Training in progress, step 6000
Browse files- last-checkpoint/optimizer.pt +1 -1
- last-checkpoint/pytorch_model.bin +1 -1
- last-checkpoint/rng_state.pth +1 -1
- last-checkpoint/scaler.pt +1 -1
- last-checkpoint/scheduler.pt +1 -1
- last-checkpoint/trainer_state.json +1211 -3
- last-checkpoint/training_args.bin +1 -1
- pytorch_model.bin +1 -1
- training_args.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 893438545
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c104283fc5be7527c7629ba1eead03f13f2fd1b9eb6038061f080de76a878df3
|
| 3 |
size 893438545
|
last-checkpoint/pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 449471589
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:023470fb6e5e8c21b4364835994b9c6ce91ee0f9ee1410357ef32e1ecfa1adb1
|
| 3 |
size 449471589
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 19603
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c2f2d29d89b8e2429a924b6aaa924afbec029bfeb253b782f8506223594c11e3
|
| 3 |
size 19603
|
last-checkpoint/scaler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 559
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0dc04459785e028b6a9c6848cbade4ff6f8a772e3ba1f8cc046c95cd9effcbd9
|
| 3 |
size 559
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 623
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:58231e07c1cb2d094566630dd322447eadb88d23d163c892c3f62365f5615dd9
|
| 3 |
size 623
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
@@ -6046,11 +6046,1219 @@
|
|
| 6046 |
"eval_samples_per_second": 120.919,
|
| 6047 |
"eval_steps_per_second": 1.282,
|
| 6048 |
"step": 5000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 6049 |
}
|
| 6050 |
],
|
| 6051 |
"max_steps": 100000,
|
| 6052 |
"num_train_epochs": 20,
|
| 6053 |
-
"total_flos":
|
| 6054 |
"trial_name": null,
|
| 6055 |
"trial_params": null
|
| 6056 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.1943823500826125,
|
| 5 |
+
"global_step": 6000,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 6046 |
"eval_samples_per_second": 120.919,
|
| 6047 |
"eval_steps_per_second": 1.282,
|
| 6048 |
"step": 5000
|
| 6049 |
+
},
|
| 6050 |
+
{
|
| 6051 |
+
"epoch": 0.0,
|
| 6052 |
+
"learning_rate": 0.00022367543977965679,
|
| 6053 |
+
"loss": 0.4648,
|
| 6054 |
+
"step": 5005
|
| 6055 |
+
},
|
| 6056 |
+
{
|
| 6057 |
+
"epoch": 0.0,
|
| 6058 |
+
"learning_rate": 0.00022367279586002289,
|
| 6059 |
+
"loss": 0.4621,
|
| 6060 |
+
"step": 5010
|
| 6061 |
+
},
|
| 6062 |
+
{
|
| 6063 |
+
"epoch": 0.0,
|
| 6064 |
+
"learning_rate": 0.00022367014932068026,
|
| 6065 |
+
"loss": 0.4606,
|
| 6066 |
+
"step": 5015
|
| 6067 |
+
},
|
| 6068 |
+
{
|
| 6069 |
+
"epoch": 0.0,
|
| 6070 |
+
"learning_rate": 0.00022366750016169416,
|
| 6071 |
+
"loss": 0.458,
|
| 6072 |
+
"step": 5020
|
| 6073 |
+
},
|
| 6074 |
+
{
|
| 6075 |
+
"epoch": 0.0,
|
| 6076 |
+
"learning_rate": 0.00022366484838313004,
|
| 6077 |
+
"loss": 0.4582,
|
| 6078 |
+
"step": 5025
|
| 6079 |
+
},
|
| 6080 |
+
{
|
| 6081 |
+
"epoch": 0.01,
|
| 6082 |
+
"learning_rate": 0.00022366219398505327,
|
| 6083 |
+
"loss": 0.4598,
|
| 6084 |
+
"step": 5030
|
| 6085 |
+
},
|
| 6086 |
+
{
|
| 6087 |
+
"epoch": 0.01,
|
| 6088 |
+
"learning_rate": 0.00022365953696752938,
|
| 6089 |
+
"loss": 0.4644,
|
| 6090 |
+
"step": 5035
|
| 6091 |
+
},
|
| 6092 |
+
{
|
| 6093 |
+
"epoch": 0.01,
|
| 6094 |
+
"learning_rate": 0.0002236568773306239,
|
| 6095 |
+
"loss": 0.4602,
|
| 6096 |
+
"step": 5040
|
| 6097 |
+
},
|
| 6098 |
+
{
|
| 6099 |
+
"epoch": 0.01,
|
| 6100 |
+
"learning_rate": 0.00022365421507440248,
|
| 6101 |
+
"loss": 0.4575,
|
| 6102 |
+
"step": 5045
|
| 6103 |
+
},
|
| 6104 |
+
{
|
| 6105 |
+
"epoch": 0.01,
|
| 6106 |
+
"learning_rate": 0.0002236515501989308,
|
| 6107 |
+
"loss": 0.4571,
|
| 6108 |
+
"step": 5050
|
| 6109 |
+
},
|
| 6110 |
+
{
|
| 6111 |
+
"epoch": 0.01,
|
| 6112 |
+
"learning_rate": 0.00022364888270427457,
|
| 6113 |
+
"loss": 0.4655,
|
| 6114 |
+
"step": 5055
|
| 6115 |
+
},
|
| 6116 |
+
{
|
| 6117 |
+
"epoch": 0.01,
|
| 6118 |
+
"learning_rate": 0.00022364621259049968,
|
| 6119 |
+
"loss": 0.4586,
|
| 6120 |
+
"step": 5060
|
| 6121 |
+
},
|
| 6122 |
+
{
|
| 6123 |
+
"epoch": 0.01,
|
| 6124 |
+
"learning_rate": 0.00022364353985767193,
|
| 6125 |
+
"loss": 0.453,
|
| 6126 |
+
"step": 5065
|
| 6127 |
+
},
|
| 6128 |
+
{
|
| 6129 |
+
"epoch": 0.01,
|
| 6130 |
+
"learning_rate": 0.00022364086450585738,
|
| 6131 |
+
"loss": 0.4537,
|
| 6132 |
+
"step": 5070
|
| 6133 |
+
},
|
| 6134 |
+
{
|
| 6135 |
+
"epoch": 0.01,
|
| 6136 |
+
"learning_rate": 0.00022363818653512192,
|
| 6137 |
+
"loss": 0.4595,
|
| 6138 |
+
"step": 5075
|
| 6139 |
+
},
|
| 6140 |
+
{
|
| 6141 |
+
"epoch": 0.02,
|
| 6142 |
+
"learning_rate": 0.00022363550594553172,
|
| 6143 |
+
"loss": 0.4605,
|
| 6144 |
+
"step": 5080
|
| 6145 |
+
},
|
| 6146 |
+
{
|
| 6147 |
+
"epoch": 0.02,
|
| 6148 |
+
"learning_rate": 0.00022363282273715288,
|
| 6149 |
+
"loss": 0.4587,
|
| 6150 |
+
"step": 5085
|
| 6151 |
+
},
|
| 6152 |
+
{
|
| 6153 |
+
"epoch": 0.02,
|
| 6154 |
+
"learning_rate": 0.0002236301369100516,
|
| 6155 |
+
"loss": 0.4588,
|
| 6156 |
+
"step": 5090
|
| 6157 |
+
},
|
| 6158 |
+
{
|
| 6159 |
+
"epoch": 0.02,
|
| 6160 |
+
"learning_rate": 0.0002236274484642942,
|
| 6161 |
+
"loss": 0.4548,
|
| 6162 |
+
"step": 5095
|
| 6163 |
+
},
|
| 6164 |
+
{
|
| 6165 |
+
"epoch": 0.02,
|
| 6166 |
+
"learning_rate": 0.0002236247573999469,
|
| 6167 |
+
"loss": 0.4513,
|
| 6168 |
+
"step": 5100
|
| 6169 |
+
},
|
| 6170 |
+
{
|
| 6171 |
+
"epoch": 0.02,
|
| 6172 |
+
"learning_rate": 0.00022362206371707622,
|
| 6173 |
+
"loss": 0.4578,
|
| 6174 |
+
"step": 5105
|
| 6175 |
+
},
|
| 6176 |
+
{
|
| 6177 |
+
"epoch": 0.02,
|
| 6178 |
+
"learning_rate": 0.00022361936741574858,
|
| 6179 |
+
"loss": 0.4566,
|
| 6180 |
+
"step": 5110
|
| 6181 |
+
},
|
| 6182 |
+
{
|
| 6183 |
+
"epoch": 0.02,
|
| 6184 |
+
"learning_rate": 0.00022361666849603047,
|
| 6185 |
+
"loss": 0.4558,
|
| 6186 |
+
"step": 5115
|
| 6187 |
+
},
|
| 6188 |
+
{
|
| 6189 |
+
"epoch": 0.02,
|
| 6190 |
+
"learning_rate": 0.00022361396695798856,
|
| 6191 |
+
"loss": 0.4545,
|
| 6192 |
+
"step": 5120
|
| 6193 |
+
},
|
| 6194 |
+
{
|
| 6195 |
+
"epoch": 0.02,
|
| 6196 |
+
"learning_rate": 0.00022361126280168945,
|
| 6197 |
+
"loss": 0.4566,
|
| 6198 |
+
"step": 5125
|
| 6199 |
+
},
|
| 6200 |
+
{
|
| 6201 |
+
"epoch": 0.03,
|
| 6202 |
+
"learning_rate": 0.00022360855602719987,
|
| 6203 |
+
"loss": 0.4567,
|
| 6204 |
+
"step": 5130
|
| 6205 |
+
},
|
| 6206 |
+
{
|
| 6207 |
+
"epoch": 0.03,
|
| 6208 |
+
"learning_rate": 0.0002236058466345866,
|
| 6209 |
+
"loss": 0.4579,
|
| 6210 |
+
"step": 5135
|
| 6211 |
+
},
|
| 6212 |
+
{
|
| 6213 |
+
"epoch": 0.03,
|
| 6214 |
+
"learning_rate": 0.0002236031346239166,
|
| 6215 |
+
"loss": 0.4592,
|
| 6216 |
+
"step": 5140
|
| 6217 |
+
},
|
| 6218 |
+
{
|
| 6219 |
+
"epoch": 0.03,
|
| 6220 |
+
"learning_rate": 0.0002236004199952566,
|
| 6221 |
+
"loss": 0.4594,
|
| 6222 |
+
"step": 5145
|
| 6223 |
+
},
|
| 6224 |
+
{
|
| 6225 |
+
"epoch": 0.03,
|
| 6226 |
+
"learning_rate": 0.0002235977027486737,
|
| 6227 |
+
"loss": 0.4626,
|
| 6228 |
+
"step": 5150
|
| 6229 |
+
},
|
| 6230 |
+
{
|
| 6231 |
+
"epoch": 0.03,
|
| 6232 |
+
"learning_rate": 0.00022359498288423494,
|
| 6233 |
+
"loss": 0.4552,
|
| 6234 |
+
"step": 5155
|
| 6235 |
+
},
|
| 6236 |
+
{
|
| 6237 |
+
"epoch": 0.03,
|
| 6238 |
+
"learning_rate": 0.0002235922604020074,
|
| 6239 |
+
"loss": 0.4586,
|
| 6240 |
+
"step": 5160
|
| 6241 |
+
},
|
| 6242 |
+
{
|
| 6243 |
+
"epoch": 0.03,
|
| 6244 |
+
"learning_rate": 0.00022358953530205827,
|
| 6245 |
+
"loss": 0.4558,
|
| 6246 |
+
"step": 5165
|
| 6247 |
+
},
|
| 6248 |
+
{
|
| 6249 |
+
"epoch": 0.03,
|
| 6250 |
+
"learning_rate": 0.00022358680758445475,
|
| 6251 |
+
"loss": 0.4571,
|
| 6252 |
+
"step": 5170
|
| 6253 |
+
},
|
| 6254 |
+
{
|
| 6255 |
+
"epoch": 0.03,
|
| 6256 |
+
"learning_rate": 0.0002235840772492642,
|
| 6257 |
+
"loss": 0.4641,
|
| 6258 |
+
"step": 5175
|
| 6259 |
+
},
|
| 6260 |
+
{
|
| 6261 |
+
"epoch": 0.03,
|
| 6262 |
+
"learning_rate": 0.00022358134429655398,
|
| 6263 |
+
"loss": 0.4559,
|
| 6264 |
+
"step": 5180
|
| 6265 |
+
},
|
| 6266 |
+
{
|
| 6267 |
+
"epoch": 0.04,
|
| 6268 |
+
"learning_rate": 0.0002235786087263915,
|
| 6269 |
+
"loss": 0.4578,
|
| 6270 |
+
"step": 5185
|
| 6271 |
+
},
|
| 6272 |
+
{
|
| 6273 |
+
"epoch": 0.04,
|
| 6274 |
+
"learning_rate": 0.00022357587053884427,
|
| 6275 |
+
"loss": 0.4545,
|
| 6276 |
+
"step": 5190
|
| 6277 |
+
},
|
| 6278 |
+
{
|
| 6279 |
+
"epoch": 0.04,
|
| 6280 |
+
"learning_rate": 0.00022357312973397982,
|
| 6281 |
+
"loss": 0.4592,
|
| 6282 |
+
"step": 5195
|
| 6283 |
+
},
|
| 6284 |
+
{
|
| 6285 |
+
"epoch": 0.04,
|
| 6286 |
+
"learning_rate": 0.00022357038631186585,
|
| 6287 |
+
"loss": 0.453,
|
| 6288 |
+
"step": 5200
|
| 6289 |
+
},
|
| 6290 |
+
{
|
| 6291 |
+
"epoch": 0.04,
|
| 6292 |
+
"learning_rate": 0.00022356764027256997,
|
| 6293 |
+
"loss": 0.4537,
|
| 6294 |
+
"step": 5205
|
| 6295 |
+
},
|
| 6296 |
+
{
|
| 6297 |
+
"epoch": 0.04,
|
| 6298 |
+
"learning_rate": 0.00022356489161616,
|
| 6299 |
+
"loss": 0.4521,
|
| 6300 |
+
"step": 5210
|
| 6301 |
+
},
|
| 6302 |
+
{
|
| 6303 |
+
"epoch": 0.04,
|
| 6304 |
+
"learning_rate": 0.00022356214034270372,
|
| 6305 |
+
"loss": 0.4589,
|
| 6306 |
+
"step": 5215
|
| 6307 |
+
},
|
| 6308 |
+
{
|
| 6309 |
+
"epoch": 0.04,
|
| 6310 |
+
"learning_rate": 0.000223559386452269,
|
| 6311 |
+
"loss": 0.4591,
|
| 6312 |
+
"step": 5220
|
| 6313 |
+
},
|
| 6314 |
+
{
|
| 6315 |
+
"epoch": 0.04,
|
| 6316 |
+
"learning_rate": 0.0002235566299449239,
|
| 6317 |
+
"loss": 0.4585,
|
| 6318 |
+
"step": 5225
|
| 6319 |
+
},
|
| 6320 |
+
{
|
| 6321 |
+
"epoch": 0.04,
|
| 6322 |
+
"learning_rate": 0.00022355387082073625,
|
| 6323 |
+
"loss": 0.4547,
|
| 6324 |
+
"step": 5230
|
| 6325 |
+
},
|
| 6326 |
+
{
|
| 6327 |
+
"epoch": 0.05,
|
| 6328 |
+
"learning_rate": 0.00022355110907977427,
|
| 6329 |
+
"loss": 0.4549,
|
| 6330 |
+
"step": 5235
|
| 6331 |
+
},
|
| 6332 |
+
{
|
| 6333 |
+
"epoch": 0.05,
|
| 6334 |
+
"learning_rate": 0.0002235483447221061,
|
| 6335 |
+
"loss": 0.4561,
|
| 6336 |
+
"step": 5240
|
| 6337 |
+
},
|
| 6338 |
+
{
|
| 6339 |
+
"epoch": 0.05,
|
| 6340 |
+
"learning_rate": 0.00022354557774779984,
|
| 6341 |
+
"loss": 0.4564,
|
| 6342 |
+
"step": 5245
|
| 6343 |
+
},
|
| 6344 |
+
{
|
| 6345 |
+
"epoch": 0.05,
|
| 6346 |
+
"learning_rate": 0.0002235428081569239,
|
| 6347 |
+
"loss": 0.4511,
|
| 6348 |
+
"step": 5250
|
| 6349 |
+
},
|
| 6350 |
+
{
|
| 6351 |
+
"epoch": 0.05,
|
| 6352 |
+
"learning_rate": 0.0002235400359495465,
|
| 6353 |
+
"loss": 0.4567,
|
| 6354 |
+
"step": 5255
|
| 6355 |
+
},
|
| 6356 |
+
{
|
| 6357 |
+
"epoch": 0.05,
|
| 6358 |
+
"learning_rate": 0.00022353726112573613,
|
| 6359 |
+
"loss": 0.4551,
|
| 6360 |
+
"step": 5260
|
| 6361 |
+
},
|
| 6362 |
+
{
|
| 6363 |
+
"epoch": 0.05,
|
| 6364 |
+
"learning_rate": 0.00022353448368556117,
|
| 6365 |
+
"loss": 0.4587,
|
| 6366 |
+
"step": 5265
|
| 6367 |
+
},
|
| 6368 |
+
{
|
| 6369 |
+
"epoch": 0.05,
|
| 6370 |
+
"learning_rate": 0.00022353170362909025,
|
| 6371 |
+
"loss": 0.4581,
|
| 6372 |
+
"step": 5270
|
| 6373 |
+
},
|
| 6374 |
+
{
|
| 6375 |
+
"epoch": 0.05,
|
| 6376 |
+
"learning_rate": 0.0002235289209563919,
|
| 6377 |
+
"loss": 0.4544,
|
| 6378 |
+
"step": 5275
|
| 6379 |
+
},
|
| 6380 |
+
{
|
| 6381 |
+
"epoch": 0.05,
|
| 6382 |
+
"learning_rate": 0.00022352613566753478,
|
| 6383 |
+
"loss": 0.4606,
|
| 6384 |
+
"step": 5280
|
| 6385 |
+
},
|
| 6386 |
+
{
|
| 6387 |
+
"epoch": 0.06,
|
| 6388 |
+
"learning_rate": 0.00022352334776258766,
|
| 6389 |
+
"loss": 0.4644,
|
| 6390 |
+
"step": 5285
|
| 6391 |
+
},
|
| 6392 |
+
{
|
| 6393 |
+
"epoch": 0.06,
|
| 6394 |
+
"learning_rate": 0.00022352055724161925,
|
| 6395 |
+
"loss": 0.4574,
|
| 6396 |
+
"step": 5290
|
| 6397 |
+
},
|
| 6398 |
+
{
|
| 6399 |
+
"epoch": 0.06,
|
| 6400 |
+
"learning_rate": 0.00022351776410469847,
|
| 6401 |
+
"loss": 0.4538,
|
| 6402 |
+
"step": 5295
|
| 6403 |
+
},
|
| 6404 |
+
{
|
| 6405 |
+
"epoch": 0.06,
|
| 6406 |
+
"learning_rate": 0.0002235149683518942,
|
| 6407 |
+
"loss": 0.4512,
|
| 6408 |
+
"step": 5300
|
| 6409 |
+
},
|
| 6410 |
+
{
|
| 6411 |
+
"epoch": 0.06,
|
| 6412 |
+
"learning_rate": 0.00022351216998327549,
|
| 6413 |
+
"loss": 0.4552,
|
| 6414 |
+
"step": 5305
|
| 6415 |
+
},
|
| 6416 |
+
{
|
| 6417 |
+
"epoch": 0.06,
|
| 6418 |
+
"learning_rate": 0.0002235093689989113,
|
| 6419 |
+
"loss": 0.4555,
|
| 6420 |
+
"step": 5310
|
| 6421 |
+
},
|
| 6422 |
+
{
|
| 6423 |
+
"epoch": 0.06,
|
| 6424 |
+
"learning_rate": 0.00022350656539887076,
|
| 6425 |
+
"loss": 0.4557,
|
| 6426 |
+
"step": 5315
|
| 6427 |
+
},
|
| 6428 |
+
{
|
| 6429 |
+
"epoch": 0.06,
|
| 6430 |
+
"learning_rate": 0.0002235037591832231,
|
| 6431 |
+
"loss": 0.459,
|
| 6432 |
+
"step": 5320
|
| 6433 |
+
},
|
| 6434 |
+
{
|
| 6435 |
+
"epoch": 0.06,
|
| 6436 |
+
"learning_rate": 0.0002235009503520375,
|
| 6437 |
+
"loss": 0.4565,
|
| 6438 |
+
"step": 5325
|
| 6439 |
+
},
|
| 6440 |
+
{
|
| 6441 |
+
"epoch": 0.06,
|
| 6442 |
+
"learning_rate": 0.00022349813890538332,
|
| 6443 |
+
"loss": 0.46,
|
| 6444 |
+
"step": 5330
|
| 6445 |
+
},
|
| 6446 |
+
{
|
| 6447 |
+
"epoch": 0.07,
|
| 6448 |
+
"learning_rate": 0.00022349532484332988,
|
| 6449 |
+
"loss": 0.4528,
|
| 6450 |
+
"step": 5335
|
| 6451 |
+
},
|
| 6452 |
+
{
|
| 6453 |
+
"epoch": 0.07,
|
| 6454 |
+
"learning_rate": 0.00022349250816594663,
|
| 6455 |
+
"loss": 0.4481,
|
| 6456 |
+
"step": 5340
|
| 6457 |
+
},
|
| 6458 |
+
{
|
| 6459 |
+
"epoch": 0.07,
|
| 6460 |
+
"learning_rate": 0.00022348968887330307,
|
| 6461 |
+
"loss": 0.4566,
|
| 6462 |
+
"step": 5345
|
| 6463 |
+
},
|
| 6464 |
+
{
|
| 6465 |
+
"epoch": 0.07,
|
| 6466 |
+
"learning_rate": 0.00022348686696546877,
|
| 6467 |
+
"loss": 0.4574,
|
| 6468 |
+
"step": 5350
|
| 6469 |
+
},
|
| 6470 |
+
{
|
| 6471 |
+
"epoch": 0.07,
|
| 6472 |
+
"learning_rate": 0.00022348404244251335,
|
| 6473 |
+
"loss": 0.4526,
|
| 6474 |
+
"step": 5355
|
| 6475 |
+
},
|
| 6476 |
+
{
|
| 6477 |
+
"epoch": 0.07,
|
| 6478 |
+
"learning_rate": 0.00022348121530450652,
|
| 6479 |
+
"loss": 0.4575,
|
| 6480 |
+
"step": 5360
|
| 6481 |
+
},
|
| 6482 |
+
{
|
| 6483 |
+
"epoch": 0.07,
|
| 6484 |
+
"learning_rate": 0.00022347838555151803,
|
| 6485 |
+
"loss": 0.4464,
|
| 6486 |
+
"step": 5365
|
| 6487 |
+
},
|
| 6488 |
+
{
|
| 6489 |
+
"epoch": 0.07,
|
| 6490 |
+
"learning_rate": 0.00022347555318361767,
|
| 6491 |
+
"loss": 0.4602,
|
| 6492 |
+
"step": 5370
|
| 6493 |
+
},
|
| 6494 |
+
{
|
| 6495 |
+
"epoch": 0.07,
|
| 6496 |
+
"learning_rate": 0.00022347271820087535,
|
| 6497 |
+
"loss": 0.4585,
|
| 6498 |
+
"step": 5375
|
| 6499 |
+
},
|
| 6500 |
+
{
|
| 6501 |
+
"epoch": 0.07,
|
| 6502 |
+
"learning_rate": 0.00022346988060336105,
|
| 6503 |
+
"loss": 0.4607,
|
| 6504 |
+
"step": 5380
|
| 6505 |
+
},
|
| 6506 |
+
{
|
| 6507 |
+
"epoch": 0.07,
|
| 6508 |
+
"learning_rate": 0.00022346704039114473,
|
| 6509 |
+
"loss": 0.4565,
|
| 6510 |
+
"step": 5385
|
| 6511 |
+
},
|
| 6512 |
+
{
|
| 6513 |
+
"epoch": 0.08,
|
| 6514 |
+
"learning_rate": 0.0002234641975642965,
|
| 6515 |
+
"loss": 0.4513,
|
| 6516 |
+
"step": 5390
|
| 6517 |
+
},
|
| 6518 |
+
{
|
| 6519 |
+
"epoch": 0.08,
|
| 6520 |
+
"learning_rate": 0.0002234613521228865,
|
| 6521 |
+
"loss": 0.4559,
|
| 6522 |
+
"step": 5395
|
| 6523 |
+
},
|
| 6524 |
+
{
|
| 6525 |
+
"epoch": 0.08,
|
| 6526 |
+
"learning_rate": 0.00022345850406698493,
|
| 6527 |
+
"loss": 0.4527,
|
| 6528 |
+
"step": 5400
|
| 6529 |
+
},
|
| 6530 |
+
{
|
| 6531 |
+
"epoch": 0.08,
|
| 6532 |
+
"learning_rate": 0.00022345565339666211,
|
| 6533 |
+
"loss": 0.4537,
|
| 6534 |
+
"step": 5405
|
| 6535 |
+
},
|
| 6536 |
+
{
|
| 6537 |
+
"epoch": 0.08,
|
| 6538 |
+
"learning_rate": 0.0002234528001119883,
|
| 6539 |
+
"loss": 0.4511,
|
| 6540 |
+
"step": 5410
|
| 6541 |
+
},
|
| 6542 |
+
{
|
| 6543 |
+
"epoch": 0.08,
|
| 6544 |
+
"learning_rate": 0.00022344994421303393,
|
| 6545 |
+
"loss": 0.451,
|
| 6546 |
+
"step": 5415
|
| 6547 |
+
},
|
| 6548 |
+
{
|
| 6549 |
+
"epoch": 0.08,
|
| 6550 |
+
"learning_rate": 0.00022344708569986947,
|
| 6551 |
+
"loss": 0.457,
|
| 6552 |
+
"step": 5420
|
| 6553 |
+
},
|
| 6554 |
+
{
|
| 6555 |
+
"epoch": 0.08,
|
| 6556 |
+
"learning_rate": 0.0002234442245725655,
|
| 6557 |
+
"loss": 0.4526,
|
| 6558 |
+
"step": 5425
|
| 6559 |
+
},
|
| 6560 |
+
{
|
| 6561 |
+
"epoch": 0.08,
|
| 6562 |
+
"learning_rate": 0.00022344136083119257,
|
| 6563 |
+
"loss": 0.4582,
|
| 6564 |
+
"step": 5430
|
| 6565 |
+
},
|
| 6566 |
+
{
|
| 6567 |
+
"epoch": 0.08,
|
| 6568 |
+
"learning_rate": 0.0002234384944758213,
|
| 6569 |
+
"loss": 0.4534,
|
| 6570 |
+
"step": 5435
|
| 6571 |
+
},
|
| 6572 |
+
{
|
| 6573 |
+
"epoch": 0.09,
|
| 6574 |
+
"learning_rate": 0.0002234356255065225,
|
| 6575 |
+
"loss": 0.4567,
|
| 6576 |
+
"step": 5440
|
| 6577 |
+
},
|
| 6578 |
+
{
|
| 6579 |
+
"epoch": 0.09,
|
| 6580 |
+
"learning_rate": 0.00022343275392336688,
|
| 6581 |
+
"loss": 0.4521,
|
| 6582 |
+
"step": 5445
|
| 6583 |
+
},
|
| 6584 |
+
{
|
| 6585 |
+
"epoch": 0.09,
|
| 6586 |
+
"learning_rate": 0.00022342987972642534,
|
| 6587 |
+
"loss": 0.4604,
|
| 6588 |
+
"step": 5450
|
| 6589 |
+
},
|
| 6590 |
+
{
|
| 6591 |
+
"epoch": 0.09,
|
| 6592 |
+
"learning_rate": 0.00022342700291576878,
|
| 6593 |
+
"loss": 0.4479,
|
| 6594 |
+
"step": 5455
|
| 6595 |
+
},
|
| 6596 |
+
{
|
| 6597 |
+
"epoch": 0.09,
|
| 6598 |
+
"learning_rate": 0.00022342412349146823,
|
| 6599 |
+
"loss": 0.4516,
|
| 6600 |
+
"step": 5460
|
| 6601 |
+
},
|
| 6602 |
+
{
|
| 6603 |
+
"epoch": 0.09,
|
| 6604 |
+
"learning_rate": 0.00022342124145359465,
|
| 6605 |
+
"loss": 0.4534,
|
| 6606 |
+
"step": 5465
|
| 6607 |
+
},
|
| 6608 |
+
{
|
| 6609 |
+
"epoch": 0.09,
|
| 6610 |
+
"learning_rate": 0.00022341835680221925,
|
| 6611 |
+
"loss": 0.4586,
|
| 6612 |
+
"step": 5470
|
| 6613 |
+
},
|
| 6614 |
+
{
|
| 6615 |
+
"epoch": 0.09,
|
| 6616 |
+
"learning_rate": 0.0002234154695374131,
|
| 6617 |
+
"loss": 0.4573,
|
| 6618 |
+
"step": 5475
|
| 6619 |
+
},
|
| 6620 |
+
{
|
| 6621 |
+
"epoch": 0.09,
|
| 6622 |
+
"learning_rate": 0.0002234125796592475,
|
| 6623 |
+
"loss": 0.4575,
|
| 6624 |
+
"step": 5480
|
| 6625 |
+
},
|
| 6626 |
+
{
|
| 6627 |
+
"epoch": 0.09,
|
| 6628 |
+
"learning_rate": 0.00022340968716779377,
|
| 6629 |
+
"loss": 0.4497,
|
| 6630 |
+
"step": 5485
|
| 6631 |
+
},
|
| 6632 |
+
{
|
| 6633 |
+
"epoch": 0.1,
|
| 6634 |
+
"learning_rate": 0.00022340679206312328,
|
| 6635 |
+
"loss": 0.4495,
|
| 6636 |
+
"step": 5490
|
| 6637 |
+
},
|
| 6638 |
+
{
|
| 6639 |
+
"epoch": 0.1,
|
| 6640 |
+
"learning_rate": 0.00022340389434530742,
|
| 6641 |
+
"loss": 0.4587,
|
| 6642 |
+
"step": 5495
|
| 6643 |
+
},
|
| 6644 |
+
{
|
| 6645 |
+
"epoch": 0.1,
|
| 6646 |
+
"learning_rate": 0.0002234009940144177,
|
| 6647 |
+
"loss": 0.4552,
|
| 6648 |
+
"step": 5500
|
| 6649 |
+
},
|
| 6650 |
+
{
|
| 6651 |
+
"epoch": 0.1,
|
| 6652 |
+
"learning_rate": 0.0002233980910705257,
|
| 6653 |
+
"loss": 0.4579,
|
| 6654 |
+
"step": 5505
|
| 6655 |
+
},
|
| 6656 |
+
{
|
| 6657 |
+
"epoch": 0.1,
|
| 6658 |
+
"learning_rate": 0.00022339518551370304,
|
| 6659 |
+
"loss": 0.4518,
|
| 6660 |
+
"step": 5510
|
| 6661 |
+
},
|
| 6662 |
+
{
|
| 6663 |
+
"epoch": 0.1,
|
| 6664 |
+
"learning_rate": 0.0002233922773440214,
|
| 6665 |
+
"loss": 0.45,
|
| 6666 |
+
"step": 5515
|
| 6667 |
+
},
|
| 6668 |
+
{
|
| 6669 |
+
"epoch": 0.1,
|
| 6670 |
+
"learning_rate": 0.0002233893665615526,
|
| 6671 |
+
"loss": 0.4591,
|
| 6672 |
+
"step": 5520
|
| 6673 |
+
},
|
| 6674 |
+
{
|
| 6675 |
+
"epoch": 0.1,
|
| 6676 |
+
"learning_rate": 0.00022338645316636837,
|
| 6677 |
+
"loss": 0.4546,
|
| 6678 |
+
"step": 5525
|
| 6679 |
+
},
|
| 6680 |
+
{
|
| 6681 |
+
"epoch": 0.1,
|
| 6682 |
+
"learning_rate": 0.00022338353715854064,
|
| 6683 |
+
"loss": 0.4596,
|
| 6684 |
+
"step": 5530
|
| 6685 |
+
},
|
| 6686 |
+
{
|
| 6687 |
+
"epoch": 0.1,
|
| 6688 |
+
"learning_rate": 0.00022338061853814135,
|
| 6689 |
+
"loss": 0.4537,
|
| 6690 |
+
"step": 5535
|
| 6691 |
+
},
|
| 6692 |
+
{
|
| 6693 |
+
"epoch": 0.1,
|
| 6694 |
+
"learning_rate": 0.00022337769730524252,
|
| 6695 |
+
"loss": 0.4606,
|
| 6696 |
+
"step": 5540
|
| 6697 |
+
},
|
| 6698 |
+
{
|
| 6699 |
+
"epoch": 0.11,
|
| 6700 |
+
"learning_rate": 0.00022337477345991624,
|
| 6701 |
+
"loss": 0.4551,
|
| 6702 |
+
"step": 5545
|
| 6703 |
+
},
|
| 6704 |
+
{
|
| 6705 |
+
"epoch": 0.11,
|
| 6706 |
+
"learning_rate": 0.00022337184700223464,
|
| 6707 |
+
"loss": 0.4545,
|
| 6708 |
+
"step": 5550
|
| 6709 |
+
},
|
| 6710 |
+
{
|
| 6711 |
+
"epoch": 0.11,
|
| 6712 |
+
"learning_rate": 0.0002233689179322699,
|
| 6713 |
+
"loss": 0.4543,
|
| 6714 |
+
"step": 5555
|
| 6715 |
+
},
|
| 6716 |
+
{
|
| 6717 |
+
"epoch": 0.11,
|
| 6718 |
+
"learning_rate": 0.00022336598625009434,
|
| 6719 |
+
"loss": 0.4531,
|
| 6720 |
+
"step": 5560
|
| 6721 |
+
},
|
| 6722 |
+
{
|
| 6723 |
+
"epoch": 0.11,
|
| 6724 |
+
"learning_rate": 0.00022336305195578028,
|
| 6725 |
+
"loss": 0.4541,
|
| 6726 |
+
"step": 5565
|
| 6727 |
+
},
|
| 6728 |
+
{
|
| 6729 |
+
"epoch": 0.11,
|
| 6730 |
+
"learning_rate": 0.0002233601150494001,
|
| 6731 |
+
"loss": 0.4593,
|
| 6732 |
+
"step": 5570
|
| 6733 |
+
},
|
| 6734 |
+
{
|
| 6735 |
+
"epoch": 0.11,
|
| 6736 |
+
"learning_rate": 0.0002233571755310263,
|
| 6737 |
+
"loss": 0.4561,
|
| 6738 |
+
"step": 5575
|
| 6739 |
+
},
|
| 6740 |
+
{
|
| 6741 |
+
"epoch": 0.11,
|
| 6742 |
+
"learning_rate": 0.00022335423340073138,
|
| 6743 |
+
"loss": 0.4502,
|
| 6744 |
+
"step": 5580
|
| 6745 |
+
},
|
| 6746 |
+
{
|
| 6747 |
+
"epoch": 0.11,
|
| 6748 |
+
"learning_rate": 0.00022335128865858797,
|
| 6749 |
+
"loss": 0.4487,
|
| 6750 |
+
"step": 5585
|
| 6751 |
+
},
|
| 6752 |
+
{
|
| 6753 |
+
"epoch": 0.11,
|
| 6754 |
+
"learning_rate": 0.0002233483413046687,
|
| 6755 |
+
"loss": 0.4479,
|
| 6756 |
+
"step": 5590
|
| 6757 |
+
},
|
| 6758 |
+
{
|
| 6759 |
+
"epoch": 0.12,
|
| 6760 |
+
"learning_rate": 0.00022334539133904625,
|
| 6761 |
+
"loss": 0.4519,
|
| 6762 |
+
"step": 5595
|
| 6763 |
+
},
|
| 6764 |
+
{
|
| 6765 |
+
"epoch": 0.12,
|
| 6766 |
+
"learning_rate": 0.0002233424387617935,
|
| 6767 |
+
"loss": 0.4529,
|
| 6768 |
+
"step": 5600
|
| 6769 |
+
},
|
| 6770 |
+
{
|
| 6771 |
+
"epoch": 0.12,
|
| 6772 |
+
"learning_rate": 0.00022333948357298322,
|
| 6773 |
+
"loss": 0.4502,
|
| 6774 |
+
"step": 5605
|
| 6775 |
+
},
|
| 6776 |
+
{
|
| 6777 |
+
"epoch": 0.12,
|
| 6778 |
+
"learning_rate": 0.0002233365257726884,
|
| 6779 |
+
"loss": 0.4492,
|
| 6780 |
+
"step": 5610
|
| 6781 |
+
},
|
| 6782 |
+
{
|
| 6783 |
+
"epoch": 0.12,
|
| 6784 |
+
"learning_rate": 0.00022333356536098197,
|
| 6785 |
+
"loss": 0.4525,
|
| 6786 |
+
"step": 5615
|
| 6787 |
+
},
|
| 6788 |
+
{
|
| 6789 |
+
"epoch": 0.12,
|
| 6790 |
+
"learning_rate": 0.000223330602337937,
|
| 6791 |
+
"loss": 0.4538,
|
| 6792 |
+
"step": 5620
|
| 6793 |
+
},
|
| 6794 |
+
{
|
| 6795 |
+
"epoch": 0.12,
|
| 6796 |
+
"learning_rate": 0.00022332763670362656,
|
| 6797 |
+
"loss": 0.4535,
|
| 6798 |
+
"step": 5625
|
| 6799 |
+
},
|
| 6800 |
+
{
|
| 6801 |
+
"epoch": 0.12,
|
| 6802 |
+
"learning_rate": 0.00022332466845812387,
|
| 6803 |
+
"loss": 0.4529,
|
| 6804 |
+
"step": 5630
|
| 6805 |
+
},
|
| 6806 |
+
{
|
| 6807 |
+
"epoch": 0.12,
|
| 6808 |
+
"learning_rate": 0.00022332169760150218,
|
| 6809 |
+
"loss": 0.4513,
|
| 6810 |
+
"step": 5635
|
| 6811 |
+
},
|
| 6812 |
+
{
|
| 6813 |
+
"epoch": 0.12,
|
| 6814 |
+
"learning_rate": 0.00022331872413383473,
|
| 6815 |
+
"loss": 0.4501,
|
| 6816 |
+
"step": 5640
|
| 6817 |
+
},
|
| 6818 |
+
{
|
| 6819 |
+
"epoch": 0.13,
|
| 6820 |
+
"learning_rate": 0.00022331574805519494,
|
| 6821 |
+
"loss": 0.4476,
|
| 6822 |
+
"step": 5645
|
| 6823 |
+
},
|
| 6824 |
+
{
|
| 6825 |
+
"epoch": 0.13,
|
| 6826 |
+
"learning_rate": 0.0002233127693656562,
|
| 6827 |
+
"loss": 0.4496,
|
| 6828 |
+
"step": 5650
|
| 6829 |
+
},
|
| 6830 |
+
{
|
| 6831 |
+
"epoch": 0.13,
|
| 6832 |
+
"learning_rate": 0.00022330978806529202,
|
| 6833 |
+
"loss": 0.4583,
|
| 6834 |
+
"step": 5655
|
| 6835 |
+
},
|
| 6836 |
+
{
|
| 6837 |
+
"epoch": 0.13,
|
| 6838 |
+
"learning_rate": 0.00022330680415417602,
|
| 6839 |
+
"loss": 0.4534,
|
| 6840 |
+
"step": 5660
|
| 6841 |
+
},
|
| 6842 |
+
{
|
| 6843 |
+
"epoch": 0.13,
|
| 6844 |
+
"learning_rate": 0.00022330381763238172,
|
| 6845 |
+
"loss": 0.4536,
|
| 6846 |
+
"step": 5665
|
| 6847 |
+
},
|
| 6848 |
+
{
|
| 6849 |
+
"epoch": 0.13,
|
| 6850 |
+
"learning_rate": 0.00022330082849998293,
|
| 6851 |
+
"loss": 0.4564,
|
| 6852 |
+
"step": 5670
|
| 6853 |
+
},
|
| 6854 |
+
{
|
| 6855 |
+
"epoch": 0.13,
|
| 6856 |
+
"learning_rate": 0.00022329783675705328,
|
| 6857 |
+
"loss": 0.4589,
|
| 6858 |
+
"step": 5675
|
| 6859 |
+
},
|
| 6860 |
+
{
|
| 6861 |
+
"epoch": 0.13,
|
| 6862 |
+
"learning_rate": 0.0002232948424036667,
|
| 6863 |
+
"loss": 0.4596,
|
| 6864 |
+
"step": 5680
|
| 6865 |
+
},
|
| 6866 |
+
{
|
| 6867 |
+
"epoch": 0.13,
|
| 6868 |
+
"learning_rate": 0.00022329184543989696,
|
| 6869 |
+
"loss": 0.4539,
|
| 6870 |
+
"step": 5685
|
| 6871 |
+
},
|
| 6872 |
+
{
|
| 6873 |
+
"epoch": 0.13,
|
| 6874 |
+
"learning_rate": 0.0002232888458658181,
|
| 6875 |
+
"loss": 0.4567,
|
| 6876 |
+
"step": 5690
|
| 6877 |
+
},
|
| 6878 |
+
{
|
| 6879 |
+
"epoch": 0.14,
|
| 6880 |
+
"learning_rate": 0.0002232858436815041,
|
| 6881 |
+
"loss": 0.4576,
|
| 6882 |
+
"step": 5695
|
| 6883 |
+
},
|
| 6884 |
+
{
|
| 6885 |
+
"epoch": 0.14,
|
| 6886 |
+
"learning_rate": 0.00022328283888702904,
|
| 6887 |
+
"loss": 0.4507,
|
| 6888 |
+
"step": 5700
|
| 6889 |
+
},
|
| 6890 |
+
{
|
| 6891 |
+
"epoch": 0.14,
|
| 6892 |
+
"learning_rate": 0.00022327983148246703,
|
| 6893 |
+
"loss": 0.4518,
|
| 6894 |
+
"step": 5705
|
| 6895 |
+
},
|
| 6896 |
+
{
|
| 6897 |
+
"epoch": 0.14,
|
| 6898 |
+
"learning_rate": 0.00022327682146789233,
|
| 6899 |
+
"loss": 0.453,
|
| 6900 |
+
"step": 5710
|
| 6901 |
+
},
|
| 6902 |
+
{
|
| 6903 |
+
"epoch": 0.14,
|
| 6904 |
+
"learning_rate": 0.00022327380884337914,
|
| 6905 |
+
"loss": 0.4573,
|
| 6906 |
+
"step": 5715
|
| 6907 |
+
},
|
| 6908 |
+
{
|
| 6909 |
+
"epoch": 0.14,
|
| 6910 |
+
"learning_rate": 0.00022327079360900186,
|
| 6911 |
+
"loss": 0.4555,
|
| 6912 |
+
"step": 5720
|
| 6913 |
+
},
|
| 6914 |
+
{
|
| 6915 |
+
"epoch": 0.14,
|
| 6916 |
+
"learning_rate": 0.00022326777576483486,
|
| 6917 |
+
"loss": 0.4567,
|
| 6918 |
+
"step": 5725
|
| 6919 |
+
},
|
| 6920 |
+
{
|
| 6921 |
+
"epoch": 0.14,
|
| 6922 |
+
"learning_rate": 0.00022326475531095257,
|
| 6923 |
+
"loss": 0.4533,
|
| 6924 |
+
"step": 5730
|
| 6925 |
+
},
|
| 6926 |
+
{
|
| 6927 |
+
"epoch": 0.14,
|
| 6928 |
+
"learning_rate": 0.00022326173224742962,
|
| 6929 |
+
"loss": 0.453,
|
| 6930 |
+
"step": 5735
|
| 6931 |
+
},
|
| 6932 |
+
{
|
| 6933 |
+
"epoch": 0.14,
|
| 6934 |
+
"learning_rate": 0.00022325870657434048,
|
| 6935 |
+
"loss": 0.4571,
|
| 6936 |
+
"step": 5740
|
| 6937 |
+
},
|
| 6938 |
+
{
|
| 6939 |
+
"epoch": 0.14,
|
| 6940 |
+
"learning_rate": 0.00022325567829175984,
|
| 6941 |
+
"loss": 0.4526,
|
| 6942 |
+
"step": 5745
|
| 6943 |
+
},
|
| 6944 |
+
{
|
| 6945 |
+
"epoch": 0.15,
|
| 6946 |
+
"learning_rate": 0.00022325264739976247,
|
| 6947 |
+
"loss": 0.4508,
|
| 6948 |
+
"step": 5750
|
| 6949 |
+
},
|
| 6950 |
+
{
|
| 6951 |
+
"epoch": 0.15,
|
| 6952 |
+
"learning_rate": 0.0002232496138984231,
|
| 6953 |
+
"loss": 0.4553,
|
| 6954 |
+
"step": 5755
|
| 6955 |
+
},
|
| 6956 |
+
{
|
| 6957 |
+
"epoch": 0.15,
|
| 6958 |
+
"learning_rate": 0.0002232465777878166,
|
| 6959 |
+
"loss": 0.4559,
|
| 6960 |
+
"step": 5760
|
| 6961 |
+
},
|
| 6962 |
+
{
|
| 6963 |
+
"epoch": 0.15,
|
| 6964 |
+
"learning_rate": 0.0002232435390680179,
|
| 6965 |
+
"loss": 0.45,
|
| 6966 |
+
"step": 5765
|
| 6967 |
+
},
|
| 6968 |
+
{
|
| 6969 |
+
"epoch": 0.15,
|
| 6970 |
+
"learning_rate": 0.00022324049773910195,
|
| 6971 |
+
"loss": 0.4571,
|
| 6972 |
+
"step": 5770
|
| 6973 |
+
},
|
| 6974 |
+
{
|
| 6975 |
+
"epoch": 0.15,
|
| 6976 |
+
"learning_rate": 0.0002232374538011438,
|
| 6977 |
+
"loss": 0.4584,
|
| 6978 |
+
"step": 5775
|
| 6979 |
+
},
|
| 6980 |
+
{
|
| 6981 |
+
"epoch": 0.15,
|
| 6982 |
+
"learning_rate": 0.00022323440725421855,
|
| 6983 |
+
"loss": 0.4529,
|
| 6984 |
+
"step": 5780
|
| 6985 |
+
},
|
| 6986 |
+
{
|
| 6987 |
+
"epoch": 0.15,
|
| 6988 |
+
"learning_rate": 0.0002232313580984014,
|
| 6989 |
+
"loss": 0.455,
|
| 6990 |
+
"step": 5785
|
| 6991 |
+
},
|
| 6992 |
+
{
|
| 6993 |
+
"epoch": 0.15,
|
| 6994 |
+
"learning_rate": 0.00022322830633376753,
|
| 6995 |
+
"loss": 0.4549,
|
| 6996 |
+
"step": 5790
|
| 6997 |
+
},
|
| 6998 |
+
{
|
| 6999 |
+
"epoch": 0.15,
|
| 7000 |
+
"learning_rate": 0.0002232252519603923,
|
| 7001 |
+
"loss": 0.4518,
|
| 7002 |
+
"step": 5795
|
| 7003 |
+
},
|
| 7004 |
+
{
|
| 7005 |
+
"epoch": 0.16,
|
| 7006 |
+
"learning_rate": 0.000223222194978351,
|
| 7007 |
+
"loss": 0.4502,
|
| 7008 |
+
"step": 5800
|
| 7009 |
+
},
|
| 7010 |
+
{
|
| 7011 |
+
"epoch": 0.16,
|
| 7012 |
+
"learning_rate": 0.00022321913538771913,
|
| 7013 |
+
"loss": 0.453,
|
| 7014 |
+
"step": 5805
|
| 7015 |
+
},
|
| 7016 |
+
{
|
| 7017 |
+
"epoch": 0.16,
|
| 7018 |
+
"learning_rate": 0.00022321607318857219,
|
| 7019 |
+
"loss": 0.4496,
|
| 7020 |
+
"step": 5810
|
| 7021 |
+
},
|
| 7022 |
+
{
|
| 7023 |
+
"epoch": 0.16,
|
| 7024 |
+
"learning_rate": 0.00022321300838098565,
|
| 7025 |
+
"loss": 0.4504,
|
| 7026 |
+
"step": 5815
|
| 7027 |
+
},
|
| 7028 |
+
{
|
| 7029 |
+
"epoch": 0.16,
|
| 7030 |
+
"learning_rate": 0.0002232099409650352,
|
| 7031 |
+
"loss": 0.4535,
|
| 7032 |
+
"step": 5820
|
| 7033 |
+
},
|
| 7034 |
+
{
|
| 7035 |
+
"epoch": 0.16,
|
| 7036 |
+
"learning_rate": 0.00022320687094079652,
|
| 7037 |
+
"loss": 0.4538,
|
| 7038 |
+
"step": 5825
|
| 7039 |
+
},
|
| 7040 |
+
{
|
| 7041 |
+
"epoch": 0.16,
|
| 7042 |
+
"learning_rate": 0.00022320379830834534,
|
| 7043 |
+
"loss": 0.4532,
|
| 7044 |
+
"step": 5830
|
| 7045 |
+
},
|
| 7046 |
+
{
|
| 7047 |
+
"epoch": 0.16,
|
| 7048 |
+
"learning_rate": 0.0002232007230677575,
|
| 7049 |
+
"loss": 0.4515,
|
| 7050 |
+
"step": 5835
|
| 7051 |
+
},
|
| 7052 |
+
{
|
| 7053 |
+
"epoch": 0.16,
|
| 7054 |
+
"learning_rate": 0.00022319764521910887,
|
| 7055 |
+
"loss": 0.4493,
|
| 7056 |
+
"step": 5840
|
| 7057 |
+
},
|
| 7058 |
+
{
|
| 7059 |
+
"epoch": 0.16,
|
| 7060 |
+
"learning_rate": 0.00022319456476247533,
|
| 7061 |
+
"loss": 0.4497,
|
| 7062 |
+
"step": 5845
|
| 7063 |
+
},
|
| 7064 |
+
{
|
| 7065 |
+
"epoch": 0.17,
|
| 7066 |
+
"learning_rate": 0.00022319148169793298,
|
| 7067 |
+
"loss": 0.4539,
|
| 7068 |
+
"step": 5850
|
| 7069 |
+
},
|
| 7070 |
+
{
|
| 7071 |
+
"epoch": 0.17,
|
| 7072 |
+
"learning_rate": 0.00022318839602555787,
|
| 7073 |
+
"loss": 0.4526,
|
| 7074 |
+
"step": 5855
|
| 7075 |
+
},
|
| 7076 |
+
{
|
| 7077 |
+
"epoch": 0.17,
|
| 7078 |
+
"learning_rate": 0.00022318530774542608,
|
| 7079 |
+
"loss": 0.4587,
|
| 7080 |
+
"step": 5860
|
| 7081 |
+
},
|
| 7082 |
+
{
|
| 7083 |
+
"epoch": 0.17,
|
| 7084 |
+
"learning_rate": 0.00022318221685761386,
|
| 7085 |
+
"loss": 0.4609,
|
| 7086 |
+
"step": 5865
|
| 7087 |
+
},
|
| 7088 |
+
{
|
| 7089 |
+
"epoch": 0.17,
|
| 7090 |
+
"learning_rate": 0.00022317912336219746,
|
| 7091 |
+
"loss": 0.454,
|
| 7092 |
+
"step": 5870
|
| 7093 |
+
},
|
| 7094 |
+
{
|
| 7095 |
+
"epoch": 0.17,
|
| 7096 |
+
"learning_rate": 0.0002231760272592532,
|
| 7097 |
+
"loss": 0.4542,
|
| 7098 |
+
"step": 5875
|
| 7099 |
+
},
|
| 7100 |
+
{
|
| 7101 |
+
"epoch": 0.17,
|
| 7102 |
+
"learning_rate": 0.00022317292854885754,
|
| 7103 |
+
"loss": 0.4562,
|
| 7104 |
+
"step": 5880
|
| 7105 |
+
},
|
| 7106 |
+
{
|
| 7107 |
+
"epoch": 0.17,
|
| 7108 |
+
"learning_rate": 0.00022316982723108686,
|
| 7109 |
+
"loss": 0.4527,
|
| 7110 |
+
"step": 5885
|
| 7111 |
+
},
|
| 7112 |
+
{
|
| 7113 |
+
"epoch": 0.17,
|
| 7114 |
+
"learning_rate": 0.00022316672330601768,
|
| 7115 |
+
"loss": 0.4572,
|
| 7116 |
+
"step": 5890
|
| 7117 |
+
},
|
| 7118 |
+
{
|
| 7119 |
+
"epoch": 0.17,
|
| 7120 |
+
"learning_rate": 0.00022316361677372662,
|
| 7121 |
+
"loss": 0.458,
|
| 7122 |
+
"step": 5895
|
| 7123 |
+
},
|
| 7124 |
+
{
|
| 7125 |
+
"epoch": 0.17,
|
| 7126 |
+
"learning_rate": 0.00022316050763429034,
|
| 7127 |
+
"loss": 0.4504,
|
| 7128 |
+
"step": 5900
|
| 7129 |
+
},
|
| 7130 |
+
{
|
| 7131 |
+
"epoch": 0.18,
|
| 7132 |
+
"learning_rate": 0.00022315739588778553,
|
| 7133 |
+
"loss": 0.4466,
|
| 7134 |
+
"step": 5905
|
| 7135 |
+
},
|
| 7136 |
+
{
|
| 7137 |
+
"epoch": 0.18,
|
| 7138 |
+
"learning_rate": 0.00022315428153428898,
|
| 7139 |
+
"loss": 0.4519,
|
| 7140 |
+
"step": 5910
|
| 7141 |
+
},
|
| 7142 |
+
{
|
| 7143 |
+
"epoch": 0.18,
|
| 7144 |
+
"learning_rate": 0.00022315116457387752,
|
| 7145 |
+
"loss": 0.452,
|
| 7146 |
+
"step": 5915
|
| 7147 |
+
},
|
| 7148 |
+
{
|
| 7149 |
+
"epoch": 0.18,
|
| 7150 |
+
"learning_rate": 0.00022314804500662806,
|
| 7151 |
+
"loss": 0.4573,
|
| 7152 |
+
"step": 5920
|
| 7153 |
+
},
|
| 7154 |
+
{
|
| 7155 |
+
"epoch": 0.18,
|
| 7156 |
+
"learning_rate": 0.0002231449228326176,
|
| 7157 |
+
"loss": 0.4556,
|
| 7158 |
+
"step": 5925
|
| 7159 |
+
},
|
| 7160 |
+
{
|
| 7161 |
+
"epoch": 0.18,
|
| 7162 |
+
"learning_rate": 0.00022314179805192313,
|
| 7163 |
+
"loss": 0.4526,
|
| 7164 |
+
"step": 5930
|
| 7165 |
+
},
|
| 7166 |
+
{
|
| 7167 |
+
"epoch": 0.18,
|
| 7168 |
+
"learning_rate": 0.00022313867066462182,
|
| 7169 |
+
"loss": 0.4481,
|
| 7170 |
+
"step": 5935
|
| 7171 |
+
},
|
| 7172 |
+
{
|
| 7173 |
+
"epoch": 0.18,
|
| 7174 |
+
"learning_rate": 0.00022313554067079078,
|
| 7175 |
+
"loss": 0.4559,
|
| 7176 |
+
"step": 5940
|
| 7177 |
+
},
|
| 7178 |
+
{
|
| 7179 |
+
"epoch": 0.18,
|
| 7180 |
+
"learning_rate": 0.00022313240807050723,
|
| 7181 |
+
"loss": 0.4524,
|
| 7182 |
+
"step": 5945
|
| 7183 |
+
},
|
| 7184 |
+
{
|
| 7185 |
+
"epoch": 0.18,
|
| 7186 |
+
"learning_rate": 0.0002231292728638485,
|
| 7187 |
+
"loss": 0.4559,
|
| 7188 |
+
"step": 5950
|
| 7189 |
+
},
|
| 7190 |
+
{
|
| 7191 |
+
"epoch": 0.19,
|
| 7192 |
+
"learning_rate": 0.00022312613505089194,
|
| 7193 |
+
"loss": 0.4572,
|
| 7194 |
+
"step": 5955
|
| 7195 |
+
},
|
| 7196 |
+
{
|
| 7197 |
+
"epoch": 0.19,
|
| 7198 |
+
"learning_rate": 0.00022312299463171499,
|
| 7199 |
+
"loss": 0.4499,
|
| 7200 |
+
"step": 5960
|
| 7201 |
+
},
|
| 7202 |
+
{
|
| 7203 |
+
"epoch": 0.19,
|
| 7204 |
+
"learning_rate": 0.00022311985160639505,
|
| 7205 |
+
"loss": 0.4555,
|
| 7206 |
+
"step": 5965
|
| 7207 |
+
},
|
| 7208 |
+
{
|
| 7209 |
+
"epoch": 0.19,
|
| 7210 |
+
"learning_rate": 0.00022311670597500977,
|
| 7211 |
+
"loss": 0.4541,
|
| 7212 |
+
"step": 5970
|
| 7213 |
+
},
|
| 7214 |
+
{
|
| 7215 |
+
"epoch": 0.19,
|
| 7216 |
+
"learning_rate": 0.00022311355773763674,
|
| 7217 |
+
"loss": 0.4472,
|
| 7218 |
+
"step": 5975
|
| 7219 |
+
},
|
| 7220 |
+
{
|
| 7221 |
+
"epoch": 0.19,
|
| 7222 |
+
"learning_rate": 0.0002231104068943536,
|
| 7223 |
+
"loss": 0.4547,
|
| 7224 |
+
"step": 5980
|
| 7225 |
+
},
|
| 7226 |
+
{
|
| 7227 |
+
"epoch": 0.19,
|
| 7228 |
+
"learning_rate": 0.00022310725344523816,
|
| 7229 |
+
"loss": 0.4562,
|
| 7230 |
+
"step": 5985
|
| 7231 |
+
},
|
| 7232 |
+
{
|
| 7233 |
+
"epoch": 0.19,
|
| 7234 |
+
"learning_rate": 0.00022310409739036816,
|
| 7235 |
+
"loss": 0.4556,
|
| 7236 |
+
"step": 5990
|
| 7237 |
+
},
|
| 7238 |
+
{
|
| 7239 |
+
"epoch": 0.19,
|
| 7240 |
+
"learning_rate": 0.00022310093872982155,
|
| 7241 |
+
"loss": 0.4533,
|
| 7242 |
+
"step": 5995
|
| 7243 |
+
},
|
| 7244 |
+
{
|
| 7245 |
+
"epoch": 0.19,
|
| 7246 |
+
"learning_rate": 0.00022309777746367618,
|
| 7247 |
+
"loss": 0.454,
|
| 7248 |
+
"step": 6000
|
| 7249 |
+
},
|
| 7250 |
+
{
|
| 7251 |
+
"epoch": 0.19,
|
| 7252 |
+
"eval_loss": 0.436722993850708,
|
| 7253 |
+
"eval_runtime": 183.1334,
|
| 7254 |
+
"eval_samples_per_second": 27.303,
|
| 7255 |
+
"eval_steps_per_second": 0.289,
|
| 7256 |
+
"step": 6000
|
| 7257 |
}
|
| 7258 |
],
|
| 7259 |
"max_steps": 100000,
|
| 7260 |
"num_train_epochs": 20,
|
| 7261 |
+
"total_flos": 6.309023572190822e+20,
|
| 7262 |
"trial_name": null,
|
| 7263 |
"trial_params": null
|
| 7264 |
}
|
last-checkpoint/training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5615
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1cad447f2e9f8cc1963ae9cac41183df426b5a7f684286ba0a46dd86d30a7627
|
| 3 |
size 5615
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 449471589
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:023470fb6e5e8c21b4364835994b9c6ce91ee0f9ee1410357ef32e1ecfa1adb1
|
| 3 |
size 449471589
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5615
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1cad447f2e9f8cc1963ae9cac41183df426b5a7f684286ba0a46dd86d30a7627
|
| 3 |
size 5615
|