Training in progress, step 400
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 +1216 -3
- 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:f8c1215989d3c7f6317e0f6c05b46382b6e29fbdc1f07b62a204c7a5a8fc3559
|
| 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 sha256:6691fb6f696ed838034568580bf11105cd7d247925aaa8c39f26ca3e936ca47e
|
| 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:b1d5d151daf5d9763ae4661efceb395b789558b09ffc0615ad5aff28047b622d
|
| 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 sha256:3aa34fdf0557bf6f0ec97e73a9c403afbf62a6c74c782a9a21d2fd1b26413d2f
|
| 3 |
size 627
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": 2.1868629455566406,
|
| 3 |
"best_model_checkpoint": "output/checkpoint-200",
|
| 4 |
-
"epoch": 0.
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
@@ -1219,11 +1219,1224 @@
|
|
| 1219 |
"eval_samples_per_second": 0.056,
|
| 1220 |
"eval_steps_per_second": 0.056,
|
| 1221 |
"step": 200
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1222 |
}
|
| 1223 |
],
|
| 1224 |
"max_steps": 2000,
|
| 1225 |
"num_train_epochs": 9223372036854775807,
|
| 1226 |
-
"total_flos":
|
| 1227 |
"trial_name": null,
|
| 1228 |
"trial_params": null
|
| 1229 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": 2.1868629455566406,
|
| 3 |
"best_model_checkpoint": "output/checkpoint-200",
|
| 4 |
+
"epoch": 0.2,
|
| 5 |
+
"global_step": 400,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 1219 |
"eval_samples_per_second": 0.056,
|
| 1220 |
"eval_steps_per_second": 0.056,
|
| 1221 |
"step": 200
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 0.1,
|
| 1225 |
+
"learning_rate": 0.0007995555555555556,
|
| 1226 |
+
"loss": 2.2859,
|
| 1227 |
+
"step": 201
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 0.1,
|
| 1231 |
+
"learning_rate": 0.0007991111111111112,
|
| 1232 |
+
"loss": 2.2128,
|
| 1233 |
+
"step": 202
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 0.1,
|
| 1237 |
+
"learning_rate": 0.0007986666666666667,
|
| 1238 |
+
"loss": 2.254,
|
| 1239 |
+
"step": 203
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 0.1,
|
| 1243 |
+
"learning_rate": 0.0007982222222222223,
|
| 1244 |
+
"loss": 2.312,
|
| 1245 |
+
"step": 204
|
| 1246 |
+
},
|
| 1247 |
+
{
|
| 1248 |
+
"epoch": 0.1,
|
| 1249 |
+
"learning_rate": 0.0007977777777777778,
|
| 1250 |
+
"loss": 2.3304,
|
| 1251 |
+
"step": 205
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 0.1,
|
| 1255 |
+
"learning_rate": 0.0007973333333333334,
|
| 1256 |
+
"loss": 2.3126,
|
| 1257 |
+
"step": 206
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"epoch": 0.1,
|
| 1261 |
+
"learning_rate": 0.0007968888888888889,
|
| 1262 |
+
"loss": 2.3429,
|
| 1263 |
+
"step": 207
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 0.1,
|
| 1267 |
+
"learning_rate": 0.0007964444444444445,
|
| 1268 |
+
"loss": 2.3222,
|
| 1269 |
+
"step": 208
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 0.1,
|
| 1273 |
+
"learning_rate": 0.000796,
|
| 1274 |
+
"loss": 2.2903,
|
| 1275 |
+
"step": 209
|
| 1276 |
+
},
|
| 1277 |
+
{
|
| 1278 |
+
"epoch": 0.1,
|
| 1279 |
+
"learning_rate": 0.0007955555555555556,
|
| 1280 |
+
"loss": 2.3405,
|
| 1281 |
+
"step": 210
|
| 1282 |
+
},
|
| 1283 |
+
{
|
| 1284 |
+
"epoch": 0.11,
|
| 1285 |
+
"learning_rate": 0.0007951111111111112,
|
| 1286 |
+
"loss": 2.3691,
|
| 1287 |
+
"step": 211
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"epoch": 0.11,
|
| 1291 |
+
"learning_rate": 0.0007946666666666667,
|
| 1292 |
+
"loss": 2.2961,
|
| 1293 |
+
"step": 212
|
| 1294 |
+
},
|
| 1295 |
+
{
|
| 1296 |
+
"epoch": 0.11,
|
| 1297 |
+
"learning_rate": 0.0007942222222222222,
|
| 1298 |
+
"loss": 2.3318,
|
| 1299 |
+
"step": 213
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 0.11,
|
| 1303 |
+
"learning_rate": 0.0007937777777777778,
|
| 1304 |
+
"loss": 2.2794,
|
| 1305 |
+
"step": 214
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 0.11,
|
| 1309 |
+
"learning_rate": 0.0007933333333333334,
|
| 1310 |
+
"loss": 2.2903,
|
| 1311 |
+
"step": 215
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 0.11,
|
| 1315 |
+
"learning_rate": 0.000792888888888889,
|
| 1316 |
+
"loss": 2.321,
|
| 1317 |
+
"step": 216
|
| 1318 |
+
},
|
| 1319 |
+
{
|
| 1320 |
+
"epoch": 0.11,
|
| 1321 |
+
"learning_rate": 0.0007924444444444445,
|
| 1322 |
+
"loss": 2.2768,
|
| 1323 |
+
"step": 217
|
| 1324 |
+
},
|
| 1325 |
+
{
|
| 1326 |
+
"epoch": 0.11,
|
| 1327 |
+
"learning_rate": 0.0007920000000000001,
|
| 1328 |
+
"loss": 2.2916,
|
| 1329 |
+
"step": 218
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 0.11,
|
| 1333 |
+
"learning_rate": 0.0007915555555555557,
|
| 1334 |
+
"loss": 2.226,
|
| 1335 |
+
"step": 219
|
| 1336 |
+
},
|
| 1337 |
+
{
|
| 1338 |
+
"epoch": 0.11,
|
| 1339 |
+
"learning_rate": 0.0007911111111111111,
|
| 1340 |
+
"loss": 2.3054,
|
| 1341 |
+
"step": 220
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 0.11,
|
| 1345 |
+
"learning_rate": 0.0007906666666666666,
|
| 1346 |
+
"loss": 2.3064,
|
| 1347 |
+
"step": 221
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 0.11,
|
| 1351 |
+
"learning_rate": 0.0007902222222222222,
|
| 1352 |
+
"loss": 2.28,
|
| 1353 |
+
"step": 222
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 0.11,
|
| 1357 |
+
"learning_rate": 0.0007897777777777778,
|
| 1358 |
+
"loss": 2.3623,
|
| 1359 |
+
"step": 223
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 0.11,
|
| 1363 |
+
"learning_rate": 0.0007893333333333334,
|
| 1364 |
+
"loss": 2.3449,
|
| 1365 |
+
"step": 224
|
| 1366 |
+
},
|
| 1367 |
+
{
|
| 1368 |
+
"epoch": 0.11,
|
| 1369 |
+
"learning_rate": 0.000788888888888889,
|
| 1370 |
+
"loss": 2.2675,
|
| 1371 |
+
"step": 225
|
| 1372 |
+
},
|
| 1373 |
+
{
|
| 1374 |
+
"epoch": 0.11,
|
| 1375 |
+
"learning_rate": 0.0007884444444444445,
|
| 1376 |
+
"loss": 2.38,
|
| 1377 |
+
"step": 226
|
| 1378 |
+
},
|
| 1379 |
+
{
|
| 1380 |
+
"epoch": 0.11,
|
| 1381 |
+
"learning_rate": 0.0007880000000000001,
|
| 1382 |
+
"loss": 2.3044,
|
| 1383 |
+
"step": 227
|
| 1384 |
+
},
|
| 1385 |
+
{
|
| 1386 |
+
"epoch": 0.11,
|
| 1387 |
+
"learning_rate": 0.0007875555555555556,
|
| 1388 |
+
"loss": 2.3083,
|
| 1389 |
+
"step": 228
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 0.11,
|
| 1393 |
+
"learning_rate": 0.0007871111111111111,
|
| 1394 |
+
"loss": 2.1913,
|
| 1395 |
+
"step": 229
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 0.12,
|
| 1399 |
+
"learning_rate": 0.0007866666666666666,
|
| 1400 |
+
"loss": 2.2425,
|
| 1401 |
+
"step": 230
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 0.12,
|
| 1405 |
+
"learning_rate": 0.0007862222222222222,
|
| 1406 |
+
"loss": 2.3341,
|
| 1407 |
+
"step": 231
|
| 1408 |
+
},
|
| 1409 |
+
{
|
| 1410 |
+
"epoch": 0.12,
|
| 1411 |
+
"learning_rate": 0.0007857777777777778,
|
| 1412 |
+
"loss": 2.2575,
|
| 1413 |
+
"step": 232
|
| 1414 |
+
},
|
| 1415 |
+
{
|
| 1416 |
+
"epoch": 0.12,
|
| 1417 |
+
"learning_rate": 0.0007853333333333334,
|
| 1418 |
+
"loss": 2.3312,
|
| 1419 |
+
"step": 233
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 0.12,
|
| 1423 |
+
"learning_rate": 0.000784888888888889,
|
| 1424 |
+
"loss": 2.2838,
|
| 1425 |
+
"step": 234
|
| 1426 |
+
},
|
| 1427 |
+
{
|
| 1428 |
+
"epoch": 0.12,
|
| 1429 |
+
"learning_rate": 0.0007844444444444445,
|
| 1430 |
+
"loss": 2.2315,
|
| 1431 |
+
"step": 235
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 0.12,
|
| 1435 |
+
"learning_rate": 0.000784,
|
| 1436 |
+
"loss": 2.3253,
|
| 1437 |
+
"step": 236
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 0.12,
|
| 1441 |
+
"learning_rate": 0.0007835555555555556,
|
| 1442 |
+
"loss": 2.2557,
|
| 1443 |
+
"step": 237
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 0.12,
|
| 1447 |
+
"learning_rate": 0.0007831111111111112,
|
| 1448 |
+
"loss": 2.3264,
|
| 1449 |
+
"step": 238
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 0.12,
|
| 1453 |
+
"learning_rate": 0.0007826666666666668,
|
| 1454 |
+
"loss": 2.3318,
|
| 1455 |
+
"step": 239
|
| 1456 |
+
},
|
| 1457 |
+
{
|
| 1458 |
+
"epoch": 0.12,
|
| 1459 |
+
"learning_rate": 0.0007822222222222222,
|
| 1460 |
+
"loss": 2.3574,
|
| 1461 |
+
"step": 240
|
| 1462 |
+
},
|
| 1463 |
+
{
|
| 1464 |
+
"epoch": 0.12,
|
| 1465 |
+
"learning_rate": 0.0007817777777777778,
|
| 1466 |
+
"loss": 2.2863,
|
| 1467 |
+
"step": 241
|
| 1468 |
+
},
|
| 1469 |
+
{
|
| 1470 |
+
"epoch": 0.12,
|
| 1471 |
+
"learning_rate": 0.0007813333333333334,
|
| 1472 |
+
"loss": 2.3765,
|
| 1473 |
+
"step": 242
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 0.12,
|
| 1477 |
+
"learning_rate": 0.0007808888888888889,
|
| 1478 |
+
"loss": 2.2848,
|
| 1479 |
+
"step": 243
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 0.12,
|
| 1483 |
+
"learning_rate": 0.0007804444444444444,
|
| 1484 |
+
"loss": 2.3497,
|
| 1485 |
+
"step": 244
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 0.12,
|
| 1489 |
+
"learning_rate": 0.00078,
|
| 1490 |
+
"loss": 2.2716,
|
| 1491 |
+
"step": 245
|
| 1492 |
+
},
|
| 1493 |
+
{
|
| 1494 |
+
"epoch": 0.12,
|
| 1495 |
+
"learning_rate": 0.0007795555555555556,
|
| 1496 |
+
"loss": 2.3409,
|
| 1497 |
+
"step": 246
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 0.12,
|
| 1501 |
+
"learning_rate": 0.0007791111111111112,
|
| 1502 |
+
"loss": 2.3397,
|
| 1503 |
+
"step": 247
|
| 1504 |
+
},
|
| 1505 |
+
{
|
| 1506 |
+
"epoch": 0.12,
|
| 1507 |
+
"learning_rate": 0.0007786666666666668,
|
| 1508 |
+
"loss": 2.232,
|
| 1509 |
+
"step": 248
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 0.12,
|
| 1513 |
+
"learning_rate": 0.0007782222222222222,
|
| 1514 |
+
"loss": 2.2983,
|
| 1515 |
+
"step": 249
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 0.12,
|
| 1519 |
+
"learning_rate": 0.0007777777777777778,
|
| 1520 |
+
"loss": 2.3148,
|
| 1521 |
+
"step": 250
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 0.13,
|
| 1525 |
+
"learning_rate": 0.0007773333333333333,
|
| 1526 |
+
"loss": 2.2546,
|
| 1527 |
+
"step": 251
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 0.13,
|
| 1531 |
+
"learning_rate": 0.0007768888888888889,
|
| 1532 |
+
"loss": 2.2874,
|
| 1533 |
+
"step": 252
|
| 1534 |
+
},
|
| 1535 |
+
{
|
| 1536 |
+
"epoch": 0.13,
|
| 1537 |
+
"learning_rate": 0.0007764444444444444,
|
| 1538 |
+
"loss": 2.3787,
|
| 1539 |
+
"step": 253
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 0.13,
|
| 1543 |
+
"learning_rate": 0.000776,
|
| 1544 |
+
"loss": 2.3286,
|
| 1545 |
+
"step": 254
|
| 1546 |
+
},
|
| 1547 |
+
{
|
| 1548 |
+
"epoch": 0.13,
|
| 1549 |
+
"learning_rate": 0.0007755555555555556,
|
| 1550 |
+
"loss": 2.3242,
|
| 1551 |
+
"step": 255
|
| 1552 |
+
},
|
| 1553 |
+
{
|
| 1554 |
+
"epoch": 0.13,
|
| 1555 |
+
"learning_rate": 0.0007751111111111112,
|
| 1556 |
+
"loss": 2.2548,
|
| 1557 |
+
"step": 256
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 0.13,
|
| 1561 |
+
"learning_rate": 0.0007746666666666668,
|
| 1562 |
+
"loss": 2.2251,
|
| 1563 |
+
"step": 257
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 0.13,
|
| 1567 |
+
"learning_rate": 0.0007742222222222223,
|
| 1568 |
+
"loss": 2.291,
|
| 1569 |
+
"step": 258
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 0.13,
|
| 1573 |
+
"learning_rate": 0.0007737777777777779,
|
| 1574 |
+
"loss": 2.2951,
|
| 1575 |
+
"step": 259
|
| 1576 |
+
},
|
| 1577 |
+
{
|
| 1578 |
+
"epoch": 0.13,
|
| 1579 |
+
"learning_rate": 0.0007733333333333333,
|
| 1580 |
+
"loss": 2.2871,
|
| 1581 |
+
"step": 260
|
| 1582 |
+
},
|
| 1583 |
+
{
|
| 1584 |
+
"epoch": 0.13,
|
| 1585 |
+
"learning_rate": 0.0007728888888888889,
|
| 1586 |
+
"loss": 2.29,
|
| 1587 |
+
"step": 261
|
| 1588 |
+
},
|
| 1589 |
+
{
|
| 1590 |
+
"epoch": 0.13,
|
| 1591 |
+
"learning_rate": 0.0007724444444444444,
|
| 1592 |
+
"loss": 2.2888,
|
| 1593 |
+
"step": 262
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 0.13,
|
| 1597 |
+
"learning_rate": 0.000772,
|
| 1598 |
+
"loss": 2.3121,
|
| 1599 |
+
"step": 263
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 0.13,
|
| 1603 |
+
"learning_rate": 0.0007715555555555556,
|
| 1604 |
+
"loss": 2.2543,
|
| 1605 |
+
"step": 264
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 0.13,
|
| 1609 |
+
"learning_rate": 0.0007711111111111112,
|
| 1610 |
+
"loss": 2.2383,
|
| 1611 |
+
"step": 265
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 0.13,
|
| 1615 |
+
"learning_rate": 0.0007706666666666668,
|
| 1616 |
+
"loss": 2.3759,
|
| 1617 |
+
"step": 266
|
| 1618 |
+
},
|
| 1619 |
+
{
|
| 1620 |
+
"epoch": 0.13,
|
| 1621 |
+
"learning_rate": 0.0007702222222222223,
|
| 1622 |
+
"loss": 2.2838,
|
| 1623 |
+
"step": 267
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 0.13,
|
| 1627 |
+
"learning_rate": 0.0007697777777777778,
|
| 1628 |
+
"loss": 2.283,
|
| 1629 |
+
"step": 268
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 0.13,
|
| 1633 |
+
"learning_rate": 0.0007693333333333333,
|
| 1634 |
+
"loss": 2.3027,
|
| 1635 |
+
"step": 269
|
| 1636 |
+
},
|
| 1637 |
+
{
|
| 1638 |
+
"epoch": 0.14,
|
| 1639 |
+
"learning_rate": 0.0007688888888888889,
|
| 1640 |
+
"loss": 2.4145,
|
| 1641 |
+
"step": 270
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 0.14,
|
| 1645 |
+
"learning_rate": 0.0007684444444444445,
|
| 1646 |
+
"loss": 2.305,
|
| 1647 |
+
"step": 271
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 0.14,
|
| 1651 |
+
"learning_rate": 0.000768,
|
| 1652 |
+
"loss": 2.335,
|
| 1653 |
+
"step": 272
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 0.14,
|
| 1657 |
+
"learning_rate": 0.0007675555555555556,
|
| 1658 |
+
"loss": 2.3401,
|
| 1659 |
+
"step": 273
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 0.14,
|
| 1663 |
+
"learning_rate": 0.0007671111111111112,
|
| 1664 |
+
"loss": 2.3155,
|
| 1665 |
+
"step": 274
|
| 1666 |
+
},
|
| 1667 |
+
{
|
| 1668 |
+
"epoch": 0.14,
|
| 1669 |
+
"learning_rate": 0.0007666666666666667,
|
| 1670 |
+
"loss": 2.3303,
|
| 1671 |
+
"step": 275
|
| 1672 |
+
},
|
| 1673 |
+
{
|
| 1674 |
+
"epoch": 0.14,
|
| 1675 |
+
"learning_rate": 0.0007662222222222222,
|
| 1676 |
+
"loss": 2.2867,
|
| 1677 |
+
"step": 276
|
| 1678 |
+
},
|
| 1679 |
+
{
|
| 1680 |
+
"epoch": 0.14,
|
| 1681 |
+
"learning_rate": 0.0007657777777777778,
|
| 1682 |
+
"loss": 2.39,
|
| 1683 |
+
"step": 277
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 0.14,
|
| 1687 |
+
"learning_rate": 0.0007653333333333334,
|
| 1688 |
+
"loss": 2.2372,
|
| 1689 |
+
"step": 278
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 0.14,
|
| 1693 |
+
"learning_rate": 0.000764888888888889,
|
| 1694 |
+
"loss": 2.3359,
|
| 1695 |
+
"step": 279
|
| 1696 |
+
},
|
| 1697 |
+
{
|
| 1698 |
+
"epoch": 0.14,
|
| 1699 |
+
"learning_rate": 0.0007644444444444445,
|
| 1700 |
+
"loss": 2.2477,
|
| 1701 |
+
"step": 280
|
| 1702 |
+
},
|
| 1703 |
+
{
|
| 1704 |
+
"epoch": 0.14,
|
| 1705 |
+
"learning_rate": 0.000764,
|
| 1706 |
+
"loss": 2.1907,
|
| 1707 |
+
"step": 281
|
| 1708 |
+
},
|
| 1709 |
+
{
|
| 1710 |
+
"epoch": 0.14,
|
| 1711 |
+
"learning_rate": 0.0007635555555555556,
|
| 1712 |
+
"loss": 2.3424,
|
| 1713 |
+
"step": 282
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 0.14,
|
| 1717 |
+
"learning_rate": 0.0007631111111111111,
|
| 1718 |
+
"loss": 2.254,
|
| 1719 |
+
"step": 283
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 0.14,
|
| 1723 |
+
"learning_rate": 0.0007626666666666667,
|
| 1724 |
+
"loss": 2.2896,
|
| 1725 |
+
"step": 284
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 0.14,
|
| 1729 |
+
"learning_rate": 0.0007622222222222222,
|
| 1730 |
+
"loss": 2.2964,
|
| 1731 |
+
"step": 285
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 0.14,
|
| 1735 |
+
"learning_rate": 0.0007617777777777778,
|
| 1736 |
+
"loss": 2.2756,
|
| 1737 |
+
"step": 286
|
| 1738 |
+
},
|
| 1739 |
+
{
|
| 1740 |
+
"epoch": 0.14,
|
| 1741 |
+
"learning_rate": 0.0007613333333333334,
|
| 1742 |
+
"loss": 2.3348,
|
| 1743 |
+
"step": 287
|
| 1744 |
+
},
|
| 1745 |
+
{
|
| 1746 |
+
"epoch": 0.14,
|
| 1747 |
+
"learning_rate": 0.000760888888888889,
|
| 1748 |
+
"loss": 2.3344,
|
| 1749 |
+
"step": 288
|
| 1750 |
+
},
|
| 1751 |
+
{
|
| 1752 |
+
"epoch": 0.14,
|
| 1753 |
+
"learning_rate": 0.0007604444444444446,
|
| 1754 |
+
"loss": 2.2948,
|
| 1755 |
+
"step": 289
|
| 1756 |
+
},
|
| 1757 |
+
{
|
| 1758 |
+
"epoch": 0.14,
|
| 1759 |
+
"learning_rate": 0.00076,
|
| 1760 |
+
"loss": 2.3696,
|
| 1761 |
+
"step": 290
|
| 1762 |
+
},
|
| 1763 |
+
{
|
| 1764 |
+
"epoch": 0.15,
|
| 1765 |
+
"learning_rate": 0.0007595555555555555,
|
| 1766 |
+
"loss": 2.3261,
|
| 1767 |
+
"step": 291
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 0.15,
|
| 1771 |
+
"learning_rate": 0.0007591111111111111,
|
| 1772 |
+
"loss": 2.3931,
|
| 1773 |
+
"step": 292
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"epoch": 0.15,
|
| 1777 |
+
"learning_rate": 0.0007586666666666667,
|
| 1778 |
+
"loss": 2.3551,
|
| 1779 |
+
"step": 293
|
| 1780 |
+
},
|
| 1781 |
+
{
|
| 1782 |
+
"epoch": 0.15,
|
| 1783 |
+
"learning_rate": 0.0007582222222222223,
|
| 1784 |
+
"loss": 2.3412,
|
| 1785 |
+
"step": 294
|
| 1786 |
+
},
|
| 1787 |
+
{
|
| 1788 |
+
"epoch": 0.15,
|
| 1789 |
+
"learning_rate": 0.0007577777777777778,
|
| 1790 |
+
"loss": 2.3517,
|
| 1791 |
+
"step": 295
|
| 1792 |
+
},
|
| 1793 |
+
{
|
| 1794 |
+
"epoch": 0.15,
|
| 1795 |
+
"learning_rate": 0.0007573333333333334,
|
| 1796 |
+
"loss": 2.3178,
|
| 1797 |
+
"step": 296
|
| 1798 |
+
},
|
| 1799 |
+
{
|
| 1800 |
+
"epoch": 0.15,
|
| 1801 |
+
"learning_rate": 0.000756888888888889,
|
| 1802 |
+
"loss": 2.3577,
|
| 1803 |
+
"step": 297
|
| 1804 |
+
},
|
| 1805 |
+
{
|
| 1806 |
+
"epoch": 0.15,
|
| 1807 |
+
"learning_rate": 0.0007564444444444445,
|
| 1808 |
+
"loss": 2.2807,
|
| 1809 |
+
"step": 298
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 0.15,
|
| 1813 |
+
"learning_rate": 0.0007559999999999999,
|
| 1814 |
+
"loss": 2.3332,
|
| 1815 |
+
"step": 299
|
| 1816 |
+
},
|
| 1817 |
+
{
|
| 1818 |
+
"epoch": 0.15,
|
| 1819 |
+
"learning_rate": 0.0007555555555555555,
|
| 1820 |
+
"loss": 2.3292,
|
| 1821 |
+
"step": 300
|
| 1822 |
+
},
|
| 1823 |
+
{
|
| 1824 |
+
"epoch": 0.15,
|
| 1825 |
+
"learning_rate": 0.0007551111111111111,
|
| 1826 |
+
"loss": 2.2887,
|
| 1827 |
+
"step": 301
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 0.15,
|
| 1831 |
+
"learning_rate": 0.0007546666666666667,
|
| 1832 |
+
"loss": 2.3498,
|
| 1833 |
+
"step": 302
|
| 1834 |
+
},
|
| 1835 |
+
{
|
| 1836 |
+
"epoch": 0.15,
|
| 1837 |
+
"learning_rate": 0.0007542222222222223,
|
| 1838 |
+
"loss": 2.3857,
|
| 1839 |
+
"step": 303
|
| 1840 |
+
},
|
| 1841 |
+
{
|
| 1842 |
+
"epoch": 0.15,
|
| 1843 |
+
"learning_rate": 0.0007537777777777778,
|
| 1844 |
+
"loss": 2.3264,
|
| 1845 |
+
"step": 304
|
| 1846 |
+
},
|
| 1847 |
+
{
|
| 1848 |
+
"epoch": 0.15,
|
| 1849 |
+
"learning_rate": 0.0007533333333333334,
|
| 1850 |
+
"loss": 2.2637,
|
| 1851 |
+
"step": 305
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 0.15,
|
| 1855 |
+
"learning_rate": 0.0007528888888888889,
|
| 1856 |
+
"loss": 2.3084,
|
| 1857 |
+
"step": 306
|
| 1858 |
+
},
|
| 1859 |
+
{
|
| 1860 |
+
"epoch": 0.15,
|
| 1861 |
+
"learning_rate": 0.0007524444444444445,
|
| 1862 |
+
"loss": 2.3164,
|
| 1863 |
+
"step": 307
|
| 1864 |
+
},
|
| 1865 |
+
{
|
| 1866 |
+
"epoch": 0.15,
|
| 1867 |
+
"learning_rate": 0.000752,
|
| 1868 |
+
"loss": 2.3356,
|
| 1869 |
+
"step": 308
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 0.15,
|
| 1873 |
+
"learning_rate": 0.0007515555555555555,
|
| 1874 |
+
"loss": 2.3407,
|
| 1875 |
+
"step": 309
|
| 1876 |
+
},
|
| 1877 |
+
{
|
| 1878 |
+
"epoch": 0.15,
|
| 1879 |
+
"learning_rate": 0.0007511111111111111,
|
| 1880 |
+
"loss": 2.3375,
|
| 1881 |
+
"step": 310
|
| 1882 |
+
},
|
| 1883 |
+
{
|
| 1884 |
+
"epoch": 0.16,
|
| 1885 |
+
"learning_rate": 0.0007506666666666667,
|
| 1886 |
+
"loss": 2.3178,
|
| 1887 |
+
"step": 311
|
| 1888 |
+
},
|
| 1889 |
+
{
|
| 1890 |
+
"epoch": 0.16,
|
| 1891 |
+
"learning_rate": 0.0007502222222222223,
|
| 1892 |
+
"loss": 2.379,
|
| 1893 |
+
"step": 312
|
| 1894 |
+
},
|
| 1895 |
+
{
|
| 1896 |
+
"epoch": 0.16,
|
| 1897 |
+
"learning_rate": 0.0007497777777777778,
|
| 1898 |
+
"loss": 2.3506,
|
| 1899 |
+
"step": 313
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 0.16,
|
| 1903 |
+
"learning_rate": 0.0007493333333333334,
|
| 1904 |
+
"loss": 2.3232,
|
| 1905 |
+
"step": 314
|
| 1906 |
+
},
|
| 1907 |
+
{
|
| 1908 |
+
"epoch": 0.16,
|
| 1909 |
+
"learning_rate": 0.0007488888888888889,
|
| 1910 |
+
"loss": 2.3266,
|
| 1911 |
+
"step": 315
|
| 1912 |
+
},
|
| 1913 |
+
{
|
| 1914 |
+
"epoch": 0.16,
|
| 1915 |
+
"learning_rate": 0.0007484444444444445,
|
| 1916 |
+
"loss": 2.2864,
|
| 1917 |
+
"step": 316
|
| 1918 |
+
},
|
| 1919 |
+
{
|
| 1920 |
+
"epoch": 0.16,
|
| 1921 |
+
"learning_rate": 0.0007480000000000001,
|
| 1922 |
+
"loss": 2.2662,
|
| 1923 |
+
"step": 317
|
| 1924 |
+
},
|
| 1925 |
+
{
|
| 1926 |
+
"epoch": 0.16,
|
| 1927 |
+
"learning_rate": 0.0007475555555555556,
|
| 1928 |
+
"loss": 2.3208,
|
| 1929 |
+
"step": 318
|
| 1930 |
+
},
|
| 1931 |
+
{
|
| 1932 |
+
"epoch": 0.16,
|
| 1933 |
+
"learning_rate": 0.0007471111111111112,
|
| 1934 |
+
"loss": 2.3149,
|
| 1935 |
+
"step": 319
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 0.16,
|
| 1939 |
+
"learning_rate": 0.0007466666666666667,
|
| 1940 |
+
"loss": 2.2972,
|
| 1941 |
+
"step": 320
|
| 1942 |
+
},
|
| 1943 |
+
{
|
| 1944 |
+
"epoch": 0.16,
|
| 1945 |
+
"learning_rate": 0.0007462222222222223,
|
| 1946 |
+
"loss": 2.4095,
|
| 1947 |
+
"step": 321
|
| 1948 |
+
},
|
| 1949 |
+
{
|
| 1950 |
+
"epoch": 0.16,
|
| 1951 |
+
"learning_rate": 0.0007457777777777778,
|
| 1952 |
+
"loss": 2.3448,
|
| 1953 |
+
"step": 322
|
| 1954 |
+
},
|
| 1955 |
+
{
|
| 1956 |
+
"epoch": 0.16,
|
| 1957 |
+
"learning_rate": 0.0007453333333333333,
|
| 1958 |
+
"loss": 2.2673,
|
| 1959 |
+
"step": 323
|
| 1960 |
+
},
|
| 1961 |
+
{
|
| 1962 |
+
"epoch": 0.16,
|
| 1963 |
+
"learning_rate": 0.0007448888888888889,
|
| 1964 |
+
"loss": 2.345,
|
| 1965 |
+
"step": 324
|
| 1966 |
+
},
|
| 1967 |
+
{
|
| 1968 |
+
"epoch": 0.16,
|
| 1969 |
+
"learning_rate": 0.0007444444444444445,
|
| 1970 |
+
"loss": 2.2822,
|
| 1971 |
+
"step": 325
|
| 1972 |
+
},
|
| 1973 |
+
{
|
| 1974 |
+
"epoch": 0.16,
|
| 1975 |
+
"learning_rate": 0.0007440000000000001,
|
| 1976 |
+
"loss": 2.2933,
|
| 1977 |
+
"step": 326
|
| 1978 |
+
},
|
| 1979 |
+
{
|
| 1980 |
+
"epoch": 0.16,
|
| 1981 |
+
"learning_rate": 0.0007435555555555556,
|
| 1982 |
+
"loss": 2.2883,
|
| 1983 |
+
"step": 327
|
| 1984 |
+
},
|
| 1985 |
+
{
|
| 1986 |
+
"epoch": 0.16,
|
| 1987 |
+
"learning_rate": 0.0007431111111111112,
|
| 1988 |
+
"loss": 2.2863,
|
| 1989 |
+
"step": 328
|
| 1990 |
+
},
|
| 1991 |
+
{
|
| 1992 |
+
"epoch": 0.16,
|
| 1993 |
+
"learning_rate": 0.0007426666666666668,
|
| 1994 |
+
"loss": 2.273,
|
| 1995 |
+
"step": 329
|
| 1996 |
+
},
|
| 1997 |
+
{
|
| 1998 |
+
"epoch": 0.17,
|
| 1999 |
+
"learning_rate": 0.0007422222222222222,
|
| 2000 |
+
"loss": 2.281,
|
| 2001 |
+
"step": 330
|
| 2002 |
+
},
|
| 2003 |
+
{
|
| 2004 |
+
"epoch": 0.17,
|
| 2005 |
+
"learning_rate": 0.0007417777777777777,
|
| 2006 |
+
"loss": 2.2877,
|
| 2007 |
+
"step": 331
|
| 2008 |
+
},
|
| 2009 |
+
{
|
| 2010 |
+
"epoch": 0.17,
|
| 2011 |
+
"learning_rate": 0.0007413333333333333,
|
| 2012 |
+
"loss": 2.2801,
|
| 2013 |
+
"step": 332
|
| 2014 |
+
},
|
| 2015 |
+
{
|
| 2016 |
+
"epoch": 0.17,
|
| 2017 |
+
"learning_rate": 0.0007408888888888889,
|
| 2018 |
+
"loss": 2.3044,
|
| 2019 |
+
"step": 333
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 0.17,
|
| 2023 |
+
"learning_rate": 0.0007404444444444445,
|
| 2024 |
+
"loss": 2.3417,
|
| 2025 |
+
"step": 334
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 0.17,
|
| 2029 |
+
"learning_rate": 0.0007400000000000001,
|
| 2030 |
+
"loss": 2.3744,
|
| 2031 |
+
"step": 335
|
| 2032 |
+
},
|
| 2033 |
+
{
|
| 2034 |
+
"epoch": 0.17,
|
| 2035 |
+
"learning_rate": 0.0007395555555555556,
|
| 2036 |
+
"loss": 2.3285,
|
| 2037 |
+
"step": 336
|
| 2038 |
+
},
|
| 2039 |
+
{
|
| 2040 |
+
"epoch": 0.17,
|
| 2041 |
+
"learning_rate": 0.0007391111111111112,
|
| 2042 |
+
"loss": 2.3032,
|
| 2043 |
+
"step": 337
|
| 2044 |
+
},
|
| 2045 |
+
{
|
| 2046 |
+
"epoch": 0.17,
|
| 2047 |
+
"learning_rate": 0.0007386666666666667,
|
| 2048 |
+
"loss": 2.275,
|
| 2049 |
+
"step": 338
|
| 2050 |
+
},
|
| 2051 |
+
{
|
| 2052 |
+
"epoch": 0.17,
|
| 2053 |
+
"learning_rate": 0.0007382222222222223,
|
| 2054 |
+
"loss": 2.3184,
|
| 2055 |
+
"step": 339
|
| 2056 |
+
},
|
| 2057 |
+
{
|
| 2058 |
+
"epoch": 0.17,
|
| 2059 |
+
"learning_rate": 0.0007377777777777778,
|
| 2060 |
+
"loss": 2.4252,
|
| 2061 |
+
"step": 340
|
| 2062 |
+
},
|
| 2063 |
+
{
|
| 2064 |
+
"epoch": 0.17,
|
| 2065 |
+
"learning_rate": 0.0007373333333333333,
|
| 2066 |
+
"loss": 2.3806,
|
| 2067 |
+
"step": 341
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 0.17,
|
| 2071 |
+
"learning_rate": 0.0007368888888888889,
|
| 2072 |
+
"loss": 2.4038,
|
| 2073 |
+
"step": 342
|
| 2074 |
+
},
|
| 2075 |
+
{
|
| 2076 |
+
"epoch": 0.17,
|
| 2077 |
+
"learning_rate": 0.0007364444444444445,
|
| 2078 |
+
"loss": 2.4387,
|
| 2079 |
+
"step": 343
|
| 2080 |
+
},
|
| 2081 |
+
{
|
| 2082 |
+
"epoch": 0.17,
|
| 2083 |
+
"learning_rate": 0.0007360000000000001,
|
| 2084 |
+
"loss": 2.2712,
|
| 2085 |
+
"step": 344
|
| 2086 |
+
},
|
| 2087 |
+
{
|
| 2088 |
+
"epoch": 0.17,
|
| 2089 |
+
"learning_rate": 0.0007355555555555556,
|
| 2090 |
+
"loss": 2.3618,
|
| 2091 |
+
"step": 345
|
| 2092 |
+
},
|
| 2093 |
+
{
|
| 2094 |
+
"epoch": 0.17,
|
| 2095 |
+
"learning_rate": 0.0007351111111111111,
|
| 2096 |
+
"loss": 2.3015,
|
| 2097 |
+
"step": 346
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"epoch": 0.17,
|
| 2101 |
+
"learning_rate": 0.0007346666666666667,
|
| 2102 |
+
"loss": 2.2618,
|
| 2103 |
+
"step": 347
|
| 2104 |
+
},
|
| 2105 |
+
{
|
| 2106 |
+
"epoch": 0.17,
|
| 2107 |
+
"learning_rate": 0.0007342222222222223,
|
| 2108 |
+
"loss": 2.3319,
|
| 2109 |
+
"step": 348
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 0.17,
|
| 2113 |
+
"learning_rate": 0.0007337777777777779,
|
| 2114 |
+
"loss": 2.2604,
|
| 2115 |
+
"step": 349
|
| 2116 |
+
},
|
| 2117 |
+
{
|
| 2118 |
+
"epoch": 0.17,
|
| 2119 |
+
"learning_rate": 0.0007333333333333333,
|
| 2120 |
+
"loss": 2.3124,
|
| 2121 |
+
"step": 350
|
| 2122 |
+
},
|
| 2123 |
+
{
|
| 2124 |
+
"epoch": 0.18,
|
| 2125 |
+
"learning_rate": 0.0007328888888888889,
|
| 2126 |
+
"loss": 2.3053,
|
| 2127 |
+
"step": 351
|
| 2128 |
+
},
|
| 2129 |
+
{
|
| 2130 |
+
"epoch": 0.18,
|
| 2131 |
+
"learning_rate": 0.0007324444444444445,
|
| 2132 |
+
"loss": 2.2509,
|
| 2133 |
+
"step": 352
|
| 2134 |
+
},
|
| 2135 |
+
{
|
| 2136 |
+
"epoch": 0.18,
|
| 2137 |
+
"learning_rate": 0.000732,
|
| 2138 |
+
"loss": 2.2795,
|
| 2139 |
+
"step": 353
|
| 2140 |
+
},
|
| 2141 |
+
{
|
| 2142 |
+
"epoch": 0.18,
|
| 2143 |
+
"learning_rate": 0.0007315555555555555,
|
| 2144 |
+
"loss": 2.3321,
|
| 2145 |
+
"step": 354
|
| 2146 |
+
},
|
| 2147 |
+
{
|
| 2148 |
+
"epoch": 0.18,
|
| 2149 |
+
"learning_rate": 0.0007311111111111111,
|
| 2150 |
+
"loss": 2.3181,
|
| 2151 |
+
"step": 355
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 0.18,
|
| 2155 |
+
"learning_rate": 0.0007306666666666667,
|
| 2156 |
+
"loss": 2.2793,
|
| 2157 |
+
"step": 356
|
| 2158 |
+
},
|
| 2159 |
+
{
|
| 2160 |
+
"epoch": 0.18,
|
| 2161 |
+
"learning_rate": 0.0007302222222222223,
|
| 2162 |
+
"loss": 2.2477,
|
| 2163 |
+
"step": 357
|
| 2164 |
+
},
|
| 2165 |
+
{
|
| 2166 |
+
"epoch": 0.18,
|
| 2167 |
+
"learning_rate": 0.0007297777777777779,
|
| 2168 |
+
"loss": 2.3584,
|
| 2169 |
+
"step": 358
|
| 2170 |
+
},
|
| 2171 |
+
{
|
| 2172 |
+
"epoch": 0.18,
|
| 2173 |
+
"learning_rate": 0.0007293333333333334,
|
| 2174 |
+
"loss": 2.2824,
|
| 2175 |
+
"step": 359
|
| 2176 |
+
},
|
| 2177 |
+
{
|
| 2178 |
+
"epoch": 0.18,
|
| 2179 |
+
"learning_rate": 0.0007288888888888889,
|
| 2180 |
+
"loss": 2.2678,
|
| 2181 |
+
"step": 360
|
| 2182 |
+
},
|
| 2183 |
+
{
|
| 2184 |
+
"epoch": 0.18,
|
| 2185 |
+
"learning_rate": 0.0007284444444444445,
|
| 2186 |
+
"loss": 2.2886,
|
| 2187 |
+
"step": 361
|
| 2188 |
+
},
|
| 2189 |
+
{
|
| 2190 |
+
"epoch": 0.18,
|
| 2191 |
+
"learning_rate": 0.000728,
|
| 2192 |
+
"loss": 2.3444,
|
| 2193 |
+
"step": 362
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 0.18,
|
| 2197 |
+
"learning_rate": 0.0007275555555555556,
|
| 2198 |
+
"loss": 2.2967,
|
| 2199 |
+
"step": 363
|
| 2200 |
+
},
|
| 2201 |
+
{
|
| 2202 |
+
"epoch": 0.18,
|
| 2203 |
+
"learning_rate": 0.0007271111111111111,
|
| 2204 |
+
"loss": 2.334,
|
| 2205 |
+
"step": 364
|
| 2206 |
+
},
|
| 2207 |
+
{
|
| 2208 |
+
"epoch": 0.18,
|
| 2209 |
+
"learning_rate": 0.0007266666666666667,
|
| 2210 |
+
"loss": 2.2629,
|
| 2211 |
+
"step": 365
|
| 2212 |
+
},
|
| 2213 |
+
{
|
| 2214 |
+
"epoch": 0.18,
|
| 2215 |
+
"learning_rate": 0.0007262222222222223,
|
| 2216 |
+
"loss": 2.3195,
|
| 2217 |
+
"step": 366
|
| 2218 |
+
},
|
| 2219 |
+
{
|
| 2220 |
+
"epoch": 0.18,
|
| 2221 |
+
"learning_rate": 0.0007257777777777779,
|
| 2222 |
+
"loss": 2.2583,
|
| 2223 |
+
"step": 367
|
| 2224 |
+
},
|
| 2225 |
+
{
|
| 2226 |
+
"epoch": 0.18,
|
| 2227 |
+
"learning_rate": 0.0007253333333333334,
|
| 2228 |
+
"loss": 2.2994,
|
| 2229 |
+
"step": 368
|
| 2230 |
+
},
|
| 2231 |
+
{
|
| 2232 |
+
"epoch": 0.18,
|
| 2233 |
+
"learning_rate": 0.000724888888888889,
|
| 2234 |
+
"loss": 2.307,
|
| 2235 |
+
"step": 369
|
| 2236 |
+
},
|
| 2237 |
+
{
|
| 2238 |
+
"epoch": 0.18,
|
| 2239 |
+
"learning_rate": 0.0007244444444444444,
|
| 2240 |
+
"loss": 2.3516,
|
| 2241 |
+
"step": 370
|
| 2242 |
+
},
|
| 2243 |
+
{
|
| 2244 |
+
"epoch": 0.19,
|
| 2245 |
+
"learning_rate": 0.000724,
|
| 2246 |
+
"loss": 2.3624,
|
| 2247 |
+
"step": 371
|
| 2248 |
+
},
|
| 2249 |
+
{
|
| 2250 |
+
"epoch": 0.19,
|
| 2251 |
+
"learning_rate": 0.0007235555555555556,
|
| 2252 |
+
"loss": 2.3008,
|
| 2253 |
+
"step": 372
|
| 2254 |
+
},
|
| 2255 |
+
{
|
| 2256 |
+
"epoch": 0.19,
|
| 2257 |
+
"learning_rate": 0.0007231111111111111,
|
| 2258 |
+
"loss": 2.3502,
|
| 2259 |
+
"step": 373
|
| 2260 |
+
},
|
| 2261 |
+
{
|
| 2262 |
+
"epoch": 0.19,
|
| 2263 |
+
"learning_rate": 0.0007226666666666667,
|
| 2264 |
+
"loss": 2.3483,
|
| 2265 |
+
"step": 374
|
| 2266 |
+
},
|
| 2267 |
+
{
|
| 2268 |
+
"epoch": 0.19,
|
| 2269 |
+
"learning_rate": 0.0007222222222222223,
|
| 2270 |
+
"loss": 2.3154,
|
| 2271 |
+
"step": 375
|
| 2272 |
+
},
|
| 2273 |
+
{
|
| 2274 |
+
"epoch": 0.19,
|
| 2275 |
+
"learning_rate": 0.0007217777777777779,
|
| 2276 |
+
"loss": 2.2634,
|
| 2277 |
+
"step": 376
|
| 2278 |
+
},
|
| 2279 |
+
{
|
| 2280 |
+
"epoch": 0.19,
|
| 2281 |
+
"learning_rate": 0.0007213333333333334,
|
| 2282 |
+
"loss": 2.2936,
|
| 2283 |
+
"step": 377
|
| 2284 |
+
},
|
| 2285 |
+
{
|
| 2286 |
+
"epoch": 0.19,
|
| 2287 |
+
"learning_rate": 0.0007208888888888889,
|
| 2288 |
+
"loss": 2.3522,
|
| 2289 |
+
"step": 378
|
| 2290 |
+
},
|
| 2291 |
+
{
|
| 2292 |
+
"epoch": 0.19,
|
| 2293 |
+
"learning_rate": 0.0007204444444444445,
|
| 2294 |
+
"loss": 2.2925,
|
| 2295 |
+
"step": 379
|
| 2296 |
+
},
|
| 2297 |
+
{
|
| 2298 |
+
"epoch": 0.19,
|
| 2299 |
+
"learning_rate": 0.00072,
|
| 2300 |
+
"loss": 2.2873,
|
| 2301 |
+
"step": 380
|
| 2302 |
+
},
|
| 2303 |
+
{
|
| 2304 |
+
"epoch": 0.19,
|
| 2305 |
+
"learning_rate": 0.0007195555555555556,
|
| 2306 |
+
"loss": 2.3467,
|
| 2307 |
+
"step": 381
|
| 2308 |
+
},
|
| 2309 |
+
{
|
| 2310 |
+
"epoch": 0.19,
|
| 2311 |
+
"learning_rate": 0.0007191111111111111,
|
| 2312 |
+
"loss": 2.3031,
|
| 2313 |
+
"step": 382
|
| 2314 |
+
},
|
| 2315 |
+
{
|
| 2316 |
+
"epoch": 0.19,
|
| 2317 |
+
"learning_rate": 0.0007186666666666667,
|
| 2318 |
+
"loss": 2.3181,
|
| 2319 |
+
"step": 383
|
| 2320 |
+
},
|
| 2321 |
+
{
|
| 2322 |
+
"epoch": 0.19,
|
| 2323 |
+
"learning_rate": 0.0007182222222222223,
|
| 2324 |
+
"loss": 2.3257,
|
| 2325 |
+
"step": 384
|
| 2326 |
+
},
|
| 2327 |
+
{
|
| 2328 |
+
"epoch": 0.19,
|
| 2329 |
+
"learning_rate": 0.0007177777777777778,
|
| 2330 |
+
"loss": 2.2915,
|
| 2331 |
+
"step": 385
|
| 2332 |
+
},
|
| 2333 |
+
{
|
| 2334 |
+
"epoch": 0.19,
|
| 2335 |
+
"learning_rate": 0.0007173333333333333,
|
| 2336 |
+
"loss": 2.3113,
|
| 2337 |
+
"step": 386
|
| 2338 |
+
},
|
| 2339 |
+
{
|
| 2340 |
+
"epoch": 0.19,
|
| 2341 |
+
"learning_rate": 0.0007168888888888889,
|
| 2342 |
+
"loss": 2.3474,
|
| 2343 |
+
"step": 387
|
| 2344 |
+
},
|
| 2345 |
+
{
|
| 2346 |
+
"epoch": 0.19,
|
| 2347 |
+
"learning_rate": 0.0007164444444444445,
|
| 2348 |
+
"loss": 2.226,
|
| 2349 |
+
"step": 388
|
| 2350 |
+
},
|
| 2351 |
+
{
|
| 2352 |
+
"epoch": 0.19,
|
| 2353 |
+
"learning_rate": 0.0007160000000000001,
|
| 2354 |
+
"loss": 2.2839,
|
| 2355 |
+
"step": 389
|
| 2356 |
+
},
|
| 2357 |
+
{
|
| 2358 |
+
"epoch": 0.2,
|
| 2359 |
+
"learning_rate": 0.0007155555555555556,
|
| 2360 |
+
"loss": 2.2699,
|
| 2361 |
+
"step": 390
|
| 2362 |
+
},
|
| 2363 |
+
{
|
| 2364 |
+
"epoch": 0.2,
|
| 2365 |
+
"learning_rate": 0.0007151111111111111,
|
| 2366 |
+
"loss": 2.2928,
|
| 2367 |
+
"step": 391
|
| 2368 |
+
},
|
| 2369 |
+
{
|
| 2370 |
+
"epoch": 0.2,
|
| 2371 |
+
"learning_rate": 0.0007146666666666667,
|
| 2372 |
+
"loss": 2.2372,
|
| 2373 |
+
"step": 392
|
| 2374 |
+
},
|
| 2375 |
+
{
|
| 2376 |
+
"epoch": 0.2,
|
| 2377 |
+
"learning_rate": 0.0007142222222222222,
|
| 2378 |
+
"loss": 2.2812,
|
| 2379 |
+
"step": 393
|
| 2380 |
+
},
|
| 2381 |
+
{
|
| 2382 |
+
"epoch": 0.2,
|
| 2383 |
+
"learning_rate": 0.0007137777777777778,
|
| 2384 |
+
"loss": 2.265,
|
| 2385 |
+
"step": 394
|
| 2386 |
+
},
|
| 2387 |
+
{
|
| 2388 |
+
"epoch": 0.2,
|
| 2389 |
+
"learning_rate": 0.0007133333333333334,
|
| 2390 |
+
"loss": 2.2571,
|
| 2391 |
+
"step": 395
|
| 2392 |
+
},
|
| 2393 |
+
{
|
| 2394 |
+
"epoch": 0.2,
|
| 2395 |
+
"learning_rate": 0.0007128888888888889,
|
| 2396 |
+
"loss": 2.3748,
|
| 2397 |
+
"step": 396
|
| 2398 |
+
},
|
| 2399 |
+
{
|
| 2400 |
+
"epoch": 0.2,
|
| 2401 |
+
"learning_rate": 0.0007124444444444445,
|
| 2402 |
+
"loss": 2.2414,
|
| 2403 |
+
"step": 397
|
| 2404 |
+
},
|
| 2405 |
+
{
|
| 2406 |
+
"epoch": 0.2,
|
| 2407 |
+
"learning_rate": 0.0007120000000000001,
|
| 2408 |
+
"loss": 2.2718,
|
| 2409 |
+
"step": 398
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 0.2,
|
| 2413 |
+
"learning_rate": 0.0007115555555555557,
|
| 2414 |
+
"loss": 2.2649,
|
| 2415 |
+
"step": 399
|
| 2416 |
+
},
|
| 2417 |
+
{
|
| 2418 |
+
"epoch": 0.2,
|
| 2419 |
+
"learning_rate": 0.0007111111111111111,
|
| 2420 |
+
"loss": 2.2645,
|
| 2421 |
+
"step": 400
|
| 2422 |
+
},
|
| 2423 |
+
{
|
| 2424 |
+
"epoch": 0.2,
|
| 2425 |
+
"eval_gen_len": 1023.0,
|
| 2426 |
+
"eval_loss": 2.1945419311523438,
|
| 2427 |
+
"eval_rouge1": 12.1435,
|
| 2428 |
+
"eval_rouge2": 4.0018,
|
| 2429 |
+
"eval_rougeL": 7.5854,
|
| 2430 |
+
"eval_rougeLsum": 7.5909,
|
| 2431 |
+
"eval_runtime": 9033.7247,
|
| 2432 |
+
"eval_samples_per_second": 0.055,
|
| 2433 |
+
"eval_steps_per_second": 0.055,
|
| 2434 |
+
"step": 400
|
| 2435 |
}
|
| 2436 |
],
|
| 2437 |
"max_steps": 2000,
|
| 2438 |
"num_train_epochs": 9223372036854775807,
|
| 2439 |
+
"total_flos": 3.2061729480431616e+17,
|
| 2440 |
"trial_name": null,
|
| 2441 |
"trial_params": null
|
| 2442 |
}
|
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 sha256:6691fb6f696ed838034568580bf11105cd7d247925aaa8c39f26ca3e936ca47e
|
| 3 |
size 2368281769
|