Training in progress, step 1447, 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 80792096
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f093a323af6c6b18fe218fd37e6aa2a686c71bf0b4bafe5d0a0512e02e64279f
|
| 3 |
size 80792096
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 41460084
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e1f9819ca68bfd335d881d0cee9cb5a88ada0e81a934bc426c0d57343c620827
|
| 3 |
size 41460084
|
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 sha256:8a16950e15a2023892910e4f74ea1f9c06da8a3ddb1e92bf5c3ec6ff9e8f0f81
|
| 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:e830029559dd1ac49acac9c5cac1f6ae0262ca4b3799ec74355143a95eb7d6cf
|
| 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,
|
|
@@ -7609,6 +7609,2533 @@
|
|
| 7609 |
"learning_rate": 1.4683195111226378e-05,
|
| 7610 |
"loss": 1.3353,
|
| 7611 |
"step": 1086
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7612 |
}
|
| 7613 |
],
|
| 7614 |
"logging_steps": 1,
|
|
@@ -7623,12 +10150,12 @@
|
|
| 7623 |
"should_evaluate": false,
|
| 7624 |
"should_log": false,
|
| 7625 |
"should_save": true,
|
| 7626 |
-
"should_training_stop":
|
| 7627 |
},
|
| 7628 |
"attributes": {}
|
| 7629 |
}
|
| 7630 |
},
|
| 7631 |
-
"total_flos":
|
| 7632 |
"train_batch_size": 4,
|
| 7633 |
"trial_name": null,
|
| 7634 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.06851001373041049,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 1447,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 7609 |
"learning_rate": 1.4683195111226378e-05,
|
| 7610 |
"loss": 1.3353,
|
| 7611 |
"step": 1086
|
| 7612 |
+
},
|
| 7613 |
+
{
|
| 7614 |
+
"epoch": 0.05146536622319019,
|
| 7615 |
+
"grad_norm": 1.421676754951477,
|
| 7616 |
+
"learning_rate": 1.460616872323236e-05,
|
| 7617 |
+
"loss": 1.559,
|
| 7618 |
+
"step": 1087
|
| 7619 |
+
},
|
| 7620 |
+
{
|
| 7621 |
+
"epoch": 0.0515127124662658,
|
| 7622 |
+
"grad_norm": 1.1935399770736694,
|
| 7623 |
+
"learning_rate": 1.4529310330343965e-05,
|
| 7624 |
+
"loss": 1.3224,
|
| 7625 |
+
"step": 1088
|
| 7626 |
+
},
|
| 7627 |
+
{
|
| 7628 |
+
"epoch": 0.05156005870934141,
|
| 7629 |
+
"grad_norm": 1.486861228942871,
|
| 7630 |
+
"learning_rate": 1.4452620297365804e-05,
|
| 7631 |
+
"loss": 1.4509,
|
| 7632 |
+
"step": 1089
|
| 7633 |
+
},
|
| 7634 |
+
{
|
| 7635 |
+
"epoch": 0.05160740495241702,
|
| 7636 |
+
"grad_norm": 1.250034213066101,
|
| 7637 |
+
"learning_rate": 1.4376098988303405e-05,
|
| 7638 |
+
"loss": 1.3847,
|
| 7639 |
+
"step": 1090
|
| 7640 |
+
},
|
| 7641 |
+
{
|
| 7642 |
+
"epoch": 0.051654751195492635,
|
| 7643 |
+
"grad_norm": 1.556272268295288,
|
| 7644 |
+
"learning_rate": 1.4299746766361477e-05,
|
| 7645 |
+
"loss": 1.5389,
|
| 7646 |
+
"step": 1091
|
| 7647 |
+
},
|
| 7648 |
+
{
|
| 7649 |
+
"epoch": 0.05170209743856825,
|
| 7650 |
+
"grad_norm": 1.4225186109542847,
|
| 7651 |
+
"learning_rate": 1.4223563993942124e-05,
|
| 7652 |
+
"loss": 1.5265,
|
| 7653 |
+
"step": 1092
|
| 7654 |
+
},
|
| 7655 |
+
{
|
| 7656 |
+
"epoch": 0.05174944368164386,
|
| 7657 |
+
"grad_norm": 1.3785078525543213,
|
| 7658 |
+
"learning_rate": 1.4147551032643192e-05,
|
| 7659 |
+
"loss": 1.2017,
|
| 7660 |
+
"step": 1093
|
| 7661 |
+
},
|
| 7662 |
+
{
|
| 7663 |
+
"epoch": 0.05179678992471947,
|
| 7664 |
+
"grad_norm": 1.3730214834213257,
|
| 7665 |
+
"learning_rate": 1.4071708243256498e-05,
|
| 7666 |
+
"loss": 1.1137,
|
| 7667 |
+
"step": 1094
|
| 7668 |
+
},
|
| 7669 |
+
{
|
| 7670 |
+
"epoch": 0.05184413616779508,
|
| 7671 |
+
"grad_norm": 1.321266531944275,
|
| 7672 |
+
"learning_rate": 1.3996035985766204e-05,
|
| 7673 |
+
"loss": 1.1409,
|
| 7674 |
+
"step": 1095
|
| 7675 |
+
},
|
| 7676 |
+
{
|
| 7677 |
+
"epoch": 0.051891482410870694,
|
| 7678 |
+
"grad_norm": 1.318037748336792,
|
| 7679 |
+
"learning_rate": 1.3920534619346997e-05,
|
| 7680 |
+
"loss": 1.1489,
|
| 7681 |
+
"step": 1096
|
| 7682 |
+
},
|
| 7683 |
+
{
|
| 7684 |
+
"epoch": 0.051938828653946306,
|
| 7685 |
+
"grad_norm": 1.3448675870895386,
|
| 7686 |
+
"learning_rate": 1.384520450236244e-05,
|
| 7687 |
+
"loss": 1.2181,
|
| 7688 |
+
"step": 1097
|
| 7689 |
+
},
|
| 7690 |
+
{
|
| 7691 |
+
"epoch": 0.05198617489702192,
|
| 7692 |
+
"grad_norm": 1.6022285223007202,
|
| 7693 |
+
"learning_rate": 1.3770045992363311e-05,
|
| 7694 |
+
"loss": 1.3204,
|
| 7695 |
+
"step": 1098
|
| 7696 |
+
},
|
| 7697 |
+
{
|
| 7698 |
+
"epoch": 0.05203352114009753,
|
| 7699 |
+
"grad_norm": 1.4788554906845093,
|
| 7700 |
+
"learning_rate": 1.3695059446085829e-05,
|
| 7701 |
+
"loss": 1.2685,
|
| 7702 |
+
"step": 1099
|
| 7703 |
+
},
|
| 7704 |
+
{
|
| 7705 |
+
"epoch": 0.05208086738317314,
|
| 7706 |
+
"grad_norm": 1.3918118476867676,
|
| 7707 |
+
"learning_rate": 1.3620245219449984e-05,
|
| 7708 |
+
"loss": 1.1012,
|
| 7709 |
+
"step": 1100
|
| 7710 |
+
},
|
| 7711 |
+
{
|
| 7712 |
+
"epoch": 0.052128213626248754,
|
| 7713 |
+
"grad_norm": 1.1865662336349487,
|
| 7714 |
+
"learning_rate": 1.354560366755791e-05,
|
| 7715 |
+
"loss": 1.5779,
|
| 7716 |
+
"step": 1101
|
| 7717 |
+
},
|
| 7718 |
+
{
|
| 7719 |
+
"epoch": 0.052175559869324366,
|
| 7720 |
+
"grad_norm": 1.157436490058899,
|
| 7721 |
+
"learning_rate": 1.3471135144692099e-05,
|
| 7722 |
+
"loss": 1.6593,
|
| 7723 |
+
"step": 1102
|
| 7724 |
+
},
|
| 7725 |
+
{
|
| 7726 |
+
"epoch": 0.05222290611239998,
|
| 7727 |
+
"grad_norm": 1.1785343885421753,
|
| 7728 |
+
"learning_rate": 1.3396840004313788e-05,
|
| 7729 |
+
"loss": 1.5321,
|
| 7730 |
+
"step": 1103
|
| 7731 |
+
},
|
| 7732 |
+
{
|
| 7733 |
+
"epoch": 0.05227025235547559,
|
| 7734 |
+
"grad_norm": 1.1837494373321533,
|
| 7735 |
+
"learning_rate": 1.3322718599061251e-05,
|
| 7736 |
+
"loss": 1.5987,
|
| 7737 |
+
"step": 1104
|
| 7738 |
+
},
|
| 7739 |
+
{
|
| 7740 |
+
"epoch": 0.0523175985985512,
|
| 7741 |
+
"grad_norm": 1.204405665397644,
|
| 7742 |
+
"learning_rate": 1.3248771280748174e-05,
|
| 7743 |
+
"loss": 1.5906,
|
| 7744 |
+
"step": 1105
|
| 7745 |
+
},
|
| 7746 |
+
{
|
| 7747 |
+
"epoch": 0.05236494484162681,
|
| 7748 |
+
"grad_norm": 1.2548047304153442,
|
| 7749 |
+
"learning_rate": 1.317499840036191e-05,
|
| 7750 |
+
"loss": 1.4687,
|
| 7751 |
+
"step": 1106
|
| 7752 |
+
},
|
| 7753 |
+
{
|
| 7754 |
+
"epoch": 0.05241229108470243,
|
| 7755 |
+
"grad_norm": 1.2715576887130737,
|
| 7756 |
+
"learning_rate": 1.310140030806184e-05,
|
| 7757 |
+
"loss": 1.6087,
|
| 7758 |
+
"step": 1107
|
| 7759 |
+
},
|
| 7760 |
+
{
|
| 7761 |
+
"epoch": 0.052459637327778044,
|
| 7762 |
+
"grad_norm": 1.2894080877304077,
|
| 7763 |
+
"learning_rate": 1.3027977353177779e-05,
|
| 7764 |
+
"loss": 1.5295,
|
| 7765 |
+
"step": 1108
|
| 7766 |
+
},
|
| 7767 |
+
{
|
| 7768 |
+
"epoch": 0.052506983570853656,
|
| 7769 |
+
"grad_norm": 1.4056642055511475,
|
| 7770 |
+
"learning_rate": 1.2954729884208211e-05,
|
| 7771 |
+
"loss": 1.4287,
|
| 7772 |
+
"step": 1109
|
| 7773 |
+
},
|
| 7774 |
+
{
|
| 7775 |
+
"epoch": 0.05255432981392927,
|
| 7776 |
+
"grad_norm": 1.352663516998291,
|
| 7777 |
+
"learning_rate": 1.2881658248818679e-05,
|
| 7778 |
+
"loss": 1.7946,
|
| 7779 |
+
"step": 1110
|
| 7780 |
+
},
|
| 7781 |
+
{
|
| 7782 |
+
"epoch": 0.05260167605700488,
|
| 7783 |
+
"grad_norm": 1.274722695350647,
|
| 7784 |
+
"learning_rate": 1.2808762793840201e-05,
|
| 7785 |
+
"loss": 1.3373,
|
| 7786 |
+
"step": 1111
|
| 7787 |
+
},
|
| 7788 |
+
{
|
| 7789 |
+
"epoch": 0.05264902230008049,
|
| 7790 |
+
"grad_norm": 1.205983281135559,
|
| 7791 |
+
"learning_rate": 1.2736043865267494e-05,
|
| 7792 |
+
"loss": 1.4965,
|
| 7793 |
+
"step": 1112
|
| 7794 |
+
},
|
| 7795 |
+
{
|
| 7796 |
+
"epoch": 0.052696368543156104,
|
| 7797 |
+
"grad_norm": 1.3179651498794556,
|
| 7798 |
+
"learning_rate": 1.2663501808257444e-05,
|
| 7799 |
+
"loss": 1.8323,
|
| 7800 |
+
"step": 1113
|
| 7801 |
+
},
|
| 7802 |
+
{
|
| 7803 |
+
"epoch": 0.052743714786231716,
|
| 7804 |
+
"grad_norm": 1.2150477170944214,
|
| 7805 |
+
"learning_rate": 1.2591136967127381e-05,
|
| 7806 |
+
"loss": 1.5991,
|
| 7807 |
+
"step": 1114
|
| 7808 |
+
},
|
| 7809 |
+
{
|
| 7810 |
+
"epoch": 0.05279106102930733,
|
| 7811 |
+
"grad_norm": 1.125106930732727,
|
| 7812 |
+
"learning_rate": 1.2518949685353559e-05,
|
| 7813 |
+
"loss": 1.3217,
|
| 7814 |
+
"step": 1115
|
| 7815 |
+
},
|
| 7816 |
+
{
|
| 7817 |
+
"epoch": 0.05283840727238294,
|
| 7818 |
+
"grad_norm": 1.0757575035095215,
|
| 7819 |
+
"learning_rate": 1.24469403055694e-05,
|
| 7820 |
+
"loss": 1.2343,
|
| 7821 |
+
"step": 1116
|
| 7822 |
+
},
|
| 7823 |
+
{
|
| 7824 |
+
"epoch": 0.05288575351545855,
|
| 7825 |
+
"grad_norm": 1.200031042098999,
|
| 7826 |
+
"learning_rate": 1.2375109169563915e-05,
|
| 7827 |
+
"loss": 1.5762,
|
| 7828 |
+
"step": 1117
|
| 7829 |
+
},
|
| 7830 |
+
{
|
| 7831 |
+
"epoch": 0.05293309975853416,
|
| 7832 |
+
"grad_norm": 1.3849049806594849,
|
| 7833 |
+
"learning_rate": 1.2303456618280141e-05,
|
| 7834 |
+
"loss": 1.5292,
|
| 7835 |
+
"step": 1118
|
| 7836 |
+
},
|
| 7837 |
+
{
|
| 7838 |
+
"epoch": 0.052980446001609775,
|
| 7839 |
+
"grad_norm": 1.3612500429153442,
|
| 7840 |
+
"learning_rate": 1.223198299181343e-05,
|
| 7841 |
+
"loss": 1.5294,
|
| 7842 |
+
"step": 1119
|
| 7843 |
+
},
|
| 7844 |
+
{
|
| 7845 |
+
"epoch": 0.05302779224468539,
|
| 7846 |
+
"grad_norm": 1.2534222602844238,
|
| 7847 |
+
"learning_rate": 1.2160688629409872e-05,
|
| 7848 |
+
"loss": 1.5581,
|
| 7849 |
+
"step": 1120
|
| 7850 |
+
},
|
| 7851 |
+
{
|
| 7852 |
+
"epoch": 0.053075138487761,
|
| 7853 |
+
"grad_norm": 1.2404296398162842,
|
| 7854 |
+
"learning_rate": 1.2089573869464738e-05,
|
| 7855 |
+
"loss": 1.5379,
|
| 7856 |
+
"step": 1121
|
| 7857 |
+
},
|
| 7858 |
+
{
|
| 7859 |
+
"epoch": 0.05312248473083661,
|
| 7860 |
+
"grad_norm": 1.2074594497680664,
|
| 7861 |
+
"learning_rate": 1.2018639049520769e-05,
|
| 7862 |
+
"loss": 1.2621,
|
| 7863 |
+
"step": 1122
|
| 7864 |
+
},
|
| 7865 |
+
{
|
| 7866 |
+
"epoch": 0.05316983097391222,
|
| 7867 |
+
"grad_norm": 1.2589749097824097,
|
| 7868 |
+
"learning_rate": 1.1947884506266655e-05,
|
| 7869 |
+
"loss": 1.6411,
|
| 7870 |
+
"step": 1123
|
| 7871 |
+
},
|
| 7872 |
+
{
|
| 7873 |
+
"epoch": 0.053217177216987835,
|
| 7874 |
+
"grad_norm": 1.301170825958252,
|
| 7875 |
+
"learning_rate": 1.1877310575535394e-05,
|
| 7876 |
+
"loss": 1.6083,
|
| 7877 |
+
"step": 1124
|
| 7878 |
+
},
|
| 7879 |
+
{
|
| 7880 |
+
"epoch": 0.05326452346006345,
|
| 7881 |
+
"grad_norm": 1.3040543794631958,
|
| 7882 |
+
"learning_rate": 1.1806917592302762e-05,
|
| 7883 |
+
"loss": 1.487,
|
| 7884 |
+
"step": 1125
|
| 7885 |
+
},
|
| 7886 |
+
{
|
| 7887 |
+
"epoch": 0.05331186970313906,
|
| 7888 |
+
"grad_norm": 1.2013649940490723,
|
| 7889 |
+
"learning_rate": 1.1736705890685623e-05,
|
| 7890 |
+
"loss": 1.5764,
|
| 7891 |
+
"step": 1126
|
| 7892 |
+
},
|
| 7893 |
+
{
|
| 7894 |
+
"epoch": 0.05335921594621467,
|
| 7895 |
+
"grad_norm": 1.2394853830337524,
|
| 7896 |
+
"learning_rate": 1.1666675803940408e-05,
|
| 7897 |
+
"loss": 1.5924,
|
| 7898 |
+
"step": 1127
|
| 7899 |
+
},
|
| 7900 |
+
{
|
| 7901 |
+
"epoch": 0.05340656218929028,
|
| 7902 |
+
"grad_norm": 1.3157364130020142,
|
| 7903 |
+
"learning_rate": 1.1596827664461558e-05,
|
| 7904 |
+
"loss": 1.2065,
|
| 7905 |
+
"step": 1128
|
| 7906 |
+
},
|
| 7907 |
+
{
|
| 7908 |
+
"epoch": 0.053453908432365894,
|
| 7909 |
+
"grad_norm": 1.2881298065185547,
|
| 7910 |
+
"learning_rate": 1.1527161803779867e-05,
|
| 7911 |
+
"loss": 1.3759,
|
| 7912 |
+
"step": 1129
|
| 7913 |
+
},
|
| 7914 |
+
{
|
| 7915 |
+
"epoch": 0.053501254675441506,
|
| 7916 |
+
"grad_norm": 1.3440982103347778,
|
| 7917 |
+
"learning_rate": 1.145767855256097e-05,
|
| 7918 |
+
"loss": 1.6574,
|
| 7919 |
+
"step": 1130
|
| 7920 |
+
},
|
| 7921 |
+
{
|
| 7922 |
+
"epoch": 0.05354860091851712,
|
| 7923 |
+
"grad_norm": 1.266324520111084,
|
| 7924 |
+
"learning_rate": 1.1388378240603741e-05,
|
| 7925 |
+
"loss": 1.4652,
|
| 7926 |
+
"step": 1131
|
| 7927 |
+
},
|
| 7928 |
+
{
|
| 7929 |
+
"epoch": 0.05359594716159273,
|
| 7930 |
+
"grad_norm": 1.4165794849395752,
|
| 7931 |
+
"learning_rate": 1.1319261196838782e-05,
|
| 7932 |
+
"loss": 1.611,
|
| 7933 |
+
"step": 1132
|
| 7934 |
+
},
|
| 7935 |
+
{
|
| 7936 |
+
"epoch": 0.05364329340466834,
|
| 7937 |
+
"grad_norm": 1.284224271774292,
|
| 7938 |
+
"learning_rate": 1.1250327749326772e-05,
|
| 7939 |
+
"loss": 1.5052,
|
| 7940 |
+
"step": 1133
|
| 7941 |
+
},
|
| 7942 |
+
{
|
| 7943 |
+
"epoch": 0.053690639647743954,
|
| 7944 |
+
"grad_norm": 1.1414461135864258,
|
| 7945 |
+
"learning_rate": 1.1181578225256972e-05,
|
| 7946 |
+
"loss": 1.1535,
|
| 7947 |
+
"step": 1134
|
| 7948 |
+
},
|
| 7949 |
+
{
|
| 7950 |
+
"epoch": 0.053737985890819566,
|
| 7951 |
+
"grad_norm": 1.4431864023208618,
|
| 7952 |
+
"learning_rate": 1.1113012950945695e-05,
|
| 7953 |
+
"loss": 1.2547,
|
| 7954 |
+
"step": 1135
|
| 7955 |
+
},
|
| 7956 |
+
{
|
| 7957 |
+
"epoch": 0.05378533213389518,
|
| 7958 |
+
"grad_norm": 1.2507365942001343,
|
| 7959 |
+
"learning_rate": 1.1044632251834686e-05,
|
| 7960 |
+
"loss": 1.3121,
|
| 7961 |
+
"step": 1136
|
| 7962 |
+
},
|
| 7963 |
+
{
|
| 7964 |
+
"epoch": 0.05383267837697079,
|
| 7965 |
+
"grad_norm": 1.1714433431625366,
|
| 7966 |
+
"learning_rate": 1.0976436452489592e-05,
|
| 7967 |
+
"loss": 1.1239,
|
| 7968 |
+
"step": 1137
|
| 7969 |
+
},
|
| 7970 |
+
{
|
| 7971 |
+
"epoch": 0.0538800246200464,
|
| 7972 |
+
"grad_norm": 1.3286577463150024,
|
| 7973 |
+
"learning_rate": 1.090842587659851e-05,
|
| 7974 |
+
"loss": 1.3356,
|
| 7975 |
+
"step": 1138
|
| 7976 |
+
},
|
| 7977 |
+
{
|
| 7978 |
+
"epoch": 0.053927370863122014,
|
| 7979 |
+
"grad_norm": 1.4974550008773804,
|
| 7980 |
+
"learning_rate": 1.0840600846970334e-05,
|
| 7981 |
+
"loss": 1.5551,
|
| 7982 |
+
"step": 1139
|
| 7983 |
+
},
|
| 7984 |
+
{
|
| 7985 |
+
"epoch": 0.053974717106197626,
|
| 7986 |
+
"grad_norm": 1.1929486989974976,
|
| 7987 |
+
"learning_rate": 1.0772961685533266e-05,
|
| 7988 |
+
"loss": 1.2629,
|
| 7989 |
+
"step": 1140
|
| 7990 |
+
},
|
| 7991 |
+
{
|
| 7992 |
+
"epoch": 0.05402206334927324,
|
| 7993 |
+
"grad_norm": 1.4195520877838135,
|
| 7994 |
+
"learning_rate": 1.0705508713333313e-05,
|
| 7995 |
+
"loss": 1.2655,
|
| 7996 |
+
"step": 1141
|
| 7997 |
+
},
|
| 7998 |
+
{
|
| 7999 |
+
"epoch": 0.05406940959234885,
|
| 8000 |
+
"grad_norm": 1.5098705291748047,
|
| 8001 |
+
"learning_rate": 1.0638242250532755e-05,
|
| 8002 |
+
"loss": 1.42,
|
| 8003 |
+
"step": 1142
|
| 8004 |
+
},
|
| 8005 |
+
{
|
| 8006 |
+
"epoch": 0.05411675583542446,
|
| 8007 |
+
"grad_norm": 1.3926045894622803,
|
| 8008 |
+
"learning_rate": 1.0571162616408587e-05,
|
| 8009 |
+
"loss": 1.4434,
|
| 8010 |
+
"step": 1143
|
| 8011 |
+
},
|
| 8012 |
+
{
|
| 8013 |
+
"epoch": 0.05416410207850007,
|
| 8014 |
+
"grad_norm": 1.1991745233535767,
|
| 8015 |
+
"learning_rate": 1.0504270129351035e-05,
|
| 8016 |
+
"loss": 1.1939,
|
| 8017 |
+
"step": 1144
|
| 8018 |
+
},
|
| 8019 |
+
{
|
| 8020 |
+
"epoch": 0.054211448321575685,
|
| 8021 |
+
"grad_norm": 1.3831684589385986,
|
| 8022 |
+
"learning_rate": 1.0437565106862074e-05,
|
| 8023 |
+
"loss": 1.492,
|
| 8024 |
+
"step": 1145
|
| 8025 |
+
},
|
| 8026 |
+
{
|
| 8027 |
+
"epoch": 0.0542587945646513,
|
| 8028 |
+
"grad_norm": 1.5574158430099487,
|
| 8029 |
+
"learning_rate": 1.0371047865553846e-05,
|
| 8030 |
+
"loss": 1.4823,
|
| 8031 |
+
"step": 1146
|
| 8032 |
+
},
|
| 8033 |
+
{
|
| 8034 |
+
"epoch": 0.05430614080772691,
|
| 8035 |
+
"grad_norm": 1.4688689708709717,
|
| 8036 |
+
"learning_rate": 1.03047187211472e-05,
|
| 8037 |
+
"loss": 1.3908,
|
| 8038 |
+
"step": 1147
|
| 8039 |
+
},
|
| 8040 |
+
{
|
| 8041 |
+
"epoch": 0.05435348705080252,
|
| 8042 |
+
"grad_norm": 1.5596418380737305,
|
| 8043 |
+
"learning_rate": 1.0238577988470232e-05,
|
| 8044 |
+
"loss": 1.247,
|
| 8045 |
+
"step": 1148
|
| 8046 |
+
},
|
| 8047 |
+
{
|
| 8048 |
+
"epoch": 0.05440083329387813,
|
| 8049 |
+
"grad_norm": 1.3878637552261353,
|
| 8050 |
+
"learning_rate": 1.0172625981456723e-05,
|
| 8051 |
+
"loss": 1.3574,
|
| 8052 |
+
"step": 1149
|
| 8053 |
+
},
|
| 8054 |
+
{
|
| 8055 |
+
"epoch": 0.054448179536953745,
|
| 8056 |
+
"grad_norm": 1.5790703296661377,
|
| 8057 |
+
"learning_rate": 1.0106863013144669e-05,
|
| 8058 |
+
"loss": 1.2905,
|
| 8059 |
+
"step": 1150
|
| 8060 |
+
},
|
| 8061 |
+
{
|
| 8062 |
+
"epoch": 0.05449552578002936,
|
| 8063 |
+
"grad_norm": 1.2336885929107666,
|
| 8064 |
+
"learning_rate": 1.0041289395674802e-05,
|
| 8065 |
+
"loss": 1.5459,
|
| 8066 |
+
"step": 1151
|
| 8067 |
+
},
|
| 8068 |
+
{
|
| 8069 |
+
"epoch": 0.05454287202310497,
|
| 8070 |
+
"grad_norm": 1.2021138668060303,
|
| 8071 |
+
"learning_rate": 9.975905440289152e-06,
|
| 8072 |
+
"loss": 1.6154,
|
| 8073 |
+
"step": 1152
|
| 8074 |
+
},
|
| 8075 |
+
{
|
| 8076 |
+
"epoch": 0.05459021826618058,
|
| 8077 |
+
"grad_norm": 1.0933634042739868,
|
| 8078 |
+
"learning_rate": 9.91071145732948e-06,
|
| 8079 |
+
"loss": 1.4471,
|
| 8080 |
+
"step": 1153
|
| 8081 |
+
},
|
| 8082 |
+
{
|
| 8083 |
+
"epoch": 0.05463756450925619,
|
| 8084 |
+
"grad_norm": 1.1646209955215454,
|
| 8085 |
+
"learning_rate": 9.84570775623585e-06,
|
| 8086 |
+
"loss": 1.4389,
|
| 8087 |
+
"step": 1154
|
| 8088 |
+
},
|
| 8089 |
+
{
|
| 8090 |
+
"epoch": 0.054684910752331804,
|
| 8091 |
+
"grad_norm": 1.2779886722564697,
|
| 8092 |
+
"learning_rate": 9.780894645545214e-06,
|
| 8093 |
+
"loss": 1.6144,
|
| 8094 |
+
"step": 1155
|
| 8095 |
+
},
|
| 8096 |
+
{
|
| 8097 |
+
"epoch": 0.054732256995407416,
|
| 8098 |
+
"grad_norm": 1.1957786083221436,
|
| 8099 |
+
"learning_rate": 9.716272432889844e-06,
|
| 8100 |
+
"loss": 1.6371,
|
| 8101 |
+
"step": 1156
|
| 8102 |
+
},
|
| 8103 |
+
{
|
| 8104 |
+
"epoch": 0.05477960323848303,
|
| 8105 |
+
"grad_norm": 1.1475330591201782,
|
| 8106 |
+
"learning_rate": 9.651841424995934e-06,
|
| 8107 |
+
"loss": 1.5913,
|
| 8108 |
+
"step": 1157
|
| 8109 |
+
},
|
| 8110 |
+
{
|
| 8111 |
+
"epoch": 0.05482694948155864,
|
| 8112 |
+
"grad_norm": 1.0740982294082642,
|
| 8113 |
+
"learning_rate": 9.58760192768216e-06,
|
| 8114 |
+
"loss": 1.223,
|
| 8115 |
+
"step": 1158
|
| 8116 |
+
},
|
| 8117 |
+
{
|
| 8118 |
+
"epoch": 0.05487429572463425,
|
| 8119 |
+
"grad_norm": 1.166710615158081,
|
| 8120 |
+
"learning_rate": 9.52355424585818e-06,
|
| 8121 |
+
"loss": 1.4922,
|
| 8122 |
+
"step": 1159
|
| 8123 |
+
},
|
| 8124 |
+
{
|
| 8125 |
+
"epoch": 0.054921641967709864,
|
| 8126 |
+
"grad_norm": 1.3334842920303345,
|
| 8127 |
+
"learning_rate": 9.459698683523204e-06,
|
| 8128 |
+
"loss": 1.5167,
|
| 8129 |
+
"step": 1160
|
| 8130 |
+
},
|
| 8131 |
+
{
|
| 8132 |
+
"epoch": 0.054968988210785476,
|
| 8133 |
+
"grad_norm": 1.399493932723999,
|
| 8134 |
+
"learning_rate": 9.39603554376456e-06,
|
| 8135 |
+
"loss": 1.6392,
|
| 8136 |
+
"step": 1161
|
| 8137 |
+
},
|
| 8138 |
+
{
|
| 8139 |
+
"epoch": 0.05501633445386109,
|
| 8140 |
+
"grad_norm": 1.229955792427063,
|
| 8141 |
+
"learning_rate": 9.332565128756283e-06,
|
| 8142 |
+
"loss": 1.5309,
|
| 8143 |
+
"step": 1162
|
| 8144 |
+
},
|
| 8145 |
+
{
|
| 8146 |
+
"epoch": 0.0550636806969367,
|
| 8147 |
+
"grad_norm": 1.2729157209396362,
|
| 8148 |
+
"learning_rate": 9.269287739757604e-06,
|
| 8149 |
+
"loss": 1.3175,
|
| 8150 |
+
"step": 1163
|
| 8151 |
+
},
|
| 8152 |
+
{
|
| 8153 |
+
"epoch": 0.05511102694001231,
|
| 8154 |
+
"grad_norm": 1.3252742290496826,
|
| 8155 |
+
"learning_rate": 9.206203677111575e-06,
|
| 8156 |
+
"loss": 1.3721,
|
| 8157 |
+
"step": 1164
|
| 8158 |
+
},
|
| 8159 |
+
{
|
| 8160 |
+
"epoch": 0.05515837318308792,
|
| 8161 |
+
"grad_norm": 1.1812686920166016,
|
| 8162 |
+
"learning_rate": 9.143313240243667e-06,
|
| 8163 |
+
"loss": 1.3929,
|
| 8164 |
+
"step": 1165
|
| 8165 |
+
},
|
| 8166 |
+
{
|
| 8167 |
+
"epoch": 0.055205719426163535,
|
| 8168 |
+
"grad_norm": 1.192521572113037,
|
| 8169 |
+
"learning_rate": 9.080616727660285e-06,
|
| 8170 |
+
"loss": 1.3355,
|
| 8171 |
+
"step": 1166
|
| 8172 |
+
},
|
| 8173 |
+
{
|
| 8174 |
+
"epoch": 0.05525306566923915,
|
| 8175 |
+
"grad_norm": 0.9685383439064026,
|
| 8176 |
+
"learning_rate": 9.018114436947373e-06,
|
| 8177 |
+
"loss": 1.1218,
|
| 8178 |
+
"step": 1167
|
| 8179 |
+
},
|
| 8180 |
+
{
|
| 8181 |
+
"epoch": 0.05530041191231476,
|
| 8182 |
+
"grad_norm": 1.3759257793426514,
|
| 8183 |
+
"learning_rate": 8.955806664769051e-06,
|
| 8184 |
+
"loss": 1.6344,
|
| 8185 |
+
"step": 1168
|
| 8186 |
+
},
|
| 8187 |
+
{
|
| 8188 |
+
"epoch": 0.05534775815539037,
|
| 8189 |
+
"grad_norm": 1.1711422204971313,
|
| 8190 |
+
"learning_rate": 8.893693706866125e-06,
|
| 8191 |
+
"loss": 1.4263,
|
| 8192 |
+
"step": 1169
|
| 8193 |
+
},
|
| 8194 |
+
{
|
| 8195 |
+
"epoch": 0.05539510439846598,
|
| 8196 |
+
"grad_norm": 1.3107439279556274,
|
| 8197 |
+
"learning_rate": 8.831775858054747e-06,
|
| 8198 |
+
"loss": 1.5476,
|
| 8199 |
+
"step": 1170
|
| 8200 |
+
},
|
| 8201 |
+
{
|
| 8202 |
+
"epoch": 0.055442450641541595,
|
| 8203 |
+
"grad_norm": 1.1891083717346191,
|
| 8204 |
+
"learning_rate": 8.770053412224971e-06,
|
| 8205 |
+
"loss": 1.3977,
|
| 8206 |
+
"step": 1171
|
| 8207 |
+
},
|
| 8208 |
+
{
|
| 8209 |
+
"epoch": 0.05548979688461721,
|
| 8210 |
+
"grad_norm": 1.472711443901062,
|
| 8211 |
+
"learning_rate": 8.708526662339428e-06,
|
| 8212 |
+
"loss": 1.7162,
|
| 8213 |
+
"step": 1172
|
| 8214 |
+
},
|
| 8215 |
+
{
|
| 8216 |
+
"epoch": 0.05553714312769282,
|
| 8217 |
+
"grad_norm": 1.3298579454421997,
|
| 8218 |
+
"learning_rate": 8.647195900431831e-06,
|
| 8219 |
+
"loss": 1.4595,
|
| 8220 |
+
"step": 1173
|
| 8221 |
+
},
|
| 8222 |
+
{
|
| 8223 |
+
"epoch": 0.05558448937076843,
|
| 8224 |
+
"grad_norm": 1.3442754745483398,
|
| 8225 |
+
"learning_rate": 8.586061417605668e-06,
|
| 8226 |
+
"loss": 1.2764,
|
| 8227 |
+
"step": 1174
|
| 8228 |
+
},
|
| 8229 |
+
{
|
| 8230 |
+
"epoch": 0.05563183561384404,
|
| 8231 |
+
"grad_norm": 1.2704945802688599,
|
| 8232 |
+
"learning_rate": 8.525123504032817e-06,
|
| 8233 |
+
"loss": 1.2696,
|
| 8234 |
+
"step": 1175
|
| 8235 |
+
},
|
| 8236 |
+
{
|
| 8237 |
+
"epoch": 0.055679181856919654,
|
| 8238 |
+
"grad_norm": 1.1754649877548218,
|
| 8239 |
+
"learning_rate": 8.46438244895213e-06,
|
| 8240 |
+
"loss": 1.3784,
|
| 8241 |
+
"step": 1176
|
| 8242 |
+
},
|
| 8243 |
+
{
|
| 8244 |
+
"epoch": 0.055726528099995266,
|
| 8245 |
+
"grad_norm": 1.1760340929031372,
|
| 8246 |
+
"learning_rate": 8.403838540668057e-06,
|
| 8247 |
+
"loss": 1.3792,
|
| 8248 |
+
"step": 1177
|
| 8249 |
+
},
|
| 8250 |
+
{
|
| 8251 |
+
"epoch": 0.05577387434307088,
|
| 8252 |
+
"grad_norm": 1.318698763847351,
|
| 8253 |
+
"learning_rate": 8.343492066549368e-06,
|
| 8254 |
+
"loss": 1.3285,
|
| 8255 |
+
"step": 1178
|
| 8256 |
+
},
|
| 8257 |
+
{
|
| 8258 |
+
"epoch": 0.05582122058614649,
|
| 8259 |
+
"grad_norm": 1.0595449209213257,
|
| 8260 |
+
"learning_rate": 8.283343313027653e-06,
|
| 8261 |
+
"loss": 1.2448,
|
| 8262 |
+
"step": 1179
|
| 8263 |
+
},
|
| 8264 |
+
{
|
| 8265 |
+
"epoch": 0.0558685668292221,
|
| 8266 |
+
"grad_norm": 1.2634952068328857,
|
| 8267 |
+
"learning_rate": 8.22339256559606e-06,
|
| 8268 |
+
"loss": 1.3498,
|
| 8269 |
+
"step": 1180
|
| 8270 |
+
},
|
| 8271 |
+
{
|
| 8272 |
+
"epoch": 0.055915913072297714,
|
| 8273 |
+
"grad_norm": 1.0653668642044067,
|
| 8274 |
+
"learning_rate": 8.163640108807896e-06,
|
| 8275 |
+
"loss": 1.2436,
|
| 8276 |
+
"step": 1181
|
| 8277 |
+
},
|
| 8278 |
+
{
|
| 8279 |
+
"epoch": 0.055963259315373326,
|
| 8280 |
+
"grad_norm": 1.3706316947937012,
|
| 8281 |
+
"learning_rate": 8.104086226275332e-06,
|
| 8282 |
+
"loss": 1.3773,
|
| 8283 |
+
"step": 1182
|
| 8284 |
+
},
|
| 8285 |
+
{
|
| 8286 |
+
"epoch": 0.05601060555844894,
|
| 8287 |
+
"grad_norm": 1.3542323112487793,
|
| 8288 |
+
"learning_rate": 8.044731200667966e-06,
|
| 8289 |
+
"loss": 1.5692,
|
| 8290 |
+
"step": 1183
|
| 8291 |
+
},
|
| 8292 |
+
{
|
| 8293 |
+
"epoch": 0.05605795180152455,
|
| 8294 |
+
"grad_norm": 1.4079458713531494,
|
| 8295 |
+
"learning_rate": 7.985575313711546e-06,
|
| 8296 |
+
"loss": 1.5482,
|
| 8297 |
+
"step": 1184
|
| 8298 |
+
},
|
| 8299 |
+
{
|
| 8300 |
+
"epoch": 0.05610529804460016,
|
| 8301 |
+
"grad_norm": 1.3803889751434326,
|
| 8302 |
+
"learning_rate": 7.926618846186645e-06,
|
| 8303 |
+
"loss": 1.4889,
|
| 8304 |
+
"step": 1185
|
| 8305 |
+
},
|
| 8306 |
+
{
|
| 8307 |
+
"epoch": 0.056152644287675774,
|
| 8308 |
+
"grad_norm": 1.2433762550354004,
|
| 8309 |
+
"learning_rate": 7.867862077927269e-06,
|
| 8310 |
+
"loss": 1.2356,
|
| 8311 |
+
"step": 1186
|
| 8312 |
+
},
|
| 8313 |
+
{
|
| 8314 |
+
"epoch": 0.056199990530751386,
|
| 8315 |
+
"grad_norm": 1.273190975189209,
|
| 8316 |
+
"learning_rate": 7.809305287819558e-06,
|
| 8317 |
+
"loss": 1.3955,
|
| 8318 |
+
"step": 1187
|
| 8319 |
+
},
|
| 8320 |
+
{
|
| 8321 |
+
"epoch": 0.056247336773827,
|
| 8322 |
+
"grad_norm": 1.3209604024887085,
|
| 8323 |
+
"learning_rate": 7.750948753800507e-06,
|
| 8324 |
+
"loss": 1.5155,
|
| 8325 |
+
"step": 1188
|
| 8326 |
+
},
|
| 8327 |
+
{
|
| 8328 |
+
"epoch": 0.05629468301690261,
|
| 8329 |
+
"grad_norm": 1.3223750591278076,
|
| 8330 |
+
"learning_rate": 7.692792752856565e-06,
|
| 8331 |
+
"loss": 1.459,
|
| 8332 |
+
"step": 1189
|
| 8333 |
+
},
|
| 8334 |
+
{
|
| 8335 |
+
"epoch": 0.05634202925997822,
|
| 8336 |
+
"grad_norm": 1.6244319677352905,
|
| 8337 |
+
"learning_rate": 7.634837561022374e-06,
|
| 8338 |
+
"loss": 1.5188,
|
| 8339 |
+
"step": 1190
|
| 8340 |
+
},
|
| 8341 |
+
{
|
| 8342 |
+
"epoch": 0.05638937550305383,
|
| 8343 |
+
"grad_norm": 1.22336745262146,
|
| 8344 |
+
"learning_rate": 7.577083453379424e-06,
|
| 8345 |
+
"loss": 1.3745,
|
| 8346 |
+
"step": 1191
|
| 8347 |
+
},
|
| 8348 |
+
{
|
| 8349 |
+
"epoch": 0.056436721746129445,
|
| 8350 |
+
"grad_norm": 1.2422270774841309,
|
| 8351 |
+
"learning_rate": 7.519530704054828e-06,
|
| 8352 |
+
"loss": 1.3194,
|
| 8353 |
+
"step": 1192
|
| 8354 |
+
},
|
| 8355 |
+
{
|
| 8356 |
+
"epoch": 0.05648406798920506,
|
| 8357 |
+
"grad_norm": 1.2926522493362427,
|
| 8358 |
+
"learning_rate": 7.462179586219897e-06,
|
| 8359 |
+
"loss": 1.274,
|
| 8360 |
+
"step": 1193
|
| 8361 |
+
},
|
| 8362 |
+
{
|
| 8363 |
+
"epoch": 0.05653141423228067,
|
| 8364 |
+
"grad_norm": 1.3746919631958008,
|
| 8365 |
+
"learning_rate": 7.405030372088933e-06,
|
| 8366 |
+
"loss": 1.2977,
|
| 8367 |
+
"step": 1194
|
| 8368 |
+
},
|
| 8369 |
+
{
|
| 8370 |
+
"epoch": 0.05657876047535628,
|
| 8371 |
+
"grad_norm": 1.4829906225204468,
|
| 8372 |
+
"learning_rate": 7.348083332917926e-06,
|
| 8373 |
+
"loss": 1.2944,
|
| 8374 |
+
"step": 1195
|
| 8375 |
+
},
|
| 8376 |
+
{
|
| 8377 |
+
"epoch": 0.05662610671843189,
|
| 8378 |
+
"grad_norm": 1.2516322135925293,
|
| 8379 |
+
"learning_rate": 7.291338739003234e-06,
|
| 8380 |
+
"loss": 1.1295,
|
| 8381 |
+
"step": 1196
|
| 8382 |
+
},
|
| 8383 |
+
{
|
| 8384 |
+
"epoch": 0.056673452961507505,
|
| 8385 |
+
"grad_norm": 1.5649276971817017,
|
| 8386 |
+
"learning_rate": 7.234796859680309e-06,
|
| 8387 |
+
"loss": 1.3443,
|
| 8388 |
+
"step": 1197
|
| 8389 |
+
},
|
| 8390 |
+
{
|
| 8391 |
+
"epoch": 0.05672079920458312,
|
| 8392 |
+
"grad_norm": 1.3753424882888794,
|
| 8393 |
+
"learning_rate": 7.178457963322471e-06,
|
| 8394 |
+
"loss": 1.2108,
|
| 8395 |
+
"step": 1198
|
| 8396 |
+
},
|
| 8397 |
+
{
|
| 8398 |
+
"epoch": 0.05676814544765873,
|
| 8399 |
+
"grad_norm": 1.6467722654342651,
|
| 8400 |
+
"learning_rate": 7.1223223173395416e-06,
|
| 8401 |
+
"loss": 1.5478,
|
| 8402 |
+
"step": 1199
|
| 8403 |
+
},
|
| 8404 |
+
{
|
| 8405 |
+
"epoch": 0.05681549169073434,
|
| 8406 |
+
"grad_norm": 1.6803184747695923,
|
| 8407 |
+
"learning_rate": 7.066390188176647e-06,
|
| 8408 |
+
"loss": 1.4304,
|
| 8409 |
+
"step": 1200
|
| 8410 |
+
},
|
| 8411 |
+
{
|
| 8412 |
+
"epoch": 0.05686283793380995,
|
| 8413 |
+
"grad_norm": 1.2419159412384033,
|
| 8414 |
+
"learning_rate": 7.010661841312921e-06,
|
| 8415 |
+
"loss": 1.5378,
|
| 8416 |
+
"step": 1201
|
| 8417 |
+
},
|
| 8418 |
+
{
|
| 8419 |
+
"epoch": 0.056910184176885564,
|
| 8420 |
+
"grad_norm": 1.4295827150344849,
|
| 8421 |
+
"learning_rate": 6.955137541260287e-06,
|
| 8422 |
+
"loss": 1.4816,
|
| 8423 |
+
"step": 1202
|
| 8424 |
+
},
|
| 8425 |
+
{
|
| 8426 |
+
"epoch": 0.056957530419961176,
|
| 8427 |
+
"grad_norm": 1.4575597047805786,
|
| 8428 |
+
"learning_rate": 6.899817551562127e-06,
|
| 8429 |
+
"loss": 1.778,
|
| 8430 |
+
"step": 1203
|
| 8431 |
+
},
|
| 8432 |
+
{
|
| 8433 |
+
"epoch": 0.05700487666303679,
|
| 8434 |
+
"grad_norm": 1.3488417863845825,
|
| 8435 |
+
"learning_rate": 6.844702134792086e-06,
|
| 8436 |
+
"loss": 1.7534,
|
| 8437 |
+
"step": 1204
|
| 8438 |
+
},
|
| 8439 |
+
{
|
| 8440 |
+
"epoch": 0.0570522229061124,
|
| 8441 |
+
"grad_norm": 1.2744368314743042,
|
| 8442 |
+
"learning_rate": 6.789791552552838e-06,
|
| 8443 |
+
"loss": 1.5629,
|
| 8444 |
+
"step": 1205
|
| 8445 |
+
},
|
| 8446 |
+
{
|
| 8447 |
+
"epoch": 0.05709956914918801,
|
| 8448 |
+
"grad_norm": 1.2052980661392212,
|
| 8449 |
+
"learning_rate": 6.735086065474799e-06,
|
| 8450 |
+
"loss": 1.4061,
|
| 8451 |
+
"step": 1206
|
| 8452 |
+
},
|
| 8453 |
+
{
|
| 8454 |
+
"epoch": 0.057146915392263624,
|
| 8455 |
+
"grad_norm": 1.2301853895187378,
|
| 8456 |
+
"learning_rate": 6.6805859332148945e-06,
|
| 8457 |
+
"loss": 1.5704,
|
| 8458 |
+
"step": 1207
|
| 8459 |
+
},
|
| 8460 |
+
{
|
| 8461 |
+
"epoch": 0.057194261635339236,
|
| 8462 |
+
"grad_norm": 1.1014851331710815,
|
| 8463 |
+
"learning_rate": 6.626291414455388e-06,
|
| 8464 |
+
"loss": 1.2843,
|
| 8465 |
+
"step": 1208
|
| 8466 |
+
},
|
| 8467 |
+
{
|
| 8468 |
+
"epoch": 0.05724160787841485,
|
| 8469 |
+
"grad_norm": 1.1493160724639893,
|
| 8470 |
+
"learning_rate": 6.572202766902569e-06,
|
| 8471 |
+
"loss": 1.3725,
|
| 8472 |
+
"step": 1209
|
| 8473 |
+
},
|
| 8474 |
+
{
|
| 8475 |
+
"epoch": 0.05728895412149046,
|
| 8476 |
+
"grad_norm": 1.1507078409194946,
|
| 8477 |
+
"learning_rate": 6.518320247285581e-06,
|
| 8478 |
+
"loss": 1.2313,
|
| 8479 |
+
"step": 1210
|
| 8480 |
+
},
|
| 8481 |
+
{
|
| 8482 |
+
"epoch": 0.05733630036456607,
|
| 8483 |
+
"grad_norm": 1.3263248205184937,
|
| 8484 |
+
"learning_rate": 6.464644111355189e-06,
|
| 8485 |
+
"loss": 1.5823,
|
| 8486 |
+
"step": 1211
|
| 8487 |
+
},
|
| 8488 |
+
{
|
| 8489 |
+
"epoch": 0.05738364660764168,
|
| 8490 |
+
"grad_norm": 1.253355622291565,
|
| 8491 |
+
"learning_rate": 6.411174613882592e-06,
|
| 8492 |
+
"loss": 1.5534,
|
| 8493 |
+
"step": 1212
|
| 8494 |
+
},
|
| 8495 |
+
{
|
| 8496 |
+
"epoch": 0.057430992850717295,
|
| 8497 |
+
"grad_norm": 1.2244974374771118,
|
| 8498 |
+
"learning_rate": 6.3579120086581515e-06,
|
| 8499 |
+
"loss": 1.5079,
|
| 8500 |
+
"step": 1213
|
| 8501 |
+
},
|
| 8502 |
+
{
|
| 8503 |
+
"epoch": 0.05747833909379291,
|
| 8504 |
+
"grad_norm": 1.1507781744003296,
|
| 8505 |
+
"learning_rate": 6.30485654849024e-06,
|
| 8506 |
+
"loss": 1.4029,
|
| 8507 |
+
"step": 1214
|
| 8508 |
+
},
|
| 8509 |
+
{
|
| 8510 |
+
"epoch": 0.05752568533686852,
|
| 8511 |
+
"grad_norm": 1.1730035543441772,
|
| 8512 |
+
"learning_rate": 6.2520084852040526e-06,
|
| 8513 |
+
"loss": 1.2631,
|
| 8514 |
+
"step": 1215
|
| 8515 |
+
},
|
| 8516 |
+
{
|
| 8517 |
+
"epoch": 0.05757303157994413,
|
| 8518 |
+
"grad_norm": 1.3316566944122314,
|
| 8519 |
+
"learning_rate": 6.199368069640343e-06,
|
| 8520 |
+
"loss": 1.6483,
|
| 8521 |
+
"step": 1216
|
| 8522 |
+
},
|
| 8523 |
+
{
|
| 8524 |
+
"epoch": 0.05762037782301974,
|
| 8525 |
+
"grad_norm": 1.1844844818115234,
|
| 8526 |
+
"learning_rate": 6.1469355516542985e-06,
|
| 8527 |
+
"loss": 1.44,
|
| 8528 |
+
"step": 1217
|
| 8529 |
+
},
|
| 8530 |
+
{
|
| 8531 |
+
"epoch": 0.057667724066095355,
|
| 8532 |
+
"grad_norm": 1.2655210494995117,
|
| 8533 |
+
"learning_rate": 6.094711180114309e-06,
|
| 8534 |
+
"loss": 1.5127,
|
| 8535 |
+
"step": 1218
|
| 8536 |
+
},
|
| 8537 |
+
{
|
| 8538 |
+
"epoch": 0.05771507030917097,
|
| 8539 |
+
"grad_norm": 1.2262639999389648,
|
| 8540 |
+
"learning_rate": 6.042695202900855e-06,
|
| 8541 |
+
"loss": 1.5044,
|
| 8542 |
+
"step": 1219
|
| 8543 |
+
},
|
| 8544 |
+
{
|
| 8545 |
+
"epoch": 0.05776241655224658,
|
| 8546 |
+
"grad_norm": 1.340762972831726,
|
| 8547 |
+
"learning_rate": 5.990887866905237e-06,
|
| 8548 |
+
"loss": 1.4632,
|
| 8549 |
+
"step": 1220
|
| 8550 |
+
},
|
| 8551 |
+
{
|
| 8552 |
+
"epoch": 0.05780976279532219,
|
| 8553 |
+
"grad_norm": 2.6526997089385986,
|
| 8554 |
+
"learning_rate": 5.939289418028454e-06,
|
| 8555 |
+
"loss": 1.4763,
|
| 8556 |
+
"step": 1221
|
| 8557 |
+
},
|
| 8558 |
+
{
|
| 8559 |
+
"epoch": 0.0578571090383978,
|
| 8560 |
+
"grad_norm": 1.28556227684021,
|
| 8561 |
+
"learning_rate": 5.887900101180066e-06,
|
| 8562 |
+
"loss": 1.4482,
|
| 8563 |
+
"step": 1222
|
| 8564 |
+
},
|
| 8565 |
+
{
|
| 8566 |
+
"epoch": 0.057904455281473415,
|
| 8567 |
+
"grad_norm": 1.3036595582962036,
|
| 8568 |
+
"learning_rate": 5.83672016027697e-06,
|
| 8569 |
+
"loss": 1.6322,
|
| 8570 |
+
"step": 1223
|
| 8571 |
+
},
|
| 8572 |
+
{
|
| 8573 |
+
"epoch": 0.057951801524549026,
|
| 8574 |
+
"grad_norm": 1.4159454107284546,
|
| 8575 |
+
"learning_rate": 5.78574983824226e-06,
|
| 8576 |
+
"loss": 1.4237,
|
| 8577 |
+
"step": 1224
|
| 8578 |
+
},
|
| 8579 |
+
{
|
| 8580 |
+
"epoch": 0.05799914776762464,
|
| 8581 |
+
"grad_norm": 1.2348464727401733,
|
| 8582 |
+
"learning_rate": 5.734989377004118e-06,
|
| 8583 |
+
"loss": 1.372,
|
| 8584 |
+
"step": 1225
|
| 8585 |
+
},
|
| 8586 |
+
{
|
| 8587 |
+
"epoch": 0.05804649401070025,
|
| 8588 |
+
"grad_norm": 1.4472335577011108,
|
| 8589 |
+
"learning_rate": 5.684439017494614e-06,
|
| 8590 |
+
"loss": 1.6784,
|
| 8591 |
+
"step": 1226
|
| 8592 |
+
},
|
| 8593 |
+
{
|
| 8594 |
+
"epoch": 0.05809384025377586,
|
| 8595 |
+
"grad_norm": 1.2965421676635742,
|
| 8596 |
+
"learning_rate": 5.634098999648579e-06,
|
| 8597 |
+
"loss": 1.5013,
|
| 8598 |
+
"step": 1227
|
| 8599 |
+
},
|
| 8600 |
+
{
|
| 8601 |
+
"epoch": 0.058141186496851474,
|
| 8602 |
+
"grad_norm": 1.2801685333251953,
|
| 8603 |
+
"learning_rate": 5.583969562402447e-06,
|
| 8604 |
+
"loss": 1.5362,
|
| 8605 |
+
"step": 1228
|
| 8606 |
+
},
|
| 8607 |
+
{
|
| 8608 |
+
"epoch": 0.058188532739927086,
|
| 8609 |
+
"grad_norm": 1.151532530784607,
|
| 8610 |
+
"learning_rate": 5.534050943693198e-06,
|
| 8611 |
+
"loss": 1.1707,
|
| 8612 |
+
"step": 1229
|
| 8613 |
+
},
|
| 8614 |
+
{
|
| 8615 |
+
"epoch": 0.0582358789830027,
|
| 8616 |
+
"grad_norm": 1.2291841506958008,
|
| 8617 |
+
"learning_rate": 5.484343380457125e-06,
|
| 8618 |
+
"loss": 1.3508,
|
| 8619 |
+
"step": 1230
|
| 8620 |
+
},
|
| 8621 |
+
{
|
| 8622 |
+
"epoch": 0.05828322522607831,
|
| 8623 |
+
"grad_norm": 1.1725528240203857,
|
| 8624 |
+
"learning_rate": 5.4348471086287486e-06,
|
| 8625 |
+
"loss": 1.3057,
|
| 8626 |
+
"step": 1231
|
| 8627 |
+
},
|
| 8628 |
+
{
|
| 8629 |
+
"epoch": 0.05833057146915392,
|
| 8630 |
+
"grad_norm": 1.1347895860671997,
|
| 8631 |
+
"learning_rate": 5.385562363139757e-06,
|
| 8632 |
+
"loss": 1.4831,
|
| 8633 |
+
"step": 1232
|
| 8634 |
+
},
|
| 8635 |
+
{
|
| 8636 |
+
"epoch": 0.058377917712229534,
|
| 8637 |
+
"grad_norm": 1.2504128217697144,
|
| 8638 |
+
"learning_rate": 5.336489377917786e-06,
|
| 8639 |
+
"loss": 1.3756,
|
| 8640 |
+
"step": 1233
|
| 8641 |
+
},
|
| 8642 |
+
{
|
| 8643 |
+
"epoch": 0.058425263955305146,
|
| 8644 |
+
"grad_norm": 1.4475090503692627,
|
| 8645 |
+
"learning_rate": 5.287628385885385e-06,
|
| 8646 |
+
"loss": 1.5545,
|
| 8647 |
+
"step": 1234
|
| 8648 |
+
},
|
| 8649 |
+
{
|
| 8650 |
+
"epoch": 0.05847261019838076,
|
| 8651 |
+
"grad_norm": 1.3991408348083496,
|
| 8652 |
+
"learning_rate": 5.2389796189589e-06,
|
| 8653 |
+
"loss": 1.6003,
|
| 8654 |
+
"step": 1235
|
| 8655 |
+
},
|
| 8656 |
+
{
|
| 8657 |
+
"epoch": 0.05851995644145637,
|
| 8658 |
+
"grad_norm": 1.089169979095459,
|
| 8659 |
+
"learning_rate": 5.190543308047335e-06,
|
| 8660 |
+
"loss": 1.2467,
|
| 8661 |
+
"step": 1236
|
| 8662 |
+
},
|
| 8663 |
+
{
|
| 8664 |
+
"epoch": 0.05856730268453198,
|
| 8665 |
+
"grad_norm": 1.4275898933410645,
|
| 8666 |
+
"learning_rate": 5.1423196830513e-06,
|
| 8667 |
+
"loss": 1.5092,
|
| 8668 |
+
"step": 1237
|
| 8669 |
+
},
|
| 8670 |
+
{
|
| 8671 |
+
"epoch": 0.05861464892760759,
|
| 8672 |
+
"grad_norm": 1.4928823709487915,
|
| 8673 |
+
"learning_rate": 5.094308972861878e-06,
|
| 8674 |
+
"loss": 1.5897,
|
| 8675 |
+
"step": 1238
|
| 8676 |
+
},
|
| 8677 |
+
{
|
| 8678 |
+
"epoch": 0.058661995170683205,
|
| 8679 |
+
"grad_norm": 1.2940183877944946,
|
| 8680 |
+
"learning_rate": 5.0465114053596e-06,
|
| 8681 |
+
"loss": 1.2045,
|
| 8682 |
+
"step": 1239
|
| 8683 |
+
},
|
| 8684 |
+
{
|
| 8685 |
+
"epoch": 0.05870934141375882,
|
| 8686 |
+
"grad_norm": 1.778552532196045,
|
| 8687 |
+
"learning_rate": 4.998927207413295e-06,
|
| 8688 |
+
"loss": 1.8119,
|
| 8689 |
+
"step": 1240
|
| 8690 |
+
},
|
| 8691 |
+
{
|
| 8692 |
+
"epoch": 0.05875668765683443,
|
| 8693 |
+
"grad_norm": 1.470236897468567,
|
| 8694 |
+
"learning_rate": 4.951556604879048e-06,
|
| 8695 |
+
"loss": 1.3587,
|
| 8696 |
+
"step": 1241
|
| 8697 |
+
},
|
| 8698 |
+
{
|
| 8699 |
+
"epoch": 0.05880403389991004,
|
| 8700 |
+
"grad_norm": 1.4887255430221558,
|
| 8701 |
+
"learning_rate": 4.904399822599142e-06,
|
| 8702 |
+
"loss": 1.3095,
|
| 8703 |
+
"step": 1242
|
| 8704 |
+
},
|
| 8705 |
+
{
|
| 8706 |
+
"epoch": 0.05885138014298565,
|
| 8707 |
+
"grad_norm": 1.263823390007019,
|
| 8708 |
+
"learning_rate": 4.857457084400957e-06,
|
| 8709 |
+
"loss": 1.2518,
|
| 8710 |
+
"step": 1243
|
| 8711 |
+
},
|
| 8712 |
+
{
|
| 8713 |
+
"epoch": 0.058898726386061265,
|
| 8714 |
+
"grad_norm": 1.330069661140442,
|
| 8715 |
+
"learning_rate": 4.81072861309591e-06,
|
| 8716 |
+
"loss": 1.1333,
|
| 8717 |
+
"step": 1244
|
| 8718 |
+
},
|
| 8719 |
+
{
|
| 8720 |
+
"epoch": 0.05894607262913688,
|
| 8721 |
+
"grad_norm": 1.332933783531189,
|
| 8722 |
+
"learning_rate": 4.764214630478447e-06,
|
| 8723 |
+
"loss": 1.3739,
|
| 8724 |
+
"step": 1245
|
| 8725 |
+
},
|
| 8726 |
+
{
|
| 8727 |
+
"epoch": 0.05899341887221249,
|
| 8728 |
+
"grad_norm": 1.5848991870880127,
|
| 8729 |
+
"learning_rate": 4.717915357324926e-06,
|
| 8730 |
+
"loss": 1.7896,
|
| 8731 |
+
"step": 1246
|
| 8732 |
+
},
|
| 8733 |
+
{
|
| 8734 |
+
"epoch": 0.0590407651152881,
|
| 8735 |
+
"grad_norm": 1.492069125175476,
|
| 8736 |
+
"learning_rate": 4.671831013392608e-06,
|
| 8737 |
+
"loss": 1.3187,
|
| 8738 |
+
"step": 1247
|
| 8739 |
+
},
|
| 8740 |
+
{
|
| 8741 |
+
"epoch": 0.05908811135836371,
|
| 8742 |
+
"grad_norm": 1.2575969696044922,
|
| 8743 |
+
"learning_rate": 4.625961817418578e-06,
|
| 8744 |
+
"loss": 1.2123,
|
| 8745 |
+
"step": 1248
|
| 8746 |
+
},
|
| 8747 |
+
{
|
| 8748 |
+
"epoch": 0.059135457601439324,
|
| 8749 |
+
"grad_norm": 1.4132999181747437,
|
| 8750 |
+
"learning_rate": 4.580307987118781e-06,
|
| 8751 |
+
"loss": 1.3083,
|
| 8752 |
+
"step": 1249
|
| 8753 |
+
},
|
| 8754 |
+
{
|
| 8755 |
+
"epoch": 0.059182803844514936,
|
| 8756 |
+
"grad_norm": 1.3583660125732422,
|
| 8757 |
+
"learning_rate": 4.534869739186903e-06,
|
| 8758 |
+
"loss": 1.2488,
|
| 8759 |
+
"step": 1250
|
| 8760 |
+
},
|
| 8761 |
+
{
|
| 8762 |
+
"epoch": 0.05923015008759055,
|
| 8763 |
+
"grad_norm": 1.1643834114074707,
|
| 8764 |
+
"learning_rate": 4.489647289293369e-06,
|
| 8765 |
+
"loss": 1.3848,
|
| 8766 |
+
"step": 1251
|
| 8767 |
+
},
|
| 8768 |
+
{
|
| 8769 |
+
"epoch": 0.05927749633066616,
|
| 8770 |
+
"grad_norm": 1.2520825862884521,
|
| 8771 |
+
"learning_rate": 4.44464085208437e-06,
|
| 8772 |
+
"loss": 1.6913,
|
| 8773 |
+
"step": 1252
|
| 8774 |
+
},
|
| 8775 |
+
{
|
| 8776 |
+
"epoch": 0.05932484257374177,
|
| 8777 |
+
"grad_norm": 1.4900792837142944,
|
| 8778 |
+
"learning_rate": 4.3998506411807695e-06,
|
| 8779 |
+
"loss": 1.5912,
|
| 8780 |
+
"step": 1253
|
| 8781 |
+
},
|
| 8782 |
+
{
|
| 8783 |
+
"epoch": 0.059372188816817384,
|
| 8784 |
+
"grad_norm": 1.1609636545181274,
|
| 8785 |
+
"learning_rate": 4.355276869177128e-06,
|
| 8786 |
+
"loss": 1.572,
|
| 8787 |
+
"step": 1254
|
| 8788 |
+
},
|
| 8789 |
+
{
|
| 8790 |
+
"epoch": 0.059419535059892996,
|
| 8791 |
+
"grad_norm": 1.1604900360107422,
|
| 8792 |
+
"learning_rate": 4.310919747640707e-06,
|
| 8793 |
+
"loss": 1.5144,
|
| 8794 |
+
"step": 1255
|
| 8795 |
+
},
|
| 8796 |
+
{
|
| 8797 |
+
"epoch": 0.05946688130296861,
|
| 8798 |
+
"grad_norm": 1.0765118598937988,
|
| 8799 |
+
"learning_rate": 4.266779487110434e-06,
|
| 8800 |
+
"loss": 1.1977,
|
| 8801 |
+
"step": 1256
|
| 8802 |
+
},
|
| 8803 |
+
{
|
| 8804 |
+
"epoch": 0.05951422754604422,
|
| 8805 |
+
"grad_norm": 1.2380905151367188,
|
| 8806 |
+
"learning_rate": 4.222856297095912e-06,
|
| 8807 |
+
"loss": 1.6044,
|
| 8808 |
+
"step": 1257
|
| 8809 |
+
},
|
| 8810 |
+
{
|
| 8811 |
+
"epoch": 0.05956157378911983,
|
| 8812 |
+
"grad_norm": 1.16346275806427,
|
| 8813 |
+
"learning_rate": 4.179150386076424e-06,
|
| 8814 |
+
"loss": 1.5277,
|
| 8815 |
+
"step": 1258
|
| 8816 |
+
},
|
| 8817 |
+
{
|
| 8818 |
+
"epoch": 0.059608920032195443,
|
| 8819 |
+
"grad_norm": 1.077487826347351,
|
| 8820 |
+
"learning_rate": 4.135661961499987e-06,
|
| 8821 |
+
"loss": 1.4139,
|
| 8822 |
+
"step": 1259
|
| 8823 |
+
},
|
| 8824 |
+
{
|
| 8825 |
+
"epoch": 0.059656266275271055,
|
| 8826 |
+
"grad_norm": 1.3056800365447998,
|
| 8827 |
+
"learning_rate": 4.0923912297822775e-06,
|
| 8828 |
+
"loss": 1.5478,
|
| 8829 |
+
"step": 1260
|
| 8830 |
+
},
|
| 8831 |
+
{
|
| 8832 |
+
"epoch": 0.05970361251834667,
|
| 8833 |
+
"grad_norm": 1.5615487098693848,
|
| 8834 |
+
"learning_rate": 4.049338396305735e-06,
|
| 8835 |
+
"loss": 1.4689,
|
| 8836 |
+
"step": 1261
|
| 8837 |
+
},
|
| 8838 |
+
{
|
| 8839 |
+
"epoch": 0.05975095876142228,
|
| 8840 |
+
"grad_norm": 1.1437926292419434,
|
| 8841 |
+
"learning_rate": 4.006503665418554e-06,
|
| 8842 |
+
"loss": 1.4132,
|
| 8843 |
+
"step": 1262
|
| 8844 |
+
},
|
| 8845 |
+
{
|
| 8846 |
+
"epoch": 0.05979830500449789,
|
| 8847 |
+
"grad_norm": 1.14958918094635,
|
| 8848 |
+
"learning_rate": 3.963887240433706e-06,
|
| 8849 |
+
"loss": 1.4433,
|
| 8850 |
+
"step": 1263
|
| 8851 |
+
},
|
| 8852 |
+
{
|
| 8853 |
+
"epoch": 0.0598456512475735,
|
| 8854 |
+
"grad_norm": 1.3017897605895996,
|
| 8855 |
+
"learning_rate": 3.921489323627975e-06,
|
| 8856 |
+
"loss": 1.5105,
|
| 8857 |
+
"step": 1264
|
| 8858 |
+
},
|
| 8859 |
+
{
|
| 8860 |
+
"epoch": 0.059892997490649115,
|
| 8861 |
+
"grad_norm": 1.1352862119674683,
|
| 8862 |
+
"learning_rate": 3.879310116241042e-06,
|
| 8863 |
+
"loss": 1.4971,
|
| 8864 |
+
"step": 1265
|
| 8865 |
+
},
|
| 8866 |
+
{
|
| 8867 |
+
"epoch": 0.05994034373372473,
|
| 8868 |
+
"grad_norm": 1.2571970224380493,
|
| 8869 |
+
"learning_rate": 3.837349818474456e-06,
|
| 8870 |
+
"loss": 1.543,
|
| 8871 |
+
"step": 1266
|
| 8872 |
+
},
|
| 8873 |
+
{
|
| 8874 |
+
"epoch": 0.05998768997680034,
|
| 8875 |
+
"grad_norm": 1.245848298072815,
|
| 8876 |
+
"learning_rate": 3.795608629490738e-06,
|
| 8877 |
+
"loss": 1.4303,
|
| 8878 |
+
"step": 1267
|
| 8879 |
+
},
|
| 8880 |
+
{
|
| 8881 |
+
"epoch": 0.06003503621987595,
|
| 8882 |
+
"grad_norm": 1.1832315921783447,
|
| 8883 |
+
"learning_rate": 3.754086747412422e-06,
|
| 8884 |
+
"loss": 1.4129,
|
| 8885 |
+
"step": 1268
|
| 8886 |
+
},
|
| 8887 |
+
{
|
| 8888 |
+
"epoch": 0.06008238246295156,
|
| 8889 |
+
"grad_norm": 1.2564343214035034,
|
| 8890 |
+
"learning_rate": 3.7127843693211206e-06,
|
| 8891 |
+
"loss": 1.367,
|
| 8892 |
+
"step": 1269
|
| 8893 |
+
},
|
| 8894 |
+
{
|
| 8895 |
+
"epoch": 0.060129728706027175,
|
| 8896 |
+
"grad_norm": 1.4539203643798828,
|
| 8897 |
+
"learning_rate": 3.671701691256557e-06,
|
| 8898 |
+
"loss": 1.6603,
|
| 8899 |
+
"step": 1270
|
| 8900 |
+
},
|
| 8901 |
+
{
|
| 8902 |
+
"epoch": 0.060177074949102786,
|
| 8903 |
+
"grad_norm": 1.374998688697815,
|
| 8904 |
+
"learning_rate": 3.6308389082156835e-06,
|
| 8905 |
+
"loss": 1.5778,
|
| 8906 |
+
"step": 1271
|
| 8907 |
+
},
|
| 8908 |
+
{
|
| 8909 |
+
"epoch": 0.0602244211921784,
|
| 8910 |
+
"grad_norm": 1.2109373807907104,
|
| 8911 |
+
"learning_rate": 3.5901962141516977e-06,
|
| 8912 |
+
"loss": 1.4751,
|
| 8913 |
+
"step": 1272
|
| 8914 |
+
},
|
| 8915 |
+
{
|
| 8916 |
+
"epoch": 0.06027176743525401,
|
| 8917 |
+
"grad_norm": 1.3299740552902222,
|
| 8918 |
+
"learning_rate": 3.549773801973211e-06,
|
| 8919 |
+
"loss": 1.4164,
|
| 8920 |
+
"step": 1273
|
| 8921 |
+
},
|
| 8922 |
+
{
|
| 8923 |
+
"epoch": 0.06031911367832962,
|
| 8924 |
+
"grad_norm": 1.5119253396987915,
|
| 8925 |
+
"learning_rate": 3.509571863543221e-06,
|
| 8926 |
+
"loss": 1.4562,
|
| 8927 |
+
"step": 1274
|
| 8928 |
+
},
|
| 8929 |
+
{
|
| 8930 |
+
"epoch": 0.060366459921405234,
|
| 8931 |
+
"grad_norm": 1.5190602540969849,
|
| 8932 |
+
"learning_rate": 3.4695905896782843e-06,
|
| 8933 |
+
"loss": 1.4357,
|
| 8934 |
+
"step": 1275
|
| 8935 |
+
},
|
| 8936 |
+
{
|
| 8937 |
+
"epoch": 0.060413806164480846,
|
| 8938 |
+
"grad_norm": 1.4870779514312744,
|
| 8939 |
+
"learning_rate": 3.4298301701475855e-06,
|
| 8940 |
+
"loss": 1.5516,
|
| 8941 |
+
"step": 1276
|
| 8942 |
+
},
|
| 8943 |
+
{
|
| 8944 |
+
"epoch": 0.06046115240755646,
|
| 8945 |
+
"grad_norm": 1.4861438274383545,
|
| 8946 |
+
"learning_rate": 3.390290793672035e-06,
|
| 8947 |
+
"loss": 1.4489,
|
| 8948 |
+
"step": 1277
|
| 8949 |
+
},
|
| 8950 |
+
{
|
| 8951 |
+
"epoch": 0.06050849865063207,
|
| 8952 |
+
"grad_norm": 1.3863167762756348,
|
| 8953 |
+
"learning_rate": 3.3509726479233504e-06,
|
| 8954 |
+
"loss": 1.3034,
|
| 8955 |
+
"step": 1278
|
| 8956 |
+
},
|
| 8957 |
+
{
|
| 8958 |
+
"epoch": 0.06055584489370768,
|
| 8959 |
+
"grad_norm": 1.258088231086731,
|
| 8960 |
+
"learning_rate": 3.3118759195232275e-06,
|
| 8961 |
+
"loss": 1.4765,
|
| 8962 |
+
"step": 1279
|
| 8963 |
+
},
|
| 8964 |
+
{
|
| 8965 |
+
"epoch": 0.060603191136783294,
|
| 8966 |
+
"grad_norm": 1.3993644714355469,
|
| 8967 |
+
"learning_rate": 3.27300079404238e-06,
|
| 8968 |
+
"loss": 1.5346,
|
| 8969 |
+
"step": 1280
|
| 8970 |
+
},
|
| 8971 |
+
{
|
| 8972 |
+
"epoch": 0.060650537379858906,
|
| 8973 |
+
"grad_norm": 1.2492510080337524,
|
| 8974 |
+
"learning_rate": 3.234347455999709e-06,
|
| 8975 |
+
"loss": 1.3976,
|
| 8976 |
+
"step": 1281
|
| 8977 |
+
},
|
| 8978 |
+
{
|
| 8979 |
+
"epoch": 0.06069788362293452,
|
| 8980 |
+
"grad_norm": 1.2434810400009155,
|
| 8981 |
+
"learning_rate": 3.195916088861406e-06,
|
| 8982 |
+
"loss": 1.3539,
|
| 8983 |
+
"step": 1282
|
| 8984 |
+
},
|
| 8985 |
+
{
|
| 8986 |
+
"epoch": 0.06074522986601013,
|
| 8987 |
+
"grad_norm": 1.376212477684021,
|
| 8988 |
+
"learning_rate": 3.1577068750401116e-06,
|
| 8989 |
+
"loss": 1.5787,
|
| 8990 |
+
"step": 1283
|
| 8991 |
+
},
|
| 8992 |
+
{
|
| 8993 |
+
"epoch": 0.06079257610908574,
|
| 8994 |
+
"grad_norm": 1.2925554513931274,
|
| 8995 |
+
"learning_rate": 3.119719995893994e-06,
|
| 8996 |
+
"loss": 1.3657,
|
| 8997 |
+
"step": 1284
|
| 8998 |
+
},
|
| 8999 |
+
{
|
| 9000 |
+
"epoch": 0.06083992235216135,
|
| 9001 |
+
"grad_norm": 1.2235181331634521,
|
| 9002 |
+
"learning_rate": 3.08195563172593e-06,
|
| 9003 |
+
"loss": 1.3284,
|
| 9004 |
+
"step": 1285
|
| 9005 |
+
},
|
| 9006 |
+
{
|
| 9007 |
+
"epoch": 0.060887268595236965,
|
| 9008 |
+
"grad_norm": 1.3936784267425537,
|
| 9009 |
+
"learning_rate": 3.0444139617826607e-06,
|
| 9010 |
+
"loss": 1.6215,
|
| 9011 |
+
"step": 1286
|
| 9012 |
+
},
|
| 9013 |
+
{
|
| 9014 |
+
"epoch": 0.06093461483831258,
|
| 9015 |
+
"grad_norm": 1.3355106115341187,
|
| 9016 |
+
"learning_rate": 3.0070951642538924e-06,
|
| 9017 |
+
"loss": 1.3555,
|
| 9018 |
+
"step": 1287
|
| 9019 |
+
},
|
| 9020 |
+
{
|
| 9021 |
+
"epoch": 0.06098196108138819,
|
| 9022 |
+
"grad_norm": 1.4832301139831543,
|
| 9023 |
+
"learning_rate": 2.969999416271496e-06,
|
| 9024 |
+
"loss": 1.5066,
|
| 9025 |
+
"step": 1288
|
| 9026 |
+
},
|
| 9027 |
+
{
|
| 9028 |
+
"epoch": 0.0610293073244638,
|
| 9029 |
+
"grad_norm": 1.284655213356018,
|
| 9030 |
+
"learning_rate": 2.933126893908633e-06,
|
| 9031 |
+
"loss": 1.3608,
|
| 9032 |
+
"step": 1289
|
| 9033 |
+
},
|
| 9034 |
+
{
|
| 9035 |
+
"epoch": 0.06107665356753941,
|
| 9036 |
+
"grad_norm": 1.4563186168670654,
|
| 9037 |
+
"learning_rate": 2.8964777721789573e-06,
|
| 9038 |
+
"loss": 1.3335,
|
| 9039 |
+
"step": 1290
|
| 9040 |
+
},
|
| 9041 |
+
{
|
| 9042 |
+
"epoch": 0.061123999810615025,
|
| 9043 |
+
"grad_norm": 1.2349377870559692,
|
| 9044 |
+
"learning_rate": 2.860052225035742e-06,
|
| 9045 |
+
"loss": 1.2591,
|
| 9046 |
+
"step": 1291
|
| 9047 |
+
},
|
| 9048 |
+
{
|
| 9049 |
+
"epoch": 0.06117134605369064,
|
| 9050 |
+
"grad_norm": 1.5449947118759155,
|
| 9051 |
+
"learning_rate": 2.823850425371083e-06,
|
| 9052 |
+
"loss": 1.6179,
|
| 9053 |
+
"step": 1292
|
| 9054 |
+
},
|
| 9055 |
+
{
|
| 9056 |
+
"epoch": 0.06121869229676625,
|
| 9057 |
+
"grad_norm": 1.526613473892212,
|
| 9058 |
+
"learning_rate": 2.787872545015069e-06,
|
| 9059 |
+
"loss": 1.4801,
|
| 9060 |
+
"step": 1293
|
| 9061 |
+
},
|
| 9062 |
+
{
|
| 9063 |
+
"epoch": 0.06126603853984186,
|
| 9064 |
+
"grad_norm": 1.374595046043396,
|
| 9065 |
+
"learning_rate": 2.752118754734967e-06,
|
| 9066 |
+
"loss": 1.4021,
|
| 9067 |
+
"step": 1294
|
| 9068 |
+
},
|
| 9069 |
+
{
|
| 9070 |
+
"epoch": 0.06131338478291747,
|
| 9071 |
+
"grad_norm": 1.4395325183868408,
|
| 9072 |
+
"learning_rate": 2.716589224234406e-06,
|
| 9073 |
+
"loss": 1.3826,
|
| 9074 |
+
"step": 1295
|
| 9075 |
+
},
|
| 9076 |
+
{
|
| 9077 |
+
"epoch": 0.061360731025993084,
|
| 9078 |
+
"grad_norm": 1.3016366958618164,
|
| 9079 |
+
"learning_rate": 2.681284122152594e-06,
|
| 9080 |
+
"loss": 1.0532,
|
| 9081 |
+
"step": 1296
|
| 9082 |
+
},
|
| 9083 |
+
{
|
| 9084 |
+
"epoch": 0.061408077269068696,
|
| 9085 |
+
"grad_norm": 1.3789076805114746,
|
| 9086 |
+
"learning_rate": 2.646203616063475e-06,
|
| 9087 |
+
"loss": 1.2517,
|
| 9088 |
+
"step": 1297
|
| 9089 |
+
},
|
| 9090 |
+
{
|
| 9091 |
+
"epoch": 0.06145542351214431,
|
| 9092 |
+
"grad_norm": 1.6451598405838013,
|
| 9093 |
+
"learning_rate": 2.6113478724749783e-06,
|
| 9094 |
+
"loss": 1.3242,
|
| 9095 |
+
"step": 1298
|
| 9096 |
+
},
|
| 9097 |
+
{
|
| 9098 |
+
"epoch": 0.06150276975521992,
|
| 9099 |
+
"grad_norm": 1.382812261581421,
|
| 9100 |
+
"learning_rate": 2.5767170568281927e-06,
|
| 9101 |
+
"loss": 1.3016,
|
| 9102 |
+
"step": 1299
|
| 9103 |
+
},
|
| 9104 |
+
{
|
| 9105 |
+
"epoch": 0.06155011599829553,
|
| 9106 |
+
"grad_norm": 1.5659140348434448,
|
| 9107 |
+
"learning_rate": 2.542311333496622e-06,
|
| 9108 |
+
"loss": 1.3122,
|
| 9109 |
+
"step": 1300
|
| 9110 |
+
},
|
| 9111 |
+
{
|
| 9112 |
+
"epoch": 0.061597462241371144,
|
| 9113 |
+
"grad_norm": 1.1755375862121582,
|
| 9114 |
+
"learning_rate": 2.5081308657853573e-06,
|
| 9115 |
+
"loss": 1.2939,
|
| 9116 |
+
"step": 1301
|
| 9117 |
+
},
|
| 9118 |
+
{
|
| 9119 |
+
"epoch": 0.061644808484446756,
|
| 9120 |
+
"grad_norm": 1.0510395765304565,
|
| 9121 |
+
"learning_rate": 2.4741758159303242e-06,
|
| 9122 |
+
"loss": 1.4157,
|
| 9123 |
+
"step": 1302
|
| 9124 |
+
},
|
| 9125 |
+
{
|
| 9126 |
+
"epoch": 0.06169215472752237,
|
| 9127 |
+
"grad_norm": 1.2231727838516235,
|
| 9128 |
+
"learning_rate": 2.440446345097541e-06,
|
| 9129 |
+
"loss": 1.454,
|
| 9130 |
+
"step": 1303
|
| 9131 |
+
},
|
| 9132 |
+
{
|
| 9133 |
+
"epoch": 0.06173950097059798,
|
| 9134 |
+
"grad_norm": 1.2164487838745117,
|
| 9135 |
+
"learning_rate": 2.4069426133822957e-06,
|
| 9136 |
+
"loss": 1.5866,
|
| 9137 |
+
"step": 1304
|
| 9138 |
+
},
|
| 9139 |
+
{
|
| 9140 |
+
"epoch": 0.06178684721367359,
|
| 9141 |
+
"grad_norm": 1.2815780639648438,
|
| 9142 |
+
"learning_rate": 2.3736647798084265e-06,
|
| 9143 |
+
"loss": 1.5007,
|
| 9144 |
+
"step": 1305
|
| 9145 |
+
},
|
| 9146 |
+
{
|
| 9147 |
+
"epoch": 0.061834193456749204,
|
| 9148 |
+
"grad_norm": 1.309321641921997,
|
| 9149 |
+
"learning_rate": 2.340613002327563e-06,
|
| 9150 |
+
"loss": 1.5988,
|
| 9151 |
+
"step": 1306
|
| 9152 |
+
},
|
| 9153 |
+
{
|
| 9154 |
+
"epoch": 0.061881539699824815,
|
| 9155 |
+
"grad_norm": 1.3088290691375732,
|
| 9156 |
+
"learning_rate": 2.307787437818365e-06,
|
| 9157 |
+
"loss": 1.5639,
|
| 9158 |
+
"step": 1307
|
| 9159 |
+
},
|
| 9160 |
+
{
|
| 9161 |
+
"epoch": 0.061928885942900434,
|
| 9162 |
+
"grad_norm": 1.0919935703277588,
|
| 9163 |
+
"learning_rate": 2.275188242085785e-06,
|
| 9164 |
+
"loss": 1.3272,
|
| 9165 |
+
"step": 1308
|
| 9166 |
+
},
|
| 9167 |
+
{
|
| 9168 |
+
"epoch": 0.061976232185976046,
|
| 9169 |
+
"grad_norm": 1.40773606300354,
|
| 9170 |
+
"learning_rate": 2.2428155698603182e-06,
|
| 9171 |
+
"loss": 1.5226,
|
| 9172 |
+
"step": 1309
|
| 9173 |
+
},
|
| 9174 |
+
{
|
| 9175 |
+
"epoch": 0.06202357842905166,
|
| 9176 |
+
"grad_norm": 1.0675700902938843,
|
| 9177 |
+
"learning_rate": 2.2106695747973026e-06,
|
| 9178 |
+
"loss": 1.3541,
|
| 9179 |
+
"step": 1310
|
| 9180 |
+
},
|
| 9181 |
+
{
|
| 9182 |
+
"epoch": 0.06207092467212727,
|
| 9183 |
+
"grad_norm": 1.1566540002822876,
|
| 9184 |
+
"learning_rate": 2.1787504094761267e-06,
|
| 9185 |
+
"loss": 1.4108,
|
| 9186 |
+
"step": 1311
|
| 9187 |
+
},
|
| 9188 |
+
{
|
| 9189 |
+
"epoch": 0.06211827091520288,
|
| 9190 |
+
"grad_norm": 1.3144512176513672,
|
| 9191 |
+
"learning_rate": 2.147058225399573e-06,
|
| 9192 |
+
"loss": 1.5544,
|
| 9193 |
+
"step": 1312
|
| 9194 |
+
},
|
| 9195 |
+
{
|
| 9196 |
+
"epoch": 0.062165617158278494,
|
| 9197 |
+
"grad_norm": 1.1639258861541748,
|
| 9198 |
+
"learning_rate": 2.115593172993058e-06,
|
| 9199 |
+
"loss": 1.4059,
|
| 9200 |
+
"step": 1313
|
| 9201 |
+
},
|
| 9202 |
+
{
|
| 9203 |
+
"epoch": 0.062212963401354106,
|
| 9204 |
+
"grad_norm": 1.3233219385147095,
|
| 9205 |
+
"learning_rate": 2.0843554016039326e-06,
|
| 9206 |
+
"loss": 1.5817,
|
| 9207 |
+
"step": 1314
|
| 9208 |
+
},
|
| 9209 |
+
{
|
| 9210 |
+
"epoch": 0.06226030964442972,
|
| 9211 |
+
"grad_norm": 1.2567836046218872,
|
| 9212 |
+
"learning_rate": 2.0533450595007498e-06,
|
| 9213 |
+
"loss": 1.5539,
|
| 9214 |
+
"step": 1315
|
| 9215 |
+
},
|
| 9216 |
+
{
|
| 9217 |
+
"epoch": 0.06230765588750533,
|
| 9218 |
+
"grad_norm": 1.1879708766937256,
|
| 9219 |
+
"learning_rate": 2.0225622938726098e-06,
|
| 9220 |
+
"loss": 1.4033,
|
| 9221 |
+
"step": 1316
|
| 9222 |
+
},
|
| 9223 |
+
{
|
| 9224 |
+
"epoch": 0.06235500213058094,
|
| 9225 |
+
"grad_norm": 1.15211021900177,
|
| 9226 |
+
"learning_rate": 1.9920072508284202e-06,
|
| 9227 |
+
"loss": 1.6171,
|
| 9228 |
+
"step": 1317
|
| 9229 |
+
},
|
| 9230 |
+
{
|
| 9231 |
+
"epoch": 0.06240234837365655,
|
| 9232 |
+
"grad_norm": 1.1985636949539185,
|
| 9233 |
+
"learning_rate": 1.961680075396205e-06,
|
| 9234 |
+
"loss": 1.6475,
|
| 9235 |
+
"step": 1318
|
| 9236 |
+
},
|
| 9237 |
+
{
|
| 9238 |
+
"epoch": 0.062449694616732165,
|
| 9239 |
+
"grad_norm": 1.2133159637451172,
|
| 9240 |
+
"learning_rate": 1.931580911522435e-06,
|
| 9241 |
+
"loss": 1.3112,
|
| 9242 |
+
"step": 1319
|
| 9243 |
+
},
|
| 9244 |
+
{
|
| 9245 |
+
"epoch": 0.06249704085980778,
|
| 9246 |
+
"grad_norm": 1.2147009372711182,
|
| 9247 |
+
"learning_rate": 1.901709902071336e-06,
|
| 9248 |
+
"loss": 1.563,
|
| 9249 |
+
"step": 1320
|
| 9250 |
+
},
|
| 9251 |
+
{
|
| 9252 |
+
"epoch": 0.06254438710288339,
|
| 9253 |
+
"grad_norm": 1.1709462404251099,
|
| 9254 |
+
"learning_rate": 1.8720671888242059e-06,
|
| 9255 |
+
"loss": 1.3305,
|
| 9256 |
+
"step": 1321
|
| 9257 |
+
},
|
| 9258 |
+
{
|
| 9259 |
+
"epoch": 0.062591733345959,
|
| 9260 |
+
"grad_norm": 1.312578797340393,
|
| 9261 |
+
"learning_rate": 1.8426529124787373e-06,
|
| 9262 |
+
"loss": 1.4542,
|
| 9263 |
+
"step": 1322
|
| 9264 |
+
},
|
| 9265 |
+
{
|
| 9266 |
+
"epoch": 0.06263907958903461,
|
| 9267 |
+
"grad_norm": 1.2657169103622437,
|
| 9268 |
+
"learning_rate": 1.81346721264839e-06,
|
| 9269 |
+
"loss": 1.4211,
|
| 9270 |
+
"step": 1323
|
| 9271 |
+
},
|
| 9272 |
+
{
|
| 9273 |
+
"epoch": 0.06268642583211022,
|
| 9274 |
+
"grad_norm": 1.2395094633102417,
|
| 9275 |
+
"learning_rate": 1.7845102278616576e-06,
|
| 9276 |
+
"loss": 1.3067,
|
| 9277 |
+
"step": 1324
|
| 9278 |
+
},
|
| 9279 |
+
{
|
| 9280 |
+
"epoch": 0.06273377207518584,
|
| 9281 |
+
"grad_norm": 1.4858688116073608,
|
| 9282 |
+
"learning_rate": 1.7557820955614701e-06,
|
| 9283 |
+
"loss": 1.5523,
|
| 9284 |
+
"step": 1325
|
| 9285 |
+
},
|
| 9286 |
+
{
|
| 9287 |
+
"epoch": 0.06278111831826144,
|
| 9288 |
+
"grad_norm": 1.1889420747756958,
|
| 9289 |
+
"learning_rate": 1.727282952104514e-06,
|
| 9290 |
+
"loss": 1.5204,
|
| 9291 |
+
"step": 1326
|
| 9292 |
+
},
|
| 9293 |
+
{
|
| 9294 |
+
"epoch": 0.06282846456133706,
|
| 9295 |
+
"grad_norm": 1.2964903116226196,
|
| 9296 |
+
"learning_rate": 1.699012932760602e-06,
|
| 9297 |
+
"loss": 1.3605,
|
| 9298 |
+
"step": 1327
|
| 9299 |
+
},
|
| 9300 |
+
{
|
| 9301 |
+
"epoch": 0.06287581080441267,
|
| 9302 |
+
"grad_norm": 1.305997371673584,
|
| 9303 |
+
"learning_rate": 1.6709721717120042e-06,
|
| 9304 |
+
"loss": 1.5661,
|
| 9305 |
+
"step": 1328
|
| 9306 |
+
},
|
| 9307 |
+
{
|
| 9308 |
+
"epoch": 0.06292315704748828,
|
| 9309 |
+
"grad_norm": 1.6361205577850342,
|
| 9310 |
+
"learning_rate": 1.6431608020528232e-06,
|
| 9311 |
+
"loss": 1.3802,
|
| 9312 |
+
"step": 1329
|
| 9313 |
+
},
|
| 9314 |
+
{
|
| 9315 |
+
"epoch": 0.06297050329056389,
|
| 9316 |
+
"grad_norm": 1.2052884101867676,
|
| 9317 |
+
"learning_rate": 1.6155789557883992e-06,
|
| 9318 |
+
"loss": 1.3603,
|
| 9319 |
+
"step": 1330
|
| 9320 |
+
},
|
| 9321 |
+
{
|
| 9322 |
+
"epoch": 0.06301784953363951,
|
| 9323 |
+
"grad_norm": 1.2012227773666382,
|
| 9324 |
+
"learning_rate": 1.5882267638346215e-06,
|
| 9325 |
+
"loss": 1.4148,
|
| 9326 |
+
"step": 1331
|
| 9327 |
+
},
|
| 9328 |
+
{
|
| 9329 |
+
"epoch": 0.06306519577671511,
|
| 9330 |
+
"grad_norm": 1.5315930843353271,
|
| 9331 |
+
"learning_rate": 1.5611043560173344e-06,
|
| 9332 |
+
"loss": 1.5159,
|
| 9333 |
+
"step": 1332
|
| 9334 |
+
},
|
| 9335 |
+
{
|
| 9336 |
+
"epoch": 0.06311254201979073,
|
| 9337 |
+
"grad_norm": 1.3208889961242676,
|
| 9338 |
+
"learning_rate": 1.5342118610717437e-06,
|
| 9339 |
+
"loss": 1.3613,
|
| 9340 |
+
"step": 1333
|
| 9341 |
+
},
|
| 9342 |
+
{
|
| 9343 |
+
"epoch": 0.06315988826286634,
|
| 9344 |
+
"grad_norm": 1.3040242195129395,
|
| 9345 |
+
"learning_rate": 1.5075494066417728e-06,
|
| 9346 |
+
"loss": 1.2185,
|
| 9347 |
+
"step": 1334
|
| 9348 |
+
},
|
| 9349 |
+
{
|
| 9350 |
+
"epoch": 0.06320723450594196,
|
| 9351 |
+
"grad_norm": 1.3122460842132568,
|
| 9352 |
+
"learning_rate": 1.4811171192794627e-06,
|
| 9353 |
+
"loss": 1.3611,
|
| 9354 |
+
"step": 1335
|
| 9355 |
+
},
|
| 9356 |
+
{
|
| 9357 |
+
"epoch": 0.06325458074901756,
|
| 9358 |
+
"grad_norm": 1.2624602317810059,
|
| 9359 |
+
"learning_rate": 1.4549151244444003e-06,
|
| 9360 |
+
"loss": 1.3265,
|
| 9361 |
+
"step": 1336
|
| 9362 |
+
},
|
| 9363 |
+
{
|
| 9364 |
+
"epoch": 0.06330192699209318,
|
| 9365 |
+
"grad_norm": 1.3999385833740234,
|
| 9366 |
+
"learning_rate": 1.4289435465030698e-06,
|
| 9367 |
+
"loss": 1.4897,
|
| 9368 |
+
"step": 1337
|
| 9369 |
+
},
|
| 9370 |
+
{
|
| 9371 |
+
"epoch": 0.06334927323516878,
|
| 9372 |
+
"grad_norm": 1.3911713361740112,
|
| 9373 |
+
"learning_rate": 1.403202508728313e-06,
|
| 9374 |
+
"loss": 1.3059,
|
| 9375 |
+
"step": 1338
|
| 9376 |
+
},
|
| 9377 |
+
{
|
| 9378 |
+
"epoch": 0.0633966194782444,
|
| 9379 |
+
"grad_norm": 1.3101698160171509,
|
| 9380 |
+
"learning_rate": 1.3776921332987192e-06,
|
| 9381 |
+
"loss": 1.2585,
|
| 9382 |
+
"step": 1339
|
| 9383 |
+
},
|
| 9384 |
+
{
|
| 9385 |
+
"epoch": 0.06344396572132001,
|
| 9386 |
+
"grad_norm": 1.3890515565872192,
|
| 9387 |
+
"learning_rate": 1.3524125412980483e-06,
|
| 9388 |
+
"loss": 1.4638,
|
| 9389 |
+
"step": 1340
|
| 9390 |
+
},
|
| 9391 |
+
{
|
| 9392 |
+
"epoch": 0.06349131196439563,
|
| 9393 |
+
"grad_norm": 1.5259530544281006,
|
| 9394 |
+
"learning_rate": 1.3273638527146637e-06,
|
| 9395 |
+
"loss": 1.3657,
|
| 9396 |
+
"step": 1341
|
| 9397 |
+
},
|
| 9398 |
+
{
|
| 9399 |
+
"epoch": 0.06353865820747123,
|
| 9400 |
+
"grad_norm": 1.3829599618911743,
|
| 9401 |
+
"learning_rate": 1.3025461864409394e-06,
|
| 9402 |
+
"loss": 1.4408,
|
| 9403 |
+
"step": 1342
|
| 9404 |
+
},
|
| 9405 |
+
{
|
| 9406 |
+
"epoch": 0.06358600445054685,
|
| 9407 |
+
"grad_norm": 1.3275704383850098,
|
| 9408 |
+
"learning_rate": 1.2779596602727428e-06,
|
| 9409 |
+
"loss": 1.5234,
|
| 9410 |
+
"step": 1343
|
| 9411 |
+
},
|
| 9412 |
+
{
|
| 9413 |
+
"epoch": 0.06363335069362246,
|
| 9414 |
+
"grad_norm": 1.581809401512146,
|
| 9415 |
+
"learning_rate": 1.2536043909088191e-06,
|
| 9416 |
+
"loss": 1.443,
|
| 9417 |
+
"step": 1344
|
| 9418 |
+
},
|
| 9419 |
+
{
|
| 9420 |
+
"epoch": 0.06368069693669808,
|
| 9421 |
+
"grad_norm": 1.5524332523345947,
|
| 9422 |
+
"learning_rate": 1.2294804939502747e-06,
|
| 9423 |
+
"loss": 1.585,
|
| 9424 |
+
"step": 1345
|
| 9425 |
+
},
|
| 9426 |
+
{
|
| 9427 |
+
"epoch": 0.06372804317977368,
|
| 9428 |
+
"grad_norm": 1.5161685943603516,
|
| 9429 |
+
"learning_rate": 1.2055880839000278e-06,
|
| 9430 |
+
"loss": 1.4337,
|
| 9431 |
+
"step": 1346
|
| 9432 |
+
},
|
| 9433 |
+
{
|
| 9434 |
+
"epoch": 0.0637753894228493,
|
| 9435 |
+
"grad_norm": 1.3019859790802002,
|
| 9436 |
+
"learning_rate": 1.1819272741622368e-06,
|
| 9437 |
+
"loss": 1.2617,
|
| 9438 |
+
"step": 1347
|
| 9439 |
+
},
|
| 9440 |
+
{
|
| 9441 |
+
"epoch": 0.0638227356659249,
|
| 9442 |
+
"grad_norm": 1.8352184295654297,
|
| 9443 |
+
"learning_rate": 1.158498177041789e-06,
|
| 9444 |
+
"loss": 1.2899,
|
| 9445 |
+
"step": 1348
|
| 9446 |
+
},
|
| 9447 |
+
{
|
| 9448 |
+
"epoch": 0.06387008190900052,
|
| 9449 |
+
"grad_norm": 1.4662976264953613,
|
| 9450 |
+
"learning_rate": 1.1353009037437523e-06,
|
| 9451 |
+
"loss": 1.2421,
|
| 9452 |
+
"step": 1349
|
| 9453 |
+
},
|
| 9454 |
+
{
|
| 9455 |
+
"epoch": 0.06391742815207613,
|
| 9456 |
+
"grad_norm": 1.6190884113311768,
|
| 9457 |
+
"learning_rate": 1.1123355643728795e-06,
|
| 9458 |
+
"loss": 1.2181,
|
| 9459 |
+
"step": 1350
|
| 9460 |
+
},
|
| 9461 |
+
{
|
| 9462 |
+
"epoch": 0.06396477439515175,
|
| 9463 |
+
"grad_norm": 1.2453593015670776,
|
| 9464 |
+
"learning_rate": 1.0896022679330265e-06,
|
| 9465 |
+
"loss": 1.8871,
|
| 9466 |
+
"step": 1351
|
| 9467 |
+
},
|
| 9468 |
+
{
|
| 9469 |
+
"epoch": 0.06401212063822735,
|
| 9470 |
+
"grad_norm": 1.1308292150497437,
|
| 9471 |
+
"learning_rate": 1.0671011223266802e-06,
|
| 9472 |
+
"loss": 1.7013,
|
| 9473 |
+
"step": 1352
|
| 9474 |
+
},
|
| 9475 |
+
{
|
| 9476 |
+
"epoch": 0.06405946688130297,
|
| 9477 |
+
"grad_norm": 1.060587763786316,
|
| 9478 |
+
"learning_rate": 1.0448322343544536e-06,
|
| 9479 |
+
"loss": 1.6145,
|
| 9480 |
+
"step": 1353
|
| 9481 |
+
},
|
| 9482 |
+
{
|
| 9483 |
+
"epoch": 0.06410681312437858,
|
| 9484 |
+
"grad_norm": 1.179922103881836,
|
| 9485 |
+
"learning_rate": 1.0227957097145413e-06,
|
| 9486 |
+
"loss": 1.4764,
|
| 9487 |
+
"step": 1354
|
| 9488 |
+
},
|
| 9489 |
+
{
|
| 9490 |
+
"epoch": 0.0641541593674542,
|
| 9491 |
+
"grad_norm": 1.139797568321228,
|
| 9492 |
+
"learning_rate": 1.0009916530022257e-06,
|
| 9493 |
+
"loss": 1.3325,
|
| 9494 |
+
"step": 1355
|
| 9495 |
+
},
|
| 9496 |
+
{
|
| 9497 |
+
"epoch": 0.0642015056105298,
|
| 9498 |
+
"grad_norm": 1.1278554201126099,
|
| 9499 |
+
"learning_rate": 9.794201677094162e-07,
|
| 9500 |
+
"loss": 1.4488,
|
| 9501 |
+
"step": 1356
|
| 9502 |
+
},
|
| 9503 |
+
{
|
| 9504 |
+
"epoch": 0.06424885185360542,
|
| 9505 |
+
"grad_norm": 1.1881204843521118,
|
| 9506 |
+
"learning_rate": 9.580813562241276e-07,
|
| 9507 |
+
"loss": 1.4232,
|
| 9508 |
+
"step": 1357
|
| 9509 |
+
},
|
| 9510 |
+
{
|
| 9511 |
+
"epoch": 0.06429619809668102,
|
| 9512 |
+
"grad_norm": 1.368537425994873,
|
| 9513 |
+
"learning_rate": 9.369753198299858e-07,
|
| 9514 |
+
"loss": 1.7749,
|
| 9515 |
+
"step": 1358
|
| 9516 |
+
},
|
| 9517 |
+
{
|
| 9518 |
+
"epoch": 0.06434354433975664,
|
| 9519 |
+
"grad_norm": 1.2621606588363647,
|
| 9520 |
+
"learning_rate": 9.161021587057727e-07,
|
| 9521 |
+
"loss": 1.5111,
|
| 9522 |
+
"step": 1359
|
| 9523 |
+
},
|
| 9524 |
+
{
|
| 9525 |
+
"epoch": 0.06439089058283225,
|
| 9526 |
+
"grad_norm": 1.2346223592758179,
|
| 9527 |
+
"learning_rate": 8.95461971924938e-07,
|
| 9528 |
+
"loss": 1.5364,
|
| 9529 |
+
"step": 1360
|
| 9530 |
+
},
|
| 9531 |
+
{
|
| 9532 |
+
"epoch": 0.06443823682590787,
|
| 9533 |
+
"grad_norm": 1.2565935850143433,
|
| 9534 |
+
"learning_rate": 8.75054857455132e-07,
|
| 9535 |
+
"loss": 1.4551,
|
| 9536 |
+
"step": 1361
|
| 9537 |
+
},
|
| 9538 |
+
{
|
| 9539 |
+
"epoch": 0.06448558306898347,
|
| 9540 |
+
"grad_norm": 1.35322904586792,
|
| 9541 |
+
"learning_rate": 8.54880912157724e-07,
|
| 9542 |
+
"loss": 1.7903,
|
| 9543 |
+
"step": 1362
|
| 9544 |
+
},
|
| 9545 |
+
{
|
| 9546 |
+
"epoch": 0.06453292931205909,
|
| 9547 |
+
"grad_norm": 1.1911216974258423,
|
| 9548 |
+
"learning_rate": 8.349402317873789e-07,
|
| 9549 |
+
"loss": 1.3712,
|
| 9550 |
+
"step": 1363
|
| 9551 |
+
},
|
| 9552 |
+
{
|
| 9553 |
+
"epoch": 0.0645802755551347,
|
| 9554 |
+
"grad_norm": 1.4313905239105225,
|
| 9555 |
+
"learning_rate": 8.152329109915646e-07,
|
| 9556 |
+
"loss": 1.5232,
|
| 9557 |
+
"step": 1364
|
| 9558 |
+
},
|
| 9559 |
+
{
|
| 9560 |
+
"epoch": 0.06462762179821031,
|
| 9561 |
+
"grad_norm": 1.163534164428711,
|
| 9562 |
+
"learning_rate": 7.957590433101292e-07,
|
| 9563 |
+
"loss": 1.4556,
|
| 9564 |
+
"step": 1365
|
| 9565 |
+
},
|
| 9566 |
+
{
|
| 9567 |
+
"epoch": 0.06467496804128592,
|
| 9568 |
+
"grad_norm": 1.2503600120544434,
|
| 9569 |
+
"learning_rate": 7.765187211748348e-07,
|
| 9570 |
+
"loss": 1.4645,
|
| 9571 |
+
"step": 1366
|
| 9572 |
+
},
|
| 9573 |
+
{
|
| 9574 |
+
"epoch": 0.06472231428436154,
|
| 9575 |
+
"grad_norm": 1.2252036333084106,
|
| 9576 |
+
"learning_rate": 7.575120359089416e-07,
|
| 9577 |
+
"loss": 1.5589,
|
| 9578 |
+
"step": 1367
|
| 9579 |
+
},
|
| 9580 |
+
{
|
| 9581 |
+
"epoch": 0.06476966052743714,
|
| 9582 |
+
"grad_norm": 1.3027896881103516,
|
| 9583 |
+
"learning_rate": 7.387390777267689e-07,
|
| 9584 |
+
"loss": 1.4416,
|
| 9585 |
+
"step": 1368
|
| 9586 |
+
},
|
| 9587 |
+
{
|
| 9588 |
+
"epoch": 0.06481700677051276,
|
| 9589 |
+
"grad_norm": 1.274207592010498,
|
| 9590 |
+
"learning_rate": 7.201999357332345e-07,
|
| 9591 |
+
"loss": 1.4932,
|
| 9592 |
+
"step": 1369
|
| 9593 |
+
},
|
| 9594 |
+
{
|
| 9595 |
+
"epoch": 0.06486435301358837,
|
| 9596 |
+
"grad_norm": 1.2771484851837158,
|
| 9597 |
+
"learning_rate": 7.018946979234997e-07,
|
| 9598 |
+
"loss": 1.4079,
|
| 9599 |
+
"step": 1370
|
| 9600 |
+
},
|
| 9601 |
+
{
|
| 9602 |
+
"epoch": 0.06491169925666398,
|
| 9603 |
+
"grad_norm": 1.2282248735427856,
|
| 9604 |
+
"learning_rate": 6.838234511824748e-07,
|
| 9605 |
+
"loss": 1.4848,
|
| 9606 |
+
"step": 1371
|
| 9607 |
+
},
|
| 9608 |
+
{
|
| 9609 |
+
"epoch": 0.06495904549973959,
|
| 9610 |
+
"grad_norm": 1.2904937267303467,
|
| 9611 |
+
"learning_rate": 6.659862812844586e-07,
|
| 9612 |
+
"loss": 1.5547,
|
| 9613 |
+
"step": 1372
|
| 9614 |
+
},
|
| 9615 |
+
{
|
| 9616 |
+
"epoch": 0.06500639174281521,
|
| 9617 |
+
"grad_norm": 1.379770040512085,
|
| 9618 |
+
"learning_rate": 6.483832728927219e-07,
|
| 9619 |
+
"loss": 1.3129,
|
| 9620 |
+
"step": 1373
|
| 9621 |
+
},
|
| 9622 |
+
{
|
| 9623 |
+
"epoch": 0.06505373798589081,
|
| 9624 |
+
"grad_norm": 1.0465701818466187,
|
| 9625 |
+
"learning_rate": 6.310145095590803e-07,
|
| 9626 |
+
"loss": 1.0869,
|
| 9627 |
+
"step": 1374
|
| 9628 |
+
},
|
| 9629 |
+
{
|
| 9630 |
+
"epoch": 0.06510108422896643,
|
| 9631 |
+
"grad_norm": 1.2646894454956055,
|
| 9632 |
+
"learning_rate": 6.138800737235384e-07,
|
| 9633 |
+
"loss": 1.3675,
|
| 9634 |
+
"step": 1375
|
| 9635 |
+
},
|
| 9636 |
+
{
|
| 9637 |
+
"epoch": 0.06514843047204204,
|
| 9638 |
+
"grad_norm": 1.0842680931091309,
|
| 9639 |
+
"learning_rate": 5.969800467138519e-07,
|
| 9640 |
+
"loss": 1.2566,
|
| 9641 |
+
"step": 1376
|
| 9642 |
+
},
|
| 9643 |
+
{
|
| 9644 |
+
"epoch": 0.06519577671511766,
|
| 9645 |
+
"grad_norm": 1.2365455627441406,
|
| 9646 |
+
"learning_rate": 5.803145087451945e-07,
|
| 9647 |
+
"loss": 1.2261,
|
| 9648 |
+
"step": 1377
|
| 9649 |
+
},
|
| 9650 |
+
{
|
| 9651 |
+
"epoch": 0.06524312295819326,
|
| 9652 |
+
"grad_norm": 1.4027780294418335,
|
| 9653 |
+
"learning_rate": 5.638835389197295e-07,
|
| 9654 |
+
"loss": 1.5185,
|
| 9655 |
+
"step": 1378
|
| 9656 |
+
},
|
| 9657 |
+
{
|
| 9658 |
+
"epoch": 0.06529046920126888,
|
| 9659 |
+
"grad_norm": 1.227062702178955,
|
| 9660 |
+
"learning_rate": 5.476872152262558e-07,
|
| 9661 |
+
"loss": 1.2785,
|
| 9662 |
+
"step": 1379
|
| 9663 |
+
},
|
| 9664 |
+
{
|
| 9665 |
+
"epoch": 0.06533781544434449,
|
| 9666 |
+
"grad_norm": 1.2006760835647583,
|
| 9667 |
+
"learning_rate": 5.317256145398409e-07,
|
| 9668 |
+
"loss": 1.3152,
|
| 9669 |
+
"step": 1380
|
| 9670 |
+
},
|
| 9671 |
+
{
|
| 9672 |
+
"epoch": 0.0653851616874201,
|
| 9673 |
+
"grad_norm": 1.1898391246795654,
|
| 9674 |
+
"learning_rate": 5.159988126214544e-07,
|
| 9675 |
+
"loss": 1.3319,
|
| 9676 |
+
"step": 1381
|
| 9677 |
+
},
|
| 9678 |
+
{
|
| 9679 |
+
"epoch": 0.06543250793049571,
|
| 9680 |
+
"grad_norm": 1.377869725227356,
|
| 9681 |
+
"learning_rate": 5.005068841175963e-07,
|
| 9682 |
+
"loss": 1.5211,
|
| 9683 |
+
"step": 1382
|
| 9684 |
+
},
|
| 9685 |
+
{
|
| 9686 |
+
"epoch": 0.06547985417357133,
|
| 9687 |
+
"grad_norm": 1.4869921207427979,
|
| 9688 |
+
"learning_rate": 4.852499025599533e-07,
|
| 9689 |
+
"loss": 1.6305,
|
| 9690 |
+
"step": 1383
|
| 9691 |
+
},
|
| 9692 |
+
{
|
| 9693 |
+
"epoch": 0.06552720041664693,
|
| 9694 |
+
"grad_norm": 1.3564372062683105,
|
| 9695 |
+
"learning_rate": 4.7022794036505335e-07,
|
| 9696 |
+
"loss": 1.5702,
|
| 9697 |
+
"step": 1384
|
| 9698 |
+
},
|
| 9699 |
+
{
|
| 9700 |
+
"epoch": 0.06557454665972255,
|
| 9701 |
+
"grad_norm": 1.1574383974075317,
|
| 9702 |
+
"learning_rate": 4.554410688339061e-07,
|
| 9703 |
+
"loss": 1.2995,
|
| 9704 |
+
"step": 1385
|
| 9705 |
+
},
|
| 9706 |
+
{
|
| 9707 |
+
"epoch": 0.06562189290279816,
|
| 9708 |
+
"grad_norm": 1.3720345497131348,
|
| 9709 |
+
"learning_rate": 4.4088935815168044e-07,
|
| 9710 |
+
"loss": 1.4227,
|
| 9711 |
+
"step": 1386
|
| 9712 |
+
},
|
| 9713 |
+
{
|
| 9714 |
+
"epoch": 0.06566923914587378,
|
| 9715 |
+
"grad_norm": 1.3558074235916138,
|
| 9716 |
+
"learning_rate": 4.2657287738737675e-07,
|
| 9717 |
+
"loss": 1.4995,
|
| 9718 |
+
"step": 1387
|
| 9719 |
+
},
|
| 9720 |
+
{
|
| 9721 |
+
"epoch": 0.06571658538894938,
|
| 9722 |
+
"grad_norm": 1.412009835243225,
|
| 9723 |
+
"learning_rate": 4.124916944934665e-07,
|
| 9724 |
+
"loss": 1.3792,
|
| 9725 |
+
"step": 1388
|
| 9726 |
+
},
|
| 9727 |
+
{
|
| 9728 |
+
"epoch": 0.065763931632025,
|
| 9729 |
+
"grad_norm": 1.3199092149734497,
|
| 9730 |
+
"learning_rate": 3.986458763056089e-07,
|
| 9731 |
+
"loss": 1.3669,
|
| 9732 |
+
"step": 1389
|
| 9733 |
+
},
|
| 9734 |
+
{
|
| 9735 |
+
"epoch": 0.0658112778751006,
|
| 9736 |
+
"grad_norm": 1.1794309616088867,
|
| 9737 |
+
"learning_rate": 3.8503548854230687e-07,
|
| 9738 |
+
"loss": 1.0283,
|
| 9739 |
+
"step": 1390
|
| 9740 |
+
},
|
| 9741 |
+
{
|
| 9742 |
+
"epoch": 0.06585862411817622,
|
| 9743 |
+
"grad_norm": 1.5401195287704468,
|
| 9744 |
+
"learning_rate": 3.716605958046071e-07,
|
| 9745 |
+
"loss": 1.474,
|
| 9746 |
+
"step": 1391
|
| 9747 |
+
},
|
| 9748 |
+
{
|
| 9749 |
+
"epoch": 0.06590597036125183,
|
| 9750 |
+
"grad_norm": 1.258670687675476,
|
| 9751 |
+
"learning_rate": 3.585212615757838e-07,
|
| 9752 |
+
"loss": 1.4567,
|
| 9753 |
+
"step": 1392
|
| 9754 |
+
},
|
| 9755 |
+
{
|
| 9756 |
+
"epoch": 0.06595331660432745,
|
| 9757 |
+
"grad_norm": 1.4355465173721313,
|
| 9758 |
+
"learning_rate": 3.456175482210611e-07,
|
| 9759 |
+
"loss": 1.4306,
|
| 9760 |
+
"step": 1393
|
| 9761 |
+
},
|
| 9762 |
+
{
|
| 9763 |
+
"epoch": 0.06600066284740305,
|
| 9764 |
+
"grad_norm": 1.3334470987319946,
|
| 9765 |
+
"learning_rate": 3.3294951698728537e-07,
|
| 9766 |
+
"loss": 1.2449,
|
| 9767 |
+
"step": 1394
|
| 9768 |
+
},
|
| 9769 |
+
{
|
| 9770 |
+
"epoch": 0.06604800909047867,
|
| 9771 |
+
"grad_norm": 1.1793415546417236,
|
| 9772 |
+
"learning_rate": 3.205172280026536e-07,
|
| 9773 |
+
"loss": 0.9879,
|
| 9774 |
+
"step": 1395
|
| 9775 |
+
},
|
| 9776 |
+
{
|
| 9777 |
+
"epoch": 0.06609535533355428,
|
| 9778 |
+
"grad_norm": 1.465490460395813,
|
| 9779 |
+
"learning_rate": 3.083207402764132e-07,
|
| 9780 |
+
"loss": 1.4966,
|
| 9781 |
+
"step": 1396
|
| 9782 |
+
},
|
| 9783 |
+
{
|
| 9784 |
+
"epoch": 0.0661427015766299,
|
| 9785 |
+
"grad_norm": 1.748968243598938,
|
| 9786 |
+
"learning_rate": 2.963601116986181e-07,
|
| 9787 |
+
"loss": 1.4836,
|
| 9788 |
+
"step": 1397
|
| 9789 |
+
},
|
| 9790 |
+
{
|
| 9791 |
+
"epoch": 0.0661900478197055,
|
| 9792 |
+
"grad_norm": 1.5985764265060425,
|
| 9793 |
+
"learning_rate": 2.846353990398065e-07,
|
| 9794 |
+
"loss": 1.5044,
|
| 9795 |
+
"step": 1398
|
| 9796 |
+
},
|
| 9797 |
+
{
|
| 9798 |
+
"epoch": 0.06623739406278112,
|
| 9799 |
+
"grad_norm": 1.5981168746948242,
|
| 9800 |
+
"learning_rate": 2.731466579507513e-07,
|
| 9801 |
+
"loss": 1.4942,
|
| 9802 |
+
"step": 1399
|
| 9803 |
+
},
|
| 9804 |
+
{
|
| 9805 |
+
"epoch": 0.06628474030585672,
|
| 9806 |
+
"grad_norm": 1.3606255054473877,
|
| 9807 |
+
"learning_rate": 2.618939429622158e-07,
|
| 9808 |
+
"loss": 1.2628,
|
| 9809 |
+
"step": 1400
|
| 9810 |
+
},
|
| 9811 |
+
{
|
| 9812 |
+
"epoch": 0.06633208654893234,
|
| 9813 |
+
"grad_norm": 1.2152913808822632,
|
| 9814 |
+
"learning_rate": 2.5087730748466485e-07,
|
| 9815 |
+
"loss": 1.67,
|
| 9816 |
+
"step": 1401
|
| 9817 |
+
},
|
| 9818 |
+
{
|
| 9819 |
+
"epoch": 0.06637943279200795,
|
| 9820 |
+
"grad_norm": 1.1747958660125732,
|
| 9821 |
+
"learning_rate": 2.400968038080209e-07,
|
| 9822 |
+
"loss": 1.6487,
|
| 9823 |
+
"step": 1402
|
| 9824 |
+
},
|
| 9825 |
+
{
|
| 9826 |
+
"epoch": 0.06642677903508357,
|
| 9827 |
+
"grad_norm": 1.3034745454788208,
|
| 9828 |
+
"learning_rate": 2.2955248310142507e-07,
|
| 9829 |
+
"loss": 1.7776,
|
| 9830 |
+
"step": 1403
|
| 9831 |
+
},
|
| 9832 |
+
{
|
| 9833 |
+
"epoch": 0.06647412527815917,
|
| 9834 |
+
"grad_norm": 1.1659724712371826,
|
| 9835 |
+
"learning_rate": 2.1924439541298193e-07,
|
| 9836 |
+
"loss": 1.6656,
|
| 9837 |
+
"step": 1404
|
| 9838 |
+
},
|
| 9839 |
+
{
|
| 9840 |
+
"epoch": 0.06652147152123479,
|
| 9841 |
+
"grad_norm": 1.4440197944641113,
|
| 9842 |
+
"learning_rate": 2.0917258966953733e-07,
|
| 9843 |
+
"loss": 1.6118,
|
| 9844 |
+
"step": 1405
|
| 9845 |
+
},
|
| 9846 |
+
{
|
| 9847 |
+
"epoch": 0.0665688177643104,
|
| 9848 |
+
"grad_norm": 1.1191264390945435,
|
| 9849 |
+
"learning_rate": 1.9933711367641771e-07,
|
| 9850 |
+
"loss": 1.391,
|
| 9851 |
+
"step": 1406
|
| 9852 |
+
},
|
| 9853 |
+
{
|
| 9854 |
+
"epoch": 0.06661616400738601,
|
| 9855 |
+
"grad_norm": 1.176639437675476,
|
| 9856 |
+
"learning_rate": 1.8973801411724667e-07,
|
| 9857 |
+
"loss": 1.4539,
|
| 9858 |
+
"step": 1407
|
| 9859 |
+
},
|
| 9860 |
+
{
|
| 9861 |
+
"epoch": 0.06666351025046162,
|
| 9862 |
+
"grad_norm": 1.211703896522522,
|
| 9863 |
+
"learning_rate": 1.8037533655367868e-07,
|
| 9864 |
+
"loss": 1.3097,
|
| 9865 |
+
"step": 1408
|
| 9866 |
+
},
|
| 9867 |
+
{
|
| 9868 |
+
"epoch": 0.06671085649353724,
|
| 9869 |
+
"grad_norm": 1.2370667457580566,
|
| 9870 |
+
"learning_rate": 1.7124912542520467e-07,
|
| 9871 |
+
"loss": 1.6055,
|
| 9872 |
+
"step": 1409
|
| 9873 |
+
},
|
| 9874 |
+
{
|
| 9875 |
+
"epoch": 0.06675820273661286,
|
| 9876 |
+
"grad_norm": 1.227429747581482,
|
| 9877 |
+
"learning_rate": 1.6235942404895788e-07,
|
| 9878 |
+
"loss": 1.6139,
|
| 9879 |
+
"step": 1410
|
| 9880 |
+
},
|
| 9881 |
+
{
|
| 9882 |
+
"epoch": 0.06680554897968846,
|
| 9883 |
+
"grad_norm": 1.2459824085235596,
|
| 9884 |
+
"learning_rate": 1.537062746194584e-07,
|
| 9885 |
+
"loss": 1.5071,
|
| 9886 |
+
"step": 1411
|
| 9887 |
+
},
|
| 9888 |
+
{
|
| 9889 |
+
"epoch": 0.06685289522276408,
|
| 9890 |
+
"grad_norm": 1.1891640424728394,
|
| 9891 |
+
"learning_rate": 1.4528971820846893e-07,
|
| 9892 |
+
"loss": 1.4231,
|
| 9893 |
+
"step": 1412
|
| 9894 |
+
},
|
| 9895 |
+
{
|
| 9896 |
+
"epoch": 0.06690024146583969,
|
| 9897 |
+
"grad_norm": 1.2420520782470703,
|
| 9898 |
+
"learning_rate": 1.3710979476476705e-07,
|
| 9899 |
+
"loss": 1.5916,
|
| 9900 |
+
"step": 1413
|
| 9901 |
+
},
|
| 9902 |
+
{
|
| 9903 |
+
"epoch": 0.0669475877089153,
|
| 9904 |
+
"grad_norm": 1.4088830947875977,
|
| 9905 |
+
"learning_rate": 1.2916654311396213e-07,
|
| 9906 |
+
"loss": 1.4991,
|
| 9907 |
+
"step": 1414
|
| 9908 |
+
},
|
| 9909 |
+
{
|
| 9910 |
+
"epoch": 0.06699493395199091,
|
| 9911 |
+
"grad_norm": 1.3840956687927246,
|
| 9912 |
+
"learning_rate": 1.2146000095831778e-07,
|
| 9913 |
+
"loss": 1.7127,
|
| 9914 |
+
"step": 1415
|
| 9915 |
+
},
|
| 9916 |
+
{
|
| 9917 |
+
"epoch": 0.06704228019506653,
|
| 9918 |
+
"grad_norm": 1.2678717374801636,
|
| 9919 |
+
"learning_rate": 1.1399020487655177e-07,
|
| 9920 |
+
"loss": 1.4063,
|
| 9921 |
+
"step": 1416
|
| 9922 |
+
},
|
| 9923 |
+
{
|
| 9924 |
+
"epoch": 0.06708962643814213,
|
| 9925 |
+
"grad_norm": 1.0719335079193115,
|
| 9926 |
+
"learning_rate": 1.0675719032370302e-07,
|
| 9927 |
+
"loss": 1.3102,
|
| 9928 |
+
"step": 1417
|
| 9929 |
+
},
|
| 9930 |
+
{
|
| 9931 |
+
"epoch": 0.06713697268121775,
|
| 9932 |
+
"grad_norm": 1.2429107427597046,
|
| 9933 |
+
"learning_rate": 9.976099163091501e-08,
|
| 9934 |
+
"loss": 1.3659,
|
| 9935 |
+
"step": 1418
|
| 9936 |
+
},
|
| 9937 |
+
{
|
| 9938 |
+
"epoch": 0.06718431892429336,
|
| 9939 |
+
"grad_norm": 1.2890995740890503,
|
| 9940 |
+
"learning_rate": 9.300164200530814e-08,
|
| 9941 |
+
"loss": 1.4734,
|
| 9942 |
+
"step": 1419
|
| 9943 |
+
},
|
| 9944 |
+
{
|
| 9945 |
+
"epoch": 0.06723166516736898,
|
| 9946 |
+
"grad_norm": 1.2928924560546875,
|
| 9947 |
+
"learning_rate": 8.647917352980762e-08,
|
| 9948 |
+
"loss": 1.4282,
|
| 9949 |
+
"step": 1420
|
| 9950 |
+
},
|
| 9951 |
+
{
|
| 9952 |
+
"epoch": 0.06727901141044458,
|
| 9953 |
+
"grad_norm": 1.3546903133392334,
|
| 9954 |
+
"learning_rate": 8.019361716299911e-08,
|
| 9955 |
+
"loss": 1.5191,
|
| 9956 |
+
"step": 1421
|
| 9957 |
+
},
|
| 9958 |
+
{
|
| 9959 |
+
"epoch": 0.0673263576535202,
|
| 9960 |
+
"grad_norm": 1.1908518075942993,
|
| 9961 |
+
"learning_rate": 7.414500273896785e-08,
|
| 9962 |
+
"loss": 1.3663,
|
| 9963 |
+
"step": 1422
|
| 9964 |
+
},
|
| 9965 |
+
{
|
| 9966 |
+
"epoch": 0.0673737038965958,
|
| 9967 |
+
"grad_norm": 1.2994002103805542,
|
| 9968 |
+
"learning_rate": 6.833335896716531e-08,
|
| 9969 |
+
"loss": 1.4845,
|
| 9970 |
+
"step": 1423
|
| 9971 |
+
},
|
| 9972 |
+
{
|
| 9973 |
+
"epoch": 0.06742105013967142,
|
| 9974 |
+
"grad_norm": 1.130214810371399,
|
| 9975 |
+
"learning_rate": 6.275871343228157e-08,
|
| 9976 |
+
"loss": 1.3068,
|
| 9977 |
+
"step": 1424
|
| 9978 |
+
},
|
| 9979 |
+
{
|
| 9980 |
+
"epoch": 0.06746839638274703,
|
| 9981 |
+
"grad_norm": 1.4080276489257812,
|
| 9982 |
+
"learning_rate": 5.7421092594101e-08,
|
| 9983 |
+
"loss": 1.3539,
|
| 9984 |
+
"step": 1425
|
| 9985 |
+
},
|
| 9986 |
+
{
|
| 9987 |
+
"epoch": 0.06751574262582265,
|
| 9988 |
+
"grad_norm": 1.3741194009780884,
|
| 9989 |
+
"learning_rate": 5.2320521787385667e-08,
|
| 9990 |
+
"loss": 1.6177,
|
| 9991 |
+
"step": 1426
|
| 9992 |
+
},
|
| 9993 |
+
{
|
| 9994 |
+
"epoch": 0.06756308886889825,
|
| 9995 |
+
"grad_norm": 1.3106088638305664,
|
| 9996 |
+
"learning_rate": 4.745702522174211e-08,
|
| 9997 |
+
"loss": 1.4852,
|
| 9998 |
+
"step": 1427
|
| 9999 |
+
},
|
| 10000 |
+
{
|
| 10001 |
+
"epoch": 0.06761043511197387,
|
| 10002 |
+
"grad_norm": 1.279340147972107,
|
| 10003 |
+
"learning_rate": 4.2830625981521435e-08,
|
| 10004 |
+
"loss": 1.4519,
|
| 10005 |
+
"step": 1428
|
| 10006 |
+
},
|
| 10007 |
+
{
|
| 10008 |
+
"epoch": 0.06765778135504948,
|
| 10009 |
+
"grad_norm": 1.298641562461853,
|
| 10010 |
+
"learning_rate": 3.8441346025708255e-08,
|
| 10011 |
+
"loss": 1.3406,
|
| 10012 |
+
"step": 1429
|
| 10013 |
+
},
|
| 10014 |
+
{
|
| 10015 |
+
"epoch": 0.0677051275981251,
|
| 10016 |
+
"grad_norm": 1.3770064115524292,
|
| 10017 |
+
"learning_rate": 3.428920618780418e-08,
|
| 10018 |
+
"loss": 1.6185,
|
| 10019 |
+
"step": 1430
|
| 10020 |
+
},
|
| 10021 |
+
{
|
| 10022 |
+
"epoch": 0.0677524738412007,
|
| 10023 |
+
"grad_norm": 1.2126370668411255,
|
| 10024 |
+
"learning_rate": 3.037422617573893e-08,
|
| 10025 |
+
"loss": 1.2262,
|
| 10026 |
+
"step": 1431
|
| 10027 |
+
},
|
| 10028 |
+
{
|
| 10029 |
+
"epoch": 0.06779982008427632,
|
| 10030 |
+
"grad_norm": 1.0952389240264893,
|
| 10031 |
+
"learning_rate": 2.6696424571776014e-08,
|
| 10032 |
+
"loss": 1.3015,
|
| 10033 |
+
"step": 1432
|
| 10034 |
+
},
|
| 10035 |
+
{
|
| 10036 |
+
"epoch": 0.06784716632735192,
|
| 10037 |
+
"grad_norm": 1.183917760848999,
|
| 10038 |
+
"learning_rate": 2.3255818832423894e-08,
|
| 10039 |
+
"loss": 1.265,
|
| 10040 |
+
"step": 1433
|
| 10041 |
+
},
|
| 10042 |
+
{
|
| 10043 |
+
"epoch": 0.06789451257042754,
|
| 10044 |
+
"grad_norm": 1.2368615865707397,
|
| 10045 |
+
"learning_rate": 2.0052425288347164e-08,
|
| 10046 |
+
"loss": 1.2985,
|
| 10047 |
+
"step": 1434
|
| 10048 |
+
},
|
| 10049 |
+
{
|
| 10050 |
+
"epoch": 0.06794185881350315,
|
| 10051 |
+
"grad_norm": 1.2383983135223389,
|
| 10052 |
+
"learning_rate": 1.7086259144305504e-08,
|
| 10053 |
+
"loss": 1.3451,
|
| 10054 |
+
"step": 1435
|
| 10055 |
+
},
|
| 10056 |
+
{
|
| 10057 |
+
"epoch": 0.06798920505657877,
|
| 10058 |
+
"grad_norm": 1.2299693822860718,
|
| 10059 |
+
"learning_rate": 1.4357334479059292e-08,
|
| 10060 |
+
"loss": 1.1355,
|
| 10061 |
+
"step": 1436
|
| 10062 |
+
},
|
| 10063 |
+
{
|
| 10064 |
+
"epoch": 0.06803655129965437,
|
| 10065 |
+
"grad_norm": 1.2785406112670898,
|
| 10066 |
+
"learning_rate": 1.1865664245314101e-08,
|
| 10067 |
+
"loss": 1.2929,
|
| 10068 |
+
"step": 1437
|
| 10069 |
+
},
|
| 10070 |
+
{
|
| 10071 |
+
"epoch": 0.06808389754272999,
|
| 10072 |
+
"grad_norm": 1.1830840110778809,
|
| 10073 |
+
"learning_rate": 9.61126026966519e-09,
|
| 10074 |
+
"loss": 1.2871,
|
| 10075 |
+
"step": 1438
|
| 10076 |
+
},
|
| 10077 |
+
{
|
| 10078 |
+
"epoch": 0.0681312437858056,
|
| 10079 |
+
"grad_norm": 1.2833662033081055,
|
| 10080 |
+
"learning_rate": 7.594133252530888e-09,
|
| 10081 |
+
"loss": 1.3836,
|
| 10082 |
+
"step": 1439
|
| 10083 |
+
},
|
| 10084 |
+
{
|
| 10085 |
+
"epoch": 0.06817859002888121,
|
| 10086 |
+
"grad_norm": 1.6154637336730957,
|
| 10087 |
+
"learning_rate": 5.814292768108187e-09,
|
| 10088 |
+
"loss": 1.6961,
|
| 10089 |
+
"step": 1440
|
| 10090 |
+
},
|
| 10091 |
+
{
|
| 10092 |
+
"epoch": 0.06822593627195682,
|
| 10093 |
+
"grad_norm": 1.3915746212005615,
|
| 10094 |
+
"learning_rate": 4.2717472643227785e-09,
|
| 10095 |
+
"loss": 1.2641,
|
| 10096 |
+
"step": 1441
|
| 10097 |
+
},
|
| 10098 |
+
{
|
| 10099 |
+
"epoch": 0.06827328251503244,
|
| 10100 |
+
"grad_norm": 1.2784771919250488,
|
| 10101 |
+
"learning_rate": 2.9665040627901984e-09,
|
| 10102 |
+
"loss": 1.3296,
|
| 10103 |
+
"step": 1442
|
| 10104 |
+
},
|
| 10105 |
+
{
|
| 10106 |
+
"epoch": 0.06832062875810804,
|
| 10107 |
+
"grad_norm": 1.3521119356155396,
|
| 10108 |
+
"learning_rate": 1.8985693587880717e-09,
|
| 10109 |
+
"loss": 1.2266,
|
| 10110 |
+
"step": 1443
|
| 10111 |
+
},
|
| 10112 |
+
{
|
| 10113 |
+
"epoch": 0.06836797500118366,
|
| 10114 |
+
"grad_norm": 1.245299220085144,
|
| 10115 |
+
"learning_rate": 1.067948221217252e-09,
|
| 10116 |
+
"loss": 1.201,
|
| 10117 |
+
"step": 1444
|
| 10118 |
+
},
|
| 10119 |
+
{
|
| 10120 |
+
"epoch": 0.06841532124425927,
|
| 10121 |
+
"grad_norm": 1.4662199020385742,
|
| 10122 |
+
"learning_rate": 4.746445925740694e-10,
|
| 10123 |
+
"loss": 1.371,
|
| 10124 |
+
"step": 1445
|
| 10125 |
+
},
|
| 10126 |
+
{
|
| 10127 |
+
"epoch": 0.06846266748733489,
|
| 10128 |
+
"grad_norm": 1.248600721359253,
|
| 10129 |
+
"learning_rate": 1.1866128895032803e-10,
|
| 10130 |
+
"loss": 1.0466,
|
| 10131 |
+
"step": 1446
|
| 10132 |
+
},
|
| 10133 |
+
{
|
| 10134 |
+
"epoch": 0.06851001373041049,
|
| 10135 |
+
"grad_norm": 1.496204137802124,
|
| 10136 |
+
"learning_rate": 0.0,
|
| 10137 |
+
"loss": 1.392,
|
| 10138 |
+
"step": 1447
|
| 10139 |
}
|
| 10140 |
],
|
| 10141 |
"logging_steps": 1,
|
|
|
|
| 10150 |
"should_evaluate": false,
|
| 10151 |
"should_log": false,
|
| 10152 |
"should_save": true,
|
| 10153 |
+
"should_training_stop": true
|
| 10154 |
},
|
| 10155 |
"attributes": {}
|
| 10156 |
}
|
| 10157 |
},
|
| 10158 |
+
"total_flos": 1.0086377434770309e+18,
|
| 10159 |
"train_batch_size": 4,
|
| 10160 |
"trial_name": null,
|
| 10161 |
"trial_params": null
|