Training in progress, step 2000
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 +1218 -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:d6f0b117cb3b5ca9f3104796b19da86da2f7d1e2c40756714f09621415d49b13
|
| 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:cdb640acc1ad0b52fbe5849e2fe9ec8023752b0a552442e328c1f4ad282eab9d
|
| 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 sha256:7f471bea3861187dceb51828819a95fc512e9746b6a62f37798a37a6d6a9d142
|
| 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:b35301bff95034222da17e187f0935a7b363754032e82a831e804e9485f6ec34
|
| 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": 1.
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
@@ -10923,11 +10923,1224 @@
|
|
| 10923 |
"eval_samples_per_second": 0.055,
|
| 10924 |
"eval_steps_per_second": 0.055,
|
| 10925 |
"step": 1800
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 10926 |
}
|
| 10927 |
],
|
| 10928 |
"max_steps": 2000,
|
| 10929 |
"num_train_epochs": 9223372036854775807,
|
| 10930 |
-
"total_flos": 1.
|
| 10931 |
"trial_name": null,
|
| 10932 |
"trial_params": null
|
| 10933 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 2.0949487686157227,
|
| 3 |
+
"best_model_checkpoint": "output/checkpoint-2000",
|
| 4 |
+
"epoch": 1.38650390625,
|
| 5 |
+
"global_step": 2000,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 10923 |
"eval_samples_per_second": 0.055,
|
| 10924 |
"eval_steps_per_second": 0.055,
|
| 10925 |
"step": 1800
|
| 10926 |
+
},
|
| 10927 |
+
{
|
| 10928 |
+
"epoch": 1.29,
|
| 10929 |
+
"learning_rate": 8.844444444444445e-05,
|
| 10930 |
+
"loss": 2.0691,
|
| 10931 |
+
"step": 1801
|
| 10932 |
+
},
|
| 10933 |
+
{
|
| 10934 |
+
"epoch": 1.29,
|
| 10935 |
+
"learning_rate": 8.800000000000001e-05,
|
| 10936 |
+
"loss": 2.0835,
|
| 10937 |
+
"step": 1802
|
| 10938 |
+
},
|
| 10939 |
+
{
|
| 10940 |
+
"epoch": 1.29,
|
| 10941 |
+
"learning_rate": 8.755555555555556e-05,
|
| 10942 |
+
"loss": 2.0686,
|
| 10943 |
+
"step": 1803
|
| 10944 |
+
},
|
| 10945 |
+
{
|
| 10946 |
+
"epoch": 1.29,
|
| 10947 |
+
"learning_rate": 8.711111111111112e-05,
|
| 10948 |
+
"loss": 2.2013,
|
| 10949 |
+
"step": 1804
|
| 10950 |
+
},
|
| 10951 |
+
{
|
| 10952 |
+
"epoch": 1.29,
|
| 10953 |
+
"learning_rate": 8.666666666666667e-05,
|
| 10954 |
+
"loss": 2.0697,
|
| 10955 |
+
"step": 1805
|
| 10956 |
+
},
|
| 10957 |
+
{
|
| 10958 |
+
"epoch": 1.29,
|
| 10959 |
+
"learning_rate": 8.622222222222222e-05,
|
| 10960 |
+
"loss": 2.15,
|
| 10961 |
+
"step": 1806
|
| 10962 |
+
},
|
| 10963 |
+
{
|
| 10964 |
+
"epoch": 1.29,
|
| 10965 |
+
"learning_rate": 8.577777777777777e-05,
|
| 10966 |
+
"loss": 2.079,
|
| 10967 |
+
"step": 1807
|
| 10968 |
+
},
|
| 10969 |
+
{
|
| 10970 |
+
"epoch": 1.29,
|
| 10971 |
+
"learning_rate": 8.533333333333334e-05,
|
| 10972 |
+
"loss": 2.1418,
|
| 10973 |
+
"step": 1808
|
| 10974 |
+
},
|
| 10975 |
+
{
|
| 10976 |
+
"epoch": 1.29,
|
| 10977 |
+
"learning_rate": 8.488888888888889e-05,
|
| 10978 |
+
"loss": 2.1518,
|
| 10979 |
+
"step": 1809
|
| 10980 |
+
},
|
| 10981 |
+
{
|
| 10982 |
+
"epoch": 1.29,
|
| 10983 |
+
"learning_rate": 8.444444444444444e-05,
|
| 10984 |
+
"loss": 2.12,
|
| 10985 |
+
"step": 1810
|
| 10986 |
+
},
|
| 10987 |
+
{
|
| 10988 |
+
"epoch": 1.29,
|
| 10989 |
+
"learning_rate": 8.4e-05,
|
| 10990 |
+
"loss": 2.1908,
|
| 10991 |
+
"step": 1811
|
| 10992 |
+
},
|
| 10993 |
+
{
|
| 10994 |
+
"epoch": 1.29,
|
| 10995 |
+
"learning_rate": 8.355555555555556e-05,
|
| 10996 |
+
"loss": 2.1359,
|
| 10997 |
+
"step": 1812
|
| 10998 |
+
},
|
| 10999 |
+
{
|
| 11000 |
+
"epoch": 1.29,
|
| 11001 |
+
"learning_rate": 8.311111111111111e-05,
|
| 11002 |
+
"loss": 2.1208,
|
| 11003 |
+
"step": 1813
|
| 11004 |
+
},
|
| 11005 |
+
{
|
| 11006 |
+
"epoch": 1.29,
|
| 11007 |
+
"learning_rate": 8.266666666666667e-05,
|
| 11008 |
+
"loss": 2.1014,
|
| 11009 |
+
"step": 1814
|
| 11010 |
+
},
|
| 11011 |
+
{
|
| 11012 |
+
"epoch": 1.29,
|
| 11013 |
+
"learning_rate": 8.222222222222222e-05,
|
| 11014 |
+
"loss": 2.1451,
|
| 11015 |
+
"step": 1815
|
| 11016 |
+
},
|
| 11017 |
+
{
|
| 11018 |
+
"epoch": 1.29,
|
| 11019 |
+
"learning_rate": 8.177777777777778e-05,
|
| 11020 |
+
"loss": 2.0757,
|
| 11021 |
+
"step": 1816
|
| 11022 |
+
},
|
| 11023 |
+
{
|
| 11024 |
+
"epoch": 1.3,
|
| 11025 |
+
"learning_rate": 8.133333333333334e-05,
|
| 11026 |
+
"loss": 2.0587,
|
| 11027 |
+
"step": 1817
|
| 11028 |
+
},
|
| 11029 |
+
{
|
| 11030 |
+
"epoch": 1.3,
|
| 11031 |
+
"learning_rate": 8.088888888888889e-05,
|
| 11032 |
+
"loss": 2.1008,
|
| 11033 |
+
"step": 1818
|
| 11034 |
+
},
|
| 11035 |
+
{
|
| 11036 |
+
"epoch": 1.3,
|
| 11037 |
+
"learning_rate": 8.044444444444444e-05,
|
| 11038 |
+
"loss": 2.0753,
|
| 11039 |
+
"step": 1819
|
| 11040 |
+
},
|
| 11041 |
+
{
|
| 11042 |
+
"epoch": 1.3,
|
| 11043 |
+
"learning_rate": 8e-05,
|
| 11044 |
+
"loss": 2.0912,
|
| 11045 |
+
"step": 1820
|
| 11046 |
+
},
|
| 11047 |
+
{
|
| 11048 |
+
"epoch": 1.3,
|
| 11049 |
+
"learning_rate": 7.955555555555556e-05,
|
| 11050 |
+
"loss": 2.0648,
|
| 11051 |
+
"step": 1821
|
| 11052 |
+
},
|
| 11053 |
+
{
|
| 11054 |
+
"epoch": 1.3,
|
| 11055 |
+
"learning_rate": 7.911111111111111e-05,
|
| 11056 |
+
"loss": 2.0948,
|
| 11057 |
+
"step": 1822
|
| 11058 |
+
},
|
| 11059 |
+
{
|
| 11060 |
+
"epoch": 1.3,
|
| 11061 |
+
"learning_rate": 7.866666666666666e-05,
|
| 11062 |
+
"loss": 2.1252,
|
| 11063 |
+
"step": 1823
|
| 11064 |
+
},
|
| 11065 |
+
{
|
| 11066 |
+
"epoch": 1.3,
|
| 11067 |
+
"learning_rate": 7.822222222222223e-05,
|
| 11068 |
+
"loss": 2.1548,
|
| 11069 |
+
"step": 1824
|
| 11070 |
+
},
|
| 11071 |
+
{
|
| 11072 |
+
"epoch": 1.3,
|
| 11073 |
+
"learning_rate": 7.777777777777778e-05,
|
| 11074 |
+
"loss": 2.1061,
|
| 11075 |
+
"step": 1825
|
| 11076 |
+
},
|
| 11077 |
+
{
|
| 11078 |
+
"epoch": 1.3,
|
| 11079 |
+
"learning_rate": 7.733333333333333e-05,
|
| 11080 |
+
"loss": 2.2017,
|
| 11081 |
+
"step": 1826
|
| 11082 |
+
},
|
| 11083 |
+
{
|
| 11084 |
+
"epoch": 1.3,
|
| 11085 |
+
"learning_rate": 7.688888888888889e-05,
|
| 11086 |
+
"loss": 2.1743,
|
| 11087 |
+
"step": 1827
|
| 11088 |
+
},
|
| 11089 |
+
{
|
| 11090 |
+
"epoch": 1.3,
|
| 11091 |
+
"learning_rate": 7.644444444444445e-05,
|
| 11092 |
+
"loss": 2.1126,
|
| 11093 |
+
"step": 1828
|
| 11094 |
+
},
|
| 11095 |
+
{
|
| 11096 |
+
"epoch": 1.3,
|
| 11097 |
+
"learning_rate": 7.6e-05,
|
| 11098 |
+
"loss": 2.0536,
|
| 11099 |
+
"step": 1829
|
| 11100 |
+
},
|
| 11101 |
+
{
|
| 11102 |
+
"epoch": 1.3,
|
| 11103 |
+
"learning_rate": 7.555555555555556e-05,
|
| 11104 |
+
"loss": 2.0527,
|
| 11105 |
+
"step": 1830
|
| 11106 |
+
},
|
| 11107 |
+
{
|
| 11108 |
+
"epoch": 1.3,
|
| 11109 |
+
"learning_rate": 7.511111111111111e-05,
|
| 11110 |
+
"loss": 2.0798,
|
| 11111 |
+
"step": 1831
|
| 11112 |
+
},
|
| 11113 |
+
{
|
| 11114 |
+
"epoch": 1.3,
|
| 11115 |
+
"learning_rate": 7.466666666666667e-05,
|
| 11116 |
+
"loss": 2.1436,
|
| 11117 |
+
"step": 1832
|
| 11118 |
+
},
|
| 11119 |
+
{
|
| 11120 |
+
"epoch": 1.3,
|
| 11121 |
+
"learning_rate": 7.422222222222223e-05,
|
| 11122 |
+
"loss": 2.1289,
|
| 11123 |
+
"step": 1833
|
| 11124 |
+
},
|
| 11125 |
+
{
|
| 11126 |
+
"epoch": 1.3,
|
| 11127 |
+
"learning_rate": 7.377777777777778e-05,
|
| 11128 |
+
"loss": 2.1487,
|
| 11129 |
+
"step": 1834
|
| 11130 |
+
},
|
| 11131 |
+
{
|
| 11132 |
+
"epoch": 1.3,
|
| 11133 |
+
"learning_rate": 7.333333333333333e-05,
|
| 11134 |
+
"loss": 2.0779,
|
| 11135 |
+
"step": 1835
|
| 11136 |
+
},
|
| 11137 |
+
{
|
| 11138 |
+
"epoch": 1.3,
|
| 11139 |
+
"learning_rate": 7.28888888888889e-05,
|
| 11140 |
+
"loss": 2.1045,
|
| 11141 |
+
"step": 1836
|
| 11142 |
+
},
|
| 11143 |
+
{
|
| 11144 |
+
"epoch": 1.31,
|
| 11145 |
+
"learning_rate": 7.244444444444445e-05,
|
| 11146 |
+
"loss": 2.0659,
|
| 11147 |
+
"step": 1837
|
| 11148 |
+
},
|
| 11149 |
+
{
|
| 11150 |
+
"epoch": 1.31,
|
| 11151 |
+
"learning_rate": 7.2e-05,
|
| 11152 |
+
"loss": 2.0704,
|
| 11153 |
+
"step": 1838
|
| 11154 |
+
},
|
| 11155 |
+
{
|
| 11156 |
+
"epoch": 1.31,
|
| 11157 |
+
"learning_rate": 7.155555555555555e-05,
|
| 11158 |
+
"loss": 2.0901,
|
| 11159 |
+
"step": 1839
|
| 11160 |
+
},
|
| 11161 |
+
{
|
| 11162 |
+
"epoch": 1.31,
|
| 11163 |
+
"learning_rate": 7.111111111111112e-05,
|
| 11164 |
+
"loss": 2.104,
|
| 11165 |
+
"step": 1840
|
| 11166 |
+
},
|
| 11167 |
+
{
|
| 11168 |
+
"epoch": 1.31,
|
| 11169 |
+
"learning_rate": 7.066666666666667e-05,
|
| 11170 |
+
"loss": 2.1784,
|
| 11171 |
+
"step": 1841
|
| 11172 |
+
},
|
| 11173 |
+
{
|
| 11174 |
+
"epoch": 1.31,
|
| 11175 |
+
"learning_rate": 7.022222222222222e-05,
|
| 11176 |
+
"loss": 2.0993,
|
| 11177 |
+
"step": 1842
|
| 11178 |
+
},
|
| 11179 |
+
{
|
| 11180 |
+
"epoch": 1.31,
|
| 11181 |
+
"learning_rate": 6.977777777777779e-05,
|
| 11182 |
+
"loss": 2.1538,
|
| 11183 |
+
"step": 1843
|
| 11184 |
+
},
|
| 11185 |
+
{
|
| 11186 |
+
"epoch": 1.31,
|
| 11187 |
+
"learning_rate": 6.933333333333334e-05,
|
| 11188 |
+
"loss": 2.0939,
|
| 11189 |
+
"step": 1844
|
| 11190 |
+
},
|
| 11191 |
+
{
|
| 11192 |
+
"epoch": 1.31,
|
| 11193 |
+
"learning_rate": 6.88888888888889e-05,
|
| 11194 |
+
"loss": 2.2134,
|
| 11195 |
+
"step": 1845
|
| 11196 |
+
},
|
| 11197 |
+
{
|
| 11198 |
+
"epoch": 1.31,
|
| 11199 |
+
"learning_rate": 6.844444444444445e-05,
|
| 11200 |
+
"loss": 2.0872,
|
| 11201 |
+
"step": 1846
|
| 11202 |
+
},
|
| 11203 |
+
{
|
| 11204 |
+
"epoch": 1.31,
|
| 11205 |
+
"learning_rate": 6.800000000000001e-05,
|
| 11206 |
+
"loss": 2.0762,
|
| 11207 |
+
"step": 1847
|
| 11208 |
+
},
|
| 11209 |
+
{
|
| 11210 |
+
"epoch": 1.31,
|
| 11211 |
+
"learning_rate": 6.755555555555557e-05,
|
| 11212 |
+
"loss": 2.1751,
|
| 11213 |
+
"step": 1848
|
| 11214 |
+
},
|
| 11215 |
+
{
|
| 11216 |
+
"epoch": 1.31,
|
| 11217 |
+
"learning_rate": 6.711111111111112e-05,
|
| 11218 |
+
"loss": 2.1107,
|
| 11219 |
+
"step": 1849
|
| 11220 |
+
},
|
| 11221 |
+
{
|
| 11222 |
+
"epoch": 1.31,
|
| 11223 |
+
"learning_rate": 6.666666666666667e-05,
|
| 11224 |
+
"loss": 2.1072,
|
| 11225 |
+
"step": 1850
|
| 11226 |
+
},
|
| 11227 |
+
{
|
| 11228 |
+
"epoch": 1.31,
|
| 11229 |
+
"learning_rate": 6.622222222222224e-05,
|
| 11230 |
+
"loss": 2.1557,
|
| 11231 |
+
"step": 1851
|
| 11232 |
+
},
|
| 11233 |
+
{
|
| 11234 |
+
"epoch": 1.31,
|
| 11235 |
+
"learning_rate": 6.577777777777779e-05,
|
| 11236 |
+
"loss": 2.1582,
|
| 11237 |
+
"step": 1852
|
| 11238 |
+
},
|
| 11239 |
+
{
|
| 11240 |
+
"epoch": 1.31,
|
| 11241 |
+
"learning_rate": 6.533333333333334e-05,
|
| 11242 |
+
"loss": 2.1279,
|
| 11243 |
+
"step": 1853
|
| 11244 |
+
},
|
| 11245 |
+
{
|
| 11246 |
+
"epoch": 1.31,
|
| 11247 |
+
"learning_rate": 6.488888888888889e-05,
|
| 11248 |
+
"loss": 2.1797,
|
| 11249 |
+
"step": 1854
|
| 11250 |
+
},
|
| 11251 |
+
{
|
| 11252 |
+
"epoch": 1.31,
|
| 11253 |
+
"learning_rate": 6.444444444444446e-05,
|
| 11254 |
+
"loss": 2.1951,
|
| 11255 |
+
"step": 1855
|
| 11256 |
+
},
|
| 11257 |
+
{
|
| 11258 |
+
"epoch": 1.31,
|
| 11259 |
+
"learning_rate": 6.400000000000001e-05,
|
| 11260 |
+
"loss": 2.018,
|
| 11261 |
+
"step": 1856
|
| 11262 |
+
},
|
| 11263 |
+
{
|
| 11264 |
+
"epoch": 1.32,
|
| 11265 |
+
"learning_rate": 6.355555555555556e-05,
|
| 11266 |
+
"loss": 2.1234,
|
| 11267 |
+
"step": 1857
|
| 11268 |
+
},
|
| 11269 |
+
{
|
| 11270 |
+
"epoch": 1.32,
|
| 11271 |
+
"learning_rate": 6.311111111111112e-05,
|
| 11272 |
+
"loss": 2.1022,
|
| 11273 |
+
"step": 1858
|
| 11274 |
+
},
|
| 11275 |
+
{
|
| 11276 |
+
"epoch": 1.32,
|
| 11277 |
+
"learning_rate": 6.266666666666667e-05,
|
| 11278 |
+
"loss": 2.0655,
|
| 11279 |
+
"step": 1859
|
| 11280 |
+
},
|
| 11281 |
+
{
|
| 11282 |
+
"epoch": 1.32,
|
| 11283 |
+
"learning_rate": 6.222222222222222e-05,
|
| 11284 |
+
"loss": 2.138,
|
| 11285 |
+
"step": 1860
|
| 11286 |
+
},
|
| 11287 |
+
{
|
| 11288 |
+
"epoch": 1.32,
|
| 11289 |
+
"learning_rate": 6.177777777777779e-05,
|
| 11290 |
+
"loss": 2.1443,
|
| 11291 |
+
"step": 1861
|
| 11292 |
+
},
|
| 11293 |
+
{
|
| 11294 |
+
"epoch": 1.32,
|
| 11295 |
+
"learning_rate": 6.133333333333334e-05,
|
| 11296 |
+
"loss": 2.1101,
|
| 11297 |
+
"step": 1862
|
| 11298 |
+
},
|
| 11299 |
+
{
|
| 11300 |
+
"epoch": 1.32,
|
| 11301 |
+
"learning_rate": 6.08888888888889e-05,
|
| 11302 |
+
"loss": 2.0674,
|
| 11303 |
+
"step": 1863
|
| 11304 |
+
},
|
| 11305 |
+
{
|
| 11306 |
+
"epoch": 1.32,
|
| 11307 |
+
"learning_rate": 6.044444444444445e-05,
|
| 11308 |
+
"loss": 2.1124,
|
| 11309 |
+
"step": 1864
|
| 11310 |
+
},
|
| 11311 |
+
{
|
| 11312 |
+
"epoch": 1.32,
|
| 11313 |
+
"learning_rate": 6e-05,
|
| 11314 |
+
"loss": 2.1006,
|
| 11315 |
+
"step": 1865
|
| 11316 |
+
},
|
| 11317 |
+
{
|
| 11318 |
+
"epoch": 1.32,
|
| 11319 |
+
"learning_rate": 5.9555555555555554e-05,
|
| 11320 |
+
"loss": 2.1313,
|
| 11321 |
+
"step": 1866
|
| 11322 |
+
},
|
| 11323 |
+
{
|
| 11324 |
+
"epoch": 1.32,
|
| 11325 |
+
"learning_rate": 5.911111111111112e-05,
|
| 11326 |
+
"loss": 2.0989,
|
| 11327 |
+
"step": 1867
|
| 11328 |
+
},
|
| 11329 |
+
{
|
| 11330 |
+
"epoch": 1.32,
|
| 11331 |
+
"learning_rate": 5.866666666666667e-05,
|
| 11332 |
+
"loss": 2.0372,
|
| 11333 |
+
"step": 1868
|
| 11334 |
+
},
|
| 11335 |
+
{
|
| 11336 |
+
"epoch": 1.32,
|
| 11337 |
+
"learning_rate": 5.8222222222222224e-05,
|
| 11338 |
+
"loss": 2.1849,
|
| 11339 |
+
"step": 1869
|
| 11340 |
+
},
|
| 11341 |
+
{
|
| 11342 |
+
"epoch": 1.32,
|
| 11343 |
+
"learning_rate": 5.7777777777777776e-05,
|
| 11344 |
+
"loss": 2.1394,
|
| 11345 |
+
"step": 1870
|
| 11346 |
+
},
|
| 11347 |
+
{
|
| 11348 |
+
"epoch": 1.32,
|
| 11349 |
+
"learning_rate": 5.7333333333333336e-05,
|
| 11350 |
+
"loss": 2.1078,
|
| 11351 |
+
"step": 1871
|
| 11352 |
+
},
|
| 11353 |
+
{
|
| 11354 |
+
"epoch": 1.32,
|
| 11355 |
+
"learning_rate": 5.6888888888888895e-05,
|
| 11356 |
+
"loss": 2.2069,
|
| 11357 |
+
"step": 1872
|
| 11358 |
+
},
|
| 11359 |
+
{
|
| 11360 |
+
"epoch": 1.32,
|
| 11361 |
+
"learning_rate": 5.644444444444445e-05,
|
| 11362 |
+
"loss": 2.0549,
|
| 11363 |
+
"step": 1873
|
| 11364 |
+
},
|
| 11365 |
+
{
|
| 11366 |
+
"epoch": 1.32,
|
| 11367 |
+
"learning_rate": 5.6000000000000006e-05,
|
| 11368 |
+
"loss": 2.0907,
|
| 11369 |
+
"step": 1874
|
| 11370 |
+
},
|
| 11371 |
+
{
|
| 11372 |
+
"epoch": 1.32,
|
| 11373 |
+
"learning_rate": 5.555555555555556e-05,
|
| 11374 |
+
"loss": 2.1449,
|
| 11375 |
+
"step": 1875
|
| 11376 |
+
},
|
| 11377 |
+
{
|
| 11378 |
+
"epoch": 1.32,
|
| 11379 |
+
"learning_rate": 5.511111111111111e-05,
|
| 11380 |
+
"loss": 2.1387,
|
| 11381 |
+
"step": 1876
|
| 11382 |
+
},
|
| 11383 |
+
{
|
| 11384 |
+
"epoch": 1.33,
|
| 11385 |
+
"learning_rate": 5.466666666666666e-05,
|
| 11386 |
+
"loss": 2.0901,
|
| 11387 |
+
"step": 1877
|
| 11388 |
+
},
|
| 11389 |
+
{
|
| 11390 |
+
"epoch": 1.33,
|
| 11391 |
+
"learning_rate": 5.422222222222223e-05,
|
| 11392 |
+
"loss": 2.0245,
|
| 11393 |
+
"step": 1878
|
| 11394 |
+
},
|
| 11395 |
+
{
|
| 11396 |
+
"epoch": 1.33,
|
| 11397 |
+
"learning_rate": 5.377777777777778e-05,
|
| 11398 |
+
"loss": 2.1397,
|
| 11399 |
+
"step": 1879
|
| 11400 |
+
},
|
| 11401 |
+
{
|
| 11402 |
+
"epoch": 1.33,
|
| 11403 |
+
"learning_rate": 5.333333333333333e-05,
|
| 11404 |
+
"loss": 2.1229,
|
| 11405 |
+
"step": 1880
|
| 11406 |
+
},
|
| 11407 |
+
{
|
| 11408 |
+
"epoch": 1.33,
|
| 11409 |
+
"learning_rate": 5.2888888888888885e-05,
|
| 11410 |
+
"loss": 2.1114,
|
| 11411 |
+
"step": 1881
|
| 11412 |
+
},
|
| 11413 |
+
{
|
| 11414 |
+
"epoch": 1.33,
|
| 11415 |
+
"learning_rate": 5.244444444444445e-05,
|
| 11416 |
+
"loss": 2.1436,
|
| 11417 |
+
"step": 1882
|
| 11418 |
+
},
|
| 11419 |
+
{
|
| 11420 |
+
"epoch": 1.33,
|
| 11421 |
+
"learning_rate": 5.2000000000000004e-05,
|
| 11422 |
+
"loss": 2.0848,
|
| 11423 |
+
"step": 1883
|
| 11424 |
+
},
|
| 11425 |
+
{
|
| 11426 |
+
"epoch": 1.33,
|
| 11427 |
+
"learning_rate": 5.1555555555555556e-05,
|
| 11428 |
+
"loss": 2.1456,
|
| 11429 |
+
"step": 1884
|
| 11430 |
+
},
|
| 11431 |
+
{
|
| 11432 |
+
"epoch": 1.33,
|
| 11433 |
+
"learning_rate": 5.111111111111111e-05,
|
| 11434 |
+
"loss": 2.0572,
|
| 11435 |
+
"step": 1885
|
| 11436 |
+
},
|
| 11437 |
+
{
|
| 11438 |
+
"epoch": 1.33,
|
| 11439 |
+
"learning_rate": 5.0666666666666674e-05,
|
| 11440 |
+
"loss": 2.1019,
|
| 11441 |
+
"step": 1886
|
| 11442 |
+
},
|
| 11443 |
+
{
|
| 11444 |
+
"epoch": 1.33,
|
| 11445 |
+
"learning_rate": 5.0222222222222226e-05,
|
| 11446 |
+
"loss": 2.0934,
|
| 11447 |
+
"step": 1887
|
| 11448 |
+
},
|
| 11449 |
+
{
|
| 11450 |
+
"epoch": 1.33,
|
| 11451 |
+
"learning_rate": 4.977777777777778e-05,
|
| 11452 |
+
"loss": 2.1123,
|
| 11453 |
+
"step": 1888
|
| 11454 |
+
},
|
| 11455 |
+
{
|
| 11456 |
+
"epoch": 1.33,
|
| 11457 |
+
"learning_rate": 4.933333333333334e-05,
|
| 11458 |
+
"loss": 2.1475,
|
| 11459 |
+
"step": 1889
|
| 11460 |
+
},
|
| 11461 |
+
{
|
| 11462 |
+
"epoch": 1.33,
|
| 11463 |
+
"learning_rate": 4.888888888888889e-05,
|
| 11464 |
+
"loss": 2.14,
|
| 11465 |
+
"step": 1890
|
| 11466 |
+
},
|
| 11467 |
+
{
|
| 11468 |
+
"epoch": 1.33,
|
| 11469 |
+
"learning_rate": 4.844444444444445e-05,
|
| 11470 |
+
"loss": 2.1103,
|
| 11471 |
+
"step": 1891
|
| 11472 |
+
},
|
| 11473 |
+
{
|
| 11474 |
+
"epoch": 1.33,
|
| 11475 |
+
"learning_rate": 4.8e-05,
|
| 11476 |
+
"loss": 2.0636,
|
| 11477 |
+
"step": 1892
|
| 11478 |
+
},
|
| 11479 |
+
{
|
| 11480 |
+
"epoch": 1.33,
|
| 11481 |
+
"learning_rate": 4.755555555555556e-05,
|
| 11482 |
+
"loss": 2.0378,
|
| 11483 |
+
"step": 1893
|
| 11484 |
+
},
|
| 11485 |
+
{
|
| 11486 |
+
"epoch": 1.33,
|
| 11487 |
+
"learning_rate": 4.711111111111111e-05,
|
| 11488 |
+
"loss": 2.1109,
|
| 11489 |
+
"step": 1894
|
| 11490 |
+
},
|
| 11491 |
+
{
|
| 11492 |
+
"epoch": 1.33,
|
| 11493 |
+
"learning_rate": 4.666666666666667e-05,
|
| 11494 |
+
"loss": 2.0696,
|
| 11495 |
+
"step": 1895
|
| 11496 |
+
},
|
| 11497 |
+
{
|
| 11498 |
+
"epoch": 1.33,
|
| 11499 |
+
"learning_rate": 4.6222222222222224e-05,
|
| 11500 |
+
"loss": 2.117,
|
| 11501 |
+
"step": 1896
|
| 11502 |
+
},
|
| 11503 |
+
{
|
| 11504 |
+
"epoch": 1.34,
|
| 11505 |
+
"learning_rate": 4.577777777777778e-05,
|
| 11506 |
+
"loss": 2.0447,
|
| 11507 |
+
"step": 1897
|
| 11508 |
+
},
|
| 11509 |
+
{
|
| 11510 |
+
"epoch": 1.34,
|
| 11511 |
+
"learning_rate": 4.5333333333333335e-05,
|
| 11512 |
+
"loss": 2.1653,
|
| 11513 |
+
"step": 1898
|
| 11514 |
+
},
|
| 11515 |
+
{
|
| 11516 |
+
"epoch": 1.34,
|
| 11517 |
+
"learning_rate": 4.4888888888888894e-05,
|
| 11518 |
+
"loss": 2.0236,
|
| 11519 |
+
"step": 1899
|
| 11520 |
+
},
|
| 11521 |
+
{
|
| 11522 |
+
"epoch": 1.34,
|
| 11523 |
+
"learning_rate": 4.4444444444444447e-05,
|
| 11524 |
+
"loss": 2.1244,
|
| 11525 |
+
"step": 1900
|
| 11526 |
+
},
|
| 11527 |
+
{
|
| 11528 |
+
"epoch": 1.34,
|
| 11529 |
+
"learning_rate": 4.4000000000000006e-05,
|
| 11530 |
+
"loss": 2.1211,
|
| 11531 |
+
"step": 1901
|
| 11532 |
+
},
|
| 11533 |
+
{
|
| 11534 |
+
"epoch": 1.34,
|
| 11535 |
+
"learning_rate": 4.355555555555556e-05,
|
| 11536 |
+
"loss": 2.2072,
|
| 11537 |
+
"step": 1902
|
| 11538 |
+
},
|
| 11539 |
+
{
|
| 11540 |
+
"epoch": 1.34,
|
| 11541 |
+
"learning_rate": 4.311111111111111e-05,
|
| 11542 |
+
"loss": 2.1041,
|
| 11543 |
+
"step": 1903
|
| 11544 |
+
},
|
| 11545 |
+
{
|
| 11546 |
+
"epoch": 1.34,
|
| 11547 |
+
"learning_rate": 4.266666666666667e-05,
|
| 11548 |
+
"loss": 2.071,
|
| 11549 |
+
"step": 1904
|
| 11550 |
+
},
|
| 11551 |
+
{
|
| 11552 |
+
"epoch": 1.34,
|
| 11553 |
+
"learning_rate": 4.222222222222222e-05,
|
| 11554 |
+
"loss": 2.0229,
|
| 11555 |
+
"step": 1905
|
| 11556 |
+
},
|
| 11557 |
+
{
|
| 11558 |
+
"epoch": 1.34,
|
| 11559 |
+
"learning_rate": 4.177777777777778e-05,
|
| 11560 |
+
"loss": 2.0974,
|
| 11561 |
+
"step": 1906
|
| 11562 |
+
},
|
| 11563 |
+
{
|
| 11564 |
+
"epoch": 1.34,
|
| 11565 |
+
"learning_rate": 4.133333333333333e-05,
|
| 11566 |
+
"loss": 2.1007,
|
| 11567 |
+
"step": 1907
|
| 11568 |
+
},
|
| 11569 |
+
{
|
| 11570 |
+
"epoch": 1.34,
|
| 11571 |
+
"learning_rate": 4.088888888888889e-05,
|
| 11572 |
+
"loss": 2.1028,
|
| 11573 |
+
"step": 1908
|
| 11574 |
+
},
|
| 11575 |
+
{
|
| 11576 |
+
"epoch": 1.34,
|
| 11577 |
+
"learning_rate": 4.0444444444444444e-05,
|
| 11578 |
+
"loss": 2.097,
|
| 11579 |
+
"step": 1909
|
| 11580 |
+
},
|
| 11581 |
+
{
|
| 11582 |
+
"epoch": 1.34,
|
| 11583 |
+
"learning_rate": 4e-05,
|
| 11584 |
+
"loss": 2.1031,
|
| 11585 |
+
"step": 1910
|
| 11586 |
+
},
|
| 11587 |
+
{
|
| 11588 |
+
"epoch": 1.34,
|
| 11589 |
+
"learning_rate": 3.9555555555555556e-05,
|
| 11590 |
+
"loss": 2.0905,
|
| 11591 |
+
"step": 1911
|
| 11592 |
+
},
|
| 11593 |
+
{
|
| 11594 |
+
"epoch": 1.34,
|
| 11595 |
+
"learning_rate": 3.9111111111111115e-05,
|
| 11596 |
+
"loss": 2.1131,
|
| 11597 |
+
"step": 1912
|
| 11598 |
+
},
|
| 11599 |
+
{
|
| 11600 |
+
"epoch": 1.34,
|
| 11601 |
+
"learning_rate": 3.866666666666667e-05,
|
| 11602 |
+
"loss": 2.1349,
|
| 11603 |
+
"step": 1913
|
| 11604 |
+
},
|
| 11605 |
+
{
|
| 11606 |
+
"epoch": 1.34,
|
| 11607 |
+
"learning_rate": 3.8222222222222226e-05,
|
| 11608 |
+
"loss": 2.1394,
|
| 11609 |
+
"step": 1914
|
| 11610 |
+
},
|
| 11611 |
+
{
|
| 11612 |
+
"epoch": 1.34,
|
| 11613 |
+
"learning_rate": 3.777777777777778e-05,
|
| 11614 |
+
"loss": 2.0631,
|
| 11615 |
+
"step": 1915
|
| 11616 |
+
},
|
| 11617 |
+
{
|
| 11618 |
+
"epoch": 1.34,
|
| 11619 |
+
"learning_rate": 3.733333333333334e-05,
|
| 11620 |
+
"loss": 2.1042,
|
| 11621 |
+
"step": 1916
|
| 11622 |
+
},
|
| 11623 |
+
{
|
| 11624 |
+
"epoch": 1.35,
|
| 11625 |
+
"learning_rate": 3.688888888888889e-05,
|
| 11626 |
+
"loss": 2.1382,
|
| 11627 |
+
"step": 1917
|
| 11628 |
+
},
|
| 11629 |
+
{
|
| 11630 |
+
"epoch": 1.35,
|
| 11631 |
+
"learning_rate": 3.644444444444445e-05,
|
| 11632 |
+
"loss": 2.1726,
|
| 11633 |
+
"step": 1918
|
| 11634 |
+
},
|
| 11635 |
+
{
|
| 11636 |
+
"epoch": 1.35,
|
| 11637 |
+
"learning_rate": 3.6e-05,
|
| 11638 |
+
"loss": 2.1671,
|
| 11639 |
+
"step": 1919
|
| 11640 |
+
},
|
| 11641 |
+
{
|
| 11642 |
+
"epoch": 1.35,
|
| 11643 |
+
"learning_rate": 3.555555555555556e-05,
|
| 11644 |
+
"loss": 2.0649,
|
| 11645 |
+
"step": 1920
|
| 11646 |
+
},
|
| 11647 |
+
{
|
| 11648 |
+
"epoch": 1.35,
|
| 11649 |
+
"learning_rate": 3.511111111111111e-05,
|
| 11650 |
+
"loss": 2.0917,
|
| 11651 |
+
"step": 1921
|
| 11652 |
+
},
|
| 11653 |
+
{
|
| 11654 |
+
"epoch": 1.35,
|
| 11655 |
+
"learning_rate": 3.466666666666667e-05,
|
| 11656 |
+
"loss": 2.0466,
|
| 11657 |
+
"step": 1922
|
| 11658 |
+
},
|
| 11659 |
+
{
|
| 11660 |
+
"epoch": 1.35,
|
| 11661 |
+
"learning_rate": 3.4222222222222224e-05,
|
| 11662 |
+
"loss": 2.0903,
|
| 11663 |
+
"step": 1923
|
| 11664 |
+
},
|
| 11665 |
+
{
|
| 11666 |
+
"epoch": 1.35,
|
| 11667 |
+
"learning_rate": 3.377777777777778e-05,
|
| 11668 |
+
"loss": 2.0701,
|
| 11669 |
+
"step": 1924
|
| 11670 |
+
},
|
| 11671 |
+
{
|
| 11672 |
+
"epoch": 1.35,
|
| 11673 |
+
"learning_rate": 3.3333333333333335e-05,
|
| 11674 |
+
"loss": 2.0987,
|
| 11675 |
+
"step": 1925
|
| 11676 |
+
},
|
| 11677 |
+
{
|
| 11678 |
+
"epoch": 1.35,
|
| 11679 |
+
"learning_rate": 3.2888888888888894e-05,
|
| 11680 |
+
"loss": 2.1408,
|
| 11681 |
+
"step": 1926
|
| 11682 |
+
},
|
| 11683 |
+
{
|
| 11684 |
+
"epoch": 1.35,
|
| 11685 |
+
"learning_rate": 3.2444444444444446e-05,
|
| 11686 |
+
"loss": 2.104,
|
| 11687 |
+
"step": 1927
|
| 11688 |
+
},
|
| 11689 |
+
{
|
| 11690 |
+
"epoch": 1.35,
|
| 11691 |
+
"learning_rate": 3.2000000000000005e-05,
|
| 11692 |
+
"loss": 2.088,
|
| 11693 |
+
"step": 1928
|
| 11694 |
+
},
|
| 11695 |
+
{
|
| 11696 |
+
"epoch": 1.35,
|
| 11697 |
+
"learning_rate": 3.155555555555556e-05,
|
| 11698 |
+
"loss": 2.0903,
|
| 11699 |
+
"step": 1929
|
| 11700 |
+
},
|
| 11701 |
+
{
|
| 11702 |
+
"epoch": 1.35,
|
| 11703 |
+
"learning_rate": 3.111111111111111e-05,
|
| 11704 |
+
"loss": 2.0852,
|
| 11705 |
+
"step": 1930
|
| 11706 |
+
},
|
| 11707 |
+
{
|
| 11708 |
+
"epoch": 1.35,
|
| 11709 |
+
"learning_rate": 3.066666666666667e-05,
|
| 11710 |
+
"loss": 2.0749,
|
| 11711 |
+
"step": 1931
|
| 11712 |
+
},
|
| 11713 |
+
{
|
| 11714 |
+
"epoch": 1.35,
|
| 11715 |
+
"learning_rate": 3.0222222222222225e-05,
|
| 11716 |
+
"loss": 2.0794,
|
| 11717 |
+
"step": 1932
|
| 11718 |
+
},
|
| 11719 |
+
{
|
| 11720 |
+
"epoch": 1.35,
|
| 11721 |
+
"learning_rate": 2.9777777777777777e-05,
|
| 11722 |
+
"loss": 2.0552,
|
| 11723 |
+
"step": 1933
|
| 11724 |
+
},
|
| 11725 |
+
{
|
| 11726 |
+
"epoch": 1.35,
|
| 11727 |
+
"learning_rate": 2.9333333333333336e-05,
|
| 11728 |
+
"loss": 2.0896,
|
| 11729 |
+
"step": 1934
|
| 11730 |
+
},
|
| 11731 |
+
{
|
| 11732 |
+
"epoch": 1.35,
|
| 11733 |
+
"learning_rate": 2.8888888888888888e-05,
|
| 11734 |
+
"loss": 2.1586,
|
| 11735 |
+
"step": 1935
|
| 11736 |
+
},
|
| 11737 |
+
{
|
| 11738 |
+
"epoch": 1.35,
|
| 11739 |
+
"learning_rate": 2.8444444444444447e-05,
|
| 11740 |
+
"loss": 2.1477,
|
| 11741 |
+
"step": 1936
|
| 11742 |
+
},
|
| 11743 |
+
{
|
| 11744 |
+
"epoch": 1.36,
|
| 11745 |
+
"learning_rate": 2.8000000000000003e-05,
|
| 11746 |
+
"loss": 2.0847,
|
| 11747 |
+
"step": 1937
|
| 11748 |
+
},
|
| 11749 |
+
{
|
| 11750 |
+
"epoch": 1.36,
|
| 11751 |
+
"learning_rate": 2.7555555555555555e-05,
|
| 11752 |
+
"loss": 2.0892,
|
| 11753 |
+
"step": 1938
|
| 11754 |
+
},
|
| 11755 |
+
{
|
| 11756 |
+
"epoch": 1.36,
|
| 11757 |
+
"learning_rate": 2.7111111111111114e-05,
|
| 11758 |
+
"loss": 2.1891,
|
| 11759 |
+
"step": 1939
|
| 11760 |
+
},
|
| 11761 |
+
{
|
| 11762 |
+
"epoch": 1.36,
|
| 11763 |
+
"learning_rate": 2.6666666666666667e-05,
|
| 11764 |
+
"loss": 2.0856,
|
| 11765 |
+
"step": 1940
|
| 11766 |
+
},
|
| 11767 |
+
{
|
| 11768 |
+
"epoch": 1.36,
|
| 11769 |
+
"learning_rate": 2.6222222222222226e-05,
|
| 11770 |
+
"loss": 2.1166,
|
| 11771 |
+
"step": 1941
|
| 11772 |
+
},
|
| 11773 |
+
{
|
| 11774 |
+
"epoch": 1.36,
|
| 11775 |
+
"learning_rate": 2.5777777777777778e-05,
|
| 11776 |
+
"loss": 2.15,
|
| 11777 |
+
"step": 1942
|
| 11778 |
+
},
|
| 11779 |
+
{
|
| 11780 |
+
"epoch": 1.36,
|
| 11781 |
+
"learning_rate": 2.5333333333333337e-05,
|
| 11782 |
+
"loss": 2.0967,
|
| 11783 |
+
"step": 1943
|
| 11784 |
+
},
|
| 11785 |
+
{
|
| 11786 |
+
"epoch": 1.36,
|
| 11787 |
+
"learning_rate": 2.488888888888889e-05,
|
| 11788 |
+
"loss": 2.1111,
|
| 11789 |
+
"step": 1944
|
| 11790 |
+
},
|
| 11791 |
+
{
|
| 11792 |
+
"epoch": 1.36,
|
| 11793 |
+
"learning_rate": 2.4444444444444445e-05,
|
| 11794 |
+
"loss": 2.097,
|
| 11795 |
+
"step": 1945
|
| 11796 |
+
},
|
| 11797 |
+
{
|
| 11798 |
+
"epoch": 1.36,
|
| 11799 |
+
"learning_rate": 2.4e-05,
|
| 11800 |
+
"loss": 2.1431,
|
| 11801 |
+
"step": 1946
|
| 11802 |
+
},
|
| 11803 |
+
{
|
| 11804 |
+
"epoch": 1.36,
|
| 11805 |
+
"learning_rate": 2.3555555555555556e-05,
|
| 11806 |
+
"loss": 2.1265,
|
| 11807 |
+
"step": 1947
|
| 11808 |
+
},
|
| 11809 |
+
{
|
| 11810 |
+
"epoch": 1.36,
|
| 11811 |
+
"learning_rate": 2.3111111111111112e-05,
|
| 11812 |
+
"loss": 2.2027,
|
| 11813 |
+
"step": 1948
|
| 11814 |
+
},
|
| 11815 |
+
{
|
| 11816 |
+
"epoch": 1.36,
|
| 11817 |
+
"learning_rate": 2.2666666666666668e-05,
|
| 11818 |
+
"loss": 2.0808,
|
| 11819 |
+
"step": 1949
|
| 11820 |
+
},
|
| 11821 |
+
{
|
| 11822 |
+
"epoch": 1.36,
|
| 11823 |
+
"learning_rate": 2.2222222222222223e-05,
|
| 11824 |
+
"loss": 2.0719,
|
| 11825 |
+
"step": 1950
|
| 11826 |
+
},
|
| 11827 |
+
{
|
| 11828 |
+
"epoch": 1.36,
|
| 11829 |
+
"learning_rate": 2.177777777777778e-05,
|
| 11830 |
+
"loss": 2.0903,
|
| 11831 |
+
"step": 1951
|
| 11832 |
+
},
|
| 11833 |
+
{
|
| 11834 |
+
"epoch": 1.36,
|
| 11835 |
+
"learning_rate": 2.1333333333333335e-05,
|
| 11836 |
+
"loss": 2.0455,
|
| 11837 |
+
"step": 1952
|
| 11838 |
+
},
|
| 11839 |
+
{
|
| 11840 |
+
"epoch": 1.36,
|
| 11841 |
+
"learning_rate": 2.088888888888889e-05,
|
| 11842 |
+
"loss": 2.1111,
|
| 11843 |
+
"step": 1953
|
| 11844 |
+
},
|
| 11845 |
+
{
|
| 11846 |
+
"epoch": 1.36,
|
| 11847 |
+
"learning_rate": 2.0444444444444446e-05,
|
| 11848 |
+
"loss": 2.1216,
|
| 11849 |
+
"step": 1954
|
| 11850 |
+
},
|
| 11851 |
+
{
|
| 11852 |
+
"epoch": 1.36,
|
| 11853 |
+
"learning_rate": 2e-05,
|
| 11854 |
+
"loss": 2.1088,
|
| 11855 |
+
"step": 1955
|
| 11856 |
+
},
|
| 11857 |
+
{
|
| 11858 |
+
"epoch": 1.36,
|
| 11859 |
+
"learning_rate": 1.9555555555555557e-05,
|
| 11860 |
+
"loss": 2.0789,
|
| 11861 |
+
"step": 1956
|
| 11862 |
+
},
|
| 11863 |
+
{
|
| 11864 |
+
"epoch": 1.37,
|
| 11865 |
+
"learning_rate": 1.9111111111111113e-05,
|
| 11866 |
+
"loss": 2.0954,
|
| 11867 |
+
"step": 1957
|
| 11868 |
+
},
|
| 11869 |
+
{
|
| 11870 |
+
"epoch": 1.37,
|
| 11871 |
+
"learning_rate": 1.866666666666667e-05,
|
| 11872 |
+
"loss": 2.0936,
|
| 11873 |
+
"step": 1958
|
| 11874 |
+
},
|
| 11875 |
+
{
|
| 11876 |
+
"epoch": 1.37,
|
| 11877 |
+
"learning_rate": 1.8222222222222224e-05,
|
| 11878 |
+
"loss": 2.1256,
|
| 11879 |
+
"step": 1959
|
| 11880 |
+
},
|
| 11881 |
+
{
|
| 11882 |
+
"epoch": 1.37,
|
| 11883 |
+
"learning_rate": 1.777777777777778e-05,
|
| 11884 |
+
"loss": 2.1112,
|
| 11885 |
+
"step": 1960
|
| 11886 |
+
},
|
| 11887 |
+
{
|
| 11888 |
+
"epoch": 1.37,
|
| 11889 |
+
"learning_rate": 1.7333333333333336e-05,
|
| 11890 |
+
"loss": 2.0945,
|
| 11891 |
+
"step": 1961
|
| 11892 |
+
},
|
| 11893 |
+
{
|
| 11894 |
+
"epoch": 1.37,
|
| 11895 |
+
"learning_rate": 1.688888888888889e-05,
|
| 11896 |
+
"loss": 2.1225,
|
| 11897 |
+
"step": 1962
|
| 11898 |
+
},
|
| 11899 |
+
{
|
| 11900 |
+
"epoch": 1.37,
|
| 11901 |
+
"learning_rate": 1.6444444444444447e-05,
|
| 11902 |
+
"loss": 2.0004,
|
| 11903 |
+
"step": 1963
|
| 11904 |
+
},
|
| 11905 |
+
{
|
| 11906 |
+
"epoch": 1.37,
|
| 11907 |
+
"learning_rate": 1.6000000000000003e-05,
|
| 11908 |
+
"loss": 2.1143,
|
| 11909 |
+
"step": 1964
|
| 11910 |
+
},
|
| 11911 |
+
{
|
| 11912 |
+
"epoch": 1.37,
|
| 11913 |
+
"learning_rate": 1.5555555555555555e-05,
|
| 11914 |
+
"loss": 2.0987,
|
| 11915 |
+
"step": 1965
|
| 11916 |
+
},
|
| 11917 |
+
{
|
| 11918 |
+
"epoch": 1.37,
|
| 11919 |
+
"learning_rate": 1.5111111111111112e-05,
|
| 11920 |
+
"loss": 2.0606,
|
| 11921 |
+
"step": 1966
|
| 11922 |
+
},
|
| 11923 |
+
{
|
| 11924 |
+
"epoch": 1.37,
|
| 11925 |
+
"learning_rate": 1.4666666666666668e-05,
|
| 11926 |
+
"loss": 2.0714,
|
| 11927 |
+
"step": 1967
|
| 11928 |
+
},
|
| 11929 |
+
{
|
| 11930 |
+
"epoch": 1.37,
|
| 11931 |
+
"learning_rate": 1.4222222222222224e-05,
|
| 11932 |
+
"loss": 2.0926,
|
| 11933 |
+
"step": 1968
|
| 11934 |
+
},
|
| 11935 |
+
{
|
| 11936 |
+
"epoch": 1.37,
|
| 11937 |
+
"learning_rate": 1.3777777777777778e-05,
|
| 11938 |
+
"loss": 2.139,
|
| 11939 |
+
"step": 1969
|
| 11940 |
+
},
|
| 11941 |
+
{
|
| 11942 |
+
"epoch": 1.37,
|
| 11943 |
+
"learning_rate": 1.3333333333333333e-05,
|
| 11944 |
+
"loss": 2.1304,
|
| 11945 |
+
"step": 1970
|
| 11946 |
+
},
|
| 11947 |
+
{
|
| 11948 |
+
"epoch": 1.37,
|
| 11949 |
+
"learning_rate": 1.2888888888888889e-05,
|
| 11950 |
+
"loss": 2.1202,
|
| 11951 |
+
"step": 1971
|
| 11952 |
+
},
|
| 11953 |
+
{
|
| 11954 |
+
"epoch": 1.37,
|
| 11955 |
+
"learning_rate": 1.2444444444444445e-05,
|
| 11956 |
+
"loss": 2.1232,
|
| 11957 |
+
"step": 1972
|
| 11958 |
+
},
|
| 11959 |
+
{
|
| 11960 |
+
"epoch": 1.37,
|
| 11961 |
+
"learning_rate": 1.2e-05,
|
| 11962 |
+
"loss": 2.1089,
|
| 11963 |
+
"step": 1973
|
| 11964 |
+
},
|
| 11965 |
+
{
|
| 11966 |
+
"epoch": 1.37,
|
| 11967 |
+
"learning_rate": 1.1555555555555556e-05,
|
| 11968 |
+
"loss": 2.1162,
|
| 11969 |
+
"step": 1974
|
| 11970 |
+
},
|
| 11971 |
+
{
|
| 11972 |
+
"epoch": 1.37,
|
| 11973 |
+
"learning_rate": 1.1111111111111112e-05,
|
| 11974 |
+
"loss": 2.0465,
|
| 11975 |
+
"step": 1975
|
| 11976 |
+
},
|
| 11977 |
+
{
|
| 11978 |
+
"epoch": 1.37,
|
| 11979 |
+
"learning_rate": 1.0666666666666667e-05,
|
| 11980 |
+
"loss": 2.1557,
|
| 11981 |
+
"step": 1976
|
| 11982 |
+
},
|
| 11983 |
+
{
|
| 11984 |
+
"epoch": 1.38,
|
| 11985 |
+
"learning_rate": 1.0222222222222223e-05,
|
| 11986 |
+
"loss": 2.1571,
|
| 11987 |
+
"step": 1977
|
| 11988 |
+
},
|
| 11989 |
+
{
|
| 11990 |
+
"epoch": 1.38,
|
| 11991 |
+
"learning_rate": 9.777777777777779e-06,
|
| 11992 |
+
"loss": 2.0609,
|
| 11993 |
+
"step": 1978
|
| 11994 |
+
},
|
| 11995 |
+
{
|
| 11996 |
+
"epoch": 1.38,
|
| 11997 |
+
"learning_rate": 9.333333333333334e-06,
|
| 11998 |
+
"loss": 2.1276,
|
| 11999 |
+
"step": 1979
|
| 12000 |
+
},
|
| 12001 |
+
{
|
| 12002 |
+
"epoch": 1.38,
|
| 12003 |
+
"learning_rate": 8.88888888888889e-06,
|
| 12004 |
+
"loss": 2.0715,
|
| 12005 |
+
"step": 1980
|
| 12006 |
+
},
|
| 12007 |
+
{
|
| 12008 |
+
"epoch": 1.38,
|
| 12009 |
+
"learning_rate": 8.444444444444446e-06,
|
| 12010 |
+
"loss": 2.0817,
|
| 12011 |
+
"step": 1981
|
| 12012 |
+
},
|
| 12013 |
+
{
|
| 12014 |
+
"epoch": 1.38,
|
| 12015 |
+
"learning_rate": 8.000000000000001e-06,
|
| 12016 |
+
"loss": 2.1797,
|
| 12017 |
+
"step": 1982
|
| 12018 |
+
},
|
| 12019 |
+
{
|
| 12020 |
+
"epoch": 1.38,
|
| 12021 |
+
"learning_rate": 7.555555555555556e-06,
|
| 12022 |
+
"loss": 2.1437,
|
| 12023 |
+
"step": 1983
|
| 12024 |
+
},
|
| 12025 |
+
{
|
| 12026 |
+
"epoch": 1.38,
|
| 12027 |
+
"learning_rate": 7.111111111111112e-06,
|
| 12028 |
+
"loss": 2.1116,
|
| 12029 |
+
"step": 1984
|
| 12030 |
+
},
|
| 12031 |
+
{
|
| 12032 |
+
"epoch": 1.38,
|
| 12033 |
+
"learning_rate": 6.666666666666667e-06,
|
| 12034 |
+
"loss": 2.0851,
|
| 12035 |
+
"step": 1985
|
| 12036 |
+
},
|
| 12037 |
+
{
|
| 12038 |
+
"epoch": 1.38,
|
| 12039 |
+
"learning_rate": 6.222222222222222e-06,
|
| 12040 |
+
"loss": 2.1341,
|
| 12041 |
+
"step": 1986
|
| 12042 |
+
},
|
| 12043 |
+
{
|
| 12044 |
+
"epoch": 1.38,
|
| 12045 |
+
"learning_rate": 5.777777777777778e-06,
|
| 12046 |
+
"loss": 2.1147,
|
| 12047 |
+
"step": 1987
|
| 12048 |
+
},
|
| 12049 |
+
{
|
| 12050 |
+
"epoch": 1.38,
|
| 12051 |
+
"learning_rate": 5.333333333333334e-06,
|
| 12052 |
+
"loss": 2.1872,
|
| 12053 |
+
"step": 1988
|
| 12054 |
+
},
|
| 12055 |
+
{
|
| 12056 |
+
"epoch": 1.38,
|
| 12057 |
+
"learning_rate": 4.888888888888889e-06,
|
| 12058 |
+
"loss": 2.1453,
|
| 12059 |
+
"step": 1989
|
| 12060 |
+
},
|
| 12061 |
+
{
|
| 12062 |
+
"epoch": 1.38,
|
| 12063 |
+
"learning_rate": 4.444444444444445e-06,
|
| 12064 |
+
"loss": 2.1466,
|
| 12065 |
+
"step": 1990
|
| 12066 |
+
},
|
| 12067 |
+
{
|
| 12068 |
+
"epoch": 1.38,
|
| 12069 |
+
"learning_rate": 4.000000000000001e-06,
|
| 12070 |
+
"loss": 2.213,
|
| 12071 |
+
"step": 1991
|
| 12072 |
+
},
|
| 12073 |
+
{
|
| 12074 |
+
"epoch": 1.38,
|
| 12075 |
+
"learning_rate": 3.555555555555556e-06,
|
| 12076 |
+
"loss": 2.0331,
|
| 12077 |
+
"step": 1992
|
| 12078 |
+
},
|
| 12079 |
+
{
|
| 12080 |
+
"epoch": 1.38,
|
| 12081 |
+
"learning_rate": 3.111111111111111e-06,
|
| 12082 |
+
"loss": 2.0973,
|
| 12083 |
+
"step": 1993
|
| 12084 |
+
},
|
| 12085 |
+
{
|
| 12086 |
+
"epoch": 1.38,
|
| 12087 |
+
"learning_rate": 2.666666666666667e-06,
|
| 12088 |
+
"loss": 2.0288,
|
| 12089 |
+
"step": 1994
|
| 12090 |
+
},
|
| 12091 |
+
{
|
| 12092 |
+
"epoch": 1.38,
|
| 12093 |
+
"learning_rate": 2.2222222222222225e-06,
|
| 12094 |
+
"loss": 2.1309,
|
| 12095 |
+
"step": 1995
|
| 12096 |
+
},
|
| 12097 |
+
{
|
| 12098 |
+
"epoch": 1.38,
|
| 12099 |
+
"learning_rate": 1.777777777777778e-06,
|
| 12100 |
+
"loss": 2.1932,
|
| 12101 |
+
"step": 1996
|
| 12102 |
+
},
|
| 12103 |
+
{
|
| 12104 |
+
"epoch": 1.39,
|
| 12105 |
+
"learning_rate": 1.3333333333333334e-06,
|
| 12106 |
+
"loss": 2.037,
|
| 12107 |
+
"step": 1997
|
| 12108 |
+
},
|
| 12109 |
+
{
|
| 12110 |
+
"epoch": 1.39,
|
| 12111 |
+
"learning_rate": 8.88888888888889e-07,
|
| 12112 |
+
"loss": 2.1351,
|
| 12113 |
+
"step": 1998
|
| 12114 |
+
},
|
| 12115 |
+
{
|
| 12116 |
+
"epoch": 1.39,
|
| 12117 |
+
"learning_rate": 4.444444444444445e-07,
|
| 12118 |
+
"loss": 2.1637,
|
| 12119 |
+
"step": 1999
|
| 12120 |
+
},
|
| 12121 |
+
{
|
| 12122 |
+
"epoch": 1.39,
|
| 12123 |
+
"learning_rate": 0.0,
|
| 12124 |
+
"loss": 2.1263,
|
| 12125 |
+
"step": 2000
|
| 12126 |
+
},
|
| 12127 |
+
{
|
| 12128 |
+
"epoch": 1.39,
|
| 12129 |
+
"eval_gen_len": 1023.0,
|
| 12130 |
+
"eval_loss": 2.0949487686157227,
|
| 12131 |
+
"eval_rouge1": 14.0237,
|
| 12132 |
+
"eval_rouge2": 4.8219,
|
| 12133 |
+
"eval_rougeL": 8.8145,
|
| 12134 |
+
"eval_rougeLsum": 8.793,
|
| 12135 |
+
"eval_runtime": 9159.2647,
|
| 12136 |
+
"eval_samples_per_second": 0.055,
|
| 12137 |
+
"eval_steps_per_second": 0.055,
|
| 12138 |
+
"step": 2000
|
| 12139 |
}
|
| 12140 |
],
|
| 12141 |
"max_steps": 2000,
|
| 12142 |
"num_train_epochs": 9223372036854775807,
|
| 12143 |
+
"total_flos": 1.608210613960658e+18,
|
| 12144 |
"trial_name": null,
|
| 12145 |
"trial_params": null
|
| 12146 |
}
|
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:cdb640acc1ad0b52fbe5849e2fe9ec8023752b0a552442e328c1f4ad282eab9d
|
| 3 |
size 2368281769
|