Training in progress, step 2200
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 sha256:844efd6c2009c5de04ac1262a4139a388ae75d1574b240173ede965955b6459a
|
| 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:ce38e42231eaec2ee57adbd3a558f89d6a9d15cb466c85343dd8a5eee549c297
|
| 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 sha256:501fe46ba428506dc04a6c17a5f3ac42a3aaaeb0ca785df8c20d998cfe0545c8
|
| 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:d32a4c0417862ee076b39c4cf863a820cfe35de7e68441945ff56dcac7983a7b
|
| 3 |
size 627
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 1.
|
| 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,
|
|
@@ -12086,11 +12086,1219 @@
|
|
| 12086 |
"eval_samples_per_second": 6.07,
|
| 12087 |
"eval_steps_per_second": 6.07,
|
| 12088 |
"step": 2000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12089 |
}
|
| 12090 |
],
|
| 12091 |
"max_steps": 4386,
|
| 12092 |
"num_train_epochs": 3,
|
| 12093 |
-
"total_flos": 2.
|
| 12094 |
"trial_name": null,
|
| 12095 |
"trial_params": null
|
| 12096 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 1.9827316999435425,
|
| 3 |
+
"best_model_checkpoint": "output/checkpoint-2200",
|
| 4 |
+
"epoch": 1.5041047745712288,
|
| 5 |
+
"global_step": 2200,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 12086 |
"eval_samples_per_second": 6.07,
|
| 12087 |
"eval_steps_per_second": 6.07,
|
| 12088 |
"step": 2000
|
| 12089 |
+
},
|
| 12090 |
+
{
|
| 12091 |
+
"epoch": 1.37,
|
| 12092 |
+
"learning_rate": 0.0003418537983755375,
|
| 12093 |
+
"loss": 2.0172,
|
| 12094 |
+
"step": 2001
|
| 12095 |
+
},
|
| 12096 |
+
{
|
| 12097 |
+
"epoch": 1.37,
|
| 12098 |
+
"learning_rate": 0.00034171046344959387,
|
| 12099 |
+
"loss": 2.0405,
|
| 12100 |
+
"step": 2002
|
| 12101 |
+
},
|
| 12102 |
+
{
|
| 12103 |
+
"epoch": 1.37,
|
| 12104 |
+
"learning_rate": 0.00034156712852365025,
|
| 12105 |
+
"loss": 1.9842,
|
| 12106 |
+
"step": 2003
|
| 12107 |
+
},
|
| 12108 |
+
{
|
| 12109 |
+
"epoch": 1.37,
|
| 12110 |
+
"learning_rate": 0.00034142379359770663,
|
| 12111 |
+
"loss": 2.0294,
|
| 12112 |
+
"step": 2004
|
| 12113 |
+
},
|
| 12114 |
+
{
|
| 12115 |
+
"epoch": 1.37,
|
| 12116 |
+
"learning_rate": 0.000341280458671763,
|
| 12117 |
+
"loss": 1.9872,
|
| 12118 |
+
"step": 2005
|
| 12119 |
+
},
|
| 12120 |
+
{
|
| 12121 |
+
"epoch": 1.37,
|
| 12122 |
+
"learning_rate": 0.00034113712374581935,
|
| 12123 |
+
"loss": 1.9517,
|
| 12124 |
+
"step": 2006
|
| 12125 |
+
},
|
| 12126 |
+
{
|
| 12127 |
+
"epoch": 1.37,
|
| 12128 |
+
"learning_rate": 0.00034099378881987573,
|
| 12129 |
+
"loss": 2.0903,
|
| 12130 |
+
"step": 2007
|
| 12131 |
+
},
|
| 12132 |
+
{
|
| 12133 |
+
"epoch": 1.37,
|
| 12134 |
+
"learning_rate": 0.0003408504538939321,
|
| 12135 |
+
"loss": 2.0047,
|
| 12136 |
+
"step": 2008
|
| 12137 |
+
},
|
| 12138 |
+
{
|
| 12139 |
+
"epoch": 1.37,
|
| 12140 |
+
"learning_rate": 0.0003407071189679885,
|
| 12141 |
+
"loss": 1.9116,
|
| 12142 |
+
"step": 2009
|
| 12143 |
+
},
|
| 12144 |
+
{
|
| 12145 |
+
"epoch": 1.37,
|
| 12146 |
+
"learning_rate": 0.0003405637840420449,
|
| 12147 |
+
"loss": 1.9933,
|
| 12148 |
+
"step": 2010
|
| 12149 |
+
},
|
| 12150 |
+
{
|
| 12151 |
+
"epoch": 1.37,
|
| 12152 |
+
"learning_rate": 0.00034042044911610127,
|
| 12153 |
+
"loss": 2.0691,
|
| 12154 |
+
"step": 2011
|
| 12155 |
+
},
|
| 12156 |
+
{
|
| 12157 |
+
"epoch": 1.38,
|
| 12158 |
+
"learning_rate": 0.00034027711419015765,
|
| 12159 |
+
"loss": 2.0249,
|
| 12160 |
+
"step": 2012
|
| 12161 |
+
},
|
| 12162 |
+
{
|
| 12163 |
+
"epoch": 1.38,
|
| 12164 |
+
"learning_rate": 0.000340133779264214,
|
| 12165 |
+
"loss": 1.9914,
|
| 12166 |
+
"step": 2013
|
| 12167 |
+
},
|
| 12168 |
+
{
|
| 12169 |
+
"epoch": 1.38,
|
| 12170 |
+
"learning_rate": 0.0003399904443382704,
|
| 12171 |
+
"loss": 2.0592,
|
| 12172 |
+
"step": 2014
|
| 12173 |
+
},
|
| 12174 |
+
{
|
| 12175 |
+
"epoch": 1.38,
|
| 12176 |
+
"learning_rate": 0.00033984710941232675,
|
| 12177 |
+
"loss": 2.0658,
|
| 12178 |
+
"step": 2015
|
| 12179 |
+
},
|
| 12180 |
+
{
|
| 12181 |
+
"epoch": 1.38,
|
| 12182 |
+
"learning_rate": 0.0003397037744863832,
|
| 12183 |
+
"loss": 1.9771,
|
| 12184 |
+
"step": 2016
|
| 12185 |
+
},
|
| 12186 |
+
{
|
| 12187 |
+
"epoch": 1.38,
|
| 12188 |
+
"learning_rate": 0.0003395604395604395,
|
| 12189 |
+
"loss": 1.9868,
|
| 12190 |
+
"step": 2017
|
| 12191 |
+
},
|
| 12192 |
+
{
|
| 12193 |
+
"epoch": 1.38,
|
| 12194 |
+
"learning_rate": 0.00033941710463449595,
|
| 12195 |
+
"loss": 1.943,
|
| 12196 |
+
"step": 2018
|
| 12197 |
+
},
|
| 12198 |
+
{
|
| 12199 |
+
"epoch": 1.38,
|
| 12200 |
+
"learning_rate": 0.0003392737697085523,
|
| 12201 |
+
"loss": 2.0233,
|
| 12202 |
+
"step": 2019
|
| 12203 |
+
},
|
| 12204 |
+
{
|
| 12205 |
+
"epoch": 1.38,
|
| 12206 |
+
"learning_rate": 0.00033913043478260867,
|
| 12207 |
+
"loss": 1.9217,
|
| 12208 |
+
"step": 2020
|
| 12209 |
+
},
|
| 12210 |
+
{
|
| 12211 |
+
"epoch": 1.38,
|
| 12212 |
+
"learning_rate": 0.00033898709985666505,
|
| 12213 |
+
"loss": 2.05,
|
| 12214 |
+
"step": 2021
|
| 12215 |
+
},
|
| 12216 |
+
{
|
| 12217 |
+
"epoch": 1.38,
|
| 12218 |
+
"learning_rate": 0.00033884376493072143,
|
| 12219 |
+
"loss": 1.9538,
|
| 12220 |
+
"step": 2022
|
| 12221 |
+
},
|
| 12222 |
+
{
|
| 12223 |
+
"epoch": 1.38,
|
| 12224 |
+
"learning_rate": 0.0003387004300047778,
|
| 12225 |
+
"loss": 1.9666,
|
| 12226 |
+
"step": 2023
|
| 12227 |
+
},
|
| 12228 |
+
{
|
| 12229 |
+
"epoch": 1.38,
|
| 12230 |
+
"learning_rate": 0.00033855709507883415,
|
| 12231 |
+
"loss": 2.0349,
|
| 12232 |
+
"step": 2024
|
| 12233 |
+
},
|
| 12234 |
+
{
|
| 12235 |
+
"epoch": 1.38,
|
| 12236 |
+
"learning_rate": 0.0003384137601528906,
|
| 12237 |
+
"loss": 2.0789,
|
| 12238 |
+
"step": 2025
|
| 12239 |
+
},
|
| 12240 |
+
{
|
| 12241 |
+
"epoch": 1.39,
|
| 12242 |
+
"learning_rate": 0.0003382704252269469,
|
| 12243 |
+
"loss": 2.0193,
|
| 12244 |
+
"step": 2026
|
| 12245 |
+
},
|
| 12246 |
+
{
|
| 12247 |
+
"epoch": 1.39,
|
| 12248 |
+
"learning_rate": 0.0003381270903010033,
|
| 12249 |
+
"loss": 2.0933,
|
| 12250 |
+
"step": 2027
|
| 12251 |
+
},
|
| 12252 |
+
{
|
| 12253 |
+
"epoch": 1.39,
|
| 12254 |
+
"learning_rate": 0.0003379837553750597,
|
| 12255 |
+
"loss": 2.0623,
|
| 12256 |
+
"step": 2028
|
| 12257 |
+
},
|
| 12258 |
+
{
|
| 12259 |
+
"epoch": 1.39,
|
| 12260 |
+
"learning_rate": 0.00033784042044911607,
|
| 12261 |
+
"loss": 2.044,
|
| 12262 |
+
"step": 2029
|
| 12263 |
+
},
|
| 12264 |
+
{
|
| 12265 |
+
"epoch": 1.39,
|
| 12266 |
+
"learning_rate": 0.00033769708552317245,
|
| 12267 |
+
"loss": 2.0534,
|
| 12268 |
+
"step": 2030
|
| 12269 |
+
},
|
| 12270 |
+
{
|
| 12271 |
+
"epoch": 1.39,
|
| 12272 |
+
"learning_rate": 0.00033755375059722883,
|
| 12273 |
+
"loss": 1.9295,
|
| 12274 |
+
"step": 2031
|
| 12275 |
+
},
|
| 12276 |
+
{
|
| 12277 |
+
"epoch": 1.39,
|
| 12278 |
+
"learning_rate": 0.0003374104156712852,
|
| 12279 |
+
"loss": 2.141,
|
| 12280 |
+
"step": 2032
|
| 12281 |
+
},
|
| 12282 |
+
{
|
| 12283 |
+
"epoch": 1.39,
|
| 12284 |
+
"learning_rate": 0.0003372670807453416,
|
| 12285 |
+
"loss": 2.0085,
|
| 12286 |
+
"step": 2033
|
| 12287 |
+
},
|
| 12288 |
+
{
|
| 12289 |
+
"epoch": 1.39,
|
| 12290 |
+
"learning_rate": 0.00033712374581939793,
|
| 12291 |
+
"loss": 2.0188,
|
| 12292 |
+
"step": 2034
|
| 12293 |
+
},
|
| 12294 |
+
{
|
| 12295 |
+
"epoch": 1.39,
|
| 12296 |
+
"learning_rate": 0.00033698041089345437,
|
| 12297 |
+
"loss": 2.0537,
|
| 12298 |
+
"step": 2035
|
| 12299 |
+
},
|
| 12300 |
+
{
|
| 12301 |
+
"epoch": 1.39,
|
| 12302 |
+
"learning_rate": 0.0003368370759675107,
|
| 12303 |
+
"loss": 1.9598,
|
| 12304 |
+
"step": 2036
|
| 12305 |
+
},
|
| 12306 |
+
{
|
| 12307 |
+
"epoch": 1.39,
|
| 12308 |
+
"learning_rate": 0.00033669374104156714,
|
| 12309 |
+
"loss": 2.004,
|
| 12310 |
+
"step": 2037
|
| 12311 |
+
},
|
| 12312 |
+
{
|
| 12313 |
+
"epoch": 1.39,
|
| 12314 |
+
"learning_rate": 0.00033655040611562347,
|
| 12315 |
+
"loss": 2.0024,
|
| 12316 |
+
"step": 2038
|
| 12317 |
+
},
|
| 12318 |
+
{
|
| 12319 |
+
"epoch": 1.39,
|
| 12320 |
+
"learning_rate": 0.0003364070711896799,
|
| 12321 |
+
"loss": 2.0414,
|
| 12322 |
+
"step": 2039
|
| 12323 |
+
},
|
| 12324 |
+
{
|
| 12325 |
+
"epoch": 1.39,
|
| 12326 |
+
"learning_rate": 0.00033626373626373623,
|
| 12327 |
+
"loss": 2.0245,
|
| 12328 |
+
"step": 2040
|
| 12329 |
+
},
|
| 12330 |
+
{
|
| 12331 |
+
"epoch": 1.4,
|
| 12332 |
+
"learning_rate": 0.00033612040133779256,
|
| 12333 |
+
"loss": 1.9819,
|
| 12334 |
+
"step": 2041
|
| 12335 |
+
},
|
| 12336 |
+
{
|
| 12337 |
+
"epoch": 1.4,
|
| 12338 |
+
"learning_rate": 0.000335977066411849,
|
| 12339 |
+
"loss": 2.0282,
|
| 12340 |
+
"step": 2042
|
| 12341 |
+
},
|
| 12342 |
+
{
|
| 12343 |
+
"epoch": 1.4,
|
| 12344 |
+
"learning_rate": 0.00033583373148590533,
|
| 12345 |
+
"loss": 1.9962,
|
| 12346 |
+
"step": 2043
|
| 12347 |
+
},
|
| 12348 |
+
{
|
| 12349 |
+
"epoch": 1.4,
|
| 12350 |
+
"learning_rate": 0.00033569039655996177,
|
| 12351 |
+
"loss": 1.9422,
|
| 12352 |
+
"step": 2044
|
| 12353 |
+
},
|
| 12354 |
+
{
|
| 12355 |
+
"epoch": 1.4,
|
| 12356 |
+
"learning_rate": 0.0003355470616340181,
|
| 12357 |
+
"loss": 2.0411,
|
| 12358 |
+
"step": 2045
|
| 12359 |
+
},
|
| 12360 |
+
{
|
| 12361 |
+
"epoch": 1.4,
|
| 12362 |
+
"learning_rate": 0.0003354037267080745,
|
| 12363 |
+
"loss": 1.9571,
|
| 12364 |
+
"step": 2046
|
| 12365 |
+
},
|
| 12366 |
+
{
|
| 12367 |
+
"epoch": 1.4,
|
| 12368 |
+
"learning_rate": 0.00033526039178213087,
|
| 12369 |
+
"loss": 1.9993,
|
| 12370 |
+
"step": 2047
|
| 12371 |
+
},
|
| 12372 |
+
{
|
| 12373 |
+
"epoch": 1.4,
|
| 12374 |
+
"learning_rate": 0.00033511705685618725,
|
| 12375 |
+
"loss": 2.0761,
|
| 12376 |
+
"step": 2048
|
| 12377 |
+
},
|
| 12378 |
+
{
|
| 12379 |
+
"epoch": 1.4,
|
| 12380 |
+
"learning_rate": 0.00033497372193024363,
|
| 12381 |
+
"loss": 2.0052,
|
| 12382 |
+
"step": 2049
|
| 12383 |
+
},
|
| 12384 |
+
{
|
| 12385 |
+
"epoch": 1.4,
|
| 12386 |
+
"learning_rate": 0.0003348303870043,
|
| 12387 |
+
"loss": 2.0984,
|
| 12388 |
+
"step": 2050
|
| 12389 |
+
},
|
| 12390 |
+
{
|
| 12391 |
+
"epoch": 1.4,
|
| 12392 |
+
"learning_rate": 0.0003346870520783564,
|
| 12393 |
+
"loss": 2.0299,
|
| 12394 |
+
"step": 2051
|
| 12395 |
+
},
|
| 12396 |
+
{
|
| 12397 |
+
"epoch": 1.4,
|
| 12398 |
+
"learning_rate": 0.0003345437171524128,
|
| 12399 |
+
"loss": 1.9776,
|
| 12400 |
+
"step": 2052
|
| 12401 |
+
},
|
| 12402 |
+
{
|
| 12403 |
+
"epoch": 1.4,
|
| 12404 |
+
"learning_rate": 0.0003344003822264691,
|
| 12405 |
+
"loss": 2.0061,
|
| 12406 |
+
"step": 2053
|
| 12407 |
+
},
|
| 12408 |
+
{
|
| 12409 |
+
"epoch": 1.4,
|
| 12410 |
+
"learning_rate": 0.00033425704730052555,
|
| 12411 |
+
"loss": 2.0546,
|
| 12412 |
+
"step": 2054
|
| 12413 |
+
},
|
| 12414 |
+
{
|
| 12415 |
+
"epoch": 1.4,
|
| 12416 |
+
"learning_rate": 0.0003341137123745819,
|
| 12417 |
+
"loss": 2.0127,
|
| 12418 |
+
"step": 2055
|
| 12419 |
+
},
|
| 12420 |
+
{
|
| 12421 |
+
"epoch": 1.41,
|
| 12422 |
+
"learning_rate": 0.0003339703774486383,
|
| 12423 |
+
"loss": 2.0911,
|
| 12424 |
+
"step": 2056
|
| 12425 |
+
},
|
| 12426 |
+
{
|
| 12427 |
+
"epoch": 1.41,
|
| 12428 |
+
"learning_rate": 0.00033382704252269465,
|
| 12429 |
+
"loss": 1.9568,
|
| 12430 |
+
"step": 2057
|
| 12431 |
+
},
|
| 12432 |
+
{
|
| 12433 |
+
"epoch": 1.41,
|
| 12434 |
+
"learning_rate": 0.0003336837075967511,
|
| 12435 |
+
"loss": 2.0353,
|
| 12436 |
+
"step": 2058
|
| 12437 |
+
},
|
| 12438 |
+
{
|
| 12439 |
+
"epoch": 1.41,
|
| 12440 |
+
"learning_rate": 0.0003335403726708074,
|
| 12441 |
+
"loss": 1.9722,
|
| 12442 |
+
"step": 2059
|
| 12443 |
+
},
|
| 12444 |
+
{
|
| 12445 |
+
"epoch": 1.41,
|
| 12446 |
+
"learning_rate": 0.00033339703774486375,
|
| 12447 |
+
"loss": 2.0115,
|
| 12448 |
+
"step": 2060
|
| 12449 |
+
},
|
| 12450 |
+
{
|
| 12451 |
+
"epoch": 1.41,
|
| 12452 |
+
"learning_rate": 0.0003332537028189202,
|
| 12453 |
+
"loss": 1.9517,
|
| 12454 |
+
"step": 2061
|
| 12455 |
+
},
|
| 12456 |
+
{
|
| 12457 |
+
"epoch": 1.41,
|
| 12458 |
+
"learning_rate": 0.0003331103678929765,
|
| 12459 |
+
"loss": 1.9787,
|
| 12460 |
+
"step": 2062
|
| 12461 |
+
},
|
| 12462 |
+
{
|
| 12463 |
+
"epoch": 1.41,
|
| 12464 |
+
"learning_rate": 0.00033296703296703295,
|
| 12465 |
+
"loss": 2.0402,
|
| 12466 |
+
"step": 2063
|
| 12467 |
+
},
|
| 12468 |
+
{
|
| 12469 |
+
"epoch": 1.41,
|
| 12470 |
+
"learning_rate": 0.0003328236980410893,
|
| 12471 |
+
"loss": 2.0362,
|
| 12472 |
+
"step": 2064
|
| 12473 |
+
},
|
| 12474 |
+
{
|
| 12475 |
+
"epoch": 1.41,
|
| 12476 |
+
"learning_rate": 0.0003326803631151457,
|
| 12477 |
+
"loss": 2.0316,
|
| 12478 |
+
"step": 2065
|
| 12479 |
+
},
|
| 12480 |
+
{
|
| 12481 |
+
"epoch": 1.41,
|
| 12482 |
+
"learning_rate": 0.00033253702818920205,
|
| 12483 |
+
"loss": 1.9957,
|
| 12484 |
+
"step": 2066
|
| 12485 |
+
},
|
| 12486 |
+
{
|
| 12487 |
+
"epoch": 1.41,
|
| 12488 |
+
"learning_rate": 0.00033239369326325843,
|
| 12489 |
+
"loss": 2.076,
|
| 12490 |
+
"step": 2067
|
| 12491 |
+
},
|
| 12492 |
+
{
|
| 12493 |
+
"epoch": 1.41,
|
| 12494 |
+
"learning_rate": 0.0003322503583373148,
|
| 12495 |
+
"loss": 1.9697,
|
| 12496 |
+
"step": 2068
|
| 12497 |
+
},
|
| 12498 |
+
{
|
| 12499 |
+
"epoch": 1.41,
|
| 12500 |
+
"learning_rate": 0.0003321070234113712,
|
| 12501 |
+
"loss": 2.063,
|
| 12502 |
+
"step": 2069
|
| 12503 |
+
},
|
| 12504 |
+
{
|
| 12505 |
+
"epoch": 1.42,
|
| 12506 |
+
"learning_rate": 0.0003319636884854276,
|
| 12507 |
+
"loss": 2.0591,
|
| 12508 |
+
"step": 2070
|
| 12509 |
+
},
|
| 12510 |
+
{
|
| 12511 |
+
"epoch": 1.42,
|
| 12512 |
+
"learning_rate": 0.00033182035355948397,
|
| 12513 |
+
"loss": 2.031,
|
| 12514 |
+
"step": 2071
|
| 12515 |
+
},
|
| 12516 |
+
{
|
| 12517 |
+
"epoch": 1.42,
|
| 12518 |
+
"learning_rate": 0.00033167701863354035,
|
| 12519 |
+
"loss": 2.0116,
|
| 12520 |
+
"step": 2072
|
| 12521 |
+
},
|
| 12522 |
+
{
|
| 12523 |
+
"epoch": 1.42,
|
| 12524 |
+
"learning_rate": 0.00033153368370759674,
|
| 12525 |
+
"loss": 2.0301,
|
| 12526 |
+
"step": 2073
|
| 12527 |
+
},
|
| 12528 |
+
{
|
| 12529 |
+
"epoch": 1.42,
|
| 12530 |
+
"learning_rate": 0.00033139034878165306,
|
| 12531 |
+
"loss": 2.0387,
|
| 12532 |
+
"step": 2074
|
| 12533 |
+
},
|
| 12534 |
+
{
|
| 12535 |
+
"epoch": 1.42,
|
| 12536 |
+
"learning_rate": 0.0003312470138557095,
|
| 12537 |
+
"loss": 2.0549,
|
| 12538 |
+
"step": 2075
|
| 12539 |
+
},
|
| 12540 |
+
{
|
| 12541 |
+
"epoch": 1.42,
|
| 12542 |
+
"learning_rate": 0.00033110367892976583,
|
| 12543 |
+
"loss": 1.9861,
|
| 12544 |
+
"step": 2076
|
| 12545 |
+
},
|
| 12546 |
+
{
|
| 12547 |
+
"epoch": 1.42,
|
| 12548 |
+
"learning_rate": 0.00033096034400382227,
|
| 12549 |
+
"loss": 1.918,
|
| 12550 |
+
"step": 2077
|
| 12551 |
+
},
|
| 12552 |
+
{
|
| 12553 |
+
"epoch": 1.42,
|
| 12554 |
+
"learning_rate": 0.0003308170090778786,
|
| 12555 |
+
"loss": 1.9836,
|
| 12556 |
+
"step": 2078
|
| 12557 |
+
},
|
| 12558 |
+
{
|
| 12559 |
+
"epoch": 1.42,
|
| 12560 |
+
"learning_rate": 0.00033067367415193504,
|
| 12561 |
+
"loss": 2.0345,
|
| 12562 |
+
"step": 2079
|
| 12563 |
+
},
|
| 12564 |
+
{
|
| 12565 |
+
"epoch": 1.42,
|
| 12566 |
+
"learning_rate": 0.00033053033922599137,
|
| 12567 |
+
"loss": 1.9827,
|
| 12568 |
+
"step": 2080
|
| 12569 |
+
},
|
| 12570 |
+
{
|
| 12571 |
+
"epoch": 1.42,
|
| 12572 |
+
"learning_rate": 0.0003303870043000477,
|
| 12573 |
+
"loss": 1.9904,
|
| 12574 |
+
"step": 2081
|
| 12575 |
+
},
|
| 12576 |
+
{
|
| 12577 |
+
"epoch": 1.42,
|
| 12578 |
+
"learning_rate": 0.00033024366937410413,
|
| 12579 |
+
"loss": 1.9661,
|
| 12580 |
+
"step": 2082
|
| 12581 |
+
},
|
| 12582 |
+
{
|
| 12583 |
+
"epoch": 1.42,
|
| 12584 |
+
"learning_rate": 0.00033010033444816046,
|
| 12585 |
+
"loss": 2.0373,
|
| 12586 |
+
"step": 2083
|
| 12587 |
+
},
|
| 12588 |
+
{
|
| 12589 |
+
"epoch": 1.42,
|
| 12590 |
+
"learning_rate": 0.0003299569995222169,
|
| 12591 |
+
"loss": 1.872,
|
| 12592 |
+
"step": 2084
|
| 12593 |
+
},
|
| 12594 |
+
{
|
| 12595 |
+
"epoch": 1.43,
|
| 12596 |
+
"learning_rate": 0.00032981366459627323,
|
| 12597 |
+
"loss": 1.9871,
|
| 12598 |
+
"step": 2085
|
| 12599 |
+
},
|
| 12600 |
+
{
|
| 12601 |
+
"epoch": 1.43,
|
| 12602 |
+
"learning_rate": 0.00032967032967032967,
|
| 12603 |
+
"loss": 1.9051,
|
| 12604 |
+
"step": 2086
|
| 12605 |
+
},
|
| 12606 |
+
{
|
| 12607 |
+
"epoch": 1.43,
|
| 12608 |
+
"learning_rate": 0.000329526994744386,
|
| 12609 |
+
"loss": 2.1045,
|
| 12610 |
+
"step": 2087
|
| 12611 |
+
},
|
| 12612 |
+
{
|
| 12613 |
+
"epoch": 1.43,
|
| 12614 |
+
"learning_rate": 0.0003293836598184424,
|
| 12615 |
+
"loss": 1.9143,
|
| 12616 |
+
"step": 2088
|
| 12617 |
+
},
|
| 12618 |
+
{
|
| 12619 |
+
"epoch": 1.43,
|
| 12620 |
+
"learning_rate": 0.00032924032489249877,
|
| 12621 |
+
"loss": 1.8973,
|
| 12622 |
+
"step": 2089
|
| 12623 |
+
},
|
| 12624 |
+
{
|
| 12625 |
+
"epoch": 1.43,
|
| 12626 |
+
"learning_rate": 0.00032909698996655515,
|
| 12627 |
+
"loss": 1.9506,
|
| 12628 |
+
"step": 2090
|
| 12629 |
+
},
|
| 12630 |
+
{
|
| 12631 |
+
"epoch": 1.43,
|
| 12632 |
+
"learning_rate": 0.00032895365504061153,
|
| 12633 |
+
"loss": 1.9271,
|
| 12634 |
+
"step": 2091
|
| 12635 |
+
},
|
| 12636 |
+
{
|
| 12637 |
+
"epoch": 1.43,
|
| 12638 |
+
"learning_rate": 0.0003288103201146679,
|
| 12639 |
+
"loss": 1.9837,
|
| 12640 |
+
"step": 2092
|
| 12641 |
+
},
|
| 12642 |
+
{
|
| 12643 |
+
"epoch": 1.43,
|
| 12644 |
+
"learning_rate": 0.0003286669851887243,
|
| 12645 |
+
"loss": 2.1084,
|
| 12646 |
+
"step": 2093
|
| 12647 |
+
},
|
| 12648 |
+
{
|
| 12649 |
+
"epoch": 1.43,
|
| 12650 |
+
"learning_rate": 0.0003285236502627807,
|
| 12651 |
+
"loss": 1.9546,
|
| 12652 |
+
"step": 2094
|
| 12653 |
+
},
|
| 12654 |
+
{
|
| 12655 |
+
"epoch": 1.43,
|
| 12656 |
+
"learning_rate": 0.000328380315336837,
|
| 12657 |
+
"loss": 2.0735,
|
| 12658 |
+
"step": 2095
|
| 12659 |
+
},
|
| 12660 |
+
{
|
| 12661 |
+
"epoch": 1.43,
|
| 12662 |
+
"learning_rate": 0.00032823698041089345,
|
| 12663 |
+
"loss": 2.0231,
|
| 12664 |
+
"step": 2096
|
| 12665 |
+
},
|
| 12666 |
+
{
|
| 12667 |
+
"epoch": 1.43,
|
| 12668 |
+
"learning_rate": 0.0003280936454849498,
|
| 12669 |
+
"loss": 2.0705,
|
| 12670 |
+
"step": 2097
|
| 12671 |
+
},
|
| 12672 |
+
{
|
| 12673 |
+
"epoch": 1.43,
|
| 12674 |
+
"learning_rate": 0.0003279503105590062,
|
| 12675 |
+
"loss": 1.9939,
|
| 12676 |
+
"step": 2098
|
| 12677 |
+
},
|
| 12678 |
+
{
|
| 12679 |
+
"epoch": 1.44,
|
| 12680 |
+
"learning_rate": 0.00032780697563306255,
|
| 12681 |
+
"loss": 1.9815,
|
| 12682 |
+
"step": 2099
|
| 12683 |
+
},
|
| 12684 |
+
{
|
| 12685 |
+
"epoch": 1.44,
|
| 12686 |
+
"learning_rate": 0.000327663640707119,
|
| 12687 |
+
"loss": 1.9705,
|
| 12688 |
+
"step": 2100
|
| 12689 |
+
},
|
| 12690 |
+
{
|
| 12691 |
+
"epoch": 1.44,
|
| 12692 |
+
"learning_rate": 0.0003275203057811753,
|
| 12693 |
+
"loss": 2.0562,
|
| 12694 |
+
"step": 2101
|
| 12695 |
+
},
|
| 12696 |
+
{
|
| 12697 |
+
"epoch": 1.44,
|
| 12698 |
+
"learning_rate": 0.00032737697085523165,
|
| 12699 |
+
"loss": 2.014,
|
| 12700 |
+
"step": 2102
|
| 12701 |
+
},
|
| 12702 |
+
{
|
| 12703 |
+
"epoch": 1.44,
|
| 12704 |
+
"learning_rate": 0.0003272336359292881,
|
| 12705 |
+
"loss": 2.0447,
|
| 12706 |
+
"step": 2103
|
| 12707 |
+
},
|
| 12708 |
+
{
|
| 12709 |
+
"epoch": 1.44,
|
| 12710 |
+
"learning_rate": 0.0003270903010033444,
|
| 12711 |
+
"loss": 1.9627,
|
| 12712 |
+
"step": 2104
|
| 12713 |
+
},
|
| 12714 |
+
{
|
| 12715 |
+
"epoch": 1.44,
|
| 12716 |
+
"learning_rate": 0.00032694696607740085,
|
| 12717 |
+
"loss": 1.9587,
|
| 12718 |
+
"step": 2105
|
| 12719 |
+
},
|
| 12720 |
+
{
|
| 12721 |
+
"epoch": 1.44,
|
| 12722 |
+
"learning_rate": 0.0003268036311514572,
|
| 12723 |
+
"loss": 2.0637,
|
| 12724 |
+
"step": 2106
|
| 12725 |
+
},
|
| 12726 |
+
{
|
| 12727 |
+
"epoch": 1.44,
|
| 12728 |
+
"learning_rate": 0.0003266602962255136,
|
| 12729 |
+
"loss": 1.9581,
|
| 12730 |
+
"step": 2107
|
| 12731 |
+
},
|
| 12732 |
+
{
|
| 12733 |
+
"epoch": 1.44,
|
| 12734 |
+
"learning_rate": 0.00032651696129956995,
|
| 12735 |
+
"loss": 1.9609,
|
| 12736 |
+
"step": 2108
|
| 12737 |
+
},
|
| 12738 |
+
{
|
| 12739 |
+
"epoch": 1.44,
|
| 12740 |
+
"learning_rate": 0.00032637362637362633,
|
| 12741 |
+
"loss": 1.9956,
|
| 12742 |
+
"step": 2109
|
| 12743 |
+
},
|
| 12744 |
+
{
|
| 12745 |
+
"epoch": 1.44,
|
| 12746 |
+
"learning_rate": 0.0003262302914476827,
|
| 12747 |
+
"loss": 1.9885,
|
| 12748 |
+
"step": 2110
|
| 12749 |
+
},
|
| 12750 |
+
{
|
| 12751 |
+
"epoch": 1.44,
|
| 12752 |
+
"learning_rate": 0.0003260869565217391,
|
| 12753 |
+
"loss": 1.9506,
|
| 12754 |
+
"step": 2111
|
| 12755 |
+
},
|
| 12756 |
+
{
|
| 12757 |
+
"epoch": 1.44,
|
| 12758 |
+
"learning_rate": 0.0003259436215957955,
|
| 12759 |
+
"loss": 2.0175,
|
| 12760 |
+
"step": 2112
|
| 12761 |
+
},
|
| 12762 |
+
{
|
| 12763 |
+
"epoch": 1.44,
|
| 12764 |
+
"learning_rate": 0.00032580028666985187,
|
| 12765 |
+
"loss": 1.9666,
|
| 12766 |
+
"step": 2113
|
| 12767 |
+
},
|
| 12768 |
+
{
|
| 12769 |
+
"epoch": 1.45,
|
| 12770 |
+
"learning_rate": 0.00032565695174390825,
|
| 12771 |
+
"loss": 2.0387,
|
| 12772 |
+
"step": 2114
|
| 12773 |
+
},
|
| 12774 |
+
{
|
| 12775 |
+
"epoch": 1.45,
|
| 12776 |
+
"learning_rate": 0.00032551361681796464,
|
| 12777 |
+
"loss": 2.0288,
|
| 12778 |
+
"step": 2115
|
| 12779 |
+
},
|
| 12780 |
+
{
|
| 12781 |
+
"epoch": 1.45,
|
| 12782 |
+
"learning_rate": 0.00032537028189202097,
|
| 12783 |
+
"loss": 2.0741,
|
| 12784 |
+
"step": 2116
|
| 12785 |
+
},
|
| 12786 |
+
{
|
| 12787 |
+
"epoch": 1.45,
|
| 12788 |
+
"learning_rate": 0.0003252269469660774,
|
| 12789 |
+
"loss": 2.0846,
|
| 12790 |
+
"step": 2117
|
| 12791 |
+
},
|
| 12792 |
+
{
|
| 12793 |
+
"epoch": 1.45,
|
| 12794 |
+
"learning_rate": 0.00032508361204013373,
|
| 12795 |
+
"loss": 1.974,
|
| 12796 |
+
"step": 2118
|
| 12797 |
+
},
|
| 12798 |
+
{
|
| 12799 |
+
"epoch": 1.45,
|
| 12800 |
+
"learning_rate": 0.00032494027711419017,
|
| 12801 |
+
"loss": 1.9899,
|
| 12802 |
+
"step": 2119
|
| 12803 |
+
},
|
| 12804 |
+
{
|
| 12805 |
+
"epoch": 1.45,
|
| 12806 |
+
"learning_rate": 0.0003247969421882465,
|
| 12807 |
+
"loss": 2.1513,
|
| 12808 |
+
"step": 2120
|
| 12809 |
+
},
|
| 12810 |
+
{
|
| 12811 |
+
"epoch": 1.45,
|
| 12812 |
+
"learning_rate": 0.00032465360726230294,
|
| 12813 |
+
"loss": 1.9642,
|
| 12814 |
+
"step": 2121
|
| 12815 |
+
},
|
| 12816 |
+
{
|
| 12817 |
+
"epoch": 1.45,
|
| 12818 |
+
"learning_rate": 0.00032451027233635927,
|
| 12819 |
+
"loss": 1.9595,
|
| 12820 |
+
"step": 2122
|
| 12821 |
+
},
|
| 12822 |
+
{
|
| 12823 |
+
"epoch": 1.45,
|
| 12824 |
+
"learning_rate": 0.0003243669374104156,
|
| 12825 |
+
"loss": 2.0706,
|
| 12826 |
+
"step": 2123
|
| 12827 |
+
},
|
| 12828 |
+
{
|
| 12829 |
+
"epoch": 1.45,
|
| 12830 |
+
"learning_rate": 0.00032422360248447204,
|
| 12831 |
+
"loss": 2.0984,
|
| 12832 |
+
"step": 2124
|
| 12833 |
+
},
|
| 12834 |
+
{
|
| 12835 |
+
"epoch": 1.45,
|
| 12836 |
+
"learning_rate": 0.00032408026755852837,
|
| 12837 |
+
"loss": 2.1254,
|
| 12838 |
+
"step": 2125
|
| 12839 |
+
},
|
| 12840 |
+
{
|
| 12841 |
+
"epoch": 1.45,
|
| 12842 |
+
"learning_rate": 0.0003239369326325848,
|
| 12843 |
+
"loss": 1.9552,
|
| 12844 |
+
"step": 2126
|
| 12845 |
+
},
|
| 12846 |
+
{
|
| 12847 |
+
"epoch": 1.45,
|
| 12848 |
+
"learning_rate": 0.00032379359770664113,
|
| 12849 |
+
"loss": 1.9694,
|
| 12850 |
+
"step": 2127
|
| 12851 |
+
},
|
| 12852 |
+
{
|
| 12853 |
+
"epoch": 1.45,
|
| 12854 |
+
"learning_rate": 0.00032365026278069757,
|
| 12855 |
+
"loss": 2.0373,
|
| 12856 |
+
"step": 2128
|
| 12857 |
+
},
|
| 12858 |
+
{
|
| 12859 |
+
"epoch": 1.46,
|
| 12860 |
+
"learning_rate": 0.0003235069278547539,
|
| 12861 |
+
"loss": 2.0331,
|
| 12862 |
+
"step": 2129
|
| 12863 |
+
},
|
| 12864 |
+
{
|
| 12865 |
+
"epoch": 1.46,
|
| 12866 |
+
"learning_rate": 0.0003233635929288103,
|
| 12867 |
+
"loss": 2.0722,
|
| 12868 |
+
"step": 2130
|
| 12869 |
+
},
|
| 12870 |
+
{
|
| 12871 |
+
"epoch": 1.46,
|
| 12872 |
+
"learning_rate": 0.00032322025800286667,
|
| 12873 |
+
"loss": 2.0406,
|
| 12874 |
+
"step": 2131
|
| 12875 |
+
},
|
| 12876 |
+
{
|
| 12877 |
+
"epoch": 1.46,
|
| 12878 |
+
"learning_rate": 0.00032307692307692305,
|
| 12879 |
+
"loss": 1.8967,
|
| 12880 |
+
"step": 2132
|
| 12881 |
+
},
|
| 12882 |
+
{
|
| 12883 |
+
"epoch": 1.46,
|
| 12884 |
+
"learning_rate": 0.00032293358815097944,
|
| 12885 |
+
"loss": 1.9515,
|
| 12886 |
+
"step": 2133
|
| 12887 |
+
},
|
| 12888 |
+
{
|
| 12889 |
+
"epoch": 1.46,
|
| 12890 |
+
"learning_rate": 0.0003227902532250358,
|
| 12891 |
+
"loss": 2.0313,
|
| 12892 |
+
"step": 2134
|
| 12893 |
+
},
|
| 12894 |
+
{
|
| 12895 |
+
"epoch": 1.46,
|
| 12896 |
+
"learning_rate": 0.0003226469182990922,
|
| 12897 |
+
"loss": 2.0825,
|
| 12898 |
+
"step": 2135
|
| 12899 |
+
},
|
| 12900 |
+
{
|
| 12901 |
+
"epoch": 1.46,
|
| 12902 |
+
"learning_rate": 0.0003225035833731486,
|
| 12903 |
+
"loss": 2.0743,
|
| 12904 |
+
"step": 2136
|
| 12905 |
+
},
|
| 12906 |
+
{
|
| 12907 |
+
"epoch": 1.46,
|
| 12908 |
+
"learning_rate": 0.0003223602484472049,
|
| 12909 |
+
"loss": 2.0133,
|
| 12910 |
+
"step": 2137
|
| 12911 |
+
},
|
| 12912 |
+
{
|
| 12913 |
+
"epoch": 1.46,
|
| 12914 |
+
"learning_rate": 0.00032221691352126136,
|
| 12915 |
+
"loss": 1.9019,
|
| 12916 |
+
"step": 2138
|
| 12917 |
+
},
|
| 12918 |
+
{
|
| 12919 |
+
"epoch": 1.46,
|
| 12920 |
+
"learning_rate": 0.0003220735785953177,
|
| 12921 |
+
"loss": 1.9543,
|
| 12922 |
+
"step": 2139
|
| 12923 |
+
},
|
| 12924 |
+
{
|
| 12925 |
+
"epoch": 1.46,
|
| 12926 |
+
"learning_rate": 0.0003219302436693741,
|
| 12927 |
+
"loss": 1.999,
|
| 12928 |
+
"step": 2140
|
| 12929 |
+
},
|
| 12930 |
+
{
|
| 12931 |
+
"epoch": 1.46,
|
| 12932 |
+
"learning_rate": 0.00032178690874343045,
|
| 12933 |
+
"loss": 1.927,
|
| 12934 |
+
"step": 2141
|
| 12935 |
+
},
|
| 12936 |
+
{
|
| 12937 |
+
"epoch": 1.46,
|
| 12938 |
+
"learning_rate": 0.0003216435738174869,
|
| 12939 |
+
"loss": 2.069,
|
| 12940 |
+
"step": 2142
|
| 12941 |
+
},
|
| 12942 |
+
{
|
| 12943 |
+
"epoch": 1.47,
|
| 12944 |
+
"learning_rate": 0.0003215002388915432,
|
| 12945 |
+
"loss": 1.9393,
|
| 12946 |
+
"step": 2143
|
| 12947 |
+
},
|
| 12948 |
+
{
|
| 12949 |
+
"epoch": 1.47,
|
| 12950 |
+
"learning_rate": 0.00032135690396559955,
|
| 12951 |
+
"loss": 1.9443,
|
| 12952 |
+
"step": 2144
|
| 12953 |
+
},
|
| 12954 |
+
{
|
| 12955 |
+
"epoch": 1.47,
|
| 12956 |
+
"learning_rate": 0.000321213569039656,
|
| 12957 |
+
"loss": 1.9659,
|
| 12958 |
+
"step": 2145
|
| 12959 |
+
},
|
| 12960 |
+
{
|
| 12961 |
+
"epoch": 1.47,
|
| 12962 |
+
"learning_rate": 0.0003210702341137123,
|
| 12963 |
+
"loss": 2.0523,
|
| 12964 |
+
"step": 2146
|
| 12965 |
+
},
|
| 12966 |
+
{
|
| 12967 |
+
"epoch": 1.47,
|
| 12968 |
+
"learning_rate": 0.00032092689918776876,
|
| 12969 |
+
"loss": 2.0378,
|
| 12970 |
+
"step": 2147
|
| 12971 |
+
},
|
| 12972 |
+
{
|
| 12973 |
+
"epoch": 1.47,
|
| 12974 |
+
"learning_rate": 0.0003207835642618251,
|
| 12975 |
+
"loss": 2.0093,
|
| 12976 |
+
"step": 2148
|
| 12977 |
+
},
|
| 12978 |
+
{
|
| 12979 |
+
"epoch": 1.47,
|
| 12980 |
+
"learning_rate": 0.00032064022933588147,
|
| 12981 |
+
"loss": 2.0546,
|
| 12982 |
+
"step": 2149
|
| 12983 |
+
},
|
| 12984 |
+
{
|
| 12985 |
+
"epoch": 1.47,
|
| 12986 |
+
"learning_rate": 0.00032049689440993785,
|
| 12987 |
+
"loss": 2.0351,
|
| 12988 |
+
"step": 2150
|
| 12989 |
+
},
|
| 12990 |
+
{
|
| 12991 |
+
"epoch": 1.47,
|
| 12992 |
+
"learning_rate": 0.00032035355948399424,
|
| 12993 |
+
"loss": 1.9175,
|
| 12994 |
+
"step": 2151
|
| 12995 |
+
},
|
| 12996 |
+
{
|
| 12997 |
+
"epoch": 1.47,
|
| 12998 |
+
"learning_rate": 0.0003202102245580506,
|
| 12999 |
+
"loss": 1.9986,
|
| 13000 |
+
"step": 2152
|
| 13001 |
+
},
|
| 13002 |
+
{
|
| 13003 |
+
"epoch": 1.47,
|
| 13004 |
+
"learning_rate": 0.000320066889632107,
|
| 13005 |
+
"loss": 1.9202,
|
| 13006 |
+
"step": 2153
|
| 13007 |
+
},
|
| 13008 |
+
{
|
| 13009 |
+
"epoch": 1.47,
|
| 13010 |
+
"learning_rate": 0.0003199235547061634,
|
| 13011 |
+
"loss": 1.9855,
|
| 13012 |
+
"step": 2154
|
| 13013 |
+
},
|
| 13014 |
+
{
|
| 13015 |
+
"epoch": 1.47,
|
| 13016 |
+
"learning_rate": 0.00031978021978021977,
|
| 13017 |
+
"loss": 1.9847,
|
| 13018 |
+
"step": 2155
|
| 13019 |
+
},
|
| 13020 |
+
{
|
| 13021 |
+
"epoch": 1.47,
|
| 13022 |
+
"learning_rate": 0.0003196368848542761,
|
| 13023 |
+
"loss": 2.096,
|
| 13024 |
+
"step": 2156
|
| 13025 |
+
},
|
| 13026 |
+
{
|
| 13027 |
+
"epoch": 1.47,
|
| 13028 |
+
"learning_rate": 0.00031949354992833254,
|
| 13029 |
+
"loss": 2.0109,
|
| 13030 |
+
"step": 2157
|
| 13031 |
+
},
|
| 13032 |
+
{
|
| 13033 |
+
"epoch": 1.48,
|
| 13034 |
+
"learning_rate": 0.00031935021500238887,
|
| 13035 |
+
"loss": 1.8784,
|
| 13036 |
+
"step": 2158
|
| 13037 |
+
},
|
| 13038 |
+
{
|
| 13039 |
+
"epoch": 1.48,
|
| 13040 |
+
"learning_rate": 0.0003192068800764453,
|
| 13041 |
+
"loss": 1.902,
|
| 13042 |
+
"step": 2159
|
| 13043 |
+
},
|
| 13044 |
+
{
|
| 13045 |
+
"epoch": 1.48,
|
| 13046 |
+
"learning_rate": 0.00031906354515050164,
|
| 13047 |
+
"loss": 2.0265,
|
| 13048 |
+
"step": 2160
|
| 13049 |
+
},
|
| 13050 |
+
{
|
| 13051 |
+
"epoch": 1.48,
|
| 13052 |
+
"learning_rate": 0.0003189202102245581,
|
| 13053 |
+
"loss": 1.9539,
|
| 13054 |
+
"step": 2161
|
| 13055 |
+
},
|
| 13056 |
+
{
|
| 13057 |
+
"epoch": 1.48,
|
| 13058 |
+
"learning_rate": 0.0003187768752986144,
|
| 13059 |
+
"loss": 1.9743,
|
| 13060 |
+
"step": 2162
|
| 13061 |
+
},
|
| 13062 |
+
{
|
| 13063 |
+
"epoch": 1.48,
|
| 13064 |
+
"learning_rate": 0.00031863354037267073,
|
| 13065 |
+
"loss": 2.0785,
|
| 13066 |
+
"step": 2163
|
| 13067 |
+
},
|
| 13068 |
+
{
|
| 13069 |
+
"epoch": 1.48,
|
| 13070 |
+
"learning_rate": 0.00031849020544672717,
|
| 13071 |
+
"loss": 2.0334,
|
| 13072 |
+
"step": 2164
|
| 13073 |
+
},
|
| 13074 |
+
{
|
| 13075 |
+
"epoch": 1.48,
|
| 13076 |
+
"learning_rate": 0.0003183468705207835,
|
| 13077 |
+
"loss": 1.8959,
|
| 13078 |
+
"step": 2165
|
| 13079 |
+
},
|
| 13080 |
+
{
|
| 13081 |
+
"epoch": 1.48,
|
| 13082 |
+
"learning_rate": 0.00031820353559483994,
|
| 13083 |
+
"loss": 1.9354,
|
| 13084 |
+
"step": 2166
|
| 13085 |
+
},
|
| 13086 |
+
{
|
| 13087 |
+
"epoch": 1.48,
|
| 13088 |
+
"learning_rate": 0.00031806020066889627,
|
| 13089 |
+
"loss": 2.1023,
|
| 13090 |
+
"step": 2167
|
| 13091 |
+
},
|
| 13092 |
+
{
|
| 13093 |
+
"epoch": 1.48,
|
| 13094 |
+
"learning_rate": 0.0003179168657429527,
|
| 13095 |
+
"loss": 2.034,
|
| 13096 |
+
"step": 2168
|
| 13097 |
+
},
|
| 13098 |
+
{
|
| 13099 |
+
"epoch": 1.48,
|
| 13100 |
+
"learning_rate": 0.00031777353081700904,
|
| 13101 |
+
"loss": 1.9602,
|
| 13102 |
+
"step": 2169
|
| 13103 |
+
},
|
| 13104 |
+
{
|
| 13105 |
+
"epoch": 1.48,
|
| 13106 |
+
"learning_rate": 0.0003176301958910654,
|
| 13107 |
+
"loss": 1.9826,
|
| 13108 |
+
"step": 2170
|
| 13109 |
+
},
|
| 13110 |
+
{
|
| 13111 |
+
"epoch": 1.48,
|
| 13112 |
+
"learning_rate": 0.0003174868609651218,
|
| 13113 |
+
"loss": 1.9921,
|
| 13114 |
+
"step": 2171
|
| 13115 |
+
},
|
| 13116 |
+
{
|
| 13117 |
+
"epoch": 1.48,
|
| 13118 |
+
"learning_rate": 0.0003173435260391782,
|
| 13119 |
+
"loss": 1.9476,
|
| 13120 |
+
"step": 2172
|
| 13121 |
+
},
|
| 13122 |
+
{
|
| 13123 |
+
"epoch": 1.49,
|
| 13124 |
+
"learning_rate": 0.00031720019111323457,
|
| 13125 |
+
"loss": 2.0478,
|
| 13126 |
+
"step": 2173
|
| 13127 |
+
},
|
| 13128 |
+
{
|
| 13129 |
+
"epoch": 1.49,
|
| 13130 |
+
"learning_rate": 0.00031705685618729095,
|
| 13131 |
+
"loss": 2.0748,
|
| 13132 |
+
"step": 2174
|
| 13133 |
+
},
|
| 13134 |
+
{
|
| 13135 |
+
"epoch": 1.49,
|
| 13136 |
+
"learning_rate": 0.00031691352126134734,
|
| 13137 |
+
"loss": 1.9633,
|
| 13138 |
+
"step": 2175
|
| 13139 |
+
},
|
| 13140 |
+
{
|
| 13141 |
+
"epoch": 1.49,
|
| 13142 |
+
"learning_rate": 0.0003167701863354037,
|
| 13143 |
+
"loss": 2.001,
|
| 13144 |
+
"step": 2176
|
| 13145 |
+
},
|
| 13146 |
+
{
|
| 13147 |
+
"epoch": 1.49,
|
| 13148 |
+
"learning_rate": 0.00031662685140946005,
|
| 13149 |
+
"loss": 2.0642,
|
| 13150 |
+
"step": 2177
|
| 13151 |
+
},
|
| 13152 |
+
{
|
| 13153 |
+
"epoch": 1.49,
|
| 13154 |
+
"learning_rate": 0.0003164835164835165,
|
| 13155 |
+
"loss": 1.9987,
|
| 13156 |
+
"step": 2178
|
| 13157 |
+
},
|
| 13158 |
+
{
|
| 13159 |
+
"epoch": 1.49,
|
| 13160 |
+
"learning_rate": 0.0003163401815575728,
|
| 13161 |
+
"loss": 1.9435,
|
| 13162 |
+
"step": 2179
|
| 13163 |
+
},
|
| 13164 |
+
{
|
| 13165 |
+
"epoch": 1.49,
|
| 13166 |
+
"learning_rate": 0.00031619684663162926,
|
| 13167 |
+
"loss": 1.9399,
|
| 13168 |
+
"step": 2180
|
| 13169 |
+
},
|
| 13170 |
+
{
|
| 13171 |
+
"epoch": 1.49,
|
| 13172 |
+
"learning_rate": 0.0003160535117056856,
|
| 13173 |
+
"loss": 1.9618,
|
| 13174 |
+
"step": 2181
|
| 13175 |
+
},
|
| 13176 |
+
{
|
| 13177 |
+
"epoch": 1.49,
|
| 13178 |
+
"learning_rate": 0.00031591017677974197,
|
| 13179 |
+
"loss": 1.9743,
|
| 13180 |
+
"step": 2182
|
| 13181 |
+
},
|
| 13182 |
+
{
|
| 13183 |
+
"epoch": 1.49,
|
| 13184 |
+
"learning_rate": 0.00031576684185379835,
|
| 13185 |
+
"loss": 1.9511,
|
| 13186 |
+
"step": 2183
|
| 13187 |
+
},
|
| 13188 |
+
{
|
| 13189 |
+
"epoch": 1.49,
|
| 13190 |
+
"learning_rate": 0.0003156235069278547,
|
| 13191 |
+
"loss": 2.0058,
|
| 13192 |
+
"step": 2184
|
| 13193 |
+
},
|
| 13194 |
+
{
|
| 13195 |
+
"epoch": 1.49,
|
| 13196 |
+
"learning_rate": 0.0003154801720019111,
|
| 13197 |
+
"loss": 2.0174,
|
| 13198 |
+
"step": 2185
|
| 13199 |
+
},
|
| 13200 |
+
{
|
| 13201 |
+
"epoch": 1.49,
|
| 13202 |
+
"learning_rate": 0.00031533683707596745,
|
| 13203 |
+
"loss": 2.0545,
|
| 13204 |
+
"step": 2186
|
| 13205 |
+
},
|
| 13206 |
+
{
|
| 13207 |
+
"epoch": 1.5,
|
| 13208 |
+
"learning_rate": 0.0003151935021500239,
|
| 13209 |
+
"loss": 1.9771,
|
| 13210 |
+
"step": 2187
|
| 13211 |
+
},
|
| 13212 |
+
{
|
| 13213 |
+
"epoch": 1.5,
|
| 13214 |
+
"learning_rate": 0.0003150501672240802,
|
| 13215 |
+
"loss": 1.9741,
|
| 13216 |
+
"step": 2188
|
| 13217 |
+
},
|
| 13218 |
+
{
|
| 13219 |
+
"epoch": 1.5,
|
| 13220 |
+
"learning_rate": 0.00031490683229813666,
|
| 13221 |
+
"loss": 1.9993,
|
| 13222 |
+
"step": 2189
|
| 13223 |
+
},
|
| 13224 |
+
{
|
| 13225 |
+
"epoch": 1.5,
|
| 13226 |
+
"learning_rate": 0.000314763497372193,
|
| 13227 |
+
"loss": 1.9776,
|
| 13228 |
+
"step": 2190
|
| 13229 |
+
},
|
| 13230 |
+
{
|
| 13231 |
+
"epoch": 1.5,
|
| 13232 |
+
"learning_rate": 0.00031462016244624937,
|
| 13233 |
+
"loss": 2.0027,
|
| 13234 |
+
"step": 2191
|
| 13235 |
+
},
|
| 13236 |
+
{
|
| 13237 |
+
"epoch": 1.5,
|
| 13238 |
+
"learning_rate": 0.00031447682752030575,
|
| 13239 |
+
"loss": 2.0119,
|
| 13240 |
+
"step": 2192
|
| 13241 |
+
},
|
| 13242 |
+
{
|
| 13243 |
+
"epoch": 1.5,
|
| 13244 |
+
"learning_rate": 0.00031433349259436214,
|
| 13245 |
+
"loss": 1.9371,
|
| 13246 |
+
"step": 2193
|
| 13247 |
+
},
|
| 13248 |
+
{
|
| 13249 |
+
"epoch": 1.5,
|
| 13250 |
+
"learning_rate": 0.0003141901576684185,
|
| 13251 |
+
"loss": 2.0221,
|
| 13252 |
+
"step": 2194
|
| 13253 |
+
},
|
| 13254 |
+
{
|
| 13255 |
+
"epoch": 1.5,
|
| 13256 |
+
"learning_rate": 0.0003140468227424749,
|
| 13257 |
+
"loss": 2.0126,
|
| 13258 |
+
"step": 2195
|
| 13259 |
+
},
|
| 13260 |
+
{
|
| 13261 |
+
"epoch": 1.5,
|
| 13262 |
+
"learning_rate": 0.0003139034878165313,
|
| 13263 |
+
"loss": 2.066,
|
| 13264 |
+
"step": 2196
|
| 13265 |
+
},
|
| 13266 |
+
{
|
| 13267 |
+
"epoch": 1.5,
|
| 13268 |
+
"learning_rate": 0.0003137601528905877,
|
| 13269 |
+
"loss": 1.9748,
|
| 13270 |
+
"step": 2197
|
| 13271 |
+
},
|
| 13272 |
+
{
|
| 13273 |
+
"epoch": 1.5,
|
| 13274 |
+
"learning_rate": 0.000313616817964644,
|
| 13275 |
+
"loss": 1.9549,
|
| 13276 |
+
"step": 2198
|
| 13277 |
+
},
|
| 13278 |
+
{
|
| 13279 |
+
"epoch": 1.5,
|
| 13280 |
+
"learning_rate": 0.0003134734830387004,
|
| 13281 |
+
"loss": 2.0066,
|
| 13282 |
+
"step": 2199
|
| 13283 |
+
},
|
| 13284 |
+
{
|
| 13285 |
+
"epoch": 1.5,
|
| 13286 |
+
"learning_rate": 0.00031333014811275677,
|
| 13287 |
+
"loss": 2.0541,
|
| 13288 |
+
"step": 2200
|
| 13289 |
+
},
|
| 13290 |
+
{
|
| 13291 |
+
"epoch": 1.5,
|
| 13292 |
+
"eval_loss": 1.9827316999435425,
|
| 13293 |
+
"eval_runtime": 1651.9566,
|
| 13294 |
+
"eval_samples_per_second": 6.053,
|
| 13295 |
+
"eval_steps_per_second": 6.053,
|
| 13296 |
+
"step": 2200
|
| 13297 |
}
|
| 13298 |
],
|
| 13299 |
"max_steps": 4386,
|
| 13300 |
"num_train_epochs": 3,
|
| 13301 |
+
"total_flos": 2.69384075772468e+18,
|
| 13302 |
"trial_name": null,
|
| 13303 |
"trial_params": null
|
| 13304 |
}
|
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:ce38e42231eaec2ee57adbd3a558f89d6a9d15cb466c85343dd8a5eee549c297
|
| 3 |
size 2368281769
|