Training in progress, step 600
Browse files- last-checkpoint/optimizer.pt +1 -1
- last-checkpoint/pytorch_model.bin +1 -1
- last-checkpoint/rng_state.pth +1 -1
- last-checkpoint/scheduler.pt +1 -1
- last-checkpoint/trainer_state.json +1213 -5
- pytorch_model.bin +1 -1
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4736616809
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:edc9882338455f81dd1e85ace29ccf7af7da229453c5afb0d6c4bc90350a8b18
|
| 3 |
size 4736616809
|
last-checkpoint/pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 2368281769
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b1ac718a5450b8f5bc3e788c6d41748c97c82f6bf2933bbce997b0073b1c8202
|
| 3 |
size 2368281769
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14575
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:425eded187a10b59dd6706ea8bd8dddc527ddb901485b28745924b7945f3098d
|
| 3 |
size 14575
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 627
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:eb5f790e340963b8140823821a1411a83e471c6359b6fd89f0bb6a4aa0276e15
|
| 3 |
size 627
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 2.
|
| 3 |
-
"best_model_checkpoint": "output/checkpoint-
|
| 4 |
-
"epoch": 0.
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
@@ -2422,11 +2422,1219 @@
|
|
| 2422 |
"eval_samples_per_second": 5.75,
|
| 2423 |
"eval_steps_per_second": 5.75,
|
| 2424 |
"step": 400
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2425 |
}
|
| 2426 |
],
|
| 2427 |
"max_steps": 4386,
|
| 2428 |
"num_train_epochs": 3,
|
| 2429 |
-
"total_flos":
|
| 2430 |
"trial_name": null,
|
| 2431 |
"trial_params": null
|
| 2432 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 2.047600030899048,
|
| 3 |
+
"best_model_checkpoint": "output/checkpoint-600",
|
| 4 |
+
"epoch": 0.41021039306488055,
|
| 5 |
+
"global_step": 600,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 2422 |
"eval_samples_per_second": 5.75,
|
| 2423 |
"eval_steps_per_second": 5.75,
|
| 2424 |
"step": 400
|
| 2425 |
+
},
|
| 2426 |
+
{
|
| 2427 |
+
"epoch": 0.27,
|
| 2428 |
+
"learning_rate": 0.000571189679885332,
|
| 2429 |
+
"loss": 2.1854,
|
| 2430 |
+
"step": 401
|
| 2431 |
+
},
|
| 2432 |
+
{
|
| 2433 |
+
"epoch": 0.27,
|
| 2434 |
+
"learning_rate": 0.0005710463449593884,
|
| 2435 |
+
"loss": 2.1751,
|
| 2436 |
+
"step": 402
|
| 2437 |
+
},
|
| 2438 |
+
{
|
| 2439 |
+
"epoch": 0.28,
|
| 2440 |
+
"learning_rate": 0.0005709030100334448,
|
| 2441 |
+
"loss": 2.2177,
|
| 2442 |
+
"step": 403
|
| 2443 |
+
},
|
| 2444 |
+
{
|
| 2445 |
+
"epoch": 0.28,
|
| 2446 |
+
"learning_rate": 0.0005707596751075012,
|
| 2447 |
+
"loss": 2.1151,
|
| 2448 |
+
"step": 404
|
| 2449 |
+
},
|
| 2450 |
+
{
|
| 2451 |
+
"epoch": 0.28,
|
| 2452 |
+
"learning_rate": 0.0005706163401815575,
|
| 2453 |
+
"loss": 2.1397,
|
| 2454 |
+
"step": 405
|
| 2455 |
+
},
|
| 2456 |
+
{
|
| 2457 |
+
"epoch": 0.28,
|
| 2458 |
+
"learning_rate": 0.0005704730052556139,
|
| 2459 |
+
"loss": 2.1307,
|
| 2460 |
+
"step": 406
|
| 2461 |
+
},
|
| 2462 |
+
{
|
| 2463 |
+
"epoch": 0.28,
|
| 2464 |
+
"learning_rate": 0.0005703296703296703,
|
| 2465 |
+
"loss": 2.1163,
|
| 2466 |
+
"step": 407
|
| 2467 |
+
},
|
| 2468 |
+
{
|
| 2469 |
+
"epoch": 0.28,
|
| 2470 |
+
"learning_rate": 0.0005701863354037267,
|
| 2471 |
+
"loss": 2.2228,
|
| 2472 |
+
"step": 408
|
| 2473 |
+
},
|
| 2474 |
+
{
|
| 2475 |
+
"epoch": 0.28,
|
| 2476 |
+
"learning_rate": 0.000570043000477783,
|
| 2477 |
+
"loss": 2.0841,
|
| 2478 |
+
"step": 409
|
| 2479 |
+
},
|
| 2480 |
+
{
|
| 2481 |
+
"epoch": 0.28,
|
| 2482 |
+
"learning_rate": 0.0005698996655518395,
|
| 2483 |
+
"loss": 2.2314,
|
| 2484 |
+
"step": 410
|
| 2485 |
+
},
|
| 2486 |
+
{
|
| 2487 |
+
"epoch": 0.28,
|
| 2488 |
+
"learning_rate": 0.0005697563306258957,
|
| 2489 |
+
"loss": 2.2701,
|
| 2490 |
+
"step": 411
|
| 2491 |
+
},
|
| 2492 |
+
{
|
| 2493 |
+
"epoch": 0.28,
|
| 2494 |
+
"learning_rate": 0.0005696129956999522,
|
| 2495 |
+
"loss": 2.1293,
|
| 2496 |
+
"step": 412
|
| 2497 |
+
},
|
| 2498 |
+
{
|
| 2499 |
+
"epoch": 0.28,
|
| 2500 |
+
"learning_rate": 0.0005694696607740085,
|
| 2501 |
+
"loss": 2.1884,
|
| 2502 |
+
"step": 413
|
| 2503 |
+
},
|
| 2504 |
+
{
|
| 2505 |
+
"epoch": 0.28,
|
| 2506 |
+
"learning_rate": 0.0005693263258480649,
|
| 2507 |
+
"loss": 2.1422,
|
| 2508 |
+
"step": 414
|
| 2509 |
+
},
|
| 2510 |
+
{
|
| 2511 |
+
"epoch": 0.28,
|
| 2512 |
+
"learning_rate": 0.0005691829909221213,
|
| 2513 |
+
"loss": 2.1939,
|
| 2514 |
+
"step": 415
|
| 2515 |
+
},
|
| 2516 |
+
{
|
| 2517 |
+
"epoch": 0.28,
|
| 2518 |
+
"learning_rate": 0.0005690396559961777,
|
| 2519 |
+
"loss": 2.1862,
|
| 2520 |
+
"step": 416
|
| 2521 |
+
},
|
| 2522 |
+
{
|
| 2523 |
+
"epoch": 0.29,
|
| 2524 |
+
"learning_rate": 0.000568896321070234,
|
| 2525 |
+
"loss": 2.2965,
|
| 2526 |
+
"step": 417
|
| 2527 |
+
},
|
| 2528 |
+
{
|
| 2529 |
+
"epoch": 0.29,
|
| 2530 |
+
"learning_rate": 0.0005687529861442904,
|
| 2531 |
+
"loss": 2.2042,
|
| 2532 |
+
"step": 418
|
| 2533 |
+
},
|
| 2534 |
+
{
|
| 2535 |
+
"epoch": 0.29,
|
| 2536 |
+
"learning_rate": 0.0005686096512183468,
|
| 2537 |
+
"loss": 2.1242,
|
| 2538 |
+
"step": 419
|
| 2539 |
+
},
|
| 2540 |
+
{
|
| 2541 |
+
"epoch": 0.29,
|
| 2542 |
+
"learning_rate": 0.0005684663162924032,
|
| 2543 |
+
"loss": 2.1837,
|
| 2544 |
+
"step": 420
|
| 2545 |
+
},
|
| 2546 |
+
{
|
| 2547 |
+
"epoch": 0.29,
|
| 2548 |
+
"learning_rate": 0.0005683229813664596,
|
| 2549 |
+
"loss": 2.1068,
|
| 2550 |
+
"step": 421
|
| 2551 |
+
},
|
| 2552 |
+
{
|
| 2553 |
+
"epoch": 0.29,
|
| 2554 |
+
"learning_rate": 0.000568179646440516,
|
| 2555 |
+
"loss": 2.2236,
|
| 2556 |
+
"step": 422
|
| 2557 |
+
},
|
| 2558 |
+
{
|
| 2559 |
+
"epoch": 0.29,
|
| 2560 |
+
"learning_rate": 0.0005680363115145723,
|
| 2561 |
+
"loss": 2.1684,
|
| 2562 |
+
"step": 423
|
| 2563 |
+
},
|
| 2564 |
+
{
|
| 2565 |
+
"epoch": 0.29,
|
| 2566 |
+
"learning_rate": 0.0005678929765886287,
|
| 2567 |
+
"loss": 2.1441,
|
| 2568 |
+
"step": 424
|
| 2569 |
+
},
|
| 2570 |
+
{
|
| 2571 |
+
"epoch": 0.29,
|
| 2572 |
+
"learning_rate": 0.0005677496416626851,
|
| 2573 |
+
"loss": 2.1091,
|
| 2574 |
+
"step": 425
|
| 2575 |
+
},
|
| 2576 |
+
{
|
| 2577 |
+
"epoch": 0.29,
|
| 2578 |
+
"learning_rate": 0.0005676063067367415,
|
| 2579 |
+
"loss": 2.2047,
|
| 2580 |
+
"step": 426
|
| 2581 |
+
},
|
| 2582 |
+
{
|
| 2583 |
+
"epoch": 0.29,
|
| 2584 |
+
"learning_rate": 0.0005674629718107979,
|
| 2585 |
+
"loss": 2.1858,
|
| 2586 |
+
"step": 427
|
| 2587 |
+
},
|
| 2588 |
+
{
|
| 2589 |
+
"epoch": 0.29,
|
| 2590 |
+
"learning_rate": 0.0005673196368848541,
|
| 2591 |
+
"loss": 2.141,
|
| 2592 |
+
"step": 428
|
| 2593 |
+
},
|
| 2594 |
+
{
|
| 2595 |
+
"epoch": 0.29,
|
| 2596 |
+
"learning_rate": 0.0005671763019589106,
|
| 2597 |
+
"loss": 2.1447,
|
| 2598 |
+
"step": 429
|
| 2599 |
+
},
|
| 2600 |
+
{
|
| 2601 |
+
"epoch": 0.29,
|
| 2602 |
+
"learning_rate": 0.0005670329670329669,
|
| 2603 |
+
"loss": 2.1216,
|
| 2604 |
+
"step": 430
|
| 2605 |
+
},
|
| 2606 |
+
{
|
| 2607 |
+
"epoch": 0.29,
|
| 2608 |
+
"learning_rate": 0.0005668896321070234,
|
| 2609 |
+
"loss": 2.1843,
|
| 2610 |
+
"step": 431
|
| 2611 |
+
},
|
| 2612 |
+
{
|
| 2613 |
+
"epoch": 0.3,
|
| 2614 |
+
"learning_rate": 0.0005667462971810797,
|
| 2615 |
+
"loss": 2.3159,
|
| 2616 |
+
"step": 432
|
| 2617 |
+
},
|
| 2618 |
+
{
|
| 2619 |
+
"epoch": 0.3,
|
| 2620 |
+
"learning_rate": 0.0005666029622551362,
|
| 2621 |
+
"loss": 2.2143,
|
| 2622 |
+
"step": 433
|
| 2623 |
+
},
|
| 2624 |
+
{
|
| 2625 |
+
"epoch": 0.3,
|
| 2626 |
+
"learning_rate": 0.0005664596273291925,
|
| 2627 |
+
"loss": 2.1456,
|
| 2628 |
+
"step": 434
|
| 2629 |
+
},
|
| 2630 |
+
{
|
| 2631 |
+
"epoch": 0.3,
|
| 2632 |
+
"learning_rate": 0.0005663162924032488,
|
| 2633 |
+
"loss": 2.1061,
|
| 2634 |
+
"step": 435
|
| 2635 |
+
},
|
| 2636 |
+
{
|
| 2637 |
+
"epoch": 0.3,
|
| 2638 |
+
"learning_rate": 0.0005661729574773052,
|
| 2639 |
+
"loss": 2.2449,
|
| 2640 |
+
"step": 436
|
| 2641 |
+
},
|
| 2642 |
+
{
|
| 2643 |
+
"epoch": 0.3,
|
| 2644 |
+
"learning_rate": 0.0005660296225513616,
|
| 2645 |
+
"loss": 2.1625,
|
| 2646 |
+
"step": 437
|
| 2647 |
+
},
|
| 2648 |
+
{
|
| 2649 |
+
"epoch": 0.3,
|
| 2650 |
+
"learning_rate": 0.000565886287625418,
|
| 2651 |
+
"loss": 2.1141,
|
| 2652 |
+
"step": 438
|
| 2653 |
+
},
|
| 2654 |
+
{
|
| 2655 |
+
"epoch": 0.3,
|
| 2656 |
+
"learning_rate": 0.0005657429526994744,
|
| 2657 |
+
"loss": 2.1314,
|
| 2658 |
+
"step": 439
|
| 2659 |
+
},
|
| 2660 |
+
{
|
| 2661 |
+
"epoch": 0.3,
|
| 2662 |
+
"learning_rate": 0.0005655996177735308,
|
| 2663 |
+
"loss": 2.1726,
|
| 2664 |
+
"step": 440
|
| 2665 |
+
},
|
| 2666 |
+
{
|
| 2667 |
+
"epoch": 0.3,
|
| 2668 |
+
"learning_rate": 0.0005654562828475871,
|
| 2669 |
+
"loss": 2.1889,
|
| 2670 |
+
"step": 441
|
| 2671 |
+
},
|
| 2672 |
+
{
|
| 2673 |
+
"epoch": 0.3,
|
| 2674 |
+
"learning_rate": 0.0005653129479216435,
|
| 2675 |
+
"loss": 2.2172,
|
| 2676 |
+
"step": 442
|
| 2677 |
+
},
|
| 2678 |
+
{
|
| 2679 |
+
"epoch": 0.3,
|
| 2680 |
+
"learning_rate": 0.0005651696129956999,
|
| 2681 |
+
"loss": 2.1766,
|
| 2682 |
+
"step": 443
|
| 2683 |
+
},
|
| 2684 |
+
{
|
| 2685 |
+
"epoch": 0.3,
|
| 2686 |
+
"learning_rate": 0.0005650262780697563,
|
| 2687 |
+
"loss": 2.0534,
|
| 2688 |
+
"step": 444
|
| 2689 |
+
},
|
| 2690 |
+
{
|
| 2691 |
+
"epoch": 0.3,
|
| 2692 |
+
"learning_rate": 0.0005648829431438127,
|
| 2693 |
+
"loss": 2.1665,
|
| 2694 |
+
"step": 445
|
| 2695 |
+
},
|
| 2696 |
+
{
|
| 2697 |
+
"epoch": 0.3,
|
| 2698 |
+
"learning_rate": 0.0005647396082178691,
|
| 2699 |
+
"loss": 2.1383,
|
| 2700 |
+
"step": 446
|
| 2701 |
+
},
|
| 2702 |
+
{
|
| 2703 |
+
"epoch": 0.31,
|
| 2704 |
+
"learning_rate": 0.0005645962732919254,
|
| 2705 |
+
"loss": 2.1737,
|
| 2706 |
+
"step": 447
|
| 2707 |
+
},
|
| 2708 |
+
{
|
| 2709 |
+
"epoch": 0.31,
|
| 2710 |
+
"learning_rate": 0.0005644529383659818,
|
| 2711 |
+
"loss": 2.1792,
|
| 2712 |
+
"step": 448
|
| 2713 |
+
},
|
| 2714 |
+
{
|
| 2715 |
+
"epoch": 0.31,
|
| 2716 |
+
"learning_rate": 0.0005643096034400381,
|
| 2717 |
+
"loss": 2.2596,
|
| 2718 |
+
"step": 449
|
| 2719 |
+
},
|
| 2720 |
+
{
|
| 2721 |
+
"epoch": 0.31,
|
| 2722 |
+
"learning_rate": 0.0005641662685140946,
|
| 2723 |
+
"loss": 2.1804,
|
| 2724 |
+
"step": 450
|
| 2725 |
+
},
|
| 2726 |
+
{
|
| 2727 |
+
"epoch": 0.31,
|
| 2728 |
+
"learning_rate": 0.0005640229335881509,
|
| 2729 |
+
"loss": 2.0674,
|
| 2730 |
+
"step": 451
|
| 2731 |
+
},
|
| 2732 |
+
{
|
| 2733 |
+
"epoch": 0.31,
|
| 2734 |
+
"learning_rate": 0.0005638795986622074,
|
| 2735 |
+
"loss": 2.0845,
|
| 2736 |
+
"step": 452
|
| 2737 |
+
},
|
| 2738 |
+
{
|
| 2739 |
+
"epoch": 0.31,
|
| 2740 |
+
"learning_rate": 0.0005637362637362636,
|
| 2741 |
+
"loss": 2.0754,
|
| 2742 |
+
"step": 453
|
| 2743 |
+
},
|
| 2744 |
+
{
|
| 2745 |
+
"epoch": 0.31,
|
| 2746 |
+
"learning_rate": 0.00056359292881032,
|
| 2747 |
+
"loss": 2.1973,
|
| 2748 |
+
"step": 454
|
| 2749 |
+
},
|
| 2750 |
+
{
|
| 2751 |
+
"epoch": 0.31,
|
| 2752 |
+
"learning_rate": 0.0005634495938843764,
|
| 2753 |
+
"loss": 2.1183,
|
| 2754 |
+
"step": 455
|
| 2755 |
+
},
|
| 2756 |
+
{
|
| 2757 |
+
"epoch": 0.31,
|
| 2758 |
+
"learning_rate": 0.0005633062589584328,
|
| 2759 |
+
"loss": 2.1111,
|
| 2760 |
+
"step": 456
|
| 2761 |
+
},
|
| 2762 |
+
{
|
| 2763 |
+
"epoch": 0.31,
|
| 2764 |
+
"learning_rate": 0.0005631629240324892,
|
| 2765 |
+
"loss": 2.1094,
|
| 2766 |
+
"step": 457
|
| 2767 |
+
},
|
| 2768 |
+
{
|
| 2769 |
+
"epoch": 0.31,
|
| 2770 |
+
"learning_rate": 0.0005630195891065456,
|
| 2771 |
+
"loss": 2.0889,
|
| 2772 |
+
"step": 458
|
| 2773 |
+
},
|
| 2774 |
+
{
|
| 2775 |
+
"epoch": 0.31,
|
| 2776 |
+
"learning_rate": 0.0005628762541806019,
|
| 2777 |
+
"loss": 2.2136,
|
| 2778 |
+
"step": 459
|
| 2779 |
+
},
|
| 2780 |
+
{
|
| 2781 |
+
"epoch": 0.31,
|
| 2782 |
+
"learning_rate": 0.0005627329192546583,
|
| 2783 |
+
"loss": 2.1521,
|
| 2784 |
+
"step": 460
|
| 2785 |
+
},
|
| 2786 |
+
{
|
| 2787 |
+
"epoch": 0.32,
|
| 2788 |
+
"learning_rate": 0.0005625895843287147,
|
| 2789 |
+
"loss": 2.1277,
|
| 2790 |
+
"step": 461
|
| 2791 |
+
},
|
| 2792 |
+
{
|
| 2793 |
+
"epoch": 0.32,
|
| 2794 |
+
"learning_rate": 0.0005624462494027711,
|
| 2795 |
+
"loss": 2.2086,
|
| 2796 |
+
"step": 462
|
| 2797 |
+
},
|
| 2798 |
+
{
|
| 2799 |
+
"epoch": 0.32,
|
| 2800 |
+
"learning_rate": 0.0005623029144768275,
|
| 2801 |
+
"loss": 2.139,
|
| 2802 |
+
"step": 463
|
| 2803 |
+
},
|
| 2804 |
+
{
|
| 2805 |
+
"epoch": 0.32,
|
| 2806 |
+
"learning_rate": 0.0005621595795508839,
|
| 2807 |
+
"loss": 2.0627,
|
| 2808 |
+
"step": 464
|
| 2809 |
+
},
|
| 2810 |
+
{
|
| 2811 |
+
"epoch": 0.32,
|
| 2812 |
+
"learning_rate": 0.0005620162446249402,
|
| 2813 |
+
"loss": 2.2356,
|
| 2814 |
+
"step": 465
|
| 2815 |
+
},
|
| 2816 |
+
{
|
| 2817 |
+
"epoch": 0.32,
|
| 2818 |
+
"learning_rate": 0.0005618729096989966,
|
| 2819 |
+
"loss": 2.2947,
|
| 2820 |
+
"step": 466
|
| 2821 |
+
},
|
| 2822 |
+
{
|
| 2823 |
+
"epoch": 0.32,
|
| 2824 |
+
"learning_rate": 0.000561729574773053,
|
| 2825 |
+
"loss": 2.1619,
|
| 2826 |
+
"step": 467
|
| 2827 |
+
},
|
| 2828 |
+
{
|
| 2829 |
+
"epoch": 0.32,
|
| 2830 |
+
"learning_rate": 0.0005615862398471093,
|
| 2831 |
+
"loss": 2.2515,
|
| 2832 |
+
"step": 468
|
| 2833 |
+
},
|
| 2834 |
+
{
|
| 2835 |
+
"epoch": 0.32,
|
| 2836 |
+
"learning_rate": 0.0005614429049211658,
|
| 2837 |
+
"loss": 2.1746,
|
| 2838 |
+
"step": 469
|
| 2839 |
+
},
|
| 2840 |
+
{
|
| 2841 |
+
"epoch": 0.32,
|
| 2842 |
+
"learning_rate": 0.000561299569995222,
|
| 2843 |
+
"loss": 2.259,
|
| 2844 |
+
"step": 470
|
| 2845 |
+
},
|
| 2846 |
+
{
|
| 2847 |
+
"epoch": 0.32,
|
| 2848 |
+
"learning_rate": 0.0005611562350692785,
|
| 2849 |
+
"loss": 2.2093,
|
| 2850 |
+
"step": 471
|
| 2851 |
+
},
|
| 2852 |
+
{
|
| 2853 |
+
"epoch": 0.32,
|
| 2854 |
+
"learning_rate": 0.0005610129001433348,
|
| 2855 |
+
"loss": 2.2077,
|
| 2856 |
+
"step": 472
|
| 2857 |
+
},
|
| 2858 |
+
{
|
| 2859 |
+
"epoch": 0.32,
|
| 2860 |
+
"learning_rate": 0.0005608695652173913,
|
| 2861 |
+
"loss": 2.2464,
|
| 2862 |
+
"step": 473
|
| 2863 |
+
},
|
| 2864 |
+
{
|
| 2865 |
+
"epoch": 0.32,
|
| 2866 |
+
"learning_rate": 0.0005607262302914476,
|
| 2867 |
+
"loss": 2.0556,
|
| 2868 |
+
"step": 474
|
| 2869 |
+
},
|
| 2870 |
+
{
|
| 2871 |
+
"epoch": 0.32,
|
| 2872 |
+
"learning_rate": 0.000560582895365504,
|
| 2873 |
+
"loss": 2.135,
|
| 2874 |
+
"step": 475
|
| 2875 |
+
},
|
| 2876 |
+
{
|
| 2877 |
+
"epoch": 0.33,
|
| 2878 |
+
"learning_rate": 0.0005604395604395604,
|
| 2879 |
+
"loss": 2.1812,
|
| 2880 |
+
"step": 476
|
| 2881 |
+
},
|
| 2882 |
+
{
|
| 2883 |
+
"epoch": 0.33,
|
| 2884 |
+
"learning_rate": 0.0005602962255136167,
|
| 2885 |
+
"loss": 2.1082,
|
| 2886 |
+
"step": 477
|
| 2887 |
+
},
|
| 2888 |
+
{
|
| 2889 |
+
"epoch": 0.33,
|
| 2890 |
+
"learning_rate": 0.0005601528905876731,
|
| 2891 |
+
"loss": 2.0997,
|
| 2892 |
+
"step": 478
|
| 2893 |
+
},
|
| 2894 |
+
{
|
| 2895 |
+
"epoch": 0.33,
|
| 2896 |
+
"learning_rate": 0.0005600095556617295,
|
| 2897 |
+
"loss": 2.1532,
|
| 2898 |
+
"step": 479
|
| 2899 |
+
},
|
| 2900 |
+
{
|
| 2901 |
+
"epoch": 0.33,
|
| 2902 |
+
"learning_rate": 0.0005598662207357859,
|
| 2903 |
+
"loss": 2.1389,
|
| 2904 |
+
"step": 480
|
| 2905 |
+
},
|
| 2906 |
+
{
|
| 2907 |
+
"epoch": 0.33,
|
| 2908 |
+
"learning_rate": 0.0005597228858098423,
|
| 2909 |
+
"loss": 2.1122,
|
| 2910 |
+
"step": 481
|
| 2911 |
+
},
|
| 2912 |
+
{
|
| 2913 |
+
"epoch": 0.33,
|
| 2914 |
+
"learning_rate": 0.0005595795508838987,
|
| 2915 |
+
"loss": 2.1091,
|
| 2916 |
+
"step": 482
|
| 2917 |
+
},
|
| 2918 |
+
{
|
| 2919 |
+
"epoch": 0.33,
|
| 2920 |
+
"learning_rate": 0.000559436215957955,
|
| 2921 |
+
"loss": 2.1699,
|
| 2922 |
+
"step": 483
|
| 2923 |
+
},
|
| 2924 |
+
{
|
| 2925 |
+
"epoch": 0.33,
|
| 2926 |
+
"learning_rate": 0.0005592928810320114,
|
| 2927 |
+
"loss": 2.1043,
|
| 2928 |
+
"step": 484
|
| 2929 |
+
},
|
| 2930 |
+
{
|
| 2931 |
+
"epoch": 0.33,
|
| 2932 |
+
"learning_rate": 0.0005591495461060678,
|
| 2933 |
+
"loss": 2.143,
|
| 2934 |
+
"step": 485
|
| 2935 |
+
},
|
| 2936 |
+
{
|
| 2937 |
+
"epoch": 0.33,
|
| 2938 |
+
"learning_rate": 0.0005590062111801242,
|
| 2939 |
+
"loss": 2.1184,
|
| 2940 |
+
"step": 486
|
| 2941 |
+
},
|
| 2942 |
+
{
|
| 2943 |
+
"epoch": 0.33,
|
| 2944 |
+
"learning_rate": 0.0005588628762541806,
|
| 2945 |
+
"loss": 2.0185,
|
| 2946 |
+
"step": 487
|
| 2947 |
+
},
|
| 2948 |
+
{
|
| 2949 |
+
"epoch": 0.33,
|
| 2950 |
+
"learning_rate": 0.000558719541328237,
|
| 2951 |
+
"loss": 2.2015,
|
| 2952 |
+
"step": 488
|
| 2953 |
+
},
|
| 2954 |
+
{
|
| 2955 |
+
"epoch": 0.33,
|
| 2956 |
+
"learning_rate": 0.0005585762064022932,
|
| 2957 |
+
"loss": 2.2501,
|
| 2958 |
+
"step": 489
|
| 2959 |
+
},
|
| 2960 |
+
{
|
| 2961 |
+
"epoch": 0.34,
|
| 2962 |
+
"learning_rate": 0.0005584328714763497,
|
| 2963 |
+
"loss": 2.1088,
|
| 2964 |
+
"step": 490
|
| 2965 |
+
},
|
| 2966 |
+
{
|
| 2967 |
+
"epoch": 0.34,
|
| 2968 |
+
"learning_rate": 0.000558289536550406,
|
| 2969 |
+
"loss": 2.0932,
|
| 2970 |
+
"step": 491
|
| 2971 |
+
},
|
| 2972 |
+
{
|
| 2973 |
+
"epoch": 0.34,
|
| 2974 |
+
"learning_rate": 0.0005581462016244625,
|
| 2975 |
+
"loss": 2.1717,
|
| 2976 |
+
"step": 492
|
| 2977 |
+
},
|
| 2978 |
+
{
|
| 2979 |
+
"epoch": 0.34,
|
| 2980 |
+
"learning_rate": 0.0005580028666985188,
|
| 2981 |
+
"loss": 2.1509,
|
| 2982 |
+
"step": 493
|
| 2983 |
+
},
|
| 2984 |
+
{
|
| 2985 |
+
"epoch": 0.34,
|
| 2986 |
+
"learning_rate": 0.0005578595317725753,
|
| 2987 |
+
"loss": 2.1354,
|
| 2988 |
+
"step": 494
|
| 2989 |
+
},
|
| 2990 |
+
{
|
| 2991 |
+
"epoch": 0.34,
|
| 2992 |
+
"learning_rate": 0.0005577161968466315,
|
| 2993 |
+
"loss": 2.2007,
|
| 2994 |
+
"step": 495
|
| 2995 |
+
},
|
| 2996 |
+
{
|
| 2997 |
+
"epoch": 0.34,
|
| 2998 |
+
"learning_rate": 0.0005575728619206879,
|
| 2999 |
+
"loss": 2.0769,
|
| 3000 |
+
"step": 496
|
| 3001 |
+
},
|
| 3002 |
+
{
|
| 3003 |
+
"epoch": 0.34,
|
| 3004 |
+
"learning_rate": 0.0005574295269947443,
|
| 3005 |
+
"loss": 2.1375,
|
| 3006 |
+
"step": 497
|
| 3007 |
+
},
|
| 3008 |
+
{
|
| 3009 |
+
"epoch": 0.34,
|
| 3010 |
+
"learning_rate": 0.0005572861920688007,
|
| 3011 |
+
"loss": 2.1743,
|
| 3012 |
+
"step": 498
|
| 3013 |
+
},
|
| 3014 |
+
{
|
| 3015 |
+
"epoch": 0.34,
|
| 3016 |
+
"learning_rate": 0.0005571428571428571,
|
| 3017 |
+
"loss": 2.1998,
|
| 3018 |
+
"step": 499
|
| 3019 |
+
},
|
| 3020 |
+
{
|
| 3021 |
+
"epoch": 0.34,
|
| 3022 |
+
"learning_rate": 0.0005569995222169135,
|
| 3023 |
+
"loss": 2.2041,
|
| 3024 |
+
"step": 500
|
| 3025 |
+
},
|
| 3026 |
+
{
|
| 3027 |
+
"epoch": 0.34,
|
| 3028 |
+
"learning_rate": 0.0005568561872909698,
|
| 3029 |
+
"loss": 2.1778,
|
| 3030 |
+
"step": 501
|
| 3031 |
+
},
|
| 3032 |
+
{
|
| 3033 |
+
"epoch": 0.34,
|
| 3034 |
+
"learning_rate": 0.0005567128523650262,
|
| 3035 |
+
"loss": 2.2309,
|
| 3036 |
+
"step": 502
|
| 3037 |
+
},
|
| 3038 |
+
{
|
| 3039 |
+
"epoch": 0.34,
|
| 3040 |
+
"learning_rate": 0.0005565695174390826,
|
| 3041 |
+
"loss": 2.1322,
|
| 3042 |
+
"step": 503
|
| 3043 |
+
},
|
| 3044 |
+
{
|
| 3045 |
+
"epoch": 0.34,
|
| 3046 |
+
"learning_rate": 0.000556426182513139,
|
| 3047 |
+
"loss": 2.1951,
|
| 3048 |
+
"step": 504
|
| 3049 |
+
},
|
| 3050 |
+
{
|
| 3051 |
+
"epoch": 0.35,
|
| 3052 |
+
"learning_rate": 0.0005562828475871954,
|
| 3053 |
+
"loss": 2.2756,
|
| 3054 |
+
"step": 505
|
| 3055 |
+
},
|
| 3056 |
+
{
|
| 3057 |
+
"epoch": 0.35,
|
| 3058 |
+
"learning_rate": 0.0005561395126612518,
|
| 3059 |
+
"loss": 2.0972,
|
| 3060 |
+
"step": 506
|
| 3061 |
+
},
|
| 3062 |
+
{
|
| 3063 |
+
"epoch": 0.35,
|
| 3064 |
+
"learning_rate": 0.0005559961777353081,
|
| 3065 |
+
"loss": 2.138,
|
| 3066 |
+
"step": 507
|
| 3067 |
+
},
|
| 3068 |
+
{
|
| 3069 |
+
"epoch": 0.35,
|
| 3070 |
+
"learning_rate": 0.0005558528428093645,
|
| 3071 |
+
"loss": 2.1627,
|
| 3072 |
+
"step": 508
|
| 3073 |
+
},
|
| 3074 |
+
{
|
| 3075 |
+
"epoch": 0.35,
|
| 3076 |
+
"learning_rate": 0.0005557095078834209,
|
| 3077 |
+
"loss": 2.1876,
|
| 3078 |
+
"step": 509
|
| 3079 |
+
},
|
| 3080 |
+
{
|
| 3081 |
+
"epoch": 0.35,
|
| 3082 |
+
"learning_rate": 0.0005555661729574772,
|
| 3083 |
+
"loss": 2.1359,
|
| 3084 |
+
"step": 510
|
| 3085 |
+
},
|
| 3086 |
+
{
|
| 3087 |
+
"epoch": 0.35,
|
| 3088 |
+
"learning_rate": 0.0005554228380315337,
|
| 3089 |
+
"loss": 2.2495,
|
| 3090 |
+
"step": 511
|
| 3091 |
+
},
|
| 3092 |
+
{
|
| 3093 |
+
"epoch": 0.35,
|
| 3094 |
+
"learning_rate": 0.00055527950310559,
|
| 3095 |
+
"loss": 2.1951,
|
| 3096 |
+
"step": 512
|
| 3097 |
+
},
|
| 3098 |
+
{
|
| 3099 |
+
"epoch": 0.35,
|
| 3100 |
+
"learning_rate": 0.0005551361681796464,
|
| 3101 |
+
"loss": 2.1939,
|
| 3102 |
+
"step": 513
|
| 3103 |
+
},
|
| 3104 |
+
{
|
| 3105 |
+
"epoch": 0.35,
|
| 3106 |
+
"learning_rate": 0.0005549928332537027,
|
| 3107 |
+
"loss": 2.12,
|
| 3108 |
+
"step": 514
|
| 3109 |
+
},
|
| 3110 |
+
{
|
| 3111 |
+
"epoch": 0.35,
|
| 3112 |
+
"learning_rate": 0.0005548494983277592,
|
| 3113 |
+
"loss": 2.1258,
|
| 3114 |
+
"step": 515
|
| 3115 |
+
},
|
| 3116 |
+
{
|
| 3117 |
+
"epoch": 0.35,
|
| 3118 |
+
"learning_rate": 0.0005547061634018155,
|
| 3119 |
+
"loss": 2.2273,
|
| 3120 |
+
"step": 516
|
| 3121 |
+
},
|
| 3122 |
+
{
|
| 3123 |
+
"epoch": 0.35,
|
| 3124 |
+
"learning_rate": 0.0005545628284758719,
|
| 3125 |
+
"loss": 2.1856,
|
| 3126 |
+
"step": 517
|
| 3127 |
+
},
|
| 3128 |
+
{
|
| 3129 |
+
"epoch": 0.35,
|
| 3130 |
+
"learning_rate": 0.0005544194935499283,
|
| 3131 |
+
"loss": 2.0875,
|
| 3132 |
+
"step": 518
|
| 3133 |
+
},
|
| 3134 |
+
{
|
| 3135 |
+
"epoch": 0.35,
|
| 3136 |
+
"learning_rate": 0.0005542761586239846,
|
| 3137 |
+
"loss": 2.0916,
|
| 3138 |
+
"step": 519
|
| 3139 |
+
},
|
| 3140 |
+
{
|
| 3141 |
+
"epoch": 0.36,
|
| 3142 |
+
"learning_rate": 0.000554132823698041,
|
| 3143 |
+
"loss": 2.094,
|
| 3144 |
+
"step": 520
|
| 3145 |
+
},
|
| 3146 |
+
{
|
| 3147 |
+
"epoch": 0.36,
|
| 3148 |
+
"learning_rate": 0.0005539894887720974,
|
| 3149 |
+
"loss": 2.1244,
|
| 3150 |
+
"step": 521
|
| 3151 |
+
},
|
| 3152 |
+
{
|
| 3153 |
+
"epoch": 0.36,
|
| 3154 |
+
"learning_rate": 0.0005538461538461538,
|
| 3155 |
+
"loss": 2.1668,
|
| 3156 |
+
"step": 522
|
| 3157 |
+
},
|
| 3158 |
+
{
|
| 3159 |
+
"epoch": 0.36,
|
| 3160 |
+
"learning_rate": 0.0005537028189202102,
|
| 3161 |
+
"loss": 2.1785,
|
| 3162 |
+
"step": 523
|
| 3163 |
+
},
|
| 3164 |
+
{
|
| 3165 |
+
"epoch": 0.36,
|
| 3166 |
+
"learning_rate": 0.0005535594839942666,
|
| 3167 |
+
"loss": 2.1497,
|
| 3168 |
+
"step": 524
|
| 3169 |
+
},
|
| 3170 |
+
{
|
| 3171 |
+
"epoch": 0.36,
|
| 3172 |
+
"learning_rate": 0.0005534161490683229,
|
| 3173 |
+
"loss": 2.2136,
|
| 3174 |
+
"step": 525
|
| 3175 |
+
},
|
| 3176 |
+
{
|
| 3177 |
+
"epoch": 0.36,
|
| 3178 |
+
"learning_rate": 0.0005532728141423793,
|
| 3179 |
+
"loss": 2.1554,
|
| 3180 |
+
"step": 526
|
| 3181 |
+
},
|
| 3182 |
+
{
|
| 3183 |
+
"epoch": 0.36,
|
| 3184 |
+
"learning_rate": 0.0005531294792164357,
|
| 3185 |
+
"loss": 2.1288,
|
| 3186 |
+
"step": 527
|
| 3187 |
+
},
|
| 3188 |
+
{
|
| 3189 |
+
"epoch": 0.36,
|
| 3190 |
+
"learning_rate": 0.0005529861442904921,
|
| 3191 |
+
"loss": 2.1221,
|
| 3192 |
+
"step": 528
|
| 3193 |
+
},
|
| 3194 |
+
{
|
| 3195 |
+
"epoch": 0.36,
|
| 3196 |
+
"learning_rate": 0.0005528428093645485,
|
| 3197 |
+
"loss": 2.1583,
|
| 3198 |
+
"step": 529
|
| 3199 |
+
},
|
| 3200 |
+
{
|
| 3201 |
+
"epoch": 0.36,
|
| 3202 |
+
"learning_rate": 0.0005526994744386049,
|
| 3203 |
+
"loss": 2.1514,
|
| 3204 |
+
"step": 530
|
| 3205 |
+
},
|
| 3206 |
+
{
|
| 3207 |
+
"epoch": 0.36,
|
| 3208 |
+
"learning_rate": 0.0005525561395126611,
|
| 3209 |
+
"loss": 2.2256,
|
| 3210 |
+
"step": 531
|
| 3211 |
+
},
|
| 3212 |
+
{
|
| 3213 |
+
"epoch": 0.36,
|
| 3214 |
+
"learning_rate": 0.0005524128045867176,
|
| 3215 |
+
"loss": 2.1435,
|
| 3216 |
+
"step": 532
|
| 3217 |
+
},
|
| 3218 |
+
{
|
| 3219 |
+
"epoch": 0.36,
|
| 3220 |
+
"learning_rate": 0.0005522694696607739,
|
| 3221 |
+
"loss": 2.169,
|
| 3222 |
+
"step": 533
|
| 3223 |
+
},
|
| 3224 |
+
{
|
| 3225 |
+
"epoch": 0.37,
|
| 3226 |
+
"learning_rate": 0.0005521261347348304,
|
| 3227 |
+
"loss": 2.1734,
|
| 3228 |
+
"step": 534
|
| 3229 |
+
},
|
| 3230 |
+
{
|
| 3231 |
+
"epoch": 0.37,
|
| 3232 |
+
"learning_rate": 0.0005519827998088867,
|
| 3233 |
+
"loss": 2.0882,
|
| 3234 |
+
"step": 535
|
| 3235 |
+
},
|
| 3236 |
+
{
|
| 3237 |
+
"epoch": 0.37,
|
| 3238 |
+
"learning_rate": 0.0005518394648829432,
|
| 3239 |
+
"loss": 2.1364,
|
| 3240 |
+
"step": 536
|
| 3241 |
+
},
|
| 3242 |
+
{
|
| 3243 |
+
"epoch": 0.37,
|
| 3244 |
+
"learning_rate": 0.0005516961299569994,
|
| 3245 |
+
"loss": 2.1544,
|
| 3246 |
+
"step": 537
|
| 3247 |
+
},
|
| 3248 |
+
{
|
| 3249 |
+
"epoch": 0.37,
|
| 3250 |
+
"learning_rate": 0.0005515527950310558,
|
| 3251 |
+
"loss": 2.1356,
|
| 3252 |
+
"step": 538
|
| 3253 |
+
},
|
| 3254 |
+
{
|
| 3255 |
+
"epoch": 0.37,
|
| 3256 |
+
"learning_rate": 0.0005514094601051122,
|
| 3257 |
+
"loss": 2.019,
|
| 3258 |
+
"step": 539
|
| 3259 |
+
},
|
| 3260 |
+
{
|
| 3261 |
+
"epoch": 0.37,
|
| 3262 |
+
"learning_rate": 0.0005512661251791686,
|
| 3263 |
+
"loss": 2.1198,
|
| 3264 |
+
"step": 540
|
| 3265 |
+
},
|
| 3266 |
+
{
|
| 3267 |
+
"epoch": 0.37,
|
| 3268 |
+
"learning_rate": 0.000551122790253225,
|
| 3269 |
+
"loss": 2.1896,
|
| 3270 |
+
"step": 541
|
| 3271 |
+
},
|
| 3272 |
+
{
|
| 3273 |
+
"epoch": 0.37,
|
| 3274 |
+
"learning_rate": 0.0005509794553272814,
|
| 3275 |
+
"loss": 2.1615,
|
| 3276 |
+
"step": 542
|
| 3277 |
+
},
|
| 3278 |
+
{
|
| 3279 |
+
"epoch": 0.37,
|
| 3280 |
+
"learning_rate": 0.0005508361204013377,
|
| 3281 |
+
"loss": 2.1163,
|
| 3282 |
+
"step": 543
|
| 3283 |
+
},
|
| 3284 |
+
{
|
| 3285 |
+
"epoch": 0.37,
|
| 3286 |
+
"learning_rate": 0.0005506927854753941,
|
| 3287 |
+
"loss": 2.2056,
|
| 3288 |
+
"step": 544
|
| 3289 |
+
},
|
| 3290 |
+
{
|
| 3291 |
+
"epoch": 0.37,
|
| 3292 |
+
"learning_rate": 0.0005505494505494505,
|
| 3293 |
+
"loss": 2.2128,
|
| 3294 |
+
"step": 545
|
| 3295 |
+
},
|
| 3296 |
+
{
|
| 3297 |
+
"epoch": 0.37,
|
| 3298 |
+
"learning_rate": 0.0005504061156235069,
|
| 3299 |
+
"loss": 2.195,
|
| 3300 |
+
"step": 546
|
| 3301 |
+
},
|
| 3302 |
+
{
|
| 3303 |
+
"epoch": 0.37,
|
| 3304 |
+
"learning_rate": 0.0005502627806975633,
|
| 3305 |
+
"loss": 2.0768,
|
| 3306 |
+
"step": 547
|
| 3307 |
+
},
|
| 3308 |
+
{
|
| 3309 |
+
"epoch": 0.37,
|
| 3310 |
+
"learning_rate": 0.0005501194457716197,
|
| 3311 |
+
"loss": 2.1667,
|
| 3312 |
+
"step": 548
|
| 3313 |
+
},
|
| 3314 |
+
{
|
| 3315 |
+
"epoch": 0.38,
|
| 3316 |
+
"learning_rate": 0.000549976110845676,
|
| 3317 |
+
"loss": 2.1035,
|
| 3318 |
+
"step": 549
|
| 3319 |
+
},
|
| 3320 |
+
{
|
| 3321 |
+
"epoch": 0.38,
|
| 3322 |
+
"learning_rate": 0.0005498327759197323,
|
| 3323 |
+
"loss": 2.1628,
|
| 3324 |
+
"step": 550
|
| 3325 |
+
},
|
| 3326 |
+
{
|
| 3327 |
+
"epoch": 0.38,
|
| 3328 |
+
"learning_rate": 0.0005496894409937888,
|
| 3329 |
+
"loss": 2.1491,
|
| 3330 |
+
"step": 551
|
| 3331 |
+
},
|
| 3332 |
+
{
|
| 3333 |
+
"epoch": 0.38,
|
| 3334 |
+
"learning_rate": 0.0005495461060678451,
|
| 3335 |
+
"loss": 2.0676,
|
| 3336 |
+
"step": 552
|
| 3337 |
+
},
|
| 3338 |
+
{
|
| 3339 |
+
"epoch": 0.38,
|
| 3340 |
+
"learning_rate": 0.0005494027711419016,
|
| 3341 |
+
"loss": 2.3462,
|
| 3342 |
+
"step": 553
|
| 3343 |
+
},
|
| 3344 |
+
{
|
| 3345 |
+
"epoch": 0.38,
|
| 3346 |
+
"learning_rate": 0.0005492594362159579,
|
| 3347 |
+
"loss": 2.107,
|
| 3348 |
+
"step": 554
|
| 3349 |
+
},
|
| 3350 |
+
{
|
| 3351 |
+
"epoch": 0.38,
|
| 3352 |
+
"learning_rate": 0.0005491161012900143,
|
| 3353 |
+
"loss": 2.1844,
|
| 3354 |
+
"step": 555
|
| 3355 |
+
},
|
| 3356 |
+
{
|
| 3357 |
+
"epoch": 0.38,
|
| 3358 |
+
"learning_rate": 0.0005489727663640706,
|
| 3359 |
+
"loss": 2.1328,
|
| 3360 |
+
"step": 556
|
| 3361 |
+
},
|
| 3362 |
+
{
|
| 3363 |
+
"epoch": 0.38,
|
| 3364 |
+
"learning_rate": 0.000548829431438127,
|
| 3365 |
+
"loss": 2.1467,
|
| 3366 |
+
"step": 557
|
| 3367 |
+
},
|
| 3368 |
+
{
|
| 3369 |
+
"epoch": 0.38,
|
| 3370 |
+
"learning_rate": 0.0005486860965121834,
|
| 3371 |
+
"loss": 2.1706,
|
| 3372 |
+
"step": 558
|
| 3373 |
+
},
|
| 3374 |
+
{
|
| 3375 |
+
"epoch": 0.38,
|
| 3376 |
+
"learning_rate": 0.0005485427615862398,
|
| 3377 |
+
"loss": 2.1649,
|
| 3378 |
+
"step": 559
|
| 3379 |
+
},
|
| 3380 |
+
{
|
| 3381 |
+
"epoch": 0.38,
|
| 3382 |
+
"learning_rate": 0.0005483994266602962,
|
| 3383 |
+
"loss": 2.154,
|
| 3384 |
+
"step": 560
|
| 3385 |
+
},
|
| 3386 |
+
{
|
| 3387 |
+
"epoch": 0.38,
|
| 3388 |
+
"learning_rate": 0.0005482560917343525,
|
| 3389 |
+
"loss": 2.1095,
|
| 3390 |
+
"step": 561
|
| 3391 |
+
},
|
| 3392 |
+
{
|
| 3393 |
+
"epoch": 0.38,
|
| 3394 |
+
"learning_rate": 0.0005481127568084089,
|
| 3395 |
+
"loss": 2.1706,
|
| 3396 |
+
"step": 562
|
| 3397 |
+
},
|
| 3398 |
+
{
|
| 3399 |
+
"epoch": 0.38,
|
| 3400 |
+
"learning_rate": 0.0005479694218824653,
|
| 3401 |
+
"loss": 2.2179,
|
| 3402 |
+
"step": 563
|
| 3403 |
+
},
|
| 3404 |
+
{
|
| 3405 |
+
"epoch": 0.39,
|
| 3406 |
+
"learning_rate": 0.0005478260869565217,
|
| 3407 |
+
"loss": 2.2188,
|
| 3408 |
+
"step": 564
|
| 3409 |
+
},
|
| 3410 |
+
{
|
| 3411 |
+
"epoch": 0.39,
|
| 3412 |
+
"learning_rate": 0.0005476827520305781,
|
| 3413 |
+
"loss": 2.1167,
|
| 3414 |
+
"step": 565
|
| 3415 |
+
},
|
| 3416 |
+
{
|
| 3417 |
+
"epoch": 0.39,
|
| 3418 |
+
"learning_rate": 0.0005475394171046345,
|
| 3419 |
+
"loss": 2.1275,
|
| 3420 |
+
"step": 566
|
| 3421 |
+
},
|
| 3422 |
+
{
|
| 3423 |
+
"epoch": 0.39,
|
| 3424 |
+
"learning_rate": 0.0005473960821786908,
|
| 3425 |
+
"loss": 2.0766,
|
| 3426 |
+
"step": 567
|
| 3427 |
+
},
|
| 3428 |
+
{
|
| 3429 |
+
"epoch": 0.39,
|
| 3430 |
+
"learning_rate": 0.0005472527472527472,
|
| 3431 |
+
"loss": 2.1719,
|
| 3432 |
+
"step": 568
|
| 3433 |
+
},
|
| 3434 |
+
{
|
| 3435 |
+
"epoch": 0.39,
|
| 3436 |
+
"learning_rate": 0.0005471094123268036,
|
| 3437 |
+
"loss": 2.2378,
|
| 3438 |
+
"step": 569
|
| 3439 |
+
},
|
| 3440 |
+
{
|
| 3441 |
+
"epoch": 0.39,
|
| 3442 |
+
"learning_rate": 0.00054696607740086,
|
| 3443 |
+
"loss": 2.1612,
|
| 3444 |
+
"step": 570
|
| 3445 |
+
},
|
| 3446 |
+
{
|
| 3447 |
+
"epoch": 0.39,
|
| 3448 |
+
"learning_rate": 0.0005468227424749163,
|
| 3449 |
+
"loss": 2.0903,
|
| 3450 |
+
"step": 571
|
| 3451 |
+
},
|
| 3452 |
+
{
|
| 3453 |
+
"epoch": 0.39,
|
| 3454 |
+
"learning_rate": 0.0005466794075489728,
|
| 3455 |
+
"loss": 2.2279,
|
| 3456 |
+
"step": 572
|
| 3457 |
+
},
|
| 3458 |
+
{
|
| 3459 |
+
"epoch": 0.39,
|
| 3460 |
+
"learning_rate": 0.000546536072623029,
|
| 3461 |
+
"loss": 2.1084,
|
| 3462 |
+
"step": 573
|
| 3463 |
+
},
|
| 3464 |
+
{
|
| 3465 |
+
"epoch": 0.39,
|
| 3466 |
+
"learning_rate": 0.0005463927376970855,
|
| 3467 |
+
"loss": 2.1652,
|
| 3468 |
+
"step": 574
|
| 3469 |
+
},
|
| 3470 |
+
{
|
| 3471 |
+
"epoch": 0.39,
|
| 3472 |
+
"learning_rate": 0.0005462494027711418,
|
| 3473 |
+
"loss": 2.1035,
|
| 3474 |
+
"step": 575
|
| 3475 |
+
},
|
| 3476 |
+
{
|
| 3477 |
+
"epoch": 0.39,
|
| 3478 |
+
"learning_rate": 0.0005461060678451983,
|
| 3479 |
+
"loss": 2.2028,
|
| 3480 |
+
"step": 576
|
| 3481 |
+
},
|
| 3482 |
+
{
|
| 3483 |
+
"epoch": 0.39,
|
| 3484 |
+
"learning_rate": 0.0005459627329192546,
|
| 3485 |
+
"loss": 2.2347,
|
| 3486 |
+
"step": 577
|
| 3487 |
+
},
|
| 3488 |
+
{
|
| 3489 |
+
"epoch": 0.4,
|
| 3490 |
+
"learning_rate": 0.000545819397993311,
|
| 3491 |
+
"loss": 2.1609,
|
| 3492 |
+
"step": 578
|
| 3493 |
+
},
|
| 3494 |
+
{
|
| 3495 |
+
"epoch": 0.4,
|
| 3496 |
+
"learning_rate": 0.0005456760630673673,
|
| 3497 |
+
"loss": 2.0815,
|
| 3498 |
+
"step": 579
|
| 3499 |
+
},
|
| 3500 |
+
{
|
| 3501 |
+
"epoch": 0.4,
|
| 3502 |
+
"learning_rate": 0.0005455327281414237,
|
| 3503 |
+
"loss": 2.0575,
|
| 3504 |
+
"step": 580
|
| 3505 |
+
},
|
| 3506 |
+
{
|
| 3507 |
+
"epoch": 0.4,
|
| 3508 |
+
"learning_rate": 0.0005453893932154801,
|
| 3509 |
+
"loss": 2.1104,
|
| 3510 |
+
"step": 581
|
| 3511 |
+
},
|
| 3512 |
+
{
|
| 3513 |
+
"epoch": 0.4,
|
| 3514 |
+
"learning_rate": 0.0005452460582895365,
|
| 3515 |
+
"loss": 2.1784,
|
| 3516 |
+
"step": 582
|
| 3517 |
+
},
|
| 3518 |
+
{
|
| 3519 |
+
"epoch": 0.4,
|
| 3520 |
+
"learning_rate": 0.0005451027233635929,
|
| 3521 |
+
"loss": 2.117,
|
| 3522 |
+
"step": 583
|
| 3523 |
+
},
|
| 3524 |
+
{
|
| 3525 |
+
"epoch": 0.4,
|
| 3526 |
+
"learning_rate": 0.0005449593884376493,
|
| 3527 |
+
"loss": 2.0947,
|
| 3528 |
+
"step": 584
|
| 3529 |
+
},
|
| 3530 |
+
{
|
| 3531 |
+
"epoch": 0.4,
|
| 3532 |
+
"learning_rate": 0.0005448160535117056,
|
| 3533 |
+
"loss": 2.1204,
|
| 3534 |
+
"step": 585
|
| 3535 |
+
},
|
| 3536 |
+
{
|
| 3537 |
+
"epoch": 0.4,
|
| 3538 |
+
"learning_rate": 0.000544672718585762,
|
| 3539 |
+
"loss": 2.0614,
|
| 3540 |
+
"step": 586
|
| 3541 |
+
},
|
| 3542 |
+
{
|
| 3543 |
+
"epoch": 0.4,
|
| 3544 |
+
"learning_rate": 0.0005445293836598184,
|
| 3545 |
+
"loss": 2.1227,
|
| 3546 |
+
"step": 587
|
| 3547 |
+
},
|
| 3548 |
+
{
|
| 3549 |
+
"epoch": 0.4,
|
| 3550 |
+
"learning_rate": 0.0005443860487338748,
|
| 3551 |
+
"loss": 2.1831,
|
| 3552 |
+
"step": 588
|
| 3553 |
+
},
|
| 3554 |
+
{
|
| 3555 |
+
"epoch": 0.4,
|
| 3556 |
+
"learning_rate": 0.0005442427138079312,
|
| 3557 |
+
"loss": 2.2267,
|
| 3558 |
+
"step": 589
|
| 3559 |
+
},
|
| 3560 |
+
{
|
| 3561 |
+
"epoch": 0.4,
|
| 3562 |
+
"learning_rate": 0.0005440993788819876,
|
| 3563 |
+
"loss": 2.1008,
|
| 3564 |
+
"step": 590
|
| 3565 |
+
},
|
| 3566 |
+
{
|
| 3567 |
+
"epoch": 0.4,
|
| 3568 |
+
"learning_rate": 0.0005439560439560439,
|
| 3569 |
+
"loss": 2.1126,
|
| 3570 |
+
"step": 591
|
| 3571 |
+
},
|
| 3572 |
+
{
|
| 3573 |
+
"epoch": 0.4,
|
| 3574 |
+
"learning_rate": 0.0005438127090301002,
|
| 3575 |
+
"loss": 2.2081,
|
| 3576 |
+
"step": 592
|
| 3577 |
+
},
|
| 3578 |
+
{
|
| 3579 |
+
"epoch": 0.41,
|
| 3580 |
+
"learning_rate": 0.0005436693741041567,
|
| 3581 |
+
"loss": 2.2372,
|
| 3582 |
+
"step": 593
|
| 3583 |
+
},
|
| 3584 |
+
{
|
| 3585 |
+
"epoch": 0.41,
|
| 3586 |
+
"learning_rate": 0.000543526039178213,
|
| 3587 |
+
"loss": 2.1783,
|
| 3588 |
+
"step": 594
|
| 3589 |
+
},
|
| 3590 |
+
{
|
| 3591 |
+
"epoch": 0.41,
|
| 3592 |
+
"learning_rate": 0.0005433827042522695,
|
| 3593 |
+
"loss": 2.1631,
|
| 3594 |
+
"step": 595
|
| 3595 |
+
},
|
| 3596 |
+
{
|
| 3597 |
+
"epoch": 0.41,
|
| 3598 |
+
"learning_rate": 0.0005432393693263258,
|
| 3599 |
+
"loss": 2.2263,
|
| 3600 |
+
"step": 596
|
| 3601 |
+
},
|
| 3602 |
+
{
|
| 3603 |
+
"epoch": 0.41,
|
| 3604 |
+
"learning_rate": 0.0005430960344003822,
|
| 3605 |
+
"loss": 2.2438,
|
| 3606 |
+
"step": 597
|
| 3607 |
+
},
|
| 3608 |
+
{
|
| 3609 |
+
"epoch": 0.41,
|
| 3610 |
+
"learning_rate": 0.0005429526994744385,
|
| 3611 |
+
"loss": 2.125,
|
| 3612 |
+
"step": 598
|
| 3613 |
+
},
|
| 3614 |
+
{
|
| 3615 |
+
"epoch": 0.41,
|
| 3616 |
+
"learning_rate": 0.0005428093645484949,
|
| 3617 |
+
"loss": 2.1655,
|
| 3618 |
+
"step": 599
|
| 3619 |
+
},
|
| 3620 |
+
{
|
| 3621 |
+
"epoch": 0.41,
|
| 3622 |
+
"learning_rate": 0.0005426660296225513,
|
| 3623 |
+
"loss": 2.241,
|
| 3624 |
+
"step": 600
|
| 3625 |
+
},
|
| 3626 |
+
{
|
| 3627 |
+
"epoch": 0.41,
|
| 3628 |
+
"eval_loss": 2.047600030899048,
|
| 3629 |
+
"eval_runtime": 1736.5603,
|
| 3630 |
+
"eval_samples_per_second": 5.759,
|
| 3631 |
+
"eval_steps_per_second": 5.759,
|
| 3632 |
+
"step": 600
|
| 3633 |
}
|
| 3634 |
],
|
| 3635 |
"max_steps": 4386,
|
| 3636 |
"num_train_epochs": 3,
|
| 3637 |
+
"total_flos": 7.368734337271511e+17,
|
| 3638 |
"trial_name": null,
|
| 3639 |
"trial_params": null
|
| 3640 |
}
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 2368281769
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b1ac718a5450b8f5bc3e788c6d41748c97c82f6bf2933bbce997b0073b1c8202
|
| 3 |
size 2368281769
|