Training in progress, step 825, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 80792096
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2788e6dfb6435e62bc41b789bd050e58a00b8e97462570125c1f3a3b3a5752c3
|
3 |
size 80792096
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 41460084
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:51f9f9871646780dbef98536e3ba6799fe3cd11660641ef8398a22395fd35d67
|
3 |
size 41460084
|
last-checkpoint/rng_state.pth
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 14244
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:37ed957e35db52d753e90b2e89a572c6b011e6c971890858f14f12a4305efd1f
|
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 sha256:797085a729e10588f0af8dfcea7980f4fc8438c6de826417968959a62c5bdc9a
|
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": 275,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -3881,6 +3881,1939 @@
|
|
3881 |
"eval_samples_per_second": 9.013,
|
3882 |
"eval_steps_per_second": 4.509,
|
3883 |
"step": 550
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
3884 |
}
|
3885 |
],
|
3886 |
"logging_steps": 1,
|
@@ -3900,7 +5833,7 @@
|
|
3900 |
"attributes": {}
|
3901 |
}
|
3902 |
},
|
3903 |
-
"total_flos": 1.
|
3904 |
"train_batch_size": 2,
|
3905 |
"trial_name": null,
|
3906 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.7509386733416771,
|
5 |
"eval_steps": 275,
|
6 |
+
"global_step": 825,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
3881 |
"eval_samples_per_second": 9.013,
|
3882 |
"eval_steps_per_second": 4.509,
|
3883 |
"step": 550
|
3884 |
+
},
|
3885 |
+
{
|
3886 |
+
"epoch": 0.5015360109227444,
|
3887 |
+
"grad_norm": 0.4380887746810913,
|
3888 |
+
"learning_rate": 0.00010100967745256766,
|
3889 |
+
"loss": 2.4596,
|
3890 |
+
"step": 551
|
3891 |
+
},
|
3892 |
+
{
|
3893 |
+
"epoch": 0.502446239617704,
|
3894 |
+
"grad_norm": 0.4522015154361725,
|
3895 |
+
"learning_rate": 0.00010072120418249745,
|
3896 |
+
"loss": 2.3217,
|
3897 |
+
"step": 552
|
3898 |
+
},
|
3899 |
+
{
|
3900 |
+
"epoch": 0.5033564683126636,
|
3901 |
+
"grad_norm": 0.4581010043621063,
|
3902 |
+
"learning_rate": 0.00010043272491034523,
|
3903 |
+
"loss": 2.4948,
|
3904 |
+
"step": 553
|
3905 |
+
},
|
3906 |
+
{
|
3907 |
+
"epoch": 0.5042666970076232,
|
3908 |
+
"grad_norm": 0.4353744685649872,
|
3909 |
+
"learning_rate": 0.00010014424203692388,
|
3910 |
+
"loss": 2.3769,
|
3911 |
+
"step": 554
|
3912 |
+
},
|
3913 |
+
{
|
3914 |
+
"epoch": 0.5051769257025828,
|
3915 |
+
"grad_norm": 0.4150161147117615,
|
3916 |
+
"learning_rate": 9.985575796307615e-05,
|
3917 |
+
"loss": 2.3557,
|
3918 |
+
"step": 555
|
3919 |
+
},
|
3920 |
+
{
|
3921 |
+
"epoch": 0.5060871543975424,
|
3922 |
+
"grad_norm": 0.4532707631587982,
|
3923 |
+
"learning_rate": 9.956727508965481e-05,
|
3924 |
+
"loss": 2.3114,
|
3925 |
+
"step": 556
|
3926 |
+
},
|
3927 |
+
{
|
3928 |
+
"epoch": 0.506997383092502,
|
3929 |
+
"grad_norm": 0.42450255155563354,
|
3930 |
+
"learning_rate": 9.927879581750259e-05,
|
3931 |
+
"loss": 2.2911,
|
3932 |
+
"step": 557
|
3933 |
+
},
|
3934 |
+
{
|
3935 |
+
"epoch": 0.5079076117874616,
|
3936 |
+
"grad_norm": 0.42910221219062805,
|
3937 |
+
"learning_rate": 9.899032254743235e-05,
|
3938 |
+
"loss": 2.3062,
|
3939 |
+
"step": 558
|
3940 |
+
},
|
3941 |
+
{
|
3942 |
+
"epoch": 0.5088178404824212,
|
3943 |
+
"grad_norm": 0.42122408747673035,
|
3944 |
+
"learning_rate": 9.870185768020693e-05,
|
3945 |
+
"loss": 2.3294,
|
3946 |
+
"step": 559
|
3947 |
+
},
|
3948 |
+
{
|
3949 |
+
"epoch": 0.5097280691773808,
|
3950 |
+
"grad_norm": 0.4562203884124756,
|
3951 |
+
"learning_rate": 9.84134036165192e-05,
|
3952 |
+
"loss": 2.2674,
|
3953 |
+
"step": 560
|
3954 |
+
},
|
3955 |
+
{
|
3956 |
+
"epoch": 0.5106382978723404,
|
3957 |
+
"grad_norm": 0.3905525207519531,
|
3958 |
+
"learning_rate": 9.812496275697226e-05,
|
3959 |
+
"loss": 2.1259,
|
3960 |
+
"step": 561
|
3961 |
+
},
|
3962 |
+
{
|
3963 |
+
"epoch": 0.5115485265673,
|
3964 |
+
"grad_norm": 0.41641199588775635,
|
3965 |
+
"learning_rate": 9.783653750205915e-05,
|
3966 |
+
"loss": 2.2191,
|
3967 |
+
"step": 562
|
3968 |
+
},
|
3969 |
+
{
|
3970 |
+
"epoch": 0.5124587552622596,
|
3971 |
+
"grad_norm": 0.4090450704097748,
|
3972 |
+
"learning_rate": 9.754813025214317e-05,
|
3973 |
+
"loss": 2.2478,
|
3974 |
+
"step": 563
|
3975 |
+
},
|
3976 |
+
{
|
3977 |
+
"epoch": 0.5133689839572193,
|
3978 |
+
"grad_norm": 0.4293293356895447,
|
3979 |
+
"learning_rate": 9.725974340743769e-05,
|
3980 |
+
"loss": 2.3854,
|
3981 |
+
"step": 564
|
3982 |
+
},
|
3983 |
+
{
|
3984 |
+
"epoch": 0.5142792126521789,
|
3985 |
+
"grad_norm": 0.4138126075267792,
|
3986 |
+
"learning_rate": 9.697137936798634e-05,
|
3987 |
+
"loss": 2.2903,
|
3988 |
+
"step": 565
|
3989 |
+
},
|
3990 |
+
{
|
3991 |
+
"epoch": 0.5151894413471385,
|
3992 |
+
"grad_norm": 0.3979492783546448,
|
3993 |
+
"learning_rate": 9.668304053364294e-05,
|
3994 |
+
"loss": 2.1878,
|
3995 |
+
"step": 566
|
3996 |
+
},
|
3997 |
+
{
|
3998 |
+
"epoch": 0.5160996700420981,
|
3999 |
+
"grad_norm": 0.38530561327934265,
|
4000 |
+
"learning_rate": 9.639472930405143e-05,
|
4001 |
+
"loss": 2.1464,
|
4002 |
+
"step": 567
|
4003 |
+
},
|
4004 |
+
{
|
4005 |
+
"epoch": 0.5170098987370577,
|
4006 |
+
"grad_norm": 0.4263143837451935,
|
4007 |
+
"learning_rate": 9.610644807862625e-05,
|
4008 |
+
"loss": 2.1856,
|
4009 |
+
"step": 568
|
4010 |
+
},
|
4011 |
+
{
|
4012 |
+
"epoch": 0.5179201274320173,
|
4013 |
+
"grad_norm": 0.41127926111221313,
|
4014 |
+
"learning_rate": 9.581819925653188e-05,
|
4015 |
+
"loss": 2.198,
|
4016 |
+
"step": 569
|
4017 |
+
},
|
4018 |
+
{
|
4019 |
+
"epoch": 0.5188303561269769,
|
4020 |
+
"grad_norm": 0.3917299509048462,
|
4021 |
+
"learning_rate": 9.552998523666326e-05,
|
4022 |
+
"loss": 2.1325,
|
4023 |
+
"step": 570
|
4024 |
+
},
|
4025 |
+
{
|
4026 |
+
"epoch": 0.5197405848219365,
|
4027 |
+
"grad_norm": 0.39180508255958557,
|
4028 |
+
"learning_rate": 9.524180841762577e-05,
|
4029 |
+
"loss": 2.0702,
|
4030 |
+
"step": 571
|
4031 |
+
},
|
4032 |
+
{
|
4033 |
+
"epoch": 0.5206508135168961,
|
4034 |
+
"grad_norm": 0.3829837739467621,
|
4035 |
+
"learning_rate": 9.495367119771503e-05,
|
4036 |
+
"loss": 1.8913,
|
4037 |
+
"step": 572
|
4038 |
+
},
|
4039 |
+
{
|
4040 |
+
"epoch": 0.5215610422118557,
|
4041 |
+
"grad_norm": 0.412706196308136,
|
4042 |
+
"learning_rate": 9.46655759748972e-05,
|
4043 |
+
"loss": 2.1221,
|
4044 |
+
"step": 573
|
4045 |
+
},
|
4046 |
+
{
|
4047 |
+
"epoch": 0.5224712709068153,
|
4048 |
+
"grad_norm": 0.39748141169548035,
|
4049 |
+
"learning_rate": 9.437752514678887e-05,
|
4050 |
+
"loss": 2.0427,
|
4051 |
+
"step": 574
|
4052 |
+
},
|
4053 |
+
{
|
4054 |
+
"epoch": 0.5233814996017749,
|
4055 |
+
"grad_norm": 0.42854538559913635,
|
4056 |
+
"learning_rate": 9.408952111063727e-05,
|
4057 |
+
"loss": 2.121,
|
4058 |
+
"step": 575
|
4059 |
+
},
|
4060 |
+
{
|
4061 |
+
"epoch": 0.5242917282967345,
|
4062 |
+
"grad_norm": 0.414654016494751,
|
4063 |
+
"learning_rate": 9.380156626330009e-05,
|
4064 |
+
"loss": 2.038,
|
4065 |
+
"step": 576
|
4066 |
+
},
|
4067 |
+
{
|
4068 |
+
"epoch": 0.5252019569916941,
|
4069 |
+
"grad_norm": 0.4241427183151245,
|
4070 |
+
"learning_rate": 9.35136630012257e-05,
|
4071 |
+
"loss": 2.1312,
|
4072 |
+
"step": 577
|
4073 |
+
},
|
4074 |
+
{
|
4075 |
+
"epoch": 0.5261121856866537,
|
4076 |
+
"grad_norm": 0.42928779125213623,
|
4077 |
+
"learning_rate": 9.322581372043321e-05,
|
4078 |
+
"loss": 2.1875,
|
4079 |
+
"step": 578
|
4080 |
+
},
|
4081 |
+
{
|
4082 |
+
"epoch": 0.5270224143816133,
|
4083 |
+
"grad_norm": 0.4133308231830597,
|
4084 |
+
"learning_rate": 9.293802081649243e-05,
|
4085 |
+
"loss": 2.0477,
|
4086 |
+
"step": 579
|
4087 |
+
},
|
4088 |
+
{
|
4089 |
+
"epoch": 0.527932643076573,
|
4090 |
+
"grad_norm": 0.427898645401001,
|
4091 |
+
"learning_rate": 9.265028668450402e-05,
|
4092 |
+
"loss": 2.0833,
|
4093 |
+
"step": 580
|
4094 |
+
},
|
4095 |
+
{
|
4096 |
+
"epoch": 0.5288428717715326,
|
4097 |
+
"grad_norm": 0.4321751892566681,
|
4098 |
+
"learning_rate": 9.23626137190794e-05,
|
4099 |
+
"loss": 2.1698,
|
4100 |
+
"step": 581
|
4101 |
+
},
|
4102 |
+
{
|
4103 |
+
"epoch": 0.5297531004664922,
|
4104 |
+
"grad_norm": 0.4715782105922699,
|
4105 |
+
"learning_rate": 9.207500431432115e-05,
|
4106 |
+
"loss": 2.1347,
|
4107 |
+
"step": 582
|
4108 |
+
},
|
4109 |
+
{
|
4110 |
+
"epoch": 0.5306633291614519,
|
4111 |
+
"grad_norm": 0.45599547028541565,
|
4112 |
+
"learning_rate": 9.178746086380275e-05,
|
4113 |
+
"loss": 2.1469,
|
4114 |
+
"step": 583
|
4115 |
+
},
|
4116 |
+
{
|
4117 |
+
"epoch": 0.5315735578564115,
|
4118 |
+
"grad_norm": 0.45286545157432556,
|
4119 |
+
"learning_rate": 9.149998576054874e-05,
|
4120 |
+
"loss": 2.2013,
|
4121 |
+
"step": 584
|
4122 |
+
},
|
4123 |
+
{
|
4124 |
+
"epoch": 0.5324837865513711,
|
4125 |
+
"grad_norm": 0.47089457511901855,
|
4126 |
+
"learning_rate": 9.121258139701502e-05,
|
4127 |
+
"loss": 2.2125,
|
4128 |
+
"step": 585
|
4129 |
+
},
|
4130 |
+
{
|
4131 |
+
"epoch": 0.5333940152463307,
|
4132 |
+
"grad_norm": 0.46750229597091675,
|
4133 |
+
"learning_rate": 9.092525016506858e-05,
|
4134 |
+
"loss": 2.1186,
|
4135 |
+
"step": 586
|
4136 |
+
},
|
4137 |
+
{
|
4138 |
+
"epoch": 0.5343042439412903,
|
4139 |
+
"grad_norm": 0.4931905269622803,
|
4140 |
+
"learning_rate": 9.063799445596795e-05,
|
4141 |
+
"loss": 2.2185,
|
4142 |
+
"step": 587
|
4143 |
+
},
|
4144 |
+
{
|
4145 |
+
"epoch": 0.5352144726362499,
|
4146 |
+
"grad_norm": 0.48538026213645935,
|
4147 |
+
"learning_rate": 9.035081666034304e-05,
|
4148 |
+
"loss": 2.2369,
|
4149 |
+
"step": 588
|
4150 |
+
},
|
4151 |
+
{
|
4152 |
+
"epoch": 0.5361247013312095,
|
4153 |
+
"grad_norm": 0.4944066107273102,
|
4154 |
+
"learning_rate": 9.006371916817534e-05,
|
4155 |
+
"loss": 2.2771,
|
4156 |
+
"step": 589
|
4157 |
+
},
|
4158 |
+
{
|
4159 |
+
"epoch": 0.5370349300261691,
|
4160 |
+
"grad_norm": 0.4564894139766693,
|
4161 |
+
"learning_rate": 8.977670436877811e-05,
|
4162 |
+
"loss": 2.0879,
|
4163 |
+
"step": 590
|
4164 |
+
},
|
4165 |
+
{
|
4166 |
+
"epoch": 0.5379451587211287,
|
4167 |
+
"grad_norm": 0.5046347379684448,
|
4168 |
+
"learning_rate": 8.948977465077632e-05,
|
4169 |
+
"loss": 2.2197,
|
4170 |
+
"step": 591
|
4171 |
+
},
|
4172 |
+
{
|
4173 |
+
"epoch": 0.5388553874160883,
|
4174 |
+
"grad_norm": 0.49683472514152527,
|
4175 |
+
"learning_rate": 8.920293240208694e-05,
|
4176 |
+
"loss": 2.2152,
|
4177 |
+
"step": 592
|
4178 |
+
},
|
4179 |
+
{
|
4180 |
+
"epoch": 0.5397656161110479,
|
4181 |
+
"grad_norm": 0.5223331451416016,
|
4182 |
+
"learning_rate": 8.891618000989891e-05,
|
4183 |
+
"loss": 2.3358,
|
4184 |
+
"step": 593
|
4185 |
+
},
|
4186 |
+
{
|
4187 |
+
"epoch": 0.5406758448060075,
|
4188 |
+
"grad_norm": 0.5552563667297363,
|
4189 |
+
"learning_rate": 8.862951986065345e-05,
|
4190 |
+
"loss": 2.1608,
|
4191 |
+
"step": 594
|
4192 |
+
},
|
4193 |
+
{
|
4194 |
+
"epoch": 0.5415860735009671,
|
4195 |
+
"grad_norm": 0.5853347778320312,
|
4196 |
+
"learning_rate": 8.83429543400241e-05,
|
4197 |
+
"loss": 2.3679,
|
4198 |
+
"step": 595
|
4199 |
+
},
|
4200 |
+
{
|
4201 |
+
"epoch": 0.5424963021959267,
|
4202 |
+
"grad_norm": 0.5858141183853149,
|
4203 |
+
"learning_rate": 8.805648583289674e-05,
|
4204 |
+
"loss": 2.341,
|
4205 |
+
"step": 596
|
4206 |
+
},
|
4207 |
+
{
|
4208 |
+
"epoch": 0.5434065308908863,
|
4209 |
+
"grad_norm": 0.6405509114265442,
|
4210 |
+
"learning_rate": 8.777011672335008e-05,
|
4211 |
+
"loss": 2.4773,
|
4212 |
+
"step": 597
|
4213 |
+
},
|
4214 |
+
{
|
4215 |
+
"epoch": 0.5443167595858459,
|
4216 |
+
"grad_norm": 0.7342801094055176,
|
4217 |
+
"learning_rate": 8.748384939463543e-05,
|
4218 |
+
"loss": 2.557,
|
4219 |
+
"step": 598
|
4220 |
+
},
|
4221 |
+
{
|
4222 |
+
"epoch": 0.5452269882808055,
|
4223 |
+
"grad_norm": 0.8813995122909546,
|
4224 |
+
"learning_rate": 8.719768622915714e-05,
|
4225 |
+
"loss": 2.5595,
|
4226 |
+
"step": 599
|
4227 |
+
},
|
4228 |
+
{
|
4229 |
+
"epoch": 0.5461372169757651,
|
4230 |
+
"grad_norm": 1.722114086151123,
|
4231 |
+
"learning_rate": 8.691162960845264e-05,
|
4232 |
+
"loss": 2.7211,
|
4233 |
+
"step": 600
|
4234 |
+
},
|
4235 |
+
{
|
4236 |
+
"epoch": 0.5470474456707247,
|
4237 |
+
"grad_norm": 0.424265056848526,
|
4238 |
+
"learning_rate": 8.662568191317273e-05,
|
4239 |
+
"loss": 2.3728,
|
4240 |
+
"step": 601
|
4241 |
+
},
|
4242 |
+
{
|
4243 |
+
"epoch": 0.5479576743656844,
|
4244 |
+
"grad_norm": 0.45933809876441956,
|
4245 |
+
"learning_rate": 8.633984552306164e-05,
|
4246 |
+
"loss": 2.4234,
|
4247 |
+
"step": 602
|
4248 |
+
},
|
4249 |
+
{
|
4250 |
+
"epoch": 0.548867903060644,
|
4251 |
+
"grad_norm": 0.45455530285835266,
|
4252 |
+
"learning_rate": 8.605412281693727e-05,
|
4253 |
+
"loss": 2.5062,
|
4254 |
+
"step": 603
|
4255 |
+
},
|
4256 |
+
{
|
4257 |
+
"epoch": 0.5497781317556036,
|
4258 |
+
"grad_norm": 0.4334143400192261,
|
4259 |
+
"learning_rate": 8.57685161726715e-05,
|
4260 |
+
"loss": 2.3087,
|
4261 |
+
"step": 604
|
4262 |
+
},
|
4263 |
+
{
|
4264 |
+
"epoch": 0.5506883604505632,
|
4265 |
+
"grad_norm": 0.4537433385848999,
|
4266 |
+
"learning_rate": 8.548302796717019e-05,
|
4267 |
+
"loss": 2.395,
|
4268 |
+
"step": 605
|
4269 |
+
},
|
4270 |
+
{
|
4271 |
+
"epoch": 0.5515985891455228,
|
4272 |
+
"grad_norm": 0.43673837184906006,
|
4273 |
+
"learning_rate": 8.519766057635355e-05,
|
4274 |
+
"loss": 2.3855,
|
4275 |
+
"step": 606
|
4276 |
+
},
|
4277 |
+
{
|
4278 |
+
"epoch": 0.5525088178404824,
|
4279 |
+
"grad_norm": 0.43078145384788513,
|
4280 |
+
"learning_rate": 8.491241637513644e-05,
|
4281 |
+
"loss": 2.2559,
|
4282 |
+
"step": 607
|
4283 |
+
},
|
4284 |
+
{
|
4285 |
+
"epoch": 0.553419046535442,
|
4286 |
+
"grad_norm": 0.4094640612602234,
|
4287 |
+
"learning_rate": 8.462729773740832e-05,
|
4288 |
+
"loss": 2.295,
|
4289 |
+
"step": 608
|
4290 |
+
},
|
4291 |
+
{
|
4292 |
+
"epoch": 0.5543292752304017,
|
4293 |
+
"grad_norm": 0.4126126170158386,
|
4294 |
+
"learning_rate": 8.434230703601384e-05,
|
4295 |
+
"loss": 2.2019,
|
4296 |
+
"step": 609
|
4297 |
+
},
|
4298 |
+
{
|
4299 |
+
"epoch": 0.5552395039253613,
|
4300 |
+
"grad_norm": 0.4372231066226959,
|
4301 |
+
"learning_rate": 8.405744664273278e-05,
|
4302 |
+
"loss": 2.4243,
|
4303 |
+
"step": 610
|
4304 |
+
},
|
4305 |
+
{
|
4306 |
+
"epoch": 0.5561497326203209,
|
4307 |
+
"grad_norm": 0.42160138487815857,
|
4308 |
+
"learning_rate": 8.37727189282606e-05,
|
4309 |
+
"loss": 2.2805,
|
4310 |
+
"step": 611
|
4311 |
+
},
|
4312 |
+
{
|
4313 |
+
"epoch": 0.5570599613152805,
|
4314 |
+
"grad_norm": 0.4336857795715332,
|
4315 |
+
"learning_rate": 8.34881262621884e-05,
|
4316 |
+
"loss": 2.4811,
|
4317 |
+
"step": 612
|
4318 |
+
},
|
4319 |
+
{
|
4320 |
+
"epoch": 0.5579701900102401,
|
4321 |
+
"grad_norm": 0.40520837903022766,
|
4322 |
+
"learning_rate": 8.320367101298351e-05,
|
4323 |
+
"loss": 2.1731,
|
4324 |
+
"step": 613
|
4325 |
+
},
|
4326 |
+
{
|
4327 |
+
"epoch": 0.5588804187051997,
|
4328 |
+
"grad_norm": 0.42664197087287903,
|
4329 |
+
"learning_rate": 8.291935554796962e-05,
|
4330 |
+
"loss": 2.3403,
|
4331 |
+
"step": 614
|
4332 |
+
},
|
4333 |
+
{
|
4334 |
+
"epoch": 0.5597906474001593,
|
4335 |
+
"grad_norm": 0.4109039902687073,
|
4336 |
+
"learning_rate": 8.263518223330697e-05,
|
4337 |
+
"loss": 2.2425,
|
4338 |
+
"step": 615
|
4339 |
+
},
|
4340 |
+
{
|
4341 |
+
"epoch": 0.5607008760951189,
|
4342 |
+
"grad_norm": 0.4032575786113739,
|
4343 |
+
"learning_rate": 8.235115343397295e-05,
|
4344 |
+
"loss": 2.2593,
|
4345 |
+
"step": 616
|
4346 |
+
},
|
4347 |
+
{
|
4348 |
+
"epoch": 0.5616111047900785,
|
4349 |
+
"grad_norm": 0.3929396867752075,
|
4350 |
+
"learning_rate": 8.206727151374207e-05,
|
4351 |
+
"loss": 2.0896,
|
4352 |
+
"step": 617
|
4353 |
+
},
|
4354 |
+
{
|
4355 |
+
"epoch": 0.5625213334850381,
|
4356 |
+
"grad_norm": 0.38567835092544556,
|
4357 |
+
"learning_rate": 8.178353883516664e-05,
|
4358 |
+
"loss": 2.0715,
|
4359 |
+
"step": 618
|
4360 |
+
},
|
4361 |
+
{
|
4362 |
+
"epoch": 0.5634315621799977,
|
4363 |
+
"grad_norm": 0.405369371175766,
|
4364 |
+
"learning_rate": 8.149995775955686e-05,
|
4365 |
+
"loss": 2.2249,
|
4366 |
+
"step": 619
|
4367 |
+
},
|
4368 |
+
{
|
4369 |
+
"epoch": 0.5643417908749573,
|
4370 |
+
"grad_norm": 0.3889697790145874,
|
4371 |
+
"learning_rate": 8.121653064696118e-05,
|
4372 |
+
"loss": 2.0797,
|
4373 |
+
"step": 620
|
4374 |
+
},
|
4375 |
+
{
|
4376 |
+
"epoch": 0.565252019569917,
|
4377 |
+
"grad_norm": 0.4065384864807129,
|
4378 |
+
"learning_rate": 8.093325985614685e-05,
|
4379 |
+
"loss": 2.2012,
|
4380 |
+
"step": 621
|
4381 |
+
},
|
4382 |
+
{
|
4383 |
+
"epoch": 0.5661622482648766,
|
4384 |
+
"grad_norm": 0.4066416323184967,
|
4385 |
+
"learning_rate": 8.065014774458003e-05,
|
4386 |
+
"loss": 2.1183,
|
4387 |
+
"step": 622
|
4388 |
+
},
|
4389 |
+
{
|
4390 |
+
"epoch": 0.5670724769598362,
|
4391 |
+
"grad_norm": 0.40575870871543884,
|
4392 |
+
"learning_rate": 8.036719666840647e-05,
|
4393 |
+
"loss": 2.0258,
|
4394 |
+
"step": 623
|
4395 |
+
},
|
4396 |
+
{
|
4397 |
+
"epoch": 0.5679827056547958,
|
4398 |
+
"grad_norm": 0.42911243438720703,
|
4399 |
+
"learning_rate": 8.008440898243149e-05,
|
4400 |
+
"loss": 2.1186,
|
4401 |
+
"step": 624
|
4402 |
+
},
|
4403 |
+
{
|
4404 |
+
"epoch": 0.5688929343497554,
|
4405 |
+
"grad_norm": 0.4009549021720886,
|
4406 |
+
"learning_rate": 7.980178704010089e-05,
|
4407 |
+
"loss": 2.0049,
|
4408 |
+
"step": 625
|
4409 |
+
},
|
4410 |
+
{
|
4411 |
+
"epoch": 0.569803163044715,
|
4412 |
+
"grad_norm": 0.41156989336013794,
|
4413 |
+
"learning_rate": 7.951933319348095e-05,
|
4414 |
+
"loss": 2.0272,
|
4415 |
+
"step": 626
|
4416 |
+
},
|
4417 |
+
{
|
4418 |
+
"epoch": 0.5707133917396746,
|
4419 |
+
"grad_norm": 0.4248954653739929,
|
4420 |
+
"learning_rate": 7.923704979323899e-05,
|
4421 |
+
"loss": 2.077,
|
4422 |
+
"step": 627
|
4423 |
+
},
|
4424 |
+
{
|
4425 |
+
"epoch": 0.5716236204346342,
|
4426 |
+
"grad_norm": 0.45484524965286255,
|
4427 |
+
"learning_rate": 7.895493918862396e-05,
|
4428 |
+
"loss": 2.2255,
|
4429 |
+
"step": 628
|
4430 |
+
},
|
4431 |
+
{
|
4432 |
+
"epoch": 0.5725338491295938,
|
4433 |
+
"grad_norm": 0.4571921229362488,
|
4434 |
+
"learning_rate": 7.867300372744657e-05,
|
4435 |
+
"loss": 2.1373,
|
4436 |
+
"step": 629
|
4437 |
+
},
|
4438 |
+
{
|
4439 |
+
"epoch": 0.5734440778245534,
|
4440 |
+
"grad_norm": 0.44238901138305664,
|
4441 |
+
"learning_rate": 7.839124575606004e-05,
|
4442 |
+
"loss": 2.1147,
|
4443 |
+
"step": 630
|
4444 |
+
},
|
4445 |
+
{
|
4446 |
+
"epoch": 0.574354306519513,
|
4447 |
+
"grad_norm": 0.4206310510635376,
|
4448 |
+
"learning_rate": 7.810966761934053e-05,
|
4449 |
+
"loss": 2.0508,
|
4450 |
+
"step": 631
|
4451 |
+
},
|
4452 |
+
{
|
4453 |
+
"epoch": 0.5752645352144726,
|
4454 |
+
"grad_norm": 0.43381330370903015,
|
4455 |
+
"learning_rate": 7.782827166066739e-05,
|
4456 |
+
"loss": 2.0847,
|
4457 |
+
"step": 632
|
4458 |
+
},
|
4459 |
+
{
|
4460 |
+
"epoch": 0.5761747639094322,
|
4461 |
+
"grad_norm": 0.4460139572620392,
|
4462 |
+
"learning_rate": 7.754706022190398e-05,
|
4463 |
+
"loss": 2.1288,
|
4464 |
+
"step": 633
|
4465 |
+
},
|
4466 |
+
{
|
4467 |
+
"epoch": 0.5770849926043918,
|
4468 |
+
"grad_norm": 0.4371720850467682,
|
4469 |
+
"learning_rate": 7.726603564337791e-05,
|
4470 |
+
"loss": 2.0476,
|
4471 |
+
"step": 634
|
4472 |
+
},
|
4473 |
+
{
|
4474 |
+
"epoch": 0.5779952212993514,
|
4475 |
+
"grad_norm": 0.4623599052429199,
|
4476 |
+
"learning_rate": 7.69852002638618e-05,
|
4477 |
+
"loss": 2.2858,
|
4478 |
+
"step": 635
|
4479 |
+
},
|
4480 |
+
{
|
4481 |
+
"epoch": 0.578905449994311,
|
4482 |
+
"grad_norm": 0.4422992765903473,
|
4483 |
+
"learning_rate": 7.670455642055361e-05,
|
4484 |
+
"loss": 2.1072,
|
4485 |
+
"step": 636
|
4486 |
+
},
|
4487 |
+
{
|
4488 |
+
"epoch": 0.5798156786892706,
|
4489 |
+
"grad_norm": 0.4804936647415161,
|
4490 |
+
"learning_rate": 7.642410644905726e-05,
|
4491 |
+
"loss": 2.2218,
|
4492 |
+
"step": 637
|
4493 |
+
},
|
4494 |
+
{
|
4495 |
+
"epoch": 0.5807259073842302,
|
4496 |
+
"grad_norm": 0.48249900341033936,
|
4497 |
+
"learning_rate": 7.614385268336336e-05,
|
4498 |
+
"loss": 2.2916,
|
4499 |
+
"step": 638
|
4500 |
+
},
|
4501 |
+
{
|
4502 |
+
"epoch": 0.5816361360791898,
|
4503 |
+
"grad_norm": 0.46635982394218445,
|
4504 |
+
"learning_rate": 7.586379745582944e-05,
|
4505 |
+
"loss": 2.1636,
|
4506 |
+
"step": 639
|
4507 |
+
},
|
4508 |
+
{
|
4509 |
+
"epoch": 0.5825463647741494,
|
4510 |
+
"grad_norm": 0.4670505225658417,
|
4511 |
+
"learning_rate": 7.558394309716088e-05,
|
4512 |
+
"loss": 2.2052,
|
4513 |
+
"step": 640
|
4514 |
+
},
|
4515 |
+
{
|
4516 |
+
"epoch": 0.5834565934691092,
|
4517 |
+
"grad_norm": 0.49475541710853577,
|
4518 |
+
"learning_rate": 7.530429193639128e-05,
|
4519 |
+
"loss": 2.18,
|
4520 |
+
"step": 641
|
4521 |
+
},
|
4522 |
+
{
|
4523 |
+
"epoch": 0.5843668221640688,
|
4524 |
+
"grad_norm": 0.5231596231460571,
|
4525 |
+
"learning_rate": 7.502484630086318e-05,
|
4526 |
+
"loss": 2.2095,
|
4527 |
+
"step": 642
|
4528 |
+
},
|
4529 |
+
{
|
4530 |
+
"epoch": 0.5852770508590284,
|
4531 |
+
"grad_norm": 0.5045900344848633,
|
4532 |
+
"learning_rate": 7.474560851620873e-05,
|
4533 |
+
"loss": 2.053,
|
4534 |
+
"step": 643
|
4535 |
+
},
|
4536 |
+
{
|
4537 |
+
"epoch": 0.586187279553988,
|
4538 |
+
"grad_norm": 0.5511046051979065,
|
4539 |
+
"learning_rate": 7.446658090633026e-05,
|
4540 |
+
"loss": 2.2706,
|
4541 |
+
"step": 644
|
4542 |
+
},
|
4543 |
+
{
|
4544 |
+
"epoch": 0.5870975082489476,
|
4545 |
+
"grad_norm": 0.5700446963310242,
|
4546 |
+
"learning_rate": 7.41877657933809e-05,
|
4547 |
+
"loss": 2.3999,
|
4548 |
+
"step": 645
|
4549 |
+
},
|
4550 |
+
{
|
4551 |
+
"epoch": 0.5880077369439072,
|
4552 |
+
"grad_norm": 0.5792605876922607,
|
4553 |
+
"learning_rate": 7.390916549774536e-05,
|
4554 |
+
"loss": 2.2391,
|
4555 |
+
"step": 646
|
4556 |
+
},
|
4557 |
+
{
|
4558 |
+
"epoch": 0.5889179656388668,
|
4559 |
+
"grad_norm": 0.6770455241203308,
|
4560 |
+
"learning_rate": 7.363078233802063e-05,
|
4561 |
+
"loss": 2.6564,
|
4562 |
+
"step": 647
|
4563 |
+
},
|
4564 |
+
{
|
4565 |
+
"epoch": 0.5898281943338264,
|
4566 |
+
"grad_norm": 0.7092955708503723,
|
4567 |
+
"learning_rate": 7.335261863099651e-05,
|
4568 |
+
"loss": 2.4722,
|
4569 |
+
"step": 648
|
4570 |
+
},
|
4571 |
+
{
|
4572 |
+
"epoch": 0.590738423028786,
|
4573 |
+
"grad_norm": 0.8125056028366089,
|
4574 |
+
"learning_rate": 7.307467669163655e-05,
|
4575 |
+
"loss": 2.3581,
|
4576 |
+
"step": 649
|
4577 |
+
},
|
4578 |
+
{
|
4579 |
+
"epoch": 0.5916486517237456,
|
4580 |
+
"grad_norm": 1.5941253900527954,
|
4581 |
+
"learning_rate": 7.279695883305866e-05,
|
4582 |
+
"loss": 2.16,
|
4583 |
+
"step": 650
|
4584 |
+
},
|
4585 |
+
{
|
4586 |
+
"epoch": 0.5925588804187052,
|
4587 |
+
"grad_norm": 0.44398507475852966,
|
4588 |
+
"learning_rate": 7.251946736651582e-05,
|
4589 |
+
"loss": 2.4689,
|
4590 |
+
"step": 651
|
4591 |
+
},
|
4592 |
+
{
|
4593 |
+
"epoch": 0.5934691091136648,
|
4594 |
+
"grad_norm": 0.4264509975910187,
|
4595 |
+
"learning_rate": 7.224220460137701e-05,
|
4596 |
+
"loss": 2.4081,
|
4597 |
+
"step": 652
|
4598 |
+
},
|
4599 |
+
{
|
4600 |
+
"epoch": 0.5943793378086244,
|
4601 |
+
"grad_norm": 0.4222484529018402,
|
4602 |
+
"learning_rate": 7.196517284510773e-05,
|
4603 |
+
"loss": 2.3827,
|
4604 |
+
"step": 653
|
4605 |
+
},
|
4606 |
+
{
|
4607 |
+
"epoch": 0.595289566503584,
|
4608 |
+
"grad_norm": 0.4516051411628723,
|
4609 |
+
"learning_rate": 7.168837440325114e-05,
|
4610 |
+
"loss": 2.399,
|
4611 |
+
"step": 654
|
4612 |
+
},
|
4613 |
+
{
|
4614 |
+
"epoch": 0.5961997951985436,
|
4615 |
+
"grad_norm": 0.4370306730270386,
|
4616 |
+
"learning_rate": 7.141181157940859e-05,
|
4617 |
+
"loss": 2.3837,
|
4618 |
+
"step": 655
|
4619 |
+
},
|
4620 |
+
{
|
4621 |
+
"epoch": 0.5971100238935032,
|
4622 |
+
"grad_norm": 0.4236253798007965,
|
4623 |
+
"learning_rate": 7.11354866752205e-05,
|
4624 |
+
"loss": 2.3066,
|
4625 |
+
"step": 656
|
4626 |
+
},
|
4627 |
+
{
|
4628 |
+
"epoch": 0.5980202525884628,
|
4629 |
+
"grad_norm": 0.41718846559524536,
|
4630 |
+
"learning_rate": 7.085940199034735e-05,
|
4631 |
+
"loss": 2.3841,
|
4632 |
+
"step": 657
|
4633 |
+
},
|
4634 |
+
{
|
4635 |
+
"epoch": 0.5989304812834224,
|
4636 |
+
"grad_norm": 0.4299750030040741,
|
4637 |
+
"learning_rate": 7.058355982245037e-05,
|
4638 |
+
"loss": 2.3842,
|
4639 |
+
"step": 658
|
4640 |
+
},
|
4641 |
+
{
|
4642 |
+
"epoch": 0.599840709978382,
|
4643 |
+
"grad_norm": 0.4180915057659149,
|
4644 |
+
"learning_rate": 7.030796246717255e-05,
|
4645 |
+
"loss": 2.0758,
|
4646 |
+
"step": 659
|
4647 |
+
},
|
4648 |
+
{
|
4649 |
+
"epoch": 0.6007509386733417,
|
4650 |
+
"grad_norm": 0.45195114612579346,
|
4651 |
+
"learning_rate": 7.003261221811934e-05,
|
4652 |
+
"loss": 2.4826,
|
4653 |
+
"step": 660
|
4654 |
+
},
|
4655 |
+
{
|
4656 |
+
"epoch": 0.6016611673683013,
|
4657 |
+
"grad_norm": 0.4253404140472412,
|
4658 |
+
"learning_rate": 6.97575113668399e-05,
|
4659 |
+
"loss": 2.3705,
|
4660 |
+
"step": 661
|
4661 |
+
},
|
4662 |
+
{
|
4663 |
+
"epoch": 0.6025713960632609,
|
4664 |
+
"grad_norm": 0.4198931157588959,
|
4665 |
+
"learning_rate": 6.948266220280771e-05,
|
4666 |
+
"loss": 2.3396,
|
4667 |
+
"step": 662
|
4668 |
+
},
|
4669 |
+
{
|
4670 |
+
"epoch": 0.6034816247582205,
|
4671 |
+
"grad_norm": 0.43457460403442383,
|
4672 |
+
"learning_rate": 6.920806701340155e-05,
|
4673 |
+
"loss": 2.1447,
|
4674 |
+
"step": 663
|
4675 |
+
},
|
4676 |
+
{
|
4677 |
+
"epoch": 0.6043918534531801,
|
4678 |
+
"grad_norm": 0.40161874890327454,
|
4679 |
+
"learning_rate": 6.893372808388675e-05,
|
4680 |
+
"loss": 2.2443,
|
4681 |
+
"step": 664
|
4682 |
+
},
|
4683 |
+
{
|
4684 |
+
"epoch": 0.6053020821481397,
|
4685 |
+
"grad_norm": 0.4039609432220459,
|
4686 |
+
"learning_rate": 6.865964769739575e-05,
|
4687 |
+
"loss": 2.1815,
|
4688 |
+
"step": 665
|
4689 |
+
},
|
4690 |
+
{
|
4691 |
+
"epoch": 0.6062123108430993,
|
4692 |
+
"grad_norm": 0.4061351716518402,
|
4693 |
+
"learning_rate": 6.838582813490947e-05,
|
4694 |
+
"loss": 2.1073,
|
4695 |
+
"step": 666
|
4696 |
+
},
|
4697 |
+
{
|
4698 |
+
"epoch": 0.607122539538059,
|
4699 |
+
"grad_norm": 0.4206211268901825,
|
4700 |
+
"learning_rate": 6.811227167523815e-05,
|
4701 |
+
"loss": 2.2549,
|
4702 |
+
"step": 667
|
4703 |
+
},
|
4704 |
+
{
|
4705 |
+
"epoch": 0.6080327682330186,
|
4706 |
+
"grad_norm": 0.3936857283115387,
|
4707 |
+
"learning_rate": 6.783898059500233e-05,
|
4708 |
+
"loss": 2.1373,
|
4709 |
+
"step": 668
|
4710 |
+
},
|
4711 |
+
{
|
4712 |
+
"epoch": 0.6089429969279782,
|
4713 |
+
"grad_norm": 0.3954029083251953,
|
4714 |
+
"learning_rate": 6.756595716861407e-05,
|
4715 |
+
"loss": 2.1001,
|
4716 |
+
"step": 669
|
4717 |
+
},
|
4718 |
+
{
|
4719 |
+
"epoch": 0.6098532256229378,
|
4720 |
+
"grad_norm": 0.407713919878006,
|
4721 |
+
"learning_rate": 6.729320366825784e-05,
|
4722 |
+
"loss": 2.0967,
|
4723 |
+
"step": 670
|
4724 |
+
},
|
4725 |
+
{
|
4726 |
+
"epoch": 0.6107634543178974,
|
4727 |
+
"grad_norm": 0.41096213459968567,
|
4728 |
+
"learning_rate": 6.702072236387182e-05,
|
4729 |
+
"loss": 2.0899,
|
4730 |
+
"step": 671
|
4731 |
+
},
|
4732 |
+
{
|
4733 |
+
"epoch": 0.611673683012857,
|
4734 |
+
"grad_norm": 0.40465790033340454,
|
4735 |
+
"learning_rate": 6.674851552312878e-05,
|
4736 |
+
"loss": 2.089,
|
4737 |
+
"step": 672
|
4738 |
+
},
|
4739 |
+
{
|
4740 |
+
"epoch": 0.6125839117078166,
|
4741 |
+
"grad_norm": 0.3991434574127197,
|
4742 |
+
"learning_rate": 6.647658541141735e-05,
|
4743 |
+
"loss": 1.9788,
|
4744 |
+
"step": 673
|
4745 |
+
},
|
4746 |
+
{
|
4747 |
+
"epoch": 0.6134941404027762,
|
4748 |
+
"grad_norm": 0.42327383160591125,
|
4749 |
+
"learning_rate": 6.620493429182323e-05,
|
4750 |
+
"loss": 2.1672,
|
4751 |
+
"step": 674
|
4752 |
+
},
|
4753 |
+
{
|
4754 |
+
"epoch": 0.6144043690977358,
|
4755 |
+
"grad_norm": 0.4061299264431,
|
4756 |
+
"learning_rate": 6.593356442511015e-05,
|
4757 |
+
"loss": 2.1617,
|
4758 |
+
"step": 675
|
4759 |
+
},
|
4760 |
+
{
|
4761 |
+
"epoch": 0.6153145977926954,
|
4762 |
+
"grad_norm": 0.4082658588886261,
|
4763 |
+
"learning_rate": 6.566247806970119e-05,
|
4764 |
+
"loss": 2.0112,
|
4765 |
+
"step": 676
|
4766 |
+
},
|
4767 |
+
{
|
4768 |
+
"epoch": 0.616224826487655,
|
4769 |
+
"grad_norm": 0.43016669154167175,
|
4770 |
+
"learning_rate": 6.539167748165994e-05,
|
4771 |
+
"loss": 2.0,
|
4772 |
+
"step": 677
|
4773 |
+
},
|
4774 |
+
{
|
4775 |
+
"epoch": 0.6171350551826146,
|
4776 |
+
"grad_norm": 0.43142226338386536,
|
4777 |
+
"learning_rate": 6.512116491467185e-05,
|
4778 |
+
"loss": 2.1585,
|
4779 |
+
"step": 678
|
4780 |
+
},
|
4781 |
+
{
|
4782 |
+
"epoch": 0.6180452838775743,
|
4783 |
+
"grad_norm": 0.4271491467952728,
|
4784 |
+
"learning_rate": 6.485094262002529e-05,
|
4785 |
+
"loss": 1.9628,
|
4786 |
+
"step": 679
|
4787 |
+
},
|
4788 |
+
{
|
4789 |
+
"epoch": 0.6189555125725339,
|
4790 |
+
"grad_norm": 0.44002971053123474,
|
4791 |
+
"learning_rate": 6.458101284659286e-05,
|
4792 |
+
"loss": 2.2214,
|
4793 |
+
"step": 680
|
4794 |
+
},
|
4795 |
+
{
|
4796 |
+
"epoch": 0.6198657412674935,
|
4797 |
+
"grad_norm": 0.4215126931667328,
|
4798 |
+
"learning_rate": 6.431137784081282e-05,
|
4799 |
+
"loss": 2.0377,
|
4800 |
+
"step": 681
|
4801 |
+
},
|
4802 |
+
{
|
4803 |
+
"epoch": 0.6207759699624531,
|
4804 |
+
"grad_norm": 0.46792343258857727,
|
4805 |
+
"learning_rate": 6.404203984667019e-05,
|
4806 |
+
"loss": 2.029,
|
4807 |
+
"step": 682
|
4808 |
+
},
|
4809 |
+
{
|
4810 |
+
"epoch": 0.6216861986574127,
|
4811 |
+
"grad_norm": 0.45737308263778687,
|
4812 |
+
"learning_rate": 6.377300110567821e-05,
|
4813 |
+
"loss": 2.2375,
|
4814 |
+
"step": 683
|
4815 |
+
},
|
4816 |
+
{
|
4817 |
+
"epoch": 0.6225964273523723,
|
4818 |
+
"grad_norm": 0.4526033401489258,
|
4819 |
+
"learning_rate": 6.350426385685957e-05,
|
4820 |
+
"loss": 2.2562,
|
4821 |
+
"step": 684
|
4822 |
+
},
|
4823 |
+
{
|
4824 |
+
"epoch": 0.6235066560473319,
|
4825 |
+
"grad_norm": 0.45917776226997375,
|
4826 |
+
"learning_rate": 6.323583033672799e-05,
|
4827 |
+
"loss": 2.1321,
|
4828 |
+
"step": 685
|
4829 |
+
},
|
4830 |
+
{
|
4831 |
+
"epoch": 0.6244168847422915,
|
4832 |
+
"grad_norm": 0.4713301658630371,
|
4833 |
+
"learning_rate": 6.296770277926937e-05,
|
4834 |
+
"loss": 2.07,
|
4835 |
+
"step": 686
|
4836 |
+
},
|
4837 |
+
{
|
4838 |
+
"epoch": 0.6253271134372511,
|
4839 |
+
"grad_norm": 0.5036799907684326,
|
4840 |
+
"learning_rate": 6.269988341592328e-05,
|
4841 |
+
"loss": 2.1103,
|
4842 |
+
"step": 687
|
4843 |
+
},
|
4844 |
+
{
|
4845 |
+
"epoch": 0.6262373421322107,
|
4846 |
+
"grad_norm": 0.4843004643917084,
|
4847 |
+
"learning_rate": 6.243237447556449e-05,
|
4848 |
+
"loss": 2.0936,
|
4849 |
+
"step": 688
|
4850 |
+
},
|
4851 |
+
{
|
4852 |
+
"epoch": 0.6271475708271703,
|
4853 |
+
"grad_norm": 0.4738497734069824,
|
4854 |
+
"learning_rate": 6.216517818448423e-05,
|
4855 |
+
"loss": 2.1004,
|
4856 |
+
"step": 689
|
4857 |
+
},
|
4858 |
+
{
|
4859 |
+
"epoch": 0.6280577995221299,
|
4860 |
+
"grad_norm": 0.5081862211227417,
|
4861 |
+
"learning_rate": 6.189829676637182e-05,
|
4862 |
+
"loss": 2.2177,
|
4863 |
+
"step": 690
|
4864 |
+
},
|
4865 |
+
{
|
4866 |
+
"epoch": 0.6289680282170895,
|
4867 |
+
"grad_norm": 0.5060831904411316,
|
4868 |
+
"learning_rate": 6.163173244229619e-05,
|
4869 |
+
"loss": 2.1342,
|
4870 |
+
"step": 691
|
4871 |
+
},
|
4872 |
+
{
|
4873 |
+
"epoch": 0.6298782569120491,
|
4874 |
+
"grad_norm": 0.5047332644462585,
|
4875 |
+
"learning_rate": 6.136548743068713e-05,
|
4876 |
+
"loss": 2.0727,
|
4877 |
+
"step": 692
|
4878 |
+
},
|
4879 |
+
{
|
4880 |
+
"epoch": 0.6307884856070087,
|
4881 |
+
"grad_norm": 0.5174648761749268,
|
4882 |
+
"learning_rate": 6.109956394731722e-05,
|
4883 |
+
"loss": 2.0623,
|
4884 |
+
"step": 693
|
4885 |
+
},
|
4886 |
+
{
|
4887 |
+
"epoch": 0.6316987143019683,
|
4888 |
+
"grad_norm": 0.5705139636993408,
|
4889 |
+
"learning_rate": 6.083396420528298e-05,
|
4890 |
+
"loss": 2.4454,
|
4891 |
+
"step": 694
|
4892 |
+
},
|
4893 |
+
{
|
4894 |
+
"epoch": 0.6326089429969279,
|
4895 |
+
"grad_norm": 0.5653088092803955,
|
4896 |
+
"learning_rate": 6.056869041498687e-05,
|
4897 |
+
"loss": 2.2071,
|
4898 |
+
"step": 695
|
4899 |
+
},
|
4900 |
+
{
|
4901 |
+
"epoch": 0.6335191716918875,
|
4902 |
+
"grad_norm": 0.6105946898460388,
|
4903 |
+
"learning_rate": 6.030374478411847e-05,
|
4904 |
+
"loss": 2.3081,
|
4905 |
+
"step": 696
|
4906 |
+
},
|
4907 |
+
{
|
4908 |
+
"epoch": 0.6344294003868471,
|
4909 |
+
"grad_norm": 0.6362658143043518,
|
4910 |
+
"learning_rate": 6.0039129517636435e-05,
|
4911 |
+
"loss": 2.3426,
|
4912 |
+
"step": 697
|
4913 |
+
},
|
4914 |
+
{
|
4915 |
+
"epoch": 0.6353396290818069,
|
4916 |
+
"grad_norm": 0.7242766618728638,
|
4917 |
+
"learning_rate": 5.9774846817750105e-05,
|
4918 |
+
"loss": 2.4877,
|
4919 |
+
"step": 698
|
4920 |
+
},
|
4921 |
+
{
|
4922 |
+
"epoch": 0.6362498577767665,
|
4923 |
+
"grad_norm": 0.9446219205856323,
|
4924 |
+
"learning_rate": 5.951089888390087e-05,
|
4925 |
+
"loss": 2.7741,
|
4926 |
+
"step": 699
|
4927 |
+
},
|
4928 |
+
{
|
4929 |
+
"epoch": 0.6371600864717261,
|
4930 |
+
"grad_norm": 1.5826219320297241,
|
4931 |
+
"learning_rate": 5.924728791274432e-05,
|
4932 |
+
"loss": 2.533,
|
4933 |
+
"step": 700
|
4934 |
+
},
|
4935 |
+
{
|
4936 |
+
"epoch": 0.6380703151666857,
|
4937 |
+
"grad_norm": 0.4994019567966461,
|
4938 |
+
"learning_rate": 5.89840160981316e-05,
|
4939 |
+
"loss": 2.4388,
|
4940 |
+
"step": 701
|
4941 |
+
},
|
4942 |
+
{
|
4943 |
+
"epoch": 0.6389805438616453,
|
4944 |
+
"grad_norm": 0.45213326811790466,
|
4945 |
+
"learning_rate": 5.872108563109131e-05,
|
4946 |
+
"loss": 2.3644,
|
4947 |
+
"step": 702
|
4948 |
+
},
|
4949 |
+
{
|
4950 |
+
"epoch": 0.6398907725566049,
|
4951 |
+
"grad_norm": 0.45655617117881775,
|
4952 |
+
"learning_rate": 5.845849869981137e-05,
|
4953 |
+
"loss": 2.5202,
|
4954 |
+
"step": 703
|
4955 |
+
},
|
4956 |
+
{
|
4957 |
+
"epoch": 0.6408010012515645,
|
4958 |
+
"grad_norm": 0.41640329360961914,
|
4959 |
+
"learning_rate": 5.819625748962049e-05,
|
4960 |
+
"loss": 2.3097,
|
4961 |
+
"step": 704
|
4962 |
+
},
|
4963 |
+
{
|
4964 |
+
"epoch": 0.6417112299465241,
|
4965 |
+
"grad_norm": 0.43625307083129883,
|
4966 |
+
"learning_rate": 5.79343641829704e-05,
|
4967 |
+
"loss": 2.3931,
|
4968 |
+
"step": 705
|
4969 |
+
},
|
4970 |
+
{
|
4971 |
+
"epoch": 0.6426214586414837,
|
4972 |
+
"grad_norm": 0.44178506731987,
|
4973 |
+
"learning_rate": 5.7672820959417254e-05,
|
4974 |
+
"loss": 2.3195,
|
4975 |
+
"step": 706
|
4976 |
+
},
|
4977 |
+
{
|
4978 |
+
"epoch": 0.6435316873364433,
|
4979 |
+
"grad_norm": 0.4416089951992035,
|
4980 |
+
"learning_rate": 5.741162999560386e-05,
|
4981 |
+
"loss": 2.2446,
|
4982 |
+
"step": 707
|
4983 |
+
},
|
4984 |
+
{
|
4985 |
+
"epoch": 0.6444419160314029,
|
4986 |
+
"grad_norm": 0.4419204890727997,
|
4987 |
+
"learning_rate": 5.7150793465241346e-05,
|
4988 |
+
"loss": 2.34,
|
4989 |
+
"step": 708
|
4990 |
+
},
|
4991 |
+
{
|
4992 |
+
"epoch": 0.6453521447263625,
|
4993 |
+
"grad_norm": 0.45422717928886414,
|
4994 |
+
"learning_rate": 5.68903135390912e-05,
|
4995 |
+
"loss": 2.1915,
|
4996 |
+
"step": 709
|
4997 |
+
},
|
4998 |
+
{
|
4999 |
+
"epoch": 0.6462623734213221,
|
5000 |
+
"grad_norm": 0.41635921597480774,
|
5001 |
+
"learning_rate": 5.663019238494704e-05,
|
5002 |
+
"loss": 2.3147,
|
5003 |
+
"step": 710
|
5004 |
+
},
|
5005 |
+
{
|
5006 |
+
"epoch": 0.6471726021162817,
|
5007 |
+
"grad_norm": 0.4240402579307556,
|
5008 |
+
"learning_rate": 5.637043216761678e-05,
|
5009 |
+
"loss": 2.1693,
|
5010 |
+
"step": 711
|
5011 |
+
},
|
5012 |
+
{
|
5013 |
+
"epoch": 0.6480828308112413,
|
5014 |
+
"grad_norm": 0.41989627480506897,
|
5015 |
+
"learning_rate": 5.611103504890444e-05,
|
5016 |
+
"loss": 2.2087,
|
5017 |
+
"step": 712
|
5018 |
+
},
|
5019 |
+
{
|
5020 |
+
"epoch": 0.6489930595062009,
|
5021 |
+
"grad_norm": 0.4187220335006714,
|
5022 |
+
"learning_rate": 5.5852003187592226e-05,
|
5023 |
+
"loss": 2.3818,
|
5024 |
+
"step": 713
|
5025 |
+
},
|
5026 |
+
{
|
5027 |
+
"epoch": 0.6499032882011605,
|
5028 |
+
"grad_norm": 0.43209201097488403,
|
5029 |
+
"learning_rate": 5.559333873942259e-05,
|
5030 |
+
"loss": 2.3176,
|
5031 |
+
"step": 714
|
5032 |
+
},
|
5033 |
+
{
|
5034 |
+
"epoch": 0.6508135168961201,
|
5035 |
+
"grad_norm": 0.416291207075119,
|
5036 |
+
"learning_rate": 5.533504385708024e-05,
|
5037 |
+
"loss": 2.2397,
|
5038 |
+
"step": 715
|
5039 |
+
},
|
5040 |
+
{
|
5041 |
+
"epoch": 0.6517237455910797,
|
5042 |
+
"grad_norm": 0.411857932806015,
|
5043 |
+
"learning_rate": 5.5077120690174246e-05,
|
5044 |
+
"loss": 2.1142,
|
5045 |
+
"step": 716
|
5046 |
+
},
|
5047 |
+
{
|
5048 |
+
"epoch": 0.6526339742860394,
|
5049 |
+
"grad_norm": 0.4142412543296814,
|
5050 |
+
"learning_rate": 5.481957138522018e-05,
|
5051 |
+
"loss": 2.2226,
|
5052 |
+
"step": 717
|
5053 |
+
},
|
5054 |
+
{
|
5055 |
+
"epoch": 0.653544202980999,
|
5056 |
+
"grad_norm": 0.4322018325328827,
|
5057 |
+
"learning_rate": 5.456239808562209e-05,
|
5058 |
+
"loss": 2.2078,
|
5059 |
+
"step": 718
|
5060 |
+
},
|
5061 |
+
{
|
5062 |
+
"epoch": 0.6544544316759586,
|
5063 |
+
"grad_norm": 0.42210081219673157,
|
5064 |
+
"learning_rate": 5.4305602931655045e-05,
|
5065 |
+
"loss": 2.0579,
|
5066 |
+
"step": 719
|
5067 |
+
},
|
5068 |
+
{
|
5069 |
+
"epoch": 0.6553646603709182,
|
5070 |
+
"grad_norm": 0.4076862335205078,
|
5071 |
+
"learning_rate": 5.404918806044679e-05,
|
5072 |
+
"loss": 2.1348,
|
5073 |
+
"step": 720
|
5074 |
+
},
|
5075 |
+
{
|
5076 |
+
"epoch": 0.6562748890658778,
|
5077 |
+
"grad_norm": 0.43610820174217224,
|
5078 |
+
"learning_rate": 5.379315560596038e-05,
|
5079 |
+
"loss": 2.2212,
|
5080 |
+
"step": 721
|
5081 |
+
},
|
5082 |
+
{
|
5083 |
+
"epoch": 0.6571851177608374,
|
5084 |
+
"grad_norm": 0.41793620586395264,
|
5085 |
+
"learning_rate": 5.3537507698976365e-05,
|
5086 |
+
"loss": 1.9606,
|
5087 |
+
"step": 722
|
5088 |
+
},
|
5089 |
+
{
|
5090 |
+
"epoch": 0.658095346455797,
|
5091 |
+
"grad_norm": 0.40680915117263794,
|
5092 |
+
"learning_rate": 5.328224646707479e-05,
|
5093 |
+
"loss": 2.0167,
|
5094 |
+
"step": 723
|
5095 |
+
},
|
5096 |
+
{
|
5097 |
+
"epoch": 0.6590055751507566,
|
5098 |
+
"grad_norm": 0.424699991941452,
|
5099 |
+
"learning_rate": 5.3027374034617785e-05,
|
5100 |
+
"loss": 2.1065,
|
5101 |
+
"step": 724
|
5102 |
+
},
|
5103 |
+
{
|
5104 |
+
"epoch": 0.6599158038457162,
|
5105 |
+
"grad_norm": 0.43258509039878845,
|
5106 |
+
"learning_rate": 5.277289252273174e-05,
|
5107 |
+
"loss": 2.0974,
|
5108 |
+
"step": 725
|
5109 |
+
},
|
5110 |
+
{
|
5111 |
+
"epoch": 0.6608260325406758,
|
5112 |
+
"grad_norm": 0.4318636655807495,
|
5113 |
+
"learning_rate": 5.251880404928971e-05,
|
5114 |
+
"loss": 2.3214,
|
5115 |
+
"step": 726
|
5116 |
+
},
|
5117 |
+
{
|
5118 |
+
"epoch": 0.6617362612356354,
|
5119 |
+
"grad_norm": 0.4147786498069763,
|
5120 |
+
"learning_rate": 5.226511072889371e-05,
|
5121 |
+
"loss": 2.1223,
|
5122 |
+
"step": 727
|
5123 |
+
},
|
5124 |
+
{
|
5125 |
+
"epoch": 0.662646489930595,
|
5126 |
+
"grad_norm": 0.4131387770175934,
|
5127 |
+
"learning_rate": 5.201181467285723e-05,
|
5128 |
+
"loss": 1.8335,
|
5129 |
+
"step": 728
|
5130 |
+
},
|
5131 |
+
{
|
5132 |
+
"epoch": 0.6635567186255547,
|
5133 |
+
"grad_norm": 0.456827849149704,
|
5134 |
+
"learning_rate": 5.175891798918757e-05,
|
5135 |
+
"loss": 2.1428,
|
5136 |
+
"step": 729
|
5137 |
+
},
|
5138 |
+
{
|
5139 |
+
"epoch": 0.6644669473205143,
|
5140 |
+
"grad_norm": 0.47478604316711426,
|
5141 |
+
"learning_rate": 5.1506422782568345e-05,
|
5142 |
+
"loss": 2.0526,
|
5143 |
+
"step": 730
|
5144 |
+
},
|
5145 |
+
{
|
5146 |
+
"epoch": 0.6653771760154739,
|
5147 |
+
"grad_norm": 0.4357281029224396,
|
5148 |
+
"learning_rate": 5.125433115434197e-05,
|
5149 |
+
"loss": 1.8949,
|
5150 |
+
"step": 731
|
5151 |
+
},
|
5152 |
+
{
|
5153 |
+
"epoch": 0.6662874047104335,
|
5154 |
+
"grad_norm": 0.45749080181121826,
|
5155 |
+
"learning_rate": 5.100264520249205e-05,
|
5156 |
+
"loss": 2.1637,
|
5157 |
+
"step": 732
|
5158 |
+
},
|
5159 |
+
{
|
5160 |
+
"epoch": 0.6671976334053931,
|
5161 |
+
"grad_norm": 0.47157686948776245,
|
5162 |
+
"learning_rate": 5.0751367021626215e-05,
|
5163 |
+
"loss": 2.1036,
|
5164 |
+
"step": 733
|
5165 |
+
},
|
5166 |
+
{
|
5167 |
+
"epoch": 0.6681078621003527,
|
5168 |
+
"grad_norm": 0.4490397274494171,
|
5169 |
+
"learning_rate": 5.050049870295841e-05,
|
5170 |
+
"loss": 1.9553,
|
5171 |
+
"step": 734
|
5172 |
+
},
|
5173 |
+
{
|
5174 |
+
"epoch": 0.6690180907953123,
|
5175 |
+
"grad_norm": 0.4798765480518341,
|
5176 |
+
"learning_rate": 5.025004233429145e-05,
|
5177 |
+
"loss": 2.0954,
|
5178 |
+
"step": 735
|
5179 |
+
},
|
5180 |
+
{
|
5181 |
+
"epoch": 0.669928319490272,
|
5182 |
+
"grad_norm": 0.5034172534942627,
|
5183 |
+
"learning_rate": 5.000000000000002e-05,
|
5184 |
+
"loss": 2.2027,
|
5185 |
+
"step": 736
|
5186 |
+
},
|
5187 |
+
{
|
5188 |
+
"epoch": 0.6708385481852316,
|
5189 |
+
"grad_norm": 0.481141060590744,
|
5190 |
+
"learning_rate": 4.9750373781012885e-05,
|
5191 |
+
"loss": 2.0822,
|
5192 |
+
"step": 737
|
5193 |
+
},
|
5194 |
+
{
|
5195 |
+
"epoch": 0.6717487768801912,
|
5196 |
+
"grad_norm": 0.49731481075286865,
|
5197 |
+
"learning_rate": 4.950116575479586e-05,
|
5198 |
+
"loss": 2.0196,
|
5199 |
+
"step": 738
|
5200 |
+
},
|
5201 |
+
{
|
5202 |
+
"epoch": 0.6726590055751508,
|
5203 |
+
"grad_norm": 0.48321208357810974,
|
5204 |
+
"learning_rate": 4.9252377995334444e-05,
|
5205 |
+
"loss": 1.9995,
|
5206 |
+
"step": 739
|
5207 |
+
},
|
5208 |
+
{
|
5209 |
+
"epoch": 0.6735692342701104,
|
5210 |
+
"grad_norm": 0.5173296332359314,
|
5211 |
+
"learning_rate": 4.90040125731165e-05,
|
5212 |
+
"loss": 2.2078,
|
5213 |
+
"step": 740
|
5214 |
+
},
|
5215 |
+
{
|
5216 |
+
"epoch": 0.67447946296507,
|
5217 |
+
"grad_norm": 0.5107961297035217,
|
5218 |
+
"learning_rate": 4.87560715551151e-05,
|
5219 |
+
"loss": 2.0684,
|
5220 |
+
"step": 741
|
5221 |
+
},
|
5222 |
+
{
|
5223 |
+
"epoch": 0.6753896916600296,
|
5224 |
+
"grad_norm": 0.522492527961731,
|
5225 |
+
"learning_rate": 4.85085570047713e-05,
|
5226 |
+
"loss": 2.2168,
|
5227 |
+
"step": 742
|
5228 |
+
},
|
5229 |
+
{
|
5230 |
+
"epoch": 0.6762999203549892,
|
5231 |
+
"grad_norm": 0.5417767763137817,
|
5232 |
+
"learning_rate": 4.826147098197691e-05,
|
5233 |
+
"loss": 2.2379,
|
5234 |
+
"step": 743
|
5235 |
+
},
|
5236 |
+
{
|
5237 |
+
"epoch": 0.6772101490499488,
|
5238 |
+
"grad_norm": 0.5735164284706116,
|
5239 |
+
"learning_rate": 4.8014815543057475e-05,
|
5240 |
+
"loss": 2.2132,
|
5241 |
+
"step": 744
|
5242 |
+
},
|
5243 |
+
{
|
5244 |
+
"epoch": 0.6781203777449084,
|
5245 |
+
"grad_norm": 0.5819071531295776,
|
5246 |
+
"learning_rate": 4.776859274075506e-05,
|
5247 |
+
"loss": 2.2469,
|
5248 |
+
"step": 745
|
5249 |
+
},
|
5250 |
+
{
|
5251 |
+
"epoch": 0.679030606439868,
|
5252 |
+
"grad_norm": 0.6113678216934204,
|
5253 |
+
"learning_rate": 4.752280462421117e-05,
|
5254 |
+
"loss": 2.3064,
|
5255 |
+
"step": 746
|
5256 |
+
},
|
5257 |
+
{
|
5258 |
+
"epoch": 0.6799408351348276,
|
5259 |
+
"grad_norm": 0.6506679654121399,
|
5260 |
+
"learning_rate": 4.727745323894976e-05,
|
5261 |
+
"loss": 2.3311,
|
5262 |
+
"step": 747
|
5263 |
+
},
|
5264 |
+
{
|
5265 |
+
"epoch": 0.6808510638297872,
|
5266 |
+
"grad_norm": 0.7372251152992249,
|
5267 |
+
"learning_rate": 4.703254062686017e-05,
|
5268 |
+
"loss": 2.575,
|
5269 |
+
"step": 748
|
5270 |
+
},
|
5271 |
+
{
|
5272 |
+
"epoch": 0.6817612925247468,
|
5273 |
+
"grad_norm": 0.8235337734222412,
|
5274 |
+
"learning_rate": 4.678806882618003e-05,
|
5275 |
+
"loss": 2.4711,
|
5276 |
+
"step": 749
|
5277 |
+
},
|
5278 |
+
{
|
5279 |
+
"epoch": 0.6826715212197064,
|
5280 |
+
"grad_norm": 1.295682430267334,
|
5281 |
+
"learning_rate": 4.654403987147865e-05,
|
5282 |
+
"loss": 2.5713,
|
5283 |
+
"step": 750
|
5284 |
+
},
|
5285 |
+
{
|
5286 |
+
"epoch": 0.683581749914666,
|
5287 |
+
"grad_norm": 0.46585023403167725,
|
5288 |
+
"learning_rate": 4.630045579363957e-05,
|
5289 |
+
"loss": 2.4203,
|
5290 |
+
"step": 751
|
5291 |
+
},
|
5292 |
+
{
|
5293 |
+
"epoch": 0.6844919786096256,
|
5294 |
+
"grad_norm": 0.4163340628147125,
|
5295 |
+
"learning_rate": 4.605731861984401e-05,
|
5296 |
+
"loss": 2.261,
|
5297 |
+
"step": 752
|
5298 |
+
},
|
5299 |
+
{
|
5300 |
+
"epoch": 0.6854022073045852,
|
5301 |
+
"grad_norm": 0.4268229007720947,
|
5302 |
+
"learning_rate": 4.5814630373554115e-05,
|
5303 |
+
"loss": 2.3078,
|
5304 |
+
"step": 753
|
5305 |
+
},
|
5306 |
+
{
|
5307 |
+
"epoch": 0.6863124359995448,
|
5308 |
+
"grad_norm": 0.43664565682411194,
|
5309 |
+
"learning_rate": 4.557239307449561e-05,
|
5310 |
+
"loss": 2.5044,
|
5311 |
+
"step": 754
|
5312 |
+
},
|
5313 |
+
{
|
5314 |
+
"epoch": 0.6872226646945045,
|
5315 |
+
"grad_norm": 0.44144201278686523,
|
5316 |
+
"learning_rate": 4.5330608738641486e-05,
|
5317 |
+
"loss": 2.4192,
|
5318 |
+
"step": 755
|
5319 |
+
},
|
5320 |
+
{
|
5321 |
+
"epoch": 0.6881328933894642,
|
5322 |
+
"grad_norm": 0.40867024660110474,
|
5323 |
+
"learning_rate": 4.508927937819499e-05,
|
5324 |
+
"loss": 2.1908,
|
5325 |
+
"step": 756
|
5326 |
+
},
|
5327 |
+
{
|
5328 |
+
"epoch": 0.6890431220844238,
|
5329 |
+
"grad_norm": 0.4147084057331085,
|
5330 |
+
"learning_rate": 4.484840700157295e-05,
|
5331 |
+
"loss": 2.2864,
|
5332 |
+
"step": 757
|
5333 |
+
},
|
5334 |
+
{
|
5335 |
+
"epoch": 0.6899533507793834,
|
5336 |
+
"grad_norm": 0.413703054189682,
|
5337 |
+
"learning_rate": 4.4607993613388976e-05,
|
5338 |
+
"loss": 2.2436,
|
5339 |
+
"step": 758
|
5340 |
+
},
|
5341 |
+
{
|
5342 |
+
"epoch": 0.690863579474343,
|
5343 |
+
"grad_norm": 0.42309460043907166,
|
5344 |
+
"learning_rate": 4.436804121443689e-05,
|
5345 |
+
"loss": 2.3444,
|
5346 |
+
"step": 759
|
5347 |
+
},
|
5348 |
+
{
|
5349 |
+
"epoch": 0.6917738081693026,
|
5350 |
+
"grad_norm": 0.4229698181152344,
|
5351 |
+
"learning_rate": 4.412855180167406e-05,
|
5352 |
+
"loss": 2.3264,
|
5353 |
+
"step": 760
|
5354 |
+
},
|
5355 |
+
{
|
5356 |
+
"epoch": 0.6926840368642622,
|
5357 |
+
"grad_norm": 0.4247763156890869,
|
5358 |
+
"learning_rate": 4.388952736820453e-05,
|
5359 |
+
"loss": 2.257,
|
5360 |
+
"step": 761
|
5361 |
+
},
|
5362 |
+
{
|
5363 |
+
"epoch": 0.6935942655592218,
|
5364 |
+
"grad_norm": 0.42383337020874023,
|
5365 |
+
"learning_rate": 4.365096990326297e-05,
|
5366 |
+
"loss": 2.1349,
|
5367 |
+
"step": 762
|
5368 |
+
},
|
5369 |
+
{
|
5370 |
+
"epoch": 0.6945044942541814,
|
5371 |
+
"grad_norm": 0.4181368350982666,
|
5372 |
+
"learning_rate": 4.3412881392197526e-05,
|
5373 |
+
"loss": 2.2587,
|
5374 |
+
"step": 763
|
5375 |
+
},
|
5376 |
+
{
|
5377 |
+
"epoch": 0.695414722949141,
|
5378 |
+
"grad_norm": 0.4086921811103821,
|
5379 |
+
"learning_rate": 4.317526381645363e-05,
|
5380 |
+
"loss": 2.2378,
|
5381 |
+
"step": 764
|
5382 |
+
},
|
5383 |
+
{
|
5384 |
+
"epoch": 0.6963249516441006,
|
5385 |
+
"grad_norm": 0.4001654088497162,
|
5386 |
+
"learning_rate": 4.293811915355761e-05,
|
5387 |
+
"loss": 2.1708,
|
5388 |
+
"step": 765
|
5389 |
+
},
|
5390 |
+
{
|
5391 |
+
"epoch": 0.6972351803390602,
|
5392 |
+
"grad_norm": 0.42960214614868164,
|
5393 |
+
"learning_rate": 4.270144937709981e-05,
|
5394 |
+
"loss": 2.1556,
|
5395 |
+
"step": 766
|
5396 |
+
},
|
5397 |
+
{
|
5398 |
+
"epoch": 0.6981454090340198,
|
5399 |
+
"grad_norm": 0.42000848054885864,
|
5400 |
+
"learning_rate": 4.2465256456718615e-05,
|
5401 |
+
"loss": 2.1182,
|
5402 |
+
"step": 767
|
5403 |
+
},
|
5404 |
+
{
|
5405 |
+
"epoch": 0.6990556377289794,
|
5406 |
+
"grad_norm": 0.394368052482605,
|
5407 |
+
"learning_rate": 4.222954235808378e-05,
|
5408 |
+
"loss": 2.0486,
|
5409 |
+
"step": 768
|
5410 |
+
},
|
5411 |
+
{
|
5412 |
+
"epoch": 0.699965866423939,
|
5413 |
+
"grad_norm": 0.3934192359447479,
|
5414 |
+
"learning_rate": 4.19943090428802e-05,
|
5415 |
+
"loss": 1.9222,
|
5416 |
+
"step": 769
|
5417 |
+
},
|
5418 |
+
{
|
5419 |
+
"epoch": 0.7008760951188986,
|
5420 |
+
"grad_norm": 0.44399651885032654,
|
5421 |
+
"learning_rate": 4.175955846879151e-05,
|
5422 |
+
"loss": 2.1621,
|
5423 |
+
"step": 770
|
5424 |
+
},
|
5425 |
+
{
|
5426 |
+
"epoch": 0.7017863238138582,
|
5427 |
+
"grad_norm": 0.4081316888332367,
|
5428 |
+
"learning_rate": 4.1525292589483843e-05,
|
5429 |
+
"loss": 1.9534,
|
5430 |
+
"step": 771
|
5431 |
+
},
|
5432 |
+
{
|
5433 |
+
"epoch": 0.7026965525088178,
|
5434 |
+
"grad_norm": 0.42008188366889954,
|
5435 |
+
"learning_rate": 4.129151335458957e-05,
|
5436 |
+
"loss": 1.9773,
|
5437 |
+
"step": 772
|
5438 |
+
},
|
5439 |
+
{
|
5440 |
+
"epoch": 0.7036067812037774,
|
5441 |
+
"grad_norm": 0.4209793508052826,
|
5442 |
+
"learning_rate": 4.105822270969102e-05,
|
5443 |
+
"loss": 2.0403,
|
5444 |
+
"step": 773
|
5445 |
+
},
|
5446 |
+
{
|
5447 |
+
"epoch": 0.7045170098987371,
|
5448 |
+
"grad_norm": 0.43969592452049255,
|
5449 |
+
"learning_rate": 4.0825422596304396e-05,
|
5450 |
+
"loss": 2.1796,
|
5451 |
+
"step": 774
|
5452 |
+
},
|
5453 |
+
{
|
5454 |
+
"epoch": 0.7054272385936967,
|
5455 |
+
"grad_norm": 0.4333605468273163,
|
5456 |
+
"learning_rate": 4.059311495186338e-05,
|
5457 |
+
"loss": 2.1115,
|
5458 |
+
"step": 775
|
5459 |
+
},
|
5460 |
+
{
|
5461 |
+
"epoch": 0.7063374672886563,
|
5462 |
+
"grad_norm": 0.42669251561164856,
|
5463 |
+
"learning_rate": 4.036130170970341e-05,
|
5464 |
+
"loss": 2.1563,
|
5465 |
+
"step": 776
|
5466 |
+
},
|
5467 |
+
{
|
5468 |
+
"epoch": 0.7072476959836159,
|
5469 |
+
"grad_norm": 0.45064857602119446,
|
5470 |
+
"learning_rate": 4.012998479904525e-05,
|
5471 |
+
"loss": 2.2003,
|
5472 |
+
"step": 777
|
5473 |
+
},
|
5474 |
+
{
|
5475 |
+
"epoch": 0.7081579246785755,
|
5476 |
+
"grad_norm": 0.4487575888633728,
|
5477 |
+
"learning_rate": 3.9899166144978904e-05,
|
5478 |
+
"loss": 2.1332,
|
5479 |
+
"step": 778
|
5480 |
+
},
|
5481 |
+
{
|
5482 |
+
"epoch": 0.7090681533735351,
|
5483 |
+
"grad_norm": 0.4423072636127472,
|
5484 |
+
"learning_rate": 3.966884766844803e-05,
|
5485 |
+
"loss": 2.1449,
|
5486 |
+
"step": 779
|
5487 |
+
},
|
5488 |
+
{
|
5489 |
+
"epoch": 0.7099783820684947,
|
5490 |
+
"grad_norm": 0.4436761736869812,
|
5491 |
+
"learning_rate": 3.943903128623335e-05,
|
5492 |
+
"loss": 2.0025,
|
5493 |
+
"step": 780
|
5494 |
+
},
|
5495 |
+
{
|
5496 |
+
"epoch": 0.7108886107634543,
|
5497 |
+
"grad_norm": 0.4364500045776367,
|
5498 |
+
"learning_rate": 3.920971891093718e-05,
|
5499 |
+
"loss": 1.9834,
|
5500 |
+
"step": 781
|
5501 |
+
},
|
5502 |
+
{
|
5503 |
+
"epoch": 0.7117988394584139,
|
5504 |
+
"grad_norm": 0.44009071588516235,
|
5505 |
+
"learning_rate": 3.8980912450967366e-05,
|
5506 |
+
"loss": 2.0204,
|
5507 |
+
"step": 782
|
5508 |
+
},
|
5509 |
+
{
|
5510 |
+
"epoch": 0.7127090681533735,
|
5511 |
+
"grad_norm": 0.4498206079006195,
|
5512 |
+
"learning_rate": 3.875261381052121e-05,
|
5513 |
+
"loss": 2.0348,
|
5514 |
+
"step": 783
|
5515 |
+
},
|
5516 |
+
{
|
5517 |
+
"epoch": 0.7136192968483331,
|
5518 |
+
"grad_norm": 0.48029908537864685,
|
5519 |
+
"learning_rate": 3.852482488956992e-05,
|
5520 |
+
"loss": 2.0383,
|
5521 |
+
"step": 784
|
5522 |
+
},
|
5523 |
+
{
|
5524 |
+
"epoch": 0.7145295255432927,
|
5525 |
+
"grad_norm": 0.4986077845096588,
|
5526 |
+
"learning_rate": 3.829754758384262e-05,
|
5527 |
+
"loss": 2.3006,
|
5528 |
+
"step": 785
|
5529 |
+
},
|
5530 |
+
{
|
5531 |
+
"epoch": 0.7154397542382523,
|
5532 |
+
"grad_norm": 0.5001522302627563,
|
5533 |
+
"learning_rate": 3.807078378481059e-05,
|
5534 |
+
"loss": 2.3416,
|
5535 |
+
"step": 786
|
5536 |
+
},
|
5537 |
+
{
|
5538 |
+
"epoch": 0.716349982933212,
|
5539 |
+
"grad_norm": 0.5049505829811096,
|
5540 |
+
"learning_rate": 3.784453537967161e-05,
|
5541 |
+
"loss": 2.1652,
|
5542 |
+
"step": 787
|
5543 |
+
},
|
5544 |
+
{
|
5545 |
+
"epoch": 0.7172602116281716,
|
5546 |
+
"grad_norm": 0.5048404932022095,
|
5547 |
+
"learning_rate": 3.761880425133413e-05,
|
5548 |
+
"loss": 2.1345,
|
5549 |
+
"step": 788
|
5550 |
+
},
|
5551 |
+
{
|
5552 |
+
"epoch": 0.7181704403231312,
|
5553 |
+
"grad_norm": 0.4869529604911804,
|
5554 |
+
"learning_rate": 3.7393592278401704e-05,
|
5555 |
+
"loss": 2.0906,
|
5556 |
+
"step": 789
|
5557 |
+
},
|
5558 |
+
{
|
5559 |
+
"epoch": 0.7190806690180908,
|
5560 |
+
"grad_norm": 0.5454188585281372,
|
5561 |
+
"learning_rate": 3.7168901335157315e-05,
|
5562 |
+
"loss": 2.4214,
|
5563 |
+
"step": 790
|
5564 |
+
},
|
5565 |
+
{
|
5566 |
+
"epoch": 0.7199908977130504,
|
5567 |
+
"grad_norm": 0.5238876938819885,
|
5568 |
+
"learning_rate": 3.694473329154778e-05,
|
5569 |
+
"loss": 1.9798,
|
5570 |
+
"step": 791
|
5571 |
+
},
|
5572 |
+
{
|
5573 |
+
"epoch": 0.72090112640801,
|
5574 |
+
"grad_norm": 0.5545910596847534,
|
5575 |
+
"learning_rate": 3.672109001316809e-05,
|
5576 |
+
"loss": 2.4726,
|
5577 |
+
"step": 792
|
5578 |
+
},
|
5579 |
+
{
|
5580 |
+
"epoch": 0.7218113551029697,
|
5581 |
+
"grad_norm": 0.542072594165802,
|
5582 |
+
"learning_rate": 3.649797336124615e-05,
|
5583 |
+
"loss": 2.016,
|
5584 |
+
"step": 793
|
5585 |
+
},
|
5586 |
+
{
|
5587 |
+
"epoch": 0.7227215837979293,
|
5588 |
+
"grad_norm": 0.5355279445648193,
|
5589 |
+
"learning_rate": 3.6275385192627056e-05,
|
5590 |
+
"loss": 2.1041,
|
5591 |
+
"step": 794
|
5592 |
+
},
|
5593 |
+
{
|
5594 |
+
"epoch": 0.7236318124928889,
|
5595 |
+
"grad_norm": 0.5673330426216125,
|
5596 |
+
"learning_rate": 3.6053327359757535e-05,
|
5597 |
+
"loss": 2.1006,
|
5598 |
+
"step": 795
|
5599 |
+
},
|
5600 |
+
{
|
5601 |
+
"epoch": 0.7245420411878485,
|
5602 |
+
"grad_norm": 0.6170483231544495,
|
5603 |
+
"learning_rate": 3.583180171067101e-05,
|
5604 |
+
"loss": 2.3275,
|
5605 |
+
"step": 796
|
5606 |
+
},
|
5607 |
+
{
|
5608 |
+
"epoch": 0.7254522698828081,
|
5609 |
+
"grad_norm": 0.6877503991127014,
|
5610 |
+
"learning_rate": 3.5610810088971625e-05,
|
5611 |
+
"loss": 2.4504,
|
5612 |
+
"step": 797
|
5613 |
+
},
|
5614 |
+
{
|
5615 |
+
"epoch": 0.7263624985777677,
|
5616 |
+
"grad_norm": 0.7676892280578613,
|
5617 |
+
"learning_rate": 3.5390354333819344e-05,
|
5618 |
+
"loss": 2.6627,
|
5619 |
+
"step": 798
|
5620 |
+
},
|
5621 |
+
{
|
5622 |
+
"epoch": 0.7272727272727273,
|
5623 |
+
"grad_norm": 0.9272985458374023,
|
5624 |
+
"learning_rate": 3.517043627991441e-05,
|
5625 |
+
"loss": 2.5253,
|
5626 |
+
"step": 799
|
5627 |
+
},
|
5628 |
+
{
|
5629 |
+
"epoch": 0.7281829559676869,
|
5630 |
+
"grad_norm": 1.9746160507202148,
|
5631 |
+
"learning_rate": 3.4951057757482205e-05,
|
5632 |
+
"loss": 2.5993,
|
5633 |
+
"step": 800
|
5634 |
+
},
|
5635 |
+
{
|
5636 |
+
"epoch": 0.7290931846626465,
|
5637 |
+
"grad_norm": 0.4283389449119568,
|
5638 |
+
"learning_rate": 3.4732220592257946e-05,
|
5639 |
+
"loss": 2.5104,
|
5640 |
+
"step": 801
|
5641 |
+
},
|
5642 |
+
{
|
5643 |
+
"epoch": 0.7300034133576061,
|
5644 |
+
"grad_norm": 0.43243661522865295,
|
5645 |
+
"learning_rate": 3.45139266054715e-05,
|
5646 |
+
"loss": 2.1707,
|
5647 |
+
"step": 802
|
5648 |
+
},
|
5649 |
+
{
|
5650 |
+
"epoch": 0.7309136420525657,
|
5651 |
+
"grad_norm": 0.4262010455131531,
|
5652 |
+
"learning_rate": 3.429617761383222e-05,
|
5653 |
+
"loss": 2.2513,
|
5654 |
+
"step": 803
|
5655 |
+
},
|
5656 |
+
{
|
5657 |
+
"epoch": 0.7318238707475253,
|
5658 |
+
"grad_norm": 0.4308786392211914,
|
5659 |
+
"learning_rate": 3.40789754295139e-05,
|
5660 |
+
"loss": 2.4042,
|
5661 |
+
"step": 804
|
5662 |
+
},
|
5663 |
+
{
|
5664 |
+
"epoch": 0.7327340994424849,
|
5665 |
+
"grad_norm": 0.42787450551986694,
|
5666 |
+
"learning_rate": 3.3862321860139576e-05,
|
5667 |
+
"loss": 2.4259,
|
5668 |
+
"step": 805
|
5669 |
+
},
|
5670 |
+
{
|
5671 |
+
"epoch": 0.7336443281374445,
|
5672 |
+
"grad_norm": 0.4381856620311737,
|
5673 |
+
"learning_rate": 3.364621870876659e-05,
|
5674 |
+
"loss": 2.4046,
|
5675 |
+
"step": 806
|
5676 |
+
},
|
5677 |
+
{
|
5678 |
+
"epoch": 0.7345545568324041,
|
5679 |
+
"grad_norm": 0.4300837814807892,
|
5680 |
+
"learning_rate": 3.343066777387148e-05,
|
5681 |
+
"loss": 2.3713,
|
5682 |
+
"step": 807
|
5683 |
+
},
|
5684 |
+
{
|
5685 |
+
"epoch": 0.7354647855273637,
|
5686 |
+
"grad_norm": 0.4228179156780243,
|
5687 |
+
"learning_rate": 3.3215670849335155e-05,
|
5688 |
+
"loss": 2.2606,
|
5689 |
+
"step": 808
|
5690 |
+
},
|
5691 |
+
{
|
5692 |
+
"epoch": 0.7363750142223233,
|
5693 |
+
"grad_norm": 0.4267946779727936,
|
5694 |
+
"learning_rate": 3.300122972442773e-05,
|
5695 |
+
"loss": 2.3377,
|
5696 |
+
"step": 809
|
5697 |
+
},
|
5698 |
+
{
|
5699 |
+
"epoch": 0.7372852429172829,
|
5700 |
+
"grad_norm": 0.41019585728645325,
|
5701 |
+
"learning_rate": 3.278734618379402e-05,
|
5702 |
+
"loss": 2.1903,
|
5703 |
+
"step": 810
|
5704 |
+
},
|
5705 |
+
{
|
5706 |
+
"epoch": 0.7381954716122425,
|
5707 |
+
"grad_norm": 0.42949211597442627,
|
5708 |
+
"learning_rate": 3.257402200743821e-05,
|
5709 |
+
"loss": 2.3309,
|
5710 |
+
"step": 811
|
5711 |
+
},
|
5712 |
+
{
|
5713 |
+
"epoch": 0.7391057003072021,
|
5714 |
+
"grad_norm": 0.4410031735897064,
|
5715 |
+
"learning_rate": 3.2361258970709397e-05,
|
5716 |
+
"loss": 2.3924,
|
5717 |
+
"step": 812
|
5718 |
+
},
|
5719 |
+
{
|
5720 |
+
"epoch": 0.7400159290021618,
|
5721 |
+
"grad_norm": 0.4223659336566925,
|
5722 |
+
"learning_rate": 3.21490588442868e-05,
|
5723 |
+
"loss": 2.2614,
|
5724 |
+
"step": 813
|
5725 |
+
},
|
5726 |
+
{
|
5727 |
+
"epoch": 0.7409261576971214,
|
5728 |
+
"grad_norm": 0.43348926305770874,
|
5729 |
+
"learning_rate": 3.19374233941647e-05,
|
5730 |
+
"loss": 2.4255,
|
5731 |
+
"step": 814
|
5732 |
+
},
|
5733 |
+
{
|
5734 |
+
"epoch": 0.741836386392081,
|
5735 |
+
"grad_norm": 0.42184650897979736,
|
5736 |
+
"learning_rate": 3.172635438163816e-05,
|
5737 |
+
"loss": 2.2794,
|
5738 |
+
"step": 815
|
5739 |
+
},
|
5740 |
+
{
|
5741 |
+
"epoch": 0.7427466150870407,
|
5742 |
+
"grad_norm": 0.4127393066883087,
|
5743 |
+
"learning_rate": 3.1515853563288076e-05,
|
5744 |
+
"loss": 2.1242,
|
5745 |
+
"step": 816
|
5746 |
+
},
|
5747 |
+
{
|
5748 |
+
"epoch": 0.7436568437820003,
|
5749 |
+
"grad_norm": 0.40478646755218506,
|
5750 |
+
"learning_rate": 3.130592269096671e-05,
|
5751 |
+
"loss": 2.035,
|
5752 |
+
"step": 817
|
5753 |
+
},
|
5754 |
+
{
|
5755 |
+
"epoch": 0.7445670724769599,
|
5756 |
+
"grad_norm": 0.41883495450019836,
|
5757 |
+
"learning_rate": 3.1096563511783014e-05,
|
5758 |
+
"loss": 2.1427,
|
5759 |
+
"step": 818
|
5760 |
+
},
|
5761 |
+
{
|
5762 |
+
"epoch": 0.7454773011719195,
|
5763 |
+
"grad_norm": 0.39868757128715515,
|
5764 |
+
"learning_rate": 3.08877777680882e-05,
|
5765 |
+
"loss": 2.0169,
|
5766 |
+
"step": 819
|
5767 |
+
},
|
5768 |
+
{
|
5769 |
+
"epoch": 0.7463875298668791,
|
5770 |
+
"grad_norm": 0.392415851354599,
|
5771 |
+
"learning_rate": 3.0679567197461134e-05,
|
5772 |
+
"loss": 2.0969,
|
5773 |
+
"step": 820
|
5774 |
+
},
|
5775 |
+
{
|
5776 |
+
"epoch": 0.7472977585618387,
|
5777 |
+
"grad_norm": 0.4181436598300934,
|
5778 |
+
"learning_rate": 3.047193353269382e-05,
|
5779 |
+
"loss": 2.1766,
|
5780 |
+
"step": 821
|
5781 |
+
},
|
5782 |
+
{
|
5783 |
+
"epoch": 0.7482079872567983,
|
5784 |
+
"grad_norm": 0.41824692487716675,
|
5785 |
+
"learning_rate": 3.0264878501777306e-05,
|
5786 |
+
"loss": 2.0897,
|
5787 |
+
"step": 822
|
5788 |
+
},
|
5789 |
+
{
|
5790 |
+
"epoch": 0.7491182159517579,
|
5791 |
+
"grad_norm": 0.442640095949173,
|
5792 |
+
"learning_rate": 3.005840382788685e-05,
|
5793 |
+
"loss": 2.0851,
|
5794 |
+
"step": 823
|
5795 |
+
},
|
5796 |
+
{
|
5797 |
+
"epoch": 0.7500284446467175,
|
5798 |
+
"grad_norm": 0.43662169575691223,
|
5799 |
+
"learning_rate": 2.9852511229367865e-05,
|
5800 |
+
"loss": 2.1546,
|
5801 |
+
"step": 824
|
5802 |
+
},
|
5803 |
+
{
|
5804 |
+
"epoch": 0.7509386733416771,
|
5805 |
+
"grad_norm": 0.44712746143341064,
|
5806 |
+
"learning_rate": 2.9647202419721687e-05,
|
5807 |
+
"loss": 2.2304,
|
5808 |
+
"step": 825
|
5809 |
+
},
|
5810 |
+
{
|
5811 |
+
"epoch": 0.7509386733416771,
|
5812 |
+
"eval_loss": 2.2184386253356934,
|
5813 |
+
"eval_runtime": 205.4094,
|
5814 |
+
"eval_samples_per_second": 9.011,
|
5815 |
+
"eval_steps_per_second": 4.508,
|
5816 |
+
"step": 825
|
5817 |
}
|
5818 |
],
|
5819 |
"logging_steps": 1,
|
|
|
5833 |
"attributes": {}
|
5834 |
}
|
5835 |
},
|
5836 |
+
"total_flos": 1.5204499502137344e+18,
|
5837 |
"train_batch_size": 2,
|
5838 |
"trial_name": null,
|
5839 |
"trial_params": null
|