Training in progress, step 1000
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:ae99f5b9471b1b88e22d76dfd54369c904f7fbe448047cf0c5671464d7f6274a
|
| 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:cc924070c94c5b619e2669e03e1453bcf47c8ac1d6edf146fae93888c7686dcc
|
| 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:c2958aedd09b3f2455955cd165d68fd32ae680af19bdad126f498dd665ace09f
|
| 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:af3121ec18c8bd79a8c50e67e6a316e1da89b03afea7df26186a8c09abf682cd
|
| 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,
|
|
@@ -4838,11 +4838,1219 @@
|
|
| 4838 |
"eval_samples_per_second": 5.918,
|
| 4839 |
"eval_steps_per_second": 5.918,
|
| 4840 |
"step": 800
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4841 |
}
|
| 4842 |
],
|
| 4843 |
"max_steps": 4386,
|
| 4844 |
"num_train_epochs": 3,
|
| 4845 |
-
"total_flos":
|
| 4846 |
"trial_name": null,
|
| 4847 |
"trial_params": null
|
| 4848 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 2.0208685398101807,
|
| 3 |
+
"best_model_checkpoint": "output/checkpoint-1000",
|
| 4 |
+
"epoch": 0.6836839884414676,
|
| 5 |
+
"global_step": 1000,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 4838 |
"eval_samples_per_second": 5.918,
|
| 4839 |
"eval_steps_per_second": 5.918,
|
| 4840 |
"step": 800
|
| 4841 |
+
},
|
| 4842 |
+
{
|
| 4843 |
+
"epoch": 0.55,
|
| 4844 |
+
"learning_rate": 0.0005138557095078833,
|
| 4845 |
+
"loss": 2.1501,
|
| 4846 |
+
"step": 801
|
| 4847 |
+
},
|
| 4848 |
+
{
|
| 4849 |
+
"epoch": 0.55,
|
| 4850 |
+
"learning_rate": 0.0005137123745819397,
|
| 4851 |
+
"loss": 2.0977,
|
| 4852 |
+
"step": 802
|
| 4853 |
+
},
|
| 4854 |
+
{
|
| 4855 |
+
"epoch": 0.55,
|
| 4856 |
+
"learning_rate": 0.0005135690396559961,
|
| 4857 |
+
"loss": 2.2608,
|
| 4858 |
+
"step": 803
|
| 4859 |
+
},
|
| 4860 |
+
{
|
| 4861 |
+
"epoch": 0.55,
|
| 4862 |
+
"learning_rate": 0.0005134257047300525,
|
| 4863 |
+
"loss": 2.0765,
|
| 4864 |
+
"step": 804
|
| 4865 |
+
},
|
| 4866 |
+
{
|
| 4867 |
+
"epoch": 0.55,
|
| 4868 |
+
"learning_rate": 0.0005132823698041089,
|
| 4869 |
+
"loss": 2.1414,
|
| 4870 |
+
"step": 805
|
| 4871 |
+
},
|
| 4872 |
+
{
|
| 4873 |
+
"epoch": 0.55,
|
| 4874 |
+
"learning_rate": 0.0005131390348781653,
|
| 4875 |
+
"loss": 2.2186,
|
| 4876 |
+
"step": 806
|
| 4877 |
+
},
|
| 4878 |
+
{
|
| 4879 |
+
"epoch": 0.55,
|
| 4880 |
+
"learning_rate": 0.0005129956999522216,
|
| 4881 |
+
"loss": 2.1596,
|
| 4882 |
+
"step": 807
|
| 4883 |
+
},
|
| 4884 |
+
{
|
| 4885 |
+
"epoch": 0.55,
|
| 4886 |
+
"learning_rate": 0.000512852365026278,
|
| 4887 |
+
"loss": 2.1241,
|
| 4888 |
+
"step": 808
|
| 4889 |
+
},
|
| 4890 |
+
{
|
| 4891 |
+
"epoch": 0.55,
|
| 4892 |
+
"learning_rate": 0.0005127090301003344,
|
| 4893 |
+
"loss": 2.0742,
|
| 4894 |
+
"step": 809
|
| 4895 |
+
},
|
| 4896 |
+
{
|
| 4897 |
+
"epoch": 0.55,
|
| 4898 |
+
"learning_rate": 0.0005125656951743908,
|
| 4899 |
+
"loss": 2.0797,
|
| 4900 |
+
"step": 810
|
| 4901 |
+
},
|
| 4902 |
+
{
|
| 4903 |
+
"epoch": 0.55,
|
| 4904 |
+
"learning_rate": 0.0005124223602484472,
|
| 4905 |
+
"loss": 2.1866,
|
| 4906 |
+
"step": 811
|
| 4907 |
+
},
|
| 4908 |
+
{
|
| 4909 |
+
"epoch": 0.56,
|
| 4910 |
+
"learning_rate": 0.0005122790253225035,
|
| 4911 |
+
"loss": 2.1895,
|
| 4912 |
+
"step": 812
|
| 4913 |
+
},
|
| 4914 |
+
{
|
| 4915 |
+
"epoch": 0.56,
|
| 4916 |
+
"learning_rate": 0.00051213569039656,
|
| 4917 |
+
"loss": 2.0698,
|
| 4918 |
+
"step": 813
|
| 4919 |
+
},
|
| 4920 |
+
{
|
| 4921 |
+
"epoch": 0.56,
|
| 4922 |
+
"learning_rate": 0.0005119923554706162,
|
| 4923 |
+
"loss": 2.1982,
|
| 4924 |
+
"step": 814
|
| 4925 |
+
},
|
| 4926 |
+
{
|
| 4927 |
+
"epoch": 0.56,
|
| 4928 |
+
"learning_rate": 0.0005118490205446727,
|
| 4929 |
+
"loss": 2.1768,
|
| 4930 |
+
"step": 815
|
| 4931 |
+
},
|
| 4932 |
+
{
|
| 4933 |
+
"epoch": 0.56,
|
| 4934 |
+
"learning_rate": 0.000511705685618729,
|
| 4935 |
+
"loss": 2.0943,
|
| 4936 |
+
"step": 816
|
| 4937 |
+
},
|
| 4938 |
+
{
|
| 4939 |
+
"epoch": 0.56,
|
| 4940 |
+
"learning_rate": 0.0005115623506927855,
|
| 4941 |
+
"loss": 2.067,
|
| 4942 |
+
"step": 817
|
| 4943 |
+
},
|
| 4944 |
+
{
|
| 4945 |
+
"epoch": 0.56,
|
| 4946 |
+
"learning_rate": 0.0005114190157668418,
|
| 4947 |
+
"loss": 2.0416,
|
| 4948 |
+
"step": 818
|
| 4949 |
+
},
|
| 4950 |
+
{
|
| 4951 |
+
"epoch": 0.56,
|
| 4952 |
+
"learning_rate": 0.0005112756808408981,
|
| 4953 |
+
"loss": 2.1266,
|
| 4954 |
+
"step": 819
|
| 4955 |
+
},
|
| 4956 |
+
{
|
| 4957 |
+
"epoch": 0.56,
|
| 4958 |
+
"learning_rate": 0.0005111323459149545,
|
| 4959 |
+
"loss": 2.1835,
|
| 4960 |
+
"step": 820
|
| 4961 |
+
},
|
| 4962 |
+
{
|
| 4963 |
+
"epoch": 0.56,
|
| 4964 |
+
"learning_rate": 0.0005109890109890109,
|
| 4965 |
+
"loss": 2.1033,
|
| 4966 |
+
"step": 821
|
| 4967 |
+
},
|
| 4968 |
+
{
|
| 4969 |
+
"epoch": 0.56,
|
| 4970 |
+
"learning_rate": 0.0005108456760630673,
|
| 4971 |
+
"loss": 2.1953,
|
| 4972 |
+
"step": 822
|
| 4973 |
+
},
|
| 4974 |
+
{
|
| 4975 |
+
"epoch": 0.56,
|
| 4976 |
+
"learning_rate": 0.0005107023411371237,
|
| 4977 |
+
"loss": 2.2002,
|
| 4978 |
+
"step": 823
|
| 4979 |
+
},
|
| 4980 |
+
{
|
| 4981 |
+
"epoch": 0.56,
|
| 4982 |
+
"learning_rate": 0.0005105590062111801,
|
| 4983 |
+
"loss": 2.0887,
|
| 4984 |
+
"step": 824
|
| 4985 |
+
},
|
| 4986 |
+
{
|
| 4987 |
+
"epoch": 0.56,
|
| 4988 |
+
"learning_rate": 0.0005104156712852364,
|
| 4989 |
+
"loss": 2.1419,
|
| 4990 |
+
"step": 825
|
| 4991 |
+
},
|
| 4992 |
+
{
|
| 4993 |
+
"epoch": 0.56,
|
| 4994 |
+
"learning_rate": 0.0005102723363592928,
|
| 4995 |
+
"loss": 2.0993,
|
| 4996 |
+
"step": 826
|
| 4997 |
+
},
|
| 4998 |
+
{
|
| 4999 |
+
"epoch": 0.57,
|
| 5000 |
+
"learning_rate": 0.0005101290014333492,
|
| 5001 |
+
"loss": 2.1632,
|
| 5002 |
+
"step": 827
|
| 5003 |
+
},
|
| 5004 |
+
{
|
| 5005 |
+
"epoch": 0.57,
|
| 5006 |
+
"learning_rate": 0.0005099856665074056,
|
| 5007 |
+
"loss": 2.11,
|
| 5008 |
+
"step": 828
|
| 5009 |
+
},
|
| 5010 |
+
{
|
| 5011 |
+
"epoch": 0.57,
|
| 5012 |
+
"learning_rate": 0.000509842331581462,
|
| 5013 |
+
"loss": 2.1375,
|
| 5014 |
+
"step": 829
|
| 5015 |
+
},
|
| 5016 |
+
{
|
| 5017 |
+
"epoch": 0.57,
|
| 5018 |
+
"learning_rate": 0.0005096989966555184,
|
| 5019 |
+
"loss": 2.0241,
|
| 5020 |
+
"step": 830
|
| 5021 |
+
},
|
| 5022 |
+
{
|
| 5023 |
+
"epoch": 0.57,
|
| 5024 |
+
"learning_rate": 0.0005095556617295748,
|
| 5025 |
+
"loss": 1.9997,
|
| 5026 |
+
"step": 831
|
| 5027 |
+
},
|
| 5028 |
+
{
|
| 5029 |
+
"epoch": 0.57,
|
| 5030 |
+
"learning_rate": 0.0005094123268036311,
|
| 5031 |
+
"loss": 2.095,
|
| 5032 |
+
"step": 832
|
| 5033 |
+
},
|
| 5034 |
+
{
|
| 5035 |
+
"epoch": 0.57,
|
| 5036 |
+
"learning_rate": 0.0005092689918776874,
|
| 5037 |
+
"loss": 2.114,
|
| 5038 |
+
"step": 833
|
| 5039 |
+
},
|
| 5040 |
+
{
|
| 5041 |
+
"epoch": 0.57,
|
| 5042 |
+
"learning_rate": 0.0005091256569517439,
|
| 5043 |
+
"loss": 2.0887,
|
| 5044 |
+
"step": 834
|
| 5045 |
+
},
|
| 5046 |
+
{
|
| 5047 |
+
"epoch": 0.57,
|
| 5048 |
+
"learning_rate": 0.0005089823220258002,
|
| 5049 |
+
"loss": 2.1666,
|
| 5050 |
+
"step": 835
|
| 5051 |
+
},
|
| 5052 |
+
{
|
| 5053 |
+
"epoch": 0.57,
|
| 5054 |
+
"learning_rate": 0.0005088389870998567,
|
| 5055 |
+
"loss": 2.1407,
|
| 5056 |
+
"step": 836
|
| 5057 |
+
},
|
| 5058 |
+
{
|
| 5059 |
+
"epoch": 0.57,
|
| 5060 |
+
"learning_rate": 0.0005086956521739129,
|
| 5061 |
+
"loss": 2.2634,
|
| 5062 |
+
"step": 837
|
| 5063 |
+
},
|
| 5064 |
+
{
|
| 5065 |
+
"epoch": 0.57,
|
| 5066 |
+
"learning_rate": 0.0005085523172479694,
|
| 5067 |
+
"loss": 2.2145,
|
| 5068 |
+
"step": 838
|
| 5069 |
+
},
|
| 5070 |
+
{
|
| 5071 |
+
"epoch": 0.57,
|
| 5072 |
+
"learning_rate": 0.0005084089823220257,
|
| 5073 |
+
"loss": 2.0569,
|
| 5074 |
+
"step": 839
|
| 5075 |
+
},
|
| 5076 |
+
{
|
| 5077 |
+
"epoch": 0.57,
|
| 5078 |
+
"learning_rate": 0.0005082656473960821,
|
| 5079 |
+
"loss": 2.1156,
|
| 5080 |
+
"step": 840
|
| 5081 |
+
},
|
| 5082 |
+
{
|
| 5083 |
+
"epoch": 0.57,
|
| 5084 |
+
"learning_rate": 0.0005081223124701385,
|
| 5085 |
+
"loss": 2.1948,
|
| 5086 |
+
"step": 841
|
| 5087 |
+
},
|
| 5088 |
+
{
|
| 5089 |
+
"epoch": 0.58,
|
| 5090 |
+
"learning_rate": 0.0005079789775441949,
|
| 5091 |
+
"loss": 2.1315,
|
| 5092 |
+
"step": 842
|
| 5093 |
+
},
|
| 5094 |
+
{
|
| 5095 |
+
"epoch": 0.58,
|
| 5096 |
+
"learning_rate": 0.0005078356426182512,
|
| 5097 |
+
"loss": 2.1392,
|
| 5098 |
+
"step": 843
|
| 5099 |
+
},
|
| 5100 |
+
{
|
| 5101 |
+
"epoch": 0.58,
|
| 5102 |
+
"learning_rate": 0.0005076923076923076,
|
| 5103 |
+
"loss": 2.1225,
|
| 5104 |
+
"step": 844
|
| 5105 |
+
},
|
| 5106 |
+
{
|
| 5107 |
+
"epoch": 0.58,
|
| 5108 |
+
"learning_rate": 0.000507548972766364,
|
| 5109 |
+
"loss": 2.1076,
|
| 5110 |
+
"step": 845
|
| 5111 |
+
},
|
| 5112 |
+
{
|
| 5113 |
+
"epoch": 0.58,
|
| 5114 |
+
"learning_rate": 0.0005074056378404204,
|
| 5115 |
+
"loss": 2.0865,
|
| 5116 |
+
"step": 846
|
| 5117 |
+
},
|
| 5118 |
+
{
|
| 5119 |
+
"epoch": 0.58,
|
| 5120 |
+
"learning_rate": 0.0005072623029144768,
|
| 5121 |
+
"loss": 2.159,
|
| 5122 |
+
"step": 847
|
| 5123 |
+
},
|
| 5124 |
+
{
|
| 5125 |
+
"epoch": 0.58,
|
| 5126 |
+
"learning_rate": 0.0005071189679885332,
|
| 5127 |
+
"loss": 2.0958,
|
| 5128 |
+
"step": 848
|
| 5129 |
+
},
|
| 5130 |
+
{
|
| 5131 |
+
"epoch": 0.58,
|
| 5132 |
+
"learning_rate": 0.0005069756330625896,
|
| 5133 |
+
"loss": 2.1292,
|
| 5134 |
+
"step": 849
|
| 5135 |
+
},
|
| 5136 |
+
{
|
| 5137 |
+
"epoch": 0.58,
|
| 5138 |
+
"learning_rate": 0.0005068322981366459,
|
| 5139 |
+
"loss": 2.1155,
|
| 5140 |
+
"step": 850
|
| 5141 |
+
},
|
| 5142 |
+
{
|
| 5143 |
+
"epoch": 0.58,
|
| 5144 |
+
"learning_rate": 0.0005066889632107023,
|
| 5145 |
+
"loss": 2.0926,
|
| 5146 |
+
"step": 851
|
| 5147 |
+
},
|
| 5148 |
+
{
|
| 5149 |
+
"epoch": 0.58,
|
| 5150 |
+
"learning_rate": 0.0005065456282847586,
|
| 5151 |
+
"loss": 2.1385,
|
| 5152 |
+
"step": 852
|
| 5153 |
+
},
|
| 5154 |
+
{
|
| 5155 |
+
"epoch": 0.58,
|
| 5156 |
+
"learning_rate": 0.0005064022933588151,
|
| 5157 |
+
"loss": 2.1187,
|
| 5158 |
+
"step": 853
|
| 5159 |
+
},
|
| 5160 |
+
{
|
| 5161 |
+
"epoch": 0.58,
|
| 5162 |
+
"learning_rate": 0.0005062589584328714,
|
| 5163 |
+
"loss": 2.0747,
|
| 5164 |
+
"step": 854
|
| 5165 |
+
},
|
| 5166 |
+
{
|
| 5167 |
+
"epoch": 0.58,
|
| 5168 |
+
"learning_rate": 0.0005061156235069279,
|
| 5169 |
+
"loss": 2.147,
|
| 5170 |
+
"step": 855
|
| 5171 |
+
},
|
| 5172 |
+
{
|
| 5173 |
+
"epoch": 0.59,
|
| 5174 |
+
"learning_rate": 0.0005059722885809841,
|
| 5175 |
+
"loss": 2.2023,
|
| 5176 |
+
"step": 856
|
| 5177 |
+
},
|
| 5178 |
+
{
|
| 5179 |
+
"epoch": 0.59,
|
| 5180 |
+
"learning_rate": 0.0005058289536550406,
|
| 5181 |
+
"loss": 2.0863,
|
| 5182 |
+
"step": 857
|
| 5183 |
+
},
|
| 5184 |
+
{
|
| 5185 |
+
"epoch": 0.59,
|
| 5186 |
+
"learning_rate": 0.0005056856187290969,
|
| 5187 |
+
"loss": 2.1402,
|
| 5188 |
+
"step": 858
|
| 5189 |
+
},
|
| 5190 |
+
{
|
| 5191 |
+
"epoch": 0.59,
|
| 5192 |
+
"learning_rate": 0.0005055422838031533,
|
| 5193 |
+
"loss": 2.1758,
|
| 5194 |
+
"step": 859
|
| 5195 |
+
},
|
| 5196 |
+
{
|
| 5197 |
+
"epoch": 0.59,
|
| 5198 |
+
"learning_rate": 0.0005053989488772097,
|
| 5199 |
+
"loss": 2.1767,
|
| 5200 |
+
"step": 860
|
| 5201 |
+
},
|
| 5202 |
+
{
|
| 5203 |
+
"epoch": 0.59,
|
| 5204 |
+
"learning_rate": 0.000505255613951266,
|
| 5205 |
+
"loss": 2.2041,
|
| 5206 |
+
"step": 861
|
| 5207 |
+
},
|
| 5208 |
+
{
|
| 5209 |
+
"epoch": 0.59,
|
| 5210 |
+
"learning_rate": 0.0005051122790253224,
|
| 5211 |
+
"loss": 2.1227,
|
| 5212 |
+
"step": 862
|
| 5213 |
+
},
|
| 5214 |
+
{
|
| 5215 |
+
"epoch": 0.59,
|
| 5216 |
+
"learning_rate": 0.0005049689440993788,
|
| 5217 |
+
"loss": 2.1992,
|
| 5218 |
+
"step": 863
|
| 5219 |
+
},
|
| 5220 |
+
{
|
| 5221 |
+
"epoch": 0.59,
|
| 5222 |
+
"learning_rate": 0.0005048256091734352,
|
| 5223 |
+
"loss": 2.0532,
|
| 5224 |
+
"step": 864
|
| 5225 |
+
},
|
| 5226 |
+
{
|
| 5227 |
+
"epoch": 0.59,
|
| 5228 |
+
"learning_rate": 0.0005046822742474916,
|
| 5229 |
+
"loss": 2.1845,
|
| 5230 |
+
"step": 865
|
| 5231 |
+
},
|
| 5232 |
+
{
|
| 5233 |
+
"epoch": 0.59,
|
| 5234 |
+
"learning_rate": 0.000504538939321548,
|
| 5235 |
+
"loss": 2.1466,
|
| 5236 |
+
"step": 866
|
| 5237 |
+
},
|
| 5238 |
+
{
|
| 5239 |
+
"epoch": 0.59,
|
| 5240 |
+
"learning_rate": 0.0005043956043956043,
|
| 5241 |
+
"loss": 2.0371,
|
| 5242 |
+
"step": 867
|
| 5243 |
+
},
|
| 5244 |
+
{
|
| 5245 |
+
"epoch": 0.59,
|
| 5246 |
+
"learning_rate": 0.0005042522694696607,
|
| 5247 |
+
"loss": 2.0987,
|
| 5248 |
+
"step": 868
|
| 5249 |
+
},
|
| 5250 |
+
{
|
| 5251 |
+
"epoch": 0.59,
|
| 5252 |
+
"learning_rate": 0.0005041089345437171,
|
| 5253 |
+
"loss": 2.1345,
|
| 5254 |
+
"step": 869
|
| 5255 |
+
},
|
| 5256 |
+
{
|
| 5257 |
+
"epoch": 0.59,
|
| 5258 |
+
"learning_rate": 0.0005039655996177735,
|
| 5259 |
+
"loss": 2.2204,
|
| 5260 |
+
"step": 870
|
| 5261 |
+
},
|
| 5262 |
+
{
|
| 5263 |
+
"epoch": 0.6,
|
| 5264 |
+
"learning_rate": 0.0005038222646918299,
|
| 5265 |
+
"loss": 2.1403,
|
| 5266 |
+
"step": 871
|
| 5267 |
+
},
|
| 5268 |
+
{
|
| 5269 |
+
"epoch": 0.6,
|
| 5270 |
+
"learning_rate": 0.0005036789297658863,
|
| 5271 |
+
"loss": 2.0582,
|
| 5272 |
+
"step": 872
|
| 5273 |
+
},
|
| 5274 |
+
{
|
| 5275 |
+
"epoch": 0.6,
|
| 5276 |
+
"learning_rate": 0.0005035355948399425,
|
| 5277 |
+
"loss": 2.1603,
|
| 5278 |
+
"step": 873
|
| 5279 |
+
},
|
| 5280 |
+
{
|
| 5281 |
+
"epoch": 0.6,
|
| 5282 |
+
"learning_rate": 0.000503392259913999,
|
| 5283 |
+
"loss": 2.0686,
|
| 5284 |
+
"step": 874
|
| 5285 |
+
},
|
| 5286 |
+
{
|
| 5287 |
+
"epoch": 0.6,
|
| 5288 |
+
"learning_rate": 0.0005032489249880553,
|
| 5289 |
+
"loss": 2.0596,
|
| 5290 |
+
"step": 875
|
| 5291 |
+
},
|
| 5292 |
+
{
|
| 5293 |
+
"epoch": 0.6,
|
| 5294 |
+
"learning_rate": 0.0005031055900621118,
|
| 5295 |
+
"loss": 2.1168,
|
| 5296 |
+
"step": 876
|
| 5297 |
+
},
|
| 5298 |
+
{
|
| 5299 |
+
"epoch": 0.6,
|
| 5300 |
+
"learning_rate": 0.0005029622551361681,
|
| 5301 |
+
"loss": 2.19,
|
| 5302 |
+
"step": 877
|
| 5303 |
+
},
|
| 5304 |
+
{
|
| 5305 |
+
"epoch": 0.6,
|
| 5306 |
+
"learning_rate": 0.0005028189202102246,
|
| 5307 |
+
"loss": 2.2216,
|
| 5308 |
+
"step": 878
|
| 5309 |
+
},
|
| 5310 |
+
{
|
| 5311 |
+
"epoch": 0.6,
|
| 5312 |
+
"learning_rate": 0.0005026755852842808,
|
| 5313 |
+
"loss": 2.1723,
|
| 5314 |
+
"step": 879
|
| 5315 |
+
},
|
| 5316 |
+
{
|
| 5317 |
+
"epoch": 0.6,
|
| 5318 |
+
"learning_rate": 0.0005025322503583372,
|
| 5319 |
+
"loss": 2.1189,
|
| 5320 |
+
"step": 880
|
| 5321 |
+
},
|
| 5322 |
+
{
|
| 5323 |
+
"epoch": 0.6,
|
| 5324 |
+
"learning_rate": 0.0005023889154323936,
|
| 5325 |
+
"loss": 2.1425,
|
| 5326 |
+
"step": 881
|
| 5327 |
+
},
|
| 5328 |
+
{
|
| 5329 |
+
"epoch": 0.6,
|
| 5330 |
+
"learning_rate": 0.00050224558050645,
|
| 5331 |
+
"loss": 2.2003,
|
| 5332 |
+
"step": 882
|
| 5333 |
+
},
|
| 5334 |
+
{
|
| 5335 |
+
"epoch": 0.6,
|
| 5336 |
+
"learning_rate": 0.0005021022455805064,
|
| 5337 |
+
"loss": 2.211,
|
| 5338 |
+
"step": 883
|
| 5339 |
+
},
|
| 5340 |
+
{
|
| 5341 |
+
"epoch": 0.6,
|
| 5342 |
+
"learning_rate": 0.0005019589106545628,
|
| 5343 |
+
"loss": 2.1302,
|
| 5344 |
+
"step": 884
|
| 5345 |
+
},
|
| 5346 |
+
{
|
| 5347 |
+
"epoch": 0.61,
|
| 5348 |
+
"learning_rate": 0.0005018155757286191,
|
| 5349 |
+
"loss": 2.1729,
|
| 5350 |
+
"step": 885
|
| 5351 |
+
},
|
| 5352 |
+
{
|
| 5353 |
+
"epoch": 0.61,
|
| 5354 |
+
"learning_rate": 0.0005016722408026755,
|
| 5355 |
+
"loss": 2.1203,
|
| 5356 |
+
"step": 886
|
| 5357 |
+
},
|
| 5358 |
+
{
|
| 5359 |
+
"epoch": 0.61,
|
| 5360 |
+
"learning_rate": 0.0005015289058767319,
|
| 5361 |
+
"loss": 2.0511,
|
| 5362 |
+
"step": 887
|
| 5363 |
+
},
|
| 5364 |
+
{
|
| 5365 |
+
"epoch": 0.61,
|
| 5366 |
+
"learning_rate": 0.0005013855709507883,
|
| 5367 |
+
"loss": 1.9758,
|
| 5368 |
+
"step": 888
|
| 5369 |
+
},
|
| 5370 |
+
{
|
| 5371 |
+
"epoch": 0.61,
|
| 5372 |
+
"learning_rate": 0.0005012422360248447,
|
| 5373 |
+
"loss": 2.1151,
|
| 5374 |
+
"step": 889
|
| 5375 |
+
},
|
| 5376 |
+
{
|
| 5377 |
+
"epoch": 0.61,
|
| 5378 |
+
"learning_rate": 0.0005010989010989011,
|
| 5379 |
+
"loss": 2.0998,
|
| 5380 |
+
"step": 890
|
| 5381 |
+
},
|
| 5382 |
+
{
|
| 5383 |
+
"epoch": 0.61,
|
| 5384 |
+
"learning_rate": 0.0005009555661729575,
|
| 5385 |
+
"loss": 2.0739,
|
| 5386 |
+
"step": 891
|
| 5387 |
+
},
|
| 5388 |
+
{
|
| 5389 |
+
"epoch": 0.61,
|
| 5390 |
+
"learning_rate": 0.0005008122312470138,
|
| 5391 |
+
"loss": 2.0421,
|
| 5392 |
+
"step": 892
|
| 5393 |
+
},
|
| 5394 |
+
{
|
| 5395 |
+
"epoch": 0.61,
|
| 5396 |
+
"learning_rate": 0.0005006688963210702,
|
| 5397 |
+
"loss": 2.0808,
|
| 5398 |
+
"step": 893
|
| 5399 |
+
},
|
| 5400 |
+
{
|
| 5401 |
+
"epoch": 0.61,
|
| 5402 |
+
"learning_rate": 0.0005005255613951265,
|
| 5403 |
+
"loss": 2.1301,
|
| 5404 |
+
"step": 894
|
| 5405 |
+
},
|
| 5406 |
+
{
|
| 5407 |
+
"epoch": 0.61,
|
| 5408 |
+
"learning_rate": 0.000500382226469183,
|
| 5409 |
+
"loss": 2.1207,
|
| 5410 |
+
"step": 895
|
| 5411 |
+
},
|
| 5412 |
+
{
|
| 5413 |
+
"epoch": 0.61,
|
| 5414 |
+
"learning_rate": 0.0005002388915432393,
|
| 5415 |
+
"loss": 2.1089,
|
| 5416 |
+
"step": 896
|
| 5417 |
+
},
|
| 5418 |
+
{
|
| 5419 |
+
"epoch": 0.61,
|
| 5420 |
+
"learning_rate": 0.0005000955566172958,
|
| 5421 |
+
"loss": 2.1796,
|
| 5422 |
+
"step": 897
|
| 5423 |
+
},
|
| 5424 |
+
{
|
| 5425 |
+
"epoch": 0.61,
|
| 5426 |
+
"learning_rate": 0.000499952221691352,
|
| 5427 |
+
"loss": 2.1123,
|
| 5428 |
+
"step": 898
|
| 5429 |
+
},
|
| 5430 |
+
{
|
| 5431 |
+
"epoch": 0.61,
|
| 5432 |
+
"learning_rate": 0.0004998088867654085,
|
| 5433 |
+
"loss": 2.1304,
|
| 5434 |
+
"step": 899
|
| 5435 |
+
},
|
| 5436 |
+
{
|
| 5437 |
+
"epoch": 0.62,
|
| 5438 |
+
"learning_rate": 0.0004996655518394648,
|
| 5439 |
+
"loss": 2.0691,
|
| 5440 |
+
"step": 900
|
| 5441 |
+
},
|
| 5442 |
+
{
|
| 5443 |
+
"epoch": 0.62,
|
| 5444 |
+
"learning_rate": 0.0004995222169135212,
|
| 5445 |
+
"loss": 2.215,
|
| 5446 |
+
"step": 901
|
| 5447 |
+
},
|
| 5448 |
+
{
|
| 5449 |
+
"epoch": 0.62,
|
| 5450 |
+
"learning_rate": 0.0004993788819875776,
|
| 5451 |
+
"loss": 2.1303,
|
| 5452 |
+
"step": 902
|
| 5453 |
+
},
|
| 5454 |
+
{
|
| 5455 |
+
"epoch": 0.62,
|
| 5456 |
+
"learning_rate": 0.000499235547061634,
|
| 5457 |
+
"loss": 2.0983,
|
| 5458 |
+
"step": 903
|
| 5459 |
+
},
|
| 5460 |
+
{
|
| 5461 |
+
"epoch": 0.62,
|
| 5462 |
+
"learning_rate": 0.0004990922121356903,
|
| 5463 |
+
"loss": 2.0761,
|
| 5464 |
+
"step": 904
|
| 5465 |
+
},
|
| 5466 |
+
{
|
| 5467 |
+
"epoch": 0.62,
|
| 5468 |
+
"learning_rate": 0.0004989488772097467,
|
| 5469 |
+
"loss": 2.1889,
|
| 5470 |
+
"step": 905
|
| 5471 |
+
},
|
| 5472 |
+
{
|
| 5473 |
+
"epoch": 0.62,
|
| 5474 |
+
"learning_rate": 0.0004988055422838031,
|
| 5475 |
+
"loss": 2.0557,
|
| 5476 |
+
"step": 906
|
| 5477 |
+
},
|
| 5478 |
+
{
|
| 5479 |
+
"epoch": 0.62,
|
| 5480 |
+
"learning_rate": 0.0004986622073578595,
|
| 5481 |
+
"loss": 2.1591,
|
| 5482 |
+
"step": 907
|
| 5483 |
+
},
|
| 5484 |
+
{
|
| 5485 |
+
"epoch": 0.62,
|
| 5486 |
+
"learning_rate": 0.0004985188724319159,
|
| 5487 |
+
"loss": 2.1217,
|
| 5488 |
+
"step": 908
|
| 5489 |
+
},
|
| 5490 |
+
{
|
| 5491 |
+
"epoch": 0.62,
|
| 5492 |
+
"learning_rate": 0.0004983755375059723,
|
| 5493 |
+
"loss": 2.0165,
|
| 5494 |
+
"step": 909
|
| 5495 |
+
},
|
| 5496 |
+
{
|
| 5497 |
+
"epoch": 0.62,
|
| 5498 |
+
"learning_rate": 0.0004982322025800286,
|
| 5499 |
+
"loss": 2.2102,
|
| 5500 |
+
"step": 910
|
| 5501 |
+
},
|
| 5502 |
+
{
|
| 5503 |
+
"epoch": 0.62,
|
| 5504 |
+
"learning_rate": 0.000498088867654085,
|
| 5505 |
+
"loss": 2.187,
|
| 5506 |
+
"step": 911
|
| 5507 |
+
},
|
| 5508 |
+
{
|
| 5509 |
+
"epoch": 0.62,
|
| 5510 |
+
"learning_rate": 0.0004979455327281414,
|
| 5511 |
+
"loss": 2.0758,
|
| 5512 |
+
"step": 912
|
| 5513 |
+
},
|
| 5514 |
+
{
|
| 5515 |
+
"epoch": 0.62,
|
| 5516 |
+
"learning_rate": 0.0004978021978021978,
|
| 5517 |
+
"loss": 2.2606,
|
| 5518 |
+
"step": 913
|
| 5519 |
+
},
|
| 5520 |
+
{
|
| 5521 |
+
"epoch": 0.62,
|
| 5522 |
+
"learning_rate": 0.0004976588628762542,
|
| 5523 |
+
"loss": 1.9795,
|
| 5524 |
+
"step": 914
|
| 5525 |
+
},
|
| 5526 |
+
{
|
| 5527 |
+
"epoch": 0.63,
|
| 5528 |
+
"learning_rate": 0.0004975155279503104,
|
| 5529 |
+
"loss": 2.1235,
|
| 5530 |
+
"step": 915
|
| 5531 |
+
},
|
| 5532 |
+
{
|
| 5533 |
+
"epoch": 0.63,
|
| 5534 |
+
"learning_rate": 0.0004973721930243669,
|
| 5535 |
+
"loss": 2.2042,
|
| 5536 |
+
"step": 916
|
| 5537 |
+
},
|
| 5538 |
+
{
|
| 5539 |
+
"epoch": 0.63,
|
| 5540 |
+
"learning_rate": 0.0004972288580984232,
|
| 5541 |
+
"loss": 2.1169,
|
| 5542 |
+
"step": 917
|
| 5543 |
+
},
|
| 5544 |
+
{
|
| 5545 |
+
"epoch": 0.63,
|
| 5546 |
+
"learning_rate": 0.0004970855231724797,
|
| 5547 |
+
"loss": 2.082,
|
| 5548 |
+
"step": 918
|
| 5549 |
+
},
|
| 5550 |
+
{
|
| 5551 |
+
"epoch": 0.63,
|
| 5552 |
+
"learning_rate": 0.000496942188246536,
|
| 5553 |
+
"loss": 2.0924,
|
| 5554 |
+
"step": 919
|
| 5555 |
+
},
|
| 5556 |
+
{
|
| 5557 |
+
"epoch": 0.63,
|
| 5558 |
+
"learning_rate": 0.0004967988533205925,
|
| 5559 |
+
"loss": 2.1969,
|
| 5560 |
+
"step": 920
|
| 5561 |
+
},
|
| 5562 |
+
{
|
| 5563 |
+
"epoch": 0.63,
|
| 5564 |
+
"learning_rate": 0.0004966555183946487,
|
| 5565 |
+
"loss": 2.1238,
|
| 5566 |
+
"step": 921
|
| 5567 |
+
},
|
| 5568 |
+
{
|
| 5569 |
+
"epoch": 0.63,
|
| 5570 |
+
"learning_rate": 0.0004965121834687051,
|
| 5571 |
+
"loss": 2.1776,
|
| 5572 |
+
"step": 922
|
| 5573 |
+
},
|
| 5574 |
+
{
|
| 5575 |
+
"epoch": 0.63,
|
| 5576 |
+
"learning_rate": 0.0004963688485427615,
|
| 5577 |
+
"loss": 2.1328,
|
| 5578 |
+
"step": 923
|
| 5579 |
+
},
|
| 5580 |
+
{
|
| 5581 |
+
"epoch": 0.63,
|
| 5582 |
+
"learning_rate": 0.0004962255136168179,
|
| 5583 |
+
"loss": 2.0723,
|
| 5584 |
+
"step": 924
|
| 5585 |
+
},
|
| 5586 |
+
{
|
| 5587 |
+
"epoch": 0.63,
|
| 5588 |
+
"learning_rate": 0.0004960821786908743,
|
| 5589 |
+
"loss": 2.2225,
|
| 5590 |
+
"step": 925
|
| 5591 |
+
},
|
| 5592 |
+
{
|
| 5593 |
+
"epoch": 0.63,
|
| 5594 |
+
"learning_rate": 0.0004959388437649307,
|
| 5595 |
+
"loss": 2.1698,
|
| 5596 |
+
"step": 926
|
| 5597 |
+
},
|
| 5598 |
+
{
|
| 5599 |
+
"epoch": 0.63,
|
| 5600 |
+
"learning_rate": 0.000495795508838987,
|
| 5601 |
+
"loss": 2.1764,
|
| 5602 |
+
"step": 927
|
| 5603 |
+
},
|
| 5604 |
+
{
|
| 5605 |
+
"epoch": 0.63,
|
| 5606 |
+
"learning_rate": 0.0004956521739130434,
|
| 5607 |
+
"loss": 2.2167,
|
| 5608 |
+
"step": 928
|
| 5609 |
+
},
|
| 5610 |
+
{
|
| 5611 |
+
"epoch": 0.64,
|
| 5612 |
+
"learning_rate": 0.0004955088389870998,
|
| 5613 |
+
"loss": 2.2408,
|
| 5614 |
+
"step": 929
|
| 5615 |
+
},
|
| 5616 |
+
{
|
| 5617 |
+
"epoch": 0.64,
|
| 5618 |
+
"learning_rate": 0.0004953655040611562,
|
| 5619 |
+
"loss": 2.049,
|
| 5620 |
+
"step": 930
|
| 5621 |
+
},
|
| 5622 |
+
{
|
| 5623 |
+
"epoch": 0.64,
|
| 5624 |
+
"learning_rate": 0.0004952221691352126,
|
| 5625 |
+
"loss": 2.0461,
|
| 5626 |
+
"step": 931
|
| 5627 |
+
},
|
| 5628 |
+
{
|
| 5629 |
+
"epoch": 0.64,
|
| 5630 |
+
"learning_rate": 0.000495078834209269,
|
| 5631 |
+
"loss": 2.1895,
|
| 5632 |
+
"step": 932
|
| 5633 |
+
},
|
| 5634 |
+
{
|
| 5635 |
+
"epoch": 0.64,
|
| 5636 |
+
"learning_rate": 0.0004949354992833254,
|
| 5637 |
+
"loss": 2.1358,
|
| 5638 |
+
"step": 933
|
| 5639 |
+
},
|
| 5640 |
+
{
|
| 5641 |
+
"epoch": 0.64,
|
| 5642 |
+
"learning_rate": 0.0004947921643573817,
|
| 5643 |
+
"loss": 2.1586,
|
| 5644 |
+
"step": 934
|
| 5645 |
+
},
|
| 5646 |
+
{
|
| 5647 |
+
"epoch": 0.64,
|
| 5648 |
+
"learning_rate": 0.0004946488294314381,
|
| 5649 |
+
"loss": 2.0463,
|
| 5650 |
+
"step": 935
|
| 5651 |
+
},
|
| 5652 |
+
{
|
| 5653 |
+
"epoch": 0.64,
|
| 5654 |
+
"learning_rate": 0.0004945054945054944,
|
| 5655 |
+
"loss": 2.006,
|
| 5656 |
+
"step": 936
|
| 5657 |
+
},
|
| 5658 |
+
{
|
| 5659 |
+
"epoch": 0.64,
|
| 5660 |
+
"learning_rate": 0.0004943621595795509,
|
| 5661 |
+
"loss": 2.0742,
|
| 5662 |
+
"step": 937
|
| 5663 |
+
},
|
| 5664 |
+
{
|
| 5665 |
+
"epoch": 0.64,
|
| 5666 |
+
"learning_rate": 0.0004942188246536072,
|
| 5667 |
+
"loss": 2.2008,
|
| 5668 |
+
"step": 938
|
| 5669 |
+
},
|
| 5670 |
+
{
|
| 5671 |
+
"epoch": 0.64,
|
| 5672 |
+
"learning_rate": 0.0004940754897276637,
|
| 5673 |
+
"loss": 2.1771,
|
| 5674 |
+
"step": 939
|
| 5675 |
+
},
|
| 5676 |
+
{
|
| 5677 |
+
"epoch": 0.64,
|
| 5678 |
+
"learning_rate": 0.0004939321548017199,
|
| 5679 |
+
"loss": 2.1418,
|
| 5680 |
+
"step": 940
|
| 5681 |
+
},
|
| 5682 |
+
{
|
| 5683 |
+
"epoch": 0.64,
|
| 5684 |
+
"learning_rate": 0.0004937888198757764,
|
| 5685 |
+
"loss": 2.1089,
|
| 5686 |
+
"step": 941
|
| 5687 |
+
},
|
| 5688 |
+
{
|
| 5689 |
+
"epoch": 0.64,
|
| 5690 |
+
"learning_rate": 0.0004936454849498327,
|
| 5691 |
+
"loss": 2.1641,
|
| 5692 |
+
"step": 942
|
| 5693 |
+
},
|
| 5694 |
+
{
|
| 5695 |
+
"epoch": 0.64,
|
| 5696 |
+
"learning_rate": 0.0004935021500238891,
|
| 5697 |
+
"loss": 2.2123,
|
| 5698 |
+
"step": 943
|
| 5699 |
+
},
|
| 5700 |
+
{
|
| 5701 |
+
"epoch": 0.65,
|
| 5702 |
+
"learning_rate": 0.0004933588150979455,
|
| 5703 |
+
"loss": 2.083,
|
| 5704 |
+
"step": 944
|
| 5705 |
+
},
|
| 5706 |
+
{
|
| 5707 |
+
"epoch": 0.65,
|
| 5708 |
+
"learning_rate": 0.0004932154801720019,
|
| 5709 |
+
"loss": 2.0856,
|
| 5710 |
+
"step": 945
|
| 5711 |
+
},
|
| 5712 |
+
{
|
| 5713 |
+
"epoch": 0.65,
|
| 5714 |
+
"learning_rate": 0.0004930721452460582,
|
| 5715 |
+
"loss": 2.0576,
|
| 5716 |
+
"step": 946
|
| 5717 |
+
},
|
| 5718 |
+
{
|
| 5719 |
+
"epoch": 0.65,
|
| 5720 |
+
"learning_rate": 0.0004929288103201146,
|
| 5721 |
+
"loss": 1.9976,
|
| 5722 |
+
"step": 947
|
| 5723 |
+
},
|
| 5724 |
+
{
|
| 5725 |
+
"epoch": 0.65,
|
| 5726 |
+
"learning_rate": 0.000492785475394171,
|
| 5727 |
+
"loss": 2.1001,
|
| 5728 |
+
"step": 948
|
| 5729 |
+
},
|
| 5730 |
+
{
|
| 5731 |
+
"epoch": 0.65,
|
| 5732 |
+
"learning_rate": 0.0004926421404682274,
|
| 5733 |
+
"loss": 2.1942,
|
| 5734 |
+
"step": 949
|
| 5735 |
+
},
|
| 5736 |
+
{
|
| 5737 |
+
"epoch": 0.65,
|
| 5738 |
+
"learning_rate": 0.0004924988055422838,
|
| 5739 |
+
"loss": 2.0722,
|
| 5740 |
+
"step": 950
|
| 5741 |
+
},
|
| 5742 |
+
{
|
| 5743 |
+
"epoch": 0.65,
|
| 5744 |
+
"learning_rate": 0.0004923554706163402,
|
| 5745 |
+
"loss": 2.1542,
|
| 5746 |
+
"step": 951
|
| 5747 |
+
},
|
| 5748 |
+
{
|
| 5749 |
+
"epoch": 0.65,
|
| 5750 |
+
"learning_rate": 0.0004922121356903965,
|
| 5751 |
+
"loss": 2.1446,
|
| 5752 |
+
"step": 952
|
| 5753 |
+
},
|
| 5754 |
+
{
|
| 5755 |
+
"epoch": 0.65,
|
| 5756 |
+
"learning_rate": 0.0004920688007644529,
|
| 5757 |
+
"loss": 2.1166,
|
| 5758 |
+
"step": 953
|
| 5759 |
+
},
|
| 5760 |
+
{
|
| 5761 |
+
"epoch": 0.65,
|
| 5762 |
+
"learning_rate": 0.0004919254658385093,
|
| 5763 |
+
"loss": 2.1406,
|
| 5764 |
+
"step": 954
|
| 5765 |
+
},
|
| 5766 |
+
{
|
| 5767 |
+
"epoch": 0.65,
|
| 5768 |
+
"learning_rate": 0.0004917821309125656,
|
| 5769 |
+
"loss": 2.1346,
|
| 5770 |
+
"step": 955
|
| 5771 |
+
},
|
| 5772 |
+
{
|
| 5773 |
+
"epoch": 0.65,
|
| 5774 |
+
"learning_rate": 0.0004916387959866221,
|
| 5775 |
+
"loss": 2.063,
|
| 5776 |
+
"step": 956
|
| 5777 |
+
},
|
| 5778 |
+
{
|
| 5779 |
+
"epoch": 0.65,
|
| 5780 |
+
"learning_rate": 0.0004914954610606783,
|
| 5781 |
+
"loss": 2.0332,
|
| 5782 |
+
"step": 957
|
| 5783 |
+
},
|
| 5784 |
+
{
|
| 5785 |
+
"epoch": 0.65,
|
| 5786 |
+
"learning_rate": 0.0004913521261347348,
|
| 5787 |
+
"loss": 2.0845,
|
| 5788 |
+
"step": 958
|
| 5789 |
+
},
|
| 5790 |
+
{
|
| 5791 |
+
"epoch": 0.66,
|
| 5792 |
+
"learning_rate": 0.0004912087912087911,
|
| 5793 |
+
"loss": 2.2177,
|
| 5794 |
+
"step": 959
|
| 5795 |
+
},
|
| 5796 |
+
{
|
| 5797 |
+
"epoch": 0.66,
|
| 5798 |
+
"learning_rate": 0.0004910654562828476,
|
| 5799 |
+
"loss": 2.1375,
|
| 5800 |
+
"step": 960
|
| 5801 |
+
},
|
| 5802 |
+
{
|
| 5803 |
+
"epoch": 0.66,
|
| 5804 |
+
"learning_rate": 0.0004909221213569039,
|
| 5805 |
+
"loss": 2.1171,
|
| 5806 |
+
"step": 961
|
| 5807 |
+
},
|
| 5808 |
+
{
|
| 5809 |
+
"epoch": 0.66,
|
| 5810 |
+
"learning_rate": 0.0004907787864309603,
|
| 5811 |
+
"loss": 2.0483,
|
| 5812 |
+
"step": 962
|
| 5813 |
+
},
|
| 5814 |
+
{
|
| 5815 |
+
"epoch": 0.66,
|
| 5816 |
+
"learning_rate": 0.0004906354515050167,
|
| 5817 |
+
"loss": 2.1325,
|
| 5818 |
+
"step": 963
|
| 5819 |
+
},
|
| 5820 |
+
{
|
| 5821 |
+
"epoch": 0.66,
|
| 5822 |
+
"learning_rate": 0.000490492116579073,
|
| 5823 |
+
"loss": 2.0789,
|
| 5824 |
+
"step": 964
|
| 5825 |
+
},
|
| 5826 |
+
{
|
| 5827 |
+
"epoch": 0.66,
|
| 5828 |
+
"learning_rate": 0.0004903487816531294,
|
| 5829 |
+
"loss": 2.1548,
|
| 5830 |
+
"step": 965
|
| 5831 |
+
},
|
| 5832 |
+
{
|
| 5833 |
+
"epoch": 0.66,
|
| 5834 |
+
"learning_rate": 0.0004902054467271858,
|
| 5835 |
+
"loss": 2.2216,
|
| 5836 |
+
"step": 966
|
| 5837 |
+
},
|
| 5838 |
+
{
|
| 5839 |
+
"epoch": 0.66,
|
| 5840 |
+
"learning_rate": 0.0004900621118012422,
|
| 5841 |
+
"loss": 2.0969,
|
| 5842 |
+
"step": 967
|
| 5843 |
+
},
|
| 5844 |
+
{
|
| 5845 |
+
"epoch": 0.66,
|
| 5846 |
+
"learning_rate": 0.0004899187768752986,
|
| 5847 |
+
"loss": 2.025,
|
| 5848 |
+
"step": 968
|
| 5849 |
+
},
|
| 5850 |
+
{
|
| 5851 |
+
"epoch": 0.66,
|
| 5852 |
+
"learning_rate": 0.000489775441949355,
|
| 5853 |
+
"loss": 2.2016,
|
| 5854 |
+
"step": 969
|
| 5855 |
+
},
|
| 5856 |
+
{
|
| 5857 |
+
"epoch": 0.66,
|
| 5858 |
+
"learning_rate": 0.0004896321070234113,
|
| 5859 |
+
"loss": 2.1138,
|
| 5860 |
+
"step": 970
|
| 5861 |
+
},
|
| 5862 |
+
{
|
| 5863 |
+
"epoch": 0.66,
|
| 5864 |
+
"learning_rate": 0.0004894887720974677,
|
| 5865 |
+
"loss": 2.0992,
|
| 5866 |
+
"step": 971
|
| 5867 |
+
},
|
| 5868 |
+
{
|
| 5869 |
+
"epoch": 0.66,
|
| 5870 |
+
"learning_rate": 0.0004893454371715241,
|
| 5871 |
+
"loss": 2.1381,
|
| 5872 |
+
"step": 972
|
| 5873 |
+
},
|
| 5874 |
+
{
|
| 5875 |
+
"epoch": 0.67,
|
| 5876 |
+
"learning_rate": 0.0004892021022455805,
|
| 5877 |
+
"loss": 2.0633,
|
| 5878 |
+
"step": 973
|
| 5879 |
+
},
|
| 5880 |
+
{
|
| 5881 |
+
"epoch": 0.67,
|
| 5882 |
+
"learning_rate": 0.0004890587673196369,
|
| 5883 |
+
"loss": 2.1169,
|
| 5884 |
+
"step": 974
|
| 5885 |
+
},
|
| 5886 |
+
{
|
| 5887 |
+
"epoch": 0.67,
|
| 5888 |
+
"learning_rate": 0.0004889154323936933,
|
| 5889 |
+
"loss": 2.0574,
|
| 5890 |
+
"step": 975
|
| 5891 |
+
},
|
| 5892 |
+
{
|
| 5893 |
+
"epoch": 0.67,
|
| 5894 |
+
"learning_rate": 0.0004887720974677495,
|
| 5895 |
+
"loss": 2.1371,
|
| 5896 |
+
"step": 976
|
| 5897 |
+
},
|
| 5898 |
+
{
|
| 5899 |
+
"epoch": 0.67,
|
| 5900 |
+
"learning_rate": 0.000488628762541806,
|
| 5901 |
+
"loss": 2.178,
|
| 5902 |
+
"step": 977
|
| 5903 |
+
},
|
| 5904 |
+
{
|
| 5905 |
+
"epoch": 0.67,
|
| 5906 |
+
"learning_rate": 0.0004884854276158623,
|
| 5907 |
+
"loss": 2.1797,
|
| 5908 |
+
"step": 978
|
| 5909 |
+
},
|
| 5910 |
+
{
|
| 5911 |
+
"epoch": 0.67,
|
| 5912 |
+
"learning_rate": 0.0004883420926899188,
|
| 5913 |
+
"loss": 2.0527,
|
| 5914 |
+
"step": 979
|
| 5915 |
+
},
|
| 5916 |
+
{
|
| 5917 |
+
"epoch": 0.67,
|
| 5918 |
+
"learning_rate": 0.0004881987577639751,
|
| 5919 |
+
"loss": 2.2112,
|
| 5920 |
+
"step": 980
|
| 5921 |
+
},
|
| 5922 |
+
{
|
| 5923 |
+
"epoch": 0.67,
|
| 5924 |
+
"learning_rate": 0.0004880554228380315,
|
| 5925 |
+
"loss": 2.1351,
|
| 5926 |
+
"step": 981
|
| 5927 |
+
},
|
| 5928 |
+
{
|
| 5929 |
+
"epoch": 0.67,
|
| 5930 |
+
"learning_rate": 0.0004879120879120879,
|
| 5931 |
+
"loss": 2.0178,
|
| 5932 |
+
"step": 982
|
| 5933 |
+
},
|
| 5934 |
+
{
|
| 5935 |
+
"epoch": 0.67,
|
| 5936 |
+
"learning_rate": 0.0004877687529861442,
|
| 5937 |
+
"loss": 2.1084,
|
| 5938 |
+
"step": 983
|
| 5939 |
+
},
|
| 5940 |
+
{
|
| 5941 |
+
"epoch": 0.67,
|
| 5942 |
+
"learning_rate": 0.00048762541806020066,
|
| 5943 |
+
"loss": 2.0922,
|
| 5944 |
+
"step": 984
|
| 5945 |
+
},
|
| 5946 |
+
{
|
| 5947 |
+
"epoch": 0.67,
|
| 5948 |
+
"learning_rate": 0.000487482083134257,
|
| 5949 |
+
"loss": 2.027,
|
| 5950 |
+
"step": 985
|
| 5951 |
+
},
|
| 5952 |
+
{
|
| 5953 |
+
"epoch": 0.67,
|
| 5954 |
+
"learning_rate": 0.0004873387482083134,
|
| 5955 |
+
"loss": 2.1061,
|
| 5956 |
+
"step": 986
|
| 5957 |
+
},
|
| 5958 |
+
{
|
| 5959 |
+
"epoch": 0.67,
|
| 5960 |
+
"learning_rate": 0.00048719541328236975,
|
| 5961 |
+
"loss": 2.0671,
|
| 5962 |
+
"step": 987
|
| 5963 |
+
},
|
| 5964 |
+
{
|
| 5965 |
+
"epoch": 0.68,
|
| 5966 |
+
"learning_rate": 0.0004870520783564262,
|
| 5967 |
+
"loss": 2.112,
|
| 5968 |
+
"step": 988
|
| 5969 |
+
},
|
| 5970 |
+
{
|
| 5971 |
+
"epoch": 0.68,
|
| 5972 |
+
"learning_rate": 0.0004869087434304825,
|
| 5973 |
+
"loss": 2.1656,
|
| 5974 |
+
"step": 989
|
| 5975 |
+
},
|
| 5976 |
+
{
|
| 5977 |
+
"epoch": 0.68,
|
| 5978 |
+
"learning_rate": 0.00048676540850453885,
|
| 5979 |
+
"loss": 2.0472,
|
| 5980 |
+
"step": 990
|
| 5981 |
+
},
|
| 5982 |
+
{
|
| 5983 |
+
"epoch": 0.68,
|
| 5984 |
+
"learning_rate": 0.0004866220735785953,
|
| 5985 |
+
"loss": 2.0411,
|
| 5986 |
+
"step": 991
|
| 5987 |
+
},
|
| 5988 |
+
{
|
| 5989 |
+
"epoch": 0.68,
|
| 5990 |
+
"learning_rate": 0.0004864787386526516,
|
| 5991 |
+
"loss": 2.0884,
|
| 5992 |
+
"step": 992
|
| 5993 |
+
},
|
| 5994 |
+
{
|
| 5995 |
+
"epoch": 0.68,
|
| 5996 |
+
"learning_rate": 0.00048633540372670806,
|
| 5997 |
+
"loss": 2.1037,
|
| 5998 |
+
"step": 993
|
| 5999 |
+
},
|
| 6000 |
+
{
|
| 6001 |
+
"epoch": 0.68,
|
| 6002 |
+
"learning_rate": 0.0004861920688007644,
|
| 6003 |
+
"loss": 2.1009,
|
| 6004 |
+
"step": 994
|
| 6005 |
+
},
|
| 6006 |
+
{
|
| 6007 |
+
"epoch": 0.68,
|
| 6008 |
+
"learning_rate": 0.0004860487338748208,
|
| 6009 |
+
"loss": 2.1671,
|
| 6010 |
+
"step": 995
|
| 6011 |
+
},
|
| 6012 |
+
{
|
| 6013 |
+
"epoch": 0.68,
|
| 6014 |
+
"learning_rate": 0.00048590539894887715,
|
| 6015 |
+
"loss": 2.0654,
|
| 6016 |
+
"step": 996
|
| 6017 |
+
},
|
| 6018 |
+
{
|
| 6019 |
+
"epoch": 0.68,
|
| 6020 |
+
"learning_rate": 0.00048576206402293354,
|
| 6021 |
+
"loss": 2.0784,
|
| 6022 |
+
"step": 997
|
| 6023 |
+
},
|
| 6024 |
+
{
|
| 6025 |
+
"epoch": 0.68,
|
| 6026 |
+
"learning_rate": 0.0004856187290969899,
|
| 6027 |
+
"loss": 2.1642,
|
| 6028 |
+
"step": 998
|
| 6029 |
+
},
|
| 6030 |
+
{
|
| 6031 |
+
"epoch": 0.68,
|
| 6032 |
+
"learning_rate": 0.0004854753941710463,
|
| 6033 |
+
"loss": 2.1117,
|
| 6034 |
+
"step": 999
|
| 6035 |
+
},
|
| 6036 |
+
{
|
| 6037 |
+
"epoch": 0.68,
|
| 6038 |
+
"learning_rate": 0.0004853320592451027,
|
| 6039 |
+
"loss": 2.0927,
|
| 6040 |
+
"step": 1000
|
| 6041 |
+
},
|
| 6042 |
+
{
|
| 6043 |
+
"epoch": 0.68,
|
| 6044 |
+
"eval_loss": 2.0208685398101807,
|
| 6045 |
+
"eval_runtime": 1688.1492,
|
| 6046 |
+
"eval_samples_per_second": 5.924,
|
| 6047 |
+
"eval_steps_per_second": 5.924,
|
| 6048 |
+
"step": 1000
|
| 6049 |
}
|
| 6050 |
],
|
| 6051 |
"max_steps": 4386,
|
| 6052 |
"num_train_epochs": 3,
|
| 6053 |
+
"total_flos": 1.2243065569981379e+18,
|
| 6054 |
"trial_name": null,
|
| 6055 |
"trial_params": null
|
| 6056 |
}
|
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:cc924070c94c5b619e2669e03e1453bcf47c8ac1d6edf146fae93888c7686dcc
|
| 3 |
size 2368281769
|