Upload 7 files
Browse files- optimizer.pt +3 -0
- pytorch_model.bin +3 -0
- rng_state.pth +3 -0
- scaler.pt +3 -0
- scheduler.pt +3 -0
- trainer_state.json +2566 -0
- training_args.bin +3 -0
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f1978f090320ee07a19ab5f8bd2ee1c6ea4121b76b34f893dac24eb0023b8ab8
|
| 3 |
+
size 160172869
|
pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cdbd7cbf8d2f9d819ce36bd3113f93c816fa31aa2538972830e00e8c4bd2a98c
|
| 3 |
+
size 80114765
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4595c03e72615b44d0aaec45543408898507c572efd56cf68af104ba53030e20
|
| 3 |
+
size 14575
|
scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ba95f40c8c77012eb193e29ef6a8f9174a2728b8ba4a3a8d413144e02f394116
|
| 3 |
+
size 557
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:da8239d37548de447d7244d16a4770f928ea309b9efcdb784e10b9e968a7a3d6
|
| 3 |
+
size 627
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2566 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 1.110647439956665,
|
| 3 |
+
"best_model_checkpoint": "outs/instuct_chat_50k/checkpoint-2000",
|
| 4 |
+
"epoch": 21.228571428571428,
|
| 5 |
+
"global_step": 38400,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.0,
|
| 12 |
+
"learning_rate": 1.5568240788790864e-07,
|
| 13 |
+
"loss": 1.1388,
|
| 14 |
+
"step": 1
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.01,
|
| 18 |
+
"learning_rate": 1.5412558380902958e-05,
|
| 19 |
+
"loss": 1.5551,
|
| 20 |
+
"step": 100
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.01,
|
| 24 |
+
"eval_loss": 1.2936956882476807,
|
| 25 |
+
"eval_runtime": 31.4406,
|
| 26 |
+
"eval_samples_per_second": 6.361,
|
| 27 |
+
"eval_steps_per_second": 0.795,
|
| 28 |
+
"step": 100
|
| 29 |
+
},
|
| 30 |
+
{
|
| 31 |
+
"epoch": 0.02,
|
| 32 |
+
"learning_rate": 3.098079916969382e-05,
|
| 33 |
+
"loss": 1.4423,
|
| 34 |
+
"step": 200
|
| 35 |
+
},
|
| 36 |
+
{
|
| 37 |
+
"epoch": 0.02,
|
| 38 |
+
"eval_loss": 1.227220892906189,
|
| 39 |
+
"eval_runtime": 31.4526,
|
| 40 |
+
"eval_samples_per_second": 6.359,
|
| 41 |
+
"eval_steps_per_second": 0.795,
|
| 42 |
+
"step": 200
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"epoch": 0.02,
|
| 46 |
+
"learning_rate": 4.6393357550596784e-05,
|
| 47 |
+
"loss": 1.3335,
|
| 48 |
+
"step": 300
|
| 49 |
+
},
|
| 50 |
+
{
|
| 51 |
+
"epoch": 0.02,
|
| 52 |
+
"eval_loss": 1.2049953937530518,
|
| 53 |
+
"eval_runtime": 31.9522,
|
| 54 |
+
"eval_samples_per_second": 6.259,
|
| 55 |
+
"eval_steps_per_second": 0.782,
|
| 56 |
+
"step": 300
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.03,
|
| 60 |
+
"learning_rate": 6.165023352361183e-05,
|
| 61 |
+
"loss": 1.4225,
|
| 62 |
+
"step": 400
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.03,
|
| 66 |
+
"eval_loss": 1.1863670349121094,
|
| 67 |
+
"eval_runtime": 31.5148,
|
| 68 |
+
"eval_samples_per_second": 6.346,
|
| 69 |
+
"eval_steps_per_second": 0.793,
|
| 70 |
+
"step": 400
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"epoch": 0.04,
|
| 74 |
+
"learning_rate": 7.721847431240269e-05,
|
| 75 |
+
"loss": 1.3618,
|
| 76 |
+
"step": 500
|
| 77 |
+
},
|
| 78 |
+
{
|
| 79 |
+
"epoch": 0.04,
|
| 80 |
+
"eval_loss": 1.1773875951766968,
|
| 81 |
+
"eval_runtime": 31.4521,
|
| 82 |
+
"eval_samples_per_second": 6.359,
|
| 83 |
+
"eval_steps_per_second": 0.795,
|
| 84 |
+
"step": 500
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.05,
|
| 88 |
+
"learning_rate": 9.278671510119357e-05,
|
| 89 |
+
"loss": 1.3262,
|
| 90 |
+
"step": 600
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.05,
|
| 94 |
+
"eval_loss": 1.168226718902588,
|
| 95 |
+
"eval_runtime": 31.5121,
|
| 96 |
+
"eval_samples_per_second": 6.347,
|
| 97 |
+
"eval_steps_per_second": 0.793,
|
| 98 |
+
"step": 600
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.05,
|
| 102 |
+
"learning_rate": 0.00010835495588998442,
|
| 103 |
+
"loss": 1.2133,
|
| 104 |
+
"step": 700
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.05,
|
| 108 |
+
"eval_loss": 1.1481703519821167,
|
| 109 |
+
"eval_runtime": 31.504,
|
| 110 |
+
"eval_samples_per_second": 6.348,
|
| 111 |
+
"eval_steps_per_second": 0.794,
|
| 112 |
+
"step": 700
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 0.06,
|
| 116 |
+
"learning_rate": 0.00012392319667877528,
|
| 117 |
+
"loss": 1.2034,
|
| 118 |
+
"step": 800
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 0.06,
|
| 122 |
+
"eval_loss": 1.13801908493042,
|
| 123 |
+
"eval_runtime": 31.5183,
|
| 124 |
+
"eval_samples_per_second": 6.346,
|
| 125 |
+
"eval_steps_per_second": 0.793,
|
| 126 |
+
"step": 800
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 0.07,
|
| 130 |
+
"learning_rate": 0.00013949143746756616,
|
| 131 |
+
"loss": 1.2159,
|
| 132 |
+
"step": 900
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"epoch": 0.07,
|
| 136 |
+
"eval_loss": 1.1303695440292358,
|
| 137 |
+
"eval_runtime": 31.4531,
|
| 138 |
+
"eval_samples_per_second": 6.359,
|
| 139 |
+
"eval_steps_per_second": 0.795,
|
| 140 |
+
"step": 900
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.08,
|
| 144 |
+
"learning_rate": 0.00015505967825635704,
|
| 145 |
+
"loss": 1.1612,
|
| 146 |
+
"step": 1000
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.08,
|
| 150 |
+
"eval_loss": 1.1280632019042969,
|
| 151 |
+
"eval_runtime": 31.434,
|
| 152 |
+
"eval_samples_per_second": 6.363,
|
| 153 |
+
"eval_steps_per_second": 0.795,
|
| 154 |
+
"step": 1000
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 0.09,
|
| 158 |
+
"learning_rate": 0.00017062791904514786,
|
| 159 |
+
"loss": 1.1549,
|
| 160 |
+
"step": 1100
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 0.09,
|
| 164 |
+
"eval_loss": 1.1211422681808472,
|
| 165 |
+
"eval_runtime": 31.5065,
|
| 166 |
+
"eval_samples_per_second": 6.348,
|
| 167 |
+
"eval_steps_per_second": 0.793,
|
| 168 |
+
"step": 1100
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 0.09,
|
| 172 |
+
"learning_rate": 0.00018619615983393875,
|
| 173 |
+
"loss": 1.2038,
|
| 174 |
+
"step": 1200
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.09,
|
| 178 |
+
"eval_loss": 1.126134991645813,
|
| 179 |
+
"eval_runtime": 31.4919,
|
| 180 |
+
"eval_samples_per_second": 6.351,
|
| 181 |
+
"eval_steps_per_second": 0.794,
|
| 182 |
+
"step": 1200
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.1,
|
| 186 |
+
"learning_rate": 0.00020176440062272963,
|
| 187 |
+
"loss": 1.1784,
|
| 188 |
+
"step": 1300
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.1,
|
| 192 |
+
"eval_loss": 1.1223646402359009,
|
| 193 |
+
"eval_runtime": 31.4108,
|
| 194 |
+
"eval_samples_per_second": 6.367,
|
| 195 |
+
"eval_steps_per_second": 0.796,
|
| 196 |
+
"step": 1300
|
| 197 |
+
},
|
| 198 |
+
{
|
| 199 |
+
"epoch": 0.11,
|
| 200 |
+
"learning_rate": 0.00021733264141152048,
|
| 201 |
+
"loss": 1.1834,
|
| 202 |
+
"step": 1400
|
| 203 |
+
},
|
| 204 |
+
{
|
| 205 |
+
"epoch": 0.11,
|
| 206 |
+
"eval_loss": 1.113663911819458,
|
| 207 |
+
"eval_runtime": 31.463,
|
| 208 |
+
"eval_samples_per_second": 6.357,
|
| 209 |
+
"eval_steps_per_second": 0.795,
|
| 210 |
+
"step": 1400
|
| 211 |
+
},
|
| 212 |
+
{
|
| 213 |
+
"epoch": 0.12,
|
| 214 |
+
"learning_rate": 0.00023290088220031133,
|
| 215 |
+
"loss": 1.1974,
|
| 216 |
+
"step": 1500
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 0.12,
|
| 220 |
+
"eval_loss": 1.1265443563461304,
|
| 221 |
+
"eval_runtime": 31.4893,
|
| 222 |
+
"eval_samples_per_second": 6.351,
|
| 223 |
+
"eval_steps_per_second": 0.794,
|
| 224 |
+
"step": 1500
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 0.12,
|
| 228 |
+
"learning_rate": 0.0002484691229891022,
|
| 229 |
+
"loss": 1.2174,
|
| 230 |
+
"step": 1600
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.12,
|
| 234 |
+
"eval_loss": 1.1237465143203735,
|
| 235 |
+
"eval_runtime": 31.5226,
|
| 236 |
+
"eval_samples_per_second": 6.345,
|
| 237 |
+
"eval_steps_per_second": 0.793,
|
| 238 |
+
"step": 1600
|
| 239 |
+
},
|
| 240 |
+
{
|
| 241 |
+
"epoch": 0.13,
|
| 242 |
+
"learning_rate": 0.00026403736377789307,
|
| 243 |
+
"loss": 1.2019,
|
| 244 |
+
"step": 1700
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 0.13,
|
| 248 |
+
"eval_loss": 1.1198475360870361,
|
| 249 |
+
"eval_runtime": 31.4658,
|
| 250 |
+
"eval_samples_per_second": 6.356,
|
| 251 |
+
"eval_steps_per_second": 0.795,
|
| 252 |
+
"step": 1700
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 0.14,
|
| 256 |
+
"learning_rate": 0.0002796056045666839,
|
| 257 |
+
"loss": 1.2114,
|
| 258 |
+
"step": 1800
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 0.14,
|
| 262 |
+
"eval_loss": 1.1201978921890259,
|
| 263 |
+
"eval_runtime": 31.5138,
|
| 264 |
+
"eval_samples_per_second": 6.346,
|
| 265 |
+
"eval_steps_per_second": 0.793,
|
| 266 |
+
"step": 1800
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.15,
|
| 270 |
+
"learning_rate": 0.00029517384535547483,
|
| 271 |
+
"loss": 1.1848,
|
| 272 |
+
"step": 1900
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 0.15,
|
| 276 |
+
"eval_loss": 1.1198968887329102,
|
| 277 |
+
"eval_runtime": 31.1932,
|
| 278 |
+
"eval_samples_per_second": 6.412,
|
| 279 |
+
"eval_steps_per_second": 0.801,
|
| 280 |
+
"step": 1900
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.16,
|
| 284 |
+
"learning_rate": 0.0002994345961596241,
|
| 285 |
+
"loss": 1.129,
|
| 286 |
+
"step": 2000
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 0.16,
|
| 290 |
+
"eval_loss": 1.110647439956665,
|
| 291 |
+
"eval_runtime": 31.1909,
|
| 292 |
+
"eval_samples_per_second": 6.412,
|
| 293 |
+
"eval_steps_per_second": 0.802,
|
| 294 |
+
"step": 2000
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.16,
|
| 298 |
+
"learning_rate": 0.00029861517030400697,
|
| 299 |
+
"loss": 1.1864,
|
| 300 |
+
"step": 2100
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 0.16,
|
| 304 |
+
"eval_loss": 1.1134731769561768,
|
| 305 |
+
"eval_runtime": 31.169,
|
| 306 |
+
"eval_samples_per_second": 6.417,
|
| 307 |
+
"eval_steps_per_second": 0.802,
|
| 308 |
+
"step": 2100
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 0.17,
|
| 312 |
+
"learning_rate": 0.0002977957444483898,
|
| 313 |
+
"loss": 1.1725,
|
| 314 |
+
"step": 2200
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.17,
|
| 318 |
+
"eval_loss": 1.1226890087127686,
|
| 319 |
+
"eval_runtime": 31.1975,
|
| 320 |
+
"eval_samples_per_second": 6.411,
|
| 321 |
+
"eval_steps_per_second": 0.801,
|
| 322 |
+
"step": 2200
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 0.18,
|
| 326 |
+
"learning_rate": 0.0002969763185927726,
|
| 327 |
+
"loss": 1.1363,
|
| 328 |
+
"step": 2300
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 0.18,
|
| 332 |
+
"eval_loss": 1.1255100965499878,
|
| 333 |
+
"eval_runtime": 31.237,
|
| 334 |
+
"eval_samples_per_second": 6.403,
|
| 335 |
+
"eval_steps_per_second": 0.8,
|
| 336 |
+
"step": 2300
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 0.19,
|
| 340 |
+
"learning_rate": 0.00029615689273715547,
|
| 341 |
+
"loss": 1.204,
|
| 342 |
+
"step": 2400
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 0.19,
|
| 346 |
+
"eval_loss": 1.1148998737335205,
|
| 347 |
+
"eval_runtime": 31.2369,
|
| 348 |
+
"eval_samples_per_second": 6.403,
|
| 349 |
+
"eval_steps_per_second": 0.8,
|
| 350 |
+
"step": 2400
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 0.19,
|
| 354 |
+
"learning_rate": 0.0002953374668815383,
|
| 355 |
+
"loss": 1.1465,
|
| 356 |
+
"step": 2500
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 0.19,
|
| 360 |
+
"eval_loss": 1.1103030443191528,
|
| 361 |
+
"eval_runtime": 31.2141,
|
| 362 |
+
"eval_samples_per_second": 6.407,
|
| 363 |
+
"eval_steps_per_second": 0.801,
|
| 364 |
+
"step": 2500
|
| 365 |
+
},
|
| 366 |
+
{
|
| 367 |
+
"epoch": 0.2,
|
| 368 |
+
"learning_rate": 0.0002945180410259212,
|
| 369 |
+
"loss": 1.2058,
|
| 370 |
+
"step": 2600
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.2,
|
| 374 |
+
"eval_loss": 1.1112116575241089,
|
| 375 |
+
"eval_runtime": 31.6225,
|
| 376 |
+
"eval_samples_per_second": 6.325,
|
| 377 |
+
"eval_steps_per_second": 0.791,
|
| 378 |
+
"step": 2600
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 0.21,
|
| 382 |
+
"learning_rate": 0.00029369861517030397,
|
| 383 |
+
"loss": 1.1662,
|
| 384 |
+
"step": 2700
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.21,
|
| 388 |
+
"eval_loss": 1.1122115850448608,
|
| 389 |
+
"eval_runtime": 31.1829,
|
| 390 |
+
"eval_samples_per_second": 6.414,
|
| 391 |
+
"eval_steps_per_second": 0.802,
|
| 392 |
+
"step": 2700
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 0.22,
|
| 396 |
+
"learning_rate": 0.0002928791893146868,
|
| 397 |
+
"loss": 1.1372,
|
| 398 |
+
"step": 2800
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 0.22,
|
| 402 |
+
"eval_loss": 1.1113629341125488,
|
| 403 |
+
"eval_runtime": 31.6588,
|
| 404 |
+
"eval_samples_per_second": 6.317,
|
| 405 |
+
"eval_steps_per_second": 0.79,
|
| 406 |
+
"step": 2800
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 0.23,
|
| 410 |
+
"learning_rate": 0.0002920597634590696,
|
| 411 |
+
"loss": 1.1693,
|
| 412 |
+
"step": 2900
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 0.23,
|
| 416 |
+
"eval_loss": 1.1073808670043945,
|
| 417 |
+
"eval_runtime": 31.5886,
|
| 418 |
+
"eval_samples_per_second": 6.331,
|
| 419 |
+
"eval_steps_per_second": 0.791,
|
| 420 |
+
"step": 2900
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 0.23,
|
| 424 |
+
"learning_rate": 0.0002912403376034525,
|
| 425 |
+
"loss": 1.1776,
|
| 426 |
+
"step": 3000
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 0.23,
|
| 430 |
+
"eval_loss": 1.1112266778945923,
|
| 431 |
+
"eval_runtime": 31.3957,
|
| 432 |
+
"eval_samples_per_second": 6.37,
|
| 433 |
+
"eval_steps_per_second": 0.796,
|
| 434 |
+
"step": 3000
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 1.06,
|
| 438 |
+
"learning_rate": 0.0002786905011532816,
|
| 439 |
+
"loss": 1.0338,
|
| 440 |
+
"step": 3100
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 1.11,
|
| 444 |
+
"learning_rate": 0.000277904172782554,
|
| 445 |
+
"loss": 0.9576,
|
| 446 |
+
"step": 3200
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 1.17,
|
| 450 |
+
"learning_rate": 0.00027711784441182633,
|
| 451 |
+
"loss": 0.9065,
|
| 452 |
+
"step": 3300
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 1.23,
|
| 456 |
+
"learning_rate": 0.00027633151604109874,
|
| 457 |
+
"loss": 0.8641,
|
| 458 |
+
"step": 3400
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 1.29,
|
| 462 |
+
"learning_rate": 0.00027554518767037114,
|
| 463 |
+
"loss": 0.8564,
|
| 464 |
+
"step": 3500
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 1.34,
|
| 468 |
+
"learning_rate": 0.00027475885929964354,
|
| 469 |
+
"loss": 0.8547,
|
| 470 |
+
"step": 3600
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 1.4,
|
| 474 |
+
"learning_rate": 0.0002739725309289159,
|
| 475 |
+
"loss": 0.8469,
|
| 476 |
+
"step": 3700
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 1.46,
|
| 480 |
+
"learning_rate": 0.00027318620255818824,
|
| 481 |
+
"loss": 0.8303,
|
| 482 |
+
"step": 3800
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 1.51,
|
| 486 |
+
"learning_rate": 0.00027239987418746064,
|
| 487 |
+
"loss": 0.8234,
|
| 488 |
+
"step": 3900
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 1.57,
|
| 492 |
+
"learning_rate": 0.00027161354581673304,
|
| 493 |
+
"loss": 0.7905,
|
| 494 |
+
"step": 4000
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 1.63,
|
| 498 |
+
"learning_rate": 0.00027082721744600544,
|
| 499 |
+
"loss": 0.7864,
|
| 500 |
+
"step": 4100
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 1.69,
|
| 504 |
+
"learning_rate": 0.0002700408890752778,
|
| 505 |
+
"loss": 0.7816,
|
| 506 |
+
"step": 4200
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 1.74,
|
| 510 |
+
"learning_rate": 0.0002692545607045502,
|
| 511 |
+
"loss": 0.7759,
|
| 512 |
+
"step": 4300
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 1.8,
|
| 516 |
+
"learning_rate": 0.0002684682323338226,
|
| 517 |
+
"loss": 0.7948,
|
| 518 |
+
"step": 4400
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 1.86,
|
| 522 |
+
"learning_rate": 0.00026768190396309494,
|
| 523 |
+
"loss": 0.7525,
|
| 524 |
+
"step": 4500
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 1.91,
|
| 528 |
+
"learning_rate": 0.00026689557559236735,
|
| 529 |
+
"loss": 0.7646,
|
| 530 |
+
"step": 4600
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 1.97,
|
| 534 |
+
"learning_rate": 0.00026610924722163975,
|
| 535 |
+
"loss": 0.7578,
|
| 536 |
+
"step": 4700
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 2.03,
|
| 540 |
+
"learning_rate": 0.0002653229188509121,
|
| 541 |
+
"loss": 0.6764,
|
| 542 |
+
"step": 4800
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 2.09,
|
| 546 |
+
"learning_rate": 0.0002645365904801845,
|
| 547 |
+
"loss": 0.6138,
|
| 548 |
+
"step": 4900
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 2.14,
|
| 552 |
+
"learning_rate": 0.0002637502621094569,
|
| 553 |
+
"loss": 0.6206,
|
| 554 |
+
"step": 5000
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 2.2,
|
| 558 |
+
"learning_rate": 0.0002629639337387293,
|
| 559 |
+
"loss": 0.6195,
|
| 560 |
+
"step": 5100
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 2.26,
|
| 564 |
+
"learning_rate": 0.00026217760536800165,
|
| 565 |
+
"loss": 0.6353,
|
| 566 |
+
"step": 5200
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 2.31,
|
| 570 |
+
"learning_rate": 0.000261391276997274,
|
| 571 |
+
"loss": 0.6084,
|
| 572 |
+
"step": 5300
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 2.37,
|
| 576 |
+
"learning_rate": 0.0002606049486265464,
|
| 577 |
+
"loss": 0.625,
|
| 578 |
+
"step": 5400
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 2.43,
|
| 582 |
+
"learning_rate": 0.0002598186202558188,
|
| 583 |
+
"loss": 0.6085,
|
| 584 |
+
"step": 5500
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 2.49,
|
| 588 |
+
"learning_rate": 0.0002590322918850912,
|
| 589 |
+
"loss": 0.6372,
|
| 590 |
+
"step": 5600
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 2.54,
|
| 594 |
+
"learning_rate": 0.00025824596351436356,
|
| 595 |
+
"loss": 0.6267,
|
| 596 |
+
"step": 5700
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 2.6,
|
| 600 |
+
"learning_rate": 0.00025745963514363596,
|
| 601 |
+
"loss": 0.6322,
|
| 602 |
+
"step": 5800
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 2.66,
|
| 606 |
+
"learning_rate": 0.00025667330677290836,
|
| 607 |
+
"loss": 0.6142,
|
| 608 |
+
"step": 5900
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 2.71,
|
| 612 |
+
"learning_rate": 0.0002558869784021807,
|
| 613 |
+
"loss": 0.6346,
|
| 614 |
+
"step": 6000
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 2.77,
|
| 618 |
+
"learning_rate": 0.0002551006500314531,
|
| 619 |
+
"loss": 0.6368,
|
| 620 |
+
"step": 6100
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 2.83,
|
| 624 |
+
"learning_rate": 0.0002543143216607255,
|
| 625 |
+
"loss": 0.638,
|
| 626 |
+
"step": 6200
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 2.89,
|
| 630 |
+
"learning_rate": 0.00025352799328999786,
|
| 631 |
+
"loss": 0.6373,
|
| 632 |
+
"step": 6300
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 2.94,
|
| 636 |
+
"learning_rate": 0.00025274166491927026,
|
| 637 |
+
"loss": 0.6196,
|
| 638 |
+
"step": 6400
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 3.0,
|
| 642 |
+
"learning_rate": 0.00025195533654854267,
|
| 643 |
+
"loss": 0.6328,
|
| 644 |
+
"step": 6500
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 3.06,
|
| 648 |
+
"learning_rate": 0.00025116900817781507,
|
| 649 |
+
"loss": 0.4419,
|
| 650 |
+
"step": 6600
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 3.11,
|
| 654 |
+
"learning_rate": 0.0002503826798070874,
|
| 655 |
+
"loss": 0.4553,
|
| 656 |
+
"step": 6700
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 3.17,
|
| 660 |
+
"learning_rate": 0.00024959635143635976,
|
| 661 |
+
"loss": 0.47,
|
| 662 |
+
"step": 6800
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 3.23,
|
| 666 |
+
"learning_rate": 0.00024881002306563217,
|
| 667 |
+
"loss": 0.4713,
|
| 668 |
+
"step": 6900
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 3.29,
|
| 672 |
+
"learning_rate": 0.00024802369469490457,
|
| 673 |
+
"loss": 0.4837,
|
| 674 |
+
"step": 7000
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 3.34,
|
| 678 |
+
"learning_rate": 0.00024723736632417697,
|
| 679 |
+
"loss": 0.4761,
|
| 680 |
+
"step": 7100
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 3.4,
|
| 684 |
+
"learning_rate": 0.0002464510379534493,
|
| 685 |
+
"loss": 0.4743,
|
| 686 |
+
"step": 7200
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 3.46,
|
| 690 |
+
"learning_rate": 0.0002456647095827217,
|
| 691 |
+
"loss": 0.4721,
|
| 692 |
+
"step": 7300
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 3.51,
|
| 696 |
+
"learning_rate": 0.0002448783812119941,
|
| 697 |
+
"loss": 0.4761,
|
| 698 |
+
"step": 7400
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 3.57,
|
| 702 |
+
"learning_rate": 0.00024409205284126647,
|
| 703 |
+
"loss": 0.4833,
|
| 704 |
+
"step": 7500
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 3.63,
|
| 708 |
+
"learning_rate": 0.0002433057244705389,
|
| 709 |
+
"loss": 0.4715,
|
| 710 |
+
"step": 7600
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 3.69,
|
| 714 |
+
"learning_rate": 0.00024251939609981125,
|
| 715 |
+
"loss": 0.4881,
|
| 716 |
+
"step": 7700
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 3.74,
|
| 720 |
+
"learning_rate": 0.00024173306772908362,
|
| 721 |
+
"loss": 0.4932,
|
| 722 |
+
"step": 7800
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 3.8,
|
| 726 |
+
"learning_rate": 0.00024094673935835603,
|
| 727 |
+
"loss": 0.49,
|
| 728 |
+
"step": 7900
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 3.86,
|
| 732 |
+
"learning_rate": 0.0002401604109876284,
|
| 733 |
+
"loss": 0.4877,
|
| 734 |
+
"step": 8000
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 3.91,
|
| 738 |
+
"learning_rate": 0.0002393740826169008,
|
| 739 |
+
"loss": 0.4959,
|
| 740 |
+
"step": 8100
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 3.97,
|
| 744 |
+
"learning_rate": 0.00023858775424617318,
|
| 745 |
+
"loss": 0.5031,
|
| 746 |
+
"step": 8200
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 4.03,
|
| 750 |
+
"learning_rate": 0.00023780142587544555,
|
| 751 |
+
"loss": 0.4162,
|
| 752 |
+
"step": 8300
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 4.09,
|
| 756 |
+
"learning_rate": 0.00023701509750471796,
|
| 757 |
+
"loss": 0.33,
|
| 758 |
+
"step": 8400
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 4.14,
|
| 762 |
+
"learning_rate": 0.00023622876913399033,
|
| 763 |
+
"loss": 0.3303,
|
| 764 |
+
"step": 8500
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 4.2,
|
| 768 |
+
"learning_rate": 0.00023544244076326273,
|
| 769 |
+
"loss": 0.3346,
|
| 770 |
+
"step": 8600
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 4.26,
|
| 774 |
+
"learning_rate": 0.0002346561123925351,
|
| 775 |
+
"loss": 0.3454,
|
| 776 |
+
"step": 8700
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 4.31,
|
| 780 |
+
"learning_rate": 0.00023386978402180748,
|
| 781 |
+
"loss": 0.3507,
|
| 782 |
+
"step": 8800
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 4.37,
|
| 786 |
+
"learning_rate": 0.0002330834556510799,
|
| 787 |
+
"loss": 0.3481,
|
| 788 |
+
"step": 8900
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 4.43,
|
| 792 |
+
"learning_rate": 0.00023229712728035223,
|
| 793 |
+
"loss": 0.3461,
|
| 794 |
+
"step": 9000
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 4.49,
|
| 798 |
+
"learning_rate": 0.00023151079890962464,
|
| 799 |
+
"loss": 0.3489,
|
| 800 |
+
"step": 9100
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 4.54,
|
| 804 |
+
"learning_rate": 0.000230724470538897,
|
| 805 |
+
"loss": 0.3534,
|
| 806 |
+
"step": 9200
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 4.6,
|
| 810 |
+
"learning_rate": 0.0002299381421681694,
|
| 811 |
+
"loss": 0.3629,
|
| 812 |
+
"step": 9300
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 4.66,
|
| 816 |
+
"learning_rate": 0.0002291518137974418,
|
| 817 |
+
"loss": 0.3788,
|
| 818 |
+
"step": 9400
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 4.71,
|
| 822 |
+
"learning_rate": 0.00022836548542671417,
|
| 823 |
+
"loss": 0.3588,
|
| 824 |
+
"step": 9500
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 4.77,
|
| 828 |
+
"learning_rate": 0.00022757915705598657,
|
| 829 |
+
"loss": 0.3655,
|
| 830 |
+
"step": 9600
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 4.83,
|
| 834 |
+
"learning_rate": 0.00022679282868525894,
|
| 835 |
+
"loss": 0.3728,
|
| 836 |
+
"step": 9700
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 4.89,
|
| 840 |
+
"learning_rate": 0.00022600650031453132,
|
| 841 |
+
"loss": 0.3782,
|
| 842 |
+
"step": 9800
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 4.94,
|
| 846 |
+
"learning_rate": 0.00022522017194380372,
|
| 847 |
+
"loss": 0.3761,
|
| 848 |
+
"step": 9900
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 5.0,
|
| 852 |
+
"learning_rate": 0.0002244338435730761,
|
| 853 |
+
"loss": 0.3783,
|
| 854 |
+
"step": 10000
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 5.06,
|
| 858 |
+
"learning_rate": 0.0002236475152023485,
|
| 859 |
+
"loss": 0.2337,
|
| 860 |
+
"step": 10100
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 5.11,
|
| 864 |
+
"learning_rate": 0.00022286118683162087,
|
| 865 |
+
"loss": 0.2369,
|
| 866 |
+
"step": 10200
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 5.17,
|
| 870 |
+
"learning_rate": 0.00022207485846089325,
|
| 871 |
+
"loss": 0.235,
|
| 872 |
+
"step": 10300
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 5.23,
|
| 876 |
+
"learning_rate": 0.00022128853009016565,
|
| 877 |
+
"loss": 0.2353,
|
| 878 |
+
"step": 10400
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 5.29,
|
| 882 |
+
"learning_rate": 0.000220502201719438,
|
| 883 |
+
"loss": 0.2448,
|
| 884 |
+
"step": 10500
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 5.34,
|
| 888 |
+
"learning_rate": 0.0002197158733487104,
|
| 889 |
+
"loss": 0.2423,
|
| 890 |
+
"step": 10600
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 5.4,
|
| 894 |
+
"learning_rate": 0.00021892954497798278,
|
| 895 |
+
"loss": 0.2471,
|
| 896 |
+
"step": 10700
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 5.46,
|
| 900 |
+
"learning_rate": 0.00021814321660725515,
|
| 901 |
+
"loss": 0.2479,
|
| 902 |
+
"step": 10800
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 5.51,
|
| 906 |
+
"learning_rate": 0.00021735688823652755,
|
| 907 |
+
"loss": 0.2591,
|
| 908 |
+
"step": 10900
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 5.57,
|
| 912 |
+
"learning_rate": 0.00021657055986579993,
|
| 913 |
+
"loss": 0.2672,
|
| 914 |
+
"step": 11000
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 5.63,
|
| 918 |
+
"learning_rate": 0.00021578423149507233,
|
| 919 |
+
"loss": 0.2676,
|
| 920 |
+
"step": 11100
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 5.69,
|
| 924 |
+
"learning_rate": 0.0002149979031243447,
|
| 925 |
+
"loss": 0.2683,
|
| 926 |
+
"step": 11200
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 5.74,
|
| 930 |
+
"learning_rate": 0.00021421157475361708,
|
| 931 |
+
"loss": 0.2646,
|
| 932 |
+
"step": 11300
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 5.8,
|
| 936 |
+
"learning_rate": 0.00021342524638288948,
|
| 937 |
+
"loss": 0.2759,
|
| 938 |
+
"step": 11400
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 5.86,
|
| 942 |
+
"learning_rate": 0.00021263891801216186,
|
| 943 |
+
"loss": 0.2805,
|
| 944 |
+
"step": 11500
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 5.91,
|
| 948 |
+
"learning_rate": 0.00021185258964143426,
|
| 949 |
+
"loss": 0.2721,
|
| 950 |
+
"step": 11600
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 5.97,
|
| 954 |
+
"learning_rate": 0.00021106626127070664,
|
| 955 |
+
"loss": 0.2785,
|
| 956 |
+
"step": 11700
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 6.03,
|
| 960 |
+
"learning_rate": 0.00021027993289997898,
|
| 961 |
+
"loss": 0.2186,
|
| 962 |
+
"step": 11800
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 6.09,
|
| 966 |
+
"learning_rate": 0.00020949360452925141,
|
| 967 |
+
"loss": 0.1612,
|
| 968 |
+
"step": 11900
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 6.14,
|
| 972 |
+
"learning_rate": 0.00020870727615852376,
|
| 973 |
+
"loss": 0.1628,
|
| 974 |
+
"step": 12000
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 6.2,
|
| 978 |
+
"learning_rate": 0.00020792094778779616,
|
| 979 |
+
"loss": 0.1661,
|
| 980 |
+
"step": 12100
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 6.26,
|
| 984 |
+
"learning_rate": 0.00020713461941706854,
|
| 985 |
+
"loss": 0.1759,
|
| 986 |
+
"step": 12200
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 6.31,
|
| 990 |
+
"learning_rate": 0.00020634829104634091,
|
| 991 |
+
"loss": 0.1803,
|
| 992 |
+
"step": 12300
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 6.37,
|
| 996 |
+
"learning_rate": 0.00020556196267561332,
|
| 997 |
+
"loss": 0.1788,
|
| 998 |
+
"step": 12400
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 6.43,
|
| 1002 |
+
"learning_rate": 0.000204783497588593,
|
| 1003 |
+
"loss": 0.1812,
|
| 1004 |
+
"step": 12500
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 6.49,
|
| 1008 |
+
"learning_rate": 0.00020399716921786537,
|
| 1009 |
+
"loss": 0.1813,
|
| 1010 |
+
"step": 12600
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 6.54,
|
| 1014 |
+
"learning_rate": 0.00020321084084713774,
|
| 1015 |
+
"loss": 0.1876,
|
| 1016 |
+
"step": 12700
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 6.6,
|
| 1020 |
+
"learning_rate": 0.00020242451247641014,
|
| 1021 |
+
"loss": 0.184,
|
| 1022 |
+
"step": 12800
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 6.66,
|
| 1026 |
+
"learning_rate": 0.00020163818410568252,
|
| 1027 |
+
"loss": 0.1964,
|
| 1028 |
+
"step": 12900
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 6.71,
|
| 1032 |
+
"learning_rate": 0.00020085185573495492,
|
| 1033 |
+
"loss": 0.194,
|
| 1034 |
+
"step": 13000
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 6.77,
|
| 1038 |
+
"learning_rate": 0.0002000655273642273,
|
| 1039 |
+
"loss": 0.1956,
|
| 1040 |
+
"step": 13100
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 6.83,
|
| 1044 |
+
"learning_rate": 0.00019927919899349964,
|
| 1045 |
+
"loss": 0.2015,
|
| 1046 |
+
"step": 13200
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 6.89,
|
| 1050 |
+
"learning_rate": 0.00019849287062277207,
|
| 1051 |
+
"loss": 0.2008,
|
| 1052 |
+
"step": 13300
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 6.94,
|
| 1056 |
+
"learning_rate": 0.00019770654225204442,
|
| 1057 |
+
"loss": 0.2035,
|
| 1058 |
+
"step": 13400
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 7.0,
|
| 1062 |
+
"learning_rate": 0.00019692021388131682,
|
| 1063 |
+
"loss": 0.2102,
|
| 1064 |
+
"step": 13500
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 7.06,
|
| 1068 |
+
"learning_rate": 0.0001961338855105892,
|
| 1069 |
+
"loss": 0.1232,
|
| 1070 |
+
"step": 13600
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 7.11,
|
| 1074 |
+
"learning_rate": 0.00019534755713986157,
|
| 1075 |
+
"loss": 0.1186,
|
| 1076 |
+
"step": 13700
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 7.17,
|
| 1080 |
+
"learning_rate": 0.00019456122876913398,
|
| 1081 |
+
"loss": 0.1213,
|
| 1082 |
+
"step": 13800
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 7.23,
|
| 1086 |
+
"learning_rate": 0.00019377490039840635,
|
| 1087 |
+
"loss": 0.1234,
|
| 1088 |
+
"step": 13900
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 7.29,
|
| 1092 |
+
"learning_rate": 0.00019298857202767875,
|
| 1093 |
+
"loss": 0.1284,
|
| 1094 |
+
"step": 14000
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 7.34,
|
| 1098 |
+
"learning_rate": 0.00019220224365695113,
|
| 1099 |
+
"loss": 0.136,
|
| 1100 |
+
"step": 14100
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 7.4,
|
| 1104 |
+
"learning_rate": 0.0001914159152862235,
|
| 1105 |
+
"loss": 0.133,
|
| 1106 |
+
"step": 14200
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 7.46,
|
| 1110 |
+
"learning_rate": 0.0001906295869154959,
|
| 1111 |
+
"loss": 0.1387,
|
| 1112 |
+
"step": 14300
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 7.51,
|
| 1116 |
+
"learning_rate": 0.00018984325854476828,
|
| 1117 |
+
"loss": 0.1373,
|
| 1118 |
+
"step": 14400
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 7.57,
|
| 1122 |
+
"learning_rate": 0.00018906479345774796,
|
| 1123 |
+
"loss": 0.1399,
|
| 1124 |
+
"step": 14500
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 7.63,
|
| 1128 |
+
"learning_rate": 0.0001882784650870203,
|
| 1129 |
+
"loss": 0.1407,
|
| 1130 |
+
"step": 14600
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 7.69,
|
| 1134 |
+
"learning_rate": 0.0001874921367162927,
|
| 1135 |
+
"loss": 0.1506,
|
| 1136 |
+
"step": 14700
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 7.74,
|
| 1140 |
+
"learning_rate": 0.00018670580834556508,
|
| 1141 |
+
"loss": 0.1471,
|
| 1142 |
+
"step": 14800
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 7.8,
|
| 1146 |
+
"learning_rate": 0.00018591947997483746,
|
| 1147 |
+
"loss": 0.1492,
|
| 1148 |
+
"step": 14900
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 7.86,
|
| 1152 |
+
"learning_rate": 0.00018513315160410986,
|
| 1153 |
+
"loss": 0.1495,
|
| 1154 |
+
"step": 15000
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 7.91,
|
| 1158 |
+
"learning_rate": 0.00018434682323338223,
|
| 1159 |
+
"loss": 0.1564,
|
| 1160 |
+
"step": 15100
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 7.97,
|
| 1164 |
+
"learning_rate": 0.00018356049486265464,
|
| 1165 |
+
"loss": 0.1541,
|
| 1166 |
+
"step": 15200
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 8.03,
|
| 1170 |
+
"learning_rate": 0.000182774166491927,
|
| 1171 |
+
"loss": 0.1303,
|
| 1172 |
+
"step": 15300
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 8.09,
|
| 1176 |
+
"learning_rate": 0.00018198783812119939,
|
| 1177 |
+
"loss": 0.0966,
|
| 1178 |
+
"step": 15400
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 8.14,
|
| 1182 |
+
"learning_rate": 0.0001812015097504718,
|
| 1183 |
+
"loss": 0.0969,
|
| 1184 |
+
"step": 15500
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 8.2,
|
| 1188 |
+
"learning_rate": 0.00018041518137974416,
|
| 1189 |
+
"loss": 0.1002,
|
| 1190 |
+
"step": 15600
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 8.26,
|
| 1194 |
+
"learning_rate": 0.00017962885300901657,
|
| 1195 |
+
"loss": 0.0999,
|
| 1196 |
+
"step": 15700
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 8.31,
|
| 1200 |
+
"learning_rate": 0.00017884252463828894,
|
| 1201 |
+
"loss": 0.1012,
|
| 1202 |
+
"step": 15800
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 8.37,
|
| 1206 |
+
"learning_rate": 0.0001780561962675613,
|
| 1207 |
+
"loss": 0.1026,
|
| 1208 |
+
"step": 15900
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 8.43,
|
| 1212 |
+
"learning_rate": 0.00017726986789683372,
|
| 1213 |
+
"loss": 0.1083,
|
| 1214 |
+
"step": 16000
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 8.49,
|
| 1218 |
+
"learning_rate": 0.00017648353952610607,
|
| 1219 |
+
"loss": 0.1086,
|
| 1220 |
+
"step": 16100
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 8.54,
|
| 1224 |
+
"learning_rate": 0.00017569721115537847,
|
| 1225 |
+
"loss": 0.1127,
|
| 1226 |
+
"step": 16200
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 8.6,
|
| 1230 |
+
"learning_rate": 0.00017491088278465084,
|
| 1231 |
+
"loss": 0.1104,
|
| 1232 |
+
"step": 16300
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 8.66,
|
| 1236 |
+
"learning_rate": 0.00017412455441392322,
|
| 1237 |
+
"loss": 0.114,
|
| 1238 |
+
"step": 16400
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 8.71,
|
| 1242 |
+
"learning_rate": 0.00017333822604319562,
|
| 1243 |
+
"loss": 0.1156,
|
| 1244 |
+
"step": 16500
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 8.77,
|
| 1248 |
+
"learning_rate": 0.000172551897672468,
|
| 1249 |
+
"loss": 0.1145,
|
| 1250 |
+
"step": 16600
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 8.83,
|
| 1254 |
+
"learning_rate": 0.00017177343258544767,
|
| 1255 |
+
"loss": 0.1181,
|
| 1256 |
+
"step": 16700
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 8.89,
|
| 1260 |
+
"learning_rate": 0.00017098710421472005,
|
| 1261 |
+
"loss": 0.1239,
|
| 1262 |
+
"step": 16800
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 8.94,
|
| 1266 |
+
"learning_rate": 0.00017020077584399245,
|
| 1267 |
+
"loss": 0.1203,
|
| 1268 |
+
"step": 16900
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 9.0,
|
| 1272 |
+
"learning_rate": 0.00016941444747326482,
|
| 1273 |
+
"loss": 0.1226,
|
| 1274 |
+
"step": 17000
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 9.06,
|
| 1278 |
+
"learning_rate": 0.00016862811910253723,
|
| 1279 |
+
"loss": 0.0842,
|
| 1280 |
+
"step": 17100
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 9.11,
|
| 1284 |
+
"learning_rate": 0.0001678417907318096,
|
| 1285 |
+
"loss": 0.0782,
|
| 1286 |
+
"step": 17200
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 9.17,
|
| 1290 |
+
"learning_rate": 0.00016705546236108195,
|
| 1291 |
+
"loss": 0.0796,
|
| 1292 |
+
"step": 17300
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 9.23,
|
| 1296 |
+
"learning_rate": 0.00016626913399035435,
|
| 1297 |
+
"loss": 0.0814,
|
| 1298 |
+
"step": 17400
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 9.29,
|
| 1302 |
+
"learning_rate": 0.00016548280561962673,
|
| 1303 |
+
"loss": 0.0836,
|
| 1304 |
+
"step": 17500
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 9.34,
|
| 1308 |
+
"learning_rate": 0.00016469647724889913,
|
| 1309 |
+
"loss": 0.0844,
|
| 1310 |
+
"step": 17600
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 9.4,
|
| 1314 |
+
"learning_rate": 0.0001639101488781715,
|
| 1315 |
+
"loss": 0.0876,
|
| 1316 |
+
"step": 17700
|
| 1317 |
+
},
|
| 1318 |
+
{
|
| 1319 |
+
"epoch": 9.46,
|
| 1320 |
+
"learning_rate": 0.00016312382050744388,
|
| 1321 |
+
"loss": 0.0853,
|
| 1322 |
+
"step": 17800
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 9.51,
|
| 1326 |
+
"learning_rate": 0.00016233749213671628,
|
| 1327 |
+
"loss": 0.0922,
|
| 1328 |
+
"step": 17900
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 9.57,
|
| 1332 |
+
"learning_rate": 0.00016155116376598866,
|
| 1333 |
+
"loss": 0.0919,
|
| 1334 |
+
"step": 18000
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 9.63,
|
| 1338 |
+
"learning_rate": 0.00016076483539526106,
|
| 1339 |
+
"loss": 0.0932,
|
| 1340 |
+
"step": 18100
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 9.69,
|
| 1344 |
+
"learning_rate": 0.00015997850702453343,
|
| 1345 |
+
"loss": 0.0919,
|
| 1346 |
+
"step": 18200
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 9.74,
|
| 1350 |
+
"learning_rate": 0.0001591921786538058,
|
| 1351 |
+
"loss": 0.0957,
|
| 1352 |
+
"step": 18300
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 9.8,
|
| 1356 |
+
"learning_rate": 0.0001584058502830782,
|
| 1357 |
+
"loss": 0.0982,
|
| 1358 |
+
"step": 18400
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 9.86,
|
| 1362 |
+
"learning_rate": 0.0001576195219123506,
|
| 1363 |
+
"loss": 0.0988,
|
| 1364 |
+
"step": 18500
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 9.91,
|
| 1368 |
+
"learning_rate": 0.000156833193541623,
|
| 1369 |
+
"loss": 0.1016,
|
| 1370 |
+
"step": 18600
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 9.97,
|
| 1374 |
+
"learning_rate": 0.00015604686517089536,
|
| 1375 |
+
"loss": 0.0967,
|
| 1376 |
+
"step": 18700
|
| 1377 |
+
},
|
| 1378 |
+
{
|
| 1379 |
+
"epoch": 10.03,
|
| 1380 |
+
"learning_rate": 0.0001552605368001677,
|
| 1381 |
+
"loss": 0.0862,
|
| 1382 |
+
"step": 18800
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 10.09,
|
| 1386 |
+
"learning_rate": 0.00015447420842944011,
|
| 1387 |
+
"loss": 0.0693,
|
| 1388 |
+
"step": 18900
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 10.14,
|
| 1392 |
+
"learning_rate": 0.0001536957433424198,
|
| 1393 |
+
"loss": 0.0722,
|
| 1394 |
+
"step": 19000
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 10.2,
|
| 1398 |
+
"learning_rate": 0.00015290941497169216,
|
| 1399 |
+
"loss": 0.0725,
|
| 1400 |
+
"step": 19100
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 10.26,
|
| 1404 |
+
"learning_rate": 0.00015212308660096454,
|
| 1405 |
+
"loss": 0.0716,
|
| 1406 |
+
"step": 19200
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 10.31,
|
| 1410 |
+
"learning_rate": 0.00015133675823023694,
|
| 1411 |
+
"loss": 0.0709,
|
| 1412 |
+
"step": 19300
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 10.37,
|
| 1416 |
+
"learning_rate": 0.00015055042985950932,
|
| 1417 |
+
"loss": 0.0734,
|
| 1418 |
+
"step": 19400
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 10.43,
|
| 1422 |
+
"learning_rate": 0.0001497641014887817,
|
| 1423 |
+
"loss": 0.0761,
|
| 1424 |
+
"step": 19500
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 10.49,
|
| 1428 |
+
"learning_rate": 0.0001489777731180541,
|
| 1429 |
+
"loss": 0.0778,
|
| 1430 |
+
"step": 19600
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 10.54,
|
| 1434 |
+
"learning_rate": 0.00014819144474732647,
|
| 1435 |
+
"loss": 0.0794,
|
| 1436 |
+
"step": 19700
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 10.6,
|
| 1440 |
+
"learning_rate": 0.00014740511637659884,
|
| 1441 |
+
"loss": 0.0811,
|
| 1442 |
+
"step": 19800
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 10.66,
|
| 1446 |
+
"learning_rate": 0.00014661878800587125,
|
| 1447 |
+
"loss": 0.0806,
|
| 1448 |
+
"step": 19900
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 10.71,
|
| 1452 |
+
"learning_rate": 0.00014583245963514362,
|
| 1453 |
+
"loss": 0.0812,
|
| 1454 |
+
"step": 20000
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 10.77,
|
| 1458 |
+
"learning_rate": 0.000145046131264416,
|
| 1459 |
+
"loss": 0.0822,
|
| 1460 |
+
"step": 20100
|
| 1461 |
+
},
|
| 1462 |
+
{
|
| 1463 |
+
"epoch": 10.83,
|
| 1464 |
+
"learning_rate": 0.0001442598028936884,
|
| 1465 |
+
"loss": 0.0828,
|
| 1466 |
+
"step": 20200
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 10.89,
|
| 1470 |
+
"learning_rate": 0.00014347347452296077,
|
| 1471 |
+
"loss": 0.0827,
|
| 1472 |
+
"step": 20300
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 10.94,
|
| 1476 |
+
"learning_rate": 0.00014268714615223315,
|
| 1477 |
+
"loss": 0.0846,
|
| 1478 |
+
"step": 20400
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 11.0,
|
| 1482 |
+
"learning_rate": 0.00014190081778150555,
|
| 1483 |
+
"loss": 0.0881,
|
| 1484 |
+
"step": 20500
|
| 1485 |
+
},
|
| 1486 |
+
{
|
| 1487 |
+
"epoch": 11.06,
|
| 1488 |
+
"learning_rate": 0.00014111448941077793,
|
| 1489 |
+
"loss": 0.0636,
|
| 1490 |
+
"step": 20600
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 11.11,
|
| 1494 |
+
"learning_rate": 0.00014032816104005033,
|
| 1495 |
+
"loss": 0.0628,
|
| 1496 |
+
"step": 20700
|
| 1497 |
+
},
|
| 1498 |
+
{
|
| 1499 |
+
"epoch": 11.17,
|
| 1500 |
+
"learning_rate": 0.00013954183266932268,
|
| 1501 |
+
"loss": 0.063,
|
| 1502 |
+
"step": 20800
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 11.23,
|
| 1506 |
+
"learning_rate": 0.00013875550429859508,
|
| 1507 |
+
"loss": 0.0621,
|
| 1508 |
+
"step": 20900
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 11.29,
|
| 1512 |
+
"learning_rate": 0.00013796917592786745,
|
| 1513 |
+
"loss": 0.0644,
|
| 1514 |
+
"step": 21000
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 11.34,
|
| 1518 |
+
"learning_rate": 0.00013718284755713986,
|
| 1519 |
+
"loss": 0.0636,
|
| 1520 |
+
"step": 21100
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 11.4,
|
| 1524 |
+
"learning_rate": 0.00013639651918641223,
|
| 1525 |
+
"loss": 0.0667,
|
| 1526 |
+
"step": 21200
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"epoch": 11.46,
|
| 1530 |
+
"learning_rate": 0.0001356180540993919,
|
| 1531 |
+
"loss": 0.0686,
|
| 1532 |
+
"step": 21300
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 11.51,
|
| 1536 |
+
"learning_rate": 0.00013483172572866428,
|
| 1537 |
+
"loss": 0.0677,
|
| 1538 |
+
"step": 21400
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 11.57,
|
| 1542 |
+
"learning_rate": 0.00013404539735793666,
|
| 1543 |
+
"loss": 0.0692,
|
| 1544 |
+
"step": 21500
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 11.63,
|
| 1548 |
+
"learning_rate": 0.00013325906898720903,
|
| 1549 |
+
"loss": 0.072,
|
| 1550 |
+
"step": 21600
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 11.69,
|
| 1554 |
+
"learning_rate": 0.00013247274061648143,
|
| 1555 |
+
"loss": 0.0708,
|
| 1556 |
+
"step": 21700
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 11.74,
|
| 1560 |
+
"learning_rate": 0.0001316864122457538,
|
| 1561 |
+
"loss": 0.0738,
|
| 1562 |
+
"step": 21800
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 11.8,
|
| 1566 |
+
"learning_rate": 0.0001309000838750262,
|
| 1567 |
+
"loss": 0.0744,
|
| 1568 |
+
"step": 21900
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 11.86,
|
| 1572 |
+
"learning_rate": 0.00013011375550429859,
|
| 1573 |
+
"loss": 0.0743,
|
| 1574 |
+
"step": 22000
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 11.91,
|
| 1578 |
+
"learning_rate": 0.00012932742713357096,
|
| 1579 |
+
"loss": 0.0742,
|
| 1580 |
+
"step": 22100
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 11.97,
|
| 1584 |
+
"learning_rate": 0.00012854109876284334,
|
| 1585 |
+
"loss": 0.0743,
|
| 1586 |
+
"step": 22200
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 12.03,
|
| 1590 |
+
"learning_rate": 0.00012775477039211574,
|
| 1591 |
+
"loss": 0.0665,
|
| 1592 |
+
"step": 22300
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 12.09,
|
| 1596 |
+
"learning_rate": 0.00012696844202138811,
|
| 1597 |
+
"loss": 0.0583,
|
| 1598 |
+
"step": 22400
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 12.14,
|
| 1602 |
+
"learning_rate": 0.00012618211365066052,
|
| 1603 |
+
"loss": 0.0555,
|
| 1604 |
+
"step": 22500
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 12.2,
|
| 1608 |
+
"learning_rate": 0.0001253957852799329,
|
| 1609 |
+
"loss": 0.0556,
|
| 1610 |
+
"step": 22600
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 12.26,
|
| 1614 |
+
"learning_rate": 0.00012460945690920527,
|
| 1615 |
+
"loss": 0.0596,
|
| 1616 |
+
"step": 22700
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 12.31,
|
| 1620 |
+
"learning_rate": 0.00012382312853847767,
|
| 1621 |
+
"loss": 0.058,
|
| 1622 |
+
"step": 22800
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 12.37,
|
| 1626 |
+
"learning_rate": 0.00012303680016775004,
|
| 1627 |
+
"loss": 0.0599,
|
| 1628 |
+
"step": 22900
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 12.43,
|
| 1632 |
+
"learning_rate": 0.00012225047179702242,
|
| 1633 |
+
"loss": 0.0574,
|
| 1634 |
+
"step": 23000
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 12.49,
|
| 1638 |
+
"learning_rate": 0.00012146414342629481,
|
| 1639 |
+
"loss": 0.0594,
|
| 1640 |
+
"step": 23100
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 12.54,
|
| 1644 |
+
"learning_rate": 0.00012067781505556718,
|
| 1645 |
+
"loss": 0.062,
|
| 1646 |
+
"step": 23200
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 12.6,
|
| 1650 |
+
"learning_rate": 0.00011989148668483957,
|
| 1651 |
+
"loss": 0.0624,
|
| 1652 |
+
"step": 23300
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 12.66,
|
| 1656 |
+
"learning_rate": 0.00011910515831411196,
|
| 1657 |
+
"loss": 0.0618,
|
| 1658 |
+
"step": 23400
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 12.71,
|
| 1662 |
+
"learning_rate": 0.00011831882994338435,
|
| 1663 |
+
"loss": 0.0645,
|
| 1664 |
+
"step": 23500
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 12.77,
|
| 1668 |
+
"learning_rate": 0.00011753250157265673,
|
| 1669 |
+
"loss": 0.0642,
|
| 1670 |
+
"step": 23600
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 12.83,
|
| 1674 |
+
"learning_rate": 0.0001167540364856364,
|
| 1675 |
+
"loss": 0.0651,
|
| 1676 |
+
"step": 23700
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 12.89,
|
| 1680 |
+
"learning_rate": 0.00011596770811490879,
|
| 1681 |
+
"loss": 0.0643,
|
| 1682 |
+
"step": 23800
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 12.94,
|
| 1686 |
+
"learning_rate": 0.00011518137974418116,
|
| 1687 |
+
"loss": 0.0649,
|
| 1688 |
+
"step": 23900
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 13.0,
|
| 1692 |
+
"learning_rate": 0.00011439505137345354,
|
| 1693 |
+
"loss": 0.0654,
|
| 1694 |
+
"step": 24000
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 13.06,
|
| 1698 |
+
"learning_rate": 0.00011360872300272593,
|
| 1699 |
+
"loss": 0.0519,
|
| 1700 |
+
"step": 24100
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 13.11,
|
| 1704 |
+
"learning_rate": 0.00011282239463199832,
|
| 1705 |
+
"loss": 0.0518,
|
| 1706 |
+
"step": 24200
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 13.17,
|
| 1710 |
+
"learning_rate": 0.0001120360662612707,
|
| 1711 |
+
"loss": 0.0517,
|
| 1712 |
+
"step": 24300
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 13.23,
|
| 1716 |
+
"learning_rate": 0.00011124973789054309,
|
| 1717 |
+
"loss": 0.0511,
|
| 1718 |
+
"step": 24400
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 13.29,
|
| 1722 |
+
"learning_rate": 0.00011046340951981545,
|
| 1723 |
+
"loss": 0.0524,
|
| 1724 |
+
"step": 24500
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 13.34,
|
| 1728 |
+
"learning_rate": 0.00010967708114908784,
|
| 1729 |
+
"loss": 0.0533,
|
| 1730 |
+
"step": 24600
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 13.4,
|
| 1734 |
+
"learning_rate": 0.00010889075277836023,
|
| 1735 |
+
"loss": 0.0541,
|
| 1736 |
+
"step": 24700
|
| 1737 |
+
},
|
| 1738 |
+
{
|
| 1739 |
+
"epoch": 13.46,
|
| 1740 |
+
"learning_rate": 0.00010810442440763262,
|
| 1741 |
+
"loss": 0.0544,
|
| 1742 |
+
"step": 24800
|
| 1743 |
+
},
|
| 1744 |
+
{
|
| 1745 |
+
"epoch": 13.51,
|
| 1746 |
+
"learning_rate": 0.00010731809603690501,
|
| 1747 |
+
"loss": 0.0549,
|
| 1748 |
+
"step": 24900
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 13.57,
|
| 1752 |
+
"learning_rate": 0.00010653176766617738,
|
| 1753 |
+
"loss": 0.0555,
|
| 1754 |
+
"step": 25000
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 13.63,
|
| 1758 |
+
"learning_rate": 0.00010574543929544977,
|
| 1759 |
+
"loss": 0.0553,
|
| 1760 |
+
"step": 25100
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 13.69,
|
| 1764 |
+
"learning_rate": 0.00010495911092472216,
|
| 1765 |
+
"loss": 0.0551,
|
| 1766 |
+
"step": 25200
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 13.74,
|
| 1770 |
+
"learning_rate": 0.00010417278255399454,
|
| 1771 |
+
"loss": 0.0565,
|
| 1772 |
+
"step": 25300
|
| 1773 |
+
},
|
| 1774 |
+
{
|
| 1775 |
+
"epoch": 13.8,
|
| 1776 |
+
"learning_rate": 0.00010338645418326693,
|
| 1777 |
+
"loss": 0.0575,
|
| 1778 |
+
"step": 25400
|
| 1779 |
+
},
|
| 1780 |
+
{
|
| 1781 |
+
"epoch": 13.86,
|
| 1782 |
+
"learning_rate": 0.0001026001258125393,
|
| 1783 |
+
"loss": 0.0563,
|
| 1784 |
+
"step": 25500
|
| 1785 |
+
},
|
| 1786 |
+
{
|
| 1787 |
+
"epoch": 13.91,
|
| 1788 |
+
"learning_rate": 0.00010181379744181169,
|
| 1789 |
+
"loss": 0.057,
|
| 1790 |
+
"step": 25600
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 13.97,
|
| 1794 |
+
"learning_rate": 0.00010102746907108408,
|
| 1795 |
+
"loss": 0.0574,
|
| 1796 |
+
"step": 25700
|
| 1797 |
+
},
|
| 1798 |
+
{
|
| 1799 |
+
"epoch": 14.03,
|
| 1800 |
+
"learning_rate": 0.00010024114070035647,
|
| 1801 |
+
"loss": 0.0556,
|
| 1802 |
+
"step": 25800
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 14.09,
|
| 1806 |
+
"learning_rate": 9.945481232962886e-05,
|
| 1807 |
+
"loss": 0.049,
|
| 1808 |
+
"step": 25900
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 14.14,
|
| 1812 |
+
"learning_rate": 9.866848395890122e-05,
|
| 1813 |
+
"loss": 0.0477,
|
| 1814 |
+
"step": 26000
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 14.2,
|
| 1818 |
+
"learning_rate": 9.78821555881736e-05,
|
| 1819 |
+
"loss": 0.0477,
|
| 1820 |
+
"step": 26100
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 14.26,
|
| 1824 |
+
"learning_rate": 9.7095827217446e-05,
|
| 1825 |
+
"loss": 0.0481,
|
| 1826 |
+
"step": 26200
|
| 1827 |
+
},
|
| 1828 |
+
{
|
| 1829 |
+
"epoch": 14.31,
|
| 1830 |
+
"learning_rate": 9.630949884671838e-05,
|
| 1831 |
+
"loss": 0.0466,
|
| 1832 |
+
"step": 26300
|
| 1833 |
+
},
|
| 1834 |
+
{
|
| 1835 |
+
"epoch": 14.37,
|
| 1836 |
+
"learning_rate": 9.552317047599077e-05,
|
| 1837 |
+
"loss": 0.049,
|
| 1838 |
+
"step": 26400
|
| 1839 |
+
},
|
| 1840 |
+
{
|
| 1841 |
+
"epoch": 14.43,
|
| 1842 |
+
"learning_rate": 9.473684210526315e-05,
|
| 1843 |
+
"loss": 0.0509,
|
| 1844 |
+
"step": 26500
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 14.49,
|
| 1848 |
+
"learning_rate": 9.395051373453554e-05,
|
| 1849 |
+
"loss": 0.0488,
|
| 1850 |
+
"step": 26600
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 14.54,
|
| 1854 |
+
"learning_rate": 9.316418536380793e-05,
|
| 1855 |
+
"loss": 0.0509,
|
| 1856 |
+
"step": 26700
|
| 1857 |
+
},
|
| 1858 |
+
{
|
| 1859 |
+
"epoch": 14.6,
|
| 1860 |
+
"learning_rate": 9.23778569930803e-05,
|
| 1861 |
+
"loss": 0.0515,
|
| 1862 |
+
"step": 26800
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 14.66,
|
| 1866 |
+
"learning_rate": 9.159152862235269e-05,
|
| 1867 |
+
"loss": 0.0523,
|
| 1868 |
+
"step": 26900
|
| 1869 |
+
},
|
| 1870 |
+
{
|
| 1871 |
+
"epoch": 14.71,
|
| 1872 |
+
"learning_rate": 9.080520025162506e-05,
|
| 1873 |
+
"loss": 0.0499,
|
| 1874 |
+
"step": 27000
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 14.77,
|
| 1878 |
+
"learning_rate": 9.001887188089745e-05,
|
| 1879 |
+
"loss": 0.0507,
|
| 1880 |
+
"step": 27100
|
| 1881 |
+
},
|
| 1882 |
+
{
|
| 1883 |
+
"epoch": 14.83,
|
| 1884 |
+
"learning_rate": 8.923254351016984e-05,
|
| 1885 |
+
"loss": 0.0524,
|
| 1886 |
+
"step": 27200
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 14.89,
|
| 1890 |
+
"learning_rate": 8.844621513944223e-05,
|
| 1891 |
+
"loss": 0.0525,
|
| 1892 |
+
"step": 27300
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 14.94,
|
| 1896 |
+
"learning_rate": 8.765988676871462e-05,
|
| 1897 |
+
"loss": 0.0522,
|
| 1898 |
+
"step": 27400
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 15.0,
|
| 1902 |
+
"learning_rate": 8.687355839798698e-05,
|
| 1903 |
+
"loss": 0.0547,
|
| 1904 |
+
"step": 27500
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 15.06,
|
| 1908 |
+
"learning_rate": 8.608723002725937e-05,
|
| 1909 |
+
"loss": 0.0449,
|
| 1910 |
+
"step": 27600
|
| 1911 |
+
},
|
| 1912 |
+
{
|
| 1913 |
+
"epoch": 15.11,
|
| 1914 |
+
"learning_rate": 8.530090165653176e-05,
|
| 1915 |
+
"loss": 0.0455,
|
| 1916 |
+
"step": 27700
|
| 1917 |
+
},
|
| 1918 |
+
{
|
| 1919 |
+
"epoch": 15.17,
|
| 1920 |
+
"learning_rate": 8.452243656951142e-05,
|
| 1921 |
+
"loss": 0.0429,
|
| 1922 |
+
"step": 27800
|
| 1923 |
+
},
|
| 1924 |
+
{
|
| 1925 |
+
"epoch": 15.23,
|
| 1926 |
+
"learning_rate": 8.373610819878381e-05,
|
| 1927 |
+
"loss": 0.0447,
|
| 1928 |
+
"step": 27900
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 15.29,
|
| 1932 |
+
"learning_rate": 8.29497798280562e-05,
|
| 1933 |
+
"loss": 0.0458,
|
| 1934 |
+
"step": 28000
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 15.34,
|
| 1938 |
+
"learning_rate": 8.216345145732857e-05,
|
| 1939 |
+
"loss": 0.0451,
|
| 1940 |
+
"step": 28100
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 15.4,
|
| 1944 |
+
"learning_rate": 8.137712308660096e-05,
|
| 1945 |
+
"loss": 0.047,
|
| 1946 |
+
"step": 28200
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 15.46,
|
| 1950 |
+
"learning_rate": 8.059079471587334e-05,
|
| 1951 |
+
"loss": 0.0465,
|
| 1952 |
+
"step": 28300
|
| 1953 |
+
},
|
| 1954 |
+
{
|
| 1955 |
+
"epoch": 15.51,
|
| 1956 |
+
"learning_rate": 7.980446634514572e-05,
|
| 1957 |
+
"loss": 0.0461,
|
| 1958 |
+
"step": 28400
|
| 1959 |
+
},
|
| 1960 |
+
{
|
| 1961 |
+
"epoch": 15.57,
|
| 1962 |
+
"learning_rate": 7.901813797441811e-05,
|
| 1963 |
+
"loss": 0.0456,
|
| 1964 |
+
"step": 28500
|
| 1965 |
+
},
|
| 1966 |
+
{
|
| 1967 |
+
"epoch": 15.63,
|
| 1968 |
+
"learning_rate": 7.82318096036905e-05,
|
| 1969 |
+
"loss": 0.0459,
|
| 1970 |
+
"step": 28600
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 15.69,
|
| 1974 |
+
"learning_rate": 7.744548123296289e-05,
|
| 1975 |
+
"loss": 0.0461,
|
| 1976 |
+
"step": 28700
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 15.74,
|
| 1980 |
+
"learning_rate": 7.665915286223525e-05,
|
| 1981 |
+
"loss": 0.0469,
|
| 1982 |
+
"step": 28800
|
| 1983 |
+
},
|
| 1984 |
+
{
|
| 1985 |
+
"epoch": 15.8,
|
| 1986 |
+
"learning_rate": 7.587282449150764e-05,
|
| 1987 |
+
"loss": 0.0478,
|
| 1988 |
+
"step": 28900
|
| 1989 |
+
},
|
| 1990 |
+
{
|
| 1991 |
+
"epoch": 15.86,
|
| 1992 |
+
"learning_rate": 7.508649612078003e-05,
|
| 1993 |
+
"loss": 0.0474,
|
| 1994 |
+
"step": 29000
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 15.91,
|
| 1998 |
+
"learning_rate": 7.430016775005242e-05,
|
| 1999 |
+
"loss": 0.0469,
|
| 2000 |
+
"step": 29100
|
| 2001 |
+
},
|
| 2002 |
+
{
|
| 2003 |
+
"epoch": 15.97,
|
| 2004 |
+
"learning_rate": 7.35138393793248e-05,
|
| 2005 |
+
"loss": 0.0471,
|
| 2006 |
+
"step": 29200
|
| 2007 |
+
},
|
| 2008 |
+
{
|
| 2009 |
+
"epoch": 16.03,
|
| 2010 |
+
"learning_rate": 7.272751100859718e-05,
|
| 2011 |
+
"loss": 0.0461,
|
| 2012 |
+
"step": 29300
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 16.09,
|
| 2016 |
+
"learning_rate": 7.194118263786957e-05,
|
| 2017 |
+
"loss": 0.0404,
|
| 2018 |
+
"step": 29400
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 16.14,
|
| 2022 |
+
"learning_rate": 7.115485426714195e-05,
|
| 2023 |
+
"loss": 0.0415,
|
| 2024 |
+
"step": 29500
|
| 2025 |
+
},
|
| 2026 |
+
{
|
| 2027 |
+
"epoch": 16.2,
|
| 2028 |
+
"learning_rate": 7.036852589641434e-05,
|
| 2029 |
+
"loss": 0.0417,
|
| 2030 |
+
"step": 29600
|
| 2031 |
+
},
|
| 2032 |
+
{
|
| 2033 |
+
"epoch": 16.26,
|
| 2034 |
+
"learning_rate": 6.958219752568672e-05,
|
| 2035 |
+
"loss": 0.0416,
|
| 2036 |
+
"step": 29700
|
| 2037 |
+
},
|
| 2038 |
+
{
|
| 2039 |
+
"epoch": 16.31,
|
| 2040 |
+
"learning_rate": 6.879586915495911e-05,
|
| 2041 |
+
"loss": 0.0419,
|
| 2042 |
+
"step": 29800
|
| 2043 |
+
},
|
| 2044 |
+
{
|
| 2045 |
+
"epoch": 16.37,
|
| 2046 |
+
"learning_rate": 6.800954078423149e-05,
|
| 2047 |
+
"loss": 0.0435,
|
| 2048 |
+
"step": 29900
|
| 2049 |
+
},
|
| 2050 |
+
{
|
| 2051 |
+
"epoch": 16.43,
|
| 2052 |
+
"learning_rate": 6.723107569721115e-05,
|
| 2053 |
+
"loss": 0.0418,
|
| 2054 |
+
"step": 30000
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 16.49,
|
| 2058 |
+
"learning_rate": 6.644474732648354e-05,
|
| 2059 |
+
"loss": 0.0429,
|
| 2060 |
+
"step": 30100
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 16.54,
|
| 2064 |
+
"learning_rate": 6.565841895575591e-05,
|
| 2065 |
+
"loss": 0.0419,
|
| 2066 |
+
"step": 30200
|
| 2067 |
+
},
|
| 2068 |
+
{
|
| 2069 |
+
"epoch": 16.6,
|
| 2070 |
+
"learning_rate": 6.48720905850283e-05,
|
| 2071 |
+
"loss": 0.0416,
|
| 2072 |
+
"step": 30300
|
| 2073 |
+
},
|
| 2074 |
+
{
|
| 2075 |
+
"epoch": 16.66,
|
| 2076 |
+
"learning_rate": 6.408576221430069e-05,
|
| 2077 |
+
"loss": 0.0423,
|
| 2078 |
+
"step": 30400
|
| 2079 |
+
},
|
| 2080 |
+
{
|
| 2081 |
+
"epoch": 16.71,
|
| 2082 |
+
"learning_rate": 6.329943384357306e-05,
|
| 2083 |
+
"loss": 0.0433,
|
| 2084 |
+
"step": 30500
|
| 2085 |
+
},
|
| 2086 |
+
{
|
| 2087 |
+
"epoch": 16.77,
|
| 2088 |
+
"learning_rate": 6.251310547284545e-05,
|
| 2089 |
+
"loss": 0.044,
|
| 2090 |
+
"step": 30600
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 16.83,
|
| 2094 |
+
"learning_rate": 6.172677710211784e-05,
|
| 2095 |
+
"loss": 0.0434,
|
| 2096 |
+
"step": 30700
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 16.89,
|
| 2100 |
+
"learning_rate": 6.0940448731390224e-05,
|
| 2101 |
+
"loss": 0.0436,
|
| 2102 |
+
"step": 30800
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 16.94,
|
| 2106 |
+
"learning_rate": 6.015412036066261e-05,
|
| 2107 |
+
"loss": 0.0441,
|
| 2108 |
+
"step": 30900
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 17.0,
|
| 2112 |
+
"learning_rate": 5.936779198993499e-05,
|
| 2113 |
+
"loss": 0.044,
|
| 2114 |
+
"step": 31000
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 17.06,
|
| 2118 |
+
"learning_rate": 5.8581463619207377e-05,
|
| 2119 |
+
"loss": 0.0369,
|
| 2120 |
+
"step": 31100
|
| 2121 |
+
},
|
| 2122 |
+
{
|
| 2123 |
+
"epoch": 17.11,
|
| 2124 |
+
"learning_rate": 5.779513524847976e-05,
|
| 2125 |
+
"loss": 0.0385,
|
| 2126 |
+
"step": 31200
|
| 2127 |
+
},
|
| 2128 |
+
{
|
| 2129 |
+
"epoch": 17.17,
|
| 2130 |
+
"learning_rate": 5.700880687775215e-05,
|
| 2131 |
+
"loss": 0.0398,
|
| 2132 |
+
"step": 31300
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 17.23,
|
| 2136 |
+
"learning_rate": 5.622247850702453e-05,
|
| 2137 |
+
"loss": 0.0389,
|
| 2138 |
+
"step": 31400
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 17.29,
|
| 2142 |
+
"learning_rate": 5.543615013629691e-05,
|
| 2143 |
+
"loss": 0.0387,
|
| 2144 |
+
"step": 31500
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 17.34,
|
| 2148 |
+
"learning_rate": 5.46498217655693e-05,
|
| 2149 |
+
"loss": 0.0398,
|
| 2150 |
+
"step": 31600
|
| 2151 |
+
},
|
| 2152 |
+
{
|
| 2153 |
+
"epoch": 17.4,
|
| 2154 |
+
"learning_rate": 5.3863493394841675e-05,
|
| 2155 |
+
"loss": 0.039,
|
| 2156 |
+
"step": 31700
|
| 2157 |
+
},
|
| 2158 |
+
{
|
| 2159 |
+
"epoch": 17.46,
|
| 2160 |
+
"learning_rate": 5.3077165024114064e-05,
|
| 2161 |
+
"loss": 0.041,
|
| 2162 |
+
"step": 31800
|
| 2163 |
+
},
|
| 2164 |
+
{
|
| 2165 |
+
"epoch": 17.51,
|
| 2166 |
+
"learning_rate": 5.229083665338645e-05,
|
| 2167 |
+
"loss": 0.0395,
|
| 2168 |
+
"step": 31900
|
| 2169 |
+
},
|
| 2170 |
+
{
|
| 2171 |
+
"epoch": 17.57,
|
| 2172 |
+
"learning_rate": 5.1504508282658835e-05,
|
| 2173 |
+
"loss": 0.0405,
|
| 2174 |
+
"step": 32000
|
| 2175 |
+
},
|
| 2176 |
+
{
|
| 2177 |
+
"epoch": 17.63,
|
| 2178 |
+
"learning_rate": 5.071817991193122e-05,
|
| 2179 |
+
"loss": 0.0401,
|
| 2180 |
+
"step": 32100
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"epoch": 17.69,
|
| 2184 |
+
"learning_rate": 4.993971482491088e-05,
|
| 2185 |
+
"loss": 0.0404,
|
| 2186 |
+
"step": 32200
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 17.74,
|
| 2190 |
+
"learning_rate": 4.915338645418326e-05,
|
| 2191 |
+
"loss": 0.0399,
|
| 2192 |
+
"step": 32300
|
| 2193 |
+
},
|
| 2194 |
+
{
|
| 2195 |
+
"epoch": 17.8,
|
| 2196 |
+
"learning_rate": 4.836705808345565e-05,
|
| 2197 |
+
"loss": 0.0384,
|
| 2198 |
+
"step": 32400
|
| 2199 |
+
},
|
| 2200 |
+
{
|
| 2201 |
+
"epoch": 17.86,
|
| 2202 |
+
"learning_rate": 4.758072971272803e-05,
|
| 2203 |
+
"loss": 0.0421,
|
| 2204 |
+
"step": 32500
|
| 2205 |
+
},
|
| 2206 |
+
{
|
| 2207 |
+
"epoch": 17.91,
|
| 2208 |
+
"learning_rate": 4.679440134200042e-05,
|
| 2209 |
+
"loss": 0.0408,
|
| 2210 |
+
"step": 32600
|
| 2211 |
+
},
|
| 2212 |
+
{
|
| 2213 |
+
"epoch": 17.97,
|
| 2214 |
+
"learning_rate": 4.60080729712728e-05,
|
| 2215 |
+
"loss": 0.0406,
|
| 2216 |
+
"step": 32700
|
| 2217 |
+
},
|
| 2218 |
+
{
|
| 2219 |
+
"epoch": 18.03,
|
| 2220 |
+
"learning_rate": 4.522174460054518e-05,
|
| 2221 |
+
"loss": 0.0388,
|
| 2222 |
+
"step": 32800
|
| 2223 |
+
},
|
| 2224 |
+
{
|
| 2225 |
+
"epoch": 18.09,
|
| 2226 |
+
"learning_rate": 4.443541622981757e-05,
|
| 2227 |
+
"loss": 0.0374,
|
| 2228 |
+
"step": 32900
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 18.14,
|
| 2232 |
+
"learning_rate": 4.3649087859089946e-05,
|
| 2233 |
+
"loss": 0.0367,
|
| 2234 |
+
"step": 33000
|
| 2235 |
+
},
|
| 2236 |
+
{
|
| 2237 |
+
"epoch": 18.2,
|
| 2238 |
+
"learning_rate": 4.2862759488362335e-05,
|
| 2239 |
+
"loss": 0.0367,
|
| 2240 |
+
"step": 33100
|
| 2241 |
+
},
|
| 2242 |
+
{
|
| 2243 |
+
"epoch": 18.26,
|
| 2244 |
+
"learning_rate": 4.2076431117634723e-05,
|
| 2245 |
+
"loss": 0.0367,
|
| 2246 |
+
"step": 33200
|
| 2247 |
+
},
|
| 2248 |
+
{
|
| 2249 |
+
"epoch": 18.31,
|
| 2250 |
+
"learning_rate": 4.1290102746907105e-05,
|
| 2251 |
+
"loss": 0.0367,
|
| 2252 |
+
"step": 33300
|
| 2253 |
+
},
|
| 2254 |
+
{
|
| 2255 |
+
"epoch": 18.37,
|
| 2256 |
+
"learning_rate": 4.050377437617949e-05,
|
| 2257 |
+
"loss": 0.0376,
|
| 2258 |
+
"step": 33400
|
| 2259 |
+
},
|
| 2260 |
+
{
|
| 2261 |
+
"epoch": 18.43,
|
| 2262 |
+
"learning_rate": 3.971744600545187e-05,
|
| 2263 |
+
"loss": 0.0381,
|
| 2264 |
+
"step": 33500
|
| 2265 |
+
},
|
| 2266 |
+
{
|
| 2267 |
+
"epoch": 18.49,
|
| 2268 |
+
"learning_rate": 3.893111763472426e-05,
|
| 2269 |
+
"loss": 0.0378,
|
| 2270 |
+
"step": 33600
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"epoch": 18.54,
|
| 2274 |
+
"learning_rate": 3.814478926399665e-05,
|
| 2275 |
+
"loss": 0.0383,
|
| 2276 |
+
"step": 33700
|
| 2277 |
+
},
|
| 2278 |
+
{
|
| 2279 |
+
"epoch": 18.6,
|
| 2280 |
+
"learning_rate": 3.735846089326902e-05,
|
| 2281 |
+
"loss": 0.0377,
|
| 2282 |
+
"step": 33800
|
| 2283 |
+
},
|
| 2284 |
+
{
|
| 2285 |
+
"epoch": 18.66,
|
| 2286 |
+
"learning_rate": 3.657213252254141e-05,
|
| 2287 |
+
"loss": 0.0381,
|
| 2288 |
+
"step": 33900
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 18.71,
|
| 2292 |
+
"learning_rate": 3.578580415181379e-05,
|
| 2293 |
+
"loss": 0.0382,
|
| 2294 |
+
"step": 34000
|
| 2295 |
+
},
|
| 2296 |
+
{
|
| 2297 |
+
"epoch": 18.77,
|
| 2298 |
+
"learning_rate": 3.4999475781086175e-05,
|
| 2299 |
+
"loss": 0.0382,
|
| 2300 |
+
"step": 34100
|
| 2301 |
+
},
|
| 2302 |
+
{
|
| 2303 |
+
"epoch": 18.83,
|
| 2304 |
+
"learning_rate": 3.4213147410358564e-05,
|
| 2305 |
+
"loss": 0.0392,
|
| 2306 |
+
"step": 34200
|
| 2307 |
+
},
|
| 2308 |
+
{
|
| 2309 |
+
"epoch": 18.89,
|
| 2310 |
+
"learning_rate": 3.3434682323338223e-05,
|
| 2311 |
+
"loss": 0.0388,
|
| 2312 |
+
"step": 34300
|
| 2313 |
+
},
|
| 2314 |
+
{
|
| 2315 |
+
"epoch": 18.94,
|
| 2316 |
+
"learning_rate": 3.2648353952610605e-05,
|
| 2317 |
+
"loss": 0.0386,
|
| 2318 |
+
"step": 34400
|
| 2319 |
+
},
|
| 2320 |
+
{
|
| 2321 |
+
"epoch": 19.0,
|
| 2322 |
+
"learning_rate": 3.186202558188299e-05,
|
| 2323 |
+
"loss": 0.0383,
|
| 2324 |
+
"step": 34500
|
| 2325 |
+
},
|
| 2326 |
+
{
|
| 2327 |
+
"epoch": 19.06,
|
| 2328 |
+
"learning_rate": 3.1075697211155376e-05,
|
| 2329 |
+
"loss": 0.0343,
|
| 2330 |
+
"step": 34600
|
| 2331 |
+
},
|
| 2332 |
+
{
|
| 2333 |
+
"epoch": 19.11,
|
| 2334 |
+
"learning_rate": 3.028936884042776e-05,
|
| 2335 |
+
"loss": 0.0349,
|
| 2336 |
+
"step": 34700
|
| 2337 |
+
},
|
| 2338 |
+
{
|
| 2339 |
+
"epoch": 19.17,
|
| 2340 |
+
"learning_rate": 2.9503040469700143e-05,
|
| 2341 |
+
"loss": 0.0355,
|
| 2342 |
+
"step": 34800
|
| 2343 |
+
},
|
| 2344 |
+
{
|
| 2345 |
+
"epoch": 19.23,
|
| 2346 |
+
"learning_rate": 2.871671209897253e-05,
|
| 2347 |
+
"loss": 0.035,
|
| 2348 |
+
"step": 34900
|
| 2349 |
+
},
|
| 2350 |
+
{
|
| 2351 |
+
"epoch": 19.29,
|
| 2352 |
+
"learning_rate": 2.793038372824491e-05,
|
| 2353 |
+
"loss": 0.0353,
|
| 2354 |
+
"step": 35000
|
| 2355 |
+
},
|
| 2356 |
+
{
|
| 2357 |
+
"epoch": 19.34,
|
| 2358 |
+
"learning_rate": 2.71440553575173e-05,
|
| 2359 |
+
"loss": 0.035,
|
| 2360 |
+
"step": 35100
|
| 2361 |
+
},
|
| 2362 |
+
{
|
| 2363 |
+
"epoch": 19.4,
|
| 2364 |
+
"learning_rate": 2.635772698678968e-05,
|
| 2365 |
+
"loss": 0.0355,
|
| 2366 |
+
"step": 35200
|
| 2367 |
+
},
|
| 2368 |
+
{
|
| 2369 |
+
"epoch": 19.46,
|
| 2370 |
+
"learning_rate": 2.5571398616062064e-05,
|
| 2371 |
+
"loss": 0.0351,
|
| 2372 |
+
"step": 35300
|
| 2373 |
+
},
|
| 2374 |
+
{
|
| 2375 |
+
"epoch": 19.51,
|
| 2376 |
+
"learning_rate": 2.478507024533445e-05,
|
| 2377 |
+
"loss": 0.0358,
|
| 2378 |
+
"step": 35400
|
| 2379 |
+
},
|
| 2380 |
+
{
|
| 2381 |
+
"epoch": 19.57,
|
| 2382 |
+
"learning_rate": 2.399874187460683e-05,
|
| 2383 |
+
"loss": 0.0386,
|
| 2384 |
+
"step": 35500
|
| 2385 |
+
},
|
| 2386 |
+
{
|
| 2387 |
+
"epoch": 19.63,
|
| 2388 |
+
"learning_rate": 2.321241350387922e-05,
|
| 2389 |
+
"loss": 0.0362,
|
| 2390 |
+
"step": 35600
|
| 2391 |
+
},
|
| 2392 |
+
{
|
| 2393 |
+
"epoch": 19.69,
|
| 2394 |
+
"learning_rate": 2.24260851331516e-05,
|
| 2395 |
+
"loss": 0.0364,
|
| 2396 |
+
"step": 35700
|
| 2397 |
+
},
|
| 2398 |
+
{
|
| 2399 |
+
"epoch": 19.74,
|
| 2400 |
+
"learning_rate": 2.1639756762423987e-05,
|
| 2401 |
+
"loss": 0.0364,
|
| 2402 |
+
"step": 35800
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 19.8,
|
| 2406 |
+
"learning_rate": 2.085342839169637e-05,
|
| 2407 |
+
"loss": 0.0344,
|
| 2408 |
+
"step": 35900
|
| 2409 |
+
},
|
| 2410 |
+
{
|
| 2411 |
+
"epoch": 19.86,
|
| 2412 |
+
"learning_rate": 2.0067100020968754e-05,
|
| 2413 |
+
"loss": 0.0372,
|
| 2414 |
+
"step": 36000
|
| 2415 |
+
},
|
| 2416 |
+
{
|
| 2417 |
+
"epoch": 19.91,
|
| 2418 |
+
"learning_rate": 1.928077165024114e-05,
|
| 2419 |
+
"loss": 0.0366,
|
| 2420 |
+
"step": 36100
|
| 2421 |
+
},
|
| 2422 |
+
{
|
| 2423 |
+
"epoch": 19.97,
|
| 2424 |
+
"learning_rate": 1.8494443279513525e-05,
|
| 2425 |
+
"loss": 0.0357,
|
| 2426 |
+
"step": 36200
|
| 2427 |
+
},
|
| 2428 |
+
{
|
| 2429 |
+
"epoch": 20.03,
|
| 2430 |
+
"learning_rate": 1.7708114908785907e-05,
|
| 2431 |
+
"loss": 0.0349,
|
| 2432 |
+
"step": 36300
|
| 2433 |
+
},
|
| 2434 |
+
{
|
| 2435 |
+
"epoch": 20.09,
|
| 2436 |
+
"learning_rate": 1.692178653805829e-05,
|
| 2437 |
+
"loss": 0.0329,
|
| 2438 |
+
"step": 36400
|
| 2439 |
+
},
|
| 2440 |
+
{
|
| 2441 |
+
"epoch": 20.14,
|
| 2442 |
+
"learning_rate": 1.6143321451037952e-05,
|
| 2443 |
+
"loss": 0.0339,
|
| 2444 |
+
"step": 36500
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 20.2,
|
| 2448 |
+
"learning_rate": 1.5356993080310334e-05,
|
| 2449 |
+
"loss": 0.0329,
|
| 2450 |
+
"step": 36600
|
| 2451 |
+
},
|
| 2452 |
+
{
|
| 2453 |
+
"epoch": 20.26,
|
| 2454 |
+
"learning_rate": 1.4570664709582721e-05,
|
| 2455 |
+
"loss": 0.0329,
|
| 2456 |
+
"step": 36700
|
| 2457 |
+
},
|
| 2458 |
+
{
|
| 2459 |
+
"epoch": 20.31,
|
| 2460 |
+
"learning_rate": 1.3784336338855105e-05,
|
| 2461 |
+
"loss": 0.0329,
|
| 2462 |
+
"step": 36800
|
| 2463 |
+
},
|
| 2464 |
+
{
|
| 2465 |
+
"epoch": 20.37,
|
| 2466 |
+
"learning_rate": 1.299800796812749e-05,
|
| 2467 |
+
"loss": 0.0342,
|
| 2468 |
+
"step": 36900
|
| 2469 |
+
},
|
| 2470 |
+
{
|
| 2471 |
+
"epoch": 20.43,
|
| 2472 |
+
"learning_rate": 1.2211679597399874e-05,
|
| 2473 |
+
"loss": 0.0344,
|
| 2474 |
+
"step": 37000
|
| 2475 |
+
},
|
| 2476 |
+
{
|
| 2477 |
+
"epoch": 20.49,
|
| 2478 |
+
"learning_rate": 1.1425351226672256e-05,
|
| 2479 |
+
"loss": 0.0336,
|
| 2480 |
+
"step": 37100
|
| 2481 |
+
},
|
| 2482 |
+
{
|
| 2483 |
+
"epoch": 20.54,
|
| 2484 |
+
"learning_rate": 1.0639022855944641e-05,
|
| 2485 |
+
"loss": 0.0339,
|
| 2486 |
+
"step": 37200
|
| 2487 |
+
},
|
| 2488 |
+
{
|
| 2489 |
+
"epoch": 20.6,
|
| 2490 |
+
"learning_rate": 9.852694485217025e-06,
|
| 2491 |
+
"loss": 0.0345,
|
| 2492 |
+
"step": 37300
|
| 2493 |
+
},
|
| 2494 |
+
{
|
| 2495 |
+
"epoch": 20.66,
|
| 2496 |
+
"learning_rate": 9.06636611448941e-06,
|
| 2497 |
+
"loss": 0.034,
|
| 2498 |
+
"step": 37400
|
| 2499 |
+
},
|
| 2500 |
+
{
|
| 2501 |
+
"epoch": 20.71,
|
| 2502 |
+
"learning_rate": 8.280037743761794e-06,
|
| 2503 |
+
"loss": 0.0343,
|
| 2504 |
+
"step": 37500
|
| 2505 |
+
},
|
| 2506 |
+
{
|
| 2507 |
+
"epoch": 20.77,
|
| 2508 |
+
"learning_rate": 7.4937093730341786e-06,
|
| 2509 |
+
"loss": 0.0339,
|
| 2510 |
+
"step": 37600
|
| 2511 |
+
},
|
| 2512 |
+
{
|
| 2513 |
+
"epoch": 20.83,
|
| 2514 |
+
"learning_rate": 6.707381002306563e-06,
|
| 2515 |
+
"loss": 0.0342,
|
| 2516 |
+
"step": 37700
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 20.89,
|
| 2520 |
+
"learning_rate": 5.921052631578947e-06,
|
| 2521 |
+
"loss": 0.036,
|
| 2522 |
+
"step": 37800
|
| 2523 |
+
},
|
| 2524 |
+
{
|
| 2525 |
+
"epoch": 20.94,
|
| 2526 |
+
"learning_rate": 5.134724260851331e-06,
|
| 2527 |
+
"loss": 0.0336,
|
| 2528 |
+
"step": 37900
|
| 2529 |
+
},
|
| 2530 |
+
{
|
| 2531 |
+
"epoch": 21.0,
|
| 2532 |
+
"learning_rate": 4.348395890123715e-06,
|
| 2533 |
+
"loss": 0.0342,
|
| 2534 |
+
"step": 38000
|
| 2535 |
+
},
|
| 2536 |
+
{
|
| 2537 |
+
"epoch": 21.06,
|
| 2538 |
+
"learning_rate": 3.5620675193960995e-06,
|
| 2539 |
+
"loss": 0.0313,
|
| 2540 |
+
"step": 38100
|
| 2541 |
+
},
|
| 2542 |
+
{
|
| 2543 |
+
"epoch": 21.11,
|
| 2544 |
+
"learning_rate": 2.7757391486684836e-06,
|
| 2545 |
+
"loss": 0.0326,
|
| 2546 |
+
"step": 38200
|
| 2547 |
+
},
|
| 2548 |
+
{
|
| 2549 |
+
"epoch": 21.17,
|
| 2550 |
+
"learning_rate": 1.9894107779408677e-06,
|
| 2551 |
+
"loss": 0.0325,
|
| 2552 |
+
"step": 38300
|
| 2553 |
+
},
|
| 2554 |
+
{
|
| 2555 |
+
"epoch": 21.23,
|
| 2556 |
+
"learning_rate": 1.2030824072132522e-06,
|
| 2557 |
+
"loss": 0.0314,
|
| 2558 |
+
"step": 38400
|
| 2559 |
+
}
|
| 2560 |
+
],
|
| 2561 |
+
"max_steps": 38538,
|
| 2562 |
+
"num_train_epochs": 23,
|
| 2563 |
+
"total_flos": 3.3292216688770744e+18,
|
| 2564 |
+
"trial_name": null,
|
| 2565 |
+
"trial_params": null
|
| 2566 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:292ba7ff28fca833429ca3c3c57b4679c4daf38549269f8c59c07f9e6fb58efd
|
| 3 |
+
size 3643
|