,
|
7474 |
+
"loss": 43.1509,
|
7475 |
+
"step": 1062
|
7476 |
+
},
|
7477 |
+
{
|
7478 |
+
"epoch": 0.10543542947827812,
|
7479 |
+
"grad_norm": 2.306077480316162,
|
7480 |
+
"learning_rate": 2.4926614367882805e-05,
|
7481 |
+
"loss": 43.5177,
|
7482 |
+
"step": 1063
|
7483 |
+
},
|
7484 |
+
{
|
7485 |
+
"epoch": 0.10553461614758976,
|
7486 |
+
"grad_norm": 1.7721586227416992,
|
7487 |
+
"learning_rate": 2.4774994822164534e-05,
|
7488 |
+
"loss": 43.3269,
|
7489 |
+
"step": 1064
|
7490 |
+
},
|
7491 |
+
{
|
7492 |
+
"epoch": 0.1056338028169014,
|
7493 |
+
"grad_norm": 2.2535550594329834,
|
7494 |
+
"learning_rate": 2.462377258210572e-05,
|
7495 |
+
"loss": 43.3561,
|
7496 |
+
"step": 1065
|
7497 |
+
},
|
7498 |
+
{
|
7499 |
+
"epoch": 0.10573298948621306,
|
7500 |
+
"grad_norm": 1.9601596593856812,
|
7501 |
+
"learning_rate": 2.447294844639624e-05,
|
7502 |
+
"loss": 43.4029,
|
7503 |
+
"step": 1066
|
7504 |
+
},
|
7505 |
+
{
|
7506 |
+
"epoch": 0.1058321761555247,
|
7507 |
+
"grad_norm": 1.8112313747406006,
|
7508 |
+
"learning_rate": 2.4322523211623327e-05,
|
7509 |
+
"loss": 43.3426,
|
7510 |
+
"step": 1067
|
7511 |
+
},
|
7512 |
+
{
|
7513 |
+
"epoch": 0.10593136282483634,
|
7514 |
+
"grad_norm": 1.8747832775115967,
|
7515 |
+
"learning_rate": 2.4172497672267514e-05,
|
7516 |
+
"loss": 43.3777,
|
7517 |
+
"step": 1068
|
7518 |
+
},
|
7519 |
+
{
|
7520 |
+
"epoch": 0.10603054949414799,
|
7521 |
+
"grad_norm": 1.791438341140747,
|
7522 |
+
"learning_rate": 2.4022872620698232e-05,
|
7523 |
+
"loss": 43.2513,
|
7524 |
+
"step": 1069
|
7525 |
+
},
|
7526 |
+
{
|
7527 |
+
"epoch": 0.10612973616345964,
|
7528 |
+
"grad_norm": 1.8881406784057617,
|
7529 |
+
"learning_rate": 2.3873648847169715e-05,
|
7530 |
+
"loss": 43.5012,
|
7531 |
+
"step": 1070
|
7532 |
+
},
|
7533 |
+
{
|
7534 |
+
"epoch": 0.10622892283277127,
|
7535 |
+
"grad_norm": 2.078063488006592,
|
7536 |
+
"learning_rate": 2.372482713981682e-05,
|
7537 |
+
"loss": 43.3562,
|
7538 |
+
"step": 1071
|
7539 |
+
},
|
7540 |
+
{
|
7541 |
+
"epoch": 0.10632810950208292,
|
7542 |
+
"grad_norm": 2.2463982105255127,
|
7543 |
+
"learning_rate": 2.357640828465092e-05,
|
7544 |
+
"loss": 43.3477,
|
7545 |
+
"step": 1072
|
7546 |
+
},
|
7547 |
+
{
|
7548 |
+
"epoch": 0.10642729617139457,
|
7549 |
+
"grad_norm": 2.4299933910369873,
|
7550 |
+
"learning_rate": 2.3428393065555632e-05,
|
7551 |
+
"loss": 43.1149,
|
7552 |
+
"step": 1073
|
7553 |
+
},
|
7554 |
+
{
|
7555 |
+
"epoch": 0.1065264828407062,
|
7556 |
+
"grad_norm": 2.2693679332733154,
|
7557 |
+
"learning_rate": 2.3280782264282753e-05,
|
7558 |
+
"loss": 43.715,
|
7559 |
+
"step": 1074
|
7560 |
+
},
|
7561 |
+
{
|
7562 |
+
"epoch": 0.10662566951001785,
|
7563 |
+
"grad_norm": 2.211836338043213,
|
7564 |
+
"learning_rate": 2.3133576660448176e-05,
|
7565 |
+
"loss": 43.4276,
|
7566 |
+
"step": 1075
|
7567 |
+
},
|
7568 |
+
{
|
7569 |
+
"epoch": 0.1067248561793295,
|
7570 |
+
"grad_norm": 1.9994254112243652,
|
7571 |
+
"learning_rate": 2.2986777031527673e-05,
|
7572 |
+
"loss": 43.316,
|
7573 |
+
"step": 1076
|
7574 |
+
},
|
7575 |
+
{
|
7576 |
+
"epoch": 0.10682404284864114,
|
7577 |
+
"grad_norm": 2.2838003635406494,
|
7578 |
+
"learning_rate": 2.28403841528528e-05,
|
7579 |
+
"loss": 43.4559,
|
7580 |
+
"step": 1077
|
7581 |
+
},
|
7582 |
+
{
|
7583 |
+
"epoch": 0.10692322951795279,
|
7584 |
+
"grad_norm": 2.2666330337524414,
|
7585 |
+
"learning_rate": 2.2694398797606954e-05,
|
7586 |
+
"loss": 43.0834,
|
7587 |
+
"step": 1078
|
7588 |
+
},
|
7589 |
+
{
|
7590 |
+
"epoch": 0.10702241618726444,
|
7591 |
+
"grad_norm": 2.0169146060943604,
|
7592 |
+
"learning_rate": 2.2548821736821002e-05,
|
7593 |
+
"loss": 43.4411,
|
7594 |
+
"step": 1079
|
7595 |
+
},
|
7596 |
+
{
|
7597 |
+
"epoch": 0.10712160285657607,
|
7598 |
+
"grad_norm": 1.9657902717590332,
|
7599 |
+
"learning_rate": 2.2403653739369535e-05,
|
7600 |
+
"loss": 43.4966,
|
7601 |
+
"step": 1080
|
7602 |
+
},
|
7603 |
+
{
|
7604 |
+
"epoch": 0.10722078952588772,
|
7605 |
+
"grad_norm": 2.2416534423828125,
|
7606 |
+
"learning_rate": 2.2258895571966564e-05,
|
7607 |
+
"loss": 43.2932,
|
7608 |
+
"step": 1081
|
7609 |
+
},
|
7610 |
+
{
|
7611 |
+
"epoch": 0.10731997619519937,
|
7612 |
+
"grad_norm": 2.406165599822998,
|
7613 |
+
"learning_rate": 2.211454799916155e-05,
|
7614 |
+
"loss": 43.3974,
|
7615 |
+
"step": 1082
|
7616 |
+
},
|
7617 |
+
{
|
7618 |
+
"epoch": 0.107419162864511,
|
7619 |
+
"grad_norm": 2.0482776165008545,
|
7620 |
+
"learning_rate": 2.197061178333545e-05,
|
7621 |
+
"loss": 43.2469,
|
7622 |
+
"step": 1083
|
7623 |
+
},
|
7624 |
+
{
|
7625 |
+
"epoch": 0.10751834953382265,
|
7626 |
+
"grad_norm": 2.2992138862609863,
|
7627 |
+
"learning_rate": 2.1827087684696522e-05,
|
7628 |
+
"loss": 43.4739,
|
7629 |
+
"step": 1084
|
7630 |
+
},
|
7631 |
+
{
|
7632 |
+
"epoch": 0.1076175362031343,
|
7633 |
+
"grad_norm": 2.0701234340667725,
|
7634 |
+
"learning_rate": 2.1683976461276446e-05,
|
7635 |
+
"loss": 43.152,
|
7636 |
+
"step": 1085
|
7637 |
+
},
|
7638 |
+
{
|
7639 |
+
"epoch": 0.10771672287244595,
|
7640 |
+
"grad_norm": 2.5112268924713135,
|
7641 |
+
"learning_rate": 2.1541278868926296e-05,
|
7642 |
+
"loss": 43.3509,
|
7643 |
+
"step": 1086
|
7644 |
+
},
|
7645 |
+
{
|
7646 |
+
"epoch": 0.10781590954175758,
|
7647 |
+
"grad_norm": 1.8819648027420044,
|
7648 |
+
"learning_rate": 2.1398995661312482e-05,
|
7649 |
+
"loss": 43.6435,
|
7650 |
+
"step": 1087
|
7651 |
+
},
|
7652 |
+
{
|
7653 |
+
"epoch": 0.10791509621106923,
|
7654 |
+
"grad_norm": 2.143810272216797,
|
7655 |
+
"learning_rate": 2.1257127589912807e-05,
|
7656 |
+
"loss": 43.3598,
|
7657 |
+
"step": 1088
|
7658 |
+
},
|
7659 |
+
{
|
7660 |
+
"epoch": 0.10801428288038088,
|
7661 |
+
"grad_norm": 2.1973557472229004,
|
7662 |
+
"learning_rate": 2.1115675404012625e-05,
|
7663 |
+
"loss": 42.8287,
|
7664 |
+
"step": 1089
|
7665 |
+
},
|
7666 |
+
{
|
7667 |
+
"epoch": 0.10811346954969252,
|
7668 |
+
"grad_norm": 2.4099574089050293,
|
7669 |
+
"learning_rate": 2.097463985070054e-05,
|
7670 |
+
"loss": 43.4581,
|
7671 |
+
"step": 1090
|
7672 |
+
},
|
7673 |
+
{
|
7674 |
+
"epoch": 0.10821265621900417,
|
7675 |
+
"grad_norm": 2.112236976623535,
|
7676 |
+
"learning_rate": 2.083402167486489e-05,
|
7677 |
+
"loss": 43.4635,
|
7678 |
+
"step": 1091
|
7679 |
+
},
|
7680 |
+
{
|
7681 |
+
"epoch": 0.10831184288831582,
|
7682 |
+
"grad_norm": 2.2669761180877686,
|
7683 |
+
"learning_rate": 2.069382161918958e-05,
|
7684 |
+
"loss": 43.2845,
|
7685 |
+
"step": 1092
|
7686 |
+
},
|
7687 |
+
{
|
7688 |
+
"epoch": 0.10841102955762745,
|
7689 |
+
"grad_norm": 2.3433635234832764,
|
7690 |
+
"learning_rate": 2.0554040424150023e-05,
|
7691 |
+
"loss": 43.3251,
|
7692 |
+
"step": 1093
|
7693 |
+
},
|
7694 |
+
{
|
7695 |
+
"epoch": 0.1085102162269391,
|
7696 |
+
"grad_norm": 1.849046230316162,
|
7697 |
+
"learning_rate": 2.0414678828009613e-05,
|
7698 |
+
"loss": 43.2783,
|
7699 |
+
"step": 1094
|
7700 |
+
},
|
7701 |
+
{
|
7702 |
+
"epoch": 0.10860940289625075,
|
7703 |
+
"grad_norm": 2.100344181060791,
|
7704 |
+
"learning_rate": 2.0275737566815455e-05,
|
7705 |
+
"loss": 43.2561,
|
7706 |
+
"step": 1095
|
7707 |
+
},
|
7708 |
+
{
|
7709 |
+
"epoch": 0.10870858956556238,
|
7710 |
+
"grad_norm": 2.0855021476745605,
|
7711 |
+
"learning_rate": 2.0137217374394655e-05,
|
7712 |
+
"loss": 43.2956,
|
7713 |
+
"step": 1096
|
7714 |
+
},
|
7715 |
+
{
|
7716 |
+
"epoch": 0.10880777623487403,
|
7717 |
+
"grad_norm": 2.3706424236297607,
|
7718 |
+
"learning_rate": 1.9999118982350485e-05,
|
7719 |
+
"loss": 43.4625,
|
7720 |
+
"step": 1097
|
7721 |
+
},
|
7722 |
+
{
|
7723 |
+
"epoch": 0.10890696290418568,
|
7724 |
+
"grad_norm": 2.0374128818511963,
|
7725 |
+
"learning_rate": 1.98614431200584e-05,
|
7726 |
+
"loss": 43.1283,
|
7727 |
+
"step": 1098
|
7728 |
+
},
|
7729 |
+
{
|
7730 |
+
"epoch": 0.10900614957349732,
|
7731 |
+
"grad_norm": 1.7949743270874023,
|
7732 |
+
"learning_rate": 1.972419051466219e-05,
|
7733 |
+
"loss": 43.2577,
|
7734 |
+
"step": 1099
|
7735 |
+
},
|
7736 |
+
{
|
7737 |
+
"epoch": 0.10910533624280896,
|
7738 |
+
"grad_norm": 2.187823534011841,
|
7739 |
+
"learning_rate": 1.95873618910703e-05,
|
7740 |
+
"loss": 43.2838,
|
7741 |
+
"step": 1100
|
7742 |
+
},
|
7743 |
+
{
|
7744 |
+
"epoch": 0.10920452291212061,
|
7745 |
+
"grad_norm": 1.766406774520874,
|
7746 |
+
"learning_rate": 1.945095797195181e-05,
|
7747 |
+
"loss": 43.5403,
|
7748 |
+
"step": 1101
|
7749 |
+
},
|
7750 |
+
{
|
7751 |
+
"epoch": 0.10930370958143226,
|
7752 |
+
"grad_norm": 1.9312326908111572,
|
7753 |
+
"learning_rate": 1.9314979477732674e-05,
|
7754 |
+
"loss": 43.451,
|
7755 |
+
"step": 1102
|
7756 |
+
},
|
7757 |
+
{
|
7758 |
+
"epoch": 0.1094028962507439,
|
7759 |
+
"grad_norm": 2.104093074798584,
|
7760 |
+
"learning_rate": 1.9179427126592054e-05,
|
7761 |
+
"loss": 43.4365,
|
7762 |
+
"step": 1103
|
7763 |
+
},
|
7764 |
+
{
|
7765 |
+
"epoch": 0.10950208292005555,
|
7766 |
+
"grad_norm": 2.419813871383667,
|
7767 |
+
"learning_rate": 1.9044301634458238e-05,
|
7768 |
+
"loss": 43.5545,
|
7769 |
+
"step": 1104
|
7770 |
+
},
|
7771 |
+
{
|
7772 |
+
"epoch": 0.1096012695893672,
|
7773 |
+
"grad_norm": 2.0978732109069824,
|
7774 |
+
"learning_rate": 1.8909603715005198e-05,
|
7775 |
+
"loss": 43.1465,
|
7776 |
+
"step": 1105
|
7777 |
+
},
|
7778 |
+
{
|
7779 |
+
"epoch": 0.10970045625867883,
|
7780 |
+
"grad_norm": 2.210231304168701,
|
7781 |
+
"learning_rate": 1.8775334079648554e-05,
|
7782 |
+
"loss": 43.5224,
|
7783 |
+
"step": 1106
|
7784 |
+
},
|
7785 |
+
{
|
7786 |
+
"epoch": 0.10979964292799048,
|
7787 |
+
"grad_norm": 2.125434160232544,
|
7788 |
+
"learning_rate": 1.864149343754191e-05,
|
7789 |
+
"loss": 43.4481,
|
7790 |
+
"step": 1107
|
7791 |
+
},
|
7792 |
+
{
|
7793 |
+
"epoch": 0.10989882959730213,
|
7794 |
+
"grad_norm": 2.2087841033935547,
|
7795 |
+
"learning_rate": 1.850808249557322e-05,
|
7796 |
+
"loss": 43.1026,
|
7797 |
+
"step": 1108
|
7798 |
+
},
|
7799 |
+
{
|
7800 |
+
"epoch": 0.10999801626661376,
|
7801 |
+
"grad_norm": 2.109090566635132,
|
7802 |
+
"learning_rate": 1.837510195836083e-05,
|
7803 |
+
"loss": 43.3705,
|
7804 |
+
"step": 1109
|
7805 |
+
},
|
7806 |
+
{
|
7807 |
+
"epoch": 0.11009720293592541,
|
7808 |
+
"grad_norm": 2.0380489826202393,
|
7809 |
+
"learning_rate": 1.82425525282499e-05,
|
7810 |
+
"loss": 43.1155,
|
7811 |
+
"step": 1110
|
7812 |
+
},
|
7813 |
+
{
|
7814 |
+
"epoch": 0.11019638960523706,
|
7815 |
+
"grad_norm": 1.8931025266647339,
|
7816 |
+
"learning_rate": 1.8110434905308758e-05,
|
7817 |
+
"loss": 43.5846,
|
7818 |
+
"step": 1111
|
7819 |
+
},
|
7820 |
+
{
|
7821 |
+
"epoch": 0.1102955762745487,
|
7822 |
+
"grad_norm": 2.193695545196533,
|
7823 |
+
"learning_rate": 1.797874978732502e-05,
|
7824 |
+
"loss": 43.1693,
|
7825 |
+
"step": 1112
|
7826 |
+
},
|
7827 |
+
{
|
7828 |
+
"epoch": 0.11039476294386034,
|
7829 |
+
"grad_norm": 2.0088858604431152,
|
7830 |
+
"learning_rate": 1.7847497869802e-05,
|
7831 |
+
"loss": 43.4337,
|
7832 |
+
"step": 1113
|
7833 |
+
},
|
7834 |
+
{
|
7835 |
+
"epoch": 0.110493949613172,
|
7836 |
+
"grad_norm": 2.075730085372925,
|
7837 |
+
"learning_rate": 1.771667984595511e-05,
|
7838 |
+
"loss": 43.2213,
|
7839 |
+
"step": 1114
|
7840 |
+
},
|
7841 |
+
{
|
7842 |
+
"epoch": 0.11059313628248363,
|
7843 |
+
"grad_norm": 1.8633036613464355,
|
7844 |
+
"learning_rate": 1.758629640670807e-05,
|
7845 |
+
"loss": 43.3132,
|
7846 |
+
"step": 1115
|
7847 |
+
},
|
7848 |
+
{
|
7849 |
+
"epoch": 0.11069232295179528,
|
7850 |
+
"grad_norm": 2.1463515758514404,
|
7851 |
+
"learning_rate": 1.7456348240689324e-05,
|
7852 |
+
"loss": 43.0615,
|
7853 |
+
"step": 1116
|
7854 |
+
},
|
7855 |
+
{
|
7856 |
+
"epoch": 0.11079150962110693,
|
7857 |
+
"grad_norm": 2.1884613037109375,
|
7858 |
+
"learning_rate": 1.7326836034228378e-05,
|
7859 |
+
"loss": 43.2174,
|
7860 |
+
"step": 1117
|
7861 |
+
},
|
7862 |
+
{
|
7863 |
+
"epoch": 0.11089069629041856,
|
7864 |
+
"grad_norm": 1.8539804220199585,
|
7865 |
+
"learning_rate": 1.7197760471352185e-05,
|
7866 |
+
"loss": 43.2283,
|
7867 |
+
"step": 1118
|
7868 |
+
},
|
7869 |
+
{
|
7870 |
+
"epoch": 0.11098988295973021,
|
7871 |
+
"grad_norm": 2.3994038105010986,
|
7872 |
+
"learning_rate": 1.706912223378162e-05,
|
7873 |
+
"loss": 43.0468,
|
7874 |
+
"step": 1119
|
7875 |
+
},
|
7876 |
+
{
|
7877 |
+
"epoch": 0.11108906962904186,
|
7878 |
+
"grad_norm": 2.16751766204834,
|
7879 |
+
"learning_rate": 1.6940922000927707e-05,
|
7880 |
+
"loss": 43.1557,
|
7881 |
+
"step": 1120
|
7882 |
+
},
|
7883 |
+
{
|
7884 |
+
"epoch": 0.11118825629835351,
|
7885 |
+
"grad_norm": 2.020477294921875,
|
7886 |
+
"learning_rate": 1.68131604498881e-05,
|
7887 |
+
"loss": 43.1987,
|
7888 |
+
"step": 1121
|
7889 |
+
},
|
7890 |
+
{
|
7891 |
+
"epoch": 0.11128744296766514,
|
7892 |
+
"grad_norm": 1.9552767276763916,
|
7893 |
+
"learning_rate": 1.6685838255443652e-05,
|
7894 |
+
"loss": 43.2363,
|
7895 |
+
"step": 1122
|
7896 |
+
},
|
7897 |
+
{
|
7898 |
+
"epoch": 0.11138662963697679,
|
7899 |
+
"grad_norm": 2.1262552738189697,
|
7900 |
+
"learning_rate": 1.65589560900546e-05,
|
7901 |
+
"loss": 43.547,
|
7902 |
+
"step": 1123
|
7903 |
+
},
|
7904 |
+
{
|
7905 |
+
"epoch": 0.11148581630628844,
|
7906 |
+
"grad_norm": 2.2838950157165527,
|
7907 |
+
"learning_rate": 1.6432514623857198e-05,
|
7908 |
+
"loss": 43.0649,
|
7909 |
+
"step": 1124
|
7910 |
+
},
|
7911 |
+
{
|
7912 |
+
"epoch": 0.11158500297560008,
|
7913 |
+
"grad_norm": 2.2327919006347656,
|
7914 |
+
"learning_rate": 1.6306514524660087e-05,
|
7915 |
+
"loss": 42.8205,
|
7916 |
+
"step": 1125
|
7917 |
+
},
|
7918 |
+
{
|
7919 |
+
"epoch": 0.11168418964491172,
|
7920 |
+
"grad_norm": 2.279003858566284,
|
7921 |
+
"learning_rate": 1.6180956457940864e-05,
|
7922 |
+
"loss": 43.2359,
|
7923 |
+
"step": 1126
|
7924 |
+
},
|
7925 |
+
{
|
7926 |
+
"epoch": 0.11178337631422337,
|
7927 |
+
"grad_norm": 2.4765570163726807,
|
7928 |
+
"learning_rate": 1.6055841086842448e-05,
|
7929 |
+
"loss": 42.9458,
|
7930 |
+
"step": 1127
|
7931 |
+
},
|
7932 |
+
{
|
7933 |
+
"epoch": 0.11188256298353501,
|
7934 |
+
"grad_norm": 2.1235129833221436,
|
7935 |
+
"learning_rate": 1.5931169072169615e-05,
|
7936 |
+
"loss": 43.5143,
|
7937 |
+
"step": 1128
|
7938 |
+
},
|
7939 |
+
{
|
7940 |
+
"epoch": 0.11198174965284666,
|
7941 |
+
"grad_norm": 2.516355276107788,
|
7942 |
+
"learning_rate": 1.5806941072385594e-05,
|
7943 |
+
"loss": 43.3404,
|
7944 |
+
"step": 1129
|
7945 |
+
},
|
7946 |
+
{
|
7947 |
+
"epoch": 0.1120809363221583,
|
7948 |
+
"grad_norm": 2.1377298831939697,
|
7949 |
+
"learning_rate": 1.5683157743608478e-05,
|
7950 |
+
"loss": 43.362,
|
7951 |
+
"step": 1130
|
7952 |
+
},
|
7953 |
+
{
|
7954 |
+
"epoch": 0.11218012299146994,
|
7955 |
+
"grad_norm": 2.019286870956421,
|
7956 |
+
"learning_rate": 1.555981973960776e-05,
|
7957 |
+
"loss": 43.4086,
|
7958 |
+
"step": 1131
|
7959 |
+
},
|
7960 |
+
{
|
7961 |
+
"epoch": 0.11227930966078159,
|
7962 |
+
"grad_norm": 1.940057635307312,
|
7963 |
+
"learning_rate": 1.5436927711801042e-05,
|
7964 |
+
"loss": 43.3762,
|
7965 |
+
"step": 1132
|
7966 |
+
},
|
7967 |
+
{
|
7968 |
+
"epoch": 0.11237849633009324,
|
7969 |
+
"grad_norm": 2.371516227722168,
|
7970 |
+
"learning_rate": 1.5314482309250365e-05,
|
7971 |
+
"loss": 43.0204,
|
7972 |
+
"step": 1133
|
7973 |
+
},
|
7974 |
+
{
|
7975 |
+
"epoch": 0.11247768299940487,
|
7976 |
+
"grad_norm": 2.1984615325927734,
|
7977 |
+
"learning_rate": 1.5192484178658939e-05,
|
7978 |
+
"loss": 42.9645,
|
7979 |
+
"step": 1134
|
7980 |
+
},
|
7981 |
+
{
|
7982 |
+
"epoch": 0.11257686966871652,
|
7983 |
+
"grad_norm": 2.0505456924438477,
|
7984 |
+
"learning_rate": 1.5070933964367651e-05,
|
7985 |
+
"loss": 43.2871,
|
7986 |
+
"step": 1135
|
7987 |
+
},
|
7988 |
+
{
|
7989 |
+
"epoch": 0.11267605633802817,
|
7990 |
+
"grad_norm": 2.359302282333374,
|
7991 |
+
"learning_rate": 1.4949832308351685e-05,
|
7992 |
+
"loss": 42.805,
|
7993 |
+
"step": 1136
|
7994 |
+
},
|
7995 |
+
{
|
7996 |
+
"epoch": 0.11277524300733982,
|
7997 |
+
"grad_norm": 2.4442696571350098,
|
7998 |
+
"learning_rate": 1.4829179850217201e-05,
|
7999 |
+
"loss": 43.5687,
|
8000 |
+
"step": 1137
|
8001 |
+
},
|
8002 |
+
{
|
8003 |
+
"epoch": 0.11287442967665146,
|
8004 |
+
"grad_norm": 2.29499888420105,
|
8005 |
+
"learning_rate": 1.470897722719783e-05,
|
8006 |
+
"loss": 43.3301,
|
8007 |
+
"step": 1138
|
8008 |
+
},
|
8009 |
+
{
|
8010 |
+
"epoch": 0.1129736163459631,
|
8011 |
+
"grad_norm": 2.022026300430298,
|
8012 |
+
"learning_rate": 1.4589225074151357e-05,
|
8013 |
+
"loss": 43.439,
|
8014 |
+
"step": 1139
|
8015 |
+
},
|
8016 |
+
{
|
8017 |
+
"epoch": 0.11307280301527475,
|
8018 |
+
"grad_norm": 2.145610809326172,
|
8019 |
+
"learning_rate": 1.446992402355647e-05,
|
8020 |
+
"loss": 43.5619,
|
8021 |
+
"step": 1140
|
8022 |
+
},
|
8023 |
+
{
|
8024 |
+
"epoch": 0.11317198968458639,
|
8025 |
+
"grad_norm": 1.9827899932861328,
|
8026 |
+
"learning_rate": 1.4351074705509237e-05,
|
8027 |
+
"loss": 43.2028,
|
8028 |
+
"step": 1141
|
8029 |
+
},
|
8030 |
+
{
|
8031 |
+
"epoch": 0.11327117635389804,
|
8032 |
+
"grad_norm": 1.884475827217102,
|
8033 |
+
"learning_rate": 1.423267774771988e-05,
|
8034 |
+
"loss": 43.3077,
|
8035 |
+
"step": 1142
|
8036 |
+
},
|
8037 |
+
{
|
8038 |
+
"epoch": 0.11337036302320969,
|
8039 |
+
"grad_norm": 2.340038776397705,
|
8040 |
+
"learning_rate": 1.4114733775509559e-05,
|
8041 |
+
"loss": 43.2573,
|
8042 |
+
"step": 1143
|
8043 |
+
},
|
8044 |
+
{
|
8045 |
+
"epoch": 0.11346954969252132,
|
8046 |
+
"grad_norm": 2.2281246185302734,
|
8047 |
+
"learning_rate": 1.3997243411806782e-05,
|
8048 |
+
"loss": 43.5854,
|
8049 |
+
"step": 1144
|
8050 |
+
},
|
8051 |
+
{
|
8052 |
+
"epoch": 0.11356873636183297,
|
8053 |
+
"grad_norm": 2.023604154586792,
|
8054 |
+
"learning_rate": 1.388020727714443e-05,
|
8055 |
+
"loss": 43.3586,
|
8056 |
+
"step": 1145
|
8057 |
+
},
|
8058 |
+
{
|
8059 |
+
"epoch": 0.11366792303114462,
|
8060 |
+
"grad_norm": 2.2741992473602295,
|
8061 |
+
"learning_rate": 1.376362598965637e-05,
|
8062 |
+
"loss": 43.455,
|
8063 |
+
"step": 1146
|
8064 |
+
},
|
8065 |
+
{
|
8066 |
+
"epoch": 0.11376710970045625,
|
8067 |
+
"grad_norm": 2.2113194465637207,
|
8068 |
+
"learning_rate": 1.3647500165074e-05,
|
8069 |
+
"loss": 43.4446,
|
8070 |
+
"step": 1147
|
8071 |
+
},
|
8072 |
+
{
|
8073 |
+
"epoch": 0.1138662963697679,
|
8074 |
+
"grad_norm": 1.9329335689544678,
|
8075 |
+
"learning_rate": 1.3531830416723346e-05,
|
8076 |
+
"loss": 43.4368,
|
8077 |
+
"step": 1148
|
8078 |
+
},
|
8079 |
+
{
|
8080 |
+
"epoch": 0.11396548303907955,
|
8081 |
+
"grad_norm": 1.7729626893997192,
|
8082 |
+
"learning_rate": 1.341661735552151e-05,
|
8083 |
+
"loss": 43.4287,
|
8084 |
+
"step": 1149
|
8085 |
+
},
|
8086 |
+
{
|
8087 |
+
"epoch": 0.11406466970839119,
|
8088 |
+
"grad_norm": 2.329885482788086,
|
8089 |
+
"learning_rate": 1.3301861589973619e-05,
|
8090 |
+
"loss": 42.897,
|
8091 |
+
"step": 1150
|
8092 |
+
},
|
8093 |
+
{
|
8094 |
+
"epoch": 0.11416385637770284,
|
8095 |
+
"grad_norm": 2.526484966278076,
|
8096 |
+
"learning_rate": 1.3187563726169593e-05,
|
8097 |
+
"loss": 43.578,
|
8098 |
+
"step": 1151
|
8099 |
+
},
|
8100 |
+
{
|
8101 |
+
"epoch": 0.11426304304701448,
|
8102 |
+
"grad_norm": 1.6455745697021484,
|
8103 |
+
"learning_rate": 1.3073724367780871e-05,
|
8104 |
+
"loss": 43.4424,
|
8105 |
+
"step": 1152
|
8106 |
+
},
|
8107 |
+
{
|
8108 |
+
"epoch": 0.11436222971632612,
|
8109 |
+
"grad_norm": 2.121734619140625,
|
8110 |
+
"learning_rate": 1.2960344116057255e-05,
|
8111 |
+
"loss": 43.2584,
|
8112 |
+
"step": 1153
|
8113 |
+
},
|
8114 |
+
{
|
8115 |
+
"epoch": 0.11446141638563777,
|
8116 |
+
"grad_norm": 2.1160125732421875,
|
8117 |
+
"learning_rate": 1.2847423569823814e-05,
|
8118 |
+
"loss": 43.4491,
|
8119 |
+
"step": 1154
|
8120 |
+
},
|
8121 |
+
{
|
8122 |
+
"epoch": 0.11456060305494942,
|
8123 |
+
"grad_norm": 2.4812018871307373,
|
8124 |
+
"learning_rate": 1.2734963325477578e-05,
|
8125 |
+
"loss": 43.2161,
|
8126 |
+
"step": 1155
|
8127 |
+
},
|
8128 |
+
{
|
8129 |
+
"epoch": 0.11465978972426107,
|
8130 |
+
"grad_norm": 2.1262412071228027,
|
8131 |
+
"learning_rate": 1.2622963976984493e-05,
|
8132 |
+
"loss": 43.2379,
|
8133 |
+
"step": 1156
|
8134 |
+
},
|
8135 |
+
{
|
8136 |
+
"epoch": 0.1147589763935727,
|
8137 |
+
"grad_norm": 2.0442497730255127,
|
8138 |
+
"learning_rate": 1.2511426115876312e-05,
|
8139 |
+
"loss": 43.0918,
|
8140 |
+
"step": 1157
|
8141 |
+
},
|
8142 |
+
{
|
8143 |
+
"epoch": 0.11485816306288435,
|
8144 |
+
"grad_norm": 2.125736713409424,
|
8145 |
+
"learning_rate": 1.2400350331247268e-05,
|
8146 |
+
"loss": 43.2273,
|
8147 |
+
"step": 1158
|
8148 |
+
},
|
8149 |
+
{
|
8150 |
+
"epoch": 0.114957349732196,
|
8151 |
+
"grad_norm": 2.3522708415985107,
|
8152 |
+
"learning_rate": 1.228973720975124e-05,
|
8153 |
+
"loss": 43.4113,
|
8154 |
+
"step": 1159
|
8155 |
+
},
|
8156 |
+
{
|
8157 |
+
"epoch": 0.11505653640150763,
|
8158 |
+
"grad_norm": 2.5320546627044678,
|
8159 |
+
"learning_rate": 1.217958733559853e-05,
|
8160 |
+
"loss": 43.2658,
|
8161 |
+
"step": 1160
|
8162 |
+
},
|
8163 |
+
{
|
8164 |
+
"epoch": 0.11515572307081928,
|
8165 |
+
"grad_norm": 2.2132513523101807,
|
8166 |
+
"learning_rate": 1.2069901290552632e-05,
|
8167 |
+
"loss": 43.6697,
|
8168 |
+
"step": 1161
|
8169 |
+
},
|
8170 |
+
{
|
8171 |
+
"epoch": 0.11525490974013093,
|
8172 |
+
"grad_norm": 2.2867939472198486,
|
8173 |
+
"learning_rate": 1.1960679653927443e-05,
|
8174 |
+
"loss": 42.9501,
|
8175 |
+
"step": 1162
|
8176 |
+
},
|
8177 |
+
{
|
8178 |
+
"epoch": 0.11535409640944257,
|
8179 |
+
"grad_norm": 2.244797706604004,
|
8180 |
+
"learning_rate": 1.1851923002583986e-05,
|
8181 |
+
"loss": 43.6076,
|
8182 |
+
"step": 1163
|
8183 |
+
},
|
8184 |
+
{
|
8185 |
+
"epoch": 0.11545328307875422,
|
8186 |
+
"grad_norm": 2.0788700580596924,
|
8187 |
+
"learning_rate": 1.1743631910927411e-05,
|
8188 |
+
"loss": 43.2903,
|
8189 |
+
"step": 1164
|
8190 |
+
},
|
8191 |
+
{
|
8192 |
+
"epoch": 0.11555246974806586,
|
8193 |
+
"grad_norm": 2.050762414932251,
|
8194 |
+
"learning_rate": 1.1635806950904072e-05,
|
8195 |
+
"loss": 43.2543,
|
8196 |
+
"step": 1165
|
8197 |
+
},
|
8198 |
+
{
|
8199 |
+
"epoch": 0.1156516564173775,
|
8200 |
+
"grad_norm": 1.866451621055603,
|
8201 |
+
"learning_rate": 1.152844869199834e-05,
|
8202 |
+
"loss": 43.4068,
|
8203 |
+
"step": 1166
|
8204 |
+
},
|
8205 |
+
{
|
8206 |
+
"epoch": 0.11575084308668915,
|
8207 |
+
"grad_norm": 2.409834861755371,
|
8208 |
+
"learning_rate": 1.142155770122969e-05,
|
8209 |
+
"loss": 43.3885,
|
8210 |
+
"step": 1167
|
8211 |
+
},
|
8212 |
+
{
|
8213 |
+
"epoch": 0.1158500297560008,
|
8214 |
+
"grad_norm": 1.9911494255065918,
|
8215 |
+
"learning_rate": 1.1315134543149741e-05,
|
8216 |
+
"loss": 43.3299,
|
8217 |
+
"step": 1168
|
8218 |
+
},
|
8219 |
+
{
|
8220 |
+
"epoch": 0.11594921642531243,
|
8221 |
+
"grad_norm": 2.2691869735717773,
|
8222 |
+
"learning_rate": 1.1209179779839174e-05,
|
8223 |
+
"loss": 43.1674,
|
8224 |
+
"step": 1169
|
8225 |
+
},
|
8226 |
+
{
|
8227 |
+
"epoch": 0.11604840309462408,
|
8228 |
+
"grad_norm": 1.9740146398544312,
|
8229 |
+
"learning_rate": 1.1103693970904805e-05,
|
8230 |
+
"loss": 43.0567,
|
8231 |
+
"step": 1170
|
8232 |
+
},
|
8233 |
+
{
|
8234 |
+
"epoch": 0.11614758976393573,
|
8235 |
+
"grad_norm": 2.2524871826171875,
|
8236 |
+
"learning_rate": 1.0998677673476642e-05,
|
8237 |
+
"loss": 43.2027,
|
8238 |
+
"step": 1171
|
8239 |
+
},
|
8240 |
+
{
|
8241 |
+
"epoch": 0.11624677643324736,
|
8242 |
+
"grad_norm": 1.826412558555603,
|
8243 |
+
"learning_rate": 1.0894131442205003e-05,
|
8244 |
+
"loss": 43.4454,
|
8245 |
+
"step": 1172
|
8246 |
+
},
|
8247 |
+
{
|
8248 |
+
"epoch": 0.11634596310255901,
|
8249 |
+
"grad_norm": 2.110656261444092,
|
8250 |
+
"learning_rate": 1.079005582925745e-05,
|
8251 |
+
"loss": 43.1023,
|
8252 |
+
"step": 1173
|
8253 |
+
},
|
8254 |
+
{
|
8255 |
+
"epoch": 0.11644514977187066,
|
8256 |
+
"grad_norm": 2.1876347064971924,
|
8257 |
+
"learning_rate": 1.0686451384315966e-05,
|
8258 |
+
"loss": 43.0696,
|
8259 |
+
"step": 1174
|
8260 |
+
},
|
8261 |
+
{
|
8262 |
+
"epoch": 0.11654433644118231,
|
8263 |
+
"grad_norm": 2.194808006286621,
|
8264 |
+
"learning_rate": 1.0583318654574004e-05,
|
8265 |
+
"loss": 43.0742,
|
8266 |
+
"step": 1175
|
8267 |
+
},
|
8268 |
+
{
|
8269 |
+
"epoch": 0.11664352311049395,
|
8270 |
+
"grad_norm": 2.1567115783691406,
|
8271 |
+
"learning_rate": 1.0480658184733694e-05,
|
8272 |
+
"loss": 43.542,
|
8273 |
+
"step": 1176
|
8274 |
+
},
|
8275 |
+
{
|
8276 |
+
"epoch": 0.1167427097798056,
|
8277 |
+
"grad_norm": 1.785593032836914,
|
8278 |
+
"learning_rate": 1.0378470517002824e-05,
|
8279 |
+
"loss": 43.488,
|
8280 |
+
"step": 1177
|
8281 |
+
},
|
8282 |
+
{
|
8283 |
+
"epoch": 0.11684189644911724,
|
8284 |
+
"grad_norm": 2.219076156616211,
|
8285 |
+
"learning_rate": 1.0276756191092064e-05,
|
8286 |
+
"loss": 43.3838,
|
8287 |
+
"step": 1178
|
8288 |
+
},
|
8289 |
+
{
|
8290 |
+
"epoch": 0.11694108311842888,
|
8291 |
+
"grad_norm": 1.965964436531067,
|
8292 |
+
"learning_rate": 1.017551574421215e-05,
|
8293 |
+
"loss": 43.5251,
|
8294 |
+
"step": 1179
|
8295 |
+
},
|
8296 |
+
{
|
8297 |
+
"epoch": 0.11704026978774053,
|
8298 |
+
"grad_norm": 2.0803263187408447,
|
8299 |
+
"learning_rate": 1.007474971107092e-05,
|
8300 |
+
"loss": 43.0246,
|
8301 |
+
"step": 1180
|
8302 |
+
},
|
8303 |
+
{
|
8304 |
+
"epoch": 0.11713945645705218,
|
8305 |
+
"grad_norm": 2.750941276550293,
|
8306 |
+
"learning_rate": 9.974458623870619e-06,
|
8307 |
+
"loss": 42.7347,
|
8308 |
+
"step": 1181
|
8309 |
+
},
|
8310 |
+
{
|
8311 |
+
"epoch": 0.11723864312636381,
|
8312 |
+
"grad_norm": 1.9899749755859375,
|
8313 |
+
"learning_rate": 9.874643012304984e-06,
|
8314 |
+
"loss": 43.1894,
|
8315 |
+
"step": 1182
|
8316 |
+
},
|
8317 |
+
{
|
8318 |
+
"epoch": 0.11733782979567546,
|
8319 |
+
"grad_norm": 2.3064157962799072,
|
8320 |
+
"learning_rate": 9.775303403556579e-06,
|
8321 |
+
"loss": 43.2956,
|
8322 |
+
"step": 1183
|
8323 |
+
},
|
8324 |
+
{
|
8325 |
+
"epoch": 0.11743701646498711,
|
8326 |
+
"grad_norm": 2.3210952281951904,
|
8327 |
+
"learning_rate": 9.676440322293845e-06,
|
8328 |
+
"loss": 43.1066,
|
8329 |
+
"step": 1184
|
8330 |
+
},
|
8331 |
+
{
|
8332 |
+
"epoch": 0.11753620313429874,
|
8333 |
+
"grad_norm": 2.1142942905426025,
|
8334 |
+
"learning_rate": 9.578054290668448e-06,
|
8335 |
+
"loss": 43.4249,
|
8336 |
+
"step": 1185
|
8337 |
+
},
|
8338 |
+
{
|
8339 |
+
"epoch": 0.1176353898036104,
|
8340 |
+
"grad_norm": 2.0365614891052246,
|
8341 |
+
"learning_rate": 9.480145828312503e-06,
|
8342 |
+
"loss": 43.4594,
|
8343 |
+
"step": 1186
|
8344 |
+
},
|
8345 |
+
{
|
8346 |
+
"epoch": 0.11773457647292204,
|
8347 |
+
"grad_norm": 2.168372392654419,
|
8348 |
+
"learning_rate": 9.382715452335777e-06,
|
8349 |
+
"loss": 43.3557,
|
8350 |
+
"step": 1187
|
8351 |
+
},
|
8352 |
+
{
|
8353 |
+
"epoch": 0.11783376314223368,
|
8354 |
+
"grad_norm": 2.244636297225952,
|
8355 |
+
"learning_rate": 9.285763677323034e-06,
|
8356 |
+
"loss": 43.191,
|
8357 |
+
"step": 1188
|
8358 |
+
},
|
8359 |
+
{
|
8360 |
+
"epoch": 0.11793294981154533,
|
8361 |
+
"grad_norm": 2.429145097732544,
|
8362 |
+
"learning_rate": 9.189291015331226e-06,
|
8363 |
+
"loss": 43.0815,
|
8364 |
+
"step": 1189
|
8365 |
+
},
|
8366 |
+
{
|
8367 |
+
"epoch": 0.11803213648085698,
|
8368 |
+
"grad_norm": 1.7875555753707886,
|
8369 |
+
"learning_rate": 9.09329797588686e-06,
|
8370 |
+
"loss": 43.2109,
|
8371 |
+
"step": 1190
|
8372 |
+
},
|
8373 |
+
{
|
8374 |
+
"epoch": 0.11813132315016862,
|
8375 |
+
"grad_norm": 2.499196767807007,
|
8376 |
+
"learning_rate": 8.997785065983322e-06,
|
8377 |
+
"loss": 43.1957,
|
8378 |
+
"step": 1191
|
8379 |
+
},
|
8380 |
+
{
|
8381 |
+
"epoch": 0.11823050981948026,
|
8382 |
+
"grad_norm": 2.726623773574829,
|
8383 |
+
"learning_rate": 8.902752790078117e-06,
|
8384 |
+
"loss": 43.4016,
|
8385 |
+
"step": 1192
|
8386 |
+
},
|
8387 |
+
{
|
8388 |
+
"epoch": 0.11832969648879191,
|
8389 |
+
"grad_norm": 2.1397037506103516,
|
8390 |
+
"learning_rate": 8.808201650090265e-06,
|
8391 |
+
"loss": 42.8903,
|
8392 |
+
"step": 1193
|
8393 |
+
},
|
8394 |
+
{
|
8395 |
+
"epoch": 0.11842888315810356,
|
8396 |
+
"grad_norm": 2.0924766063690186,
|
8397 |
+
"learning_rate": 8.714132145397647e-06,
|
8398 |
+
"loss": 43.3431,
|
8399 |
+
"step": 1194
|
8400 |
+
},
|
8401 |
+
{
|
8402 |
+
"epoch": 0.11852806982741519,
|
8403 |
+
"grad_norm": 2.0014185905456543,
|
8404 |
+
"learning_rate": 8.62054477283436e-06,
|
8405 |
+
"loss": 43.4458,
|
8406 |
+
"step": 1195
|
8407 |
+
},
|
8408 |
+
{
|
8409 |
+
"epoch": 0.11862725649672684,
|
8410 |
+
"grad_norm": 2.0613114833831787,
|
8411 |
+
"learning_rate": 8.527440026688028e-06,
|
8412 |
+
"loss": 43.2646,
|
8413 |
+
"step": 1196
|
8414 |
+
},
|
8415 |
+
{
|
8416 |
+
"epoch": 0.11872644316603849,
|
8417 |
+
"grad_norm": 2.042891263961792,
|
8418 |
+
"learning_rate": 8.434818398697375e-06,
|
8419 |
+
"loss": 43.5922,
|
8420 |
+
"step": 1197
|
8421 |
+
},
|
8422 |
+
{
|
8423 |
+
"epoch": 0.11882562983535012,
|
8424 |
+
"grad_norm": 1.7818388938903809,
|
8425 |
+
"learning_rate": 8.34268037804936e-06,
|
8426 |
+
"loss": 43.3734,
|
8427 |
+
"step": 1198
|
8428 |
+
},
|
8429 |
+
{
|
8430 |
+
"epoch": 0.11892481650466177,
|
8431 |
+
"grad_norm": 1.8697561025619507,
|
8432 |
+
"learning_rate": 8.251026451376841e-06,
|
8433 |
+
"loss": 43.3112,
|
8434 |
+
"step": 1199
|
8435 |
+
},
|
8436 |
+
{
|
8437 |
+
"epoch": 0.11902400317397342,
|
8438 |
+
"grad_norm": 2.050682544708252,
|
8439 |
+
"learning_rate": 8.159857102755918e-06,
|
8440 |
+
"loss": 43.5161,
|
8441 |
+
"step": 1200
|
8442 |
+
},
|
8443 |
+
{
|
8444 |
+
"epoch": 0.11912318984328506,
|
8445 |
+
"grad_norm": 2.090418577194214,
|
8446 |
+
"learning_rate": 8.06917281370323e-06,
|
8447 |
+
"loss": 43.2041,
|
8448 |
+
"step": 1201
|
8449 |
+
},
|
8450 |
+
{
|
8451 |
+
"epoch": 0.1192223765125967,
|
8452 |
+
"grad_norm": 2.0807907581329346,
|
8453 |
+
"learning_rate": 7.978974063173684e-06,
|
8454 |
+
"loss": 43.4141,
|
8455 |
+
"step": 1202
|
8456 |
+
},
|
8457 |
+
{
|
8458 |
+
"epoch": 0.11932156318190835,
|
8459 |
+
"grad_norm": 2.35972261428833,
|
8460 |
+
"learning_rate": 7.889261327557752e-06,
|
8461 |
+
"loss": 43.2823,
|
8462 |
+
"step": 1203
|
8463 |
+
},
|
8464 |
+
{
|
8465 |
+
"epoch": 0.11942074985121999,
|
8466 |
+
"grad_norm": 2.20231556892395,
|
8467 |
+
"learning_rate": 7.800035080678891e-06,
|
8468 |
+
"loss": 43.4965,
|
8469 |
+
"step": 1204
|
8470 |
+
},
|
8471 |
+
{
|
8472 |
+
"epoch": 0.11951993652053164,
|
8473 |
+
"grad_norm": 2.4803483486175537,
|
8474 |
+
"learning_rate": 7.711295793791285e-06,
|
8475 |
+
"loss": 43.3348,
|
8476 |
+
"step": 1205
|
8477 |
+
},
|
8478 |
+
{
|
8479 |
+
"epoch": 0.11961912318984329,
|
8480 |
+
"grad_norm": 2.5327541828155518,
|
8481 |
+
"learning_rate": 7.62304393557709e-06,
|
8482 |
+
"loss": 43.2968,
|
8483 |
+
"step": 1206
|
8484 |
+
},
|
8485 |
+
{
|
8486 |
+
"epoch": 0.11971830985915492,
|
8487 |
+
"grad_norm": 1.9711940288543701,
|
8488 |
+
"learning_rate": 7.535279972144116e-06,
|
8489 |
+
"loss": 43.1184,
|
8490 |
+
"step": 1207
|
8491 |
+
},
|
8492 |
+
{
|
8493 |
+
"epoch": 0.11981749652846657,
|
8494 |
+
"grad_norm": 2.119109869003296,
|
8495 |
+
"learning_rate": 7.4480043670233755e-06,
|
8496 |
+
"loss": 43.4068,
|
8497 |
+
"step": 1208
|
8498 |
+
},
|
8499 |
+
{
|
8500 |
+
"epoch": 0.11991668319777822,
|
8501 |
+
"grad_norm": 2.3444743156433105,
|
8502 |
+
"learning_rate": 7.36121758116649e-06,
|
8503 |
+
"loss": 42.9626,
|
8504 |
+
"step": 1209
|
8505 |
+
},
|
8506 |
+
{
|
8507 |
+
"epoch": 0.12001586986708987,
|
8508 |
+
"grad_norm": 2.241438150405884,
|
8509 |
+
"learning_rate": 7.274920072943425e-06,
|
8510 |
+
"loss": 43.5055,
|
8511 |
+
"step": 1210
|
8512 |
+
},
|
8513 |
+
{
|
8514 |
+
"epoch": 0.1201150565364015,
|
8515 |
+
"grad_norm": 2.450463056564331,
|
8516 |
+
"learning_rate": 7.189112298139988e-06,
|
8517 |
+
"loss": 43.307,
|
8518 |
+
"step": 1211
|
8519 |
+
},
|
8520 |
+
{
|
8521 |
+
"epoch": 0.12021424320571315,
|
8522 |
+
"grad_norm": 1.971405029296875,
|
8523 |
+
"learning_rate": 7.103794709955347e-06,
|
8524 |
+
"loss": 42.8255,
|
8525 |
+
"step": 1212
|
8526 |
+
},
|
8527 |
+
{
|
8528 |
+
"epoch": 0.1203134298750248,
|
8529 |
+
"grad_norm": 2.300004243850708,
|
8530 |
+
"learning_rate": 7.01896775899984e-06,
|
8531 |
+
"loss": 43.5371,
|
8532 |
+
"step": 1213
|
8533 |
+
},
|
8534 |
+
{
|
8535 |
+
"epoch": 0.12041261654433644,
|
8536 |
+
"grad_norm": 2.3139774799346924,
|
8537 |
+
"learning_rate": 6.93463189329242e-06,
|
8538 |
+
"loss": 43.2636,
|
8539 |
+
"step": 1214
|
8540 |
+
},
|
8541 |
+
{
|
8542 |
+
"epoch": 0.12051180321364809,
|
8543 |
+
"grad_norm": 2.28747296333313,
|
8544 |
+
"learning_rate": 6.850787558258287e-06,
|
8545 |
+
"loss": 43.4247,
|
8546 |
+
"step": 1215
|
8547 |
+
},
|
8548 |
+
{
|
8549 |
+
"epoch": 0.12061098988295973,
|
8550 |
+
"grad_norm": 1.9092282056808472,
|
8551 |
+
"learning_rate": 6.767435196726679e-06,
|
8552 |
+
"loss": 43.3869,
|
8553 |
+
"step": 1216
|
8554 |
+
},
|
8555 |
+
{
|
8556 |
+
"epoch": 0.12071017655227137,
|
8557 |
+
"grad_norm": 2.101717472076416,
|
8558 |
+
"learning_rate": 6.684575248928404e-06,
|
8559 |
+
"loss": 43.273,
|
8560 |
+
"step": 1217
|
8561 |
+
},
|
8562 |
+
{
|
8563 |
+
"epoch": 0.12080936322158302,
|
8564 |
+
"grad_norm": 2.3089897632598877,
|
8565 |
+
"learning_rate": 6.602208152493527e-06,
|
8566 |
+
"loss": 43.1768,
|
8567 |
+
"step": 1218
|
8568 |
+
},
|
8569 |
+
{
|
8570 |
+
"epoch": 0.12090854989089467,
|
8571 |
+
"grad_norm": 2.346907138824463,
|
8572 |
+
"learning_rate": 6.520334342449163e-06,
|
8573 |
+
"loss": 43.0728,
|
8574 |
+
"step": 1219
|
8575 |
+
},
|
8576 |
+
{
|
8577 |
+
"epoch": 0.1210077365602063,
|
8578 |
+
"grad_norm": 1.7760286331176758,
|
8579 |
+
"learning_rate": 6.438954251217033e-06,
|
8580 |
+
"loss": 43.3931,
|
8581 |
+
"step": 1220
|
8582 |
+
},
|
8583 |
+
{
|
8584 |
+
"epoch": 0.12110692322951795,
|
8585 |
+
"grad_norm": 1.7887672185897827,
|
8586 |
+
"learning_rate": 6.358068308611265e-06,
|
8587 |
+
"loss": 43.1055,
|
8588 |
+
"step": 1221
|
8589 |
+
},
|
8590 |
+
{
|
8591 |
+
"epoch": 0.1212061098988296,
|
8592 |
+
"grad_norm": 2.1375975608825684,
|
8593 |
+
"learning_rate": 6.277676941836141e-06,
|
8594 |
+
"loss": 43.3121,
|
8595 |
+
"step": 1222
|
8596 |
+
},
|
8597 |
+
{
|
8598 |
+
"epoch": 0.12130529656814124,
|
8599 |
+
"grad_norm": 2.0237154960632324,
|
8600 |
+
"learning_rate": 6.19778057548378e-06,
|
8601 |
+
"loss": 43.2777,
|
8602 |
+
"step": 1223
|
8603 |
+
},
|
8604 |
+
{
|
8605 |
+
"epoch": 0.12140448323745288,
|
8606 |
+
"grad_norm": 2.1526854038238525,
|
8607 |
+
"learning_rate": 6.118379631531923e-06,
|
8608 |
+
"loss": 43.238,
|
8609 |
+
"step": 1224
|
8610 |
+
},
|
8611 |
+
{
|
8612 |
+
"epoch": 0.12150366990676453,
|
8613 |
+
"grad_norm": 2.29803729057312,
|
8614 |
+
"learning_rate": 6.039474529341738e-06,
|
8615 |
+
"loss": 43.3275,
|
8616 |
+
"step": 1225
|
8617 |
+
},
|
8618 |
+
{
|
8619 |
+
"epoch": 0.12160285657607618,
|
8620 |
+
"grad_norm": 2.1667020320892334,
|
8621 |
+
"learning_rate": 5.961065685655543e-06,
|
8622 |
+
"loss": 43.5573,
|
8623 |
+
"step": 1226
|
8624 |
+
},
|
8625 |
+
{
|
8626 |
+
"epoch": 0.12170204324538782,
|
8627 |
+
"grad_norm": 2.034217357635498,
|
8628 |
+
"learning_rate": 5.883153514594641e-06,
|
8629 |
+
"loss": 43.5801,
|
8630 |
+
"step": 1227
|
8631 |
+
},
|
8632 |
+
{
|
8633 |
+
"epoch": 0.12180122991469947,
|
8634 |
+
"grad_norm": 1.9061791896820068,
|
8635 |
+
"learning_rate": 5.8057384276571365e-06,
|
8636 |
+
"loss": 42.9946,
|
8637 |
+
"step": 1228
|
8638 |
+
},
|
8639 |
+
{
|
8640 |
+
"epoch": 0.12190041658401111,
|
8641 |
+
"grad_norm": 2.0376012325286865,
|
8642 |
+
"learning_rate": 5.728820833715709e-06,
|
8643 |
+
"loss": 43.3239,
|
8644 |
+
"step": 1229
|
8645 |
+
},
|
8646 |
+
{
|
8647 |
+
"epoch": 0.12199960325332275,
|
8648 |
+
"grad_norm": 2.0292599201202393,
|
8649 |
+
"learning_rate": 5.652401139015584e-06,
|
8650 |
+
"loss": 43.2072,
|
8651 |
+
"step": 1230
|
8652 |
+
},
|
8653 |
+
{
|
8654 |
+
"epoch": 0.1220987899226344,
|
8655 |
+
"grad_norm": 2.264747142791748,
|
8656 |
+
"learning_rate": 5.576479747172225e-06,
|
8657 |
+
"loss": 42.9874,
|
8658 |
+
"step": 1231
|
8659 |
+
},
|
8660 |
+
{
|
8661 |
+
"epoch": 0.12219797659194605,
|
8662 |
+
"grad_norm": 2.0120532512664795,
|
8663 |
+
"learning_rate": 5.5010570591693015e-06,
|
8664 |
+
"loss": 43.1233,
|
8665 |
+
"step": 1232
|
8666 |
+
},
|
8667 |
+
{
|
8668 |
+
"epoch": 0.12229716326125768,
|
8669 |
+
"grad_norm": 2.6142144203186035,
|
8670 |
+
"learning_rate": 5.426133473356554e-06,
|
8671 |
+
"loss": 43.3979,
|
8672 |
+
"step": 1233
|
8673 |
+
},
|
8674 |
+
{
|
8675 |
+
"epoch": 0.12239634993056933,
|
8676 |
+
"grad_norm": 1.9740287065505981,
|
8677 |
+
"learning_rate": 5.3517093854476785e-06,
|
8678 |
+
"loss": 43.0373,
|
8679 |
+
"step": 1234
|
8680 |
+
},
|
8681 |
+
{
|
8682 |
+
"epoch": 0.12249553659988098,
|
8683 |
+
"grad_norm": 2.3029367923736572,
|
8684 |
+
"learning_rate": 5.277785188518225e-06,
|
8685 |
+
"loss": 43.1026,
|
8686 |
+
"step": 1235
|
8687 |
+
},
|
8688 |
+
{
|
8689 |
+
"epoch": 0.12259472326919262,
|
8690 |
+
"grad_norm": 2.306999444961548,
|
8691 |
+
"learning_rate": 5.204361273003555e-06,
|
8692 |
+
"loss": 43.4458,
|
8693 |
+
"step": 1236
|
8694 |
+
},
|
8695 |
+
{
|
8696 |
+
"epoch": 0.12269390993850426,
|
8697 |
+
"grad_norm": 2.1386356353759766,
|
8698 |
+
"learning_rate": 5.131438026696778e-06,
|
8699 |
+
"loss": 43.4215,
|
8700 |
+
"step": 1237
|
8701 |
+
},
|
8702 |
+
{
|
8703 |
+
"epoch": 0.12279309660781591,
|
8704 |
+
"grad_norm": 2.4391958713531494,
|
8705 |
+
"learning_rate": 5.059015834746639e-06,
|
8706 |
+
"loss": 43.1421,
|
8707 |
+
"step": 1238
|
8708 |
+
},
|
8709 |
+
{
|
8710 |
+
"epoch": 0.12289228327712755,
|
8711 |
+
"grad_norm": 1.901392936706543,
|
8712 |
+
"learning_rate": 4.9870950796555685e-06,
|
8713 |
+
"loss": 43.5324,
|
8714 |
+
"step": 1239
|
8715 |
+
},
|
8716 |
+
{
|
8717 |
+
"epoch": 0.1229914699464392,
|
8718 |
+
"grad_norm": 2.04669451713562,
|
8719 |
+
"learning_rate": 4.915676141277636e-06,
|
8720 |
+
"loss": 43.406,
|
8721 |
+
"step": 1240
|
8722 |
+
},
|
8723 |
+
{
|
8724 |
+
"epoch": 0.12309065661575085,
|
8725 |
+
"grad_norm": 2.3022654056549072,
|
8726 |
+
"learning_rate": 4.844759396816512e-06,
|
8727 |
+
"loss": 42.9503,
|
8728 |
+
"step": 1241
|
8729 |
+
},
|
8730 |
+
{
|
8731 |
+
"epoch": 0.12318984328506248,
|
8732 |
+
"grad_norm": 2.032792568206787,
|
8733 |
+
"learning_rate": 4.774345220823484e-06,
|
8734 |
+
"loss": 43.2597,
|
8735 |
+
"step": 1242
|
8736 |
+
},
|
8737 |
+
{
|
8738 |
+
"epoch": 0.12328902995437413,
|
8739 |
+
"grad_norm": 2.073387384414673,
|
8740 |
+
"learning_rate": 4.7044339851955115e-06,
|
8741 |
+
"loss": 43.4015,
|
8742 |
+
"step": 1243
|
8743 |
+
},
|
8744 |
+
{
|
8745 |
+
"epoch": 0.12338821662368578,
|
8746 |
+
"grad_norm": 2.2737884521484375,
|
8747 |
+
"learning_rate": 4.635026059173264e-06,
|
8748 |
+
"loss": 43.0576,
|
8749 |
+
"step": 1244
|
8750 |
+
},
|
8751 |
+
{
|
8752 |
+
"epoch": 0.12348740329299743,
|
8753 |
+
"grad_norm": 1.879482626914978,
|
8754 |
+
"learning_rate": 4.566121809339119e-06,
|
8755 |
+
"loss": 43.5151,
|
8756 |
+
"step": 1245
|
8757 |
+
},
|
8758 |
+
{
|
8759 |
+
"epoch": 0.12358658996230906,
|
8760 |
+
"grad_norm": 2.631582260131836,
|
8761 |
+
"learning_rate": 4.497721599615257e-06,
|
8762 |
+
"loss": 43.4226,
|
8763 |
+
"step": 1246
|
8764 |
+
},
|
8765 |
+
{
|
8766 |
+
"epoch": 0.12368577663162071,
|
8767 |
+
"grad_norm": 2.4985082149505615,
|
8768 |
+
"learning_rate": 4.429825791261733e-06,
|
8769 |
+
"loss": 43.2806,
|
8770 |
+
"step": 1247
|
8771 |
+
},
|
8772 |
+
{
|
8773 |
+
"epoch": 0.12378496330093236,
|
8774 |
+
"grad_norm": 2.055318593978882,
|
8775 |
+
"learning_rate": 4.3624347428746e-06,
|
8776 |
+
"loss": 43.2033,
|
8777 |
+
"step": 1248
|
8778 |
+
},
|
8779 |
+
{
|
8780 |
+
"epoch": 0.123884149970244,
|
8781 |
+
"grad_norm": 2.548267126083374,
|
8782 |
+
"learning_rate": 4.295548810383987e-06,
|
8783 |
+
"loss": 43.1069,
|
8784 |
+
"step": 1249
|
8785 |
+
},
|
8786 |
+
{
|
8787 |
+
"epoch": 0.12398333663955564,
|
8788 |
+
"grad_norm": 2.289154529571533,
|
8789 |
+
"learning_rate": 4.229168347052181e-06,
|
8790 |
+
"loss": 43.5297,
|
8791 |
+
"step": 1250
|
8792 |
+
},
|
8793 |
+
{
|
8794 |
+
"epoch": 0.12408252330886729,
|
8795 |
+
"grad_norm": 2.0629653930664062,
|
8796 |
+
"learning_rate": 4.163293703471871e-06,
|
8797 |
+
"loss": 43.3083,
|
8798 |
+
"step": 1251
|
8799 |
+
},
|
8800 |
+
{
|
8801 |
+
"epoch": 0.12418170997817893,
|
8802 |
+
"grad_norm": 2.202407121658325,
|
8803 |
+
"learning_rate": 4.097925227564148e-06,
|
8804 |
+
"loss": 43.3239,
|
8805 |
+
"step": 1252
|
8806 |
+
},
|
8807 |
+
{
|
8808 |
+
"epoch": 0.12428089664749058,
|
8809 |
+
"grad_norm": 2.0026113986968994,
|
8810 |
+
"learning_rate": 4.033063264576786e-06,
|
8811 |
+
"loss": 43.4304,
|
8812 |
+
"step": 1253
|
8813 |
+
},
|
8814 |
+
{
|
8815 |
+
"epoch": 0.12438008331680223,
|
8816 |
+
"grad_norm": 2.017918586730957,
|
8817 |
+
"learning_rate": 3.9687081570824126e-06,
|
8818 |
+
"loss": 43.1982,
|
8819 |
+
"step": 1254
|
8820 |
+
},
|
8821 |
+
{
|
8822 |
+
"epoch": 0.12447926998611386,
|
8823 |
+
"grad_norm": 2.305312156677246,
|
8824 |
+
"learning_rate": 3.904860244976582e-06,
|
8825 |
+
"loss": 43.1873,
|
8826 |
+
"step": 1255
|
8827 |
+
},
|
8828 |
+
{
|
8829 |
+
"epoch": 0.12457845665542551,
|
8830 |
+
"grad_norm": 2.2610886096954346,
|
8831 |
+
"learning_rate": 3.84151986547614e-06,
|
8832 |
+
"loss": 43.4435,
|
8833 |
+
"step": 1256
|
8834 |
+
},
|
8835 |
+
{
|
8836 |
+
"epoch": 0.12467764332473716,
|
8837 |
+
"grad_norm": 2.3130605220794678,
|
8838 |
+
"learning_rate": 3.7786873531173493e-06,
|
8839 |
+
"loss": 43.1467,
|
8840 |
+
"step": 1257
|
8841 |
+
},
|
8842 |
+
{
|
8843 |
+
"epoch": 0.1247768299940488,
|
8844 |
+
"grad_norm": 2.211864948272705,
|
8845 |
+
"learning_rate": 3.7163630397540893e-06,
|
8846 |
+
"loss": 43.3073,
|
8847 |
+
"step": 1258
|
8848 |
+
},
|
8849 |
+
{
|
8850 |
+
"epoch": 0.12487601666336044,
|
8851 |
+
"grad_norm": 2.169344663619995,
|
8852 |
+
"learning_rate": 3.6545472545562288e-06,
|
8853 |
+
"loss": 43.0595,
|
8854 |
+
"step": 1259
|
8855 |
+
},
|
8856 |
+
{
|
8857 |
+
"epoch": 0.12497520333267209,
|
8858 |
+
"grad_norm": 2.3061749935150146,
|
8859 |
+
"learning_rate": 3.593240324007774e-06,
|
8860 |
+
"loss": 43.1083,
|
8861 |
+
"step": 1260
|
8862 |
+
},
|
8863 |
+
{
|
8864 |
+
"epoch": 0.12507439000198373,
|
8865 |
+
"grad_norm": 2.1812314987182617,
|
8866 |
+
"learning_rate": 3.532442571905148e-06,
|
8867 |
+
"loss": 43.2467,
|
8868 |
+
"step": 1261
|
8869 |
+
},
|
8870 |
+
{
|
8871 |
+
"epoch": 0.1251735766712954,
|
8872 |
+
"grad_norm": 1.8057440519332886,
|
8873 |
+
"learning_rate": 3.4721543193556026e-06,
|
8874 |
+
"loss": 43.4519,
|
8875 |
+
"step": 1262
|
8876 |
+
},
|
8877 |
+
{
|
8878 |
+
"epoch": 0.12527276334060702,
|
8879 |
+
"grad_norm": 1.6972638368606567,
|
8880 |
+
"learning_rate": 3.412375884775365e-06,
|
8881 |
+
"loss": 43.1531,
|
8882 |
+
"step": 1263
|
8883 |
+
},
|
8884 |
+
{
|
8885 |
+
"epoch": 0.12537195000991866,
|
8886 |
+
"grad_norm": 2.08941912651062,
|
8887 |
+
"learning_rate": 3.3531075838880266e-06,
|
8888 |
+
"loss": 43.3175,
|
8889 |
+
"step": 1264
|
8890 |
+
},
|
8891 |
+
{
|
8892 |
+
"epoch": 0.12547113667923032,
|
8893 |
+
"grad_norm": 2.2320053577423096,
|
8894 |
+
"learning_rate": 3.2943497297229564e-06,
|
8895 |
+
"loss": 43.3922,
|
8896 |
+
"step": 1265
|
8897 |
+
},
|
8898 |
+
{
|
8899 |
+
"epoch": 0.12557032334854196,
|
8900 |
+
"grad_norm": 2.365736246109009,
|
8901 |
+
"learning_rate": 3.2361026326134357e-06,
|
8902 |
+
"loss": 43.5396,
|
8903 |
+
"step": 1266
|
8904 |
+
},
|
8905 |
+
{
|
8906 |
+
"epoch": 0.1256695100178536,
|
8907 |
+
"grad_norm": 2.5101640224456787,
|
8908 |
+
"learning_rate": 3.178366600195248e-06,
|
8909 |
+
"loss": 43.0945,
|
8910 |
+
"step": 1267
|
8911 |
+
},
|
8912 |
+
{
|
8913 |
+
"epoch": 0.12576869668716525,
|
8914 |
+
"grad_norm": 2.3893067836761475,
|
8915 |
+
"learning_rate": 3.121141937404959e-06,
|
8916 |
+
"loss": 43.1998,
|
8917 |
+
"step": 1268
|
8918 |
+
},
|
8919 |
+
{
|
8920 |
+
"epoch": 0.1258678833564769,
|
8921 |
+
"grad_norm": 1.9934214353561401,
|
8922 |
+
"learning_rate": 3.064428946478226e-06,
|
8923 |
+
"loss": 43.5163,
|
8924 |
+
"step": 1269
|
8925 |
+
},
|
8926 |
+
{
|
8927 |
+
"epoch": 0.12596707002578852,
|
8928 |
+
"grad_norm": 2.342822313308716,
|
8929 |
+
"learning_rate": 3.0082279269483484e-06,
|
8930 |
+
"loss": 43.2763,
|
8931 |
+
"step": 1270
|
8932 |
+
},
|
8933 |
+
{
|
8934 |
+
"epoch": 0.1260662566951002,
|
8935 |
+
"grad_norm": 2.3785641193389893,
|
8936 |
+
"learning_rate": 2.9525391756446086e-06,
|
8937 |
+
"loss": 43.1885,
|
8938 |
+
"step": 1271
|
8939 |
+
},
|
8940 |
+
{
|
8941 |
+
"epoch": 0.12616544336441182,
|
8942 |
+
"grad_norm": 2.2609450817108154,
|
8943 |
+
"learning_rate": 2.8973629866906548e-06,
|
8944 |
+
"loss": 43.0992,
|
8945 |
+
"step": 1272
|
8946 |
+
},
|
8947 |
+
{
|
8948 |
+
"epoch": 0.12626463003372346,
|
8949 |
+
"grad_norm": 2.203937292098999,
|
8950 |
+
"learning_rate": 2.842699651503067e-06,
|
8951 |
+
"loss": 43.2518,
|
8952 |
+
"step": 1273
|
8953 |
+
},
|
8954 |
+
{
|
8955 |
+
"epoch": 0.12636381670303512,
|
8956 |
+
"grad_norm": 1.8921879529953003,
|
8957 |
+
"learning_rate": 2.7885494587897465e-06,
|
8958 |
+
"loss": 42.9348,
|
8959 |
+
"step": 1274
|
8960 |
+
},
|
8961 |
+
{
|
8962 |
+
"epoch": 0.12646300337234676,
|
8963 |
+
"grad_norm": 2.4242756366729736,
|
8964 |
+
"learning_rate": 2.734912694548364e-06,
|
8965 |
+
"loss": 43.0746,
|
8966 |
+
"step": 1275
|
8967 |
+
},
|
8968 |
+
{
|
8969 |
+
"epoch": 0.1265621900416584,
|
8970 |
+
"grad_norm": 1.9764368534088135,
|
8971 |
+
"learning_rate": 2.681789642064936e-06,
|
8972 |
+
"loss": 43.1359,
|
8973 |
+
"step": 1276
|
8974 |
+
},
|
8975 |
+
{
|
8976 |
+
"epoch": 0.12666137671097005,
|
8977 |
+
"grad_norm": 2.145023822784424,
|
8978 |
+
"learning_rate": 2.6291805819122273e-06,
|
8979 |
+
"loss": 43.594,
|
8980 |
+
"step": 1277
|
8981 |
+
},
|
8982 |
+
{
|
8983 |
+
"epoch": 0.1267605633802817,
|
8984 |
+
"grad_norm": 2.1979358196258545,
|
8985 |
+
"learning_rate": 2.5770857919483416e-06,
|
8986 |
+
"loss": 43.0942,
|
8987 |
+
"step": 1278
|
8988 |
+
},
|
8989 |
+
{
|
8990 |
+
"epoch": 0.12685975004959332,
|
8991 |
+
"grad_norm": 2.368830680847168,
|
8992 |
+
"learning_rate": 2.5255055473152543e-06,
|
8993 |
+
"loss": 43.3192,
|
8994 |
+
"step": 1279
|
8995 |
+
},
|
8996 |
+
{
|
8997 |
+
"epoch": 0.12695893671890499,
|
8998 |
+
"grad_norm": 2.20819354057312,
|
8999 |
+
"learning_rate": 2.4744401204372937e-06,
|
9000 |
+
"loss": 43.2989,
|
9001 |
+
"step": 1280
|
9002 |
+
},
|
9003 |
+
{
|
9004 |
+
"epoch": 0.12705812338821662,
|
9005 |
+
"grad_norm": 2.263983964920044,
|
9006 |
+
"learning_rate": 2.4238897810197726e-06,
|
9007 |
+
"loss": 43.3038,
|
9008 |
+
"step": 1281
|
9009 |
+
},
|
9010 |
+
{
|
9011 |
+
"epoch": 0.12715731005752826,
|
9012 |
+
"grad_norm": 2.213432788848877,
|
9013 |
+
"learning_rate": 2.3738547960475365e-06,
|
9014 |
+
"loss": 43.1987,
|
9015 |
+
"step": 1282
|
9016 |
+
},
|
9017 |
+
{
|
9018 |
+
"epoch": 0.12725649672683992,
|
9019 |
+
"grad_norm": 2.0271692276000977,
|
9020 |
+
"learning_rate": 2.324335429783542e-06,
|
9021 |
+
"loss": 43.5205,
|
9022 |
+
"step": 1283
|
9023 |
+
},
|
9024 |
+
{
|
9025 |
+
"epoch": 0.12735568339615155,
|
9026 |
+
"grad_norm": 1.8703464269638062,
|
9027 |
+
"learning_rate": 2.275331943767478e-06,
|
9028 |
+
"loss": 43.2703,
|
9029 |
+
"step": 1284
|
9030 |
+
},
|
9031 |
+
{
|
9032 |
+
"epoch": 0.1274548700654632,
|
9033 |
+
"grad_norm": 2.126736879348755,
|
9034 |
+
"learning_rate": 2.2268445968143815e-06,
|
9035 |
+
"loss": 43.4308,
|
9036 |
+
"step": 1285
|
9037 |
+
},
|
9038 |
+
{
|
9039 |
+
"epoch": 0.12755405673477485,
|
9040 |
+
"grad_norm": 2.1270225048065186,
|
9041 |
+
"learning_rate": 2.1788736450132574e-06,
|
9042 |
+
"loss": 42.7397,
|
9043 |
+
"step": 1286
|
9044 |
+
},
|
9045 |
+
{
|
9046 |
+
"epoch": 0.12765324340408649,
|
9047 |
+
"grad_norm": 2.07893705368042,
|
9048 |
+
"learning_rate": 2.131419341725749e-06,
|
9049 |
+
"loss": 43.1869,
|
9050 |
+
"step": 1287
|
9051 |
+
},
|
9052 |
+
{
|
9053 |
+
"epoch": 0.12775243007339815,
|
9054 |
+
"grad_norm": 1.8926030397415161,
|
9055 |
+
"learning_rate": 2.0844819375848035e-06,
|
9056 |
+
"loss": 43.4108,
|
9057 |
+
"step": 1288
|
9058 |
+
},
|
9059 |
+
{
|
9060 |
+
"epoch": 0.12785161674270978,
|
9061 |
+
"grad_norm": 2.1941773891448975,
|
9062 |
+
"learning_rate": 2.038061680493275e-06,
|
9063 |
+
"loss": 43.2738,
|
9064 |
+
"step": 1289
|
9065 |
+
},
|
9066 |
+
{
|
9067 |
+
"epoch": 0.12795080341202142,
|
9068 |
+
"grad_norm": 2.055089235305786,
|
9069 |
+
"learning_rate": 1.9921588156227466e-06,
|
9070 |
+
"loss": 43.1291,
|
9071 |
+
"step": 1290
|
9072 |
+
},
|
9073 |
+
{
|
9074 |
+
"epoch": 0.12804999008133308,
|
9075 |
+
"grad_norm": 2.207758665084839,
|
9076 |
+
"learning_rate": 1.9467735854121094e-06,
|
9077 |
+
"loss": 43.4069,
|
9078 |
+
"step": 1291
|
9079 |
+
},
|
9080 |
+
{
|
9081 |
+
"epoch": 0.12814917675064472,
|
9082 |
+
"grad_norm": 2.0695319175720215,
|
9083 |
+
"learning_rate": 1.9019062295663525e-06,
|
9084 |
+
"loss": 43.5762,
|
9085 |
+
"step": 1292
|
9086 |
+
},
|
9087 |
+
{
|
9088 |
+
"epoch": 0.12824836341995635,
|
9089 |
+
"grad_norm": 2.1350889205932617,
|
9090 |
+
"learning_rate": 1.8575569850552644e-06,
|
9091 |
+
"loss": 42.9807,
|
9092 |
+
"step": 1293
|
9093 |
+
},
|
9094 |
+
{
|
9095 |
+
"epoch": 0.12834755008926801,
|
9096 |
+
"grad_norm": 2.0793445110321045,
|
9097 |
+
"learning_rate": 1.8137260861122108e-06,
|
9098 |
+
"loss": 43.2064,
|
9099 |
+
"step": 1294
|
9100 |
+
},
|
9101 |
+
{
|
9102 |
+
"epoch": 0.12844673675857965,
|
9103 |
+
"grad_norm": 1.8201216459274292,
|
9104 |
+
"learning_rate": 1.7704137642328921e-06,
|
9105 |
+
"loss": 43.3127,
|
9106 |
+
"step": 1295
|
9107 |
+
},
|
9108 |
+
{
|
9109 |
+
"epoch": 0.12854592342789128,
|
9110 |
+
"grad_norm": 1.9616897106170654,
|
9111 |
+
"learning_rate": 1.7276202481740777e-06,
|
9112 |
+
"loss": 43.2338,
|
9113 |
+
"step": 1296
|
9114 |
+
},
|
9115 |
+
{
|
9116 |
+
"epoch": 0.12864511009720295,
|
9117 |
+
"grad_norm": 2.18774151802063,
|
9118 |
+
"learning_rate": 1.6853457639524617e-06,
|
9119 |
+
"loss": 43.0747,
|
9120 |
+
"step": 1297
|
9121 |
+
},
|
9122 |
+
{
|
9123 |
+
"epoch": 0.12874429676651458,
|
9124 |
+
"grad_norm": 2.114123582839966,
|
9125 |
+
"learning_rate": 1.6435905348434311e-06,
|
9126 |
+
"loss": 43.5173,
|
9127 |
+
"step": 1298
|
9128 |
+
},
|
9129 |
+
{
|
9130 |
+
"epoch": 0.12884348343582622,
|
9131 |
+
"grad_norm": 1.9579181671142578,
|
9132 |
+
"learning_rate": 1.6023547813798778e-06,
|
9133 |
+
"loss": 43.745,
|
9134 |
+
"step": 1299
|
9135 |
+
},
|
9136 |
+
{
|
9137 |
+
"epoch": 0.12894267010513788,
|
9138 |
+
"grad_norm": 2.646709680557251,
|
9139 |
+
"learning_rate": 1.5616387213510663e-06,
|
9140 |
+
"loss": 43.2046,
|
9141 |
+
"step": 1300
|
9142 |
+
},
|
9143 |
+
{
|
9144 |
+
"epoch": 0.12904185677444951,
|
9145 |
+
"grad_norm": 2.432769775390625,
|
9146 |
+
"learning_rate": 1.5214425698014568e-06,
|
9147 |
+
"loss": 43.4605,
|
9148 |
+
"step": 1301
|
9149 |
+
},
|
9150 |
+
{
|
9151 |
+
"epoch": 0.12914104344376115,
|
9152 |
+
"grad_norm": 1.9655414819717407,
|
9153 |
+
"learning_rate": 1.4817665390296053e-06,
|
9154 |
+
"loss": 43.1898,
|
9155 |
+
"step": 1302
|
9156 |
+
},
|
9157 |
+
{
|
9158 |
+
"epoch": 0.1292402301130728,
|
9159 |
+
"grad_norm": 2.2822318077087402,
|
9160 |
+
"learning_rate": 1.4426108385869774e-06,
|
9161 |
+
"loss": 43.1048,
|
9162 |
+
"step": 1303
|
9163 |
+
},
|
9164 |
+
{
|
9165 |
+
"epoch": 0.12933941678238445,
|
9166 |
+
"grad_norm": 2.103745698928833,
|
9167 |
+
"learning_rate": 1.4039756752769029e-06,
|
9168 |
+
"loss": 43.425,
|
9169 |
+
"step": 1304
|
9170 |
+
},
|
9171 |
+
{
|
9172 |
+
"epoch": 0.12943860345169608,
|
9173 |
+
"grad_norm": 2.0974647998809814,
|
9174 |
+
"learning_rate": 1.3658612531534666e-06,
|
9175 |
+
"loss": 43.4766,
|
9176 |
+
"step": 1305
|
9177 |
+
},
|
9178 |
+
{
|
9179 |
+
"epoch": 0.12953779012100775,
|
9180 |
+
"grad_norm": 2.3900954723358154,
|
9181 |
+
"learning_rate": 1.3282677735204197e-06,
|
9182 |
+
"loss": 42.895,
|
9183 |
+
"step": 1306
|
9184 |
+
},
|
9185 |
+
{
|
9186 |
+
"epoch": 0.12963697679031938,
|
9187 |
+
"grad_norm": 1.7264951467514038,
|
9188 |
+
"learning_rate": 1.2911954349301148e-06,
|
9189 |
+
"loss": 43.3442,
|
9190 |
+
"step": 1307
|
9191 |
+
},
|
9192 |
+
{
|
9193 |
+
"epoch": 0.12973616345963102,
|
9194 |
+
"grad_norm": 2.0513551235198975,
|
9195 |
+
"learning_rate": 1.2546444331824837e-06,
|
9196 |
+
"loss": 43.3262,
|
9197 |
+
"step": 1308
|
9198 |
+
},
|
9199 |
+
{
|
9200 |
+
"epoch": 0.12983535012894268,
|
9201 |
+
"grad_norm": 1.866257667541504,
|
9202 |
+
"learning_rate": 1.2186149613239385e-06,
|
9203 |
+
"loss": 43.4503,
|
9204 |
+
"step": 1309
|
9205 |
+
},
|
9206 |
+
{
|
9207 |
+
"epoch": 0.1299345367982543,
|
9208 |
+
"grad_norm": 1.9009063243865967,
|
9209 |
+
"learning_rate": 1.1831072096464612e-06,
|
9210 |
+
"loss": 43.0886,
|
9211 |
+
"step": 1310
|
9212 |
+
},
|
9213 |
+
{
|
9214 |
+
"epoch": 0.13003372346756595,
|
9215 |
+
"grad_norm": 2.2211618423461914,
|
9216 |
+
"learning_rate": 1.1481213656865164e-06,
|
9217 |
+
"loss": 43.3954,
|
9218 |
+
"step": 1311
|
9219 |
+
},
|
9220 |
+
{
|
9221 |
+
"epoch": 0.1301329101368776,
|
9222 |
+
"grad_norm": 2.0510098934173584,
|
9223 |
+
"learning_rate": 1.1136576142240286e-06,
|
9224 |
+
"loss": 43.5047,
|
9225 |
+
"step": 1312
|
9226 |
+
},
|
9227 |
+
{
|
9228 |
+
"epoch": 0.13023209680618925,
|
9229 |
+
"grad_norm": 1.779088020324707,
|
9230 |
+
"learning_rate": 1.079716137281528e-06,
|
9231 |
+
"loss": 43.3659,
|
9232 |
+
"step": 1313
|
9233 |
+
},
|
9234 |
+
{
|
9235 |
+
"epoch": 0.13033128347550088,
|
9236 |
+
"grad_norm": 2.585233211517334,
|
9237 |
+
"learning_rate": 1.0462971141230849e-06,
|
9238 |
+
"loss": 43.1262,
|
9239 |
+
"step": 1314
|
9240 |
+
},
|
9241 |
+
{
|
9242 |
+
"epoch": 0.13043047014481254,
|
9243 |
+
"grad_norm": 2.646188497543335,
|
9244 |
+
"learning_rate": 1.0134007212533769e-06,
|
9245 |
+
"loss": 43.0961,
|
9246 |
+
"step": 1315
|
9247 |
+
},
|
9248 |
+
{
|
9249 |
+
"epoch": 0.13052965681412418,
|
9250 |
+
"grad_norm": 2.085897207260132,
|
9251 |
+
"learning_rate": 9.81027132416834e-07,
|
9252 |
+
"loss": 43.4507,
|
9253 |
+
"step": 1316
|
9254 |
+
},
|
9255 |
+
{
|
9256 |
+
"epoch": 0.1306288434834358,
|
9257 |
+
"grad_norm": 2.3529210090637207,
|
9258 |
+
"learning_rate": 9.491765185966173e-07,
|
9259 |
+
"loss": 43.3753,
|
9260 |
+
"step": 1317
|
9261 |
+
},
|
9262 |
+
{
|
9263 |
+
"epoch": 0.13072803015274748,
|
9264 |
+
"grad_norm": 2.244891881942749,
|
9265 |
+
"learning_rate": 9.178490480137525e-07,
|
9266 |
+
"loss": 43.365,
|
9267 |
+
"step": 1318
|
9268 |
+
},
|
9269 |
+
{
|
9270 |
+
"epoch": 0.1308272168220591,
|
9271 |
+
"grad_norm": 2.257857084274292,
|
9272 |
+
"learning_rate": 8.870448861263092e-07,
|
9273 |
+
"loss": 43.436,
|
9274 |
+
"step": 1319
|
9275 |
+
},
|
9276 |
+
{
|
9277 |
+
"epoch": 0.13092640349137075,
|
9278 |
+
"grad_norm": 2.1860482692718506,
|
9279 |
+
"learning_rate": 8.567641956283901e-07,
|
9280 |
+
"loss": 43.3999,
|
9281 |
+
"step": 1320
|
9282 |
+
},
|
9283 |
+
{
|
9284 |
+
"epoch": 0.1310255901606824,
|
9285 |
+
"grad_norm": 2.1375296115875244,
|
9286 |
+
"learning_rate": 8.270071364494203e-07,
|
9287 |
+
"loss": 43.4638,
|
9288 |
+
"step": 1321
|
9289 |
+
},
|
9290 |
+
{
|
9291 |
+
"epoch": 0.13112477682999404,
|
9292 |
+
"grad_norm": 2.1761748790740967,
|
9293 |
+
"learning_rate": 7.977738657532152e-07,
|
9294 |
+
"loss": 43.5287,
|
9295 |
+
"step": 1322
|
9296 |
+
},
|
9297 |
+
{
|
9298 |
+
"epoch": 0.1312239634993057,
|
9299 |
+
"grad_norm": 2.061898708343506,
|
9300 |
+
"learning_rate": 7.690645379371253e-07,
|
9301 |
+
"loss": 43.471,
|
9302 |
+
"step": 1323
|
9303 |
+
},
|
9304 |
+
{
|
9305 |
+
"epoch": 0.13132315016861734,
|
9306 |
+
"grad_norm": 2.035172700881958,
|
9307 |
+
"learning_rate": 7.408793046313034e-07,
|
9308 |
+
"loss": 43.3537,
|
9309 |
+
"step": 1324
|
9310 |
+
},
|
9311 |
+
{
|
9312 |
+
"epoch": 0.13142233683792898,
|
9313 |
+
"grad_norm": 2.6309173107147217,
|
9314 |
+
"learning_rate": 7.132183146978722e-07,
|
9315 |
+
"loss": 42.725,
|
9316 |
+
"step": 1325
|
9317 |
+
},
|
9318 |
+
{
|
9319 |
+
"epoch": 0.13152152350724064,
|
9320 |
+
"grad_norm": 2.411343574523926,
|
9321 |
+
"learning_rate": 6.860817142301024e-07,
|
9322 |
+
"loss": 43.3168,
|
9323 |
+
"step": 1326
|
9324 |
+
},
|
9325 |
+
{
|
9326 |
+
"epoch": 0.13162071017655227,
|
9327 |
+
"grad_norm": 2.1707561016082764,
|
9328 |
+
"learning_rate": 6.594696465516692e-07,
|
9329 |
+
"loss": 43.1672,
|
9330 |
+
"step": 1327
|
9331 |
+
},
|
9332 |
+
{
|
9333 |
+
"epoch": 0.1317198968458639,
|
9334 |
+
"grad_norm": 2.18800950050354,
|
9335 |
+
"learning_rate": 6.333822522159306e-07,
|
9336 |
+
"loss": 43.314,
|
9337 |
+
"step": 1328
|
9338 |
+
},
|
9339 |
+
{
|
9340 |
+
"epoch": 0.13181908351517557,
|
9341 |
+
"grad_norm": 2.0691611766815186,
|
9342 |
+
"learning_rate": 6.078196690050942e-07,
|
9343 |
+
"loss": 43.3585,
|
9344 |
+
"step": 1329
|
9345 |
+
},
|
9346 |
+
{
|
9347 |
+
"epoch": 0.1319182701844872,
|
9348 |
+
"grad_norm": 2.176452398300171,
|
9349 |
+
"learning_rate": 5.827820319295851e-07,
|
9350 |
+
"loss": 43.3425,
|
9351 |
+
"step": 1330
|
9352 |
+
},
|
9353 |
+
{
|
9354 |
+
"epoch": 0.13201745685379884,
|
9355 |
+
"grad_norm": 2.120635747909546,
|
9356 |
+
"learning_rate": 5.582694732272797e-07,
|
9357 |
+
"loss": 43.2628,
|
9358 |
+
"step": 1331
|
9359 |
+
},
|
9360 |
+
{
|
9361 |
+
"epoch": 0.1321166435231105,
|
9362 |
+
"grad_norm": 2.1084201335906982,
|
9363 |
+
"learning_rate": 5.342821223628169e-07,
|
9364 |
+
"loss": 43.3292,
|
9365 |
+
"step": 1332
|
9366 |
+
},
|
9367 |
+
{
|
9368 |
+
"epoch": 0.13221583019242214,
|
9369 |
+
"grad_norm": 2.155085802078247,
|
9370 |
+
"learning_rate": 5.108201060269324e-07,
|
9371 |
+
"loss": 43.587,
|
9372 |
+
"step": 1333
|
9373 |
+
},
|
9374 |
+
{
|
9375 |
+
"epoch": 0.13231501686173378,
|
9376 |
+
"grad_norm": 2.179335355758667,
|
9377 |
+
"learning_rate": 4.878835481357369e-07,
|
9378 |
+
"loss": 43.4465,
|
9379 |
+
"step": 1334
|
9380 |
+
},
|
9381 |
+
{
|
9382 |
+
"epoch": 0.13241420353104544,
|
9383 |
+
"grad_norm": 2.027651786804199,
|
9384 |
+
"learning_rate": 4.6547256983013877e-07,
|
9385 |
+
"loss": 43.3556,
|
9386 |
+
"step": 1335
|
9387 |
+
},
|
9388 |
+
{
|
9389 |
+
"epoch": 0.13251339020035707,
|
9390 |
+
"grad_norm": 1.9948636293411255,
|
9391 |
+
"learning_rate": 4.435872894751447e-07,
|
9392 |
+
"loss": 43.4829,
|
9393 |
+
"step": 1336
|
9394 |
+
},
|
9395 |
+
{
|
9396 |
+
"epoch": 0.1326125768696687,
|
9397 |
+
"grad_norm": 1.868950605392456,
|
9398 |
+
"learning_rate": 4.222278226592491e-07,
|
9399 |
+
"loss": 43.2998,
|
9400 |
+
"step": 1337
|
9401 |
+
},
|
9402 |
+
{
|
9403 |
+
"epoch": 0.13271176353898037,
|
9404 |
+
"grad_norm": 1.9464658498764038,
|
9405 |
+
"learning_rate": 4.0139428219385656e-07,
|
9406 |
+
"loss": 43.1973,
|
9407 |
+
"step": 1338
|
9408 |
+
},
|
9409 |
+
{
|
9410 |
+
"epoch": 0.132810950208292,
|
9411 |
+
"grad_norm": 2.3970725536346436,
|
9412 |
+
"learning_rate": 3.8108677811262705e-07,
|
9413 |
+
"loss": 43.1581,
|
9414 |
+
"step": 1339
|
9415 |
+
},
|
9416 |
+
{
|
9417 |
+
"epoch": 0.13291013687760364,
|
9418 |
+
"grad_norm": 2.359151840209961,
|
9419 |
+
"learning_rate": 3.6130541767093183e-07,
|
9420 |
+
"loss": 43.3907,
|
9421 |
+
"step": 1340
|
9422 |
+
},
|
9423 |
+
{
|
9424 |
+
"epoch": 0.1330093235469153,
|
9425 |
+
"grad_norm": 2.001293659210205,
|
9426 |
+
"learning_rate": 3.4205030534529834e-07,
|
9427 |
+
"loss": 43.6698,
|
9428 |
+
"step": 1341
|
9429 |
+
},
|
9430 |
+
{
|
9431 |
+
"epoch": 0.13310851021622694,
|
9432 |
+
"grad_norm": 2.3657517433166504,
|
9433 |
+
"learning_rate": 3.233215428328329e-07,
|
9434 |
+
"loss": 42.9345,
|
9435 |
+
"step": 1342
|
9436 |
+
},
|
9437 |
+
{
|
9438 |
+
"epoch": 0.13320769688553857,
|
9439 |
+
"grad_norm": 2.348754644393921,
|
9440 |
+
"learning_rate": 3.0511922905066547e-07,
|
9441 |
+
"loss": 43.3322,
|
9442 |
+
"step": 1343
|
9443 |
+
},
|
9444 |
+
{
|
9445 |
+
"epoch": 0.13330688355485024,
|
9446 |
+
"grad_norm": 2.3029611110687256,
|
9447 |
+
"learning_rate": 2.8744346013550583e-07,
|
9448 |
+
"loss": 43.4514,
|
9449 |
+
"step": 1344
|
9450 |
+
},
|
9451 |
+
{
|
9452 |
+
"epoch": 0.13340607022416187,
|
9453 |
+
"grad_norm": 2.1175925731658936,
|
9454 |
+
"learning_rate": 2.702943294430327e-07,
|
9455 |
+
"loss": 43.2903,
|
9456 |
+
"step": 1345
|
9457 |
+
},
|
9458 |
+
{
|
9459 |
+
"epoch": 0.1335052568934735,
|
9460 |
+
"grad_norm": 2.0833818912506104,
|
9461 |
+
"learning_rate": 2.5367192754747193e-07,
|
9462 |
+
"loss": 43.3412,
|
9463 |
+
"step": 1346
|
9464 |
+
},
|
9465 |
+
{
|
9466 |
+
"epoch": 0.13360444356278517,
|
9467 |
+
"grad_norm": 2.0652518272399902,
|
9468 |
+
"learning_rate": 2.3757634224111924e-07,
|
9469 |
+
"loss": 43.4184,
|
9470 |
+
"step": 1347
|
9471 |
+
},
|
9472 |
+
{
|
9473 |
+
"epoch": 0.1337036302320968,
|
9474 |
+
"grad_norm": 1.9658844470977783,
|
9475 |
+
"learning_rate": 2.2200765853381823e-07,
|
9476 |
+
"loss": 43.2789,
|
9477 |
+
"step": 1348
|
9478 |
+
},
|
9479 |
+
{
|
9480 |
+
"epoch": 0.13380281690140844,
|
9481 |
+
"grad_norm": 2.1758694648742676,
|
9482 |
+
"learning_rate": 2.0696595865256076e-07,
|
9483 |
+
"loss": 43.5539,
|
9484 |
+
"step": 1349
|
9485 |
+
},
|
9486 |
+
{
|
9487 |
+
"epoch": 0.1339020035707201,
|
9488 |
+
"grad_norm": 2.309638500213623,
|
9489 |
+
"learning_rate": 1.9245132204105398e-07,
|
9490 |
+
"loss": 43.2878,
|
9491 |
+
"step": 1350
|
9492 |
+
},
|
9493 |
+
{
|
9494 |
+
"epoch": 0.13400119024003174,
|
9495 |
+
"grad_norm": 2.347503900527954,
|
9496 |
+
"learning_rate": 1.784638253592763e-07,
|
9497 |
+
"loss": 42.9477,
|
9498 |
+
"step": 1351
|
9499 |
+
},
|
9500 |
+
{
|
9501 |
+
"epoch": 0.13410037690934337,
|
9502 |
+
"grad_norm": 2.0042808055877686,
|
9503 |
+
"learning_rate": 1.650035424830776e-07,
|
9504 |
+
"loss": 43.1693,
|
9505 |
+
"step": 1352
|
9506 |
+
},
|
9507 |
+
{
|
9508 |
+
"epoch": 0.13419956357865503,
|
9509 |
+
"grad_norm": 1.9504557847976685,
|
9510 |
+
"learning_rate": 1.5207054450380175e-07,
|
9511 |
+
"loss": 43.3698,
|
9512 |
+
"step": 1353
|
9513 |
+
},
|
9514 |
+
{
|
9515 |
+
"epoch": 0.13429875024796667,
|
9516 |
+
"grad_norm": 2.324517011642456,
|
9517 |
+
"learning_rate": 1.396648997278982e-07,
|
9518 |
+
"loss": 43.4977,
|
9519 |
+
"step": 1354
|
9520 |
+
},
|
9521 |
+
{
|
9522 |
+
"epoch": 0.1343979369172783,
|
9523 |
+
"grad_norm": 2.1884117126464844,
|
9524 |
+
"learning_rate": 1.277866736765776e-07,
|
9525 |
+
"loss": 42.9123,
|
9526 |
+
"step": 1355
|
9527 |
+
},
|
9528 |
+
{
|
9529 |
+
"epoch": 0.13449712358658997,
|
9530 |
+
"grad_norm": 2.5567128658294678,
|
9531 |
+
"learning_rate": 1.1643592908544554e-07,
|
9532 |
+
"loss": 43.4158,
|
9533 |
+
"step": 1356
|
9534 |
+
},
|
9535 |
+
{
|
9536 |
+
"epoch": 0.1345963102559016,
|
9537 |
+
"grad_norm": 1.7337242364883423,
|
9538 |
+
"learning_rate": 1.0561272590418058e-07,
|
9539 |
+
"loss": 43.1967,
|
9540 |
+
"step": 1357
|
9541 |
+
},
|
9542 |
+
{
|
9543 |
+
"epoch": 0.13469549692521327,
|
9544 |
+
"grad_norm": 1.987276315689087,
|
9545 |
+
"learning_rate": 9.531712129621229e-08,
|
9546 |
+
"loss": 43.4314,
|
9547 |
+
"step": 1358
|
9548 |
+
},
|
9549 |
+
{
|
9550 |
+
"epoch": 0.1347946835945249,
|
9551 |
+
"grad_norm": 1.984566569328308,
|
9552 |
+
"learning_rate": 8.554916963845472e-08,
|
9553 |
+
"loss": 43.4567,
|
9554 |
+
"step": 1359
|
9555 |
+
},
|
9556 |
+
{
|
9557 |
+
"epoch": 0.13489387026383654,
|
9558 |
+
"grad_norm": 2.2458252906799316,
|
9559 |
+
"learning_rate": 7.630892252095124e-08,
|
9560 |
+
"loss": 43.2742,
|
9561 |
+
"step": 1360
|
9562 |
+
},
|
9563 |
+
{
|
9564 |
+
"epoch": 0.1349930569331482,
|
9565 |
+
"grad_norm": 2.194035530090332,
|
9566 |
+
"learning_rate": 6.759642874665239e-08,
|
9567 |
+
"loss": 43.3974,
|
9568 |
+
"step": 1361
|
9569 |
+
},
|
9570 |
+
{
|
9571 |
+
"epoch": 0.13509224360245983,
|
9572 |
+
"grad_norm": 2.0920443534851074,
|
9573 |
+
"learning_rate": 5.941173433116065e-08,
|
9574 |
+
"loss": 43.4842,
|
9575 |
+
"step": 1362
|
9576 |
+
},
|
9577 |
+
{
|
9578 |
+
"epoch": 0.13519143027177147,
|
9579 |
+
"grad_norm": 1.9753786325454712,
|
9580 |
+
"learning_rate": 5.175488250245275e-08,
|
9581 |
+
"loss": 43.3207,
|
9582 |
+
"step": 1363
|
9583 |
+
},
|
9584 |
+
{
|
9585 |
+
"epoch": 0.13529061694108313,
|
9586 |
+
"grad_norm": 1.636340856552124,
|
9587 |
+
"learning_rate": 4.4625913700691026e-08,
|
9588 |
+
"loss": 43.2878,
|
9589 |
+
"step": 1364
|
9590 |
+
},
|
9591 |
+
{
|
9592 |
+
"epoch": 0.13538980361039477,
|
9593 |
+
"grad_norm": 2.0113155841827393,
|
9594 |
+
"learning_rate": 3.8024865577968026e-08,
|
9595 |
+
"loss": 43.4514,
|
9596 |
+
"step": 1365
|
9597 |
+
},
|
9598 |
+
{
|
9599 |
+
"epoch": 0.1354889902797064,
|
9600 |
+
"grad_norm": 1.9771004915237427,
|
9601 |
+
"learning_rate": 3.19517729981289e-08,
|
9602 |
+
"loss": 43.3993,
|
9603 |
+
"step": 1366
|
9604 |
+
},
|
9605 |
+
{
|
9606 |
+
"epoch": 0.13558817694901806,
|
9607 |
+
"grad_norm": 1.7332797050476074,
|
9608 |
+
"learning_rate": 2.6406668036627057e-08,
|
9609 |
+
"loss": 43.168,
|
9610 |
+
"step": 1367
|
9611 |
+
},
|
9612 |
+
{
|
9613 |
+
"epoch": 0.1356873636183297,
|
9614 |
+
"grad_norm": 2.1644415855407715,
|
9615 |
+
"learning_rate": 2.1389579980268804e-08,
|
9616 |
+
"loss": 43.456,
|
9617 |
+
"step": 1368
|
9618 |
+
},
|
9619 |
+
{
|
9620 |
+
"epoch": 0.13578655028764133,
|
9621 |
+
"grad_norm": 2.1394221782684326,
|
9622 |
+
"learning_rate": 1.6900535327135646e-08,
|
9623 |
+
"loss": 43.2602,
|
9624 |
+
"step": 1369
|
9625 |
+
},
|
9626 |
+
{
|
9627 |
+
"epoch": 0.135885736956953,
|
9628 |
+
"grad_norm": 2.1935205459594727,
|
9629 |
+
"learning_rate": 1.293955778639555e-08,
|
9630 |
+
"loss": 43.4416,
|
9631 |
+
"step": 1370
|
9632 |
+
},
|
9633 |
+
{
|
9634 |
+
"epoch": 0.13598492362626463,
|
9635 |
+
"grad_norm": 2.318765878677368,
|
9636 |
+
"learning_rate": 9.506668278203012e-09,
|
9637 |
+
"loss": 43.3051,
|
9638 |
+
"step": 1371
|
9639 |
+
},
|
9640 |
+
{
|
9641 |
+
"epoch": 0.13608411029557627,
|
9642 |
+
"grad_norm": 1.8142633438110352,
|
9643 |
+
"learning_rate": 6.601884933576941e-09,
|
9644 |
+
"loss": 43.3015,
|
9645 |
+
"step": 1372
|
9646 |
+
},
|
9647 |
+
{
|
9648 |
+
"epoch": 0.13618329696488793,
|
9649 |
+
"grad_norm": 2.0885353088378906,
|
9650 |
+
"learning_rate": 4.225223094334041e-09,
|
9651 |
+
"loss": 43.3403,
|
9652 |
+
"step": 1373
|
9653 |
+
},
|
9654 |
+
{
|
9655 |
+
"epoch": 0.13628248363419956,
|
9656 |
+
"grad_norm": 2.292125940322876,
|
9657 |
+
"learning_rate": 2.3766953129555902e-09,
|
9658 |
+
"loss": 43.0529,
|
9659 |
+
"step": 1374
|
9660 |
+
},
|
9661 |
+
{
|
9662 |
+
"epoch": 0.1363816703035112,
|
9663 |
+
"grad_norm": 1.9363865852355957,
|
9664 |
+
"learning_rate": 1.056311352554129e-09,
|
9665 |
+
"loss": 43.3281,
|
9666 |
+
"step": 1375
|
9667 |
+
},
|
9668 |
+
{
|
9669 |
+
"epoch": 0.13648085697282286,
|
9670 |
+
"grad_norm": 2.5582120418548584,
|
9671 |
+
"learning_rate": 2.6407818682905315e-10,
|
9672 |
+
"loss": 43.1195,
|
9673 |
+
"step": 1376
|
9674 |
+
},
|
9675 |
+
{
|
9676 |
+
"epoch": 0.1365800436421345,
|
9677 |
+
"grad_norm": 2.156623125076294,
|
9678 |
+
"learning_rate": 0.0,
|
9679 |
+
"loss": 43.1981,
|
9680 |
+
"step": 1377
|
9681 |
}
|
9682 |
],
|
9683 |
"logging_steps": 1,
|
|
|
9692 |
"should_evaluate": false,
|
9693 |
"should_log": false,
|
9694 |
"should_save": true,
|
9695 |
+
"should_training_stop": true
|
9696 |
},
|
9697 |
"attributes": {}
|
9698 |
}
|
9699 |
},
|
9700 |
+
"total_flos": 13374929240064.0,
|
9701 |
"train_batch_size": 2,
|
9702 |
"trial_name": null,
|
9703 |
"trial_params": null
|