Training in progress, step 1327, 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 sha256:83dbaeaee01846f48df9ea221a857d412285b2811d32fb18577887a8512871c7
|
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 sha256:045f7efac05b92e8f94870de407382375541f5e340ad11048df478584ec32234
|
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 sha256:12322feca1d403db4eb8cb16effd4b34acca4309e1a2cdbfc72b9e408f6de635
|
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:ef39c5a61850b95d896c6a6cd037808d522ac6c3ed4aad3f7a25052c8146ef07
|
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":
|
5 |
"eval_steps": 332,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -7003,6 +7003,2323 @@
|
|
7003 |
"eval_samples_per_second": 17.59,
|
7004 |
"eval_steps_per_second": 8.799,
|
7005 |
"step": 996
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
7006 |
}
|
7007 |
],
|
7008 |
"logging_steps": 1,
|
@@ -7017,12 +9334,12 @@
|
|
7017 |
"should_evaluate": false,
|
7018 |
"should_log": false,
|
7019 |
"should_save": true,
|
7020 |
-
"should_training_stop":
|
7021 |
},
|
7022 |
"attributes": {}
|
7023 |
}
|
7024 |
},
|
7025 |
-
"total_flos": 1.
|
7026 |
"train_batch_size": 2,
|
7027 |
"trial_name": null,
|
7028 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 1.0007067137809187,
|
5 |
"eval_steps": 332,
|
6 |
+
"global_step": 1327,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
7003 |
"eval_samples_per_second": 17.59,
|
7004 |
"eval_steps_per_second": 8.799,
|
7005 |
"step": 996
|
7006 |
+
},
|
7007 |
+
{
|
7008 |
+
"epoch": 0.7515665488810365,
|
7009 |
+
"grad_norm": 9.90727710723877,
|
7010 |
+
"learning_rate": 2.941594075558366e-05,
|
7011 |
+
"loss": 0.4791,
|
7012 |
+
"step": 997
|
7013 |
+
},
|
7014 |
+
{
|
7015 |
+
"epoch": 0.7523203769140165,
|
7016 |
+
"grad_norm": 9.148994445800781,
|
7017 |
+
"learning_rate": 2.9247165864740856e-05,
|
7018 |
+
"loss": 0.4488,
|
7019 |
+
"step": 998
|
7020 |
+
},
|
7021 |
+
{
|
7022 |
+
"epoch": 0.7530742049469965,
|
7023 |
+
"grad_norm": 10.751917839050293,
|
7024 |
+
"learning_rate": 2.9078793572099616e-05,
|
7025 |
+
"loss": 0.4695,
|
7026 |
+
"step": 999
|
7027 |
+
},
|
7028 |
+
{
|
7029 |
+
"epoch": 0.7538280329799765,
|
7030 |
+
"grad_norm": 12.66123104095459,
|
7031 |
+
"learning_rate": 2.8910824835732952e-05,
|
7032 |
+
"loss": 0.4773,
|
7033 |
+
"step": 1000
|
7034 |
+
},
|
7035 |
+
{
|
7036 |
+
"epoch": 0.7545818610129564,
|
7037 |
+
"grad_norm": 5.507136821746826,
|
7038 |
+
"learning_rate": 2.8743260611417665e-05,
|
7039 |
+
"loss": 0.5073,
|
7040 |
+
"step": 1001
|
7041 |
+
},
|
7042 |
+
{
|
7043 |
+
"epoch": 0.7553356890459364,
|
7044 |
+
"grad_norm": 5.805990695953369,
|
7045 |
+
"learning_rate": 2.857610185262859e-05,
|
7046 |
+
"loss": 0.4735,
|
7047 |
+
"step": 1002
|
7048 |
+
},
|
7049 |
+
{
|
7050 |
+
"epoch": 0.7560895170789164,
|
7051 |
+
"grad_norm": 5.612555980682373,
|
7052 |
+
"learning_rate": 2.8409349510533578e-05,
|
7053 |
+
"loss": 0.4536,
|
7054 |
+
"step": 1003
|
7055 |
+
},
|
7056 |
+
{
|
7057 |
+
"epoch": 0.7568433451118963,
|
7058 |
+
"grad_norm": 5.850246906280518,
|
7059 |
+
"learning_rate": 2.8243004533987793e-05,
|
7060 |
+
"loss": 0.4578,
|
7061 |
+
"step": 1004
|
7062 |
+
},
|
7063 |
+
{
|
7064 |
+
"epoch": 0.7575971731448763,
|
7065 |
+
"grad_norm": 5.569720268249512,
|
7066 |
+
"learning_rate": 2.8077067869528417e-05,
|
7067 |
+
"loss": 0.4135,
|
7068 |
+
"step": 1005
|
7069 |
+
},
|
7070 |
+
{
|
7071 |
+
"epoch": 0.7583510011778563,
|
7072 |
+
"grad_norm": 5.9112114906311035,
|
7073 |
+
"learning_rate": 2.7911540461369222e-05,
|
7074 |
+
"loss": 0.4445,
|
7075 |
+
"step": 1006
|
7076 |
+
},
|
7077 |
+
{
|
7078 |
+
"epoch": 0.7591048292108363,
|
7079 |
+
"grad_norm": 5.9236249923706055,
|
7080 |
+
"learning_rate": 2.774642325139535e-05,
|
7081 |
+
"loss": 0.4402,
|
7082 |
+
"step": 1007
|
7083 |
+
},
|
7084 |
+
{
|
7085 |
+
"epoch": 0.7598586572438163,
|
7086 |
+
"grad_norm": 6.210232257843018,
|
7087 |
+
"learning_rate": 2.7581717179157606e-05,
|
7088 |
+
"loss": 0.4605,
|
7089 |
+
"step": 1008
|
7090 |
+
},
|
7091 |
+
{
|
7092 |
+
"epoch": 0.7606124852767963,
|
7093 |
+
"grad_norm": 5.880030155181885,
|
7094 |
+
"learning_rate": 2.7417423181867585e-05,
|
7095 |
+
"loss": 0.4227,
|
7096 |
+
"step": 1009
|
7097 |
+
},
|
7098 |
+
{
|
7099 |
+
"epoch": 0.7613663133097762,
|
7100 |
+
"grad_norm": 5.549881458282471,
|
7101 |
+
"learning_rate": 2.72535421943919e-05,
|
7102 |
+
"loss": 0.4168,
|
7103 |
+
"step": 1010
|
7104 |
+
},
|
7105 |
+
{
|
7106 |
+
"epoch": 0.7621201413427562,
|
7107 |
+
"grad_norm": 5.586158275604248,
|
7108 |
+
"learning_rate": 2.7090075149247217e-05,
|
7109 |
+
"loss": 0.4334,
|
7110 |
+
"step": 1011
|
7111 |
+
},
|
7112 |
+
{
|
7113 |
+
"epoch": 0.7628739693757361,
|
7114 |
+
"grad_norm": 5.5952348709106445,
|
7115 |
+
"learning_rate": 2.6927022976594607e-05,
|
7116 |
+
"loss": 0.4232,
|
7117 |
+
"step": 1012
|
7118 |
+
},
|
7119 |
+
{
|
7120 |
+
"epoch": 0.7636277974087161,
|
7121 |
+
"grad_norm": 5.478029727935791,
|
7122 |
+
"learning_rate": 2.676438660423457e-05,
|
7123 |
+
"loss": 0.4053,
|
7124 |
+
"step": 1013
|
7125 |
+
},
|
7126 |
+
{
|
7127 |
+
"epoch": 0.7643816254416961,
|
7128 |
+
"grad_norm": 5.441522121429443,
|
7129 |
+
"learning_rate": 2.660216695760157e-05,
|
7130 |
+
"loss": 0.3847,
|
7131 |
+
"step": 1014
|
7132 |
+
},
|
7133 |
+
{
|
7134 |
+
"epoch": 0.765135453474676,
|
7135 |
+
"grad_norm": 5.584785461425781,
|
7136 |
+
"learning_rate": 2.6440364959758813e-05,
|
7137 |
+
"loss": 0.4098,
|
7138 |
+
"step": 1015
|
7139 |
+
},
|
7140 |
+
{
|
7141 |
+
"epoch": 0.7658892815076561,
|
7142 |
+
"grad_norm": 5.545854091644287,
|
7143 |
+
"learning_rate": 2.6278981531392945e-05,
|
7144 |
+
"loss": 0.4002,
|
7145 |
+
"step": 1016
|
7146 |
+
},
|
7147 |
+
{
|
7148 |
+
"epoch": 0.7666431095406361,
|
7149 |
+
"grad_norm": 5.697778701782227,
|
7150 |
+
"learning_rate": 2.6118017590809017e-05,
|
7151 |
+
"loss": 0.4013,
|
7152 |
+
"step": 1017
|
7153 |
+
},
|
7154 |
+
{
|
7155 |
+
"epoch": 0.767396937573616,
|
7156 |
+
"grad_norm": 6.265735626220703,
|
7157 |
+
"learning_rate": 2.595747405392491e-05,
|
7158 |
+
"loss": 0.4102,
|
7159 |
+
"step": 1018
|
7160 |
+
},
|
7161 |
+
{
|
7162 |
+
"epoch": 0.768150765606596,
|
7163 |
+
"grad_norm": 5.284882545471191,
|
7164 |
+
"learning_rate": 2.579735183426649e-05,
|
7165 |
+
"loss": 0.3747,
|
7166 |
+
"step": 1019
|
7167 |
+
},
|
7168 |
+
{
|
7169 |
+
"epoch": 0.768904593639576,
|
7170 |
+
"grad_norm": 5.939345359802246,
|
7171 |
+
"learning_rate": 2.5637651842962164e-05,
|
7172 |
+
"loss": 0.4019,
|
7173 |
+
"step": 1020
|
7174 |
+
},
|
7175 |
+
{
|
7176 |
+
"epoch": 0.7696584216725559,
|
7177 |
+
"grad_norm": 5.655182838439941,
|
7178 |
+
"learning_rate": 2.5478374988737753e-05,
|
7179 |
+
"loss": 0.4038,
|
7180 |
+
"step": 1021
|
7181 |
+
},
|
7182 |
+
{
|
7183 |
+
"epoch": 0.7704122497055359,
|
7184 |
+
"grad_norm": 5.510229587554932,
|
7185 |
+
"learning_rate": 2.531952217791136e-05,
|
7186 |
+
"loss": 0.3912,
|
7187 |
+
"step": 1022
|
7188 |
+
},
|
7189 |
+
{
|
7190 |
+
"epoch": 0.7711660777385159,
|
7191 |
+
"grad_norm": 5.720643997192383,
|
7192 |
+
"learning_rate": 2.5161094314388278e-05,
|
7193 |
+
"loss": 0.3995,
|
7194 |
+
"step": 1023
|
7195 |
+
},
|
7196 |
+
{
|
7197 |
+
"epoch": 0.7719199057714958,
|
7198 |
+
"grad_norm": 5.860435962677002,
|
7199 |
+
"learning_rate": 2.5003092299655584e-05,
|
7200 |
+
"loss": 0.3995,
|
7201 |
+
"step": 1024
|
7202 |
+
},
|
7203 |
+
{
|
7204 |
+
"epoch": 0.7726737338044759,
|
7205 |
+
"grad_norm": 6.223293304443359,
|
7206 |
+
"learning_rate": 2.4845517032777364e-05,
|
7207 |
+
"loss": 0.4424,
|
7208 |
+
"step": 1025
|
7209 |
+
},
|
7210 |
+
{
|
7211 |
+
"epoch": 0.7734275618374559,
|
7212 |
+
"grad_norm": 6.027644157409668,
|
7213 |
+
"learning_rate": 2.4688369410389334e-05,
|
7214 |
+
"loss": 0.4299,
|
7215 |
+
"step": 1026
|
7216 |
+
},
|
7217 |
+
{
|
7218 |
+
"epoch": 0.7741813898704358,
|
7219 |
+
"grad_norm": 5.946674346923828,
|
7220 |
+
"learning_rate": 2.4531650326693822e-05,
|
7221 |
+
"loss": 0.3849,
|
7222 |
+
"step": 1027
|
7223 |
+
},
|
7224 |
+
{
|
7225 |
+
"epoch": 0.7749352179034158,
|
7226 |
+
"grad_norm": 6.277134895324707,
|
7227 |
+
"learning_rate": 2.4375360673454718e-05,
|
7228 |
+
"loss": 0.4147,
|
7229 |
+
"step": 1028
|
7230 |
+
},
|
7231 |
+
{
|
7232 |
+
"epoch": 0.7756890459363958,
|
7233 |
+
"grad_norm": 6.024038314819336,
|
7234 |
+
"learning_rate": 2.4219501339992334e-05,
|
7235 |
+
"loss": 0.3774,
|
7236 |
+
"step": 1029
|
7237 |
+
},
|
7238 |
+
{
|
7239 |
+
"epoch": 0.7764428739693757,
|
7240 |
+
"grad_norm": 5.8574910163879395,
|
7241 |
+
"learning_rate": 2.406407321317835e-05,
|
7242 |
+
"loss": 0.3865,
|
7243 |
+
"step": 1030
|
7244 |
+
},
|
7245 |
+
{
|
7246 |
+
"epoch": 0.7771967020023557,
|
7247 |
+
"grad_norm": 6.022578239440918,
|
7248 |
+
"learning_rate": 2.3909077177430893e-05,
|
7249 |
+
"loss": 0.3957,
|
7250 |
+
"step": 1031
|
7251 |
+
},
|
7252 |
+
{
|
7253 |
+
"epoch": 0.7779505300353357,
|
7254 |
+
"grad_norm": 5.923416614532471,
|
7255 |
+
"learning_rate": 2.3754514114709304e-05,
|
7256 |
+
"loss": 0.3836,
|
7257 |
+
"step": 1032
|
7258 |
+
},
|
7259 |
+
{
|
7260 |
+
"epoch": 0.7787043580683156,
|
7261 |
+
"grad_norm": 6.270270824432373,
|
7262 |
+
"learning_rate": 2.3600384904509254e-05,
|
7263 |
+
"loss": 0.3979,
|
7264 |
+
"step": 1033
|
7265 |
+
},
|
7266 |
+
{
|
7267 |
+
"epoch": 0.7794581861012957,
|
7268 |
+
"grad_norm": 6.285928726196289,
|
7269 |
+
"learning_rate": 2.3446690423857685e-05,
|
7270 |
+
"loss": 0.4098,
|
7271 |
+
"step": 1034
|
7272 |
+
},
|
7273 |
+
{
|
7274 |
+
"epoch": 0.7802120141342757,
|
7275 |
+
"grad_norm": 6.104770660400391,
|
7276 |
+
"learning_rate": 2.3293431547307887e-05,
|
7277 |
+
"loss": 0.3746,
|
7278 |
+
"step": 1035
|
7279 |
+
},
|
7280 |
+
{
|
7281 |
+
"epoch": 0.7809658421672556,
|
7282 |
+
"grad_norm": 6.284374237060547,
|
7283 |
+
"learning_rate": 2.31406091469344e-05,
|
7284 |
+
"loss": 0.3933,
|
7285 |
+
"step": 1036
|
7286 |
+
},
|
7287 |
+
{
|
7288 |
+
"epoch": 0.7817196702002356,
|
7289 |
+
"grad_norm": 6.502585411071777,
|
7290 |
+
"learning_rate": 2.298822409232817e-05,
|
7291 |
+
"loss": 0.3964,
|
7292 |
+
"step": 1037
|
7293 |
+
},
|
7294 |
+
{
|
7295 |
+
"epoch": 0.7824734982332155,
|
7296 |
+
"grad_norm": 6.121708869934082,
|
7297 |
+
"learning_rate": 2.2836277250591574e-05,
|
7298 |
+
"loss": 0.3822,
|
7299 |
+
"step": 1038
|
7300 |
+
},
|
7301 |
+
{
|
7302 |
+
"epoch": 0.7832273262661955,
|
7303 |
+
"grad_norm": 7.069113731384277,
|
7304 |
+
"learning_rate": 2.2684769486333445e-05,
|
7305 |
+
"loss": 0.3919,
|
7306 |
+
"step": 1039
|
7307 |
+
},
|
7308 |
+
{
|
7309 |
+
"epoch": 0.7839811542991755,
|
7310 |
+
"grad_norm": 6.825623035430908,
|
7311 |
+
"learning_rate": 2.2533701661664154e-05,
|
7312 |
+
"loss": 0.4296,
|
7313 |
+
"step": 1040
|
7314 |
+
},
|
7315 |
+
{
|
7316 |
+
"epoch": 0.7847349823321554,
|
7317 |
+
"grad_norm": 7.632999897003174,
|
7318 |
+
"learning_rate": 2.2383074636190748e-05,
|
7319 |
+
"loss": 0.4266,
|
7320 |
+
"step": 1041
|
7321 |
+
},
|
7322 |
+
{
|
7323 |
+
"epoch": 0.7854888103651354,
|
7324 |
+
"grad_norm": 7.41874885559082,
|
7325 |
+
"learning_rate": 2.2232889267012038e-05,
|
7326 |
+
"loss": 0.4263,
|
7327 |
+
"step": 1042
|
7328 |
+
},
|
7329 |
+
{
|
7330 |
+
"epoch": 0.7862426383981155,
|
7331 |
+
"grad_norm": 7.6582417488098145,
|
7332 |
+
"learning_rate": 2.2083146408713673e-05,
|
7333 |
+
"loss": 0.4351,
|
7334 |
+
"step": 1043
|
7335 |
+
},
|
7336 |
+
{
|
7337 |
+
"epoch": 0.7869964664310954,
|
7338 |
+
"grad_norm": 9.17532730102539,
|
7339 |
+
"learning_rate": 2.1933846913363466e-05,
|
7340 |
+
"loss": 0.5107,
|
7341 |
+
"step": 1044
|
7342 |
+
},
|
7343 |
+
{
|
7344 |
+
"epoch": 0.7877502944640754,
|
7345 |
+
"grad_norm": 9.609545707702637,
|
7346 |
+
"learning_rate": 2.178499163050617e-05,
|
7347 |
+
"loss": 0.4606,
|
7348 |
+
"step": 1045
|
7349 |
+
},
|
7350 |
+
{
|
7351 |
+
"epoch": 0.7885041224970554,
|
7352 |
+
"grad_norm": 9.567949295043945,
|
7353 |
+
"learning_rate": 2.1636581407159105e-05,
|
7354 |
+
"loss": 0.4663,
|
7355 |
+
"step": 1046
|
7356 |
+
},
|
7357 |
+
{
|
7358 |
+
"epoch": 0.7892579505300353,
|
7359 |
+
"grad_norm": 9.527708053588867,
|
7360 |
+
"learning_rate": 2.1488617087806982e-05,
|
7361 |
+
"loss": 0.4712,
|
7362 |
+
"step": 1047
|
7363 |
+
},
|
7364 |
+
{
|
7365 |
+
"epoch": 0.7900117785630153,
|
7366 |
+
"grad_norm": 9.680562973022461,
|
7367 |
+
"learning_rate": 2.1341099514397266e-05,
|
7368 |
+
"loss": 0.4975,
|
7369 |
+
"step": 1048
|
7370 |
+
},
|
7371 |
+
{
|
7372 |
+
"epoch": 0.7907656065959953,
|
7373 |
+
"grad_norm": 10.399216651916504,
|
7374 |
+
"learning_rate": 2.1194029526335303e-05,
|
7375 |
+
"loss": 0.4586,
|
7376 |
+
"step": 1049
|
7377 |
+
},
|
7378 |
+
{
|
7379 |
+
"epoch": 0.7915194346289752,
|
7380 |
+
"grad_norm": 10.869539260864258,
|
7381 |
+
"learning_rate": 2.1047407960479702e-05,
|
7382 |
+
"loss": 0.4429,
|
7383 |
+
"step": 1050
|
7384 |
+
},
|
7385 |
+
{
|
7386 |
+
"epoch": 0.7922732626619552,
|
7387 |
+
"grad_norm": 5.385607719421387,
|
7388 |
+
"learning_rate": 2.0901235651137284e-05,
|
7389 |
+
"loss": 0.5019,
|
7390 |
+
"step": 1051
|
7391 |
+
},
|
7392 |
+
{
|
7393 |
+
"epoch": 0.7930270906949353,
|
7394 |
+
"grad_norm": 5.6260223388671875,
|
7395 |
+
"learning_rate": 2.0755513430058672e-05,
|
7396 |
+
"loss": 0.4988,
|
7397 |
+
"step": 1052
|
7398 |
+
},
|
7399 |
+
{
|
7400 |
+
"epoch": 0.7937809187279152,
|
7401 |
+
"grad_norm": 5.487570762634277,
|
7402 |
+
"learning_rate": 2.0610242126433297e-05,
|
7403 |
+
"loss": 0.4594,
|
7404 |
+
"step": 1053
|
7405 |
+
},
|
7406 |
+
{
|
7407 |
+
"epoch": 0.7945347467608952,
|
7408 |
+
"grad_norm": 5.6461591720581055,
|
7409 |
+
"learning_rate": 2.0465422566884805e-05,
|
7410 |
+
"loss": 0.4642,
|
7411 |
+
"step": 1054
|
7412 |
+
},
|
7413 |
+
{
|
7414 |
+
"epoch": 0.7952885747938752,
|
7415 |
+
"grad_norm": 5.7345123291015625,
|
7416 |
+
"learning_rate": 2.0321055575466284e-05,
|
7417 |
+
"loss": 0.4442,
|
7418 |
+
"step": 1055
|
7419 |
+
},
|
7420 |
+
{
|
7421 |
+
"epoch": 0.7960424028268551,
|
7422 |
+
"grad_norm": 5.918202877044678,
|
7423 |
+
"learning_rate": 2.0177141973655766e-05,
|
7424 |
+
"loss": 0.4708,
|
7425 |
+
"step": 1056
|
7426 |
+
},
|
7427 |
+
{
|
7428 |
+
"epoch": 0.7967962308598351,
|
7429 |
+
"grad_norm": 5.593347549438477,
|
7430 |
+
"learning_rate": 2.0033682580351144e-05,
|
7431 |
+
"loss": 0.4277,
|
7432 |
+
"step": 1057
|
7433 |
+
},
|
7434 |
+
{
|
7435 |
+
"epoch": 0.797550058892815,
|
7436 |
+
"grad_norm": 5.557769775390625,
|
7437 |
+
"learning_rate": 1.9890678211866033e-05,
|
7438 |
+
"loss": 0.4267,
|
7439 |
+
"step": 1058
|
7440 |
+
},
|
7441 |
+
{
|
7442 |
+
"epoch": 0.798303886925795,
|
7443 |
+
"grad_norm": 5.38918924331665,
|
7444 |
+
"learning_rate": 1.9748129681924675e-05,
|
7445 |
+
"loss": 0.4112,
|
7446 |
+
"step": 1059
|
7447 |
+
},
|
7448 |
+
{
|
7449 |
+
"epoch": 0.799057714958775,
|
7450 |
+
"grad_norm": 5.82417631149292,
|
7451 |
+
"learning_rate": 1.9606037801657673e-05,
|
7452 |
+
"loss": 0.4104,
|
7453 |
+
"step": 1060
|
7454 |
+
},
|
7455 |
+
{
|
7456 |
+
"epoch": 0.799811542991755,
|
7457 |
+
"grad_norm": 5.548363208770752,
|
7458 |
+
"learning_rate": 1.9464403379596963e-05,
|
7459 |
+
"loss": 0.4127,
|
7460 |
+
"step": 1061
|
7461 |
+
},
|
7462 |
+
{
|
7463 |
+
"epoch": 0.800565371024735,
|
7464 |
+
"grad_norm": 5.548163890838623,
|
7465 |
+
"learning_rate": 1.932322722167168e-05,
|
7466 |
+
"loss": 0.4198,
|
7467 |
+
"step": 1062
|
7468 |
+
},
|
7469 |
+
{
|
7470 |
+
"epoch": 0.801319199057715,
|
7471 |
+
"grad_norm": 5.443014621734619,
|
7472 |
+
"learning_rate": 1.9182510131203224e-05,
|
7473 |
+
"loss": 0.4012,
|
7474 |
+
"step": 1063
|
7475 |
+
},
|
7476 |
+
{
|
7477 |
+
"epoch": 0.802073027090695,
|
7478 |
+
"grad_norm": 5.750105381011963,
|
7479 |
+
"learning_rate": 1.9042252908900814e-05,
|
7480 |
+
"loss": 0.4075,
|
7481 |
+
"step": 1064
|
7482 |
+
},
|
7483 |
+
{
|
7484 |
+
"epoch": 0.8028268551236749,
|
7485 |
+
"grad_norm": 5.6281418800354,
|
7486 |
+
"learning_rate": 1.8902456352856925e-05,
|
7487 |
+
"loss": 0.3896,
|
7488 |
+
"step": 1065
|
7489 |
+
},
|
7490 |
+
{
|
7491 |
+
"epoch": 0.8035806831566549,
|
7492 |
+
"grad_norm": 5.443961143493652,
|
7493 |
+
"learning_rate": 1.8763121258542815e-05,
|
7494 |
+
"loss": 0.4057,
|
7495 |
+
"step": 1066
|
7496 |
+
},
|
7497 |
+
{
|
7498 |
+
"epoch": 0.8043345111896348,
|
7499 |
+
"grad_norm": 5.808502674102783,
|
7500 |
+
"learning_rate": 1.86242484188038e-05,
|
7501 |
+
"loss": 0.4137,
|
7502 |
+
"step": 1067
|
7503 |
+
},
|
7504 |
+
{
|
7505 |
+
"epoch": 0.8050883392226148,
|
7506 |
+
"grad_norm": 5.866790294647217,
|
7507 |
+
"learning_rate": 1.848583862385501e-05,
|
7508 |
+
"loss": 0.4129,
|
7509 |
+
"step": 1068
|
7510 |
+
},
|
7511 |
+
{
|
7512 |
+
"epoch": 0.8058421672555948,
|
7513 |
+
"grad_norm": 5.517582893371582,
|
7514 |
+
"learning_rate": 1.8347892661276656e-05,
|
7515 |
+
"loss": 0.3901,
|
7516 |
+
"step": 1069
|
7517 |
+
},
|
7518 |
+
{
|
7519 |
+
"epoch": 0.8065959952885748,
|
7520 |
+
"grad_norm": 6.088197231292725,
|
7521 |
+
"learning_rate": 1.82104113160097e-05,
|
7522 |
+
"loss": 0.4125,
|
7523 |
+
"step": 1070
|
7524 |
+
},
|
7525 |
+
{
|
7526 |
+
"epoch": 0.8073498233215548,
|
7527 |
+
"grad_norm": 5.613511562347412,
|
7528 |
+
"learning_rate": 1.8073395370351287e-05,
|
7529 |
+
"loss": 0.3968,
|
7530 |
+
"step": 1071
|
7531 |
+
},
|
7532 |
+
{
|
7533 |
+
"epoch": 0.8081036513545348,
|
7534 |
+
"grad_norm": 5.712565898895264,
|
7535 |
+
"learning_rate": 1.7936845603950447e-05,
|
7536 |
+
"loss": 0.3925,
|
7537 |
+
"step": 1072
|
7538 |
+
},
|
7539 |
+
{
|
7540 |
+
"epoch": 0.8088574793875147,
|
7541 |
+
"grad_norm": 5.371545314788818,
|
7542 |
+
"learning_rate": 1.780076279380337e-05,
|
7543 |
+
"loss": 0.3589,
|
7544 |
+
"step": 1073
|
7545 |
+
},
|
7546 |
+
{
|
7547 |
+
"epoch": 0.8096113074204947,
|
7548 |
+
"grad_norm": 5.599592208862305,
|
7549 |
+
"learning_rate": 1.7665147714249376e-05,
|
7550 |
+
"loss": 0.3838,
|
7551 |
+
"step": 1074
|
7552 |
+
},
|
7553 |
+
{
|
7554 |
+
"epoch": 0.8103651354534747,
|
7555 |
+
"grad_norm": 6.015298843383789,
|
7556 |
+
"learning_rate": 1.753000113696617e-05,
|
7557 |
+
"loss": 0.386,
|
7558 |
+
"step": 1075
|
7559 |
+
},
|
7560 |
+
{
|
7561 |
+
"epoch": 0.8111189634864546,
|
7562 |
+
"grad_norm": 5.434444427490234,
|
7563 |
+
"learning_rate": 1.7395323830965605e-05,
|
7564 |
+
"loss": 0.3771,
|
7565 |
+
"step": 1076
|
7566 |
+
},
|
7567 |
+
{
|
7568 |
+
"epoch": 0.8118727915194346,
|
7569 |
+
"grad_norm": 6.145053863525391,
|
7570 |
+
"learning_rate": 1.726111656258932e-05,
|
7571 |
+
"loss": 0.4039,
|
7572 |
+
"step": 1077
|
7573 |
+
},
|
7574 |
+
{
|
7575 |
+
"epoch": 0.8126266195524146,
|
7576 |
+
"grad_norm": 5.7801384925842285,
|
7577 |
+
"learning_rate": 1.7127380095504296e-05,
|
7578 |
+
"loss": 0.3955,
|
7579 |
+
"step": 1078
|
7580 |
+
},
|
7581 |
+
{
|
7582 |
+
"epoch": 0.8133804475853946,
|
7583 |
+
"grad_norm": 5.640938758850098,
|
7584 |
+
"learning_rate": 1.699411519069858e-05,
|
7585 |
+
"loss": 0.3788,
|
7586 |
+
"step": 1079
|
7587 |
+
},
|
7588 |
+
{
|
7589 |
+
"epoch": 0.8141342756183746,
|
7590 |
+
"grad_norm": 5.714921951293945,
|
7591 |
+
"learning_rate": 1.686132260647696e-05,
|
7592 |
+
"loss": 0.3637,
|
7593 |
+
"step": 1080
|
7594 |
+
},
|
7595 |
+
{
|
7596 |
+
"epoch": 0.8148881036513546,
|
7597 |
+
"grad_norm": 6.3913750648498535,
|
7598 |
+
"learning_rate": 1.6729003098456576e-05,
|
7599 |
+
"loss": 0.3815,
|
7600 |
+
"step": 1081
|
7601 |
+
},
|
7602 |
+
{
|
7603 |
+
"epoch": 0.8156419316843345,
|
7604 |
+
"grad_norm": 5.981407642364502,
|
7605 |
+
"learning_rate": 1.6597157419562703e-05,
|
7606 |
+
"loss": 0.3756,
|
7607 |
+
"step": 1082
|
7608 |
+
},
|
7609 |
+
{
|
7610 |
+
"epoch": 0.8163957597173145,
|
7611 |
+
"grad_norm": 6.408857822418213,
|
7612 |
+
"learning_rate": 1.646578632002439e-05,
|
7613 |
+
"loss": 0.4219,
|
7614 |
+
"step": 1083
|
7615 |
+
},
|
7616 |
+
{
|
7617 |
+
"epoch": 0.8171495877502944,
|
7618 |
+
"grad_norm": 6.3557329177856445,
|
7619 |
+
"learning_rate": 1.6334890547370286e-05,
|
7620 |
+
"loss": 0.387,
|
7621 |
+
"step": 1084
|
7622 |
+
},
|
7623 |
+
{
|
7624 |
+
"epoch": 0.8179034157832744,
|
7625 |
+
"grad_norm": 6.406612873077393,
|
7626 |
+
"learning_rate": 1.6204470846424268e-05,
|
7627 |
+
"loss": 0.3736,
|
7628 |
+
"step": 1085
|
7629 |
+
},
|
7630 |
+
{
|
7631 |
+
"epoch": 0.8186572438162544,
|
7632 |
+
"grad_norm": 6.225420951843262,
|
7633 |
+
"learning_rate": 1.607452795930131e-05,
|
7634 |
+
"loss": 0.3886,
|
7635 |
+
"step": 1086
|
7636 |
+
},
|
7637 |
+
{
|
7638 |
+
"epoch": 0.8194110718492343,
|
7639 |
+
"grad_norm": 6.3113789558410645,
|
7640 |
+
"learning_rate": 1.594506262540324e-05,
|
7641 |
+
"loss": 0.402,
|
7642 |
+
"step": 1087
|
7643 |
+
},
|
7644 |
+
{
|
7645 |
+
"epoch": 0.8201648998822144,
|
7646 |
+
"grad_norm": 6.504429817199707,
|
7647 |
+
"learning_rate": 1.5816075581414458e-05,
|
7648 |
+
"loss": 0.3911,
|
7649 |
+
"step": 1088
|
7650 |
+
},
|
7651 |
+
{
|
7652 |
+
"epoch": 0.8209187279151944,
|
7653 |
+
"grad_norm": 7.651139736175537,
|
7654 |
+
"learning_rate": 1.56875675612978e-05,
|
7655 |
+
"loss": 0.4127,
|
7656 |
+
"step": 1089
|
7657 |
+
},
|
7658 |
+
{
|
7659 |
+
"epoch": 0.8216725559481743,
|
7660 |
+
"grad_norm": 6.864494800567627,
|
7661 |
+
"learning_rate": 1.5559539296290403e-05,
|
7662 |
+
"loss": 0.3841,
|
7663 |
+
"step": 1090
|
7664 |
+
},
|
7665 |
+
{
|
7666 |
+
"epoch": 0.8224263839811543,
|
7667 |
+
"grad_norm": 7.120053291320801,
|
7668 |
+
"learning_rate": 1.5431991514899446e-05,
|
7669 |
+
"loss": 0.4185,
|
7670 |
+
"step": 1091
|
7671 |
+
},
|
7672 |
+
{
|
7673 |
+
"epoch": 0.8231802120141343,
|
7674 |
+
"grad_norm": 7.861664295196533,
|
7675 |
+
"learning_rate": 1.5304924942898068e-05,
|
7676 |
+
"loss": 0.4293,
|
7677 |
+
"step": 1092
|
7678 |
+
},
|
7679 |
+
{
|
7680 |
+
"epoch": 0.8239340400471142,
|
7681 |
+
"grad_norm": 8.355661392211914,
|
7682 |
+
"learning_rate": 1.5178340303321314e-05,
|
7683 |
+
"loss": 0.4559,
|
7684 |
+
"step": 1093
|
7685 |
+
},
|
7686 |
+
{
|
7687 |
+
"epoch": 0.8246878680800942,
|
7688 |
+
"grad_norm": 8.859525680541992,
|
7689 |
+
"learning_rate": 1.5052238316461753e-05,
|
7690 |
+
"loss": 0.4503,
|
7691 |
+
"step": 1094
|
7692 |
+
},
|
7693 |
+
{
|
7694 |
+
"epoch": 0.8254416961130742,
|
7695 |
+
"grad_norm": 9.211348533630371,
|
7696 |
+
"learning_rate": 1.492661969986574e-05,
|
7697 |
+
"loss": 0.4435,
|
7698 |
+
"step": 1095
|
7699 |
+
},
|
7700 |
+
{
|
7701 |
+
"epoch": 0.8261955241460541,
|
7702 |
+
"grad_norm": 8.610541343688965,
|
7703 |
+
"learning_rate": 1.4801485168329066e-05,
|
7704 |
+
"loss": 0.4625,
|
7705 |
+
"step": 1096
|
7706 |
+
},
|
7707 |
+
{
|
7708 |
+
"epoch": 0.8269493521790342,
|
7709 |
+
"grad_norm": 10.033802032470703,
|
7710 |
+
"learning_rate": 1.4676835433892989e-05,
|
7711 |
+
"loss": 0.437,
|
7712 |
+
"step": 1097
|
7713 |
+
},
|
7714 |
+
{
|
7715 |
+
"epoch": 0.8277031802120142,
|
7716 |
+
"grad_norm": 10.607207298278809,
|
7717 |
+
"learning_rate": 1.4552671205840163e-05,
|
7718 |
+
"loss": 0.4369,
|
7719 |
+
"step": 1098
|
7720 |
+
},
|
7721 |
+
{
|
7722 |
+
"epoch": 0.8284570082449941,
|
7723 |
+
"grad_norm": 10.07897663116455,
|
7724 |
+
"learning_rate": 1.4428993190690677e-05,
|
7725 |
+
"loss": 0.4563,
|
7726 |
+
"step": 1099
|
7727 |
+
},
|
7728 |
+
{
|
7729 |
+
"epoch": 0.8292108362779741,
|
7730 |
+
"grad_norm": 12.518508911132812,
|
7731 |
+
"learning_rate": 1.4305802092197829e-05,
|
7732 |
+
"loss": 0.4645,
|
7733 |
+
"step": 1100
|
7734 |
+
},
|
7735 |
+
{
|
7736 |
+
"epoch": 0.8299646643109541,
|
7737 |
+
"grad_norm": 5.578033924102783,
|
7738 |
+
"learning_rate": 1.4183098611344415e-05,
|
7739 |
+
"loss": 0.51,
|
7740 |
+
"step": 1101
|
7741 |
+
},
|
7742 |
+
{
|
7743 |
+
"epoch": 0.830718492343934,
|
7744 |
+
"grad_norm": 5.301563739776611,
|
7745 |
+
"learning_rate": 1.4060883446338502e-05,
|
7746 |
+
"loss": 0.4486,
|
7747 |
+
"step": 1102
|
7748 |
+
},
|
7749 |
+
{
|
7750 |
+
"epoch": 0.831472320376914,
|
7751 |
+
"grad_norm": 5.3994293212890625,
|
7752 |
+
"learning_rate": 1.393915729260955e-05,
|
7753 |
+
"loss": 0.4536,
|
7754 |
+
"step": 1103
|
7755 |
+
},
|
7756 |
+
{
|
7757 |
+
"epoch": 0.832226148409894,
|
7758 |
+
"grad_norm": 5.560753345489502,
|
7759 |
+
"learning_rate": 1.3817920842804433e-05,
|
7760 |
+
"loss": 0.455,
|
7761 |
+
"step": 1104
|
7762 |
+
},
|
7763 |
+
{
|
7764 |
+
"epoch": 0.8329799764428739,
|
7765 |
+
"grad_norm": 5.810977935791016,
|
7766 |
+
"learning_rate": 1.3697174786783584e-05,
|
7767 |
+
"loss": 0.4373,
|
7768 |
+
"step": 1105
|
7769 |
+
},
|
7770 |
+
{
|
7771 |
+
"epoch": 0.833733804475854,
|
7772 |
+
"grad_norm": 5.4894256591796875,
|
7773 |
+
"learning_rate": 1.3576919811616862e-05,
|
7774 |
+
"loss": 0.4106,
|
7775 |
+
"step": 1106
|
7776 |
+
},
|
7777 |
+
{
|
7778 |
+
"epoch": 0.834487632508834,
|
7779 |
+
"grad_norm": 5.865782737731934,
|
7780 |
+
"learning_rate": 1.345715660157989e-05,
|
7781 |
+
"loss": 0.4151,
|
7782 |
+
"step": 1107
|
7783 |
+
},
|
7784 |
+
{
|
7785 |
+
"epoch": 0.8352414605418139,
|
7786 |
+
"grad_norm": 5.4949469566345215,
|
7787 |
+
"learning_rate": 1.3337885838149988e-05,
|
7788 |
+
"loss": 0.4422,
|
7789 |
+
"step": 1108
|
7790 |
+
},
|
7791 |
+
{
|
7792 |
+
"epoch": 0.8359952885747939,
|
7793 |
+
"grad_norm": 5.45637845993042,
|
7794 |
+
"learning_rate": 1.3219108200002418e-05,
|
7795 |
+
"loss": 0.4237,
|
7796 |
+
"step": 1109
|
7797 |
+
},
|
7798 |
+
{
|
7799 |
+
"epoch": 0.8367491166077738,
|
7800 |
+
"grad_norm": 5.681154251098633,
|
7801 |
+
"learning_rate": 1.3100824363006326e-05,
|
7802 |
+
"loss": 0.443,
|
7803 |
+
"step": 1110
|
7804 |
+
},
|
7805 |
+
{
|
7806 |
+
"epoch": 0.8375029446407538,
|
7807 |
+
"grad_norm": 5.729828357696533,
|
7808 |
+
"learning_rate": 1.2983035000221177e-05,
|
7809 |
+
"loss": 0.4053,
|
7810 |
+
"step": 1111
|
7811 |
+
},
|
7812 |
+
{
|
7813 |
+
"epoch": 0.8382567726737338,
|
7814 |
+
"grad_norm": 6.101329326629639,
|
7815 |
+
"learning_rate": 1.2865740781892699e-05,
|
7816 |
+
"loss": 0.4384,
|
7817 |
+
"step": 1112
|
7818 |
+
},
|
7819 |
+
{
|
7820 |
+
"epoch": 0.8390106007067137,
|
7821 |
+
"grad_norm": 5.694645881652832,
|
7822 |
+
"learning_rate": 1.2748942375449135e-05,
|
7823 |
+
"loss": 0.4013,
|
7824 |
+
"step": 1113
|
7825 |
+
},
|
7826 |
+
{
|
7827 |
+
"epoch": 0.8397644287396937,
|
7828 |
+
"grad_norm": 5.564671516418457,
|
7829 |
+
"learning_rate": 1.263264044549748e-05,
|
7830 |
+
"loss": 0.4148,
|
7831 |
+
"step": 1114
|
7832 |
+
},
|
7833 |
+
{
|
7834 |
+
"epoch": 0.8405182567726738,
|
7835 |
+
"grad_norm": 5.393068313598633,
|
7836 |
+
"learning_rate": 1.2516835653819725e-05,
|
7837 |
+
"loss": 0.3981,
|
7838 |
+
"step": 1115
|
7839 |
+
},
|
7840 |
+
{
|
7841 |
+
"epoch": 0.8412720848056537,
|
7842 |
+
"grad_norm": 5.637123107910156,
|
7843 |
+
"learning_rate": 1.2401528659368911e-05,
|
7844 |
+
"loss": 0.406,
|
7845 |
+
"step": 1116
|
7846 |
+
},
|
7847 |
+
{
|
7848 |
+
"epoch": 0.8420259128386337,
|
7849 |
+
"grad_norm": 5.908216953277588,
|
7850 |
+
"learning_rate": 1.2286720118265659e-05,
|
7851 |
+
"loss": 0.3637,
|
7852 |
+
"step": 1117
|
7853 |
+
},
|
7854 |
+
{
|
7855 |
+
"epoch": 0.8427797408716137,
|
7856 |
+
"grad_norm": 5.7352070808410645,
|
7857 |
+
"learning_rate": 1.2172410683794177e-05,
|
7858 |
+
"loss": 0.4082,
|
7859 |
+
"step": 1118
|
7860 |
+
},
|
7861 |
+
{
|
7862 |
+
"epoch": 0.8435335689045936,
|
7863 |
+
"grad_norm": 5.5727858543396,
|
7864 |
+
"learning_rate": 1.2058601006398718e-05,
|
7865 |
+
"loss": 0.3828,
|
7866 |
+
"step": 1119
|
7867 |
+
},
|
7868 |
+
{
|
7869 |
+
"epoch": 0.8442873969375736,
|
7870 |
+
"grad_norm": 6.22990608215332,
|
7871 |
+
"learning_rate": 1.1945291733679764e-05,
|
7872 |
+
"loss": 0.4306,
|
7873 |
+
"step": 1120
|
7874 |
+
},
|
7875 |
+
{
|
7876 |
+
"epoch": 0.8450412249705536,
|
7877 |
+
"grad_norm": 5.981517314910889,
|
7878 |
+
"learning_rate": 1.1832483510390469e-05,
|
7879 |
+
"loss": 0.4177,
|
7880 |
+
"step": 1121
|
7881 |
+
},
|
7882 |
+
{
|
7883 |
+
"epoch": 0.8457950530035335,
|
7884 |
+
"grad_norm": 5.5717973709106445,
|
7885 |
+
"learning_rate": 1.1720176978432795e-05,
|
7886 |
+
"loss": 0.375,
|
7887 |
+
"step": 1122
|
7888 |
+
},
|
7889 |
+
{
|
7890 |
+
"epoch": 0.8465488810365136,
|
7891 |
+
"grad_norm": 5.83533239364624,
|
7892 |
+
"learning_rate": 1.1608372776854103e-05,
|
7893 |
+
"loss": 0.4141,
|
7894 |
+
"step": 1123
|
7895 |
+
},
|
7896 |
+
{
|
7897 |
+
"epoch": 0.8473027090694936,
|
7898 |
+
"grad_norm": 5.770301342010498,
|
7899 |
+
"learning_rate": 1.1497071541843306e-05,
|
7900 |
+
"loss": 0.3698,
|
7901 |
+
"step": 1124
|
7902 |
+
},
|
7903 |
+
{
|
7904 |
+
"epoch": 0.8480565371024735,
|
7905 |
+
"grad_norm": 5.999599933624268,
|
7906 |
+
"learning_rate": 1.1386273906727363e-05,
|
7907 |
+
"loss": 0.4177,
|
7908 |
+
"step": 1125
|
7909 |
+
},
|
7910 |
+
{
|
7911 |
+
"epoch": 0.8488103651354535,
|
7912 |
+
"grad_norm": 5.716385841369629,
|
7913 |
+
"learning_rate": 1.1275980501967642e-05,
|
7914 |
+
"loss": 0.3931,
|
7915 |
+
"step": 1126
|
7916 |
+
},
|
7917 |
+
{
|
7918 |
+
"epoch": 0.8495641931684335,
|
7919 |
+
"grad_norm": 6.15166711807251,
|
7920 |
+
"learning_rate": 1.1166191955156346e-05,
|
7921 |
+
"loss": 0.4025,
|
7922 |
+
"step": 1127
|
7923 |
+
},
|
7924 |
+
{
|
7925 |
+
"epoch": 0.8503180212014134,
|
7926 |
+
"grad_norm": 6.117612361907959,
|
7927 |
+
"learning_rate": 1.1056908891012884e-05,
|
7928 |
+
"loss": 0.4186,
|
7929 |
+
"step": 1128
|
7930 |
+
},
|
7931 |
+
{
|
7932 |
+
"epoch": 0.8510718492343934,
|
7933 |
+
"grad_norm": 6.109333038330078,
|
7934 |
+
"learning_rate": 1.0948131931380457e-05,
|
7935 |
+
"loss": 0.3863,
|
7936 |
+
"step": 1129
|
7937 |
+
},
|
7938 |
+
{
|
7939 |
+
"epoch": 0.8518256772673733,
|
7940 |
+
"grad_norm": 5.863979816436768,
|
7941 |
+
"learning_rate": 1.0839861695222354e-05,
|
7942 |
+
"loss": 0.3737,
|
7943 |
+
"step": 1130
|
7944 |
+
},
|
7945 |
+
{
|
7946 |
+
"epoch": 0.8525795053003533,
|
7947 |
+
"grad_norm": 5.980686664581299,
|
7948 |
+
"learning_rate": 1.0732098798618517e-05,
|
7949 |
+
"loss": 0.3739,
|
7950 |
+
"step": 1131
|
7951 |
+
},
|
7952 |
+
{
|
7953 |
+
"epoch": 0.8533333333333334,
|
7954 |
+
"grad_norm": 6.321891784667969,
|
7955 |
+
"learning_rate": 1.0624843854762034e-05,
|
7956 |
+
"loss": 0.416,
|
7957 |
+
"step": 1132
|
7958 |
+
},
|
7959 |
+
{
|
7960 |
+
"epoch": 0.8540871613663134,
|
7961 |
+
"grad_norm": 6.081487655639648,
|
7962 |
+
"learning_rate": 1.0518097473955624e-05,
|
7963 |
+
"loss": 0.3922,
|
7964 |
+
"step": 1133
|
7965 |
+
},
|
7966 |
+
{
|
7967 |
+
"epoch": 0.8548409893992933,
|
7968 |
+
"grad_norm": 6.287003040313721,
|
7969 |
+
"learning_rate": 1.0411860263608186e-05,
|
7970 |
+
"loss": 0.3747,
|
7971 |
+
"step": 1134
|
7972 |
+
},
|
7973 |
+
{
|
7974 |
+
"epoch": 0.8555948174322733,
|
7975 |
+
"grad_norm": 6.175232887268066,
|
7976 |
+
"learning_rate": 1.0306132828231318e-05,
|
7977 |
+
"loss": 0.3708,
|
7978 |
+
"step": 1135
|
7979 |
+
},
|
7980 |
+
{
|
7981 |
+
"epoch": 0.8563486454652532,
|
7982 |
+
"grad_norm": 6.49648904800415,
|
7983 |
+
"learning_rate": 1.0200915769435937e-05,
|
7984 |
+
"loss": 0.373,
|
7985 |
+
"step": 1136
|
7986 |
+
},
|
7987 |
+
{
|
7988 |
+
"epoch": 0.8571024734982332,
|
7989 |
+
"grad_norm": 6.249892234802246,
|
7990 |
+
"learning_rate": 1.009620968592876e-05,
|
7991 |
+
"loss": 0.3807,
|
7992 |
+
"step": 1137
|
7993 |
+
},
|
7994 |
+
{
|
7995 |
+
"epoch": 0.8578563015312132,
|
7996 |
+
"grad_norm": 6.616731643676758,
|
7997 |
+
"learning_rate": 9.992015173508995e-06,
|
7998 |
+
"loss": 0.3981,
|
7999 |
+
"step": 1138
|
8000 |
+
},
|
8001 |
+
{
|
8002 |
+
"epoch": 0.8586101295641931,
|
8003 |
+
"grad_norm": 6.801102638244629,
|
8004 |
+
"learning_rate": 9.88833282506486e-06,
|
8005 |
+
"loss": 0.3968,
|
8006 |
+
"step": 1139
|
8007 |
+
},
|
8008 |
+
{
|
8009 |
+
"epoch": 0.8593639575971731,
|
8010 |
+
"grad_norm": 6.820323467254639,
|
8011 |
+
"learning_rate": 9.785163230570282e-06,
|
8012 |
+
"loss": 0.3939,
|
8013 |
+
"step": 1140
|
8014 |
+
},
|
8015 |
+
{
|
8016 |
+
"epoch": 0.8601177856301532,
|
8017 |
+
"grad_norm": 8.20490837097168,
|
8018 |
+
"learning_rate": 9.682506977081496e-06,
|
8019 |
+
"loss": 0.4353,
|
8020 |
+
"step": 1141
|
8021 |
+
},
|
8022 |
+
{
|
8023 |
+
"epoch": 0.8608716136631331,
|
8024 |
+
"grad_norm": 7.587864398956299,
|
8025 |
+
"learning_rate": 9.580364648733775e-06,
|
8026 |
+
"loss": 0.4369,
|
8027 |
+
"step": 1142
|
8028 |
+
},
|
8029 |
+
{
|
8030 |
+
"epoch": 0.8616254416961131,
|
8031 |
+
"grad_norm": 7.294688701629639,
|
8032 |
+
"learning_rate": 9.478736826737944e-06,
|
8033 |
+
"loss": 0.411,
|
8034 |
+
"step": 1143
|
8035 |
+
},
|
8036 |
+
{
|
8037 |
+
"epoch": 0.8623792697290931,
|
8038 |
+
"grad_norm": 7.802835464477539,
|
8039 |
+
"learning_rate": 9.37762408937729e-06,
|
8040 |
+
"loss": 0.424,
|
8041 |
+
"step": 1144
|
8042 |
+
},
|
8043 |
+
{
|
8044 |
+
"epoch": 0.863133097762073,
|
8045 |
+
"grad_norm": 8.21778678894043,
|
8046 |
+
"learning_rate": 9.277027012004125e-06,
|
8047 |
+
"loss": 0.4752,
|
8048 |
+
"step": 1145
|
8049 |
+
},
|
8050 |
+
{
|
8051 |
+
"epoch": 0.863886925795053,
|
8052 |
+
"grad_norm": 8.805744171142578,
|
8053 |
+
"learning_rate": 9.176946167036516e-06,
|
8054 |
+
"loss": 0.4736,
|
8055 |
+
"step": 1146
|
8056 |
+
},
|
8057 |
+
{
|
8058 |
+
"epoch": 0.864640753828033,
|
8059 |
+
"grad_norm": 10.24565601348877,
|
8060 |
+
"learning_rate": 9.07738212395508e-06,
|
8061 |
+
"loss": 0.4635,
|
8062 |
+
"step": 1147
|
8063 |
+
},
|
8064 |
+
{
|
8065 |
+
"epoch": 0.8653945818610129,
|
8066 |
+
"grad_norm": 9.218001365661621,
|
8067 |
+
"learning_rate": 8.978335449299791e-06,
|
8068 |
+
"loss": 0.4313,
|
8069 |
+
"step": 1148
|
8070 |
+
},
|
8071 |
+
{
|
8072 |
+
"epoch": 0.8661484098939929,
|
8073 |
+
"grad_norm": 10.276748657226562,
|
8074 |
+
"learning_rate": 8.87980670666655e-06,
|
8075 |
+
"loss": 0.421,
|
8076 |
+
"step": 1149
|
8077 |
+
},
|
8078 |
+
{
|
8079 |
+
"epoch": 0.866902237926973,
|
8080 |
+
"grad_norm": 11.982145309448242,
|
8081 |
+
"learning_rate": 8.781796456704262e-06,
|
8082 |
+
"loss": 0.4486,
|
8083 |
+
"step": 1150
|
8084 |
+
},
|
8085 |
+
{
|
8086 |
+
"epoch": 0.8676560659599529,
|
8087 |
+
"grad_norm": 5.365624904632568,
|
8088 |
+
"learning_rate": 8.684305257111425e-06,
|
8089 |
+
"loss": 0.5014,
|
8090 |
+
"step": 1151
|
8091 |
+
},
|
8092 |
+
{
|
8093 |
+
"epoch": 0.8684098939929329,
|
8094 |
+
"grad_norm": 5.599196910858154,
|
8095 |
+
"learning_rate": 8.587333662633035e-06,
|
8096 |
+
"loss": 0.4984,
|
8097 |
+
"step": 1152
|
8098 |
+
},
|
8099 |
+
{
|
8100 |
+
"epoch": 0.8691637220259129,
|
8101 |
+
"grad_norm": 5.679477214813232,
|
8102 |
+
"learning_rate": 8.490882225057428e-06,
|
8103 |
+
"loss": 0.5011,
|
8104 |
+
"step": 1153
|
8105 |
+
},
|
8106 |
+
{
|
8107 |
+
"epoch": 0.8699175500588928,
|
8108 |
+
"grad_norm": 5.679898738861084,
|
8109 |
+
"learning_rate": 8.39495149321322e-06,
|
8110 |
+
"loss": 0.443,
|
8111 |
+
"step": 1154
|
8112 |
+
},
|
8113 |
+
{
|
8114 |
+
"epoch": 0.8706713780918728,
|
8115 |
+
"grad_norm": 5.414709091186523,
|
8116 |
+
"learning_rate": 8.299542012965944e-06,
|
8117 |
+
"loss": 0.4269,
|
8118 |
+
"step": 1155
|
8119 |
+
},
|
8120 |
+
{
|
8121 |
+
"epoch": 0.8714252061248527,
|
8122 |
+
"grad_norm": 5.3179426193237305,
|
8123 |
+
"learning_rate": 8.204654327215267e-06,
|
8124 |
+
"loss": 0.4395,
|
8125 |
+
"step": 1156
|
8126 |
+
},
|
8127 |
+
{
|
8128 |
+
"epoch": 0.8721790341578327,
|
8129 |
+
"grad_norm": 5.2444963455200195,
|
8130 |
+
"learning_rate": 8.110288975891634e-06,
|
8131 |
+
"loss": 0.4217,
|
8132 |
+
"step": 1157
|
8133 |
+
},
|
8134 |
+
{
|
8135 |
+
"epoch": 0.8729328621908127,
|
8136 |
+
"grad_norm": 5.733283996582031,
|
8137 |
+
"learning_rate": 8.016446495953367e-06,
|
8138 |
+
"loss": 0.4395,
|
8139 |
+
"step": 1158
|
8140 |
+
},
|
8141 |
+
{
|
8142 |
+
"epoch": 0.8736866902237928,
|
8143 |
+
"grad_norm": 5.545217037200928,
|
8144 |
+
"learning_rate": 7.923127421383458e-06,
|
8145 |
+
"loss": 0.436,
|
8146 |
+
"step": 1159
|
8147 |
+
},
|
8148 |
+
{
|
8149 |
+
"epoch": 0.8744405182567727,
|
8150 |
+
"grad_norm": 5.759894371032715,
|
8151 |
+
"learning_rate": 7.830332283186714e-06,
|
8152 |
+
"loss": 0.4376,
|
8153 |
+
"step": 1160
|
8154 |
+
},
|
8155 |
+
{
|
8156 |
+
"epoch": 0.8751943462897527,
|
8157 |
+
"grad_norm": 5.31406831741333,
|
8158 |
+
"learning_rate": 7.73806160938656e-06,
|
8159 |
+
"loss": 0.4097,
|
8160 |
+
"step": 1161
|
8161 |
+
},
|
8162 |
+
{
|
8163 |
+
"epoch": 0.8759481743227326,
|
8164 |
+
"grad_norm": 5.372743129730225,
|
8165 |
+
"learning_rate": 7.646315925022152e-06,
|
8166 |
+
"loss": 0.4264,
|
8167 |
+
"step": 1162
|
8168 |
+
},
|
8169 |
+
{
|
8170 |
+
"epoch": 0.8767020023557126,
|
8171 |
+
"grad_norm": 5.223913192749023,
|
8172 |
+
"learning_rate": 7.555095752145313e-06,
|
8173 |
+
"loss": 0.3879,
|
8174 |
+
"step": 1163
|
8175 |
+
},
|
8176 |
+
{
|
8177 |
+
"epoch": 0.8774558303886926,
|
8178 |
+
"grad_norm": 5.493069171905518,
|
8179 |
+
"learning_rate": 7.4644016098176615e-06,
|
8180 |
+
"loss": 0.4099,
|
8181 |
+
"step": 1164
|
8182 |
+
},
|
8183 |
+
{
|
8184 |
+
"epoch": 0.8782096584216725,
|
8185 |
+
"grad_norm": 5.413908004760742,
|
8186 |
+
"learning_rate": 7.374234014107484e-06,
|
8187 |
+
"loss": 0.4041,
|
8188 |
+
"step": 1165
|
8189 |
+
},
|
8190 |
+
{
|
8191 |
+
"epoch": 0.8789634864546525,
|
8192 |
+
"grad_norm": 5.9703288078308105,
|
8193 |
+
"learning_rate": 7.284593478087043e-06,
|
8194 |
+
"loss": 0.4391,
|
8195 |
+
"step": 1166
|
8196 |
+
},
|
8197 |
+
{
|
8198 |
+
"epoch": 0.8797173144876325,
|
8199 |
+
"grad_norm": 6.033265590667725,
|
8200 |
+
"learning_rate": 7.195480511829411e-06,
|
8201 |
+
"loss": 0.4356,
|
8202 |
+
"step": 1167
|
8203 |
+
},
|
8204 |
+
{
|
8205 |
+
"epoch": 0.8804711425206125,
|
8206 |
+
"grad_norm": 5.589619159698486,
|
8207 |
+
"learning_rate": 7.106895622405752e-06,
|
8208 |
+
"loss": 0.4029,
|
8209 |
+
"step": 1168
|
8210 |
+
},
|
8211 |
+
{
|
8212 |
+
"epoch": 0.8812249705535925,
|
8213 |
+
"grad_norm": 5.580582141876221,
|
8214 |
+
"learning_rate": 7.018839313882286e-06,
|
8215 |
+
"loss": 0.4039,
|
8216 |
+
"step": 1169
|
8217 |
+
},
|
8218 |
+
{
|
8219 |
+
"epoch": 0.8819787985865725,
|
8220 |
+
"grad_norm": 5.605942726135254,
|
8221 |
+
"learning_rate": 6.931312087317632e-06,
|
8222 |
+
"loss": 0.3915,
|
8223 |
+
"step": 1170
|
8224 |
+
},
|
8225 |
+
{
|
8226 |
+
"epoch": 0.8827326266195524,
|
8227 |
+
"grad_norm": 5.954355239868164,
|
8228 |
+
"learning_rate": 6.844314440759647e-06,
|
8229 |
+
"loss": 0.4119,
|
8230 |
+
"step": 1171
|
8231 |
+
},
|
8232 |
+
{
|
8233 |
+
"epoch": 0.8834864546525324,
|
8234 |
+
"grad_norm": 5.943442344665527,
|
8235 |
+
"learning_rate": 6.7578468692429345e-06,
|
8236 |
+
"loss": 0.4227,
|
8237 |
+
"step": 1172
|
8238 |
+
},
|
8239 |
+
{
|
8240 |
+
"epoch": 0.8842402826855124,
|
8241 |
+
"grad_norm": 6.070568561553955,
|
8242 |
+
"learning_rate": 6.6719098647857525e-06,
|
8243 |
+
"loss": 0.3824,
|
8244 |
+
"step": 1173
|
8245 |
+
},
|
8246 |
+
{
|
8247 |
+
"epoch": 0.8849941107184923,
|
8248 |
+
"grad_norm": 5.827738285064697,
|
8249 |
+
"learning_rate": 6.586503916387366e-06,
|
8250 |
+
"loss": 0.4358,
|
8251 |
+
"step": 1174
|
8252 |
+
},
|
8253 |
+
{
|
8254 |
+
"epoch": 0.8857479387514723,
|
8255 |
+
"grad_norm": 5.9503655433654785,
|
8256 |
+
"learning_rate": 6.501629510025231e-06,
|
8257 |
+
"loss": 0.3862,
|
8258 |
+
"step": 1175
|
8259 |
+
},
|
8260 |
+
{
|
8261 |
+
"epoch": 0.8865017667844522,
|
8262 |
+
"grad_norm": 5.86431360244751,
|
8263 |
+
"learning_rate": 6.417287128652172e-06,
|
8264 |
+
"loss": 0.3849,
|
8265 |
+
"step": 1176
|
8266 |
+
},
|
8267 |
+
{
|
8268 |
+
"epoch": 0.8872555948174323,
|
8269 |
+
"grad_norm": 5.833621978759766,
|
8270 |
+
"learning_rate": 6.333477252193731e-06,
|
8271 |
+
"loss": 0.3935,
|
8272 |
+
"step": 1177
|
8273 |
+
},
|
8274 |
+
{
|
8275 |
+
"epoch": 0.8880094228504123,
|
8276 |
+
"grad_norm": 6.094554901123047,
|
8277 |
+
"learning_rate": 6.250200357545377e-06,
|
8278 |
+
"loss": 0.3911,
|
8279 |
+
"step": 1178
|
8280 |
+
},
|
8281 |
+
{
|
8282 |
+
"epoch": 0.8887632508833923,
|
8283 |
+
"grad_norm": 5.814612865447998,
|
8284 |
+
"learning_rate": 6.167456918569792e-06,
|
8285 |
+
"loss": 0.3738,
|
8286 |
+
"step": 1179
|
8287 |
+
},
|
8288 |
+
{
|
8289 |
+
"epoch": 0.8895170789163722,
|
8290 |
+
"grad_norm": 6.395360946655273,
|
8291 |
+
"learning_rate": 6.085247406094197e-06,
|
8292 |
+
"loss": 0.3692,
|
8293 |
+
"step": 1180
|
8294 |
+
},
|
8295 |
+
{
|
8296 |
+
"epoch": 0.8902709069493522,
|
8297 |
+
"grad_norm": 5.914385795593262,
|
8298 |
+
"learning_rate": 6.003572287907633e-06,
|
8299 |
+
"loss": 0.4008,
|
8300 |
+
"step": 1181
|
8301 |
+
},
|
8302 |
+
{
|
8303 |
+
"epoch": 0.8910247349823321,
|
8304 |
+
"grad_norm": 6.416135787963867,
|
8305 |
+
"learning_rate": 5.922432028758362e-06,
|
8306 |
+
"loss": 0.3997,
|
8307 |
+
"step": 1182
|
8308 |
+
},
|
8309 |
+
{
|
8310 |
+
"epoch": 0.8917785630153121,
|
8311 |
+
"grad_norm": 5.680757522583008,
|
8312 |
+
"learning_rate": 5.841827090351171e-06,
|
8313 |
+
"loss": 0.347,
|
8314 |
+
"step": 1183
|
8315 |
+
},
|
8316 |
+
{
|
8317 |
+
"epoch": 0.8925323910482921,
|
8318 |
+
"grad_norm": 5.837109088897705,
|
8319 |
+
"learning_rate": 5.761757931344758e-06,
|
8320 |
+
"loss": 0.3623,
|
8321 |
+
"step": 1184
|
8322 |
+
},
|
8323 |
+
{
|
8324 |
+
"epoch": 0.893286219081272,
|
8325 |
+
"grad_norm": 5.914787769317627,
|
8326 |
+
"learning_rate": 5.68222500734914e-06,
|
8327 |
+
"loss": 0.3632,
|
8328 |
+
"step": 1185
|
8329 |
+
},
|
8330 |
+
{
|
8331 |
+
"epoch": 0.8940400471142521,
|
8332 |
+
"grad_norm": 6.179137229919434,
|
8333 |
+
"learning_rate": 5.603228770923041e-06,
|
8334 |
+
"loss": 0.3864,
|
8335 |
+
"step": 1186
|
8336 |
+
},
|
8337 |
+
{
|
8338 |
+
"epoch": 0.8947938751472321,
|
8339 |
+
"grad_norm": 5.854869365692139,
|
8340 |
+
"learning_rate": 5.524769671571317e-06,
|
8341 |
+
"loss": 0.3318,
|
8342 |
+
"step": 1187
|
8343 |
+
},
|
8344 |
+
{
|
8345 |
+
"epoch": 0.895547703180212,
|
8346 |
+
"grad_norm": 6.880571365356445,
|
8347 |
+
"learning_rate": 5.446848155742401e-06,
|
8348 |
+
"loss": 0.4063,
|
8349 |
+
"step": 1188
|
8350 |
+
},
|
8351 |
+
{
|
8352 |
+
"epoch": 0.896301531213192,
|
8353 |
+
"grad_norm": 6.602806568145752,
|
8354 |
+
"learning_rate": 5.3694646668257855e-06,
|
8355 |
+
"loss": 0.3698,
|
8356 |
+
"step": 1189
|
8357 |
+
},
|
8358 |
+
{
|
8359 |
+
"epoch": 0.897055359246172,
|
8360 |
+
"grad_norm": 7.17775821685791,
|
8361 |
+
"learning_rate": 5.292619645149433e-06,
|
8362 |
+
"loss": 0.4266,
|
8363 |
+
"step": 1190
|
8364 |
+
},
|
8365 |
+
{
|
8366 |
+
"epoch": 0.8978091872791519,
|
8367 |
+
"grad_norm": 7.022253036499023,
|
8368 |
+
"learning_rate": 5.2163135279773904e-06,
|
8369 |
+
"loss": 0.3885,
|
8370 |
+
"step": 1191
|
8371 |
+
},
|
8372 |
+
{
|
8373 |
+
"epoch": 0.8985630153121319,
|
8374 |
+
"grad_norm": 7.834957599639893,
|
8375 |
+
"learning_rate": 5.140546749507136e-06,
|
8376 |
+
"loss": 0.4484,
|
8377 |
+
"step": 1192
|
8378 |
+
},
|
8379 |
+
{
|
8380 |
+
"epoch": 0.8993168433451119,
|
8381 |
+
"grad_norm": 8.505350112915039,
|
8382 |
+
"learning_rate": 5.06531974086728e-06,
|
8383 |
+
"loss": 0.4535,
|
8384 |
+
"step": 1193
|
8385 |
+
},
|
8386 |
+
{
|
8387 |
+
"epoch": 0.9000706713780918,
|
8388 |
+
"grad_norm": 8.074254035949707,
|
8389 |
+
"learning_rate": 4.9906329301149914e-06,
|
8390 |
+
"loss": 0.4528,
|
8391 |
+
"step": 1194
|
8392 |
+
},
|
8393 |
+
{
|
8394 |
+
"epoch": 0.9008244994110719,
|
8395 |
+
"grad_norm": 8.195548057556152,
|
8396 |
+
"learning_rate": 4.916486742233606e-06,
|
8397 |
+
"loss": 0.447,
|
8398 |
+
"step": 1195
|
8399 |
+
},
|
8400 |
+
{
|
8401 |
+
"epoch": 0.9015783274440519,
|
8402 |
+
"grad_norm": 9.020340919494629,
|
8403 |
+
"learning_rate": 4.8428815991302005e-06,
|
8404 |
+
"loss": 0.4507,
|
8405 |
+
"step": 1196
|
8406 |
+
},
|
8407 |
+
{
|
8408 |
+
"epoch": 0.9023321554770318,
|
8409 |
+
"grad_norm": 9.480902671813965,
|
8410 |
+
"learning_rate": 4.769817919633235e-06,
|
8411 |
+
"loss": 0.4905,
|
8412 |
+
"step": 1197
|
8413 |
+
},
|
8414 |
+
{
|
8415 |
+
"epoch": 0.9030859835100118,
|
8416 |
+
"grad_norm": 9.953953742980957,
|
8417 |
+
"learning_rate": 4.697296119490047e-06,
|
8418 |
+
"loss": 0.4291,
|
8419 |
+
"step": 1198
|
8420 |
+
},
|
8421 |
+
{
|
8422 |
+
"epoch": 0.9038398115429918,
|
8423 |
+
"grad_norm": 9.974310874938965,
|
8424 |
+
"learning_rate": 4.625316611364661e-06,
|
8425 |
+
"loss": 0.4283,
|
8426 |
+
"step": 1199
|
8427 |
+
},
|
8428 |
+
{
|
8429 |
+
"epoch": 0.9045936395759717,
|
8430 |
+
"grad_norm": 12.497854232788086,
|
8431 |
+
"learning_rate": 4.553879804835282e-06,
|
8432 |
+
"loss": 0.4614,
|
8433 |
+
"step": 1200
|
8434 |
+
},
|
8435 |
+
{
|
8436 |
+
"epoch": 0.9053474676089517,
|
8437 |
+
"grad_norm": 4.8798136711120605,
|
8438 |
+
"learning_rate": 4.482986106392073e-06,
|
8439 |
+
"loss": 0.4771,
|
8440 |
+
"step": 1201
|
8441 |
+
},
|
8442 |
+
{
|
8443 |
+
"epoch": 0.9061012956419316,
|
8444 |
+
"grad_norm": 4.956184387207031,
|
8445 |
+
"learning_rate": 4.412635919434749e-06,
|
8446 |
+
"loss": 0.4444,
|
8447 |
+
"step": 1202
|
8448 |
+
},
|
8449 |
+
{
|
8450 |
+
"epoch": 0.9068551236749116,
|
8451 |
+
"grad_norm": 5.346173286437988,
|
8452 |
+
"learning_rate": 4.342829644270429e-06,
|
8453 |
+
"loss": 0.4442,
|
8454 |
+
"step": 1203
|
8455 |
+
},
|
8456 |
+
{
|
8457 |
+
"epoch": 0.9076089517078917,
|
8458 |
+
"grad_norm": 5.293701648712158,
|
8459 |
+
"learning_rate": 4.273567678111123e-06,
|
8460 |
+
"loss": 0.4614,
|
8461 |
+
"step": 1204
|
8462 |
+
},
|
8463 |
+
{
|
8464 |
+
"epoch": 0.9083627797408717,
|
8465 |
+
"grad_norm": 5.237243175506592,
|
8466 |
+
"learning_rate": 4.204850415071748e-06,
|
8467 |
+
"loss": 0.4512,
|
8468 |
+
"step": 1205
|
8469 |
+
},
|
8470 |
+
{
|
8471 |
+
"epoch": 0.9091166077738516,
|
8472 |
+
"grad_norm": 5.3798604011535645,
|
8473 |
+
"learning_rate": 4.136678246167636e-06,
|
8474 |
+
"loss": 0.4286,
|
8475 |
+
"step": 1206
|
8476 |
+
},
|
8477 |
+
{
|
8478 |
+
"epoch": 0.9098704358068316,
|
8479 |
+
"grad_norm": 5.367835998535156,
|
8480 |
+
"learning_rate": 4.069051559312531e-06,
|
8481 |
+
"loss": 0.4139,
|
8482 |
+
"step": 1207
|
8483 |
+
},
|
8484 |
+
{
|
8485 |
+
"epoch": 0.9106242638398115,
|
8486 |
+
"grad_norm": 5.50463342666626,
|
8487 |
+
"learning_rate": 4.001970739316163e-06,
|
8488 |
+
"loss": 0.4407,
|
8489 |
+
"step": 1208
|
8490 |
+
},
|
8491 |
+
{
|
8492 |
+
"epoch": 0.9113780918727915,
|
8493 |
+
"grad_norm": 5.295793056488037,
|
8494 |
+
"learning_rate": 3.935436167882234e-06,
|
8495 |
+
"loss": 0.418,
|
8496 |
+
"step": 1209
|
8497 |
+
},
|
8498 |
+
{
|
8499 |
+
"epoch": 0.9121319199057715,
|
8500 |
+
"grad_norm": 5.284564018249512,
|
8501 |
+
"learning_rate": 3.869448223606165e-06,
|
8502 |
+
"loss": 0.4096,
|
8503 |
+
"step": 1210
|
8504 |
+
},
|
8505 |
+
{
|
8506 |
+
"epoch": 0.9128857479387514,
|
8507 |
+
"grad_norm": 5.553956031799316,
|
8508 |
+
"learning_rate": 3.8040072819729545e-06,
|
8509 |
+
"loss": 0.4141,
|
8510 |
+
"step": 1211
|
8511 |
+
},
|
8512 |
+
{
|
8513 |
+
"epoch": 0.9136395759717314,
|
8514 |
+
"grad_norm": 5.626007080078125,
|
8515 |
+
"learning_rate": 3.7391137153550137e-06,
|
8516 |
+
"loss": 0.4138,
|
8517 |
+
"step": 1212
|
8518 |
+
},
|
8519 |
+
{
|
8520 |
+
"epoch": 0.9143934040047115,
|
8521 |
+
"grad_norm": 5.603013038635254,
|
8522 |
+
"learning_rate": 3.6747678930101558e-06,
|
8523 |
+
"loss": 0.4148,
|
8524 |
+
"step": 1213
|
8525 |
+
},
|
8526 |
+
{
|
8527 |
+
"epoch": 0.9151472320376914,
|
8528 |
+
"grad_norm": 5.539734363555908,
|
8529 |
+
"learning_rate": 3.6109701810793208e-06,
|
8530 |
+
"loss": 0.4181,
|
8531 |
+
"step": 1214
|
8532 |
+
},
|
8533 |
+
{
|
8534 |
+
"epoch": 0.9159010600706714,
|
8535 |
+
"grad_norm": 5.379584789276123,
|
8536 |
+
"learning_rate": 3.5477209425846538e-06,
|
8537 |
+
"loss": 0.4015,
|
8538 |
+
"step": 1215
|
8539 |
+
},
|
8540 |
+
{
|
8541 |
+
"epoch": 0.9166548881036514,
|
8542 |
+
"grad_norm": 5.433023929595947,
|
8543 |
+
"learning_rate": 3.4850205374273416e-06,
|
8544 |
+
"loss": 0.398,
|
8545 |
+
"step": 1216
|
8546 |
+
},
|
8547 |
+
{
|
8548 |
+
"epoch": 0.9174087161366313,
|
8549 |
+
"grad_norm": 5.5849199295043945,
|
8550 |
+
"learning_rate": 3.4228693223856136e-06,
|
8551 |
+
"loss": 0.4165,
|
8552 |
+
"step": 1217
|
8553 |
+
},
|
8554 |
+
{
|
8555 |
+
"epoch": 0.9181625441696113,
|
8556 |
+
"grad_norm": 5.703511714935303,
|
8557 |
+
"learning_rate": 3.361267651112676e-06,
|
8558 |
+
"loss": 0.422,
|
8559 |
+
"step": 1218
|
8560 |
+
},
|
8561 |
+
{
|
8562 |
+
"epoch": 0.9189163722025913,
|
8563 |
+
"grad_norm": 5.733764171600342,
|
8564 |
+
"learning_rate": 3.30021587413476e-06,
|
8565 |
+
"loss": 0.4017,
|
8566 |
+
"step": 1219
|
8567 |
+
},
|
8568 |
+
{
|
8569 |
+
"epoch": 0.9196702002355712,
|
8570 |
+
"grad_norm": 5.802048206329346,
|
8571 |
+
"learning_rate": 3.2397143388489983e-06,
|
8572 |
+
"loss": 0.3935,
|
8573 |
+
"step": 1220
|
8574 |
+
},
|
8575 |
+
{
|
8576 |
+
"epoch": 0.9204240282685512,
|
8577 |
+
"grad_norm": 5.458968639373779,
|
8578 |
+
"learning_rate": 3.1797633895216394e-06,
|
8579 |
+
"loss": 0.3783,
|
8580 |
+
"step": 1221
|
8581 |
+
},
|
8582 |
+
{
|
8583 |
+
"epoch": 0.9211778563015313,
|
8584 |
+
"grad_norm": 5.353023052215576,
|
8585 |
+
"learning_rate": 3.120363367285917e-06,
|
8586 |
+
"loss": 0.3788,
|
8587 |
+
"step": 1222
|
8588 |
+
},
|
8589 |
+
{
|
8590 |
+
"epoch": 0.9219316843345112,
|
8591 |
+
"grad_norm": 5.518474578857422,
|
8592 |
+
"learning_rate": 3.0615146101401925e-06,
|
8593 |
+
"loss": 0.3944,
|
8594 |
+
"step": 1223
|
8595 |
+
},
|
8596 |
+
{
|
8597 |
+
"epoch": 0.9226855123674912,
|
8598 |
+
"grad_norm": 5.713134765625,
|
8599 |
+
"learning_rate": 3.0032174529460165e-06,
|
8600 |
+
"loss": 0.3953,
|
8601 |
+
"step": 1224
|
8602 |
+
},
|
8603 |
+
{
|
8604 |
+
"epoch": 0.9234393404004712,
|
8605 |
+
"grad_norm": 6.142655372619629,
|
8606 |
+
"learning_rate": 2.945472227426227e-06,
|
8607 |
+
"loss": 0.4168,
|
8608 |
+
"step": 1225
|
8609 |
+
},
|
8610 |
+
{
|
8611 |
+
"epoch": 0.9241931684334511,
|
8612 |
+
"grad_norm": 5.580604553222656,
|
8613 |
+
"learning_rate": 2.8882792621630406e-06,
|
8614 |
+
"loss": 0.3642,
|
8615 |
+
"step": 1226
|
8616 |
+
},
|
8617 |
+
{
|
8618 |
+
"epoch": 0.9249469964664311,
|
8619 |
+
"grad_norm": 5.7619757652282715,
|
8620 |
+
"learning_rate": 2.8316388825962324e-06,
|
8621 |
+
"loss": 0.3708,
|
8622 |
+
"step": 1227
|
8623 |
+
},
|
8624 |
+
{
|
8625 |
+
"epoch": 0.925700824499411,
|
8626 |
+
"grad_norm": 6.232563018798828,
|
8627 |
+
"learning_rate": 2.7755514110212264e-06,
|
8628 |
+
"loss": 0.4063,
|
8629 |
+
"step": 1228
|
8630 |
+
},
|
8631 |
+
{
|
8632 |
+
"epoch": 0.926454652532391,
|
8633 |
+
"grad_norm": 5.895346164703369,
|
8634 |
+
"learning_rate": 2.7200171665872742e-06,
|
8635 |
+
"loss": 0.399,
|
8636 |
+
"step": 1229
|
8637 |
+
},
|
8638 |
+
{
|
8639 |
+
"epoch": 0.927208480565371,
|
8640 |
+
"grad_norm": 5.760490894317627,
|
8641 |
+
"learning_rate": 2.6650364652956894e-06,
|
8642 |
+
"loss": 0.3785,
|
8643 |
+
"step": 1230
|
8644 |
+
},
|
8645 |
+
{
|
8646 |
+
"epoch": 0.927962308598351,
|
8647 |
+
"grad_norm": 5.620173454284668,
|
8648 |
+
"learning_rate": 2.6106096199979614e-06,
|
8649 |
+
"loss": 0.3564,
|
8650 |
+
"step": 1231
|
8651 |
+
},
|
8652 |
+
{
|
8653 |
+
"epoch": 0.928716136631331,
|
8654 |
+
"grad_norm": 5.84246826171875,
|
8655 |
+
"learning_rate": 2.5567369403940776e-06,
|
8656 |
+
"loss": 0.3575,
|
8657 |
+
"step": 1232
|
8658 |
+
},
|
8659 |
+
{
|
8660 |
+
"epoch": 0.929469964664311,
|
8661 |
+
"grad_norm": 5.908325672149658,
|
8662 |
+
"learning_rate": 2.50341873303066e-06,
|
8663 |
+
"loss": 0.384,
|
8664 |
+
"step": 1233
|
8665 |
+
},
|
8666 |
+
{
|
8667 |
+
"epoch": 0.9302237926972909,
|
8668 |
+
"grad_norm": 5.850981712341309,
|
8669 |
+
"learning_rate": 2.4506553012993093e-06,
|
8670 |
+
"loss": 0.3704,
|
8671 |
+
"step": 1234
|
8672 |
+
},
|
8673 |
+
{
|
8674 |
+
"epoch": 0.9309776207302709,
|
8675 |
+
"grad_norm": 6.301943778991699,
|
8676 |
+
"learning_rate": 2.398446945434818e-06,
|
8677 |
+
"loss": 0.385,
|
8678 |
+
"step": 1235
|
8679 |
+
},
|
8680 |
+
{
|
8681 |
+
"epoch": 0.9317314487632509,
|
8682 |
+
"grad_norm": 6.557477951049805,
|
8683 |
+
"learning_rate": 2.346793962513483e-06,
|
8684 |
+
"loss": 0.3607,
|
8685 |
+
"step": 1236
|
8686 |
+
},
|
8687 |
+
{
|
8688 |
+
"epoch": 0.9324852767962308,
|
8689 |
+
"grad_norm": 6.442347049713135,
|
8690 |
+
"learning_rate": 2.2956966464514175e-06,
|
8691 |
+
"loss": 0.3829,
|
8692 |
+
"step": 1237
|
8693 |
+
},
|
8694 |
+
{
|
8695 |
+
"epoch": 0.9332391048292108,
|
8696 |
+
"grad_norm": 7.224841594696045,
|
8697 |
+
"learning_rate": 2.245155288002876e-06,
|
8698 |
+
"loss": 0.3964,
|
8699 |
+
"step": 1238
|
8700 |
+
},
|
8701 |
+
{
|
8702 |
+
"epoch": 0.9339929328621908,
|
8703 |
+
"grad_norm": 7.129518032073975,
|
8704 |
+
"learning_rate": 2.1951701747585982e-06,
|
8705 |
+
"loss": 0.3682,
|
8706 |
+
"step": 1239
|
8707 |
+
},
|
8708 |
+
{
|
8709 |
+
"epoch": 0.9347467608951708,
|
8710 |
+
"grad_norm": 6.685035228729248,
|
8711 |
+
"learning_rate": 2.1457415911442013e-06,
|
8712 |
+
"loss": 0.4049,
|
8713 |
+
"step": 1240
|
8714 |
+
},
|
8715 |
+
{
|
8716 |
+
"epoch": 0.9355005889281508,
|
8717 |
+
"grad_norm": 7.421708583831787,
|
8718 |
+
"learning_rate": 2.0968698184184565e-06,
|
8719 |
+
"loss": 0.4029,
|
8720 |
+
"step": 1241
|
8721 |
+
},
|
8722 |
+
{
|
8723 |
+
"epoch": 0.9362544169611308,
|
8724 |
+
"grad_norm": 7.260560989379883,
|
8725 |
+
"learning_rate": 2.04855513467187e-06,
|
8726 |
+
"loss": 0.4232,
|
8727 |
+
"step": 1242
|
8728 |
+
},
|
8729 |
+
{
|
8730 |
+
"epoch": 0.9370082449941107,
|
8731 |
+
"grad_norm": 8.069437980651855,
|
8732 |
+
"learning_rate": 2.000797814824906e-06,
|
8733 |
+
"loss": 0.4409,
|
8734 |
+
"step": 1243
|
8735 |
+
},
|
8736 |
+
{
|
8737 |
+
"epoch": 0.9377620730270907,
|
8738 |
+
"grad_norm": 7.945827960968018,
|
8739 |
+
"learning_rate": 1.9535981306265884e-06,
|
8740 |
+
"loss": 0.4244,
|
8741 |
+
"step": 1244
|
8742 |
+
},
|
8743 |
+
{
|
8744 |
+
"epoch": 0.9385159010600707,
|
8745 |
+
"grad_norm": 8.818882942199707,
|
8746 |
+
"learning_rate": 1.9069563506527998e-06,
|
8747 |
+
"loss": 0.4722,
|
8748 |
+
"step": 1245
|
8749 |
+
},
|
8750 |
+
{
|
8751 |
+
"epoch": 0.9392697290930506,
|
8752 |
+
"grad_norm": 8.6805419921875,
|
8753 |
+
"learning_rate": 1.8608727403049309e-06,
|
8754 |
+
"loss": 0.4574,
|
8755 |
+
"step": 1246
|
8756 |
+
},
|
8757 |
+
{
|
8758 |
+
"epoch": 0.9400235571260306,
|
8759 |
+
"grad_norm": 8.550375938415527,
|
8760 |
+
"learning_rate": 1.8153475618081673e-06,
|
8761 |
+
"loss": 0.4289,
|
8762 |
+
"step": 1247
|
8763 |
+
},
|
8764 |
+
{
|
8765 |
+
"epoch": 0.9407773851590105,
|
8766 |
+
"grad_norm": 9.816337585449219,
|
8767 |
+
"learning_rate": 1.7703810742101813e-06,
|
8768 |
+
"loss": 0.4884,
|
8769 |
+
"step": 1248
|
8770 |
+
},
|
8771 |
+
{
|
8772 |
+
"epoch": 0.9415312131919906,
|
8773 |
+
"grad_norm": 9.228532791137695,
|
8774 |
+
"learning_rate": 1.7259735333795545e-06,
|
8775 |
+
"loss": 0.4282,
|
8776 |
+
"step": 1249
|
8777 |
+
},
|
8778 |
+
{
|
8779 |
+
"epoch": 0.9422850412249706,
|
8780 |
+
"grad_norm": 12.300414085388184,
|
8781 |
+
"learning_rate": 1.6821251920043246e-06,
|
8782 |
+
"loss": 0.4527,
|
8783 |
+
"step": 1250
|
8784 |
+
},
|
8785 |
+
{
|
8786 |
+
"epoch": 0.9430388692579506,
|
8787 |
+
"grad_norm": 5.250865459442139,
|
8788 |
+
"learning_rate": 1.6388362995905848e-06,
|
8789 |
+
"loss": 0.509,
|
8790 |
+
"step": 1251
|
8791 |
+
},
|
8792 |
+
{
|
8793 |
+
"epoch": 0.9437926972909305,
|
8794 |
+
"grad_norm": 5.213113307952881,
|
8795 |
+
"learning_rate": 1.5961071024610752e-06,
|
8796 |
+
"loss": 0.4615,
|
8797 |
+
"step": 1252
|
8798 |
+
},
|
8799 |
+
{
|
8800 |
+
"epoch": 0.9445465253239105,
|
8801 |
+
"grad_norm": 5.200348377227783,
|
8802 |
+
"learning_rate": 1.5539378437536944e-06,
|
8803 |
+
"loss": 0.4463,
|
8804 |
+
"step": 1253
|
8805 |
+
},
|
8806 |
+
{
|
8807 |
+
"epoch": 0.9453003533568904,
|
8808 |
+
"grad_norm": 5.2860941886901855,
|
8809 |
+
"learning_rate": 1.5123287634202454e-06,
|
8810 |
+
"loss": 0.4441,
|
8811 |
+
"step": 1254
|
8812 |
+
},
|
8813 |
+
{
|
8814 |
+
"epoch": 0.9460541813898704,
|
8815 |
+
"grad_norm": 5.183274269104004,
|
8816 |
+
"learning_rate": 1.4712800982249474e-06,
|
8817 |
+
"loss": 0.4292,
|
8818 |
+
"step": 1255
|
8819 |
+
},
|
8820 |
+
{
|
8821 |
+
"epoch": 0.9468080094228504,
|
8822 |
+
"grad_norm": 5.593634605407715,
|
8823 |
+
"learning_rate": 1.430792081743182e-06,
|
8824 |
+
"loss": 0.4589,
|
8825 |
+
"step": 1256
|
8826 |
+
},
|
8827 |
+
{
|
8828 |
+
"epoch": 0.9475618374558303,
|
8829 |
+
"grad_norm": 5.3267388343811035,
|
8830 |
+
"learning_rate": 1.3908649443600707e-06,
|
8831 |
+
"loss": 0.4336,
|
8832 |
+
"step": 1257
|
8833 |
+
},
|
8834 |
+
{
|
8835 |
+
"epoch": 0.9483156654888104,
|
8836 |
+
"grad_norm": 5.741166114807129,
|
8837 |
+
"learning_rate": 1.351498913269289e-06,
|
8838 |
+
"loss": 0.4008,
|
8839 |
+
"step": 1258
|
8840 |
+
},
|
8841 |
+
{
|
8842 |
+
"epoch": 0.9490694935217904,
|
8843 |
+
"grad_norm": 5.336604118347168,
|
8844 |
+
"learning_rate": 1.3126942124716213e-06,
|
8845 |
+
"loss": 0.4218,
|
8846 |
+
"step": 1259
|
8847 |
+
},
|
8848 |
+
{
|
8849 |
+
"epoch": 0.9498233215547703,
|
8850 |
+
"grad_norm": 5.611804962158203,
|
8851 |
+
"learning_rate": 1.2744510627738516e-06,
|
8852 |
+
"loss": 0.4434,
|
8853 |
+
"step": 1260
|
8854 |
+
},
|
8855 |
+
{
|
8856 |
+
"epoch": 0.9505771495877503,
|
8857 |
+
"grad_norm": 5.724870204925537,
|
8858 |
+
"learning_rate": 1.2367696817873419e-06,
|
8859 |
+
"loss": 0.4227,
|
8860 |
+
"step": 1261
|
8861 |
+
},
|
8862 |
+
{
|
8863 |
+
"epoch": 0.9513309776207303,
|
8864 |
+
"grad_norm": 5.307777404785156,
|
8865 |
+
"learning_rate": 1.1996502839269453e-06,
|
8866 |
+
"loss": 0.4002,
|
8867 |
+
"step": 1262
|
8868 |
+
},
|
8869 |
+
{
|
8870 |
+
"epoch": 0.9520848056537102,
|
8871 |
+
"grad_norm": 5.79971170425415,
|
8872 |
+
"learning_rate": 1.1630930804096495e-06,
|
8873 |
+
"loss": 0.405,
|
8874 |
+
"step": 1263
|
8875 |
+
},
|
8876 |
+
{
|
8877 |
+
"epoch": 0.9528386336866902,
|
8878 |
+
"grad_norm": 5.324243068695068,
|
8879 |
+
"learning_rate": 1.127098279253491e-06,
|
8880 |
+
"loss": 0.4043,
|
8881 |
+
"step": 1264
|
8882 |
+
},
|
8883 |
+
{
|
8884 |
+
"epoch": 0.9535924617196702,
|
8885 |
+
"grad_norm": 5.532378673553467,
|
8886 |
+
"learning_rate": 1.0916660852763216e-06,
|
8887 |
+
"loss": 0.4068,
|
8888 |
+
"step": 1265
|
8889 |
+
},
|
8890 |
+
{
|
8891 |
+
"epoch": 0.9543462897526501,
|
8892 |
+
"grad_norm": 5.695662021636963,
|
8893 |
+
"learning_rate": 1.0567967000945866e-06,
|
8894 |
+
"loss": 0.4286,
|
8895 |
+
"step": 1266
|
8896 |
+
},
|
8897 |
+
{
|
8898 |
+
"epoch": 0.9551001177856302,
|
8899 |
+
"grad_norm": 5.8561482429504395,
|
8900 |
+
"learning_rate": 1.0224903221222938e-06,
|
8901 |
+
"loss": 0.4249,
|
8902 |
+
"step": 1267
|
8903 |
+
},
|
8904 |
+
{
|
8905 |
+
"epoch": 0.9558539458186102,
|
8906 |
+
"grad_norm": 5.72511625289917,
|
8907 |
+
"learning_rate": 9.88747146569813e-07,
|
8908 |
+
"loss": 0.4021,
|
8909 |
+
"step": 1268
|
8910 |
+
},
|
8911 |
+
{
|
8912 |
+
"epoch": 0.9566077738515901,
|
8913 |
+
"grad_norm": 5.385478973388672,
|
8914 |
+
"learning_rate": 9.555673654427332e-07,
|
8915 |
+
"loss": 0.3788,
|
8916 |
+
"step": 1269
|
8917 |
+
},
|
8918 |
+
{
|
8919 |
+
"epoch": 0.9573616018845701,
|
8920 |
+
"grad_norm": 5.669264316558838,
|
8921 |
+
"learning_rate": 9.229511675408642e-07,
|
8922 |
+
"loss": 0.4148,
|
8923 |
+
"step": 1270
|
8924 |
+
},
|
8925 |
+
{
|
8926 |
+
"epoch": 0.95811542991755,
|
8927 |
+
"grad_norm": 5.313277244567871,
|
8928 |
+
"learning_rate": 8.90898738457091e-07,
|
8929 |
+
"loss": 0.3641,
|
8930 |
+
"step": 1271
|
8931 |
+
},
|
8932 |
+
{
|
8933 |
+
"epoch": 0.95886925795053,
|
8934 |
+
"grad_norm": 5.480482578277588,
|
8935 |
+
"learning_rate": 8.59410260576321e-07,
|
8936 |
+
"loss": 0.3971,
|
8937 |
+
"step": 1272
|
8938 |
+
},
|
8939 |
+
{
|
8940 |
+
"epoch": 0.95962308598351,
|
8941 |
+
"grad_norm": 5.8209757804870605,
|
8942 |
+
"learning_rate": 8.28485913074506e-07,
|
8943 |
+
"loss": 0.3919,
|
8944 |
+
"step": 1273
|
8945 |
+
},
|
8946 |
+
{
|
8947 |
+
"epoch": 0.96037691401649,
|
8948 |
+
"grad_norm": 5.919877052307129,
|
8949 |
+
"learning_rate": 7.981258719175322e-07,
|
8950 |
+
"loss": 0.3863,
|
8951 |
+
"step": 1274
|
8952 |
+
},
|
8953 |
+
{
|
8954 |
+
"epoch": 0.9611307420494699,
|
8955 |
+
"grad_norm": 5.9404144287109375,
|
8956 |
+
"learning_rate": 7.683303098602989e-07,
|
8957 |
+
"loss": 0.4059,
|
8958 |
+
"step": 1275
|
8959 |
+
},
|
8960 |
+
{
|
8961 |
+
"epoch": 0.96188457008245,
|
8962 |
+
"grad_norm": 5.609850883483887,
|
8963 |
+
"learning_rate": 7.39099396445686e-07,
|
8964 |
+
"loss": 0.3697,
|
8965 |
+
"step": 1276
|
8966 |
+
},
|
8967 |
+
{
|
8968 |
+
"epoch": 0.96263839811543,
|
8969 |
+
"grad_norm": 5.695891857147217,
|
8970 |
+
"learning_rate": 7.104332980036211e-07,
|
8971 |
+
"loss": 0.3917,
|
8972 |
+
"step": 1277
|
8973 |
+
},
|
8974 |
+
{
|
8975 |
+
"epoch": 0.9633922261484099,
|
8976 |
+
"grad_norm": 5.932850360870361,
|
8977 |
+
"learning_rate": 6.823321776501024e-07,
|
8978 |
+
"loss": 0.415,
|
8979 |
+
"step": 1278
|
8980 |
+
},
|
8981 |
+
{
|
8982 |
+
"epoch": 0.9641460541813899,
|
8983 |
+
"grad_norm": 6.023778438568115,
|
8984 |
+
"learning_rate": 6.547961952863002e-07,
|
8985 |
+
"loss": 0.3817,
|
8986 |
+
"step": 1279
|
8987 |
+
},
|
8988 |
+
{
|
8989 |
+
"epoch": 0.9648998822143698,
|
8990 |
+
"grad_norm": 5.926705360412598,
|
8991 |
+
"learning_rate": 6.278255075976125e-07,
|
8992 |
+
"loss": 0.3884,
|
8993 |
+
"step": 1280
|
8994 |
+
},
|
8995 |
+
{
|
8996 |
+
"epoch": 0.9656537102473498,
|
8997 |
+
"grad_norm": 5.837738513946533,
|
8998 |
+
"learning_rate": 6.014202680528324e-07,
|
8999 |
+
"loss": 0.3598,
|
9000 |
+
"step": 1281
|
9001 |
+
},
|
9002 |
+
{
|
9003 |
+
"epoch": 0.9664075382803298,
|
9004 |
+
"grad_norm": 6.178413391113281,
|
9005 |
+
"learning_rate": 5.755806269031827e-07,
|
9006 |
+
"loss": 0.3917,
|
9007 |
+
"step": 1282
|
9008 |
+
},
|
9009 |
+
{
|
9010 |
+
"epoch": 0.9671613663133097,
|
9011 |
+
"grad_norm": 6.282332897186279,
|
9012 |
+
"learning_rate": 5.503067311815713e-07,
|
9013 |
+
"loss": 0.4286,
|
9014 |
+
"step": 1283
|
9015 |
+
},
|
9016 |
+
{
|
9017 |
+
"epoch": 0.9679151943462897,
|
9018 |
+
"grad_norm": 6.746578216552734,
|
9019 |
+
"learning_rate": 5.255987247016591e-07,
|
9020 |
+
"loss": 0.4118,
|
9021 |
+
"step": 1284
|
9022 |
+
},
|
9023 |
+
{
|
9024 |
+
"epoch": 0.9686690223792698,
|
9025 |
+
"grad_norm": 6.075422763824463,
|
9026 |
+
"learning_rate": 5.014567480570831e-07,
|
9027 |
+
"loss": 0.3829,
|
9028 |
+
"step": 1285
|
9029 |
+
},
|
9030 |
+
{
|
9031 |
+
"epoch": 0.9694228504122497,
|
9032 |
+
"grad_norm": 6.149974346160889,
|
9033 |
+
"learning_rate": 4.778809386206895e-07,
|
9034 |
+
"loss": 0.3847,
|
9035 |
+
"step": 1286
|
9036 |
+
},
|
9037 |
+
{
|
9038 |
+
"epoch": 0.9701766784452297,
|
9039 |
+
"grad_norm": 6.333911418914795,
|
9040 |
+
"learning_rate": 4.548714305436685e-07,
|
9041 |
+
"loss": 0.3638,
|
9042 |
+
"step": 1287
|
9043 |
+
},
|
9044 |
+
{
|
9045 |
+
"epoch": 0.9709305064782097,
|
9046 |
+
"grad_norm": 6.391441345214844,
|
9047 |
+
"learning_rate": 4.324283547548658e-07,
|
9048 |
+
"loss": 0.3893,
|
9049 |
+
"step": 1288
|
9050 |
+
},
|
9051 |
+
{
|
9052 |
+
"epoch": 0.9716843345111896,
|
9053 |
+
"grad_norm": 6.624934196472168,
|
9054 |
+
"learning_rate": 4.1055183896001606e-07,
|
9055 |
+
"loss": 0.378,
|
9056 |
+
"step": 1289
|
9057 |
+
},
|
9058 |
+
{
|
9059 |
+
"epoch": 0.9724381625441696,
|
9060 |
+
"grad_norm": 6.473977565765381,
|
9061 |
+
"learning_rate": 3.892420076409886e-07,
|
9062 |
+
"loss": 0.366,
|
9063 |
+
"step": 1290
|
9064 |
+
},
|
9065 |
+
{
|
9066 |
+
"epoch": 0.9731919905771496,
|
9067 |
+
"grad_norm": 6.985432147979736,
|
9068 |
+
"learning_rate": 3.68498982055121e-07,
|
9069 |
+
"loss": 0.4335,
|
9070 |
+
"step": 1291
|
9071 |
+
},
|
9072 |
+
{
|
9073 |
+
"epoch": 0.9739458186101295,
|
9074 |
+
"grad_norm": 7.089210510253906,
|
9075 |
+
"learning_rate": 3.483228802344973e-07,
|
9076 |
+
"loss": 0.4066,
|
9077 |
+
"step": 1292
|
9078 |
+
},
|
9079 |
+
{
|
9080 |
+
"epoch": 0.9746996466431095,
|
9081 |
+
"grad_norm": 7.46934175491333,
|
9082 |
+
"learning_rate": 3.2871381698529324e-07,
|
9083 |
+
"loss": 0.4253,
|
9084 |
+
"step": 1293
|
9085 |
+
},
|
9086 |
+
{
|
9087 |
+
"epoch": 0.9754534746760896,
|
9088 |
+
"grad_norm": 8.461312294006348,
|
9089 |
+
"learning_rate": 3.0967190388712097e-07,
|
9090 |
+
"loss": 0.4596,
|
9091 |
+
"step": 1294
|
9092 |
+
},
|
9093 |
+
{
|
9094 |
+
"epoch": 0.9762073027090695,
|
9095 |
+
"grad_norm": 8.289325714111328,
|
9096 |
+
"learning_rate": 2.9119724929239645e-07,
|
9097 |
+
"loss": 0.4382,
|
9098 |
+
"step": 1295
|
9099 |
+
},
|
9100 |
+
{
|
9101 |
+
"epoch": 0.9769611307420495,
|
9102 |
+
"grad_norm": 8.890064239501953,
|
9103 |
+
"learning_rate": 2.7328995832568426e-07,
|
9104 |
+
"loss": 0.4469,
|
9105 |
+
"step": 1296
|
9106 |
+
},
|
9107 |
+
{
|
9108 |
+
"epoch": 0.9777149587750295,
|
9109 |
+
"grad_norm": 8.737083435058594,
|
9110 |
+
"learning_rate": 2.5595013288318703e-07,
|
9111 |
+
"loss": 0.4262,
|
9112 |
+
"step": 1297
|
9113 |
+
},
|
9114 |
+
{
|
9115 |
+
"epoch": 0.9784687868080094,
|
9116 |
+
"grad_norm": 9.281461715698242,
|
9117 |
+
"learning_rate": 2.391778716320792e-07,
|
9118 |
+
"loss": 0.4036,
|
9119 |
+
"step": 1298
|
9120 |
+
},
|
9121 |
+
{
|
9122 |
+
"epoch": 0.9792226148409894,
|
9123 |
+
"grad_norm": 9.91952896118164,
|
9124 |
+
"learning_rate": 2.2297327000996293e-07,
|
9125 |
+
"loss": 0.4469,
|
9126 |
+
"step": 1299
|
9127 |
+
},
|
9128 |
+
{
|
9129 |
+
"epoch": 0.9799764428739693,
|
9130 |
+
"grad_norm": 11.952555656433105,
|
9131 |
+
"learning_rate": 2.0733642022437994e-07,
|
9132 |
+
"loss": 0.4597,
|
9133 |
+
"step": 1300
|
9134 |
+
},
|
9135 |
+
{
|
9136 |
+
"epoch": 0.9807302709069493,
|
9137 |
+
"grad_norm": 5.1298322677612305,
|
9138 |
+
"learning_rate": 1.922674112522227e-07,
|
9139 |
+
"loss": 0.478,
|
9140 |
+
"step": 1301
|
9141 |
+
},
|
9142 |
+
{
|
9143 |
+
"epoch": 0.9814840989399294,
|
9144 |
+
"grad_norm": 5.572525501251221,
|
9145 |
+
"learning_rate": 1.7776632883924615e-07,
|
9146 |
+
"loss": 0.4829,
|
9147 |
+
"step": 1302
|
9148 |
+
},
|
9149 |
+
{
|
9150 |
+
"epoch": 0.9822379269729093,
|
9151 |
+
"grad_norm": 5.343718528747559,
|
9152 |
+
"learning_rate": 1.638332554996125e-07,
|
9153 |
+
"loss": 0.4319,
|
9154 |
+
"step": 1303
|
9155 |
+
},
|
9156 |
+
{
|
9157 |
+
"epoch": 0.9829917550058893,
|
9158 |
+
"grad_norm": 5.716027736663818,
|
9159 |
+
"learning_rate": 1.5046827051536928e-07,
|
9160 |
+
"loss": 0.4378,
|
9161 |
+
"step": 1304
|
9162 |
+
},
|
9163 |
+
{
|
9164 |
+
"epoch": 0.9837455830388693,
|
9165 |
+
"grad_norm": 5.513693809509277,
|
9166 |
+
"learning_rate": 1.3767144993602766e-07,
|
9167 |
+
"loss": 0.4235,
|
9168 |
+
"step": 1305
|
9169 |
+
},
|
9170 |
+
{
|
9171 |
+
"epoch": 0.9844994110718492,
|
9172 |
+
"grad_norm": 5.508944988250732,
|
9173 |
+
"learning_rate": 1.254428665781515e-07,
|
9174 |
+
"loss": 0.4007,
|
9175 |
+
"step": 1306
|
9176 |
+
},
|
9177 |
+
{
|
9178 |
+
"epoch": 0.9852532391048292,
|
9179 |
+
"grad_norm": 5.180131435394287,
|
9180 |
+
"learning_rate": 1.1378259002488013e-07,
|
9181 |
+
"loss": 0.3939,
|
9182 |
+
"step": 1307
|
9183 |
+
},
|
9184 |
+
{
|
9185 |
+
"epoch": 0.9860070671378092,
|
9186 |
+
"grad_norm": 5.590184688568115,
|
9187 |
+
"learning_rate": 1.0269068662560611e-07,
|
9188 |
+
"loss": 0.4166,
|
9189 |
+
"step": 1308
|
9190 |
+
},
|
9191 |
+
{
|
9192 |
+
"epoch": 0.9867608951707891,
|
9193 |
+
"grad_norm": 5.44436502456665,
|
9194 |
+
"learning_rate": 9.216721949553142e-08,
|
9195 |
+
"loss": 0.4047,
|
9196 |
+
"step": 1309
|
9197 |
+
},
|
9198 |
+
{
|
9199 |
+
"epoch": 0.9875147232037691,
|
9200 |
+
"grad_norm": 5.489165782928467,
|
9201 |
+
"learning_rate": 8.221224851535647e-08,
|
9202 |
+
"loss": 0.3999,
|
9203 |
+
"step": 1310
|
9204 |
+
},
|
9205 |
+
{
|
9206 |
+
"epoch": 0.9882685512367492,
|
9207 |
+
"grad_norm": 5.663797855377197,
|
9208 |
+
"learning_rate": 7.282583033091372e-08,
|
9209 |
+
"loss": 0.3842,
|
9210 |
+
"step": 1311
|
9211 |
+
},
|
9212 |
+
{
|
9213 |
+
"epoch": 0.9890223792697291,
|
9214 |
+
"grad_norm": 5.638896942138672,
|
9215 |
+
"learning_rate": 6.400801835286796e-08,
|
9216 |
+
"loss": 0.3977,
|
9217 |
+
"step": 1312
|
9218 |
+
},
|
9219 |
+
{
|
9220 |
+
"epoch": 0.9897762073027091,
|
9221 |
+
"grad_norm": 5.8632307052612305,
|
9222 |
+
"learning_rate": 5.57588627563721e-08,
|
9223 |
+
"loss": 0.3579,
|
9224 |
+
"step": 1313
|
9225 |
+
},
|
9226 |
+
{
|
9227 |
+
"epoch": 0.9905300353356891,
|
9228 |
+
"grad_norm": 5.826532363891602,
|
9229 |
+
"learning_rate": 4.807841048082296e-08,
|
9230 |
+
"loss": 0.4088,
|
9231 |
+
"step": 1314
|
9232 |
+
},
|
9233 |
+
{
|
9234 |
+
"epoch": 0.991283863368669,
|
9235 |
+
"grad_norm": 5.712516784667969,
|
9236 |
+
"learning_rate": 4.096670522959478e-08,
|
9237 |
+
"loss": 0.3853,
|
9238 |
+
"step": 1315
|
9239 |
+
},
|
9240 |
+
{
|
9241 |
+
"epoch": 0.992037691401649,
|
9242 |
+
"grad_norm": 6.0777459144592285,
|
9243 |
+
"learning_rate": 3.442378746972841e-08,
|
9244 |
+
"loss": 0.4111,
|
9245 |
+
"step": 1316
|
9246 |
+
},
|
9247 |
+
{
|
9248 |
+
"epoch": 0.992791519434629,
|
9249 |
+
"grad_norm": 5.916062831878662,
|
9250 |
+
"learning_rate": 2.844969443178691e-08,
|
9251 |
+
"loss": 0.3821,
|
9252 |
+
"step": 1317
|
9253 |
+
},
|
9254 |
+
{
|
9255 |
+
"epoch": 0.9935453474676089,
|
9256 |
+
"grad_norm": 5.911341190338135,
|
9257 |
+
"learning_rate": 2.304446010958916e-08,
|
9258 |
+
"loss": 0.38,
|
9259 |
+
"step": 1318
|
9260 |
+
},
|
9261 |
+
{
|
9262 |
+
"epoch": 0.9942991755005889,
|
9263 |
+
"grad_norm": 6.334498405456543,
|
9264 |
+
"learning_rate": 1.8208115260032187e-08,
|
9265 |
+
"loss": 0.3812,
|
9266 |
+
"step": 1319
|
9267 |
+
},
|
9268 |
+
{
|
9269 |
+
"epoch": 0.995053003533569,
|
9270 |
+
"grad_norm": 6.576707363128662,
|
9271 |
+
"learning_rate": 1.3940687402924646e-08,
|
9272 |
+
"loss": 0.3858,
|
9273 |
+
"step": 1320
|
9274 |
+
},
|
9275 |
+
{
|
9276 |
+
"epoch": 0.9958068315665489,
|
9277 |
+
"grad_norm": 6.39242696762085,
|
9278 |
+
"learning_rate": 1.0242200820786974e-08,
|
9279 |
+
"loss": 0.3661,
|
9280 |
+
"step": 1321
|
9281 |
+
},
|
9282 |
+
{
|
9283 |
+
"epoch": 0.9965606595995289,
|
9284 |
+
"grad_norm": 7.869157791137695,
|
9285 |
+
"learning_rate": 7.112676558784781e-09,
|
9286 |
+
"loss": 0.3966,
|
9287 |
+
"step": 1322
|
9288 |
+
},
|
9289 |
+
{
|
9290 |
+
"epoch": 0.9973144876325089,
|
9291 |
+
"grad_norm": 7.689291954040527,
|
9292 |
+
"learning_rate": 4.552132424562317e-09,
|
9293 |
+
"loss": 0.4297,
|
9294 |
+
"step": 1323
|
9295 |
+
},
|
9296 |
+
{
|
9297 |
+
"epoch": 0.9980683156654888,
|
9298 |
+
"grad_norm": 8.572519302368164,
|
9299 |
+
"learning_rate": 2.5605829881203414e-09,
|
9300 |
+
"loss": 0.451,
|
9301 |
+
"step": 1324
|
9302 |
+
},
|
9303 |
+
{
|
9304 |
+
"epoch": 0.9988221436984688,
|
9305 |
+
"grad_norm": 9.072525024414062,
|
9306 |
+
"learning_rate": 1.1380395818050282e-09,
|
9307 |
+
"loss": 0.4373,
|
9308 |
+
"step": 1325
|
9309 |
+
},
|
9310 |
+
{
|
9311 |
+
"epoch": 0.9995759717314487,
|
9312 |
+
"grad_norm": 9.224164962768555,
|
9313 |
+
"learning_rate": 2.8451030018583623e-10,
|
9314 |
+
"loss": 0.4368,
|
9315 |
+
"step": 1326
|
9316 |
+
},
|
9317 |
+
{
|
9318 |
+
"epoch": 1.0007067137809187,
|
9319 |
+
"grad_norm": 5.4062819480896,
|
9320 |
+
"learning_rate": 0.0,
|
9321 |
+
"loss": 0.4839,
|
9322 |
+
"step": 1327
|
9323 |
}
|
9324 |
],
|
9325 |
"logging_steps": 1,
|
|
|
9334 |
"should_evaluate": false,
|
9335 |
"should_log": false,
|
9336 |
"should_save": true,
|
9337 |
+
"should_training_stop": true
|
9338 |
},
|
9339 |
"attributes": {}
|
9340 |
}
|
9341 |
},
|
9342 |
+
"total_flos": 1.8606342447625667e+18,
|
9343 |
"train_batch_size": 2,
|
9344 |
"trial_name": null,
|
9345 |
"trial_params": null
|