Training in progress, step 800
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 sha256:2b93cb0f5af4a676996d113ae67c14903f845be14efc3d75962cc5c86990b4be
|
| 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 sha256:6b06061cdd59a61c03b74896e78f938e36d6d587093dddf2c3beb4c518798564
|
| 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 sha256:78177ec8a4e9181f496a71815f95534c1ccdd07dd8b38c39977074346212cd45
|
| 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:bc38c7f71de03e75da249f8cf736366cbbae8af7a495e0547a8a03a22691e8a0
|
| 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,
|
|
@@ -3630,11 +3630,1219 @@
|
|
| 3630 |
"eval_samples_per_second": 5.759,
|
| 3631 |
"eval_steps_per_second": 5.759,
|
| 3632 |
"step": 600
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3633 |
}
|
| 3634 |
],
|
| 3635 |
"max_steps": 4386,
|
| 3636 |
"num_train_epochs": 3,
|
| 3637 |
-
"total_flos":
|
| 3638 |
"trial_name": null,
|
| 3639 |
"trial_params": null
|
| 3640 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 2.0316452980041504,
|
| 3 |
+
"best_model_checkpoint": "output/checkpoint-800",
|
| 4 |
+
"epoch": 0.5469471907531741,
|
| 5 |
+
"global_step": 800,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 3630 |
"eval_samples_per_second": 5.759,
|
| 3631 |
"eval_steps_per_second": 5.759,
|
| 3632 |
"step": 600
|
| 3633 |
+
},
|
| 3634 |
+
{
|
| 3635 |
+
"epoch": 0.41,
|
| 3636 |
+
"learning_rate": 0.0005425226946966077,
|
| 3637 |
+
"loss": 2.0728,
|
| 3638 |
+
"step": 601
|
| 3639 |
+
},
|
| 3640 |
+
{
|
| 3641 |
+
"epoch": 0.41,
|
| 3642 |
+
"learning_rate": 0.0005423793597706641,
|
| 3643 |
+
"loss": 2.051,
|
| 3644 |
+
"step": 602
|
| 3645 |
+
},
|
| 3646 |
+
{
|
| 3647 |
+
"epoch": 0.41,
|
| 3648 |
+
"learning_rate": 0.0005422360248447204,
|
| 3649 |
+
"loss": 2.1657,
|
| 3650 |
+
"step": 603
|
| 3651 |
+
},
|
| 3652 |
+
{
|
| 3653 |
+
"epoch": 0.41,
|
| 3654 |
+
"learning_rate": 0.0005420926899187768,
|
| 3655 |
+
"loss": 2.1084,
|
| 3656 |
+
"step": 604
|
| 3657 |
+
},
|
| 3658 |
+
{
|
| 3659 |
+
"epoch": 0.41,
|
| 3660 |
+
"learning_rate": 0.0005419493549928332,
|
| 3661 |
+
"loss": 2.1312,
|
| 3662 |
+
"step": 605
|
| 3663 |
+
},
|
| 3664 |
+
{
|
| 3665 |
+
"epoch": 0.41,
|
| 3666 |
+
"learning_rate": 0.0005418060200668896,
|
| 3667 |
+
"loss": 2.0724,
|
| 3668 |
+
"step": 606
|
| 3669 |
+
},
|
| 3670 |
+
{
|
| 3671 |
+
"epoch": 0.41,
|
| 3672 |
+
"learning_rate": 0.000541662685140946,
|
| 3673 |
+
"loss": 2.0999,
|
| 3674 |
+
"step": 607
|
| 3675 |
+
},
|
| 3676 |
+
{
|
| 3677 |
+
"epoch": 0.42,
|
| 3678 |
+
"learning_rate": 0.0005415193502150024,
|
| 3679 |
+
"loss": 2.1649,
|
| 3680 |
+
"step": 608
|
| 3681 |
+
},
|
| 3682 |
+
{
|
| 3683 |
+
"epoch": 0.42,
|
| 3684 |
+
"learning_rate": 0.0005413760152890587,
|
| 3685 |
+
"loss": 2.1458,
|
| 3686 |
+
"step": 609
|
| 3687 |
+
},
|
| 3688 |
+
{
|
| 3689 |
+
"epoch": 0.42,
|
| 3690 |
+
"learning_rate": 0.0005412326803631151,
|
| 3691 |
+
"loss": 2.071,
|
| 3692 |
+
"step": 610
|
| 3693 |
+
},
|
| 3694 |
+
{
|
| 3695 |
+
"epoch": 0.42,
|
| 3696 |
+
"learning_rate": 0.0005410893454371715,
|
| 3697 |
+
"loss": 2.1931,
|
| 3698 |
+
"step": 611
|
| 3699 |
+
},
|
| 3700 |
+
{
|
| 3701 |
+
"epoch": 0.42,
|
| 3702 |
+
"learning_rate": 0.0005409460105112279,
|
| 3703 |
+
"loss": 2.1968,
|
| 3704 |
+
"step": 612
|
| 3705 |
+
},
|
| 3706 |
+
{
|
| 3707 |
+
"epoch": 0.42,
|
| 3708 |
+
"learning_rate": 0.0005408026755852842,
|
| 3709 |
+
"loss": 2.105,
|
| 3710 |
+
"step": 613
|
| 3711 |
+
},
|
| 3712 |
+
{
|
| 3713 |
+
"epoch": 0.42,
|
| 3714 |
+
"learning_rate": 0.0005406593406593407,
|
| 3715 |
+
"loss": 2.1429,
|
| 3716 |
+
"step": 614
|
| 3717 |
+
},
|
| 3718 |
+
{
|
| 3719 |
+
"epoch": 0.42,
|
| 3720 |
+
"learning_rate": 0.0005405160057333969,
|
| 3721 |
+
"loss": 2.0812,
|
| 3722 |
+
"step": 615
|
| 3723 |
+
},
|
| 3724 |
+
{
|
| 3725 |
+
"epoch": 0.42,
|
| 3726 |
+
"learning_rate": 0.0005403726708074534,
|
| 3727 |
+
"loss": 2.1389,
|
| 3728 |
+
"step": 616
|
| 3729 |
+
},
|
| 3730 |
+
{
|
| 3731 |
+
"epoch": 0.42,
|
| 3732 |
+
"learning_rate": 0.0005402293358815097,
|
| 3733 |
+
"loss": 2.1685,
|
| 3734 |
+
"step": 617
|
| 3735 |
+
},
|
| 3736 |
+
{
|
| 3737 |
+
"epoch": 0.42,
|
| 3738 |
+
"learning_rate": 0.0005400860009555662,
|
| 3739 |
+
"loss": 2.2248,
|
| 3740 |
+
"step": 618
|
| 3741 |
+
},
|
| 3742 |
+
{
|
| 3743 |
+
"epoch": 0.42,
|
| 3744 |
+
"learning_rate": 0.0005399426660296225,
|
| 3745 |
+
"loss": 2.1771,
|
| 3746 |
+
"step": 619
|
| 3747 |
+
},
|
| 3748 |
+
{
|
| 3749 |
+
"epoch": 0.42,
|
| 3750 |
+
"learning_rate": 0.0005397993311036789,
|
| 3751 |
+
"loss": 2.1094,
|
| 3752 |
+
"step": 620
|
| 3753 |
+
},
|
| 3754 |
+
{
|
| 3755 |
+
"epoch": 0.42,
|
| 3756 |
+
"learning_rate": 0.0005396559961777352,
|
| 3757 |
+
"loss": 2.141,
|
| 3758 |
+
"step": 621
|
| 3759 |
+
},
|
| 3760 |
+
{
|
| 3761 |
+
"epoch": 0.43,
|
| 3762 |
+
"learning_rate": 0.0005395126612517916,
|
| 3763 |
+
"loss": 2.2442,
|
| 3764 |
+
"step": 622
|
| 3765 |
+
},
|
| 3766 |
+
{
|
| 3767 |
+
"epoch": 0.43,
|
| 3768 |
+
"learning_rate": 0.000539369326325848,
|
| 3769 |
+
"loss": 2.1703,
|
| 3770 |
+
"step": 623
|
| 3771 |
+
},
|
| 3772 |
+
{
|
| 3773 |
+
"epoch": 0.43,
|
| 3774 |
+
"learning_rate": 0.0005392259913999044,
|
| 3775 |
+
"loss": 2.1351,
|
| 3776 |
+
"step": 624
|
| 3777 |
+
},
|
| 3778 |
+
{
|
| 3779 |
+
"epoch": 0.43,
|
| 3780 |
+
"learning_rate": 0.0005390826564739608,
|
| 3781 |
+
"loss": 2.2008,
|
| 3782 |
+
"step": 625
|
| 3783 |
+
},
|
| 3784 |
+
{
|
| 3785 |
+
"epoch": 0.43,
|
| 3786 |
+
"learning_rate": 0.0005389393215480172,
|
| 3787 |
+
"loss": 2.2114,
|
| 3788 |
+
"step": 626
|
| 3789 |
+
},
|
| 3790 |
+
{
|
| 3791 |
+
"epoch": 0.43,
|
| 3792 |
+
"learning_rate": 0.0005387959866220735,
|
| 3793 |
+
"loss": 2.1368,
|
| 3794 |
+
"step": 627
|
| 3795 |
+
},
|
| 3796 |
+
{
|
| 3797 |
+
"epoch": 0.43,
|
| 3798 |
+
"learning_rate": 0.0005386526516961299,
|
| 3799 |
+
"loss": 2.2127,
|
| 3800 |
+
"step": 628
|
| 3801 |
+
},
|
| 3802 |
+
{
|
| 3803 |
+
"epoch": 0.43,
|
| 3804 |
+
"learning_rate": 0.0005385093167701863,
|
| 3805 |
+
"loss": 2.2554,
|
| 3806 |
+
"step": 629
|
| 3807 |
+
},
|
| 3808 |
+
{
|
| 3809 |
+
"epoch": 0.43,
|
| 3810 |
+
"learning_rate": 0.0005383659818442427,
|
| 3811 |
+
"loss": 2.143,
|
| 3812 |
+
"step": 630
|
| 3813 |
+
},
|
| 3814 |
+
{
|
| 3815 |
+
"epoch": 0.43,
|
| 3816 |
+
"learning_rate": 0.0005382226469182991,
|
| 3817 |
+
"loss": 2.1306,
|
| 3818 |
+
"step": 631
|
| 3819 |
+
},
|
| 3820 |
+
{
|
| 3821 |
+
"epoch": 0.43,
|
| 3822 |
+
"learning_rate": 0.0005380793119923555,
|
| 3823 |
+
"loss": 2.1257,
|
| 3824 |
+
"step": 632
|
| 3825 |
+
},
|
| 3826 |
+
{
|
| 3827 |
+
"epoch": 0.43,
|
| 3828 |
+
"learning_rate": 0.0005379359770664118,
|
| 3829 |
+
"loss": 2.2013,
|
| 3830 |
+
"step": 633
|
| 3831 |
+
},
|
| 3832 |
+
{
|
| 3833 |
+
"epoch": 0.43,
|
| 3834 |
+
"learning_rate": 0.0005377926421404681,
|
| 3835 |
+
"loss": 2.1856,
|
| 3836 |
+
"step": 634
|
| 3837 |
+
},
|
| 3838 |
+
{
|
| 3839 |
+
"epoch": 0.43,
|
| 3840 |
+
"learning_rate": 0.0005376493072145246,
|
| 3841 |
+
"loss": 2.1917,
|
| 3842 |
+
"step": 635
|
| 3843 |
+
},
|
| 3844 |
+
{
|
| 3845 |
+
"epoch": 0.43,
|
| 3846 |
+
"learning_rate": 0.0005375059722885809,
|
| 3847 |
+
"loss": 2.1044,
|
| 3848 |
+
"step": 636
|
| 3849 |
+
},
|
| 3850 |
+
{
|
| 3851 |
+
"epoch": 0.44,
|
| 3852 |
+
"learning_rate": 0.0005373626373626374,
|
| 3853 |
+
"loss": 2.1909,
|
| 3854 |
+
"step": 637
|
| 3855 |
+
},
|
| 3856 |
+
{
|
| 3857 |
+
"epoch": 0.44,
|
| 3858 |
+
"learning_rate": 0.0005372193024366937,
|
| 3859 |
+
"loss": 2.0706,
|
| 3860 |
+
"step": 638
|
| 3861 |
+
},
|
| 3862 |
+
{
|
| 3863 |
+
"epoch": 0.44,
|
| 3864 |
+
"learning_rate": 0.0005370759675107501,
|
| 3865 |
+
"loss": 2.142,
|
| 3866 |
+
"step": 639
|
| 3867 |
+
},
|
| 3868 |
+
{
|
| 3869 |
+
"epoch": 0.44,
|
| 3870 |
+
"learning_rate": 0.0005369326325848064,
|
| 3871 |
+
"loss": 2.1813,
|
| 3872 |
+
"step": 640
|
| 3873 |
+
},
|
| 3874 |
+
{
|
| 3875 |
+
"epoch": 0.44,
|
| 3876 |
+
"learning_rate": 0.0005367892976588628,
|
| 3877 |
+
"loss": 2.1802,
|
| 3878 |
+
"step": 641
|
| 3879 |
+
},
|
| 3880 |
+
{
|
| 3881 |
+
"epoch": 0.44,
|
| 3882 |
+
"learning_rate": 0.0005366459627329192,
|
| 3883 |
+
"loss": 2.1461,
|
| 3884 |
+
"step": 642
|
| 3885 |
+
},
|
| 3886 |
+
{
|
| 3887 |
+
"epoch": 0.44,
|
| 3888 |
+
"learning_rate": 0.0005365026278069756,
|
| 3889 |
+
"loss": 2.045,
|
| 3890 |
+
"step": 643
|
| 3891 |
+
},
|
| 3892 |
+
{
|
| 3893 |
+
"epoch": 0.44,
|
| 3894 |
+
"learning_rate": 0.000536359292881032,
|
| 3895 |
+
"loss": 2.1449,
|
| 3896 |
+
"step": 644
|
| 3897 |
+
},
|
| 3898 |
+
{
|
| 3899 |
+
"epoch": 0.44,
|
| 3900 |
+
"learning_rate": 0.0005362159579550883,
|
| 3901 |
+
"loss": 2.0893,
|
| 3902 |
+
"step": 645
|
| 3903 |
+
},
|
| 3904 |
+
{
|
| 3905 |
+
"epoch": 0.44,
|
| 3906 |
+
"learning_rate": 0.0005360726230291447,
|
| 3907 |
+
"loss": 2.1477,
|
| 3908 |
+
"step": 646
|
| 3909 |
+
},
|
| 3910 |
+
{
|
| 3911 |
+
"epoch": 0.44,
|
| 3912 |
+
"learning_rate": 0.0005359292881032011,
|
| 3913 |
+
"loss": 2.172,
|
| 3914 |
+
"step": 647
|
| 3915 |
+
},
|
| 3916 |
+
{
|
| 3917 |
+
"epoch": 0.44,
|
| 3918 |
+
"learning_rate": 0.0005357859531772575,
|
| 3919 |
+
"loss": 2.2154,
|
| 3920 |
+
"step": 648
|
| 3921 |
+
},
|
| 3922 |
+
{
|
| 3923 |
+
"epoch": 0.44,
|
| 3924 |
+
"learning_rate": 0.0005356426182513139,
|
| 3925 |
+
"loss": 2.0651,
|
| 3926 |
+
"step": 649
|
| 3927 |
+
},
|
| 3928 |
+
{
|
| 3929 |
+
"epoch": 0.44,
|
| 3930 |
+
"learning_rate": 0.0005354992833253703,
|
| 3931 |
+
"loss": 2.2091,
|
| 3932 |
+
"step": 650
|
| 3933 |
+
},
|
| 3934 |
+
{
|
| 3935 |
+
"epoch": 0.45,
|
| 3936 |
+
"learning_rate": 0.0005353559483994266,
|
| 3937 |
+
"loss": 2.1892,
|
| 3938 |
+
"step": 651
|
| 3939 |
+
},
|
| 3940 |
+
{
|
| 3941 |
+
"epoch": 0.45,
|
| 3942 |
+
"learning_rate": 0.000535212613473483,
|
| 3943 |
+
"loss": 2.1727,
|
| 3944 |
+
"step": 652
|
| 3945 |
+
},
|
| 3946 |
+
{
|
| 3947 |
+
"epoch": 0.45,
|
| 3948 |
+
"learning_rate": 0.0005350692785475393,
|
| 3949 |
+
"loss": 2.1532,
|
| 3950 |
+
"step": 653
|
| 3951 |
+
},
|
| 3952 |
+
{
|
| 3953 |
+
"epoch": 0.45,
|
| 3954 |
+
"learning_rate": 0.0005349259436215958,
|
| 3955 |
+
"loss": 2.0929,
|
| 3956 |
+
"step": 654
|
| 3957 |
+
},
|
| 3958 |
+
{
|
| 3959 |
+
"epoch": 0.45,
|
| 3960 |
+
"learning_rate": 0.0005347826086956521,
|
| 3961 |
+
"loss": 2.1364,
|
| 3962 |
+
"step": 655
|
| 3963 |
+
},
|
| 3964 |
+
{
|
| 3965 |
+
"epoch": 0.45,
|
| 3966 |
+
"learning_rate": 0.0005346392737697086,
|
| 3967 |
+
"loss": 2.1704,
|
| 3968 |
+
"step": 656
|
| 3969 |
+
},
|
| 3970 |
+
{
|
| 3971 |
+
"epoch": 0.45,
|
| 3972 |
+
"learning_rate": 0.0005344959388437648,
|
| 3973 |
+
"loss": 2.116,
|
| 3974 |
+
"step": 657
|
| 3975 |
+
},
|
| 3976 |
+
{
|
| 3977 |
+
"epoch": 0.45,
|
| 3978 |
+
"learning_rate": 0.0005343526039178213,
|
| 3979 |
+
"loss": 2.2498,
|
| 3980 |
+
"step": 658
|
| 3981 |
+
},
|
| 3982 |
+
{
|
| 3983 |
+
"epoch": 0.45,
|
| 3984 |
+
"learning_rate": 0.0005342092689918776,
|
| 3985 |
+
"loss": 2.2484,
|
| 3986 |
+
"step": 659
|
| 3987 |
+
},
|
| 3988 |
+
{
|
| 3989 |
+
"epoch": 0.45,
|
| 3990 |
+
"learning_rate": 0.000534065934065934,
|
| 3991 |
+
"loss": 2.0954,
|
| 3992 |
+
"step": 660
|
| 3993 |
+
},
|
| 3994 |
+
{
|
| 3995 |
+
"epoch": 0.45,
|
| 3996 |
+
"learning_rate": 0.0005339225991399904,
|
| 3997 |
+
"loss": 2.1747,
|
| 3998 |
+
"step": 661
|
| 3999 |
+
},
|
| 4000 |
+
{
|
| 4001 |
+
"epoch": 0.45,
|
| 4002 |
+
"learning_rate": 0.0005337792642140468,
|
| 4003 |
+
"loss": 2.1775,
|
| 4004 |
+
"step": 662
|
| 4005 |
+
},
|
| 4006 |
+
{
|
| 4007 |
+
"epoch": 0.45,
|
| 4008 |
+
"learning_rate": 0.0005336359292881031,
|
| 4009 |
+
"loss": 2.2085,
|
| 4010 |
+
"step": 663
|
| 4011 |
+
},
|
| 4012 |
+
{
|
| 4013 |
+
"epoch": 0.45,
|
| 4014 |
+
"learning_rate": 0.0005334925943621595,
|
| 4015 |
+
"loss": 2.1576,
|
| 4016 |
+
"step": 664
|
| 4017 |
+
},
|
| 4018 |
+
{
|
| 4019 |
+
"epoch": 0.45,
|
| 4020 |
+
"learning_rate": 0.0005333492594362159,
|
| 4021 |
+
"loss": 2.1618,
|
| 4022 |
+
"step": 665
|
| 4023 |
+
},
|
| 4024 |
+
{
|
| 4025 |
+
"epoch": 0.46,
|
| 4026 |
+
"learning_rate": 0.0005332059245102723,
|
| 4027 |
+
"loss": 2.1005,
|
| 4028 |
+
"step": 666
|
| 4029 |
+
},
|
| 4030 |
+
{
|
| 4031 |
+
"epoch": 0.46,
|
| 4032 |
+
"learning_rate": 0.0005330625895843287,
|
| 4033 |
+
"loss": 2.1803,
|
| 4034 |
+
"step": 667
|
| 4035 |
+
},
|
| 4036 |
+
{
|
| 4037 |
+
"epoch": 0.46,
|
| 4038 |
+
"learning_rate": 0.0005329192546583851,
|
| 4039 |
+
"loss": 2.036,
|
| 4040 |
+
"step": 668
|
| 4041 |
+
},
|
| 4042 |
+
{
|
| 4043 |
+
"epoch": 0.46,
|
| 4044 |
+
"learning_rate": 0.0005327759197324414,
|
| 4045 |
+
"loss": 2.1275,
|
| 4046 |
+
"step": 669
|
| 4047 |
+
},
|
| 4048 |
+
{
|
| 4049 |
+
"epoch": 0.46,
|
| 4050 |
+
"learning_rate": 0.0005326325848064978,
|
| 4051 |
+
"loss": 2.1677,
|
| 4052 |
+
"step": 670
|
| 4053 |
+
},
|
| 4054 |
+
{
|
| 4055 |
+
"epoch": 0.46,
|
| 4056 |
+
"learning_rate": 0.0005324892498805542,
|
| 4057 |
+
"loss": 2.1649,
|
| 4058 |
+
"step": 671
|
| 4059 |
+
},
|
| 4060 |
+
{
|
| 4061 |
+
"epoch": 0.46,
|
| 4062 |
+
"learning_rate": 0.0005323459149546106,
|
| 4063 |
+
"loss": 2.1187,
|
| 4064 |
+
"step": 672
|
| 4065 |
+
},
|
| 4066 |
+
{
|
| 4067 |
+
"epoch": 0.46,
|
| 4068 |
+
"learning_rate": 0.000532202580028667,
|
| 4069 |
+
"loss": 2.0904,
|
| 4070 |
+
"step": 673
|
| 4071 |
+
},
|
| 4072 |
+
{
|
| 4073 |
+
"epoch": 0.46,
|
| 4074 |
+
"learning_rate": 0.0005320592451027233,
|
| 4075 |
+
"loss": 2.1749,
|
| 4076 |
+
"step": 674
|
| 4077 |
+
},
|
| 4078 |
+
{
|
| 4079 |
+
"epoch": 0.46,
|
| 4080 |
+
"learning_rate": 0.0005319159101767797,
|
| 4081 |
+
"loss": 2.1062,
|
| 4082 |
+
"step": 675
|
| 4083 |
+
},
|
| 4084 |
+
{
|
| 4085 |
+
"epoch": 0.46,
|
| 4086 |
+
"learning_rate": 0.000531772575250836,
|
| 4087 |
+
"loss": 2.1854,
|
| 4088 |
+
"step": 676
|
| 4089 |
+
},
|
| 4090 |
+
{
|
| 4091 |
+
"epoch": 0.46,
|
| 4092 |
+
"learning_rate": 0.0005316292403248925,
|
| 4093 |
+
"loss": 2.1493,
|
| 4094 |
+
"step": 677
|
| 4095 |
+
},
|
| 4096 |
+
{
|
| 4097 |
+
"epoch": 0.46,
|
| 4098 |
+
"learning_rate": 0.0005314859053989488,
|
| 4099 |
+
"loss": 2.2119,
|
| 4100 |
+
"step": 678
|
| 4101 |
+
},
|
| 4102 |
+
{
|
| 4103 |
+
"epoch": 0.46,
|
| 4104 |
+
"learning_rate": 0.0005313425704730053,
|
| 4105 |
+
"loss": 2.104,
|
| 4106 |
+
"step": 679
|
| 4107 |
+
},
|
| 4108 |
+
{
|
| 4109 |
+
"epoch": 0.46,
|
| 4110 |
+
"learning_rate": 0.0005311992355470616,
|
| 4111 |
+
"loss": 2.2132,
|
| 4112 |
+
"step": 680
|
| 4113 |
+
},
|
| 4114 |
+
{
|
| 4115 |
+
"epoch": 0.47,
|
| 4116 |
+
"learning_rate": 0.0005310559006211179,
|
| 4117 |
+
"loss": 2.2275,
|
| 4118 |
+
"step": 681
|
| 4119 |
+
},
|
| 4120 |
+
{
|
| 4121 |
+
"epoch": 0.47,
|
| 4122 |
+
"learning_rate": 0.0005309125656951743,
|
| 4123 |
+
"loss": 2.146,
|
| 4124 |
+
"step": 682
|
| 4125 |
+
},
|
| 4126 |
+
{
|
| 4127 |
+
"epoch": 0.47,
|
| 4128 |
+
"learning_rate": 0.0005307692307692307,
|
| 4129 |
+
"loss": 2.2401,
|
| 4130 |
+
"step": 683
|
| 4131 |
+
},
|
| 4132 |
+
{
|
| 4133 |
+
"epoch": 0.47,
|
| 4134 |
+
"learning_rate": 0.0005306258958432871,
|
| 4135 |
+
"loss": 2.1453,
|
| 4136 |
+
"step": 684
|
| 4137 |
+
},
|
| 4138 |
+
{
|
| 4139 |
+
"epoch": 0.47,
|
| 4140 |
+
"learning_rate": 0.0005304825609173435,
|
| 4141 |
+
"loss": 2.1154,
|
| 4142 |
+
"step": 685
|
| 4143 |
+
},
|
| 4144 |
+
{
|
| 4145 |
+
"epoch": 0.47,
|
| 4146 |
+
"learning_rate": 0.0005303392259913999,
|
| 4147 |
+
"loss": 2.2073,
|
| 4148 |
+
"step": 686
|
| 4149 |
+
},
|
| 4150 |
+
{
|
| 4151 |
+
"epoch": 0.47,
|
| 4152 |
+
"learning_rate": 0.0005301958910654562,
|
| 4153 |
+
"loss": 2.3166,
|
| 4154 |
+
"step": 687
|
| 4155 |
+
},
|
| 4156 |
+
{
|
| 4157 |
+
"epoch": 0.47,
|
| 4158 |
+
"learning_rate": 0.0005300525561395126,
|
| 4159 |
+
"loss": 2.1223,
|
| 4160 |
+
"step": 688
|
| 4161 |
+
},
|
| 4162 |
+
{
|
| 4163 |
+
"epoch": 0.47,
|
| 4164 |
+
"learning_rate": 0.000529909221213569,
|
| 4165 |
+
"loss": 2.1857,
|
| 4166 |
+
"step": 689
|
| 4167 |
+
},
|
| 4168 |
+
{
|
| 4169 |
+
"epoch": 0.47,
|
| 4170 |
+
"learning_rate": 0.0005297658862876254,
|
| 4171 |
+
"loss": 2.1932,
|
| 4172 |
+
"step": 690
|
| 4173 |
+
},
|
| 4174 |
+
{
|
| 4175 |
+
"epoch": 0.47,
|
| 4176 |
+
"learning_rate": 0.0005296225513616818,
|
| 4177 |
+
"loss": 2.1361,
|
| 4178 |
+
"step": 691
|
| 4179 |
+
},
|
| 4180 |
+
{
|
| 4181 |
+
"epoch": 0.47,
|
| 4182 |
+
"learning_rate": 0.0005294792164357382,
|
| 4183 |
+
"loss": 2.1133,
|
| 4184 |
+
"step": 692
|
| 4185 |
+
},
|
| 4186 |
+
{
|
| 4187 |
+
"epoch": 0.47,
|
| 4188 |
+
"learning_rate": 0.0005293358815097945,
|
| 4189 |
+
"loss": 2.137,
|
| 4190 |
+
"step": 693
|
| 4191 |
+
},
|
| 4192 |
+
{
|
| 4193 |
+
"epoch": 0.47,
|
| 4194 |
+
"learning_rate": 0.0005291925465838509,
|
| 4195 |
+
"loss": 2.1077,
|
| 4196 |
+
"step": 694
|
| 4197 |
+
},
|
| 4198 |
+
{
|
| 4199 |
+
"epoch": 0.48,
|
| 4200 |
+
"learning_rate": 0.0005290492116579072,
|
| 4201 |
+
"loss": 2.2026,
|
| 4202 |
+
"step": 695
|
| 4203 |
+
},
|
| 4204 |
+
{
|
| 4205 |
+
"epoch": 0.48,
|
| 4206 |
+
"learning_rate": 0.0005289058767319637,
|
| 4207 |
+
"loss": 2.1443,
|
| 4208 |
+
"step": 696
|
| 4209 |
+
},
|
| 4210 |
+
{
|
| 4211 |
+
"epoch": 0.48,
|
| 4212 |
+
"learning_rate": 0.00052876254180602,
|
| 4213 |
+
"loss": 2.0933,
|
| 4214 |
+
"step": 697
|
| 4215 |
+
},
|
| 4216 |
+
{
|
| 4217 |
+
"epoch": 0.48,
|
| 4218 |
+
"learning_rate": 0.0005286192068800765,
|
| 4219 |
+
"loss": 2.1728,
|
| 4220 |
+
"step": 698
|
| 4221 |
+
},
|
| 4222 |
+
{
|
| 4223 |
+
"epoch": 0.48,
|
| 4224 |
+
"learning_rate": 0.0005284758719541327,
|
| 4225 |
+
"loss": 2.1026,
|
| 4226 |
+
"step": 699
|
| 4227 |
+
},
|
| 4228 |
+
{
|
| 4229 |
+
"epoch": 0.48,
|
| 4230 |
+
"learning_rate": 0.0005283325370281892,
|
| 4231 |
+
"loss": 2.1373,
|
| 4232 |
+
"step": 700
|
| 4233 |
+
},
|
| 4234 |
+
{
|
| 4235 |
+
"epoch": 0.48,
|
| 4236 |
+
"learning_rate": 0.0005281892021022455,
|
| 4237 |
+
"loss": 2.0439,
|
| 4238 |
+
"step": 701
|
| 4239 |
+
},
|
| 4240 |
+
{
|
| 4241 |
+
"epoch": 0.48,
|
| 4242 |
+
"learning_rate": 0.0005280458671763019,
|
| 4243 |
+
"loss": 2.1893,
|
| 4244 |
+
"step": 702
|
| 4245 |
+
},
|
| 4246 |
+
{
|
| 4247 |
+
"epoch": 0.48,
|
| 4248 |
+
"learning_rate": 0.0005279025322503583,
|
| 4249 |
+
"loss": 2.2672,
|
| 4250 |
+
"step": 703
|
| 4251 |
+
},
|
| 4252 |
+
{
|
| 4253 |
+
"epoch": 0.48,
|
| 4254 |
+
"learning_rate": 0.0005277591973244147,
|
| 4255 |
+
"loss": 2.176,
|
| 4256 |
+
"step": 704
|
| 4257 |
+
},
|
| 4258 |
+
{
|
| 4259 |
+
"epoch": 0.48,
|
| 4260 |
+
"learning_rate": 0.000527615862398471,
|
| 4261 |
+
"loss": 2.1351,
|
| 4262 |
+
"step": 705
|
| 4263 |
+
},
|
| 4264 |
+
{
|
| 4265 |
+
"epoch": 0.48,
|
| 4266 |
+
"learning_rate": 0.0005274725274725274,
|
| 4267 |
+
"loss": 2.0995,
|
| 4268 |
+
"step": 706
|
| 4269 |
+
},
|
| 4270 |
+
{
|
| 4271 |
+
"epoch": 0.48,
|
| 4272 |
+
"learning_rate": 0.0005273291925465838,
|
| 4273 |
+
"loss": 2.1155,
|
| 4274 |
+
"step": 707
|
| 4275 |
+
},
|
| 4276 |
+
{
|
| 4277 |
+
"epoch": 0.48,
|
| 4278 |
+
"learning_rate": 0.0005271858576206402,
|
| 4279 |
+
"loss": 2.1424,
|
| 4280 |
+
"step": 708
|
| 4281 |
+
},
|
| 4282 |
+
{
|
| 4283 |
+
"epoch": 0.48,
|
| 4284 |
+
"learning_rate": 0.0005270425226946966,
|
| 4285 |
+
"loss": 2.2112,
|
| 4286 |
+
"step": 709
|
| 4287 |
+
},
|
| 4288 |
+
{
|
| 4289 |
+
"epoch": 0.49,
|
| 4290 |
+
"learning_rate": 0.000526899187768753,
|
| 4291 |
+
"loss": 2.1792,
|
| 4292 |
+
"step": 710
|
| 4293 |
+
},
|
| 4294 |
+
{
|
| 4295 |
+
"epoch": 0.49,
|
| 4296 |
+
"learning_rate": 0.0005267558528428093,
|
| 4297 |
+
"loss": 2.133,
|
| 4298 |
+
"step": 711
|
| 4299 |
+
},
|
| 4300 |
+
{
|
| 4301 |
+
"epoch": 0.49,
|
| 4302 |
+
"learning_rate": 0.0005266125179168657,
|
| 4303 |
+
"loss": 2.1542,
|
| 4304 |
+
"step": 712
|
| 4305 |
+
},
|
| 4306 |
+
{
|
| 4307 |
+
"epoch": 0.49,
|
| 4308 |
+
"learning_rate": 0.0005264691829909221,
|
| 4309 |
+
"loss": 2.1184,
|
| 4310 |
+
"step": 713
|
| 4311 |
+
},
|
| 4312 |
+
{
|
| 4313 |
+
"epoch": 0.49,
|
| 4314 |
+
"learning_rate": 0.0005263258480649785,
|
| 4315 |
+
"loss": 2.0157,
|
| 4316 |
+
"step": 714
|
| 4317 |
+
},
|
| 4318 |
+
{
|
| 4319 |
+
"epoch": 0.49,
|
| 4320 |
+
"learning_rate": 0.0005261825131390349,
|
| 4321 |
+
"loss": 2.1661,
|
| 4322 |
+
"step": 715
|
| 4323 |
+
},
|
| 4324 |
+
{
|
| 4325 |
+
"epoch": 0.49,
|
| 4326 |
+
"learning_rate": 0.0005260391782130912,
|
| 4327 |
+
"loss": 2.2518,
|
| 4328 |
+
"step": 716
|
| 4329 |
+
},
|
| 4330 |
+
{
|
| 4331 |
+
"epoch": 0.49,
|
| 4332 |
+
"learning_rate": 0.0005258958432871477,
|
| 4333 |
+
"loss": 2.0832,
|
| 4334 |
+
"step": 717
|
| 4335 |
+
},
|
| 4336 |
+
{
|
| 4337 |
+
"epoch": 0.49,
|
| 4338 |
+
"learning_rate": 0.0005257525083612039,
|
| 4339 |
+
"loss": 2.2078,
|
| 4340 |
+
"step": 718
|
| 4341 |
+
},
|
| 4342 |
+
{
|
| 4343 |
+
"epoch": 0.49,
|
| 4344 |
+
"learning_rate": 0.0005256091734352604,
|
| 4345 |
+
"loss": 2.085,
|
| 4346 |
+
"step": 719
|
| 4347 |
+
},
|
| 4348 |
+
{
|
| 4349 |
+
"epoch": 0.49,
|
| 4350 |
+
"learning_rate": 0.0005254658385093167,
|
| 4351 |
+
"loss": 2.1,
|
| 4352 |
+
"step": 720
|
| 4353 |
+
},
|
| 4354 |
+
{
|
| 4355 |
+
"epoch": 0.49,
|
| 4356 |
+
"learning_rate": 0.0005253225035833732,
|
| 4357 |
+
"loss": 2.0984,
|
| 4358 |
+
"step": 721
|
| 4359 |
+
},
|
| 4360 |
+
{
|
| 4361 |
+
"epoch": 0.49,
|
| 4362 |
+
"learning_rate": 0.0005251791686574295,
|
| 4363 |
+
"loss": 2.1313,
|
| 4364 |
+
"step": 722
|
| 4365 |
+
},
|
| 4366 |
+
{
|
| 4367 |
+
"epoch": 0.49,
|
| 4368 |
+
"learning_rate": 0.0005250358337314858,
|
| 4369 |
+
"loss": 2.0344,
|
| 4370 |
+
"step": 723
|
| 4371 |
+
},
|
| 4372 |
+
{
|
| 4373 |
+
"epoch": 0.49,
|
| 4374 |
+
"learning_rate": 0.0005248924988055422,
|
| 4375 |
+
"loss": 2.2206,
|
| 4376 |
+
"step": 724
|
| 4377 |
+
},
|
| 4378 |
+
{
|
| 4379 |
+
"epoch": 0.5,
|
| 4380 |
+
"learning_rate": 0.0005247491638795986,
|
| 4381 |
+
"loss": 2.2121,
|
| 4382 |
+
"step": 725
|
| 4383 |
+
},
|
| 4384 |
+
{
|
| 4385 |
+
"epoch": 0.5,
|
| 4386 |
+
"learning_rate": 0.000524605828953655,
|
| 4387 |
+
"loss": 2.1267,
|
| 4388 |
+
"step": 726
|
| 4389 |
+
},
|
| 4390 |
+
{
|
| 4391 |
+
"epoch": 0.5,
|
| 4392 |
+
"learning_rate": 0.0005244624940277114,
|
| 4393 |
+
"loss": 2.0358,
|
| 4394 |
+
"step": 727
|
| 4395 |
+
},
|
| 4396 |
+
{
|
| 4397 |
+
"epoch": 0.5,
|
| 4398 |
+
"learning_rate": 0.0005243191591017678,
|
| 4399 |
+
"loss": 2.1301,
|
| 4400 |
+
"step": 728
|
| 4401 |
+
},
|
| 4402 |
+
{
|
| 4403 |
+
"epoch": 0.5,
|
| 4404 |
+
"learning_rate": 0.0005241758241758241,
|
| 4405 |
+
"loss": 2.1678,
|
| 4406 |
+
"step": 729
|
| 4407 |
+
},
|
| 4408 |
+
{
|
| 4409 |
+
"epoch": 0.5,
|
| 4410 |
+
"learning_rate": 0.0005240324892498805,
|
| 4411 |
+
"loss": 2.0717,
|
| 4412 |
+
"step": 730
|
| 4413 |
+
},
|
| 4414 |
+
{
|
| 4415 |
+
"epoch": 0.5,
|
| 4416 |
+
"learning_rate": 0.0005238891543239369,
|
| 4417 |
+
"loss": 2.1041,
|
| 4418 |
+
"step": 731
|
| 4419 |
+
},
|
| 4420 |
+
{
|
| 4421 |
+
"epoch": 0.5,
|
| 4422 |
+
"learning_rate": 0.0005237458193979933,
|
| 4423 |
+
"loss": 2.1661,
|
| 4424 |
+
"step": 732
|
| 4425 |
+
},
|
| 4426 |
+
{
|
| 4427 |
+
"epoch": 0.5,
|
| 4428 |
+
"learning_rate": 0.0005236024844720497,
|
| 4429 |
+
"loss": 2.2038,
|
| 4430 |
+
"step": 733
|
| 4431 |
+
},
|
| 4432 |
+
{
|
| 4433 |
+
"epoch": 0.5,
|
| 4434 |
+
"learning_rate": 0.0005234591495461061,
|
| 4435 |
+
"loss": 2.0797,
|
| 4436 |
+
"step": 734
|
| 4437 |
+
},
|
| 4438 |
+
{
|
| 4439 |
+
"epoch": 0.5,
|
| 4440 |
+
"learning_rate": 0.0005233158146201624,
|
| 4441 |
+
"loss": 1.9513,
|
| 4442 |
+
"step": 735
|
| 4443 |
+
},
|
| 4444 |
+
{
|
| 4445 |
+
"epoch": 0.5,
|
| 4446 |
+
"learning_rate": 0.0005231724796942188,
|
| 4447 |
+
"loss": 2.0952,
|
| 4448 |
+
"step": 736
|
| 4449 |
+
},
|
| 4450 |
+
{
|
| 4451 |
+
"epoch": 0.5,
|
| 4452 |
+
"learning_rate": 0.0005230291447682751,
|
| 4453 |
+
"loss": 2.0112,
|
| 4454 |
+
"step": 737
|
| 4455 |
+
},
|
| 4456 |
+
{
|
| 4457 |
+
"epoch": 0.5,
|
| 4458 |
+
"learning_rate": 0.0005228858098423316,
|
| 4459 |
+
"loss": 2.1552,
|
| 4460 |
+
"step": 738
|
| 4461 |
+
},
|
| 4462 |
+
{
|
| 4463 |
+
"epoch": 0.51,
|
| 4464 |
+
"learning_rate": 0.0005227424749163879,
|
| 4465 |
+
"loss": 2.1163,
|
| 4466 |
+
"step": 739
|
| 4467 |
+
},
|
| 4468 |
+
{
|
| 4469 |
+
"epoch": 0.51,
|
| 4470 |
+
"learning_rate": 0.0005225991399904444,
|
| 4471 |
+
"loss": 2.1879,
|
| 4472 |
+
"step": 740
|
| 4473 |
+
},
|
| 4474 |
+
{
|
| 4475 |
+
"epoch": 0.51,
|
| 4476 |
+
"learning_rate": 0.0005224558050645006,
|
| 4477 |
+
"loss": 2.1494,
|
| 4478 |
+
"step": 741
|
| 4479 |
+
},
|
| 4480 |
+
{
|
| 4481 |
+
"epoch": 0.51,
|
| 4482 |
+
"learning_rate": 0.000522312470138557,
|
| 4483 |
+
"loss": 2.1554,
|
| 4484 |
+
"step": 742
|
| 4485 |
+
},
|
| 4486 |
+
{
|
| 4487 |
+
"epoch": 0.51,
|
| 4488 |
+
"learning_rate": 0.0005221691352126134,
|
| 4489 |
+
"loss": 2.1672,
|
| 4490 |
+
"step": 743
|
| 4491 |
+
},
|
| 4492 |
+
{
|
| 4493 |
+
"epoch": 0.51,
|
| 4494 |
+
"learning_rate": 0.0005220258002866698,
|
| 4495 |
+
"loss": 2.1297,
|
| 4496 |
+
"step": 744
|
| 4497 |
+
},
|
| 4498 |
+
{
|
| 4499 |
+
"epoch": 0.51,
|
| 4500 |
+
"learning_rate": 0.0005218824653607262,
|
| 4501 |
+
"loss": 2.1553,
|
| 4502 |
+
"step": 745
|
| 4503 |
+
},
|
| 4504 |
+
{
|
| 4505 |
+
"epoch": 0.51,
|
| 4506 |
+
"learning_rate": 0.0005217391304347826,
|
| 4507 |
+
"loss": 2.1184,
|
| 4508 |
+
"step": 746
|
| 4509 |
+
},
|
| 4510 |
+
{
|
| 4511 |
+
"epoch": 0.51,
|
| 4512 |
+
"learning_rate": 0.000521595795508839,
|
| 4513 |
+
"loss": 2.1945,
|
| 4514 |
+
"step": 747
|
| 4515 |
+
},
|
| 4516 |
+
{
|
| 4517 |
+
"epoch": 0.51,
|
| 4518 |
+
"learning_rate": 0.0005214524605828953,
|
| 4519 |
+
"loss": 2.1973,
|
| 4520 |
+
"step": 748
|
| 4521 |
+
},
|
| 4522 |
+
{
|
| 4523 |
+
"epoch": 0.51,
|
| 4524 |
+
"learning_rate": 0.0005213091256569517,
|
| 4525 |
+
"loss": 2.076,
|
| 4526 |
+
"step": 749
|
| 4527 |
+
},
|
| 4528 |
+
{
|
| 4529 |
+
"epoch": 0.51,
|
| 4530 |
+
"learning_rate": 0.0005211657907310081,
|
| 4531 |
+
"loss": 2.1019,
|
| 4532 |
+
"step": 750
|
| 4533 |
+
},
|
| 4534 |
+
{
|
| 4535 |
+
"epoch": 0.51,
|
| 4536 |
+
"learning_rate": 0.0005210224558050645,
|
| 4537 |
+
"loss": 2.1234,
|
| 4538 |
+
"step": 751
|
| 4539 |
+
},
|
| 4540 |
+
{
|
| 4541 |
+
"epoch": 0.51,
|
| 4542 |
+
"learning_rate": 0.0005208791208791209,
|
| 4543 |
+
"loss": 2.221,
|
| 4544 |
+
"step": 752
|
| 4545 |
+
},
|
| 4546 |
+
{
|
| 4547 |
+
"epoch": 0.51,
|
| 4548 |
+
"learning_rate": 0.0005207357859531772,
|
| 4549 |
+
"loss": 2.0994,
|
| 4550 |
+
"step": 753
|
| 4551 |
+
},
|
| 4552 |
+
{
|
| 4553 |
+
"epoch": 0.52,
|
| 4554 |
+
"learning_rate": 0.0005205924510272336,
|
| 4555 |
+
"loss": 2.0699,
|
| 4556 |
+
"step": 754
|
| 4557 |
+
},
|
| 4558 |
+
{
|
| 4559 |
+
"epoch": 0.52,
|
| 4560 |
+
"learning_rate": 0.00052044911610129,
|
| 4561 |
+
"loss": 2.1554,
|
| 4562 |
+
"step": 755
|
| 4563 |
+
},
|
| 4564 |
+
{
|
| 4565 |
+
"epoch": 0.52,
|
| 4566 |
+
"learning_rate": 0.0005203057811753463,
|
| 4567 |
+
"loss": 2.1254,
|
| 4568 |
+
"step": 756
|
| 4569 |
+
},
|
| 4570 |
+
{
|
| 4571 |
+
"epoch": 0.52,
|
| 4572 |
+
"learning_rate": 0.0005201624462494028,
|
| 4573 |
+
"loss": 2.2119,
|
| 4574 |
+
"step": 757
|
| 4575 |
+
},
|
| 4576 |
+
{
|
| 4577 |
+
"epoch": 0.52,
|
| 4578 |
+
"learning_rate": 0.0005200191113234591,
|
| 4579 |
+
"loss": 2.0509,
|
| 4580 |
+
"step": 758
|
| 4581 |
+
},
|
| 4582 |
+
{
|
| 4583 |
+
"epoch": 0.52,
|
| 4584 |
+
"learning_rate": 0.0005198757763975154,
|
| 4585 |
+
"loss": 2.2616,
|
| 4586 |
+
"step": 759
|
| 4587 |
+
},
|
| 4588 |
+
{
|
| 4589 |
+
"epoch": 0.52,
|
| 4590 |
+
"learning_rate": 0.0005197324414715718,
|
| 4591 |
+
"loss": 2.0417,
|
| 4592 |
+
"step": 760
|
| 4593 |
+
},
|
| 4594 |
+
{
|
| 4595 |
+
"epoch": 0.52,
|
| 4596 |
+
"learning_rate": 0.0005195891065456282,
|
| 4597 |
+
"loss": 2.1356,
|
| 4598 |
+
"step": 761
|
| 4599 |
+
},
|
| 4600 |
+
{
|
| 4601 |
+
"epoch": 0.52,
|
| 4602 |
+
"learning_rate": 0.0005194457716196846,
|
| 4603 |
+
"loss": 2.0701,
|
| 4604 |
+
"step": 762
|
| 4605 |
+
},
|
| 4606 |
+
{
|
| 4607 |
+
"epoch": 0.52,
|
| 4608 |
+
"learning_rate": 0.000519302436693741,
|
| 4609 |
+
"loss": 2.1677,
|
| 4610 |
+
"step": 763
|
| 4611 |
+
},
|
| 4612 |
+
{
|
| 4613 |
+
"epoch": 0.52,
|
| 4614 |
+
"learning_rate": 0.0005191591017677974,
|
| 4615 |
+
"loss": 2.0766,
|
| 4616 |
+
"step": 764
|
| 4617 |
+
},
|
| 4618 |
+
{
|
| 4619 |
+
"epoch": 0.52,
|
| 4620 |
+
"learning_rate": 0.0005190157668418537,
|
| 4621 |
+
"loss": 2.1733,
|
| 4622 |
+
"step": 765
|
| 4623 |
+
},
|
| 4624 |
+
{
|
| 4625 |
+
"epoch": 0.52,
|
| 4626 |
+
"learning_rate": 0.0005188724319159101,
|
| 4627 |
+
"loss": 2.1307,
|
| 4628 |
+
"step": 766
|
| 4629 |
+
},
|
| 4630 |
+
{
|
| 4631 |
+
"epoch": 0.52,
|
| 4632 |
+
"learning_rate": 0.0005187290969899665,
|
| 4633 |
+
"loss": 2.0885,
|
| 4634 |
+
"step": 767
|
| 4635 |
+
},
|
| 4636 |
+
{
|
| 4637 |
+
"epoch": 0.53,
|
| 4638 |
+
"learning_rate": 0.0005185857620640229,
|
| 4639 |
+
"loss": 2.1836,
|
| 4640 |
+
"step": 768
|
| 4641 |
+
},
|
| 4642 |
+
{
|
| 4643 |
+
"epoch": 0.53,
|
| 4644 |
+
"learning_rate": 0.0005184424271380793,
|
| 4645 |
+
"loss": 2.1477,
|
| 4646 |
+
"step": 769
|
| 4647 |
+
},
|
| 4648 |
+
{
|
| 4649 |
+
"epoch": 0.53,
|
| 4650 |
+
"learning_rate": 0.0005182990922121357,
|
| 4651 |
+
"loss": 2.1539,
|
| 4652 |
+
"step": 770
|
| 4653 |
+
},
|
| 4654 |
+
{
|
| 4655 |
+
"epoch": 0.53,
|
| 4656 |
+
"learning_rate": 0.000518155757286192,
|
| 4657 |
+
"loss": 2.1302,
|
| 4658 |
+
"step": 771
|
| 4659 |
+
},
|
| 4660 |
+
{
|
| 4661 |
+
"epoch": 0.53,
|
| 4662 |
+
"learning_rate": 0.0005180124223602484,
|
| 4663 |
+
"loss": 2.1686,
|
| 4664 |
+
"step": 772
|
| 4665 |
+
},
|
| 4666 |
+
{
|
| 4667 |
+
"epoch": 0.53,
|
| 4668 |
+
"learning_rate": 0.0005178690874343048,
|
| 4669 |
+
"loss": 2.1822,
|
| 4670 |
+
"step": 773
|
| 4671 |
+
},
|
| 4672 |
+
{
|
| 4673 |
+
"epoch": 0.53,
|
| 4674 |
+
"learning_rate": 0.0005177257525083612,
|
| 4675 |
+
"loss": 2.1349,
|
| 4676 |
+
"step": 774
|
| 4677 |
+
},
|
| 4678 |
+
{
|
| 4679 |
+
"epoch": 0.53,
|
| 4680 |
+
"learning_rate": 0.0005175824175824176,
|
| 4681 |
+
"loss": 2.1254,
|
| 4682 |
+
"step": 775
|
| 4683 |
+
},
|
| 4684 |
+
{
|
| 4685 |
+
"epoch": 0.53,
|
| 4686 |
+
"learning_rate": 0.0005174390826564739,
|
| 4687 |
+
"loss": 2.1617,
|
| 4688 |
+
"step": 776
|
| 4689 |
+
},
|
| 4690 |
+
{
|
| 4691 |
+
"epoch": 0.53,
|
| 4692 |
+
"learning_rate": 0.0005172957477305302,
|
| 4693 |
+
"loss": 2.0366,
|
| 4694 |
+
"step": 777
|
| 4695 |
+
},
|
| 4696 |
+
{
|
| 4697 |
+
"epoch": 0.53,
|
| 4698 |
+
"learning_rate": 0.0005171524128045866,
|
| 4699 |
+
"loss": 2.1776,
|
| 4700 |
+
"step": 778
|
| 4701 |
+
},
|
| 4702 |
+
{
|
| 4703 |
+
"epoch": 0.53,
|
| 4704 |
+
"learning_rate": 0.000517009077878643,
|
| 4705 |
+
"loss": 2.0636,
|
| 4706 |
+
"step": 779
|
| 4707 |
+
},
|
| 4708 |
+
{
|
| 4709 |
+
"epoch": 0.53,
|
| 4710 |
+
"learning_rate": 0.0005168657429526994,
|
| 4711 |
+
"loss": 2.0966,
|
| 4712 |
+
"step": 780
|
| 4713 |
+
},
|
| 4714 |
+
{
|
| 4715 |
+
"epoch": 0.53,
|
| 4716 |
+
"learning_rate": 0.0005167224080267558,
|
| 4717 |
+
"loss": 2.0935,
|
| 4718 |
+
"step": 781
|
| 4719 |
+
},
|
| 4720 |
+
{
|
| 4721 |
+
"epoch": 0.53,
|
| 4722 |
+
"learning_rate": 0.0005165790731008122,
|
| 4723 |
+
"loss": 2.0194,
|
| 4724 |
+
"step": 782
|
| 4725 |
+
},
|
| 4726 |
+
{
|
| 4727 |
+
"epoch": 0.54,
|
| 4728 |
+
"learning_rate": 0.0005164357381748685,
|
| 4729 |
+
"loss": 2.1995,
|
| 4730 |
+
"step": 783
|
| 4731 |
+
},
|
| 4732 |
+
{
|
| 4733 |
+
"epoch": 0.54,
|
| 4734 |
+
"learning_rate": 0.0005162924032489249,
|
| 4735 |
+
"loss": 2.0758,
|
| 4736 |
+
"step": 784
|
| 4737 |
+
},
|
| 4738 |
+
{
|
| 4739 |
+
"epoch": 0.54,
|
| 4740 |
+
"learning_rate": 0.0005161490683229813,
|
| 4741 |
+
"loss": 2.134,
|
| 4742 |
+
"step": 785
|
| 4743 |
+
},
|
| 4744 |
+
{
|
| 4745 |
+
"epoch": 0.54,
|
| 4746 |
+
"learning_rate": 0.0005160057333970377,
|
| 4747 |
+
"loss": 2.1708,
|
| 4748 |
+
"step": 786
|
| 4749 |
+
},
|
| 4750 |
+
{
|
| 4751 |
+
"epoch": 0.54,
|
| 4752 |
+
"learning_rate": 0.0005158623984710941,
|
| 4753 |
+
"loss": 2.155,
|
| 4754 |
+
"step": 787
|
| 4755 |
+
},
|
| 4756 |
+
{
|
| 4757 |
+
"epoch": 0.54,
|
| 4758 |
+
"learning_rate": 0.0005157190635451505,
|
| 4759 |
+
"loss": 2.2275,
|
| 4760 |
+
"step": 788
|
| 4761 |
+
},
|
| 4762 |
+
{
|
| 4763 |
+
"epoch": 0.54,
|
| 4764 |
+
"learning_rate": 0.0005155757286192068,
|
| 4765 |
+
"loss": 2.151,
|
| 4766 |
+
"step": 789
|
| 4767 |
+
},
|
| 4768 |
+
{
|
| 4769 |
+
"epoch": 0.54,
|
| 4770 |
+
"learning_rate": 0.0005154323936932632,
|
| 4771 |
+
"loss": 2.1853,
|
| 4772 |
+
"step": 790
|
| 4773 |
+
},
|
| 4774 |
+
{
|
| 4775 |
+
"epoch": 0.54,
|
| 4776 |
+
"learning_rate": 0.0005152890587673196,
|
| 4777 |
+
"loss": 2.1455,
|
| 4778 |
+
"step": 791
|
| 4779 |
+
},
|
| 4780 |
+
{
|
| 4781 |
+
"epoch": 0.54,
|
| 4782 |
+
"learning_rate": 0.000515145723841376,
|
| 4783 |
+
"loss": 2.1934,
|
| 4784 |
+
"step": 792
|
| 4785 |
+
},
|
| 4786 |
+
{
|
| 4787 |
+
"epoch": 0.54,
|
| 4788 |
+
"learning_rate": 0.0005150023889154323,
|
| 4789 |
+
"loss": 2.0501,
|
| 4790 |
+
"step": 793
|
| 4791 |
+
},
|
| 4792 |
+
{
|
| 4793 |
+
"epoch": 0.54,
|
| 4794 |
+
"learning_rate": 0.0005148590539894888,
|
| 4795 |
+
"loss": 2.197,
|
| 4796 |
+
"step": 794
|
| 4797 |
+
},
|
| 4798 |
+
{
|
| 4799 |
+
"epoch": 0.54,
|
| 4800 |
+
"learning_rate": 0.000514715719063545,
|
| 4801 |
+
"loss": 2.0713,
|
| 4802 |
+
"step": 795
|
| 4803 |
+
},
|
| 4804 |
+
{
|
| 4805 |
+
"epoch": 0.54,
|
| 4806 |
+
"learning_rate": 0.0005145723841376015,
|
| 4807 |
+
"loss": 2.1562,
|
| 4808 |
+
"step": 796
|
| 4809 |
+
},
|
| 4810 |
+
{
|
| 4811 |
+
"epoch": 0.54,
|
| 4812 |
+
"learning_rate": 0.0005144290492116578,
|
| 4813 |
+
"loss": 2.158,
|
| 4814 |
+
"step": 797
|
| 4815 |
+
},
|
| 4816 |
+
{
|
| 4817 |
+
"epoch": 0.55,
|
| 4818 |
+
"learning_rate": 0.0005142857142857142,
|
| 4819 |
+
"loss": 2.0456,
|
| 4820 |
+
"step": 798
|
| 4821 |
+
},
|
| 4822 |
+
{
|
| 4823 |
+
"epoch": 0.55,
|
| 4824 |
+
"learning_rate": 0.0005141423793597706,
|
| 4825 |
+
"loss": 2.157,
|
| 4826 |
+
"step": 799
|
| 4827 |
+
},
|
| 4828 |
+
{
|
| 4829 |
+
"epoch": 0.55,
|
| 4830 |
+
"learning_rate": 0.000513999044433827,
|
| 4831 |
+
"loss": 2.0934,
|
| 4832 |
+
"step": 800
|
| 4833 |
+
},
|
| 4834 |
+
{
|
| 4835 |
+
"epoch": 0.55,
|
| 4836 |
+
"eval_loss": 2.0316452980041504,
|
| 4837 |
+
"eval_runtime": 1689.8048,
|
| 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": 9.813169375148851e+17,
|
| 4846 |
"trial_name": null,
|
| 4847 |
"trial_params": null
|
| 4848 |
}
|
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 sha256:6b06061cdd59a61c03b74896e78f938e36d6d587093dddf2c3beb4c518798564
|
| 3 |
size 2368281769
|