Training in progress, step 800
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 +2413 -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:ac64079a0330d63c2055aac2d6c2fdd2bdd9007c85e42939ced7054a4deb4102
|
| 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 sha256:18d064e377257d8b4962000d02ab7ff4990aabdaf29ea0066e9c7e06266ac53f
|
| 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:c99e0c5c4d03d18376b499a14aa2de3c961e0b17bd1755ddc51098c9bdc845ed
|
| 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:e958ac3a412653c5a55518335f9da8afd5387ca22d1d3dbfc5f94de4bbb91f7f
|
| 3 |
size 627
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric":
|
| 3 |
-
"best_model_checkpoint": "output/checkpoint-
|
| 4 |
-
"epoch": 0.
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
@@ -2414,11 +2414,2419 @@
|
|
| 2414 |
"eval_samples_per_second": 9.879,
|
| 2415 |
"eval_steps_per_second": 9.879,
|
| 2416 |
"step": 400
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2417 |
}
|
| 2418 |
],
|
| 2419 |
"max_steps": 5848,
|
| 2420 |
"num_train_epochs": 4,
|
| 2421 |
-
"total_flos":
|
| 2422 |
"trial_name": null,
|
| 2423 |
"trial_params": null
|
| 2424 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 1.9985228776931763,
|
| 3 |
+
"best_model_checkpoint": "output/checkpoint-800",
|
| 4 |
+
"epoch": 0.5469471907531741,
|
| 5 |
+
"global_step": 800,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 2414 |
"eval_samples_per_second": 9.879,
|
| 2415 |
"eval_steps_per_second": 9.879,
|
| 2416 |
"step": 400
|
| 2417 |
+
},
|
| 2418 |
+
{
|
| 2419 |
+
"epoch": 0.27,
|
| 2420 |
+
"learning_rate": 0.0005786473087818696,
|
| 2421 |
+
"loss": 1.9149,
|
| 2422 |
+
"step": 401
|
| 2423 |
+
},
|
| 2424 |
+
{
|
| 2425 |
+
"epoch": 0.27,
|
| 2426 |
+
"learning_rate": 0.0005785410764872521,
|
| 2427 |
+
"loss": 1.9261,
|
| 2428 |
+
"step": 402
|
| 2429 |
+
},
|
| 2430 |
+
{
|
| 2431 |
+
"epoch": 0.28,
|
| 2432 |
+
"learning_rate": 0.0005784348441926345,
|
| 2433 |
+
"loss": 1.9515,
|
| 2434 |
+
"step": 403
|
| 2435 |
+
},
|
| 2436 |
+
{
|
| 2437 |
+
"epoch": 0.28,
|
| 2438 |
+
"learning_rate": 0.0005783286118980169,
|
| 2439 |
+
"loss": 1.8682,
|
| 2440 |
+
"step": 404
|
| 2441 |
+
},
|
| 2442 |
+
{
|
| 2443 |
+
"epoch": 0.28,
|
| 2444 |
+
"learning_rate": 0.0005782223796033994,
|
| 2445 |
+
"loss": 1.8787,
|
| 2446 |
+
"step": 405
|
| 2447 |
+
},
|
| 2448 |
+
{
|
| 2449 |
+
"epoch": 0.28,
|
| 2450 |
+
"learning_rate": 0.0005781161473087819,
|
| 2451 |
+
"loss": 1.8882,
|
| 2452 |
+
"step": 406
|
| 2453 |
+
},
|
| 2454 |
+
{
|
| 2455 |
+
"epoch": 0.28,
|
| 2456 |
+
"learning_rate": 0.0005780099150141643,
|
| 2457 |
+
"loss": 1.8623,
|
| 2458 |
+
"step": 407
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"epoch": 0.28,
|
| 2462 |
+
"learning_rate": 0.0005779036827195467,
|
| 2463 |
+
"loss": 1.9533,
|
| 2464 |
+
"step": 408
|
| 2465 |
+
},
|
| 2466 |
+
{
|
| 2467 |
+
"epoch": 0.28,
|
| 2468 |
+
"learning_rate": 0.0005777974504249291,
|
| 2469 |
+
"loss": 1.8139,
|
| 2470 |
+
"step": 409
|
| 2471 |
+
},
|
| 2472 |
+
{
|
| 2473 |
+
"epoch": 0.28,
|
| 2474 |
+
"learning_rate": 0.0005776912181303115,
|
| 2475 |
+
"loss": 1.9436,
|
| 2476 |
+
"step": 410
|
| 2477 |
+
},
|
| 2478 |
+
{
|
| 2479 |
+
"epoch": 0.28,
|
| 2480 |
+
"learning_rate": 0.000577584985835694,
|
| 2481 |
+
"loss": 2.012,
|
| 2482 |
+
"step": 411
|
| 2483 |
+
},
|
| 2484 |
+
{
|
| 2485 |
+
"epoch": 0.28,
|
| 2486 |
+
"learning_rate": 0.0005774787535410764,
|
| 2487 |
+
"loss": 1.8593,
|
| 2488 |
+
"step": 412
|
| 2489 |
+
},
|
| 2490 |
+
{
|
| 2491 |
+
"epoch": 0.28,
|
| 2492 |
+
"learning_rate": 0.0005773725212464589,
|
| 2493 |
+
"loss": 1.9534,
|
| 2494 |
+
"step": 413
|
| 2495 |
+
},
|
| 2496 |
+
{
|
| 2497 |
+
"epoch": 0.28,
|
| 2498 |
+
"learning_rate": 0.0005772662889518413,
|
| 2499 |
+
"loss": 1.8901,
|
| 2500 |
+
"step": 414
|
| 2501 |
+
},
|
| 2502 |
+
{
|
| 2503 |
+
"epoch": 0.28,
|
| 2504 |
+
"learning_rate": 0.0005771600566572237,
|
| 2505 |
+
"loss": 1.9311,
|
| 2506 |
+
"step": 415
|
| 2507 |
+
},
|
| 2508 |
+
{
|
| 2509 |
+
"epoch": 0.28,
|
| 2510 |
+
"learning_rate": 0.0005770538243626062,
|
| 2511 |
+
"loss": 1.9357,
|
| 2512 |
+
"step": 416
|
| 2513 |
+
},
|
| 2514 |
+
{
|
| 2515 |
+
"epoch": 0.29,
|
| 2516 |
+
"learning_rate": 0.0005769475920679887,
|
| 2517 |
+
"loss": 2.0402,
|
| 2518 |
+
"step": 417
|
| 2519 |
+
},
|
| 2520 |
+
{
|
| 2521 |
+
"epoch": 0.29,
|
| 2522 |
+
"learning_rate": 0.0005768413597733711,
|
| 2523 |
+
"loss": 1.9272,
|
| 2524 |
+
"step": 418
|
| 2525 |
+
},
|
| 2526 |
+
{
|
| 2527 |
+
"epoch": 0.29,
|
| 2528 |
+
"learning_rate": 0.0005767351274787535,
|
| 2529 |
+
"loss": 1.8699,
|
| 2530 |
+
"step": 419
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"epoch": 0.29,
|
| 2534 |
+
"learning_rate": 0.000576628895184136,
|
| 2535 |
+
"loss": 1.9146,
|
| 2536 |
+
"step": 420
|
| 2537 |
+
},
|
| 2538 |
+
{
|
| 2539 |
+
"epoch": 0.29,
|
| 2540 |
+
"learning_rate": 0.0005765226628895183,
|
| 2541 |
+
"loss": 1.8632,
|
| 2542 |
+
"step": 421
|
| 2543 |
+
},
|
| 2544 |
+
{
|
| 2545 |
+
"epoch": 0.29,
|
| 2546 |
+
"learning_rate": 0.0005764164305949008,
|
| 2547 |
+
"loss": 1.9577,
|
| 2548 |
+
"step": 422
|
| 2549 |
+
},
|
| 2550 |
+
{
|
| 2551 |
+
"epoch": 0.29,
|
| 2552 |
+
"learning_rate": 0.0005763101983002832,
|
| 2553 |
+
"loss": 1.9124,
|
| 2554 |
+
"step": 423
|
| 2555 |
+
},
|
| 2556 |
+
{
|
| 2557 |
+
"epoch": 0.29,
|
| 2558 |
+
"learning_rate": 0.0005762039660056656,
|
| 2559 |
+
"loss": 1.8922,
|
| 2560 |
+
"step": 424
|
| 2561 |
+
},
|
| 2562 |
+
{
|
| 2563 |
+
"epoch": 0.29,
|
| 2564 |
+
"learning_rate": 0.0005760977337110481,
|
| 2565 |
+
"loss": 1.8515,
|
| 2566 |
+
"step": 425
|
| 2567 |
+
},
|
| 2568 |
+
{
|
| 2569 |
+
"epoch": 0.29,
|
| 2570 |
+
"learning_rate": 0.0005759915014164305,
|
| 2571 |
+
"loss": 1.9232,
|
| 2572 |
+
"step": 426
|
| 2573 |
+
},
|
| 2574 |
+
{
|
| 2575 |
+
"epoch": 0.29,
|
| 2576 |
+
"learning_rate": 0.000575885269121813,
|
| 2577 |
+
"loss": 1.9402,
|
| 2578 |
+
"step": 427
|
| 2579 |
+
},
|
| 2580 |
+
{
|
| 2581 |
+
"epoch": 0.29,
|
| 2582 |
+
"learning_rate": 0.0005757790368271955,
|
| 2583 |
+
"loss": 1.885,
|
| 2584 |
+
"step": 428
|
| 2585 |
+
},
|
| 2586 |
+
{
|
| 2587 |
+
"epoch": 0.29,
|
| 2588 |
+
"learning_rate": 0.0005756728045325779,
|
| 2589 |
+
"loss": 1.8847,
|
| 2590 |
+
"step": 429
|
| 2591 |
+
},
|
| 2592 |
+
{
|
| 2593 |
+
"epoch": 0.29,
|
| 2594 |
+
"learning_rate": 0.0005755665722379603,
|
| 2595 |
+
"loss": 1.8776,
|
| 2596 |
+
"step": 430
|
| 2597 |
+
},
|
| 2598 |
+
{
|
| 2599 |
+
"epoch": 0.29,
|
| 2600 |
+
"learning_rate": 0.0005754603399433428,
|
| 2601 |
+
"loss": 1.9292,
|
| 2602 |
+
"step": 431
|
| 2603 |
+
},
|
| 2604 |
+
{
|
| 2605 |
+
"epoch": 0.3,
|
| 2606 |
+
"learning_rate": 0.0005753541076487251,
|
| 2607 |
+
"loss": 2.037,
|
| 2608 |
+
"step": 432
|
| 2609 |
+
},
|
| 2610 |
+
{
|
| 2611 |
+
"epoch": 0.3,
|
| 2612 |
+
"learning_rate": 0.0005752478753541076,
|
| 2613 |
+
"loss": 1.9526,
|
| 2614 |
+
"step": 433
|
| 2615 |
+
},
|
| 2616 |
+
{
|
| 2617 |
+
"epoch": 0.3,
|
| 2618 |
+
"learning_rate": 0.00057514164305949,
|
| 2619 |
+
"loss": 1.8857,
|
| 2620 |
+
"step": 434
|
| 2621 |
+
},
|
| 2622 |
+
{
|
| 2623 |
+
"epoch": 0.3,
|
| 2624 |
+
"learning_rate": 0.0005750354107648724,
|
| 2625 |
+
"loss": 1.8571,
|
| 2626 |
+
"step": 435
|
| 2627 |
+
},
|
| 2628 |
+
{
|
| 2629 |
+
"epoch": 0.3,
|
| 2630 |
+
"learning_rate": 0.0005749291784702549,
|
| 2631 |
+
"loss": 1.99,
|
| 2632 |
+
"step": 436
|
| 2633 |
+
},
|
| 2634 |
+
{
|
| 2635 |
+
"epoch": 0.3,
|
| 2636 |
+
"learning_rate": 0.0005748229461756373,
|
| 2637 |
+
"loss": 1.9094,
|
| 2638 |
+
"step": 437
|
| 2639 |
+
},
|
| 2640 |
+
{
|
| 2641 |
+
"epoch": 0.3,
|
| 2642 |
+
"learning_rate": 0.0005747167138810198,
|
| 2643 |
+
"loss": 1.8634,
|
| 2644 |
+
"step": 438
|
| 2645 |
+
},
|
| 2646 |
+
{
|
| 2647 |
+
"epoch": 0.3,
|
| 2648 |
+
"learning_rate": 0.0005746104815864022,
|
| 2649 |
+
"loss": 1.85,
|
| 2650 |
+
"step": 439
|
| 2651 |
+
},
|
| 2652 |
+
{
|
| 2653 |
+
"epoch": 0.3,
|
| 2654 |
+
"learning_rate": 0.0005745042492917847,
|
| 2655 |
+
"loss": 1.9329,
|
| 2656 |
+
"step": 440
|
| 2657 |
+
},
|
| 2658 |
+
{
|
| 2659 |
+
"epoch": 0.3,
|
| 2660 |
+
"learning_rate": 0.0005743980169971671,
|
| 2661 |
+
"loss": 1.9566,
|
| 2662 |
+
"step": 441
|
| 2663 |
+
},
|
| 2664 |
+
{
|
| 2665 |
+
"epoch": 0.3,
|
| 2666 |
+
"learning_rate": 0.0005742917847025496,
|
| 2667 |
+
"loss": 1.9756,
|
| 2668 |
+
"step": 442
|
| 2669 |
+
},
|
| 2670 |
+
{
|
| 2671 |
+
"epoch": 0.3,
|
| 2672 |
+
"learning_rate": 0.000574185552407932,
|
| 2673 |
+
"loss": 1.9327,
|
| 2674 |
+
"step": 443
|
| 2675 |
+
},
|
| 2676 |
+
{
|
| 2677 |
+
"epoch": 0.3,
|
| 2678 |
+
"learning_rate": 0.0005740793201133143,
|
| 2679 |
+
"loss": 1.8133,
|
| 2680 |
+
"step": 444
|
| 2681 |
+
},
|
| 2682 |
+
{
|
| 2683 |
+
"epoch": 0.3,
|
| 2684 |
+
"learning_rate": 0.0005739730878186968,
|
| 2685 |
+
"loss": 1.9292,
|
| 2686 |
+
"step": 445
|
| 2687 |
+
},
|
| 2688 |
+
{
|
| 2689 |
+
"epoch": 0.3,
|
| 2690 |
+
"learning_rate": 0.0005738668555240792,
|
| 2691 |
+
"loss": 1.8839,
|
| 2692 |
+
"step": 446
|
| 2693 |
+
},
|
| 2694 |
+
{
|
| 2695 |
+
"epoch": 0.31,
|
| 2696 |
+
"learning_rate": 0.0005737606232294617,
|
| 2697 |
+
"loss": 1.9216,
|
| 2698 |
+
"step": 447
|
| 2699 |
+
},
|
| 2700 |
+
{
|
| 2701 |
+
"epoch": 0.31,
|
| 2702 |
+
"learning_rate": 0.0005736543909348441,
|
| 2703 |
+
"loss": 1.9303,
|
| 2704 |
+
"step": 448
|
| 2705 |
+
},
|
| 2706 |
+
{
|
| 2707 |
+
"epoch": 0.31,
|
| 2708 |
+
"learning_rate": 0.0005735481586402266,
|
| 2709 |
+
"loss": 2.0129,
|
| 2710 |
+
"step": 449
|
| 2711 |
+
},
|
| 2712 |
+
{
|
| 2713 |
+
"epoch": 0.31,
|
| 2714 |
+
"learning_rate": 0.000573441926345609,
|
| 2715 |
+
"loss": 1.9233,
|
| 2716 |
+
"step": 450
|
| 2717 |
+
},
|
| 2718 |
+
{
|
| 2719 |
+
"epoch": 0.31,
|
| 2720 |
+
"learning_rate": 0.0005733356940509915,
|
| 2721 |
+
"loss": 1.809,
|
| 2722 |
+
"step": 451
|
| 2723 |
+
},
|
| 2724 |
+
{
|
| 2725 |
+
"epoch": 0.31,
|
| 2726 |
+
"learning_rate": 0.0005732294617563739,
|
| 2727 |
+
"loss": 1.822,
|
| 2728 |
+
"step": 452
|
| 2729 |
+
},
|
| 2730 |
+
{
|
| 2731 |
+
"epoch": 0.31,
|
| 2732 |
+
"learning_rate": 0.0005731232294617564,
|
| 2733 |
+
"loss": 1.829,
|
| 2734 |
+
"step": 453
|
| 2735 |
+
},
|
| 2736 |
+
{
|
| 2737 |
+
"epoch": 0.31,
|
| 2738 |
+
"learning_rate": 0.0005730169971671388,
|
| 2739 |
+
"loss": 1.9524,
|
| 2740 |
+
"step": 454
|
| 2741 |
+
},
|
| 2742 |
+
{
|
| 2743 |
+
"epoch": 0.31,
|
| 2744 |
+
"learning_rate": 0.0005729107648725212,
|
| 2745 |
+
"loss": 1.8695,
|
| 2746 |
+
"step": 455
|
| 2747 |
+
},
|
| 2748 |
+
{
|
| 2749 |
+
"epoch": 0.31,
|
| 2750 |
+
"learning_rate": 0.0005728045325779037,
|
| 2751 |
+
"loss": 1.8614,
|
| 2752 |
+
"step": 456
|
| 2753 |
+
},
|
| 2754 |
+
{
|
| 2755 |
+
"epoch": 0.31,
|
| 2756 |
+
"learning_rate": 0.000572698300283286,
|
| 2757 |
+
"loss": 1.8697,
|
| 2758 |
+
"step": 457
|
| 2759 |
+
},
|
| 2760 |
+
{
|
| 2761 |
+
"epoch": 0.31,
|
| 2762 |
+
"learning_rate": 0.0005725920679886684,
|
| 2763 |
+
"loss": 1.8378,
|
| 2764 |
+
"step": 458
|
| 2765 |
+
},
|
| 2766 |
+
{
|
| 2767 |
+
"epoch": 0.31,
|
| 2768 |
+
"learning_rate": 0.0005724858356940509,
|
| 2769 |
+
"loss": 1.9554,
|
| 2770 |
+
"step": 459
|
| 2771 |
+
},
|
| 2772 |
+
{
|
| 2773 |
+
"epoch": 0.31,
|
| 2774 |
+
"learning_rate": 0.0005723796033994334,
|
| 2775 |
+
"loss": 1.9066,
|
| 2776 |
+
"step": 460
|
| 2777 |
+
},
|
| 2778 |
+
{
|
| 2779 |
+
"epoch": 0.32,
|
| 2780 |
+
"learning_rate": 0.0005722733711048158,
|
| 2781 |
+
"loss": 1.8836,
|
| 2782 |
+
"step": 461
|
| 2783 |
+
},
|
| 2784 |
+
{
|
| 2785 |
+
"epoch": 0.32,
|
| 2786 |
+
"learning_rate": 0.0005721671388101983,
|
| 2787 |
+
"loss": 1.9541,
|
| 2788 |
+
"step": 462
|
| 2789 |
+
},
|
| 2790 |
+
{
|
| 2791 |
+
"epoch": 0.32,
|
| 2792 |
+
"learning_rate": 0.0005720609065155807,
|
| 2793 |
+
"loss": 1.8879,
|
| 2794 |
+
"step": 463
|
| 2795 |
+
},
|
| 2796 |
+
{
|
| 2797 |
+
"epoch": 0.32,
|
| 2798 |
+
"learning_rate": 0.0005719546742209631,
|
| 2799 |
+
"loss": 1.8126,
|
| 2800 |
+
"step": 464
|
| 2801 |
+
},
|
| 2802 |
+
{
|
| 2803 |
+
"epoch": 0.32,
|
| 2804 |
+
"learning_rate": 0.0005718484419263456,
|
| 2805 |
+
"loss": 1.9861,
|
| 2806 |
+
"step": 465
|
| 2807 |
+
},
|
| 2808 |
+
{
|
| 2809 |
+
"epoch": 0.32,
|
| 2810 |
+
"learning_rate": 0.000571742209631728,
|
| 2811 |
+
"loss": 2.0253,
|
| 2812 |
+
"step": 466
|
| 2813 |
+
},
|
| 2814 |
+
{
|
| 2815 |
+
"epoch": 0.32,
|
| 2816 |
+
"learning_rate": 0.0005716359773371105,
|
| 2817 |
+
"loss": 1.9213,
|
| 2818 |
+
"step": 467
|
| 2819 |
+
},
|
| 2820 |
+
{
|
| 2821 |
+
"epoch": 0.32,
|
| 2822 |
+
"learning_rate": 0.0005715297450424929,
|
| 2823 |
+
"loss": 1.9988,
|
| 2824 |
+
"step": 468
|
| 2825 |
+
},
|
| 2826 |
+
{
|
| 2827 |
+
"epoch": 0.32,
|
| 2828 |
+
"learning_rate": 0.0005714235127478752,
|
| 2829 |
+
"loss": 1.9285,
|
| 2830 |
+
"step": 469
|
| 2831 |
+
},
|
| 2832 |
+
{
|
| 2833 |
+
"epoch": 0.32,
|
| 2834 |
+
"learning_rate": 0.0005713172804532577,
|
| 2835 |
+
"loss": 1.9973,
|
| 2836 |
+
"step": 470
|
| 2837 |
+
},
|
| 2838 |
+
{
|
| 2839 |
+
"epoch": 0.32,
|
| 2840 |
+
"learning_rate": 0.0005712110481586402,
|
| 2841 |
+
"loss": 1.945,
|
| 2842 |
+
"step": 471
|
| 2843 |
+
},
|
| 2844 |
+
{
|
| 2845 |
+
"epoch": 0.32,
|
| 2846 |
+
"learning_rate": 0.0005711048158640226,
|
| 2847 |
+
"loss": 1.9327,
|
| 2848 |
+
"step": 472
|
| 2849 |
+
},
|
| 2850 |
+
{
|
| 2851 |
+
"epoch": 0.32,
|
| 2852 |
+
"learning_rate": 0.0005709985835694051,
|
| 2853 |
+
"loss": 2.0014,
|
| 2854 |
+
"step": 473
|
| 2855 |
+
},
|
| 2856 |
+
{
|
| 2857 |
+
"epoch": 0.32,
|
| 2858 |
+
"learning_rate": 0.0005708923512747875,
|
| 2859 |
+
"loss": 1.8221,
|
| 2860 |
+
"step": 474
|
| 2861 |
+
},
|
| 2862 |
+
{
|
| 2863 |
+
"epoch": 0.32,
|
| 2864 |
+
"learning_rate": 0.0005707861189801699,
|
| 2865 |
+
"loss": 1.8905,
|
| 2866 |
+
"step": 475
|
| 2867 |
+
},
|
| 2868 |
+
{
|
| 2869 |
+
"epoch": 0.33,
|
| 2870 |
+
"learning_rate": 0.0005706798866855524,
|
| 2871 |
+
"loss": 1.9236,
|
| 2872 |
+
"step": 476
|
| 2873 |
+
},
|
| 2874 |
+
{
|
| 2875 |
+
"epoch": 0.33,
|
| 2876 |
+
"learning_rate": 0.0005705736543909348,
|
| 2877 |
+
"loss": 1.8436,
|
| 2878 |
+
"step": 477
|
| 2879 |
+
},
|
| 2880 |
+
{
|
| 2881 |
+
"epoch": 0.33,
|
| 2882 |
+
"learning_rate": 0.0005704674220963172,
|
| 2883 |
+
"loss": 1.8545,
|
| 2884 |
+
"step": 478
|
| 2885 |
+
},
|
| 2886 |
+
{
|
| 2887 |
+
"epoch": 0.33,
|
| 2888 |
+
"learning_rate": 0.0005703611898016997,
|
| 2889 |
+
"loss": 1.8886,
|
| 2890 |
+
"step": 479
|
| 2891 |
+
},
|
| 2892 |
+
{
|
| 2893 |
+
"epoch": 0.33,
|
| 2894 |
+
"learning_rate": 0.000570254957507082,
|
| 2895 |
+
"loss": 1.873,
|
| 2896 |
+
"step": 480
|
| 2897 |
+
},
|
| 2898 |
+
{
|
| 2899 |
+
"epoch": 0.33,
|
| 2900 |
+
"learning_rate": 0.0005701487252124645,
|
| 2901 |
+
"loss": 1.8624,
|
| 2902 |
+
"step": 481
|
| 2903 |
+
},
|
| 2904 |
+
{
|
| 2905 |
+
"epoch": 0.33,
|
| 2906 |
+
"learning_rate": 0.000570042492917847,
|
| 2907 |
+
"loss": 1.8614,
|
| 2908 |
+
"step": 482
|
| 2909 |
+
},
|
| 2910 |
+
{
|
| 2911 |
+
"epoch": 0.33,
|
| 2912 |
+
"learning_rate": 0.0005699362606232294,
|
| 2913 |
+
"loss": 1.9153,
|
| 2914 |
+
"step": 483
|
| 2915 |
+
},
|
| 2916 |
+
{
|
| 2917 |
+
"epoch": 0.33,
|
| 2918 |
+
"learning_rate": 0.0005698300283286118,
|
| 2919 |
+
"loss": 1.8524,
|
| 2920 |
+
"step": 484
|
| 2921 |
+
},
|
| 2922 |
+
{
|
| 2923 |
+
"epoch": 0.33,
|
| 2924 |
+
"learning_rate": 0.0005697237960339943,
|
| 2925 |
+
"loss": 1.9039,
|
| 2926 |
+
"step": 485
|
| 2927 |
+
},
|
| 2928 |
+
{
|
| 2929 |
+
"epoch": 0.33,
|
| 2930 |
+
"learning_rate": 0.0005696175637393767,
|
| 2931 |
+
"loss": 1.8866,
|
| 2932 |
+
"step": 486
|
| 2933 |
+
},
|
| 2934 |
+
{
|
| 2935 |
+
"epoch": 0.33,
|
| 2936 |
+
"learning_rate": 0.0005695113314447592,
|
| 2937 |
+
"loss": 1.7717,
|
| 2938 |
+
"step": 487
|
| 2939 |
+
},
|
| 2940 |
+
{
|
| 2941 |
+
"epoch": 0.33,
|
| 2942 |
+
"learning_rate": 0.0005694050991501416,
|
| 2943 |
+
"loss": 1.9695,
|
| 2944 |
+
"step": 488
|
| 2945 |
+
},
|
| 2946 |
+
{
|
| 2947 |
+
"epoch": 0.33,
|
| 2948 |
+
"learning_rate": 0.000569298866855524,
|
| 2949 |
+
"loss": 1.9931,
|
| 2950 |
+
"step": 489
|
| 2951 |
+
},
|
| 2952 |
+
{
|
| 2953 |
+
"epoch": 0.34,
|
| 2954 |
+
"learning_rate": 0.0005691926345609065,
|
| 2955 |
+
"loss": 1.8473,
|
| 2956 |
+
"step": 490
|
| 2957 |
+
},
|
| 2958 |
+
{
|
| 2959 |
+
"epoch": 0.34,
|
| 2960 |
+
"learning_rate": 0.0005690864022662889,
|
| 2961 |
+
"loss": 1.8366,
|
| 2962 |
+
"step": 491
|
| 2963 |
+
},
|
| 2964 |
+
{
|
| 2965 |
+
"epoch": 0.34,
|
| 2966 |
+
"learning_rate": 0.0005689801699716714,
|
| 2967 |
+
"loss": 1.9242,
|
| 2968 |
+
"step": 492
|
| 2969 |
+
},
|
| 2970 |
+
{
|
| 2971 |
+
"epoch": 0.34,
|
| 2972 |
+
"learning_rate": 0.0005688739376770539,
|
| 2973 |
+
"loss": 1.9081,
|
| 2974 |
+
"step": 493
|
| 2975 |
+
},
|
| 2976 |
+
{
|
| 2977 |
+
"epoch": 0.34,
|
| 2978 |
+
"learning_rate": 0.0005687677053824362,
|
| 2979 |
+
"loss": 1.8658,
|
| 2980 |
+
"step": 494
|
| 2981 |
+
},
|
| 2982 |
+
{
|
| 2983 |
+
"epoch": 0.34,
|
| 2984 |
+
"learning_rate": 0.0005686614730878186,
|
| 2985 |
+
"loss": 1.9603,
|
| 2986 |
+
"step": 495
|
| 2987 |
+
},
|
| 2988 |
+
{
|
| 2989 |
+
"epoch": 0.34,
|
| 2990 |
+
"learning_rate": 0.0005685552407932011,
|
| 2991 |
+
"loss": 1.8292,
|
| 2992 |
+
"step": 496
|
| 2993 |
+
},
|
| 2994 |
+
{
|
| 2995 |
+
"epoch": 0.34,
|
| 2996 |
+
"learning_rate": 0.0005684490084985835,
|
| 2997 |
+
"loss": 1.8917,
|
| 2998 |
+
"step": 497
|
| 2999 |
+
},
|
| 3000 |
+
{
|
| 3001 |
+
"epoch": 0.34,
|
| 3002 |
+
"learning_rate": 0.0005683427762039659,
|
| 3003 |
+
"loss": 1.9171,
|
| 3004 |
+
"step": 498
|
| 3005 |
+
},
|
| 3006 |
+
{
|
| 3007 |
+
"epoch": 0.34,
|
| 3008 |
+
"learning_rate": 0.0005682365439093484,
|
| 3009 |
+
"loss": 1.9373,
|
| 3010 |
+
"step": 499
|
| 3011 |
+
},
|
| 3012 |
+
{
|
| 3013 |
+
"epoch": 0.34,
|
| 3014 |
+
"learning_rate": 0.0005681303116147308,
|
| 3015 |
+
"loss": 1.9529,
|
| 3016 |
+
"step": 500
|
| 3017 |
+
},
|
| 3018 |
+
{
|
| 3019 |
+
"epoch": 0.34,
|
| 3020 |
+
"learning_rate": 0.0005680240793201133,
|
| 3021 |
+
"loss": 1.9459,
|
| 3022 |
+
"step": 501
|
| 3023 |
+
},
|
| 3024 |
+
{
|
| 3025 |
+
"epoch": 0.34,
|
| 3026 |
+
"learning_rate": 0.0005679178470254957,
|
| 3027 |
+
"loss": 1.9941,
|
| 3028 |
+
"step": 502
|
| 3029 |
+
},
|
| 3030 |
+
{
|
| 3031 |
+
"epoch": 0.34,
|
| 3032 |
+
"learning_rate": 0.0005678116147308781,
|
| 3033 |
+
"loss": 1.8738,
|
| 3034 |
+
"step": 503
|
| 3035 |
+
},
|
| 3036 |
+
{
|
| 3037 |
+
"epoch": 0.34,
|
| 3038 |
+
"learning_rate": 0.0005677053824362606,
|
| 3039 |
+
"loss": 1.9666,
|
| 3040 |
+
"step": 504
|
| 3041 |
+
},
|
| 3042 |
+
{
|
| 3043 |
+
"epoch": 0.35,
|
| 3044 |
+
"learning_rate": 0.000567599150141643,
|
| 3045 |
+
"loss": 2.0352,
|
| 3046 |
+
"step": 505
|
| 3047 |
+
},
|
| 3048 |
+
{
|
| 3049 |
+
"epoch": 0.35,
|
| 3050 |
+
"learning_rate": 0.0005674929178470254,
|
| 3051 |
+
"loss": 1.8621,
|
| 3052 |
+
"step": 506
|
| 3053 |
+
},
|
| 3054 |
+
{
|
| 3055 |
+
"epoch": 0.35,
|
| 3056 |
+
"learning_rate": 0.0005673866855524079,
|
| 3057 |
+
"loss": 1.8733,
|
| 3058 |
+
"step": 507
|
| 3059 |
+
},
|
| 3060 |
+
{
|
| 3061 |
+
"epoch": 0.35,
|
| 3062 |
+
"learning_rate": 0.0005672804532577903,
|
| 3063 |
+
"loss": 1.9414,
|
| 3064 |
+
"step": 508
|
| 3065 |
+
},
|
| 3066 |
+
{
|
| 3067 |
+
"epoch": 0.35,
|
| 3068 |
+
"learning_rate": 0.0005671742209631727,
|
| 3069 |
+
"loss": 1.9515,
|
| 3070 |
+
"step": 509
|
| 3071 |
+
},
|
| 3072 |
+
{
|
| 3073 |
+
"epoch": 0.35,
|
| 3074 |
+
"learning_rate": 0.0005670679886685552,
|
| 3075 |
+
"loss": 1.891,
|
| 3076 |
+
"step": 510
|
| 3077 |
+
},
|
| 3078 |
+
{
|
| 3079 |
+
"epoch": 0.35,
|
| 3080 |
+
"learning_rate": 0.0005669617563739376,
|
| 3081 |
+
"loss": 2.0231,
|
| 3082 |
+
"step": 511
|
| 3083 |
+
},
|
| 3084 |
+
{
|
| 3085 |
+
"epoch": 0.35,
|
| 3086 |
+
"learning_rate": 0.0005668555240793201,
|
| 3087 |
+
"loss": 1.9534,
|
| 3088 |
+
"step": 512
|
| 3089 |
+
},
|
| 3090 |
+
{
|
| 3091 |
+
"epoch": 0.35,
|
| 3092 |
+
"learning_rate": 0.0005667492917847025,
|
| 3093 |
+
"loss": 1.9437,
|
| 3094 |
+
"step": 513
|
| 3095 |
+
},
|
| 3096 |
+
{
|
| 3097 |
+
"epoch": 0.35,
|
| 3098 |
+
"learning_rate": 0.0005666430594900849,
|
| 3099 |
+
"loss": 1.8914,
|
| 3100 |
+
"step": 514
|
| 3101 |
+
},
|
| 3102 |
+
{
|
| 3103 |
+
"epoch": 0.35,
|
| 3104 |
+
"learning_rate": 0.0005665368271954674,
|
| 3105 |
+
"loss": 1.8711,
|
| 3106 |
+
"step": 515
|
| 3107 |
+
},
|
| 3108 |
+
{
|
| 3109 |
+
"epoch": 0.35,
|
| 3110 |
+
"learning_rate": 0.0005664305949008499,
|
| 3111 |
+
"loss": 1.9708,
|
| 3112 |
+
"step": 516
|
| 3113 |
+
},
|
| 3114 |
+
{
|
| 3115 |
+
"epoch": 0.35,
|
| 3116 |
+
"learning_rate": 0.0005663243626062323,
|
| 3117 |
+
"loss": 1.9462,
|
| 3118 |
+
"step": 517
|
| 3119 |
+
},
|
| 3120 |
+
{
|
| 3121 |
+
"epoch": 0.35,
|
| 3122 |
+
"learning_rate": 0.0005662181303116146,
|
| 3123 |
+
"loss": 1.8547,
|
| 3124 |
+
"step": 518
|
| 3125 |
+
},
|
| 3126 |
+
{
|
| 3127 |
+
"epoch": 0.35,
|
| 3128 |
+
"learning_rate": 0.0005661118980169971,
|
| 3129 |
+
"loss": 1.842,
|
| 3130 |
+
"step": 519
|
| 3131 |
+
},
|
| 3132 |
+
{
|
| 3133 |
+
"epoch": 0.36,
|
| 3134 |
+
"learning_rate": 0.0005660056657223795,
|
| 3135 |
+
"loss": 1.8544,
|
| 3136 |
+
"step": 520
|
| 3137 |
+
},
|
| 3138 |
+
{
|
| 3139 |
+
"epoch": 0.36,
|
| 3140 |
+
"learning_rate": 0.000565899433427762,
|
| 3141 |
+
"loss": 1.8744,
|
| 3142 |
+
"step": 521
|
| 3143 |
+
},
|
| 3144 |
+
{
|
| 3145 |
+
"epoch": 0.36,
|
| 3146 |
+
"learning_rate": 0.0005657932011331444,
|
| 3147 |
+
"loss": 1.9211,
|
| 3148 |
+
"step": 522
|
| 3149 |
+
},
|
| 3150 |
+
{
|
| 3151 |
+
"epoch": 0.36,
|
| 3152 |
+
"learning_rate": 0.0005656869688385268,
|
| 3153 |
+
"loss": 1.9399,
|
| 3154 |
+
"step": 523
|
| 3155 |
+
},
|
| 3156 |
+
{
|
| 3157 |
+
"epoch": 0.36,
|
| 3158 |
+
"learning_rate": 0.0005655807365439093,
|
| 3159 |
+
"loss": 1.9049,
|
| 3160 |
+
"step": 524
|
| 3161 |
+
},
|
| 3162 |
+
{
|
| 3163 |
+
"epoch": 0.36,
|
| 3164 |
+
"learning_rate": 0.0005654745042492917,
|
| 3165 |
+
"loss": 1.9623,
|
| 3166 |
+
"step": 525
|
| 3167 |
+
},
|
| 3168 |
+
{
|
| 3169 |
+
"epoch": 0.36,
|
| 3170 |
+
"learning_rate": 0.0005653682719546742,
|
| 3171 |
+
"loss": 1.9142,
|
| 3172 |
+
"step": 526
|
| 3173 |
+
},
|
| 3174 |
+
{
|
| 3175 |
+
"epoch": 0.36,
|
| 3176 |
+
"learning_rate": 0.0005652620396600567,
|
| 3177 |
+
"loss": 1.8899,
|
| 3178 |
+
"step": 527
|
| 3179 |
+
},
|
| 3180 |
+
{
|
| 3181 |
+
"epoch": 0.36,
|
| 3182 |
+
"learning_rate": 0.0005651558073654391,
|
| 3183 |
+
"loss": 1.8878,
|
| 3184 |
+
"step": 528
|
| 3185 |
+
},
|
| 3186 |
+
{
|
| 3187 |
+
"epoch": 0.36,
|
| 3188 |
+
"learning_rate": 0.0005650495750708215,
|
| 3189 |
+
"loss": 1.924,
|
| 3190 |
+
"step": 529
|
| 3191 |
+
},
|
| 3192 |
+
{
|
| 3193 |
+
"epoch": 0.36,
|
| 3194 |
+
"learning_rate": 0.000564943342776204,
|
| 3195 |
+
"loss": 1.9087,
|
| 3196 |
+
"step": 530
|
| 3197 |
+
},
|
| 3198 |
+
{
|
| 3199 |
+
"epoch": 0.36,
|
| 3200 |
+
"learning_rate": 0.0005648371104815863,
|
| 3201 |
+
"loss": 1.9945,
|
| 3202 |
+
"step": 531
|
| 3203 |
+
},
|
| 3204 |
+
{
|
| 3205 |
+
"epoch": 0.36,
|
| 3206 |
+
"learning_rate": 0.0005647308781869688,
|
| 3207 |
+
"loss": 1.8945,
|
| 3208 |
+
"step": 532
|
| 3209 |
+
},
|
| 3210 |
+
{
|
| 3211 |
+
"epoch": 0.36,
|
| 3212 |
+
"learning_rate": 0.0005646246458923512,
|
| 3213 |
+
"loss": 1.9209,
|
| 3214 |
+
"step": 533
|
| 3215 |
+
},
|
| 3216 |
+
{
|
| 3217 |
+
"epoch": 0.37,
|
| 3218 |
+
"learning_rate": 0.0005645184135977336,
|
| 3219 |
+
"loss": 1.9301,
|
| 3220 |
+
"step": 534
|
| 3221 |
+
},
|
| 3222 |
+
{
|
| 3223 |
+
"epoch": 0.37,
|
| 3224 |
+
"learning_rate": 0.0005644121813031161,
|
| 3225 |
+
"loss": 1.8516,
|
| 3226 |
+
"step": 535
|
| 3227 |
+
},
|
| 3228 |
+
{
|
| 3229 |
+
"epoch": 0.37,
|
| 3230 |
+
"learning_rate": 0.0005643059490084985,
|
| 3231 |
+
"loss": 1.906,
|
| 3232 |
+
"step": 536
|
| 3233 |
+
},
|
| 3234 |
+
{
|
| 3235 |
+
"epoch": 0.37,
|
| 3236 |
+
"learning_rate": 0.000564199716713881,
|
| 3237 |
+
"loss": 1.91,
|
| 3238 |
+
"step": 537
|
| 3239 |
+
},
|
| 3240 |
+
{
|
| 3241 |
+
"epoch": 0.37,
|
| 3242 |
+
"learning_rate": 0.0005640934844192634,
|
| 3243 |
+
"loss": 1.9069,
|
| 3244 |
+
"step": 538
|
| 3245 |
+
},
|
| 3246 |
+
{
|
| 3247 |
+
"epoch": 0.37,
|
| 3248 |
+
"learning_rate": 0.0005639872521246459,
|
| 3249 |
+
"loss": 1.7911,
|
| 3250 |
+
"step": 539
|
| 3251 |
+
},
|
| 3252 |
+
{
|
| 3253 |
+
"epoch": 0.37,
|
| 3254 |
+
"learning_rate": 0.0005638810198300283,
|
| 3255 |
+
"loss": 1.8925,
|
| 3256 |
+
"step": 540
|
| 3257 |
+
},
|
| 3258 |
+
{
|
| 3259 |
+
"epoch": 0.37,
|
| 3260 |
+
"learning_rate": 0.0005637747875354108,
|
| 3261 |
+
"loss": 1.9526,
|
| 3262 |
+
"step": 541
|
| 3263 |
+
},
|
| 3264 |
+
{
|
| 3265 |
+
"epoch": 0.37,
|
| 3266 |
+
"learning_rate": 0.0005636685552407931,
|
| 3267 |
+
"loss": 1.9279,
|
| 3268 |
+
"step": 542
|
| 3269 |
+
},
|
| 3270 |
+
{
|
| 3271 |
+
"epoch": 0.37,
|
| 3272 |
+
"learning_rate": 0.0005635623229461755,
|
| 3273 |
+
"loss": 1.8829,
|
| 3274 |
+
"step": 543
|
| 3275 |
+
},
|
| 3276 |
+
{
|
| 3277 |
+
"epoch": 0.37,
|
| 3278 |
+
"learning_rate": 0.000563456090651558,
|
| 3279 |
+
"loss": 1.9501,
|
| 3280 |
+
"step": 544
|
| 3281 |
+
},
|
| 3282 |
+
{
|
| 3283 |
+
"epoch": 0.37,
|
| 3284 |
+
"learning_rate": 0.0005633498583569404,
|
| 3285 |
+
"loss": 1.9602,
|
| 3286 |
+
"step": 545
|
| 3287 |
+
},
|
| 3288 |
+
{
|
| 3289 |
+
"epoch": 0.37,
|
| 3290 |
+
"learning_rate": 0.0005632436260623229,
|
| 3291 |
+
"loss": 1.9497,
|
| 3292 |
+
"step": 546
|
| 3293 |
+
},
|
| 3294 |
+
{
|
| 3295 |
+
"epoch": 0.37,
|
| 3296 |
+
"learning_rate": 0.0005631373937677053,
|
| 3297 |
+
"loss": 1.84,
|
| 3298 |
+
"step": 547
|
| 3299 |
+
},
|
| 3300 |
+
{
|
| 3301 |
+
"epoch": 0.37,
|
| 3302 |
+
"learning_rate": 0.0005630311614730878,
|
| 3303 |
+
"loss": 1.9337,
|
| 3304 |
+
"step": 548
|
| 3305 |
+
},
|
| 3306 |
+
{
|
| 3307 |
+
"epoch": 0.38,
|
| 3308 |
+
"learning_rate": 0.0005629249291784702,
|
| 3309 |
+
"loss": 1.8709,
|
| 3310 |
+
"step": 549
|
| 3311 |
+
},
|
| 3312 |
+
{
|
| 3313 |
+
"epoch": 0.38,
|
| 3314 |
+
"learning_rate": 0.0005628186968838527,
|
| 3315 |
+
"loss": 1.9231,
|
| 3316 |
+
"step": 550
|
| 3317 |
+
},
|
| 3318 |
+
{
|
| 3319 |
+
"epoch": 0.38,
|
| 3320 |
+
"learning_rate": 0.0005627124645892351,
|
| 3321 |
+
"loss": 1.9035,
|
| 3322 |
+
"step": 551
|
| 3323 |
+
},
|
| 3324 |
+
{
|
| 3325 |
+
"epoch": 0.38,
|
| 3326 |
+
"learning_rate": 0.0005626062322946176,
|
| 3327 |
+
"loss": 1.8365,
|
| 3328 |
+
"step": 552
|
| 3329 |
+
},
|
| 3330 |
+
{
|
| 3331 |
+
"epoch": 0.38,
|
| 3332 |
+
"learning_rate": 0.0005625,
|
| 3333 |
+
"loss": 2.1028,
|
| 3334 |
+
"step": 553
|
| 3335 |
+
},
|
| 3336 |
+
{
|
| 3337 |
+
"epoch": 0.38,
|
| 3338 |
+
"learning_rate": 0.0005623937677053823,
|
| 3339 |
+
"loss": 1.8692,
|
| 3340 |
+
"step": 554
|
| 3341 |
+
},
|
| 3342 |
+
{
|
| 3343 |
+
"epoch": 0.38,
|
| 3344 |
+
"learning_rate": 0.0005622875354107648,
|
| 3345 |
+
"loss": 1.9452,
|
| 3346 |
+
"step": 555
|
| 3347 |
+
},
|
| 3348 |
+
{
|
| 3349 |
+
"epoch": 0.38,
|
| 3350 |
+
"learning_rate": 0.0005621813031161472,
|
| 3351 |
+
"loss": 1.883,
|
| 3352 |
+
"step": 556
|
| 3353 |
+
},
|
| 3354 |
+
{
|
| 3355 |
+
"epoch": 0.38,
|
| 3356 |
+
"learning_rate": 0.0005620750708215296,
|
| 3357 |
+
"loss": 1.8987,
|
| 3358 |
+
"step": 557
|
| 3359 |
+
},
|
| 3360 |
+
{
|
| 3361 |
+
"epoch": 0.38,
|
| 3362 |
+
"learning_rate": 0.0005619688385269121,
|
| 3363 |
+
"loss": 1.9188,
|
| 3364 |
+
"step": 558
|
| 3365 |
+
},
|
| 3366 |
+
{
|
| 3367 |
+
"epoch": 0.38,
|
| 3368 |
+
"learning_rate": 0.0005618626062322946,
|
| 3369 |
+
"loss": 1.9373,
|
| 3370 |
+
"step": 559
|
| 3371 |
+
},
|
| 3372 |
+
{
|
| 3373 |
+
"epoch": 0.38,
|
| 3374 |
+
"learning_rate": 0.000561756373937677,
|
| 3375 |
+
"loss": 1.9167,
|
| 3376 |
+
"step": 560
|
| 3377 |
+
},
|
| 3378 |
+
{
|
| 3379 |
+
"epoch": 0.38,
|
| 3380 |
+
"learning_rate": 0.0005616501416430595,
|
| 3381 |
+
"loss": 1.8628,
|
| 3382 |
+
"step": 561
|
| 3383 |
+
},
|
| 3384 |
+
{
|
| 3385 |
+
"epoch": 0.38,
|
| 3386 |
+
"learning_rate": 0.0005615439093484419,
|
| 3387 |
+
"loss": 1.9258,
|
| 3388 |
+
"step": 562
|
| 3389 |
+
},
|
| 3390 |
+
{
|
| 3391 |
+
"epoch": 0.38,
|
| 3392 |
+
"learning_rate": 0.0005614376770538243,
|
| 3393 |
+
"loss": 1.9783,
|
| 3394 |
+
"step": 563
|
| 3395 |
+
},
|
| 3396 |
+
{
|
| 3397 |
+
"epoch": 0.39,
|
| 3398 |
+
"learning_rate": 0.0005613314447592068,
|
| 3399 |
+
"loss": 1.9738,
|
| 3400 |
+
"step": 564
|
| 3401 |
+
},
|
| 3402 |
+
{
|
| 3403 |
+
"epoch": 0.39,
|
| 3404 |
+
"learning_rate": 0.0005612252124645892,
|
| 3405 |
+
"loss": 1.9351,
|
| 3406 |
+
"step": 565
|
| 3407 |
+
},
|
| 3408 |
+
{
|
| 3409 |
+
"epoch": 0.39,
|
| 3410 |
+
"learning_rate": 0.0005611189801699717,
|
| 3411 |
+
"loss": 1.9109,
|
| 3412 |
+
"step": 566
|
| 3413 |
+
},
|
| 3414 |
+
{
|
| 3415 |
+
"epoch": 0.39,
|
| 3416 |
+
"learning_rate": 0.000561012747875354,
|
| 3417 |
+
"loss": 1.832,
|
| 3418 |
+
"step": 567
|
| 3419 |
+
},
|
| 3420 |
+
{
|
| 3421 |
+
"epoch": 0.39,
|
| 3422 |
+
"learning_rate": 0.0005609065155807364,
|
| 3423 |
+
"loss": 1.9258,
|
| 3424 |
+
"step": 568
|
| 3425 |
+
},
|
| 3426 |
+
{
|
| 3427 |
+
"epoch": 0.39,
|
| 3428 |
+
"learning_rate": 0.0005608002832861189,
|
| 3429 |
+
"loss": 1.9888,
|
| 3430 |
+
"step": 569
|
| 3431 |
+
},
|
| 3432 |
+
{
|
| 3433 |
+
"epoch": 0.39,
|
| 3434 |
+
"learning_rate": 0.0005606940509915014,
|
| 3435 |
+
"loss": 1.9195,
|
| 3436 |
+
"step": 570
|
| 3437 |
+
},
|
| 3438 |
+
{
|
| 3439 |
+
"epoch": 0.39,
|
| 3440 |
+
"learning_rate": 0.0005605878186968838,
|
| 3441 |
+
"loss": 1.8607,
|
| 3442 |
+
"step": 571
|
| 3443 |
+
},
|
| 3444 |
+
{
|
| 3445 |
+
"epoch": 0.39,
|
| 3446 |
+
"learning_rate": 0.0005604815864022663,
|
| 3447 |
+
"loss": 1.9868,
|
| 3448 |
+
"step": 572
|
| 3449 |
+
},
|
| 3450 |
+
{
|
| 3451 |
+
"epoch": 0.39,
|
| 3452 |
+
"learning_rate": 0.0005603753541076487,
|
| 3453 |
+
"loss": 1.885,
|
| 3454 |
+
"step": 573
|
| 3455 |
+
},
|
| 3456 |
+
{
|
| 3457 |
+
"epoch": 0.39,
|
| 3458 |
+
"learning_rate": 0.0005602691218130311,
|
| 3459 |
+
"loss": 1.9379,
|
| 3460 |
+
"step": 574
|
| 3461 |
+
},
|
| 3462 |
+
{
|
| 3463 |
+
"epoch": 0.39,
|
| 3464 |
+
"learning_rate": 0.0005601628895184136,
|
| 3465 |
+
"loss": 1.8685,
|
| 3466 |
+
"step": 575
|
| 3467 |
+
},
|
| 3468 |
+
{
|
| 3469 |
+
"epoch": 0.39,
|
| 3470 |
+
"learning_rate": 0.000560056657223796,
|
| 3471 |
+
"loss": 1.9636,
|
| 3472 |
+
"step": 576
|
| 3473 |
+
},
|
| 3474 |
+
{
|
| 3475 |
+
"epoch": 0.39,
|
| 3476 |
+
"learning_rate": 0.0005599504249291784,
|
| 3477 |
+
"loss": 1.9925,
|
| 3478 |
+
"step": 577
|
| 3479 |
+
},
|
| 3480 |
+
{
|
| 3481 |
+
"epoch": 0.4,
|
| 3482 |
+
"learning_rate": 0.0005598441926345609,
|
| 3483 |
+
"loss": 1.9332,
|
| 3484 |
+
"step": 578
|
| 3485 |
+
},
|
| 3486 |
+
{
|
| 3487 |
+
"epoch": 0.4,
|
| 3488 |
+
"learning_rate": 0.0005597379603399432,
|
| 3489 |
+
"loss": 1.8458,
|
| 3490 |
+
"step": 579
|
| 3491 |
+
},
|
| 3492 |
+
{
|
| 3493 |
+
"epoch": 0.4,
|
| 3494 |
+
"learning_rate": 0.0005596317280453257,
|
| 3495 |
+
"loss": 1.8204,
|
| 3496 |
+
"step": 580
|
| 3497 |
+
},
|
| 3498 |
+
{
|
| 3499 |
+
"epoch": 0.4,
|
| 3500 |
+
"learning_rate": 0.0005595254957507082,
|
| 3501 |
+
"loss": 1.8656,
|
| 3502 |
+
"step": 581
|
| 3503 |
+
},
|
| 3504 |
+
{
|
| 3505 |
+
"epoch": 0.4,
|
| 3506 |
+
"learning_rate": 0.0005594192634560906,
|
| 3507 |
+
"loss": 1.9399,
|
| 3508 |
+
"step": 582
|
| 3509 |
+
},
|
| 3510 |
+
{
|
| 3511 |
+
"epoch": 0.4,
|
| 3512 |
+
"learning_rate": 0.000559313031161473,
|
| 3513 |
+
"loss": 1.891,
|
| 3514 |
+
"step": 583
|
| 3515 |
+
},
|
| 3516 |
+
{
|
| 3517 |
+
"epoch": 0.4,
|
| 3518 |
+
"learning_rate": 0.0005592067988668555,
|
| 3519 |
+
"loss": 1.8595,
|
| 3520 |
+
"step": 584
|
| 3521 |
+
},
|
| 3522 |
+
{
|
| 3523 |
+
"epoch": 0.4,
|
| 3524 |
+
"learning_rate": 0.0005591005665722379,
|
| 3525 |
+
"loss": 1.8989,
|
| 3526 |
+
"step": 585
|
| 3527 |
+
},
|
| 3528 |
+
{
|
| 3529 |
+
"epoch": 0.4,
|
| 3530 |
+
"learning_rate": 0.0005589943342776204,
|
| 3531 |
+
"loss": 1.8275,
|
| 3532 |
+
"step": 586
|
| 3533 |
+
},
|
| 3534 |
+
{
|
| 3535 |
+
"epoch": 0.4,
|
| 3536 |
+
"learning_rate": 0.0005588881019830028,
|
| 3537 |
+
"loss": 1.8716,
|
| 3538 |
+
"step": 587
|
| 3539 |
+
},
|
| 3540 |
+
{
|
| 3541 |
+
"epoch": 0.4,
|
| 3542 |
+
"learning_rate": 0.0005587818696883852,
|
| 3543 |
+
"loss": 1.933,
|
| 3544 |
+
"step": 588
|
| 3545 |
+
},
|
| 3546 |
+
{
|
| 3547 |
+
"epoch": 0.4,
|
| 3548 |
+
"learning_rate": 0.0005586756373937677,
|
| 3549 |
+
"loss": 1.982,
|
| 3550 |
+
"step": 589
|
| 3551 |
+
},
|
| 3552 |
+
{
|
| 3553 |
+
"epoch": 0.4,
|
| 3554 |
+
"learning_rate": 0.00055856940509915,
|
| 3555 |
+
"loss": 1.8773,
|
| 3556 |
+
"step": 590
|
| 3557 |
+
},
|
| 3558 |
+
{
|
| 3559 |
+
"epoch": 0.4,
|
| 3560 |
+
"learning_rate": 0.0005584631728045325,
|
| 3561 |
+
"loss": 1.8793,
|
| 3562 |
+
"step": 591
|
| 3563 |
+
},
|
| 3564 |
+
{
|
| 3565 |
+
"epoch": 0.4,
|
| 3566 |
+
"learning_rate": 0.000558356940509915,
|
| 3567 |
+
"loss": 1.9778,
|
| 3568 |
+
"step": 592
|
| 3569 |
+
},
|
| 3570 |
+
{
|
| 3571 |
+
"epoch": 0.41,
|
| 3572 |
+
"learning_rate": 0.0005582507082152974,
|
| 3573 |
+
"loss": 1.9904,
|
| 3574 |
+
"step": 593
|
| 3575 |
+
},
|
| 3576 |
+
{
|
| 3577 |
+
"epoch": 0.41,
|
| 3578 |
+
"learning_rate": 0.0005581444759206798,
|
| 3579 |
+
"loss": 1.9301,
|
| 3580 |
+
"step": 594
|
| 3581 |
+
},
|
| 3582 |
+
{
|
| 3583 |
+
"epoch": 0.41,
|
| 3584 |
+
"learning_rate": 0.0005580382436260623,
|
| 3585 |
+
"loss": 1.9367,
|
| 3586 |
+
"step": 595
|
| 3587 |
+
},
|
| 3588 |
+
{
|
| 3589 |
+
"epoch": 0.41,
|
| 3590 |
+
"learning_rate": 0.0005579320113314447,
|
| 3591 |
+
"loss": 1.9853,
|
| 3592 |
+
"step": 596
|
| 3593 |
+
},
|
| 3594 |
+
{
|
| 3595 |
+
"epoch": 0.41,
|
| 3596 |
+
"learning_rate": 0.0005578257790368272,
|
| 3597 |
+
"loss": 1.9987,
|
| 3598 |
+
"step": 597
|
| 3599 |
+
},
|
| 3600 |
+
{
|
| 3601 |
+
"epoch": 0.41,
|
| 3602 |
+
"learning_rate": 0.0005577195467422096,
|
| 3603 |
+
"loss": 1.88,
|
| 3604 |
+
"step": 598
|
| 3605 |
+
},
|
| 3606 |
+
{
|
| 3607 |
+
"epoch": 0.41,
|
| 3608 |
+
"learning_rate": 0.000557613314447592,
|
| 3609 |
+
"loss": 1.9391,
|
| 3610 |
+
"step": 599
|
| 3611 |
+
},
|
| 3612 |
+
{
|
| 3613 |
+
"epoch": 0.41,
|
| 3614 |
+
"learning_rate": 0.0005575070821529745,
|
| 3615 |
+
"loss": 2.0134,
|
| 3616 |
+
"step": 600
|
| 3617 |
+
},
|
| 3618 |
+
{
|
| 3619 |
+
"epoch": 0.41,
|
| 3620 |
+
"learning_rate": 0.0005574008498583569,
|
| 3621 |
+
"loss": 1.8383,
|
| 3622 |
+
"step": 601
|
| 3623 |
+
},
|
| 3624 |
+
{
|
| 3625 |
+
"epoch": 0.41,
|
| 3626 |
+
"learning_rate": 0.0005572946175637392,
|
| 3627 |
+
"loss": 1.8189,
|
| 3628 |
+
"step": 602
|
| 3629 |
+
},
|
| 3630 |
+
{
|
| 3631 |
+
"epoch": 0.41,
|
| 3632 |
+
"learning_rate": 0.0005571883852691217,
|
| 3633 |
+
"loss": 1.9324,
|
| 3634 |
+
"step": 603
|
| 3635 |
+
},
|
| 3636 |
+
{
|
| 3637 |
+
"epoch": 0.41,
|
| 3638 |
+
"learning_rate": 0.0005570821529745042,
|
| 3639 |
+
"loss": 1.8702,
|
| 3640 |
+
"step": 604
|
| 3641 |
+
},
|
| 3642 |
+
{
|
| 3643 |
+
"epoch": 0.41,
|
| 3644 |
+
"learning_rate": 0.0005569759206798866,
|
| 3645 |
+
"loss": 1.8803,
|
| 3646 |
+
"step": 605
|
| 3647 |
+
},
|
| 3648 |
+
{
|
| 3649 |
+
"epoch": 0.41,
|
| 3650 |
+
"learning_rate": 0.0005568696883852691,
|
| 3651 |
+
"loss": 1.8409,
|
| 3652 |
+
"step": 606
|
| 3653 |
+
},
|
| 3654 |
+
{
|
| 3655 |
+
"epoch": 0.41,
|
| 3656 |
+
"learning_rate": 0.0005567634560906515,
|
| 3657 |
+
"loss": 1.8715,
|
| 3658 |
+
"step": 607
|
| 3659 |
+
},
|
| 3660 |
+
{
|
| 3661 |
+
"epoch": 0.42,
|
| 3662 |
+
"learning_rate": 0.0005566572237960339,
|
| 3663 |
+
"loss": 1.9359,
|
| 3664 |
+
"step": 608
|
| 3665 |
+
},
|
| 3666 |
+
{
|
| 3667 |
+
"epoch": 0.42,
|
| 3668 |
+
"learning_rate": 0.0005565509915014164,
|
| 3669 |
+
"loss": 1.922,
|
| 3670 |
+
"step": 609
|
| 3671 |
+
},
|
| 3672 |
+
{
|
| 3673 |
+
"epoch": 0.42,
|
| 3674 |
+
"learning_rate": 0.0005564447592067988,
|
| 3675 |
+
"loss": 1.8341,
|
| 3676 |
+
"step": 610
|
| 3677 |
+
},
|
| 3678 |
+
{
|
| 3679 |
+
"epoch": 0.42,
|
| 3680 |
+
"learning_rate": 0.0005563385269121813,
|
| 3681 |
+
"loss": 1.9494,
|
| 3682 |
+
"step": 611
|
| 3683 |
+
},
|
| 3684 |
+
{
|
| 3685 |
+
"epoch": 0.42,
|
| 3686 |
+
"learning_rate": 0.0005562322946175637,
|
| 3687 |
+
"loss": 1.9585,
|
| 3688 |
+
"step": 612
|
| 3689 |
+
},
|
| 3690 |
+
{
|
| 3691 |
+
"epoch": 0.42,
|
| 3692 |
+
"learning_rate": 0.0005561260623229461,
|
| 3693 |
+
"loss": 1.8668,
|
| 3694 |
+
"step": 613
|
| 3695 |
+
},
|
| 3696 |
+
{
|
| 3697 |
+
"epoch": 0.42,
|
| 3698 |
+
"learning_rate": 0.0005560198300283286,
|
| 3699 |
+
"loss": 1.9067,
|
| 3700 |
+
"step": 614
|
| 3701 |
+
},
|
| 3702 |
+
{
|
| 3703 |
+
"epoch": 0.42,
|
| 3704 |
+
"learning_rate": 0.000555913597733711,
|
| 3705 |
+
"loss": 1.8475,
|
| 3706 |
+
"step": 615
|
| 3707 |
+
},
|
| 3708 |
+
{
|
| 3709 |
+
"epoch": 0.42,
|
| 3710 |
+
"learning_rate": 0.0005558073654390934,
|
| 3711 |
+
"loss": 1.9009,
|
| 3712 |
+
"step": 616
|
| 3713 |
+
},
|
| 3714 |
+
{
|
| 3715 |
+
"epoch": 0.42,
|
| 3716 |
+
"learning_rate": 0.0005557011331444759,
|
| 3717 |
+
"loss": 1.941,
|
| 3718 |
+
"step": 617
|
| 3719 |
+
},
|
| 3720 |
+
{
|
| 3721 |
+
"epoch": 0.42,
|
| 3722 |
+
"learning_rate": 0.0005555949008498583,
|
| 3723 |
+
"loss": 1.9756,
|
| 3724 |
+
"step": 618
|
| 3725 |
+
},
|
| 3726 |
+
{
|
| 3727 |
+
"epoch": 0.42,
|
| 3728 |
+
"learning_rate": 0.0005554886685552407,
|
| 3729 |
+
"loss": 1.9568,
|
| 3730 |
+
"step": 619
|
| 3731 |
+
},
|
| 3732 |
+
{
|
| 3733 |
+
"epoch": 0.42,
|
| 3734 |
+
"learning_rate": 0.0005553824362606232,
|
| 3735 |
+
"loss": 1.8794,
|
| 3736 |
+
"step": 620
|
| 3737 |
+
},
|
| 3738 |
+
{
|
| 3739 |
+
"epoch": 0.42,
|
| 3740 |
+
"learning_rate": 0.0005552762039660056,
|
| 3741 |
+
"loss": 1.9152,
|
| 3742 |
+
"step": 621
|
| 3743 |
+
},
|
| 3744 |
+
{
|
| 3745 |
+
"epoch": 0.43,
|
| 3746 |
+
"learning_rate": 0.000555169971671388,
|
| 3747 |
+
"loss": 2.0171,
|
| 3748 |
+
"step": 622
|
| 3749 |
+
},
|
| 3750 |
+
{
|
| 3751 |
+
"epoch": 0.43,
|
| 3752 |
+
"learning_rate": 0.0005550637393767705,
|
| 3753 |
+
"loss": 1.9388,
|
| 3754 |
+
"step": 623
|
| 3755 |
+
},
|
| 3756 |
+
{
|
| 3757 |
+
"epoch": 0.43,
|
| 3758 |
+
"learning_rate": 0.0005549575070821529,
|
| 3759 |
+
"loss": 1.9048,
|
| 3760 |
+
"step": 624
|
| 3761 |
+
},
|
| 3762 |
+
{
|
| 3763 |
+
"epoch": 0.43,
|
| 3764 |
+
"learning_rate": 0.0005548512747875354,
|
| 3765 |
+
"loss": 1.9667,
|
| 3766 |
+
"step": 625
|
| 3767 |
+
},
|
| 3768 |
+
{
|
| 3769 |
+
"epoch": 0.43,
|
| 3770 |
+
"learning_rate": 0.0005547450424929179,
|
| 3771 |
+
"loss": 1.9878,
|
| 3772 |
+
"step": 626
|
| 3773 |
+
},
|
| 3774 |
+
{
|
| 3775 |
+
"epoch": 0.43,
|
| 3776 |
+
"learning_rate": 0.0005546388101983003,
|
| 3777 |
+
"loss": 1.9145,
|
| 3778 |
+
"step": 627
|
| 3779 |
+
},
|
| 3780 |
+
{
|
| 3781 |
+
"epoch": 0.43,
|
| 3782 |
+
"learning_rate": 0.0005545325779036826,
|
| 3783 |
+
"loss": 1.9873,
|
| 3784 |
+
"step": 628
|
| 3785 |
+
},
|
| 3786 |
+
{
|
| 3787 |
+
"epoch": 0.43,
|
| 3788 |
+
"learning_rate": 0.0005544263456090651,
|
| 3789 |
+
"loss": 2.015,
|
| 3790 |
+
"step": 629
|
| 3791 |
+
},
|
| 3792 |
+
{
|
| 3793 |
+
"epoch": 0.43,
|
| 3794 |
+
"learning_rate": 0.0005543201133144475,
|
| 3795 |
+
"loss": 1.9204,
|
| 3796 |
+
"step": 630
|
| 3797 |
+
},
|
| 3798 |
+
{
|
| 3799 |
+
"epoch": 0.43,
|
| 3800 |
+
"learning_rate": 0.00055421388101983,
|
| 3801 |
+
"loss": 1.8899,
|
| 3802 |
+
"step": 631
|
| 3803 |
+
},
|
| 3804 |
+
{
|
| 3805 |
+
"epoch": 0.43,
|
| 3806 |
+
"learning_rate": 0.0005541076487252124,
|
| 3807 |
+
"loss": 1.9057,
|
| 3808 |
+
"step": 632
|
| 3809 |
+
},
|
| 3810 |
+
{
|
| 3811 |
+
"epoch": 0.43,
|
| 3812 |
+
"learning_rate": 0.0005540014164305948,
|
| 3813 |
+
"loss": 1.9663,
|
| 3814 |
+
"step": 633
|
| 3815 |
+
},
|
| 3816 |
+
{
|
| 3817 |
+
"epoch": 0.43,
|
| 3818 |
+
"learning_rate": 0.0005538951841359773,
|
| 3819 |
+
"loss": 1.9447,
|
| 3820 |
+
"step": 634
|
| 3821 |
+
},
|
| 3822 |
+
{
|
| 3823 |
+
"epoch": 0.43,
|
| 3824 |
+
"learning_rate": 0.0005537889518413597,
|
| 3825 |
+
"loss": 1.957,
|
| 3826 |
+
"step": 635
|
| 3827 |
+
},
|
| 3828 |
+
{
|
| 3829 |
+
"epoch": 0.43,
|
| 3830 |
+
"learning_rate": 0.0005536827195467422,
|
| 3831 |
+
"loss": 1.8772,
|
| 3832 |
+
"step": 636
|
| 3833 |
+
},
|
| 3834 |
+
{
|
| 3835 |
+
"epoch": 0.44,
|
| 3836 |
+
"learning_rate": 0.0005535764872521247,
|
| 3837 |
+
"loss": 1.9601,
|
| 3838 |
+
"step": 637
|
| 3839 |
+
},
|
| 3840 |
+
{
|
| 3841 |
+
"epoch": 0.44,
|
| 3842 |
+
"learning_rate": 0.0005534702549575071,
|
| 3843 |
+
"loss": 1.8354,
|
| 3844 |
+
"step": 638
|
| 3845 |
+
},
|
| 3846 |
+
{
|
| 3847 |
+
"epoch": 0.44,
|
| 3848 |
+
"learning_rate": 0.0005533640226628895,
|
| 3849 |
+
"loss": 1.9107,
|
| 3850 |
+
"step": 639
|
| 3851 |
+
},
|
| 3852 |
+
{
|
| 3853 |
+
"epoch": 0.44,
|
| 3854 |
+
"learning_rate": 0.000553257790368272,
|
| 3855 |
+
"loss": 1.9511,
|
| 3856 |
+
"step": 640
|
| 3857 |
+
},
|
| 3858 |
+
{
|
| 3859 |
+
"epoch": 0.44,
|
| 3860 |
+
"learning_rate": 0.0005531515580736543,
|
| 3861 |
+
"loss": 1.9354,
|
| 3862 |
+
"step": 641
|
| 3863 |
+
},
|
| 3864 |
+
{
|
| 3865 |
+
"epoch": 0.44,
|
| 3866 |
+
"learning_rate": 0.0005530453257790367,
|
| 3867 |
+
"loss": 1.9189,
|
| 3868 |
+
"step": 642
|
| 3869 |
+
},
|
| 3870 |
+
{
|
| 3871 |
+
"epoch": 0.44,
|
| 3872 |
+
"learning_rate": 0.0005529390934844192,
|
| 3873 |
+
"loss": 1.803,
|
| 3874 |
+
"step": 643
|
| 3875 |
+
},
|
| 3876 |
+
{
|
| 3877 |
+
"epoch": 0.44,
|
| 3878 |
+
"learning_rate": 0.0005528328611898016,
|
| 3879 |
+
"loss": 1.9278,
|
| 3880 |
+
"step": 644
|
| 3881 |
+
},
|
| 3882 |
+
{
|
| 3883 |
+
"epoch": 0.44,
|
| 3884 |
+
"learning_rate": 0.0005527266288951841,
|
| 3885 |
+
"loss": 1.8629,
|
| 3886 |
+
"step": 645
|
| 3887 |
+
},
|
| 3888 |
+
{
|
| 3889 |
+
"epoch": 0.44,
|
| 3890 |
+
"learning_rate": 0.0005526203966005665,
|
| 3891 |
+
"loss": 1.9348,
|
| 3892 |
+
"step": 646
|
| 3893 |
+
},
|
| 3894 |
+
{
|
| 3895 |
+
"epoch": 0.44,
|
| 3896 |
+
"learning_rate": 0.000552514164305949,
|
| 3897 |
+
"loss": 1.9472,
|
| 3898 |
+
"step": 647
|
| 3899 |
+
},
|
| 3900 |
+
{
|
| 3901 |
+
"epoch": 0.44,
|
| 3902 |
+
"learning_rate": 0.0005524079320113314,
|
| 3903 |
+
"loss": 1.9808,
|
| 3904 |
+
"step": 648
|
| 3905 |
+
},
|
| 3906 |
+
{
|
| 3907 |
+
"epoch": 0.44,
|
| 3908 |
+
"learning_rate": 0.0005523016997167139,
|
| 3909 |
+
"loss": 1.8421,
|
| 3910 |
+
"step": 649
|
| 3911 |
+
},
|
| 3912 |
+
{
|
| 3913 |
+
"epoch": 0.44,
|
| 3914 |
+
"learning_rate": 0.0005521954674220963,
|
| 3915 |
+
"loss": 1.9773,
|
| 3916 |
+
"step": 650
|
| 3917 |
+
},
|
| 3918 |
+
{
|
| 3919 |
+
"epoch": 0.45,
|
| 3920 |
+
"learning_rate": 0.0005520892351274788,
|
| 3921 |
+
"loss": 1.9566,
|
| 3922 |
+
"step": 651
|
| 3923 |
+
},
|
| 3924 |
+
{
|
| 3925 |
+
"epoch": 0.45,
|
| 3926 |
+
"learning_rate": 0.0005519830028328611,
|
| 3927 |
+
"loss": 1.9461,
|
| 3928 |
+
"step": 652
|
| 3929 |
+
},
|
| 3930 |
+
{
|
| 3931 |
+
"epoch": 0.45,
|
| 3932 |
+
"learning_rate": 0.0005518767705382435,
|
| 3933 |
+
"loss": 1.9117,
|
| 3934 |
+
"step": 653
|
| 3935 |
+
},
|
| 3936 |
+
{
|
| 3937 |
+
"epoch": 0.45,
|
| 3938 |
+
"learning_rate": 0.000551770538243626,
|
| 3939 |
+
"loss": 1.8791,
|
| 3940 |
+
"step": 654
|
| 3941 |
+
},
|
| 3942 |
+
{
|
| 3943 |
+
"epoch": 0.45,
|
| 3944 |
+
"learning_rate": 0.0005516643059490084,
|
| 3945 |
+
"loss": 1.9065,
|
| 3946 |
+
"step": 655
|
| 3947 |
+
},
|
| 3948 |
+
{
|
| 3949 |
+
"epoch": 0.45,
|
| 3950 |
+
"learning_rate": 0.0005515580736543909,
|
| 3951 |
+
"loss": 1.9535,
|
| 3952 |
+
"step": 656
|
| 3953 |
+
},
|
| 3954 |
+
{
|
| 3955 |
+
"epoch": 0.45,
|
| 3956 |
+
"learning_rate": 0.0005514518413597733,
|
| 3957 |
+
"loss": 1.8854,
|
| 3958 |
+
"step": 657
|
| 3959 |
+
},
|
| 3960 |
+
{
|
| 3961 |
+
"epoch": 0.45,
|
| 3962 |
+
"learning_rate": 0.0005513456090651558,
|
| 3963 |
+
"loss": 2.023,
|
| 3964 |
+
"step": 658
|
| 3965 |
+
},
|
| 3966 |
+
{
|
| 3967 |
+
"epoch": 0.45,
|
| 3968 |
+
"learning_rate": 0.0005512393767705382,
|
| 3969 |
+
"loss": 2.0061,
|
| 3970 |
+
"step": 659
|
| 3971 |
+
},
|
| 3972 |
+
{
|
| 3973 |
+
"epoch": 0.45,
|
| 3974 |
+
"learning_rate": 0.0005511331444759207,
|
| 3975 |
+
"loss": 1.8768,
|
| 3976 |
+
"step": 660
|
| 3977 |
+
},
|
| 3978 |
+
{
|
| 3979 |
+
"epoch": 0.45,
|
| 3980 |
+
"learning_rate": 0.0005510269121813031,
|
| 3981 |
+
"loss": 1.948,
|
| 3982 |
+
"step": 661
|
| 3983 |
+
},
|
| 3984 |
+
{
|
| 3985 |
+
"epoch": 0.45,
|
| 3986 |
+
"learning_rate": 0.0005509206798866855,
|
| 3987 |
+
"loss": 1.9478,
|
| 3988 |
+
"step": 662
|
| 3989 |
+
},
|
| 3990 |
+
{
|
| 3991 |
+
"epoch": 0.45,
|
| 3992 |
+
"learning_rate": 0.000550814447592068,
|
| 3993 |
+
"loss": 1.977,
|
| 3994 |
+
"step": 663
|
| 3995 |
+
},
|
| 3996 |
+
{
|
| 3997 |
+
"epoch": 0.45,
|
| 3998 |
+
"learning_rate": 0.0005507082152974503,
|
| 3999 |
+
"loss": 1.9208,
|
| 4000 |
+
"step": 664
|
| 4001 |
+
},
|
| 4002 |
+
{
|
| 4003 |
+
"epoch": 0.45,
|
| 4004 |
+
"learning_rate": 0.0005506019830028328,
|
| 4005 |
+
"loss": 1.9287,
|
| 4006 |
+
"step": 665
|
| 4007 |
+
},
|
| 4008 |
+
{
|
| 4009 |
+
"epoch": 0.46,
|
| 4010 |
+
"learning_rate": 0.0005504957507082152,
|
| 4011 |
+
"loss": 1.8817,
|
| 4012 |
+
"step": 666
|
| 4013 |
+
},
|
| 4014 |
+
{
|
| 4015 |
+
"epoch": 0.46,
|
| 4016 |
+
"learning_rate": 0.0005503895184135976,
|
| 4017 |
+
"loss": 1.9567,
|
| 4018 |
+
"step": 667
|
| 4019 |
+
},
|
| 4020 |
+
{
|
| 4021 |
+
"epoch": 0.46,
|
| 4022 |
+
"learning_rate": 0.0005502832861189801,
|
| 4023 |
+
"loss": 1.8139,
|
| 4024 |
+
"step": 668
|
| 4025 |
+
},
|
| 4026 |
+
{
|
| 4027 |
+
"epoch": 0.46,
|
| 4028 |
+
"learning_rate": 0.0005501770538243626,
|
| 4029 |
+
"loss": 1.913,
|
| 4030 |
+
"step": 669
|
| 4031 |
+
},
|
| 4032 |
+
{
|
| 4033 |
+
"epoch": 0.46,
|
| 4034 |
+
"learning_rate": 0.000550070821529745,
|
| 4035 |
+
"loss": 1.9434,
|
| 4036 |
+
"step": 670
|
| 4037 |
+
},
|
| 4038 |
+
{
|
| 4039 |
+
"epoch": 0.46,
|
| 4040 |
+
"learning_rate": 0.0005499645892351275,
|
| 4041 |
+
"loss": 1.9377,
|
| 4042 |
+
"step": 671
|
| 4043 |
+
},
|
| 4044 |
+
{
|
| 4045 |
+
"epoch": 0.46,
|
| 4046 |
+
"learning_rate": 0.0005498583569405099,
|
| 4047 |
+
"loss": 1.8781,
|
| 4048 |
+
"step": 672
|
| 4049 |
+
},
|
| 4050 |
+
{
|
| 4051 |
+
"epoch": 0.46,
|
| 4052 |
+
"learning_rate": 0.0005497521246458923,
|
| 4053 |
+
"loss": 1.8837,
|
| 4054 |
+
"step": 673
|
| 4055 |
+
},
|
| 4056 |
+
{
|
| 4057 |
+
"epoch": 0.46,
|
| 4058 |
+
"learning_rate": 0.0005496458923512748,
|
| 4059 |
+
"loss": 1.9478,
|
| 4060 |
+
"step": 674
|
| 4061 |
+
},
|
| 4062 |
+
{
|
| 4063 |
+
"epoch": 0.46,
|
| 4064 |
+
"learning_rate": 0.0005495396600566572,
|
| 4065 |
+
"loss": 1.8756,
|
| 4066 |
+
"step": 675
|
| 4067 |
+
},
|
| 4068 |
+
{
|
| 4069 |
+
"epoch": 0.46,
|
| 4070 |
+
"learning_rate": 0.0005494334277620397,
|
| 4071 |
+
"loss": 1.9507,
|
| 4072 |
+
"step": 676
|
| 4073 |
+
},
|
| 4074 |
+
{
|
| 4075 |
+
"epoch": 0.46,
|
| 4076 |
+
"learning_rate": 0.000549327195467422,
|
| 4077 |
+
"loss": 1.9027,
|
| 4078 |
+
"step": 677
|
| 4079 |
+
},
|
| 4080 |
+
{
|
| 4081 |
+
"epoch": 0.46,
|
| 4082 |
+
"learning_rate": 0.0005492209631728044,
|
| 4083 |
+
"loss": 1.9749,
|
| 4084 |
+
"step": 678
|
| 4085 |
+
},
|
| 4086 |
+
{
|
| 4087 |
+
"epoch": 0.46,
|
| 4088 |
+
"learning_rate": 0.0005491147308781869,
|
| 4089 |
+
"loss": 1.8564,
|
| 4090 |
+
"step": 679
|
| 4091 |
+
},
|
| 4092 |
+
{
|
| 4093 |
+
"epoch": 0.46,
|
| 4094 |
+
"learning_rate": 0.0005490084985835694,
|
| 4095 |
+
"loss": 1.9649,
|
| 4096 |
+
"step": 680
|
| 4097 |
+
},
|
| 4098 |
+
{
|
| 4099 |
+
"epoch": 0.47,
|
| 4100 |
+
"learning_rate": 0.0005489022662889518,
|
| 4101 |
+
"loss": 2.0128,
|
| 4102 |
+
"step": 681
|
| 4103 |
+
},
|
| 4104 |
+
{
|
| 4105 |
+
"epoch": 0.47,
|
| 4106 |
+
"learning_rate": 0.0005487960339943342,
|
| 4107 |
+
"loss": 1.9041,
|
| 4108 |
+
"step": 682
|
| 4109 |
+
},
|
| 4110 |
+
{
|
| 4111 |
+
"epoch": 0.47,
|
| 4112 |
+
"learning_rate": 0.0005486898016997167,
|
| 4113 |
+
"loss": 2.0121,
|
| 4114 |
+
"step": 683
|
| 4115 |
+
},
|
| 4116 |
+
{
|
| 4117 |
+
"epoch": 0.47,
|
| 4118 |
+
"learning_rate": 0.0005485835694050991,
|
| 4119 |
+
"loss": 1.9203,
|
| 4120 |
+
"step": 684
|
| 4121 |
+
},
|
| 4122 |
+
{
|
| 4123 |
+
"epoch": 0.47,
|
| 4124 |
+
"learning_rate": 0.0005484773371104816,
|
| 4125 |
+
"loss": 1.9,
|
| 4126 |
+
"step": 685
|
| 4127 |
+
},
|
| 4128 |
+
{
|
| 4129 |
+
"epoch": 0.47,
|
| 4130 |
+
"learning_rate": 0.000548371104815864,
|
| 4131 |
+
"loss": 1.9857,
|
| 4132 |
+
"step": 686
|
| 4133 |
+
},
|
| 4134 |
+
{
|
| 4135 |
+
"epoch": 0.47,
|
| 4136 |
+
"learning_rate": 0.0005482648725212464,
|
| 4137 |
+
"loss": 2.0838,
|
| 4138 |
+
"step": 687
|
| 4139 |
+
},
|
| 4140 |
+
{
|
| 4141 |
+
"epoch": 0.47,
|
| 4142 |
+
"learning_rate": 0.0005481586402266289,
|
| 4143 |
+
"loss": 1.8973,
|
| 4144 |
+
"step": 688
|
| 4145 |
+
},
|
| 4146 |
+
{
|
| 4147 |
+
"epoch": 0.47,
|
| 4148 |
+
"learning_rate": 0.0005480524079320112,
|
| 4149 |
+
"loss": 1.9636,
|
| 4150 |
+
"step": 689
|
| 4151 |
+
},
|
| 4152 |
+
{
|
| 4153 |
+
"epoch": 0.47,
|
| 4154 |
+
"learning_rate": 0.0005479461756373937,
|
| 4155 |
+
"loss": 1.9726,
|
| 4156 |
+
"step": 690
|
| 4157 |
+
},
|
| 4158 |
+
{
|
| 4159 |
+
"epoch": 0.47,
|
| 4160 |
+
"learning_rate": 0.0005478399433427762,
|
| 4161 |
+
"loss": 1.9062,
|
| 4162 |
+
"step": 691
|
| 4163 |
+
},
|
| 4164 |
+
{
|
| 4165 |
+
"epoch": 0.47,
|
| 4166 |
+
"learning_rate": 0.0005477337110481586,
|
| 4167 |
+
"loss": 1.8709,
|
| 4168 |
+
"step": 692
|
| 4169 |
+
},
|
| 4170 |
+
{
|
| 4171 |
+
"epoch": 0.47,
|
| 4172 |
+
"learning_rate": 0.000547627478753541,
|
| 4173 |
+
"loss": 1.9234,
|
| 4174 |
+
"step": 693
|
| 4175 |
+
},
|
| 4176 |
+
{
|
| 4177 |
+
"epoch": 0.47,
|
| 4178 |
+
"learning_rate": 0.0005475212464589235,
|
| 4179 |
+
"loss": 1.8859,
|
| 4180 |
+
"step": 694
|
| 4181 |
+
},
|
| 4182 |
+
{
|
| 4183 |
+
"epoch": 0.48,
|
| 4184 |
+
"learning_rate": 0.0005474150141643059,
|
| 4185 |
+
"loss": 1.9746,
|
| 4186 |
+
"step": 695
|
| 4187 |
+
},
|
| 4188 |
+
{
|
| 4189 |
+
"epoch": 0.48,
|
| 4190 |
+
"learning_rate": 0.0005473087818696884,
|
| 4191 |
+
"loss": 1.9229,
|
| 4192 |
+
"step": 696
|
| 4193 |
+
},
|
| 4194 |
+
{
|
| 4195 |
+
"epoch": 0.48,
|
| 4196 |
+
"learning_rate": 0.0005472025495750708,
|
| 4197 |
+
"loss": 1.8827,
|
| 4198 |
+
"step": 697
|
| 4199 |
+
},
|
| 4200 |
+
{
|
| 4201 |
+
"epoch": 0.48,
|
| 4202 |
+
"learning_rate": 0.0005470963172804532,
|
| 4203 |
+
"loss": 1.9455,
|
| 4204 |
+
"step": 698
|
| 4205 |
+
},
|
| 4206 |
+
{
|
| 4207 |
+
"epoch": 0.48,
|
| 4208 |
+
"learning_rate": 0.0005469900849858357,
|
| 4209 |
+
"loss": 1.8779,
|
| 4210 |
+
"step": 699
|
| 4211 |
+
},
|
| 4212 |
+
{
|
| 4213 |
+
"epoch": 0.48,
|
| 4214 |
+
"learning_rate": 0.000546883852691218,
|
| 4215 |
+
"loss": 1.9151,
|
| 4216 |
+
"step": 700
|
| 4217 |
+
},
|
| 4218 |
+
{
|
| 4219 |
+
"epoch": 0.48,
|
| 4220 |
+
"learning_rate": 0.0005467776203966004,
|
| 4221 |
+
"loss": 1.8283,
|
| 4222 |
+
"step": 701
|
| 4223 |
+
},
|
| 4224 |
+
{
|
| 4225 |
+
"epoch": 0.48,
|
| 4226 |
+
"learning_rate": 0.0005466713881019829,
|
| 4227 |
+
"loss": 1.9657,
|
| 4228 |
+
"step": 702
|
| 4229 |
+
},
|
| 4230 |
+
{
|
| 4231 |
+
"epoch": 0.48,
|
| 4232 |
+
"learning_rate": 0.0005465651558073654,
|
| 4233 |
+
"loss": 2.0174,
|
| 4234 |
+
"step": 703
|
| 4235 |
+
},
|
| 4236 |
+
{
|
| 4237 |
+
"epoch": 0.48,
|
| 4238 |
+
"learning_rate": 0.0005464589235127478,
|
| 4239 |
+
"loss": 1.9462,
|
| 4240 |
+
"step": 704
|
| 4241 |
+
},
|
| 4242 |
+
{
|
| 4243 |
+
"epoch": 0.48,
|
| 4244 |
+
"learning_rate": 0.0005463526912181303,
|
| 4245 |
+
"loss": 1.9257,
|
| 4246 |
+
"step": 705
|
| 4247 |
+
},
|
| 4248 |
+
{
|
| 4249 |
+
"epoch": 0.48,
|
| 4250 |
+
"learning_rate": 0.0005462464589235127,
|
| 4251 |
+
"loss": 1.8828,
|
| 4252 |
+
"step": 706
|
| 4253 |
+
},
|
| 4254 |
+
{
|
| 4255 |
+
"epoch": 0.48,
|
| 4256 |
+
"learning_rate": 0.0005461402266288951,
|
| 4257 |
+
"loss": 1.8831,
|
| 4258 |
+
"step": 707
|
| 4259 |
+
},
|
| 4260 |
+
{
|
| 4261 |
+
"epoch": 0.48,
|
| 4262 |
+
"learning_rate": 0.0005460339943342776,
|
| 4263 |
+
"loss": 1.9266,
|
| 4264 |
+
"step": 708
|
| 4265 |
+
},
|
| 4266 |
+
{
|
| 4267 |
+
"epoch": 0.48,
|
| 4268 |
+
"learning_rate": 0.00054592776203966,
|
| 4269 |
+
"loss": 2.0017,
|
| 4270 |
+
"step": 709
|
| 4271 |
+
},
|
| 4272 |
+
{
|
| 4273 |
+
"epoch": 0.49,
|
| 4274 |
+
"learning_rate": 0.0005458215297450425,
|
| 4275 |
+
"loss": 1.9537,
|
| 4276 |
+
"step": 710
|
| 4277 |
+
},
|
| 4278 |
+
{
|
| 4279 |
+
"epoch": 0.49,
|
| 4280 |
+
"learning_rate": 0.0005457152974504249,
|
| 4281 |
+
"loss": 1.9141,
|
| 4282 |
+
"step": 711
|
| 4283 |
+
},
|
| 4284 |
+
{
|
| 4285 |
+
"epoch": 0.49,
|
| 4286 |
+
"learning_rate": 0.0005456090651558072,
|
| 4287 |
+
"loss": 1.9393,
|
| 4288 |
+
"step": 712
|
| 4289 |
+
},
|
| 4290 |
+
{
|
| 4291 |
+
"epoch": 0.49,
|
| 4292 |
+
"learning_rate": 0.0005455028328611897,
|
| 4293 |
+
"loss": 1.8896,
|
| 4294 |
+
"step": 713
|
| 4295 |
+
},
|
| 4296 |
+
{
|
| 4297 |
+
"epoch": 0.49,
|
| 4298 |
+
"learning_rate": 0.0005453966005665722,
|
| 4299 |
+
"loss": 1.8186,
|
| 4300 |
+
"step": 714
|
| 4301 |
+
},
|
| 4302 |
+
{
|
| 4303 |
+
"epoch": 0.49,
|
| 4304 |
+
"learning_rate": 0.0005452903682719546,
|
| 4305 |
+
"loss": 1.9527,
|
| 4306 |
+
"step": 715
|
| 4307 |
+
},
|
| 4308 |
+
{
|
| 4309 |
+
"epoch": 0.49,
|
| 4310 |
+
"learning_rate": 0.0005451841359773371,
|
| 4311 |
+
"loss": 2.0319,
|
| 4312 |
+
"step": 716
|
| 4313 |
+
},
|
| 4314 |
+
{
|
| 4315 |
+
"epoch": 0.49,
|
| 4316 |
+
"learning_rate": 0.0005450779036827195,
|
| 4317 |
+
"loss": 1.8668,
|
| 4318 |
+
"step": 717
|
| 4319 |
+
},
|
| 4320 |
+
{
|
| 4321 |
+
"epoch": 0.49,
|
| 4322 |
+
"learning_rate": 0.0005449716713881019,
|
| 4323 |
+
"loss": 1.9775,
|
| 4324 |
+
"step": 718
|
| 4325 |
+
},
|
| 4326 |
+
{
|
| 4327 |
+
"epoch": 0.49,
|
| 4328 |
+
"learning_rate": 0.0005448654390934844,
|
| 4329 |
+
"loss": 1.8624,
|
| 4330 |
+
"step": 719
|
| 4331 |
+
},
|
| 4332 |
+
{
|
| 4333 |
+
"epoch": 0.49,
|
| 4334 |
+
"learning_rate": 0.0005447592067988668,
|
| 4335 |
+
"loss": 1.8725,
|
| 4336 |
+
"step": 720
|
| 4337 |
+
},
|
| 4338 |
+
{
|
| 4339 |
+
"epoch": 0.49,
|
| 4340 |
+
"learning_rate": 0.0005446529745042492,
|
| 4341 |
+
"loss": 1.8682,
|
| 4342 |
+
"step": 721
|
| 4343 |
+
},
|
| 4344 |
+
{
|
| 4345 |
+
"epoch": 0.49,
|
| 4346 |
+
"learning_rate": 0.0005445467422096317,
|
| 4347 |
+
"loss": 1.9078,
|
| 4348 |
+
"step": 722
|
| 4349 |
+
},
|
| 4350 |
+
{
|
| 4351 |
+
"epoch": 0.49,
|
| 4352 |
+
"learning_rate": 0.0005444405099150141,
|
| 4353 |
+
"loss": 1.8206,
|
| 4354 |
+
"step": 723
|
| 4355 |
+
},
|
| 4356 |
+
{
|
| 4357 |
+
"epoch": 0.49,
|
| 4358 |
+
"learning_rate": 0.0005443342776203966,
|
| 4359 |
+
"loss": 1.9923,
|
| 4360 |
+
"step": 724
|
| 4361 |
+
},
|
| 4362 |
+
{
|
| 4363 |
+
"epoch": 0.5,
|
| 4364 |
+
"learning_rate": 0.000544228045325779,
|
| 4365 |
+
"loss": 1.9765,
|
| 4366 |
+
"step": 725
|
| 4367 |
+
},
|
| 4368 |
+
{
|
| 4369 |
+
"epoch": 0.5,
|
| 4370 |
+
"learning_rate": 0.0005441218130311614,
|
| 4371 |
+
"loss": 1.9154,
|
| 4372 |
+
"step": 726
|
| 4373 |
+
},
|
| 4374 |
+
{
|
| 4375 |
+
"epoch": 0.5,
|
| 4376 |
+
"learning_rate": 0.0005440155807365438,
|
| 4377 |
+
"loss": 1.8111,
|
| 4378 |
+
"step": 727
|
| 4379 |
+
},
|
| 4380 |
+
{
|
| 4381 |
+
"epoch": 0.5,
|
| 4382 |
+
"learning_rate": 0.0005439093484419263,
|
| 4383 |
+
"loss": 1.9149,
|
| 4384 |
+
"step": 728
|
| 4385 |
+
},
|
| 4386 |
+
{
|
| 4387 |
+
"epoch": 0.5,
|
| 4388 |
+
"learning_rate": 0.0005438031161473087,
|
| 4389 |
+
"loss": 1.927,
|
| 4390 |
+
"step": 729
|
| 4391 |
+
},
|
| 4392 |
+
{
|
| 4393 |
+
"epoch": 0.5,
|
| 4394 |
+
"learning_rate": 0.0005436968838526912,
|
| 4395 |
+
"loss": 1.855,
|
| 4396 |
+
"step": 730
|
| 4397 |
+
},
|
| 4398 |
+
{
|
| 4399 |
+
"epoch": 0.5,
|
| 4400 |
+
"learning_rate": 0.0005435906515580736,
|
| 4401 |
+
"loss": 1.8842,
|
| 4402 |
+
"step": 731
|
| 4403 |
+
},
|
| 4404 |
+
{
|
| 4405 |
+
"epoch": 0.5,
|
| 4406 |
+
"learning_rate": 0.000543484419263456,
|
| 4407 |
+
"loss": 1.955,
|
| 4408 |
+
"step": 732
|
| 4409 |
+
},
|
| 4410 |
+
{
|
| 4411 |
+
"epoch": 0.5,
|
| 4412 |
+
"learning_rate": 0.0005433781869688385,
|
| 4413 |
+
"loss": 1.9599,
|
| 4414 |
+
"step": 733
|
| 4415 |
+
},
|
| 4416 |
+
{
|
| 4417 |
+
"epoch": 0.5,
|
| 4418 |
+
"learning_rate": 0.0005432719546742209,
|
| 4419 |
+
"loss": 1.8462,
|
| 4420 |
+
"step": 734
|
| 4421 |
+
},
|
| 4422 |
+
{
|
| 4423 |
+
"epoch": 0.5,
|
| 4424 |
+
"learning_rate": 0.0005431657223796034,
|
| 4425 |
+
"loss": 1.7453,
|
| 4426 |
+
"step": 735
|
| 4427 |
+
},
|
| 4428 |
+
{
|
| 4429 |
+
"epoch": 0.5,
|
| 4430 |
+
"learning_rate": 0.0005430594900849859,
|
| 4431 |
+
"loss": 1.8518,
|
| 4432 |
+
"step": 736
|
| 4433 |
+
},
|
| 4434 |
+
{
|
| 4435 |
+
"epoch": 0.5,
|
| 4436 |
+
"learning_rate": 0.0005429532577903683,
|
| 4437 |
+
"loss": 1.7791,
|
| 4438 |
+
"step": 737
|
| 4439 |
+
},
|
| 4440 |
+
{
|
| 4441 |
+
"epoch": 0.5,
|
| 4442 |
+
"learning_rate": 0.0005428470254957506,
|
| 4443 |
+
"loss": 1.918,
|
| 4444 |
+
"step": 738
|
| 4445 |
+
},
|
| 4446 |
+
{
|
| 4447 |
+
"epoch": 0.51,
|
| 4448 |
+
"learning_rate": 0.0005427407932011331,
|
| 4449 |
+
"loss": 1.8775,
|
| 4450 |
+
"step": 739
|
| 4451 |
+
},
|
| 4452 |
+
{
|
| 4453 |
+
"epoch": 0.51,
|
| 4454 |
+
"learning_rate": 0.0005426345609065155,
|
| 4455 |
+
"loss": 1.9502,
|
| 4456 |
+
"step": 740
|
| 4457 |
+
},
|
| 4458 |
+
{
|
| 4459 |
+
"epoch": 0.51,
|
| 4460 |
+
"learning_rate": 0.0005425283286118979,
|
| 4461 |
+
"loss": 1.9256,
|
| 4462 |
+
"step": 741
|
| 4463 |
+
},
|
| 4464 |
+
{
|
| 4465 |
+
"epoch": 0.51,
|
| 4466 |
+
"learning_rate": 0.0005424220963172804,
|
| 4467 |
+
"loss": 1.9275,
|
| 4468 |
+
"step": 742
|
| 4469 |
+
},
|
| 4470 |
+
{
|
| 4471 |
+
"epoch": 0.51,
|
| 4472 |
+
"learning_rate": 0.0005423158640226628,
|
| 4473 |
+
"loss": 1.9383,
|
| 4474 |
+
"step": 743
|
| 4475 |
+
},
|
| 4476 |
+
{
|
| 4477 |
+
"epoch": 0.51,
|
| 4478 |
+
"learning_rate": 0.0005422096317280453,
|
| 4479 |
+
"loss": 1.8964,
|
| 4480 |
+
"step": 744
|
| 4481 |
+
},
|
| 4482 |
+
{
|
| 4483 |
+
"epoch": 0.51,
|
| 4484 |
+
"learning_rate": 0.0005421033994334277,
|
| 4485 |
+
"loss": 1.9313,
|
| 4486 |
+
"step": 745
|
| 4487 |
+
},
|
| 4488 |
+
{
|
| 4489 |
+
"epoch": 0.51,
|
| 4490 |
+
"learning_rate": 0.0005419971671388101,
|
| 4491 |
+
"loss": 1.8931,
|
| 4492 |
+
"step": 746
|
| 4493 |
+
},
|
| 4494 |
+
{
|
| 4495 |
+
"epoch": 0.51,
|
| 4496 |
+
"learning_rate": 0.0005418909348441926,
|
| 4497 |
+
"loss": 1.9748,
|
| 4498 |
+
"step": 747
|
| 4499 |
+
},
|
| 4500 |
+
{
|
| 4501 |
+
"epoch": 0.51,
|
| 4502 |
+
"learning_rate": 0.0005417847025495751,
|
| 4503 |
+
"loss": 1.985,
|
| 4504 |
+
"step": 748
|
| 4505 |
+
},
|
| 4506 |
+
{
|
| 4507 |
+
"epoch": 0.51,
|
| 4508 |
+
"learning_rate": 0.0005416784702549575,
|
| 4509 |
+
"loss": 1.8515,
|
| 4510 |
+
"step": 749
|
| 4511 |
+
},
|
| 4512 |
+
{
|
| 4513 |
+
"epoch": 0.51,
|
| 4514 |
+
"learning_rate": 0.00054157223796034,
|
| 4515 |
+
"loss": 1.8717,
|
| 4516 |
+
"step": 750
|
| 4517 |
+
},
|
| 4518 |
+
{
|
| 4519 |
+
"epoch": 0.51,
|
| 4520 |
+
"learning_rate": 0.0005414660056657223,
|
| 4521 |
+
"loss": 1.9044,
|
| 4522 |
+
"step": 751
|
| 4523 |
+
},
|
| 4524 |
+
{
|
| 4525 |
+
"epoch": 0.51,
|
| 4526 |
+
"learning_rate": 0.0005413597733711047,
|
| 4527 |
+
"loss": 1.9972,
|
| 4528 |
+
"step": 752
|
| 4529 |
+
},
|
| 4530 |
+
{
|
| 4531 |
+
"epoch": 0.51,
|
| 4532 |
+
"learning_rate": 0.0005412535410764872,
|
| 4533 |
+
"loss": 1.8846,
|
| 4534 |
+
"step": 753
|
| 4535 |
+
},
|
| 4536 |
+
{
|
| 4537 |
+
"epoch": 0.52,
|
| 4538 |
+
"learning_rate": 0.0005411473087818696,
|
| 4539 |
+
"loss": 1.8465,
|
| 4540 |
+
"step": 754
|
| 4541 |
+
},
|
| 4542 |
+
{
|
| 4543 |
+
"epoch": 0.52,
|
| 4544 |
+
"learning_rate": 0.0005410410764872521,
|
| 4545 |
+
"loss": 1.9254,
|
| 4546 |
+
"step": 755
|
| 4547 |
+
},
|
| 4548 |
+
{
|
| 4549 |
+
"epoch": 0.52,
|
| 4550 |
+
"learning_rate": 0.0005409348441926345,
|
| 4551 |
+
"loss": 1.89,
|
| 4552 |
+
"step": 756
|
| 4553 |
+
},
|
| 4554 |
+
{
|
| 4555 |
+
"epoch": 0.52,
|
| 4556 |
+
"learning_rate": 0.0005408286118980169,
|
| 4557 |
+
"loss": 1.9923,
|
| 4558 |
+
"step": 757
|
| 4559 |
+
},
|
| 4560 |
+
{
|
| 4561 |
+
"epoch": 0.52,
|
| 4562 |
+
"learning_rate": 0.0005407223796033994,
|
| 4563 |
+
"loss": 1.8325,
|
| 4564 |
+
"step": 758
|
| 4565 |
+
},
|
| 4566 |
+
{
|
| 4567 |
+
"epoch": 0.52,
|
| 4568 |
+
"learning_rate": 0.0005406161473087819,
|
| 4569 |
+
"loss": 2.023,
|
| 4570 |
+
"step": 759
|
| 4571 |
+
},
|
| 4572 |
+
{
|
| 4573 |
+
"epoch": 0.52,
|
| 4574 |
+
"learning_rate": 0.0005405099150141643,
|
| 4575 |
+
"loss": 1.8279,
|
| 4576 |
+
"step": 760
|
| 4577 |
+
},
|
| 4578 |
+
{
|
| 4579 |
+
"epoch": 0.52,
|
| 4580 |
+
"learning_rate": 0.0005404036827195467,
|
| 4581 |
+
"loss": 1.9296,
|
| 4582 |
+
"step": 761
|
| 4583 |
+
},
|
| 4584 |
+
{
|
| 4585 |
+
"epoch": 0.52,
|
| 4586 |
+
"learning_rate": 0.0005402974504249291,
|
| 4587 |
+
"loss": 1.8728,
|
| 4588 |
+
"step": 762
|
| 4589 |
+
},
|
| 4590 |
+
{
|
| 4591 |
+
"epoch": 0.52,
|
| 4592 |
+
"learning_rate": 0.0005401912181303115,
|
| 4593 |
+
"loss": 1.9336,
|
| 4594 |
+
"step": 763
|
| 4595 |
+
},
|
| 4596 |
+
{
|
| 4597 |
+
"epoch": 0.52,
|
| 4598 |
+
"learning_rate": 0.000540084985835694,
|
| 4599 |
+
"loss": 1.8554,
|
| 4600 |
+
"step": 764
|
| 4601 |
+
},
|
| 4602 |
+
{
|
| 4603 |
+
"epoch": 0.52,
|
| 4604 |
+
"learning_rate": 0.0005399787535410764,
|
| 4605 |
+
"loss": 1.9388,
|
| 4606 |
+
"step": 765
|
| 4607 |
+
},
|
| 4608 |
+
{
|
| 4609 |
+
"epoch": 0.52,
|
| 4610 |
+
"learning_rate": 0.0005398725212464588,
|
| 4611 |
+
"loss": 1.9101,
|
| 4612 |
+
"step": 766
|
| 4613 |
+
},
|
| 4614 |
+
{
|
| 4615 |
+
"epoch": 0.52,
|
| 4616 |
+
"learning_rate": 0.0005397662889518413,
|
| 4617 |
+
"loss": 1.866,
|
| 4618 |
+
"step": 767
|
| 4619 |
+
},
|
| 4620 |
+
{
|
| 4621 |
+
"epoch": 0.53,
|
| 4622 |
+
"learning_rate": 0.0005396600566572237,
|
| 4623 |
+
"loss": 1.9731,
|
| 4624 |
+
"step": 768
|
| 4625 |
+
},
|
| 4626 |
+
{
|
| 4627 |
+
"epoch": 0.53,
|
| 4628 |
+
"learning_rate": 0.0005395538243626062,
|
| 4629 |
+
"loss": 1.9263,
|
| 4630 |
+
"step": 769
|
| 4631 |
+
},
|
| 4632 |
+
{
|
| 4633 |
+
"epoch": 0.53,
|
| 4634 |
+
"learning_rate": 0.0005394475920679887,
|
| 4635 |
+
"loss": 1.9337,
|
| 4636 |
+
"step": 770
|
| 4637 |
+
},
|
| 4638 |
+
{
|
| 4639 |
+
"epoch": 0.53,
|
| 4640 |
+
"learning_rate": 0.0005393413597733711,
|
| 4641 |
+
"loss": 1.896,
|
| 4642 |
+
"step": 771
|
| 4643 |
+
},
|
| 4644 |
+
{
|
| 4645 |
+
"epoch": 0.53,
|
| 4646 |
+
"learning_rate": 0.0005392351274787535,
|
| 4647 |
+
"loss": 1.9376,
|
| 4648 |
+
"step": 772
|
| 4649 |
+
},
|
| 4650 |
+
{
|
| 4651 |
+
"epoch": 0.53,
|
| 4652 |
+
"learning_rate": 0.000539128895184136,
|
| 4653 |
+
"loss": 1.9601,
|
| 4654 |
+
"step": 773
|
| 4655 |
+
},
|
| 4656 |
+
{
|
| 4657 |
+
"epoch": 0.53,
|
| 4658 |
+
"learning_rate": 0.0005390226628895183,
|
| 4659 |
+
"loss": 1.9177,
|
| 4660 |
+
"step": 774
|
| 4661 |
+
},
|
| 4662 |
+
{
|
| 4663 |
+
"epoch": 0.53,
|
| 4664 |
+
"learning_rate": 0.0005389164305949008,
|
| 4665 |
+
"loss": 1.9044,
|
| 4666 |
+
"step": 775
|
| 4667 |
+
},
|
| 4668 |
+
{
|
| 4669 |
+
"epoch": 0.53,
|
| 4670 |
+
"learning_rate": 0.0005388101983002832,
|
| 4671 |
+
"loss": 1.9378,
|
| 4672 |
+
"step": 776
|
| 4673 |
+
},
|
| 4674 |
+
{
|
| 4675 |
+
"epoch": 0.53,
|
| 4676 |
+
"learning_rate": 0.0005387039660056656,
|
| 4677 |
+
"loss": 1.8153,
|
| 4678 |
+
"step": 777
|
| 4679 |
+
},
|
| 4680 |
+
{
|
| 4681 |
+
"epoch": 0.53,
|
| 4682 |
+
"learning_rate": 0.0005385977337110481,
|
| 4683 |
+
"loss": 1.9445,
|
| 4684 |
+
"step": 778
|
| 4685 |
+
},
|
| 4686 |
+
{
|
| 4687 |
+
"epoch": 0.53,
|
| 4688 |
+
"learning_rate": 0.0005384915014164305,
|
| 4689 |
+
"loss": 1.8467,
|
| 4690 |
+
"step": 779
|
| 4691 |
+
},
|
| 4692 |
+
{
|
| 4693 |
+
"epoch": 0.53,
|
| 4694 |
+
"learning_rate": 0.000538385269121813,
|
| 4695 |
+
"loss": 1.8743,
|
| 4696 |
+
"step": 780
|
| 4697 |
+
},
|
| 4698 |
+
{
|
| 4699 |
+
"epoch": 0.53,
|
| 4700 |
+
"learning_rate": 0.0005382790368271955,
|
| 4701 |
+
"loss": 1.8839,
|
| 4702 |
+
"step": 781
|
| 4703 |
+
},
|
| 4704 |
+
{
|
| 4705 |
+
"epoch": 0.53,
|
| 4706 |
+
"learning_rate": 0.0005381728045325779,
|
| 4707 |
+
"loss": 1.8023,
|
| 4708 |
+
"step": 782
|
| 4709 |
+
},
|
| 4710 |
+
{
|
| 4711 |
+
"epoch": 0.54,
|
| 4712 |
+
"learning_rate": 0.0005380665722379603,
|
| 4713 |
+
"loss": 1.9814,
|
| 4714 |
+
"step": 783
|
| 4715 |
+
},
|
| 4716 |
+
{
|
| 4717 |
+
"epoch": 0.54,
|
| 4718 |
+
"learning_rate": 0.0005379603399433428,
|
| 4719 |
+
"loss": 1.8621,
|
| 4720 |
+
"step": 784
|
| 4721 |
+
},
|
| 4722 |
+
{
|
| 4723 |
+
"epoch": 0.54,
|
| 4724 |
+
"learning_rate": 0.0005378541076487252,
|
| 4725 |
+
"loss": 1.9371,
|
| 4726 |
+
"step": 785
|
| 4727 |
+
},
|
| 4728 |
+
{
|
| 4729 |
+
"epoch": 0.54,
|
| 4730 |
+
"learning_rate": 0.0005377478753541075,
|
| 4731 |
+
"loss": 1.9692,
|
| 4732 |
+
"step": 786
|
| 4733 |
+
},
|
| 4734 |
+
{
|
| 4735 |
+
"epoch": 0.54,
|
| 4736 |
+
"learning_rate": 0.00053764164305949,
|
| 4737 |
+
"loss": 1.9392,
|
| 4738 |
+
"step": 787
|
| 4739 |
+
},
|
| 4740 |
+
{
|
| 4741 |
+
"epoch": 0.54,
|
| 4742 |
+
"learning_rate": 0.0005375354107648724,
|
| 4743 |
+
"loss": 2.0,
|
| 4744 |
+
"step": 788
|
| 4745 |
+
},
|
| 4746 |
+
{
|
| 4747 |
+
"epoch": 0.54,
|
| 4748 |
+
"learning_rate": 0.0005374291784702549,
|
| 4749 |
+
"loss": 1.9281,
|
| 4750 |
+
"step": 789
|
| 4751 |
+
},
|
| 4752 |
+
{
|
| 4753 |
+
"epoch": 0.54,
|
| 4754 |
+
"learning_rate": 0.0005373229461756373,
|
| 4755 |
+
"loss": 1.9512,
|
| 4756 |
+
"step": 790
|
| 4757 |
+
},
|
| 4758 |
+
{
|
| 4759 |
+
"epoch": 0.54,
|
| 4760 |
+
"learning_rate": 0.0005372167138810198,
|
| 4761 |
+
"loss": 1.9222,
|
| 4762 |
+
"step": 791
|
| 4763 |
+
},
|
| 4764 |
+
{
|
| 4765 |
+
"epoch": 0.54,
|
| 4766 |
+
"learning_rate": 0.0005371104815864022,
|
| 4767 |
+
"loss": 1.9698,
|
| 4768 |
+
"step": 792
|
| 4769 |
+
},
|
| 4770 |
+
{
|
| 4771 |
+
"epoch": 0.54,
|
| 4772 |
+
"learning_rate": 0.0005370042492917847,
|
| 4773 |
+
"loss": 1.8392,
|
| 4774 |
+
"step": 793
|
| 4775 |
+
},
|
| 4776 |
+
{
|
| 4777 |
+
"epoch": 0.54,
|
| 4778 |
+
"learning_rate": 0.0005368980169971671,
|
| 4779 |
+
"loss": 1.9714,
|
| 4780 |
+
"step": 794
|
| 4781 |
+
},
|
| 4782 |
+
{
|
| 4783 |
+
"epoch": 0.54,
|
| 4784 |
+
"learning_rate": 0.0005367917847025496,
|
| 4785 |
+
"loss": 1.8521,
|
| 4786 |
+
"step": 795
|
| 4787 |
+
},
|
| 4788 |
+
{
|
| 4789 |
+
"epoch": 0.54,
|
| 4790 |
+
"learning_rate": 0.000536685552407932,
|
| 4791 |
+
"loss": 1.9495,
|
| 4792 |
+
"step": 796
|
| 4793 |
+
},
|
| 4794 |
+
{
|
| 4795 |
+
"epoch": 0.54,
|
| 4796 |
+
"learning_rate": 0.0005365793201133144,
|
| 4797 |
+
"loss": 1.9401,
|
| 4798 |
+
"step": 797
|
| 4799 |
+
},
|
| 4800 |
+
{
|
| 4801 |
+
"epoch": 0.55,
|
| 4802 |
+
"learning_rate": 0.0005364730878186969,
|
| 4803 |
+
"loss": 1.8449,
|
| 4804 |
+
"step": 798
|
| 4805 |
+
},
|
| 4806 |
+
{
|
| 4807 |
+
"epoch": 0.55,
|
| 4808 |
+
"learning_rate": 0.0005363668555240792,
|
| 4809 |
+
"loss": 1.9425,
|
| 4810 |
+
"step": 799
|
| 4811 |
+
},
|
| 4812 |
+
{
|
| 4813 |
+
"epoch": 0.55,
|
| 4814 |
+
"learning_rate": 0.0005362606232294617,
|
| 4815 |
+
"loss": 1.8826,
|
| 4816 |
+
"step": 800
|
| 4817 |
+
},
|
| 4818 |
+
{
|
| 4819 |
+
"epoch": 0.55,
|
| 4820 |
+
"eval_loss": 1.9985228776931763,
|
| 4821 |
+
"eval_runtime": 1468.32,
|
| 4822 |
+
"eval_samples_per_second": 9.904,
|
| 4823 |
+
"eval_steps_per_second": 9.904,
|
| 4824 |
+
"step": 800
|
| 4825 |
}
|
| 4826 |
],
|
| 4827 |
"max_steps": 5848,
|
| 4828 |
"num_train_epochs": 4,
|
| 4829 |
+
"total_flos": 9.813169375148851e+17,
|
| 4830 |
"trial_name": null,
|
| 4831 |
"trial_params": null
|
| 4832 |
}
|
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 sha256:18d064e377257d8b4962000d02ab7ff4990aabdaf29ea0066e9c7e06266ac53f
|
| 3 |
size 2368281769
|