Training in progress, step 1800
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:ad303116119ac3dd30faa387e82a893e400cc6b5394fd8a2316ed68e06379fbf
|
| 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:f420a02b8b31a90067fa5983f6a9bf2cf10ec0e98b96f411b86d4b5b802edba4
|
| 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:d3a5865aa75d1a342f9e4f08604e7cf4a9b9aefed59e5d9faabd23163aa4ef04
|
| 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 sha256:9dca826faf822cb98c83b9c364c8db05cdbd5491b1fff17afae58f1f6a5090de
|
| 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": 1.
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
@@ -9710,11 +9710,1224 @@
|
|
| 9710 |
"eval_samples_per_second": 0.055,
|
| 9711 |
"eval_steps_per_second": 0.055,
|
| 9712 |
"step": 1600
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9713 |
}
|
| 9714 |
],
|
| 9715 |
"max_steps": 2000,
|
| 9716 |
"num_train_epochs": 9223372036854775807,
|
| 9717 |
-
"total_flos": 1.
|
| 9718 |
"trial_name": null,
|
| 9719 |
"trial_params": null
|
| 9720 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 2.099489688873291,
|
| 3 |
+
"best_model_checkpoint": "output/checkpoint-1800",
|
| 4 |
+
"epoch": 1.28650390625,
|
| 5 |
+
"global_step": 1800,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 9710 |
"eval_samples_per_second": 0.055,
|
| 9711 |
"eval_steps_per_second": 0.055,
|
| 9712 |
"step": 1600
|
| 9713 |
+
},
|
| 9714 |
+
{
|
| 9715 |
+
"epoch": 1.19,
|
| 9716 |
+
"learning_rate": 0.00017733333333333335,
|
| 9717 |
+
"loss": 2.1553,
|
| 9718 |
+
"step": 1601
|
| 9719 |
+
},
|
| 9720 |
+
{
|
| 9721 |
+
"epoch": 1.19,
|
| 9722 |
+
"learning_rate": 0.0001768888888888889,
|
| 9723 |
+
"loss": 2.1205,
|
| 9724 |
+
"step": 1602
|
| 9725 |
+
},
|
| 9726 |
+
{
|
| 9727 |
+
"epoch": 1.19,
|
| 9728 |
+
"learning_rate": 0.00017644444444444446,
|
| 9729 |
+
"loss": 2.0923,
|
| 9730 |
+
"step": 1603
|
| 9731 |
+
},
|
| 9732 |
+
{
|
| 9733 |
+
"epoch": 1.19,
|
| 9734 |
+
"learning_rate": 0.00017600000000000002,
|
| 9735 |
+
"loss": 2.1086,
|
| 9736 |
+
"step": 1604
|
| 9737 |
+
},
|
| 9738 |
+
{
|
| 9739 |
+
"epoch": 1.19,
|
| 9740 |
+
"learning_rate": 0.00017555555555555556,
|
| 9741 |
+
"loss": 2.1597,
|
| 9742 |
+
"step": 1605
|
| 9743 |
+
},
|
| 9744 |
+
{
|
| 9745 |
+
"epoch": 1.19,
|
| 9746 |
+
"learning_rate": 0.00017511111111111113,
|
| 9747 |
+
"loss": 2.1147,
|
| 9748 |
+
"step": 1606
|
| 9749 |
+
},
|
| 9750 |
+
{
|
| 9751 |
+
"epoch": 1.19,
|
| 9752 |
+
"learning_rate": 0.00017466666666666667,
|
| 9753 |
+
"loss": 2.1185,
|
| 9754 |
+
"step": 1607
|
| 9755 |
+
},
|
| 9756 |
+
{
|
| 9757 |
+
"epoch": 1.19,
|
| 9758 |
+
"learning_rate": 0.00017422222222222223,
|
| 9759 |
+
"loss": 2.1641,
|
| 9760 |
+
"step": 1608
|
| 9761 |
+
},
|
| 9762 |
+
{
|
| 9763 |
+
"epoch": 1.19,
|
| 9764 |
+
"learning_rate": 0.0001737777777777778,
|
| 9765 |
+
"loss": 2.1046,
|
| 9766 |
+
"step": 1609
|
| 9767 |
+
},
|
| 9768 |
+
{
|
| 9769 |
+
"epoch": 1.19,
|
| 9770 |
+
"learning_rate": 0.00017333333333333334,
|
| 9771 |
+
"loss": 2.136,
|
| 9772 |
+
"step": 1610
|
| 9773 |
+
},
|
| 9774 |
+
{
|
| 9775 |
+
"epoch": 1.19,
|
| 9776 |
+
"learning_rate": 0.0001728888888888889,
|
| 9777 |
+
"loss": 2.1428,
|
| 9778 |
+
"step": 1611
|
| 9779 |
+
},
|
| 9780 |
+
{
|
| 9781 |
+
"epoch": 1.19,
|
| 9782 |
+
"learning_rate": 0.00017244444444444444,
|
| 9783 |
+
"loss": 2.1183,
|
| 9784 |
+
"step": 1612
|
| 9785 |
+
},
|
| 9786 |
+
{
|
| 9787 |
+
"epoch": 1.19,
|
| 9788 |
+
"learning_rate": 0.000172,
|
| 9789 |
+
"loss": 2.1328,
|
| 9790 |
+
"step": 1613
|
| 9791 |
+
},
|
| 9792 |
+
{
|
| 9793 |
+
"epoch": 1.19,
|
| 9794 |
+
"learning_rate": 0.00017155555555555555,
|
| 9795 |
+
"loss": 2.1679,
|
| 9796 |
+
"step": 1614
|
| 9797 |
+
},
|
| 9798 |
+
{
|
| 9799 |
+
"epoch": 1.19,
|
| 9800 |
+
"learning_rate": 0.0001711111111111111,
|
| 9801 |
+
"loss": 2.0474,
|
| 9802 |
+
"step": 1615
|
| 9803 |
+
},
|
| 9804 |
+
{
|
| 9805 |
+
"epoch": 1.19,
|
| 9806 |
+
"learning_rate": 0.00017066666666666668,
|
| 9807 |
+
"loss": 2.1221,
|
| 9808 |
+
"step": 1616
|
| 9809 |
+
},
|
| 9810 |
+
{
|
| 9811 |
+
"epoch": 1.2,
|
| 9812 |
+
"learning_rate": 0.00017022222222222224,
|
| 9813 |
+
"loss": 2.1004,
|
| 9814 |
+
"step": 1617
|
| 9815 |
+
},
|
| 9816 |
+
{
|
| 9817 |
+
"epoch": 1.2,
|
| 9818 |
+
"learning_rate": 0.00016977777777777778,
|
| 9819 |
+
"loss": 2.1145,
|
| 9820 |
+
"step": 1618
|
| 9821 |
+
},
|
| 9822 |
+
{
|
| 9823 |
+
"epoch": 1.2,
|
| 9824 |
+
"learning_rate": 0.00016933333333333335,
|
| 9825 |
+
"loss": 2.0614,
|
| 9826 |
+
"step": 1619
|
| 9827 |
+
},
|
| 9828 |
+
{
|
| 9829 |
+
"epoch": 1.2,
|
| 9830 |
+
"learning_rate": 0.00016888888888888889,
|
| 9831 |
+
"loss": 2.0831,
|
| 9832 |
+
"step": 1620
|
| 9833 |
+
},
|
| 9834 |
+
{
|
| 9835 |
+
"epoch": 1.2,
|
| 9836 |
+
"learning_rate": 0.00016844444444444445,
|
| 9837 |
+
"loss": 2.1027,
|
| 9838 |
+
"step": 1621
|
| 9839 |
+
},
|
| 9840 |
+
{
|
| 9841 |
+
"epoch": 1.2,
|
| 9842 |
+
"learning_rate": 0.000168,
|
| 9843 |
+
"loss": 2.0872,
|
| 9844 |
+
"step": 1622
|
| 9845 |
+
},
|
| 9846 |
+
{
|
| 9847 |
+
"epoch": 1.2,
|
| 9848 |
+
"learning_rate": 0.00016755555555555556,
|
| 9849 |
+
"loss": 2.1903,
|
| 9850 |
+
"step": 1623
|
| 9851 |
+
},
|
| 9852 |
+
{
|
| 9853 |
+
"epoch": 1.2,
|
| 9854 |
+
"learning_rate": 0.00016711111111111112,
|
| 9855 |
+
"loss": 2.0553,
|
| 9856 |
+
"step": 1624
|
| 9857 |
+
},
|
| 9858 |
+
{
|
| 9859 |
+
"epoch": 1.2,
|
| 9860 |
+
"learning_rate": 0.0001666666666666667,
|
| 9861 |
+
"loss": 2.0946,
|
| 9862 |
+
"step": 1625
|
| 9863 |
+
},
|
| 9864 |
+
{
|
| 9865 |
+
"epoch": 1.2,
|
| 9866 |
+
"learning_rate": 0.00016622222222222223,
|
| 9867 |
+
"loss": 2.0989,
|
| 9868 |
+
"step": 1626
|
| 9869 |
+
},
|
| 9870 |
+
{
|
| 9871 |
+
"epoch": 1.2,
|
| 9872 |
+
"learning_rate": 0.0001657777777777778,
|
| 9873 |
+
"loss": 2.0825,
|
| 9874 |
+
"step": 1627
|
| 9875 |
+
},
|
| 9876 |
+
{
|
| 9877 |
+
"epoch": 1.2,
|
| 9878 |
+
"learning_rate": 0.00016533333333333333,
|
| 9879 |
+
"loss": 2.1103,
|
| 9880 |
+
"step": 1628
|
| 9881 |
+
},
|
| 9882 |
+
{
|
| 9883 |
+
"epoch": 1.2,
|
| 9884 |
+
"learning_rate": 0.0001648888888888889,
|
| 9885 |
+
"loss": 2.0972,
|
| 9886 |
+
"step": 1629
|
| 9887 |
+
},
|
| 9888 |
+
{
|
| 9889 |
+
"epoch": 1.2,
|
| 9890 |
+
"learning_rate": 0.00016444444444444444,
|
| 9891 |
+
"loss": 2.1631,
|
| 9892 |
+
"step": 1630
|
| 9893 |
+
},
|
| 9894 |
+
{
|
| 9895 |
+
"epoch": 1.2,
|
| 9896 |
+
"learning_rate": 0.000164,
|
| 9897 |
+
"loss": 2.0751,
|
| 9898 |
+
"step": 1631
|
| 9899 |
+
},
|
| 9900 |
+
{
|
| 9901 |
+
"epoch": 1.2,
|
| 9902 |
+
"learning_rate": 0.00016355555555555557,
|
| 9903 |
+
"loss": 2.1255,
|
| 9904 |
+
"step": 1632
|
| 9905 |
+
},
|
| 9906 |
+
{
|
| 9907 |
+
"epoch": 1.2,
|
| 9908 |
+
"learning_rate": 0.00016311111111111113,
|
| 9909 |
+
"loss": 2.159,
|
| 9910 |
+
"step": 1633
|
| 9911 |
+
},
|
| 9912 |
+
{
|
| 9913 |
+
"epoch": 1.2,
|
| 9914 |
+
"learning_rate": 0.00016266666666666667,
|
| 9915 |
+
"loss": 2.1844,
|
| 9916 |
+
"step": 1634
|
| 9917 |
+
},
|
| 9918 |
+
{
|
| 9919 |
+
"epoch": 1.2,
|
| 9920 |
+
"learning_rate": 0.00016222222222222224,
|
| 9921 |
+
"loss": 2.1239,
|
| 9922 |
+
"step": 1635
|
| 9923 |
+
},
|
| 9924 |
+
{
|
| 9925 |
+
"epoch": 1.2,
|
| 9926 |
+
"learning_rate": 0.00016177777777777778,
|
| 9927 |
+
"loss": 2.1313,
|
| 9928 |
+
"step": 1636
|
| 9929 |
+
},
|
| 9930 |
+
{
|
| 9931 |
+
"epoch": 1.21,
|
| 9932 |
+
"learning_rate": 0.00016133333333333334,
|
| 9933 |
+
"loss": 2.075,
|
| 9934 |
+
"step": 1637
|
| 9935 |
+
},
|
| 9936 |
+
{
|
| 9937 |
+
"epoch": 1.21,
|
| 9938 |
+
"learning_rate": 0.00016088888888888888,
|
| 9939 |
+
"loss": 2.1068,
|
| 9940 |
+
"step": 1638
|
| 9941 |
+
},
|
| 9942 |
+
{
|
| 9943 |
+
"epoch": 1.21,
|
| 9944 |
+
"learning_rate": 0.00016044444444444445,
|
| 9945 |
+
"loss": 2.1633,
|
| 9946 |
+
"step": 1639
|
| 9947 |
+
},
|
| 9948 |
+
{
|
| 9949 |
+
"epoch": 1.21,
|
| 9950 |
+
"learning_rate": 0.00016,
|
| 9951 |
+
"loss": 2.073,
|
| 9952 |
+
"step": 1640
|
| 9953 |
+
},
|
| 9954 |
+
{
|
| 9955 |
+
"epoch": 1.21,
|
| 9956 |
+
"learning_rate": 0.00015955555555555558,
|
| 9957 |
+
"loss": 2.0468,
|
| 9958 |
+
"step": 1641
|
| 9959 |
+
},
|
| 9960 |
+
{
|
| 9961 |
+
"epoch": 1.21,
|
| 9962 |
+
"learning_rate": 0.00015911111111111112,
|
| 9963 |
+
"loss": 2.0787,
|
| 9964 |
+
"step": 1642
|
| 9965 |
+
},
|
| 9966 |
+
{
|
| 9967 |
+
"epoch": 1.21,
|
| 9968 |
+
"learning_rate": 0.00015866666666666668,
|
| 9969 |
+
"loss": 2.123,
|
| 9970 |
+
"step": 1643
|
| 9971 |
+
},
|
| 9972 |
+
{
|
| 9973 |
+
"epoch": 1.21,
|
| 9974 |
+
"learning_rate": 0.00015822222222222222,
|
| 9975 |
+
"loss": 2.1059,
|
| 9976 |
+
"step": 1644
|
| 9977 |
+
},
|
| 9978 |
+
{
|
| 9979 |
+
"epoch": 1.21,
|
| 9980 |
+
"learning_rate": 0.0001577777777777778,
|
| 9981 |
+
"loss": 2.0936,
|
| 9982 |
+
"step": 1645
|
| 9983 |
+
},
|
| 9984 |
+
{
|
| 9985 |
+
"epoch": 1.21,
|
| 9986 |
+
"learning_rate": 0.00015733333333333333,
|
| 9987 |
+
"loss": 2.0886,
|
| 9988 |
+
"step": 1646
|
| 9989 |
+
},
|
| 9990 |
+
{
|
| 9991 |
+
"epoch": 1.21,
|
| 9992 |
+
"learning_rate": 0.00015688888888888892,
|
| 9993 |
+
"loss": 2.1275,
|
| 9994 |
+
"step": 1647
|
| 9995 |
+
},
|
| 9996 |
+
{
|
| 9997 |
+
"epoch": 1.21,
|
| 9998 |
+
"learning_rate": 0.00015644444444444446,
|
| 9999 |
+
"loss": 2.1048,
|
| 10000 |
+
"step": 1648
|
| 10001 |
+
},
|
| 10002 |
+
{
|
| 10003 |
+
"epoch": 1.21,
|
| 10004 |
+
"learning_rate": 0.00015600000000000002,
|
| 10005 |
+
"loss": 2.1402,
|
| 10006 |
+
"step": 1649
|
| 10007 |
+
},
|
| 10008 |
+
{
|
| 10009 |
+
"epoch": 1.21,
|
| 10010 |
+
"learning_rate": 0.00015555555555555556,
|
| 10011 |
+
"loss": 2.0868,
|
| 10012 |
+
"step": 1650
|
| 10013 |
+
},
|
| 10014 |
+
{
|
| 10015 |
+
"epoch": 1.21,
|
| 10016 |
+
"learning_rate": 0.00015511111111111113,
|
| 10017 |
+
"loss": 2.154,
|
| 10018 |
+
"step": 1651
|
| 10019 |
+
},
|
| 10020 |
+
{
|
| 10021 |
+
"epoch": 1.21,
|
| 10022 |
+
"learning_rate": 0.00015466666666666667,
|
| 10023 |
+
"loss": 2.172,
|
| 10024 |
+
"step": 1652
|
| 10025 |
+
},
|
| 10026 |
+
{
|
| 10027 |
+
"epoch": 1.21,
|
| 10028 |
+
"learning_rate": 0.00015422222222222223,
|
| 10029 |
+
"loss": 2.1129,
|
| 10030 |
+
"step": 1653
|
| 10031 |
+
},
|
| 10032 |
+
{
|
| 10033 |
+
"epoch": 1.21,
|
| 10034 |
+
"learning_rate": 0.00015377777777777777,
|
| 10035 |
+
"loss": 2.1185,
|
| 10036 |
+
"step": 1654
|
| 10037 |
+
},
|
| 10038 |
+
{
|
| 10039 |
+
"epoch": 1.21,
|
| 10040 |
+
"learning_rate": 0.00015333333333333334,
|
| 10041 |
+
"loss": 2.0523,
|
| 10042 |
+
"step": 1655
|
| 10043 |
+
},
|
| 10044 |
+
{
|
| 10045 |
+
"epoch": 1.21,
|
| 10046 |
+
"learning_rate": 0.0001528888888888889,
|
| 10047 |
+
"loss": 2.0779,
|
| 10048 |
+
"step": 1656
|
| 10049 |
+
},
|
| 10050 |
+
{
|
| 10051 |
+
"epoch": 1.22,
|
| 10052 |
+
"learning_rate": 0.00015244444444444447,
|
| 10053 |
+
"loss": 2.149,
|
| 10054 |
+
"step": 1657
|
| 10055 |
+
},
|
| 10056 |
+
{
|
| 10057 |
+
"epoch": 1.22,
|
| 10058 |
+
"learning_rate": 0.000152,
|
| 10059 |
+
"loss": 2.0786,
|
| 10060 |
+
"step": 1658
|
| 10061 |
+
},
|
| 10062 |
+
{
|
| 10063 |
+
"epoch": 1.22,
|
| 10064 |
+
"learning_rate": 0.00015155555555555557,
|
| 10065 |
+
"loss": 2.1615,
|
| 10066 |
+
"step": 1659
|
| 10067 |
+
},
|
| 10068 |
+
{
|
| 10069 |
+
"epoch": 1.22,
|
| 10070 |
+
"learning_rate": 0.0001511111111111111,
|
| 10071 |
+
"loss": 2.0473,
|
| 10072 |
+
"step": 1660
|
| 10073 |
+
},
|
| 10074 |
+
{
|
| 10075 |
+
"epoch": 1.22,
|
| 10076 |
+
"learning_rate": 0.00015066666666666668,
|
| 10077 |
+
"loss": 2.1122,
|
| 10078 |
+
"step": 1661
|
| 10079 |
+
},
|
| 10080 |
+
{
|
| 10081 |
+
"epoch": 1.22,
|
| 10082 |
+
"learning_rate": 0.00015022222222222222,
|
| 10083 |
+
"loss": 2.1615,
|
| 10084 |
+
"step": 1662
|
| 10085 |
+
},
|
| 10086 |
+
{
|
| 10087 |
+
"epoch": 1.22,
|
| 10088 |
+
"learning_rate": 0.00014977777777777778,
|
| 10089 |
+
"loss": 2.1331,
|
| 10090 |
+
"step": 1663
|
| 10091 |
+
},
|
| 10092 |
+
{
|
| 10093 |
+
"epoch": 1.22,
|
| 10094 |
+
"learning_rate": 0.00014933333333333335,
|
| 10095 |
+
"loss": 2.1782,
|
| 10096 |
+
"step": 1664
|
| 10097 |
+
},
|
| 10098 |
+
{
|
| 10099 |
+
"epoch": 1.22,
|
| 10100 |
+
"learning_rate": 0.0001488888888888889,
|
| 10101 |
+
"loss": 2.126,
|
| 10102 |
+
"step": 1665
|
| 10103 |
+
},
|
| 10104 |
+
{
|
| 10105 |
+
"epoch": 1.22,
|
| 10106 |
+
"learning_rate": 0.00014844444444444445,
|
| 10107 |
+
"loss": 2.1666,
|
| 10108 |
+
"step": 1666
|
| 10109 |
+
},
|
| 10110 |
+
{
|
| 10111 |
+
"epoch": 1.22,
|
| 10112 |
+
"learning_rate": 0.000148,
|
| 10113 |
+
"loss": 2.0928,
|
| 10114 |
+
"step": 1667
|
| 10115 |
+
},
|
| 10116 |
+
{
|
| 10117 |
+
"epoch": 1.22,
|
| 10118 |
+
"learning_rate": 0.00014755555555555556,
|
| 10119 |
+
"loss": 2.111,
|
| 10120 |
+
"step": 1668
|
| 10121 |
+
},
|
| 10122 |
+
{
|
| 10123 |
+
"epoch": 1.22,
|
| 10124 |
+
"learning_rate": 0.00014711111111111112,
|
| 10125 |
+
"loss": 2.1106,
|
| 10126 |
+
"step": 1669
|
| 10127 |
+
},
|
| 10128 |
+
{
|
| 10129 |
+
"epoch": 1.22,
|
| 10130 |
+
"learning_rate": 0.00014666666666666666,
|
| 10131 |
+
"loss": 2.1035,
|
| 10132 |
+
"step": 1670
|
| 10133 |
+
},
|
| 10134 |
+
{
|
| 10135 |
+
"epoch": 1.22,
|
| 10136 |
+
"learning_rate": 0.00014622222222222223,
|
| 10137 |
+
"loss": 2.0571,
|
| 10138 |
+
"step": 1671
|
| 10139 |
+
},
|
| 10140 |
+
{
|
| 10141 |
+
"epoch": 1.22,
|
| 10142 |
+
"learning_rate": 0.0001457777777777778,
|
| 10143 |
+
"loss": 2.1375,
|
| 10144 |
+
"step": 1672
|
| 10145 |
+
},
|
| 10146 |
+
{
|
| 10147 |
+
"epoch": 1.22,
|
| 10148 |
+
"learning_rate": 0.00014533333333333333,
|
| 10149 |
+
"loss": 2.1541,
|
| 10150 |
+
"step": 1673
|
| 10151 |
+
},
|
| 10152 |
+
{
|
| 10153 |
+
"epoch": 1.22,
|
| 10154 |
+
"learning_rate": 0.0001448888888888889,
|
| 10155 |
+
"loss": 2.1483,
|
| 10156 |
+
"step": 1674
|
| 10157 |
+
},
|
| 10158 |
+
{
|
| 10159 |
+
"epoch": 1.22,
|
| 10160 |
+
"learning_rate": 0.00014444444444444444,
|
| 10161 |
+
"loss": 2.1686,
|
| 10162 |
+
"step": 1675
|
| 10163 |
+
},
|
| 10164 |
+
{
|
| 10165 |
+
"epoch": 1.22,
|
| 10166 |
+
"learning_rate": 0.000144,
|
| 10167 |
+
"loss": 2.1511,
|
| 10168 |
+
"step": 1676
|
| 10169 |
+
},
|
| 10170 |
+
{
|
| 10171 |
+
"epoch": 1.23,
|
| 10172 |
+
"learning_rate": 0.00014355555555555554,
|
| 10173 |
+
"loss": 2.1709,
|
| 10174 |
+
"step": 1677
|
| 10175 |
+
},
|
| 10176 |
+
{
|
| 10177 |
+
"epoch": 1.23,
|
| 10178 |
+
"learning_rate": 0.0001431111111111111,
|
| 10179 |
+
"loss": 2.1034,
|
| 10180 |
+
"step": 1678
|
| 10181 |
+
},
|
| 10182 |
+
{
|
| 10183 |
+
"epoch": 1.23,
|
| 10184 |
+
"learning_rate": 0.00014266666666666667,
|
| 10185 |
+
"loss": 2.1392,
|
| 10186 |
+
"step": 1679
|
| 10187 |
+
},
|
| 10188 |
+
{
|
| 10189 |
+
"epoch": 1.23,
|
| 10190 |
+
"learning_rate": 0.00014222222222222224,
|
| 10191 |
+
"loss": 2.0913,
|
| 10192 |
+
"step": 1680
|
| 10193 |
+
},
|
| 10194 |
+
{
|
| 10195 |
+
"epoch": 1.23,
|
| 10196 |
+
"learning_rate": 0.00014177777777777778,
|
| 10197 |
+
"loss": 2.1404,
|
| 10198 |
+
"step": 1681
|
| 10199 |
+
},
|
| 10200 |
+
{
|
| 10201 |
+
"epoch": 1.23,
|
| 10202 |
+
"learning_rate": 0.00014133333333333334,
|
| 10203 |
+
"loss": 2.0701,
|
| 10204 |
+
"step": 1682
|
| 10205 |
+
},
|
| 10206 |
+
{
|
| 10207 |
+
"epoch": 1.23,
|
| 10208 |
+
"learning_rate": 0.00014088888888888888,
|
| 10209 |
+
"loss": 2.0879,
|
| 10210 |
+
"step": 1683
|
| 10211 |
+
},
|
| 10212 |
+
{
|
| 10213 |
+
"epoch": 1.23,
|
| 10214 |
+
"learning_rate": 0.00014044444444444445,
|
| 10215 |
+
"loss": 2.1121,
|
| 10216 |
+
"step": 1684
|
| 10217 |
+
},
|
| 10218 |
+
{
|
| 10219 |
+
"epoch": 1.23,
|
| 10220 |
+
"learning_rate": 0.00014,
|
| 10221 |
+
"loss": 2.1578,
|
| 10222 |
+
"step": 1685
|
| 10223 |
+
},
|
| 10224 |
+
{
|
| 10225 |
+
"epoch": 1.23,
|
| 10226 |
+
"learning_rate": 0.00013955555555555558,
|
| 10227 |
+
"loss": 2.1933,
|
| 10228 |
+
"step": 1686
|
| 10229 |
+
},
|
| 10230 |
+
{
|
| 10231 |
+
"epoch": 1.23,
|
| 10232 |
+
"learning_rate": 0.00013911111111111112,
|
| 10233 |
+
"loss": 2.1029,
|
| 10234 |
+
"step": 1687
|
| 10235 |
+
},
|
| 10236 |
+
{
|
| 10237 |
+
"epoch": 1.23,
|
| 10238 |
+
"learning_rate": 0.00013866666666666669,
|
| 10239 |
+
"loss": 2.1216,
|
| 10240 |
+
"step": 1688
|
| 10241 |
+
},
|
| 10242 |
+
{
|
| 10243 |
+
"epoch": 1.23,
|
| 10244 |
+
"learning_rate": 0.00013822222222222222,
|
| 10245 |
+
"loss": 2.1357,
|
| 10246 |
+
"step": 1689
|
| 10247 |
+
},
|
| 10248 |
+
{
|
| 10249 |
+
"epoch": 1.23,
|
| 10250 |
+
"learning_rate": 0.0001377777777777778,
|
| 10251 |
+
"loss": 2.1176,
|
| 10252 |
+
"step": 1690
|
| 10253 |
+
},
|
| 10254 |
+
{
|
| 10255 |
+
"epoch": 1.23,
|
| 10256 |
+
"learning_rate": 0.00013733333333333333,
|
| 10257 |
+
"loss": 2.1144,
|
| 10258 |
+
"step": 1691
|
| 10259 |
+
},
|
| 10260 |
+
{
|
| 10261 |
+
"epoch": 1.23,
|
| 10262 |
+
"learning_rate": 0.0001368888888888889,
|
| 10263 |
+
"loss": 2.1074,
|
| 10264 |
+
"step": 1692
|
| 10265 |
+
},
|
| 10266 |
+
{
|
| 10267 |
+
"epoch": 1.23,
|
| 10268 |
+
"learning_rate": 0.00013644444444444443,
|
| 10269 |
+
"loss": 2.0608,
|
| 10270 |
+
"step": 1693
|
| 10271 |
+
},
|
| 10272 |
+
{
|
| 10273 |
+
"epoch": 1.23,
|
| 10274 |
+
"learning_rate": 0.00013600000000000003,
|
| 10275 |
+
"loss": 2.0729,
|
| 10276 |
+
"step": 1694
|
| 10277 |
+
},
|
| 10278 |
+
{
|
| 10279 |
+
"epoch": 1.23,
|
| 10280 |
+
"learning_rate": 0.00013555555555555556,
|
| 10281 |
+
"loss": 2.1246,
|
| 10282 |
+
"step": 1695
|
| 10283 |
+
},
|
| 10284 |
+
{
|
| 10285 |
+
"epoch": 1.23,
|
| 10286 |
+
"learning_rate": 0.00013511111111111113,
|
| 10287 |
+
"loss": 2.0802,
|
| 10288 |
+
"step": 1696
|
| 10289 |
+
},
|
| 10290 |
+
{
|
| 10291 |
+
"epoch": 1.24,
|
| 10292 |
+
"learning_rate": 0.00013466666666666667,
|
| 10293 |
+
"loss": 2.1442,
|
| 10294 |
+
"step": 1697
|
| 10295 |
+
},
|
| 10296 |
+
{
|
| 10297 |
+
"epoch": 1.24,
|
| 10298 |
+
"learning_rate": 0.00013422222222222224,
|
| 10299 |
+
"loss": 2.1373,
|
| 10300 |
+
"step": 1698
|
| 10301 |
+
},
|
| 10302 |
+
{
|
| 10303 |
+
"epoch": 1.24,
|
| 10304 |
+
"learning_rate": 0.00013377777777777777,
|
| 10305 |
+
"loss": 2.141,
|
| 10306 |
+
"step": 1699
|
| 10307 |
+
},
|
| 10308 |
+
{
|
| 10309 |
+
"epoch": 1.24,
|
| 10310 |
+
"learning_rate": 0.00013333333333333334,
|
| 10311 |
+
"loss": 2.0219,
|
| 10312 |
+
"step": 1700
|
| 10313 |
+
},
|
| 10314 |
+
{
|
| 10315 |
+
"epoch": 1.24,
|
| 10316 |
+
"learning_rate": 0.00013288888888888888,
|
| 10317 |
+
"loss": 2.0535,
|
| 10318 |
+
"step": 1701
|
| 10319 |
+
},
|
| 10320 |
+
{
|
| 10321 |
+
"epoch": 1.24,
|
| 10322 |
+
"learning_rate": 0.00013244444444444447,
|
| 10323 |
+
"loss": 2.1072,
|
| 10324 |
+
"step": 1702
|
| 10325 |
+
},
|
| 10326 |
+
{
|
| 10327 |
+
"epoch": 1.24,
|
| 10328 |
+
"learning_rate": 0.000132,
|
| 10329 |
+
"loss": 2.174,
|
| 10330 |
+
"step": 1703
|
| 10331 |
+
},
|
| 10332 |
+
{
|
| 10333 |
+
"epoch": 1.24,
|
| 10334 |
+
"learning_rate": 0.00013155555555555558,
|
| 10335 |
+
"loss": 2.0816,
|
| 10336 |
+
"step": 1704
|
| 10337 |
+
},
|
| 10338 |
+
{
|
| 10339 |
+
"epoch": 1.24,
|
| 10340 |
+
"learning_rate": 0.00013111111111111111,
|
| 10341 |
+
"loss": 2.0743,
|
| 10342 |
+
"step": 1705
|
| 10343 |
+
},
|
| 10344 |
+
{
|
| 10345 |
+
"epoch": 1.24,
|
| 10346 |
+
"learning_rate": 0.00013066666666666668,
|
| 10347 |
+
"loss": 2.1471,
|
| 10348 |
+
"step": 1706
|
| 10349 |
+
},
|
| 10350 |
+
{
|
| 10351 |
+
"epoch": 1.24,
|
| 10352 |
+
"learning_rate": 0.00013022222222222222,
|
| 10353 |
+
"loss": 2.1651,
|
| 10354 |
+
"step": 1707
|
| 10355 |
+
},
|
| 10356 |
+
{
|
| 10357 |
+
"epoch": 1.24,
|
| 10358 |
+
"learning_rate": 0.00012977777777777779,
|
| 10359 |
+
"loss": 2.1085,
|
| 10360 |
+
"step": 1708
|
| 10361 |
+
},
|
| 10362 |
+
{
|
| 10363 |
+
"epoch": 1.24,
|
| 10364 |
+
"learning_rate": 0.00012933333333333332,
|
| 10365 |
+
"loss": 2.1157,
|
| 10366 |
+
"step": 1709
|
| 10367 |
+
},
|
| 10368 |
+
{
|
| 10369 |
+
"epoch": 1.24,
|
| 10370 |
+
"learning_rate": 0.00012888888888888892,
|
| 10371 |
+
"loss": 2.0453,
|
| 10372 |
+
"step": 1710
|
| 10373 |
+
},
|
| 10374 |
+
{
|
| 10375 |
+
"epoch": 1.24,
|
| 10376 |
+
"learning_rate": 0.00012844444444444446,
|
| 10377 |
+
"loss": 2.0658,
|
| 10378 |
+
"step": 1711
|
| 10379 |
+
},
|
| 10380 |
+
{
|
| 10381 |
+
"epoch": 1.24,
|
| 10382 |
+
"learning_rate": 0.00012800000000000002,
|
| 10383 |
+
"loss": 2.1744,
|
| 10384 |
+
"step": 1712
|
| 10385 |
+
},
|
| 10386 |
+
{
|
| 10387 |
+
"epoch": 1.24,
|
| 10388 |
+
"learning_rate": 0.00012755555555555556,
|
| 10389 |
+
"loss": 2.1362,
|
| 10390 |
+
"step": 1713
|
| 10391 |
+
},
|
| 10392 |
+
{
|
| 10393 |
+
"epoch": 1.24,
|
| 10394 |
+
"learning_rate": 0.00012711111111111113,
|
| 10395 |
+
"loss": 2.1377,
|
| 10396 |
+
"step": 1714
|
| 10397 |
+
},
|
| 10398 |
+
{
|
| 10399 |
+
"epoch": 1.24,
|
| 10400 |
+
"learning_rate": 0.00012666666666666666,
|
| 10401 |
+
"loss": 2.1357,
|
| 10402 |
+
"step": 1715
|
| 10403 |
+
},
|
| 10404 |
+
{
|
| 10405 |
+
"epoch": 1.24,
|
| 10406 |
+
"learning_rate": 0.00012622222222222223,
|
| 10407 |
+
"loss": 2.1446,
|
| 10408 |
+
"step": 1716
|
| 10409 |
+
},
|
| 10410 |
+
{
|
| 10411 |
+
"epoch": 1.25,
|
| 10412 |
+
"learning_rate": 0.0001257777777777778,
|
| 10413 |
+
"loss": 2.1889,
|
| 10414 |
+
"step": 1717
|
| 10415 |
+
},
|
| 10416 |
+
{
|
| 10417 |
+
"epoch": 1.25,
|
| 10418 |
+
"learning_rate": 0.00012533333333333334,
|
| 10419 |
+
"loss": 2.1327,
|
| 10420 |
+
"step": 1718
|
| 10421 |
+
},
|
| 10422 |
+
{
|
| 10423 |
+
"epoch": 1.25,
|
| 10424 |
+
"learning_rate": 0.0001248888888888889,
|
| 10425 |
+
"loss": 2.1451,
|
| 10426 |
+
"step": 1719
|
| 10427 |
+
},
|
| 10428 |
+
{
|
| 10429 |
+
"epoch": 1.25,
|
| 10430 |
+
"learning_rate": 0.00012444444444444444,
|
| 10431 |
+
"loss": 2.1374,
|
| 10432 |
+
"step": 1720
|
| 10433 |
+
},
|
| 10434 |
+
{
|
| 10435 |
+
"epoch": 1.25,
|
| 10436 |
+
"learning_rate": 0.000124,
|
| 10437 |
+
"loss": 2.1527,
|
| 10438 |
+
"step": 1721
|
| 10439 |
+
},
|
| 10440 |
+
{
|
| 10441 |
+
"epoch": 1.25,
|
| 10442 |
+
"learning_rate": 0.00012355555555555557,
|
| 10443 |
+
"loss": 2.1332,
|
| 10444 |
+
"step": 1722
|
| 10445 |
+
},
|
| 10446 |
+
{
|
| 10447 |
+
"epoch": 1.25,
|
| 10448 |
+
"learning_rate": 0.0001231111111111111,
|
| 10449 |
+
"loss": 2.1022,
|
| 10450 |
+
"step": 1723
|
| 10451 |
+
},
|
| 10452 |
+
{
|
| 10453 |
+
"epoch": 1.25,
|
| 10454 |
+
"learning_rate": 0.00012266666666666668,
|
| 10455 |
+
"loss": 2.1166,
|
| 10456 |
+
"step": 1724
|
| 10457 |
+
},
|
| 10458 |
+
{
|
| 10459 |
+
"epoch": 1.25,
|
| 10460 |
+
"learning_rate": 0.00012222222222222224,
|
| 10461 |
+
"loss": 2.1562,
|
| 10462 |
+
"step": 1725
|
| 10463 |
+
},
|
| 10464 |
+
{
|
| 10465 |
+
"epoch": 1.25,
|
| 10466 |
+
"learning_rate": 0.0001217777777777778,
|
| 10467 |
+
"loss": 2.1021,
|
| 10468 |
+
"step": 1726
|
| 10469 |
+
},
|
| 10470 |
+
{
|
| 10471 |
+
"epoch": 1.25,
|
| 10472 |
+
"learning_rate": 0.00012133333333333335,
|
| 10473 |
+
"loss": 2.1367,
|
| 10474 |
+
"step": 1727
|
| 10475 |
+
},
|
| 10476 |
+
{
|
| 10477 |
+
"epoch": 1.25,
|
| 10478 |
+
"learning_rate": 0.0001208888888888889,
|
| 10479 |
+
"loss": 2.1476,
|
| 10480 |
+
"step": 1728
|
| 10481 |
+
},
|
| 10482 |
+
{
|
| 10483 |
+
"epoch": 1.25,
|
| 10484 |
+
"learning_rate": 0.00012044444444444445,
|
| 10485 |
+
"loss": 2.1072,
|
| 10486 |
+
"step": 1729
|
| 10487 |
+
},
|
| 10488 |
+
{
|
| 10489 |
+
"epoch": 1.25,
|
| 10490 |
+
"learning_rate": 0.00012,
|
| 10491 |
+
"loss": 2.1132,
|
| 10492 |
+
"step": 1730
|
| 10493 |
+
},
|
| 10494 |
+
{
|
| 10495 |
+
"epoch": 1.25,
|
| 10496 |
+
"learning_rate": 0.00011955555555555556,
|
| 10497 |
+
"loss": 2.0795,
|
| 10498 |
+
"step": 1731
|
| 10499 |
+
},
|
| 10500 |
+
{
|
| 10501 |
+
"epoch": 1.25,
|
| 10502 |
+
"learning_rate": 0.00011911111111111111,
|
| 10503 |
+
"loss": 2.0882,
|
| 10504 |
+
"step": 1732
|
| 10505 |
+
},
|
| 10506 |
+
{
|
| 10507 |
+
"epoch": 1.25,
|
| 10508 |
+
"learning_rate": 0.00011866666666666669,
|
| 10509 |
+
"loss": 2.1346,
|
| 10510 |
+
"step": 1733
|
| 10511 |
+
},
|
| 10512 |
+
{
|
| 10513 |
+
"epoch": 1.25,
|
| 10514 |
+
"learning_rate": 0.00011822222222222224,
|
| 10515 |
+
"loss": 2.0931,
|
| 10516 |
+
"step": 1734
|
| 10517 |
+
},
|
| 10518 |
+
{
|
| 10519 |
+
"epoch": 1.25,
|
| 10520 |
+
"learning_rate": 0.00011777777777777779,
|
| 10521 |
+
"loss": 2.0273,
|
| 10522 |
+
"step": 1735
|
| 10523 |
+
},
|
| 10524 |
+
{
|
| 10525 |
+
"epoch": 1.25,
|
| 10526 |
+
"learning_rate": 0.00011733333333333334,
|
| 10527 |
+
"loss": 2.0873,
|
| 10528 |
+
"step": 1736
|
| 10529 |
+
},
|
| 10530 |
+
{
|
| 10531 |
+
"epoch": 1.26,
|
| 10532 |
+
"learning_rate": 0.0001168888888888889,
|
| 10533 |
+
"loss": 2.0911,
|
| 10534 |
+
"step": 1737
|
| 10535 |
+
},
|
| 10536 |
+
{
|
| 10537 |
+
"epoch": 1.26,
|
| 10538 |
+
"learning_rate": 0.00011644444444444445,
|
| 10539 |
+
"loss": 2.1585,
|
| 10540 |
+
"step": 1738
|
| 10541 |
+
},
|
| 10542 |
+
{
|
| 10543 |
+
"epoch": 1.26,
|
| 10544 |
+
"learning_rate": 0.000116,
|
| 10545 |
+
"loss": 2.0838,
|
| 10546 |
+
"step": 1739
|
| 10547 |
+
},
|
| 10548 |
+
{
|
| 10549 |
+
"epoch": 1.26,
|
| 10550 |
+
"learning_rate": 0.00011555555555555555,
|
| 10551 |
+
"loss": 2.0459,
|
| 10552 |
+
"step": 1740
|
| 10553 |
+
},
|
| 10554 |
+
{
|
| 10555 |
+
"epoch": 1.26,
|
| 10556 |
+
"learning_rate": 0.00011511111111111112,
|
| 10557 |
+
"loss": 2.1807,
|
| 10558 |
+
"step": 1741
|
| 10559 |
+
},
|
| 10560 |
+
{
|
| 10561 |
+
"epoch": 1.26,
|
| 10562 |
+
"learning_rate": 0.00011466666666666667,
|
| 10563 |
+
"loss": 2.0983,
|
| 10564 |
+
"step": 1742
|
| 10565 |
+
},
|
| 10566 |
+
{
|
| 10567 |
+
"epoch": 1.26,
|
| 10568 |
+
"learning_rate": 0.00011422222222222224,
|
| 10569 |
+
"loss": 2.0333,
|
| 10570 |
+
"step": 1743
|
| 10571 |
+
},
|
| 10572 |
+
{
|
| 10573 |
+
"epoch": 1.26,
|
| 10574 |
+
"learning_rate": 0.00011377777777777779,
|
| 10575 |
+
"loss": 2.0963,
|
| 10576 |
+
"step": 1744
|
| 10577 |
+
},
|
| 10578 |
+
{
|
| 10579 |
+
"epoch": 1.26,
|
| 10580 |
+
"learning_rate": 0.00011333333333333334,
|
| 10581 |
+
"loss": 2.0992,
|
| 10582 |
+
"step": 1745
|
| 10583 |
+
},
|
| 10584 |
+
{
|
| 10585 |
+
"epoch": 1.26,
|
| 10586 |
+
"learning_rate": 0.0001128888888888889,
|
| 10587 |
+
"loss": 2.0909,
|
| 10588 |
+
"step": 1746
|
| 10589 |
+
},
|
| 10590 |
+
{
|
| 10591 |
+
"epoch": 1.26,
|
| 10592 |
+
"learning_rate": 0.00011244444444444445,
|
| 10593 |
+
"loss": 2.1078,
|
| 10594 |
+
"step": 1747
|
| 10595 |
+
},
|
| 10596 |
+
{
|
| 10597 |
+
"epoch": 1.26,
|
| 10598 |
+
"learning_rate": 0.00011200000000000001,
|
| 10599 |
+
"loss": 2.0909,
|
| 10600 |
+
"step": 1748
|
| 10601 |
+
},
|
| 10602 |
+
{
|
| 10603 |
+
"epoch": 1.26,
|
| 10604 |
+
"learning_rate": 0.00011155555555555556,
|
| 10605 |
+
"loss": 2.1248,
|
| 10606 |
+
"step": 1749
|
| 10607 |
+
},
|
| 10608 |
+
{
|
| 10609 |
+
"epoch": 1.26,
|
| 10610 |
+
"learning_rate": 0.00011111111111111112,
|
| 10611 |
+
"loss": 1.9917,
|
| 10612 |
+
"step": 1750
|
| 10613 |
+
},
|
| 10614 |
+
{
|
| 10615 |
+
"epoch": 1.26,
|
| 10616 |
+
"learning_rate": 0.00011066666666666667,
|
| 10617 |
+
"loss": 2.046,
|
| 10618 |
+
"step": 1751
|
| 10619 |
+
},
|
| 10620 |
+
{
|
| 10621 |
+
"epoch": 1.26,
|
| 10622 |
+
"learning_rate": 0.00011022222222222222,
|
| 10623 |
+
"loss": 2.0792,
|
| 10624 |
+
"step": 1752
|
| 10625 |
+
},
|
| 10626 |
+
{
|
| 10627 |
+
"epoch": 1.26,
|
| 10628 |
+
"learning_rate": 0.00010977777777777777,
|
| 10629 |
+
"loss": 2.2136,
|
| 10630 |
+
"step": 1753
|
| 10631 |
+
},
|
| 10632 |
+
{
|
| 10633 |
+
"epoch": 1.26,
|
| 10634 |
+
"learning_rate": 0.00010933333333333333,
|
| 10635 |
+
"loss": 2.1238,
|
| 10636 |
+
"step": 1754
|
| 10637 |
+
},
|
| 10638 |
+
{
|
| 10639 |
+
"epoch": 1.26,
|
| 10640 |
+
"learning_rate": 0.00010888888888888889,
|
| 10641 |
+
"loss": 2.1477,
|
| 10642 |
+
"step": 1755
|
| 10643 |
+
},
|
| 10644 |
+
{
|
| 10645 |
+
"epoch": 1.26,
|
| 10646 |
+
"learning_rate": 0.00010844444444444446,
|
| 10647 |
+
"loss": 2.0884,
|
| 10648 |
+
"step": 1756
|
| 10649 |
+
},
|
| 10650 |
+
{
|
| 10651 |
+
"epoch": 1.27,
|
| 10652 |
+
"learning_rate": 0.00010800000000000001,
|
| 10653 |
+
"loss": 2.1012,
|
| 10654 |
+
"step": 1757
|
| 10655 |
+
},
|
| 10656 |
+
{
|
| 10657 |
+
"epoch": 1.27,
|
| 10658 |
+
"learning_rate": 0.00010755555555555556,
|
| 10659 |
+
"loss": 2.1251,
|
| 10660 |
+
"step": 1758
|
| 10661 |
+
},
|
| 10662 |
+
{
|
| 10663 |
+
"epoch": 1.27,
|
| 10664 |
+
"learning_rate": 0.00010711111111111111,
|
| 10665 |
+
"loss": 2.1066,
|
| 10666 |
+
"step": 1759
|
| 10667 |
+
},
|
| 10668 |
+
{
|
| 10669 |
+
"epoch": 1.27,
|
| 10670 |
+
"learning_rate": 0.00010666666666666667,
|
| 10671 |
+
"loss": 2.2252,
|
| 10672 |
+
"step": 1760
|
| 10673 |
+
},
|
| 10674 |
+
{
|
| 10675 |
+
"epoch": 1.27,
|
| 10676 |
+
"learning_rate": 0.00010622222222222222,
|
| 10677 |
+
"loss": 2.1278,
|
| 10678 |
+
"step": 1761
|
| 10679 |
+
},
|
| 10680 |
+
{
|
| 10681 |
+
"epoch": 1.27,
|
| 10682 |
+
"learning_rate": 0.00010577777777777777,
|
| 10683 |
+
"loss": 2.1932,
|
| 10684 |
+
"step": 1762
|
| 10685 |
+
},
|
| 10686 |
+
{
|
| 10687 |
+
"epoch": 1.27,
|
| 10688 |
+
"learning_rate": 0.00010533333333333332,
|
| 10689 |
+
"loss": 2.1579,
|
| 10690 |
+
"step": 1763
|
| 10691 |
+
},
|
| 10692 |
+
{
|
| 10693 |
+
"epoch": 1.27,
|
| 10694 |
+
"learning_rate": 0.0001048888888888889,
|
| 10695 |
+
"loss": 2.1337,
|
| 10696 |
+
"step": 1764
|
| 10697 |
+
},
|
| 10698 |
+
{
|
| 10699 |
+
"epoch": 1.27,
|
| 10700 |
+
"learning_rate": 0.00010444444444444445,
|
| 10701 |
+
"loss": 2.1615,
|
| 10702 |
+
"step": 1765
|
| 10703 |
+
},
|
| 10704 |
+
{
|
| 10705 |
+
"epoch": 1.27,
|
| 10706 |
+
"learning_rate": 0.00010400000000000001,
|
| 10707 |
+
"loss": 2.1607,
|
| 10708 |
+
"step": 1766
|
| 10709 |
+
},
|
| 10710 |
+
{
|
| 10711 |
+
"epoch": 1.27,
|
| 10712 |
+
"learning_rate": 0.00010355555555555556,
|
| 10713 |
+
"loss": 2.1803,
|
| 10714 |
+
"step": 1767
|
| 10715 |
+
},
|
| 10716 |
+
{
|
| 10717 |
+
"epoch": 1.27,
|
| 10718 |
+
"learning_rate": 0.00010311111111111111,
|
| 10719 |
+
"loss": 2.0833,
|
| 10720 |
+
"step": 1768
|
| 10721 |
+
},
|
| 10722 |
+
{
|
| 10723 |
+
"epoch": 1.27,
|
| 10724 |
+
"learning_rate": 0.00010266666666666666,
|
| 10725 |
+
"loss": 2.1253,
|
| 10726 |
+
"step": 1769
|
| 10727 |
+
},
|
| 10728 |
+
{
|
| 10729 |
+
"epoch": 1.27,
|
| 10730 |
+
"learning_rate": 0.00010222222222222222,
|
| 10731 |
+
"loss": 2.1848,
|
| 10732 |
+
"step": 1770
|
| 10733 |
+
},
|
| 10734 |
+
{
|
| 10735 |
+
"epoch": 1.27,
|
| 10736 |
+
"learning_rate": 0.00010177777777777777,
|
| 10737 |
+
"loss": 2.0917,
|
| 10738 |
+
"step": 1771
|
| 10739 |
+
},
|
| 10740 |
+
{
|
| 10741 |
+
"epoch": 1.27,
|
| 10742 |
+
"learning_rate": 0.00010133333333333335,
|
| 10743 |
+
"loss": 2.1113,
|
| 10744 |
+
"step": 1772
|
| 10745 |
+
},
|
| 10746 |
+
{
|
| 10747 |
+
"epoch": 1.27,
|
| 10748 |
+
"learning_rate": 0.0001008888888888889,
|
| 10749 |
+
"loss": 2.1489,
|
| 10750 |
+
"step": 1773
|
| 10751 |
+
},
|
| 10752 |
+
{
|
| 10753 |
+
"epoch": 1.27,
|
| 10754 |
+
"learning_rate": 0.00010044444444444445,
|
| 10755 |
+
"loss": 2.182,
|
| 10756 |
+
"step": 1774
|
| 10757 |
+
},
|
| 10758 |
+
{
|
| 10759 |
+
"epoch": 1.27,
|
| 10760 |
+
"learning_rate": 0.0001,
|
| 10761 |
+
"loss": 2.103,
|
| 10762 |
+
"step": 1775
|
| 10763 |
+
},
|
| 10764 |
+
{
|
| 10765 |
+
"epoch": 1.27,
|
| 10766 |
+
"learning_rate": 9.955555555555556e-05,
|
| 10767 |
+
"loss": 2.1152,
|
| 10768 |
+
"step": 1776
|
| 10769 |
+
},
|
| 10770 |
+
{
|
| 10771 |
+
"epoch": 1.28,
|
| 10772 |
+
"learning_rate": 9.911111111111112e-05,
|
| 10773 |
+
"loss": 2.0967,
|
| 10774 |
+
"step": 1777
|
| 10775 |
+
},
|
| 10776 |
+
{
|
| 10777 |
+
"epoch": 1.28,
|
| 10778 |
+
"learning_rate": 9.866666666666668e-05,
|
| 10779 |
+
"loss": 2.097,
|
| 10780 |
+
"step": 1778
|
| 10781 |
+
},
|
| 10782 |
+
{
|
| 10783 |
+
"epoch": 1.28,
|
| 10784 |
+
"learning_rate": 9.822222222222223e-05,
|
| 10785 |
+
"loss": 2.1097,
|
| 10786 |
+
"step": 1779
|
| 10787 |
+
},
|
| 10788 |
+
{
|
| 10789 |
+
"epoch": 1.28,
|
| 10790 |
+
"learning_rate": 9.777777777777778e-05,
|
| 10791 |
+
"loss": 2.1377,
|
| 10792 |
+
"step": 1780
|
| 10793 |
+
},
|
| 10794 |
+
{
|
| 10795 |
+
"epoch": 1.28,
|
| 10796 |
+
"learning_rate": 9.733333333333335e-05,
|
| 10797 |
+
"loss": 2.0993,
|
| 10798 |
+
"step": 1781
|
| 10799 |
+
},
|
| 10800 |
+
{
|
| 10801 |
+
"epoch": 1.28,
|
| 10802 |
+
"learning_rate": 9.68888888888889e-05,
|
| 10803 |
+
"loss": 2.1712,
|
| 10804 |
+
"step": 1782
|
| 10805 |
+
},
|
| 10806 |
+
{
|
| 10807 |
+
"epoch": 1.28,
|
| 10808 |
+
"learning_rate": 9.644444444444445e-05,
|
| 10809 |
+
"loss": 2.1424,
|
| 10810 |
+
"step": 1783
|
| 10811 |
+
},
|
| 10812 |
+
{
|
| 10813 |
+
"epoch": 1.28,
|
| 10814 |
+
"learning_rate": 9.6e-05,
|
| 10815 |
+
"loss": 2.1878,
|
| 10816 |
+
"step": 1784
|
| 10817 |
+
},
|
| 10818 |
+
{
|
| 10819 |
+
"epoch": 1.28,
|
| 10820 |
+
"learning_rate": 9.555555555555557e-05,
|
| 10821 |
+
"loss": 2.1306,
|
| 10822 |
+
"step": 1785
|
| 10823 |
+
},
|
| 10824 |
+
{
|
| 10825 |
+
"epoch": 1.28,
|
| 10826 |
+
"learning_rate": 9.511111111111112e-05,
|
| 10827 |
+
"loss": 2.0727,
|
| 10828 |
+
"step": 1786
|
| 10829 |
+
},
|
| 10830 |
+
{
|
| 10831 |
+
"epoch": 1.28,
|
| 10832 |
+
"learning_rate": 9.466666666666667e-05,
|
| 10833 |
+
"loss": 2.1407,
|
| 10834 |
+
"step": 1787
|
| 10835 |
+
},
|
| 10836 |
+
{
|
| 10837 |
+
"epoch": 1.28,
|
| 10838 |
+
"learning_rate": 9.422222222222223e-05,
|
| 10839 |
+
"loss": 2.0953,
|
| 10840 |
+
"step": 1788
|
| 10841 |
+
},
|
| 10842 |
+
{
|
| 10843 |
+
"epoch": 1.28,
|
| 10844 |
+
"learning_rate": 9.377777777777779e-05,
|
| 10845 |
+
"loss": 2.1386,
|
| 10846 |
+
"step": 1789
|
| 10847 |
+
},
|
| 10848 |
+
{
|
| 10849 |
+
"epoch": 1.28,
|
| 10850 |
+
"learning_rate": 9.333333333333334e-05,
|
| 10851 |
+
"loss": 2.0762,
|
| 10852 |
+
"step": 1790
|
| 10853 |
+
},
|
| 10854 |
+
{
|
| 10855 |
+
"epoch": 1.28,
|
| 10856 |
+
"learning_rate": 9.28888888888889e-05,
|
| 10857 |
+
"loss": 2.1777,
|
| 10858 |
+
"step": 1791
|
| 10859 |
+
},
|
| 10860 |
+
{
|
| 10861 |
+
"epoch": 1.28,
|
| 10862 |
+
"learning_rate": 9.244444444444445e-05,
|
| 10863 |
+
"loss": 2.1934,
|
| 10864 |
+
"step": 1792
|
| 10865 |
+
},
|
| 10866 |
+
{
|
| 10867 |
+
"epoch": 1.28,
|
| 10868 |
+
"learning_rate": 9.200000000000001e-05,
|
| 10869 |
+
"loss": 2.0668,
|
| 10870 |
+
"step": 1793
|
| 10871 |
+
},
|
| 10872 |
+
{
|
| 10873 |
+
"epoch": 1.28,
|
| 10874 |
+
"learning_rate": 9.155555555555557e-05,
|
| 10875 |
+
"loss": 2.1243,
|
| 10876 |
+
"step": 1794
|
| 10877 |
+
},
|
| 10878 |
+
{
|
| 10879 |
+
"epoch": 1.28,
|
| 10880 |
+
"learning_rate": 9.111111111111112e-05,
|
| 10881 |
+
"loss": 2.0199,
|
| 10882 |
+
"step": 1795
|
| 10883 |
+
},
|
| 10884 |
+
{
|
| 10885 |
+
"epoch": 1.28,
|
| 10886 |
+
"learning_rate": 9.066666666666667e-05,
|
| 10887 |
+
"loss": 2.0717,
|
| 10888 |
+
"step": 1796
|
| 10889 |
+
},
|
| 10890 |
+
{
|
| 10891 |
+
"epoch": 1.29,
|
| 10892 |
+
"learning_rate": 9.022222222222224e-05,
|
| 10893 |
+
"loss": 2.1475,
|
| 10894 |
+
"step": 1797
|
| 10895 |
+
},
|
| 10896 |
+
{
|
| 10897 |
+
"epoch": 1.29,
|
| 10898 |
+
"learning_rate": 8.977777777777779e-05,
|
| 10899 |
+
"loss": 2.0687,
|
| 10900 |
+
"step": 1798
|
| 10901 |
+
},
|
| 10902 |
+
{
|
| 10903 |
+
"epoch": 1.29,
|
| 10904 |
+
"learning_rate": 8.933333333333334e-05,
|
| 10905 |
+
"loss": 2.0979,
|
| 10906 |
+
"step": 1799
|
| 10907 |
+
},
|
| 10908 |
+
{
|
| 10909 |
+
"epoch": 1.29,
|
| 10910 |
+
"learning_rate": 8.888888888888889e-05,
|
| 10911 |
+
"loss": 2.105,
|
| 10912 |
+
"step": 1800
|
| 10913 |
+
},
|
| 10914 |
+
{
|
| 10915 |
+
"epoch": 1.29,
|
| 10916 |
+
"eval_gen_len": 1023.0,
|
| 10917 |
+
"eval_loss": 2.099489688873291,
|
| 10918 |
+
"eval_rouge1": 13.9638,
|
| 10919 |
+
"eval_rouge2": 4.7209,
|
| 10920 |
+
"eval_rougeL": 8.7889,
|
| 10921 |
+
"eval_rougeLsum": 8.7778,
|
| 10922 |
+
"eval_runtime": 9040.193,
|
| 10923 |
+
"eval_samples_per_second": 0.055,
|
| 10924 |
+
"eval_steps_per_second": 0.055,
|
| 10925 |
+
"step": 1800
|
| 10926 |
}
|
| 10927 |
],
|
| 10928 |
"max_steps": 2000,
|
| 10929 |
"num_train_epochs": 9223372036854775807,
|
| 10930 |
+
"total_flos": 1.4463981997036646e+18,
|
| 10931 |
"trial_name": null,
|
| 10932 |
"trial_params": null
|
| 10933 |
}
|
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:f420a02b8b31a90067fa5983f6a9bf2cf10ec0e98b96f411b86d4b5b802edba4
|
| 3 |
size 2368281769
|