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 +1213 -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 sha256:052f70623aa8868a83cba6c72bff784cd35f1627f7830c96a8037c50efd03161
|
| 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:e0581c601cbb195614ebff9150719cfc4b2d8be4e3802ef0fd4f340d4b33ab78
|
| 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:f2bf5cd3985dc56adcf7cef3650c97c2e1e8cbdf593411958cc282a06a916078
|
| 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:e94ec64346c3eafc3d12675fe356be8b89d123bd9cf0ee96e17680b54394d6f7
|
| 3 |
size 627
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric":
|
| 3 |
-
"best_model_checkpoint": "output/checkpoint-
|
| 4 |
-
"epoch": 1.
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
@@ -9670,11 +9670,1219 @@
|
|
| 9670 |
"eval_samples_per_second": 6.083,
|
| 9671 |
"eval_steps_per_second": 6.083,
|
| 9672 |
"step": 1600
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9673 |
}
|
| 9674 |
],
|
| 9675 |
"max_steps": 4386,
|
| 9676 |
"num_train_epochs": 3,
|
| 9677 |
-
"total_flos":
|
| 9678 |
"trial_name": null,
|
| 9679 |
"trial_params": null
|
| 9680 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 1.9992964267730713,
|
| 3 |
+
"best_model_checkpoint": "output/checkpoint-1800",
|
| 4 |
+
"epoch": 1.2306311791946416,
|
| 5 |
+
"global_step": 1800,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 9670 |
"eval_samples_per_second": 6.083,
|
| 9671 |
"eval_steps_per_second": 6.083,
|
| 9672 |
"step": 1600
|
| 9673 |
+
},
|
| 9674 |
+
{
|
| 9675 |
+
"epoch": 1.09,
|
| 9676 |
+
"learning_rate": 0.0003991877687529861,
|
| 9677 |
+
"loss": 1.9763,
|
| 9678 |
+
"step": 1601
|
| 9679 |
+
},
|
| 9680 |
+
{
|
| 9681 |
+
"epoch": 1.1,
|
| 9682 |
+
"learning_rate": 0.0003990444338270425,
|
| 9683 |
+
"loss": 2.0162,
|
| 9684 |
+
"step": 1602
|
| 9685 |
+
},
|
| 9686 |
+
{
|
| 9687 |
+
"epoch": 1.1,
|
| 9688 |
+
"learning_rate": 0.00039890109890109885,
|
| 9689 |
+
"loss": 1.8628,
|
| 9690 |
+
"step": 1603
|
| 9691 |
+
},
|
| 9692 |
+
{
|
| 9693 |
+
"epoch": 1.1,
|
| 9694 |
+
"learning_rate": 0.0003987577639751553,
|
| 9695 |
+
"loss": 2.0199,
|
| 9696 |
+
"step": 1604
|
| 9697 |
+
},
|
| 9698 |
+
{
|
| 9699 |
+
"epoch": 1.1,
|
| 9700 |
+
"learning_rate": 0.0003986144290492116,
|
| 9701 |
+
"loss": 1.9488,
|
| 9702 |
+
"step": 1605
|
| 9703 |
+
},
|
| 9704 |
+
{
|
| 9705 |
+
"epoch": 1.1,
|
| 9706 |
+
"learning_rate": 0.000398471094123268,
|
| 9707 |
+
"loss": 2.0393,
|
| 9708 |
+
"step": 1606
|
| 9709 |
+
},
|
| 9710 |
+
{
|
| 9711 |
+
"epoch": 1.1,
|
| 9712 |
+
"learning_rate": 0.0003983277591973244,
|
| 9713 |
+
"loss": 1.9423,
|
| 9714 |
+
"step": 1607
|
| 9715 |
+
},
|
| 9716 |
+
{
|
| 9717 |
+
"epoch": 1.1,
|
| 9718 |
+
"learning_rate": 0.0003981844242713807,
|
| 9719 |
+
"loss": 2.1568,
|
| 9720 |
+
"step": 1608
|
| 9721 |
+
},
|
| 9722 |
+
{
|
| 9723 |
+
"epoch": 1.1,
|
| 9724 |
+
"learning_rate": 0.00039804108934543716,
|
| 9725 |
+
"loss": 2.0384,
|
| 9726 |
+
"step": 1609
|
| 9727 |
+
},
|
| 9728 |
+
{
|
| 9729 |
+
"epoch": 1.1,
|
| 9730 |
+
"learning_rate": 0.0003978977544194935,
|
| 9731 |
+
"loss": 1.9644,
|
| 9732 |
+
"step": 1610
|
| 9733 |
+
},
|
| 9734 |
+
{
|
| 9735 |
+
"epoch": 1.1,
|
| 9736 |
+
"learning_rate": 0.0003977544194935499,
|
| 9737 |
+
"loss": 2.0689,
|
| 9738 |
+
"step": 1611
|
| 9739 |
+
},
|
| 9740 |
+
{
|
| 9741 |
+
"epoch": 1.1,
|
| 9742 |
+
"learning_rate": 0.00039761108456760625,
|
| 9743 |
+
"loss": 2.0017,
|
| 9744 |
+
"step": 1612
|
| 9745 |
+
},
|
| 9746 |
+
{
|
| 9747 |
+
"epoch": 1.1,
|
| 9748 |
+
"learning_rate": 0.0003974677496416627,
|
| 9749 |
+
"loss": 1.9427,
|
| 9750 |
+
"step": 1613
|
| 9751 |
+
},
|
| 9752 |
+
{
|
| 9753 |
+
"epoch": 1.1,
|
| 9754 |
+
"learning_rate": 0.000397324414715719,
|
| 9755 |
+
"loss": 1.9709,
|
| 9756 |
+
"step": 1614
|
| 9757 |
+
},
|
| 9758 |
+
{
|
| 9759 |
+
"epoch": 1.1,
|
| 9760 |
+
"learning_rate": 0.0003971810797897754,
|
| 9761 |
+
"loss": 2.0033,
|
| 9762 |
+
"step": 1615
|
| 9763 |
+
},
|
| 9764 |
+
{
|
| 9765 |
+
"epoch": 1.1,
|
| 9766 |
+
"learning_rate": 0.0003970377448638318,
|
| 9767 |
+
"loss": 1.9937,
|
| 9768 |
+
"step": 1616
|
| 9769 |
+
},
|
| 9770 |
+
{
|
| 9771 |
+
"epoch": 1.11,
|
| 9772 |
+
"learning_rate": 0.00039689440993788817,
|
| 9773 |
+
"loss": 2.005,
|
| 9774 |
+
"step": 1617
|
| 9775 |
+
},
|
| 9776 |
+
{
|
| 9777 |
+
"epoch": 1.11,
|
| 9778 |
+
"learning_rate": 0.00039675107501194456,
|
| 9779 |
+
"loss": 2.0023,
|
| 9780 |
+
"step": 1618
|
| 9781 |
+
},
|
| 9782 |
+
{
|
| 9783 |
+
"epoch": 1.11,
|
| 9784 |
+
"learning_rate": 0.00039660774008600094,
|
| 9785 |
+
"loss": 2.0233,
|
| 9786 |
+
"step": 1619
|
| 9787 |
+
},
|
| 9788 |
+
{
|
| 9789 |
+
"epoch": 1.11,
|
| 9790 |
+
"learning_rate": 0.0003964644051600573,
|
| 9791 |
+
"loss": 1.962,
|
| 9792 |
+
"step": 1620
|
| 9793 |
+
},
|
| 9794 |
+
{
|
| 9795 |
+
"epoch": 1.11,
|
| 9796 |
+
"learning_rate": 0.0003963210702341137,
|
| 9797 |
+
"loss": 2.026,
|
| 9798 |
+
"step": 1621
|
| 9799 |
+
},
|
| 9800 |
+
{
|
| 9801 |
+
"epoch": 1.11,
|
| 9802 |
+
"learning_rate": 0.00039617773530817004,
|
| 9803 |
+
"loss": 2.0934,
|
| 9804 |
+
"step": 1622
|
| 9805 |
+
},
|
| 9806 |
+
{
|
| 9807 |
+
"epoch": 1.11,
|
| 9808 |
+
"learning_rate": 0.0003960344003822264,
|
| 9809 |
+
"loss": 2.0144,
|
| 9810 |
+
"step": 1623
|
| 9811 |
+
},
|
| 9812 |
+
{
|
| 9813 |
+
"epoch": 1.11,
|
| 9814 |
+
"learning_rate": 0.0003958910654562828,
|
| 9815 |
+
"loss": 1.9919,
|
| 9816 |
+
"step": 1624
|
| 9817 |
+
},
|
| 9818 |
+
{
|
| 9819 |
+
"epoch": 1.11,
|
| 9820 |
+
"learning_rate": 0.0003957477305303392,
|
| 9821 |
+
"loss": 1.9533,
|
| 9822 |
+
"step": 1625
|
| 9823 |
+
},
|
| 9824 |
+
{
|
| 9825 |
+
"epoch": 1.11,
|
| 9826 |
+
"learning_rate": 0.00039560439560439557,
|
| 9827 |
+
"loss": 1.9303,
|
| 9828 |
+
"step": 1626
|
| 9829 |
+
},
|
| 9830 |
+
{
|
| 9831 |
+
"epoch": 1.11,
|
| 9832 |
+
"learning_rate": 0.00039546106067845196,
|
| 9833 |
+
"loss": 1.9299,
|
| 9834 |
+
"step": 1627
|
| 9835 |
+
},
|
| 9836 |
+
{
|
| 9837 |
+
"epoch": 1.11,
|
| 9838 |
+
"learning_rate": 0.00039531772575250834,
|
| 9839 |
+
"loss": 2.0134,
|
| 9840 |
+
"step": 1628
|
| 9841 |
+
},
|
| 9842 |
+
{
|
| 9843 |
+
"epoch": 1.11,
|
| 9844 |
+
"learning_rate": 0.00039517439082656467,
|
| 9845 |
+
"loss": 1.9218,
|
| 9846 |
+
"step": 1629
|
| 9847 |
+
},
|
| 9848 |
+
{
|
| 9849 |
+
"epoch": 1.11,
|
| 9850 |
+
"learning_rate": 0.0003950310559006211,
|
| 9851 |
+
"loss": 2.0074,
|
| 9852 |
+
"step": 1630
|
| 9853 |
+
},
|
| 9854 |
+
{
|
| 9855 |
+
"epoch": 1.12,
|
| 9856 |
+
"learning_rate": 0.00039488772097467744,
|
| 9857 |
+
"loss": 2.1142,
|
| 9858 |
+
"step": 1631
|
| 9859 |
+
},
|
| 9860 |
+
{
|
| 9861 |
+
"epoch": 1.12,
|
| 9862 |
+
"learning_rate": 0.0003947443860487339,
|
| 9863 |
+
"loss": 2.0237,
|
| 9864 |
+
"step": 1632
|
| 9865 |
+
},
|
| 9866 |
+
{
|
| 9867 |
+
"epoch": 1.12,
|
| 9868 |
+
"learning_rate": 0.0003946010511227902,
|
| 9869 |
+
"loss": 2.0208,
|
| 9870 |
+
"step": 1633
|
| 9871 |
+
},
|
| 9872 |
+
{
|
| 9873 |
+
"epoch": 1.12,
|
| 9874 |
+
"learning_rate": 0.00039445771619684664,
|
| 9875 |
+
"loss": 1.9773,
|
| 9876 |
+
"step": 1634
|
| 9877 |
+
},
|
| 9878 |
+
{
|
| 9879 |
+
"epoch": 1.12,
|
| 9880 |
+
"learning_rate": 0.00039431438127090297,
|
| 9881 |
+
"loss": 1.9945,
|
| 9882 |
+
"step": 1635
|
| 9883 |
+
},
|
| 9884 |
+
{
|
| 9885 |
+
"epoch": 1.12,
|
| 9886 |
+
"learning_rate": 0.00039417104634495936,
|
| 9887 |
+
"loss": 1.9521,
|
| 9888 |
+
"step": 1636
|
| 9889 |
+
},
|
| 9890 |
+
{
|
| 9891 |
+
"epoch": 1.12,
|
| 9892 |
+
"learning_rate": 0.00039402771141901574,
|
| 9893 |
+
"loss": 2.0321,
|
| 9894 |
+
"step": 1637
|
| 9895 |
+
},
|
| 9896 |
+
{
|
| 9897 |
+
"epoch": 1.12,
|
| 9898 |
+
"learning_rate": 0.0003938843764930721,
|
| 9899 |
+
"loss": 2.0056,
|
| 9900 |
+
"step": 1638
|
| 9901 |
+
},
|
| 9902 |
+
{
|
| 9903 |
+
"epoch": 1.12,
|
| 9904 |
+
"learning_rate": 0.0003937410415671285,
|
| 9905 |
+
"loss": 2.008,
|
| 9906 |
+
"step": 1639
|
| 9907 |
+
},
|
| 9908 |
+
{
|
| 9909 |
+
"epoch": 1.12,
|
| 9910 |
+
"learning_rate": 0.00039359770664118484,
|
| 9911 |
+
"loss": 2.0351,
|
| 9912 |
+
"step": 1640
|
| 9913 |
+
},
|
| 9914 |
+
{
|
| 9915 |
+
"epoch": 1.12,
|
| 9916 |
+
"learning_rate": 0.0003934543717152413,
|
| 9917 |
+
"loss": 2.0366,
|
| 9918 |
+
"step": 1641
|
| 9919 |
+
},
|
| 9920 |
+
{
|
| 9921 |
+
"epoch": 1.12,
|
| 9922 |
+
"learning_rate": 0.0003933110367892976,
|
| 9923 |
+
"loss": 2.0498,
|
| 9924 |
+
"step": 1642
|
| 9925 |
+
},
|
| 9926 |
+
{
|
| 9927 |
+
"epoch": 1.12,
|
| 9928 |
+
"learning_rate": 0.000393167701863354,
|
| 9929 |
+
"loss": 2.0764,
|
| 9930 |
+
"step": 1643
|
| 9931 |
+
},
|
| 9932 |
+
{
|
| 9933 |
+
"epoch": 1.12,
|
| 9934 |
+
"learning_rate": 0.00039302436693741037,
|
| 9935 |
+
"loss": 1.9468,
|
| 9936 |
+
"step": 1644
|
| 9937 |
+
},
|
| 9938 |
+
{
|
| 9939 |
+
"epoch": 1.12,
|
| 9940 |
+
"learning_rate": 0.00039288103201146675,
|
| 9941 |
+
"loss": 2.0169,
|
| 9942 |
+
"step": 1645
|
| 9943 |
+
},
|
| 9944 |
+
{
|
| 9945 |
+
"epoch": 1.13,
|
| 9946 |
+
"learning_rate": 0.00039273769708552314,
|
| 9947 |
+
"loss": 1.9899,
|
| 9948 |
+
"step": 1646
|
| 9949 |
+
},
|
| 9950 |
+
{
|
| 9951 |
+
"epoch": 1.13,
|
| 9952 |
+
"learning_rate": 0.0003925943621595795,
|
| 9953 |
+
"loss": 2.0945,
|
| 9954 |
+
"step": 1647
|
| 9955 |
+
},
|
| 9956 |
+
{
|
| 9957 |
+
"epoch": 1.13,
|
| 9958 |
+
"learning_rate": 0.00039245102723363585,
|
| 9959 |
+
"loss": 2.055,
|
| 9960 |
+
"step": 1648
|
| 9961 |
+
},
|
| 9962 |
+
{
|
| 9963 |
+
"epoch": 1.13,
|
| 9964 |
+
"learning_rate": 0.0003923076923076923,
|
| 9965 |
+
"loss": 1.9816,
|
| 9966 |
+
"step": 1649
|
| 9967 |
+
},
|
| 9968 |
+
{
|
| 9969 |
+
"epoch": 1.13,
|
| 9970 |
+
"learning_rate": 0.0003921643573817486,
|
| 9971 |
+
"loss": 1.9407,
|
| 9972 |
+
"step": 1650
|
| 9973 |
+
},
|
| 9974 |
+
{
|
| 9975 |
+
"epoch": 1.13,
|
| 9976 |
+
"learning_rate": 0.00039202102245580506,
|
| 9977 |
+
"loss": 2.1146,
|
| 9978 |
+
"step": 1651
|
| 9979 |
+
},
|
| 9980 |
+
{
|
| 9981 |
+
"epoch": 1.13,
|
| 9982 |
+
"learning_rate": 0.0003918776875298614,
|
| 9983 |
+
"loss": 2.096,
|
| 9984 |
+
"step": 1652
|
| 9985 |
+
},
|
| 9986 |
+
{
|
| 9987 |
+
"epoch": 1.13,
|
| 9988 |
+
"learning_rate": 0.0003917343526039178,
|
| 9989 |
+
"loss": 2.0008,
|
| 9990 |
+
"step": 1653
|
| 9991 |
+
},
|
| 9992 |
+
{
|
| 9993 |
+
"epoch": 1.13,
|
| 9994 |
+
"learning_rate": 0.00039159101767797415,
|
| 9995 |
+
"loss": 2.0295,
|
| 9996 |
+
"step": 1654
|
| 9997 |
+
},
|
| 9998 |
+
{
|
| 9999 |
+
"epoch": 1.13,
|
| 10000 |
+
"learning_rate": 0.00039144768275203054,
|
| 10001 |
+
"loss": 1.9967,
|
| 10002 |
+
"step": 1655
|
| 10003 |
+
},
|
| 10004 |
+
{
|
| 10005 |
+
"epoch": 1.13,
|
| 10006 |
+
"learning_rate": 0.0003913043478260869,
|
| 10007 |
+
"loss": 2.0425,
|
| 10008 |
+
"step": 1656
|
| 10009 |
+
},
|
| 10010 |
+
{
|
| 10011 |
+
"epoch": 1.13,
|
| 10012 |
+
"learning_rate": 0.00039116101290014325,
|
| 10013 |
+
"loss": 2.062,
|
| 10014 |
+
"step": 1657
|
| 10015 |
+
},
|
| 10016 |
+
{
|
| 10017 |
+
"epoch": 1.13,
|
| 10018 |
+
"learning_rate": 0.0003910176779741997,
|
| 10019 |
+
"loss": 2.0158,
|
| 10020 |
+
"step": 1658
|
| 10021 |
+
},
|
| 10022 |
+
{
|
| 10023 |
+
"epoch": 1.13,
|
| 10024 |
+
"learning_rate": 0.000390874343048256,
|
| 10025 |
+
"loss": 1.9327,
|
| 10026 |
+
"step": 1659
|
| 10027 |
+
},
|
| 10028 |
+
{
|
| 10029 |
+
"epoch": 1.13,
|
| 10030 |
+
"learning_rate": 0.00039073100812231246,
|
| 10031 |
+
"loss": 2.0324,
|
| 10032 |
+
"step": 1660
|
| 10033 |
+
},
|
| 10034 |
+
{
|
| 10035 |
+
"epoch": 1.14,
|
| 10036 |
+
"learning_rate": 0.0003905876731963688,
|
| 10037 |
+
"loss": 1.9874,
|
| 10038 |
+
"step": 1661
|
| 10039 |
+
},
|
| 10040 |
+
{
|
| 10041 |
+
"epoch": 1.14,
|
| 10042 |
+
"learning_rate": 0.00039044433827042517,
|
| 10043 |
+
"loss": 1.9798,
|
| 10044 |
+
"step": 1662
|
| 10045 |
+
},
|
| 10046 |
+
{
|
| 10047 |
+
"epoch": 1.14,
|
| 10048 |
+
"learning_rate": 0.00039030100334448155,
|
| 10049 |
+
"loss": 1.8957,
|
| 10050 |
+
"step": 1663
|
| 10051 |
+
},
|
| 10052 |
+
{
|
| 10053 |
+
"epoch": 1.14,
|
| 10054 |
+
"learning_rate": 0.00039015766841853794,
|
| 10055 |
+
"loss": 2.0757,
|
| 10056 |
+
"step": 1664
|
| 10057 |
+
},
|
| 10058 |
+
{
|
| 10059 |
+
"epoch": 1.14,
|
| 10060 |
+
"learning_rate": 0.0003900143334925943,
|
| 10061 |
+
"loss": 2.0211,
|
| 10062 |
+
"step": 1665
|
| 10063 |
+
},
|
| 10064 |
+
{
|
| 10065 |
+
"epoch": 1.14,
|
| 10066 |
+
"learning_rate": 0.0003898709985666507,
|
| 10067 |
+
"loss": 2.0094,
|
| 10068 |
+
"step": 1666
|
| 10069 |
+
},
|
| 10070 |
+
{
|
| 10071 |
+
"epoch": 1.14,
|
| 10072 |
+
"learning_rate": 0.0003897276636407071,
|
| 10073 |
+
"loss": 2.0978,
|
| 10074 |
+
"step": 1667
|
| 10075 |
+
},
|
| 10076 |
+
{
|
| 10077 |
+
"epoch": 1.14,
|
| 10078 |
+
"learning_rate": 0.0003895843287147635,
|
| 10079 |
+
"loss": 2.0842,
|
| 10080 |
+
"step": 1668
|
| 10081 |
+
},
|
| 10082 |
+
{
|
| 10083 |
+
"epoch": 1.14,
|
| 10084 |
+
"learning_rate": 0.0003894409937888198,
|
| 10085 |
+
"loss": 2.1968,
|
| 10086 |
+
"step": 1669
|
| 10087 |
+
},
|
| 10088 |
+
{
|
| 10089 |
+
"epoch": 1.14,
|
| 10090 |
+
"learning_rate": 0.00038929765886287624,
|
| 10091 |
+
"loss": 2.0997,
|
| 10092 |
+
"step": 1670
|
| 10093 |
+
},
|
| 10094 |
+
{
|
| 10095 |
+
"epoch": 1.14,
|
| 10096 |
+
"learning_rate": 0.00038915432393693257,
|
| 10097 |
+
"loss": 1.9782,
|
| 10098 |
+
"step": 1671
|
| 10099 |
+
},
|
| 10100 |
+
{
|
| 10101 |
+
"epoch": 1.14,
|
| 10102 |
+
"learning_rate": 0.000389010989010989,
|
| 10103 |
+
"loss": 2.0218,
|
| 10104 |
+
"step": 1672
|
| 10105 |
+
},
|
| 10106 |
+
{
|
| 10107 |
+
"epoch": 1.14,
|
| 10108 |
+
"learning_rate": 0.00038886765408504534,
|
| 10109 |
+
"loss": 2.038,
|
| 10110 |
+
"step": 1673
|
| 10111 |
+
},
|
| 10112 |
+
{
|
| 10113 |
+
"epoch": 1.14,
|
| 10114 |
+
"learning_rate": 0.0003887243191591018,
|
| 10115 |
+
"loss": 2.049,
|
| 10116 |
+
"step": 1674
|
| 10117 |
+
},
|
| 10118 |
+
{
|
| 10119 |
+
"epoch": 1.15,
|
| 10120 |
+
"learning_rate": 0.0003885809842331581,
|
| 10121 |
+
"loss": 2.1634,
|
| 10122 |
+
"step": 1675
|
| 10123 |
+
},
|
| 10124 |
+
{
|
| 10125 |
+
"epoch": 1.15,
|
| 10126 |
+
"learning_rate": 0.00038843764930721444,
|
| 10127 |
+
"loss": 2.0358,
|
| 10128 |
+
"step": 1676
|
| 10129 |
+
},
|
| 10130 |
+
{
|
| 10131 |
+
"epoch": 1.15,
|
| 10132 |
+
"learning_rate": 0.0003882943143812709,
|
| 10133 |
+
"loss": 2.0982,
|
| 10134 |
+
"step": 1677
|
| 10135 |
+
},
|
| 10136 |
+
{
|
| 10137 |
+
"epoch": 1.15,
|
| 10138 |
+
"learning_rate": 0.0003881509794553272,
|
| 10139 |
+
"loss": 2.1026,
|
| 10140 |
+
"step": 1678
|
| 10141 |
+
},
|
| 10142 |
+
{
|
| 10143 |
+
"epoch": 1.15,
|
| 10144 |
+
"learning_rate": 0.00038800764452938364,
|
| 10145 |
+
"loss": 1.9847,
|
| 10146 |
+
"step": 1679
|
| 10147 |
+
},
|
| 10148 |
+
{
|
| 10149 |
+
"epoch": 1.15,
|
| 10150 |
+
"learning_rate": 0.00038786430960343997,
|
| 10151 |
+
"loss": 1.9671,
|
| 10152 |
+
"step": 1680
|
| 10153 |
+
},
|
| 10154 |
+
{
|
| 10155 |
+
"epoch": 1.15,
|
| 10156 |
+
"learning_rate": 0.0003877209746774964,
|
| 10157 |
+
"loss": 2.0934,
|
| 10158 |
+
"step": 1681
|
| 10159 |
+
},
|
| 10160 |
+
{
|
| 10161 |
+
"epoch": 1.15,
|
| 10162 |
+
"learning_rate": 0.00038757763975155274,
|
| 10163 |
+
"loss": 1.9865,
|
| 10164 |
+
"step": 1682
|
| 10165 |
+
},
|
| 10166 |
+
{
|
| 10167 |
+
"epoch": 1.15,
|
| 10168 |
+
"learning_rate": 0.0003874343048256091,
|
| 10169 |
+
"loss": 2.0684,
|
| 10170 |
+
"step": 1683
|
| 10171 |
+
},
|
| 10172 |
+
{
|
| 10173 |
+
"epoch": 1.15,
|
| 10174 |
+
"learning_rate": 0.0003872909698996655,
|
| 10175 |
+
"loss": 2.1101,
|
| 10176 |
+
"step": 1684
|
| 10177 |
+
},
|
| 10178 |
+
{
|
| 10179 |
+
"epoch": 1.15,
|
| 10180 |
+
"learning_rate": 0.0003871476349737219,
|
| 10181 |
+
"loss": 1.9562,
|
| 10182 |
+
"step": 1685
|
| 10183 |
+
},
|
| 10184 |
+
{
|
| 10185 |
+
"epoch": 1.15,
|
| 10186 |
+
"learning_rate": 0.0003870043000477783,
|
| 10187 |
+
"loss": 1.9814,
|
| 10188 |
+
"step": 1686
|
| 10189 |
+
},
|
| 10190 |
+
{
|
| 10191 |
+
"epoch": 1.15,
|
| 10192 |
+
"learning_rate": 0.00038686096512183466,
|
| 10193 |
+
"loss": 2.0065,
|
| 10194 |
+
"step": 1687
|
| 10195 |
+
},
|
| 10196 |
+
{
|
| 10197 |
+
"epoch": 1.15,
|
| 10198 |
+
"learning_rate": 0.00038671763019589104,
|
| 10199 |
+
"loss": 2.1368,
|
| 10200 |
+
"step": 1688
|
| 10201 |
+
},
|
| 10202 |
+
{
|
| 10203 |
+
"epoch": 1.15,
|
| 10204 |
+
"learning_rate": 0.0003865742952699474,
|
| 10205 |
+
"loss": 1.9981,
|
| 10206 |
+
"step": 1689
|
| 10207 |
+
},
|
| 10208 |
+
{
|
| 10209 |
+
"epoch": 1.16,
|
| 10210 |
+
"learning_rate": 0.00038643096034400375,
|
| 10211 |
+
"loss": 1.993,
|
| 10212 |
+
"step": 1690
|
| 10213 |
+
},
|
| 10214 |
+
{
|
| 10215 |
+
"epoch": 1.16,
|
| 10216 |
+
"learning_rate": 0.0003862876254180602,
|
| 10217 |
+
"loss": 2.0458,
|
| 10218 |
+
"step": 1691
|
| 10219 |
+
},
|
| 10220 |
+
{
|
| 10221 |
+
"epoch": 1.16,
|
| 10222 |
+
"learning_rate": 0.0003861442904921165,
|
| 10223 |
+
"loss": 1.9745,
|
| 10224 |
+
"step": 1692
|
| 10225 |
+
},
|
| 10226 |
+
{
|
| 10227 |
+
"epoch": 1.16,
|
| 10228 |
+
"learning_rate": 0.00038600095556617296,
|
| 10229 |
+
"loss": 1.932,
|
| 10230 |
+
"step": 1693
|
| 10231 |
+
},
|
| 10232 |
+
{
|
| 10233 |
+
"epoch": 1.16,
|
| 10234 |
+
"learning_rate": 0.0003858576206402293,
|
| 10235 |
+
"loss": 2.0137,
|
| 10236 |
+
"step": 1694
|
| 10237 |
+
},
|
| 10238 |
+
{
|
| 10239 |
+
"epoch": 1.16,
|
| 10240 |
+
"learning_rate": 0.0003857142857142857,
|
| 10241 |
+
"loss": 2.0512,
|
| 10242 |
+
"step": 1695
|
| 10243 |
+
},
|
| 10244 |
+
{
|
| 10245 |
+
"epoch": 1.16,
|
| 10246 |
+
"learning_rate": 0.00038557095078834206,
|
| 10247 |
+
"loss": 1.9725,
|
| 10248 |
+
"step": 1696
|
| 10249 |
+
},
|
| 10250 |
+
{
|
| 10251 |
+
"epoch": 1.16,
|
| 10252 |
+
"learning_rate": 0.0003854276158623984,
|
| 10253 |
+
"loss": 1.9678,
|
| 10254 |
+
"step": 1697
|
| 10255 |
+
},
|
| 10256 |
+
{
|
| 10257 |
+
"epoch": 1.16,
|
| 10258 |
+
"learning_rate": 0.0003852842809364548,
|
| 10259 |
+
"loss": 1.9604,
|
| 10260 |
+
"step": 1698
|
| 10261 |
+
},
|
| 10262 |
+
{
|
| 10263 |
+
"epoch": 1.16,
|
| 10264 |
+
"learning_rate": 0.00038514094601051115,
|
| 10265 |
+
"loss": 2.0289,
|
| 10266 |
+
"step": 1699
|
| 10267 |
+
},
|
| 10268 |
+
{
|
| 10269 |
+
"epoch": 1.16,
|
| 10270 |
+
"learning_rate": 0.0003849976110845676,
|
| 10271 |
+
"loss": 1.9483,
|
| 10272 |
+
"step": 1700
|
| 10273 |
+
},
|
| 10274 |
+
{
|
| 10275 |
+
"epoch": 1.16,
|
| 10276 |
+
"learning_rate": 0.0003848542761586239,
|
| 10277 |
+
"loss": 2.0639,
|
| 10278 |
+
"step": 1701
|
| 10279 |
+
},
|
| 10280 |
+
{
|
| 10281 |
+
"epoch": 1.16,
|
| 10282 |
+
"learning_rate": 0.00038471094123268036,
|
| 10283 |
+
"loss": 2.0865,
|
| 10284 |
+
"step": 1702
|
| 10285 |
+
},
|
| 10286 |
+
{
|
| 10287 |
+
"epoch": 1.16,
|
| 10288 |
+
"learning_rate": 0.0003845676063067367,
|
| 10289 |
+
"loss": 2.014,
|
| 10290 |
+
"step": 1703
|
| 10291 |
+
},
|
| 10292 |
+
{
|
| 10293 |
+
"epoch": 1.16,
|
| 10294 |
+
"learning_rate": 0.00038442427138079307,
|
| 10295 |
+
"loss": 2.0076,
|
| 10296 |
+
"step": 1704
|
| 10297 |
+
},
|
| 10298 |
+
{
|
| 10299 |
+
"epoch": 1.17,
|
| 10300 |
+
"learning_rate": 0.00038428093645484946,
|
| 10301 |
+
"loss": 2.042,
|
| 10302 |
+
"step": 1705
|
| 10303 |
+
},
|
| 10304 |
+
{
|
| 10305 |
+
"epoch": 1.17,
|
| 10306 |
+
"learning_rate": 0.00038413760152890584,
|
| 10307 |
+
"loss": 1.974,
|
| 10308 |
+
"step": 1706
|
| 10309 |
+
},
|
| 10310 |
+
{
|
| 10311 |
+
"epoch": 1.17,
|
| 10312 |
+
"learning_rate": 0.0003839942666029622,
|
| 10313 |
+
"loss": 2.0785,
|
| 10314 |
+
"step": 1707
|
| 10315 |
+
},
|
| 10316 |
+
{
|
| 10317 |
+
"epoch": 1.17,
|
| 10318 |
+
"learning_rate": 0.0003838509316770186,
|
| 10319 |
+
"loss": 1.971,
|
| 10320 |
+
"step": 1708
|
| 10321 |
+
},
|
| 10322 |
+
{
|
| 10323 |
+
"epoch": 1.17,
|
| 10324 |
+
"learning_rate": 0.000383707596751075,
|
| 10325 |
+
"loss": 2.0806,
|
| 10326 |
+
"step": 1709
|
| 10327 |
+
},
|
| 10328 |
+
{
|
| 10329 |
+
"epoch": 1.17,
|
| 10330 |
+
"learning_rate": 0.0003835642618251314,
|
| 10331 |
+
"loss": 2.0394,
|
| 10332 |
+
"step": 1710
|
| 10333 |
+
},
|
| 10334 |
+
{
|
| 10335 |
+
"epoch": 1.17,
|
| 10336 |
+
"learning_rate": 0.0003834209268991877,
|
| 10337 |
+
"loss": 2.0122,
|
| 10338 |
+
"step": 1711
|
| 10339 |
+
},
|
| 10340 |
+
{
|
| 10341 |
+
"epoch": 1.17,
|
| 10342 |
+
"learning_rate": 0.00038327759197324414,
|
| 10343 |
+
"loss": 1.9853,
|
| 10344 |
+
"step": 1712
|
| 10345 |
+
},
|
| 10346 |
+
{
|
| 10347 |
+
"epoch": 1.17,
|
| 10348 |
+
"learning_rate": 0.00038313425704730047,
|
| 10349 |
+
"loss": 1.9518,
|
| 10350 |
+
"step": 1713
|
| 10351 |
+
},
|
| 10352 |
+
{
|
| 10353 |
+
"epoch": 1.17,
|
| 10354 |
+
"learning_rate": 0.0003829909221213569,
|
| 10355 |
+
"loss": 2.0134,
|
| 10356 |
+
"step": 1714
|
| 10357 |
+
},
|
| 10358 |
+
{
|
| 10359 |
+
"epoch": 1.17,
|
| 10360 |
+
"learning_rate": 0.00038284758719541324,
|
| 10361 |
+
"loss": 2.0334,
|
| 10362 |
+
"step": 1715
|
| 10363 |
+
},
|
| 10364 |
+
{
|
| 10365 |
+
"epoch": 1.17,
|
| 10366 |
+
"learning_rate": 0.0003827042522694697,
|
| 10367 |
+
"loss": 2.0515,
|
| 10368 |
+
"step": 1716
|
| 10369 |
+
},
|
| 10370 |
+
{
|
| 10371 |
+
"epoch": 1.17,
|
| 10372 |
+
"learning_rate": 0.000382560917343526,
|
| 10373 |
+
"loss": 2.0012,
|
| 10374 |
+
"step": 1717
|
| 10375 |
+
},
|
| 10376 |
+
{
|
| 10377 |
+
"epoch": 1.17,
|
| 10378 |
+
"learning_rate": 0.00038241758241758234,
|
| 10379 |
+
"loss": 1.9745,
|
| 10380 |
+
"step": 1718
|
| 10381 |
+
},
|
| 10382 |
+
{
|
| 10383 |
+
"epoch": 1.18,
|
| 10384 |
+
"learning_rate": 0.0003822742474916388,
|
| 10385 |
+
"loss": 2.0406,
|
| 10386 |
+
"step": 1719
|
| 10387 |
+
},
|
| 10388 |
+
{
|
| 10389 |
+
"epoch": 1.18,
|
| 10390 |
+
"learning_rate": 0.0003821309125656951,
|
| 10391 |
+
"loss": 2.0851,
|
| 10392 |
+
"step": 1720
|
| 10393 |
+
},
|
| 10394 |
+
{
|
| 10395 |
+
"epoch": 1.18,
|
| 10396 |
+
"learning_rate": 0.00038198757763975154,
|
| 10397 |
+
"loss": 1.991,
|
| 10398 |
+
"step": 1721
|
| 10399 |
+
},
|
| 10400 |
+
{
|
| 10401 |
+
"epoch": 1.18,
|
| 10402 |
+
"learning_rate": 0.00038184424271380787,
|
| 10403 |
+
"loss": 1.9414,
|
| 10404 |
+
"step": 1722
|
| 10405 |
+
},
|
| 10406 |
+
{
|
| 10407 |
+
"epoch": 1.18,
|
| 10408 |
+
"learning_rate": 0.0003817009077878643,
|
| 10409 |
+
"loss": 1.9876,
|
| 10410 |
+
"step": 1723
|
| 10411 |
+
},
|
| 10412 |
+
{
|
| 10413 |
+
"epoch": 1.18,
|
| 10414 |
+
"learning_rate": 0.00038155757286192064,
|
| 10415 |
+
"loss": 1.9759,
|
| 10416 |
+
"step": 1724
|
| 10417 |
+
},
|
| 10418 |
+
{
|
| 10419 |
+
"epoch": 1.18,
|
| 10420 |
+
"learning_rate": 0.000381414237935977,
|
| 10421 |
+
"loss": 2.0176,
|
| 10422 |
+
"step": 1725
|
| 10423 |
+
},
|
| 10424 |
+
{
|
| 10425 |
+
"epoch": 1.18,
|
| 10426 |
+
"learning_rate": 0.0003812709030100334,
|
| 10427 |
+
"loss": 2.063,
|
| 10428 |
+
"step": 1726
|
| 10429 |
+
},
|
| 10430 |
+
{
|
| 10431 |
+
"epoch": 1.18,
|
| 10432 |
+
"learning_rate": 0.0003811275680840898,
|
| 10433 |
+
"loss": 2.0379,
|
| 10434 |
+
"step": 1727
|
| 10435 |
+
},
|
| 10436 |
+
{
|
| 10437 |
+
"epoch": 1.18,
|
| 10438 |
+
"learning_rate": 0.0003809842331581462,
|
| 10439 |
+
"loss": 2.0511,
|
| 10440 |
+
"step": 1728
|
| 10441 |
+
},
|
| 10442 |
+
{
|
| 10443 |
+
"epoch": 1.18,
|
| 10444 |
+
"learning_rate": 0.00038084089823220256,
|
| 10445 |
+
"loss": 2.0851,
|
| 10446 |
+
"step": 1729
|
| 10447 |
+
},
|
| 10448 |
+
{
|
| 10449 |
+
"epoch": 1.18,
|
| 10450 |
+
"learning_rate": 0.00038069756330625894,
|
| 10451 |
+
"loss": 2.0056,
|
| 10452 |
+
"step": 1730
|
| 10453 |
+
},
|
| 10454 |
+
{
|
| 10455 |
+
"epoch": 1.18,
|
| 10456 |
+
"learning_rate": 0.0003805542283803153,
|
| 10457 |
+
"loss": 2.0117,
|
| 10458 |
+
"step": 1731
|
| 10459 |
+
},
|
| 10460 |
+
{
|
| 10461 |
+
"epoch": 1.18,
|
| 10462 |
+
"learning_rate": 0.00038041089345437166,
|
| 10463 |
+
"loss": 1.924,
|
| 10464 |
+
"step": 1732
|
| 10465 |
+
},
|
| 10466 |
+
{
|
| 10467 |
+
"epoch": 1.18,
|
| 10468 |
+
"learning_rate": 0.0003802675585284281,
|
| 10469 |
+
"loss": 1.9935,
|
| 10470 |
+
"step": 1733
|
| 10471 |
+
},
|
| 10472 |
+
{
|
| 10473 |
+
"epoch": 1.19,
|
| 10474 |
+
"learning_rate": 0.0003801242236024844,
|
| 10475 |
+
"loss": 1.9831,
|
| 10476 |
+
"step": 1734
|
| 10477 |
+
},
|
| 10478 |
+
{
|
| 10479 |
+
"epoch": 1.19,
|
| 10480 |
+
"learning_rate": 0.00037998088867654086,
|
| 10481 |
+
"loss": 2.0066,
|
| 10482 |
+
"step": 1735
|
| 10483 |
+
},
|
| 10484 |
+
{
|
| 10485 |
+
"epoch": 1.19,
|
| 10486 |
+
"learning_rate": 0.0003798375537505972,
|
| 10487 |
+
"loss": 1.9368,
|
| 10488 |
+
"step": 1736
|
| 10489 |
+
},
|
| 10490 |
+
{
|
| 10491 |
+
"epoch": 1.19,
|
| 10492 |
+
"learning_rate": 0.00037969421882465363,
|
| 10493 |
+
"loss": 2.1333,
|
| 10494 |
+
"step": 1737
|
| 10495 |
+
},
|
| 10496 |
+
{
|
| 10497 |
+
"epoch": 1.19,
|
| 10498 |
+
"learning_rate": 0.00037955088389870996,
|
| 10499 |
+
"loss": 2.0094,
|
| 10500 |
+
"step": 1738
|
| 10501 |
+
},
|
| 10502 |
+
{
|
| 10503 |
+
"epoch": 1.19,
|
| 10504 |
+
"learning_rate": 0.0003794075489727663,
|
| 10505 |
+
"loss": 2.0104,
|
| 10506 |
+
"step": 1739
|
| 10507 |
+
},
|
| 10508 |
+
{
|
| 10509 |
+
"epoch": 1.19,
|
| 10510 |
+
"learning_rate": 0.0003792642140468227,
|
| 10511 |
+
"loss": 2.1044,
|
| 10512 |
+
"step": 1740
|
| 10513 |
+
},
|
| 10514 |
+
{
|
| 10515 |
+
"epoch": 1.19,
|
| 10516 |
+
"learning_rate": 0.00037912087912087906,
|
| 10517 |
+
"loss": 1.9823,
|
| 10518 |
+
"step": 1741
|
| 10519 |
+
},
|
| 10520 |
+
{
|
| 10521 |
+
"epoch": 1.19,
|
| 10522 |
+
"learning_rate": 0.0003789775441949355,
|
| 10523 |
+
"loss": 2.0191,
|
| 10524 |
+
"step": 1742
|
| 10525 |
+
},
|
| 10526 |
+
{
|
| 10527 |
+
"epoch": 1.19,
|
| 10528 |
+
"learning_rate": 0.0003788342092689918,
|
| 10529 |
+
"loss": 1.9903,
|
| 10530 |
+
"step": 1743
|
| 10531 |
+
},
|
| 10532 |
+
{
|
| 10533 |
+
"epoch": 1.19,
|
| 10534 |
+
"learning_rate": 0.00037869087434304826,
|
| 10535 |
+
"loss": 2.0378,
|
| 10536 |
+
"step": 1744
|
| 10537 |
+
},
|
| 10538 |
+
{
|
| 10539 |
+
"epoch": 1.19,
|
| 10540 |
+
"learning_rate": 0.0003785475394171046,
|
| 10541 |
+
"loss": 1.9787,
|
| 10542 |
+
"step": 1745
|
| 10543 |
+
},
|
| 10544 |
+
{
|
| 10545 |
+
"epoch": 1.19,
|
| 10546 |
+
"learning_rate": 0.000378404204491161,
|
| 10547 |
+
"loss": 2.016,
|
| 10548 |
+
"step": 1746
|
| 10549 |
+
},
|
| 10550 |
+
{
|
| 10551 |
+
"epoch": 1.19,
|
| 10552 |
+
"learning_rate": 0.00037826086956521736,
|
| 10553 |
+
"loss": 2.0677,
|
| 10554 |
+
"step": 1747
|
| 10555 |
+
},
|
| 10556 |
+
{
|
| 10557 |
+
"epoch": 1.2,
|
| 10558 |
+
"learning_rate": 0.00037811753463927374,
|
| 10559 |
+
"loss": 2.0095,
|
| 10560 |
+
"step": 1748
|
| 10561 |
+
},
|
| 10562 |
+
{
|
| 10563 |
+
"epoch": 1.2,
|
| 10564 |
+
"learning_rate": 0.0003779741997133301,
|
| 10565 |
+
"loss": 2.0386,
|
| 10566 |
+
"step": 1749
|
| 10567 |
+
},
|
| 10568 |
+
{
|
| 10569 |
+
"epoch": 1.2,
|
| 10570 |
+
"learning_rate": 0.0003778308647873865,
|
| 10571 |
+
"loss": 2.0146,
|
| 10572 |
+
"step": 1750
|
| 10573 |
+
},
|
| 10574 |
+
{
|
| 10575 |
+
"epoch": 1.2,
|
| 10576 |
+
"learning_rate": 0.00037768752986144284,
|
| 10577 |
+
"loss": 2.0429,
|
| 10578 |
+
"step": 1751
|
| 10579 |
+
},
|
| 10580 |
+
{
|
| 10581 |
+
"epoch": 1.2,
|
| 10582 |
+
"learning_rate": 0.0003775441949354993,
|
| 10583 |
+
"loss": 2.0113,
|
| 10584 |
+
"step": 1752
|
| 10585 |
+
},
|
| 10586 |
+
{
|
| 10587 |
+
"epoch": 1.2,
|
| 10588 |
+
"learning_rate": 0.0003774008600095556,
|
| 10589 |
+
"loss": 2.0442,
|
| 10590 |
+
"step": 1753
|
| 10591 |
+
},
|
| 10592 |
+
{
|
| 10593 |
+
"epoch": 1.2,
|
| 10594 |
+
"learning_rate": 0.00037725752508361204,
|
| 10595 |
+
"loss": 2.0165,
|
| 10596 |
+
"step": 1754
|
| 10597 |
+
},
|
| 10598 |
+
{
|
| 10599 |
+
"epoch": 1.2,
|
| 10600 |
+
"learning_rate": 0.0003771141901576684,
|
| 10601 |
+
"loss": 2.0509,
|
| 10602 |
+
"step": 1755
|
| 10603 |
+
},
|
| 10604 |
+
{
|
| 10605 |
+
"epoch": 1.2,
|
| 10606 |
+
"learning_rate": 0.0003769708552317248,
|
| 10607 |
+
"loss": 2.0619,
|
| 10608 |
+
"step": 1756
|
| 10609 |
+
},
|
| 10610 |
+
{
|
| 10611 |
+
"epoch": 1.2,
|
| 10612 |
+
"learning_rate": 0.00037682752030578114,
|
| 10613 |
+
"loss": 1.9896,
|
| 10614 |
+
"step": 1757
|
| 10615 |
+
},
|
| 10616 |
+
{
|
| 10617 |
+
"epoch": 1.2,
|
| 10618 |
+
"learning_rate": 0.00037668418537983747,
|
| 10619 |
+
"loss": 1.9639,
|
| 10620 |
+
"step": 1758
|
| 10621 |
+
},
|
| 10622 |
+
{
|
| 10623 |
+
"epoch": 1.2,
|
| 10624 |
+
"learning_rate": 0.0003765408504538939,
|
| 10625 |
+
"loss": 1.9989,
|
| 10626 |
+
"step": 1759
|
| 10627 |
+
},
|
| 10628 |
+
{
|
| 10629 |
+
"epoch": 1.2,
|
| 10630 |
+
"learning_rate": 0.00037639751552795024,
|
| 10631 |
+
"loss": 1.9954,
|
| 10632 |
+
"step": 1760
|
| 10633 |
+
},
|
| 10634 |
+
{
|
| 10635 |
+
"epoch": 1.2,
|
| 10636 |
+
"learning_rate": 0.0003762541806020067,
|
| 10637 |
+
"loss": 2.0313,
|
| 10638 |
+
"step": 1761
|
| 10639 |
+
},
|
| 10640 |
+
{
|
| 10641 |
+
"epoch": 1.2,
|
| 10642 |
+
"learning_rate": 0.000376110845676063,
|
| 10643 |
+
"loss": 1.9543,
|
| 10644 |
+
"step": 1762
|
| 10645 |
+
},
|
| 10646 |
+
{
|
| 10647 |
+
"epoch": 1.21,
|
| 10648 |
+
"learning_rate": 0.00037596751075011944,
|
| 10649 |
+
"loss": 1.982,
|
| 10650 |
+
"step": 1763
|
| 10651 |
+
},
|
| 10652 |
+
{
|
| 10653 |
+
"epoch": 1.21,
|
| 10654 |
+
"learning_rate": 0.0003758241758241758,
|
| 10655 |
+
"loss": 2.0651,
|
| 10656 |
+
"step": 1764
|
| 10657 |
+
},
|
| 10658 |
+
{
|
| 10659 |
+
"epoch": 1.21,
|
| 10660 |
+
"learning_rate": 0.00037568084089823216,
|
| 10661 |
+
"loss": 2.0458,
|
| 10662 |
+
"step": 1765
|
| 10663 |
+
},
|
| 10664 |
+
{
|
| 10665 |
+
"epoch": 1.21,
|
| 10666 |
+
"learning_rate": 0.00037553750597228854,
|
| 10667 |
+
"loss": 2.0584,
|
| 10668 |
+
"step": 1766
|
| 10669 |
+
},
|
| 10670 |
+
{
|
| 10671 |
+
"epoch": 1.21,
|
| 10672 |
+
"learning_rate": 0.0003753941710463449,
|
| 10673 |
+
"loss": 2.0646,
|
| 10674 |
+
"step": 1767
|
| 10675 |
+
},
|
| 10676 |
+
{
|
| 10677 |
+
"epoch": 1.21,
|
| 10678 |
+
"learning_rate": 0.0003752508361204013,
|
| 10679 |
+
"loss": 1.9832,
|
| 10680 |
+
"step": 1768
|
| 10681 |
+
},
|
| 10682 |
+
{
|
| 10683 |
+
"epoch": 1.21,
|
| 10684 |
+
"learning_rate": 0.0003751075011944577,
|
| 10685 |
+
"loss": 2.0284,
|
| 10686 |
+
"step": 1769
|
| 10687 |
+
},
|
| 10688 |
+
{
|
| 10689 |
+
"epoch": 1.21,
|
| 10690 |
+
"learning_rate": 0.0003749641662685141,
|
| 10691 |
+
"loss": 2.002,
|
| 10692 |
+
"step": 1770
|
| 10693 |
+
},
|
| 10694 |
+
{
|
| 10695 |
+
"epoch": 1.21,
|
| 10696 |
+
"learning_rate": 0.00037482083134257046,
|
| 10697 |
+
"loss": 2.0567,
|
| 10698 |
+
"step": 1771
|
| 10699 |
+
},
|
| 10700 |
+
{
|
| 10701 |
+
"epoch": 1.21,
|
| 10702 |
+
"learning_rate": 0.0003746774964166268,
|
| 10703 |
+
"loss": 1.9435,
|
| 10704 |
+
"step": 1772
|
| 10705 |
+
},
|
| 10706 |
+
{
|
| 10707 |
+
"epoch": 1.21,
|
| 10708 |
+
"learning_rate": 0.00037453416149068323,
|
| 10709 |
+
"loss": 2.0795,
|
| 10710 |
+
"step": 1773
|
| 10711 |
+
},
|
| 10712 |
+
{
|
| 10713 |
+
"epoch": 1.21,
|
| 10714 |
+
"learning_rate": 0.00037439082656473956,
|
| 10715 |
+
"loss": 1.9883,
|
| 10716 |
+
"step": 1774
|
| 10717 |
+
},
|
| 10718 |
+
{
|
| 10719 |
+
"epoch": 1.21,
|
| 10720 |
+
"learning_rate": 0.000374247491638796,
|
| 10721 |
+
"loss": 1.9813,
|
| 10722 |
+
"step": 1775
|
| 10723 |
+
},
|
| 10724 |
+
{
|
| 10725 |
+
"epoch": 1.21,
|
| 10726 |
+
"learning_rate": 0.0003741041567128523,
|
| 10727 |
+
"loss": 2.0155,
|
| 10728 |
+
"step": 1776
|
| 10729 |
+
},
|
| 10730 |
+
{
|
| 10731 |
+
"epoch": 1.21,
|
| 10732 |
+
"learning_rate": 0.00037396082178690876,
|
| 10733 |
+
"loss": 2.0065,
|
| 10734 |
+
"step": 1777
|
| 10735 |
+
},
|
| 10736 |
+
{
|
| 10737 |
+
"epoch": 1.22,
|
| 10738 |
+
"learning_rate": 0.0003738174868609651,
|
| 10739 |
+
"loss": 1.9658,
|
| 10740 |
+
"step": 1778
|
| 10741 |
+
},
|
| 10742 |
+
{
|
| 10743 |
+
"epoch": 1.22,
|
| 10744 |
+
"learning_rate": 0.0003736741519350214,
|
| 10745 |
+
"loss": 2.0513,
|
| 10746 |
+
"step": 1779
|
| 10747 |
+
},
|
| 10748 |
+
{
|
| 10749 |
+
"epoch": 1.22,
|
| 10750 |
+
"learning_rate": 0.00037353081700907786,
|
| 10751 |
+
"loss": 2.1119,
|
| 10752 |
+
"step": 1780
|
| 10753 |
+
},
|
| 10754 |
+
{
|
| 10755 |
+
"epoch": 1.22,
|
| 10756 |
+
"learning_rate": 0.0003733874820831342,
|
| 10757 |
+
"loss": 1.9894,
|
| 10758 |
+
"step": 1781
|
| 10759 |
+
},
|
| 10760 |
+
{
|
| 10761 |
+
"epoch": 1.22,
|
| 10762 |
+
"learning_rate": 0.00037324414715719063,
|
| 10763 |
+
"loss": 2.0833,
|
| 10764 |
+
"step": 1782
|
| 10765 |
+
},
|
| 10766 |
+
{
|
| 10767 |
+
"epoch": 1.22,
|
| 10768 |
+
"learning_rate": 0.00037310081223124696,
|
| 10769 |
+
"loss": 2.0896,
|
| 10770 |
+
"step": 1783
|
| 10771 |
+
},
|
| 10772 |
+
{
|
| 10773 |
+
"epoch": 1.22,
|
| 10774 |
+
"learning_rate": 0.0003729574773053034,
|
| 10775 |
+
"loss": 1.9927,
|
| 10776 |
+
"step": 1784
|
| 10777 |
+
},
|
| 10778 |
+
{
|
| 10779 |
+
"epoch": 1.22,
|
| 10780 |
+
"learning_rate": 0.0003728141423793597,
|
| 10781 |
+
"loss": 1.9954,
|
| 10782 |
+
"step": 1785
|
| 10783 |
+
},
|
| 10784 |
+
{
|
| 10785 |
+
"epoch": 1.22,
|
| 10786 |
+
"learning_rate": 0.0003726708074534161,
|
| 10787 |
+
"loss": 2.0618,
|
| 10788 |
+
"step": 1786
|
| 10789 |
+
},
|
| 10790 |
+
{
|
| 10791 |
+
"epoch": 1.22,
|
| 10792 |
+
"learning_rate": 0.0003725274725274725,
|
| 10793 |
+
"loss": 1.9595,
|
| 10794 |
+
"step": 1787
|
| 10795 |
+
},
|
| 10796 |
+
{
|
| 10797 |
+
"epoch": 1.22,
|
| 10798 |
+
"learning_rate": 0.0003723841376015289,
|
| 10799 |
+
"loss": 2.0487,
|
| 10800 |
+
"step": 1788
|
| 10801 |
+
},
|
| 10802 |
+
{
|
| 10803 |
+
"epoch": 1.22,
|
| 10804 |
+
"learning_rate": 0.00037224080267558526,
|
| 10805 |
+
"loss": 2.1212,
|
| 10806 |
+
"step": 1789
|
| 10807 |
+
},
|
| 10808 |
+
{
|
| 10809 |
+
"epoch": 1.22,
|
| 10810 |
+
"learning_rate": 0.00037209746774964164,
|
| 10811 |
+
"loss": 1.976,
|
| 10812 |
+
"step": 1790
|
| 10813 |
+
},
|
| 10814 |
+
{
|
| 10815 |
+
"epoch": 1.22,
|
| 10816 |
+
"learning_rate": 0.00037195413282369803,
|
| 10817 |
+
"loss": 1.8774,
|
| 10818 |
+
"step": 1791
|
| 10819 |
+
},
|
| 10820 |
+
{
|
| 10821 |
+
"epoch": 1.23,
|
| 10822 |
+
"learning_rate": 0.0003718107978977544,
|
| 10823 |
+
"loss": 2.0226,
|
| 10824 |
+
"step": 1792
|
| 10825 |
+
},
|
| 10826 |
+
{
|
| 10827 |
+
"epoch": 1.23,
|
| 10828 |
+
"learning_rate": 0.00037166746297181074,
|
| 10829 |
+
"loss": 1.9768,
|
| 10830 |
+
"step": 1793
|
| 10831 |
+
},
|
| 10832 |
+
{
|
| 10833 |
+
"epoch": 1.23,
|
| 10834 |
+
"learning_rate": 0.0003715241280458672,
|
| 10835 |
+
"loss": 2.0637,
|
| 10836 |
+
"step": 1794
|
| 10837 |
+
},
|
| 10838 |
+
{
|
| 10839 |
+
"epoch": 1.23,
|
| 10840 |
+
"learning_rate": 0.0003713807931199235,
|
| 10841 |
+
"loss": 2.0001,
|
| 10842 |
+
"step": 1795
|
| 10843 |
+
},
|
| 10844 |
+
{
|
| 10845 |
+
"epoch": 1.23,
|
| 10846 |
+
"learning_rate": 0.00037123745819397995,
|
| 10847 |
+
"loss": 2.0729,
|
| 10848 |
+
"step": 1796
|
| 10849 |
+
},
|
| 10850 |
+
{
|
| 10851 |
+
"epoch": 1.23,
|
| 10852 |
+
"learning_rate": 0.0003710941232680363,
|
| 10853 |
+
"loss": 1.9483,
|
| 10854 |
+
"step": 1797
|
| 10855 |
+
},
|
| 10856 |
+
{
|
| 10857 |
+
"epoch": 1.23,
|
| 10858 |
+
"learning_rate": 0.00037095078834209266,
|
| 10859 |
+
"loss": 2.0153,
|
| 10860 |
+
"step": 1798
|
| 10861 |
+
},
|
| 10862 |
+
{
|
| 10863 |
+
"epoch": 1.23,
|
| 10864 |
+
"learning_rate": 0.00037080745341614904,
|
| 10865 |
+
"loss": 2.0948,
|
| 10866 |
+
"step": 1799
|
| 10867 |
+
},
|
| 10868 |
+
{
|
| 10869 |
+
"epoch": 1.23,
|
| 10870 |
+
"learning_rate": 0.0003706641184902054,
|
| 10871 |
+
"loss": 2.0054,
|
| 10872 |
+
"step": 1800
|
| 10873 |
+
},
|
| 10874 |
+
{
|
| 10875 |
+
"epoch": 1.23,
|
| 10876 |
+
"eval_loss": 1.9992964267730713,
|
| 10877 |
+
"eval_runtime": 1649.7967,
|
| 10878 |
+
"eval_samples_per_second": 6.061,
|
| 10879 |
+
"eval_steps_per_second": 6.061,
|
| 10880 |
+
"step": 1800
|
| 10881 |
}
|
| 10882 |
],
|
| 10883 |
"max_steps": 4386,
|
| 10884 |
"num_train_epochs": 3,
|
| 10885 |
+
"total_flos": 2.2063711021239214e+18,
|
| 10886 |
"trial_name": null,
|
| 10887 |
"trial_params": null
|
| 10888 |
}
|
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:e0581c601cbb195614ebff9150719cfc4b2d8be4e3802ef0fd4f340d4b33ab78
|
| 3 |
size 2368281769
|