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 +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 sha256:2840dcbe13ebc59d5938192139ea78cb7c1c1d2a91ba80b7d1dc7ea55876d800
|
| 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:af24d837954601d8f9729f848ea03545b9388763d421c6d2dce0a7e8ab73e67c
|
| 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:d22bcbbfffa1691c652a8c7287d2c77ac5f4da8a53b1ef656dda3633adb8346e
|
| 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:e13dfeda80fac3770a6173d870ebee06f4c0adc8fe9cfc980b6faa21c8de1112
|
| 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,
|
|
@@ -10878,11 +10878,1219 @@
|
|
| 10878 |
"eval_samples_per_second": 6.061,
|
| 10879 |
"eval_steps_per_second": 6.061,
|
| 10880 |
"step": 1800
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 10881 |
}
|
| 10882 |
],
|
| 10883 |
"max_steps": 4386,
|
| 10884 |
"num_train_epochs": 3,
|
| 10885 |
-
"total_flos": 2.
|
| 10886 |
"trial_name": null,
|
| 10887 |
"trial_params": null
|
| 10888 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 1.9946362972259521,
|
| 3 |
+
"best_model_checkpoint": "output/checkpoint-2000",
|
| 4 |
+
"epoch": 1.3673679768829352,
|
| 5 |
+
"global_step": 2000,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 10878 |
"eval_samples_per_second": 6.061,
|
| 10879 |
"eval_steps_per_second": 6.061,
|
| 10880 |
"step": 1800
|
| 10881 |
+
},
|
| 10882 |
+
{
|
| 10883 |
+
"epoch": 1.23,
|
| 10884 |
+
"learning_rate": 0.0003705207835642618,
|
| 10885 |
+
"loss": 1.9406,
|
| 10886 |
+
"step": 1801
|
| 10887 |
+
},
|
| 10888 |
+
{
|
| 10889 |
+
"epoch": 1.23,
|
| 10890 |
+
"learning_rate": 0.00037037744863831814,
|
| 10891 |
+
"loss": 2.164,
|
| 10892 |
+
"step": 1802
|
| 10893 |
+
},
|
| 10894 |
+
{
|
| 10895 |
+
"epoch": 1.23,
|
| 10896 |
+
"learning_rate": 0.0003702341137123746,
|
| 10897 |
+
"loss": 2.0001,
|
| 10898 |
+
"step": 1803
|
| 10899 |
+
},
|
| 10900 |
+
{
|
| 10901 |
+
"epoch": 1.23,
|
| 10902 |
+
"learning_rate": 0.0003700907787864309,
|
| 10903 |
+
"loss": 1.9451,
|
| 10904 |
+
"step": 1804
|
| 10905 |
+
},
|
| 10906 |
+
{
|
| 10907 |
+
"epoch": 1.23,
|
| 10908 |
+
"learning_rate": 0.00036994744386048735,
|
| 10909 |
+
"loss": 2.003,
|
| 10910 |
+
"step": 1805
|
| 10911 |
+
},
|
| 10912 |
+
{
|
| 10913 |
+
"epoch": 1.23,
|
| 10914 |
+
"learning_rate": 0.0003698041089345437,
|
| 10915 |
+
"loss": 2.0272,
|
| 10916 |
+
"step": 1806
|
| 10917 |
+
},
|
| 10918 |
+
{
|
| 10919 |
+
"epoch": 1.24,
|
| 10920 |
+
"learning_rate": 0.00036966077400860006,
|
| 10921 |
+
"loss": 1.9824,
|
| 10922 |
+
"step": 1807
|
| 10923 |
+
},
|
| 10924 |
+
{
|
| 10925 |
+
"epoch": 1.24,
|
| 10926 |
+
"learning_rate": 0.00036951743908265644,
|
| 10927 |
+
"loss": 2.0841,
|
| 10928 |
+
"step": 1808
|
| 10929 |
+
},
|
| 10930 |
+
{
|
| 10931 |
+
"epoch": 1.24,
|
| 10932 |
+
"learning_rate": 0.00036937410415671283,
|
| 10933 |
+
"loss": 2.0329,
|
| 10934 |
+
"step": 1809
|
| 10935 |
+
},
|
| 10936 |
+
{
|
| 10937 |
+
"epoch": 1.24,
|
| 10938 |
+
"learning_rate": 0.0003692307692307692,
|
| 10939 |
+
"loss": 2.0209,
|
| 10940 |
+
"step": 1810
|
| 10941 |
+
},
|
| 10942 |
+
{
|
| 10943 |
+
"epoch": 1.24,
|
| 10944 |
+
"learning_rate": 0.0003690874343048256,
|
| 10945 |
+
"loss": 1.8782,
|
| 10946 |
+
"step": 1811
|
| 10947 |
+
},
|
| 10948 |
+
{
|
| 10949 |
+
"epoch": 1.24,
|
| 10950 |
+
"learning_rate": 0.000368944099378882,
|
| 10951 |
+
"loss": 1.9467,
|
| 10952 |
+
"step": 1812
|
| 10953 |
+
},
|
| 10954 |
+
{
|
| 10955 |
+
"epoch": 1.24,
|
| 10956 |
+
"learning_rate": 0.00036880076445293836,
|
| 10957 |
+
"loss": 1.915,
|
| 10958 |
+
"step": 1813
|
| 10959 |
+
},
|
| 10960 |
+
{
|
| 10961 |
+
"epoch": 1.24,
|
| 10962 |
+
"learning_rate": 0.0003686574295269947,
|
| 10963 |
+
"loss": 2.019,
|
| 10964 |
+
"step": 1814
|
| 10965 |
+
},
|
| 10966 |
+
{
|
| 10967 |
+
"epoch": 1.24,
|
| 10968 |
+
"learning_rate": 0.0003685140946010511,
|
| 10969 |
+
"loss": 2.0315,
|
| 10970 |
+
"step": 1815
|
| 10971 |
+
},
|
| 10972 |
+
{
|
| 10973 |
+
"epoch": 1.24,
|
| 10974 |
+
"learning_rate": 0.00036837075967510746,
|
| 10975 |
+
"loss": 2.0391,
|
| 10976 |
+
"step": 1816
|
| 10977 |
+
},
|
| 10978 |
+
{
|
| 10979 |
+
"epoch": 1.24,
|
| 10980 |
+
"learning_rate": 0.00036822742474916384,
|
| 10981 |
+
"loss": 1.9748,
|
| 10982 |
+
"step": 1817
|
| 10983 |
+
},
|
| 10984 |
+
{
|
| 10985 |
+
"epoch": 1.24,
|
| 10986 |
+
"learning_rate": 0.0003680840898232202,
|
| 10987 |
+
"loss": 2.0068,
|
| 10988 |
+
"step": 1818
|
| 10989 |
+
},
|
| 10990 |
+
{
|
| 10991 |
+
"epoch": 1.24,
|
| 10992 |
+
"learning_rate": 0.0003679407548972766,
|
| 10993 |
+
"loss": 1.9701,
|
| 10994 |
+
"step": 1819
|
| 10995 |
+
},
|
| 10996 |
+
{
|
| 10997 |
+
"epoch": 1.24,
|
| 10998 |
+
"learning_rate": 0.000367797419971333,
|
| 10999 |
+
"loss": 1.9574,
|
| 11000 |
+
"step": 1820
|
| 11001 |
+
},
|
| 11002 |
+
{
|
| 11003 |
+
"epoch": 1.24,
|
| 11004 |
+
"learning_rate": 0.0003676540850453893,
|
| 11005 |
+
"loss": 2.0035,
|
| 11006 |
+
"step": 1821
|
| 11007 |
+
},
|
| 11008 |
+
{
|
| 11009 |
+
"epoch": 1.25,
|
| 11010 |
+
"learning_rate": 0.00036751075011944576,
|
| 11011 |
+
"loss": 1.9546,
|
| 11012 |
+
"step": 1822
|
| 11013 |
+
},
|
| 11014 |
+
{
|
| 11015 |
+
"epoch": 1.25,
|
| 11016 |
+
"learning_rate": 0.0003673674151935021,
|
| 11017 |
+
"loss": 1.9779,
|
| 11018 |
+
"step": 1823
|
| 11019 |
+
},
|
| 11020 |
+
{
|
| 11021 |
+
"epoch": 1.25,
|
| 11022 |
+
"learning_rate": 0.00036722408026755853,
|
| 11023 |
+
"loss": 2.1025,
|
| 11024 |
+
"step": 1824
|
| 11025 |
+
},
|
| 11026 |
+
{
|
| 11027 |
+
"epoch": 1.25,
|
| 11028 |
+
"learning_rate": 0.00036708074534161486,
|
| 11029 |
+
"loss": 2.028,
|
| 11030 |
+
"step": 1825
|
| 11031 |
+
},
|
| 11032 |
+
{
|
| 11033 |
+
"epoch": 1.25,
|
| 11034 |
+
"learning_rate": 0.0003669374104156713,
|
| 11035 |
+
"loss": 1.9185,
|
| 11036 |
+
"step": 1826
|
| 11037 |
+
},
|
| 11038 |
+
{
|
| 11039 |
+
"epoch": 1.25,
|
| 11040 |
+
"learning_rate": 0.0003667940754897276,
|
| 11041 |
+
"loss": 1.865,
|
| 11042 |
+
"step": 1827
|
| 11043 |
+
},
|
| 11044 |
+
{
|
| 11045 |
+
"epoch": 1.25,
|
| 11046 |
+
"learning_rate": 0.000366650740563784,
|
| 11047 |
+
"loss": 1.9609,
|
| 11048 |
+
"step": 1828
|
| 11049 |
+
},
|
| 11050 |
+
{
|
| 11051 |
+
"epoch": 1.25,
|
| 11052 |
+
"learning_rate": 0.0003665074056378404,
|
| 11053 |
+
"loss": 1.9569,
|
| 11054 |
+
"step": 1829
|
| 11055 |
+
},
|
| 11056 |
+
{
|
| 11057 |
+
"epoch": 1.25,
|
| 11058 |
+
"learning_rate": 0.0003663640707118968,
|
| 11059 |
+
"loss": 2.0257,
|
| 11060 |
+
"step": 1830
|
| 11061 |
+
},
|
| 11062 |
+
{
|
| 11063 |
+
"epoch": 1.25,
|
| 11064 |
+
"learning_rate": 0.00036622073578595316,
|
| 11065 |
+
"loss": 2.0884,
|
| 11066 |
+
"step": 1831
|
| 11067 |
+
},
|
| 11068 |
+
{
|
| 11069 |
+
"epoch": 1.25,
|
| 11070 |
+
"learning_rate": 0.0003660774008600095,
|
| 11071 |
+
"loss": 2.0377,
|
| 11072 |
+
"step": 1832
|
| 11073 |
+
},
|
| 11074 |
+
{
|
| 11075 |
+
"epoch": 1.25,
|
| 11076 |
+
"learning_rate": 0.00036593406593406593,
|
| 11077 |
+
"loss": 1.9751,
|
| 11078 |
+
"step": 1833
|
| 11079 |
+
},
|
| 11080 |
+
{
|
| 11081 |
+
"epoch": 1.25,
|
| 11082 |
+
"learning_rate": 0.00036579073100812226,
|
| 11083 |
+
"loss": 2.0617,
|
| 11084 |
+
"step": 1834
|
| 11085 |
+
},
|
| 11086 |
+
{
|
| 11087 |
+
"epoch": 1.25,
|
| 11088 |
+
"learning_rate": 0.00036564739608217864,
|
| 11089 |
+
"loss": 2.0659,
|
| 11090 |
+
"step": 1835
|
| 11091 |
+
},
|
| 11092 |
+
{
|
| 11093 |
+
"epoch": 1.26,
|
| 11094 |
+
"learning_rate": 0.000365504061156235,
|
| 11095 |
+
"loss": 1.9916,
|
| 11096 |
+
"step": 1836
|
| 11097 |
+
},
|
| 11098 |
+
{
|
| 11099 |
+
"epoch": 1.26,
|
| 11100 |
+
"learning_rate": 0.0003653607262302914,
|
| 11101 |
+
"loss": 2.0324,
|
| 11102 |
+
"step": 1837
|
| 11103 |
+
},
|
| 11104 |
+
{
|
| 11105 |
+
"epoch": 1.26,
|
| 11106 |
+
"learning_rate": 0.0003652173913043478,
|
| 11107 |
+
"loss": 2.0329,
|
| 11108 |
+
"step": 1838
|
| 11109 |
+
},
|
| 11110 |
+
{
|
| 11111 |
+
"epoch": 1.26,
|
| 11112 |
+
"learning_rate": 0.0003650740563784042,
|
| 11113 |
+
"loss": 1.9504,
|
| 11114 |
+
"step": 1839
|
| 11115 |
+
},
|
| 11116 |
+
{
|
| 11117 |
+
"epoch": 1.26,
|
| 11118 |
+
"learning_rate": 0.00036493072145246056,
|
| 11119 |
+
"loss": 1.8959,
|
| 11120 |
+
"step": 1840
|
| 11121 |
+
},
|
| 11122 |
+
{
|
| 11123 |
+
"epoch": 1.26,
|
| 11124 |
+
"learning_rate": 0.00036478738652651695,
|
| 11125 |
+
"loss": 2.0087,
|
| 11126 |
+
"step": 1841
|
| 11127 |
+
},
|
| 11128 |
+
{
|
| 11129 |
+
"epoch": 1.26,
|
| 11130 |
+
"learning_rate": 0.0003646440516005733,
|
| 11131 |
+
"loss": 2.1006,
|
| 11132 |
+
"step": 1842
|
| 11133 |
+
},
|
| 11134 |
+
{
|
| 11135 |
+
"epoch": 1.26,
|
| 11136 |
+
"learning_rate": 0.0003645007166746297,
|
| 11137 |
+
"loss": 2.0928,
|
| 11138 |
+
"step": 1843
|
| 11139 |
+
},
|
| 11140 |
+
{
|
| 11141 |
+
"epoch": 1.26,
|
| 11142 |
+
"learning_rate": 0.00036435738174868604,
|
| 11143 |
+
"loss": 2.0526,
|
| 11144 |
+
"step": 1844
|
| 11145 |
+
},
|
| 11146 |
+
{
|
| 11147 |
+
"epoch": 1.26,
|
| 11148 |
+
"learning_rate": 0.0003642140468227425,
|
| 11149 |
+
"loss": 1.9657,
|
| 11150 |
+
"step": 1845
|
| 11151 |
+
},
|
| 11152 |
+
{
|
| 11153 |
+
"epoch": 1.26,
|
| 11154 |
+
"learning_rate": 0.0003640707118967988,
|
| 11155 |
+
"loss": 1.9081,
|
| 11156 |
+
"step": 1846
|
| 11157 |
+
},
|
| 11158 |
+
{
|
| 11159 |
+
"epoch": 1.26,
|
| 11160 |
+
"learning_rate": 0.00036392737697085514,
|
| 11161 |
+
"loss": 2.007,
|
| 11162 |
+
"step": 1847
|
| 11163 |
+
},
|
| 11164 |
+
{
|
| 11165 |
+
"epoch": 1.26,
|
| 11166 |
+
"learning_rate": 0.0003637840420449116,
|
| 11167 |
+
"loss": 2.0273,
|
| 11168 |
+
"step": 1848
|
| 11169 |
+
},
|
| 11170 |
+
{
|
| 11171 |
+
"epoch": 1.26,
|
| 11172 |
+
"learning_rate": 0.0003636407071189679,
|
| 11173 |
+
"loss": 1.9917,
|
| 11174 |
+
"step": 1849
|
| 11175 |
+
},
|
| 11176 |
+
{
|
| 11177 |
+
"epoch": 1.26,
|
| 11178 |
+
"learning_rate": 0.00036349737219302435,
|
| 11179 |
+
"loss": 1.9588,
|
| 11180 |
+
"step": 1850
|
| 11181 |
+
},
|
| 11182 |
+
{
|
| 11183 |
+
"epoch": 1.27,
|
| 11184 |
+
"learning_rate": 0.0003633540372670807,
|
| 11185 |
+
"loss": 1.9464,
|
| 11186 |
+
"step": 1851
|
| 11187 |
+
},
|
| 11188 |
+
{
|
| 11189 |
+
"epoch": 1.27,
|
| 11190 |
+
"learning_rate": 0.0003632107023411371,
|
| 11191 |
+
"loss": 1.9609,
|
| 11192 |
+
"step": 1852
|
| 11193 |
+
},
|
| 11194 |
+
{
|
| 11195 |
+
"epoch": 1.27,
|
| 11196 |
+
"learning_rate": 0.00036306736741519344,
|
| 11197 |
+
"loss": 1.9257,
|
| 11198 |
+
"step": 1853
|
| 11199 |
+
},
|
| 11200 |
+
{
|
| 11201 |
+
"epoch": 1.27,
|
| 11202 |
+
"learning_rate": 0.0003629240324892498,
|
| 11203 |
+
"loss": 2.0676,
|
| 11204 |
+
"step": 1854
|
| 11205 |
+
},
|
| 11206 |
+
{
|
| 11207 |
+
"epoch": 1.27,
|
| 11208 |
+
"learning_rate": 0.0003627806975633062,
|
| 11209 |
+
"loss": 2.0322,
|
| 11210 |
+
"step": 1855
|
| 11211 |
+
},
|
| 11212 |
+
{
|
| 11213 |
+
"epoch": 1.27,
|
| 11214 |
+
"learning_rate": 0.0003626373626373626,
|
| 11215 |
+
"loss": 2.023,
|
| 11216 |
+
"step": 1856
|
| 11217 |
+
},
|
| 11218 |
+
{
|
| 11219 |
+
"epoch": 1.27,
|
| 11220 |
+
"learning_rate": 0.000362494027711419,
|
| 11221 |
+
"loss": 1.9282,
|
| 11222 |
+
"step": 1857
|
| 11223 |
+
},
|
| 11224 |
+
{
|
| 11225 |
+
"epoch": 1.27,
|
| 11226 |
+
"learning_rate": 0.00036235069278547536,
|
| 11227 |
+
"loss": 2.0074,
|
| 11228 |
+
"step": 1858
|
| 11229 |
+
},
|
| 11230 |
+
{
|
| 11231 |
+
"epoch": 1.27,
|
| 11232 |
+
"learning_rate": 0.00036220735785953175,
|
| 11233 |
+
"loss": 2.037,
|
| 11234 |
+
"step": 1859
|
| 11235 |
+
},
|
| 11236 |
+
{
|
| 11237 |
+
"epoch": 1.27,
|
| 11238 |
+
"learning_rate": 0.00036206402293358813,
|
| 11239 |
+
"loss": 1.9538,
|
| 11240 |
+
"step": 1860
|
| 11241 |
+
},
|
| 11242 |
+
{
|
| 11243 |
+
"epoch": 1.27,
|
| 11244 |
+
"learning_rate": 0.00036192068800764446,
|
| 11245 |
+
"loss": 1.852,
|
| 11246 |
+
"step": 1861
|
| 11247 |
+
},
|
| 11248 |
+
{
|
| 11249 |
+
"epoch": 1.27,
|
| 11250 |
+
"learning_rate": 0.0003617773530817009,
|
| 11251 |
+
"loss": 2.0184,
|
| 11252 |
+
"step": 1862
|
| 11253 |
+
},
|
| 11254 |
+
{
|
| 11255 |
+
"epoch": 1.27,
|
| 11256 |
+
"learning_rate": 0.0003616340181557572,
|
| 11257 |
+
"loss": 2.0041,
|
| 11258 |
+
"step": 1863
|
| 11259 |
+
},
|
| 11260 |
+
{
|
| 11261 |
+
"epoch": 1.27,
|
| 11262 |
+
"learning_rate": 0.00036149068322981366,
|
| 11263 |
+
"loss": 1.988,
|
| 11264 |
+
"step": 1864
|
| 11265 |
+
},
|
| 11266 |
+
{
|
| 11267 |
+
"epoch": 1.28,
|
| 11268 |
+
"learning_rate": 0.00036134734830387,
|
| 11269 |
+
"loss": 2.0504,
|
| 11270 |
+
"step": 1865
|
| 11271 |
+
},
|
| 11272 |
+
{
|
| 11273 |
+
"epoch": 1.28,
|
| 11274 |
+
"learning_rate": 0.00036120401337792643,
|
| 11275 |
+
"loss": 2.0778,
|
| 11276 |
+
"step": 1866
|
| 11277 |
+
},
|
| 11278 |
+
{
|
| 11279 |
+
"epoch": 1.28,
|
| 11280 |
+
"learning_rate": 0.00036106067845198276,
|
| 11281 |
+
"loss": 2.066,
|
| 11282 |
+
"step": 1867
|
| 11283 |
+
},
|
| 11284 |
+
{
|
| 11285 |
+
"epoch": 1.28,
|
| 11286 |
+
"learning_rate": 0.0003609173435260391,
|
| 11287 |
+
"loss": 2.0418,
|
| 11288 |
+
"step": 1868
|
| 11289 |
+
},
|
| 11290 |
+
{
|
| 11291 |
+
"epoch": 1.28,
|
| 11292 |
+
"learning_rate": 0.00036077400860009553,
|
| 11293 |
+
"loss": 2.0154,
|
| 11294 |
+
"step": 1869
|
| 11295 |
+
},
|
| 11296 |
+
{
|
| 11297 |
+
"epoch": 1.28,
|
| 11298 |
+
"learning_rate": 0.00036063067367415186,
|
| 11299 |
+
"loss": 1.9723,
|
| 11300 |
+
"step": 1870
|
| 11301 |
+
},
|
| 11302 |
+
{
|
| 11303 |
+
"epoch": 1.28,
|
| 11304 |
+
"learning_rate": 0.0003604873387482083,
|
| 11305 |
+
"loss": 1.9196,
|
| 11306 |
+
"step": 1871
|
| 11307 |
+
},
|
| 11308 |
+
{
|
| 11309 |
+
"epoch": 1.28,
|
| 11310 |
+
"learning_rate": 0.0003603440038222646,
|
| 11311 |
+
"loss": 2.0374,
|
| 11312 |
+
"step": 1872
|
| 11313 |
+
},
|
| 11314 |
+
{
|
| 11315 |
+
"epoch": 1.28,
|
| 11316 |
+
"learning_rate": 0.00036020066889632106,
|
| 11317 |
+
"loss": 2.1146,
|
| 11318 |
+
"step": 1873
|
| 11319 |
+
},
|
| 11320 |
+
{
|
| 11321 |
+
"epoch": 1.28,
|
| 11322 |
+
"learning_rate": 0.0003600573339703774,
|
| 11323 |
+
"loss": 2.0713,
|
| 11324 |
+
"step": 1874
|
| 11325 |
+
},
|
| 11326 |
+
{
|
| 11327 |
+
"epoch": 1.28,
|
| 11328 |
+
"learning_rate": 0.0003599139990444338,
|
| 11329 |
+
"loss": 2.0524,
|
| 11330 |
+
"step": 1875
|
| 11331 |
+
},
|
| 11332 |
+
{
|
| 11333 |
+
"epoch": 1.28,
|
| 11334 |
+
"learning_rate": 0.00035977066411849016,
|
| 11335 |
+
"loss": 1.9778,
|
| 11336 |
+
"step": 1876
|
| 11337 |
+
},
|
| 11338 |
+
{
|
| 11339 |
+
"epoch": 1.28,
|
| 11340 |
+
"learning_rate": 0.00035962732919254654,
|
| 11341 |
+
"loss": 2.0138,
|
| 11342 |
+
"step": 1877
|
| 11343 |
+
},
|
| 11344 |
+
{
|
| 11345 |
+
"epoch": 1.28,
|
| 11346 |
+
"learning_rate": 0.00035948399426660293,
|
| 11347 |
+
"loss": 1.9767,
|
| 11348 |
+
"step": 1878
|
| 11349 |
+
},
|
| 11350 |
+
{
|
| 11351 |
+
"epoch": 1.28,
|
| 11352 |
+
"learning_rate": 0.0003593406593406593,
|
| 11353 |
+
"loss": 1.9536,
|
| 11354 |
+
"step": 1879
|
| 11355 |
+
},
|
| 11356 |
+
{
|
| 11357 |
+
"epoch": 1.29,
|
| 11358 |
+
"learning_rate": 0.0003591973244147157,
|
| 11359 |
+
"loss": 2.1092,
|
| 11360 |
+
"step": 1880
|
| 11361 |
+
},
|
| 11362 |
+
{
|
| 11363 |
+
"epoch": 1.29,
|
| 11364 |
+
"learning_rate": 0.0003590539894887721,
|
| 11365 |
+
"loss": 1.9906,
|
| 11366 |
+
"step": 1881
|
| 11367 |
+
},
|
| 11368 |
+
{
|
| 11369 |
+
"epoch": 1.29,
|
| 11370 |
+
"learning_rate": 0.0003589106545628284,
|
| 11371 |
+
"loss": 2.0383,
|
| 11372 |
+
"step": 1882
|
| 11373 |
+
},
|
| 11374 |
+
{
|
| 11375 |
+
"epoch": 1.29,
|
| 11376 |
+
"learning_rate": 0.00035876731963688485,
|
| 11377 |
+
"loss": 1.9626,
|
| 11378 |
+
"step": 1883
|
| 11379 |
+
},
|
| 11380 |
+
{
|
| 11381 |
+
"epoch": 1.29,
|
| 11382 |
+
"learning_rate": 0.0003586239847109412,
|
| 11383 |
+
"loss": 2.0121,
|
| 11384 |
+
"step": 1884
|
| 11385 |
+
},
|
| 11386 |
+
{
|
| 11387 |
+
"epoch": 1.29,
|
| 11388 |
+
"learning_rate": 0.0003584806497849976,
|
| 11389 |
+
"loss": 2.0007,
|
| 11390 |
+
"step": 1885
|
| 11391 |
+
},
|
| 11392 |
+
{
|
| 11393 |
+
"epoch": 1.29,
|
| 11394 |
+
"learning_rate": 0.00035833731485905394,
|
| 11395 |
+
"loss": 2.0846,
|
| 11396 |
+
"step": 1886
|
| 11397 |
+
},
|
| 11398 |
+
{
|
| 11399 |
+
"epoch": 1.29,
|
| 11400 |
+
"learning_rate": 0.0003581939799331104,
|
| 11401 |
+
"loss": 2.1493,
|
| 11402 |
+
"step": 1887
|
| 11403 |
+
},
|
| 11404 |
+
{
|
| 11405 |
+
"epoch": 1.29,
|
| 11406 |
+
"learning_rate": 0.0003580506450071667,
|
| 11407 |
+
"loss": 1.9866,
|
| 11408 |
+
"step": 1888
|
| 11409 |
+
},
|
| 11410 |
+
{
|
| 11411 |
+
"epoch": 1.29,
|
| 11412 |
+
"learning_rate": 0.00035790731008122304,
|
| 11413 |
+
"loss": 1.9876,
|
| 11414 |
+
"step": 1889
|
| 11415 |
+
},
|
| 11416 |
+
{
|
| 11417 |
+
"epoch": 1.29,
|
| 11418 |
+
"learning_rate": 0.0003577639751552795,
|
| 11419 |
+
"loss": 2.0256,
|
| 11420 |
+
"step": 1890
|
| 11421 |
+
},
|
| 11422 |
+
{
|
| 11423 |
+
"epoch": 1.29,
|
| 11424 |
+
"learning_rate": 0.0003576206402293358,
|
| 11425 |
+
"loss": 2.0551,
|
| 11426 |
+
"step": 1891
|
| 11427 |
+
},
|
| 11428 |
+
{
|
| 11429 |
+
"epoch": 1.29,
|
| 11430 |
+
"learning_rate": 0.00035747730530339225,
|
| 11431 |
+
"loss": 1.9491,
|
| 11432 |
+
"step": 1892
|
| 11433 |
+
},
|
| 11434 |
+
{
|
| 11435 |
+
"epoch": 1.29,
|
| 11436 |
+
"learning_rate": 0.0003573339703774486,
|
| 11437 |
+
"loss": 1.976,
|
| 11438 |
+
"step": 1893
|
| 11439 |
+
},
|
| 11440 |
+
{
|
| 11441 |
+
"epoch": 1.29,
|
| 11442 |
+
"learning_rate": 0.000357190635451505,
|
| 11443 |
+
"loss": 2.0242,
|
| 11444 |
+
"step": 1894
|
| 11445 |
+
},
|
| 11446 |
+
{
|
| 11447 |
+
"epoch": 1.3,
|
| 11448 |
+
"learning_rate": 0.00035704730052556134,
|
| 11449 |
+
"loss": 1.9942,
|
| 11450 |
+
"step": 1895
|
| 11451 |
+
},
|
| 11452 |
+
{
|
| 11453 |
+
"epoch": 1.3,
|
| 11454 |
+
"learning_rate": 0.00035690396559961773,
|
| 11455 |
+
"loss": 1.9462,
|
| 11456 |
+
"step": 1896
|
| 11457 |
+
},
|
| 11458 |
+
{
|
| 11459 |
+
"epoch": 1.3,
|
| 11460 |
+
"learning_rate": 0.0003567606306736741,
|
| 11461 |
+
"loss": 1.9941,
|
| 11462 |
+
"step": 1897
|
| 11463 |
+
},
|
| 11464 |
+
{
|
| 11465 |
+
"epoch": 1.3,
|
| 11466 |
+
"learning_rate": 0.0003566172957477305,
|
| 11467 |
+
"loss": 1.987,
|
| 11468 |
+
"step": 1898
|
| 11469 |
+
},
|
| 11470 |
+
{
|
| 11471 |
+
"epoch": 1.3,
|
| 11472 |
+
"learning_rate": 0.0003564739608217869,
|
| 11473 |
+
"loss": 1.9921,
|
| 11474 |
+
"step": 1899
|
| 11475 |
+
},
|
| 11476 |
+
{
|
| 11477 |
+
"epoch": 1.3,
|
| 11478 |
+
"learning_rate": 0.00035633062589584326,
|
| 11479 |
+
"loss": 1.9913,
|
| 11480 |
+
"step": 1900
|
| 11481 |
+
},
|
| 11482 |
+
{
|
| 11483 |
+
"epoch": 1.3,
|
| 11484 |
+
"learning_rate": 0.00035618729096989965,
|
| 11485 |
+
"loss": 2.0761,
|
| 11486 |
+
"step": 1901
|
| 11487 |
+
},
|
| 11488 |
+
{
|
| 11489 |
+
"epoch": 1.3,
|
| 11490 |
+
"learning_rate": 0.00035604395604395603,
|
| 11491 |
+
"loss": 2.0225,
|
| 11492 |
+
"step": 1902
|
| 11493 |
+
},
|
| 11494 |
+
{
|
| 11495 |
+
"epoch": 1.3,
|
| 11496 |
+
"learning_rate": 0.00035590062111801236,
|
| 11497 |
+
"loss": 2.1646,
|
| 11498 |
+
"step": 1903
|
| 11499 |
+
},
|
| 11500 |
+
{
|
| 11501 |
+
"epoch": 1.3,
|
| 11502 |
+
"learning_rate": 0.0003557572861920688,
|
| 11503 |
+
"loss": 1.9713,
|
| 11504 |
+
"step": 1904
|
| 11505 |
+
},
|
| 11506 |
+
{
|
| 11507 |
+
"epoch": 1.3,
|
| 11508 |
+
"learning_rate": 0.00035561395126612513,
|
| 11509 |
+
"loss": 2.0381,
|
| 11510 |
+
"step": 1905
|
| 11511 |
+
},
|
| 11512 |
+
{
|
| 11513 |
+
"epoch": 1.3,
|
| 11514 |
+
"learning_rate": 0.00035547061634018157,
|
| 11515 |
+
"loss": 2.1288,
|
| 11516 |
+
"step": 1906
|
| 11517 |
+
},
|
| 11518 |
+
{
|
| 11519 |
+
"epoch": 1.3,
|
| 11520 |
+
"learning_rate": 0.0003553272814142379,
|
| 11521 |
+
"loss": 1.9958,
|
| 11522 |
+
"step": 1907
|
| 11523 |
+
},
|
| 11524 |
+
{
|
| 11525 |
+
"epoch": 1.3,
|
| 11526 |
+
"learning_rate": 0.00035518394648829433,
|
| 11527 |
+
"loss": 2.0927,
|
| 11528 |
+
"step": 1908
|
| 11529 |
+
},
|
| 11530 |
+
{
|
| 11531 |
+
"epoch": 1.31,
|
| 11532 |
+
"learning_rate": 0.00035504061156235066,
|
| 11533 |
+
"loss": 2.0493,
|
| 11534 |
+
"step": 1909
|
| 11535 |
+
},
|
| 11536 |
+
{
|
| 11537 |
+
"epoch": 1.31,
|
| 11538 |
+
"learning_rate": 0.000354897276636407,
|
| 11539 |
+
"loss": 1.9751,
|
| 11540 |
+
"step": 1910
|
| 11541 |
+
},
|
| 11542 |
+
{
|
| 11543 |
+
"epoch": 1.31,
|
| 11544 |
+
"learning_rate": 0.00035475394171046343,
|
| 11545 |
+
"loss": 1.9947,
|
| 11546 |
+
"step": 1911
|
| 11547 |
+
},
|
| 11548 |
+
{
|
| 11549 |
+
"epoch": 1.31,
|
| 11550 |
+
"learning_rate": 0.00035461060678451976,
|
| 11551 |
+
"loss": 1.9878,
|
| 11552 |
+
"step": 1912
|
| 11553 |
+
},
|
| 11554 |
+
{
|
| 11555 |
+
"epoch": 1.31,
|
| 11556 |
+
"learning_rate": 0.0003544672718585762,
|
| 11557 |
+
"loss": 2.0322,
|
| 11558 |
+
"step": 1913
|
| 11559 |
+
},
|
| 11560 |
+
{
|
| 11561 |
+
"epoch": 1.31,
|
| 11562 |
+
"learning_rate": 0.00035432393693263253,
|
| 11563 |
+
"loss": 1.9825,
|
| 11564 |
+
"step": 1914
|
| 11565 |
+
},
|
| 11566 |
+
{
|
| 11567 |
+
"epoch": 1.31,
|
| 11568 |
+
"learning_rate": 0.00035418060200668897,
|
| 11569 |
+
"loss": 2.0174,
|
| 11570 |
+
"step": 1915
|
| 11571 |
+
},
|
| 11572 |
+
{
|
| 11573 |
+
"epoch": 1.31,
|
| 11574 |
+
"learning_rate": 0.0003540372670807453,
|
| 11575 |
+
"loss": 1.9631,
|
| 11576 |
+
"step": 1916
|
| 11577 |
+
},
|
| 11578 |
+
{
|
| 11579 |
+
"epoch": 1.31,
|
| 11580 |
+
"learning_rate": 0.0003538939321548017,
|
| 11581 |
+
"loss": 1.9678,
|
| 11582 |
+
"step": 1917
|
| 11583 |
+
},
|
| 11584 |
+
{
|
| 11585 |
+
"epoch": 1.31,
|
| 11586 |
+
"learning_rate": 0.00035375059722885806,
|
| 11587 |
+
"loss": 2.0006,
|
| 11588 |
+
"step": 1918
|
| 11589 |
+
},
|
| 11590 |
+
{
|
| 11591 |
+
"epoch": 1.31,
|
| 11592 |
+
"learning_rate": 0.00035360726230291445,
|
| 11593 |
+
"loss": 1.9347,
|
| 11594 |
+
"step": 1919
|
| 11595 |
+
},
|
| 11596 |
+
{
|
| 11597 |
+
"epoch": 1.31,
|
| 11598 |
+
"learning_rate": 0.00035346392737697083,
|
| 11599 |
+
"loss": 2.0409,
|
| 11600 |
+
"step": 1920
|
| 11601 |
+
},
|
| 11602 |
+
{
|
| 11603 |
+
"epoch": 1.31,
|
| 11604 |
+
"learning_rate": 0.0003533205924510272,
|
| 11605 |
+
"loss": 2.0187,
|
| 11606 |
+
"step": 1921
|
| 11607 |
+
},
|
| 11608 |
+
{
|
| 11609 |
+
"epoch": 1.31,
|
| 11610 |
+
"learning_rate": 0.0003531772575250836,
|
| 11611 |
+
"loss": 1.9847,
|
| 11612 |
+
"step": 1922
|
| 11613 |
+
},
|
| 11614 |
+
{
|
| 11615 |
+
"epoch": 1.31,
|
| 11616 |
+
"learning_rate": 0.00035303392259914,
|
| 11617 |
+
"loss": 2.0057,
|
| 11618 |
+
"step": 1923
|
| 11619 |
+
},
|
| 11620 |
+
{
|
| 11621 |
+
"epoch": 1.32,
|
| 11622 |
+
"learning_rate": 0.0003528905876731963,
|
| 11623 |
+
"loss": 2.0196,
|
| 11624 |
+
"step": 1924
|
| 11625 |
+
},
|
| 11626 |
+
{
|
| 11627 |
+
"epoch": 1.32,
|
| 11628 |
+
"learning_rate": 0.00035274725274725275,
|
| 11629 |
+
"loss": 1.8575,
|
| 11630 |
+
"step": 1925
|
| 11631 |
+
},
|
| 11632 |
+
{
|
| 11633 |
+
"epoch": 1.32,
|
| 11634 |
+
"learning_rate": 0.0003526039178213091,
|
| 11635 |
+
"loss": 1.9818,
|
| 11636 |
+
"step": 1926
|
| 11637 |
+
},
|
| 11638 |
+
{
|
| 11639 |
+
"epoch": 1.32,
|
| 11640 |
+
"learning_rate": 0.0003524605828953655,
|
| 11641 |
+
"loss": 2.0204,
|
| 11642 |
+
"step": 1927
|
| 11643 |
+
},
|
| 11644 |
+
{
|
| 11645 |
+
"epoch": 1.32,
|
| 11646 |
+
"learning_rate": 0.00035231724796942185,
|
| 11647 |
+
"loss": 2.0446,
|
| 11648 |
+
"step": 1928
|
| 11649 |
+
},
|
| 11650 |
+
{
|
| 11651 |
+
"epoch": 1.32,
|
| 11652 |
+
"learning_rate": 0.0003521739130434783,
|
| 11653 |
+
"loss": 2.0144,
|
| 11654 |
+
"step": 1929
|
| 11655 |
+
},
|
| 11656 |
+
{
|
| 11657 |
+
"epoch": 1.32,
|
| 11658 |
+
"learning_rate": 0.0003520305781175346,
|
| 11659 |
+
"loss": 1.9512,
|
| 11660 |
+
"step": 1930
|
| 11661 |
+
},
|
| 11662 |
+
{
|
| 11663 |
+
"epoch": 1.32,
|
| 11664 |
+
"learning_rate": 0.00035188724319159094,
|
| 11665 |
+
"loss": 1.9937,
|
| 11666 |
+
"step": 1931
|
| 11667 |
+
},
|
| 11668 |
+
{
|
| 11669 |
+
"epoch": 1.32,
|
| 11670 |
+
"learning_rate": 0.0003517439082656474,
|
| 11671 |
+
"loss": 1.9763,
|
| 11672 |
+
"step": 1932
|
| 11673 |
+
},
|
| 11674 |
+
{
|
| 11675 |
+
"epoch": 1.32,
|
| 11676 |
+
"learning_rate": 0.0003516005733397037,
|
| 11677 |
+
"loss": 2.0291,
|
| 11678 |
+
"step": 1933
|
| 11679 |
+
},
|
| 11680 |
+
{
|
| 11681 |
+
"epoch": 1.32,
|
| 11682 |
+
"learning_rate": 0.00035145723841376015,
|
| 11683 |
+
"loss": 2.0598,
|
| 11684 |
+
"step": 1934
|
| 11685 |
+
},
|
| 11686 |
+
{
|
| 11687 |
+
"epoch": 1.32,
|
| 11688 |
+
"learning_rate": 0.0003513139034878165,
|
| 11689 |
+
"loss": 2.0737,
|
| 11690 |
+
"step": 1935
|
| 11691 |
+
},
|
| 11692 |
+
{
|
| 11693 |
+
"epoch": 1.32,
|
| 11694 |
+
"learning_rate": 0.0003511705685618729,
|
| 11695 |
+
"loss": 2.1272,
|
| 11696 |
+
"step": 1936
|
| 11697 |
+
},
|
| 11698 |
+
{
|
| 11699 |
+
"epoch": 1.32,
|
| 11700 |
+
"learning_rate": 0.00035102723363592925,
|
| 11701 |
+
"loss": 2.0258,
|
| 11702 |
+
"step": 1937
|
| 11703 |
+
},
|
| 11704 |
+
{
|
| 11705 |
+
"epoch": 1.32,
|
| 11706 |
+
"learning_rate": 0.00035088389870998563,
|
| 11707 |
+
"loss": 2.0811,
|
| 11708 |
+
"step": 1938
|
| 11709 |
+
},
|
| 11710 |
+
{
|
| 11711 |
+
"epoch": 1.33,
|
| 11712 |
+
"learning_rate": 0.000350740563784042,
|
| 11713 |
+
"loss": 1.9712,
|
| 11714 |
+
"step": 1939
|
| 11715 |
+
},
|
| 11716 |
+
{
|
| 11717 |
+
"epoch": 1.33,
|
| 11718 |
+
"learning_rate": 0.0003505972288580984,
|
| 11719 |
+
"loss": 1.9419,
|
| 11720 |
+
"step": 1940
|
| 11721 |
+
},
|
| 11722 |
+
{
|
| 11723 |
+
"epoch": 1.33,
|
| 11724 |
+
"learning_rate": 0.0003504538939321548,
|
| 11725 |
+
"loss": 2.0849,
|
| 11726 |
+
"step": 1941
|
| 11727 |
+
},
|
| 11728 |
+
{
|
| 11729 |
+
"epoch": 1.33,
|
| 11730 |
+
"learning_rate": 0.00035031055900621116,
|
| 11731 |
+
"loss": 2.0123,
|
| 11732 |
+
"step": 1942
|
| 11733 |
+
},
|
| 11734 |
+
{
|
| 11735 |
+
"epoch": 1.33,
|
| 11736 |
+
"learning_rate": 0.00035016722408026755,
|
| 11737 |
+
"loss": 1.9185,
|
| 11738 |
+
"step": 1943
|
| 11739 |
+
},
|
| 11740 |
+
{
|
| 11741 |
+
"epoch": 1.33,
|
| 11742 |
+
"learning_rate": 0.00035002388915432393,
|
| 11743 |
+
"loss": 2.0759,
|
| 11744 |
+
"step": 1944
|
| 11745 |
+
},
|
| 11746 |
+
{
|
| 11747 |
+
"epoch": 1.33,
|
| 11748 |
+
"learning_rate": 0.00034988055422838026,
|
| 11749 |
+
"loss": 2.0272,
|
| 11750 |
+
"step": 1945
|
| 11751 |
+
},
|
| 11752 |
+
{
|
| 11753 |
+
"epoch": 1.33,
|
| 11754 |
+
"learning_rate": 0.0003497372193024367,
|
| 11755 |
+
"loss": 2.0296,
|
| 11756 |
+
"step": 1946
|
| 11757 |
+
},
|
| 11758 |
+
{
|
| 11759 |
+
"epoch": 1.33,
|
| 11760 |
+
"learning_rate": 0.00034959388437649303,
|
| 11761 |
+
"loss": 2.095,
|
| 11762 |
+
"step": 1947
|
| 11763 |
+
},
|
| 11764 |
+
{
|
| 11765 |
+
"epoch": 1.33,
|
| 11766 |
+
"learning_rate": 0.00034945054945054947,
|
| 11767 |
+
"loss": 2.1172,
|
| 11768 |
+
"step": 1948
|
| 11769 |
+
},
|
| 11770 |
+
{
|
| 11771 |
+
"epoch": 1.33,
|
| 11772 |
+
"learning_rate": 0.0003493072145246058,
|
| 11773 |
+
"loss": 1.9911,
|
| 11774 |
+
"step": 1949
|
| 11775 |
+
},
|
| 11776 |
+
{
|
| 11777 |
+
"epoch": 1.33,
|
| 11778 |
+
"learning_rate": 0.0003491638795986621,
|
| 11779 |
+
"loss": 1.9346,
|
| 11780 |
+
"step": 1950
|
| 11781 |
+
},
|
| 11782 |
+
{
|
| 11783 |
+
"epoch": 1.33,
|
| 11784 |
+
"learning_rate": 0.00034902054467271856,
|
| 11785 |
+
"loss": 2.0437,
|
| 11786 |
+
"step": 1951
|
| 11787 |
+
},
|
| 11788 |
+
{
|
| 11789 |
+
"epoch": 1.33,
|
| 11790 |
+
"learning_rate": 0.0003488772097467749,
|
| 11791 |
+
"loss": 1.9926,
|
| 11792 |
+
"step": 1952
|
| 11793 |
+
},
|
| 11794 |
+
{
|
| 11795 |
+
"epoch": 1.34,
|
| 11796 |
+
"learning_rate": 0.00034873387482083133,
|
| 11797 |
+
"loss": 1.8601,
|
| 11798 |
+
"step": 1953
|
| 11799 |
+
},
|
| 11800 |
+
{
|
| 11801 |
+
"epoch": 1.34,
|
| 11802 |
+
"learning_rate": 0.00034859053989488766,
|
| 11803 |
+
"loss": 2.0288,
|
| 11804 |
+
"step": 1954
|
| 11805 |
+
},
|
| 11806 |
+
{
|
| 11807 |
+
"epoch": 1.34,
|
| 11808 |
+
"learning_rate": 0.0003484472049689441,
|
| 11809 |
+
"loss": 1.9644,
|
| 11810 |
+
"step": 1955
|
| 11811 |
+
},
|
| 11812 |
+
{
|
| 11813 |
+
"epoch": 1.34,
|
| 11814 |
+
"learning_rate": 0.00034830387004300043,
|
| 11815 |
+
"loss": 2.0219,
|
| 11816 |
+
"step": 1956
|
| 11817 |
+
},
|
| 11818 |
+
{
|
| 11819 |
+
"epoch": 1.34,
|
| 11820 |
+
"learning_rate": 0.0003481605351170568,
|
| 11821 |
+
"loss": 2.0288,
|
| 11822 |
+
"step": 1957
|
| 11823 |
+
},
|
| 11824 |
+
{
|
| 11825 |
+
"epoch": 1.34,
|
| 11826 |
+
"learning_rate": 0.0003480172001911132,
|
| 11827 |
+
"loss": 2.0549,
|
| 11828 |
+
"step": 1958
|
| 11829 |
+
},
|
| 11830 |
+
{
|
| 11831 |
+
"epoch": 1.34,
|
| 11832 |
+
"learning_rate": 0.0003478738652651696,
|
| 11833 |
+
"loss": 2.0246,
|
| 11834 |
+
"step": 1959
|
| 11835 |
+
},
|
| 11836 |
+
{
|
| 11837 |
+
"epoch": 1.34,
|
| 11838 |
+
"learning_rate": 0.00034773053033922596,
|
| 11839 |
+
"loss": 2.0496,
|
| 11840 |
+
"step": 1960
|
| 11841 |
+
},
|
| 11842 |
+
{
|
| 11843 |
+
"epoch": 1.34,
|
| 11844 |
+
"learning_rate": 0.00034758719541328235,
|
| 11845 |
+
"loss": 1.9183,
|
| 11846 |
+
"step": 1961
|
| 11847 |
+
},
|
| 11848 |
+
{
|
| 11849 |
+
"epoch": 1.34,
|
| 11850 |
+
"learning_rate": 0.00034744386048733873,
|
| 11851 |
+
"loss": 2.0135,
|
| 11852 |
+
"step": 1962
|
| 11853 |
+
},
|
| 11854 |
+
{
|
| 11855 |
+
"epoch": 1.34,
|
| 11856 |
+
"learning_rate": 0.0003473005255613951,
|
| 11857 |
+
"loss": 1.9164,
|
| 11858 |
+
"step": 1963
|
| 11859 |
+
},
|
| 11860 |
+
{
|
| 11861 |
+
"epoch": 1.34,
|
| 11862 |
+
"learning_rate": 0.00034715719063545145,
|
| 11863 |
+
"loss": 2.0452,
|
| 11864 |
+
"step": 1964
|
| 11865 |
+
},
|
| 11866 |
+
{
|
| 11867 |
+
"epoch": 1.34,
|
| 11868 |
+
"learning_rate": 0.0003470138557095079,
|
| 11869 |
+
"loss": 2.1023,
|
| 11870 |
+
"step": 1965
|
| 11871 |
+
},
|
| 11872 |
+
{
|
| 11873 |
+
"epoch": 1.34,
|
| 11874 |
+
"learning_rate": 0.0003468705207835642,
|
| 11875 |
+
"loss": 1.9568,
|
| 11876 |
+
"step": 1966
|
| 11877 |
+
},
|
| 11878 |
+
{
|
| 11879 |
+
"epoch": 1.34,
|
| 11880 |
+
"learning_rate": 0.00034672718585762065,
|
| 11881 |
+
"loss": 2.0042,
|
| 11882 |
+
"step": 1967
|
| 11883 |
+
},
|
| 11884 |
+
{
|
| 11885 |
+
"epoch": 1.35,
|
| 11886 |
+
"learning_rate": 0.000346583850931677,
|
| 11887 |
+
"loss": 1.9825,
|
| 11888 |
+
"step": 1968
|
| 11889 |
+
},
|
| 11890 |
+
{
|
| 11891 |
+
"epoch": 1.35,
|
| 11892 |
+
"learning_rate": 0.0003464405160057334,
|
| 11893 |
+
"loss": 2.0362,
|
| 11894 |
+
"step": 1969
|
| 11895 |
+
},
|
| 11896 |
+
{
|
| 11897 |
+
"epoch": 1.35,
|
| 11898 |
+
"learning_rate": 0.00034629718107978975,
|
| 11899 |
+
"loss": 1.9395,
|
| 11900 |
+
"step": 1970
|
| 11901 |
+
},
|
| 11902 |
+
{
|
| 11903 |
+
"epoch": 1.35,
|
| 11904 |
+
"learning_rate": 0.0003461538461538461,
|
| 11905 |
+
"loss": 1.9586,
|
| 11906 |
+
"step": 1971
|
| 11907 |
+
},
|
| 11908 |
+
{
|
| 11909 |
+
"epoch": 1.35,
|
| 11910 |
+
"learning_rate": 0.0003460105112279025,
|
| 11911 |
+
"loss": 2.0692,
|
| 11912 |
+
"step": 1972
|
| 11913 |
+
},
|
| 11914 |
+
{
|
| 11915 |
+
"epoch": 1.35,
|
| 11916 |
+
"learning_rate": 0.00034586717630195885,
|
| 11917 |
+
"loss": 2.0178,
|
| 11918 |
+
"step": 1973
|
| 11919 |
+
},
|
| 11920 |
+
{
|
| 11921 |
+
"epoch": 1.35,
|
| 11922 |
+
"learning_rate": 0.0003457238413760153,
|
| 11923 |
+
"loss": 1.9349,
|
| 11924 |
+
"step": 1974
|
| 11925 |
+
},
|
| 11926 |
+
{
|
| 11927 |
+
"epoch": 1.35,
|
| 11928 |
+
"learning_rate": 0.0003455805064500716,
|
| 11929 |
+
"loss": 2.0868,
|
| 11930 |
+
"step": 1975
|
| 11931 |
+
},
|
| 11932 |
+
{
|
| 11933 |
+
"epoch": 1.35,
|
| 11934 |
+
"learning_rate": 0.00034543717152412805,
|
| 11935 |
+
"loss": 1.9723,
|
| 11936 |
+
"step": 1976
|
| 11937 |
+
},
|
| 11938 |
+
{
|
| 11939 |
+
"epoch": 1.35,
|
| 11940 |
+
"learning_rate": 0.0003452938365981844,
|
| 11941 |
+
"loss": 2.0752,
|
| 11942 |
+
"step": 1977
|
| 11943 |
+
},
|
| 11944 |
+
{
|
| 11945 |
+
"epoch": 1.35,
|
| 11946 |
+
"learning_rate": 0.00034515050167224076,
|
| 11947 |
+
"loss": 1.9482,
|
| 11948 |
+
"step": 1978
|
| 11949 |
+
},
|
| 11950 |
+
{
|
| 11951 |
+
"epoch": 1.35,
|
| 11952 |
+
"learning_rate": 0.00034500716674629715,
|
| 11953 |
+
"loss": 1.9869,
|
| 11954 |
+
"step": 1979
|
| 11955 |
+
},
|
| 11956 |
+
{
|
| 11957 |
+
"epoch": 1.35,
|
| 11958 |
+
"learning_rate": 0.00034486383182035353,
|
| 11959 |
+
"loss": 2.0023,
|
| 11960 |
+
"step": 1980
|
| 11961 |
+
},
|
| 11962 |
+
{
|
| 11963 |
+
"epoch": 1.35,
|
| 11964 |
+
"learning_rate": 0.0003447204968944099,
|
| 11965 |
+
"loss": 2.0806,
|
| 11966 |
+
"step": 1981
|
| 11967 |
+
},
|
| 11968 |
+
{
|
| 11969 |
+
"epoch": 1.36,
|
| 11970 |
+
"learning_rate": 0.0003445771619684663,
|
| 11971 |
+
"loss": 2.0252,
|
| 11972 |
+
"step": 1982
|
| 11973 |
+
},
|
| 11974 |
+
{
|
| 11975 |
+
"epoch": 1.36,
|
| 11976 |
+
"learning_rate": 0.0003444338270425227,
|
| 11977 |
+
"loss": 1.9967,
|
| 11978 |
+
"step": 1983
|
| 11979 |
+
},
|
| 11980 |
+
{
|
| 11981 |
+
"epoch": 1.36,
|
| 11982 |
+
"learning_rate": 0.00034429049211657907,
|
| 11983 |
+
"loss": 1.9377,
|
| 11984 |
+
"step": 1984
|
| 11985 |
+
},
|
| 11986 |
+
{
|
| 11987 |
+
"epoch": 1.36,
|
| 11988 |
+
"learning_rate": 0.0003441471571906354,
|
| 11989 |
+
"loss": 1.9739,
|
| 11990 |
+
"step": 1985
|
| 11991 |
+
},
|
| 11992 |
+
{
|
| 11993 |
+
"epoch": 1.36,
|
| 11994 |
+
"learning_rate": 0.00034400382226469183,
|
| 11995 |
+
"loss": 2.0232,
|
| 11996 |
+
"step": 1986
|
| 11997 |
+
},
|
| 11998 |
+
{
|
| 11999 |
+
"epoch": 1.36,
|
| 12000 |
+
"learning_rate": 0.00034386048733874816,
|
| 12001 |
+
"loss": 1.9893,
|
| 12002 |
+
"step": 1987
|
| 12003 |
+
},
|
| 12004 |
+
{
|
| 12005 |
+
"epoch": 1.36,
|
| 12006 |
+
"learning_rate": 0.0003437171524128046,
|
| 12007 |
+
"loss": 2.1028,
|
| 12008 |
+
"step": 1988
|
| 12009 |
+
},
|
| 12010 |
+
{
|
| 12011 |
+
"epoch": 1.36,
|
| 12012 |
+
"learning_rate": 0.00034357381748686093,
|
| 12013 |
+
"loss": 2.0027,
|
| 12014 |
+
"step": 1989
|
| 12015 |
+
},
|
| 12016 |
+
{
|
| 12017 |
+
"epoch": 1.36,
|
| 12018 |
+
"learning_rate": 0.0003434304825609173,
|
| 12019 |
+
"loss": 1.9481,
|
| 12020 |
+
"step": 1990
|
| 12021 |
+
},
|
| 12022 |
+
{
|
| 12023 |
+
"epoch": 1.36,
|
| 12024 |
+
"learning_rate": 0.0003432871476349737,
|
| 12025 |
+
"loss": 2.0907,
|
| 12026 |
+
"step": 1991
|
| 12027 |
+
},
|
| 12028 |
+
{
|
| 12029 |
+
"epoch": 1.36,
|
| 12030 |
+
"learning_rate": 0.00034314381270903003,
|
| 12031 |
+
"loss": 2.0744,
|
| 12032 |
+
"step": 1992
|
| 12033 |
+
},
|
| 12034 |
+
{
|
| 12035 |
+
"epoch": 1.36,
|
| 12036 |
+
"learning_rate": 0.00034300047778308647,
|
| 12037 |
+
"loss": 2.0442,
|
| 12038 |
+
"step": 1993
|
| 12039 |
+
},
|
| 12040 |
+
{
|
| 12041 |
+
"epoch": 1.36,
|
| 12042 |
+
"learning_rate": 0.0003428571428571428,
|
| 12043 |
+
"loss": 2.0232,
|
| 12044 |
+
"step": 1994
|
| 12045 |
+
},
|
| 12046 |
+
{
|
| 12047 |
+
"epoch": 1.36,
|
| 12048 |
+
"learning_rate": 0.00034271380793119923,
|
| 12049 |
+
"loss": 2.0709,
|
| 12050 |
+
"step": 1995
|
| 12051 |
+
},
|
| 12052 |
+
{
|
| 12053 |
+
"epoch": 1.36,
|
| 12054 |
+
"learning_rate": 0.00034257047300525556,
|
| 12055 |
+
"loss": 1.9478,
|
| 12056 |
+
"step": 1996
|
| 12057 |
+
},
|
| 12058 |
+
{
|
| 12059 |
+
"epoch": 1.37,
|
| 12060 |
+
"learning_rate": 0.000342427138079312,
|
| 12061 |
+
"loss": 1.9676,
|
| 12062 |
+
"step": 1997
|
| 12063 |
+
},
|
| 12064 |
+
{
|
| 12065 |
+
"epoch": 1.37,
|
| 12066 |
+
"learning_rate": 0.00034228380315336833,
|
| 12067 |
+
"loss": 2.1258,
|
| 12068 |
+
"step": 1998
|
| 12069 |
+
},
|
| 12070 |
+
{
|
| 12071 |
+
"epoch": 1.37,
|
| 12072 |
+
"learning_rate": 0.0003421404682274247,
|
| 12073 |
+
"loss": 1.9892,
|
| 12074 |
+
"step": 1999
|
| 12075 |
+
},
|
| 12076 |
+
{
|
| 12077 |
+
"epoch": 1.37,
|
| 12078 |
+
"learning_rate": 0.0003419971333014811,
|
| 12079 |
+
"loss": 2.0512,
|
| 12080 |
+
"step": 2000
|
| 12081 |
+
},
|
| 12082 |
+
{
|
| 12083 |
+
"epoch": 1.37,
|
| 12084 |
+
"eval_loss": 1.9946362972259521,
|
| 12085 |
+
"eval_runtime": 1647.5347,
|
| 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": 2.4507225531653345e+18,
|
| 12094 |
"trial_name": null,
|
| 12095 |
"trial_params": null
|
| 12096 |
}
|
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:af24d837954601d8f9729f848ea03545b9388763d421c6d2dce0a7e8ab73e67c
|
| 3 |
size 2368281769
|