Training in progress, step 1200
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 sha256:0c6d17006cf5a6e9a9c57de80abad493938bbba39dca6e7012dfe15e101d2c93
|
| 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 sha256:951861143d15fa1c5e1fffde87d6bb38f0764793caa011f89d710cfbb0713166
|
| 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 sha256:4faea06187a773dac9e201feb077e07ac6b09d5a724adf5bebe4d787443e0fc2
|
| 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:be47a0e08dd26ff8097dc5611aa5b8367c04e0c171f4ca560845a27c3c3d8339
|
| 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,
|
|
@@ -6071,11 +6071,1224 @@
|
|
| 6071 |
"eval_samples_per_second": 0.054,
|
| 6072 |
"eval_steps_per_second": 0.054,
|
| 6073 |
"step": 1000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 6074 |
}
|
| 6075 |
],
|
| 6076 |
"max_steps": 2000,
|
| 6077 |
"num_train_epochs": 9223372036854775807,
|
| 6078 |
-
"total_flos":
|
| 6079 |
"trial_name": null,
|
| 6080 |
"trial_params": null
|
| 6081 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 2.126681089401245,
|
| 3 |
+
"best_model_checkpoint": "output/checkpoint-1200",
|
| 4 |
+
"epoch": 0.6,
|
| 5 |
+
"global_step": 1200,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 6071 |
"eval_samples_per_second": 0.054,
|
| 6072 |
"eval_steps_per_second": 0.054,
|
| 6073 |
"step": 1000
|
| 6074 |
+
},
|
| 6075 |
+
{
|
| 6076 |
+
"epoch": 0.5,
|
| 6077 |
+
"learning_rate": 0.00044400000000000006,
|
| 6078 |
+
"loss": 2.2323,
|
| 6079 |
+
"step": 1001
|
| 6080 |
+
},
|
| 6081 |
+
{
|
| 6082 |
+
"epoch": 0.5,
|
| 6083 |
+
"learning_rate": 0.00044355555555555554,
|
| 6084 |
+
"loss": 2.3168,
|
| 6085 |
+
"step": 1002
|
| 6086 |
+
},
|
| 6087 |
+
{
|
| 6088 |
+
"epoch": 0.5,
|
| 6089 |
+
"learning_rate": 0.00044311111111111114,
|
| 6090 |
+
"loss": 2.2113,
|
| 6091 |
+
"step": 1003
|
| 6092 |
+
},
|
| 6093 |
+
{
|
| 6094 |
+
"epoch": 0.5,
|
| 6095 |
+
"learning_rate": 0.0004426666666666667,
|
| 6096 |
+
"loss": 2.2115,
|
| 6097 |
+
"step": 1004
|
| 6098 |
+
},
|
| 6099 |
+
{
|
| 6100 |
+
"epoch": 0.5,
|
| 6101 |
+
"learning_rate": 0.00044222222222222227,
|
| 6102 |
+
"loss": 2.2126,
|
| 6103 |
+
"step": 1005
|
| 6104 |
+
},
|
| 6105 |
+
{
|
| 6106 |
+
"epoch": 0.5,
|
| 6107 |
+
"learning_rate": 0.00044177777777777775,
|
| 6108 |
+
"loss": 2.1807,
|
| 6109 |
+
"step": 1006
|
| 6110 |
+
},
|
| 6111 |
+
{
|
| 6112 |
+
"epoch": 0.5,
|
| 6113 |
+
"learning_rate": 0.00044133333333333335,
|
| 6114 |
+
"loss": 2.2357,
|
| 6115 |
+
"step": 1007
|
| 6116 |
+
},
|
| 6117 |
+
{
|
| 6118 |
+
"epoch": 0.5,
|
| 6119 |
+
"learning_rate": 0.0004408888888888889,
|
| 6120 |
+
"loss": 2.1952,
|
| 6121 |
+
"step": 1008
|
| 6122 |
+
},
|
| 6123 |
+
{
|
| 6124 |
+
"epoch": 0.5,
|
| 6125 |
+
"learning_rate": 0.0004404444444444445,
|
| 6126 |
+
"loss": 2.2606,
|
| 6127 |
+
"step": 1009
|
| 6128 |
+
},
|
| 6129 |
+
{
|
| 6130 |
+
"epoch": 0.51,
|
| 6131 |
+
"learning_rate": 0.00044000000000000007,
|
| 6132 |
+
"loss": 2.1503,
|
| 6133 |
+
"step": 1010
|
| 6134 |
+
},
|
| 6135 |
+
{
|
| 6136 |
+
"epoch": 0.51,
|
| 6137 |
+
"learning_rate": 0.00043955555555555555,
|
| 6138 |
+
"loss": 2.2449,
|
| 6139 |
+
"step": 1011
|
| 6140 |
+
},
|
| 6141 |
+
{
|
| 6142 |
+
"epoch": 0.51,
|
| 6143 |
+
"learning_rate": 0.0004391111111111111,
|
| 6144 |
+
"loss": 2.1983,
|
| 6145 |
+
"step": 1012
|
| 6146 |
+
},
|
| 6147 |
+
{
|
| 6148 |
+
"epoch": 0.51,
|
| 6149 |
+
"learning_rate": 0.0004386666666666667,
|
| 6150 |
+
"loss": 2.1528,
|
| 6151 |
+
"step": 1013
|
| 6152 |
+
},
|
| 6153 |
+
{
|
| 6154 |
+
"epoch": 0.51,
|
| 6155 |
+
"learning_rate": 0.0004382222222222223,
|
| 6156 |
+
"loss": 2.1265,
|
| 6157 |
+
"step": 1014
|
| 6158 |
+
},
|
| 6159 |
+
{
|
| 6160 |
+
"epoch": 0.51,
|
| 6161 |
+
"learning_rate": 0.0004377777777777778,
|
| 6162 |
+
"loss": 2.2405,
|
| 6163 |
+
"step": 1015
|
| 6164 |
+
},
|
| 6165 |
+
{
|
| 6166 |
+
"epoch": 0.51,
|
| 6167 |
+
"learning_rate": 0.0004373333333333333,
|
| 6168 |
+
"loss": 2.2103,
|
| 6169 |
+
"step": 1016
|
| 6170 |
+
},
|
| 6171 |
+
{
|
| 6172 |
+
"epoch": 0.51,
|
| 6173 |
+
"learning_rate": 0.0004368888888888889,
|
| 6174 |
+
"loss": 2.266,
|
| 6175 |
+
"step": 1017
|
| 6176 |
+
},
|
| 6177 |
+
{
|
| 6178 |
+
"epoch": 0.51,
|
| 6179 |
+
"learning_rate": 0.0004364444444444445,
|
| 6180 |
+
"loss": 2.255,
|
| 6181 |
+
"step": 1018
|
| 6182 |
+
},
|
| 6183 |
+
{
|
| 6184 |
+
"epoch": 0.51,
|
| 6185 |
+
"learning_rate": 0.000436,
|
| 6186 |
+
"loss": 2.2256,
|
| 6187 |
+
"step": 1019
|
| 6188 |
+
},
|
| 6189 |
+
{
|
| 6190 |
+
"epoch": 0.51,
|
| 6191 |
+
"learning_rate": 0.00043555555555555557,
|
| 6192 |
+
"loss": 2.2561,
|
| 6193 |
+
"step": 1020
|
| 6194 |
+
},
|
| 6195 |
+
{
|
| 6196 |
+
"epoch": 0.51,
|
| 6197 |
+
"learning_rate": 0.0004351111111111111,
|
| 6198 |
+
"loss": 2.2569,
|
| 6199 |
+
"step": 1021
|
| 6200 |
+
},
|
| 6201 |
+
{
|
| 6202 |
+
"epoch": 0.51,
|
| 6203 |
+
"learning_rate": 0.0004346666666666667,
|
| 6204 |
+
"loss": 2.1995,
|
| 6205 |
+
"step": 1022
|
| 6206 |
+
},
|
| 6207 |
+
{
|
| 6208 |
+
"epoch": 0.51,
|
| 6209 |
+
"learning_rate": 0.0004342222222222223,
|
| 6210 |
+
"loss": 2.193,
|
| 6211 |
+
"step": 1023
|
| 6212 |
+
},
|
| 6213 |
+
{
|
| 6214 |
+
"epoch": 0.51,
|
| 6215 |
+
"learning_rate": 0.00043377777777777783,
|
| 6216 |
+
"loss": 2.1382,
|
| 6217 |
+
"step": 1024
|
| 6218 |
+
},
|
| 6219 |
+
{
|
| 6220 |
+
"epoch": 0.51,
|
| 6221 |
+
"learning_rate": 0.0004333333333333333,
|
| 6222 |
+
"loss": 2.2134,
|
| 6223 |
+
"step": 1025
|
| 6224 |
+
},
|
| 6225 |
+
{
|
| 6226 |
+
"epoch": 0.51,
|
| 6227 |
+
"learning_rate": 0.0004328888888888889,
|
| 6228 |
+
"loss": 2.173,
|
| 6229 |
+
"step": 1026
|
| 6230 |
+
},
|
| 6231 |
+
{
|
| 6232 |
+
"epoch": 0.51,
|
| 6233 |
+
"learning_rate": 0.0004324444444444445,
|
| 6234 |
+
"loss": 2.2619,
|
| 6235 |
+
"step": 1027
|
| 6236 |
+
},
|
| 6237 |
+
{
|
| 6238 |
+
"epoch": 0.51,
|
| 6239 |
+
"learning_rate": 0.00043200000000000004,
|
| 6240 |
+
"loss": 2.1414,
|
| 6241 |
+
"step": 1028
|
| 6242 |
+
},
|
| 6243 |
+
{
|
| 6244 |
+
"epoch": 0.51,
|
| 6245 |
+
"learning_rate": 0.0004315555555555555,
|
| 6246 |
+
"loss": 2.1653,
|
| 6247 |
+
"step": 1029
|
| 6248 |
+
},
|
| 6249 |
+
{
|
| 6250 |
+
"epoch": 0.52,
|
| 6251 |
+
"learning_rate": 0.0004311111111111111,
|
| 6252 |
+
"loss": 2.2585,
|
| 6253 |
+
"step": 1030
|
| 6254 |
+
},
|
| 6255 |
+
{
|
| 6256 |
+
"epoch": 0.52,
|
| 6257 |
+
"learning_rate": 0.0004306666666666667,
|
| 6258 |
+
"loss": 2.1986,
|
| 6259 |
+
"step": 1031
|
| 6260 |
+
},
|
| 6261 |
+
{
|
| 6262 |
+
"epoch": 0.52,
|
| 6263 |
+
"learning_rate": 0.00043022222222222225,
|
| 6264 |
+
"loss": 2.1942,
|
| 6265 |
+
"step": 1032
|
| 6266 |
+
},
|
| 6267 |
+
{
|
| 6268 |
+
"epoch": 0.52,
|
| 6269 |
+
"learning_rate": 0.00042977777777777784,
|
| 6270 |
+
"loss": 2.2071,
|
| 6271 |
+
"step": 1033
|
| 6272 |
+
},
|
| 6273 |
+
{
|
| 6274 |
+
"epoch": 0.52,
|
| 6275 |
+
"learning_rate": 0.0004293333333333333,
|
| 6276 |
+
"loss": 2.2311,
|
| 6277 |
+
"step": 1034
|
| 6278 |
+
},
|
| 6279 |
+
{
|
| 6280 |
+
"epoch": 0.52,
|
| 6281 |
+
"learning_rate": 0.0004288888888888889,
|
| 6282 |
+
"loss": 2.2495,
|
| 6283 |
+
"step": 1035
|
| 6284 |
+
},
|
| 6285 |
+
{
|
| 6286 |
+
"epoch": 0.52,
|
| 6287 |
+
"learning_rate": 0.00042844444444444446,
|
| 6288 |
+
"loss": 2.1843,
|
| 6289 |
+
"step": 1036
|
| 6290 |
+
},
|
| 6291 |
+
{
|
| 6292 |
+
"epoch": 0.52,
|
| 6293 |
+
"learning_rate": 0.00042800000000000005,
|
| 6294 |
+
"loss": 2.2238,
|
| 6295 |
+
"step": 1037
|
| 6296 |
+
},
|
| 6297 |
+
{
|
| 6298 |
+
"epoch": 0.52,
|
| 6299 |
+
"learning_rate": 0.00042755555555555564,
|
| 6300 |
+
"loss": 2.2338,
|
| 6301 |
+
"step": 1038
|
| 6302 |
+
},
|
| 6303 |
+
{
|
| 6304 |
+
"epoch": 0.52,
|
| 6305 |
+
"learning_rate": 0.0004271111111111111,
|
| 6306 |
+
"loss": 2.2446,
|
| 6307 |
+
"step": 1039
|
| 6308 |
+
},
|
| 6309 |
+
{
|
| 6310 |
+
"epoch": 0.52,
|
| 6311 |
+
"learning_rate": 0.00042666666666666667,
|
| 6312 |
+
"loss": 2.1949,
|
| 6313 |
+
"step": 1040
|
| 6314 |
+
},
|
| 6315 |
+
{
|
| 6316 |
+
"epoch": 0.52,
|
| 6317 |
+
"learning_rate": 0.00042622222222222226,
|
| 6318 |
+
"loss": 2.2357,
|
| 6319 |
+
"step": 1041
|
| 6320 |
+
},
|
| 6321 |
+
{
|
| 6322 |
+
"epoch": 0.52,
|
| 6323 |
+
"learning_rate": 0.00042577777777777785,
|
| 6324 |
+
"loss": 2.1674,
|
| 6325 |
+
"step": 1042
|
| 6326 |
+
},
|
| 6327 |
+
{
|
| 6328 |
+
"epoch": 0.52,
|
| 6329 |
+
"learning_rate": 0.00042533333333333334,
|
| 6330 |
+
"loss": 2.2584,
|
| 6331 |
+
"step": 1043
|
| 6332 |
+
},
|
| 6333 |
+
{
|
| 6334 |
+
"epoch": 0.52,
|
| 6335 |
+
"learning_rate": 0.0004248888888888889,
|
| 6336 |
+
"loss": 2.2447,
|
| 6337 |
+
"step": 1044
|
| 6338 |
+
},
|
| 6339 |
+
{
|
| 6340 |
+
"epoch": 0.52,
|
| 6341 |
+
"learning_rate": 0.00042444444444444447,
|
| 6342 |
+
"loss": 2.1795,
|
| 6343 |
+
"step": 1045
|
| 6344 |
+
},
|
| 6345 |
+
{
|
| 6346 |
+
"epoch": 0.52,
|
| 6347 |
+
"learning_rate": 0.00042400000000000006,
|
| 6348 |
+
"loss": 2.2055,
|
| 6349 |
+
"step": 1046
|
| 6350 |
+
},
|
| 6351 |
+
{
|
| 6352 |
+
"epoch": 0.52,
|
| 6353 |
+
"learning_rate": 0.0004235555555555556,
|
| 6354 |
+
"loss": 2.2359,
|
| 6355 |
+
"step": 1047
|
| 6356 |
+
},
|
| 6357 |
+
{
|
| 6358 |
+
"epoch": 0.52,
|
| 6359 |
+
"learning_rate": 0.0004231111111111111,
|
| 6360 |
+
"loss": 2.1674,
|
| 6361 |
+
"step": 1048
|
| 6362 |
+
},
|
| 6363 |
+
{
|
| 6364 |
+
"epoch": 0.52,
|
| 6365 |
+
"learning_rate": 0.0004226666666666667,
|
| 6366 |
+
"loss": 2.2057,
|
| 6367 |
+
"step": 1049
|
| 6368 |
+
},
|
| 6369 |
+
{
|
| 6370 |
+
"epoch": 0.53,
|
| 6371 |
+
"learning_rate": 0.00042222222222222227,
|
| 6372 |
+
"loss": 2.2296,
|
| 6373 |
+
"step": 1050
|
| 6374 |
+
},
|
| 6375 |
+
{
|
| 6376 |
+
"epoch": 0.53,
|
| 6377 |
+
"learning_rate": 0.0004217777777777778,
|
| 6378 |
+
"loss": 2.2603,
|
| 6379 |
+
"step": 1051
|
| 6380 |
+
},
|
| 6381 |
+
{
|
| 6382 |
+
"epoch": 0.53,
|
| 6383 |
+
"learning_rate": 0.0004213333333333333,
|
| 6384 |
+
"loss": 2.3112,
|
| 6385 |
+
"step": 1052
|
| 6386 |
+
},
|
| 6387 |
+
{
|
| 6388 |
+
"epoch": 0.53,
|
| 6389 |
+
"learning_rate": 0.0004208888888888889,
|
| 6390 |
+
"loss": 2.317,
|
| 6391 |
+
"step": 1053
|
| 6392 |
+
},
|
| 6393 |
+
{
|
| 6394 |
+
"epoch": 0.53,
|
| 6395 |
+
"learning_rate": 0.0004204444444444445,
|
| 6396 |
+
"loss": 2.226,
|
| 6397 |
+
"step": 1054
|
| 6398 |
+
},
|
| 6399 |
+
{
|
| 6400 |
+
"epoch": 0.53,
|
| 6401 |
+
"learning_rate": 0.00042,
|
| 6402 |
+
"loss": 2.2722,
|
| 6403 |
+
"step": 1055
|
| 6404 |
+
},
|
| 6405 |
+
{
|
| 6406 |
+
"epoch": 0.53,
|
| 6407 |
+
"learning_rate": 0.0004195555555555556,
|
| 6408 |
+
"loss": 2.2752,
|
| 6409 |
+
"step": 1056
|
| 6410 |
+
},
|
| 6411 |
+
{
|
| 6412 |
+
"epoch": 0.53,
|
| 6413 |
+
"learning_rate": 0.0004191111111111111,
|
| 6414 |
+
"loss": 2.2753,
|
| 6415 |
+
"step": 1057
|
| 6416 |
+
},
|
| 6417 |
+
{
|
| 6418 |
+
"epoch": 0.53,
|
| 6419 |
+
"learning_rate": 0.0004186666666666667,
|
| 6420 |
+
"loss": 2.2863,
|
| 6421 |
+
"step": 1058
|
| 6422 |
+
},
|
| 6423 |
+
{
|
| 6424 |
+
"epoch": 0.53,
|
| 6425 |
+
"learning_rate": 0.0004182222222222222,
|
| 6426 |
+
"loss": 2.2846,
|
| 6427 |
+
"step": 1059
|
| 6428 |
+
},
|
| 6429 |
+
{
|
| 6430 |
+
"epoch": 0.53,
|
| 6431 |
+
"learning_rate": 0.0004177777777777778,
|
| 6432 |
+
"loss": 2.2366,
|
| 6433 |
+
"step": 1060
|
| 6434 |
+
},
|
| 6435 |
+
{
|
| 6436 |
+
"epoch": 0.53,
|
| 6437 |
+
"learning_rate": 0.0004173333333333333,
|
| 6438 |
+
"loss": 2.2447,
|
| 6439 |
+
"step": 1061
|
| 6440 |
+
},
|
| 6441 |
+
{
|
| 6442 |
+
"epoch": 0.53,
|
| 6443 |
+
"learning_rate": 0.0004168888888888889,
|
| 6444 |
+
"loss": 2.3256,
|
| 6445 |
+
"step": 1062
|
| 6446 |
+
},
|
| 6447 |
+
{
|
| 6448 |
+
"epoch": 0.53,
|
| 6449 |
+
"learning_rate": 0.00041644444444444444,
|
| 6450 |
+
"loss": 2.2203,
|
| 6451 |
+
"step": 1063
|
| 6452 |
+
},
|
| 6453 |
+
{
|
| 6454 |
+
"epoch": 0.53,
|
| 6455 |
+
"learning_rate": 0.00041600000000000003,
|
| 6456 |
+
"loss": 2.1463,
|
| 6457 |
+
"step": 1064
|
| 6458 |
+
},
|
| 6459 |
+
{
|
| 6460 |
+
"epoch": 0.53,
|
| 6461 |
+
"learning_rate": 0.0004155555555555556,
|
| 6462 |
+
"loss": 2.2303,
|
| 6463 |
+
"step": 1065
|
| 6464 |
+
},
|
| 6465 |
+
{
|
| 6466 |
+
"epoch": 0.53,
|
| 6467 |
+
"learning_rate": 0.0004151111111111111,
|
| 6468 |
+
"loss": 2.1686,
|
| 6469 |
+
"step": 1066
|
| 6470 |
+
},
|
| 6471 |
+
{
|
| 6472 |
+
"epoch": 0.53,
|
| 6473 |
+
"learning_rate": 0.00041466666666666664,
|
| 6474 |
+
"loss": 2.2296,
|
| 6475 |
+
"step": 1067
|
| 6476 |
+
},
|
| 6477 |
+
{
|
| 6478 |
+
"epoch": 0.53,
|
| 6479 |
+
"learning_rate": 0.00041422222222222224,
|
| 6480 |
+
"loss": 2.2437,
|
| 6481 |
+
"step": 1068
|
| 6482 |
+
},
|
| 6483 |
+
{
|
| 6484 |
+
"epoch": 0.53,
|
| 6485 |
+
"learning_rate": 0.00041377777777777783,
|
| 6486 |
+
"loss": 2.2731,
|
| 6487 |
+
"step": 1069
|
| 6488 |
+
},
|
| 6489 |
+
{
|
| 6490 |
+
"epoch": 0.54,
|
| 6491 |
+
"learning_rate": 0.00041333333333333337,
|
| 6492 |
+
"loss": 2.249,
|
| 6493 |
+
"step": 1070
|
| 6494 |
+
},
|
| 6495 |
+
{
|
| 6496 |
+
"epoch": 0.54,
|
| 6497 |
+
"learning_rate": 0.00041288888888888885,
|
| 6498 |
+
"loss": 2.2399,
|
| 6499 |
+
"step": 1071
|
| 6500 |
+
},
|
| 6501 |
+
{
|
| 6502 |
+
"epoch": 0.54,
|
| 6503 |
+
"learning_rate": 0.00041244444444444445,
|
| 6504 |
+
"loss": 2.1833,
|
| 6505 |
+
"step": 1072
|
| 6506 |
+
},
|
| 6507 |
+
{
|
| 6508 |
+
"epoch": 0.54,
|
| 6509 |
+
"learning_rate": 0.00041200000000000004,
|
| 6510 |
+
"loss": 2.18,
|
| 6511 |
+
"step": 1073
|
| 6512 |
+
},
|
| 6513 |
+
{
|
| 6514 |
+
"epoch": 0.54,
|
| 6515 |
+
"learning_rate": 0.0004115555555555556,
|
| 6516 |
+
"loss": 2.1966,
|
| 6517 |
+
"step": 1074
|
| 6518 |
+
},
|
| 6519 |
+
{
|
| 6520 |
+
"epoch": 0.54,
|
| 6521 |
+
"learning_rate": 0.0004111111111111111,
|
| 6522 |
+
"loss": 2.1802,
|
| 6523 |
+
"step": 1075
|
| 6524 |
+
},
|
| 6525 |
+
{
|
| 6526 |
+
"epoch": 0.54,
|
| 6527 |
+
"learning_rate": 0.00041066666666666666,
|
| 6528 |
+
"loss": 2.2319,
|
| 6529 |
+
"step": 1076
|
| 6530 |
+
},
|
| 6531 |
+
{
|
| 6532 |
+
"epoch": 0.54,
|
| 6533 |
+
"learning_rate": 0.00041022222222222225,
|
| 6534 |
+
"loss": 2.2461,
|
| 6535 |
+
"step": 1077
|
| 6536 |
+
},
|
| 6537 |
+
{
|
| 6538 |
+
"epoch": 0.54,
|
| 6539 |
+
"learning_rate": 0.00040977777777777784,
|
| 6540 |
+
"loss": 2.2113,
|
| 6541 |
+
"step": 1078
|
| 6542 |
+
},
|
| 6543 |
+
{
|
| 6544 |
+
"epoch": 0.54,
|
| 6545 |
+
"learning_rate": 0.0004093333333333334,
|
| 6546 |
+
"loss": 2.2787,
|
| 6547 |
+
"step": 1079
|
| 6548 |
+
},
|
| 6549 |
+
{
|
| 6550 |
+
"epoch": 0.54,
|
| 6551 |
+
"learning_rate": 0.00040888888888888887,
|
| 6552 |
+
"loss": 2.1956,
|
| 6553 |
+
"step": 1080
|
| 6554 |
+
},
|
| 6555 |
+
{
|
| 6556 |
+
"epoch": 0.54,
|
| 6557 |
+
"learning_rate": 0.00040844444444444446,
|
| 6558 |
+
"loss": 2.2832,
|
| 6559 |
+
"step": 1081
|
| 6560 |
+
},
|
| 6561 |
+
{
|
| 6562 |
+
"epoch": 0.54,
|
| 6563 |
+
"learning_rate": 0.00040800000000000005,
|
| 6564 |
+
"loss": 2.2853,
|
| 6565 |
+
"step": 1082
|
| 6566 |
+
},
|
| 6567 |
+
{
|
| 6568 |
+
"epoch": 0.54,
|
| 6569 |
+
"learning_rate": 0.0004075555555555556,
|
| 6570 |
+
"loss": 2.1744,
|
| 6571 |
+
"step": 1083
|
| 6572 |
+
},
|
| 6573 |
+
{
|
| 6574 |
+
"epoch": 0.54,
|
| 6575 |
+
"learning_rate": 0.0004071111111111111,
|
| 6576 |
+
"loss": 2.214,
|
| 6577 |
+
"step": 1084
|
| 6578 |
+
},
|
| 6579 |
+
{
|
| 6580 |
+
"epoch": 0.54,
|
| 6581 |
+
"learning_rate": 0.00040666666666666667,
|
| 6582 |
+
"loss": 2.2348,
|
| 6583 |
+
"step": 1085
|
| 6584 |
+
},
|
| 6585 |
+
{
|
| 6586 |
+
"epoch": 0.54,
|
| 6587 |
+
"learning_rate": 0.00040622222222222226,
|
| 6588 |
+
"loss": 2.1959,
|
| 6589 |
+
"step": 1086
|
| 6590 |
+
},
|
| 6591 |
+
{
|
| 6592 |
+
"epoch": 0.54,
|
| 6593 |
+
"learning_rate": 0.0004057777777777778,
|
| 6594 |
+
"loss": 2.2039,
|
| 6595 |
+
"step": 1087
|
| 6596 |
+
},
|
| 6597 |
+
{
|
| 6598 |
+
"epoch": 0.54,
|
| 6599 |
+
"learning_rate": 0.0004053333333333334,
|
| 6600 |
+
"loss": 2.1746,
|
| 6601 |
+
"step": 1088
|
| 6602 |
+
},
|
| 6603 |
+
{
|
| 6604 |
+
"epoch": 0.54,
|
| 6605 |
+
"learning_rate": 0.0004048888888888889,
|
| 6606 |
+
"loss": 2.2353,
|
| 6607 |
+
"step": 1089
|
| 6608 |
+
},
|
| 6609 |
+
{
|
| 6610 |
+
"epoch": 0.55,
|
| 6611 |
+
"learning_rate": 0.00040444444444444447,
|
| 6612 |
+
"loss": 2.238,
|
| 6613 |
+
"step": 1090
|
| 6614 |
+
},
|
| 6615 |
+
{
|
| 6616 |
+
"epoch": 0.55,
|
| 6617 |
+
"learning_rate": 0.000404,
|
| 6618 |
+
"loss": 2.254,
|
| 6619 |
+
"step": 1091
|
| 6620 |
+
},
|
| 6621 |
+
{
|
| 6622 |
+
"epoch": 0.55,
|
| 6623 |
+
"learning_rate": 0.0004035555555555556,
|
| 6624 |
+
"loss": 2.214,
|
| 6625 |
+
"step": 1092
|
| 6626 |
+
},
|
| 6627 |
+
{
|
| 6628 |
+
"epoch": 0.55,
|
| 6629 |
+
"learning_rate": 0.0004031111111111112,
|
| 6630 |
+
"loss": 2.2567,
|
| 6631 |
+
"step": 1093
|
| 6632 |
+
},
|
| 6633 |
+
{
|
| 6634 |
+
"epoch": 0.55,
|
| 6635 |
+
"learning_rate": 0.0004026666666666667,
|
| 6636 |
+
"loss": 2.1921,
|
| 6637 |
+
"step": 1094
|
| 6638 |
+
},
|
| 6639 |
+
{
|
| 6640 |
+
"epoch": 0.55,
|
| 6641 |
+
"learning_rate": 0.0004022222222222222,
|
| 6642 |
+
"loss": 2.2253,
|
| 6643 |
+
"step": 1095
|
| 6644 |
+
},
|
| 6645 |
+
{
|
| 6646 |
+
"epoch": 0.55,
|
| 6647 |
+
"learning_rate": 0.0004017777777777778,
|
| 6648 |
+
"loss": 2.2083,
|
| 6649 |
+
"step": 1096
|
| 6650 |
+
},
|
| 6651 |
+
{
|
| 6652 |
+
"epoch": 0.55,
|
| 6653 |
+
"learning_rate": 0.0004013333333333334,
|
| 6654 |
+
"loss": 2.2661,
|
| 6655 |
+
"step": 1097
|
| 6656 |
+
},
|
| 6657 |
+
{
|
| 6658 |
+
"epoch": 0.55,
|
| 6659 |
+
"learning_rate": 0.0004008888888888889,
|
| 6660 |
+
"loss": 2.2081,
|
| 6661 |
+
"step": 1098
|
| 6662 |
+
},
|
| 6663 |
+
{
|
| 6664 |
+
"epoch": 0.55,
|
| 6665 |
+
"learning_rate": 0.0004004444444444444,
|
| 6666 |
+
"loss": 2.105,
|
| 6667 |
+
"step": 1099
|
| 6668 |
+
},
|
| 6669 |
+
{
|
| 6670 |
+
"epoch": 0.55,
|
| 6671 |
+
"learning_rate": 0.0004,
|
| 6672 |
+
"loss": 2.2621,
|
| 6673 |
+
"step": 1100
|
| 6674 |
+
},
|
| 6675 |
+
{
|
| 6676 |
+
"epoch": 0.55,
|
| 6677 |
+
"learning_rate": 0.0003995555555555556,
|
| 6678 |
+
"loss": 2.2716,
|
| 6679 |
+
"step": 1101
|
| 6680 |
+
},
|
| 6681 |
+
{
|
| 6682 |
+
"epoch": 0.55,
|
| 6683 |
+
"learning_rate": 0.00039911111111111115,
|
| 6684 |
+
"loss": 2.2039,
|
| 6685 |
+
"step": 1102
|
| 6686 |
+
},
|
| 6687 |
+
{
|
| 6688 |
+
"epoch": 0.55,
|
| 6689 |
+
"learning_rate": 0.0003986666666666667,
|
| 6690 |
+
"loss": 2.2143,
|
| 6691 |
+
"step": 1103
|
| 6692 |
+
},
|
| 6693 |
+
{
|
| 6694 |
+
"epoch": 0.55,
|
| 6695 |
+
"learning_rate": 0.00039822222222222223,
|
| 6696 |
+
"loss": 2.3052,
|
| 6697 |
+
"step": 1104
|
| 6698 |
+
},
|
| 6699 |
+
{
|
| 6700 |
+
"epoch": 0.55,
|
| 6701 |
+
"learning_rate": 0.0003977777777777778,
|
| 6702 |
+
"loss": 2.2629,
|
| 6703 |
+
"step": 1105
|
| 6704 |
+
},
|
| 6705 |
+
{
|
| 6706 |
+
"epoch": 0.55,
|
| 6707 |
+
"learning_rate": 0.00039733333333333336,
|
| 6708 |
+
"loss": 2.2976,
|
| 6709 |
+
"step": 1106
|
| 6710 |
+
},
|
| 6711 |
+
{
|
| 6712 |
+
"epoch": 0.55,
|
| 6713 |
+
"learning_rate": 0.0003968888888888889,
|
| 6714 |
+
"loss": 2.2596,
|
| 6715 |
+
"step": 1107
|
| 6716 |
+
},
|
| 6717 |
+
{
|
| 6718 |
+
"epoch": 0.55,
|
| 6719 |
+
"learning_rate": 0.0003964444444444445,
|
| 6720 |
+
"loss": 2.2815,
|
| 6721 |
+
"step": 1108
|
| 6722 |
+
},
|
| 6723 |
+
{
|
| 6724 |
+
"epoch": 0.55,
|
| 6725 |
+
"learning_rate": 0.00039600000000000003,
|
| 6726 |
+
"loss": 2.2405,
|
| 6727 |
+
"step": 1109
|
| 6728 |
+
},
|
| 6729 |
+
{
|
| 6730 |
+
"epoch": 0.56,
|
| 6731 |
+
"learning_rate": 0.00039555555555555557,
|
| 6732 |
+
"loss": 2.2578,
|
| 6733 |
+
"step": 1110
|
| 6734 |
+
},
|
| 6735 |
+
{
|
| 6736 |
+
"epoch": 0.56,
|
| 6737 |
+
"learning_rate": 0.0003951111111111111,
|
| 6738 |
+
"loss": 2.1971,
|
| 6739 |
+
"step": 1111
|
| 6740 |
+
},
|
| 6741 |
+
{
|
| 6742 |
+
"epoch": 0.56,
|
| 6743 |
+
"learning_rate": 0.0003946666666666667,
|
| 6744 |
+
"loss": 2.2151,
|
| 6745 |
+
"step": 1112
|
| 6746 |
+
},
|
| 6747 |
+
{
|
| 6748 |
+
"epoch": 0.56,
|
| 6749 |
+
"learning_rate": 0.00039422222222222224,
|
| 6750 |
+
"loss": 2.2425,
|
| 6751 |
+
"step": 1113
|
| 6752 |
+
},
|
| 6753 |
+
{
|
| 6754 |
+
"epoch": 0.56,
|
| 6755 |
+
"learning_rate": 0.0003937777777777778,
|
| 6756 |
+
"loss": 2.2328,
|
| 6757 |
+
"step": 1114
|
| 6758 |
+
},
|
| 6759 |
+
{
|
| 6760 |
+
"epoch": 0.56,
|
| 6761 |
+
"learning_rate": 0.0003933333333333333,
|
| 6762 |
+
"loss": 2.2582,
|
| 6763 |
+
"step": 1115
|
| 6764 |
+
},
|
| 6765 |
+
{
|
| 6766 |
+
"epoch": 0.56,
|
| 6767 |
+
"learning_rate": 0.0003928888888888889,
|
| 6768 |
+
"loss": 2.1567,
|
| 6769 |
+
"step": 1116
|
| 6770 |
+
},
|
| 6771 |
+
{
|
| 6772 |
+
"epoch": 0.56,
|
| 6773 |
+
"learning_rate": 0.0003924444444444445,
|
| 6774 |
+
"loss": 2.2728,
|
| 6775 |
+
"step": 1117
|
| 6776 |
+
},
|
| 6777 |
+
{
|
| 6778 |
+
"epoch": 0.56,
|
| 6779 |
+
"learning_rate": 0.000392,
|
| 6780 |
+
"loss": 2.1876,
|
| 6781 |
+
"step": 1118
|
| 6782 |
+
},
|
| 6783 |
+
{
|
| 6784 |
+
"epoch": 0.56,
|
| 6785 |
+
"learning_rate": 0.0003915555555555556,
|
| 6786 |
+
"loss": 2.2007,
|
| 6787 |
+
"step": 1119
|
| 6788 |
+
},
|
| 6789 |
+
{
|
| 6790 |
+
"epoch": 0.56,
|
| 6791 |
+
"learning_rate": 0.0003911111111111111,
|
| 6792 |
+
"loss": 2.2632,
|
| 6793 |
+
"step": 1120
|
| 6794 |
+
},
|
| 6795 |
+
{
|
| 6796 |
+
"epoch": 0.56,
|
| 6797 |
+
"learning_rate": 0.0003906666666666667,
|
| 6798 |
+
"loss": 2.1838,
|
| 6799 |
+
"step": 1121
|
| 6800 |
+
},
|
| 6801 |
+
{
|
| 6802 |
+
"epoch": 0.56,
|
| 6803 |
+
"learning_rate": 0.0003902222222222222,
|
| 6804 |
+
"loss": 2.1546,
|
| 6805 |
+
"step": 1122
|
| 6806 |
+
},
|
| 6807 |
+
{
|
| 6808 |
+
"epoch": 0.56,
|
| 6809 |
+
"learning_rate": 0.0003897777777777778,
|
| 6810 |
+
"loss": 2.2672,
|
| 6811 |
+
"step": 1123
|
| 6812 |
+
},
|
| 6813 |
+
{
|
| 6814 |
+
"epoch": 0.56,
|
| 6815 |
+
"learning_rate": 0.0003893333333333334,
|
| 6816 |
+
"loss": 2.2595,
|
| 6817 |
+
"step": 1124
|
| 6818 |
+
},
|
| 6819 |
+
{
|
| 6820 |
+
"epoch": 0.56,
|
| 6821 |
+
"learning_rate": 0.0003888888888888889,
|
| 6822 |
+
"loss": 2.1636,
|
| 6823 |
+
"step": 1125
|
| 6824 |
+
},
|
| 6825 |
+
{
|
| 6826 |
+
"epoch": 0.56,
|
| 6827 |
+
"learning_rate": 0.00038844444444444446,
|
| 6828 |
+
"loss": 2.2421,
|
| 6829 |
+
"step": 1126
|
| 6830 |
+
},
|
| 6831 |
+
{
|
| 6832 |
+
"epoch": 0.56,
|
| 6833 |
+
"learning_rate": 0.000388,
|
| 6834 |
+
"loss": 2.2063,
|
| 6835 |
+
"step": 1127
|
| 6836 |
+
},
|
| 6837 |
+
{
|
| 6838 |
+
"epoch": 0.56,
|
| 6839 |
+
"learning_rate": 0.0003875555555555556,
|
| 6840 |
+
"loss": 2.3393,
|
| 6841 |
+
"step": 1128
|
| 6842 |
+
},
|
| 6843 |
+
{
|
| 6844 |
+
"epoch": 0.56,
|
| 6845 |
+
"learning_rate": 0.00038711111111111113,
|
| 6846 |
+
"loss": 2.2483,
|
| 6847 |
+
"step": 1129
|
| 6848 |
+
},
|
| 6849 |
+
{
|
| 6850 |
+
"epoch": 0.56,
|
| 6851 |
+
"learning_rate": 0.00038666666666666667,
|
| 6852 |
+
"loss": 2.2506,
|
| 6853 |
+
"step": 1130
|
| 6854 |
+
},
|
| 6855 |
+
{
|
| 6856 |
+
"epoch": 0.57,
|
| 6857 |
+
"learning_rate": 0.0003862222222222222,
|
| 6858 |
+
"loss": 2.2516,
|
| 6859 |
+
"step": 1131
|
| 6860 |
+
},
|
| 6861 |
+
{
|
| 6862 |
+
"epoch": 0.57,
|
| 6863 |
+
"learning_rate": 0.0003857777777777778,
|
| 6864 |
+
"loss": 2.2543,
|
| 6865 |
+
"step": 1132
|
| 6866 |
+
},
|
| 6867 |
+
{
|
| 6868 |
+
"epoch": 0.57,
|
| 6869 |
+
"learning_rate": 0.0003853333333333334,
|
| 6870 |
+
"loss": 2.2858,
|
| 6871 |
+
"step": 1133
|
| 6872 |
+
},
|
| 6873 |
+
{
|
| 6874 |
+
"epoch": 0.57,
|
| 6875 |
+
"learning_rate": 0.0003848888888888889,
|
| 6876 |
+
"loss": 2.1998,
|
| 6877 |
+
"step": 1134
|
| 6878 |
+
},
|
| 6879 |
+
{
|
| 6880 |
+
"epoch": 0.57,
|
| 6881 |
+
"learning_rate": 0.00038444444444444447,
|
| 6882 |
+
"loss": 2.2668,
|
| 6883 |
+
"step": 1135
|
| 6884 |
+
},
|
| 6885 |
+
{
|
| 6886 |
+
"epoch": 0.57,
|
| 6887 |
+
"learning_rate": 0.000384,
|
| 6888 |
+
"loss": 2.2263,
|
| 6889 |
+
"step": 1136
|
| 6890 |
+
},
|
| 6891 |
+
{
|
| 6892 |
+
"epoch": 0.57,
|
| 6893 |
+
"learning_rate": 0.0003835555555555556,
|
| 6894 |
+
"loss": 2.2531,
|
| 6895 |
+
"step": 1137
|
| 6896 |
+
},
|
| 6897 |
+
{
|
| 6898 |
+
"epoch": 0.57,
|
| 6899 |
+
"learning_rate": 0.0003831111111111111,
|
| 6900 |
+
"loss": 2.2176,
|
| 6901 |
+
"step": 1138
|
| 6902 |
+
},
|
| 6903 |
+
{
|
| 6904 |
+
"epoch": 0.57,
|
| 6905 |
+
"learning_rate": 0.0003826666666666667,
|
| 6906 |
+
"loss": 2.2106,
|
| 6907 |
+
"step": 1139
|
| 6908 |
+
},
|
| 6909 |
+
{
|
| 6910 |
+
"epoch": 0.57,
|
| 6911 |
+
"learning_rate": 0.0003822222222222223,
|
| 6912 |
+
"loss": 2.2093,
|
| 6913 |
+
"step": 1140
|
| 6914 |
+
},
|
| 6915 |
+
{
|
| 6916 |
+
"epoch": 0.57,
|
| 6917 |
+
"learning_rate": 0.0003817777777777778,
|
| 6918 |
+
"loss": 2.2806,
|
| 6919 |
+
"step": 1141
|
| 6920 |
+
},
|
| 6921 |
+
{
|
| 6922 |
+
"epoch": 0.57,
|
| 6923 |
+
"learning_rate": 0.00038133333333333335,
|
| 6924 |
+
"loss": 2.2486,
|
| 6925 |
+
"step": 1142
|
| 6926 |
+
},
|
| 6927 |
+
{
|
| 6928 |
+
"epoch": 0.57,
|
| 6929 |
+
"learning_rate": 0.0003808888888888889,
|
| 6930 |
+
"loss": 2.1462,
|
| 6931 |
+
"step": 1143
|
| 6932 |
+
},
|
| 6933 |
+
{
|
| 6934 |
+
"epoch": 0.57,
|
| 6935 |
+
"learning_rate": 0.0003804444444444445,
|
| 6936 |
+
"loss": 2.1784,
|
| 6937 |
+
"step": 1144
|
| 6938 |
+
},
|
| 6939 |
+
{
|
| 6940 |
+
"epoch": 0.57,
|
| 6941 |
+
"learning_rate": 0.00038,
|
| 6942 |
+
"loss": 2.2601,
|
| 6943 |
+
"step": 1145
|
| 6944 |
+
},
|
| 6945 |
+
{
|
| 6946 |
+
"epoch": 0.57,
|
| 6947 |
+
"learning_rate": 0.00037955555555555556,
|
| 6948 |
+
"loss": 2.1816,
|
| 6949 |
+
"step": 1146
|
| 6950 |
+
},
|
| 6951 |
+
{
|
| 6952 |
+
"epoch": 0.57,
|
| 6953 |
+
"learning_rate": 0.00037911111111111115,
|
| 6954 |
+
"loss": 2.2372,
|
| 6955 |
+
"step": 1147
|
| 6956 |
+
},
|
| 6957 |
+
{
|
| 6958 |
+
"epoch": 0.57,
|
| 6959 |
+
"learning_rate": 0.0003786666666666667,
|
| 6960 |
+
"loss": 2.1934,
|
| 6961 |
+
"step": 1148
|
| 6962 |
+
},
|
| 6963 |
+
{
|
| 6964 |
+
"epoch": 0.57,
|
| 6965 |
+
"learning_rate": 0.00037822222222222223,
|
| 6966 |
+
"loss": 2.151,
|
| 6967 |
+
"step": 1149
|
| 6968 |
+
},
|
| 6969 |
+
{
|
| 6970 |
+
"epoch": 0.57,
|
| 6971 |
+
"learning_rate": 0.00037777777777777777,
|
| 6972 |
+
"loss": 2.2499,
|
| 6973 |
+
"step": 1150
|
| 6974 |
+
},
|
| 6975 |
+
{
|
| 6976 |
+
"epoch": 0.58,
|
| 6977 |
+
"learning_rate": 0.00037733333333333336,
|
| 6978 |
+
"loss": 2.1494,
|
| 6979 |
+
"step": 1151
|
| 6980 |
+
},
|
| 6981 |
+
{
|
| 6982 |
+
"epoch": 0.58,
|
| 6983 |
+
"learning_rate": 0.0003768888888888889,
|
| 6984 |
+
"loss": 2.2536,
|
| 6985 |
+
"step": 1152
|
| 6986 |
+
},
|
| 6987 |
+
{
|
| 6988 |
+
"epoch": 0.58,
|
| 6989 |
+
"learning_rate": 0.00037644444444444444,
|
| 6990 |
+
"loss": 2.186,
|
| 6991 |
+
"step": 1153
|
| 6992 |
+
},
|
| 6993 |
+
{
|
| 6994 |
+
"epoch": 0.58,
|
| 6995 |
+
"learning_rate": 0.000376,
|
| 6996 |
+
"loss": 2.2463,
|
| 6997 |
+
"step": 1154
|
| 6998 |
+
},
|
| 6999 |
+
{
|
| 7000 |
+
"epoch": 0.58,
|
| 7001 |
+
"learning_rate": 0.00037555555555555557,
|
| 7002 |
+
"loss": 2.2183,
|
| 7003 |
+
"step": 1155
|
| 7004 |
+
},
|
| 7005 |
+
{
|
| 7006 |
+
"epoch": 0.58,
|
| 7007 |
+
"learning_rate": 0.00037511111111111116,
|
| 7008 |
+
"loss": 2.2226,
|
| 7009 |
+
"step": 1156
|
| 7010 |
+
},
|
| 7011 |
+
{
|
| 7012 |
+
"epoch": 0.58,
|
| 7013 |
+
"learning_rate": 0.0003746666666666667,
|
| 7014 |
+
"loss": 2.2353,
|
| 7015 |
+
"step": 1157
|
| 7016 |
+
},
|
| 7017 |
+
{
|
| 7018 |
+
"epoch": 0.58,
|
| 7019 |
+
"learning_rate": 0.00037422222222222224,
|
| 7020 |
+
"loss": 2.2144,
|
| 7021 |
+
"step": 1158
|
| 7022 |
+
},
|
| 7023 |
+
{
|
| 7024 |
+
"epoch": 0.58,
|
| 7025 |
+
"learning_rate": 0.0003737777777777778,
|
| 7026 |
+
"loss": 2.232,
|
| 7027 |
+
"step": 1159
|
| 7028 |
+
},
|
| 7029 |
+
{
|
| 7030 |
+
"epoch": 0.58,
|
| 7031 |
+
"learning_rate": 0.0003733333333333334,
|
| 7032 |
+
"loss": 2.2233,
|
| 7033 |
+
"step": 1160
|
| 7034 |
+
},
|
| 7035 |
+
{
|
| 7036 |
+
"epoch": 0.58,
|
| 7037 |
+
"learning_rate": 0.0003728888888888889,
|
| 7038 |
+
"loss": 2.1968,
|
| 7039 |
+
"step": 1161
|
| 7040 |
+
},
|
| 7041 |
+
{
|
| 7042 |
+
"epoch": 0.58,
|
| 7043 |
+
"learning_rate": 0.00037244444444444445,
|
| 7044 |
+
"loss": 2.2326,
|
| 7045 |
+
"step": 1162
|
| 7046 |
+
},
|
| 7047 |
+
{
|
| 7048 |
+
"epoch": 0.58,
|
| 7049 |
+
"learning_rate": 0.00037200000000000004,
|
| 7050 |
+
"loss": 2.1779,
|
| 7051 |
+
"step": 1163
|
| 7052 |
+
},
|
| 7053 |
+
{
|
| 7054 |
+
"epoch": 0.58,
|
| 7055 |
+
"learning_rate": 0.0003715555555555556,
|
| 7056 |
+
"loss": 2.1174,
|
| 7057 |
+
"step": 1164
|
| 7058 |
+
},
|
| 7059 |
+
{
|
| 7060 |
+
"epoch": 0.58,
|
| 7061 |
+
"learning_rate": 0.0003711111111111111,
|
| 7062 |
+
"loss": 2.2201,
|
| 7063 |
+
"step": 1165
|
| 7064 |
+
},
|
| 7065 |
+
{
|
| 7066 |
+
"epoch": 0.58,
|
| 7067 |
+
"learning_rate": 0.00037066666666666666,
|
| 7068 |
+
"loss": 2.1531,
|
| 7069 |
+
"step": 1166
|
| 7070 |
+
},
|
| 7071 |
+
{
|
| 7072 |
+
"epoch": 0.58,
|
| 7073 |
+
"learning_rate": 0.00037022222222222225,
|
| 7074 |
+
"loss": 2.2818,
|
| 7075 |
+
"step": 1167
|
| 7076 |
+
},
|
| 7077 |
+
{
|
| 7078 |
+
"epoch": 0.58,
|
| 7079 |
+
"learning_rate": 0.0003697777777777778,
|
| 7080 |
+
"loss": 2.2397,
|
| 7081 |
+
"step": 1168
|
| 7082 |
+
},
|
| 7083 |
+
{
|
| 7084 |
+
"epoch": 0.58,
|
| 7085 |
+
"learning_rate": 0.00036933333333333333,
|
| 7086 |
+
"loss": 2.2505,
|
| 7087 |
+
"step": 1169
|
| 7088 |
+
},
|
| 7089 |
+
{
|
| 7090 |
+
"epoch": 0.58,
|
| 7091 |
+
"learning_rate": 0.0003688888888888889,
|
| 7092 |
+
"loss": 2.3194,
|
| 7093 |
+
"step": 1170
|
| 7094 |
+
},
|
| 7095 |
+
{
|
| 7096 |
+
"epoch": 0.59,
|
| 7097 |
+
"learning_rate": 0.00036844444444444446,
|
| 7098 |
+
"loss": 2.2095,
|
| 7099 |
+
"step": 1171
|
| 7100 |
+
},
|
| 7101 |
+
{
|
| 7102 |
+
"epoch": 0.59,
|
| 7103 |
+
"learning_rate": 0.00036800000000000005,
|
| 7104 |
+
"loss": 2.2788,
|
| 7105 |
+
"step": 1172
|
| 7106 |
+
},
|
| 7107 |
+
{
|
| 7108 |
+
"epoch": 0.59,
|
| 7109 |
+
"learning_rate": 0.00036755555555555554,
|
| 7110 |
+
"loss": 2.1899,
|
| 7111 |
+
"step": 1173
|
| 7112 |
+
},
|
| 7113 |
+
{
|
| 7114 |
+
"epoch": 0.59,
|
| 7115 |
+
"learning_rate": 0.00036711111111111113,
|
| 7116 |
+
"loss": 2.1734,
|
| 7117 |
+
"step": 1174
|
| 7118 |
+
},
|
| 7119 |
+
{
|
| 7120 |
+
"epoch": 0.59,
|
| 7121 |
+
"learning_rate": 0.00036666666666666667,
|
| 7122 |
+
"loss": 2.2329,
|
| 7123 |
+
"step": 1175
|
| 7124 |
+
},
|
| 7125 |
+
{
|
| 7126 |
+
"epoch": 0.59,
|
| 7127 |
+
"learning_rate": 0.00036622222222222226,
|
| 7128 |
+
"loss": 2.2439,
|
| 7129 |
+
"step": 1176
|
| 7130 |
+
},
|
| 7131 |
+
{
|
| 7132 |
+
"epoch": 0.59,
|
| 7133 |
+
"learning_rate": 0.00036577777777777775,
|
| 7134 |
+
"loss": 2.1828,
|
| 7135 |
+
"step": 1177
|
| 7136 |
+
},
|
| 7137 |
+
{
|
| 7138 |
+
"epoch": 0.59,
|
| 7139 |
+
"learning_rate": 0.00036533333333333334,
|
| 7140 |
+
"loss": 2.2352,
|
| 7141 |
+
"step": 1178
|
| 7142 |
+
},
|
| 7143 |
+
{
|
| 7144 |
+
"epoch": 0.59,
|
| 7145 |
+
"learning_rate": 0.00036488888888888893,
|
| 7146 |
+
"loss": 2.2226,
|
| 7147 |
+
"step": 1179
|
| 7148 |
+
},
|
| 7149 |
+
{
|
| 7150 |
+
"epoch": 0.59,
|
| 7151 |
+
"learning_rate": 0.00036444444444444447,
|
| 7152 |
+
"loss": 2.1416,
|
| 7153 |
+
"step": 1180
|
| 7154 |
+
},
|
| 7155 |
+
{
|
| 7156 |
+
"epoch": 0.59,
|
| 7157 |
+
"learning_rate": 0.000364,
|
| 7158 |
+
"loss": 2.2727,
|
| 7159 |
+
"step": 1181
|
| 7160 |
+
},
|
| 7161 |
+
{
|
| 7162 |
+
"epoch": 0.59,
|
| 7163 |
+
"learning_rate": 0.00036355555555555555,
|
| 7164 |
+
"loss": 2.2861,
|
| 7165 |
+
"step": 1182
|
| 7166 |
+
},
|
| 7167 |
+
{
|
| 7168 |
+
"epoch": 0.59,
|
| 7169 |
+
"learning_rate": 0.00036311111111111114,
|
| 7170 |
+
"loss": 2.1931,
|
| 7171 |
+
"step": 1183
|
| 7172 |
+
},
|
| 7173 |
+
{
|
| 7174 |
+
"epoch": 0.59,
|
| 7175 |
+
"learning_rate": 0.0003626666666666667,
|
| 7176 |
+
"loss": 2.2435,
|
| 7177 |
+
"step": 1184
|
| 7178 |
+
},
|
| 7179 |
+
{
|
| 7180 |
+
"epoch": 0.59,
|
| 7181 |
+
"learning_rate": 0.0003622222222222222,
|
| 7182 |
+
"loss": 2.2815,
|
| 7183 |
+
"step": 1185
|
| 7184 |
+
},
|
| 7185 |
+
{
|
| 7186 |
+
"epoch": 0.59,
|
| 7187 |
+
"learning_rate": 0.0003617777777777778,
|
| 7188 |
+
"loss": 2.257,
|
| 7189 |
+
"step": 1186
|
| 7190 |
+
},
|
| 7191 |
+
{
|
| 7192 |
+
"epoch": 0.59,
|
| 7193 |
+
"learning_rate": 0.00036133333333333335,
|
| 7194 |
+
"loss": 2.2182,
|
| 7195 |
+
"step": 1187
|
| 7196 |
+
},
|
| 7197 |
+
{
|
| 7198 |
+
"epoch": 0.59,
|
| 7199 |
+
"learning_rate": 0.00036088888888888895,
|
| 7200 |
+
"loss": 2.2526,
|
| 7201 |
+
"step": 1188
|
| 7202 |
+
},
|
| 7203 |
+
{
|
| 7204 |
+
"epoch": 0.59,
|
| 7205 |
+
"learning_rate": 0.00036044444444444443,
|
| 7206 |
+
"loss": 2.2212,
|
| 7207 |
+
"step": 1189
|
| 7208 |
+
},
|
| 7209 |
+
{
|
| 7210 |
+
"epoch": 0.59,
|
| 7211 |
+
"learning_rate": 0.00036,
|
| 7212 |
+
"loss": 2.2718,
|
| 7213 |
+
"step": 1190
|
| 7214 |
+
},
|
| 7215 |
+
{
|
| 7216 |
+
"epoch": 0.6,
|
| 7217 |
+
"learning_rate": 0.00035955555555555556,
|
| 7218 |
+
"loss": 2.3178,
|
| 7219 |
+
"step": 1191
|
| 7220 |
+
},
|
| 7221 |
+
{
|
| 7222 |
+
"epoch": 0.6,
|
| 7223 |
+
"learning_rate": 0.00035911111111111115,
|
| 7224 |
+
"loss": 2.189,
|
| 7225 |
+
"step": 1192
|
| 7226 |
+
},
|
| 7227 |
+
{
|
| 7228 |
+
"epoch": 0.6,
|
| 7229 |
+
"learning_rate": 0.00035866666666666664,
|
| 7230 |
+
"loss": 2.2615,
|
| 7231 |
+
"step": 1193
|
| 7232 |
+
},
|
| 7233 |
+
{
|
| 7234 |
+
"epoch": 0.6,
|
| 7235 |
+
"learning_rate": 0.00035822222222222223,
|
| 7236 |
+
"loss": 2.1864,
|
| 7237 |
+
"step": 1194
|
| 7238 |
+
},
|
| 7239 |
+
{
|
| 7240 |
+
"epoch": 0.6,
|
| 7241 |
+
"learning_rate": 0.0003577777777777778,
|
| 7242 |
+
"loss": 2.1476,
|
| 7243 |
+
"step": 1195
|
| 7244 |
+
},
|
| 7245 |
+
{
|
| 7246 |
+
"epoch": 0.6,
|
| 7247 |
+
"learning_rate": 0.00035733333333333336,
|
| 7248 |
+
"loss": 2.2484,
|
| 7249 |
+
"step": 1196
|
| 7250 |
+
},
|
| 7251 |
+
{
|
| 7252 |
+
"epoch": 0.6,
|
| 7253 |
+
"learning_rate": 0.0003568888888888889,
|
| 7254 |
+
"loss": 2.2128,
|
| 7255 |
+
"step": 1197
|
| 7256 |
+
},
|
| 7257 |
+
{
|
| 7258 |
+
"epoch": 0.6,
|
| 7259 |
+
"learning_rate": 0.00035644444444444444,
|
| 7260 |
+
"loss": 2.1948,
|
| 7261 |
+
"step": 1198
|
| 7262 |
+
},
|
| 7263 |
+
{
|
| 7264 |
+
"epoch": 0.6,
|
| 7265 |
+
"learning_rate": 0.00035600000000000003,
|
| 7266 |
+
"loss": 2.292,
|
| 7267 |
+
"step": 1199
|
| 7268 |
+
},
|
| 7269 |
+
{
|
| 7270 |
+
"epoch": 0.6,
|
| 7271 |
+
"learning_rate": 0.00035555555555555557,
|
| 7272 |
+
"loss": 2.1969,
|
| 7273 |
+
"step": 1200
|
| 7274 |
+
},
|
| 7275 |
+
{
|
| 7276 |
+
"epoch": 0.6,
|
| 7277 |
+
"eval_gen_len": 1023.0,
|
| 7278 |
+
"eval_loss": 2.126681089401245,
|
| 7279 |
+
"eval_rouge1": 13.5606,
|
| 7280 |
+
"eval_rouge2": 4.5005,
|
| 7281 |
+
"eval_rougeL": 8.5459,
|
| 7282 |
+
"eval_rougeLsum": 8.5241,
|
| 7283 |
+
"eval_runtime": 9294.4875,
|
| 7284 |
+
"eval_samples_per_second": 0.054,
|
| 7285 |
+
"eval_steps_per_second": 0.054,
|
| 7286 |
+
"step": 1200
|
| 7287 |
}
|
| 7288 |
],
|
| 7289 |
"max_steps": 2000,
|
| 7290 |
"num_train_epochs": 9223372036854775807,
|
| 7291 |
+
"total_flos": 9.642800147560151e+17,
|
| 7292 |
"trial_name": null,
|
| 7293 |
"trial_params": null
|
| 7294 |
}
|
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 sha256:951861143d15fa1c5e1fffde87d6bb38f0764793caa011f89d710cfbb0713166
|
| 3 |
size 2368281769
|