Training in progress, step 10000, checkpoint
Browse files
last-checkpoint/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 560983656
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1c8e6799b936bbb72f55d38f901d37d73eb48aa2c7687d5081feb69086f1ca29
|
| 3 |
size 560983656
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 246053739
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8459e376ea931998d33c038080ed53ca14e8b5c5ec8f7e930035f963ea0160eb
|
| 3 |
size 246053739
|
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 sha256:713b57f62ca2b5f71e37e51f362f0a8f37b5504e2e104de0c5deeb68a0b57d4c
|
| 3 |
size 14244
|
last-checkpoint/scaler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 988
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:581a3b97e73590097d6ffd55f019c06921d3c41b5bf808069d22c75dbddbf820
|
| 3 |
size 988
|
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:b24df43f2a8c3c00021031b8f6e5b0ecb8e2da1059013876673704ecfc7bfbf7
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -2,9 +2,9 @@
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
-
"epoch": 0.
|
| 6 |
"eval_steps": 50,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
@@ -1608,6 +1608,1606 @@
|
|
| 1608 |
"eval_samples_per_second": 5.614,
|
| 1609 |
"eval_steps_per_second": 2.245,
|
| 1610 |
"step": 5000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1611 |
}
|
| 1612 |
],
|
| 1613 |
"logging_steps": 50,
|
|
@@ -1627,7 +3227,7 @@
|
|
| 1627 |
"attributes": {}
|
| 1628 |
}
|
| 1629 |
},
|
| 1630 |
-
"total_flos":
|
| 1631 |
"train_batch_size": 2,
|
| 1632 |
"trial_name": null,
|
| 1633 |
"trial_params": null
|
|
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.016784409297891375,
|
| 6 |
"eval_steps": 50,
|
| 7 |
+
"global_step": 10000,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 1608 |
"eval_samples_per_second": 5.614,
|
| 1609 |
"eval_steps_per_second": 2.245,
|
| 1610 |
"step": 5000
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 0.008476126695435144,
|
| 1614 |
+
"grad_norm": 6.474329471588135,
|
| 1615 |
+
"learning_rate": 4.9576445431367715e-05,
|
| 1616 |
+
"loss": 1.8228,
|
| 1617 |
+
"step": 5050
|
| 1618 |
+
},
|
| 1619 |
+
{
|
| 1620 |
+
"epoch": 0.008476126695435144,
|
| 1621 |
+
"eval_loss": 1.4911173582077026,
|
| 1622 |
+
"eval_masked_accuracy": 0.71875,
|
| 1623 |
+
"eval_runtime": 1.8052,
|
| 1624 |
+
"eval_samples_per_second": 5.54,
|
| 1625 |
+
"eval_steps_per_second": 2.216,
|
| 1626 |
+
"step": 5050
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 0.0085600487419246,
|
| 1630 |
+
"grad_norm": 4.493051052093506,
|
| 1631 |
+
"learning_rate": 4.9572249329043244e-05,
|
| 1632 |
+
"loss": 1.5526,
|
| 1633 |
+
"step": 5100
|
| 1634 |
+
},
|
| 1635 |
+
{
|
| 1636 |
+
"epoch": 0.0085600487419246,
|
| 1637 |
+
"eval_loss": 1.4060901403427124,
|
| 1638 |
+
"eval_masked_accuracy": 0.7131474018096924,
|
| 1639 |
+
"eval_runtime": 1.8193,
|
| 1640 |
+
"eval_samples_per_second": 5.497,
|
| 1641 |
+
"eval_steps_per_second": 2.199,
|
| 1642 |
+
"step": 5100
|
| 1643 |
+
},
|
| 1644 |
+
{
|
| 1645 |
+
"epoch": 0.008643970788414057,
|
| 1646 |
+
"grad_norm": 5.657381057739258,
|
| 1647 |
+
"learning_rate": 4.956805322671877e-05,
|
| 1648 |
+
"loss": 1.5743,
|
| 1649 |
+
"step": 5150
|
| 1650 |
+
},
|
| 1651 |
+
{
|
| 1652 |
+
"epoch": 0.008643970788414057,
|
| 1653 |
+
"eval_loss": 1.7347627878189087,
|
| 1654 |
+
"eval_masked_accuracy": 0.6392694115638733,
|
| 1655 |
+
"eval_runtime": 1.7632,
|
| 1656 |
+
"eval_samples_per_second": 5.671,
|
| 1657 |
+
"eval_steps_per_second": 2.269,
|
| 1658 |
+
"step": 5150
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 0.008727892834903515,
|
| 1662 |
+
"grad_norm": 5.059664726257324,
|
| 1663 |
+
"learning_rate": 4.9563941046440784e-05,
|
| 1664 |
+
"loss": 1.5923,
|
| 1665 |
+
"step": 5200
|
| 1666 |
+
},
|
| 1667 |
+
{
|
| 1668 |
+
"epoch": 0.008727892834903515,
|
| 1669 |
+
"eval_loss": 1.7108001708984375,
|
| 1670 |
+
"eval_masked_accuracy": 0.6759999990463257,
|
| 1671 |
+
"eval_runtime": 1.7312,
|
| 1672 |
+
"eval_samples_per_second": 5.776,
|
| 1673 |
+
"eval_steps_per_second": 2.311,
|
| 1674 |
+
"step": 5200
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"epoch": 0.008811814881392972,
|
| 1678 |
+
"grad_norm": 6.256536483764648,
|
| 1679 |
+
"learning_rate": 4.955974494411631e-05,
|
| 1680 |
+
"loss": 1.5454,
|
| 1681 |
+
"step": 5250
|
| 1682 |
+
},
|
| 1683 |
+
{
|
| 1684 |
+
"epoch": 0.008811814881392972,
|
| 1685 |
+
"eval_loss": 1.8423763513565063,
|
| 1686 |
+
"eval_masked_accuracy": 0.6590909361839294,
|
| 1687 |
+
"eval_runtime": 1.7323,
|
| 1688 |
+
"eval_samples_per_second": 5.773,
|
| 1689 |
+
"eval_steps_per_second": 2.309,
|
| 1690 |
+
"step": 5250
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 0.008895736927882428,
|
| 1694 |
+
"grad_norm": 6.45760440826416,
|
| 1695 |
+
"learning_rate": 4.955554884179184e-05,
|
| 1696 |
+
"loss": 1.5381,
|
| 1697 |
+
"step": 5300
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 0.008895736927882428,
|
| 1701 |
+
"eval_loss": 1.8820030689239502,
|
| 1702 |
+
"eval_masked_accuracy": 0.6486486196517944,
|
| 1703 |
+
"eval_runtime": 1.7529,
|
| 1704 |
+
"eval_samples_per_second": 5.705,
|
| 1705 |
+
"eval_steps_per_second": 2.282,
|
| 1706 |
+
"step": 5300
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 0.008979658974371885,
|
| 1710 |
+
"grad_norm": 7.668667793273926,
|
| 1711 |
+
"learning_rate": 4.955135273946737e-05,
|
| 1712 |
+
"loss": 1.6363,
|
| 1713 |
+
"step": 5350
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 0.008979658974371885,
|
| 1717 |
+
"eval_loss": 1.631400465965271,
|
| 1718 |
+
"eval_masked_accuracy": 0.7160493731498718,
|
| 1719 |
+
"eval_runtime": 1.7511,
|
| 1720 |
+
"eval_samples_per_second": 5.711,
|
| 1721 |
+
"eval_steps_per_second": 2.284,
|
| 1722 |
+
"step": 5350
|
| 1723 |
+
},
|
| 1724 |
+
{
|
| 1725 |
+
"epoch": 0.009063581020861342,
|
| 1726 |
+
"grad_norm": 7.2050018310546875,
|
| 1727 |
+
"learning_rate": 4.954715663714289e-05,
|
| 1728 |
+
"loss": 1.5738,
|
| 1729 |
+
"step": 5400
|
| 1730 |
+
},
|
| 1731 |
+
{
|
| 1732 |
+
"epoch": 0.009063581020861342,
|
| 1733 |
+
"eval_loss": 1.5917881727218628,
|
| 1734 |
+
"eval_masked_accuracy": 0.7405857443809509,
|
| 1735 |
+
"eval_runtime": 1.75,
|
| 1736 |
+
"eval_samples_per_second": 5.714,
|
| 1737 |
+
"eval_steps_per_second": 2.286,
|
| 1738 |
+
"step": 5400
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 0.0091475030673508,
|
| 1742 |
+
"grad_norm": 6.094969749450684,
|
| 1743 |
+
"learning_rate": 4.954296053481842e-05,
|
| 1744 |
+
"loss": 1.7321,
|
| 1745 |
+
"step": 5450
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 0.0091475030673508,
|
| 1749 |
+
"eval_loss": 1.5327577590942383,
|
| 1750 |
+
"eval_masked_accuracy": 0.707317054271698,
|
| 1751 |
+
"eval_runtime": 1.7423,
|
| 1752 |
+
"eval_samples_per_second": 5.74,
|
| 1753 |
+
"eval_steps_per_second": 2.296,
|
| 1754 |
+
"step": 5450
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 0.009231425113840256,
|
| 1758 |
+
"grad_norm": 8.869881629943848,
|
| 1759 |
+
"learning_rate": 4.953876443249395e-05,
|
| 1760 |
+
"loss": 1.5768,
|
| 1761 |
+
"step": 5500
|
| 1762 |
+
},
|
| 1763 |
+
{
|
| 1764 |
+
"epoch": 0.009231425113840256,
|
| 1765 |
+
"eval_loss": 1.3501726388931274,
|
| 1766 |
+
"eval_masked_accuracy": 0.7801724076271057,
|
| 1767 |
+
"eval_runtime": 1.7732,
|
| 1768 |
+
"eval_samples_per_second": 5.64,
|
| 1769 |
+
"eval_steps_per_second": 2.256,
|
| 1770 |
+
"step": 5500
|
| 1771 |
+
},
|
| 1772 |
+
{
|
| 1773 |
+
"epoch": 0.009315347160329713,
|
| 1774 |
+
"grad_norm": 4.408574104309082,
|
| 1775 |
+
"learning_rate": 4.9534568330169476e-05,
|
| 1776 |
+
"loss": 1.5802,
|
| 1777 |
+
"step": 5550
|
| 1778 |
+
},
|
| 1779 |
+
{
|
| 1780 |
+
"epoch": 0.009315347160329713,
|
| 1781 |
+
"eval_loss": 1.7055152654647827,
|
| 1782 |
+
"eval_masked_accuracy": 0.6707317233085632,
|
| 1783 |
+
"eval_runtime": 1.7418,
|
| 1784 |
+
"eval_samples_per_second": 5.741,
|
| 1785 |
+
"eval_steps_per_second": 2.296,
|
| 1786 |
+
"step": 5550
|
| 1787 |
+
},
|
| 1788 |
+
{
|
| 1789 |
+
"epoch": 0.00939926920681917,
|
| 1790 |
+
"grad_norm": 5.3869147300720215,
|
| 1791 |
+
"learning_rate": 4.9530372227845e-05,
|
| 1792 |
+
"loss": 1.5547,
|
| 1793 |
+
"step": 5600
|
| 1794 |
+
},
|
| 1795 |
+
{
|
| 1796 |
+
"epoch": 0.00939926920681917,
|
| 1797 |
+
"eval_loss": 1.3663699626922607,
|
| 1798 |
+
"eval_masked_accuracy": 0.6974790096282959,
|
| 1799 |
+
"eval_runtime": 1.7338,
|
| 1800 |
+
"eval_samples_per_second": 5.768,
|
| 1801 |
+
"eval_steps_per_second": 2.307,
|
| 1802 |
+
"step": 5600
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 0.009483191253308626,
|
| 1806 |
+
"grad_norm": 4.417982578277588,
|
| 1807 |
+
"learning_rate": 4.9526176125520526e-05,
|
| 1808 |
+
"loss": 1.5658,
|
| 1809 |
+
"step": 5650
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 0.009483191253308626,
|
| 1813 |
+
"eval_loss": 1.6572059392929077,
|
| 1814 |
+
"eval_masked_accuracy": 0.6520000100135803,
|
| 1815 |
+
"eval_runtime": 1.7583,
|
| 1816 |
+
"eval_samples_per_second": 5.687,
|
| 1817 |
+
"eval_steps_per_second": 2.275,
|
| 1818 |
+
"step": 5650
|
| 1819 |
+
},
|
| 1820 |
+
{
|
| 1821 |
+
"epoch": 0.009567113299798084,
|
| 1822 |
+
"grad_norm": 5.2137861251831055,
|
| 1823 |
+
"learning_rate": 4.9521980023196055e-05,
|
| 1824 |
+
"loss": 1.5929,
|
| 1825 |
+
"step": 5700
|
| 1826 |
+
},
|
| 1827 |
+
{
|
| 1828 |
+
"epoch": 0.009567113299798084,
|
| 1829 |
+
"eval_loss": 1.4574190378189087,
|
| 1830 |
+
"eval_masked_accuracy": 0.6694560647010803,
|
| 1831 |
+
"eval_runtime": 1.7352,
|
| 1832 |
+
"eval_samples_per_second": 5.763,
|
| 1833 |
+
"eval_steps_per_second": 2.305,
|
| 1834 |
+
"step": 5700
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 0.00965103534628754,
|
| 1838 |
+
"grad_norm": 6.848864555358887,
|
| 1839 |
+
"learning_rate": 4.951778392087158e-05,
|
| 1840 |
+
"loss": 1.6008,
|
| 1841 |
+
"step": 5750
|
| 1842 |
+
},
|
| 1843 |
+
{
|
| 1844 |
+
"epoch": 0.00965103534628754,
|
| 1845 |
+
"eval_loss": 2.133417844772339,
|
| 1846 |
+
"eval_masked_accuracy": 0.6540084481239319,
|
| 1847 |
+
"eval_runtime": 1.8568,
|
| 1848 |
+
"eval_samples_per_second": 5.386,
|
| 1849 |
+
"eval_steps_per_second": 2.154,
|
| 1850 |
+
"step": 5750
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 0.009734957392776997,
|
| 1854 |
+
"grad_norm": 3.9827840328216553,
|
| 1855 |
+
"learning_rate": 4.9513587818547105e-05,
|
| 1856 |
+
"loss": 1.5811,
|
| 1857 |
+
"step": 5800
|
| 1858 |
+
},
|
| 1859 |
+
{
|
| 1860 |
+
"epoch": 0.009734957392776997,
|
| 1861 |
+
"eval_loss": 1.403198003768921,
|
| 1862 |
+
"eval_masked_accuracy": 0.7085201740264893,
|
| 1863 |
+
"eval_runtime": 1.749,
|
| 1864 |
+
"eval_samples_per_second": 5.717,
|
| 1865 |
+
"eval_steps_per_second": 2.287,
|
| 1866 |
+
"step": 5800
|
| 1867 |
+
},
|
| 1868 |
+
{
|
| 1869 |
+
"epoch": 0.009818879439266454,
|
| 1870 |
+
"grad_norm": 4.541887283325195,
|
| 1871 |
+
"learning_rate": 4.950939171622263e-05,
|
| 1872 |
+
"loss": 1.558,
|
| 1873 |
+
"step": 5850
|
| 1874 |
+
},
|
| 1875 |
+
{
|
| 1876 |
+
"epoch": 0.009818879439266454,
|
| 1877 |
+
"eval_loss": 1.4281632900238037,
|
| 1878 |
+
"eval_masked_accuracy": 0.7195122241973877,
|
| 1879 |
+
"eval_runtime": 1.7523,
|
| 1880 |
+
"eval_samples_per_second": 5.707,
|
| 1881 |
+
"eval_steps_per_second": 2.283,
|
| 1882 |
+
"step": 5850
|
| 1883 |
+
},
|
| 1884 |
+
{
|
| 1885 |
+
"epoch": 0.00990280148575591,
|
| 1886 |
+
"grad_norm": 8.121429443359375,
|
| 1887 |
+
"learning_rate": 4.950519561389816e-05,
|
| 1888 |
+
"loss": 1.5583,
|
| 1889 |
+
"step": 5900
|
| 1890 |
+
},
|
| 1891 |
+
{
|
| 1892 |
+
"epoch": 0.00990280148575591,
|
| 1893 |
+
"eval_loss": 1.608547568321228,
|
| 1894 |
+
"eval_masked_accuracy": 0.6582278609275818,
|
| 1895 |
+
"eval_runtime": 1.7405,
|
| 1896 |
+
"eval_samples_per_second": 5.745,
|
| 1897 |
+
"eval_steps_per_second": 2.298,
|
| 1898 |
+
"step": 5900
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 0.009986723532245369,
|
| 1902 |
+
"grad_norm": 4.750977039337158,
|
| 1903 |
+
"learning_rate": 4.950099951157369e-05,
|
| 1904 |
+
"loss": 1.5378,
|
| 1905 |
+
"step": 5950
|
| 1906 |
+
},
|
| 1907 |
+
{
|
| 1908 |
+
"epoch": 0.009986723532245369,
|
| 1909 |
+
"eval_loss": 1.3912121057510376,
|
| 1910 |
+
"eval_masked_accuracy": 0.701298713684082,
|
| 1911 |
+
"eval_runtime": 1.7623,
|
| 1912 |
+
"eval_samples_per_second": 5.674,
|
| 1913 |
+
"eval_steps_per_second": 2.27,
|
| 1914 |
+
"step": 5950
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 0.010070645578734825,
|
| 1918 |
+
"grad_norm": 4.445640563964844,
|
| 1919 |
+
"learning_rate": 4.949680340924922e-05,
|
| 1920 |
+
"loss": 1.5063,
|
| 1921 |
+
"step": 6000
|
| 1922 |
+
},
|
| 1923 |
+
{
|
| 1924 |
+
"epoch": 0.010070645578734825,
|
| 1925 |
+
"eval_loss": 1.6513465642929077,
|
| 1926 |
+
"eval_masked_accuracy": 0.6796537041664124,
|
| 1927 |
+
"eval_runtime": 1.7424,
|
| 1928 |
+
"eval_samples_per_second": 5.739,
|
| 1929 |
+
"eval_steps_per_second": 2.296,
|
| 1930 |
+
"step": 6000
|
| 1931 |
+
},
|
| 1932 |
+
{
|
| 1933 |
+
"epoch": 0.010154567625224282,
|
| 1934 |
+
"grad_norm": 13.394184112548828,
|
| 1935 |
+
"learning_rate": 4.949260730692475e-05,
|
| 1936 |
+
"loss": 1.5155,
|
| 1937 |
+
"step": 6050
|
| 1938 |
+
},
|
| 1939 |
+
{
|
| 1940 |
+
"epoch": 0.010154567625224282,
|
| 1941 |
+
"eval_loss": 1.5842430591583252,
|
| 1942 |
+
"eval_masked_accuracy": 0.6853448152542114,
|
| 1943 |
+
"eval_runtime": 1.7416,
|
| 1944 |
+
"eval_samples_per_second": 5.742,
|
| 1945 |
+
"eval_steps_per_second": 2.297,
|
| 1946 |
+
"step": 6050
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 0.010238489671713738,
|
| 1950 |
+
"grad_norm": 7.441386699676514,
|
| 1951 |
+
"learning_rate": 4.948841120460027e-05,
|
| 1952 |
+
"loss": 1.5009,
|
| 1953 |
+
"step": 6100
|
| 1954 |
+
},
|
| 1955 |
+
{
|
| 1956 |
+
"epoch": 0.010238489671713738,
|
| 1957 |
+
"eval_loss": 1.512109637260437,
|
| 1958 |
+
"eval_masked_accuracy": 0.6987447738647461,
|
| 1959 |
+
"eval_runtime": 1.7546,
|
| 1960 |
+
"eval_samples_per_second": 5.699,
|
| 1961 |
+
"eval_steps_per_second": 2.28,
|
| 1962 |
+
"step": 6100
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 0.010322411718203195,
|
| 1966 |
+
"grad_norm": 6.1988749504089355,
|
| 1967 |
+
"learning_rate": 4.94842151022758e-05,
|
| 1968 |
+
"loss": 1.5567,
|
| 1969 |
+
"step": 6150
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 0.010322411718203195,
|
| 1973 |
+
"eval_loss": 1.5210555791854858,
|
| 1974 |
+
"eval_masked_accuracy": 0.7109375,
|
| 1975 |
+
"eval_runtime": 1.7524,
|
| 1976 |
+
"eval_samples_per_second": 5.707,
|
| 1977 |
+
"eval_steps_per_second": 2.283,
|
| 1978 |
+
"step": 6150
|
| 1979 |
+
},
|
| 1980 |
+
{
|
| 1981 |
+
"epoch": 0.010406333764692651,
|
| 1982 |
+
"grad_norm": 4.782381057739258,
|
| 1983 |
+
"learning_rate": 4.9480018999951325e-05,
|
| 1984 |
+
"loss": 1.6125,
|
| 1985 |
+
"step": 6200
|
| 1986 |
+
},
|
| 1987 |
+
{
|
| 1988 |
+
"epoch": 0.010406333764692651,
|
| 1989 |
+
"eval_loss": 1.6434142589569092,
|
| 1990 |
+
"eval_masked_accuracy": 0.6638655662536621,
|
| 1991 |
+
"eval_runtime": 1.7489,
|
| 1992 |
+
"eval_samples_per_second": 5.718,
|
| 1993 |
+
"eval_steps_per_second": 2.287,
|
| 1994 |
+
"step": 6200
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 0.01049025581118211,
|
| 1998 |
+
"grad_norm": 5.14832878112793,
|
| 1999 |
+
"learning_rate": 4.9475822897626854e-05,
|
| 2000 |
+
"loss": 1.6089,
|
| 2001 |
+
"step": 6250
|
| 2002 |
+
},
|
| 2003 |
+
{
|
| 2004 |
+
"epoch": 0.01049025581118211,
|
| 2005 |
+
"eval_loss": 1.239379644393921,
|
| 2006 |
+
"eval_masked_accuracy": 0.7427386045455933,
|
| 2007 |
+
"eval_runtime": 1.7532,
|
| 2008 |
+
"eval_samples_per_second": 5.704,
|
| 2009 |
+
"eval_steps_per_second": 2.282,
|
| 2010 |
+
"step": 6250
|
| 2011 |
+
},
|
| 2012 |
+
{
|
| 2013 |
+
"epoch": 0.010574177857671566,
|
| 2014 |
+
"grad_norm": 5.390649795532227,
|
| 2015 |
+
"learning_rate": 4.947162679530238e-05,
|
| 2016 |
+
"loss": 1.6357,
|
| 2017 |
+
"step": 6300
|
| 2018 |
+
},
|
| 2019 |
+
{
|
| 2020 |
+
"epoch": 0.010574177857671566,
|
| 2021 |
+
"eval_loss": 1.5129663944244385,
|
| 2022 |
+
"eval_masked_accuracy": 0.692307710647583,
|
| 2023 |
+
"eval_runtime": 1.7523,
|
| 2024 |
+
"eval_samples_per_second": 5.707,
|
| 2025 |
+
"eval_steps_per_second": 2.283,
|
| 2026 |
+
"step": 6300
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 0.010658099904161023,
|
| 2030 |
+
"grad_norm": 4.3327412605285645,
|
| 2031 |
+
"learning_rate": 4.9467430692977904e-05,
|
| 2032 |
+
"loss": 1.5318,
|
| 2033 |
+
"step": 6350
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 0.010658099904161023,
|
| 2037 |
+
"eval_loss": 1.7716737985610962,
|
| 2038 |
+
"eval_masked_accuracy": 0.6942148804664612,
|
| 2039 |
+
"eval_runtime": 1.7284,
|
| 2040 |
+
"eval_samples_per_second": 5.786,
|
| 2041 |
+
"eval_steps_per_second": 2.314,
|
| 2042 |
+
"step": 6350
|
| 2043 |
+
},
|
| 2044 |
+
{
|
| 2045 |
+
"epoch": 0.01074202195065048,
|
| 2046 |
+
"grad_norm": 5.145776271820068,
|
| 2047 |
+
"learning_rate": 4.946323459065343e-05,
|
| 2048 |
+
"loss": 1.6081,
|
| 2049 |
+
"step": 6400
|
| 2050 |
+
},
|
| 2051 |
+
{
|
| 2052 |
+
"epoch": 0.01074202195065048,
|
| 2053 |
+
"eval_loss": 1.6661970615386963,
|
| 2054 |
+
"eval_masked_accuracy": 0.6882591247558594,
|
| 2055 |
+
"eval_runtime": 1.7486,
|
| 2056 |
+
"eval_samples_per_second": 5.719,
|
| 2057 |
+
"eval_steps_per_second": 2.288,
|
| 2058 |
+
"step": 6400
|
| 2059 |
+
},
|
| 2060 |
+
{
|
| 2061 |
+
"epoch": 0.010825943997139936,
|
| 2062 |
+
"grad_norm": 5.037006855010986,
|
| 2063 |
+
"learning_rate": 4.945903848832896e-05,
|
| 2064 |
+
"loss": 1.5028,
|
| 2065 |
+
"step": 6450
|
| 2066 |
+
},
|
| 2067 |
+
{
|
| 2068 |
+
"epoch": 0.010825943997139936,
|
| 2069 |
+
"eval_loss": 1.4679136276245117,
|
| 2070 |
+
"eval_masked_accuracy": 0.714893639087677,
|
| 2071 |
+
"eval_runtime": 1.7514,
|
| 2072 |
+
"eval_samples_per_second": 5.71,
|
| 2073 |
+
"eval_steps_per_second": 2.284,
|
| 2074 |
+
"step": 6450
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 0.010909866043629394,
|
| 2078 |
+
"grad_norm": 5.618253707885742,
|
| 2079 |
+
"learning_rate": 4.945484238600449e-05,
|
| 2080 |
+
"loss": 1.5477,
|
| 2081 |
+
"step": 6500
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 0.010909866043629394,
|
| 2085 |
+
"eval_loss": 1.6666347980499268,
|
| 2086 |
+
"eval_masked_accuracy": 0.7094017267227173,
|
| 2087 |
+
"eval_runtime": 1.7486,
|
| 2088 |
+
"eval_samples_per_second": 5.719,
|
| 2089 |
+
"eval_steps_per_second": 2.288,
|
| 2090 |
+
"step": 6500
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 0.01099378809011885,
|
| 2094 |
+
"grad_norm": 14.34435749053955,
|
| 2095 |
+
"learning_rate": 4.945064628368002e-05,
|
| 2096 |
+
"loss": 1.6291,
|
| 2097 |
+
"step": 6550
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"epoch": 0.01099378809011885,
|
| 2101 |
+
"eval_loss": 1.8381481170654297,
|
| 2102 |
+
"eval_masked_accuracy": 0.6547085046768188,
|
| 2103 |
+
"eval_runtime": 1.7548,
|
| 2104 |
+
"eval_samples_per_second": 5.699,
|
| 2105 |
+
"eval_steps_per_second": 2.279,
|
| 2106 |
+
"step": 6550
|
| 2107 |
+
},
|
| 2108 |
+
{
|
| 2109 |
+
"epoch": 0.011077710136608307,
|
| 2110 |
+
"grad_norm": 4.846654891967773,
|
| 2111 |
+
"learning_rate": 4.9446450181355546e-05,
|
| 2112 |
+
"loss": 1.6077,
|
| 2113 |
+
"step": 6600
|
| 2114 |
+
},
|
| 2115 |
+
{
|
| 2116 |
+
"epoch": 0.011077710136608307,
|
| 2117 |
+
"eval_loss": 1.5568077564239502,
|
| 2118 |
+
"eval_masked_accuracy": 0.6872428059577942,
|
| 2119 |
+
"eval_runtime": 1.7324,
|
| 2120 |
+
"eval_samples_per_second": 5.772,
|
| 2121 |
+
"eval_steps_per_second": 2.309,
|
| 2122 |
+
"step": 6600
|
| 2123 |
+
},
|
| 2124 |
+
{
|
| 2125 |
+
"epoch": 0.011161632183097764,
|
| 2126 |
+
"grad_norm": 5.304859161376953,
|
| 2127 |
+
"learning_rate": 4.944225407903107e-05,
|
| 2128 |
+
"loss": 1.5758,
|
| 2129 |
+
"step": 6650
|
| 2130 |
+
},
|
| 2131 |
+
{
|
| 2132 |
+
"epoch": 0.011161632183097764,
|
| 2133 |
+
"eval_loss": 1.3110054731369019,
|
| 2134 |
+
"eval_masked_accuracy": 0.7312775254249573,
|
| 2135 |
+
"eval_runtime": 1.7439,
|
| 2136 |
+
"eval_samples_per_second": 5.734,
|
| 2137 |
+
"eval_steps_per_second": 2.294,
|
| 2138 |
+
"step": 6650
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 0.01124555422958722,
|
| 2142 |
+
"grad_norm": 6.187143802642822,
|
| 2143 |
+
"learning_rate": 4.9438057976706596e-05,
|
| 2144 |
+
"loss": 1.5817,
|
| 2145 |
+
"step": 6700
|
| 2146 |
+
},
|
| 2147 |
+
{
|
| 2148 |
+
"epoch": 0.01124555422958722,
|
| 2149 |
+
"eval_loss": 1.7989356517791748,
|
| 2150 |
+
"eval_masked_accuracy": 0.6666666865348816,
|
| 2151 |
+
"eval_runtime": 1.754,
|
| 2152 |
+
"eval_samples_per_second": 5.701,
|
| 2153 |
+
"eval_steps_per_second": 2.28,
|
| 2154 |
+
"step": 6700
|
| 2155 |
+
},
|
| 2156 |
+
{
|
| 2157 |
+
"epoch": 0.011329476276076679,
|
| 2158 |
+
"grad_norm": 5.595826148986816,
|
| 2159 |
+
"learning_rate": 4.9433861874382124e-05,
|
| 2160 |
+
"loss": 1.6367,
|
| 2161 |
+
"step": 6750
|
| 2162 |
+
},
|
| 2163 |
+
{
|
| 2164 |
+
"epoch": 0.011329476276076679,
|
| 2165 |
+
"eval_loss": 1.7425569295883179,
|
| 2166 |
+
"eval_masked_accuracy": 0.6583333611488342,
|
| 2167 |
+
"eval_runtime": 1.7467,
|
| 2168 |
+
"eval_samples_per_second": 5.725,
|
| 2169 |
+
"eval_steps_per_second": 2.29,
|
| 2170 |
+
"step": 6750
|
| 2171 |
+
},
|
| 2172 |
+
{
|
| 2173 |
+
"epoch": 0.011413398322566135,
|
| 2174 |
+
"grad_norm": 4.125125408172607,
|
| 2175 |
+
"learning_rate": 4.942966577205765e-05,
|
| 2176 |
+
"loss": 1.641,
|
| 2177 |
+
"step": 6800
|
| 2178 |
+
},
|
| 2179 |
+
{
|
| 2180 |
+
"epoch": 0.011413398322566135,
|
| 2181 |
+
"eval_loss": 1.728715181350708,
|
| 2182 |
+
"eval_masked_accuracy": 0.6652892827987671,
|
| 2183 |
+
"eval_runtime": 1.772,
|
| 2184 |
+
"eval_samples_per_second": 5.643,
|
| 2185 |
+
"eval_steps_per_second": 2.257,
|
| 2186 |
+
"step": 6800
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 0.011497320369055592,
|
| 2190 |
+
"grad_norm": 6.3898844718933105,
|
| 2191 |
+
"learning_rate": 4.942546966973318e-05,
|
| 2192 |
+
"loss": 1.6574,
|
| 2193 |
+
"step": 6850
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 0.011497320369055592,
|
| 2197 |
+
"eval_loss": 1.8261781930923462,
|
| 2198 |
+
"eval_masked_accuracy": 0.6752136945724487,
|
| 2199 |
+
"eval_runtime": 1.7446,
|
| 2200 |
+
"eval_samples_per_second": 5.732,
|
| 2201 |
+
"eval_steps_per_second": 2.293,
|
| 2202 |
+
"step": 6850
|
| 2203 |
+
},
|
| 2204 |
+
{
|
| 2205 |
+
"epoch": 0.011581242415545048,
|
| 2206 |
+
"grad_norm": 5.9191155433654785,
|
| 2207 |
+
"learning_rate": 4.942127356740871e-05,
|
| 2208 |
+
"loss": 1.5732,
|
| 2209 |
+
"step": 6900
|
| 2210 |
+
},
|
| 2211 |
+
{
|
| 2212 |
+
"epoch": 0.011581242415545048,
|
| 2213 |
+
"eval_loss": 1.2290430068969727,
|
| 2214 |
+
"eval_masked_accuracy": 0.7573221921920776,
|
| 2215 |
+
"eval_runtime": 1.7438,
|
| 2216 |
+
"eval_samples_per_second": 5.735,
|
| 2217 |
+
"eval_steps_per_second": 2.294,
|
| 2218 |
+
"step": 6900
|
| 2219 |
+
},
|
| 2220 |
+
{
|
| 2221 |
+
"epoch": 0.011665164462034505,
|
| 2222 |
+
"grad_norm": 5.910600185394287,
|
| 2223 |
+
"learning_rate": 4.941707746508423e-05,
|
| 2224 |
+
"loss": 1.5018,
|
| 2225 |
+
"step": 6950
|
| 2226 |
+
},
|
| 2227 |
+
{
|
| 2228 |
+
"epoch": 0.011665164462034505,
|
| 2229 |
+
"eval_loss": 1.3011202812194824,
|
| 2230 |
+
"eval_masked_accuracy": 0.746835470199585,
|
| 2231 |
+
"eval_runtime": 1.739,
|
| 2232 |
+
"eval_samples_per_second": 5.751,
|
| 2233 |
+
"eval_steps_per_second": 2.3,
|
| 2234 |
+
"step": 6950
|
| 2235 |
+
},
|
| 2236 |
+
{
|
| 2237 |
+
"epoch": 0.011749086508523963,
|
| 2238 |
+
"grad_norm": 7.273187637329102,
|
| 2239 |
+
"learning_rate": 4.941288136275976e-05,
|
| 2240 |
+
"loss": 1.6083,
|
| 2241 |
+
"step": 7000
|
| 2242 |
+
},
|
| 2243 |
+
{
|
| 2244 |
+
"epoch": 0.011749086508523963,
|
| 2245 |
+
"eval_loss": 1.7945482730865479,
|
| 2246 |
+
"eval_masked_accuracy": 0.6719367504119873,
|
| 2247 |
+
"eval_runtime": 1.7495,
|
| 2248 |
+
"eval_samples_per_second": 5.716,
|
| 2249 |
+
"eval_steps_per_second": 2.286,
|
| 2250 |
+
"step": 7000
|
| 2251 |
+
},
|
| 2252 |
+
{
|
| 2253 |
+
"epoch": 0.01183300855501342,
|
| 2254 |
+
"grad_norm": 5.980038642883301,
|
| 2255 |
+
"learning_rate": 4.940868526043529e-05,
|
| 2256 |
+
"loss": 1.7157,
|
| 2257 |
+
"step": 7050
|
| 2258 |
+
},
|
| 2259 |
+
{
|
| 2260 |
+
"epoch": 0.01183300855501342,
|
| 2261 |
+
"eval_loss": 1.6633656024932861,
|
| 2262 |
+
"eval_masked_accuracy": 0.6859503984451294,
|
| 2263 |
+
"eval_runtime": 1.7603,
|
| 2264 |
+
"eval_samples_per_second": 5.681,
|
| 2265 |
+
"eval_steps_per_second": 2.272,
|
| 2266 |
+
"step": 7050
|
| 2267 |
+
},
|
| 2268 |
+
{
|
| 2269 |
+
"epoch": 0.011916930601502876,
|
| 2270 |
+
"grad_norm": 4.222002029418945,
|
| 2271 |
+
"learning_rate": 4.9404489158110817e-05,
|
| 2272 |
+
"loss": 1.4124,
|
| 2273 |
+
"step": 7100
|
| 2274 |
+
},
|
| 2275 |
+
{
|
| 2276 |
+
"epoch": 0.011916930601502876,
|
| 2277 |
+
"eval_loss": 1.7207615375518799,
|
| 2278 |
+
"eval_masked_accuracy": 0.6793248653411865,
|
| 2279 |
+
"eval_runtime": 1.753,
|
| 2280 |
+
"eval_samples_per_second": 5.704,
|
| 2281 |
+
"eval_steps_per_second": 2.282,
|
| 2282 |
+
"step": 7100
|
| 2283 |
+
},
|
| 2284 |
+
{
|
| 2285 |
+
"epoch": 0.012000852647992333,
|
| 2286 |
+
"grad_norm": 8.79937744140625,
|
| 2287 |
+
"learning_rate": 4.9400293055786345e-05,
|
| 2288 |
+
"loss": 1.5698,
|
| 2289 |
+
"step": 7150
|
| 2290 |
+
},
|
| 2291 |
+
{
|
| 2292 |
+
"epoch": 0.012000852647992333,
|
| 2293 |
+
"eval_loss": 1.5078874826431274,
|
| 2294 |
+
"eval_masked_accuracy": 0.7276119589805603,
|
| 2295 |
+
"eval_runtime": 1.866,
|
| 2296 |
+
"eval_samples_per_second": 5.359,
|
| 2297 |
+
"eval_steps_per_second": 2.144,
|
| 2298 |
+
"step": 7150
|
| 2299 |
+
},
|
| 2300 |
+
{
|
| 2301 |
+
"epoch": 0.01208477469448179,
|
| 2302 |
+
"grad_norm": 6.331279754638672,
|
| 2303 |
+
"learning_rate": 4.939609695346187e-05,
|
| 2304 |
+
"loss": 1.5354,
|
| 2305 |
+
"step": 7200
|
| 2306 |
+
},
|
| 2307 |
+
{
|
| 2308 |
+
"epoch": 0.01208477469448179,
|
| 2309 |
+
"eval_loss": 1.3983685970306396,
|
| 2310 |
+
"eval_masked_accuracy": 0.7590909004211426,
|
| 2311 |
+
"eval_runtime": 1.7632,
|
| 2312 |
+
"eval_samples_per_second": 5.672,
|
| 2313 |
+
"eval_steps_per_second": 2.269,
|
| 2314 |
+
"step": 7200
|
| 2315 |
+
},
|
| 2316 |
+
{
|
| 2317 |
+
"epoch": 0.012168696740971246,
|
| 2318 |
+
"grad_norm": 4.12935733795166,
|
| 2319 |
+
"learning_rate": 4.9391900851137395e-05,
|
| 2320 |
+
"loss": 1.4778,
|
| 2321 |
+
"step": 7250
|
| 2322 |
+
},
|
| 2323 |
+
{
|
| 2324 |
+
"epoch": 0.012168696740971246,
|
| 2325 |
+
"eval_loss": 1.7603422403335571,
|
| 2326 |
+
"eval_masked_accuracy": 0.686956524848938,
|
| 2327 |
+
"eval_runtime": 1.7504,
|
| 2328 |
+
"eval_samples_per_second": 5.713,
|
| 2329 |
+
"eval_steps_per_second": 2.285,
|
| 2330 |
+
"step": 7250
|
| 2331 |
+
},
|
| 2332 |
+
{
|
| 2333 |
+
"epoch": 0.012252618787460704,
|
| 2334 |
+
"grad_norm": 5.025778293609619,
|
| 2335 |
+
"learning_rate": 4.9387704748812923e-05,
|
| 2336 |
+
"loss": 1.5175,
|
| 2337 |
+
"step": 7300
|
| 2338 |
+
},
|
| 2339 |
+
{
|
| 2340 |
+
"epoch": 0.012252618787460704,
|
| 2341 |
+
"eval_loss": 1.7313247919082642,
|
| 2342 |
+
"eval_masked_accuracy": 0.6872428059577942,
|
| 2343 |
+
"eval_runtime": 1.745,
|
| 2344 |
+
"eval_samples_per_second": 5.731,
|
| 2345 |
+
"eval_steps_per_second": 2.292,
|
| 2346 |
+
"step": 7300
|
| 2347 |
+
},
|
| 2348 |
+
{
|
| 2349 |
+
"epoch": 0.01233654083395016,
|
| 2350 |
+
"grad_norm": 9.704473495483398,
|
| 2351 |
+
"learning_rate": 4.938350864648845e-05,
|
| 2352 |
+
"loss": 1.4634,
|
| 2353 |
+
"step": 7350
|
| 2354 |
+
},
|
| 2355 |
+
{
|
| 2356 |
+
"epoch": 0.01233654083395016,
|
| 2357 |
+
"eval_loss": 1.271333932876587,
|
| 2358 |
+
"eval_masked_accuracy": 0.7397260069847107,
|
| 2359 |
+
"eval_runtime": 1.7484,
|
| 2360 |
+
"eval_samples_per_second": 5.72,
|
| 2361 |
+
"eval_steps_per_second": 2.288,
|
| 2362 |
+
"step": 7350
|
| 2363 |
+
},
|
| 2364 |
+
{
|
| 2365 |
+
"epoch": 0.012420462880439617,
|
| 2366 |
+
"grad_norm": 6.080599308013916,
|
| 2367 |
+
"learning_rate": 4.937931254416398e-05,
|
| 2368 |
+
"loss": 1.5937,
|
| 2369 |
+
"step": 7400
|
| 2370 |
+
},
|
| 2371 |
+
{
|
| 2372 |
+
"epoch": 0.012420462880439617,
|
| 2373 |
+
"eval_loss": 1.4850938320159912,
|
| 2374 |
+
"eval_masked_accuracy": 0.7280701994895935,
|
| 2375 |
+
"eval_runtime": 1.7517,
|
| 2376 |
+
"eval_samples_per_second": 5.709,
|
| 2377 |
+
"eval_steps_per_second": 2.284,
|
| 2378 |
+
"step": 7400
|
| 2379 |
+
},
|
| 2380 |
+
{
|
| 2381 |
+
"epoch": 0.012504384926929074,
|
| 2382 |
+
"grad_norm": 3.824946880340576,
|
| 2383 |
+
"learning_rate": 4.937511644183951e-05,
|
| 2384 |
+
"loss": 1.6026,
|
| 2385 |
+
"step": 7450
|
| 2386 |
+
},
|
| 2387 |
+
{
|
| 2388 |
+
"epoch": 0.012504384926929074,
|
| 2389 |
+
"eval_loss": 1.5267841815948486,
|
| 2390 |
+
"eval_masked_accuracy": 0.7058823704719543,
|
| 2391 |
+
"eval_runtime": 1.7438,
|
| 2392 |
+
"eval_samples_per_second": 5.734,
|
| 2393 |
+
"eval_steps_per_second": 2.294,
|
| 2394 |
+
"step": 7450
|
| 2395 |
+
},
|
| 2396 |
+
{
|
| 2397 |
+
"epoch": 0.01258830697341853,
|
| 2398 |
+
"grad_norm": 4.5395989418029785,
|
| 2399 |
+
"learning_rate": 4.937092033951503e-05,
|
| 2400 |
+
"loss": 1.4575,
|
| 2401 |
+
"step": 7500
|
| 2402 |
+
},
|
| 2403 |
+
{
|
| 2404 |
+
"epoch": 0.01258830697341853,
|
| 2405 |
+
"eval_loss": 1.4801056385040283,
|
| 2406 |
+
"eval_masked_accuracy": 0.680672287940979,
|
| 2407 |
+
"eval_runtime": 1.7409,
|
| 2408 |
+
"eval_samples_per_second": 5.744,
|
| 2409 |
+
"eval_steps_per_second": 2.298,
|
| 2410 |
+
"step": 7500
|
| 2411 |
+
},
|
| 2412 |
+
{
|
| 2413 |
+
"epoch": 0.012672229019907989,
|
| 2414 |
+
"grad_norm": 6.853204250335693,
|
| 2415 |
+
"learning_rate": 4.936672423719056e-05,
|
| 2416 |
+
"loss": 1.4224,
|
| 2417 |
+
"step": 7550
|
| 2418 |
+
},
|
| 2419 |
+
{
|
| 2420 |
+
"epoch": 0.012672229019907989,
|
| 2421 |
+
"eval_loss": 1.6892282962799072,
|
| 2422 |
+
"eval_masked_accuracy": 0.6551724076271057,
|
| 2423 |
+
"eval_runtime": 1.7414,
|
| 2424 |
+
"eval_samples_per_second": 5.742,
|
| 2425 |
+
"eval_steps_per_second": 2.297,
|
| 2426 |
+
"step": 7550
|
| 2427 |
+
},
|
| 2428 |
+
{
|
| 2429 |
+
"epoch": 0.012756151066397445,
|
| 2430 |
+
"grad_norm": 5.53077507019043,
|
| 2431 |
+
"learning_rate": 4.936252813486609e-05,
|
| 2432 |
+
"loss": 1.6706,
|
| 2433 |
+
"step": 7600
|
| 2434 |
+
},
|
| 2435 |
+
{
|
| 2436 |
+
"epoch": 0.012756151066397445,
|
| 2437 |
+
"eval_loss": 1.4235472679138184,
|
| 2438 |
+
"eval_masked_accuracy": 0.7426160573959351,
|
| 2439 |
+
"eval_runtime": 1.8082,
|
| 2440 |
+
"eval_samples_per_second": 5.53,
|
| 2441 |
+
"eval_steps_per_second": 2.212,
|
| 2442 |
+
"step": 7600
|
| 2443 |
+
},
|
| 2444 |
+
{
|
| 2445 |
+
"epoch": 0.012840073112886902,
|
| 2446 |
+
"grad_norm": 4.5907087326049805,
|
| 2447 |
+
"learning_rate": 4.9358332032541616e-05,
|
| 2448 |
+
"loss": 1.6674,
|
| 2449 |
+
"step": 7650
|
| 2450 |
+
},
|
| 2451 |
+
{
|
| 2452 |
+
"epoch": 0.012840073112886902,
|
| 2453 |
+
"eval_loss": 1.4942524433135986,
|
| 2454 |
+
"eval_masked_accuracy": 0.7172995805740356,
|
| 2455 |
+
"eval_runtime": 1.7449,
|
| 2456 |
+
"eval_samples_per_second": 5.731,
|
| 2457 |
+
"eval_steps_per_second": 2.292,
|
| 2458 |
+
"step": 7650
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"epoch": 0.012923995159376358,
|
| 2462 |
+
"grad_norm": 8.004353523254395,
|
| 2463 |
+
"learning_rate": 4.9354135930217144e-05,
|
| 2464 |
+
"loss": 1.4294,
|
| 2465 |
+
"step": 7700
|
| 2466 |
+
},
|
| 2467 |
+
{
|
| 2468 |
+
"epoch": 0.012923995159376358,
|
| 2469 |
+
"eval_loss": 1.7548024654388428,
|
| 2470 |
+
"eval_masked_accuracy": 0.6547619104385376,
|
| 2471 |
+
"eval_runtime": 1.7767,
|
| 2472 |
+
"eval_samples_per_second": 5.628,
|
| 2473 |
+
"eval_steps_per_second": 2.251,
|
| 2474 |
+
"step": 7700
|
| 2475 |
+
},
|
| 2476 |
+
{
|
| 2477 |
+
"epoch": 0.013007917205865815,
|
| 2478 |
+
"grad_norm": 6.963031768798828,
|
| 2479 |
+
"learning_rate": 4.934993982789267e-05,
|
| 2480 |
+
"loss": 1.5078,
|
| 2481 |
+
"step": 7750
|
| 2482 |
+
},
|
| 2483 |
+
{
|
| 2484 |
+
"epoch": 0.013007917205865815,
|
| 2485 |
+
"eval_loss": 1.4269187450408936,
|
| 2486 |
+
"eval_masked_accuracy": 0.7027027010917664,
|
| 2487 |
+
"eval_runtime": 1.7471,
|
| 2488 |
+
"eval_samples_per_second": 5.724,
|
| 2489 |
+
"eval_steps_per_second": 2.29,
|
| 2490 |
+
"step": 7750
|
| 2491 |
+
},
|
| 2492 |
+
{
|
| 2493 |
+
"epoch": 0.013091839252355273,
|
| 2494 |
+
"grad_norm": 6.4043288230896,
|
| 2495 |
+
"learning_rate": 4.9345743725568194e-05,
|
| 2496 |
+
"loss": 1.604,
|
| 2497 |
+
"step": 7800
|
| 2498 |
+
},
|
| 2499 |
+
{
|
| 2500 |
+
"epoch": 0.013091839252355273,
|
| 2501 |
+
"eval_loss": 1.4502145051956177,
|
| 2502 |
+
"eval_masked_accuracy": 0.7172995805740356,
|
| 2503 |
+
"eval_runtime": 1.748,
|
| 2504 |
+
"eval_samples_per_second": 5.721,
|
| 2505 |
+
"eval_steps_per_second": 2.288,
|
| 2506 |
+
"step": 7800
|
| 2507 |
+
},
|
| 2508 |
+
{
|
| 2509 |
+
"epoch": 0.01317576129884473,
|
| 2510 |
+
"grad_norm": 5.293691158294678,
|
| 2511 |
+
"learning_rate": 4.934154762324372e-05,
|
| 2512 |
+
"loss": 1.6301,
|
| 2513 |
+
"step": 7850
|
| 2514 |
+
},
|
| 2515 |
+
{
|
| 2516 |
+
"epoch": 0.01317576129884473,
|
| 2517 |
+
"eval_loss": 1.3547624349594116,
|
| 2518 |
+
"eval_masked_accuracy": 0.7759336233139038,
|
| 2519 |
+
"eval_runtime": 1.7437,
|
| 2520 |
+
"eval_samples_per_second": 5.735,
|
| 2521 |
+
"eval_steps_per_second": 2.294,
|
| 2522 |
+
"step": 7850
|
| 2523 |
+
},
|
| 2524 |
+
{
|
| 2525 |
+
"epoch": 0.013259683345334186,
|
| 2526 |
+
"grad_norm": 7.364100933074951,
|
| 2527 |
+
"learning_rate": 4.933735152091925e-05,
|
| 2528 |
+
"loss": 1.5163,
|
| 2529 |
+
"step": 7900
|
| 2530 |
+
},
|
| 2531 |
+
{
|
| 2532 |
+
"epoch": 0.013259683345334186,
|
| 2533 |
+
"eval_loss": 1.6089417934417725,
|
| 2534 |
+
"eval_masked_accuracy": 0.6610878705978394,
|
| 2535 |
+
"eval_runtime": 1.753,
|
| 2536 |
+
"eval_samples_per_second": 5.704,
|
| 2537 |
+
"eval_steps_per_second": 2.282,
|
| 2538 |
+
"step": 7900
|
| 2539 |
+
},
|
| 2540 |
+
{
|
| 2541 |
+
"epoch": 0.013343605391823643,
|
| 2542 |
+
"grad_norm": 7.704033851623535,
|
| 2543 |
+
"learning_rate": 4.933315541859478e-05,
|
| 2544 |
+
"loss": 1.6564,
|
| 2545 |
+
"step": 7950
|
| 2546 |
+
},
|
| 2547 |
+
{
|
| 2548 |
+
"epoch": 0.013343605391823643,
|
| 2549 |
+
"eval_loss": 1.4759953022003174,
|
| 2550 |
+
"eval_masked_accuracy": 0.6958333253860474,
|
| 2551 |
+
"eval_runtime": 1.7614,
|
| 2552 |
+
"eval_samples_per_second": 5.677,
|
| 2553 |
+
"eval_steps_per_second": 2.271,
|
| 2554 |
+
"step": 7950
|
| 2555 |
+
},
|
| 2556 |
+
{
|
| 2557 |
+
"epoch": 0.0134275274383131,
|
| 2558 |
+
"grad_norm": 5.562460899353027,
|
| 2559 |
+
"learning_rate": 4.932895931627031e-05,
|
| 2560 |
+
"loss": 1.5703,
|
| 2561 |
+
"step": 8000
|
| 2562 |
+
},
|
| 2563 |
+
{
|
| 2564 |
+
"epoch": 0.0134275274383131,
|
| 2565 |
+
"eval_loss": 1.735896348953247,
|
| 2566 |
+
"eval_masked_accuracy": 0.6875,
|
| 2567 |
+
"eval_runtime": 1.7493,
|
| 2568 |
+
"eval_samples_per_second": 5.717,
|
| 2569 |
+
"eval_steps_per_second": 2.287,
|
| 2570 |
+
"step": 8000
|
| 2571 |
+
},
|
| 2572 |
+
{
|
| 2573 |
+
"epoch": 0.013511449484802556,
|
| 2574 |
+
"grad_norm": 8.801225662231445,
|
| 2575 |
+
"learning_rate": 4.9324763213945836e-05,
|
| 2576 |
+
"loss": 1.5328,
|
| 2577 |
+
"step": 8050
|
| 2578 |
+
},
|
| 2579 |
+
{
|
| 2580 |
+
"epoch": 0.013511449484802556,
|
| 2581 |
+
"eval_loss": 1.2792503833770752,
|
| 2582 |
+
"eval_masked_accuracy": 0.7292576432228088,
|
| 2583 |
+
"eval_runtime": 1.7802,
|
| 2584 |
+
"eval_samples_per_second": 5.617,
|
| 2585 |
+
"eval_steps_per_second": 2.247,
|
| 2586 |
+
"step": 8050
|
| 2587 |
+
},
|
| 2588 |
+
{
|
| 2589 |
+
"epoch": 0.013595371531292014,
|
| 2590 |
+
"grad_norm": 5.510076999664307,
|
| 2591 |
+
"learning_rate": 4.932056711162136e-05,
|
| 2592 |
+
"loss": 1.5086,
|
| 2593 |
+
"step": 8100
|
| 2594 |
+
},
|
| 2595 |
+
{
|
| 2596 |
+
"epoch": 0.013595371531292014,
|
| 2597 |
+
"eval_loss": 1.811342477798462,
|
| 2598 |
+
"eval_masked_accuracy": 0.6508620977401733,
|
| 2599 |
+
"eval_runtime": 1.7772,
|
| 2600 |
+
"eval_samples_per_second": 5.627,
|
| 2601 |
+
"eval_steps_per_second": 2.251,
|
| 2602 |
+
"step": 8100
|
| 2603 |
+
},
|
| 2604 |
+
{
|
| 2605 |
+
"epoch": 0.01367929357778147,
|
| 2606 |
+
"grad_norm": 4.370019912719727,
|
| 2607 |
+
"learning_rate": 4.9316371009296886e-05,
|
| 2608 |
+
"loss": 1.5992,
|
| 2609 |
+
"step": 8150
|
| 2610 |
+
},
|
| 2611 |
+
{
|
| 2612 |
+
"epoch": 0.01367929357778147,
|
| 2613 |
+
"eval_loss": 1.7015224695205688,
|
| 2614 |
+
"eval_masked_accuracy": 0.6945606470108032,
|
| 2615 |
+
"eval_runtime": 1.7399,
|
| 2616 |
+
"eval_samples_per_second": 5.747,
|
| 2617 |
+
"eval_steps_per_second": 2.299,
|
| 2618 |
+
"step": 8150
|
| 2619 |
+
},
|
| 2620 |
+
{
|
| 2621 |
+
"epoch": 0.013763215624270927,
|
| 2622 |
+
"grad_norm": 5.960280895233154,
|
| 2623 |
+
"learning_rate": 4.9312174906972415e-05,
|
| 2624 |
+
"loss": 1.6392,
|
| 2625 |
+
"step": 8200
|
| 2626 |
+
},
|
| 2627 |
+
{
|
| 2628 |
+
"epoch": 0.013763215624270927,
|
| 2629 |
+
"eval_loss": 1.4644631147384644,
|
| 2630 |
+
"eval_masked_accuracy": 0.7004830837249756,
|
| 2631 |
+
"eval_runtime": 1.7493,
|
| 2632 |
+
"eval_samples_per_second": 5.717,
|
| 2633 |
+
"eval_steps_per_second": 2.287,
|
| 2634 |
+
"step": 8200
|
| 2635 |
+
},
|
| 2636 |
+
{
|
| 2637 |
+
"epoch": 0.013847137670760384,
|
| 2638 |
+
"grad_norm": 5.401033878326416,
|
| 2639 |
+
"learning_rate": 4.930797880464794e-05,
|
| 2640 |
+
"loss": 1.6492,
|
| 2641 |
+
"step": 8250
|
| 2642 |
+
},
|
| 2643 |
+
{
|
| 2644 |
+
"epoch": 0.013847137670760384,
|
| 2645 |
+
"eval_loss": 1.5244245529174805,
|
| 2646 |
+
"eval_masked_accuracy": 0.688034176826477,
|
| 2647 |
+
"eval_runtime": 1.7597,
|
| 2648 |
+
"eval_samples_per_second": 5.683,
|
| 2649 |
+
"eval_steps_per_second": 2.273,
|
| 2650 |
+
"step": 8250
|
| 2651 |
+
},
|
| 2652 |
+
{
|
| 2653 |
+
"epoch": 0.01393105971724984,
|
| 2654 |
+
"grad_norm": 7.356916427612305,
|
| 2655 |
+
"learning_rate": 4.930378270232347e-05,
|
| 2656 |
+
"loss": 1.5673,
|
| 2657 |
+
"step": 8300
|
| 2658 |
+
},
|
| 2659 |
+
{
|
| 2660 |
+
"epoch": 0.01393105971724984,
|
| 2661 |
+
"eval_loss": 1.4024368524551392,
|
| 2662 |
+
"eval_masked_accuracy": 0.7016806602478027,
|
| 2663 |
+
"eval_runtime": 1.7463,
|
| 2664 |
+
"eval_samples_per_second": 5.726,
|
| 2665 |
+
"eval_steps_per_second": 2.291,
|
| 2666 |
+
"step": 8300
|
| 2667 |
+
},
|
| 2668 |
+
{
|
| 2669 |
+
"epoch": 0.014014981763739299,
|
| 2670 |
+
"grad_norm": 5.370472431182861,
|
| 2671 |
+
"learning_rate": 4.929958659999899e-05,
|
| 2672 |
+
"loss": 1.5267,
|
| 2673 |
+
"step": 8350
|
| 2674 |
+
},
|
| 2675 |
+
{
|
| 2676 |
+
"epoch": 0.014014981763739299,
|
| 2677 |
+
"eval_loss": 1.7430174350738525,
|
| 2678 |
+
"eval_masked_accuracy": 0.6653061509132385,
|
| 2679 |
+
"eval_runtime": 1.7353,
|
| 2680 |
+
"eval_samples_per_second": 5.763,
|
| 2681 |
+
"eval_steps_per_second": 2.305,
|
| 2682 |
+
"step": 8350
|
| 2683 |
+
},
|
| 2684 |
+
{
|
| 2685 |
+
"epoch": 0.014098903810228755,
|
| 2686 |
+
"grad_norm": 6.4656500816345215,
|
| 2687 |
+
"learning_rate": 4.929539049767452e-05,
|
| 2688 |
+
"loss": 1.5918,
|
| 2689 |
+
"step": 8400
|
| 2690 |
+
},
|
| 2691 |
+
{
|
| 2692 |
+
"epoch": 0.014098903810228755,
|
| 2693 |
+
"eval_loss": 1.691054344177246,
|
| 2694 |
+
"eval_masked_accuracy": 0.6849315166473389,
|
| 2695 |
+
"eval_runtime": 1.746,
|
| 2696 |
+
"eval_samples_per_second": 5.727,
|
| 2697 |
+
"eval_steps_per_second": 2.291,
|
| 2698 |
+
"step": 8400
|
| 2699 |
+
},
|
| 2700 |
+
{
|
| 2701 |
+
"epoch": 0.014182825856718212,
|
| 2702 |
+
"grad_norm": 5.481358051300049,
|
| 2703 |
+
"learning_rate": 4.929119439535005e-05,
|
| 2704 |
+
"loss": 1.5156,
|
| 2705 |
+
"step": 8450
|
| 2706 |
+
},
|
| 2707 |
+
{
|
| 2708 |
+
"epoch": 0.014182825856718212,
|
| 2709 |
+
"eval_loss": 1.6469824314117432,
|
| 2710 |
+
"eval_masked_accuracy": 0.6516393423080444,
|
| 2711 |
+
"eval_runtime": 1.7423,
|
| 2712 |
+
"eval_samples_per_second": 5.74,
|
| 2713 |
+
"eval_steps_per_second": 2.296,
|
| 2714 |
+
"step": 8450
|
| 2715 |
+
},
|
| 2716 |
+
{
|
| 2717 |
+
"epoch": 0.014266747903207668,
|
| 2718 |
+
"grad_norm": 4.755044937133789,
|
| 2719 |
+
"learning_rate": 4.928699829302558e-05,
|
| 2720 |
+
"loss": 1.5223,
|
| 2721 |
+
"step": 8500
|
| 2722 |
+
},
|
| 2723 |
+
{
|
| 2724 |
+
"epoch": 0.014266747903207668,
|
| 2725 |
+
"eval_loss": 1.667824387550354,
|
| 2726 |
+
"eval_masked_accuracy": 0.7068965435028076,
|
| 2727 |
+
"eval_runtime": 1.7979,
|
| 2728 |
+
"eval_samples_per_second": 5.562,
|
| 2729 |
+
"eval_steps_per_second": 2.225,
|
| 2730 |
+
"step": 8500
|
| 2731 |
+
},
|
| 2732 |
+
{
|
| 2733 |
+
"epoch": 0.014350669949697125,
|
| 2734 |
+
"grad_norm": 6.595943450927734,
|
| 2735 |
+
"learning_rate": 4.928280219070111e-05,
|
| 2736 |
+
"loss": 1.4699,
|
| 2737 |
+
"step": 8550
|
| 2738 |
+
},
|
| 2739 |
+
{
|
| 2740 |
+
"epoch": 0.014350669949697125,
|
| 2741 |
+
"eval_loss": 1.2367641925811768,
|
| 2742 |
+
"eval_masked_accuracy": 0.7447698712348938,
|
| 2743 |
+
"eval_runtime": 1.7387,
|
| 2744 |
+
"eval_samples_per_second": 5.752,
|
| 2745 |
+
"eval_steps_per_second": 2.301,
|
| 2746 |
+
"step": 8550
|
| 2747 |
+
},
|
| 2748 |
+
{
|
| 2749 |
+
"epoch": 0.014434591996186583,
|
| 2750 |
+
"grad_norm": 3.9210710525512695,
|
| 2751 |
+
"learning_rate": 4.9278606088376635e-05,
|
| 2752 |
+
"loss": 1.5695,
|
| 2753 |
+
"step": 8600
|
| 2754 |
+
},
|
| 2755 |
+
{
|
| 2756 |
+
"epoch": 0.014434591996186583,
|
| 2757 |
+
"eval_loss": 1.3033006191253662,
|
| 2758 |
+
"eval_masked_accuracy": 0.693965494632721,
|
| 2759 |
+
"eval_runtime": 1.7554,
|
| 2760 |
+
"eval_samples_per_second": 5.697,
|
| 2761 |
+
"eval_steps_per_second": 2.279,
|
| 2762 |
+
"step": 8600
|
| 2763 |
+
},
|
| 2764 |
+
{
|
| 2765 |
+
"epoch": 0.01451851404267604,
|
| 2766 |
+
"grad_norm": 4.682461261749268,
|
| 2767 |
+
"learning_rate": 4.927440998605216e-05,
|
| 2768 |
+
"loss": 1.5371,
|
| 2769 |
+
"step": 8650
|
| 2770 |
+
},
|
| 2771 |
+
{
|
| 2772 |
+
"epoch": 0.01451851404267604,
|
| 2773 |
+
"eval_loss": 1.727216124534607,
|
| 2774 |
+
"eval_masked_accuracy": 0.6639004349708557,
|
| 2775 |
+
"eval_runtime": 1.7387,
|
| 2776 |
+
"eval_samples_per_second": 5.751,
|
| 2777 |
+
"eval_steps_per_second": 2.301,
|
| 2778 |
+
"step": 8650
|
| 2779 |
+
},
|
| 2780 |
+
{
|
| 2781 |
+
"epoch": 0.014602436089165496,
|
| 2782 |
+
"grad_norm": 4.478100776672363,
|
| 2783 |
+
"learning_rate": 4.9270213883727685e-05,
|
| 2784 |
+
"loss": 1.5679,
|
| 2785 |
+
"step": 8700
|
| 2786 |
+
},
|
| 2787 |
+
{
|
| 2788 |
+
"epoch": 0.014602436089165496,
|
| 2789 |
+
"eval_loss": 1.4694969654083252,
|
| 2790 |
+
"eval_masked_accuracy": 0.7364016771316528,
|
| 2791 |
+
"eval_runtime": 1.7474,
|
| 2792 |
+
"eval_samples_per_second": 5.723,
|
| 2793 |
+
"eval_steps_per_second": 2.289,
|
| 2794 |
+
"step": 8700
|
| 2795 |
+
},
|
| 2796 |
+
{
|
| 2797 |
+
"epoch": 0.014686358135654953,
|
| 2798 |
+
"grad_norm": 8.149710655212402,
|
| 2799 |
+
"learning_rate": 4.9266017781403214e-05,
|
| 2800 |
+
"loss": 1.4814,
|
| 2801 |
+
"step": 8750
|
| 2802 |
+
},
|
| 2803 |
+
{
|
| 2804 |
+
"epoch": 0.014686358135654953,
|
| 2805 |
+
"eval_loss": 1.9258610010147095,
|
| 2806 |
+
"eval_masked_accuracy": 0.6228070259094238,
|
| 2807 |
+
"eval_runtime": 1.7513,
|
| 2808 |
+
"eval_samples_per_second": 5.71,
|
| 2809 |
+
"eval_steps_per_second": 2.284,
|
| 2810 |
+
"step": 8750
|
| 2811 |
+
},
|
| 2812 |
+
{
|
| 2813 |
+
"epoch": 0.01477028018214441,
|
| 2814 |
+
"grad_norm": 4.727016925811768,
|
| 2815 |
+
"learning_rate": 4.926182167907874e-05,
|
| 2816 |
+
"loss": 1.609,
|
| 2817 |
+
"step": 8800
|
| 2818 |
+
},
|
| 2819 |
+
{
|
| 2820 |
+
"epoch": 0.01477028018214441,
|
| 2821 |
+
"eval_loss": 1.6111774444580078,
|
| 2822 |
+
"eval_masked_accuracy": 0.6590038537979126,
|
| 2823 |
+
"eval_runtime": 1.7579,
|
| 2824 |
+
"eval_samples_per_second": 5.689,
|
| 2825 |
+
"eval_steps_per_second": 2.275,
|
| 2826 |
+
"step": 8800
|
| 2827 |
+
},
|
| 2828 |
+
{
|
| 2829 |
+
"epoch": 0.014854202228633867,
|
| 2830 |
+
"grad_norm": 5.348945140838623,
|
| 2831 |
+
"learning_rate": 4.925762557675427e-05,
|
| 2832 |
+
"loss": 1.5557,
|
| 2833 |
+
"step": 8850
|
| 2834 |
+
},
|
| 2835 |
+
{
|
| 2836 |
+
"epoch": 0.014854202228633867,
|
| 2837 |
+
"eval_loss": 1.3535053730010986,
|
| 2838 |
+
"eval_masked_accuracy": 0.7245762944221497,
|
| 2839 |
+
"eval_runtime": 1.8639,
|
| 2840 |
+
"eval_samples_per_second": 5.365,
|
| 2841 |
+
"eval_steps_per_second": 2.146,
|
| 2842 |
+
"step": 8850
|
| 2843 |
+
},
|
| 2844 |
+
{
|
| 2845 |
+
"epoch": 0.014938124275123324,
|
| 2846 |
+
"grad_norm": 6.573589324951172,
|
| 2847 |
+
"learning_rate": 4.92534294744298e-05,
|
| 2848 |
+
"loss": 1.6389,
|
| 2849 |
+
"step": 8900
|
| 2850 |
+
},
|
| 2851 |
+
{
|
| 2852 |
+
"epoch": 0.014938124275123324,
|
| 2853 |
+
"eval_loss": 1.8509418964385986,
|
| 2854 |
+
"eval_masked_accuracy": 0.7085201740264893,
|
| 2855 |
+
"eval_runtime": 1.7536,
|
| 2856 |
+
"eval_samples_per_second": 5.703,
|
| 2857 |
+
"eval_steps_per_second": 2.281,
|
| 2858 |
+
"step": 8900
|
| 2859 |
+
},
|
| 2860 |
+
{
|
| 2861 |
+
"epoch": 0.01502204632161278,
|
| 2862 |
+
"grad_norm": 7.373574256896973,
|
| 2863 |
+
"learning_rate": 4.924923337210532e-05,
|
| 2864 |
+
"loss": 1.4773,
|
| 2865 |
+
"step": 8950
|
| 2866 |
+
},
|
| 2867 |
+
{
|
| 2868 |
+
"epoch": 0.01502204632161278,
|
| 2869 |
+
"eval_loss": 1.7772554159164429,
|
| 2870 |
+
"eval_masked_accuracy": 0.6640625,
|
| 2871 |
+
"eval_runtime": 1.7655,
|
| 2872 |
+
"eval_samples_per_second": 5.664,
|
| 2873 |
+
"eval_steps_per_second": 2.266,
|
| 2874 |
+
"step": 8950
|
| 2875 |
+
},
|
| 2876 |
+
{
|
| 2877 |
+
"epoch": 0.015105968368102237,
|
| 2878 |
+
"grad_norm": 5.861003875732422,
|
| 2879 |
+
"learning_rate": 4.924503726978085e-05,
|
| 2880 |
+
"loss": 1.3842,
|
| 2881 |
+
"step": 9000
|
| 2882 |
+
},
|
| 2883 |
+
{
|
| 2884 |
+
"epoch": 0.015105968368102237,
|
| 2885 |
+
"eval_loss": 1.6182334423065186,
|
| 2886 |
+
"eval_masked_accuracy": 0.7183098793029785,
|
| 2887 |
+
"eval_runtime": 1.7386,
|
| 2888 |
+
"eval_samples_per_second": 5.752,
|
| 2889 |
+
"eval_steps_per_second": 2.301,
|
| 2890 |
+
"step": 9000
|
| 2891 |
+
},
|
| 2892 |
+
{
|
| 2893 |
+
"epoch": 0.015189890414591694,
|
| 2894 |
+
"grad_norm": 5.086306571960449,
|
| 2895 |
+
"learning_rate": 4.924084116745638e-05,
|
| 2896 |
+
"loss": 1.6445,
|
| 2897 |
+
"step": 9050
|
| 2898 |
+
},
|
| 2899 |
+
{
|
| 2900 |
+
"epoch": 0.015189890414591694,
|
| 2901 |
+
"eval_loss": 1.3457679748535156,
|
| 2902 |
+
"eval_masked_accuracy": 0.752293586730957,
|
| 2903 |
+
"eval_runtime": 1.7595,
|
| 2904 |
+
"eval_samples_per_second": 5.684,
|
| 2905 |
+
"eval_steps_per_second": 2.273,
|
| 2906 |
+
"step": 9050
|
| 2907 |
+
},
|
| 2908 |
+
{
|
| 2909 |
+
"epoch": 0.01527381246108115,
|
| 2910 |
+
"grad_norm": 7.099021911621094,
|
| 2911 |
+
"learning_rate": 4.9236645065131906e-05,
|
| 2912 |
+
"loss": 1.5536,
|
| 2913 |
+
"step": 9100
|
| 2914 |
+
},
|
| 2915 |
+
{
|
| 2916 |
+
"epoch": 0.01527381246108115,
|
| 2917 |
+
"eval_loss": 1.8317623138427734,
|
| 2918 |
+
"eval_masked_accuracy": 0.6588628888130188,
|
| 2919 |
+
"eval_runtime": 1.8424,
|
| 2920 |
+
"eval_samples_per_second": 5.428,
|
| 2921 |
+
"eval_steps_per_second": 2.171,
|
| 2922 |
+
"step": 9100
|
| 2923 |
+
},
|
| 2924 |
+
{
|
| 2925 |
+
"epoch": 0.015357734507570608,
|
| 2926 |
+
"grad_norm": 6.620283126831055,
|
| 2927 |
+
"learning_rate": 4.9232448962807434e-05,
|
| 2928 |
+
"loss": 1.5151,
|
| 2929 |
+
"step": 9150
|
| 2930 |
+
},
|
| 2931 |
+
{
|
| 2932 |
+
"epoch": 0.015357734507570608,
|
| 2933 |
+
"eval_loss": 1.4230843782424927,
|
| 2934 |
+
"eval_masked_accuracy": 0.700421929359436,
|
| 2935 |
+
"eval_runtime": 1.7611,
|
| 2936 |
+
"eval_samples_per_second": 5.678,
|
| 2937 |
+
"eval_steps_per_second": 2.271,
|
| 2938 |
+
"step": 9150
|
| 2939 |
+
},
|
| 2940 |
+
{
|
| 2941 |
+
"epoch": 0.015441656554060065,
|
| 2942 |
+
"grad_norm": 7.231357097625732,
|
| 2943 |
+
"learning_rate": 4.922825286048296e-05,
|
| 2944 |
+
"loss": 1.6078,
|
| 2945 |
+
"step": 9200
|
| 2946 |
+
},
|
| 2947 |
+
{
|
| 2948 |
+
"epoch": 0.015441656554060065,
|
| 2949 |
+
"eval_loss": 1.7547998428344727,
|
| 2950 |
+
"eval_masked_accuracy": 0.6745283007621765,
|
| 2951 |
+
"eval_runtime": 1.8328,
|
| 2952 |
+
"eval_samples_per_second": 5.456,
|
| 2953 |
+
"eval_steps_per_second": 2.182,
|
| 2954 |
+
"step": 9200
|
| 2955 |
+
},
|
| 2956 |
+
{
|
| 2957 |
+
"epoch": 0.015525578600549522,
|
| 2958 |
+
"grad_norm": 4.755532264709473,
|
| 2959 |
+
"learning_rate": 4.9224140680204975e-05,
|
| 2960 |
+
"loss": 1.5938,
|
| 2961 |
+
"step": 9250
|
| 2962 |
+
},
|
| 2963 |
+
{
|
| 2964 |
+
"epoch": 0.015525578600549522,
|
| 2965 |
+
"eval_loss": 1.3346257209777832,
|
| 2966 |
+
"eval_masked_accuracy": 0.7244444489479065,
|
| 2967 |
+
"eval_runtime": 1.7553,
|
| 2968 |
+
"eval_samples_per_second": 5.697,
|
| 2969 |
+
"eval_steps_per_second": 2.279,
|
| 2970 |
+
"step": 9250
|
| 2971 |
+
},
|
| 2972 |
+
{
|
| 2973 |
+
"epoch": 0.015609500647038978,
|
| 2974 |
+
"grad_norm": 5.728196620941162,
|
| 2975 |
+
"learning_rate": 4.92199445778805e-05,
|
| 2976 |
+
"loss": 1.5542,
|
| 2977 |
+
"step": 9300
|
| 2978 |
+
},
|
| 2979 |
+
{
|
| 2980 |
+
"epoch": 0.015609500647038978,
|
| 2981 |
+
"eval_loss": 1.6833394765853882,
|
| 2982 |
+
"eval_masked_accuracy": 0.6654929518699646,
|
| 2983 |
+
"eval_runtime": 1.7516,
|
| 2984 |
+
"eval_samples_per_second": 5.709,
|
| 2985 |
+
"eval_steps_per_second": 2.284,
|
| 2986 |
+
"step": 9300
|
| 2987 |
+
},
|
| 2988 |
+
{
|
| 2989 |
+
"epoch": 0.015693422693528435,
|
| 2990 |
+
"grad_norm": 5.66224479675293,
|
| 2991 |
+
"learning_rate": 4.921574847555603e-05,
|
| 2992 |
+
"loss": 1.6099,
|
| 2993 |
+
"step": 9350
|
| 2994 |
+
},
|
| 2995 |
+
{
|
| 2996 |
+
"epoch": 0.015693422693528435,
|
| 2997 |
+
"eval_loss": 1.442452311515808,
|
| 2998 |
+
"eval_masked_accuracy": 0.6905829310417175,
|
| 2999 |
+
"eval_runtime": 1.7553,
|
| 3000 |
+
"eval_samples_per_second": 5.697,
|
| 3001 |
+
"eval_steps_per_second": 2.279,
|
| 3002 |
+
"step": 9350
|
| 3003 |
+
},
|
| 3004 |
+
{
|
| 3005 |
+
"epoch": 0.015777344740017893,
|
| 3006 |
+
"grad_norm": 6.560795307159424,
|
| 3007 |
+
"learning_rate": 4.921155237323155e-05,
|
| 3008 |
+
"loss": 1.4188,
|
| 3009 |
+
"step": 9400
|
| 3010 |
+
},
|
| 3011 |
+
{
|
| 3012 |
+
"epoch": 0.015777344740017893,
|
| 3013 |
+
"eval_loss": 1.539738416671753,
|
| 3014 |
+
"eval_masked_accuracy": 0.68359375,
|
| 3015 |
+
"eval_runtime": 1.7406,
|
| 3016 |
+
"eval_samples_per_second": 5.745,
|
| 3017 |
+
"eval_steps_per_second": 2.298,
|
| 3018 |
+
"step": 9400
|
| 3019 |
+
},
|
| 3020 |
+
{
|
| 3021 |
+
"epoch": 0.015861266786507348,
|
| 3022 |
+
"grad_norm": 4.9847025871276855,
|
| 3023 |
+
"learning_rate": 4.920735627090708e-05,
|
| 3024 |
+
"loss": 1.6344,
|
| 3025 |
+
"step": 9450
|
| 3026 |
+
},
|
| 3027 |
+
{
|
| 3028 |
+
"epoch": 0.015861266786507348,
|
| 3029 |
+
"eval_loss": 1.244769811630249,
|
| 3030 |
+
"eval_masked_accuracy": 0.7078189253807068,
|
| 3031 |
+
"eval_runtime": 1.77,
|
| 3032 |
+
"eval_samples_per_second": 5.65,
|
| 3033 |
+
"eval_steps_per_second": 2.26,
|
| 3034 |
+
"step": 9450
|
| 3035 |
+
},
|
| 3036 |
+
{
|
| 3037 |
+
"epoch": 0.015945188832996806,
|
| 3038 |
+
"grad_norm": 6.173788070678711,
|
| 3039 |
+
"learning_rate": 4.920316016858261e-05,
|
| 3040 |
+
"loss": 1.6249,
|
| 3041 |
+
"step": 9500
|
| 3042 |
+
},
|
| 3043 |
+
{
|
| 3044 |
+
"epoch": 0.015945188832996806,
|
| 3045 |
+
"eval_loss": 2.0483577251434326,
|
| 3046 |
+
"eval_masked_accuracy": 0.607594907283783,
|
| 3047 |
+
"eval_runtime": 1.7538,
|
| 3048 |
+
"eval_samples_per_second": 5.702,
|
| 3049 |
+
"eval_steps_per_second": 2.281,
|
| 3050 |
+
"step": 9500
|
| 3051 |
+
},
|
| 3052 |
+
{
|
| 3053 |
+
"epoch": 0.016029110879486264,
|
| 3054 |
+
"grad_norm": 4.4076828956604,
|
| 3055 |
+
"learning_rate": 4.919896406625814e-05,
|
| 3056 |
+
"loss": 1.505,
|
| 3057 |
+
"step": 9550
|
| 3058 |
+
},
|
| 3059 |
+
{
|
| 3060 |
+
"epoch": 0.016029110879486264,
|
| 3061 |
+
"eval_loss": 1.7403160333633423,
|
| 3062 |
+
"eval_masked_accuracy": 0.7048457860946655,
|
| 3063 |
+
"eval_runtime": 1.7491,
|
| 3064 |
+
"eval_samples_per_second": 5.717,
|
| 3065 |
+
"eval_steps_per_second": 2.287,
|
| 3066 |
+
"step": 9550
|
| 3067 |
+
},
|
| 3068 |
+
{
|
| 3069 |
+
"epoch": 0.01611303292597572,
|
| 3070 |
+
"grad_norm": 6.358312129974365,
|
| 3071 |
+
"learning_rate": 4.919476796393366e-05,
|
| 3072 |
+
"loss": 1.655,
|
| 3073 |
+
"step": 9600
|
| 3074 |
+
},
|
| 3075 |
+
{
|
| 3076 |
+
"epoch": 0.01611303292597572,
|
| 3077 |
+
"eval_loss": 1.8444688320159912,
|
| 3078 |
+
"eval_masked_accuracy": 0.6808510422706604,
|
| 3079 |
+
"eval_runtime": 1.7573,
|
| 3080 |
+
"eval_samples_per_second": 5.691,
|
| 3081 |
+
"eval_steps_per_second": 2.276,
|
| 3082 |
+
"step": 9600
|
| 3083 |
+
},
|
| 3084 |
+
{
|
| 3085 |
+
"epoch": 0.016196954972465177,
|
| 3086 |
+
"grad_norm": 6.645698547363281,
|
| 3087 |
+
"learning_rate": 4.919057186160919e-05,
|
| 3088 |
+
"loss": 1.5926,
|
| 3089 |
+
"step": 9650
|
| 3090 |
+
},
|
| 3091 |
+
{
|
| 3092 |
+
"epoch": 0.016196954972465177,
|
| 3093 |
+
"eval_loss": 1.6228317022323608,
|
| 3094 |
+
"eval_masked_accuracy": 0.65625,
|
| 3095 |
+
"eval_runtime": 1.8422,
|
| 3096 |
+
"eval_samples_per_second": 5.428,
|
| 3097 |
+
"eval_steps_per_second": 2.171,
|
| 3098 |
+
"step": 9650
|
| 3099 |
+
},
|
| 3100 |
+
{
|
| 3101 |
+
"epoch": 0.016280877018954632,
|
| 3102 |
+
"grad_norm": 5.672697067260742,
|
| 3103 |
+
"learning_rate": 4.918637575928472e-05,
|
| 3104 |
+
"loss": 1.4762,
|
| 3105 |
+
"step": 9700
|
| 3106 |
+
},
|
| 3107 |
+
{
|
| 3108 |
+
"epoch": 0.016280877018954632,
|
| 3109 |
+
"eval_loss": 1.5051512718200684,
|
| 3110 |
+
"eval_masked_accuracy": 0.6943231225013733,
|
| 3111 |
+
"eval_runtime": 1.7515,
|
| 3112 |
+
"eval_samples_per_second": 5.709,
|
| 3113 |
+
"eval_steps_per_second": 2.284,
|
| 3114 |
+
"step": 9700
|
| 3115 |
+
},
|
| 3116 |
+
{
|
| 3117 |
+
"epoch": 0.01636479906544409,
|
| 3118 |
+
"grad_norm": 5.369190216064453,
|
| 3119 |
+
"learning_rate": 4.9182179656960245e-05,
|
| 3120 |
+
"loss": 1.5021,
|
| 3121 |
+
"step": 9750
|
| 3122 |
+
},
|
| 3123 |
+
{
|
| 3124 |
+
"epoch": 0.01636479906544409,
|
| 3125 |
+
"eval_loss": 1.7301708459854126,
|
| 3126 |
+
"eval_masked_accuracy": 0.6593886613845825,
|
| 3127 |
+
"eval_runtime": 1.7374,
|
| 3128 |
+
"eval_samples_per_second": 5.756,
|
| 3129 |
+
"eval_steps_per_second": 2.302,
|
| 3130 |
+
"step": 9750
|
| 3131 |
+
},
|
| 3132 |
+
{
|
| 3133 |
+
"epoch": 0.01644872111193355,
|
| 3134 |
+
"grad_norm": 4.986740589141846,
|
| 3135 |
+
"learning_rate": 4.917798355463577e-05,
|
| 3136 |
+
"loss": 1.5618,
|
| 3137 |
+
"step": 9800
|
| 3138 |
+
},
|
| 3139 |
+
{
|
| 3140 |
+
"epoch": 0.01644872111193355,
|
| 3141 |
+
"eval_loss": 1.3315510749816895,
|
| 3142 |
+
"eval_masked_accuracy": 0.700421929359436,
|
| 3143 |
+
"eval_runtime": 1.7373,
|
| 3144 |
+
"eval_samples_per_second": 5.756,
|
| 3145 |
+
"eval_steps_per_second": 2.302,
|
| 3146 |
+
"step": 9800
|
| 3147 |
+
},
|
| 3148 |
+
{
|
| 3149 |
+
"epoch": 0.016532643158423004,
|
| 3150 |
+
"grad_norm": 7.441061973571777,
|
| 3151 |
+
"learning_rate": 4.9173787452311295e-05,
|
| 3152 |
+
"loss": 1.5428,
|
| 3153 |
+
"step": 9850
|
| 3154 |
+
},
|
| 3155 |
+
{
|
| 3156 |
+
"epoch": 0.016532643158423004,
|
| 3157 |
+
"eval_loss": 1.6381117105484009,
|
| 3158 |
+
"eval_masked_accuracy": 0.6695652008056641,
|
| 3159 |
+
"eval_runtime": 1.7386,
|
| 3160 |
+
"eval_samples_per_second": 5.752,
|
| 3161 |
+
"eval_steps_per_second": 2.301,
|
| 3162 |
+
"step": 9850
|
| 3163 |
+
},
|
| 3164 |
+
{
|
| 3165 |
+
"epoch": 0.016616565204912462,
|
| 3166 |
+
"grad_norm": 6.459640979766846,
|
| 3167 |
+
"learning_rate": 4.9169591349986824e-05,
|
| 3168 |
+
"loss": 1.4702,
|
| 3169 |
+
"step": 9900
|
| 3170 |
+
},
|
| 3171 |
+
{
|
| 3172 |
+
"epoch": 0.016616565204912462,
|
| 3173 |
+
"eval_loss": 1.537841796875,
|
| 3174 |
+
"eval_masked_accuracy": 0.6741573214530945,
|
| 3175 |
+
"eval_runtime": 1.7482,
|
| 3176 |
+
"eval_samples_per_second": 5.72,
|
| 3177 |
+
"eval_steps_per_second": 2.288,
|
| 3178 |
+
"step": 9900
|
| 3179 |
+
},
|
| 3180 |
+
{
|
| 3181 |
+
"epoch": 0.016700487251401917,
|
| 3182 |
+
"grad_norm": 6.058482646942139,
|
| 3183 |
+
"learning_rate": 4.916539524766235e-05,
|
| 3184 |
+
"loss": 1.5765,
|
| 3185 |
+
"step": 9950
|
| 3186 |
+
},
|
| 3187 |
+
{
|
| 3188 |
+
"epoch": 0.016700487251401917,
|
| 3189 |
+
"eval_loss": 1.688913345336914,
|
| 3190 |
+
"eval_masked_accuracy": 0.692307710647583,
|
| 3191 |
+
"eval_runtime": 1.7482,
|
| 3192 |
+
"eval_samples_per_second": 5.72,
|
| 3193 |
+
"eval_steps_per_second": 2.288,
|
| 3194 |
+
"step": 9950
|
| 3195 |
+
},
|
| 3196 |
+
{
|
| 3197 |
+
"epoch": 0.016784409297891375,
|
| 3198 |
+
"grad_norm": 4.960835933685303,
|
| 3199 |
+
"learning_rate": 4.916119914533788e-05,
|
| 3200 |
+
"loss": 1.544,
|
| 3201 |
+
"step": 10000
|
| 3202 |
+
},
|
| 3203 |
+
{
|
| 3204 |
+
"epoch": 0.016784409297891375,
|
| 3205 |
+
"eval_loss": 1.7901655435562134,
|
| 3206 |
+
"eval_masked_accuracy": 0.6443514823913574,
|
| 3207 |
+
"eval_runtime": 1.7882,
|
| 3208 |
+
"eval_samples_per_second": 5.592,
|
| 3209 |
+
"eval_steps_per_second": 2.237,
|
| 3210 |
+
"step": 10000
|
| 3211 |
}
|
| 3212 |
],
|
| 3213 |
"logging_steps": 50,
|
|
|
|
| 3227 |
"attributes": {}
|
| 3228 |
}
|
| 3229 |
},
|
| 3230 |
+
"total_flos": 1.617791736784392e+16,
|
| 3231 |
"train_batch_size": 2,
|
| 3232 |
"trial_name": null,
|
| 3233 |
"trial_params": null
|