Training in progress, step 1544, 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 10624
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dab47beb528b31b2a4f5c050fbcde0c1f6bb6c3618a6980eb81143fa3c739ef8
|
| 3 |
size 10624
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 26562
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6ef645c75d4c09cffd6de316dd4b1e057fadd5794ff26a91130ec4b92a5d6337
|
| 3 |
size 26562
|
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:d6b21f7a1c3a1b7249d3ca424a2568e0e334163bc2d986441a59f4393fa09ba7
|
| 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 sha256:191d3ebd8ea7d9c5929065b4868984ecccee07ef09841bd9c112aee3f2baf5e9
|
| 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,
|
|
@@ -8113,6 +8113,2708 @@
|
|
| 8113 |
"learning_rate": 1.4734990425263179e-05,
|
| 8114 |
"loss": 12.3751,
|
| 8115 |
"step": 1158
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8116 |
}
|
| 8117 |
],
|
| 8118 |
"logging_steps": 1,
|
|
@@ -8127,12 +10829,12 @@
|
|
| 8127 |
"should_evaluate": false,
|
| 8128 |
"should_log": false,
|
| 8129 |
"should_save": true,
|
| 8130 |
-
"should_training_stop":
|
| 8131 |
},
|
| 8132 |
"attributes": {}
|
| 8133 |
}
|
| 8134 |
},
|
| 8135 |
-
"total_flos":
|
| 8136 |
"train_batch_size": 4,
|
| 8137 |
"trial_name": null,
|
| 8138 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.5833018511522479,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 1544,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 8113 |
"learning_rate": 1.4734990425263179e-05,
|
| 8114 |
"loss": 12.3751,
|
| 8115 |
"step": 1158
|
| 8116 |
+
},
|
| 8117 |
+
{
|
| 8118 |
+
"epoch": 0.4378541745372119,
|
| 8119 |
+
"grad_norm": 0.03931385651230812,
|
| 8120 |
+
"learning_rate": 1.4662708440729218e-05,
|
| 8121 |
+
"loss": 12.3822,
|
| 8122 |
+
"step": 1159
|
| 8123 |
+
},
|
| 8124 |
+
{
|
| 8125 |
+
"epoch": 0.438231960710238,
|
| 8126 |
+
"grad_norm": 0.04073174670338631,
|
| 8127 |
+
"learning_rate": 1.4590573706300781e-05,
|
| 8128 |
+
"loss": 12.3726,
|
| 8129 |
+
"step": 1160
|
| 8130 |
+
},
|
| 8131 |
+
{
|
| 8132 |
+
"epoch": 0.4386097468832641,
|
| 8133 |
+
"grad_norm": 0.036465421319007874,
|
| 8134 |
+
"learning_rate": 1.4518586522562511e-05,
|
| 8135 |
+
"loss": 12.3707,
|
| 8136 |
+
"step": 1161
|
| 8137 |
+
},
|
| 8138 |
+
{
|
| 8139 |
+
"epoch": 0.4389875330562901,
|
| 8140 |
+
"grad_norm": 0.037270959466695786,
|
| 8141 |
+
"learning_rate": 1.4446747189484217e-05,
|
| 8142 |
+
"loss": 12.3651,
|
| 8143 |
+
"step": 1162
|
| 8144 |
+
},
|
| 8145 |
+
{
|
| 8146 |
+
"epoch": 0.4393653192293162,
|
| 8147 |
+
"grad_norm": 0.03188575431704521,
|
| 8148 |
+
"learning_rate": 1.4375056006419607e-05,
|
| 8149 |
+
"loss": 12.3732,
|
| 8150 |
+
"step": 1163
|
| 8151 |
+
},
|
| 8152 |
+
{
|
| 8153 |
+
"epoch": 0.4397431054023423,
|
| 8154 |
+
"grad_norm": 0.03198220580816269,
|
| 8155 |
+
"learning_rate": 1.4303513272105057e-05,
|
| 8156 |
+
"loss": 12.3689,
|
| 8157 |
+
"step": 1164
|
| 8158 |
+
},
|
| 8159 |
+
{
|
| 8160 |
+
"epoch": 0.4401208915753683,
|
| 8161 |
+
"grad_norm": 0.030993863940238953,
|
| 8162 |
+
"learning_rate": 1.4232119284658347e-05,
|
| 8163 |
+
"loss": 12.3709,
|
| 8164 |
+
"step": 1165
|
| 8165 |
+
},
|
| 8166 |
+
{
|
| 8167 |
+
"epoch": 0.4404986777483944,
|
| 8168 |
+
"grad_norm": 0.03314605727791786,
|
| 8169 |
+
"learning_rate": 1.4160874341577446e-05,
|
| 8170 |
+
"loss": 12.3757,
|
| 8171 |
+
"step": 1166
|
| 8172 |
+
},
|
| 8173 |
+
{
|
| 8174 |
+
"epoch": 0.4408764639214205,
|
| 8175 |
+
"grad_norm": 0.03950593248009682,
|
| 8176 |
+
"learning_rate": 1.408977873973924e-05,
|
| 8177 |
+
"loss": 12.3715,
|
| 8178 |
+
"step": 1167
|
| 8179 |
+
},
|
| 8180 |
+
{
|
| 8181 |
+
"epoch": 0.4412542500944465,
|
| 8182 |
+
"grad_norm": 0.0374479666352272,
|
| 8183 |
+
"learning_rate": 1.4018832775398334e-05,
|
| 8184 |
+
"loss": 12.365,
|
| 8185 |
+
"step": 1168
|
| 8186 |
+
},
|
| 8187 |
+
{
|
| 8188 |
+
"epoch": 0.4416320362674726,
|
| 8189 |
+
"grad_norm": 0.038041599094867706,
|
| 8190 |
+
"learning_rate": 1.3948036744185766e-05,
|
| 8191 |
+
"loss": 12.3724,
|
| 8192 |
+
"step": 1169
|
| 8193 |
+
},
|
| 8194 |
+
{
|
| 8195 |
+
"epoch": 0.4420098224404987,
|
| 8196 |
+
"grad_norm": 0.03593967482447624,
|
| 8197 |
+
"learning_rate": 1.3877390941107826e-05,
|
| 8198 |
+
"loss": 12.3724,
|
| 8199 |
+
"step": 1170
|
| 8200 |
+
},
|
| 8201 |
+
{
|
| 8202 |
+
"epoch": 0.4423876086135247,
|
| 8203 |
+
"grad_norm": 0.04763636738061905,
|
| 8204 |
+
"learning_rate": 1.3806895660544805e-05,
|
| 8205 |
+
"loss": 12.3657,
|
| 8206 |
+
"step": 1171
|
| 8207 |
+
},
|
| 8208 |
+
{
|
| 8209 |
+
"epoch": 0.4427653947865508,
|
| 8210 |
+
"grad_norm": 0.03423771634697914,
|
| 8211 |
+
"learning_rate": 1.3736551196249759e-05,
|
| 8212 |
+
"loss": 12.3661,
|
| 8213 |
+
"step": 1172
|
| 8214 |
+
},
|
| 8215 |
+
{
|
| 8216 |
+
"epoch": 0.4431431809595769,
|
| 8217 |
+
"grad_norm": 0.04524846374988556,
|
| 8218 |
+
"learning_rate": 1.3666357841347289e-05,
|
| 8219 |
+
"loss": 12.3692,
|
| 8220 |
+
"step": 1173
|
| 8221 |
+
},
|
| 8222 |
+
{
|
| 8223 |
+
"epoch": 0.443520967132603,
|
| 8224 |
+
"grad_norm": 0.03168905898928642,
|
| 8225 |
+
"learning_rate": 1.3596315888332344e-05,
|
| 8226 |
+
"loss": 12.3664,
|
| 8227 |
+
"step": 1174
|
| 8228 |
+
},
|
| 8229 |
+
{
|
| 8230 |
+
"epoch": 0.443898753305629,
|
| 8231 |
+
"grad_norm": 0.043227244168519974,
|
| 8232 |
+
"learning_rate": 1.3526425629068967e-05,
|
| 8233 |
+
"loss": 12.364,
|
| 8234 |
+
"step": 1175
|
| 8235 |
+
},
|
| 8236 |
+
{
|
| 8237 |
+
"epoch": 0.4442765394786551,
|
| 8238 |
+
"grad_norm": 0.038131628185510635,
|
| 8239 |
+
"learning_rate": 1.3456687354789105e-05,
|
| 8240 |
+
"loss": 12.3692,
|
| 8241 |
+
"step": 1176
|
| 8242 |
+
},
|
| 8243 |
+
{
|
| 8244 |
+
"epoch": 0.44465432565168117,
|
| 8245 |
+
"grad_norm": 0.03136202692985535,
|
| 8246 |
+
"learning_rate": 1.3387101356091376e-05,
|
| 8247 |
+
"loss": 12.3723,
|
| 8248 |
+
"step": 1177
|
| 8249 |
+
},
|
| 8250 |
+
{
|
| 8251 |
+
"epoch": 0.4450321118247072,
|
| 8252 |
+
"grad_norm": 0.03408299386501312,
|
| 8253 |
+
"learning_rate": 1.3317667922939868e-05,
|
| 8254 |
+
"loss": 12.3697,
|
| 8255 |
+
"step": 1178
|
| 8256 |
+
},
|
| 8257 |
+
{
|
| 8258 |
+
"epoch": 0.4454098979977333,
|
| 8259 |
+
"grad_norm": 0.03337300196290016,
|
| 8260 |
+
"learning_rate": 1.3248387344662944e-05,
|
| 8261 |
+
"loss": 12.3668,
|
| 8262 |
+
"step": 1179
|
| 8263 |
+
},
|
| 8264 |
+
{
|
| 8265 |
+
"epoch": 0.44578768417075937,
|
| 8266 |
+
"grad_norm": 0.03062303178012371,
|
| 8267 |
+
"learning_rate": 1.3179259909952002e-05,
|
| 8268 |
+
"loss": 12.367,
|
| 8269 |
+
"step": 1180
|
| 8270 |
+
},
|
| 8271 |
+
{
|
| 8272 |
+
"epoch": 0.4461654703437854,
|
| 8273 |
+
"grad_norm": 0.04291022941470146,
|
| 8274 |
+
"learning_rate": 1.3110285906860315e-05,
|
| 8275 |
+
"loss": 12.3636,
|
| 8276 |
+
"step": 1181
|
| 8277 |
+
},
|
| 8278 |
+
{
|
| 8279 |
+
"epoch": 0.4465432565168115,
|
| 8280 |
+
"grad_norm": 0.0346299447119236,
|
| 8281 |
+
"learning_rate": 1.3041465622801779e-05,
|
| 8282 |
+
"loss": 12.362,
|
| 8283 |
+
"step": 1182
|
| 8284 |
+
},
|
| 8285 |
+
{
|
| 8286 |
+
"epoch": 0.44692104268983757,
|
| 8287 |
+
"grad_norm": 0.03581158071756363,
|
| 8288 |
+
"learning_rate": 1.297279934454978e-05,
|
| 8289 |
+
"loss": 12.371,
|
| 8290 |
+
"step": 1183
|
| 8291 |
+
},
|
| 8292 |
+
{
|
| 8293 |
+
"epoch": 0.4472988288628636,
|
| 8294 |
+
"grad_norm": 0.04244088754057884,
|
| 8295 |
+
"learning_rate": 1.2904287358235928e-05,
|
| 8296 |
+
"loss": 12.3672,
|
| 8297 |
+
"step": 1184
|
| 8298 |
+
},
|
| 8299 |
+
{
|
| 8300 |
+
"epoch": 0.4476766150358897,
|
| 8301 |
+
"grad_norm": 0.0474376305937767,
|
| 8302 |
+
"learning_rate": 1.283592994934893e-05,
|
| 8303 |
+
"loss": 12.3618,
|
| 8304 |
+
"step": 1185
|
| 8305 |
+
},
|
| 8306 |
+
{
|
| 8307 |
+
"epoch": 0.44805440120891576,
|
| 8308 |
+
"grad_norm": 0.04155411943793297,
|
| 8309 |
+
"learning_rate": 1.2767727402733348e-05,
|
| 8310 |
+
"loss": 12.3741,
|
| 8311 |
+
"step": 1186
|
| 8312 |
+
},
|
| 8313 |
+
{
|
| 8314 |
+
"epoch": 0.4484321873819418,
|
| 8315 |
+
"grad_norm": 0.03525369614362717,
|
| 8316 |
+
"learning_rate": 1.2699680002588443e-05,
|
| 8317 |
+
"loss": 12.3639,
|
| 8318 |
+
"step": 1187
|
| 8319 |
+
},
|
| 8320 |
+
{
|
| 8321 |
+
"epoch": 0.4488099735549679,
|
| 8322 |
+
"grad_norm": 0.0417000912129879,
|
| 8323 |
+
"learning_rate": 1.2631788032466995e-05,
|
| 8324 |
+
"loss": 12.3664,
|
| 8325 |
+
"step": 1188
|
| 8326 |
+
},
|
| 8327 |
+
{
|
| 8328 |
+
"epoch": 0.44918775972799396,
|
| 8329 |
+
"grad_norm": 0.036169346421957016,
|
| 8330 |
+
"learning_rate": 1.2564051775274088e-05,
|
| 8331 |
+
"loss": 12.3697,
|
| 8332 |
+
"step": 1189
|
| 8333 |
+
},
|
| 8334 |
+
{
|
| 8335 |
+
"epoch": 0.44956554590102005,
|
| 8336 |
+
"grad_norm": 0.04191224277019501,
|
| 8337 |
+
"learning_rate": 1.2496471513265968e-05,
|
| 8338 |
+
"loss": 12.3767,
|
| 8339 |
+
"step": 1190
|
| 8340 |
+
},
|
| 8341 |
+
{
|
| 8342 |
+
"epoch": 0.4499433320740461,
|
| 8343 |
+
"grad_norm": 0.04688515514135361,
|
| 8344 |
+
"learning_rate": 1.2429047528048838e-05,
|
| 8345 |
+
"loss": 12.3704,
|
| 8346 |
+
"step": 1191
|
| 8347 |
+
},
|
| 8348 |
+
{
|
| 8349 |
+
"epoch": 0.45032111824707216,
|
| 8350 |
+
"grad_norm": 0.03686659410595894,
|
| 8351 |
+
"learning_rate": 1.2361780100577703e-05,
|
| 8352 |
+
"loss": 12.3591,
|
| 8353 |
+
"step": 1192
|
| 8354 |
+
},
|
| 8355 |
+
{
|
| 8356 |
+
"epoch": 0.45069890442009825,
|
| 8357 |
+
"grad_norm": 0.05007659271359444,
|
| 8358 |
+
"learning_rate": 1.2294669511155193e-05,
|
| 8359 |
+
"loss": 12.3692,
|
| 8360 |
+
"step": 1193
|
| 8361 |
+
},
|
| 8362 |
+
{
|
| 8363 |
+
"epoch": 0.4510766905931243,
|
| 8364 |
+
"grad_norm": 0.058174509555101395,
|
| 8365 |
+
"learning_rate": 1.2227716039430404e-05,
|
| 8366 |
+
"loss": 12.3632,
|
| 8367 |
+
"step": 1194
|
| 8368 |
+
},
|
| 8369 |
+
{
|
| 8370 |
+
"epoch": 0.45145447676615036,
|
| 8371 |
+
"grad_norm": 0.056341782212257385,
|
| 8372 |
+
"learning_rate": 1.216091996439771e-05,
|
| 8373 |
+
"loss": 12.373,
|
| 8374 |
+
"step": 1195
|
| 8375 |
+
},
|
| 8376 |
+
{
|
| 8377 |
+
"epoch": 0.45183226293917644,
|
| 8378 |
+
"grad_norm": 0.05682500824332237,
|
| 8379 |
+
"learning_rate": 1.2094281564395627e-05,
|
| 8380 |
+
"loss": 12.3637,
|
| 8381 |
+
"step": 1196
|
| 8382 |
+
},
|
| 8383 |
+
{
|
| 8384 |
+
"epoch": 0.4522100491122025,
|
| 8385 |
+
"grad_norm": 0.06348198652267456,
|
| 8386 |
+
"learning_rate": 1.202780111710563e-05,
|
| 8387 |
+
"loss": 12.3755,
|
| 8388 |
+
"step": 1197
|
| 8389 |
+
},
|
| 8390 |
+
{
|
| 8391 |
+
"epoch": 0.45258783528522856,
|
| 8392 |
+
"grad_norm": 0.06290506571531296,
|
| 8393 |
+
"learning_rate": 1.1961478899551015e-05,
|
| 8394 |
+
"loss": 12.381,
|
| 8395 |
+
"step": 1198
|
| 8396 |
+
},
|
| 8397 |
+
{
|
| 8398 |
+
"epoch": 0.45296562145825464,
|
| 8399 |
+
"grad_norm": 0.09010813385248184,
|
| 8400 |
+
"learning_rate": 1.189531518809573e-05,
|
| 8401 |
+
"loss": 12.3777,
|
| 8402 |
+
"step": 1199
|
| 8403 |
+
},
|
| 8404 |
+
{
|
| 8405 |
+
"epoch": 0.45334340763128067,
|
| 8406 |
+
"grad_norm": 0.15625080466270447,
|
| 8407 |
+
"learning_rate": 1.1829310258443238e-05,
|
| 8408 |
+
"loss": 12.3885,
|
| 8409 |
+
"step": 1200
|
| 8410 |
+
},
|
| 8411 |
+
{
|
| 8412 |
+
"epoch": 0.45372119380430675,
|
| 8413 |
+
"grad_norm": 0.047231078147888184,
|
| 8414 |
+
"learning_rate": 1.1763464385635353e-05,
|
| 8415 |
+
"loss": 12.381,
|
| 8416 |
+
"step": 1201
|
| 8417 |
+
},
|
| 8418 |
+
{
|
| 8419 |
+
"epoch": 0.45409897997733284,
|
| 8420 |
+
"grad_norm": 0.03812883049249649,
|
| 8421 |
+
"learning_rate": 1.1697777844051105e-05,
|
| 8422 |
+
"loss": 12.3728,
|
| 8423 |
+
"step": 1202
|
| 8424 |
+
},
|
| 8425 |
+
{
|
| 8426 |
+
"epoch": 0.4544767661503589,
|
| 8427 |
+
"grad_norm": 0.03945963829755783,
|
| 8428 |
+
"learning_rate": 1.163225090740559e-05,
|
| 8429 |
+
"loss": 12.3772,
|
| 8430 |
+
"step": 1203
|
| 8431 |
+
},
|
| 8432 |
+
{
|
| 8433 |
+
"epoch": 0.45485455232338495,
|
| 8434 |
+
"grad_norm": 0.04512109234929085,
|
| 8435 |
+
"learning_rate": 1.1566883848748832e-05,
|
| 8436 |
+
"loss": 12.3682,
|
| 8437 |
+
"step": 1204
|
| 8438 |
+
},
|
| 8439 |
+
{
|
| 8440 |
+
"epoch": 0.45523233849641104,
|
| 8441 |
+
"grad_norm": 0.03057880327105522,
|
| 8442 |
+
"learning_rate": 1.1501676940464645e-05,
|
| 8443 |
+
"loss": 12.3762,
|
| 8444 |
+
"step": 1205
|
| 8445 |
+
},
|
| 8446 |
+
{
|
| 8447 |
+
"epoch": 0.4556101246694371,
|
| 8448 |
+
"grad_norm": 0.038191623985767365,
|
| 8449 |
+
"learning_rate": 1.1436630454269504e-05,
|
| 8450 |
+
"loss": 12.3757,
|
| 8451 |
+
"step": 1206
|
| 8452 |
+
},
|
| 8453 |
+
{
|
| 8454 |
+
"epoch": 0.45598791084246315,
|
| 8455 |
+
"grad_norm": 0.04440156742930412,
|
| 8456 |
+
"learning_rate": 1.1371744661211391e-05,
|
| 8457 |
+
"loss": 12.3754,
|
| 8458 |
+
"step": 1207
|
| 8459 |
+
},
|
| 8460 |
+
{
|
| 8461 |
+
"epoch": 0.45636569701548924,
|
| 8462 |
+
"grad_norm": 0.03836330398917198,
|
| 8463 |
+
"learning_rate": 1.1307019831668709e-05,
|
| 8464 |
+
"loss": 12.3745,
|
| 8465 |
+
"step": 1208
|
| 8466 |
+
},
|
| 8467 |
+
{
|
| 8468 |
+
"epoch": 0.4567434831885153,
|
| 8469 |
+
"grad_norm": 0.036225590854883194,
|
| 8470 |
+
"learning_rate": 1.1242456235349096e-05,
|
| 8471 |
+
"loss": 12.3757,
|
| 8472 |
+
"step": 1209
|
| 8473 |
+
},
|
| 8474 |
+
{
|
| 8475 |
+
"epoch": 0.45712126936154135,
|
| 8476 |
+
"grad_norm": 0.032095544040203094,
|
| 8477 |
+
"learning_rate": 1.1178054141288358e-05,
|
| 8478 |
+
"loss": 12.3712,
|
| 8479 |
+
"step": 1210
|
| 8480 |
+
},
|
| 8481 |
+
{
|
| 8482 |
+
"epoch": 0.45749905553456743,
|
| 8483 |
+
"grad_norm": 0.037354450672864914,
|
| 8484 |
+
"learning_rate": 1.111381381784931e-05,
|
| 8485 |
+
"loss": 12.3723,
|
| 8486 |
+
"step": 1211
|
| 8487 |
+
},
|
| 8488 |
+
{
|
| 8489 |
+
"epoch": 0.4578768417075935,
|
| 8490 |
+
"grad_norm": 0.03643963858485222,
|
| 8491 |
+
"learning_rate": 1.1049735532720673e-05,
|
| 8492 |
+
"loss": 12.3675,
|
| 8493 |
+
"step": 1212
|
| 8494 |
+
},
|
| 8495 |
+
{
|
| 8496 |
+
"epoch": 0.45825462788061955,
|
| 8497 |
+
"grad_norm": 0.04525665566325188,
|
| 8498 |
+
"learning_rate": 1.0985819552915955e-05,
|
| 8499 |
+
"loss": 12.3668,
|
| 8500 |
+
"step": 1213
|
| 8501 |
+
},
|
| 8502 |
+
{
|
| 8503 |
+
"epoch": 0.45863241405364563,
|
| 8504 |
+
"grad_norm": 0.03201952949166298,
|
| 8505 |
+
"learning_rate": 1.092206614477234e-05,
|
| 8506 |
+
"loss": 12.37,
|
| 8507 |
+
"step": 1214
|
| 8508 |
+
},
|
| 8509 |
+
{
|
| 8510 |
+
"epoch": 0.4590102002266717,
|
| 8511 |
+
"grad_norm": 0.04047270119190216,
|
| 8512 |
+
"learning_rate": 1.0858475573949589e-05,
|
| 8513 |
+
"loss": 12.3769,
|
| 8514 |
+
"step": 1215
|
| 8515 |
+
},
|
| 8516 |
+
{
|
| 8517 |
+
"epoch": 0.45938798639969775,
|
| 8518 |
+
"grad_norm": 0.040772419422864914,
|
| 8519 |
+
"learning_rate": 1.0795048105428907e-05,
|
| 8520 |
+
"loss": 12.3792,
|
| 8521 |
+
"step": 1216
|
| 8522 |
+
},
|
| 8523 |
+
{
|
| 8524 |
+
"epoch": 0.45976577257272383,
|
| 8525 |
+
"grad_norm": 0.02995879203081131,
|
| 8526 |
+
"learning_rate": 1.0731784003511852e-05,
|
| 8527 |
+
"loss": 12.3721,
|
| 8528 |
+
"step": 1217
|
| 8529 |
+
},
|
| 8530 |
+
{
|
| 8531 |
+
"epoch": 0.4601435587457499,
|
| 8532 |
+
"grad_norm": 0.03294239193201065,
|
| 8533 |
+
"learning_rate": 1.0668683531819247e-05,
|
| 8534 |
+
"loss": 12.3715,
|
| 8535 |
+
"step": 1218
|
| 8536 |
+
},
|
| 8537 |
+
{
|
| 8538 |
+
"epoch": 0.460521344918776,
|
| 8539 |
+
"grad_norm": 0.03151008486747742,
|
| 8540 |
+
"learning_rate": 1.0605746953290057e-05,
|
| 8541 |
+
"loss": 12.3708,
|
| 8542 |
+
"step": 1219
|
| 8543 |
+
},
|
| 8544 |
+
{
|
| 8545 |
+
"epoch": 0.46089913109180203,
|
| 8546 |
+
"grad_norm": 0.03497641906142235,
|
| 8547 |
+
"learning_rate": 1.0542974530180327e-05,
|
| 8548 |
+
"loss": 12.3742,
|
| 8549 |
+
"step": 1220
|
| 8550 |
+
},
|
| 8551 |
+
{
|
| 8552 |
+
"epoch": 0.4612769172648281,
|
| 8553 |
+
"grad_norm": 0.04114446043968201,
|
| 8554 |
+
"learning_rate": 1.0480366524062042e-05,
|
| 8555 |
+
"loss": 12.3713,
|
| 8556 |
+
"step": 1221
|
| 8557 |
+
},
|
| 8558 |
+
{
|
| 8559 |
+
"epoch": 0.4616547034378542,
|
| 8560 |
+
"grad_norm": 0.03543113172054291,
|
| 8561 |
+
"learning_rate": 1.0417923195822076e-05,
|
| 8562 |
+
"loss": 12.3751,
|
| 8563 |
+
"step": 1222
|
| 8564 |
+
},
|
| 8565 |
+
{
|
| 8566 |
+
"epoch": 0.4620324896108802,
|
| 8567 |
+
"grad_norm": 0.041865747421979904,
|
| 8568 |
+
"learning_rate": 1.0355644805661102e-05,
|
| 8569 |
+
"loss": 12.3751,
|
| 8570 |
+
"step": 1223
|
| 8571 |
+
},
|
| 8572 |
+
{
|
| 8573 |
+
"epoch": 0.4624102757839063,
|
| 8574 |
+
"grad_norm": 0.04181693121790886,
|
| 8575 |
+
"learning_rate": 1.0293531613092478e-05,
|
| 8576 |
+
"loss": 12.3672,
|
| 8577 |
+
"step": 1224
|
| 8578 |
+
},
|
| 8579 |
+
{
|
| 8580 |
+
"epoch": 0.4627880619569324,
|
| 8581 |
+
"grad_norm": 0.041295990347862244,
|
| 8582 |
+
"learning_rate": 1.0231583876941192e-05,
|
| 8583 |
+
"loss": 12.3661,
|
| 8584 |
+
"step": 1225
|
| 8585 |
+
},
|
| 8586 |
+
{
|
| 8587 |
+
"epoch": 0.4631658481299584,
|
| 8588 |
+
"grad_norm": 0.05063636973500252,
|
| 8589 |
+
"learning_rate": 1.0169801855342786e-05,
|
| 8590 |
+
"loss": 12.3639,
|
| 8591 |
+
"step": 1226
|
| 8592 |
+
},
|
| 8593 |
+
{
|
| 8594 |
+
"epoch": 0.4635436343029845,
|
| 8595 |
+
"grad_norm": 0.051273223012685776,
|
| 8596 |
+
"learning_rate": 1.0108185805742264e-05,
|
| 8597 |
+
"loss": 12.3657,
|
| 8598 |
+
"step": 1227
|
| 8599 |
+
},
|
| 8600 |
+
{
|
| 8601 |
+
"epoch": 0.4639214204760106,
|
| 8602 |
+
"grad_norm": 0.03442752733826637,
|
| 8603 |
+
"learning_rate": 1.0046735984893023e-05,
|
| 8604 |
+
"loss": 12.3626,
|
| 8605 |
+
"step": 1228
|
| 8606 |
+
},
|
| 8607 |
+
{
|
| 8608 |
+
"epoch": 0.4642992066490366,
|
| 8609 |
+
"grad_norm": 0.055842552334070206,
|
| 8610 |
+
"learning_rate": 9.985452648855803e-06,
|
| 8611 |
+
"loss": 12.3613,
|
| 8612 |
+
"step": 1229
|
| 8613 |
+
},
|
| 8614 |
+
{
|
| 8615 |
+
"epoch": 0.4646769928220627,
|
| 8616 |
+
"grad_norm": 0.03904302045702934,
|
| 8617 |
+
"learning_rate": 9.924336052997579e-06,
|
| 8618 |
+
"loss": 12.3667,
|
| 8619 |
+
"step": 1230
|
| 8620 |
+
},
|
| 8621 |
+
{
|
| 8622 |
+
"epoch": 0.4650547789950888,
|
| 8623 |
+
"grad_norm": 0.039607588201761246,
|
| 8624 |
+
"learning_rate": 9.863386451990536e-06,
|
| 8625 |
+
"loss": 12.3627,
|
| 8626 |
+
"step": 1231
|
| 8627 |
+
},
|
| 8628 |
+
{
|
| 8629 |
+
"epoch": 0.4654325651681148,
|
| 8630 |
+
"grad_norm": 0.03496779128909111,
|
| 8631 |
+
"learning_rate": 9.802604099810997e-06,
|
| 8632 |
+
"loss": 12.3662,
|
| 8633 |
+
"step": 1232
|
| 8634 |
+
},
|
| 8635 |
+
{
|
| 8636 |
+
"epoch": 0.4658103513411409,
|
| 8637 |
+
"grad_norm": 0.03711122274398804,
|
| 8638 |
+
"learning_rate": 9.741989249738348e-06,
|
| 8639 |
+
"loss": 12.3758,
|
| 8640 |
+
"step": 1233
|
| 8641 |
+
},
|
| 8642 |
+
{
|
| 8643 |
+
"epoch": 0.466188137514167,
|
| 8644 |
+
"grad_norm": 0.03672249615192413,
|
| 8645 |
+
"learning_rate": 9.681542154354011e-06,
|
| 8646 |
+
"loss": 12.3726,
|
| 8647 |
+
"step": 1234
|
| 8648 |
+
},
|
| 8649 |
+
{
|
| 8650 |
+
"epoch": 0.4665659236871931,
|
| 8651 |
+
"grad_norm": 0.03551850467920303,
|
| 8652 |
+
"learning_rate": 9.621263065540365e-06,
|
| 8653 |
+
"loss": 12.3642,
|
| 8654 |
+
"step": 1235
|
| 8655 |
+
},
|
| 8656 |
+
{
|
| 8657 |
+
"epoch": 0.4669437098602191,
|
| 8658 |
+
"grad_norm": 0.039268869906663895,
|
| 8659 |
+
"learning_rate": 9.561152234479714e-06,
|
| 8660 |
+
"loss": 12.3699,
|
| 8661 |
+
"step": 1236
|
| 8662 |
+
},
|
| 8663 |
+
{
|
| 8664 |
+
"epoch": 0.4673214960332452,
|
| 8665 |
+
"grad_norm": 0.04356931895017624,
|
| 8666 |
+
"learning_rate": 9.501209911653219e-06,
|
| 8667 |
+
"loss": 12.3671,
|
| 8668 |
+
"step": 1237
|
| 8669 |
+
},
|
| 8670 |
+
{
|
| 8671 |
+
"epoch": 0.46769928220627127,
|
| 8672 |
+
"grad_norm": 0.0456683486700058,
|
| 8673 |
+
"learning_rate": 9.441436346839894e-06,
|
| 8674 |
+
"loss": 12.3664,
|
| 8675 |
+
"step": 1238
|
| 8676 |
+
},
|
| 8677 |
+
{
|
| 8678 |
+
"epoch": 0.4680770683792973,
|
| 8679 |
+
"grad_norm": 0.031490568071603775,
|
| 8680 |
+
"learning_rate": 9.381831789115515e-06,
|
| 8681 |
+
"loss": 12.3667,
|
| 8682 |
+
"step": 1239
|
| 8683 |
+
},
|
| 8684 |
+
{
|
| 8685 |
+
"epoch": 0.4684548545523234,
|
| 8686 |
+
"grad_norm": 0.0355166494846344,
|
| 8687 |
+
"learning_rate": 9.322396486851626e-06,
|
| 8688 |
+
"loss": 12.3749,
|
| 8689 |
+
"step": 1240
|
| 8690 |
+
},
|
| 8691 |
+
{
|
| 8692 |
+
"epoch": 0.46883264072534947,
|
| 8693 |
+
"grad_norm": 0.04162101075053215,
|
| 8694 |
+
"learning_rate": 9.263130687714478e-06,
|
| 8695 |
+
"loss": 12.3661,
|
| 8696 |
+
"step": 1241
|
| 8697 |
+
},
|
| 8698 |
+
{
|
| 8699 |
+
"epoch": 0.4692104268983755,
|
| 8700 |
+
"grad_norm": 0.04071405902504921,
|
| 8701 |
+
"learning_rate": 9.204034638664006e-06,
|
| 8702 |
+
"loss": 12.3747,
|
| 8703 |
+
"step": 1242
|
| 8704 |
+
},
|
| 8705 |
+
{
|
| 8706 |
+
"epoch": 0.4695882130714016,
|
| 8707 |
+
"grad_norm": 0.0722724124789238,
|
| 8708 |
+
"learning_rate": 9.145108585952788e-06,
|
| 8709 |
+
"loss": 12.3785,
|
| 8710 |
+
"step": 1243
|
| 8711 |
+
},
|
| 8712 |
+
{
|
| 8713 |
+
"epoch": 0.46996599924442767,
|
| 8714 |
+
"grad_norm": 0.050802621990442276,
|
| 8715 |
+
"learning_rate": 9.086352775125046e-06,
|
| 8716 |
+
"loss": 12.3668,
|
| 8717 |
+
"step": 1244
|
| 8718 |
+
},
|
| 8719 |
+
{
|
| 8720 |
+
"epoch": 0.4703437854174537,
|
| 8721 |
+
"grad_norm": 0.055558983236551285,
|
| 8722 |
+
"learning_rate": 9.027767451015595e-06,
|
| 8723 |
+
"loss": 12.3663,
|
| 8724 |
+
"step": 1245
|
| 8725 |
+
},
|
| 8726 |
+
{
|
| 8727 |
+
"epoch": 0.4707215715904798,
|
| 8728 |
+
"grad_norm": 0.05985091254115105,
|
| 8729 |
+
"learning_rate": 8.969352857748842e-06,
|
| 8730 |
+
"loss": 12.381,
|
| 8731 |
+
"step": 1246
|
| 8732 |
+
},
|
| 8733 |
+
{
|
| 8734 |
+
"epoch": 0.47109935776350587,
|
| 8735 |
+
"grad_norm": 0.05538693815469742,
|
| 8736 |
+
"learning_rate": 8.911109238737747e-06,
|
| 8737 |
+
"loss": 12.3641,
|
| 8738 |
+
"step": 1247
|
| 8739 |
+
},
|
| 8740 |
+
{
|
| 8741 |
+
"epoch": 0.47147714393653195,
|
| 8742 |
+
"grad_norm": 0.07532553374767303,
|
| 8743 |
+
"learning_rate": 8.853036836682838e-06,
|
| 8744 |
+
"loss": 12.367,
|
| 8745 |
+
"step": 1248
|
| 8746 |
+
},
|
| 8747 |
+
{
|
| 8748 |
+
"epoch": 0.471854930109558,
|
| 8749 |
+
"grad_norm": 0.08992443978786469,
|
| 8750 |
+
"learning_rate": 8.795135893571166e-06,
|
| 8751 |
+
"loss": 12.3734,
|
| 8752 |
+
"step": 1249
|
| 8753 |
+
},
|
| 8754 |
+
{
|
| 8755 |
+
"epoch": 0.47223271628258406,
|
| 8756 |
+
"grad_norm": 0.12376437336206436,
|
| 8757 |
+
"learning_rate": 8.737406650675333e-06,
|
| 8758 |
+
"loss": 12.3813,
|
| 8759 |
+
"step": 1250
|
| 8760 |
+
},
|
| 8761 |
+
{
|
| 8762 |
+
"epoch": 0.47261050245561015,
|
| 8763 |
+
"grad_norm": 0.04168854281306267,
|
| 8764 |
+
"learning_rate": 8.679849348552455e-06,
|
| 8765 |
+
"loss": 12.3736,
|
| 8766 |
+
"step": 1251
|
| 8767 |
+
},
|
| 8768 |
+
{
|
| 8769 |
+
"epoch": 0.4729882886286362,
|
| 8770 |
+
"grad_norm": 0.03898398578166962,
|
| 8771 |
+
"learning_rate": 8.622464227043181e-06,
|
| 8772 |
+
"loss": 12.3767,
|
| 8773 |
+
"step": 1252
|
| 8774 |
+
},
|
| 8775 |
+
{
|
| 8776 |
+
"epoch": 0.47336607480166226,
|
| 8777 |
+
"grad_norm": 0.04223041981458664,
|
| 8778 |
+
"learning_rate": 8.565251525270684e-06,
|
| 8779 |
+
"loss": 12.3719,
|
| 8780 |
+
"step": 1253
|
| 8781 |
+
},
|
| 8782 |
+
{
|
| 8783 |
+
"epoch": 0.47374386097468835,
|
| 8784 |
+
"grad_norm": 0.03358473256230354,
|
| 8785 |
+
"learning_rate": 8.508211481639655e-06,
|
| 8786 |
+
"loss": 12.3685,
|
| 8787 |
+
"step": 1254
|
| 8788 |
+
},
|
| 8789 |
+
{
|
| 8790 |
+
"epoch": 0.4741216471477144,
|
| 8791 |
+
"grad_norm": 0.03558828681707382,
|
| 8792 |
+
"learning_rate": 8.451344333835338e-06,
|
| 8793 |
+
"loss": 12.3719,
|
| 8794 |
+
"step": 1255
|
| 8795 |
+
},
|
| 8796 |
+
{
|
| 8797 |
+
"epoch": 0.47449943332074046,
|
| 8798 |
+
"grad_norm": 0.03825223818421364,
|
| 8799 |
+
"learning_rate": 8.3946503188225e-06,
|
| 8800 |
+
"loss": 12.3784,
|
| 8801 |
+
"step": 1256
|
| 8802 |
+
},
|
| 8803 |
+
{
|
| 8804 |
+
"epoch": 0.47487721949376654,
|
| 8805 |
+
"grad_norm": 0.03340928256511688,
|
| 8806 |
+
"learning_rate": 8.338129672844491e-06,
|
| 8807 |
+
"loss": 12.3721,
|
| 8808 |
+
"step": 1257
|
| 8809 |
+
},
|
| 8810 |
+
{
|
| 8811 |
+
"epoch": 0.4752550056667926,
|
| 8812 |
+
"grad_norm": 0.04322151467204094,
|
| 8813 |
+
"learning_rate": 8.28178263142222e-06,
|
| 8814 |
+
"loss": 12.3694,
|
| 8815 |
+
"step": 1258
|
| 8816 |
+
},
|
| 8817 |
+
{
|
| 8818 |
+
"epoch": 0.47563279183981866,
|
| 8819 |
+
"grad_norm": 0.03472879156470299,
|
| 8820 |
+
"learning_rate": 8.225609429353187e-06,
|
| 8821 |
+
"loss": 12.3715,
|
| 8822 |
+
"step": 1259
|
| 8823 |
+
},
|
| 8824 |
+
{
|
| 8825 |
+
"epoch": 0.47601057801284474,
|
| 8826 |
+
"grad_norm": 0.04734859988093376,
|
| 8827 |
+
"learning_rate": 8.1696103007105e-06,
|
| 8828 |
+
"loss": 12.3765,
|
| 8829 |
+
"step": 1260
|
| 8830 |
+
},
|
| 8831 |
+
{
|
| 8832 |
+
"epoch": 0.47638836418587077,
|
| 8833 |
+
"grad_norm": 0.042171910405159,
|
| 8834 |
+
"learning_rate": 8.113785478841924e-06,
|
| 8835 |
+
"loss": 12.3667,
|
| 8836 |
+
"step": 1261
|
| 8837 |
+
},
|
| 8838 |
+
{
|
| 8839 |
+
"epoch": 0.47676615035889686,
|
| 8840 |
+
"grad_norm": 0.03718302771449089,
|
| 8841 |
+
"learning_rate": 8.058135196368877e-06,
|
| 8842 |
+
"loss": 12.3759,
|
| 8843 |
+
"step": 1262
|
| 8844 |
+
},
|
| 8845 |
+
{
|
| 8846 |
+
"epoch": 0.47714393653192294,
|
| 8847 |
+
"grad_norm": 0.04056549444794655,
|
| 8848 |
+
"learning_rate": 8.002659685185464e-06,
|
| 8849 |
+
"loss": 12.3742,
|
| 8850 |
+
"step": 1263
|
| 8851 |
+
},
|
| 8852 |
+
{
|
| 8853 |
+
"epoch": 0.477521722704949,
|
| 8854 |
+
"grad_norm": 0.042442917823791504,
|
| 8855 |
+
"learning_rate": 7.947359176457541e-06,
|
| 8856 |
+
"loss": 12.3682,
|
| 8857 |
+
"step": 1264
|
| 8858 |
+
},
|
| 8859 |
+
{
|
| 8860 |
+
"epoch": 0.47789950887797505,
|
| 8861 |
+
"grad_norm": 0.041739899665117264,
|
| 8862 |
+
"learning_rate": 7.89223390062172e-06,
|
| 8863 |
+
"loss": 12.3742,
|
| 8864 |
+
"step": 1265
|
| 8865 |
+
},
|
| 8866 |
+
{
|
| 8867 |
+
"epoch": 0.47827729505100114,
|
| 8868 |
+
"grad_norm": 0.033437564969062805,
|
| 8869 |
+
"learning_rate": 7.837284087384411e-06,
|
| 8870 |
+
"loss": 12.3677,
|
| 8871 |
+
"step": 1266
|
| 8872 |
+
},
|
| 8873 |
+
{
|
| 8874 |
+
"epoch": 0.4786550812240272,
|
| 8875 |
+
"grad_norm": 0.03466346487402916,
|
| 8876 |
+
"learning_rate": 7.782509965720891e-06,
|
| 8877 |
+
"loss": 12.3674,
|
| 8878 |
+
"step": 1267
|
| 8879 |
+
},
|
| 8880 |
+
{
|
| 8881 |
+
"epoch": 0.47903286739705325,
|
| 8882 |
+
"grad_norm": 0.03019503317773342,
|
| 8883 |
+
"learning_rate": 7.727911763874313e-06,
|
| 8884 |
+
"loss": 12.3779,
|
| 8885 |
+
"step": 1268
|
| 8886 |
+
},
|
| 8887 |
+
{
|
| 8888 |
+
"epoch": 0.47941065357007934,
|
| 8889 |
+
"grad_norm": 0.03774089366197586,
|
| 8890 |
+
"learning_rate": 7.673489709354798e-06,
|
| 8891 |
+
"loss": 12.3672,
|
| 8892 |
+
"step": 1269
|
| 8893 |
+
},
|
| 8894 |
+
{
|
| 8895 |
+
"epoch": 0.4797884397431054,
|
| 8896 |
+
"grad_norm": 0.03035971336066723,
|
| 8897 |
+
"learning_rate": 7.619244028938433e-06,
|
| 8898 |
+
"loss": 12.3657,
|
| 8899 |
+
"step": 1270
|
| 8900 |
+
},
|
| 8901 |
+
{
|
| 8902 |
+
"epoch": 0.48016622591613145,
|
| 8903 |
+
"grad_norm": 0.041345734149217606,
|
| 8904 |
+
"learning_rate": 7.565174948666381e-06,
|
| 8905 |
+
"loss": 12.3704,
|
| 8906 |
+
"step": 1271
|
| 8907 |
+
},
|
| 8908 |
+
{
|
| 8909 |
+
"epoch": 0.48054401208915754,
|
| 8910 |
+
"grad_norm": 0.034325577318668365,
|
| 8911 |
+
"learning_rate": 7.511282693843908e-06,
|
| 8912 |
+
"loss": 12.3673,
|
| 8913 |
+
"step": 1272
|
| 8914 |
+
},
|
| 8915 |
+
{
|
| 8916 |
+
"epoch": 0.4809217982621836,
|
| 8917 |
+
"grad_norm": 0.031203050166368484,
|
| 8918 |
+
"learning_rate": 7.457567489039441e-06,
|
| 8919 |
+
"loss": 12.3683,
|
| 8920 |
+
"step": 1273
|
| 8921 |
+
},
|
| 8922 |
+
{
|
| 8923 |
+
"epoch": 0.48129958443520965,
|
| 8924 |
+
"grad_norm": 0.0467066615819931,
|
| 8925 |
+
"learning_rate": 7.404029558083653e-06,
|
| 8926 |
+
"loss": 12.3686,
|
| 8927 |
+
"step": 1274
|
| 8928 |
+
},
|
| 8929 |
+
{
|
| 8930 |
+
"epoch": 0.48167737060823573,
|
| 8931 |
+
"grad_norm": 0.039916858077049255,
|
| 8932 |
+
"learning_rate": 7.35066912406851e-06,
|
| 8933 |
+
"loss": 12.3674,
|
| 8934 |
+
"step": 1275
|
| 8935 |
+
},
|
| 8936 |
+
{
|
| 8937 |
+
"epoch": 0.4820551567812618,
|
| 8938 |
+
"grad_norm": 0.035765450447797775,
|
| 8939 |
+
"learning_rate": 7.2974864093463545e-06,
|
| 8940 |
+
"loss": 12.3702,
|
| 8941 |
+
"step": 1276
|
| 8942 |
+
},
|
| 8943 |
+
{
|
| 8944 |
+
"epoch": 0.48243294295428785,
|
| 8945 |
+
"grad_norm": 0.03454265743494034,
|
| 8946 |
+
"learning_rate": 7.2444816355289715e-06,
|
| 8947 |
+
"loss": 12.3753,
|
| 8948 |
+
"step": 1277
|
| 8949 |
+
},
|
| 8950 |
+
{
|
| 8951 |
+
"epoch": 0.48281072912731393,
|
| 8952 |
+
"grad_norm": 0.03991926088929176,
|
| 8953 |
+
"learning_rate": 7.191655023486682e-06,
|
| 8954 |
+
"loss": 12.3691,
|
| 8955 |
+
"step": 1278
|
| 8956 |
+
},
|
| 8957 |
+
{
|
| 8958 |
+
"epoch": 0.48318851530034,
|
| 8959 |
+
"grad_norm": 0.047338444739580154,
|
| 8960 |
+
"learning_rate": 7.139006793347386e-06,
|
| 8961 |
+
"loss": 12.3618,
|
| 8962 |
+
"step": 1279
|
| 8963 |
+
},
|
| 8964 |
+
{
|
| 8965 |
+
"epoch": 0.4835663014733661,
|
| 8966 |
+
"grad_norm": 0.035819366574287415,
|
| 8967 |
+
"learning_rate": 7.086537164495688e-06,
|
| 8968 |
+
"loss": 12.368,
|
| 8969 |
+
"step": 1280
|
| 8970 |
+
},
|
| 8971 |
+
{
|
| 8972 |
+
"epoch": 0.48394408764639213,
|
| 8973 |
+
"grad_norm": 0.03620653226971626,
|
| 8974 |
+
"learning_rate": 7.034246355571944e-06,
|
| 8975 |
+
"loss": 12.3732,
|
| 8976 |
+
"step": 1281
|
| 8977 |
+
},
|
| 8978 |
+
{
|
| 8979 |
+
"epoch": 0.4843218738194182,
|
| 8980 |
+
"grad_norm": 0.037103258073329926,
|
| 8981 |
+
"learning_rate": 6.982134584471395e-06,
|
| 8982 |
+
"loss": 12.3646,
|
| 8983 |
+
"step": 1282
|
| 8984 |
+
},
|
| 8985 |
+
{
|
| 8986 |
+
"epoch": 0.4846996599924443,
|
| 8987 |
+
"grad_norm": 0.055807821452617645,
|
| 8988 |
+
"learning_rate": 6.9302020683432055e-06,
|
| 8989 |
+
"loss": 12.3622,
|
| 8990 |
+
"step": 1283
|
| 8991 |
+
},
|
| 8992 |
+
{
|
| 8993 |
+
"epoch": 0.4850774461654703,
|
| 8994 |
+
"grad_norm": 0.038796085864305496,
|
| 8995 |
+
"learning_rate": 6.878449023589601e-06,
|
| 8996 |
+
"loss": 12.3686,
|
| 8997 |
+
"step": 1284
|
| 8998 |
+
},
|
| 8999 |
+
{
|
| 9000 |
+
"epoch": 0.4854552323384964,
|
| 9001 |
+
"grad_norm": 0.04036309942603111,
|
| 9002 |
+
"learning_rate": 6.826875665864957e-06,
|
| 9003 |
+
"loss": 12.3711,
|
| 9004 |
+
"step": 1285
|
| 9005 |
+
},
|
| 9006 |
+
{
|
| 9007 |
+
"epoch": 0.4858330185115225,
|
| 9008 |
+
"grad_norm": 0.03805702179670334,
|
| 9009 |
+
"learning_rate": 6.775482210074885e-06,
|
| 9010 |
+
"loss": 12.3679,
|
| 9011 |
+
"step": 1286
|
| 9012 |
+
},
|
| 9013 |
+
{
|
| 9014 |
+
"epoch": 0.4862108046845485,
|
| 9015 |
+
"grad_norm": 0.03731246292591095,
|
| 9016 |
+
"learning_rate": 6.72426887037535e-06,
|
| 9017 |
+
"loss": 12.3666,
|
| 9018 |
+
"step": 1287
|
| 9019 |
+
},
|
| 9020 |
+
{
|
| 9021 |
+
"epoch": 0.4865885908575746,
|
| 9022 |
+
"grad_norm": 0.036057423800230026,
|
| 9023 |
+
"learning_rate": 6.673235860171778e-06,
|
| 9024 |
+
"loss": 12.3575,
|
| 9025 |
+
"step": 1288
|
| 9026 |
+
},
|
| 9027 |
+
{
|
| 9028 |
+
"epoch": 0.4869663770306007,
|
| 9029 |
+
"grad_norm": 0.04278253763914108,
|
| 9030 |
+
"learning_rate": 6.622383392118154e-06,
|
| 9031 |
+
"loss": 12.3602,
|
| 9032 |
+
"step": 1289
|
| 9033 |
+
},
|
| 9034 |
+
{
|
| 9035 |
+
"epoch": 0.4873441632036267,
|
| 9036 |
+
"grad_norm": 0.03653295338153839,
|
| 9037 |
+
"learning_rate": 6.57171167811616e-06,
|
| 9038 |
+
"loss": 12.3671,
|
| 9039 |
+
"step": 1290
|
| 9040 |
+
},
|
| 9041 |
+
{
|
| 9042 |
+
"epoch": 0.4877219493766528,
|
| 9043 |
+
"grad_norm": 0.04199488088488579,
|
| 9044 |
+
"learning_rate": 6.521220929314276e-06,
|
| 9045 |
+
"loss": 12.3753,
|
| 9046 |
+
"step": 1291
|
| 9047 |
+
},
|
| 9048 |
+
{
|
| 9049 |
+
"epoch": 0.4880997355496789,
|
| 9050 |
+
"grad_norm": 0.041755374521017075,
|
| 9051 |
+
"learning_rate": 6.470911356106885e-06,
|
| 9052 |
+
"loss": 12.3735,
|
| 9053 |
+
"step": 1292
|
| 9054 |
+
},
|
| 9055 |
+
{
|
| 9056 |
+
"epoch": 0.4884775217227049,
|
| 9057 |
+
"grad_norm": 0.04074932634830475,
|
| 9058 |
+
"learning_rate": 6.420783168133432e-06,
|
| 9059 |
+
"loss": 12.3671,
|
| 9060 |
+
"step": 1293
|
| 9061 |
+
},
|
| 9062 |
+
{
|
| 9063 |
+
"epoch": 0.488855307895731,
|
| 9064 |
+
"grad_norm": 0.04746726527810097,
|
| 9065 |
+
"learning_rate": 6.370836574277528e-06,
|
| 9066 |
+
"loss": 12.3705,
|
| 9067 |
+
"step": 1294
|
| 9068 |
+
},
|
| 9069 |
+
{
|
| 9070 |
+
"epoch": 0.4892330940687571,
|
| 9071 |
+
"grad_norm": 0.05516129359602928,
|
| 9072 |
+
"learning_rate": 6.321071782666077e-06,
|
| 9073 |
+
"loss": 12.3579,
|
| 9074 |
+
"step": 1295
|
| 9075 |
+
},
|
| 9076 |
+
{
|
| 9077 |
+
"epoch": 0.4896108802417832,
|
| 9078 |
+
"grad_norm": 0.046728022396564484,
|
| 9079 |
+
"learning_rate": 6.271489000668418e-06,
|
| 9080 |
+
"loss": 12.3693,
|
| 9081 |
+
"step": 1296
|
| 9082 |
+
},
|
| 9083 |
+
{
|
| 9084 |
+
"epoch": 0.4899886664148092,
|
| 9085 |
+
"grad_norm": 0.055551476776599884,
|
| 9086 |
+
"learning_rate": 6.222088434895462e-06,
|
| 9087 |
+
"loss": 12.3703,
|
| 9088 |
+
"step": 1297
|
| 9089 |
+
},
|
| 9090 |
+
{
|
| 9091 |
+
"epoch": 0.4903664525878353,
|
| 9092 |
+
"grad_norm": 0.06050414592027664,
|
| 9093 |
+
"learning_rate": 6.172870291198824e-06,
|
| 9094 |
+
"loss": 12.3741,
|
| 9095 |
+
"step": 1298
|
| 9096 |
+
},
|
| 9097 |
+
{
|
| 9098 |
+
"epoch": 0.4907442387608614,
|
| 9099 |
+
"grad_norm": 0.07734154164791107,
|
| 9100 |
+
"learning_rate": 6.123834774669956e-06,
|
| 9101 |
+
"loss": 12.3748,
|
| 9102 |
+
"step": 1299
|
| 9103 |
+
},
|
| 9104 |
+
{
|
| 9105 |
+
"epoch": 0.4911220249338874,
|
| 9106 |
+
"grad_norm": 0.11719764024019241,
|
| 9107 |
+
"learning_rate": 6.074982089639336e-06,
|
| 9108 |
+
"loss": 12.3771,
|
| 9109 |
+
"step": 1300
|
| 9110 |
+
},
|
| 9111 |
+
{
|
| 9112 |
+
"epoch": 0.4914998111069135,
|
| 9113 |
+
"grad_norm": 0.051039863377809525,
|
| 9114 |
+
"learning_rate": 6.026312439675552e-06,
|
| 9115 |
+
"loss": 12.3685,
|
| 9116 |
+
"step": 1301
|
| 9117 |
+
},
|
| 9118 |
+
{
|
| 9119 |
+
"epoch": 0.49187759727993957,
|
| 9120 |
+
"grad_norm": 0.04042213782668114,
|
| 9121 |
+
"learning_rate": 5.977826027584505e-06,
|
| 9122 |
+
"loss": 12.3719,
|
| 9123 |
+
"step": 1302
|
| 9124 |
+
},
|
| 9125 |
+
{
|
| 9126 |
+
"epoch": 0.4922553834529656,
|
| 9127 |
+
"grad_norm": 0.0355064831674099,
|
| 9128 |
+
"learning_rate": 5.929523055408537e-06,
|
| 9129 |
+
"loss": 12.3732,
|
| 9130 |
+
"step": 1303
|
| 9131 |
+
},
|
| 9132 |
+
{
|
| 9133 |
+
"epoch": 0.4926331696259917,
|
| 9134 |
+
"grad_norm": 0.04373539611697197,
|
| 9135 |
+
"learning_rate": 5.881403724425605e-06,
|
| 9136 |
+
"loss": 12.3731,
|
| 9137 |
+
"step": 1304
|
| 9138 |
+
},
|
| 9139 |
+
{
|
| 9140 |
+
"epoch": 0.49301095579901777,
|
| 9141 |
+
"grad_norm": 0.04197046160697937,
|
| 9142 |
+
"learning_rate": 5.833468235148437e-06,
|
| 9143 |
+
"loss": 12.3823,
|
| 9144 |
+
"step": 1305
|
| 9145 |
+
},
|
| 9146 |
+
{
|
| 9147 |
+
"epoch": 0.4933887419720438,
|
| 9148 |
+
"grad_norm": 0.03871211037039757,
|
| 9149 |
+
"learning_rate": 5.7857167873236805e-06,
|
| 9150 |
+
"loss": 12.3787,
|
| 9151 |
+
"step": 1306
|
| 9152 |
+
},
|
| 9153 |
+
{
|
| 9154 |
+
"epoch": 0.4937665281450699,
|
| 9155 |
+
"grad_norm": 0.034230586141347885,
|
| 9156 |
+
"learning_rate": 5.7381495799311015e-06,
|
| 9157 |
+
"loss": 12.3719,
|
| 9158 |
+
"step": 1307
|
| 9159 |
+
},
|
| 9160 |
+
{
|
| 9161 |
+
"epoch": 0.49414431431809597,
|
| 9162 |
+
"grad_norm": 0.04650005325675011,
|
| 9163 |
+
"learning_rate": 5.690766811182729e-06,
|
| 9164 |
+
"loss": 12.373,
|
| 9165 |
+
"step": 1308
|
| 9166 |
+
},
|
| 9167 |
+
{
|
| 9168 |
+
"epoch": 0.49452210049112205,
|
| 9169 |
+
"grad_norm": 0.03966010734438896,
|
| 9170 |
+
"learning_rate": 5.643568678522038e-06,
|
| 9171 |
+
"loss": 12.3681,
|
| 9172 |
+
"step": 1309
|
| 9173 |
+
},
|
| 9174 |
+
{
|
| 9175 |
+
"epoch": 0.4948998866641481,
|
| 9176 |
+
"grad_norm": 0.03580264374613762,
|
| 9177 |
+
"learning_rate": 5.596555378623125e-06,
|
| 9178 |
+
"loss": 12.3739,
|
| 9179 |
+
"step": 1310
|
| 9180 |
+
},
|
| 9181 |
+
{
|
| 9182 |
+
"epoch": 0.49527767283717417,
|
| 9183 |
+
"grad_norm": 0.039899274706840515,
|
| 9184 |
+
"learning_rate": 5.549727107389902e-06,
|
| 9185 |
+
"loss": 12.3718,
|
| 9186 |
+
"step": 1311
|
| 9187 |
+
},
|
| 9188 |
+
{
|
| 9189 |
+
"epoch": 0.49565545901020025,
|
| 9190 |
+
"grad_norm": 0.03877566382288933,
|
| 9191 |
+
"learning_rate": 5.503084059955255e-06,
|
| 9192 |
+
"loss": 12.3793,
|
| 9193 |
+
"step": 1312
|
| 9194 |
+
},
|
| 9195 |
+
{
|
| 9196 |
+
"epoch": 0.4960332451832263,
|
| 9197 |
+
"grad_norm": 0.03266843035817146,
|
| 9198 |
+
"learning_rate": 5.45662643068025e-06,
|
| 9199 |
+
"loss": 12.3717,
|
| 9200 |
+
"step": 1313
|
| 9201 |
+
},
|
| 9202 |
+
{
|
| 9203 |
+
"epoch": 0.49641103135625236,
|
| 9204 |
+
"grad_norm": 0.04085678979754448,
|
| 9205 |
+
"learning_rate": 5.4103544131533236e-06,
|
| 9206 |
+
"loss": 12.3661,
|
| 9207 |
+
"step": 1314
|
| 9208 |
+
},
|
| 9209 |
+
{
|
| 9210 |
+
"epoch": 0.49678881752927845,
|
| 9211 |
+
"grad_norm": 0.03333406522870064,
|
| 9212 |
+
"learning_rate": 5.364268200189454e-06,
|
| 9213 |
+
"loss": 12.372,
|
| 9214 |
+
"step": 1315
|
| 9215 |
+
},
|
| 9216 |
+
{
|
| 9217 |
+
"epoch": 0.4971666037023045,
|
| 9218 |
+
"grad_norm": 0.03501317277550697,
|
| 9219 |
+
"learning_rate": 5.318367983829392e-06,
|
| 9220 |
+
"loss": 12.3746,
|
| 9221 |
+
"step": 1316
|
| 9222 |
+
},
|
| 9223 |
+
{
|
| 9224 |
+
"epoch": 0.49754438987533056,
|
| 9225 |
+
"grad_norm": 0.036452457308769226,
|
| 9226 |
+
"learning_rate": 5.2726539553388296e-06,
|
| 9227 |
+
"loss": 12.369,
|
| 9228 |
+
"step": 1317
|
| 9229 |
+
},
|
| 9230 |
+
{
|
| 9231 |
+
"epoch": 0.49792217604835665,
|
| 9232 |
+
"grad_norm": 0.03703896328806877,
|
| 9233 |
+
"learning_rate": 5.227126305207625e-06,
|
| 9234 |
+
"loss": 12.3688,
|
| 9235 |
+
"step": 1318
|
| 9236 |
+
},
|
| 9237 |
+
{
|
| 9238 |
+
"epoch": 0.4982999622213827,
|
| 9239 |
+
"grad_norm": 0.03834862634539604,
|
| 9240 |
+
"learning_rate": 5.181785223148999e-06,
|
| 9241 |
+
"loss": 12.3725,
|
| 9242 |
+
"step": 1319
|
| 9243 |
+
},
|
| 9244 |
+
{
|
| 9245 |
+
"epoch": 0.49867774839440876,
|
| 9246 |
+
"grad_norm": 0.03437526896595955,
|
| 9247 |
+
"learning_rate": 5.136630898098732e-06,
|
| 9248 |
+
"loss": 12.3682,
|
| 9249 |
+
"step": 1320
|
| 9250 |
+
},
|
| 9251 |
+
{
|
| 9252 |
+
"epoch": 0.49905553456743484,
|
| 9253 |
+
"grad_norm": 0.04603976383805275,
|
| 9254 |
+
"learning_rate": 5.091663518214407e-06,
|
| 9255 |
+
"loss": 12.3646,
|
| 9256 |
+
"step": 1321
|
| 9257 |
+
},
|
| 9258 |
+
{
|
| 9259 |
+
"epoch": 0.4994333207404609,
|
| 9260 |
+
"grad_norm": 0.03163118287920952,
|
| 9261 |
+
"learning_rate": 5.046883270874603e-06,
|
| 9262 |
+
"loss": 12.3669,
|
| 9263 |
+
"step": 1322
|
| 9264 |
+
},
|
| 9265 |
+
{
|
| 9266 |
+
"epoch": 0.49981110691348696,
|
| 9267 |
+
"grad_norm": 0.035730570554733276,
|
| 9268 |
+
"learning_rate": 5.002290342678112e-06,
|
| 9269 |
+
"loss": 12.363,
|
| 9270 |
+
"step": 1323
|
| 9271 |
+
},
|
| 9272 |
+
{
|
| 9273 |
+
"epoch": 0.500188893086513,
|
| 9274 |
+
"grad_norm": 0.03787192702293396,
|
| 9275 |
+
"learning_rate": 4.957884919443184e-06,
|
| 9276 |
+
"loss": 12.3627,
|
| 9277 |
+
"step": 1324
|
| 9278 |
+
},
|
| 9279 |
+
{
|
| 9280 |
+
"epoch": 0.5005666792595391,
|
| 9281 |
+
"grad_norm": 0.06199578195810318,
|
| 9282 |
+
"learning_rate": 4.913667186206722e-06,
|
| 9283 |
+
"loss": 12.365,
|
| 9284 |
+
"step": 1325
|
| 9285 |
+
},
|
| 9286 |
+
{
|
| 9287 |
+
"epoch": 0.5009444654325652,
|
| 9288 |
+
"grad_norm": 0.031377822160720825,
|
| 9289 |
+
"learning_rate": 4.869637327223542e-06,
|
| 9290 |
+
"loss": 12.3698,
|
| 9291 |
+
"step": 1326
|
| 9292 |
+
},
|
| 9293 |
+
{
|
| 9294 |
+
"epoch": 0.5013222516055912,
|
| 9295 |
+
"grad_norm": 0.03355984762310982,
|
| 9296 |
+
"learning_rate": 4.825795525965576e-06,
|
| 9297 |
+
"loss": 12.3644,
|
| 9298 |
+
"step": 1327
|
| 9299 |
+
},
|
| 9300 |
+
{
|
| 9301 |
+
"epoch": 0.5017000377786173,
|
| 9302 |
+
"grad_norm": 0.04313547536730766,
|
| 9303 |
+
"learning_rate": 4.782141965121128e-06,
|
| 9304 |
+
"loss": 12.3668,
|
| 9305 |
+
"step": 1328
|
| 9306 |
+
},
|
| 9307 |
+
{
|
| 9308 |
+
"epoch": 0.5020778239516434,
|
| 9309 |
+
"grad_norm": 0.034874800592660904,
|
| 9310 |
+
"learning_rate": 4.738676826594113e-06,
|
| 9311 |
+
"loss": 12.374,
|
| 9312 |
+
"step": 1329
|
| 9313 |
+
},
|
| 9314 |
+
{
|
| 9315 |
+
"epoch": 0.5024556101246694,
|
| 9316 |
+
"grad_norm": 0.046843621879816055,
|
| 9317 |
+
"learning_rate": 4.695400291503277e-06,
|
| 9318 |
+
"loss": 12.3678,
|
| 9319 |
+
"step": 1330
|
| 9320 |
+
},
|
| 9321 |
+
{
|
| 9322 |
+
"epoch": 0.5028333962976955,
|
| 9323 |
+
"grad_norm": 0.047022756189107895,
|
| 9324 |
+
"learning_rate": 4.65231254018147e-06,
|
| 9325 |
+
"loss": 12.3689,
|
| 9326 |
+
"step": 1331
|
| 9327 |
+
},
|
| 9328 |
+
{
|
| 9329 |
+
"epoch": 0.5032111824707216,
|
| 9330 |
+
"grad_norm": 0.037727728486061096,
|
| 9331 |
+
"learning_rate": 4.6094137521748795e-06,
|
| 9332 |
+
"loss": 12.3677,
|
| 9333 |
+
"step": 1332
|
| 9334 |
+
},
|
| 9335 |
+
{
|
| 9336 |
+
"epoch": 0.5035889686437477,
|
| 9337 |
+
"grad_norm": 0.03536062315106392,
|
| 9338 |
+
"learning_rate": 4.566704106242281e-06,
|
| 9339 |
+
"loss": 12.3677,
|
| 9340 |
+
"step": 1333
|
| 9341 |
+
},
|
| 9342 |
+
{
|
| 9343 |
+
"epoch": 0.5039667548167737,
|
| 9344 |
+
"grad_norm": 0.03846399486064911,
|
| 9345 |
+
"learning_rate": 4.524183780354302e-06,
|
| 9346 |
+
"loss": 12.3621,
|
| 9347 |
+
"step": 1334
|
| 9348 |
+
},
|
| 9349 |
+
{
|
| 9350 |
+
"epoch": 0.5043445409897998,
|
| 9351 |
+
"grad_norm": 0.03362063318490982,
|
| 9352 |
+
"learning_rate": 4.4818529516926726e-06,
|
| 9353 |
+
"loss": 12.3681,
|
| 9354 |
+
"step": 1335
|
| 9355 |
+
},
|
| 9356 |
+
{
|
| 9357 |
+
"epoch": 0.5047223271628258,
|
| 9358 |
+
"grad_norm": 0.03910623863339424,
|
| 9359 |
+
"learning_rate": 4.439711796649487e-06,
|
| 9360 |
+
"loss": 12.375,
|
| 9361 |
+
"step": 1336
|
| 9362 |
+
},
|
| 9363 |
+
{
|
| 9364 |
+
"epoch": 0.5051001133358519,
|
| 9365 |
+
"grad_norm": 0.049038905650377274,
|
| 9366 |
+
"learning_rate": 4.397760490826481e-06,
|
| 9367 |
+
"loss": 12.3676,
|
| 9368 |
+
"step": 1337
|
| 9369 |
+
},
|
| 9370 |
+
{
|
| 9371 |
+
"epoch": 0.505477899508878,
|
| 9372 |
+
"grad_norm": 0.040389277040958405,
|
| 9373 |
+
"learning_rate": 4.355999209034278e-06,
|
| 9374 |
+
"loss": 12.3617,
|
| 9375 |
+
"step": 1338
|
| 9376 |
+
},
|
| 9377 |
+
{
|
| 9378 |
+
"epoch": 0.5058556856819041,
|
| 9379 |
+
"grad_norm": 0.04509030282497406,
|
| 9380 |
+
"learning_rate": 4.314428125291686e-06,
|
| 9381 |
+
"loss": 12.3689,
|
| 9382 |
+
"step": 1339
|
| 9383 |
+
},
|
| 9384 |
+
{
|
| 9385 |
+
"epoch": 0.5062334718549301,
|
| 9386 |
+
"grad_norm": 0.04911265894770622,
|
| 9387 |
+
"learning_rate": 4.273047412824954e-06,
|
| 9388 |
+
"loss": 12.359,
|
| 9389 |
+
"step": 1340
|
| 9390 |
+
},
|
| 9391 |
+
{
|
| 9392 |
+
"epoch": 0.5066112580279561,
|
| 9393 |
+
"grad_norm": 0.05804325267672539,
|
| 9394 |
+
"learning_rate": 4.231857244067056e-06,
|
| 9395 |
+
"loss": 12.3664,
|
| 9396 |
+
"step": 1341
|
| 9397 |
+
},
|
| 9398 |
+
{
|
| 9399 |
+
"epoch": 0.5069890442009822,
|
| 9400 |
+
"grad_norm": 0.04763133078813553,
|
| 9401 |
+
"learning_rate": 4.190857790656966e-06,
|
| 9402 |
+
"loss": 12.3685,
|
| 9403 |
+
"step": 1342
|
| 9404 |
+
},
|
| 9405 |
+
{
|
| 9406 |
+
"epoch": 0.5073668303740083,
|
| 9407 |
+
"grad_norm": 0.050545379519462585,
|
| 9408 |
+
"learning_rate": 4.150049223438962e-06,
|
| 9409 |
+
"loss": 12.3736,
|
| 9410 |
+
"step": 1343
|
| 9411 |
+
},
|
| 9412 |
+
{
|
| 9413 |
+
"epoch": 0.5077446165470344,
|
| 9414 |
+
"grad_norm": 0.060127004981040955,
|
| 9415 |
+
"learning_rate": 4.109431712461892e-06,
|
| 9416 |
+
"loss": 12.3616,
|
| 9417 |
+
"step": 1344
|
| 9418 |
+
},
|
| 9419 |
+
{
|
| 9420 |
+
"epoch": 0.5081224027200605,
|
| 9421 |
+
"grad_norm": 0.04868752881884575,
|
| 9422 |
+
"learning_rate": 4.0690054269784835e-06,
|
| 9423 |
+
"loss": 12.3658,
|
| 9424 |
+
"step": 1345
|
| 9425 |
+
},
|
| 9426 |
+
{
|
| 9427 |
+
"epoch": 0.5085001888930865,
|
| 9428 |
+
"grad_norm": 0.04811963066458702,
|
| 9429 |
+
"learning_rate": 4.028770535444615e-06,
|
| 9430 |
+
"loss": 12.3687,
|
| 9431 |
+
"step": 1346
|
| 9432 |
+
},
|
| 9433 |
+
{
|
| 9434 |
+
"epoch": 0.5088779750661125,
|
| 9435 |
+
"grad_norm": 0.06719520688056946,
|
| 9436 |
+
"learning_rate": 3.988727205518644e-06,
|
| 9437 |
+
"loss": 12.3734,
|
| 9438 |
+
"step": 1347
|
| 9439 |
+
},
|
| 9440 |
+
{
|
| 9441 |
+
"epoch": 0.5092557612391386,
|
| 9442 |
+
"grad_norm": 0.06329771876335144,
|
| 9443 |
+
"learning_rate": 3.948875604060681e-06,
|
| 9444 |
+
"loss": 12.3814,
|
| 9445 |
+
"step": 1348
|
| 9446 |
+
},
|
| 9447 |
+
{
|
| 9448 |
+
"epoch": 0.5096335474121647,
|
| 9449 |
+
"grad_norm": 0.07885588705539703,
|
| 9450 |
+
"learning_rate": 3.909215897131918e-06,
|
| 9451 |
+
"loss": 12.3883,
|
| 9452 |
+
"step": 1349
|
| 9453 |
+
},
|
| 9454 |
+
{
|
| 9455 |
+
"epoch": 0.5100113335851908,
|
| 9456 |
+
"grad_norm": 0.11217532306909561,
|
| 9457 |
+
"learning_rate": 3.86974824999391e-06,
|
| 9458 |
+
"loss": 12.3774,
|
| 9459 |
+
"step": 1350
|
| 9460 |
+
},
|
| 9461 |
+
{
|
| 9462 |
+
"epoch": 0.5103891197582169,
|
| 9463 |
+
"grad_norm": 0.045559655874967575,
|
| 9464 |
+
"learning_rate": 3.830472827107917e-06,
|
| 9465 |
+
"loss": 12.3755,
|
| 9466 |
+
"step": 1351
|
| 9467 |
+
},
|
| 9468 |
+
{
|
| 9469 |
+
"epoch": 0.510766905931243,
|
| 9470 |
+
"grad_norm": 0.049998242408037186,
|
| 9471 |
+
"learning_rate": 3.7913897921341866e-06,
|
| 9472 |
+
"loss": 12.3713,
|
| 9473 |
+
"step": 1352
|
| 9474 |
+
},
|
| 9475 |
+
{
|
| 9476 |
+
"epoch": 0.5111446921042689,
|
| 9477 |
+
"grad_norm": 0.05028686672449112,
|
| 9478 |
+
"learning_rate": 3.7524993079313066e-06,
|
| 9479 |
+
"loss": 12.37,
|
| 9480 |
+
"step": 1353
|
| 9481 |
+
},
|
| 9482 |
+
{
|
| 9483 |
+
"epoch": 0.511522478277295,
|
| 9484 |
+
"grad_norm": 0.03987390920519829,
|
| 9485 |
+
"learning_rate": 3.7138015365554833e-06,
|
| 9486 |
+
"loss": 12.3683,
|
| 9487 |
+
"step": 1354
|
| 9488 |
+
},
|
| 9489 |
+
{
|
| 9490 |
+
"epoch": 0.5119002644503211,
|
| 9491 |
+
"grad_norm": 0.03649371489882469,
|
| 9492 |
+
"learning_rate": 3.675296639259912e-06,
|
| 9493 |
+
"loss": 12.3795,
|
| 9494 |
+
"step": 1355
|
| 9495 |
+
},
|
| 9496 |
+
{
|
| 9497 |
+
"epoch": 0.5122780506233472,
|
| 9498 |
+
"grad_norm": 0.03349265828728676,
|
| 9499 |
+
"learning_rate": 3.636984776494079e-06,
|
| 9500 |
+
"loss": 12.3686,
|
| 9501 |
+
"step": 1356
|
| 9502 |
+
},
|
| 9503 |
+
{
|
| 9504 |
+
"epoch": 0.5126558367963733,
|
| 9505 |
+
"grad_norm": 0.04325801134109497,
|
| 9506 |
+
"learning_rate": 3.598866107903093e-06,
|
| 9507 |
+
"loss": 12.379,
|
| 9508 |
+
"step": 1357
|
| 9509 |
+
},
|
| 9510 |
+
{
|
| 9511 |
+
"epoch": 0.5130336229693994,
|
| 9512 |
+
"grad_norm": 0.049634095281362534,
|
| 9513 |
+
"learning_rate": 3.560940792327028e-06,
|
| 9514 |
+
"loss": 12.3746,
|
| 9515 |
+
"step": 1358
|
| 9516 |
+
},
|
| 9517 |
+
{
|
| 9518 |
+
"epoch": 0.5134114091424253,
|
| 9519 |
+
"grad_norm": 0.03438926115632057,
|
| 9520 |
+
"learning_rate": 3.5232089878002527e-06,
|
| 9521 |
+
"loss": 12.3732,
|
| 9522 |
+
"step": 1359
|
| 9523 |
+
},
|
| 9524 |
+
{
|
| 9525 |
+
"epoch": 0.5137891953154514,
|
| 9526 |
+
"grad_norm": 0.03937823325395584,
|
| 9527 |
+
"learning_rate": 3.485670851550782e-06,
|
| 9528 |
+
"loss": 12.3664,
|
| 9529 |
+
"step": 1360
|
| 9530 |
+
},
|
| 9531 |
+
{
|
| 9532 |
+
"epoch": 0.5141669814884775,
|
| 9533 |
+
"grad_norm": 0.031588733196258545,
|
| 9534 |
+
"learning_rate": 3.4483265399996246e-06,
|
| 9535 |
+
"loss": 12.369,
|
| 9536 |
+
"step": 1361
|
| 9537 |
+
},
|
| 9538 |
+
{
|
| 9539 |
+
"epoch": 0.5145447676615036,
|
| 9540 |
+
"grad_norm": 0.02943914756178856,
|
| 9541 |
+
"learning_rate": 3.411176208760114e-06,
|
| 9542 |
+
"loss": 12.3721,
|
| 9543 |
+
"step": 1362
|
| 9544 |
+
},
|
| 9545 |
+
{
|
| 9546 |
+
"epoch": 0.5149225538345297,
|
| 9547 |
+
"grad_norm": 0.03561398386955261,
|
| 9548 |
+
"learning_rate": 3.3742200126372635e-06,
|
| 9549 |
+
"loss": 12.3723,
|
| 9550 |
+
"step": 1363
|
| 9551 |
+
},
|
| 9552 |
+
{
|
| 9553 |
+
"epoch": 0.5153003400075558,
|
| 9554 |
+
"grad_norm": 0.03551902249455452,
|
| 9555 |
+
"learning_rate": 3.3374581056271447e-06,
|
| 9556 |
+
"loss": 12.3743,
|
| 9557 |
+
"step": 1364
|
| 9558 |
+
},
|
| 9559 |
+
{
|
| 9560 |
+
"epoch": 0.5156781261805818,
|
| 9561 |
+
"grad_norm": 0.030797170475125313,
|
| 9562 |
+
"learning_rate": 3.300890640916221e-06,
|
| 9563 |
+
"loss": 12.3748,
|
| 9564 |
+
"step": 1365
|
| 9565 |
+
},
|
| 9566 |
+
{
|
| 9567 |
+
"epoch": 0.5160559123536078,
|
| 9568 |
+
"grad_norm": 0.03474782407283783,
|
| 9569 |
+
"learning_rate": 3.2645177708807094e-06,
|
| 9570 |
+
"loss": 12.3704,
|
| 9571 |
+
"step": 1366
|
| 9572 |
+
},
|
| 9573 |
+
{
|
| 9574 |
+
"epoch": 0.5164336985266339,
|
| 9575 |
+
"grad_norm": 0.03498007729649544,
|
| 9576 |
+
"learning_rate": 3.228339647085965e-06,
|
| 9577 |
+
"loss": 12.369,
|
| 9578 |
+
"step": 1367
|
| 9579 |
+
},
|
| 9580 |
+
{
|
| 9581 |
+
"epoch": 0.51681148469966,
|
| 9582 |
+
"grad_norm": 0.0353989377617836,
|
| 9583 |
+
"learning_rate": 3.1923564202858304e-06,
|
| 9584 |
+
"loss": 12.3753,
|
| 9585 |
+
"step": 1368
|
| 9586 |
+
},
|
| 9587 |
+
{
|
| 9588 |
+
"epoch": 0.5171892708726861,
|
| 9589 |
+
"grad_norm": 0.03804413601756096,
|
| 9590 |
+
"learning_rate": 3.1565682404220087e-06,
|
| 9591 |
+
"loss": 12.3658,
|
| 9592 |
+
"step": 1369
|
| 9593 |
+
},
|
| 9594 |
+
{
|
| 9595 |
+
"epoch": 0.5175670570457122,
|
| 9596 |
+
"grad_norm": 0.03596872463822365,
|
| 9597 |
+
"learning_rate": 3.1209752566234653e-06,
|
| 9598 |
+
"loss": 12.3666,
|
| 9599 |
+
"step": 1370
|
| 9600 |
+
},
|
| 9601 |
+
{
|
| 9602 |
+
"epoch": 0.5179448432187382,
|
| 9603 |
+
"grad_norm": 0.037172719836235046,
|
| 9604 |
+
"learning_rate": 3.085577617205765e-06,
|
| 9605 |
+
"loss": 12.3758,
|
| 9606 |
+
"step": 1371
|
| 9607 |
+
},
|
| 9608 |
+
{
|
| 9609 |
+
"epoch": 0.5183226293917642,
|
| 9610 |
+
"grad_norm": 0.03441259264945984,
|
| 9611 |
+
"learning_rate": 3.0503754696704858e-06,
|
| 9612 |
+
"loss": 12.3764,
|
| 9613 |
+
"step": 1372
|
| 9614 |
+
},
|
| 9615 |
+
{
|
| 9616 |
+
"epoch": 0.5187004155647903,
|
| 9617 |
+
"grad_norm": 0.027058640494942665,
|
| 9618 |
+
"learning_rate": 3.0153689607045845e-06,
|
| 9619 |
+
"loss": 12.3705,
|
| 9620 |
+
"step": 1373
|
| 9621 |
+
},
|
| 9622 |
+
{
|
| 9623 |
+
"epoch": 0.5190782017378164,
|
| 9624 |
+
"grad_norm": 0.03889136761426926,
|
| 9625 |
+
"learning_rate": 2.9805582361798035e-06,
|
| 9626 |
+
"loss": 12.3753,
|
| 9627 |
+
"step": 1374
|
| 9628 |
+
},
|
| 9629 |
+
{
|
| 9630 |
+
"epoch": 0.5194559879108425,
|
| 9631 |
+
"grad_norm": 0.029406381770968437,
|
| 9632 |
+
"learning_rate": 2.9459434411520436e-06,
|
| 9633 |
+
"loss": 12.3662,
|
| 9634 |
+
"step": 1375
|
| 9635 |
+
},
|
| 9636 |
+
{
|
| 9637 |
+
"epoch": 0.5198337740838685,
|
| 9638 |
+
"grad_norm": 0.03538326919078827,
|
| 9639 |
+
"learning_rate": 2.9115247198607807e-06,
|
| 9640 |
+
"loss": 12.3644,
|
| 9641 |
+
"step": 1376
|
| 9642 |
+
},
|
| 9643 |
+
{
|
| 9644 |
+
"epoch": 0.5202115602568946,
|
| 9645 |
+
"grad_norm": 0.0315384678542614,
|
| 9646 |
+
"learning_rate": 2.87730221572845e-06,
|
| 9647 |
+
"loss": 12.3737,
|
| 9648 |
+
"step": 1377
|
| 9649 |
+
},
|
| 9650 |
+
{
|
| 9651 |
+
"epoch": 0.5205893464299207,
|
| 9652 |
+
"grad_norm": 0.035543229430913925,
|
| 9653 |
+
"learning_rate": 2.843276071359846e-06,
|
| 9654 |
+
"loss": 12.3755,
|
| 9655 |
+
"step": 1378
|
| 9656 |
+
},
|
| 9657 |
+
{
|
| 9658 |
+
"epoch": 0.5209671326029467,
|
| 9659 |
+
"grad_norm": 0.03708391636610031,
|
| 9660 |
+
"learning_rate": 2.8094464285415344e-06,
|
| 9661 |
+
"loss": 12.3723,
|
| 9662 |
+
"step": 1379
|
| 9663 |
+
},
|
| 9664 |
+
{
|
| 9665 |
+
"epoch": 0.5213449187759728,
|
| 9666 |
+
"grad_norm": 0.031074371188879013,
|
| 9667 |
+
"learning_rate": 2.775813428241275e-06,
|
| 9668 |
+
"loss": 12.3678,
|
| 9669 |
+
"step": 1380
|
| 9670 |
+
},
|
| 9671 |
+
{
|
| 9672 |
+
"epoch": 0.5217227049489989,
|
| 9673 |
+
"grad_norm": 0.031949955970048904,
|
| 9674 |
+
"learning_rate": 2.7423772106073996e-06,
|
| 9675 |
+
"loss": 12.3707,
|
| 9676 |
+
"step": 1381
|
| 9677 |
+
},
|
| 9678 |
+
{
|
| 9679 |
+
"epoch": 0.522100491122025,
|
| 9680 |
+
"grad_norm": 0.03809535875916481,
|
| 9681 |
+
"learning_rate": 2.7091379149682685e-06,
|
| 9682 |
+
"loss": 12.3634,
|
| 9683 |
+
"step": 1382
|
| 9684 |
+
},
|
| 9685 |
+
{
|
| 9686 |
+
"epoch": 0.522478277295051,
|
| 9687 |
+
"grad_norm": 0.03852887079119682,
|
| 9688 |
+
"learning_rate": 2.6760956798316582e-06,
|
| 9689 |
+
"loss": 12.3645,
|
| 9690 |
+
"step": 1383
|
| 9691 |
+
},
|
| 9692 |
+
{
|
| 9693 |
+
"epoch": 0.5228560634680771,
|
| 9694 |
+
"grad_norm": 0.04403216764330864,
|
| 9695 |
+
"learning_rate": 2.6432506428841987e-06,
|
| 9696 |
+
"loss": 12.3605,
|
| 9697 |
+
"step": 1384
|
| 9698 |
+
},
|
| 9699 |
+
{
|
| 9700 |
+
"epoch": 0.5232338496411031,
|
| 9701 |
+
"grad_norm": 0.03586175665259361,
|
| 9702 |
+
"learning_rate": 2.6106029409907974e-06,
|
| 9703 |
+
"loss": 12.3634,
|
| 9704 |
+
"step": 1385
|
| 9705 |
+
},
|
| 9706 |
+
{
|
| 9707 |
+
"epoch": 0.5236116358141292,
|
| 9708 |
+
"grad_norm": 0.039137307554483414,
|
| 9709 |
+
"learning_rate": 2.578152710194065e-06,
|
| 9710 |
+
"loss": 12.3582,
|
| 9711 |
+
"step": 1386
|
| 9712 |
+
},
|
| 9713 |
+
{
|
| 9714 |
+
"epoch": 0.5239894219871553,
|
| 9715 |
+
"grad_norm": 0.040835149586200714,
|
| 9716 |
+
"learning_rate": 2.5459000857137704e-06,
|
| 9717 |
+
"loss": 12.3626,
|
| 9718 |
+
"step": 1387
|
| 9719 |
+
},
|
| 9720 |
+
{
|
| 9721 |
+
"epoch": 0.5243672081601813,
|
| 9722 |
+
"grad_norm": 0.03839719668030739,
|
| 9723 |
+
"learning_rate": 2.51384520194623e-06,
|
| 9724 |
+
"loss": 12.3714,
|
| 9725 |
+
"step": 1388
|
| 9726 |
+
},
|
| 9727 |
+
{
|
| 9728 |
+
"epoch": 0.5247449943332074,
|
| 9729 |
+
"grad_norm": 0.05415833368897438,
|
| 9730 |
+
"learning_rate": 2.4819881924638023e-06,
|
| 9731 |
+
"loss": 12.3675,
|
| 9732 |
+
"step": 1389
|
| 9733 |
+
},
|
| 9734 |
+
{
|
| 9735 |
+
"epoch": 0.5251227805062335,
|
| 9736 |
+
"grad_norm": 0.048753414303064346,
|
| 9737 |
+
"learning_rate": 2.450329190014289e-06,
|
| 9738 |
+
"loss": 12.3703,
|
| 9739 |
+
"step": 1390
|
| 9740 |
+
},
|
| 9741 |
+
{
|
| 9742 |
+
"epoch": 0.5255005666792595,
|
| 9743 |
+
"grad_norm": 0.04843161255121231,
|
| 9744 |
+
"learning_rate": 2.4188683265204127e-06,
|
| 9745 |
+
"loss": 12.3663,
|
| 9746 |
+
"step": 1391
|
| 9747 |
+
},
|
| 9748 |
+
{
|
| 9749 |
+
"epoch": 0.5258783528522856,
|
| 9750 |
+
"grad_norm": 0.04117654263973236,
|
| 9751 |
+
"learning_rate": 2.3876057330792346e-06,
|
| 9752 |
+
"loss": 12.3666,
|
| 9753 |
+
"step": 1392
|
| 9754 |
+
},
|
| 9755 |
+
{
|
| 9756 |
+
"epoch": 0.5262561390253117,
|
| 9757 |
+
"grad_norm": 0.05515278875827789,
|
| 9758 |
+
"learning_rate": 2.3565415399616486e-06,
|
| 9759 |
+
"loss": 12.3725,
|
| 9760 |
+
"step": 1393
|
| 9761 |
+
},
|
| 9762 |
+
{
|
| 9763 |
+
"epoch": 0.5266339251983377,
|
| 9764 |
+
"grad_norm": 0.04353925958275795,
|
| 9765 |
+
"learning_rate": 2.325675876611805e-06,
|
| 9766 |
+
"loss": 12.3679,
|
| 9767 |
+
"step": 1394
|
| 9768 |
+
},
|
| 9769 |
+
{
|
| 9770 |
+
"epoch": 0.5270117113713638,
|
| 9771 |
+
"grad_norm": 0.06032606214284897,
|
| 9772 |
+
"learning_rate": 2.2950088716465755e-06,
|
| 9773 |
+
"loss": 12.3663,
|
| 9774 |
+
"step": 1395
|
| 9775 |
+
},
|
| 9776 |
+
{
|
| 9777 |
+
"epoch": 0.5273894975443899,
|
| 9778 |
+
"grad_norm": 0.04356678947806358,
|
| 9779 |
+
"learning_rate": 2.2645406528550405e-06,
|
| 9780 |
+
"loss": 12.3604,
|
| 9781 |
+
"step": 1396
|
| 9782 |
+
},
|
| 9783 |
+
{
|
| 9784 |
+
"epoch": 0.527767283717416,
|
| 9785 |
+
"grad_norm": 0.05580585077404976,
|
| 9786 |
+
"learning_rate": 2.2342713471979303e-06,
|
| 9787 |
+
"loss": 12.3688,
|
| 9788 |
+
"step": 1397
|
| 9789 |
+
},
|
| 9790 |
+
{
|
| 9791 |
+
"epoch": 0.528145069890442,
|
| 9792 |
+
"grad_norm": 0.06988968700170517,
|
| 9793 |
+
"learning_rate": 2.204201080807111e-06,
|
| 9794 |
+
"loss": 12.3727,
|
| 9795 |
+
"step": 1398
|
| 9796 |
+
},
|
| 9797 |
+
{
|
| 9798 |
+
"epoch": 0.528522856063468,
|
| 9799 |
+
"grad_norm": 0.10209229588508606,
|
| 9800 |
+
"learning_rate": 2.1743299789850457e-06,
|
| 9801 |
+
"loss": 12.3752,
|
| 9802 |
+
"step": 1399
|
| 9803 |
+
},
|
| 9804 |
+
{
|
| 9805 |
+
"epoch": 0.5289006422364941,
|
| 9806 |
+
"grad_norm": 0.12823429703712463,
|
| 9807 |
+
"learning_rate": 2.1446581662042942e-06,
|
| 9808 |
+
"loss": 12.3812,
|
| 9809 |
+
"step": 1400
|
| 9810 |
+
},
|
| 9811 |
+
{
|
| 9812 |
+
"epoch": 0.5292784284095202,
|
| 9813 |
+
"grad_norm": 0.05512097477912903,
|
| 9814 |
+
"learning_rate": 2.1151857661069696e-06,
|
| 9815 |
+
"loss": 12.3741,
|
| 9816 |
+
"step": 1401
|
| 9817 |
+
},
|
| 9818 |
+
{
|
| 9819 |
+
"epoch": 0.5296562145825463,
|
| 9820 |
+
"grad_norm": 0.03914400935173035,
|
| 9821 |
+
"learning_rate": 2.0859129015042445e-06,
|
| 9822 |
+
"loss": 12.3728,
|
| 9823 |
+
"step": 1402
|
| 9824 |
+
},
|
| 9825 |
+
{
|
| 9826 |
+
"epoch": 0.5300340007555724,
|
| 9827 |
+
"grad_norm": 0.051340021193027496,
|
| 9828 |
+
"learning_rate": 2.0568396943758172e-06,
|
| 9829 |
+
"loss": 12.3701,
|
| 9830 |
+
"step": 1403
|
| 9831 |
+
},
|
| 9832 |
+
{
|
| 9833 |
+
"epoch": 0.5304117869285984,
|
| 9834 |
+
"grad_norm": 0.05863821879029274,
|
| 9835 |
+
"learning_rate": 2.02796626586943e-06,
|
| 9836 |
+
"loss": 12.3688,
|
| 9837 |
+
"step": 1404
|
| 9838 |
+
},
|
| 9839 |
+
{
|
| 9840 |
+
"epoch": 0.5307895731016244,
|
| 9841 |
+
"grad_norm": 0.0406498908996582,
|
| 9842 |
+
"learning_rate": 1.9992927363003346e-06,
|
| 9843 |
+
"loss": 12.3717,
|
| 9844 |
+
"step": 1405
|
| 9845 |
+
},
|
| 9846 |
+
{
|
| 9847 |
+
"epoch": 0.5311673592746505,
|
| 9848 |
+
"grad_norm": 0.03591388091444969,
|
| 9849 |
+
"learning_rate": 1.970819225150822e-06,
|
| 9850 |
+
"loss": 12.3745,
|
| 9851 |
+
"step": 1406
|
| 9852 |
+
},
|
| 9853 |
+
{
|
| 9854 |
+
"epoch": 0.5315451454476766,
|
| 9855 |
+
"grad_norm": 0.03432812541723251,
|
| 9856 |
+
"learning_rate": 1.942545851069688e-06,
|
| 9857 |
+
"loss": 12.3726,
|
| 9858 |
+
"step": 1407
|
| 9859 |
+
},
|
| 9860 |
+
{
|
| 9861 |
+
"epoch": 0.5319229316207027,
|
| 9862 |
+
"grad_norm": 0.03600994497537613,
|
| 9863 |
+
"learning_rate": 1.9144727318717857e-06,
|
| 9864 |
+
"loss": 12.3738,
|
| 9865 |
+
"step": 1408
|
| 9866 |
+
},
|
| 9867 |
+
{
|
| 9868 |
+
"epoch": 0.5323007177937288,
|
| 9869 |
+
"grad_norm": 0.03433964028954506,
|
| 9870 |
+
"learning_rate": 1.8865999845374793e-06,
|
| 9871 |
+
"loss": 12.3706,
|
| 9872 |
+
"step": 1409
|
| 9873 |
+
},
|
| 9874 |
+
{
|
| 9875 |
+
"epoch": 0.5326785039667549,
|
| 9876 |
+
"grad_norm": 0.049884695559740067,
|
| 9877 |
+
"learning_rate": 1.8589277252122173e-06,
|
| 9878 |
+
"loss": 12.3731,
|
| 9879 |
+
"step": 1410
|
| 9880 |
+
},
|
| 9881 |
+
{
|
| 9882 |
+
"epoch": 0.5330562901397808,
|
| 9883 |
+
"grad_norm": 0.034618549048900604,
|
| 9884 |
+
"learning_rate": 1.8314560692059835e-06,
|
| 9885 |
+
"loss": 12.379,
|
| 9886 |
+
"step": 1411
|
| 9887 |
+
},
|
| 9888 |
+
{
|
| 9889 |
+
"epoch": 0.5334340763128069,
|
| 9890 |
+
"grad_norm": 0.04899175092577934,
|
| 9891 |
+
"learning_rate": 1.8041851309928802e-06,
|
| 9892 |
+
"loss": 12.3695,
|
| 9893 |
+
"step": 1412
|
| 9894 |
+
},
|
| 9895 |
+
{
|
| 9896 |
+
"epoch": 0.533811862485833,
|
| 9897 |
+
"grad_norm": 0.04382387176156044,
|
| 9898 |
+
"learning_rate": 1.7771150242106016e-06,
|
| 9899 |
+
"loss": 12.3714,
|
| 9900 |
+
"step": 1413
|
| 9901 |
+
},
|
| 9902 |
+
{
|
| 9903 |
+
"epoch": 0.5341896486588591,
|
| 9904 |
+
"grad_norm": 0.04348282516002655,
|
| 9905 |
+
"learning_rate": 1.7502458616599827e-06,
|
| 9906 |
+
"loss": 12.3659,
|
| 9907 |
+
"step": 1414
|
| 9908 |
+
},
|
| 9909 |
+
{
|
| 9910 |
+
"epoch": 0.5345674348318852,
|
| 9911 |
+
"grad_norm": 0.03303327038884163,
|
| 9912 |
+
"learning_rate": 1.7235777553045284e-06,
|
| 9913 |
+
"loss": 12.3722,
|
| 9914 |
+
"step": 1415
|
| 9915 |
+
},
|
| 9916 |
+
{
|
| 9917 |
+
"epoch": 0.5349452210049113,
|
| 9918 |
+
"grad_norm": 0.033907659351825714,
|
| 9919 |
+
"learning_rate": 1.6971108162699478e-06,
|
| 9920 |
+
"loss": 12.3677,
|
| 9921 |
+
"step": 1416
|
| 9922 |
+
},
|
| 9923 |
+
{
|
| 9924 |
+
"epoch": 0.5353230071779372,
|
| 9925 |
+
"grad_norm": 0.03980765864253044,
|
| 9926 |
+
"learning_rate": 1.6708451548436866e-06,
|
| 9927 |
+
"loss": 12.3773,
|
| 9928 |
+
"step": 1417
|
| 9929 |
+
},
|
| 9930 |
+
{
|
| 9931 |
+
"epoch": 0.5357007933509633,
|
| 9932 |
+
"grad_norm": 0.028395546600222588,
|
| 9933 |
+
"learning_rate": 1.6447808804744668e-06,
|
| 9934 |
+
"loss": 12.3671,
|
| 9935 |
+
"step": 1418
|
| 9936 |
+
},
|
| 9937 |
+
{
|
| 9938 |
+
"epoch": 0.5360785795239894,
|
| 9939 |
+
"grad_norm": 0.03201286494731903,
|
| 9940 |
+
"learning_rate": 1.618918101771827e-06,
|
| 9941 |
+
"loss": 12.3765,
|
| 9942 |
+
"step": 1419
|
| 9943 |
+
},
|
| 9944 |
+
{
|
| 9945 |
+
"epoch": 0.5364563656970155,
|
| 9946 |
+
"grad_norm": 0.03632766008377075,
|
| 9947 |
+
"learning_rate": 1.5932569265056874e-06,
|
| 9948 |
+
"loss": 12.3747,
|
| 9949 |
+
"step": 1420
|
| 9950 |
+
},
|
| 9951 |
+
{
|
| 9952 |
+
"epoch": 0.5368341518700416,
|
| 9953 |
+
"grad_norm": 0.039473291486501694,
|
| 9954 |
+
"learning_rate": 1.5677974616058855e-06,
|
| 9955 |
+
"loss": 12.3676,
|
| 9956 |
+
"step": 1421
|
| 9957 |
+
},
|
| 9958 |
+
{
|
| 9959 |
+
"epoch": 0.5372119380430677,
|
| 9960 |
+
"grad_norm": 0.04315740242600441,
|
| 9961 |
+
"learning_rate": 1.5425398131617252e-06,
|
| 9962 |
+
"loss": 12.3648,
|
| 9963 |
+
"step": 1422
|
| 9964 |
+
},
|
| 9965 |
+
{
|
| 9966 |
+
"epoch": 0.5375897242160937,
|
| 9967 |
+
"grad_norm": 0.03224621340632439,
|
| 9968 |
+
"learning_rate": 1.5174840864215555e-06,
|
| 9969 |
+
"loss": 12.3657,
|
| 9970 |
+
"step": 1423
|
| 9971 |
+
},
|
| 9972 |
+
{
|
| 9973 |
+
"epoch": 0.5379675103891197,
|
| 9974 |
+
"grad_norm": 0.04244806617498398,
|
| 9975 |
+
"learning_rate": 1.4926303857923097e-06,
|
| 9976 |
+
"loss": 12.3734,
|
| 9977 |
+
"step": 1424
|
| 9978 |
+
},
|
| 9979 |
+
{
|
| 9980 |
+
"epoch": 0.5383452965621458,
|
| 9981 |
+
"grad_norm": 0.031122561544179916,
|
| 9982 |
+
"learning_rate": 1.4679788148390827e-06,
|
| 9983 |
+
"loss": 12.3766,
|
| 9984 |
+
"step": 1425
|
| 9985 |
+
},
|
| 9986 |
+
{
|
| 9987 |
+
"epoch": 0.5387230827351719,
|
| 9988 |
+
"grad_norm": 0.045771513134241104,
|
| 9989 |
+
"learning_rate": 1.4435294762847107e-06,
|
| 9990 |
+
"loss": 12.3758,
|
| 9991 |
+
"step": 1426
|
| 9992 |
+
},
|
| 9993 |
+
{
|
| 9994 |
+
"epoch": 0.539100868908198,
|
| 9995 |
+
"grad_norm": 0.03772364929318428,
|
| 9996 |
+
"learning_rate": 1.419282472009309e-06,
|
| 9997 |
+
"loss": 12.3608,
|
| 9998 |
+
"step": 1427
|
| 9999 |
+
},
|
| 10000 |
+
{
|
| 10001 |
+
"epoch": 0.5394786550812241,
|
| 10002 |
+
"grad_norm": 0.0344366617500782,
|
| 10003 |
+
"learning_rate": 1.395237903049884e-06,
|
| 10004 |
+
"loss": 12.3669,
|
| 10005 |
+
"step": 1428
|
| 10006 |
+
},
|
| 10007 |
+
{
|
| 10008 |
+
"epoch": 0.5398564412542501,
|
| 10009 |
+
"grad_norm": 0.037167783826589584,
|
| 10010 |
+
"learning_rate": 1.3713958695998895e-06,
|
| 10011 |
+
"loss": 12.3657,
|
| 10012 |
+
"step": 1429
|
| 10013 |
+
},
|
| 10014 |
+
{
|
| 10015 |
+
"epoch": 0.5402342274272761,
|
| 10016 |
+
"grad_norm": 0.036133185029029846,
|
| 10017 |
+
"learning_rate": 1.3477564710088098e-06,
|
| 10018 |
+
"loss": 12.3755,
|
| 10019 |
+
"step": 1430
|
| 10020 |
+
},
|
| 10021 |
+
{
|
| 10022 |
+
"epoch": 0.5406120136003022,
|
| 10023 |
+
"grad_norm": 0.035620953887701035,
|
| 10024 |
+
"learning_rate": 1.3243198057817652e-06,
|
| 10025 |
+
"loss": 12.3698,
|
| 10026 |
+
"step": 1431
|
| 10027 |
+
},
|
| 10028 |
+
{
|
| 10029 |
+
"epoch": 0.5409897997733283,
|
| 10030 |
+
"grad_norm": 0.036473289132118225,
|
| 10031 |
+
"learning_rate": 1.3010859715790802e-06,
|
| 10032 |
+
"loss": 12.3725,
|
| 10033 |
+
"step": 1432
|
| 10034 |
+
},
|
| 10035 |
+
{
|
| 10036 |
+
"epoch": 0.5413675859463544,
|
| 10037 |
+
"grad_norm": 0.04385390877723694,
|
| 10038 |
+
"learning_rate": 1.278055065215894e-06,
|
| 10039 |
+
"loss": 12.373,
|
| 10040 |
+
"step": 1433
|
| 10041 |
+
},
|
| 10042 |
+
{
|
| 10043 |
+
"epoch": 0.5417453721193805,
|
| 10044 |
+
"grad_norm": 0.034310031682252884,
|
| 10045 |
+
"learning_rate": 1.255227182661728e-06,
|
| 10046 |
+
"loss": 12.3711,
|
| 10047 |
+
"step": 1434
|
| 10048 |
+
},
|
| 10049 |
+
{
|
| 10050 |
+
"epoch": 0.5421231582924065,
|
| 10051 |
+
"grad_norm": 0.04797558858990669,
|
| 10052 |
+
"learning_rate": 1.2326024190401242e-06,
|
| 10053 |
+
"loss": 12.3723,
|
| 10054 |
+
"step": 1435
|
| 10055 |
+
},
|
| 10056 |
+
{
|
| 10057 |
+
"epoch": 0.5425009444654325,
|
| 10058 |
+
"grad_norm": 0.03860705345869064,
|
| 10059 |
+
"learning_rate": 1.210180868628219e-06,
|
| 10060 |
+
"loss": 12.3709,
|
| 10061 |
+
"step": 1436
|
| 10062 |
+
},
|
| 10063 |
+
{
|
| 10064 |
+
"epoch": 0.5428787306384586,
|
| 10065 |
+
"grad_norm": 0.035169027745723724,
|
| 10066 |
+
"learning_rate": 1.1879626248563758e-06,
|
| 10067 |
+
"loss": 12.3664,
|
| 10068 |
+
"step": 1437
|
| 10069 |
+
},
|
| 10070 |
+
{
|
| 10071 |
+
"epoch": 0.5432565168114847,
|
| 10072 |
+
"grad_norm": 0.044533584266901016,
|
| 10073 |
+
"learning_rate": 1.165947780307769e-06,
|
| 10074 |
+
"loss": 12.3663,
|
| 10075 |
+
"step": 1438
|
| 10076 |
+
},
|
| 10077 |
+
{
|
| 10078 |
+
"epoch": 0.5436343029845108,
|
| 10079 |
+
"grad_norm": 0.035262878984212875,
|
| 10080 |
+
"learning_rate": 1.1441364267180065e-06,
|
| 10081 |
+
"loss": 12.375,
|
| 10082 |
+
"step": 1439
|
| 10083 |
+
},
|
| 10084 |
+
{
|
| 10085 |
+
"epoch": 0.5440120891575368,
|
| 10086 |
+
"grad_norm": 0.03809209540486336,
|
| 10087 |
+
"learning_rate": 1.1225286549747693e-06,
|
| 10088 |
+
"loss": 12.3636,
|
| 10089 |
+
"step": 1440
|
| 10090 |
+
},
|
| 10091 |
+
{
|
| 10092 |
+
"epoch": 0.5443898753305629,
|
| 10093 |
+
"grad_norm": 0.042231276631355286,
|
| 10094 |
+
"learning_rate": 1.1011245551173943e-06,
|
| 10095 |
+
"loss": 12.3714,
|
| 10096 |
+
"step": 1441
|
| 10097 |
+
},
|
| 10098 |
+
{
|
| 10099 |
+
"epoch": 0.544767661503589,
|
| 10100 |
+
"grad_norm": 0.043087128549814224,
|
| 10101 |
+
"learning_rate": 1.0799242163365419e-06,
|
| 10102 |
+
"loss": 12.3682,
|
| 10103 |
+
"step": 1442
|
| 10104 |
+
},
|
| 10105 |
+
{
|
| 10106 |
+
"epoch": 0.545145447676615,
|
| 10107 |
+
"grad_norm": 0.04966861009597778,
|
| 10108 |
+
"learning_rate": 1.0589277269737797e-06,
|
| 10109 |
+
"loss": 12.3644,
|
| 10110 |
+
"step": 1443
|
| 10111 |
+
},
|
| 10112 |
+
{
|
| 10113 |
+
"epoch": 0.5455232338496411,
|
| 10114 |
+
"grad_norm": 0.04546117037534714,
|
| 10115 |
+
"learning_rate": 1.038135174521254e-06,
|
| 10116 |
+
"loss": 12.3602,
|
| 10117 |
+
"step": 1444
|
| 10118 |
+
},
|
| 10119 |
+
{
|
| 10120 |
+
"epoch": 0.5459010200226672,
|
| 10121 |
+
"grad_norm": 0.03941427171230316,
|
| 10122 |
+
"learning_rate": 1.0175466456213034e-06,
|
| 10123 |
+
"loss": 12.3638,
|
| 10124 |
+
"step": 1445
|
| 10125 |
+
},
|
| 10126 |
+
{
|
| 10127 |
+
"epoch": 0.5462788061956932,
|
| 10128 |
+
"grad_norm": 0.04676871374249458,
|
| 10129 |
+
"learning_rate": 9.971622260661007e-07,
|
| 10130 |
+
"loss": 12.3773,
|
| 10131 |
+
"step": 1446
|
| 10132 |
+
},
|
| 10133 |
+
{
|
| 10134 |
+
"epoch": 0.5466565923687193,
|
| 10135 |
+
"grad_norm": 0.060187239199876785,
|
| 10136 |
+
"learning_rate": 9.769820007972997e-07,
|
| 10137 |
+
"loss": 12.3734,
|
| 10138 |
+
"step": 1447
|
| 10139 |
+
},
|
| 10140 |
+
{
|
| 10141 |
+
"epoch": 0.5470343785417454,
|
| 10142 |
+
"grad_norm": 0.0665447860956192,
|
| 10143 |
+
"learning_rate": 9.57006053905668e-07,
|
| 10144 |
+
"loss": 12.3815,
|
| 10145 |
+
"step": 1448
|
| 10146 |
+
},
|
| 10147 |
+
{
|
| 10148 |
+
"epoch": 0.5474121647147714,
|
| 10149 |
+
"grad_norm": 0.06303698569536209,
|
| 10150 |
+
"learning_rate": 9.372344686307655e-07,
|
| 10151 |
+
"loss": 12.3693,
|
| 10152 |
+
"step": 1449
|
| 10153 |
+
},
|
| 10154 |
+
{
|
| 10155 |
+
"epoch": 0.5477899508877975,
|
| 10156 |
+
"grad_norm": 0.11176969110965729,
|
| 10157 |
+
"learning_rate": 9.176673273605718e-07,
|
| 10158 |
+
"loss": 12.3815,
|
| 10159 |
+
"step": 1450
|
| 10160 |
+
},
|
| 10161 |
+
{
|
| 10162 |
+
"epoch": 0.5481677370608236,
|
| 10163 |
+
"grad_norm": 0.04446045309305191,
|
| 10164 |
+
"learning_rate": 8.983047116311428e-07,
|
| 10165 |
+
"loss": 12.3757,
|
| 10166 |
+
"step": 1451
|
| 10167 |
+
},
|
| 10168 |
+
{
|
| 10169 |
+
"epoch": 0.5485455232338496,
|
| 10170 |
+
"grad_norm": 0.03576750308275223,
|
| 10171 |
+
"learning_rate": 8.791467021262933e-07,
|
| 10172 |
+
"loss": 12.3792,
|
| 10173 |
+
"step": 1452
|
| 10174 |
+
},
|
| 10175 |
+
{
|
| 10176 |
+
"epoch": 0.5489233094068757,
|
| 10177 |
+
"grad_norm": 0.04086080193519592,
|
| 10178 |
+
"learning_rate": 8.60193378677232e-07,
|
| 10179 |
+
"loss": 12.3676,
|
| 10180 |
+
"step": 1453
|
| 10181 |
+
},
|
| 10182 |
+
{
|
| 10183 |
+
"epoch": 0.5493010955799018,
|
| 10184 |
+
"grad_norm": 0.03202464058995247,
|
| 10185 |
+
"learning_rate": 8.414448202622494e-07,
|
| 10186 |
+
"loss": 12.3765,
|
| 10187 |
+
"step": 1454
|
| 10188 |
+
},
|
| 10189 |
+
{
|
| 10190 |
+
"epoch": 0.5496788817529279,
|
| 10191 |
+
"grad_norm": 0.04358900710940361,
|
| 10192 |
+
"learning_rate": 8.22901105006385e-07,
|
| 10193 |
+
"loss": 12.3755,
|
| 10194 |
+
"step": 1455
|
| 10195 |
+
},
|
| 10196 |
+
{
|
| 10197 |
+
"epoch": 0.5500566679259539,
|
| 10198 |
+
"grad_norm": 0.03576086089015007,
|
| 10199 |
+
"learning_rate": 8.045623101810951e-07,
|
| 10200 |
+
"loss": 12.3721,
|
| 10201 |
+
"step": 1456
|
| 10202 |
+
},
|
| 10203 |
+
{
|
| 10204 |
+
"epoch": 0.55043445409898,
|
| 10205 |
+
"grad_norm": 0.038851261138916016,
|
| 10206 |
+
"learning_rate": 7.864285122039405e-07,
|
| 10207 |
+
"loss": 12.3728,
|
| 10208 |
+
"step": 1457
|
| 10209 |
+
},
|
| 10210 |
+
{
|
| 10211 |
+
"epoch": 0.550812240272006,
|
| 10212 |
+
"grad_norm": 0.03601044416427612,
|
| 10213 |
+
"learning_rate": 7.684997866382549e-07,
|
| 10214 |
+
"loss": 12.3739,
|
| 10215 |
+
"step": 1458
|
| 10216 |
+
},
|
| 10217 |
+
{
|
| 10218 |
+
"epoch": 0.5511900264450321,
|
| 10219 |
+
"grad_norm": 0.04070388153195381,
|
| 10220 |
+
"learning_rate": 7.507762081928327e-07,
|
| 10221 |
+
"loss": 12.3711,
|
| 10222 |
+
"step": 1459
|
| 10223 |
+
},
|
| 10224 |
+
{
|
| 10225 |
+
"epoch": 0.5515678126180582,
|
| 10226 |
+
"grad_norm": 0.04098959267139435,
|
| 10227 |
+
"learning_rate": 7.33257850721647e-07,
|
| 10228 |
+
"loss": 12.3686,
|
| 10229 |
+
"step": 1460
|
| 10230 |
+
},
|
| 10231 |
+
{
|
| 10232 |
+
"epoch": 0.5519455987910843,
|
| 10233 |
+
"grad_norm": 0.03851509839296341,
|
| 10234 |
+
"learning_rate": 7.159447872234881e-07,
|
| 10235 |
+
"loss": 12.365,
|
| 10236 |
+
"step": 1461
|
| 10237 |
+
},
|
| 10238 |
+
{
|
| 10239 |
+
"epoch": 0.5523233849641103,
|
| 10240 |
+
"grad_norm": 0.027653610333800316,
|
| 10241 |
+
"learning_rate": 6.988370898417029e-07,
|
| 10242 |
+
"loss": 12.374,
|
| 10243 |
+
"step": 1462
|
| 10244 |
+
},
|
| 10245 |
+
{
|
| 10246 |
+
"epoch": 0.5527011711371363,
|
| 10247 |
+
"grad_norm": 0.028537223115563393,
|
| 10248 |
+
"learning_rate": 6.819348298638839e-07,
|
| 10249 |
+
"loss": 12.3747,
|
| 10250 |
+
"step": 1463
|
| 10251 |
+
},
|
| 10252 |
+
{
|
| 10253 |
+
"epoch": 0.5530789573101624,
|
| 10254 |
+
"grad_norm": 0.0324607715010643,
|
| 10255 |
+
"learning_rate": 6.652380777215583e-07,
|
| 10256 |
+
"loss": 12.371,
|
| 10257 |
+
"step": 1464
|
| 10258 |
+
},
|
| 10259 |
+
{
|
| 10260 |
+
"epoch": 0.5534567434831885,
|
| 10261 |
+
"grad_norm": 0.02856702171266079,
|
| 10262 |
+
"learning_rate": 6.487469029899052e-07,
|
| 10263 |
+
"loss": 12.3648,
|
| 10264 |
+
"step": 1465
|
| 10265 |
+
},
|
| 10266 |
+
{
|
| 10267 |
+
"epoch": 0.5538345296562146,
|
| 10268 |
+
"grad_norm": 0.03662535548210144,
|
| 10269 |
+
"learning_rate": 6.324613743874774e-07,
|
| 10270 |
+
"loss": 12.369,
|
| 10271 |
+
"step": 1466
|
| 10272 |
+
},
|
| 10273 |
+
{
|
| 10274 |
+
"epoch": 0.5542123158292407,
|
| 10275 |
+
"grad_norm": 0.03720373660326004,
|
| 10276 |
+
"learning_rate": 6.163815597758916e-07,
|
| 10277 |
+
"loss": 12.3691,
|
| 10278 |
+
"step": 1467
|
| 10279 |
+
},
|
| 10280 |
+
{
|
| 10281 |
+
"epoch": 0.5545901020022668,
|
| 10282 |
+
"grad_norm": 0.03894321992993355,
|
| 10283 |
+
"learning_rate": 6.005075261595494e-07,
|
| 10284 |
+
"loss": 12.3679,
|
| 10285 |
+
"step": 1468
|
| 10286 |
+
},
|
| 10287 |
+
{
|
| 10288 |
+
"epoch": 0.5549678881752927,
|
| 10289 |
+
"grad_norm": 0.03958560526371002,
|
| 10290 |
+
"learning_rate": 5.848393396853891e-07,
|
| 10291 |
+
"loss": 12.3729,
|
| 10292 |
+
"step": 1469
|
| 10293 |
+
},
|
| 10294 |
+
{
|
| 10295 |
+
"epoch": 0.5553456743483188,
|
| 10296 |
+
"grad_norm": 0.03394071385264397,
|
| 10297 |
+
"learning_rate": 5.693770656425568e-07,
|
| 10298 |
+
"loss": 12.371,
|
| 10299 |
+
"step": 1470
|
| 10300 |
+
},
|
| 10301 |
+
{
|
| 10302 |
+
"epoch": 0.5557234605213449,
|
| 10303 |
+
"grad_norm": 0.0403885617852211,
|
| 10304 |
+
"learning_rate": 5.541207684621907e-07,
|
| 10305 |
+
"loss": 12.3771,
|
| 10306 |
+
"step": 1471
|
| 10307 |
+
},
|
| 10308 |
+
{
|
| 10309 |
+
"epoch": 0.556101246694371,
|
| 10310 |
+
"grad_norm": 0.03610556200146675,
|
| 10311 |
+
"learning_rate": 5.390705117171047e-07,
|
| 10312 |
+
"loss": 12.3619,
|
| 10313 |
+
"step": 1472
|
| 10314 |
+
},
|
| 10315 |
+
{
|
| 10316 |
+
"epoch": 0.5564790328673971,
|
| 10317 |
+
"grad_norm": 0.03496142476797104,
|
| 10318 |
+
"learning_rate": 5.242263581215545e-07,
|
| 10319 |
+
"loss": 12.3744,
|
| 10320 |
+
"step": 1473
|
| 10321 |
+
},
|
| 10322 |
+
{
|
| 10323 |
+
"epoch": 0.5568568190404232,
|
| 10324 |
+
"grad_norm": 0.032318051904439926,
|
| 10325 |
+
"learning_rate": 5.095883695309722e-07,
|
| 10326 |
+
"loss": 12.3707,
|
| 10327 |
+
"step": 1474
|
| 10328 |
+
},
|
| 10329 |
+
{
|
| 10330 |
+
"epoch": 0.5572346052134491,
|
| 10331 |
+
"grad_norm": 0.041582122445106506,
|
| 10332 |
+
"learning_rate": 4.95156606941688e-07,
|
| 10333 |
+
"loss": 12.3724,
|
| 10334 |
+
"step": 1475
|
| 10335 |
+
},
|
| 10336 |
+
{
|
| 10337 |
+
"epoch": 0.5576123913864752,
|
| 10338 |
+
"grad_norm": 0.04513188824057579,
|
| 10339 |
+
"learning_rate": 4.809311304906972e-07,
|
| 10340 |
+
"loss": 12.3598,
|
| 10341 |
+
"step": 1476
|
| 10342 |
+
},
|
| 10343 |
+
{
|
| 10344 |
+
"epoch": 0.5579901775595013,
|
| 10345 |
+
"grad_norm": 0.0343400277197361,
|
| 10346 |
+
"learning_rate": 4.6691199945541054e-07,
|
| 10347 |
+
"loss": 12.3681,
|
| 10348 |
+
"step": 1477
|
| 10349 |
+
},
|
| 10350 |
+
{
|
| 10351 |
+
"epoch": 0.5583679637325274,
|
| 10352 |
+
"grad_norm": 0.03323856741189957,
|
| 10353 |
+
"learning_rate": 4.530992722533878e-07,
|
| 10354 |
+
"loss": 12.3615,
|
| 10355 |
+
"step": 1478
|
| 10356 |
+
},
|
| 10357 |
+
{
|
| 10358 |
+
"epoch": 0.5587457499055535,
|
| 10359 |
+
"grad_norm": 0.036309778690338135,
|
| 10360 |
+
"learning_rate": 4.394930064421099e-07,
|
| 10361 |
+
"loss": 12.3667,
|
| 10362 |
+
"step": 1479
|
| 10363 |
+
},
|
| 10364 |
+
{
|
| 10365 |
+
"epoch": 0.5591235360785796,
|
| 10366 |
+
"grad_norm": 0.03436039388179779,
|
| 10367 |
+
"learning_rate": 4.2609325871874603e-07,
|
| 10368 |
+
"loss": 12.3606,
|
| 10369 |
+
"step": 1480
|
| 10370 |
+
},
|
| 10371 |
+
{
|
| 10372 |
+
"epoch": 0.5595013222516055,
|
| 10373 |
+
"grad_norm": 0.043769899755716324,
|
| 10374 |
+
"learning_rate": 4.129000849198872e-07,
|
| 10375 |
+
"loss": 12.3743,
|
| 10376 |
+
"step": 1481
|
| 10377 |
+
},
|
| 10378 |
+
{
|
| 10379 |
+
"epoch": 0.5598791084246316,
|
| 10380 |
+
"grad_norm": 0.03071453608572483,
|
| 10381 |
+
"learning_rate": 3.999135400213461e-07,
|
| 10382 |
+
"loss": 12.3697,
|
| 10383 |
+
"step": 1482
|
| 10384 |
+
},
|
| 10385 |
+
{
|
| 10386 |
+
"epoch": 0.5602568945976577,
|
| 10387 |
+
"grad_norm": 0.039880815893411636,
|
| 10388 |
+
"learning_rate": 3.8713367813791867e-07,
|
| 10389 |
+
"loss": 12.3648,
|
| 10390 |
+
"step": 1483
|
| 10391 |
+
},
|
| 10392 |
+
{
|
| 10393 |
+
"epoch": 0.5606346807706838,
|
| 10394 |
+
"grad_norm": 0.042634375393390656,
|
| 10395 |
+
"learning_rate": 3.745605525231399e-07,
|
| 10396 |
+
"loss": 12.3593,
|
| 10397 |
+
"step": 1484
|
| 10398 |
+
},
|
| 10399 |
+
{
|
| 10400 |
+
"epoch": 0.5610124669437099,
|
| 10401 |
+
"grad_norm": 0.036256734281778336,
|
| 10402 |
+
"learning_rate": 3.6219421556909495e-07,
|
| 10403 |
+
"loss": 12.3669,
|
| 10404 |
+
"step": 1485
|
| 10405 |
+
},
|
| 10406 |
+
{
|
| 10407 |
+
"epoch": 0.561390253116736,
|
| 10408 |
+
"grad_norm": 0.037956301122903824,
|
| 10409 |
+
"learning_rate": 3.5003471880616925e-07,
|
| 10410 |
+
"loss": 12.376,
|
| 10411 |
+
"step": 1486
|
| 10412 |
+
},
|
| 10413 |
+
{
|
| 10414 |
+
"epoch": 0.561768039289762,
|
| 10415 |
+
"grad_norm": 0.0457519106566906,
|
| 10416 |
+
"learning_rate": 3.380821129028489e-07,
|
| 10417 |
+
"loss": 12.367,
|
| 10418 |
+
"step": 1487
|
| 10419 |
+
},
|
| 10420 |
+
{
|
| 10421 |
+
"epoch": 0.562145825462788,
|
| 10422 |
+
"grad_norm": 0.04063735157251358,
|
| 10423 |
+
"learning_rate": 3.2633644766552064e-07,
|
| 10424 |
+
"loss": 12.3653,
|
| 10425 |
+
"step": 1488
|
| 10426 |
+
},
|
| 10427 |
+
{
|
| 10428 |
+
"epoch": 0.5625236116358141,
|
| 10429 |
+
"grad_norm": 0.04077863693237305,
|
| 10430 |
+
"learning_rate": 3.147977720382334e-07,
|
| 10431 |
+
"loss": 12.3769,
|
| 10432 |
+
"step": 1489
|
| 10433 |
+
},
|
| 10434 |
+
{
|
| 10435 |
+
"epoch": 0.5629013978088402,
|
| 10436 |
+
"grad_norm": 0.04038354009389877,
|
| 10437 |
+
"learning_rate": 3.034661341025258e-07,
|
| 10438 |
+
"loss": 12.3686,
|
| 10439 |
+
"step": 1490
|
| 10440 |
+
},
|
| 10441 |
+
{
|
| 10442 |
+
"epoch": 0.5632791839818663,
|
| 10443 |
+
"grad_norm": 0.03477098047733307,
|
| 10444 |
+
"learning_rate": 2.923415810772101e-07,
|
| 10445 |
+
"loss": 12.3716,
|
| 10446 |
+
"step": 1491
|
| 10447 |
+
},
|
| 10448 |
+
{
|
| 10449 |
+
"epoch": 0.5636569701548924,
|
| 10450 |
+
"grad_norm": 0.042445071041584015,
|
| 10451 |
+
"learning_rate": 2.8142415931817214e-07,
|
| 10452 |
+
"loss": 12.3724,
|
| 10453 |
+
"step": 1492
|
| 10454 |
+
},
|
| 10455 |
+
{
|
| 10456 |
+
"epoch": 0.5640347563279184,
|
| 10457 |
+
"grad_norm": 0.04494417831301689,
|
| 10458 |
+
"learning_rate": 2.7071391431818806e-07,
|
| 10459 |
+
"loss": 12.368,
|
| 10460 |
+
"step": 1493
|
| 10461 |
+
},
|
| 10462 |
+
{
|
| 10463 |
+
"epoch": 0.5644125425009444,
|
| 10464 |
+
"grad_norm": 0.050600260496139526,
|
| 10465 |
+
"learning_rate": 2.602108907067358e-07,
|
| 10466 |
+
"loss": 12.3698,
|
| 10467 |
+
"step": 1494
|
| 10468 |
+
},
|
| 10469 |
+
{
|
| 10470 |
+
"epoch": 0.5647903286739705,
|
| 10471 |
+
"grad_norm": 0.046449873596429825,
|
| 10472 |
+
"learning_rate": 2.499151322498006e-07,
|
| 10473 |
+
"loss": 12.3713,
|
| 10474 |
+
"step": 1495
|
| 10475 |
+
},
|
| 10476 |
+
{
|
| 10477 |
+
"epoch": 0.5651681148469966,
|
| 10478 |
+
"grad_norm": 0.05416286364197731,
|
| 10479 |
+
"learning_rate": 2.3982668184968636e-07,
|
| 10480 |
+
"loss": 12.376,
|
| 10481 |
+
"step": 1496
|
| 10482 |
+
},
|
| 10483 |
+
{
|
| 10484 |
+
"epoch": 0.5655459010200227,
|
| 10485 |
+
"grad_norm": 0.06181178241968155,
|
| 10486 |
+
"learning_rate": 2.2994558154486588e-07,
|
| 10487 |
+
"loss": 12.3673,
|
| 10488 |
+
"step": 1497
|
| 10489 |
+
},
|
| 10490 |
+
{
|
| 10491 |
+
"epoch": 0.5659236871930488,
|
| 10492 |
+
"grad_norm": 0.07762554287910461,
|
| 10493 |
+
"learning_rate": 2.202718725097641e-07,
|
| 10494 |
+
"loss": 12.364,
|
| 10495 |
+
"step": 1498
|
| 10496 |
+
},
|
| 10497 |
+
{
|
| 10498 |
+
"epoch": 0.5663014733660748,
|
| 10499 |
+
"grad_norm": 0.08430846035480499,
|
| 10500 |
+
"learning_rate": 2.1080559505462505e-07,
|
| 10501 |
+
"loss": 12.3749,
|
| 10502 |
+
"step": 1499
|
| 10503 |
+
},
|
| 10504 |
+
{
|
| 10505 |
+
"epoch": 0.5666792595391009,
|
| 10506 |
+
"grad_norm": 0.12796476483345032,
|
| 10507 |
+
"learning_rate": 2.015467886253286e-07,
|
| 10508 |
+
"loss": 12.3818,
|
| 10509 |
+
"step": 1500
|
| 10510 |
+
},
|
| 10511 |
+
{
|
| 10512 |
+
"epoch": 0.5670570457121269,
|
| 10513 |
+
"grad_norm": 0.04819419980049133,
|
| 10514 |
+
"learning_rate": 1.9249549180320737e-07,
|
| 10515 |
+
"loss": 12.3688,
|
| 10516 |
+
"step": 1501
|
| 10517 |
+
},
|
| 10518 |
+
{
|
| 10519 |
+
"epoch": 0.567434831885153,
|
| 10520 |
+
"grad_norm": 0.04547456279397011,
|
| 10521 |
+
"learning_rate": 1.8365174230492998e-07,
|
| 10522 |
+
"loss": 12.3766,
|
| 10523 |
+
"step": 1502
|
| 10524 |
+
},
|
| 10525 |
+
{
|
| 10526 |
+
"epoch": 0.5678126180581791,
|
| 10527 |
+
"grad_norm": 0.04163209721446037,
|
| 10528 |
+
"learning_rate": 1.7501557698229033e-07,
|
| 10529 |
+
"loss": 12.3779,
|
| 10530 |
+
"step": 1503
|
| 10531 |
+
},
|
| 10532 |
+
{
|
| 10533 |
+
"epoch": 0.5681904042312051,
|
| 10534 |
+
"grad_norm": 0.04335664585232735,
|
| 10535 |
+
"learning_rate": 1.665870318221019e-07,
|
| 10536 |
+
"loss": 12.3705,
|
| 10537 |
+
"step": 1504
|
| 10538 |
+
},
|
| 10539 |
+
{
|
| 10540 |
+
"epoch": 0.5685681904042312,
|
| 10541 |
+
"grad_norm": 0.04164076969027519,
|
| 10542 |
+
"learning_rate": 1.5836614194602027e-07,
|
| 10543 |
+
"loss": 12.3701,
|
| 10544 |
+
"step": 1505
|
| 10545 |
+
},
|
| 10546 |
+
{
|
| 10547 |
+
"epoch": 0.5689459765772573,
|
| 10548 |
+
"grad_norm": 0.040243785828351974,
|
| 10549 |
+
"learning_rate": 1.503529416103988e-07,
|
| 10550 |
+
"loss": 12.3806,
|
| 10551 |
+
"step": 1506
|
| 10552 |
+
},
|
| 10553 |
+
{
|
| 10554 |
+
"epoch": 0.5693237627502833,
|
| 10555 |
+
"grad_norm": 0.048739392310380936,
|
| 10556 |
+
"learning_rate": 1.4254746420615527e-07,
|
| 10557 |
+
"loss": 12.374,
|
| 10558 |
+
"step": 1507
|
| 10559 |
+
},
|
| 10560 |
+
{
|
| 10561 |
+
"epoch": 0.5697015489233094,
|
| 10562 |
+
"grad_norm": 0.0307659562677145,
|
| 10563 |
+
"learning_rate": 1.3494974225863322e-07,
|
| 10564 |
+
"loss": 12.3795,
|
| 10565 |
+
"step": 1508
|
| 10566 |
+
},
|
| 10567 |
+
{
|
| 10568 |
+
"epoch": 0.5700793350963355,
|
| 10569 |
+
"grad_norm": 0.0348367840051651,
|
| 10570 |
+
"learning_rate": 1.27559807427452e-07,
|
| 10571 |
+
"loss": 12.3701,
|
| 10572 |
+
"step": 1509
|
| 10573 |
+
},
|
| 10574 |
+
{
|
| 10575 |
+
"epoch": 0.5704571212693615,
|
| 10576 |
+
"grad_norm": 0.04481798782944679,
|
| 10577 |
+
"learning_rate": 1.2037769050639024e-07,
|
| 10578 |
+
"loss": 12.3769,
|
| 10579 |
+
"step": 1510
|
| 10580 |
+
},
|
| 10581 |
+
{
|
| 10582 |
+
"epoch": 0.5708349074423876,
|
| 10583 |
+
"grad_norm": 0.04808036610484123,
|
| 10584 |
+
"learning_rate": 1.1340342142325267e-07,
|
| 10585 |
+
"loss": 12.3664,
|
| 10586 |
+
"step": 1511
|
| 10587 |
+
},
|
| 10588 |
+
{
|
| 10589 |
+
"epoch": 0.5712126936154137,
|
| 10590 |
+
"grad_norm": 0.037592723965644836,
|
| 10591 |
+
"learning_rate": 1.0663702923973673e-07,
|
| 10592 |
+
"loss": 12.3692,
|
| 10593 |
+
"step": 1512
|
| 10594 |
+
},
|
| 10595 |
+
{
|
| 10596 |
+
"epoch": 0.5715904797884398,
|
| 10597 |
+
"grad_norm": 0.04113468527793884,
|
| 10598 |
+
"learning_rate": 1.0007854215132728e-07,
|
| 10599 |
+
"loss": 12.3699,
|
| 10600 |
+
"step": 1513
|
| 10601 |
+
},
|
| 10602 |
+
{
|
| 10603 |
+
"epoch": 0.5719682659614658,
|
| 10604 |
+
"grad_norm": 0.029787473380565643,
|
| 10605 |
+
"learning_rate": 9.372798748716883e-08,
|
| 10606 |
+
"loss": 12.3756,
|
| 10607 |
+
"step": 1514
|
| 10608 |
+
},
|
| 10609 |
+
{
|
| 10610 |
+
"epoch": 0.5723460521344919,
|
| 10611 |
+
"grad_norm": 0.03630399331450462,
|
| 10612 |
+
"learning_rate": 8.75853917099545e-08,
|
| 10613 |
+
"loss": 12.3705,
|
| 10614 |
+
"step": 1515
|
| 10615 |
+
},
|
| 10616 |
+
{
|
| 10617 |
+
"epoch": 0.5727238383075179,
|
| 10618 |
+
"grad_norm": 0.032823652029037476,
|
| 10619 |
+
"learning_rate": 8.165078041580399e-08,
|
| 10620 |
+
"loss": 12.3706,
|
| 10621 |
+
"step": 1516
|
| 10622 |
+
},
|
| 10623 |
+
{
|
| 10624 |
+
"epoch": 0.573101624480544,
|
| 10625 |
+
"grad_norm": 0.056236084550619125,
|
| 10626 |
+
"learning_rate": 7.59241783341913e-08,
|
| 10627 |
+
"loss": 12.3624,
|
| 10628 |
+
"step": 1517
|
| 10629 |
+
},
|
| 10630 |
+
{
|
| 10631 |
+
"epoch": 0.5734794106535701,
|
| 10632 |
+
"grad_norm": 0.03066546469926834,
|
| 10633 |
+
"learning_rate": 7.040560932778939e-08,
|
| 10634 |
+
"loss": 12.3657,
|
| 10635 |
+
"step": 1518
|
| 10636 |
+
},
|
| 10637 |
+
{
|
| 10638 |
+
"epoch": 0.5738571968265962,
|
| 10639 |
+
"grad_norm": 0.028224581852555275,
|
| 10640 |
+
"learning_rate": 6.509509639242572e-08,
|
| 10641 |
+
"loss": 12.3712,
|
| 10642 |
+
"step": 1519
|
| 10643 |
+
},
|
| 10644 |
+
{
|
| 10645 |
+
"epoch": 0.5742349829996222,
|
| 10646 |
+
"grad_norm": 0.040541455149650574,
|
| 10647 |
+
"learning_rate": 5.999266165694905e-08,
|
| 10648 |
+
"loss": 12.3674,
|
| 10649 |
+
"step": 1520
|
| 10650 |
+
},
|
| 10651 |
+
{
|
| 10652 |
+
"epoch": 0.5746127691726483,
|
| 10653 |
+
"grad_norm": 0.03858514502644539,
|
| 10654 |
+
"learning_rate": 5.509832638314061e-08,
|
| 10655 |
+
"loss": 12.3672,
|
| 10656 |
+
"step": 1521
|
| 10657 |
+
},
|
| 10658 |
+
{
|
| 10659 |
+
"epoch": 0.5749905553456743,
|
| 10660 |
+
"grad_norm": 0.03187035024166107,
|
| 10661 |
+
"learning_rate": 5.041211096565301e-08,
|
| 10662 |
+
"loss": 12.367,
|
| 10663 |
+
"step": 1522
|
| 10664 |
+
},
|
| 10665 |
+
{
|
| 10666 |
+
"epoch": 0.5753683415187004,
|
| 10667 |
+
"grad_norm": 0.03534197807312012,
|
| 10668 |
+
"learning_rate": 4.5934034931882644e-08,
|
| 10669 |
+
"loss": 12.3738,
|
| 10670 |
+
"step": 1523
|
| 10671 |
+
},
|
| 10672 |
+
{
|
| 10673 |
+
"epoch": 0.5757461276917265,
|
| 10674 |
+
"grad_norm": 0.05696889013051987,
|
| 10675 |
+
"learning_rate": 4.166411694193073e-08,
|
| 10676 |
+
"loss": 12.3633,
|
| 10677 |
+
"step": 1524
|
| 10678 |
+
},
|
| 10679 |
+
{
|
| 10680 |
+
"epoch": 0.5761239138647526,
|
| 10681 |
+
"grad_norm": 0.0424017570912838,
|
| 10682 |
+
"learning_rate": 3.760237478849793e-08,
|
| 10683 |
+
"loss": 12.3683,
|
| 10684 |
+
"step": 1525
|
| 10685 |
+
},
|
| 10686 |
+
{
|
| 10687 |
+
"epoch": 0.5765017000377786,
|
| 10688 |
+
"grad_norm": 0.030759789049625397,
|
| 10689 |
+
"learning_rate": 3.374882539681767e-08,
|
| 10690 |
+
"loss": 12.3768,
|
| 10691 |
+
"step": 1526
|
| 10692 |
+
},
|
| 10693 |
+
{
|
| 10694 |
+
"epoch": 0.5768794862108046,
|
| 10695 |
+
"grad_norm": 0.034990064799785614,
|
| 10696 |
+
"learning_rate": 3.010348482458403e-08,
|
| 10697 |
+
"loss": 12.3652,
|
| 10698 |
+
"step": 1527
|
| 10699 |
+
},
|
| 10700 |
+
{
|
| 10701 |
+
"epoch": 0.5772572723838307,
|
| 10702 |
+
"grad_norm": 0.037047334015369415,
|
| 10703 |
+
"learning_rate": 2.666636826189617e-08,
|
| 10704 |
+
"loss": 12.3693,
|
| 10705 |
+
"step": 1528
|
| 10706 |
+
},
|
| 10707 |
+
{
|
| 10708 |
+
"epoch": 0.5776350585568568,
|
| 10709 |
+
"grad_norm": 0.03725414723157883,
|
| 10710 |
+
"learning_rate": 2.3437490031180676e-08,
|
| 10711 |
+
"loss": 12.3682,
|
| 10712 |
+
"step": 1529
|
| 10713 |
+
},
|
| 10714 |
+
{
|
| 10715 |
+
"epoch": 0.5780128447298829,
|
| 10716 |
+
"grad_norm": 0.037986818701028824,
|
| 10717 |
+
"learning_rate": 2.0416863587141566e-08,
|
| 10718 |
+
"loss": 12.373,
|
| 10719 |
+
"step": 1530
|
| 10720 |
+
},
|
| 10721 |
+
{
|
| 10722 |
+
"epoch": 0.578390630902909,
|
| 10723 |
+
"grad_norm": 0.04022598639130592,
|
| 10724 |
+
"learning_rate": 1.760450151668258e-08,
|
| 10725 |
+
"loss": 12.3734,
|
| 10726 |
+
"step": 1531
|
| 10727 |
+
},
|
| 10728 |
+
{
|
| 10729 |
+
"epoch": 0.5787684170759351,
|
| 10730 |
+
"grad_norm": 0.050697073340415955,
|
| 10731 |
+
"learning_rate": 1.5000415538901636e-08,
|
| 10732 |
+
"loss": 12.3647,
|
| 10733 |
+
"step": 1532
|
| 10734 |
+
},
|
| 10735 |
+
{
|
| 10736 |
+
"epoch": 0.579146203248961,
|
| 10737 |
+
"grad_norm": 0.05417259410023689,
|
| 10738 |
+
"learning_rate": 1.2604616504985344e-08,
|
| 10739 |
+
"loss": 12.3676,
|
| 10740 |
+
"step": 1533
|
| 10741 |
+
},
|
| 10742 |
+
{
|
| 10743 |
+
"epoch": 0.5795239894219871,
|
| 10744 |
+
"grad_norm": 0.04395093396306038,
|
| 10745 |
+
"learning_rate": 1.0417114398209027e-08,
|
| 10746 |
+
"loss": 12.3633,
|
| 10747 |
+
"step": 1534
|
| 10748 |
+
},
|
| 10749 |
+
{
|
| 10750 |
+
"epoch": 0.5799017755950132,
|
| 10751 |
+
"grad_norm": 0.047466449439525604,
|
| 10752 |
+
"learning_rate": 8.437918333864536e-09,
|
| 10753 |
+
"loss": 12.3668,
|
| 10754 |
+
"step": 1535
|
| 10755 |
+
},
|
| 10756 |
+
{
|
| 10757 |
+
"epoch": 0.5802795617680393,
|
| 10758 |
+
"grad_norm": 0.03220509737730026,
|
| 10759 |
+
"learning_rate": 6.66703655923806e-09,
|
| 10760 |
+
"loss": 12.3653,
|
| 10761 |
+
"step": 1536
|
| 10762 |
+
},
|
| 10763 |
+
{
|
| 10764 |
+
"epoch": 0.5806573479410654,
|
| 10765 |
+
"grad_norm": 0.0401943065226078,
|
| 10766 |
+
"learning_rate": 5.104476453582363e-09,
|
| 10767 |
+
"loss": 12.365,
|
| 10768 |
+
"step": 1537
|
| 10769 |
+
},
|
| 10770 |
+
{
|
| 10771 |
+
"epoch": 0.5810351341140915,
|
| 10772 |
+
"grad_norm": 0.03627855330705643,
|
| 10773 |
+
"learning_rate": 3.750244528066826e-09,
|
| 10774 |
+
"loss": 12.3664,
|
| 10775 |
+
"step": 1538
|
| 10776 |
+
},
|
| 10777 |
+
{
|
| 10778 |
+
"epoch": 0.5814129202871174,
|
| 10779 |
+
"grad_norm": 0.03609207645058632,
|
| 10780 |
+
"learning_rate": 2.604346425755244e-09,
|
| 10781 |
+
"loss": 12.3618,
|
| 10782 |
+
"step": 1539
|
| 10783 |
+
},
|
| 10784 |
+
{
|
| 10785 |
+
"epoch": 0.5817907064601435,
|
| 10786 |
+
"grad_norm": 0.04659217223525047,
|
| 10787 |
+
"learning_rate": 1.6667869215947208e-09,
|
| 10788 |
+
"loss": 12.3656,
|
| 10789 |
+
"step": 1540
|
| 10790 |
+
},
|
| 10791 |
+
{
|
| 10792 |
+
"epoch": 0.5821684926331696,
|
| 10793 |
+
"grad_norm": 0.03406490385532379,
|
| 10794 |
+
"learning_rate": 9.375699223879153e-10,
|
| 10795 |
+
"loss": 12.3705,
|
| 10796 |
+
"step": 1541
|
| 10797 |
+
},
|
| 10798 |
+
{
|
| 10799 |
+
"epoch": 0.5825462788061957,
|
| 10800 |
+
"grad_norm": 0.04274613410234451,
|
| 10801 |
+
"learning_rate": 4.1669846677638847e-10,
|
| 10802 |
+
"loss": 12.3621,
|
| 10803 |
+
"step": 1542
|
| 10804 |
+
},
|
| 10805 |
+
{
|
| 10806 |
+
"epoch": 0.5829240649792218,
|
| 10807 |
+
"grad_norm": 0.04796082153916359,
|
| 10808 |
+
"learning_rate": 1.0417472521839778e-10,
|
| 10809 |
+
"loss": 12.3658,
|
| 10810 |
+
"step": 1543
|
| 10811 |
+
},
|
| 10812 |
+
{
|
| 10813 |
+
"epoch": 0.5833018511522479,
|
| 10814 |
+
"grad_norm": 0.05695313960313797,
|
| 10815 |
+
"learning_rate": 0.0,
|
| 10816 |
+
"loss": 12.369,
|
| 10817 |
+
"step": 1544
|
| 10818 |
}
|
| 10819 |
],
|
| 10820 |
"logging_steps": 1,
|
|
|
|
| 10829 |
"should_evaluate": false,
|
| 10830 |
"should_log": false,
|
| 10831 |
"should_save": true,
|
| 10832 |
+
"should_training_stop": true
|
| 10833 |
},
|
| 10834 |
"attributes": {}
|
| 10835 |
}
|
| 10836 |
},
|
| 10837 |
+
"total_flos": 727424434176.0,
|
| 10838 |
"train_batch_size": 4,
|
| 10839 |
"trial_name": null,
|
| 10840 |
"trial_params": null
|