Training in progress, step 978, 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 5752
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e8ea9b5902789889db7ce7eb9bf6b5fb17fc70fbab5ab040891645a983cec7d7
|
3 |
size 5752
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 15814
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9c5d5209b356cdbecaaa88e0c3c7c4e1cd85d90fe9a1673a6ff88a9c283fe663
|
3 |
size 15814
|
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 sha256:5be5bb0fa11a21ec6104b5340813d5d711ba7d8332dcb1a72ea37f04651fdf53
|
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:fe6ddabd32cdc41f4c81c1d53be4d94bd11320d8500ca8b0b69995845949f304
|
3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch": 0.
|
5 |
"eval_steps": 326,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -4595,6 +4595,2296 @@
|
|
4595 |
"eval_samples_per_second": 165.434,
|
4596 |
"eval_steps_per_second": 82.717,
|
4597 |
"step": 652
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
4598 |
}
|
4599 |
],
|
4600 |
"logging_steps": 1,
|
@@ -4614,7 +6904,7 @@
|
|
4614 |
"attributes": {}
|
4615 |
}
|
4616 |
},
|
4617 |
-
"total_flos":
|
4618 |
"train_batch_size": 2,
|
4619 |
"trial_name": null,
|
4620 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.019686584739877413,
|
5 |
"eval_steps": 326,
|
6 |
+
"global_step": 978,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
4595 |
"eval_samples_per_second": 165.434,
|
4596 |
"eval_steps_per_second": 82.717,
|
4597 |
"step": 652
|
4598 |
+
},
|
4599 |
+
{
|
4600 |
+
"epoch": 0.013144519258834305,
|
4601 |
+
"grad_norm": 0.00034002333995886147,
|
4602 |
+
"learning_rate": 0.00010085038219340393,
|
4603 |
+
"loss": 46.0,
|
4604 |
+
"step": 653
|
4605 |
+
},
|
4606 |
+
{
|
4607 |
+
"epoch": 0.013164648691083669,
|
4608 |
+
"grad_norm": 0.00015378076932393014,
|
4609 |
+
"learning_rate": 0.00010060741943827776,
|
4610 |
+
"loss": 46.0,
|
4611 |
+
"step": 654
|
4612 |
+
},
|
4613 |
+
{
|
4614 |
+
"epoch": 0.01318477812333303,
|
4615 |
+
"grad_norm": 0.0002593511308077723,
|
4616 |
+
"learning_rate": 0.00010036445309730944,
|
4617 |
+
"loss": 46.0,
|
4618 |
+
"step": 655
|
4619 |
+
},
|
4620 |
+
{
|
4621 |
+
"epoch": 0.013204907555582394,
|
4622 |
+
"grad_norm": 0.00023333106946665794,
|
4623 |
+
"learning_rate": 0.00010012148460482738,
|
4624 |
+
"loss": 46.0,
|
4625 |
+
"step": 656
|
4626 |
+
},
|
4627 |
+
{
|
4628 |
+
"epoch": 0.013225036987831758,
|
4629 |
+
"grad_norm": 0.00023424337268806994,
|
4630 |
+
"learning_rate": 9.987851539517262e-05,
|
4631 |
+
"loss": 46.0,
|
4632 |
+
"step": 657
|
4633 |
+
},
|
4634 |
+
{
|
4635 |
+
"epoch": 0.013245166420081122,
|
4636 |
+
"grad_norm": 0.00014827096310909837,
|
4637 |
+
"learning_rate": 9.963554690269058e-05,
|
4638 |
+
"loss": 46.0,
|
4639 |
+
"step": 658
|
4640 |
+
},
|
4641 |
+
{
|
4642 |
+
"epoch": 0.013265295852330486,
|
4643 |
+
"grad_norm": 0.00013894452422391623,
|
4644 |
+
"learning_rate": 9.939258056172225e-05,
|
4645 |
+
"loss": 46.0,
|
4646 |
+
"step": 659
|
4647 |
+
},
|
4648 |
+
{
|
4649 |
+
"epoch": 0.01328542528457985,
|
4650 |
+
"grad_norm": 0.00020886877609882504,
|
4651 |
+
"learning_rate": 9.914961780659609e-05,
|
4652 |
+
"loss": 46.0,
|
4653 |
+
"step": 660
|
4654 |
+
},
|
4655 |
+
{
|
4656 |
+
"epoch": 0.013305554716829211,
|
4657 |
+
"grad_norm": 0.00015650583372917026,
|
4658 |
+
"learning_rate": 9.890666007161929e-05,
|
4659 |
+
"loss": 46.0,
|
4660 |
+
"step": 661
|
4661 |
+
},
|
4662 |
+
{
|
4663 |
+
"epoch": 0.013325684149078575,
|
4664 |
+
"grad_norm": 0.00018213962903246284,
|
4665 |
+
"learning_rate": 9.866370879106947e-05,
|
4666 |
+
"loss": 46.0,
|
4667 |
+
"step": 662
|
4668 |
+
},
|
4669 |
+
{
|
4670 |
+
"epoch": 0.013345813581327939,
|
4671 |
+
"grad_norm": 0.0003600477648433298,
|
4672 |
+
"learning_rate": 9.84207653991861e-05,
|
4673 |
+
"loss": 46.0,
|
4674 |
+
"step": 663
|
4675 |
+
},
|
4676 |
+
{
|
4677 |
+
"epoch": 0.013365943013577302,
|
4678 |
+
"grad_norm": 0.0002795457548927516,
|
4679 |
+
"learning_rate": 9.817783133016206e-05,
|
4680 |
+
"loss": 46.0,
|
4681 |
+
"step": 664
|
4682 |
+
},
|
4683 |
+
{
|
4684 |
+
"epoch": 0.013386072445826666,
|
4685 |
+
"grad_norm": 0.0002572405501268804,
|
4686 |
+
"learning_rate": 9.793490801813528e-05,
|
4687 |
+
"loss": 46.0,
|
4688 |
+
"step": 665
|
4689 |
+
},
|
4690 |
+
{
|
4691 |
+
"epoch": 0.013406201878076028,
|
4692 |
+
"grad_norm": 0.0002055472432402894,
|
4693 |
+
"learning_rate": 9.769199689718009e-05,
|
4694 |
+
"loss": 46.0,
|
4695 |
+
"step": 666
|
4696 |
+
},
|
4697 |
+
{
|
4698 |
+
"epoch": 0.013426331310325392,
|
4699 |
+
"grad_norm": 0.0004195273795630783,
|
4700 |
+
"learning_rate": 9.744909940129895e-05,
|
4701 |
+
"loss": 46.0,
|
4702 |
+
"step": 667
|
4703 |
+
},
|
4704 |
+
{
|
4705 |
+
"epoch": 0.013446460742574756,
|
4706 |
+
"grad_norm": 0.0004413472779560834,
|
4707 |
+
"learning_rate": 9.720621696441378e-05,
|
4708 |
+
"loss": 46.0,
|
4709 |
+
"step": 668
|
4710 |
+
},
|
4711 |
+
{
|
4712 |
+
"epoch": 0.01346659017482412,
|
4713 |
+
"grad_norm": 0.00018565787468105555,
|
4714 |
+
"learning_rate": 9.69633510203577e-05,
|
4715 |
+
"loss": 46.0,
|
4716 |
+
"step": 669
|
4717 |
+
},
|
4718 |
+
{
|
4719 |
+
"epoch": 0.013486719607073483,
|
4720 |
+
"grad_norm": 0.00014439223741646856,
|
4721 |
+
"learning_rate": 9.672050300286636e-05,
|
4722 |
+
"loss": 46.0,
|
4723 |
+
"step": 670
|
4724 |
+
},
|
4725 |
+
{
|
4726 |
+
"epoch": 0.013506849039322847,
|
4727 |
+
"grad_norm": 0.0003282301186118275,
|
4728 |
+
"learning_rate": 9.64776743455697e-05,
|
4729 |
+
"loss": 46.0,
|
4730 |
+
"step": 671
|
4731 |
+
},
|
4732 |
+
{
|
4733 |
+
"epoch": 0.013526978471572209,
|
4734 |
+
"grad_norm": 0.00015751754108350724,
|
4735 |
+
"learning_rate": 9.623486648198326e-05,
|
4736 |
+
"loss": 46.0,
|
4737 |
+
"step": 672
|
4738 |
+
},
|
4739 |
+
{
|
4740 |
+
"epoch": 0.013547107903821572,
|
4741 |
+
"grad_norm": 0.00012771219189744443,
|
4742 |
+
"learning_rate": 9.599208084549993e-05,
|
4743 |
+
"loss": 46.0,
|
4744 |
+
"step": 673
|
4745 |
+
},
|
4746 |
+
{
|
4747 |
+
"epoch": 0.013567237336070936,
|
4748 |
+
"grad_norm": 0.00033567333593964577,
|
4749 |
+
"learning_rate": 9.574931886938128e-05,
|
4750 |
+
"loss": 46.0,
|
4751 |
+
"step": 674
|
4752 |
+
},
|
4753 |
+
{
|
4754 |
+
"epoch": 0.0135873667683203,
|
4755 |
+
"grad_norm": 0.00019254954531788826,
|
4756 |
+
"learning_rate": 9.550658198674931e-05,
|
4757 |
+
"loss": 46.0,
|
4758 |
+
"step": 675
|
4759 |
+
},
|
4760 |
+
{
|
4761 |
+
"epoch": 0.013607496200569663,
|
4762 |
+
"grad_norm": 0.0002889351744670421,
|
4763 |
+
"learning_rate": 9.526387163057777e-05,
|
4764 |
+
"loss": 46.0,
|
4765 |
+
"step": 676
|
4766 |
+
},
|
4767 |
+
{
|
4768 |
+
"epoch": 0.013627625632819027,
|
4769 |
+
"grad_norm": 0.00026451353915035725,
|
4770 |
+
"learning_rate": 9.502118923368388e-05,
|
4771 |
+
"loss": 46.0,
|
4772 |
+
"step": 677
|
4773 |
+
},
|
4774 |
+
{
|
4775 |
+
"epoch": 0.01364775506506839,
|
4776 |
+
"grad_norm": 0.00015573820564895868,
|
4777 |
+
"learning_rate": 9.477853622871984e-05,
|
4778 |
+
"loss": 46.0,
|
4779 |
+
"step": 678
|
4780 |
+
},
|
4781 |
+
{
|
4782 |
+
"epoch": 0.013667884497317753,
|
4783 |
+
"grad_norm": 0.00014926907897461206,
|
4784 |
+
"learning_rate": 9.453591404816423e-05,
|
4785 |
+
"loss": 46.0,
|
4786 |
+
"step": 679
|
4787 |
+
},
|
4788 |
+
{
|
4789 |
+
"epoch": 0.013688013929567117,
|
4790 |
+
"grad_norm": 0.0003569158725440502,
|
4791 |
+
"learning_rate": 9.429332412431377e-05,
|
4792 |
+
"loss": 46.0,
|
4793 |
+
"step": 680
|
4794 |
+
},
|
4795 |
+
{
|
4796 |
+
"epoch": 0.01370814336181648,
|
4797 |
+
"grad_norm": 0.00013413053238764405,
|
4798 |
+
"learning_rate": 9.405076788927469e-05,
|
4799 |
+
"loss": 46.0,
|
4800 |
+
"step": 681
|
4801 |
+
},
|
4802 |
+
{
|
4803 |
+
"epoch": 0.013728272794065844,
|
4804 |
+
"grad_norm": 0.000353492476278916,
|
4805 |
+
"learning_rate": 9.380824677495441e-05,
|
4806 |
+
"loss": 46.0,
|
4807 |
+
"step": 682
|
4808 |
+
},
|
4809 |
+
{
|
4810 |
+
"epoch": 0.013748402226315206,
|
4811 |
+
"grad_norm": 0.0003761777188628912,
|
4812 |
+
"learning_rate": 9.356576221305289e-05,
|
4813 |
+
"loss": 46.0,
|
4814 |
+
"step": 683
|
4815 |
+
},
|
4816 |
+
{
|
4817 |
+
"epoch": 0.01376853165856457,
|
4818 |
+
"grad_norm": 0.0002234268467873335,
|
4819 |
+
"learning_rate": 9.332331563505444e-05,
|
4820 |
+
"loss": 46.0,
|
4821 |
+
"step": 684
|
4822 |
+
},
|
4823 |
+
{
|
4824 |
+
"epoch": 0.013788661090813933,
|
4825 |
+
"grad_norm": 0.00026313794660381973,
|
4826 |
+
"learning_rate": 9.308090847221905e-05,
|
4827 |
+
"loss": 46.0,
|
4828 |
+
"step": 685
|
4829 |
+
},
|
4830 |
+
{
|
4831 |
+
"epoch": 0.013808790523063297,
|
4832 |
+
"grad_norm": 0.0005154896061867476,
|
4833 |
+
"learning_rate": 9.283854215557409e-05,
|
4834 |
+
"loss": 46.0,
|
4835 |
+
"step": 686
|
4836 |
+
},
|
4837 |
+
{
|
4838 |
+
"epoch": 0.01382891995531266,
|
4839 |
+
"grad_norm": 0.00024182203924283385,
|
4840 |
+
"learning_rate": 9.259621811590578e-05,
|
4841 |
+
"loss": 46.0,
|
4842 |
+
"step": 687
|
4843 |
+
},
|
4844 |
+
{
|
4845 |
+
"epoch": 0.013849049387562025,
|
4846 |
+
"grad_norm": 0.00014097105304244906,
|
4847 |
+
"learning_rate": 9.235393778375068e-05,
|
4848 |
+
"loss": 46.0,
|
4849 |
+
"step": 688
|
4850 |
+
},
|
4851 |
+
{
|
4852 |
+
"epoch": 0.013869178819811387,
|
4853 |
+
"grad_norm": 0.0003665017429739237,
|
4854 |
+
"learning_rate": 9.211170258938747e-05,
|
4855 |
+
"loss": 46.0,
|
4856 |
+
"step": 689
|
4857 |
+
},
|
4858 |
+
{
|
4859 |
+
"epoch": 0.01388930825206075,
|
4860 |
+
"grad_norm": 0.00021189030667301267,
|
4861 |
+
"learning_rate": 9.18695139628282e-05,
|
4862 |
+
"loss": 46.0,
|
4863 |
+
"step": 690
|
4864 |
+
},
|
4865 |
+
{
|
4866 |
+
"epoch": 0.013909437684310114,
|
4867 |
+
"grad_norm": 0.0005137083935551345,
|
4868 |
+
"learning_rate": 9.162737333381019e-05,
|
4869 |
+
"loss": 46.0,
|
4870 |
+
"step": 691
|
4871 |
+
},
|
4872 |
+
{
|
4873 |
+
"epoch": 0.013929567116559478,
|
4874 |
+
"grad_norm": 0.00035794309224002063,
|
4875 |
+
"learning_rate": 9.138528213178727e-05,
|
4876 |
+
"loss": 46.0,
|
4877 |
+
"step": 692
|
4878 |
+
},
|
4879 |
+
{
|
4880 |
+
"epoch": 0.013949696548808841,
|
4881 |
+
"grad_norm": 0.00044030786375515163,
|
4882 |
+
"learning_rate": 9.11432417859216e-05,
|
4883 |
+
"loss": 46.0,
|
4884 |
+
"step": 693
|
4885 |
+
},
|
4886 |
+
{
|
4887 |
+
"epoch": 0.013969825981058205,
|
4888 |
+
"grad_norm": 0.0002957424148917198,
|
4889 |
+
"learning_rate": 9.090125372507492e-05,
|
4890 |
+
"loss": 46.0,
|
4891 |
+
"step": 694
|
4892 |
+
},
|
4893 |
+
{
|
4894 |
+
"epoch": 0.013989955413307567,
|
4895 |
+
"grad_norm": 0.00022510747658088803,
|
4896 |
+
"learning_rate": 9.065931937780059e-05,
|
4897 |
+
"loss": 46.0,
|
4898 |
+
"step": 695
|
4899 |
+
},
|
4900 |
+
{
|
4901 |
+
"epoch": 0.01401008484555693,
|
4902 |
+
"grad_norm": 0.00029219602583907545,
|
4903 |
+
"learning_rate": 9.041744017233462e-05,
|
4904 |
+
"loss": 46.0,
|
4905 |
+
"step": 696
|
4906 |
+
},
|
4907 |
+
{
|
4908 |
+
"epoch": 0.014030214277806295,
|
4909 |
+
"grad_norm": 0.0005821465747430921,
|
4910 |
+
"learning_rate": 9.017561753658764e-05,
|
4911 |
+
"loss": 46.0,
|
4912 |
+
"step": 697
|
4913 |
+
},
|
4914 |
+
{
|
4915 |
+
"epoch": 0.014050343710055658,
|
4916 |
+
"grad_norm": 0.00017983518773689866,
|
4917 |
+
"learning_rate": 8.993385289813627e-05,
|
4918 |
+
"loss": 46.0,
|
4919 |
+
"step": 698
|
4920 |
+
},
|
4921 |
+
{
|
4922 |
+
"epoch": 0.014070473142305022,
|
4923 |
+
"grad_norm": 0.0003665021213237196,
|
4924 |
+
"learning_rate": 8.969214768421483e-05,
|
4925 |
+
"loss": 46.0,
|
4926 |
+
"step": 699
|
4927 |
+
},
|
4928 |
+
{
|
4929 |
+
"epoch": 0.014090602574554384,
|
4930 |
+
"grad_norm": 0.00022963494120631367,
|
4931 |
+
"learning_rate": 8.945050332170672e-05,
|
4932 |
+
"loss": 46.0,
|
4933 |
+
"step": 700
|
4934 |
+
},
|
4935 |
+
{
|
4936 |
+
"epoch": 0.014110732006803748,
|
4937 |
+
"grad_norm": 0.00016433373093605042,
|
4938 |
+
"learning_rate": 8.920892123713614e-05,
|
4939 |
+
"loss": 46.0,
|
4940 |
+
"step": 701
|
4941 |
+
},
|
4942 |
+
{
|
4943 |
+
"epoch": 0.014130861439053111,
|
4944 |
+
"grad_norm": 0.0002634669654071331,
|
4945 |
+
"learning_rate": 8.89674028566597e-05,
|
4946 |
+
"loss": 46.0,
|
4947 |
+
"step": 702
|
4948 |
+
},
|
4949 |
+
{
|
4950 |
+
"epoch": 0.014150990871302475,
|
4951 |
+
"grad_norm": 0.00027517983107827604,
|
4952 |
+
"learning_rate": 8.872594960605785e-05,
|
4953 |
+
"loss": 46.0,
|
4954 |
+
"step": 703
|
4955 |
+
},
|
4956 |
+
{
|
4957 |
+
"epoch": 0.014171120303551839,
|
4958 |
+
"grad_norm": 0.0004251411010045558,
|
4959 |
+
"learning_rate": 8.848456291072666e-05,
|
4960 |
+
"loss": 46.0,
|
4961 |
+
"step": 704
|
4962 |
+
},
|
4963 |
+
{
|
4964 |
+
"epoch": 0.014191249735801202,
|
4965 |
+
"grad_norm": 0.00023084439453668892,
|
4966 |
+
"learning_rate": 8.82432441956692e-05,
|
4967 |
+
"loss": 46.0,
|
4968 |
+
"step": 705
|
4969 |
+
},
|
4970 |
+
{
|
4971 |
+
"epoch": 0.014211379168050564,
|
4972 |
+
"grad_norm": 0.0002691158442758024,
|
4973 |
+
"learning_rate": 8.80019948854873e-05,
|
4974 |
+
"loss": 46.0,
|
4975 |
+
"step": 706
|
4976 |
+
},
|
4977 |
+
{
|
4978 |
+
"epoch": 0.014231508600299928,
|
4979 |
+
"grad_norm": 0.000261798471910879,
|
4980 |
+
"learning_rate": 8.776081640437294e-05,
|
4981 |
+
"loss": 46.0,
|
4982 |
+
"step": 707
|
4983 |
+
},
|
4984 |
+
{
|
4985 |
+
"epoch": 0.014251638032549292,
|
4986 |
+
"grad_norm": 0.0004933233722113073,
|
4987 |
+
"learning_rate": 8.751971017610012e-05,
|
4988 |
+
"loss": 46.0,
|
4989 |
+
"step": 708
|
4990 |
+
},
|
4991 |
+
{
|
4992 |
+
"epoch": 0.014271767464798656,
|
4993 |
+
"grad_norm": 0.00015099802112672478,
|
4994 |
+
"learning_rate": 8.727867762401623e-05,
|
4995 |
+
"loss": 46.0,
|
4996 |
+
"step": 709
|
4997 |
+
},
|
4998 |
+
{
|
4999 |
+
"epoch": 0.01429189689704802,
|
5000 |
+
"grad_norm": 0.00032548594754189253,
|
5001 |
+
"learning_rate": 8.703772017103372e-05,
|
5002 |
+
"loss": 46.0,
|
5003 |
+
"step": 710
|
5004 |
+
},
|
5005 |
+
{
|
5006 |
+
"epoch": 0.014312026329297383,
|
5007 |
+
"grad_norm": 0.00044292572420090437,
|
5008 |
+
"learning_rate": 8.679683923962174e-05,
|
5009 |
+
"loss": 46.0,
|
5010 |
+
"step": 711
|
5011 |
+
},
|
5012 |
+
{
|
5013 |
+
"epoch": 0.014332155761546745,
|
5014 |
+
"grad_norm": 0.00028910860419273376,
|
5015 |
+
"learning_rate": 8.655603625179759e-05,
|
5016 |
+
"loss": 46.0,
|
5017 |
+
"step": 712
|
5018 |
+
},
|
5019 |
+
{
|
5020 |
+
"epoch": 0.014352285193796109,
|
5021 |
+
"grad_norm": 0.0002870987809728831,
|
5022 |
+
"learning_rate": 8.631531262911857e-05,
|
5023 |
+
"loss": 46.0,
|
5024 |
+
"step": 713
|
5025 |
+
},
|
5026 |
+
{
|
5027 |
+
"epoch": 0.014372414626045472,
|
5028 |
+
"grad_norm": 0.00021710267174057662,
|
5029 |
+
"learning_rate": 8.607466979267338e-05,
|
5030 |
+
"loss": 46.0,
|
5031 |
+
"step": 714
|
5032 |
+
},
|
5033 |
+
{
|
5034 |
+
"epoch": 0.014392544058294836,
|
5035 |
+
"grad_norm": 0.00029545449069701135,
|
5036 |
+
"learning_rate": 8.583410916307386e-05,
|
5037 |
+
"loss": 46.0,
|
5038 |
+
"step": 715
|
5039 |
+
},
|
5040 |
+
{
|
5041 |
+
"epoch": 0.0144126734905442,
|
5042 |
+
"grad_norm": 0.0002639677841216326,
|
5043 |
+
"learning_rate": 8.559363216044647e-05,
|
5044 |
+
"loss": 46.0,
|
5045 |
+
"step": 716
|
5046 |
+
},
|
5047 |
+
{
|
5048 |
+
"epoch": 0.014432802922793562,
|
5049 |
+
"grad_norm": 0.00033363461261615157,
|
5050 |
+
"learning_rate": 8.53532402044241e-05,
|
5051 |
+
"loss": 46.0,
|
5052 |
+
"step": 717
|
5053 |
+
},
|
5054 |
+
{
|
5055 |
+
"epoch": 0.014452932355042926,
|
5056 |
+
"grad_norm": 0.00029945329879410565,
|
5057 |
+
"learning_rate": 8.51129347141374e-05,
|
5058 |
+
"loss": 46.0,
|
5059 |
+
"step": 718
|
5060 |
+
},
|
5061 |
+
{
|
5062 |
+
"epoch": 0.01447306178729229,
|
5063 |
+
"grad_norm": 0.0004964692052453756,
|
5064 |
+
"learning_rate": 8.487271710820681e-05,
|
5065 |
+
"loss": 46.0,
|
5066 |
+
"step": 719
|
5067 |
+
},
|
5068 |
+
{
|
5069 |
+
"epoch": 0.014493191219541653,
|
5070 |
+
"grad_norm": 0.00026980109396390617,
|
5071 |
+
"learning_rate": 8.463258880473373e-05,
|
5072 |
+
"loss": 46.0,
|
5073 |
+
"step": 720
|
5074 |
+
},
|
5075 |
+
{
|
5076 |
+
"epoch": 0.014513320651791017,
|
5077 |
+
"grad_norm": 0.0004024615336675197,
|
5078 |
+
"learning_rate": 8.439255122129254e-05,
|
5079 |
+
"loss": 46.0,
|
5080 |
+
"step": 721
|
5081 |
+
},
|
5082 |
+
{
|
5083 |
+
"epoch": 0.01453345008404038,
|
5084 |
+
"grad_norm": 0.0006139858742244542,
|
5085 |
+
"learning_rate": 8.415260577492195e-05,
|
5086 |
+
"loss": 46.0,
|
5087 |
+
"step": 722
|
5088 |
+
},
|
5089 |
+
{
|
5090 |
+
"epoch": 0.014553579516289742,
|
5091 |
+
"grad_norm": 0.000292949698632583,
|
5092 |
+
"learning_rate": 8.391275388211684e-05,
|
5093 |
+
"loss": 46.0,
|
5094 |
+
"step": 723
|
5095 |
+
},
|
5096 |
+
{
|
5097 |
+
"epoch": 0.014573708948539106,
|
5098 |
+
"grad_norm": 0.00030586167122237384,
|
5099 |
+
"learning_rate": 8.367299695881973e-05,
|
5100 |
+
"loss": 46.0,
|
5101 |
+
"step": 724
|
5102 |
+
},
|
5103 |
+
{
|
5104 |
+
"epoch": 0.01459383838078847,
|
5105 |
+
"grad_norm": 0.0002910511684603989,
|
5106 |
+
"learning_rate": 8.34333364204125e-05,
|
5107 |
+
"loss": 46.0,
|
5108 |
+
"step": 725
|
5109 |
+
},
|
5110 |
+
{
|
5111 |
+
"epoch": 0.014613967813037834,
|
5112 |
+
"grad_norm": 0.0002732513239607215,
|
5113 |
+
"learning_rate": 8.319377368170808e-05,
|
5114 |
+
"loss": 46.0,
|
5115 |
+
"step": 726
|
5116 |
+
},
|
5117 |
+
{
|
5118 |
+
"epoch": 0.014634097245287197,
|
5119 |
+
"grad_norm": 0.0002025508729275316,
|
5120 |
+
"learning_rate": 8.295431015694202e-05,
|
5121 |
+
"loss": 46.0,
|
5122 |
+
"step": 727
|
5123 |
+
},
|
5124 |
+
{
|
5125 |
+
"epoch": 0.014654226677536561,
|
5126 |
+
"grad_norm": 0.0005748551338911057,
|
5127 |
+
"learning_rate": 8.271494725976418e-05,
|
5128 |
+
"loss": 46.0,
|
5129 |
+
"step": 728
|
5130 |
+
},
|
5131 |
+
{
|
5132 |
+
"epoch": 0.014674356109785923,
|
5133 |
+
"grad_norm": 0.0005893989582546055,
|
5134 |
+
"learning_rate": 8.247568640323036e-05,
|
5135 |
+
"loss": 46.0,
|
5136 |
+
"step": 729
|
5137 |
+
},
|
5138 |
+
{
|
5139 |
+
"epoch": 0.014694485542035287,
|
5140 |
+
"grad_norm": 0.0007098098867572844,
|
5141 |
+
"learning_rate": 8.223652899979402e-05,
|
5142 |
+
"loss": 46.0,
|
5143 |
+
"step": 730
|
5144 |
+
},
|
5145 |
+
{
|
5146 |
+
"epoch": 0.01471461497428465,
|
5147 |
+
"grad_norm": 0.00044822742347605526,
|
5148 |
+
"learning_rate": 8.199747646129775e-05,
|
5149 |
+
"loss": 46.0,
|
5150 |
+
"step": 731
|
5151 |
+
},
|
5152 |
+
{
|
5153 |
+
"epoch": 0.014734744406534014,
|
5154 |
+
"grad_norm": 0.0006507154321298003,
|
5155 |
+
"learning_rate": 8.175853019896534e-05,
|
5156 |
+
"loss": 46.0,
|
5157 |
+
"step": 732
|
5158 |
+
},
|
5159 |
+
{
|
5160 |
+
"epoch": 0.014754873838783378,
|
5161 |
+
"grad_norm": 0.00015137386799324304,
|
5162 |
+
"learning_rate": 8.15196916233929e-05,
|
5163 |
+
"loss": 46.0,
|
5164 |
+
"step": 733
|
5165 |
+
},
|
5166 |
+
{
|
5167 |
+
"epoch": 0.01477500327103274,
|
5168 |
+
"grad_norm": 0.0002277817839058116,
|
5169 |
+
"learning_rate": 8.128096214454105e-05,
|
5170 |
+
"loss": 46.0,
|
5171 |
+
"step": 734
|
5172 |
+
},
|
5173 |
+
{
|
5174 |
+
"epoch": 0.014795132703282103,
|
5175 |
+
"grad_norm": 0.00034286073059774935,
|
5176 |
+
"learning_rate": 8.104234317172621e-05,
|
5177 |
+
"loss": 46.0,
|
5178 |
+
"step": 735
|
5179 |
+
},
|
5180 |
+
{
|
5181 |
+
"epoch": 0.014815262135531467,
|
5182 |
+
"grad_norm": 0.00032818439649417996,
|
5183 |
+
"learning_rate": 8.080383611361254e-05,
|
5184 |
+
"loss": 46.0,
|
5185 |
+
"step": 736
|
5186 |
+
},
|
5187 |
+
{
|
5188 |
+
"epoch": 0.014835391567780831,
|
5189 |
+
"grad_norm": 0.0006836484535597265,
|
5190 |
+
"learning_rate": 8.056544237820351e-05,
|
5191 |
+
"loss": 46.0,
|
5192 |
+
"step": 737
|
5193 |
+
},
|
5194 |
+
{
|
5195 |
+
"epoch": 0.014855521000030195,
|
5196 |
+
"grad_norm": 0.00038759320159442723,
|
5197 |
+
"learning_rate": 8.03271633728335e-05,
|
5198 |
+
"loss": 46.0,
|
5199 |
+
"step": 738
|
5200 |
+
},
|
5201 |
+
{
|
5202 |
+
"epoch": 0.014875650432279558,
|
5203 |
+
"grad_norm": 0.0003174393787048757,
|
5204 |
+
"learning_rate": 8.008900050415973e-05,
|
5205 |
+
"loss": 46.0,
|
5206 |
+
"step": 739
|
5207 |
+
},
|
5208 |
+
{
|
5209 |
+
"epoch": 0.01489577986452892,
|
5210 |
+
"grad_norm": 0.0004829028621315956,
|
5211 |
+
"learning_rate": 7.985095517815371e-05,
|
5212 |
+
"loss": 46.0,
|
5213 |
+
"step": 740
|
5214 |
+
},
|
5215 |
+
{
|
5216 |
+
"epoch": 0.014915909296778284,
|
5217 |
+
"grad_norm": 0.0003439805586822331,
|
5218 |
+
"learning_rate": 7.961302880009314e-05,
|
5219 |
+
"loss": 46.0,
|
5220 |
+
"step": 741
|
5221 |
+
},
|
5222 |
+
{
|
5223 |
+
"epoch": 0.014936038729027648,
|
5224 |
+
"grad_norm": 0.00036893304786644876,
|
5225 |
+
"learning_rate": 7.937522277455343e-05,
|
5226 |
+
"loss": 46.0,
|
5227 |
+
"step": 742
|
5228 |
+
},
|
5229 |
+
{
|
5230 |
+
"epoch": 0.014956168161277011,
|
5231 |
+
"grad_norm": 0.00026861962396651506,
|
5232 |
+
"learning_rate": 7.913753850539964e-05,
|
5233 |
+
"loss": 46.0,
|
5234 |
+
"step": 743
|
5235 |
+
},
|
5236 |
+
{
|
5237 |
+
"epoch": 0.014976297593526375,
|
5238 |
+
"grad_norm": 0.0002473319473210722,
|
5239 |
+
"learning_rate": 7.889997739577783e-05,
|
5240 |
+
"loss": 46.0,
|
5241 |
+
"step": 744
|
5242 |
+
},
|
5243 |
+
{
|
5244 |
+
"epoch": 0.014996427025775739,
|
5245 |
+
"grad_norm": 0.0005373766180127859,
|
5246 |
+
"learning_rate": 7.866254084810724e-05,
|
5247 |
+
"loss": 46.0,
|
5248 |
+
"step": 745
|
5249 |
+
},
|
5250 |
+
{
|
5251 |
+
"epoch": 0.0150165564580251,
|
5252 |
+
"grad_norm": 0.0004670672060456127,
|
5253 |
+
"learning_rate": 7.842523026407159e-05,
|
5254 |
+
"loss": 46.0,
|
5255 |
+
"step": 746
|
5256 |
+
},
|
5257 |
+
{
|
5258 |
+
"epoch": 0.015036685890274465,
|
5259 |
+
"grad_norm": 0.00029645231552422047,
|
5260 |
+
"learning_rate": 7.818804704461108e-05,
|
5261 |
+
"loss": 46.0,
|
5262 |
+
"step": 747
|
5263 |
+
},
|
5264 |
+
{
|
5265 |
+
"epoch": 0.015056815322523828,
|
5266 |
+
"grad_norm": 0.00033488948247395456,
|
5267 |
+
"learning_rate": 7.795099258991404e-05,
|
5268 |
+
"loss": 46.0,
|
5269 |
+
"step": 748
|
5270 |
+
},
|
5271 |
+
{
|
5272 |
+
"epoch": 0.015076944754773192,
|
5273 |
+
"grad_norm": 0.00025332943187095225,
|
5274 |
+
"learning_rate": 7.771406829940852e-05,
|
5275 |
+
"loss": 46.0,
|
5276 |
+
"step": 749
|
5277 |
+
},
|
5278 |
+
{
|
5279 |
+
"epoch": 0.015097074187022556,
|
5280 |
+
"grad_norm": 0.0003477553545963019,
|
5281 |
+
"learning_rate": 7.747727557175434e-05,
|
5282 |
+
"loss": 46.0,
|
5283 |
+
"step": 750
|
5284 |
+
},
|
5285 |
+
{
|
5286 |
+
"epoch": 0.015117203619271918,
|
5287 |
+
"grad_norm": 0.0003306921571493149,
|
5288 |
+
"learning_rate": 7.724061580483449e-05,
|
5289 |
+
"loss": 46.0,
|
5290 |
+
"step": 751
|
5291 |
+
},
|
5292 |
+
{
|
5293 |
+
"epoch": 0.015137333051521281,
|
5294 |
+
"grad_norm": 0.0003417869738768786,
|
5295 |
+
"learning_rate": 7.700409039574717e-05,
|
5296 |
+
"loss": 46.0,
|
5297 |
+
"step": 752
|
5298 |
+
},
|
5299 |
+
{
|
5300 |
+
"epoch": 0.015157462483770645,
|
5301 |
+
"grad_norm": 0.0004558273358270526,
|
5302 |
+
"learning_rate": 7.676770074079732e-05,
|
5303 |
+
"loss": 46.0,
|
5304 |
+
"step": 753
|
5305 |
+
},
|
5306 |
+
{
|
5307 |
+
"epoch": 0.015177591916020009,
|
5308 |
+
"grad_norm": 0.00022576648916583508,
|
5309 |
+
"learning_rate": 7.653144823548852e-05,
|
5310 |
+
"loss": 46.0,
|
5311 |
+
"step": 754
|
5312 |
+
},
|
5313 |
+
{
|
5314 |
+
"epoch": 0.015197721348269373,
|
5315 |
+
"grad_norm": 0.0003926701901946217,
|
5316 |
+
"learning_rate": 7.62953342745146e-05,
|
5317 |
+
"loss": 46.0,
|
5318 |
+
"step": 755
|
5319 |
+
},
|
5320 |
+
{
|
5321 |
+
"epoch": 0.015217850780518736,
|
5322 |
+
"grad_norm": 0.0004790101374965161,
|
5323 |
+
"learning_rate": 7.605936025175174e-05,
|
5324 |
+
"loss": 46.0,
|
5325 |
+
"step": 756
|
5326 |
+
},
|
5327 |
+
{
|
5328 |
+
"epoch": 0.015237980212768098,
|
5329 |
+
"grad_norm": 0.00037941025220789015,
|
5330 |
+
"learning_rate": 7.582352756024971e-05,
|
5331 |
+
"loss": 46.0,
|
5332 |
+
"step": 757
|
5333 |
+
},
|
5334 |
+
{
|
5335 |
+
"epoch": 0.015258109645017462,
|
5336 |
+
"grad_norm": 0.0003313767083454877,
|
5337 |
+
"learning_rate": 7.558783759222417e-05,
|
5338 |
+
"loss": 46.0,
|
5339 |
+
"step": 758
|
5340 |
+
},
|
5341 |
+
{
|
5342 |
+
"epoch": 0.015278239077266826,
|
5343 |
+
"grad_norm": 0.00023061798128765076,
|
5344 |
+
"learning_rate": 7.535229173904811e-05,
|
5345 |
+
"loss": 46.0,
|
5346 |
+
"step": 759
|
5347 |
+
},
|
5348 |
+
{
|
5349 |
+
"epoch": 0.01529836850951619,
|
5350 |
+
"grad_norm": 0.00033232785062864423,
|
5351 |
+
"learning_rate": 7.511689139124382e-05,
|
5352 |
+
"loss": 46.0,
|
5353 |
+
"step": 760
|
5354 |
+
},
|
5355 |
+
{
|
5356 |
+
"epoch": 0.015318497941765553,
|
5357 |
+
"grad_norm": 0.0007354258559644222,
|
5358 |
+
"learning_rate": 7.488163793847458e-05,
|
5359 |
+
"loss": 46.0,
|
5360 |
+
"step": 761
|
5361 |
+
},
|
5362 |
+
{
|
5363 |
+
"epoch": 0.015338627374014917,
|
5364 |
+
"grad_norm": 0.00019556190818548203,
|
5365 |
+
"learning_rate": 7.464653276953643e-05,
|
5366 |
+
"loss": 46.0,
|
5367 |
+
"step": 762
|
5368 |
+
},
|
5369 |
+
{
|
5370 |
+
"epoch": 0.015358756806264279,
|
5371 |
+
"grad_norm": 0.0005126126925460994,
|
5372 |
+
"learning_rate": 7.441157727235015e-05,
|
5373 |
+
"loss": 46.0,
|
5374 |
+
"step": 763
|
5375 |
+
},
|
5376 |
+
{
|
5377 |
+
"epoch": 0.015378886238513642,
|
5378 |
+
"grad_norm": 0.00027637736639007926,
|
5379 |
+
"learning_rate": 7.417677283395284e-05,
|
5380 |
+
"loss": 46.0,
|
5381 |
+
"step": 764
|
5382 |
+
},
|
5383 |
+
{
|
5384 |
+
"epoch": 0.015399015670763006,
|
5385 |
+
"grad_norm": 0.0008606024202890694,
|
5386 |
+
"learning_rate": 7.394212084048995e-05,
|
5387 |
+
"loss": 46.0,
|
5388 |
+
"step": 765
|
5389 |
+
},
|
5390 |
+
{
|
5391 |
+
"epoch": 0.01541914510301237,
|
5392 |
+
"grad_norm": 0.0005743610672652721,
|
5393 |
+
"learning_rate": 7.370762267720685e-05,
|
5394 |
+
"loss": 46.0,
|
5395 |
+
"step": 766
|
5396 |
+
},
|
5397 |
+
{
|
5398 |
+
"epoch": 0.015439274535261734,
|
5399 |
+
"grad_norm": 0.0006563942297361791,
|
5400 |
+
"learning_rate": 7.347327972844096e-05,
|
5401 |
+
"loss": 46.0,
|
5402 |
+
"step": 767
|
5403 |
+
},
|
5404 |
+
{
|
5405 |
+
"epoch": 0.015459403967511096,
|
5406 |
+
"grad_norm": 0.00022202875697985291,
|
5407 |
+
"learning_rate": 7.323909337761317e-05,
|
5408 |
+
"loss": 46.0,
|
5409 |
+
"step": 768
|
5410 |
+
},
|
5411 |
+
{
|
5412 |
+
"epoch": 0.01547953339976046,
|
5413 |
+
"grad_norm": 0.0002219324087491259,
|
5414 |
+
"learning_rate": 7.30050650072202e-05,
|
5415 |
+
"loss": 46.0,
|
5416 |
+
"step": 769
|
5417 |
+
},
|
5418 |
+
{
|
5419 |
+
"epoch": 0.015499662832009823,
|
5420 |
+
"grad_norm": 0.00032914732582867146,
|
5421 |
+
"learning_rate": 7.277119599882586e-05,
|
5422 |
+
"loss": 46.0,
|
5423 |
+
"step": 770
|
5424 |
+
},
|
5425 |
+
{
|
5426 |
+
"epoch": 0.015519792264259187,
|
5427 |
+
"grad_norm": 0.0005626956117339432,
|
5428 |
+
"learning_rate": 7.25374877330534e-05,
|
5429 |
+
"loss": 46.0,
|
5430 |
+
"step": 771
|
5431 |
+
},
|
5432 |
+
{
|
5433 |
+
"epoch": 0.01553992169650855,
|
5434 |
+
"grad_norm": 0.0002807240525726229,
|
5435 |
+
"learning_rate": 7.230394158957705e-05,
|
5436 |
+
"loss": 46.0,
|
5437 |
+
"step": 772
|
5438 |
+
},
|
5439 |
+
{
|
5440 |
+
"epoch": 0.015560051128757914,
|
5441 |
+
"grad_norm": 0.0005751781282015145,
|
5442 |
+
"learning_rate": 7.20705589471139e-05,
|
5443 |
+
"loss": 46.0,
|
5444 |
+
"step": 773
|
5445 |
+
},
|
5446 |
+
{
|
5447 |
+
"epoch": 0.015580180561007276,
|
5448 |
+
"grad_norm": 0.0006733541958965361,
|
5449 |
+
"learning_rate": 7.1837341183416e-05,
|
5450 |
+
"loss": 46.0,
|
5451 |
+
"step": 774
|
5452 |
+
},
|
5453 |
+
{
|
5454 |
+
"epoch": 0.01560030999325664,
|
5455 |
+
"grad_norm": 0.00039969501085579395,
|
5456 |
+
"learning_rate": 7.160428967526187e-05,
|
5457 |
+
"loss": 46.0,
|
5458 |
+
"step": 775
|
5459 |
+
},
|
5460 |
+
{
|
5461 |
+
"epoch": 0.015620439425506004,
|
5462 |
+
"grad_norm": 0.0005381114315241575,
|
5463 |
+
"learning_rate": 7.137140579844871e-05,
|
5464 |
+
"loss": 46.0,
|
5465 |
+
"step": 776
|
5466 |
+
},
|
5467 |
+
{
|
5468 |
+
"epoch": 0.015640568857755367,
|
5469 |
+
"grad_norm": 0.0006002363516017795,
|
5470 |
+
"learning_rate": 7.1138690927784e-05,
|
5471 |
+
"loss": 46.0,
|
5472 |
+
"step": 777
|
5473 |
+
},
|
5474 |
+
{
|
5475 |
+
"epoch": 0.01566069829000473,
|
5476 |
+
"grad_norm": 0.000290636089630425,
|
5477 |
+
"learning_rate": 7.090614643707762e-05,
|
5478 |
+
"loss": 46.0,
|
5479 |
+
"step": 778
|
5480 |
+
},
|
5481 |
+
{
|
5482 |
+
"epoch": 0.015680827722254095,
|
5483 |
+
"grad_norm": 0.00021310077863745391,
|
5484 |
+
"learning_rate": 7.067377369913352e-05,
|
5485 |
+
"loss": 46.0,
|
5486 |
+
"step": 779
|
5487 |
+
},
|
5488 |
+
{
|
5489 |
+
"epoch": 0.015700957154503457,
|
5490 |
+
"grad_norm": 0.00011205086775589734,
|
5491 |
+
"learning_rate": 7.044157408574185e-05,
|
5492 |
+
"loss": 46.0,
|
5493 |
+
"step": 780
|
5494 |
+
},
|
5495 |
+
{
|
5496 |
+
"epoch": 0.015721086586752822,
|
5497 |
+
"grad_norm": 0.0011209280928596854,
|
5498 |
+
"learning_rate": 7.020954896767058e-05,
|
5499 |
+
"loss": 46.0,
|
5500 |
+
"step": 781
|
5501 |
+
},
|
5502 |
+
{
|
5503 |
+
"epoch": 0.015741216019002184,
|
5504 |
+
"grad_norm": 0.0005297974566929042,
|
5505 |
+
"learning_rate": 6.997769971465769e-05,
|
5506 |
+
"loss": 46.0,
|
5507 |
+
"step": 782
|
5508 |
+
},
|
5509 |
+
{
|
5510 |
+
"epoch": 0.015761345451251546,
|
5511 |
+
"grad_norm": 0.0007235727971419692,
|
5512 |
+
"learning_rate": 6.974602769540289e-05,
|
5513 |
+
"loss": 46.0,
|
5514 |
+
"step": 783
|
5515 |
+
},
|
5516 |
+
{
|
5517 |
+
"epoch": 0.01578147488350091,
|
5518 |
+
"grad_norm": 0.0008870816673152149,
|
5519 |
+
"learning_rate": 6.951453427755968e-05,
|
5520 |
+
"loss": 46.0,
|
5521 |
+
"step": 784
|
5522 |
+
},
|
5523 |
+
{
|
5524 |
+
"epoch": 0.015801604315750273,
|
5525 |
+
"grad_norm": 0.00041311554377898574,
|
5526 |
+
"learning_rate": 6.928322082772712e-05,
|
5527 |
+
"loss": 46.0,
|
5528 |
+
"step": 785
|
5529 |
+
},
|
5530 |
+
{
|
5531 |
+
"epoch": 0.01582173374799964,
|
5532 |
+
"grad_norm": 0.0003360177797731012,
|
5533 |
+
"learning_rate": 6.905208871144187e-05,
|
5534 |
+
"loss": 46.0,
|
5535 |
+
"step": 786
|
5536 |
+
},
|
5537 |
+
{
|
5538 |
+
"epoch": 0.015841863180249,
|
5539 |
+
"grad_norm": 0.00035951961763203144,
|
5540 |
+
"learning_rate": 6.882113929317015e-05,
|
5541 |
+
"loss": 46.0,
|
5542 |
+
"step": 787
|
5543 |
+
},
|
5544 |
+
{
|
5545 |
+
"epoch": 0.015861992612498363,
|
5546 |
+
"grad_norm": 0.0004993542679585516,
|
5547 |
+
"learning_rate": 6.859037393629957e-05,
|
5548 |
+
"loss": 46.0,
|
5549 |
+
"step": 788
|
5550 |
+
},
|
5551 |
+
{
|
5552 |
+
"epoch": 0.01588212204474773,
|
5553 |
+
"grad_norm": 0.0004983010003343225,
|
5554 |
+
"learning_rate": 6.835979400313122e-05,
|
5555 |
+
"loss": 46.0,
|
5556 |
+
"step": 789
|
5557 |
+
},
|
5558 |
+
{
|
5559 |
+
"epoch": 0.01590225147699709,
|
5560 |
+
"grad_norm": 0.00040434906259179115,
|
5561 |
+
"learning_rate": 6.81294008548715e-05,
|
5562 |
+
"loss": 46.0,
|
5563 |
+
"step": 790
|
5564 |
+
},
|
5565 |
+
{
|
5566 |
+
"epoch": 0.015922380909246456,
|
5567 |
+
"grad_norm": 0.0005002523539587855,
|
5568 |
+
"learning_rate": 6.789919585162423e-05,
|
5569 |
+
"loss": 46.0,
|
5570 |
+
"step": 791
|
5571 |
+
},
|
5572 |
+
{
|
5573 |
+
"epoch": 0.015942510341495818,
|
5574 |
+
"grad_norm": 0.0006788838654756546,
|
5575 |
+
"learning_rate": 6.766918035238237e-05,
|
5576 |
+
"loss": 46.0,
|
5577 |
+
"step": 792
|
5578 |
+
},
|
5579 |
+
{
|
5580 |
+
"epoch": 0.015962639773745183,
|
5581 |
+
"grad_norm": 0.0002748892002273351,
|
5582 |
+
"learning_rate": 6.743935571502038e-05,
|
5583 |
+
"loss": 46.0,
|
5584 |
+
"step": 793
|
5585 |
+
},
|
5586 |
+
{
|
5587 |
+
"epoch": 0.015982769205994545,
|
5588 |
+
"grad_norm": 0.00030586449429392815,
|
5589 |
+
"learning_rate": 6.720972329628577e-05,
|
5590 |
+
"loss": 46.0,
|
5591 |
+
"step": 794
|
5592 |
+
},
|
5593 |
+
{
|
5594 |
+
"epoch": 0.016002898638243907,
|
5595 |
+
"grad_norm": 0.0005200284067541361,
|
5596 |
+
"learning_rate": 6.698028445179148e-05,
|
5597 |
+
"loss": 46.0,
|
5598 |
+
"step": 795
|
5599 |
+
},
|
5600 |
+
{
|
5601 |
+
"epoch": 0.016023028070493273,
|
5602 |
+
"grad_norm": 0.00024440689594484866,
|
5603 |
+
"learning_rate": 6.675104053600763e-05,
|
5604 |
+
"loss": 46.0,
|
5605 |
+
"step": 796
|
5606 |
+
},
|
5607 |
+
{
|
5608 |
+
"epoch": 0.016043157502742635,
|
5609 |
+
"grad_norm": 0.0006268357392400503,
|
5610 |
+
"learning_rate": 6.65219929022535e-05,
|
5611 |
+
"loss": 46.0,
|
5612 |
+
"step": 797
|
5613 |
+
},
|
5614 |
+
{
|
5615 |
+
"epoch": 0.016063286934992,
|
5616 |
+
"grad_norm": 0.00042199273593723774,
|
5617 |
+
"learning_rate": 6.629314290268987e-05,
|
5618 |
+
"loss": 46.0,
|
5619 |
+
"step": 798
|
5620 |
+
},
|
5621 |
+
{
|
5622 |
+
"epoch": 0.016083416367241362,
|
5623 |
+
"grad_norm": 0.00035459554055705667,
|
5624 |
+
"learning_rate": 6.606449188831057e-05,
|
5625 |
+
"loss": 46.0,
|
5626 |
+
"step": 799
|
5627 |
+
},
|
5628 |
+
{
|
5629 |
+
"epoch": 0.016103545799490724,
|
5630 |
+
"grad_norm": 0.00029721111059188843,
|
5631 |
+
"learning_rate": 6.583604120893488e-05,
|
5632 |
+
"loss": 46.0,
|
5633 |
+
"step": 800
|
5634 |
+
},
|
5635 |
+
{
|
5636 |
+
"epoch": 0.01612367523174009,
|
5637 |
+
"grad_norm": 0.0003098206070717424,
|
5638 |
+
"learning_rate": 6.560779221319938e-05,
|
5639 |
+
"loss": 46.0,
|
5640 |
+
"step": 801
|
5641 |
+
},
|
5642 |
+
{
|
5643 |
+
"epoch": 0.01614380466398945,
|
5644 |
+
"grad_norm": 0.00051769835408777,
|
5645 |
+
"learning_rate": 6.537974624855003e-05,
|
5646 |
+
"loss": 46.0,
|
5647 |
+
"step": 802
|
5648 |
+
},
|
5649 |
+
{
|
5650 |
+
"epoch": 0.016163934096238817,
|
5651 |
+
"grad_norm": 0.0006196299218572676,
|
5652 |
+
"learning_rate": 6.51519046612342e-05,
|
5653 |
+
"loss": 46.0,
|
5654 |
+
"step": 803
|
5655 |
+
},
|
5656 |
+
{
|
5657 |
+
"epoch": 0.01618406352848818,
|
5658 |
+
"grad_norm": 0.0002488850150257349,
|
5659 |
+
"learning_rate": 6.492426879629282e-05,
|
5660 |
+
"loss": 46.0,
|
5661 |
+
"step": 804
|
5662 |
+
},
|
5663 |
+
{
|
5664 |
+
"epoch": 0.01620419296073754,
|
5665 |
+
"grad_norm": 0.0009111189283430576,
|
5666 |
+
"learning_rate": 6.46968399975522e-05,
|
5667 |
+
"loss": 46.0,
|
5668 |
+
"step": 805
|
5669 |
+
},
|
5670 |
+
{
|
5671 |
+
"epoch": 0.016224322392986906,
|
5672 |
+
"grad_norm": 0.000280485866824165,
|
5673 |
+
"learning_rate": 6.446961960761643e-05,
|
5674 |
+
"loss": 46.0,
|
5675 |
+
"step": 806
|
5676 |
+
},
|
5677 |
+
{
|
5678 |
+
"epoch": 0.016244451825236268,
|
5679 |
+
"grad_norm": 0.0004848411481361836,
|
5680 |
+
"learning_rate": 6.424260896785914e-05,
|
5681 |
+
"loss": 46.0,
|
5682 |
+
"step": 807
|
5683 |
+
},
|
5684 |
+
{
|
5685 |
+
"epoch": 0.016264581257485634,
|
5686 |
+
"grad_norm": 0.0007137816864997149,
|
5687 |
+
"learning_rate": 6.40158094184158e-05,
|
5688 |
+
"loss": 46.0,
|
5689 |
+
"step": 808
|
5690 |
+
},
|
5691 |
+
{
|
5692 |
+
"epoch": 0.016284710689734996,
|
5693 |
+
"grad_norm": 0.0005314049194566905,
|
5694 |
+
"learning_rate": 6.378922229817575e-05,
|
5695 |
+
"loss": 46.0,
|
5696 |
+
"step": 809
|
5697 |
+
},
|
5698 |
+
{
|
5699 |
+
"epoch": 0.01630484012198436,
|
5700 |
+
"grad_norm": 0.0008834300679154694,
|
5701 |
+
"learning_rate": 6.356284894477412e-05,
|
5702 |
+
"loss": 46.0,
|
5703 |
+
"step": 810
|
5704 |
+
},
|
5705 |
+
{
|
5706 |
+
"epoch": 0.016324969554233723,
|
5707 |
+
"grad_norm": 0.0008476293878629804,
|
5708 |
+
"learning_rate": 6.333669069458432e-05,
|
5709 |
+
"loss": 46.0,
|
5710 |
+
"step": 811
|
5711 |
+
},
|
5712 |
+
{
|
5713 |
+
"epoch": 0.016345098986483085,
|
5714 |
+
"grad_norm": 0.0003947268414776772,
|
5715 |
+
"learning_rate": 6.311074888270971e-05,
|
5716 |
+
"loss": 46.0,
|
5717 |
+
"step": 812
|
5718 |
+
},
|
5719 |
+
{
|
5720 |
+
"epoch": 0.01636522841873245,
|
5721 |
+
"grad_norm": 0.0003291845496278256,
|
5722 |
+
"learning_rate": 6.288502484297607e-05,
|
5723 |
+
"loss": 46.0,
|
5724 |
+
"step": 813
|
5725 |
+
},
|
5726 |
+
{
|
5727 |
+
"epoch": 0.016385357850981812,
|
5728 |
+
"grad_norm": 0.0005163813475519419,
|
5729 |
+
"learning_rate": 6.265951990792347e-05,
|
5730 |
+
"loss": 46.0,
|
5731 |
+
"step": 814
|
5732 |
+
},
|
5733 |
+
{
|
5734 |
+
"epoch": 0.016405487283231178,
|
5735 |
+
"grad_norm": 0.00034108126419596374,
|
5736 |
+
"learning_rate": 6.243423540879865e-05,
|
5737 |
+
"loss": 46.0,
|
5738 |
+
"step": 815
|
5739 |
+
},
|
5740 |
+
{
|
5741 |
+
"epoch": 0.01642561671548054,
|
5742 |
+
"grad_norm": 0.0003455891564954072,
|
5743 |
+
"learning_rate": 6.220917267554686e-05,
|
5744 |
+
"loss": 46.0,
|
5745 |
+
"step": 816
|
5746 |
+
},
|
5747 |
+
{
|
5748 |
+
"epoch": 0.016445746147729902,
|
5749 |
+
"grad_norm": 0.001084479154087603,
|
5750 |
+
"learning_rate": 6.198433303680439e-05,
|
5751 |
+
"loss": 46.0,
|
5752 |
+
"step": 817
|
5753 |
+
},
|
5754 |
+
{
|
5755 |
+
"epoch": 0.016465875579979267,
|
5756 |
+
"grad_norm": 0.0003289970045443624,
|
5757 |
+
"learning_rate": 6.175971781989025e-05,
|
5758 |
+
"loss": 46.0,
|
5759 |
+
"step": 818
|
5760 |
+
},
|
5761 |
+
{
|
5762 |
+
"epoch": 0.01648600501222863,
|
5763 |
+
"grad_norm": 0.0003814552037511021,
|
5764 |
+
"learning_rate": 6.153532835079886e-05,
|
5765 |
+
"loss": 46.0,
|
5766 |
+
"step": 819
|
5767 |
+
},
|
5768 |
+
{
|
5769 |
+
"epoch": 0.016506134444477995,
|
5770 |
+
"grad_norm": 0.0013550389558076859,
|
5771 |
+
"learning_rate": 6.131116595419178e-05,
|
5772 |
+
"loss": 46.0,
|
5773 |
+
"step": 820
|
5774 |
+
},
|
5775 |
+
{
|
5776 |
+
"epoch": 0.016526263876727357,
|
5777 |
+
"grad_norm": 0.0005670760874636471,
|
5778 |
+
"learning_rate": 6.108723195339011e-05,
|
5779 |
+
"loss": 46.0,
|
5780 |
+
"step": 821
|
5781 |
+
},
|
5782 |
+
{
|
5783 |
+
"epoch": 0.01654639330897672,
|
5784 |
+
"grad_norm": 0.0008530982304364443,
|
5785 |
+
"learning_rate": 6.086352767036673e-05,
|
5786 |
+
"loss": 46.0,
|
5787 |
+
"step": 822
|
5788 |
+
},
|
5789 |
+
{
|
5790 |
+
"epoch": 0.016566522741226084,
|
5791 |
+
"grad_norm": 0.0002651612740010023,
|
5792 |
+
"learning_rate": 6.064005442573824e-05,
|
5793 |
+
"loss": 46.0,
|
5794 |
+
"step": 823
|
5795 |
+
},
|
5796 |
+
{
|
5797 |
+
"epoch": 0.016586652173475446,
|
5798 |
+
"grad_norm": 0.0009935569250956178,
|
5799 |
+
"learning_rate": 6.041681353875746e-05,
|
5800 |
+
"loss": 46.0,
|
5801 |
+
"step": 824
|
5802 |
+
},
|
5803 |
+
{
|
5804 |
+
"epoch": 0.01660678160572481,
|
5805 |
+
"grad_norm": 0.00029908836586400867,
|
5806 |
+
"learning_rate": 6.019380632730546e-05,
|
5807 |
+
"loss": 46.0,
|
5808 |
+
"step": 825
|
5809 |
+
},
|
5810 |
+
{
|
5811 |
+
"epoch": 0.016626911037974174,
|
5812 |
+
"grad_norm": 0.00025813686079345644,
|
5813 |
+
"learning_rate": 5.997103410788385e-05,
|
5814 |
+
"loss": 46.0,
|
5815 |
+
"step": 826
|
5816 |
+
},
|
5817 |
+
{
|
5818 |
+
"epoch": 0.01664704047022354,
|
5819 |
+
"grad_norm": 0.0006937151192687452,
|
5820 |
+
"learning_rate": 5.9748498195606925e-05,
|
5821 |
+
"loss": 46.0,
|
5822 |
+
"step": 827
|
5823 |
+
},
|
5824 |
+
{
|
5825 |
+
"epoch": 0.0166671699024729,
|
5826 |
+
"grad_norm": 0.0004930637078359723,
|
5827 |
+
"learning_rate": 5.952619990419408e-05,
|
5828 |
+
"loss": 46.0,
|
5829 |
+
"step": 828
|
5830 |
+
},
|
5831 |
+
{
|
5832 |
+
"epoch": 0.016687299334722263,
|
5833 |
+
"grad_norm": 0.0004578740627039224,
|
5834 |
+
"learning_rate": 5.9304140545961784e-05,
|
5835 |
+
"loss": 46.0,
|
5836 |
+
"step": 829
|
5837 |
+
},
|
5838 |
+
{
|
5839 |
+
"epoch": 0.01670742876697163,
|
5840 |
+
"grad_norm": 0.00044765419443137944,
|
5841 |
+
"learning_rate": 5.9082321431816156e-05,
|
5842 |
+
"loss": 46.0,
|
5843 |
+
"step": 830
|
5844 |
+
},
|
5845 |
+
{
|
5846 |
+
"epoch": 0.01672755819922099,
|
5847 |
+
"grad_norm": 0.00078958785161376,
|
5848 |
+
"learning_rate": 5.88607438712449e-05,
|
5849 |
+
"loss": 46.0,
|
5850 |
+
"step": 831
|
5851 |
+
},
|
5852 |
+
{
|
5853 |
+
"epoch": 0.016747687631470356,
|
5854 |
+
"grad_norm": 0.0006307225557975471,
|
5855 |
+
"learning_rate": 5.863940917230986e-05,
|
5856 |
+
"loss": 46.0,
|
5857 |
+
"step": 832
|
5858 |
+
},
|
5859 |
+
{
|
5860 |
+
"epoch": 0.016767817063719718,
|
5861 |
+
"grad_norm": 0.0006907914648763835,
|
5862 |
+
"learning_rate": 5.841831864163909e-05,
|
5863 |
+
"loss": 46.0,
|
5864 |
+
"step": 833
|
5865 |
+
},
|
5866 |
+
{
|
5867 |
+
"epoch": 0.01678794649596908,
|
5868 |
+
"grad_norm": 0.0006139145698398352,
|
5869 |
+
"learning_rate": 5.8197473584419184e-05,
|
5870 |
+
"loss": 46.0,
|
5871 |
+
"step": 834
|
5872 |
+
},
|
5873 |
+
{
|
5874 |
+
"epoch": 0.016808075928218445,
|
5875 |
+
"grad_norm": 0.0009209556155838072,
|
5876 |
+
"learning_rate": 5.7976875304387756e-05,
|
5877 |
+
"loss": 46.0,
|
5878 |
+
"step": 835
|
5879 |
+
},
|
5880 |
+
{
|
5881 |
+
"epoch": 0.016828205360467807,
|
5882 |
+
"grad_norm": 0.0005617816932499409,
|
5883 |
+
"learning_rate": 5.7756525103825474e-05,
|
5884 |
+
"loss": 46.0,
|
5885 |
+
"step": 836
|
5886 |
+
},
|
5887 |
+
{
|
5888 |
+
"epoch": 0.016848334792717173,
|
5889 |
+
"grad_norm": 0.0006842644652351737,
|
5890 |
+
"learning_rate": 5.753642428354852e-05,
|
5891 |
+
"loss": 46.0,
|
5892 |
+
"step": 837
|
5893 |
+
},
|
5894 |
+
{
|
5895 |
+
"epoch": 0.016868464224966535,
|
5896 |
+
"grad_norm": 0.0002727070532273501,
|
5897 |
+
"learning_rate": 5.731657414290085e-05,
|
5898 |
+
"loss": 46.0,
|
5899 |
+
"step": 838
|
5900 |
+
},
|
5901 |
+
{
|
5902 |
+
"epoch": 0.016888593657215897,
|
5903 |
+
"grad_norm": 0.0005894514033570886,
|
5904 |
+
"learning_rate": 5.7096975979746704e-05,
|
5905 |
+
"loss": 46.0,
|
5906 |
+
"step": 839
|
5907 |
+
},
|
5908 |
+
{
|
5909 |
+
"epoch": 0.016908723089465262,
|
5910 |
+
"grad_norm": 0.00021878795814700425,
|
5911 |
+
"learning_rate": 5.687763109046255e-05,
|
5912 |
+
"loss": 46.0,
|
5913 |
+
"step": 840
|
5914 |
+
},
|
5915 |
+
{
|
5916 |
+
"epoch": 0.016928852521714624,
|
5917 |
+
"grad_norm": 0.0005657190340571105,
|
5918 |
+
"learning_rate": 5.665854076992991e-05,
|
5919 |
+
"loss": 46.0,
|
5920 |
+
"step": 841
|
5921 |
+
},
|
5922 |
+
{
|
5923 |
+
"epoch": 0.01694898195396399,
|
5924 |
+
"grad_norm": 0.0005830818554386497,
|
5925 |
+
"learning_rate": 5.643970631152735e-05,
|
5926 |
+
"loss": 46.0,
|
5927 |
+
"step": 842
|
5928 |
+
},
|
5929 |
+
{
|
5930 |
+
"epoch": 0.01696911138621335,
|
5931 |
+
"grad_norm": 0.0007164289709180593,
|
5932 |
+
"learning_rate": 5.622112900712304e-05,
|
5933 |
+
"loss": 46.0,
|
5934 |
+
"step": 843
|
5935 |
+
},
|
5936 |
+
{
|
5937 |
+
"epoch": 0.016989240818462717,
|
5938 |
+
"grad_norm": 0.0004906615940853953,
|
5939 |
+
"learning_rate": 5.600281014706703e-05,
|
5940 |
+
"loss": 46.0,
|
5941 |
+
"step": 844
|
5942 |
+
},
|
5943 |
+
{
|
5944 |
+
"epoch": 0.01700937025071208,
|
5945 |
+
"grad_norm": 0.001047360710799694,
|
5946 |
+
"learning_rate": 5.57847510201837e-05,
|
5947 |
+
"loss": 46.0,
|
5948 |
+
"step": 845
|
5949 |
+
},
|
5950 |
+
{
|
5951 |
+
"epoch": 0.01702949968296144,
|
5952 |
+
"grad_norm": 0.00040790237835608423,
|
5953 |
+
"learning_rate": 5.556695291376406e-05,
|
5954 |
+
"loss": 46.0,
|
5955 |
+
"step": 846
|
5956 |
+
},
|
5957 |
+
{
|
5958 |
+
"epoch": 0.017049629115210806,
|
5959 |
+
"grad_norm": 0.0006064849440008402,
|
5960 |
+
"learning_rate": 5.5349417113558254e-05,
|
5961 |
+
"loss": 46.0,
|
5962 |
+
"step": 847
|
5963 |
+
},
|
5964 |
+
{
|
5965 |
+
"epoch": 0.01706975854746017,
|
5966 |
+
"grad_norm": 0.0001834803551901132,
|
5967 |
+
"learning_rate": 5.5132144903768e-05,
|
5968 |
+
"loss": 46.0,
|
5969 |
+
"step": 848
|
5970 |
+
},
|
5971 |
+
{
|
5972 |
+
"epoch": 0.017089887979709534,
|
5973 |
+
"grad_norm": 0.00027216278249397874,
|
5974 |
+
"learning_rate": 5.491513756703881e-05,
|
5975 |
+
"loss": 46.0,
|
5976 |
+
"step": 849
|
5977 |
+
},
|
5978 |
+
{
|
5979 |
+
"epoch": 0.017110017411958896,
|
5980 |
+
"grad_norm": 0.0005655901040881872,
|
5981 |
+
"learning_rate": 5.46983963844526e-05,
|
5982 |
+
"loss": 46.0,
|
5983 |
+
"step": 850
|
5984 |
+
},
|
5985 |
+
{
|
5986 |
+
"epoch": 0.017130146844208258,
|
5987 |
+
"grad_norm": 0.0013301552971825004,
|
5988 |
+
"learning_rate": 5.448192263552006e-05,
|
5989 |
+
"loss": 46.0,
|
5990 |
+
"step": 851
|
5991 |
+
},
|
5992 |
+
{
|
5993 |
+
"epoch": 0.017150276276457623,
|
5994 |
+
"grad_norm": 0.00028807963826693594,
|
5995 |
+
"learning_rate": 5.426571759817314e-05,
|
5996 |
+
"loss": 46.0,
|
5997 |
+
"step": 852
|
5998 |
+
},
|
5999 |
+
{
|
6000 |
+
"epoch": 0.017170405708706985,
|
6001 |
+
"grad_norm": 0.0008422695682384074,
|
6002 |
+
"learning_rate": 5.4049782548757386e-05,
|
6003 |
+
"loss": 46.0,
|
6004 |
+
"step": 853
|
6005 |
+
},
|
6006 |
+
{
|
6007 |
+
"epoch": 0.01719053514095635,
|
6008 |
+
"grad_norm": 0.0004891850403510034,
|
6009 |
+
"learning_rate": 5.383411876202464e-05,
|
6010 |
+
"loss": 46.0,
|
6011 |
+
"step": 854
|
6012 |
+
},
|
6013 |
+
{
|
6014 |
+
"epoch": 0.017210664573205713,
|
6015 |
+
"grad_norm": 0.00028051427216269076,
|
6016 |
+
"learning_rate": 5.36187275111253e-05,
|
6017 |
+
"loss": 46.0,
|
6018 |
+
"step": 855
|
6019 |
+
},
|
6020 |
+
{
|
6021 |
+
"epoch": 0.017230794005455075,
|
6022 |
+
"grad_norm": 0.0012559060705825686,
|
6023 |
+
"learning_rate": 5.340361006760082e-05,
|
6024 |
+
"loss": 46.0,
|
6025 |
+
"step": 856
|
6026 |
+
},
|
6027 |
+
{
|
6028 |
+
"epoch": 0.01725092343770444,
|
6029 |
+
"grad_norm": 0.0007434745784848928,
|
6030 |
+
"learning_rate": 5.318876770137634e-05,
|
6031 |
+
"loss": 46.0,
|
6032 |
+
"step": 857
|
6033 |
+
},
|
6034 |
+
{
|
6035 |
+
"epoch": 0.017271052869953802,
|
6036 |
+
"grad_norm": 0.0007889253320172429,
|
6037 |
+
"learning_rate": 5.297420168075307e-05,
|
6038 |
+
"loss": 46.0,
|
6039 |
+
"step": 858
|
6040 |
+
},
|
6041 |
+
{
|
6042 |
+
"epoch": 0.017291182302203167,
|
6043 |
+
"grad_norm": 0.0008923442219384015,
|
6044 |
+
"learning_rate": 5.275991327240082e-05,
|
6045 |
+
"loss": 46.0,
|
6046 |
+
"step": 859
|
6047 |
+
},
|
6048 |
+
{
|
6049 |
+
"epoch": 0.01731131173445253,
|
6050 |
+
"grad_norm": 0.0007664742297492921,
|
6051 |
+
"learning_rate": 5.254590374135058e-05,
|
6052 |
+
"loss": 46.0,
|
6053 |
+
"step": 860
|
6054 |
+
},
|
6055 |
+
{
|
6056 |
+
"epoch": 0.017331441166701895,
|
6057 |
+
"grad_norm": 0.0003808206529356539,
|
6058 |
+
"learning_rate": 5.233217435098707e-05,
|
6059 |
+
"loss": 46.0,
|
6060 |
+
"step": 861
|
6061 |
+
},
|
6062 |
+
{
|
6063 |
+
"epoch": 0.017351570598951257,
|
6064 |
+
"grad_norm": 0.0008439875091426075,
|
6065 |
+
"learning_rate": 5.2118726363041036e-05,
|
6066 |
+
"loss": 46.0,
|
6067 |
+
"step": 862
|
6068 |
+
},
|
6069 |
+
{
|
6070 |
+
"epoch": 0.01737170003120062,
|
6071 |
+
"grad_norm": 0.0005331309512257576,
|
6072 |
+
"learning_rate": 5.190556103758223e-05,
|
6073 |
+
"loss": 46.0,
|
6074 |
+
"step": 863
|
6075 |
+
},
|
6076 |
+
{
|
6077 |
+
"epoch": 0.017391829463449984,
|
6078 |
+
"grad_norm": 0.0005547069013118744,
|
6079 |
+
"learning_rate": 5.1692679633011564e-05,
|
6080 |
+
"loss": 46.0,
|
6081 |
+
"step": 864
|
6082 |
+
},
|
6083 |
+
{
|
6084 |
+
"epoch": 0.017411958895699346,
|
6085 |
+
"grad_norm": 0.000526057556271553,
|
6086 |
+
"learning_rate": 5.148008340605392e-05,
|
6087 |
+
"loss": 46.0,
|
6088 |
+
"step": 865
|
6089 |
+
},
|
6090 |
+
{
|
6091 |
+
"epoch": 0.01743208832794871,
|
6092 |
+
"grad_norm": 0.0012951105600222945,
|
6093 |
+
"learning_rate": 5.1267773611750624e-05,
|
6094 |
+
"loss": 46.0,
|
6095 |
+
"step": 866
|
6096 |
+
},
|
6097 |
+
{
|
6098 |
+
"epoch": 0.017452217760198074,
|
6099 |
+
"grad_norm": 0.0010913871228694916,
|
6100 |
+
"learning_rate": 5.1055751503452144e-05,
|
6101 |
+
"loss": 46.0,
|
6102 |
+
"step": 867
|
6103 |
+
},
|
6104 |
+
{
|
6105 |
+
"epoch": 0.017472347192447436,
|
6106 |
+
"grad_norm": 0.0005590246873907745,
|
6107 |
+
"learning_rate": 5.0844018332810594e-05,
|
6108 |
+
"loss": 46.0,
|
6109 |
+
"step": 868
|
6110 |
+
},
|
6111 |
+
{
|
6112 |
+
"epoch": 0.0174924766246968,
|
6113 |
+
"grad_norm": 0.0006127689266577363,
|
6114 |
+
"learning_rate": 5.0632575349772225e-05,
|
6115 |
+
"loss": 46.0,
|
6116 |
+
"step": 869
|
6117 |
+
},
|
6118 |
+
{
|
6119 |
+
"epoch": 0.017512606056946163,
|
6120 |
+
"grad_norm": 0.0004413559508975595,
|
6121 |
+
"learning_rate": 5.0421423802570454e-05,
|
6122 |
+
"loss": 46.0,
|
6123 |
+
"step": 870
|
6124 |
+
},
|
6125 |
+
{
|
6126 |
+
"epoch": 0.01753273548919553,
|
6127 |
+
"grad_norm": 0.00039344895048998296,
|
6128 |
+
"learning_rate": 5.0210564937718055e-05,
|
6129 |
+
"loss": 46.0,
|
6130 |
+
"step": 871
|
6131 |
+
},
|
6132 |
+
{
|
6133 |
+
"epoch": 0.01755286492144489,
|
6134 |
+
"grad_norm": 0.0009540827013552189,
|
6135 |
+
"learning_rate": 5.000000000000002e-05,
|
6136 |
+
"loss": 46.0,
|
6137 |
+
"step": 872
|
6138 |
+
},
|
6139 |
+
{
|
6140 |
+
"epoch": 0.017572994353694252,
|
6141 |
+
"grad_norm": 0.0008320124470628798,
|
6142 |
+
"learning_rate": 4.978973023246616e-05,
|
6143 |
+
"loss": 46.0,
|
6144 |
+
"step": 873
|
6145 |
+
},
|
6146 |
+
{
|
6147 |
+
"epoch": 0.017593123785943618,
|
6148 |
+
"grad_norm": 0.001188238151371479,
|
6149 |
+
"learning_rate": 4.957975687642389e-05,
|
6150 |
+
"loss": 46.0,
|
6151 |
+
"step": 874
|
6152 |
+
},
|
6153 |
+
{
|
6154 |
+
"epoch": 0.01761325321819298,
|
6155 |
+
"grad_norm": 0.0008758010808378458,
|
6156 |
+
"learning_rate": 4.937008117143055e-05,
|
6157 |
+
"loss": 46.0,
|
6158 |
+
"step": 875
|
6159 |
+
},
|
6160 |
+
{
|
6161 |
+
"epoch": 0.017633382650442345,
|
6162 |
+
"grad_norm": 0.0005536804674193263,
|
6163 |
+
"learning_rate": 4.9160704355286577e-05,
|
6164 |
+
"loss": 46.0,
|
6165 |
+
"step": 876
|
6166 |
+
},
|
6167 |
+
{
|
6168 |
+
"epoch": 0.017653512082691707,
|
6169 |
+
"grad_norm": 0.00037108969991095364,
|
6170 |
+
"learning_rate": 4.895162766402781e-05,
|
6171 |
+
"loss": 46.0,
|
6172 |
+
"step": 877
|
6173 |
+
},
|
6174 |
+
{
|
6175 |
+
"epoch": 0.017673641514941073,
|
6176 |
+
"grad_norm": 0.0011138038244098425,
|
6177 |
+
"learning_rate": 4.8742852331918364e-05,
|
6178 |
+
"loss": 46.0,
|
6179 |
+
"step": 878
|
6180 |
+
},
|
6181 |
+
{
|
6182 |
+
"epoch": 0.017693770947190435,
|
6183 |
+
"grad_norm": 0.0004553370818030089,
|
6184 |
+
"learning_rate": 4.8534379591443246e-05,
|
6185 |
+
"loss": 46.0,
|
6186 |
+
"step": 879
|
6187 |
+
},
|
6188 |
+
{
|
6189 |
+
"epoch": 0.017713900379439797,
|
6190 |
+
"grad_norm": 0.0010616140207275748,
|
6191 |
+
"learning_rate": 4.8326210673301284e-05,
|
6192 |
+
"loss": 46.0,
|
6193 |
+
"step": 880
|
6194 |
+
},
|
6195 |
+
{
|
6196 |
+
"epoch": 0.017734029811689162,
|
6197 |
+
"grad_norm": 0.00036784596159122884,
|
6198 |
+
"learning_rate": 4.811834680639765e-05,
|
6199 |
+
"loss": 46.0,
|
6200 |
+
"step": 881
|
6201 |
+
},
|
6202 |
+
{
|
6203 |
+
"epoch": 0.017754159243938524,
|
6204 |
+
"grad_norm": 0.00037242978578433394,
|
6205 |
+
"learning_rate": 4.791078921783653e-05,
|
6206 |
+
"loss": 46.0,
|
6207 |
+
"step": 882
|
6208 |
+
},
|
6209 |
+
{
|
6210 |
+
"epoch": 0.01777428867618789,
|
6211 |
+
"grad_norm": 0.0014223081525415182,
|
6212 |
+
"learning_rate": 4.770353913291428e-05,
|
6213 |
+
"loss": 46.0,
|
6214 |
+
"step": 883
|
6215 |
+
},
|
6216 |
+
{
|
6217 |
+
"epoch": 0.01779441810843725,
|
6218 |
+
"grad_norm": 0.0004683698061853647,
|
6219 |
+
"learning_rate": 4.749659777511177e-05,
|
6220 |
+
"loss": 46.0,
|
6221 |
+
"step": 884
|
6222 |
+
},
|
6223 |
+
{
|
6224 |
+
"epoch": 0.017814547540686614,
|
6225 |
+
"grad_norm": 0.0004508081474341452,
|
6226 |
+
"learning_rate": 4.728996636608738e-05,
|
6227 |
+
"loss": 46.0,
|
6228 |
+
"step": 885
|
6229 |
+
},
|
6230 |
+
{
|
6231 |
+
"epoch": 0.01783467697293598,
|
6232 |
+
"grad_norm": 0.000599596940446645,
|
6233 |
+
"learning_rate": 4.708364612566969e-05,
|
6234 |
+
"loss": 46.0,
|
6235 |
+
"step": 886
|
6236 |
+
},
|
6237 |
+
{
|
6238 |
+
"epoch": 0.01785480640518534,
|
6239 |
+
"grad_norm": 0.0004682582803070545,
|
6240 |
+
"learning_rate": 4.6877638271850485e-05,
|
6241 |
+
"loss": 46.0,
|
6242 |
+
"step": 887
|
6243 |
+
},
|
6244 |
+
{
|
6245 |
+
"epoch": 0.017874935837434706,
|
6246 |
+
"grad_norm": 0.00035204822779633105,
|
6247 |
+
"learning_rate": 4.667194402077714e-05,
|
6248 |
+
"loss": 46.0,
|
6249 |
+
"step": 888
|
6250 |
+
},
|
6251 |
+
{
|
6252 |
+
"epoch": 0.01789506526968407,
|
6253 |
+
"grad_norm": 0.0009401136194355786,
|
6254 |
+
"learning_rate": 4.646656458674595e-05,
|
6255 |
+
"loss": 46.0,
|
6256 |
+
"step": 889
|
6257 |
+
},
|
6258 |
+
{
|
6259 |
+
"epoch": 0.01791519470193343,
|
6260 |
+
"grad_norm": 0.0006439790595322847,
|
6261 |
+
"learning_rate": 4.62615011821946e-05,
|
6262 |
+
"loss": 46.0,
|
6263 |
+
"step": 890
|
6264 |
+
},
|
6265 |
+
{
|
6266 |
+
"epoch": 0.017935324134182796,
|
6267 |
+
"grad_norm": 0.0006773846689611673,
|
6268 |
+
"learning_rate": 4.6056755017695155e-05,
|
6269 |
+
"loss": 46.0,
|
6270 |
+
"step": 891
|
6271 |
+
},
|
6272 |
+
{
|
6273 |
+
"epoch": 0.017955453566432158,
|
6274 |
+
"grad_norm": 0.0009548621601425111,
|
6275 |
+
"learning_rate": 4.585232730194682e-05,
|
6276 |
+
"loss": 46.0,
|
6277 |
+
"step": 892
|
6278 |
+
},
|
6279 |
+
{
|
6280 |
+
"epoch": 0.017975582998681523,
|
6281 |
+
"grad_norm": 0.0008568911580368876,
|
6282 |
+
"learning_rate": 4.5648219241769054e-05,
|
6283 |
+
"loss": 46.0,
|
6284 |
+
"step": 893
|
6285 |
+
},
|
6286 |
+
{
|
6287 |
+
"epoch": 0.017995712430930885,
|
6288 |
+
"grad_norm": 0.00021023969748057425,
|
6289 |
+
"learning_rate": 4.5444432042093996e-05,
|
6290 |
+
"loss": 46.0,
|
6291 |
+
"step": 894
|
6292 |
+
},
|
6293 |
+
{
|
6294 |
+
"epoch": 0.01801584186318025,
|
6295 |
+
"grad_norm": 0.0003755395009648055,
|
6296 |
+
"learning_rate": 4.524096690595978e-05,
|
6297 |
+
"loss": 46.0,
|
6298 |
+
"step": 895
|
6299 |
+
},
|
6300 |
+
{
|
6301 |
+
"epoch": 0.018035971295429613,
|
6302 |
+
"grad_norm": 0.0010003233328461647,
|
6303 |
+
"learning_rate": 4.5037825034503304e-05,
|
6304 |
+
"loss": 46.0,
|
6305 |
+
"step": 896
|
6306 |
+
},
|
6307 |
+
{
|
6308 |
+
"epoch": 0.018056100727678975,
|
6309 |
+
"grad_norm": 0.0007916453178040683,
|
6310 |
+
"learning_rate": 4.4835007626953e-05,
|
6311 |
+
"loss": 46.0,
|
6312 |
+
"step": 897
|
6313 |
+
},
|
6314 |
+
{
|
6315 |
+
"epoch": 0.01807623015992834,
|
6316 |
+
"grad_norm": 0.0002515468222554773,
|
6317 |
+
"learning_rate": 4.4632515880621894e-05,
|
6318 |
+
"loss": 46.0,
|
6319 |
+
"step": 898
|
6320 |
+
},
|
6321 |
+
{
|
6322 |
+
"epoch": 0.018096359592177702,
|
6323 |
+
"grad_norm": 0.0010216659866273403,
|
6324 |
+
"learning_rate": 4.443035099090048e-05,
|
6325 |
+
"loss": 46.0,
|
6326 |
+
"step": 899
|
6327 |
+
},
|
6328 |
+
{
|
6329 |
+
"epoch": 0.018116489024427068,
|
6330 |
+
"grad_norm": 0.0006039583240635693,
|
6331 |
+
"learning_rate": 4.4228514151249825e-05,
|
6332 |
+
"loss": 46.0,
|
6333 |
+
"step": 900
|
6334 |
+
},
|
6335 |
+
{
|
6336 |
+
"epoch": 0.01813661845667643,
|
6337 |
+
"grad_norm": 0.0011847359128296375,
|
6338 |
+
"learning_rate": 4.4027006553194115e-05,
|
6339 |
+
"loss": 46.0,
|
6340 |
+
"step": 901
|
6341 |
+
},
|
6342 |
+
{
|
6343 |
+
"epoch": 0.01815674788892579,
|
6344 |
+
"grad_norm": 0.0013753952225670218,
|
6345 |
+
"learning_rate": 4.3825829386314166e-05,
|
6346 |
+
"loss": 46.0,
|
6347 |
+
"step": 902
|
6348 |
+
},
|
6349 |
+
{
|
6350 |
+
"epoch": 0.018176877321175157,
|
6351 |
+
"grad_norm": 0.0006966555956751108,
|
6352 |
+
"learning_rate": 4.362498383823996e-05,
|
6353 |
+
"loss": 46.0,
|
6354 |
+
"step": 903
|
6355 |
+
},
|
6356 |
+
{
|
6357 |
+
"epoch": 0.01819700675342452,
|
6358 |
+
"grad_norm": 0.000611517985817045,
|
6359 |
+
"learning_rate": 4.342447109464385e-05,
|
6360 |
+
"loss": 46.0,
|
6361 |
+
"step": 904
|
6362 |
+
},
|
6363 |
+
{
|
6364 |
+
"epoch": 0.018217136185673884,
|
6365 |
+
"grad_norm": 0.001369222765788436,
|
6366 |
+
"learning_rate": 4.322429233923351e-05,
|
6367 |
+
"loss": 46.0,
|
6368 |
+
"step": 905
|
6369 |
+
},
|
6370 |
+
{
|
6371 |
+
"epoch": 0.018237265617923246,
|
6372 |
+
"grad_norm": 0.00040184592944569886,
|
6373 |
+
"learning_rate": 4.3024448753744925e-05,
|
6374 |
+
"loss": 46.0,
|
6375 |
+
"step": 906
|
6376 |
+
},
|
6377 |
+
{
|
6378 |
+
"epoch": 0.01825739505017261,
|
6379 |
+
"grad_norm": 0.0007477894541807473,
|
6380 |
+
"learning_rate": 4.282494151793548e-05,
|
6381 |
+
"loss": 46.0,
|
6382 |
+
"step": 907
|
6383 |
+
},
|
6384 |
+
{
|
6385 |
+
"epoch": 0.018277524482421974,
|
6386 |
+
"grad_norm": 0.0008548839250579476,
|
6387 |
+
"learning_rate": 4.2625771809576874e-05,
|
6388 |
+
"loss": 46.0,
|
6389 |
+
"step": 908
|
6390 |
+
},
|
6391 |
+
{
|
6392 |
+
"epoch": 0.018297653914671336,
|
6393 |
+
"grad_norm": 0.0004542908282019198,
|
6394 |
+
"learning_rate": 4.242694080444837e-05,
|
6395 |
+
"loss": 46.0,
|
6396 |
+
"step": 909
|
6397 |
+
},
|
6398 |
+
{
|
6399 |
+
"epoch": 0.0183177833469207,
|
6400 |
+
"grad_norm": 0.0004831959377042949,
|
6401 |
+
"learning_rate": 4.2228449676329616e-05,
|
6402 |
+
"loss": 46.0,
|
6403 |
+
"step": 910
|
6404 |
+
},
|
6405 |
+
{
|
6406 |
+
"epoch": 0.018337912779170063,
|
6407 |
+
"grad_norm": 0.0005519616534002125,
|
6408 |
+
"learning_rate": 4.2030299596993883e-05,
|
6409 |
+
"loss": 46.0,
|
6410 |
+
"step": 911
|
6411 |
+
},
|
6412 |
+
{
|
6413 |
+
"epoch": 0.01835804221141943,
|
6414 |
+
"grad_norm": 0.0007433863938786089,
|
6415 |
+
"learning_rate": 4.1832491736201077e-05,
|
6416 |
+
"loss": 46.0,
|
6417 |
+
"step": 912
|
6418 |
+
},
|
6419 |
+
{
|
6420 |
+
"epoch": 0.01837817164366879,
|
6421 |
+
"grad_norm": 0.0009301244281232357,
|
6422 |
+
"learning_rate": 4.1635027261690827e-05,
|
6423 |
+
"loss": 46.0,
|
6424 |
+
"step": 913
|
6425 |
+
},
|
6426 |
+
{
|
6427 |
+
"epoch": 0.018398301075918153,
|
6428 |
+
"grad_norm": 0.0016711597563698888,
|
6429 |
+
"learning_rate": 4.143790733917564e-05,
|
6430 |
+
"loss": 46.0,
|
6431 |
+
"step": 914
|
6432 |
+
},
|
6433 |
+
{
|
6434 |
+
"epoch": 0.018418430508167518,
|
6435 |
+
"grad_norm": 0.0014043203555047512,
|
6436 |
+
"learning_rate": 4.124113313233404e-05,
|
6437 |
+
"loss": 46.0,
|
6438 |
+
"step": 915
|
6439 |
+
},
|
6440 |
+
{
|
6441 |
+
"epoch": 0.01843855994041688,
|
6442 |
+
"grad_norm": 0.001079570734873414,
|
6443 |
+
"learning_rate": 4.1044705802803574e-05,
|
6444 |
+
"loss": 46.0,
|
6445 |
+
"step": 916
|
6446 |
+
},
|
6447 |
+
{
|
6448 |
+
"epoch": 0.018458689372666245,
|
6449 |
+
"grad_norm": 0.0006708208820782602,
|
6450 |
+
"learning_rate": 4.084862651017406e-05,
|
6451 |
+
"loss": 46.0,
|
6452 |
+
"step": 917
|
6453 |
+
},
|
6454 |
+
{
|
6455 |
+
"epoch": 0.018478818804915607,
|
6456 |
+
"grad_norm": 0.00034618499921634793,
|
6457 |
+
"learning_rate": 4.065289641198073e-05,
|
6458 |
+
"loss": 46.0,
|
6459 |
+
"step": 918
|
6460 |
+
},
|
6461 |
+
{
|
6462 |
+
"epoch": 0.01849894823716497,
|
6463 |
+
"grad_norm": 0.0003828117623925209,
|
6464 |
+
"learning_rate": 4.045751666369736e-05,
|
6465 |
+
"loss": 46.0,
|
6466 |
+
"step": 919
|
6467 |
+
},
|
6468 |
+
{
|
6469 |
+
"epoch": 0.018519077669414335,
|
6470 |
+
"grad_norm": 0.00042931470670737326,
|
6471 |
+
"learning_rate": 4.026248841872946e-05,
|
6472 |
+
"loss": 46.0,
|
6473 |
+
"step": 920
|
6474 |
+
},
|
6475 |
+
{
|
6476 |
+
"epoch": 0.018539207101663697,
|
6477 |
+
"grad_norm": 0.0014193730894476175,
|
6478 |
+
"learning_rate": 4.006781282840748e-05,
|
6479 |
+
"loss": 46.0,
|
6480 |
+
"step": 921
|
6481 |
+
},
|
6482 |
+
{
|
6483 |
+
"epoch": 0.018559336533913062,
|
6484 |
+
"grad_norm": 0.0003629166749306023,
|
6485 |
+
"learning_rate": 3.987349104198007e-05,
|
6486 |
+
"loss": 46.0,
|
6487 |
+
"step": 922
|
6488 |
+
},
|
6489 |
+
{
|
6490 |
+
"epoch": 0.018579465966162424,
|
6491 |
+
"grad_norm": 0.0006790644838474691,
|
6492 |
+
"learning_rate": 3.9679524206607156e-05,
|
6493 |
+
"loss": 46.0,
|
6494 |
+
"step": 923
|
6495 |
+
},
|
6496 |
+
{
|
6497 |
+
"epoch": 0.018599595398411786,
|
6498 |
+
"grad_norm": 0.0003729971940629184,
|
6499 |
+
"learning_rate": 3.948591346735325e-05,
|
6500 |
+
"loss": 46.0,
|
6501 |
+
"step": 924
|
6502 |
+
},
|
6503 |
+
{
|
6504 |
+
"epoch": 0.01861972483066115,
|
6505 |
+
"grad_norm": 0.0007552574970759451,
|
6506 |
+
"learning_rate": 3.929265996718072e-05,
|
6507 |
+
"loss": 46.0,
|
6508 |
+
"step": 925
|
6509 |
+
},
|
6510 |
+
{
|
6511 |
+
"epoch": 0.018639854262910514,
|
6512 |
+
"grad_norm": 0.0008475988288410008,
|
6513 |
+
"learning_rate": 3.9099764846943e-05,
|
6514 |
+
"loss": 46.0,
|
6515 |
+
"step": 926
|
6516 |
+
},
|
6517 |
+
{
|
6518 |
+
"epoch": 0.01865998369515988,
|
6519 |
+
"grad_norm": 0.0007365219644270837,
|
6520 |
+
"learning_rate": 3.89072292453778e-05,
|
6521 |
+
"loss": 46.0,
|
6522 |
+
"step": 927
|
6523 |
+
},
|
6524 |
+
{
|
6525 |
+
"epoch": 0.01868011312740924,
|
6526 |
+
"grad_norm": 0.0006838237750343978,
|
6527 |
+
"learning_rate": 3.871505429910057e-05,
|
6528 |
+
"loss": 46.0,
|
6529 |
+
"step": 928
|
6530 |
+
},
|
6531 |
+
{
|
6532 |
+
"epoch": 0.018700242559658607,
|
6533 |
+
"grad_norm": 0.001090203644707799,
|
6534 |
+
"learning_rate": 3.85232411425976e-05,
|
6535 |
+
"loss": 46.0,
|
6536 |
+
"step": 929
|
6537 |
+
},
|
6538 |
+
{
|
6539 |
+
"epoch": 0.01872037199190797,
|
6540 |
+
"grad_norm": 0.0004651647468563169,
|
6541 |
+
"learning_rate": 3.833179090821929e-05,
|
6542 |
+
"loss": 46.0,
|
6543 |
+
"step": 930
|
6544 |
+
},
|
6545 |
+
{
|
6546 |
+
"epoch": 0.01874050142415733,
|
6547 |
+
"grad_norm": 0.0019039801554754376,
|
6548 |
+
"learning_rate": 3.814070472617375e-05,
|
6549 |
+
"loss": 46.0,
|
6550 |
+
"step": 931
|
6551 |
+
},
|
6552 |
+
{
|
6553 |
+
"epoch": 0.018760630856406696,
|
6554 |
+
"grad_norm": 0.0003899121074937284,
|
6555 |
+
"learning_rate": 3.794998372451981e-05,
|
6556 |
+
"loss": 46.0,
|
6557 |
+
"step": 932
|
6558 |
+
},
|
6559 |
+
{
|
6560 |
+
"epoch": 0.018780760288656058,
|
6561 |
+
"grad_norm": 0.0008817263296805322,
|
6562 |
+
"learning_rate": 3.775962902916056e-05,
|
6563 |
+
"loss": 46.0,
|
6564 |
+
"step": 933
|
6565 |
+
},
|
6566 |
+
{
|
6567 |
+
"epoch": 0.018800889720905423,
|
6568 |
+
"grad_norm": 0.00196042750030756,
|
6569 |
+
"learning_rate": 3.756964176383655e-05,
|
6570 |
+
"loss": 46.0,
|
6571 |
+
"step": 934
|
6572 |
+
},
|
6573 |
+
{
|
6574 |
+
"epoch": 0.018821019153154785,
|
6575 |
+
"grad_norm": 0.0003646701225079596,
|
6576 |
+
"learning_rate": 3.7380023050119415e-05,
|
6577 |
+
"loss": 46.0,
|
6578 |
+
"step": 935
|
6579 |
+
},
|
6580 |
+
{
|
6581 |
+
"epoch": 0.018841148585404147,
|
6582 |
+
"grad_norm": 0.0007882235804572701,
|
6583 |
+
"learning_rate": 3.7190774007404835e-05,
|
6584 |
+
"loss": 46.0,
|
6585 |
+
"step": 936
|
6586 |
+
},
|
6587 |
+
{
|
6588 |
+
"epoch": 0.018861278017653513,
|
6589 |
+
"grad_norm": 0.0008490802138112485,
|
6590 |
+
"learning_rate": 3.700189575290641e-05,
|
6591 |
+
"loss": 46.0,
|
6592 |
+
"step": 937
|
6593 |
+
},
|
6594 |
+
{
|
6595 |
+
"epoch": 0.018881407449902875,
|
6596 |
+
"grad_norm": 0.0003721664543263614,
|
6597 |
+
"learning_rate": 3.681338940164868e-05,
|
6598 |
+
"loss": 46.0,
|
6599 |
+
"step": 938
|
6600 |
+
},
|
6601 |
+
{
|
6602 |
+
"epoch": 0.01890153688215224,
|
6603 |
+
"grad_norm": 0.00035278795985504985,
|
6604 |
+
"learning_rate": 3.6625256066460735e-05,
|
6605 |
+
"loss": 46.0,
|
6606 |
+
"step": 939
|
6607 |
+
},
|
6608 |
+
{
|
6609 |
+
"epoch": 0.018921666314401602,
|
6610 |
+
"grad_norm": 0.0005430346354842186,
|
6611 |
+
"learning_rate": 3.6437496857969566e-05,
|
6612 |
+
"loss": 46.0,
|
6613 |
+
"step": 940
|
6614 |
+
},
|
6615 |
+
{
|
6616 |
+
"epoch": 0.018941795746650964,
|
6617 |
+
"grad_norm": 0.0013423273339867592,
|
6618 |
+
"learning_rate": 3.625011288459365e-05,
|
6619 |
+
"loss": 46.0,
|
6620 |
+
"step": 941
|
6621 |
+
},
|
6622 |
+
{
|
6623 |
+
"epoch": 0.01896192517890033,
|
6624 |
+
"grad_norm": 0.0003662093076854944,
|
6625 |
+
"learning_rate": 3.606310525253621e-05,
|
6626 |
+
"loss": 46.0,
|
6627 |
+
"step": 942
|
6628 |
+
},
|
6629 |
+
{
|
6630 |
+
"epoch": 0.01898205461114969,
|
6631 |
+
"grad_norm": 0.0005090291379019618,
|
6632 |
+
"learning_rate": 3.5876475065778715e-05,
|
6633 |
+
"loss": 46.0,
|
6634 |
+
"step": 943
|
6635 |
+
},
|
6636 |
+
{
|
6637 |
+
"epoch": 0.019002184043399057,
|
6638 |
+
"grad_norm": 0.0002727884566411376,
|
6639 |
+
"learning_rate": 3.5690223426074576e-05,
|
6640 |
+
"loss": 46.0,
|
6641 |
+
"step": 944
|
6642 |
+
},
|
6643 |
+
{
|
6644 |
+
"epoch": 0.01902231347564842,
|
6645 |
+
"grad_norm": 0.0005081517156213522,
|
6646 |
+
"learning_rate": 3.550435143294238e-05,
|
6647 |
+
"loss": 46.0,
|
6648 |
+
"step": 945
|
6649 |
+
},
|
6650 |
+
{
|
6651 |
+
"epoch": 0.019042442907897784,
|
6652 |
+
"grad_norm": 0.0008462371188215911,
|
6653 |
+
"learning_rate": 3.531886018365954e-05,
|
6654 |
+
"loss": 46.0,
|
6655 |
+
"step": 946
|
6656 |
+
},
|
6657 |
+
{
|
6658 |
+
"epoch": 0.019062572340147146,
|
6659 |
+
"grad_norm": 0.0013481457717716694,
|
6660 |
+
"learning_rate": 3.513375077325575e-05,
|
6661 |
+
"loss": 46.0,
|
6662 |
+
"step": 947
|
6663 |
+
},
|
6664 |
+
{
|
6665 |
+
"epoch": 0.01908270177239651,
|
6666 |
+
"grad_norm": 0.0006557535380125046,
|
6667 |
+
"learning_rate": 3.4949024294506674e-05,
|
6668 |
+
"loss": 46.0,
|
6669 |
+
"step": 948
|
6670 |
+
},
|
6671 |
+
{
|
6672 |
+
"epoch": 0.019102831204645874,
|
6673 |
+
"grad_norm": 0.0005414308398030698,
|
6674 |
+
"learning_rate": 3.476468183792716e-05,
|
6675 |
+
"loss": 46.0,
|
6676 |
+
"step": 949
|
6677 |
+
},
|
6678 |
+
{
|
6679 |
+
"epoch": 0.019122960636895236,
|
6680 |
+
"grad_norm": 0.0014492279151454568,
|
6681 |
+
"learning_rate": 3.458072449176525e-05,
|
6682 |
+
"loss": 46.0,
|
6683 |
+
"step": 950
|
6684 |
+
},
|
6685 |
+
{
|
6686 |
+
"epoch": 0.0191430900691446,
|
6687 |
+
"grad_norm": 0.0007959827198646963,
|
6688 |
+
"learning_rate": 3.439715334199538e-05,
|
6689 |
+
"loss": 46.0,
|
6690 |
+
"step": 951
|
6691 |
+
},
|
6692 |
+
{
|
6693 |
+
"epoch": 0.019163219501393963,
|
6694 |
+
"grad_norm": 0.0005054565845057368,
|
6695 |
+
"learning_rate": 3.4213969472312154e-05,
|
6696 |
+
"loss": 46.0,
|
6697 |
+
"step": 952
|
6698 |
+
},
|
6699 |
+
{
|
6700 |
+
"epoch": 0.019183348933643325,
|
6701 |
+
"grad_norm": 0.0010196175426244736,
|
6702 |
+
"learning_rate": 3.403117396412391e-05,
|
6703 |
+
"loss": 46.0,
|
6704 |
+
"step": 953
|
6705 |
+
},
|
6706 |
+
{
|
6707 |
+
"epoch": 0.01920347836589269,
|
6708 |
+
"grad_norm": 0.0007254026713781059,
|
6709 |
+
"learning_rate": 3.384876789654631e-05,
|
6710 |
+
"loss": 46.0,
|
6711 |
+
"step": 954
|
6712 |
+
},
|
6713 |
+
{
|
6714 |
+
"epoch": 0.019223607798142053,
|
6715 |
+
"grad_norm": 0.000904410204384476,
|
6716 |
+
"learning_rate": 3.366675234639601e-05,
|
6717 |
+
"loss": 46.0,
|
6718 |
+
"step": 955
|
6719 |
+
},
|
6720 |
+
{
|
6721 |
+
"epoch": 0.019243737230391418,
|
6722 |
+
"grad_norm": 0.0011299613397568464,
|
6723 |
+
"learning_rate": 3.348512838818425e-05,
|
6724 |
+
"loss": 46.0,
|
6725 |
+
"step": 956
|
6726 |
+
},
|
6727 |
+
{
|
6728 |
+
"epoch": 0.01926386666264078,
|
6729 |
+
"grad_norm": 0.0011055005015805364,
|
6730 |
+
"learning_rate": 3.3303897094110636e-05,
|
6731 |
+
"loss": 46.0,
|
6732 |
+
"step": 957
|
6733 |
+
},
|
6734 |
+
{
|
6735 |
+
"epoch": 0.019283996094890142,
|
6736 |
+
"grad_norm": 0.0005328382831066847,
|
6737 |
+
"learning_rate": 3.3123059534056634e-05,
|
6738 |
+
"loss": 46.0,
|
6739 |
+
"step": 958
|
6740 |
+
},
|
6741 |
+
{
|
6742 |
+
"epoch": 0.019304125527139507,
|
6743 |
+
"grad_norm": 0.0006177676841616631,
|
6744 |
+
"learning_rate": 3.294261677557935e-05,
|
6745 |
+
"loss": 46.0,
|
6746 |
+
"step": 959
|
6747 |
+
},
|
6748 |
+
{
|
6749 |
+
"epoch": 0.01932425495938887,
|
6750 |
+
"grad_norm": 0.0007813276024535298,
|
6751 |
+
"learning_rate": 3.2762569883905205e-05,
|
6752 |
+
"loss": 46.0,
|
6753 |
+
"step": 960
|
6754 |
+
},
|
6755 |
+
{
|
6756 |
+
"epoch": 0.019344384391638235,
|
6757 |
+
"grad_norm": 0.000981914228759706,
|
6758 |
+
"learning_rate": 3.258291992192377e-05,
|
6759 |
+
"loss": 46.0,
|
6760 |
+
"step": 961
|
6761 |
+
},
|
6762 |
+
{
|
6763 |
+
"epoch": 0.019364513823887597,
|
6764 |
+
"grad_norm": 0.000979884178377688,
|
6765 |
+
"learning_rate": 3.240366795018117e-05,
|
6766 |
+
"loss": 46.0,
|
6767 |
+
"step": 962
|
6768 |
+
},
|
6769 |
+
{
|
6770 |
+
"epoch": 0.019384643256136962,
|
6771 |
+
"grad_norm": 0.0005548804765567183,
|
6772 |
+
"learning_rate": 3.222481502687425e-05,
|
6773 |
+
"loss": 46.0,
|
6774 |
+
"step": 963
|
6775 |
+
},
|
6776 |
+
{
|
6777 |
+
"epoch": 0.019404772688386324,
|
6778 |
+
"grad_norm": 0.0007180179818533361,
|
6779 |
+
"learning_rate": 3.2046362207844e-05,
|
6780 |
+
"loss": 46.0,
|
6781 |
+
"step": 964
|
6782 |
+
},
|
6783 |
+
{
|
6784 |
+
"epoch": 0.019424902120635686,
|
6785 |
+
"grad_norm": 0.0011889664456248283,
|
6786 |
+
"learning_rate": 3.1868310546569424e-05,
|
6787 |
+
"loss": 46.0,
|
6788 |
+
"step": 965
|
6789 |
+
},
|
6790 |
+
{
|
6791 |
+
"epoch": 0.019445031552885052,
|
6792 |
+
"grad_norm": 0.00041984309791587293,
|
6793 |
+
"learning_rate": 3.1690661094161364e-05,
|
6794 |
+
"loss": 46.0,
|
6795 |
+
"step": 966
|
6796 |
+
},
|
6797 |
+
{
|
6798 |
+
"epoch": 0.019465160985134414,
|
6799 |
+
"grad_norm": 0.0005576315452344716,
|
6800 |
+
"learning_rate": 3.151341489935627e-05,
|
6801 |
+
"loss": 46.0,
|
6802 |
+
"step": 967
|
6803 |
+
},
|
6804 |
+
{
|
6805 |
+
"epoch": 0.01948529041738378,
|
6806 |
+
"grad_norm": 0.0008044608402997255,
|
6807 |
+
"learning_rate": 3.133657300850995e-05,
|
6808 |
+
"loss": 46.0,
|
6809 |
+
"step": 968
|
6810 |
+
},
|
6811 |
+
{
|
6812 |
+
"epoch": 0.01950541984963314,
|
6813 |
+
"grad_norm": 0.0006808895850554109,
|
6814 |
+
"learning_rate": 3.116013646559146e-05,
|
6815 |
+
"loss": 46.0,
|
6816 |
+
"step": 969
|
6817 |
+
},
|
6818 |
+
{
|
6819 |
+
"epoch": 0.019525549281882503,
|
6820 |
+
"grad_norm": 0.000420604192186147,
|
6821 |
+
"learning_rate": 3.0984106312177e-05,
|
6822 |
+
"loss": 46.0,
|
6823 |
+
"step": 970
|
6824 |
+
},
|
6825 |
+
{
|
6826 |
+
"epoch": 0.01954567871413187,
|
6827 |
+
"grad_norm": 0.0008205743506550789,
|
6828 |
+
"learning_rate": 3.0808483587443595e-05,
|
6829 |
+
"loss": 46.0,
|
6830 |
+
"step": 971
|
6831 |
+
},
|
6832 |
+
{
|
6833 |
+
"epoch": 0.01956580814638123,
|
6834 |
+
"grad_norm": 0.0006578009924851358,
|
6835 |
+
"learning_rate": 3.063326932816307e-05,
|
6836 |
+
"loss": 46.0,
|
6837 |
+
"step": 972
|
6838 |
+
},
|
6839 |
+
{
|
6840 |
+
"epoch": 0.019585937578630596,
|
6841 |
+
"grad_norm": 0.0017218554858118296,
|
6842 |
+
"learning_rate": 3.045846456869592e-05,
|
6843 |
+
"loss": 46.0,
|
6844 |
+
"step": 973
|
6845 |
+
},
|
6846 |
+
{
|
6847 |
+
"epoch": 0.019606067010879958,
|
6848 |
+
"grad_norm": 0.0008508884930051863,
|
6849 |
+
"learning_rate": 3.0284070340985295e-05,
|
6850 |
+
"loss": 46.0,
|
6851 |
+
"step": 974
|
6852 |
+
},
|
6853 |
+
{
|
6854 |
+
"epoch": 0.01962619644312932,
|
6855 |
+
"grad_norm": 0.0006469383952207863,
|
6856 |
+
"learning_rate": 3.011008767455059e-05,
|
6857 |
+
"loss": 46.0,
|
6858 |
+
"step": 975
|
6859 |
+
},
|
6860 |
+
{
|
6861 |
+
"epoch": 0.019646325875378685,
|
6862 |
+
"grad_norm": 0.00034800561843439937,
|
6863 |
+
"learning_rate": 2.9936517596481818e-05,
|
6864 |
+
"loss": 46.0,
|
6865 |
+
"step": 976
|
6866 |
+
},
|
6867 |
+
{
|
6868 |
+
"epoch": 0.019666455307628047,
|
6869 |
+
"grad_norm": 0.0005060382536612451,
|
6870 |
+
"learning_rate": 2.9763361131433208e-05,
|
6871 |
+
"loss": 46.0,
|
6872 |
+
"step": 977
|
6873 |
+
},
|
6874 |
+
{
|
6875 |
+
"epoch": 0.019686584739877413,
|
6876 |
+
"grad_norm": 0.0005251033580861986,
|
6877 |
+
"learning_rate": 2.9590619301617183e-05,
|
6878 |
+
"loss": 46.0,
|
6879 |
+
"step": 978
|
6880 |
+
},
|
6881 |
+
{
|
6882 |
+
"epoch": 0.019686584739877413,
|
6883 |
+
"eval_loss": 11.5,
|
6884 |
+
"eval_runtime": 126.6494,
|
6885 |
+
"eval_samples_per_second": 165.165,
|
6886 |
+
"eval_steps_per_second": 82.582,
|
6887 |
+
"step": 978
|
6888 |
}
|
6889 |
],
|
6890 |
"logging_steps": 1,
|
|
|
6904 |
"attributes": {}
|
6905 |
}
|
6906 |
},
|
6907 |
+
"total_flos": 22552698421248.0,
|
6908 |
"train_batch_size": 2,
|
6909 |
"trial_name": null,
|
6910 |
"trial_params": null
|