Training in progress, step 1448, 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 83945296
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6aa6546fba9dcd47deba6f7055379b5cdcb66a1299886df001c25d123421facf
|
3 |
size 83945296
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 43123028
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:36bbcb1498222146b5e834880def6206a530c6a7d4d93d6ba8e0c4e719556f56
|
3 |
size 43123028
|
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:b929f4efc1f1f93715fbd6870bf3ae9300c900f729416c7379dfe7931621b103
|
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 sha256:5683862389b99783149902b66144f96581cbba27088e78b8f32dd047db577f30
|
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": 362,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -7641,6 +7641,2548 @@
|
|
7641 |
"eval_samples_per_second": 29.737,
|
7642 |
"eval_steps_per_second": 14.889,
|
7643 |
"step": 1086
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
7644 |
}
|
7645 |
],
|
7646 |
"logging_steps": 1,
|
@@ -7655,12 +10197,12 @@
|
|
7655 |
"should_evaluate": false,
|
7656 |
"should_log": false,
|
7657 |
"should_save": true,
|
7658 |
-
"should_training_stop":
|
7659 |
},
|
7660 |
"attributes": {}
|
7661 |
}
|
7662 |
},
|
7663 |
-
"total_flos": 2.
|
7664 |
"train_batch_size": 2,
|
7665 |
"trial_name": null,
|
7666 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.8388124547429399,
|
5 |
"eval_steps": 362,
|
6 |
+
"global_step": 1448,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
7641 |
"eval_samples_per_second": 29.737,
|
7642 |
"eval_steps_per_second": 14.889,
|
7643 |
"step": 1086
|
7644 |
+
},
|
7645 |
+
{
|
7646 |
+
"epoch": 0.6296886314265026,
|
7647 |
+
"grad_norm": 2.324061155319214,
|
7648 |
+
"learning_rate": 2.9521423136939997e-05,
|
7649 |
+
"loss": 0.0747,
|
7650 |
+
"step": 1087
|
7651 |
+
},
|
7652 |
+
{
|
7653 |
+
"epoch": 0.6302679217958002,
|
7654 |
+
"grad_norm": 2.1686408519744873,
|
7655 |
+
"learning_rate": 2.9366604715861156e-05,
|
7656 |
+
"loss": 0.0941,
|
7657 |
+
"step": 1088
|
7658 |
+
},
|
7659 |
+
{
|
7660 |
+
"epoch": 0.6308472121650978,
|
7661 |
+
"grad_norm": 1.4898357391357422,
|
7662 |
+
"learning_rate": 2.9212123420499536e-05,
|
7663 |
+
"loss": 0.0588,
|
7664 |
+
"step": 1089
|
7665 |
+
},
|
7666 |
+
{
|
7667 |
+
"epoch": 0.6314265025343954,
|
7668 |
+
"grad_norm": 1.3927000761032104,
|
7669 |
+
"learning_rate": 2.9057979988178087e-05,
|
7670 |
+
"loss": 0.0703,
|
7671 |
+
"step": 1090
|
7672 |
+
},
|
7673 |
+
{
|
7674 |
+
"epoch": 0.632005792903693,
|
7675 |
+
"grad_norm": 0.9695125222206116,
|
7676 |
+
"learning_rate": 2.890417515460704e-05,
|
7677 |
+
"loss": 0.0264,
|
7678 |
+
"step": 1091
|
7679 |
+
},
|
7680 |
+
{
|
7681 |
+
"epoch": 0.6325850832729906,
|
7682 |
+
"grad_norm": 1.4306727647781372,
|
7683 |
+
"learning_rate": 2.8750709653880592e-05,
|
7684 |
+
"loss": 0.0791,
|
7685 |
+
"step": 1092
|
7686 |
+
},
|
7687 |
+
{
|
7688 |
+
"epoch": 0.6331643736422882,
|
7689 |
+
"grad_norm": 2.2921833992004395,
|
7690 |
+
"learning_rate": 2.8597584218473316e-05,
|
7691 |
+
"loss": 0.0832,
|
7692 |
+
"step": 1093
|
7693 |
+
},
|
7694 |
+
{
|
7695 |
+
"epoch": 0.6337436640115858,
|
7696 |
+
"grad_norm": 3.3682353496551514,
|
7697 |
+
"learning_rate": 2.8444799579236704e-05,
|
7698 |
+
"loss": 0.136,
|
7699 |
+
"step": 1094
|
7700 |
+
},
|
7701 |
+
{
|
7702 |
+
"epoch": 0.6343229543808834,
|
7703 |
+
"grad_norm": 0.894691526889801,
|
7704 |
+
"learning_rate": 2.8292356465395687e-05,
|
7705 |
+
"loss": 0.0425,
|
7706 |
+
"step": 1095
|
7707 |
+
},
|
7708 |
+
{
|
7709 |
+
"epoch": 0.634902244750181,
|
7710 |
+
"grad_norm": 1.9159088134765625,
|
7711 |
+
"learning_rate": 2.8140255604545106e-05,
|
7712 |
+
"loss": 0.1879,
|
7713 |
+
"step": 1096
|
7714 |
+
},
|
7715 |
+
{
|
7716 |
+
"epoch": 0.6354815351194787,
|
7717 |
+
"grad_norm": 1.968672752380371,
|
7718 |
+
"learning_rate": 2.7988497722646247e-05,
|
7719 |
+
"loss": 0.1133,
|
7720 |
+
"step": 1097
|
7721 |
+
},
|
7722 |
+
{
|
7723 |
+
"epoch": 0.6360608254887763,
|
7724 |
+
"grad_norm": 3.2951698303222656,
|
7725 |
+
"learning_rate": 2.7837083544023422e-05,
|
7726 |
+
"loss": 0.1449,
|
7727 |
+
"step": 1098
|
7728 |
+
},
|
7729 |
+
{
|
7730 |
+
"epoch": 0.6366401158580739,
|
7731 |
+
"grad_norm": 1.5269850492477417,
|
7732 |
+
"learning_rate": 2.7686013791360464e-05,
|
7733 |
+
"loss": 0.0537,
|
7734 |
+
"step": 1099
|
7735 |
+
},
|
7736 |
+
{
|
7737 |
+
"epoch": 0.6372194062273715,
|
7738 |
+
"grad_norm": 0.7901954650878906,
|
7739 |
+
"learning_rate": 2.753528918569732e-05,
|
7740 |
+
"loss": 0.0587,
|
7741 |
+
"step": 1100
|
7742 |
+
},
|
7743 |
+
{
|
7744 |
+
"epoch": 0.6377986965966691,
|
7745 |
+
"grad_norm": 0.9255656003952026,
|
7746 |
+
"learning_rate": 2.738491044642657e-05,
|
7747 |
+
"loss": 0.0396,
|
7748 |
+
"step": 1101
|
7749 |
+
},
|
7750 |
+
{
|
7751 |
+
"epoch": 0.6383779869659667,
|
7752 |
+
"grad_norm": 2.195528984069824,
|
7753 |
+
"learning_rate": 2.723487829129001e-05,
|
7754 |
+
"loss": 0.1936,
|
7755 |
+
"step": 1102
|
7756 |
+
},
|
7757 |
+
{
|
7758 |
+
"epoch": 0.6389572773352643,
|
7759 |
+
"grad_norm": 0.6521852016448975,
|
7760 |
+
"learning_rate": 2.708519343637521e-05,
|
7761 |
+
"loss": 0.0407,
|
7762 |
+
"step": 1103
|
7763 |
+
},
|
7764 |
+
{
|
7765 |
+
"epoch": 0.6395365677045619,
|
7766 |
+
"grad_norm": 0.5633903741836548,
|
7767 |
+
"learning_rate": 2.6935856596112197e-05,
|
7768 |
+
"loss": 0.0325,
|
7769 |
+
"step": 1104
|
7770 |
+
},
|
7771 |
+
{
|
7772 |
+
"epoch": 0.6401158580738595,
|
7773 |
+
"grad_norm": 0.8536943197250366,
|
7774 |
+
"learning_rate": 2.6786868483269856e-05,
|
7775 |
+
"loss": 0.0421,
|
7776 |
+
"step": 1105
|
7777 |
+
},
|
7778 |
+
{
|
7779 |
+
"epoch": 0.6406951484431571,
|
7780 |
+
"grad_norm": 1.3334097862243652,
|
7781 |
+
"learning_rate": 2.6638229808952685e-05,
|
7782 |
+
"loss": 0.0505,
|
7783 |
+
"step": 1106
|
7784 |
+
},
|
7785 |
+
{
|
7786 |
+
"epoch": 0.6412744388124547,
|
7787 |
+
"grad_norm": 1.8636704683303833,
|
7788 |
+
"learning_rate": 2.648994128259733e-05,
|
7789 |
+
"loss": 0.1267,
|
7790 |
+
"step": 1107
|
7791 |
+
},
|
7792 |
+
{
|
7793 |
+
"epoch": 0.6418537291817523,
|
7794 |
+
"grad_norm": 1.0794011354446411,
|
7795 |
+
"learning_rate": 2.6342003611969223e-05,
|
7796 |
+
"loss": 0.0526,
|
7797 |
+
"step": 1108
|
7798 |
+
},
|
7799 |
+
{
|
7800 |
+
"epoch": 0.6424330195510499,
|
7801 |
+
"grad_norm": 2.190622329711914,
|
7802 |
+
"learning_rate": 2.619441750315923e-05,
|
7803 |
+
"loss": 0.0816,
|
7804 |
+
"step": 1109
|
7805 |
+
},
|
7806 |
+
{
|
7807 |
+
"epoch": 0.6430123099203475,
|
7808 |
+
"grad_norm": 1.3400945663452148,
|
7809 |
+
"learning_rate": 2.60471836605802e-05,
|
7810 |
+
"loss": 0.0385,
|
7811 |
+
"step": 1110
|
7812 |
+
},
|
7813 |
+
{
|
7814 |
+
"epoch": 0.6435916002896452,
|
7815 |
+
"grad_norm": 1.6180418729782104,
|
7816 |
+
"learning_rate": 2.5900302786963636e-05,
|
7817 |
+
"loss": 0.0597,
|
7818 |
+
"step": 1111
|
7819 |
+
},
|
7820 |
+
{
|
7821 |
+
"epoch": 0.6441708906589428,
|
7822 |
+
"grad_norm": 1.4665288925170898,
|
7823 |
+
"learning_rate": 2.575377558335649e-05,
|
7824 |
+
"loss": 0.0355,
|
7825 |
+
"step": 1112
|
7826 |
+
},
|
7827 |
+
{
|
7828 |
+
"epoch": 0.6447501810282404,
|
7829 |
+
"grad_norm": 1.2391499280929565,
|
7830 |
+
"learning_rate": 2.560760274911752e-05,
|
7831 |
+
"loss": 0.0786,
|
7832 |
+
"step": 1113
|
7833 |
+
},
|
7834 |
+
{
|
7835 |
+
"epoch": 0.645329471397538,
|
7836 |
+
"grad_norm": 0.8473105430603027,
|
7837 |
+
"learning_rate": 2.5461784981914193e-05,
|
7838 |
+
"loss": 0.0298,
|
7839 |
+
"step": 1114
|
7840 |
+
},
|
7841 |
+
{
|
7842 |
+
"epoch": 0.6459087617668356,
|
7843 |
+
"grad_norm": 0.7986234426498413,
|
7844 |
+
"learning_rate": 2.531632297771931e-05,
|
7845 |
+
"loss": 0.0407,
|
7846 |
+
"step": 1115
|
7847 |
+
},
|
7848 |
+
{
|
7849 |
+
"epoch": 0.6464880521361333,
|
7850 |
+
"grad_norm": 2.9595439434051514,
|
7851 |
+
"learning_rate": 2.5171217430807605e-05,
|
7852 |
+
"loss": 0.1765,
|
7853 |
+
"step": 1116
|
7854 |
+
},
|
7855 |
+
{
|
7856 |
+
"epoch": 0.6470673425054309,
|
7857 |
+
"grad_norm": 0.8639671206474304,
|
7858 |
+
"learning_rate": 2.5026469033752464e-05,
|
7859 |
+
"loss": 0.0331,
|
7860 |
+
"step": 1117
|
7861 |
+
},
|
7862 |
+
{
|
7863 |
+
"epoch": 0.6476466328747285,
|
7864 |
+
"grad_norm": 2.8039450645446777,
|
7865 |
+
"learning_rate": 2.4882078477422687e-05,
|
7866 |
+
"loss": 0.4577,
|
7867 |
+
"step": 1118
|
7868 |
+
},
|
7869 |
+
{
|
7870 |
+
"epoch": 0.6482259232440261,
|
7871 |
+
"grad_norm": 1.1273934841156006,
|
7872 |
+
"learning_rate": 2.4738046450979103e-05,
|
7873 |
+
"loss": 0.0531,
|
7874 |
+
"step": 1119
|
7875 |
+
},
|
7876 |
+
{
|
7877 |
+
"epoch": 0.6488052136133237,
|
7878 |
+
"grad_norm": 2.667005777359009,
|
7879 |
+
"learning_rate": 2.4594373641871314e-05,
|
7880 |
+
"loss": 0.091,
|
7881 |
+
"step": 1120
|
7882 |
+
},
|
7883 |
+
{
|
7884 |
+
"epoch": 0.6493845039826213,
|
7885 |
+
"grad_norm": 5.183987140655518,
|
7886 |
+
"learning_rate": 2.4451060735834443e-05,
|
7887 |
+
"loss": 0.2782,
|
7888 |
+
"step": 1121
|
7889 |
+
},
|
7890 |
+
{
|
7891 |
+
"epoch": 0.6499637943519189,
|
7892 |
+
"grad_norm": 1.5411016941070557,
|
7893 |
+
"learning_rate": 2.4308108416885778e-05,
|
7894 |
+
"loss": 0.0559,
|
7895 |
+
"step": 1122
|
7896 |
+
},
|
7897 |
+
{
|
7898 |
+
"epoch": 0.6505430847212165,
|
7899 |
+
"grad_norm": 1.374833106994629,
|
7900 |
+
"learning_rate": 2.4165517367321656e-05,
|
7901 |
+
"loss": 0.0556,
|
7902 |
+
"step": 1123
|
7903 |
+
},
|
7904 |
+
{
|
7905 |
+
"epoch": 0.6511223750905141,
|
7906 |
+
"grad_norm": 1.1232033967971802,
|
7907 |
+
"learning_rate": 2.4023288267714038e-05,
|
7908 |
+
"loss": 0.0563,
|
7909 |
+
"step": 1124
|
7910 |
+
},
|
7911 |
+
{
|
7912 |
+
"epoch": 0.6517016654598118,
|
7913 |
+
"grad_norm": 2.193145751953125,
|
7914 |
+
"learning_rate": 2.3881421796907366e-05,
|
7915 |
+
"loss": 0.1394,
|
7916 |
+
"step": 1125
|
7917 |
+
},
|
7918 |
+
{
|
7919 |
+
"epoch": 0.6522809558291094,
|
7920 |
+
"grad_norm": 0.6777664422988892,
|
7921 |
+
"learning_rate": 2.373991863201527e-05,
|
7922 |
+
"loss": 0.0179,
|
7923 |
+
"step": 1126
|
7924 |
+
},
|
7925 |
+
{
|
7926 |
+
"epoch": 0.652860246198407,
|
7927 |
+
"grad_norm": 1.298601746559143,
|
7928 |
+
"learning_rate": 2.3598779448417373e-05,
|
7929 |
+
"loss": 0.0605,
|
7930 |
+
"step": 1127
|
7931 |
+
},
|
7932 |
+
{
|
7933 |
+
"epoch": 0.6534395365677046,
|
7934 |
+
"grad_norm": 1.3922193050384521,
|
7935 |
+
"learning_rate": 2.345800491975606e-05,
|
7936 |
+
"loss": 0.1004,
|
7937 |
+
"step": 1128
|
7938 |
+
},
|
7939 |
+
{
|
7940 |
+
"epoch": 0.6540188269370022,
|
7941 |
+
"grad_norm": 2.807434320449829,
|
7942 |
+
"learning_rate": 2.331759571793326e-05,
|
7943 |
+
"loss": 0.1556,
|
7944 |
+
"step": 1129
|
7945 |
+
},
|
7946 |
+
{
|
7947 |
+
"epoch": 0.6545981173062998,
|
7948 |
+
"grad_norm": 2.1901803016662598,
|
7949 |
+
"learning_rate": 2.317755251310719e-05,
|
7950 |
+
"loss": 0.1278,
|
7951 |
+
"step": 1130
|
7952 |
+
},
|
7953 |
+
{
|
7954 |
+
"epoch": 0.6551774076755974,
|
7955 |
+
"grad_norm": 1.5145044326782227,
|
7956 |
+
"learning_rate": 2.3037875973689293e-05,
|
7957 |
+
"loss": 0.0493,
|
7958 |
+
"step": 1131
|
7959 |
+
},
|
7960 |
+
{
|
7961 |
+
"epoch": 0.655756698044895,
|
7962 |
+
"grad_norm": 2.6977388858795166,
|
7963 |
+
"learning_rate": 2.289856676634089e-05,
|
7964 |
+
"loss": 0.1319,
|
7965 |
+
"step": 1132
|
7966 |
+
},
|
7967 |
+
{
|
7968 |
+
"epoch": 0.6563359884141926,
|
7969 |
+
"grad_norm": 1.313981533050537,
|
7970 |
+
"learning_rate": 2.2759625555970055e-05,
|
7971 |
+
"loss": 0.0809,
|
7972 |
+
"step": 1133
|
7973 |
+
},
|
7974 |
+
{
|
7975 |
+
"epoch": 0.6569152787834902,
|
7976 |
+
"grad_norm": 0.6499395370483398,
|
7977 |
+
"learning_rate": 2.2621053005728498e-05,
|
7978 |
+
"loss": 0.0255,
|
7979 |
+
"step": 1134
|
7980 |
+
},
|
7981 |
+
{
|
7982 |
+
"epoch": 0.6574945691527878,
|
7983 |
+
"grad_norm": 0.598595917224884,
|
7984 |
+
"learning_rate": 2.2482849777008308e-05,
|
7985 |
+
"loss": 0.026,
|
7986 |
+
"step": 1135
|
7987 |
+
},
|
7988 |
+
{
|
7989 |
+
"epoch": 0.6580738595220854,
|
7990 |
+
"grad_norm": 0.5190146565437317,
|
7991 |
+
"learning_rate": 2.234501652943888e-05,
|
7992 |
+
"loss": 0.019,
|
7993 |
+
"step": 1136
|
7994 |
+
},
|
7995 |
+
{
|
7996 |
+
"epoch": 0.658653149891383,
|
7997 |
+
"grad_norm": 1.525578498840332,
|
7998 |
+
"learning_rate": 2.2207553920883706e-05,
|
7999 |
+
"loss": 0.0738,
|
8000 |
+
"step": 1137
|
8001 |
+
},
|
8002 |
+
{
|
8003 |
+
"epoch": 0.6592324402606806,
|
8004 |
+
"grad_norm": 2.3997116088867188,
|
8005 |
+
"learning_rate": 2.2070462607437258e-05,
|
8006 |
+
"loss": 0.0979,
|
8007 |
+
"step": 1138
|
8008 |
+
},
|
8009 |
+
{
|
8010 |
+
"epoch": 0.6598117306299782,
|
8011 |
+
"grad_norm": 1.328265905380249,
|
8012 |
+
"learning_rate": 2.1933743243421833e-05,
|
8013 |
+
"loss": 0.0646,
|
8014 |
+
"step": 1139
|
8015 |
+
},
|
8016 |
+
{
|
8017 |
+
"epoch": 0.6603910209992759,
|
8018 |
+
"grad_norm": 1.3342244625091553,
|
8019 |
+
"learning_rate": 2.1797396481384546e-05,
|
8020 |
+
"loss": 0.0574,
|
8021 |
+
"step": 1140
|
8022 |
+
},
|
8023 |
+
{
|
8024 |
+
"epoch": 0.6609703113685735,
|
8025 |
+
"grad_norm": 0.6104873418807983,
|
8026 |
+
"learning_rate": 2.1661422972093993e-05,
|
8027 |
+
"loss": 0.021,
|
8028 |
+
"step": 1141
|
8029 |
+
},
|
8030 |
+
{
|
8031 |
+
"epoch": 0.6615496017378711,
|
8032 |
+
"grad_norm": 1.49978506565094,
|
8033 |
+
"learning_rate": 2.152582336453739e-05,
|
8034 |
+
"loss": 0.0781,
|
8035 |
+
"step": 1142
|
8036 |
+
},
|
8037 |
+
{
|
8038 |
+
"epoch": 0.6621288921071687,
|
8039 |
+
"grad_norm": 1.9663605690002441,
|
8040 |
+
"learning_rate": 2.1390598305917275e-05,
|
8041 |
+
"loss": 0.1421,
|
8042 |
+
"step": 1143
|
8043 |
+
},
|
8044 |
+
{
|
8045 |
+
"epoch": 0.6627081824764663,
|
8046 |
+
"grad_norm": 0.8139146566390991,
|
8047 |
+
"learning_rate": 2.125574844164856e-05,
|
8048 |
+
"loss": 0.0387,
|
8049 |
+
"step": 1144
|
8050 |
+
},
|
8051 |
+
{
|
8052 |
+
"epoch": 0.663287472845764,
|
8053 |
+
"grad_norm": 1.3449358940124512,
|
8054 |
+
"learning_rate": 2.112127441535534e-05,
|
8055 |
+
"loss": 0.0643,
|
8056 |
+
"step": 1145
|
8057 |
+
},
|
8058 |
+
{
|
8059 |
+
"epoch": 0.6638667632150616,
|
8060 |
+
"grad_norm": 2.343512535095215,
|
8061 |
+
"learning_rate": 2.098717686886793e-05,
|
8062 |
+
"loss": 0.1408,
|
8063 |
+
"step": 1146
|
8064 |
+
},
|
8065 |
+
{
|
8066 |
+
"epoch": 0.6644460535843592,
|
8067 |
+
"grad_norm": 2.189060688018799,
|
8068 |
+
"learning_rate": 2.08534564422197e-05,
|
8069 |
+
"loss": 0.2214,
|
8070 |
+
"step": 1147
|
8071 |
+
},
|
8072 |
+
{
|
8073 |
+
"epoch": 0.6650253439536568,
|
8074 |
+
"grad_norm": 1.148862600326538,
|
8075 |
+
"learning_rate": 2.0720113773644102e-05,
|
8076 |
+
"loss": 0.055,
|
8077 |
+
"step": 1148
|
8078 |
+
},
|
8079 |
+
{
|
8080 |
+
"epoch": 0.6656046343229544,
|
8081 |
+
"grad_norm": 1.1233261823654175,
|
8082 |
+
"learning_rate": 2.0587149499571534e-05,
|
8083 |
+
"loss": 0.0397,
|
8084 |
+
"step": 1149
|
8085 |
+
},
|
8086 |
+
{
|
8087 |
+
"epoch": 0.666183924692252,
|
8088 |
+
"grad_norm": 1.1604281663894653,
|
8089 |
+
"learning_rate": 2.0454564254626473e-05,
|
8090 |
+
"loss": 0.0478,
|
8091 |
+
"step": 1150
|
8092 |
+
},
|
8093 |
+
{
|
8094 |
+
"epoch": 0.6667632150615496,
|
8095 |
+
"grad_norm": 5.3524489402771,
|
8096 |
+
"learning_rate": 2.032235867162422e-05,
|
8097 |
+
"loss": 0.1517,
|
8098 |
+
"step": 1151
|
8099 |
+
},
|
8100 |
+
{
|
8101 |
+
"epoch": 0.6673425054308472,
|
8102 |
+
"grad_norm": 1.4170546531677246,
|
8103 |
+
"learning_rate": 2.019053338156801e-05,
|
8104 |
+
"loss": 0.0728,
|
8105 |
+
"step": 1152
|
8106 |
+
},
|
8107 |
+
{
|
8108 |
+
"epoch": 0.6679217958001448,
|
8109 |
+
"grad_norm": 1.3472294807434082,
|
8110 |
+
"learning_rate": 2.0059089013646025e-05,
|
8111 |
+
"loss": 0.0378,
|
8112 |
+
"step": 1153
|
8113 |
+
},
|
8114 |
+
{
|
8115 |
+
"epoch": 0.6685010861694425,
|
8116 |
+
"grad_norm": 2.448164224624634,
|
8117 |
+
"learning_rate": 1.9928026195228288e-05,
|
8118 |
+
"loss": 0.1915,
|
8119 |
+
"step": 1154
|
8120 |
+
},
|
8121 |
+
{
|
8122 |
+
"epoch": 0.6690803765387401,
|
8123 |
+
"grad_norm": 1.3396458625793457,
|
8124 |
+
"learning_rate": 1.9797345551863765e-05,
|
8125 |
+
"loss": 0.0409,
|
8126 |
+
"step": 1155
|
8127 |
+
},
|
8128 |
+
{
|
8129 |
+
"epoch": 0.6696596669080377,
|
8130 |
+
"grad_norm": 0.6571703553199768,
|
8131 |
+
"learning_rate": 1.966704770727734e-05,
|
8132 |
+
"loss": 0.0265,
|
8133 |
+
"step": 1156
|
8134 |
+
},
|
8135 |
+
{
|
8136 |
+
"epoch": 0.6702389572773353,
|
8137 |
+
"grad_norm": 2.5409204959869385,
|
8138 |
+
"learning_rate": 1.9537133283366826e-05,
|
8139 |
+
"loss": 0.0757,
|
8140 |
+
"step": 1157
|
8141 |
+
},
|
8142 |
+
{
|
8143 |
+
"epoch": 0.6708182476466329,
|
8144 |
+
"grad_norm": 0.7289441227912903,
|
8145 |
+
"learning_rate": 1.940760290019997e-05,
|
8146 |
+
"loss": 0.0229,
|
8147 |
+
"step": 1158
|
8148 |
+
},
|
8149 |
+
{
|
8150 |
+
"epoch": 0.6713975380159305,
|
8151 |
+
"grad_norm": 0.6556041240692139,
|
8152 |
+
"learning_rate": 1.9278457176011645e-05,
|
8153 |
+
"loss": 0.0391,
|
8154 |
+
"step": 1159
|
8155 |
+
},
|
8156 |
+
{
|
8157 |
+
"epoch": 0.6719768283852281,
|
8158 |
+
"grad_norm": 1.4001652002334595,
|
8159 |
+
"learning_rate": 1.9149696727200695e-05,
|
8160 |
+
"loss": 0.0467,
|
8161 |
+
"step": 1160
|
8162 |
+
},
|
8163 |
+
{
|
8164 |
+
"epoch": 0.6725561187545257,
|
8165 |
+
"grad_norm": 0.9999477863311768,
|
8166 |
+
"learning_rate": 1.9021322168327104e-05,
|
8167 |
+
"loss": 0.0421,
|
8168 |
+
"step": 1161
|
8169 |
+
},
|
8170 |
+
{
|
8171 |
+
"epoch": 0.6731354091238233,
|
8172 |
+
"grad_norm": 2.330754518508911,
|
8173 |
+
"learning_rate": 1.889333411210905e-05,
|
8174 |
+
"loss": 0.0763,
|
8175 |
+
"step": 1162
|
8176 |
+
},
|
8177 |
+
{
|
8178 |
+
"epoch": 0.6737146994931209,
|
8179 |
+
"grad_norm": 2.5705575942993164,
|
8180 |
+
"learning_rate": 1.876573316941994e-05,
|
8181 |
+
"loss": 0.0581,
|
8182 |
+
"step": 1163
|
8183 |
+
},
|
8184 |
+
{
|
8185 |
+
"epoch": 0.6742939898624185,
|
8186 |
+
"grad_norm": 1.7020872831344604,
|
8187 |
+
"learning_rate": 1.8638519949285636e-05,
|
8188 |
+
"loss": 0.0565,
|
8189 |
+
"step": 1164
|
8190 |
+
},
|
8191 |
+
{
|
8192 |
+
"epoch": 0.6748732802317161,
|
8193 |
+
"grad_norm": 3.051485776901245,
|
8194 |
+
"learning_rate": 1.8511695058881316e-05,
|
8195 |
+
"loss": 0.1276,
|
8196 |
+
"step": 1165
|
8197 |
+
},
|
8198 |
+
{
|
8199 |
+
"epoch": 0.6754525706010137,
|
8200 |
+
"grad_norm": 3.2355425357818604,
|
8201 |
+
"learning_rate": 1.8385259103528784e-05,
|
8202 |
+
"loss": 0.1606,
|
8203 |
+
"step": 1166
|
8204 |
+
},
|
8205 |
+
{
|
8206 |
+
"epoch": 0.6760318609703113,
|
8207 |
+
"grad_norm": 2.5328848361968994,
|
8208 |
+
"learning_rate": 1.8259212686693438e-05,
|
8209 |
+
"loss": 0.0442,
|
8210 |
+
"step": 1167
|
8211 |
+
},
|
8212 |
+
{
|
8213 |
+
"epoch": 0.676611151339609,
|
8214 |
+
"grad_norm": 2.2313077449798584,
|
8215 |
+
"learning_rate": 1.813355640998149e-05,
|
8216 |
+
"loss": 0.1209,
|
8217 |
+
"step": 1168
|
8218 |
+
},
|
8219 |
+
{
|
8220 |
+
"epoch": 0.6771904417089066,
|
8221 |
+
"grad_norm": 0.7647517323493958,
|
8222 |
+
"learning_rate": 1.8008290873137057e-05,
|
8223 |
+
"loss": 0.0367,
|
8224 |
+
"step": 1169
|
8225 |
+
},
|
8226 |
+
{
|
8227 |
+
"epoch": 0.6777697320782042,
|
8228 |
+
"grad_norm": 2.191480875015259,
|
8229 |
+
"learning_rate": 1.7883416674039278e-05,
|
8230 |
+
"loss": 0.0886,
|
8231 |
+
"step": 1170
|
8232 |
+
},
|
8233 |
+
{
|
8234 |
+
"epoch": 0.6783490224475018,
|
8235 |
+
"grad_norm": 2.841825485229492,
|
8236 |
+
"learning_rate": 1.775893440869949e-05,
|
8237 |
+
"loss": 0.3078,
|
8238 |
+
"step": 1171
|
8239 |
+
},
|
8240 |
+
{
|
8241 |
+
"epoch": 0.6789283128167994,
|
8242 |
+
"grad_norm": 1.2976844310760498,
|
8243 |
+
"learning_rate": 1.7634844671258354e-05,
|
8244 |
+
"loss": 0.0464,
|
8245 |
+
"step": 1172
|
8246 |
+
},
|
8247 |
+
{
|
8248 |
+
"epoch": 0.679507603186097,
|
8249 |
+
"grad_norm": 1.7949457168579102,
|
8250 |
+
"learning_rate": 1.751114805398305e-05,
|
8251 |
+
"loss": 0.0586,
|
8252 |
+
"step": 1173
|
8253 |
+
},
|
8254 |
+
{
|
8255 |
+
"epoch": 0.6800868935553946,
|
8256 |
+
"grad_norm": 1.2988979816436768,
|
8257 |
+
"learning_rate": 1.7387845147264446e-05,
|
8258 |
+
"loss": 0.0535,
|
8259 |
+
"step": 1174
|
8260 |
+
},
|
8261 |
+
{
|
8262 |
+
"epoch": 0.6806661839246922,
|
8263 |
+
"grad_norm": 2.0930166244506836,
|
8264 |
+
"learning_rate": 1.726493653961425e-05,
|
8265 |
+
"loss": 0.0488,
|
8266 |
+
"step": 1175
|
8267 |
+
},
|
8268 |
+
{
|
8269 |
+
"epoch": 0.6812454742939898,
|
8270 |
+
"grad_norm": 0.5041794776916504,
|
8271 |
+
"learning_rate": 1.714242281766224e-05,
|
8272 |
+
"loss": 0.0272,
|
8273 |
+
"step": 1176
|
8274 |
+
},
|
8275 |
+
{
|
8276 |
+
"epoch": 0.6818247646632875,
|
8277 |
+
"grad_norm": 1.4483470916748047,
|
8278 |
+
"learning_rate": 1.702030456615341e-05,
|
8279 |
+
"loss": 0.0622,
|
8280 |
+
"step": 1177
|
8281 |
+
},
|
8282 |
+
{
|
8283 |
+
"epoch": 0.682404055032585,
|
8284 |
+
"grad_norm": 0.9361678957939148,
|
8285 |
+
"learning_rate": 1.6898582367945293e-05,
|
8286 |
+
"loss": 0.027,
|
8287 |
+
"step": 1178
|
8288 |
+
},
|
8289 |
+
{
|
8290 |
+
"epoch": 0.6829833454018827,
|
8291 |
+
"grad_norm": 1.59813392162323,
|
8292 |
+
"learning_rate": 1.6777256804005024e-05,
|
8293 |
+
"loss": 0.0738,
|
8294 |
+
"step": 1179
|
8295 |
+
},
|
8296 |
+
{
|
8297 |
+
"epoch": 0.6835626357711803,
|
8298 |
+
"grad_norm": 1.2701432704925537,
|
8299 |
+
"learning_rate": 1.665632845340669e-05,
|
8300 |
+
"loss": 0.0645,
|
8301 |
+
"step": 1180
|
8302 |
+
},
|
8303 |
+
{
|
8304 |
+
"epoch": 0.6841419261404779,
|
8305 |
+
"grad_norm": 1.3925986289978027,
|
8306 |
+
"learning_rate": 1.65357978933285e-05,
|
8307 |
+
"loss": 0.0608,
|
8308 |
+
"step": 1181
|
8309 |
+
},
|
8310 |
+
{
|
8311 |
+
"epoch": 0.6847212165097756,
|
8312 |
+
"grad_norm": 1.780645728111267,
|
8313 |
+
"learning_rate": 1.6415665699050044e-05,
|
8314 |
+
"loss": 0.0785,
|
8315 |
+
"step": 1182
|
8316 |
+
},
|
8317 |
+
{
|
8318 |
+
"epoch": 0.6853005068790732,
|
8319 |
+
"grad_norm": 1.0147874355316162,
|
8320 |
+
"learning_rate": 1.629593244394958e-05,
|
8321 |
+
"loss": 0.038,
|
8322 |
+
"step": 1183
|
8323 |
+
},
|
8324 |
+
{
|
8325 |
+
"epoch": 0.6858797972483708,
|
8326 |
+
"grad_norm": 1.7671834230422974,
|
8327 |
+
"learning_rate": 1.6176598699501256e-05,
|
8328 |
+
"loss": 0.1112,
|
8329 |
+
"step": 1184
|
8330 |
+
},
|
8331 |
+
{
|
8332 |
+
"epoch": 0.6864590876176684,
|
8333 |
+
"grad_norm": 0.9414635896682739,
|
8334 |
+
"learning_rate": 1.605766503527236e-05,
|
8335 |
+
"loss": 0.0509,
|
8336 |
+
"step": 1185
|
8337 |
+
},
|
8338 |
+
{
|
8339 |
+
"epoch": 0.687038377986966,
|
8340 |
+
"grad_norm": 0.9722913503646851,
|
8341 |
+
"learning_rate": 1.5939132018920743e-05,
|
8342 |
+
"loss": 0.0451,
|
8343 |
+
"step": 1186
|
8344 |
+
},
|
8345 |
+
{
|
8346 |
+
"epoch": 0.6876176683562636,
|
8347 |
+
"grad_norm": 0.6239556670188904,
|
8348 |
+
"learning_rate": 1.582100021619193e-05,
|
8349 |
+
"loss": 0.0273,
|
8350 |
+
"step": 1187
|
8351 |
+
},
|
8352 |
+
{
|
8353 |
+
"epoch": 0.6881969587255612,
|
8354 |
+
"grad_norm": 0.5656236410140991,
|
8355 |
+
"learning_rate": 1.5703270190916453e-05,
|
8356 |
+
"loss": 0.0187,
|
8357 |
+
"step": 1188
|
8358 |
+
},
|
8359 |
+
{
|
8360 |
+
"epoch": 0.6887762490948588,
|
8361 |
+
"grad_norm": 1.9651774168014526,
|
8362 |
+
"learning_rate": 1.5585942505007312e-05,
|
8363 |
+
"loss": 0.22,
|
8364 |
+
"step": 1189
|
8365 |
+
},
|
8366 |
+
{
|
8367 |
+
"epoch": 0.6893555394641564,
|
8368 |
+
"grad_norm": 1.4058241844177246,
|
8369 |
+
"learning_rate": 1.5469017718457124e-05,
|
8370 |
+
"loss": 0.0863,
|
8371 |
+
"step": 1190
|
8372 |
+
},
|
8373 |
+
{
|
8374 |
+
"epoch": 0.689934829833454,
|
8375 |
+
"grad_norm": 1.7083112001419067,
|
8376 |
+
"learning_rate": 1.5352496389335514e-05,
|
8377 |
+
"loss": 0.0547,
|
8378 |
+
"step": 1191
|
8379 |
+
},
|
8380 |
+
{
|
8381 |
+
"epoch": 0.6905141202027516,
|
8382 |
+
"grad_norm": 1.5525312423706055,
|
8383 |
+
"learning_rate": 1.5236379073786434e-05,
|
8384 |
+
"loss": 0.0517,
|
8385 |
+
"step": 1192
|
8386 |
+
},
|
8387 |
+
{
|
8388 |
+
"epoch": 0.6910934105720492,
|
8389 |
+
"grad_norm": 2.710247039794922,
|
8390 |
+
"learning_rate": 1.5120666326025546e-05,
|
8391 |
+
"loss": 0.1076,
|
8392 |
+
"step": 1193
|
8393 |
+
},
|
8394 |
+
{
|
8395 |
+
"epoch": 0.6916727009413468,
|
8396 |
+
"grad_norm": 2.1334500312805176,
|
8397 |
+
"learning_rate": 1.5005358698337535e-05,
|
8398 |
+
"loss": 0.1462,
|
8399 |
+
"step": 1194
|
8400 |
+
},
|
8401 |
+
{
|
8402 |
+
"epoch": 0.6922519913106444,
|
8403 |
+
"grad_norm": 0.808712363243103,
|
8404 |
+
"learning_rate": 1.4890456741073488e-05,
|
8405 |
+
"loss": 0.0273,
|
8406 |
+
"step": 1195
|
8407 |
+
},
|
8408 |
+
{
|
8409 |
+
"epoch": 0.6928312816799421,
|
8410 |
+
"grad_norm": 1.0377944707870483,
|
8411 |
+
"learning_rate": 1.477596100264832e-05,
|
8412 |
+
"loss": 0.0517,
|
8413 |
+
"step": 1196
|
8414 |
+
},
|
8415 |
+
{
|
8416 |
+
"epoch": 0.6934105720492397,
|
8417 |
+
"grad_norm": 1.4817993640899658,
|
8418 |
+
"learning_rate": 1.4661872029538049e-05,
|
8419 |
+
"loss": 0.044,
|
8420 |
+
"step": 1197
|
8421 |
+
},
|
8422 |
+
{
|
8423 |
+
"epoch": 0.6939898624185373,
|
8424 |
+
"grad_norm": 0.8048624396324158,
|
8425 |
+
"learning_rate": 1.454819036627727e-05,
|
8426 |
+
"loss": 0.0292,
|
8427 |
+
"step": 1198
|
8428 |
+
},
|
8429 |
+
{
|
8430 |
+
"epoch": 0.6945691527878349,
|
8431 |
+
"grad_norm": 1.120234727859497,
|
8432 |
+
"learning_rate": 1.4434916555456512e-05,
|
8433 |
+
"loss": 0.0266,
|
8434 |
+
"step": 1199
|
8435 |
+
},
|
8436 |
+
{
|
8437 |
+
"epoch": 0.6951484431571325,
|
8438 |
+
"grad_norm": 2.329502582550049,
|
8439 |
+
"learning_rate": 1.4322051137719684e-05,
|
8440 |
+
"loss": 0.13,
|
8441 |
+
"step": 1200
|
8442 |
+
},
|
8443 |
+
{
|
8444 |
+
"epoch": 0.6957277335264301,
|
8445 |
+
"grad_norm": 0.7854491472244263,
|
8446 |
+
"learning_rate": 1.4209594651761482e-05,
|
8447 |
+
"loss": 0.0357,
|
8448 |
+
"step": 1201
|
8449 |
+
},
|
8450 |
+
{
|
8451 |
+
"epoch": 0.6963070238957277,
|
8452 |
+
"grad_norm": 0.8537886142730713,
|
8453 |
+
"learning_rate": 1.4097547634324814e-05,
|
8454 |
+
"loss": 0.0326,
|
8455 |
+
"step": 1202
|
8456 |
+
},
|
8457 |
+
{
|
8458 |
+
"epoch": 0.6968863142650253,
|
8459 |
+
"grad_norm": 1.1249948740005493,
|
8460 |
+
"learning_rate": 1.3985910620198207e-05,
|
8461 |
+
"loss": 0.0508,
|
8462 |
+
"step": 1203
|
8463 |
+
},
|
8464 |
+
{
|
8465 |
+
"epoch": 0.6974656046343229,
|
8466 |
+
"grad_norm": 0.44697949290275574,
|
8467 |
+
"learning_rate": 1.3874684142213312e-05,
|
8468 |
+
"loss": 0.0133,
|
8469 |
+
"step": 1204
|
8470 |
+
},
|
8471 |
+
{
|
8472 |
+
"epoch": 0.6980448950036205,
|
8473 |
+
"grad_norm": 2.169045925140381,
|
8474 |
+
"learning_rate": 1.3763868731242357e-05,
|
8475 |
+
"loss": 0.0668,
|
8476 |
+
"step": 1205
|
8477 |
+
},
|
8478 |
+
{
|
8479 |
+
"epoch": 0.6986241853729181,
|
8480 |
+
"grad_norm": 1.6234321594238281,
|
8481 |
+
"learning_rate": 1.3653464916195558e-05,
|
8482 |
+
"loss": 0.0475,
|
8483 |
+
"step": 1206
|
8484 |
+
},
|
8485 |
+
{
|
8486 |
+
"epoch": 0.6992034757422158,
|
8487 |
+
"grad_norm": 1.4020620584487915,
|
8488 |
+
"learning_rate": 1.3543473224018621e-05,
|
8489 |
+
"loss": 0.0514,
|
8490 |
+
"step": 1207
|
8491 |
+
},
|
8492 |
+
{
|
8493 |
+
"epoch": 0.6997827661115134,
|
8494 |
+
"grad_norm": 0.6575192809104919,
|
8495 |
+
"learning_rate": 1.3433894179690265e-05,
|
8496 |
+
"loss": 0.0323,
|
8497 |
+
"step": 1208
|
8498 |
+
},
|
8499 |
+
{
|
8500 |
+
"epoch": 0.700362056480811,
|
8501 |
+
"grad_norm": 0.9925378561019897,
|
8502 |
+
"learning_rate": 1.3324728306219659e-05,
|
8503 |
+
"loss": 0.041,
|
8504 |
+
"step": 1209
|
8505 |
+
},
|
8506 |
+
{
|
8507 |
+
"epoch": 0.7009413468501087,
|
8508 |
+
"grad_norm": 1.427495002746582,
|
8509 |
+
"learning_rate": 1.3215976124643947e-05,
|
8510 |
+
"loss": 0.0633,
|
8511 |
+
"step": 1210
|
8512 |
+
},
|
8513 |
+
{
|
8514 |
+
"epoch": 0.7015206372194063,
|
8515 |
+
"grad_norm": 1.141708493232727,
|
8516 |
+
"learning_rate": 1.3107638154025793e-05,
|
8517 |
+
"loss": 0.033,
|
8518 |
+
"step": 1211
|
8519 |
+
},
|
8520 |
+
{
|
8521 |
+
"epoch": 0.7020999275887039,
|
8522 |
+
"grad_norm": 1.6439590454101562,
|
8523 |
+
"learning_rate": 1.2999714911450833e-05,
|
8524 |
+
"loss": 0.0741,
|
8525 |
+
"step": 1212
|
8526 |
+
},
|
8527 |
+
{
|
8528 |
+
"epoch": 0.7026792179580015,
|
8529 |
+
"grad_norm": 0.6117269992828369,
|
8530 |
+
"learning_rate": 1.289220691202533e-05,
|
8531 |
+
"loss": 0.0223,
|
8532 |
+
"step": 1213
|
8533 |
+
},
|
8534 |
+
{
|
8535 |
+
"epoch": 0.7032585083272991,
|
8536 |
+
"grad_norm": 4.183342933654785,
|
8537 |
+
"learning_rate": 1.278511466887351e-05,
|
8538 |
+
"loss": 0.197,
|
8539 |
+
"step": 1214
|
8540 |
+
},
|
8541 |
+
{
|
8542 |
+
"epoch": 0.7038377986965967,
|
8543 |
+
"grad_norm": 1.6271179914474487,
|
8544 |
+
"learning_rate": 1.2678438693135386e-05,
|
8545 |
+
"loss": 0.0874,
|
8546 |
+
"step": 1215
|
8547 |
+
},
|
8548 |
+
{
|
8549 |
+
"epoch": 0.7044170890658943,
|
8550 |
+
"grad_norm": 2.17236065864563,
|
8551 |
+
"learning_rate": 1.2572179493964066e-05,
|
8552 |
+
"loss": 0.125,
|
8553 |
+
"step": 1216
|
8554 |
+
},
|
8555 |
+
{
|
8556 |
+
"epoch": 0.7049963794351919,
|
8557 |
+
"grad_norm": 1.1981199979782104,
|
8558 |
+
"learning_rate": 1.2466337578523456e-05,
|
8559 |
+
"loss": 0.0453,
|
8560 |
+
"step": 1217
|
8561 |
+
},
|
8562 |
+
{
|
8563 |
+
"epoch": 0.7055756698044895,
|
8564 |
+
"grad_norm": 1.2022173404693604,
|
8565 |
+
"learning_rate": 1.2360913451985801e-05,
|
8566 |
+
"loss": 0.0696,
|
8567 |
+
"step": 1218
|
8568 |
+
},
|
8569 |
+
{
|
8570 |
+
"epoch": 0.7061549601737871,
|
8571 |
+
"grad_norm": 1.0011115074157715,
|
8572 |
+
"learning_rate": 1.2255907617529306e-05,
|
8573 |
+
"loss": 0.0331,
|
8574 |
+
"step": 1219
|
8575 |
+
},
|
8576 |
+
{
|
8577 |
+
"epoch": 0.7067342505430847,
|
8578 |
+
"grad_norm": 1.7974885702133179,
|
8579 |
+
"learning_rate": 1.2151320576335701e-05,
|
8580 |
+
"loss": 0.1402,
|
8581 |
+
"step": 1220
|
8582 |
+
},
|
8583 |
+
{
|
8584 |
+
"epoch": 0.7073135409123823,
|
8585 |
+
"grad_norm": 0.9930383563041687,
|
8586 |
+
"learning_rate": 1.204715282758786e-05,
|
8587 |
+
"loss": 0.0354,
|
8588 |
+
"step": 1221
|
8589 |
+
},
|
8590 |
+
{
|
8591 |
+
"epoch": 0.7078928312816799,
|
8592 |
+
"grad_norm": 2.908696174621582,
|
8593 |
+
"learning_rate": 1.1943404868467412e-05,
|
8594 |
+
"loss": 0.2207,
|
8595 |
+
"step": 1222
|
8596 |
+
},
|
8597 |
+
{
|
8598 |
+
"epoch": 0.7084721216509775,
|
8599 |
+
"grad_norm": 1.904938817024231,
|
8600 |
+
"learning_rate": 1.1840077194152377e-05,
|
8601 |
+
"loss": 0.0381,
|
8602 |
+
"step": 1223
|
8603 |
+
},
|
8604 |
+
{
|
8605 |
+
"epoch": 0.7090514120202752,
|
8606 |
+
"grad_norm": 2.9847166538238525,
|
8607 |
+
"learning_rate": 1.1737170297814826e-05,
|
8608 |
+
"loss": 0.0851,
|
8609 |
+
"step": 1224
|
8610 |
+
},
|
8611 |
+
{
|
8612 |
+
"epoch": 0.7096307023895728,
|
8613 |
+
"grad_norm": 2.0300652980804443,
|
8614 |
+
"learning_rate": 1.1634684670618468e-05,
|
8615 |
+
"loss": 0.0678,
|
8616 |
+
"step": 1225
|
8617 |
+
},
|
8618 |
+
{
|
8619 |
+
"epoch": 0.7102099927588704,
|
8620 |
+
"grad_norm": 1.1873992681503296,
|
8621 |
+
"learning_rate": 1.1532620801716354e-05,
|
8622 |
+
"loss": 0.0354,
|
8623 |
+
"step": 1226
|
8624 |
+
},
|
8625 |
+
{
|
8626 |
+
"epoch": 0.710789283128168,
|
8627 |
+
"grad_norm": 2.6163318157196045,
|
8628 |
+
"learning_rate": 1.1430979178248513e-05,
|
8629 |
+
"loss": 0.131,
|
8630 |
+
"step": 1227
|
8631 |
+
},
|
8632 |
+
{
|
8633 |
+
"epoch": 0.7113685734974656,
|
8634 |
+
"grad_norm": 1.7732253074645996,
|
8635 |
+
"learning_rate": 1.1329760285339674e-05,
|
8636 |
+
"loss": 0.0999,
|
8637 |
+
"step": 1228
|
8638 |
+
},
|
8639 |
+
{
|
8640 |
+
"epoch": 0.7119478638667632,
|
8641 |
+
"grad_norm": 2.5668795108795166,
|
8642 |
+
"learning_rate": 1.122896460609687e-05,
|
8643 |
+
"loss": 0.1939,
|
8644 |
+
"step": 1229
|
8645 |
+
},
|
8646 |
+
{
|
8647 |
+
"epoch": 0.7125271542360608,
|
8648 |
+
"grad_norm": 0.7375625371932983,
|
8649 |
+
"learning_rate": 1.1128592621607226e-05,
|
8650 |
+
"loss": 0.0278,
|
8651 |
+
"step": 1230
|
8652 |
+
},
|
8653 |
+
{
|
8654 |
+
"epoch": 0.7131064446053584,
|
8655 |
+
"grad_norm": 1.7102712392807007,
|
8656 |
+
"learning_rate": 1.1028644810935563e-05,
|
8657 |
+
"loss": 0.0665,
|
8658 |
+
"step": 1231
|
8659 |
+
},
|
8660 |
+
{
|
8661 |
+
"epoch": 0.713685734974656,
|
8662 |
+
"grad_norm": 0.9485417604446411,
|
8663 |
+
"learning_rate": 1.0929121651122254e-05,
|
8664 |
+
"loss": 0.0376,
|
8665 |
+
"step": 1232
|
8666 |
+
},
|
8667 |
+
{
|
8668 |
+
"epoch": 0.7142650253439536,
|
8669 |
+
"grad_norm": 0.8667916655540466,
|
8670 |
+
"learning_rate": 1.0830023617180773e-05,
|
8671 |
+
"loss": 0.0427,
|
8672 |
+
"step": 1233
|
8673 |
+
},
|
8674 |
+
{
|
8675 |
+
"epoch": 0.7148443157132512,
|
8676 |
+
"grad_norm": 1.996333122253418,
|
8677 |
+
"learning_rate": 1.0731351182095539e-05,
|
8678 |
+
"loss": 0.0902,
|
8679 |
+
"step": 1234
|
8680 |
+
},
|
8681 |
+
{
|
8682 |
+
"epoch": 0.7154236060825488,
|
8683 |
+
"grad_norm": 1.2158366441726685,
|
8684 |
+
"learning_rate": 1.063310481681965e-05,
|
8685 |
+
"loss": 0.0491,
|
8686 |
+
"step": 1235
|
8687 |
+
},
|
8688 |
+
{
|
8689 |
+
"epoch": 0.7160028964518464,
|
8690 |
+
"grad_norm": 2.8177034854888916,
|
8691 |
+
"learning_rate": 1.0535284990272599e-05,
|
8692 |
+
"loss": 0.1753,
|
8693 |
+
"step": 1236
|
8694 |
+
},
|
8695 |
+
{
|
8696 |
+
"epoch": 0.716582186821144,
|
8697 |
+
"grad_norm": 1.7381958961486816,
|
8698 |
+
"learning_rate": 1.043789216933806e-05,
|
8699 |
+
"loss": 0.0869,
|
8700 |
+
"step": 1237
|
8701 |
+
},
|
8702 |
+
{
|
8703 |
+
"epoch": 0.7171614771904417,
|
8704 |
+
"grad_norm": 0.5652958750724792,
|
8705 |
+
"learning_rate": 1.034092681886164e-05,
|
8706 |
+
"loss": 0.0198,
|
8707 |
+
"step": 1238
|
8708 |
+
},
|
8709 |
+
{
|
8710 |
+
"epoch": 0.7177407675597394,
|
8711 |
+
"grad_norm": 2.561673402786255,
|
8712 |
+
"learning_rate": 1.0244389401648735e-05,
|
8713 |
+
"loss": 0.122,
|
8714 |
+
"step": 1239
|
8715 |
+
},
|
8716 |
+
{
|
8717 |
+
"epoch": 0.718320057929037,
|
8718 |
+
"grad_norm": 1.161051630973816,
|
8719 |
+
"learning_rate": 1.0148280378462182e-05,
|
8720 |
+
"loss": 0.0602,
|
8721 |
+
"step": 1240
|
8722 |
+
},
|
8723 |
+
{
|
8724 |
+
"epoch": 0.7188993482983346,
|
8725 |
+
"grad_norm": 1.1027592420578003,
|
8726 |
+
"learning_rate": 1.005260020802018e-05,
|
8727 |
+
"loss": 0.0735,
|
8728 |
+
"step": 1241
|
8729 |
+
},
|
8730 |
+
{
|
8731 |
+
"epoch": 0.7194786386676322,
|
8732 |
+
"grad_norm": 0.8367759585380554,
|
8733 |
+
"learning_rate": 9.957349346994038e-06,
|
8734 |
+
"loss": 0.0302,
|
8735 |
+
"step": 1242
|
8736 |
+
},
|
8737 |
+
{
|
8738 |
+
"epoch": 0.7200579290369298,
|
8739 |
+
"grad_norm": 1.3736345767974854,
|
8740 |
+
"learning_rate": 9.862528250006065e-06,
|
8741 |
+
"loss": 0.0503,
|
8742 |
+
"step": 1243
|
8743 |
+
},
|
8744 |
+
{
|
8745 |
+
"epoch": 0.7206372194062274,
|
8746 |
+
"grad_norm": 0.475423663854599,
|
8747 |
+
"learning_rate": 9.768137369627296e-06,
|
8748 |
+
"loss": 0.0184,
|
8749 |
+
"step": 1244
|
8750 |
+
},
|
8751 |
+
{
|
8752 |
+
"epoch": 0.721216509775525,
|
8753 |
+
"grad_norm": 1.0663799047470093,
|
8754 |
+
"learning_rate": 9.67417715637542e-06,
|
8755 |
+
"loss": 0.0543,
|
8756 |
+
"step": 1245
|
8757 |
+
},
|
8758 |
+
{
|
8759 |
+
"epoch": 0.7217958001448226,
|
8760 |
+
"grad_norm": 1.056991696357727,
|
8761 |
+
"learning_rate": 9.58064805871256e-06,
|
8762 |
+
"loss": 0.0565,
|
8763 |
+
"step": 1246
|
8764 |
+
},
|
8765 |
+
{
|
8766 |
+
"epoch": 0.7223750905141202,
|
8767 |
+
"grad_norm": 0.9765681624412537,
|
8768 |
+
"learning_rate": 9.487550523043232e-06,
|
8769 |
+
"loss": 0.0423,
|
8770 |
+
"step": 1247
|
8771 |
+
},
|
8772 |
+
{
|
8773 |
+
"epoch": 0.7229543808834178,
|
8774 |
+
"grad_norm": 1.9340802431106567,
|
8775 |
+
"learning_rate": 9.394884993712094e-06,
|
8776 |
+
"loss": 0.0771,
|
8777 |
+
"step": 1248
|
8778 |
+
},
|
8779 |
+
{
|
8780 |
+
"epoch": 0.7235336712527154,
|
8781 |
+
"grad_norm": 0.8700636625289917,
|
8782 |
+
"learning_rate": 9.302651913001914e-06,
|
8783 |
+
"loss": 0.0325,
|
8784 |
+
"step": 1249
|
8785 |
+
},
|
8786 |
+
{
|
8787 |
+
"epoch": 0.724112961622013,
|
8788 |
+
"grad_norm": 1.980541706085205,
|
8789 |
+
"learning_rate": 9.210851721131398e-06,
|
8790 |
+
"loss": 0.1062,
|
8791 |
+
"step": 1250
|
8792 |
+
},
|
8793 |
+
{
|
8794 |
+
"epoch": 0.7246922519913106,
|
8795 |
+
"grad_norm": 0.6983858346939087,
|
8796 |
+
"learning_rate": 9.11948485625318e-06,
|
8797 |
+
"loss": 0.0345,
|
8798 |
+
"step": 1251
|
8799 |
+
},
|
8800 |
+
{
|
8801 |
+
"epoch": 0.7252715423606082,
|
8802 |
+
"grad_norm": 1.797666311264038,
|
8803 |
+
"learning_rate": 9.028551754451619e-06,
|
8804 |
+
"loss": 0.1083,
|
8805 |
+
"step": 1252
|
8806 |
+
},
|
8807 |
+
{
|
8808 |
+
"epoch": 0.7258508327299059,
|
8809 |
+
"grad_norm": 2.497971296310425,
|
8810 |
+
"learning_rate": 8.938052849740775e-06,
|
8811 |
+
"loss": 0.0968,
|
8812 |
+
"step": 1253
|
8813 |
+
},
|
8814 |
+
{
|
8815 |
+
"epoch": 0.7264301230992035,
|
8816 |
+
"grad_norm": 1.3927079439163208,
|
8817 |
+
"learning_rate": 8.84798857406235e-06,
|
8818 |
+
"loss": 0.0743,
|
8819 |
+
"step": 1254
|
8820 |
+
},
|
8821 |
+
{
|
8822 |
+
"epoch": 0.7270094134685011,
|
8823 |
+
"grad_norm": 0.7377947568893433,
|
8824 |
+
"learning_rate": 8.758359357283574e-06,
|
8825 |
+
"loss": 0.028,
|
8826 |
+
"step": 1255
|
8827 |
+
},
|
8828 |
+
{
|
8829 |
+
"epoch": 0.7275887038377987,
|
8830 |
+
"grad_norm": 1.269887924194336,
|
8831 |
+
"learning_rate": 8.669165627195242e-06,
|
8832 |
+
"loss": 0.0372,
|
8833 |
+
"step": 1256
|
8834 |
+
},
|
8835 |
+
{
|
8836 |
+
"epoch": 0.7281679942070963,
|
8837 |
+
"grad_norm": 1.1663178205490112,
|
8838 |
+
"learning_rate": 8.58040780950955e-06,
|
8839 |
+
"loss": 0.0304,
|
8840 |
+
"step": 1257
|
8841 |
+
},
|
8842 |
+
{
|
8843 |
+
"epoch": 0.7287472845763939,
|
8844 |
+
"grad_norm": 0.9889265894889832,
|
8845 |
+
"learning_rate": 8.492086327858185e-06,
|
8846 |
+
"loss": 0.0344,
|
8847 |
+
"step": 1258
|
8848 |
+
},
|
8849 |
+
{
|
8850 |
+
"epoch": 0.7293265749456915,
|
8851 |
+
"grad_norm": 3.2795023918151855,
|
8852 |
+
"learning_rate": 8.404201603790207e-06,
|
8853 |
+
"loss": 0.1686,
|
8854 |
+
"step": 1259
|
8855 |
+
},
|
8856 |
+
{
|
8857 |
+
"epoch": 0.7299058653149891,
|
8858 |
+
"grad_norm": 1.1463748216629028,
|
8859 |
+
"learning_rate": 8.316754056770138e-06,
|
8860 |
+
"loss": 0.0428,
|
8861 |
+
"step": 1260
|
8862 |
+
},
|
8863 |
+
{
|
8864 |
+
"epoch": 0.7304851556842867,
|
8865 |
+
"grad_norm": 1.3406840562820435,
|
8866 |
+
"learning_rate": 8.229744104175829e-06,
|
8867 |
+
"loss": 0.061,
|
8868 |
+
"step": 1261
|
8869 |
+
},
|
8870 |
+
{
|
8871 |
+
"epoch": 0.7310644460535843,
|
8872 |
+
"grad_norm": 1.2390825748443604,
|
8873 |
+
"learning_rate": 8.143172161296586e-06,
|
8874 |
+
"loss": 0.0476,
|
8875 |
+
"step": 1262
|
8876 |
+
},
|
8877 |
+
{
|
8878 |
+
"epoch": 0.7316437364228819,
|
8879 |
+
"grad_norm": 1.9425374269485474,
|
8880 |
+
"learning_rate": 8.057038641331104e-06,
|
8881 |
+
"loss": 0.0919,
|
8882 |
+
"step": 1263
|
8883 |
+
},
|
8884 |
+
{
|
8885 |
+
"epoch": 0.7322230267921795,
|
8886 |
+
"grad_norm": 1.4705849885940552,
|
8887 |
+
"learning_rate": 7.97134395538559e-06,
|
8888 |
+
"loss": 0.1943,
|
8889 |
+
"step": 1264
|
8890 |
+
},
|
8891 |
+
{
|
8892 |
+
"epoch": 0.7328023171614771,
|
8893 |
+
"grad_norm": 1.2691655158996582,
|
8894 |
+
"learning_rate": 7.886088512471678e-06,
|
8895 |
+
"loss": 0.05,
|
8896 |
+
"step": 1265
|
8897 |
+
},
|
8898 |
+
{
|
8899 |
+
"epoch": 0.7333816075307747,
|
8900 |
+
"grad_norm": 1.0049604177474976,
|
8901 |
+
"learning_rate": 7.801272719504593e-06,
|
8902 |
+
"loss": 0.0441,
|
8903 |
+
"step": 1266
|
8904 |
+
},
|
8905 |
+
{
|
8906 |
+
"epoch": 0.7339608979000725,
|
8907 |
+
"grad_norm": 1.5805842876434326,
|
8908 |
+
"learning_rate": 7.716896981301113e-06,
|
8909 |
+
"loss": 0.0606,
|
8910 |
+
"step": 1267
|
8911 |
+
},
|
8912 |
+
{
|
8913 |
+
"epoch": 0.7345401882693701,
|
8914 |
+
"grad_norm": 0.8633338809013367,
|
8915 |
+
"learning_rate": 7.632961700577723e-06,
|
8916 |
+
"loss": 0.0257,
|
8917 |
+
"step": 1268
|
8918 |
+
},
|
8919 |
+
{
|
8920 |
+
"epoch": 0.7351194786386677,
|
8921 |
+
"grad_norm": 2.1949973106384277,
|
8922 |
+
"learning_rate": 7.5494672779485965e-06,
|
8923 |
+
"loss": 0.0517,
|
8924 |
+
"step": 1269
|
8925 |
+
},
|
8926 |
+
{
|
8927 |
+
"epoch": 0.7356987690079653,
|
8928 |
+
"grad_norm": 1.3663002252578735,
|
8929 |
+
"learning_rate": 7.466414111923814e-06,
|
8930 |
+
"loss": 0.067,
|
8931 |
+
"step": 1270
|
8932 |
+
},
|
8933 |
+
{
|
8934 |
+
"epoch": 0.7362780593772629,
|
8935 |
+
"grad_norm": 2.0407004356384277,
|
8936 |
+
"learning_rate": 7.38380259890733e-06,
|
8937 |
+
"loss": 0.09,
|
8938 |
+
"step": 1271
|
8939 |
+
},
|
8940 |
+
{
|
8941 |
+
"epoch": 0.7368573497465605,
|
8942 |
+
"grad_norm": 0.6014885902404785,
|
8943 |
+
"learning_rate": 7.301633133195141e-06,
|
8944 |
+
"loss": 0.0231,
|
8945 |
+
"step": 1272
|
8946 |
+
},
|
8947 |
+
{
|
8948 |
+
"epoch": 0.7374366401158581,
|
8949 |
+
"grad_norm": 0.9976663589477539,
|
8950 |
+
"learning_rate": 7.219906106973418e-06,
|
8951 |
+
"loss": 0.0332,
|
8952 |
+
"step": 1273
|
8953 |
+
},
|
8954 |
+
{
|
8955 |
+
"epoch": 0.7380159304851557,
|
8956 |
+
"grad_norm": 1.190585732460022,
|
8957 |
+
"learning_rate": 7.138621910316589e-06,
|
8958 |
+
"loss": 0.0513,
|
8959 |
+
"step": 1274
|
8960 |
+
},
|
8961 |
+
{
|
8962 |
+
"epoch": 0.7385952208544533,
|
8963 |
+
"grad_norm": 1.752960205078125,
|
8964 |
+
"learning_rate": 7.0577809311855425e-06,
|
8965 |
+
"loss": 0.0647,
|
8966 |
+
"step": 1275
|
8967 |
+
},
|
8968 |
+
{
|
8969 |
+
"epoch": 0.7391745112237509,
|
8970 |
+
"grad_norm": 0.5577161312103271,
|
8971 |
+
"learning_rate": 6.9773835554256875e-06,
|
8972 |
+
"loss": 0.0201,
|
8973 |
+
"step": 1276
|
8974 |
+
},
|
8975 |
+
{
|
8976 |
+
"epoch": 0.7397538015930485,
|
8977 |
+
"grad_norm": 3.0362846851348877,
|
8978 |
+
"learning_rate": 6.897430166765195e-06,
|
8979 |
+
"loss": 0.1115,
|
8980 |
+
"step": 1277
|
8981 |
+
},
|
8982 |
+
{
|
8983 |
+
"epoch": 0.7403330919623461,
|
8984 |
+
"grad_norm": 1.245192289352417,
|
8985 |
+
"learning_rate": 6.817921146813111e-06,
|
8986 |
+
"loss": 0.0389,
|
8987 |
+
"step": 1278
|
8988 |
+
},
|
8989 |
+
{
|
8990 |
+
"epoch": 0.7409123823316437,
|
8991 |
+
"grad_norm": 2.4826786518096924,
|
8992 |
+
"learning_rate": 6.738856875057598e-06,
|
8993 |
+
"loss": 0.0736,
|
8994 |
+
"step": 1279
|
8995 |
+
},
|
8996 |
+
{
|
8997 |
+
"epoch": 0.7414916727009413,
|
8998 |
+
"grad_norm": 0.8865545392036438,
|
8999 |
+
"learning_rate": 6.660237728864039e-06,
|
9000 |
+
"loss": 0.0248,
|
9001 |
+
"step": 1280
|
9002 |
+
},
|
9003 |
+
{
|
9004 |
+
"epoch": 0.742070963070239,
|
9005 |
+
"grad_norm": 1.8779815435409546,
|
9006 |
+
"learning_rate": 6.5820640834733045e-06,
|
9007 |
+
"loss": 0.125,
|
9008 |
+
"step": 1281
|
9009 |
+
},
|
9010 |
+
{
|
9011 |
+
"epoch": 0.7426502534395366,
|
9012 |
+
"grad_norm": 1.601940631866455,
|
9013 |
+
"learning_rate": 6.504336311999915e-06,
|
9014 |
+
"loss": 0.0529,
|
9015 |
+
"step": 1282
|
9016 |
+
},
|
9017 |
+
{
|
9018 |
+
"epoch": 0.7432295438088342,
|
9019 |
+
"grad_norm": 2.002863645553589,
|
9020 |
+
"learning_rate": 6.4270547854302974e-06,
|
9021 |
+
"loss": 0.0634,
|
9022 |
+
"step": 1283
|
9023 |
+
},
|
9024 |
+
{
|
9025 |
+
"epoch": 0.7438088341781318,
|
9026 |
+
"grad_norm": 0.6799012422561646,
|
9027 |
+
"learning_rate": 6.3502198726209885e-06,
|
9028 |
+
"loss": 0.0368,
|
9029 |
+
"step": 1284
|
9030 |
+
},
|
9031 |
+
{
|
9032 |
+
"epoch": 0.7443881245474294,
|
9033 |
+
"grad_norm": 1.7904435396194458,
|
9034 |
+
"learning_rate": 6.273831940296904e-06,
|
9035 |
+
"loss": 0.0613,
|
9036 |
+
"step": 1285
|
9037 |
+
},
|
9038 |
+
{
|
9039 |
+
"epoch": 0.744967414916727,
|
9040 |
+
"grad_norm": 1.3958054780960083,
|
9041 |
+
"learning_rate": 6.197891353049545e-06,
|
9042 |
+
"loss": 0.0754,
|
9043 |
+
"step": 1286
|
9044 |
+
},
|
9045 |
+
{
|
9046 |
+
"epoch": 0.7455467052860246,
|
9047 |
+
"grad_norm": 0.5189615488052368,
|
9048 |
+
"learning_rate": 6.1223984733353114e-06,
|
9049 |
+
"loss": 0.0256,
|
9050 |
+
"step": 1287
|
9051 |
+
},
|
9052 |
+
{
|
9053 |
+
"epoch": 0.7461259956553222,
|
9054 |
+
"grad_norm": 0.9718292355537415,
|
9055 |
+
"learning_rate": 6.047353661473698e-06,
|
9056 |
+
"loss": 0.0299,
|
9057 |
+
"step": 1288
|
9058 |
+
},
|
9059 |
+
{
|
9060 |
+
"epoch": 0.7467052860246198,
|
9061 |
+
"grad_norm": 0.8095974922180176,
|
9062 |
+
"learning_rate": 5.97275727564568e-06,
|
9063 |
+
"loss": 0.0368,
|
9064 |
+
"step": 1289
|
9065 |
+
},
|
9066 |
+
{
|
9067 |
+
"epoch": 0.7472845763939174,
|
9068 |
+
"grad_norm": 0.6793989539146423,
|
9069 |
+
"learning_rate": 5.898609671891897e-06,
|
9070 |
+
"loss": 0.0276,
|
9071 |
+
"step": 1290
|
9072 |
+
},
|
9073 |
+
{
|
9074 |
+
"epoch": 0.747863866763215,
|
9075 |
+
"grad_norm": 0.8708273768424988,
|
9076 |
+
"learning_rate": 5.824911204111006e-06,
|
9077 |
+
"loss": 0.0374,
|
9078 |
+
"step": 1291
|
9079 |
+
},
|
9080 |
+
{
|
9081 |
+
"epoch": 0.7484431571325126,
|
9082 |
+
"grad_norm": 1.5355048179626465,
|
9083 |
+
"learning_rate": 5.751662224057997e-06,
|
9084 |
+
"loss": 0.0386,
|
9085 |
+
"step": 1292
|
9086 |
+
},
|
9087 |
+
{
|
9088 |
+
"epoch": 0.7490224475018102,
|
9089 |
+
"grad_norm": 1.0437235832214355,
|
9090 |
+
"learning_rate": 5.6788630813424785e-06,
|
9091 |
+
"loss": 0.0582,
|
9092 |
+
"step": 1293
|
9093 |
+
},
|
9094 |
+
{
|
9095 |
+
"epoch": 0.7496017378711078,
|
9096 |
+
"grad_norm": 0.8630734086036682,
|
9097 |
+
"learning_rate": 5.606514123427076e-06,
|
9098 |
+
"loss": 0.0304,
|
9099 |
+
"step": 1294
|
9100 |
+
},
|
9101 |
+
{
|
9102 |
+
"epoch": 0.7501810282404056,
|
9103 |
+
"grad_norm": 1.2198039293289185,
|
9104 |
+
"learning_rate": 5.534615695625689e-06,
|
9105 |
+
"loss": 0.0347,
|
9106 |
+
"step": 1295
|
9107 |
+
},
|
9108 |
+
{
|
9109 |
+
"epoch": 0.7507603186097032,
|
9110 |
+
"grad_norm": 1.917325496673584,
|
9111 |
+
"learning_rate": 5.463168141101904e-06,
|
9112 |
+
"loss": 0.1005,
|
9113 |
+
"step": 1296
|
9114 |
+
},
|
9115 |
+
{
|
9116 |
+
"epoch": 0.7513396089790008,
|
9117 |
+
"grad_norm": 1.0141617059707642,
|
9118 |
+
"learning_rate": 5.3921718008673185e-06,
|
9119 |
+
"loss": 0.0625,
|
9120 |
+
"step": 1297
|
9121 |
+
},
|
9122 |
+
{
|
9123 |
+
"epoch": 0.7519188993482984,
|
9124 |
+
"grad_norm": 0.7988129258155823,
|
9125 |
+
"learning_rate": 5.321627013779984e-06,
|
9126 |
+
"loss": 0.03,
|
9127 |
+
"step": 1298
|
9128 |
+
},
|
9129 |
+
{
|
9130 |
+
"epoch": 0.752498189717596,
|
9131 |
+
"grad_norm": 1.2913761138916016,
|
9132 |
+
"learning_rate": 5.251534116542678e-06,
|
9133 |
+
"loss": 0.0555,
|
9134 |
+
"step": 1299
|
9135 |
+
},
|
9136 |
+
{
|
9137 |
+
"epoch": 0.7530774800868936,
|
9138 |
+
"grad_norm": 2.239727020263672,
|
9139 |
+
"learning_rate": 5.18189344370138e-06,
|
9140 |
+
"loss": 0.0896,
|
9141 |
+
"step": 1300
|
9142 |
+
},
|
9143 |
+
{
|
9144 |
+
"epoch": 0.7536567704561912,
|
9145 |
+
"grad_norm": 0.9208332300186157,
|
9146 |
+
"learning_rate": 5.1127053276436524e-06,
|
9147 |
+
"loss": 0.0509,
|
9148 |
+
"step": 1301
|
9149 |
+
},
|
9150 |
+
{
|
9151 |
+
"epoch": 0.7542360608254888,
|
9152 |
+
"grad_norm": 1.2946892976760864,
|
9153 |
+
"learning_rate": 5.04397009859704e-06,
|
9154 |
+
"loss": 0.0425,
|
9155 |
+
"step": 1302
|
9156 |
+
},
|
9157 |
+
{
|
9158 |
+
"epoch": 0.7548153511947864,
|
9159 |
+
"grad_norm": 1.2358165979385376,
|
9160 |
+
"learning_rate": 4.975688084627528e-06,
|
9161 |
+
"loss": 0.0503,
|
9162 |
+
"step": 1303
|
9163 |
+
},
|
9164 |
+
{
|
9165 |
+
"epoch": 0.755394641564084,
|
9166 |
+
"grad_norm": 1.4641838073730469,
|
9167 |
+
"learning_rate": 4.907859611637911e-06,
|
9168 |
+
"loss": 0.0636,
|
9169 |
+
"step": 1304
|
9170 |
+
},
|
9171 |
+
{
|
9172 |
+
"epoch": 0.7559739319333816,
|
9173 |
+
"grad_norm": 0.713836133480072,
|
9174 |
+
"learning_rate": 4.840485003366324e-06,
|
9175 |
+
"loss": 0.035,
|
9176 |
+
"step": 1305
|
9177 |
+
},
|
9178 |
+
{
|
9179 |
+
"epoch": 0.7565532223026792,
|
9180 |
+
"grad_norm": 1.1142395734786987,
|
9181 |
+
"learning_rate": 4.773564581384659e-06,
|
9182 |
+
"loss": 0.0671,
|
9183 |
+
"step": 1306
|
9184 |
+
},
|
9185 |
+
{
|
9186 |
+
"epoch": 0.7571325126719768,
|
9187 |
+
"grad_norm": 1.5183186531066895,
|
9188 |
+
"learning_rate": 4.707098665096987e-06,
|
9189 |
+
"loss": 0.0708,
|
9190 |
+
"step": 1307
|
9191 |
+
},
|
9192 |
+
{
|
9193 |
+
"epoch": 0.7577118030412744,
|
9194 |
+
"grad_norm": 2.752210855484009,
|
9195 |
+
"learning_rate": 4.6410875717380895e-06,
|
9196 |
+
"loss": 0.0989,
|
9197 |
+
"step": 1308
|
9198 |
+
},
|
9199 |
+
{
|
9200 |
+
"epoch": 0.7582910934105721,
|
9201 |
+
"grad_norm": 1.4349359273910522,
|
9202 |
+
"learning_rate": 4.5755316163719194e-06,
|
9203 |
+
"loss": 0.0413,
|
9204 |
+
"step": 1309
|
9205 |
+
},
|
9206 |
+
{
|
9207 |
+
"epoch": 0.7588703837798697,
|
9208 |
+
"grad_norm": 1.7319257259368896,
|
9209 |
+
"learning_rate": 4.510431111890134e-06,
|
9210 |
+
"loss": 0.078,
|
9211 |
+
"step": 1310
|
9212 |
+
},
|
9213 |
+
{
|
9214 |
+
"epoch": 0.7594496741491673,
|
9215 |
+
"grad_norm": 2.1485910415649414,
|
9216 |
+
"learning_rate": 4.445786369010518e-06,
|
9217 |
+
"loss": 0.1709,
|
9218 |
+
"step": 1311
|
9219 |
+
},
|
9220 |
+
{
|
9221 |
+
"epoch": 0.7600289645184649,
|
9222 |
+
"grad_norm": 1.6092329025268555,
|
9223 |
+
"learning_rate": 4.381597696275596e-06,
|
9224 |
+
"loss": 0.0631,
|
9225 |
+
"step": 1312
|
9226 |
+
},
|
9227 |
+
{
|
9228 |
+
"epoch": 0.7606082548877625,
|
9229 |
+
"grad_norm": 1.7530405521392822,
|
9230 |
+
"learning_rate": 4.317865400051113e-06,
|
9231 |
+
"loss": 0.0524,
|
9232 |
+
"step": 1313
|
9233 |
+
},
|
9234 |
+
{
|
9235 |
+
"epoch": 0.7611875452570601,
|
9236 |
+
"grad_norm": 1.321370244026184,
|
9237 |
+
"learning_rate": 4.254589784524576e-06,
|
9238 |
+
"loss": 0.0573,
|
9239 |
+
"step": 1314
|
9240 |
+
},
|
9241 |
+
{
|
9242 |
+
"epoch": 0.7617668356263577,
|
9243 |
+
"grad_norm": 1.1057686805725098,
|
9244 |
+
"learning_rate": 4.191771151703794e-06,
|
9245 |
+
"loss": 0.0262,
|
9246 |
+
"step": 1315
|
9247 |
+
},
|
9248 |
+
{
|
9249 |
+
"epoch": 0.7623461259956553,
|
9250 |
+
"grad_norm": 1.0524578094482422,
|
9251 |
+
"learning_rate": 4.129409801415429e-06,
|
9252 |
+
"loss": 0.0635,
|
9253 |
+
"step": 1316
|
9254 |
+
},
|
9255 |
+
{
|
9256 |
+
"epoch": 0.7629254163649529,
|
9257 |
+
"grad_norm": 0.9433290958404541,
|
9258 |
+
"learning_rate": 4.067506031303636e-06,
|
9259 |
+
"loss": 0.0378,
|
9260 |
+
"step": 1317
|
9261 |
+
},
|
9262 |
+
{
|
9263 |
+
"epoch": 0.7635047067342505,
|
9264 |
+
"grad_norm": 0.4127956032752991,
|
9265 |
+
"learning_rate": 4.006060136828549e-06,
|
9266 |
+
"loss": 0.0149,
|
9267 |
+
"step": 1318
|
9268 |
+
},
|
9269 |
+
{
|
9270 |
+
"epoch": 0.7640839971035481,
|
9271 |
+
"grad_norm": 1.5808672904968262,
|
9272 |
+
"learning_rate": 3.945072411264905e-06,
|
9273 |
+
"loss": 0.1073,
|
9274 |
+
"step": 1319
|
9275 |
+
},
|
9276 |
+
{
|
9277 |
+
"epoch": 0.7646632874728457,
|
9278 |
+
"grad_norm": 0.6743228435516357,
|
9279 |
+
"learning_rate": 3.884543145700659e-06,
|
9280 |
+
"loss": 0.0289,
|
9281 |
+
"step": 1320
|
9282 |
+
},
|
9283 |
+
{
|
9284 |
+
"epoch": 0.7652425778421433,
|
9285 |
+
"grad_norm": 0.8340603709220886,
|
9286 |
+
"learning_rate": 3.824472629035603e-06,
|
9287 |
+
"loss": 0.0643,
|
9288 |
+
"step": 1321
|
9289 |
+
},
|
9290 |
+
{
|
9291 |
+
"epoch": 0.7658218682114409,
|
9292 |
+
"grad_norm": 0.6846963763237,
|
9293 |
+
"learning_rate": 3.7648611479799366e-06,
|
9294 |
+
"loss": 0.0276,
|
9295 |
+
"step": 1322
|
9296 |
+
},
|
9297 |
+
{
|
9298 |
+
"epoch": 0.7664011585807387,
|
9299 |
+
"grad_norm": 1.7906545400619507,
|
9300 |
+
"learning_rate": 3.7057089870529472e-06,
|
9301 |
+
"loss": 0.0938,
|
9302 |
+
"step": 1323
|
9303 |
+
},
|
9304 |
+
{
|
9305 |
+
"epoch": 0.7669804489500363,
|
9306 |
+
"grad_norm": 3.829669237136841,
|
9307 |
+
"learning_rate": 3.64701642858164e-06,
|
9308 |
+
"loss": 0.1381,
|
9309 |
+
"step": 1324
|
9310 |
+
},
|
9311 |
+
{
|
9312 |
+
"epoch": 0.7675597393193339,
|
9313 |
+
"grad_norm": 1.4311552047729492,
|
9314 |
+
"learning_rate": 3.5887837526993983e-06,
|
9315 |
+
"loss": 0.0644,
|
9316 |
+
"step": 1325
|
9317 |
+
},
|
9318 |
+
{
|
9319 |
+
"epoch": 0.7681390296886315,
|
9320 |
+
"grad_norm": 2.025052070617676,
|
9321 |
+
"learning_rate": 3.5310112373445948e-06,
|
9322 |
+
"loss": 0.0486,
|
9323 |
+
"step": 1326
|
9324 |
+
},
|
9325 |
+
{
|
9326 |
+
"epoch": 0.7687183200579291,
|
9327 |
+
"grad_norm": 0.40631604194641113,
|
9328 |
+
"learning_rate": 3.473699158259347e-06,
|
9329 |
+
"loss": 0.0223,
|
9330 |
+
"step": 1327
|
9331 |
+
},
|
9332 |
+
{
|
9333 |
+
"epoch": 0.7692976104272267,
|
9334 |
+
"grad_norm": 0.6289862990379333,
|
9335 |
+
"learning_rate": 3.416847788988131e-06,
|
9336 |
+
"loss": 0.0277,
|
9337 |
+
"step": 1328
|
9338 |
+
},
|
9339 |
+
{
|
9340 |
+
"epoch": 0.7698769007965243,
|
9341 |
+
"grad_norm": 0.9489920735359192,
|
9342 |
+
"learning_rate": 3.3604574008765155e-06,
|
9343 |
+
"loss": 0.0327,
|
9344 |
+
"step": 1329
|
9345 |
+
},
|
9346 |
+
{
|
9347 |
+
"epoch": 0.7704561911658219,
|
9348 |
+
"grad_norm": 1.060785174369812,
|
9349 |
+
"learning_rate": 3.3045282630698506e-06,
|
9350 |
+
"loss": 0.0688,
|
9351 |
+
"step": 1330
|
9352 |
+
},
|
9353 |
+
{
|
9354 |
+
"epoch": 0.7710354815351195,
|
9355 |
+
"grad_norm": 2.090697765350342,
|
9356 |
+
"learning_rate": 3.2490606425119917e-06,
|
9357 |
+
"loss": 0.0631,
|
9358 |
+
"step": 1331
|
9359 |
+
},
|
9360 |
+
{
|
9361 |
+
"epoch": 0.7716147719044171,
|
9362 |
+
"grad_norm": 0.9930267930030823,
|
9363 |
+
"learning_rate": 3.1940548039440132e-06,
|
9364 |
+
"loss": 0.0402,
|
9365 |
+
"step": 1332
|
9366 |
+
},
|
9367 |
+
{
|
9368 |
+
"epoch": 0.7721940622737147,
|
9369 |
+
"grad_norm": 0.5885985493659973,
|
9370 |
+
"learning_rate": 3.1395110099029623e-06,
|
9371 |
+
"loss": 0.0257,
|
9372 |
+
"step": 1333
|
9373 |
+
},
|
9374 |
+
{
|
9375 |
+
"epoch": 0.7727733526430123,
|
9376 |
+
"grad_norm": 0.8741908669471741,
|
9377 |
+
"learning_rate": 3.085429520720595e-06,
|
9378 |
+
"loss": 0.0274,
|
9379 |
+
"step": 1334
|
9380 |
+
},
|
9381 |
+
{
|
9382 |
+
"epoch": 0.7733526430123099,
|
9383 |
+
"grad_norm": 1.159607172012329,
|
9384 |
+
"learning_rate": 3.031810594522133e-06,
|
9385 |
+
"loss": 0.024,
|
9386 |
+
"step": 1335
|
9387 |
+
},
|
9388 |
+
{
|
9389 |
+
"epoch": 0.7739319333816075,
|
9390 |
+
"grad_norm": 1.7260888814926147,
|
9391 |
+
"learning_rate": 2.978654487225041e-06,
|
9392 |
+
"loss": 0.0742,
|
9393 |
+
"step": 1336
|
9394 |
+
},
|
9395 |
+
{
|
9396 |
+
"epoch": 0.7745112237509051,
|
9397 |
+
"grad_norm": 2.15091609954834,
|
9398 |
+
"learning_rate": 2.925961452537773e-06,
|
9399 |
+
"loss": 0.1026,
|
9400 |
+
"step": 1337
|
9401 |
+
},
|
9402 |
+
{
|
9403 |
+
"epoch": 0.7750905141202028,
|
9404 |
+
"grad_norm": 3.3468515872955322,
|
9405 |
+
"learning_rate": 2.8737317419586185e-06,
|
9406 |
+
"loss": 0.1434,
|
9407 |
+
"step": 1338
|
9408 |
+
},
|
9409 |
+
{
|
9410 |
+
"epoch": 0.7756698044895004,
|
9411 |
+
"grad_norm": 0.5634710788726807,
|
9412 |
+
"learning_rate": 2.821965604774457e-06,
|
9413 |
+
"loss": 0.0251,
|
9414 |
+
"step": 1339
|
9415 |
+
},
|
9416 |
+
{
|
9417 |
+
"epoch": 0.776249094858798,
|
9418 |
+
"grad_norm": 1.5229746103286743,
|
9419 |
+
"learning_rate": 2.7706632880595716e-06,
|
9420 |
+
"loss": 0.0652,
|
9421 |
+
"step": 1340
|
9422 |
+
},
|
9423 |
+
{
|
9424 |
+
"epoch": 0.7768283852280956,
|
9425 |
+
"grad_norm": 2.156667709350586,
|
9426 |
+
"learning_rate": 2.719825036674495e-06,
|
9427 |
+
"loss": 0.0475,
|
9428 |
+
"step": 1341
|
9429 |
+
},
|
9430 |
+
{
|
9431 |
+
"epoch": 0.7774076755973932,
|
9432 |
+
"grad_norm": 1.208805799484253,
|
9433 |
+
"learning_rate": 2.669451093264819e-06,
|
9434 |
+
"loss": 0.0505,
|
9435 |
+
"step": 1342
|
9436 |
+
},
|
9437 |
+
{
|
9438 |
+
"epoch": 0.7779869659666908,
|
9439 |
+
"grad_norm": 1.2439669370651245,
|
9440 |
+
"learning_rate": 2.6195416982600196e-06,
|
9441 |
+
"loss": 0.0375,
|
9442 |
+
"step": 1343
|
9443 |
+
},
|
9444 |
+
{
|
9445 |
+
"epoch": 0.7785662563359884,
|
9446 |
+
"grad_norm": 1.2280553579330444,
|
9447 |
+
"learning_rate": 2.570097089872392e-06,
|
9448 |
+
"loss": 0.0444,
|
9449 |
+
"step": 1344
|
9450 |
+
},
|
9451 |
+
{
|
9452 |
+
"epoch": 0.779145546705286,
|
9453 |
+
"grad_norm": 1.43001389503479,
|
9454 |
+
"learning_rate": 2.5211175040958048e-06,
|
9455 |
+
"loss": 0.0563,
|
9456 |
+
"step": 1345
|
9457 |
+
},
|
9458 |
+
{
|
9459 |
+
"epoch": 0.7797248370745836,
|
9460 |
+
"grad_norm": 1.2540639638900757,
|
9461 |
+
"learning_rate": 2.472603174704624e-06,
|
9462 |
+
"loss": 0.0696,
|
9463 |
+
"step": 1346
|
9464 |
+
},
|
9465 |
+
{
|
9466 |
+
"epoch": 0.7803041274438812,
|
9467 |
+
"grad_norm": 1.1704764366149902,
|
9468 |
+
"learning_rate": 2.4245543332526156e-06,
|
9469 |
+
"loss": 0.0371,
|
9470 |
+
"step": 1347
|
9471 |
+
},
|
9472 |
+
{
|
9473 |
+
"epoch": 0.7808834178131788,
|
9474 |
+
"grad_norm": 0.6710085272789001,
|
9475 |
+
"learning_rate": 2.37697120907181e-06,
|
9476 |
+
"loss": 0.0276,
|
9477 |
+
"step": 1348
|
9478 |
+
},
|
9479 |
+
{
|
9480 |
+
"epoch": 0.7814627081824764,
|
9481 |
+
"grad_norm": 0.4951589107513428,
|
9482 |
+
"learning_rate": 2.329854029271417e-06,
|
9483 |
+
"loss": 0.0182,
|
9484 |
+
"step": 1349
|
9485 |
+
},
|
9486 |
+
{
|
9487 |
+
"epoch": 0.782041998551774,
|
9488 |
+
"grad_norm": 0.790463387966156,
|
9489 |
+
"learning_rate": 2.283203018736757e-06,
|
9490 |
+
"loss": 0.0361,
|
9491 |
+
"step": 1350
|
9492 |
+
},
|
9493 |
+
{
|
9494 |
+
"epoch": 0.7826212889210716,
|
9495 |
+
"grad_norm": 0.7865434885025024,
|
9496 |
+
"learning_rate": 2.2370184001281546e-06,
|
9497 |
+
"loss": 0.0364,
|
9498 |
+
"step": 1351
|
9499 |
+
},
|
9500 |
+
{
|
9501 |
+
"epoch": 0.7832005792903693,
|
9502 |
+
"grad_norm": 0.5698888301849365,
|
9503 |
+
"learning_rate": 2.1913003938798916e-06,
|
9504 |
+
"loss": 0.0236,
|
9505 |
+
"step": 1352
|
9506 |
+
},
|
9507 |
+
{
|
9508 |
+
"epoch": 0.783779869659667,
|
9509 |
+
"grad_norm": 1.5720646381378174,
|
9510 |
+
"learning_rate": 2.1460492181992085e-06,
|
9511 |
+
"loss": 0.0317,
|
9512 |
+
"step": 1353
|
9513 |
+
},
|
9514 |
+
{
|
9515 |
+
"epoch": 0.7843591600289646,
|
9516 |
+
"grad_norm": 0.7214342355728149,
|
9517 |
+
"learning_rate": 2.101265089065152e-06,
|
9518 |
+
"loss": 0.0403,
|
9519 |
+
"step": 1354
|
9520 |
+
},
|
9521 |
+
{
|
9522 |
+
"epoch": 0.7849384503982622,
|
9523 |
+
"grad_norm": 1.176257848739624,
|
9524 |
+
"learning_rate": 2.056948220227639e-06,
|
9525 |
+
"loss": 0.0414,
|
9526 |
+
"step": 1355
|
9527 |
+
},
|
9528 |
+
{
|
9529 |
+
"epoch": 0.7855177407675598,
|
9530 |
+
"grad_norm": 1.4957693815231323,
|
9531 |
+
"learning_rate": 2.013098823206372e-06,
|
9532 |
+
"loss": 0.0758,
|
9533 |
+
"step": 1356
|
9534 |
+
},
|
9535 |
+
{
|
9536 |
+
"epoch": 0.7860970311368574,
|
9537 |
+
"grad_norm": 0.7468437552452087,
|
9538 |
+
"learning_rate": 1.9697171072899032e-06,
|
9539 |
+
"loss": 0.0278,
|
9540 |
+
"step": 1357
|
9541 |
+
},
|
9542 |
+
{
|
9543 |
+
"epoch": 0.786676321506155,
|
9544 |
+
"grad_norm": 0.5401385426521301,
|
9545 |
+
"learning_rate": 1.9268032795345613e-06,
|
9546 |
+
"loss": 0.0238,
|
9547 |
+
"step": 1358
|
9548 |
+
},
|
9549 |
+
{
|
9550 |
+
"epoch": 0.7872556118754526,
|
9551 |
+
"grad_norm": 0.7670809626579285,
|
9552 |
+
"learning_rate": 1.8843575447635043e-06,
|
9553 |
+
"loss": 0.0207,
|
9554 |
+
"step": 1359
|
9555 |
+
},
|
9556 |
+
{
|
9557 |
+
"epoch": 0.7878349022447502,
|
9558 |
+
"grad_norm": 1.0698966979980469,
|
9559 |
+
"learning_rate": 1.842380105565711e-06,
|
9560 |
+
"loss": 0.0346,
|
9561 |
+
"step": 1360
|
9562 |
+
},
|
9563 |
+
{
|
9564 |
+
"epoch": 0.7884141926140478,
|
9565 |
+
"grad_norm": 0.7072330117225647,
|
9566 |
+
"learning_rate": 1.8008711622950813e-06,
|
9567 |
+
"loss": 0.023,
|
9568 |
+
"step": 1361
|
9569 |
+
},
|
9570 |
+
{
|
9571 |
+
"epoch": 0.7889934829833454,
|
9572 |
+
"grad_norm": 0.8675196170806885,
|
9573 |
+
"learning_rate": 1.7598309130693825e-06,
|
9574 |
+
"loss": 0.0522,
|
9575 |
+
"step": 1362
|
9576 |
+
},
|
9577 |
+
{
|
9578 |
+
"epoch": 0.789572773352643,
|
9579 |
+
"grad_norm": 3.1522326469421387,
|
9580 |
+
"learning_rate": 1.7192595537694146e-06,
|
9581 |
+
"loss": 0.1146,
|
9582 |
+
"step": 1363
|
9583 |
+
},
|
9584 |
+
{
|
9585 |
+
"epoch": 0.7901520637219406,
|
9586 |
+
"grad_norm": 0.7621691823005676,
|
9587 |
+
"learning_rate": 1.679157278037946e-06,
|
9588 |
+
"loss": 0.0247,
|
9589 |
+
"step": 1364
|
9590 |
+
},
|
9591 |
+
{
|
9592 |
+
"epoch": 0.7907313540912382,
|
9593 |
+
"grad_norm": 0.5931485295295715,
|
9594 |
+
"learning_rate": 1.6395242772789144e-06,
|
9595 |
+
"loss": 0.0257,
|
9596 |
+
"step": 1365
|
9597 |
+
},
|
9598 |
+
{
|
9599 |
+
"epoch": 0.7913106444605359,
|
9600 |
+
"grad_norm": 0.9685682058334351,
|
9601 |
+
"learning_rate": 1.6003607406564148e-06,
|
9602 |
+
"loss": 0.0533,
|
9603 |
+
"step": 1366
|
9604 |
+
},
|
9605 |
+
{
|
9606 |
+
"epoch": 0.7918899348298335,
|
9607 |
+
"grad_norm": 0.7118789553642273,
|
9608 |
+
"learning_rate": 1.5616668550938796e-06,
|
9609 |
+
"loss": 0.0373,
|
9610 |
+
"step": 1367
|
9611 |
+
},
|
9612 |
+
{
|
9613 |
+
"epoch": 0.7924692251991311,
|
9614 |
+
"grad_norm": 1.1625103950500488,
|
9615 |
+
"learning_rate": 1.523442805273123e-06,
|
9616 |
+
"loss": 0.0813,
|
9617 |
+
"step": 1368
|
9618 |
+
},
|
9619 |
+
{
|
9620 |
+
"epoch": 0.7930485155684287,
|
9621 |
+
"grad_norm": 0.6741803288459778,
|
9622 |
+
"learning_rate": 1.485688773633487e-06,
|
9623 |
+
"loss": 0.0348,
|
9624 |
+
"step": 1369
|
9625 |
+
},
|
9626 |
+
{
|
9627 |
+
"epoch": 0.7936278059377263,
|
9628 |
+
"grad_norm": 2.2038042545318604,
|
9629 |
+
"learning_rate": 1.448404940370951e-06,
|
9630 |
+
"loss": 0.1115,
|
9631 |
+
"step": 1370
|
9632 |
+
},
|
9633 |
+
{
|
9634 |
+
"epoch": 0.7942070963070239,
|
9635 |
+
"grad_norm": 1.7105461359024048,
|
9636 |
+
"learning_rate": 1.4115914834373245e-06,
|
9637 |
+
"loss": 0.0618,
|
9638 |
+
"step": 1371
|
9639 |
+
},
|
9640 |
+
{
|
9641 |
+
"epoch": 0.7947863866763215,
|
9642 |
+
"grad_norm": 1.4145504236221313,
|
9643 |
+
"learning_rate": 1.3752485785393455e-06,
|
9644 |
+
"loss": 0.0513,
|
9645 |
+
"step": 1372
|
9646 |
+
},
|
9647 |
+
{
|
9648 |
+
"epoch": 0.7953656770456191,
|
9649 |
+
"grad_norm": 0.9347787499427795,
|
9650 |
+
"learning_rate": 1.3393763991378371e-06,
|
9651 |
+
"loss": 0.0435,
|
9652 |
+
"step": 1373
|
9653 |
+
},
|
9654 |
+
{
|
9655 |
+
"epoch": 0.7959449674149167,
|
9656 |
+
"grad_norm": 1.513407826423645,
|
9657 |
+
"learning_rate": 1.3039751164469095e-06,
|
9658 |
+
"loss": 0.0339,
|
9659 |
+
"step": 1374
|
9660 |
+
},
|
9661 |
+
{
|
9662 |
+
"epoch": 0.7965242577842143,
|
9663 |
+
"grad_norm": 2.0904335975646973,
|
9664 |
+
"learning_rate": 1.2690448994331472e-06,
|
9665 |
+
"loss": 0.0687,
|
9666 |
+
"step": 1375
|
9667 |
+
},
|
9668 |
+
{
|
9669 |
+
"epoch": 0.7971035481535119,
|
9670 |
+
"grad_norm": 0.7092882394790649,
|
9671 |
+
"learning_rate": 1.2345859148147787e-06,
|
9672 |
+
"loss": 0.0213,
|
9673 |
+
"step": 1376
|
9674 |
+
},
|
9675 |
+
{
|
9676 |
+
"epoch": 0.7976828385228095,
|
9677 |
+
"grad_norm": 1.844236969947815,
|
9678 |
+
"learning_rate": 1.2005983270608756e-06,
|
9679 |
+
"loss": 0.0714,
|
9680 |
+
"step": 1377
|
9681 |
+
},
|
9682 |
+
{
|
9683 |
+
"epoch": 0.7982621288921071,
|
9684 |
+
"grad_norm": 2.195664405822754,
|
9685 |
+
"learning_rate": 1.1670822983906094e-06,
|
9686 |
+
"loss": 0.0482,
|
9687 |
+
"step": 1378
|
9688 |
+
},
|
9689 |
+
{
|
9690 |
+
"epoch": 0.7988414192614047,
|
9691 |
+
"grad_norm": 0.9162880778312683,
|
9692 |
+
"learning_rate": 1.1340379887724295e-06,
|
9693 |
+
"loss": 0.0235,
|
9694 |
+
"step": 1379
|
9695 |
+
},
|
9696 |
+
{
|
9697 |
+
"epoch": 0.7994207096307024,
|
9698 |
+
"grad_norm": 0.8054811358451843,
|
9699 |
+
"learning_rate": 1.1014655559233312e-06,
|
9700 |
+
"loss": 0.0337,
|
9701 |
+
"step": 1380
|
9702 |
+
},
|
9703 |
+
{
|
9704 |
+
"epoch": 0.8,
|
9705 |
+
"grad_norm": 0.36062493920326233,
|
9706 |
+
"learning_rate": 1.0693651553080775e-06,
|
9707 |
+
"loss": 0.0276,
|
9708 |
+
"step": 1381
|
9709 |
+
},
|
9710 |
+
{
|
9711 |
+
"epoch": 0.8005792903692976,
|
9712 |
+
"grad_norm": 0.8155601024627686,
|
9713 |
+
"learning_rate": 1.0377369401385007e-06,
|
9714 |
+
"loss": 0.0241,
|
9715 |
+
"step": 1382
|
9716 |
+
},
|
9717 |
+
{
|
9718 |
+
"epoch": 0.8011585807385952,
|
9719 |
+
"grad_norm": 0.6971447467803955,
|
9720 |
+
"learning_rate": 1.0065810613727134e-06,
|
9721 |
+
"loss": 0.033,
|
9722 |
+
"step": 1383
|
9723 |
+
},
|
9724 |
+
{
|
9725 |
+
"epoch": 0.8017378711078929,
|
9726 |
+
"grad_norm": 0.7152788043022156,
|
9727 |
+
"learning_rate": 9.758976677144206e-07,
|
9728 |
+
"loss": 0.0271,
|
9729 |
+
"step": 1384
|
9730 |
+
},
|
9731 |
+
{
|
9732 |
+
"epoch": 0.8023171614771905,
|
9733 |
+
"grad_norm": 1.6824337244033813,
|
9734 |
+
"learning_rate": 9.456869056122197e-07,
|
9735 |
+
"loss": 0.0795,
|
9736 |
+
"step": 1385
|
9737 |
+
},
|
9738 |
+
{
|
9739 |
+
"epoch": 0.8028964518464881,
|
9740 |
+
"grad_norm": 1.2086037397384644,
|
9741 |
+
"learning_rate": 9.159489192588688e-07,
|
9742 |
+
"loss": 0.0413,
|
9743 |
+
"step": 1386
|
9744 |
+
},
|
9745 |
+
{
|
9746 |
+
"epoch": 0.8034757422157857,
|
9747 |
+
"grad_norm": 1.3253053426742554,
|
9748 |
+
"learning_rate": 8.866838505906417e-07,
|
9749 |
+
"loss": 0.0664,
|
9750 |
+
"step": 1387
|
9751 |
+
},
|
9752 |
+
{
|
9753 |
+
"epoch": 0.8040550325850833,
|
9754 |
+
"grad_norm": 3.2883996963500977,
|
9755 |
+
"learning_rate": 8.578918392866064e-07,
|
9756 |
+
"loss": 0.1577,
|
9757 |
+
"step": 1388
|
9758 |
+
},
|
9759 |
+
{
|
9760 |
+
"epoch": 0.8046343229543809,
|
9761 |
+
"grad_norm": 1.6165320873260498,
|
9762 |
+
"learning_rate": 8.29573022767982e-07,
|
9763 |
+
"loss": 0.1023,
|
9764 |
+
"step": 1389
|
9765 |
+
},
|
9766 |
+
{
|
9767 |
+
"epoch": 0.8052136133236785,
|
9768 |
+
"grad_norm": 0.6786884069442749,
|
9769 |
+
"learning_rate": 8.01727536197483e-07,
|
9770 |
+
"loss": 0.0231,
|
9771 |
+
"step": 1390
|
9772 |
+
},
|
9773 |
+
{
|
9774 |
+
"epoch": 0.8057929036929761,
|
9775 |
+
"grad_norm": 0.8029406070709229,
|
9776 |
+
"learning_rate": 7.74355512478675e-07,
|
9777 |
+
"loss": 0.0275,
|
9778 |
+
"step": 1391
|
9779 |
+
},
|
9780 |
+
{
|
9781 |
+
"epoch": 0.8063721940622737,
|
9782 |
+
"grad_norm": 2.034668445587158,
|
9783 |
+
"learning_rate": 7.474570822553317e-07,
|
9784 |
+
"loss": 0.1057,
|
9785 |
+
"step": 1392
|
9786 |
+
},
|
9787 |
+
{
|
9788 |
+
"epoch": 0.8069514844315713,
|
9789 |
+
"grad_norm": 2.6389472484588623,
|
9790 |
+
"learning_rate": 7.210323739107905e-07,
|
9791 |
+
"loss": 0.1175,
|
9792 |
+
"step": 1393
|
9793 |
+
},
|
9794 |
+
{
|
9795 |
+
"epoch": 0.807530774800869,
|
9796 |
+
"grad_norm": 0.9846882820129395,
|
9797 |
+
"learning_rate": 6.950815135674083e-07,
|
9798 |
+
"loss": 0.0479,
|
9799 |
+
"step": 1394
|
9800 |
+
},
|
9801 |
+
{
|
9802 |
+
"epoch": 0.8081100651701666,
|
9803 |
+
"grad_norm": 0.9405648708343506,
|
9804 |
+
"learning_rate": 6.696046250858845e-07,
|
9805 |
+
"loss": 0.0409,
|
9806 |
+
"step": 1395
|
9807 |
+
},
|
9808 |
+
{
|
9809 |
+
"epoch": 0.8086893555394642,
|
9810 |
+
"grad_norm": 2.827239990234375,
|
9811 |
+
"learning_rate": 6.446018300647172e-07,
|
9812 |
+
"loss": 0.1431,
|
9813 |
+
"step": 1396
|
9814 |
+
},
|
9815 |
+
{
|
9816 |
+
"epoch": 0.8092686459087618,
|
9817 |
+
"grad_norm": 1.554103136062622,
|
9818 |
+
"learning_rate": 6.200732478395921e-07,
|
9819 |
+
"loss": 0.1219,
|
9820 |
+
"step": 1397
|
9821 |
+
},
|
9822 |
+
{
|
9823 |
+
"epoch": 0.8098479362780594,
|
9824 |
+
"grad_norm": 1.1416951417922974,
|
9825 |
+
"learning_rate": 5.960189954828277e-07,
|
9826 |
+
"loss": 0.0586,
|
9827 |
+
"step": 1398
|
9828 |
+
},
|
9829 |
+
{
|
9830 |
+
"epoch": 0.810427226647357,
|
9831 |
+
"grad_norm": 1.481079339981079,
|
9832 |
+
"learning_rate": 5.724391878028535e-07,
|
9833 |
+
"loss": 0.0893,
|
9834 |
+
"step": 1399
|
9835 |
+
},
|
9836 |
+
{
|
9837 |
+
"epoch": 0.8110065170166546,
|
9838 |
+
"grad_norm": 0.8359711766242981,
|
9839 |
+
"learning_rate": 5.493339373435657e-07,
|
9840 |
+
"loss": 0.0211,
|
9841 |
+
"step": 1400
|
9842 |
+
},
|
9843 |
+
{
|
9844 |
+
"epoch": 0.8115858073859522,
|
9845 |
+
"grad_norm": 0.7014052867889404,
|
9846 |
+
"learning_rate": 5.26703354383884e-07,
|
9847 |
+
"loss": 0.0402,
|
9848 |
+
"step": 1401
|
9849 |
+
},
|
9850 |
+
{
|
9851 |
+
"epoch": 0.8121650977552498,
|
9852 |
+
"grad_norm": 1.0758659839630127,
|
9853 |
+
"learning_rate": 5.045475469371952e-07,
|
9854 |
+
"loss": 0.0365,
|
9855 |
+
"step": 1402
|
9856 |
+
},
|
9857 |
+
{
|
9858 |
+
"epoch": 0.8127443881245474,
|
9859 |
+
"grad_norm": 1.2663283348083496,
|
9860 |
+
"learning_rate": 4.828666207508214e-07,
|
9861 |
+
"loss": 0.053,
|
9862 |
+
"step": 1403
|
9863 |
+
},
|
9864 |
+
{
|
9865 |
+
"epoch": 0.813323678493845,
|
9866 |
+
"grad_norm": 0.5185549259185791,
|
9867 |
+
"learning_rate": 4.6166067930550895e-07,
|
9868 |
+
"loss": 0.0255,
|
9869 |
+
"step": 1404
|
9870 |
+
},
|
9871 |
+
{
|
9872 |
+
"epoch": 0.8139029688631426,
|
9873 |
+
"grad_norm": 0.9175311326980591,
|
9874 |
+
"learning_rate": 4.4092982381499505e-07,
|
9875 |
+
"loss": 0.0322,
|
9876 |
+
"step": 1405
|
9877 |
+
},
|
9878 |
+
{
|
9879 |
+
"epoch": 0.8144822592324402,
|
9880 |
+
"grad_norm": 0.8697713613510132,
|
9881 |
+
"learning_rate": 4.206741532254421e-07,
|
9882 |
+
"loss": 0.0499,
|
9883 |
+
"step": 1406
|
9884 |
+
},
|
9885 |
+
{
|
9886 |
+
"epoch": 0.8150615496017378,
|
9887 |
+
"grad_norm": 1.4453476667404175,
|
9888 |
+
"learning_rate": 4.008937642150379e-07,
|
9889 |
+
"loss": 0.0519,
|
9890 |
+
"step": 1407
|
9891 |
+
},
|
9892 |
+
{
|
9893 |
+
"epoch": 0.8156408399710355,
|
9894 |
+
"grad_norm": 1.0735323429107666,
|
9895 |
+
"learning_rate": 3.815887511934735e-07,
|
9896 |
+
"loss": 0.0478,
|
9897 |
+
"step": 1408
|
9898 |
+
},
|
9899 |
+
{
|
9900 |
+
"epoch": 0.8162201303403331,
|
9901 |
+
"grad_norm": 0.8734245896339417,
|
9902 |
+
"learning_rate": 3.6275920630152174e-07,
|
9903 |
+
"loss": 0.0356,
|
9904 |
+
"step": 1409
|
9905 |
+
},
|
9906 |
+
{
|
9907 |
+
"epoch": 0.8167994207096307,
|
9908 |
+
"grad_norm": 0.7365018129348755,
|
9909 |
+
"learning_rate": 3.444052194106262e-07,
|
9910 |
+
"loss": 0.0499,
|
9911 |
+
"step": 1410
|
9912 |
+
},
|
9913 |
+
{
|
9914 |
+
"epoch": 0.8173787110789283,
|
9915 |
+
"grad_norm": 1.058528184890747,
|
9916 |
+
"learning_rate": 3.265268781224129e-07,
|
9917 |
+
"loss": 0.0759,
|
9918 |
+
"step": 1411
|
9919 |
+
},
|
9920 |
+
{
|
9921 |
+
"epoch": 0.817958001448226,
|
9922 |
+
"grad_norm": 2.3366315364837646,
|
9923 |
+
"learning_rate": 3.091242677683126e-07,
|
9924 |
+
"loss": 0.1013,
|
9925 |
+
"step": 1412
|
9926 |
+
},
|
9927 |
+
{
|
9928 |
+
"epoch": 0.8185372918175235,
|
9929 |
+
"grad_norm": 0.6572679877281189,
|
9930 |
+
"learning_rate": 2.921974714091391e-07,
|
9931 |
+
"loss": 0.0375,
|
9932 |
+
"step": 1413
|
9933 |
+
},
|
9934 |
+
{
|
9935 |
+
"epoch": 0.8191165821868212,
|
9936 |
+
"grad_norm": 0.8239521384239197,
|
9937 |
+
"learning_rate": 2.757465698347117e-07,
|
9938 |
+
"loss": 0.0303,
|
9939 |
+
"step": 1414
|
9940 |
+
},
|
9941 |
+
{
|
9942 |
+
"epoch": 0.8196958725561188,
|
9943 |
+
"grad_norm": 1.0088764429092407,
|
9944 |
+
"learning_rate": 2.5977164156343327e-07,
|
9945 |
+
"loss": 0.0568,
|
9946 |
+
"step": 1415
|
9947 |
+
},
|
9948 |
+
{
|
9949 |
+
"epoch": 0.8202751629254164,
|
9950 |
+
"grad_norm": 1.2730189561843872,
|
9951 |
+
"learning_rate": 2.44272762841935e-07,
|
9952 |
+
"loss": 0.0607,
|
9953 |
+
"step": 1416
|
9954 |
+
},
|
9955 |
+
{
|
9956 |
+
"epoch": 0.820854453294714,
|
9957 |
+
"grad_norm": 0.6496265530586243,
|
9958 |
+
"learning_rate": 2.2925000764474347e-07,
|
9959 |
+
"loss": 0.0204,
|
9960 |
+
"step": 1417
|
9961 |
+
},
|
9962 |
+
{
|
9963 |
+
"epoch": 0.8214337436640116,
|
9964 |
+
"grad_norm": 1.9408307075500488,
|
9965 |
+
"learning_rate": 2.1470344767386963e-07,
|
9966 |
+
"loss": 0.0817,
|
9967 |
+
"step": 1418
|
9968 |
+
},
|
9969 |
+
{
|
9970 |
+
"epoch": 0.8220130340333092,
|
9971 |
+
"grad_norm": 1.823900580406189,
|
9972 |
+
"learning_rate": 2.006331523584981e-07,
|
9973 |
+
"loss": 0.081,
|
9974 |
+
"step": 1419
|
9975 |
+
},
|
9976 |
+
{
|
9977 |
+
"epoch": 0.8225923244026068,
|
9978 |
+
"grad_norm": 0.9203438758850098,
|
9979 |
+
"learning_rate": 1.870391888546652e-07,
|
9980 |
+
"loss": 0.0247,
|
9981 |
+
"step": 1420
|
9982 |
+
},
|
9983 |
+
{
|
9984 |
+
"epoch": 0.8231716147719044,
|
9985 |
+
"grad_norm": 1.078161597251892,
|
9986 |
+
"learning_rate": 1.739216220449258e-07,
|
9987 |
+
"loss": 0.053,
|
9988 |
+
"step": 1421
|
9989 |
+
},
|
9990 |
+
{
|
9991 |
+
"epoch": 0.823750905141202,
|
9992 |
+
"grad_norm": 1.4152048826217651,
|
9993 |
+
"learning_rate": 1.612805145380314e-07,
|
9994 |
+
"loss": 0.0628,
|
9995 |
+
"step": 1422
|
9996 |
+
},
|
9997 |
+
{
|
9998 |
+
"epoch": 0.8243301955104997,
|
9999 |
+
"grad_norm": 1.100527048110962,
|
10000 |
+
"learning_rate": 1.491159266686415e-07,
|
10001 |
+
"loss": 0.0391,
|
10002 |
+
"step": 1423
|
10003 |
+
},
|
10004 |
+
{
|
10005 |
+
"epoch": 0.8249094858797973,
|
10006 |
+
"grad_norm": 1.7918452024459839,
|
10007 |
+
"learning_rate": 1.374279164970571e-07,
|
10008 |
+
"loss": 0.0628,
|
10009 |
+
"step": 1424
|
10010 |
+
},
|
10011 |
+
{
|
10012 |
+
"epoch": 0.8254887762490949,
|
10013 |
+
"grad_norm": 1.4456841945648193,
|
10014 |
+
"learning_rate": 1.262165398089099e-07,
|
10015 |
+
"loss": 0.0454,
|
10016 |
+
"step": 1425
|
10017 |
+
},
|
10018 |
+
{
|
10019 |
+
"epoch": 0.8260680666183925,
|
10020 |
+
"grad_norm": 0.4844144284725189,
|
10021 |
+
"learning_rate": 1.1548185011490686e-07,
|
10022 |
+
"loss": 0.0209,
|
10023 |
+
"step": 1426
|
10024 |
+
},
|
10025 |
+
{
|
10026 |
+
"epoch": 0.8266473569876901,
|
10027 |
+
"grad_norm": 2.1798360347747803,
|
10028 |
+
"learning_rate": 1.0522389865059712e-07,
|
10029 |
+
"loss": 0.0744,
|
10030 |
+
"step": 1427
|
10031 |
+
},
|
10032 |
+
{
|
10033 |
+
"epoch": 0.8272266473569877,
|
10034 |
+
"grad_norm": 0.8768436908721924,
|
10035 |
+
"learning_rate": 9.544273437609441e-08,
|
10036 |
+
"loss": 0.0356,
|
10037 |
+
"step": 1428
|
10038 |
+
},
|
10039 |
+
{
|
10040 |
+
"epoch": 0.8278059377262853,
|
10041 |
+
"grad_norm": 1.1398060321807861,
|
10042 |
+
"learning_rate": 8.613840397585504e-08,
|
10043 |
+
"loss": 0.0376,
|
10044 |
+
"step": 1429
|
10045 |
+
},
|
10046 |
+
{
|
10047 |
+
"epoch": 0.8283852280955829,
|
10048 |
+
"grad_norm": 1.3790842294692993,
|
10049 |
+
"learning_rate": 7.731095185846693e-08,
|
10050 |
+
"loss": 0.0345,
|
10051 |
+
"step": 1430
|
10052 |
+
},
|
10053 |
+
{
|
10054 |
+
"epoch": 0.8289645184648805,
|
10055 |
+
"grad_norm": 1.3248194456100464,
|
10056 |
+
"learning_rate": 6.896042015641646e-08,
|
10057 |
+
"loss": 0.0505,
|
10058 |
+
"step": 1431
|
10059 |
+
},
|
10060 |
+
{
|
10061 |
+
"epoch": 0.8295438088341781,
|
10062 |
+
"grad_norm": 0.7751725912094116,
|
10063 |
+
"learning_rate": 6.108684872592197e-08,
|
10064 |
+
"loss": 0.0579,
|
10065 |
+
"step": 1432
|
10066 |
+
},
|
10067 |
+
{
|
10068 |
+
"epoch": 0.8301230992034757,
|
10069 |
+
"grad_norm": 1.2538723945617676,
|
10070 |
+
"learning_rate": 5.369027514668945e-08,
|
10071 |
+
"loss": 0.0634,
|
10072 |
+
"step": 1433
|
10073 |
+
},
|
10074 |
+
{
|
10075 |
+
"epoch": 0.8307023895727733,
|
10076 |
+
"grad_norm": 1.2392035722732544,
|
10077 |
+
"learning_rate": 4.67707347218016e-08,
|
10078 |
+
"loss": 0.1109,
|
10079 |
+
"step": 1434
|
10080 |
+
},
|
10081 |
+
{
|
10082 |
+
"epoch": 0.8312816799420709,
|
10083 |
+
"grad_norm": 1.1726109981536865,
|
10084 |
+
"learning_rate": 4.032826047747351e-08,
|
10085 |
+
"loss": 0.0548,
|
10086 |
+
"step": 1435
|
10087 |
+
},
|
10088 |
+
{
|
10089 |
+
"epoch": 0.8318609703113685,
|
10090 |
+
"grad_norm": 1.5222654342651367,
|
10091 |
+
"learning_rate": 3.436288316296388e-08,
|
10092 |
+
"loss": 0.0938,
|
10093 |
+
"step": 1436
|
10094 |
+
},
|
10095 |
+
{
|
10096 |
+
"epoch": 0.8324402606806662,
|
10097 |
+
"grad_norm": 0.9183850884437561,
|
10098 |
+
"learning_rate": 2.887463125037515e-08,
|
10099 |
+
"loss": 0.0465,
|
10100 |
+
"step": 1437
|
10101 |
+
},
|
10102 |
+
{
|
10103 |
+
"epoch": 0.8330195510499638,
|
10104 |
+
"grad_norm": 1.2826752662658691,
|
10105 |
+
"learning_rate": 2.3863530934564727e-08,
|
10106 |
+
"loss": 0.0635,
|
10107 |
+
"step": 1438
|
10108 |
+
},
|
10109 |
+
{
|
10110 |
+
"epoch": 0.8335988414192614,
|
10111 |
+
"grad_norm": 2.2082648277282715,
|
10112 |
+
"learning_rate": 1.9329606132978407e-08,
|
10113 |
+
"loss": 0.1066,
|
10114 |
+
"step": 1439
|
10115 |
+
},
|
10116 |
+
{
|
10117 |
+
"epoch": 0.834178131788559,
|
10118 |
+
"grad_norm": 1.5816879272460938,
|
10119 |
+
"learning_rate": 1.5272878485561582e-08,
|
10120 |
+
"loss": 0.0464,
|
10121 |
+
"step": 1440
|
10122 |
+
},
|
10123 |
+
{
|
10124 |
+
"epoch": 0.8347574221578566,
|
10125 |
+
"grad_norm": 1.03940749168396,
|
10126 |
+
"learning_rate": 1.169336735464821e-08,
|
10127 |
+
"loss": 0.0283,
|
10128 |
+
"step": 1441
|
10129 |
+
},
|
10130 |
+
{
|
10131 |
+
"epoch": 0.8353367125271542,
|
10132 |
+
"grad_norm": 0.4027893841266632,
|
10133 |
+
"learning_rate": 8.591089824860898e-09,
|
10134 |
+
"loss": 0.0184,
|
10135 |
+
"step": 1442
|
10136 |
+
},
|
10137 |
+
{
|
10138 |
+
"epoch": 0.8359160028964518,
|
10139 |
+
"grad_norm": 2.225884437561035,
|
10140 |
+
"learning_rate": 5.966060703044285e-09,
|
10141 |
+
"loss": 0.073,
|
10142 |
+
"step": 1443
|
10143 |
+
},
|
10144 |
+
{
|
10145 |
+
"epoch": 0.8364952932657495,
|
10146 |
+
"grad_norm": 1.024481177330017,
|
10147 |
+
"learning_rate": 3.818292518176225e-09,
|
10148 |
+
"loss": 0.0508,
|
10149 |
+
"step": 1444
|
10150 |
+
},
|
10151 |
+
{
|
10152 |
+
"epoch": 0.837074583635047,
|
10153 |
+
"grad_norm": 1.035207748413086,
|
10154 |
+
"learning_rate": 2.1477955213455857e-09,
|
10155 |
+
"loss": 0.0431,
|
10156 |
+
"step": 1445
|
10157 |
+
},
|
10158 |
+
{
|
10159 |
+
"epoch": 0.8376538740043447,
|
10160 |
+
"grad_norm": 1.363139271736145,
|
10161 |
+
"learning_rate": 9.545776856412225e-10,
|
10162 |
+
"loss": 0.0548,
|
10163 |
+
"step": 1446
|
10164 |
+
},
|
10165 |
+
{
|
10166 |
+
"epoch": 0.8382331643736423,
|
10167 |
+
"grad_norm": 1.283840537071228,
|
10168 |
+
"learning_rate": 2.3864470616308257e-10,
|
10169 |
+
"loss": 0.049,
|
10170 |
+
"step": 1447
|
10171 |
+
},
|
10172 |
+
{
|
10173 |
+
"epoch": 0.8388124547429399,
|
10174 |
+
"grad_norm": 1.5410807132720947,
|
10175 |
+
"learning_rate": 0.0,
|
10176 |
+
"loss": 0.0373,
|
10177 |
+
"step": 1448
|
10178 |
+
},
|
10179 |
+
{
|
10180 |
+
"epoch": 0.8388124547429399,
|
10181 |
+
"eval_loss": 0.059544339776039124,
|
10182 |
+
"eval_runtime": 24.4338,
|
10183 |
+
"eval_samples_per_second": 29.754,
|
10184 |
+
"eval_steps_per_second": 14.897,
|
10185 |
+
"step": 1448
|
10186 |
}
|
10187 |
],
|
10188 |
"logging_steps": 1,
|
|
|
10197 |
"should_evaluate": false,
|
10198 |
"should_log": false,
|
10199 |
"should_save": true,
|
10200 |
+
"should_training_stop": true
|
10201 |
},
|
10202 |
"attributes": {}
|
10203 |
}
|
10204 |
},
|
10205 |
+
"total_flos": 2.717472271588393e+17,
|
10206 |
"train_batch_size": 2,
|
10207 |
"trial_name": null,
|
10208 |
"trial_params": null
|