checkpoint 1200
Browse files- optimizer.pt +2 -2
- pytorch_model.bin +1 -1
- rng_state.pth +1 -1
- scaler.pt +1 -1
- scheduler.pt +1 -1
- trainer_state.json +2284 -11
- training_args.bin +2 -2
optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:149216126f79d89f96f2dd88f83738259f91516192c7af864119c935c4938c86
|
| 3 |
+
size 1980951621
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 990452905
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3260f3d235a97524002927c60208f78db38bb196586f469bb4ec57219d2d76c9
|
| 3 |
size 990452905
|
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 sha256:183919bd52d5297a8eede950b5f2c3f25481765f8a5ff32894b60f70f5b64581
|
| 3 |
size 14575
|
scaler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 557
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:741cefeca9ef427f92406d2d10b81996655e2a9d50eb7aaa9614e6fdd1c9f529
|
| 3 |
size 557
|
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 sha256:9edbd55c449de80bf26fda741f0d114df8fd97ac2c7e81667c83f44ee9234726
|
| 3 |
size 627
|
trainer_state.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch":
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
@@ -5550,18 +5550,2291 @@
|
|
| 5550 |
"step": 850
|
| 5551 |
},
|
| 5552 |
{
|
| 5553 |
-
"epoch": 13.
|
| 5554 |
-
"
|
| 5555 |
-
"
|
| 5556 |
-
"
|
| 5557 |
-
|
| 5558 |
-
|
| 5559 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5560 |
}
|
| 5561 |
],
|
| 5562 |
-
"max_steps":
|
| 5563 |
"num_train_epochs": 25,
|
| 5564 |
-
"total_flos": 1.
|
| 5565 |
"trial_name": null,
|
| 5566 |
"trial_params": null
|
| 5567 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 18.749268292682927,
|
| 5 |
+
"global_step": 1200,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 5550 |
"step": 850
|
| 5551 |
},
|
| 5552 |
{
|
| 5553 |
+
"epoch": 13.3,
|
| 5554 |
+
"learning_rate": 0.000244316687138972,
|
| 5555 |
+
"loss": 1.2264,
|
| 5556 |
+
"step": 851
|
| 5557 |
+
},
|
| 5558 |
+
{
|
| 5559 |
+
"epoch": 13.31,
|
| 5560 |
+
"learning_rate": 0.00024380012386289248,
|
| 5561 |
+
"loss": 1.2144,
|
| 5562 |
+
"step": 852
|
| 5563 |
+
},
|
| 5564 |
+
{
|
| 5565 |
+
"epoch": 13.33,
|
| 5566 |
+
"learning_rate": 0.00024328358707153557,
|
| 5567 |
+
"loss": 1.2739,
|
| 5568 |
+
"step": 853
|
| 5569 |
+
},
|
| 5570 |
+
{
|
| 5571 |
+
"epoch": 13.34,
|
| 5572 |
+
"learning_rate": 0.00024276707897145073,
|
| 5573 |
+
"loss": 1.247,
|
| 5574 |
+
"step": 854
|
| 5575 |
+
},
|
| 5576 |
+
{
|
| 5577 |
+
"epoch": 13.36,
|
| 5578 |
+
"learning_rate": 0.0002422506017690649,
|
| 5579 |
+
"loss": 1.232,
|
| 5580 |
+
"step": 855
|
| 5581 |
+
},
|
| 5582 |
+
{
|
| 5583 |
+
"epoch": 13.37,
|
| 5584 |
+
"learning_rate": 0.00024173415767067295,
|
| 5585 |
+
"loss": 1.3068,
|
| 5586 |
+
"step": 856
|
| 5587 |
+
},
|
| 5588 |
+
{
|
| 5589 |
+
"epoch": 13.39,
|
| 5590 |
+
"learning_rate": 0.0002412177488824286,
|
| 5591 |
+
"loss": 1.2301,
|
| 5592 |
+
"step": 857
|
| 5593 |
+
},
|
| 5594 |
+
{
|
| 5595 |
+
"epoch": 13.41,
|
| 5596 |
+
"learning_rate": 0.00024070137761033445,
|
| 5597 |
+
"loss": 1.2606,
|
| 5598 |
+
"step": 858
|
| 5599 |
+
},
|
| 5600 |
+
{
|
| 5601 |
+
"epoch": 13.42,
|
| 5602 |
+
"learning_rate": 0.00024018504606023293,
|
| 5603 |
+
"loss": 1.2397,
|
| 5604 |
+
"step": 859
|
| 5605 |
+
},
|
| 5606 |
+
{
|
| 5607 |
+
"epoch": 13.44,
|
| 5608 |
+
"learning_rate": 0.0002396687564377967,
|
| 5609 |
+
"loss": 1.3012,
|
| 5610 |
+
"step": 860
|
| 5611 |
+
},
|
| 5612 |
+
{
|
| 5613 |
+
"epoch": 13.45,
|
| 5614 |
+
"learning_rate": 0.0002391525109485196,
|
| 5615 |
+
"loss": 1.2032,
|
| 5616 |
+
"step": 861
|
| 5617 |
+
},
|
| 5618 |
+
{
|
| 5619 |
+
"epoch": 13.47,
|
| 5620 |
+
"learning_rate": 0.00023863631179770666,
|
| 5621 |
+
"loss": 1.3246,
|
| 5622 |
+
"step": 862
|
| 5623 |
+
},
|
| 5624 |
+
{
|
| 5625 |
+
"epoch": 13.48,
|
| 5626 |
+
"learning_rate": 0.00023812016119046503,
|
| 5627 |
+
"loss": 1.1968,
|
| 5628 |
+
"step": 863
|
| 5629 |
+
},
|
| 5630 |
+
{
|
| 5631 |
+
"epoch": 13.5,
|
| 5632 |
+
"learning_rate": 0.0002376040613316944,
|
| 5633 |
+
"loss": 1.2726,
|
| 5634 |
+
"step": 864
|
| 5635 |
+
},
|
| 5636 |
+
{
|
| 5637 |
+
"epoch": 13.52,
|
| 5638 |
+
"learning_rate": 0.00023708801442607803,
|
| 5639 |
+
"loss": 1.2574,
|
| 5640 |
+
"step": 865
|
| 5641 |
+
},
|
| 5642 |
+
{
|
| 5643 |
+
"epoch": 13.53,
|
| 5644 |
+
"learning_rate": 0.00023657202267807258,
|
| 5645 |
+
"loss": 1.2215,
|
| 5646 |
+
"step": 866
|
| 5647 |
+
},
|
| 5648 |
+
{
|
| 5649 |
+
"epoch": 13.55,
|
| 5650 |
+
"learning_rate": 0.00023605608829189921,
|
| 5651 |
+
"loss": 1.2118,
|
| 5652 |
+
"step": 867
|
| 5653 |
+
},
|
| 5654 |
+
{
|
| 5655 |
+
"epoch": 13.56,
|
| 5656 |
+
"learning_rate": 0.00023554021347153403,
|
| 5657 |
+
"loss": 1.2433,
|
| 5658 |
+
"step": 868
|
| 5659 |
+
},
|
| 5660 |
+
{
|
| 5661 |
+
"epoch": 13.58,
|
| 5662 |
+
"learning_rate": 0.0002350244004206988,
|
| 5663 |
+
"loss": 1.2434,
|
| 5664 |
+
"step": 869
|
| 5665 |
+
},
|
| 5666 |
+
{
|
| 5667 |
+
"epoch": 13.59,
|
| 5668 |
+
"learning_rate": 0.00023450865134285137,
|
| 5669 |
+
"loss": 1.2695,
|
| 5670 |
+
"step": 870
|
| 5671 |
+
},
|
| 5672 |
+
{
|
| 5673 |
+
"epoch": 13.61,
|
| 5674 |
+
"learning_rate": 0.00023399296844117624,
|
| 5675 |
+
"loss": 1.2494,
|
| 5676 |
+
"step": 871
|
| 5677 |
+
},
|
| 5678 |
+
{
|
| 5679 |
+
"epoch": 13.62,
|
| 5680 |
+
"learning_rate": 0.00023347735391857517,
|
| 5681 |
+
"loss": 1.2801,
|
| 5682 |
+
"step": 872
|
| 5683 |
+
},
|
| 5684 |
+
{
|
| 5685 |
+
"epoch": 13.64,
|
| 5686 |
+
"learning_rate": 0.00023296180997765798,
|
| 5687 |
+
"loss": 1.3121,
|
| 5688 |
+
"step": 873
|
| 5689 |
+
},
|
| 5690 |
+
{
|
| 5691 |
+
"epoch": 13.66,
|
| 5692 |
+
"learning_rate": 0.00023244633882073295,
|
| 5693 |
+
"loss": 1.3019,
|
| 5694 |
+
"step": 874
|
| 5695 |
+
},
|
| 5696 |
+
{
|
| 5697 |
+
"epoch": 13.67,
|
| 5698 |
+
"learning_rate": 0.00023193094264979737,
|
| 5699 |
+
"loss": 1.2655,
|
| 5700 |
+
"step": 875
|
| 5701 |
+
},
|
| 5702 |
+
{
|
| 5703 |
+
"epoch": 13.67,
|
| 5704 |
+
"eval_gen_len": 805.352,
|
| 5705 |
+
"eval_loss": 1.5386887788772583,
|
| 5706 |
+
"eval_rouge1": 56.6127,
|
| 5707 |
+
"eval_rouge2": 24.6239,
|
| 5708 |
+
"eval_rougeL": 26.4088,
|
| 5709 |
+
"eval_rougeLsum": 53.708,
|
| 5710 |
+
"eval_runtime": 896.6875,
|
| 5711 |
+
"eval_samples_per_second": 0.279,
|
| 5712 |
+
"eval_steps_per_second": 0.279,
|
| 5713 |
+
"step": 875
|
| 5714 |
+
},
|
| 5715 |
+
{
|
| 5716 |
+
"epoch": 13.69,
|
| 5717 |
+
"learning_rate": 0.00023141562366652812,
|
| 5718 |
+
"loss": 1.2454,
|
| 5719 |
+
"step": 876
|
| 5720 |
+
},
|
| 5721 |
+
{
|
| 5722 |
+
"epoch": 13.7,
|
| 5723 |
+
"learning_rate": 0.00023090038407227255,
|
| 5724 |
+
"loss": 1.2877,
|
| 5725 |
+
"step": 877
|
| 5726 |
+
},
|
| 5727 |
+
{
|
| 5728 |
+
"epoch": 13.72,
|
| 5729 |
+
"learning_rate": 0.0002303852260680388,
|
| 5730 |
+
"loss": 1.2915,
|
| 5731 |
+
"step": 878
|
| 5732 |
+
},
|
| 5733 |
+
{
|
| 5734 |
+
"epoch": 13.73,
|
| 5735 |
+
"learning_rate": 0.0002298701518544864,
|
| 5736 |
+
"loss": 1.3163,
|
| 5737 |
+
"step": 879
|
| 5738 |
+
},
|
| 5739 |
+
{
|
| 5740 |
+
"epoch": 13.75,
|
| 5741 |
+
"learning_rate": 0.00022935516363191695,
|
| 5742 |
+
"loss": 1.2792,
|
| 5743 |
+
"step": 880
|
| 5744 |
+
},
|
| 5745 |
+
{
|
| 5746 |
+
"epoch": 13.76,
|
| 5747 |
+
"learning_rate": 0.00022884026360026486,
|
| 5748 |
+
"loss": 1.2367,
|
| 5749 |
+
"step": 881
|
| 5750 |
+
},
|
| 5751 |
+
{
|
| 5752 |
+
"epoch": 13.78,
|
| 5753 |
+
"learning_rate": 0.00022832545395908759,
|
| 5754 |
+
"loss": 1.309,
|
| 5755 |
+
"step": 882
|
| 5756 |
+
},
|
| 5757 |
+
{
|
| 5758 |
+
"epoch": 13.8,
|
| 5759 |
+
"learning_rate": 0.00022781073690755668,
|
| 5760 |
+
"loss": 1.2333,
|
| 5761 |
+
"step": 883
|
| 5762 |
+
},
|
| 5763 |
+
{
|
| 5764 |
+
"epoch": 13.81,
|
| 5765 |
+
"learning_rate": 0.00022729611464444795,
|
| 5766 |
+
"loss": 1.249,
|
| 5767 |
+
"step": 884
|
| 5768 |
+
},
|
| 5769 |
+
{
|
| 5770 |
+
"epoch": 13.83,
|
| 5771 |
+
"learning_rate": 0.0002267815893681325,
|
| 5772 |
+
"loss": 1.2565,
|
| 5773 |
+
"step": 885
|
| 5774 |
+
},
|
| 5775 |
+
{
|
| 5776 |
+
"epoch": 13.84,
|
| 5777 |
+
"learning_rate": 0.0002262671632765669,
|
| 5778 |
+
"loss": 1.2797,
|
| 5779 |
+
"step": 886
|
| 5780 |
+
},
|
| 5781 |
+
{
|
| 5782 |
+
"epoch": 13.86,
|
| 5783 |
+
"learning_rate": 0.00022575283856728423,
|
| 5784 |
+
"loss": 1.2498,
|
| 5785 |
+
"step": 887
|
| 5786 |
+
},
|
| 5787 |
+
{
|
| 5788 |
+
"epoch": 13.87,
|
| 5789 |
+
"learning_rate": 0.00022523861743738434,
|
| 5790 |
+
"loss": 1.202,
|
| 5791 |
+
"step": 888
|
| 5792 |
+
},
|
| 5793 |
+
{
|
| 5794 |
+
"epoch": 13.89,
|
| 5795 |
+
"learning_rate": 0.0002247245020835247,
|
| 5796 |
+
"loss": 1.2963,
|
| 5797 |
+
"step": 889
|
| 5798 |
+
},
|
| 5799 |
+
{
|
| 5800 |
+
"epoch": 13.91,
|
| 5801 |
+
"learning_rate": 0.00022421049470191078,
|
| 5802 |
+
"loss": 1.2997,
|
| 5803 |
+
"step": 890
|
| 5804 |
+
},
|
| 5805 |
+
{
|
| 5806 |
+
"epoch": 13.92,
|
| 5807 |
+
"learning_rate": 0.00022369659748828704,
|
| 5808 |
+
"loss": 1.2842,
|
| 5809 |
+
"step": 891
|
| 5810 |
+
},
|
| 5811 |
+
{
|
| 5812 |
+
"epoch": 13.94,
|
| 5813 |
+
"learning_rate": 0.0002231828126379271,
|
| 5814 |
+
"loss": 1.2524,
|
| 5815 |
+
"step": 892
|
| 5816 |
+
},
|
| 5817 |
+
{
|
| 5818 |
+
"epoch": 13.95,
|
| 5819 |
+
"learning_rate": 0.00022266914234562478,
|
| 5820 |
+
"loss": 1.3074,
|
| 5821 |
+
"step": 893
|
| 5822 |
+
},
|
| 5823 |
+
{
|
| 5824 |
+
"epoch": 13.97,
|
| 5825 |
+
"learning_rate": 0.0002221555888056843,
|
| 5826 |
+
"loss": 1.2566,
|
| 5827 |
+
"step": 894
|
| 5828 |
+
},
|
| 5829 |
+
{
|
| 5830 |
+
"epoch": 13.98,
|
| 5831 |
+
"learning_rate": 0.00022164215421191142,
|
| 5832 |
+
"loss": 1.2684,
|
| 5833 |
+
"step": 895
|
| 5834 |
+
},
|
| 5835 |
+
{
|
| 5836 |
+
"epoch": 14.0,
|
| 5837 |
+
"learning_rate": 0.00022112884075760347,
|
| 5838 |
+
"loss": 1.2607,
|
| 5839 |
+
"step": 896
|
| 5840 |
+
},
|
| 5841 |
+
{
|
| 5842 |
+
"epoch": 14.02,
|
| 5843 |
+
"learning_rate": 0.00022061565063554063,
|
| 5844 |
+
"loss": 1.3246,
|
| 5845 |
+
"step": 897
|
| 5846 |
+
},
|
| 5847 |
+
{
|
| 5848 |
+
"epoch": 14.03,
|
| 5849 |
+
"learning_rate": 0.00022010258603797592,
|
| 5850 |
+
"loss": 1.2884,
|
| 5851 |
+
"step": 898
|
| 5852 |
+
},
|
| 5853 |
+
{
|
| 5854 |
+
"epoch": 14.05,
|
| 5855 |
+
"learning_rate": 0.00021958964915662628,
|
| 5856 |
+
"loss": 1.2892,
|
| 5857 |
+
"step": 899
|
| 5858 |
+
},
|
| 5859 |
+
{
|
| 5860 |
+
"epoch": 14.06,
|
| 5861 |
+
"learning_rate": 0.00021907684218266309,
|
| 5862 |
+
"loss": 1.294,
|
| 5863 |
+
"step": 900
|
| 5864 |
+
},
|
| 5865 |
+
{
|
| 5866 |
+
"epoch": 14.06,
|
| 5867 |
+
"eval_gen_len": 817.576,
|
| 5868 |
+
"eval_loss": 1.5446990728378296,
|
| 5869 |
+
"eval_rouge1": 56.7767,
|
| 5870 |
+
"eval_rouge2": 24.7515,
|
| 5871 |
+
"eval_rougeL": 26.682,
|
| 5872 |
+
"eval_rougeLsum": 53.7647,
|
| 5873 |
+
"eval_runtime": 909.4637,
|
| 5874 |
+
"eval_samples_per_second": 0.275,
|
| 5875 |
+
"eval_steps_per_second": 0.275,
|
| 5876 |
+
"step": 900
|
| 5877 |
+
},
|
| 5878 |
+
{
|
| 5879 |
+
"epoch": 14.08,
|
| 5880 |
+
"learning_rate": 0.00021856416730670284,
|
| 5881 |
+
"loss": 1.233,
|
| 5882 |
+
"step": 901
|
| 5883 |
+
},
|
| 5884 |
+
{
|
| 5885 |
+
"epoch": 14.09,
|
| 5886 |
+
"learning_rate": 0.00021805162671879758,
|
| 5887 |
+
"loss": 1.238,
|
| 5888 |
+
"step": 902
|
| 5889 |
+
},
|
| 5890 |
+
{
|
| 5891 |
+
"epoch": 14.11,
|
| 5892 |
+
"learning_rate": 0.00021753922260842572,
|
| 5893 |
+
"loss": 1.2278,
|
| 5894 |
+
"step": 903
|
| 5895 |
+
},
|
| 5896 |
+
{
|
| 5897 |
+
"epoch": 14.12,
|
| 5898 |
+
"learning_rate": 0.00021702695716448276,
|
| 5899 |
+
"loss": 1.269,
|
| 5900 |
+
"step": 904
|
| 5901 |
+
},
|
| 5902 |
+
{
|
| 5903 |
+
"epoch": 14.14,
|
| 5904 |
+
"learning_rate": 0.0002165148325752719,
|
| 5905 |
+
"loss": 1.279,
|
| 5906 |
+
"step": 905
|
| 5907 |
+
},
|
| 5908 |
+
{
|
| 5909 |
+
"epoch": 14.16,
|
| 5910 |
+
"learning_rate": 0.0002160028510284944,
|
| 5911 |
+
"loss": 1.2223,
|
| 5912 |
+
"step": 906
|
| 5913 |
+
},
|
| 5914 |
+
{
|
| 5915 |
+
"epoch": 14.17,
|
| 5916 |
+
"learning_rate": 0.0002154910147112406,
|
| 5917 |
+
"loss": 1.1541,
|
| 5918 |
+
"step": 907
|
| 5919 |
+
},
|
| 5920 |
+
{
|
| 5921 |
+
"epoch": 14.19,
|
| 5922 |
+
"learning_rate": 0.00021497932580998052,
|
| 5923 |
+
"loss": 1.2241,
|
| 5924 |
+
"step": 908
|
| 5925 |
+
},
|
| 5926 |
+
{
|
| 5927 |
+
"epoch": 14.2,
|
| 5928 |
+
"learning_rate": 0.0002144677865105544,
|
| 5929 |
+
"loss": 1.2448,
|
| 5930 |
+
"step": 909
|
| 5931 |
+
},
|
| 5932 |
+
{
|
| 5933 |
+
"epoch": 14.22,
|
| 5934 |
+
"learning_rate": 0.00021395639899816332,
|
| 5935 |
+
"loss": 1.2976,
|
| 5936 |
+
"step": 910
|
| 5937 |
+
},
|
| 5938 |
+
{
|
| 5939 |
+
"epoch": 14.23,
|
| 5940 |
+
"learning_rate": 0.00021344516545736007,
|
| 5941 |
+
"loss": 1.1451,
|
| 5942 |
+
"step": 911
|
| 5943 |
+
},
|
| 5944 |
+
{
|
| 5945 |
+
"epoch": 14.25,
|
| 5946 |
+
"learning_rate": 0.00021293408807203948,
|
| 5947 |
+
"loss": 1.2736,
|
| 5948 |
+
"step": 912
|
| 5949 |
+
},
|
| 5950 |
+
{
|
| 5951 |
+
"epoch": 14.27,
|
| 5952 |
+
"learning_rate": 0.00021242316902542974,
|
| 5953 |
+
"loss": 1.308,
|
| 5954 |
+
"step": 913
|
| 5955 |
+
},
|
| 5956 |
+
{
|
| 5957 |
+
"epoch": 14.28,
|
| 5958 |
+
"learning_rate": 0.00021191241050008225,
|
| 5959 |
+
"loss": 1.2435,
|
| 5960 |
+
"step": 914
|
| 5961 |
+
},
|
| 5962 |
+
{
|
| 5963 |
+
"epoch": 14.3,
|
| 5964 |
+
"learning_rate": 0.00021140181467786288,
|
| 5965 |
+
"loss": 1.2843,
|
| 5966 |
+
"step": 915
|
| 5967 |
+
},
|
| 5968 |
+
{
|
| 5969 |
+
"epoch": 14.31,
|
| 5970 |
+
"learning_rate": 0.00021089138373994224,
|
| 5971 |
+
"loss": 1.216,
|
| 5972 |
+
"step": 916
|
| 5973 |
+
},
|
| 5974 |
+
{
|
| 5975 |
+
"epoch": 14.33,
|
| 5976 |
+
"learning_rate": 0.00021038111986678703,
|
| 5977 |
+
"loss": 1.1999,
|
| 5978 |
+
"step": 917
|
| 5979 |
+
},
|
| 5980 |
+
{
|
| 5981 |
+
"epoch": 14.34,
|
| 5982 |
+
"learning_rate": 0.00020987102523814988,
|
| 5983 |
+
"loss": 1.244,
|
| 5984 |
+
"step": 918
|
| 5985 |
+
},
|
| 5986 |
+
{
|
| 5987 |
+
"epoch": 14.36,
|
| 5988 |
+
"learning_rate": 0.00020936110203306058,
|
| 5989 |
+
"loss": 1.2254,
|
| 5990 |
+
"step": 919
|
| 5991 |
+
},
|
| 5992 |
+
{
|
| 5993 |
+
"epoch": 14.37,
|
| 5994 |
+
"learning_rate": 0.00020885135242981647,
|
| 5995 |
+
"loss": 1.2589,
|
| 5996 |
+
"step": 920
|
| 5997 |
+
},
|
| 5998 |
+
{
|
| 5999 |
+
"epoch": 14.39,
|
| 6000 |
+
"learning_rate": 0.00020834177860597374,
|
| 6001 |
+
"loss": 1.2369,
|
| 6002 |
+
"step": 921
|
| 6003 |
+
},
|
| 6004 |
+
{
|
| 6005 |
+
"epoch": 14.41,
|
| 6006 |
+
"learning_rate": 0.0002078323827383372,
|
| 6007 |
+
"loss": 1.2228,
|
| 6008 |
+
"step": 922
|
| 6009 |
+
},
|
| 6010 |
+
{
|
| 6011 |
+
"epoch": 14.42,
|
| 6012 |
+
"learning_rate": 0.00020732316700295168,
|
| 6013 |
+
"loss": 1.3307,
|
| 6014 |
+
"step": 923
|
| 6015 |
+
},
|
| 6016 |
+
{
|
| 6017 |
+
"epoch": 14.44,
|
| 6018 |
+
"learning_rate": 0.0002068141335750925,
|
| 6019 |
+
"loss": 1.2378,
|
| 6020 |
+
"step": 924
|
| 6021 |
+
},
|
| 6022 |
+
{
|
| 6023 |
+
"epoch": 14.45,
|
| 6024 |
+
"learning_rate": 0.00020630528462925603,
|
| 6025 |
+
"loss": 1.2525,
|
| 6026 |
+
"step": 925
|
| 6027 |
+
},
|
| 6028 |
+
{
|
| 6029 |
+
"epoch": 14.45,
|
| 6030 |
+
"eval_gen_len": 800.032,
|
| 6031 |
+
"eval_loss": 1.543728232383728,
|
| 6032 |
+
"eval_rouge1": 56.4674,
|
| 6033 |
+
"eval_rouge2": 24.6172,
|
| 6034 |
+
"eval_rougeL": 26.5266,
|
| 6035 |
+
"eval_rougeLsum": 53.5669,
|
| 6036 |
+
"eval_runtime": 891.0212,
|
| 6037 |
+
"eval_samples_per_second": 0.281,
|
| 6038 |
+
"eval_steps_per_second": 0.281,
|
| 6039 |
+
"step": 925
|
| 6040 |
+
},
|
| 6041 |
+
{
|
| 6042 |
+
"epoch": 14.47,
|
| 6043 |
+
"learning_rate": 0.000205796622339151,
|
| 6044 |
+
"loss": 1.2453,
|
| 6045 |
+
"step": 926
|
| 6046 |
+
},
|
| 6047 |
+
{
|
| 6048 |
+
"epoch": 14.48,
|
| 6049 |
+
"learning_rate": 0.00020528814887768837,
|
| 6050 |
+
"loss": 1.2626,
|
| 6051 |
+
"step": 927
|
| 6052 |
+
},
|
| 6053 |
+
{
|
| 6054 |
+
"epoch": 14.5,
|
| 6055 |
+
"learning_rate": 0.0002047798664169726,
|
| 6056 |
+
"loss": 1.252,
|
| 6057 |
+
"step": 928
|
| 6058 |
+
},
|
| 6059 |
+
{
|
| 6060 |
+
"epoch": 14.52,
|
| 6061 |
+
"learning_rate": 0.0002042717771282921,
|
| 6062 |
+
"loss": 1.1964,
|
| 6063 |
+
"step": 929
|
| 6064 |
+
},
|
| 6065 |
+
{
|
| 6066 |
+
"epoch": 14.53,
|
| 6067 |
+
"learning_rate": 0.00020376388318211043,
|
| 6068 |
+
"loss": 1.2156,
|
| 6069 |
+
"step": 930
|
| 6070 |
+
},
|
| 6071 |
+
{
|
| 6072 |
+
"epoch": 14.55,
|
| 6073 |
+
"learning_rate": 0.00020325618674805636,
|
| 6074 |
+
"loss": 1.1582,
|
| 6075 |
+
"step": 931
|
| 6076 |
+
},
|
| 6077 |
+
{
|
| 6078 |
+
"epoch": 14.56,
|
| 6079 |
+
"learning_rate": 0.00020274868999491498,
|
| 6080 |
+
"loss": 1.2684,
|
| 6081 |
+
"step": 932
|
| 6082 |
+
},
|
| 6083 |
+
{
|
| 6084 |
+
"epoch": 14.58,
|
| 6085 |
+
"learning_rate": 0.00020224139509061837,
|
| 6086 |
+
"loss": 1.1987,
|
| 6087 |
+
"step": 933
|
| 6088 |
+
},
|
| 6089 |
+
{
|
| 6090 |
+
"epoch": 14.59,
|
| 6091 |
+
"learning_rate": 0.00020173430420223663,
|
| 6092 |
+
"loss": 1.2351,
|
| 6093 |
+
"step": 934
|
| 6094 |
+
},
|
| 6095 |
+
{
|
| 6096 |
+
"epoch": 14.61,
|
| 6097 |
+
"learning_rate": 0.00020122741949596797,
|
| 6098 |
+
"loss": 1.2513,
|
| 6099 |
+
"step": 935
|
| 6100 |
+
},
|
| 6101 |
+
{
|
| 6102 |
+
"epoch": 14.62,
|
| 6103 |
+
"learning_rate": 0.00020072074313712995,
|
| 6104 |
+
"loss": 1.2079,
|
| 6105 |
+
"step": 936
|
| 6106 |
+
},
|
| 6107 |
+
{
|
| 6108 |
+
"epoch": 14.64,
|
| 6109 |
+
"learning_rate": 0.0002002142772901502,
|
| 6110 |
+
"loss": 1.1933,
|
| 6111 |
+
"step": 937
|
| 6112 |
+
},
|
| 6113 |
+
{
|
| 6114 |
+
"epoch": 14.66,
|
| 6115 |
+
"learning_rate": 0.0001997080241185569,
|
| 6116 |
+
"loss": 1.1643,
|
| 6117 |
+
"step": 938
|
| 6118 |
+
},
|
| 6119 |
+
{
|
| 6120 |
+
"epoch": 14.67,
|
| 6121 |
+
"learning_rate": 0.00019920198578497002,
|
| 6122 |
+
"loss": 1.264,
|
| 6123 |
+
"step": 939
|
| 6124 |
+
},
|
| 6125 |
+
{
|
| 6126 |
+
"epoch": 14.69,
|
| 6127 |
+
"learning_rate": 0.00019869616445109147,
|
| 6128 |
+
"loss": 1.2386,
|
| 6129 |
+
"step": 940
|
| 6130 |
+
},
|
| 6131 |
+
{
|
| 6132 |
+
"epoch": 14.7,
|
| 6133 |
+
"learning_rate": 0.00019819056227769633,
|
| 6134 |
+
"loss": 1.2513,
|
| 6135 |
+
"step": 941
|
| 6136 |
+
},
|
| 6137 |
+
{
|
| 6138 |
+
"epoch": 14.72,
|
| 6139 |
+
"learning_rate": 0.00019768518142462336,
|
| 6140 |
+
"loss": 1.2206,
|
| 6141 |
+
"step": 942
|
| 6142 |
+
},
|
| 6143 |
+
{
|
| 6144 |
+
"epoch": 14.73,
|
| 6145 |
+
"learning_rate": 0.00019718002405076616,
|
| 6146 |
+
"loss": 1.2685,
|
| 6147 |
+
"step": 943
|
| 6148 |
+
},
|
| 6149 |
+
{
|
| 6150 |
+
"epoch": 14.75,
|
| 6151 |
+
"learning_rate": 0.00019667509231406332,
|
| 6152 |
+
"loss": 1.2362,
|
| 6153 |
+
"step": 944
|
| 6154 |
+
},
|
| 6155 |
+
{
|
| 6156 |
+
"epoch": 14.76,
|
| 6157 |
+
"learning_rate": 0.00019617038837148977,
|
| 6158 |
+
"loss": 1.2189,
|
| 6159 |
+
"step": 945
|
| 6160 |
+
},
|
| 6161 |
+
{
|
| 6162 |
+
"epoch": 14.78,
|
| 6163 |
+
"learning_rate": 0.0001956659143790471,
|
| 6164 |
+
"loss": 1.2744,
|
| 6165 |
+
"step": 946
|
| 6166 |
+
},
|
| 6167 |
+
{
|
| 6168 |
+
"epoch": 14.8,
|
| 6169 |
+
"learning_rate": 0.00019516167249175503,
|
| 6170 |
+
"loss": 1.2063,
|
| 6171 |
+
"step": 947
|
| 6172 |
+
},
|
| 6173 |
+
{
|
| 6174 |
+
"epoch": 14.81,
|
| 6175 |
+
"learning_rate": 0.00019465766486364143,
|
| 6176 |
+
"loss": 1.2603,
|
| 6177 |
+
"step": 948
|
| 6178 |
+
},
|
| 6179 |
+
{
|
| 6180 |
+
"epoch": 14.83,
|
| 6181 |
+
"learning_rate": 0.00019415389364773345,
|
| 6182 |
+
"loss": 1.2424,
|
| 6183 |
+
"step": 949
|
| 6184 |
+
},
|
| 6185 |
+
{
|
| 6186 |
+
"epoch": 14.84,
|
| 6187 |
+
"learning_rate": 0.00019365036099604853,
|
| 6188 |
+
"loss": 1.3088,
|
| 6189 |
+
"step": 950
|
| 6190 |
+
},
|
| 6191 |
+
{
|
| 6192 |
+
"epoch": 14.84,
|
| 6193 |
+
"eval_gen_len": 789.032,
|
| 6194 |
+
"eval_loss": 1.5428788661956787,
|
| 6195 |
+
"eval_rouge1": 56.5205,
|
| 6196 |
+
"eval_rouge2": 24.6994,
|
| 6197 |
+
"eval_rougeL": 26.701,
|
| 6198 |
+
"eval_rougeLsum": 53.6155,
|
| 6199 |
+
"eval_runtime": 880.5733,
|
| 6200 |
+
"eval_samples_per_second": 0.284,
|
| 6201 |
+
"eval_steps_per_second": 0.284,
|
| 6202 |
+
"step": 950
|
| 6203 |
+
},
|
| 6204 |
+
{
|
| 6205 |
+
"epoch": 14.86,
|
| 6206 |
+
"learning_rate": 0.00019314706905958467,
|
| 6207 |
+
"loss": 1.3325,
|
| 6208 |
+
"step": 951
|
| 6209 |
+
},
|
| 6210 |
+
{
|
| 6211 |
+
"epoch": 14.87,
|
| 6212 |
+
"learning_rate": 0.00019264401998831212,
|
| 6213 |
+
"loss": 1.2646,
|
| 6214 |
+
"step": 952
|
| 6215 |
+
},
|
| 6216 |
+
{
|
| 6217 |
+
"epoch": 14.89,
|
| 6218 |
+
"learning_rate": 0.0001921412159311632,
|
| 6219 |
+
"loss": 1.2284,
|
| 6220 |
+
"step": 953
|
| 6221 |
+
},
|
| 6222 |
+
{
|
| 6223 |
+
"epoch": 14.91,
|
| 6224 |
+
"learning_rate": 0.00019163865903602372,
|
| 6225 |
+
"loss": 1.2927,
|
| 6226 |
+
"step": 954
|
| 6227 |
+
},
|
| 6228 |
+
{
|
| 6229 |
+
"epoch": 14.92,
|
| 6230 |
+
"learning_rate": 0.00019113635144972355,
|
| 6231 |
+
"loss": 1.2324,
|
| 6232 |
+
"step": 955
|
| 6233 |
+
},
|
| 6234 |
+
{
|
| 6235 |
+
"epoch": 14.94,
|
| 6236 |
+
"learning_rate": 0.00019063429531802786,
|
| 6237 |
+
"loss": 1.2576,
|
| 6238 |
+
"step": 956
|
| 6239 |
+
},
|
| 6240 |
+
{
|
| 6241 |
+
"epoch": 14.95,
|
| 6242 |
+
"learning_rate": 0.00019013249278562738,
|
| 6243 |
+
"loss": 1.2009,
|
| 6244 |
+
"step": 957
|
| 6245 |
+
},
|
| 6246 |
+
{
|
| 6247 |
+
"epoch": 14.97,
|
| 6248 |
+
"learning_rate": 0.00018963094599612956,
|
| 6249 |
+
"loss": 1.2031,
|
| 6250 |
+
"step": 958
|
| 6251 |
+
},
|
| 6252 |
+
{
|
| 6253 |
+
"epoch": 14.98,
|
| 6254 |
+
"learning_rate": 0.00018912965709204932,
|
| 6255 |
+
"loss": 1.2318,
|
| 6256 |
+
"step": 959
|
| 6257 |
+
},
|
| 6258 |
+
{
|
| 6259 |
+
"epoch": 15.0,
|
| 6260 |
+
"learning_rate": 0.00018862862821480023,
|
| 6261 |
+
"loss": 1.2538,
|
| 6262 |
+
"step": 960
|
| 6263 |
+
},
|
| 6264 |
+
{
|
| 6265 |
+
"epoch": 15.02,
|
| 6266 |
+
"learning_rate": 0.00018812786150468468,
|
| 6267 |
+
"loss": 1.2972,
|
| 6268 |
+
"step": 961
|
| 6269 |
+
},
|
| 6270 |
+
{
|
| 6271 |
+
"epoch": 15.03,
|
| 6272 |
+
"learning_rate": 0.00018762735910088542,
|
| 6273 |
+
"loss": 1.2474,
|
| 6274 |
+
"step": 962
|
| 6275 |
+
},
|
| 6276 |
+
{
|
| 6277 |
+
"epoch": 15.05,
|
| 6278 |
+
"learning_rate": 0.00018712712314145587,
|
| 6279 |
+
"loss": 1.2064,
|
| 6280 |
+
"step": 963
|
| 6281 |
+
},
|
| 6282 |
+
{
|
| 6283 |
+
"epoch": 15.06,
|
| 6284 |
+
"learning_rate": 0.00018662715576331148,
|
| 6285 |
+
"loss": 1.2912,
|
| 6286 |
+
"step": 964
|
| 6287 |
+
},
|
| 6288 |
+
{
|
| 6289 |
+
"epoch": 15.08,
|
| 6290 |
+
"learning_rate": 0.00018612745910222037,
|
| 6291 |
+
"loss": 1.186,
|
| 6292 |
+
"step": 965
|
| 6293 |
+
},
|
| 6294 |
+
{
|
| 6295 |
+
"epoch": 15.09,
|
| 6296 |
+
"learning_rate": 0.00018562803529279404,
|
| 6297 |
+
"loss": 1.1969,
|
| 6298 |
+
"step": 966
|
| 6299 |
+
},
|
| 6300 |
+
{
|
| 6301 |
+
"epoch": 15.11,
|
| 6302 |
+
"learning_rate": 0.00018512888646847846,
|
| 6303 |
+
"loss": 1.1615,
|
| 6304 |
+
"step": 967
|
| 6305 |
+
},
|
| 6306 |
+
{
|
| 6307 |
+
"epoch": 15.12,
|
| 6308 |
+
"learning_rate": 0.00018463001476154507,
|
| 6309 |
+
"loss": 1.216,
|
| 6310 |
+
"step": 968
|
| 6311 |
+
},
|
| 6312 |
+
{
|
| 6313 |
+
"epoch": 15.14,
|
| 6314 |
+
"learning_rate": 0.00018413142230308137,
|
| 6315 |
+
"loss": 1.2415,
|
| 6316 |
+
"step": 969
|
| 6317 |
+
},
|
| 6318 |
+
{
|
| 6319 |
+
"epoch": 15.16,
|
| 6320 |
+
"learning_rate": 0.000183633111222982,
|
| 6321 |
+
"loss": 1.2929,
|
| 6322 |
+
"step": 970
|
| 6323 |
+
},
|
| 6324 |
+
{
|
| 6325 |
+
"epoch": 15.17,
|
| 6326 |
+
"learning_rate": 0.00018313508364993947,
|
| 6327 |
+
"loss": 1.2413,
|
| 6328 |
+
"step": 971
|
| 6329 |
+
},
|
| 6330 |
+
{
|
| 6331 |
+
"epoch": 15.19,
|
| 6332 |
+
"learning_rate": 0.0001826373417114355,
|
| 6333 |
+
"loss": 1.2335,
|
| 6334 |
+
"step": 972
|
| 6335 |
+
},
|
| 6336 |
+
{
|
| 6337 |
+
"epoch": 15.2,
|
| 6338 |
+
"learning_rate": 0.00018213988753373146,
|
| 6339 |
+
"loss": 1.1665,
|
| 6340 |
+
"step": 973
|
| 6341 |
+
},
|
| 6342 |
+
{
|
| 6343 |
+
"epoch": 15.22,
|
| 6344 |
+
"learning_rate": 0.0001816427232418594,
|
| 6345 |
+
"loss": 1.2116,
|
| 6346 |
+
"step": 974
|
| 6347 |
+
},
|
| 6348 |
+
{
|
| 6349 |
+
"epoch": 15.23,
|
| 6350 |
+
"learning_rate": 0.00018114585095961306,
|
| 6351 |
+
"loss": 1.2124,
|
| 6352 |
+
"step": 975
|
| 6353 |
+
},
|
| 6354 |
+
{
|
| 6355 |
+
"epoch": 15.23,
|
| 6356 |
+
"eval_gen_len": 790.376,
|
| 6357 |
+
"eval_loss": 1.5507651567459106,
|
| 6358 |
+
"eval_rouge1": 56.7924,
|
| 6359 |
+
"eval_rouge2": 24.6834,
|
| 6360 |
+
"eval_rougeL": 26.3234,
|
| 6361 |
+
"eval_rougeLsum": 53.8648,
|
| 6362 |
+
"eval_runtime": 883.2222,
|
| 6363 |
+
"eval_samples_per_second": 0.283,
|
| 6364 |
+
"eval_steps_per_second": 0.283,
|
| 6365 |
+
"step": 975
|
| 6366 |
+
},
|
| 6367 |
+
{
|
| 6368 |
+
"epoch": 15.25,
|
| 6369 |
+
"learning_rate": 0.00018064927280953891,
|
| 6370 |
+
"loss": 1.2773,
|
| 6371 |
+
"step": 976
|
| 6372 |
+
},
|
| 6373 |
+
{
|
| 6374 |
+
"epoch": 15.27,
|
| 6375 |
+
"learning_rate": 0.00018015299091292668,
|
| 6376 |
+
"loss": 1.2097,
|
| 6377 |
+
"step": 977
|
| 6378 |
+
},
|
| 6379 |
+
{
|
| 6380 |
+
"epoch": 15.28,
|
| 6381 |
+
"learning_rate": 0.0001796570073898009,
|
| 6382 |
+
"loss": 1.215,
|
| 6383 |
+
"step": 978
|
| 6384 |
+
},
|
| 6385 |
+
{
|
| 6386 |
+
"epoch": 15.3,
|
| 6387 |
+
"learning_rate": 0.0001791613243589112,
|
| 6388 |
+
"loss": 1.3315,
|
| 6389 |
+
"step": 979
|
| 6390 |
+
},
|
| 6391 |
+
{
|
| 6392 |
+
"epoch": 15.31,
|
| 6393 |
+
"learning_rate": 0.00017866594393772372,
|
| 6394 |
+
"loss": 1.2595,
|
| 6395 |
+
"step": 980
|
| 6396 |
+
},
|
| 6397 |
+
{
|
| 6398 |
+
"epoch": 15.33,
|
| 6399 |
+
"learning_rate": 0.0001781708682424118,
|
| 6400 |
+
"loss": 1.234,
|
| 6401 |
+
"step": 981
|
| 6402 |
+
},
|
| 6403 |
+
{
|
| 6404 |
+
"epoch": 15.34,
|
| 6405 |
+
"learning_rate": 0.00017767609938784727,
|
| 6406 |
+
"loss": 1.1581,
|
| 6407 |
+
"step": 982
|
| 6408 |
+
},
|
| 6409 |
+
{
|
| 6410 |
+
"epoch": 15.36,
|
| 6411 |
+
"learning_rate": 0.00017718163948759086,
|
| 6412 |
+
"loss": 1.231,
|
| 6413 |
+
"step": 983
|
| 6414 |
+
},
|
| 6415 |
+
{
|
| 6416 |
+
"epoch": 15.37,
|
| 6417 |
+
"learning_rate": 0.00017668749065388384,
|
| 6418 |
+
"loss": 1.2202,
|
| 6419 |
+
"step": 984
|
| 6420 |
+
},
|
| 6421 |
+
{
|
| 6422 |
+
"epoch": 15.39,
|
| 6423 |
+
"learning_rate": 0.0001761936549976384,
|
| 6424 |
+
"loss": 1.2636,
|
| 6425 |
+
"step": 985
|
| 6426 |
+
},
|
| 6427 |
+
{
|
| 6428 |
+
"epoch": 15.41,
|
| 6429 |
+
"learning_rate": 0.00017570013462842905,
|
| 6430 |
+
"loss": 1.2869,
|
| 6431 |
+
"step": 986
|
| 6432 |
+
},
|
| 6433 |
+
{
|
| 6434 |
+
"epoch": 15.42,
|
| 6435 |
+
"learning_rate": 0.00017520693165448337,
|
| 6436 |
+
"loss": 1.2041,
|
| 6437 |
+
"step": 987
|
| 6438 |
+
},
|
| 6439 |
+
{
|
| 6440 |
+
"epoch": 15.44,
|
| 6441 |
+
"learning_rate": 0.00017471404818267318,
|
| 6442 |
+
"loss": 1.2307,
|
| 6443 |
+
"step": 988
|
| 6444 |
+
},
|
| 6445 |
+
{
|
| 6446 |
+
"epoch": 15.45,
|
| 6447 |
+
"learning_rate": 0.00017422148631850525,
|
| 6448 |
+
"loss": 1.2112,
|
| 6449 |
+
"step": 989
|
| 6450 |
+
},
|
| 6451 |
+
{
|
| 6452 |
+
"epoch": 15.47,
|
| 6453 |
+
"learning_rate": 0.0001737292481661128,
|
| 6454 |
+
"loss": 1.2476,
|
| 6455 |
+
"step": 990
|
| 6456 |
+
},
|
| 6457 |
+
{
|
| 6458 |
+
"epoch": 15.48,
|
| 6459 |
+
"learning_rate": 0.00017323733582824596,
|
| 6460 |
+
"loss": 1.3425,
|
| 6461 |
+
"step": 991
|
| 6462 |
+
},
|
| 6463 |
+
{
|
| 6464 |
+
"epoch": 15.5,
|
| 6465 |
+
"learning_rate": 0.00017274575140626317,
|
| 6466 |
+
"loss": 1.3227,
|
| 6467 |
+
"step": 992
|
| 6468 |
+
},
|
| 6469 |
+
{
|
| 6470 |
+
"epoch": 15.52,
|
| 6471 |
+
"learning_rate": 0.000172254497000122,
|
| 6472 |
+
"loss": 1.2707,
|
| 6473 |
+
"step": 993
|
| 6474 |
+
},
|
| 6475 |
+
{
|
| 6476 |
+
"epoch": 15.53,
|
| 6477 |
+
"learning_rate": 0.00017176357470837024,
|
| 6478 |
+
"loss": 1.2576,
|
| 6479 |
+
"step": 994
|
| 6480 |
+
},
|
| 6481 |
+
{
|
| 6482 |
+
"epoch": 15.55,
|
| 6483 |
+
"learning_rate": 0.00017127298662813706,
|
| 6484 |
+
"loss": 1.1688,
|
| 6485 |
+
"step": 995
|
| 6486 |
+
},
|
| 6487 |
+
{
|
| 6488 |
+
"epoch": 15.56,
|
| 6489 |
+
"learning_rate": 0.0001707827348551239,
|
| 6490 |
+
"loss": 1.2615,
|
| 6491 |
+
"step": 996
|
| 6492 |
+
},
|
| 6493 |
+
{
|
| 6494 |
+
"epoch": 15.58,
|
| 6495 |
+
"learning_rate": 0.00017029282148359553,
|
| 6496 |
+
"loss": 1.2438,
|
| 6497 |
+
"step": 997
|
| 6498 |
+
},
|
| 6499 |
+
{
|
| 6500 |
+
"epoch": 15.59,
|
| 6501 |
+
"learning_rate": 0.00016980324860637106,
|
| 6502 |
+
"loss": 1.1954,
|
| 6503 |
+
"step": 998
|
| 6504 |
+
},
|
| 6505 |
+
{
|
| 6506 |
+
"epoch": 15.61,
|
| 6507 |
+
"learning_rate": 0.00016931401831481519,
|
| 6508 |
+
"loss": 1.1599,
|
| 6509 |
+
"step": 999
|
| 6510 |
+
},
|
| 6511 |
+
{
|
| 6512 |
+
"epoch": 15.62,
|
| 6513 |
+
"learning_rate": 0.00016882513269882916,
|
| 6514 |
+
"loss": 1.2046,
|
| 6515 |
+
"step": 1000
|
| 6516 |
+
},
|
| 6517 |
+
{
|
| 6518 |
+
"epoch": 15.62,
|
| 6519 |
+
"eval_gen_len": 822.16,
|
| 6520 |
+
"eval_loss": 1.5468982458114624,
|
| 6521 |
+
"eval_rouge1": 57.0844,
|
| 6522 |
+
"eval_rouge2": 24.9798,
|
| 6523 |
+
"eval_rougeL": 26.8697,
|
| 6524 |
+
"eval_rougeLsum": 54.1281,
|
| 6525 |
+
"eval_runtime": 915.7316,
|
| 6526 |
+
"eval_samples_per_second": 0.273,
|
| 6527 |
+
"eval_steps_per_second": 0.273,
|
| 6528 |
+
"step": 1000
|
| 6529 |
+
},
|
| 6530 |
+
{
|
| 6531 |
+
"epoch": 15.64,
|
| 6532 |
+
"learning_rate": 0.00016833659384684168,
|
| 6533 |
+
"loss": 1.2254,
|
| 6534 |
+
"step": 1001
|
| 6535 |
+
},
|
| 6536 |
+
{
|
| 6537 |
+
"epoch": 15.66,
|
| 6538 |
+
"learning_rate": 0.0001678484038458003,
|
| 6539 |
+
"loss": 1.1956,
|
| 6540 |
+
"step": 1002
|
| 6541 |
+
},
|
| 6542 |
+
{
|
| 6543 |
+
"epoch": 15.67,
|
| 6544 |
+
"learning_rate": 0.0001673605647811623,
|
| 6545 |
+
"loss": 1.2027,
|
| 6546 |
+
"step": 1003
|
| 6547 |
+
},
|
| 6548 |
+
{
|
| 6549 |
+
"epoch": 15.69,
|
| 6550 |
+
"learning_rate": 0.00016687307873688583,
|
| 6551 |
+
"loss": 1.2073,
|
| 6552 |
+
"step": 1004
|
| 6553 |
+
},
|
| 6554 |
+
{
|
| 6555 |
+
"epoch": 15.7,
|
| 6556 |
+
"learning_rate": 0.00016638594779542094,
|
| 6557 |
+
"loss": 1.2126,
|
| 6558 |
+
"step": 1005
|
| 6559 |
+
},
|
| 6560 |
+
{
|
| 6561 |
+
"epoch": 15.72,
|
| 6562 |
+
"learning_rate": 0.00016589917403770076,
|
| 6563 |
+
"loss": 1.1758,
|
| 6564 |
+
"step": 1006
|
| 6565 |
+
},
|
| 6566 |
+
{
|
| 6567 |
+
"epoch": 15.73,
|
| 6568 |
+
"learning_rate": 0.00016541275954313258,
|
| 6569 |
+
"loss": 1.2036,
|
| 6570 |
+
"step": 1007
|
| 6571 |
+
},
|
| 6572 |
+
{
|
| 6573 |
+
"epoch": 15.75,
|
| 6574 |
+
"learning_rate": 0.00016492670638958924,
|
| 6575 |
+
"loss": 1.2371,
|
| 6576 |
+
"step": 1008
|
| 6577 |
+
},
|
| 6578 |
+
{
|
| 6579 |
+
"epoch": 15.76,
|
| 6580 |
+
"learning_rate": 0.00016444101665339967,
|
| 6581 |
+
"loss": 1.2794,
|
| 6582 |
+
"step": 1009
|
| 6583 |
+
},
|
| 6584 |
+
{
|
| 6585 |
+
"epoch": 15.78,
|
| 6586 |
+
"learning_rate": 0.00016395569240934042,
|
| 6587 |
+
"loss": 1.2055,
|
| 6588 |
+
"step": 1010
|
| 6589 |
+
},
|
| 6590 |
+
{
|
| 6591 |
+
"epoch": 15.8,
|
| 6592 |
+
"learning_rate": 0.0001634707357306267,
|
| 6593 |
+
"loss": 1.2402,
|
| 6594 |
+
"step": 1011
|
| 6595 |
+
},
|
| 6596 |
+
{
|
| 6597 |
+
"epoch": 15.81,
|
| 6598 |
+
"learning_rate": 0.00016298614868890388,
|
| 6599 |
+
"loss": 1.2223,
|
| 6600 |
+
"step": 1012
|
| 6601 |
+
},
|
| 6602 |
+
{
|
| 6603 |
+
"epoch": 15.83,
|
| 6604 |
+
"learning_rate": 0.0001625019333542379,
|
| 6605 |
+
"loss": 1.2268,
|
| 6606 |
+
"step": 1013
|
| 6607 |
+
},
|
| 6608 |
+
{
|
| 6609 |
+
"epoch": 15.84,
|
| 6610 |
+
"learning_rate": 0.00016201809179510702,
|
| 6611 |
+
"loss": 1.1779,
|
| 6612 |
+
"step": 1014
|
| 6613 |
+
},
|
| 6614 |
+
{
|
| 6615 |
+
"epoch": 15.86,
|
| 6616 |
+
"learning_rate": 0.0001615346260783926,
|
| 6617 |
+
"loss": 1.263,
|
| 6618 |
+
"step": 1015
|
| 6619 |
+
},
|
| 6620 |
+
{
|
| 6621 |
+
"epoch": 15.87,
|
| 6622 |
+
"learning_rate": 0.00016105153826937086,
|
| 6623 |
+
"loss": 1.2173,
|
| 6624 |
+
"step": 1016
|
| 6625 |
+
},
|
| 6626 |
+
{
|
| 6627 |
+
"epoch": 15.89,
|
| 6628 |
+
"learning_rate": 0.0001605688304317034,
|
| 6629 |
+
"loss": 1.2335,
|
| 6630 |
+
"step": 1017
|
| 6631 |
+
},
|
| 6632 |
+
{
|
| 6633 |
+
"epoch": 15.91,
|
| 6634 |
+
"learning_rate": 0.0001600865046274287,
|
| 6635 |
+
"loss": 1.1882,
|
| 6636 |
+
"step": 1018
|
| 6637 |
+
},
|
| 6638 |
+
{
|
| 6639 |
+
"epoch": 15.92,
|
| 6640 |
+
"learning_rate": 0.0001596045629169532,
|
| 6641 |
+
"loss": 1.2563,
|
| 6642 |
+
"step": 1019
|
| 6643 |
+
},
|
| 6644 |
+
{
|
| 6645 |
+
"epoch": 15.94,
|
| 6646 |
+
"learning_rate": 0.0001591230073590425,
|
| 6647 |
+
"loss": 1.2362,
|
| 6648 |
+
"step": 1020
|
| 6649 |
+
},
|
| 6650 |
+
{
|
| 6651 |
+
"epoch": 15.95,
|
| 6652 |
+
"learning_rate": 0.00015864184001081308,
|
| 6653 |
+
"loss": 1.1643,
|
| 6654 |
+
"step": 1021
|
| 6655 |
+
},
|
| 6656 |
+
{
|
| 6657 |
+
"epoch": 15.97,
|
| 6658 |
+
"learning_rate": 0.00015816106292772253,
|
| 6659 |
+
"loss": 1.2467,
|
| 6660 |
+
"step": 1022
|
| 6661 |
+
},
|
| 6662 |
+
{
|
| 6663 |
+
"epoch": 15.98,
|
| 6664 |
+
"learning_rate": 0.00015768067816356157,
|
| 6665 |
+
"loss": 1.2112,
|
| 6666 |
+
"step": 1023
|
| 6667 |
+
},
|
| 6668 |
+
{
|
| 6669 |
+
"epoch": 16.0,
|
| 6670 |
+
"learning_rate": 0.00015720068777044476,
|
| 6671 |
+
"loss": 1.2221,
|
| 6672 |
+
"step": 1024
|
| 6673 |
+
},
|
| 6674 |
+
{
|
| 6675 |
+
"epoch": 16.02,
|
| 6676 |
+
"learning_rate": 0.00015672109379880248,
|
| 6677 |
+
"loss": 1.3096,
|
| 6678 |
+
"step": 1025
|
| 6679 |
+
},
|
| 6680 |
+
{
|
| 6681 |
+
"epoch": 16.02,
|
| 6682 |
+
"eval_gen_len": 809.748,
|
| 6683 |
+
"eval_loss": 1.5421111583709717,
|
| 6684 |
+
"eval_rouge1": 57.2115,
|
| 6685 |
+
"eval_rouge2": 25.1487,
|
| 6686 |
+
"eval_rougeL": 26.7475,
|
| 6687 |
+
"eval_rougeLsum": 54.3755,
|
| 6688 |
+
"eval_runtime": 902.5744,
|
| 6689 |
+
"eval_samples_per_second": 0.277,
|
| 6690 |
+
"eval_steps_per_second": 0.277,
|
| 6691 |
+
"step": 1025
|
| 6692 |
+
},
|
| 6693 |
+
{
|
| 6694 |
+
"epoch": 16.03,
|
| 6695 |
+
"learning_rate": 0.0001562418982973711,
|
| 6696 |
+
"loss": 1.2714,
|
| 6697 |
+
"step": 1026
|
| 6698 |
+
},
|
| 6699 |
+
{
|
| 6700 |
+
"epoch": 16.05,
|
| 6701 |
+
"learning_rate": 0.0001557631033131852,
|
| 6702 |
+
"loss": 1.2757,
|
| 6703 |
+
"step": 1027
|
| 6704 |
+
},
|
| 6705 |
+
{
|
| 6706 |
+
"epoch": 16.06,
|
| 6707 |
+
"learning_rate": 0.00015528471089156805,
|
| 6708 |
+
"loss": 1.2074,
|
| 6709 |
+
"step": 1028
|
| 6710 |
+
},
|
| 6711 |
+
{
|
| 6712 |
+
"epoch": 16.08,
|
| 6713 |
+
"learning_rate": 0.00015480672307612375,
|
| 6714 |
+
"loss": 1.1999,
|
| 6715 |
+
"step": 1029
|
| 6716 |
+
},
|
| 6717 |
+
{
|
| 6718 |
+
"epoch": 16.09,
|
| 6719 |
+
"learning_rate": 0.00015432914190872756,
|
| 6720 |
+
"loss": 1.2254,
|
| 6721 |
+
"step": 1030
|
| 6722 |
+
},
|
| 6723 |
+
{
|
| 6724 |
+
"epoch": 16.11,
|
| 6725 |
+
"learning_rate": 0.00015385196942951785,
|
| 6726 |
+
"loss": 1.1717,
|
| 6727 |
+
"step": 1031
|
| 6728 |
+
},
|
| 6729 |
+
{
|
| 6730 |
+
"epoch": 16.12,
|
| 6731 |
+
"learning_rate": 0.000153375207676887,
|
| 6732 |
+
"loss": 1.2264,
|
| 6733 |
+
"step": 1032
|
| 6734 |
+
},
|
| 6735 |
+
{
|
| 6736 |
+
"epoch": 16.14,
|
| 6737 |
+
"learning_rate": 0.00015289885868747282,
|
| 6738 |
+
"loss": 1.2429,
|
| 6739 |
+
"step": 1033
|
| 6740 |
+
},
|
| 6741 |
+
{
|
| 6742 |
+
"epoch": 16.16,
|
| 6743 |
+
"learning_rate": 0.00015242292449615025,
|
| 6744 |
+
"loss": 1.2632,
|
| 6745 |
+
"step": 1034
|
| 6746 |
+
},
|
| 6747 |
+
{
|
| 6748 |
+
"epoch": 16.17,
|
| 6749 |
+
"learning_rate": 0.00015194740713602185,
|
| 6750 |
+
"loss": 1.106,
|
| 6751 |
+
"step": 1035
|
| 6752 |
+
},
|
| 6753 |
+
{
|
| 6754 |
+
"epoch": 16.19,
|
| 6755 |
+
"learning_rate": 0.00015147230863840966,
|
| 6756 |
+
"loss": 1.1791,
|
| 6757 |
+
"step": 1036
|
| 6758 |
+
},
|
| 6759 |
+
{
|
| 6760 |
+
"epoch": 16.2,
|
| 6761 |
+
"learning_rate": 0.0001509976310328464,
|
| 6762 |
+
"loss": 1.2164,
|
| 6763 |
+
"step": 1037
|
| 6764 |
+
},
|
| 6765 |
+
{
|
| 6766 |
+
"epoch": 16.22,
|
| 6767 |
+
"learning_rate": 0.0001505233763470672,
|
| 6768 |
+
"loss": 1.1441,
|
| 6769 |
+
"step": 1038
|
| 6770 |
+
},
|
| 6771 |
+
{
|
| 6772 |
+
"epoch": 16.23,
|
| 6773 |
+
"learning_rate": 0.00015004954660699998,
|
| 6774 |
+
"loss": 1.2251,
|
| 6775 |
+
"step": 1039
|
| 6776 |
+
},
|
| 6777 |
+
{
|
| 6778 |
+
"epoch": 16.25,
|
| 6779 |
+
"learning_rate": 0.0001495761438367577,
|
| 6780 |
+
"loss": 1.2343,
|
| 6781 |
+
"step": 1040
|
| 6782 |
+
},
|
| 6783 |
+
{
|
| 6784 |
+
"epoch": 16.27,
|
| 6785 |
+
"learning_rate": 0.00014910317005862922,
|
| 6786 |
+
"loss": 1.1249,
|
| 6787 |
+
"step": 1041
|
| 6788 |
+
},
|
| 6789 |
+
{
|
| 6790 |
+
"epoch": 16.28,
|
| 6791 |
+
"learning_rate": 0.00014863062729307108,
|
| 6792 |
+
"loss": 1.1402,
|
| 6793 |
+
"step": 1042
|
| 6794 |
+
},
|
| 6795 |
+
{
|
| 6796 |
+
"epoch": 16.3,
|
| 6797 |
+
"learning_rate": 0.0001481585175586984,
|
| 6798 |
+
"loss": 1.254,
|
| 6799 |
+
"step": 1043
|
| 6800 |
+
},
|
| 6801 |
+
{
|
| 6802 |
+
"epoch": 16.31,
|
| 6803 |
+
"learning_rate": 0.0001476868428722765,
|
| 6804 |
+
"loss": 1.1773,
|
| 6805 |
+
"step": 1044
|
| 6806 |
+
},
|
| 6807 |
+
{
|
| 6808 |
+
"epoch": 16.33,
|
| 6809 |
+
"learning_rate": 0.00014721560524871236,
|
| 6810 |
+
"loss": 1.2251,
|
| 6811 |
+
"step": 1045
|
| 6812 |
+
},
|
| 6813 |
+
{
|
| 6814 |
+
"epoch": 16.34,
|
| 6815 |
+
"learning_rate": 0.0001467448067010456,
|
| 6816 |
+
"loss": 1.2379,
|
| 6817 |
+
"step": 1046
|
| 6818 |
+
},
|
| 6819 |
+
{
|
| 6820 |
+
"epoch": 16.36,
|
| 6821 |
+
"learning_rate": 0.00014627444924044067,
|
| 6822 |
+
"loss": 1.2196,
|
| 6823 |
+
"step": 1047
|
| 6824 |
+
},
|
| 6825 |
+
{
|
| 6826 |
+
"epoch": 16.37,
|
| 6827 |
+
"learning_rate": 0.00014580453487617745,
|
| 6828 |
+
"loss": 1.2115,
|
| 6829 |
+
"step": 1048
|
| 6830 |
+
},
|
| 6831 |
+
{
|
| 6832 |
+
"epoch": 16.39,
|
| 6833 |
+
"learning_rate": 0.00014533506561564306,
|
| 6834 |
+
"loss": 1.2273,
|
| 6835 |
+
"step": 1049
|
| 6836 |
+
},
|
| 6837 |
+
{
|
| 6838 |
+
"epoch": 16.41,
|
| 6839 |
+
"learning_rate": 0.0001448660434643231,
|
| 6840 |
+
"loss": 1.2388,
|
| 6841 |
+
"step": 1050
|
| 6842 |
+
},
|
| 6843 |
+
{
|
| 6844 |
+
"epoch": 16.41,
|
| 6845 |
+
"eval_gen_len": 796.94,
|
| 6846 |
+
"eval_loss": 1.5498594045639038,
|
| 6847 |
+
"eval_rouge1": 57.0084,
|
| 6848 |
+
"eval_rouge2": 25.0799,
|
| 6849 |
+
"eval_rougeL": 26.7496,
|
| 6850 |
+
"eval_rougeLsum": 54.1173,
|
| 6851 |
+
"eval_runtime": 870.1848,
|
| 6852 |
+
"eval_samples_per_second": 0.287,
|
| 6853 |
+
"eval_steps_per_second": 0.287,
|
| 6854 |
+
"step": 1050
|
| 6855 |
+
},
|
| 6856 |
+
{
|
| 6857 |
+
"epoch": 16.42,
|
| 6858 |
+
"learning_rate": 0.0001443974704257936,
|
| 6859 |
+
"loss": 1.2485,
|
| 6860 |
+
"step": 1051
|
| 6861 |
+
},
|
| 6862 |
+
{
|
| 6863 |
+
"epoch": 16.44,
|
| 6864 |
+
"learning_rate": 0.00014392934850171162,
|
| 6865 |
+
"loss": 1.2033,
|
| 6866 |
+
"step": 1052
|
| 6867 |
+
},
|
| 6868 |
+
{
|
| 6869 |
+
"epoch": 16.45,
|
| 6870 |
+
"learning_rate": 0.00014346167969180734,
|
| 6871 |
+
"loss": 1.2272,
|
| 6872 |
+
"step": 1053
|
| 6873 |
+
},
|
| 6874 |
+
{
|
| 6875 |
+
"epoch": 16.47,
|
| 6876 |
+
"learning_rate": 0.00014299446599387524,
|
| 6877 |
+
"loss": 1.2266,
|
| 6878 |
+
"step": 1054
|
| 6879 |
+
},
|
| 6880 |
+
{
|
| 6881 |
+
"epoch": 16.48,
|
| 6882 |
+
"learning_rate": 0.00014252770940376586,
|
| 6883 |
+
"loss": 1.1934,
|
| 6884 |
+
"step": 1055
|
| 6885 |
+
},
|
| 6886 |
+
{
|
| 6887 |
+
"epoch": 16.5,
|
| 6888 |
+
"learning_rate": 0.0001420614119153768,
|
| 6889 |
+
"loss": 1.19,
|
| 6890 |
+
"step": 1056
|
| 6891 |
+
},
|
| 6892 |
+
{
|
| 6893 |
+
"epoch": 16.52,
|
| 6894 |
+
"learning_rate": 0.00014159557552064463,
|
| 6895 |
+
"loss": 1.1861,
|
| 6896 |
+
"step": 1057
|
| 6897 |
+
},
|
| 6898 |
+
{
|
| 6899 |
+
"epoch": 16.53,
|
| 6900 |
+
"learning_rate": 0.00014113020220953603,
|
| 6901 |
+
"loss": 1.2589,
|
| 6902 |
+
"step": 1058
|
| 6903 |
+
},
|
| 6904 |
+
{
|
| 6905 |
+
"epoch": 16.55,
|
| 6906 |
+
"learning_rate": 0.00014066529397003963,
|
| 6907 |
+
"loss": 1.1822,
|
| 6908 |
+
"step": 1059
|
| 6909 |
+
},
|
| 6910 |
+
{
|
| 6911 |
+
"epoch": 16.56,
|
| 6912 |
+
"learning_rate": 0.00014020085278815744,
|
| 6913 |
+
"loss": 1.2299,
|
| 6914 |
+
"step": 1060
|
| 6915 |
+
},
|
| 6916 |
+
{
|
| 6917 |
+
"epoch": 16.58,
|
| 6918 |
+
"learning_rate": 0.00013973688064789608,
|
| 6919 |
+
"loss": 1.2684,
|
| 6920 |
+
"step": 1061
|
| 6921 |
+
},
|
| 6922 |
+
{
|
| 6923 |
+
"epoch": 16.59,
|
| 6924 |
+
"learning_rate": 0.00013927337953125861,
|
| 6925 |
+
"loss": 1.2632,
|
| 6926 |
+
"step": 1062
|
| 6927 |
+
},
|
| 6928 |
+
{
|
| 6929 |
+
"epoch": 16.61,
|
| 6930 |
+
"learning_rate": 0.00013881035141823588,
|
| 6931 |
+
"loss": 1.2404,
|
| 6932 |
+
"step": 1063
|
| 6933 |
+
},
|
| 6934 |
+
{
|
| 6935 |
+
"epoch": 16.62,
|
| 6936 |
+
"learning_rate": 0.00013834779828679838,
|
| 6937 |
+
"loss": 1.1814,
|
| 6938 |
+
"step": 1064
|
| 6939 |
+
},
|
| 6940 |
+
{
|
| 6941 |
+
"epoch": 16.64,
|
| 6942 |
+
"learning_rate": 0.00013788572211288736,
|
| 6943 |
+
"loss": 1.2235,
|
| 6944 |
+
"step": 1065
|
| 6945 |
+
},
|
| 6946 |
+
{
|
| 6947 |
+
"epoch": 16.66,
|
| 6948 |
+
"learning_rate": 0.0001374241248704066,
|
| 6949 |
+
"loss": 1.2549,
|
| 6950 |
+
"step": 1066
|
| 6951 |
+
},
|
| 6952 |
+
{
|
| 6953 |
+
"epoch": 16.67,
|
| 6954 |
+
"learning_rate": 0.00013696300853121402,
|
| 6955 |
+
"loss": 1.2114,
|
| 6956 |
+
"step": 1067
|
| 6957 |
+
},
|
| 6958 |
+
{
|
| 6959 |
+
"epoch": 16.69,
|
| 6960 |
+
"learning_rate": 0.00013650237506511331,
|
| 6961 |
+
"loss": 1.2368,
|
| 6962 |
+
"step": 1068
|
| 6963 |
+
},
|
| 6964 |
+
{
|
| 6965 |
+
"epoch": 16.7,
|
| 6966 |
+
"learning_rate": 0.00013604222643984532,
|
| 6967 |
+
"loss": 1.254,
|
| 6968 |
+
"step": 1069
|
| 6969 |
+
},
|
| 6970 |
+
{
|
| 6971 |
+
"epoch": 16.72,
|
| 6972 |
+
"learning_rate": 0.00013558256462107966,
|
| 6973 |
+
"loss": 1.2721,
|
| 6974 |
+
"step": 1070
|
| 6975 |
+
},
|
| 6976 |
+
{
|
| 6977 |
+
"epoch": 16.73,
|
| 6978 |
+
"learning_rate": 0.00013512339157240654,
|
| 6979 |
+
"loss": 1.2246,
|
| 6980 |
+
"step": 1071
|
| 6981 |
+
},
|
| 6982 |
+
{
|
| 6983 |
+
"epoch": 16.75,
|
| 6984 |
+
"learning_rate": 0.00013466470925532808,
|
| 6985 |
+
"loss": 1.1829,
|
| 6986 |
+
"step": 1072
|
| 6987 |
+
},
|
| 6988 |
+
{
|
| 6989 |
+
"epoch": 16.76,
|
| 6990 |
+
"learning_rate": 0.00013420651962925035,
|
| 6991 |
+
"loss": 1.2187,
|
| 6992 |
+
"step": 1073
|
| 6993 |
+
},
|
| 6994 |
+
{
|
| 6995 |
+
"epoch": 16.78,
|
| 6996 |
+
"learning_rate": 0.00013374882465147454,
|
| 6997 |
+
"loss": 1.1972,
|
| 6998 |
+
"step": 1074
|
| 6999 |
+
},
|
| 7000 |
+
{
|
| 7001 |
+
"epoch": 16.8,
|
| 7002 |
+
"learning_rate": 0.00013329162627718873,
|
| 7003 |
+
"loss": 1.2536,
|
| 7004 |
+
"step": 1075
|
| 7005 |
+
},
|
| 7006 |
+
{
|
| 7007 |
+
"epoch": 16.8,
|
| 7008 |
+
"eval_gen_len": 810.896,
|
| 7009 |
+
"eval_loss": 1.5420339107513428,
|
| 7010 |
+
"eval_rouge1": 57.5194,
|
| 7011 |
+
"eval_rouge2": 25.2201,
|
| 7012 |
+
"eval_rougeL": 27.0883,
|
| 7013 |
+
"eval_rougeLsum": 54.5868,
|
| 7014 |
+
"eval_runtime": 880.8422,
|
| 7015 |
+
"eval_samples_per_second": 0.284,
|
| 7016 |
+
"eval_steps_per_second": 0.284,
|
| 7017 |
+
"step": 1075
|
| 7018 |
+
},
|
| 7019 |
+
{
|
| 7020 |
+
"epoch": 16.81,
|
| 7021 |
+
"learning_rate": 0.00013283492645945965,
|
| 7022 |
+
"loss": 1.2711,
|
| 7023 |
+
"step": 1076
|
| 7024 |
+
},
|
| 7025 |
+
{
|
| 7026 |
+
"epoch": 16.83,
|
| 7027 |
+
"learning_rate": 0.00013237872714922457,
|
| 7028 |
+
"loss": 1.1944,
|
| 7029 |
+
"step": 1077
|
| 7030 |
+
},
|
| 7031 |
+
{
|
| 7032 |
+
"epoch": 16.84,
|
| 7033 |
+
"learning_rate": 0.0001319230302952823,
|
| 7034 |
+
"loss": 1.216,
|
| 7035 |
+
"step": 1078
|
| 7036 |
+
},
|
| 7037 |
+
{
|
| 7038 |
+
"epoch": 16.86,
|
| 7039 |
+
"learning_rate": 0.0001314678378442854,
|
| 7040 |
+
"loss": 1.1859,
|
| 7041 |
+
"step": 1079
|
| 7042 |
+
},
|
| 7043 |
+
{
|
| 7044 |
+
"epoch": 16.87,
|
| 7045 |
+
"learning_rate": 0.0001310131517407316,
|
| 7046 |
+
"loss": 1.1937,
|
| 7047 |
+
"step": 1080
|
| 7048 |
+
},
|
| 7049 |
+
{
|
| 7050 |
+
"epoch": 16.89,
|
| 7051 |
+
"learning_rate": 0.00013055897392695598,
|
| 7052 |
+
"loss": 1.1958,
|
| 7053 |
+
"step": 1081
|
| 7054 |
+
},
|
| 7055 |
+
{
|
| 7056 |
+
"epoch": 16.91,
|
| 7057 |
+
"learning_rate": 0.0001301053063431219,
|
| 7058 |
+
"loss": 1.1696,
|
| 7059 |
+
"step": 1082
|
| 7060 |
+
},
|
| 7061 |
+
{
|
| 7062 |
+
"epoch": 16.92,
|
| 7063 |
+
"learning_rate": 0.00012965215092721328,
|
| 7064 |
+
"loss": 1.1705,
|
| 7065 |
+
"step": 1083
|
| 7066 |
+
},
|
| 7067 |
+
{
|
| 7068 |
+
"epoch": 16.94,
|
| 7069 |
+
"learning_rate": 0.00012919950961502602,
|
| 7070 |
+
"loss": 1.1804,
|
| 7071 |
+
"step": 1084
|
| 7072 |
+
},
|
| 7073 |
+
{
|
| 7074 |
+
"epoch": 16.95,
|
| 7075 |
+
"learning_rate": 0.0001287473843401601,
|
| 7076 |
+
"loss": 1.2113,
|
| 7077 |
+
"step": 1085
|
| 7078 |
+
},
|
| 7079 |
+
{
|
| 7080 |
+
"epoch": 16.97,
|
| 7081 |
+
"learning_rate": 0.00012829577703401095,
|
| 7082 |
+
"loss": 1.292,
|
| 7083 |
+
"step": 1086
|
| 7084 |
+
},
|
| 7085 |
+
{
|
| 7086 |
+
"epoch": 16.98,
|
| 7087 |
+
"learning_rate": 0.00012784468962576134,
|
| 7088 |
+
"loss": 1.1778,
|
| 7089 |
+
"step": 1087
|
| 7090 |
+
},
|
| 7091 |
+
{
|
| 7092 |
+
"epoch": 17.0,
|
| 7093 |
+
"learning_rate": 0.00012739412404237305,
|
| 7094 |
+
"loss": 1.2141,
|
| 7095 |
+
"step": 1088
|
| 7096 |
+
},
|
| 7097 |
+
{
|
| 7098 |
+
"epoch": 17.02,
|
| 7099 |
+
"learning_rate": 0.0001269440822085788,
|
| 7100 |
+
"loss": 1.3028,
|
| 7101 |
+
"step": 1089
|
| 7102 |
+
},
|
| 7103 |
+
{
|
| 7104 |
+
"epoch": 17.03,
|
| 7105 |
+
"learning_rate": 0.00012649456604687403,
|
| 7106 |
+
"loss": 1.2569,
|
| 7107 |
+
"step": 1090
|
| 7108 |
+
},
|
| 7109 |
+
{
|
| 7110 |
+
"epoch": 17.05,
|
| 7111 |
+
"learning_rate": 0.00012604557747750855,
|
| 7112 |
+
"loss": 1.2092,
|
| 7113 |
+
"step": 1091
|
| 7114 |
+
},
|
| 7115 |
+
{
|
| 7116 |
+
"epoch": 17.06,
|
| 7117 |
+
"learning_rate": 0.00012559711841847828,
|
| 7118 |
+
"loss": 1.1945,
|
| 7119 |
+
"step": 1092
|
| 7120 |
+
},
|
| 7121 |
+
{
|
| 7122 |
+
"epoch": 17.08,
|
| 7123 |
+
"learning_rate": 0.00012514919078551723,
|
| 7124 |
+
"loss": 1.2009,
|
| 7125 |
+
"step": 1093
|
| 7126 |
+
},
|
| 7127 |
+
{
|
| 7128 |
+
"epoch": 17.09,
|
| 7129 |
+
"learning_rate": 0.00012470179649208947,
|
| 7130 |
+
"loss": 1.1729,
|
| 7131 |
+
"step": 1094
|
| 7132 |
+
},
|
| 7133 |
+
{
|
| 7134 |
+
"epoch": 17.11,
|
| 7135 |
+
"learning_rate": 0.0001242549374493804,
|
| 7136 |
+
"loss": 1.2607,
|
| 7137 |
+
"step": 1095
|
| 7138 |
+
},
|
| 7139 |
+
{
|
| 7140 |
+
"epoch": 17.12,
|
| 7141 |
+
"learning_rate": 0.00012380861556628915,
|
| 7142 |
+
"loss": 1.2086,
|
| 7143 |
+
"step": 1096
|
| 7144 |
+
},
|
| 7145 |
+
{
|
| 7146 |
+
"epoch": 17.14,
|
| 7147 |
+
"learning_rate": 0.00012336283274942003,
|
| 7148 |
+
"loss": 1.1583,
|
| 7149 |
+
"step": 1097
|
| 7150 |
+
},
|
| 7151 |
+
{
|
| 7152 |
+
"epoch": 17.16,
|
| 7153 |
+
"learning_rate": 0.00012291759090307487,
|
| 7154 |
+
"loss": 1.1823,
|
| 7155 |
+
"step": 1098
|
| 7156 |
+
},
|
| 7157 |
+
{
|
| 7158 |
+
"epoch": 17.17,
|
| 7159 |
+
"learning_rate": 0.00012247289192924424,
|
| 7160 |
+
"loss": 1.1805,
|
| 7161 |
+
"step": 1099
|
| 7162 |
+
},
|
| 7163 |
+
{
|
| 7164 |
+
"epoch": 17.19,
|
| 7165 |
+
"learning_rate": 0.00012202873772759981,
|
| 7166 |
+
"loss": 1.2539,
|
| 7167 |
+
"step": 1100
|
| 7168 |
+
},
|
| 7169 |
+
{
|
| 7170 |
+
"epoch": 17.19,
|
| 7171 |
+
"eval_gen_len": 809.712,
|
| 7172 |
+
"eval_loss": 1.5523115396499634,
|
| 7173 |
+
"eval_rouge1": 58.3301,
|
| 7174 |
+
"eval_rouge2": 25.7175,
|
| 7175 |
+
"eval_rougeL": 27.2238,
|
| 7176 |
+
"eval_rougeLsum": 55.3734,
|
| 7177 |
+
"eval_runtime": 879.0327,
|
| 7178 |
+
"eval_samples_per_second": 0.284,
|
| 7179 |
+
"eval_steps_per_second": 0.284,
|
| 7180 |
+
"step": 1100
|
| 7181 |
+
},
|
| 7182 |
+
{
|
| 7183 |
+
"epoch": 17.2,
|
| 7184 |
+
"learning_rate": 0.00012158513019548612,
|
| 7185 |
+
"loss": 1.1831,
|
| 7186 |
+
"step": 1101
|
| 7187 |
+
},
|
| 7188 |
+
{
|
| 7189 |
+
"epoch": 17.22,
|
| 7190 |
+
"learning_rate": 0.00012114207122791221,
|
| 7191 |
+
"loss": 1.1519,
|
| 7192 |
+
"step": 1102
|
| 7193 |
+
},
|
| 7194 |
+
{
|
| 7195 |
+
"epoch": 17.23,
|
| 7196 |
+
"learning_rate": 0.00012069956271754417,
|
| 7197 |
+
"loss": 1.1788,
|
| 7198 |
+
"step": 1103
|
| 7199 |
+
},
|
| 7200 |
+
{
|
| 7201 |
+
"epoch": 17.25,
|
| 7202 |
+
"learning_rate": 0.00012025760655469628,
|
| 7203 |
+
"loss": 1.2795,
|
| 7204 |
+
"step": 1104
|
| 7205 |
+
},
|
| 7206 |
+
{
|
| 7207 |
+
"epoch": 17.27,
|
| 7208 |
+
"learning_rate": 0.00011981620462732343,
|
| 7209 |
+
"loss": 1.2211,
|
| 7210 |
+
"step": 1105
|
| 7211 |
+
},
|
| 7212 |
+
{
|
| 7213 |
+
"epoch": 17.28,
|
| 7214 |
+
"learning_rate": 0.00011937535882101281,
|
| 7215 |
+
"loss": 1.2403,
|
| 7216 |
+
"step": 1106
|
| 7217 |
+
},
|
| 7218 |
+
{
|
| 7219 |
+
"epoch": 17.3,
|
| 7220 |
+
"learning_rate": 0.00011893507101897622,
|
| 7221 |
+
"loss": 1.1747,
|
| 7222 |
+
"step": 1107
|
| 7223 |
+
},
|
| 7224 |
+
{
|
| 7225 |
+
"epoch": 17.31,
|
| 7226 |
+
"learning_rate": 0.00011849534310204153,
|
| 7227 |
+
"loss": 1.2143,
|
| 7228 |
+
"step": 1108
|
| 7229 |
+
},
|
| 7230 |
+
{
|
| 7231 |
+
"epoch": 17.33,
|
| 7232 |
+
"learning_rate": 0.00011805617694864493,
|
| 7233 |
+
"loss": 1.2278,
|
| 7234 |
+
"step": 1109
|
| 7235 |
+
},
|
| 7236 |
+
{
|
| 7237 |
+
"epoch": 17.34,
|
| 7238 |
+
"learning_rate": 0.00011761757443482285,
|
| 7239 |
+
"loss": 1.1928,
|
| 7240 |
+
"step": 1110
|
| 7241 |
+
},
|
| 7242 |
+
{
|
| 7243 |
+
"epoch": 17.36,
|
| 7244 |
+
"learning_rate": 0.00011717953743420412,
|
| 7245 |
+
"loss": 1.1998,
|
| 7246 |
+
"step": 1111
|
| 7247 |
+
},
|
| 7248 |
+
{
|
| 7249 |
+
"epoch": 17.37,
|
| 7250 |
+
"learning_rate": 0.00011674206781800162,
|
| 7251 |
+
"loss": 1.2648,
|
| 7252 |
+
"step": 1112
|
| 7253 |
+
},
|
| 7254 |
+
{
|
| 7255 |
+
"epoch": 17.39,
|
| 7256 |
+
"learning_rate": 0.00011630516745500453,
|
| 7257 |
+
"loss": 1.1509,
|
| 7258 |
+
"step": 1113
|
| 7259 |
+
},
|
| 7260 |
+
{
|
| 7261 |
+
"epoch": 17.41,
|
| 7262 |
+
"learning_rate": 0.00011586883821157015,
|
| 7263 |
+
"loss": 1.2491,
|
| 7264 |
+
"step": 1114
|
| 7265 |
+
},
|
| 7266 |
+
{
|
| 7267 |
+
"epoch": 17.42,
|
| 7268 |
+
"learning_rate": 0.00011543308195161642,
|
| 7269 |
+
"loss": 1.2086,
|
| 7270 |
+
"step": 1115
|
| 7271 |
+
},
|
| 7272 |
+
{
|
| 7273 |
+
"epoch": 17.44,
|
| 7274 |
+
"learning_rate": 0.00011499790053661327,
|
| 7275 |
+
"loss": 1.1833,
|
| 7276 |
+
"step": 1116
|
| 7277 |
+
},
|
| 7278 |
+
{
|
| 7279 |
+
"epoch": 17.45,
|
| 7280 |
+
"learning_rate": 0.00011456329582557512,
|
| 7281 |
+
"loss": 1.1709,
|
| 7282 |
+
"step": 1117
|
| 7283 |
+
},
|
| 7284 |
+
{
|
| 7285 |
+
"epoch": 17.47,
|
| 7286 |
+
"learning_rate": 0.00011412926967505269,
|
| 7287 |
+
"loss": 1.1348,
|
| 7288 |
+
"step": 1118
|
| 7289 |
+
},
|
| 7290 |
+
{
|
| 7291 |
+
"epoch": 17.48,
|
| 7292 |
+
"learning_rate": 0.00011369582393912548,
|
| 7293 |
+
"loss": 1.2343,
|
| 7294 |
+
"step": 1119
|
| 7295 |
+
},
|
| 7296 |
+
{
|
| 7297 |
+
"epoch": 17.5,
|
| 7298 |
+
"learning_rate": 0.00011326296046939332,
|
| 7299 |
+
"loss": 1.1591,
|
| 7300 |
+
"step": 1120
|
| 7301 |
+
},
|
| 7302 |
+
{
|
| 7303 |
+
"epoch": 17.52,
|
| 7304 |
+
"learning_rate": 0.0001128306811149688,
|
| 7305 |
+
"loss": 1.2393,
|
| 7306 |
+
"step": 1121
|
| 7307 |
+
},
|
| 7308 |
+
{
|
| 7309 |
+
"epoch": 17.53,
|
| 7310 |
+
"learning_rate": 0.00011239898772246915,
|
| 7311 |
+
"loss": 1.2146,
|
| 7312 |
+
"step": 1122
|
| 7313 |
+
},
|
| 7314 |
+
{
|
| 7315 |
+
"epoch": 17.55,
|
| 7316 |
+
"learning_rate": 0.00011196788213600875,
|
| 7317 |
+
"loss": 1.2021,
|
| 7318 |
+
"step": 1123
|
| 7319 |
+
},
|
| 7320 |
+
{
|
| 7321 |
+
"epoch": 17.56,
|
| 7322 |
+
"learning_rate": 0.00011153736619719076,
|
| 7323 |
+
"loss": 1.1951,
|
| 7324 |
+
"step": 1124
|
| 7325 |
+
},
|
| 7326 |
+
{
|
| 7327 |
+
"epoch": 17.58,
|
| 7328 |
+
"learning_rate": 0.00011110744174509952,
|
| 7329 |
+
"loss": 1.2294,
|
| 7330 |
+
"step": 1125
|
| 7331 |
+
},
|
| 7332 |
+
{
|
| 7333 |
+
"epoch": 17.58,
|
| 7334 |
+
"eval_gen_len": 793.548,
|
| 7335 |
+
"eval_loss": 1.5482228994369507,
|
| 7336 |
+
"eval_rouge1": 57.6922,
|
| 7337 |
+
"eval_rouge2": 25.454,
|
| 7338 |
+
"eval_rougeL": 27.1896,
|
| 7339 |
+
"eval_rougeLsum": 54.7084,
|
| 7340 |
+
"eval_runtime": 864.7323,
|
| 7341 |
+
"eval_samples_per_second": 0.289,
|
| 7342 |
+
"eval_steps_per_second": 0.289,
|
| 7343 |
+
"step": 1125
|
| 7344 |
+
},
|
| 7345 |
+
{
|
| 7346 |
+
"epoch": 17.59,
|
| 7347 |
+
"learning_rate": 0.00011067811061629257,
|
| 7348 |
+
"loss": 1.159,
|
| 7349 |
+
"step": 1126
|
| 7350 |
+
},
|
| 7351 |
+
{
|
| 7352 |
+
"epoch": 17.61,
|
| 7353 |
+
"learning_rate": 0.00011024937464479314,
|
| 7354 |
+
"loss": 1.1791,
|
| 7355 |
+
"step": 1127
|
| 7356 |
+
},
|
| 7357 |
+
{
|
| 7358 |
+
"epoch": 17.62,
|
| 7359 |
+
"learning_rate": 0.00010982123566208185,
|
| 7360 |
+
"loss": 1.1825,
|
| 7361 |
+
"step": 1128
|
| 7362 |
+
},
|
| 7363 |
+
{
|
| 7364 |
+
"epoch": 17.64,
|
| 7365 |
+
"learning_rate": 0.00010939369549708903,
|
| 7366 |
+
"loss": 1.217,
|
| 7367 |
+
"step": 1129
|
| 7368 |
+
},
|
| 7369 |
+
{
|
| 7370 |
+
"epoch": 17.66,
|
| 7371 |
+
"learning_rate": 0.00010896675597618725,
|
| 7372 |
+
"loss": 1.2057,
|
| 7373 |
+
"step": 1130
|
| 7374 |
+
},
|
| 7375 |
+
{
|
| 7376 |
+
"epoch": 17.67,
|
| 7377 |
+
"learning_rate": 0.00010854041892318296,
|
| 7378 |
+
"loss": 1.2306,
|
| 7379 |
+
"step": 1131
|
| 7380 |
+
},
|
| 7381 |
+
{
|
| 7382 |
+
"epoch": 17.69,
|
| 7383 |
+
"learning_rate": 0.00010811468615930911,
|
| 7384 |
+
"loss": 1.1588,
|
| 7385 |
+
"step": 1132
|
| 7386 |
+
},
|
| 7387 |
+
{
|
| 7388 |
+
"epoch": 17.7,
|
| 7389 |
+
"learning_rate": 0.00010768955950321713,
|
| 7390 |
+
"loss": 1.2247,
|
| 7391 |
+
"step": 1133
|
| 7392 |
+
},
|
| 7393 |
+
{
|
| 7394 |
+
"epoch": 17.72,
|
| 7395 |
+
"learning_rate": 0.00010726504077096952,
|
| 7396 |
+
"loss": 1.2917,
|
| 7397 |
+
"step": 1134
|
| 7398 |
+
},
|
| 7399 |
+
{
|
| 7400 |
+
"epoch": 17.73,
|
| 7401 |
+
"learning_rate": 0.0001068411317760316,
|
| 7402 |
+
"loss": 1.2074,
|
| 7403 |
+
"step": 1135
|
| 7404 |
+
},
|
| 7405 |
+
{
|
| 7406 |
+
"epoch": 17.75,
|
| 7407 |
+
"learning_rate": 0.0001064178343292641,
|
| 7408 |
+
"loss": 1.1683,
|
| 7409 |
+
"step": 1136
|
| 7410 |
+
},
|
| 7411 |
+
{
|
| 7412 |
+
"epoch": 17.76,
|
| 7413 |
+
"learning_rate": 0.00010599515023891521,
|
| 7414 |
+
"loss": 1.1775,
|
| 7415 |
+
"step": 1137
|
| 7416 |
+
},
|
| 7417 |
+
{
|
| 7418 |
+
"epoch": 17.78,
|
| 7419 |
+
"learning_rate": 0.00010557308131061325,
|
| 7420 |
+
"loss": 1.1825,
|
| 7421 |
+
"step": 1138
|
| 7422 |
+
},
|
| 7423 |
+
{
|
| 7424 |
+
"epoch": 17.8,
|
| 7425 |
+
"learning_rate": 0.00010515162934735842,
|
| 7426 |
+
"loss": 1.2372,
|
| 7427 |
+
"step": 1139
|
| 7428 |
+
},
|
| 7429 |
+
{
|
| 7430 |
+
"epoch": 17.81,
|
| 7431 |
+
"learning_rate": 0.00010473079614951545,
|
| 7432 |
+
"loss": 1.2204,
|
| 7433 |
+
"step": 1140
|
| 7434 |
+
},
|
| 7435 |
+
{
|
| 7436 |
+
"epoch": 17.83,
|
| 7437 |
+
"learning_rate": 0.00010431058351480574,
|
| 7438 |
+
"loss": 1.2157,
|
| 7439 |
+
"step": 1141
|
| 7440 |
+
},
|
| 7441 |
+
{
|
| 7442 |
+
"epoch": 17.84,
|
| 7443 |
+
"learning_rate": 0.00010389099323830001,
|
| 7444 |
+
"loss": 1.2588,
|
| 7445 |
+
"step": 1142
|
| 7446 |
+
},
|
| 7447 |
+
{
|
| 7448 |
+
"epoch": 17.86,
|
| 7449 |
+
"learning_rate": 0.00010347202711241011,
|
| 7450 |
+
"loss": 1.2763,
|
| 7451 |
+
"step": 1143
|
| 7452 |
+
},
|
| 7453 |
+
{
|
| 7454 |
+
"epoch": 17.87,
|
| 7455 |
+
"learning_rate": 0.00010305368692688174,
|
| 7456 |
+
"loss": 1.201,
|
| 7457 |
+
"step": 1144
|
| 7458 |
+
},
|
| 7459 |
+
{
|
| 7460 |
+
"epoch": 17.89,
|
| 7461 |
+
"learning_rate": 0.00010263597446878661,
|
| 7462 |
+
"loss": 1.1619,
|
| 7463 |
+
"step": 1145
|
| 7464 |
+
},
|
| 7465 |
+
{
|
| 7466 |
+
"epoch": 17.91,
|
| 7467 |
+
"learning_rate": 0.00010221889152251512,
|
| 7468 |
+
"loss": 1.1512,
|
| 7469 |
+
"step": 1146
|
| 7470 |
+
},
|
| 7471 |
+
{
|
| 7472 |
+
"epoch": 17.92,
|
| 7473 |
+
"learning_rate": 0.00010180243986976833,
|
| 7474 |
+
"loss": 1.2321,
|
| 7475 |
+
"step": 1147
|
| 7476 |
+
},
|
| 7477 |
+
{
|
| 7478 |
+
"epoch": 17.94,
|
| 7479 |
+
"learning_rate": 0.00010138662128955054,
|
| 7480 |
+
"loss": 1.2074,
|
| 7481 |
+
"step": 1148
|
| 7482 |
+
},
|
| 7483 |
+
{
|
| 7484 |
+
"epoch": 17.95,
|
| 7485 |
+
"learning_rate": 0.00010097143755816163,
|
| 7486 |
+
"loss": 1.1581,
|
| 7487 |
+
"step": 1149
|
| 7488 |
+
},
|
| 7489 |
+
{
|
| 7490 |
+
"epoch": 17.97,
|
| 7491 |
+
"learning_rate": 0.00010055689044918978,
|
| 7492 |
+
"loss": 1.1568,
|
| 7493 |
+
"step": 1150
|
| 7494 |
+
},
|
| 7495 |
+
{
|
| 7496 |
+
"epoch": 17.97,
|
| 7497 |
+
"eval_gen_len": 788.808,
|
| 7498 |
+
"eval_loss": 1.5476616621017456,
|
| 7499 |
+
"eval_rouge1": 57.3929,
|
| 7500 |
+
"eval_rouge2": 25.4513,
|
| 7501 |
+
"eval_rougeL": 27.1555,
|
| 7502 |
+
"eval_rougeLsum": 54.5364,
|
| 7503 |
+
"eval_runtime": 859.8638,
|
| 7504 |
+
"eval_samples_per_second": 0.291,
|
| 7505 |
+
"eval_steps_per_second": 0.291,
|
| 7506 |
+
"step": 1150
|
| 7507 |
+
},
|
| 7508 |
+
{
|
| 7509 |
+
"epoch": 17.98,
|
| 7510 |
+
"learning_rate": 0.00010014298173350333,
|
| 7511 |
+
"loss": 1.2697,
|
| 7512 |
+
"step": 1151
|
| 7513 |
+
},
|
| 7514 |
+
{
|
| 7515 |
+
"epoch": 18.0,
|
| 7516 |
+
"learning_rate": 9.972971317924374e-05,
|
| 7517 |
+
"loss": 1.1648,
|
| 7518 |
+
"step": 1152
|
| 7519 |
+
},
|
| 7520 |
+
{
|
| 7521 |
+
"epoch": 18.02,
|
| 7522 |
+
"learning_rate": 9.931708655181765e-05,
|
| 7523 |
+
"loss": 1.3094,
|
| 7524 |
+
"step": 1153
|
| 7525 |
+
},
|
| 7526 |
+
{
|
| 7527 |
+
"epoch": 18.03,
|
| 7528 |
+
"learning_rate": 9.890510361388955e-05,
|
| 7529 |
+
"loss": 1.2276,
|
| 7530 |
+
"step": 1154
|
| 7531 |
+
},
|
| 7532 |
+
{
|
| 7533 |
+
"epoch": 18.05,
|
| 7534 |
+
"learning_rate": 9.849376612537442e-05,
|
| 7535 |
+
"loss": 1.165,
|
| 7536 |
+
"step": 1155
|
| 7537 |
+
},
|
| 7538 |
+
{
|
| 7539 |
+
"epoch": 18.06,
|
| 7540 |
+
"learning_rate": 9.808307584342971e-05,
|
| 7541 |
+
"loss": 1.1136,
|
| 7542 |
+
"step": 1156
|
| 7543 |
+
},
|
| 7544 |
+
{
|
| 7545 |
+
"epoch": 18.08,
|
| 7546 |
+
"learning_rate": 9.767303452244824e-05,
|
| 7547 |
+
"loss": 1.1616,
|
| 7548 |
+
"step": 1157
|
| 7549 |
+
},
|
| 7550 |
+
{
|
| 7551 |
+
"epoch": 18.09,
|
| 7552 |
+
"learning_rate": 9.726364391405055e-05,
|
| 7553 |
+
"loss": 1.1733,
|
| 7554 |
+
"step": 1158
|
| 7555 |
+
},
|
| 7556 |
+
{
|
| 7557 |
+
"epoch": 18.11,
|
| 7558 |
+
"learning_rate": 9.685490576707761e-05,
|
| 7559 |
+
"loss": 1.1681,
|
| 7560 |
+
"step": 1159
|
| 7561 |
+
},
|
| 7562 |
+
{
|
| 7563 |
+
"epoch": 18.12,
|
| 7564 |
+
"learning_rate": 9.644682182758304e-05,
|
| 7565 |
+
"loss": 1.2383,
|
| 7566 |
+
"step": 1160
|
| 7567 |
+
},
|
| 7568 |
+
{
|
| 7569 |
+
"epoch": 18.14,
|
| 7570 |
+
"learning_rate": 9.603939383882582e-05,
|
| 7571 |
+
"loss": 1.1642,
|
| 7572 |
+
"step": 1161
|
| 7573 |
+
},
|
| 7574 |
+
{
|
| 7575 |
+
"epoch": 18.16,
|
| 7576 |
+
"learning_rate": 9.563262354126278e-05,
|
| 7577 |
+
"loss": 1.1681,
|
| 7578 |
+
"step": 1162
|
| 7579 |
+
},
|
| 7580 |
+
{
|
| 7581 |
+
"epoch": 18.17,
|
| 7582 |
+
"learning_rate": 9.522651267254148e-05,
|
| 7583 |
+
"loss": 1.243,
|
| 7584 |
+
"step": 1163
|
| 7585 |
+
},
|
| 7586 |
+
{
|
| 7587 |
+
"epoch": 18.19,
|
| 7588 |
+
"learning_rate": 9.482106296749222e-05,
|
| 7589 |
+
"loss": 1.242,
|
| 7590 |
+
"step": 1164
|
| 7591 |
+
},
|
| 7592 |
+
{
|
| 7593 |
+
"epoch": 18.2,
|
| 7594 |
+
"learning_rate": 9.441627615812107e-05,
|
| 7595 |
+
"loss": 1.1943,
|
| 7596 |
+
"step": 1165
|
| 7597 |
+
},
|
| 7598 |
+
{
|
| 7599 |
+
"epoch": 18.22,
|
| 7600 |
+
"learning_rate": 9.401215397360227e-05,
|
| 7601 |
+
"loss": 1.2083,
|
| 7602 |
+
"step": 1166
|
| 7603 |
+
},
|
| 7604 |
+
{
|
| 7605 |
+
"epoch": 18.23,
|
| 7606 |
+
"learning_rate": 9.360869814027092e-05,
|
| 7607 |
+
"loss": 1.1749,
|
| 7608 |
+
"step": 1167
|
| 7609 |
+
},
|
| 7610 |
+
{
|
| 7611 |
+
"epoch": 18.25,
|
| 7612 |
+
"learning_rate": 9.320591038161574e-05,
|
| 7613 |
+
"loss": 1.203,
|
| 7614 |
+
"step": 1168
|
| 7615 |
+
},
|
| 7616 |
+
{
|
| 7617 |
+
"epoch": 18.27,
|
| 7618 |
+
"learning_rate": 9.280379241827137e-05,
|
| 7619 |
+
"loss": 1.1599,
|
| 7620 |
+
"step": 1169
|
| 7621 |
+
},
|
| 7622 |
+
{
|
| 7623 |
+
"epoch": 18.28,
|
| 7624 |
+
"learning_rate": 9.240234596801125e-05,
|
| 7625 |
+
"loss": 1.1176,
|
| 7626 |
+
"step": 1170
|
| 7627 |
+
},
|
| 7628 |
+
{
|
| 7629 |
+
"epoch": 18.3,
|
| 7630 |
+
"learning_rate": 9.20015727457402e-05,
|
| 7631 |
+
"loss": 1.2722,
|
| 7632 |
+
"step": 1171
|
| 7633 |
+
},
|
| 7634 |
+
{
|
| 7635 |
+
"epoch": 18.31,
|
| 7636 |
+
"learning_rate": 9.160147446348737e-05,
|
| 7637 |
+
"loss": 1.2143,
|
| 7638 |
+
"step": 1172
|
| 7639 |
+
},
|
| 7640 |
+
{
|
| 7641 |
+
"epoch": 18.33,
|
| 7642 |
+
"learning_rate": 9.120205283039842e-05,
|
| 7643 |
+
"loss": 1.2345,
|
| 7644 |
+
"step": 1173
|
| 7645 |
+
},
|
| 7646 |
+
{
|
| 7647 |
+
"epoch": 18.34,
|
| 7648 |
+
"learning_rate": 9.080330955272859e-05,
|
| 7649 |
+
"loss": 1.1591,
|
| 7650 |
+
"step": 1174
|
| 7651 |
+
},
|
| 7652 |
+
{
|
| 7653 |
+
"epoch": 18.36,
|
| 7654 |
+
"learning_rate": 9.040524633383523e-05,
|
| 7655 |
+
"loss": 1.2142,
|
| 7656 |
+
"step": 1175
|
| 7657 |
+
},
|
| 7658 |
+
{
|
| 7659 |
+
"epoch": 18.36,
|
| 7660 |
+
"eval_gen_len": 814.364,
|
| 7661 |
+
"eval_loss": 1.5497373342514038,
|
| 7662 |
+
"eval_rouge1": 57.1257,
|
| 7663 |
+
"eval_rouge2": 25.1404,
|
| 7664 |
+
"eval_rougeL": 26.7918,
|
| 7665 |
+
"eval_rougeLsum": 54.2231,
|
| 7666 |
+
"eval_runtime": 887.3864,
|
| 7667 |
+
"eval_samples_per_second": 0.282,
|
| 7668 |
+
"eval_steps_per_second": 0.282,
|
| 7669 |
+
"step": 1175
|
| 7670 |
+
},
|
| 7671 |
+
{
|
| 7672 |
+
"epoch": 18.37,
|
| 7673 |
+
"learning_rate": 9.000786487417084e-05,
|
| 7674 |
+
"loss": 1.1998,
|
| 7675 |
+
"step": 1176
|
| 7676 |
+
},
|
| 7677 |
+
{
|
| 7678 |
+
"epoch": 18.39,
|
| 7679 |
+
"learning_rate": 8.961116687127538e-05,
|
| 7680 |
+
"loss": 1.1858,
|
| 7681 |
+
"step": 1177
|
| 7682 |
+
},
|
| 7683 |
+
{
|
| 7684 |
+
"epoch": 18.41,
|
| 7685 |
+
"learning_rate": 8.921515401976918e-05,
|
| 7686 |
+
"loss": 1.1863,
|
| 7687 |
+
"step": 1178
|
| 7688 |
+
},
|
| 7689 |
+
{
|
| 7690 |
+
"epoch": 18.42,
|
| 7691 |
+
"learning_rate": 8.881982801134574e-05,
|
| 7692 |
+
"loss": 1.207,
|
| 7693 |
+
"step": 1179
|
| 7694 |
+
},
|
| 7695 |
+
{
|
| 7696 |
+
"epoch": 18.44,
|
| 7697 |
+
"learning_rate": 8.842519053476475e-05,
|
| 7698 |
+
"loss": 1.241,
|
| 7699 |
+
"step": 1180
|
| 7700 |
+
},
|
| 7701 |
+
{
|
| 7702 |
+
"epoch": 18.45,
|
| 7703 |
+
"learning_rate": 8.80312432758443e-05,
|
| 7704 |
+
"loss": 1.1988,
|
| 7705 |
+
"step": 1181
|
| 7706 |
+
},
|
| 7707 |
+
{
|
| 7708 |
+
"epoch": 18.47,
|
| 7709 |
+
"learning_rate": 8.763798791745412e-05,
|
| 7710 |
+
"loss": 1.227,
|
| 7711 |
+
"step": 1182
|
| 7712 |
+
},
|
| 7713 |
+
{
|
| 7714 |
+
"epoch": 18.48,
|
| 7715 |
+
"learning_rate": 8.724542613950825e-05,
|
| 7716 |
+
"loss": 1.2041,
|
| 7717 |
+
"step": 1183
|
| 7718 |
+
},
|
| 7719 |
+
{
|
| 7720 |
+
"epoch": 18.5,
|
| 7721 |
+
"learning_rate": 8.685355961895783e-05,
|
| 7722 |
+
"loss": 1.17,
|
| 7723 |
+
"step": 1184
|
| 7724 |
+
},
|
| 7725 |
+
{
|
| 7726 |
+
"epoch": 18.52,
|
| 7727 |
+
"learning_rate": 8.646239002978423e-05,
|
| 7728 |
+
"loss": 1.1184,
|
| 7729 |
+
"step": 1185
|
| 7730 |
+
},
|
| 7731 |
+
{
|
| 7732 |
+
"epoch": 18.53,
|
| 7733 |
+
"learning_rate": 8.607191904299142e-05,
|
| 7734 |
+
"loss": 1.2092,
|
| 7735 |
+
"step": 1186
|
| 7736 |
+
},
|
| 7737 |
+
{
|
| 7738 |
+
"epoch": 18.55,
|
| 7739 |
+
"learning_rate": 8.56821483265991e-05,
|
| 7740 |
+
"loss": 1.1561,
|
| 7741 |
+
"step": 1187
|
| 7742 |
+
},
|
| 7743 |
+
{
|
| 7744 |
+
"epoch": 18.56,
|
| 7745 |
+
"learning_rate": 8.52930795456355e-05,
|
| 7746 |
+
"loss": 1.196,
|
| 7747 |
+
"step": 1188
|
| 7748 |
+
},
|
| 7749 |
+
{
|
| 7750 |
+
"epoch": 18.58,
|
| 7751 |
+
"learning_rate": 8.490471436213056e-05,
|
| 7752 |
+
"loss": 1.2142,
|
| 7753 |
+
"step": 1189
|
| 7754 |
+
},
|
| 7755 |
+
{
|
| 7756 |
+
"epoch": 18.59,
|
| 7757 |
+
"learning_rate": 8.451705443510838e-05,
|
| 7758 |
+
"loss": 1.2558,
|
| 7759 |
+
"step": 1190
|
| 7760 |
+
},
|
| 7761 |
+
{
|
| 7762 |
+
"epoch": 18.61,
|
| 7763 |
+
"learning_rate": 8.41301014205803e-05,
|
| 7764 |
+
"loss": 1.1887,
|
| 7765 |
+
"step": 1191
|
| 7766 |
+
},
|
| 7767 |
+
{
|
| 7768 |
+
"epoch": 18.62,
|
| 7769 |
+
"learning_rate": 8.374385697153791e-05,
|
| 7770 |
+
"loss": 1.2226,
|
| 7771 |
+
"step": 1192
|
| 7772 |
+
},
|
| 7773 |
+
{
|
| 7774 |
+
"epoch": 18.64,
|
| 7775 |
+
"learning_rate": 8.335832273794608e-05,
|
| 7776 |
+
"loss": 1.2365,
|
| 7777 |
+
"step": 1193
|
| 7778 |
+
},
|
| 7779 |
+
{
|
| 7780 |
+
"epoch": 18.66,
|
| 7781 |
+
"learning_rate": 8.297350036673556e-05,
|
| 7782 |
+
"loss": 1.1281,
|
| 7783 |
+
"step": 1194
|
| 7784 |
+
},
|
| 7785 |
+
{
|
| 7786 |
+
"epoch": 18.67,
|
| 7787 |
+
"learning_rate": 8.258939150179628e-05,
|
| 7788 |
+
"loss": 1.2167,
|
| 7789 |
+
"step": 1195
|
| 7790 |
+
},
|
| 7791 |
+
{
|
| 7792 |
+
"epoch": 18.69,
|
| 7793 |
+
"learning_rate": 8.220599778397017e-05,
|
| 7794 |
+
"loss": 1.1748,
|
| 7795 |
+
"step": 1196
|
| 7796 |
+
},
|
| 7797 |
+
{
|
| 7798 |
+
"epoch": 18.7,
|
| 7799 |
+
"learning_rate": 8.182332085104408e-05,
|
| 7800 |
+
"loss": 1.1774,
|
| 7801 |
+
"step": 1197
|
| 7802 |
+
},
|
| 7803 |
+
{
|
| 7804 |
+
"epoch": 18.72,
|
| 7805 |
+
"learning_rate": 8.144136233774315e-05,
|
| 7806 |
+
"loss": 1.2056,
|
| 7807 |
+
"step": 1198
|
| 7808 |
+
},
|
| 7809 |
+
{
|
| 7810 |
+
"epoch": 18.73,
|
| 7811 |
+
"learning_rate": 8.106012387572335e-05,
|
| 7812 |
+
"loss": 1.2122,
|
| 7813 |
+
"step": 1199
|
| 7814 |
+
},
|
| 7815 |
+
{
|
| 7816 |
+
"epoch": 18.75,
|
| 7817 |
+
"learning_rate": 8.067960709356478e-05,
|
| 7818 |
+
"loss": 1.2302,
|
| 7819 |
+
"step": 1200
|
| 7820 |
+
},
|
| 7821 |
+
{
|
| 7822 |
+
"epoch": 18.75,
|
| 7823 |
+
"eval_gen_len": 800.512,
|
| 7824 |
+
"eval_loss": 1.5504323244094849,
|
| 7825 |
+
"eval_rouge1": 57.5314,
|
| 7826 |
+
"eval_rouge2": 25.3115,
|
| 7827 |
+
"eval_rougeL": 27.066,
|
| 7828 |
+
"eval_rougeLsum": 54.5668,
|
| 7829 |
+
"eval_runtime": 872.4747,
|
| 7830 |
+
"eval_samples_per_second": 0.287,
|
| 7831 |
+
"eval_steps_per_second": 0.287,
|
| 7832 |
+
"step": 1200
|
| 7833 |
}
|
| 7834 |
],
|
| 7835 |
+
"max_steps": 1600,
|
| 7836 |
"num_train_epochs": 25,
|
| 7837 |
+
"total_flos": 1.684506994016256e+18,
|
| 7838 |
"trial_name": null,
|
| 7839 |
"trial_params": null
|
| 7840 |
}
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:534557a54ea44920b6ae5900bae5a3085b1082d4f98e6868703c07aee1f352a0
|
| 3 |
+
size 3579
|