Training in progress, step 1107, checkpoint
Browse files
last-checkpoint/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3582214344
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:11b33a906b7a0ee7ff522149f85febe58b0f353b6651b2713c9413108ecdfd0e
|
| 3 |
size 3582214344
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5116251049
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:91c061110c7da0939f19b0bb0c7d1f994922960a408802bc9117f394f7f59b90
|
| 3 |
size 5116251049
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14645
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3d6d8fafcd1ee268414be5acf0366296af5b03d60871978712eac1979cb42d65
|
| 3 |
size 14645
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1465
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3167866cb3f20cacb55eee6970efd9d14816cf9b28eba6c543fb2982e8286d1b
|
| 3 |
size 1465
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -2,9 +2,9 @@
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
-
"epoch": 0.
|
| 6 |
"eval_steps": 369,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
@@ -5190,6 +5190,2597 @@
|
|
| 5190 |
"eval_samples_per_second": 11.931,
|
| 5191 |
"eval_steps_per_second": 2.983,
|
| 5192 |
"step": 738
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5193 |
}
|
| 5194 |
],
|
| 5195 |
"logging_steps": 1,
|
|
@@ -5209,7 +7800,7 @@
|
|
| 5209 |
"attributes": {}
|
| 5210 |
}
|
| 5211 |
},
|
| 5212 |
-
"total_flos":
|
| 5213 |
"train_batch_size": 8,
|
| 5214 |
"trial_name": null,
|
| 5215 |
"trial_params": null
|
|
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.631578947368421,
|
| 6 |
"eval_steps": 369,
|
| 7 |
+
"global_step": 1107,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 5190 |
"eval_samples_per_second": 11.931,
|
| 5191 |
"eval_steps_per_second": 2.983,
|
| 5192 |
"step": 738
|
| 5193 |
+
},
|
| 5194 |
+
{
|
| 5195 |
+
"epoch": 0.421623163600057,
|
| 5196 |
+
"grad_norm": 0.98828125,
|
| 5197 |
+
"learning_rate": 3.05709534368071e-05,
|
| 5198 |
+
"loss": 1.193061351776123,
|
| 5199 |
+
"step": 739
|
| 5200 |
+
},
|
| 5201 |
+
{
|
| 5202 |
+
"epoch": 0.42219369562116676,
|
| 5203 |
+
"grad_norm": 1.03125,
|
| 5204 |
+
"learning_rate": 3.054323725055432e-05,
|
| 5205 |
+
"loss": 1.1793735027313232,
|
| 5206 |
+
"step": 740
|
| 5207 |
+
},
|
| 5208 |
+
{
|
| 5209 |
+
"epoch": 0.42276422764227645,
|
| 5210 |
+
"grad_norm": 0.95703125,
|
| 5211 |
+
"learning_rate": 3.0515521064301554e-05,
|
| 5212 |
+
"loss": 1.1607141494750977,
|
| 5213 |
+
"step": 741
|
| 5214 |
+
},
|
| 5215 |
+
{
|
| 5216 |
+
"epoch": 0.42333475966338613,
|
| 5217 |
+
"grad_norm": 0.99609375,
|
| 5218 |
+
"learning_rate": 3.048780487804878e-05,
|
| 5219 |
+
"loss": 1.1790132522583008,
|
| 5220 |
+
"step": 742
|
| 5221 |
+
},
|
| 5222 |
+
{
|
| 5223 |
+
"epoch": 0.4239052916844958,
|
| 5224 |
+
"grad_norm": 0.98046875,
|
| 5225 |
+
"learning_rate": 3.0460088691796013e-05,
|
| 5226 |
+
"loss": 1.155259132385254,
|
| 5227 |
+
"step": 743
|
| 5228 |
+
},
|
| 5229 |
+
{
|
| 5230 |
+
"epoch": 0.4244758237056055,
|
| 5231 |
+
"grad_norm": 0.9609375,
|
| 5232 |
+
"learning_rate": 3.043237250554324e-05,
|
| 5233 |
+
"loss": 1.1134623289108276,
|
| 5234 |
+
"step": 744
|
| 5235 |
+
},
|
| 5236 |
+
{
|
| 5237 |
+
"epoch": 0.4250463557267152,
|
| 5238 |
+
"grad_norm": 0.93359375,
|
| 5239 |
+
"learning_rate": 3.0404656319290465e-05,
|
| 5240 |
+
"loss": 1.198337435722351,
|
| 5241 |
+
"step": 745
|
| 5242 |
+
},
|
| 5243 |
+
{
|
| 5244 |
+
"epoch": 0.42561688774782486,
|
| 5245 |
+
"grad_norm": 0.9921875,
|
| 5246 |
+
"learning_rate": 3.0376940133037695e-05,
|
| 5247 |
+
"loss": 1.1744345426559448,
|
| 5248 |
+
"step": 746
|
| 5249 |
+
},
|
| 5250 |
+
{
|
| 5251 |
+
"epoch": 0.42618741976893454,
|
| 5252 |
+
"grad_norm": 0.984375,
|
| 5253 |
+
"learning_rate": 3.034922394678492e-05,
|
| 5254 |
+
"loss": 1.1646068096160889,
|
| 5255 |
+
"step": 747
|
| 5256 |
+
},
|
| 5257 |
+
{
|
| 5258 |
+
"epoch": 0.4267579517900442,
|
| 5259 |
+
"grad_norm": 0.984375,
|
| 5260 |
+
"learning_rate": 3.0321507760532154e-05,
|
| 5261 |
+
"loss": 1.1827648878097534,
|
| 5262 |
+
"step": 748
|
| 5263 |
+
},
|
| 5264 |
+
{
|
| 5265 |
+
"epoch": 0.4273284838111539,
|
| 5266 |
+
"grad_norm": 0.953125,
|
| 5267 |
+
"learning_rate": 3.029379157427938e-05,
|
| 5268 |
+
"loss": 1.1942888498306274,
|
| 5269 |
+
"step": 749
|
| 5270 |
+
},
|
| 5271 |
+
{
|
| 5272 |
+
"epoch": 0.4278990158322636,
|
| 5273 |
+
"grad_norm": 0.9765625,
|
| 5274 |
+
"learning_rate": 3.026607538802661e-05,
|
| 5275 |
+
"loss": 1.1896655559539795,
|
| 5276 |
+
"step": 750
|
| 5277 |
+
},
|
| 5278 |
+
{
|
| 5279 |
+
"epoch": 0.42846954785337327,
|
| 5280 |
+
"grad_norm": 0.98046875,
|
| 5281 |
+
"learning_rate": 3.0238359201773835e-05,
|
| 5282 |
+
"loss": 1.197471022605896,
|
| 5283 |
+
"step": 751
|
| 5284 |
+
},
|
| 5285 |
+
{
|
| 5286 |
+
"epoch": 0.42904007987448295,
|
| 5287 |
+
"grad_norm": 1.0078125,
|
| 5288 |
+
"learning_rate": 3.021064301552107e-05,
|
| 5289 |
+
"loss": 1.1281297206878662,
|
| 5290 |
+
"step": 752
|
| 5291 |
+
},
|
| 5292 |
+
{
|
| 5293 |
+
"epoch": 0.42961061189559263,
|
| 5294 |
+
"grad_norm": 0.99609375,
|
| 5295 |
+
"learning_rate": 3.0182926829268294e-05,
|
| 5296 |
+
"loss": 1.1960434913635254,
|
| 5297 |
+
"step": 753
|
| 5298 |
+
},
|
| 5299 |
+
{
|
| 5300 |
+
"epoch": 0.4301811439167023,
|
| 5301 |
+
"grad_norm": 0.95703125,
|
| 5302 |
+
"learning_rate": 3.0155210643015524e-05,
|
| 5303 |
+
"loss": 1.1772822141647339,
|
| 5304 |
+
"step": 754
|
| 5305 |
+
},
|
| 5306 |
+
{
|
| 5307 |
+
"epoch": 0.430751675937812,
|
| 5308 |
+
"grad_norm": 0.98046875,
|
| 5309 |
+
"learning_rate": 3.012749445676275e-05,
|
| 5310 |
+
"loss": 1.2077326774597168,
|
| 5311 |
+
"step": 755
|
| 5312 |
+
},
|
| 5313 |
+
{
|
| 5314 |
+
"epoch": 0.4313222079589217,
|
| 5315 |
+
"grad_norm": 1.0,
|
| 5316 |
+
"learning_rate": 3.0099778270509983e-05,
|
| 5317 |
+
"loss": 1.216168999671936,
|
| 5318 |
+
"step": 756
|
| 5319 |
+
},
|
| 5320 |
+
{
|
| 5321 |
+
"epoch": 0.43189273998003136,
|
| 5322 |
+
"grad_norm": 0.97265625,
|
| 5323 |
+
"learning_rate": 3.007206208425721e-05,
|
| 5324 |
+
"loss": 1.1528898477554321,
|
| 5325 |
+
"step": 757
|
| 5326 |
+
},
|
| 5327 |
+
{
|
| 5328 |
+
"epoch": 0.43246327200114104,
|
| 5329 |
+
"grad_norm": 1.0,
|
| 5330 |
+
"learning_rate": 3.0044345898004435e-05,
|
| 5331 |
+
"loss": 1.1724753379821777,
|
| 5332 |
+
"step": 758
|
| 5333 |
+
},
|
| 5334 |
+
{
|
| 5335 |
+
"epoch": 0.4330338040222507,
|
| 5336 |
+
"grad_norm": 0.96484375,
|
| 5337 |
+
"learning_rate": 3.0016629711751664e-05,
|
| 5338 |
+
"loss": 1.1700730323791504,
|
| 5339 |
+
"step": 759
|
| 5340 |
+
},
|
| 5341 |
+
{
|
| 5342 |
+
"epoch": 0.43360433604336046,
|
| 5343 |
+
"grad_norm": 0.9609375,
|
| 5344 |
+
"learning_rate": 2.998891352549889e-05,
|
| 5345 |
+
"loss": 1.1328129768371582,
|
| 5346 |
+
"step": 760
|
| 5347 |
+
},
|
| 5348 |
+
{
|
| 5349 |
+
"epoch": 0.43417486806447014,
|
| 5350 |
+
"grad_norm": 0.9765625,
|
| 5351 |
+
"learning_rate": 2.9961197339246123e-05,
|
| 5352 |
+
"loss": 1.191325306892395,
|
| 5353 |
+
"step": 761
|
| 5354 |
+
},
|
| 5355 |
+
{
|
| 5356 |
+
"epoch": 0.4347454000855798,
|
| 5357 |
+
"grad_norm": 0.97265625,
|
| 5358 |
+
"learning_rate": 2.993348115299335e-05,
|
| 5359 |
+
"loss": 1.160369873046875,
|
| 5360 |
+
"step": 762
|
| 5361 |
+
},
|
| 5362 |
+
{
|
| 5363 |
+
"epoch": 0.4353159321066895,
|
| 5364 |
+
"grad_norm": 0.96484375,
|
| 5365 |
+
"learning_rate": 2.990576496674058e-05,
|
| 5366 |
+
"loss": 1.196010947227478,
|
| 5367 |
+
"step": 763
|
| 5368 |
+
},
|
| 5369 |
+
{
|
| 5370 |
+
"epoch": 0.4358864641277992,
|
| 5371 |
+
"grad_norm": 0.96875,
|
| 5372 |
+
"learning_rate": 2.9878048780487805e-05,
|
| 5373 |
+
"loss": 1.1497125625610352,
|
| 5374 |
+
"step": 764
|
| 5375 |
+
},
|
| 5376 |
+
{
|
| 5377 |
+
"epoch": 0.43645699614890887,
|
| 5378 |
+
"grad_norm": 1.0078125,
|
| 5379 |
+
"learning_rate": 2.9850332594235038e-05,
|
| 5380 |
+
"loss": 1.152623176574707,
|
| 5381 |
+
"step": 765
|
| 5382 |
+
},
|
| 5383 |
+
{
|
| 5384 |
+
"epoch": 0.43702752817001855,
|
| 5385 |
+
"grad_norm": 1.015625,
|
| 5386 |
+
"learning_rate": 2.9822616407982264e-05,
|
| 5387 |
+
"loss": 1.1713566780090332,
|
| 5388 |
+
"step": 766
|
| 5389 |
+
},
|
| 5390 |
+
{
|
| 5391 |
+
"epoch": 0.43759806019112824,
|
| 5392 |
+
"grad_norm": 1.1640625,
|
| 5393 |
+
"learning_rate": 2.9794900221729493e-05,
|
| 5394 |
+
"loss": 1.263333797454834,
|
| 5395 |
+
"step": 767
|
| 5396 |
+
},
|
| 5397 |
+
{
|
| 5398 |
+
"epoch": 0.4381685922122379,
|
| 5399 |
+
"grad_norm": 0.96875,
|
| 5400 |
+
"learning_rate": 2.976718403547672e-05,
|
| 5401 |
+
"loss": 1.144421935081482,
|
| 5402 |
+
"step": 768
|
| 5403 |
+
},
|
| 5404 |
+
{
|
| 5405 |
+
"epoch": 0.4387391242333476,
|
| 5406 |
+
"grad_norm": 0.953125,
|
| 5407 |
+
"learning_rate": 2.9739467849223952e-05,
|
| 5408 |
+
"loss": 1.2290055751800537,
|
| 5409 |
+
"step": 769
|
| 5410 |
+
},
|
| 5411 |
+
{
|
| 5412 |
+
"epoch": 0.4393096562544573,
|
| 5413 |
+
"grad_norm": 0.9921875,
|
| 5414 |
+
"learning_rate": 2.971175166297118e-05,
|
| 5415 |
+
"loss": 1.1050488948822021,
|
| 5416 |
+
"step": 770
|
| 5417 |
+
},
|
| 5418 |
+
{
|
| 5419 |
+
"epoch": 0.43988018827556696,
|
| 5420 |
+
"grad_norm": 0.9765625,
|
| 5421 |
+
"learning_rate": 2.96840354767184e-05,
|
| 5422 |
+
"loss": 1.2218358516693115,
|
| 5423 |
+
"step": 771
|
| 5424 |
+
},
|
| 5425 |
+
{
|
| 5426 |
+
"epoch": 0.44045072029667665,
|
| 5427 |
+
"grad_norm": 0.94921875,
|
| 5428 |
+
"learning_rate": 2.9656319290465634e-05,
|
| 5429 |
+
"loss": 1.1308021545410156,
|
| 5430 |
+
"step": 772
|
| 5431 |
+
},
|
| 5432 |
+
{
|
| 5433 |
+
"epoch": 0.4410212523177863,
|
| 5434 |
+
"grad_norm": 0.99609375,
|
| 5435 |
+
"learning_rate": 2.962860310421286e-05,
|
| 5436 |
+
"loss": 1.2299238443374634,
|
| 5437 |
+
"step": 773
|
| 5438 |
+
},
|
| 5439 |
+
{
|
| 5440 |
+
"epoch": 0.441591784338896,
|
| 5441 |
+
"grad_norm": 0.98046875,
|
| 5442 |
+
"learning_rate": 2.960088691796009e-05,
|
| 5443 |
+
"loss": 1.1389673948287964,
|
| 5444 |
+
"step": 774
|
| 5445 |
+
},
|
| 5446 |
+
{
|
| 5447 |
+
"epoch": 0.4421623163600057,
|
| 5448 |
+
"grad_norm": 1.0078125,
|
| 5449 |
+
"learning_rate": 2.9573170731707316e-05,
|
| 5450 |
+
"loss": 1.2660845518112183,
|
| 5451 |
+
"step": 775
|
| 5452 |
+
},
|
| 5453 |
+
{
|
| 5454 |
+
"epoch": 0.4427328483811154,
|
| 5455 |
+
"grad_norm": 0.96484375,
|
| 5456 |
+
"learning_rate": 2.954545454545455e-05,
|
| 5457 |
+
"loss": 1.099113941192627,
|
| 5458 |
+
"step": 776
|
| 5459 |
+
},
|
| 5460 |
+
{
|
| 5461 |
+
"epoch": 0.44330338040222506,
|
| 5462 |
+
"grad_norm": 0.953125,
|
| 5463 |
+
"learning_rate": 2.9517738359201774e-05,
|
| 5464 |
+
"loss": 1.2134381532669067,
|
| 5465 |
+
"step": 777
|
| 5466 |
+
},
|
| 5467 |
+
{
|
| 5468 |
+
"epoch": 0.44387391242333474,
|
| 5469 |
+
"grad_norm": 0.96875,
|
| 5470 |
+
"learning_rate": 2.9490022172949004e-05,
|
| 5471 |
+
"loss": 1.1754953861236572,
|
| 5472 |
+
"step": 778
|
| 5473 |
+
},
|
| 5474 |
+
{
|
| 5475 |
+
"epoch": 0.4444444444444444,
|
| 5476 |
+
"grad_norm": 0.98046875,
|
| 5477 |
+
"learning_rate": 2.946230598669623e-05,
|
| 5478 |
+
"loss": 1.1886742115020752,
|
| 5479 |
+
"step": 779
|
| 5480 |
+
},
|
| 5481 |
+
{
|
| 5482 |
+
"epoch": 0.4450149764655541,
|
| 5483 |
+
"grad_norm": 0.953125,
|
| 5484 |
+
"learning_rate": 2.9434589800443463e-05,
|
| 5485 |
+
"loss": 1.192276954650879,
|
| 5486 |
+
"step": 780
|
| 5487 |
+
},
|
| 5488 |
+
{
|
| 5489 |
+
"epoch": 0.44558550848666384,
|
| 5490 |
+
"grad_norm": 1.0078125,
|
| 5491 |
+
"learning_rate": 2.940687361419069e-05,
|
| 5492 |
+
"loss": 1.2006890773773193,
|
| 5493 |
+
"step": 781
|
| 5494 |
+
},
|
| 5495 |
+
{
|
| 5496 |
+
"epoch": 0.4461560405077735,
|
| 5497 |
+
"grad_norm": 0.98828125,
|
| 5498 |
+
"learning_rate": 2.9379157427937915e-05,
|
| 5499 |
+
"loss": 1.1819924116134644,
|
| 5500 |
+
"step": 782
|
| 5501 |
+
},
|
| 5502 |
+
{
|
| 5503 |
+
"epoch": 0.4467265725288832,
|
| 5504 |
+
"grad_norm": 0.9453125,
|
| 5505 |
+
"learning_rate": 2.9351441241685145e-05,
|
| 5506 |
+
"loss": 1.1743961572647095,
|
| 5507 |
+
"step": 783
|
| 5508 |
+
},
|
| 5509 |
+
{
|
| 5510 |
+
"epoch": 0.4472971045499929,
|
| 5511 |
+
"grad_norm": 0.98046875,
|
| 5512 |
+
"learning_rate": 2.932372505543237e-05,
|
| 5513 |
+
"loss": 1.2021007537841797,
|
| 5514 |
+
"step": 784
|
| 5515 |
+
},
|
| 5516 |
+
{
|
| 5517 |
+
"epoch": 0.44786763657110257,
|
| 5518 |
+
"grad_norm": 1.0,
|
| 5519 |
+
"learning_rate": 2.9296008869179603e-05,
|
| 5520 |
+
"loss": 1.2032489776611328,
|
| 5521 |
+
"step": 785
|
| 5522 |
+
},
|
| 5523 |
+
{
|
| 5524 |
+
"epoch": 0.44843816859221225,
|
| 5525 |
+
"grad_norm": 1.015625,
|
| 5526 |
+
"learning_rate": 2.926829268292683e-05,
|
| 5527 |
+
"loss": 1.1912821531295776,
|
| 5528 |
+
"step": 786
|
| 5529 |
+
},
|
| 5530 |
+
{
|
| 5531 |
+
"epoch": 0.44900870061332193,
|
| 5532 |
+
"grad_norm": 0.9609375,
|
| 5533 |
+
"learning_rate": 2.924057649667406e-05,
|
| 5534 |
+
"loss": 1.184190034866333,
|
| 5535 |
+
"step": 787
|
| 5536 |
+
},
|
| 5537 |
+
{
|
| 5538 |
+
"epoch": 0.4495792326344316,
|
| 5539 |
+
"grad_norm": 1.015625,
|
| 5540 |
+
"learning_rate": 2.9212860310421285e-05,
|
| 5541 |
+
"loss": 1.272563099861145,
|
| 5542 |
+
"step": 788
|
| 5543 |
+
},
|
| 5544 |
+
{
|
| 5545 |
+
"epoch": 0.4501497646555413,
|
| 5546 |
+
"grad_norm": 0.98046875,
|
| 5547 |
+
"learning_rate": 2.9185144124168518e-05,
|
| 5548 |
+
"loss": 1.2212070226669312,
|
| 5549 |
+
"step": 789
|
| 5550 |
+
},
|
| 5551 |
+
{
|
| 5552 |
+
"epoch": 0.450720296676651,
|
| 5553 |
+
"grad_norm": 1.03125,
|
| 5554 |
+
"learning_rate": 2.9157427937915744e-05,
|
| 5555 |
+
"loss": 1.1937004327774048,
|
| 5556 |
+
"step": 790
|
| 5557 |
+
},
|
| 5558 |
+
{
|
| 5559 |
+
"epoch": 0.45129082869776066,
|
| 5560 |
+
"grad_norm": 1.0390625,
|
| 5561 |
+
"learning_rate": 2.9129711751662973e-05,
|
| 5562 |
+
"loss": 1.1712844371795654,
|
| 5563 |
+
"step": 791
|
| 5564 |
+
},
|
| 5565 |
+
{
|
| 5566 |
+
"epoch": 0.45186136071887034,
|
| 5567 |
+
"grad_norm": 0.9609375,
|
| 5568 |
+
"learning_rate": 2.91019955654102e-05,
|
| 5569 |
+
"loss": 1.1701891422271729,
|
| 5570 |
+
"step": 792
|
| 5571 |
+
},
|
| 5572 |
+
{
|
| 5573 |
+
"epoch": 0.45243189273998,
|
| 5574 |
+
"grad_norm": 1.015625,
|
| 5575 |
+
"learning_rate": 2.9074279379157432e-05,
|
| 5576 |
+
"loss": 1.2575602531433105,
|
| 5577 |
+
"step": 793
|
| 5578 |
+
},
|
| 5579 |
+
{
|
| 5580 |
+
"epoch": 0.4530024247610897,
|
| 5581 |
+
"grad_norm": 0.9765625,
|
| 5582 |
+
"learning_rate": 2.904656319290466e-05,
|
| 5583 |
+
"loss": 1.1968649625778198,
|
| 5584 |
+
"step": 794
|
| 5585 |
+
},
|
| 5586 |
+
{
|
| 5587 |
+
"epoch": 0.4535729567821994,
|
| 5588 |
+
"grad_norm": 0.97265625,
|
| 5589 |
+
"learning_rate": 2.9018847006651885e-05,
|
| 5590 |
+
"loss": 1.205810546875,
|
| 5591 |
+
"step": 795
|
| 5592 |
+
},
|
| 5593 |
+
{
|
| 5594 |
+
"epoch": 0.45414348880330907,
|
| 5595 |
+
"grad_norm": 0.96875,
|
| 5596 |
+
"learning_rate": 2.8991130820399114e-05,
|
| 5597 |
+
"loss": 1.1697238683700562,
|
| 5598 |
+
"step": 796
|
| 5599 |
+
},
|
| 5600 |
+
{
|
| 5601 |
+
"epoch": 0.45471402082441875,
|
| 5602 |
+
"grad_norm": 1.0703125,
|
| 5603 |
+
"learning_rate": 2.896341463414634e-05,
|
| 5604 |
+
"loss": 1.27318274974823,
|
| 5605 |
+
"step": 797
|
| 5606 |
+
},
|
| 5607 |
+
{
|
| 5608 |
+
"epoch": 0.45528455284552843,
|
| 5609 |
+
"grad_norm": 1.015625,
|
| 5610 |
+
"learning_rate": 2.8935698447893573e-05,
|
| 5611 |
+
"loss": 1.2104084491729736,
|
| 5612 |
+
"step": 798
|
| 5613 |
+
},
|
| 5614 |
+
{
|
| 5615 |
+
"epoch": 0.4558550848666381,
|
| 5616 |
+
"grad_norm": 1.0234375,
|
| 5617 |
+
"learning_rate": 2.89079822616408e-05,
|
| 5618 |
+
"loss": 1.2579401731491089,
|
| 5619 |
+
"step": 799
|
| 5620 |
+
},
|
| 5621 |
+
{
|
| 5622 |
+
"epoch": 0.4564256168877478,
|
| 5623 |
+
"grad_norm": 0.97265625,
|
| 5624 |
+
"learning_rate": 2.888026607538803e-05,
|
| 5625 |
+
"loss": 1.1750009059906006,
|
| 5626 |
+
"step": 800
|
| 5627 |
+
},
|
| 5628 |
+
{
|
| 5629 |
+
"epoch": 0.45699614890885754,
|
| 5630 |
+
"grad_norm": 1.03125,
|
| 5631 |
+
"learning_rate": 2.8852549889135255e-05,
|
| 5632 |
+
"loss": 1.1911466121673584,
|
| 5633 |
+
"step": 801
|
| 5634 |
+
},
|
| 5635 |
+
{
|
| 5636 |
+
"epoch": 0.4575666809299672,
|
| 5637 |
+
"grad_norm": 0.96875,
|
| 5638 |
+
"learning_rate": 2.8824833702882487e-05,
|
| 5639 |
+
"loss": 1.0935354232788086,
|
| 5640 |
+
"step": 802
|
| 5641 |
+
},
|
| 5642 |
+
{
|
| 5643 |
+
"epoch": 0.4581372129510769,
|
| 5644 |
+
"grad_norm": 0.9453125,
|
| 5645 |
+
"learning_rate": 2.8797117516629713e-05,
|
| 5646 |
+
"loss": 1.1621028184890747,
|
| 5647 |
+
"step": 803
|
| 5648 |
+
},
|
| 5649 |
+
{
|
| 5650 |
+
"epoch": 0.4587077449721866,
|
| 5651 |
+
"grad_norm": 0.98828125,
|
| 5652 |
+
"learning_rate": 2.8769401330376943e-05,
|
| 5653 |
+
"loss": 1.1952382326126099,
|
| 5654 |
+
"step": 804
|
| 5655 |
+
},
|
| 5656 |
+
{
|
| 5657 |
+
"epoch": 0.45927827699329626,
|
| 5658 |
+
"grad_norm": 0.984375,
|
| 5659 |
+
"learning_rate": 2.874168514412417e-05,
|
| 5660 |
+
"loss": 1.2074031829833984,
|
| 5661 |
+
"step": 805
|
| 5662 |
+
},
|
| 5663 |
+
{
|
| 5664 |
+
"epoch": 0.45984880901440595,
|
| 5665 |
+
"grad_norm": 0.94921875,
|
| 5666 |
+
"learning_rate": 2.8713968957871395e-05,
|
| 5667 |
+
"loss": 1.191246509552002,
|
| 5668 |
+
"step": 806
|
| 5669 |
+
},
|
| 5670 |
+
{
|
| 5671 |
+
"epoch": 0.46041934103551563,
|
| 5672 |
+
"grad_norm": 0.921875,
|
| 5673 |
+
"learning_rate": 2.8686252771618628e-05,
|
| 5674 |
+
"loss": 1.2298707962036133,
|
| 5675 |
+
"step": 807
|
| 5676 |
+
},
|
| 5677 |
+
{
|
| 5678 |
+
"epoch": 0.4609898730566253,
|
| 5679 |
+
"grad_norm": 1.015625,
|
| 5680 |
+
"learning_rate": 2.8658536585365854e-05,
|
| 5681 |
+
"loss": 1.2514528036117554,
|
| 5682 |
+
"step": 808
|
| 5683 |
+
},
|
| 5684 |
+
{
|
| 5685 |
+
"epoch": 0.461560405077735,
|
| 5686 |
+
"grad_norm": 0.96484375,
|
| 5687 |
+
"learning_rate": 2.8630820399113084e-05,
|
| 5688 |
+
"loss": 1.2710151672363281,
|
| 5689 |
+
"step": 809
|
| 5690 |
+
},
|
| 5691 |
+
{
|
| 5692 |
+
"epoch": 0.4621309370988447,
|
| 5693 |
+
"grad_norm": 0.93359375,
|
| 5694 |
+
"learning_rate": 2.860310421286031e-05,
|
| 5695 |
+
"loss": 1.1337497234344482,
|
| 5696 |
+
"step": 810
|
| 5697 |
+
},
|
| 5698 |
+
{
|
| 5699 |
+
"epoch": 0.46270146911995436,
|
| 5700 |
+
"grad_norm": 0.96875,
|
| 5701 |
+
"learning_rate": 2.8575388026607542e-05,
|
| 5702 |
+
"loss": 1.1267883777618408,
|
| 5703 |
+
"step": 811
|
| 5704 |
+
},
|
| 5705 |
+
{
|
| 5706 |
+
"epoch": 0.46327200114106404,
|
| 5707 |
+
"grad_norm": 0.9609375,
|
| 5708 |
+
"learning_rate": 2.854767184035477e-05,
|
| 5709 |
+
"loss": 1.1755304336547852,
|
| 5710 |
+
"step": 812
|
| 5711 |
+
},
|
| 5712 |
+
{
|
| 5713 |
+
"epoch": 0.4638425331621737,
|
| 5714 |
+
"grad_norm": 0.96875,
|
| 5715 |
+
"learning_rate": 2.8519955654101998e-05,
|
| 5716 |
+
"loss": 1.1366599798202515,
|
| 5717 |
+
"step": 813
|
| 5718 |
+
},
|
| 5719 |
+
{
|
| 5720 |
+
"epoch": 0.4644130651832834,
|
| 5721 |
+
"grad_norm": 1.0234375,
|
| 5722 |
+
"learning_rate": 2.8492239467849224e-05,
|
| 5723 |
+
"loss": 1.2038339376449585,
|
| 5724 |
+
"step": 814
|
| 5725 |
+
},
|
| 5726 |
+
{
|
| 5727 |
+
"epoch": 0.4649835972043931,
|
| 5728 |
+
"grad_norm": 0.96875,
|
| 5729 |
+
"learning_rate": 2.8464523281596457e-05,
|
| 5730 |
+
"loss": 1.2154085636138916,
|
| 5731 |
+
"step": 815
|
| 5732 |
+
},
|
| 5733 |
+
{
|
| 5734 |
+
"epoch": 0.46555412922550277,
|
| 5735 |
+
"grad_norm": 1.0078125,
|
| 5736 |
+
"learning_rate": 2.8436807095343683e-05,
|
| 5737 |
+
"loss": 1.1818276643753052,
|
| 5738 |
+
"step": 816
|
| 5739 |
+
},
|
| 5740 |
+
{
|
| 5741 |
+
"epoch": 0.46612466124661245,
|
| 5742 |
+
"grad_norm": 1.0078125,
|
| 5743 |
+
"learning_rate": 2.8409090909090912e-05,
|
| 5744 |
+
"loss": 1.2436468601226807,
|
| 5745 |
+
"step": 817
|
| 5746 |
+
},
|
| 5747 |
+
{
|
| 5748 |
+
"epoch": 0.46669519326772213,
|
| 5749 |
+
"grad_norm": 0.953125,
|
| 5750 |
+
"learning_rate": 2.838137472283814e-05,
|
| 5751 |
+
"loss": 1.1363047361373901,
|
| 5752 |
+
"step": 818
|
| 5753 |
+
},
|
| 5754 |
+
{
|
| 5755 |
+
"epoch": 0.4672657252888318,
|
| 5756 |
+
"grad_norm": 0.984375,
|
| 5757 |
+
"learning_rate": 2.8353658536585365e-05,
|
| 5758 |
+
"loss": 1.1960558891296387,
|
| 5759 |
+
"step": 819
|
| 5760 |
+
},
|
| 5761 |
+
{
|
| 5762 |
+
"epoch": 0.4678362573099415,
|
| 5763 |
+
"grad_norm": 0.96875,
|
| 5764 |
+
"learning_rate": 2.8325942350332597e-05,
|
| 5765 |
+
"loss": 1.171709418296814,
|
| 5766 |
+
"step": 820
|
| 5767 |
+
},
|
| 5768 |
+
{
|
| 5769 |
+
"epoch": 0.46840678933105123,
|
| 5770 |
+
"grad_norm": 0.953125,
|
| 5771 |
+
"learning_rate": 2.8298226164079824e-05,
|
| 5772 |
+
"loss": 1.1537501811981201,
|
| 5773 |
+
"step": 821
|
| 5774 |
+
},
|
| 5775 |
+
{
|
| 5776 |
+
"epoch": 0.4689773213521609,
|
| 5777 |
+
"grad_norm": 0.9609375,
|
| 5778 |
+
"learning_rate": 2.8270509977827053e-05,
|
| 5779 |
+
"loss": 1.1839423179626465,
|
| 5780 |
+
"step": 822
|
| 5781 |
+
},
|
| 5782 |
+
{
|
| 5783 |
+
"epoch": 0.4695478533732706,
|
| 5784 |
+
"grad_norm": 0.98046875,
|
| 5785 |
+
"learning_rate": 2.824279379157428e-05,
|
| 5786 |
+
"loss": 1.1610156297683716,
|
| 5787 |
+
"step": 823
|
| 5788 |
+
},
|
| 5789 |
+
{
|
| 5790 |
+
"epoch": 0.4701183853943803,
|
| 5791 |
+
"grad_norm": 0.984375,
|
| 5792 |
+
"learning_rate": 2.8215077605321512e-05,
|
| 5793 |
+
"loss": 1.1708459854125977,
|
| 5794 |
+
"step": 824
|
| 5795 |
+
},
|
| 5796 |
+
{
|
| 5797 |
+
"epoch": 0.47068891741548996,
|
| 5798 |
+
"grad_norm": 1.015625,
|
| 5799 |
+
"learning_rate": 2.8187361419068735e-05,
|
| 5800 |
+
"loss": 1.251354455947876,
|
| 5801 |
+
"step": 825
|
| 5802 |
+
},
|
| 5803 |
+
{
|
| 5804 |
+
"epoch": 0.47125944943659964,
|
| 5805 |
+
"grad_norm": 0.984375,
|
| 5806 |
+
"learning_rate": 2.8159645232815967e-05,
|
| 5807 |
+
"loss": 1.2049927711486816,
|
| 5808 |
+
"step": 826
|
| 5809 |
+
},
|
| 5810 |
+
{
|
| 5811 |
+
"epoch": 0.4718299814577093,
|
| 5812 |
+
"grad_norm": 0.99609375,
|
| 5813 |
+
"learning_rate": 2.8131929046563194e-05,
|
| 5814 |
+
"loss": 1.230988621711731,
|
| 5815 |
+
"step": 827
|
| 5816 |
+
},
|
| 5817 |
+
{
|
| 5818 |
+
"epoch": 0.472400513478819,
|
| 5819 |
+
"grad_norm": 0.96484375,
|
| 5820 |
+
"learning_rate": 2.8104212860310426e-05,
|
| 5821 |
+
"loss": 1.1739616394042969,
|
| 5822 |
+
"step": 828
|
| 5823 |
+
},
|
| 5824 |
+
{
|
| 5825 |
+
"epoch": 0.4729710454999287,
|
| 5826 |
+
"grad_norm": 0.99609375,
|
| 5827 |
+
"learning_rate": 2.807649667405765e-05,
|
| 5828 |
+
"loss": 1.1999741792678833,
|
| 5829 |
+
"step": 829
|
| 5830 |
+
},
|
| 5831 |
+
{
|
| 5832 |
+
"epoch": 0.47354157752103837,
|
| 5833 |
+
"grad_norm": 1.0,
|
| 5834 |
+
"learning_rate": 2.8048780487804882e-05,
|
| 5835 |
+
"loss": 1.2062275409698486,
|
| 5836 |
+
"step": 830
|
| 5837 |
+
},
|
| 5838 |
+
{
|
| 5839 |
+
"epoch": 0.47411210954214805,
|
| 5840 |
+
"grad_norm": 1.0078125,
|
| 5841 |
+
"learning_rate": 2.8021064301552108e-05,
|
| 5842 |
+
"loss": 1.1344287395477295,
|
| 5843 |
+
"step": 831
|
| 5844 |
+
},
|
| 5845 |
+
{
|
| 5846 |
+
"epoch": 0.47468264156325773,
|
| 5847 |
+
"grad_norm": 0.96484375,
|
| 5848 |
+
"learning_rate": 2.7993348115299334e-05,
|
| 5849 |
+
"loss": 1.2056477069854736,
|
| 5850 |
+
"step": 832
|
| 5851 |
+
},
|
| 5852 |
+
{
|
| 5853 |
+
"epoch": 0.4752531735843674,
|
| 5854 |
+
"grad_norm": 1.015625,
|
| 5855 |
+
"learning_rate": 2.7965631929046564e-05,
|
| 5856 |
+
"loss": 1.1727713346481323,
|
| 5857 |
+
"step": 833
|
| 5858 |
+
},
|
| 5859 |
+
{
|
| 5860 |
+
"epoch": 0.4758237056054771,
|
| 5861 |
+
"grad_norm": 0.99609375,
|
| 5862 |
+
"learning_rate": 2.793791574279379e-05,
|
| 5863 |
+
"loss": 1.2081948518753052,
|
| 5864 |
+
"step": 834
|
| 5865 |
+
},
|
| 5866 |
+
{
|
| 5867 |
+
"epoch": 0.4763942376265868,
|
| 5868 |
+
"grad_norm": 0.98046875,
|
| 5869 |
+
"learning_rate": 2.7910199556541023e-05,
|
| 5870 |
+
"loss": 1.255791187286377,
|
| 5871 |
+
"step": 835
|
| 5872 |
+
},
|
| 5873 |
+
{
|
| 5874 |
+
"epoch": 0.47696476964769646,
|
| 5875 |
+
"grad_norm": 0.9921875,
|
| 5876 |
+
"learning_rate": 2.788248337028825e-05,
|
| 5877 |
+
"loss": 1.1889286041259766,
|
| 5878 |
+
"step": 836
|
| 5879 |
+
},
|
| 5880 |
+
{
|
| 5881 |
+
"epoch": 0.47753530166880614,
|
| 5882 |
+
"grad_norm": 0.9921875,
|
| 5883 |
+
"learning_rate": 2.7854767184035478e-05,
|
| 5884 |
+
"loss": 1.241337776184082,
|
| 5885 |
+
"step": 837
|
| 5886 |
+
},
|
| 5887 |
+
{
|
| 5888 |
+
"epoch": 0.4781058336899158,
|
| 5889 |
+
"grad_norm": 0.98828125,
|
| 5890 |
+
"learning_rate": 2.7827050997782704e-05,
|
| 5891 |
+
"loss": 1.2144089937210083,
|
| 5892 |
+
"step": 838
|
| 5893 |
+
},
|
| 5894 |
+
{
|
| 5895 |
+
"epoch": 0.4786763657110255,
|
| 5896 |
+
"grad_norm": 0.95703125,
|
| 5897 |
+
"learning_rate": 2.7799334811529937e-05,
|
| 5898 |
+
"loss": 1.1527715921401978,
|
| 5899 |
+
"step": 839
|
| 5900 |
+
},
|
| 5901 |
+
{
|
| 5902 |
+
"epoch": 0.4792468977321352,
|
| 5903 |
+
"grad_norm": 0.96875,
|
| 5904 |
+
"learning_rate": 2.7771618625277163e-05,
|
| 5905 |
+
"loss": 1.181959629058838,
|
| 5906 |
+
"step": 840
|
| 5907 |
+
},
|
| 5908 |
+
{
|
| 5909 |
+
"epoch": 0.47981742975324493,
|
| 5910 |
+
"grad_norm": 0.9921875,
|
| 5911 |
+
"learning_rate": 2.7743902439024393e-05,
|
| 5912 |
+
"loss": 1.1999069452285767,
|
| 5913 |
+
"step": 841
|
| 5914 |
+
},
|
| 5915 |
+
{
|
| 5916 |
+
"epoch": 0.4803879617743546,
|
| 5917 |
+
"grad_norm": 0.984375,
|
| 5918 |
+
"learning_rate": 2.771618625277162e-05,
|
| 5919 |
+
"loss": 1.2098867893218994,
|
| 5920 |
+
"step": 842
|
| 5921 |
+
},
|
| 5922 |
+
{
|
| 5923 |
+
"epoch": 0.4809584937954643,
|
| 5924 |
+
"grad_norm": 0.9765625,
|
| 5925 |
+
"learning_rate": 2.7688470066518845e-05,
|
| 5926 |
+
"loss": 1.1860891580581665,
|
| 5927 |
+
"step": 843
|
| 5928 |
+
},
|
| 5929 |
+
{
|
| 5930 |
+
"epoch": 0.481529025816574,
|
| 5931 |
+
"grad_norm": 0.9609375,
|
| 5932 |
+
"learning_rate": 2.7660753880266078e-05,
|
| 5933 |
+
"loss": 1.1108654737472534,
|
| 5934 |
+
"step": 844
|
| 5935 |
+
},
|
| 5936 |
+
{
|
| 5937 |
+
"epoch": 0.48209955783768366,
|
| 5938 |
+
"grad_norm": 0.953125,
|
| 5939 |
+
"learning_rate": 2.7633037694013304e-05,
|
| 5940 |
+
"loss": 1.2157371044158936,
|
| 5941 |
+
"step": 845
|
| 5942 |
+
},
|
| 5943 |
+
{
|
| 5944 |
+
"epoch": 0.48267008985879334,
|
| 5945 |
+
"grad_norm": 0.96875,
|
| 5946 |
+
"learning_rate": 2.7605321507760533e-05,
|
| 5947 |
+
"loss": 1.2216970920562744,
|
| 5948 |
+
"step": 846
|
| 5949 |
+
},
|
| 5950 |
+
{
|
| 5951 |
+
"epoch": 0.483240621879903,
|
| 5952 |
+
"grad_norm": 0.98046875,
|
| 5953 |
+
"learning_rate": 2.757760532150776e-05,
|
| 5954 |
+
"loss": 1.1434253454208374,
|
| 5955 |
+
"step": 847
|
| 5956 |
+
},
|
| 5957 |
+
{
|
| 5958 |
+
"epoch": 0.4838111539010127,
|
| 5959 |
+
"grad_norm": 0.99609375,
|
| 5960 |
+
"learning_rate": 2.7549889135254992e-05,
|
| 5961 |
+
"loss": 1.1241540908813477,
|
| 5962 |
+
"step": 848
|
| 5963 |
+
},
|
| 5964 |
+
{
|
| 5965 |
+
"epoch": 0.4843816859221224,
|
| 5966 |
+
"grad_norm": 0.9921875,
|
| 5967 |
+
"learning_rate": 2.7522172949002218e-05,
|
| 5968 |
+
"loss": 1.186653971672058,
|
| 5969 |
+
"step": 849
|
| 5970 |
+
},
|
| 5971 |
+
{
|
| 5972 |
+
"epoch": 0.48495221794323207,
|
| 5973 |
+
"grad_norm": 1.015625,
|
| 5974 |
+
"learning_rate": 2.7494456762749448e-05,
|
| 5975 |
+
"loss": 1.2525804042816162,
|
| 5976 |
+
"step": 850
|
| 5977 |
+
},
|
| 5978 |
+
{
|
| 5979 |
+
"epoch": 0.48552274996434175,
|
| 5980 |
+
"grad_norm": 0.98046875,
|
| 5981 |
+
"learning_rate": 2.7466740576496674e-05,
|
| 5982 |
+
"loss": 1.1987820863723755,
|
| 5983 |
+
"step": 851
|
| 5984 |
+
},
|
| 5985 |
+
{
|
| 5986 |
+
"epoch": 0.48609328198545143,
|
| 5987 |
+
"grad_norm": 0.96875,
|
| 5988 |
+
"learning_rate": 2.7439024390243906e-05,
|
| 5989 |
+
"loss": 1.2217812538146973,
|
| 5990 |
+
"step": 852
|
| 5991 |
+
},
|
| 5992 |
+
{
|
| 5993 |
+
"epoch": 0.4866638140065611,
|
| 5994 |
+
"grad_norm": 0.97265625,
|
| 5995 |
+
"learning_rate": 2.7411308203991133e-05,
|
| 5996 |
+
"loss": 1.201343297958374,
|
| 5997 |
+
"step": 853
|
| 5998 |
+
},
|
| 5999 |
+
{
|
| 6000 |
+
"epoch": 0.4872343460276708,
|
| 6001 |
+
"grad_norm": 0.953125,
|
| 6002 |
+
"learning_rate": 2.7383592017738362e-05,
|
| 6003 |
+
"loss": 1.1668754816055298,
|
| 6004 |
+
"step": 854
|
| 6005 |
+
},
|
| 6006 |
+
{
|
| 6007 |
+
"epoch": 0.4878048780487805,
|
| 6008 |
+
"grad_norm": 0.96875,
|
| 6009 |
+
"learning_rate": 2.7355875831485588e-05,
|
| 6010 |
+
"loss": 1.1264851093292236,
|
| 6011 |
+
"step": 855
|
| 6012 |
+
},
|
| 6013 |
+
{
|
| 6014 |
+
"epoch": 0.48837541006989016,
|
| 6015 |
+
"grad_norm": 0.9921875,
|
| 6016 |
+
"learning_rate": 2.7328159645232814e-05,
|
| 6017 |
+
"loss": 1.202168345451355,
|
| 6018 |
+
"step": 856
|
| 6019 |
+
},
|
| 6020 |
+
{
|
| 6021 |
+
"epoch": 0.48894594209099984,
|
| 6022 |
+
"grad_norm": 1.015625,
|
| 6023 |
+
"learning_rate": 2.7300443458980047e-05,
|
| 6024 |
+
"loss": 1.2231934070587158,
|
| 6025 |
+
"step": 857
|
| 6026 |
+
},
|
| 6027 |
+
{
|
| 6028 |
+
"epoch": 0.4895164741121095,
|
| 6029 |
+
"grad_norm": 1.0078125,
|
| 6030 |
+
"learning_rate": 2.7272727272727273e-05,
|
| 6031 |
+
"loss": 1.1511149406433105,
|
| 6032 |
+
"step": 858
|
| 6033 |
+
},
|
| 6034 |
+
{
|
| 6035 |
+
"epoch": 0.4900870061332192,
|
| 6036 |
+
"grad_norm": 1.0,
|
| 6037 |
+
"learning_rate": 2.7245011086474503e-05,
|
| 6038 |
+
"loss": 1.1898903846740723,
|
| 6039 |
+
"step": 859
|
| 6040 |
+
},
|
| 6041 |
+
{
|
| 6042 |
+
"epoch": 0.4906575381543289,
|
| 6043 |
+
"grad_norm": 1.0,
|
| 6044 |
+
"learning_rate": 2.721729490022173e-05,
|
| 6045 |
+
"loss": 1.1848946809768677,
|
| 6046 |
+
"step": 860
|
| 6047 |
+
},
|
| 6048 |
+
{
|
| 6049 |
+
"epoch": 0.49122807017543857,
|
| 6050 |
+
"grad_norm": 0.97265625,
|
| 6051 |
+
"learning_rate": 2.718957871396896e-05,
|
| 6052 |
+
"loss": 1.1898174285888672,
|
| 6053 |
+
"step": 861
|
| 6054 |
+
},
|
| 6055 |
+
{
|
| 6056 |
+
"epoch": 0.4917986021965483,
|
| 6057 |
+
"grad_norm": 1.0234375,
|
| 6058 |
+
"learning_rate": 2.7161862527716188e-05,
|
| 6059 |
+
"loss": 1.2187345027923584,
|
| 6060 |
+
"step": 862
|
| 6061 |
+
},
|
| 6062 |
+
{
|
| 6063 |
+
"epoch": 0.492369134217658,
|
| 6064 |
+
"grad_norm": 0.9765625,
|
| 6065 |
+
"learning_rate": 2.7134146341463417e-05,
|
| 6066 |
+
"loss": 1.1753157377243042,
|
| 6067 |
+
"step": 863
|
| 6068 |
+
},
|
| 6069 |
+
{
|
| 6070 |
+
"epoch": 0.49293966623876767,
|
| 6071 |
+
"grad_norm": 1.0078125,
|
| 6072 |
+
"learning_rate": 2.7106430155210643e-05,
|
| 6073 |
+
"loss": 1.2812843322753906,
|
| 6074 |
+
"step": 864
|
| 6075 |
+
},
|
| 6076 |
+
{
|
| 6077 |
+
"epoch": 0.49351019825987735,
|
| 6078 |
+
"grad_norm": 0.9921875,
|
| 6079 |
+
"learning_rate": 2.7078713968957876e-05,
|
| 6080 |
+
"loss": 1.2476832866668701,
|
| 6081 |
+
"step": 865
|
| 6082 |
+
},
|
| 6083 |
+
{
|
| 6084 |
+
"epoch": 0.49408073028098703,
|
| 6085 |
+
"grad_norm": 0.95703125,
|
| 6086 |
+
"learning_rate": 2.7050997782705102e-05,
|
| 6087 |
+
"loss": 1.1763570308685303,
|
| 6088 |
+
"step": 866
|
| 6089 |
+
},
|
| 6090 |
+
{
|
| 6091 |
+
"epoch": 0.4946512623020967,
|
| 6092 |
+
"grad_norm": 0.9609375,
|
| 6093 |
+
"learning_rate": 2.7023281596452328e-05,
|
| 6094 |
+
"loss": 1.159504771232605,
|
| 6095 |
+
"step": 867
|
| 6096 |
+
},
|
| 6097 |
+
{
|
| 6098 |
+
"epoch": 0.4952217943232064,
|
| 6099 |
+
"grad_norm": 0.94140625,
|
| 6100 |
+
"learning_rate": 2.6995565410199558e-05,
|
| 6101 |
+
"loss": 1.2344439029693604,
|
| 6102 |
+
"step": 868
|
| 6103 |
+
},
|
| 6104 |
+
{
|
| 6105 |
+
"epoch": 0.4957923263443161,
|
| 6106 |
+
"grad_norm": 0.98046875,
|
| 6107 |
+
"learning_rate": 2.6967849223946784e-05,
|
| 6108 |
+
"loss": 1.2668113708496094,
|
| 6109 |
+
"step": 869
|
| 6110 |
+
},
|
| 6111 |
+
{
|
| 6112 |
+
"epoch": 0.49636285836542576,
|
| 6113 |
+
"grad_norm": 0.96875,
|
| 6114 |
+
"learning_rate": 2.6940133037694017e-05,
|
| 6115 |
+
"loss": 1.2388842105865479,
|
| 6116 |
+
"step": 870
|
| 6117 |
+
},
|
| 6118 |
+
{
|
| 6119 |
+
"epoch": 0.49693339038653545,
|
| 6120 |
+
"grad_norm": 1.0,
|
| 6121 |
+
"learning_rate": 2.6912416851441243e-05,
|
| 6122 |
+
"loss": 1.197232723236084,
|
| 6123 |
+
"step": 871
|
| 6124 |
+
},
|
| 6125 |
+
{
|
| 6126 |
+
"epoch": 0.4975039224076451,
|
| 6127 |
+
"grad_norm": 0.98046875,
|
| 6128 |
+
"learning_rate": 2.6884700665188472e-05,
|
| 6129 |
+
"loss": 1.1960959434509277,
|
| 6130 |
+
"step": 872
|
| 6131 |
+
},
|
| 6132 |
+
{
|
| 6133 |
+
"epoch": 0.4980744544287548,
|
| 6134 |
+
"grad_norm": 0.99609375,
|
| 6135 |
+
"learning_rate": 2.6856984478935698e-05,
|
| 6136 |
+
"loss": 1.222888469696045,
|
| 6137 |
+
"step": 873
|
| 6138 |
+
},
|
| 6139 |
+
{
|
| 6140 |
+
"epoch": 0.4986449864498645,
|
| 6141 |
+
"grad_norm": 0.98828125,
|
| 6142 |
+
"learning_rate": 2.682926829268293e-05,
|
| 6143 |
+
"loss": 1.239640474319458,
|
| 6144 |
+
"step": 874
|
| 6145 |
+
},
|
| 6146 |
+
{
|
| 6147 |
+
"epoch": 0.4992155184709742,
|
| 6148 |
+
"grad_norm": 0.953125,
|
| 6149 |
+
"learning_rate": 2.6801552106430157e-05,
|
| 6150 |
+
"loss": 1.1557681560516357,
|
| 6151 |
+
"step": 875
|
| 6152 |
+
},
|
| 6153 |
+
{
|
| 6154 |
+
"epoch": 0.49978605049208386,
|
| 6155 |
+
"grad_norm": 1.0,
|
| 6156 |
+
"learning_rate": 2.6773835920177387e-05,
|
| 6157 |
+
"loss": 1.1697707176208496,
|
| 6158 |
+
"step": 876
|
| 6159 |
+
},
|
| 6160 |
+
{
|
| 6161 |
+
"epoch": 0.5003565825131936,
|
| 6162 |
+
"grad_norm": 1.0234375,
|
| 6163 |
+
"learning_rate": 2.6746119733924613e-05,
|
| 6164 |
+
"loss": 1.2065680027008057,
|
| 6165 |
+
"step": 877
|
| 6166 |
+
},
|
| 6167 |
+
{
|
| 6168 |
+
"epoch": 0.5009271145343033,
|
| 6169 |
+
"grad_norm": 0.9921875,
|
| 6170 |
+
"learning_rate": 2.6718403547671845e-05,
|
| 6171 |
+
"loss": 1.2194795608520508,
|
| 6172 |
+
"step": 878
|
| 6173 |
+
},
|
| 6174 |
+
{
|
| 6175 |
+
"epoch": 0.501497646555413,
|
| 6176 |
+
"grad_norm": 0.9609375,
|
| 6177 |
+
"learning_rate": 2.669068736141907e-05,
|
| 6178 |
+
"loss": 1.1722071170806885,
|
| 6179 |
+
"step": 879
|
| 6180 |
+
},
|
| 6181 |
+
{
|
| 6182 |
+
"epoch": 0.5020681785765226,
|
| 6183 |
+
"grad_norm": 0.99609375,
|
| 6184 |
+
"learning_rate": 2.6662971175166294e-05,
|
| 6185 |
+
"loss": 1.1860017776489258,
|
| 6186 |
+
"step": 880
|
| 6187 |
+
},
|
| 6188 |
+
{
|
| 6189 |
+
"epoch": 0.5026387105976323,
|
| 6190 |
+
"grad_norm": 0.9921875,
|
| 6191 |
+
"learning_rate": 2.6635254988913527e-05,
|
| 6192 |
+
"loss": 1.173937439918518,
|
| 6193 |
+
"step": 881
|
| 6194 |
+
},
|
| 6195 |
+
{
|
| 6196 |
+
"epoch": 0.503209242618742,
|
| 6197 |
+
"grad_norm": 1.0234375,
|
| 6198 |
+
"learning_rate": 2.6607538802660753e-05,
|
| 6199 |
+
"loss": 1.1348332166671753,
|
| 6200 |
+
"step": 882
|
| 6201 |
+
},
|
| 6202 |
+
{
|
| 6203 |
+
"epoch": 0.5037797746398517,
|
| 6204 |
+
"grad_norm": 0.97265625,
|
| 6205 |
+
"learning_rate": 2.6579822616407986e-05,
|
| 6206 |
+
"loss": 1.205221176147461,
|
| 6207 |
+
"step": 883
|
| 6208 |
+
},
|
| 6209 |
+
{
|
| 6210 |
+
"epoch": 0.5043503066609614,
|
| 6211 |
+
"grad_norm": 0.95703125,
|
| 6212 |
+
"learning_rate": 2.655210643015521e-05,
|
| 6213 |
+
"loss": 1.1510381698608398,
|
| 6214 |
+
"step": 884
|
| 6215 |
+
},
|
| 6216 |
+
{
|
| 6217 |
+
"epoch": 0.504920838682071,
|
| 6218 |
+
"grad_norm": 0.9921875,
|
| 6219 |
+
"learning_rate": 2.652439024390244e-05,
|
| 6220 |
+
"loss": 1.194382905960083,
|
| 6221 |
+
"step": 885
|
| 6222 |
+
},
|
| 6223 |
+
{
|
| 6224 |
+
"epoch": 0.5054913707031807,
|
| 6225 |
+
"grad_norm": 1.015625,
|
| 6226 |
+
"learning_rate": 2.6496674057649668e-05,
|
| 6227 |
+
"loss": 1.2697436809539795,
|
| 6228 |
+
"step": 886
|
| 6229 |
+
},
|
| 6230 |
+
{
|
| 6231 |
+
"epoch": 0.5060619027242904,
|
| 6232 |
+
"grad_norm": 0.9609375,
|
| 6233 |
+
"learning_rate": 2.64689578713969e-05,
|
| 6234 |
+
"loss": 1.1560388803482056,
|
| 6235 |
+
"step": 887
|
| 6236 |
+
},
|
| 6237 |
+
{
|
| 6238 |
+
"epoch": 0.5066324347454001,
|
| 6239 |
+
"grad_norm": 0.984375,
|
| 6240 |
+
"learning_rate": 2.6441241685144123e-05,
|
| 6241 |
+
"loss": 1.2498875856399536,
|
| 6242 |
+
"step": 888
|
| 6243 |
+
},
|
| 6244 |
+
{
|
| 6245 |
+
"epoch": 0.5072029667665098,
|
| 6246 |
+
"grad_norm": 0.9609375,
|
| 6247 |
+
"learning_rate": 2.6413525498891356e-05,
|
| 6248 |
+
"loss": 1.1706441640853882,
|
| 6249 |
+
"step": 889
|
| 6250 |
+
},
|
| 6251 |
+
{
|
| 6252 |
+
"epoch": 0.5077734987876195,
|
| 6253 |
+
"grad_norm": 0.99609375,
|
| 6254 |
+
"learning_rate": 2.6385809312638582e-05,
|
| 6255 |
+
"loss": 1.1960177421569824,
|
| 6256 |
+
"step": 890
|
| 6257 |
+
},
|
| 6258 |
+
{
|
| 6259 |
+
"epoch": 0.5083440308087291,
|
| 6260 |
+
"grad_norm": 1.0,
|
| 6261 |
+
"learning_rate": 2.6358093126385815e-05,
|
| 6262 |
+
"loss": 1.1732114553451538,
|
| 6263 |
+
"step": 891
|
| 6264 |
+
},
|
| 6265 |
+
{
|
| 6266 |
+
"epoch": 0.5089145628298388,
|
| 6267 |
+
"grad_norm": 0.984375,
|
| 6268 |
+
"learning_rate": 2.6330376940133038e-05,
|
| 6269 |
+
"loss": 1.1812173128128052,
|
| 6270 |
+
"step": 892
|
| 6271 |
+
},
|
| 6272 |
+
{
|
| 6273 |
+
"epoch": 0.5094850948509485,
|
| 6274 |
+
"grad_norm": 1.0,
|
| 6275 |
+
"learning_rate": 2.6302660753880264e-05,
|
| 6276 |
+
"loss": 1.243033528327942,
|
| 6277 |
+
"step": 893
|
| 6278 |
+
},
|
| 6279 |
+
{
|
| 6280 |
+
"epoch": 0.5100556268720582,
|
| 6281 |
+
"grad_norm": 0.984375,
|
| 6282 |
+
"learning_rate": 2.6274944567627497e-05,
|
| 6283 |
+
"loss": 1.1132174730300903,
|
| 6284 |
+
"step": 894
|
| 6285 |
+
},
|
| 6286 |
+
{
|
| 6287 |
+
"epoch": 0.5106261588931679,
|
| 6288 |
+
"grad_norm": 0.9296875,
|
| 6289 |
+
"learning_rate": 2.6247228381374723e-05,
|
| 6290 |
+
"loss": 1.129286289215088,
|
| 6291 |
+
"step": 895
|
| 6292 |
+
},
|
| 6293 |
+
{
|
| 6294 |
+
"epoch": 0.5111966909142776,
|
| 6295 |
+
"grad_norm": 1.0078125,
|
| 6296 |
+
"learning_rate": 2.6219512195121952e-05,
|
| 6297 |
+
"loss": 1.1969499588012695,
|
| 6298 |
+
"step": 896
|
| 6299 |
+
},
|
| 6300 |
+
{
|
| 6301 |
+
"epoch": 0.5117672229353872,
|
| 6302 |
+
"grad_norm": 0.9921875,
|
| 6303 |
+
"learning_rate": 2.6191796008869178e-05,
|
| 6304 |
+
"loss": 1.1295521259307861,
|
| 6305 |
+
"step": 897
|
| 6306 |
+
},
|
| 6307 |
+
{
|
| 6308 |
+
"epoch": 0.5123377549564969,
|
| 6309 |
+
"grad_norm": 1.0390625,
|
| 6310 |
+
"learning_rate": 2.616407982261641e-05,
|
| 6311 |
+
"loss": 1.1657040119171143,
|
| 6312 |
+
"step": 898
|
| 6313 |
+
},
|
| 6314 |
+
{
|
| 6315 |
+
"epoch": 0.5129082869776066,
|
| 6316 |
+
"grad_norm": 0.953125,
|
| 6317 |
+
"learning_rate": 2.6136363636363637e-05,
|
| 6318 |
+
"loss": 1.182844638824463,
|
| 6319 |
+
"step": 899
|
| 6320 |
+
},
|
| 6321 |
+
{
|
| 6322 |
+
"epoch": 0.5134788189987163,
|
| 6323 |
+
"grad_norm": 0.92578125,
|
| 6324 |
+
"learning_rate": 2.6108647450110867e-05,
|
| 6325 |
+
"loss": 1.11708664894104,
|
| 6326 |
+
"step": 900
|
| 6327 |
+
},
|
| 6328 |
+
{
|
| 6329 |
+
"epoch": 0.514049351019826,
|
| 6330 |
+
"grad_norm": 0.953125,
|
| 6331 |
+
"learning_rate": 2.6080931263858093e-05,
|
| 6332 |
+
"loss": 1.1282655000686646,
|
| 6333 |
+
"step": 901
|
| 6334 |
+
},
|
| 6335 |
+
{
|
| 6336 |
+
"epoch": 0.5146198830409356,
|
| 6337 |
+
"grad_norm": 0.98046875,
|
| 6338 |
+
"learning_rate": 2.6053215077605326e-05,
|
| 6339 |
+
"loss": 1.1830154657363892,
|
| 6340 |
+
"step": 902
|
| 6341 |
+
},
|
| 6342 |
+
{
|
| 6343 |
+
"epoch": 0.5151904150620453,
|
| 6344 |
+
"grad_norm": 0.9765625,
|
| 6345 |
+
"learning_rate": 2.602549889135255e-05,
|
| 6346 |
+
"loss": 1.1873393058776855,
|
| 6347 |
+
"step": 903
|
| 6348 |
+
},
|
| 6349 |
+
{
|
| 6350 |
+
"epoch": 0.515760947083155,
|
| 6351 |
+
"grad_norm": 0.953125,
|
| 6352 |
+
"learning_rate": 2.5997782705099778e-05,
|
| 6353 |
+
"loss": 1.1280049085617065,
|
| 6354 |
+
"step": 904
|
| 6355 |
+
},
|
| 6356 |
+
{
|
| 6357 |
+
"epoch": 0.5163314791042647,
|
| 6358 |
+
"grad_norm": 0.96875,
|
| 6359 |
+
"learning_rate": 2.5970066518847007e-05,
|
| 6360 |
+
"loss": 1.1866214275360107,
|
| 6361 |
+
"step": 905
|
| 6362 |
+
},
|
| 6363 |
+
{
|
| 6364 |
+
"epoch": 0.5169020111253744,
|
| 6365 |
+
"grad_norm": 0.9296875,
|
| 6366 |
+
"learning_rate": 2.5942350332594233e-05,
|
| 6367 |
+
"loss": 1.132464051246643,
|
| 6368 |
+
"step": 906
|
| 6369 |
+
},
|
| 6370 |
+
{
|
| 6371 |
+
"epoch": 0.517472543146484,
|
| 6372 |
+
"grad_norm": 0.9921875,
|
| 6373 |
+
"learning_rate": 2.5914634146341466e-05,
|
| 6374 |
+
"loss": 1.2057054042816162,
|
| 6375 |
+
"step": 907
|
| 6376 |
+
},
|
| 6377 |
+
{
|
| 6378 |
+
"epoch": 0.5180430751675937,
|
| 6379 |
+
"grad_norm": 0.96875,
|
| 6380 |
+
"learning_rate": 2.5886917960088692e-05,
|
| 6381 |
+
"loss": 1.1725504398345947,
|
| 6382 |
+
"step": 908
|
| 6383 |
+
},
|
| 6384 |
+
{
|
| 6385 |
+
"epoch": 0.5186136071887034,
|
| 6386 |
+
"grad_norm": 1.0078125,
|
| 6387 |
+
"learning_rate": 2.585920177383592e-05,
|
| 6388 |
+
"loss": 1.2105215787887573,
|
| 6389 |
+
"step": 909
|
| 6390 |
+
},
|
| 6391 |
+
{
|
| 6392 |
+
"epoch": 0.5191841392098131,
|
| 6393 |
+
"grad_norm": 0.9375,
|
| 6394 |
+
"learning_rate": 2.5831485587583148e-05,
|
| 6395 |
+
"loss": 1.126555323600769,
|
| 6396 |
+
"step": 910
|
| 6397 |
+
},
|
| 6398 |
+
{
|
| 6399 |
+
"epoch": 0.5197546712309228,
|
| 6400 |
+
"grad_norm": 0.953125,
|
| 6401 |
+
"learning_rate": 2.580376940133038e-05,
|
| 6402 |
+
"loss": 1.117220401763916,
|
| 6403 |
+
"step": 911
|
| 6404 |
+
},
|
| 6405 |
+
{
|
| 6406 |
+
"epoch": 0.5203252032520326,
|
| 6407 |
+
"grad_norm": 0.98828125,
|
| 6408 |
+
"learning_rate": 2.5776053215077607e-05,
|
| 6409 |
+
"loss": 1.1578710079193115,
|
| 6410 |
+
"step": 912
|
| 6411 |
+
},
|
| 6412 |
+
{
|
| 6413 |
+
"epoch": 0.5208957352731423,
|
| 6414 |
+
"grad_norm": 0.97265625,
|
| 6415 |
+
"learning_rate": 2.5748337028824836e-05,
|
| 6416 |
+
"loss": 1.1631922721862793,
|
| 6417 |
+
"step": 913
|
| 6418 |
+
},
|
| 6419 |
+
{
|
| 6420 |
+
"epoch": 0.5214662672942519,
|
| 6421 |
+
"grad_norm": 0.9921875,
|
| 6422 |
+
"learning_rate": 2.5720620842572062e-05,
|
| 6423 |
+
"loss": 1.2013893127441406,
|
| 6424 |
+
"step": 914
|
| 6425 |
+
},
|
| 6426 |
+
{
|
| 6427 |
+
"epoch": 0.5220367993153616,
|
| 6428 |
+
"grad_norm": 1.0,
|
| 6429 |
+
"learning_rate": 2.5692904656319295e-05,
|
| 6430 |
+
"loss": 1.159932017326355,
|
| 6431 |
+
"step": 915
|
| 6432 |
+
},
|
| 6433 |
+
{
|
| 6434 |
+
"epoch": 0.5226073313364713,
|
| 6435 |
+
"grad_norm": 0.9453125,
|
| 6436 |
+
"learning_rate": 2.566518847006652e-05,
|
| 6437 |
+
"loss": 1.1213711500167847,
|
| 6438 |
+
"step": 916
|
| 6439 |
+
},
|
| 6440 |
+
{
|
| 6441 |
+
"epoch": 0.523177863357581,
|
| 6442 |
+
"grad_norm": 1.0,
|
| 6443 |
+
"learning_rate": 2.5637472283813747e-05,
|
| 6444 |
+
"loss": 1.2035624980926514,
|
| 6445 |
+
"step": 917
|
| 6446 |
+
},
|
| 6447 |
+
{
|
| 6448 |
+
"epoch": 0.5237483953786907,
|
| 6449 |
+
"grad_norm": 0.921875,
|
| 6450 |
+
"learning_rate": 2.5609756097560977e-05,
|
| 6451 |
+
"loss": 1.100569725036621,
|
| 6452 |
+
"step": 918
|
| 6453 |
+
},
|
| 6454 |
+
{
|
| 6455 |
+
"epoch": 0.5243189273998003,
|
| 6456 |
+
"grad_norm": 0.99609375,
|
| 6457 |
+
"learning_rate": 2.5582039911308203e-05,
|
| 6458 |
+
"loss": 1.1802055835723877,
|
| 6459 |
+
"step": 919
|
| 6460 |
+
},
|
| 6461 |
+
{
|
| 6462 |
+
"epoch": 0.52488945942091,
|
| 6463 |
+
"grad_norm": 0.9453125,
|
| 6464 |
+
"learning_rate": 2.5554323725055436e-05,
|
| 6465 |
+
"loss": 1.2129563093185425,
|
| 6466 |
+
"step": 920
|
| 6467 |
+
},
|
| 6468 |
+
{
|
| 6469 |
+
"epoch": 0.5254599914420197,
|
| 6470 |
+
"grad_norm": 0.984375,
|
| 6471 |
+
"learning_rate": 2.552660753880266e-05,
|
| 6472 |
+
"loss": 1.2040753364562988,
|
| 6473 |
+
"step": 921
|
| 6474 |
+
},
|
| 6475 |
+
{
|
| 6476 |
+
"epoch": 0.5260305234631294,
|
| 6477 |
+
"grad_norm": 0.9921875,
|
| 6478 |
+
"learning_rate": 2.549889135254989e-05,
|
| 6479 |
+
"loss": 1.1266067028045654,
|
| 6480 |
+
"step": 922
|
| 6481 |
+
},
|
| 6482 |
+
{
|
| 6483 |
+
"epoch": 0.5266010554842391,
|
| 6484 |
+
"grad_norm": 1.0078125,
|
| 6485 |
+
"learning_rate": 2.5471175166297117e-05,
|
| 6486 |
+
"loss": 1.1967592239379883,
|
| 6487 |
+
"step": 923
|
| 6488 |
+
},
|
| 6489 |
+
{
|
| 6490 |
+
"epoch": 0.5271715875053488,
|
| 6491 |
+
"grad_norm": 0.97265625,
|
| 6492 |
+
"learning_rate": 2.544345898004435e-05,
|
| 6493 |
+
"loss": 1.1658574342727661,
|
| 6494 |
+
"step": 924
|
| 6495 |
+
},
|
| 6496 |
+
{
|
| 6497 |
+
"epoch": 0.5277421195264584,
|
| 6498 |
+
"grad_norm": 0.96875,
|
| 6499 |
+
"learning_rate": 2.5415742793791576e-05,
|
| 6500 |
+
"loss": 1.1974247694015503,
|
| 6501 |
+
"step": 925
|
| 6502 |
+
},
|
| 6503 |
+
{
|
| 6504 |
+
"epoch": 0.5283126515475681,
|
| 6505 |
+
"grad_norm": 0.96484375,
|
| 6506 |
+
"learning_rate": 2.5388026607538806e-05,
|
| 6507 |
+
"loss": 1.175785779953003,
|
| 6508 |
+
"step": 926
|
| 6509 |
+
},
|
| 6510 |
+
{
|
| 6511 |
+
"epoch": 0.5288831835686778,
|
| 6512 |
+
"grad_norm": 0.98828125,
|
| 6513 |
+
"learning_rate": 2.5360310421286032e-05,
|
| 6514 |
+
"loss": 1.2295399904251099,
|
| 6515 |
+
"step": 927
|
| 6516 |
+
},
|
| 6517 |
+
{
|
| 6518 |
+
"epoch": 0.5294537155897875,
|
| 6519 |
+
"grad_norm": 1.015625,
|
| 6520 |
+
"learning_rate": 2.5332594235033258e-05,
|
| 6521 |
+
"loss": 1.1797332763671875,
|
| 6522 |
+
"step": 928
|
| 6523 |
+
},
|
| 6524 |
+
{
|
| 6525 |
+
"epoch": 0.5300242476108972,
|
| 6526 |
+
"grad_norm": 0.9375,
|
| 6527 |
+
"learning_rate": 2.530487804878049e-05,
|
| 6528 |
+
"loss": 1.1036921739578247,
|
| 6529 |
+
"step": 929
|
| 6530 |
+
},
|
| 6531 |
+
{
|
| 6532 |
+
"epoch": 0.5305947796320069,
|
| 6533 |
+
"grad_norm": 1.0,
|
| 6534 |
+
"learning_rate": 2.5277161862527717e-05,
|
| 6535 |
+
"loss": 1.1661919355392456,
|
| 6536 |
+
"step": 930
|
| 6537 |
+
},
|
| 6538 |
+
{
|
| 6539 |
+
"epoch": 0.5311653116531165,
|
| 6540 |
+
"grad_norm": 1.015625,
|
| 6541 |
+
"learning_rate": 2.5249445676274946e-05,
|
| 6542 |
+
"loss": 1.220758318901062,
|
| 6543 |
+
"step": 931
|
| 6544 |
+
},
|
| 6545 |
+
{
|
| 6546 |
+
"epoch": 0.5317358436742262,
|
| 6547 |
+
"grad_norm": 1.015625,
|
| 6548 |
+
"learning_rate": 2.5221729490022172e-05,
|
| 6549 |
+
"loss": 1.2072967290878296,
|
| 6550 |
+
"step": 932
|
| 6551 |
+
},
|
| 6552 |
+
{
|
| 6553 |
+
"epoch": 0.5323063756953359,
|
| 6554 |
+
"grad_norm": 0.98046875,
|
| 6555 |
+
"learning_rate": 2.5194013303769405e-05,
|
| 6556 |
+
"loss": 1.211767315864563,
|
| 6557 |
+
"step": 933
|
| 6558 |
+
},
|
| 6559 |
+
{
|
| 6560 |
+
"epoch": 0.5328769077164456,
|
| 6561 |
+
"grad_norm": 1.0,
|
| 6562 |
+
"learning_rate": 2.516629711751663e-05,
|
| 6563 |
+
"loss": 1.196463942527771,
|
| 6564 |
+
"step": 934
|
| 6565 |
+
},
|
| 6566 |
+
{
|
| 6567 |
+
"epoch": 0.5334474397375553,
|
| 6568 |
+
"grad_norm": 0.96484375,
|
| 6569 |
+
"learning_rate": 2.513858093126386e-05,
|
| 6570 |
+
"loss": 1.1342837810516357,
|
| 6571 |
+
"step": 935
|
| 6572 |
+
},
|
| 6573 |
+
{
|
| 6574 |
+
"epoch": 0.5340179717586649,
|
| 6575 |
+
"grad_norm": 0.9765625,
|
| 6576 |
+
"learning_rate": 2.5110864745011087e-05,
|
| 6577 |
+
"loss": 1.155871868133545,
|
| 6578 |
+
"step": 936
|
| 6579 |
+
},
|
| 6580 |
+
{
|
| 6581 |
+
"epoch": 0.5345885037797746,
|
| 6582 |
+
"grad_norm": 1.0,
|
| 6583 |
+
"learning_rate": 2.508314855875832e-05,
|
| 6584 |
+
"loss": 1.1863211393356323,
|
| 6585 |
+
"step": 937
|
| 6586 |
+
},
|
| 6587 |
+
{
|
| 6588 |
+
"epoch": 0.5351590358008843,
|
| 6589 |
+
"grad_norm": 0.96484375,
|
| 6590 |
+
"learning_rate": 2.5055432372505546e-05,
|
| 6591 |
+
"loss": 1.1399109363555908,
|
| 6592 |
+
"step": 938
|
| 6593 |
+
},
|
| 6594 |
+
{
|
| 6595 |
+
"epoch": 0.535729567821994,
|
| 6596 |
+
"grad_norm": 0.96875,
|
| 6597 |
+
"learning_rate": 2.5027716186252775e-05,
|
| 6598 |
+
"loss": 1.148442268371582,
|
| 6599 |
+
"step": 939
|
| 6600 |
+
},
|
| 6601 |
+
{
|
| 6602 |
+
"epoch": 0.5363000998431037,
|
| 6603 |
+
"grad_norm": 1.0234375,
|
| 6604 |
+
"learning_rate": 2.5e-05,
|
| 6605 |
+
"loss": 1.2298827171325684,
|
| 6606 |
+
"step": 940
|
| 6607 |
+
},
|
| 6608 |
+
{
|
| 6609 |
+
"epoch": 0.5368706318642134,
|
| 6610 |
+
"grad_norm": 0.953125,
|
| 6611 |
+
"learning_rate": 2.497228381374723e-05,
|
| 6612 |
+
"loss": 1.1379940509796143,
|
| 6613 |
+
"step": 941
|
| 6614 |
+
},
|
| 6615 |
+
{
|
| 6616 |
+
"epoch": 0.537441163885323,
|
| 6617 |
+
"grad_norm": 0.9453125,
|
| 6618 |
+
"learning_rate": 2.4944567627494457e-05,
|
| 6619 |
+
"loss": 1.1394915580749512,
|
| 6620 |
+
"step": 942
|
| 6621 |
+
},
|
| 6622 |
+
{
|
| 6623 |
+
"epoch": 0.5380116959064327,
|
| 6624 |
+
"grad_norm": 0.9921875,
|
| 6625 |
+
"learning_rate": 2.4916851441241686e-05,
|
| 6626 |
+
"loss": 1.180498480796814,
|
| 6627 |
+
"step": 943
|
| 6628 |
+
},
|
| 6629 |
+
{
|
| 6630 |
+
"epoch": 0.5385822279275424,
|
| 6631 |
+
"grad_norm": 1.0546875,
|
| 6632 |
+
"learning_rate": 2.4889135254988916e-05,
|
| 6633 |
+
"loss": 1.2175443172454834,
|
| 6634 |
+
"step": 944
|
| 6635 |
+
},
|
| 6636 |
+
{
|
| 6637 |
+
"epoch": 0.5391527599486521,
|
| 6638 |
+
"grad_norm": 0.98828125,
|
| 6639 |
+
"learning_rate": 2.4861419068736145e-05,
|
| 6640 |
+
"loss": 1.1404181718826294,
|
| 6641 |
+
"step": 945
|
| 6642 |
+
},
|
| 6643 |
+
{
|
| 6644 |
+
"epoch": 0.5397232919697618,
|
| 6645 |
+
"grad_norm": 0.9765625,
|
| 6646 |
+
"learning_rate": 2.483370288248337e-05,
|
| 6647 |
+
"loss": 1.1929075717926025,
|
| 6648 |
+
"step": 946
|
| 6649 |
+
},
|
| 6650 |
+
{
|
| 6651 |
+
"epoch": 0.5402938239908714,
|
| 6652 |
+
"grad_norm": 0.96484375,
|
| 6653 |
+
"learning_rate": 2.4805986696230597e-05,
|
| 6654 |
+
"loss": 1.1470379829406738,
|
| 6655 |
+
"step": 947
|
| 6656 |
+
},
|
| 6657 |
+
{
|
| 6658 |
+
"epoch": 0.5408643560119811,
|
| 6659 |
+
"grad_norm": 1.0,
|
| 6660 |
+
"learning_rate": 2.4778270509977827e-05,
|
| 6661 |
+
"loss": 1.1692397594451904,
|
| 6662 |
+
"step": 948
|
| 6663 |
+
},
|
| 6664 |
+
{
|
| 6665 |
+
"epoch": 0.5414348880330908,
|
| 6666 |
+
"grad_norm": 1.0078125,
|
| 6667 |
+
"learning_rate": 2.4750554323725056e-05,
|
| 6668 |
+
"loss": 1.2243307828903198,
|
| 6669 |
+
"step": 949
|
| 6670 |
+
},
|
| 6671 |
+
{
|
| 6672 |
+
"epoch": 0.5420054200542005,
|
| 6673 |
+
"grad_norm": 0.99609375,
|
| 6674 |
+
"learning_rate": 2.4722838137472286e-05,
|
| 6675 |
+
"loss": 1.1853331327438354,
|
| 6676 |
+
"step": 950
|
| 6677 |
+
},
|
| 6678 |
+
{
|
| 6679 |
+
"epoch": 0.5425759520753102,
|
| 6680 |
+
"grad_norm": 1.015625,
|
| 6681 |
+
"learning_rate": 2.4695121951219512e-05,
|
| 6682 |
+
"loss": 1.2312514781951904,
|
| 6683 |
+
"step": 951
|
| 6684 |
+
},
|
| 6685 |
+
{
|
| 6686 |
+
"epoch": 0.5431464840964199,
|
| 6687 |
+
"grad_norm": 0.953125,
|
| 6688 |
+
"learning_rate": 2.466740576496674e-05,
|
| 6689 |
+
"loss": 1.1487960815429688,
|
| 6690 |
+
"step": 952
|
| 6691 |
+
},
|
| 6692 |
+
{
|
| 6693 |
+
"epoch": 0.5437170161175297,
|
| 6694 |
+
"grad_norm": 0.96875,
|
| 6695 |
+
"learning_rate": 2.463968957871397e-05,
|
| 6696 |
+
"loss": 1.1434435844421387,
|
| 6697 |
+
"step": 953
|
| 6698 |
+
},
|
| 6699 |
+
{
|
| 6700 |
+
"epoch": 0.5442875481386393,
|
| 6701 |
+
"grad_norm": 0.97265625,
|
| 6702 |
+
"learning_rate": 2.46119733924612e-05,
|
| 6703 |
+
"loss": 1.2065646648406982,
|
| 6704 |
+
"step": 954
|
| 6705 |
+
},
|
| 6706 |
+
{
|
| 6707 |
+
"epoch": 0.544858080159749,
|
| 6708 |
+
"grad_norm": 0.96875,
|
| 6709 |
+
"learning_rate": 2.4584257206208426e-05,
|
| 6710 |
+
"loss": 1.1631767749786377,
|
| 6711 |
+
"step": 955
|
| 6712 |
+
},
|
| 6713 |
+
{
|
| 6714 |
+
"epoch": 0.5454286121808587,
|
| 6715 |
+
"grad_norm": 0.94921875,
|
| 6716 |
+
"learning_rate": 2.4556541019955656e-05,
|
| 6717 |
+
"loss": 1.19287109375,
|
| 6718 |
+
"step": 956
|
| 6719 |
+
},
|
| 6720 |
+
{
|
| 6721 |
+
"epoch": 0.5459991442019684,
|
| 6722 |
+
"grad_norm": 0.98828125,
|
| 6723 |
+
"learning_rate": 2.4528824833702885e-05,
|
| 6724 |
+
"loss": 1.183131456375122,
|
| 6725 |
+
"step": 957
|
| 6726 |
+
},
|
| 6727 |
+
{
|
| 6728 |
+
"epoch": 0.5465696762230781,
|
| 6729 |
+
"grad_norm": 0.953125,
|
| 6730 |
+
"learning_rate": 2.4501108647450115e-05,
|
| 6731 |
+
"loss": 1.1865886449813843,
|
| 6732 |
+
"step": 958
|
| 6733 |
+
},
|
| 6734 |
+
{
|
| 6735 |
+
"epoch": 0.5471402082441877,
|
| 6736 |
+
"grad_norm": 0.98046875,
|
| 6737 |
+
"learning_rate": 2.447339246119734e-05,
|
| 6738 |
+
"loss": 1.1511285305023193,
|
| 6739 |
+
"step": 959
|
| 6740 |
+
},
|
| 6741 |
+
{
|
| 6742 |
+
"epoch": 0.5477107402652974,
|
| 6743 |
+
"grad_norm": 0.96484375,
|
| 6744 |
+
"learning_rate": 2.4445676274944567e-05,
|
| 6745 |
+
"loss": 1.1591591835021973,
|
| 6746 |
+
"step": 960
|
| 6747 |
+
},
|
| 6748 |
+
{
|
| 6749 |
+
"epoch": 0.5482812722864071,
|
| 6750 |
+
"grad_norm": 0.9375,
|
| 6751 |
+
"learning_rate": 2.4417960088691796e-05,
|
| 6752 |
+
"loss": 1.1885075569152832,
|
| 6753 |
+
"step": 961
|
| 6754 |
+
},
|
| 6755 |
+
{
|
| 6756 |
+
"epoch": 0.5488518043075168,
|
| 6757 |
+
"grad_norm": 0.97265625,
|
| 6758 |
+
"learning_rate": 2.4390243902439026e-05,
|
| 6759 |
+
"loss": 1.1785187721252441,
|
| 6760 |
+
"step": 962
|
| 6761 |
+
},
|
| 6762 |
+
{
|
| 6763 |
+
"epoch": 0.5494223363286265,
|
| 6764 |
+
"grad_norm": 0.96484375,
|
| 6765 |
+
"learning_rate": 2.4362527716186255e-05,
|
| 6766 |
+
"loss": 1.1689701080322266,
|
| 6767 |
+
"step": 963
|
| 6768 |
+
},
|
| 6769 |
+
{
|
| 6770 |
+
"epoch": 0.5499928683497362,
|
| 6771 |
+
"grad_norm": 0.94921875,
|
| 6772 |
+
"learning_rate": 2.433481152993348e-05,
|
| 6773 |
+
"loss": 1.1543480157852173,
|
| 6774 |
+
"step": 964
|
| 6775 |
+
},
|
| 6776 |
+
{
|
| 6777 |
+
"epoch": 0.5505634003708458,
|
| 6778 |
+
"grad_norm": 1.0078125,
|
| 6779 |
+
"learning_rate": 2.430709534368071e-05,
|
| 6780 |
+
"loss": 1.196134328842163,
|
| 6781 |
+
"step": 965
|
| 6782 |
+
},
|
| 6783 |
+
{
|
| 6784 |
+
"epoch": 0.5511339323919555,
|
| 6785 |
+
"grad_norm": 0.98828125,
|
| 6786 |
+
"learning_rate": 2.427937915742794e-05,
|
| 6787 |
+
"loss": 1.2235426902770996,
|
| 6788 |
+
"step": 966
|
| 6789 |
+
},
|
| 6790 |
+
{
|
| 6791 |
+
"epoch": 0.5517044644130652,
|
| 6792 |
+
"grad_norm": 0.96484375,
|
| 6793 |
+
"learning_rate": 2.425166297117517e-05,
|
| 6794 |
+
"loss": 1.2253239154815674,
|
| 6795 |
+
"step": 967
|
| 6796 |
+
},
|
| 6797 |
+
{
|
| 6798 |
+
"epoch": 0.5522749964341749,
|
| 6799 |
+
"grad_norm": 0.953125,
|
| 6800 |
+
"learning_rate": 2.4223946784922396e-05,
|
| 6801 |
+
"loss": 1.1899304389953613,
|
| 6802 |
+
"step": 968
|
| 6803 |
+
},
|
| 6804 |
+
{
|
| 6805 |
+
"epoch": 0.5528455284552846,
|
| 6806 |
+
"grad_norm": 1.0234375,
|
| 6807 |
+
"learning_rate": 2.4196230598669625e-05,
|
| 6808 |
+
"loss": 1.1620666980743408,
|
| 6809 |
+
"step": 969
|
| 6810 |
+
},
|
| 6811 |
+
{
|
| 6812 |
+
"epoch": 0.5534160604763942,
|
| 6813 |
+
"grad_norm": 0.9765625,
|
| 6814 |
+
"learning_rate": 2.4168514412416855e-05,
|
| 6815 |
+
"loss": 1.1896693706512451,
|
| 6816 |
+
"step": 970
|
| 6817 |
+
},
|
| 6818 |
+
{
|
| 6819 |
+
"epoch": 0.5539865924975039,
|
| 6820 |
+
"grad_norm": 0.921875,
|
| 6821 |
+
"learning_rate": 2.414079822616408e-05,
|
| 6822 |
+
"loss": 1.1168513298034668,
|
| 6823 |
+
"step": 971
|
| 6824 |
+
},
|
| 6825 |
+
{
|
| 6826 |
+
"epoch": 0.5545571245186136,
|
| 6827 |
+
"grad_norm": 0.9453125,
|
| 6828 |
+
"learning_rate": 2.4113082039911307e-05,
|
| 6829 |
+
"loss": 1.1533100605010986,
|
| 6830 |
+
"step": 972
|
| 6831 |
+
},
|
| 6832 |
+
{
|
| 6833 |
+
"epoch": 0.5551276565397233,
|
| 6834 |
+
"grad_norm": 0.953125,
|
| 6835 |
+
"learning_rate": 2.4085365853658536e-05,
|
| 6836 |
+
"loss": 1.11790132522583,
|
| 6837 |
+
"step": 973
|
| 6838 |
+
},
|
| 6839 |
+
{
|
| 6840 |
+
"epoch": 0.555698188560833,
|
| 6841 |
+
"grad_norm": 0.96875,
|
| 6842 |
+
"learning_rate": 2.4057649667405766e-05,
|
| 6843 |
+
"loss": 1.1832971572875977,
|
| 6844 |
+
"step": 974
|
| 6845 |
+
},
|
| 6846 |
+
{
|
| 6847 |
+
"epoch": 0.5562687205819427,
|
| 6848 |
+
"grad_norm": 0.9375,
|
| 6849 |
+
"learning_rate": 2.4029933481152995e-05,
|
| 6850 |
+
"loss": 1.136374592781067,
|
| 6851 |
+
"step": 975
|
| 6852 |
+
},
|
| 6853 |
+
{
|
| 6854 |
+
"epoch": 0.5568392526030523,
|
| 6855 |
+
"grad_norm": 0.94140625,
|
| 6856 |
+
"learning_rate": 2.400221729490022e-05,
|
| 6857 |
+
"loss": 1.13529634475708,
|
| 6858 |
+
"step": 976
|
| 6859 |
+
},
|
| 6860 |
+
{
|
| 6861 |
+
"epoch": 0.557409784624162,
|
| 6862 |
+
"grad_norm": 0.94140625,
|
| 6863 |
+
"learning_rate": 2.397450110864745e-05,
|
| 6864 |
+
"loss": 1.152282476425171,
|
| 6865 |
+
"step": 977
|
| 6866 |
+
},
|
| 6867 |
+
{
|
| 6868 |
+
"epoch": 0.5579803166452717,
|
| 6869 |
+
"grad_norm": 0.9375,
|
| 6870 |
+
"learning_rate": 2.394678492239468e-05,
|
| 6871 |
+
"loss": 1.1445283889770508,
|
| 6872 |
+
"step": 978
|
| 6873 |
+
},
|
| 6874 |
+
{
|
| 6875 |
+
"epoch": 0.5585508486663814,
|
| 6876 |
+
"grad_norm": 0.96875,
|
| 6877 |
+
"learning_rate": 2.391906873614191e-05,
|
| 6878 |
+
"loss": 1.1682907342910767,
|
| 6879 |
+
"step": 979
|
| 6880 |
+
},
|
| 6881 |
+
{
|
| 6882 |
+
"epoch": 0.5591213806874911,
|
| 6883 |
+
"grad_norm": 0.96484375,
|
| 6884 |
+
"learning_rate": 2.3891352549889136e-05,
|
| 6885 |
+
"loss": 1.2181129455566406,
|
| 6886 |
+
"step": 980
|
| 6887 |
+
},
|
| 6888 |
+
{
|
| 6889 |
+
"epoch": 0.5596919127086007,
|
| 6890 |
+
"grad_norm": 0.94140625,
|
| 6891 |
+
"learning_rate": 2.3863636363636365e-05,
|
| 6892 |
+
"loss": 1.1683390140533447,
|
| 6893 |
+
"step": 981
|
| 6894 |
+
},
|
| 6895 |
+
{
|
| 6896 |
+
"epoch": 0.5602624447297104,
|
| 6897 |
+
"grad_norm": 0.9609375,
|
| 6898 |
+
"learning_rate": 2.3835920177383595e-05,
|
| 6899 |
+
"loss": 1.1526210308074951,
|
| 6900 |
+
"step": 982
|
| 6901 |
+
},
|
| 6902 |
+
{
|
| 6903 |
+
"epoch": 0.5608329767508201,
|
| 6904 |
+
"grad_norm": 0.94140625,
|
| 6905 |
+
"learning_rate": 2.380820399113082e-05,
|
| 6906 |
+
"loss": 1.1839709281921387,
|
| 6907 |
+
"step": 983
|
| 6908 |
+
},
|
| 6909 |
+
{
|
| 6910 |
+
"epoch": 0.5614035087719298,
|
| 6911 |
+
"grad_norm": 0.96875,
|
| 6912 |
+
"learning_rate": 2.378048780487805e-05,
|
| 6913 |
+
"loss": 1.171961784362793,
|
| 6914 |
+
"step": 984
|
| 6915 |
+
},
|
| 6916 |
+
{
|
| 6917 |
+
"epoch": 0.5619740407930395,
|
| 6918 |
+
"grad_norm": 0.96484375,
|
| 6919 |
+
"learning_rate": 2.3752771618625276e-05,
|
| 6920 |
+
"loss": 1.1404699087142944,
|
| 6921 |
+
"step": 985
|
| 6922 |
+
},
|
| 6923 |
+
{
|
| 6924 |
+
"epoch": 0.5625445728141492,
|
| 6925 |
+
"grad_norm": 0.9375,
|
| 6926 |
+
"learning_rate": 2.3725055432372506e-05,
|
| 6927 |
+
"loss": 1.1446641683578491,
|
| 6928 |
+
"step": 986
|
| 6929 |
+
},
|
| 6930 |
+
{
|
| 6931 |
+
"epoch": 0.5631151048352588,
|
| 6932 |
+
"grad_norm": 0.9375,
|
| 6933 |
+
"learning_rate": 2.3697339246119735e-05,
|
| 6934 |
+
"loss": 1.1063508987426758,
|
| 6935 |
+
"step": 987
|
| 6936 |
+
},
|
| 6937 |
+
{
|
| 6938 |
+
"epoch": 0.5636856368563685,
|
| 6939 |
+
"grad_norm": 0.9921875,
|
| 6940 |
+
"learning_rate": 2.3669623059866965e-05,
|
| 6941 |
+
"loss": 1.1023223400115967,
|
| 6942 |
+
"step": 988
|
| 6943 |
+
},
|
| 6944 |
+
{
|
| 6945 |
+
"epoch": 0.5642561688774782,
|
| 6946 |
+
"grad_norm": 0.9453125,
|
| 6947 |
+
"learning_rate": 2.364190687361419e-05,
|
| 6948 |
+
"loss": 1.157923698425293,
|
| 6949 |
+
"step": 989
|
| 6950 |
+
},
|
| 6951 |
+
{
|
| 6952 |
+
"epoch": 0.5648267008985879,
|
| 6953 |
+
"grad_norm": 0.9609375,
|
| 6954 |
+
"learning_rate": 2.361419068736142e-05,
|
| 6955 |
+
"loss": 1.1578837633132935,
|
| 6956 |
+
"step": 990
|
| 6957 |
+
},
|
| 6958 |
+
{
|
| 6959 |
+
"epoch": 0.5653972329196976,
|
| 6960 |
+
"grad_norm": 0.94140625,
|
| 6961 |
+
"learning_rate": 2.358647450110865e-05,
|
| 6962 |
+
"loss": 1.110813856124878,
|
| 6963 |
+
"step": 991
|
| 6964 |
+
},
|
| 6965 |
+
{
|
| 6966 |
+
"epoch": 0.5659677649408072,
|
| 6967 |
+
"grad_norm": 0.91796875,
|
| 6968 |
+
"learning_rate": 2.355875831485588e-05,
|
| 6969 |
+
"loss": 1.1383073329925537,
|
| 6970 |
+
"step": 992
|
| 6971 |
+
},
|
| 6972 |
+
{
|
| 6973 |
+
"epoch": 0.566538296961917,
|
| 6974 |
+
"grad_norm": 0.94921875,
|
| 6975 |
+
"learning_rate": 2.3531042128603105e-05,
|
| 6976 |
+
"loss": 1.1709469556808472,
|
| 6977 |
+
"step": 993
|
| 6978 |
+
},
|
| 6979 |
+
{
|
| 6980 |
+
"epoch": 0.5671088289830267,
|
| 6981 |
+
"grad_norm": 0.984375,
|
| 6982 |
+
"learning_rate": 2.3503325942350335e-05,
|
| 6983 |
+
"loss": 1.1664437055587769,
|
| 6984 |
+
"step": 994
|
| 6985 |
+
},
|
| 6986 |
+
{
|
| 6987 |
+
"epoch": 0.5676793610041364,
|
| 6988 |
+
"grad_norm": 0.953125,
|
| 6989 |
+
"learning_rate": 2.347560975609756e-05,
|
| 6990 |
+
"loss": 1.1766831874847412,
|
| 6991 |
+
"step": 995
|
| 6992 |
+
},
|
| 6993 |
+
{
|
| 6994 |
+
"epoch": 0.5682498930252461,
|
| 6995 |
+
"grad_norm": 0.92578125,
|
| 6996 |
+
"learning_rate": 2.344789356984479e-05,
|
| 6997 |
+
"loss": 1.1888954639434814,
|
| 6998 |
+
"step": 996
|
| 6999 |
+
},
|
| 7000 |
+
{
|
| 7001 |
+
"epoch": 0.5688204250463558,
|
| 7002 |
+
"grad_norm": 1.0078125,
|
| 7003 |
+
"learning_rate": 2.3420177383592016e-05,
|
| 7004 |
+
"loss": 1.1901835203170776,
|
| 7005 |
+
"step": 997
|
| 7006 |
+
},
|
| 7007 |
+
{
|
| 7008 |
+
"epoch": 0.5693909570674655,
|
| 7009 |
+
"grad_norm": 0.9140625,
|
| 7010 |
+
"learning_rate": 2.3392461197339246e-05,
|
| 7011 |
+
"loss": 1.13261079788208,
|
| 7012 |
+
"step": 998
|
| 7013 |
+
},
|
| 7014 |
+
{
|
| 7015 |
+
"epoch": 0.5699614890885751,
|
| 7016 |
+
"grad_norm": 0.99609375,
|
| 7017 |
+
"learning_rate": 2.3364745011086475e-05,
|
| 7018 |
+
"loss": 1.2113161087036133,
|
| 7019 |
+
"step": 999
|
| 7020 |
+
},
|
| 7021 |
+
{
|
| 7022 |
+
"epoch": 0.5705320211096848,
|
| 7023 |
+
"grad_norm": 0.9609375,
|
| 7024 |
+
"learning_rate": 2.3337028824833705e-05,
|
| 7025 |
+
"loss": 1.1643033027648926,
|
| 7026 |
+
"step": 1000
|
| 7027 |
+
},
|
| 7028 |
+
{
|
| 7029 |
+
"epoch": 0.5711025531307945,
|
| 7030 |
+
"grad_norm": 1.0,
|
| 7031 |
+
"learning_rate": 2.330931263858093e-05,
|
| 7032 |
+
"loss": 1.2085559368133545,
|
| 7033 |
+
"step": 1001
|
| 7034 |
+
},
|
| 7035 |
+
{
|
| 7036 |
+
"epoch": 0.5716730851519042,
|
| 7037 |
+
"grad_norm": 0.97265625,
|
| 7038 |
+
"learning_rate": 2.328159645232816e-05,
|
| 7039 |
+
"loss": 1.1837122440338135,
|
| 7040 |
+
"step": 1002
|
| 7041 |
+
},
|
| 7042 |
+
{
|
| 7043 |
+
"epoch": 0.5722436171730139,
|
| 7044 |
+
"grad_norm": 1.03125,
|
| 7045 |
+
"learning_rate": 2.325388026607539e-05,
|
| 7046 |
+
"loss": 1.2685991525650024,
|
| 7047 |
+
"step": 1003
|
| 7048 |
+
},
|
| 7049 |
+
{
|
| 7050 |
+
"epoch": 0.5728141491941235,
|
| 7051 |
+
"grad_norm": 0.95703125,
|
| 7052 |
+
"learning_rate": 2.322616407982262e-05,
|
| 7053 |
+
"loss": 1.1660895347595215,
|
| 7054 |
+
"step": 1004
|
| 7055 |
+
},
|
| 7056 |
+
{
|
| 7057 |
+
"epoch": 0.5733846812152332,
|
| 7058 |
+
"grad_norm": 1.0,
|
| 7059 |
+
"learning_rate": 2.3198447893569845e-05,
|
| 7060 |
+
"loss": 1.1840052604675293,
|
| 7061 |
+
"step": 1005
|
| 7062 |
+
},
|
| 7063 |
+
{
|
| 7064 |
+
"epoch": 0.5739552132363429,
|
| 7065 |
+
"grad_norm": 0.9296875,
|
| 7066 |
+
"learning_rate": 2.3170731707317075e-05,
|
| 7067 |
+
"loss": 1.1665326356887817,
|
| 7068 |
+
"step": 1006
|
| 7069 |
+
},
|
| 7070 |
+
{
|
| 7071 |
+
"epoch": 0.5745257452574526,
|
| 7072 |
+
"grad_norm": 0.99609375,
|
| 7073 |
+
"learning_rate": 2.3143015521064304e-05,
|
| 7074 |
+
"loss": 1.1994144916534424,
|
| 7075 |
+
"step": 1007
|
| 7076 |
+
},
|
| 7077 |
+
{
|
| 7078 |
+
"epoch": 0.5750962772785623,
|
| 7079 |
+
"grad_norm": 0.9921875,
|
| 7080 |
+
"learning_rate": 2.311529933481153e-05,
|
| 7081 |
+
"loss": 1.1023156642913818,
|
| 7082 |
+
"step": 1008
|
| 7083 |
+
},
|
| 7084 |
+
{
|
| 7085 |
+
"epoch": 0.575666809299672,
|
| 7086 |
+
"grad_norm": 0.9453125,
|
| 7087 |
+
"learning_rate": 2.308758314855876e-05,
|
| 7088 |
+
"loss": 1.2176637649536133,
|
| 7089 |
+
"step": 1009
|
| 7090 |
+
},
|
| 7091 |
+
{
|
| 7092 |
+
"epoch": 0.5762373413207816,
|
| 7093 |
+
"grad_norm": 1.0390625,
|
| 7094 |
+
"learning_rate": 2.3059866962305986e-05,
|
| 7095 |
+
"loss": 1.2663724422454834,
|
| 7096 |
+
"step": 1010
|
| 7097 |
+
},
|
| 7098 |
+
{
|
| 7099 |
+
"epoch": 0.5768078733418913,
|
| 7100 |
+
"grad_norm": 1.0,
|
| 7101 |
+
"learning_rate": 2.3032150776053215e-05,
|
| 7102 |
+
"loss": 1.1681220531463623,
|
| 7103 |
+
"step": 1011
|
| 7104 |
+
},
|
| 7105 |
+
{
|
| 7106 |
+
"epoch": 0.577378405363001,
|
| 7107 |
+
"grad_norm": 1.0078125,
|
| 7108 |
+
"learning_rate": 2.3004434589800445e-05,
|
| 7109 |
+
"loss": 1.221947431564331,
|
| 7110 |
+
"step": 1012
|
| 7111 |
+
},
|
| 7112 |
+
{
|
| 7113 |
+
"epoch": 0.5779489373841107,
|
| 7114 |
+
"grad_norm": 0.9921875,
|
| 7115 |
+
"learning_rate": 2.2976718403547674e-05,
|
| 7116 |
+
"loss": 1.1309971809387207,
|
| 7117 |
+
"step": 1013
|
| 7118 |
+
},
|
| 7119 |
+
{
|
| 7120 |
+
"epoch": 0.5785194694052204,
|
| 7121 |
+
"grad_norm": 0.98828125,
|
| 7122 |
+
"learning_rate": 2.29490022172949e-05,
|
| 7123 |
+
"loss": 1.1859217882156372,
|
| 7124 |
+
"step": 1014
|
| 7125 |
+
},
|
| 7126 |
+
{
|
| 7127 |
+
"epoch": 0.57909000142633,
|
| 7128 |
+
"grad_norm": 1.0,
|
| 7129 |
+
"learning_rate": 2.292128603104213e-05,
|
| 7130 |
+
"loss": 1.1979272365570068,
|
| 7131 |
+
"step": 1015
|
| 7132 |
+
},
|
| 7133 |
+
{
|
| 7134 |
+
"epoch": 0.5796605334474397,
|
| 7135 |
+
"grad_norm": 1.0,
|
| 7136 |
+
"learning_rate": 2.289356984478936e-05,
|
| 7137 |
+
"loss": 1.1865754127502441,
|
| 7138 |
+
"step": 1016
|
| 7139 |
+
},
|
| 7140 |
+
{
|
| 7141 |
+
"epoch": 0.5802310654685494,
|
| 7142 |
+
"grad_norm": 0.95703125,
|
| 7143 |
+
"learning_rate": 2.286585365853659e-05,
|
| 7144 |
+
"loss": 1.1868486404418945,
|
| 7145 |
+
"step": 1017
|
| 7146 |
+
},
|
| 7147 |
+
{
|
| 7148 |
+
"epoch": 0.5808015974896591,
|
| 7149 |
+
"grad_norm": 0.921875,
|
| 7150 |
+
"learning_rate": 2.2838137472283815e-05,
|
| 7151 |
+
"loss": 1.129669427871704,
|
| 7152 |
+
"step": 1018
|
| 7153 |
+
},
|
| 7154 |
+
{
|
| 7155 |
+
"epoch": 0.5813721295107688,
|
| 7156 |
+
"grad_norm": 0.9375,
|
| 7157 |
+
"learning_rate": 2.2810421286031044e-05,
|
| 7158 |
+
"loss": 1.1734843254089355,
|
| 7159 |
+
"step": 1019
|
| 7160 |
+
},
|
| 7161 |
+
{
|
| 7162 |
+
"epoch": 0.5819426615318785,
|
| 7163 |
+
"grad_norm": 1.0390625,
|
| 7164 |
+
"learning_rate": 2.278270509977827e-05,
|
| 7165 |
+
"loss": 1.2343952655792236,
|
| 7166 |
+
"step": 1020
|
| 7167 |
+
},
|
| 7168 |
+
{
|
| 7169 |
+
"epoch": 0.5825131935529881,
|
| 7170 |
+
"grad_norm": 0.96484375,
|
| 7171 |
+
"learning_rate": 2.27549889135255e-05,
|
| 7172 |
+
"loss": 1.21380615234375,
|
| 7173 |
+
"step": 1021
|
| 7174 |
+
},
|
| 7175 |
+
{
|
| 7176 |
+
"epoch": 0.5830837255740978,
|
| 7177 |
+
"grad_norm": 0.9296875,
|
| 7178 |
+
"learning_rate": 2.272727272727273e-05,
|
| 7179 |
+
"loss": 1.1312305927276611,
|
| 7180 |
+
"step": 1022
|
| 7181 |
+
},
|
| 7182 |
+
{
|
| 7183 |
+
"epoch": 0.5836542575952075,
|
| 7184 |
+
"grad_norm": 0.9609375,
|
| 7185 |
+
"learning_rate": 2.2699556541019955e-05,
|
| 7186 |
+
"loss": 1.1510472297668457,
|
| 7187 |
+
"step": 1023
|
| 7188 |
+
},
|
| 7189 |
+
{
|
| 7190 |
+
"epoch": 0.5842247896163172,
|
| 7191 |
+
"grad_norm": 1.0,
|
| 7192 |
+
"learning_rate": 2.2671840354767185e-05,
|
| 7193 |
+
"loss": 1.1997393369674683,
|
| 7194 |
+
"step": 1024
|
| 7195 |
+
},
|
| 7196 |
+
{
|
| 7197 |
+
"epoch": 0.5847953216374269,
|
| 7198 |
+
"grad_norm": 0.953125,
|
| 7199 |
+
"learning_rate": 2.2644124168514414e-05,
|
| 7200 |
+
"loss": 1.1844977140426636,
|
| 7201 |
+
"step": 1025
|
| 7202 |
+
},
|
| 7203 |
+
{
|
| 7204 |
+
"epoch": 0.5853658536585366,
|
| 7205 |
+
"grad_norm": 0.95703125,
|
| 7206 |
+
"learning_rate": 2.261640798226164e-05,
|
| 7207 |
+
"loss": 1.1642664670944214,
|
| 7208 |
+
"step": 1026
|
| 7209 |
+
},
|
| 7210 |
+
{
|
| 7211 |
+
"epoch": 0.5859363856796462,
|
| 7212 |
+
"grad_norm": 0.95703125,
|
| 7213 |
+
"learning_rate": 2.258869179600887e-05,
|
| 7214 |
+
"loss": 1.1929872035980225,
|
| 7215 |
+
"step": 1027
|
| 7216 |
+
},
|
| 7217 |
+
{
|
| 7218 |
+
"epoch": 0.5865069177007559,
|
| 7219 |
+
"grad_norm": 0.98046875,
|
| 7220 |
+
"learning_rate": 2.25609756097561e-05,
|
| 7221 |
+
"loss": 1.2264790534973145,
|
| 7222 |
+
"step": 1028
|
| 7223 |
+
},
|
| 7224 |
+
{
|
| 7225 |
+
"epoch": 0.5870774497218656,
|
| 7226 |
+
"grad_norm": 0.96484375,
|
| 7227 |
+
"learning_rate": 2.253325942350333e-05,
|
| 7228 |
+
"loss": 1.208320140838623,
|
| 7229 |
+
"step": 1029
|
| 7230 |
+
},
|
| 7231 |
+
{
|
| 7232 |
+
"epoch": 0.5876479817429753,
|
| 7233 |
+
"grad_norm": 0.9140625,
|
| 7234 |
+
"learning_rate": 2.2505543237250555e-05,
|
| 7235 |
+
"loss": 1.1017545461654663,
|
| 7236 |
+
"step": 1030
|
| 7237 |
+
},
|
| 7238 |
+
{
|
| 7239 |
+
"epoch": 0.588218513764085,
|
| 7240 |
+
"grad_norm": 0.91015625,
|
| 7241 |
+
"learning_rate": 2.2477827050997784e-05,
|
| 7242 |
+
"loss": 1.0866947174072266,
|
| 7243 |
+
"step": 1031
|
| 7244 |
+
},
|
| 7245 |
+
{
|
| 7246 |
+
"epoch": 0.5887890457851946,
|
| 7247 |
+
"grad_norm": 0.94140625,
|
| 7248 |
+
"learning_rate": 2.245011086474501e-05,
|
| 7249 |
+
"loss": 1.134414553642273,
|
| 7250 |
+
"step": 1032
|
| 7251 |
+
},
|
| 7252 |
+
{
|
| 7253 |
+
"epoch": 0.5893595778063043,
|
| 7254 |
+
"grad_norm": 0.9609375,
|
| 7255 |
+
"learning_rate": 2.242239467849224e-05,
|
| 7256 |
+
"loss": 1.1386680603027344,
|
| 7257 |
+
"step": 1033
|
| 7258 |
+
},
|
| 7259 |
+
{
|
| 7260 |
+
"epoch": 0.5899301098274141,
|
| 7261 |
+
"grad_norm": 0.93359375,
|
| 7262 |
+
"learning_rate": 2.239467849223947e-05,
|
| 7263 |
+
"loss": 1.098857045173645,
|
| 7264 |
+
"step": 1034
|
| 7265 |
+
},
|
| 7266 |
+
{
|
| 7267 |
+
"epoch": 0.5905006418485238,
|
| 7268 |
+
"grad_norm": 0.98046875,
|
| 7269 |
+
"learning_rate": 2.2366962305986695e-05,
|
| 7270 |
+
"loss": 1.1710071563720703,
|
| 7271 |
+
"step": 1035
|
| 7272 |
+
},
|
| 7273 |
+
{
|
| 7274 |
+
"epoch": 0.5910711738696335,
|
| 7275 |
+
"grad_norm": 0.90234375,
|
| 7276 |
+
"learning_rate": 2.2339246119733925e-05,
|
| 7277 |
+
"loss": 1.1196489334106445,
|
| 7278 |
+
"step": 1036
|
| 7279 |
+
},
|
| 7280 |
+
{
|
| 7281 |
+
"epoch": 0.5916417058907432,
|
| 7282 |
+
"grad_norm": 0.94140625,
|
| 7283 |
+
"learning_rate": 2.2311529933481154e-05,
|
| 7284 |
+
"loss": 1.132148265838623,
|
| 7285 |
+
"step": 1037
|
| 7286 |
+
},
|
| 7287 |
+
{
|
| 7288 |
+
"epoch": 0.5922122379118528,
|
| 7289 |
+
"grad_norm": 0.9453125,
|
| 7290 |
+
"learning_rate": 2.2283813747228384e-05,
|
| 7291 |
+
"loss": 1.1694618463516235,
|
| 7292 |
+
"step": 1038
|
| 7293 |
+
},
|
| 7294 |
+
{
|
| 7295 |
+
"epoch": 0.5927827699329625,
|
| 7296 |
+
"grad_norm": 0.94140625,
|
| 7297 |
+
"learning_rate": 2.225609756097561e-05,
|
| 7298 |
+
"loss": 1.141546607017517,
|
| 7299 |
+
"step": 1039
|
| 7300 |
+
},
|
| 7301 |
+
{
|
| 7302 |
+
"epoch": 0.5933533019540722,
|
| 7303 |
+
"grad_norm": 1.015625,
|
| 7304 |
+
"learning_rate": 2.222838137472284e-05,
|
| 7305 |
+
"loss": 1.214141607284546,
|
| 7306 |
+
"step": 1040
|
| 7307 |
+
},
|
| 7308 |
+
{
|
| 7309 |
+
"epoch": 0.5939238339751819,
|
| 7310 |
+
"grad_norm": 0.9375,
|
| 7311 |
+
"learning_rate": 2.220066518847007e-05,
|
| 7312 |
+
"loss": 1.142057180404663,
|
| 7313 |
+
"step": 1041
|
| 7314 |
+
},
|
| 7315 |
+
{
|
| 7316 |
+
"epoch": 0.5944943659962916,
|
| 7317 |
+
"grad_norm": 0.9609375,
|
| 7318 |
+
"learning_rate": 2.2172949002217298e-05,
|
| 7319 |
+
"loss": 1.1707711219787598,
|
| 7320 |
+
"step": 1042
|
| 7321 |
+
},
|
| 7322 |
+
{
|
| 7323 |
+
"epoch": 0.5950648980174013,
|
| 7324 |
+
"grad_norm": 0.91796875,
|
| 7325 |
+
"learning_rate": 2.2145232815964524e-05,
|
| 7326 |
+
"loss": 1.164795994758606,
|
| 7327 |
+
"step": 1043
|
| 7328 |
+
},
|
| 7329 |
+
{
|
| 7330 |
+
"epoch": 0.5956354300385109,
|
| 7331 |
+
"grad_norm": 0.97265625,
|
| 7332 |
+
"learning_rate": 2.211751662971175e-05,
|
| 7333 |
+
"loss": 1.1659691333770752,
|
| 7334 |
+
"step": 1044
|
| 7335 |
+
},
|
| 7336 |
+
{
|
| 7337 |
+
"epoch": 0.5962059620596206,
|
| 7338 |
+
"grad_norm": 0.94921875,
|
| 7339 |
+
"learning_rate": 2.208980044345898e-05,
|
| 7340 |
+
"loss": 1.1294951438903809,
|
| 7341 |
+
"step": 1045
|
| 7342 |
+
},
|
| 7343 |
+
{
|
| 7344 |
+
"epoch": 0.5967764940807303,
|
| 7345 |
+
"grad_norm": 0.96875,
|
| 7346 |
+
"learning_rate": 2.206208425720621e-05,
|
| 7347 |
+
"loss": 1.1925092935562134,
|
| 7348 |
+
"step": 1046
|
| 7349 |
+
},
|
| 7350 |
+
{
|
| 7351 |
+
"epoch": 0.59734702610184,
|
| 7352 |
+
"grad_norm": 0.93359375,
|
| 7353 |
+
"learning_rate": 2.203436807095344e-05,
|
| 7354 |
+
"loss": 1.1600418090820312,
|
| 7355 |
+
"step": 1047
|
| 7356 |
+
},
|
| 7357 |
+
{
|
| 7358 |
+
"epoch": 0.5979175581229497,
|
| 7359 |
+
"grad_norm": 0.98046875,
|
| 7360 |
+
"learning_rate": 2.2006651884700665e-05,
|
| 7361 |
+
"loss": 1.157020092010498,
|
| 7362 |
+
"step": 1048
|
| 7363 |
+
},
|
| 7364 |
+
{
|
| 7365 |
+
"epoch": 0.5984880901440593,
|
| 7366 |
+
"grad_norm": 1.0234375,
|
| 7367 |
+
"learning_rate": 2.1978935698447894e-05,
|
| 7368 |
+
"loss": 1.1589795351028442,
|
| 7369 |
+
"step": 1049
|
| 7370 |
+
},
|
| 7371 |
+
{
|
| 7372 |
+
"epoch": 0.599058622165169,
|
| 7373 |
+
"grad_norm": 0.9453125,
|
| 7374 |
+
"learning_rate": 2.1951219512195124e-05,
|
| 7375 |
+
"loss": 1.1546876430511475,
|
| 7376 |
+
"step": 1050
|
| 7377 |
+
},
|
| 7378 |
+
{
|
| 7379 |
+
"epoch": 0.5996291541862787,
|
| 7380 |
+
"grad_norm": 0.94140625,
|
| 7381 |
+
"learning_rate": 2.1923503325942353e-05,
|
| 7382 |
+
"loss": 1.1549787521362305,
|
| 7383 |
+
"step": 1051
|
| 7384 |
+
},
|
| 7385 |
+
{
|
| 7386 |
+
"epoch": 0.6001996862073884,
|
| 7387 |
+
"grad_norm": 0.9921875,
|
| 7388 |
+
"learning_rate": 2.189578713968958e-05,
|
| 7389 |
+
"loss": 1.1518681049346924,
|
| 7390 |
+
"step": 1052
|
| 7391 |
+
},
|
| 7392 |
+
{
|
| 7393 |
+
"epoch": 0.6007702182284981,
|
| 7394 |
+
"grad_norm": 0.96484375,
|
| 7395 |
+
"learning_rate": 2.186807095343681e-05,
|
| 7396 |
+
"loss": 1.1609306335449219,
|
| 7397 |
+
"step": 1053
|
| 7398 |
+
},
|
| 7399 |
+
{
|
| 7400 |
+
"epoch": 0.6013407502496078,
|
| 7401 |
+
"grad_norm": 0.97265625,
|
| 7402 |
+
"learning_rate": 2.1840354767184038e-05,
|
| 7403 |
+
"loss": 1.1526927947998047,
|
| 7404 |
+
"step": 1054
|
| 7405 |
+
},
|
| 7406 |
+
{
|
| 7407 |
+
"epoch": 0.6019112822707174,
|
| 7408 |
+
"grad_norm": 0.98046875,
|
| 7409 |
+
"learning_rate": 2.1812638580931268e-05,
|
| 7410 |
+
"loss": 1.2030518054962158,
|
| 7411 |
+
"step": 1055
|
| 7412 |
+
},
|
| 7413 |
+
{
|
| 7414 |
+
"epoch": 0.6024818142918271,
|
| 7415 |
+
"grad_norm": 0.94921875,
|
| 7416 |
+
"learning_rate": 2.178492239467849e-05,
|
| 7417 |
+
"loss": 1.087314248085022,
|
| 7418 |
+
"step": 1056
|
| 7419 |
+
},
|
| 7420 |
+
{
|
| 7421 |
+
"epoch": 0.6030523463129368,
|
| 7422 |
+
"grad_norm": 0.93359375,
|
| 7423 |
+
"learning_rate": 2.175720620842572e-05,
|
| 7424 |
+
"loss": 1.120784044265747,
|
| 7425 |
+
"step": 1057
|
| 7426 |
+
},
|
| 7427 |
+
{
|
| 7428 |
+
"epoch": 0.6036228783340465,
|
| 7429 |
+
"grad_norm": 0.921875,
|
| 7430 |
+
"learning_rate": 2.172949002217295e-05,
|
| 7431 |
+
"loss": 1.0867156982421875,
|
| 7432 |
+
"step": 1058
|
| 7433 |
+
},
|
| 7434 |
+
{
|
| 7435 |
+
"epoch": 0.6041934103551562,
|
| 7436 |
+
"grad_norm": 0.96484375,
|
| 7437 |
+
"learning_rate": 2.170177383592018e-05,
|
| 7438 |
+
"loss": 1.2083582878112793,
|
| 7439 |
+
"step": 1059
|
| 7440 |
+
},
|
| 7441 |
+
{
|
| 7442 |
+
"epoch": 0.6047639423762659,
|
| 7443 |
+
"grad_norm": 0.94921875,
|
| 7444 |
+
"learning_rate": 2.1674057649667405e-05,
|
| 7445 |
+
"loss": 1.1944574117660522,
|
| 7446 |
+
"step": 1060
|
| 7447 |
+
},
|
| 7448 |
+
{
|
| 7449 |
+
"epoch": 0.6053344743973755,
|
| 7450 |
+
"grad_norm": 0.92578125,
|
| 7451 |
+
"learning_rate": 2.1646341463414634e-05,
|
| 7452 |
+
"loss": 1.118787169456482,
|
| 7453 |
+
"step": 1061
|
| 7454 |
+
},
|
| 7455 |
+
{
|
| 7456 |
+
"epoch": 0.6059050064184852,
|
| 7457 |
+
"grad_norm": 0.94921875,
|
| 7458 |
+
"learning_rate": 2.1618625277161864e-05,
|
| 7459 |
+
"loss": 1.1591801643371582,
|
| 7460 |
+
"step": 1062
|
| 7461 |
+
},
|
| 7462 |
+
{
|
| 7463 |
+
"epoch": 0.6064755384395949,
|
| 7464 |
+
"grad_norm": 0.95703125,
|
| 7465 |
+
"learning_rate": 2.1590909090909093e-05,
|
| 7466 |
+
"loss": 1.1802964210510254,
|
| 7467 |
+
"step": 1063
|
| 7468 |
+
},
|
| 7469 |
+
{
|
| 7470 |
+
"epoch": 0.6070460704607046,
|
| 7471 |
+
"grad_norm": 0.97265625,
|
| 7472 |
+
"learning_rate": 2.156319290465632e-05,
|
| 7473 |
+
"loss": 1.1993342638015747,
|
| 7474 |
+
"step": 1064
|
| 7475 |
+
},
|
| 7476 |
+
{
|
| 7477 |
+
"epoch": 0.6076166024818143,
|
| 7478 |
+
"grad_norm": 0.96484375,
|
| 7479 |
+
"learning_rate": 2.153547671840355e-05,
|
| 7480 |
+
"loss": 1.2244541645050049,
|
| 7481 |
+
"step": 1065
|
| 7482 |
+
},
|
| 7483 |
+
{
|
| 7484 |
+
"epoch": 0.6081871345029239,
|
| 7485 |
+
"grad_norm": 0.9375,
|
| 7486 |
+
"learning_rate": 2.150776053215078e-05,
|
| 7487 |
+
"loss": 1.1696969270706177,
|
| 7488 |
+
"step": 1066
|
| 7489 |
+
},
|
| 7490 |
+
{
|
| 7491 |
+
"epoch": 0.6087576665240336,
|
| 7492 |
+
"grad_norm": 0.9609375,
|
| 7493 |
+
"learning_rate": 2.1480044345898008e-05,
|
| 7494 |
+
"loss": 1.204698085784912,
|
| 7495 |
+
"step": 1067
|
| 7496 |
+
},
|
| 7497 |
+
{
|
| 7498 |
+
"epoch": 0.6093281985451433,
|
| 7499 |
+
"grad_norm": 0.96875,
|
| 7500 |
+
"learning_rate": 2.1452328159645234e-05,
|
| 7501 |
+
"loss": 1.167772650718689,
|
| 7502 |
+
"step": 1068
|
| 7503 |
+
},
|
| 7504 |
+
{
|
| 7505 |
+
"epoch": 0.609898730566253,
|
| 7506 |
+
"grad_norm": 0.93359375,
|
| 7507 |
+
"learning_rate": 2.142461197339246e-05,
|
| 7508 |
+
"loss": 1.1064563989639282,
|
| 7509 |
+
"step": 1069
|
| 7510 |
+
},
|
| 7511 |
+
{
|
| 7512 |
+
"epoch": 0.6104692625873627,
|
| 7513 |
+
"grad_norm": 0.9296875,
|
| 7514 |
+
"learning_rate": 2.139689578713969e-05,
|
| 7515 |
+
"loss": 1.1095709800720215,
|
| 7516 |
+
"step": 1070
|
| 7517 |
+
},
|
| 7518 |
+
{
|
| 7519 |
+
"epoch": 0.6110397946084724,
|
| 7520 |
+
"grad_norm": 0.953125,
|
| 7521 |
+
"learning_rate": 2.136917960088692e-05,
|
| 7522 |
+
"loss": 1.1526896953582764,
|
| 7523 |
+
"step": 1071
|
| 7524 |
+
},
|
| 7525 |
+
{
|
| 7526 |
+
"epoch": 0.611610326629582,
|
| 7527 |
+
"grad_norm": 0.98828125,
|
| 7528 |
+
"learning_rate": 2.134146341463415e-05,
|
| 7529 |
+
"loss": 1.1842620372772217,
|
| 7530 |
+
"step": 1072
|
| 7531 |
+
},
|
| 7532 |
+
{
|
| 7533 |
+
"epoch": 0.6121808586506917,
|
| 7534 |
+
"grad_norm": 0.96484375,
|
| 7535 |
+
"learning_rate": 2.1313747228381374e-05,
|
| 7536 |
+
"loss": 1.1854032278060913,
|
| 7537 |
+
"step": 1073
|
| 7538 |
+
},
|
| 7539 |
+
{
|
| 7540 |
+
"epoch": 0.6127513906718015,
|
| 7541 |
+
"grad_norm": 0.94140625,
|
| 7542 |
+
"learning_rate": 2.1286031042128604e-05,
|
| 7543 |
+
"loss": 1.1536649465560913,
|
| 7544 |
+
"step": 1074
|
| 7545 |
+
},
|
| 7546 |
+
{
|
| 7547 |
+
"epoch": 0.6133219226929112,
|
| 7548 |
+
"grad_norm": 0.99609375,
|
| 7549 |
+
"learning_rate": 2.1258314855875833e-05,
|
| 7550 |
+
"loss": 1.162165641784668,
|
| 7551 |
+
"step": 1075
|
| 7552 |
+
},
|
| 7553 |
+
{
|
| 7554 |
+
"epoch": 0.6138924547140209,
|
| 7555 |
+
"grad_norm": 0.95703125,
|
| 7556 |
+
"learning_rate": 2.1230598669623063e-05,
|
| 7557 |
+
"loss": 1.1589579582214355,
|
| 7558 |
+
"step": 1076
|
| 7559 |
+
},
|
| 7560 |
+
{
|
| 7561 |
+
"epoch": 0.6144629867351306,
|
| 7562 |
+
"grad_norm": 0.99609375,
|
| 7563 |
+
"learning_rate": 2.120288248337029e-05,
|
| 7564 |
+
"loss": 1.2380765676498413,
|
| 7565 |
+
"step": 1077
|
| 7566 |
+
},
|
| 7567 |
+
{
|
| 7568 |
+
"epoch": 0.6150335187562402,
|
| 7569 |
+
"grad_norm": 0.9921875,
|
| 7570 |
+
"learning_rate": 2.117516629711752e-05,
|
| 7571 |
+
"loss": 1.1789859533309937,
|
| 7572 |
+
"step": 1078
|
| 7573 |
+
},
|
| 7574 |
+
{
|
| 7575 |
+
"epoch": 0.6156040507773499,
|
| 7576 |
+
"grad_norm": 0.92578125,
|
| 7577 |
+
"learning_rate": 2.1147450110864748e-05,
|
| 7578 |
+
"loss": 1.1379293203353882,
|
| 7579 |
+
"step": 1079
|
| 7580 |
+
},
|
| 7581 |
+
{
|
| 7582 |
+
"epoch": 0.6161745827984596,
|
| 7583 |
+
"grad_norm": 0.984375,
|
| 7584 |
+
"learning_rate": 2.1119733924611977e-05,
|
| 7585 |
+
"loss": 1.176946759223938,
|
| 7586 |
+
"step": 1080
|
| 7587 |
+
},
|
| 7588 |
+
{
|
| 7589 |
+
"epoch": 0.6167451148195693,
|
| 7590 |
+
"grad_norm": 0.98046875,
|
| 7591 |
+
"learning_rate": 2.10920177383592e-05,
|
| 7592 |
+
"loss": 1.232793927192688,
|
| 7593 |
+
"step": 1081
|
| 7594 |
+
},
|
| 7595 |
+
{
|
| 7596 |
+
"epoch": 0.617315646840679,
|
| 7597 |
+
"grad_norm": 0.94140625,
|
| 7598 |
+
"learning_rate": 2.106430155210643e-05,
|
| 7599 |
+
"loss": 1.1333751678466797,
|
| 7600 |
+
"step": 1082
|
| 7601 |
+
},
|
| 7602 |
+
{
|
| 7603 |
+
"epoch": 0.6178861788617886,
|
| 7604 |
+
"grad_norm": 0.98046875,
|
| 7605 |
+
"learning_rate": 2.103658536585366e-05,
|
| 7606 |
+
"loss": 1.1847493648529053,
|
| 7607 |
+
"step": 1083
|
| 7608 |
+
},
|
| 7609 |
+
{
|
| 7610 |
+
"epoch": 0.6184567108828983,
|
| 7611 |
+
"grad_norm": 0.98828125,
|
| 7612 |
+
"learning_rate": 2.100886917960089e-05,
|
| 7613 |
+
"loss": 1.1365629434585571,
|
| 7614 |
+
"step": 1084
|
| 7615 |
+
},
|
| 7616 |
+
{
|
| 7617 |
+
"epoch": 0.619027242904008,
|
| 7618 |
+
"grad_norm": 0.9609375,
|
| 7619 |
+
"learning_rate": 2.0981152993348114e-05,
|
| 7620 |
+
"loss": 1.1531561613082886,
|
| 7621 |
+
"step": 1085
|
| 7622 |
+
},
|
| 7623 |
+
{
|
| 7624 |
+
"epoch": 0.6195977749251177,
|
| 7625 |
+
"grad_norm": 0.9765625,
|
| 7626 |
+
"learning_rate": 2.0953436807095344e-05,
|
| 7627 |
+
"loss": 1.1419352293014526,
|
| 7628 |
+
"step": 1086
|
| 7629 |
+
},
|
| 7630 |
+
{
|
| 7631 |
+
"epoch": 0.6201683069462274,
|
| 7632 |
+
"grad_norm": 0.95703125,
|
| 7633 |
+
"learning_rate": 2.0925720620842573e-05,
|
| 7634 |
+
"loss": 1.2071990966796875,
|
| 7635 |
+
"step": 1087
|
| 7636 |
+
},
|
| 7637 |
+
{
|
| 7638 |
+
"epoch": 0.6207388389673371,
|
| 7639 |
+
"grad_norm": 1.0078125,
|
| 7640 |
+
"learning_rate": 2.0898004434589803e-05,
|
| 7641 |
+
"loss": 1.146884799003601,
|
| 7642 |
+
"step": 1088
|
| 7643 |
+
},
|
| 7644 |
+
{
|
| 7645 |
+
"epoch": 0.6213093709884467,
|
| 7646 |
+
"grad_norm": 1.0,
|
| 7647 |
+
"learning_rate": 2.087028824833703e-05,
|
| 7648 |
+
"loss": 1.1956453323364258,
|
| 7649 |
+
"step": 1089
|
| 7650 |
+
},
|
| 7651 |
+
{
|
| 7652 |
+
"epoch": 0.6218799030095564,
|
| 7653 |
+
"grad_norm": 0.97265625,
|
| 7654 |
+
"learning_rate": 2.084257206208426e-05,
|
| 7655 |
+
"loss": 1.182574987411499,
|
| 7656 |
+
"step": 1090
|
| 7657 |
+
},
|
| 7658 |
+
{
|
| 7659 |
+
"epoch": 0.6224504350306661,
|
| 7660 |
+
"grad_norm": 0.9765625,
|
| 7661 |
+
"learning_rate": 2.0814855875831488e-05,
|
| 7662 |
+
"loss": 1.1805145740509033,
|
| 7663 |
+
"step": 1091
|
| 7664 |
+
},
|
| 7665 |
+
{
|
| 7666 |
+
"epoch": 0.6230209670517758,
|
| 7667 |
+
"grad_norm": 0.96484375,
|
| 7668 |
+
"learning_rate": 2.0787139689578717e-05,
|
| 7669 |
+
"loss": 1.173978567123413,
|
| 7670 |
+
"step": 1092
|
| 7671 |
+
},
|
| 7672 |
+
{
|
| 7673 |
+
"epoch": 0.6235914990728855,
|
| 7674 |
+
"grad_norm": 0.9375,
|
| 7675 |
+
"learning_rate": 2.0759423503325943e-05,
|
| 7676 |
+
"loss": 1.1732361316680908,
|
| 7677 |
+
"step": 1093
|
| 7678 |
+
},
|
| 7679 |
+
{
|
| 7680 |
+
"epoch": 0.6241620310939952,
|
| 7681 |
+
"grad_norm": 0.94921875,
|
| 7682 |
+
"learning_rate": 2.073170731707317e-05,
|
| 7683 |
+
"loss": 1.1978164911270142,
|
| 7684 |
+
"step": 1094
|
| 7685 |
+
},
|
| 7686 |
+
{
|
| 7687 |
+
"epoch": 0.6247325631151048,
|
| 7688 |
+
"grad_norm": 0.96484375,
|
| 7689 |
+
"learning_rate": 2.07039911308204e-05,
|
| 7690 |
+
"loss": 1.161289930343628,
|
| 7691 |
+
"step": 1095
|
| 7692 |
+
},
|
| 7693 |
+
{
|
| 7694 |
+
"epoch": 0.6253030951362145,
|
| 7695 |
+
"grad_norm": 0.953125,
|
| 7696 |
+
"learning_rate": 2.067627494456763e-05,
|
| 7697 |
+
"loss": 1.1583458185195923,
|
| 7698 |
+
"step": 1096
|
| 7699 |
+
},
|
| 7700 |
+
{
|
| 7701 |
+
"epoch": 0.6258736271573242,
|
| 7702 |
+
"grad_norm": 0.9765625,
|
| 7703 |
+
"learning_rate": 2.0648558758314858e-05,
|
| 7704 |
+
"loss": 1.1835911273956299,
|
| 7705 |
+
"step": 1097
|
| 7706 |
+
},
|
| 7707 |
+
{
|
| 7708 |
+
"epoch": 0.6264441591784339,
|
| 7709 |
+
"grad_norm": 0.9921875,
|
| 7710 |
+
"learning_rate": 2.0620842572062084e-05,
|
| 7711 |
+
"loss": 1.1692794561386108,
|
| 7712 |
+
"step": 1098
|
| 7713 |
+
},
|
| 7714 |
+
{
|
| 7715 |
+
"epoch": 0.6270146911995436,
|
| 7716 |
+
"grad_norm": 0.97265625,
|
| 7717 |
+
"learning_rate": 2.0593126385809313e-05,
|
| 7718 |
+
"loss": 1.1748257875442505,
|
| 7719 |
+
"step": 1099
|
| 7720 |
+
},
|
| 7721 |
+
{
|
| 7722 |
+
"epoch": 0.6275852232206532,
|
| 7723 |
+
"grad_norm": 0.9765625,
|
| 7724 |
+
"learning_rate": 2.0565410199556543e-05,
|
| 7725 |
+
"loss": 1.172876238822937,
|
| 7726 |
+
"step": 1100
|
| 7727 |
+
},
|
| 7728 |
+
{
|
| 7729 |
+
"epoch": 0.6281557552417629,
|
| 7730 |
+
"grad_norm": 1.0078125,
|
| 7731 |
+
"learning_rate": 2.0537694013303772e-05,
|
| 7732 |
+
"loss": 1.1829420328140259,
|
| 7733 |
+
"step": 1101
|
| 7734 |
+
},
|
| 7735 |
+
{
|
| 7736 |
+
"epoch": 0.6287262872628726,
|
| 7737 |
+
"grad_norm": 0.9375,
|
| 7738 |
+
"learning_rate": 2.0509977827051e-05,
|
| 7739 |
+
"loss": 1.163160800933838,
|
| 7740 |
+
"step": 1102
|
| 7741 |
+
},
|
| 7742 |
+
{
|
| 7743 |
+
"epoch": 0.6292968192839823,
|
| 7744 |
+
"grad_norm": 0.96484375,
|
| 7745 |
+
"learning_rate": 2.0482261640798228e-05,
|
| 7746 |
+
"loss": 1.144565463066101,
|
| 7747 |
+
"step": 1103
|
| 7748 |
+
},
|
| 7749 |
+
{
|
| 7750 |
+
"epoch": 0.629867351305092,
|
| 7751 |
+
"grad_norm": 0.953125,
|
| 7752 |
+
"learning_rate": 2.0454545454545457e-05,
|
| 7753 |
+
"loss": 1.1199369430541992,
|
| 7754 |
+
"step": 1104
|
| 7755 |
+
},
|
| 7756 |
+
{
|
| 7757 |
+
"epoch": 0.6304378833262017,
|
| 7758 |
+
"grad_norm": 0.9765625,
|
| 7759 |
+
"learning_rate": 2.0426829268292683e-05,
|
| 7760 |
+
"loss": 1.1951239109039307,
|
| 7761 |
+
"step": 1105
|
| 7762 |
+
},
|
| 7763 |
+
{
|
| 7764 |
+
"epoch": 0.6310084153473113,
|
| 7765 |
+
"grad_norm": 0.96484375,
|
| 7766 |
+
"learning_rate": 2.0399113082039913e-05,
|
| 7767 |
+
"loss": 1.1440958976745605,
|
| 7768 |
+
"step": 1106
|
| 7769 |
+
},
|
| 7770 |
+
{
|
| 7771 |
+
"epoch": 0.631578947368421,
|
| 7772 |
+
"grad_norm": 0.94140625,
|
| 7773 |
+
"learning_rate": 2.037139689578714e-05,
|
| 7774 |
+
"loss": 1.1329402923583984,
|
| 7775 |
+
"step": 1107
|
| 7776 |
+
},
|
| 7777 |
+
{
|
| 7778 |
+
"epoch": 0.631578947368421,
|
| 7779 |
+
"eval_loss": 1.1687453985214233,
|
| 7780 |
+
"eval_runtime": 80.1565,
|
| 7781 |
+
"eval_samples_per_second": 11.927,
|
| 7782 |
+
"eval_steps_per_second": 2.982,
|
| 7783 |
+
"step": 1107
|
| 7784 |
}
|
| 7785 |
],
|
| 7786 |
"logging_steps": 1,
|
|
|
|
| 7800 |
"attributes": {}
|
| 7801 |
}
|
| 7802 |
},
|
| 7803 |
+
"total_flos": 9.75350724523733e+17,
|
| 7804 |
"train_batch_size": 8,
|
| 7805 |
"trial_name": null,
|
| 7806 |
"trial_params": null
|