Training in progress, step 1464, 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:d8e5b1075c740bf718f5cb6a5fe7dbba07216a2ebe0d29fad75adeb68908ccc0
|
| 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 sha256:3c4cac79f054b9b90a9dbe3d71344698b44883ffe8d2df6f5ae633d917746406
|
| 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:d569a9c96ff4e89d32dfe8ca0b97952cc2edcd76f89c14caa7e57238c4b46c8a
|
| 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 sha256:191bb8328007897e523a3d4a5522dce5544f51821431d7a2777792d9f33b45b6
|
| 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": 488,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
@@ -6856,6 +6856,3430 @@
|
|
| 6856 |
"eval_samples_per_second": 77.928,
|
| 6857 |
"eval_steps_per_second": 19.482,
|
| 6858 |
"step": 976
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 6859 |
}
|
| 6860 |
],
|
| 6861 |
"logging_steps": 1,
|
|
@@ -6875,7 +10299,7 @@
|
|
| 6875 |
"attributes": {}
|
| 6876 |
}
|
| 6877 |
},
|
| 6878 |
-
"total_flos":
|
| 6879 |
"train_batch_size": 4,
|
| 6880 |
"trial_name": null,
|
| 6881 |
"trial_params": null
|
|
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.6006153846153847,
|
| 6 |
"eval_steps": 488,
|
| 7 |
+
"global_step": 1464,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 6856 |
"eval_samples_per_second": 77.928,
|
| 6857 |
"eval_steps_per_second": 19.482,
|
| 6858 |
"step": 976
|
| 6859 |
+
},
|
| 6860 |
+
{
|
| 6861 |
+
"epoch": 0.40082051282051284,
|
| 6862 |
+
"grad_norm": 3.953125,
|
| 6863 |
+
"learning_rate": 3.0598576810380913e-06,
|
| 6864 |
+
"loss": 1.1163733005523682,
|
| 6865 |
+
"step": 977
|
| 6866 |
+
},
|
| 6867 |
+
{
|
| 6868 |
+
"epoch": 0.4012307692307692,
|
| 6869 |
+
"grad_norm": 4.375,
|
| 6870 |
+
"learning_rate": 3.0577647551276685e-06,
|
| 6871 |
+
"loss": 1.3030307292938232,
|
| 6872 |
+
"step": 978
|
| 6873 |
+
},
|
| 6874 |
+
{
|
| 6875 |
+
"epoch": 0.40164102564102566,
|
| 6876 |
+
"grad_norm": 4.28125,
|
| 6877 |
+
"learning_rate": 3.055671829217246e-06,
|
| 6878 |
+
"loss": 1.1708984375,
|
| 6879 |
+
"step": 979
|
| 6880 |
+
},
|
| 6881 |
+
{
|
| 6882 |
+
"epoch": 0.40205128205128204,
|
| 6883 |
+
"grad_norm": 3.921875,
|
| 6884 |
+
"learning_rate": 3.053578903306823e-06,
|
| 6885 |
+
"loss": 1.2434278726577759,
|
| 6886 |
+
"step": 980
|
| 6887 |
+
},
|
| 6888 |
+
{
|
| 6889 |
+
"epoch": 0.4024615384615385,
|
| 6890 |
+
"grad_norm": 3.8125,
|
| 6891 |
+
"learning_rate": 3.0514859773964006e-06,
|
| 6892 |
+
"loss": 1.2517290115356445,
|
| 6893 |
+
"step": 981
|
| 6894 |
+
},
|
| 6895 |
+
{
|
| 6896 |
+
"epoch": 0.40287179487179486,
|
| 6897 |
+
"grad_norm": 4.15625,
|
| 6898 |
+
"learning_rate": 3.049393051485978e-06,
|
| 6899 |
+
"loss": 1.2692155838012695,
|
| 6900 |
+
"step": 982
|
| 6901 |
+
},
|
| 6902 |
+
{
|
| 6903 |
+
"epoch": 0.4032820512820513,
|
| 6904 |
+
"grad_norm": 4.46875,
|
| 6905 |
+
"learning_rate": 3.0473001255755546e-06,
|
| 6906 |
+
"loss": 1.2814083099365234,
|
| 6907 |
+
"step": 983
|
| 6908 |
+
},
|
| 6909 |
+
{
|
| 6910 |
+
"epoch": 0.4036923076923077,
|
| 6911 |
+
"grad_norm": 4.375,
|
| 6912 |
+
"learning_rate": 3.0452071996651322e-06,
|
| 6913 |
+
"loss": 1.152833342552185,
|
| 6914 |
+
"step": 984
|
| 6915 |
+
},
|
| 6916 |
+
{
|
| 6917 |
+
"epoch": 0.4041025641025641,
|
| 6918 |
+
"grad_norm": 3.34375,
|
| 6919 |
+
"learning_rate": 3.0431142737547095e-06,
|
| 6920 |
+
"loss": 1.1034908294677734,
|
| 6921 |
+
"step": 985
|
| 6922 |
+
},
|
| 6923 |
+
{
|
| 6924 |
+
"epoch": 0.4045128205128205,
|
| 6925 |
+
"grad_norm": 4.375,
|
| 6926 |
+
"learning_rate": 3.0410213478442867e-06,
|
| 6927 |
+
"loss": 1.4281508922576904,
|
| 6928 |
+
"step": 986
|
| 6929 |
+
},
|
| 6930 |
+
{
|
| 6931 |
+
"epoch": 0.40492307692307694,
|
| 6932 |
+
"grad_norm": 4.15625,
|
| 6933 |
+
"learning_rate": 3.038928421933864e-06,
|
| 6934 |
+
"loss": 1.1934211254119873,
|
| 6935 |
+
"step": 987
|
| 6936 |
+
},
|
| 6937 |
+
{
|
| 6938 |
+
"epoch": 0.4053333333333333,
|
| 6939 |
+
"grad_norm": 4.15625,
|
| 6940 |
+
"learning_rate": 3.0368354960234407e-06,
|
| 6941 |
+
"loss": 1.07750403881073,
|
| 6942 |
+
"step": 988
|
| 6943 |
+
},
|
| 6944 |
+
{
|
| 6945 |
+
"epoch": 0.40574358974358976,
|
| 6946 |
+
"grad_norm": 4.125,
|
| 6947 |
+
"learning_rate": 3.0347425701130183e-06,
|
| 6948 |
+
"loss": 1.1173770427703857,
|
| 6949 |
+
"step": 989
|
| 6950 |
+
},
|
| 6951 |
+
{
|
| 6952 |
+
"epoch": 0.40615384615384614,
|
| 6953 |
+
"grad_norm": 3.78125,
|
| 6954 |
+
"learning_rate": 3.0326496442025955e-06,
|
| 6955 |
+
"loss": 1.1600890159606934,
|
| 6956 |
+
"step": 990
|
| 6957 |
+
},
|
| 6958 |
+
{
|
| 6959 |
+
"epoch": 0.4065641025641026,
|
| 6960 |
+
"grad_norm": 3.78125,
|
| 6961 |
+
"learning_rate": 3.0305567182921723e-06,
|
| 6962 |
+
"loss": 1.2492793798446655,
|
| 6963 |
+
"step": 991
|
| 6964 |
+
},
|
| 6965 |
+
{
|
| 6966 |
+
"epoch": 0.40697435897435896,
|
| 6967 |
+
"grad_norm": 4.03125,
|
| 6968 |
+
"learning_rate": 3.02846379238175e-06,
|
| 6969 |
+
"loss": 1.3283929824829102,
|
| 6970 |
+
"step": 992
|
| 6971 |
+
},
|
| 6972 |
+
{
|
| 6973 |
+
"epoch": 0.4073846153846154,
|
| 6974 |
+
"grad_norm": 3.765625,
|
| 6975 |
+
"learning_rate": 3.026370866471327e-06,
|
| 6976 |
+
"loss": 1.016139030456543,
|
| 6977 |
+
"step": 993
|
| 6978 |
+
},
|
| 6979 |
+
{
|
| 6980 |
+
"epoch": 0.4077948717948718,
|
| 6981 |
+
"grad_norm": 4.375,
|
| 6982 |
+
"learning_rate": 3.024277940560905e-06,
|
| 6983 |
+
"loss": 1.0495033264160156,
|
| 6984 |
+
"step": 994
|
| 6985 |
+
},
|
| 6986 |
+
{
|
| 6987 |
+
"epoch": 0.4082051282051282,
|
| 6988 |
+
"grad_norm": 4.125,
|
| 6989 |
+
"learning_rate": 3.0221850146504816e-06,
|
| 6990 |
+
"loss": 1.2140045166015625,
|
| 6991 |
+
"step": 995
|
| 6992 |
+
},
|
| 6993 |
+
{
|
| 6994 |
+
"epoch": 0.4086153846153846,
|
| 6995 |
+
"grad_norm": 3.515625,
|
| 6996 |
+
"learning_rate": 3.020092088740059e-06,
|
| 6997 |
+
"loss": 1.2161176204681396,
|
| 6998 |
+
"step": 996
|
| 6999 |
+
},
|
| 7000 |
+
{
|
| 7001 |
+
"epoch": 0.40902564102564104,
|
| 7002 |
+
"grad_norm": 3.859375,
|
| 7003 |
+
"learning_rate": 3.017999162829636e-06,
|
| 7004 |
+
"loss": 1.0736242532730103,
|
| 7005 |
+
"step": 997
|
| 7006 |
+
},
|
| 7007 |
+
{
|
| 7008 |
+
"epoch": 0.4094358974358974,
|
| 7009 |
+
"grad_norm": 3.78125,
|
| 7010 |
+
"learning_rate": 3.0159062369192132e-06,
|
| 7011 |
+
"loss": 1.1534308195114136,
|
| 7012 |
+
"step": 998
|
| 7013 |
+
},
|
| 7014 |
+
{
|
| 7015 |
+
"epoch": 0.40984615384615386,
|
| 7016 |
+
"grad_norm": 3.671875,
|
| 7017 |
+
"learning_rate": 3.0138133110087905e-06,
|
| 7018 |
+
"loss": 1.082121729850769,
|
| 7019 |
+
"step": 999
|
| 7020 |
+
},
|
| 7021 |
+
{
|
| 7022 |
+
"epoch": 0.41025641025641024,
|
| 7023 |
+
"grad_norm": 4.125,
|
| 7024 |
+
"learning_rate": 3.0117203850983677e-06,
|
| 7025 |
+
"loss": 1.2772135734558105,
|
| 7026 |
+
"step": 1000
|
| 7027 |
+
},
|
| 7028 |
+
{
|
| 7029 |
+
"epoch": 0.4106666666666667,
|
| 7030 |
+
"grad_norm": 3.8125,
|
| 7031 |
+
"learning_rate": 3.009627459187945e-06,
|
| 7032 |
+
"loss": 1.0698857307434082,
|
| 7033 |
+
"step": 1001
|
| 7034 |
+
},
|
| 7035 |
+
{
|
| 7036 |
+
"epoch": 0.41107692307692306,
|
| 7037 |
+
"grad_norm": 3.5,
|
| 7038 |
+
"learning_rate": 3.0075345332775225e-06,
|
| 7039 |
+
"loss": 1.1434152126312256,
|
| 7040 |
+
"step": 1002
|
| 7041 |
+
},
|
| 7042 |
+
{
|
| 7043 |
+
"epoch": 0.4114871794871795,
|
| 7044 |
+
"grad_norm": 4.125,
|
| 7045 |
+
"learning_rate": 3.0054416073670993e-06,
|
| 7046 |
+
"loss": 1.0693262815475464,
|
| 7047 |
+
"step": 1003
|
| 7048 |
+
},
|
| 7049 |
+
{
|
| 7050 |
+
"epoch": 0.4118974358974359,
|
| 7051 |
+
"grad_norm": 4.25,
|
| 7052 |
+
"learning_rate": 3.0033486814566765e-06,
|
| 7053 |
+
"loss": 1.2686125040054321,
|
| 7054 |
+
"step": 1004
|
| 7055 |
+
},
|
| 7056 |
+
{
|
| 7057 |
+
"epoch": 0.4123076923076923,
|
| 7058 |
+
"grad_norm": 4.3125,
|
| 7059 |
+
"learning_rate": 3.001255755546254e-06,
|
| 7060 |
+
"loss": 1.2773319482803345,
|
| 7061 |
+
"step": 1005
|
| 7062 |
+
},
|
| 7063 |
+
{
|
| 7064 |
+
"epoch": 0.4127179487179487,
|
| 7065 |
+
"grad_norm": 3.765625,
|
| 7066 |
+
"learning_rate": 2.999162829635831e-06,
|
| 7067 |
+
"loss": 1.201353907585144,
|
| 7068 |
+
"step": 1006
|
| 7069 |
+
},
|
| 7070 |
+
{
|
| 7071 |
+
"epoch": 0.41312820512820514,
|
| 7072 |
+
"grad_norm": 4.125,
|
| 7073 |
+
"learning_rate": 2.9970699037254086e-06,
|
| 7074 |
+
"loss": 1.277069091796875,
|
| 7075 |
+
"step": 1007
|
| 7076 |
+
},
|
| 7077 |
+
{
|
| 7078 |
+
"epoch": 0.4135384615384615,
|
| 7079 |
+
"grad_norm": 3.6875,
|
| 7080 |
+
"learning_rate": 2.994976977814986e-06,
|
| 7081 |
+
"loss": 1.0054022073745728,
|
| 7082 |
+
"step": 1008
|
| 7083 |
+
},
|
| 7084 |
+
{
|
| 7085 |
+
"epoch": 0.41394871794871796,
|
| 7086 |
+
"grad_norm": 4.03125,
|
| 7087 |
+
"learning_rate": 2.9928840519045626e-06,
|
| 7088 |
+
"loss": 1.1745253801345825,
|
| 7089 |
+
"step": 1009
|
| 7090 |
+
},
|
| 7091 |
+
{
|
| 7092 |
+
"epoch": 0.41435897435897434,
|
| 7093 |
+
"grad_norm": 4.0,
|
| 7094 |
+
"learning_rate": 2.9907911259941402e-06,
|
| 7095 |
+
"loss": 1.1786189079284668,
|
| 7096 |
+
"step": 1010
|
| 7097 |
+
},
|
| 7098 |
+
{
|
| 7099 |
+
"epoch": 0.4147692307692308,
|
| 7100 |
+
"grad_norm": 4.125,
|
| 7101 |
+
"learning_rate": 2.9886982000837175e-06,
|
| 7102 |
+
"loss": 1.2094573974609375,
|
| 7103 |
+
"step": 1011
|
| 7104 |
+
},
|
| 7105 |
+
{
|
| 7106 |
+
"epoch": 0.41517948717948716,
|
| 7107 |
+
"grad_norm": 3.8125,
|
| 7108 |
+
"learning_rate": 2.9866052741732942e-06,
|
| 7109 |
+
"loss": 1.269519567489624,
|
| 7110 |
+
"step": 1012
|
| 7111 |
+
},
|
| 7112 |
+
{
|
| 7113 |
+
"epoch": 0.4155897435897436,
|
| 7114 |
+
"grad_norm": 3.765625,
|
| 7115 |
+
"learning_rate": 2.984512348262872e-06,
|
| 7116 |
+
"loss": 1.0968009233474731,
|
| 7117 |
+
"step": 1013
|
| 7118 |
+
},
|
| 7119 |
+
{
|
| 7120 |
+
"epoch": 0.416,
|
| 7121 |
+
"grad_norm": 3.578125,
|
| 7122 |
+
"learning_rate": 2.9824194223524487e-06,
|
| 7123 |
+
"loss": 1.1730378866195679,
|
| 7124 |
+
"step": 1014
|
| 7125 |
+
},
|
| 7126 |
+
{
|
| 7127 |
+
"epoch": 0.4164102564102564,
|
| 7128 |
+
"grad_norm": 3.921875,
|
| 7129 |
+
"learning_rate": 2.9803264964420263e-06,
|
| 7130 |
+
"loss": 1.308756947517395,
|
| 7131 |
+
"step": 1015
|
| 7132 |
+
},
|
| 7133 |
+
{
|
| 7134 |
+
"epoch": 0.4168205128205128,
|
| 7135 |
+
"grad_norm": 4.0625,
|
| 7136 |
+
"learning_rate": 2.9782335705316035e-06,
|
| 7137 |
+
"loss": 1.2382745742797852,
|
| 7138 |
+
"step": 1016
|
| 7139 |
+
},
|
| 7140 |
+
{
|
| 7141 |
+
"epoch": 0.41723076923076924,
|
| 7142 |
+
"grad_norm": 4.28125,
|
| 7143 |
+
"learning_rate": 2.9761406446211803e-06,
|
| 7144 |
+
"loss": 1.2725597620010376,
|
| 7145 |
+
"step": 1017
|
| 7146 |
+
},
|
| 7147 |
+
{
|
| 7148 |
+
"epoch": 0.4176410256410256,
|
| 7149 |
+
"grad_norm": 3.625,
|
| 7150 |
+
"learning_rate": 2.974047718710758e-06,
|
| 7151 |
+
"loss": 1.1540956497192383,
|
| 7152 |
+
"step": 1018
|
| 7153 |
+
},
|
| 7154 |
+
{
|
| 7155 |
+
"epoch": 0.41805128205128206,
|
| 7156 |
+
"grad_norm": 3.859375,
|
| 7157 |
+
"learning_rate": 2.971954792800335e-06,
|
| 7158 |
+
"loss": 1.1573514938354492,
|
| 7159 |
+
"step": 1019
|
| 7160 |
+
},
|
| 7161 |
+
{
|
| 7162 |
+
"epoch": 0.41846153846153844,
|
| 7163 |
+
"grad_norm": 3.875,
|
| 7164 |
+
"learning_rate": 2.969861866889912e-06,
|
| 7165 |
+
"loss": 1.1249496936798096,
|
| 7166 |
+
"step": 1020
|
| 7167 |
+
},
|
| 7168 |
+
{
|
| 7169 |
+
"epoch": 0.4188717948717949,
|
| 7170 |
+
"grad_norm": 3.9375,
|
| 7171 |
+
"learning_rate": 2.9677689409794896e-06,
|
| 7172 |
+
"loss": 1.1744952201843262,
|
| 7173 |
+
"step": 1021
|
| 7174 |
+
},
|
| 7175 |
+
{
|
| 7176 |
+
"epoch": 0.41928205128205126,
|
| 7177 |
+
"grad_norm": 3.875,
|
| 7178 |
+
"learning_rate": 2.965676015069067e-06,
|
| 7179 |
+
"loss": 1.2501696348190308,
|
| 7180 |
+
"step": 1022
|
| 7181 |
+
},
|
| 7182 |
+
{
|
| 7183 |
+
"epoch": 0.4196923076923077,
|
| 7184 |
+
"grad_norm": 3.65625,
|
| 7185 |
+
"learning_rate": 2.963583089158644e-06,
|
| 7186 |
+
"loss": 1.2378509044647217,
|
| 7187 |
+
"step": 1023
|
| 7188 |
+
},
|
| 7189 |
+
{
|
| 7190 |
+
"epoch": 0.4201025641025641,
|
| 7191 |
+
"grad_norm": 4.25,
|
| 7192 |
+
"learning_rate": 2.9614901632482212e-06,
|
| 7193 |
+
"loss": 1.1466659307479858,
|
| 7194 |
+
"step": 1024
|
| 7195 |
+
},
|
| 7196 |
+
{
|
| 7197 |
+
"epoch": 0.4205128205128205,
|
| 7198 |
+
"grad_norm": 4.125,
|
| 7199 |
+
"learning_rate": 2.9593972373377985e-06,
|
| 7200 |
+
"loss": 1.2907891273498535,
|
| 7201 |
+
"step": 1025
|
| 7202 |
+
},
|
| 7203 |
+
{
|
| 7204 |
+
"epoch": 0.4209230769230769,
|
| 7205 |
+
"grad_norm": 4.28125,
|
| 7206 |
+
"learning_rate": 2.9573043114273757e-06,
|
| 7207 |
+
"loss": 1.1832275390625,
|
| 7208 |
+
"step": 1026
|
| 7209 |
+
},
|
| 7210 |
+
{
|
| 7211 |
+
"epoch": 0.42133333333333334,
|
| 7212 |
+
"grad_norm": 3.828125,
|
| 7213 |
+
"learning_rate": 2.955211385516953e-06,
|
| 7214 |
+
"loss": 1.220141053199768,
|
| 7215 |
+
"step": 1027
|
| 7216 |
+
},
|
| 7217 |
+
{
|
| 7218 |
+
"epoch": 0.4217435897435897,
|
| 7219 |
+
"grad_norm": 3.953125,
|
| 7220 |
+
"learning_rate": 2.9531184596065305e-06,
|
| 7221 |
+
"loss": 1.2819666862487793,
|
| 7222 |
+
"step": 1028
|
| 7223 |
+
},
|
| 7224 |
+
{
|
| 7225 |
+
"epoch": 0.42215384615384616,
|
| 7226 |
+
"grad_norm": 4.125,
|
| 7227 |
+
"learning_rate": 2.9510255336961073e-06,
|
| 7228 |
+
"loss": 1.2700068950653076,
|
| 7229 |
+
"step": 1029
|
| 7230 |
+
},
|
| 7231 |
+
{
|
| 7232 |
+
"epoch": 0.42256410256410254,
|
| 7233 |
+
"grad_norm": 3.953125,
|
| 7234 |
+
"learning_rate": 2.9489326077856845e-06,
|
| 7235 |
+
"loss": 1.1419286727905273,
|
| 7236 |
+
"step": 1030
|
| 7237 |
+
},
|
| 7238 |
+
{
|
| 7239 |
+
"epoch": 0.422974358974359,
|
| 7240 |
+
"grad_norm": 4.0,
|
| 7241 |
+
"learning_rate": 2.946839681875262e-06,
|
| 7242 |
+
"loss": 1.1791789531707764,
|
| 7243 |
+
"step": 1031
|
| 7244 |
+
},
|
| 7245 |
+
{
|
| 7246 |
+
"epoch": 0.42338461538461536,
|
| 7247 |
+
"grad_norm": 4.34375,
|
| 7248 |
+
"learning_rate": 2.944746755964839e-06,
|
| 7249 |
+
"loss": 1.1738345623016357,
|
| 7250 |
+
"step": 1032
|
| 7251 |
+
},
|
| 7252 |
+
{
|
| 7253 |
+
"epoch": 0.4237948717948718,
|
| 7254 |
+
"grad_norm": 4.0625,
|
| 7255 |
+
"learning_rate": 2.942653830054416e-06,
|
| 7256 |
+
"loss": 1.102777123451233,
|
| 7257 |
+
"step": 1033
|
| 7258 |
+
},
|
| 7259 |
+
{
|
| 7260 |
+
"epoch": 0.4242051282051282,
|
| 7261 |
+
"grad_norm": 3.8125,
|
| 7262 |
+
"learning_rate": 2.940560904143994e-06,
|
| 7263 |
+
"loss": 1.298481822013855,
|
| 7264 |
+
"step": 1034
|
| 7265 |
+
},
|
| 7266 |
+
{
|
| 7267 |
+
"epoch": 0.4246153846153846,
|
| 7268 |
+
"grad_norm": 3.859375,
|
| 7269 |
+
"learning_rate": 2.9384679782335706e-06,
|
| 7270 |
+
"loss": 1.1880578994750977,
|
| 7271 |
+
"step": 1035
|
| 7272 |
+
},
|
| 7273 |
+
{
|
| 7274 |
+
"epoch": 0.425025641025641,
|
| 7275 |
+
"grad_norm": 4.09375,
|
| 7276 |
+
"learning_rate": 2.9363750523231482e-06,
|
| 7277 |
+
"loss": 1.2760035991668701,
|
| 7278 |
+
"step": 1036
|
| 7279 |
+
},
|
| 7280 |
+
{
|
| 7281 |
+
"epoch": 0.42543589743589744,
|
| 7282 |
+
"grad_norm": 4.21875,
|
| 7283 |
+
"learning_rate": 2.9342821264127255e-06,
|
| 7284 |
+
"loss": 1.3583061695098877,
|
| 7285 |
+
"step": 1037
|
| 7286 |
+
},
|
| 7287 |
+
{
|
| 7288 |
+
"epoch": 0.4258461538461538,
|
| 7289 |
+
"grad_norm": 4.09375,
|
| 7290 |
+
"learning_rate": 2.9321892005023022e-06,
|
| 7291 |
+
"loss": 1.3997758626937866,
|
| 7292 |
+
"step": 1038
|
| 7293 |
+
},
|
| 7294 |
+
{
|
| 7295 |
+
"epoch": 0.42625641025641026,
|
| 7296 |
+
"grad_norm": 3.6875,
|
| 7297 |
+
"learning_rate": 2.93009627459188e-06,
|
| 7298 |
+
"loss": 1.1049081087112427,
|
| 7299 |
+
"step": 1039
|
| 7300 |
+
},
|
| 7301 |
+
{
|
| 7302 |
+
"epoch": 0.4266666666666667,
|
| 7303 |
+
"grad_norm": 4.03125,
|
| 7304 |
+
"learning_rate": 2.9280033486814567e-06,
|
| 7305 |
+
"loss": 1.2833629846572876,
|
| 7306 |
+
"step": 1040
|
| 7307 |
+
},
|
| 7308 |
+
{
|
| 7309 |
+
"epoch": 0.4270769230769231,
|
| 7310 |
+
"grad_norm": 4.03125,
|
| 7311 |
+
"learning_rate": 2.925910422771034e-06,
|
| 7312 |
+
"loss": 1.2144968509674072,
|
| 7313 |
+
"step": 1041
|
| 7314 |
+
},
|
| 7315 |
+
{
|
| 7316 |
+
"epoch": 0.4274871794871795,
|
| 7317 |
+
"grad_norm": 3.859375,
|
| 7318 |
+
"learning_rate": 2.9238174968606115e-06,
|
| 7319 |
+
"loss": 1.276278018951416,
|
| 7320 |
+
"step": 1042
|
| 7321 |
+
},
|
| 7322 |
+
{
|
| 7323 |
+
"epoch": 0.4278974358974359,
|
| 7324 |
+
"grad_norm": 3.796875,
|
| 7325 |
+
"learning_rate": 2.9217245709501883e-06,
|
| 7326 |
+
"loss": 1.0377084016799927,
|
| 7327 |
+
"step": 1043
|
| 7328 |
+
},
|
| 7329 |
+
{
|
| 7330 |
+
"epoch": 0.42830769230769233,
|
| 7331 |
+
"grad_norm": 3.78125,
|
| 7332 |
+
"learning_rate": 2.919631645039766e-06,
|
| 7333 |
+
"loss": 1.2340822219848633,
|
| 7334 |
+
"step": 1044
|
| 7335 |
+
},
|
| 7336 |
+
{
|
| 7337 |
+
"epoch": 0.4287179487179487,
|
| 7338 |
+
"grad_norm": 4.03125,
|
| 7339 |
+
"learning_rate": 2.917538719129343e-06,
|
| 7340 |
+
"loss": 1.0763278007507324,
|
| 7341 |
+
"step": 1045
|
| 7342 |
+
},
|
| 7343 |
+
{
|
| 7344 |
+
"epoch": 0.42912820512820515,
|
| 7345 |
+
"grad_norm": 3.546875,
|
| 7346 |
+
"learning_rate": 2.91544579321892e-06,
|
| 7347 |
+
"loss": 1.2491021156311035,
|
| 7348 |
+
"step": 1046
|
| 7349 |
+
},
|
| 7350 |
+
{
|
| 7351 |
+
"epoch": 0.42953846153846154,
|
| 7352 |
+
"grad_norm": 3.8125,
|
| 7353 |
+
"learning_rate": 2.9133528673084976e-06,
|
| 7354 |
+
"loss": 1.2443821430206299,
|
| 7355 |
+
"step": 1047
|
| 7356 |
+
},
|
| 7357 |
+
{
|
| 7358 |
+
"epoch": 0.429948717948718,
|
| 7359 |
+
"grad_norm": 3.90625,
|
| 7360 |
+
"learning_rate": 2.911259941398075e-06,
|
| 7361 |
+
"loss": 1.2136783599853516,
|
| 7362 |
+
"step": 1048
|
| 7363 |
+
},
|
| 7364 |
+
{
|
| 7365 |
+
"epoch": 0.43035897435897436,
|
| 7366 |
+
"grad_norm": 4.09375,
|
| 7367 |
+
"learning_rate": 2.909167015487652e-06,
|
| 7368 |
+
"loss": 1.066545009613037,
|
| 7369 |
+
"step": 1049
|
| 7370 |
+
},
|
| 7371 |
+
{
|
| 7372 |
+
"epoch": 0.4307692307692308,
|
| 7373 |
+
"grad_norm": 4.34375,
|
| 7374 |
+
"learning_rate": 2.9070740895772292e-06,
|
| 7375 |
+
"loss": 1.218514323234558,
|
| 7376 |
+
"step": 1050
|
| 7377 |
+
},
|
| 7378 |
+
{
|
| 7379 |
+
"epoch": 0.4311794871794872,
|
| 7380 |
+
"grad_norm": 4.0625,
|
| 7381 |
+
"learning_rate": 2.9049811636668065e-06,
|
| 7382 |
+
"loss": 1.2703115940093994,
|
| 7383 |
+
"step": 1051
|
| 7384 |
+
},
|
| 7385 |
+
{
|
| 7386 |
+
"epoch": 0.4315897435897436,
|
| 7387 |
+
"grad_norm": 4.53125,
|
| 7388 |
+
"learning_rate": 2.9028882377563837e-06,
|
| 7389 |
+
"loss": 1.3395044803619385,
|
| 7390 |
+
"step": 1052
|
| 7391 |
+
},
|
| 7392 |
+
{
|
| 7393 |
+
"epoch": 0.432,
|
| 7394 |
+
"grad_norm": 4.25,
|
| 7395 |
+
"learning_rate": 2.900795311845961e-06,
|
| 7396 |
+
"loss": 1.2730753421783447,
|
| 7397 |
+
"step": 1053
|
| 7398 |
+
},
|
| 7399 |
+
{
|
| 7400 |
+
"epoch": 0.43241025641025643,
|
| 7401 |
+
"grad_norm": 4.46875,
|
| 7402 |
+
"learning_rate": 2.8987023859355377e-06,
|
| 7403 |
+
"loss": 1.2495253086090088,
|
| 7404 |
+
"step": 1054
|
| 7405 |
+
},
|
| 7406 |
+
{
|
| 7407 |
+
"epoch": 0.4328205128205128,
|
| 7408 |
+
"grad_norm": 3.953125,
|
| 7409 |
+
"learning_rate": 2.8966094600251153e-06,
|
| 7410 |
+
"loss": 1.212220549583435,
|
| 7411 |
+
"step": 1055
|
| 7412 |
+
},
|
| 7413 |
+
{
|
| 7414 |
+
"epoch": 0.43323076923076925,
|
| 7415 |
+
"grad_norm": 4.09375,
|
| 7416 |
+
"learning_rate": 2.8945165341146925e-06,
|
| 7417 |
+
"loss": 1.1558529138565063,
|
| 7418 |
+
"step": 1056
|
| 7419 |
+
},
|
| 7420 |
+
{
|
| 7421 |
+
"epoch": 0.43364102564102563,
|
| 7422 |
+
"grad_norm": 3.75,
|
| 7423 |
+
"learning_rate": 2.89242360820427e-06,
|
| 7424 |
+
"loss": 1.0970323085784912,
|
| 7425 |
+
"step": 1057
|
| 7426 |
+
},
|
| 7427 |
+
{
|
| 7428 |
+
"epoch": 0.4340512820512821,
|
| 7429 |
+
"grad_norm": 3.859375,
|
| 7430 |
+
"learning_rate": 2.890330682293847e-06,
|
| 7431 |
+
"loss": 1.2792545557022095,
|
| 7432 |
+
"step": 1058
|
| 7433 |
+
},
|
| 7434 |
+
{
|
| 7435 |
+
"epoch": 0.43446153846153845,
|
| 7436 |
+
"grad_norm": 4.21875,
|
| 7437 |
+
"learning_rate": 2.888237756383424e-06,
|
| 7438 |
+
"loss": 1.2390037775039673,
|
| 7439 |
+
"step": 1059
|
| 7440 |
+
},
|
| 7441 |
+
{
|
| 7442 |
+
"epoch": 0.4348717948717949,
|
| 7443 |
+
"grad_norm": 3.765625,
|
| 7444 |
+
"learning_rate": 2.886144830473002e-06,
|
| 7445 |
+
"loss": 1.2574375867843628,
|
| 7446 |
+
"step": 1060
|
| 7447 |
+
},
|
| 7448 |
+
{
|
| 7449 |
+
"epoch": 0.4352820512820513,
|
| 7450 |
+
"grad_norm": 4.125,
|
| 7451 |
+
"learning_rate": 2.8840519045625786e-06,
|
| 7452 |
+
"loss": 1.1660637855529785,
|
| 7453 |
+
"step": 1061
|
| 7454 |
+
},
|
| 7455 |
+
{
|
| 7456 |
+
"epoch": 0.4356923076923077,
|
| 7457 |
+
"grad_norm": 3.75,
|
| 7458 |
+
"learning_rate": 2.881958978652156e-06,
|
| 7459 |
+
"loss": 1.1375092267990112,
|
| 7460 |
+
"step": 1062
|
| 7461 |
+
},
|
| 7462 |
+
{
|
| 7463 |
+
"epoch": 0.4361025641025641,
|
| 7464 |
+
"grad_norm": 3.96875,
|
| 7465 |
+
"learning_rate": 2.8798660527417335e-06,
|
| 7466 |
+
"loss": 1.2119207382202148,
|
| 7467 |
+
"step": 1063
|
| 7468 |
+
},
|
| 7469 |
+
{
|
| 7470 |
+
"epoch": 0.43651282051282053,
|
| 7471 |
+
"grad_norm": 4.53125,
|
| 7472 |
+
"learning_rate": 2.8777731268313102e-06,
|
| 7473 |
+
"loss": 1.1683343648910522,
|
| 7474 |
+
"step": 1064
|
| 7475 |
+
},
|
| 7476 |
+
{
|
| 7477 |
+
"epoch": 0.4369230769230769,
|
| 7478 |
+
"grad_norm": 4.0625,
|
| 7479 |
+
"learning_rate": 2.875680200920888e-06,
|
| 7480 |
+
"loss": 1.393289566040039,
|
| 7481 |
+
"step": 1065
|
| 7482 |
+
},
|
| 7483 |
+
{
|
| 7484 |
+
"epoch": 0.43733333333333335,
|
| 7485 |
+
"grad_norm": 3.984375,
|
| 7486 |
+
"learning_rate": 2.8735872750104647e-06,
|
| 7487 |
+
"loss": 1.449811339378357,
|
| 7488 |
+
"step": 1066
|
| 7489 |
+
},
|
| 7490 |
+
{
|
| 7491 |
+
"epoch": 0.43774358974358973,
|
| 7492 |
+
"grad_norm": 3.859375,
|
| 7493 |
+
"learning_rate": 2.871494349100042e-06,
|
| 7494 |
+
"loss": 1.2296123504638672,
|
| 7495 |
+
"step": 1067
|
| 7496 |
+
},
|
| 7497 |
+
{
|
| 7498 |
+
"epoch": 0.43815384615384617,
|
| 7499 |
+
"grad_norm": 4.1875,
|
| 7500 |
+
"learning_rate": 2.8694014231896195e-06,
|
| 7501 |
+
"loss": 1.248206377029419,
|
| 7502 |
+
"step": 1068
|
| 7503 |
+
},
|
| 7504 |
+
{
|
| 7505 |
+
"epoch": 0.43856410256410255,
|
| 7506 |
+
"grad_norm": 4.40625,
|
| 7507 |
+
"learning_rate": 2.8673084972791963e-06,
|
| 7508 |
+
"loss": 1.294684886932373,
|
| 7509 |
+
"step": 1069
|
| 7510 |
+
},
|
| 7511 |
+
{
|
| 7512 |
+
"epoch": 0.438974358974359,
|
| 7513 |
+
"grad_norm": 3.96875,
|
| 7514 |
+
"learning_rate": 2.865215571368774e-06,
|
| 7515 |
+
"loss": 1.2600420713424683,
|
| 7516 |
+
"step": 1070
|
| 7517 |
+
},
|
| 7518 |
+
{
|
| 7519 |
+
"epoch": 0.4393846153846154,
|
| 7520 |
+
"grad_norm": 3.703125,
|
| 7521 |
+
"learning_rate": 2.863122645458351e-06,
|
| 7522 |
+
"loss": 1.327905297279358,
|
| 7523 |
+
"step": 1071
|
| 7524 |
+
},
|
| 7525 |
+
{
|
| 7526 |
+
"epoch": 0.4397948717948718,
|
| 7527 |
+
"grad_norm": 4.65625,
|
| 7528 |
+
"learning_rate": 2.861029719547928e-06,
|
| 7529 |
+
"loss": 1.30216646194458,
|
| 7530 |
+
"step": 1072
|
| 7531 |
+
},
|
| 7532 |
+
{
|
| 7533 |
+
"epoch": 0.4402051282051282,
|
| 7534 |
+
"grad_norm": 3.921875,
|
| 7535 |
+
"learning_rate": 2.8589367936375056e-06,
|
| 7536 |
+
"loss": 1.115570306777954,
|
| 7537 |
+
"step": 1073
|
| 7538 |
+
},
|
| 7539 |
+
{
|
| 7540 |
+
"epoch": 0.44061538461538463,
|
| 7541 |
+
"grad_norm": 3.8125,
|
| 7542 |
+
"learning_rate": 2.856843867727083e-06,
|
| 7543 |
+
"loss": 1.2897348403930664,
|
| 7544 |
+
"step": 1074
|
| 7545 |
+
},
|
| 7546 |
+
{
|
| 7547 |
+
"epoch": 0.441025641025641,
|
| 7548 |
+
"grad_norm": 4.09375,
|
| 7549 |
+
"learning_rate": 2.8547509418166596e-06,
|
| 7550 |
+
"loss": 1.1873122453689575,
|
| 7551 |
+
"step": 1075
|
| 7552 |
+
},
|
| 7553 |
+
{
|
| 7554 |
+
"epoch": 0.44143589743589745,
|
| 7555 |
+
"grad_norm": 3.890625,
|
| 7556 |
+
"learning_rate": 2.8526580159062372e-06,
|
| 7557 |
+
"loss": 1.3173024654388428,
|
| 7558 |
+
"step": 1076
|
| 7559 |
+
},
|
| 7560 |
+
{
|
| 7561 |
+
"epoch": 0.44184615384615383,
|
| 7562 |
+
"grad_norm": 3.859375,
|
| 7563 |
+
"learning_rate": 2.8505650899958145e-06,
|
| 7564 |
+
"loss": 1.2999330759048462,
|
| 7565 |
+
"step": 1077
|
| 7566 |
+
},
|
| 7567 |
+
{
|
| 7568 |
+
"epoch": 0.44225641025641027,
|
| 7569 |
+
"grad_norm": 4.28125,
|
| 7570 |
+
"learning_rate": 2.8484721640853917e-06,
|
| 7571 |
+
"loss": 1.3054252862930298,
|
| 7572 |
+
"step": 1078
|
| 7573 |
+
},
|
| 7574 |
+
{
|
| 7575 |
+
"epoch": 0.44266666666666665,
|
| 7576 |
+
"grad_norm": 4.0625,
|
| 7577 |
+
"learning_rate": 2.846379238174969e-06,
|
| 7578 |
+
"loss": 1.2864702939987183,
|
| 7579 |
+
"step": 1079
|
| 7580 |
+
},
|
| 7581 |
+
{
|
| 7582 |
+
"epoch": 0.4430769230769231,
|
| 7583 |
+
"grad_norm": 4.3125,
|
| 7584 |
+
"learning_rate": 2.8442863122645457e-06,
|
| 7585 |
+
"loss": 1.1695835590362549,
|
| 7586 |
+
"step": 1080
|
| 7587 |
+
},
|
| 7588 |
+
{
|
| 7589 |
+
"epoch": 0.4434871794871795,
|
| 7590 |
+
"grad_norm": 4.625,
|
| 7591 |
+
"learning_rate": 2.8421933863541233e-06,
|
| 7592 |
+
"loss": 1.3510276079177856,
|
| 7593 |
+
"step": 1081
|
| 7594 |
+
},
|
| 7595 |
+
{
|
| 7596 |
+
"epoch": 0.4438974358974359,
|
| 7597 |
+
"grad_norm": 4.5625,
|
| 7598 |
+
"learning_rate": 2.8401004604437005e-06,
|
| 7599 |
+
"loss": 1.3043633699417114,
|
| 7600 |
+
"step": 1082
|
| 7601 |
+
},
|
| 7602 |
+
{
|
| 7603 |
+
"epoch": 0.4443076923076923,
|
| 7604 |
+
"grad_norm": 4.40625,
|
| 7605 |
+
"learning_rate": 2.8380075345332773e-06,
|
| 7606 |
+
"loss": 1.3284595012664795,
|
| 7607 |
+
"step": 1083
|
| 7608 |
+
},
|
| 7609 |
+
{
|
| 7610 |
+
"epoch": 0.44471794871794873,
|
| 7611 |
+
"grad_norm": 3.90625,
|
| 7612 |
+
"learning_rate": 2.835914608622855e-06,
|
| 7613 |
+
"loss": 1.1851894855499268,
|
| 7614 |
+
"step": 1084
|
| 7615 |
+
},
|
| 7616 |
+
{
|
| 7617 |
+
"epoch": 0.4451282051282051,
|
| 7618 |
+
"grad_norm": 4.0625,
|
| 7619 |
+
"learning_rate": 2.833821682712432e-06,
|
| 7620 |
+
"loss": 1.3271174430847168,
|
| 7621 |
+
"step": 1085
|
| 7622 |
+
},
|
| 7623 |
+
{
|
| 7624 |
+
"epoch": 0.44553846153846155,
|
| 7625 |
+
"grad_norm": 4.34375,
|
| 7626 |
+
"learning_rate": 2.83172875680201e-06,
|
| 7627 |
+
"loss": 1.2851948738098145,
|
| 7628 |
+
"step": 1086
|
| 7629 |
+
},
|
| 7630 |
+
{
|
| 7631 |
+
"epoch": 0.44594871794871793,
|
| 7632 |
+
"grad_norm": 4.0625,
|
| 7633 |
+
"learning_rate": 2.8296358308915866e-06,
|
| 7634 |
+
"loss": 1.2113120555877686,
|
| 7635 |
+
"step": 1087
|
| 7636 |
+
},
|
| 7637 |
+
{
|
| 7638 |
+
"epoch": 0.44635897435897437,
|
| 7639 |
+
"grad_norm": 3.796875,
|
| 7640 |
+
"learning_rate": 2.827542904981164e-06,
|
| 7641 |
+
"loss": 1.2024807929992676,
|
| 7642 |
+
"step": 1088
|
| 7643 |
+
},
|
| 7644 |
+
{
|
| 7645 |
+
"epoch": 0.44676923076923075,
|
| 7646 |
+
"grad_norm": 4.125,
|
| 7647 |
+
"learning_rate": 2.8254499790707415e-06,
|
| 7648 |
+
"loss": 1.2510011196136475,
|
| 7649 |
+
"step": 1089
|
| 7650 |
+
},
|
| 7651 |
+
{
|
| 7652 |
+
"epoch": 0.4471794871794872,
|
| 7653 |
+
"grad_norm": 4.03125,
|
| 7654 |
+
"learning_rate": 2.8233570531603182e-06,
|
| 7655 |
+
"loss": 1.2534303665161133,
|
| 7656 |
+
"step": 1090
|
| 7657 |
+
},
|
| 7658 |
+
{
|
| 7659 |
+
"epoch": 0.44758974358974357,
|
| 7660 |
+
"grad_norm": 4.3125,
|
| 7661 |
+
"learning_rate": 2.821264127249896e-06,
|
| 7662 |
+
"loss": 1.2300729751586914,
|
| 7663 |
+
"step": 1091
|
| 7664 |
+
},
|
| 7665 |
+
{
|
| 7666 |
+
"epoch": 0.448,
|
| 7667 |
+
"grad_norm": 4.0,
|
| 7668 |
+
"learning_rate": 2.8191712013394727e-06,
|
| 7669 |
+
"loss": 1.3683750629425049,
|
| 7670 |
+
"step": 1092
|
| 7671 |
+
},
|
| 7672 |
+
{
|
| 7673 |
+
"epoch": 0.4484102564102564,
|
| 7674 |
+
"grad_norm": 3.609375,
|
| 7675 |
+
"learning_rate": 2.81707827542905e-06,
|
| 7676 |
+
"loss": 1.1301509141921997,
|
| 7677 |
+
"step": 1093
|
| 7678 |
+
},
|
| 7679 |
+
{
|
| 7680 |
+
"epoch": 0.44882051282051283,
|
| 7681 |
+
"grad_norm": 4.125,
|
| 7682 |
+
"learning_rate": 2.8149853495186275e-06,
|
| 7683 |
+
"loss": 1.335710048675537,
|
| 7684 |
+
"step": 1094
|
| 7685 |
+
},
|
| 7686 |
+
{
|
| 7687 |
+
"epoch": 0.4492307692307692,
|
| 7688 |
+
"grad_norm": 3.625,
|
| 7689 |
+
"learning_rate": 2.8128924236082043e-06,
|
| 7690 |
+
"loss": 1.1548030376434326,
|
| 7691 |
+
"step": 1095
|
| 7692 |
+
},
|
| 7693 |
+
{
|
| 7694 |
+
"epoch": 0.44964102564102565,
|
| 7695 |
+
"grad_norm": 3.953125,
|
| 7696 |
+
"learning_rate": 2.8107994976977815e-06,
|
| 7697 |
+
"loss": 1.2068957090377808,
|
| 7698 |
+
"step": 1096
|
| 7699 |
+
},
|
| 7700 |
+
{
|
| 7701 |
+
"epoch": 0.45005128205128203,
|
| 7702 |
+
"grad_norm": 4.0,
|
| 7703 |
+
"learning_rate": 2.808706571787359e-06,
|
| 7704 |
+
"loss": 1.182767629623413,
|
| 7705 |
+
"step": 1097
|
| 7706 |
+
},
|
| 7707 |
+
{
|
| 7708 |
+
"epoch": 0.45046153846153847,
|
| 7709 |
+
"grad_norm": 4.0625,
|
| 7710 |
+
"learning_rate": 2.806613645876936e-06,
|
| 7711 |
+
"loss": 1.2844791412353516,
|
| 7712 |
+
"step": 1098
|
| 7713 |
+
},
|
| 7714 |
+
{
|
| 7715 |
+
"epoch": 0.45087179487179485,
|
| 7716 |
+
"grad_norm": 3.671875,
|
| 7717 |
+
"learning_rate": 2.8045207199665136e-06,
|
| 7718 |
+
"loss": 1.1864662170410156,
|
| 7719 |
+
"step": 1099
|
| 7720 |
+
},
|
| 7721 |
+
{
|
| 7722 |
+
"epoch": 0.4512820512820513,
|
| 7723 |
+
"grad_norm": 3.796875,
|
| 7724 |
+
"learning_rate": 2.802427794056091e-06,
|
| 7725 |
+
"loss": 1.208574652671814,
|
| 7726 |
+
"step": 1100
|
| 7727 |
+
},
|
| 7728 |
+
{
|
| 7729 |
+
"epoch": 0.45169230769230767,
|
| 7730 |
+
"grad_norm": 4.09375,
|
| 7731 |
+
"learning_rate": 2.8003348681456676e-06,
|
| 7732 |
+
"loss": 1.2983245849609375,
|
| 7733 |
+
"step": 1101
|
| 7734 |
+
},
|
| 7735 |
+
{
|
| 7736 |
+
"epoch": 0.4521025641025641,
|
| 7737 |
+
"grad_norm": 3.953125,
|
| 7738 |
+
"learning_rate": 2.7982419422352452e-06,
|
| 7739 |
+
"loss": 1.0818623304367065,
|
| 7740 |
+
"step": 1102
|
| 7741 |
+
},
|
| 7742 |
+
{
|
| 7743 |
+
"epoch": 0.4525128205128205,
|
| 7744 |
+
"grad_norm": 3.875,
|
| 7745 |
+
"learning_rate": 2.7961490163248225e-06,
|
| 7746 |
+
"loss": 1.1839005947113037,
|
| 7747 |
+
"step": 1103
|
| 7748 |
+
},
|
| 7749 |
+
{
|
| 7750 |
+
"epoch": 0.45292307692307693,
|
| 7751 |
+
"grad_norm": 4.0,
|
| 7752 |
+
"learning_rate": 2.7940560904143992e-06,
|
| 7753 |
+
"loss": 1.3109443187713623,
|
| 7754 |
+
"step": 1104
|
| 7755 |
+
},
|
| 7756 |
+
{
|
| 7757 |
+
"epoch": 0.4533333333333333,
|
| 7758 |
+
"grad_norm": 3.65625,
|
| 7759 |
+
"learning_rate": 2.791963164503977e-06,
|
| 7760 |
+
"loss": 1.1283392906188965,
|
| 7761 |
+
"step": 1105
|
| 7762 |
+
},
|
| 7763 |
+
{
|
| 7764 |
+
"epoch": 0.45374358974358975,
|
| 7765 |
+
"grad_norm": 3.796875,
|
| 7766 |
+
"learning_rate": 2.7898702385935537e-06,
|
| 7767 |
+
"loss": 1.210160732269287,
|
| 7768 |
+
"step": 1106
|
| 7769 |
+
},
|
| 7770 |
+
{
|
| 7771 |
+
"epoch": 0.45415384615384613,
|
| 7772 |
+
"grad_norm": 4.125,
|
| 7773 |
+
"learning_rate": 2.7877773126831313e-06,
|
| 7774 |
+
"loss": 1.2524406909942627,
|
| 7775 |
+
"step": 1107
|
| 7776 |
+
},
|
| 7777 |
+
{
|
| 7778 |
+
"epoch": 0.45456410256410257,
|
| 7779 |
+
"grad_norm": 3.90625,
|
| 7780 |
+
"learning_rate": 2.7856843867727085e-06,
|
| 7781 |
+
"loss": 1.1924431324005127,
|
| 7782 |
+
"step": 1108
|
| 7783 |
+
},
|
| 7784 |
+
{
|
| 7785 |
+
"epoch": 0.45497435897435895,
|
| 7786 |
+
"grad_norm": 3.84375,
|
| 7787 |
+
"learning_rate": 2.7835914608622853e-06,
|
| 7788 |
+
"loss": 1.4012980461120605,
|
| 7789 |
+
"step": 1109
|
| 7790 |
+
},
|
| 7791 |
+
{
|
| 7792 |
+
"epoch": 0.4553846153846154,
|
| 7793 |
+
"grad_norm": 4.15625,
|
| 7794 |
+
"learning_rate": 2.781498534951863e-06,
|
| 7795 |
+
"loss": 1.2820216417312622,
|
| 7796 |
+
"step": 1110
|
| 7797 |
+
},
|
| 7798 |
+
{
|
| 7799 |
+
"epoch": 0.45579487179487177,
|
| 7800 |
+
"grad_norm": 4.0625,
|
| 7801 |
+
"learning_rate": 2.77940560904144e-06,
|
| 7802 |
+
"loss": 1.2549073696136475,
|
| 7803 |
+
"step": 1111
|
| 7804 |
+
},
|
| 7805 |
+
{
|
| 7806 |
+
"epoch": 0.4562051282051282,
|
| 7807 |
+
"grad_norm": 4.0,
|
| 7808 |
+
"learning_rate": 2.777312683131018e-06,
|
| 7809 |
+
"loss": 1.2605948448181152,
|
| 7810 |
+
"step": 1112
|
| 7811 |
+
},
|
| 7812 |
+
{
|
| 7813 |
+
"epoch": 0.4566153846153846,
|
| 7814 |
+
"grad_norm": 4.09375,
|
| 7815 |
+
"learning_rate": 2.7752197572205946e-06,
|
| 7816 |
+
"loss": 1.1646690368652344,
|
| 7817 |
+
"step": 1113
|
| 7818 |
+
},
|
| 7819 |
+
{
|
| 7820 |
+
"epoch": 0.457025641025641,
|
| 7821 |
+
"grad_norm": 4.09375,
|
| 7822 |
+
"learning_rate": 2.773126831310172e-06,
|
| 7823 |
+
"loss": 1.3407741785049438,
|
| 7824 |
+
"step": 1114
|
| 7825 |
+
},
|
| 7826 |
+
{
|
| 7827 |
+
"epoch": 0.4574358974358974,
|
| 7828 |
+
"grad_norm": 3.96875,
|
| 7829 |
+
"learning_rate": 2.7710339053997495e-06,
|
| 7830 |
+
"loss": 1.2757973670959473,
|
| 7831 |
+
"step": 1115
|
| 7832 |
+
},
|
| 7833 |
+
{
|
| 7834 |
+
"epoch": 0.45784615384615385,
|
| 7835 |
+
"grad_norm": 3.9375,
|
| 7836 |
+
"learning_rate": 2.7689409794893262e-06,
|
| 7837 |
+
"loss": 1.1767247915267944,
|
| 7838 |
+
"step": 1116
|
| 7839 |
+
},
|
| 7840 |
+
{
|
| 7841 |
+
"epoch": 0.45825641025641023,
|
| 7842 |
+
"grad_norm": 3.78125,
|
| 7843 |
+
"learning_rate": 2.7668480535789035e-06,
|
| 7844 |
+
"loss": 1.2286181449890137,
|
| 7845 |
+
"step": 1117
|
| 7846 |
+
},
|
| 7847 |
+
{
|
| 7848 |
+
"epoch": 0.45866666666666667,
|
| 7849 |
+
"grad_norm": 4.25,
|
| 7850 |
+
"learning_rate": 2.7647551276684807e-06,
|
| 7851 |
+
"loss": 1.3293908834457397,
|
| 7852 |
+
"step": 1118
|
| 7853 |
+
},
|
| 7854 |
+
{
|
| 7855 |
+
"epoch": 0.4590769230769231,
|
| 7856 |
+
"grad_norm": 4.40625,
|
| 7857 |
+
"learning_rate": 2.762662201758058e-06,
|
| 7858 |
+
"loss": 1.136747121810913,
|
| 7859 |
+
"step": 1119
|
| 7860 |
+
},
|
| 7861 |
+
{
|
| 7862 |
+
"epoch": 0.4594871794871795,
|
| 7863 |
+
"grad_norm": 4.0625,
|
| 7864 |
+
"learning_rate": 2.7605692758476355e-06,
|
| 7865 |
+
"loss": 1.2828847169876099,
|
| 7866 |
+
"step": 1120
|
| 7867 |
+
},
|
| 7868 |
+
{
|
| 7869 |
+
"epoch": 0.4598974358974359,
|
| 7870 |
+
"grad_norm": 4.34375,
|
| 7871 |
+
"learning_rate": 2.7584763499372123e-06,
|
| 7872 |
+
"loss": 1.2202463150024414,
|
| 7873 |
+
"step": 1121
|
| 7874 |
+
},
|
| 7875 |
+
{
|
| 7876 |
+
"epoch": 0.4603076923076923,
|
| 7877 |
+
"grad_norm": 4.15625,
|
| 7878 |
+
"learning_rate": 2.7563834240267895e-06,
|
| 7879 |
+
"loss": 1.1882953643798828,
|
| 7880 |
+
"step": 1122
|
| 7881 |
+
},
|
| 7882 |
+
{
|
| 7883 |
+
"epoch": 0.46071794871794874,
|
| 7884 |
+
"grad_norm": 4.15625,
|
| 7885 |
+
"learning_rate": 2.754290498116367e-06,
|
| 7886 |
+
"loss": 1.3246879577636719,
|
| 7887 |
+
"step": 1123
|
| 7888 |
+
},
|
| 7889 |
+
{
|
| 7890 |
+
"epoch": 0.4611282051282051,
|
| 7891 |
+
"grad_norm": 3.953125,
|
| 7892 |
+
"learning_rate": 2.752197572205944e-06,
|
| 7893 |
+
"loss": 1.0927704572677612,
|
| 7894 |
+
"step": 1124
|
| 7895 |
+
},
|
| 7896 |
+
{
|
| 7897 |
+
"epoch": 0.46153846153846156,
|
| 7898 |
+
"grad_norm": 4.09375,
|
| 7899 |
+
"learning_rate": 2.750104646295521e-06,
|
| 7900 |
+
"loss": 1.3022692203521729,
|
| 7901 |
+
"step": 1125
|
| 7902 |
+
},
|
| 7903 |
+
{
|
| 7904 |
+
"epoch": 0.46194871794871795,
|
| 7905 |
+
"grad_norm": 3.703125,
|
| 7906 |
+
"learning_rate": 2.748011720385099e-06,
|
| 7907 |
+
"loss": 1.1929343938827515,
|
| 7908 |
+
"step": 1126
|
| 7909 |
+
},
|
| 7910 |
+
{
|
| 7911 |
+
"epoch": 0.4623589743589744,
|
| 7912 |
+
"grad_norm": 3.796875,
|
| 7913 |
+
"learning_rate": 2.7459187944746756e-06,
|
| 7914 |
+
"loss": 1.1554863452911377,
|
| 7915 |
+
"step": 1127
|
| 7916 |
+
},
|
| 7917 |
+
{
|
| 7918 |
+
"epoch": 0.46276923076923077,
|
| 7919 |
+
"grad_norm": 3.96875,
|
| 7920 |
+
"learning_rate": 2.7438258685642532e-06,
|
| 7921 |
+
"loss": 1.3685423135757446,
|
| 7922 |
+
"step": 1128
|
| 7923 |
+
},
|
| 7924 |
+
{
|
| 7925 |
+
"epoch": 0.4631794871794872,
|
| 7926 |
+
"grad_norm": 4.5625,
|
| 7927 |
+
"learning_rate": 2.7417329426538305e-06,
|
| 7928 |
+
"loss": 1.3145275115966797,
|
| 7929 |
+
"step": 1129
|
| 7930 |
+
},
|
| 7931 |
+
{
|
| 7932 |
+
"epoch": 0.4635897435897436,
|
| 7933 |
+
"grad_norm": 4.15625,
|
| 7934 |
+
"learning_rate": 2.7396400167434072e-06,
|
| 7935 |
+
"loss": 1.197896957397461,
|
| 7936 |
+
"step": 1130
|
| 7937 |
+
},
|
| 7938 |
+
{
|
| 7939 |
+
"epoch": 0.464,
|
| 7940 |
+
"grad_norm": 4.5,
|
| 7941 |
+
"learning_rate": 2.737547090832985e-06,
|
| 7942 |
+
"loss": 1.3405158519744873,
|
| 7943 |
+
"step": 1131
|
| 7944 |
+
},
|
| 7945 |
+
{
|
| 7946 |
+
"epoch": 0.4644102564102564,
|
| 7947 |
+
"grad_norm": 3.703125,
|
| 7948 |
+
"learning_rate": 2.7354541649225617e-06,
|
| 7949 |
+
"loss": 1.1004047393798828,
|
| 7950 |
+
"step": 1132
|
| 7951 |
+
},
|
| 7952 |
+
{
|
| 7953 |
+
"epoch": 0.46482051282051284,
|
| 7954 |
+
"grad_norm": 4.25,
|
| 7955 |
+
"learning_rate": 2.7333612390121393e-06,
|
| 7956 |
+
"loss": 1.2117571830749512,
|
| 7957 |
+
"step": 1133
|
| 7958 |
+
},
|
| 7959 |
+
{
|
| 7960 |
+
"epoch": 0.4652307692307692,
|
| 7961 |
+
"grad_norm": 3.65625,
|
| 7962 |
+
"learning_rate": 2.7312683131017165e-06,
|
| 7963 |
+
"loss": 1.2892667055130005,
|
| 7964 |
+
"step": 1134
|
| 7965 |
+
},
|
| 7966 |
+
{
|
| 7967 |
+
"epoch": 0.46564102564102566,
|
| 7968 |
+
"grad_norm": 3.828125,
|
| 7969 |
+
"learning_rate": 2.7291753871912933e-06,
|
| 7970 |
+
"loss": 1.2790594100952148,
|
| 7971 |
+
"step": 1135
|
| 7972 |
+
},
|
| 7973 |
+
{
|
| 7974 |
+
"epoch": 0.46605128205128205,
|
| 7975 |
+
"grad_norm": 4.0625,
|
| 7976 |
+
"learning_rate": 2.727082461280871e-06,
|
| 7977 |
+
"loss": 1.1410282850265503,
|
| 7978 |
+
"step": 1136
|
| 7979 |
+
},
|
| 7980 |
+
{
|
| 7981 |
+
"epoch": 0.4664615384615385,
|
| 7982 |
+
"grad_norm": 4.375,
|
| 7983 |
+
"learning_rate": 2.724989535370448e-06,
|
| 7984 |
+
"loss": 1.228535532951355,
|
| 7985 |
+
"step": 1137
|
| 7986 |
+
},
|
| 7987 |
+
{
|
| 7988 |
+
"epoch": 0.46687179487179487,
|
| 7989 |
+
"grad_norm": 3.75,
|
| 7990 |
+
"learning_rate": 2.722896609460025e-06,
|
| 7991 |
+
"loss": 1.0462685823440552,
|
| 7992 |
+
"step": 1138
|
| 7993 |
+
},
|
| 7994 |
+
{
|
| 7995 |
+
"epoch": 0.4672820512820513,
|
| 7996 |
+
"grad_norm": 4.1875,
|
| 7997 |
+
"learning_rate": 2.7208036835496026e-06,
|
| 7998 |
+
"loss": 1.254673719406128,
|
| 7999 |
+
"step": 1139
|
| 8000 |
+
},
|
| 8001 |
+
{
|
| 8002 |
+
"epoch": 0.4676923076923077,
|
| 8003 |
+
"grad_norm": 3.9375,
|
| 8004 |
+
"learning_rate": 2.71871075763918e-06,
|
| 8005 |
+
"loss": 1.2705061435699463,
|
| 8006 |
+
"step": 1140
|
| 8007 |
+
},
|
| 8008 |
+
{
|
| 8009 |
+
"epoch": 0.4681025641025641,
|
| 8010 |
+
"grad_norm": 3.875,
|
| 8011 |
+
"learning_rate": 2.7166178317287575e-06,
|
| 8012 |
+
"loss": 1.1526612043380737,
|
| 8013 |
+
"step": 1141
|
| 8014 |
+
},
|
| 8015 |
+
{
|
| 8016 |
+
"epoch": 0.4685128205128205,
|
| 8017 |
+
"grad_norm": 4.1875,
|
| 8018 |
+
"learning_rate": 2.7145249058183342e-06,
|
| 8019 |
+
"loss": 1.2010191679000854,
|
| 8020 |
+
"step": 1142
|
| 8021 |
+
},
|
| 8022 |
+
{
|
| 8023 |
+
"epoch": 0.46892307692307694,
|
| 8024 |
+
"grad_norm": 4.21875,
|
| 8025 |
+
"learning_rate": 2.7124319799079115e-06,
|
| 8026 |
+
"loss": 1.27230703830719,
|
| 8027 |
+
"step": 1143
|
| 8028 |
+
},
|
| 8029 |
+
{
|
| 8030 |
+
"epoch": 0.4693333333333333,
|
| 8031 |
+
"grad_norm": 3.828125,
|
| 8032 |
+
"learning_rate": 2.7103390539974887e-06,
|
| 8033 |
+
"loss": 1.235460877418518,
|
| 8034 |
+
"step": 1144
|
| 8035 |
+
},
|
| 8036 |
+
{
|
| 8037 |
+
"epoch": 0.46974358974358976,
|
| 8038 |
+
"grad_norm": 4.0,
|
| 8039 |
+
"learning_rate": 2.708246128087066e-06,
|
| 8040 |
+
"loss": 1.3169441223144531,
|
| 8041 |
+
"step": 1145
|
| 8042 |
+
},
|
| 8043 |
+
{
|
| 8044 |
+
"epoch": 0.47015384615384614,
|
| 8045 |
+
"grad_norm": 4.0,
|
| 8046 |
+
"learning_rate": 2.706153202176643e-06,
|
| 8047 |
+
"loss": 1.30753493309021,
|
| 8048 |
+
"step": 1146
|
| 8049 |
+
},
|
| 8050 |
+
{
|
| 8051 |
+
"epoch": 0.4705641025641026,
|
| 8052 |
+
"grad_norm": 4.125,
|
| 8053 |
+
"learning_rate": 2.7040602762662203e-06,
|
| 8054 |
+
"loss": 1.279885172843933,
|
| 8055 |
+
"step": 1147
|
| 8056 |
+
},
|
| 8057 |
+
{
|
| 8058 |
+
"epoch": 0.47097435897435896,
|
| 8059 |
+
"grad_norm": 4.28125,
|
| 8060 |
+
"learning_rate": 2.7019673503557975e-06,
|
| 8061 |
+
"loss": 1.2754309177398682,
|
| 8062 |
+
"step": 1148
|
| 8063 |
+
},
|
| 8064 |
+
{
|
| 8065 |
+
"epoch": 0.4713846153846154,
|
| 8066 |
+
"grad_norm": 3.953125,
|
| 8067 |
+
"learning_rate": 2.699874424445375e-06,
|
| 8068 |
+
"loss": 1.2124710083007812,
|
| 8069 |
+
"step": 1149
|
| 8070 |
+
},
|
| 8071 |
+
{
|
| 8072 |
+
"epoch": 0.4717948717948718,
|
| 8073 |
+
"grad_norm": 3.9375,
|
| 8074 |
+
"learning_rate": 2.697781498534952e-06,
|
| 8075 |
+
"loss": 1.1597471237182617,
|
| 8076 |
+
"step": 1150
|
| 8077 |
+
},
|
| 8078 |
+
{
|
| 8079 |
+
"epoch": 0.4722051282051282,
|
| 8080 |
+
"grad_norm": 3.96875,
|
| 8081 |
+
"learning_rate": 2.695688572624529e-06,
|
| 8082 |
+
"loss": 1.1590266227722168,
|
| 8083 |
+
"step": 1151
|
| 8084 |
+
},
|
| 8085 |
+
{
|
| 8086 |
+
"epoch": 0.4726153846153846,
|
| 8087 |
+
"grad_norm": 4.15625,
|
| 8088 |
+
"learning_rate": 2.693595646714107e-06,
|
| 8089 |
+
"loss": 1.2750378847122192,
|
| 8090 |
+
"step": 1152
|
| 8091 |
+
},
|
| 8092 |
+
{
|
| 8093 |
+
"epoch": 0.47302564102564104,
|
| 8094 |
+
"grad_norm": 4.09375,
|
| 8095 |
+
"learning_rate": 2.6915027208036836e-06,
|
| 8096 |
+
"loss": 1.2880383729934692,
|
| 8097 |
+
"step": 1153
|
| 8098 |
+
},
|
| 8099 |
+
{
|
| 8100 |
+
"epoch": 0.4734358974358974,
|
| 8101 |
+
"grad_norm": 3.765625,
|
| 8102 |
+
"learning_rate": 2.6894097948932612e-06,
|
| 8103 |
+
"loss": 1.1571751832962036,
|
| 8104 |
+
"step": 1154
|
| 8105 |
+
},
|
| 8106 |
+
{
|
| 8107 |
+
"epoch": 0.47384615384615386,
|
| 8108 |
+
"grad_norm": 4.21875,
|
| 8109 |
+
"learning_rate": 2.6873168689828385e-06,
|
| 8110 |
+
"loss": 1.1115379333496094,
|
| 8111 |
+
"step": 1155
|
| 8112 |
+
},
|
| 8113 |
+
{
|
| 8114 |
+
"epoch": 0.47425641025641024,
|
| 8115 |
+
"grad_norm": 3.703125,
|
| 8116 |
+
"learning_rate": 2.6852239430724152e-06,
|
| 8117 |
+
"loss": 1.183795690536499,
|
| 8118 |
+
"step": 1156
|
| 8119 |
+
},
|
| 8120 |
+
{
|
| 8121 |
+
"epoch": 0.4746666666666667,
|
| 8122 |
+
"grad_norm": 4.125,
|
| 8123 |
+
"learning_rate": 2.683131017161993e-06,
|
| 8124 |
+
"loss": 0.972062885761261,
|
| 8125 |
+
"step": 1157
|
| 8126 |
+
},
|
| 8127 |
+
{
|
| 8128 |
+
"epoch": 0.47507692307692306,
|
| 8129 |
+
"grad_norm": 4.125,
|
| 8130 |
+
"learning_rate": 2.68103809125157e-06,
|
| 8131 |
+
"loss": 1.2399051189422607,
|
| 8132 |
+
"step": 1158
|
| 8133 |
+
},
|
| 8134 |
+
{
|
| 8135 |
+
"epoch": 0.4754871794871795,
|
| 8136 |
+
"grad_norm": 4.0,
|
| 8137 |
+
"learning_rate": 2.678945165341147e-06,
|
| 8138 |
+
"loss": 1.2703721523284912,
|
| 8139 |
+
"step": 1159
|
| 8140 |
+
},
|
| 8141 |
+
{
|
| 8142 |
+
"epoch": 0.4758974358974359,
|
| 8143 |
+
"grad_norm": 3.8125,
|
| 8144 |
+
"learning_rate": 2.6768522394307245e-06,
|
| 8145 |
+
"loss": 1.201429009437561,
|
| 8146 |
+
"step": 1160
|
| 8147 |
+
},
|
| 8148 |
+
{
|
| 8149 |
+
"epoch": 0.4763076923076923,
|
| 8150 |
+
"grad_norm": 4.75,
|
| 8151 |
+
"learning_rate": 2.6747593135203013e-06,
|
| 8152 |
+
"loss": 1.3240954875946045,
|
| 8153 |
+
"step": 1161
|
| 8154 |
+
},
|
| 8155 |
+
{
|
| 8156 |
+
"epoch": 0.4767179487179487,
|
| 8157 |
+
"grad_norm": 3.90625,
|
| 8158 |
+
"learning_rate": 2.672666387609879e-06,
|
| 8159 |
+
"loss": 1.3501107692718506,
|
| 8160 |
+
"step": 1162
|
| 8161 |
+
},
|
| 8162 |
+
{
|
| 8163 |
+
"epoch": 0.47712820512820514,
|
| 8164 |
+
"grad_norm": 4.1875,
|
| 8165 |
+
"learning_rate": 2.670573461699456e-06,
|
| 8166 |
+
"loss": 1.0815821886062622,
|
| 8167 |
+
"step": 1163
|
| 8168 |
+
},
|
| 8169 |
+
{
|
| 8170 |
+
"epoch": 0.4775384615384615,
|
| 8171 |
+
"grad_norm": 4.3125,
|
| 8172 |
+
"learning_rate": 2.668480535789033e-06,
|
| 8173 |
+
"loss": 1.1292061805725098,
|
| 8174 |
+
"step": 1164
|
| 8175 |
+
},
|
| 8176 |
+
{
|
| 8177 |
+
"epoch": 0.47794871794871796,
|
| 8178 |
+
"grad_norm": 3.8125,
|
| 8179 |
+
"learning_rate": 2.6663876098786106e-06,
|
| 8180 |
+
"loss": 1.2564899921417236,
|
| 8181 |
+
"step": 1165
|
| 8182 |
+
},
|
| 8183 |
+
{
|
| 8184 |
+
"epoch": 0.47835897435897434,
|
| 8185 |
+
"grad_norm": 3.96875,
|
| 8186 |
+
"learning_rate": 2.664294683968188e-06,
|
| 8187 |
+
"loss": 1.2183459997177124,
|
| 8188 |
+
"step": 1166
|
| 8189 |
+
},
|
| 8190 |
+
{
|
| 8191 |
+
"epoch": 0.4787692307692308,
|
| 8192 |
+
"grad_norm": 4.28125,
|
| 8193 |
+
"learning_rate": 2.6622017580577646e-06,
|
| 8194 |
+
"loss": 1.3162555694580078,
|
| 8195 |
+
"step": 1167
|
| 8196 |
+
},
|
| 8197 |
+
{
|
| 8198 |
+
"epoch": 0.47917948717948716,
|
| 8199 |
+
"grad_norm": 4.1875,
|
| 8200 |
+
"learning_rate": 2.6601088321473422e-06,
|
| 8201 |
+
"loss": 1.3321352005004883,
|
| 8202 |
+
"step": 1168
|
| 8203 |
+
},
|
| 8204 |
+
{
|
| 8205 |
+
"epoch": 0.4795897435897436,
|
| 8206 |
+
"grad_norm": 3.984375,
|
| 8207 |
+
"learning_rate": 2.6580159062369195e-06,
|
| 8208 |
+
"loss": 1.3019156455993652,
|
| 8209 |
+
"step": 1169
|
| 8210 |
+
},
|
| 8211 |
+
{
|
| 8212 |
+
"epoch": 0.48,
|
| 8213 |
+
"grad_norm": 3.9375,
|
| 8214 |
+
"learning_rate": 2.6559229803264967e-06,
|
| 8215 |
+
"loss": 1.1561474800109863,
|
| 8216 |
+
"step": 1170
|
| 8217 |
+
},
|
| 8218 |
+
{
|
| 8219 |
+
"epoch": 0.4804102564102564,
|
| 8220 |
+
"grad_norm": 3.828125,
|
| 8221 |
+
"learning_rate": 2.653830054416074e-06,
|
| 8222 |
+
"loss": 1.0528982877731323,
|
| 8223 |
+
"step": 1171
|
| 8224 |
+
},
|
| 8225 |
+
{
|
| 8226 |
+
"epoch": 0.4808205128205128,
|
| 8227 |
+
"grad_norm": 4.0625,
|
| 8228 |
+
"learning_rate": 2.651737128505651e-06,
|
| 8229 |
+
"loss": 1.211184024810791,
|
| 8230 |
+
"step": 1172
|
| 8231 |
+
},
|
| 8232 |
+
{
|
| 8233 |
+
"epoch": 0.48123076923076924,
|
| 8234 |
+
"grad_norm": 4.15625,
|
| 8235 |
+
"learning_rate": 2.6496442025952283e-06,
|
| 8236 |
+
"loss": 1.3078688383102417,
|
| 8237 |
+
"step": 1173
|
| 8238 |
+
},
|
| 8239 |
+
{
|
| 8240 |
+
"epoch": 0.4816410256410256,
|
| 8241 |
+
"grad_norm": 4.34375,
|
| 8242 |
+
"learning_rate": 2.6475512766848055e-06,
|
| 8243 |
+
"loss": 1.0660418272018433,
|
| 8244 |
+
"step": 1174
|
| 8245 |
+
},
|
| 8246 |
+
{
|
| 8247 |
+
"epoch": 0.48205128205128206,
|
| 8248 |
+
"grad_norm": 4.90625,
|
| 8249 |
+
"learning_rate": 2.645458350774383e-06,
|
| 8250 |
+
"loss": 1.42172110080719,
|
| 8251 |
+
"step": 1175
|
| 8252 |
+
},
|
| 8253 |
+
{
|
| 8254 |
+
"epoch": 0.48246153846153844,
|
| 8255 |
+
"grad_norm": 3.96875,
|
| 8256 |
+
"learning_rate": 2.64336542486396e-06,
|
| 8257 |
+
"loss": 1.12493896484375,
|
| 8258 |
+
"step": 1176
|
| 8259 |
+
},
|
| 8260 |
+
{
|
| 8261 |
+
"epoch": 0.4828717948717949,
|
| 8262 |
+
"grad_norm": 4.15625,
|
| 8263 |
+
"learning_rate": 2.641272498953537e-06,
|
| 8264 |
+
"loss": 1.1366249322891235,
|
| 8265 |
+
"step": 1177
|
| 8266 |
+
},
|
| 8267 |
+
{
|
| 8268 |
+
"epoch": 0.48328205128205126,
|
| 8269 |
+
"grad_norm": 3.65625,
|
| 8270 |
+
"learning_rate": 2.639179573043115e-06,
|
| 8271 |
+
"loss": 1.2538044452667236,
|
| 8272 |
+
"step": 1178
|
| 8273 |
+
},
|
| 8274 |
+
{
|
| 8275 |
+
"epoch": 0.4836923076923077,
|
| 8276 |
+
"grad_norm": 4.09375,
|
| 8277 |
+
"learning_rate": 2.6370866471326916e-06,
|
| 8278 |
+
"loss": 1.2119715213775635,
|
| 8279 |
+
"step": 1179
|
| 8280 |
+
},
|
| 8281 |
+
{
|
| 8282 |
+
"epoch": 0.4841025641025641,
|
| 8283 |
+
"grad_norm": 4.21875,
|
| 8284 |
+
"learning_rate": 2.634993721222269e-06,
|
| 8285 |
+
"loss": 1.2946739196777344,
|
| 8286 |
+
"step": 1180
|
| 8287 |
+
},
|
| 8288 |
+
{
|
| 8289 |
+
"epoch": 0.4845128205128205,
|
| 8290 |
+
"grad_norm": 4.0625,
|
| 8291 |
+
"learning_rate": 2.6329007953118465e-06,
|
| 8292 |
+
"loss": 1.2397691011428833,
|
| 8293 |
+
"step": 1181
|
| 8294 |
+
},
|
| 8295 |
+
{
|
| 8296 |
+
"epoch": 0.4849230769230769,
|
| 8297 |
+
"grad_norm": 4.28125,
|
| 8298 |
+
"learning_rate": 2.6308078694014232e-06,
|
| 8299 |
+
"loss": 1.149743676185608,
|
| 8300 |
+
"step": 1182
|
| 8301 |
+
},
|
| 8302 |
+
{
|
| 8303 |
+
"epoch": 0.48533333333333334,
|
| 8304 |
+
"grad_norm": 4.0625,
|
| 8305 |
+
"learning_rate": 2.628714943491001e-06,
|
| 8306 |
+
"loss": 1.173339605331421,
|
| 8307 |
+
"step": 1183
|
| 8308 |
+
},
|
| 8309 |
+
{
|
| 8310 |
+
"epoch": 0.4857435897435897,
|
| 8311 |
+
"grad_norm": 4.40625,
|
| 8312 |
+
"learning_rate": 2.626622017580578e-06,
|
| 8313 |
+
"loss": 1.2582676410675049,
|
| 8314 |
+
"step": 1184
|
| 8315 |
+
},
|
| 8316 |
+
{
|
| 8317 |
+
"epoch": 0.48615384615384616,
|
| 8318 |
+
"grad_norm": 4.4375,
|
| 8319 |
+
"learning_rate": 2.624529091670155e-06,
|
| 8320 |
+
"loss": 1.2588952779769897,
|
| 8321 |
+
"step": 1185
|
| 8322 |
+
},
|
| 8323 |
+
{
|
| 8324 |
+
"epoch": 0.48656410256410254,
|
| 8325 |
+
"grad_norm": 3.578125,
|
| 8326 |
+
"learning_rate": 2.6224361657597325e-06,
|
| 8327 |
+
"loss": 1.168443202972412,
|
| 8328 |
+
"step": 1186
|
| 8329 |
+
},
|
| 8330 |
+
{
|
| 8331 |
+
"epoch": 0.486974358974359,
|
| 8332 |
+
"grad_norm": 4.0,
|
| 8333 |
+
"learning_rate": 2.6203432398493093e-06,
|
| 8334 |
+
"loss": 1.1664633750915527,
|
| 8335 |
+
"step": 1187
|
| 8336 |
+
},
|
| 8337 |
+
{
|
| 8338 |
+
"epoch": 0.48738461538461536,
|
| 8339 |
+
"grad_norm": 3.5625,
|
| 8340 |
+
"learning_rate": 2.6182503139388865e-06,
|
| 8341 |
+
"loss": 1.0813941955566406,
|
| 8342 |
+
"step": 1188
|
| 8343 |
+
},
|
| 8344 |
+
{
|
| 8345 |
+
"epoch": 0.4877948717948718,
|
| 8346 |
+
"grad_norm": 4.03125,
|
| 8347 |
+
"learning_rate": 2.616157388028464e-06,
|
| 8348 |
+
"loss": 1.1876239776611328,
|
| 8349 |
+
"step": 1189
|
| 8350 |
+
},
|
| 8351 |
+
{
|
| 8352 |
+
"epoch": 0.4882051282051282,
|
| 8353 |
+
"grad_norm": 3.875,
|
| 8354 |
+
"learning_rate": 2.614064462118041e-06,
|
| 8355 |
+
"loss": 1.088321328163147,
|
| 8356 |
+
"step": 1190
|
| 8357 |
+
},
|
| 8358 |
+
{
|
| 8359 |
+
"epoch": 0.4886153846153846,
|
| 8360 |
+
"grad_norm": 3.890625,
|
| 8361 |
+
"learning_rate": 2.6119715362076186e-06,
|
| 8362 |
+
"loss": 1.1338192224502563,
|
| 8363 |
+
"step": 1191
|
| 8364 |
+
},
|
| 8365 |
+
{
|
| 8366 |
+
"epoch": 0.489025641025641,
|
| 8367 |
+
"grad_norm": 3.953125,
|
| 8368 |
+
"learning_rate": 2.609878610297196e-06,
|
| 8369 |
+
"loss": 1.2346575260162354,
|
| 8370 |
+
"step": 1192
|
| 8371 |
+
},
|
| 8372 |
+
{
|
| 8373 |
+
"epoch": 0.48943589743589744,
|
| 8374 |
+
"grad_norm": 4.0,
|
| 8375 |
+
"learning_rate": 2.6077856843867726e-06,
|
| 8376 |
+
"loss": 1.2607864141464233,
|
| 8377 |
+
"step": 1193
|
| 8378 |
+
},
|
| 8379 |
+
{
|
| 8380 |
+
"epoch": 0.4898461538461538,
|
| 8381 |
+
"grad_norm": 4.46875,
|
| 8382 |
+
"learning_rate": 2.6056927584763502e-06,
|
| 8383 |
+
"loss": 1.3188421726226807,
|
| 8384 |
+
"step": 1194
|
| 8385 |
+
},
|
| 8386 |
+
{
|
| 8387 |
+
"epoch": 0.49025641025641026,
|
| 8388 |
+
"grad_norm": 4.40625,
|
| 8389 |
+
"learning_rate": 2.6035998325659275e-06,
|
| 8390 |
+
"loss": 1.1725444793701172,
|
| 8391 |
+
"step": 1195
|
| 8392 |
+
},
|
| 8393 |
+
{
|
| 8394 |
+
"epoch": 0.49066666666666664,
|
| 8395 |
+
"grad_norm": 3.875,
|
| 8396 |
+
"learning_rate": 2.6015069066555047e-06,
|
| 8397 |
+
"loss": 1.0022180080413818,
|
| 8398 |
+
"step": 1196
|
| 8399 |
+
},
|
| 8400 |
+
{
|
| 8401 |
+
"epoch": 0.4910769230769231,
|
| 8402 |
+
"grad_norm": 3.859375,
|
| 8403 |
+
"learning_rate": 2.599413980745082e-06,
|
| 8404 |
+
"loss": 0.971603274345398,
|
| 8405 |
+
"step": 1197
|
| 8406 |
+
},
|
| 8407 |
+
{
|
| 8408 |
+
"epoch": 0.49148717948717946,
|
| 8409 |
+
"grad_norm": 4.28125,
|
| 8410 |
+
"learning_rate": 2.597321054834659e-06,
|
| 8411 |
+
"loss": 1.2265708446502686,
|
| 8412 |
+
"step": 1198
|
| 8413 |
+
},
|
| 8414 |
+
{
|
| 8415 |
+
"epoch": 0.4918974358974359,
|
| 8416 |
+
"grad_norm": 3.875,
|
| 8417 |
+
"learning_rate": 2.5952281289242363e-06,
|
| 8418 |
+
"loss": 1.2679684162139893,
|
| 8419 |
+
"step": 1199
|
| 8420 |
+
},
|
| 8421 |
+
{
|
| 8422 |
+
"epoch": 0.49230769230769234,
|
| 8423 |
+
"grad_norm": 4.0625,
|
| 8424 |
+
"learning_rate": 2.5931352030138135e-06,
|
| 8425 |
+
"loss": 1.0979076623916626,
|
| 8426 |
+
"step": 1200
|
| 8427 |
+
},
|
| 8428 |
+
{
|
| 8429 |
+
"epoch": 0.4927179487179487,
|
| 8430 |
+
"grad_norm": 4.21875,
|
| 8431 |
+
"learning_rate": 2.5910422771033903e-06,
|
| 8432 |
+
"loss": 1.3017851114273071,
|
| 8433 |
+
"step": 1201
|
| 8434 |
+
},
|
| 8435 |
+
{
|
| 8436 |
+
"epoch": 0.49312820512820515,
|
| 8437 |
+
"grad_norm": 4.625,
|
| 8438 |
+
"learning_rate": 2.588949351192968e-06,
|
| 8439 |
+
"loss": 1.3456947803497314,
|
| 8440 |
+
"step": 1202
|
| 8441 |
+
},
|
| 8442 |
+
{
|
| 8443 |
+
"epoch": 0.49353846153846154,
|
| 8444 |
+
"grad_norm": 4.21875,
|
| 8445 |
+
"learning_rate": 2.586856425282545e-06,
|
| 8446 |
+
"loss": 1.1733098030090332,
|
| 8447 |
+
"step": 1203
|
| 8448 |
+
},
|
| 8449 |
+
{
|
| 8450 |
+
"epoch": 0.493948717948718,
|
| 8451 |
+
"grad_norm": 4.34375,
|
| 8452 |
+
"learning_rate": 2.584763499372123e-06,
|
| 8453 |
+
"loss": 1.0947279930114746,
|
| 8454 |
+
"step": 1204
|
| 8455 |
+
},
|
| 8456 |
+
{
|
| 8457 |
+
"epoch": 0.49435897435897436,
|
| 8458 |
+
"grad_norm": 3.953125,
|
| 8459 |
+
"learning_rate": 2.5826705734616996e-06,
|
| 8460 |
+
"loss": 0.9675426483154297,
|
| 8461 |
+
"step": 1205
|
| 8462 |
+
},
|
| 8463 |
+
{
|
| 8464 |
+
"epoch": 0.4947692307692308,
|
| 8465 |
+
"grad_norm": 3.703125,
|
| 8466 |
+
"learning_rate": 2.580577647551277e-06,
|
| 8467 |
+
"loss": 1.143105387687683,
|
| 8468 |
+
"step": 1206
|
| 8469 |
+
},
|
| 8470 |
+
{
|
| 8471 |
+
"epoch": 0.4951794871794872,
|
| 8472 |
+
"grad_norm": 3.546875,
|
| 8473 |
+
"learning_rate": 2.5784847216408545e-06,
|
| 8474 |
+
"loss": 1.179133653640747,
|
| 8475 |
+
"step": 1207
|
| 8476 |
+
},
|
| 8477 |
+
{
|
| 8478 |
+
"epoch": 0.4955897435897436,
|
| 8479 |
+
"grad_norm": 3.921875,
|
| 8480 |
+
"learning_rate": 2.5763917957304312e-06,
|
| 8481 |
+
"loss": 1.1521447896957397,
|
| 8482 |
+
"step": 1208
|
| 8483 |
+
},
|
| 8484 |
+
{
|
| 8485 |
+
"epoch": 0.496,
|
| 8486 |
+
"grad_norm": 3.875,
|
| 8487 |
+
"learning_rate": 2.5742988698200085e-06,
|
| 8488 |
+
"loss": 1.1445086002349854,
|
| 8489 |
+
"step": 1209
|
| 8490 |
+
},
|
| 8491 |
+
{
|
| 8492 |
+
"epoch": 0.49641025641025643,
|
| 8493 |
+
"grad_norm": 4.09375,
|
| 8494 |
+
"learning_rate": 2.572205943909586e-06,
|
| 8495 |
+
"loss": 1.2033452987670898,
|
| 8496 |
+
"step": 1210
|
| 8497 |
+
},
|
| 8498 |
+
{
|
| 8499 |
+
"epoch": 0.4968205128205128,
|
| 8500 |
+
"grad_norm": 3.9375,
|
| 8501 |
+
"learning_rate": 2.570113017999163e-06,
|
| 8502 |
+
"loss": 1.1786425113677979,
|
| 8503 |
+
"step": 1211
|
| 8504 |
+
},
|
| 8505 |
+
{
|
| 8506 |
+
"epoch": 0.49723076923076925,
|
| 8507 |
+
"grad_norm": 4.40625,
|
| 8508 |
+
"learning_rate": 2.5680200920887405e-06,
|
| 8509 |
+
"loss": 1.2034565210342407,
|
| 8510 |
+
"step": 1212
|
| 8511 |
+
},
|
| 8512 |
+
{
|
| 8513 |
+
"epoch": 0.49764102564102564,
|
| 8514 |
+
"grad_norm": 3.84375,
|
| 8515 |
+
"learning_rate": 2.5659271661783173e-06,
|
| 8516 |
+
"loss": 1.2323551177978516,
|
| 8517 |
+
"step": 1213
|
| 8518 |
+
},
|
| 8519 |
+
{
|
| 8520 |
+
"epoch": 0.4980512820512821,
|
| 8521 |
+
"grad_norm": 3.703125,
|
| 8522 |
+
"learning_rate": 2.5638342402678945e-06,
|
| 8523 |
+
"loss": 1.1349540948867798,
|
| 8524 |
+
"step": 1214
|
| 8525 |
+
},
|
| 8526 |
+
{
|
| 8527 |
+
"epoch": 0.49846153846153846,
|
| 8528 |
+
"grad_norm": 3.921875,
|
| 8529 |
+
"learning_rate": 2.561741314357472e-06,
|
| 8530 |
+
"loss": 1.1817114353179932,
|
| 8531 |
+
"step": 1215
|
| 8532 |
+
},
|
| 8533 |
+
{
|
| 8534 |
+
"epoch": 0.4988717948717949,
|
| 8535 |
+
"grad_norm": 4.1875,
|
| 8536 |
+
"learning_rate": 2.559648388447049e-06,
|
| 8537 |
+
"loss": 1.3258721828460693,
|
| 8538 |
+
"step": 1216
|
| 8539 |
+
},
|
| 8540 |
+
{
|
| 8541 |
+
"epoch": 0.4992820512820513,
|
| 8542 |
+
"grad_norm": 4.4375,
|
| 8543 |
+
"learning_rate": 2.5575554625366266e-06,
|
| 8544 |
+
"loss": 1.2787827253341675,
|
| 8545 |
+
"step": 1217
|
| 8546 |
+
},
|
| 8547 |
+
{
|
| 8548 |
+
"epoch": 0.4996923076923077,
|
| 8549 |
+
"grad_norm": 3.96875,
|
| 8550 |
+
"learning_rate": 2.555462536626204e-06,
|
| 8551 |
+
"loss": 1.4378894567489624,
|
| 8552 |
+
"step": 1218
|
| 8553 |
+
},
|
| 8554 |
+
{
|
| 8555 |
+
"epoch": 0.5001025641025642,
|
| 8556 |
+
"grad_norm": 3.921875,
|
| 8557 |
+
"learning_rate": 2.5533696107157806e-06,
|
| 8558 |
+
"loss": 1.2369979619979858,
|
| 8559 |
+
"step": 1219
|
| 8560 |
+
},
|
| 8561 |
+
{
|
| 8562 |
+
"epoch": 0.5005128205128205,
|
| 8563 |
+
"grad_norm": 4.09375,
|
| 8564 |
+
"learning_rate": 2.5512766848053582e-06,
|
| 8565 |
+
"loss": 1.1508114337921143,
|
| 8566 |
+
"step": 1220
|
| 8567 |
+
},
|
| 8568 |
+
{
|
| 8569 |
+
"epoch": 0.5009230769230769,
|
| 8570 |
+
"grad_norm": 3.65625,
|
| 8571 |
+
"learning_rate": 2.5491837588949355e-06,
|
| 8572 |
+
"loss": 1.1524691581726074,
|
| 8573 |
+
"step": 1221
|
| 8574 |
+
},
|
| 8575 |
+
{
|
| 8576 |
+
"epoch": 0.5013333333333333,
|
| 8577 |
+
"grad_norm": 3.421875,
|
| 8578 |
+
"learning_rate": 2.5470908329845122e-06,
|
| 8579 |
+
"loss": 1.047582983970642,
|
| 8580 |
+
"step": 1222
|
| 8581 |
+
},
|
| 8582 |
+
{
|
| 8583 |
+
"epoch": 0.5017435897435898,
|
| 8584 |
+
"grad_norm": 3.890625,
|
| 8585 |
+
"learning_rate": 2.54499790707409e-06,
|
| 8586 |
+
"loss": 1.2225149869918823,
|
| 8587 |
+
"step": 1223
|
| 8588 |
+
},
|
| 8589 |
+
{
|
| 8590 |
+
"epoch": 0.5021538461538462,
|
| 8591 |
+
"grad_norm": 4.125,
|
| 8592 |
+
"learning_rate": 2.542904981163667e-06,
|
| 8593 |
+
"loss": 1.1423752307891846,
|
| 8594 |
+
"step": 1224
|
| 8595 |
+
},
|
| 8596 |
+
{
|
| 8597 |
+
"epoch": 0.5025641025641026,
|
| 8598 |
+
"grad_norm": 3.890625,
|
| 8599 |
+
"learning_rate": 2.5408120552532443e-06,
|
| 8600 |
+
"loss": 1.114809274673462,
|
| 8601 |
+
"step": 1225
|
| 8602 |
+
},
|
| 8603 |
+
{
|
| 8604 |
+
"epoch": 0.5029743589743589,
|
| 8605 |
+
"grad_norm": 4.21875,
|
| 8606 |
+
"learning_rate": 2.5387191293428215e-06,
|
| 8607 |
+
"loss": 1.1651010513305664,
|
| 8608 |
+
"step": 1226
|
| 8609 |
+
},
|
| 8610 |
+
{
|
| 8611 |
+
"epoch": 0.5033846153846154,
|
| 8612 |
+
"grad_norm": 4.3125,
|
| 8613 |
+
"learning_rate": 2.5366262034323983e-06,
|
| 8614 |
+
"loss": 1.1728034019470215,
|
| 8615 |
+
"step": 1227
|
| 8616 |
+
},
|
| 8617 |
+
{
|
| 8618 |
+
"epoch": 0.5037948717948718,
|
| 8619 |
+
"grad_norm": 4.03125,
|
| 8620 |
+
"learning_rate": 2.534533277521976e-06,
|
| 8621 |
+
"loss": 1.1965000629425049,
|
| 8622 |
+
"step": 1228
|
| 8623 |
+
},
|
| 8624 |
+
{
|
| 8625 |
+
"epoch": 0.5042051282051282,
|
| 8626 |
+
"grad_norm": 3.984375,
|
| 8627 |
+
"learning_rate": 2.532440351611553e-06,
|
| 8628 |
+
"loss": 1.3585489988327026,
|
| 8629 |
+
"step": 1229
|
| 8630 |
+
},
|
| 8631 |
+
{
|
| 8632 |
+
"epoch": 0.5046153846153846,
|
| 8633 |
+
"grad_norm": 4.0625,
|
| 8634 |
+
"learning_rate": 2.53034742570113e-06,
|
| 8635 |
+
"loss": 1.411143183708191,
|
| 8636 |
+
"step": 1230
|
| 8637 |
+
},
|
| 8638 |
+
{
|
| 8639 |
+
"epoch": 0.5050256410256411,
|
| 8640 |
+
"grad_norm": 4.1875,
|
| 8641 |
+
"learning_rate": 2.5282544997907076e-06,
|
| 8642 |
+
"loss": 1.41710364818573,
|
| 8643 |
+
"step": 1231
|
| 8644 |
+
},
|
| 8645 |
+
{
|
| 8646 |
+
"epoch": 0.5054358974358975,
|
| 8647 |
+
"grad_norm": 3.890625,
|
| 8648 |
+
"learning_rate": 2.526161573880285e-06,
|
| 8649 |
+
"loss": 1.0846178531646729,
|
| 8650 |
+
"step": 1232
|
| 8651 |
+
},
|
| 8652 |
+
{
|
| 8653 |
+
"epoch": 0.5058461538461538,
|
| 8654 |
+
"grad_norm": 4.21875,
|
| 8655 |
+
"learning_rate": 2.5240686479698625e-06,
|
| 8656 |
+
"loss": 1.1966477632522583,
|
| 8657 |
+
"step": 1233
|
| 8658 |
+
},
|
| 8659 |
+
{
|
| 8660 |
+
"epoch": 0.5062564102564102,
|
| 8661 |
+
"grad_norm": 4.59375,
|
| 8662 |
+
"learning_rate": 2.5219757220594392e-06,
|
| 8663 |
+
"loss": 1.2536089420318604,
|
| 8664 |
+
"step": 1234
|
| 8665 |
+
},
|
| 8666 |
+
{
|
| 8667 |
+
"epoch": 0.5066666666666667,
|
| 8668 |
+
"grad_norm": 3.875,
|
| 8669 |
+
"learning_rate": 2.5198827961490165e-06,
|
| 8670 |
+
"loss": 1.1332647800445557,
|
| 8671 |
+
"step": 1235
|
| 8672 |
+
},
|
| 8673 |
+
{
|
| 8674 |
+
"epoch": 0.5070769230769231,
|
| 8675 |
+
"grad_norm": 4.1875,
|
| 8676 |
+
"learning_rate": 2.517789870238594e-06,
|
| 8677 |
+
"loss": 1.3640671968460083,
|
| 8678 |
+
"step": 1236
|
| 8679 |
+
},
|
| 8680 |
+
{
|
| 8681 |
+
"epoch": 0.5074871794871795,
|
| 8682 |
+
"grad_norm": 3.875,
|
| 8683 |
+
"learning_rate": 2.515696944328171e-06,
|
| 8684 |
+
"loss": 1.1429142951965332,
|
| 8685 |
+
"step": 1237
|
| 8686 |
+
},
|
| 8687 |
+
{
|
| 8688 |
+
"epoch": 0.5078974358974359,
|
| 8689 |
+
"grad_norm": 3.828125,
|
| 8690 |
+
"learning_rate": 2.5136040184177485e-06,
|
| 8691 |
+
"loss": 1.1598784923553467,
|
| 8692 |
+
"step": 1238
|
| 8693 |
+
},
|
| 8694 |
+
{
|
| 8695 |
+
"epoch": 0.5083076923076923,
|
| 8696 |
+
"grad_norm": 4.84375,
|
| 8697 |
+
"learning_rate": 2.5115110925073253e-06,
|
| 8698 |
+
"loss": 1.3633712530136108,
|
| 8699 |
+
"step": 1239
|
| 8700 |
+
},
|
| 8701 |
+
{
|
| 8702 |
+
"epoch": 0.5087179487179487,
|
| 8703 |
+
"grad_norm": 3.75,
|
| 8704 |
+
"learning_rate": 2.5094181665969025e-06,
|
| 8705 |
+
"loss": 1.2264748811721802,
|
| 8706 |
+
"step": 1240
|
| 8707 |
+
},
|
| 8708 |
+
{
|
| 8709 |
+
"epoch": 0.5091282051282051,
|
| 8710 |
+
"grad_norm": 3.875,
|
| 8711 |
+
"learning_rate": 2.50732524068648e-06,
|
| 8712 |
+
"loss": 1.1186363697052002,
|
| 8713 |
+
"step": 1241
|
| 8714 |
+
},
|
| 8715 |
+
{
|
| 8716 |
+
"epoch": 0.5095384615384615,
|
| 8717 |
+
"grad_norm": 3.9375,
|
| 8718 |
+
"learning_rate": 2.505232314776057e-06,
|
| 8719 |
+
"loss": 1.0515657663345337,
|
| 8720 |
+
"step": 1242
|
| 8721 |
+
},
|
| 8722 |
+
{
|
| 8723 |
+
"epoch": 0.509948717948718,
|
| 8724 |
+
"grad_norm": 4.28125,
|
| 8725 |
+
"learning_rate": 2.503139388865634e-06,
|
| 8726 |
+
"loss": 1.322778582572937,
|
| 8727 |
+
"step": 1243
|
| 8728 |
+
},
|
| 8729 |
+
{
|
| 8730 |
+
"epoch": 0.5103589743589744,
|
| 8731 |
+
"grad_norm": 4.3125,
|
| 8732 |
+
"learning_rate": 2.501046462955212e-06,
|
| 8733 |
+
"loss": 1.3161189556121826,
|
| 8734 |
+
"step": 1244
|
| 8735 |
+
},
|
| 8736 |
+
{
|
| 8737 |
+
"epoch": 0.5107692307692308,
|
| 8738 |
+
"grad_norm": 3.578125,
|
| 8739 |
+
"learning_rate": 2.4989535370447886e-06,
|
| 8740 |
+
"loss": 1.1348364353179932,
|
| 8741 |
+
"step": 1245
|
| 8742 |
+
},
|
| 8743 |
+
{
|
| 8744 |
+
"epoch": 0.5111794871794871,
|
| 8745 |
+
"grad_norm": 4.28125,
|
| 8746 |
+
"learning_rate": 2.496860611134366e-06,
|
| 8747 |
+
"loss": 1.2516742944717407,
|
| 8748 |
+
"step": 1246
|
| 8749 |
+
},
|
| 8750 |
+
{
|
| 8751 |
+
"epoch": 0.5115897435897436,
|
| 8752 |
+
"grad_norm": 3.90625,
|
| 8753 |
+
"learning_rate": 2.4947676852239435e-06,
|
| 8754 |
+
"loss": 1.3633825778961182,
|
| 8755 |
+
"step": 1247
|
| 8756 |
+
},
|
| 8757 |
+
{
|
| 8758 |
+
"epoch": 0.512,
|
| 8759 |
+
"grad_norm": 3.828125,
|
| 8760 |
+
"learning_rate": 2.4926747593135207e-06,
|
| 8761 |
+
"loss": 1.209015130996704,
|
| 8762 |
+
"step": 1248
|
| 8763 |
+
},
|
| 8764 |
+
{
|
| 8765 |
+
"epoch": 0.5124102564102564,
|
| 8766 |
+
"grad_norm": 4.09375,
|
| 8767 |
+
"learning_rate": 2.4905818334030975e-06,
|
| 8768 |
+
"loss": 1.1872485876083374,
|
| 8769 |
+
"step": 1249
|
| 8770 |
+
},
|
| 8771 |
+
{
|
| 8772 |
+
"epoch": 0.5128205128205128,
|
| 8773 |
+
"grad_norm": 3.828125,
|
| 8774 |
+
"learning_rate": 2.488488907492675e-06,
|
| 8775 |
+
"loss": 0.9834448099136353,
|
| 8776 |
+
"step": 1250
|
| 8777 |
+
},
|
| 8778 |
+
{
|
| 8779 |
+
"epoch": 0.5132307692307693,
|
| 8780 |
+
"grad_norm": 4.3125,
|
| 8781 |
+
"learning_rate": 2.4863959815822523e-06,
|
| 8782 |
+
"loss": 1.0878517627716064,
|
| 8783 |
+
"step": 1251
|
| 8784 |
+
},
|
| 8785 |
+
{
|
| 8786 |
+
"epoch": 0.5136410256410257,
|
| 8787 |
+
"grad_norm": 3.65625,
|
| 8788 |
+
"learning_rate": 2.4843030556718295e-06,
|
| 8789 |
+
"loss": 1.2621769905090332,
|
| 8790 |
+
"step": 1252
|
| 8791 |
+
},
|
| 8792 |
+
{
|
| 8793 |
+
"epoch": 0.514051282051282,
|
| 8794 |
+
"grad_norm": 3.75,
|
| 8795 |
+
"learning_rate": 2.4822101297614063e-06,
|
| 8796 |
+
"loss": 1.1476291418075562,
|
| 8797 |
+
"step": 1253
|
| 8798 |
+
},
|
| 8799 |
+
{
|
| 8800 |
+
"epoch": 0.5144615384615384,
|
| 8801 |
+
"grad_norm": 4.3125,
|
| 8802 |
+
"learning_rate": 2.480117203850984e-06,
|
| 8803 |
+
"loss": 1.3722760677337646,
|
| 8804 |
+
"step": 1254
|
| 8805 |
+
},
|
| 8806 |
+
{
|
| 8807 |
+
"epoch": 0.5148717948717949,
|
| 8808 |
+
"grad_norm": 4.09375,
|
| 8809 |
+
"learning_rate": 2.478024277940561e-06,
|
| 8810 |
+
"loss": 1.2697558403015137,
|
| 8811 |
+
"step": 1255
|
| 8812 |
+
},
|
| 8813 |
+
{
|
| 8814 |
+
"epoch": 0.5152820512820513,
|
| 8815 |
+
"grad_norm": 4.25,
|
| 8816 |
+
"learning_rate": 2.4759313520301384e-06,
|
| 8817 |
+
"loss": 1.2076574563980103,
|
| 8818 |
+
"step": 1256
|
| 8819 |
+
},
|
| 8820 |
+
{
|
| 8821 |
+
"epoch": 0.5156923076923077,
|
| 8822 |
+
"grad_norm": 4.0,
|
| 8823 |
+
"learning_rate": 2.4738384261197156e-06,
|
| 8824 |
+
"loss": 1.20155668258667,
|
| 8825 |
+
"step": 1257
|
| 8826 |
+
},
|
| 8827 |
+
{
|
| 8828 |
+
"epoch": 0.516102564102564,
|
| 8829 |
+
"grad_norm": 3.828125,
|
| 8830 |
+
"learning_rate": 2.471745500209293e-06,
|
| 8831 |
+
"loss": 0.9523393511772156,
|
| 8832 |
+
"step": 1258
|
| 8833 |
+
},
|
| 8834 |
+
{
|
| 8835 |
+
"epoch": 0.5165128205128205,
|
| 8836 |
+
"grad_norm": 4.09375,
|
| 8837 |
+
"learning_rate": 2.46965257429887e-06,
|
| 8838 |
+
"loss": 1.2442148923873901,
|
| 8839 |
+
"step": 1259
|
| 8840 |
+
},
|
| 8841 |
+
{
|
| 8842 |
+
"epoch": 0.5169230769230769,
|
| 8843 |
+
"grad_norm": 3.6875,
|
| 8844 |
+
"learning_rate": 2.4675596483884472e-06,
|
| 8845 |
+
"loss": 1.2221131324768066,
|
| 8846 |
+
"step": 1260
|
| 8847 |
+
},
|
| 8848 |
+
{
|
| 8849 |
+
"epoch": 0.5173333333333333,
|
| 8850 |
+
"grad_norm": 3.421875,
|
| 8851 |
+
"learning_rate": 2.4654667224780245e-06,
|
| 8852 |
+
"loss": 1.2359682321548462,
|
| 8853 |
+
"step": 1261
|
| 8854 |
+
},
|
| 8855 |
+
{
|
| 8856 |
+
"epoch": 0.5177435897435897,
|
| 8857 |
+
"grad_norm": 3.796875,
|
| 8858 |
+
"learning_rate": 2.4633737965676017e-06,
|
| 8859 |
+
"loss": 1.3138307332992554,
|
| 8860 |
+
"step": 1262
|
| 8861 |
+
},
|
| 8862 |
+
{
|
| 8863 |
+
"epoch": 0.5181538461538462,
|
| 8864 |
+
"grad_norm": 3.5,
|
| 8865 |
+
"learning_rate": 2.461280870657179e-06,
|
| 8866 |
+
"loss": 1.2189525365829468,
|
| 8867 |
+
"step": 1263
|
| 8868 |
+
},
|
| 8869 |
+
{
|
| 8870 |
+
"epoch": 0.5185641025641026,
|
| 8871 |
+
"grad_norm": 4.59375,
|
| 8872 |
+
"learning_rate": 2.459187944746756e-06,
|
| 8873 |
+
"loss": 1.4329938888549805,
|
| 8874 |
+
"step": 1264
|
| 8875 |
+
},
|
| 8876 |
+
{
|
| 8877 |
+
"epoch": 0.518974358974359,
|
| 8878 |
+
"grad_norm": 3.53125,
|
| 8879 |
+
"learning_rate": 2.4570950188363333e-06,
|
| 8880 |
+
"loss": 1.2080436944961548,
|
| 8881 |
+
"step": 1265
|
| 8882 |
+
},
|
| 8883 |
+
{
|
| 8884 |
+
"epoch": 0.5193846153846153,
|
| 8885 |
+
"grad_norm": 3.828125,
|
| 8886 |
+
"learning_rate": 2.4550020929259105e-06,
|
| 8887 |
+
"loss": 1.3450018167495728,
|
| 8888 |
+
"step": 1266
|
| 8889 |
+
},
|
| 8890 |
+
{
|
| 8891 |
+
"epoch": 0.5197948717948718,
|
| 8892 |
+
"grad_norm": 4.53125,
|
| 8893 |
+
"learning_rate": 2.4529091670154877e-06,
|
| 8894 |
+
"loss": 1.3941442966461182,
|
| 8895 |
+
"step": 1267
|
| 8896 |
+
},
|
| 8897 |
+
{
|
| 8898 |
+
"epoch": 0.5202051282051282,
|
| 8899 |
+
"grad_norm": 3.90625,
|
| 8900 |
+
"learning_rate": 2.450816241105065e-06,
|
| 8901 |
+
"loss": 1.2751364707946777,
|
| 8902 |
+
"step": 1268
|
| 8903 |
+
},
|
| 8904 |
+
{
|
| 8905 |
+
"epoch": 0.5206153846153846,
|
| 8906 |
+
"grad_norm": 4.1875,
|
| 8907 |
+
"learning_rate": 2.4487233151946426e-06,
|
| 8908 |
+
"loss": 1.1873536109924316,
|
| 8909 |
+
"step": 1269
|
| 8910 |
+
},
|
| 8911 |
+
{
|
| 8912 |
+
"epoch": 0.521025641025641,
|
| 8913 |
+
"grad_norm": 4.1875,
|
| 8914 |
+
"learning_rate": 2.4466303892842194e-06,
|
| 8915 |
+
"loss": 1.2449404001235962,
|
| 8916 |
+
"step": 1270
|
| 8917 |
+
},
|
| 8918 |
+
{
|
| 8919 |
+
"epoch": 0.5214358974358975,
|
| 8920 |
+
"grad_norm": 4.0625,
|
| 8921 |
+
"learning_rate": 2.4445374633737966e-06,
|
| 8922 |
+
"loss": 1.302126169204712,
|
| 8923 |
+
"step": 1271
|
| 8924 |
+
},
|
| 8925 |
+
{
|
| 8926 |
+
"epoch": 0.5218461538461538,
|
| 8927 |
+
"grad_norm": 3.796875,
|
| 8928 |
+
"learning_rate": 2.442444537463374e-06,
|
| 8929 |
+
"loss": 1.1582050323486328,
|
| 8930 |
+
"step": 1272
|
| 8931 |
+
},
|
| 8932 |
+
{
|
| 8933 |
+
"epoch": 0.5222564102564102,
|
| 8934 |
+
"grad_norm": 3.75,
|
| 8935 |
+
"learning_rate": 2.4403516115529515e-06,
|
| 8936 |
+
"loss": 1.365095615386963,
|
| 8937 |
+
"step": 1273
|
| 8938 |
+
},
|
| 8939 |
+
{
|
| 8940 |
+
"epoch": 0.5226666666666666,
|
| 8941 |
+
"grad_norm": 3.984375,
|
| 8942 |
+
"learning_rate": 2.4382586856425282e-06,
|
| 8943 |
+
"loss": 1.3592795133590698,
|
| 8944 |
+
"step": 1274
|
| 8945 |
+
},
|
| 8946 |
+
{
|
| 8947 |
+
"epoch": 0.5230769230769231,
|
| 8948 |
+
"grad_norm": 4.125,
|
| 8949 |
+
"learning_rate": 2.4361657597321055e-06,
|
| 8950 |
+
"loss": 1.2852964401245117,
|
| 8951 |
+
"step": 1275
|
| 8952 |
+
},
|
| 8953 |
+
{
|
| 8954 |
+
"epoch": 0.5234871794871795,
|
| 8955 |
+
"grad_norm": 3.953125,
|
| 8956 |
+
"learning_rate": 2.434072833821683e-06,
|
| 8957 |
+
"loss": 1.1792867183685303,
|
| 8958 |
+
"step": 1276
|
| 8959 |
+
},
|
| 8960 |
+
{
|
| 8961 |
+
"epoch": 0.5238974358974359,
|
| 8962 |
+
"grad_norm": 3.875,
|
| 8963 |
+
"learning_rate": 2.4319799079112603e-06,
|
| 8964 |
+
"loss": 1.1743160486221313,
|
| 8965 |
+
"step": 1277
|
| 8966 |
+
},
|
| 8967 |
+
{
|
| 8968 |
+
"epoch": 0.5243076923076923,
|
| 8969 |
+
"grad_norm": 3.796875,
|
| 8970 |
+
"learning_rate": 2.4298869820008375e-06,
|
| 8971 |
+
"loss": 1.2896336317062378,
|
| 8972 |
+
"step": 1278
|
| 8973 |
+
},
|
| 8974 |
+
{
|
| 8975 |
+
"epoch": 0.5247179487179487,
|
| 8976 |
+
"grad_norm": 4.625,
|
| 8977 |
+
"learning_rate": 2.4277940560904143e-06,
|
| 8978 |
+
"loss": 1.1750905513763428,
|
| 8979 |
+
"step": 1279
|
| 8980 |
+
},
|
| 8981 |
+
{
|
| 8982 |
+
"epoch": 0.5251282051282051,
|
| 8983 |
+
"grad_norm": 4.0,
|
| 8984 |
+
"learning_rate": 2.425701130179992e-06,
|
| 8985 |
+
"loss": 1.2609091997146606,
|
| 8986 |
+
"step": 1280
|
| 8987 |
+
},
|
| 8988 |
+
{
|
| 8989 |
+
"epoch": 0.5255384615384615,
|
| 8990 |
+
"grad_norm": 3.96875,
|
| 8991 |
+
"learning_rate": 2.423608204269569e-06,
|
| 8992 |
+
"loss": 1.1478030681610107,
|
| 8993 |
+
"step": 1281
|
| 8994 |
+
},
|
| 8995 |
+
{
|
| 8996 |
+
"epoch": 0.525948717948718,
|
| 8997 |
+
"grad_norm": 3.90625,
|
| 8998 |
+
"learning_rate": 2.4215152783591464e-06,
|
| 8999 |
+
"loss": 1.1259759664535522,
|
| 9000 |
+
"step": 1282
|
| 9001 |
+
},
|
| 9002 |
+
{
|
| 9003 |
+
"epoch": 0.5263589743589744,
|
| 9004 |
+
"grad_norm": 4.1875,
|
| 9005 |
+
"learning_rate": 2.4194223524487236e-06,
|
| 9006 |
+
"loss": 1.1587246656417847,
|
| 9007 |
+
"step": 1283
|
| 9008 |
+
},
|
| 9009 |
+
{
|
| 9010 |
+
"epoch": 0.5267692307692308,
|
| 9011 |
+
"grad_norm": 4.1875,
|
| 9012 |
+
"learning_rate": 2.417329426538301e-06,
|
| 9013 |
+
"loss": 1.1886944770812988,
|
| 9014 |
+
"step": 1284
|
| 9015 |
+
},
|
| 9016 |
+
{
|
| 9017 |
+
"epoch": 0.5271794871794871,
|
| 9018 |
+
"grad_norm": 3.6875,
|
| 9019 |
+
"learning_rate": 2.415236500627878e-06,
|
| 9020 |
+
"loss": 1.209748387336731,
|
| 9021 |
+
"step": 1285
|
| 9022 |
+
},
|
| 9023 |
+
{
|
| 9024 |
+
"epoch": 0.5275897435897436,
|
| 9025 |
+
"grad_norm": 4.0,
|
| 9026 |
+
"learning_rate": 2.4131435747174552e-06,
|
| 9027 |
+
"loss": 1.1581147909164429,
|
| 9028 |
+
"step": 1286
|
| 9029 |
+
},
|
| 9030 |
+
{
|
| 9031 |
+
"epoch": 0.528,
|
| 9032 |
+
"grad_norm": 4.09375,
|
| 9033 |
+
"learning_rate": 2.4110506488070325e-06,
|
| 9034 |
+
"loss": 1.2428579330444336,
|
| 9035 |
+
"step": 1287
|
| 9036 |
+
},
|
| 9037 |
+
{
|
| 9038 |
+
"epoch": 0.5284102564102564,
|
| 9039 |
+
"grad_norm": 4.8125,
|
| 9040 |
+
"learning_rate": 2.4089577228966097e-06,
|
| 9041 |
+
"loss": 1.45995032787323,
|
| 9042 |
+
"step": 1288
|
| 9043 |
+
},
|
| 9044 |
+
{
|
| 9045 |
+
"epoch": 0.5288205128205128,
|
| 9046 |
+
"grad_norm": 3.8125,
|
| 9047 |
+
"learning_rate": 2.406864796986187e-06,
|
| 9048 |
+
"loss": 1.289202094078064,
|
| 9049 |
+
"step": 1289
|
| 9050 |
+
},
|
| 9051 |
+
{
|
| 9052 |
+
"epoch": 0.5292307692307693,
|
| 9053 |
+
"grad_norm": 4.03125,
|
| 9054 |
+
"learning_rate": 2.404771871075764e-06,
|
| 9055 |
+
"loss": 1.2943933010101318,
|
| 9056 |
+
"step": 1290
|
| 9057 |
+
},
|
| 9058 |
+
{
|
| 9059 |
+
"epoch": 0.5296410256410257,
|
| 9060 |
+
"grad_norm": 3.875,
|
| 9061 |
+
"learning_rate": 2.4026789451653413e-06,
|
| 9062 |
+
"loss": 1.1371855735778809,
|
| 9063 |
+
"step": 1291
|
| 9064 |
+
},
|
| 9065 |
+
{
|
| 9066 |
+
"epoch": 0.530051282051282,
|
| 9067 |
+
"grad_norm": 4.46875,
|
| 9068 |
+
"learning_rate": 2.4005860192549185e-06,
|
| 9069 |
+
"loss": 1.1848924160003662,
|
| 9070 |
+
"step": 1292
|
| 9071 |
+
},
|
| 9072 |
+
{
|
| 9073 |
+
"epoch": 0.5304615384615384,
|
| 9074 |
+
"grad_norm": 3.5,
|
| 9075 |
+
"learning_rate": 2.3984930933444957e-06,
|
| 9076 |
+
"loss": 1.0555386543273926,
|
| 9077 |
+
"step": 1293
|
| 9078 |
+
},
|
| 9079 |
+
{
|
| 9080 |
+
"epoch": 0.5308717948717949,
|
| 9081 |
+
"grad_norm": 3.734375,
|
| 9082 |
+
"learning_rate": 2.396400167434073e-06,
|
| 9083 |
+
"loss": 1.1866416931152344,
|
| 9084 |
+
"step": 1294
|
| 9085 |
+
},
|
| 9086 |
+
{
|
| 9087 |
+
"epoch": 0.5312820512820513,
|
| 9088 |
+
"grad_norm": 3.75,
|
| 9089 |
+
"learning_rate": 2.39430724152365e-06,
|
| 9090 |
+
"loss": 1.0859930515289307,
|
| 9091 |
+
"step": 1295
|
| 9092 |
+
},
|
| 9093 |
+
{
|
| 9094 |
+
"epoch": 0.5316923076923077,
|
| 9095 |
+
"grad_norm": 3.953125,
|
| 9096 |
+
"learning_rate": 2.3922143156132274e-06,
|
| 9097 |
+
"loss": 1.2078888416290283,
|
| 9098 |
+
"step": 1296
|
| 9099 |
+
},
|
| 9100 |
+
{
|
| 9101 |
+
"epoch": 0.5321025641025641,
|
| 9102 |
+
"grad_norm": 3.953125,
|
| 9103 |
+
"learning_rate": 2.3901213897028046e-06,
|
| 9104 |
+
"loss": 1.1704161167144775,
|
| 9105 |
+
"step": 1297
|
| 9106 |
+
},
|
| 9107 |
+
{
|
| 9108 |
+
"epoch": 0.5325128205128206,
|
| 9109 |
+
"grad_norm": 4.1875,
|
| 9110 |
+
"learning_rate": 2.388028463792382e-06,
|
| 9111 |
+
"loss": 1.445304036140442,
|
| 9112 |
+
"step": 1298
|
| 9113 |
+
},
|
| 9114 |
+
{
|
| 9115 |
+
"epoch": 0.5329230769230769,
|
| 9116 |
+
"grad_norm": 3.921875,
|
| 9117 |
+
"learning_rate": 2.3859355378819595e-06,
|
| 9118 |
+
"loss": 1.238620638847351,
|
| 9119 |
+
"step": 1299
|
| 9120 |
+
},
|
| 9121 |
+
{
|
| 9122 |
+
"epoch": 0.5333333333333333,
|
| 9123 |
+
"grad_norm": 4.1875,
|
| 9124 |
+
"learning_rate": 2.3838426119715362e-06,
|
| 9125 |
+
"loss": 1.177973985671997,
|
| 9126 |
+
"step": 1300
|
| 9127 |
+
},
|
| 9128 |
+
{
|
| 9129 |
+
"epoch": 0.5337435897435897,
|
| 9130 |
+
"grad_norm": 4.28125,
|
| 9131 |
+
"learning_rate": 2.3817496860611135e-06,
|
| 9132 |
+
"loss": 1.283602237701416,
|
| 9133 |
+
"step": 1301
|
| 9134 |
+
},
|
| 9135 |
+
{
|
| 9136 |
+
"epoch": 0.5341538461538462,
|
| 9137 |
+
"grad_norm": 4.4375,
|
| 9138 |
+
"learning_rate": 2.379656760150691e-06,
|
| 9139 |
+
"loss": 1.173832654953003,
|
| 9140 |
+
"step": 1302
|
| 9141 |
+
},
|
| 9142 |
+
{
|
| 9143 |
+
"epoch": 0.5345641025641026,
|
| 9144 |
+
"grad_norm": 3.890625,
|
| 9145 |
+
"learning_rate": 2.3775638342402683e-06,
|
| 9146 |
+
"loss": 1.2819654941558838,
|
| 9147 |
+
"step": 1303
|
| 9148 |
+
},
|
| 9149 |
+
{
|
| 9150 |
+
"epoch": 0.534974358974359,
|
| 9151 |
+
"grad_norm": 4.28125,
|
| 9152 |
+
"learning_rate": 2.375470908329845e-06,
|
| 9153 |
+
"loss": 1.3683156967163086,
|
| 9154 |
+
"step": 1304
|
| 9155 |
+
},
|
| 9156 |
+
{
|
| 9157 |
+
"epoch": 0.5353846153846153,
|
| 9158 |
+
"grad_norm": 3.96875,
|
| 9159 |
+
"learning_rate": 2.3733779824194223e-06,
|
| 9160 |
+
"loss": 1.1089121103286743,
|
| 9161 |
+
"step": 1305
|
| 9162 |
+
},
|
| 9163 |
+
{
|
| 9164 |
+
"epoch": 0.5357948717948718,
|
| 9165 |
+
"grad_norm": 4.125,
|
| 9166 |
+
"learning_rate": 2.371285056509e-06,
|
| 9167 |
+
"loss": 1.2259081602096558,
|
| 9168 |
+
"step": 1306
|
| 9169 |
+
},
|
| 9170 |
+
{
|
| 9171 |
+
"epoch": 0.5362051282051282,
|
| 9172 |
+
"grad_norm": 3.90625,
|
| 9173 |
+
"learning_rate": 2.369192130598577e-06,
|
| 9174 |
+
"loss": 1.16845703125,
|
| 9175 |
+
"step": 1307
|
| 9176 |
+
},
|
| 9177 |
+
{
|
| 9178 |
+
"epoch": 0.5366153846153846,
|
| 9179 |
+
"grad_norm": 3.671875,
|
| 9180 |
+
"learning_rate": 2.367099204688154e-06,
|
| 9181 |
+
"loss": 1.069218397140503,
|
| 9182 |
+
"step": 1308
|
| 9183 |
+
},
|
| 9184 |
+
{
|
| 9185 |
+
"epoch": 0.537025641025641,
|
| 9186 |
+
"grad_norm": 3.65625,
|
| 9187 |
+
"learning_rate": 2.3650062787777316e-06,
|
| 9188 |
+
"loss": 1.2612122297286987,
|
| 9189 |
+
"step": 1309
|
| 9190 |
+
},
|
| 9191 |
+
{
|
| 9192 |
+
"epoch": 0.5374358974358975,
|
| 9193 |
+
"grad_norm": 4.75,
|
| 9194 |
+
"learning_rate": 2.362913352867309e-06,
|
| 9195 |
+
"loss": 1.247633695602417,
|
| 9196 |
+
"step": 1310
|
| 9197 |
+
},
|
| 9198 |
+
{
|
| 9199 |
+
"epoch": 0.5378461538461539,
|
| 9200 |
+
"grad_norm": 3.828125,
|
| 9201 |
+
"learning_rate": 2.360820426956886e-06,
|
| 9202 |
+
"loss": 1.1097643375396729,
|
| 9203 |
+
"step": 1311
|
| 9204 |
+
},
|
| 9205 |
+
{
|
| 9206 |
+
"epoch": 0.5382564102564102,
|
| 9207 |
+
"grad_norm": 3.671875,
|
| 9208 |
+
"learning_rate": 2.3587275010464632e-06,
|
| 9209 |
+
"loss": 1.1084036827087402,
|
| 9210 |
+
"step": 1312
|
| 9211 |
+
},
|
| 9212 |
+
{
|
| 9213 |
+
"epoch": 0.5386666666666666,
|
| 9214 |
+
"grad_norm": 3.9375,
|
| 9215 |
+
"learning_rate": 2.3566345751360405e-06,
|
| 9216 |
+
"loss": 1.1544528007507324,
|
| 9217 |
+
"step": 1313
|
| 9218 |
+
},
|
| 9219 |
+
{
|
| 9220 |
+
"epoch": 0.5390769230769231,
|
| 9221 |
+
"grad_norm": 3.90625,
|
| 9222 |
+
"learning_rate": 2.3545416492256177e-06,
|
| 9223 |
+
"loss": 1.1434739828109741,
|
| 9224 |
+
"step": 1314
|
| 9225 |
+
},
|
| 9226 |
+
{
|
| 9227 |
+
"epoch": 0.5394871794871795,
|
| 9228 |
+
"grad_norm": 4.21875,
|
| 9229 |
+
"learning_rate": 2.352448723315195e-06,
|
| 9230 |
+
"loss": 1.3218271732330322,
|
| 9231 |
+
"step": 1315
|
| 9232 |
+
},
|
| 9233 |
+
{
|
| 9234 |
+
"epoch": 0.5398974358974359,
|
| 9235 |
+
"grad_norm": 4.09375,
|
| 9236 |
+
"learning_rate": 2.350355797404772e-06,
|
| 9237 |
+
"loss": 1.1885273456573486,
|
| 9238 |
+
"step": 1316
|
| 9239 |
+
},
|
| 9240 |
+
{
|
| 9241 |
+
"epoch": 0.5403076923076923,
|
| 9242 |
+
"grad_norm": 4.28125,
|
| 9243 |
+
"learning_rate": 2.3482628714943493e-06,
|
| 9244 |
+
"loss": 1.1027476787567139,
|
| 9245 |
+
"step": 1317
|
| 9246 |
+
},
|
| 9247 |
+
{
|
| 9248 |
+
"epoch": 0.5407179487179488,
|
| 9249 |
+
"grad_norm": 4.0625,
|
| 9250 |
+
"learning_rate": 2.3461699455839265e-06,
|
| 9251 |
+
"loss": 1.3202548027038574,
|
| 9252 |
+
"step": 1318
|
| 9253 |
+
},
|
| 9254 |
+
{
|
| 9255 |
+
"epoch": 0.5411282051282051,
|
| 9256 |
+
"grad_norm": 4.3125,
|
| 9257 |
+
"learning_rate": 2.3440770196735037e-06,
|
| 9258 |
+
"loss": 1.2963600158691406,
|
| 9259 |
+
"step": 1319
|
| 9260 |
+
},
|
| 9261 |
+
{
|
| 9262 |
+
"epoch": 0.5415384615384615,
|
| 9263 |
+
"grad_norm": 4.03125,
|
| 9264 |
+
"learning_rate": 2.341984093763081e-06,
|
| 9265 |
+
"loss": 1.3234796524047852,
|
| 9266 |
+
"step": 1320
|
| 9267 |
+
},
|
| 9268 |
+
{
|
| 9269 |
+
"epoch": 0.5419487179487179,
|
| 9270 |
+
"grad_norm": 4.3125,
|
| 9271 |
+
"learning_rate": 2.339891167852658e-06,
|
| 9272 |
+
"loss": 1.3363049030303955,
|
| 9273 |
+
"step": 1321
|
| 9274 |
+
},
|
| 9275 |
+
{
|
| 9276 |
+
"epoch": 0.5423589743589744,
|
| 9277 |
+
"grad_norm": 3.671875,
|
| 9278 |
+
"learning_rate": 2.3377982419422354e-06,
|
| 9279 |
+
"loss": 1.1250463724136353,
|
| 9280 |
+
"step": 1322
|
| 9281 |
+
},
|
| 9282 |
+
{
|
| 9283 |
+
"epoch": 0.5427692307692308,
|
| 9284 |
+
"grad_norm": 4.59375,
|
| 9285 |
+
"learning_rate": 2.3357053160318126e-06,
|
| 9286 |
+
"loss": 1.313450574874878,
|
| 9287 |
+
"step": 1323
|
| 9288 |
+
},
|
| 9289 |
+
{
|
| 9290 |
+
"epoch": 0.5431794871794872,
|
| 9291 |
+
"grad_norm": 4.0625,
|
| 9292 |
+
"learning_rate": 2.33361239012139e-06,
|
| 9293 |
+
"loss": 1.191622257232666,
|
| 9294 |
+
"step": 1324
|
| 9295 |
+
},
|
| 9296 |
+
{
|
| 9297 |
+
"epoch": 0.5435897435897435,
|
| 9298 |
+
"grad_norm": 4.0625,
|
| 9299 |
+
"learning_rate": 2.331519464210967e-06,
|
| 9300 |
+
"loss": 1.349053144454956,
|
| 9301 |
+
"step": 1325
|
| 9302 |
+
},
|
| 9303 |
+
{
|
| 9304 |
+
"epoch": 0.544,
|
| 9305 |
+
"grad_norm": 4.1875,
|
| 9306 |
+
"learning_rate": 2.3294265383005442e-06,
|
| 9307 |
+
"loss": 1.2662193775177002,
|
| 9308 |
+
"step": 1326
|
| 9309 |
+
},
|
| 9310 |
+
{
|
| 9311 |
+
"epoch": 0.5444102564102564,
|
| 9312 |
+
"grad_norm": 3.890625,
|
| 9313 |
+
"learning_rate": 2.3273336123901215e-06,
|
| 9314 |
+
"loss": 1.2077032327651978,
|
| 9315 |
+
"step": 1327
|
| 9316 |
+
},
|
| 9317 |
+
{
|
| 9318 |
+
"epoch": 0.5448205128205128,
|
| 9319 |
+
"grad_norm": 4.0625,
|
| 9320 |
+
"learning_rate": 2.325240686479699e-06,
|
| 9321 |
+
"loss": 1.2089680433273315,
|
| 9322 |
+
"step": 1328
|
| 9323 |
+
},
|
| 9324 |
+
{
|
| 9325 |
+
"epoch": 0.5452307692307692,
|
| 9326 |
+
"grad_norm": 3.828125,
|
| 9327 |
+
"learning_rate": 2.323147760569276e-06,
|
| 9328 |
+
"loss": 1.2732915878295898,
|
| 9329 |
+
"step": 1329
|
| 9330 |
+
},
|
| 9331 |
+
{
|
| 9332 |
+
"epoch": 0.5456410256410257,
|
| 9333 |
+
"grad_norm": 3.859375,
|
| 9334 |
+
"learning_rate": 2.321054834658853e-06,
|
| 9335 |
+
"loss": 1.1631616353988647,
|
| 9336 |
+
"step": 1330
|
| 9337 |
+
},
|
| 9338 |
+
{
|
| 9339 |
+
"epoch": 0.5460512820512821,
|
| 9340 |
+
"grad_norm": 3.9375,
|
| 9341 |
+
"learning_rate": 2.3189619087484303e-06,
|
| 9342 |
+
"loss": 1.1962820291519165,
|
| 9343 |
+
"step": 1331
|
| 9344 |
+
},
|
| 9345 |
+
{
|
| 9346 |
+
"epoch": 0.5464615384615384,
|
| 9347 |
+
"grad_norm": 3.8125,
|
| 9348 |
+
"learning_rate": 2.316868982838008e-06,
|
| 9349 |
+
"loss": 1.1445090770721436,
|
| 9350 |
+
"step": 1332
|
| 9351 |
+
},
|
| 9352 |
+
{
|
| 9353 |
+
"epoch": 0.5468717948717948,
|
| 9354 |
+
"grad_norm": 4.09375,
|
| 9355 |
+
"learning_rate": 2.3147760569275847e-06,
|
| 9356 |
+
"loss": 1.2740085124969482,
|
| 9357 |
+
"step": 1333
|
| 9358 |
+
},
|
| 9359 |
+
{
|
| 9360 |
+
"epoch": 0.5472820512820513,
|
| 9361 |
+
"grad_norm": 3.625,
|
| 9362 |
+
"learning_rate": 2.312683131017162e-06,
|
| 9363 |
+
"loss": 0.999878466129303,
|
| 9364 |
+
"step": 1334
|
| 9365 |
+
},
|
| 9366 |
+
{
|
| 9367 |
+
"epoch": 0.5476923076923077,
|
| 9368 |
+
"grad_norm": 3.4375,
|
| 9369 |
+
"learning_rate": 2.3105902051067396e-06,
|
| 9370 |
+
"loss": 1.1936888694763184,
|
| 9371 |
+
"step": 1335
|
| 9372 |
+
},
|
| 9373 |
+
{
|
| 9374 |
+
"epoch": 0.5481025641025641,
|
| 9375 |
+
"grad_norm": 4.125,
|
| 9376 |
+
"learning_rate": 2.308497279196317e-06,
|
| 9377 |
+
"loss": 1.2396663427352905,
|
| 9378 |
+
"step": 1336
|
| 9379 |
+
},
|
| 9380 |
+
{
|
| 9381 |
+
"epoch": 0.5485128205128205,
|
| 9382 |
+
"grad_norm": 4.09375,
|
| 9383 |
+
"learning_rate": 2.3064043532858936e-06,
|
| 9384 |
+
"loss": 1.2868750095367432,
|
| 9385 |
+
"step": 1337
|
| 9386 |
+
},
|
| 9387 |
+
{
|
| 9388 |
+
"epoch": 0.548923076923077,
|
| 9389 |
+
"grad_norm": 4.46875,
|
| 9390 |
+
"learning_rate": 2.3043114273754712e-06,
|
| 9391 |
+
"loss": 1.2976571321487427,
|
| 9392 |
+
"step": 1338
|
| 9393 |
+
},
|
| 9394 |
+
{
|
| 9395 |
+
"epoch": 0.5493333333333333,
|
| 9396 |
+
"grad_norm": 4.0,
|
| 9397 |
+
"learning_rate": 2.3022185014650485e-06,
|
| 9398 |
+
"loss": 1.2845680713653564,
|
| 9399 |
+
"step": 1339
|
| 9400 |
+
},
|
| 9401 |
+
{
|
| 9402 |
+
"epoch": 0.5497435897435897,
|
| 9403 |
+
"grad_norm": 3.828125,
|
| 9404 |
+
"learning_rate": 2.3001255755546257e-06,
|
| 9405 |
+
"loss": 1.3165431022644043,
|
| 9406 |
+
"step": 1340
|
| 9407 |
+
},
|
| 9408 |
+
{
|
| 9409 |
+
"epoch": 0.5501538461538461,
|
| 9410 |
+
"grad_norm": 3.84375,
|
| 9411 |
+
"learning_rate": 2.298032649644203e-06,
|
| 9412 |
+
"loss": 1.216347575187683,
|
| 9413 |
+
"step": 1341
|
| 9414 |
+
},
|
| 9415 |
+
{
|
| 9416 |
+
"epoch": 0.5505641025641026,
|
| 9417 |
+
"grad_norm": 3.703125,
|
| 9418 |
+
"learning_rate": 2.29593972373378e-06,
|
| 9419 |
+
"loss": 1.2590183019638062,
|
| 9420 |
+
"step": 1342
|
| 9421 |
+
},
|
| 9422 |
+
{
|
| 9423 |
+
"epoch": 0.550974358974359,
|
| 9424 |
+
"grad_norm": 4.03125,
|
| 9425 |
+
"learning_rate": 2.2938467978233573e-06,
|
| 9426 |
+
"loss": 1.2649482488632202,
|
| 9427 |
+
"step": 1343
|
| 9428 |
+
},
|
| 9429 |
+
{
|
| 9430 |
+
"epoch": 0.5513846153846154,
|
| 9431 |
+
"grad_norm": 3.578125,
|
| 9432 |
+
"learning_rate": 2.2917538719129345e-06,
|
| 9433 |
+
"loss": 1.214444637298584,
|
| 9434 |
+
"step": 1344
|
| 9435 |
+
},
|
| 9436 |
+
{
|
| 9437 |
+
"epoch": 0.5517948717948717,
|
| 9438 |
+
"grad_norm": 4.0625,
|
| 9439 |
+
"learning_rate": 2.2896609460025117e-06,
|
| 9440 |
+
"loss": 1.1526480913162231,
|
| 9441 |
+
"step": 1345
|
| 9442 |
+
},
|
| 9443 |
+
{
|
| 9444 |
+
"epoch": 0.5522051282051282,
|
| 9445 |
+
"grad_norm": 3.796875,
|
| 9446 |
+
"learning_rate": 2.287568020092089e-06,
|
| 9447 |
+
"loss": 0.9375726580619812,
|
| 9448 |
+
"step": 1346
|
| 9449 |
+
},
|
| 9450 |
+
{
|
| 9451 |
+
"epoch": 0.5526153846153846,
|
| 9452 |
+
"grad_norm": 4.03125,
|
| 9453 |
+
"learning_rate": 2.285475094181666e-06,
|
| 9454 |
+
"loss": 1.2299563884735107,
|
| 9455 |
+
"step": 1347
|
| 9456 |
+
},
|
| 9457 |
+
{
|
| 9458 |
+
"epoch": 0.553025641025641,
|
| 9459 |
+
"grad_norm": 4.15625,
|
| 9460 |
+
"learning_rate": 2.2833821682712434e-06,
|
| 9461 |
+
"loss": 1.2389065027236938,
|
| 9462 |
+
"step": 1348
|
| 9463 |
+
},
|
| 9464 |
+
{
|
| 9465 |
+
"epoch": 0.5534358974358974,
|
| 9466 |
+
"grad_norm": 3.953125,
|
| 9467 |
+
"learning_rate": 2.2812892423608206e-06,
|
| 9468 |
+
"loss": 1.3055462837219238,
|
| 9469 |
+
"step": 1349
|
| 9470 |
+
},
|
| 9471 |
+
{
|
| 9472 |
+
"epoch": 0.5538461538461539,
|
| 9473 |
+
"grad_norm": 3.90625,
|
| 9474 |
+
"learning_rate": 2.279196316450398e-06,
|
| 9475 |
+
"loss": 1.134075403213501,
|
| 9476 |
+
"step": 1350
|
| 9477 |
+
},
|
| 9478 |
+
{
|
| 9479 |
+
"epoch": 0.5542564102564103,
|
| 9480 |
+
"grad_norm": 3.84375,
|
| 9481 |
+
"learning_rate": 2.277103390539975e-06,
|
| 9482 |
+
"loss": 1.1117699146270752,
|
| 9483 |
+
"step": 1351
|
| 9484 |
+
},
|
| 9485 |
+
{
|
| 9486 |
+
"epoch": 0.5546666666666666,
|
| 9487 |
+
"grad_norm": 4.09375,
|
| 9488 |
+
"learning_rate": 2.2750104646295522e-06,
|
| 9489 |
+
"loss": 1.3207385540008545,
|
| 9490 |
+
"step": 1352
|
| 9491 |
+
},
|
| 9492 |
+
{
|
| 9493 |
+
"epoch": 0.555076923076923,
|
| 9494 |
+
"grad_norm": 3.84375,
|
| 9495 |
+
"learning_rate": 2.2729175387191295e-06,
|
| 9496 |
+
"loss": 1.2645072937011719,
|
| 9497 |
+
"step": 1353
|
| 9498 |
+
},
|
| 9499 |
+
{
|
| 9500 |
+
"epoch": 0.5554871794871795,
|
| 9501 |
+
"grad_norm": 4.0,
|
| 9502 |
+
"learning_rate": 2.2708246128087067e-06,
|
| 9503 |
+
"loss": 1.2555599212646484,
|
| 9504 |
+
"step": 1354
|
| 9505 |
+
},
|
| 9506 |
+
{
|
| 9507 |
+
"epoch": 0.5558974358974359,
|
| 9508 |
+
"grad_norm": 3.65625,
|
| 9509 |
+
"learning_rate": 2.268731686898284e-06,
|
| 9510 |
+
"loss": 1.1866891384124756,
|
| 9511 |
+
"step": 1355
|
| 9512 |
+
},
|
| 9513 |
+
{
|
| 9514 |
+
"epoch": 0.5563076923076923,
|
| 9515 |
+
"grad_norm": 4.15625,
|
| 9516 |
+
"learning_rate": 2.266638760987861e-06,
|
| 9517 |
+
"loss": 1.1923153400421143,
|
| 9518 |
+
"step": 1356
|
| 9519 |
+
},
|
| 9520 |
+
{
|
| 9521 |
+
"epoch": 0.5567179487179487,
|
| 9522 |
+
"grad_norm": 3.453125,
|
| 9523 |
+
"learning_rate": 2.2645458350774387e-06,
|
| 9524 |
+
"loss": 1.2041971683502197,
|
| 9525 |
+
"step": 1357
|
| 9526 |
+
},
|
| 9527 |
+
{
|
| 9528 |
+
"epoch": 0.5571282051282052,
|
| 9529 |
+
"grad_norm": 4.09375,
|
| 9530 |
+
"learning_rate": 2.2624529091670155e-06,
|
| 9531 |
+
"loss": 1.218790888786316,
|
| 9532 |
+
"step": 1358
|
| 9533 |
+
},
|
| 9534 |
+
{
|
| 9535 |
+
"epoch": 0.5575384615384615,
|
| 9536 |
+
"grad_norm": 4.21875,
|
| 9537 |
+
"learning_rate": 2.2603599832565927e-06,
|
| 9538 |
+
"loss": 1.2315562963485718,
|
| 9539 |
+
"step": 1359
|
| 9540 |
+
},
|
| 9541 |
+
{
|
| 9542 |
+
"epoch": 0.5579487179487179,
|
| 9543 |
+
"grad_norm": 4.21875,
|
| 9544 |
+
"learning_rate": 2.25826705734617e-06,
|
| 9545 |
+
"loss": 1.3012369871139526,
|
| 9546 |
+
"step": 1360
|
| 9547 |
+
},
|
| 9548 |
+
{
|
| 9549 |
+
"epoch": 0.5583589743589744,
|
| 9550 |
+
"grad_norm": 4.1875,
|
| 9551 |
+
"learning_rate": 2.2561741314357476e-06,
|
| 9552 |
+
"loss": 1.1454898118972778,
|
| 9553 |
+
"step": 1361
|
| 9554 |
+
},
|
| 9555 |
+
{
|
| 9556 |
+
"epoch": 0.5587692307692308,
|
| 9557 |
+
"grad_norm": 4.53125,
|
| 9558 |
+
"learning_rate": 2.254081205525325e-06,
|
| 9559 |
+
"loss": 1.4520554542541504,
|
| 9560 |
+
"step": 1362
|
| 9561 |
+
},
|
| 9562 |
+
{
|
| 9563 |
+
"epoch": 0.5591794871794872,
|
| 9564 |
+
"grad_norm": 4.15625,
|
| 9565 |
+
"learning_rate": 2.2519882796149016e-06,
|
| 9566 |
+
"loss": 1.3536001443862915,
|
| 9567 |
+
"step": 1363
|
| 9568 |
+
},
|
| 9569 |
+
{
|
| 9570 |
+
"epoch": 0.5595897435897436,
|
| 9571 |
+
"grad_norm": 4.0625,
|
| 9572 |
+
"learning_rate": 2.2498953537044792e-06,
|
| 9573 |
+
"loss": 1.2245821952819824,
|
| 9574 |
+
"step": 1364
|
| 9575 |
+
},
|
| 9576 |
+
{
|
| 9577 |
+
"epoch": 0.56,
|
| 9578 |
+
"grad_norm": 3.859375,
|
| 9579 |
+
"learning_rate": 2.2478024277940564e-06,
|
| 9580 |
+
"loss": 1.2086546421051025,
|
| 9581 |
+
"step": 1365
|
| 9582 |
+
},
|
| 9583 |
+
{
|
| 9584 |
+
"epoch": 0.5604102564102564,
|
| 9585 |
+
"grad_norm": 3.921875,
|
| 9586 |
+
"learning_rate": 2.2457095018836337e-06,
|
| 9587 |
+
"loss": 1.2366352081298828,
|
| 9588 |
+
"step": 1366
|
| 9589 |
+
},
|
| 9590 |
+
{
|
| 9591 |
+
"epoch": 0.5608205128205128,
|
| 9592 |
+
"grad_norm": 3.9375,
|
| 9593 |
+
"learning_rate": 2.2436165759732105e-06,
|
| 9594 |
+
"loss": 1.2329105138778687,
|
| 9595 |
+
"step": 1367
|
| 9596 |
+
},
|
| 9597 |
+
{
|
| 9598 |
+
"epoch": 0.5612307692307692,
|
| 9599 |
+
"grad_norm": 3.671875,
|
| 9600 |
+
"learning_rate": 2.241523650062788e-06,
|
| 9601 |
+
"loss": 1.1722228527069092,
|
| 9602 |
+
"step": 1368
|
| 9603 |
+
},
|
| 9604 |
+
{
|
| 9605 |
+
"epoch": 0.5616410256410257,
|
| 9606 |
+
"grad_norm": 3.828125,
|
| 9607 |
+
"learning_rate": 2.2394307241523653e-06,
|
| 9608 |
+
"loss": 1.2727577686309814,
|
| 9609 |
+
"step": 1369
|
| 9610 |
+
},
|
| 9611 |
+
{
|
| 9612 |
+
"epoch": 0.5620512820512821,
|
| 9613 |
+
"grad_norm": 4.03125,
|
| 9614 |
+
"learning_rate": 2.2373377982419425e-06,
|
| 9615 |
+
"loss": 1.2854326963424683,
|
| 9616 |
+
"step": 1370
|
| 9617 |
+
},
|
| 9618 |
+
{
|
| 9619 |
+
"epoch": 0.5624615384615385,
|
| 9620 |
+
"grad_norm": 4.21875,
|
| 9621 |
+
"learning_rate": 2.2352448723315197e-06,
|
| 9622 |
+
"loss": 1.2284642457962036,
|
| 9623 |
+
"step": 1371
|
| 9624 |
+
},
|
| 9625 |
+
{
|
| 9626 |
+
"epoch": 0.5628717948717948,
|
| 9627 |
+
"grad_norm": 3.515625,
|
| 9628 |
+
"learning_rate": 2.233151946421097e-06,
|
| 9629 |
+
"loss": 1.2309677600860596,
|
| 9630 |
+
"step": 1372
|
| 9631 |
+
},
|
| 9632 |
+
{
|
| 9633 |
+
"epoch": 0.5632820512820513,
|
| 9634 |
+
"grad_norm": 4.25,
|
| 9635 |
+
"learning_rate": 2.231059020510674e-06,
|
| 9636 |
+
"loss": 1.2395477294921875,
|
| 9637 |
+
"step": 1373
|
| 9638 |
+
},
|
| 9639 |
+
{
|
| 9640 |
+
"epoch": 0.5636923076923077,
|
| 9641 |
+
"grad_norm": 3.65625,
|
| 9642 |
+
"learning_rate": 2.2289660946002514e-06,
|
| 9643 |
+
"loss": 1.0826879739761353,
|
| 9644 |
+
"step": 1374
|
| 9645 |
+
},
|
| 9646 |
+
{
|
| 9647 |
+
"epoch": 0.5641025641025641,
|
| 9648 |
+
"grad_norm": 4.21875,
|
| 9649 |
+
"learning_rate": 2.2268731686898286e-06,
|
| 9650 |
+
"loss": 1.192272424697876,
|
| 9651 |
+
"step": 1375
|
| 9652 |
+
},
|
| 9653 |
+
{
|
| 9654 |
+
"epoch": 0.5645128205128205,
|
| 9655 |
+
"grad_norm": 4.03125,
|
| 9656 |
+
"learning_rate": 2.224780242779406e-06,
|
| 9657 |
+
"loss": 1.3193955421447754,
|
| 9658 |
+
"step": 1376
|
| 9659 |
+
},
|
| 9660 |
+
{
|
| 9661 |
+
"epoch": 0.564923076923077,
|
| 9662 |
+
"grad_norm": 4.125,
|
| 9663 |
+
"learning_rate": 2.222687316868983e-06,
|
| 9664 |
+
"loss": 1.3623859882354736,
|
| 9665 |
+
"step": 1377
|
| 9666 |
+
},
|
| 9667 |
+
{
|
| 9668 |
+
"epoch": 0.5653333333333334,
|
| 9669 |
+
"grad_norm": 3.921875,
|
| 9670 |
+
"learning_rate": 2.2205943909585602e-06,
|
| 9671 |
+
"loss": 1.3031232357025146,
|
| 9672 |
+
"step": 1378
|
| 9673 |
+
},
|
| 9674 |
+
{
|
| 9675 |
+
"epoch": 0.5657435897435897,
|
| 9676 |
+
"grad_norm": 3.640625,
|
| 9677 |
+
"learning_rate": 2.2185014650481375e-06,
|
| 9678 |
+
"loss": 1.2850788831710815,
|
| 9679 |
+
"step": 1379
|
| 9680 |
+
},
|
| 9681 |
+
{
|
| 9682 |
+
"epoch": 0.5661538461538461,
|
| 9683 |
+
"grad_norm": 3.84375,
|
| 9684 |
+
"learning_rate": 2.2164085391377147e-06,
|
| 9685 |
+
"loss": 1.0947902202606201,
|
| 9686 |
+
"step": 1380
|
| 9687 |
+
},
|
| 9688 |
+
{
|
| 9689 |
+
"epoch": 0.5665641025641026,
|
| 9690 |
+
"grad_norm": 3.953125,
|
| 9691 |
+
"learning_rate": 2.214315613227292e-06,
|
| 9692 |
+
"loss": 1.05146062374115,
|
| 9693 |
+
"step": 1381
|
| 9694 |
+
},
|
| 9695 |
+
{
|
| 9696 |
+
"epoch": 0.566974358974359,
|
| 9697 |
+
"grad_norm": 3.796875,
|
| 9698 |
+
"learning_rate": 2.212222687316869e-06,
|
| 9699 |
+
"loss": 1.1703437566757202,
|
| 9700 |
+
"step": 1382
|
| 9701 |
+
},
|
| 9702 |
+
{
|
| 9703 |
+
"epoch": 0.5673846153846154,
|
| 9704 |
+
"grad_norm": 4.21875,
|
| 9705 |
+
"learning_rate": 2.2101297614064467e-06,
|
| 9706 |
+
"loss": 1.2748526334762573,
|
| 9707 |
+
"step": 1383
|
| 9708 |
+
},
|
| 9709 |
+
{
|
| 9710 |
+
"epoch": 0.5677948717948718,
|
| 9711 |
+
"grad_norm": 4.0625,
|
| 9712 |
+
"learning_rate": 2.2080368354960235e-06,
|
| 9713 |
+
"loss": 1.3482277393341064,
|
| 9714 |
+
"step": 1384
|
| 9715 |
+
},
|
| 9716 |
+
{
|
| 9717 |
+
"epoch": 0.5682051282051283,
|
| 9718 |
+
"grad_norm": 4.40625,
|
| 9719 |
+
"learning_rate": 2.2059439095856007e-06,
|
| 9720 |
+
"loss": 1.2656503915786743,
|
| 9721 |
+
"step": 1385
|
| 9722 |
+
},
|
| 9723 |
+
{
|
| 9724 |
+
"epoch": 0.5686153846153846,
|
| 9725 |
+
"grad_norm": 4.21875,
|
| 9726 |
+
"learning_rate": 2.203850983675178e-06,
|
| 9727 |
+
"loss": 1.1904926300048828,
|
| 9728 |
+
"step": 1386
|
| 9729 |
+
},
|
| 9730 |
+
{
|
| 9731 |
+
"epoch": 0.569025641025641,
|
| 9732 |
+
"grad_norm": 4.34375,
|
| 9733 |
+
"learning_rate": 2.2017580577647556e-06,
|
| 9734 |
+
"loss": 1.258373737335205,
|
| 9735 |
+
"step": 1387
|
| 9736 |
+
},
|
| 9737 |
+
{
|
| 9738 |
+
"epoch": 0.5694358974358974,
|
| 9739 |
+
"grad_norm": 4.1875,
|
| 9740 |
+
"learning_rate": 2.1996651318543324e-06,
|
| 9741 |
+
"loss": 1.0907816886901855,
|
| 9742 |
+
"step": 1388
|
| 9743 |
+
},
|
| 9744 |
+
{
|
| 9745 |
+
"epoch": 0.5698461538461539,
|
| 9746 |
+
"grad_norm": 4.25,
|
| 9747 |
+
"learning_rate": 2.1975722059439096e-06,
|
| 9748 |
+
"loss": 1.316295862197876,
|
| 9749 |
+
"step": 1389
|
| 9750 |
+
},
|
| 9751 |
+
{
|
| 9752 |
+
"epoch": 0.5702564102564103,
|
| 9753 |
+
"grad_norm": 3.65625,
|
| 9754 |
+
"learning_rate": 2.1954792800334872e-06,
|
| 9755 |
+
"loss": 1.1051268577575684,
|
| 9756 |
+
"step": 1390
|
| 9757 |
+
},
|
| 9758 |
+
{
|
| 9759 |
+
"epoch": 0.5706666666666667,
|
| 9760 |
+
"grad_norm": 4.1875,
|
| 9761 |
+
"learning_rate": 2.1933863541230644e-06,
|
| 9762 |
+
"loss": 1.3238565921783447,
|
| 9763 |
+
"step": 1391
|
| 9764 |
+
},
|
| 9765 |
+
{
|
| 9766 |
+
"epoch": 0.571076923076923,
|
| 9767 |
+
"grad_norm": 3.8125,
|
| 9768 |
+
"learning_rate": 2.1912934282126412e-06,
|
| 9769 |
+
"loss": 1.360011339187622,
|
| 9770 |
+
"step": 1392
|
| 9771 |
+
},
|
| 9772 |
+
{
|
| 9773 |
+
"epoch": 0.5714871794871795,
|
| 9774 |
+
"grad_norm": 3.8125,
|
| 9775 |
+
"learning_rate": 2.1892005023022185e-06,
|
| 9776 |
+
"loss": 1.145668625831604,
|
| 9777 |
+
"step": 1393
|
| 9778 |
+
},
|
| 9779 |
+
{
|
| 9780 |
+
"epoch": 0.5718974358974359,
|
| 9781 |
+
"grad_norm": 3.625,
|
| 9782 |
+
"learning_rate": 2.187107576391796e-06,
|
| 9783 |
+
"loss": 1.1856372356414795,
|
| 9784 |
+
"step": 1394
|
| 9785 |
+
},
|
| 9786 |
+
{
|
| 9787 |
+
"epoch": 0.5723076923076923,
|
| 9788 |
+
"grad_norm": 4.15625,
|
| 9789 |
+
"learning_rate": 2.1850146504813733e-06,
|
| 9790 |
+
"loss": 1.282850980758667,
|
| 9791 |
+
"step": 1395
|
| 9792 |
+
},
|
| 9793 |
+
{
|
| 9794 |
+
"epoch": 0.5727179487179487,
|
| 9795 |
+
"grad_norm": 3.90625,
|
| 9796 |
+
"learning_rate": 2.18292172457095e-06,
|
| 9797 |
+
"loss": 1.1707267761230469,
|
| 9798 |
+
"step": 1396
|
| 9799 |
+
},
|
| 9800 |
+
{
|
| 9801 |
+
"epoch": 0.5731282051282052,
|
| 9802 |
+
"grad_norm": 4.09375,
|
| 9803 |
+
"learning_rate": 2.1808287986605277e-06,
|
| 9804 |
+
"loss": 1.1335796117782593,
|
| 9805 |
+
"step": 1397
|
| 9806 |
+
},
|
| 9807 |
+
{
|
| 9808 |
+
"epoch": 0.5735384615384616,
|
| 9809 |
+
"grad_norm": 4.5,
|
| 9810 |
+
"learning_rate": 2.178735872750105e-06,
|
| 9811 |
+
"loss": 1.3201531171798706,
|
| 9812 |
+
"step": 1398
|
| 9813 |
+
},
|
| 9814 |
+
{
|
| 9815 |
+
"epoch": 0.5739487179487179,
|
| 9816 |
+
"grad_norm": 4.03125,
|
| 9817 |
+
"learning_rate": 2.176642946839682e-06,
|
| 9818 |
+
"loss": 1.239987850189209,
|
| 9819 |
+
"step": 1399
|
| 9820 |
+
},
|
| 9821 |
+
{
|
| 9822 |
+
"epoch": 0.5743589743589743,
|
| 9823 |
+
"grad_norm": 3.734375,
|
| 9824 |
+
"learning_rate": 2.174550020929259e-06,
|
| 9825 |
+
"loss": 1.1450743675231934,
|
| 9826 |
+
"step": 1400
|
| 9827 |
+
},
|
| 9828 |
+
{
|
| 9829 |
+
"epoch": 0.5747692307692308,
|
| 9830 |
+
"grad_norm": 4.0625,
|
| 9831 |
+
"learning_rate": 2.1724570950188366e-06,
|
| 9832 |
+
"loss": 1.277755856513977,
|
| 9833 |
+
"step": 1401
|
| 9834 |
+
},
|
| 9835 |
+
{
|
| 9836 |
+
"epoch": 0.5751794871794872,
|
| 9837 |
+
"grad_norm": 3.859375,
|
| 9838 |
+
"learning_rate": 2.170364169108414e-06,
|
| 9839 |
+
"loss": 1.1827813386917114,
|
| 9840 |
+
"step": 1402
|
| 9841 |
+
},
|
| 9842 |
+
{
|
| 9843 |
+
"epoch": 0.5755897435897436,
|
| 9844 |
+
"grad_norm": 3.625,
|
| 9845 |
+
"learning_rate": 2.168271243197991e-06,
|
| 9846 |
+
"loss": 1.3474993705749512,
|
| 9847 |
+
"step": 1403
|
| 9848 |
+
},
|
| 9849 |
+
{
|
| 9850 |
+
"epoch": 0.576,
|
| 9851 |
+
"grad_norm": 3.875,
|
| 9852 |
+
"learning_rate": 2.1661783172875682e-06,
|
| 9853 |
+
"loss": 1.3035471439361572,
|
| 9854 |
+
"step": 1404
|
| 9855 |
+
},
|
| 9856 |
+
{
|
| 9857 |
+
"epoch": 0.5764102564102564,
|
| 9858 |
+
"grad_norm": 4.28125,
|
| 9859 |
+
"learning_rate": 2.1640853913771455e-06,
|
| 9860 |
+
"loss": 1.3824542760849,
|
| 9861 |
+
"step": 1405
|
| 9862 |
+
},
|
| 9863 |
+
{
|
| 9864 |
+
"epoch": 0.5768205128205128,
|
| 9865 |
+
"grad_norm": 4.3125,
|
| 9866 |
+
"learning_rate": 2.1619924654667227e-06,
|
| 9867 |
+
"loss": 1.302233099937439,
|
| 9868 |
+
"step": 1406
|
| 9869 |
+
},
|
| 9870 |
+
{
|
| 9871 |
+
"epoch": 0.5772307692307692,
|
| 9872 |
+
"grad_norm": 4.1875,
|
| 9873 |
+
"learning_rate": 2.1598995395563e-06,
|
| 9874 |
+
"loss": 1.2546019554138184,
|
| 9875 |
+
"step": 1407
|
| 9876 |
+
},
|
| 9877 |
+
{
|
| 9878 |
+
"epoch": 0.5776410256410256,
|
| 9879 |
+
"grad_norm": 4.09375,
|
| 9880 |
+
"learning_rate": 2.157806613645877e-06,
|
| 9881 |
+
"loss": 1.342165231704712,
|
| 9882 |
+
"step": 1408
|
| 9883 |
+
},
|
| 9884 |
+
{
|
| 9885 |
+
"epoch": 0.5780512820512821,
|
| 9886 |
+
"grad_norm": 3.84375,
|
| 9887 |
+
"learning_rate": 2.1557136877354543e-06,
|
| 9888 |
+
"loss": 1.2095009088516235,
|
| 9889 |
+
"step": 1409
|
| 9890 |
+
},
|
| 9891 |
+
{
|
| 9892 |
+
"epoch": 0.5784615384615385,
|
| 9893 |
+
"grad_norm": 3.921875,
|
| 9894 |
+
"learning_rate": 2.1536207618250315e-06,
|
| 9895 |
+
"loss": 1.03236985206604,
|
| 9896 |
+
"step": 1410
|
| 9897 |
+
},
|
| 9898 |
+
{
|
| 9899 |
+
"epoch": 0.5788717948717949,
|
| 9900 |
+
"grad_norm": 3.71875,
|
| 9901 |
+
"learning_rate": 2.1515278359146087e-06,
|
| 9902 |
+
"loss": 1.1905386447906494,
|
| 9903 |
+
"step": 1411
|
| 9904 |
+
},
|
| 9905 |
+
{
|
| 9906 |
+
"epoch": 0.5792820512820512,
|
| 9907 |
+
"grad_norm": 4.3125,
|
| 9908 |
+
"learning_rate": 2.149434910004186e-06,
|
| 9909 |
+
"loss": 1.3496975898742676,
|
| 9910 |
+
"step": 1412
|
| 9911 |
+
},
|
| 9912 |
+
{
|
| 9913 |
+
"epoch": 0.5796923076923077,
|
| 9914 |
+
"grad_norm": 3.71875,
|
| 9915 |
+
"learning_rate": 2.147341984093763e-06,
|
| 9916 |
+
"loss": 1.2973510026931763,
|
| 9917 |
+
"step": 1413
|
| 9918 |
+
},
|
| 9919 |
+
{
|
| 9920 |
+
"epoch": 0.5801025641025641,
|
| 9921 |
+
"grad_norm": 3.796875,
|
| 9922 |
+
"learning_rate": 2.1452490581833404e-06,
|
| 9923 |
+
"loss": 1.1965060234069824,
|
| 9924 |
+
"step": 1414
|
| 9925 |
+
},
|
| 9926 |
+
{
|
| 9927 |
+
"epoch": 0.5805128205128205,
|
| 9928 |
+
"grad_norm": 3.84375,
|
| 9929 |
+
"learning_rate": 2.1431561322729176e-06,
|
| 9930 |
+
"loss": 1.1148426532745361,
|
| 9931 |
+
"step": 1415
|
| 9932 |
+
},
|
| 9933 |
+
{
|
| 9934 |
+
"epoch": 0.5809230769230769,
|
| 9935 |
+
"grad_norm": 4.40625,
|
| 9936 |
+
"learning_rate": 2.1410632063624952e-06,
|
| 9937 |
+
"loss": 1.1427874565124512,
|
| 9938 |
+
"step": 1416
|
| 9939 |
+
},
|
| 9940 |
+
{
|
| 9941 |
+
"epoch": 0.5813333333333334,
|
| 9942 |
+
"grad_norm": 4.0,
|
| 9943 |
+
"learning_rate": 2.138970280452072e-06,
|
| 9944 |
+
"loss": 1.3478374481201172,
|
| 9945 |
+
"step": 1417
|
| 9946 |
+
},
|
| 9947 |
+
{
|
| 9948 |
+
"epoch": 0.5817435897435898,
|
| 9949 |
+
"grad_norm": 4.34375,
|
| 9950 |
+
"learning_rate": 2.1368773545416492e-06,
|
| 9951 |
+
"loss": 1.2258148193359375,
|
| 9952 |
+
"step": 1418
|
| 9953 |
+
},
|
| 9954 |
+
{
|
| 9955 |
+
"epoch": 0.5821538461538461,
|
| 9956 |
+
"grad_norm": 3.921875,
|
| 9957 |
+
"learning_rate": 2.1347844286312265e-06,
|
| 9958 |
+
"loss": 1.1401121616363525,
|
| 9959 |
+
"step": 1419
|
| 9960 |
+
},
|
| 9961 |
+
{
|
| 9962 |
+
"epoch": 0.5825641025641025,
|
| 9963 |
+
"grad_norm": 4.28125,
|
| 9964 |
+
"learning_rate": 2.132691502720804e-06,
|
| 9965 |
+
"loss": 1.0413143634796143,
|
| 9966 |
+
"step": 1420
|
| 9967 |
+
},
|
| 9968 |
+
{
|
| 9969 |
+
"epoch": 0.582974358974359,
|
| 9970 |
+
"grad_norm": 4.15625,
|
| 9971 |
+
"learning_rate": 2.130598576810381e-06,
|
| 9972 |
+
"loss": 1.3585100173950195,
|
| 9973 |
+
"step": 1421
|
| 9974 |
+
},
|
| 9975 |
+
{
|
| 9976 |
+
"epoch": 0.5833846153846154,
|
| 9977 |
+
"grad_norm": 3.90625,
|
| 9978 |
+
"learning_rate": 2.128505650899958e-06,
|
| 9979 |
+
"loss": 1.2155094146728516,
|
| 9980 |
+
"step": 1422
|
| 9981 |
+
},
|
| 9982 |
+
{
|
| 9983 |
+
"epoch": 0.5837948717948718,
|
| 9984 |
+
"grad_norm": 4.09375,
|
| 9985 |
+
"learning_rate": 2.1264127249895357e-06,
|
| 9986 |
+
"loss": 1.318897008895874,
|
| 9987 |
+
"step": 1423
|
| 9988 |
+
},
|
| 9989 |
+
{
|
| 9990 |
+
"epoch": 0.5842051282051282,
|
| 9991 |
+
"grad_norm": 3.734375,
|
| 9992 |
+
"learning_rate": 2.124319799079113e-06,
|
| 9993 |
+
"loss": 1.1209454536437988,
|
| 9994 |
+
"step": 1424
|
| 9995 |
+
},
|
| 9996 |
+
{
|
| 9997 |
+
"epoch": 0.5846153846153846,
|
| 9998 |
+
"grad_norm": 4.625,
|
| 9999 |
+
"learning_rate": 2.12222687316869e-06,
|
| 10000 |
+
"loss": 1.1922168731689453,
|
| 10001 |
+
"step": 1425
|
| 10002 |
+
},
|
| 10003 |
+
{
|
| 10004 |
+
"epoch": 0.585025641025641,
|
| 10005 |
+
"grad_norm": 3.875,
|
| 10006 |
+
"learning_rate": 2.120133947258267e-06,
|
| 10007 |
+
"loss": 1.2534267902374268,
|
| 10008 |
+
"step": 1426
|
| 10009 |
+
},
|
| 10010 |
+
{
|
| 10011 |
+
"epoch": 0.5854358974358974,
|
| 10012 |
+
"grad_norm": 4.09375,
|
| 10013 |
+
"learning_rate": 2.1180410213478446e-06,
|
| 10014 |
+
"loss": 1.3108831644058228,
|
| 10015 |
+
"step": 1427
|
| 10016 |
+
},
|
| 10017 |
+
{
|
| 10018 |
+
"epoch": 0.5858461538461538,
|
| 10019 |
+
"grad_norm": 3.71875,
|
| 10020 |
+
"learning_rate": 2.115948095437422e-06,
|
| 10021 |
+
"loss": 1.2236104011535645,
|
| 10022 |
+
"step": 1428
|
| 10023 |
+
},
|
| 10024 |
+
{
|
| 10025 |
+
"epoch": 0.5862564102564103,
|
| 10026 |
+
"grad_norm": 4.28125,
|
| 10027 |
+
"learning_rate": 2.113855169526999e-06,
|
| 10028 |
+
"loss": 1.0898399353027344,
|
| 10029 |
+
"step": 1429
|
| 10030 |
+
},
|
| 10031 |
+
{
|
| 10032 |
+
"epoch": 0.5866666666666667,
|
| 10033 |
+
"grad_norm": 3.5625,
|
| 10034 |
+
"learning_rate": 2.1117622436165762e-06,
|
| 10035 |
+
"loss": 1.1293773651123047,
|
| 10036 |
+
"step": 1430
|
| 10037 |
+
},
|
| 10038 |
+
{
|
| 10039 |
+
"epoch": 0.587076923076923,
|
| 10040 |
+
"grad_norm": 3.75,
|
| 10041 |
+
"learning_rate": 2.1096693177061534e-06,
|
| 10042 |
+
"loss": 1.2320749759674072,
|
| 10043 |
+
"step": 1431
|
| 10044 |
+
},
|
| 10045 |
+
{
|
| 10046 |
+
"epoch": 0.5874871794871794,
|
| 10047 |
+
"grad_norm": 4.125,
|
| 10048 |
+
"learning_rate": 2.1075763917957307e-06,
|
| 10049 |
+
"loss": 1.3349491357803345,
|
| 10050 |
+
"step": 1432
|
| 10051 |
+
},
|
| 10052 |
+
{
|
| 10053 |
+
"epoch": 0.5878974358974359,
|
| 10054 |
+
"grad_norm": 3.765625,
|
| 10055 |
+
"learning_rate": 2.105483465885308e-06,
|
| 10056 |
+
"loss": 1.1549761295318604,
|
| 10057 |
+
"step": 1433
|
| 10058 |
+
},
|
| 10059 |
+
{
|
| 10060 |
+
"epoch": 0.5883076923076923,
|
| 10061 |
+
"grad_norm": 3.796875,
|
| 10062 |
+
"learning_rate": 2.103390539974885e-06,
|
| 10063 |
+
"loss": 1.2241703271865845,
|
| 10064 |
+
"step": 1434
|
| 10065 |
+
},
|
| 10066 |
+
{
|
| 10067 |
+
"epoch": 0.5887179487179487,
|
| 10068 |
+
"grad_norm": 3.875,
|
| 10069 |
+
"learning_rate": 2.1012976140644623e-06,
|
| 10070 |
+
"loss": 1.1840665340423584,
|
| 10071 |
+
"step": 1435
|
| 10072 |
+
},
|
| 10073 |
+
{
|
| 10074 |
+
"epoch": 0.5891282051282051,
|
| 10075 |
+
"grad_norm": 3.921875,
|
| 10076 |
+
"learning_rate": 2.0992046881540395e-06,
|
| 10077 |
+
"loss": 1.2154953479766846,
|
| 10078 |
+
"step": 1436
|
| 10079 |
+
},
|
| 10080 |
+
{
|
| 10081 |
+
"epoch": 0.5895384615384616,
|
| 10082 |
+
"grad_norm": 3.625,
|
| 10083 |
+
"learning_rate": 2.0971117622436167e-06,
|
| 10084 |
+
"loss": 1.0303895473480225,
|
| 10085 |
+
"step": 1437
|
| 10086 |
+
},
|
| 10087 |
+
{
|
| 10088 |
+
"epoch": 0.589948717948718,
|
| 10089 |
+
"grad_norm": 3.875,
|
| 10090 |
+
"learning_rate": 2.095018836333194e-06,
|
| 10091 |
+
"loss": 1.1743472814559937,
|
| 10092 |
+
"step": 1438
|
| 10093 |
+
},
|
| 10094 |
+
{
|
| 10095 |
+
"epoch": 0.5903589743589743,
|
| 10096 |
+
"grad_norm": 3.609375,
|
| 10097 |
+
"learning_rate": 2.092925910422771e-06,
|
| 10098 |
+
"loss": 1.130566954612732,
|
| 10099 |
+
"step": 1439
|
| 10100 |
+
},
|
| 10101 |
+
{
|
| 10102 |
+
"epoch": 0.5907692307692308,
|
| 10103 |
+
"grad_norm": 4.0625,
|
| 10104 |
+
"learning_rate": 2.0908329845123484e-06,
|
| 10105 |
+
"loss": 1.2543385028839111,
|
| 10106 |
+
"step": 1440
|
| 10107 |
+
},
|
| 10108 |
+
{
|
| 10109 |
+
"epoch": 0.5911794871794872,
|
| 10110 |
+
"grad_norm": 3.515625,
|
| 10111 |
+
"learning_rate": 2.0887400586019256e-06,
|
| 10112 |
+
"loss": 1.2341110706329346,
|
| 10113 |
+
"step": 1441
|
| 10114 |
+
},
|
| 10115 |
+
{
|
| 10116 |
+
"epoch": 0.5915897435897436,
|
| 10117 |
+
"grad_norm": 3.65625,
|
| 10118 |
+
"learning_rate": 2.086647132691503e-06,
|
| 10119 |
+
"loss": 0.9825805425643921,
|
| 10120 |
+
"step": 1442
|
| 10121 |
+
},
|
| 10122 |
+
{
|
| 10123 |
+
"epoch": 0.592,
|
| 10124 |
+
"grad_norm": 4.125,
|
| 10125 |
+
"learning_rate": 2.08455420678108e-06,
|
| 10126 |
+
"loss": 1.2022192478179932,
|
| 10127 |
+
"step": 1443
|
| 10128 |
+
},
|
| 10129 |
+
{
|
| 10130 |
+
"epoch": 0.5924102564102565,
|
| 10131 |
+
"grad_norm": 3.828125,
|
| 10132 |
+
"learning_rate": 2.0824612808706572e-06,
|
| 10133 |
+
"loss": 1.2230455875396729,
|
| 10134 |
+
"step": 1444
|
| 10135 |
+
},
|
| 10136 |
+
{
|
| 10137 |
+
"epoch": 0.5928205128205128,
|
| 10138 |
+
"grad_norm": 3.8125,
|
| 10139 |
+
"learning_rate": 2.0803683549602345e-06,
|
| 10140 |
+
"loss": 1.2542235851287842,
|
| 10141 |
+
"step": 1445
|
| 10142 |
+
},
|
| 10143 |
+
{
|
| 10144 |
+
"epoch": 0.5932307692307692,
|
| 10145 |
+
"grad_norm": 3.859375,
|
| 10146 |
+
"learning_rate": 2.078275429049812e-06,
|
| 10147 |
+
"loss": 1.1766939163208008,
|
| 10148 |
+
"step": 1446
|
| 10149 |
+
},
|
| 10150 |
+
{
|
| 10151 |
+
"epoch": 0.5936410256410256,
|
| 10152 |
+
"grad_norm": 4.09375,
|
| 10153 |
+
"learning_rate": 2.076182503139389e-06,
|
| 10154 |
+
"loss": 1.1811625957489014,
|
| 10155 |
+
"step": 1447
|
| 10156 |
+
},
|
| 10157 |
+
{
|
| 10158 |
+
"epoch": 0.5940512820512821,
|
| 10159 |
+
"grad_norm": 3.953125,
|
| 10160 |
+
"learning_rate": 2.074089577228966e-06,
|
| 10161 |
+
"loss": 1.3194475173950195,
|
| 10162 |
+
"step": 1448
|
| 10163 |
+
},
|
| 10164 |
+
{
|
| 10165 |
+
"epoch": 0.5944615384615385,
|
| 10166 |
+
"grad_norm": 3.90625,
|
| 10167 |
+
"learning_rate": 2.0719966513185437e-06,
|
| 10168 |
+
"loss": 1.1240209341049194,
|
| 10169 |
+
"step": 1449
|
| 10170 |
+
},
|
| 10171 |
+
{
|
| 10172 |
+
"epoch": 0.5948717948717949,
|
| 10173 |
+
"grad_norm": 4.5,
|
| 10174 |
+
"learning_rate": 2.069903725408121e-06,
|
| 10175 |
+
"loss": 1.2608507871627808,
|
| 10176 |
+
"step": 1450
|
| 10177 |
+
},
|
| 10178 |
+
{
|
| 10179 |
+
"epoch": 0.5952820512820512,
|
| 10180 |
+
"grad_norm": 3.796875,
|
| 10181 |
+
"learning_rate": 2.0678107994976977e-06,
|
| 10182 |
+
"loss": 1.0938096046447754,
|
| 10183 |
+
"step": 1451
|
| 10184 |
+
},
|
| 10185 |
+
{
|
| 10186 |
+
"epoch": 0.5956923076923077,
|
| 10187 |
+
"grad_norm": 3.890625,
|
| 10188 |
+
"learning_rate": 2.065717873587275e-06,
|
| 10189 |
+
"loss": 1.2084927558898926,
|
| 10190 |
+
"step": 1452
|
| 10191 |
+
},
|
| 10192 |
+
{
|
| 10193 |
+
"epoch": 0.5961025641025641,
|
| 10194 |
+
"grad_norm": 3.796875,
|
| 10195 |
+
"learning_rate": 2.0636249476768526e-06,
|
| 10196 |
+
"loss": 1.2875339984893799,
|
| 10197 |
+
"step": 1453
|
| 10198 |
+
},
|
| 10199 |
+
{
|
| 10200 |
+
"epoch": 0.5965128205128205,
|
| 10201 |
+
"grad_norm": 3.90625,
|
| 10202 |
+
"learning_rate": 2.06153202176643e-06,
|
| 10203 |
+
"loss": 1.095679759979248,
|
| 10204 |
+
"step": 1454
|
| 10205 |
+
},
|
| 10206 |
+
{
|
| 10207 |
+
"epoch": 0.5969230769230769,
|
| 10208 |
+
"grad_norm": 3.71875,
|
| 10209 |
+
"learning_rate": 2.0594390958560066e-06,
|
| 10210 |
+
"loss": 1.0941301584243774,
|
| 10211 |
+
"step": 1455
|
| 10212 |
+
},
|
| 10213 |
+
{
|
| 10214 |
+
"epoch": 0.5973333333333334,
|
| 10215 |
+
"grad_norm": 4.09375,
|
| 10216 |
+
"learning_rate": 2.0573461699455842e-06,
|
| 10217 |
+
"loss": 1.1964309215545654,
|
| 10218 |
+
"step": 1456
|
| 10219 |
+
},
|
| 10220 |
+
{
|
| 10221 |
+
"epoch": 0.5977435897435898,
|
| 10222 |
+
"grad_norm": 4.0,
|
| 10223 |
+
"learning_rate": 2.0552532440351614e-06,
|
| 10224 |
+
"loss": 1.3319376707077026,
|
| 10225 |
+
"step": 1457
|
| 10226 |
+
},
|
| 10227 |
+
{
|
| 10228 |
+
"epoch": 0.5981538461538461,
|
| 10229 |
+
"grad_norm": 4.0,
|
| 10230 |
+
"learning_rate": 2.0531603181247387e-06,
|
| 10231 |
+
"loss": 1.2013217210769653,
|
| 10232 |
+
"step": 1458
|
| 10233 |
+
},
|
| 10234 |
+
{
|
| 10235 |
+
"epoch": 0.5985641025641025,
|
| 10236 |
+
"grad_norm": 3.765625,
|
| 10237 |
+
"learning_rate": 2.051067392214316e-06,
|
| 10238 |
+
"loss": 1.1405173540115356,
|
| 10239 |
+
"step": 1459
|
| 10240 |
+
},
|
| 10241 |
+
{
|
| 10242 |
+
"epoch": 0.598974358974359,
|
| 10243 |
+
"grad_norm": 3.75,
|
| 10244 |
+
"learning_rate": 2.048974466303893e-06,
|
| 10245 |
+
"loss": 1.0882655382156372,
|
| 10246 |
+
"step": 1460
|
| 10247 |
+
},
|
| 10248 |
+
{
|
| 10249 |
+
"epoch": 0.5993846153846154,
|
| 10250 |
+
"grad_norm": 4.03125,
|
| 10251 |
+
"learning_rate": 2.0468815403934703e-06,
|
| 10252 |
+
"loss": 1.1837613582611084,
|
| 10253 |
+
"step": 1461
|
| 10254 |
+
},
|
| 10255 |
+
{
|
| 10256 |
+
"epoch": 0.5997948717948718,
|
| 10257 |
+
"grad_norm": 4.03125,
|
| 10258 |
+
"learning_rate": 2.0447886144830475e-06,
|
| 10259 |
+
"loss": 1.284795880317688,
|
| 10260 |
+
"step": 1462
|
| 10261 |
+
},
|
| 10262 |
+
{
|
| 10263 |
+
"epoch": 0.6002051282051282,
|
| 10264 |
+
"grad_norm": 3.90625,
|
| 10265 |
+
"learning_rate": 2.0426956885726247e-06,
|
| 10266 |
+
"loss": 1.0082780122756958,
|
| 10267 |
+
"step": 1463
|
| 10268 |
+
},
|
| 10269 |
+
{
|
| 10270 |
+
"epoch": 0.6006153846153847,
|
| 10271 |
+
"grad_norm": 3.734375,
|
| 10272 |
+
"learning_rate": 2.040602762662202e-06,
|
| 10273 |
+
"loss": 1.2614877223968506,
|
| 10274 |
+
"step": 1464
|
| 10275 |
+
},
|
| 10276 |
+
{
|
| 10277 |
+
"epoch": 0.6006153846153847,
|
| 10278 |
+
"eval_loss": 1.1979458332061768,
|
| 10279 |
+
"eval_runtime": 12.8503,
|
| 10280 |
+
"eval_samples_per_second": 77.819,
|
| 10281 |
+
"eval_steps_per_second": 19.455,
|
| 10282 |
+
"step": 1464
|
| 10283 |
}
|
| 10284 |
],
|
| 10285 |
"logging_steps": 1,
|
|
|
|
| 10299 |
"attributes": {}
|
| 10300 |
}
|
| 10301 |
},
|
| 10302 |
+
"total_flos": 7.912387772346778e+16,
|
| 10303 |
"train_batch_size": 4,
|
| 10304 |
"trial_name": null,
|
| 10305 |
"trial_params": null
|