Training in progress, step 1500, 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 91951912
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3a844b7ad2f678ba9ab667141220d793a89f23247f72091b4105b3d009194030
|
| 3 |
size 91951912
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 183991226
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1b6f57ee131250f20ef25a1bdeccb9cbf2435d64cbd9cbfecd208c58d056fe59
|
| 3 |
size 183991226
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8d138cfe3a4adf21f048848ee35837c9a757a0a3616ff7adbb45b69aac247435
|
| 3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ba660c20f165b96727b9ef97a65b0525be383413124af3e512d8edda94be4ac5
|
| 3 |
size 1064
|
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": 500,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
@@ -7024,6 +7024,3514 @@
|
|
| 7024 |
"eval_samples_per_second": 27.462,
|
| 7025 |
"eval_steps_per_second": 1.716,
|
| 7026 |
"step": 1000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7027 |
}
|
| 7028 |
],
|
| 7029 |
"logging_steps": 1,
|
|
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.00015,
|
| 6 |
"eval_steps": 500,
|
| 7 |
+
"global_step": 1500,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 7024 |
"eval_samples_per_second": 27.462,
|
| 7025 |
"eval_steps_per_second": 1.716,
|
| 7026 |
"step": 1000
|
| 7027 |
+
},
|
| 7028 |
+
{
|
| 7029 |
+
"epoch": 0.0001001,
|
| 7030 |
+
"grad_norm": 9.124345779418945,
|
| 7031 |
+
"learning_rate": 1e-05,
|
| 7032 |
+
"loss": 60.4375,
|
| 7033 |
+
"step": 1001
|
| 7034 |
+
},
|
| 7035 |
+
{
|
| 7036 |
+
"epoch": 0.0001002,
|
| 7037 |
+
"grad_norm": 9.41726303100586,
|
| 7038 |
+
"learning_rate": 1.001e-05,
|
| 7039 |
+
"loss": 60.0938,
|
| 7040 |
+
"step": 1002
|
| 7041 |
+
},
|
| 7042 |
+
{
|
| 7043 |
+
"epoch": 0.0001003,
|
| 7044 |
+
"grad_norm": 8.93173599243164,
|
| 7045 |
+
"learning_rate": 1.002e-05,
|
| 7046 |
+
"loss": 60.6875,
|
| 7047 |
+
"step": 1003
|
| 7048 |
+
},
|
| 7049 |
+
{
|
| 7050 |
+
"epoch": 0.0001004,
|
| 7051 |
+
"grad_norm": 9.111990928649902,
|
| 7052 |
+
"learning_rate": 1.0029999999999999e-05,
|
| 7053 |
+
"loss": 60.3438,
|
| 7054 |
+
"step": 1004
|
| 7055 |
+
},
|
| 7056 |
+
{
|
| 7057 |
+
"epoch": 0.0001005,
|
| 7058 |
+
"grad_norm": 8.782609939575195,
|
| 7059 |
+
"learning_rate": 1.004e-05,
|
| 7060 |
+
"loss": 61.1562,
|
| 7061 |
+
"step": 1005
|
| 7062 |
+
},
|
| 7063 |
+
{
|
| 7064 |
+
"epoch": 0.0001006,
|
| 7065 |
+
"grad_norm": 9.252752304077148,
|
| 7066 |
+
"learning_rate": 1.005e-05,
|
| 7067 |
+
"loss": 60.1562,
|
| 7068 |
+
"step": 1006
|
| 7069 |
+
},
|
| 7070 |
+
{
|
| 7071 |
+
"epoch": 0.0001007,
|
| 7072 |
+
"grad_norm": 8.887770652770996,
|
| 7073 |
+
"learning_rate": 1.006e-05,
|
| 7074 |
+
"loss": 60.375,
|
| 7075 |
+
"step": 1007
|
| 7076 |
+
},
|
| 7077 |
+
{
|
| 7078 |
+
"epoch": 0.0001008,
|
| 7079 |
+
"grad_norm": 8.99946117401123,
|
| 7080 |
+
"learning_rate": 1.007e-05,
|
| 7081 |
+
"loss": 60.375,
|
| 7082 |
+
"step": 1008
|
| 7083 |
+
},
|
| 7084 |
+
{
|
| 7085 |
+
"epoch": 0.0001009,
|
| 7086 |
+
"grad_norm": 9.170857429504395,
|
| 7087 |
+
"learning_rate": 1.008e-05,
|
| 7088 |
+
"loss": 60.2812,
|
| 7089 |
+
"step": 1009
|
| 7090 |
+
},
|
| 7091 |
+
{
|
| 7092 |
+
"epoch": 0.000101,
|
| 7093 |
+
"grad_norm": 9.04532241821289,
|
| 7094 |
+
"learning_rate": 1.009e-05,
|
| 7095 |
+
"loss": 60.125,
|
| 7096 |
+
"step": 1010
|
| 7097 |
+
},
|
| 7098 |
+
{
|
| 7099 |
+
"epoch": 0.0001011,
|
| 7100 |
+
"grad_norm": 8.962970733642578,
|
| 7101 |
+
"learning_rate": 1.01e-05,
|
| 7102 |
+
"loss": 60.25,
|
| 7103 |
+
"step": 1011
|
| 7104 |
+
},
|
| 7105 |
+
{
|
| 7106 |
+
"epoch": 0.0001012,
|
| 7107 |
+
"grad_norm": 8.685812950134277,
|
| 7108 |
+
"learning_rate": 1.0110000000000001e-05,
|
| 7109 |
+
"loss": 60.6562,
|
| 7110 |
+
"step": 1012
|
| 7111 |
+
},
|
| 7112 |
+
{
|
| 7113 |
+
"epoch": 0.0001013,
|
| 7114 |
+
"grad_norm": 8.879267692565918,
|
| 7115 |
+
"learning_rate": 1.012e-05,
|
| 7116 |
+
"loss": 60.125,
|
| 7117 |
+
"step": 1013
|
| 7118 |
+
},
|
| 7119 |
+
{
|
| 7120 |
+
"epoch": 0.0001014,
|
| 7121 |
+
"grad_norm": 8.940817832946777,
|
| 7122 |
+
"learning_rate": 1.013e-05,
|
| 7123 |
+
"loss": 60.25,
|
| 7124 |
+
"step": 1014
|
| 7125 |
+
},
|
| 7126 |
+
{
|
| 7127 |
+
"epoch": 0.0001015,
|
| 7128 |
+
"grad_norm": 8.88219165802002,
|
| 7129 |
+
"learning_rate": 1.0140000000000001e-05,
|
| 7130 |
+
"loss": 60.25,
|
| 7131 |
+
"step": 1015
|
| 7132 |
+
},
|
| 7133 |
+
{
|
| 7134 |
+
"epoch": 0.0001016,
|
| 7135 |
+
"grad_norm": 9.43167781829834,
|
| 7136 |
+
"learning_rate": 1.015e-05,
|
| 7137 |
+
"loss": 60.0,
|
| 7138 |
+
"step": 1016
|
| 7139 |
+
},
|
| 7140 |
+
{
|
| 7141 |
+
"epoch": 0.0001017,
|
| 7142 |
+
"grad_norm": 9.163987159729004,
|
| 7143 |
+
"learning_rate": 1.0160000000000001e-05,
|
| 7144 |
+
"loss": 59.6562,
|
| 7145 |
+
"step": 1017
|
| 7146 |
+
},
|
| 7147 |
+
{
|
| 7148 |
+
"epoch": 0.0001018,
|
| 7149 |
+
"grad_norm": 8.8132905960083,
|
| 7150 |
+
"learning_rate": 1.017e-05,
|
| 7151 |
+
"loss": 60.0312,
|
| 7152 |
+
"step": 1018
|
| 7153 |
+
},
|
| 7154 |
+
{
|
| 7155 |
+
"epoch": 0.0001019,
|
| 7156 |
+
"grad_norm": 9.031733512878418,
|
| 7157 |
+
"learning_rate": 1.018e-05,
|
| 7158 |
+
"loss": 59.9688,
|
| 7159 |
+
"step": 1019
|
| 7160 |
+
},
|
| 7161 |
+
{
|
| 7162 |
+
"epoch": 0.000102,
|
| 7163 |
+
"grad_norm": 9.099145889282227,
|
| 7164 |
+
"learning_rate": 1.019e-05,
|
| 7165 |
+
"loss": 59.6562,
|
| 7166 |
+
"step": 1020
|
| 7167 |
+
},
|
| 7168 |
+
{
|
| 7169 |
+
"epoch": 0.0001021,
|
| 7170 |
+
"grad_norm": 9.057586669921875,
|
| 7171 |
+
"learning_rate": 1.0199999999999999e-05,
|
| 7172 |
+
"loss": 59.75,
|
| 7173 |
+
"step": 1021
|
| 7174 |
+
},
|
| 7175 |
+
{
|
| 7176 |
+
"epoch": 0.0001022,
|
| 7177 |
+
"grad_norm": 8.728660583496094,
|
| 7178 |
+
"learning_rate": 1.021e-05,
|
| 7179 |
+
"loss": 60.0938,
|
| 7180 |
+
"step": 1022
|
| 7181 |
+
},
|
| 7182 |
+
{
|
| 7183 |
+
"epoch": 0.0001023,
|
| 7184 |
+
"grad_norm": 9.054150581359863,
|
| 7185 |
+
"learning_rate": 1.022e-05,
|
| 7186 |
+
"loss": 59.5,
|
| 7187 |
+
"step": 1023
|
| 7188 |
+
},
|
| 7189 |
+
{
|
| 7190 |
+
"epoch": 0.0001024,
|
| 7191 |
+
"grad_norm": 8.890925407409668,
|
| 7192 |
+
"learning_rate": 1.023e-05,
|
| 7193 |
+
"loss": 59.875,
|
| 7194 |
+
"step": 1024
|
| 7195 |
+
},
|
| 7196 |
+
{
|
| 7197 |
+
"epoch": 0.0001025,
|
| 7198 |
+
"grad_norm": 8.602814674377441,
|
| 7199 |
+
"learning_rate": 1.024e-05,
|
| 7200 |
+
"loss": 60.0938,
|
| 7201 |
+
"step": 1025
|
| 7202 |
+
},
|
| 7203 |
+
{
|
| 7204 |
+
"epoch": 0.0001026,
|
| 7205 |
+
"grad_norm": 8.891504287719727,
|
| 7206 |
+
"learning_rate": 1.025e-05,
|
| 7207 |
+
"loss": 59.4375,
|
| 7208 |
+
"step": 1026
|
| 7209 |
+
},
|
| 7210 |
+
{
|
| 7211 |
+
"epoch": 0.0001027,
|
| 7212 |
+
"grad_norm": 8.879639625549316,
|
| 7213 |
+
"learning_rate": 1.026e-05,
|
| 7214 |
+
"loss": 59.9062,
|
| 7215 |
+
"step": 1027
|
| 7216 |
+
},
|
| 7217 |
+
{
|
| 7218 |
+
"epoch": 0.0001028,
|
| 7219 |
+
"grad_norm": 9.04471206665039,
|
| 7220 |
+
"learning_rate": 1.027e-05,
|
| 7221 |
+
"loss": 59.3125,
|
| 7222 |
+
"step": 1028
|
| 7223 |
+
},
|
| 7224 |
+
{
|
| 7225 |
+
"epoch": 0.0001029,
|
| 7226 |
+
"grad_norm": 8.917312622070312,
|
| 7227 |
+
"learning_rate": 1.0280000000000002e-05,
|
| 7228 |
+
"loss": 59.9375,
|
| 7229 |
+
"step": 1029
|
| 7230 |
+
},
|
| 7231 |
+
{
|
| 7232 |
+
"epoch": 0.000103,
|
| 7233 |
+
"grad_norm": 8.734441757202148,
|
| 7234 |
+
"learning_rate": 1.029e-05,
|
| 7235 |
+
"loss": 59.9688,
|
| 7236 |
+
"step": 1030
|
| 7237 |
+
},
|
| 7238 |
+
{
|
| 7239 |
+
"epoch": 0.0001031,
|
| 7240 |
+
"grad_norm": 9.322687149047852,
|
| 7241 |
+
"learning_rate": 1.03e-05,
|
| 7242 |
+
"loss": 59.0312,
|
| 7243 |
+
"step": 1031
|
| 7244 |
+
},
|
| 7245 |
+
{
|
| 7246 |
+
"epoch": 0.0001032,
|
| 7247 |
+
"grad_norm": 8.85023307800293,
|
| 7248 |
+
"learning_rate": 1.0310000000000001e-05,
|
| 7249 |
+
"loss": 59.5312,
|
| 7250 |
+
"step": 1032
|
| 7251 |
+
},
|
| 7252 |
+
{
|
| 7253 |
+
"epoch": 0.0001033,
|
| 7254 |
+
"grad_norm": 8.855387687683105,
|
| 7255 |
+
"learning_rate": 1.032e-05,
|
| 7256 |
+
"loss": 59.5312,
|
| 7257 |
+
"step": 1033
|
| 7258 |
+
},
|
| 7259 |
+
{
|
| 7260 |
+
"epoch": 0.0001034,
|
| 7261 |
+
"grad_norm": 9.083724021911621,
|
| 7262 |
+
"learning_rate": 1.0330000000000001e-05,
|
| 7263 |
+
"loss": 59.125,
|
| 7264 |
+
"step": 1034
|
| 7265 |
+
},
|
| 7266 |
+
{
|
| 7267 |
+
"epoch": 0.0001035,
|
| 7268 |
+
"grad_norm": 8.690860748291016,
|
| 7269 |
+
"learning_rate": 1.034e-05,
|
| 7270 |
+
"loss": 59.5938,
|
| 7271 |
+
"step": 1035
|
| 7272 |
+
},
|
| 7273 |
+
{
|
| 7274 |
+
"epoch": 0.0001036,
|
| 7275 |
+
"grad_norm": 8.5335693359375,
|
| 7276 |
+
"learning_rate": 1.0350000000000001e-05,
|
| 7277 |
+
"loss": 59.875,
|
| 7278 |
+
"step": 1036
|
| 7279 |
+
},
|
| 7280 |
+
{
|
| 7281 |
+
"epoch": 0.0001037,
|
| 7282 |
+
"grad_norm": 8.61517333984375,
|
| 7283 |
+
"learning_rate": 1.036e-05,
|
| 7284 |
+
"loss": 59.7812,
|
| 7285 |
+
"step": 1037
|
| 7286 |
+
},
|
| 7287 |
+
{
|
| 7288 |
+
"epoch": 0.0001038,
|
| 7289 |
+
"grad_norm": 8.838473320007324,
|
| 7290 |
+
"learning_rate": 1.0369999999999999e-05,
|
| 7291 |
+
"loss": 59.3125,
|
| 7292 |
+
"step": 1038
|
| 7293 |
+
},
|
| 7294 |
+
{
|
| 7295 |
+
"epoch": 0.0001039,
|
| 7296 |
+
"grad_norm": 8.471953392028809,
|
| 7297 |
+
"learning_rate": 1.038e-05,
|
| 7298 |
+
"loss": 59.9375,
|
| 7299 |
+
"step": 1039
|
| 7300 |
+
},
|
| 7301 |
+
{
|
| 7302 |
+
"epoch": 0.000104,
|
| 7303 |
+
"grad_norm": 8.951537132263184,
|
| 7304 |
+
"learning_rate": 1.039e-05,
|
| 7305 |
+
"loss": 59.4375,
|
| 7306 |
+
"step": 1040
|
| 7307 |
+
},
|
| 7308 |
+
{
|
| 7309 |
+
"epoch": 0.0001041,
|
| 7310 |
+
"grad_norm": 9.035890579223633,
|
| 7311 |
+
"learning_rate": 1.04e-05,
|
| 7312 |
+
"loss": 59.0938,
|
| 7313 |
+
"step": 1041
|
| 7314 |
+
},
|
| 7315 |
+
{
|
| 7316 |
+
"epoch": 0.0001042,
|
| 7317 |
+
"grad_norm": 9.102971076965332,
|
| 7318 |
+
"learning_rate": 1.041e-05,
|
| 7319 |
+
"loss": 58.9062,
|
| 7320 |
+
"step": 1042
|
| 7321 |
+
},
|
| 7322 |
+
{
|
| 7323 |
+
"epoch": 0.0001043,
|
| 7324 |
+
"grad_norm": 9.211016654968262,
|
| 7325 |
+
"learning_rate": 1.042e-05,
|
| 7326 |
+
"loss": 58.5625,
|
| 7327 |
+
"step": 1043
|
| 7328 |
+
},
|
| 7329 |
+
{
|
| 7330 |
+
"epoch": 0.0001044,
|
| 7331 |
+
"grad_norm": 8.802969932556152,
|
| 7332 |
+
"learning_rate": 1.043e-05,
|
| 7333 |
+
"loss": 59.125,
|
| 7334 |
+
"step": 1044
|
| 7335 |
+
},
|
| 7336 |
+
{
|
| 7337 |
+
"epoch": 0.0001045,
|
| 7338 |
+
"grad_norm": 8.844217300415039,
|
| 7339 |
+
"learning_rate": 1.044e-05,
|
| 7340 |
+
"loss": 59.0,
|
| 7341 |
+
"step": 1045
|
| 7342 |
+
},
|
| 7343 |
+
{
|
| 7344 |
+
"epoch": 0.0001046,
|
| 7345 |
+
"grad_norm": 8.752573013305664,
|
| 7346 |
+
"learning_rate": 1.045e-05,
|
| 7347 |
+
"loss": 59.3125,
|
| 7348 |
+
"step": 1046
|
| 7349 |
+
},
|
| 7350 |
+
{
|
| 7351 |
+
"epoch": 0.0001047,
|
| 7352 |
+
"grad_norm": 8.805431365966797,
|
| 7353 |
+
"learning_rate": 1.046e-05,
|
| 7354 |
+
"loss": 58.9688,
|
| 7355 |
+
"step": 1047
|
| 7356 |
+
},
|
| 7357 |
+
{
|
| 7358 |
+
"epoch": 0.0001048,
|
| 7359 |
+
"grad_norm": 8.648407936096191,
|
| 7360 |
+
"learning_rate": 1.047e-05,
|
| 7361 |
+
"loss": 58.9688,
|
| 7362 |
+
"step": 1048
|
| 7363 |
+
},
|
| 7364 |
+
{
|
| 7365 |
+
"epoch": 0.0001049,
|
| 7366 |
+
"grad_norm": 8.364664077758789,
|
| 7367 |
+
"learning_rate": 1.0480000000000001e-05,
|
| 7368 |
+
"loss": 59.5938,
|
| 7369 |
+
"step": 1049
|
| 7370 |
+
},
|
| 7371 |
+
{
|
| 7372 |
+
"epoch": 0.000105,
|
| 7373 |
+
"grad_norm": 8.843344688415527,
|
| 7374 |
+
"learning_rate": 1.049e-05,
|
| 7375 |
+
"loss": 59.0,
|
| 7376 |
+
"step": 1050
|
| 7377 |
+
},
|
| 7378 |
+
{
|
| 7379 |
+
"epoch": 0.0001051,
|
| 7380 |
+
"grad_norm": 8.329179763793945,
|
| 7381 |
+
"learning_rate": 1.0500000000000001e-05,
|
| 7382 |
+
"loss": 59.4062,
|
| 7383 |
+
"step": 1051
|
| 7384 |
+
},
|
| 7385 |
+
{
|
| 7386 |
+
"epoch": 0.0001052,
|
| 7387 |
+
"grad_norm": 8.471250534057617,
|
| 7388 |
+
"learning_rate": 1.051e-05,
|
| 7389 |
+
"loss": 59.3438,
|
| 7390 |
+
"step": 1052
|
| 7391 |
+
},
|
| 7392 |
+
{
|
| 7393 |
+
"epoch": 0.0001053,
|
| 7394 |
+
"grad_norm": 8.373708724975586,
|
| 7395 |
+
"learning_rate": 1.052e-05,
|
| 7396 |
+
"loss": 59.375,
|
| 7397 |
+
"step": 1053
|
| 7398 |
+
},
|
| 7399 |
+
{
|
| 7400 |
+
"epoch": 0.0001054,
|
| 7401 |
+
"grad_norm": 8.504606246948242,
|
| 7402 |
+
"learning_rate": 1.053e-05,
|
| 7403 |
+
"loss": 59.6562,
|
| 7404 |
+
"step": 1054
|
| 7405 |
+
},
|
| 7406 |
+
{
|
| 7407 |
+
"epoch": 0.0001055,
|
| 7408 |
+
"grad_norm": 8.886446952819824,
|
| 7409 |
+
"learning_rate": 1.0539999999999999e-05,
|
| 7410 |
+
"loss": 58.7188,
|
| 7411 |
+
"step": 1055
|
| 7412 |
+
},
|
| 7413 |
+
{
|
| 7414 |
+
"epoch": 0.0001056,
|
| 7415 |
+
"grad_norm": 8.751822471618652,
|
| 7416 |
+
"learning_rate": 1.055e-05,
|
| 7417 |
+
"loss": 58.8438,
|
| 7418 |
+
"step": 1056
|
| 7419 |
+
},
|
| 7420 |
+
{
|
| 7421 |
+
"epoch": 0.0001057,
|
| 7422 |
+
"grad_norm": 8.406770706176758,
|
| 7423 |
+
"learning_rate": 1.056e-05,
|
| 7424 |
+
"loss": 59.2188,
|
| 7425 |
+
"step": 1057
|
| 7426 |
+
},
|
| 7427 |
+
{
|
| 7428 |
+
"epoch": 0.0001058,
|
| 7429 |
+
"grad_norm": 8.74715805053711,
|
| 7430 |
+
"learning_rate": 1.057e-05,
|
| 7431 |
+
"loss": 58.6562,
|
| 7432 |
+
"step": 1058
|
| 7433 |
+
},
|
| 7434 |
+
{
|
| 7435 |
+
"epoch": 0.0001059,
|
| 7436 |
+
"grad_norm": 8.3496732711792,
|
| 7437 |
+
"learning_rate": 1.058e-05,
|
| 7438 |
+
"loss": 59.0312,
|
| 7439 |
+
"step": 1059
|
| 7440 |
+
},
|
| 7441 |
+
{
|
| 7442 |
+
"epoch": 0.000106,
|
| 7443 |
+
"grad_norm": 8.718629837036133,
|
| 7444 |
+
"learning_rate": 1.059e-05,
|
| 7445 |
+
"loss": 58.4688,
|
| 7446 |
+
"step": 1060
|
| 7447 |
+
},
|
| 7448 |
+
{
|
| 7449 |
+
"epoch": 0.0001061,
|
| 7450 |
+
"grad_norm": 8.428848266601562,
|
| 7451 |
+
"learning_rate": 1.06e-05,
|
| 7452 |
+
"loss": 58.9688,
|
| 7453 |
+
"step": 1061
|
| 7454 |
+
},
|
| 7455 |
+
{
|
| 7456 |
+
"epoch": 0.0001062,
|
| 7457 |
+
"grad_norm": 9.044172286987305,
|
| 7458 |
+
"learning_rate": 1.061e-05,
|
| 7459 |
+
"loss": 58.0625,
|
| 7460 |
+
"step": 1062
|
| 7461 |
+
},
|
| 7462 |
+
{
|
| 7463 |
+
"epoch": 0.0001063,
|
| 7464 |
+
"grad_norm": 8.45057487487793,
|
| 7465 |
+
"learning_rate": 1.062e-05,
|
| 7466 |
+
"loss": 58.7188,
|
| 7467 |
+
"step": 1063
|
| 7468 |
+
},
|
| 7469 |
+
{
|
| 7470 |
+
"epoch": 0.0001064,
|
| 7471 |
+
"grad_norm": 8.13118839263916,
|
| 7472 |
+
"learning_rate": 1.063e-05,
|
| 7473 |
+
"loss": 58.9062,
|
| 7474 |
+
"step": 1064
|
| 7475 |
+
},
|
| 7476 |
+
{
|
| 7477 |
+
"epoch": 0.0001065,
|
| 7478 |
+
"grad_norm": 8.686187744140625,
|
| 7479 |
+
"learning_rate": 1.064e-05,
|
| 7480 |
+
"loss": 58.9688,
|
| 7481 |
+
"step": 1065
|
| 7482 |
+
},
|
| 7483 |
+
{
|
| 7484 |
+
"epoch": 0.0001066,
|
| 7485 |
+
"grad_norm": 8.561543464660645,
|
| 7486 |
+
"learning_rate": 1.0650000000000001e-05,
|
| 7487 |
+
"loss": 58.3438,
|
| 7488 |
+
"step": 1066
|
| 7489 |
+
},
|
| 7490 |
+
{
|
| 7491 |
+
"epoch": 0.0001067,
|
| 7492 |
+
"grad_norm": 8.650547981262207,
|
| 7493 |
+
"learning_rate": 1.066e-05,
|
| 7494 |
+
"loss": 58.5,
|
| 7495 |
+
"step": 1067
|
| 7496 |
+
},
|
| 7497 |
+
{
|
| 7498 |
+
"epoch": 0.0001068,
|
| 7499 |
+
"grad_norm": 8.45773983001709,
|
| 7500 |
+
"learning_rate": 1.0670000000000001e-05,
|
| 7501 |
+
"loss": 58.625,
|
| 7502 |
+
"step": 1068
|
| 7503 |
+
},
|
| 7504 |
+
{
|
| 7505 |
+
"epoch": 0.0001069,
|
| 7506 |
+
"grad_norm": 8.8767671585083,
|
| 7507 |
+
"learning_rate": 1.068e-05,
|
| 7508 |
+
"loss": 57.8438,
|
| 7509 |
+
"step": 1069
|
| 7510 |
+
},
|
| 7511 |
+
{
|
| 7512 |
+
"epoch": 0.000107,
|
| 7513 |
+
"grad_norm": 8.721537590026855,
|
| 7514 |
+
"learning_rate": 1.069e-05,
|
| 7515 |
+
"loss": 57.7812,
|
| 7516 |
+
"step": 1070
|
| 7517 |
+
},
|
| 7518 |
+
{
|
| 7519 |
+
"epoch": 0.0001071,
|
| 7520 |
+
"grad_norm": 8.56977653503418,
|
| 7521 |
+
"learning_rate": 1.0700000000000001e-05,
|
| 7522 |
+
"loss": 58.2812,
|
| 7523 |
+
"step": 1071
|
| 7524 |
+
},
|
| 7525 |
+
{
|
| 7526 |
+
"epoch": 0.0001072,
|
| 7527 |
+
"grad_norm": 8.458526611328125,
|
| 7528 |
+
"learning_rate": 1.0709999999999999e-05,
|
| 7529 |
+
"loss": 58.875,
|
| 7530 |
+
"step": 1072
|
| 7531 |
+
},
|
| 7532 |
+
{
|
| 7533 |
+
"epoch": 0.0001073,
|
| 7534 |
+
"grad_norm": 8.45875358581543,
|
| 7535 |
+
"learning_rate": 1.072e-05,
|
| 7536 |
+
"loss": 58.7188,
|
| 7537 |
+
"step": 1073
|
| 7538 |
+
},
|
| 7539 |
+
{
|
| 7540 |
+
"epoch": 0.0001074,
|
| 7541 |
+
"grad_norm": 8.583601951599121,
|
| 7542 |
+
"learning_rate": 1.073e-05,
|
| 7543 |
+
"loss": 58.2812,
|
| 7544 |
+
"step": 1074
|
| 7545 |
+
},
|
| 7546 |
+
{
|
| 7547 |
+
"epoch": 0.0001075,
|
| 7548 |
+
"grad_norm": 8.412371635437012,
|
| 7549 |
+
"learning_rate": 1.074e-05,
|
| 7550 |
+
"loss": 58.4062,
|
| 7551 |
+
"step": 1075
|
| 7552 |
+
},
|
| 7553 |
+
{
|
| 7554 |
+
"epoch": 0.0001076,
|
| 7555 |
+
"grad_norm": 8.315399169921875,
|
| 7556 |
+
"learning_rate": 1.075e-05,
|
| 7557 |
+
"loss": 58.3438,
|
| 7558 |
+
"step": 1076
|
| 7559 |
+
},
|
| 7560 |
+
{
|
| 7561 |
+
"epoch": 0.0001077,
|
| 7562 |
+
"grad_norm": 8.517126083374023,
|
| 7563 |
+
"learning_rate": 1.076e-05,
|
| 7564 |
+
"loss": 58.3125,
|
| 7565 |
+
"step": 1077
|
| 7566 |
+
},
|
| 7567 |
+
{
|
| 7568 |
+
"epoch": 0.0001078,
|
| 7569 |
+
"grad_norm": 8.215516090393066,
|
| 7570 |
+
"learning_rate": 1.077e-05,
|
| 7571 |
+
"loss": 58.5938,
|
| 7572 |
+
"step": 1078
|
| 7573 |
+
},
|
| 7574 |
+
{
|
| 7575 |
+
"epoch": 0.0001079,
|
| 7576 |
+
"grad_norm": 8.485785484313965,
|
| 7577 |
+
"learning_rate": 1.078e-05,
|
| 7578 |
+
"loss": 58.0312,
|
| 7579 |
+
"step": 1079
|
| 7580 |
+
},
|
| 7581 |
+
{
|
| 7582 |
+
"epoch": 0.000108,
|
| 7583 |
+
"grad_norm": 8.737732887268066,
|
| 7584 |
+
"learning_rate": 1.079e-05,
|
| 7585 |
+
"loss": 57.9375,
|
| 7586 |
+
"step": 1080
|
| 7587 |
+
},
|
| 7588 |
+
{
|
| 7589 |
+
"epoch": 0.0001081,
|
| 7590 |
+
"grad_norm": 8.925642967224121,
|
| 7591 |
+
"learning_rate": 1.08e-05,
|
| 7592 |
+
"loss": 57.2812,
|
| 7593 |
+
"step": 1081
|
| 7594 |
+
},
|
| 7595 |
+
{
|
| 7596 |
+
"epoch": 0.0001082,
|
| 7597 |
+
"grad_norm": 8.599075317382812,
|
| 7598 |
+
"learning_rate": 1.081e-05,
|
| 7599 |
+
"loss": 57.9062,
|
| 7600 |
+
"step": 1082
|
| 7601 |
+
},
|
| 7602 |
+
{
|
| 7603 |
+
"epoch": 0.0001083,
|
| 7604 |
+
"grad_norm": 8.235535621643066,
|
| 7605 |
+
"learning_rate": 1.0820000000000001e-05,
|
| 7606 |
+
"loss": 58.8125,
|
| 7607 |
+
"step": 1083
|
| 7608 |
+
},
|
| 7609 |
+
{
|
| 7610 |
+
"epoch": 0.0001084,
|
| 7611 |
+
"grad_norm": 8.46491527557373,
|
| 7612 |
+
"learning_rate": 1.083e-05,
|
| 7613 |
+
"loss": 58.0938,
|
| 7614 |
+
"step": 1084
|
| 7615 |
+
},
|
| 7616 |
+
{
|
| 7617 |
+
"epoch": 0.0001085,
|
| 7618 |
+
"grad_norm": 8.536995887756348,
|
| 7619 |
+
"learning_rate": 1.0840000000000001e-05,
|
| 7620 |
+
"loss": 57.75,
|
| 7621 |
+
"step": 1085
|
| 7622 |
+
},
|
| 7623 |
+
{
|
| 7624 |
+
"epoch": 0.0001086,
|
| 7625 |
+
"grad_norm": 8.33989143371582,
|
| 7626 |
+
"learning_rate": 1.085e-05,
|
| 7627 |
+
"loss": 58.4375,
|
| 7628 |
+
"step": 1086
|
| 7629 |
+
},
|
| 7630 |
+
{
|
| 7631 |
+
"epoch": 0.0001087,
|
| 7632 |
+
"grad_norm": 8.58334732055664,
|
| 7633 |
+
"learning_rate": 1.086e-05,
|
| 7634 |
+
"loss": 58.0312,
|
| 7635 |
+
"step": 1087
|
| 7636 |
+
},
|
| 7637 |
+
{
|
| 7638 |
+
"epoch": 0.0001088,
|
| 7639 |
+
"grad_norm": 8.356263160705566,
|
| 7640 |
+
"learning_rate": 1.0870000000000001e-05,
|
| 7641 |
+
"loss": 58.0,
|
| 7642 |
+
"step": 1088
|
| 7643 |
+
},
|
| 7644 |
+
{
|
| 7645 |
+
"epoch": 0.0001089,
|
| 7646 |
+
"grad_norm": 8.417243003845215,
|
| 7647 |
+
"learning_rate": 1.088e-05,
|
| 7648 |
+
"loss": 57.9375,
|
| 7649 |
+
"step": 1089
|
| 7650 |
+
},
|
| 7651 |
+
{
|
| 7652 |
+
"epoch": 0.000109,
|
| 7653 |
+
"grad_norm": 8.363197326660156,
|
| 7654 |
+
"learning_rate": 1.089e-05,
|
| 7655 |
+
"loss": 57.875,
|
| 7656 |
+
"step": 1090
|
| 7657 |
+
},
|
| 7658 |
+
{
|
| 7659 |
+
"epoch": 0.0001091,
|
| 7660 |
+
"grad_norm": 8.385573387145996,
|
| 7661 |
+
"learning_rate": 1.09e-05,
|
| 7662 |
+
"loss": 57.625,
|
| 7663 |
+
"step": 1091
|
| 7664 |
+
},
|
| 7665 |
+
{
|
| 7666 |
+
"epoch": 0.0001092,
|
| 7667 |
+
"grad_norm": 8.56344223022461,
|
| 7668 |
+
"learning_rate": 1.091e-05,
|
| 7669 |
+
"loss": 58.0938,
|
| 7670 |
+
"step": 1092
|
| 7671 |
+
},
|
| 7672 |
+
{
|
| 7673 |
+
"epoch": 0.0001093,
|
| 7674 |
+
"grad_norm": 8.359785079956055,
|
| 7675 |
+
"learning_rate": 1.092e-05,
|
| 7676 |
+
"loss": 57.7812,
|
| 7677 |
+
"step": 1093
|
| 7678 |
+
},
|
| 7679 |
+
{
|
| 7680 |
+
"epoch": 0.0001094,
|
| 7681 |
+
"grad_norm": 8.217353820800781,
|
| 7682 |
+
"learning_rate": 1.093e-05,
|
| 7683 |
+
"loss": 58.125,
|
| 7684 |
+
"step": 1094
|
| 7685 |
+
},
|
| 7686 |
+
{
|
| 7687 |
+
"epoch": 0.0001095,
|
| 7688 |
+
"grad_norm": 8.240918159484863,
|
| 7689 |
+
"learning_rate": 1.094e-05,
|
| 7690 |
+
"loss": 58.2188,
|
| 7691 |
+
"step": 1095
|
| 7692 |
+
},
|
| 7693 |
+
{
|
| 7694 |
+
"epoch": 0.0001096,
|
| 7695 |
+
"grad_norm": 8.025176048278809,
|
| 7696 |
+
"learning_rate": 1.095e-05,
|
| 7697 |
+
"loss": 57.6562,
|
| 7698 |
+
"step": 1096
|
| 7699 |
+
},
|
| 7700 |
+
{
|
| 7701 |
+
"epoch": 0.0001097,
|
| 7702 |
+
"grad_norm": 7.992165565490723,
|
| 7703 |
+
"learning_rate": 1.096e-05,
|
| 7704 |
+
"loss": 57.875,
|
| 7705 |
+
"step": 1097
|
| 7706 |
+
},
|
| 7707 |
+
{
|
| 7708 |
+
"epoch": 0.0001098,
|
| 7709 |
+
"grad_norm": 8.216459274291992,
|
| 7710 |
+
"learning_rate": 1.097e-05,
|
| 7711 |
+
"loss": 57.9062,
|
| 7712 |
+
"step": 1098
|
| 7713 |
+
},
|
| 7714 |
+
{
|
| 7715 |
+
"epoch": 0.0001099,
|
| 7716 |
+
"grad_norm": 8.325234413146973,
|
| 7717 |
+
"learning_rate": 1.098e-05,
|
| 7718 |
+
"loss": 57.5625,
|
| 7719 |
+
"step": 1099
|
| 7720 |
+
},
|
| 7721 |
+
{
|
| 7722 |
+
"epoch": 0.00011,
|
| 7723 |
+
"grad_norm": 8.516204833984375,
|
| 7724 |
+
"learning_rate": 1.0990000000000002e-05,
|
| 7725 |
+
"loss": 57.1875,
|
| 7726 |
+
"step": 1100
|
| 7727 |
+
},
|
| 7728 |
+
{
|
| 7729 |
+
"epoch": 0.0001101,
|
| 7730 |
+
"grad_norm": 8.354333877563477,
|
| 7731 |
+
"learning_rate": 1.1e-05,
|
| 7732 |
+
"loss": 57.4062,
|
| 7733 |
+
"step": 1101
|
| 7734 |
+
},
|
| 7735 |
+
{
|
| 7736 |
+
"epoch": 0.0001102,
|
| 7737 |
+
"grad_norm": 8.326759338378906,
|
| 7738 |
+
"learning_rate": 1.1010000000000001e-05,
|
| 7739 |
+
"loss": 56.9375,
|
| 7740 |
+
"step": 1102
|
| 7741 |
+
},
|
| 7742 |
+
{
|
| 7743 |
+
"epoch": 0.0001103,
|
| 7744 |
+
"grad_norm": 8.392999649047852,
|
| 7745 |
+
"learning_rate": 1.102e-05,
|
| 7746 |
+
"loss": 57.0,
|
| 7747 |
+
"step": 1103
|
| 7748 |
+
},
|
| 7749 |
+
{
|
| 7750 |
+
"epoch": 0.0001104,
|
| 7751 |
+
"grad_norm": 8.289926528930664,
|
| 7752 |
+
"learning_rate": 1.103e-05,
|
| 7753 |
+
"loss": 57.5625,
|
| 7754 |
+
"step": 1104
|
| 7755 |
+
},
|
| 7756 |
+
{
|
| 7757 |
+
"epoch": 0.0001105,
|
| 7758 |
+
"grad_norm": 8.27065658569336,
|
| 7759 |
+
"learning_rate": 1.1040000000000001e-05,
|
| 7760 |
+
"loss": 56.9375,
|
| 7761 |
+
"step": 1105
|
| 7762 |
+
},
|
| 7763 |
+
{
|
| 7764 |
+
"epoch": 0.0001106,
|
| 7765 |
+
"grad_norm": 7.944595813751221,
|
| 7766 |
+
"learning_rate": 1.105e-05,
|
| 7767 |
+
"loss": 58.125,
|
| 7768 |
+
"step": 1106
|
| 7769 |
+
},
|
| 7770 |
+
{
|
| 7771 |
+
"epoch": 0.0001107,
|
| 7772 |
+
"grad_norm": 8.095520973205566,
|
| 7773 |
+
"learning_rate": 1.106e-05,
|
| 7774 |
+
"loss": 57.6875,
|
| 7775 |
+
"step": 1107
|
| 7776 |
+
},
|
| 7777 |
+
{
|
| 7778 |
+
"epoch": 0.0001108,
|
| 7779 |
+
"grad_norm": 8.457866668701172,
|
| 7780 |
+
"learning_rate": 1.107e-05,
|
| 7781 |
+
"loss": 56.8438,
|
| 7782 |
+
"step": 1108
|
| 7783 |
+
},
|
| 7784 |
+
{
|
| 7785 |
+
"epoch": 0.0001109,
|
| 7786 |
+
"grad_norm": 8.350686073303223,
|
| 7787 |
+
"learning_rate": 1.108e-05,
|
| 7788 |
+
"loss": 57.0,
|
| 7789 |
+
"step": 1109
|
| 7790 |
+
},
|
| 7791 |
+
{
|
| 7792 |
+
"epoch": 0.000111,
|
| 7793 |
+
"grad_norm": 8.33674430847168,
|
| 7794 |
+
"learning_rate": 1.109e-05,
|
| 7795 |
+
"loss": 56.75,
|
| 7796 |
+
"step": 1110
|
| 7797 |
+
},
|
| 7798 |
+
{
|
| 7799 |
+
"epoch": 0.0001111,
|
| 7800 |
+
"grad_norm": 8.197669982910156,
|
| 7801 |
+
"learning_rate": 1.11e-05,
|
| 7802 |
+
"loss": 57.2812,
|
| 7803 |
+
"step": 1111
|
| 7804 |
+
},
|
| 7805 |
+
{
|
| 7806 |
+
"epoch": 0.0001112,
|
| 7807 |
+
"grad_norm": 8.191856384277344,
|
| 7808 |
+
"learning_rate": 1.111e-05,
|
| 7809 |
+
"loss": 56.9688,
|
| 7810 |
+
"step": 1112
|
| 7811 |
+
},
|
| 7812 |
+
{
|
| 7813 |
+
"epoch": 0.0001113,
|
| 7814 |
+
"grad_norm": 8.121025085449219,
|
| 7815 |
+
"learning_rate": 1.112e-05,
|
| 7816 |
+
"loss": 57.1875,
|
| 7817 |
+
"step": 1113
|
| 7818 |
+
},
|
| 7819 |
+
{
|
| 7820 |
+
"epoch": 0.0001114,
|
| 7821 |
+
"grad_norm": 8.034751892089844,
|
| 7822 |
+
"learning_rate": 1.113e-05,
|
| 7823 |
+
"loss": 57.2188,
|
| 7824 |
+
"step": 1114
|
| 7825 |
+
},
|
| 7826 |
+
{
|
| 7827 |
+
"epoch": 0.0001115,
|
| 7828 |
+
"grad_norm": 8.12697982788086,
|
| 7829 |
+
"learning_rate": 1.114e-05,
|
| 7830 |
+
"loss": 57.3125,
|
| 7831 |
+
"step": 1115
|
| 7832 |
+
},
|
| 7833 |
+
{
|
| 7834 |
+
"epoch": 0.0001116,
|
| 7835 |
+
"grad_norm": 7.979389667510986,
|
| 7836 |
+
"learning_rate": 1.115e-05,
|
| 7837 |
+
"loss": 57.1562,
|
| 7838 |
+
"step": 1116
|
| 7839 |
+
},
|
| 7840 |
+
{
|
| 7841 |
+
"epoch": 0.0001117,
|
| 7842 |
+
"grad_norm": 8.33296012878418,
|
| 7843 |
+
"learning_rate": 1.1160000000000002e-05,
|
| 7844 |
+
"loss": 56.9062,
|
| 7845 |
+
"step": 1117
|
| 7846 |
+
},
|
| 7847 |
+
{
|
| 7848 |
+
"epoch": 0.0001118,
|
| 7849 |
+
"grad_norm": 8.188544273376465,
|
| 7850 |
+
"learning_rate": 1.117e-05,
|
| 7851 |
+
"loss": 56.9062,
|
| 7852 |
+
"step": 1118
|
| 7853 |
+
},
|
| 7854 |
+
{
|
| 7855 |
+
"epoch": 0.0001119,
|
| 7856 |
+
"grad_norm": 7.855072975158691,
|
| 7857 |
+
"learning_rate": 1.118e-05,
|
| 7858 |
+
"loss": 57.1875,
|
| 7859 |
+
"step": 1119
|
| 7860 |
+
},
|
| 7861 |
+
{
|
| 7862 |
+
"epoch": 0.000112,
|
| 7863 |
+
"grad_norm": 8.365545272827148,
|
| 7864 |
+
"learning_rate": 1.119e-05,
|
| 7865 |
+
"loss": 56.4688,
|
| 7866 |
+
"step": 1120
|
| 7867 |
+
},
|
| 7868 |
+
{
|
| 7869 |
+
"epoch": 0.0001121,
|
| 7870 |
+
"grad_norm": 8.095410346984863,
|
| 7871 |
+
"learning_rate": 1.12e-05,
|
| 7872 |
+
"loss": 56.875,
|
| 7873 |
+
"step": 1121
|
| 7874 |
+
},
|
| 7875 |
+
{
|
| 7876 |
+
"epoch": 0.0001122,
|
| 7877 |
+
"grad_norm": 7.993049144744873,
|
| 7878 |
+
"learning_rate": 1.1210000000000001e-05,
|
| 7879 |
+
"loss": 57.2188,
|
| 7880 |
+
"step": 1122
|
| 7881 |
+
},
|
| 7882 |
+
{
|
| 7883 |
+
"epoch": 0.0001123,
|
| 7884 |
+
"grad_norm": 8.100505828857422,
|
| 7885 |
+
"learning_rate": 1.122e-05,
|
| 7886 |
+
"loss": 56.2812,
|
| 7887 |
+
"step": 1123
|
| 7888 |
+
},
|
| 7889 |
+
{
|
| 7890 |
+
"epoch": 0.0001124,
|
| 7891 |
+
"grad_norm": 8.246110916137695,
|
| 7892 |
+
"learning_rate": 1.1230000000000001e-05,
|
| 7893 |
+
"loss": 56.4375,
|
| 7894 |
+
"step": 1124
|
| 7895 |
+
},
|
| 7896 |
+
{
|
| 7897 |
+
"epoch": 0.0001125,
|
| 7898 |
+
"grad_norm": 8.12435531616211,
|
| 7899 |
+
"learning_rate": 1.124e-05,
|
| 7900 |
+
"loss": 56.5,
|
| 7901 |
+
"step": 1125
|
| 7902 |
+
},
|
| 7903 |
+
{
|
| 7904 |
+
"epoch": 0.0001126,
|
| 7905 |
+
"grad_norm": 7.739667892456055,
|
| 7906 |
+
"learning_rate": 1.1249999999999999e-05,
|
| 7907 |
+
"loss": 57.3438,
|
| 7908 |
+
"step": 1126
|
| 7909 |
+
},
|
| 7910 |
+
{
|
| 7911 |
+
"epoch": 0.0001127,
|
| 7912 |
+
"grad_norm": 8.352540016174316,
|
| 7913 |
+
"learning_rate": 1.126e-05,
|
| 7914 |
+
"loss": 56.1562,
|
| 7915 |
+
"step": 1127
|
| 7916 |
+
},
|
| 7917 |
+
{
|
| 7918 |
+
"epoch": 0.0001128,
|
| 7919 |
+
"grad_norm": 7.80290412902832,
|
| 7920 |
+
"learning_rate": 1.127e-05,
|
| 7921 |
+
"loss": 57.125,
|
| 7922 |
+
"step": 1128
|
| 7923 |
+
},
|
| 7924 |
+
{
|
| 7925 |
+
"epoch": 0.0001129,
|
| 7926 |
+
"grad_norm": 7.911022663116455,
|
| 7927 |
+
"learning_rate": 1.128e-05,
|
| 7928 |
+
"loss": 57.0625,
|
| 7929 |
+
"step": 1129
|
| 7930 |
+
},
|
| 7931 |
+
{
|
| 7932 |
+
"epoch": 0.000113,
|
| 7933 |
+
"grad_norm": 8.22894287109375,
|
| 7934 |
+
"learning_rate": 1.129e-05,
|
| 7935 |
+
"loss": 56.25,
|
| 7936 |
+
"step": 1130
|
| 7937 |
+
},
|
| 7938 |
+
{
|
| 7939 |
+
"epoch": 0.0001131,
|
| 7940 |
+
"grad_norm": 8.299407005310059,
|
| 7941 |
+
"learning_rate": 1.13e-05,
|
| 7942 |
+
"loss": 55.9688,
|
| 7943 |
+
"step": 1131
|
| 7944 |
+
},
|
| 7945 |
+
{
|
| 7946 |
+
"epoch": 0.0001132,
|
| 7947 |
+
"grad_norm": 8.011561393737793,
|
| 7948 |
+
"learning_rate": 1.131e-05,
|
| 7949 |
+
"loss": 56.3125,
|
| 7950 |
+
"step": 1132
|
| 7951 |
+
},
|
| 7952 |
+
{
|
| 7953 |
+
"epoch": 0.0001133,
|
| 7954 |
+
"grad_norm": 8.105046272277832,
|
| 7955 |
+
"learning_rate": 1.132e-05,
|
| 7956 |
+
"loss": 55.875,
|
| 7957 |
+
"step": 1133
|
| 7958 |
+
},
|
| 7959 |
+
{
|
| 7960 |
+
"epoch": 0.0001134,
|
| 7961 |
+
"grad_norm": 7.755193710327148,
|
| 7962 |
+
"learning_rate": 1.133e-05,
|
| 7963 |
+
"loss": 57.2188,
|
| 7964 |
+
"step": 1134
|
| 7965 |
+
},
|
| 7966 |
+
{
|
| 7967 |
+
"epoch": 0.0001135,
|
| 7968 |
+
"grad_norm": 7.714022636413574,
|
| 7969 |
+
"learning_rate": 1.134e-05,
|
| 7970 |
+
"loss": 56.8125,
|
| 7971 |
+
"step": 1135
|
| 7972 |
+
},
|
| 7973 |
+
{
|
| 7974 |
+
"epoch": 0.0001136,
|
| 7975 |
+
"grad_norm": 8.133477210998535,
|
| 7976 |
+
"learning_rate": 1.135e-05,
|
| 7977 |
+
"loss": 55.6875,
|
| 7978 |
+
"step": 1136
|
| 7979 |
+
},
|
| 7980 |
+
{
|
| 7981 |
+
"epoch": 0.0001137,
|
| 7982 |
+
"grad_norm": 8.036331176757812,
|
| 7983 |
+
"learning_rate": 1.136e-05,
|
| 7984 |
+
"loss": 56.0625,
|
| 7985 |
+
"step": 1137
|
| 7986 |
+
},
|
| 7987 |
+
{
|
| 7988 |
+
"epoch": 0.0001138,
|
| 7989 |
+
"grad_norm": 7.8458356857299805,
|
| 7990 |
+
"learning_rate": 1.137e-05,
|
| 7991 |
+
"loss": 56.375,
|
| 7992 |
+
"step": 1138
|
| 7993 |
+
},
|
| 7994 |
+
{
|
| 7995 |
+
"epoch": 0.0001139,
|
| 7996 |
+
"grad_norm": 8.083093643188477,
|
| 7997 |
+
"learning_rate": 1.1380000000000001e-05,
|
| 7998 |
+
"loss": 56.0625,
|
| 7999 |
+
"step": 1139
|
| 8000 |
+
},
|
| 8001 |
+
{
|
| 8002 |
+
"epoch": 0.000114,
|
| 8003 |
+
"grad_norm": 8.058805465698242,
|
| 8004 |
+
"learning_rate": 1.139e-05,
|
| 8005 |
+
"loss": 56.0,
|
| 8006 |
+
"step": 1140
|
| 8007 |
+
},
|
| 8008 |
+
{
|
| 8009 |
+
"epoch": 0.0001141,
|
| 8010 |
+
"grad_norm": 7.845074653625488,
|
| 8011 |
+
"learning_rate": 1.1400000000000001e-05,
|
| 8012 |
+
"loss": 56.625,
|
| 8013 |
+
"step": 1141
|
| 8014 |
+
},
|
| 8015 |
+
{
|
| 8016 |
+
"epoch": 0.0001142,
|
| 8017 |
+
"grad_norm": 7.745107173919678,
|
| 8018 |
+
"learning_rate": 1.1410000000000001e-05,
|
| 8019 |
+
"loss": 57.0,
|
| 8020 |
+
"step": 1142
|
| 8021 |
+
},
|
| 8022 |
+
{
|
| 8023 |
+
"epoch": 0.0001143,
|
| 8024 |
+
"grad_norm": 8.069292068481445,
|
| 8025 |
+
"learning_rate": 1.1419999999999999e-05,
|
| 8026 |
+
"loss": 55.7812,
|
| 8027 |
+
"step": 1143
|
| 8028 |
+
},
|
| 8029 |
+
{
|
| 8030 |
+
"epoch": 0.0001144,
|
| 8031 |
+
"grad_norm": 7.823101043701172,
|
| 8032 |
+
"learning_rate": 1.143e-05,
|
| 8033 |
+
"loss": 56.25,
|
| 8034 |
+
"step": 1144
|
| 8035 |
+
},
|
| 8036 |
+
{
|
| 8037 |
+
"epoch": 0.0001145,
|
| 8038 |
+
"grad_norm": 7.864339828491211,
|
| 8039 |
+
"learning_rate": 1.144e-05,
|
| 8040 |
+
"loss": 55.9375,
|
| 8041 |
+
"step": 1145
|
| 8042 |
+
},
|
| 8043 |
+
{
|
| 8044 |
+
"epoch": 0.0001146,
|
| 8045 |
+
"grad_norm": 7.925984859466553,
|
| 8046 |
+
"learning_rate": 1.145e-05,
|
| 8047 |
+
"loss": 56.4688,
|
| 8048 |
+
"step": 1146
|
| 8049 |
+
},
|
| 8050 |
+
{
|
| 8051 |
+
"epoch": 0.0001147,
|
| 8052 |
+
"grad_norm": 7.802701473236084,
|
| 8053 |
+
"learning_rate": 1.146e-05,
|
| 8054 |
+
"loss": 56.375,
|
| 8055 |
+
"step": 1147
|
| 8056 |
+
},
|
| 8057 |
+
{
|
| 8058 |
+
"epoch": 0.0001148,
|
| 8059 |
+
"grad_norm": 8.001948356628418,
|
| 8060 |
+
"learning_rate": 1.147e-05,
|
| 8061 |
+
"loss": 55.6875,
|
| 8062 |
+
"step": 1148
|
| 8063 |
+
},
|
| 8064 |
+
{
|
| 8065 |
+
"epoch": 0.0001149,
|
| 8066 |
+
"grad_norm": 8.064249038696289,
|
| 8067 |
+
"learning_rate": 1.148e-05,
|
| 8068 |
+
"loss": 55.5625,
|
| 8069 |
+
"step": 1149
|
| 8070 |
+
},
|
| 8071 |
+
{
|
| 8072 |
+
"epoch": 0.000115,
|
| 8073 |
+
"grad_norm": 8.19449520111084,
|
| 8074 |
+
"learning_rate": 1.149e-05,
|
| 8075 |
+
"loss": 55.4375,
|
| 8076 |
+
"step": 1150
|
| 8077 |
+
},
|
| 8078 |
+
{
|
| 8079 |
+
"epoch": 0.0001151,
|
| 8080 |
+
"grad_norm": 8.22554874420166,
|
| 8081 |
+
"learning_rate": 1.15e-05,
|
| 8082 |
+
"loss": 55.0312,
|
| 8083 |
+
"step": 1151
|
| 8084 |
+
},
|
| 8085 |
+
{
|
| 8086 |
+
"epoch": 0.0001152,
|
| 8087 |
+
"grad_norm": 7.867717742919922,
|
| 8088 |
+
"learning_rate": 1.151e-05,
|
| 8089 |
+
"loss": 55.8125,
|
| 8090 |
+
"step": 1152
|
| 8091 |
+
},
|
| 8092 |
+
{
|
| 8093 |
+
"epoch": 0.0001153,
|
| 8094 |
+
"grad_norm": 7.817559242248535,
|
| 8095 |
+
"learning_rate": 1.152e-05,
|
| 8096 |
+
"loss": 55.7188,
|
| 8097 |
+
"step": 1153
|
| 8098 |
+
},
|
| 8099 |
+
{
|
| 8100 |
+
"epoch": 0.0001154,
|
| 8101 |
+
"grad_norm": 7.744104385375977,
|
| 8102 |
+
"learning_rate": 1.153e-05,
|
| 8103 |
+
"loss": 55.6562,
|
| 8104 |
+
"step": 1154
|
| 8105 |
+
},
|
| 8106 |
+
{
|
| 8107 |
+
"epoch": 0.0001155,
|
| 8108 |
+
"grad_norm": 7.854978084564209,
|
| 8109 |
+
"learning_rate": 1.154e-05,
|
| 8110 |
+
"loss": 55.875,
|
| 8111 |
+
"step": 1155
|
| 8112 |
+
},
|
| 8113 |
+
{
|
| 8114 |
+
"epoch": 0.0001156,
|
| 8115 |
+
"grad_norm": 7.853444576263428,
|
| 8116 |
+
"learning_rate": 1.1550000000000001e-05,
|
| 8117 |
+
"loss": 55.75,
|
| 8118 |
+
"step": 1156
|
| 8119 |
+
},
|
| 8120 |
+
{
|
| 8121 |
+
"epoch": 0.0001157,
|
| 8122 |
+
"grad_norm": 7.83394718170166,
|
| 8123 |
+
"learning_rate": 1.156e-05,
|
| 8124 |
+
"loss": 55.8438,
|
| 8125 |
+
"step": 1157
|
| 8126 |
+
},
|
| 8127 |
+
{
|
| 8128 |
+
"epoch": 0.0001158,
|
| 8129 |
+
"grad_norm": 7.851531505584717,
|
| 8130 |
+
"learning_rate": 1.1570000000000001e-05,
|
| 8131 |
+
"loss": 55.4375,
|
| 8132 |
+
"step": 1158
|
| 8133 |
+
},
|
| 8134 |
+
{
|
| 8135 |
+
"epoch": 0.0001159,
|
| 8136 |
+
"grad_norm": 7.822463035583496,
|
| 8137 |
+
"learning_rate": 1.1580000000000001e-05,
|
| 8138 |
+
"loss": 55.0938,
|
| 8139 |
+
"step": 1159
|
| 8140 |
+
},
|
| 8141 |
+
{
|
| 8142 |
+
"epoch": 0.000116,
|
| 8143 |
+
"grad_norm": 8.192028045654297,
|
| 8144 |
+
"learning_rate": 1.1589999999999999e-05,
|
| 8145 |
+
"loss": 54.6875,
|
| 8146 |
+
"step": 1160
|
| 8147 |
+
},
|
| 8148 |
+
{
|
| 8149 |
+
"epoch": 0.0001161,
|
| 8150 |
+
"grad_norm": 7.832354545593262,
|
| 8151 |
+
"learning_rate": 1.16e-05,
|
| 8152 |
+
"loss": 55.0938,
|
| 8153 |
+
"step": 1161
|
| 8154 |
+
},
|
| 8155 |
+
{
|
| 8156 |
+
"epoch": 0.0001162,
|
| 8157 |
+
"grad_norm": 7.876185894012451,
|
| 8158 |
+
"learning_rate": 1.161e-05,
|
| 8159 |
+
"loss": 54.875,
|
| 8160 |
+
"step": 1162
|
| 8161 |
+
},
|
| 8162 |
+
{
|
| 8163 |
+
"epoch": 0.0001163,
|
| 8164 |
+
"grad_norm": 7.61892557144165,
|
| 8165 |
+
"learning_rate": 1.162e-05,
|
| 8166 |
+
"loss": 56.3125,
|
| 8167 |
+
"step": 1163
|
| 8168 |
+
},
|
| 8169 |
+
{
|
| 8170 |
+
"epoch": 0.0001164,
|
| 8171 |
+
"grad_norm": 7.596429824829102,
|
| 8172 |
+
"learning_rate": 1.163e-05,
|
| 8173 |
+
"loss": 55.375,
|
| 8174 |
+
"step": 1164
|
| 8175 |
+
},
|
| 8176 |
+
{
|
| 8177 |
+
"epoch": 0.0001165,
|
| 8178 |
+
"grad_norm": 7.708346843719482,
|
| 8179 |
+
"learning_rate": 1.164e-05,
|
| 8180 |
+
"loss": 55.5312,
|
| 8181 |
+
"step": 1165
|
| 8182 |
+
},
|
| 8183 |
+
{
|
| 8184 |
+
"epoch": 0.0001166,
|
| 8185 |
+
"grad_norm": 7.724869251251221,
|
| 8186 |
+
"learning_rate": 1.165e-05,
|
| 8187 |
+
"loss": 55.1562,
|
| 8188 |
+
"step": 1166
|
| 8189 |
+
},
|
| 8190 |
+
{
|
| 8191 |
+
"epoch": 0.0001167,
|
| 8192 |
+
"grad_norm": 7.922326564788818,
|
| 8193 |
+
"learning_rate": 1.166e-05,
|
| 8194 |
+
"loss": 55.125,
|
| 8195 |
+
"step": 1167
|
| 8196 |
+
},
|
| 8197 |
+
{
|
| 8198 |
+
"epoch": 0.0001168,
|
| 8199 |
+
"grad_norm": 7.799662113189697,
|
| 8200 |
+
"learning_rate": 1.167e-05,
|
| 8201 |
+
"loss": 55.0625,
|
| 8202 |
+
"step": 1168
|
| 8203 |
+
},
|
| 8204 |
+
{
|
| 8205 |
+
"epoch": 0.0001169,
|
| 8206 |
+
"grad_norm": 7.678478717803955,
|
| 8207 |
+
"learning_rate": 1.168e-05,
|
| 8208 |
+
"loss": 55.1562,
|
| 8209 |
+
"step": 1169
|
| 8210 |
+
},
|
| 8211 |
+
{
|
| 8212 |
+
"epoch": 0.000117,
|
| 8213 |
+
"grad_norm": 7.4621901512146,
|
| 8214 |
+
"learning_rate": 1.169e-05,
|
| 8215 |
+
"loss": 55.9375,
|
| 8216 |
+
"step": 1170
|
| 8217 |
+
},
|
| 8218 |
+
{
|
| 8219 |
+
"epoch": 0.0001171,
|
| 8220 |
+
"grad_norm": 7.68449068069458,
|
| 8221 |
+
"learning_rate": 1.17e-05,
|
| 8222 |
+
"loss": 55.2812,
|
| 8223 |
+
"step": 1171
|
| 8224 |
+
},
|
| 8225 |
+
{
|
| 8226 |
+
"epoch": 0.0001172,
|
| 8227 |
+
"grad_norm": 7.555746078491211,
|
| 8228 |
+
"learning_rate": 1.171e-05,
|
| 8229 |
+
"loss": 55.2188,
|
| 8230 |
+
"step": 1172
|
| 8231 |
+
},
|
| 8232 |
+
{
|
| 8233 |
+
"epoch": 0.0001173,
|
| 8234 |
+
"grad_norm": 7.522375583648682,
|
| 8235 |
+
"learning_rate": 1.1720000000000001e-05,
|
| 8236 |
+
"loss": 55.6562,
|
| 8237 |
+
"step": 1173
|
| 8238 |
+
},
|
| 8239 |
+
{
|
| 8240 |
+
"epoch": 0.0001174,
|
| 8241 |
+
"grad_norm": 7.526357650756836,
|
| 8242 |
+
"learning_rate": 1.1730000000000001e-05,
|
| 8243 |
+
"loss": 54.9688,
|
| 8244 |
+
"step": 1174
|
| 8245 |
+
},
|
| 8246 |
+
{
|
| 8247 |
+
"epoch": 0.0001175,
|
| 8248 |
+
"grad_norm": 7.813880443572998,
|
| 8249 |
+
"learning_rate": 1.174e-05,
|
| 8250 |
+
"loss": 54.5,
|
| 8251 |
+
"step": 1175
|
| 8252 |
+
},
|
| 8253 |
+
{
|
| 8254 |
+
"epoch": 0.0001176,
|
| 8255 |
+
"grad_norm": 7.758258819580078,
|
| 8256 |
+
"learning_rate": 1.175e-05,
|
| 8257 |
+
"loss": 54.5,
|
| 8258 |
+
"step": 1176
|
| 8259 |
+
},
|
| 8260 |
+
{
|
| 8261 |
+
"epoch": 0.0001177,
|
| 8262 |
+
"grad_norm": 7.648372173309326,
|
| 8263 |
+
"learning_rate": 1.176e-05,
|
| 8264 |
+
"loss": 54.9375,
|
| 8265 |
+
"step": 1177
|
| 8266 |
+
},
|
| 8267 |
+
{
|
| 8268 |
+
"epoch": 0.0001178,
|
| 8269 |
+
"grad_norm": 7.699697971343994,
|
| 8270 |
+
"learning_rate": 1.177e-05,
|
| 8271 |
+
"loss": 54.3125,
|
| 8272 |
+
"step": 1178
|
| 8273 |
+
},
|
| 8274 |
+
{
|
| 8275 |
+
"epoch": 0.0001179,
|
| 8276 |
+
"grad_norm": 7.569566249847412,
|
| 8277 |
+
"learning_rate": 1.178e-05,
|
| 8278 |
+
"loss": 55.1562,
|
| 8279 |
+
"step": 1179
|
| 8280 |
+
},
|
| 8281 |
+
{
|
| 8282 |
+
"epoch": 0.000118,
|
| 8283 |
+
"grad_norm": 7.436944484710693,
|
| 8284 |
+
"learning_rate": 1.179e-05,
|
| 8285 |
+
"loss": 55.3438,
|
| 8286 |
+
"step": 1180
|
| 8287 |
+
},
|
| 8288 |
+
{
|
| 8289 |
+
"epoch": 0.0001181,
|
| 8290 |
+
"grad_norm": 7.739448070526123,
|
| 8291 |
+
"learning_rate": 1.1799999999999999e-05,
|
| 8292 |
+
"loss": 54.2188,
|
| 8293 |
+
"step": 1181
|
| 8294 |
+
},
|
| 8295 |
+
{
|
| 8296 |
+
"epoch": 0.0001182,
|
| 8297 |
+
"grad_norm": 7.439329624176025,
|
| 8298 |
+
"learning_rate": 1.1809999999999999e-05,
|
| 8299 |
+
"loss": 55.8125,
|
| 8300 |
+
"step": 1182
|
| 8301 |
+
},
|
| 8302 |
+
{
|
| 8303 |
+
"epoch": 0.0001183,
|
| 8304 |
+
"grad_norm": 8.071717262268066,
|
| 8305 |
+
"learning_rate": 1.182e-05,
|
| 8306 |
+
"loss": 53.7812,
|
| 8307 |
+
"step": 1183
|
| 8308 |
+
},
|
| 8309 |
+
{
|
| 8310 |
+
"epoch": 0.0001184,
|
| 8311 |
+
"grad_norm": 7.629164218902588,
|
| 8312 |
+
"learning_rate": 1.183e-05,
|
| 8313 |
+
"loss": 54.25,
|
| 8314 |
+
"step": 1184
|
| 8315 |
+
},
|
| 8316 |
+
{
|
| 8317 |
+
"epoch": 0.0001185,
|
| 8318 |
+
"grad_norm": 7.748568534851074,
|
| 8319 |
+
"learning_rate": 1.184e-05,
|
| 8320 |
+
"loss": 54.2812,
|
| 8321 |
+
"step": 1185
|
| 8322 |
+
},
|
| 8323 |
+
{
|
| 8324 |
+
"epoch": 0.0001186,
|
| 8325 |
+
"grad_norm": 7.587509632110596,
|
| 8326 |
+
"learning_rate": 1.1850000000000002e-05,
|
| 8327 |
+
"loss": 54.5625,
|
| 8328 |
+
"step": 1186
|
| 8329 |
+
},
|
| 8330 |
+
{
|
| 8331 |
+
"epoch": 0.0001187,
|
| 8332 |
+
"grad_norm": 7.9441351890563965,
|
| 8333 |
+
"learning_rate": 1.186e-05,
|
| 8334 |
+
"loss": 53.9375,
|
| 8335 |
+
"step": 1187
|
| 8336 |
+
},
|
| 8337 |
+
{
|
| 8338 |
+
"epoch": 0.0001188,
|
| 8339 |
+
"grad_norm": 7.724300861358643,
|
| 8340 |
+
"learning_rate": 1.187e-05,
|
| 8341 |
+
"loss": 54.1562,
|
| 8342 |
+
"step": 1188
|
| 8343 |
+
},
|
| 8344 |
+
{
|
| 8345 |
+
"epoch": 0.0001189,
|
| 8346 |
+
"grad_norm": 7.837823867797852,
|
| 8347 |
+
"learning_rate": 1.188e-05,
|
| 8348 |
+
"loss": 53.5,
|
| 8349 |
+
"step": 1189
|
| 8350 |
+
},
|
| 8351 |
+
{
|
| 8352 |
+
"epoch": 0.000119,
|
| 8353 |
+
"grad_norm": 7.670633316040039,
|
| 8354 |
+
"learning_rate": 1.1890000000000001e-05,
|
| 8355 |
+
"loss": 53.9062,
|
| 8356 |
+
"step": 1190
|
| 8357 |
+
},
|
| 8358 |
+
{
|
| 8359 |
+
"epoch": 0.0001191,
|
| 8360 |
+
"grad_norm": 7.673746585845947,
|
| 8361 |
+
"learning_rate": 1.1900000000000001e-05,
|
| 8362 |
+
"loss": 54.0938,
|
| 8363 |
+
"step": 1191
|
| 8364 |
+
},
|
| 8365 |
+
{
|
| 8366 |
+
"epoch": 0.0001192,
|
| 8367 |
+
"grad_norm": 7.649930000305176,
|
| 8368 |
+
"learning_rate": 1.191e-05,
|
| 8369 |
+
"loss": 54.625,
|
| 8370 |
+
"step": 1192
|
| 8371 |
+
},
|
| 8372 |
+
{
|
| 8373 |
+
"epoch": 0.0001193,
|
| 8374 |
+
"grad_norm": 7.45327615737915,
|
| 8375 |
+
"learning_rate": 1.192e-05,
|
| 8376 |
+
"loss": 54.4375,
|
| 8377 |
+
"step": 1193
|
| 8378 |
+
},
|
| 8379 |
+
{
|
| 8380 |
+
"epoch": 0.0001194,
|
| 8381 |
+
"grad_norm": 7.5371856689453125,
|
| 8382 |
+
"learning_rate": 1.193e-05,
|
| 8383 |
+
"loss": 53.7188,
|
| 8384 |
+
"step": 1194
|
| 8385 |
+
},
|
| 8386 |
+
{
|
| 8387 |
+
"epoch": 0.0001195,
|
| 8388 |
+
"grad_norm": 7.465251922607422,
|
| 8389 |
+
"learning_rate": 1.1940000000000001e-05,
|
| 8390 |
+
"loss": 54.5,
|
| 8391 |
+
"step": 1195
|
| 8392 |
+
},
|
| 8393 |
+
{
|
| 8394 |
+
"epoch": 0.0001196,
|
| 8395 |
+
"grad_norm": 7.773534774780273,
|
| 8396 |
+
"learning_rate": 1.195e-05,
|
| 8397 |
+
"loss": 53.4375,
|
| 8398 |
+
"step": 1196
|
| 8399 |
+
},
|
| 8400 |
+
{
|
| 8401 |
+
"epoch": 0.0001197,
|
| 8402 |
+
"grad_norm": 7.400661945343018,
|
| 8403 |
+
"learning_rate": 1.196e-05,
|
| 8404 |
+
"loss": 54.25,
|
| 8405 |
+
"step": 1197
|
| 8406 |
+
},
|
| 8407 |
+
{
|
| 8408 |
+
"epoch": 0.0001198,
|
| 8409 |
+
"grad_norm": 7.420455455780029,
|
| 8410 |
+
"learning_rate": 1.1969999999999999e-05,
|
| 8411 |
+
"loss": 54.5938,
|
| 8412 |
+
"step": 1198
|
| 8413 |
+
},
|
| 8414 |
+
{
|
| 8415 |
+
"epoch": 0.0001199,
|
| 8416 |
+
"grad_norm": 7.698644161224365,
|
| 8417 |
+
"learning_rate": 1.1979999999999999e-05,
|
| 8418 |
+
"loss": 53.8125,
|
| 8419 |
+
"step": 1199
|
| 8420 |
+
},
|
| 8421 |
+
{
|
| 8422 |
+
"epoch": 0.00012,
|
| 8423 |
+
"grad_norm": 7.344296932220459,
|
| 8424 |
+
"learning_rate": 1.199e-05,
|
| 8425 |
+
"loss": 54.4688,
|
| 8426 |
+
"step": 1200
|
| 8427 |
+
},
|
| 8428 |
+
{
|
| 8429 |
+
"epoch": 0.0001201,
|
| 8430 |
+
"grad_norm": 7.562631130218506,
|
| 8431 |
+
"learning_rate": 1.2e-05,
|
| 8432 |
+
"loss": 53.8125,
|
| 8433 |
+
"step": 1201
|
| 8434 |
+
},
|
| 8435 |
+
{
|
| 8436 |
+
"epoch": 0.0001202,
|
| 8437 |
+
"grad_norm": 7.408926486968994,
|
| 8438 |
+
"learning_rate": 1.201e-05,
|
| 8439 |
+
"loss": 53.8125,
|
| 8440 |
+
"step": 1202
|
| 8441 |
+
},
|
| 8442 |
+
{
|
| 8443 |
+
"epoch": 0.0001203,
|
| 8444 |
+
"grad_norm": 7.729759216308594,
|
| 8445 |
+
"learning_rate": 1.2020000000000002e-05,
|
| 8446 |
+
"loss": 53.375,
|
| 8447 |
+
"step": 1203
|
| 8448 |
+
},
|
| 8449 |
+
{
|
| 8450 |
+
"epoch": 0.0001204,
|
| 8451 |
+
"grad_norm": 7.490784168243408,
|
| 8452 |
+
"learning_rate": 1.203e-05,
|
| 8453 |
+
"loss": 53.5938,
|
| 8454 |
+
"step": 1204
|
| 8455 |
+
},
|
| 8456 |
+
{
|
| 8457 |
+
"epoch": 0.0001205,
|
| 8458 |
+
"grad_norm": 7.493447780609131,
|
| 8459 |
+
"learning_rate": 1.204e-05,
|
| 8460 |
+
"loss": 53.7188,
|
| 8461 |
+
"step": 1205
|
| 8462 |
+
},
|
| 8463 |
+
{
|
| 8464 |
+
"epoch": 0.0001206,
|
| 8465 |
+
"grad_norm": 7.282375812530518,
|
| 8466 |
+
"learning_rate": 1.205e-05,
|
| 8467 |
+
"loss": 54.3438,
|
| 8468 |
+
"step": 1206
|
| 8469 |
+
},
|
| 8470 |
+
{
|
| 8471 |
+
"epoch": 0.0001207,
|
| 8472 |
+
"grad_norm": 7.591827392578125,
|
| 8473 |
+
"learning_rate": 1.2060000000000001e-05,
|
| 8474 |
+
"loss": 53.5,
|
| 8475 |
+
"step": 1207
|
| 8476 |
+
},
|
| 8477 |
+
{
|
| 8478 |
+
"epoch": 0.0001208,
|
| 8479 |
+
"grad_norm": 7.682456016540527,
|
| 8480 |
+
"learning_rate": 1.2070000000000001e-05,
|
| 8481 |
+
"loss": 53.7812,
|
| 8482 |
+
"step": 1208
|
| 8483 |
+
},
|
| 8484 |
+
{
|
| 8485 |
+
"epoch": 0.0001209,
|
| 8486 |
+
"grad_norm": 7.2463202476501465,
|
| 8487 |
+
"learning_rate": 1.208e-05,
|
| 8488 |
+
"loss": 53.8125,
|
| 8489 |
+
"step": 1209
|
| 8490 |
+
},
|
| 8491 |
+
{
|
| 8492 |
+
"epoch": 0.000121,
|
| 8493 |
+
"grad_norm": 7.508241653442383,
|
| 8494 |
+
"learning_rate": 1.209e-05,
|
| 8495 |
+
"loss": 53.3125,
|
| 8496 |
+
"step": 1210
|
| 8497 |
+
},
|
| 8498 |
+
{
|
| 8499 |
+
"epoch": 0.0001211,
|
| 8500 |
+
"grad_norm": 7.579042911529541,
|
| 8501 |
+
"learning_rate": 1.21e-05,
|
| 8502 |
+
"loss": 52.8125,
|
| 8503 |
+
"step": 1211
|
| 8504 |
+
},
|
| 8505 |
+
{
|
| 8506 |
+
"epoch": 0.0001212,
|
| 8507 |
+
"grad_norm": 7.6091084480285645,
|
| 8508 |
+
"learning_rate": 1.2110000000000001e-05,
|
| 8509 |
+
"loss": 53.1562,
|
| 8510 |
+
"step": 1212
|
| 8511 |
+
},
|
| 8512 |
+
{
|
| 8513 |
+
"epoch": 0.0001213,
|
| 8514 |
+
"grad_norm": 7.4466657638549805,
|
| 8515 |
+
"learning_rate": 1.2120000000000001e-05,
|
| 8516 |
+
"loss": 53.2812,
|
| 8517 |
+
"step": 1213
|
| 8518 |
+
},
|
| 8519 |
+
{
|
| 8520 |
+
"epoch": 0.0001214,
|
| 8521 |
+
"grad_norm": 7.076037406921387,
|
| 8522 |
+
"learning_rate": 1.213e-05,
|
| 8523 |
+
"loss": 54.3438,
|
| 8524 |
+
"step": 1214
|
| 8525 |
+
},
|
| 8526 |
+
{
|
| 8527 |
+
"epoch": 0.0001215,
|
| 8528 |
+
"grad_norm": 7.500237941741943,
|
| 8529 |
+
"learning_rate": 1.2139999999999999e-05,
|
| 8530 |
+
"loss": 53.0,
|
| 8531 |
+
"step": 1215
|
| 8532 |
+
},
|
| 8533 |
+
{
|
| 8534 |
+
"epoch": 0.0001216,
|
| 8535 |
+
"grad_norm": 7.2276997566223145,
|
| 8536 |
+
"learning_rate": 1.2149999999999999e-05,
|
| 8537 |
+
"loss": 53.8125,
|
| 8538 |
+
"step": 1216
|
| 8539 |
+
},
|
| 8540 |
+
{
|
| 8541 |
+
"epoch": 0.0001217,
|
| 8542 |
+
"grad_norm": 7.537385940551758,
|
| 8543 |
+
"learning_rate": 1.216e-05,
|
| 8544 |
+
"loss": 52.7812,
|
| 8545 |
+
"step": 1217
|
| 8546 |
+
},
|
| 8547 |
+
{
|
| 8548 |
+
"epoch": 0.0001218,
|
| 8549 |
+
"grad_norm": 7.664484024047852,
|
| 8550 |
+
"learning_rate": 1.217e-05,
|
| 8551 |
+
"loss": 52.7812,
|
| 8552 |
+
"step": 1218
|
| 8553 |
+
},
|
| 8554 |
+
{
|
| 8555 |
+
"epoch": 0.0001219,
|
| 8556 |
+
"grad_norm": 7.303295612335205,
|
| 8557 |
+
"learning_rate": 1.218e-05,
|
| 8558 |
+
"loss": 53.7188,
|
| 8559 |
+
"step": 1219
|
| 8560 |
+
},
|
| 8561 |
+
{
|
| 8562 |
+
"epoch": 0.000122,
|
| 8563 |
+
"grad_norm": 7.290215492248535,
|
| 8564 |
+
"learning_rate": 1.2190000000000002e-05,
|
| 8565 |
+
"loss": 53.7188,
|
| 8566 |
+
"step": 1220
|
| 8567 |
+
},
|
| 8568 |
+
{
|
| 8569 |
+
"epoch": 0.0001221,
|
| 8570 |
+
"grad_norm": 7.22444486618042,
|
| 8571 |
+
"learning_rate": 1.22e-05,
|
| 8572 |
+
"loss": 53.5,
|
| 8573 |
+
"step": 1221
|
| 8574 |
+
},
|
| 8575 |
+
{
|
| 8576 |
+
"epoch": 0.0001222,
|
| 8577 |
+
"grad_norm": 7.554375648498535,
|
| 8578 |
+
"learning_rate": 1.221e-05,
|
| 8579 |
+
"loss": 52.5625,
|
| 8580 |
+
"step": 1222
|
| 8581 |
+
},
|
| 8582 |
+
{
|
| 8583 |
+
"epoch": 0.0001223,
|
| 8584 |
+
"grad_norm": 7.6661787033081055,
|
| 8585 |
+
"learning_rate": 1.222e-05,
|
| 8586 |
+
"loss": 52.6562,
|
| 8587 |
+
"step": 1223
|
| 8588 |
+
},
|
| 8589 |
+
{
|
| 8590 |
+
"epoch": 0.0001224,
|
| 8591 |
+
"grad_norm": 7.28986120223999,
|
| 8592 |
+
"learning_rate": 1.2230000000000001e-05,
|
| 8593 |
+
"loss": 53.375,
|
| 8594 |
+
"step": 1224
|
| 8595 |
+
},
|
| 8596 |
+
{
|
| 8597 |
+
"epoch": 0.0001225,
|
| 8598 |
+
"grad_norm": 7.4713215827941895,
|
| 8599 |
+
"learning_rate": 1.2240000000000001e-05,
|
| 8600 |
+
"loss": 53.375,
|
| 8601 |
+
"step": 1225
|
| 8602 |
+
},
|
| 8603 |
+
{
|
| 8604 |
+
"epoch": 0.0001226,
|
| 8605 |
+
"grad_norm": 7.248449325561523,
|
| 8606 |
+
"learning_rate": 1.225e-05,
|
| 8607 |
+
"loss": 53.0625,
|
| 8608 |
+
"step": 1226
|
| 8609 |
+
},
|
| 8610 |
+
{
|
| 8611 |
+
"epoch": 0.0001227,
|
| 8612 |
+
"grad_norm": 7.284693717956543,
|
| 8613 |
+
"learning_rate": 1.226e-05,
|
| 8614 |
+
"loss": 53.3125,
|
| 8615 |
+
"step": 1227
|
| 8616 |
+
},
|
| 8617 |
+
{
|
| 8618 |
+
"epoch": 0.0001228,
|
| 8619 |
+
"grad_norm": 7.41956901550293,
|
| 8620 |
+
"learning_rate": 1.227e-05,
|
| 8621 |
+
"loss": 52.5625,
|
| 8622 |
+
"step": 1228
|
| 8623 |
+
},
|
| 8624 |
+
{
|
| 8625 |
+
"epoch": 0.0001229,
|
| 8626 |
+
"grad_norm": 7.39863920211792,
|
| 8627 |
+
"learning_rate": 1.2280000000000001e-05,
|
| 8628 |
+
"loss": 53.0312,
|
| 8629 |
+
"step": 1229
|
| 8630 |
+
},
|
| 8631 |
+
{
|
| 8632 |
+
"epoch": 0.000123,
|
| 8633 |
+
"grad_norm": 7.424099922180176,
|
| 8634 |
+
"learning_rate": 1.2290000000000001e-05,
|
| 8635 |
+
"loss": 53.25,
|
| 8636 |
+
"step": 1230
|
| 8637 |
+
},
|
| 8638 |
+
{
|
| 8639 |
+
"epoch": 0.0001231,
|
| 8640 |
+
"grad_norm": 7.63134241104126,
|
| 8641 |
+
"learning_rate": 1.23e-05,
|
| 8642 |
+
"loss": 52.3125,
|
| 8643 |
+
"step": 1231
|
| 8644 |
+
},
|
| 8645 |
+
{
|
| 8646 |
+
"epoch": 0.0001232,
|
| 8647 |
+
"grad_norm": 7.46926212310791,
|
| 8648 |
+
"learning_rate": 1.2309999999999999e-05,
|
| 8649 |
+
"loss": 52.9688,
|
| 8650 |
+
"step": 1232
|
| 8651 |
+
},
|
| 8652 |
+
{
|
| 8653 |
+
"epoch": 0.0001233,
|
| 8654 |
+
"grad_norm": 7.465059280395508,
|
| 8655 |
+
"learning_rate": 1.2319999999999999e-05,
|
| 8656 |
+
"loss": 52.0,
|
| 8657 |
+
"step": 1233
|
| 8658 |
+
},
|
| 8659 |
+
{
|
| 8660 |
+
"epoch": 0.0001234,
|
| 8661 |
+
"grad_norm": 7.630496501922607,
|
| 8662 |
+
"learning_rate": 1.233e-05,
|
| 8663 |
+
"loss": 52.125,
|
| 8664 |
+
"step": 1234
|
| 8665 |
+
},
|
| 8666 |
+
{
|
| 8667 |
+
"epoch": 0.0001235,
|
| 8668 |
+
"grad_norm": 7.242209434509277,
|
| 8669 |
+
"learning_rate": 1.234e-05,
|
| 8670 |
+
"loss": 52.9062,
|
| 8671 |
+
"step": 1235
|
| 8672 |
+
},
|
| 8673 |
+
{
|
| 8674 |
+
"epoch": 0.0001236,
|
| 8675 |
+
"grad_norm": 7.2926740646362305,
|
| 8676 |
+
"learning_rate": 1.235e-05,
|
| 8677 |
+
"loss": 53.6562,
|
| 8678 |
+
"step": 1236
|
| 8679 |
+
},
|
| 8680 |
+
{
|
| 8681 |
+
"epoch": 0.0001237,
|
| 8682 |
+
"grad_norm": 7.212370872497559,
|
| 8683 |
+
"learning_rate": 1.2360000000000002e-05,
|
| 8684 |
+
"loss": 52.8125,
|
| 8685 |
+
"step": 1237
|
| 8686 |
+
},
|
| 8687 |
+
{
|
| 8688 |
+
"epoch": 0.0001238,
|
| 8689 |
+
"grad_norm": 7.0620598793029785,
|
| 8690 |
+
"learning_rate": 1.237e-05,
|
| 8691 |
+
"loss": 53.2188,
|
| 8692 |
+
"step": 1238
|
| 8693 |
+
},
|
| 8694 |
+
{
|
| 8695 |
+
"epoch": 0.0001239,
|
| 8696 |
+
"grad_norm": 7.333443641662598,
|
| 8697 |
+
"learning_rate": 1.238e-05,
|
| 8698 |
+
"loss": 52.8125,
|
| 8699 |
+
"step": 1239
|
| 8700 |
+
},
|
| 8701 |
+
{
|
| 8702 |
+
"epoch": 0.000124,
|
| 8703 |
+
"grad_norm": 7.3336873054504395,
|
| 8704 |
+
"learning_rate": 1.239e-05,
|
| 8705 |
+
"loss": 52.2188,
|
| 8706 |
+
"step": 1240
|
| 8707 |
+
},
|
| 8708 |
+
{
|
| 8709 |
+
"epoch": 0.0001241,
|
| 8710 |
+
"grad_norm": 7.147743225097656,
|
| 8711 |
+
"learning_rate": 1.24e-05,
|
| 8712 |
+
"loss": 52.5,
|
| 8713 |
+
"step": 1241
|
| 8714 |
+
},
|
| 8715 |
+
{
|
| 8716 |
+
"epoch": 0.0001242,
|
| 8717 |
+
"grad_norm": 7.3426618576049805,
|
| 8718 |
+
"learning_rate": 1.2410000000000001e-05,
|
| 8719 |
+
"loss": 52.3438,
|
| 8720 |
+
"step": 1242
|
| 8721 |
+
},
|
| 8722 |
+
{
|
| 8723 |
+
"epoch": 0.0001243,
|
| 8724 |
+
"grad_norm": 7.250710487365723,
|
| 8725 |
+
"learning_rate": 1.242e-05,
|
| 8726 |
+
"loss": 52.1562,
|
| 8727 |
+
"step": 1243
|
| 8728 |
+
},
|
| 8729 |
+
{
|
| 8730 |
+
"epoch": 0.0001244,
|
| 8731 |
+
"grad_norm": 7.301087379455566,
|
| 8732 |
+
"learning_rate": 1.243e-05,
|
| 8733 |
+
"loss": 52.625,
|
| 8734 |
+
"step": 1244
|
| 8735 |
+
},
|
| 8736 |
+
{
|
| 8737 |
+
"epoch": 0.0001245,
|
| 8738 |
+
"grad_norm": 7.3164191246032715,
|
| 8739 |
+
"learning_rate": 1.244e-05,
|
| 8740 |
+
"loss": 52.0938,
|
| 8741 |
+
"step": 1245
|
| 8742 |
+
},
|
| 8743 |
+
{
|
| 8744 |
+
"epoch": 0.0001246,
|
| 8745 |
+
"grad_norm": 7.23098087310791,
|
| 8746 |
+
"learning_rate": 1.2450000000000001e-05,
|
| 8747 |
+
"loss": 52.7812,
|
| 8748 |
+
"step": 1246
|
| 8749 |
+
},
|
| 8750 |
+
{
|
| 8751 |
+
"epoch": 0.0001247,
|
| 8752 |
+
"grad_norm": 7.254575252532959,
|
| 8753 |
+
"learning_rate": 1.2460000000000001e-05,
|
| 8754 |
+
"loss": 52.25,
|
| 8755 |
+
"step": 1247
|
| 8756 |
+
},
|
| 8757 |
+
{
|
| 8758 |
+
"epoch": 0.0001248,
|
| 8759 |
+
"grad_norm": 7.3596625328063965,
|
| 8760 |
+
"learning_rate": 1.2470000000000001e-05,
|
| 8761 |
+
"loss": 51.7812,
|
| 8762 |
+
"step": 1248
|
| 8763 |
+
},
|
| 8764 |
+
{
|
| 8765 |
+
"epoch": 0.0001249,
|
| 8766 |
+
"grad_norm": 7.4117536544799805,
|
| 8767 |
+
"learning_rate": 1.2479999999999999e-05,
|
| 8768 |
+
"loss": 51.4688,
|
| 8769 |
+
"step": 1249
|
| 8770 |
+
},
|
| 8771 |
+
{
|
| 8772 |
+
"epoch": 0.000125,
|
| 8773 |
+
"grad_norm": 7.203137397766113,
|
| 8774 |
+
"learning_rate": 1.2489999999999999e-05,
|
| 8775 |
+
"loss": 52.1562,
|
| 8776 |
+
"step": 1250
|
| 8777 |
+
},
|
| 8778 |
+
{
|
| 8779 |
+
"epoch": 0.0001251,
|
| 8780 |
+
"grad_norm": 7.224808692932129,
|
| 8781 |
+
"learning_rate": 1.25e-05,
|
| 8782 |
+
"loss": 52.1562,
|
| 8783 |
+
"step": 1251
|
| 8784 |
+
},
|
| 8785 |
+
{
|
| 8786 |
+
"epoch": 0.0001252,
|
| 8787 |
+
"grad_norm": 7.343545913696289,
|
| 8788 |
+
"learning_rate": 1.251e-05,
|
| 8789 |
+
"loss": 51.75,
|
| 8790 |
+
"step": 1252
|
| 8791 |
+
},
|
| 8792 |
+
{
|
| 8793 |
+
"epoch": 0.0001253,
|
| 8794 |
+
"grad_norm": 7.2418365478515625,
|
| 8795 |
+
"learning_rate": 1.252e-05,
|
| 8796 |
+
"loss": 51.6562,
|
| 8797 |
+
"step": 1253
|
| 8798 |
+
},
|
| 8799 |
+
{
|
| 8800 |
+
"epoch": 0.0001254,
|
| 8801 |
+
"grad_norm": 7.007531642913818,
|
| 8802 |
+
"learning_rate": 1.2530000000000002e-05,
|
| 8803 |
+
"loss": 52.4375,
|
| 8804 |
+
"step": 1254
|
| 8805 |
+
},
|
| 8806 |
+
{
|
| 8807 |
+
"epoch": 0.0001255,
|
| 8808 |
+
"grad_norm": 6.968234539031982,
|
| 8809 |
+
"learning_rate": 1.2539999999999999e-05,
|
| 8810 |
+
"loss": 52.5312,
|
| 8811 |
+
"step": 1255
|
| 8812 |
+
},
|
| 8813 |
+
{
|
| 8814 |
+
"epoch": 0.0001256,
|
| 8815 |
+
"grad_norm": 7.113062381744385,
|
| 8816 |
+
"learning_rate": 1.255e-05,
|
| 8817 |
+
"loss": 52.375,
|
| 8818 |
+
"step": 1256
|
| 8819 |
+
},
|
| 8820 |
+
{
|
| 8821 |
+
"epoch": 0.0001257,
|
| 8822 |
+
"grad_norm": 7.2013702392578125,
|
| 8823 |
+
"learning_rate": 1.256e-05,
|
| 8824 |
+
"loss": 51.8125,
|
| 8825 |
+
"step": 1257
|
| 8826 |
+
},
|
| 8827 |
+
{
|
| 8828 |
+
"epoch": 0.0001258,
|
| 8829 |
+
"grad_norm": 7.480557441711426,
|
| 8830 |
+
"learning_rate": 1.257e-05,
|
| 8831 |
+
"loss": 51.9375,
|
| 8832 |
+
"step": 1258
|
| 8833 |
+
},
|
| 8834 |
+
{
|
| 8835 |
+
"epoch": 0.0001259,
|
| 8836 |
+
"grad_norm": 7.298722743988037,
|
| 8837 |
+
"learning_rate": 1.2580000000000002e-05,
|
| 8838 |
+
"loss": 51.3438,
|
| 8839 |
+
"step": 1259
|
| 8840 |
+
},
|
| 8841 |
+
{
|
| 8842 |
+
"epoch": 0.000126,
|
| 8843 |
+
"grad_norm": 7.417407035827637,
|
| 8844 |
+
"learning_rate": 1.259e-05,
|
| 8845 |
+
"loss": 50.875,
|
| 8846 |
+
"step": 1260
|
| 8847 |
+
},
|
| 8848 |
+
{
|
| 8849 |
+
"epoch": 0.0001261,
|
| 8850 |
+
"grad_norm": 7.077691078186035,
|
| 8851 |
+
"learning_rate": 1.26e-05,
|
| 8852 |
+
"loss": 51.6562,
|
| 8853 |
+
"step": 1261
|
| 8854 |
+
},
|
| 8855 |
+
{
|
| 8856 |
+
"epoch": 0.0001262,
|
| 8857 |
+
"grad_norm": 7.0805535316467285,
|
| 8858 |
+
"learning_rate": 1.261e-05,
|
| 8859 |
+
"loss": 51.75,
|
| 8860 |
+
"step": 1262
|
| 8861 |
+
},
|
| 8862 |
+
{
|
| 8863 |
+
"epoch": 0.0001263,
|
| 8864 |
+
"grad_norm": 6.998329162597656,
|
| 8865 |
+
"learning_rate": 1.2620000000000001e-05,
|
| 8866 |
+
"loss": 51.875,
|
| 8867 |
+
"step": 1263
|
| 8868 |
+
},
|
| 8869 |
+
{
|
| 8870 |
+
"epoch": 0.0001264,
|
| 8871 |
+
"grad_norm": 7.11593770980835,
|
| 8872 |
+
"learning_rate": 1.2630000000000001e-05,
|
| 8873 |
+
"loss": 51.6875,
|
| 8874 |
+
"step": 1264
|
| 8875 |
+
},
|
| 8876 |
+
{
|
| 8877 |
+
"epoch": 0.0001265,
|
| 8878 |
+
"grad_norm": 7.197646617889404,
|
| 8879 |
+
"learning_rate": 1.2640000000000001e-05,
|
| 8880 |
+
"loss": 51.3438,
|
| 8881 |
+
"step": 1265
|
| 8882 |
+
},
|
| 8883 |
+
{
|
| 8884 |
+
"epoch": 0.0001266,
|
| 8885 |
+
"grad_norm": 7.040606498718262,
|
| 8886 |
+
"learning_rate": 1.265e-05,
|
| 8887 |
+
"loss": 51.5312,
|
| 8888 |
+
"step": 1266
|
| 8889 |
+
},
|
| 8890 |
+
{
|
| 8891 |
+
"epoch": 0.0001267,
|
| 8892 |
+
"grad_norm": 7.308192253112793,
|
| 8893 |
+
"learning_rate": 1.2659999999999999e-05,
|
| 8894 |
+
"loss": 50.9375,
|
| 8895 |
+
"step": 1267
|
| 8896 |
+
},
|
| 8897 |
+
{
|
| 8898 |
+
"epoch": 0.0001268,
|
| 8899 |
+
"grad_norm": 7.215775012969971,
|
| 8900 |
+
"learning_rate": 1.267e-05,
|
| 8901 |
+
"loss": 51.2188,
|
| 8902 |
+
"step": 1268
|
| 8903 |
+
},
|
| 8904 |
+
{
|
| 8905 |
+
"epoch": 0.0001269,
|
| 8906 |
+
"grad_norm": 7.334770679473877,
|
| 8907 |
+
"learning_rate": 1.268e-05,
|
| 8908 |
+
"loss": 50.6875,
|
| 8909 |
+
"step": 1269
|
| 8910 |
+
},
|
| 8911 |
+
{
|
| 8912 |
+
"epoch": 0.000127,
|
| 8913 |
+
"grad_norm": 7.020657062530518,
|
| 8914 |
+
"learning_rate": 1.269e-05,
|
| 8915 |
+
"loss": 51.5938,
|
| 8916 |
+
"step": 1270
|
| 8917 |
+
},
|
| 8918 |
+
{
|
| 8919 |
+
"epoch": 0.0001271,
|
| 8920 |
+
"grad_norm": 7.122906684875488,
|
| 8921 |
+
"learning_rate": 1.2700000000000002e-05,
|
| 8922 |
+
"loss": 51.6875,
|
| 8923 |
+
"step": 1271
|
| 8924 |
+
},
|
| 8925 |
+
{
|
| 8926 |
+
"epoch": 0.0001272,
|
| 8927 |
+
"grad_norm": 6.957503795623779,
|
| 8928 |
+
"learning_rate": 1.2709999999999999e-05,
|
| 8929 |
+
"loss": 51.5312,
|
| 8930 |
+
"step": 1272
|
| 8931 |
+
},
|
| 8932 |
+
{
|
| 8933 |
+
"epoch": 0.0001273,
|
| 8934 |
+
"grad_norm": 7.044886589050293,
|
| 8935 |
+
"learning_rate": 1.272e-05,
|
| 8936 |
+
"loss": 51.0938,
|
| 8937 |
+
"step": 1273
|
| 8938 |
+
},
|
| 8939 |
+
{
|
| 8940 |
+
"epoch": 0.0001274,
|
| 8941 |
+
"grad_norm": 6.931931972503662,
|
| 8942 |
+
"learning_rate": 1.273e-05,
|
| 8943 |
+
"loss": 51.8125,
|
| 8944 |
+
"step": 1274
|
| 8945 |
+
},
|
| 8946 |
+
{
|
| 8947 |
+
"epoch": 0.0001275,
|
| 8948 |
+
"grad_norm": 7.079439640045166,
|
| 8949 |
+
"learning_rate": 1.274e-05,
|
| 8950 |
+
"loss": 51.375,
|
| 8951 |
+
"step": 1275
|
| 8952 |
+
},
|
| 8953 |
+
{
|
| 8954 |
+
"epoch": 0.0001276,
|
| 8955 |
+
"grad_norm": 7.171270847320557,
|
| 8956 |
+
"learning_rate": 1.2750000000000002e-05,
|
| 8957 |
+
"loss": 50.5625,
|
| 8958 |
+
"step": 1276
|
| 8959 |
+
},
|
| 8960 |
+
{
|
| 8961 |
+
"epoch": 0.0001277,
|
| 8962 |
+
"grad_norm": 7.095261573791504,
|
| 8963 |
+
"learning_rate": 1.276e-05,
|
| 8964 |
+
"loss": 51.5312,
|
| 8965 |
+
"step": 1277
|
| 8966 |
+
},
|
| 8967 |
+
{
|
| 8968 |
+
"epoch": 0.0001278,
|
| 8969 |
+
"grad_norm": 6.948635578155518,
|
| 8970 |
+
"learning_rate": 1.277e-05,
|
| 8971 |
+
"loss": 51.4062,
|
| 8972 |
+
"step": 1278
|
| 8973 |
+
},
|
| 8974 |
+
{
|
| 8975 |
+
"epoch": 0.0001279,
|
| 8976 |
+
"grad_norm": 7.283115863800049,
|
| 8977 |
+
"learning_rate": 1.278e-05,
|
| 8978 |
+
"loss": 50.2188,
|
| 8979 |
+
"step": 1279
|
| 8980 |
+
},
|
| 8981 |
+
{
|
| 8982 |
+
"epoch": 0.000128,
|
| 8983 |
+
"grad_norm": 7.020801544189453,
|
| 8984 |
+
"learning_rate": 1.2790000000000001e-05,
|
| 8985 |
+
"loss": 51.0938,
|
| 8986 |
+
"step": 1280
|
| 8987 |
+
},
|
| 8988 |
+
{
|
| 8989 |
+
"epoch": 0.0001281,
|
| 8990 |
+
"grad_norm": 6.878104209899902,
|
| 8991 |
+
"learning_rate": 1.2800000000000001e-05,
|
| 8992 |
+
"loss": 51.5,
|
| 8993 |
+
"step": 1281
|
| 8994 |
+
},
|
| 8995 |
+
{
|
| 8996 |
+
"epoch": 0.0001282,
|
| 8997 |
+
"grad_norm": 7.140067100524902,
|
| 8998 |
+
"learning_rate": 1.2810000000000001e-05,
|
| 8999 |
+
"loss": 50.6562,
|
| 9000 |
+
"step": 1282
|
| 9001 |
+
},
|
| 9002 |
+
{
|
| 9003 |
+
"epoch": 0.0001283,
|
| 9004 |
+
"grad_norm": 7.064198017120361,
|
| 9005 |
+
"learning_rate": 1.282e-05,
|
| 9006 |
+
"loss": 50.6562,
|
| 9007 |
+
"step": 1283
|
| 9008 |
+
},
|
| 9009 |
+
{
|
| 9010 |
+
"epoch": 0.0001284,
|
| 9011 |
+
"grad_norm": 7.170623302459717,
|
| 9012 |
+
"learning_rate": 1.2829999999999999e-05,
|
| 9013 |
+
"loss": 50.0625,
|
| 9014 |
+
"step": 1284
|
| 9015 |
+
},
|
| 9016 |
+
{
|
| 9017 |
+
"epoch": 0.0001285,
|
| 9018 |
+
"grad_norm": 6.972163200378418,
|
| 9019 |
+
"learning_rate": 1.284e-05,
|
| 9020 |
+
"loss": 51.0625,
|
| 9021 |
+
"step": 1285
|
| 9022 |
+
},
|
| 9023 |
+
{
|
| 9024 |
+
"epoch": 0.0001286,
|
| 9025 |
+
"grad_norm": 7.093153953552246,
|
| 9026 |
+
"learning_rate": 1.285e-05,
|
| 9027 |
+
"loss": 50.4062,
|
| 9028 |
+
"step": 1286
|
| 9029 |
+
},
|
| 9030 |
+
{
|
| 9031 |
+
"epoch": 0.0001287,
|
| 9032 |
+
"grad_norm": 6.831297874450684,
|
| 9033 |
+
"learning_rate": 1.286e-05,
|
| 9034 |
+
"loss": 51.8125,
|
| 9035 |
+
"step": 1287
|
| 9036 |
+
},
|
| 9037 |
+
{
|
| 9038 |
+
"epoch": 0.0001288,
|
| 9039 |
+
"grad_norm": 7.019623279571533,
|
| 9040 |
+
"learning_rate": 1.2870000000000002e-05,
|
| 9041 |
+
"loss": 50.8125,
|
| 9042 |
+
"step": 1288
|
| 9043 |
+
},
|
| 9044 |
+
{
|
| 9045 |
+
"epoch": 0.0001289,
|
| 9046 |
+
"grad_norm": 6.93576717376709,
|
| 9047 |
+
"learning_rate": 1.2879999999999999e-05,
|
| 9048 |
+
"loss": 50.625,
|
| 9049 |
+
"step": 1289
|
| 9050 |
+
},
|
| 9051 |
+
{
|
| 9052 |
+
"epoch": 0.000129,
|
| 9053 |
+
"grad_norm": 7.083858489990234,
|
| 9054 |
+
"learning_rate": 1.289e-05,
|
| 9055 |
+
"loss": 50.1875,
|
| 9056 |
+
"step": 1290
|
| 9057 |
+
},
|
| 9058 |
+
{
|
| 9059 |
+
"epoch": 0.0001291,
|
| 9060 |
+
"grad_norm": 7.067101955413818,
|
| 9061 |
+
"learning_rate": 1.29e-05,
|
| 9062 |
+
"loss": 50.0312,
|
| 9063 |
+
"step": 1291
|
| 9064 |
+
},
|
| 9065 |
+
{
|
| 9066 |
+
"epoch": 0.0001292,
|
| 9067 |
+
"grad_norm": 7.251895427703857,
|
| 9068 |
+
"learning_rate": 1.291e-05,
|
| 9069 |
+
"loss": 49.7188,
|
| 9070 |
+
"step": 1292
|
| 9071 |
+
},
|
| 9072 |
+
{
|
| 9073 |
+
"epoch": 0.0001293,
|
| 9074 |
+
"grad_norm": 6.953596115112305,
|
| 9075 |
+
"learning_rate": 1.2920000000000002e-05,
|
| 9076 |
+
"loss": 50.9062,
|
| 9077 |
+
"step": 1293
|
| 9078 |
+
},
|
| 9079 |
+
{
|
| 9080 |
+
"epoch": 0.0001294,
|
| 9081 |
+
"grad_norm": 6.8302788734436035,
|
| 9082 |
+
"learning_rate": 1.293e-05,
|
| 9083 |
+
"loss": 50.7812,
|
| 9084 |
+
"step": 1294
|
| 9085 |
+
},
|
| 9086 |
+
{
|
| 9087 |
+
"epoch": 0.0001295,
|
| 9088 |
+
"grad_norm": 6.787026405334473,
|
| 9089 |
+
"learning_rate": 1.294e-05,
|
| 9090 |
+
"loss": 50.8438,
|
| 9091 |
+
"step": 1295
|
| 9092 |
+
},
|
| 9093 |
+
{
|
| 9094 |
+
"epoch": 0.0001296,
|
| 9095 |
+
"grad_norm": 6.825897693634033,
|
| 9096 |
+
"learning_rate": 1.295e-05,
|
| 9097 |
+
"loss": 51.25,
|
| 9098 |
+
"step": 1296
|
| 9099 |
+
},
|
| 9100 |
+
{
|
| 9101 |
+
"epoch": 0.0001297,
|
| 9102 |
+
"grad_norm": 7.1078362464904785,
|
| 9103 |
+
"learning_rate": 1.2960000000000001e-05,
|
| 9104 |
+
"loss": 50.4062,
|
| 9105 |
+
"step": 1297
|
| 9106 |
+
},
|
| 9107 |
+
{
|
| 9108 |
+
"epoch": 0.0001298,
|
| 9109 |
+
"grad_norm": 6.9669599533081055,
|
| 9110 |
+
"learning_rate": 1.2970000000000001e-05,
|
| 9111 |
+
"loss": 50.125,
|
| 9112 |
+
"step": 1298
|
| 9113 |
+
},
|
| 9114 |
+
{
|
| 9115 |
+
"epoch": 0.0001299,
|
| 9116 |
+
"grad_norm": 6.832136154174805,
|
| 9117 |
+
"learning_rate": 1.2980000000000001e-05,
|
| 9118 |
+
"loss": 50.4375,
|
| 9119 |
+
"step": 1299
|
| 9120 |
+
},
|
| 9121 |
+
{
|
| 9122 |
+
"epoch": 0.00013,
|
| 9123 |
+
"grad_norm": 6.841429710388184,
|
| 9124 |
+
"learning_rate": 1.299e-05,
|
| 9125 |
+
"loss": 50.3125,
|
| 9126 |
+
"step": 1300
|
| 9127 |
+
},
|
| 9128 |
+
{
|
| 9129 |
+
"epoch": 0.0001301,
|
| 9130 |
+
"grad_norm": 6.736574649810791,
|
| 9131 |
+
"learning_rate": 1.3e-05,
|
| 9132 |
+
"loss": 51.0625,
|
| 9133 |
+
"step": 1301
|
| 9134 |
+
},
|
| 9135 |
+
{
|
| 9136 |
+
"epoch": 0.0001302,
|
| 9137 |
+
"grad_norm": 6.874002456665039,
|
| 9138 |
+
"learning_rate": 1.301e-05,
|
| 9139 |
+
"loss": 50.0938,
|
| 9140 |
+
"step": 1302
|
| 9141 |
+
},
|
| 9142 |
+
{
|
| 9143 |
+
"epoch": 0.0001303,
|
| 9144 |
+
"grad_norm": 6.825052738189697,
|
| 9145 |
+
"learning_rate": 1.302e-05,
|
| 9146 |
+
"loss": 50.3125,
|
| 9147 |
+
"step": 1303
|
| 9148 |
+
},
|
| 9149 |
+
{
|
| 9150 |
+
"epoch": 0.0001304,
|
| 9151 |
+
"grad_norm": 6.87752103805542,
|
| 9152 |
+
"learning_rate": 1.303e-05,
|
| 9153 |
+
"loss": 50.3125,
|
| 9154 |
+
"step": 1304
|
| 9155 |
+
},
|
| 9156 |
+
{
|
| 9157 |
+
"epoch": 0.0001305,
|
| 9158 |
+
"grad_norm": 6.828380107879639,
|
| 9159 |
+
"learning_rate": 1.3040000000000002e-05,
|
| 9160 |
+
"loss": 50.1562,
|
| 9161 |
+
"step": 1305
|
| 9162 |
+
},
|
| 9163 |
+
{
|
| 9164 |
+
"epoch": 0.0001306,
|
| 9165 |
+
"grad_norm": 6.835124969482422,
|
| 9166 |
+
"learning_rate": 1.3049999999999999e-05,
|
| 9167 |
+
"loss": 49.9688,
|
| 9168 |
+
"step": 1306
|
| 9169 |
+
},
|
| 9170 |
+
{
|
| 9171 |
+
"epoch": 0.0001307,
|
| 9172 |
+
"grad_norm": 7.016767978668213,
|
| 9173 |
+
"learning_rate": 1.306e-05,
|
| 9174 |
+
"loss": 50.0,
|
| 9175 |
+
"step": 1307
|
| 9176 |
+
},
|
| 9177 |
+
{
|
| 9178 |
+
"epoch": 0.0001308,
|
| 9179 |
+
"grad_norm": 6.881340503692627,
|
| 9180 |
+
"learning_rate": 1.307e-05,
|
| 9181 |
+
"loss": 49.9062,
|
| 9182 |
+
"step": 1308
|
| 9183 |
+
},
|
| 9184 |
+
{
|
| 9185 |
+
"epoch": 0.0001309,
|
| 9186 |
+
"grad_norm": 6.8892316818237305,
|
| 9187 |
+
"learning_rate": 1.308e-05,
|
| 9188 |
+
"loss": 49.8125,
|
| 9189 |
+
"step": 1309
|
| 9190 |
+
},
|
| 9191 |
+
{
|
| 9192 |
+
"epoch": 0.000131,
|
| 9193 |
+
"grad_norm": 6.788465976715088,
|
| 9194 |
+
"learning_rate": 1.3090000000000002e-05,
|
| 9195 |
+
"loss": 49.9062,
|
| 9196 |
+
"step": 1310
|
| 9197 |
+
},
|
| 9198 |
+
{
|
| 9199 |
+
"epoch": 0.0001311,
|
| 9200 |
+
"grad_norm": 6.9352264404296875,
|
| 9201 |
+
"learning_rate": 1.3099999999999998e-05,
|
| 9202 |
+
"loss": 49.7812,
|
| 9203 |
+
"step": 1311
|
| 9204 |
+
},
|
| 9205 |
+
{
|
| 9206 |
+
"epoch": 0.0001312,
|
| 9207 |
+
"grad_norm": 6.924383163452148,
|
| 9208 |
+
"learning_rate": 1.311e-05,
|
| 9209 |
+
"loss": 49.8438,
|
| 9210 |
+
"step": 1312
|
| 9211 |
+
},
|
| 9212 |
+
{
|
| 9213 |
+
"epoch": 0.0001313,
|
| 9214 |
+
"grad_norm": 6.814214706420898,
|
| 9215 |
+
"learning_rate": 1.312e-05,
|
| 9216 |
+
"loss": 50.0312,
|
| 9217 |
+
"step": 1313
|
| 9218 |
+
},
|
| 9219 |
+
{
|
| 9220 |
+
"epoch": 0.0001314,
|
| 9221 |
+
"grad_norm": 6.863121509552002,
|
| 9222 |
+
"learning_rate": 1.313e-05,
|
| 9223 |
+
"loss": 50.5938,
|
| 9224 |
+
"step": 1314
|
| 9225 |
+
},
|
| 9226 |
+
{
|
| 9227 |
+
"epoch": 0.0001315,
|
| 9228 |
+
"grad_norm": 6.9984354972839355,
|
| 9229 |
+
"learning_rate": 1.3140000000000001e-05,
|
| 9230 |
+
"loss": 49.2188,
|
| 9231 |
+
"step": 1315
|
| 9232 |
+
},
|
| 9233 |
+
{
|
| 9234 |
+
"epoch": 0.0001316,
|
| 9235 |
+
"grad_norm": 6.792233467102051,
|
| 9236 |
+
"learning_rate": 1.3150000000000001e-05,
|
| 9237 |
+
"loss": 49.6562,
|
| 9238 |
+
"step": 1316
|
| 9239 |
+
},
|
| 9240 |
+
{
|
| 9241 |
+
"epoch": 0.0001317,
|
| 9242 |
+
"grad_norm": 6.951003551483154,
|
| 9243 |
+
"learning_rate": 1.316e-05,
|
| 9244 |
+
"loss": 49.3438,
|
| 9245 |
+
"step": 1317
|
| 9246 |
+
},
|
| 9247 |
+
{
|
| 9248 |
+
"epoch": 0.0001318,
|
| 9249 |
+
"grad_norm": 6.749107837677002,
|
| 9250 |
+
"learning_rate": 1.317e-05,
|
| 9251 |
+
"loss": 49.5312,
|
| 9252 |
+
"step": 1318
|
| 9253 |
+
},
|
| 9254 |
+
{
|
| 9255 |
+
"epoch": 0.0001319,
|
| 9256 |
+
"grad_norm": 6.893924713134766,
|
| 9257 |
+
"learning_rate": 1.3180000000000001e-05,
|
| 9258 |
+
"loss": 49.8438,
|
| 9259 |
+
"step": 1319
|
| 9260 |
+
},
|
| 9261 |
+
{
|
| 9262 |
+
"epoch": 0.000132,
|
| 9263 |
+
"grad_norm": 6.869941234588623,
|
| 9264 |
+
"learning_rate": 1.319e-05,
|
| 9265 |
+
"loss": 48.9688,
|
| 9266 |
+
"step": 1320
|
| 9267 |
+
},
|
| 9268 |
+
{
|
| 9269 |
+
"epoch": 0.0001321,
|
| 9270 |
+
"grad_norm": 7.000456809997559,
|
| 9271 |
+
"learning_rate": 1.32e-05,
|
| 9272 |
+
"loss": 49.5312,
|
| 9273 |
+
"step": 1321
|
| 9274 |
+
},
|
| 9275 |
+
{
|
| 9276 |
+
"epoch": 0.0001322,
|
| 9277 |
+
"grad_norm": 6.8329386711120605,
|
| 9278 |
+
"learning_rate": 1.3210000000000002e-05,
|
| 9279 |
+
"loss": 49.2812,
|
| 9280 |
+
"step": 1322
|
| 9281 |
+
},
|
| 9282 |
+
{
|
| 9283 |
+
"epoch": 0.0001323,
|
| 9284 |
+
"grad_norm": 6.7939043045043945,
|
| 9285 |
+
"learning_rate": 1.3219999999999999e-05,
|
| 9286 |
+
"loss": 49.6562,
|
| 9287 |
+
"step": 1323
|
| 9288 |
+
},
|
| 9289 |
+
{
|
| 9290 |
+
"epoch": 0.0001324,
|
| 9291 |
+
"grad_norm": 6.9336137771606445,
|
| 9292 |
+
"learning_rate": 1.323e-05,
|
| 9293 |
+
"loss": 49.4688,
|
| 9294 |
+
"step": 1324
|
| 9295 |
+
},
|
| 9296 |
+
{
|
| 9297 |
+
"epoch": 0.0001325,
|
| 9298 |
+
"grad_norm": 6.958576202392578,
|
| 9299 |
+
"learning_rate": 1.324e-05,
|
| 9300 |
+
"loss": 49.1875,
|
| 9301 |
+
"step": 1325
|
| 9302 |
+
},
|
| 9303 |
+
{
|
| 9304 |
+
"epoch": 0.0001326,
|
| 9305 |
+
"grad_norm": 6.707515716552734,
|
| 9306 |
+
"learning_rate": 1.325e-05,
|
| 9307 |
+
"loss": 50.0312,
|
| 9308 |
+
"step": 1326
|
| 9309 |
+
},
|
| 9310 |
+
{
|
| 9311 |
+
"epoch": 0.0001327,
|
| 9312 |
+
"grad_norm": 6.904212474822998,
|
| 9313 |
+
"learning_rate": 1.3260000000000002e-05,
|
| 9314 |
+
"loss": 48.9062,
|
| 9315 |
+
"step": 1327
|
| 9316 |
+
},
|
| 9317 |
+
{
|
| 9318 |
+
"epoch": 0.0001328,
|
| 9319 |
+
"grad_norm": 6.822237491607666,
|
| 9320 |
+
"learning_rate": 1.3269999999999998e-05,
|
| 9321 |
+
"loss": 48.9375,
|
| 9322 |
+
"step": 1328
|
| 9323 |
+
},
|
| 9324 |
+
{
|
| 9325 |
+
"epoch": 0.0001329,
|
| 9326 |
+
"grad_norm": 6.981447696685791,
|
| 9327 |
+
"learning_rate": 1.328e-05,
|
| 9328 |
+
"loss": 49.0625,
|
| 9329 |
+
"step": 1329
|
| 9330 |
+
},
|
| 9331 |
+
{
|
| 9332 |
+
"epoch": 0.000133,
|
| 9333 |
+
"grad_norm": 6.96505880355835,
|
| 9334 |
+
"learning_rate": 1.329e-05,
|
| 9335 |
+
"loss": 48.9062,
|
| 9336 |
+
"step": 1330
|
| 9337 |
+
},
|
| 9338 |
+
{
|
| 9339 |
+
"epoch": 0.0001331,
|
| 9340 |
+
"grad_norm": 6.847214221954346,
|
| 9341 |
+
"learning_rate": 1.33e-05,
|
| 9342 |
+
"loss": 49.375,
|
| 9343 |
+
"step": 1331
|
| 9344 |
+
},
|
| 9345 |
+
{
|
| 9346 |
+
"epoch": 0.0001332,
|
| 9347 |
+
"grad_norm": 6.874561786651611,
|
| 9348 |
+
"learning_rate": 1.3310000000000001e-05,
|
| 9349 |
+
"loss": 48.9062,
|
| 9350 |
+
"step": 1332
|
| 9351 |
+
},
|
| 9352 |
+
{
|
| 9353 |
+
"epoch": 0.0001333,
|
| 9354 |
+
"grad_norm": 6.889916896820068,
|
| 9355 |
+
"learning_rate": 1.3320000000000001e-05,
|
| 9356 |
+
"loss": 48.5625,
|
| 9357 |
+
"step": 1333
|
| 9358 |
+
},
|
| 9359 |
+
{
|
| 9360 |
+
"epoch": 0.0001334,
|
| 9361 |
+
"grad_norm": 6.986288070678711,
|
| 9362 |
+
"learning_rate": 1.333e-05,
|
| 9363 |
+
"loss": 48.9688,
|
| 9364 |
+
"step": 1334
|
| 9365 |
+
},
|
| 9366 |
+
{
|
| 9367 |
+
"epoch": 0.0001335,
|
| 9368 |
+
"grad_norm": 7.319582462310791,
|
| 9369 |
+
"learning_rate": 1.334e-05,
|
| 9370 |
+
"loss": 49.3125,
|
| 9371 |
+
"step": 1335
|
| 9372 |
+
},
|
| 9373 |
+
{
|
| 9374 |
+
"epoch": 0.0001336,
|
| 9375 |
+
"grad_norm": 6.805470943450928,
|
| 9376 |
+
"learning_rate": 1.3350000000000001e-05,
|
| 9377 |
+
"loss": 48.6562,
|
| 9378 |
+
"step": 1336
|
| 9379 |
+
},
|
| 9380 |
+
{
|
| 9381 |
+
"epoch": 0.0001337,
|
| 9382 |
+
"grad_norm": 6.76211404800415,
|
| 9383 |
+
"learning_rate": 1.336e-05,
|
| 9384 |
+
"loss": 49.0625,
|
| 9385 |
+
"step": 1337
|
| 9386 |
+
},
|
| 9387 |
+
{
|
| 9388 |
+
"epoch": 0.0001338,
|
| 9389 |
+
"grad_norm": 6.805838108062744,
|
| 9390 |
+
"learning_rate": 1.337e-05,
|
| 9391 |
+
"loss": 48.8438,
|
| 9392 |
+
"step": 1338
|
| 9393 |
+
},
|
| 9394 |
+
{
|
| 9395 |
+
"epoch": 0.0001339,
|
| 9396 |
+
"grad_norm": 6.896024227142334,
|
| 9397 |
+
"learning_rate": 1.3380000000000002e-05,
|
| 9398 |
+
"loss": 48.6562,
|
| 9399 |
+
"step": 1339
|
| 9400 |
+
},
|
| 9401 |
+
{
|
| 9402 |
+
"epoch": 0.000134,
|
| 9403 |
+
"grad_norm": 6.7913737297058105,
|
| 9404 |
+
"learning_rate": 1.3389999999999999e-05,
|
| 9405 |
+
"loss": 48.25,
|
| 9406 |
+
"step": 1340
|
| 9407 |
+
},
|
| 9408 |
+
{
|
| 9409 |
+
"epoch": 0.0001341,
|
| 9410 |
+
"grad_norm": 6.825916767120361,
|
| 9411 |
+
"learning_rate": 1.34e-05,
|
| 9412 |
+
"loss": 48.6875,
|
| 9413 |
+
"step": 1341
|
| 9414 |
+
},
|
| 9415 |
+
{
|
| 9416 |
+
"epoch": 0.0001342,
|
| 9417 |
+
"grad_norm": 6.727726459503174,
|
| 9418 |
+
"learning_rate": 1.341e-05,
|
| 9419 |
+
"loss": 49.0625,
|
| 9420 |
+
"step": 1342
|
| 9421 |
+
},
|
| 9422 |
+
{
|
| 9423 |
+
"epoch": 0.0001343,
|
| 9424 |
+
"grad_norm": 6.678559303283691,
|
| 9425 |
+
"learning_rate": 1.342e-05,
|
| 9426 |
+
"loss": 48.6562,
|
| 9427 |
+
"step": 1343
|
| 9428 |
+
},
|
| 9429 |
+
{
|
| 9430 |
+
"epoch": 0.0001344,
|
| 9431 |
+
"grad_norm": 6.7459564208984375,
|
| 9432 |
+
"learning_rate": 1.3430000000000002e-05,
|
| 9433 |
+
"loss": 49.3438,
|
| 9434 |
+
"step": 1344
|
| 9435 |
+
},
|
| 9436 |
+
{
|
| 9437 |
+
"epoch": 0.0001345,
|
| 9438 |
+
"grad_norm": 6.67067289352417,
|
| 9439 |
+
"learning_rate": 1.3439999999999998e-05,
|
| 9440 |
+
"loss": 49.3438,
|
| 9441 |
+
"step": 1345
|
| 9442 |
+
},
|
| 9443 |
+
{
|
| 9444 |
+
"epoch": 0.0001346,
|
| 9445 |
+
"grad_norm": 6.752988338470459,
|
| 9446 |
+
"learning_rate": 1.345e-05,
|
| 9447 |
+
"loss": 48.5312,
|
| 9448 |
+
"step": 1346
|
| 9449 |
+
},
|
| 9450 |
+
{
|
| 9451 |
+
"epoch": 0.0001347,
|
| 9452 |
+
"grad_norm": 6.771851539611816,
|
| 9453 |
+
"learning_rate": 1.346e-05,
|
| 9454 |
+
"loss": 48.9375,
|
| 9455 |
+
"step": 1347
|
| 9456 |
+
},
|
| 9457 |
+
{
|
| 9458 |
+
"epoch": 0.0001348,
|
| 9459 |
+
"grad_norm": 6.481204032897949,
|
| 9460 |
+
"learning_rate": 1.347e-05,
|
| 9461 |
+
"loss": 49.5938,
|
| 9462 |
+
"step": 1348
|
| 9463 |
+
},
|
| 9464 |
+
{
|
| 9465 |
+
"epoch": 0.0001349,
|
| 9466 |
+
"grad_norm": 6.8165388107299805,
|
| 9467 |
+
"learning_rate": 1.3480000000000001e-05,
|
| 9468 |
+
"loss": 48.875,
|
| 9469 |
+
"step": 1349
|
| 9470 |
+
},
|
| 9471 |
+
{
|
| 9472 |
+
"epoch": 0.000135,
|
| 9473 |
+
"grad_norm": 6.703948497772217,
|
| 9474 |
+
"learning_rate": 1.3490000000000001e-05,
|
| 9475 |
+
"loss": 48.1875,
|
| 9476 |
+
"step": 1350
|
| 9477 |
+
},
|
| 9478 |
+
{
|
| 9479 |
+
"epoch": 0.0001351,
|
| 9480 |
+
"grad_norm": 6.681395053863525,
|
| 9481 |
+
"learning_rate": 1.35e-05,
|
| 9482 |
+
"loss": 48.125,
|
| 9483 |
+
"step": 1351
|
| 9484 |
+
},
|
| 9485 |
+
{
|
| 9486 |
+
"epoch": 0.0001352,
|
| 9487 |
+
"grad_norm": 6.673883438110352,
|
| 9488 |
+
"learning_rate": 1.351e-05,
|
| 9489 |
+
"loss": 48.3438,
|
| 9490 |
+
"step": 1352
|
| 9491 |
+
},
|
| 9492 |
+
{
|
| 9493 |
+
"epoch": 0.0001353,
|
| 9494 |
+
"grad_norm": 6.701688766479492,
|
| 9495 |
+
"learning_rate": 1.3520000000000001e-05,
|
| 9496 |
+
"loss": 48.9688,
|
| 9497 |
+
"step": 1353
|
| 9498 |
+
},
|
| 9499 |
+
{
|
| 9500 |
+
"epoch": 0.0001354,
|
| 9501 |
+
"grad_norm": 6.5370025634765625,
|
| 9502 |
+
"learning_rate": 1.3530000000000001e-05,
|
| 9503 |
+
"loss": 48.5312,
|
| 9504 |
+
"step": 1354
|
| 9505 |
+
},
|
| 9506 |
+
{
|
| 9507 |
+
"epoch": 0.0001355,
|
| 9508 |
+
"grad_norm": 6.6904706954956055,
|
| 9509 |
+
"learning_rate": 1.354e-05,
|
| 9510 |
+
"loss": 47.7188,
|
| 9511 |
+
"step": 1355
|
| 9512 |
+
},
|
| 9513 |
+
{
|
| 9514 |
+
"epoch": 0.0001356,
|
| 9515 |
+
"grad_norm": 6.671186923980713,
|
| 9516 |
+
"learning_rate": 1.3550000000000002e-05,
|
| 9517 |
+
"loss": 48.4062,
|
| 9518 |
+
"step": 1356
|
| 9519 |
+
},
|
| 9520 |
+
{
|
| 9521 |
+
"epoch": 0.0001357,
|
| 9522 |
+
"grad_norm": 6.853418827056885,
|
| 9523 |
+
"learning_rate": 1.3559999999999999e-05,
|
| 9524 |
+
"loss": 49.3438,
|
| 9525 |
+
"step": 1357
|
| 9526 |
+
},
|
| 9527 |
+
{
|
| 9528 |
+
"epoch": 0.0001358,
|
| 9529 |
+
"grad_norm": 6.607544422149658,
|
| 9530 |
+
"learning_rate": 1.357e-05,
|
| 9531 |
+
"loss": 49.0312,
|
| 9532 |
+
"step": 1358
|
| 9533 |
+
},
|
| 9534 |
+
{
|
| 9535 |
+
"epoch": 0.0001359,
|
| 9536 |
+
"grad_norm": 6.757382869720459,
|
| 9537 |
+
"learning_rate": 1.358e-05,
|
| 9538 |
+
"loss": 48.125,
|
| 9539 |
+
"step": 1359
|
| 9540 |
+
},
|
| 9541 |
+
{
|
| 9542 |
+
"epoch": 0.000136,
|
| 9543 |
+
"grad_norm": 6.590203285217285,
|
| 9544 |
+
"learning_rate": 1.359e-05,
|
| 9545 |
+
"loss": 48.2812,
|
| 9546 |
+
"step": 1360
|
| 9547 |
+
},
|
| 9548 |
+
{
|
| 9549 |
+
"epoch": 0.0001361,
|
| 9550 |
+
"grad_norm": 6.539251804351807,
|
| 9551 |
+
"learning_rate": 1.3600000000000002e-05,
|
| 9552 |
+
"loss": 48.0938,
|
| 9553 |
+
"step": 1361
|
| 9554 |
+
},
|
| 9555 |
+
{
|
| 9556 |
+
"epoch": 0.0001362,
|
| 9557 |
+
"grad_norm": 6.598182201385498,
|
| 9558 |
+
"learning_rate": 1.3609999999999998e-05,
|
| 9559 |
+
"loss": 48.125,
|
| 9560 |
+
"step": 1362
|
| 9561 |
+
},
|
| 9562 |
+
{
|
| 9563 |
+
"epoch": 0.0001363,
|
| 9564 |
+
"grad_norm": 6.8250226974487305,
|
| 9565 |
+
"learning_rate": 1.362e-05,
|
| 9566 |
+
"loss": 47.6562,
|
| 9567 |
+
"step": 1363
|
| 9568 |
+
},
|
| 9569 |
+
{
|
| 9570 |
+
"epoch": 0.0001364,
|
| 9571 |
+
"grad_norm": 6.718772888183594,
|
| 9572 |
+
"learning_rate": 1.363e-05,
|
| 9573 |
+
"loss": 48.1875,
|
| 9574 |
+
"step": 1364
|
| 9575 |
+
},
|
| 9576 |
+
{
|
| 9577 |
+
"epoch": 0.0001365,
|
| 9578 |
+
"grad_norm": 6.544684886932373,
|
| 9579 |
+
"learning_rate": 1.364e-05,
|
| 9580 |
+
"loss": 48.0312,
|
| 9581 |
+
"step": 1365
|
| 9582 |
+
},
|
| 9583 |
+
{
|
| 9584 |
+
"epoch": 0.0001366,
|
| 9585 |
+
"grad_norm": 6.549186706542969,
|
| 9586 |
+
"learning_rate": 1.3650000000000001e-05,
|
| 9587 |
+
"loss": 47.9375,
|
| 9588 |
+
"step": 1366
|
| 9589 |
+
},
|
| 9590 |
+
{
|
| 9591 |
+
"epoch": 0.0001367,
|
| 9592 |
+
"grad_norm": 6.598300457000732,
|
| 9593 |
+
"learning_rate": 1.3660000000000001e-05,
|
| 9594 |
+
"loss": 48.25,
|
| 9595 |
+
"step": 1367
|
| 9596 |
+
},
|
| 9597 |
+
{
|
| 9598 |
+
"epoch": 0.0001368,
|
| 9599 |
+
"grad_norm": 6.672146797180176,
|
| 9600 |
+
"learning_rate": 1.367e-05,
|
| 9601 |
+
"loss": 47.125,
|
| 9602 |
+
"step": 1368
|
| 9603 |
+
},
|
| 9604 |
+
{
|
| 9605 |
+
"epoch": 0.0001369,
|
| 9606 |
+
"grad_norm": 6.587193012237549,
|
| 9607 |
+
"learning_rate": 1.368e-05,
|
| 9608 |
+
"loss": 48.1875,
|
| 9609 |
+
"step": 1369
|
| 9610 |
+
},
|
| 9611 |
+
{
|
| 9612 |
+
"epoch": 0.000137,
|
| 9613 |
+
"grad_norm": 6.673519611358643,
|
| 9614 |
+
"learning_rate": 1.369e-05,
|
| 9615 |
+
"loss": 47.5,
|
| 9616 |
+
"step": 1370
|
| 9617 |
+
},
|
| 9618 |
+
{
|
| 9619 |
+
"epoch": 0.0001371,
|
| 9620 |
+
"grad_norm": 6.656841278076172,
|
| 9621 |
+
"learning_rate": 1.3700000000000001e-05,
|
| 9622 |
+
"loss": 49.6875,
|
| 9623 |
+
"step": 1371
|
| 9624 |
+
},
|
| 9625 |
+
{
|
| 9626 |
+
"epoch": 0.0001372,
|
| 9627 |
+
"grad_norm": 6.472851753234863,
|
| 9628 |
+
"learning_rate": 1.3710000000000001e-05,
|
| 9629 |
+
"loss": 48.3438,
|
| 9630 |
+
"step": 1372
|
| 9631 |
+
},
|
| 9632 |
+
{
|
| 9633 |
+
"epoch": 0.0001373,
|
| 9634 |
+
"grad_norm": 6.7288970947265625,
|
| 9635 |
+
"learning_rate": 1.372e-05,
|
| 9636 |
+
"loss": 47.9688,
|
| 9637 |
+
"step": 1373
|
| 9638 |
+
},
|
| 9639 |
+
{
|
| 9640 |
+
"epoch": 0.0001374,
|
| 9641 |
+
"grad_norm": 6.470300674438477,
|
| 9642 |
+
"learning_rate": 1.3729999999999999e-05,
|
| 9643 |
+
"loss": 48.9062,
|
| 9644 |
+
"step": 1374
|
| 9645 |
+
},
|
| 9646 |
+
{
|
| 9647 |
+
"epoch": 0.0001375,
|
| 9648 |
+
"grad_norm": 6.735988140106201,
|
| 9649 |
+
"learning_rate": 1.374e-05,
|
| 9650 |
+
"loss": 47.75,
|
| 9651 |
+
"step": 1375
|
| 9652 |
+
},
|
| 9653 |
+
{
|
| 9654 |
+
"epoch": 0.0001376,
|
| 9655 |
+
"grad_norm": 6.431521892547607,
|
| 9656 |
+
"learning_rate": 1.375e-05,
|
| 9657 |
+
"loss": 48.3438,
|
| 9658 |
+
"step": 1376
|
| 9659 |
+
},
|
| 9660 |
+
{
|
| 9661 |
+
"epoch": 0.0001377,
|
| 9662 |
+
"grad_norm": 6.473598957061768,
|
| 9663 |
+
"learning_rate": 1.376e-05,
|
| 9664 |
+
"loss": 48.3438,
|
| 9665 |
+
"step": 1377
|
| 9666 |
+
},
|
| 9667 |
+
{
|
| 9668 |
+
"epoch": 0.0001378,
|
| 9669 |
+
"grad_norm": 6.585727214813232,
|
| 9670 |
+
"learning_rate": 1.3770000000000002e-05,
|
| 9671 |
+
"loss": 47.4375,
|
| 9672 |
+
"step": 1378
|
| 9673 |
+
},
|
| 9674 |
+
{
|
| 9675 |
+
"epoch": 0.0001379,
|
| 9676 |
+
"grad_norm": 6.473487854003906,
|
| 9677 |
+
"learning_rate": 1.3779999999999999e-05,
|
| 9678 |
+
"loss": 47.6875,
|
| 9679 |
+
"step": 1379
|
| 9680 |
+
},
|
| 9681 |
+
{
|
| 9682 |
+
"epoch": 0.000138,
|
| 9683 |
+
"grad_norm": 6.589076519012451,
|
| 9684 |
+
"learning_rate": 1.379e-05,
|
| 9685 |
+
"loss": 47.0,
|
| 9686 |
+
"step": 1380
|
| 9687 |
+
},
|
| 9688 |
+
{
|
| 9689 |
+
"epoch": 0.0001381,
|
| 9690 |
+
"grad_norm": 6.590631484985352,
|
| 9691 |
+
"learning_rate": 1.38e-05,
|
| 9692 |
+
"loss": 47.6875,
|
| 9693 |
+
"step": 1381
|
| 9694 |
+
},
|
| 9695 |
+
{
|
| 9696 |
+
"epoch": 0.0001382,
|
| 9697 |
+
"grad_norm": 6.402490139007568,
|
| 9698 |
+
"learning_rate": 1.381e-05,
|
| 9699 |
+
"loss": 48.125,
|
| 9700 |
+
"step": 1382
|
| 9701 |
+
},
|
| 9702 |
+
{
|
| 9703 |
+
"epoch": 0.0001383,
|
| 9704 |
+
"grad_norm": 6.432614326477051,
|
| 9705 |
+
"learning_rate": 1.3820000000000002e-05,
|
| 9706 |
+
"loss": 47.4688,
|
| 9707 |
+
"step": 1383
|
| 9708 |
+
},
|
| 9709 |
+
{
|
| 9710 |
+
"epoch": 0.0001384,
|
| 9711 |
+
"grad_norm": 6.533024311065674,
|
| 9712 |
+
"learning_rate": 1.3830000000000001e-05,
|
| 9713 |
+
"loss": 47.75,
|
| 9714 |
+
"step": 1384
|
| 9715 |
+
},
|
| 9716 |
+
{
|
| 9717 |
+
"epoch": 0.0001385,
|
| 9718 |
+
"grad_norm": 6.51165246963501,
|
| 9719 |
+
"learning_rate": 1.384e-05,
|
| 9720 |
+
"loss": 47.625,
|
| 9721 |
+
"step": 1385
|
| 9722 |
+
},
|
| 9723 |
+
{
|
| 9724 |
+
"epoch": 0.0001386,
|
| 9725 |
+
"grad_norm": 6.475950717926025,
|
| 9726 |
+
"learning_rate": 1.385e-05,
|
| 9727 |
+
"loss": 47.6562,
|
| 9728 |
+
"step": 1386
|
| 9729 |
+
},
|
| 9730 |
+
{
|
| 9731 |
+
"epoch": 0.0001387,
|
| 9732 |
+
"grad_norm": 6.405846118927002,
|
| 9733 |
+
"learning_rate": 1.386e-05,
|
| 9734 |
+
"loss": 48.1562,
|
| 9735 |
+
"step": 1387
|
| 9736 |
+
},
|
| 9737 |
+
{
|
| 9738 |
+
"epoch": 0.0001388,
|
| 9739 |
+
"grad_norm": 6.439004898071289,
|
| 9740 |
+
"learning_rate": 1.3870000000000001e-05,
|
| 9741 |
+
"loss": 47.8438,
|
| 9742 |
+
"step": 1388
|
| 9743 |
+
},
|
| 9744 |
+
{
|
| 9745 |
+
"epoch": 0.0001389,
|
| 9746 |
+
"grad_norm": 6.551534175872803,
|
| 9747 |
+
"learning_rate": 1.3880000000000001e-05,
|
| 9748 |
+
"loss": 47.125,
|
| 9749 |
+
"step": 1389
|
| 9750 |
+
},
|
| 9751 |
+
{
|
| 9752 |
+
"epoch": 0.000139,
|
| 9753 |
+
"grad_norm": 6.39193868637085,
|
| 9754 |
+
"learning_rate": 1.3889999999999999e-05,
|
| 9755 |
+
"loss": 47.4375,
|
| 9756 |
+
"step": 1390
|
| 9757 |
+
},
|
| 9758 |
+
{
|
| 9759 |
+
"epoch": 0.0001391,
|
| 9760 |
+
"grad_norm": 6.597825050354004,
|
| 9761 |
+
"learning_rate": 1.3899999999999999e-05,
|
| 9762 |
+
"loss": 46.6562,
|
| 9763 |
+
"step": 1391
|
| 9764 |
+
},
|
| 9765 |
+
{
|
| 9766 |
+
"epoch": 0.0001392,
|
| 9767 |
+
"grad_norm": 6.471019268035889,
|
| 9768 |
+
"learning_rate": 1.391e-05,
|
| 9769 |
+
"loss": 47.5625,
|
| 9770 |
+
"step": 1392
|
| 9771 |
+
},
|
| 9772 |
+
{
|
| 9773 |
+
"epoch": 0.0001393,
|
| 9774 |
+
"grad_norm": 6.471146106719971,
|
| 9775 |
+
"learning_rate": 1.392e-05,
|
| 9776 |
+
"loss": 47.0,
|
| 9777 |
+
"step": 1393
|
| 9778 |
+
},
|
| 9779 |
+
{
|
| 9780 |
+
"epoch": 0.0001394,
|
| 9781 |
+
"grad_norm": 6.441690921783447,
|
| 9782 |
+
"learning_rate": 1.393e-05,
|
| 9783 |
+
"loss": 47.6562,
|
| 9784 |
+
"step": 1394
|
| 9785 |
+
},
|
| 9786 |
+
{
|
| 9787 |
+
"epoch": 0.0001395,
|
| 9788 |
+
"grad_norm": 6.480464935302734,
|
| 9789 |
+
"learning_rate": 1.3940000000000002e-05,
|
| 9790 |
+
"loss": 47.625,
|
| 9791 |
+
"step": 1395
|
| 9792 |
+
},
|
| 9793 |
+
{
|
| 9794 |
+
"epoch": 0.0001396,
|
| 9795 |
+
"grad_norm": 6.39990758895874,
|
| 9796 |
+
"learning_rate": 1.3949999999999999e-05,
|
| 9797 |
+
"loss": 47.5312,
|
| 9798 |
+
"step": 1396
|
| 9799 |
+
},
|
| 9800 |
+
{
|
| 9801 |
+
"epoch": 0.0001397,
|
| 9802 |
+
"grad_norm": 6.367557048797607,
|
| 9803 |
+
"learning_rate": 1.396e-05,
|
| 9804 |
+
"loss": 47.4688,
|
| 9805 |
+
"step": 1397
|
| 9806 |
+
},
|
| 9807 |
+
{
|
| 9808 |
+
"epoch": 0.0001398,
|
| 9809 |
+
"grad_norm": 6.547006607055664,
|
| 9810 |
+
"learning_rate": 1.397e-05,
|
| 9811 |
+
"loss": 46.4375,
|
| 9812 |
+
"step": 1398
|
| 9813 |
+
},
|
| 9814 |
+
{
|
| 9815 |
+
"epoch": 0.0001399,
|
| 9816 |
+
"grad_norm": 6.42279052734375,
|
| 9817 |
+
"learning_rate": 1.398e-05,
|
| 9818 |
+
"loss": 48.0312,
|
| 9819 |
+
"step": 1399
|
| 9820 |
+
},
|
| 9821 |
+
{
|
| 9822 |
+
"epoch": 0.00014,
|
| 9823 |
+
"grad_norm": 6.570935249328613,
|
| 9824 |
+
"learning_rate": 1.3990000000000002e-05,
|
| 9825 |
+
"loss": 47.4375,
|
| 9826 |
+
"step": 1400
|
| 9827 |
+
},
|
| 9828 |
+
{
|
| 9829 |
+
"epoch": 0.0001401,
|
| 9830 |
+
"grad_norm": 6.599931240081787,
|
| 9831 |
+
"learning_rate": 1.4000000000000001e-05,
|
| 9832 |
+
"loss": 47.0312,
|
| 9833 |
+
"step": 1401
|
| 9834 |
+
},
|
| 9835 |
+
{
|
| 9836 |
+
"epoch": 0.0001402,
|
| 9837 |
+
"grad_norm": 6.265009880065918,
|
| 9838 |
+
"learning_rate": 1.401e-05,
|
| 9839 |
+
"loss": 47.1875,
|
| 9840 |
+
"step": 1402
|
| 9841 |
+
},
|
| 9842 |
+
{
|
| 9843 |
+
"epoch": 0.0001403,
|
| 9844 |
+
"grad_norm": 6.438925266265869,
|
| 9845 |
+
"learning_rate": 1.402e-05,
|
| 9846 |
+
"loss": 46.6875,
|
| 9847 |
+
"step": 1403
|
| 9848 |
+
},
|
| 9849 |
+
{
|
| 9850 |
+
"epoch": 0.0001404,
|
| 9851 |
+
"grad_norm": 6.475961208343506,
|
| 9852 |
+
"learning_rate": 1.403e-05,
|
| 9853 |
+
"loss": 46.9062,
|
| 9854 |
+
"step": 1404
|
| 9855 |
+
},
|
| 9856 |
+
{
|
| 9857 |
+
"epoch": 0.0001405,
|
| 9858 |
+
"grad_norm": 6.573430061340332,
|
| 9859 |
+
"learning_rate": 1.4040000000000001e-05,
|
| 9860 |
+
"loss": 46.3125,
|
| 9861 |
+
"step": 1405
|
| 9862 |
+
},
|
| 9863 |
+
{
|
| 9864 |
+
"epoch": 0.0001406,
|
| 9865 |
+
"grad_norm": 6.558287143707275,
|
| 9866 |
+
"learning_rate": 1.4050000000000001e-05,
|
| 9867 |
+
"loss": 46.4062,
|
| 9868 |
+
"step": 1406
|
| 9869 |
+
},
|
| 9870 |
+
{
|
| 9871 |
+
"epoch": 0.0001407,
|
| 9872 |
+
"grad_norm": 6.389150619506836,
|
| 9873 |
+
"learning_rate": 1.406e-05,
|
| 9874 |
+
"loss": 47.1562,
|
| 9875 |
+
"step": 1407
|
| 9876 |
+
},
|
| 9877 |
+
{
|
| 9878 |
+
"epoch": 0.0001408,
|
| 9879 |
+
"grad_norm": 6.322193145751953,
|
| 9880 |
+
"learning_rate": 1.4069999999999999e-05,
|
| 9881 |
+
"loss": 48.1562,
|
| 9882 |
+
"step": 1408
|
| 9883 |
+
},
|
| 9884 |
+
{
|
| 9885 |
+
"epoch": 0.0001409,
|
| 9886 |
+
"grad_norm": 6.224388122558594,
|
| 9887 |
+
"learning_rate": 1.408e-05,
|
| 9888 |
+
"loss": 47.625,
|
| 9889 |
+
"step": 1409
|
| 9890 |
+
},
|
| 9891 |
+
{
|
| 9892 |
+
"epoch": 0.000141,
|
| 9893 |
+
"grad_norm": 6.360954761505127,
|
| 9894 |
+
"learning_rate": 1.409e-05,
|
| 9895 |
+
"loss": 47.25,
|
| 9896 |
+
"step": 1410
|
| 9897 |
+
},
|
| 9898 |
+
{
|
| 9899 |
+
"epoch": 0.0001411,
|
| 9900 |
+
"grad_norm": 6.193026542663574,
|
| 9901 |
+
"learning_rate": 1.41e-05,
|
| 9902 |
+
"loss": 47.9375,
|
| 9903 |
+
"step": 1411
|
| 9904 |
+
},
|
| 9905 |
+
{
|
| 9906 |
+
"epoch": 0.0001412,
|
| 9907 |
+
"grad_norm": 6.543434143066406,
|
| 9908 |
+
"learning_rate": 1.4110000000000002e-05,
|
| 9909 |
+
"loss": 47.7188,
|
| 9910 |
+
"step": 1412
|
| 9911 |
+
},
|
| 9912 |
+
{
|
| 9913 |
+
"epoch": 0.0001413,
|
| 9914 |
+
"grad_norm": 6.402957916259766,
|
| 9915 |
+
"learning_rate": 1.4119999999999999e-05,
|
| 9916 |
+
"loss": 47.0625,
|
| 9917 |
+
"step": 1413
|
| 9918 |
+
},
|
| 9919 |
+
{
|
| 9920 |
+
"epoch": 0.0001414,
|
| 9921 |
+
"grad_norm": 6.295007705688477,
|
| 9922 |
+
"learning_rate": 1.413e-05,
|
| 9923 |
+
"loss": 47.3125,
|
| 9924 |
+
"step": 1414
|
| 9925 |
+
},
|
| 9926 |
+
{
|
| 9927 |
+
"epoch": 0.0001415,
|
| 9928 |
+
"grad_norm": 6.175797462463379,
|
| 9929 |
+
"learning_rate": 1.414e-05,
|
| 9930 |
+
"loss": 48.0625,
|
| 9931 |
+
"step": 1415
|
| 9932 |
+
},
|
| 9933 |
+
{
|
| 9934 |
+
"epoch": 0.0001416,
|
| 9935 |
+
"grad_norm": 6.613070011138916,
|
| 9936 |
+
"learning_rate": 1.415e-05,
|
| 9937 |
+
"loss": 45.8125,
|
| 9938 |
+
"step": 1416
|
| 9939 |
+
},
|
| 9940 |
+
{
|
| 9941 |
+
"epoch": 0.0001417,
|
| 9942 |
+
"grad_norm": 6.471121311187744,
|
| 9943 |
+
"learning_rate": 1.4160000000000002e-05,
|
| 9944 |
+
"loss": 46.625,
|
| 9945 |
+
"step": 1417
|
| 9946 |
+
},
|
| 9947 |
+
{
|
| 9948 |
+
"epoch": 0.0001418,
|
| 9949 |
+
"grad_norm": 6.47176456451416,
|
| 9950 |
+
"learning_rate": 1.4170000000000002e-05,
|
| 9951 |
+
"loss": 46.3438,
|
| 9952 |
+
"step": 1418
|
| 9953 |
+
},
|
| 9954 |
+
{
|
| 9955 |
+
"epoch": 0.0001419,
|
| 9956 |
+
"grad_norm": 6.427188396453857,
|
| 9957 |
+
"learning_rate": 1.418e-05,
|
| 9958 |
+
"loss": 46.125,
|
| 9959 |
+
"step": 1419
|
| 9960 |
+
},
|
| 9961 |
+
{
|
| 9962 |
+
"epoch": 0.000142,
|
| 9963 |
+
"grad_norm": 6.524816989898682,
|
| 9964 |
+
"learning_rate": 1.419e-05,
|
| 9965 |
+
"loss": 45.9375,
|
| 9966 |
+
"step": 1420
|
| 9967 |
+
},
|
| 9968 |
+
{
|
| 9969 |
+
"epoch": 0.0001421,
|
| 9970 |
+
"grad_norm": 6.358154296875,
|
| 9971 |
+
"learning_rate": 1.42e-05,
|
| 9972 |
+
"loss": 47.2812,
|
| 9973 |
+
"step": 1421
|
| 9974 |
+
},
|
| 9975 |
+
{
|
| 9976 |
+
"epoch": 0.0001422,
|
| 9977 |
+
"grad_norm": 6.3673624992370605,
|
| 9978 |
+
"learning_rate": 1.4210000000000001e-05,
|
| 9979 |
+
"loss": 46.4375,
|
| 9980 |
+
"step": 1422
|
| 9981 |
+
},
|
| 9982 |
+
{
|
| 9983 |
+
"epoch": 0.0001423,
|
| 9984 |
+
"grad_norm": 6.6436357498168945,
|
| 9985 |
+
"learning_rate": 1.4220000000000001e-05,
|
| 9986 |
+
"loss": 46.5938,
|
| 9987 |
+
"step": 1423
|
| 9988 |
+
},
|
| 9989 |
+
{
|
| 9990 |
+
"epoch": 0.0001424,
|
| 9991 |
+
"grad_norm": 6.492889404296875,
|
| 9992 |
+
"learning_rate": 1.423e-05,
|
| 9993 |
+
"loss": 46.4062,
|
| 9994 |
+
"step": 1424
|
| 9995 |
+
},
|
| 9996 |
+
{
|
| 9997 |
+
"epoch": 0.0001425,
|
| 9998 |
+
"grad_norm": 6.439519882202148,
|
| 9999 |
+
"learning_rate": 1.424e-05,
|
| 10000 |
+
"loss": 46.3125,
|
| 10001 |
+
"step": 1425
|
| 10002 |
+
},
|
| 10003 |
+
{
|
| 10004 |
+
"epoch": 0.0001426,
|
| 10005 |
+
"grad_norm": 6.359393119812012,
|
| 10006 |
+
"learning_rate": 1.425e-05,
|
| 10007 |
+
"loss": 46.4688,
|
| 10008 |
+
"step": 1426
|
| 10009 |
+
},
|
| 10010 |
+
{
|
| 10011 |
+
"epoch": 0.0001427,
|
| 10012 |
+
"grad_norm": 6.562929153442383,
|
| 10013 |
+
"learning_rate": 1.426e-05,
|
| 10014 |
+
"loss": 46.0312,
|
| 10015 |
+
"step": 1427
|
| 10016 |
+
},
|
| 10017 |
+
{
|
| 10018 |
+
"epoch": 0.0001428,
|
| 10019 |
+
"grad_norm": 6.442821025848389,
|
| 10020 |
+
"learning_rate": 1.427e-05,
|
| 10021 |
+
"loss": 45.4375,
|
| 10022 |
+
"step": 1428
|
| 10023 |
+
},
|
| 10024 |
+
{
|
| 10025 |
+
"epoch": 0.0001429,
|
| 10026 |
+
"grad_norm": 6.317464351654053,
|
| 10027 |
+
"learning_rate": 1.428e-05,
|
| 10028 |
+
"loss": 46.0625,
|
| 10029 |
+
"step": 1429
|
| 10030 |
+
},
|
| 10031 |
+
{
|
| 10032 |
+
"epoch": 0.000143,
|
| 10033 |
+
"grad_norm": 6.314460754394531,
|
| 10034 |
+
"learning_rate": 1.4289999999999999e-05,
|
| 10035 |
+
"loss": 45.9688,
|
| 10036 |
+
"step": 1430
|
| 10037 |
+
},
|
| 10038 |
+
{
|
| 10039 |
+
"epoch": 0.0001431,
|
| 10040 |
+
"grad_norm": 6.241281032562256,
|
| 10041 |
+
"learning_rate": 1.43e-05,
|
| 10042 |
+
"loss": 46.4688,
|
| 10043 |
+
"step": 1431
|
| 10044 |
+
},
|
| 10045 |
+
{
|
| 10046 |
+
"epoch": 0.0001432,
|
| 10047 |
+
"grad_norm": 6.5193963050842285,
|
| 10048 |
+
"learning_rate": 1.431e-05,
|
| 10049 |
+
"loss": 45.625,
|
| 10050 |
+
"step": 1432
|
| 10051 |
+
},
|
| 10052 |
+
{
|
| 10053 |
+
"epoch": 0.0001433,
|
| 10054 |
+
"grad_norm": 6.031685829162598,
|
| 10055 |
+
"learning_rate": 1.432e-05,
|
| 10056 |
+
"loss": 47.7188,
|
| 10057 |
+
"step": 1433
|
| 10058 |
+
},
|
| 10059 |
+
{
|
| 10060 |
+
"epoch": 0.0001434,
|
| 10061 |
+
"grad_norm": 6.39438533782959,
|
| 10062 |
+
"learning_rate": 1.4330000000000002e-05,
|
| 10063 |
+
"loss": 45.5312,
|
| 10064 |
+
"step": 1434
|
| 10065 |
+
},
|
| 10066 |
+
{
|
| 10067 |
+
"epoch": 0.0001435,
|
| 10068 |
+
"grad_norm": 6.2862324714660645,
|
| 10069 |
+
"learning_rate": 1.4340000000000002e-05,
|
| 10070 |
+
"loss": 46.125,
|
| 10071 |
+
"step": 1435
|
| 10072 |
+
},
|
| 10073 |
+
{
|
| 10074 |
+
"epoch": 0.0001436,
|
| 10075 |
+
"grad_norm": 6.472970485687256,
|
| 10076 |
+
"learning_rate": 1.435e-05,
|
| 10077 |
+
"loss": 45.75,
|
| 10078 |
+
"step": 1436
|
| 10079 |
+
},
|
| 10080 |
+
{
|
| 10081 |
+
"epoch": 0.0001437,
|
| 10082 |
+
"grad_norm": 6.28604793548584,
|
| 10083 |
+
"learning_rate": 1.436e-05,
|
| 10084 |
+
"loss": 46.0,
|
| 10085 |
+
"step": 1437
|
| 10086 |
+
},
|
| 10087 |
+
{
|
| 10088 |
+
"epoch": 0.0001438,
|
| 10089 |
+
"grad_norm": 6.293455600738525,
|
| 10090 |
+
"learning_rate": 1.437e-05,
|
| 10091 |
+
"loss": 47.375,
|
| 10092 |
+
"step": 1438
|
| 10093 |
+
},
|
| 10094 |
+
{
|
| 10095 |
+
"epoch": 0.0001439,
|
| 10096 |
+
"grad_norm": 6.2602057456970215,
|
| 10097 |
+
"learning_rate": 1.4380000000000001e-05,
|
| 10098 |
+
"loss": 46.25,
|
| 10099 |
+
"step": 1439
|
| 10100 |
+
},
|
| 10101 |
+
{
|
| 10102 |
+
"epoch": 0.000144,
|
| 10103 |
+
"grad_norm": 6.504815101623535,
|
| 10104 |
+
"learning_rate": 1.4390000000000001e-05,
|
| 10105 |
+
"loss": 45.5938,
|
| 10106 |
+
"step": 1440
|
| 10107 |
+
},
|
| 10108 |
+
{
|
| 10109 |
+
"epoch": 0.0001441,
|
| 10110 |
+
"grad_norm": 6.3342366218566895,
|
| 10111 |
+
"learning_rate": 1.44e-05,
|
| 10112 |
+
"loss": 45.5,
|
| 10113 |
+
"step": 1441
|
| 10114 |
+
},
|
| 10115 |
+
{
|
| 10116 |
+
"epoch": 0.0001442,
|
| 10117 |
+
"grad_norm": 6.301247596740723,
|
| 10118 |
+
"learning_rate": 1.441e-05,
|
| 10119 |
+
"loss": 45.9688,
|
| 10120 |
+
"step": 1442
|
| 10121 |
+
},
|
| 10122 |
+
{
|
| 10123 |
+
"epoch": 0.0001443,
|
| 10124 |
+
"grad_norm": 6.273526191711426,
|
| 10125 |
+
"learning_rate": 1.442e-05,
|
| 10126 |
+
"loss": 46.0312,
|
| 10127 |
+
"step": 1443
|
| 10128 |
+
},
|
| 10129 |
+
{
|
| 10130 |
+
"epoch": 0.0001444,
|
| 10131 |
+
"grad_norm": 6.2740678787231445,
|
| 10132 |
+
"learning_rate": 1.443e-05,
|
| 10133 |
+
"loss": 47.0,
|
| 10134 |
+
"step": 1444
|
| 10135 |
+
},
|
| 10136 |
+
{
|
| 10137 |
+
"epoch": 0.0001445,
|
| 10138 |
+
"grad_norm": 6.325813293457031,
|
| 10139 |
+
"learning_rate": 1.444e-05,
|
| 10140 |
+
"loss": 45.0938,
|
| 10141 |
+
"step": 1445
|
| 10142 |
+
},
|
| 10143 |
+
{
|
| 10144 |
+
"epoch": 0.0001446,
|
| 10145 |
+
"grad_norm": 6.318421363830566,
|
| 10146 |
+
"learning_rate": 1.445e-05,
|
| 10147 |
+
"loss": 46.4688,
|
| 10148 |
+
"step": 1446
|
| 10149 |
+
},
|
| 10150 |
+
{
|
| 10151 |
+
"epoch": 0.0001447,
|
| 10152 |
+
"grad_norm": 6.3151445388793945,
|
| 10153 |
+
"learning_rate": 1.4459999999999999e-05,
|
| 10154 |
+
"loss": 45.75,
|
| 10155 |
+
"step": 1447
|
| 10156 |
+
},
|
| 10157 |
+
{
|
| 10158 |
+
"epoch": 0.0001448,
|
| 10159 |
+
"grad_norm": 6.330325126647949,
|
| 10160 |
+
"learning_rate": 1.447e-05,
|
| 10161 |
+
"loss": 45.7188,
|
| 10162 |
+
"step": 1448
|
| 10163 |
+
},
|
| 10164 |
+
{
|
| 10165 |
+
"epoch": 0.0001449,
|
| 10166 |
+
"grad_norm": 6.304328918457031,
|
| 10167 |
+
"learning_rate": 1.448e-05,
|
| 10168 |
+
"loss": 46.5938,
|
| 10169 |
+
"step": 1449
|
| 10170 |
+
},
|
| 10171 |
+
{
|
| 10172 |
+
"epoch": 0.000145,
|
| 10173 |
+
"grad_norm": 6.322455883026123,
|
| 10174 |
+
"learning_rate": 1.449e-05,
|
| 10175 |
+
"loss": 45.7812,
|
| 10176 |
+
"step": 1450
|
| 10177 |
+
},
|
| 10178 |
+
{
|
| 10179 |
+
"epoch": 0.0001451,
|
| 10180 |
+
"grad_norm": 6.178551197052002,
|
| 10181 |
+
"learning_rate": 1.4500000000000002e-05,
|
| 10182 |
+
"loss": 46.4062,
|
| 10183 |
+
"step": 1451
|
| 10184 |
+
},
|
| 10185 |
+
{
|
| 10186 |
+
"epoch": 0.0001452,
|
| 10187 |
+
"grad_norm": 6.315935134887695,
|
| 10188 |
+
"learning_rate": 1.4510000000000002e-05,
|
| 10189 |
+
"loss": 45.2812,
|
| 10190 |
+
"step": 1452
|
| 10191 |
+
},
|
| 10192 |
+
{
|
| 10193 |
+
"epoch": 0.0001453,
|
| 10194 |
+
"grad_norm": 6.263783931732178,
|
| 10195 |
+
"learning_rate": 1.452e-05,
|
| 10196 |
+
"loss": 45.0312,
|
| 10197 |
+
"step": 1453
|
| 10198 |
+
},
|
| 10199 |
+
{
|
| 10200 |
+
"epoch": 0.0001454,
|
| 10201 |
+
"grad_norm": 6.470102787017822,
|
| 10202 |
+
"learning_rate": 1.453e-05,
|
| 10203 |
+
"loss": 45.1562,
|
| 10204 |
+
"step": 1454
|
| 10205 |
+
},
|
| 10206 |
+
{
|
| 10207 |
+
"epoch": 0.0001455,
|
| 10208 |
+
"grad_norm": 6.220240592956543,
|
| 10209 |
+
"learning_rate": 1.454e-05,
|
| 10210 |
+
"loss": 45.8438,
|
| 10211 |
+
"step": 1455
|
| 10212 |
+
},
|
| 10213 |
+
{
|
| 10214 |
+
"epoch": 0.0001456,
|
| 10215 |
+
"grad_norm": 6.216402530670166,
|
| 10216 |
+
"learning_rate": 1.4550000000000001e-05,
|
| 10217 |
+
"loss": 45.6875,
|
| 10218 |
+
"step": 1456
|
| 10219 |
+
},
|
| 10220 |
+
{
|
| 10221 |
+
"epoch": 0.0001457,
|
| 10222 |
+
"grad_norm": 6.349188327789307,
|
| 10223 |
+
"learning_rate": 1.4560000000000001e-05,
|
| 10224 |
+
"loss": 44.9375,
|
| 10225 |
+
"step": 1457
|
| 10226 |
+
},
|
| 10227 |
+
{
|
| 10228 |
+
"epoch": 0.0001458,
|
| 10229 |
+
"grad_norm": 6.409142017364502,
|
| 10230 |
+
"learning_rate": 1.457e-05,
|
| 10231 |
+
"loss": 44.7188,
|
| 10232 |
+
"step": 1458
|
| 10233 |
+
},
|
| 10234 |
+
{
|
| 10235 |
+
"epoch": 0.0001459,
|
| 10236 |
+
"grad_norm": 6.316352367401123,
|
| 10237 |
+
"learning_rate": 1.458e-05,
|
| 10238 |
+
"loss": 45.5625,
|
| 10239 |
+
"step": 1459
|
| 10240 |
+
},
|
| 10241 |
+
{
|
| 10242 |
+
"epoch": 0.000146,
|
| 10243 |
+
"grad_norm": 6.413899898529053,
|
| 10244 |
+
"learning_rate": 1.459e-05,
|
| 10245 |
+
"loss": 44.75,
|
| 10246 |
+
"step": 1460
|
| 10247 |
+
},
|
| 10248 |
+
{
|
| 10249 |
+
"epoch": 0.0001461,
|
| 10250 |
+
"grad_norm": 6.37227201461792,
|
| 10251 |
+
"learning_rate": 1.46e-05,
|
| 10252 |
+
"loss": 44.9062,
|
| 10253 |
+
"step": 1461
|
| 10254 |
+
},
|
| 10255 |
+
{
|
| 10256 |
+
"epoch": 0.0001462,
|
| 10257 |
+
"grad_norm": 6.28816032409668,
|
| 10258 |
+
"learning_rate": 1.461e-05,
|
| 10259 |
+
"loss": 45.0,
|
| 10260 |
+
"step": 1462
|
| 10261 |
+
},
|
| 10262 |
+
{
|
| 10263 |
+
"epoch": 0.0001463,
|
| 10264 |
+
"grad_norm": 6.313782215118408,
|
| 10265 |
+
"learning_rate": 1.462e-05,
|
| 10266 |
+
"loss": 44.9062,
|
| 10267 |
+
"step": 1463
|
| 10268 |
+
},
|
| 10269 |
+
{
|
| 10270 |
+
"epoch": 0.0001464,
|
| 10271 |
+
"grad_norm": 6.226417064666748,
|
| 10272 |
+
"learning_rate": 1.4629999999999999e-05,
|
| 10273 |
+
"loss": 45.5,
|
| 10274 |
+
"step": 1464
|
| 10275 |
+
},
|
| 10276 |
+
{
|
| 10277 |
+
"epoch": 0.0001465,
|
| 10278 |
+
"grad_norm": 6.16673469543457,
|
| 10279 |
+
"learning_rate": 1.464e-05,
|
| 10280 |
+
"loss": 46.0938,
|
| 10281 |
+
"step": 1465
|
| 10282 |
+
},
|
| 10283 |
+
{
|
| 10284 |
+
"epoch": 0.0001466,
|
| 10285 |
+
"grad_norm": 6.2313361167907715,
|
| 10286 |
+
"learning_rate": 1.465e-05,
|
| 10287 |
+
"loss": 45.375,
|
| 10288 |
+
"step": 1466
|
| 10289 |
+
},
|
| 10290 |
+
{
|
| 10291 |
+
"epoch": 0.0001467,
|
| 10292 |
+
"grad_norm": 6.184562683105469,
|
| 10293 |
+
"learning_rate": 1.466e-05,
|
| 10294 |
+
"loss": 45.0938,
|
| 10295 |
+
"step": 1467
|
| 10296 |
+
},
|
| 10297 |
+
{
|
| 10298 |
+
"epoch": 0.0001468,
|
| 10299 |
+
"grad_norm": 6.402483940124512,
|
| 10300 |
+
"learning_rate": 1.4670000000000002e-05,
|
| 10301 |
+
"loss": 44.6875,
|
| 10302 |
+
"step": 1468
|
| 10303 |
+
},
|
| 10304 |
+
{
|
| 10305 |
+
"epoch": 0.0001469,
|
| 10306 |
+
"grad_norm": 6.2219624519348145,
|
| 10307 |
+
"learning_rate": 1.4680000000000002e-05,
|
| 10308 |
+
"loss": 45.875,
|
| 10309 |
+
"step": 1469
|
| 10310 |
+
},
|
| 10311 |
+
{
|
| 10312 |
+
"epoch": 0.000147,
|
| 10313 |
+
"grad_norm": 6.251800537109375,
|
| 10314 |
+
"learning_rate": 1.469e-05,
|
| 10315 |
+
"loss": 44.7188,
|
| 10316 |
+
"step": 1470
|
| 10317 |
+
},
|
| 10318 |
+
{
|
| 10319 |
+
"epoch": 0.0001471,
|
| 10320 |
+
"grad_norm": 6.082367897033691,
|
| 10321 |
+
"learning_rate": 1.47e-05,
|
| 10322 |
+
"loss": 45.2812,
|
| 10323 |
+
"step": 1471
|
| 10324 |
+
},
|
| 10325 |
+
{
|
| 10326 |
+
"epoch": 0.0001472,
|
| 10327 |
+
"grad_norm": 6.209146499633789,
|
| 10328 |
+
"learning_rate": 1.471e-05,
|
| 10329 |
+
"loss": 45.2812,
|
| 10330 |
+
"step": 1472
|
| 10331 |
+
},
|
| 10332 |
+
{
|
| 10333 |
+
"epoch": 0.0001473,
|
| 10334 |
+
"grad_norm": 6.131628036499023,
|
| 10335 |
+
"learning_rate": 1.4720000000000001e-05,
|
| 10336 |
+
"loss": 45.6562,
|
| 10337 |
+
"step": 1473
|
| 10338 |
+
},
|
| 10339 |
+
{
|
| 10340 |
+
"epoch": 0.0001474,
|
| 10341 |
+
"grad_norm": 6.09556245803833,
|
| 10342 |
+
"learning_rate": 1.4730000000000001e-05,
|
| 10343 |
+
"loss": 46.0,
|
| 10344 |
+
"step": 1474
|
| 10345 |
+
},
|
| 10346 |
+
{
|
| 10347 |
+
"epoch": 0.0001475,
|
| 10348 |
+
"grad_norm": 6.235780239105225,
|
| 10349 |
+
"learning_rate": 1.474e-05,
|
| 10350 |
+
"loss": 44.3125,
|
| 10351 |
+
"step": 1475
|
| 10352 |
+
},
|
| 10353 |
+
{
|
| 10354 |
+
"epoch": 0.0001476,
|
| 10355 |
+
"grad_norm": 6.3033037185668945,
|
| 10356 |
+
"learning_rate": 1.475e-05,
|
| 10357 |
+
"loss": 44.9062,
|
| 10358 |
+
"step": 1476
|
| 10359 |
+
},
|
| 10360 |
+
{
|
| 10361 |
+
"epoch": 0.0001477,
|
| 10362 |
+
"grad_norm": 6.262908458709717,
|
| 10363 |
+
"learning_rate": 1.476e-05,
|
| 10364 |
+
"loss": 44.7812,
|
| 10365 |
+
"step": 1477
|
| 10366 |
+
},
|
| 10367 |
+
{
|
| 10368 |
+
"epoch": 0.0001478,
|
| 10369 |
+
"grad_norm": 6.337582588195801,
|
| 10370 |
+
"learning_rate": 1.4770000000000001e-05,
|
| 10371 |
+
"loss": 44.25,
|
| 10372 |
+
"step": 1478
|
| 10373 |
+
},
|
| 10374 |
+
{
|
| 10375 |
+
"epoch": 0.0001479,
|
| 10376 |
+
"grad_norm": 6.241846561431885,
|
| 10377 |
+
"learning_rate": 1.478e-05,
|
| 10378 |
+
"loss": 44.875,
|
| 10379 |
+
"step": 1479
|
| 10380 |
+
},
|
| 10381 |
+
{
|
| 10382 |
+
"epoch": 0.000148,
|
| 10383 |
+
"grad_norm": 6.219179153442383,
|
| 10384 |
+
"learning_rate": 1.479e-05,
|
| 10385 |
+
"loss": 44.8125,
|
| 10386 |
+
"step": 1480
|
| 10387 |
+
},
|
| 10388 |
+
{
|
| 10389 |
+
"epoch": 0.0001481,
|
| 10390 |
+
"grad_norm": 6.218071937561035,
|
| 10391 |
+
"learning_rate": 1.4799999999999999e-05,
|
| 10392 |
+
"loss": 45.2188,
|
| 10393 |
+
"step": 1481
|
| 10394 |
+
},
|
| 10395 |
+
{
|
| 10396 |
+
"epoch": 0.0001482,
|
| 10397 |
+
"grad_norm": 6.311007022857666,
|
| 10398 |
+
"learning_rate": 1.481e-05,
|
| 10399 |
+
"loss": 44.625,
|
| 10400 |
+
"step": 1482
|
| 10401 |
+
},
|
| 10402 |
+
{
|
| 10403 |
+
"epoch": 0.0001483,
|
| 10404 |
+
"grad_norm": 6.308942794799805,
|
| 10405 |
+
"learning_rate": 1.482e-05,
|
| 10406 |
+
"loss": 44.6875,
|
| 10407 |
+
"step": 1483
|
| 10408 |
+
},
|
| 10409 |
+
{
|
| 10410 |
+
"epoch": 0.0001484,
|
| 10411 |
+
"grad_norm": 6.2515339851379395,
|
| 10412 |
+
"learning_rate": 1.483e-05,
|
| 10413 |
+
"loss": 44.1875,
|
| 10414 |
+
"step": 1484
|
| 10415 |
+
},
|
| 10416 |
+
{
|
| 10417 |
+
"epoch": 0.0001485,
|
| 10418 |
+
"grad_norm": 6.152646541595459,
|
| 10419 |
+
"learning_rate": 1.4840000000000002e-05,
|
| 10420 |
+
"loss": 44.9062,
|
| 10421 |
+
"step": 1485
|
| 10422 |
+
},
|
| 10423 |
+
{
|
| 10424 |
+
"epoch": 0.0001486,
|
| 10425 |
+
"grad_norm": 6.309966087341309,
|
| 10426 |
+
"learning_rate": 1.4850000000000002e-05,
|
| 10427 |
+
"loss": 44.3125,
|
| 10428 |
+
"step": 1486
|
| 10429 |
+
},
|
| 10430 |
+
{
|
| 10431 |
+
"epoch": 0.0001487,
|
| 10432 |
+
"grad_norm": 6.248706340789795,
|
| 10433 |
+
"learning_rate": 1.486e-05,
|
| 10434 |
+
"loss": 44.4062,
|
| 10435 |
+
"step": 1487
|
| 10436 |
+
},
|
| 10437 |
+
{
|
| 10438 |
+
"epoch": 0.0001488,
|
| 10439 |
+
"grad_norm": 6.111617565155029,
|
| 10440 |
+
"learning_rate": 1.487e-05,
|
| 10441 |
+
"loss": 44.875,
|
| 10442 |
+
"step": 1488
|
| 10443 |
+
},
|
| 10444 |
+
{
|
| 10445 |
+
"epoch": 0.0001489,
|
| 10446 |
+
"grad_norm": 6.138513565063477,
|
| 10447 |
+
"learning_rate": 1.488e-05,
|
| 10448 |
+
"loss": 45.375,
|
| 10449 |
+
"step": 1489
|
| 10450 |
+
},
|
| 10451 |
+
{
|
| 10452 |
+
"epoch": 0.000149,
|
| 10453 |
+
"grad_norm": 6.269906520843506,
|
| 10454 |
+
"learning_rate": 1.4890000000000001e-05,
|
| 10455 |
+
"loss": 44.3125,
|
| 10456 |
+
"step": 1490
|
| 10457 |
+
},
|
| 10458 |
+
{
|
| 10459 |
+
"epoch": 0.0001491,
|
| 10460 |
+
"grad_norm": 6.300713062286377,
|
| 10461 |
+
"learning_rate": 1.4900000000000001e-05,
|
| 10462 |
+
"loss": 44.3125,
|
| 10463 |
+
"step": 1491
|
| 10464 |
+
},
|
| 10465 |
+
{
|
| 10466 |
+
"epoch": 0.0001492,
|
| 10467 |
+
"grad_norm": 6.14302921295166,
|
| 10468 |
+
"learning_rate": 1.491e-05,
|
| 10469 |
+
"loss": 45.0,
|
| 10470 |
+
"step": 1492
|
| 10471 |
+
},
|
| 10472 |
+
{
|
| 10473 |
+
"epoch": 0.0001493,
|
| 10474 |
+
"grad_norm": 6.299230575561523,
|
| 10475 |
+
"learning_rate": 1.492e-05,
|
| 10476 |
+
"loss": 44.625,
|
| 10477 |
+
"step": 1493
|
| 10478 |
+
},
|
| 10479 |
+
{
|
| 10480 |
+
"epoch": 0.0001494,
|
| 10481 |
+
"grad_norm": 6.294898509979248,
|
| 10482 |
+
"learning_rate": 1.493e-05,
|
| 10483 |
+
"loss": 43.7188,
|
| 10484 |
+
"step": 1494
|
| 10485 |
+
},
|
| 10486 |
+
{
|
| 10487 |
+
"epoch": 0.0001495,
|
| 10488 |
+
"grad_norm": 6.1497063636779785,
|
| 10489 |
+
"learning_rate": 1.4940000000000001e-05,
|
| 10490 |
+
"loss": 45.25,
|
| 10491 |
+
"step": 1495
|
| 10492 |
+
},
|
| 10493 |
+
{
|
| 10494 |
+
"epoch": 0.0001496,
|
| 10495 |
+
"grad_norm": 6.63415002822876,
|
| 10496 |
+
"learning_rate": 1.4950000000000001e-05,
|
| 10497 |
+
"loss": 44.8438,
|
| 10498 |
+
"step": 1496
|
| 10499 |
+
},
|
| 10500 |
+
{
|
| 10501 |
+
"epoch": 0.0001497,
|
| 10502 |
+
"grad_norm": 6.380233287811279,
|
| 10503 |
+
"learning_rate": 1.496e-05,
|
| 10504 |
+
"loss": 43.7812,
|
| 10505 |
+
"step": 1497
|
| 10506 |
+
},
|
| 10507 |
+
{
|
| 10508 |
+
"epoch": 0.0001498,
|
| 10509 |
+
"grad_norm": 6.056626319885254,
|
| 10510 |
+
"learning_rate": 1.4969999999999999e-05,
|
| 10511 |
+
"loss": 44.8125,
|
| 10512 |
+
"step": 1498
|
| 10513 |
+
},
|
| 10514 |
+
{
|
| 10515 |
+
"epoch": 0.0001499,
|
| 10516 |
+
"grad_norm": 6.122725486755371,
|
| 10517 |
+
"learning_rate": 1.4979999999999999e-05,
|
| 10518 |
+
"loss": 44.9062,
|
| 10519 |
+
"step": 1499
|
| 10520 |
+
},
|
| 10521 |
+
{
|
| 10522 |
+
"epoch": 0.00015,
|
| 10523 |
+
"grad_norm": 6.298846244812012,
|
| 10524 |
+
"learning_rate": 1.499e-05,
|
| 10525 |
+
"loss": 43.6875,
|
| 10526 |
+
"step": 1500
|
| 10527 |
+
},
|
| 10528 |
+
{
|
| 10529 |
+
"epoch": 0.00015,
|
| 10530 |
+
"eval_loss": 5.494104862213135,
|
| 10531 |
+
"eval_runtime": 364.1692,
|
| 10532 |
+
"eval_samples_per_second": 27.46,
|
| 10533 |
+
"eval_steps_per_second": 1.716,
|
| 10534 |
+
"step": 1500
|
| 10535 |
}
|
| 10536 |
],
|
| 10537 |
"logging_steps": 1,
|