Commit
•
c2bb426
1
Parent(s):
b1cea36
Training in progress, step 8256, 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 439648328
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8555c6c30b01b7b518e204262ad49bdfc8a647ffac30864d51ee8c8057b5b58b
|
| 3 |
size 439648328
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 879415866
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c65419cfd6a880659ccadbe3db88894c9a0bc93ac16d9c1b7a0c6cf2cbc2b395
|
| 3 |
size 879415866
|
last-checkpoint/rng_state_0.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14512
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5528a8a5438254c67bb6f375f3876eeca26717fef489265e3b041c5387c9fb8f
|
| 3 |
size 14512
|
last-checkpoint/rng_state_1.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14512
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bc1d87868b3d95ab9fb053bc3e7b7216c1360a2d6ef559d5a4f71fdb1eb48e41
|
| 3 |
size 14512
|
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:c699c45754dba9f295f88b976126b3ed2ecc4605b1af134d5e1f2b88049fd75b
|
| 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": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -12649,6 +12649,1812 @@
|
|
| 12649 |
"learning_rate": 1.4979639325189065e-05,
|
| 12650 |
"loss": 5.2557,
|
| 12651 |
"step": 7224
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12652 |
}
|
| 12653 |
],
|
| 12654 |
"logging_steps": 4,
|
|
@@ -12656,7 +14462,7 @@
|
|
| 12656 |
"num_input_tokens_seen": 0,
|
| 12657 |
"num_train_epochs": 1,
|
| 12658 |
"save_steps": 1032,
|
| 12659 |
-
"total_flos": 6.
|
| 12660 |
"train_batch_size": 8,
|
| 12661 |
"trial_name": null,
|
| 12662 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.8004653868528214,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 8256,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 12649 |
"learning_rate": 1.4979639325189065e-05,
|
| 12650 |
"loss": 5.2557,
|
| 12651 |
"step": 7224
|
| 12652 |
+
},
|
| 12653 |
+
{
|
| 12654 |
+
"epoch": 0.7,
|
| 12655 |
+
"grad_norm": 1.0959160327911377,
|
| 12656 |
+
"learning_rate": 1.4960248206321506e-05,
|
| 12657 |
+
"loss": 5.2666,
|
| 12658 |
+
"step": 7228
|
| 12659 |
+
},
|
| 12660 |
+
{
|
| 12661 |
+
"epoch": 0.7,
|
| 12662 |
+
"grad_norm": 1.0157089233398438,
|
| 12663 |
+
"learning_rate": 1.4940857087453946e-05,
|
| 12664 |
+
"loss": 5.3451,
|
| 12665 |
+
"step": 7232
|
| 12666 |
+
},
|
| 12667 |
+
{
|
| 12668 |
+
"epoch": 0.7,
|
| 12669 |
+
"grad_norm": 1.1046866178512573,
|
| 12670 |
+
"learning_rate": 1.4921465968586387e-05,
|
| 12671 |
+
"loss": 5.2349,
|
| 12672 |
+
"step": 7236
|
| 12673 |
+
},
|
| 12674 |
+
{
|
| 12675 |
+
"epoch": 0.7,
|
| 12676 |
+
"grad_norm": 1.0781642198562622,
|
| 12677 |
+
"learning_rate": 1.4902074849718831e-05,
|
| 12678 |
+
"loss": 5.2609,
|
| 12679 |
+
"step": 7240
|
| 12680 |
+
},
|
| 12681 |
+
{
|
| 12682 |
+
"epoch": 0.7,
|
| 12683 |
+
"grad_norm": 1.0019387006759644,
|
| 12684 |
+
"learning_rate": 1.4882683730851271e-05,
|
| 12685 |
+
"loss": 5.3954,
|
| 12686 |
+
"step": 7244
|
| 12687 |
+
},
|
| 12688 |
+
{
|
| 12689 |
+
"epoch": 0.7,
|
| 12690 |
+
"grad_norm": 1.1083266735076904,
|
| 12691 |
+
"learning_rate": 1.4863292611983712e-05,
|
| 12692 |
+
"loss": 5.2636,
|
| 12693 |
+
"step": 7248
|
| 12694 |
+
},
|
| 12695 |
+
{
|
| 12696 |
+
"epoch": 0.7,
|
| 12697 |
+
"grad_norm": 1.2309002876281738,
|
| 12698 |
+
"learning_rate": 1.4843901493116152e-05,
|
| 12699 |
+
"loss": 5.2955,
|
| 12700 |
+
"step": 7252
|
| 12701 |
+
},
|
| 12702 |
+
{
|
| 12703 |
+
"epoch": 0.7,
|
| 12704 |
+
"grad_norm": 1.087774634361267,
|
| 12705 |
+
"learning_rate": 1.4824510374248596e-05,
|
| 12706 |
+
"loss": 5.3,
|
| 12707 |
+
"step": 7256
|
| 12708 |
+
},
|
| 12709 |
+
{
|
| 12710 |
+
"epoch": 0.7,
|
| 12711 |
+
"grad_norm": 1.075287938117981,
|
| 12712 |
+
"learning_rate": 1.4805119255381037e-05,
|
| 12713 |
+
"loss": 5.3727,
|
| 12714 |
+
"step": 7260
|
| 12715 |
+
},
|
| 12716 |
+
{
|
| 12717 |
+
"epoch": 0.7,
|
| 12718 |
+
"grad_norm": 1.0246081352233887,
|
| 12719 |
+
"learning_rate": 1.4785728136513477e-05,
|
| 12720 |
+
"loss": 5.3558,
|
| 12721 |
+
"step": 7264
|
| 12722 |
+
},
|
| 12723 |
+
{
|
| 12724 |
+
"epoch": 0.7,
|
| 12725 |
+
"grad_norm": 1.124543309211731,
|
| 12726 |
+
"learning_rate": 1.4766337017645918e-05,
|
| 12727 |
+
"loss": 5.3379,
|
| 12728 |
+
"step": 7268
|
| 12729 |
+
},
|
| 12730 |
+
{
|
| 12731 |
+
"epoch": 0.71,
|
| 12732 |
+
"grad_norm": 1.047892689704895,
|
| 12733 |
+
"learning_rate": 1.474694589877836e-05,
|
| 12734 |
+
"loss": 5.3975,
|
| 12735 |
+
"step": 7272
|
| 12736 |
+
},
|
| 12737 |
+
{
|
| 12738 |
+
"epoch": 0.71,
|
| 12739 |
+
"grad_norm": 1.0381947755813599,
|
| 12740 |
+
"learning_rate": 1.47275547799108e-05,
|
| 12741 |
+
"loss": 5.3515,
|
| 12742 |
+
"step": 7276
|
| 12743 |
+
},
|
| 12744 |
+
{
|
| 12745 |
+
"epoch": 0.71,
|
| 12746 |
+
"grad_norm": 1.0230307579040527,
|
| 12747 |
+
"learning_rate": 1.4708163661043243e-05,
|
| 12748 |
+
"loss": 5.2925,
|
| 12749 |
+
"step": 7280
|
| 12750 |
+
},
|
| 12751 |
+
{
|
| 12752 |
+
"epoch": 0.71,
|
| 12753 |
+
"grad_norm": 1.0596458911895752,
|
| 12754 |
+
"learning_rate": 1.4688772542175685e-05,
|
| 12755 |
+
"loss": 5.3191,
|
| 12756 |
+
"step": 7284
|
| 12757 |
+
},
|
| 12758 |
+
{
|
| 12759 |
+
"epoch": 0.71,
|
| 12760 |
+
"grad_norm": 1.1031346321105957,
|
| 12761 |
+
"learning_rate": 1.4669381423308126e-05,
|
| 12762 |
+
"loss": 5.402,
|
| 12763 |
+
"step": 7288
|
| 12764 |
+
},
|
| 12765 |
+
{
|
| 12766 |
+
"epoch": 0.71,
|
| 12767 |
+
"grad_norm": 1.0289580821990967,
|
| 12768 |
+
"learning_rate": 1.4649990304440566e-05,
|
| 12769 |
+
"loss": 5.3957,
|
| 12770 |
+
"step": 7292
|
| 12771 |
+
},
|
| 12772 |
+
{
|
| 12773 |
+
"epoch": 0.71,
|
| 12774 |
+
"grad_norm": 1.1469511985778809,
|
| 12775 |
+
"learning_rate": 1.4630599185573007e-05,
|
| 12776 |
+
"loss": 5.3244,
|
| 12777 |
+
"step": 7296
|
| 12778 |
+
},
|
| 12779 |
+
{
|
| 12780 |
+
"epoch": 0.71,
|
| 12781 |
+
"grad_norm": 1.0669410228729248,
|
| 12782 |
+
"learning_rate": 1.461120806670545e-05,
|
| 12783 |
+
"loss": 5.431,
|
| 12784 |
+
"step": 7300
|
| 12785 |
+
},
|
| 12786 |
+
{
|
| 12787 |
+
"epoch": 0.71,
|
| 12788 |
+
"grad_norm": 1.05574631690979,
|
| 12789 |
+
"learning_rate": 1.4591816947837891e-05,
|
| 12790 |
+
"loss": 5.3382,
|
| 12791 |
+
"step": 7304
|
| 12792 |
+
},
|
| 12793 |
+
{
|
| 12794 |
+
"epoch": 0.71,
|
| 12795 |
+
"grad_norm": 1.0296452045440674,
|
| 12796 |
+
"learning_rate": 1.4572425828970332e-05,
|
| 12797 |
+
"loss": 5.337,
|
| 12798 |
+
"step": 7308
|
| 12799 |
+
},
|
| 12800 |
+
{
|
| 12801 |
+
"epoch": 0.71,
|
| 12802 |
+
"grad_norm": 1.0180591344833374,
|
| 12803 |
+
"learning_rate": 1.4553034710102772e-05,
|
| 12804 |
+
"loss": 5.3239,
|
| 12805 |
+
"step": 7312
|
| 12806 |
+
},
|
| 12807 |
+
{
|
| 12808 |
+
"epoch": 0.71,
|
| 12809 |
+
"grad_norm": 1.0508371591567993,
|
| 12810 |
+
"learning_rate": 1.4533643591235216e-05,
|
| 12811 |
+
"loss": 5.2944,
|
| 12812 |
+
"step": 7316
|
| 12813 |
+
},
|
| 12814 |
+
{
|
| 12815 |
+
"epoch": 0.71,
|
| 12816 |
+
"grad_norm": 1.0255225896835327,
|
| 12817 |
+
"learning_rate": 1.4514252472367657e-05,
|
| 12818 |
+
"loss": 5.3574,
|
| 12819 |
+
"step": 7320
|
| 12820 |
+
},
|
| 12821 |
+
{
|
| 12822 |
+
"epoch": 0.71,
|
| 12823 |
+
"grad_norm": 1.0599967241287231,
|
| 12824 |
+
"learning_rate": 1.4494861353500097e-05,
|
| 12825 |
+
"loss": 5.3568,
|
| 12826 |
+
"step": 7324
|
| 12827 |
+
},
|
| 12828 |
+
{
|
| 12829 |
+
"epoch": 0.71,
|
| 12830 |
+
"grad_norm": 0.9832557439804077,
|
| 12831 |
+
"learning_rate": 1.4475470234632538e-05,
|
| 12832 |
+
"loss": 5.2891,
|
| 12833 |
+
"step": 7328
|
| 12834 |
+
},
|
| 12835 |
+
{
|
| 12836 |
+
"epoch": 0.71,
|
| 12837 |
+
"grad_norm": 1.0541869401931763,
|
| 12838 |
+
"learning_rate": 1.4456079115764982e-05,
|
| 12839 |
+
"loss": 5.2502,
|
| 12840 |
+
"step": 7332
|
| 12841 |
+
},
|
| 12842 |
+
{
|
| 12843 |
+
"epoch": 0.71,
|
| 12844 |
+
"grad_norm": 1.069973111152649,
|
| 12845 |
+
"learning_rate": 1.4436687996897422e-05,
|
| 12846 |
+
"loss": 5.3376,
|
| 12847 |
+
"step": 7336
|
| 12848 |
+
},
|
| 12849 |
+
{
|
| 12850 |
+
"epoch": 0.71,
|
| 12851 |
+
"grad_norm": 1.0768502950668335,
|
| 12852 |
+
"learning_rate": 1.4417296878029863e-05,
|
| 12853 |
+
"loss": 5.2392,
|
| 12854 |
+
"step": 7340
|
| 12855 |
+
},
|
| 12856 |
+
{
|
| 12857 |
+
"epoch": 0.71,
|
| 12858 |
+
"grad_norm": 1.000628113746643,
|
| 12859 |
+
"learning_rate": 1.4397905759162305e-05,
|
| 12860 |
+
"loss": 5.3293,
|
| 12861 |
+
"step": 7344
|
| 12862 |
+
},
|
| 12863 |
+
{
|
| 12864 |
+
"epoch": 0.71,
|
| 12865 |
+
"grad_norm": 0.9960778951644897,
|
| 12866 |
+
"learning_rate": 1.4378514640294746e-05,
|
| 12867 |
+
"loss": 5.2655,
|
| 12868 |
+
"step": 7348
|
| 12869 |
+
},
|
| 12870 |
+
{
|
| 12871 |
+
"epoch": 0.71,
|
| 12872 |
+
"grad_norm": 1.0327279567718506,
|
| 12873 |
+
"learning_rate": 1.4359123521427186e-05,
|
| 12874 |
+
"loss": 5.3477,
|
| 12875 |
+
"step": 7352
|
| 12876 |
+
},
|
| 12877 |
+
{
|
| 12878 |
+
"epoch": 0.71,
|
| 12879 |
+
"grad_norm": 1.0809035301208496,
|
| 12880 |
+
"learning_rate": 1.4339732402559627e-05,
|
| 12881 |
+
"loss": 5.331,
|
| 12882 |
+
"step": 7356
|
| 12883 |
+
},
|
| 12884 |
+
{
|
| 12885 |
+
"epoch": 0.71,
|
| 12886 |
+
"grad_norm": 1.0690807104110718,
|
| 12887 |
+
"learning_rate": 1.432034128369207e-05,
|
| 12888 |
+
"loss": 5.3986,
|
| 12889 |
+
"step": 7360
|
| 12890 |
+
},
|
| 12891 |
+
{
|
| 12892 |
+
"epoch": 0.71,
|
| 12893 |
+
"grad_norm": 1.0245548486709595,
|
| 12894 |
+
"learning_rate": 1.4300950164824511e-05,
|
| 12895 |
+
"loss": 5.3024,
|
| 12896 |
+
"step": 7364
|
| 12897 |
+
},
|
| 12898 |
+
{
|
| 12899 |
+
"epoch": 0.71,
|
| 12900 |
+
"grad_norm": 0.9999493956565857,
|
| 12901 |
+
"learning_rate": 1.4281559045956952e-05,
|
| 12902 |
+
"loss": 5.2878,
|
| 12903 |
+
"step": 7368
|
| 12904 |
+
},
|
| 12905 |
+
{
|
| 12906 |
+
"epoch": 0.71,
|
| 12907 |
+
"grad_norm": 1.0037769079208374,
|
| 12908 |
+
"learning_rate": 1.4262167927089392e-05,
|
| 12909 |
+
"loss": 5.2216,
|
| 12910 |
+
"step": 7372
|
| 12911 |
+
},
|
| 12912 |
+
{
|
| 12913 |
+
"epoch": 0.72,
|
| 12914 |
+
"grad_norm": 1.002113699913025,
|
| 12915 |
+
"learning_rate": 1.4242776808221836e-05,
|
| 12916 |
+
"loss": 5.4347,
|
| 12917 |
+
"step": 7376
|
| 12918 |
+
},
|
| 12919 |
+
{
|
| 12920 |
+
"epoch": 0.72,
|
| 12921 |
+
"grad_norm": 1.04192054271698,
|
| 12922 |
+
"learning_rate": 1.4223385689354277e-05,
|
| 12923 |
+
"loss": 5.3812,
|
| 12924 |
+
"step": 7380
|
| 12925 |
+
},
|
| 12926 |
+
{
|
| 12927 |
+
"epoch": 0.72,
|
| 12928 |
+
"grad_norm": 1.0271295309066772,
|
| 12929 |
+
"learning_rate": 1.4203994570486717e-05,
|
| 12930 |
+
"loss": 5.427,
|
| 12931 |
+
"step": 7384
|
| 12932 |
+
},
|
| 12933 |
+
{
|
| 12934 |
+
"epoch": 0.72,
|
| 12935 |
+
"grad_norm": 1.0071477890014648,
|
| 12936 |
+
"learning_rate": 1.4184603451619158e-05,
|
| 12937 |
+
"loss": 5.2727,
|
| 12938 |
+
"step": 7388
|
| 12939 |
+
},
|
| 12940 |
+
{
|
| 12941 |
+
"epoch": 0.72,
|
| 12942 |
+
"grad_norm": 1.0228255987167358,
|
| 12943 |
+
"learning_rate": 1.4165212332751602e-05,
|
| 12944 |
+
"loss": 5.3244,
|
| 12945 |
+
"step": 7392
|
| 12946 |
+
},
|
| 12947 |
+
{
|
| 12948 |
+
"epoch": 0.72,
|
| 12949 |
+
"grad_norm": 1.0034810304641724,
|
| 12950 |
+
"learning_rate": 1.4145821213884042e-05,
|
| 12951 |
+
"loss": 5.3883,
|
| 12952 |
+
"step": 7396
|
| 12953 |
+
},
|
| 12954 |
+
{
|
| 12955 |
+
"epoch": 0.72,
|
| 12956 |
+
"grad_norm": 1.0305688381195068,
|
| 12957 |
+
"learning_rate": 1.4126430095016483e-05,
|
| 12958 |
+
"loss": 5.3851,
|
| 12959 |
+
"step": 7400
|
| 12960 |
+
},
|
| 12961 |
+
{
|
| 12962 |
+
"epoch": 0.72,
|
| 12963 |
+
"grad_norm": 1.074646234512329,
|
| 12964 |
+
"learning_rate": 1.4107038976148925e-05,
|
| 12965 |
+
"loss": 5.2406,
|
| 12966 |
+
"step": 7404
|
| 12967 |
+
},
|
| 12968 |
+
{
|
| 12969 |
+
"epoch": 0.72,
|
| 12970 |
+
"grad_norm": 0.9961770176887512,
|
| 12971 |
+
"learning_rate": 1.4087647857281366e-05,
|
| 12972 |
+
"loss": 5.2848,
|
| 12973 |
+
"step": 7408
|
| 12974 |
+
},
|
| 12975 |
+
{
|
| 12976 |
+
"epoch": 0.72,
|
| 12977 |
+
"grad_norm": 1.034627079963684,
|
| 12978 |
+
"learning_rate": 1.4068256738413808e-05,
|
| 12979 |
+
"loss": 5.2984,
|
| 12980 |
+
"step": 7412
|
| 12981 |
+
},
|
| 12982 |
+
{
|
| 12983 |
+
"epoch": 0.72,
|
| 12984 |
+
"grad_norm": 1.072096824645996,
|
| 12985 |
+
"learning_rate": 1.4048865619546248e-05,
|
| 12986 |
+
"loss": 5.3182,
|
| 12987 |
+
"step": 7416
|
| 12988 |
+
},
|
| 12989 |
+
{
|
| 12990 |
+
"epoch": 0.72,
|
| 12991 |
+
"grad_norm": 1.0899096727371216,
|
| 12992 |
+
"learning_rate": 1.402947450067869e-05,
|
| 12993 |
+
"loss": 5.3017,
|
| 12994 |
+
"step": 7420
|
| 12995 |
+
},
|
| 12996 |
+
{
|
| 12997 |
+
"epoch": 0.72,
|
| 12998 |
+
"grad_norm": 1.1063376665115356,
|
| 12999 |
+
"learning_rate": 1.4010083381811131e-05,
|
| 13000 |
+
"loss": 5.311,
|
| 13001 |
+
"step": 7424
|
| 13002 |
+
},
|
| 13003 |
+
{
|
| 13004 |
+
"epoch": 0.72,
|
| 13005 |
+
"grad_norm": 1.0748202800750732,
|
| 13006 |
+
"learning_rate": 1.3990692262943572e-05,
|
| 13007 |
+
"loss": 5.318,
|
| 13008 |
+
"step": 7428
|
| 13009 |
+
},
|
| 13010 |
+
{
|
| 13011 |
+
"epoch": 0.72,
|
| 13012 |
+
"grad_norm": 1.0851057767868042,
|
| 13013 |
+
"learning_rate": 1.3971301144076012e-05,
|
| 13014 |
+
"loss": 5.3425,
|
| 13015 |
+
"step": 7432
|
| 13016 |
+
},
|
| 13017 |
+
{
|
| 13018 |
+
"epoch": 0.72,
|
| 13019 |
+
"grad_norm": 0.9942495822906494,
|
| 13020 |
+
"learning_rate": 1.3951910025208456e-05,
|
| 13021 |
+
"loss": 5.3258,
|
| 13022 |
+
"step": 7436
|
| 13023 |
+
},
|
| 13024 |
+
{
|
| 13025 |
+
"epoch": 0.72,
|
| 13026 |
+
"grad_norm": 1.0894228219985962,
|
| 13027 |
+
"learning_rate": 1.3932518906340897e-05,
|
| 13028 |
+
"loss": 5.2769,
|
| 13029 |
+
"step": 7440
|
| 13030 |
+
},
|
| 13031 |
+
{
|
| 13032 |
+
"epoch": 0.72,
|
| 13033 |
+
"grad_norm": 0.9603523015975952,
|
| 13034 |
+
"learning_rate": 1.3913127787473337e-05,
|
| 13035 |
+
"loss": 5.4087,
|
| 13036 |
+
"step": 7444
|
| 13037 |
+
},
|
| 13038 |
+
{
|
| 13039 |
+
"epoch": 0.72,
|
| 13040 |
+
"grad_norm": 1.1049365997314453,
|
| 13041 |
+
"learning_rate": 1.3893736668605778e-05,
|
| 13042 |
+
"loss": 5.339,
|
| 13043 |
+
"step": 7448
|
| 13044 |
+
},
|
| 13045 |
+
{
|
| 13046 |
+
"epoch": 0.72,
|
| 13047 |
+
"grad_norm": 1.021468162536621,
|
| 13048 |
+
"learning_rate": 1.3874345549738222e-05,
|
| 13049 |
+
"loss": 5.3447,
|
| 13050 |
+
"step": 7452
|
| 13051 |
+
},
|
| 13052 |
+
{
|
| 13053 |
+
"epoch": 0.72,
|
| 13054 |
+
"grad_norm": 1.088437557220459,
|
| 13055 |
+
"learning_rate": 1.3854954430870662e-05,
|
| 13056 |
+
"loss": 5.3546,
|
| 13057 |
+
"step": 7456
|
| 13058 |
+
},
|
| 13059 |
+
{
|
| 13060 |
+
"epoch": 0.72,
|
| 13061 |
+
"grad_norm": 1.0266107320785522,
|
| 13062 |
+
"learning_rate": 1.3835563312003103e-05,
|
| 13063 |
+
"loss": 5.1867,
|
| 13064 |
+
"step": 7460
|
| 13065 |
+
},
|
| 13066 |
+
{
|
| 13067 |
+
"epoch": 0.72,
|
| 13068 |
+
"grad_norm": 1.088911533355713,
|
| 13069 |
+
"learning_rate": 1.3816172193135547e-05,
|
| 13070 |
+
"loss": 5.3247,
|
| 13071 |
+
"step": 7464
|
| 13072 |
+
},
|
| 13073 |
+
{
|
| 13074 |
+
"epoch": 0.72,
|
| 13075 |
+
"grad_norm": 1.0675928592681885,
|
| 13076 |
+
"learning_rate": 1.3796781074267987e-05,
|
| 13077 |
+
"loss": 5.2807,
|
| 13078 |
+
"step": 7468
|
| 13079 |
+
},
|
| 13080 |
+
{
|
| 13081 |
+
"epoch": 0.72,
|
| 13082 |
+
"grad_norm": 0.9511438608169556,
|
| 13083 |
+
"learning_rate": 1.3777389955400428e-05,
|
| 13084 |
+
"loss": 5.3774,
|
| 13085 |
+
"step": 7472
|
| 13086 |
+
},
|
| 13087 |
+
{
|
| 13088 |
+
"epoch": 0.72,
|
| 13089 |
+
"grad_norm": 1.0546114444732666,
|
| 13090 |
+
"learning_rate": 1.3757998836532868e-05,
|
| 13091 |
+
"loss": 5.2856,
|
| 13092 |
+
"step": 7476
|
| 13093 |
+
},
|
| 13094 |
+
{
|
| 13095 |
+
"epoch": 0.73,
|
| 13096 |
+
"grad_norm": 1.0445231199264526,
|
| 13097 |
+
"learning_rate": 1.373860771766531e-05,
|
| 13098 |
+
"loss": 5.3805,
|
| 13099 |
+
"step": 7480
|
| 13100 |
+
},
|
| 13101 |
+
{
|
| 13102 |
+
"epoch": 0.73,
|
| 13103 |
+
"grad_norm": 1.0493693351745605,
|
| 13104 |
+
"learning_rate": 1.3719216598797751e-05,
|
| 13105 |
+
"loss": 5.3603,
|
| 13106 |
+
"step": 7484
|
| 13107 |
+
},
|
| 13108 |
+
{
|
| 13109 |
+
"epoch": 0.73,
|
| 13110 |
+
"grad_norm": 1.0047773122787476,
|
| 13111 |
+
"learning_rate": 1.3699825479930192e-05,
|
| 13112 |
+
"loss": 5.3224,
|
| 13113 |
+
"step": 7488
|
| 13114 |
+
},
|
| 13115 |
+
{
|
| 13116 |
+
"epoch": 0.73,
|
| 13117 |
+
"grad_norm": 1.0536508560180664,
|
| 13118 |
+
"learning_rate": 1.3680434361062632e-05,
|
| 13119 |
+
"loss": 5.4235,
|
| 13120 |
+
"step": 7492
|
| 13121 |
+
},
|
| 13122 |
+
{
|
| 13123 |
+
"epoch": 0.73,
|
| 13124 |
+
"grad_norm": 1.0347819328308105,
|
| 13125 |
+
"learning_rate": 1.3661043242195076e-05,
|
| 13126 |
+
"loss": 5.2528,
|
| 13127 |
+
"step": 7496
|
| 13128 |
+
},
|
| 13129 |
+
{
|
| 13130 |
+
"epoch": 0.73,
|
| 13131 |
+
"grad_norm": 1.0392394065856934,
|
| 13132 |
+
"learning_rate": 1.3641652123327517e-05,
|
| 13133 |
+
"loss": 5.4127,
|
| 13134 |
+
"step": 7500
|
| 13135 |
+
},
|
| 13136 |
+
{
|
| 13137 |
+
"epoch": 0.73,
|
| 13138 |
+
"grad_norm": 1.0930792093276978,
|
| 13139 |
+
"learning_rate": 1.3622261004459957e-05,
|
| 13140 |
+
"loss": 5.3221,
|
| 13141 |
+
"step": 7504
|
| 13142 |
+
},
|
| 13143 |
+
{
|
| 13144 |
+
"epoch": 0.73,
|
| 13145 |
+
"grad_norm": 0.9962918758392334,
|
| 13146 |
+
"learning_rate": 1.3602869885592398e-05,
|
| 13147 |
+
"loss": 5.3108,
|
| 13148 |
+
"step": 7508
|
| 13149 |
+
},
|
| 13150 |
+
{
|
| 13151 |
+
"epoch": 0.73,
|
| 13152 |
+
"grad_norm": 1.030331015586853,
|
| 13153 |
+
"learning_rate": 1.3583478766724842e-05,
|
| 13154 |
+
"loss": 5.2632,
|
| 13155 |
+
"step": 7512
|
| 13156 |
+
},
|
| 13157 |
+
{
|
| 13158 |
+
"epoch": 0.73,
|
| 13159 |
+
"grad_norm": 1.008636236190796,
|
| 13160 |
+
"learning_rate": 1.3564087647857282e-05,
|
| 13161 |
+
"loss": 5.3089,
|
| 13162 |
+
"step": 7516
|
| 13163 |
+
},
|
| 13164 |
+
{
|
| 13165 |
+
"epoch": 0.73,
|
| 13166 |
+
"grad_norm": 1.006934642791748,
|
| 13167 |
+
"learning_rate": 1.3544696528989723e-05,
|
| 13168 |
+
"loss": 5.2903,
|
| 13169 |
+
"step": 7520
|
| 13170 |
+
},
|
| 13171 |
+
{
|
| 13172 |
+
"epoch": 0.73,
|
| 13173 |
+
"grad_norm": 1.0194462537765503,
|
| 13174 |
+
"learning_rate": 1.3525305410122167e-05,
|
| 13175 |
+
"loss": 5.3827,
|
| 13176 |
+
"step": 7524
|
| 13177 |
+
},
|
| 13178 |
+
{
|
| 13179 |
+
"epoch": 0.73,
|
| 13180 |
+
"grad_norm": 0.9879323840141296,
|
| 13181 |
+
"learning_rate": 1.3505914291254607e-05,
|
| 13182 |
+
"loss": 5.4307,
|
| 13183 |
+
"step": 7528
|
| 13184 |
+
},
|
| 13185 |
+
{
|
| 13186 |
+
"epoch": 0.73,
|
| 13187 |
+
"grad_norm": 1.0651185512542725,
|
| 13188 |
+
"learning_rate": 1.3486523172387048e-05,
|
| 13189 |
+
"loss": 5.3134,
|
| 13190 |
+
"step": 7532
|
| 13191 |
+
},
|
| 13192 |
+
{
|
| 13193 |
+
"epoch": 0.73,
|
| 13194 |
+
"grad_norm": 1.0858125686645508,
|
| 13195 |
+
"learning_rate": 1.3467132053519488e-05,
|
| 13196 |
+
"loss": 5.3024,
|
| 13197 |
+
"step": 7536
|
| 13198 |
+
},
|
| 13199 |
+
{
|
| 13200 |
+
"epoch": 0.73,
|
| 13201 |
+
"grad_norm": 1.0240780115127563,
|
| 13202 |
+
"learning_rate": 1.344774093465193e-05,
|
| 13203 |
+
"loss": 5.3555,
|
| 13204 |
+
"step": 7540
|
| 13205 |
+
},
|
| 13206 |
+
{
|
| 13207 |
+
"epoch": 0.73,
|
| 13208 |
+
"grad_norm": 1.0629985332489014,
|
| 13209 |
+
"learning_rate": 1.3428349815784373e-05,
|
| 13210 |
+
"loss": 5.3652,
|
| 13211 |
+
"step": 7544
|
| 13212 |
+
},
|
| 13213 |
+
{
|
| 13214 |
+
"epoch": 0.73,
|
| 13215 |
+
"grad_norm": 1.1222716569900513,
|
| 13216 |
+
"learning_rate": 1.3408958696916813e-05,
|
| 13217 |
+
"loss": 5.3845,
|
| 13218 |
+
"step": 7548
|
| 13219 |
+
},
|
| 13220 |
+
{
|
| 13221 |
+
"epoch": 0.73,
|
| 13222 |
+
"grad_norm": 1.0008291006088257,
|
| 13223 |
+
"learning_rate": 1.3389567578049254e-05,
|
| 13224 |
+
"loss": 5.3552,
|
| 13225 |
+
"step": 7552
|
| 13226 |
+
},
|
| 13227 |
+
{
|
| 13228 |
+
"epoch": 0.73,
|
| 13229 |
+
"grad_norm": 1.0860753059387207,
|
| 13230 |
+
"learning_rate": 1.3370176459181696e-05,
|
| 13231 |
+
"loss": 5.3044,
|
| 13232 |
+
"step": 7556
|
| 13233 |
+
},
|
| 13234 |
+
{
|
| 13235 |
+
"epoch": 0.73,
|
| 13236 |
+
"grad_norm": 1.0495448112487793,
|
| 13237 |
+
"learning_rate": 1.3350785340314136e-05,
|
| 13238 |
+
"loss": 5.3259,
|
| 13239 |
+
"step": 7560
|
| 13240 |
+
},
|
| 13241 |
+
{
|
| 13242 |
+
"epoch": 0.73,
|
| 13243 |
+
"grad_norm": 1.0432664155960083,
|
| 13244 |
+
"learning_rate": 1.3331394221446577e-05,
|
| 13245 |
+
"loss": 5.3482,
|
| 13246 |
+
"step": 7564
|
| 13247 |
+
},
|
| 13248 |
+
{
|
| 13249 |
+
"epoch": 0.73,
|
| 13250 |
+
"grad_norm": 1.0440526008605957,
|
| 13251 |
+
"learning_rate": 1.3312003102579018e-05,
|
| 13252 |
+
"loss": 5.2362,
|
| 13253 |
+
"step": 7568
|
| 13254 |
+
},
|
| 13255 |
+
{
|
| 13256 |
+
"epoch": 0.73,
|
| 13257 |
+
"grad_norm": 1.0317118167877197,
|
| 13258 |
+
"learning_rate": 1.3292611983711461e-05,
|
| 13259 |
+
"loss": 5.3579,
|
| 13260 |
+
"step": 7572
|
| 13261 |
+
},
|
| 13262 |
+
{
|
| 13263 |
+
"epoch": 0.73,
|
| 13264 |
+
"grad_norm": 0.9613714218139648,
|
| 13265 |
+
"learning_rate": 1.3273220864843902e-05,
|
| 13266 |
+
"loss": 5.3631,
|
| 13267 |
+
"step": 7576
|
| 13268 |
+
},
|
| 13269 |
+
{
|
| 13270 |
+
"epoch": 0.73,
|
| 13271 |
+
"grad_norm": 1.1608860492706299,
|
| 13272 |
+
"learning_rate": 1.3253829745976343e-05,
|
| 13273 |
+
"loss": 5.3547,
|
| 13274 |
+
"step": 7580
|
| 13275 |
+
},
|
| 13276 |
+
{
|
| 13277 |
+
"epoch": 0.74,
|
| 13278 |
+
"grad_norm": 1.06599760055542,
|
| 13279 |
+
"learning_rate": 1.3234438627108786e-05,
|
| 13280 |
+
"loss": 5.3728,
|
| 13281 |
+
"step": 7584
|
| 13282 |
+
},
|
| 13283 |
+
{
|
| 13284 |
+
"epoch": 0.74,
|
| 13285 |
+
"grad_norm": 1.0115044116973877,
|
| 13286 |
+
"learning_rate": 1.3215047508241227e-05,
|
| 13287 |
+
"loss": 5.2954,
|
| 13288 |
+
"step": 7588
|
| 13289 |
+
},
|
| 13290 |
+
{
|
| 13291 |
+
"epoch": 0.74,
|
| 13292 |
+
"grad_norm": 1.0504167079925537,
|
| 13293 |
+
"learning_rate": 1.3195656389373668e-05,
|
| 13294 |
+
"loss": 5.3378,
|
| 13295 |
+
"step": 7592
|
| 13296 |
+
},
|
| 13297 |
+
{
|
| 13298 |
+
"epoch": 0.74,
|
| 13299 |
+
"grad_norm": 1.1047917604446411,
|
| 13300 |
+
"learning_rate": 1.3176265270506108e-05,
|
| 13301 |
+
"loss": 5.3153,
|
| 13302 |
+
"step": 7596
|
| 13303 |
+
},
|
| 13304 |
+
{
|
| 13305 |
+
"epoch": 0.74,
|
| 13306 |
+
"grad_norm": 1.0862175226211548,
|
| 13307 |
+
"learning_rate": 1.3156874151638552e-05,
|
| 13308 |
+
"loss": 5.4315,
|
| 13309 |
+
"step": 7600
|
| 13310 |
+
},
|
| 13311 |
+
{
|
| 13312 |
+
"epoch": 0.74,
|
| 13313 |
+
"grad_norm": 1.05397629737854,
|
| 13314 |
+
"learning_rate": 1.3137483032770993e-05,
|
| 13315 |
+
"loss": 5.3807,
|
| 13316 |
+
"step": 7604
|
| 13317 |
+
},
|
| 13318 |
+
{
|
| 13319 |
+
"epoch": 0.74,
|
| 13320 |
+
"grad_norm": 1.0143108367919922,
|
| 13321 |
+
"learning_rate": 1.3118091913903433e-05,
|
| 13322 |
+
"loss": 5.346,
|
| 13323 |
+
"step": 7608
|
| 13324 |
+
},
|
| 13325 |
+
{
|
| 13326 |
+
"epoch": 0.74,
|
| 13327 |
+
"grad_norm": 1.0487464666366577,
|
| 13328 |
+
"learning_rate": 1.3098700795035874e-05,
|
| 13329 |
+
"loss": 5.2966,
|
| 13330 |
+
"step": 7612
|
| 13331 |
+
},
|
| 13332 |
+
{
|
| 13333 |
+
"epoch": 0.74,
|
| 13334 |
+
"grad_norm": 1.156467318534851,
|
| 13335 |
+
"learning_rate": 1.3079309676168316e-05,
|
| 13336 |
+
"loss": 5.24,
|
| 13337 |
+
"step": 7616
|
| 13338 |
+
},
|
| 13339 |
+
{
|
| 13340 |
+
"epoch": 0.74,
|
| 13341 |
+
"grad_norm": 1.0586912631988525,
|
| 13342 |
+
"learning_rate": 1.3059918557300756e-05,
|
| 13343 |
+
"loss": 5.3162,
|
| 13344 |
+
"step": 7620
|
| 13345 |
+
},
|
| 13346 |
+
{
|
| 13347 |
+
"epoch": 0.74,
|
| 13348 |
+
"grad_norm": 1.0472930669784546,
|
| 13349 |
+
"learning_rate": 1.3040527438433197e-05,
|
| 13350 |
+
"loss": 5.4309,
|
| 13351 |
+
"step": 7624
|
| 13352 |
+
},
|
| 13353 |
+
{
|
| 13354 |
+
"epoch": 0.74,
|
| 13355 |
+
"grad_norm": 1.0239797830581665,
|
| 13356 |
+
"learning_rate": 1.3021136319565637e-05,
|
| 13357 |
+
"loss": 5.3124,
|
| 13358 |
+
"step": 7628
|
| 13359 |
+
},
|
| 13360 |
+
{
|
| 13361 |
+
"epoch": 0.74,
|
| 13362 |
+
"grad_norm": 1.084915041923523,
|
| 13363 |
+
"learning_rate": 1.3001745200698081e-05,
|
| 13364 |
+
"loss": 5.382,
|
| 13365 |
+
"step": 7632
|
| 13366 |
+
},
|
| 13367 |
+
{
|
| 13368 |
+
"epoch": 0.74,
|
| 13369 |
+
"grad_norm": 1.121639370918274,
|
| 13370 |
+
"learning_rate": 1.2982354081830522e-05,
|
| 13371 |
+
"loss": 5.2888,
|
| 13372 |
+
"step": 7636
|
| 13373 |
+
},
|
| 13374 |
+
{
|
| 13375 |
+
"epoch": 0.74,
|
| 13376 |
+
"grad_norm": 1.06790030002594,
|
| 13377 |
+
"learning_rate": 1.2962962962962962e-05,
|
| 13378 |
+
"loss": 5.3484,
|
| 13379 |
+
"step": 7640
|
| 13380 |
+
},
|
| 13381 |
+
{
|
| 13382 |
+
"epoch": 0.74,
|
| 13383 |
+
"grad_norm": 1.0664012432098389,
|
| 13384 |
+
"learning_rate": 1.2943571844095406e-05,
|
| 13385 |
+
"loss": 5.3624,
|
| 13386 |
+
"step": 7644
|
| 13387 |
+
},
|
| 13388 |
+
{
|
| 13389 |
+
"epoch": 0.74,
|
| 13390 |
+
"grad_norm": 1.032593846321106,
|
| 13391 |
+
"learning_rate": 1.2924180725227847e-05,
|
| 13392 |
+
"loss": 5.3109,
|
| 13393 |
+
"step": 7648
|
| 13394 |
+
},
|
| 13395 |
+
{
|
| 13396 |
+
"epoch": 0.74,
|
| 13397 |
+
"grad_norm": 1.0182029008865356,
|
| 13398 |
+
"learning_rate": 1.2904789606360287e-05,
|
| 13399 |
+
"loss": 5.3785,
|
| 13400 |
+
"step": 7652
|
| 13401 |
+
},
|
| 13402 |
+
{
|
| 13403 |
+
"epoch": 0.74,
|
| 13404 |
+
"grad_norm": 0.9787065982818604,
|
| 13405 |
+
"learning_rate": 1.2885398487492728e-05,
|
| 13406 |
+
"loss": 5.2586,
|
| 13407 |
+
"step": 7656
|
| 13408 |
+
},
|
| 13409 |
+
{
|
| 13410 |
+
"epoch": 0.74,
|
| 13411 |
+
"grad_norm": 1.14923095703125,
|
| 13412 |
+
"learning_rate": 1.2866007368625172e-05,
|
| 13413 |
+
"loss": 5.3378,
|
| 13414 |
+
"step": 7660
|
| 13415 |
+
},
|
| 13416 |
+
{
|
| 13417 |
+
"epoch": 0.74,
|
| 13418 |
+
"grad_norm": 1.0064685344696045,
|
| 13419 |
+
"learning_rate": 1.2846616249757612e-05,
|
| 13420 |
+
"loss": 5.3339,
|
| 13421 |
+
"step": 7664
|
| 13422 |
+
},
|
| 13423 |
+
{
|
| 13424 |
+
"epoch": 0.74,
|
| 13425 |
+
"grad_norm": 1.015594720840454,
|
| 13426 |
+
"learning_rate": 1.2827225130890053e-05,
|
| 13427 |
+
"loss": 5.2412,
|
| 13428 |
+
"step": 7668
|
| 13429 |
+
},
|
| 13430 |
+
{
|
| 13431 |
+
"epoch": 0.74,
|
| 13432 |
+
"grad_norm": 1.1527953147888184,
|
| 13433 |
+
"learning_rate": 1.2807834012022494e-05,
|
| 13434 |
+
"loss": 5.4026,
|
| 13435 |
+
"step": 7672
|
| 13436 |
+
},
|
| 13437 |
+
{
|
| 13438 |
+
"epoch": 0.74,
|
| 13439 |
+
"grad_norm": 1.024170994758606,
|
| 13440 |
+
"learning_rate": 1.2788442893154936e-05,
|
| 13441 |
+
"loss": 5.2431,
|
| 13442 |
+
"step": 7676
|
| 13443 |
+
},
|
| 13444 |
+
{
|
| 13445 |
+
"epoch": 0.74,
|
| 13446 |
+
"grad_norm": 1.0834672451019287,
|
| 13447 |
+
"learning_rate": 1.2769051774287378e-05,
|
| 13448 |
+
"loss": 5.3018,
|
| 13449 |
+
"step": 7680
|
| 13450 |
+
},
|
| 13451 |
+
{
|
| 13452 |
+
"epoch": 0.75,
|
| 13453 |
+
"grad_norm": 1.00787353515625,
|
| 13454 |
+
"learning_rate": 1.2749660655419819e-05,
|
| 13455 |
+
"loss": 5.3129,
|
| 13456 |
+
"step": 7684
|
| 13457 |
+
},
|
| 13458 |
+
{
|
| 13459 |
+
"epoch": 0.75,
|
| 13460 |
+
"grad_norm": 1.0804412364959717,
|
| 13461 |
+
"learning_rate": 1.2730269536552259e-05,
|
| 13462 |
+
"loss": 5.3974,
|
| 13463 |
+
"step": 7688
|
| 13464 |
+
},
|
| 13465 |
+
{
|
| 13466 |
+
"epoch": 0.75,
|
| 13467 |
+
"grad_norm": 1.0361813306808472,
|
| 13468 |
+
"learning_rate": 1.2710878417684701e-05,
|
| 13469 |
+
"loss": 5.291,
|
| 13470 |
+
"step": 7692
|
| 13471 |
+
},
|
| 13472 |
+
{
|
| 13473 |
+
"epoch": 0.75,
|
| 13474 |
+
"grad_norm": 0.9639879465103149,
|
| 13475 |
+
"learning_rate": 1.2691487298817142e-05,
|
| 13476 |
+
"loss": 5.3649,
|
| 13477 |
+
"step": 7696
|
| 13478 |
+
},
|
| 13479 |
+
{
|
| 13480 |
+
"epoch": 0.75,
|
| 13481 |
+
"grad_norm": 1.035768985748291,
|
| 13482 |
+
"learning_rate": 1.2672096179949582e-05,
|
| 13483 |
+
"loss": 5.3159,
|
| 13484 |
+
"step": 7700
|
| 13485 |
+
},
|
| 13486 |
+
{
|
| 13487 |
+
"epoch": 0.75,
|
| 13488 |
+
"grad_norm": 1.002968430519104,
|
| 13489 |
+
"learning_rate": 1.2652705061082026e-05,
|
| 13490 |
+
"loss": 5.1975,
|
| 13491 |
+
"step": 7704
|
| 13492 |
+
},
|
| 13493 |
+
{
|
| 13494 |
+
"epoch": 0.75,
|
| 13495 |
+
"grad_norm": 1.0666626691818237,
|
| 13496 |
+
"learning_rate": 1.2633313942214467e-05,
|
| 13497 |
+
"loss": 5.2216,
|
| 13498 |
+
"step": 7708
|
| 13499 |
+
},
|
| 13500 |
+
{
|
| 13501 |
+
"epoch": 0.75,
|
| 13502 |
+
"grad_norm": 1.031752347946167,
|
| 13503 |
+
"learning_rate": 1.2613922823346907e-05,
|
| 13504 |
+
"loss": 5.4174,
|
| 13505 |
+
"step": 7712
|
| 13506 |
+
},
|
| 13507 |
+
{
|
| 13508 |
+
"epoch": 0.75,
|
| 13509 |
+
"grad_norm": 1.0436227321624756,
|
| 13510 |
+
"learning_rate": 1.2594531704479348e-05,
|
| 13511 |
+
"loss": 5.2295,
|
| 13512 |
+
"step": 7716
|
| 13513 |
+
},
|
| 13514 |
+
{
|
| 13515 |
+
"epoch": 0.75,
|
| 13516 |
+
"grad_norm": 1.0823688507080078,
|
| 13517 |
+
"learning_rate": 1.2575140585611792e-05,
|
| 13518 |
+
"loss": 5.3274,
|
| 13519 |
+
"step": 7720
|
| 13520 |
+
},
|
| 13521 |
+
{
|
| 13522 |
+
"epoch": 0.75,
|
| 13523 |
+
"grad_norm": 1.0667200088500977,
|
| 13524 |
+
"learning_rate": 1.2555749466744232e-05,
|
| 13525 |
+
"loss": 5.3149,
|
| 13526 |
+
"step": 7724
|
| 13527 |
+
},
|
| 13528 |
+
{
|
| 13529 |
+
"epoch": 0.75,
|
| 13530 |
+
"grad_norm": 1.0479573011398315,
|
| 13531 |
+
"learning_rate": 1.2536358347876673e-05,
|
| 13532 |
+
"loss": 5.2653,
|
| 13533 |
+
"step": 7728
|
| 13534 |
+
},
|
| 13535 |
+
{
|
| 13536 |
+
"epoch": 0.75,
|
| 13537 |
+
"grad_norm": 1.1973553895950317,
|
| 13538 |
+
"learning_rate": 1.2516967229009113e-05,
|
| 13539 |
+
"loss": 5.2504,
|
| 13540 |
+
"step": 7732
|
| 13541 |
+
},
|
| 13542 |
+
{
|
| 13543 |
+
"epoch": 0.75,
|
| 13544 |
+
"grad_norm": 1.0791184902191162,
|
| 13545 |
+
"learning_rate": 1.2497576110141556e-05,
|
| 13546 |
+
"loss": 5.2797,
|
| 13547 |
+
"step": 7736
|
| 13548 |
+
},
|
| 13549 |
+
{
|
| 13550 |
+
"epoch": 0.75,
|
| 13551 |
+
"grad_norm": 1.091112732887268,
|
| 13552 |
+
"learning_rate": 1.2478184991273998e-05,
|
| 13553 |
+
"loss": 5.3129,
|
| 13554 |
+
"step": 7740
|
| 13555 |
+
},
|
| 13556 |
+
{
|
| 13557 |
+
"epoch": 0.75,
|
| 13558 |
+
"grad_norm": 1.0607527494430542,
|
| 13559 |
+
"learning_rate": 1.2458793872406438e-05,
|
| 13560 |
+
"loss": 5.2997,
|
| 13561 |
+
"step": 7744
|
| 13562 |
+
},
|
| 13563 |
+
{
|
| 13564 |
+
"epoch": 0.75,
|
| 13565 |
+
"grad_norm": 1.0152305364608765,
|
| 13566 |
+
"learning_rate": 1.243940275353888e-05,
|
| 13567 |
+
"loss": 5.3655,
|
| 13568 |
+
"step": 7748
|
| 13569 |
+
},
|
| 13570 |
+
{
|
| 13571 |
+
"epoch": 0.75,
|
| 13572 |
+
"grad_norm": 1.009064793586731,
|
| 13573 |
+
"learning_rate": 1.2420011634671321e-05,
|
| 13574 |
+
"loss": 5.3058,
|
| 13575 |
+
"step": 7752
|
| 13576 |
+
},
|
| 13577 |
+
{
|
| 13578 |
+
"epoch": 0.75,
|
| 13579 |
+
"grad_norm": 1.0338549613952637,
|
| 13580 |
+
"learning_rate": 1.2400620515803762e-05,
|
| 13581 |
+
"loss": 5.2527,
|
| 13582 |
+
"step": 7756
|
| 13583 |
+
},
|
| 13584 |
+
{
|
| 13585 |
+
"epoch": 0.75,
|
| 13586 |
+
"grad_norm": 1.0737717151641846,
|
| 13587 |
+
"learning_rate": 1.2381229396936204e-05,
|
| 13588 |
+
"loss": 5.3859,
|
| 13589 |
+
"step": 7760
|
| 13590 |
+
},
|
| 13591 |
+
{
|
| 13592 |
+
"epoch": 0.75,
|
| 13593 |
+
"grad_norm": 1.0524028539657593,
|
| 13594 |
+
"learning_rate": 1.2361838278068645e-05,
|
| 13595 |
+
"loss": 5.407,
|
| 13596 |
+
"step": 7764
|
| 13597 |
+
},
|
| 13598 |
+
{
|
| 13599 |
+
"epoch": 0.75,
|
| 13600 |
+
"grad_norm": 1.0182230472564697,
|
| 13601 |
+
"learning_rate": 1.2342447159201087e-05,
|
| 13602 |
+
"loss": 5.3243,
|
| 13603 |
+
"step": 7768
|
| 13604 |
+
},
|
| 13605 |
+
{
|
| 13606 |
+
"epoch": 0.75,
|
| 13607 |
+
"grad_norm": 1.042325496673584,
|
| 13608 |
+
"learning_rate": 1.2323056040333527e-05,
|
| 13609 |
+
"loss": 5.372,
|
| 13610 |
+
"step": 7772
|
| 13611 |
+
},
|
| 13612 |
+
{
|
| 13613 |
+
"epoch": 0.75,
|
| 13614 |
+
"grad_norm": 1.099138855934143,
|
| 13615 |
+
"learning_rate": 1.230366492146597e-05,
|
| 13616 |
+
"loss": 5.3577,
|
| 13617 |
+
"step": 7776
|
| 13618 |
+
},
|
| 13619 |
+
{
|
| 13620 |
+
"epoch": 0.75,
|
| 13621 |
+
"grad_norm": 1.007309913635254,
|
| 13622 |
+
"learning_rate": 1.228427380259841e-05,
|
| 13623 |
+
"loss": 5.252,
|
| 13624 |
+
"step": 7780
|
| 13625 |
+
},
|
| 13626 |
+
{
|
| 13627 |
+
"epoch": 0.75,
|
| 13628 |
+
"grad_norm": 1.1510696411132812,
|
| 13629 |
+
"learning_rate": 1.2264882683730852e-05,
|
| 13630 |
+
"loss": 5.3726,
|
| 13631 |
+
"step": 7784
|
| 13632 |
+
},
|
| 13633 |
+
{
|
| 13634 |
+
"epoch": 0.76,
|
| 13635 |
+
"grad_norm": 1.0110242366790771,
|
| 13636 |
+
"learning_rate": 1.2245491564863293e-05,
|
| 13637 |
+
"loss": 5.3932,
|
| 13638 |
+
"step": 7788
|
| 13639 |
+
},
|
| 13640 |
+
{
|
| 13641 |
+
"epoch": 0.76,
|
| 13642 |
+
"grad_norm": 1.0620808601379395,
|
| 13643 |
+
"learning_rate": 1.2226100445995735e-05,
|
| 13644 |
+
"loss": 5.2691,
|
| 13645 |
+
"step": 7792
|
| 13646 |
+
},
|
| 13647 |
+
{
|
| 13648 |
+
"epoch": 0.76,
|
| 13649 |
+
"grad_norm": 1.052786946296692,
|
| 13650 |
+
"learning_rate": 1.2206709327128176e-05,
|
| 13651 |
+
"loss": 5.3597,
|
| 13652 |
+
"step": 7796
|
| 13653 |
+
},
|
| 13654 |
+
{
|
| 13655 |
+
"epoch": 0.76,
|
| 13656 |
+
"grad_norm": 1.1259844303131104,
|
| 13657 |
+
"learning_rate": 1.2187318208260618e-05,
|
| 13658 |
+
"loss": 5.3324,
|
| 13659 |
+
"step": 7800
|
| 13660 |
+
},
|
| 13661 |
+
{
|
| 13662 |
+
"epoch": 0.76,
|
| 13663 |
+
"grad_norm": 1.0244724750518799,
|
| 13664 |
+
"learning_rate": 1.2167927089393058e-05,
|
| 13665 |
+
"loss": 5.3643,
|
| 13666 |
+
"step": 7804
|
| 13667 |
+
},
|
| 13668 |
+
{
|
| 13669 |
+
"epoch": 0.76,
|
| 13670 |
+
"grad_norm": 1.1225011348724365,
|
| 13671 |
+
"learning_rate": 1.21485359705255e-05,
|
| 13672 |
+
"loss": 5.3501,
|
| 13673 |
+
"step": 7808
|
| 13674 |
+
},
|
| 13675 |
+
{
|
| 13676 |
+
"epoch": 0.76,
|
| 13677 |
+
"grad_norm": 0.9966182112693787,
|
| 13678 |
+
"learning_rate": 1.2129144851657941e-05,
|
| 13679 |
+
"loss": 5.3966,
|
| 13680 |
+
"step": 7812
|
| 13681 |
+
},
|
| 13682 |
+
{
|
| 13683 |
+
"epoch": 0.76,
|
| 13684 |
+
"grad_norm": 1.102308988571167,
|
| 13685 |
+
"learning_rate": 1.2109753732790383e-05,
|
| 13686 |
+
"loss": 5.2625,
|
| 13687 |
+
"step": 7816
|
| 13688 |
+
},
|
| 13689 |
+
{
|
| 13690 |
+
"epoch": 0.76,
|
| 13691 |
+
"grad_norm": 1.0900803804397583,
|
| 13692 |
+
"learning_rate": 1.2095210393639714e-05,
|
| 13693 |
+
"loss": 5.365,
|
| 13694 |
+
"step": 7820
|
| 13695 |
+
},
|
| 13696 |
+
{
|
| 13697 |
+
"epoch": 0.76,
|
| 13698 |
+
"grad_norm": 1.1339807510375977,
|
| 13699 |
+
"learning_rate": 1.2075819274772154e-05,
|
| 13700 |
+
"loss": 5.3816,
|
| 13701 |
+
"step": 7824
|
| 13702 |
+
},
|
| 13703 |
+
{
|
| 13704 |
+
"epoch": 0.76,
|
| 13705 |
+
"grad_norm": 1.077379822731018,
|
| 13706 |
+
"learning_rate": 1.2056428155904597e-05,
|
| 13707 |
+
"loss": 5.1332,
|
| 13708 |
+
"step": 7828
|
| 13709 |
+
},
|
| 13710 |
+
{
|
| 13711 |
+
"epoch": 0.76,
|
| 13712 |
+
"grad_norm": 1.1201417446136475,
|
| 13713 |
+
"learning_rate": 1.2037037037037037e-05,
|
| 13714 |
+
"loss": 5.3037,
|
| 13715 |
+
"step": 7832
|
| 13716 |
+
},
|
| 13717 |
+
{
|
| 13718 |
+
"epoch": 0.76,
|
| 13719 |
+
"grad_norm": 1.1047320365905762,
|
| 13720 |
+
"learning_rate": 1.201764591816948e-05,
|
| 13721 |
+
"loss": 5.4453,
|
| 13722 |
+
"step": 7836
|
| 13723 |
+
},
|
| 13724 |
+
{
|
| 13725 |
+
"epoch": 0.76,
|
| 13726 |
+
"grad_norm": 1.0625344514846802,
|
| 13727 |
+
"learning_rate": 1.1998254799301922e-05,
|
| 13728 |
+
"loss": 5.2676,
|
| 13729 |
+
"step": 7840
|
| 13730 |
+
},
|
| 13731 |
+
{
|
| 13732 |
+
"epoch": 0.76,
|
| 13733 |
+
"grad_norm": 1.0774505138397217,
|
| 13734 |
+
"learning_rate": 1.1978863680434362e-05,
|
| 13735 |
+
"loss": 5.3222,
|
| 13736 |
+
"step": 7844
|
| 13737 |
+
},
|
| 13738 |
+
{
|
| 13739 |
+
"epoch": 0.76,
|
| 13740 |
+
"grad_norm": 1.0557003021240234,
|
| 13741 |
+
"learning_rate": 1.1959472561566804e-05,
|
| 13742 |
+
"loss": 5.3235,
|
| 13743 |
+
"step": 7848
|
| 13744 |
+
},
|
| 13745 |
+
{
|
| 13746 |
+
"epoch": 0.76,
|
| 13747 |
+
"grad_norm": 0.9856312274932861,
|
| 13748 |
+
"learning_rate": 1.1940081442699245e-05,
|
| 13749 |
+
"loss": 5.2947,
|
| 13750 |
+
"step": 7852
|
| 13751 |
+
},
|
| 13752 |
+
{
|
| 13753 |
+
"epoch": 0.76,
|
| 13754 |
+
"grad_norm": 1.0185786485671997,
|
| 13755 |
+
"learning_rate": 1.1920690323831685e-05,
|
| 13756 |
+
"loss": 5.3262,
|
| 13757 |
+
"step": 7856
|
| 13758 |
+
},
|
| 13759 |
+
{
|
| 13760 |
+
"epoch": 0.76,
|
| 13761 |
+
"grad_norm": 1.0777360200881958,
|
| 13762 |
+
"learning_rate": 1.1901299204964126e-05,
|
| 13763 |
+
"loss": 5.254,
|
| 13764 |
+
"step": 7860
|
| 13765 |
+
},
|
| 13766 |
+
{
|
| 13767 |
+
"epoch": 0.76,
|
| 13768 |
+
"grad_norm": 1.0238205194473267,
|
| 13769 |
+
"learning_rate": 1.1881908086096568e-05,
|
| 13770 |
+
"loss": 5.356,
|
| 13771 |
+
"step": 7864
|
| 13772 |
+
},
|
| 13773 |
+
{
|
| 13774 |
+
"epoch": 0.76,
|
| 13775 |
+
"grad_norm": 1.0025471448898315,
|
| 13776 |
+
"learning_rate": 1.1862516967229009e-05,
|
| 13777 |
+
"loss": 5.2541,
|
| 13778 |
+
"step": 7868
|
| 13779 |
+
},
|
| 13780 |
+
{
|
| 13781 |
+
"epoch": 0.76,
|
| 13782 |
+
"grad_norm": 1.030316710472107,
|
| 13783 |
+
"learning_rate": 1.1843125848361451e-05,
|
| 13784 |
+
"loss": 5.2578,
|
| 13785 |
+
"step": 7872
|
| 13786 |
+
},
|
| 13787 |
+
{
|
| 13788 |
+
"epoch": 0.76,
|
| 13789 |
+
"grad_norm": 1.091535210609436,
|
| 13790 |
+
"learning_rate": 1.1823734729493891e-05,
|
| 13791 |
+
"loss": 5.3234,
|
| 13792 |
+
"step": 7876
|
| 13793 |
+
},
|
| 13794 |
+
{
|
| 13795 |
+
"epoch": 0.76,
|
| 13796 |
+
"grad_norm": 1.0827471017837524,
|
| 13797 |
+
"learning_rate": 1.1804343610626334e-05,
|
| 13798 |
+
"loss": 5.2876,
|
| 13799 |
+
"step": 7880
|
| 13800 |
+
},
|
| 13801 |
+
{
|
| 13802 |
+
"epoch": 0.76,
|
| 13803 |
+
"grad_norm": 1.0991337299346924,
|
| 13804 |
+
"learning_rate": 1.1784952491758774e-05,
|
| 13805 |
+
"loss": 5.473,
|
| 13806 |
+
"step": 7884
|
| 13807 |
+
},
|
| 13808 |
+
{
|
| 13809 |
+
"epoch": 0.76,
|
| 13810 |
+
"grad_norm": 1.0421675443649292,
|
| 13811 |
+
"learning_rate": 1.1765561372891216e-05,
|
| 13812 |
+
"loss": 5.3059,
|
| 13813 |
+
"step": 7888
|
| 13814 |
+
},
|
| 13815 |
+
{
|
| 13816 |
+
"epoch": 0.77,
|
| 13817 |
+
"grad_norm": 1.1006362438201904,
|
| 13818 |
+
"learning_rate": 1.1746170254023657e-05,
|
| 13819 |
+
"loss": 5.2836,
|
| 13820 |
+
"step": 7892
|
| 13821 |
+
},
|
| 13822 |
+
{
|
| 13823 |
+
"epoch": 0.77,
|
| 13824 |
+
"grad_norm": 1.0210435390472412,
|
| 13825 |
+
"learning_rate": 1.17267791351561e-05,
|
| 13826 |
+
"loss": 5.3632,
|
| 13827 |
+
"step": 7896
|
| 13828 |
+
},
|
| 13829 |
+
{
|
| 13830 |
+
"epoch": 0.77,
|
| 13831 |
+
"grad_norm": 0.9662442207336426,
|
| 13832 |
+
"learning_rate": 1.1707388016288541e-05,
|
| 13833 |
+
"loss": 5.3246,
|
| 13834 |
+
"step": 7900
|
| 13835 |
+
},
|
| 13836 |
+
{
|
| 13837 |
+
"epoch": 0.77,
|
| 13838 |
+
"grad_norm": 1.1299954652786255,
|
| 13839 |
+
"learning_rate": 1.1687996897420982e-05,
|
| 13840 |
+
"loss": 5.2499,
|
| 13841 |
+
"step": 7904
|
| 13842 |
+
},
|
| 13843 |
+
{
|
| 13844 |
+
"epoch": 0.77,
|
| 13845 |
+
"grad_norm": 1.0942116975784302,
|
| 13846 |
+
"learning_rate": 1.1668605778553424e-05,
|
| 13847 |
+
"loss": 5.2834,
|
| 13848 |
+
"step": 7908
|
| 13849 |
+
},
|
| 13850 |
+
{
|
| 13851 |
+
"epoch": 0.77,
|
| 13852 |
+
"grad_norm": 1.0262593030929565,
|
| 13853 |
+
"learning_rate": 1.1649214659685865e-05,
|
| 13854 |
+
"loss": 5.3228,
|
| 13855 |
+
"step": 7912
|
| 13856 |
+
},
|
| 13857 |
+
{
|
| 13858 |
+
"epoch": 0.77,
|
| 13859 |
+
"grad_norm": 1.0624995231628418,
|
| 13860 |
+
"learning_rate": 1.1629823540818307e-05,
|
| 13861 |
+
"loss": 5.2603,
|
| 13862 |
+
"step": 7916
|
| 13863 |
+
},
|
| 13864 |
+
{
|
| 13865 |
+
"epoch": 0.77,
|
| 13866 |
+
"grad_norm": 1.0851109027862549,
|
| 13867 |
+
"learning_rate": 1.1610432421950747e-05,
|
| 13868 |
+
"loss": 5.2343,
|
| 13869 |
+
"step": 7920
|
| 13870 |
+
},
|
| 13871 |
+
{
|
| 13872 |
+
"epoch": 0.77,
|
| 13873 |
+
"grad_norm": 1.0987358093261719,
|
| 13874 |
+
"learning_rate": 1.1591041303083188e-05,
|
| 13875 |
+
"loss": 5.2615,
|
| 13876 |
+
"step": 7924
|
| 13877 |
+
},
|
| 13878 |
+
{
|
| 13879 |
+
"epoch": 0.77,
|
| 13880 |
+
"grad_norm": 1.0852704048156738,
|
| 13881 |
+
"learning_rate": 1.1571650184215629e-05,
|
| 13882 |
+
"loss": 5.2886,
|
| 13883 |
+
"step": 7928
|
| 13884 |
+
},
|
| 13885 |
+
{
|
| 13886 |
+
"epoch": 0.77,
|
| 13887 |
+
"grad_norm": 1.0782544612884521,
|
| 13888 |
+
"learning_rate": 1.155225906534807e-05,
|
| 13889 |
+
"loss": 5.3585,
|
| 13890 |
+
"step": 7932
|
| 13891 |
+
},
|
| 13892 |
+
{
|
| 13893 |
+
"epoch": 0.77,
|
| 13894 |
+
"grad_norm": 1.0928773880004883,
|
| 13895 |
+
"learning_rate": 1.1532867946480511e-05,
|
| 13896 |
+
"loss": 5.2083,
|
| 13897 |
+
"step": 7936
|
| 13898 |
+
},
|
| 13899 |
+
{
|
| 13900 |
+
"epoch": 0.77,
|
| 13901 |
+
"grad_norm": 0.9779551029205322,
|
| 13902 |
+
"learning_rate": 1.1513476827612954e-05,
|
| 13903 |
+
"loss": 5.2969,
|
| 13904 |
+
"step": 7940
|
| 13905 |
+
},
|
| 13906 |
+
{
|
| 13907 |
+
"epoch": 0.77,
|
| 13908 |
+
"grad_norm": 1.0859659910202026,
|
| 13909 |
+
"learning_rate": 1.1494085708745394e-05,
|
| 13910 |
+
"loss": 5.3044,
|
| 13911 |
+
"step": 7944
|
| 13912 |
+
},
|
| 13913 |
+
{
|
| 13914 |
+
"epoch": 0.77,
|
| 13915 |
+
"grad_norm": 1.0626839399337769,
|
| 13916 |
+
"learning_rate": 1.1474694589877836e-05,
|
| 13917 |
+
"loss": 5.3455,
|
| 13918 |
+
"step": 7948
|
| 13919 |
+
},
|
| 13920 |
+
{
|
| 13921 |
+
"epoch": 0.77,
|
| 13922 |
+
"grad_norm": 1.1160836219787598,
|
| 13923 |
+
"learning_rate": 1.1455303471010277e-05,
|
| 13924 |
+
"loss": 5.3082,
|
| 13925 |
+
"step": 7952
|
| 13926 |
+
},
|
| 13927 |
+
{
|
| 13928 |
+
"epoch": 0.77,
|
| 13929 |
+
"grad_norm": 1.089357614517212,
|
| 13930 |
+
"learning_rate": 1.1435912352142719e-05,
|
| 13931 |
+
"loss": 5.3013,
|
| 13932 |
+
"step": 7956
|
| 13933 |
+
},
|
| 13934 |
+
{
|
| 13935 |
+
"epoch": 0.77,
|
| 13936 |
+
"grad_norm": 0.9616773128509521,
|
| 13937 |
+
"learning_rate": 1.1416521233275161e-05,
|
| 13938 |
+
"loss": 5.2995,
|
| 13939 |
+
"step": 7960
|
| 13940 |
+
},
|
| 13941 |
+
{
|
| 13942 |
+
"epoch": 0.77,
|
| 13943 |
+
"grad_norm": 1.0657833814620972,
|
| 13944 |
+
"learning_rate": 1.1397130114407602e-05,
|
| 13945 |
+
"loss": 5.3208,
|
| 13946 |
+
"step": 7964
|
| 13947 |
+
},
|
| 13948 |
+
{
|
| 13949 |
+
"epoch": 0.77,
|
| 13950 |
+
"grad_norm": 1.0845454931259155,
|
| 13951 |
+
"learning_rate": 1.1377738995540044e-05,
|
| 13952 |
+
"loss": 5.3213,
|
| 13953 |
+
"step": 7968
|
| 13954 |
+
},
|
| 13955 |
+
{
|
| 13956 |
+
"epoch": 0.77,
|
| 13957 |
+
"grad_norm": 1.0332484245300293,
|
| 13958 |
+
"learning_rate": 1.1358347876672485e-05,
|
| 13959 |
+
"loss": 5.2833,
|
| 13960 |
+
"step": 7972
|
| 13961 |
+
},
|
| 13962 |
+
{
|
| 13963 |
+
"epoch": 0.77,
|
| 13964 |
+
"grad_norm": 1.0877047777175903,
|
| 13965 |
+
"learning_rate": 1.1338956757804927e-05,
|
| 13966 |
+
"loss": 5.2889,
|
| 13967 |
+
"step": 7976
|
| 13968 |
+
},
|
| 13969 |
+
{
|
| 13970 |
+
"epoch": 0.77,
|
| 13971 |
+
"grad_norm": 1.0782090425491333,
|
| 13972 |
+
"learning_rate": 1.1319565638937367e-05,
|
| 13973 |
+
"loss": 5.2914,
|
| 13974 |
+
"step": 7980
|
| 13975 |
+
},
|
| 13976 |
+
{
|
| 13977 |
+
"epoch": 0.77,
|
| 13978 |
+
"grad_norm": 1.0773468017578125,
|
| 13979 |
+
"learning_rate": 1.130017452006981e-05,
|
| 13980 |
+
"loss": 5.2853,
|
| 13981 |
+
"step": 7984
|
| 13982 |
+
},
|
| 13983 |
+
{
|
| 13984 |
+
"epoch": 0.77,
|
| 13985 |
+
"grad_norm": 1.1031116247177124,
|
| 13986 |
+
"learning_rate": 1.128078340120225e-05,
|
| 13987 |
+
"loss": 5.2285,
|
| 13988 |
+
"step": 7988
|
| 13989 |
+
},
|
| 13990 |
+
{
|
| 13991 |
+
"epoch": 0.77,
|
| 13992 |
+
"grad_norm": 1.0579017400741577,
|
| 13993 |
+
"learning_rate": 1.126139228233469e-05,
|
| 13994 |
+
"loss": 5.2608,
|
| 13995 |
+
"step": 7992
|
| 13996 |
+
},
|
| 13997 |
+
{
|
| 13998 |
+
"epoch": 0.78,
|
| 13999 |
+
"grad_norm": 1.0039610862731934,
|
| 14000 |
+
"learning_rate": 1.1242001163467133e-05,
|
| 14001 |
+
"loss": 5.2801,
|
| 14002 |
+
"step": 7996
|
| 14003 |
+
},
|
| 14004 |
+
{
|
| 14005 |
+
"epoch": 0.78,
|
| 14006 |
+
"grad_norm": 1.1101170778274536,
|
| 14007 |
+
"learning_rate": 1.1222610044599573e-05,
|
| 14008 |
+
"loss": 5.3003,
|
| 14009 |
+
"step": 8000
|
| 14010 |
+
},
|
| 14011 |
+
{
|
| 14012 |
+
"epoch": 0.78,
|
| 14013 |
+
"grad_norm": 1.1231920719146729,
|
| 14014 |
+
"learning_rate": 1.1203218925732014e-05,
|
| 14015 |
+
"loss": 5.2951,
|
| 14016 |
+
"step": 8004
|
| 14017 |
+
},
|
| 14018 |
+
{
|
| 14019 |
+
"epoch": 0.78,
|
| 14020 |
+
"grad_norm": 1.104062557220459,
|
| 14021 |
+
"learning_rate": 1.1183827806864456e-05,
|
| 14022 |
+
"loss": 5.2419,
|
| 14023 |
+
"step": 8008
|
| 14024 |
+
},
|
| 14025 |
+
{
|
| 14026 |
+
"epoch": 0.78,
|
| 14027 |
+
"grad_norm": 1.0631533861160278,
|
| 14028 |
+
"learning_rate": 1.1164436687996897e-05,
|
| 14029 |
+
"loss": 5.2927,
|
| 14030 |
+
"step": 8012
|
| 14031 |
+
},
|
| 14032 |
+
{
|
| 14033 |
+
"epoch": 0.78,
|
| 14034 |
+
"grad_norm": 1.0923023223876953,
|
| 14035 |
+
"learning_rate": 1.1145045569129339e-05,
|
| 14036 |
+
"loss": 5.3566,
|
| 14037 |
+
"step": 8016
|
| 14038 |
+
},
|
| 14039 |
+
{
|
| 14040 |
+
"epoch": 0.78,
|
| 14041 |
+
"grad_norm": 1.1646323204040527,
|
| 14042 |
+
"learning_rate": 1.1125654450261781e-05,
|
| 14043 |
+
"loss": 5.229,
|
| 14044 |
+
"step": 8020
|
| 14045 |
+
},
|
| 14046 |
+
{
|
| 14047 |
+
"epoch": 0.78,
|
| 14048 |
+
"grad_norm": 1.0775333642959595,
|
| 14049 |
+
"learning_rate": 1.1106263331394222e-05,
|
| 14050 |
+
"loss": 5.3121,
|
| 14051 |
+
"step": 8024
|
| 14052 |
+
},
|
| 14053 |
+
{
|
| 14054 |
+
"epoch": 0.78,
|
| 14055 |
+
"grad_norm": 1.101682186126709,
|
| 14056 |
+
"learning_rate": 1.1086872212526664e-05,
|
| 14057 |
+
"loss": 5.3746,
|
| 14058 |
+
"step": 8028
|
| 14059 |
+
},
|
| 14060 |
+
{
|
| 14061 |
+
"epoch": 0.78,
|
| 14062 |
+
"grad_norm": 1.06061589717865,
|
| 14063 |
+
"learning_rate": 1.1067481093659105e-05,
|
| 14064 |
+
"loss": 5.2878,
|
| 14065 |
+
"step": 8032
|
| 14066 |
+
},
|
| 14067 |
+
{
|
| 14068 |
+
"epoch": 0.78,
|
| 14069 |
+
"grad_norm": 1.0500963926315308,
|
| 14070 |
+
"learning_rate": 1.1048089974791547e-05,
|
| 14071 |
+
"loss": 5.3188,
|
| 14072 |
+
"step": 8036
|
| 14073 |
+
},
|
| 14074 |
+
{
|
| 14075 |
+
"epoch": 0.78,
|
| 14076 |
+
"grad_norm": 1.1280819177627563,
|
| 14077 |
+
"learning_rate": 1.1028698855923987e-05,
|
| 14078 |
+
"loss": 5.2829,
|
| 14079 |
+
"step": 8040
|
| 14080 |
+
},
|
| 14081 |
+
{
|
| 14082 |
+
"epoch": 0.78,
|
| 14083 |
+
"grad_norm": 1.0025454759597778,
|
| 14084 |
+
"learning_rate": 1.100930773705643e-05,
|
| 14085 |
+
"loss": 5.3121,
|
| 14086 |
+
"step": 8044
|
| 14087 |
+
},
|
| 14088 |
+
{
|
| 14089 |
+
"epoch": 0.78,
|
| 14090 |
+
"grad_norm": 1.0445626974105835,
|
| 14091 |
+
"learning_rate": 1.098991661818887e-05,
|
| 14092 |
+
"loss": 5.3264,
|
| 14093 |
+
"step": 8048
|
| 14094 |
+
},
|
| 14095 |
+
{
|
| 14096 |
+
"epoch": 0.78,
|
| 14097 |
+
"grad_norm": 1.0949641466140747,
|
| 14098 |
+
"learning_rate": 1.0970525499321312e-05,
|
| 14099 |
+
"loss": 5.3306,
|
| 14100 |
+
"step": 8052
|
| 14101 |
+
},
|
| 14102 |
+
{
|
| 14103 |
+
"epoch": 0.78,
|
| 14104 |
+
"grad_norm": 1.044668197631836,
|
| 14105 |
+
"learning_rate": 1.0951134380453753e-05,
|
| 14106 |
+
"loss": 5.3335,
|
| 14107 |
+
"step": 8056
|
| 14108 |
+
},
|
| 14109 |
+
{
|
| 14110 |
+
"epoch": 0.78,
|
| 14111 |
+
"grad_norm": 1.079334020614624,
|
| 14112 |
+
"learning_rate": 1.0931743261586193e-05,
|
| 14113 |
+
"loss": 5.3383,
|
| 14114 |
+
"step": 8060
|
| 14115 |
+
},
|
| 14116 |
+
{
|
| 14117 |
+
"epoch": 0.78,
|
| 14118 |
+
"grad_norm": 1.0334477424621582,
|
| 14119 |
+
"learning_rate": 1.0912352142718636e-05,
|
| 14120 |
+
"loss": 5.3128,
|
| 14121 |
+
"step": 8064
|
| 14122 |
+
},
|
| 14123 |
+
{
|
| 14124 |
+
"epoch": 0.78,
|
| 14125 |
+
"grad_norm": 0.9899519681930542,
|
| 14126 |
+
"learning_rate": 1.0892961023851076e-05,
|
| 14127 |
+
"loss": 5.3245,
|
| 14128 |
+
"step": 8068
|
| 14129 |
+
},
|
| 14130 |
+
{
|
| 14131 |
+
"epoch": 0.78,
|
| 14132 |
+
"grad_norm": 1.0130146741867065,
|
| 14133 |
+
"learning_rate": 1.0873569904983517e-05,
|
| 14134 |
+
"loss": 5.2891,
|
| 14135 |
+
"step": 8072
|
| 14136 |
+
},
|
| 14137 |
+
{
|
| 14138 |
+
"epoch": 0.78,
|
| 14139 |
+
"grad_norm": 1.0125179290771484,
|
| 14140 |
+
"learning_rate": 1.0854178786115959e-05,
|
| 14141 |
+
"loss": 5.3007,
|
| 14142 |
+
"step": 8076
|
| 14143 |
+
},
|
| 14144 |
+
{
|
| 14145 |
+
"epoch": 0.78,
|
| 14146 |
+
"grad_norm": 1.0482258796691895,
|
| 14147 |
+
"learning_rate": 1.0834787667248401e-05,
|
| 14148 |
+
"loss": 5.2417,
|
| 14149 |
+
"step": 8080
|
| 14150 |
+
},
|
| 14151 |
+
{
|
| 14152 |
+
"epoch": 0.78,
|
| 14153 |
+
"grad_norm": 1.055640697479248,
|
| 14154 |
+
"learning_rate": 1.0815396548380842e-05,
|
| 14155 |
+
"loss": 5.3779,
|
| 14156 |
+
"step": 8084
|
| 14157 |
+
},
|
| 14158 |
+
{
|
| 14159 |
+
"epoch": 0.78,
|
| 14160 |
+
"grad_norm": 1.0579723119735718,
|
| 14161 |
+
"learning_rate": 1.0796005429513284e-05,
|
| 14162 |
+
"loss": 5.232,
|
| 14163 |
+
"step": 8088
|
| 14164 |
+
},
|
| 14165 |
+
{
|
| 14166 |
+
"epoch": 0.78,
|
| 14167 |
+
"grad_norm": 1.083598017692566,
|
| 14168 |
+
"learning_rate": 1.0776614310645724e-05,
|
| 14169 |
+
"loss": 5.4277,
|
| 14170 |
+
"step": 8092
|
| 14171 |
+
},
|
| 14172 |
+
{
|
| 14173 |
+
"epoch": 0.78,
|
| 14174 |
+
"grad_norm": 0.9834340214729309,
|
| 14175 |
+
"learning_rate": 1.0757223191778167e-05,
|
| 14176 |
+
"loss": 5.2498,
|
| 14177 |
+
"step": 8096
|
| 14178 |
+
},
|
| 14179 |
+
{
|
| 14180 |
+
"epoch": 0.79,
|
| 14181 |
+
"grad_norm": 1.0400574207305908,
|
| 14182 |
+
"learning_rate": 1.0737832072910607e-05,
|
| 14183 |
+
"loss": 5.3321,
|
| 14184 |
+
"step": 8100
|
| 14185 |
+
},
|
| 14186 |
+
{
|
| 14187 |
+
"epoch": 0.79,
|
| 14188 |
+
"grad_norm": 1.083775281906128,
|
| 14189 |
+
"learning_rate": 1.071844095404305e-05,
|
| 14190 |
+
"loss": 5.1589,
|
| 14191 |
+
"step": 8104
|
| 14192 |
+
},
|
| 14193 |
+
{
|
| 14194 |
+
"epoch": 0.79,
|
| 14195 |
+
"grad_norm": 1.035290241241455,
|
| 14196 |
+
"learning_rate": 1.069904983517549e-05,
|
| 14197 |
+
"loss": 5.4106,
|
| 14198 |
+
"step": 8108
|
| 14199 |
+
},
|
| 14200 |
+
{
|
| 14201 |
+
"epoch": 0.79,
|
| 14202 |
+
"grad_norm": 1.0348436832427979,
|
| 14203 |
+
"learning_rate": 1.0679658716307932e-05,
|
| 14204 |
+
"loss": 5.2728,
|
| 14205 |
+
"step": 8112
|
| 14206 |
+
},
|
| 14207 |
+
{
|
| 14208 |
+
"epoch": 0.79,
|
| 14209 |
+
"grad_norm": 1.058597445487976,
|
| 14210 |
+
"learning_rate": 1.0660267597440373e-05,
|
| 14211 |
+
"loss": 5.2081,
|
| 14212 |
+
"step": 8116
|
| 14213 |
+
},
|
| 14214 |
+
{
|
| 14215 |
+
"epoch": 0.79,
|
| 14216 |
+
"grad_norm": 1.0467309951782227,
|
| 14217 |
+
"learning_rate": 1.0640876478572815e-05,
|
| 14218 |
+
"loss": 5.232,
|
| 14219 |
+
"step": 8120
|
| 14220 |
+
},
|
| 14221 |
+
{
|
| 14222 |
+
"epoch": 0.79,
|
| 14223 |
+
"grad_norm": 0.9697101712226868,
|
| 14224 |
+
"learning_rate": 1.0621485359705256e-05,
|
| 14225 |
+
"loss": 5.2641,
|
| 14226 |
+
"step": 8124
|
| 14227 |
+
},
|
| 14228 |
+
{
|
| 14229 |
+
"epoch": 0.79,
|
| 14230 |
+
"grad_norm": 1.0596665143966675,
|
| 14231 |
+
"learning_rate": 1.0602094240837698e-05,
|
| 14232 |
+
"loss": 5.2865,
|
| 14233 |
+
"step": 8128
|
| 14234 |
+
},
|
| 14235 |
+
{
|
| 14236 |
+
"epoch": 0.79,
|
| 14237 |
+
"grad_norm": 1.0948309898376465,
|
| 14238 |
+
"learning_rate": 1.0582703121970138e-05,
|
| 14239 |
+
"loss": 5.2826,
|
| 14240 |
+
"step": 8132
|
| 14241 |
+
},
|
| 14242 |
+
{
|
| 14243 |
+
"epoch": 0.79,
|
| 14244 |
+
"grad_norm": 1.0270111560821533,
|
| 14245 |
+
"learning_rate": 1.0563312003102579e-05,
|
| 14246 |
+
"loss": 5.2726,
|
| 14247 |
+
"step": 8136
|
| 14248 |
+
},
|
| 14249 |
+
{
|
| 14250 |
+
"epoch": 0.79,
|
| 14251 |
+
"grad_norm": 1.0012414455413818,
|
| 14252 |
+
"learning_rate": 1.0543920884235021e-05,
|
| 14253 |
+
"loss": 5.3184,
|
| 14254 |
+
"step": 8140
|
| 14255 |
+
},
|
| 14256 |
+
{
|
| 14257 |
+
"epoch": 0.79,
|
| 14258 |
+
"grad_norm": 1.019332766532898,
|
| 14259 |
+
"learning_rate": 1.0524529765367462e-05,
|
| 14260 |
+
"loss": 5.2908,
|
| 14261 |
+
"step": 8144
|
| 14262 |
+
},
|
| 14263 |
+
{
|
| 14264 |
+
"epoch": 0.79,
|
| 14265 |
+
"grad_norm": 1.0300483703613281,
|
| 14266 |
+
"learning_rate": 1.0505138646499904e-05,
|
| 14267 |
+
"loss": 5.3333,
|
| 14268 |
+
"step": 8148
|
| 14269 |
+
},
|
| 14270 |
+
{
|
| 14271 |
+
"epoch": 0.79,
|
| 14272 |
+
"grad_norm": 1.0536975860595703,
|
| 14273 |
+
"learning_rate": 1.0485747527632344e-05,
|
| 14274 |
+
"loss": 5.2944,
|
| 14275 |
+
"step": 8152
|
| 14276 |
+
},
|
| 14277 |
+
{
|
| 14278 |
+
"epoch": 0.79,
|
| 14279 |
+
"grad_norm": 0.9881764650344849,
|
| 14280 |
+
"learning_rate": 1.0466356408764787e-05,
|
| 14281 |
+
"loss": 5.2767,
|
| 14282 |
+
"step": 8156
|
| 14283 |
+
},
|
| 14284 |
+
{
|
| 14285 |
+
"epoch": 0.79,
|
| 14286 |
+
"grad_norm": 1.0163639783859253,
|
| 14287 |
+
"learning_rate": 1.0446965289897227e-05,
|
| 14288 |
+
"loss": 5.2723,
|
| 14289 |
+
"step": 8160
|
| 14290 |
+
},
|
| 14291 |
+
{
|
| 14292 |
+
"epoch": 0.79,
|
| 14293 |
+
"grad_norm": 1.0919207334518433,
|
| 14294 |
+
"learning_rate": 1.042757417102967e-05,
|
| 14295 |
+
"loss": 5.3327,
|
| 14296 |
+
"step": 8164
|
| 14297 |
+
},
|
| 14298 |
+
{
|
| 14299 |
+
"epoch": 0.79,
|
| 14300 |
+
"grad_norm": 1.0306917428970337,
|
| 14301 |
+
"learning_rate": 1.040818305216211e-05,
|
| 14302 |
+
"loss": 5.2565,
|
| 14303 |
+
"step": 8168
|
| 14304 |
+
},
|
| 14305 |
+
{
|
| 14306 |
+
"epoch": 0.79,
|
| 14307 |
+
"grad_norm": 1.0125017166137695,
|
| 14308 |
+
"learning_rate": 1.0388791933294552e-05,
|
| 14309 |
+
"loss": 5.342,
|
| 14310 |
+
"step": 8172
|
| 14311 |
+
},
|
| 14312 |
+
{
|
| 14313 |
+
"epoch": 0.79,
|
| 14314 |
+
"grad_norm": 1.0879104137420654,
|
| 14315 |
+
"learning_rate": 1.0369400814426993e-05,
|
| 14316 |
+
"loss": 5.3122,
|
| 14317 |
+
"step": 8176
|
| 14318 |
+
},
|
| 14319 |
+
{
|
| 14320 |
+
"epoch": 0.79,
|
| 14321 |
+
"grad_norm": 0.9527262449264526,
|
| 14322 |
+
"learning_rate": 1.0350009695559435e-05,
|
| 14323 |
+
"loss": 5.284,
|
| 14324 |
+
"step": 8180
|
| 14325 |
+
},
|
| 14326 |
+
{
|
| 14327 |
+
"epoch": 0.79,
|
| 14328 |
+
"grad_norm": 1.0486680269241333,
|
| 14329 |
+
"learning_rate": 1.0330618576691875e-05,
|
| 14330 |
+
"loss": 5.2428,
|
| 14331 |
+
"step": 8184
|
| 14332 |
+
},
|
| 14333 |
+
{
|
| 14334 |
+
"epoch": 0.79,
|
| 14335 |
+
"grad_norm": 1.0278397798538208,
|
| 14336 |
+
"learning_rate": 1.0311227457824318e-05,
|
| 14337 |
+
"loss": 5.3466,
|
| 14338 |
+
"step": 8188
|
| 14339 |
+
},
|
| 14340 |
+
{
|
| 14341 |
+
"epoch": 0.79,
|
| 14342 |
+
"grad_norm": 1.0635344982147217,
|
| 14343 |
+
"learning_rate": 1.0291836338956758e-05,
|
| 14344 |
+
"loss": 5.2645,
|
| 14345 |
+
"step": 8192
|
| 14346 |
+
},
|
| 14347 |
+
{
|
| 14348 |
+
"epoch": 0.79,
|
| 14349 |
+
"grad_norm": 1.0180613994598389,
|
| 14350 |
+
"learning_rate": 1.02724452200892e-05,
|
| 14351 |
+
"loss": 5.3552,
|
| 14352 |
+
"step": 8196
|
| 14353 |
+
},
|
| 14354 |
+
{
|
| 14355 |
+
"epoch": 0.8,
|
| 14356 |
+
"grad_norm": 1.0265159606933594,
|
| 14357 |
+
"learning_rate": 1.0253054101221641e-05,
|
| 14358 |
+
"loss": 5.3422,
|
| 14359 |
+
"step": 8200
|
| 14360 |
+
},
|
| 14361 |
+
{
|
| 14362 |
+
"epoch": 0.8,
|
| 14363 |
+
"grad_norm": 1.029842495918274,
|
| 14364 |
+
"learning_rate": 1.0233662982354082e-05,
|
| 14365 |
+
"loss": 5.4009,
|
| 14366 |
+
"step": 8204
|
| 14367 |
+
},
|
| 14368 |
+
{
|
| 14369 |
+
"epoch": 0.8,
|
| 14370 |
+
"grad_norm": 1.0848013162612915,
|
| 14371 |
+
"learning_rate": 1.0214271863486524e-05,
|
| 14372 |
+
"loss": 5.3982,
|
| 14373 |
+
"step": 8208
|
| 14374 |
+
},
|
| 14375 |
+
{
|
| 14376 |
+
"epoch": 0.8,
|
| 14377 |
+
"grad_norm": 1.0277658700942993,
|
| 14378 |
+
"learning_rate": 1.0194880744618964e-05,
|
| 14379 |
+
"loss": 5.3567,
|
| 14380 |
+
"step": 8212
|
| 14381 |
+
},
|
| 14382 |
+
{
|
| 14383 |
+
"epoch": 0.8,
|
| 14384 |
+
"grad_norm": 1.0273760557174683,
|
| 14385 |
+
"learning_rate": 1.0175489625751407e-05,
|
| 14386 |
+
"loss": 5.2762,
|
| 14387 |
+
"step": 8216
|
| 14388 |
+
},
|
| 14389 |
+
{
|
| 14390 |
+
"epoch": 0.8,
|
| 14391 |
+
"grad_norm": 0.9951087832450867,
|
| 14392 |
+
"learning_rate": 1.0156098506883847e-05,
|
| 14393 |
+
"loss": 5.3267,
|
| 14394 |
+
"step": 8220
|
| 14395 |
+
},
|
| 14396 |
+
{
|
| 14397 |
+
"epoch": 0.8,
|
| 14398 |
+
"grad_norm": 1.0347890853881836,
|
| 14399 |
+
"learning_rate": 1.013670738801629e-05,
|
| 14400 |
+
"loss": 5.2681,
|
| 14401 |
+
"step": 8224
|
| 14402 |
+
},
|
| 14403 |
+
{
|
| 14404 |
+
"epoch": 0.8,
|
| 14405 |
+
"grad_norm": 1.076242446899414,
|
| 14406 |
+
"learning_rate": 1.011731626914873e-05,
|
| 14407 |
+
"loss": 5.3621,
|
| 14408 |
+
"step": 8228
|
| 14409 |
+
},
|
| 14410 |
+
{
|
| 14411 |
+
"epoch": 0.8,
|
| 14412 |
+
"grad_norm": 1.02762770652771,
|
| 14413 |
+
"learning_rate": 1.0097925150281172e-05,
|
| 14414 |
+
"loss": 5.3412,
|
| 14415 |
+
"step": 8232
|
| 14416 |
+
},
|
| 14417 |
+
{
|
| 14418 |
+
"epoch": 0.8,
|
| 14419 |
+
"grad_norm": 1.0394989252090454,
|
| 14420 |
+
"learning_rate": 1.0078534031413613e-05,
|
| 14421 |
+
"loss": 5.2555,
|
| 14422 |
+
"step": 8236
|
| 14423 |
+
},
|
| 14424 |
+
{
|
| 14425 |
+
"epoch": 0.8,
|
| 14426 |
+
"grad_norm": 1.0912150144577026,
|
| 14427 |
+
"learning_rate": 1.0059142912546055e-05,
|
| 14428 |
+
"loss": 5.3784,
|
| 14429 |
+
"step": 8240
|
| 14430 |
+
},
|
| 14431 |
+
{
|
| 14432 |
+
"epoch": 0.8,
|
| 14433 |
+
"grad_norm": 1.0681991577148438,
|
| 14434 |
+
"learning_rate": 1.0039751793678495e-05,
|
| 14435 |
+
"loss": 5.3226,
|
| 14436 |
+
"step": 8244
|
| 14437 |
+
},
|
| 14438 |
+
{
|
| 14439 |
+
"epoch": 0.8,
|
| 14440 |
+
"grad_norm": 1.0572381019592285,
|
| 14441 |
+
"learning_rate": 1.0020360674810938e-05,
|
| 14442 |
+
"loss": 5.2935,
|
| 14443 |
+
"step": 8248
|
| 14444 |
+
},
|
| 14445 |
+
{
|
| 14446 |
+
"epoch": 0.8,
|
| 14447 |
+
"grad_norm": 1.0432649850845337,
|
| 14448 |
+
"learning_rate": 1.0000969555943378e-05,
|
| 14449 |
+
"loss": 5.3087,
|
| 14450 |
+
"step": 8252
|
| 14451 |
+
},
|
| 14452 |
+
{
|
| 14453 |
+
"epoch": 0.8,
|
| 14454 |
+
"grad_norm": 1.0084688663482666,
|
| 14455 |
+
"learning_rate": 9.98157843707582e-06,
|
| 14456 |
+
"loss": 5.2778,
|
| 14457 |
+
"step": 8256
|
| 14458 |
}
|
| 14459 |
],
|
| 14460 |
"logging_steps": 4,
|
|
|
|
| 14462 |
"num_input_tokens_seen": 0,
|
| 14463 |
"num_train_epochs": 1,
|
| 14464 |
"save_steps": 1032,
|
| 14465 |
+
"total_flos": 6.953660644432282e+16,
|
| 14466 |
"train_batch_size": 8,
|
| 14467 |
"trial_name": null,
|
| 14468 |
"trial_params": null
|