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 +1218 -5
- pytorch_model.bin +1 -1
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4736616809
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4f65a9357a3b13a32cb1771d8d3d54fda21610e2f927e2b097b8984f87c58bb1
|
| 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 sha256:13d97f15adc41397e040b1276339005dcc4183c282bc569b80bbabc18a3e5da7
|
| 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 sha256:1185b9f269b6f6c5c36dd83644735ab968d01c77551fcdab8caad4927690c5dd
|
| 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 sha256:96705910f49b796948adf4805354cef92505b767c8e6d0a6e64c4e33427725e4
|
| 3 |
size 627
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 2.
|
| 3 |
-
"best_model_checkpoint": "output/checkpoint-
|
| 4 |
-
"epoch": 0.
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
@@ -3645,11 +3645,1224 @@
|
|
| 3645 |
"eval_samples_per_second": 0.055,
|
| 3646 |
"eval_steps_per_second": 0.055,
|
| 3647 |
"step": 600
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3648 |
}
|
| 3649 |
],
|
| 3650 |
"max_steps": 2000,
|
| 3651 |
"num_train_epochs": 9223372036854775807,
|
| 3652 |
-
"total_flos":
|
| 3653 |
"trial_name": null,
|
| 3654 |
"trial_params": null
|
| 3655 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 2.158494710922241,
|
| 3 |
+
"best_model_checkpoint": "output/checkpoint-800",
|
| 4 |
+
"epoch": 0.4,
|
| 5 |
+
"global_step": 800,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 3645 |
"eval_samples_per_second": 0.055,
|
| 3646 |
"eval_steps_per_second": 0.055,
|
| 3647 |
"step": 600
|
| 3648 |
+
},
|
| 3649 |
+
{
|
| 3650 |
+
"epoch": 0.3,
|
| 3651 |
+
"learning_rate": 0.0006217777777777778,
|
| 3652 |
+
"loss": 2.2629,
|
| 3653 |
+
"step": 601
|
| 3654 |
+
},
|
| 3655 |
+
{
|
| 3656 |
+
"epoch": 0.3,
|
| 3657 |
+
"learning_rate": 0.0006213333333333333,
|
| 3658 |
+
"loss": 2.2242,
|
| 3659 |
+
"step": 602
|
| 3660 |
+
},
|
| 3661 |
+
{
|
| 3662 |
+
"epoch": 0.3,
|
| 3663 |
+
"learning_rate": 0.0006208888888888889,
|
| 3664 |
+
"loss": 2.2483,
|
| 3665 |
+
"step": 603
|
| 3666 |
+
},
|
| 3667 |
+
{
|
| 3668 |
+
"epoch": 0.3,
|
| 3669 |
+
"learning_rate": 0.0006204444444444445,
|
| 3670 |
+
"loss": 2.2495,
|
| 3671 |
+
"step": 604
|
| 3672 |
+
},
|
| 3673 |
+
{
|
| 3674 |
+
"epoch": 0.3,
|
| 3675 |
+
"learning_rate": 0.00062,
|
| 3676 |
+
"loss": 2.313,
|
| 3677 |
+
"step": 605
|
| 3678 |
+
},
|
| 3679 |
+
{
|
| 3680 |
+
"epoch": 0.3,
|
| 3681 |
+
"learning_rate": 0.0006195555555555556,
|
| 3682 |
+
"loss": 2.2913,
|
| 3683 |
+
"step": 606
|
| 3684 |
+
},
|
| 3685 |
+
{
|
| 3686 |
+
"epoch": 0.3,
|
| 3687 |
+
"learning_rate": 0.0006191111111111111,
|
| 3688 |
+
"loss": 2.3303,
|
| 3689 |
+
"step": 607
|
| 3690 |
+
},
|
| 3691 |
+
{
|
| 3692 |
+
"epoch": 0.3,
|
| 3693 |
+
"learning_rate": 0.0006186666666666667,
|
| 3694 |
+
"loss": 2.2479,
|
| 3695 |
+
"step": 608
|
| 3696 |
+
},
|
| 3697 |
+
{
|
| 3698 |
+
"epoch": 0.3,
|
| 3699 |
+
"learning_rate": 0.0006182222222222223,
|
| 3700 |
+
"loss": 2.2751,
|
| 3701 |
+
"step": 609
|
| 3702 |
+
},
|
| 3703 |
+
{
|
| 3704 |
+
"epoch": 0.3,
|
| 3705 |
+
"learning_rate": 0.0006177777777777779,
|
| 3706 |
+
"loss": 2.2311,
|
| 3707 |
+
"step": 610
|
| 3708 |
+
},
|
| 3709 |
+
{
|
| 3710 |
+
"epoch": 0.31,
|
| 3711 |
+
"learning_rate": 0.0006173333333333333,
|
| 3712 |
+
"loss": 2.2294,
|
| 3713 |
+
"step": 611
|
| 3714 |
+
},
|
| 3715 |
+
{
|
| 3716 |
+
"epoch": 0.31,
|
| 3717 |
+
"learning_rate": 0.0006168888888888889,
|
| 3718 |
+
"loss": 2.2305,
|
| 3719 |
+
"step": 612
|
| 3720 |
+
},
|
| 3721 |
+
{
|
| 3722 |
+
"epoch": 0.31,
|
| 3723 |
+
"learning_rate": 0.0006164444444444444,
|
| 3724 |
+
"loss": 2.277,
|
| 3725 |
+
"step": 613
|
| 3726 |
+
},
|
| 3727 |
+
{
|
| 3728 |
+
"epoch": 0.31,
|
| 3729 |
+
"learning_rate": 0.000616,
|
| 3730 |
+
"loss": 2.347,
|
| 3731 |
+
"step": 614
|
| 3732 |
+
},
|
| 3733 |
+
{
|
| 3734 |
+
"epoch": 0.31,
|
| 3735 |
+
"learning_rate": 0.0006155555555555556,
|
| 3736 |
+
"loss": 2.2763,
|
| 3737 |
+
"step": 615
|
| 3738 |
+
},
|
| 3739 |
+
{
|
| 3740 |
+
"epoch": 0.31,
|
| 3741 |
+
"learning_rate": 0.0006151111111111111,
|
| 3742 |
+
"loss": 2.3201,
|
| 3743 |
+
"step": 616
|
| 3744 |
+
},
|
| 3745 |
+
{
|
| 3746 |
+
"epoch": 0.31,
|
| 3747 |
+
"learning_rate": 0.0006146666666666667,
|
| 3748 |
+
"loss": 2.2555,
|
| 3749 |
+
"step": 617
|
| 3750 |
+
},
|
| 3751 |
+
{
|
| 3752 |
+
"epoch": 0.31,
|
| 3753 |
+
"learning_rate": 0.0006142222222222223,
|
| 3754 |
+
"loss": 2.3702,
|
| 3755 |
+
"step": 618
|
| 3756 |
+
},
|
| 3757 |
+
{
|
| 3758 |
+
"epoch": 0.31,
|
| 3759 |
+
"learning_rate": 0.0006137777777777779,
|
| 3760 |
+
"loss": 2.2319,
|
| 3761 |
+
"step": 619
|
| 3762 |
+
},
|
| 3763 |
+
{
|
| 3764 |
+
"epoch": 0.31,
|
| 3765 |
+
"learning_rate": 0.0006133333333333334,
|
| 3766 |
+
"loss": 2.2326,
|
| 3767 |
+
"step": 620
|
| 3768 |
+
},
|
| 3769 |
+
{
|
| 3770 |
+
"epoch": 0.31,
|
| 3771 |
+
"learning_rate": 0.0006128888888888888,
|
| 3772 |
+
"loss": 2.3561,
|
| 3773 |
+
"step": 621
|
| 3774 |
+
},
|
| 3775 |
+
{
|
| 3776 |
+
"epoch": 0.31,
|
| 3777 |
+
"learning_rate": 0.0006124444444444444,
|
| 3778 |
+
"loss": 2.2702,
|
| 3779 |
+
"step": 622
|
| 3780 |
+
},
|
| 3781 |
+
{
|
| 3782 |
+
"epoch": 0.31,
|
| 3783 |
+
"learning_rate": 0.000612,
|
| 3784 |
+
"loss": 2.2726,
|
| 3785 |
+
"step": 623
|
| 3786 |
+
},
|
| 3787 |
+
{
|
| 3788 |
+
"epoch": 0.31,
|
| 3789 |
+
"learning_rate": 0.0006115555555555556,
|
| 3790 |
+
"loss": 2.3036,
|
| 3791 |
+
"step": 624
|
| 3792 |
+
},
|
| 3793 |
+
{
|
| 3794 |
+
"epoch": 0.31,
|
| 3795 |
+
"learning_rate": 0.0006111111111111111,
|
| 3796 |
+
"loss": 2.316,
|
| 3797 |
+
"step": 625
|
| 3798 |
+
},
|
| 3799 |
+
{
|
| 3800 |
+
"epoch": 0.31,
|
| 3801 |
+
"learning_rate": 0.0006106666666666667,
|
| 3802 |
+
"loss": 2.2936,
|
| 3803 |
+
"step": 626
|
| 3804 |
+
},
|
| 3805 |
+
{
|
| 3806 |
+
"epoch": 0.31,
|
| 3807 |
+
"learning_rate": 0.0006102222222222223,
|
| 3808 |
+
"loss": 2.3333,
|
| 3809 |
+
"step": 627
|
| 3810 |
+
},
|
| 3811 |
+
{
|
| 3812 |
+
"epoch": 0.31,
|
| 3813 |
+
"learning_rate": 0.0006097777777777779,
|
| 3814 |
+
"loss": 2.3695,
|
| 3815 |
+
"step": 628
|
| 3816 |
+
},
|
| 3817 |
+
{
|
| 3818 |
+
"epoch": 0.31,
|
| 3819 |
+
"learning_rate": 0.0006093333333333334,
|
| 3820 |
+
"loss": 2.1621,
|
| 3821 |
+
"step": 629
|
| 3822 |
+
},
|
| 3823 |
+
{
|
| 3824 |
+
"epoch": 0.32,
|
| 3825 |
+
"learning_rate": 0.0006088888888888888,
|
| 3826 |
+
"loss": 2.288,
|
| 3827 |
+
"step": 630
|
| 3828 |
+
},
|
| 3829 |
+
{
|
| 3830 |
+
"epoch": 0.32,
|
| 3831 |
+
"learning_rate": 0.0006084444444444444,
|
| 3832 |
+
"loss": 2.275,
|
| 3833 |
+
"step": 631
|
| 3834 |
+
},
|
| 3835 |
+
{
|
| 3836 |
+
"epoch": 0.32,
|
| 3837 |
+
"learning_rate": 0.000608,
|
| 3838 |
+
"loss": 2.2326,
|
| 3839 |
+
"step": 632
|
| 3840 |
+
},
|
| 3841 |
+
{
|
| 3842 |
+
"epoch": 0.32,
|
| 3843 |
+
"learning_rate": 0.0006075555555555556,
|
| 3844 |
+
"loss": 2.3058,
|
| 3845 |
+
"step": 633
|
| 3846 |
+
},
|
| 3847 |
+
{
|
| 3848 |
+
"epoch": 0.32,
|
| 3849 |
+
"learning_rate": 0.0006071111111111111,
|
| 3850 |
+
"loss": 2.3021,
|
| 3851 |
+
"step": 634
|
| 3852 |
+
},
|
| 3853 |
+
{
|
| 3854 |
+
"epoch": 0.32,
|
| 3855 |
+
"learning_rate": 0.0006066666666666667,
|
| 3856 |
+
"loss": 2.2709,
|
| 3857 |
+
"step": 635
|
| 3858 |
+
},
|
| 3859 |
+
{
|
| 3860 |
+
"epoch": 0.32,
|
| 3861 |
+
"learning_rate": 0.0006062222222222223,
|
| 3862 |
+
"loss": 2.2401,
|
| 3863 |
+
"step": 636
|
| 3864 |
+
},
|
| 3865 |
+
{
|
| 3866 |
+
"epoch": 0.32,
|
| 3867 |
+
"learning_rate": 0.0006057777777777778,
|
| 3868 |
+
"loss": 2.2679,
|
| 3869 |
+
"step": 637
|
| 3870 |
+
},
|
| 3871 |
+
{
|
| 3872 |
+
"epoch": 0.32,
|
| 3873 |
+
"learning_rate": 0.0006053333333333334,
|
| 3874 |
+
"loss": 2.2649,
|
| 3875 |
+
"step": 638
|
| 3876 |
+
},
|
| 3877 |
+
{
|
| 3878 |
+
"epoch": 0.32,
|
| 3879 |
+
"learning_rate": 0.0006048888888888889,
|
| 3880 |
+
"loss": 2.2832,
|
| 3881 |
+
"step": 639
|
| 3882 |
+
},
|
| 3883 |
+
{
|
| 3884 |
+
"epoch": 0.32,
|
| 3885 |
+
"learning_rate": 0.0006044444444444445,
|
| 3886 |
+
"loss": 2.2549,
|
| 3887 |
+
"step": 640
|
| 3888 |
+
},
|
| 3889 |
+
{
|
| 3890 |
+
"epoch": 0.32,
|
| 3891 |
+
"learning_rate": 0.000604,
|
| 3892 |
+
"loss": 2.2033,
|
| 3893 |
+
"step": 641
|
| 3894 |
+
},
|
| 3895 |
+
{
|
| 3896 |
+
"epoch": 0.32,
|
| 3897 |
+
"learning_rate": 0.0006035555555555556,
|
| 3898 |
+
"loss": 2.3478,
|
| 3899 |
+
"step": 642
|
| 3900 |
+
},
|
| 3901 |
+
{
|
| 3902 |
+
"epoch": 0.32,
|
| 3903 |
+
"learning_rate": 0.0006031111111111112,
|
| 3904 |
+
"loss": 2.2811,
|
| 3905 |
+
"step": 643
|
| 3906 |
+
},
|
| 3907 |
+
{
|
| 3908 |
+
"epoch": 0.32,
|
| 3909 |
+
"learning_rate": 0.0006026666666666667,
|
| 3910 |
+
"loss": 2.2695,
|
| 3911 |
+
"step": 644
|
| 3912 |
+
},
|
| 3913 |
+
{
|
| 3914 |
+
"epoch": 0.32,
|
| 3915 |
+
"learning_rate": 0.0006022222222222222,
|
| 3916 |
+
"loss": 2.3959,
|
| 3917 |
+
"step": 645
|
| 3918 |
+
},
|
| 3919 |
+
{
|
| 3920 |
+
"epoch": 0.32,
|
| 3921 |
+
"learning_rate": 0.0006017777777777778,
|
| 3922 |
+
"loss": 2.2135,
|
| 3923 |
+
"step": 646
|
| 3924 |
+
},
|
| 3925 |
+
{
|
| 3926 |
+
"epoch": 0.32,
|
| 3927 |
+
"learning_rate": 0.0006013333333333334,
|
| 3928 |
+
"loss": 2.2501,
|
| 3929 |
+
"step": 647
|
| 3930 |
+
},
|
| 3931 |
+
{
|
| 3932 |
+
"epoch": 0.32,
|
| 3933 |
+
"learning_rate": 0.0006008888888888889,
|
| 3934 |
+
"loss": 2.2801,
|
| 3935 |
+
"step": 648
|
| 3936 |
+
},
|
| 3937 |
+
{
|
| 3938 |
+
"epoch": 0.32,
|
| 3939 |
+
"learning_rate": 0.0006004444444444445,
|
| 3940 |
+
"loss": 2.2968,
|
| 3941 |
+
"step": 649
|
| 3942 |
+
},
|
| 3943 |
+
{
|
| 3944 |
+
"epoch": 0.33,
|
| 3945 |
+
"learning_rate": 0.0006000000000000001,
|
| 3946 |
+
"loss": 2.2578,
|
| 3947 |
+
"step": 650
|
| 3948 |
+
},
|
| 3949 |
+
{
|
| 3950 |
+
"epoch": 0.33,
|
| 3951 |
+
"learning_rate": 0.0005995555555555556,
|
| 3952 |
+
"loss": 2.1764,
|
| 3953 |
+
"step": 651
|
| 3954 |
+
},
|
| 3955 |
+
{
|
| 3956 |
+
"epoch": 0.33,
|
| 3957 |
+
"learning_rate": 0.0005991111111111111,
|
| 3958 |
+
"loss": 2.3049,
|
| 3959 |
+
"step": 652
|
| 3960 |
+
},
|
| 3961 |
+
{
|
| 3962 |
+
"epoch": 0.33,
|
| 3963 |
+
"learning_rate": 0.0005986666666666666,
|
| 3964 |
+
"loss": 2.2893,
|
| 3965 |
+
"step": 653
|
| 3966 |
+
},
|
| 3967 |
+
{
|
| 3968 |
+
"epoch": 0.33,
|
| 3969 |
+
"learning_rate": 0.0005982222222222222,
|
| 3970 |
+
"loss": 2.2749,
|
| 3971 |
+
"step": 654
|
| 3972 |
+
},
|
| 3973 |
+
{
|
| 3974 |
+
"epoch": 0.33,
|
| 3975 |
+
"learning_rate": 0.0005977777777777778,
|
| 3976 |
+
"loss": 2.3037,
|
| 3977 |
+
"step": 655
|
| 3978 |
+
},
|
| 3979 |
+
{
|
| 3980 |
+
"epoch": 0.33,
|
| 3981 |
+
"learning_rate": 0.0005973333333333334,
|
| 3982 |
+
"loss": 2.2434,
|
| 3983 |
+
"step": 656
|
| 3984 |
+
},
|
| 3985 |
+
{
|
| 3986 |
+
"epoch": 0.33,
|
| 3987 |
+
"learning_rate": 0.0005968888888888889,
|
| 3988 |
+
"loss": 2.3198,
|
| 3989 |
+
"step": 657
|
| 3990 |
+
},
|
| 3991 |
+
{
|
| 3992 |
+
"epoch": 0.33,
|
| 3993 |
+
"learning_rate": 0.0005964444444444445,
|
| 3994 |
+
"loss": 2.2131,
|
| 3995 |
+
"step": 658
|
| 3996 |
+
},
|
| 3997 |
+
{
|
| 3998 |
+
"epoch": 0.33,
|
| 3999 |
+
"learning_rate": 0.0005960000000000001,
|
| 4000 |
+
"loss": 2.2619,
|
| 4001 |
+
"step": 659
|
| 4002 |
+
},
|
| 4003 |
+
{
|
| 4004 |
+
"epoch": 0.33,
|
| 4005 |
+
"learning_rate": 0.0005955555555555556,
|
| 4006 |
+
"loss": 2.2657,
|
| 4007 |
+
"step": 660
|
| 4008 |
+
},
|
| 4009 |
+
{
|
| 4010 |
+
"epoch": 0.33,
|
| 4011 |
+
"learning_rate": 0.0005951111111111111,
|
| 4012 |
+
"loss": 2.2787,
|
| 4013 |
+
"step": 661
|
| 4014 |
+
},
|
| 4015 |
+
{
|
| 4016 |
+
"epoch": 0.33,
|
| 4017 |
+
"learning_rate": 0.0005946666666666666,
|
| 4018 |
+
"loss": 2.3317,
|
| 4019 |
+
"step": 662
|
| 4020 |
+
},
|
| 4021 |
+
{
|
| 4022 |
+
"epoch": 0.33,
|
| 4023 |
+
"learning_rate": 0.0005942222222222222,
|
| 4024 |
+
"loss": 2.3017,
|
| 4025 |
+
"step": 663
|
| 4026 |
+
},
|
| 4027 |
+
{
|
| 4028 |
+
"epoch": 0.33,
|
| 4029 |
+
"learning_rate": 0.0005937777777777778,
|
| 4030 |
+
"loss": 2.2839,
|
| 4031 |
+
"step": 664
|
| 4032 |
+
},
|
| 4033 |
+
{
|
| 4034 |
+
"epoch": 0.33,
|
| 4035 |
+
"learning_rate": 0.0005933333333333334,
|
| 4036 |
+
"loss": 2.2235,
|
| 4037 |
+
"step": 665
|
| 4038 |
+
},
|
| 4039 |
+
{
|
| 4040 |
+
"epoch": 0.33,
|
| 4041 |
+
"learning_rate": 0.000592888888888889,
|
| 4042 |
+
"loss": 2.191,
|
| 4043 |
+
"step": 666
|
| 4044 |
+
},
|
| 4045 |
+
{
|
| 4046 |
+
"epoch": 0.33,
|
| 4047 |
+
"learning_rate": 0.0005924444444444445,
|
| 4048 |
+
"loss": 2.2485,
|
| 4049 |
+
"step": 667
|
| 4050 |
+
},
|
| 4051 |
+
{
|
| 4052 |
+
"epoch": 0.33,
|
| 4053 |
+
"learning_rate": 0.000592,
|
| 4054 |
+
"loss": 2.2507,
|
| 4055 |
+
"step": 668
|
| 4056 |
+
},
|
| 4057 |
+
{
|
| 4058 |
+
"epoch": 0.33,
|
| 4059 |
+
"learning_rate": 0.0005915555555555556,
|
| 4060 |
+
"loss": 2.2717,
|
| 4061 |
+
"step": 669
|
| 4062 |
+
},
|
| 4063 |
+
{
|
| 4064 |
+
"epoch": 0.34,
|
| 4065 |
+
"learning_rate": 0.0005911111111111112,
|
| 4066 |
+
"loss": 2.1896,
|
| 4067 |
+
"step": 670
|
| 4068 |
+
},
|
| 4069 |
+
{
|
| 4070 |
+
"epoch": 0.34,
|
| 4071 |
+
"learning_rate": 0.0005906666666666666,
|
| 4072 |
+
"loss": 2.3243,
|
| 4073 |
+
"step": 671
|
| 4074 |
+
},
|
| 4075 |
+
{
|
| 4076 |
+
"epoch": 0.34,
|
| 4077 |
+
"learning_rate": 0.0005902222222222222,
|
| 4078 |
+
"loss": 2.1593,
|
| 4079 |
+
"step": 672
|
| 4080 |
+
},
|
| 4081 |
+
{
|
| 4082 |
+
"epoch": 0.34,
|
| 4083 |
+
"learning_rate": 0.0005897777777777778,
|
| 4084 |
+
"loss": 2.2796,
|
| 4085 |
+
"step": 673
|
| 4086 |
+
},
|
| 4087 |
+
{
|
| 4088 |
+
"epoch": 0.34,
|
| 4089 |
+
"learning_rate": 0.0005893333333333334,
|
| 4090 |
+
"loss": 2.2769,
|
| 4091 |
+
"step": 674
|
| 4092 |
+
},
|
| 4093 |
+
{
|
| 4094 |
+
"epoch": 0.34,
|
| 4095 |
+
"learning_rate": 0.0005888888888888889,
|
| 4096 |
+
"loss": 2.3809,
|
| 4097 |
+
"step": 675
|
| 4098 |
+
},
|
| 4099 |
+
{
|
| 4100 |
+
"epoch": 0.34,
|
| 4101 |
+
"learning_rate": 0.0005884444444444445,
|
| 4102 |
+
"loss": 2.2652,
|
| 4103 |
+
"step": 676
|
| 4104 |
+
},
|
| 4105 |
+
{
|
| 4106 |
+
"epoch": 0.34,
|
| 4107 |
+
"learning_rate": 0.000588,
|
| 4108 |
+
"loss": 2.2152,
|
| 4109 |
+
"step": 677
|
| 4110 |
+
},
|
| 4111 |
+
{
|
| 4112 |
+
"epoch": 0.34,
|
| 4113 |
+
"learning_rate": 0.0005875555555555556,
|
| 4114 |
+
"loss": 2.18,
|
| 4115 |
+
"step": 678
|
| 4116 |
+
},
|
| 4117 |
+
{
|
| 4118 |
+
"epoch": 0.34,
|
| 4119 |
+
"learning_rate": 0.0005871111111111112,
|
| 4120 |
+
"loss": 2.2659,
|
| 4121 |
+
"step": 679
|
| 4122 |
+
},
|
| 4123 |
+
{
|
| 4124 |
+
"epoch": 0.34,
|
| 4125 |
+
"learning_rate": 0.0005866666666666667,
|
| 4126 |
+
"loss": 2.2558,
|
| 4127 |
+
"step": 680
|
| 4128 |
+
},
|
| 4129 |
+
{
|
| 4130 |
+
"epoch": 0.34,
|
| 4131 |
+
"learning_rate": 0.0005862222222222222,
|
| 4132 |
+
"loss": 2.2737,
|
| 4133 |
+
"step": 681
|
| 4134 |
+
},
|
| 4135 |
+
{
|
| 4136 |
+
"epoch": 0.34,
|
| 4137 |
+
"learning_rate": 0.0005857777777777778,
|
| 4138 |
+
"loss": 2.2537,
|
| 4139 |
+
"step": 682
|
| 4140 |
+
},
|
| 4141 |
+
{
|
| 4142 |
+
"epoch": 0.34,
|
| 4143 |
+
"learning_rate": 0.0005853333333333334,
|
| 4144 |
+
"loss": 2.2587,
|
| 4145 |
+
"step": 683
|
| 4146 |
+
},
|
| 4147 |
+
{
|
| 4148 |
+
"epoch": 0.34,
|
| 4149 |
+
"learning_rate": 0.0005848888888888889,
|
| 4150 |
+
"loss": 2.2707,
|
| 4151 |
+
"step": 684
|
| 4152 |
+
},
|
| 4153 |
+
{
|
| 4154 |
+
"epoch": 0.34,
|
| 4155 |
+
"learning_rate": 0.0005844444444444444,
|
| 4156 |
+
"loss": 2.2646,
|
| 4157 |
+
"step": 685
|
| 4158 |
+
},
|
| 4159 |
+
{
|
| 4160 |
+
"epoch": 0.34,
|
| 4161 |
+
"learning_rate": 0.000584,
|
| 4162 |
+
"loss": 2.2837,
|
| 4163 |
+
"step": 686
|
| 4164 |
+
},
|
| 4165 |
+
{
|
| 4166 |
+
"epoch": 0.34,
|
| 4167 |
+
"learning_rate": 0.0005835555555555556,
|
| 4168 |
+
"loss": 2.2998,
|
| 4169 |
+
"step": 687
|
| 4170 |
+
},
|
| 4171 |
+
{
|
| 4172 |
+
"epoch": 0.34,
|
| 4173 |
+
"learning_rate": 0.0005831111111111112,
|
| 4174 |
+
"loss": 2.2286,
|
| 4175 |
+
"step": 688
|
| 4176 |
+
},
|
| 4177 |
+
{
|
| 4178 |
+
"epoch": 0.34,
|
| 4179 |
+
"learning_rate": 0.0005826666666666668,
|
| 4180 |
+
"loss": 2.2666,
|
| 4181 |
+
"step": 689
|
| 4182 |
+
},
|
| 4183 |
+
{
|
| 4184 |
+
"epoch": 0.34,
|
| 4185 |
+
"learning_rate": 0.0005822222222222223,
|
| 4186 |
+
"loss": 2.2959,
|
| 4187 |
+
"step": 690
|
| 4188 |
+
},
|
| 4189 |
+
{
|
| 4190 |
+
"epoch": 0.35,
|
| 4191 |
+
"learning_rate": 0.0005817777777777778,
|
| 4192 |
+
"loss": 2.3227,
|
| 4193 |
+
"step": 691
|
| 4194 |
+
},
|
| 4195 |
+
{
|
| 4196 |
+
"epoch": 0.35,
|
| 4197 |
+
"learning_rate": 0.0005813333333333333,
|
| 4198 |
+
"loss": 2.3145,
|
| 4199 |
+
"step": 692
|
| 4200 |
+
},
|
| 4201 |
+
{
|
| 4202 |
+
"epoch": 0.35,
|
| 4203 |
+
"learning_rate": 0.0005808888888888889,
|
| 4204 |
+
"loss": 2.2154,
|
| 4205 |
+
"step": 693
|
| 4206 |
+
},
|
| 4207 |
+
{
|
| 4208 |
+
"epoch": 0.35,
|
| 4209 |
+
"learning_rate": 0.0005804444444444444,
|
| 4210 |
+
"loss": 2.2541,
|
| 4211 |
+
"step": 694
|
| 4212 |
+
},
|
| 4213 |
+
{
|
| 4214 |
+
"epoch": 0.35,
|
| 4215 |
+
"learning_rate": 0.00058,
|
| 4216 |
+
"loss": 2.2052,
|
| 4217 |
+
"step": 695
|
| 4218 |
+
},
|
| 4219 |
+
{
|
| 4220 |
+
"epoch": 0.35,
|
| 4221 |
+
"learning_rate": 0.0005795555555555556,
|
| 4222 |
+
"loss": 2.236,
|
| 4223 |
+
"step": 696
|
| 4224 |
+
},
|
| 4225 |
+
{
|
| 4226 |
+
"epoch": 0.35,
|
| 4227 |
+
"learning_rate": 0.0005791111111111112,
|
| 4228 |
+
"loss": 2.2407,
|
| 4229 |
+
"step": 697
|
| 4230 |
+
},
|
| 4231 |
+
{
|
| 4232 |
+
"epoch": 0.35,
|
| 4233 |
+
"learning_rate": 0.0005786666666666668,
|
| 4234 |
+
"loss": 2.2687,
|
| 4235 |
+
"step": 698
|
| 4236 |
+
},
|
| 4237 |
+
{
|
| 4238 |
+
"epoch": 0.35,
|
| 4239 |
+
"learning_rate": 0.0005782222222222223,
|
| 4240 |
+
"loss": 2.2907,
|
| 4241 |
+
"step": 699
|
| 4242 |
+
},
|
| 4243 |
+
{
|
| 4244 |
+
"epoch": 0.35,
|
| 4245 |
+
"learning_rate": 0.0005777777777777778,
|
| 4246 |
+
"loss": 2.2697,
|
| 4247 |
+
"step": 700
|
| 4248 |
+
},
|
| 4249 |
+
{
|
| 4250 |
+
"epoch": 0.35,
|
| 4251 |
+
"learning_rate": 0.0005773333333333333,
|
| 4252 |
+
"loss": 2.2348,
|
| 4253 |
+
"step": 701
|
| 4254 |
+
},
|
| 4255 |
+
{
|
| 4256 |
+
"epoch": 0.35,
|
| 4257 |
+
"learning_rate": 0.0005768888888888889,
|
| 4258 |
+
"loss": 2.2328,
|
| 4259 |
+
"step": 702
|
| 4260 |
+
},
|
| 4261 |
+
{
|
| 4262 |
+
"epoch": 0.35,
|
| 4263 |
+
"learning_rate": 0.0005764444444444444,
|
| 4264 |
+
"loss": 2.2347,
|
| 4265 |
+
"step": 703
|
| 4266 |
+
},
|
| 4267 |
+
{
|
| 4268 |
+
"epoch": 0.35,
|
| 4269 |
+
"learning_rate": 0.000576,
|
| 4270 |
+
"loss": 2.2297,
|
| 4271 |
+
"step": 704
|
| 4272 |
+
},
|
| 4273 |
+
{
|
| 4274 |
+
"epoch": 0.35,
|
| 4275 |
+
"learning_rate": 0.0005755555555555556,
|
| 4276 |
+
"loss": 2.2458,
|
| 4277 |
+
"step": 705
|
| 4278 |
+
},
|
| 4279 |
+
{
|
| 4280 |
+
"epoch": 0.35,
|
| 4281 |
+
"learning_rate": 0.0005751111111111112,
|
| 4282 |
+
"loss": 2.2093,
|
| 4283 |
+
"step": 706
|
| 4284 |
+
},
|
| 4285 |
+
{
|
| 4286 |
+
"epoch": 0.35,
|
| 4287 |
+
"learning_rate": 0.0005746666666666667,
|
| 4288 |
+
"loss": 2.227,
|
| 4289 |
+
"step": 707
|
| 4290 |
+
},
|
| 4291 |
+
{
|
| 4292 |
+
"epoch": 0.35,
|
| 4293 |
+
"learning_rate": 0.0005742222222222222,
|
| 4294 |
+
"loss": 2.3247,
|
| 4295 |
+
"step": 708
|
| 4296 |
+
},
|
| 4297 |
+
{
|
| 4298 |
+
"epoch": 0.35,
|
| 4299 |
+
"learning_rate": 0.0005737777777777778,
|
| 4300 |
+
"loss": 2.2955,
|
| 4301 |
+
"step": 709
|
| 4302 |
+
},
|
| 4303 |
+
{
|
| 4304 |
+
"epoch": 0.35,
|
| 4305 |
+
"learning_rate": 0.0005733333333333334,
|
| 4306 |
+
"loss": 2.2264,
|
| 4307 |
+
"step": 710
|
| 4308 |
+
},
|
| 4309 |
+
{
|
| 4310 |
+
"epoch": 0.36,
|
| 4311 |
+
"learning_rate": 0.000572888888888889,
|
| 4312 |
+
"loss": 2.2543,
|
| 4313 |
+
"step": 711
|
| 4314 |
+
},
|
| 4315 |
+
{
|
| 4316 |
+
"epoch": 0.36,
|
| 4317 |
+
"learning_rate": 0.0005724444444444444,
|
| 4318 |
+
"loss": 2.3499,
|
| 4319 |
+
"step": 712
|
| 4320 |
+
},
|
| 4321 |
+
{
|
| 4322 |
+
"epoch": 0.36,
|
| 4323 |
+
"learning_rate": 0.000572,
|
| 4324 |
+
"loss": 2.2414,
|
| 4325 |
+
"step": 713
|
| 4326 |
+
},
|
| 4327 |
+
{
|
| 4328 |
+
"epoch": 0.36,
|
| 4329 |
+
"learning_rate": 0.0005715555555555556,
|
| 4330 |
+
"loss": 2.2703,
|
| 4331 |
+
"step": 714
|
| 4332 |
+
},
|
| 4333 |
+
{
|
| 4334 |
+
"epoch": 0.36,
|
| 4335 |
+
"learning_rate": 0.0005711111111111111,
|
| 4336 |
+
"loss": 2.3098,
|
| 4337 |
+
"step": 715
|
| 4338 |
+
},
|
| 4339 |
+
{
|
| 4340 |
+
"epoch": 0.36,
|
| 4341 |
+
"learning_rate": 0.0005706666666666667,
|
| 4342 |
+
"loss": 2.2562,
|
| 4343 |
+
"step": 716
|
| 4344 |
+
},
|
| 4345 |
+
{
|
| 4346 |
+
"epoch": 0.36,
|
| 4347 |
+
"learning_rate": 0.0005702222222222222,
|
| 4348 |
+
"loss": 2.2553,
|
| 4349 |
+
"step": 717
|
| 4350 |
+
},
|
| 4351 |
+
{
|
| 4352 |
+
"epoch": 0.36,
|
| 4353 |
+
"learning_rate": 0.0005697777777777778,
|
| 4354 |
+
"loss": 2.2428,
|
| 4355 |
+
"step": 718
|
| 4356 |
+
},
|
| 4357 |
+
{
|
| 4358 |
+
"epoch": 0.36,
|
| 4359 |
+
"learning_rate": 0.0005693333333333334,
|
| 4360 |
+
"loss": 2.2984,
|
| 4361 |
+
"step": 719
|
| 4362 |
+
},
|
| 4363 |
+
{
|
| 4364 |
+
"epoch": 0.36,
|
| 4365 |
+
"learning_rate": 0.000568888888888889,
|
| 4366 |
+
"loss": 2.2823,
|
| 4367 |
+
"step": 720
|
| 4368 |
+
},
|
| 4369 |
+
{
|
| 4370 |
+
"epoch": 0.36,
|
| 4371 |
+
"learning_rate": 0.0005684444444444446,
|
| 4372 |
+
"loss": 2.3694,
|
| 4373 |
+
"step": 721
|
| 4374 |
+
},
|
| 4375 |
+
{
|
| 4376 |
+
"epoch": 0.36,
|
| 4377 |
+
"learning_rate": 0.000568,
|
| 4378 |
+
"loss": 2.2332,
|
| 4379 |
+
"step": 722
|
| 4380 |
+
},
|
| 4381 |
+
{
|
| 4382 |
+
"epoch": 0.36,
|
| 4383 |
+
"learning_rate": 0.0005675555555555555,
|
| 4384 |
+
"loss": 2.2421,
|
| 4385 |
+
"step": 723
|
| 4386 |
+
},
|
| 4387 |
+
{
|
| 4388 |
+
"epoch": 0.36,
|
| 4389 |
+
"learning_rate": 0.0005671111111111111,
|
| 4390 |
+
"loss": 2.251,
|
| 4391 |
+
"step": 724
|
| 4392 |
+
},
|
| 4393 |
+
{
|
| 4394 |
+
"epoch": 0.36,
|
| 4395 |
+
"learning_rate": 0.0005666666666666667,
|
| 4396 |
+
"loss": 2.1915,
|
| 4397 |
+
"step": 725
|
| 4398 |
+
},
|
| 4399 |
+
{
|
| 4400 |
+
"epoch": 0.36,
|
| 4401 |
+
"learning_rate": 0.0005662222222222222,
|
| 4402 |
+
"loss": 2.2584,
|
| 4403 |
+
"step": 726
|
| 4404 |
+
},
|
| 4405 |
+
{
|
| 4406 |
+
"epoch": 0.36,
|
| 4407 |
+
"learning_rate": 0.0005657777777777778,
|
| 4408 |
+
"loss": 2.2784,
|
| 4409 |
+
"step": 727
|
| 4410 |
+
},
|
| 4411 |
+
{
|
| 4412 |
+
"epoch": 0.36,
|
| 4413 |
+
"learning_rate": 0.0005653333333333334,
|
| 4414 |
+
"loss": 2.2584,
|
| 4415 |
+
"step": 728
|
| 4416 |
+
},
|
| 4417 |
+
{
|
| 4418 |
+
"epoch": 0.36,
|
| 4419 |
+
"learning_rate": 0.000564888888888889,
|
| 4420 |
+
"loss": 2.2316,
|
| 4421 |
+
"step": 729
|
| 4422 |
+
},
|
| 4423 |
+
{
|
| 4424 |
+
"epoch": 0.36,
|
| 4425 |
+
"learning_rate": 0.0005644444444444445,
|
| 4426 |
+
"loss": 2.2502,
|
| 4427 |
+
"step": 730
|
| 4428 |
+
},
|
| 4429 |
+
{
|
| 4430 |
+
"epoch": 0.37,
|
| 4431 |
+
"learning_rate": 0.000564,
|
| 4432 |
+
"loss": 2.2364,
|
| 4433 |
+
"step": 731
|
| 4434 |
+
},
|
| 4435 |
+
{
|
| 4436 |
+
"epoch": 0.37,
|
| 4437 |
+
"learning_rate": 0.0005635555555555555,
|
| 4438 |
+
"loss": 2.2743,
|
| 4439 |
+
"step": 732
|
| 4440 |
+
},
|
| 4441 |
+
{
|
| 4442 |
+
"epoch": 0.37,
|
| 4443 |
+
"learning_rate": 0.0005631111111111111,
|
| 4444 |
+
"loss": 2.2735,
|
| 4445 |
+
"step": 733
|
| 4446 |
+
},
|
| 4447 |
+
{
|
| 4448 |
+
"epoch": 0.37,
|
| 4449 |
+
"learning_rate": 0.0005626666666666667,
|
| 4450 |
+
"loss": 2.2353,
|
| 4451 |
+
"step": 734
|
| 4452 |
+
},
|
| 4453 |
+
{
|
| 4454 |
+
"epoch": 0.37,
|
| 4455 |
+
"learning_rate": 0.0005622222222222222,
|
| 4456 |
+
"loss": 2.2888,
|
| 4457 |
+
"step": 735
|
| 4458 |
+
},
|
| 4459 |
+
{
|
| 4460 |
+
"epoch": 0.37,
|
| 4461 |
+
"learning_rate": 0.0005617777777777778,
|
| 4462 |
+
"loss": 2.1532,
|
| 4463 |
+
"step": 736
|
| 4464 |
+
},
|
| 4465 |
+
{
|
| 4466 |
+
"epoch": 0.37,
|
| 4467 |
+
"learning_rate": 0.0005613333333333334,
|
| 4468 |
+
"loss": 2.2746,
|
| 4469 |
+
"step": 737
|
| 4470 |
+
},
|
| 4471 |
+
{
|
| 4472 |
+
"epoch": 0.37,
|
| 4473 |
+
"learning_rate": 0.000560888888888889,
|
| 4474 |
+
"loss": 2.2409,
|
| 4475 |
+
"step": 738
|
| 4476 |
+
},
|
| 4477 |
+
{
|
| 4478 |
+
"epoch": 0.37,
|
| 4479 |
+
"learning_rate": 0.0005604444444444445,
|
| 4480 |
+
"loss": 2.2122,
|
| 4481 |
+
"step": 739
|
| 4482 |
+
},
|
| 4483 |
+
{
|
| 4484 |
+
"epoch": 0.37,
|
| 4485 |
+
"learning_rate": 0.00056,
|
| 4486 |
+
"loss": 2.2223,
|
| 4487 |
+
"step": 740
|
| 4488 |
+
},
|
| 4489 |
+
{
|
| 4490 |
+
"epoch": 0.37,
|
| 4491 |
+
"learning_rate": 0.0005595555555555555,
|
| 4492 |
+
"loss": 2.228,
|
| 4493 |
+
"step": 741
|
| 4494 |
+
},
|
| 4495 |
+
{
|
| 4496 |
+
"epoch": 0.37,
|
| 4497 |
+
"learning_rate": 0.0005591111111111111,
|
| 4498 |
+
"loss": 2.2854,
|
| 4499 |
+
"step": 742
|
| 4500 |
+
},
|
| 4501 |
+
{
|
| 4502 |
+
"epoch": 0.37,
|
| 4503 |
+
"learning_rate": 0.0005586666666666667,
|
| 4504 |
+
"loss": 2.2992,
|
| 4505 |
+
"step": 743
|
| 4506 |
+
},
|
| 4507 |
+
{
|
| 4508 |
+
"epoch": 0.37,
|
| 4509 |
+
"learning_rate": 0.0005582222222222223,
|
| 4510 |
+
"loss": 2.2765,
|
| 4511 |
+
"step": 744
|
| 4512 |
+
},
|
| 4513 |
+
{
|
| 4514 |
+
"epoch": 0.37,
|
| 4515 |
+
"learning_rate": 0.0005577777777777778,
|
| 4516 |
+
"loss": 2.2645,
|
| 4517 |
+
"step": 745
|
| 4518 |
+
},
|
| 4519 |
+
{
|
| 4520 |
+
"epoch": 0.37,
|
| 4521 |
+
"learning_rate": 0.0005573333333333334,
|
| 4522 |
+
"loss": 2.2738,
|
| 4523 |
+
"step": 746
|
| 4524 |
+
},
|
| 4525 |
+
{
|
| 4526 |
+
"epoch": 0.37,
|
| 4527 |
+
"learning_rate": 0.0005568888888888889,
|
| 4528 |
+
"loss": 2.2665,
|
| 4529 |
+
"step": 747
|
| 4530 |
+
},
|
| 4531 |
+
{
|
| 4532 |
+
"epoch": 0.37,
|
| 4533 |
+
"learning_rate": 0.0005564444444444445,
|
| 4534 |
+
"loss": 2.1959,
|
| 4535 |
+
"step": 748
|
| 4536 |
+
},
|
| 4537 |
+
{
|
| 4538 |
+
"epoch": 0.37,
|
| 4539 |
+
"learning_rate": 0.000556,
|
| 4540 |
+
"loss": 2.3094,
|
| 4541 |
+
"step": 749
|
| 4542 |
+
},
|
| 4543 |
+
{
|
| 4544 |
+
"epoch": 0.38,
|
| 4545 |
+
"learning_rate": 0.0005555555555555556,
|
| 4546 |
+
"loss": 2.3302,
|
| 4547 |
+
"step": 750
|
| 4548 |
+
},
|
| 4549 |
+
{
|
| 4550 |
+
"epoch": 0.38,
|
| 4551 |
+
"learning_rate": 0.0005551111111111111,
|
| 4552 |
+
"loss": 2.2088,
|
| 4553 |
+
"step": 751
|
| 4554 |
+
},
|
| 4555 |
+
{
|
| 4556 |
+
"epoch": 0.38,
|
| 4557 |
+
"learning_rate": 0.0005546666666666667,
|
| 4558 |
+
"loss": 2.2774,
|
| 4559 |
+
"step": 752
|
| 4560 |
+
},
|
| 4561 |
+
{
|
| 4562 |
+
"epoch": 0.38,
|
| 4563 |
+
"learning_rate": 0.0005542222222222223,
|
| 4564 |
+
"loss": 2.2279,
|
| 4565 |
+
"step": 753
|
| 4566 |
+
},
|
| 4567 |
+
{
|
| 4568 |
+
"epoch": 0.38,
|
| 4569 |
+
"learning_rate": 0.0005537777777777778,
|
| 4570 |
+
"loss": 2.2091,
|
| 4571 |
+
"step": 754
|
| 4572 |
+
},
|
| 4573 |
+
{
|
| 4574 |
+
"epoch": 0.38,
|
| 4575 |
+
"learning_rate": 0.0005533333333333333,
|
| 4576 |
+
"loss": 2.3221,
|
| 4577 |
+
"step": 755
|
| 4578 |
+
},
|
| 4579 |
+
{
|
| 4580 |
+
"epoch": 0.38,
|
| 4581 |
+
"learning_rate": 0.0005528888888888889,
|
| 4582 |
+
"loss": 2.3202,
|
| 4583 |
+
"step": 756
|
| 4584 |
+
},
|
| 4585 |
+
{
|
| 4586 |
+
"epoch": 0.38,
|
| 4587 |
+
"learning_rate": 0.0005524444444444445,
|
| 4588 |
+
"loss": 2.2552,
|
| 4589 |
+
"step": 757
|
| 4590 |
+
},
|
| 4591 |
+
{
|
| 4592 |
+
"epoch": 0.38,
|
| 4593 |
+
"learning_rate": 0.000552,
|
| 4594 |
+
"loss": 2.2166,
|
| 4595 |
+
"step": 758
|
| 4596 |
+
},
|
| 4597 |
+
{
|
| 4598 |
+
"epoch": 0.38,
|
| 4599 |
+
"learning_rate": 0.0005515555555555556,
|
| 4600 |
+
"loss": 2.286,
|
| 4601 |
+
"step": 759
|
| 4602 |
+
},
|
| 4603 |
+
{
|
| 4604 |
+
"epoch": 0.38,
|
| 4605 |
+
"learning_rate": 0.0005511111111111112,
|
| 4606 |
+
"loss": 2.2642,
|
| 4607 |
+
"step": 760
|
| 4608 |
+
},
|
| 4609 |
+
{
|
| 4610 |
+
"epoch": 0.38,
|
| 4611 |
+
"learning_rate": 0.0005506666666666668,
|
| 4612 |
+
"loss": 2.3546,
|
| 4613 |
+
"step": 761
|
| 4614 |
+
},
|
| 4615 |
+
{
|
| 4616 |
+
"epoch": 0.38,
|
| 4617 |
+
"learning_rate": 0.0005502222222222222,
|
| 4618 |
+
"loss": 2.2763,
|
| 4619 |
+
"step": 762
|
| 4620 |
+
},
|
| 4621 |
+
{
|
| 4622 |
+
"epoch": 0.38,
|
| 4623 |
+
"learning_rate": 0.0005497777777777777,
|
| 4624 |
+
"loss": 2.2846,
|
| 4625 |
+
"step": 763
|
| 4626 |
+
},
|
| 4627 |
+
{
|
| 4628 |
+
"epoch": 0.38,
|
| 4629 |
+
"learning_rate": 0.0005493333333333333,
|
| 4630 |
+
"loss": 2.3388,
|
| 4631 |
+
"step": 764
|
| 4632 |
+
},
|
| 4633 |
+
{
|
| 4634 |
+
"epoch": 0.38,
|
| 4635 |
+
"learning_rate": 0.0005488888888888889,
|
| 4636 |
+
"loss": 2.1978,
|
| 4637 |
+
"step": 765
|
| 4638 |
+
},
|
| 4639 |
+
{
|
| 4640 |
+
"epoch": 0.38,
|
| 4641 |
+
"learning_rate": 0.0005484444444444445,
|
| 4642 |
+
"loss": 2.2363,
|
| 4643 |
+
"step": 766
|
| 4644 |
+
},
|
| 4645 |
+
{
|
| 4646 |
+
"epoch": 0.38,
|
| 4647 |
+
"learning_rate": 0.0005480000000000001,
|
| 4648 |
+
"loss": 2.1851,
|
| 4649 |
+
"step": 767
|
| 4650 |
+
},
|
| 4651 |
+
{
|
| 4652 |
+
"epoch": 0.38,
|
| 4653 |
+
"learning_rate": 0.0005475555555555556,
|
| 4654 |
+
"loss": 2.2744,
|
| 4655 |
+
"step": 768
|
| 4656 |
+
},
|
| 4657 |
+
{
|
| 4658 |
+
"epoch": 0.38,
|
| 4659 |
+
"learning_rate": 0.0005471111111111112,
|
| 4660 |
+
"loss": 2.3183,
|
| 4661 |
+
"step": 769
|
| 4662 |
+
},
|
| 4663 |
+
{
|
| 4664 |
+
"epoch": 0.39,
|
| 4665 |
+
"learning_rate": 0.0005466666666666667,
|
| 4666 |
+
"loss": 2.1745,
|
| 4667 |
+
"step": 770
|
| 4668 |
+
},
|
| 4669 |
+
{
|
| 4670 |
+
"epoch": 0.39,
|
| 4671 |
+
"learning_rate": 0.0005462222222222222,
|
| 4672 |
+
"loss": 2.2923,
|
| 4673 |
+
"step": 771
|
| 4674 |
+
},
|
| 4675 |
+
{
|
| 4676 |
+
"epoch": 0.39,
|
| 4677 |
+
"learning_rate": 0.0005457777777777777,
|
| 4678 |
+
"loss": 2.3098,
|
| 4679 |
+
"step": 772
|
| 4680 |
+
},
|
| 4681 |
+
{
|
| 4682 |
+
"epoch": 0.39,
|
| 4683 |
+
"learning_rate": 0.0005453333333333333,
|
| 4684 |
+
"loss": 2.2963,
|
| 4685 |
+
"step": 773
|
| 4686 |
+
},
|
| 4687 |
+
{
|
| 4688 |
+
"epoch": 0.39,
|
| 4689 |
+
"learning_rate": 0.0005448888888888889,
|
| 4690 |
+
"loss": 2.2082,
|
| 4691 |
+
"step": 774
|
| 4692 |
+
},
|
| 4693 |
+
{
|
| 4694 |
+
"epoch": 0.39,
|
| 4695 |
+
"learning_rate": 0.0005444444444444445,
|
| 4696 |
+
"loss": 2.222,
|
| 4697 |
+
"step": 775
|
| 4698 |
+
},
|
| 4699 |
+
{
|
| 4700 |
+
"epoch": 0.39,
|
| 4701 |
+
"learning_rate": 0.0005440000000000001,
|
| 4702 |
+
"loss": 2.2265,
|
| 4703 |
+
"step": 776
|
| 4704 |
+
},
|
| 4705 |
+
{
|
| 4706 |
+
"epoch": 0.39,
|
| 4707 |
+
"learning_rate": 0.0005435555555555556,
|
| 4708 |
+
"loss": 2.2619,
|
| 4709 |
+
"step": 777
|
| 4710 |
+
},
|
| 4711 |
+
{
|
| 4712 |
+
"epoch": 0.39,
|
| 4713 |
+
"learning_rate": 0.0005431111111111111,
|
| 4714 |
+
"loss": 2.2613,
|
| 4715 |
+
"step": 778
|
| 4716 |
+
},
|
| 4717 |
+
{
|
| 4718 |
+
"epoch": 0.39,
|
| 4719 |
+
"learning_rate": 0.0005426666666666667,
|
| 4720 |
+
"loss": 2.3053,
|
| 4721 |
+
"step": 779
|
| 4722 |
+
},
|
| 4723 |
+
{
|
| 4724 |
+
"epoch": 0.39,
|
| 4725 |
+
"learning_rate": 0.0005422222222222223,
|
| 4726 |
+
"loss": 2.203,
|
| 4727 |
+
"step": 780
|
| 4728 |
+
},
|
| 4729 |
+
{
|
| 4730 |
+
"epoch": 0.39,
|
| 4731 |
+
"learning_rate": 0.0005417777777777777,
|
| 4732 |
+
"loss": 2.2989,
|
| 4733 |
+
"step": 781
|
| 4734 |
+
},
|
| 4735 |
+
{
|
| 4736 |
+
"epoch": 0.39,
|
| 4737 |
+
"learning_rate": 0.0005413333333333333,
|
| 4738 |
+
"loss": 2.2543,
|
| 4739 |
+
"step": 782
|
| 4740 |
+
},
|
| 4741 |
+
{
|
| 4742 |
+
"epoch": 0.39,
|
| 4743 |
+
"learning_rate": 0.0005408888888888889,
|
| 4744 |
+
"loss": 2.2812,
|
| 4745 |
+
"step": 783
|
| 4746 |
+
},
|
| 4747 |
+
{
|
| 4748 |
+
"epoch": 0.39,
|
| 4749 |
+
"learning_rate": 0.0005404444444444445,
|
| 4750 |
+
"loss": 2.2044,
|
| 4751 |
+
"step": 784
|
| 4752 |
+
},
|
| 4753 |
+
{
|
| 4754 |
+
"epoch": 0.39,
|
| 4755 |
+
"learning_rate": 0.0005400000000000001,
|
| 4756 |
+
"loss": 2.2387,
|
| 4757 |
+
"step": 785
|
| 4758 |
+
},
|
| 4759 |
+
{
|
| 4760 |
+
"epoch": 0.39,
|
| 4761 |
+
"learning_rate": 0.0005395555555555556,
|
| 4762 |
+
"loss": 2.2621,
|
| 4763 |
+
"step": 786
|
| 4764 |
+
},
|
| 4765 |
+
{
|
| 4766 |
+
"epoch": 0.39,
|
| 4767 |
+
"learning_rate": 0.0005391111111111111,
|
| 4768 |
+
"loss": 2.2329,
|
| 4769 |
+
"step": 787
|
| 4770 |
+
},
|
| 4771 |
+
{
|
| 4772 |
+
"epoch": 0.39,
|
| 4773 |
+
"learning_rate": 0.0005386666666666667,
|
| 4774 |
+
"loss": 2.2585,
|
| 4775 |
+
"step": 788
|
| 4776 |
+
},
|
| 4777 |
+
{
|
| 4778 |
+
"epoch": 0.39,
|
| 4779 |
+
"learning_rate": 0.0005382222222222223,
|
| 4780 |
+
"loss": 2.2237,
|
| 4781 |
+
"step": 789
|
| 4782 |
+
},
|
| 4783 |
+
{
|
| 4784 |
+
"epoch": 0.4,
|
| 4785 |
+
"learning_rate": 0.0005377777777777779,
|
| 4786 |
+
"loss": 2.2546,
|
| 4787 |
+
"step": 790
|
| 4788 |
+
},
|
| 4789 |
+
{
|
| 4790 |
+
"epoch": 0.4,
|
| 4791 |
+
"learning_rate": 0.0005373333333333333,
|
| 4792 |
+
"loss": 2.1964,
|
| 4793 |
+
"step": 791
|
| 4794 |
+
},
|
| 4795 |
+
{
|
| 4796 |
+
"epoch": 0.4,
|
| 4797 |
+
"learning_rate": 0.0005368888888888889,
|
| 4798 |
+
"loss": 2.179,
|
| 4799 |
+
"step": 792
|
| 4800 |
+
},
|
| 4801 |
+
{
|
| 4802 |
+
"epoch": 0.4,
|
| 4803 |
+
"learning_rate": 0.0005364444444444445,
|
| 4804 |
+
"loss": 2.2903,
|
| 4805 |
+
"step": 793
|
| 4806 |
+
},
|
| 4807 |
+
{
|
| 4808 |
+
"epoch": 0.4,
|
| 4809 |
+
"learning_rate": 0.000536,
|
| 4810 |
+
"loss": 2.2472,
|
| 4811 |
+
"step": 794
|
| 4812 |
+
},
|
| 4813 |
+
{
|
| 4814 |
+
"epoch": 0.4,
|
| 4815 |
+
"learning_rate": 0.0005355555555555555,
|
| 4816 |
+
"loss": 2.2579,
|
| 4817 |
+
"step": 795
|
| 4818 |
+
},
|
| 4819 |
+
{
|
| 4820 |
+
"epoch": 0.4,
|
| 4821 |
+
"learning_rate": 0.0005351111111111111,
|
| 4822 |
+
"loss": 2.2336,
|
| 4823 |
+
"step": 796
|
| 4824 |
+
},
|
| 4825 |
+
{
|
| 4826 |
+
"epoch": 0.4,
|
| 4827 |
+
"learning_rate": 0.0005346666666666667,
|
| 4828 |
+
"loss": 2.2474,
|
| 4829 |
+
"step": 797
|
| 4830 |
+
},
|
| 4831 |
+
{
|
| 4832 |
+
"epoch": 0.4,
|
| 4833 |
+
"learning_rate": 0.0005342222222222223,
|
| 4834 |
+
"loss": 2.2482,
|
| 4835 |
+
"step": 798
|
| 4836 |
+
},
|
| 4837 |
+
{
|
| 4838 |
+
"epoch": 0.4,
|
| 4839 |
+
"learning_rate": 0.0005337777777777779,
|
| 4840 |
+
"loss": 2.2413,
|
| 4841 |
+
"step": 799
|
| 4842 |
+
},
|
| 4843 |
+
{
|
| 4844 |
+
"epoch": 0.4,
|
| 4845 |
+
"learning_rate": 0.0005333333333333334,
|
| 4846 |
+
"loss": 2.307,
|
| 4847 |
+
"step": 800
|
| 4848 |
+
},
|
| 4849 |
+
{
|
| 4850 |
+
"epoch": 0.4,
|
| 4851 |
+
"eval_gen_len": 1023.0,
|
| 4852 |
+
"eval_loss": 2.158494710922241,
|
| 4853 |
+
"eval_rouge1": 13.1246,
|
| 4854 |
+
"eval_rouge2": 4.3442,
|
| 4855 |
+
"eval_rougeL": 8.3035,
|
| 4856 |
+
"eval_rougeLsum": 8.2848,
|
| 4857 |
+
"eval_runtime": 9293.9809,
|
| 4858 |
+
"eval_samples_per_second": 0.054,
|
| 4859 |
+
"eval_steps_per_second": 0.054,
|
| 4860 |
+
"step": 800
|
| 4861 |
}
|
| 4862 |
],
|
| 4863 |
"max_steps": 2000,
|
| 4864 |
"num_train_epochs": 9223372036854775807,
|
| 4865 |
+
"total_flos": 6.440623327611003e+17,
|
| 4866 |
"trial_name": null,
|
| 4867 |
"trial_params": null
|
| 4868 |
}
|
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 sha256:13d97f15adc41397e040b1276339005dcc4183c282bc569b80bbabc18a3e5da7
|
| 3 |
size 2368281769
|