Training in progress, step 1000, 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:cc6b44951f71e38932dd738d841d6f489b985f24e3c3523e2a7adb893f8fdc5d
|
| 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:c02cb680c39a21e145a8cbcca42081ce685165e39759570b73fb2f82a2e28f37
|
| 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 sha256:9196a1e708bf24d6abba41cce3f8558820acc3e50f9394c5955e29eb41ffea3d
|
| 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:cc0ead927c2d87154ea1c8600519341bfb2c3090b1e4a3776c279253dd2e14c0
|
| 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":
|
| 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,
|
|
@@ -3516,6 +3516,3514 @@
|
|
| 3516 |
"eval_samples_per_second": 27.452,
|
| 3517 |
"eval_steps_per_second": 1.716,
|
| 3518 |
"step": 500
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3519 |
}
|
| 3520 |
],
|
| 3521 |
"logging_steps": 1,
|
|
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.0001,
|
| 6 |
"eval_steps": 500,
|
| 7 |
+
"global_step": 1000,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 3516 |
"eval_samples_per_second": 27.452,
|
| 3517 |
"eval_steps_per_second": 1.716,
|
| 3518 |
"step": 500
|
| 3519 |
+
},
|
| 3520 |
+
{
|
| 3521 |
+
"epoch": 5.01e-05,
|
| 3522 |
+
"grad_norm": 12.755992889404297,
|
| 3523 |
+
"learning_rate": 5e-06,
|
| 3524 |
+
"loss": 72.125,
|
| 3525 |
+
"step": 501
|
| 3526 |
+
},
|
| 3527 |
+
{
|
| 3528 |
+
"epoch": 5.02e-05,
|
| 3529 |
+
"grad_norm": 13.567408561706543,
|
| 3530 |
+
"learning_rate": 5.01e-06,
|
| 3531 |
+
"loss": 72.1875,
|
| 3532 |
+
"step": 502
|
| 3533 |
+
},
|
| 3534 |
+
{
|
| 3535 |
+
"epoch": 5.03e-05,
|
| 3536 |
+
"grad_norm": 13.13305377960205,
|
| 3537 |
+
"learning_rate": 5.02e-06,
|
| 3538 |
+
"loss": 72.125,
|
| 3539 |
+
"step": 503
|
| 3540 |
+
},
|
| 3541 |
+
{
|
| 3542 |
+
"epoch": 5.04e-05,
|
| 3543 |
+
"grad_norm": 13.306023597717285,
|
| 3544 |
+
"learning_rate": 5.03e-06,
|
| 3545 |
+
"loss": 72.0625,
|
| 3546 |
+
"step": 504
|
| 3547 |
+
},
|
| 3548 |
+
{
|
| 3549 |
+
"epoch": 5.05e-05,
|
| 3550 |
+
"grad_norm": 13.41368579864502,
|
| 3551 |
+
"learning_rate": 5.04e-06,
|
| 3552 |
+
"loss": 72.1875,
|
| 3553 |
+
"step": 505
|
| 3554 |
+
},
|
| 3555 |
+
{
|
| 3556 |
+
"epoch": 5.06e-05,
|
| 3557 |
+
"grad_norm": 13.642829895019531,
|
| 3558 |
+
"learning_rate": 5.05e-06,
|
| 3559 |
+
"loss": 72.0625,
|
| 3560 |
+
"step": 506
|
| 3561 |
+
},
|
| 3562 |
+
{
|
| 3563 |
+
"epoch": 5.07e-05,
|
| 3564 |
+
"grad_norm": 13.747808456420898,
|
| 3565 |
+
"learning_rate": 5.06e-06,
|
| 3566 |
+
"loss": 72.0625,
|
| 3567 |
+
"step": 507
|
| 3568 |
+
},
|
| 3569 |
+
{
|
| 3570 |
+
"epoch": 5.08e-05,
|
| 3571 |
+
"grad_norm": 13.421524047851562,
|
| 3572 |
+
"learning_rate": 5.070000000000001e-06,
|
| 3573 |
+
"loss": 72.125,
|
| 3574 |
+
"step": 508
|
| 3575 |
+
},
|
| 3576 |
+
{
|
| 3577 |
+
"epoch": 5.09e-05,
|
| 3578 |
+
"grad_norm": 13.79678726196289,
|
| 3579 |
+
"learning_rate": 5.0800000000000005e-06,
|
| 3580 |
+
"loss": 72.0,
|
| 3581 |
+
"step": 509
|
| 3582 |
+
},
|
| 3583 |
+
{
|
| 3584 |
+
"epoch": 5.1e-05,
|
| 3585 |
+
"grad_norm": 13.362953186035156,
|
| 3586 |
+
"learning_rate": 5.09e-06,
|
| 3587 |
+
"loss": 72.0,
|
| 3588 |
+
"step": 510
|
| 3589 |
+
},
|
| 3590 |
+
{
|
| 3591 |
+
"epoch": 5.11e-05,
|
| 3592 |
+
"grad_norm": 13.886173248291016,
|
| 3593 |
+
"learning_rate": 5.0999999999999995e-06,
|
| 3594 |
+
"loss": 72.0,
|
| 3595 |
+
"step": 511
|
| 3596 |
+
},
|
| 3597 |
+
{
|
| 3598 |
+
"epoch": 5.12e-05,
|
| 3599 |
+
"grad_norm": 13.49032211303711,
|
| 3600 |
+
"learning_rate": 5.11e-06,
|
| 3601 |
+
"loss": 72.0625,
|
| 3602 |
+
"step": 512
|
| 3603 |
+
},
|
| 3604 |
+
{
|
| 3605 |
+
"epoch": 5.13e-05,
|
| 3606 |
+
"grad_norm": 13.661921501159668,
|
| 3607 |
+
"learning_rate": 5.12e-06,
|
| 3608 |
+
"loss": 72.0625,
|
| 3609 |
+
"step": 513
|
| 3610 |
+
},
|
| 3611 |
+
{
|
| 3612 |
+
"epoch": 5.14e-05,
|
| 3613 |
+
"grad_norm": 13.80624771118164,
|
| 3614 |
+
"learning_rate": 5.13e-06,
|
| 3615 |
+
"loss": 72.0625,
|
| 3616 |
+
"step": 514
|
| 3617 |
+
},
|
| 3618 |
+
{
|
| 3619 |
+
"epoch": 5.15e-05,
|
| 3620 |
+
"grad_norm": 13.84620475769043,
|
| 3621 |
+
"learning_rate": 5.140000000000001e-06,
|
| 3622 |
+
"loss": 72.0,
|
| 3623 |
+
"step": 515
|
| 3624 |
+
},
|
| 3625 |
+
{
|
| 3626 |
+
"epoch": 5.16e-05,
|
| 3627 |
+
"grad_norm": 13.160120010375977,
|
| 3628 |
+
"learning_rate": 5.15e-06,
|
| 3629 |
+
"loss": 72.0625,
|
| 3630 |
+
"step": 516
|
| 3631 |
+
},
|
| 3632 |
+
{
|
| 3633 |
+
"epoch": 5.17e-05,
|
| 3634 |
+
"grad_norm": 13.24973201751709,
|
| 3635 |
+
"learning_rate": 5.16e-06,
|
| 3636 |
+
"loss": 72.125,
|
| 3637 |
+
"step": 517
|
| 3638 |
+
},
|
| 3639 |
+
{
|
| 3640 |
+
"epoch": 5.18e-05,
|
| 3641 |
+
"grad_norm": 14.063905715942383,
|
| 3642 |
+
"learning_rate": 5.17e-06,
|
| 3643 |
+
"loss": 72.0,
|
| 3644 |
+
"step": 518
|
| 3645 |
+
},
|
| 3646 |
+
{
|
| 3647 |
+
"epoch": 5.19e-05,
|
| 3648 |
+
"grad_norm": 13.477499008178711,
|
| 3649 |
+
"learning_rate": 5.18e-06,
|
| 3650 |
+
"loss": 72.0,
|
| 3651 |
+
"step": 519
|
| 3652 |
+
},
|
| 3653 |
+
{
|
| 3654 |
+
"epoch": 5.2e-05,
|
| 3655 |
+
"grad_norm": 13.442255020141602,
|
| 3656 |
+
"learning_rate": 5.19e-06,
|
| 3657 |
+
"loss": 72.0,
|
| 3658 |
+
"step": 520
|
| 3659 |
+
},
|
| 3660 |
+
{
|
| 3661 |
+
"epoch": 5.21e-05,
|
| 3662 |
+
"grad_norm": 13.900424003601074,
|
| 3663 |
+
"learning_rate": 5.2e-06,
|
| 3664 |
+
"loss": 72.0,
|
| 3665 |
+
"step": 521
|
| 3666 |
+
},
|
| 3667 |
+
{
|
| 3668 |
+
"epoch": 5.22e-05,
|
| 3669 |
+
"grad_norm": 14.005378723144531,
|
| 3670 |
+
"learning_rate": 5.21e-06,
|
| 3671 |
+
"loss": 72.0,
|
| 3672 |
+
"step": 522
|
| 3673 |
+
},
|
| 3674 |
+
{
|
| 3675 |
+
"epoch": 5.23e-05,
|
| 3676 |
+
"grad_norm": 13.34553337097168,
|
| 3677 |
+
"learning_rate": 5.22e-06,
|
| 3678 |
+
"loss": 71.9375,
|
| 3679 |
+
"step": 523
|
| 3680 |
+
},
|
| 3681 |
+
{
|
| 3682 |
+
"epoch": 5.24e-05,
|
| 3683 |
+
"grad_norm": 13.30509090423584,
|
| 3684 |
+
"learning_rate": 5.23e-06,
|
| 3685 |
+
"loss": 71.9375,
|
| 3686 |
+
"step": 524
|
| 3687 |
+
},
|
| 3688 |
+
{
|
| 3689 |
+
"epoch": 5.25e-05,
|
| 3690 |
+
"grad_norm": 13.428346633911133,
|
| 3691 |
+
"learning_rate": 5.240000000000001e-06,
|
| 3692 |
+
"loss": 72.0,
|
| 3693 |
+
"step": 525
|
| 3694 |
+
},
|
| 3695 |
+
{
|
| 3696 |
+
"epoch": 5.26e-05,
|
| 3697 |
+
"grad_norm": 13.481391906738281,
|
| 3698 |
+
"learning_rate": 5.2500000000000006e-06,
|
| 3699 |
+
"loss": 72.0,
|
| 3700 |
+
"step": 526
|
| 3701 |
+
},
|
| 3702 |
+
{
|
| 3703 |
+
"epoch": 5.27e-05,
|
| 3704 |
+
"grad_norm": 13.900728225708008,
|
| 3705 |
+
"learning_rate": 5.26e-06,
|
| 3706 |
+
"loss": 72.0,
|
| 3707 |
+
"step": 527
|
| 3708 |
+
},
|
| 3709 |
+
{
|
| 3710 |
+
"epoch": 5.28e-05,
|
| 3711 |
+
"grad_norm": 12.912371635437012,
|
| 3712 |
+
"learning_rate": 5.2699999999999995e-06,
|
| 3713 |
+
"loss": 72.0,
|
| 3714 |
+
"step": 528
|
| 3715 |
+
},
|
| 3716 |
+
{
|
| 3717 |
+
"epoch": 5.29e-05,
|
| 3718 |
+
"grad_norm": 13.814693450927734,
|
| 3719 |
+
"learning_rate": 5.28e-06,
|
| 3720 |
+
"loss": 72.0,
|
| 3721 |
+
"step": 529
|
| 3722 |
+
},
|
| 3723 |
+
{
|
| 3724 |
+
"epoch": 5.3e-05,
|
| 3725 |
+
"grad_norm": 13.151686668395996,
|
| 3726 |
+
"learning_rate": 5.29e-06,
|
| 3727 |
+
"loss": 72.0,
|
| 3728 |
+
"step": 530
|
| 3729 |
+
},
|
| 3730 |
+
{
|
| 3731 |
+
"epoch": 5.31e-05,
|
| 3732 |
+
"grad_norm": 13.657590866088867,
|
| 3733 |
+
"learning_rate": 5.3e-06,
|
| 3734 |
+
"loss": 71.9375,
|
| 3735 |
+
"step": 531
|
| 3736 |
+
},
|
| 3737 |
+
{
|
| 3738 |
+
"epoch": 5.32e-05,
|
| 3739 |
+
"grad_norm": 13.253593444824219,
|
| 3740 |
+
"learning_rate": 5.31e-06,
|
| 3741 |
+
"loss": 71.9375,
|
| 3742 |
+
"step": 532
|
| 3743 |
+
},
|
| 3744 |
+
{
|
| 3745 |
+
"epoch": 5.33e-05,
|
| 3746 |
+
"grad_norm": 13.740532875061035,
|
| 3747 |
+
"learning_rate": 5.32e-06,
|
| 3748 |
+
"loss": 71.9375,
|
| 3749 |
+
"step": 533
|
| 3750 |
+
},
|
| 3751 |
+
{
|
| 3752 |
+
"epoch": 5.34e-05,
|
| 3753 |
+
"grad_norm": 14.07413387298584,
|
| 3754 |
+
"learning_rate": 5.33e-06,
|
| 3755 |
+
"loss": 71.875,
|
| 3756 |
+
"step": 534
|
| 3757 |
+
},
|
| 3758 |
+
{
|
| 3759 |
+
"epoch": 5.35e-05,
|
| 3760 |
+
"grad_norm": 13.681927680969238,
|
| 3761 |
+
"learning_rate": 5.34e-06,
|
| 3762 |
+
"loss": 72.0,
|
| 3763 |
+
"step": 535
|
| 3764 |
+
},
|
| 3765 |
+
{
|
| 3766 |
+
"epoch": 5.36e-05,
|
| 3767 |
+
"grad_norm": 13.324553489685059,
|
| 3768 |
+
"learning_rate": 5.3500000000000004e-06,
|
| 3769 |
+
"loss": 71.9375,
|
| 3770 |
+
"step": 536
|
| 3771 |
+
},
|
| 3772 |
+
{
|
| 3773 |
+
"epoch": 5.37e-05,
|
| 3774 |
+
"grad_norm": 13.417908668518066,
|
| 3775 |
+
"learning_rate": 5.36e-06,
|
| 3776 |
+
"loss": 71.9375,
|
| 3777 |
+
"step": 537
|
| 3778 |
+
},
|
| 3779 |
+
{
|
| 3780 |
+
"epoch": 5.38e-05,
|
| 3781 |
+
"grad_norm": 12.827329635620117,
|
| 3782 |
+
"learning_rate": 5.37e-06,
|
| 3783 |
+
"loss": 71.9375,
|
| 3784 |
+
"step": 538
|
| 3785 |
+
},
|
| 3786 |
+
{
|
| 3787 |
+
"epoch": 5.39e-05,
|
| 3788 |
+
"grad_norm": 13.676140785217285,
|
| 3789 |
+
"learning_rate": 5.38e-06,
|
| 3790 |
+
"loss": 71.9375,
|
| 3791 |
+
"step": 539
|
| 3792 |
+
},
|
| 3793 |
+
{
|
| 3794 |
+
"epoch": 5.4e-05,
|
| 3795 |
+
"grad_norm": 13.923197746276855,
|
| 3796 |
+
"learning_rate": 5.39e-06,
|
| 3797 |
+
"loss": 71.875,
|
| 3798 |
+
"step": 540
|
| 3799 |
+
},
|
| 3800 |
+
{
|
| 3801 |
+
"epoch": 5.41e-05,
|
| 3802 |
+
"grad_norm": 13.1552152633667,
|
| 3803 |
+
"learning_rate": 5.4e-06,
|
| 3804 |
+
"loss": 71.875,
|
| 3805 |
+
"step": 541
|
| 3806 |
+
},
|
| 3807 |
+
{
|
| 3808 |
+
"epoch": 5.42e-05,
|
| 3809 |
+
"grad_norm": 13.265161514282227,
|
| 3810 |
+
"learning_rate": 5.410000000000001e-06,
|
| 3811 |
+
"loss": 71.9375,
|
| 3812 |
+
"step": 542
|
| 3813 |
+
},
|
| 3814 |
+
{
|
| 3815 |
+
"epoch": 5.43e-05,
|
| 3816 |
+
"grad_norm": 13.648545265197754,
|
| 3817 |
+
"learning_rate": 5.420000000000001e-06,
|
| 3818 |
+
"loss": 71.875,
|
| 3819 |
+
"step": 543
|
| 3820 |
+
},
|
| 3821 |
+
{
|
| 3822 |
+
"epoch": 5.44e-05,
|
| 3823 |
+
"grad_norm": 14.054911613464355,
|
| 3824 |
+
"learning_rate": 5.43e-06,
|
| 3825 |
+
"loss": 71.75,
|
| 3826 |
+
"step": 544
|
| 3827 |
+
},
|
| 3828 |
+
{
|
| 3829 |
+
"epoch": 5.45e-05,
|
| 3830 |
+
"grad_norm": 13.410968780517578,
|
| 3831 |
+
"learning_rate": 5.44e-06,
|
| 3832 |
+
"loss": 71.8125,
|
| 3833 |
+
"step": 545
|
| 3834 |
+
},
|
| 3835 |
+
{
|
| 3836 |
+
"epoch": 5.46e-05,
|
| 3837 |
+
"grad_norm": 14.150003433227539,
|
| 3838 |
+
"learning_rate": 5.45e-06,
|
| 3839 |
+
"loss": 71.8125,
|
| 3840 |
+
"step": 546
|
| 3841 |
+
},
|
| 3842 |
+
{
|
| 3843 |
+
"epoch": 5.47e-05,
|
| 3844 |
+
"grad_norm": 13.140257835388184,
|
| 3845 |
+
"learning_rate": 5.46e-06,
|
| 3846 |
+
"loss": 71.75,
|
| 3847 |
+
"step": 547
|
| 3848 |
+
},
|
| 3849 |
+
{
|
| 3850 |
+
"epoch": 5.48e-05,
|
| 3851 |
+
"grad_norm": 13.407617568969727,
|
| 3852 |
+
"learning_rate": 5.47e-06,
|
| 3853 |
+
"loss": 71.875,
|
| 3854 |
+
"step": 548
|
| 3855 |
+
},
|
| 3856 |
+
{
|
| 3857 |
+
"epoch": 5.49e-05,
|
| 3858 |
+
"grad_norm": 13.361193656921387,
|
| 3859 |
+
"learning_rate": 5.48e-06,
|
| 3860 |
+
"loss": 71.8125,
|
| 3861 |
+
"step": 549
|
| 3862 |
+
},
|
| 3863 |
+
{
|
| 3864 |
+
"epoch": 5.5e-05,
|
| 3865 |
+
"grad_norm": 13.742301940917969,
|
| 3866 |
+
"learning_rate": 5.49e-06,
|
| 3867 |
+
"loss": 71.75,
|
| 3868 |
+
"step": 550
|
| 3869 |
+
},
|
| 3870 |
+
{
|
| 3871 |
+
"epoch": 5.51e-05,
|
| 3872 |
+
"grad_norm": 13.647250175476074,
|
| 3873 |
+
"learning_rate": 5.5e-06,
|
| 3874 |
+
"loss": 71.8125,
|
| 3875 |
+
"step": 551
|
| 3876 |
+
},
|
| 3877 |
+
{
|
| 3878 |
+
"epoch": 5.52e-05,
|
| 3879 |
+
"grad_norm": 13.841861724853516,
|
| 3880 |
+
"learning_rate": 5.51e-06,
|
| 3881 |
+
"loss": 71.625,
|
| 3882 |
+
"step": 552
|
| 3883 |
+
},
|
| 3884 |
+
{
|
| 3885 |
+
"epoch": 5.53e-05,
|
| 3886 |
+
"grad_norm": 13.940226554870605,
|
| 3887 |
+
"learning_rate": 5.5200000000000005e-06,
|
| 3888 |
+
"loss": 71.5625,
|
| 3889 |
+
"step": 553
|
| 3890 |
+
},
|
| 3891 |
+
{
|
| 3892 |
+
"epoch": 5.54e-05,
|
| 3893 |
+
"grad_norm": 14.098892211914062,
|
| 3894 |
+
"learning_rate": 5.53e-06,
|
| 3895 |
+
"loss": 71.875,
|
| 3896 |
+
"step": 554
|
| 3897 |
+
},
|
| 3898 |
+
{
|
| 3899 |
+
"epoch": 5.55e-05,
|
| 3900 |
+
"grad_norm": 13.767340660095215,
|
| 3901 |
+
"learning_rate": 5.54e-06,
|
| 3902 |
+
"loss": 71.8125,
|
| 3903 |
+
"step": 555
|
| 3904 |
+
},
|
| 3905 |
+
{
|
| 3906 |
+
"epoch": 5.56e-05,
|
| 3907 |
+
"grad_norm": 13.485721588134766,
|
| 3908 |
+
"learning_rate": 5.55e-06,
|
| 3909 |
+
"loss": 71.5625,
|
| 3910 |
+
"step": 556
|
| 3911 |
+
},
|
| 3912 |
+
{
|
| 3913 |
+
"epoch": 5.57e-05,
|
| 3914 |
+
"grad_norm": 13.09888744354248,
|
| 3915 |
+
"learning_rate": 5.56e-06,
|
| 3916 |
+
"loss": 71.75,
|
| 3917 |
+
"step": 557
|
| 3918 |
+
},
|
| 3919 |
+
{
|
| 3920 |
+
"epoch": 5.58e-05,
|
| 3921 |
+
"grad_norm": 13.47565746307373,
|
| 3922 |
+
"learning_rate": 5.57e-06,
|
| 3923 |
+
"loss": 71.75,
|
| 3924 |
+
"step": 558
|
| 3925 |
+
},
|
| 3926 |
+
{
|
| 3927 |
+
"epoch": 5.59e-05,
|
| 3928 |
+
"grad_norm": 13.305706977844238,
|
| 3929 |
+
"learning_rate": 5.580000000000001e-06,
|
| 3930 |
+
"loss": 71.5625,
|
| 3931 |
+
"step": 559
|
| 3932 |
+
},
|
| 3933 |
+
{
|
| 3934 |
+
"epoch": 5.6e-05,
|
| 3935 |
+
"grad_norm": 13.260443687438965,
|
| 3936 |
+
"learning_rate": 5.59e-06,
|
| 3937 |
+
"loss": 71.5625,
|
| 3938 |
+
"step": 560
|
| 3939 |
+
},
|
| 3940 |
+
{
|
| 3941 |
+
"epoch": 5.61e-05,
|
| 3942 |
+
"grad_norm": 13.876309394836426,
|
| 3943 |
+
"learning_rate": 5.6e-06,
|
| 3944 |
+
"loss": 71.625,
|
| 3945 |
+
"step": 561
|
| 3946 |
+
},
|
| 3947 |
+
{
|
| 3948 |
+
"epoch": 5.62e-05,
|
| 3949 |
+
"grad_norm": 13.163431167602539,
|
| 3950 |
+
"learning_rate": 5.61e-06,
|
| 3951 |
+
"loss": 71.5625,
|
| 3952 |
+
"step": 562
|
| 3953 |
+
},
|
| 3954 |
+
{
|
| 3955 |
+
"epoch": 5.63e-05,
|
| 3956 |
+
"grad_norm": 13.940244674682617,
|
| 3957 |
+
"learning_rate": 5.62e-06,
|
| 3958 |
+
"loss": 71.5625,
|
| 3959 |
+
"step": 563
|
| 3960 |
+
},
|
| 3961 |
+
{
|
| 3962 |
+
"epoch": 5.64e-05,
|
| 3963 |
+
"grad_norm": 14.002641677856445,
|
| 3964 |
+
"learning_rate": 5.63e-06,
|
| 3965 |
+
"loss": 71.625,
|
| 3966 |
+
"step": 564
|
| 3967 |
+
},
|
| 3968 |
+
{
|
| 3969 |
+
"epoch": 5.65e-05,
|
| 3970 |
+
"grad_norm": 13.585519790649414,
|
| 3971 |
+
"learning_rate": 5.64e-06,
|
| 3972 |
+
"loss": 71.5,
|
| 3973 |
+
"step": 565
|
| 3974 |
+
},
|
| 3975 |
+
{
|
| 3976 |
+
"epoch": 5.66e-05,
|
| 3977 |
+
"grad_norm": 13.694162368774414,
|
| 3978 |
+
"learning_rate": 5.65e-06,
|
| 3979 |
+
"loss": 71.625,
|
| 3980 |
+
"step": 566
|
| 3981 |
+
},
|
| 3982 |
+
{
|
| 3983 |
+
"epoch": 5.67e-05,
|
| 3984 |
+
"grad_norm": 13.502567291259766,
|
| 3985 |
+
"learning_rate": 5.66e-06,
|
| 3986 |
+
"loss": 71.625,
|
| 3987 |
+
"step": 567
|
| 3988 |
+
},
|
| 3989 |
+
{
|
| 3990 |
+
"epoch": 5.68e-05,
|
| 3991 |
+
"grad_norm": 13.50383186340332,
|
| 3992 |
+
"learning_rate": 5.67e-06,
|
| 3993 |
+
"loss": 71.625,
|
| 3994 |
+
"step": 568
|
| 3995 |
+
},
|
| 3996 |
+
{
|
| 3997 |
+
"epoch": 5.69e-05,
|
| 3998 |
+
"grad_norm": 13.493083953857422,
|
| 3999 |
+
"learning_rate": 5.68e-06,
|
| 4000 |
+
"loss": 71.5625,
|
| 4001 |
+
"step": 569
|
| 4002 |
+
},
|
| 4003 |
+
{
|
| 4004 |
+
"epoch": 5.7e-05,
|
| 4005 |
+
"grad_norm": 13.770458221435547,
|
| 4006 |
+
"learning_rate": 5.690000000000001e-06,
|
| 4007 |
+
"loss": 71.4375,
|
| 4008 |
+
"step": 570
|
| 4009 |
+
},
|
| 4010 |
+
{
|
| 4011 |
+
"epoch": 5.71e-05,
|
| 4012 |
+
"grad_norm": 13.740925788879395,
|
| 4013 |
+
"learning_rate": 5.7000000000000005e-06,
|
| 4014 |
+
"loss": 71.5625,
|
| 4015 |
+
"step": 571
|
| 4016 |
+
},
|
| 4017 |
+
{
|
| 4018 |
+
"epoch": 5.72e-05,
|
| 4019 |
+
"grad_norm": 13.495709419250488,
|
| 4020 |
+
"learning_rate": 5.7099999999999995e-06,
|
| 4021 |
+
"loss": 71.5625,
|
| 4022 |
+
"step": 572
|
| 4023 |
+
},
|
| 4024 |
+
{
|
| 4025 |
+
"epoch": 5.73e-05,
|
| 4026 |
+
"grad_norm": 14.037687301635742,
|
| 4027 |
+
"learning_rate": 5.72e-06,
|
| 4028 |
+
"loss": 71.5,
|
| 4029 |
+
"step": 573
|
| 4030 |
+
},
|
| 4031 |
+
{
|
| 4032 |
+
"epoch": 5.74e-05,
|
| 4033 |
+
"grad_norm": 13.584688186645508,
|
| 4034 |
+
"learning_rate": 5.73e-06,
|
| 4035 |
+
"loss": 71.5625,
|
| 4036 |
+
"step": 574
|
| 4037 |
+
},
|
| 4038 |
+
{
|
| 4039 |
+
"epoch": 5.75e-05,
|
| 4040 |
+
"grad_norm": 13.670408248901367,
|
| 4041 |
+
"learning_rate": 5.74e-06,
|
| 4042 |
+
"loss": 71.5,
|
| 4043 |
+
"step": 575
|
| 4044 |
+
},
|
| 4045 |
+
{
|
| 4046 |
+
"epoch": 5.76e-05,
|
| 4047 |
+
"grad_norm": 13.831013679504395,
|
| 4048 |
+
"learning_rate": 5.75e-06,
|
| 4049 |
+
"loss": 71.5,
|
| 4050 |
+
"step": 576
|
| 4051 |
+
},
|
| 4052 |
+
{
|
| 4053 |
+
"epoch": 5.77e-05,
|
| 4054 |
+
"grad_norm": 13.588455200195312,
|
| 4055 |
+
"learning_rate": 5.76e-06,
|
| 4056 |
+
"loss": 71.5,
|
| 4057 |
+
"step": 577
|
| 4058 |
+
},
|
| 4059 |
+
{
|
| 4060 |
+
"epoch": 5.78e-05,
|
| 4061 |
+
"grad_norm": 14.016668319702148,
|
| 4062 |
+
"learning_rate": 5.77e-06,
|
| 4063 |
+
"loss": 71.5,
|
| 4064 |
+
"step": 578
|
| 4065 |
+
},
|
| 4066 |
+
{
|
| 4067 |
+
"epoch": 5.79e-05,
|
| 4068 |
+
"grad_norm": 13.188124656677246,
|
| 4069 |
+
"learning_rate": 5.78e-06,
|
| 4070 |
+
"loss": 71.5625,
|
| 4071 |
+
"step": 579
|
| 4072 |
+
},
|
| 4073 |
+
{
|
| 4074 |
+
"epoch": 5.8e-05,
|
| 4075 |
+
"grad_norm": 14.161450386047363,
|
| 4076 |
+
"learning_rate": 5.7900000000000005e-06,
|
| 4077 |
+
"loss": 71.5,
|
| 4078 |
+
"step": 580
|
| 4079 |
+
},
|
| 4080 |
+
{
|
| 4081 |
+
"epoch": 5.81e-05,
|
| 4082 |
+
"grad_norm": 13.826251983642578,
|
| 4083 |
+
"learning_rate": 5.8e-06,
|
| 4084 |
+
"loss": 71.4375,
|
| 4085 |
+
"step": 581
|
| 4086 |
+
},
|
| 4087 |
+
{
|
| 4088 |
+
"epoch": 5.82e-05,
|
| 4089 |
+
"grad_norm": 13.522049903869629,
|
| 4090 |
+
"learning_rate": 5.81e-06,
|
| 4091 |
+
"loss": 71.5,
|
| 4092 |
+
"step": 582
|
| 4093 |
+
},
|
| 4094 |
+
{
|
| 4095 |
+
"epoch": 5.83e-05,
|
| 4096 |
+
"grad_norm": 13.675570487976074,
|
| 4097 |
+
"learning_rate": 5.82e-06,
|
| 4098 |
+
"loss": 71.5625,
|
| 4099 |
+
"step": 583
|
| 4100 |
+
},
|
| 4101 |
+
{
|
| 4102 |
+
"epoch": 5.84e-05,
|
| 4103 |
+
"grad_norm": 13.925761222839355,
|
| 4104 |
+
"learning_rate": 5.83e-06,
|
| 4105 |
+
"loss": 71.5,
|
| 4106 |
+
"step": 584
|
| 4107 |
+
},
|
| 4108 |
+
{
|
| 4109 |
+
"epoch": 5.85e-05,
|
| 4110 |
+
"grad_norm": 13.596080780029297,
|
| 4111 |
+
"learning_rate": 5.84e-06,
|
| 4112 |
+
"loss": 71.5,
|
| 4113 |
+
"step": 585
|
| 4114 |
+
},
|
| 4115 |
+
{
|
| 4116 |
+
"epoch": 5.86e-05,
|
| 4117 |
+
"grad_norm": 13.48724365234375,
|
| 4118 |
+
"learning_rate": 5.85e-06,
|
| 4119 |
+
"loss": 71.5,
|
| 4120 |
+
"step": 586
|
| 4121 |
+
},
|
| 4122 |
+
{
|
| 4123 |
+
"epoch": 5.87e-05,
|
| 4124 |
+
"grad_norm": 13.393336296081543,
|
| 4125 |
+
"learning_rate": 5.860000000000001e-06,
|
| 4126 |
+
"loss": 71.3125,
|
| 4127 |
+
"step": 587
|
| 4128 |
+
},
|
| 4129 |
+
{
|
| 4130 |
+
"epoch": 5.88e-05,
|
| 4131 |
+
"grad_norm": 13.0198392868042,
|
| 4132 |
+
"learning_rate": 5.87e-06,
|
| 4133 |
+
"loss": 71.5,
|
| 4134 |
+
"step": 588
|
| 4135 |
+
},
|
| 4136 |
+
{
|
| 4137 |
+
"epoch": 5.89e-05,
|
| 4138 |
+
"grad_norm": 14.366602897644043,
|
| 4139 |
+
"learning_rate": 5.88e-06,
|
| 4140 |
+
"loss": 71.3125,
|
| 4141 |
+
"step": 589
|
| 4142 |
+
},
|
| 4143 |
+
{
|
| 4144 |
+
"epoch": 5.9e-05,
|
| 4145 |
+
"grad_norm": 13.80266284942627,
|
| 4146 |
+
"learning_rate": 5.89e-06,
|
| 4147 |
+
"loss": 71.4375,
|
| 4148 |
+
"step": 590
|
| 4149 |
+
},
|
| 4150 |
+
{
|
| 4151 |
+
"epoch": 5.91e-05,
|
| 4152 |
+
"grad_norm": 14.043784141540527,
|
| 4153 |
+
"learning_rate": 5.899999999999999e-06,
|
| 4154 |
+
"loss": 71.4375,
|
| 4155 |
+
"step": 591
|
| 4156 |
+
},
|
| 4157 |
+
{
|
| 4158 |
+
"epoch": 5.92e-05,
|
| 4159 |
+
"grad_norm": 14.366748809814453,
|
| 4160 |
+
"learning_rate": 5.91e-06,
|
| 4161 |
+
"loss": 71.5,
|
| 4162 |
+
"step": 592
|
| 4163 |
+
},
|
| 4164 |
+
{
|
| 4165 |
+
"epoch": 5.93e-05,
|
| 4166 |
+
"grad_norm": 13.664860725402832,
|
| 4167 |
+
"learning_rate": 5.92e-06,
|
| 4168 |
+
"loss": 71.4375,
|
| 4169 |
+
"step": 593
|
| 4170 |
+
},
|
| 4171 |
+
{
|
| 4172 |
+
"epoch": 5.94e-05,
|
| 4173 |
+
"grad_norm": 12.838698387145996,
|
| 4174 |
+
"learning_rate": 5.93e-06,
|
| 4175 |
+
"loss": 71.4375,
|
| 4176 |
+
"step": 594
|
| 4177 |
+
},
|
| 4178 |
+
{
|
| 4179 |
+
"epoch": 5.95e-05,
|
| 4180 |
+
"grad_norm": 13.87798023223877,
|
| 4181 |
+
"learning_rate": 5.94e-06,
|
| 4182 |
+
"loss": 71.3125,
|
| 4183 |
+
"step": 595
|
| 4184 |
+
},
|
| 4185 |
+
{
|
| 4186 |
+
"epoch": 5.96e-05,
|
| 4187 |
+
"grad_norm": 13.455665588378906,
|
| 4188 |
+
"learning_rate": 5.950000000000001e-06,
|
| 4189 |
+
"loss": 71.3125,
|
| 4190 |
+
"step": 596
|
| 4191 |
+
},
|
| 4192 |
+
{
|
| 4193 |
+
"epoch": 5.97e-05,
|
| 4194 |
+
"grad_norm": 13.53802490234375,
|
| 4195 |
+
"learning_rate": 5.96e-06,
|
| 4196 |
+
"loss": 71.4375,
|
| 4197 |
+
"step": 597
|
| 4198 |
+
},
|
| 4199 |
+
{
|
| 4200 |
+
"epoch": 5.98e-05,
|
| 4201 |
+
"grad_norm": 14.010590553283691,
|
| 4202 |
+
"learning_rate": 5.9700000000000004e-06,
|
| 4203 |
+
"loss": 71.5,
|
| 4204 |
+
"step": 598
|
| 4205 |
+
},
|
| 4206 |
+
{
|
| 4207 |
+
"epoch": 5.99e-05,
|
| 4208 |
+
"grad_norm": 13.441306114196777,
|
| 4209 |
+
"learning_rate": 5.98e-06,
|
| 4210 |
+
"loss": 71.375,
|
| 4211 |
+
"step": 599
|
| 4212 |
+
},
|
| 4213 |
+
{
|
| 4214 |
+
"epoch": 6e-05,
|
| 4215 |
+
"grad_norm": 13.627511024475098,
|
| 4216 |
+
"learning_rate": 5.989999999999999e-06,
|
| 4217 |
+
"loss": 71.4375,
|
| 4218 |
+
"step": 600
|
| 4219 |
+
},
|
| 4220 |
+
{
|
| 4221 |
+
"epoch": 6.01e-05,
|
| 4222 |
+
"grad_norm": 13.531881332397461,
|
| 4223 |
+
"learning_rate": 6e-06,
|
| 4224 |
+
"loss": 71.4375,
|
| 4225 |
+
"step": 601
|
| 4226 |
+
},
|
| 4227 |
+
{
|
| 4228 |
+
"epoch": 6.02e-05,
|
| 4229 |
+
"grad_norm": 13.889076232910156,
|
| 4230 |
+
"learning_rate": 6.010000000000001e-06,
|
| 4231 |
+
"loss": 71.3125,
|
| 4232 |
+
"step": 602
|
| 4233 |
+
},
|
| 4234 |
+
{
|
| 4235 |
+
"epoch": 6.03e-05,
|
| 4236 |
+
"grad_norm": 13.621227264404297,
|
| 4237 |
+
"learning_rate": 6.02e-06,
|
| 4238 |
+
"loss": 71.25,
|
| 4239 |
+
"step": 603
|
| 4240 |
+
},
|
| 4241 |
+
{
|
| 4242 |
+
"epoch": 6.04e-05,
|
| 4243 |
+
"grad_norm": 13.696484565734863,
|
| 4244 |
+
"learning_rate": 6.030000000000001e-06,
|
| 4245 |
+
"loss": 71.375,
|
| 4246 |
+
"step": 604
|
| 4247 |
+
},
|
| 4248 |
+
{
|
| 4249 |
+
"epoch": 6.05e-05,
|
| 4250 |
+
"grad_norm": 13.138842582702637,
|
| 4251 |
+
"learning_rate": 6.04e-06,
|
| 4252 |
+
"loss": 71.3125,
|
| 4253 |
+
"step": 605
|
| 4254 |
+
},
|
| 4255 |
+
{
|
| 4256 |
+
"epoch": 6.06e-05,
|
| 4257 |
+
"grad_norm": 14.300238609313965,
|
| 4258 |
+
"learning_rate": 6.05e-06,
|
| 4259 |
+
"loss": 71.25,
|
| 4260 |
+
"step": 606
|
| 4261 |
+
},
|
| 4262 |
+
{
|
| 4263 |
+
"epoch": 6.07e-05,
|
| 4264 |
+
"grad_norm": 13.133456230163574,
|
| 4265 |
+
"learning_rate": 6.0600000000000004e-06,
|
| 4266 |
+
"loss": 71.4375,
|
| 4267 |
+
"step": 607
|
| 4268 |
+
},
|
| 4269 |
+
{
|
| 4270 |
+
"epoch": 6.08e-05,
|
| 4271 |
+
"grad_norm": 13.548200607299805,
|
| 4272 |
+
"learning_rate": 6.0699999999999995e-06,
|
| 4273 |
+
"loss": 71.375,
|
| 4274 |
+
"step": 608
|
| 4275 |
+
},
|
| 4276 |
+
{
|
| 4277 |
+
"epoch": 6.09e-05,
|
| 4278 |
+
"grad_norm": 13.798404693603516,
|
| 4279 |
+
"learning_rate": 6.08e-06,
|
| 4280 |
+
"loss": 71.25,
|
| 4281 |
+
"step": 609
|
| 4282 |
+
},
|
| 4283 |
+
{
|
| 4284 |
+
"epoch": 6.1e-05,
|
| 4285 |
+
"grad_norm": 13.428898811340332,
|
| 4286 |
+
"learning_rate": 6.09e-06,
|
| 4287 |
+
"loss": 71.375,
|
| 4288 |
+
"step": 610
|
| 4289 |
+
},
|
| 4290 |
+
{
|
| 4291 |
+
"epoch": 6.11e-05,
|
| 4292 |
+
"grad_norm": 13.93244457244873,
|
| 4293 |
+
"learning_rate": 6.1e-06,
|
| 4294 |
+
"loss": 71.25,
|
| 4295 |
+
"step": 611
|
| 4296 |
+
},
|
| 4297 |
+
{
|
| 4298 |
+
"epoch": 6.12e-05,
|
| 4299 |
+
"grad_norm": 14.050407409667969,
|
| 4300 |
+
"learning_rate": 6.11e-06,
|
| 4301 |
+
"loss": 71.375,
|
| 4302 |
+
"step": 612
|
| 4303 |
+
},
|
| 4304 |
+
{
|
| 4305 |
+
"epoch": 6.13e-05,
|
| 4306 |
+
"grad_norm": 13.533824920654297,
|
| 4307 |
+
"learning_rate": 6.120000000000001e-06,
|
| 4308 |
+
"loss": 71.3125,
|
| 4309 |
+
"step": 613
|
| 4310 |
+
},
|
| 4311 |
+
{
|
| 4312 |
+
"epoch": 6.14e-05,
|
| 4313 |
+
"grad_norm": 13.453178405761719,
|
| 4314 |
+
"learning_rate": 6.13e-06,
|
| 4315 |
+
"loss": 71.25,
|
| 4316 |
+
"step": 614
|
| 4317 |
+
},
|
| 4318 |
+
{
|
| 4319 |
+
"epoch": 6.15e-05,
|
| 4320 |
+
"grad_norm": 13.044660568237305,
|
| 4321 |
+
"learning_rate": 6.1400000000000005e-06,
|
| 4322 |
+
"loss": 71.4375,
|
| 4323 |
+
"step": 615
|
| 4324 |
+
},
|
| 4325 |
+
{
|
| 4326 |
+
"epoch": 6.16e-05,
|
| 4327 |
+
"grad_norm": 13.782415390014648,
|
| 4328 |
+
"learning_rate": 6.15e-06,
|
| 4329 |
+
"loss": 71.25,
|
| 4330 |
+
"step": 616
|
| 4331 |
+
},
|
| 4332 |
+
{
|
| 4333 |
+
"epoch": 6.17e-05,
|
| 4334 |
+
"grad_norm": 14.078405380249023,
|
| 4335 |
+
"learning_rate": 6.1599999999999995e-06,
|
| 4336 |
+
"loss": 71.25,
|
| 4337 |
+
"step": 617
|
| 4338 |
+
},
|
| 4339 |
+
{
|
| 4340 |
+
"epoch": 6.18e-05,
|
| 4341 |
+
"grad_norm": 13.621548652648926,
|
| 4342 |
+
"learning_rate": 6.17e-06,
|
| 4343 |
+
"loss": 71.3125,
|
| 4344 |
+
"step": 618
|
| 4345 |
+
},
|
| 4346 |
+
{
|
| 4347 |
+
"epoch": 6.19e-05,
|
| 4348 |
+
"grad_norm": 13.817481994628906,
|
| 4349 |
+
"learning_rate": 6.180000000000001e-06,
|
| 4350 |
+
"loss": 71.3125,
|
| 4351 |
+
"step": 619
|
| 4352 |
+
},
|
| 4353 |
+
{
|
| 4354 |
+
"epoch": 6.2e-05,
|
| 4355 |
+
"grad_norm": 13.902446746826172,
|
| 4356 |
+
"learning_rate": 6.19e-06,
|
| 4357 |
+
"loss": 71.125,
|
| 4358 |
+
"step": 620
|
| 4359 |
+
},
|
| 4360 |
+
{
|
| 4361 |
+
"epoch": 6.21e-05,
|
| 4362 |
+
"grad_norm": 14.227055549621582,
|
| 4363 |
+
"learning_rate": 6.2e-06,
|
| 4364 |
+
"loss": 71.0625,
|
| 4365 |
+
"step": 621
|
| 4366 |
+
},
|
| 4367 |
+
{
|
| 4368 |
+
"epoch": 6.22e-05,
|
| 4369 |
+
"grad_norm": 14.249224662780762,
|
| 4370 |
+
"learning_rate": 6.21e-06,
|
| 4371 |
+
"loss": 71.0,
|
| 4372 |
+
"step": 622
|
| 4373 |
+
},
|
| 4374 |
+
{
|
| 4375 |
+
"epoch": 6.23e-05,
|
| 4376 |
+
"grad_norm": 13.914899826049805,
|
| 4377 |
+
"learning_rate": 6.22e-06,
|
| 4378 |
+
"loss": 71.1875,
|
| 4379 |
+
"step": 623
|
| 4380 |
+
},
|
| 4381 |
+
{
|
| 4382 |
+
"epoch": 6.24e-05,
|
| 4383 |
+
"grad_norm": 13.97021770477295,
|
| 4384 |
+
"learning_rate": 6.2300000000000005e-06,
|
| 4385 |
+
"loss": 71.0,
|
| 4386 |
+
"step": 624
|
| 4387 |
+
},
|
| 4388 |
+
{
|
| 4389 |
+
"epoch": 6.25e-05,
|
| 4390 |
+
"grad_norm": 13.667683601379395,
|
| 4391 |
+
"learning_rate": 6.2399999999999995e-06,
|
| 4392 |
+
"loss": 71.0625,
|
| 4393 |
+
"step": 625
|
| 4394 |
+
},
|
| 4395 |
+
{
|
| 4396 |
+
"epoch": 6.26e-05,
|
| 4397 |
+
"grad_norm": 13.566893577575684,
|
| 4398 |
+
"learning_rate": 6.25e-06,
|
| 4399 |
+
"loss": 71.1875,
|
| 4400 |
+
"step": 626
|
| 4401 |
+
},
|
| 4402 |
+
{
|
| 4403 |
+
"epoch": 6.27e-05,
|
| 4404 |
+
"grad_norm": 14.130186080932617,
|
| 4405 |
+
"learning_rate": 6.26e-06,
|
| 4406 |
+
"loss": 71.0625,
|
| 4407 |
+
"step": 627
|
| 4408 |
+
},
|
| 4409 |
+
{
|
| 4410 |
+
"epoch": 6.28e-05,
|
| 4411 |
+
"grad_norm": 13.867085456848145,
|
| 4412 |
+
"learning_rate": 6.269999999999999e-06,
|
| 4413 |
+
"loss": 71.1875,
|
| 4414 |
+
"step": 628
|
| 4415 |
+
},
|
| 4416 |
+
{
|
| 4417 |
+
"epoch": 6.29e-05,
|
| 4418 |
+
"grad_norm": 14.15766716003418,
|
| 4419 |
+
"learning_rate": 6.28e-06,
|
| 4420 |
+
"loss": 71.0625,
|
| 4421 |
+
"step": 629
|
| 4422 |
+
},
|
| 4423 |
+
{
|
| 4424 |
+
"epoch": 6.3e-05,
|
| 4425 |
+
"grad_norm": 13.474671363830566,
|
| 4426 |
+
"learning_rate": 6.290000000000001e-06,
|
| 4427 |
+
"loss": 71.1875,
|
| 4428 |
+
"step": 630
|
| 4429 |
+
},
|
| 4430 |
+
{
|
| 4431 |
+
"epoch": 6.31e-05,
|
| 4432 |
+
"grad_norm": 13.981532096862793,
|
| 4433 |
+
"learning_rate": 6.3e-06,
|
| 4434 |
+
"loss": 71.0,
|
| 4435 |
+
"step": 631
|
| 4436 |
+
},
|
| 4437 |
+
{
|
| 4438 |
+
"epoch": 6.32e-05,
|
| 4439 |
+
"grad_norm": 13.78842544555664,
|
| 4440 |
+
"learning_rate": 6.3100000000000006e-06,
|
| 4441 |
+
"loss": 71.0625,
|
| 4442 |
+
"step": 632
|
| 4443 |
+
},
|
| 4444 |
+
{
|
| 4445 |
+
"epoch": 6.33e-05,
|
| 4446 |
+
"grad_norm": 14.002915382385254,
|
| 4447 |
+
"learning_rate": 6.3200000000000005e-06,
|
| 4448 |
+
"loss": 71.1875,
|
| 4449 |
+
"step": 633
|
| 4450 |
+
},
|
| 4451 |
+
{
|
| 4452 |
+
"epoch": 6.34e-05,
|
| 4453 |
+
"grad_norm": 13.505497932434082,
|
| 4454 |
+
"learning_rate": 6.3299999999999995e-06,
|
| 4455 |
+
"loss": 71.125,
|
| 4456 |
+
"step": 634
|
| 4457 |
+
},
|
| 4458 |
+
{
|
| 4459 |
+
"epoch": 6.35e-05,
|
| 4460 |
+
"grad_norm": 13.425030708312988,
|
| 4461 |
+
"learning_rate": 6.34e-06,
|
| 4462 |
+
"loss": 71.0625,
|
| 4463 |
+
"step": 635
|
| 4464 |
+
},
|
| 4465 |
+
{
|
| 4466 |
+
"epoch": 6.36e-05,
|
| 4467 |
+
"grad_norm": 13.330291748046875,
|
| 4468 |
+
"learning_rate": 6.350000000000001e-06,
|
| 4469 |
+
"loss": 71.1875,
|
| 4470 |
+
"step": 636
|
| 4471 |
+
},
|
| 4472 |
+
{
|
| 4473 |
+
"epoch": 6.37e-05,
|
| 4474 |
+
"grad_norm": 14.178326606750488,
|
| 4475 |
+
"learning_rate": 6.36e-06,
|
| 4476 |
+
"loss": 71.0625,
|
| 4477 |
+
"step": 637
|
| 4478 |
+
},
|
| 4479 |
+
{
|
| 4480 |
+
"epoch": 6.38e-05,
|
| 4481 |
+
"grad_norm": 13.922074317932129,
|
| 4482 |
+
"learning_rate": 6.37e-06,
|
| 4483 |
+
"loss": 71.0,
|
| 4484 |
+
"step": 638
|
| 4485 |
+
},
|
| 4486 |
+
{
|
| 4487 |
+
"epoch": 6.39e-05,
|
| 4488 |
+
"grad_norm": 13.808489799499512,
|
| 4489 |
+
"learning_rate": 6.38e-06,
|
| 4490 |
+
"loss": 71.0625,
|
| 4491 |
+
"step": 639
|
| 4492 |
+
},
|
| 4493 |
+
{
|
| 4494 |
+
"epoch": 6.4e-05,
|
| 4495 |
+
"grad_norm": 13.859967231750488,
|
| 4496 |
+
"learning_rate": 6.39e-06,
|
| 4497 |
+
"loss": 71.0625,
|
| 4498 |
+
"step": 640
|
| 4499 |
+
},
|
| 4500 |
+
{
|
| 4501 |
+
"epoch": 6.41e-05,
|
| 4502 |
+
"grad_norm": 14.012020111083984,
|
| 4503 |
+
"learning_rate": 6.4000000000000006e-06,
|
| 4504 |
+
"loss": 71.0,
|
| 4505 |
+
"step": 641
|
| 4506 |
+
},
|
| 4507 |
+
{
|
| 4508 |
+
"epoch": 6.42e-05,
|
| 4509 |
+
"grad_norm": 13.656329154968262,
|
| 4510 |
+
"learning_rate": 6.41e-06,
|
| 4511 |
+
"loss": 71.0,
|
| 4512 |
+
"step": 642
|
| 4513 |
+
},
|
| 4514 |
+
{
|
| 4515 |
+
"epoch": 6.43e-05,
|
| 4516 |
+
"grad_norm": 13.934961318969727,
|
| 4517 |
+
"learning_rate": 6.42e-06,
|
| 4518 |
+
"loss": 71.0625,
|
| 4519 |
+
"step": 643
|
| 4520 |
+
},
|
| 4521 |
+
{
|
| 4522 |
+
"epoch": 6.44e-05,
|
| 4523 |
+
"grad_norm": 13.582793235778809,
|
| 4524 |
+
"learning_rate": 6.43e-06,
|
| 4525 |
+
"loss": 71.0,
|
| 4526 |
+
"step": 644
|
| 4527 |
+
},
|
| 4528 |
+
{
|
| 4529 |
+
"epoch": 6.45e-05,
|
| 4530 |
+
"grad_norm": 13.44519329071045,
|
| 4531 |
+
"learning_rate": 6.439999999999999e-06,
|
| 4532 |
+
"loss": 71.0625,
|
| 4533 |
+
"step": 645
|
| 4534 |
+
},
|
| 4535 |
+
{
|
| 4536 |
+
"epoch": 6.46e-05,
|
| 4537 |
+
"grad_norm": 14.572875022888184,
|
| 4538 |
+
"learning_rate": 6.45e-06,
|
| 4539 |
+
"loss": 70.9375,
|
| 4540 |
+
"step": 646
|
| 4541 |
+
},
|
| 4542 |
+
{
|
| 4543 |
+
"epoch": 6.47e-05,
|
| 4544 |
+
"grad_norm": 13.89630126953125,
|
| 4545 |
+
"learning_rate": 6.460000000000001e-06,
|
| 4546 |
+
"loss": 70.9375,
|
| 4547 |
+
"step": 647
|
| 4548 |
+
},
|
| 4549 |
+
{
|
| 4550 |
+
"epoch": 6.48e-05,
|
| 4551 |
+
"grad_norm": 13.762505531311035,
|
| 4552 |
+
"learning_rate": 6.47e-06,
|
| 4553 |
+
"loss": 70.875,
|
| 4554 |
+
"step": 648
|
| 4555 |
+
},
|
| 4556 |
+
{
|
| 4557 |
+
"epoch": 6.49e-05,
|
| 4558 |
+
"grad_norm": 13.604170799255371,
|
| 4559 |
+
"learning_rate": 6.480000000000001e-06,
|
| 4560 |
+
"loss": 70.9375,
|
| 4561 |
+
"step": 649
|
| 4562 |
+
},
|
| 4563 |
+
{
|
| 4564 |
+
"epoch": 6.5e-05,
|
| 4565 |
+
"grad_norm": 13.724708557128906,
|
| 4566 |
+
"learning_rate": 6.4900000000000005e-06,
|
| 4567 |
+
"loss": 70.9375,
|
| 4568 |
+
"step": 650
|
| 4569 |
+
},
|
| 4570 |
+
{
|
| 4571 |
+
"epoch": 6.51e-05,
|
| 4572 |
+
"grad_norm": 13.948774337768555,
|
| 4573 |
+
"learning_rate": 6.5e-06,
|
| 4574 |
+
"loss": 70.9375,
|
| 4575 |
+
"step": 651
|
| 4576 |
+
},
|
| 4577 |
+
{
|
| 4578 |
+
"epoch": 6.52e-05,
|
| 4579 |
+
"grad_norm": 14.045595169067383,
|
| 4580 |
+
"learning_rate": 6.51e-06,
|
| 4581 |
+
"loss": 70.9375,
|
| 4582 |
+
"step": 652
|
| 4583 |
+
},
|
| 4584 |
+
{
|
| 4585 |
+
"epoch": 6.53e-05,
|
| 4586 |
+
"grad_norm": 13.71200180053711,
|
| 4587 |
+
"learning_rate": 6.520000000000001e-06,
|
| 4588 |
+
"loss": 70.9375,
|
| 4589 |
+
"step": 653
|
| 4590 |
+
},
|
| 4591 |
+
{
|
| 4592 |
+
"epoch": 6.54e-05,
|
| 4593 |
+
"grad_norm": 14.408919334411621,
|
| 4594 |
+
"learning_rate": 6.53e-06,
|
| 4595 |
+
"loss": 70.8125,
|
| 4596 |
+
"step": 654
|
| 4597 |
+
},
|
| 4598 |
+
{
|
| 4599 |
+
"epoch": 6.55e-05,
|
| 4600 |
+
"grad_norm": 13.958176612854004,
|
| 4601 |
+
"learning_rate": 6.54e-06,
|
| 4602 |
+
"loss": 70.9375,
|
| 4603 |
+
"step": 655
|
| 4604 |
+
},
|
| 4605 |
+
{
|
| 4606 |
+
"epoch": 6.56e-05,
|
| 4607 |
+
"grad_norm": 13.734806060791016,
|
| 4608 |
+
"learning_rate": 6.549999999999999e-06,
|
| 4609 |
+
"loss": 71.0625,
|
| 4610 |
+
"step": 656
|
| 4611 |
+
},
|
| 4612 |
+
{
|
| 4613 |
+
"epoch": 6.57e-05,
|
| 4614 |
+
"grad_norm": 14.043981552124023,
|
| 4615 |
+
"learning_rate": 6.56e-06,
|
| 4616 |
+
"loss": 70.8125,
|
| 4617 |
+
"step": 657
|
| 4618 |
+
},
|
| 4619 |
+
{
|
| 4620 |
+
"epoch": 6.58e-05,
|
| 4621 |
+
"grad_norm": 13.461383819580078,
|
| 4622 |
+
"learning_rate": 6.570000000000001e-06,
|
| 4623 |
+
"loss": 71.0,
|
| 4624 |
+
"step": 658
|
| 4625 |
+
},
|
| 4626 |
+
{
|
| 4627 |
+
"epoch": 6.59e-05,
|
| 4628 |
+
"grad_norm": 13.893697738647461,
|
| 4629 |
+
"learning_rate": 6.58e-06,
|
| 4630 |
+
"loss": 70.9375,
|
| 4631 |
+
"step": 659
|
| 4632 |
+
},
|
| 4633 |
+
{
|
| 4634 |
+
"epoch": 6.6e-05,
|
| 4635 |
+
"grad_norm": 14.152552604675293,
|
| 4636 |
+
"learning_rate": 6.5900000000000004e-06,
|
| 4637 |
+
"loss": 70.8125,
|
| 4638 |
+
"step": 660
|
| 4639 |
+
},
|
| 4640 |
+
{
|
| 4641 |
+
"epoch": 6.61e-05,
|
| 4642 |
+
"grad_norm": 14.377528190612793,
|
| 4643 |
+
"learning_rate": 6.6e-06,
|
| 4644 |
+
"loss": 70.75,
|
| 4645 |
+
"step": 661
|
| 4646 |
+
},
|
| 4647 |
+
{
|
| 4648 |
+
"epoch": 6.62e-05,
|
| 4649 |
+
"grad_norm": 13.767778396606445,
|
| 4650 |
+
"learning_rate": 6.609999999999999e-06,
|
| 4651 |
+
"loss": 70.8125,
|
| 4652 |
+
"step": 662
|
| 4653 |
+
},
|
| 4654 |
+
{
|
| 4655 |
+
"epoch": 6.63e-05,
|
| 4656 |
+
"grad_norm": 13.831917762756348,
|
| 4657 |
+
"learning_rate": 6.62e-06,
|
| 4658 |
+
"loss": 70.8125,
|
| 4659 |
+
"step": 663
|
| 4660 |
+
},
|
| 4661 |
+
{
|
| 4662 |
+
"epoch": 6.64e-05,
|
| 4663 |
+
"grad_norm": 14.295366287231445,
|
| 4664 |
+
"learning_rate": 6.630000000000001e-06,
|
| 4665 |
+
"loss": 70.8125,
|
| 4666 |
+
"step": 664
|
| 4667 |
+
},
|
| 4668 |
+
{
|
| 4669 |
+
"epoch": 6.65e-05,
|
| 4670 |
+
"grad_norm": 13.494677543640137,
|
| 4671 |
+
"learning_rate": 6.64e-06,
|
| 4672 |
+
"loss": 71.0,
|
| 4673 |
+
"step": 665
|
| 4674 |
+
},
|
| 4675 |
+
{
|
| 4676 |
+
"epoch": 6.66e-05,
|
| 4677 |
+
"grad_norm": 14.52921199798584,
|
| 4678 |
+
"learning_rate": 6.65e-06,
|
| 4679 |
+
"loss": 70.6875,
|
| 4680 |
+
"step": 666
|
| 4681 |
+
},
|
| 4682 |
+
{
|
| 4683 |
+
"epoch": 6.67e-05,
|
| 4684 |
+
"grad_norm": 14.000243186950684,
|
| 4685 |
+
"learning_rate": 6.660000000000001e-06,
|
| 4686 |
+
"loss": 70.625,
|
| 4687 |
+
"step": 667
|
| 4688 |
+
},
|
| 4689 |
+
{
|
| 4690 |
+
"epoch": 6.68e-05,
|
| 4691 |
+
"grad_norm": 13.905571937561035,
|
| 4692 |
+
"learning_rate": 6.67e-06,
|
| 4693 |
+
"loss": 70.8125,
|
| 4694 |
+
"step": 668
|
| 4695 |
+
},
|
| 4696 |
+
{
|
| 4697 |
+
"epoch": 6.69e-05,
|
| 4698 |
+
"grad_norm": 13.985931396484375,
|
| 4699 |
+
"learning_rate": 6.68e-06,
|
| 4700 |
+
"loss": 70.8125,
|
| 4701 |
+
"step": 669
|
| 4702 |
+
},
|
| 4703 |
+
{
|
| 4704 |
+
"epoch": 6.7e-05,
|
| 4705 |
+
"grad_norm": 13.165407180786133,
|
| 4706 |
+
"learning_rate": 6.690000000000001e-06,
|
| 4707 |
+
"loss": 70.875,
|
| 4708 |
+
"step": 670
|
| 4709 |
+
},
|
| 4710 |
+
{
|
| 4711 |
+
"epoch": 6.71e-05,
|
| 4712 |
+
"grad_norm": 14.570108413696289,
|
| 4713 |
+
"learning_rate": 6.7e-06,
|
| 4714 |
+
"loss": 70.5625,
|
| 4715 |
+
"step": 671
|
| 4716 |
+
},
|
| 4717 |
+
{
|
| 4718 |
+
"epoch": 6.72e-05,
|
| 4719 |
+
"grad_norm": 14.332723617553711,
|
| 4720 |
+
"learning_rate": 6.71e-06,
|
| 4721 |
+
"loss": 70.625,
|
| 4722 |
+
"step": 672
|
| 4723 |
+
},
|
| 4724 |
+
{
|
| 4725 |
+
"epoch": 6.73e-05,
|
| 4726 |
+
"grad_norm": 13.870111465454102,
|
| 4727 |
+
"learning_rate": 6.719999999999999e-06,
|
| 4728 |
+
"loss": 70.5625,
|
| 4729 |
+
"step": 673
|
| 4730 |
+
},
|
| 4731 |
+
{
|
| 4732 |
+
"epoch": 6.74e-05,
|
| 4733 |
+
"grad_norm": 14.580185890197754,
|
| 4734 |
+
"learning_rate": 6.73e-06,
|
| 4735 |
+
"loss": 70.75,
|
| 4736 |
+
"step": 674
|
| 4737 |
+
},
|
| 4738 |
+
{
|
| 4739 |
+
"epoch": 6.75e-05,
|
| 4740 |
+
"grad_norm": 13.757917404174805,
|
| 4741 |
+
"learning_rate": 6.740000000000001e-06,
|
| 4742 |
+
"loss": 70.5625,
|
| 4743 |
+
"step": 675
|
| 4744 |
+
},
|
| 4745 |
+
{
|
| 4746 |
+
"epoch": 6.76e-05,
|
| 4747 |
+
"grad_norm": 14.098215103149414,
|
| 4748 |
+
"learning_rate": 6.75e-06,
|
| 4749 |
+
"loss": 70.625,
|
| 4750 |
+
"step": 676
|
| 4751 |
+
},
|
| 4752 |
+
{
|
| 4753 |
+
"epoch": 6.77e-05,
|
| 4754 |
+
"grad_norm": 13.889370918273926,
|
| 4755 |
+
"learning_rate": 6.7600000000000005e-06,
|
| 4756 |
+
"loss": 70.625,
|
| 4757 |
+
"step": 677
|
| 4758 |
+
},
|
| 4759 |
+
{
|
| 4760 |
+
"epoch": 6.78e-05,
|
| 4761 |
+
"grad_norm": 13.543923377990723,
|
| 4762 |
+
"learning_rate": 6.77e-06,
|
| 4763 |
+
"loss": 70.75,
|
| 4764 |
+
"step": 678
|
| 4765 |
+
},
|
| 4766 |
+
{
|
| 4767 |
+
"epoch": 6.79e-05,
|
| 4768 |
+
"grad_norm": 14.494497299194336,
|
| 4769 |
+
"learning_rate": 6.7799999999999995e-06,
|
| 4770 |
+
"loss": 70.4375,
|
| 4771 |
+
"step": 679
|
| 4772 |
+
},
|
| 4773 |
+
{
|
| 4774 |
+
"epoch": 6.8e-05,
|
| 4775 |
+
"grad_norm": 13.892583847045898,
|
| 4776 |
+
"learning_rate": 6.79e-06,
|
| 4777 |
+
"loss": 70.6875,
|
| 4778 |
+
"step": 680
|
| 4779 |
+
},
|
| 4780 |
+
{
|
| 4781 |
+
"epoch": 6.81e-05,
|
| 4782 |
+
"grad_norm": 14.069755554199219,
|
| 4783 |
+
"learning_rate": 6.800000000000001e-06,
|
| 4784 |
+
"loss": 70.5,
|
| 4785 |
+
"step": 681
|
| 4786 |
+
},
|
| 4787 |
+
{
|
| 4788 |
+
"epoch": 6.82e-05,
|
| 4789 |
+
"grad_norm": 14.10586166381836,
|
| 4790 |
+
"learning_rate": 6.81e-06,
|
| 4791 |
+
"loss": 70.4375,
|
| 4792 |
+
"step": 682
|
| 4793 |
+
},
|
| 4794 |
+
{
|
| 4795 |
+
"epoch": 6.83e-05,
|
| 4796 |
+
"grad_norm": 14.291821479797363,
|
| 4797 |
+
"learning_rate": 6.82e-06,
|
| 4798 |
+
"loss": 70.5625,
|
| 4799 |
+
"step": 683
|
| 4800 |
+
},
|
| 4801 |
+
{
|
| 4802 |
+
"epoch": 6.84e-05,
|
| 4803 |
+
"grad_norm": 14.351411819458008,
|
| 4804 |
+
"learning_rate": 6.830000000000001e-06,
|
| 4805 |
+
"loss": 70.5625,
|
| 4806 |
+
"step": 684
|
| 4807 |
+
},
|
| 4808 |
+
{
|
| 4809 |
+
"epoch": 6.85e-05,
|
| 4810 |
+
"grad_norm": 13.590216636657715,
|
| 4811 |
+
"learning_rate": 6.84e-06,
|
| 4812 |
+
"loss": 70.4375,
|
| 4813 |
+
"step": 685
|
| 4814 |
+
},
|
| 4815 |
+
{
|
| 4816 |
+
"epoch": 6.86e-05,
|
| 4817 |
+
"grad_norm": 14.20474910736084,
|
| 4818 |
+
"learning_rate": 6.8500000000000005e-06,
|
| 4819 |
+
"loss": 70.5,
|
| 4820 |
+
"step": 686
|
| 4821 |
+
},
|
| 4822 |
+
{
|
| 4823 |
+
"epoch": 6.87e-05,
|
| 4824 |
+
"grad_norm": 14.491683959960938,
|
| 4825 |
+
"learning_rate": 6.86e-06,
|
| 4826 |
+
"loss": 70.25,
|
| 4827 |
+
"step": 687
|
| 4828 |
+
},
|
| 4829 |
+
{
|
| 4830 |
+
"epoch": 6.88e-05,
|
| 4831 |
+
"grad_norm": 14.437154769897461,
|
| 4832 |
+
"learning_rate": 6.87e-06,
|
| 4833 |
+
"loss": 70.3125,
|
| 4834 |
+
"step": 688
|
| 4835 |
+
},
|
| 4836 |
+
{
|
| 4837 |
+
"epoch": 6.89e-05,
|
| 4838 |
+
"grad_norm": 14.214030265808105,
|
| 4839 |
+
"learning_rate": 6.88e-06,
|
| 4840 |
+
"loss": 70.375,
|
| 4841 |
+
"step": 689
|
| 4842 |
+
},
|
| 4843 |
+
{
|
| 4844 |
+
"epoch": 6.9e-05,
|
| 4845 |
+
"grad_norm": 13.905532836914062,
|
| 4846 |
+
"learning_rate": 6.889999999999999e-06,
|
| 4847 |
+
"loss": 70.4375,
|
| 4848 |
+
"step": 690
|
| 4849 |
+
},
|
| 4850 |
+
{
|
| 4851 |
+
"epoch": 6.91e-05,
|
| 4852 |
+
"grad_norm": 14.396810531616211,
|
| 4853 |
+
"learning_rate": 6.9e-06,
|
| 4854 |
+
"loss": 70.3125,
|
| 4855 |
+
"step": 691
|
| 4856 |
+
},
|
| 4857 |
+
{
|
| 4858 |
+
"epoch": 6.92e-05,
|
| 4859 |
+
"grad_norm": 13.368425369262695,
|
| 4860 |
+
"learning_rate": 6.910000000000001e-06,
|
| 4861 |
+
"loss": 70.625,
|
| 4862 |
+
"step": 692
|
| 4863 |
+
},
|
| 4864 |
+
{
|
| 4865 |
+
"epoch": 6.93e-05,
|
| 4866 |
+
"grad_norm": 14.467592239379883,
|
| 4867 |
+
"learning_rate": 6.92e-06,
|
| 4868 |
+
"loss": 70.125,
|
| 4869 |
+
"step": 693
|
| 4870 |
+
},
|
| 4871 |
+
{
|
| 4872 |
+
"epoch": 6.94e-05,
|
| 4873 |
+
"grad_norm": 14.07659912109375,
|
| 4874 |
+
"learning_rate": 6.93e-06,
|
| 4875 |
+
"loss": 70.25,
|
| 4876 |
+
"step": 694
|
| 4877 |
+
},
|
| 4878 |
+
{
|
| 4879 |
+
"epoch": 6.95e-05,
|
| 4880 |
+
"grad_norm": 13.678894996643066,
|
| 4881 |
+
"learning_rate": 6.9400000000000005e-06,
|
| 4882 |
+
"loss": 70.25,
|
| 4883 |
+
"step": 695
|
| 4884 |
+
},
|
| 4885 |
+
{
|
| 4886 |
+
"epoch": 6.96e-05,
|
| 4887 |
+
"grad_norm": 13.846305847167969,
|
| 4888 |
+
"learning_rate": 6.9499999999999995e-06,
|
| 4889 |
+
"loss": 70.4375,
|
| 4890 |
+
"step": 696
|
| 4891 |
+
},
|
| 4892 |
+
{
|
| 4893 |
+
"epoch": 6.97e-05,
|
| 4894 |
+
"grad_norm": 14.344511032104492,
|
| 4895 |
+
"learning_rate": 6.96e-06,
|
| 4896 |
+
"loss": 70.125,
|
| 4897 |
+
"step": 697
|
| 4898 |
+
},
|
| 4899 |
+
{
|
| 4900 |
+
"epoch": 6.98e-05,
|
| 4901 |
+
"grad_norm": 13.975683212280273,
|
| 4902 |
+
"learning_rate": 6.970000000000001e-06,
|
| 4903 |
+
"loss": 70.3125,
|
| 4904 |
+
"step": 698
|
| 4905 |
+
},
|
| 4906 |
+
{
|
| 4907 |
+
"epoch": 6.99e-05,
|
| 4908 |
+
"grad_norm": 13.725882530212402,
|
| 4909 |
+
"learning_rate": 6.98e-06,
|
| 4910 |
+
"loss": 70.25,
|
| 4911 |
+
"step": 699
|
| 4912 |
+
},
|
| 4913 |
+
{
|
| 4914 |
+
"epoch": 7e-05,
|
| 4915 |
+
"grad_norm": 14.02601146697998,
|
| 4916 |
+
"learning_rate": 6.99e-06,
|
| 4917 |
+
"loss": 70.1875,
|
| 4918 |
+
"step": 700
|
| 4919 |
+
},
|
| 4920 |
+
{
|
| 4921 |
+
"epoch": 7.01e-05,
|
| 4922 |
+
"grad_norm": 14.149601936340332,
|
| 4923 |
+
"learning_rate": 7.000000000000001e-06,
|
| 4924 |
+
"loss": 70.25,
|
| 4925 |
+
"step": 701
|
| 4926 |
+
},
|
| 4927 |
+
{
|
| 4928 |
+
"epoch": 7.02e-05,
|
| 4929 |
+
"grad_norm": 14.683643341064453,
|
| 4930 |
+
"learning_rate": 7.01e-06,
|
| 4931 |
+
"loss": 69.9375,
|
| 4932 |
+
"step": 702
|
| 4933 |
+
},
|
| 4934 |
+
{
|
| 4935 |
+
"epoch": 7.03e-05,
|
| 4936 |
+
"grad_norm": 14.377092361450195,
|
| 4937 |
+
"learning_rate": 7.0200000000000006e-06,
|
| 4938 |
+
"loss": 70.0625,
|
| 4939 |
+
"step": 703
|
| 4940 |
+
},
|
| 4941 |
+
{
|
| 4942 |
+
"epoch": 7.04e-05,
|
| 4943 |
+
"grad_norm": 13.508744239807129,
|
| 4944 |
+
"learning_rate": 7.03e-06,
|
| 4945 |
+
"loss": 70.125,
|
| 4946 |
+
"step": 704
|
| 4947 |
+
},
|
| 4948 |
+
{
|
| 4949 |
+
"epoch": 7.05e-05,
|
| 4950 |
+
"grad_norm": 14.27863883972168,
|
| 4951 |
+
"learning_rate": 7.04e-06,
|
| 4952 |
+
"loss": 70.0,
|
| 4953 |
+
"step": 705
|
| 4954 |
+
},
|
| 4955 |
+
{
|
| 4956 |
+
"epoch": 7.06e-05,
|
| 4957 |
+
"grad_norm": 14.04149055480957,
|
| 4958 |
+
"learning_rate": 7.05e-06,
|
| 4959 |
+
"loss": 70.0625,
|
| 4960 |
+
"step": 706
|
| 4961 |
+
},
|
| 4962 |
+
{
|
| 4963 |
+
"epoch": 7.07e-05,
|
| 4964 |
+
"grad_norm": 13.799717903137207,
|
| 4965 |
+
"learning_rate": 7.059999999999999e-06,
|
| 4966 |
+
"loss": 70.125,
|
| 4967 |
+
"step": 707
|
| 4968 |
+
},
|
| 4969 |
+
{
|
| 4970 |
+
"epoch": 7.08e-05,
|
| 4971 |
+
"grad_norm": 13.9593505859375,
|
| 4972 |
+
"learning_rate": 7.07e-06,
|
| 4973 |
+
"loss": 70.0,
|
| 4974 |
+
"step": 708
|
| 4975 |
+
},
|
| 4976 |
+
{
|
| 4977 |
+
"epoch": 7.09e-05,
|
| 4978 |
+
"grad_norm": 14.294474601745605,
|
| 4979 |
+
"learning_rate": 7.080000000000001e-06,
|
| 4980 |
+
"loss": 69.875,
|
| 4981 |
+
"step": 709
|
| 4982 |
+
},
|
| 4983 |
+
{
|
| 4984 |
+
"epoch": 7.1e-05,
|
| 4985 |
+
"grad_norm": 14.307499885559082,
|
| 4986 |
+
"learning_rate": 7.09e-06,
|
| 4987 |
+
"loss": 70.0,
|
| 4988 |
+
"step": 710
|
| 4989 |
+
},
|
| 4990 |
+
{
|
| 4991 |
+
"epoch": 7.11e-05,
|
| 4992 |
+
"grad_norm": 13.888057708740234,
|
| 4993 |
+
"learning_rate": 7.1e-06,
|
| 4994 |
+
"loss": 70.0,
|
| 4995 |
+
"step": 711
|
| 4996 |
+
},
|
| 4997 |
+
{
|
| 4998 |
+
"epoch": 7.12e-05,
|
| 4999 |
+
"grad_norm": 13.905348777770996,
|
| 5000 |
+
"learning_rate": 7.1100000000000005e-06,
|
| 5001 |
+
"loss": 70.0,
|
| 5002 |
+
"step": 712
|
| 5003 |
+
},
|
| 5004 |
+
{
|
| 5005 |
+
"epoch": 7.13e-05,
|
| 5006 |
+
"grad_norm": 14.086119651794434,
|
| 5007 |
+
"learning_rate": 7.12e-06,
|
| 5008 |
+
"loss": 69.875,
|
| 5009 |
+
"step": 713
|
| 5010 |
+
},
|
| 5011 |
+
{
|
| 5012 |
+
"epoch": 7.14e-05,
|
| 5013 |
+
"grad_norm": 14.221412658691406,
|
| 5014 |
+
"learning_rate": 7.13e-06,
|
| 5015 |
+
"loss": 69.9375,
|
| 5016 |
+
"step": 714
|
| 5017 |
+
},
|
| 5018 |
+
{
|
| 5019 |
+
"epoch": 7.15e-05,
|
| 5020 |
+
"grad_norm": 14.204227447509766,
|
| 5021 |
+
"learning_rate": 7.14e-06,
|
| 5022 |
+
"loss": 69.875,
|
| 5023 |
+
"step": 715
|
| 5024 |
+
},
|
| 5025 |
+
{
|
| 5026 |
+
"epoch": 7.16e-05,
|
| 5027 |
+
"grad_norm": 14.413752555847168,
|
| 5028 |
+
"learning_rate": 7.15e-06,
|
| 5029 |
+
"loss": 69.75,
|
| 5030 |
+
"step": 716
|
| 5031 |
+
},
|
| 5032 |
+
{
|
| 5033 |
+
"epoch": 7.17e-05,
|
| 5034 |
+
"grad_norm": 14.447648048400879,
|
| 5035 |
+
"learning_rate": 7.16e-06,
|
| 5036 |
+
"loss": 69.625,
|
| 5037 |
+
"step": 717
|
| 5038 |
+
},
|
| 5039 |
+
{
|
| 5040 |
+
"epoch": 7.18e-05,
|
| 5041 |
+
"grad_norm": 13.275869369506836,
|
| 5042 |
+
"learning_rate": 7.170000000000001e-06,
|
| 5043 |
+
"loss": 69.9375,
|
| 5044 |
+
"step": 718
|
| 5045 |
+
},
|
| 5046 |
+
{
|
| 5047 |
+
"epoch": 7.19e-05,
|
| 5048 |
+
"grad_norm": 14.1619873046875,
|
| 5049 |
+
"learning_rate": 7.18e-06,
|
| 5050 |
+
"loss": 69.6875,
|
| 5051 |
+
"step": 719
|
| 5052 |
+
},
|
| 5053 |
+
{
|
| 5054 |
+
"epoch": 7.2e-05,
|
| 5055 |
+
"grad_norm": 14.165518760681152,
|
| 5056 |
+
"learning_rate": 7.190000000000001e-06,
|
| 5057 |
+
"loss": 69.6875,
|
| 5058 |
+
"step": 720
|
| 5059 |
+
},
|
| 5060 |
+
{
|
| 5061 |
+
"epoch": 7.21e-05,
|
| 5062 |
+
"grad_norm": 13.807438850402832,
|
| 5063 |
+
"learning_rate": 7.2e-06,
|
| 5064 |
+
"loss": 69.75,
|
| 5065 |
+
"step": 721
|
| 5066 |
+
},
|
| 5067 |
+
{
|
| 5068 |
+
"epoch": 7.22e-05,
|
| 5069 |
+
"grad_norm": 14.538020133972168,
|
| 5070 |
+
"learning_rate": 7.21e-06,
|
| 5071 |
+
"loss": 69.6875,
|
| 5072 |
+
"step": 722
|
| 5073 |
+
},
|
| 5074 |
+
{
|
| 5075 |
+
"epoch": 7.23e-05,
|
| 5076 |
+
"grad_norm": 14.57617473602295,
|
| 5077 |
+
"learning_rate": 7.22e-06,
|
| 5078 |
+
"loss": 69.5625,
|
| 5079 |
+
"step": 723
|
| 5080 |
+
},
|
| 5081 |
+
{
|
| 5082 |
+
"epoch": 7.24e-05,
|
| 5083 |
+
"grad_norm": 13.881351470947266,
|
| 5084 |
+
"learning_rate": 7.229999999999999e-06,
|
| 5085 |
+
"loss": 69.625,
|
| 5086 |
+
"step": 724
|
| 5087 |
+
},
|
| 5088 |
+
{
|
| 5089 |
+
"epoch": 7.25e-05,
|
| 5090 |
+
"grad_norm": 14.827073097229004,
|
| 5091 |
+
"learning_rate": 7.24e-06,
|
| 5092 |
+
"loss": 69.5625,
|
| 5093 |
+
"step": 725
|
| 5094 |
+
},
|
| 5095 |
+
{
|
| 5096 |
+
"epoch": 7.26e-05,
|
| 5097 |
+
"grad_norm": 14.633003234863281,
|
| 5098 |
+
"learning_rate": 7.250000000000001e-06,
|
| 5099 |
+
"loss": 69.375,
|
| 5100 |
+
"step": 726
|
| 5101 |
+
},
|
| 5102 |
+
{
|
| 5103 |
+
"epoch": 7.27e-05,
|
| 5104 |
+
"grad_norm": 13.844290733337402,
|
| 5105 |
+
"learning_rate": 7.26e-06,
|
| 5106 |
+
"loss": 69.4375,
|
| 5107 |
+
"step": 727
|
| 5108 |
+
},
|
| 5109 |
+
{
|
| 5110 |
+
"epoch": 7.28e-05,
|
| 5111 |
+
"grad_norm": 14.124848365783691,
|
| 5112 |
+
"learning_rate": 7.27e-06,
|
| 5113 |
+
"loss": 69.4375,
|
| 5114 |
+
"step": 728
|
| 5115 |
+
},
|
| 5116 |
+
{
|
| 5117 |
+
"epoch": 7.29e-05,
|
| 5118 |
+
"grad_norm": 14.088972091674805,
|
| 5119 |
+
"learning_rate": 7.280000000000001e-06,
|
| 5120 |
+
"loss": 69.5,
|
| 5121 |
+
"step": 729
|
| 5122 |
+
},
|
| 5123 |
+
{
|
| 5124 |
+
"epoch": 7.3e-05,
|
| 5125 |
+
"grad_norm": 14.448264122009277,
|
| 5126 |
+
"learning_rate": 7.29e-06,
|
| 5127 |
+
"loss": 69.4375,
|
| 5128 |
+
"step": 730
|
| 5129 |
+
},
|
| 5130 |
+
{
|
| 5131 |
+
"epoch": 7.31e-05,
|
| 5132 |
+
"grad_norm": 14.05547046661377,
|
| 5133 |
+
"learning_rate": 7.3e-06,
|
| 5134 |
+
"loss": 69.375,
|
| 5135 |
+
"step": 731
|
| 5136 |
+
},
|
| 5137 |
+
{
|
| 5138 |
+
"epoch": 7.32e-05,
|
| 5139 |
+
"grad_norm": 14.225979804992676,
|
| 5140 |
+
"learning_rate": 7.31e-06,
|
| 5141 |
+
"loss": 69.375,
|
| 5142 |
+
"step": 732
|
| 5143 |
+
},
|
| 5144 |
+
{
|
| 5145 |
+
"epoch": 7.33e-05,
|
| 5146 |
+
"grad_norm": 14.301802635192871,
|
| 5147 |
+
"learning_rate": 7.32e-06,
|
| 5148 |
+
"loss": 69.25,
|
| 5149 |
+
"step": 733
|
| 5150 |
+
},
|
| 5151 |
+
{
|
| 5152 |
+
"epoch": 7.34e-05,
|
| 5153 |
+
"grad_norm": 14.079911231994629,
|
| 5154 |
+
"learning_rate": 7.33e-06,
|
| 5155 |
+
"loss": 69.375,
|
| 5156 |
+
"step": 734
|
| 5157 |
+
},
|
| 5158 |
+
{
|
| 5159 |
+
"epoch": 7.35e-05,
|
| 5160 |
+
"grad_norm": 13.954413414001465,
|
| 5161 |
+
"learning_rate": 7.340000000000001e-06,
|
| 5162 |
+
"loss": 69.375,
|
| 5163 |
+
"step": 735
|
| 5164 |
+
},
|
| 5165 |
+
{
|
| 5166 |
+
"epoch": 7.36e-05,
|
| 5167 |
+
"grad_norm": 14.633395195007324,
|
| 5168 |
+
"learning_rate": 7.35e-06,
|
| 5169 |
+
"loss": 69.1875,
|
| 5170 |
+
"step": 736
|
| 5171 |
+
},
|
| 5172 |
+
{
|
| 5173 |
+
"epoch": 7.37e-05,
|
| 5174 |
+
"grad_norm": 13.73357105255127,
|
| 5175 |
+
"learning_rate": 7.360000000000001e-06,
|
| 5176 |
+
"loss": 69.3125,
|
| 5177 |
+
"step": 737
|
| 5178 |
+
},
|
| 5179 |
+
{
|
| 5180 |
+
"epoch": 7.38e-05,
|
| 5181 |
+
"grad_norm": 13.533805847167969,
|
| 5182 |
+
"learning_rate": 7.37e-06,
|
| 5183 |
+
"loss": 69.375,
|
| 5184 |
+
"step": 738
|
| 5185 |
+
},
|
| 5186 |
+
{
|
| 5187 |
+
"epoch": 7.39e-05,
|
| 5188 |
+
"grad_norm": 13.967336654663086,
|
| 5189 |
+
"learning_rate": 7.38e-06,
|
| 5190 |
+
"loss": 69.125,
|
| 5191 |
+
"step": 739
|
| 5192 |
+
},
|
| 5193 |
+
{
|
| 5194 |
+
"epoch": 7.4e-05,
|
| 5195 |
+
"grad_norm": 13.907071113586426,
|
| 5196 |
+
"learning_rate": 7.39e-06,
|
| 5197 |
+
"loss": 69.25,
|
| 5198 |
+
"step": 740
|
| 5199 |
+
},
|
| 5200 |
+
{
|
| 5201 |
+
"epoch": 7.41e-05,
|
| 5202 |
+
"grad_norm": 14.33962345123291,
|
| 5203 |
+
"learning_rate": 7.3999999999999995e-06,
|
| 5204 |
+
"loss": 69.0625,
|
| 5205 |
+
"step": 741
|
| 5206 |
+
},
|
| 5207 |
+
{
|
| 5208 |
+
"epoch": 7.42e-05,
|
| 5209 |
+
"grad_norm": 14.036276817321777,
|
| 5210 |
+
"learning_rate": 7.41e-06,
|
| 5211 |
+
"loss": 69.0625,
|
| 5212 |
+
"step": 742
|
| 5213 |
+
},
|
| 5214 |
+
{
|
| 5215 |
+
"epoch": 7.43e-05,
|
| 5216 |
+
"grad_norm": 14.373995780944824,
|
| 5217 |
+
"learning_rate": 7.420000000000001e-06,
|
| 5218 |
+
"loss": 69.0,
|
| 5219 |
+
"step": 743
|
| 5220 |
+
},
|
| 5221 |
+
{
|
| 5222 |
+
"epoch": 7.44e-05,
|
| 5223 |
+
"grad_norm": 13.279586791992188,
|
| 5224 |
+
"learning_rate": 7.43e-06,
|
| 5225 |
+
"loss": 69.25,
|
| 5226 |
+
"step": 744
|
| 5227 |
+
},
|
| 5228 |
+
{
|
| 5229 |
+
"epoch": 7.45e-05,
|
| 5230 |
+
"grad_norm": 13.835576057434082,
|
| 5231 |
+
"learning_rate": 7.44e-06,
|
| 5232 |
+
"loss": 69.0625,
|
| 5233 |
+
"step": 745
|
| 5234 |
+
},
|
| 5235 |
+
{
|
| 5236 |
+
"epoch": 7.46e-05,
|
| 5237 |
+
"grad_norm": 14.343767166137695,
|
| 5238 |
+
"learning_rate": 7.450000000000001e-06,
|
| 5239 |
+
"loss": 69.0,
|
| 5240 |
+
"step": 746
|
| 5241 |
+
},
|
| 5242 |
+
{
|
| 5243 |
+
"epoch": 7.47e-05,
|
| 5244 |
+
"grad_norm": 14.20533275604248,
|
| 5245 |
+
"learning_rate": 7.46e-06,
|
| 5246 |
+
"loss": 68.875,
|
| 5247 |
+
"step": 747
|
| 5248 |
+
},
|
| 5249 |
+
{
|
| 5250 |
+
"epoch": 7.48e-05,
|
| 5251 |
+
"grad_norm": 14.370162010192871,
|
| 5252 |
+
"learning_rate": 7.4700000000000005e-06,
|
| 5253 |
+
"loss": 68.75,
|
| 5254 |
+
"step": 748
|
| 5255 |
+
},
|
| 5256 |
+
{
|
| 5257 |
+
"epoch": 7.49e-05,
|
| 5258 |
+
"grad_norm": 14.102258682250977,
|
| 5259 |
+
"learning_rate": 7.48e-06,
|
| 5260 |
+
"loss": 68.8125,
|
| 5261 |
+
"step": 749
|
| 5262 |
+
},
|
| 5263 |
+
{
|
| 5264 |
+
"epoch": 7.5e-05,
|
| 5265 |
+
"grad_norm": 14.238746643066406,
|
| 5266 |
+
"learning_rate": 7.4899999999999994e-06,
|
| 5267 |
+
"loss": 68.75,
|
| 5268 |
+
"step": 750
|
| 5269 |
+
},
|
| 5270 |
+
{
|
| 5271 |
+
"epoch": 7.51e-05,
|
| 5272 |
+
"grad_norm": 14.459688186645508,
|
| 5273 |
+
"learning_rate": 7.5e-06,
|
| 5274 |
+
"loss": 68.75,
|
| 5275 |
+
"step": 751
|
| 5276 |
+
},
|
| 5277 |
+
{
|
| 5278 |
+
"epoch": 7.52e-05,
|
| 5279 |
+
"grad_norm": 13.407461166381836,
|
| 5280 |
+
"learning_rate": 7.510000000000001e-06,
|
| 5281 |
+
"loss": 68.9375,
|
| 5282 |
+
"step": 752
|
| 5283 |
+
},
|
| 5284 |
+
{
|
| 5285 |
+
"epoch": 7.53e-05,
|
| 5286 |
+
"grad_norm": 13.310074806213379,
|
| 5287 |
+
"learning_rate": 7.52e-06,
|
| 5288 |
+
"loss": 68.9375,
|
| 5289 |
+
"step": 753
|
| 5290 |
+
},
|
| 5291 |
+
{
|
| 5292 |
+
"epoch": 7.54e-05,
|
| 5293 |
+
"grad_norm": 14.59634017944336,
|
| 5294 |
+
"learning_rate": 7.530000000000001e-06,
|
| 5295 |
+
"loss": 68.5625,
|
| 5296 |
+
"step": 754
|
| 5297 |
+
},
|
| 5298 |
+
{
|
| 5299 |
+
"epoch": 7.55e-05,
|
| 5300 |
+
"grad_norm": 13.75051212310791,
|
| 5301 |
+
"learning_rate": 7.54e-06,
|
| 5302 |
+
"loss": 68.6875,
|
| 5303 |
+
"step": 755
|
| 5304 |
+
},
|
| 5305 |
+
{
|
| 5306 |
+
"epoch": 7.56e-05,
|
| 5307 |
+
"grad_norm": 14.305959701538086,
|
| 5308 |
+
"learning_rate": 7.55e-06,
|
| 5309 |
+
"loss": 68.5625,
|
| 5310 |
+
"step": 756
|
| 5311 |
+
},
|
| 5312 |
+
{
|
| 5313 |
+
"epoch": 7.57e-05,
|
| 5314 |
+
"grad_norm": 14.588299751281738,
|
| 5315 |
+
"learning_rate": 7.5600000000000005e-06,
|
| 5316 |
+
"loss": 68.375,
|
| 5317 |
+
"step": 757
|
| 5318 |
+
},
|
| 5319 |
+
{
|
| 5320 |
+
"epoch": 7.58e-05,
|
| 5321 |
+
"grad_norm": 14.081939697265625,
|
| 5322 |
+
"learning_rate": 7.5699999999999995e-06,
|
| 5323 |
+
"loss": 68.5,
|
| 5324 |
+
"step": 758
|
| 5325 |
+
},
|
| 5326 |
+
{
|
| 5327 |
+
"epoch": 7.59e-05,
|
| 5328 |
+
"grad_norm": 13.225693702697754,
|
| 5329 |
+
"learning_rate": 7.58e-06,
|
| 5330 |
+
"loss": 68.9375,
|
| 5331 |
+
"step": 759
|
| 5332 |
+
},
|
| 5333 |
+
{
|
| 5334 |
+
"epoch": 7.6e-05,
|
| 5335 |
+
"grad_norm": 13.850139617919922,
|
| 5336 |
+
"learning_rate": 7.59e-06,
|
| 5337 |
+
"loss": 68.625,
|
| 5338 |
+
"step": 760
|
| 5339 |
+
},
|
| 5340 |
+
{
|
| 5341 |
+
"epoch": 7.61e-05,
|
| 5342 |
+
"grad_norm": 13.827278137207031,
|
| 5343 |
+
"learning_rate": 7.6e-06,
|
| 5344 |
+
"loss": 68.5,
|
| 5345 |
+
"step": 761
|
| 5346 |
+
},
|
| 5347 |
+
{
|
| 5348 |
+
"epoch": 7.62e-05,
|
| 5349 |
+
"grad_norm": 14.146428108215332,
|
| 5350 |
+
"learning_rate": 7.61e-06,
|
| 5351 |
+
"loss": 68.3125,
|
| 5352 |
+
"step": 762
|
| 5353 |
+
},
|
| 5354 |
+
{
|
| 5355 |
+
"epoch": 7.63e-05,
|
| 5356 |
+
"grad_norm": 13.658434867858887,
|
| 5357 |
+
"learning_rate": 7.620000000000001e-06,
|
| 5358 |
+
"loss": 68.4375,
|
| 5359 |
+
"step": 763
|
| 5360 |
+
},
|
| 5361 |
+
{
|
| 5362 |
+
"epoch": 7.64e-05,
|
| 5363 |
+
"grad_norm": 14.356409072875977,
|
| 5364 |
+
"learning_rate": 7.63e-06,
|
| 5365 |
+
"loss": 68.3125,
|
| 5366 |
+
"step": 764
|
| 5367 |
+
},
|
| 5368 |
+
{
|
| 5369 |
+
"epoch": 7.65e-05,
|
| 5370 |
+
"grad_norm": 14.146637916564941,
|
| 5371 |
+
"learning_rate": 7.64e-06,
|
| 5372 |
+
"loss": 68.3125,
|
| 5373 |
+
"step": 765
|
| 5374 |
+
},
|
| 5375 |
+
{
|
| 5376 |
+
"epoch": 7.66e-05,
|
| 5377 |
+
"grad_norm": 13.119014739990234,
|
| 5378 |
+
"learning_rate": 7.650000000000001e-06,
|
| 5379 |
+
"loss": 68.5625,
|
| 5380 |
+
"step": 766
|
| 5381 |
+
},
|
| 5382 |
+
{
|
| 5383 |
+
"epoch": 7.67e-05,
|
| 5384 |
+
"grad_norm": 14.340291976928711,
|
| 5385 |
+
"learning_rate": 7.66e-06,
|
| 5386 |
+
"loss": 68.1875,
|
| 5387 |
+
"step": 767
|
| 5388 |
+
},
|
| 5389 |
+
{
|
| 5390 |
+
"epoch": 7.68e-05,
|
| 5391 |
+
"grad_norm": 13.567488670349121,
|
| 5392 |
+
"learning_rate": 7.670000000000001e-06,
|
| 5393 |
+
"loss": 68.375,
|
| 5394 |
+
"step": 768
|
| 5395 |
+
},
|
| 5396 |
+
{
|
| 5397 |
+
"epoch": 7.69e-05,
|
| 5398 |
+
"grad_norm": 14.072280883789062,
|
| 5399 |
+
"learning_rate": 7.680000000000001e-06,
|
| 5400 |
+
"loss": 68.125,
|
| 5401 |
+
"step": 769
|
| 5402 |
+
},
|
| 5403 |
+
{
|
| 5404 |
+
"epoch": 7.7e-05,
|
| 5405 |
+
"grad_norm": 13.906455993652344,
|
| 5406 |
+
"learning_rate": 7.69e-06,
|
| 5407 |
+
"loss": 68.0625,
|
| 5408 |
+
"step": 770
|
| 5409 |
+
},
|
| 5410 |
+
{
|
| 5411 |
+
"epoch": 7.71e-05,
|
| 5412 |
+
"grad_norm": 14.400522232055664,
|
| 5413 |
+
"learning_rate": 7.7e-06,
|
| 5414 |
+
"loss": 67.9375,
|
| 5415 |
+
"step": 771
|
| 5416 |
+
},
|
| 5417 |
+
{
|
| 5418 |
+
"epoch": 7.72e-05,
|
| 5419 |
+
"grad_norm": 14.000092506408691,
|
| 5420 |
+
"learning_rate": 7.709999999999999e-06,
|
| 5421 |
+
"loss": 68.0,
|
| 5422 |
+
"step": 772
|
| 5423 |
+
},
|
| 5424 |
+
{
|
| 5425 |
+
"epoch": 7.73e-05,
|
| 5426 |
+
"grad_norm": 13.759057998657227,
|
| 5427 |
+
"learning_rate": 7.72e-06,
|
| 5428 |
+
"loss": 68.125,
|
| 5429 |
+
"step": 773
|
| 5430 |
+
},
|
| 5431 |
+
{
|
| 5432 |
+
"epoch": 7.74e-05,
|
| 5433 |
+
"grad_norm": 13.556885719299316,
|
| 5434 |
+
"learning_rate": 7.73e-06,
|
| 5435 |
+
"loss": 68.1875,
|
| 5436 |
+
"step": 774
|
| 5437 |
+
},
|
| 5438 |
+
{
|
| 5439 |
+
"epoch": 7.75e-05,
|
| 5440 |
+
"grad_norm": 13.875627517700195,
|
| 5441 |
+
"learning_rate": 7.74e-06,
|
| 5442 |
+
"loss": 68.0,
|
| 5443 |
+
"step": 775
|
| 5444 |
+
},
|
| 5445 |
+
{
|
| 5446 |
+
"epoch": 7.76e-05,
|
| 5447 |
+
"grad_norm": 14.190862655639648,
|
| 5448 |
+
"learning_rate": 7.75e-06,
|
| 5449 |
+
"loss": 67.75,
|
| 5450 |
+
"step": 776
|
| 5451 |
+
},
|
| 5452 |
+
{
|
| 5453 |
+
"epoch": 7.77e-05,
|
| 5454 |
+
"grad_norm": 14.008695602416992,
|
| 5455 |
+
"learning_rate": 7.76e-06,
|
| 5456 |
+
"loss": 67.8125,
|
| 5457 |
+
"step": 777
|
| 5458 |
+
},
|
| 5459 |
+
{
|
| 5460 |
+
"epoch": 7.78e-05,
|
| 5461 |
+
"grad_norm": 13.735383033752441,
|
| 5462 |
+
"learning_rate": 7.77e-06,
|
| 5463 |
+
"loss": 67.6875,
|
| 5464 |
+
"step": 778
|
| 5465 |
+
},
|
| 5466 |
+
{
|
| 5467 |
+
"epoch": 7.79e-05,
|
| 5468 |
+
"grad_norm": 13.905145645141602,
|
| 5469 |
+
"learning_rate": 7.78e-06,
|
| 5470 |
+
"loss": 67.875,
|
| 5471 |
+
"step": 779
|
| 5472 |
+
},
|
| 5473 |
+
{
|
| 5474 |
+
"epoch": 7.8e-05,
|
| 5475 |
+
"grad_norm": 13.315927505493164,
|
| 5476 |
+
"learning_rate": 7.79e-06,
|
| 5477 |
+
"loss": 68.0,
|
| 5478 |
+
"step": 780
|
| 5479 |
+
},
|
| 5480 |
+
{
|
| 5481 |
+
"epoch": 7.81e-05,
|
| 5482 |
+
"grad_norm": 14.015225410461426,
|
| 5483 |
+
"learning_rate": 7.8e-06,
|
| 5484 |
+
"loss": 67.75,
|
| 5485 |
+
"step": 781
|
| 5486 |
+
},
|
| 5487 |
+
{
|
| 5488 |
+
"epoch": 7.82e-05,
|
| 5489 |
+
"grad_norm": 13.459209442138672,
|
| 5490 |
+
"learning_rate": 7.81e-06,
|
| 5491 |
+
"loss": 67.875,
|
| 5492 |
+
"step": 782
|
| 5493 |
+
},
|
| 5494 |
+
{
|
| 5495 |
+
"epoch": 7.83e-05,
|
| 5496 |
+
"grad_norm": 13.911338806152344,
|
| 5497 |
+
"learning_rate": 7.820000000000001e-06,
|
| 5498 |
+
"loss": 67.75,
|
| 5499 |
+
"step": 783
|
| 5500 |
+
},
|
| 5501 |
+
{
|
| 5502 |
+
"epoch": 7.84e-05,
|
| 5503 |
+
"grad_norm": 13.783612251281738,
|
| 5504 |
+
"learning_rate": 7.83e-06,
|
| 5505 |
+
"loss": 67.8125,
|
| 5506 |
+
"step": 784
|
| 5507 |
+
},
|
| 5508 |
+
{
|
| 5509 |
+
"epoch": 7.85e-05,
|
| 5510 |
+
"grad_norm": 14.066487312316895,
|
| 5511 |
+
"learning_rate": 7.840000000000001e-06,
|
| 5512 |
+
"loss": 67.375,
|
| 5513 |
+
"step": 785
|
| 5514 |
+
},
|
| 5515 |
+
{
|
| 5516 |
+
"epoch": 7.86e-05,
|
| 5517 |
+
"grad_norm": 13.927901268005371,
|
| 5518 |
+
"learning_rate": 7.850000000000001e-06,
|
| 5519 |
+
"loss": 67.5,
|
| 5520 |
+
"step": 786
|
| 5521 |
+
},
|
| 5522 |
+
{
|
| 5523 |
+
"epoch": 7.87e-05,
|
| 5524 |
+
"grad_norm": 13.599596977233887,
|
| 5525 |
+
"learning_rate": 7.86e-06,
|
| 5526 |
+
"loss": 67.5,
|
| 5527 |
+
"step": 787
|
| 5528 |
+
},
|
| 5529 |
+
{
|
| 5530 |
+
"epoch": 7.88e-05,
|
| 5531 |
+
"grad_norm": 13.849498748779297,
|
| 5532 |
+
"learning_rate": 7.870000000000001e-06,
|
| 5533 |
+
"loss": 67.4375,
|
| 5534 |
+
"step": 788
|
| 5535 |
+
},
|
| 5536 |
+
{
|
| 5537 |
+
"epoch": 7.89e-05,
|
| 5538 |
+
"grad_norm": 13.334601402282715,
|
| 5539 |
+
"learning_rate": 7.879999999999999e-06,
|
| 5540 |
+
"loss": 67.5,
|
| 5541 |
+
"step": 789
|
| 5542 |
+
},
|
| 5543 |
+
{
|
| 5544 |
+
"epoch": 7.9e-05,
|
| 5545 |
+
"grad_norm": 13.693037986755371,
|
| 5546 |
+
"learning_rate": 7.89e-06,
|
| 5547 |
+
"loss": 67.375,
|
| 5548 |
+
"step": 790
|
| 5549 |
+
},
|
| 5550 |
+
{
|
| 5551 |
+
"epoch": 7.91e-05,
|
| 5552 |
+
"grad_norm": 13.442035675048828,
|
| 5553 |
+
"learning_rate": 7.9e-06,
|
| 5554 |
+
"loss": 67.5,
|
| 5555 |
+
"step": 791
|
| 5556 |
+
},
|
| 5557 |
+
{
|
| 5558 |
+
"epoch": 7.92e-05,
|
| 5559 |
+
"grad_norm": 13.878938674926758,
|
| 5560 |
+
"learning_rate": 7.91e-06,
|
| 5561 |
+
"loss": 67.0625,
|
| 5562 |
+
"step": 792
|
| 5563 |
+
},
|
| 5564 |
+
{
|
| 5565 |
+
"epoch": 7.93e-05,
|
| 5566 |
+
"grad_norm": 13.525701522827148,
|
| 5567 |
+
"learning_rate": 7.92e-06,
|
| 5568 |
+
"loss": 67.5,
|
| 5569 |
+
"step": 793
|
| 5570 |
+
},
|
| 5571 |
+
{
|
| 5572 |
+
"epoch": 7.94e-05,
|
| 5573 |
+
"grad_norm": 13.480079650878906,
|
| 5574 |
+
"learning_rate": 7.93e-06,
|
| 5575 |
+
"loss": 67.3125,
|
| 5576 |
+
"step": 794
|
| 5577 |
+
},
|
| 5578 |
+
{
|
| 5579 |
+
"epoch": 7.95e-05,
|
| 5580 |
+
"grad_norm": 13.224320411682129,
|
| 5581 |
+
"learning_rate": 7.94e-06,
|
| 5582 |
+
"loss": 67.375,
|
| 5583 |
+
"step": 795
|
| 5584 |
+
},
|
| 5585 |
+
{
|
| 5586 |
+
"epoch": 7.96e-05,
|
| 5587 |
+
"grad_norm": 13.158392906188965,
|
| 5588 |
+
"learning_rate": 7.95e-06,
|
| 5589 |
+
"loss": 67.3125,
|
| 5590 |
+
"step": 796
|
| 5591 |
+
},
|
| 5592 |
+
{
|
| 5593 |
+
"epoch": 7.97e-05,
|
| 5594 |
+
"grad_norm": 13.183201789855957,
|
| 5595 |
+
"learning_rate": 7.96e-06,
|
| 5596 |
+
"loss": 67.375,
|
| 5597 |
+
"step": 797
|
| 5598 |
+
},
|
| 5599 |
+
{
|
| 5600 |
+
"epoch": 7.98e-05,
|
| 5601 |
+
"grad_norm": 13.698829650878906,
|
| 5602 |
+
"learning_rate": 7.97e-06,
|
| 5603 |
+
"loss": 67.0,
|
| 5604 |
+
"step": 798
|
| 5605 |
+
},
|
| 5606 |
+
{
|
| 5607 |
+
"epoch": 7.99e-05,
|
| 5608 |
+
"grad_norm": 13.33995532989502,
|
| 5609 |
+
"learning_rate": 7.98e-06,
|
| 5610 |
+
"loss": 67.0,
|
| 5611 |
+
"step": 799
|
| 5612 |
+
},
|
| 5613 |
+
{
|
| 5614 |
+
"epoch": 8e-05,
|
| 5615 |
+
"grad_norm": 13.223627090454102,
|
| 5616 |
+
"learning_rate": 7.990000000000001e-06,
|
| 5617 |
+
"loss": 67.1875,
|
| 5618 |
+
"step": 800
|
| 5619 |
+
},
|
| 5620 |
+
{
|
| 5621 |
+
"epoch": 8.01e-05,
|
| 5622 |
+
"grad_norm": 13.501330375671387,
|
| 5623 |
+
"learning_rate": 8e-06,
|
| 5624 |
+
"loss": 66.9375,
|
| 5625 |
+
"step": 801
|
| 5626 |
+
},
|
| 5627 |
+
{
|
| 5628 |
+
"epoch": 8.02e-05,
|
| 5629 |
+
"grad_norm": 13.263901710510254,
|
| 5630 |
+
"learning_rate": 8.01e-06,
|
| 5631 |
+
"loss": 66.9375,
|
| 5632 |
+
"step": 802
|
| 5633 |
+
},
|
| 5634 |
+
{
|
| 5635 |
+
"epoch": 8.03e-05,
|
| 5636 |
+
"grad_norm": 13.32174301147461,
|
| 5637 |
+
"learning_rate": 8.02e-06,
|
| 5638 |
+
"loss": 66.875,
|
| 5639 |
+
"step": 803
|
| 5640 |
+
},
|
| 5641 |
+
{
|
| 5642 |
+
"epoch": 8.04e-05,
|
| 5643 |
+
"grad_norm": 13.487302780151367,
|
| 5644 |
+
"learning_rate": 8.03e-06,
|
| 5645 |
+
"loss": 66.75,
|
| 5646 |
+
"step": 804
|
| 5647 |
+
},
|
| 5648 |
+
{
|
| 5649 |
+
"epoch": 8.05e-05,
|
| 5650 |
+
"grad_norm": 13.301477432250977,
|
| 5651 |
+
"learning_rate": 8.040000000000001e-06,
|
| 5652 |
+
"loss": 66.875,
|
| 5653 |
+
"step": 805
|
| 5654 |
+
},
|
| 5655 |
+
{
|
| 5656 |
+
"epoch": 8.06e-05,
|
| 5657 |
+
"grad_norm": 13.382040023803711,
|
| 5658 |
+
"learning_rate": 8.05e-06,
|
| 5659 |
+
"loss": 66.625,
|
| 5660 |
+
"step": 806
|
| 5661 |
+
},
|
| 5662 |
+
{
|
| 5663 |
+
"epoch": 8.07e-05,
|
| 5664 |
+
"grad_norm": 13.217137336730957,
|
| 5665 |
+
"learning_rate": 8.06e-06,
|
| 5666 |
+
"loss": 66.875,
|
| 5667 |
+
"step": 807
|
| 5668 |
+
},
|
| 5669 |
+
{
|
| 5670 |
+
"epoch": 8.08e-05,
|
| 5671 |
+
"grad_norm": 13.310396194458008,
|
| 5672 |
+
"learning_rate": 8.07e-06,
|
| 5673 |
+
"loss": 66.6875,
|
| 5674 |
+
"step": 808
|
| 5675 |
+
},
|
| 5676 |
+
{
|
| 5677 |
+
"epoch": 8.09e-05,
|
| 5678 |
+
"grad_norm": 12.70387077331543,
|
| 5679 |
+
"learning_rate": 8.079999999999999e-06,
|
| 5680 |
+
"loss": 67.0,
|
| 5681 |
+
"step": 809
|
| 5682 |
+
},
|
| 5683 |
+
{
|
| 5684 |
+
"epoch": 8.1e-05,
|
| 5685 |
+
"grad_norm": 13.554513931274414,
|
| 5686 |
+
"learning_rate": 8.09e-06,
|
| 5687 |
+
"loss": 66.4375,
|
| 5688 |
+
"step": 810
|
| 5689 |
+
},
|
| 5690 |
+
{
|
| 5691 |
+
"epoch": 8.11e-05,
|
| 5692 |
+
"grad_norm": 13.117542266845703,
|
| 5693 |
+
"learning_rate": 8.1e-06,
|
| 5694 |
+
"loss": 66.5625,
|
| 5695 |
+
"step": 811
|
| 5696 |
+
},
|
| 5697 |
+
{
|
| 5698 |
+
"epoch": 8.12e-05,
|
| 5699 |
+
"grad_norm": 12.712786674499512,
|
| 5700 |
+
"learning_rate": 8.11e-06,
|
| 5701 |
+
"loss": 66.6875,
|
| 5702 |
+
"step": 812
|
| 5703 |
+
},
|
| 5704 |
+
{
|
| 5705 |
+
"epoch": 8.13e-05,
|
| 5706 |
+
"grad_norm": 13.098764419555664,
|
| 5707 |
+
"learning_rate": 8.12e-06,
|
| 5708 |
+
"loss": 66.5,
|
| 5709 |
+
"step": 813
|
| 5710 |
+
},
|
| 5711 |
+
{
|
| 5712 |
+
"epoch": 8.14e-05,
|
| 5713 |
+
"grad_norm": 13.022997856140137,
|
| 5714 |
+
"learning_rate": 8.13e-06,
|
| 5715 |
+
"loss": 66.4375,
|
| 5716 |
+
"step": 814
|
| 5717 |
+
},
|
| 5718 |
+
{
|
| 5719 |
+
"epoch": 8.15e-05,
|
| 5720 |
+
"grad_norm": 12.858806610107422,
|
| 5721 |
+
"learning_rate": 8.14e-06,
|
| 5722 |
+
"loss": 66.625,
|
| 5723 |
+
"step": 815
|
| 5724 |
+
},
|
| 5725 |
+
{
|
| 5726 |
+
"epoch": 8.16e-05,
|
| 5727 |
+
"grad_norm": 13.087048530578613,
|
| 5728 |
+
"learning_rate": 8.15e-06,
|
| 5729 |
+
"loss": 66.5,
|
| 5730 |
+
"step": 816
|
| 5731 |
+
},
|
| 5732 |
+
{
|
| 5733 |
+
"epoch": 8.17e-05,
|
| 5734 |
+
"grad_norm": 12.747725486755371,
|
| 5735 |
+
"learning_rate": 8.160000000000001e-06,
|
| 5736 |
+
"loss": 66.4375,
|
| 5737 |
+
"step": 817
|
| 5738 |
+
},
|
| 5739 |
+
{
|
| 5740 |
+
"epoch": 8.18e-05,
|
| 5741 |
+
"grad_norm": 13.237361907958984,
|
| 5742 |
+
"learning_rate": 8.17e-06,
|
| 5743 |
+
"loss": 66.1875,
|
| 5744 |
+
"step": 818
|
| 5745 |
+
},
|
| 5746 |
+
{
|
| 5747 |
+
"epoch": 8.19e-05,
|
| 5748 |
+
"grad_norm": 12.671786308288574,
|
| 5749 |
+
"learning_rate": 8.18e-06,
|
| 5750 |
+
"loss": 66.4375,
|
| 5751 |
+
"step": 819
|
| 5752 |
+
},
|
| 5753 |
+
{
|
| 5754 |
+
"epoch": 8.2e-05,
|
| 5755 |
+
"grad_norm": 12.85875129699707,
|
| 5756 |
+
"learning_rate": 8.19e-06,
|
| 5757 |
+
"loss": 66.3125,
|
| 5758 |
+
"step": 820
|
| 5759 |
+
},
|
| 5760 |
+
{
|
| 5761 |
+
"epoch": 8.21e-05,
|
| 5762 |
+
"grad_norm": 12.405821800231934,
|
| 5763 |
+
"learning_rate": 8.2e-06,
|
| 5764 |
+
"loss": 66.5,
|
| 5765 |
+
"step": 821
|
| 5766 |
+
},
|
| 5767 |
+
{
|
| 5768 |
+
"epoch": 8.22e-05,
|
| 5769 |
+
"grad_norm": 13.07856559753418,
|
| 5770 |
+
"learning_rate": 8.210000000000001e-06,
|
| 5771 |
+
"loss": 66.0625,
|
| 5772 |
+
"step": 822
|
| 5773 |
+
},
|
| 5774 |
+
{
|
| 5775 |
+
"epoch": 8.23e-05,
|
| 5776 |
+
"grad_norm": 13.018475532531738,
|
| 5777 |
+
"learning_rate": 8.22e-06,
|
| 5778 |
+
"loss": 66.1875,
|
| 5779 |
+
"step": 823
|
| 5780 |
+
},
|
| 5781 |
+
{
|
| 5782 |
+
"epoch": 8.24e-05,
|
| 5783 |
+
"grad_norm": 13.105154037475586,
|
| 5784 |
+
"learning_rate": 8.23e-06,
|
| 5785 |
+
"loss": 65.9375,
|
| 5786 |
+
"step": 824
|
| 5787 |
+
},
|
| 5788 |
+
{
|
| 5789 |
+
"epoch": 8.25e-05,
|
| 5790 |
+
"grad_norm": 12.811441421508789,
|
| 5791 |
+
"learning_rate": 8.24e-06,
|
| 5792 |
+
"loss": 66.0,
|
| 5793 |
+
"step": 825
|
| 5794 |
+
},
|
| 5795 |
+
{
|
| 5796 |
+
"epoch": 8.26e-05,
|
| 5797 |
+
"grad_norm": 12.24394702911377,
|
| 5798 |
+
"learning_rate": 8.249999999999999e-06,
|
| 5799 |
+
"loss": 66.5,
|
| 5800 |
+
"step": 826
|
| 5801 |
+
},
|
| 5802 |
+
{
|
| 5803 |
+
"epoch": 8.27e-05,
|
| 5804 |
+
"grad_norm": 12.856313705444336,
|
| 5805 |
+
"learning_rate": 8.26e-06,
|
| 5806 |
+
"loss": 66.1875,
|
| 5807 |
+
"step": 827
|
| 5808 |
+
},
|
| 5809 |
+
{
|
| 5810 |
+
"epoch": 8.28e-05,
|
| 5811 |
+
"grad_norm": 12.286417007446289,
|
| 5812 |
+
"learning_rate": 8.27e-06,
|
| 5813 |
+
"loss": 66.0625,
|
| 5814 |
+
"step": 828
|
| 5815 |
+
},
|
| 5816 |
+
{
|
| 5817 |
+
"epoch": 8.29e-05,
|
| 5818 |
+
"grad_norm": 12.537483215332031,
|
| 5819 |
+
"learning_rate": 8.28e-06,
|
| 5820 |
+
"loss": 65.9375,
|
| 5821 |
+
"step": 829
|
| 5822 |
+
},
|
| 5823 |
+
{
|
| 5824 |
+
"epoch": 8.3e-05,
|
| 5825 |
+
"grad_norm": 12.375907897949219,
|
| 5826 |
+
"learning_rate": 8.29e-06,
|
| 5827 |
+
"loss": 65.875,
|
| 5828 |
+
"step": 830
|
| 5829 |
+
},
|
| 5830 |
+
{
|
| 5831 |
+
"epoch": 8.31e-05,
|
| 5832 |
+
"grad_norm": 12.420113563537598,
|
| 5833 |
+
"learning_rate": 8.3e-06,
|
| 5834 |
+
"loss": 66.0,
|
| 5835 |
+
"step": 831
|
| 5836 |
+
},
|
| 5837 |
+
{
|
| 5838 |
+
"epoch": 8.32e-05,
|
| 5839 |
+
"grad_norm": 12.373382568359375,
|
| 5840 |
+
"learning_rate": 8.31e-06,
|
| 5841 |
+
"loss": 65.6875,
|
| 5842 |
+
"step": 832
|
| 5843 |
+
},
|
| 5844 |
+
{
|
| 5845 |
+
"epoch": 8.33e-05,
|
| 5846 |
+
"grad_norm": 11.9176607131958,
|
| 5847 |
+
"learning_rate": 8.32e-06,
|
| 5848 |
+
"loss": 65.875,
|
| 5849 |
+
"step": 833
|
| 5850 |
+
},
|
| 5851 |
+
{
|
| 5852 |
+
"epoch": 8.34e-05,
|
| 5853 |
+
"grad_norm": 12.355897903442383,
|
| 5854 |
+
"learning_rate": 8.330000000000002e-06,
|
| 5855 |
+
"loss": 65.9375,
|
| 5856 |
+
"step": 834
|
| 5857 |
+
},
|
| 5858 |
+
{
|
| 5859 |
+
"epoch": 8.35e-05,
|
| 5860 |
+
"grad_norm": 12.412797927856445,
|
| 5861 |
+
"learning_rate": 8.34e-06,
|
| 5862 |
+
"loss": 65.75,
|
| 5863 |
+
"step": 835
|
| 5864 |
+
},
|
| 5865 |
+
{
|
| 5866 |
+
"epoch": 8.36e-05,
|
| 5867 |
+
"grad_norm": 12.29367733001709,
|
| 5868 |
+
"learning_rate": 8.35e-06,
|
| 5869 |
+
"loss": 65.8125,
|
| 5870 |
+
"step": 836
|
| 5871 |
+
},
|
| 5872 |
+
{
|
| 5873 |
+
"epoch": 8.37e-05,
|
| 5874 |
+
"grad_norm": 12.362329483032227,
|
| 5875 |
+
"learning_rate": 8.36e-06,
|
| 5876 |
+
"loss": 65.5625,
|
| 5877 |
+
"step": 837
|
| 5878 |
+
},
|
| 5879 |
+
{
|
| 5880 |
+
"epoch": 8.38e-05,
|
| 5881 |
+
"grad_norm": 12.16931438446045,
|
| 5882 |
+
"learning_rate": 8.37e-06,
|
| 5883 |
+
"loss": 65.875,
|
| 5884 |
+
"step": 838
|
| 5885 |
+
},
|
| 5886 |
+
{
|
| 5887 |
+
"epoch": 8.39e-05,
|
| 5888 |
+
"grad_norm": 12.073834419250488,
|
| 5889 |
+
"learning_rate": 8.380000000000001e-06,
|
| 5890 |
+
"loss": 65.6875,
|
| 5891 |
+
"step": 839
|
| 5892 |
+
},
|
| 5893 |
+
{
|
| 5894 |
+
"epoch": 8.4e-05,
|
| 5895 |
+
"grad_norm": 12.422995567321777,
|
| 5896 |
+
"learning_rate": 8.39e-06,
|
| 5897 |
+
"loss": 65.5,
|
| 5898 |
+
"step": 840
|
| 5899 |
+
},
|
| 5900 |
+
{
|
| 5901 |
+
"epoch": 8.41e-05,
|
| 5902 |
+
"grad_norm": 12.235881805419922,
|
| 5903 |
+
"learning_rate": 8.400000000000001e-06,
|
| 5904 |
+
"loss": 65.5,
|
| 5905 |
+
"step": 841
|
| 5906 |
+
},
|
| 5907 |
+
{
|
| 5908 |
+
"epoch": 8.42e-05,
|
| 5909 |
+
"grad_norm": 11.909436225891113,
|
| 5910 |
+
"learning_rate": 8.41e-06,
|
| 5911 |
+
"loss": 65.6875,
|
| 5912 |
+
"step": 842
|
| 5913 |
+
},
|
| 5914 |
+
{
|
| 5915 |
+
"epoch": 8.43e-05,
|
| 5916 |
+
"grad_norm": 12.046891212463379,
|
| 5917 |
+
"learning_rate": 8.419999999999999e-06,
|
| 5918 |
+
"loss": 65.6875,
|
| 5919 |
+
"step": 843
|
| 5920 |
+
},
|
| 5921 |
+
{
|
| 5922 |
+
"epoch": 8.44e-05,
|
| 5923 |
+
"grad_norm": 12.069416999816895,
|
| 5924 |
+
"learning_rate": 8.43e-06,
|
| 5925 |
+
"loss": 65.25,
|
| 5926 |
+
"step": 844
|
| 5927 |
+
},
|
| 5928 |
+
{
|
| 5929 |
+
"epoch": 8.45e-05,
|
| 5930 |
+
"grad_norm": 11.80577278137207,
|
| 5931 |
+
"learning_rate": 8.44e-06,
|
| 5932 |
+
"loss": 65.375,
|
| 5933 |
+
"step": 845
|
| 5934 |
+
},
|
| 5935 |
+
{
|
| 5936 |
+
"epoch": 8.46e-05,
|
| 5937 |
+
"grad_norm": 11.65450382232666,
|
| 5938 |
+
"learning_rate": 8.45e-06,
|
| 5939 |
+
"loss": 65.5,
|
| 5940 |
+
"step": 846
|
| 5941 |
+
},
|
| 5942 |
+
{
|
| 5943 |
+
"epoch": 8.47e-05,
|
| 5944 |
+
"grad_norm": 11.70303726196289,
|
| 5945 |
+
"learning_rate": 8.46e-06,
|
| 5946 |
+
"loss": 65.4375,
|
| 5947 |
+
"step": 847
|
| 5948 |
+
},
|
| 5949 |
+
{
|
| 5950 |
+
"epoch": 8.48e-05,
|
| 5951 |
+
"grad_norm": 12.221776008605957,
|
| 5952 |
+
"learning_rate": 8.47e-06,
|
| 5953 |
+
"loss": 65.25,
|
| 5954 |
+
"step": 848
|
| 5955 |
+
},
|
| 5956 |
+
{
|
| 5957 |
+
"epoch": 8.49e-05,
|
| 5958 |
+
"grad_norm": 11.785761833190918,
|
| 5959 |
+
"learning_rate": 8.48e-06,
|
| 5960 |
+
"loss": 65.4375,
|
| 5961 |
+
"step": 849
|
| 5962 |
+
},
|
| 5963 |
+
{
|
| 5964 |
+
"epoch": 8.5e-05,
|
| 5965 |
+
"grad_norm": 12.039600372314453,
|
| 5966 |
+
"learning_rate": 8.49e-06,
|
| 5967 |
+
"loss": 65.25,
|
| 5968 |
+
"step": 850
|
| 5969 |
+
},
|
| 5970 |
+
{
|
| 5971 |
+
"epoch": 8.51e-05,
|
| 5972 |
+
"grad_norm": 11.632035255432129,
|
| 5973 |
+
"learning_rate": 8.500000000000002e-06,
|
| 5974 |
+
"loss": 65.4375,
|
| 5975 |
+
"step": 851
|
| 5976 |
+
},
|
| 5977 |
+
{
|
| 5978 |
+
"epoch": 8.52e-05,
|
| 5979 |
+
"grad_norm": 11.960968017578125,
|
| 5980 |
+
"learning_rate": 8.51e-06,
|
| 5981 |
+
"loss": 65.125,
|
| 5982 |
+
"step": 852
|
| 5983 |
+
},
|
| 5984 |
+
{
|
| 5985 |
+
"epoch": 8.53e-05,
|
| 5986 |
+
"grad_norm": 11.791102409362793,
|
| 5987 |
+
"learning_rate": 8.52e-06,
|
| 5988 |
+
"loss": 65.0625,
|
| 5989 |
+
"step": 853
|
| 5990 |
+
},
|
| 5991 |
+
{
|
| 5992 |
+
"epoch": 8.54e-05,
|
| 5993 |
+
"grad_norm": 11.28836727142334,
|
| 5994 |
+
"learning_rate": 8.53e-06,
|
| 5995 |
+
"loss": 65.5,
|
| 5996 |
+
"step": 854
|
| 5997 |
+
},
|
| 5998 |
+
{
|
| 5999 |
+
"epoch": 8.55e-05,
|
| 6000 |
+
"grad_norm": 11.553174018859863,
|
| 6001 |
+
"learning_rate": 8.54e-06,
|
| 6002 |
+
"loss": 65.125,
|
| 6003 |
+
"step": 855
|
| 6004 |
+
},
|
| 6005 |
+
{
|
| 6006 |
+
"epoch": 8.56e-05,
|
| 6007 |
+
"grad_norm": 11.61713981628418,
|
| 6008 |
+
"learning_rate": 8.550000000000001e-06,
|
| 6009 |
+
"loss": 65.0625,
|
| 6010 |
+
"step": 856
|
| 6011 |
+
},
|
| 6012 |
+
{
|
| 6013 |
+
"epoch": 8.57e-05,
|
| 6014 |
+
"grad_norm": 11.704262733459473,
|
| 6015 |
+
"learning_rate": 8.56e-06,
|
| 6016 |
+
"loss": 64.8125,
|
| 6017 |
+
"step": 857
|
| 6018 |
+
},
|
| 6019 |
+
{
|
| 6020 |
+
"epoch": 8.58e-05,
|
| 6021 |
+
"grad_norm": 11.702911376953125,
|
| 6022 |
+
"learning_rate": 8.57e-06,
|
| 6023 |
+
"loss": 64.875,
|
| 6024 |
+
"step": 858
|
| 6025 |
+
},
|
| 6026 |
+
{
|
| 6027 |
+
"epoch": 8.59e-05,
|
| 6028 |
+
"grad_norm": 11.293460845947266,
|
| 6029 |
+
"learning_rate": 8.580000000000001e-06,
|
| 6030 |
+
"loss": 65.0,
|
| 6031 |
+
"step": 859
|
| 6032 |
+
},
|
| 6033 |
+
{
|
| 6034 |
+
"epoch": 8.6e-05,
|
| 6035 |
+
"grad_norm": 11.456600189208984,
|
| 6036 |
+
"learning_rate": 8.589999999999999e-06,
|
| 6037 |
+
"loss": 64.7812,
|
| 6038 |
+
"step": 860
|
| 6039 |
+
},
|
| 6040 |
+
{
|
| 6041 |
+
"epoch": 8.61e-05,
|
| 6042 |
+
"grad_norm": 11.203218460083008,
|
| 6043 |
+
"learning_rate": 8.6e-06,
|
| 6044 |
+
"loss": 65.125,
|
| 6045 |
+
"step": 861
|
| 6046 |
+
},
|
| 6047 |
+
{
|
| 6048 |
+
"epoch": 8.62e-05,
|
| 6049 |
+
"grad_norm": 11.105107307434082,
|
| 6050 |
+
"learning_rate": 8.61e-06,
|
| 6051 |
+
"loss": 65.0,
|
| 6052 |
+
"step": 862
|
| 6053 |
+
},
|
| 6054 |
+
{
|
| 6055 |
+
"epoch": 8.63e-05,
|
| 6056 |
+
"grad_norm": 11.6665620803833,
|
| 6057 |
+
"learning_rate": 8.62e-06,
|
| 6058 |
+
"loss": 64.625,
|
| 6059 |
+
"step": 863
|
| 6060 |
+
},
|
| 6061 |
+
{
|
| 6062 |
+
"epoch": 8.64e-05,
|
| 6063 |
+
"grad_norm": 11.519858360290527,
|
| 6064 |
+
"learning_rate": 8.63e-06,
|
| 6065 |
+
"loss": 64.8125,
|
| 6066 |
+
"step": 864
|
| 6067 |
+
},
|
| 6068 |
+
{
|
| 6069 |
+
"epoch": 8.65e-05,
|
| 6070 |
+
"grad_norm": 11.364030838012695,
|
| 6071 |
+
"learning_rate": 8.64e-06,
|
| 6072 |
+
"loss": 64.8125,
|
| 6073 |
+
"step": 865
|
| 6074 |
+
},
|
| 6075 |
+
{
|
| 6076 |
+
"epoch": 8.66e-05,
|
| 6077 |
+
"grad_norm": 11.232035636901855,
|
| 6078 |
+
"learning_rate": 8.65e-06,
|
| 6079 |
+
"loss": 64.8125,
|
| 6080 |
+
"step": 866
|
| 6081 |
+
},
|
| 6082 |
+
{
|
| 6083 |
+
"epoch": 8.67e-05,
|
| 6084 |
+
"grad_norm": 11.68748950958252,
|
| 6085 |
+
"learning_rate": 8.66e-06,
|
| 6086 |
+
"loss": 64.5,
|
| 6087 |
+
"step": 867
|
| 6088 |
+
},
|
| 6089 |
+
{
|
| 6090 |
+
"epoch": 8.68e-05,
|
| 6091 |
+
"grad_norm": 10.944489479064941,
|
| 6092 |
+
"learning_rate": 8.67e-06,
|
| 6093 |
+
"loss": 64.875,
|
| 6094 |
+
"step": 868
|
| 6095 |
+
},
|
| 6096 |
+
{
|
| 6097 |
+
"epoch": 8.69e-05,
|
| 6098 |
+
"grad_norm": 11.102535247802734,
|
| 6099 |
+
"learning_rate": 8.68e-06,
|
| 6100 |
+
"loss": 64.5,
|
| 6101 |
+
"step": 869
|
| 6102 |
+
},
|
| 6103 |
+
{
|
| 6104 |
+
"epoch": 8.7e-05,
|
| 6105 |
+
"grad_norm": 10.927079200744629,
|
| 6106 |
+
"learning_rate": 8.69e-06,
|
| 6107 |
+
"loss": 64.9375,
|
| 6108 |
+
"step": 870
|
| 6109 |
+
},
|
| 6110 |
+
{
|
| 6111 |
+
"epoch": 8.71e-05,
|
| 6112 |
+
"grad_norm": 10.939054489135742,
|
| 6113 |
+
"learning_rate": 8.7e-06,
|
| 6114 |
+
"loss": 64.4688,
|
| 6115 |
+
"step": 871
|
| 6116 |
+
},
|
| 6117 |
+
{
|
| 6118 |
+
"epoch": 8.72e-05,
|
| 6119 |
+
"grad_norm": 11.497695922851562,
|
| 6120 |
+
"learning_rate": 8.71e-06,
|
| 6121 |
+
"loss": 64.625,
|
| 6122 |
+
"step": 872
|
| 6123 |
+
},
|
| 6124 |
+
{
|
| 6125 |
+
"epoch": 8.73e-05,
|
| 6126 |
+
"grad_norm": 11.157756805419922,
|
| 6127 |
+
"learning_rate": 8.720000000000001e-06,
|
| 6128 |
+
"loss": 64.8125,
|
| 6129 |
+
"step": 873
|
| 6130 |
+
},
|
| 6131 |
+
{
|
| 6132 |
+
"epoch": 8.74e-05,
|
| 6133 |
+
"grad_norm": 10.892260551452637,
|
| 6134 |
+
"learning_rate": 8.73e-06,
|
| 6135 |
+
"loss": 64.5,
|
| 6136 |
+
"step": 874
|
| 6137 |
+
},
|
| 6138 |
+
{
|
| 6139 |
+
"epoch": 8.75e-05,
|
| 6140 |
+
"grad_norm": 10.81840991973877,
|
| 6141 |
+
"learning_rate": 8.74e-06,
|
| 6142 |
+
"loss": 64.625,
|
| 6143 |
+
"step": 875
|
| 6144 |
+
},
|
| 6145 |
+
{
|
| 6146 |
+
"epoch": 8.76e-05,
|
| 6147 |
+
"grad_norm": 10.906977653503418,
|
| 6148 |
+
"learning_rate": 8.750000000000001e-06,
|
| 6149 |
+
"loss": 64.4062,
|
| 6150 |
+
"step": 876
|
| 6151 |
+
},
|
| 6152 |
+
{
|
| 6153 |
+
"epoch": 8.77e-05,
|
| 6154 |
+
"grad_norm": 10.821674346923828,
|
| 6155 |
+
"learning_rate": 8.759999999999999e-06,
|
| 6156 |
+
"loss": 64.5,
|
| 6157 |
+
"step": 877
|
| 6158 |
+
},
|
| 6159 |
+
{
|
| 6160 |
+
"epoch": 8.78e-05,
|
| 6161 |
+
"grad_norm": 10.746541023254395,
|
| 6162 |
+
"learning_rate": 8.77e-06,
|
| 6163 |
+
"loss": 64.5625,
|
| 6164 |
+
"step": 878
|
| 6165 |
+
},
|
| 6166 |
+
{
|
| 6167 |
+
"epoch": 8.79e-05,
|
| 6168 |
+
"grad_norm": 11.237996101379395,
|
| 6169 |
+
"learning_rate": 8.78e-06,
|
| 6170 |
+
"loss": 64.125,
|
| 6171 |
+
"step": 879
|
| 6172 |
+
},
|
| 6173 |
+
{
|
| 6174 |
+
"epoch": 8.8e-05,
|
| 6175 |
+
"grad_norm": 11.085122108459473,
|
| 6176 |
+
"learning_rate": 8.79e-06,
|
| 6177 |
+
"loss": 64.25,
|
| 6178 |
+
"step": 880
|
| 6179 |
+
},
|
| 6180 |
+
{
|
| 6181 |
+
"epoch": 8.81e-05,
|
| 6182 |
+
"grad_norm": 10.644852638244629,
|
| 6183 |
+
"learning_rate": 8.8e-06,
|
| 6184 |
+
"loss": 64.4062,
|
| 6185 |
+
"step": 881
|
| 6186 |
+
},
|
| 6187 |
+
{
|
| 6188 |
+
"epoch": 8.82e-05,
|
| 6189 |
+
"grad_norm": 10.829193115234375,
|
| 6190 |
+
"learning_rate": 8.81e-06,
|
| 6191 |
+
"loss": 64.0938,
|
| 6192 |
+
"step": 882
|
| 6193 |
+
},
|
| 6194 |
+
{
|
| 6195 |
+
"epoch": 8.83e-05,
|
| 6196 |
+
"grad_norm": 10.780670166015625,
|
| 6197 |
+
"learning_rate": 8.82e-06,
|
| 6198 |
+
"loss": 64.4062,
|
| 6199 |
+
"step": 883
|
| 6200 |
+
},
|
| 6201 |
+
{
|
| 6202 |
+
"epoch": 8.84e-05,
|
| 6203 |
+
"grad_norm": 10.732144355773926,
|
| 6204 |
+
"learning_rate": 8.83e-06,
|
| 6205 |
+
"loss": 64.4375,
|
| 6206 |
+
"step": 884
|
| 6207 |
+
},
|
| 6208 |
+
{
|
| 6209 |
+
"epoch": 8.85e-05,
|
| 6210 |
+
"grad_norm": 10.749310493469238,
|
| 6211 |
+
"learning_rate": 8.84e-06,
|
| 6212 |
+
"loss": 64.0625,
|
| 6213 |
+
"step": 885
|
| 6214 |
+
},
|
| 6215 |
+
{
|
| 6216 |
+
"epoch": 8.86e-05,
|
| 6217 |
+
"grad_norm": 10.903584480285645,
|
| 6218 |
+
"learning_rate": 8.85e-06,
|
| 6219 |
+
"loss": 63.8125,
|
| 6220 |
+
"step": 886
|
| 6221 |
+
},
|
| 6222 |
+
{
|
| 6223 |
+
"epoch": 8.87e-05,
|
| 6224 |
+
"grad_norm": 10.870782852172852,
|
| 6225 |
+
"learning_rate": 8.86e-06,
|
| 6226 |
+
"loss": 63.5625,
|
| 6227 |
+
"step": 887
|
| 6228 |
+
},
|
| 6229 |
+
{
|
| 6230 |
+
"epoch": 8.88e-05,
|
| 6231 |
+
"grad_norm": 10.494131088256836,
|
| 6232 |
+
"learning_rate": 8.87e-06,
|
| 6233 |
+
"loss": 64.2188,
|
| 6234 |
+
"step": 888
|
| 6235 |
+
},
|
| 6236 |
+
{
|
| 6237 |
+
"epoch": 8.89e-05,
|
| 6238 |
+
"grad_norm": 10.62818717956543,
|
| 6239 |
+
"learning_rate": 8.88e-06,
|
| 6240 |
+
"loss": 64.3438,
|
| 6241 |
+
"step": 889
|
| 6242 |
+
},
|
| 6243 |
+
{
|
| 6244 |
+
"epoch": 8.9e-05,
|
| 6245 |
+
"grad_norm": 10.806772232055664,
|
| 6246 |
+
"learning_rate": 8.890000000000001e-06,
|
| 6247 |
+
"loss": 63.7812,
|
| 6248 |
+
"step": 890
|
| 6249 |
+
},
|
| 6250 |
+
{
|
| 6251 |
+
"epoch": 8.91e-05,
|
| 6252 |
+
"grad_norm": 10.955464363098145,
|
| 6253 |
+
"learning_rate": 8.9e-06,
|
| 6254 |
+
"loss": 63.75,
|
| 6255 |
+
"step": 891
|
| 6256 |
+
},
|
| 6257 |
+
{
|
| 6258 |
+
"epoch": 8.92e-05,
|
| 6259 |
+
"grad_norm": 10.537474632263184,
|
| 6260 |
+
"learning_rate": 8.91e-06,
|
| 6261 |
+
"loss": 63.6875,
|
| 6262 |
+
"step": 892
|
| 6263 |
+
},
|
| 6264 |
+
{
|
| 6265 |
+
"epoch": 8.93e-05,
|
| 6266 |
+
"grad_norm": 10.190065383911133,
|
| 6267 |
+
"learning_rate": 8.920000000000001e-06,
|
| 6268 |
+
"loss": 64.1562,
|
| 6269 |
+
"step": 893
|
| 6270 |
+
},
|
| 6271 |
+
{
|
| 6272 |
+
"epoch": 8.94e-05,
|
| 6273 |
+
"grad_norm": 10.386507987976074,
|
| 6274 |
+
"learning_rate": 8.93e-06,
|
| 6275 |
+
"loss": 63.9375,
|
| 6276 |
+
"step": 894
|
| 6277 |
+
},
|
| 6278 |
+
{
|
| 6279 |
+
"epoch": 8.95e-05,
|
| 6280 |
+
"grad_norm": 10.566305160522461,
|
| 6281 |
+
"learning_rate": 8.94e-06,
|
| 6282 |
+
"loss": 63.5,
|
| 6283 |
+
"step": 895
|
| 6284 |
+
},
|
| 6285 |
+
{
|
| 6286 |
+
"epoch": 8.96e-05,
|
| 6287 |
+
"grad_norm": 10.46068000793457,
|
| 6288 |
+
"learning_rate": 8.95e-06,
|
| 6289 |
+
"loss": 63.6562,
|
| 6290 |
+
"step": 896
|
| 6291 |
+
},
|
| 6292 |
+
{
|
| 6293 |
+
"epoch": 8.97e-05,
|
| 6294 |
+
"grad_norm": 10.569717407226562,
|
| 6295 |
+
"learning_rate": 8.96e-06,
|
| 6296 |
+
"loss": 63.5625,
|
| 6297 |
+
"step": 897
|
| 6298 |
+
},
|
| 6299 |
+
{
|
| 6300 |
+
"epoch": 8.98e-05,
|
| 6301 |
+
"grad_norm": 10.547991752624512,
|
| 6302 |
+
"learning_rate": 8.97e-06,
|
| 6303 |
+
"loss": 63.625,
|
| 6304 |
+
"step": 898
|
| 6305 |
+
},
|
| 6306 |
+
{
|
| 6307 |
+
"epoch": 8.99e-05,
|
| 6308 |
+
"grad_norm": 10.377073287963867,
|
| 6309 |
+
"learning_rate": 8.98e-06,
|
| 6310 |
+
"loss": 63.4688,
|
| 6311 |
+
"step": 899
|
| 6312 |
+
},
|
| 6313 |
+
{
|
| 6314 |
+
"epoch": 9e-05,
|
| 6315 |
+
"grad_norm": 10.192441940307617,
|
| 6316 |
+
"learning_rate": 8.99e-06,
|
| 6317 |
+
"loss": 63.75,
|
| 6318 |
+
"step": 900
|
| 6319 |
+
},
|
| 6320 |
+
{
|
| 6321 |
+
"epoch": 9.01e-05,
|
| 6322 |
+
"grad_norm": 10.346672058105469,
|
| 6323 |
+
"learning_rate": 9e-06,
|
| 6324 |
+
"loss": 63.5312,
|
| 6325 |
+
"step": 901
|
| 6326 |
+
},
|
| 6327 |
+
{
|
| 6328 |
+
"epoch": 9.02e-05,
|
| 6329 |
+
"grad_norm": 10.225927352905273,
|
| 6330 |
+
"learning_rate": 9.01e-06,
|
| 6331 |
+
"loss": 63.5,
|
| 6332 |
+
"step": 902
|
| 6333 |
+
},
|
| 6334 |
+
{
|
| 6335 |
+
"epoch": 9.03e-05,
|
| 6336 |
+
"grad_norm": 10.153120994567871,
|
| 6337 |
+
"learning_rate": 9.02e-06,
|
| 6338 |
+
"loss": 63.6562,
|
| 6339 |
+
"step": 903
|
| 6340 |
+
},
|
| 6341 |
+
{
|
| 6342 |
+
"epoch": 9.04e-05,
|
| 6343 |
+
"grad_norm": 10.05332088470459,
|
| 6344 |
+
"learning_rate": 9.03e-06,
|
| 6345 |
+
"loss": 63.6875,
|
| 6346 |
+
"step": 904
|
| 6347 |
+
},
|
| 6348 |
+
{
|
| 6349 |
+
"epoch": 9.05e-05,
|
| 6350 |
+
"grad_norm": 10.070221900939941,
|
| 6351 |
+
"learning_rate": 9.04e-06,
|
| 6352 |
+
"loss": 63.5938,
|
| 6353 |
+
"step": 905
|
| 6354 |
+
},
|
| 6355 |
+
{
|
| 6356 |
+
"epoch": 9.06e-05,
|
| 6357 |
+
"grad_norm": 10.313018798828125,
|
| 6358 |
+
"learning_rate": 9.05e-06,
|
| 6359 |
+
"loss": 63.3438,
|
| 6360 |
+
"step": 906
|
| 6361 |
+
},
|
| 6362 |
+
{
|
| 6363 |
+
"epoch": 9.07e-05,
|
| 6364 |
+
"grad_norm": 10.222672462463379,
|
| 6365 |
+
"learning_rate": 9.060000000000001e-06,
|
| 6366 |
+
"loss": 63.3438,
|
| 6367 |
+
"step": 907
|
| 6368 |
+
},
|
| 6369 |
+
{
|
| 6370 |
+
"epoch": 9.08e-05,
|
| 6371 |
+
"grad_norm": 10.298966407775879,
|
| 6372 |
+
"learning_rate": 9.07e-06,
|
| 6373 |
+
"loss": 63.3125,
|
| 6374 |
+
"step": 908
|
| 6375 |
+
},
|
| 6376 |
+
{
|
| 6377 |
+
"epoch": 9.09e-05,
|
| 6378 |
+
"grad_norm": 9.815750122070312,
|
| 6379 |
+
"learning_rate": 9.08e-06,
|
| 6380 |
+
"loss": 63.625,
|
| 6381 |
+
"step": 909
|
| 6382 |
+
},
|
| 6383 |
+
{
|
| 6384 |
+
"epoch": 9.1e-05,
|
| 6385 |
+
"grad_norm": 9.667336463928223,
|
| 6386 |
+
"learning_rate": 9.090000000000001e-06,
|
| 6387 |
+
"loss": 63.625,
|
| 6388 |
+
"step": 910
|
| 6389 |
+
},
|
| 6390 |
+
{
|
| 6391 |
+
"epoch": 9.11e-05,
|
| 6392 |
+
"grad_norm": 9.815160751342773,
|
| 6393 |
+
"learning_rate": 9.1e-06,
|
| 6394 |
+
"loss": 63.4062,
|
| 6395 |
+
"step": 911
|
| 6396 |
+
},
|
| 6397 |
+
{
|
| 6398 |
+
"epoch": 9.12e-05,
|
| 6399 |
+
"grad_norm": 10.02402114868164,
|
| 6400 |
+
"learning_rate": 9.110000000000001e-06,
|
| 6401 |
+
"loss": 63.1562,
|
| 6402 |
+
"step": 912
|
| 6403 |
+
},
|
| 6404 |
+
{
|
| 6405 |
+
"epoch": 9.13e-05,
|
| 6406 |
+
"grad_norm": 10.039457321166992,
|
| 6407 |
+
"learning_rate": 9.12e-06,
|
| 6408 |
+
"loss": 63.0,
|
| 6409 |
+
"step": 913
|
| 6410 |
+
},
|
| 6411 |
+
{
|
| 6412 |
+
"epoch": 9.14e-05,
|
| 6413 |
+
"grad_norm": 10.062280654907227,
|
| 6414 |
+
"learning_rate": 9.13e-06,
|
| 6415 |
+
"loss": 63.2812,
|
| 6416 |
+
"step": 914
|
| 6417 |
+
},
|
| 6418 |
+
{
|
| 6419 |
+
"epoch": 9.15e-05,
|
| 6420 |
+
"grad_norm": 9.866779327392578,
|
| 6421 |
+
"learning_rate": 9.14e-06,
|
| 6422 |
+
"loss": 63.5,
|
| 6423 |
+
"step": 915
|
| 6424 |
+
},
|
| 6425 |
+
{
|
| 6426 |
+
"epoch": 9.16e-05,
|
| 6427 |
+
"grad_norm": 9.92796516418457,
|
| 6428 |
+
"learning_rate": 9.15e-06,
|
| 6429 |
+
"loss": 63.0938,
|
| 6430 |
+
"step": 916
|
| 6431 |
+
},
|
| 6432 |
+
{
|
| 6433 |
+
"epoch": 9.17e-05,
|
| 6434 |
+
"grad_norm": 9.809439659118652,
|
| 6435 |
+
"learning_rate": 9.16e-06,
|
| 6436 |
+
"loss": 63.3438,
|
| 6437 |
+
"step": 917
|
| 6438 |
+
},
|
| 6439 |
+
{
|
| 6440 |
+
"epoch": 9.18e-05,
|
| 6441 |
+
"grad_norm": 9.726333618164062,
|
| 6442 |
+
"learning_rate": 9.17e-06,
|
| 6443 |
+
"loss": 63.4062,
|
| 6444 |
+
"step": 918
|
| 6445 |
+
},
|
| 6446 |
+
{
|
| 6447 |
+
"epoch": 9.19e-05,
|
| 6448 |
+
"grad_norm": 10.27943229675293,
|
| 6449 |
+
"learning_rate": 9.18e-06,
|
| 6450 |
+
"loss": 62.7812,
|
| 6451 |
+
"step": 919
|
| 6452 |
+
},
|
| 6453 |
+
{
|
| 6454 |
+
"epoch": 9.2e-05,
|
| 6455 |
+
"grad_norm": 9.923724174499512,
|
| 6456 |
+
"learning_rate": 9.19e-06,
|
| 6457 |
+
"loss": 63.125,
|
| 6458 |
+
"step": 920
|
| 6459 |
+
},
|
| 6460 |
+
{
|
| 6461 |
+
"epoch": 9.21e-05,
|
| 6462 |
+
"grad_norm": 9.776262283325195,
|
| 6463 |
+
"learning_rate": 9.2e-06,
|
| 6464 |
+
"loss": 63.1562,
|
| 6465 |
+
"step": 921
|
| 6466 |
+
},
|
| 6467 |
+
{
|
| 6468 |
+
"epoch": 9.22e-05,
|
| 6469 |
+
"grad_norm": 10.05753231048584,
|
| 6470 |
+
"learning_rate": 9.21e-06,
|
| 6471 |
+
"loss": 62.5312,
|
| 6472 |
+
"step": 922
|
| 6473 |
+
},
|
| 6474 |
+
{
|
| 6475 |
+
"epoch": 9.23e-05,
|
| 6476 |
+
"grad_norm": 10.067216873168945,
|
| 6477 |
+
"learning_rate": 9.22e-06,
|
| 6478 |
+
"loss": 62.6875,
|
| 6479 |
+
"step": 923
|
| 6480 |
+
},
|
| 6481 |
+
{
|
| 6482 |
+
"epoch": 9.24e-05,
|
| 6483 |
+
"grad_norm": 9.938775062561035,
|
| 6484 |
+
"learning_rate": 9.23e-06,
|
| 6485 |
+
"loss": 62.5312,
|
| 6486 |
+
"step": 924
|
| 6487 |
+
},
|
| 6488 |
+
{
|
| 6489 |
+
"epoch": 9.25e-05,
|
| 6490 |
+
"grad_norm": 9.898990631103516,
|
| 6491 |
+
"learning_rate": 9.24e-06,
|
| 6492 |
+
"loss": 62.5938,
|
| 6493 |
+
"step": 925
|
| 6494 |
+
},
|
| 6495 |
+
{
|
| 6496 |
+
"epoch": 9.26e-05,
|
| 6497 |
+
"grad_norm": 9.89505672454834,
|
| 6498 |
+
"learning_rate": 9.25e-06,
|
| 6499 |
+
"loss": 62.75,
|
| 6500 |
+
"step": 926
|
| 6501 |
+
},
|
| 6502 |
+
{
|
| 6503 |
+
"epoch": 9.27e-05,
|
| 6504 |
+
"grad_norm": 9.820269584655762,
|
| 6505 |
+
"learning_rate": 9.260000000000001e-06,
|
| 6506 |
+
"loss": 62.8125,
|
| 6507 |
+
"step": 927
|
| 6508 |
+
},
|
| 6509 |
+
{
|
| 6510 |
+
"epoch": 9.28e-05,
|
| 6511 |
+
"grad_norm": 9.785930633544922,
|
| 6512 |
+
"learning_rate": 9.27e-06,
|
| 6513 |
+
"loss": 62.7188,
|
| 6514 |
+
"step": 928
|
| 6515 |
+
},
|
| 6516 |
+
{
|
| 6517 |
+
"epoch": 9.29e-05,
|
| 6518 |
+
"grad_norm": 9.649805068969727,
|
| 6519 |
+
"learning_rate": 9.280000000000001e-06,
|
| 6520 |
+
"loss": 62.75,
|
| 6521 |
+
"step": 929
|
| 6522 |
+
},
|
| 6523 |
+
{
|
| 6524 |
+
"epoch": 9.3e-05,
|
| 6525 |
+
"grad_norm": 9.778607368469238,
|
| 6526 |
+
"learning_rate": 9.29e-06,
|
| 6527 |
+
"loss": 62.5312,
|
| 6528 |
+
"step": 930
|
| 6529 |
+
},
|
| 6530 |
+
{
|
| 6531 |
+
"epoch": 9.31e-05,
|
| 6532 |
+
"grad_norm": 9.530630111694336,
|
| 6533 |
+
"learning_rate": 9.299999999999999e-06,
|
| 6534 |
+
"loss": 62.75,
|
| 6535 |
+
"step": 931
|
| 6536 |
+
},
|
| 6537 |
+
{
|
| 6538 |
+
"epoch": 9.32e-05,
|
| 6539 |
+
"grad_norm": 9.770853042602539,
|
| 6540 |
+
"learning_rate": 9.31e-06,
|
| 6541 |
+
"loss": 62.375,
|
| 6542 |
+
"step": 932
|
| 6543 |
+
},
|
| 6544 |
+
{
|
| 6545 |
+
"epoch": 9.33e-05,
|
| 6546 |
+
"grad_norm": 9.653392791748047,
|
| 6547 |
+
"learning_rate": 9.32e-06,
|
| 6548 |
+
"loss": 62.6875,
|
| 6549 |
+
"step": 933
|
| 6550 |
+
},
|
| 6551 |
+
{
|
| 6552 |
+
"epoch": 9.34e-05,
|
| 6553 |
+
"grad_norm": 9.52197265625,
|
| 6554 |
+
"learning_rate": 9.33e-06,
|
| 6555 |
+
"loss": 62.4688,
|
| 6556 |
+
"step": 934
|
| 6557 |
+
},
|
| 6558 |
+
{
|
| 6559 |
+
"epoch": 9.35e-05,
|
| 6560 |
+
"grad_norm": 9.621892929077148,
|
| 6561 |
+
"learning_rate": 9.34e-06,
|
| 6562 |
+
"loss": 62.4375,
|
| 6563 |
+
"step": 935
|
| 6564 |
+
},
|
| 6565 |
+
{
|
| 6566 |
+
"epoch": 9.36e-05,
|
| 6567 |
+
"grad_norm": 9.49044132232666,
|
| 6568 |
+
"learning_rate": 9.35e-06,
|
| 6569 |
+
"loss": 63.0938,
|
| 6570 |
+
"step": 936
|
| 6571 |
+
},
|
| 6572 |
+
{
|
| 6573 |
+
"epoch": 9.37e-05,
|
| 6574 |
+
"grad_norm": 9.625767707824707,
|
| 6575 |
+
"learning_rate": 9.36e-06,
|
| 6576 |
+
"loss": 62.6875,
|
| 6577 |
+
"step": 937
|
| 6578 |
+
},
|
| 6579 |
+
{
|
| 6580 |
+
"epoch": 9.38e-05,
|
| 6581 |
+
"grad_norm": 9.668294906616211,
|
| 6582 |
+
"learning_rate": 9.37e-06,
|
| 6583 |
+
"loss": 62.2812,
|
| 6584 |
+
"step": 938
|
| 6585 |
+
},
|
| 6586 |
+
{
|
| 6587 |
+
"epoch": 9.39e-05,
|
| 6588 |
+
"grad_norm": 9.578293800354004,
|
| 6589 |
+
"learning_rate": 9.38e-06,
|
| 6590 |
+
"loss": 62.3125,
|
| 6591 |
+
"step": 939
|
| 6592 |
+
},
|
| 6593 |
+
{
|
| 6594 |
+
"epoch": 9.4e-05,
|
| 6595 |
+
"grad_norm": 9.879005432128906,
|
| 6596 |
+
"learning_rate": 9.39e-06,
|
| 6597 |
+
"loss": 62.3125,
|
| 6598 |
+
"step": 940
|
| 6599 |
+
},
|
| 6600 |
+
{
|
| 6601 |
+
"epoch": 9.41e-05,
|
| 6602 |
+
"grad_norm": 9.26992130279541,
|
| 6603 |
+
"learning_rate": 9.4e-06,
|
| 6604 |
+
"loss": 62.5938,
|
| 6605 |
+
"step": 941
|
| 6606 |
+
},
|
| 6607 |
+
{
|
| 6608 |
+
"epoch": 9.42e-05,
|
| 6609 |
+
"grad_norm": 9.741432189941406,
|
| 6610 |
+
"learning_rate": 9.41e-06,
|
| 6611 |
+
"loss": 62.125,
|
| 6612 |
+
"step": 942
|
| 6613 |
+
},
|
| 6614 |
+
{
|
| 6615 |
+
"epoch": 9.43e-05,
|
| 6616 |
+
"grad_norm": 9.542763710021973,
|
| 6617 |
+
"learning_rate": 9.42e-06,
|
| 6618 |
+
"loss": 62.4062,
|
| 6619 |
+
"step": 943
|
| 6620 |
+
},
|
| 6621 |
+
{
|
| 6622 |
+
"epoch": 9.44e-05,
|
| 6623 |
+
"grad_norm": 9.331867218017578,
|
| 6624 |
+
"learning_rate": 9.430000000000001e-06,
|
| 6625 |
+
"loss": 62.375,
|
| 6626 |
+
"step": 944
|
| 6627 |
+
},
|
| 6628 |
+
{
|
| 6629 |
+
"epoch": 9.45e-05,
|
| 6630 |
+
"grad_norm": 9.504682540893555,
|
| 6631 |
+
"learning_rate": 9.44e-06,
|
| 6632 |
+
"loss": 62.0312,
|
| 6633 |
+
"step": 945
|
| 6634 |
+
},
|
| 6635 |
+
{
|
| 6636 |
+
"epoch": 9.46e-05,
|
| 6637 |
+
"grad_norm": 9.386909484863281,
|
| 6638 |
+
"learning_rate": 9.450000000000001e-06,
|
| 6639 |
+
"loss": 62.375,
|
| 6640 |
+
"step": 946
|
| 6641 |
+
},
|
| 6642 |
+
{
|
| 6643 |
+
"epoch": 9.47e-05,
|
| 6644 |
+
"grad_norm": 9.543030738830566,
|
| 6645 |
+
"learning_rate": 9.460000000000001e-06,
|
| 6646 |
+
"loss": 62.0938,
|
| 6647 |
+
"step": 947
|
| 6648 |
+
},
|
| 6649 |
+
{
|
| 6650 |
+
"epoch": 9.48e-05,
|
| 6651 |
+
"grad_norm": 9.453919410705566,
|
| 6652 |
+
"learning_rate": 9.469999999999999e-06,
|
| 6653 |
+
"loss": 62.375,
|
| 6654 |
+
"step": 948
|
| 6655 |
+
},
|
| 6656 |
+
{
|
| 6657 |
+
"epoch": 9.49e-05,
|
| 6658 |
+
"grad_norm": 9.36674690246582,
|
| 6659 |
+
"learning_rate": 9.48e-06,
|
| 6660 |
+
"loss": 62.0312,
|
| 6661 |
+
"step": 949
|
| 6662 |
+
},
|
| 6663 |
+
{
|
| 6664 |
+
"epoch": 9.5e-05,
|
| 6665 |
+
"grad_norm": 9.571072578430176,
|
| 6666 |
+
"learning_rate": 9.49e-06,
|
| 6667 |
+
"loss": 62.0312,
|
| 6668 |
+
"step": 950
|
| 6669 |
+
},
|
| 6670 |
+
{
|
| 6671 |
+
"epoch": 9.51e-05,
|
| 6672 |
+
"grad_norm": 9.615150451660156,
|
| 6673 |
+
"learning_rate": 9.5e-06,
|
| 6674 |
+
"loss": 61.8125,
|
| 6675 |
+
"step": 951
|
| 6676 |
+
},
|
| 6677 |
+
{
|
| 6678 |
+
"epoch": 9.52e-05,
|
| 6679 |
+
"grad_norm": 9.266009330749512,
|
| 6680 |
+
"learning_rate": 9.51e-06,
|
| 6681 |
+
"loss": 62.375,
|
| 6682 |
+
"step": 952
|
| 6683 |
+
},
|
| 6684 |
+
{
|
| 6685 |
+
"epoch": 9.53e-05,
|
| 6686 |
+
"grad_norm": 9.239862442016602,
|
| 6687 |
+
"learning_rate": 9.52e-06,
|
| 6688 |
+
"loss": 62.3438,
|
| 6689 |
+
"step": 953
|
| 6690 |
+
},
|
| 6691 |
+
{
|
| 6692 |
+
"epoch": 9.54e-05,
|
| 6693 |
+
"grad_norm": 9.388384819030762,
|
| 6694 |
+
"learning_rate": 9.53e-06,
|
| 6695 |
+
"loss": 61.9062,
|
| 6696 |
+
"step": 954
|
| 6697 |
+
},
|
| 6698 |
+
{
|
| 6699 |
+
"epoch": 9.55e-05,
|
| 6700 |
+
"grad_norm": 9.385138511657715,
|
| 6701 |
+
"learning_rate": 9.54e-06,
|
| 6702 |
+
"loss": 61.9375,
|
| 6703 |
+
"step": 955
|
| 6704 |
+
},
|
| 6705 |
+
{
|
| 6706 |
+
"epoch": 9.56e-05,
|
| 6707 |
+
"grad_norm": 9.419622421264648,
|
| 6708 |
+
"learning_rate": 9.55e-06,
|
| 6709 |
+
"loss": 61.7188,
|
| 6710 |
+
"step": 956
|
| 6711 |
+
},
|
| 6712 |
+
{
|
| 6713 |
+
"epoch": 9.57e-05,
|
| 6714 |
+
"grad_norm": 9.560093879699707,
|
| 6715 |
+
"learning_rate": 9.56e-06,
|
| 6716 |
+
"loss": 61.6562,
|
| 6717 |
+
"step": 957
|
| 6718 |
+
},
|
| 6719 |
+
{
|
| 6720 |
+
"epoch": 9.58e-05,
|
| 6721 |
+
"grad_norm": 9.4889554977417,
|
| 6722 |
+
"learning_rate": 9.57e-06,
|
| 6723 |
+
"loss": 62.0625,
|
| 6724 |
+
"step": 958
|
| 6725 |
+
},
|
| 6726 |
+
{
|
| 6727 |
+
"epoch": 9.59e-05,
|
| 6728 |
+
"grad_norm": 8.988265991210938,
|
| 6729 |
+
"learning_rate": 9.58e-06,
|
| 6730 |
+
"loss": 62.3125,
|
| 6731 |
+
"step": 959
|
| 6732 |
+
},
|
| 6733 |
+
{
|
| 6734 |
+
"epoch": 9.6e-05,
|
| 6735 |
+
"grad_norm": 9.52734661102295,
|
| 6736 |
+
"learning_rate": 9.59e-06,
|
| 6737 |
+
"loss": 61.5312,
|
| 6738 |
+
"step": 960
|
| 6739 |
+
},
|
| 6740 |
+
{
|
| 6741 |
+
"epoch": 9.61e-05,
|
| 6742 |
+
"grad_norm": 9.097179412841797,
|
| 6743 |
+
"learning_rate": 9.600000000000001e-06,
|
| 6744 |
+
"loss": 62.1562,
|
| 6745 |
+
"step": 961
|
| 6746 |
+
},
|
| 6747 |
+
{
|
| 6748 |
+
"epoch": 9.62e-05,
|
| 6749 |
+
"grad_norm": 9.454554557800293,
|
| 6750 |
+
"learning_rate": 9.61e-06,
|
| 6751 |
+
"loss": 61.4688,
|
| 6752 |
+
"step": 962
|
| 6753 |
+
},
|
| 6754 |
+
{
|
| 6755 |
+
"epoch": 9.63e-05,
|
| 6756 |
+
"grad_norm": 9.153898239135742,
|
| 6757 |
+
"learning_rate": 9.620000000000001e-06,
|
| 6758 |
+
"loss": 62.125,
|
| 6759 |
+
"step": 963
|
| 6760 |
+
},
|
| 6761 |
+
{
|
| 6762 |
+
"epoch": 9.64e-05,
|
| 6763 |
+
"grad_norm": 9.330815315246582,
|
| 6764 |
+
"learning_rate": 9.630000000000001e-06,
|
| 6765 |
+
"loss": 61.6562,
|
| 6766 |
+
"step": 964
|
| 6767 |
+
},
|
| 6768 |
+
{
|
| 6769 |
+
"epoch": 9.65e-05,
|
| 6770 |
+
"grad_norm": 9.72275161743164,
|
| 6771 |
+
"learning_rate": 9.64e-06,
|
| 6772 |
+
"loss": 61.3125,
|
| 6773 |
+
"step": 965
|
| 6774 |
+
},
|
| 6775 |
+
{
|
| 6776 |
+
"epoch": 9.66e-05,
|
| 6777 |
+
"grad_norm": 9.388081550598145,
|
| 6778 |
+
"learning_rate": 9.65e-06,
|
| 6779 |
+
"loss": 61.4688,
|
| 6780 |
+
"step": 966
|
| 6781 |
+
},
|
| 6782 |
+
{
|
| 6783 |
+
"epoch": 9.67e-05,
|
| 6784 |
+
"grad_norm": 9.236139297485352,
|
| 6785 |
+
"learning_rate": 9.66e-06,
|
| 6786 |
+
"loss": 61.5938,
|
| 6787 |
+
"step": 967
|
| 6788 |
+
},
|
| 6789 |
+
{
|
| 6790 |
+
"epoch": 9.68e-05,
|
| 6791 |
+
"grad_norm": 9.027604103088379,
|
| 6792 |
+
"learning_rate": 9.67e-06,
|
| 6793 |
+
"loss": 62.0312,
|
| 6794 |
+
"step": 968
|
| 6795 |
+
},
|
| 6796 |
+
{
|
| 6797 |
+
"epoch": 9.69e-05,
|
| 6798 |
+
"grad_norm": 9.193624496459961,
|
| 6799 |
+
"learning_rate": 9.68e-06,
|
| 6800 |
+
"loss": 61.5,
|
| 6801 |
+
"step": 969
|
| 6802 |
+
},
|
| 6803 |
+
{
|
| 6804 |
+
"epoch": 9.7e-05,
|
| 6805 |
+
"grad_norm": 9.14177131652832,
|
| 6806 |
+
"learning_rate": 9.69e-06,
|
| 6807 |
+
"loss": 61.75,
|
| 6808 |
+
"step": 970
|
| 6809 |
+
},
|
| 6810 |
+
{
|
| 6811 |
+
"epoch": 9.71e-05,
|
| 6812 |
+
"grad_norm": 9.21058177947998,
|
| 6813 |
+
"learning_rate": 9.7e-06,
|
| 6814 |
+
"loss": 61.75,
|
| 6815 |
+
"step": 971
|
| 6816 |
+
},
|
| 6817 |
+
{
|
| 6818 |
+
"epoch": 9.72e-05,
|
| 6819 |
+
"grad_norm": 9.118620872497559,
|
| 6820 |
+
"learning_rate": 9.71e-06,
|
| 6821 |
+
"loss": 61.75,
|
| 6822 |
+
"step": 972
|
| 6823 |
+
},
|
| 6824 |
+
{
|
| 6825 |
+
"epoch": 9.73e-05,
|
| 6826 |
+
"grad_norm": 9.088716506958008,
|
| 6827 |
+
"learning_rate": 9.72e-06,
|
| 6828 |
+
"loss": 61.625,
|
| 6829 |
+
"step": 973
|
| 6830 |
+
},
|
| 6831 |
+
{
|
| 6832 |
+
"epoch": 9.74e-05,
|
| 6833 |
+
"grad_norm": 9.206241607666016,
|
| 6834 |
+
"learning_rate": 9.73e-06,
|
| 6835 |
+
"loss": 61.25,
|
| 6836 |
+
"step": 974
|
| 6837 |
+
},
|
| 6838 |
+
{
|
| 6839 |
+
"epoch": 9.75e-05,
|
| 6840 |
+
"grad_norm": 9.151651382446289,
|
| 6841 |
+
"learning_rate": 9.74e-06,
|
| 6842 |
+
"loss": 61.2188,
|
| 6843 |
+
"step": 975
|
| 6844 |
+
},
|
| 6845 |
+
{
|
| 6846 |
+
"epoch": 9.76e-05,
|
| 6847 |
+
"grad_norm": 9.240182876586914,
|
| 6848 |
+
"learning_rate": 9.75e-06,
|
| 6849 |
+
"loss": 61.4688,
|
| 6850 |
+
"step": 976
|
| 6851 |
+
},
|
| 6852 |
+
{
|
| 6853 |
+
"epoch": 9.77e-05,
|
| 6854 |
+
"grad_norm": 9.352987289428711,
|
| 6855 |
+
"learning_rate": 9.76e-06,
|
| 6856 |
+
"loss": 61.25,
|
| 6857 |
+
"step": 977
|
| 6858 |
+
},
|
| 6859 |
+
{
|
| 6860 |
+
"epoch": 9.78e-05,
|
| 6861 |
+
"grad_norm": 9.508001327514648,
|
| 6862 |
+
"learning_rate": 9.770000000000001e-06,
|
| 6863 |
+
"loss": 61.0,
|
| 6864 |
+
"step": 978
|
| 6865 |
+
},
|
| 6866 |
+
{
|
| 6867 |
+
"epoch": 9.79e-05,
|
| 6868 |
+
"grad_norm": 8.883227348327637,
|
| 6869 |
+
"learning_rate": 9.78e-06,
|
| 6870 |
+
"loss": 62.0,
|
| 6871 |
+
"step": 979
|
| 6872 |
+
},
|
| 6873 |
+
{
|
| 6874 |
+
"epoch": 9.8e-05,
|
| 6875 |
+
"grad_norm": 9.131720542907715,
|
| 6876 |
+
"learning_rate": 9.790000000000001e-06,
|
| 6877 |
+
"loss": 61.125,
|
| 6878 |
+
"step": 980
|
| 6879 |
+
},
|
| 6880 |
+
{
|
| 6881 |
+
"epoch": 9.81e-05,
|
| 6882 |
+
"grad_norm": 9.582850456237793,
|
| 6883 |
+
"learning_rate": 9.800000000000001e-06,
|
| 6884 |
+
"loss": 60.6875,
|
| 6885 |
+
"step": 981
|
| 6886 |
+
},
|
| 6887 |
+
{
|
| 6888 |
+
"epoch": 9.82e-05,
|
| 6889 |
+
"grad_norm": 9.125800132751465,
|
| 6890 |
+
"learning_rate": 9.81e-06,
|
| 6891 |
+
"loss": 61.0625,
|
| 6892 |
+
"step": 982
|
| 6893 |
+
},
|
| 6894 |
+
{
|
| 6895 |
+
"epoch": 9.83e-05,
|
| 6896 |
+
"grad_norm": 9.01922607421875,
|
| 6897 |
+
"learning_rate": 9.820000000000001e-06,
|
| 6898 |
+
"loss": 61.125,
|
| 6899 |
+
"step": 983
|
| 6900 |
+
},
|
| 6901 |
+
{
|
| 6902 |
+
"epoch": 9.84e-05,
|
| 6903 |
+
"grad_norm": 9.03088092803955,
|
| 6904 |
+
"learning_rate": 9.83e-06,
|
| 6905 |
+
"loss": 61.0625,
|
| 6906 |
+
"step": 984
|
| 6907 |
+
},
|
| 6908 |
+
{
|
| 6909 |
+
"epoch": 9.85e-05,
|
| 6910 |
+
"grad_norm": 9.083083152770996,
|
| 6911 |
+
"learning_rate": 9.84e-06,
|
| 6912 |
+
"loss": 61.0312,
|
| 6913 |
+
"step": 985
|
| 6914 |
+
},
|
| 6915 |
+
{
|
| 6916 |
+
"epoch": 9.86e-05,
|
| 6917 |
+
"grad_norm": 9.077742576599121,
|
| 6918 |
+
"learning_rate": 9.85e-06,
|
| 6919 |
+
"loss": 61.0312,
|
| 6920 |
+
"step": 986
|
| 6921 |
+
},
|
| 6922 |
+
{
|
| 6923 |
+
"epoch": 9.87e-05,
|
| 6924 |
+
"grad_norm": 9.057887077331543,
|
| 6925 |
+
"learning_rate": 9.859999999999999e-06,
|
| 6926 |
+
"loss": 60.875,
|
| 6927 |
+
"step": 987
|
| 6928 |
+
},
|
| 6929 |
+
{
|
| 6930 |
+
"epoch": 9.88e-05,
|
| 6931 |
+
"grad_norm": 8.55154800415039,
|
| 6932 |
+
"learning_rate": 9.87e-06,
|
| 6933 |
+
"loss": 62.375,
|
| 6934 |
+
"step": 988
|
| 6935 |
+
},
|
| 6936 |
+
{
|
| 6937 |
+
"epoch": 9.89e-05,
|
| 6938 |
+
"grad_norm": 8.761177062988281,
|
| 6939 |
+
"learning_rate": 9.88e-06,
|
| 6940 |
+
"loss": 61.125,
|
| 6941 |
+
"step": 989
|
| 6942 |
+
},
|
| 6943 |
+
{
|
| 6944 |
+
"epoch": 9.9e-05,
|
| 6945 |
+
"grad_norm": 9.016249656677246,
|
| 6946 |
+
"learning_rate": 9.89e-06,
|
| 6947 |
+
"loss": 61.2188,
|
| 6948 |
+
"step": 990
|
| 6949 |
+
},
|
| 6950 |
+
{
|
| 6951 |
+
"epoch": 9.91e-05,
|
| 6952 |
+
"grad_norm": 8.754048347473145,
|
| 6953 |
+
"learning_rate": 9.9e-06,
|
| 6954 |
+
"loss": 61.1875,
|
| 6955 |
+
"step": 991
|
| 6956 |
+
},
|
| 6957 |
+
{
|
| 6958 |
+
"epoch": 9.92e-05,
|
| 6959 |
+
"grad_norm": 9.200346946716309,
|
| 6960 |
+
"learning_rate": 9.91e-06,
|
| 6961 |
+
"loss": 60.5,
|
| 6962 |
+
"step": 992
|
| 6963 |
+
},
|
| 6964 |
+
{
|
| 6965 |
+
"epoch": 9.93e-05,
|
| 6966 |
+
"grad_norm": 8.876838684082031,
|
| 6967 |
+
"learning_rate": 9.92e-06,
|
| 6968 |
+
"loss": 61.0,
|
| 6969 |
+
"step": 993
|
| 6970 |
+
},
|
| 6971 |
+
{
|
| 6972 |
+
"epoch": 9.94e-05,
|
| 6973 |
+
"grad_norm": 8.757495880126953,
|
| 6974 |
+
"learning_rate": 9.93e-06,
|
| 6975 |
+
"loss": 61.4062,
|
| 6976 |
+
"step": 994
|
| 6977 |
+
},
|
| 6978 |
+
{
|
| 6979 |
+
"epoch": 9.95e-05,
|
| 6980 |
+
"grad_norm": 8.726040840148926,
|
| 6981 |
+
"learning_rate": 9.940000000000001e-06,
|
| 6982 |
+
"loss": 61.6562,
|
| 6983 |
+
"step": 995
|
| 6984 |
+
},
|
| 6985 |
+
{
|
| 6986 |
+
"epoch": 9.96e-05,
|
| 6987 |
+
"grad_norm": 8.834831237792969,
|
| 6988 |
+
"learning_rate": 9.95e-06,
|
| 6989 |
+
"loss": 61.125,
|
| 6990 |
+
"step": 996
|
| 6991 |
+
},
|
| 6992 |
+
{
|
| 6993 |
+
"epoch": 9.97e-05,
|
| 6994 |
+
"grad_norm": 8.835776329040527,
|
| 6995 |
+
"learning_rate": 9.96e-06,
|
| 6996 |
+
"loss": 60.6562,
|
| 6997 |
+
"step": 997
|
| 6998 |
+
},
|
| 6999 |
+
{
|
| 7000 |
+
"epoch": 9.98e-05,
|
| 7001 |
+
"grad_norm": 8.862153053283691,
|
| 7002 |
+
"learning_rate": 9.970000000000001e-06,
|
| 7003 |
+
"loss": 61.0312,
|
| 7004 |
+
"step": 998
|
| 7005 |
+
},
|
| 7006 |
+
{
|
| 7007 |
+
"epoch": 9.99e-05,
|
| 7008 |
+
"grad_norm": 9.10541820526123,
|
| 7009 |
+
"learning_rate": 9.98e-06,
|
| 7010 |
+
"loss": 60.5625,
|
| 7011 |
+
"step": 999
|
| 7012 |
+
},
|
| 7013 |
+
{
|
| 7014 |
+
"epoch": 0.0001,
|
| 7015 |
+
"grad_norm": 9.29044246673584,
|
| 7016 |
+
"learning_rate": 9.990000000000001e-06,
|
| 7017 |
+
"loss": 60.375,
|
| 7018 |
+
"step": 1000
|
| 7019 |
+
},
|
| 7020 |
+
{
|
| 7021 |
+
"epoch": 0.0001,
|
| 7022 |
+
"eval_loss": 7.538768768310547,
|
| 7023 |
+
"eval_runtime": 364.1443,
|
| 7024 |
+
"eval_samples_per_second": 27.462,
|
| 7025 |
+
"eval_steps_per_second": 1.716,
|
| 7026 |
+
"step": 1000
|
| 7027 |
}
|
| 7028 |
],
|
| 7029 |
"logging_steps": 1,
|